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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
4cb98cd3-4506-49ed-a9c5-625eb9c3ace5 | glosh-global-local-spherical-harmonics-for | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Zhou_GLoSH_Global-Local_Spherical_Harmonics_for_Intrinsic_Image_Decomposition_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Zhou_GLoSH_Global-Local_Spherical_Harmonics_for_Intrinsic_Image_Decomposition_ICCV_2019_paper.pdf | GLoSH: Global-Local Spherical Harmonics for Intrinsic Image Decomposition | Traditional intrinsic image decomposition focuses on decomposing images into reflectance and shading, leaving surfaces normals and lighting entangled in shading. In this work, we propose a Global-Local Spherical Harmonics (GLoSH) lighting model to improve the lighting component, and jointly predict reflectance and surface normals. The global SH models the holistic lighting while local SH account for the spatial variation of lighting. Also, a novel non-negative lighting constraint is proposed to encourage the estimated SH to be physically meaningful. To seamlessly reflect the GLoSH model, we design a coarse-to-fine network structure. The coarse network predicts global SH, reflectance and normals, and the fine network predicts their local residuals. Lacking labels for reflectance and lighting, we apply synthetic data for model pre-training and fine-tune the model with real data in a self-supervised way. Compared to the state-of-the-art methods only targeting normals or reflectance and shading, our method recovers all components and achieves consistently better results on three real datasets, IIW, SAW and NYUv2.
| [' David W. Jacobs', ' Xiang Yu', 'Hao Zhou'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['intrinsic-image-decomposition'] | ['computer-vision'] | [ 4.06807095e-01 -5.00737131e-02 2.06520259e-01 -5.32748997e-01
-3.70626420e-01 -3.17206413e-01 4.89033937e-01 -5.19295573e-01
2.90658444e-01 3.88789088e-01 3.56455892e-01 -8.55628476e-02
1.54365122e-01 -8.14809978e-01 -7.36575365e-01 -8.69819045e-01
5.05784333e-01 1.88906174e-02 6.30236045e-02 -1.17227659e-01
-1.62738711e-01 5.50880611e-01 -1.68696475e+00 3.28360140e-01
1.00672507e+00 9.06243801e-01 2.06738412e-01 7.15181947e-01
-2.15081424e-01 8.16150606e-01 -2.47821793e-01 9.73792523e-02
6.10264063e-01 -4.74364519e-01 -1.56957284e-01 3.18711907e-01
8.12629461e-01 -3.33761692e-01 -1.03611395e-01 8.37442160e-01
4.01555926e-01 8.85482803e-02 5.31251490e-01 -1.02994382e+00
-1.06678200e+00 -6.19253889e-02 -9.25959945e-01 -7.42113233e-01
2.01468810e-01 2.56132245e-01 1.02992857e+00 -1.01811814e+00
2.97570318e-01 1.22366977e+00 8.29283834e-01 4.51826394e-01
-1.49331427e+00 -5.18041790e-01 4.11984175e-01 -3.20337415e-01
-1.31678963e+00 -4.49834943e-01 1.12428820e+00 -4.71323520e-01
6.19718909e-01 4.96188432e-01 7.13595927e-01 9.53023911e-01
-1.48752779e-01 5.43236792e-01 1.41497946e+00 -4.74561602e-01
1.13108955e-01 1.77904908e-02 2.84643956e-02 7.55354524e-01
-9.87760127e-02 1.78332984e-01 -7.19790220e-01 -1.39285386e-01
9.19170737e-01 2.03055561e-01 -6.90804601e-01 -3.93053889e-01
-1.02040577e+00 4.83100027e-01 5.34334004e-01 -2.77857810e-01
-3.90600026e-01 1.30520970e-01 -2.47885495e-01 -5.27992779e-05
7.93003082e-01 1.50126800e-01 -7.51737893e-01 4.37679738e-01
-8.87929618e-01 -3.29549275e-02 6.32165790e-01 7.10848153e-01
1.31335044e+00 1.61671996e-01 -1.91085622e-01 1.18394566e+00
7.18821347e-01 8.18430841e-01 4.60434034e-02 -1.31033242e+00
8.42187181e-02 7.26280093e-01 1.75334617e-01 -9.63053048e-01
-4.30839121e-01 -3.41543138e-01 -1.01309168e+00 3.72114062e-01
1.05734617e-01 1.28055990e-01 -1.28118980e+00 1.66195917e+00
4.71212000e-01 4.79928613e-01 -1.30734697e-01 9.83163178e-01
1.02370048e+00 7.11092174e-01 -3.89653862e-01 -4.01936844e-02
9.84081507e-01 -1.43009651e+00 -5.90196192e-01 -3.53999615e-01
8.75011235e-02 -9.27956760e-01 1.70054793e+00 5.04631400e-01
-1.00111461e+00 -6.77701771e-01 -8.47281873e-01 -3.85284752e-01
2.21246034e-02 4.90577251e-01 5.04413426e-01 3.79403472e-01
-1.26990128e+00 5.77435076e-01 -8.15586805e-01 -1.61208838e-01
1.71508178e-01 -9.11020562e-02 8.26951936e-02 -3.19542080e-01
-8.03892910e-01 4.76266205e-01 -2.94707447e-01 5.04363298e-01
-7.95406222e-01 -8.58076394e-01 -9.15130734e-01 -1.12256613e-02
1.76054060e-01 -8.02132487e-01 7.78816819e-01 -1.03167033e+00
-1.93266559e+00 7.48891652e-01 -5.18927991e-01 2.70987660e-01
3.07843953e-01 -1.74886435e-01 -2.35682473e-01 -4.26096730e-02
-2.04310000e-01 6.47499382e-01 1.11844349e+00 -2.07781768e+00
-1.17756821e-01 -3.29973668e-01 1.40015979e-03 2.68162310e-01
-3.17197740e-01 -2.49389052e-01 -6.13272667e-01 -4.58957583e-01
4.87587005e-01 -6.39587402e-01 -1.03542335e-01 3.52115512e-01
-5.61527967e-01 2.04476252e-01 6.71967089e-01 -8.06316376e-01
8.44878137e-01 -2.16193914e+00 1.70480013e-02 2.42541835e-01
3.47852916e-01 -7.39842728e-02 -5.30465484e-01 2.26065889e-01
-1.50542617e-01 -1.05672970e-01 -3.06215376e-01 -9.16995347e-01
5.82607239e-02 6.54395700e-01 -4.34192508e-01 4.97354150e-01
1.31980926e-01 7.93298960e-01 -6.20301604e-01 -7.10001588e-02
3.41255188e-01 1.09791708e+00 -4.94969457e-01 4.60958779e-01
-2.84159064e-01 6.47147179e-01 -3.14203650e-02 7.20194340e-01
9.64217484e-01 -3.05798978e-01 -6.81062648e-03 -5.32418549e-01
-1.50456771e-01 2.51953393e-01 -1.18098152e+00 1.50655973e+00
-6.81467772e-01 3.47624213e-01 3.52573931e-01 -4.23027158e-01
1.25281525e+00 -7.87050724e-02 3.39635879e-01 -8.24839652e-01
-1.63519189e-01 -1.23323211e-02 -6.40033782e-01 -2.54176438e-01
1.74692467e-01 -8.83833412e-03 7.62446105e-01 2.98180491e-01
-3.97735804e-01 -2.35873803e-01 -2.95858562e-01 -6.48367032e-02
7.00483561e-01 8.79579306e-01 -1.53260529e-01 -2.33638257e-01
6.47236586e-01 -3.55612695e-01 6.95828140e-01 2.66402662e-01
3.07462275e-01 1.24185836e+00 3.04722786e-01 -5.45389891e-01
-9.60944772e-01 -1.10367227e+00 1.89176574e-01 1.26145363e+00
2.31150955e-01 -2.69056350e-01 -8.12510133e-01 -2.27393225e-01
-3.63689326e-02 7.19583631e-01 -7.26616919e-01 3.90229858e-02
-3.37382376e-01 -6.62731111e-01 9.87742841e-02 4.00019169e-01
4.94396627e-01 -9.81251478e-01 -2.22550124e-01 -3.97474498e-01
-1.28779158e-01 -1.00816548e+00 -5.10355592e-01 1.75255351e-02
-4.81587023e-01 -9.77934718e-01 -4.96138930e-01 -3.91248912e-01
1.02017128e+00 6.48893356e-01 1.30001116e+00 2.87091613e-01
-3.64220291e-01 3.30850571e-01 -1.45616651e-01 -1.62561163e-01
-8.46031830e-02 -4.22901630e-01 -2.79516429e-01 4.37442541e-01
4.16318513e-03 -6.94188058e-01 -7.76177347e-01 4.11579967e-01
-7.84476519e-01 6.18882656e-01 2.46426448e-01 7.22776055e-01
8.55787098e-01 -1.14986755e-01 -1.37838572e-01 -8.60776067e-01
2.44459271e-01 -2.64474284e-02 -6.30766273e-01 2.97171295e-01
-7.62081027e-01 -2.63193429e-01 7.29259551e-01 -2.84154117e-01
-1.34205103e+00 2.25005522e-01 -3.26672792e-02 -6.24882281e-01
-3.43878955e-01 1.46149129e-01 -3.53819758e-01 -1.94394737e-01
6.18202150e-01 2.33578622e-01 -2.29671255e-01 -7.81307638e-01
4.09816772e-01 1.70480072e-01 4.74297732e-01 -5.98447442e-01
1.06785238e+00 8.04974914e-01 1.05196007e-01 -9.93787050e-01
-1.26612008e+00 -4.75207806e-01 -5.60563862e-01 -1.44513816e-01
7.61472642e-01 -1.05958521e+00 -6.11770570e-01 7.17161059e-01
-1.18085527e+00 -8.78978968e-01 -6.00528002e-01 -7.65770581e-03
-3.07319194e-01 3.81117582e-01 -5.37990153e-01 -8.62531662e-01
-2.12604985e-01 -9.82838869e-01 1.49675369e+00 1.06599227e-01
2.55669028e-01 -1.05175459e+00 2.44825900e-01 4.50070947e-01
4.65163678e-01 3.14220876e-01 6.34131253e-01 4.34955537e-01
-6.15036666e-01 1.85248747e-01 -7.17464030e-01 7.96742499e-01
3.84789556e-01 4.17445958e-01 -1.40648377e+00 -2.16874465e-01
1.63712859e-01 -2.63759762e-01 1.05531490e+00 4.73892927e-01
1.34442246e+00 -2.65760005e-01 2.84715682e-01 1.34344542e+00
1.62636554e+00 -3.62068683e-01 6.92793667e-01 1.84949398e-01
1.32285511e+00 7.92531967e-01 1.85026526e-01 5.30101836e-01
8.56444478e-01 2.70968258e-01 6.57887340e-01 -8.90255690e-01
-4.04026389e-01 -2.69844711e-01 3.84471744e-01 7.68764436e-01
-1.94358274e-01 -1.57214120e-01 -7.07134485e-01 2.75091499e-01
-1.75216913e+00 -5.13150454e-01 -5.53439379e-01 2.09093118e+00
8.51789117e-01 -3.42666894e-01 -3.58430773e-01 -2.26163954e-01
3.82044137e-01 5.80208004e-01 -6.22014880e-01 -2.97819078e-02
-4.81685817e-01 1.17679268e-01 5.19337058e-01 9.38920140e-01
-6.37481809e-01 8.68154168e-01 6.43093443e+00 3.91354531e-01
-1.17685843e+00 5.28189354e-02 7.17323899e-01 -5.90623952e-02
-9.85044479e-01 1.68243721e-01 -6.03376567e-01 2.02871889e-01
5.16154289e-01 7.25698590e-01 1.11732197e+00 6.15950108e-01
5.64823151e-01 3.47382985e-02 -7.61240721e-01 9.13231492e-01
2.82958746e-01 -9.11244214e-01 -4.27308232e-02 -5.74460328e-02
1.06019711e+00 1.95344910e-01 -1.27003416e-01 -1.36768341e-01
4.42393333e-01 -1.09489763e+00 5.26131630e-01 9.77617562e-01
8.78413737e-01 -3.53117913e-01 2.94850647e-01 3.76233250e-01
-1.21716416e+00 1.30271405e-01 -4.91433591e-01 -1.13992676e-01
1.32165983e-01 8.72005343e-01 -2.61843085e-01 5.12978911e-01
7.06841469e-01 9.23438430e-01 -6.61750615e-01 4.09454554e-01
-7.91422486e-01 6.28708422e-01 -4.12223458e-01 5.80429852e-01
-1.01667315e-01 -7.58606672e-01 9.96738374e-02 1.01844347e+00
1.67423129e-01 -8.92329067e-02 5.05368710e-01 1.15430176e+00
-8.37609097e-02 1.14524113e-02 -1.96376473e-01 3.36711913e-01
1.01466432e-01 1.61442888e+00 -3.97355527e-01 -1.83404475e-01
-4.15712237e-01 1.11838460e+00 2.43805677e-01 9.47251678e-01
-7.54464507e-01 1.48266882e-01 8.20807517e-01 2.07163095e-01
4.42314111e-02 -2.79620856e-01 -7.28889704e-01 -1.34622598e+00
2.32395560e-01 -6.74653649e-01 -2.49174550e-01 -1.31604004e+00
-1.37516725e+00 3.88200849e-01 -4.78090078e-01 -1.00415826e+00
3.26544851e-01 -7.31611967e-01 -7.61539936e-01 1.20744526e+00
-2.28444648e+00 -1.63805246e+00 -1.10389960e+00 6.83740854e-01
3.94825876e-01 2.69120902e-01 7.80188620e-01 5.26319332e-02
-7.12915361e-01 1.12493470e-01 2.02987403e-01 -2.63387948e-01
9.80943799e-01 -1.37843120e+00 4.01210546e-01 8.29610944e-01
-3.93894129e-02 7.41641223e-01 5.46480596e-01 -4.96267796e-01
-1.47921848e+00 -1.18501413e+00 5.79451382e-01 -1.92166910e-01
5.43582618e-01 -4.98308748e-01 -9.43949461e-01 5.16969383e-01
2.19242603e-01 3.58936995e-01 5.24301887e-01 2.17265505e-02
-7.68769026e-01 -4.16570276e-01 -9.59644020e-01 5.88972270e-01
9.76058602e-01 -6.05952024e-01 3.52230221e-02 6.09145761e-01
9.83368933e-01 -4.34247494e-01 -6.68454230e-01 4.55904573e-01
6.64653063e-01 -1.34273541e+00 1.14727604e+00 -1.06581308e-01
6.08901203e-01 -6.71676397e-01 -3.94626558e-01 -1.29730988e+00
-4.71965462e-01 -5.76673329e-01 -2.71361470e-01 1.25177205e+00
2.18490928e-01 -7.80796826e-01 7.41421878e-01 5.75933278e-01
-3.38741153e-01 -7.65846372e-01 -2.75757611e-01 -4.17076945e-01
-3.19756985e-01 -4.25740480e-01 7.87727296e-01 1.05268919e+00
-9.32913899e-01 1.47439346e-01 -6.73372567e-01 4.31994766e-01
8.97698104e-01 2.89076447e-01 1.08508706e+00 -1.35262740e+00
-4.74339962e-01 -2.43380398e-01 4.00927752e-01 -1.27858472e+00
1.29755363e-01 -4.21055555e-01 3.26414257e-01 -1.77156985e+00
2.30848387e-01 -2.64693648e-01 -2.06496000e-01 7.86682427e-01
-2.33192310e-01 6.38335824e-01 2.89660506e-02 3.70677829e-01
-2.52780408e-01 7.77804792e-01 1.42842543e+00 1.55587755e-02
-3.98975790e-01 -2.92192250e-01 -8.47704053e-01 1.07718694e+00
7.11173773e-01 -2.82603782e-02 -5.40514052e-01 -7.68312633e-01
4.64308977e-01 -4.13294137e-01 5.25461197e-01 -5.90378463e-01
4.50893380e-02 -5.17276585e-01 5.09606123e-01 -7.23380327e-01
5.84341586e-01 -9.08798575e-01 1.46857589e-01 -8.90942737e-02
-1.86334938e-01 -3.88724148e-01 -1.88172296e-01 3.73228490e-01
6.98126629e-02 1.63440824e-01 8.76799703e-01 2.00910494e-03
-2.38526091e-01 4.40388978e-01 3.39234889e-01 -2.32433245e-01
4.79843557e-01 -3.56278419e-01 -4.97779727e-01 -3.01991314e-01
-4.14824665e-01 3.39330375e-01 1.02879715e+00 1.74178407e-01
7.01968312e-01 -1.09368980e+00 -6.36117101e-01 5.91201842e-01
8.36880803e-02 5.03785312e-01 3.19873244e-01 7.87137091e-01
-6.64205074e-01 -2.31552720e-01 2.31648460e-01 -6.86740160e-01
-1.18171716e+00 2.69788891e-01 3.76310438e-01 -9.50837284e-02
-6.93708837e-01 7.78325558e-01 7.79079318e-01 -1.01521087e+00
2.58476734e-01 -3.94468188e-01 3.07536609e-02 -2.86554456e-01
5.02794445e-01 3.72339249e-01 -2.06822917e-01 -5.67667544e-01
-7.58402869e-02 1.00642645e+00 5.03528237e-01 3.00638288e-01
1.58052254e+00 -4.54005569e-01 -6.72867119e-01 6.05598152e-01
1.11630630e+00 3.14818710e-01 -1.78866673e+00 -3.40265900e-01
-6.31398678e-01 -6.48028553e-01 2.68321574e-01 -9.02489662e-01
-1.24686301e+00 9.52874243e-01 3.38388294e-01 1.75795540e-01
1.54864168e+00 -2.93751687e-01 7.09083259e-01 2.05181673e-01
-1.86535921e-02 -1.03848481e+00 9.97706503e-02 6.86725736e-01
9.51203942e-01 -1.22388256e+00 2.03744739e-01 -6.34804308e-01
-6.45388424e-01 1.07988250e+00 6.73038185e-01 -2.10639104e-01
5.21552742e-01 4.05559212e-01 3.18461955e-01 -2.45804831e-01
-5.63909292e-01 -1.46063522e-01 6.20690823e-01 5.67694545e-01
4.36674535e-01 -4.36434196e-03 3.21839720e-01 1.19280905e-01
-2.06437260e-02 -1.92533717e-01 2.00290799e-01 3.67076695e-01
-3.50685716e-01 -8.79073560e-01 -5.23063719e-01 2.50814766e-01
7.83563405e-02 -3.64214003e-01 -5.66868842e-01 2.73317426e-01
2.74581045e-01 8.15039515e-01 -5.63980155e-02 -2.99120903e-01
3.27299237e-01 -1.40041232e-01 1.83883369e-01 -6.55187905e-01
-3.46620768e-01 5.77364445e-01 -1.30242765e-01 -9.30670798e-01
-6.30702078e-01 -3.94104213e-01 -1.13356638e+00 -1.31818488e-01
-3.92430156e-01 -4.98457432e-01 9.41157043e-01 7.70500064e-01
3.21007371e-01 7.65995264e-01 1.07879388e+00 -1.22234774e+00
-2.23887250e-01 -8.23115170e-01 -6.80306971e-01 3.96281719e-01
6.82982981e-01 -4.35095102e-01 -8.04281235e-01 2.82101721e-01] | [9.899665832519531, -2.9699509143829346] |
39e4ab8c-f5c3-4c89-8276-05462c02c14a | improving-paraphrase-detection-with-the | 2106.07691 | null | https://arxiv.org/abs/2106.07691v1 | https://arxiv.org/pdf/2106.07691v1.pdf | Improving Paraphrase Detection with the Adversarial Paraphrasing Task | If two sentences have the same meaning, it should follow that they are equivalent in their inferential properties, i.e., each sentence should textually entail the other. However, many paraphrase datasets currently in widespread use rely on a sense of paraphrase based on word overlap and syntax. Can we teach them instead to identify paraphrases in a way that draws on the inferential properties of the sentences, and is not over-reliant on lexical and syntactic similarities of a sentence pair? We apply the adversarial paradigm to this question, and introduce a new adversarial method of dataset creation for paraphrase identification: the Adversarial Paraphrasing Task (APT), which asks participants to generate semantically equivalent (in the sense of mutually implicative) but lexically and syntactically disparate paraphrases. These sentence pairs can then be used both to test paraphrase identification models (which get barely random accuracy) and then improve their performance. To accelerate dataset generation, we explore automation of APT using T5, and show that the resulting dataset also improves accuracy. We discuss implications for paraphrase detection and release our dataset in the hope of making paraphrase detection models better able to detect sentence-level meaning equivalence. | ['John Licato', 'Animesh Nighojkar'] | 2021-06-14 | null | https://aclanthology.org/2021.acl-long.552 | https://aclanthology.org/2021.acl-long.552.pdf | acl-2021-5 | ['paraphrase-identification'] | ['natural-language-processing'] | [ 5.83402157e-01 1.32239804e-01 1.17193803e-01 -5.46383977e-01
-7.00876296e-01 -1.25099635e+00 5.94248295e-01 5.47143519e-01
-1.94239244e-01 6.50737524e-01 6.83698595e-01 -7.61424899e-01
1.53813669e-02 -8.41273248e-01 -7.20509171e-01 1.91132091e-02
5.26122808e-01 4.15024698e-01 -2.30323081e-03 -5.80554903e-01
4.70521033e-01 3.21239352e-01 -1.38011467e+00 1.01203346e+00
1.06815314e+00 2.94497848e-01 -1.35124087e-01 6.93791986e-01
-1.21604607e-01 1.01110196e+00 -9.13347960e-01 -8.22909594e-01
3.36070240e-01 -9.18557465e-01 -1.32269549e+00 -4.49229598e-01
7.35924542e-01 -4.25836854e-02 -1.38040990e-01 9.98361588e-01
3.75423372e-01 -1.06270105e-01 5.99247396e-01 -1.12744617e+00
-9.55602586e-01 7.87438095e-01 1.36822630e-02 2.76147962e-01
1.22047353e+00 2.22896084e-01 1.20294380e+00 -6.85649097e-01
6.84492052e-01 1.14752245e+00 9.12831187e-01 5.54244220e-01
-1.57047558e+00 -7.27040112e-01 -2.22987995e-01 3.22209716e-01
-8.91300023e-01 -4.94415104e-01 7.48753786e-01 -3.26817423e-01
1.23246300e+00 6.67561710e-01 8.30196321e-01 1.53141010e+00
3.06780666e-01 4.10423934e-01 1.20672059e+00 -6.18462265e-01
1.47347867e-01 2.05198839e-01 1.11051172e-01 4.20717508e-01
2.22932085e-01 -2.52394527e-02 -6.24446452e-01 -3.83860201e-01
1.99267626e-01 -3.24022800e-01 -2.24831223e-01 -1.37890577e-01
-1.21284735e+00 1.07700610e+00 3.39761496e-01 5.41126251e-01
4.27957550e-02 -2.85345435e-01 6.20980501e-01 1.04232299e+00
1.71485335e-01 1.31084442e+00 -1.87487513e-01 -7.51064718e-03
-9.14424479e-01 7.06397414e-01 1.03170168e+00 8.44410837e-01
3.80420625e-01 -4.06177878e-01 -2.61395723e-01 9.25820827e-01
-2.31228527e-02 4.22283441e-01 1.04680550e+00 -8.94893408e-01
5.31443357e-01 7.12284327e-01 -1.97751559e-02 -9.78748083e-01
-6.13892414e-02 -1.05306115e-02 -3.97741497e-01 -1.00235015e-01
3.90984893e-01 2.73389667e-02 -3.02388191e-01 1.95935702e+00
-9.46851000e-02 -2.03428715e-01 1.62682757e-01 5.59969068e-01
5.96309364e-01 1.76653266e-01 -5.45797916e-03 3.83832082e-02
1.24320674e+00 -2.52924472e-01 -1.24899529e-01 -6.26860023e-01
9.88603234e-01 -1.03158271e+00 1.58624768e+00 5.95129207e-02
-1.19506788e+00 -4.71939832e-01 -1.27371907e+00 -1.57100692e-01
-2.36699060e-01 -6.20344818e-01 6.00179315e-01 8.96398544e-01
-9.05503690e-01 9.27842975e-01 -9.04189572e-02 -5.07100284e-01
1.84854493e-01 4.49582078e-02 -5.29260039e-01 -2.95100752e-02
-1.61541021e+00 1.24102283e+00 3.81546110e-01 -4.10693139e-01
-2.14899346e-01 -8.52274716e-01 -9.89669919e-01 2.03770455e-02
-2.44739130e-01 -1.19997144e+00 1.26090407e+00 -1.45990348e+00
-9.74264681e-01 1.39673197e+00 -2.58117169e-01 -5.17908990e-01
3.56788963e-01 2.95491200e-02 -3.08009207e-01 1.02155574e-01
4.53079581e-01 4.26454067e-01 8.02725971e-01 -6.57842219e-01
-1.81197807e-01 -4.82538581e-01 2.74983585e-01 2.27053180e-01
-2.01589495e-01 4.21741158e-01 6.59751654e-01 -8.59064162e-01
-2.56519150e-02 -8.84667575e-01 2.45267406e-01 -1.49879307e-01
-5.05657077e-01 -1.57555476e-01 2.24437371e-01 -7.31375396e-01
1.10241103e+00 -1.93608677e+00 -1.74333602e-02 1.49134442e-01
1.05834864e-01 2.58829474e-01 -2.66054600e-01 7.28330314e-01
-7.45706022e-01 5.70401013e-01 -2.63100177e-01 9.86040803e-04
7.45409429e-02 2.41640899e-02 -7.82336831e-01 1.61197424e-01
2.58175403e-01 1.21922600e+00 -1.07573617e+00 -4.61377531e-01
5.19371480e-02 -4.17795122e-01 -7.46395290e-01 1.92623392e-01
1.25613064e-02 -8.41206610e-02 -6.50628135e-02 7.79637694e-02
4.10746515e-01 2.46819347e-01 1.55438185e-01 -1.76435672e-02
3.72569054e-01 7.46380210e-01 -6.52536750e-01 1.48827136e+00
-7.63192475e-01 7.86022484e-01 -4.31643575e-01 -9.95612800e-01
7.38802195e-01 1.35628521e-01 -8.27850997e-02 -6.88286006e-01
-2.01507986e-01 3.21710646e-01 2.10312679e-01 -5.90553582e-01
4.03617561e-01 -8.80199790e-01 -3.71190459e-01 8.84716511e-01
-3.18531603e-01 -7.39605784e-01 7.87332654e-02 6.30863011e-01
1.37760067e+00 -2.71707416e-01 4.09549981e-01 -2.26174623e-01
4.27805215e-01 1.80780247e-01 1.78962082e-01 1.12682462e+00
-9.00887772e-02 6.69044197e-01 5.42063117e-01 -1.70841083e-01
-1.38326478e+00 -1.61275649e+00 -1.08483866e-01 8.40267241e-01
8.42168884e-05 -5.11707127e-01 -6.57433391e-01 -8.07617724e-01
8.34071487e-02 1.42864108e+00 -6.67868197e-01 -7.08077252e-01
-5.78339994e-01 -2.56215304e-01 1.10724699e+00 4.88778651e-01
7.88011998e-02 -1.28984964e+00 -3.65029097e-01 -3.10703777e-02
-4.89559323e-01 -7.56523073e-01 -5.00786066e-01 -8.64616483e-02
-6.42926574e-01 -1.23861170e+00 -1.88094914e-01 -8.41952384e-01
4.64364618e-01 1.87845990e-01 1.42873454e+00 1.71586320e-01
-1.09758064e-01 1.40929177e-01 -4.86084878e-01 -3.35927606e-01
-1.35789251e+00 -2.33122095e-01 4.27023554e-03 -4.86620277e-01
6.44116819e-01 -8.64395380e-01 -3.73719394e-01 8.64353180e-02
-9.73046601e-01 2.19382882e-01 3.55179161e-01 9.55741644e-01
-8.41359496e-02 -3.66871536e-01 8.72486591e-01 -1.11182415e+00
1.27518249e+00 -5.58999181e-01 1.90958083e-01 4.89892215e-01
-4.17040855e-01 1.52359232e-01 1.07635260e+00 -4.06203687e-01
-7.66598642e-01 -2.11421266e-01 -3.79821420e-01 -2.73166746e-01
-2.41556987e-01 2.69152939e-01 3.57491560e-02 1.32269204e-01
1.23407543e+00 3.95351470e-01 1.60156742e-01 -6.15734495e-02
5.81876278e-01 8.43555510e-01 5.57003319e-01 -5.94990849e-01
8.85012567e-01 1.21775463e-01 -3.74092013e-01 -5.36013603e-01
-8.45241606e-01 -1.66115299e-01 -2.87746340e-01 2.11806372e-01
5.31252384e-01 -5.90989053e-01 -5.10270298e-01 1.57618523e-01
-1.33324885e+00 -5.13307191e-02 -6.67143226e-01 8.48724693e-02
-7.26816654e-01 8.06298673e-01 -5.05198002e-01 -1.12500094e-01
-3.93039227e-01 -8.34840596e-01 7.83144832e-01 -3.46029639e-01
-1.31717014e+00 -9.98132229e-01 2.47771621e-01 6.04961693e-01
3.13689291e-01 -2.24984507e-03 1.54891539e+00 -1.28939772e+00
1.63807720e-01 -3.80260229e-01 -3.49846333e-02 5.60057878e-01
2.32784569e-01 -1.99925467e-01 -8.02186847e-01 -6.61867335e-02
5.24159074e-01 -6.89856231e-01 4.40556854e-01 -1.91121951e-01
7.36695051e-01 -5.95368147e-01 3.33013982e-02 2.39337862e-01
8.83965552e-01 -2.87308335e-01 7.06826091e-01 1.88796908e-01
4.59647804e-01 9.77102280e-01 1.59310520e-01 -4.81855609e-02
4.46479768e-03 6.59340322e-01 -2.14933306e-01 3.45626950e-01
-1.81248739e-01 -8.29339743e-01 5.86039901e-01 5.95763803e-01
9.09038365e-01 -7.99703971e-02 -7.22189486e-01 6.45264626e-01
-1.46080625e+00 -1.52650058e+00 -1.78607196e-01 2.13858509e+00
1.16239417e+00 2.40565047e-01 9.64307487e-02 2.56533384e-01
6.91403091e-01 6.99724481e-02 -4.04651552e-01 -1.13251019e+00
-2.70939589e-01 6.44698620e-01 -7.85920012e-04 5.02558887e-01
-5.99175513e-01 7.83366680e-01 6.62933493e+00 7.31304049e-01
-6.21425450e-01 -1.06930442e-01 4.10129905e-01 -3.77426073e-02
-9.82927799e-01 2.62435794e-01 -1.35079682e-01 6.22532606e-01
9.06756878e-01 -7.69020021e-01 4.84764546e-01 7.39102185e-01
7.56845623e-02 -5.46384826e-02 -1.71122932e+00 7.49992728e-01
2.90230125e-01 -1.27428997e+00 4.05678779e-01 -5.60389936e-01
4.33108002e-01 -2.25273147e-01 -1.16681859e-01 4.11774755e-01
4.22267079e-01 -1.26316023e+00 6.48562789e-01 4.11321297e-02
5.99261522e-01 -6.14970803e-01 6.99154317e-01 4.70839620e-01
-7.75297880e-01 -2.50450522e-02 -3.47004354e-01 -4.36316878e-01
-7.26028979e-02 1.97029158e-01 -8.94783378e-01 2.87478089e-01
1.86564296e-01 5.62475801e-01 -1.03038132e+00 6.25581801e-01
-5.08842528e-01 5.74177980e-01 -1.33794457e-01 -2.78781205e-01
-2.47451231e-01 4.20287922e-02 6.83616519e-01 1.21467674e+00
8.13371986e-02 -2.28682995e-01 -3.34859133e-01 1.23262525e+00
-1.53694740e-02 2.19737127e-01 -1.12317908e+00 -4.33396883e-02
8.24604154e-01 6.93932116e-01 -1.82111859e-01 -3.71142298e-01
-3.76273066e-01 1.34066558e+00 4.07675445e-01 -4.72162403e-02
-5.91884613e-01 -5.85586607e-01 7.01462090e-01 3.51581462e-02
-3.00163060e-01 4.12348539e-01 -6.67566001e-01 -1.37117755e+00
3.14726084e-01 -1.29172027e+00 3.19397628e-01 -1.15912008e+00
-1.84112728e+00 2.58595288e-01 -1.18055462e-03 -9.28412497e-01
-7.22361565e-01 -3.93140525e-01 -1.08731115e+00 1.30360985e+00
-6.51515543e-01 -9.97923613e-01 -1.49711952e-01 5.21413088e-01
6.50663733e-01 4.08736914e-02 1.01972115e+00 -2.49032408e-01
-8.75471830e-02 9.20043588e-01 -2.65219152e-01 2.30872542e-01
9.87458944e-01 -1.28452170e+00 1.01657212e+00 8.39051247e-01
4.64898705e-01 1.17992473e+00 9.60196376e-01 -7.15985537e-01
-9.52535093e-01 -8.47557485e-01 1.54365301e+00 -9.51963544e-01
1.03757310e+00 -4.23933357e-01 -9.56988454e-01 6.22470796e-01
1.61918908e-01 -7.70931065e-01 8.73613656e-01 2.25667864e-01
-9.01226938e-01 2.78156698e-01 -1.30252743e+00 9.69144821e-01
1.35684752e+00 -1.12551582e+00 -1.57855117e+00 6.90716386e-01
9.13515031e-01 6.46756962e-02 -5.75594306e-01 1.99319333e-01
6.21673346e-01 -1.27925014e+00 1.09176421e+00 -1.32829690e+00
9.33961153e-01 1.10339321e-01 -8.89127776e-02 -1.54271173e+00
-3.60566914e-01 -4.91232932e-01 5.12428522e-01 1.16832840e+00
4.64332551e-01 -6.27539039e-01 6.92061603e-01 9.37193453e-01
-6.13447353e-02 -5.00647843e-01 -9.65477288e-01 -9.78585124e-01
5.13866305e-01 -4.07266557e-01 5.35762370e-01 1.18673289e+00
8.13219786e-01 8.79063964e-01 1.11103423e-01 -3.40168178e-01
2.36492842e-01 3.61510366e-01 6.98027730e-01 -8.17934096e-01
-7.10425138e-01 -5.76457858e-01 -4.58356291e-01 -6.72819555e-01
6.57222688e-01 -1.50735497e+00 -2.62236476e-01 -1.24434853e+00
4.50058222e-01 -1.58212617e-01 2.17107028e-01 2.78773576e-01
-4.78159815e-01 1.64684281e-01 3.11502606e-01 3.08848560e-01
-1.35425851e-01 2.79996395e-01 6.86232805e-01 -1.21585093e-01
2.91563421e-02 7.33786598e-02 -1.08758175e+00 7.39430547e-01
8.94910872e-01 -5.47122002e-01 -6.45712554e-01 -2.82562196e-01
6.43236220e-01 -6.10018559e-02 7.03273952e-01 -7.50939608e-01
3.18721076e-03 -1.53543308e-01 4.30633336e-01 6.66206107e-02
4.73136827e-03 -4.40241992e-01 2.26013139e-01 6.51724100e-01
-9.62812364e-01 4.59755987e-01 2.67276734e-01 3.80464464e-01
-6.09831326e-03 -8.30872297e-01 7.91366160e-01 -2.59352118e-01
-2.74074435e-01 -5.72333872e-01 -5.55313051e-01 6.45368457e-01
8.71442199e-01 -5.35993338e-01 -4.28186715e-01 -4.14143860e-01
-4.29124445e-01 -1.42310947e-01 9.74944472e-01 5.31994104e-01
6.79900289e-01 -1.15474319e+00 -9.85642254e-01 3.39181513e-01
5.13365388e-01 -7.00810611e-01 7.67190456e-02 2.66656667e-01
-3.89643133e-01 1.54184297e-01 -2.34251410e-01 -1.45735219e-02
-1.44765794e+00 6.99923813e-01 4.43557948e-01 -2.01503396e-01
-2.66282201e-01 8.88000369e-01 2.44970962e-01 -5.59939146e-01
-3.74194324e-01 -1.21856906e-01 1.39279544e-01 -1.53255716e-01
4.29055035e-01 -3.29919606e-02 1.12211950e-01 -4.40848857e-01
-3.79670918e-01 1.71986461e-01 -2.43803665e-01 -2.39075556e-01
8.24410439e-01 1.48515984e-01 -1.29616380e-01 3.09244514e-01
1.38741767e+00 4.59014714e-01 -2.01466262e-01 8.14568624e-02
-8.84633735e-02 -6.09256506e-01 -6.90246582e-01 -9.24309134e-01
-3.54019292e-02 5.85775554e-01 -3.46065052e-02 3.90902609e-01
8.62557888e-01 6.01665676e-02 9.03072059e-01 4.63080138e-01
2.28961229e-01 -7.25565076e-01 2.15186626e-01 4.16383594e-01
1.18703055e+00 -8.29262257e-01 -4.50544953e-02 -4.19378281e-01
-7.10698426e-01 9.28157628e-01 4.83222514e-01 -3.24069798e-01
-6.47279248e-02 -1.50569499e-01 -2.86052108e-01 -3.19109224e-02
-7.43865609e-01 3.11125576e-01 1.16775200e-01 7.54333556e-01
5.48366368e-01 -1.58921614e-01 -4.96464014e-01 5.62331021e-01
-9.88920093e-01 -2.11513922e-01 5.44711173e-01 6.74892843e-01
-2.31236726e-01 -1.38606083e+00 -2.08879679e-01 7.41157115e-01
-1.86539307e-01 -6.05100811e-01 -1.33327138e+00 4.83464926e-01
-1.37599692e-01 1.16121602e+00 -1.62596684e-02 -5.85750163e-01
3.55442852e-01 3.46401095e-01 7.34049082e-01 -9.47328210e-01
-1.12326467e+00 -1.04646564e+00 5.18293738e-01 -4.59093422e-01
1.89011946e-01 -6.29839838e-01 -7.87741542e-01 -7.27004588e-01
-9.34968740e-02 2.56719559e-01 2.61766285e-01 1.00603998e+00
4.73091573e-01 -8.16061571e-02 6.74491167e-01 -2.74962485e-01
-1.02609789e+00 -9.38073337e-01 -1.55265599e-01 1.36778939e+00
-9.84979272e-02 6.05671890e-02 -4.92120713e-01 -6.32905886e-02] | [11.34171199798584, 9.22806453704834] |
33f1e652-8c5d-4d21-86cb-d2b846fbe9cd | language-modeling-for-formal-mathematics | 2006.04757 | null | https://arxiv.org/abs/2006.04757v3 | https://arxiv.org/pdf/2006.04757v3.pdf | Mathematical Reasoning via Self-supervised Skip-tree Training | We examine whether self-supervised language modeling applied to mathematical formulas enables logical reasoning. We suggest several logical reasoning tasks that can be used to evaluate language models trained on formal mathematical statements, such as type inference, suggesting missing assumptions and completing equalities. To train language models for formal mathematics, we propose a novel skip-tree task. We find that models trained on the skip-tree task show surprisingly strong mathematical reasoning abilities, and outperform models trained on standard skip-sequence tasks. We also analyze the models' ability to formulate new conjectures by measuring how often the predictions are provable and useful in other proofs. | ['Dennis Lee', 'Kshitij Bansal', 'Christian Szegedy', 'Markus N. Rabe'] | 2020-06-08 | null | https://openreview.net/forum?id=YmqAnY0CMEy | https://openreview.net/pdf?id=YmqAnY0CMEy | iclr-2021-1 | ['mathematical-reasoning'] | ['natural-language-processing'] | [-1.03961341e-01 6.63028657e-01 -3.35189462e-01 -5.49546123e-01
-3.47795665e-01 -5.08674562e-01 5.37944019e-01 3.75031143e-01
1.11852854e-01 9.72453535e-01 -8.80150199e-02 -1.54946959e+00
-2.99926043e-01 -1.02410734e+00 -1.12251675e+00 3.14747244e-01
-4.28448290e-01 4.57965434e-01 -2.21785121e-02 -1.91363275e-01
4.94140178e-01 1.66076198e-01 -1.08305216e+00 6.35421336e-01
1.10881138e+00 4.00420040e-01 -1.60857722e-01 9.66115713e-01
-5.87347686e-01 2.06921625e+00 -4.97850835e-01 -1.01741374e+00
-1.79416865e-01 -2.84357041e-01 -1.24611366e+00 -6.59724712e-01
8.63362193e-01 -5.15774369e-01 -1.89850941e-01 1.10524452e+00
-4.03907537e-01 -2.06021115e-01 4.43645895e-01 -1.71357608e+00
-9.40447569e-01 1.65843797e+00 1.19827308e-01 3.21668983e-01
9.22660053e-01 8.49506035e-02 1.45412743e+00 -5.47204733e-01
6.00245833e-01 1.74230170e+00 1.04968011e+00 5.96154988e-01
-1.56390011e+00 -6.45369291e-01 1.38980463e-01 6.74333751e-01
-9.36197102e-01 -4.31089967e-01 8.44247043e-01 -4.59586233e-01
1.11449456e+00 2.97652096e-01 4.65455681e-01 7.54105449e-01
5.18058121e-01 8.91531885e-01 1.42239869e+00 -6.52486622e-01
7.46834725e-02 8.63953382e-02 7.67943263e-01 1.35763526e+00
4.08841044e-01 -1.20890215e-01 -5.64251423e-01 -2.20895037e-01
4.31609124e-01 -4.05341178e-01 2.62426827e-02 -6.37709117e-03
-1.32307053e+00 8.91833603e-01 -1.01213485e-01 2.76638895e-01
6.77862242e-02 6.10320270e-01 4.60444480e-01 7.71932662e-01
1.54968560e-01 1.10758722e+00 -1.09937596e+00 -1.03810281e-01
-1.06153405e+00 7.32150674e-01 1.23505235e+00 8.85603249e-01
4.62873578e-01 -2.62755938e-02 3.30812149e-02 1.00753613e-01
3.14351678e-01 4.73877311e-01 3.25260088e-02 -1.54474807e+00
3.20410013e-01 4.66922939e-01 1.65068905e-03 -6.54448152e-01
-3.65179598e-01 -2.65681863e-01 -4.10903007e-01 8.81483871e-03
6.10765576e-01 -2.26236954e-02 -2.37975568e-01 1.90029991e+00
-2.37696737e-01 3.37234318e-01 2.49782711e-01 5.07922396e-02
9.00005817e-01 4.61030364e-01 1.99592307e-01 -3.00330281e-01
9.73477244e-01 -6.29106462e-01 -6.34896100e-01 2.27923751e-01
1.34552383e+00 -3.40874285e-01 1.05813539e+00 6.31496429e-01
-1.57276714e+00 -4.35155243e-01 -7.56848276e-01 -3.71856898e-01
-4.24839944e-01 -1.80896595e-02 1.31313622e+00 2.60206401e-01
-1.13219202e+00 7.84982741e-01 -4.53393489e-01 3.31949592e-01
3.96335036e-01 -5.84276803e-02 7.60488138e-02 -1.54630408e-01
-1.42477620e+00 1.13376713e+00 4.79008168e-01 -1.25707090e-01
-9.13829803e-01 -1.29654908e+00 -1.19702244e+00 1.87045008e-01
4.67153430e-01 -9.62999046e-01 1.63971543e+00 -5.85712075e-01
-1.26870704e+00 9.00498688e-01 -4.84534919e-01 -9.92875278e-01
4.69770551e-01 1.51561484e-01 -1.93332031e-01 2.17632726e-02
1.65285692e-01 4.18605447e-01 3.16522598e-01 -9.82805669e-01
-3.24916273e-01 -4.70108539e-02 9.45475698e-01 -6.66767716e-01
1.41083911e-01 2.04122365e-01 6.71207428e-01 -3.60976458e-01
5.40460832e-02 -3.71372253e-01 -6.39082789e-02 2.55417019e-01
-6.26246691e-01 -9.94601488e-01 2.86692023e-01 -5.86176932e-01
1.27242362e+00 -1.30724406e+00 8.17997605e-02 1.37452081e-01
6.52926803e-01 -7.52035454e-02 5.78086963e-03 4.86964174e-02
-3.34935337e-01 7.16569185e-01 3.05604517e-01 7.09805340e-02
6.98714375e-01 2.62380421e-01 -8.30772817e-01 -1.75677493e-01
3.66072774e-01 1.44385850e+00 -1.05228817e+00 -7.79239655e-01
3.73388790e-02 -4.59532619e-01 -7.63425767e-01 2.35670760e-01
-1.08554065e+00 -2.43144140e-01 -1.82567701e-01 5.09620488e-01
6.50363088e-01 -6.08796179e-01 2.83373863e-01 2.18645036e-01
1.92153640e-02 1.30852282e+00 -7.44383693e-01 1.28800178e+00
-5.23200810e-01 8.69525194e-01 -2.92939126e-01 -1.27482557e+00
5.77037632e-01 1.38140947e-01 -3.19343984e-01 -4.74888444e-01
-2.69390404e-01 5.62089449e-03 4.30018872e-01 -7.81240284e-01
2.11585574e-02 -5.44664443e-01 -2.33259331e-02 6.81154072e-01
3.62670124e-02 -6.51795566e-01 6.40705764e-01 6.03766799e-01
9.79739964e-01 3.10010254e-01 2.26618126e-01 -4.72738385e-01
7.80052662e-01 2.49714881e-01 4.15142328e-01 1.07053781e+00
1.31411090e-01 -6.27484977e-01 1.08601654e+00 -8.50884199e-01
-1.06374288e+00 -1.18720675e+00 -3.80317979e-02 1.09062207e+00
-3.98867756e-01 -1.03415191e+00 -1.98508590e-01 -6.80332065e-01
3.57535690e-01 1.36652732e+00 -4.08524841e-01 -6.64659664e-02
-3.83578092e-01 1.81694403e-01 9.43355083e-01 6.14106417e-01
3.34323585e-01 -7.17972219e-01 -2.90817171e-01 5.48976660e-03
-1.27947465e-01 -1.15434015e+00 5.81536174e-01 2.29076654e-01
-1.17747772e+00 -1.25227261e+00 6.51358664e-02 -6.66757941e-01
5.81635773e-01 -1.80596694e-01 1.49793482e+00 8.51888835e-01
1.02692343e-01 3.01616222e-01 -2.72159185e-02 -6.07930958e-01
-9.11121726e-01 -2.85511762e-01 1.69509530e-01 -1.16320813e+00
4.70461220e-01 -5.24016738e-01 5.80348611e-01 -2.79855222e-01
-4.16865855e-01 5.87921500e-01 1.90490842e-01 8.13195884e-01
-2.29732111e-01 3.92671973e-01 1.92426503e-01 -7.81461835e-01
7.34029651e-01 -2.40914389e-01 -7.42981791e-01 7.78305531e-01
-4.95462954e-01 7.15436637e-01 1.22220564e+00 -3.58688593e-01
-7.85361707e-01 -8.88409019e-01 6.43472746e-02 -2.84198791e-01
5.74923418e-02 1.06470370e+00 3.75561357e-01 -3.85106690e-02
4.72935081e-01 3.82856667e-01 1.20998966e-02 2.96145678e-02
3.18710834e-01 -9.71205756e-02 4.75908637e-01 -1.56563497e+00
1.44104934e+00 -5.93180545e-02 6.01617098e-01 -5.58213472e-01
-1.39048028e+00 4.60610062e-01 -5.36638439e-01 1.43632948e-01
2.96117872e-01 -7.40465224e-01 -1.37363493e+00 6.40808418e-02
-1.60053873e+00 -8.77155662e-01 2.54018828e-02 4.50268775e-01
-6.71671331e-01 7.47690380e-01 -7.21416831e-01 -1.25289047e+00
-2.68894508e-02 -7.13138938e-01 5.30868411e-01 -9.94871035e-02
-8.46810162e-01 -1.62228751e+00 -1.88745931e-01 3.94134015e-01
1.44128591e-01 -2.06936926e-01 1.96015096e+00 -6.48433208e-01
-6.28637493e-01 3.03891003e-01 -3.54536921e-01 2.75603116e-01
-2.86218941e-01 3.74753594e-01 -5.22325277e-01 2.86076665e-01
-2.06473038e-01 -8.66531909e-01 6.55052841e-01 1.17560834e-01
1.74057674e+00 -8.66816223e-01 -6.83988677e-03 3.84632915e-01
8.36636782e-01 -4.73309815e-01 5.75920820e-01 1.65770039e-01
1.89498201e-01 3.67850453e-01 2.39505932e-01 7.45243579e-02
8.45044553e-01 -6.59631565e-02 -1.34830147e-01 3.31921756e-01
2.06232548e-01 -8.32059443e-01 4.45305586e-01 4.19328421e-01
-2.84818672e-02 3.99591386e-01 -1.27448916e+00 3.19368362e-01
-1.50018895e+00 -1.44722378e+00 -4.20825243e-01 1.59895694e+00
1.64332652e+00 7.12171435e-01 -1.39785275e-01 3.46289665e-01
3.81682254e-02 -1.68096244e-01 -1.78515613e-01 -8.78157914e-01
-2.18055546e-01 7.36551106e-01 -3.46577913e-02 1.02586770e+00
-6.68073356e-01 9.65832591e-01 8.04949951e+00 5.50173461e-01
-7.48259306e-01 -2.68612534e-01 4.55804467e-01 3.09385121e-01
-9.28040028e-01 4.23913896e-01 -4.83615190e-01 1.39065400e-01
9.95915651e-01 -4.80422229e-01 5.46550751e-01 8.68948042e-01
2.30047718e-01 -1.59696788e-01 -2.00456166e+00 5.27776122e-01
4.24795002e-02 -1.81205869e+00 4.48718458e-01 -5.29248893e-01
4.73428369e-01 -6.35872006e-01 -7.20831230e-02 7.99249113e-01
7.84312606e-01 -1.44993341e+00 7.06096768e-01 6.65092826e-01
4.03554589e-01 -1.73745662e-01 5.96936285e-01 7.08387077e-01
-6.19539857e-01 -1.21342815e-01 -3.29543263e-01 -1.15411341e+00
-3.45649779e-01 4.50838953e-01 -1.04346502e+00 3.87842745e-01
1.49721622e-01 8.06036353e-01 -8.37278903e-01 5.59692740e-01
-9.24387753e-01 8.04982543e-01 -1.92072928e-01 -6.23953760e-01
-8.41506869e-02 -1.59876011e-02 2.91572422e-01 1.13673401e+00
-2.58504897e-01 1.67380691e-01 3.29448208e-02 1.87427378e+00
-1.50509970e-02 -5.24354160e-01 -6.31280601e-01 -4.07729179e-01
3.23699087e-01 6.05611026e-01 2.49429289e-02 -9.00343657e-01
-3.70288223e-01 2.46298909e-01 5.43190539e-01 2.74690330e-01
-9.43214834e-01 -1.54457584e-01 3.66157293e-01 2.63925083e-02
-1.69377849e-01 -7.64908195e-01 -7.28570998e-01 -1.57706082e+00
3.38346995e-02 -9.92198586e-01 3.18435907e-01 -1.54554284e+00
-1.38643861e+00 -3.86715710e-01 4.19952124e-01 -3.76469821e-01
-4.56497610e-01 -1.30503452e+00 -1.06454396e+00 6.90693259e-01
-1.75696743e+00 -1.13896811e+00 1.96273446e-01 3.45223635e-01
2.82983840e-01 -4.24212553e-02 9.73505318e-01 -4.45427746e-01
-2.95784444e-01 6.99809492e-01 -6.72935307e-01 5.15132308e-01
2.03903452e-01 -1.41629744e+00 5.45768291e-02 8.57162356e-01
1.80266127e-01 1.45002794e+00 1.04208207e+00 -3.99217427e-01
-1.85425019e+00 -6.97414100e-01 1.58580053e+00 -6.06298327e-01
1.36916804e+00 -2.33929202e-01 -7.93266833e-01 1.23012960e+00
1.34959975e-02 -2.81692803e-01 5.21513283e-01 6.71783209e-01
-1.12918329e+00 1.92251295e-01 -8.79012525e-01 8.13781738e-01
1.06056094e+00 -1.05492175e+00 -1.69706357e+00 6.39217079e-01
8.52343380e-01 -3.08623523e-01 -8.90381098e-01 3.03520441e-01
5.66389501e-01 -6.01988375e-01 8.67152154e-01 -1.81645250e+00
1.34398007e+00 -1.18956611e-01 -8.25347379e-03 -9.00047660e-01
-3.68677318e-01 -7.84831166e-01 -6.30617917e-01 7.33653665e-01
7.33552337e-01 -5.66185594e-01 6.06017590e-01 8.88482392e-01
2.30320990e-01 -7.25124717e-01 -7.59793460e-01 -8.47370982e-01
9.08281088e-01 -8.88450503e-01 3.71708989e-01 1.20932353e+00
9.55877125e-01 4.88575906e-01 3.05970103e-01 2.02303529e-02
8.12750399e-01 5.77478826e-01 1.00329506e+00 -1.42687821e+00
-3.59306455e-01 -7.97537744e-01 -3.15373152e-01 -1.14767063e+00
1.30366886e+00 -1.44750488e+00 -5.18666863e-01 -1.54102850e+00
4.78614926e-01 -3.93900007e-01 1.49784148e-01 8.65138650e-01
-1.01134792e-01 -4.46815789e-01 -1.44947115e-02 -3.77468020e-01
-8.62172663e-01 1.82225004e-01 1.06954134e+00 -4.86995071e-01
5.32445252e-01 -2.69122958e-01 -7.09779084e-01 1.05673015e+00
5.97678423e-01 6.75611049e-02 -9.59870145e-02 -4.31299537e-01
1.08328223e+00 1.66654706e-01 9.25064325e-01 -7.17971683e-01
2.96537638e-01 -8.14392090e-01 2.91103363e-01 -4.93622690e-01
-3.09439749e-01 -2.98314095e-01 -4.49790567e-01 7.31350005e-01
-9.71814334e-01 1.66278183e-01 3.91627312e-01 -1.18031420e-01
7.27327242e-02 -4.28550333e-01 3.59642535e-01 -3.28330547e-01
-6.06302083e-01 -2.36835480e-01 -4.30018812e-01 3.30106109e-01
6.17003918e-01 1.69731200e-01 -4.66724396e-01 -5.04813254e-01
-6.25170827e-01 6.45960867e-01 2.67364204e-01 -3.56447063e-02
8.40635717e-01 -9.45737183e-01 -8.26795280e-01 -1.37269601e-01
-2.59955041e-02 -7.79884756e-02 -2.83478588e-01 1.08231711e+00
-5.53777993e-01 7.43442297e-01 2.75773089e-02 -3.18833888e-01
-1.31708670e+00 7.41649032e-01 5.19022167e-01 -5.88301897e-01
-2.63583124e-01 8.22907209e-01 -2.26612240e-01 -7.52468407e-01
2.36014754e-01 -1.35594928e+00 2.94472605e-01 -6.78901792e-01
4.81011540e-01 1.45064607e-01 -4.54094678e-01 3.62749040e-01
-3.27977777e-01 2.76631713e-01 1.02157995e-01 2.66379595e-01
1.34655941e+00 3.18052053e-01 -9.47522402e-01 8.10375333e-01
6.98333263e-01 9.81814563e-02 -2.53925413e-01 -4.19840395e-01
3.82727951e-01 1.46644944e-02 -2.26273715e-01 -1.06902194e+00
-1.37514714e-02 1.06877732e+00 -7.01382518e-01 2.53485352e-01
4.61155057e-01 4.45783064e-02 3.96258146e-01 1.26489782e+00
5.74292779e-01 -7.83228457e-01 -4.99700420e-02 1.12607086e+00
7.06940353e-01 -1.19093037e+00 3.74535888e-01 -4.51658189e-01
2.21171901e-02 1.80893302e+00 5.12591302e-01 -3.67421657e-02
4.44099426e-01 6.79646254e-01 -4.03678715e-01 -1.98736981e-01
-1.38016903e+00 2.35893354e-01 -6.44237362e-03 3.38484824e-01
8.48836243e-01 2.22434878e-01 -1.18587650e-01 8.54661882e-01
-8.87674928e-01 5.81768394e-01 9.09474254e-01 6.90128803e-01
-2.78706998e-01 -8.34720910e-01 -3.83604109e-01 5.00010729e-01
-4.57616061e-01 -5.25290191e-01 -5.50421536e-01 7.44134486e-01
-5.11428602e-02 8.21815431e-01 -1.79920360e-01 -1.42347053e-01
-4.75396276e-01 6.38079226e-01 1.22932363e+00 -6.92801833e-01
-4.88804467e-02 -9.72695708e-01 5.75811327e-01 -2.57702231e-01
-4.29917574e-01 -5.45495272e-01 -1.39839339e+00 -1.01670074e+00
-1.91708609e-01 4.89493191e-01 1.23873971e-01 1.33939803e+00
-1.60172015e-01 5.55624068e-01 2.78315634e-01 1.48047442e-02
-1.16706324e+00 -9.91579771e-01 -1.43088251e-01 2.16459438e-01
4.25060004e-01 -3.28225851e-01 -5.72762907e-01 2.12776065e-01] | [9.151759147644043, 7.1607232093811035] |
e86c5497-010c-4782-9eea-640917b77029 | prediction-of-brain-tumor-recurrence-location | 2304.13725 | null | https://arxiv.org/abs/2304.13725v1 | https://arxiv.org/pdf/2304.13725v1.pdf | Prediction of brain tumor recurrence location based on multi-modal fusion and nonlinear correlation learning | Brain tumor is one of the leading causes of cancer death. The high-grade brain tumors are easier to recurrent even after standard treatment. Therefore, developing a method to predict brain tumor recurrence location plays an important role in the treatment planning and it can potentially prolong patient's survival time. There is still little work to deal with this issue. In this paper, we present a deep learning-based brain tumor recurrence location prediction network. Since the dataset is usually small, we propose to use transfer learning to improve the prediction. We first train a multi-modal brain tumor segmentation network on the public dataset BraTS 2021. Then, the pre-trained encoder is transferred to our private dataset for extracting the rich semantic features. Following that, a multi-scale multi-channel feature fusion model and a nonlinear correlation learning module are developed to learn the effective features. The correlation between multi-channel features is modeled by a nonlinear equation. To measure the similarity between the distributions of original features of one modality and the estimated correlated features of another modality, we propose to use Kullback-Leibler divergence. Based on this divergence, a correlation loss function is designed to maximize the similarity between the two feature distributions. Finally, two decoders are constructed to jointly segment the present brain tumor and predict its future tumor recurrence location. To the best of our knowledge, this is the first work that can segment the present tumor and at the same time predict future tumor recurrence location, making the treatment planning more efficient and precise. The experimental results demonstrated the effectiveness of our proposed method to predict the brain tumor recurrence location from the limited dataset. | ['Su Ruan', 'Maxime Fontanilles', 'Sébastien Thureau', 'Fethi Ghazouani', 'Romain Modzelewski', 'Alexandra Noeuveglise', 'Tongxue Zhou'] | 2023-04-11 | null | null | null | null | ['tumor-segmentation', 'brain-tumor-segmentation'] | ['computer-vision', 'medical'] | [ 1.77368056e-02 2.05370169e-02 -3.00768942e-01 -5.11219680e-01
-1.12428832e+00 1.55323476e-01 2.35535920e-01 -2.37386040e-02
-4.76429552e-01 6.12628222e-01 1.56284794e-01 -1.47518829e-01
-1.13030612e-01 -7.37749040e-01 -3.81630152e-01 -9.82350349e-01
1.69099793e-01 3.08715671e-01 2.82840371e-01 1.75574437e-01
1.68787286e-01 2.63095140e-01 -8.80713582e-01 1.76211238e-01
1.02298439e+00 1.29705203e+00 8.20118785e-01 2.12983072e-01
-1.96852788e-01 6.65373027e-01 -3.40356916e-01 1.73108533e-01
-5.39797097e-02 -5.67973912e-01 -7.73622274e-01 -3.45320553e-02
-2.93347865e-01 -2.04281837e-01 -4.50419277e-01 1.20046151e+00
5.41921794e-01 -1.94665387e-01 6.38778746e-01 -1.09281635e+00
-2.50941008e-01 5.38025737e-01 -6.09893441e-01 1.38577133e-01
-1.42537877e-02 -9.38258544e-02 6.46208704e-01 -6.37099504e-01
3.36760908e-01 6.84566319e-01 4.91401225e-01 5.59936225e-01
-6.69679344e-01 -8.93492877e-01 1.82547383e-02 4.09973770e-01
-1.38218820e+00 -2.13056743e-01 7.23852932e-01 -4.20672774e-01
3.91684860e-01 -1.71386395e-02 8.18859935e-01 9.65724766e-01
7.33910739e-01 8.92657876e-01 1.06155634e+00 -1.54471815e-01
5.85854724e-02 1.50011227e-01 -7.09916428e-02 9.36339378e-01
-2.56647408e-01 7.42354766e-02 -2.57444799e-01 1.09833032e-01
5.37190914e-01 4.53778148e-01 -5.99648178e-01 -1.26549080e-01
-1.23527861e+00 8.75427246e-01 9.06070948e-01 6.52991712e-01
-1.93946049e-01 6.09342605e-02 3.79618406e-01 -7.84446597e-02
4.20539081e-01 -1.79926306e-01 -4.23834920e-01 -8.40221904e-03
-8.31115544e-01 -2.38926843e-01 5.73231757e-01 6.53410554e-01
5.88558972e-01 -5.54545701e-01 -2.01595128e-01 8.23907316e-01
6.02688968e-01 4.87899125e-01 1.17632794e+00 -1.75552458e-01
3.88337731e-01 4.66432780e-01 -4.06048506e-01 -6.88465536e-01
-8.53987396e-01 -6.21403098e-01 -1.05080795e+00 -9.34082121e-02
2.00089023e-01 -2.86103368e-01 -1.02842045e+00 1.65213966e+00
3.02418768e-02 5.37167549e-01 1.19060159e-01 9.59735692e-01
8.46084476e-01 5.11622071e-01 -1.72377527e-02 -2.69384235e-01
1.25435603e+00 -1.00997686e+00 -6.29245937e-01 -9.60250795e-02
1.05267406e+00 -6.04077816e-01 6.09589875e-01 -1.31805509e-01
-5.75996399e-01 -1.56677350e-01 -7.61218965e-01 1.71183199e-01
-1.35246471e-01 3.36960584e-01 6.91467881e-01 4.75284547e-01
-7.72841215e-01 1.71144679e-01 -1.09588218e+00 -3.39883149e-01
6.58412278e-01 4.49824959e-01 -2.84886748e-01 -1.22974858e-01
-1.28909218e+00 1.02251422e+00 5.29742181e-01 1.25811845e-01
-8.67596686e-01 -6.52349055e-01 -5.50002873e-01 -4.38682130e-03
2.38667682e-01 -5.81604600e-01 1.16463828e+00 -1.08994555e+00
-1.54120064e+00 5.68890393e-01 -2.43529558e-01 -3.54086548e-01
3.19578886e-01 3.86976749e-01 -5.73859692e-01 -1.05137311e-01
3.17662746e-01 4.75447327e-01 5.74407279e-01 -7.75443792e-01
-1.00121701e+00 -5.96650302e-01 -5.91583729e-01 4.18738127e-01
-3.82769287e-01 -3.33303213e-01 -5.39896846e-01 -3.64186227e-01
2.28882641e-01 -8.96637619e-01 -2.51224518e-01 -1.88436627e-01
-4.08278704e-01 -1.10596314e-01 8.75783741e-01 -8.21344435e-01
9.50849295e-01 -2.21639276e+00 2.42248580e-01 4.43476111e-01
3.10335249e-01 -6.45790771e-02 -3.15146968e-02 -2.67157763e-01
-3.96542214e-02 -5.22817597e-02 -3.51980358e-01 -1.43995449e-01
-3.56543988e-01 -3.71702388e-03 2.17730209e-01 5.87552607e-01
-1.73318863e-01 9.64070320e-01 -6.59525633e-01 -4.42333668e-01
1.26627356e-01 3.91234398e-01 -2.56314129e-01 2.27318212e-01
4.81070392e-02 7.68155396e-01 -7.66713023e-01 4.94151145e-01
5.38829029e-01 -1.49446651e-01 -3.11899751e-01 -2.72978514e-01
-9.24208015e-03 -1.63941041e-01 -5.39066553e-01 1.99161291e+00
-6.13631368e-01 4.80897516e-01 -1.64874375e-01 -1.20477903e+00
9.85393107e-01 3.20346594e-01 9.00038779e-01 -1.00729489e+00
4.08935934e-01 4.79926229e-01 2.13369712e-01 -6.73666596e-01
-2.05868706e-01 -3.63832206e-01 4.01516519e-02 6.67757168e-02
-1.92924410e-01 -1.37112439e-01 -3.34993571e-01 -1.17154293e-01
1.09609950e+00 -4.06177998e-01 1.26073226e-01 -4.66741920e-02
7.53822923e-01 -1.72576338e-01 8.30372274e-01 2.92377919e-01
-2.37107724e-01 5.05107224e-01 4.07249808e-01 -1.40573412e-01
-4.37169373e-01 -8.70833576e-01 -4.74253565e-01 3.85937005e-01
3.76349300e-01 2.90690437e-02 -6.10724270e-01 -8.73310268e-01
-1.80172697e-01 5.78889370e-01 -5.27119398e-01 -5.36418736e-01
-2.11456701e-01 -8.99761200e-01 3.08874339e-01 4.36970860e-01
8.56032431e-01 -7.88722694e-01 -3.28606218e-01 2.14995727e-01
-3.98662001e-01 -8.37895751e-01 -6.00264907e-01 2.31068566e-01
-9.02995229e-01 -1.06473231e+00 -1.06642652e+00 -1.03829908e+00
7.40759194e-01 -1.90798752e-02 2.53559351e-01 -1.45091889e-02
-2.86030740e-01 7.86185712e-02 -3.36153686e-01 -3.72486860e-01
-2.00067014e-01 3.09295952e-01 -3.40134799e-01 5.87226935e-02
4.57119495e-01 -2.43499607e-01 -6.09683037e-01 1.91977918e-01
-6.64813638e-01 3.81264687e-01 9.45379376e-01 1.06588244e+00
6.32235289e-01 2.98402786e-01 6.60106122e-01 -5.41655362e-01
5.80057263e-01 -6.05037987e-01 -3.96334678e-01 2.23426744e-01
-4.53254104e-01 1.40997756e-03 6.30643606e-01 -2.23751530e-01
-9.91091430e-01 2.68019170e-01 -2.60723203e-01 -1.35839567e-01
-1.08842596e-01 9.93912816e-01 -2.05744877e-01 -1.08783804e-01
1.17164388e-01 3.31914127e-01 4.74118032e-02 -1.68666571e-01
-5.10076899e-03 9.61624920e-01 2.71157652e-01 -7.85917714e-02
3.51492912e-01 2.45829910e-01 2.10275427e-02 -5.67534447e-01
-7.62615442e-01 -5.20283818e-01 -3.66361409e-01 -1.55223325e-01
1.01747179e+00 -8.41077864e-01 -6.15435421e-01 7.11200237e-01
-9.10422325e-01 -1.99596956e-01 1.09741092e-01 1.04547262e+00
-5.65837920e-01 2.25560755e-01 -4.39630598e-01 -3.21169198e-01
-3.56819689e-01 -1.60674858e+00 1.01194191e+00 4.85946357e-01
3.51473957e-01 -9.85697150e-01 -3.49321812e-02 2.87586838e-01
4.06507552e-01 -7.23387152e-02 8.36885035e-01 -6.74971044e-01
-5.42893469e-01 -3.87235790e-01 -5.03607273e-01 1.33904964e-02
3.21822375e-01 -3.24219823e-01 -7.21637905e-01 -2.84030646e-01
1.84717342e-01 -6.76816702e-02 1.04933357e+00 6.35684550e-01
1.38137603e+00 2.66522884e-01 -8.56577396e-01 8.42413664e-01
1.28459334e+00 5.60586512e-01 5.98231733e-01 3.39877635e-01
7.82638252e-01 1.30501837e-01 4.56302315e-01 4.54603076e-01
6.29631102e-01 5.56248724e-01 5.16587377e-01 2.52610166e-03
-8.59848112e-02 -1.43285453e-01 2.60463893e-01 9.10009801e-01
1.85714975e-01 -1.41035363e-01 -1.06268585e+00 3.78797114e-01
-1.75313532e+00 -8.50720882e-01 1.07441675e-02 2.18364763e+00
6.85867250e-01 -9.81268138e-02 -4.20327902e-01 -7.49154538e-02
7.97888398e-01 -1.42530233e-01 -6.47222519e-01 1.18877918e-01
1.60286099e-01 -1.67580545e-01 5.98645866e-01 3.55672181e-01
-1.07237482e+00 6.89859331e-01 5.12465286e+00 1.07334590e+00
-1.61570716e+00 3.26149523e-01 7.78905928e-01 6.83013722e-02
-1.62334248e-01 7.11408257e-03 -6.38954520e-01 7.46426880e-01
7.68091321e-01 -2.25363106e-01 2.27876335e-01 5.71353734e-01
3.00297320e-01 -3.20959151e-01 -9.24426556e-01 1.24054527e+00
1.60629079e-01 -1.08222866e+00 -4.27226067e-01 1.74708933e-01
5.84020495e-01 2.69128531e-01 3.63414325e-02 2.63544619e-01
-2.40758389e-01 -1.07542467e+00 3.43042910e-01 1.04354417e+00
5.92604220e-01 -8.71598661e-01 1.06952703e+00 6.52392030e-01
-1.14674795e+00 -1.91609710e-01 -2.19799414e-01 4.70606655e-01
1.67332888e-01 8.28380406e-01 -7.91636288e-01 4.84684914e-01
3.54427218e-01 1.03864491e+00 -4.86446828e-01 1.32582760e+00
-6.51672632e-02 4.08919573e-01 -3.14187735e-01 -5.20466119e-02
7.33248815e-02 -8.12798440e-02 3.17949802e-01 7.44568050e-01
8.35764587e-01 1.21598216e-02 3.78568292e-01 7.26018369e-01
2.82465350e-02 2.98377484e-01 -2.59624183e-01 2.07919687e-01
2.34105587e-01 1.21242857e+00 -7.67613113e-01 -6.60376847e-02
-6.22293115e-01 1.12137079e+00 2.82439470e-01 1.23645768e-01
-1.02500045e+00 -3.16010863e-01 8.57958943e-02 -1.94252566e-01
7.79559687e-02 1.87279191e-02 -4.45356160e-01 -1.36711872e+00
-1.50364667e-01 -2.83803791e-01 2.36366585e-01 -5.16803861e-01
-1.21206105e+00 5.93041718e-01 -2.99543172e-01 -1.20778167e+00
1.46638870e-03 -3.73385578e-01 -8.75474095e-01 1.05880010e+00
-1.58465672e+00 -1.21541727e+00 -3.88691813e-01 7.93129861e-01
3.84125262e-01 -4.10649449e-01 9.45145667e-01 2.30840415e-01
-8.20623696e-01 5.46379566e-01 2.15716168e-01 1.70933828e-01
6.67786002e-01 -9.29761648e-01 -5.89088678e-01 4.23584938e-01
-4.07897204e-01 1.82471666e-02 1.54092699e-01 -6.65114164e-01
-1.16328561e+00 -1.08775091e+00 5.32824516e-01 2.77293354e-01
6.23869777e-01 5.61853535e-02 -7.97073245e-01 5.37284374e-01
-4.09949049e-02 2.62705028e-01 6.97014570e-01 -2.31892899e-01
2.83827037e-01 -2.91740626e-01 -1.16381061e+00 3.14028591e-01
6.37932301e-01 -5.40644526e-01 -3.09372872e-01 4.75323081e-01
4.23891574e-01 -4.90865618e-01 -1.02749586e+00 6.27603829e-01
4.44055617e-01 -6.73938692e-01 6.04305208e-01 8.62055421e-02
4.07617152e-01 -2.03864932e-01 -5.23241051e-02 -1.74565268e+00
-2.22292736e-01 1.70147941e-01 3.77437383e-01 1.06768143e+00
5.77314138e-01 -7.87818849e-01 8.24551821e-01 7.40248561e-01
-2.81060219e-01 -9.87698019e-01 -1.00094521e+00 -4.43468302e-01
2.23202631e-01 -3.70372057e-01 5.12629330e-01 6.52578056e-01
2.06118345e-01 1.05857626e-01 -3.25757288e-03 3.00552934e-01
4.17747617e-01 1.28896043e-01 1.93170950e-01 -1.05393481e+00
2.17649974e-02 -5.97290099e-01 -6.18046641e-01 -9.52238977e-01
3.94369721e-01 -1.30833960e+00 1.54462695e-01 -1.61719823e+00
5.07310867e-01 -6.18904829e-01 -6.10340297e-01 4.27487403e-01
-5.49207889e-02 -3.25643480e-01 -1.44436657e-01 1.99884757e-01
-2.74906397e-01 9.30462062e-01 1.55972469e+00 -3.39044809e-01
-2.09867701e-01 2.44456694e-01 -5.04604638e-01 6.89435840e-01
9.04296279e-01 -4.83448416e-01 -4.11718130e-01 -3.60671818e-01
-4.15092915e-01 6.10567331e-01 2.55962372e-01 -1.05794716e+00
6.57684326e-01 -2.43375778e-01 5.14883816e-01 -6.13251746e-01
1.47399858e-01 -1.12030053e+00 -1.40229955e-01 6.00102544e-01
-2.71859765e-01 -4.57052946e-01 -7.71036185e-03 4.82982814e-01
-4.60466504e-01 -2.34140277e-01 8.28107417e-01 1.70697764e-01
-7.55201221e-01 8.61623287e-01 -5.13344109e-01 -2.47471005e-01
1.42167306e+00 -8.39253142e-03 -9.51923206e-02 -1.78516835e-01
-7.06207991e-01 3.70431572e-01 4.56605479e-02 4.40978944e-01
9.05734897e-01 -1.43774331e+00 -6.11761808e-01 3.37418914e-01
1.45190805e-01 9.57780033e-02 3.63483787e-01 1.33387101e+00
-2.80690581e-01 3.14892262e-01 -1.31904632e-01 -5.16939044e-01
-1.04418910e+00 2.43729651e-01 7.08119571e-01 -2.94824362e-01
-6.78129852e-01 8.24344993e-01 2.20023990e-01 -2.92203575e-01
1.78901941e-01 -4.10388678e-01 -4.10042912e-01 -1.29404113e-01
5.63826621e-01 -2.63469089e-02 -6.58286214e-02 -6.71303868e-01
-4.06832159e-01 6.75271749e-01 -3.28037709e-01 -8.07855725e-02
1.25096405e+00 -2.47618407e-01 -2.86217988e-01 1.75096378e-01
1.59648860e+00 -3.10412288e-01 -1.06357360e+00 -2.86162794e-01
-1.68762639e-01 -2.51875877e-01 6.35314584e-01 -8.70069385e-01
-1.65159547e+00 8.20401192e-01 1.09577394e+00 -2.14711055e-01
1.42719042e+00 1.72543734e-01 1.13200903e+00 8.99867043e-02
2.87374079e-01 -8.43626440e-01 -1.60008073e-01 6.16853952e-01
6.59933150e-01 -1.47256708e+00 -3.97645682e-01 -4.17433172e-01
-6.08277500e-01 1.32523584e+00 6.80883169e-01 4.54944931e-02
1.03388476e+00 3.07492554e-01 8.68875980e-02 -1.41292229e-01
-4.13170218e-01 -2.53231853e-01 1.29527271e-01 4.19339806e-01
3.64795595e-01 3.70458186e-01 -3.39390725e-01 1.02995253e+00
-8.92024934e-02 3.92321974e-01 2.58734912e-01 5.80611706e-01
-6.02098882e-01 -1.11931980e+00 -2.05676243e-01 8.98402750e-01
-2.70021647e-01 -1.50717404e-02 3.27442996e-02 3.00199538e-01
2.46272087e-01 8.15314412e-01 -7.76577368e-02 -4.96552736e-01
1.16067268e-01 -7.00029060e-02 3.13537568e-01 -4.74336892e-01
-1.01990774e-01 1.37132853e-01 -4.03858006e-01 -3.57047588e-01
-3.96493256e-01 -5.62374413e-01 -1.74222183e+00 1.60434291e-01
-6.39562726e-01 -1.45643987e-02 9.22761798e-01 1.27583003e+00
1.06196575e-01 7.82611787e-01 8.30581605e-01 -6.03384376e-01
-3.98834169e-01 -8.25485587e-01 -7.46260405e-01 1.48817301e-01
1.63386822e-01 -6.56452954e-01 -1.54523373e-01 -3.28148425e-01] | [14.51521110534668, -2.3234496116638184] |
89e57b27-4396-4fee-9d58-015dc5afb390 | hyperntf-a-hypergraph-regularized-nonnegative | 2101.06827 | null | https://arxiv.org/abs/2101.06827v3 | https://arxiv.org/pdf/2101.06827v3.pdf | HyperNTF: A Hypergraph Regularized Nonnegative Tensor Factorization for Dimensionality Reduction | Tensor decomposition is an effective tool for learning multi-way structures and heterogeneous features from high-dimensional data, such as the multi-view images and multichannel electroencephalography (EEG) signals, are often represented by tensors. However, most of tensor decomposition methods are the linear feature extraction techniques, which are unable to reveal the nonlinear structure within high-dimensional data. To address such problem, a lot of algorithms have been proposed for simultaneously performs linear and non-linear feature extraction. A representative algorithm is the Graph Regularized Non-negative Matrix Factorization (GNMF) for image clustering. However, the normal 2-order graph can only models the pairwise similarity of objects, which cannot sufficiently exploit the complex structures of samples. Thus, we propose a novel method, named Hypergraph Regularized Non-negative Tensor Factorization (HyperNTF), which utilizes hypergraph to encode the complex connections among samples and employs the factor matrix corresponding with last mode of Canonical Polyadic (CP) decomposition as low-dimensional representation. Extensive experiments on synthetic manifolds, real-world image datasets, and EEG signals, demonstrating that HyperNTF outperforms the state-of-the-art methods in terms of dimensionality reduction, clustering, and classification. | ['Zhengming Ma', 'Youzhi Qu', 'Quanying Liu', 'Wanguang Yin'] | 2021-01-18 | null | null | null | null | ['image-clustering'] | ['computer-vision'] | [-3.93070936e-01 -6.20883465e-01 5.98612949e-02 -1.13998845e-01
-3.91273648e-02 -4.61814016e-01 2.97577471e-01 -3.87283653e-01
-8.42323005e-02 3.22686851e-01 3.30935359e-01 -2.21988722e-03
-6.40994072e-01 -4.44676965e-01 -1.53278619e-01 -1.02897823e+00
-3.79564762e-01 7.13107884e-02 -1.99076504e-01 -1.82583764e-01
2.69436538e-01 3.75486553e-01 -1.19678771e+00 3.05480272e-01
7.17245817e-01 8.11120927e-01 6.52350252e-03 2.30692983e-01
-2.11076468e-01 2.28652507e-01 -1.26148194e-01 -1.94425911e-01
3.80172469e-02 -1.82045609e-01 -5.51505625e-01 3.52490366e-01
-1.13284374e-02 1.51716694e-02 -7.29748130e-01 1.23815048e+00
2.93476254e-01 -5.69957718e-02 6.17489100e-01 -1.49467194e+00
-7.43590593e-01 4.44380671e-01 -8.46984029e-01 4.31840777e-01
2.55319566e-01 -2.60152280e-01 8.79372954e-01 -1.16161311e+00
3.75467956e-01 1.46472716e+00 1.96234941e-01 2.78896809e-01
-1.28778791e+00 -8.13255370e-01 2.12385297e-01 5.31448841e-01
-1.39403379e+00 -3.60326506e-02 1.02067268e+00 -8.39706004e-01
5.96020937e-01 4.24162209e-01 9.86942828e-01 8.84059787e-01
3.28923732e-01 7.44262695e-01 1.25511599e+00 5.24271652e-02
-3.20186689e-02 -2.33202904e-01 5.20889878e-01 6.23172820e-01
3.24410439e-01 -2.99473643e-01 -5.52178502e-01 -5.95454156e-01
7.29186654e-01 4.95271355e-01 -6.80064857e-01 -6.22488678e-01
-1.92355978e+00 6.07187808e-01 3.49225700e-01 6.77894950e-01
-5.20081639e-01 -3.25507134e-01 6.95421159e-01 2.96265841e-01
4.28957134e-01 3.56429875e-01 -3.25974703e-01 -2.14740232e-01
-2.20778763e-01 -1.03807688e-01 4.53128815e-01 8.06539834e-01
8.66150379e-01 1.51820704e-01 1.64711624e-01 6.49115801e-01
3.18232864e-01 4.17247057e-01 9.35631275e-01 -5.14823794e-01
6.37311876e-01 1.26514912e+00 -3.49771440e-01 -1.71011698e+00
-7.54091322e-01 -3.21169168e-01 -1.55305481e+00 -5.88781595e-01
-2.23928213e-01 -2.40513459e-02 -6.02064490e-01 1.36081719e+00
3.24377924e-01 3.49964857e-01 9.70324036e-03 1.00394928e+00
7.24164486e-01 4.71913248e-01 -5.05061805e-01 -5.33166409e-01
1.45533454e+00 -4.47320968e-01 -9.41572011e-01 3.87857497e-01
6.50909245e-01 -5.08624077e-01 7.71340489e-01 5.40992320e-01
-6.48346305e-01 -2.52049357e-01 -9.34421897e-01 2.60164022e-01
-2.52758950e-01 1.98688298e-01 1.05873549e+00 5.08092582e-01
-6.26935124e-01 4.01882291e-01 -1.02016282e+00 -4.06626351e-02
2.46916354e-01 3.65239471e-01 -8.94769669e-01 -4.51242089e-01
-9.34188485e-01 2.22348899e-01 2.25421563e-01 4.21584278e-01
-5.06518424e-01 -4.96474057e-01 -6.60290837e-01 -8.70305002e-02
1.67308182e-01 -5.19895792e-01 1.29921198e-01 -5.95499694e-01
-1.18946528e+00 1.89966530e-01 -1.89133331e-01 2.97509402e-01
-9.40537676e-02 2.17576534e-01 -6.31755412e-01 3.67195427e-01
6.49982989e-02 -7.21516600e-03 1.14239073e+00 -1.12083828e+00
3.81933004e-02 -9.88146007e-01 -7.66842514e-02 2.36968651e-01
-9.05439913e-01 -7.85699561e-02 -1.30795449e-01 -6.20197594e-01
9.95674133e-01 -1.06082475e+00 -1.86244130e-01 -5.06093919e-01
-5.45468748e-01 -2.69491047e-01 1.06675899e+00 -6.50359869e-01
1.23334134e+00 -2.32477784e+00 9.28193390e-01 4.39781189e-01
7.85786211e-01 2.52980180e-02 -1.73677832e-01 3.60970289e-01
-4.72312808e-01 1.43967167e-01 -3.48622836e-02 -1.54852405e-01
-3.13290983e-01 2.30983824e-01 -1.37623563e-01 7.00580895e-01
-3.79235186e-02 6.11165464e-01 -9.76425111e-01 -2.72800595e-01
6.50454611e-02 5.41292489e-01 -3.47395003e-01 8.88145640e-02
5.45841098e-01 6.97110951e-01 -6.86732769e-01 4.30021822e-01
8.28779519e-01 -3.50455403e-01 8.60650614e-02 -8.90254319e-01
1.40381521e-02 -2.22705022e-01 -1.33143306e+00 1.88686013e+00
1.32010505e-01 1.89162448e-01 1.12885006e-01 -1.24820673e+00
6.61012173e-01 4.51436847e-01 9.77924407e-01 -3.41444373e-01
1.59260660e-01 4.32149842e-02 3.60248297e-01 -8.54721665e-01
2.88214594e-01 1.26865759e-01 2.57839382e-01 5.27666509e-01
1.65015712e-01 4.69557613e-01 2.85132349e-01 5.48390687e-01
1.03736234e+00 -3.58199149e-01 -1.31451279e-01 -4.43986267e-01
7.01551437e-01 -5.68815708e-01 7.50202715e-01 -2.26992190e-01
1.30859718e-01 3.77005786e-01 7.17149913e-01 -5.36970139e-01
-7.23185956e-01 -7.59995401e-01 -2.80936271e-01 5.04102468e-01
1.45950168e-01 -9.32997286e-01 -6.63616419e-01 -5.56425989e-01
-6.29678965e-02 -1.55908376e-01 -6.70553088e-01 -2.86865801e-01
-3.15520525e-01 -1.20279348e+00 7.86939263e-02 1.78403780e-01
3.62314045e-01 -3.58981133e-01 1.75874680e-02 -7.12916330e-02
-3.41544181e-01 -1.28009677e+00 -5.26161015e-01 -2.54844874e-01
-1.09853017e+00 -1.27727902e+00 -6.22839093e-01 -4.87876385e-01
1.03628528e+00 8.53440821e-01 4.77981091e-01 7.61879832e-02
-3.15844774e-01 6.33933306e-01 -5.14836133e-01 2.30511174e-01
2.60781437e-01 -1.63844258e-01 7.50047326e-01 7.71350205e-01
3.48536968e-01 -9.91977572e-01 -6.33296013e-01 4.95666683e-01
-1.06505799e+00 1.36834025e-01 6.67778254e-01 1.03442085e+00
6.00093603e-01 2.53248066e-01 4.97085065e-01 -4.95225430e-01
8.62194359e-01 -5.73817074e-01 -1.49266556e-01 3.26599121e-01
-5.23757458e-01 2.11996466e-01 8.00849736e-01 -5.88376403e-01
-3.61071885e-01 -1.77853286e-01 5.32719731e-01 -9.66064274e-01
1.63791254e-01 7.22606003e-01 -3.32527786e-01 -4.93409604e-01
3.52697849e-01 5.89852870e-01 6.65240213e-02 -5.65793812e-01
3.18767637e-01 6.83136344e-01 -7.94193372e-02 -5.68386674e-01
7.65433967e-01 6.72970533e-01 4.34531063e-01 -1.02198422e+00
-3.57890964e-01 -5.86311162e-01 -1.01171279e+00 -7.31400028e-02
8.19266975e-01 -8.52043688e-01 -9.73363280e-01 4.45922613e-01
-1.07245278e+00 6.05559349e-01 2.62848109e-01 1.05984890e+00
-1.99436069e-01 8.58911216e-01 -4.79976624e-01 -5.61533689e-01
-2.49009147e-01 -1.10646057e+00 7.84064174e-01 -2.38766015e-01
4.56441790e-01 -7.72321761e-01 8.01229384e-03 3.39880705e-01
6.72632009e-02 2.96197206e-01 1.23522604e+00 -1.83638558e-01
-5.33767581e-01 -1.83886185e-01 -2.47888133e-01 4.46030170e-01
3.00866783e-01 -1.72195017e-01 -5.86385012e-01 -5.41096032e-01
2.21647099e-01 -1.07324116e-01 5.90007603e-01 7.54983798e-02
1.20827889e+00 -2.41134301e-01 -4.54437375e-01 7.62593210e-01
1.09052157e+00 -1.98924858e-02 3.38508159e-01 4.66315337e-02
1.43085349e+00 5.23229957e-01 1.51942208e-01 5.95171928e-01
6.35247827e-01 3.96994442e-01 3.27780753e-01 1.50579184e-01
5.40975988e-01 8.30238760e-02 1.21899255e-01 1.90171421e+00
-4.81722027e-01 4.73165959e-01 -8.08577001e-01 2.87220836e-01
-1.98911679e+00 -7.57113695e-01 -4.56088811e-01 2.03872776e+00
1.64085820e-01 -2.56403565e-01 1.66368529e-01 3.88554841e-01
7.17757881e-01 -1.21124117e-02 -5.83752751e-01 1.61974460e-01
-2.26274669e-01 -5.38854539e-01 1.83012873e-01 -1.14948586e-01
-7.61231363e-01 4.10328805e-01 5.31607866e+00 5.00544369e-01
-1.06551456e+00 1.01755530e-01 6.65246472e-02 1.65118337e-01
-5.48980832e-01 -3.96404490e-02 -2.41843820e-01 4.84435439e-01
5.67081630e-01 -3.17235112e-01 9.21973348e-01 4.78803366e-01
1.42233446e-01 4.03845429e-01 -9.87769425e-01 1.72779608e+00
3.57638419e-01 -1.04915142e+00 5.30181766e-01 3.54993403e-01
5.87513447e-01 -5.36921211e-02 6.17899969e-02 1.26539826e-01
-3.48193169e-01 -6.86758757e-01 9.80299860e-02 7.16141343e-01
4.74201977e-01 -7.70509422e-01 4.48227197e-01 3.64542902e-01
-1.42094696e+00 -1.67750791e-01 -7.07494497e-01 -2.04792097e-02
-4.08360474e-02 7.78010309e-01 -2.63882428e-01 1.05746329e+00
7.87145376e-01 1.35686541e+00 -7.06480503e-01 9.79898393e-01
2.57411182e-01 3.84330302e-01 -1.82755496e-02 1.29389346e-01
2.54897714e-01 -8.54251087e-01 7.78932154e-01 6.09646201e-01
3.17380071e-01 5.56788981e-01 2.71231234e-01 6.47779882e-01
8.41506794e-02 3.52489889e-01 -6.04688942e-01 -3.52337033e-01
2.65014078e-03 1.67228758e+00 -6.95689321e-01 -8.38330761e-02
-5.97148597e-01 8.38187099e-01 2.80330241e-01 5.59001684e-01
-3.31925124e-01 -3.18202615e-01 7.92113543e-01 -1.73774794e-01
-3.27049941e-02 -7.75251329e-01 -8.87332112e-02 -1.93868184e+00
3.63757849e-01 -8.89424145e-01 2.85022676e-01 -5.48700869e-01
-1.49560404e+00 8.18024576e-01 -4.11279798e-02 -1.57216442e+00
1.19394027e-01 -6.56294763e-01 -3.13397408e-01 7.09358931e-01
-1.21239960e+00 -1.15956926e+00 -4.18355107e-01 1.33090699e+00
4.23720013e-03 -3.52023214e-01 9.42171276e-01 4.67294395e-01
-8.21182489e-01 3.12355399e-01 4.34679180e-01 2.61184186e-01
2.40104720e-01 -9.91393089e-01 -2.51274966e-02 6.17178798e-01
9.96478349e-02 9.30454314e-01 1.42394081e-01 -3.98294032e-01
-2.36548638e+00 -8.30001831e-01 9.21577960e-02 -1.59089044e-01
1.03496683e+00 -5.35281897e-01 -9.93456364e-01 5.68670034e-01
-1.00512393e-01 3.52644205e-01 1.01428854e+00 2.58093774e-01
-5.01976073e-01 -1.74281046e-01 -8.71517003e-01 4.62920278e-01
1.07320738e+00 -5.48285365e-01 -3.46236885e-01 6.65857077e-01
7.40618765e-01 -2.60103364e-02 -1.44448924e+00 3.52455020e-01
4.16842997e-01 -7.49939442e-01 8.07896078e-01 -8.80588949e-01
8.84721521e-03 -4.93683726e-01 -3.38777214e-01 -1.59227693e+00
-7.52736211e-01 -5.95455289e-01 -3.02107602e-01 1.06583786e+00
-2.36920249e-02 -8.43654335e-01 5.78665853e-01 5.26086211e-01
-6.52968884e-02 -8.62567484e-01 -1.04588354e+00 -7.35319674e-01
-3.00871283e-01 -3.12634379e-01 6.94675148e-01 1.49587369e+00
4.78993684e-01 6.32395685e-01 -4.00559396e-01 2.60453731e-01
8.44483554e-01 2.26475403e-01 5.95752299e-01 -1.39803362e+00
6.80798367e-02 -1.87582001e-01 -1.30897224e+00 -6.87807739e-01
3.31111491e-01 -1.07819629e+00 -7.49449432e-01 -1.33535743e+00
4.20970261e-01 -3.27530563e-01 -3.58171403e-01 1.55552089e-01
1.45541029e-02 -1.25709951e-01 6.95802420e-02 5.83010972e-01
-5.36460876e-01 9.73212659e-01 1.58794463e+00 -1.84985518e-01
-1.89942956e-01 -2.89112031e-01 -4.69110131e-01 5.51101267e-01
3.62533659e-01 -1.07054099e-01 -4.93477583e-01 -5.06710649e-01
3.96258026e-01 1.10401213e-01 3.86207998e-02 -8.40714991e-01
3.14758450e-01 -1.27457067e-01 3.40070963e-01 -4.48186785e-01
4.63768870e-01 -9.57282841e-01 2.01734677e-01 1.70531392e-01
1.90987840e-01 5.90775549e-01 -1.57277346e-01 8.28951895e-01
-4.30893451e-01 4.15170759e-01 2.89289683e-01 -1.40783921e-01
-4.28111136e-01 8.86059523e-01 -1.15338191e-01 -9.56522003e-02
8.74325812e-01 -7.95874521e-02 -4.70937967e-01 -4.55763191e-02
-7.73612261e-01 3.56790006e-01 1.08873434e-01 5.93884110e-01
1.11250329e+00 -1.83087039e+00 -5.93296885e-01 5.61471641e-01
2.54839420e-01 -1.17768213e-01 6.12770438e-01 1.38941121e+00
4.47744085e-03 4.24652398e-01 -3.77417266e-01 -9.10367072e-01
-1.15636981e+00 7.84722447e-01 1.06020741e-01 -5.19555388e-03
-7.76169538e-01 3.36679190e-01 3.74327809e-01 -3.93491894e-01
-1.38143629e-01 -2.37773642e-01 -6.12360895e-01 2.41511956e-01
6.32146001e-01 5.65463901e-01 1.01846457e-01 -1.14303756e+00
-4.01899517e-01 9.33619082e-01 -1.60385653e-01 1.32810012e-01
1.47567844e+00 -2.61031955e-01 -9.62166131e-01 7.71974385e-01
1.59170282e+00 -4.66060340e-01 -6.90066576e-01 -3.30934107e-01
-2.19383374e-01 -7.99093366e-01 2.15547264e-01 9.39756110e-02
-1.33557606e+00 1.10058331e+00 5.16095817e-01 4.93998885e-01
1.08784294e+00 -3.19987178e-01 6.14673376e-01 7.05933511e-01
7.79329777e-01 -6.56375766e-01 2.53448129e-01 3.47094715e-01
9.74876702e-01 -8.55008304e-01 5.22225685e-02 -4.74757820e-01
-6.26923919e-01 1.38088977e+00 5.55675626e-01 -2.37937957e-01
1.18462276e+00 -2.81054288e-01 -3.29340130e-01 -5.45769751e-01
-5.78400910e-01 2.00436532e-01 5.49416780e-01 4.14394915e-01
1.55436382e-01 3.72644126e-01 -3.08103889e-01 5.81085086e-01
-1.33051679e-01 -3.84157687e-01 5.55869758e-01 6.04371011e-01
1.31565961e-03 -8.65808368e-01 -5.49190998e-01 8.60367775e-01
-2.91276723e-01 4.39788215e-03 -2.92570978e-01 2.97118038e-01
-5.54025657e-02 8.74291539e-01 -2.23243877e-01 -1.06203032e+00
2.42079109e-01 -2.05060378e-01 4.13235247e-01 -4.37723994e-01
-1.38708934e-01 7.07180128e-02 -4.37243581e-01 -8.06789696e-01
-7.31272221e-01 -7.44535744e-01 -1.08089817e+00 -1.08331338e-01
-4.31992233e-01 4.60547864e-01 8.35377336e-01 9.04633284e-01
6.42226815e-01 3.85169476e-01 1.05079913e+00 -9.84057069e-01
-2.66410649e-01 -9.64958131e-01 -1.16546559e+00 6.22642457e-01
3.67724508e-01 -1.03248346e+00 -6.95817769e-01 -2.40079433e-01] | [8.132966041564941, 4.575297832489014] |
104b7d15-695a-402b-8a64-ba0fafeb8b07 | reference-based-image-composition-with-sketch | 2304.09748 | null | https://arxiv.org/abs/2304.09748v1 | https://arxiv.org/pdf/2304.09748v1.pdf | Reference-based Image Composition with Sketch via Structure-aware Diffusion Model | Recent remarkable improvements in large-scale text-to-image generative models have shown promising results in generating high-fidelity images. To further enhance editability and enable fine-grained generation, we introduce a multi-input-conditioned image composition model that incorporates a sketch as a novel modal, alongside a reference image. Thanks to the edge-level controllability using sketches, our method enables a user to edit or complete an image sub-part with a desired structure (i.e., sketch) and content (i.e., reference image). Our framework fine-tunes a pre-trained diffusion model to complete missing regions using the reference image while maintaining sketch guidance. Albeit simple, this leads to wide opportunities to fulfill user needs for obtaining the in-demand images. Through extensive experiments, we demonstrate that our proposed method offers unique use cases for image manipulation, enabling user-driven modifications of arbitrary scenes. | ['Jaegul Choo', 'Junsoo Lee', 'Sunghyun Park', 'Kangyeol Kim'] | 2023-03-31 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 7.79644132e-01 2.18112320e-01 -5.66130038e-04 -4.68511954e-02
-4.82327759e-01 -7.61424839e-01 9.11805451e-01 -4.15549457e-01
1.53229222e-01 6.42357588e-01 1.62294969e-01 -8.15018415e-02
8.15967992e-02 -1.04764807e+00 -8.42922449e-01 -6.77917421e-01
4.82637703e-01 2.17203006e-01 -3.54987755e-02 -3.14458370e-01
3.62816989e-01 8.85235786e-01 -1.67558098e+00 3.26842964e-01
1.00331640e+00 6.25167787e-01 7.48105884e-01 8.70908856e-01
-1.45802379e-01 4.38700616e-01 -4.43999320e-01 -5.12232661e-01
3.34361345e-01 -4.64900017e-01 -3.68692577e-01 6.79438412e-01
6.82519615e-01 -4.99361992e-01 -2.65666485e-01 7.61863530e-01
4.10442322e-01 9.12687406e-02 7.49122679e-01 -9.86260712e-01
-1.12497520e+00 3.21356416e-01 -5.35721183e-01 -6.64159775e-01
4.67847943e-01 5.33299983e-01 8.05035830e-01 -9.84800696e-01
1.07104266e+00 1.08317769e+00 1.08395196e-01 8.10559630e-01
-1.68781126e+00 -4.85957146e-01 1.67772636e-01 -4.56642359e-01
-1.31659853e+00 -6.94576681e-01 1.00488424e+00 -3.56688797e-01
5.78742802e-01 4.33708102e-01 7.99897373e-01 1.04106021e+00
1.73651159e-01 6.64965928e-01 1.03080809e+00 -4.12080109e-01
1.12610288e-01 2.68113881e-01 -8.15840542e-01 8.59054327e-01
-1.23240434e-01 4.62655984e-02 -4.75779772e-01 3.97194503e-03
1.46700180e+00 1.48731887e-01 -4.22370911e-01 -6.77778959e-01
-1.36983895e+00 5.27226806e-01 3.65290344e-01 8.79361331e-02
-4.84014571e-01 3.13906640e-01 -6.99656904e-02 1.40907466e-01
1.96161285e-01 5.78024805e-01 -1.64155215e-02 -6.73065707e-02
-1.23628569e+00 3.34464788e-01 4.85122770e-01 1.47110355e+00
8.25901449e-01 3.61533105e-01 -5.76314747e-01 8.47503662e-01
4.46436694e-03 7.47271597e-01 6.93089589e-02 -1.13166428e+00
2.50197113e-01 3.90774816e-01 4.27750289e-01 -8.36533666e-01
2.75475085e-01 -4.25523698e-01 -1.03989434e+00 5.27522922e-01
-5.09333946e-02 2.51542151e-01 -1.11075926e+00 1.78051269e+00
2.72593737e-01 -1.47602350e-01 -2.74204105e-01 7.21480727e-01
4.50056881e-01 6.67235255e-01 -1.01893749e-02 -1.38338000e-01
1.13322067e+00 -1.14090514e+00 -5.12550294e-01 7.60307908e-02
-1.09735802e-02 -8.90504062e-01 1.43398619e+00 4.00944501e-01
-1.48563099e+00 -7.95680881e-01 -9.79604363e-01 -7.44557455e-02
-7.70446360e-02 3.26145917e-01 3.43593568e-01 5.97717464e-01
-1.13038850e+00 5.73020339e-01 -3.71066630e-01 -2.19722435e-01
4.05540109e-01 1.78582340e-01 -3.63701671e-01 -1.59068182e-01
-6.26088381e-01 5.52350819e-01 1.68284163e-01 -1.34881109e-01
-1.20046341e+00 -8.39317620e-01 -7.29115844e-01 8.86972025e-02
2.84313679e-01 -1.27165878e+00 8.44494522e-01 -8.15977275e-01
-1.94206917e+00 8.28549445e-01 -2.18383640e-01 1.22706585e-01
7.57263482e-01 3.27958614e-02 -1.02199994e-01 2.32778594e-01
-2.36300658e-02 1.17274809e+00 1.55662727e+00 -1.84883976e+00
-5.15188932e-01 1.75009400e-01 2.38556236e-01 1.38210118e-01
-4.48891431e-01 -2.67306656e-01 -6.72828794e-01 -1.05666351e+00
-2.62843668e-01 -9.08500791e-01 -1.66415200e-01 3.90636027e-01
-6.32145584e-01 2.68571913e-01 9.68964458e-01 -3.44689876e-01
1.32474196e+00 -2.05877852e+00 5.13202906e-01 3.64099771e-01
3.61225367e-01 1.46316990e-01 -4.82724190e-01 7.02652395e-01
7.99230710e-02 3.52857620e-01 -3.75074655e-01 -6.61523104e-01
4.77456860e-02 2.20613226e-01 -4.03327465e-01 -3.94057250e-03
5.17876327e-01 1.13001943e+00 -8.58260870e-01 -4.96102631e-01
4.87902135e-01 6.03714705e-01 -6.97118580e-01 4.60892946e-01
-3.88800561e-01 7.31068850e-01 -4.39173490e-01 5.87415457e-01
6.57989323e-01 -2.44189382e-01 -9.31918025e-02 -4.20907587e-01
-1.31553888e-01 -3.33372712e-01 -1.08905327e+00 2.05298138e+00
-8.79480958e-01 2.88266540e-01 4.17124301e-01 -3.46418381e-01
1.00819230e+00 4.55019586e-02 2.09624574e-01 -4.95692134e-01
-2.03432113e-01 -2.87002772e-02 -4.47746903e-01 -2.28063077e-01
9.76857066e-01 -8.75276178e-02 -1.54121751e-02 7.26671696e-01
-1.50796250e-01 -8.47967207e-01 2.83082277e-01 5.35094082e-01
6.46989763e-01 7.08220541e-01 -6.06273394e-03 -2.88425595e-01
4.62918460e-01 -2.36661315e-01 2.51023914e-03 8.31177831e-01
3.93120736e-01 1.01876843e+00 7.47885555e-02 -6.62552267e-02
-1.63305414e+00 -1.26579845e+00 1.43577799e-01 7.23036170e-01
1.06659852e-01 -3.14473480e-01 -8.05391252e-01 -2.94968098e-01
-1.52120739e-01 7.69912481e-01 -6.28678024e-01 -4.70472649e-02
-5.55434525e-01 -2.63390001e-02 3.24737579e-01 3.09650540e-01
4.71207291e-01 -1.04501569e+00 -5.09546041e-01 2.31419429e-01
-9.35927182e-02 -8.06831717e-01 -1.11162758e+00 -3.21671963e-01
-7.68990934e-01 -4.52229798e-01 -1.19237876e+00 -9.05962586e-01
1.13503325e+00 4.34036106e-01 1.08772576e+00 1.69854581e-01
-4.60759789e-01 5.82475901e-01 -2.37399023e-02 2.60246307e-01
-8.09364140e-01 -3.75931337e-02 -1.94750860e-01 2.64549106e-01
-7.88414121e-01 -7.94383168e-01 -8.15256715e-01 3.44910532e-01
-1.53819692e+00 7.39982486e-01 5.30514300e-01 1.02595711e+00
7.65144527e-01 -9.91968662e-02 4.25153077e-01 -6.71363592e-01
8.41010213e-01 7.02703372e-02 -5.46151996e-01 4.63388771e-01
-6.11212969e-01 2.50763744e-01 7.76072681e-01 -6.58560157e-01
-1.44023871e+00 1.70954645e-01 1.24725260e-01 -4.41683233e-01
-1.54366940e-01 1.66009992e-01 -2.32254639e-01 -2.26575777e-01
4.68171954e-01 5.06361842e-01 -6.69189170e-03 -2.05400810e-01
1.10509956e+00 4.63981837e-01 6.24808848e-01 -9.35127437e-01
1.13937104e+00 5.32224417e-01 -4.46505733e-02 -6.37994707e-01
-2.00096741e-01 1.39804736e-01 -7.96946168e-01 -3.24603885e-01
7.69561291e-01 -8.09425235e-01 -5.24155736e-01 3.97442162e-01
-1.10322285e+00 -4.58118945e-01 -5.09927154e-01 -2.29105413e-01
-7.93438435e-01 2.94928640e-01 -5.42647481e-01 -6.69243336e-01
-3.99845630e-01 -1.34422553e+00 1.48055661e+00 3.77553612e-01
-2.34242097e-01 -8.10165823e-01 -2.69122988e-01 2.38133177e-01
7.24942684e-01 3.77345651e-01 9.49951768e-01 4.54828352e-01
-1.22968554e+00 -6.55752867e-02 -3.29535782e-01 1.54890805e-01
4.14895087e-01 4.36959743e-01 -6.82426512e-01 -3.67534101e-01
-3.89516443e-01 -2.88656384e-01 6.24734640e-01 -1.97409075e-02
1.19557095e+00 -2.34040409e-01 -2.98586547e-01 6.01921916e-01
1.44303954e+00 8.73627663e-02 9.82968032e-01 1.07273255e-02
9.05803263e-01 2.88559616e-01 4.25505072e-01 7.25307047e-01
2.32556313e-01 6.99000895e-01 1.98654100e-01 -2.63507575e-01
-5.47045767e-01 -7.46483326e-01 1.80250734e-01 7.08842576e-01
-2.26441920e-01 -4.58613724e-01 -2.38038868e-01 5.42730808e-01
-1.50919020e+00 -1.04226363e+00 1.66256636e-01 2.17078447e+00
1.02761316e+00 -1.34504929e-01 -1.42539218e-01 -3.63756448e-01
7.51373768e-01 2.09522262e-01 -6.52716160e-01 -2.99857229e-01
-5.73237613e-02 4.27577376e-01 1.43678352e-01 4.45542186e-01
-6.22933805e-01 1.07107663e+00 6.35900640e+00 1.23904300e+00
-1.13054037e+00 -1.67980373e-01 5.56002140e-01 -1.91444471e-01
-1.08598185e+00 -5.26869856e-03 -5.00248432e-01 2.24940658e-01
3.44795525e-01 -1.67022482e-01 8.42535436e-01 4.83590692e-01
5.03252923e-01 -5.04896007e-02 -9.24802184e-01 9.22859251e-01
1.17528453e-01 -1.74779844e+00 6.23640418e-01 1.77889049e-01
1.17926514e+00 -7.70723224e-01 3.05954427e-01 -7.18026981e-02
2.55534768e-01 -8.69733214e-01 1.01878202e+00 8.09965074e-01
1.57680345e+00 -7.19277442e-01 -2.31447354e-01 3.04924548e-01
-1.15299582e+00 2.90141255e-01 -5.56042269e-02 2.80690372e-01
5.82322776e-01 4.41702545e-01 -5.51945984e-01 5.64315081e-01
1.63960174e-01 4.37550545e-01 -3.52609247e-01 5.83878815e-01
-2.74203509e-01 1.13199569e-01 -1.73175156e-01 1.83744967e-01
4.76842672e-02 -4.11851943e-01 5.55878758e-01 1.08622026e+00
4.72389013e-01 2.46542972e-02 8.08009729e-02 1.57212627e+00
-2.39214599e-01 -5.26733100e-02 -7.62631238e-01 -2.31742367e-01
3.60373110e-01 1.51685643e+00 -6.09996319e-01 -5.17057478e-01
-3.23573016e-02 1.63009012e+00 2.19255909e-01 4.62964565e-01
-9.33937192e-01 -5.21013856e-01 5.13653457e-01 2.12809458e-01
5.03309727e-01 -5.41479826e-01 -3.19884688e-01 -1.24248767e+00
7.65088387e-03 -8.70292783e-01 -5.09600818e-01 -1.25886750e+00
-9.63128865e-01 5.87968230e-01 -1.67929038e-01 -1.15189838e+00
-1.34675518e-01 -2.28727251e-01 -5.95142126e-01 1.08059013e+00
-1.31109798e+00 -1.77015698e+00 -5.53365767e-01 6.51636481e-01
5.37651360e-01 -1.29440084e-01 8.56970549e-01 2.41538122e-01
-2.62581080e-01 6.88325584e-01 7.44586810e-03 -2.37355784e-01
6.69307470e-01 -1.01021600e+00 6.01970732e-01 7.66899645e-01
6.97522387e-02 8.13336074e-01 4.72418815e-01 -9.08575356e-01
-1.66834617e+00 -1.10057139e+00 4.03451532e-01 -3.32864970e-01
3.91983032e-01 -4.89353925e-01 -4.97223347e-01 2.96030879e-01
4.77129310e-01 -3.87814611e-01 2.57655025e-01 -5.92696190e-01
-1.84712216e-01 -3.01987380e-02 -1.27060664e+00 1.13353026e+00
1.37516403e+00 -7.22673535e-01 -8.50908831e-02 4.03496018e-03
8.52536380e-01 -4.39856172e-01 -9.67904747e-01 1.97270066e-01
7.48339236e-01 -8.56594801e-01 9.97931659e-01 -1.93464875e-01
8.59032929e-01 -6.53852642e-01 3.05436049e-02 -1.31464970e+00
-6.14399135e-01 -1.12854278e+00 1.11242525e-01 1.38637984e+00
3.34161520e-01 -3.13483417e-01 6.86626256e-01 8.53888988e-01
-1.89357772e-01 -5.71095169e-01 -4.02565986e-01 -5.29772699e-01
-2.43962735e-01 -2.23946229e-01 8.39526176e-01 6.01945817e-01
-3.54803473e-01 8.58346149e-02 -8.87961268e-01 -5.41471690e-02
6.33370936e-01 4.95280743e-01 1.04343545e+00 -6.28928483e-01
-4.68870282e-01 -5.39764166e-01 7.51635730e-02 -1.50207841e+00
-2.11508229e-01 -7.08617687e-01 3.70756071e-03 -1.53026021e+00
2.70925462e-01 -6.31650627e-01 2.17884719e-01 2.72051454e-01
-9.22769159e-02 5.27817607e-01 5.00379741e-01 2.07072720e-01
-2.43602812e-01 7.82000124e-01 2.02340555e+00 -2.81924278e-01
-1.77249804e-01 -2.60252476e-01 -7.48362184e-01 1.82129562e-01
6.03200078e-01 -1.62877142e-01 -5.29747307e-01 -4.35441613e-01
6.36802986e-02 2.16189966e-01 2.82931149e-01 -6.95799530e-01
7.43905315e-03 -4.60588962e-01 3.80780637e-01 -4.29419696e-01
4.33667183e-01 -7.84559309e-01 6.86083138e-01 2.40923509e-01
-5.20292282e-01 -4.22081165e-02 2.75544152e-02 5.96319079e-01
1.19914422e-02 -2.53093958e-01 7.38624215e-01 -2.79415160e-01
-4.31823730e-01 4.85513300e-01 -2.24336594e-01 -3.69483620e-01
1.06254721e+00 -4.66338336e-01 -2.07033455e-01 -5.81349432e-01
-6.98594093e-01 -7.95009211e-02 1.09401584e+00 5.71245670e-01
8.68503988e-01 -1.55731332e+00 -6.41993880e-01 5.14868319e-01
2.58876115e-01 -8.51388723e-02 3.89044076e-01 3.14199090e-01
-5.23094952e-01 1.20327935e-01 -2.52252519e-01 -4.77354407e-01
-1.13352990e+00 6.20256364e-01 4.75952215e-02 -1.82757184e-01
-5.44658780e-01 5.38136840e-01 4.90089446e-01 -2.39261672e-01
-1.35384053e-01 -2.76665092e-01 4.57707077e-01 -2.76935607e-01
4.56082523e-01 -2.44954340e-02 -2.01634377e-01 -3.55242848e-01
2.10228026e-01 6.36279106e-01 -3.10024470e-02 -4.12660658e-01
1.15289807e+00 -3.56390029e-01 -5.84230274e-02 -7.84161836e-02
9.15459156e-01 3.59570086e-01 -1.76271093e+00 -7.09135085e-02
-7.29408622e-01 -7.60481238e-01 -6.39280900e-02 -8.81674349e-01
-1.15499270e+00 6.90314651e-01 2.72325158e-01 -4.95506339e-02
1.12725008e+00 -1.48889005e-01 8.21354508e-01 5.30795753e-02
6.21205688e-01 -9.25845563e-01 3.83448690e-01 1.24144837e-01
1.34479284e+00 -9.14747536e-01 -7.80465454e-02 -3.63258660e-01
-8.06458116e-01 1.13090813e+00 3.86496872e-01 -6.54983148e-02
2.71088958e-01 3.15818787e-01 -2.30387971e-01 -4.08787653e-02
-6.13402367e-01 -3.62658612e-02 3.92193705e-01 8.52698743e-01
2.48971239e-01 1.81013182e-01 4.46983278e-02 -8.40711966e-02
-4.31707464e-02 1.35534331e-01 5.33910155e-01 7.41346240e-01
-2.35272765e-01 -1.41159141e+00 -3.10199827e-01 3.15409005e-01
7.05318451e-02 -3.17750782e-01 -1.10863224e-01 4.63790774e-01
6.52309731e-02 7.23360777e-01 -9.69844833e-02 -2.67310053e-01
3.15307587e-01 -1.64625525e-01 7.82005787e-01 -5.49036920e-01
-4.70297903e-01 1.45179823e-01 -3.75280925e-03 -4.89937305e-01
-3.38245928e-01 -3.53428304e-01 -9.00997221e-01 -3.98253620e-01
-2.97141016e-01 -2.44115323e-01 6.17336035e-01 5.75451255e-01
8.19773316e-01 5.32339036e-01 6.95000470e-01 -1.23736739e+00
-2.02864662e-01 -6.46199763e-01 -5.35859525e-01 6.76924288e-01
1.48516178e-01 -4.36507881e-01 4.07882109e-02 6.42185569e-01] | [11.450334548950195, -0.4219026565551758] |
144b010f-cc12-4bd9-a98b-01c9b3204419 | incorporating-global-visual-features-into | 1701.06521 | null | http://arxiv.org/abs/1701.06521v1 | http://arxiv.org/pdf/1701.06521v1.pdf | Incorporating Global Visual Features into Attention-Based Neural Machine Translation | We introduce multi-modal, attention-based neural machine translation (NMT)
models which incorporate visual features into different parts of both the
encoder and the decoder. We utilise global image features extracted using a
pre-trained convolutional neural network and incorporate them (i) as words in
the source sentence, (ii) to initialise the encoder hidden state, and (iii) as
additional data to initialise the decoder hidden state. In our experiments, we
evaluate how these different strategies to incorporate global image features
compare and which ones perform best. We also study the impact that adding
synthetic multi-modal, multilingual data brings and find that the additional
data have a positive impact on multi-modal models. We report new
state-of-the-art results and our best models also significantly improve on a
comparable phrase-based Statistical MT (PBSMT) model trained on the Multi30k
data set according to all metrics evaluated. To the best of our knowledge, it
is the first time a purely neural model significantly improves over a PBSMT
model on all metrics evaluated on this data set. | ['Qun Liu', 'Nick Campbell', 'Iacer Calixto'] | 2017-01-23 | null | null | null | null | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 3.97702068e-01 2.88820595e-01 -4.37969677e-02 -1.37543947e-01
-1.17603433e+00 -3.83034557e-01 1.25368845e+00 -8.60237554e-02
-9.54467237e-01 8.54527593e-01 3.97129625e-01 -4.58468914e-01
4.83977795e-01 -4.94231761e-01 -1.34678280e+00 -5.67425191e-01
3.44828010e-01 8.09077144e-01 2.15039745e-01 -2.24764749e-01
1.47772670e-01 4.71426994e-02 -1.09376836e+00 7.30885267e-01
6.31510019e-01 5.80633461e-01 3.29502255e-01 6.52180076e-01
1.10310204e-01 6.86554909e-01 -4.43373591e-01 -6.25449121e-01
2.15107813e-01 -4.42571700e-01 -9.62566376e-01 -6.81401789e-02
6.93795204e-01 -2.42532656e-01 -2.86230415e-01 7.31990993e-01
6.77860737e-01 -3.56926441e-01 8.40574205e-01 -8.90593290e-01
-9.96438622e-01 5.78461409e-01 -4.84799623e-01 2.85123676e-01
2.85825312e-01 5.37665665e-01 8.47996950e-01 -1.06986105e+00
1.14205933e+00 1.29219675e+00 5.13564169e-01 4.20105875e-01
-1.47166634e+00 -2.46423766e-01 -2.13214964e-01 4.16868538e-01
-9.90559757e-01 -8.29060614e-01 4.84086871e-01 -3.39973003e-01
1.44847596e+00 -5.71212582e-02 3.41743439e-01 1.42781091e+00
6.31202161e-01 8.04851234e-01 1.49791086e+00 -9.76198912e-01
-9.57920626e-02 2.28028700e-01 -2.70168066e-01 8.45962942e-01
-2.33421922e-01 3.25922370e-01 -5.91776609e-01 2.46844620e-01
4.11920488e-01 -5.90785205e-01 -1.07514016e-01 -2.87610322e-01
-1.69350505e+00 8.87357891e-01 4.83764917e-01 5.34988761e-01
-4.75131422e-01 3.76996189e-01 3.95974457e-01 5.09217918e-01
5.08247912e-01 2.81111896e-01 -5.18113732e-01 -1.14381231e-01
-1.21208429e+00 -2.19661817e-02 6.29287422e-01 6.30616546e-01
8.94584298e-01 -1.00258909e-01 -6.37814879e-01 7.94758558e-01
2.07981586e-01 6.18379295e-01 6.29977942e-01 -8.73291492e-01
8.80300879e-01 3.15450668e-01 -1.24073491e-01 -5.30538082e-01
-1.97780013e-01 -3.35234821e-01 -4.59985107e-01 2.04291224e-01
3.54735047e-01 -1.43827483e-01 -1.27014828e+00 1.75297010e+00
-1.56742945e-01 -1.57552749e-01 2.62443364e-01 8.35506856e-01
8.69831800e-01 9.64335263e-01 1.08938433e-01 -5.08409366e-02
1.24172044e+00 -1.34812796e+00 -5.52981675e-01 -3.63872439e-01
6.55286252e-01 -1.08931649e+00 1.00008297e+00 5.44175804e-02
-1.24842215e+00 -6.28357351e-01 -1.05787385e+00 -1.58117726e-01
-5.95927596e-01 4.26753551e-01 7.51566887e-02 2.96180338e-01
-1.59716058e+00 5.50634205e-01 -8.06491435e-01 -6.08394146e-01
2.28398293e-01 4.40535963e-01 -5.38279414e-01 -2.56091386e-01
-1.30460751e+00 1.48040211e+00 3.00999820e-01 -1.39621437e-01
-9.89478111e-01 -2.04428226e-01 -9.60074365e-01 -2.72484571e-01
1.21176608e-01 -9.61452007e-01 1.35489976e+00 -1.40397751e+00
-1.50366604e+00 1.09187531e+00 -4.55974281e-01 -6.14660025e-01
5.13103962e-01 1.01232186e-01 -1.19885258e-01 2.20703557e-01
2.88930416e-01 1.40873039e+00 8.91174257e-01 -1.30040455e+00
-2.92597532e-01 2.68397555e-02 2.61780489e-02 2.54528195e-01
-6.41701296e-02 2.22660393e-01 -6.16261005e-01 -5.46140611e-01
-4.07120228e-01 -1.23576844e+00 -7.75286704e-02 -2.71688372e-01
-4.33954954e-01 5.15325628e-02 6.69933915e-01 -1.02679062e+00
7.55629659e-01 -1.79869354e+00 6.58027709e-01 -2.86416203e-01
-1.84621230e-01 2.17691198e-01 -5.47874153e-01 6.29459858e-01
-1.70429304e-01 1.37264237e-01 -3.01828057e-01 -7.04978883e-01
3.65480185e-02 3.88146549e-01 3.44481803e-02 2.27991030e-01
6.11990869e-01 1.50286770e+00 -4.81131494e-01 -6.27861261e-01
2.86203325e-01 5.88683307e-01 -2.45278001e-01 1.77735984e-02
-2.24485412e-01 5.05533218e-01 1.01355284e-01 1.50577769e-01
3.21223348e-01 -1.82672635e-01 -1.52409866e-01 -1.98269054e-01
-3.06955755e-01 3.82010162e-01 -4.29831445e-01 1.94504845e+00
-6.23360991e-01 1.07550192e+00 -3.06155771e-01 -9.50986564e-01
5.08020043e-01 5.62735081e-01 -1.71163641e-02 -1.15092933e+00
9.21011195e-02 3.95460665e-01 2.10780144e-01 -3.63953859e-01
5.64281642e-01 -2.51909137e-01 2.47263405e-02 5.05242109e-01
5.60240090e-01 2.84889698e-01 3.85678858e-01 2.25940183e-01
1.04749739e+00 3.94533366e-01 7.58115575e-02 -2.61698246e-01
5.19015849e-01 9.97534767e-02 3.32911382e-03 5.27185559e-01
8.08637962e-03 8.47832799e-01 5.11072695e-01 -3.42517167e-01
-1.45021462e+00 -8.37265074e-01 1.03587635e-01 9.33620572e-01
-4.18623149e-01 -1.31349161e-01 -8.76060188e-01 -7.58405209e-01
-4.75256383e-01 8.83575916e-01 -7.77384043e-01 -1.61072791e-01
-6.30827308e-01 -8.39629054e-01 6.28123522e-01 4.49604928e-01
4.12477255e-01 -1.35959768e+00 -5.04108727e-01 2.71648675e-01
-4.68046844e-01 -1.27025366e+00 -4.02829230e-01 3.97018284e-01
-7.26992488e-01 -4.65070397e-01 -1.07891095e+00 -8.94415319e-01
5.32804787e-01 -2.15487167e-01 1.29695773e+00 -1.63414747e-01
2.71563698e-02 2.60814935e-01 -3.48358124e-01 -2.44922221e-01
-8.60173285e-01 3.94758314e-01 -2.20733717e-01 -5.98621182e-02
2.34889582e-01 -2.11925104e-01 -2.59528399e-01 6.28412422e-03
-7.61088729e-01 4.84484524e-01 1.09828973e+00 7.85172403e-01
3.28949332e-01 -6.31090820e-01 1.95549428e-01 -6.00219011e-01
3.87862742e-01 -2.06405297e-01 -2.44459093e-01 2.19761610e-01
-4.49744105e-01 4.86137658e-01 4.48293924e-01 -2.60557562e-01
-7.63145626e-01 4.82542738e-02 -2.55202234e-01 -3.28390956e-01
-2.60445327e-01 5.57615280e-01 5.11070304e-02 -5.10029607e-02
5.83058417e-01 4.51467425e-01 -3.11715119e-02 -3.89258921e-01
5.53418636e-01 6.48061812e-01 3.87967050e-01 -3.78752887e-01
7.69540071e-01 3.22587937e-01 4.46237363e-02 -5.99134445e-01
-4.53531414e-01 -1.28928244e-01 -1.00219798e+00 -1.02538735e-01
1.22837567e+00 -9.81143296e-01 3.33080590e-02 4.70543146e-01
-1.44606698e+00 -6.63963437e-01 -4.19299118e-02 2.41672754e-01
-8.09149504e-01 2.47643813e-01 -7.33976781e-01 -3.50545943e-01
-3.32356006e-01 -1.47214842e+00 1.35173500e+00 -1.78322390e-01
-1.16938680e-01 -1.07323527e+00 2.79206216e-01 4.47842568e-01
5.01830518e-01 1.48121282e-01 1.00710177e+00 -4.11963224e-01
-7.24287629e-01 1.16487756e-01 -4.35028255e-01 1.92968011e-01
-2.02820972e-01 -7.84770027e-02 -1.02594697e+00 -3.50534201e-01
-4.44698811e-01 -4.76560444e-01 1.24805522e+00 3.12046707e-01
2.66583055e-01 -3.39922339e-01 -2.60805100e-01 3.19470137e-01
1.46182239e+00 -1.08712181e-01 8.89659524e-01 6.66263759e-01
6.68900669e-01 4.80612844e-01 2.71981686e-01 -3.03617388e-01
5.42984366e-01 1.08413470e+00 3.84327650e-01 -3.41737717e-01
-5.18792152e-01 -1.33266836e-01 9.18436587e-01 1.09787583e+00
-1.07047059e-01 -5.01124501e-01 -9.58581150e-01 7.57812560e-01
-1.83964527e+00 -6.93155706e-01 -1.03884578e-01 1.85293734e+00
9.83916938e-01 1.49531603e-01 6.53837398e-02 -1.11305438e-01
4.88551289e-01 1.61030650e-01 -6.69198297e-03 -9.33628023e-01
-3.05141628e-01 4.73933518e-01 5.93398154e-01 5.89125097e-01
-1.10915625e+00 1.20812750e+00 6.01058388e+00 7.75763392e-01
-1.36605668e+00 6.15224600e-01 5.92397630e-01 -1.21781223e-01
-1.46685600e-01 9.83428303e-03 -5.63787580e-01 3.16216528e-01
1.57264972e+00 4.27393109e-01 5.27323663e-01 2.31665105e-01
1.42552570e-01 -1.05216809e-01 -1.19585657e+00 6.43792808e-01
4.45817679e-01 -1.26707006e+00 2.42295027e-01 3.04781914e-01
7.37711728e-01 6.36019170e-01 1.55144647e-01 4.32125002e-01
2.03970566e-01 -1.07916045e+00 1.02720153e+00 6.67219400e-01
8.74159873e-01 -5.28081775e-01 8.04124534e-01 3.09629768e-01
-6.67043626e-01 3.60933930e-01 -2.96879381e-01 4.50209752e-02
2.57893562e-01 1.91929564e-01 -8.30499709e-01 6.64223313e-01
4.26815093e-01 6.69166803e-01 -9.33731437e-01 7.84626603e-01
-3.64735842e-01 7.07524538e-01 -1.53858796e-01 1.16628535e-01
7.56769896e-01 2.53082454e-01 5.80328941e-01 1.31445038e+00
2.69075692e-01 -6.56143248e-01 -1.02446362e-01 7.88801134e-01
-4.15636934e-02 2.46212125e-01 -5.99537551e-01 -2.13603035e-01
-2.59036958e-01 1.08440471e+00 -5.39188862e-01 -4.35713410e-01
-7.75479138e-01 1.47865677e+00 5.54655850e-01 4.47654903e-01
-8.87573600e-01 1.51949236e-02 1.26085341e-01 1.16371371e-01
5.89394927e-01 -2.54808724e-01 -9.13982838e-02 -1.22066784e+00
-2.30688900e-02 -9.61485565e-01 7.10938126e-02 -1.25595868e+00
-8.95444572e-01 7.38714755e-01 -1.73082575e-01 -1.06812191e+00
-5.64072967e-01 -7.16337979e-01 -3.65250677e-01 1.14861393e+00
-1.66616499e+00 -1.49940717e+00 2.85384804e-01 3.68504077e-01
6.41905487e-01 -1.94638088e-01 9.29993868e-01 1.62809253e-01
-4.03624922e-01 5.83812475e-01 1.30007029e-01 2.74275571e-01
8.45503271e-01 -1.15069854e+00 6.81040287e-01 9.34346735e-01
5.67171574e-01 3.21843743e-01 5.02905786e-01 -4.40157950e-01
-1.20076799e+00 -1.12422907e+00 1.31632125e+00 -7.50452518e-01
4.41156983e-01 -3.85823578e-01 -6.40447736e-01 8.09284806e-01
1.08313310e+00 -4.20023680e-01 2.39458814e-01 -1.03507653e-01
-2.64015555e-01 3.26683193e-01 -7.14662254e-01 5.86970389e-01
5.01702070e-01 -4.14610326e-01 -7.86003411e-01 2.11092070e-01
8.81119967e-01 -3.71733755e-01 -7.63178706e-01 3.45508069e-01
4.33517456e-01 -8.05265784e-01 8.47128630e-01 -4.76549119e-01
1.06525779e+00 -1.82651117e-01 -2.05836028e-01 -1.59642160e+00
-4.74271566e-01 -3.61286253e-01 -1.35452691e-02 1.04467082e+00
1.07203376e+00 -5.12383878e-01 3.94747227e-01 1.11012518e-01
-1.44565299e-01 -7.25723147e-01 -1.25752294e+00 -6.47307634e-01
4.80714291e-01 -2.53753066e-01 1.41776249e-01 6.82968974e-01
-2.78697431e-01 7.97884941e-01 -5.68971038e-01 -2.50715405e-01
3.16476554e-01 -1.03843927e-01 6.60657883e-01 -6.50507808e-01
-3.82583082e-01 -4.55513895e-01 -3.05015802e-01 -8.52663338e-01
1.41815066e-01 -1.20390940e+00 1.05533443e-01 -1.92053533e+00
5.05332708e-01 8.07663873e-02 -2.02352196e-01 7.17521369e-01
-1.96776509e-01 7.12879181e-01 4.58124995e-01 2.52905488e-01
-6.25227392e-01 4.86308128e-01 1.19416702e+00 -2.75565654e-01
1.93658710e-01 -5.06522000e-01 -4.43249911e-01 2.34949470e-01
6.31532967e-01 -5.39068103e-01 6.67222962e-03 -1.01138282e+00
2.23067418e-01 -8.81827064e-03 5.52101135e-01 -9.25666571e-01
-3.43836546e-02 2.58576691e-01 5.37549913e-01 -3.74369234e-01
4.19053108e-01 -5.38228869e-01 -1.33551240e-01 5.60535192e-01
-3.74612272e-01 2.62807399e-01 4.55294281e-01 2.87550151e-01
-1.32298201e-01 -2.48502985e-01 7.93392539e-01 -2.60868937e-01
-5.93948364e-01 -6.21576607e-02 -6.19459987e-01 -1.27489030e-01
6.32568240e-01 -1.31476045e-01 -3.05006355e-01 -4.53438073e-01
-7.25349903e-01 -6.69603497e-02 6.78019822e-01 6.67814493e-01
2.34753653e-01 -1.41906428e+00 -1.01795518e+00 5.14154844e-02
2.21102163e-01 -5.50860822e-01 -1.81942746e-01 1.21657276e+00
-4.90798533e-01 8.49504709e-01 -2.25804195e-01 -8.14305723e-01
-1.12993634e+00 5.48757315e-01 1.82544813e-01 -4.55170035e-01
-3.35467368e-01 5.38257837e-01 -6.28023269e-03 -6.25143528e-01
-1.13644026e-01 -2.39446402e-01 -1.42170891e-01 9.20701846e-02
1.49400383e-01 -3.53522114e-02 3.47528130e-01 -1.25089550e+00
-3.79153460e-01 5.57244956e-01 -1.77126467e-01 -7.47074604e-01
1.25965941e+00 -1.23693310e-01 -1.52595237e-01 5.67316532e-01
1.32659936e+00 -3.01291168e-01 -1.00769794e+00 -3.06146353e-01
-1.06519535e-01 6.44853339e-02 1.60813242e-01 -1.22078991e+00
-9.39197958e-01 1.18464720e+00 6.81425095e-01 -2.81968683e-01
9.88008916e-01 4.86752838e-02 7.41292477e-01 2.21391514e-01
2.66991705e-01 -9.36701715e-01 7.84900635e-02 6.88892782e-01
9.63479936e-01 -1.25691462e+00 -3.05671901e-01 1.20008498e-01
-7.81922579e-01 1.03900087e+00 3.18337977e-01 -1.11441955e-01
1.92099169e-01 8.01928043e-02 2.66647965e-01 8.93624965e-03
-1.07329965e+00 -3.81281942e-01 5.03070652e-01 4.83607650e-01
6.17911398e-01 -8.46366510e-02 -3.69092375e-01 -5.78389317e-03
-1.15996286e-01 3.03683393e-02 4.02019411e-01 7.65595615e-01
-2.37338066e-01 -1.34104931e+00 -2.89640695e-01 2.32076153e-01
-5.35202265e-01 -6.01277888e-01 -6.37273073e-01 8.64171863e-01
1.69725150e-01 8.24444473e-01 -1.07270040e-01 -3.16641301e-01
1.93091720e-01 4.92182970e-01 7.25285828e-01 -6.22466266e-01
-6.65874481e-01 9.81042162e-02 3.71790618e-01 -3.75214905e-01
-6.30921602e-01 -8.35009158e-01 -8.19205940e-01 -5.00560813e-02
-1.62262887e-01 -1.70262843e-01 9.29081202e-01 1.26159358e+00
4.52833205e-01 6.25016570e-01 2.24116832e-01 -1.14019191e+00
-2.74001420e-01 -1.34260845e+00 4.21151996e-01 2.32541338e-01
2.43270561e-01 -3.91471326e-01 -5.54808788e-02 3.79221499e-01] | [11.417673110961914, 1.5016207695007324] |
349204cc-8da4-490f-95d1-8e620a0f9703 | the-role-of-context-and-uncertainty-in | null | null | https://aclanthology.org/2022.coling-1.67 | https://aclanthology.org/2022.coling-1.67.pdf | The Role of Context and Uncertainty in Shallow Discourse Parsing | Discourse parsing has proven to be useful for a number of NLP tasks that require complex reasoning. However, over a decade since the advent of the Penn Discourse Treebank, predicting implicit discourse relations in text remains challenging. There are several possible reasons for this, and we hypothesize that models should be exposed to more context as it plays an important role in accurate human annotation; meanwhile adding uncertainty measures can improve model accuracy and calibration. To thoroughly investigate this phenomenon, we perform a series of experiments to determine 1) the effects of context on human judgments, and 2) the effect of quantifying uncertainty with annotator confidence ratings on model accuracy and calibration (which we measure using the Brier score (Brier et al, 1950)). We find that including annotator accuracy and confidence improves model accuracy, and incorporating confidence in the model’s temperature function can lead to models with significantly better-calibrated confidence measures. We also find some insightful qualitative results regarding human and model behavior on these datasets. | ['Malihe Alikhani', 'Junyi Jessy Li', 'Remi Choi', 'Katherine Atwell'] | null | null | null | null | coling-2022-10 | ['discourse-parsing'] | ['natural-language-processing'] | [-8.54789764e-02 7.58665144e-01 -4.07528758e-01 -5.53125441e-01
-9.02941048e-01 -7.08943427e-01 8.17732334e-01 7.11397707e-01
-7.31433809e-01 1.09501827e+00 7.89616346e-01 -6.01787269e-01
-3.51077095e-02 -5.35945475e-01 -5.05203605e-01 -2.00721785e-01
1.54306650e-01 4.82418656e-01 2.63828307e-01 -5.86316176e-02
4.91195947e-01 3.49653028e-02 -1.07075059e+00 1.60044000e-01
1.02813673e+00 5.11971533e-01 -7.25889429e-02 5.95742702e-01
-5.44719025e-02 1.53444374e+00 -8.79234612e-01 -8.17157090e-01
-2.35701501e-01 -3.23990405e-01 -1.28461444e+00 -1.27614632e-01
5.39822951e-02 -2.49108285e-01 -1.98048446e-03 9.15862262e-01
1.46704316e-01 2.34025270e-01 9.04190004e-01 -8.37141573e-01
-4.21289861e-01 1.05029976e+00 -3.48932654e-01 2.99649060e-01
5.68349183e-01 1.99589565e-01 1.23945892e+00 -3.95556986e-01
8.27053726e-01 1.38648474e+00 7.46472061e-01 2.69169539e-01
-1.20490074e+00 -4.38292295e-01 1.84236348e-01 2.29609236e-01
-1.02445519e+00 -5.38951039e-01 5.80755711e-01 -5.36845148e-01
6.57493532e-01 2.76655227e-01 3.22333187e-01 8.61159563e-01
7.93441013e-02 5.13913512e-01 1.39523375e+00 -6.23755813e-01
2.86866188e-01 4.99729872e-01 5.04317224e-01 3.59162211e-01
4.41217750e-01 -1.30448520e-01 -4.25282896e-01 -3.47864419e-01
6.15048766e-01 -8.08380306e-01 -3.90863121e-01 9.57986563e-02
-9.20730829e-01 9.78695154e-01 3.39682162e-01 5.84304571e-01
-1.33636564e-01 -3.13274972e-02 4.34652179e-01 1.80352792e-01
5.74320734e-01 1.09891343e+00 -3.60928833e-01 -5.06962597e-01
-6.50829077e-01 5.41491270e-01 1.20728791e+00 5.01493752e-01
2.58208185e-01 -4.24681574e-01 -7.11882636e-02 7.54146039e-01
3.24944109e-01 -9.76864696e-02 3.28515530e-01 -1.45623076e+00
6.36303008e-01 5.77225447e-01 7.07268476e-01 -1.06538808e+00
-5.12630999e-01 -1.30802825e-01 -1.55534938e-01 3.73442570e-04
1.06668937e+00 -3.05669039e-01 -4.88961875e-01 1.91307342e+00
1.95894256e-01 -4.35964435e-01 1.19461007e-02 1.02941966e+00
4.81045663e-01 5.29893756e-01 5.76922953e-01 -5.01513779e-01
1.51611102e+00 -6.60001993e-01 -1.07371438e+00 -5.19716799e-01
1.12625110e+00 -7.56634593e-01 1.26530409e+00 6.44021705e-02
-1.25890911e+00 -1.82976961e-01 -1.03269339e+00 -3.22073132e-01
5.90433031e-02 -3.03635210e-01 7.92272270e-01 6.81886017e-01
-5.12795687e-01 6.44133151e-01 -1.06873345e+00 -2.99146652e-01
1.27571300e-01 -2.80951764e-02 -1.27171576e-01 1.23971961e-01
-1.38484132e+00 1.61470091e+00 4.36116964e-01 -1.12969456e-02
-2.38386933e-02 -2.66962558e-01 -9.74726677e-01 1.72198668e-01
6.70692503e-01 -4.90551591e-01 1.52182508e+00 -1.09431815e+00
-1.26462340e+00 7.74136007e-01 -3.29349011e-01 -5.27082145e-01
6.21416628e-01 -2.77560771e-01 1.26416376e-02 -6.97060749e-02
9.74006653e-02 3.55741918e-01 7.58186877e-02 -1.23729420e+00
-3.85086358e-01 -3.63752663e-01 3.64047319e-01 4.25700039e-01
-3.68153229e-02 3.14014584e-01 8.91518146e-02 -4.66534644e-01
1.78778142e-01 -8.84079576e-01 -5.53254448e-02 -2.76024610e-01
-9.93254036e-02 -4.41667706e-01 1.72275633e-01 -9.14245546e-01
1.54819918e+00 -1.72884369e+00 -5.90207241e-02 5.18705696e-02
2.85815686e-01 2.36135557e-01 3.52416992e-01 2.27063626e-01
2.46180475e-01 5.92878461e-01 -2.18910933e-01 -1.54069379e-01
6.04961514e-02 2.03939810e-01 -8.76983535e-03 1.73616931e-01
2.62938201e-01 8.46337855e-01 -6.93179548e-01 -8.64812672e-01
-1.25671983e-01 1.50165230e-01 -4.77136284e-01 1.15774110e-01
-3.83287907e-01 3.25927764e-01 -3.81704926e-01 3.57275277e-01
1.50068328e-01 -5.17536461e-01 5.73961794e-01 5.89164998e-03
-1.68931469e-01 9.50269699e-01 -8.64069819e-01 9.89715397e-01
-1.29007131e-01 9.19754505e-01 1.55694410e-01 -7.04387188e-01
6.79497361e-01 4.50032055e-01 -1.66043267e-01 -4.30322856e-01
4.31616396e-01 1.50703490e-01 5.84956467e-01 -6.27389133e-01
8.67629945e-01 -4.07228708e-01 -8.81905928e-02 6.26599252e-01
-4.92310196e-01 -3.02797168e-01 2.56908447e-01 3.22331041e-01
9.42529202e-01 -6.03419021e-02 3.93224895e-01 -5.81693113e-01
1.87929630e-01 5.36907077e-01 6.41312420e-01 6.95117950e-01
-4.85910296e-01 2.17786878e-01 9.38941658e-01 -2.65504241e-01
-1.07721519e+00 -5.37552416e-01 -3.19008261e-01 8.75988662e-01
-8.79417211e-02 -3.54650289e-01 -8.86605799e-01 -4.49757993e-01
-2.06590354e-01 1.41900325e+00 -7.01008677e-01 8.27685222e-02
-7.59897351e-01 -7.40547538e-01 5.52857220e-01 7.80891299e-01
4.52449083e-01 -7.75650859e-01 -7.65037656e-01 1.44046858e-01
-6.47458494e-01 -1.06271803e+00 -1.42562807e-01 1.29617766e-01
-8.64534438e-01 -1.21062374e+00 -2.31251583e-01 -4.29613054e-01
4.16814953e-01 -2.21601322e-01 1.36110759e+00 5.16203046e-01
4.21243101e-01 3.02658170e-01 -4.92627233e-01 -5.77201366e-01
-9.28755343e-01 2.24916086e-01 -2.03375503e-01 -9.09411192e-01
5.26479959e-01 -1.57574654e-01 -1.66464835e-01 2.26781636e-01
-6.15418315e-01 7.59834498e-02 1.66703835e-01 9.29733634e-01
3.97809688e-03 -1.99224830e-01 5.83413601e-01 -1.33487594e+00
1.06246185e+00 -3.11383516e-01 -5.34490407e-01 3.93363446e-01
-8.56754601e-01 1.65204525e-01 -2.39227097e-02 -3.78161877e-01
-1.26413059e+00 -5.85992098e-01 -1.50493579e-02 2.83721179e-01
9.16528329e-02 8.61545861e-01 2.07995743e-01 2.92824566e-01
9.43071544e-01 -6.57167196e-01 9.37104896e-02 -1.21068932e-01
5.20812757e-02 5.64936996e-01 2.86194891e-01 -1.01371133e+00
4.25092310e-01 -1.27968177e-01 -3.20799917e-01 -5.60854554e-01
-1.36893547e+00 -5.07991128e-02 -5.21001518e-01 -2.04623520e-01
1.07155633e+00 -8.30517769e-01 -7.24188387e-01 -6.00715727e-02
-1.25668645e+00 -6.26567006e-01 8.18176009e-03 7.00908482e-01
-3.82313281e-01 2.79308677e-01 -9.27299440e-01 -1.09248483e+00
1.42314390e-03 -1.18744719e+00 4.51513559e-01 3.86409283e-01
-1.19029701e+00 -1.16932499e+00 1.37825897e-02 7.30030775e-01
2.94102758e-01 2.91635424e-01 1.30689085e+00 -9.39886332e-01
-3.34438562e-01 -3.05545032e-02 -2.11041123e-01 1.79594189e-01
-5.66633791e-02 -1.38548305e-02 -1.00741851e+00 1.03634663e-01
3.09020489e-01 -6.82543755e-01 5.19353688e-01 2.78515249e-01
6.78119779e-01 -4.78977531e-01 -2.67225474e-01 -3.65503430e-01
1.11775172e+00 4.49637473e-02 5.84112585e-01 5.81666648e-01
4.04197812e-01 8.67530406e-01 7.85935283e-01 -3.01012932e-03
8.08953226e-01 5.26996732e-01 -7.30179623e-02 3.40359092e-01
2.70813406e-01 -1.84726626e-01 5.69016337e-02 8.38302374e-01
-3.36118162e-01 -3.03524941e-01 -1.20314956e+00 3.48392576e-01
-1.89516604e+00 -9.57092881e-01 -8.83090720e-02 2.10871291e+00
1.19646907e+00 6.31327748e-01 3.94278429e-02 2.51234651e-01
6.22208178e-01 2.52102584e-01 -2.15790734e-01 -4.77803051e-01
-8.02481845e-02 -1.16399392e-01 2.27725849e-01 1.01835930e+00
-6.64855719e-01 8.06412220e-01 7.12386417e+00 2.32739642e-01
-6.10453486e-01 3.85704963e-03 1.13849556e+00 1.07080705e-01
-4.10800457e-01 1.30559698e-01 -4.36884612e-01 3.06106925e-01
1.01415348e+00 -3.91157359e-01 9.44124609e-02 6.99439049e-01
6.12435751e-02 -6.99853182e-01 -1.29681122e+00 1.62447602e-01
-1.30643293e-01 -1.15464687e+00 -4.71032023e-01 -2.61282474e-02
6.78826869e-01 -4.28882688e-01 -4.41167742e-01 5.20577610e-01
5.99186242e-01 -1.11579764e+00 7.48242736e-01 2.67621249e-01
1.35319084e-01 -5.97887754e-01 1.14361668e+00 6.49798989e-01
-5.16439438e-01 7.54017159e-02 -2.13274628e-01 -6.57344043e-01
2.77219236e-01 3.82028252e-01 -1.09222019e+00 6.11843802e-02
3.82910907e-01 4.20534387e-02 -4.85243797e-01 5.20979762e-01
-4.19268876e-01 9.48938131e-01 -2.21424013e-01 -2.82422632e-01
1.39568552e-01 -8.73381272e-02 3.73262346e-01 9.53320384e-01
-1.39516607e-01 6.90147519e-01 -1.10739537e-01 8.00134242e-01
-3.86940166e-02 -9.65729058e-02 -1.87073246e-01 -3.54207337e-01
9.30780292e-01 8.10151994e-01 -8.80000055e-01 -3.93922389e-01
-3.19499969e-01 3.88940871e-01 5.95038354e-01 2.43403435e-01
-7.80212402e-01 3.50648584e-03 3.30916524e-01 2.09024206e-01
-1.53943032e-01 -2.61529058e-01 -7.93875933e-01 -9.43590581e-01
2.33706157e-03 -9.43065405e-01 3.48755866e-01 -7.98068762e-01
-1.08235514e+00 3.79550785e-01 1.34791031e-01 -3.88020843e-01
-4.25022215e-01 -4.55914289e-01 -4.56952900e-01 8.46541882e-01
-1.19214845e+00 -6.61675930e-01 -4.60080579e-02 -1.58606529e-01
3.34658951e-01 5.50902843e-01 8.27091992e-01 -8.97103697e-02
-3.92820626e-01 5.28680444e-01 -3.63484740e-01 3.05927366e-01
1.01654148e+00 -1.30885816e+00 3.19406480e-01 5.22524297e-01
-1.52581155e-01 8.92254829e-01 1.20523548e+00 -7.69476652e-01
-7.73102641e-01 -4.46486652e-01 1.09871280e+00 -1.10156262e+00
7.31610298e-01 1.31280348e-01 -1.25526488e+00 1.10445523e+00
2.42149740e-01 -7.54361272e-01 6.39410138e-01 6.86367989e-01
-3.83622587e-01 5.70133924e-01 -1.14106321e+00 6.26032114e-01
7.42498040e-01 -5.74813485e-01 -1.13758790e+00 1.58174723e-01
9.44552481e-01 -8.48365188e-01 -1.19449341e+00 4.41916972e-01
5.00278413e-01 -8.94102454e-01 4.53656465e-01 -5.23839831e-01
6.95047498e-01 2.58470699e-02 -1.02022618e-01 -1.23905241e+00
-3.64399672e-01 -3.96344215e-02 1.37955487e-01 1.21579576e+00
9.42432880e-01 -5.64210892e-01 6.27196908e-01 1.76924312e+00
5.74042201e-02 -7.86421120e-01 -6.32741272e-01 -3.82324845e-01
4.94466454e-01 -4.01455700e-01 4.05464470e-01 1.24011970e+00
5.29216230e-01 5.48879564e-01 -2.49246117e-02 5.87363206e-02
1.71523035e-01 -2.40629971e-01 5.05934954e-01 -1.32735384e+00
-2.36169443e-01 -3.47004980e-01 -6.47499412e-02 -8.28432798e-01
1.95257202e-01 -3.35637182e-01 1.05431795e-01 -1.56659245e+00
2.03884080e-01 -7.72568524e-01 2.97816932e-01 2.29413509e-01
-6.14219308e-01 -3.06111097e-01 4.20566261e-01 3.74474674e-01
-5.14792025e-01 2.05008805e-01 1.11034954e+00 2.91215777e-01
-2.04182342e-01 -4.45934944e-02 -9.12730336e-01 9.14993048e-01
7.11018264e-01 -3.97159815e-01 -3.36779505e-01 -6.51890337e-01
4.88518596e-01 3.40154767e-01 1.03326671e-01 -6.74643517e-01
1.73444003e-01 -3.81768316e-01 2.84698784e-01 -9.25364494e-02
4.89111900e-01 -5.57312608e-01 2.47508679e-02 3.32097769e-01
-7.28707910e-01 2.70953804e-01 2.72046387e-01 3.01912010e-01
-1.62261695e-01 -5.76696873e-01 5.66033602e-01 -2.18933642e-01
-2.27949500e-01 -6.95509732e-01 -4.92037982e-01 3.78341526e-01
7.40458727e-01 5.02801128e-02 -6.36610568e-01 -5.43305814e-01
-5.58827221e-01 3.23565066e-01 8.25945973e-01 1.08598530e-01
-1.76580474e-01 -9.37540412e-01 -5.40636778e-01 -6.28042221e-01
-1.67990968e-01 9.21234265e-02 -2.31239811e-01 8.15892816e-01
-5.11288822e-01 5.60655534e-01 2.14197591e-01 -3.91782612e-01
-1.20527899e+00 5.28431609e-02 2.94712037e-01 -5.39291859e-01
-2.20211282e-01 8.64176869e-01 -1.79926142e-01 -1.84862182e-01
2.30328664e-01 -4.92951661e-01 -2.67117739e-01 1.59829408e-01
4.94050235e-01 5.31110764e-01 -2.05278441e-01 -5.54122746e-01
-1.71889529e-01 1.33722007e-01 -2.71503150e-01 -4.27992314e-01
1.08823276e+00 -2.12775692e-01 5.28964177e-02 8.22330058e-01
4.15135384e-01 1.22536577e-01 -1.14208460e+00 -1.80735514e-01
3.83277744e-01 -3.52870613e-01 -7.01053813e-02 -1.11940575e+00
-1.74160898e-01 5.17328143e-01 -8.09963718e-02 4.64328855e-01
4.74924773e-01 9.33933482e-02 3.55785400e-01 4.70721751e-01
3.61115068e-01 -1.21699107e+00 -8.21715146e-02 5.00295222e-01
7.18058228e-01 -1.61298954e+00 4.36922461e-01 -5.77508092e-01
-1.00052953e+00 7.78392911e-01 8.29439402e-01 2.72325367e-01
2.49569491e-01 2.28579953e-01 2.69348830e-01 -2.60405660e-01
-9.33971405e-01 3.01456034e-01 2.57228967e-03 2.20765918e-01
1.13871086e+00 2.07890958e-01 -7.21155226e-01 7.39768386e-01
-5.85062444e-01 -1.07396878e-01 6.95114493e-01 9.36498582e-01
-7.01745570e-01 -9.59704936e-01 -5.94399869e-01 5.13867617e-01
-4.53134954e-01 1.30543202e-01 -5.77521265e-01 1.05499482e+00
-4.21488702e-01 1.12054610e+00 1.35420740e-01 -1.65345922e-01
1.30985752e-01 4.92400885e-01 4.91866142e-01 -6.20717466e-01
-6.30743861e-01 -2.61530727e-01 9.67242539e-01 -2.18269885e-01
-6.11952722e-01 -7.49299943e-01 -1.34922969e+00 -6.09867573e-01
-9.10836160e-01 7.12900341e-01 5.19647837e-01 1.39327800e+00
-1.30642895e-02 3.25602889e-01 4.67620976e-02 -4.51355815e-01
-8.39963973e-01 -1.44814467e+00 -1.91249579e-01 3.57663691e-01
-2.10213903e-02 -7.15523601e-01 -5.81790209e-01 3.18306684e-02] | [10.603748321533203, 8.739092826843262] |
a20438a7-94da-4bab-9647-6402fbd30122 | faster-gradient-free-algorithms-for-nonsmooth | 2301.06428 | null | https://arxiv.org/abs/2301.06428v2 | https://arxiv.org/pdf/2301.06428v2.pdf | Faster Gradient-Free Algorithms for Nonsmooth Nonconvex Stochastic Optimization | We consider the optimization problem of the form $\min_{x \in \mathbb{R}^d} f(x) \triangleq \mathbb{E}_{\xi} [F(x; \xi)]$, where the component $F(x;\xi)$ is $L$-mean-squared Lipschitz but possibly nonconvex and nonsmooth. The recently proposed gradient-free method requires at most $\mathcal{O}( L^4 d^{3/2} \epsilon^{-4} + \Delta L^3 d^{3/2} \delta^{-1} \epsilon^{-4})$ stochastic zeroth-order oracle complexity to find a $(\delta,\epsilon)$-Goldstein stationary point of objective function, where $\Delta = f(x_0) - \inf_{x \in \mathbb{R}^d} f(x)$ and $x_0$ is the initial point of the algorithm. This paper proposes a more efficient algorithm using stochastic recursive gradient estimators, which improves the complexity to $\mathcal{O}(L^3 d^{3/2} \epsilon^{-3}+ \Delta L^2 d^{3/2} \delta^{-1} \epsilon^{-3})$. | ['Luo Luo', 'Jing Xu', 'Lesi Chen'] | 2023-01-16 | null | null | null | null | ['stochastic-optimization'] | ['methodology'] | [-1.06752217e-01 2.60656208e-01 4.46288511e-02 -8.24109539e-02
-1.14142764e+00 -6.34708285e-01 -4.40045297e-01 -4.91499193e-02
-8.78541827e-01 1.20890284e+00 -5.55769801e-01 -6.13753319e-01
-8.92718315e-01 -8.06138813e-01 -6.79046690e-01 -9.95643616e-01
-8.34142625e-01 1.81376711e-01 -7.67648891e-02 -3.23466301e-01
2.66617060e-01 3.83570284e-01 -1.34330392e+00 -5.26770651e-01
1.15955567e+00 1.90376520e+00 -2.54061390e-02 8.21423054e-01
2.04502698e-02 5.82043529e-01 -6.06472492e-01 -3.31295758e-01
6.86436653e-01 -8.29580128e-01 -5.87086320e-01 -1.46134913e-01
1.18457250e-01 4.85450886e-02 -1.56304434e-01 1.78649509e+00
4.25984889e-01 5.96070409e-01 8.01399350e-01 -6.67566180e-01
-3.37819785e-01 -5.82581758e-03 -8.63874376e-01 2.72237629e-01
4.55316752e-02 1.71490148e-01 5.79229593e-01 -6.24257088e-01
2.70357549e-01 7.01073050e-01 6.75020516e-01 3.57776403e-01
-8.61500323e-01 -8.55691195e-01 2.65297532e-01 -5.82173705e-01
-1.82476664e+00 1.40610382e-01 3.70567501e-01 -4.57088709e-01
4.69981641e-01 5.97559690e-01 4.44569886e-01 -2.26690829e-01
5.29051304e-01 1.93878829e-01 1.45312095e+00 -2.91552722e-01
3.07146251e-01 1.30093649e-01 -8.90536383e-02 1.30598652e+00
1.06862590e-01 1.11778222e-01 1.66527793e-01 1.24378003e-01
1.18791974e+00 -1.77887846e-02 -3.75560999e-01 8.04498911e-01
-5.10660529e-01 1.03762722e+00 1.97189674e-01 3.11642408e-01
-1.95019752e-01 5.30251861e-01 -9.64387283e-02 3.88649821e-01
5.10721862e-01 5.25007606e-01 -4.84171331e-01 -3.97213280e-01
-1.09780276e+00 4.03111309e-01 6.68882966e-01 1.30199182e+00
8.14863265e-01 3.53617460e-01 -6.34941012e-02 6.81330025e-01
2.95863152e-01 1.08022094e+00 -1.14442535e-01 -1.60185480e+00
5.77639222e-01 5.27416587e-01 2.88504571e-01 -7.37770319e-01
-2.55160511e-01 -5.06871045e-01 -9.48684335e-01 4.62861300e-01
6.78878427e-01 -7.10849404e-01 -4.80548710e-01 1.70534170e+00
2.65546795e-02 -4.81701821e-01 -4.79952872e-01 8.93432736e-01
1.63874954e-01 7.39454806e-01 -3.96491766e-01 -1.05916417e+00
1.15154994e+00 -2.57373124e-01 -2.76455671e-01 7.32648969e-02
1.52049080e-01 -9.07146633e-01 1.21304107e+00 4.77799803e-01
-1.95076632e+00 -3.29366833e-01 -8.08226466e-01 5.53021848e-01
-6.26394898e-02 -1.42125413e-01 3.78795654e-01 8.10031831e-01
-8.83908153e-01 6.09565437e-01 -5.63415110e-01 7.41924644e-01
4.55431104e-01 7.64333844e-01 -7.39744976e-02 -5.58030680e-02
-9.41617966e-01 3.21824253e-01 -2.71779507e-01 1.25555560e-01
-6.16249621e-01 -7.81674087e-01 -5.41855931e-01 -2.00094372e-01
4.38721597e-01 -4.71516214e-02 7.54054606e-01 -6.56236112e-01
-1.34414279e+00 6.92138910e-01 -6.49692193e-02 -8.58141948e-03
6.70329094e-01 2.34584451e-01 -1.99822187e-01 3.74289975e-02
5.24084806e-01 2.39236906e-01 3.53776366e-01 -1.08375180e+00
-8.21984947e-01 -8.75997126e-01 8.80512372e-02 -4.87476354e-03
3.07095833e-02 1.58699363e-01 -4.45996314e-01 -7.91214287e-01
2.64060885e-01 -8.75964642e-01 -3.98404479e-01 -2.19541967e-01
-2.84819603e-01 7.70226046e-02 3.62333268e-01 -7.05168307e-01
1.57677948e+00 -1.96956229e+00 -9.51817259e-02 6.73347950e-01
5.76243214e-02 2.97333486e-03 7.09706485e-01 -3.77282267e-04
2.36798510e-01 3.16038191e-01 -6.49820864e-01 8.34858641e-02
-4.40589413e-02 6.11795448e-02 1.56181455e-01 8.43401611e-01
-3.68236095e-01 1.60885274e-01 -8.44448447e-01 -2.09155381e-01
-1.04456529e-01 5.43088615e-01 -3.28216463e-01 -3.47500890e-01
-1.19329095e-01 5.06563962e-01 -7.94748724e-01 6.63002610e-01
9.98165190e-01 -1.93954363e-01 -1.96652517e-01 4.45289791e-01
-6.43214464e-01 -3.92246664e-01 -1.98748994e+00 1.23254728e+00
-3.30972522e-01 1.42773435e-01 9.26988542e-01 -1.40185535e+00
9.03920829e-01 5.15600927e-02 1.08033001e+00 -7.76011407e-01
4.75570828e-01 4.44275111e-01 -4.32851285e-01 -3.65509719e-01
-1.14513285e-01 -9.36476707e-01 -1.44752324e-01 4.25704807e-01
-3.68043810e-01 -4.23211128e-01 1.86959952e-01 -4.60739493e-01
1.31543458e+00 -2.69259930e-01 -1.82133287e-01 -8.15844119e-01
6.25707150e-01 -1.00955278e-01 6.77747011e-01 8.20417047e-01
-2.78141618e-01 2.33772054e-01 9.94933963e-01 -1.11450963e-01
-7.29138315e-01 -1.01775563e+00 -7.30638027e-01 1.12216175e+00
5.57307303e-01 1.97457328e-01 -7.83842385e-01 -7.69250691e-01
-4.24877629e-02 8.31417143e-01 -6.55802846e-01 2.04585209e-01
-6.15725219e-01 -1.07907534e+00 5.44967532e-01 5.74754775e-01
7.70800591e-01 -1.01035178e+00 -6.09904826e-01 2.23243937e-01
1.95721656e-01 -1.51081309e-01 -8.02627742e-01 6.96710467e-01
-7.89947033e-01 -9.13771331e-01 -8.56280804e-01 -4.77026612e-01
1.37748325e+00 -4.58833963e-01 8.05677176e-01 -1.38186082e-01
-4.65940386e-01 3.11132133e-01 -5.97182848e-02 -6.78530812e-01
2.29168877e-01 -6.24327540e-01 1.85974196e-01 -3.06496382e-01
-1.91204116e-01 -4.09464896e-01 -1.08809161e+00 5.45027196e-01
-8.75464797e-01 -7.81479716e-01 1.11253746e-01 8.46201956e-01
1.32758498e+00 5.40860176e-01 4.98053193e-01 -5.50081789e-01
5.78444660e-01 -1.27284870e-01 -1.57659662e+00 -1.21968307e-01
-9.00244713e-01 6.73067430e-03 1.06923056e+00 -1.86371282e-01
-6.76623642e-01 -2.65420973e-01 -3.06350946e-01 -5.65613866e-01
3.26317549e-01 4.19328719e-01 2.36132741e-01 -2.22493470e-01
7.21494019e-01 3.11920315e-01 -1.38334274e-01 -5.42018354e-01
1.42384782e-01 3.33806425e-01 6.33653402e-01 -7.37087607e-01
4.74019587e-01 6.42055869e-01 5.02908826e-01 -4.96310890e-01
-8.15049171e-01 6.16503283e-02 5.88290878e-02 -6.09243847e-03
9.89987254e-01 -4.25842375e-01 -1.41118503e+00 -3.05100918e-01
-4.33943361e-01 -3.75233442e-01 -1.03121185e+00 7.70470619e-01
-8.13161731e-01 9.17749181e-02 -2.11777523e-01 -1.28952098e+00
-3.05806965e-01 -1.33827806e+00 5.35578370e-01 3.76677185e-01
1.38974413e-01 -1.05962694e+00 -3.42606813e-01 4.33280021e-01
4.34027106e-01 3.07923317e-01 8.70332897e-01 -9.90521163e-02
-5.18941879e-01 -6.55236661e-01 -3.44646990e-01 1.12640381e+00
-1.06815502e-01 -4.03091192e-01 -2.29829967e-01 -2.73242235e-01
5.45484960e-01 1.18759900e-01 5.05681872e-01 8.45617414e-01
1.48458433e+00 -8.88846338e-01 -8.90373811e-02 6.85248494e-01
1.74561977e+00 8.08578432e-01 7.03500926e-01 -8.19771215e-02
-8.39325339e-02 1.97837260e-02 5.56682944e-01 8.58454704e-01
1.23867914e-01 2.87476122e-01 5.69719672e-01 -4.45753708e-02
2.73060411e-01 3.85778517e-01 2.59536982e-01 2.94855773e-01
-3.25533122e-01 -8.17876533e-02 -7.14154720e-01 6.23960912e-01
-1.26085818e+00 -8.13465536e-01 -1.53906018e-01 2.66609240e+00
9.87750649e-01 3.23795468e-01 -1.44411111e-02 9.70006660e-02
5.21469176e-01 -3.01062554e-01 -2.77325362e-01 -6.21087551e-01
-4.18165997e-02 1.20580542e+00 8.43067706e-01 7.00517416e-01
-7.79410005e-01 3.56775790e-01 3.66992116e+00 1.30774105e+00
-1.12580931e+00 -4.28662263e-02 7.63367951e-01 -5.68757534e-01
-2.98975885e-01 2.05467731e-01 -7.20452785e-01 1.12694275e+00
3.76831800e-01 -1.75110586e-02 5.50075173e-01 9.85568643e-01
2.52326280e-01 -7.16098130e-01 -6.45482838e-01 8.47891450e-01
-1.18015774e-01 -1.32314014e+00 -9.42341447e-01 3.97832900e-01
1.21659815e+00 -3.07891309e-01 3.53768677e-01 1.54917017e-01
6.48115456e-01 -1.39384103e+00 4.19189990e-01 3.00463706e-01
1.46914637e+00 -1.00887632e+00 7.15754032e-01 4.66321170e-01
-1.31880713e+00 -3.95059884e-01 -1.53706670e-01 1.10526852e-01
3.08925599e-01 9.79326963e-01 -1.68916881e-01 5.07551908e-01
1.26871610e+00 -4.47562337e-01 5.20441115e-01 6.84744060e-01
-4.55149775e-03 1.91503942e-01 -6.68794930e-01 -6.05723411e-02
4.79481161e-01 -6.76534534e-01 6.70546830e-01 9.01824355e-01
5.98073721e-01 1.02684891e+00 5.84766984e-01 7.87329018e-01
-3.84513289e-01 3.87803108e-01 -2.06076279e-01 3.72201055e-01
1.99914336e-01 5.96638858e-01 -9.23383594e-01 -6.39058417e-03
-2.78186768e-01 4.54285949e-01 -2.04520658e-01 3.81037027e-01
-1.02585387e+00 -1.28714359e+00 5.61435044e-01 7.48153269e-01
3.87795359e-01 -2.08277568e-01 -5.80416083e-01 -6.14475191e-01
3.59802276e-01 -3.37190777e-01 5.77620745e-01 -2.20121831e-01
-1.14870465e+00 4.31777716e-01 2.08099246e-01 -1.00048983e+00
-9.92059708e-02 -6.76800072e-01 -3.07997197e-01 1.15169632e+00
-9.16990399e-01 -2.36512676e-01 1.12249881e-01 9.00049448e-01
-8.20138231e-02 -5.59859872e-01 5.87257326e-01 4.77392644e-01
-1.22915536e-01 8.03580999e-01 6.60996497e-01 5.25643304e-02
-7.91469142e-02 -9.88726974e-01 -7.57293522e-01 4.29540783e-01
-5.91729105e-01 3.80308658e-01 5.33627331e-01 -6.50489211e-01
-1.34971297e+00 -8.63574088e-01 6.70061648e-01 -4.20231074e-02
3.16751033e-01 1.56787530e-01 -4.97634739e-01 5.52351236e-01
-1.26882374e-01 3.48171383e-01 5.44177353e-01 -6.06867313e-01
4.64705676e-01 -4.43228662e-01 -1.83725369e+00 5.59162557e-01
7.31153429e-01 -1.22981481e-01 1.86871439e-01 3.62245232e-01
7.42115304e-02 -3.93933833e-01 -1.43970239e+00 7.71439850e-01
2.85864383e-01 -1.00499630e+00 6.31605923e-01 -2.38399446e-01
-1.25012919e-01 -2.79594958e-01 -5.50408661e-01 -5.73160648e-01
1.71807155e-01 -1.15007794e+00 2.95104474e-01 5.18040895e-01
6.66171610e-01 -7.24129081e-01 8.99248838e-01 5.33054471e-01
-3.06675762e-01 -1.37813222e+00 -1.80414152e+00 -7.69926965e-01
8.27724993e-01 -3.86541992e-01 -2.00506389e-01 6.58235908e-01
-1.12796769e-01 -5.82162738e-01 8.38250294e-02 -1.76141664e-01
4.29482967e-01 -1.52035639e-01 1.90372795e-01 -8.54047656e-01
-7.42772877e-01 -6.26515090e-01 -2.29064859e-02 -1.01496851e+00
-2.80983955e-01 -6.67720139e-01 7.31173391e-03 -1.31806207e+00
-1.36040211e-01 -1.07176673e+00 -3.35078478e-01 2.58759469e-01
2.78668433e-01 -9.79339555e-02 -6.01253957e-02 -3.56291592e-01
-3.06797594e-01 2.70649493e-01 1.49805415e+00 -1.62666902e-01
-4.60271478e-01 3.44800383e-01 -7.80877411e-01 1.01179695e+00
2.88927406e-01 -4.89057362e-01 -4.19257045e-01 -4.91508067e-01
4.92581189e-01 9.54493403e-01 5.19174039e-02 -8.15731525e-01
4.64099161e-02 -5.71656048e-01 3.27069104e-01 -3.45029414e-01
4.55329716e-01 -5.36968291e-01 7.82780573e-02 2.40249902e-01
6.09489605e-02 3.52194011e-01 1.39525354e-01 1.56771958e-01
2.37213168e-02 -5.22228599e-01 1.25488734e+00 -5.87777674e-01
2.08139032e-01 3.24511588e-01 -2.03995094e-01 4.72128302e-01
1.30011308e+00 -3.31006795e-01 -1.01200216e-01 -5.91543198e-01
-8.30428898e-01 1.82989821e-01 1.93380147e-01 -4.90172356e-01
2.99112082e-01 -1.01475644e+00 -6.91508532e-01 2.95078337e-01
-6.36232555e-01 6.89751327e-01 3.40686709e-01 1.13868332e+00
-8.16773534e-01 -1.06558613e-02 3.78741562e-01 -5.00583708e-01
-3.20102096e-01 2.30773136e-01 6.45814359e-01 -4.02189374e-01
-8.30444917e-02 1.81934285e+00 -2.58421570e-01 7.45992959e-02
3.18199635e-01 -3.25944722e-01 6.82295263e-01 -2.38022715e-01
-3.24461125e-02 1.12524128e+00 -3.47660575e-03 -1.95498273e-01
-4.72559482e-01 6.10120296e-01 1.47563607e-01 -6.46220446e-01
1.11489081e+00 3.69311720e-02 -4.00247902e-01 5.76343536e-02
1.48774660e+00 3.12062353e-01 -1.46796227e+00 4.30339217e-01
-5.32621324e-01 -3.67884427e-01 -1.65091604e-01 -9.22562480e-01
-1.19356632e+00 7.77960002e-01 9.13701773e-01 3.95610213e-01
1.30265594e+00 5.63168153e-02 6.52772546e-01 -2.10002214e-01
3.67355973e-01 -2.03608584e+00 1.19731173e-01 6.03880823e-01
7.09271729e-01 -1.09734273e+00 2.84167707e-01 2.10996121e-02
-6.03128016e-01 7.97097921e-01 4.02444690e-01 -4.13421631e-01
1.34856546e+00 3.13297868e-01 -6.62913024e-02 -2.91494250e-01
1.84954509e-01 1.34819284e-01 2.83638269e-01 -1.87511861e-01
6.51559949e-01 1.13013178e-01 -7.85060048e-01 9.59711611e-01
-3.72406840e-01 1.41127240e-02 7.67634436e-02 1.38883615e+00
-4.27277505e-01 -9.36233997e-01 -3.76325607e-01 7.82936633e-01
-9.49587166e-01 1.44507006e-01 4.00844991e-01 8.10826242e-01
8.29704583e-01 9.51953411e-01 -1.00229062e-01 2.19024807e-01
3.40004951e-01 3.96760032e-02 7.26805031e-01 -2.43720144e-01
-4.53862846e-01 6.44053817e-01 -3.87981564e-01 -2.69608349e-01
2.05226585e-01 -6.90493584e-01 -2.07076454e+00 -3.59683037e-01
-2.27810472e-01 4.35793787e-01 8.41396093e-01 8.53168190e-01
1.74341258e-02 1.82914719e-01 9.32321668e-01 -2.06128970e-01
-5.54806113e-01 -5.57397485e-01 -1.10432339e+00 7.69029558e-02
1.93942606e-01 -6.68250203e-01 -5.38050532e-01 -1.50285706e-01] | [6.396448612213135, 4.549167156219482] |
92432293-d102-418e-a57d-4619694d36b3 | physically-interpretable-neural-networks-for | 1912.01752 | null | https://arxiv.org/abs/1912.01752v2 | https://arxiv.org/pdf/1912.01752v2.pdf | Physically Interpretable Neural Networks for the Geosciences: Applications to Earth System Variability | Neural networks have become increasingly prevalent within the geosciences, although a common limitation of their usage has been a lack of methods to interpret what the networks learn and how they make decisions. As such, neural networks have often been used within the geosciences to most accurately identify a desired output given a set of inputs, with the interpretation of what the network learns used as a secondary metric to ensure the network is making the right decision for the right reason. Neural network interpretation techniques have become more advanced in recent years, however, and we therefore propose that the ultimate objective of using a neural network can also be the interpretation of what the network has learned rather than the output itself. We show that the interpretation of neural networks can enable the discovery of scientifically meaningful connections within geoscientific data. In particular, we use two methods for neural network interpretation called backwards optimization and layerwise relevance propagation, both of which project the decision pathways of a network back onto the original input dimensions. To the best of our knowledge, LRP has not yet been applied to geoscientific research, and we believe it has great potential in this area. We show how these interpretation techniques can be used to reliably infer scientifically meaningful information from neural networks by applying them to common climate patterns. These results suggest that combining interpretable neural networks with novel scientific hypotheses will open the door to many new avenues in neural network-related geoscience research. | ['Imme Ebert-Uphoff', 'Benjamin A. Toms', 'Elizabeth A. Barnes'] | 2019-12-04 | null | null | null | null | ['network-interpretation'] | ['computer-vision'] | [ 4.47834522e-01 1.66658282e-01 -6.61453903e-02 -4.24499303e-01
2.80720323e-01 -6.77479446e-01 7.35580564e-01 2.12896809e-01
-3.83509189e-01 5.66975176e-01 2.06194311e-01 -1.06829762e+00
-5.16675770e-01 -8.88278067e-01 -7.42916524e-01 -7.27251768e-01
-2.39300460e-01 2.49266192e-01 -2.51340896e-01 -3.27991128e-01
4.86551225e-01 6.91750944e-01 -1.42341125e+00 -2.43384377e-04
7.17997849e-01 8.32233846e-01 1.52591243e-01 5.84052324e-01
-4.38007340e-02 6.24078333e-01 -3.10978740e-01 1.01647831e-01
1.97109446e-01 -5.73805869e-01 -8.36502314e-01 -4.53008711e-01
6.60481602e-02 -3.17732207e-02 -3.56303481e-03 9.57335353e-01
6.05779141e-02 3.83546859e-01 8.66228938e-01 -6.23191237e-01
-5.91037273e-01 7.25651920e-01 -3.02227408e-01 2.43400142e-01
-5.17848492e-01 1.30220190e-01 1.22141290e+00 -7.96015501e-01
1.89847842e-01 1.27019930e+00 5.88407636e-01 9.88664851e-02
-1.53123581e+00 -5.02761960e-01 5.28582990e-01 2.94764861e-02
-1.14125037e+00 -4.61443245e-01 6.77295864e-01 -8.32725286e-01
8.04691195e-01 3.45835805e-01 8.32542539e-01 5.88706672e-01
2.57133633e-01 3.32372248e-01 9.46692526e-01 -4.67407882e-01
4.85238761e-01 -2.68901195e-02 1.06329322e-01 4.87615794e-01
4.51513469e-01 5.36600292e-01 -2.64841259e-01 2.28522956e-01
1.03719461e+00 -9.52989683e-02 -3.88244241e-01 2.63911933e-01
-1.25681710e+00 9.76899803e-01 8.52474391e-01 4.66532499e-01
-3.71237367e-01 4.56400335e-01 3.01709864e-02 2.53687114e-01
8.31091821e-01 1.37126267e+00 -7.41447687e-01 4.11638707e-01
-1.00325990e+00 3.41608047e-01 8.37304413e-01 9.03977677e-02
7.26830065e-01 4.51584071e-01 4.56940502e-01 6.97175860e-01
6.24525428e-01 3.68290603e-01 2.72584677e-01 -1.20493639e+00
2.25202546e-01 5.74893355e-01 -1.19763412e-01 -1.43817008e+00
-3.96534264e-01 -6.14091992e-01 -8.02490532e-01 6.06323063e-01
6.73752606e-01 -5.34422159e-01 -9.21141088e-01 1.76180911e+00
-1.18046544e-01 6.41318187e-02 1.86079293e-01 1.05670238e+00
2.66622990e-01 1.00134420e+00 9.68756154e-02 3.28658789e-01
9.45065558e-01 -4.58871096e-01 -2.69082546e-01 -7.17457891e-01
4.44385111e-01 -3.11180830e-01 7.23473668e-01 2.55871266e-01
-6.80898249e-01 -3.13026816e-01 -1.10968709e+00 3.10940713e-01
-9.28402364e-01 -1.45139053e-01 9.43982124e-01 2.27071986e-01
-1.02831602e+00 1.01690984e+00 -8.66134822e-01 -4.30134416e-01
3.08311582e-01 4.23142016e-01 6.43832088e-02 4.15174812e-01
-1.19999743e+00 1.18494225e+00 7.79764831e-01 7.23026693e-01
-6.35741532e-01 -5.98622441e-01 -6.33131921e-01 2.57476658e-01
2.88312584e-01 -7.53025651e-01 1.09183395e+00 -1.31785762e+00
-1.23744345e+00 4.88697886e-01 -1.29546508e-01 -3.91286165e-01
3.44352089e-02 3.85158099e-02 -2.37563819e-01 -2.54647136e-01
-9.19026658e-02 7.58544147e-01 5.34796596e-01 -1.25507963e+00
-7.24186599e-01 -5.42121887e-01 -1.82138756e-01 2.81457961e-01
-3.72719973e-01 -1.94069088e-01 -4.09754850e-02 -9.52798724e-01
5.29676259e-01 -9.06913698e-01 -4.57608730e-01 4.27160233e-01
-1.98447391e-01 -4.87868600e-02 5.00136197e-01 -7.19805360e-01
9.60451066e-01 -1.78475285e+00 4.82558131e-01 6.60002053e-01
4.26383734e-01 2.04219490e-01 -1.80346513e-04 1.20073430e-01
-1.13867752e-01 6.91335857e-01 -6.26376092e-01 1.19530551e-01
-3.98532391e-01 4.57015514e-01 -6.58780277e-01 3.86025190e-01
4.86018300e-01 7.15651333e-01 -1.05181432e+00 1.54619768e-01
3.38247359e-01 3.38283241e-01 -2.60855377e-01 1.73185870e-01
-3.76982808e-01 4.53689635e-01 -5.45867383e-01 3.48778427e-01
1.13308340e-01 -2.13704407e-01 2.54318893e-01 2.89783239e-01
-4.68252718e-01 2.13773683e-01 -9.67545152e-01 1.07135308e+00
-3.95344138e-01 1.14316463e+00 -5.34104109e-02 -1.35433292e+00
9.20704842e-01 2.97078669e-01 -1.29225431e-02 -3.81258398e-01
6.36560842e-02 4.21550870e-01 4.79502052e-01 -4.10105884e-01
2.60860473e-01 -4.68895316e-01 3.96326929e-01 6.95484221e-01
-1.50383130e-01 -1.09868571e-01 -5.05591603e-03 -3.33048612e-01
4.59060520e-01 1.43128678e-01 2.08712071e-01 -5.90941012e-01
1.74709842e-01 3.12515736e-01 3.90731484e-01 9.35405970e-01
2.98025578e-01 4.71012384e-01 3.92361760e-01 -7.54821599e-01
-1.13047194e+00 -8.55544031e-01 -2.23641470e-01 1.14524651e+00
-3.00261229e-01 3.53209794e-01 -3.65133822e-01 -2.85834759e-01
2.26047903e-01 9.69945431e-01 -8.97037566e-01 -2.26180360e-01
-3.76290560e-01 -1.05521131e+00 5.90978444e-01 6.33390784e-01
1.89798057e-01 -1.11225879e+00 -3.86827618e-01 3.54145199e-01
3.35737437e-01 -5.56437731e-01 2.31213644e-01 5.33882678e-01
-1.44645882e+00 -9.31526065e-01 -6.21762574e-01 -6.16332889e-01
1.03714705e+00 2.55206078e-01 8.41394901e-01 1.71781391e-01
2.57419676e-01 -1.89119041e-01 2.58965306e-02 -6.89746201e-01
-4.78578061e-01 2.38405779e-01 2.08614871e-01 -1.70603752e-01
2.71786630e-01 -6.98614478e-01 -5.25973737e-01 2.60131449e-01
-9.77770805e-01 3.47889930e-01 5.92779636e-01 6.62744999e-01
1.88137874e-01 3.51909027e-02 7.22131491e-01 -8.21061313e-01
7.74193645e-01 -9.30129647e-01 -7.11108088e-01 2.28922665e-01
-1.08201349e+00 5.17121673e-01 8.13051105e-01 -1.59792855e-01
-9.31260586e-01 -2.15843797e-01 1.10905647e-01 -1.34391651e-01
-2.93356121e-01 1.35282815e+00 3.36240470e-01 1.17007615e-02
8.16078603e-01 1.29164532e-01 1.36988476e-01 -6.99144661e-01
3.73520285e-01 4.54051614e-01 5.43664753e-01 -6.89995825e-01
7.22078800e-01 2.18106434e-01 1.63630828e-01 -9.94351685e-01
-7.02483654e-01 -1.90654084e-01 -5.47511101e-01 -3.85994256e-01
8.48884225e-01 -7.51927555e-01 -4.14272040e-01 3.41682732e-01
-1.16978085e+00 -3.72378170e-01 -1.61110252e-01 7.06891835e-01
-9.52473655e-02 -2.45269746e-01 -1.56210586e-01 -8.32138300e-01
-1.43683016e-01 -1.07321489e+00 3.61878663e-01 3.07342738e-01
-5.74724436e-01 -1.70651472e+00 1.55247785e-02 -1.19402558e-01
5.10069489e-01 2.93004781e-01 1.13847697e+00 -6.73295736e-01
-4.31508958e-01 -1.08078483e-03 -2.83053815e-01 4.83730376e-01
3.20529878e-01 3.44276100e-01 -1.25007045e+00 1.39766499e-01
-3.86127457e-02 1.55373320e-01 1.17105687e+00 6.53270066e-01
1.19321418e+00 -5.27550817e-01 -2.08319306e-01 7.71092653e-01
1.32371414e+00 4.06896174e-01 3.27168882e-01 5.28786302e-01
6.91709518e-01 1.10569942e+00 -2.21123978e-01 -1.63064733e-01
8.46339613e-02 1.97460979e-01 3.68567765e-01 -4.57541049e-01
2.49987602e-01 -2.07824573e-01 1.68642834e-01 6.65699720e-01
-5.12483478e-01 -1.86310247e-01 -1.37570751e+00 5.89817584e-01
-2.00865507e+00 -9.06768680e-01 3.49088088e-02 2.13027954e+00
7.20271409e-01 1.10307030e-01 -3.85090709e-01 1.61993757e-01
5.70601463e-01 1.50678948e-01 -8.71152520e-01 -5.92052102e-01
-2.04811528e-01 4.70212586e-02 7.30546474e-01 7.14418411e-01
-9.24331188e-01 8.37733448e-01 7.17484570e+00 2.27321163e-01
-1.37192667e+00 -5.85914612e-01 1.03305423e+00 3.05621058e-01
-4.33089226e-01 2.23288476e-01 -4.73757774e-01 1.97435260e-01
9.56612349e-01 -1.17011115e-01 6.45408869e-01 5.98071635e-01
7.67512619e-01 -1.60943702e-01 -1.29014480e+00 2.51800835e-01
-3.39413762e-01 -1.59032202e+00 1.41991004e-01 1.54292196e-01
7.10426331e-01 1.43770263e-01 1.71761915e-01 -1.45226315e-01
6.16331995e-01 -1.38728237e+00 6.47576571e-01 6.76862419e-01
4.68355089e-01 -5.08695245e-01 5.13128638e-01 3.36775243e-01
-9.08842683e-01 4.29800898e-02 -2.14121833e-01 -7.15367615e-01
-2.89403439e-01 6.63403869e-01 -1.21278572e+00 3.04207974e-03
4.91528690e-01 9.27537978e-01 -4.70281124e-01 9.13807571e-01
-5.73682964e-01 1.09752083e+00 -5.25943041e-01 -2.08944842e-01
4.17408258e-01 -3.09176803e-01 6.17666483e-01 9.41458941e-01
3.15982312e-01 -9.72787589e-02 -3.33242536e-01 1.42997110e+00
9.91479605e-02 -1.03245072e-01 -9.20899272e-01 -4.75134283e-01
3.52476001e-01 7.65589416e-01 -8.78473520e-01 -1.59548953e-01
-1.79142550e-01 2.69037753e-01 1.82860866e-01 8.04109037e-01
-8.78958181e-02 -4.29670781e-01 9.10687804e-01 -7.30513707e-02
3.03858873e-02 -2.96324760e-01 -1.05692661e+00 -1.01399136e+00
-3.09866607e-01 -6.05563402e-01 1.01272993e-01 -7.73255765e-01
-1.11009812e+00 2.23698452e-01 1.98116601e-02 -6.73062086e-01
-3.13164622e-01 -9.54394221e-01 -7.70450294e-01 1.25523961e+00
-1.52559710e+00 -6.60980284e-01 6.32177517e-02 -2.98059255e-01
3.33935767e-01 -1.96175158e-01 9.02311862e-01 -3.97695541e-01
-7.62290895e-01 -9.83257964e-02 6.34842157e-01 2.12464422e-01
1.98403373e-01 -1.25790560e+00 7.45866954e-01 9.77985919e-01
1.16981037e-01 8.17207634e-01 8.94336879e-01 -6.06327772e-01
-1.17281115e+00 -1.05133188e+00 7.41218746e-01 -2.61037499e-01
8.02976489e-01 -9.64949746e-03 -1.14486003e+00 8.06844294e-01
-1.78107843e-01 -3.92901003e-01 6.16338372e-01 2.90927142e-01
-1.18968122e-01 -1.38188079e-01 -6.44653320e-01 8.92784238e-01
4.50203508e-01 -3.59055728e-01 -7.55175114e-01 -4.19821665e-02
5.90745986e-01 8.09538215e-02 -8.35499525e-01 3.13502938e-01
6.75156415e-01 -4.22167391e-01 9.74116623e-01 -8.43337417e-01
7.44209826e-01 -6.04527295e-01 -5.90052875e-03 -1.69282019e+00
-3.86372834e-01 -1.15887158e-01 1.22820683e-01 8.88986945e-01
1.02001476e+00 -1.00068283e+00 6.45456553e-01 8.02068710e-01
-9.13355500e-02 -6.89244866e-01 -4.61201251e-01 -3.95222723e-01
4.83132780e-01 -5.95950603e-01 6.17493987e-01 1.16853952e+00
-1.10240638e-01 2.77367681e-01 -1.10462531e-01 3.45372826e-01
4.66600180e-01 1.52162313e-02 3.21278840e-01 -1.65606797e+00
-2.60611415e-01 -1.02395260e+00 -9.67770368e-02 -8.61573100e-01
4.36493978e-02 -1.03858352e+00 2.09316701e-01 -1.45958483e+00
-3.42858404e-01 -5.93712986e-01 -3.18772018e-01 6.49660766e-01
-2.93992788e-01 -1.12986460e-01 1.02812476e-01 3.56851995e-01
4.88240987e-01 2.61239290e-01 1.16840065e+00 -1.02138512e-01
-3.70657653e-01 7.12355673e-02 -1.15399671e+00 9.02065337e-01
1.06711960e+00 -5.27121425e-01 -2.85775065e-01 -8.43579531e-01
6.35353923e-01 -2.63382822e-01 5.20464838e-01 -7.98957348e-01
1.96940929e-01 -7.47518659e-01 7.51788557e-01 1.39952168e-01
-6.82518333e-02 -7.65337586e-01 2.34492362e-01 4.40386683e-01
-7.57688582e-01 -7.83460364e-02 4.23227906e-01 4.54478592e-01
-1.51905432e-01 -2.16394559e-01 4.80128706e-01 -3.33248705e-01
-8.28698397e-01 -2.50754948e-03 -8.32580328e-01 -4.08983380e-02
6.65739954e-01 -2.19468981e-01 -2.49476597e-01 -2.64958143e-01
-6.95602715e-01 3.77074450e-01 3.15709502e-01 3.65455240e-01
6.41527057e-01 -9.15806472e-01 -8.00592542e-01 1.90805718e-01
-1.36334985e-01 2.09978774e-01 -4.04747903e-01 4.17753249e-01
-7.23928809e-01 5.13696909e-01 -6.49053678e-02 -4.65633124e-01
-6.44104838e-01 -2.04810575e-02 7.38320410e-01 1.42907307e-01
-3.84093374e-01 6.79789782e-01 3.33200634e-01 -4.22338128e-01
4.52364720e-02 -5.13187706e-01 -4.23044264e-01 1.68720961e-01
3.92839938e-01 2.99962103e-01 -1.37710214e-01 -3.95186514e-01
-8.29207823e-02 4.40879285e-01 3.11335534e-01 -3.32079411e-01
1.84699380e+00 1.54028744e-01 -2.14363277e-01 8.41266632e-01
9.84279811e-01 -6.93650186e-01 -1.23184407e+00 -1.47485420e-01
1.74206153e-01 -2.86831856e-01 6.16946101e-01 -8.71152520e-01
-1.26678920e+00 1.09475005e+00 3.37145567e-01 2.31625959e-01
9.99695778e-01 -2.31697485e-01 2.18151450e-01 8.00236940e-01
-3.53228539e-01 -1.10301971e+00 -4.15975153e-01 5.56087136e-01
9.96680319e-01 -1.28565562e+00 -4.87226471e-02 2.65989333e-01
-2.01781034e-01 1.54739368e+00 4.01438892e-01 -8.59587044e-02
8.25652003e-01 4.18928787e-02 9.66101661e-02 -2.92420775e-01
-6.10216022e-01 -1.65278435e-01 5.39945126e-01 3.38607222e-01
6.00076556e-01 1.87399849e-01 7.05642402e-02 8.06982964e-02
-1.45145163e-01 1.39232939e-02 2.54038811e-01 7.45706022e-01
-7.56363630e-01 -7.79780269e-01 -6.04181468e-01 6.91371799e-01
-3.93786788e-01 -3.92724991e-01 -5.47329664e-01 6.25340521e-01
-9.95495543e-02 7.98953593e-01 1.51942670e-01 -4.20645207e-01
4.00885083e-02 6.38043806e-02 -9.96081978e-02 -7.35520422e-01
-4.77822185e-01 -2.35268101e-01 6.57741055e-02 -5.10546230e-02
-3.93237412e-01 -7.07944870e-01 -1.17516983e+00 -3.64100635e-01
-2.82719433e-01 2.23417297e-01 1.05449700e+00 1.18300867e+00
1.80831075e-01 5.66154659e-01 5.07517576e-01 -1.13695359e+00
-3.61797124e-01 -8.40920448e-01 -3.77751261e-01 -8.27571601e-02
5.14942527e-01 -3.56287122e-01 -6.46190643e-01 1.57611638e-01] | [6.8022260665893555, 2.9716856479644775] |
b0f2a81b-2492-4637-b1db-49609bda8c82 | the-cat-set-on-the-mat-cross-attention-for | 2111.00243 | null | https://arxiv.org/abs/2111.00243v1 | https://arxiv.org/pdf/2111.00243v1.pdf | The CAT SET on the MAT: Cross Attention for Set Matching in Bipartite Hypergraphs | Usual relations between entities could be captured using graphs; but those of a higher-order -- more so between two different types of entities (which we term "left" and "right") -- calls for a "bipartite hypergraph". For example, given a left set of symptoms and right set of diseases, the relation between a set subset of symptoms (that a patient experiences at a given point of time) and a subset of diseases (that he/she might be diagnosed with) could be well-represented using a bipartite hyperedge. The state-of-the-art in embedding nodes of a hypergraph is based on learning the self-attention structure between node-pairs from a hyperedge. In the present work, given a bipartite hypergraph, we aim at capturing relations between node pairs from the cross-product between the left and right hyperedges, and term it a "cross-attention" (CAT) based model. More precisely, we pose "bipartite hyperedge link prediction" as a set-matching (SETMAT) problem and propose a novel neural network architecture called CATSETMAT for the same. We perform extensive experiments on multiple bipartite hypergraph datasets to show the superior performance of CATSETMAT, which we compare with multiple techniques from the state-of-the-art. Our results also elucidate information flow in self- and cross-attention scenarios. | ['M. Narasimha Murty', 'V. Susheela Devi', 'Swyam Prakash Singh', 'Govind Sharma'] | 2021-10-30 | null | null | null | null | ['set-matching'] | ['computer-vision'] | [ 2.16989979e-01 7.78181851e-01 -3.19429070e-01 -4.08743948e-01
-1.44638792e-01 -3.88367474e-01 4.58858192e-01 5.56181669e-01
1.05579160e-01 6.02295220e-01 3.39780301e-01 -3.91396552e-01
-7.43385553e-01 -1.37445045e+00 -9.98469234e-01 -4.23089385e-01
-4.33118284e-01 7.82622635e-01 -2.20616870e-02 -4.53410000e-01
-4.28454190e-01 2.72127241e-01 -1.00148499e+00 4.39666897e-01
5.67920089e-01 7.13055968e-01 -1.39224499e-01 2.33660698e-01
-1.45389721e-01 6.35598898e-01 -2.87220806e-01 -1.04513538e+00
1.75372213e-01 -3.91946673e-01 -1.32546842e+00 -4.55692261e-02
5.13117671e-01 1.69584319e-01 -1.10540390e+00 1.12297523e+00
1.47699475e-01 -1.44104868e-01 7.56946564e-01 -1.81444728e+00
-1.15957618e+00 1.10606766e+00 -5.18613398e-01 1.69566199e-01
5.08770466e-01 -1.64100930e-01 1.64663982e+00 -3.25358033e-01
8.44775915e-01 1.29660010e+00 8.21384192e-01 4.58333760e-01
-1.46004164e+00 -4.51318592e-01 2.35663876e-01 4.05596942e-01
-1.29527807e+00 1.14561729e-01 7.90310025e-01 -5.43359876e-01
9.39377487e-01 4.50425148e-01 7.97085822e-01 9.74968016e-01
2.31493652e-01 6.13274097e-01 7.12526381e-01 -1.96959183e-01
-2.74148077e-01 3.60592008e-01 6.26311302e-01 9.87081766e-01
6.96666181e-01 1.41347304e-01 -1.38438940e-01 -1.20797195e-01
5.24169624e-01 2.72037089e-01 -6.43472254e-01 -5.97448051e-01
-1.04989684e+00 7.61454284e-01 1.24699807e+00 7.03701556e-01
-3.86728108e-01 1.42874613e-01 4.12462920e-01 5.18096387e-01
2.62261868e-01 4.07580018e-01 -5.71517050e-01 8.95379722e-01
-4.52381730e-01 -2.05675840e-01 1.20082104e+00 1.07672298e+00
6.86902463e-01 -3.25280964e-01 -1.89685062e-01 4.64275390e-01
3.08618635e-01 1.74374878e-01 2.53649235e-01 -5.86937517e-02
6.21506035e-01 1.21033621e+00 -2.50851363e-01 -1.52163398e+00
-7.57425427e-01 -5.15215039e-01 -1.39110804e+00 -3.94698650e-01
1.89948425e-01 4.15090024e-02 -8.98156106e-01 2.10069513e+00
2.52313614e-01 3.93030256e-01 6.15821332e-02 7.42085338e-01
1.34897578e+00 5.13334155e-01 -1.20259926e-01 -8.61351863e-02
1.76377201e+00 -8.02575648e-01 -5.92789590e-01 -1.53665051e-01
6.58459723e-01 2.02321067e-01 5.20395398e-01 -1.60586253e-01
-9.83496785e-01 -3.92738789e-01 -8.41875553e-01 -3.04509159e-02
-1.06929505e+00 -3.01693201e-01 7.10165441e-01 3.15935403e-01
-1.10021722e+00 8.97364974e-01 -4.34450835e-01 -7.08515942e-01
2.32381612e-01 5.55249393e-01 -7.73573935e-01 5.74677512e-02
-1.79977953e+00 8.94487023e-01 5.60285807e-01 1.64563194e-01
-5.45874417e-01 -9.10066009e-01 -1.04189193e+00 6.14558518e-01
6.18038297e-01 -1.29563248e+00 5.30556440e-01 -8.54050457e-01
-4.53038782e-01 1.32104897e+00 1.55817538e-01 -4.76440310e-01
9.18698832e-02 1.24408789e-01 -9.40361321e-01 2.63554323e-03
2.51910836e-02 4.55754101e-01 4.20426339e-01 -1.40239179e+00
-4.01343554e-01 -5.72154343e-01 4.53991294e-01 -4.20043282e-02
-4.55630153e-01 -1.97306529e-01 -6.26261950e-01 -2.88322270e-01
-6.03295816e-03 -1.01354384e+00 -7.90550839e-03 -2.47811660e-01
-9.70239401e-01 -2.60613948e-01 6.43387735e-01 -5.41178524e-01
1.71347034e+00 -1.74132693e+00 4.55616236e-01 5.20720243e-01
8.88001382e-01 2.34731033e-01 -2.05700055e-01 8.41792405e-01
-8.49517167e-01 3.53404552e-01 -2.57758409e-01 6.37594759e-02
1.02721795e-01 4.84159648e-01 1.59603544e-02 3.73450011e-01
2.75770634e-01 1.14030623e+00 -1.10307467e+00 -4.52257901e-01
-2.13491693e-01 5.02272069e-01 -3.90017450e-01 8.89151245e-02
6.75313547e-02 2.74434295e-02 -4.03083265e-01 3.09698164e-01
5.94523549e-01 -1.00203764e+00 6.28629088e-01 -7.71664500e-01
5.00602007e-01 1.35550603e-01 -1.07983446e+00 1.25719762e+00
-3.41220140e-01 4.93463546e-01 -1.59821987e-01 -1.16515076e+00
6.88823283e-01 4.72205400e-01 6.10414207e-01 -4.77530956e-01
7.18912855e-02 -1.63835898e-01 4.45397377e-01 -7.13149607e-01
3.02932352e-01 1.36984974e-01 8.63009971e-03 3.80985662e-02
3.36597621e-01 5.12803555e-01 3.18849832e-01 6.34730279e-01
1.33495092e+00 -2.59687483e-01 6.77375853e-01 -2.45950580e-01
5.67542136e-01 -1.75328925e-01 4.83711988e-01 6.49889350e-01
3.73245887e-02 1.35496065e-01 1.03337157e+00 -4.87646341e-01
-9.10378218e-01 -1.00458527e+00 -2.03461573e-01 7.66985595e-01
1.43528640e-01 -5.72962523e-01 -3.12617391e-01 -9.91816640e-01
5.36640167e-01 2.71572739e-01 -1.18686092e+00 -4.33183014e-01
-3.68680745e-01 -6.79196715e-01 3.74321699e-01 6.87500298e-01
1.61412284e-01 -9.32175219e-01 -2.45486759e-02 3.03713381e-02
-8.03629756e-02 -9.69114065e-01 -5.97741365e-01 2.94231713e-01
-5.74122548e-01 -1.50747657e+00 -5.37617445e-01 -9.82342958e-01
7.59505153e-01 -9.35677364e-02 1.76200664e+00 3.77313048e-01
-2.42974490e-01 3.55085254e-01 -2.84292579e-01 -1.15265191e-01
-1.15970559e-01 -2.00378671e-02 6.36271462e-02 4.01979744e-01
3.52569133e-01 -6.15451753e-01 -5.25325775e-01 2.05333903e-01
-1.04632044e+00 6.21661581e-02 7.29522169e-01 1.15753520e+00
3.47139001e-01 7.56482705e-02 5.53377271e-01 -1.63211536e+00
4.98539716e-01 -1.04528379e+00 -3.80890548e-01 7.26349115e-01
-6.75558329e-01 3.72148812e-01 6.68644071e-01 -2.08822623e-01
-6.28453434e-01 -3.09397727e-01 2.11125866e-01 -5.42559445e-01
-1.81777254e-02 8.53964031e-01 -3.99041355e-01 9.86091867e-02
2.60041595e-01 -5.70128411e-02 -4.53317672e-01 -3.16092372e-01
5.75055599e-01 3.01929981e-01 4.42821205e-01 -2.38714993e-01
7.03907073e-01 2.95879275e-01 6.07248187e-01 -3.47446859e-01
-8.01889777e-01 -6.11484587e-01 -6.50527656e-01 5.76604605e-02
9.09386933e-01 -4.90909666e-01 -1.16796398e+00 -6.01860620e-02
-1.12796474e+00 2.53872693e-01 -1.74143776e-01 2.55763352e-01
-1.96216211e-01 1.77165896e-01 -6.98218048e-01 -5.12205839e-01
-2.68195540e-01 -8.52480531e-01 9.02545571e-01 1.64776802e-01
-7.15189427e-02 -1.65839016e+00 3.49244416e-01 1.08193681e-01
-9.58546065e-03 3.72485340e-01 1.41183686e+00 -1.24047399e+00
-5.58705151e-01 -1.59736916e-01 -7.03425765e-01 -3.07758510e-01
8.81164744e-02 -2.52040118e-01 -5.57290912e-01 -3.41168016e-01
-5.56398988e-01 -2.49902923e-02 9.31985080e-01 2.00280800e-01
9.39783394e-01 -6.80728137e-01 -1.04556000e+00 6.27210855e-01
1.76197565e+00 -5.33290301e-03 5.25273740e-01 -4.81884144e-02
9.56568718e-01 8.83288145e-01 -1.52315535e-02 1.79991335e-01
7.64932632e-01 7.10767031e-01 5.57178020e-01 -4.42456305e-01
-1.00970991e-01 -4.47627127e-01 -1.67722240e-01 8.56342494e-01
-5.81570268e-02 -6.75569296e-01 -8.32191586e-01 8.58733118e-01
-1.99982750e+00 -1.04308999e+00 -5.50195277e-01 2.00543404e+00
7.84059107e-01 -1.35382771e-01 1.24112934e-01 -2.00329483e-01
9.36818659e-01 8.19883123e-02 -5.28220832e-01 -2.78400630e-01
-6.68472871e-02 2.08548993e-01 3.38472337e-01 4.00811046e-01
-1.14629948e+00 5.79677284e-01 5.17457199e+00 4.93312001e-01
-6.48116946e-01 -1.54945537e-01 5.29611707e-01 4.31466907e-01
-6.90966070e-01 2.15292107e-02 -6.67819083e-01 4.32289660e-01
9.35568273e-01 -3.46037328e-01 1.96315736e-01 4.28981304e-01
-4.65834916e-01 4.68854040e-01 -1.72795296e+00 7.70729423e-01
1.01763122e-01 -1.18127739e+00 1.13297701e-01 2.60991842e-01
7.37565100e-01 -5.28665967e-02 -7.30690658e-02 5.16397297e-01
6.34277582e-01 -9.87566352e-01 -3.70900370e-02 7.82875478e-01
6.14684165e-01 -2.73117602e-01 8.07664812e-01 -3.27369198e-02
-1.68785930e+00 -1.47391617e-01 -1.81357160e-01 6.04524851e-01
1.01301149e-01 4.96462464e-01 -7.65915871e-01 1.36479700e+00
6.09660447e-01 8.87086034e-01 -5.24274886e-01 1.20924723e+00
-1.19276196e-01 3.06276470e-01 -7.35709295e-02 -9.02475882e-03
2.56363064e-01 -2.93477803e-01 5.63228846e-01 1.26959968e+00
1.55894116e-01 2.23520860e-01 1.22728683e-01 1.06563759e+00
-5.96286833e-01 -8.80886093e-02 -1.10175097e+00 -1.35252655e-01
2.25900814e-01 1.38437152e+00 -4.46457833e-01 -4.53984916e-01
-5.57397962e-01 8.20127547e-01 6.59118295e-01 2.87618667e-01
-8.85209620e-01 -5.80778658e-01 4.88045543e-01 6.37237057e-02
1.94824442e-01 5.40342212e-01 8.50461200e-02 -1.20020938e+00
-1.75697967e-01 -5.15207291e-01 1.03841841e+00 -1.02851951e+00
-1.83160961e+00 8.34254742e-01 3.79220769e-02 -1.14946496e+00
7.42528588e-02 -7.11998463e-01 -6.22150421e-01 6.37338340e-01
-1.36850190e+00 -1.32557642e+00 -2.60295302e-01 7.28142083e-01
-2.13356450e-01 5.52126626e-03 9.39670324e-01 4.82347429e-01
-5.10467887e-01 6.96185648e-01 -5.53831132e-03 4.63963836e-01
3.36087167e-01 -1.69116545e+00 2.09379941e-01 4.76582527e-01
4.26563174e-01 6.12519264e-01 5.18772125e-01 -7.88396418e-01
-1.32570529e+00 -1.24413216e+00 1.34442830e+00 -5.00938475e-01
1.02750373e+00 -1.86154887e-01 -1.09002507e+00 1.24002981e+00
5.03543258e-01 8.39569718e-02 6.91201091e-01 5.34124494e-01
-5.79060674e-01 -1.56652763e-01 -1.00727010e+00 6.41270101e-01
1.39640069e+00 -6.15867138e-01 -7.04603136e-01 6.25545442e-01
1.02839673e+00 -6.09529577e-02 -1.33679414e+00 5.70954800e-01
3.81918997e-01 -6.37586474e-01 1.15137386e+00 -1.48231947e+00
4.14494336e-01 -4.61501256e-02 -6.96273579e-04 -1.74846649e+00
-9.08078372e-01 -3.94654423e-01 -3.19590211e-01 1.24549878e+00
7.23642945e-01 -6.35205686e-01 7.99399197e-01 4.25231487e-01
-5.98606504e-02 -9.44532275e-01 -5.56776464e-01 -6.81564629e-01
-1.59368128e-01 9.82427970e-02 8.53517354e-01 1.57634377e+00
5.72888076e-01 7.01350212e-01 -4.44489390e-01 5.04448116e-01
5.23530602e-01 3.70755523e-01 2.56215841e-01 -1.50690079e+00
-4.13230866e-01 -5.32146573e-01 -6.97580040e-01 -6.18897021e-01
3.03759396e-01 -1.43580294e+00 -4.93921697e-01 -1.86033249e+00
5.15718281e-01 -2.76188731e-01 -5.48111498e-01 7.20428705e-01
-2.91389465e-01 -2.45578941e-02 1.21785156e-01 -1.61872640e-01
-6.04077697e-01 4.53140408e-01 1.16055501e+00 -5.98935366e-01
-2.53692865e-02 -1.94824133e-02 -7.47706115e-01 5.45198143e-01
2.77662396e-01 -5.81593931e-01 -4.67820197e-01 -3.21216643e-01
6.93631411e-01 6.66934848e-01 4.34850186e-01 -5.03819108e-01
5.23575962e-01 2.38587588e-01 2.43444052e-02 -3.79302949e-01
3.80926311e-01 -1.30552399e+00 3.40733111e-01 5.71776330e-01
-5.51213503e-01 2.45284066e-01 -4.09238320e-03 8.40224922e-01
-2.55197465e-01 -1.51385456e-01 2.66617656e-01 -7.33729675e-02
-5.39774776e-01 5.64941108e-01 4.09889966e-01 2.31638163e-01
1.05943120e+00 -6.24078102e-02 -7.39232600e-01 -3.94865453e-01
-1.25533211e+00 5.67970991e-01 -7.20750913e-02 3.28682661e-01
6.59931242e-01 -1.47581732e+00 -7.85380363e-01 9.59949121e-02
5.70593119e-01 -4.43302929e-01 2.74720281e-01 1.12778747e+00
4.78900708e-02 5.66731036e-01 -1.25191301e-01 -2.69676775e-01
-1.33551550e+00 1.09310889e+00 4.85088706e-01 -8.44518840e-01
-5.71067512e-01 8.49163592e-01 6.27912939e-01 -5.67135751e-01
2.32489109e-01 -2.19174117e-01 -5.73778272e-01 5.20513803e-02
6.97034448e-02 2.75683403e-01 2.91222031e-03 -7.38388896e-01
-4.74828631e-01 3.86153162e-01 -1.68245479e-01 4.78126585e-01
1.32981753e+00 9.77745578e-02 -3.79878551e-01 3.02180767e-01
1.33541727e+00 -2.18448430e-01 -3.51427227e-01 -5.39133728e-01
-1.79621987e-02 -2.54478186e-01 -2.79042572e-01 -9.83862281e-01
-1.55176044e+00 6.46819770e-01 2.62977064e-01 9.12511110e-01
1.09490561e+00 4.81724739e-01 6.75089061e-01 4.46480602e-01
9.90462005e-02 -4.35066670e-01 -3.42908859e-01 2.26072386e-01
9.31463122e-01 -1.22523665e+00 -3.43949437e-01 -6.09349489e-01
-6.90259933e-01 9.18587625e-01 6.49300277e-01 -8.46685320e-02
1.06673694e+00 4.97550964e-02 -5.30214906e-01 -7.95844316e-01
-9.48223352e-01 -4.62227792e-01 9.35167432e-01 5.62727571e-01
3.70775014e-01 2.77943134e-01 -2.39295423e-01 6.13631666e-01
1.11366481e-01 -1.21699639e-01 2.61674583e-01 3.36139262e-01
1.28608882e-01 -9.84246433e-01 9.82285850e-03 7.50726283e-01
-2.34184369e-01 -3.67075145e-01 -8.79593372e-01 1.01904249e+00
3.25668007e-01 8.21380019e-01 3.09522599e-01 -6.81856751e-01
5.14201939e-01 -9.70289633e-02 4.81394976e-01 -5.20752490e-01
-1.00480604e+00 -3.73524636e-01 3.99112016e-01 -5.35680473e-01
-5.13493180e-01 -2.99739122e-01 -1.06433845e+00 -4.39926296e-01
-4.82232481e-01 2.10403010e-01 2.99995262e-02 7.48425901e-01
5.30767322e-01 9.40863013e-01 4.97699589e-01 -6.74568638e-02
-1.27913192e-01 -8.05543065e-01 -9.44870770e-01 1.13184822e+00
2.21413106e-01 -5.32864153e-01 -1.21331982e-01 -1.31199077e-01] | [7.355543613433838, 6.530724048614502] |
b63d89d4-b7e4-4492-bad6-447d5edba9f7 | deep-sparse-and-low-rank-prior-for | null | null | https://ieeexplore.ieee.org/document/9884071 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9884071 | Deep Sparse and Low-Rank Prior for Hyperspectral Image Denoising | Spectral and spatial correlation in hyperspectral images (HSIs) can be exploited in HSI processing because it directly induces a sparse and low-rank prior via linear transformations. Researchers have used the sparse and low-rank prior as an image prior for HSI restoration, such as denoising, deblurring, and super-resolution. This paper proposes a HSI denoising method that incorporates a sparse and low-rank prior with a deep image prior (DIP). The sparse and low-rank prior is obtained using the 2-dimensional discrete wavelet transform (2-D DWT), and singular value decomposition (SVD), while the DIP is provided by the structure of a convolutional neural network (CNN). The combination of a sparse and low-rank prior with a DIP views the CNN-based denoising method similar to a model-based method, inheriting the advantages of both model-based and CNN-based methods. Experimental results with simulated and real HSI datasets show that the proposed method outperforms the conventional sparse and low-rank based methods in both quantitative and qualitative performance. Codes are available at https://github.com/hvn2/DIP-SLR | ['Han V. Nguyen; Magnus O. Ulfarsson; Jakob Sigurdsson; Johannes R. Sveinsson'] | 2022-09-28 | null | null | null | ieee-international-geoscience-and-remote-1 | ['deblurring'] | ['computer-vision'] | [ 4.30657476e-01 -6.23446941e-01 1.73490852e-01 -1.99018940e-01
-7.09181905e-01 -3.04420710e-01 2.33211502e-01 -4.78878886e-01
-7.00522885e-02 6.26854181e-01 7.26939440e-01 2.69448459e-01
-3.90459985e-01 -9.05928195e-01 -4.49181587e-01 -1.20973575e+00
-2.73300931e-02 -3.46108109e-01 9.73107740e-02 -2.46645898e-01
-4.03229259e-02 4.15046811e-01 -1.52198696e+00 3.43207598e-01
1.12359977e+00 1.03726792e+00 4.41781193e-01 3.30915183e-01
3.73580486e-01 6.68473005e-01 5.15480600e-02 3.30001354e-01
5.06131947e-01 -5.35884738e-01 -2.51699567e-01 2.06364080e-01
2.66324610e-01 -4.40160304e-01 -9.60202992e-01 1.47889888e+00
6.05957568e-01 4.59998161e-01 4.07360673e-01 -8.77139270e-01
-9.35918808e-01 1.32727042e-01 -1.00808275e+00 1.46160036e-01
2.94228524e-01 3.30029465e-02 5.21687567e-01 -1.12600207e+00
1.91569254e-01 1.12941360e+00 1.05202579e+00 -2.96440750e-01
-1.12338245e+00 -5.84687293e-01 -4.52069879e-01 4.65371430e-01
-1.65701985e+00 -4.62585390e-01 1.16598117e+00 -2.99470097e-01
4.90321040e-01 2.66137719e-01 5.92326045e-01 5.71116149e-01
5.77430613e-02 4.78679806e-01 1.37782109e+00 -4.30877179e-01
-3.48210633e-02 -5.80852211e-01 -6.41391845e-03 5.77661693e-01
4.95363593e-01 5.17311573e-01 -5.26305437e-01 -1.28263488e-01
1.04657722e+00 4.23809707e-01 -7.52521276e-01 1.47774190e-01
-1.03530550e+00 6.82972252e-01 8.71333361e-01 5.89564979e-01
-7.57460117e-01 -1.50686800e-01 1.13058932e-01 7.21102655e-02
5.70796609e-01 3.86945461e-03 1.21196873e-01 6.10560417e-01
-1.38149333e+00 -2.82833606e-01 5.24636745e-01 6.33417904e-01
9.82376993e-01 4.25939113e-01 -3.19752425e-01 1.01754797e+00
4.76583540e-01 6.91207886e-01 5.08849144e-01 -1.13719249e+00
-4.10663672e-02 1.48749858e-01 1.16265312e-01 -1.51226783e+00
-5.11064604e-02 -6.04750276e-01 -1.75857687e+00 1.02052964e-01
-2.22243652e-01 7.44985715e-02 -8.03825915e-01 1.42320633e+00
3.89777720e-02 7.99125969e-01 2.47550771e-01 1.24103558e+00
1.06628108e+00 1.13019741e+00 -3.68455589e-01 -4.11654115e-01
1.18692935e+00 -9.06886220e-01 -1.11749387e+00 -2.01327011e-01
-1.84897393e-01 -9.94312406e-01 6.80020988e-01 4.11018848e-01
-9.31207180e-01 -7.62372613e-01 -1.04225552e+00 -8.77674222e-02
1.97615800e-03 4.89909172e-01 3.47330838e-01 2.97220856e-01
-1.07924640e+00 5.50551414e-01 -7.65718222e-01 -3.26994121e-01
3.57357889e-01 -5.55370525e-02 -4.85081494e-01 -8.19473803e-01
-1.15363991e+00 5.94495058e-01 2.78414518e-01 6.53911889e-01
-9.32243228e-01 -6.25124514e-01 -9.05951023e-01 1.15054503e-01
4.25825007e-02 -4.81215268e-01 4.39489573e-01 -7.12015748e-01
-1.29578876e+00 4.02124643e-01 -5.92923105e-01 7.32349232e-02
-4.70016785e-02 -1.87688068e-01 -5.49237370e-01 5.33262253e-01
1.98084742e-01 3.49195227e-02 1.08648682e+00 -1.46045673e+00
-1.66824460e-01 -2.93529660e-01 -4.55630660e-01 2.27098748e-01
-3.74552459e-01 -2.09344476e-01 -1.61226302e-01 -7.56218135e-01
8.63588095e-01 -4.54729110e-01 -1.84577003e-01 1.32680729e-01
-3.65083128e-01 4.64591950e-01 1.08256721e+00 -1.27919567e+00
8.99548113e-01 -2.49593616e+00 1.42756686e-01 1.55313283e-01
1.44860730e-01 3.61268073e-01 -4.07950580e-01 7.19347060e-01
-3.85069251e-01 -1.88393027e-01 -6.86825395e-01 1.90822594e-02
-4.57240433e-01 2.42150679e-01 -9.19170603e-02 9.19265807e-01
-1.88314337e-02 5.10917544e-01 -8.98644030e-01 -9.16907713e-02
4.38958198e-01 9.59835947e-01 -2.06592143e-01 4.27790552e-01
4.87610906e-01 6.94757581e-01 -2.67858028e-01 7.87852943e-01
1.26036954e+00 -1.52102903e-01 -1.38181239e-01 -1.16423023e+00
-6.19447887e-01 -4.37777877e-01 -1.46773612e+00 1.50599575e+00
-1.86480239e-01 5.02388060e-01 7.79700398e-01 -1.08178818e+00
1.04179430e+00 4.16905016e-01 4.08916414e-01 -5.10780752e-01
-6.68840110e-02 3.85651797e-01 -4.46734786e-01 -7.02566266e-01
1.95897102e-01 -4.39391047e-01 7.34520137e-01 -4.40678634e-02
-4.80854735e-02 -2.05718577e-01 8.39548558e-02 9.11801010e-02
8.59921932e-01 -9.08065811e-02 1.40144557e-01 -4.55779374e-01
8.39154601e-01 -1.72952324e-01 6.48704171e-01 5.94155610e-01
7.22255111e-02 9.66925025e-01 -2.20241435e-02 -1.89447820e-01
-1.03436077e+00 -6.34755731e-01 -2.29206353e-01 6.23440266e-01
2.79447019e-01 1.83525950e-01 -1.87323064e-01 3.41866016e-01
-1.54436171e-01 5.06490111e-01 -3.37281317e-01 -1.53821170e-01
-1.65079802e-01 -8.78427744e-01 2.92619735e-01 1.94571033e-01
1.26959670e+00 -6.18875861e-01 3.09960954e-02 1.49804220e-01
-6.41505957e-01 -1.08127677e+00 -2.37712681e-01 1.42679781e-01
-9.20275748e-01 -9.38830495e-01 -8.07204247e-01 -8.44067156e-01
7.06587195e-01 1.02544618e+00 5.94278753e-01 5.44721149e-02
-2.15363204e-01 3.53004277e-01 -6.36212468e-01 1.48893595e-01
6.17615990e-02 -5.04641950e-01 -1.12051703e-01 4.49743718e-01
-1.09927326e-01 -1.16622233e+00 -7.77959228e-01 2.14278758e-01
-1.41629326e+00 1.41433164e-01 8.57114375e-01 1.10805798e+00
6.03241503e-01 7.90411592e-01 8.36980119e-02 -4.48626608e-01
4.83804882e-01 -4.20763761e-01 -6.85835421e-01 3.91967176e-03
-4.09677029e-01 -3.93052042e-01 5.19211769e-01 -8.35103244e-02
-1.42776310e+00 1.67313471e-01 -1.10912368e-01 -5.69707274e-01
-2.62865752e-01 1.15589678e+00 -2.85220176e-01 -5.49329162e-01
5.93333483e-01 8.93281043e-01 1.71007246e-01 -7.32142687e-01
1.84292078e-01 5.08882403e-01 8.15301180e-01 -2.92500168e-01
1.46904361e+00 8.61977518e-01 2.20249787e-01 -1.33440077e+00
-9.85710680e-01 -7.25981891e-01 -6.81879997e-01 -2.45585546e-01
6.61879241e-01 -1.52067208e+00 -3.27902019e-01 8.17133486e-01
-1.01420355e+00 -1.53032467e-01 -2.20617175e-01 7.41132379e-01
-1.47499084e-01 9.00887907e-01 -7.61827290e-01 -5.29576540e-01
-3.25027823e-01 -7.64086068e-01 7.87538290e-01 4.87977743e-01
5.81854105e-01 -8.71918738e-01 -1.28193289e-01 3.72537076e-01
9.74079788e-01 4.96422619e-01 6.49699092e-01 9.05131027e-02
-4.77982908e-01 -3.54042917e-01 -6.13898218e-01 6.82293892e-01
3.20416987e-01 -2.97076434e-01 -9.54018712e-01 -6.08602405e-01
4.49575990e-01 -2.19350696e-01 1.02858937e+00 8.28835487e-01
1.01390302e+00 -4.21772629e-01 1.44548059e-01 1.16242743e+00
2.13784623e+00 -1.31825656e-01 1.00821042e+00 2.06882134e-01
8.37077618e-01 2.87245601e-01 2.98653513e-01 6.40504301e-01
-6.16283640e-02 1.35376036e-01 6.66800380e-01 -6.43986046e-01
-3.41821015e-01 -1.95754860e-02 2.45926321e-01 1.04486144e+00
-4.37864214e-01 -2.31067725e-02 -6.95000887e-01 5.92276990e-01
-1.78958607e+00 -1.14247668e+00 -6.36257648e-01 2.01543665e+00
7.09134638e-01 -6.29937410e-01 -5.61476171e-01 3.17231327e-01
9.00063396e-01 6.04433358e-01 -4.16732132e-01 5.15551388e-01
-6.53871417e-01 1.15871027e-01 6.32975876e-01 6.63181305e-01
-9.73752797e-01 6.73470795e-01 5.52194881e+00 7.47659922e-01
-1.00833344e+00 1.67122304e-01 3.41191113e-01 5.02348959e-01
-2.43850991e-01 5.88663034e-02 -2.07066923e-01 1.74629688e-01
4.27152365e-01 -1.51128784e-01 8.62410843e-01 3.29098225e-01
8.02449346e-01 -2.29590803e-01 -4.40268397e-01 1.12693226e+00
1.29118174e-01 -1.18140829e+00 3.00354753e-02 -6.97908700e-02
9.71326888e-01 5.90171516e-02 1.31224424e-01 -1.41121134e-01
9.80246514e-02 -9.05238032e-01 4.18402433e-01 8.86408687e-01
8.60887229e-01 -5.61395228e-01 1.01572502e+00 2.49283224e-01
-1.45349431e+00 -1.07655831e-01 -7.87549138e-01 -4.38154824e-02
4.78173494e-02 1.32705557e+00 2.03087434e-01 1.05626619e+00
8.54914427e-01 1.55563998e+00 -1.60207689e-01 1.30819976e+00
-4.86715943e-01 8.64910424e-01 -1.97467744e-01 8.48452270e-01
2.88307309e-01 -8.37298691e-01 8.26129317e-01 1.24653113e+00
8.22817802e-01 9.15625930e-01 3.33217770e-01 9.18433070e-01
8.66630748e-02 -2.71088421e-01 -5.77000201e-01 -5.45549504e-02
3.88323814e-01 1.48073661e+00 -3.28785121e-01 -1.71784729e-01
-4.38611835e-01 8.11527967e-01 -6.39215708e-01 7.75267422e-01
-3.63193542e-01 -4.52484965e-01 4.01981086e-01 2.70879176e-02
4.25351769e-01 -4.64640707e-01 -1.42267153e-01 -1.24813998e+00
-1.11143030e-01 -9.89452004e-01 2.31585205e-01 -1.36423695e+00
-1.56959510e+00 4.86261159e-01 -1.79453954e-01 -1.52914238e+00
4.99260604e-01 -5.57248116e-01 -7.03957856e-01 1.40053403e+00
-2.28414059e+00 -1.35737062e+00 -1.10777342e+00 9.75610673e-01
2.64765710e-01 1.05908066e-02 6.67441845e-01 3.87772948e-01
-6.23860180e-01 -3.09668452e-01 6.06749475e-01 1.56837866e-01
6.25988960e-01 -6.22679353e-01 -5.24260581e-01 1.37979960e+00
-5.50347924e-01 4.02365357e-01 7.28789449e-01 -6.43644571e-01
-1.71254563e+00 -1.23994052e+00 3.87960553e-01 5.73083401e-01
5.15352607e-01 4.78105813e-01 -1.18694270e+00 4.05780375e-01
3.99429142e-01 3.72923583e-01 7.60392189e-01 -4.19203281e-01
-3.67455393e-01 -5.22049248e-01 -1.31776261e+00 1.32225201e-01
5.75497925e-01 -6.40548468e-01 -3.21520299e-01 5.78176558e-01
5.38206041e-01 -1.22874327e-01 -9.68069613e-01 4.76375163e-01
2.61870414e-01 -1.09618294e+00 1.16305816e+00 1.99847579e-01
7.21307576e-01 -8.45375299e-01 -6.61236286e-01 -1.40237546e+00
-1.18995118e+00 -3.04300815e-01 1.63757280e-01 1.04523170e+00
-1.65148199e-01 -6.59516871e-01 2.20234245e-01 1.22971512e-01
-4.13755059e-01 -1.26059297e-02 -7.47107267e-01 -8.02430093e-01
-4.82964635e-01 7.76433647e-02 2.51129389e-01 1.24892068e+00
-4.96017992e-01 2.19885781e-02 -7.17934549e-01 1.00279474e+00
1.14482641e+00 6.68408275e-02 2.31683299e-01 -1.20187438e+00
2.04369891e-03 -4.83817905e-02 -2.11424530e-01 -8.55105102e-01
-6.65770322e-02 -8.11659038e-01 2.70878971e-01 -1.87746978e+00
3.73219520e-01 2.11600661e-01 -4.49649483e-01 6.66827381e-01
4.33981754e-02 4.30462778e-01 -3.19535397e-02 6.56819463e-01
-9.48492140e-02 9.25230861e-01 1.21787238e+00 -2.78304994e-01
-1.26032308e-02 -4.20543462e-01 -6.82264626e-01 6.27383232e-01
6.68019354e-01 -5.10768950e-01 -6.36585578e-02 -6.87289596e-01
-7.21027190e-03 3.33092839e-01 5.95110178e-01 -1.30163574e+00
6.20678246e-01 -2.28186890e-01 6.45791948e-01 -6.17234409e-01
3.57365131e-01 -1.07048523e+00 4.10176158e-01 4.94100899e-01
-6.17303327e-03 -5.38379908e-01 2.49734432e-01 6.98569000e-01
-8.17228556e-01 -1.78120703e-01 1.10135722e+00 -3.57950538e-01
-7.80789018e-01 3.59608531e-01 -2.99980938e-01 -5.00869513e-01
4.28201139e-01 -3.83991361e-01 -3.30870032e-01 -5.86445153e-01
-6.83036149e-01 -8.34412724e-02 1.30289555e-01 -2.79919088e-01
1.07255888e+00 -1.33696425e+00 -1.19062030e+00 4.92826194e-01
-1.74135387e-01 -5.35979792e-02 5.33143580e-01 1.00328207e+00
-6.49636388e-01 -4.77057174e-02 -4.96445328e-01 -5.75268269e-01
-9.87855613e-01 1.97451800e-01 3.34393114e-01 -3.76571864e-02
-5.64853072e-01 6.62599981e-01 6.41831476e-03 -3.33799541e-01
-1.50673181e-01 1.93899333e-01 -3.41423452e-01 -4.29650880e-02
6.45512462e-01 6.22127593e-01 -1.92962125e-01 -9.64733541e-01
-6.26107305e-02 8.00373852e-01 5.77487946e-01 -1.76690109e-02
2.09033775e+00 -2.39578485e-01 -1.00326824e+00 1.29027069e-01
1.25543606e+00 -1.36364758e-01 -1.22114003e+00 -6.45080626e-01
-5.04475534e-01 -8.92916501e-01 8.35799336e-01 -5.83825707e-01
-1.40524900e+00 7.67551839e-01 7.57911384e-01 1.51416123e-01
1.73074949e+00 -5.93285561e-01 6.78839624e-01 3.21447283e-01
1.23650871e-01 -5.87078750e-01 2.22790092e-01 6.24033272e-01
1.26049960e+00 -1.11466348e+00 3.21099699e-01 -6.28899515e-01
-2.18612626e-01 1.21077836e+00 1.83316201e-01 -2.37742141e-01
9.21308279e-01 7.19920620e-02 -2.17785835e-01 -1.39095768e-01
3.95148015e-03 -2.55510062e-01 2.59917527e-01 6.89476430e-01
4.39198434e-01 -6.80037364e-02 -1.44280508e-01 3.75971079e-01
3.24616194e-01 2.70828575e-01 5.27766109e-01 7.84804642e-01
-7.33627439e-01 -6.81235135e-01 -8.97069633e-01 3.18652183e-01
-2.18707025e-02 -4.53348786e-01 2.02511847e-01 1.65159404e-01
1.06715277e-01 1.28735769e+00 -3.30537021e-01 -3.16955686e-01
2.99862772e-01 -2.74806678e-01 1.93078890e-02 -5.00728786e-01
1.24832736e-02 5.60898006e-01 -3.62994075e-01 -5.46933055e-01
-6.80404365e-01 -4.83993351e-01 -6.96126938e-01 -3.25939387e-01
-3.55055928e-01 2.23956048e-01 4.07862097e-01 5.76664686e-01
1.14742912e-01 4.09465671e-01 8.82240176e-01 -1.28283918e+00
-2.72706747e-01 -1.07115805e+00 -1.28367996e+00 1.34373084e-01
5.81248105e-01 -3.33314538e-01 -8.55872929e-01 4.07321066e-01] | [10.488300323486328, -2.1039810180664062] |
7b546a0c-ceab-444a-b36e-7b38c2bc83d0 | differentiable-programming-of-reaction | 2107.06862 | null | https://arxiv.org/abs/2107.06862v1 | https://arxiv.org/pdf/2107.06862v1.pdf | Differentiable Programming of Reaction-Diffusion Patterns | Reaction-Diffusion (RD) systems provide a computational framework that governs many pattern formation processes in nature. Current RD system design practices boil down to trial-and-error parameter search. We propose a differentiable optimization method for learning the RD system parameters to perform example-based texture synthesis on a 2D plane. We do this by representing the RD system as a variant of Neural Cellular Automata and using task-specific differentiable loss functions. RD systems generated by our method exhibit robust, non-trivial 'life-like' behavior. | ['Eyvind Niklasson', 'Ettore Randazzo', 'Alexander Mordvintsev'] | 2021-06-22 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 2.34033331e-01 1.96068157e-02 6.29845336e-02 1.99327439e-01
-3.31823915e-01 -5.01054466e-01 1.17472458e+00 -3.11958879e-01
-1.32496819e-01 1.04955268e+00 -6.82491958e-02 -2.53737867e-01
-5.88642918e-02 -8.75431776e-01 -6.55553818e-01 -1.37825835e+00
-5.08094840e-02 4.47487235e-01 2.65836507e-01 -4.67917055e-01
3.66114020e-01 1.00322044e+00 -1.32677507e+00 3.85536477e-02
3.06913108e-01 1.21000981e+00 -7.93894231e-02 1.04427862e+00
2.09769368e-01 7.80621707e-01 -4.78380263e-01 1.24366330e-02
3.51305276e-01 -5.77742279e-01 -4.23013061e-01 3.71151745e-01
-1.13579713e-01 4.15707260e-01 -5.53949833e-01 4.82302725e-01
6.39922678e-01 2.76066273e-01 1.40636373e+00 -7.98340201e-01
-1.01674843e+00 -2.94447005e-01 -2.58504331e-01 1.58139333e-01
8.27475265e-02 4.76757616e-01 5.59778094e-01 -9.16848361e-01
7.18213558e-01 1.11910033e+00 7.17453837e-01 9.40656424e-01
-1.71081054e+00 -6.90287650e-02 -2.51886517e-01 -5.89757264e-01
-1.42609072e+00 -6.93089366e-01 6.68775439e-01 -7.63815641e-01
1.10269117e+00 3.25541377e-01 7.84048736e-01 1.15147841e+00
1.00626922e+00 6.23022616e-01 1.49192750e+00 -5.75882375e-01
6.54951036e-01 -3.76124978e-01 -3.21949035e-01 1.00882387e+00
2.78495759e-01 4.41131353e-01 -2.48155490e-01 -3.59388888e-02
1.63523197e+00 -4.88415331e-01 8.94745160e-03 -4.30553406e-01
-1.11868155e+00 4.05657083e-01 1.39721274e-01 1.50445834e-01
-4.09465969e-01 6.10771477e-01 1.27390707e-02 3.99043918e-01
4.85287786e-01 9.31656182e-01 -1.24236785e-01 1.88281909e-02
-4.47124422e-01 6.31000936e-01 9.84373748e-01 8.18677068e-01
3.93737406e-01 2.77020305e-01 -3.91463906e-01 1.06966639e+00
2.70187557e-01 4.98307079e-01 3.21525455e-01 -1.16632378e+00
-2.38261282e-01 2.94420481e-01 3.09015483e-01 -8.37512195e-01
-4.72060591e-01 -9.38511491e-02 -1.20656312e+00 4.66441989e-01
5.93358934e-01 -4.09613222e-01 -9.65237379e-01 1.42924929e+00
4.33779925e-01 3.55839729e-02 2.00358525e-01 6.68770194e-01
2.69111514e-01 9.45914328e-01 -3.05593103e-01 -2.64118403e-01
8.09304416e-01 -9.78461444e-01 -5.20387411e-01 3.76398146e-01
4.46623385e-01 -4.77111310e-01 1.18705881e+00 5.21761000e-01
-1.24544919e+00 2.15257891e-02 -1.05268764e+00 3.19300920e-01
-3.95116866e-01 -4.30884287e-02 3.99150640e-01 5.36038935e-01
-9.59451318e-01 7.69824266e-01 -8.44375134e-01 -2.81167537e-01
5.82125127e-01 3.52195799e-01 -1.64326936e-01 3.23791295e-01
-6.38967037e-01 8.05893660e-01 -3.13439488e-01 1.63780957e-01
-1.09888065e+00 -5.41044950e-01 -3.13666850e-01 -4.74245280e-01
2.30317891e-01 -9.84146595e-01 1.31993330e+00 -5.39493501e-01
-2.48987532e+00 9.46834445e-01 2.76362579e-02 -3.49356264e-01
6.87992454e-01 1.44054323e-01 -1.90781713e-01 -1.77589551e-01
-3.36116761e-01 2.86348701e-01 1.03168964e+00 -1.56220794e+00
-2.22286478e-01 1.82922184e-01 -3.25129747e-01 1.04137667e-01
2.60642380e-01 -3.12586516e-01 -1.64328232e-01 -1.13846254e+00
2.29162071e-02 -1.03269351e+00 -6.32708669e-01 5.72792768e-01
-4.53506261e-01 -1.33488074e-01 7.14980960e-01 6.82231933e-02
1.05997539e+00 -1.68210065e+00 3.42239588e-01 3.62986922e-01
1.43112659e-01 1.97482854e-02 3.13586071e-02 6.22403622e-01
4.63196069e-01 9.25733987e-03 -4.25260991e-01 -1.29276842e-01
4.42454629e-02 3.07803720e-01 -2.11327419e-01 5.36916852e-01
4.21156079e-01 9.39333737e-01 -7.94995785e-01 -2.34013200e-01
-1.66992217e-01 4.33014691e-01 -5.24177074e-01 2.35181674e-01
-5.87513983e-01 5.75091660e-01 -5.36561310e-01 7.70335793e-01
4.74108048e-02 -3.39979023e-01 -2.58896202e-01 2.26495653e-01
-3.64310861e-01 -2.02871606e-01 -7.73818910e-01 1.24950314e+00
-5.35178840e-01 7.79118299e-01 7.19367489e-02 -8.83162320e-01
1.27475989e+00 1.09231718e-01 5.66346884e-01 -7.01764345e-01
2.05509573e-01 4.17196482e-01 3.65992524e-02 -5.55683672e-01
5.00419475e-02 -3.88157785e-01 -1.42827407e-01 7.16041028e-01
-3.06055337e-01 -8.53098214e-01 -8.64479020e-02 -5.25082350e-01
1.44684625e+00 -4.24737856e-02 -1.28553480e-01 -8.66382182e-01
4.88423944e-01 2.38023642e-02 1.39205009e-01 9.78242815e-01
1.02429405e-01 6.65863454e-01 5.19169390e-01 -8.00707757e-01
-1.63947976e+00 -8.74491096e-01 -3.02706361e-01 3.71139109e-01
9.22663286e-02 3.18474054e-01 -9.53635633e-01 -1.33367509e-01
1.68472871e-01 3.39996904e-01 -9.84358191e-01 -2.58604199e-01
-8.89888644e-01 -9.17920411e-01 7.46021271e-01 1.08893216e-01
5.87831914e-01 -1.27972281e+00 -4.43537384e-01 7.67654300e-01
6.68844342e-01 -5.88484883e-01 -4.65478688e-01 2.63309628e-01
-5.08861661e-01 -7.92598307e-01 -1.12225735e+00 -8.45946550e-01
8.07498991e-01 -1.79685131e-01 9.56972778e-01 6.28990829e-02
-5.29601634e-01 3.15796226e-01 1.18877515e-01 -2.31241867e-01
-6.99133396e-01 4.93220538e-02 4.03931797e-01 1.31584570e-01
-3.29335093e-01 -4.59470898e-01 -8.33502233e-01 7.46344090e-01
-8.48152518e-01 2.47430518e-01 3.81123960e-01 1.07464147e+00
9.62059557e-01 2.25596130e-01 7.93438375e-01 -6.67065740e-01
1.25231004e+00 -1.90641955e-01 -7.25797951e-01 1.95760414e-01
-5.67115366e-01 3.07146072e-01 7.21445203e-01 -7.84257412e-01
-1.00986612e+00 6.45371825e-02 -1.38038909e-02 -2.43003860e-01
-9.06699821e-02 1.12171516e-01 1.10904738e-01 -6.11878872e-01
8.87131572e-01 4.06749845e-01 3.28692228e-01 -1.53796658e-01
1.72458082e-01 3.86745095e-01 4.27381277e-01 -1.33591568e+00
5.43972075e-01 5.06780863e-01 5.96439064e-01 -1.32210767e+00
-4.81374174e-01 3.81207407e-01 -3.43766600e-01 -3.63718182e-01
6.39168203e-01 -2.27091685e-01 -9.09787238e-01 1.10445356e+00
-8.92758846e-01 -1.41826403e+00 -7.42341220e-01 -1.43949702e-01
-1.23698282e+00 -4.10749406e-01 -7.90015042e-01 -9.97836053e-01
-1.83316320e-01 -9.28986013e-01 1.15185153e+00 1.03223741e-01
-3.21728081e-01 -1.11502075e+00 6.46285355e-01 -4.27505136e-01
7.00975358e-01 5.32167733e-01 9.79832530e-01 1.73325121e-01
-6.09474301e-01 -8.45592842e-02 -5.47306836e-02 4.86850999e-02
3.07562292e-01 4.59851265e-01 -6.19335830e-01 -1.43288374e-01
9.99908745e-02 -8.66028294e-02 5.45391083e-01 6.92973375e-01
1.20386755e+00 -8.78767669e-01 -5.26742637e-01 7.18006432e-01
1.33497667e+00 4.91204232e-01 7.04682827e-01 3.64273071e-01
5.44298112e-01 5.68571627e-01 5.27799837e-02 2.90218711e-01
1.14282362e-01 7.69743800e-01 -2.55538851e-01 -5.07560223e-02
-3.07520658e-01 -2.40237147e-01 9.41685364e-02 5.23031175e-01
-2.37064719e-01 -4.96587604e-01 -9.52036560e-01 3.18693846e-01
-1.86581695e+00 -7.84316719e-01 6.41776472e-02 1.90965831e+00
9.46310818e-01 2.00818881e-01 3.73827428e-01 -8.19096416e-02
4.59279507e-01 -1.75319076e-01 -6.64970696e-01 -4.98835325e-01
-6.38550341e-01 2.55249143e-01 5.78490019e-01 5.77956617e-01
-9.00088549e-01 1.19495225e+00 8.14473057e+00 9.36903238e-01
-1.42157817e+00 -3.39511931e-01 9.17699099e-01 -3.53954248e-02
-2.74815947e-01 -3.07237685e-01 -7.37595856e-01 3.75051141e-01
7.49914885e-01 -1.82597071e-01 5.18581092e-01 1.29490048e-01
4.97182071e-01 -8.53111371e-02 -8.28138292e-01 9.69119608e-01
-5.87751091e-01 -2.03985119e+00 -4.69511822e-02 2.36007348e-01
1.21074605e+00 -3.84416014e-01 3.09460610e-01 -2.55527169e-01
5.94006360e-01 -1.17032290e+00 8.03362250e-01 9.97918367e-01
8.09065223e-01 -4.48194921e-01 -9.74528678e-03 1.50533065e-01
-7.90886760e-01 2.03625448e-02 -3.03682894e-01 -2.33565662e-02
3.07592213e-01 5.96211016e-01 -5.65561414e-01 -4.40100789e-01
5.01756907e-01 4.48098361e-01 3.41115072e-02 9.69509423e-01
1.48553342e-01 5.27795374e-01 -6.04606748e-01 -7.79426813e-01
2.55108565e-01 -4.97993290e-01 7.99066186e-01 1.24055195e+00
5.11700332e-01 6.84114218e-01 -4.10557389e-01 1.03486240e+00
-7.57576004e-02 -4.72944416e-02 -1.07533360e+00 -2.00111523e-01
2.96128035e-01 7.92582929e-01 -1.09932339e+00 -1.14830714e-02
2.00457573e-01 8.02135170e-01 2.59830207e-01 5.03362775e-01
-3.90112132e-01 -2.59951472e-01 8.47961187e-01 5.30803978e-01
3.25035304e-01 -7.11971402e-01 -4.18860853e-01 -9.78138804e-01
-3.01554292e-01 -6.72251105e-01 -3.84599298e-01 -5.63768506e-01
-1.57473779e+00 5.39708257e-01 -4.18083251e-01 -8.16451907e-01
7.68561214e-02 -8.81678343e-01 -5.08937776e-01 5.46547651e-01
-8.44826877e-01 -8.46594512e-01 2.02408224e-01 3.97371560e-01
4.23557431e-01 -5.39760232e-01 8.44637811e-01 -9.11344513e-02
-7.85440922e-01 4.96305227e-01 5.18003166e-01 -4.15562600e-01
7.22175688e-02 -1.15294468e+00 7.84871340e-01 2.99572408e-01
-3.78623694e-01 2.09403887e-01 9.34960842e-01 -6.41955614e-01
-1.70537090e+00 -8.79127383e-01 4.30732995e-01 -4.50740188e-01
7.97369897e-01 -6.30436778e-01 -8.31085324e-01 5.34169748e-02
-1.40858516e-01 8.32997784e-02 2.15754583e-01 -5.86433291e-01
1.29773647e-01 8.55309367e-02 -1.25718355e+00 1.25149214e+00
1.19360113e+00 -3.64537954e-01 9.30798128e-02 4.65700358e-01
4.22147751e-01 -3.76823992e-01 -1.02478898e+00 1.55242890e-01
7.34176338e-01 -2.53282189e-01 8.85638118e-01 -6.55257642e-01
6.34432495e-01 -3.45409811e-01 -5.83197884e-02 -1.42047811e+00
-5.12669981e-01 -1.43612289e+00 -4.21834379e-01 6.14541471e-01
6.00484848e-01 -8.15345705e-01 1.10682881e+00 7.15835571e-01
-8.44683573e-02 -1.32808208e+00 -9.00489867e-01 -9.36324835e-01
5.27702689e-01 1.24585927e-01 5.09421170e-01 6.20158494e-01
-1.90768138e-01 1.38834551e-01 -1.09906606e-01 -1.68270797e-01
6.53542042e-01 -1.95461452e-01 7.37819552e-01 -9.40477312e-01
-2.41040617e-01 -1.10841775e+00 -6.92377239e-02 -1.30591822e+00
-6.92378059e-02 -4.60594147e-01 4.23808903e-01 -1.08882546e+00
-5.28932333e-01 -1.19093692e+00 6.09443933e-02 6.58591092e-02
4.00827557e-01 1.96376204e-01 -4.87590760e-01 4.35239762e-01
-2.69514114e-01 6.74379587e-01 1.69864881e+00 -9.64488834e-02
-4.21038121e-01 -6.99067721e-03 -3.35568845e-01 2.67280459e-01
8.39190245e-01 -3.53742003e-01 -1.74699619e-01 -1.33236974e-01
3.95044535e-01 2.46543571e-01 1.65692925e-01 -9.64037418e-01
3.12384844e-01 -6.54529035e-01 2.28985846e-01 1.37039006e-01
2.34420910e-01 -3.94822478e-01 5.70555031e-01 4.16066915e-01
-6.08575165e-01 -1.18944339e-01 1.66539311e-01 5.23930192e-01
1.48785248e-01 1.09157652e-01 1.16695595e+00 -7.99569190e-02
-3.01643647e-02 3.97248834e-01 -1.03880918e+00 -4.89898771e-02
1.05241954e+00 -3.94870579e-01 -4.68348116e-01 -1.41396880e-01
-9.31656599e-01 -1.97214290e-01 7.47555196e-01 2.63446737e-02
6.18964314e-01 -1.51755416e+00 -5.51446080e-01 3.66978705e-01
-1.01181827e-01 1.74091920e-01 -2.04561591e-01 4.69083369e-01
-1.19843125e+00 3.30932587e-01 -2.87404656e-01 -5.93981564e-01
-4.81068075e-01 2.99472779e-01 9.15126503e-01 -4.14288044e-02
-6.69770122e-01 1.15434468e+00 2.25858748e-01 -4.21459615e-01
-2.61500686e-01 -1.72351524e-01 1.96041182e-01 -5.03819585e-01
1.14365876e-01 3.98068577e-01 3.23296078e-02 -3.32517117e-01
1.21286936e-01 6.70304954e-01 1.15053937e-01 -5.88643909e-01
1.51224518e+00 7.58413880e-05 -6.02388643e-02 5.55064738e-01
9.38169122e-01 -4.37386632e-01 -1.89385581e+00 2.43855007e-02
-8.79883468e-02 -1.78370580e-01 7.69828781e-02 -4.32842463e-01
-9.30876315e-01 4.80824113e-01 3.37562144e-01 5.95097423e-01
6.49849236e-01 -4.15388271e-02 6.30014718e-01 6.09912097e-01
4.10573810e-01 -1.23595059e+00 4.07260835e-01 6.26300514e-01
1.17401111e+00 -6.88928068e-01 -2.56848305e-01 -7.35525042e-02
-2.38665089e-01 1.08915401e+00 5.09522915e-01 -8.94634366e-01
1.18077850e+00 7.14270532e-01 -9.30764750e-02 -3.99134696e-01
-1.24908900e+00 3.32353950e-01 1.92607388e-01 6.54644549e-01
2.96301246e-01 3.07497308e-02 -1.94694594e-01 1.50641903e-01
-1.11427628e-01 2.34933244e-03 4.13730025e-01 1.12442601e+00
-5.95762908e-01 -1.10592437e+00 -1.41021550e-01 5.41614115e-01
-2.83523232e-01 3.37676436e-01 -3.53803605e-01 5.02192020e-01
-4.90686178e-01 3.11327875e-01 6.62929639e-02 -1.61521807e-01
4.82917219e-01 -2.58161128e-01 9.79670107e-01 -3.01084667e-01
-2.54027903e-01 1.36035696e-01 -2.49255281e-02 -2.31679946e-01
-5.86856641e-02 -7.49114454e-01 -9.32433724e-01 -6.18734360e-01
-1.29062876e-01 -7.84971416e-02 4.60941255e-01 7.73114502e-01
4.40396696e-01 4.16175246e-01 6.29584014e-01 -1.08445275e+00
-7.03588203e-02 -3.30185562e-01 -6.96790755e-01 1.88318864e-02
5.64553380e-01 -9.15665209e-01 -2.66139448e-01 1.45859852e-01] | [11.427066802978516, -0.35613998770713806] |
35d7622b-3421-42d1-9fa7-84e9d20107ec | an-adapter-based-multi-label-pre-training-for | 2211.06041 | null | https://arxiv.org/abs/2211.06041v1 | https://arxiv.org/pdf/2211.06041v1.pdf | An Adapter based Multi-label Pre-training for Speech Separation and Enhancement | In recent years, self-supervised learning (SSL) has achieved tremendous success in various speech tasks due to its power to extract representations from massive unlabeled data. However, compared with tasks such as speech recognition (ASR), the improvements from SSL representation in speech separation (SS) and enhancement (SE) are considerably smaller. Based on HuBERT, this work investigates improving the SSL model for SS and SE. We first update HuBERT's masked speech prediction (MSP) objective by integrating the separation and denoising terms, resulting in a multiple pseudo label pre-training scheme, which significantly improves HuBERT's performance on SS and SE but degrades the performance on ASR. To maintain its performance gain on ASR, we further propose an adapter-based architecture for HuBERT's Transformer encoder, where only a few parameters of each layer are adjusted to the multiple pseudo label MSP while other parameters remain frozen as default HuBERT. Experimental results show that our proposed adapter-based multiple pseudo label HuBERT yield consistent and significant performance improvements on SE, SS, and ASR tasks, with a faster pre-training speed, at only marginal parameters increase. | ['Weibin Zhu', 'Shu Yu', 'Zhuo Chen', 'Xie Chen', 'Tianrui Wang'] | 2022-11-11 | null | null | null | null | ['speech-separation'] | ['speech'] | [ 5.13281524e-01 2.32337028e-01 -1.80181786e-01 -7.37089038e-01
-1.11786854e+00 -1.60337135e-01 6.02864385e-01 -8.40295926e-02
-3.82179618e-01 5.01298368e-01 3.80271763e-01 -4.54917878e-01
2.94166535e-01 -9.87993255e-02 -3.68078649e-01 -7.62792945e-01
1.68337479e-01 6.32497892e-02 3.49293023e-01 -2.34213978e-01
-2.92500347e-01 2.38167942e-01 -1.35188019e+00 3.09955418e-01
9.33226407e-01 1.19505227e+00 3.90578479e-01 6.01110697e-01
-1.10149220e-01 1.25116527e+00 -7.89695323e-01 -4.08031970e-01
2.02473756e-02 -4.70026016e-01 -7.72983968e-01 3.03353041e-01
2.67031759e-01 -2.06920117e-01 -6.49408698e-01 1.08948314e+00
6.47480667e-01 4.01360214e-01 4.75020438e-01 -1.04717743e+00
-8.10628831e-01 1.01443219e+00 -4.99895841e-01 1.31352082e-01
-1.08667381e-01 -1.57376587e-01 1.03002417e+00 -1.19626176e+00
-1.08296823e-04 1.38021612e+00 7.55191088e-01 7.14693487e-01
-1.05513811e+00 -8.55699539e-01 3.55539978e-01 3.80026758e-01
-1.44453859e+00 -1.23447323e+00 6.79745555e-01 1.00220308e-01
1.10040879e+00 3.67012888e-01 3.20377643e-03 1.16750336e+00
-4.28390771e-01 1.15949893e+00 1.08245575e+00 -4.97440666e-01
3.36685687e-01 2.14220494e-01 2.45880708e-01 6.38733149e-01
-4.87405747e-01 1.65144764e-02 -7.50583410e-01 1.52256310e-01
3.80086631e-01 -1.82110965e-01 -3.87293756e-01 2.12483376e-01
-1.09981441e+00 6.61676109e-01 2.85511613e-01 2.94257313e-01
-1.33692756e-01 -1.21732041e-01 2.31025696e-01 4.72603679e-01
8.40173483e-01 6.87803030e-02 -5.11373818e-01 -1.90917268e-01
-1.26872480e+00 -5.20859480e-01 5.06834328e-01 1.09059238e+00
7.32828617e-01 6.54959798e-01 -3.30561846e-01 1.51217151e+00
2.65341043e-01 6.07802570e-01 8.15744460e-01 -7.90659428e-01
6.51114583e-01 2.55600929e-01 -2.95331538e-01 -4.82785940e-01
-3.15889388e-01 -9.16465640e-01 -1.10678101e+00 -2.18375087e-01
-2.82103270e-01 -7.82756433e-02 -1.39067721e+00 1.88132405e+00
-1.89603046e-02 5.22249341e-01 3.23259562e-01 8.84696424e-01
1.05379319e+00 1.01415527e+00 1.54415667e-01 -3.89067680e-01
1.08081353e+00 -1.48111939e+00 -1.04595304e+00 -7.34449327e-01
7.77101457e-01 -8.14167082e-01 9.77752805e-01 2.18972147e-01
-1.11391973e+00 -6.14455223e-01 -1.06170642e+00 -2.09756687e-01
-7.64918700e-02 5.00651240e-01 2.86078483e-01 8.35135102e-01
-1.28849399e+00 4.99285161e-01 -7.94813156e-01 -2.03486413e-01
4.00115907e-01 3.42299402e-01 -2.69923925e-01 1.09680049e-01
-1.34040177e+00 9.81045604e-01 2.52248079e-01 1.51617244e-01
-9.44993973e-01 -5.48057497e-01 -9.17012691e-01 4.27398890e-01
3.61012071e-01 -1.21800706e-01 1.47917223e+00 -8.67931247e-01
-1.96054602e+00 6.25689685e-01 -5.24371803e-01 -6.70011759e-01
-3.58259212e-03 9.85195339e-02 -8.91258478e-01 1.45862743e-01
-3.14674407e-01 6.79931223e-01 1.15011704e+00 -1.28243840e+00
-6.52714312e-01 -2.68166959e-01 -2.71372974e-01 2.95217693e-01
-6.70331776e-01 3.22636634e-01 -2.90706605e-01 -9.59153056e-01
5.22853553e-01 -8.02405894e-01 5.57715409e-02 -4.79705364e-01
-1.07708253e-01 -2.46928573e-01 8.40683401e-01 -9.49902892e-01
1.45739067e+00 -2.45727777e+00 5.76034822e-02 -1.29325867e-01
2.63137549e-01 8.10194194e-01 -3.78232002e-01 2.33547643e-01
-2.55972624e-01 -7.23513961e-02 -4.94409770e-01 -1.11199510e+00
1.21294416e-01 3.15916836e-01 -3.99196506e-01 2.70050615e-01
4.08127874e-01 7.44588912e-01 -7.48745382e-01 -4.42543179e-01
2.13098243e-01 6.22088611e-01 -3.01260442e-01 5.34221232e-01
9.88276452e-02 3.91098052e-01 1.76324248e-01 5.19317806e-01
7.94424295e-01 -2.82950670e-01 8.82574823e-03 -1.43879324e-01
3.64026017e-02 9.24018204e-01 -1.07210052e+00 1.47940540e+00
-7.59333014e-01 6.67550266e-01 5.08221090e-01 -1.09676433e+00
1.21780539e+00 7.07701981e-01 6.22599646e-02 -6.89039290e-01
7.02767372e-02 2.80660301e-01 -2.92056147e-02 -8.82550180e-02
4.21726644e-01 -3.52401525e-01 3.87857020e-01 4.04545754e-01
3.65512520e-01 2.26675123e-02 -1.21848963e-01 3.83957893e-01
1.15252388e+00 -4.39081669e-01 1.50201786e-02 7.87671506e-02
8.12266767e-01 -5.93979061e-01 5.70113003e-01 4.89512414e-01
-3.92873555e-01 7.60917246e-01 -4.18720730e-02 2.06273109e-01
-7.02193677e-01 -1.07085228e+00 -1.13994546e-01 1.25525355e+00
8.17763209e-02 -5.85740626e-01 -7.87701309e-01 -6.12992942e-01
-4.26241338e-01 7.56068051e-01 -1.48990318e-01 -5.03640056e-01
-6.44932449e-01 -8.09345961e-01 6.78824484e-01 5.00093520e-01
7.50100493e-01 -8.99241805e-01 3.15465778e-01 5.37103340e-02
-3.62948596e-01 -1.24458635e+00 -6.77224398e-01 5.88072062e-01
-7.18397319e-01 -1.82765260e-01 -7.93938339e-01 -9.64011490e-01
4.86802608e-01 6.23550951e-01 7.26755083e-01 -1.37718618e-01
3.53348553e-01 -4.10394371e-02 -6.96501374e-01 -2.09640115e-02
-8.63993943e-01 2.95132279e-01 1.51588559e-01 3.57095301e-01
1.92475289e-01 -6.92067206e-01 -2.57971197e-01 4.95066345e-01
-8.03140461e-01 -4.65581007e-02 6.76656783e-01 9.81473386e-01
2.65115499e-01 1.39929637e-01 9.39437151e-01 -7.76135504e-01
3.35264713e-01 -3.35346460e-01 -1.76213264e-01 2.41514623e-01
-9.42761183e-01 7.50909820e-02 8.01383793e-01 -5.00255466e-01
-1.26707685e+00 -3.44508111e-01 -5.81783593e-01 -6.31000638e-01
-1.07608862e-01 3.21878791e-01 -3.49258274e-01 1.37433307e-02
3.38836700e-01 3.80275488e-01 -2.19846368e-02 -8.23179185e-01
3.30212653e-01 1.48030996e+00 5.64155340e-01 -1.76594362e-01
7.05722511e-01 2.11682469e-02 -5.52374840e-01 -8.76962662e-01
-1.18724751e+00 -7.06938863e-01 -2.73372978e-01 7.46622607e-02
4.08971310e-01 -1.28853595e+00 -3.29857081e-01 5.65548241e-01
-9.84438658e-01 -2.15882674e-01 -2.85492569e-01 5.92236578e-01
-3.37456882e-01 4.40833122e-01 -8.34079981e-01 -9.26775336e-01
-4.58239496e-01 -1.18446720e+00 1.18475401e+00 -3.97473462e-02
-1.55755773e-01 -7.77593851e-01 -2.25881845e-01 6.94234908e-01
7.14451492e-01 -9.23888266e-01 7.68734992e-01 -7.73542583e-01
-3.71336222e-01 1.93019733e-02 -3.37013662e-01 1.00514483e+00
3.65343481e-01 -5.49638033e-01 -1.59506702e+00 -7.08513081e-01
3.95820528e-01 -2.76324213e-01 1.03373098e+00 6.64762557e-02
9.60867107e-01 -2.70066321e-01 -1.08564887e-02 8.03095400e-01
8.26371849e-01 3.43578786e-01 4.72872913e-01 1.43534556e-01
7.36887157e-01 4.09785420e-01 3.01095843e-01 1.08845882e-01
2.32659876e-01 6.76707029e-01 1.03814691e-01 -2.89638370e-01
-8.07397902e-01 -2.30848178e-01 8.20262909e-01 1.71730280e+00
3.35413724e-01 -1.72509149e-01 -6.34118140e-01 4.83485013e-01
-1.68186605e+00 -7.86586404e-01 3.16208214e-01 2.25669718e+00
1.12833142e+00 5.48430495e-02 -9.34020653e-02 3.87276828e-01
8.03173900e-01 5.07458031e-01 -4.68910933e-01 -2.31162548e-01
-3.50913376e-01 3.18290919e-01 4.55283761e-01 7.60277390e-01
-1.11419511e+00 1.10443115e+00 6.16878366e+00 1.31329167e+00
-1.20963347e+00 5.93715668e-01 5.47430694e-01 -1.54874213e-02
-1.81770712e-01 -6.99047968e-02 -9.19635534e-01 4.74729568e-01
1.51012540e+00 4.73446921e-02 5.90534329e-01 6.91044867e-01
3.07645857e-01 3.74319136e-01 -1.00238347e+00 1.14792144e+00
1.54205680e-01 -9.04675841e-01 4.15025800e-02 -3.58695507e-01
6.17859423e-01 1.78064868e-01 3.11927587e-01 6.58475161e-01
2.21565083e-01 -8.38476717e-01 7.88876295e-01 -2.12669313e-01
8.30165803e-01 -7.28995204e-01 6.73929572e-01 4.45045382e-01
-1.10330510e+00 -2.80167580e-01 -4.13097352e-01 2.10950688e-01
1.77528322e-01 6.26384079e-01 -9.84537065e-01 3.79876494e-01
4.04651672e-01 7.69466162e-01 -4.96937573e-01 5.61790109e-01
-4.09385324e-01 1.27433741e+00 -1.34928077e-01 3.57029110e-01
1.65762097e-01 -7.55474120e-02 5.03347635e-01 1.42713988e+00
1.28194138e-01 -2.91459747e-02 4.31367606e-02 6.27151191e-01
-3.47603202e-01 2.14602090e-02 -1.73673164e-02 -8.67113173e-02
8.59888077e-01 1.06915843e+00 -3.29487473e-01 -4.07110810e-01
-4.75446731e-01 1.03199422e+00 4.32649016e-01 4.41232324e-01
-8.40121865e-01 -3.93140823e-01 6.25365019e-01 9.26128775e-03
3.58059943e-01 -1.82819128e-01 -1.60961032e-01 -1.26743531e+00
4.68032295e-03 -1.08652556e+00 -1.48605835e-02 -6.99546397e-01
-1.03826058e+00 1.01460469e+00 -4.04891133e-01 -1.17885196e+00
-1.09984510e-01 -2.55698502e-01 -2.94152111e-01 9.66342151e-01
-1.88104486e+00 -1.21573997e+00 1.21416748e-01 4.69673395e-01
9.29810762e-01 -4.91761357e-01 8.09967577e-01 6.39665723e-01
-1.03074014e+00 1.01839626e+00 2.97095537e-01 9.48078483e-02
8.51030052e-01 -1.19095719e+00 5.96404016e-01 1.07368267e+00
3.04892361e-01 5.22563994e-01 2.89172292e-01 -4.21352893e-01
-9.69431281e-01 -1.29606712e+00 1.03406191e+00 -5.40367933e-03
6.82473540e-01 -5.22816241e-01 -1.02836728e+00 8.00652802e-01
1.51075751e-01 -8.25299919e-02 5.32646894e-01 1.46347374e-01
-5.91103375e-01 -3.36543411e-01 -8.72858047e-01 4.98070836e-01
1.06255412e+00 -9.61475134e-01 -5.15748680e-01 2.27371648e-01
1.21332097e+00 -1.81935400e-01 -8.38639259e-01 4.70985115e-01
-4.56187800e-02 -7.23677993e-01 9.39847648e-01 -2.21498564e-01
1.46289751e-01 -3.52137715e-01 -4.24040824e-01 -1.53806889e+00
-5.16779840e-01 -8.70114982e-01 -3.79687250e-01 1.69928789e+00
5.95227361e-01 -5.65794587e-01 6.45680487e-01 3.51533175e-01
-6.53250635e-01 -5.03235757e-01 -1.10836291e+00 -1.03464794e+00
-2.70366699e-01 -3.58202130e-01 6.30588293e-01 9.25666213e-01
-1.15114525e-01 7.44440794e-01 -6.18536115e-01 4.66537863e-01
5.45418143e-01 -1.58623293e-01 2.19740912e-01 -8.52631390e-01
-3.35043192e-01 -1.79828465e-01 -1.38992086e-01 -1.42409015e+00
4.54750061e-01 -1.18534338e+00 3.24111223e-01 -1.26859498e+00
2.85600759e-02 -5.56198120e-01 -6.20673537e-01 6.06541693e-01
-3.87514353e-01 1.91970527e-01 2.38816246e-01 3.54801387e-01
-6.83960199e-01 9.60833371e-01 9.04852033e-01 -4.58043188e-01
-1.01271972e-01 1.59885615e-01 -5.45599222e-01 6.54883683e-01
6.42819405e-01 -4.89134312e-01 -6.14948809e-01 -7.67259538e-01
-4.57213014e-01 -1.47232994e-01 -1.81391522e-01 -1.02395093e+00
3.19070160e-01 3.09629589e-01 -7.39055723e-02 -3.86551946e-01
7.21686423e-01 -7.07393408e-01 -1.98871717e-01 1.87806226e-02
-5.92095971e-01 -4.65078145e-01 3.87348756e-02 3.15710306e-01
-6.66641414e-01 -2.80013114e-01 1.24237216e+00 2.38704339e-01
-6.13755941e-01 1.30195543e-01 -3.69440019e-01 -7.07317367e-02
4.76232916e-01 5.47075793e-02 -3.89131010e-01 -5.63017249e-01
-7.94645607e-01 7.19858855e-02 1.06927328e-01 4.77155507e-01
7.08983362e-01 -1.25937557e+00 -7.23912656e-01 5.22454262e-01
-8.05940256e-02 -5.78296036e-02 3.28406364e-01 8.53666961e-01
1.07006133e-01 4.63686675e-01 4.83431160e-01 -4.39091653e-01
-1.19907248e+00 4.32720870e-01 2.08098754e-01 -1.43949464e-01
-5.90197086e-01 1.13505399e+00 2.08276138e-01 -6.10147655e-01
6.89337075e-01 -1.59239352e-01 -1.47302926e-01 -8.93077478e-02
6.90785885e-01 4.56797332e-01 5.08599579e-01 -8.02174926e-01
-3.07225585e-01 2.46858820e-01 -3.56442332e-01 -1.09699719e-01
1.37692249e+00 -6.06707752e-01 9.44777206e-03 4.88075644e-01
1.29883206e+00 2.77579408e-02 -1.06228173e+00 -5.08781374e-01
-3.54487523e-02 -1.57984927e-01 6.46440864e-01 -8.15478146e-01
-1.21292436e+00 1.10840392e+00 7.01511681e-01 1.56802148e-01
1.22648549e+00 6.54094517e-02 1.23021209e+00 2.59144098e-01
2.48307139e-01 -1.04837573e+00 -1.37354165e-01 6.21542931e-01
6.12164855e-01 -1.29204428e+00 -3.63614529e-01 -6.09270215e-01
-6.47433519e-01 5.85242212e-01 3.69537503e-01 4.65891361e-01
6.64310038e-01 5.44186592e-01 2.45917782e-01 3.31884444e-01
-8.90221417e-01 -3.11259806e-01 1.46479681e-01 5.47293782e-01
4.33921665e-01 2.43151598e-02 1.19053602e-01 6.52159989e-01
-1.37293234e-01 -2.43280649e-01 2.97613800e-01 6.57956898e-01
-7.26479530e-01 -1.19371402e+00 -2.93003052e-01 2.56383687e-01
-3.55403423e-01 -5.89601219e-01 -1.08711444e-01 3.44535299e-02
-1.06662832e-01 1.40907848e+00 -3.34668532e-02 -5.97117782e-01
2.64680982e-01 2.77467877e-01 6.85614571e-02 -7.98713207e-01
-5.52238584e-01 3.22592497e-01 2.27002248e-01 -2.65341759e-01
-4.86979067e-01 -2.95352280e-01 -1.22477090e+00 -1.57449022e-01
-6.85080171e-01 3.49815249e-01 7.17937648e-01 1.03158557e+00
3.42177302e-01 6.46113336e-01 1.06060946e+00 -5.53944230e-01
-9.01154399e-01 -1.34186339e+00 -7.61083007e-01 3.20334673e-01
6.54924393e-01 -6.01172268e-01 -8.03633273e-01 1.88214496e-01] | [14.567848205566406, 6.3883562088012695] |
dfc6ec59-011a-43e4-85c7-e24f49b413c3 | multi-level-adaptive-region-of-interest-and | 2102.12154 | null | https://arxiv.org/abs/2102.12154v1 | https://arxiv.org/pdf/2102.12154v1.pdf | Multi-Level Adaptive Region of Interest and Graph Learning for Facial Action Unit Recognition | In facial action unit (AU) recognition tasks, regional feature learning and AU relation modeling are two effective aspects which are worth exploring. However, the limited representation capacity of regional features makes it difficult for relation models to embed AU relationship knowledge. In this paper, we propose a novel multi-level adaptive ROI and graph learning (MARGL) framework to tackle this problem. Specifically, an adaptive ROI learning module is designed to automatically adjust the location and size of the predefined AU regions. Meanwhile, besides relationship between AUs, there exists strong relevance between regional features across multiple levels of the backbone network as level-wise features focus on different aspects of representation. In order to incorporate the intra-level AU relation and inter-level AU regional relevance simultaneously, a multi-level AU relation graph is constructed and graph convolution is performed to further enhance AU regional features of each level. Experiments on BP4D and DISFA demonstrate the proposed MARGL significantly outperforms the previous state-of-the-art methods. | ['ShiLiang Pu', 'Chunmao Wang', 'Qiang Li', 'Jingjing Wang', 'Boyuan Jiang', 'Jingwei Yan'] | 2021-02-24 | null | null | null | null | ['facial-action-unit-detection'] | ['computer-vision'] | [ 1.24522448e-01 3.04435492e-01 -4.22065228e-01 -4.01654720e-01
-2.85224259e-01 -7.31139258e-02 4.00414616e-01 -4.27290089e-02
-6.49545714e-03 3.57290208e-01 2.16388986e-01 2.15274602e-01
-1.95483699e-01 -1.01236701e+00 -4.96779412e-01 -6.63669884e-01
-1.23278268e-01 -1.96889699e-01 5.12540936e-01 -3.35629880e-01
-6.02864511e-02 8.34858775e-01 -1.35949957e+00 3.48899573e-01
4.71335948e-01 1.49018955e+00 -7.21131265e-02 1.29966184e-01
-7.30621517e-02 1.03757679e+00 -2.95662463e-01 -1.42992914e-01
4.02559787e-01 -5.08491814e-01 -5.56106031e-01 4.26821440e-01
4.57964033e-01 -2.24863127e-01 -6.35782480e-01 8.80931616e-01
4.27442163e-01 1.85246050e-01 7.09354103e-01 -1.21671569e+00
-5.73867202e-01 5.46463490e-01 -1.23630846e+00 5.07521331e-01
-6.36550272e-03 -2.87949014e-03 1.31573200e+00 -8.74388635e-01
4.18944627e-01 1.35981226e+00 2.60652125e-01 3.89819950e-01
-1.02742422e+00 -8.70387495e-01 5.88516235e-01 3.61793816e-01
-1.53597367e+00 -1.24305092e-01 1.09146857e+00 -1.54968008e-01
7.74080038e-01 -3.59259807e-02 5.74757695e-01 5.81923425e-01
2.22662985e-01 6.01538360e-01 9.73051667e-01 -2.51055449e-01
-3.00815254e-01 -2.32087836e-01 -4.18173708e-03 1.18084192e+00
-3.67904752e-01 -1.47578374e-01 -5.57363451e-01 7.86374733e-02
1.32248020e+00 -2.51160525e-02 -1.71453401e-01 -2.54366428e-01
-7.24308670e-01 7.56076813e-01 1.02635002e+00 4.67531979e-01
-3.55905801e-01 3.30923766e-01 2.81395853e-01 9.98633653e-02
4.22624528e-01 -2.20165133e-01 -3.08444619e-01 3.05458099e-01
-5.22119045e-01 -4.16554213e-01 1.24471024e-01 8.70256960e-01
1.21778524e+00 7.09894821e-02 -4.79338229e-01 9.06626642e-01
5.37573040e-01 -2.83199221e-01 1.37207642e-01 -6.37270033e-01
5.55456340e-01 1.24301648e+00 -6.01011336e-01 -1.35722005e+00
-4.85039175e-01 -6.14227474e-01 -9.82039392e-01 8.59155133e-02
1.19424984e-01 -2.65678205e-02 -7.62793005e-01 1.62901187e+00
5.89240849e-01 6.49033725e-01 -1.53590232e-01 8.31966519e-01
1.17664361e+00 6.65658951e-01 7.93334469e-02 -2.06613332e-01
1.37750185e+00 -1.13851094e+00 -4.39645082e-01 -1.21117063e-01
6.13355398e-01 -4.92980868e-01 8.13747168e-01 9.74262226e-03
-9.16492879e-01 -8.47098172e-01 -8.84802580e-01 -4.55282889e-02
-1.61162049e-01 4.64480132e-01 7.77296841e-01 2.91340023e-01
-8.90283167e-01 2.37648353e-01 -3.49110514e-01 -1.03209853e-01
8.74707639e-01 6.31694496e-01 -4.86442268e-01 -1.40301347e-01
-1.32484508e+00 5.03150463e-01 2.95056760e-01 5.04654944e-01
-8.00392985e-01 -5.01746297e-01 -1.01816297e+00 1.22075841e-01
3.60545963e-01 -9.43176225e-02 6.89850330e-01 -9.64622796e-01
-1.36948240e+00 6.98792100e-01 6.76267669e-02 -9.58283693e-02
1.87868088e-01 3.53694737e-01 -3.77606183e-01 4.79999691e-01
-2.39225805e-01 8.13027978e-01 1.08267260e+00 -1.02201378e+00
-6.57029450e-01 -4.76787597e-01 3.35578263e-01 4.09296960e-01
-5.42521954e-01 1.75123796e-01 -6.73311710e-01 -6.19568467e-01
2.23758370e-01 -3.96546870e-01 -2.10564047e-01 2.59862304e-01
5.30141555e-02 -7.01686561e-01 9.21705067e-01 -3.72872978e-01
1.51320136e+00 -2.20442820e+00 1.06688321e-01 6.57972991e-01
5.16515791e-01 2.11717486e-01 -3.98516476e-01 -6.11304166e-03
-1.54970512e-01 -2.14092463e-01 2.76842937e-02 5.29501066e-02
-3.18376064e-01 2.44166464e-01 3.48821044e-01 5.26603043e-01
6.03790462e-01 8.91356647e-01 -6.18954301e-01 -7.32303083e-01
2.92719543e-01 5.33133745e-01 -5.07569909e-01 1.02741979e-01
1.01373419e-02 4.80503261e-01 -7.87589371e-01 7.92004585e-01
6.83354974e-01 -2.12723255e-01 5.70045561e-02 -6.95645332e-01
5.95689528e-02 -6.98758115e-04 -1.03840494e+00 1.44652498e+00
-3.00873756e-01 3.63389552e-01 1.80036500e-01 -1.29132628e+00
1.37428641e+00 1.88880697e-01 7.86145568e-01 -6.35359943e-01
6.06586814e-01 -2.27350459e-01 2.40507036e-01 -1.69839710e-01
2.36968379e-02 1.48316115e-01 1.47947997e-01 3.70397061e-01
4.09362577e-02 4.71206576e-01 -1.11394279e-01 3.77531014e-02
1.08589423e+00 9.81613025e-02 3.99177432e-01 -1.52677655e-01
1.11997807e+00 -5.66284060e-01 7.59068787e-01 2.26957574e-01
-5.12792528e-01 4.70848382e-01 8.73629153e-01 -3.95556718e-01
-2.49650195e-01 -8.00783694e-01 -1.41607434e-01 1.09175706e+00
2.76384503e-01 -5.39660215e-01 -7.64961898e-01 -1.13598168e+00
-1.03817031e-01 -3.13256644e-02 -1.06546593e+00 -3.44416469e-01
-6.10238016e-01 -5.54172277e-01 3.16157401e-01 6.73192978e-01
9.61557925e-01 -1.13340318e+00 -2.84067571e-01 1.67535633e-01
2.01352954e-01 -1.17531264e+00 -7.43423998e-01 -2.42147334e-02
-6.26377940e-01 -1.19959068e+00 -4.31679010e-01 -1.04768014e+00
1.01024628e+00 3.05864364e-01 7.33258426e-01 1.92117184e-01
-5.58467090e-01 3.67857814e-01 -4.73157823e-01 6.78226799e-02
1.89997721e-02 3.84767801e-02 -2.98933566e-01 5.96940160e-01
3.61330986e-01 -6.73417926e-01 -7.59195149e-01 6.94467187e-01
-8.15158784e-01 -1.26028284e-01 7.85246730e-01 7.75080919e-01
8.16396415e-01 9.17479768e-02 6.33450210e-01 -7.77840555e-01
4.70075458e-01 -2.77689725e-01 -4.97949213e-01 4.69456315e-01
-2.53423452e-01 -2.17537269e-01 3.91390085e-01 -4.37465191e-01
-9.33558166e-01 1.07999168e-01 -7.36382678e-02 -6.41056597e-01
7.53053427e-02 4.52265441e-01 -5.19029677e-01 -5.48337519e-01
5.64631999e-01 -2.19126865e-02 1.50476754e-01 -7.02415183e-02
3.83172870e-01 5.98691225e-01 -7.07465084e-03 -5.91351151e-01
6.07912302e-01 2.86788255e-01 2.65977800e-01 -7.71140695e-01
-8.90109718e-01 -2.68094093e-01 -9.22122657e-01 -5.26517391e-01
8.76749635e-01 -9.29442883e-01 -6.80728316e-01 3.31003994e-01
-8.97276938e-01 -3.54304522e-01 -2.46322840e-01 2.56476313e-01
-3.66347015e-01 3.66318911e-01 -6.13723457e-01 -3.41054291e-01
-3.79716039e-01 -1.19863689e+00 9.23125982e-01 4.55006808e-01
1.34723678e-01 -8.69176507e-01 -4.99679565e-01 3.05701405e-01
-1.33647406e-02 2.92124093e-01 7.32294917e-01 -2.35516969e-02
-4.68907356e-01 -1.79646060e-01 -8.63171697e-01 4.85540837e-01
6.47783875e-01 3.21579874e-02 -6.65872216e-01 -9.17551219e-02
-2.95675725e-01 -3.44291121e-01 8.19827259e-01 2.77691841e-01
1.47522473e+00 -2.17455804e-01 -2.97429353e-01 5.26838064e-01
1.00771666e+00 -1.66483838e-02 7.01540232e-01 -9.93773434e-03
9.29672599e-01 6.91905558e-01 8.86290133e-01 6.00752652e-01
3.39629203e-01 7.36913025e-01 6.47698462e-01 -2.37309322e-01
-3.66107434e-01 7.95885623e-02 4.70647693e-01 4.29580927e-01
-4.02819008e-01 2.33859420e-01 -3.79283428e-01 1.35519668e-01
-1.89347827e+00 -8.02988231e-01 6.73119202e-02 1.83459640e+00
7.17297733e-01 9.81973708e-02 2.08707109e-01 -1.75295115e-01
7.90027320e-01 5.09776711e-01 -4.35565650e-01 -3.55228752e-01
-1.09595992e-01 1.99486971e-01 2.60540515e-01 3.22044492e-01
-1.06750894e+00 1.32388830e+00 5.41365194e+00 1.13262844e+00
-1.01482499e+00 1.30474284e-01 8.88273597e-01 2.54333526e-01
-1.22739218e-01 -2.33849913e-01 -9.68124926e-01 -1.25796432e-02
2.70729363e-01 2.33389288e-01 2.66889185e-01 6.62615955e-01
8.19722936e-02 1.40464053e-01 -7.54765809e-01 1.06022966e+00
2.29928821e-01 -1.16760564e+00 2.14576140e-01 9.33189765e-02
6.89252794e-01 -9.83732566e-02 -9.11609009e-02 1.07815914e-01
9.23814178e-02 -1.00593281e+00 1.88720092e-01 3.46573800e-01
8.68152380e-01 -1.03588820e+00 5.28188646e-01 -4.66674892e-03
-1.86755228e+00 -1.40401553e-02 -7.18357086e-01 1.10387489e-01
-3.50751728e-01 7.97373503e-02 -4.27317411e-01 5.19603789e-01
6.96561873e-01 1.26136100e+00 -6.70553923e-01 4.85251188e-01
-7.14695275e-01 3.87071818e-01 -3.10326338e-01 7.87957236e-02
5.87632477e-01 -5.05282938e-01 1.14346378e-01 9.34700191e-01
8.47099647e-02 3.87414783e-01 1.31102160e-01 7.52297938e-01
-2.95991778e-01 3.58617514e-01 -4.49285239e-01 2.23056242e-01
9.14943870e-03 1.67626643e+00 -5.82513273e-01 1.37033582e-01
-9.15798306e-01 8.46790552e-01 8.46416175e-01 1.83004797e-01
-1.00746119e+00 -4.52907905e-02 7.43410945e-01 1.12542965e-01
3.80498409e-01 -1.75590470e-01 1.21024683e-01 -9.24033999e-01
-1.84221387e-01 -7.12049425e-01 8.10849607e-01 -5.07352769e-01
-1.25437510e+00 6.45562232e-01 -2.08258554e-01 -1.43026233e+00
2.46839643e-01 -5.94437242e-01 -7.31840491e-01 6.31688595e-01
-1.76959133e+00 -1.60918963e+00 -3.58865023e-01 1.22785985e+00
4.35174495e-01 -2.31544718e-01 4.59469140e-01 2.44357467e-01
-8.15128803e-01 9.98926044e-01 -6.45766973e-01 4.56487328e-01
3.37117493e-01 -6.30034745e-01 -2.84649074e-01 7.70767212e-01
2.99941182e-01 3.60655427e-01 -6.51385412e-02 -5.11535287e-01
-1.05192184e+00 -1.32848847e+00 2.54121840e-01 1.08899407e-01
9.04008269e-01 -3.02206963e-01 -7.93145597e-01 6.93395257e-01
4.38367948e-02 7.46049523e-01 5.69627106e-01 -5.19215204e-02
-4.07322943e-01 -6.53037667e-01 -1.00796211e+00 6.24420643e-01
1.21976662e+00 -3.08609396e-01 -1.12147957e-01 2.14204907e-01
3.12597275e-01 -2.27081448e-01 -1.25381374e+00 5.88628471e-01
4.12693977e-01 -8.62100124e-01 8.21271896e-01 -4.32261169e-01
2.77567506e-01 -3.93835932e-01 -1.06787443e-01 -1.01125729e+00
-4.34409142e-01 -5.76730371e-01 -1.79958299e-01 1.41022062e+00
3.17505389e-01 -6.89973533e-01 9.88268971e-01 3.45954239e-01
-6.48927763e-02 -1.13679969e+00 -1.03202093e+00 -3.82490635e-01
-3.26812357e-01 -3.72914076e-01 5.33643126e-01 7.51690328e-01
1.75599363e-02 6.37455761e-01 -2.95278490e-01 3.43630493e-01
4.72951502e-01 2.24821344e-01 6.46270573e-01 -1.02975738e+00
-1.67968556e-01 -6.65411472e-01 -8.98477674e-01 -1.12420070e+00
3.48183602e-01 -9.51915622e-01 -2.31175989e-01 -1.62963772e+00
-9.46188420e-02 -6.91149354e-01 -6.96237206e-01 7.95452237e-01
-2.00144053e-01 4.61640626e-01 -6.84377775e-02 9.74831134e-02
-5.07310092e-01 8.28795016e-01 1.80152500e+00 -2.04022512e-01
-2.16755778e-01 5.30603677e-02 -5.80741942e-01 6.10482693e-01
8.85492742e-01 -6.56152815e-02 -5.94383717e-01 -1.45000279e-01
-1.73135325e-01 -1.05936222e-01 1.86902076e-01 -9.63800967e-01
1.66313246e-01 -1.77571774e-01 4.61882710e-01 -5.18463016e-01
4.67675984e-01 -1.05630136e+00 -5.43060124e-01 1.49609640e-01
-2.00920984e-01 -1.93456873e-01 1.83243603e-01 5.75896263e-01
-3.97649258e-01 2.24800780e-01 9.54237282e-01 -1.22659385e-01
-8.93897116e-01 1.03032148e+00 -1.61492854e-01 -2.84278154e-01
1.41297674e+00 -3.99204701e-01 -1.47872731e-01 -3.15346420e-01
-7.96279192e-01 4.39381957e-01 -7.88531303e-02 4.23326224e-01
9.01434839e-01 -1.31277692e+00 -5.63998878e-01 3.81241500e-01
3.97270381e-01 1.48254514e-01 4.44913089e-01 1.00387466e+00
-4.13098752e-01 6.67094141e-02 -3.88916641e-01 -4.49000955e-01
-1.45215809e+00 3.52094829e-01 5.45005918e-01 -3.11079890e-01
-5.69477379e-01 1.26826215e+00 5.76305807e-01 -2.44907826e-01
1.96652412e-01 1.43622514e-02 -7.33566582e-01 2.33659759e-01
4.37392414e-01 3.32316533e-02 -1.36468336e-01 -1.15518236e+00
-2.11524650e-01 9.54732358e-01 -2.46675700e-01 5.04651010e-01
1.25060833e+00 -1.86339721e-01 -4.78340030e-01 -5.71670718e-02
1.16405916e+00 -1.85268328e-01 -1.60809755e+00 -5.90362132e-01
-2.99414486e-01 -3.54228139e-01 1.92395419e-01 -2.51972914e-01
-1.67851186e+00 9.20195699e-01 5.07944345e-01 -1.61515310e-01
1.42866957e+00 3.56586009e-01 4.94841337e-01 -8.28054994e-02
2.42787063e-01 -1.00839865e+00 4.16328162e-01 2.83693433e-01
9.57479060e-01 -1.02148938e+00 1.16467938e-01 -1.09809148e+00
-5.63328862e-01 1.29264367e+00 1.30786169e+00 -2.65664041e-01
1.01950431e+00 1.31098628e-02 1.95335727e-02 -5.21314859e-01
-4.66190040e-01 -4.50759053e-01 6.50860906e-01 5.08213282e-01
4.09777075e-01 -2.25345120e-01 -3.80557179e-01 3.87438834e-01
2.06960469e-01 -3.51004332e-01 4.79009040e-02 6.35577500e-01
-3.66476655e-01 -1.15809560e+00 -7.17451647e-02 6.32442117e-01
-2.81679809e-01 -4.43496741e-02 -3.90415609e-01 9.25401390e-01
3.19265574e-01 8.94532621e-01 1.50039226e-01 -6.05217874e-01
1.82166860e-01 -4.10900086e-01 6.97186828e-01 -6.79525077e-01
-6.28646016e-01 2.95578241e-01 -8.72376189e-02 -7.88758695e-01
-6.42257869e-01 -6.17096364e-01 -1.60574126e+00 1.21539824e-01
-3.64386022e-01 -2.15332881e-02 1.11817777e-01 9.01867747e-01
3.86474997e-01 7.23442256e-01 9.42388654e-01 -5.49601197e-01
6.99369535e-02 -9.98986244e-01 -8.40328872e-01 1.57249905e-02
-3.01385280e-02 -8.62740755e-01 -1.53208286e-01 -3.53620797e-01] | [13.652281761169434, 1.5612372159957886] |
7186df4c-0175-43d5-8387-2eb08ce2aadc | gmote-gaussian-based-minority-oversampling | 2105.03855 | null | https://arxiv.org/abs/2105.03855v1 | https://arxiv.org/pdf/2105.03855v1.pdf | GMOTE: Gaussian based minority oversampling technique for imbalanced classification adapting tail probability of outliers | Classification of imbalanced data is one of the common problems in the recent field of data mining. Imbalanced data substantially affects the performance of standard classification models. Data-level approaches mainly use the oversampling methods to solve the problem, such as synthetic minority oversampling Technique (SMOTE). However, since the methods such as SMOTE generate instances by linear interpolation, synthetic data space may look like a polygonal. Also, the oversampling methods generate outliers of the minority class. In this paper, we proposed Gaussian based minority oversampling technique (GMOTE) with a statistical perspective for imbalanced datasets. To avoid linear interpolation and to consider outliers, this proposed method generates instances by the Gaussian Mixture Model. Motivated by clustering-based multivariate Gaussian outlier score (CMGOS), we propose to adapt tail probability of instances through the Mahalanobis distance to consider local outliers. The experiment was carried out on a representative set of benchmark datasets. The performance of the GMOTE is compared with other methods such as SMOTE. When the GMOTE is combined with classification and regression tree (CART) or support vector machine (SVM), it shows better accuracy and F1-Score. Experimental results demonstrate the robust performance. | ['Kyung Joon Cha', 'Seung Jee Yang'] | 2021-05-09 | null | null | null | null | ['classification'] | ['methodology'] | [ 5.36693784e-04 -2.07454890e-01 -1.13960020e-01 -4.46465641e-01
-4.78516251e-01 1.36757299e-01 4.09778118e-01 5.25060892e-01
-1.64227948e-01 9.78684604e-01 -1.89494863e-01 -1.04767159e-01
-1.90300167e-01 -1.01878941e+00 -5.20739734e-01 -8.34338307e-01
1.58206090e-01 5.45660377e-01 1.30831376e-01 4.07673903e-02
5.99241793e-01 2.59019732e-01 -1.91846001e+00 5.38812518e-01
1.61125326e+00 9.65183437e-01 -4.39244628e-01 1.89704955e-01
-5.17585933e-01 4.79008734e-01 -9.59352076e-01 -2.33416453e-01
4.19295222e-01 -3.75233680e-01 -2.98043787e-01 5.28508015e-02
5.98346144e-02 9.15855020e-02 2.14314401e-01 1.10941541e+00
3.75215292e-01 1.07590064e-01 9.46912289e-01 -1.86049104e+00
-2.65972853e-01 2.53512412e-01 -1.04828203e+00 1.09940067e-01
1.14596123e-02 -2.32775003e-01 2.31590763e-01 -9.70524609e-01
2.74624139e-01 1.32419884e+00 9.13691342e-01 -3.57163064e-02
-1.09116662e+00 -8.64271224e-01 -8.49052519e-02 4.50548142e-01
-1.54289639e+00 5.01151849e-03 7.68747985e-01 -4.51600194e-01
4.58693177e-01 4.33340520e-01 4.33578581e-01 4.78382647e-01
3.28046054e-01 6.48662686e-01 1.30685627e+00 -3.89340907e-01
6.17789268e-01 2.64066577e-01 3.66272330e-01 -1.53770577e-03
9.21558321e-01 -1.91696823e-01 -3.13686460e-01 -5.42055964e-01
1.53202206e-01 4.70905066e-01 1.56664938e-01 -3.61233652e-01
-7.26744056e-01 9.30517435e-01 2.30294511e-01 1.16646685e-01
-2.65716195e-01 -4.89590675e-01 7.50797689e-01 2.03292057e-01
6.96537554e-01 -4.52522300e-02 -3.34129572e-01 1.29203200e-01
-1.27811050e+00 4.79006618e-01 6.60444319e-01 8.76783133e-01
6.03489459e-01 1.54642925e-01 -3.35348397e-02 8.61626863e-01
2.42602065e-01 3.01521838e-01 8.62455428e-01 -3.22351784e-01
6.92458630e-01 1.12164676e+00 7.83194900e-02 -1.31587362e+00
-4.47309017e-01 -6.38404012e-01 -1.09146810e+00 4.24827129e-01
4.66860324e-01 1.29803643e-01 -8.23628843e-01 9.97812927e-01
6.77740335e-01 1.94064662e-01 7.97080621e-02 6.48810148e-01
6.67887986e-01 6.58340991e-01 6.78806305e-02 -5.36814392e-01
9.27120030e-01 -6.65146887e-01 -7.87220299e-01 2.21319571e-01
7.61646450e-01 -8.17945540e-01 8.66215944e-01 8.94488752e-01
-6.90909088e-01 -6.28960311e-01 -1.03630543e+00 4.33650643e-01
-4.54915792e-01 -3.87797244e-02 2.62494713e-01 9.21822667e-01
-2.23568410e-01 6.31550491e-01 -7.41470695e-01 -3.10992926e-01
5.37937403e-01 1.64124027e-01 -3.72470707e-01 -2.06086472e-01
-8.94240320e-01 6.36558950e-01 7.04693198e-01 -2.45432518e-02
-1.82468385e-01 -6.22864664e-01 -7.36809194e-01 -2.70043314e-01
-1.31325005e-02 -3.37473929e-01 6.61032677e-01 -1.01035988e+00
-9.97143269e-01 5.19458890e-01 -1.20408826e-01 -5.89513838e-01
7.28435338e-01 -9.93201435e-02 -4.74779218e-01 -2.44191334e-01
3.89432013e-02 -1.74340799e-01 5.88789463e-01 -1.26411903e+00
-6.34933949e-01 -9.00684118e-01 -7.35286951e-01 5.63590825e-02
1.00872191e-02 -2.41271164e-02 2.91723162e-01 -7.46631384e-01
7.45059729e-01 -6.92630827e-01 -3.46635342e-01 -4.89649147e-01
-6.68218553e-01 -1.94792628e-01 9.71536219e-01 -6.84953272e-01
1.49171019e+00 -2.14365077e+00 -4.17788237e-01 5.64977229e-01
-2.56588589e-02 2.59878725e-01 5.72047770e-01 5.30420482e-01
-2.42550641e-01 1.17942549e-01 -6.19089544e-01 -8.01342875e-02
-2.05474570e-01 8.52959901e-02 -8.23246241e-02 8.28990161e-01
6.27736822e-02 9.96448398e-02 -6.04100287e-01 -5.35710633e-01
1.93414897e-01 1.73394278e-01 -5.84289670e-01 6.98701367e-02
4.21828665e-02 4.23511595e-01 -2.59786129e-01 7.81718314e-01
1.52443898e+00 2.69625723e-01 -2.82506257e-01 -2.13981852e-01
-1.28946109e-02 -1.70230642e-01 -1.84096444e+00 9.31687772e-01
-8.37906003e-02 8.87754187e-03 -3.38236123e-01 -1.34585452e+00
1.54581285e+00 1.10773459e-01 4.11942393e-01 -3.55211586e-01
8.41908827e-02 6.45138860e-01 -3.55130397e-02 -5.71271837e-01
3.55452299e-01 -2.54550457e-01 8.56667385e-02 -5.70714772e-02
-3.23811650e-01 2.93087238e-03 1.34365216e-01 -1.47992611e-01
7.28242338e-01 1.53498799e-01 5.51891148e-01 -3.12307328e-01
6.62102997e-01 2.48191759e-01 9.75238323e-01 4.51173812e-01
-1.29077554e-01 7.85936177e-01 4.57585543e-01 -5.76423526e-01
-1.09732127e+00 -9.02064860e-01 -6.18848622e-01 3.65218222e-01
8.94666687e-02 -1.76074699e-01 -6.75826669e-01 -6.21519625e-01
3.26384574e-01 7.92493284e-01 -3.76850456e-01 -2.63091743e-01
-3.41459960e-01 -1.51753867e+00 4.16868240e-01 1.61497995e-01
7.27478504e-01 -6.92702591e-01 -2.35565647e-01 2.38571480e-01
-3.53099778e-02 -6.23106658e-01 1.11965574e-01 -9.37732961e-03
-1.26431453e+00 -1.24203491e+00 -4.68820959e-01 -3.87895226e-01
7.95545399e-01 4.56472114e-02 8.85673463e-01 4.64851074e-02
-3.81421953e-01 -6.17187262e-01 -6.92569375e-01 -6.63125515e-01
-3.34820092e-01 -1.75033018e-01 2.90904045e-01 3.02906781e-01
8.42049479e-01 -6.17194235e-01 -6.61114573e-01 4.65430409e-01
-1.04810870e+00 -2.99739063e-01 3.92847598e-01 7.87274778e-01
7.00001895e-01 5.35223901e-01 9.40151453e-01 -1.13190186e+00
5.24790466e-01 -9.83044922e-01 -3.90293211e-01 -1.85775861e-01
-8.14747155e-01 -3.03756684e-01 9.13504958e-01 -2.81548917e-01
-8.76038730e-01 -3.13265234e-01 1.91466250e-02 -3.28848928e-01
-3.83128285e-01 3.81296426e-01 -4.34948564e-01 2.30237842e-01
7.70551205e-01 1.61031559e-01 5.45686148e-02 -6.83021426e-01
-2.35626265e-01 1.22112346e+00 2.36162975e-01 -4.26910639e-01
6.29961610e-01 5.15664935e-01 1.05601110e-01 -7.64091432e-01
-5.37339449e-01 -5.90038061e-01 -3.95365000e-01 5.63221239e-02
5.08484185e-01 -8.30023229e-01 -6.15832329e-01 6.48084939e-01
-7.80561686e-01 2.93210775e-01 -1.06439382e-01 7.60178089e-01
-4.47886646e-01 3.29044998e-01 -1.73627689e-01 -1.30832112e+00
-2.00348705e-01 -1.10186052e+00 6.84581161e-01 2.95687556e-01
-2.10939541e-01 -6.32208586e-01 -9.59522575e-02 4.92459238e-01
2.11355731e-01 1.04781592e+00 7.95573652e-01 -1.18296707e+00
-1.94596916e-01 -5.52917957e-01 -9.84745771e-02 5.22666633e-01
5.08977920e-02 3.73125672e-01 -9.11485434e-01 -1.72850773e-01
3.14211130e-01 1.51317984e-01 5.54088950e-01 3.49629372e-01
1.46871734e+00 -3.91787142e-01 -2.55303174e-01 6.71923876e-01
1.62134027e+00 3.49755138e-01 8.46995413e-01 6.56542778e-01
4.84137863e-01 6.80088341e-01 1.12839246e+00 6.90289378e-01
2.79262871e-01 4.22435582e-01 4.32214856e-01 4.30676565e-02
3.63728732e-01 -2.53718913e-01 1.18447743e-01 8.46759915e-01
1.04919247e-01 1.52093217e-01 -9.64309573e-01 4.49867874e-01
-1.89264452e+00 -9.44607913e-01 -7.99662054e-01 2.69476604e+00
6.64917886e-01 3.17384750e-01 2.71746784e-01 1.05861628e+00
1.06888008e+00 -2.14516506e-01 -3.94813657e-01 -7.46772110e-01
-1.35405257e-01 1.01450011e-02 5.91128111e-01 2.33692661e-01
-1.00459695e+00 2.72844911e-01 5.12052011e+00 1.22627258e+00
-7.50340879e-01 2.34589586e-03 9.11971152e-01 4.05772440e-02
-1.34169549e-01 2.78468039e-02 -8.49944711e-01 1.10796952e+00
7.12720275e-01 -2.86307722e-01 -1.33302107e-01 8.86956096e-01
4.39948976e-01 -4.37898368e-01 -6.52261376e-01 9.42470133e-01
1.15026303e-01 -7.42960989e-01 4.40321341e-02 -3.93412150e-02
9.70162272e-01 -3.47562522e-01 -1.70893863e-01 3.73777717e-01
-1.38263106e-01 -1.07636762e+00 3.76114637e-01 5.66649497e-01
4.13969547e-01 -1.12237942e+00 1.14409602e+00 7.31027126e-01
-8.03476572e-01 -1.69733658e-01 -6.64900243e-01 -2.51768440e-01
-2.13731483e-01 1.42863405e+00 -7.48690486e-01 8.48264575e-01
7.74309337e-01 4.25024956e-01 -5.66215932e-01 1.55223227e+00
3.77998382e-01 5.83116829e-01 -5.00830173e-01 1.15583465e-01
-1.37317657e-01 -6.64435208e-01 6.20294333e-01 8.41727793e-01
5.04015744e-01 -1.86612993e-01 1.06907576e-01 6.43268645e-01
2.58815974e-01 7.45180964e-01 -6.23828828e-01 5.61935246e-01
6.37792826e-01 8.41087937e-01 -6.51721895e-01 -3.67112577e-01
-2.10773319e-01 5.19258201e-01 -2.49136955e-01 1.49156928e-01
-7.14968204e-01 -6.96491659e-01 3.05711269e-01 4.08374935e-01
-7.37764239e-02 9.91148353e-02 -8.52734089e-01 -8.83955419e-01
4.27356005e-01 -1.06252956e+00 5.36171019e-01 -3.99548590e-01
-1.42943275e+00 5.72264910e-01 1.84087604e-02 -1.78179634e+00
2.07978874e-01 -2.38083497e-01 -8.65328431e-01 8.50696325e-01
-1.15108585e+00 -7.01563895e-01 -5.71951985e-01 3.92963588e-01
5.00126719e-01 -2.35443175e-01 4.92273599e-01 4.88312066e-01
-6.70545876e-01 5.45679331e-01 6.32415712e-01 -2.86231816e-01
8.11419070e-01 -1.05427217e+00 4.58350638e-03 6.90510690e-01
-4.52253819e-01 3.88642848e-01 1.10447311e+00 -9.14666891e-01
-7.13500500e-01 -1.26902127e+00 7.25066066e-01 -2.02551812e-01
1.29296392e-01 -4.36198525e-02 -1.19129360e+00 3.62656891e-01
-3.32121253e-01 2.42362633e-01 8.61138225e-01 -1.85175836e-02
-1.36814669e-01 -4.44190383e-01 -1.88381135e+00 1.51328713e-01
4.42684710e-01 1.35277137e-01 -6.21974587e-01 4.27478105e-01
4.02627617e-01 -2.96904713e-01 -9.91128743e-01 8.00278723e-01
3.45920801e-01 -1.42638755e+00 7.42441595e-01 -6.09913111e-01
3.78241867e-01 -7.36116230e-01 -2.36461356e-01 -1.33645976e+00
1.30690515e-01 -7.92953372e-02 -1.09145842e-01 1.40658367e+00
2.23161563e-01 -8.51593733e-01 8.51931095e-01 4.47757602e-01
1.14780910e-01 -1.01565659e+00 -1.03559315e+00 -9.91096020e-01
2.05159754e-01 -2.81684637e-01 9.62441564e-01 9.10116017e-01
-3.28193568e-02 -4.33875054e-01 -2.08318964e-01 1.42054543e-01
9.44601595e-01 1.43692028e-02 1.07141936e+00 -1.55117655e+00
6.39822930e-02 -1.80417858e-02 -8.58263314e-01 -1.23103894e-01
-2.84284890e-01 -8.67012501e-01 -3.55156541e-01 -1.38030231e+00
6.22911192e-03 -7.70211160e-01 -2.68974483e-01 -1.53832540e-01
-3.27894866e-01 3.56951267e-01 -7.99305290e-02 3.62778902e-01
-2.53553480e-01 6.22442603e-01 8.57368290e-01 1.02569073e-01
-3.85833740e-01 4.18486983e-01 -5.46209812e-01 9.29430544e-01
1.06430793e+00 -7.91646302e-01 -1.95683107e-01 2.22943351e-01
3.92683670e-02 -1.44550323e-01 6.18081242e-02 -1.29245424e+00
4.76562232e-02 -3.03139478e-01 5.83440125e-01 -9.51755464e-01
-1.80267498e-01 -9.50549841e-01 4.41757828e-01 5.46219110e-01
7.28572011e-02 1.44645303e-01 -1.94421820e-02 5.98562956e-01
-5.77358305e-01 -2.04550967e-01 9.94233131e-01 9.85915363e-02
-2.09820941e-02 1.48993403e-01 -1.10288508e-01 1.18422680e-01
1.40524411e+00 -6.72867179e-01 -2.09296823e-01 3.49852890e-02
-5.23487270e-01 2.24166349e-01 6.02742493e-01 2.31910095e-01
5.43949246e-01 -1.47242725e+00 -8.83466005e-01 4.37611490e-01
3.26936007e-01 3.01913798e-01 2.97017843e-01 1.05830896e+00
-7.64397383e-01 -6.09049648e-02 -1.80447280e-01 -7.25012898e-01
-1.09800732e+00 6.08144760e-01 1.52193248e-01 -1.26335993e-01
-3.65052015e-01 4.05495375e-01 -1.92619979e-01 -7.25948274e-01
6.19339347e-02 -3.23441297e-01 -3.28317553e-01 1.38495684e-01
5.37395418e-01 9.46350455e-01 3.96783471e-01 -5.03317654e-01
-4.07769114e-01 3.76719832e-01 5.72186187e-02 2.75313973e-01
1.16075397e+00 1.19831800e-01 -3.90612036e-01 7.28854656e-01
1.18101740e+00 1.89518213e-01 -7.07217097e-01 -1.28594805e-02
2.09793732e-01 -9.34051573e-01 -3.19469362e-01 -3.53870273e-01
-8.56325090e-01 7.02843666e-01 6.76484942e-01 3.45050246e-01
1.01267946e+00 -6.70007467e-01 6.36169612e-01 -4.91370261e-02
3.70582283e-01 -1.27657962e+00 -4.24081624e-01 1.32622540e-01
6.94551528e-01 -1.34632015e+00 1.54354781e-01 -4.68525022e-01
-6.13700211e-01 1.06273150e+00 7.48830795e-01 -5.89057207e-01
9.13827240e-01 1.41556293e-01 -3.98621298e-02 1.77911118e-01
-4.18491155e-01 2.24228159e-01 -4.11844701e-02 5.36229670e-01
3.41368854e-01 1.12870559e-01 -1.06142414e+00 9.39751983e-01
-3.17223877e-01 2.61117280e-01 5.94943702e-01 8.44889939e-01
-5.92127562e-01 -9.35464263e-01 -9.76105630e-01 9.19487596e-01
-5.99824727e-01 1.68262914e-01 1.43555433e-01 6.93494797e-01
6.76201165e-01 1.08267486e+00 1.68759316e-01 -2.29759440e-01
4.70997512e-01 2.91110337e-01 -9.02020093e-03 -3.39942396e-01
-4.74553794e-01 -2.33802855e-01 -2.03221530e-01 -3.56532097e-01
-2.34468892e-01 -6.77888989e-01 -1.23468196e+00 -3.71080101e-01
-5.13326168e-01 4.54576880e-01 9.62826908e-01 8.13354492e-01
2.14407653e-01 4.39129770e-01 9.21911716e-01 -3.88821125e-01
-6.75784826e-01 -1.05427015e+00 -8.95627499e-01 6.86386287e-01
7.12780356e-02 -7.66476989e-01 -8.07266891e-01 -1.94789916e-01] | [8.579293251037598, 4.184606075286865] |
040e2d80-cb0b-4915-a6d8-efe5d12ded78 | adaptive-depth-graph-attention-networks | 2301.06265 | null | https://arxiv.org/abs/2301.06265v1 | https://arxiv.org/pdf/2301.06265v1.pdf | Adaptive Depth Graph Attention Networks | As one of the most popular GNN architectures, the graph attention networks (GAT) is considered the most advanced learning architecture for graph representation and has been widely used in various graph mining tasks with impressive results. However, since GAT was proposed, none of the existing studies have provided systematic insight into the relationship between the performance of GAT and the number of layers, which is a critical issue in guiding model performance improvement. In this paper, we perform a systematic experimental evaluation and based on the experimental results, we find two important facts: (1) the main factor limiting the accuracy of the GAT model as the number of layers increases is the oversquashing phenomenon; (2) among the previous improvements applied to the GNN model, only the residual connection can significantly improve the GAT model performance. We combine these two important findings to provide a theoretical explanation that it is the residual connection that mitigates the loss of original feature information due to oversquashing and thus improves the deep GAT model performance. This provides empirical insights and guidelines for researchers to design the GAT variant model with appropriate depth and well performance. To demonstrate the effectiveness of our proposed guidelines, we propose a GAT variant model-ADGAT that adaptively selects the number of layers based on the sparsity of the graph, and experimentally demonstrate that the effectiveness of our model is significantly improved over the original GAT. | ['Rui Zhang', 'Ruqiong Zhang', 'Yixuan Du', 'Jingbo Zhou'] | 2023-01-16 | null | null | null | null | ['graph-mining'] | ['graphs'] | [-1.18391484e-01 1.77737489e-01 -4.89292324e-01 3.03784329e-02
2.29624003e-01 -5.47600677e-03 1.79438323e-01 1.07329100e-01
-2.63500839e-01 3.96866828e-01 1.93472177e-01 -4.72192079e-01
-2.51404315e-01 -1.05668640e+00 -6.14493668e-01 -6.03808463e-01
-1.78033724e-01 2.56923381e-02 3.95280182e-01 -3.63021672e-01
2.10925460e-01 3.92567545e-01 -1.15767777e+00 -7.20226252e-03
1.13945961e+00 9.50938046e-01 1.97714448e-01 2.64979124e-01
-2.01598957e-01 7.55675375e-01 -5.31212211e-01 -4.70793843e-01
3.69090110e-01 -5.24900675e-01 -6.03721857e-01 -1.42074019e-01
-1.16375787e-03 -2.66532600e-01 -5.31391263e-01 1.22644639e+00
3.49122494e-01 3.25497091e-02 4.52550709e-01 -1.21065474e+00
-8.18951070e-01 1.16491902e+00 -7.39191055e-01 5.87042451e-01
-2.30363011e-01 5.27343936e-02 1.29713917e+00 -4.82254297e-01
2.50920475e-01 1.19721174e+00 6.34809971e-01 2.70099759e-01
-9.29652274e-01 -8.87407959e-01 7.01315165e-01 2.94848502e-01
-1.48114944e+00 5.60518587e-03 1.02411342e+00 -5.16230911e-02
9.62264359e-01 -8.04629922e-02 9.88478720e-01 8.54058743e-01
3.28610480e-01 7.36787319e-01 6.57754421e-01 -4.87046182e-01
1.29517838e-02 5.89945875e-02 3.51655245e-01 8.91168594e-01
8.55448008e-01 -1.01680614e-01 -3.67381960e-01 1.40425682e-01
1.13162279e+00 -8.02338943e-02 -2.67059833e-01 -3.44213337e-01
-5.92659652e-01 1.06930852e+00 1.16328764e+00 4.31940377e-01
-3.00338686e-01 4.08073127e-01 3.68601233e-01 2.86545187e-01
5.04948735e-01 6.74583793e-01 -3.11969846e-01 1.00706130e-01
-6.41681135e-01 -2.73769256e-02 4.67848539e-01 6.90761685e-01
6.39581442e-01 4.08382356e-01 -6.51257336e-02 7.77276874e-01
4.53785002e-01 -1.77536104e-02 4.62401778e-01 -2.40120217e-01
4.88442361e-01 1.31244707e+00 -5.80823779e-01 -1.33094454e+00
-4.28487867e-01 -1.13997114e+00 -9.85441208e-01 -1.85682893e-01
1.76652044e-01 -2.67811447e-01 -8.78686011e-01 1.86997604e+00
-3.04393005e-02 2.18202114e-01 -2.45234832e-01 8.62606406e-01
1.00843227e+00 3.91022027e-01 1.27373680e-01 9.55841392e-02
1.17310154e+00 -1.06599081e+00 -6.47777200e-01 -5.23982584e-01
8.04845572e-01 -3.33555549e-01 1.17361522e+00 1.07985474e-01
-7.30458558e-01 -4.59985107e-01 -1.43864620e+00 7.65480027e-02
-3.24045300e-01 -8.51972699e-02 1.13514066e+00 6.90889895e-01
-1.13633490e+00 7.11461127e-01 -7.47783661e-01 -5.37881613e-01
4.84375179e-01 5.53211987e-01 -1.10607423e-01 -9.01943445e-02
-1.35080850e+00 7.01907575e-01 4.86565024e-01 2.10193887e-01
-5.08085668e-01 -3.61872286e-01 -7.13112831e-01 4.88402426e-01
5.25968075e-01 -6.81858301e-01 7.49549270e-01 -1.08416879e+00
-1.03929031e+00 3.46120119e-01 4.31759208e-02 -6.55566752e-01
1.30251301e-02 -1.24790110e-01 -2.05616534e-01 1.92507021e-02
-2.81338722e-01 4.00925249e-01 4.75829303e-01 -1.07383931e+00
-3.41697454e-01 -4.64686841e-01 3.57917428e-01 3.03404063e-01
-7.14184463e-01 -3.07432562e-01 -7.52911985e-01 -7.28412092e-01
2.06678331e-01 -7.62235701e-01 -3.36450756e-01 -4.05957460e-01
-4.95875478e-01 -3.22592169e-01 5.91859281e-01 -4.02129740e-01
1.71245635e+00 -2.16126561e+00 6.14290647e-02 2.48674110e-01
5.88242590e-01 4.48078126e-01 -2.20422834e-01 5.09037197e-01
-3.38009559e-02 4.83520269e-01 5.97552843e-02 7.20476061e-02
-2.34101117e-01 6.97001517e-02 -1.67623043e-01 2.80669034e-01
1.17915519e-01 1.24053335e+00 -7.56564260e-01 -1.88207418e-01
1.41386718e-01 3.99818748e-01 -6.29460156e-01 1.28037840e-01
-1.17617629e-01 2.38401413e-01 -7.11500883e-01 2.51735747e-01
6.12534940e-01 -6.05069101e-01 3.99612188e-01 -3.78182739e-01
2.05251068e-01 4.95527446e-01 -9.39013600e-01 1.25228488e+00
2.31527630e-02 5.53242385e-01 -1.57688767e-01 -1.08074224e+00
1.05491519e+00 3.45431007e-02 2.77935594e-01 -8.36746871e-01
3.72993827e-01 -5.84158674e-03 5.88426352e-01 -2.69164771e-01
4.37313348e-01 7.96893910e-02 1.98985934e-01 4.06018674e-01
-7.84570500e-02 3.96279126e-01 1.87189937e-01 4.44989711e-01
1.10522521e+00 -2.92340040e-01 4.23610926e-01 -4.57498997e-01
2.76636004e-01 -3.34192216e-01 5.74256241e-01 7.25488782e-01
-2.22587287e-01 2.16132507e-01 9.72789109e-01 -5.47932744e-01
-8.11361969e-01 -5.08745253e-01 2.75636911e-01 1.04655719e+00
2.34597042e-01 -6.70442164e-01 -8.24418128e-01 -7.57989109e-01
-1.06374413e-01 4.10189688e-01 -8.26399863e-01 -5.80803812e-01
-5.26088417e-01 -1.07458329e+00 5.53238451e-01 7.08270133e-01
7.96346009e-01 -1.08568668e+00 -1.55745164e-01 2.01157499e-02
9.94980931e-02 -1.02100480e+00 -2.97228783e-01 8.40701982e-02
-1.13728201e+00 -1.21384275e+00 -2.93191165e-01 -7.96511233e-01
8.77688348e-01 6.47457242e-01 9.60503757e-01 8.81676197e-01
2.07283020e-01 -1.69587418e-01 -5.40235162e-01 -3.64239067e-01
-1.78340867e-01 6.14886522e-01 -2.76070893e-01 -1.59295112e-01
3.34673643e-01 -7.27768481e-01 -6.19627893e-01 6.29626662e-02
-8.49022567e-01 2.16518939e-01 8.21750879e-01 6.31821334e-01
1.16357505e-01 3.78648460e-01 7.77745068e-01 -1.24945354e+00
9.42947149e-01 -3.22519273e-01 -4.56760854e-01 -2.32760352e-03
-1.03315449e+00 2.48273566e-01 7.47394323e-01 -1.24391586e-01
-4.49228346e-01 -6.40113115e-01 -1.56404406e-01 -2.95311868e-01
3.58576268e-01 9.92818713e-01 -9.95457545e-02 -3.60508233e-01
2.89536148e-01 4.47782539e-02 -1.85149357e-01 -4.74428862e-01
6.52047768e-02 1.89158320e-01 -9.25095677e-02 -3.65648448e-01
5.37262082e-01 1.95885688e-01 -2.78997254e-02 -7.70541489e-01
-9.37898636e-01 -9.81604233e-02 -4.15589690e-01 2.09415657e-03
5.77714086e-01 -8.02649081e-01 -6.50809526e-01 3.99262667e-01
-8.39605927e-01 -4.68360841e-01 1.18210554e-01 3.45309973e-01
1.01482444e-01 4.31997150e-01 -7.54610419e-01 -7.74863839e-01
-6.07008696e-01 -1.18902385e+00 3.93572509e-01 3.28249812e-01
2.10430170e-03 -1.21001029e+00 -3.34547609e-01 1.08578444e-01
5.64920843e-01 1.46163240e-01 1.45073628e+00 -6.68144226e-01
-5.90783954e-01 -2.41655875e-02 -6.12778604e-01 2.03711256e-01
4.79617454e-02 -9.18125808e-02 -5.98750889e-01 -4.97013420e-01
-1.53802425e-01 4.99944389e-02 1.07663751e+00 4.91384387e-01
1.31506979e+00 -3.04001600e-01 -4.22739357e-01 8.02810907e-01
1.59189200e+00 9.71664116e-02 7.00379789e-01 3.93298447e-01
1.04150903e+00 2.11500525e-01 1.19651236e-01 8.39760602e-02
5.73385954e-01 4.29363817e-01 8.82227242e-01 -2.96346247e-01
-2.64795631e-01 -5.41156292e-01 2.92727858e-01 1.06701660e+00
-2.02729687e-01 -4.00387049e-01 -7.94190407e-01 5.92673779e-01
-1.89194310e+00 -4.24564034e-01 -5.96875921e-02 2.22980857e+00
3.67241353e-01 4.81671274e-01 4.08658758e-02 1.89146295e-01
6.86812699e-01 4.19206858e-01 -6.38502002e-01 -3.66036743e-01
-1.89047530e-02 2.01567769e-01 6.31345510e-01 1.89101472e-01
-7.33298957e-01 1.15547049e+00 6.88976908e+00 7.35655606e-01
-1.26455700e+00 -1.99187592e-01 6.12174213e-01 9.50327367e-02
-2.99661696e-01 2.32873093e-02 -6.89641297e-01 4.47529733e-01
7.65339673e-01 -2.90313929e-01 4.88178492e-01 6.85150027e-01
2.78359354e-01 1.76176757e-01 -9.72560227e-01 7.04767108e-01
-4.90626844e-04 -1.25549650e+00 4.48544174e-01 2.97175348e-01
6.42289877e-01 1.25250921e-01 1.09490314e-02 5.13318241e-01
3.55560184e-01 -1.15876436e+00 4.38932747e-01 -4.20905314e-02
3.67530167e-01 -8.96343589e-01 9.52701211e-01 2.81124234e-01
-1.29332530e+00 -3.54235232e-01 -5.24707019e-01 -3.23430121e-01
-7.98505321e-02 6.43043816e-01 -8.19077969e-01 7.05119014e-01
5.69519699e-01 7.84636974e-01 -8.67484808e-01 1.06471145e+00
-5.76237381e-01 1.00603282e+00 -1.58772320e-02 -1.01276636e-01
4.17089194e-01 -2.81874716e-01 4.48345721e-01 9.14769888e-01
1.96925953e-01 -1.12165272e-01 1.46405563e-01 9.49320316e-01
-3.25927705e-01 1.17587522e-01 -5.79724491e-01 -3.72000188e-01
6.13293946e-01 1.08407080e+00 -9.29868698e-01 -1.61125101e-02
-4.46557760e-01 3.51871520e-01 7.70356536e-01 4.46653664e-01
-8.12792182e-01 -3.39585572e-01 4.89256233e-01 4.53736693e-01
5.19546628e-01 -2.74037927e-01 -5.01205921e-01 -9.11614835e-01
-9.93080288e-02 -8.08660388e-01 5.48694313e-01 -5.45283139e-01
-1.17073691e+00 7.50971019e-01 -2.29180485e-01 -5.88278174e-01
2.16951564e-01 -5.04616022e-01 -7.90843844e-01 7.50305653e-01
-1.63869059e+00 -1.02095401e+00 -2.82429248e-01 4.05338019e-01
2.72631407e-01 -8.58452916e-02 5.73075414e-01 4.02766824e-01
-1.07296550e+00 9.31346178e-01 -2.79136777e-01 4.04177070e-01
2.20040172e-01 -9.82223988e-01 7.14636981e-01 1.13688934e+00
9.73599590e-03 9.74618793e-01 5.07649899e-01 -7.20902741e-01
-1.28855264e+00 -1.12386930e+00 4.75361437e-01 1.35658070e-01
5.90256095e-01 -3.98608446e-01 -1.04758859e+00 8.41042697e-01
1.44671217e-01 -2.05894560e-01 4.41617906e-01 3.85480642e-01
-2.48945832e-01 -2.73939013e-01 -8.00541461e-01 6.65682197e-01
1.08107924e+00 -1.30047038e-01 -2.19636798e-01 -1.33756518e-01
1.03376055e+00 -2.92711467e-01 -6.50154948e-01 5.20383596e-01
2.91306525e-01 -1.13234735e+00 6.68522775e-01 -5.27159452e-01
3.19360852e-01 -4.97578941e-02 9.94108990e-02 -1.41856134e+00
-7.06126511e-01 -3.46879363e-01 -2.72686183e-01 9.80952859e-01
4.24084961e-01 -9.24418688e-01 9.26815033e-01 2.32749104e-01
-1.05909027e-01 -1.28162134e+00 -5.08112729e-01 -5.89039207e-01
4.08154763e-02 -2.65177429e-01 9.49488938e-01 8.65723193e-01
-1.95108697e-01 6.44045353e-01 -2.77911127e-01 1.47518277e-01
3.48098487e-01 -2.62867454e-02 6.77279830e-01 -1.27825499e+00
-1.13711812e-01 -5.44117033e-01 -4.92404938e-01 -1.28432035e+00
-1.35665193e-01 -9.79203820e-01 -4.39163029e-01 -1.89335072e+00
3.70883912e-01 -5.87032974e-01 -5.71080565e-01 5.09438753e-01
-4.23763543e-01 7.71333054e-02 2.24154443e-01 1.49613768e-01
-5.43075204e-01 5.95152676e-01 1.43961155e+00 1.66403651e-01
-2.22684771e-01 -5.00514433e-02 -1.47994030e+00 6.18504703e-01
7.38943994e-01 -3.16520751e-01 -6.87999845e-01 -6.83146060e-01
5.88732243e-01 -4.59440559e-01 1.41004592e-01 -7.51127779e-01
9.43685547e-02 -5.80246635e-02 1.93465620e-01 -2.88382083e-01
-9.58040655e-02 -6.36020780e-01 -2.08834499e-01 5.52222550e-01
-1.38362974e-01 2.39763737e-01 3.08929175e-01 6.51643276e-01
-8.45132098e-02 -2.97919158e-02 7.40729690e-01 -1.54242054e-01
-7.44902372e-01 5.60846925e-01 -1.33052960e-01 -5.75236529e-02
6.67034209e-01 -3.60354275e-01 -4.57266599e-01 -4.87701535e-01
-2.35592976e-01 3.41391832e-01 3.90596241e-01 5.59138119e-01
4.14587051e-01 -1.30443668e+00 -6.12650216e-01 4.20277745e-01
-1.13176964e-01 9.02186409e-02 1.54803455e-01 8.30814600e-01
-4.79345769e-01 5.47858953e-01 -4.50498350e-02 -3.33725095e-01
-9.18809950e-01 5.14503121e-01 3.99560213e-01 -6.98845088e-01
-6.99291706e-01 8.59892011e-01 5.12530386e-01 -9.24282894e-02
3.55443239e-01 -4.03977007e-01 -5.23370862e-01 -4.01298255e-01
3.57354015e-01 2.29509324e-01 2.16300488e-01 -3.72520328e-01
-2.68417239e-01 3.24481636e-01 -3.39143991e-01 5.21901131e-01
1.54851627e+00 -9.09799933e-02 -2.24336743e-01 2.11856842e-01
7.62727618e-01 -9.36571956e-02 -1.09683228e+00 -2.27252215e-01
-2.60294855e-01 -3.21720928e-01 2.40703702e-01 -5.13393641e-01
-1.65851116e+00 8.76129746e-01 1.30562261e-01 4.98130679e-01
1.13522995e+00 -2.04654947e-01 7.65194535e-01 1.22575806e-02
3.13693434e-01 -7.14552939e-01 2.10016161e-01 6.52206540e-01
6.90680981e-01 -9.10241902e-01 2.36886472e-01 -6.77168071e-01
-5.20574689e-01 7.18410790e-01 9.97280478e-01 -3.62667948e-01
5.67898691e-01 1.92370974e-02 -2.14429498e-01 -6.36313677e-01
-6.95071518e-01 -7.63473809e-02 3.90168875e-01 4.49977368e-01
5.52451134e-01 -3.95356156e-02 -4.50616032e-01 9.21759188e-01
-2.98304230e-01 -2.72950709e-01 4.23533648e-01 4.49628502e-01
-4.76925582e-01 -1.05712640e+00 1.37499735e-01 7.42843032e-01
-4.98653173e-01 -2.73410797e-01 -5.88654041e-01 9.25062299e-01
-4.64581847e-02 9.07664180e-01 -1.22879900e-01 -6.21006250e-01
3.20805013e-01 -2.33610854e-01 4.26903039e-01 -7.41134346e-01
-7.17567801e-01 -1.55661553e-01 -4.82425690e-02 -4.72514540e-01
-1.44044787e-01 6.14119843e-02 -1.21414006e+00 -7.02421963e-01
-8.50005090e-01 3.03689271e-01 3.25987369e-01 1.01457036e+00
5.96534073e-01 9.86738741e-01 4.25877661e-01 -4.06284094e-01
-2.53811955e-01 -1.15815723e+00 -7.12205350e-01 3.24389458e-01
1.04283050e-01 -8.24271142e-01 -4.04270649e-01 -7.11904883e-01] | [7.1935930252075195, 6.250176429748535] |
ed1561c3-c45f-4999-a17c-902713cbd9c9 | entity-level-text-guided-image-manipulation | 2302.11383 | null | https://arxiv.org/abs/2302.11383v1 | https://arxiv.org/pdf/2302.11383v1.pdf | Entity-Level Text-Guided Image Manipulation | Existing text-guided image manipulation methods aim to modify the appearance of the image or to edit a few objects in a virtual or simple scenario, which is far from practical applications. In this work, we study a novel task on text-guided image manipulation on the entity level in the real world (eL-TGIM). The task imposes three basic requirements, (1) to edit the entity consistent with the text descriptions, (2) to preserve the entity-irrelevant regions, and (3) to merge the manipulated entity into the image naturally. To this end, we propose an elegant framework, dubbed as SeMani, forming the Semantic Manipulation of real-world images that can not only edit the appearance of entities but also generate new entities corresponding to the text guidance. To solve eL-TGIM, SeMani decomposes the task into two phases: the semantic alignment phase and the image manipulation phase. In the semantic alignment phase, SeMani incorporates a semantic alignment module to locate the entity-relevant region to be manipulated. In the image manipulation phase, SeMani adopts a generative model to synthesize new images conditioned on the entity-irrelevant regions and target text descriptions. We discuss and propose two popular generation processes that can be utilized in SeMani, the discrete auto-regressive generation with transformers and the continuous denoising generation with diffusion models, yielding SeMani-Trans and SeMani-Diff, respectively. We conduct extensive experiments on the real datasets CUB, Oxford, and COCO datasets to verify that SeMani can distinguish the entity-relevant and -irrelevant regions and achieve more precise and flexible manipulation in a zero-shot manner compared with baseline methods. Our codes and models will be released at https://github.com/Yikai-Wang/SeMani. | ['Yanwei Fu', 'Wei zhang', 'Zhenguo Li', 'Hang Xu', 'Guansong Lu', 'Jianan Wang', 'Yikai Wang'] | 2023-02-22 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 5.79981625e-01 5.23710288e-02 3.99698615e-01 -2.89276391e-01
-4.31801379e-01 -4.03379381e-01 7.30999768e-01 -3.36014062e-01
-2.65873581e-01 3.99407595e-01 -3.06057297e-02 1.05282880e-01
-2.57123590e-01 -9.43238437e-01 -7.85428941e-01 -8.29084754e-01
3.84230793e-01 4.63195980e-01 2.36382917e-01 -2.55483598e-01
2.48343378e-01 2.12529346e-01 -1.50861001e+00 3.18936646e-01
1.14791739e+00 8.44012856e-01 5.57779133e-01 4.07462120e-01
-2.87495613e-01 5.30853093e-01 -5.90306759e-01 -4.18392628e-01
4.13153440e-01 -7.11311102e-01 -6.59141183e-01 4.18087602e-01
3.37980062e-01 -2.41517007e-01 -2.07105353e-01 1.39808095e+00
4.59056705e-01 1.63943678e-01 7.78985083e-01 -1.46783090e+00
-9.92104709e-01 5.25005400e-01 -7.84097910e-01 -1.52702466e-01
2.27581188e-01 3.05127293e-01 5.81218660e-01 -1.05746973e+00
1.07602274e+00 1.35477364e+00 1.64246053e-01 5.68286657e-01
-1.07208025e+00 -6.86626017e-01 3.66534233e-01 2.32621923e-01
-1.44906712e+00 -3.45338851e-01 8.87821138e-01 -5.46911538e-01
3.80912662e-01 2.30758831e-01 4.20121402e-01 9.98134613e-01
1.31841138e-01 7.05918908e-01 1.09973979e+00 -4.11004364e-01
5.00977188e-02 1.53525949e-01 -6.14949763e-02 7.24003196e-01
2.52995659e-02 5.18322624e-02 -4.79941368e-01 1.94539398e-01
7.65819669e-01 -1.09491628e-02 -4.58820015e-01 -3.06954503e-01
-1.57112682e+00 6.54334784e-01 3.58297527e-01 3.77789140e-01
-4.94612753e-01 -1.29897371e-02 2.19854981e-01 3.34229738e-01
5.12753367e-01 2.97673315e-01 -1.84651241e-01 3.09065968e-01
-8.13087046e-01 3.09804857e-01 3.92829925e-01 1.31602716e+00
8.88650119e-01 -8.27280153e-03 -4.29448962e-01 6.49088085e-01
2.73602337e-01 6.21340454e-01 5.07367909e-01 -8.55867088e-01
6.21998370e-01 6.86440527e-01 7.43384659e-02 -1.25158262e+00
-7.93195069e-02 -2.46404573e-01 -1.14851320e+00 2.17251703e-01
2.68119961e-01 -9.95976627e-02 -1.17715240e+00 1.94884932e+00
5.25365472e-01 3.59319478e-01 -5.10988161e-02 8.12983871e-01
1.02982152e+00 8.80351901e-01 -4.89414074e-02 -1.89914539e-01
1.48277724e+00 -1.13263547e+00 -8.46419394e-01 -2.25076228e-01
2.75192142e-01 -8.11989784e-01 1.07189691e+00 1.92287654e-01
-1.31771433e+00 -6.51969790e-01 -8.18120599e-01 -1.66587621e-01
-5.20064533e-01 3.27721685e-01 1.45903409e-01 1.63626790e-01
-1.05965018e+00 3.71474832e-01 -4.45217729e-01 -3.75771344e-01
2.50411272e-01 9.38972086e-02 -4.72039610e-01 9.99413729e-02
-1.16770911e+00 6.49018824e-01 5.67641973e-01 3.63712311e-01
-8.33949685e-01 -5.13275921e-01 -8.95944297e-01 3.42067890e-02
7.31859803e-01 -1.04045606e+00 8.50658596e-01 -1.30655682e+00
-1.37678826e+00 1.02439320e+00 -1.89176589e-01 -1.12255789e-01
6.73657417e-01 1.27870351e-01 -2.75849581e-01 1.20870970e-01
3.40511709e-01 8.50428343e-01 1.29111469e+00 -1.54966950e+00
-7.35546768e-01 -3.87607932e-01 4.09266278e-02 5.20779490e-01
-7.28778020e-02 3.57958674e-02 -1.02290714e+00 -1.08086312e+00
2.00251654e-01 -8.87572050e-01 -6.40116483e-02 1.63170412e-01
-6.41533971e-01 2.96047237e-02 8.89651299e-01 -9.58586693e-01
1.14701927e+00 -2.32024097e+00 5.70487559e-01 2.43369997e-01
3.72196704e-01 1.75351366e-01 -4.75202441e-01 2.41230428e-01
-2.19259277e-01 2.69024700e-01 -4.55114514e-01 -4.03002441e-01
-1.33318678e-01 1.52718738e-01 -2.53024340e-01 1.25969321e-01
7.70390630e-02 1.01970387e+00 -9.11959589e-01 -5.72276771e-01
3.60573351e-01 2.48502105e-01 -3.27108890e-01 1.96487725e-01
-2.85048425e-01 6.16433620e-01 -5.42883992e-01 4.60391551e-01
7.13488638e-01 -1.22477844e-01 -7.12857768e-02 -7.10100532e-01
-2.87876800e-02 -2.07524762e-01 -1.38807118e+00 1.67343521e+00
-1.91073284e-01 2.80130923e-01 2.55491942e-01 -1.04755116e+00
7.84671009e-01 1.27319306e-01 4.60055053e-01 -7.85247982e-01
-3.47995083e-03 5.93410060e-02 -1.94296837e-01 -6.21394575e-01
5.65987527e-01 7.49208778e-03 -5.36109693e-03 4.76212770e-01
2.12710053e-02 -2.62361705e-01 4.06989723e-01 3.60790759e-01
6.08911276e-01 3.59058946e-01 2.28134006e-01 -1.51645795e-01
6.69081748e-01 -3.84611525e-02 6.22343421e-01 7.77680576e-01
6.87980354e-02 7.82834589e-01 2.50866055e-01 -1.12293243e-01
-9.07891154e-01 -1.09347749e+00 2.42726997e-01 6.61406815e-01
7.05412269e-01 -2.33262092e-01 -9.23079729e-01 -7.48339653e-01
-3.41304570e-01 8.14563096e-01 -6.48884654e-01 -2.58903056e-01
-5.49549282e-01 -5.82829118e-01 1.20147809e-01 2.13895276e-01
8.57005656e-01 -1.32784688e+00 -2.97166616e-01 4.72838292e-03
-5.68486571e-01 -1.00055337e+00 -9.32345510e-01 -2.11936414e-01
-4.46344107e-01 -8.93138230e-01 -7.23799527e-01 -9.58871722e-01
1.03229558e+00 4.12111342e-01 8.11680079e-01 1.42445877e-01
-3.54238540e-01 2.79074490e-01 -3.87697518e-01 -2.00260848e-01
-4.58696574e-01 -1.09080091e-01 -2.93726623e-01 3.88486654e-01
-2.65909731e-01 -4.87243742e-01 -4.80305105e-01 5.20002842e-01
-1.29863417e+00 7.26782501e-01 5.73091865e-01 8.60269129e-01
7.81339109e-01 2.92151392e-01 3.00289690e-01 -9.25937295e-01
5.16751349e-01 -2.03896701e-01 -4.99621302e-01 6.29252076e-01
-4.11758453e-01 1.15836546e-01 5.27087510e-01 -5.27045429e-01
-1.38900137e+00 -3.73161435e-02 1.17722951e-01 -4.32165831e-01
2.02952996e-02 5.46462655e-01 -6.00919664e-01 1.04632854e-01
4.91212130e-01 5.27416527e-01 -6.76435754e-02 -2.13147432e-01
7.12118506e-01 4.87128347e-01 6.60998106e-01 -6.88243449e-01
1.29660380e+00 6.00802898e-01 -2.44881809e-01 -6.13606989e-01
-5.94406843e-01 -2.41391763e-01 -6.83442295e-01 -2.94145674e-01
1.08872926e+00 -7.33408034e-01 -2.58225262e-01 8.51403892e-01
-1.21725798e+00 -4.42814559e-01 -4.61916029e-01 1.21758379e-01
-6.65072560e-01 5.89350343e-01 -3.63254756e-01 -2.65343726e-01
-4.89769816e-01 -1.36354744e+00 1.13817310e+00 3.50506574e-01
1.24732248e-01 -7.87103772e-01 -2.76535511e-01 3.30783188e-01
2.96061397e-01 3.85600895e-01 9.51551676e-01 -2.43369907e-01
-8.74180198e-01 2.16794282e-01 -4.01882917e-01 2.65829444e-01
3.09744537e-01 1.61016896e-01 -4.80865389e-01 -1.68279022e-01
2.42844950e-02 4.89644669e-02 6.79144382e-01 3.10406446e-01
1.06209207e+00 -3.01325649e-01 -2.94257730e-01 8.63465607e-01
1.28649771e+00 5.30896187e-01 8.86509717e-01 3.18142980e-01
6.48965716e-01 6.12563789e-01 9.80496705e-01 2.65560687e-01
2.98533738e-01 8.64540398e-01 2.16009408e-01 -2.52841115e-01
-4.34747428e-01 -3.51235539e-01 3.40712100e-01 8.26233745e-01
-5.14236167e-02 -4.88968104e-01 -5.70173979e-01 4.94491965e-01
-2.16908693e+00 -1.01955271e+00 -1.14789300e-01 1.99992108e+00
8.33396792e-01 -2.48659432e-01 -3.08816075e-01 -3.28154802e-01
1.06917083e+00 1.40949443e-01 -6.00707412e-01 -2.98870113e-02
-1.32335588e-01 3.63226570e-02 6.78450838e-02 4.99196738e-01
-9.73537803e-01 1.11996448e+00 4.59355879e+00 1.01852417e+00
-1.13536727e+00 2.30812788e-01 5.70746183e-01 1.46021485e-01
-3.41744542e-01 1.12288006e-01 -6.11000657e-01 6.24471962e-01
2.19088331e-01 -4.76841033e-01 6.31035149e-01 5.35635591e-01
4.47404951e-01 -9.23158675e-02 -8.71308982e-01 1.03469944e+00
2.81785280e-01 -1.16189480e+00 5.42911053e-01 -2.51589507e-01
8.12133610e-01 -4.60716218e-01 1.02335446e-01 1.05205655e-01
1.55318782e-01 -7.39154279e-01 1.12650228e+00 8.55352402e-01
9.82063651e-01 -4.42333251e-01 3.10407251e-01 3.27685416e-01
-1.24919665e+00 2.15677962e-01 -2.20714852e-01 5.41717172e-01
3.80578518e-01 5.97088933e-01 -3.40776682e-01 9.70956445e-01
6.99720860e-01 8.71400476e-01 -5.64007342e-01 7.78818011e-01
-5.73547304e-01 1.01079956e-01 -2.14659199e-01 4.66250032e-01
2.10741200e-02 -6.88099205e-01 7.08832443e-01 9.35791612e-01
5.50335288e-01 9.03351307e-02 9.14320126e-02 1.34587467e+00
-9.36847180e-02 1.66193113e-01 -4.74676371e-01 1.12058580e-01
3.48844856e-01 1.27906144e+00 -7.72835314e-01 -5.29671073e-01
-2.21764416e-01 1.53978384e+00 6.06658533e-02 7.20886767e-01
-1.17717254e+00 -6.15927041e-01 2.83320785e-01 -6.98344633e-02
2.62710601e-01 -1.24738127e-01 -2.17847437e-01 -1.46890879e+00
3.72268111e-01 -1.13773811e+00 1.39684558e-01 -1.26460481e+00
-1.04562914e+00 6.04805946e-01 2.96428055e-01 -1.23412836e+00
-5.41370967e-03 -3.40097874e-01 -6.82478368e-01 9.52617586e-01
-1.33219087e+00 -1.47773814e+00 -7.68633485e-01 8.14854681e-01
5.94351530e-01 5.27549237e-02 3.52348894e-01 3.45767230e-01
-7.15475261e-01 3.73323828e-01 3.97750773e-02 1.30091384e-01
8.53604734e-01 -8.92874479e-01 4.08730328e-01 1.11279666e+00
1.63617119e-01 5.76927185e-01 4.90201950e-01 -8.67445648e-01
-1.00608671e+00 -1.45525801e+00 7.09672451e-01 -1.21807769e-01
3.79586101e-01 -3.81741971e-01 -8.52826715e-01 8.17479551e-01
2.18144193e-01 -3.13222080e-01 -1.27062589e-01 -5.66037059e-01
-4.66684327e-02 -1.26512334e-01 -1.01584172e+00 9.52488184e-01
1.31815779e+00 -3.10759515e-01 -4.90805656e-01 3.37378919e-01
6.96460724e-01 -5.17131746e-01 -6.37244225e-01 4.66613054e-01
2.66385823e-01 -8.77404869e-01 1.08084512e+00 -2.93762118e-01
5.16354680e-01 -7.14806616e-01 1.53754637e-01 -1.33811533e+00
-2.27890998e-01 -6.59824014e-01 3.09247762e-01 1.52151871e+00
3.29479039e-01 -6.87809110e-01 2.69031703e-01 5.58833957e-01
-1.77837517e-02 -5.21181941e-01 -7.43258238e-01 -7.10433722e-01
-1.10002302e-01 -2.14807764e-01 6.33998573e-01 9.72379267e-01
-4.40197349e-01 2.03481644e-01 -4.58111137e-01 1.67504504e-01
5.59265614e-01 2.81532615e-01 9.75673079e-01 -7.93555140e-01
-2.80106306e-01 -4.74187583e-01 -2.97644854e-01 -1.03901815e+00
1.21946476e-01 -8.90152514e-01 1.76324397e-01 -1.72648180e+00
4.35859710e-01 -4.26084548e-01 6.99171126e-02 5.74064732e-01
-3.75689924e-01 4.63043787e-02 3.10553432e-01 3.34617883e-01
-4.27222461e-01 8.32359552e-01 1.50503039e+00 -4.08397377e-01
-1.93481356e-01 -3.35580885e-01 -7.42946804e-01 6.29606605e-01
6.60423219e-01 -3.37737054e-01 -6.37128294e-01 -5.21782815e-01
1.13773018e-01 8.14098865e-03 5.15882254e-01 -6.98691070e-01
2.13588074e-01 -4.17692989e-01 5.86961024e-02 -4.90573376e-01
1.51268914e-01 -7.65505970e-01 5.16696453e-01 1.88054651e-01
-2.40904227e-01 1.79791544e-02 -4.03636210e-02 4.29186463e-01
-2.71869928e-01 -3.82504642e-01 8.88197720e-01 -1.98564917e-01
-8.95241082e-01 5.05296826e-01 -1.31887615e-01 1.29196689e-01
1.35849965e+00 -2.27631778e-01 -4.05043334e-01 -4.45116222e-01
-9.53710616e-01 2.37613559e-01 6.48481786e-01 5.72574258e-01
7.44468570e-01 -1.30000532e+00 -7.97870457e-01 2.79989213e-01
1.27500132e-01 2.20929265e-01 5.30230820e-01 8.76454711e-01
-4.57464546e-01 -1.20441936e-01 -5.07709756e-02 -4.91503835e-01
-1.40076327e+00 7.82451510e-01 4.64354277e-01 -4.17045772e-01
-7.09057987e-01 5.88206530e-01 9.70908403e-01 -4.03357714e-01
7.14730620e-02 -3.42215061e-01 -9.37913544e-03 -8.42088386e-02
3.64631981e-01 3.06133437e-03 -2.64769644e-01 -6.90717757e-01
-1.20216329e-02 8.08805943e-01 -1.26127318e-01 -2.06904456e-01
1.08616149e+00 -4.64520603e-01 -3.89211386e-01 9.50108394e-02
9.17091310e-01 -1.54031903e-01 -1.06893110e+00 -3.16741258e-01
-2.78478384e-01 -4.66733932e-01 -6.36844933e-02 -8.93374324e-01
-1.35293400e+00 7.30280399e-01 5.22946477e-01 -1.48641199e-01
1.42037416e+00 -7.10633174e-02 7.51791418e-01 1.08245030e-01
2.40676016e-01 -1.01991212e+00 1.54101685e-01 2.84645498e-01
1.32143974e+00 -8.79369378e-01 -1.28971189e-01 -7.01202631e-01
-7.89670467e-01 9.06675279e-01 7.29525030e-01 1.75958499e-01
6.06682062e-01 1.55240238e-01 -4.14885301e-03 -3.13521594e-01
-4.34947759e-01 -2.75557071e-01 4.57450002e-01 5.33999562e-01
1.75597128e-02 -1.38026744e-01 -3.34400177e-01 4.51113850e-01
8.56765360e-02 -3.48625369e-02 2.88921654e-01 7.22558856e-01
-1.51965186e-01 -1.06421459e+00 -4.05656606e-01 3.08223575e-01
4.25371863e-02 -2.34591171e-01 -3.54828805e-01 7.89436281e-01
4.81666327e-01 8.49811196e-01 -6.63904026e-02 -1.89510554e-01
3.97583336e-01 1.60931293e-02 3.92416745e-01 -5.60847461e-01
-3.74458671e-01 1.36266217e-01 -1.05924718e-01 -5.79794884e-01
-6.05676055e-01 -6.36642694e-01 -1.30406082e+00 -1.04614727e-01
-3.93563986e-01 -1.67227149e-01 5.14130592e-01 8.72688234e-01
5.76655686e-01 5.87106824e-01 5.44865847e-01 -7.79143870e-01
-3.36866885e-01 -7.73662090e-01 -5.17903686e-01 9.94540513e-01
-1.29970089e-01 -6.66613042e-01 -3.34382623e-01 5.09354293e-01] | [11.282054901123047, -0.9698629975318909] |
bd01d11b-7cbe-4c43-8218-f013aec1e34b | hole-filling-method-for-depth-image-based | null | null | https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7836315 | https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7836315 | Hole Filling Method for Depth Image Based Rendering Based on Boundary Decision | In three-dimensional display systems, depth image based rendering is the most commonly used technique for generating images captured from a virtual viewpoint through reference views and depth maps. However, disoccluded hole filling remain a challenging issue as newly exposed area appears in the virtual view. Image inpainting based hole filling is a popular approach for filling in the hole, but the conventional methods generate annoying artifacts along the boundaries of the foreground objects. This letter presents a robust hole filling method that generates a virtual view with high quality. First, by using the local and neighboring property of the depth map, the background region is filled prior to the foreground region. Also, in the matching step to find the data to fill in, depth homogeneity is considered. Finally, based on the boundary decision using the depth property, the boundary regions are filled with background-relevant data to reduce boundary artifacts. Experimental results show that the views synthesized by the proposed method achieve higher visual comfort than the view synthesized by the existing methods. | ['Taejeong Kim', 'Hyuk Choi', 'Wonseok Song', 'Jea-Hyung Cho'] | 2017-03-01 | null | null | null | signal-processing-letters-2017-3 | ['image-inpainting'] | ['computer-vision'] | [ 6.04645669e-01 3.25381868e-02 2.50828326e-01 6.15244098e-02
-1.88387513e-01 -2.44805753e-01 1.16670847e-01 5.84372655e-02
2.32766252e-02 7.64231086e-01 1.68815836e-01 -2.75444030e-03
2.27044374e-01 -9.96364772e-01 -3.57262492e-01 -8.44811082e-01
5.47094703e-01 -4.87018637e-02 7.63951242e-01 1.16059184e-02
7.31906652e-01 3.12615275e-01 -1.76892233e+00 3.76622915e-01
1.12804854e+00 1.13296449e+00 6.35973811e-01 2.68601328e-01
-5.06420135e-01 5.77853322e-01 -6.23407900e-01 2.03394890e-02
4.58875090e-01 -6.02569342e-01 -3.08553338e-01 7.22353995e-01
1.86933905e-01 -7.41767168e-01 1.11160584e-01 1.11259747e+00
4.75556403e-01 1.42991424e-01 3.01981628e-01 -1.03612840e+00
-1.51437342e-01 -2.93250322e-01 -1.08036649e+00 1.96093515e-01
5.98920107e-01 -9.80052724e-02 9.31820199e-02 -1.05713153e+00
8.77164483e-01 1.10514438e+00 1.87316850e-01 4.43434060e-01
-8.86791170e-01 -6.04614317e-01 3.26563329e-01 1.36239782e-01
-1.35816801e+00 -2.63904810e-01 1.21983004e+00 -4.52591240e-01
4.35581297e-01 5.50449908e-01 7.04546213e-01 2.18396947e-01
4.51122820e-01 3.11437994e-01 1.36327505e+00 -4.83300775e-01
2.53203958e-01 4.41817939e-01 -2.25449577e-01 7.56509483e-01
1.51240543e-01 9.29856896e-02 -5.09141326e-01 -2.15454251e-01
1.08182156e+00 1.68183237e-01 -9.21146393e-01 -7.33904243e-01
-8.25181246e-01 4.08853948e-01 7.51962792e-03 6.50375411e-02
-2.22577706e-01 -5.11749208e-01 2.41764948e-01 -2.29641840e-01
6.66499317e-01 1.65035445e-02 -1.37477964e-01 -6.34141639e-02
-1.02306926e+00 1.30530551e-01 3.50003272e-01 8.06419849e-01
8.96051049e-01 1.67106241e-02 -2.12273449e-02 5.38727939e-01
3.51572007e-01 8.25805664e-02 2.61631370e-01 -1.01965129e+00
7.54271686e-01 9.22468543e-01 4.02717054e-01 -1.15788722e+00
4.24165390e-02 -5.46485819e-02 -7.61446059e-01 8.76343846e-01
1.26530468e-01 -4.61964086e-02 -9.11639094e-01 9.40122783e-01
8.75084519e-01 1.04709886e-01 -8.58079493e-02 9.79315341e-01
9.41113532e-01 8.60584736e-01 -3.97504419e-01 -7.50155628e-01
1.23572052e+00 -8.00152898e-01 -1.35145390e+00 -2.01354176e-01
1.76627617e-02 -9.60768044e-01 8.37538004e-01 7.32487798e-01
-1.30323231e+00 -6.38233066e-01 -1.25682163e+00 -1.77283287e-01
2.56542638e-02 2.04730574e-02 1.68841541e-01 6.26709998e-01
-8.31973433e-01 1.85938835e-01 -4.36163753e-01 -8.54277685e-02
-2.07289178e-02 -3.12411021e-02 -1.93043783e-01 -7.30447471e-02
-7.61855483e-01 5.46176553e-01 2.41132438e-01 2.31743321e-01
-4.28057700e-01 -6.65719807e-01 -7.36553013e-01 -2.33523101e-01
3.57626647e-01 -5.13614118e-01 6.66457534e-01 -8.17850053e-01
-1.57027769e+00 8.99437428e-01 -6.60335720e-01 1.78335875e-01
9.03434753e-01 8.11327249e-03 -2.32560009e-01 3.22341144e-01
5.76009266e-02 4.04727072e-01 8.72694910e-01 -1.76306963e+00
-7.93331981e-01 -5.38167357e-01 -1.94436029e-01 7.06195295e-01
2.71434009e-01 -2.89441586e-01 -6.32158875e-01 -4.06256169e-01
8.37996066e-01 -3.10784131e-01 -8.93027186e-02 4.48111802e-01
-3.56743842e-01 3.37618858e-01 1.24456120e+00 -9.52127993e-01
1.43259871e+00 -2.34491038e+00 -1.15596831e-01 2.49951556e-01
3.51671755e-01 -2.29182490e-03 6.62677705e-01 2.39746287e-01
1.56616300e-01 -2.08870173e-02 -2.73381919e-01 -2.49252364e-01
-6.26861513e-01 -6.74702525e-02 -1.09687440e-01 2.96603799e-01
-2.81461537e-01 1.69828221e-01 -5.13117909e-01 -6.35692954e-01
5.84299445e-01 5.71692050e-01 -4.20648277e-01 3.53172004e-01
6.99447468e-02 9.80515957e-01 -2.90949821e-01 4.75974888e-01
1.38961518e+00 1.40107930e-01 -9.59098414e-02 -1.90620378e-01
-5.02818167e-01 -4.45703268e-02 -1.56908417e+00 1.44658673e+00
-3.25958431e-01 5.22193432e-01 4.26668674e-01 -2.37471983e-01
1.31092310e+00 2.76268393e-01 4.00164574e-01 -8.34777832e-01
1.60596043e-01 1.43905893e-01 -4.14701104e-01 -6.29591465e-01
6.86762750e-01 -5.98655567e-02 5.74151814e-01 2.38975585e-01
-9.16206360e-01 -2.39833266e-01 -1.12278767e-01 -6.28083423e-02
4.31822956e-01 2.53163904e-01 2.28075936e-01 -5.32689467e-02
6.99507475e-01 -2.39796862e-01 9.60114002e-01 1.66839644e-01
-1.07168823e-01 1.14096391e+00 2.60643721e-01 -4.54595476e-01
-9.19240057e-01 -9.45084512e-01 -2.57974118e-01 5.22374436e-02
9.43933845e-01 -1.14815362e-01 -8.14898074e-01 -7.35376403e-02
-3.52959335e-01 4.46900874e-01 -5.62325537e-01 1.18807897e-01
-6.04616821e-01 -3.71040195e-01 -5.11327684e-01 1.34149998e-01
1.12965345e+00 -9.39218998e-01 -1.11193144e+00 2.46152803e-01
-3.82037520e-01 -8.55257332e-01 -4.26029652e-01 -3.28756332e-01
-1.17278624e+00 -1.12185073e+00 -9.56699967e-01 -8.73973846e-01
9.75325465e-01 6.24115229e-01 7.97533274e-01 3.46211344e-01
-3.88308346e-01 -1.40646204e-01 -2.05548942e-01 -5.53210340e-02
-3.12843740e-01 -7.29475617e-01 -5.17051995e-01 1.28480226e-01
-9.69762430e-02 -4.40368801e-01 -1.09328735e+00 5.26084363e-01
-8.15749049e-01 6.78651452e-01 7.62814805e-02 4.69908357e-01
7.96817482e-01 3.75489384e-01 9.05229002e-02 -7.33777165e-01
4.15681064e-01 -5.36716357e-02 -7.83911884e-01 1.38538666e-02
-3.58184367e-01 -4.82937843e-01 3.20789427e-01 -2.16121554e-01
-1.61554503e+00 -2.63130248e-01 1.72152817e-01 -2.69526452e-01
-1.24853134e-01 -4.36784588e-02 -7.58907318e-01 1.13783255e-02
2.48182818e-01 3.60371202e-01 8.88503864e-02 -4.22183871e-01
-7.02474564e-02 5.93155384e-01 2.61294067e-01 -5.21744117e-02
4.86694247e-01 8.19634497e-01 -7.98128620e-02 -7.24529326e-01
-4.84271087e-02 -3.18837315e-01 -4.52967227e-01 -4.78824615e-01
1.15304351e+00 -7.61855781e-01 -5.98456860e-01 4.14019763e-01
-1.32286835e+00 5.79555929e-02 -8.18215162e-02 3.49525958e-01
-3.36562246e-01 6.85558677e-01 -4.36300993e-01 -1.09859884e+00
-2.88748503e-01 -1.29505956e+00 1.00919461e+00 5.15263498e-01
1.53846011e-01 -6.94799185e-01 -1.60328105e-01 4.67137098e-01
1.12975694e-01 4.40923929e-01 1.05277181e+00 5.83825290e-01
-1.09888089e+00 1.66776523e-01 -2.95287400e-01 6.63185492e-02
5.99627137e-01 5.54544628e-02 -9.86895919e-01 3.46801728e-02
3.79446268e-01 4.18727368e-01 3.67428601e-01 5.47724783e-01
1.10901117e+00 -2.66064465e-01 -5.58797240e-01 6.76374674e-01
1.74229217e+00 1.00960231e+00 1.09942389e+00 3.61298978e-01
5.76023459e-01 7.38252580e-01 1.02202117e+00 6.15609229e-01
9.53862369e-02 6.44310772e-01 4.07664597e-01 -3.42349559e-01
-1.09495677e-01 -3.86628658e-01 -5.50265461e-02 6.23322010e-01
3.22122723e-02 -4.15285408e-01 -4.13778841e-01 3.19063097e-01
-1.53504610e+00 -7.85790384e-01 -4.57388431e-01 2.61616731e+00
7.29834497e-01 1.74951747e-01 -3.51858228e-01 5.13049006e-01
9.36992288e-01 1.36932030e-01 -3.55282217e-01 -5.67780972e-01
-1.06924139e-02 -6.31615669e-02 2.24893764e-01 8.82267714e-01
-6.34738386e-01 5.25736868e-01 5.12538528e+00 6.50915980e-01
-1.11512423e+00 -3.62517014e-02 6.48109138e-01 3.77673991e-02
-6.28652573e-01 1.06210805e-01 -7.38807440e-01 7.61668563e-01
-1.43106982e-01 8.72445107e-02 -4.68000732e-02 4.77947474e-01
7.35171378e-01 -1.22750020e+00 -6.04973555e-01 1.28313482e+00
2.32312739e-01 -1.14668167e+00 3.00969202e-02 1.22547314e-01
9.62129891e-01 -9.26706135e-01 7.71206394e-02 -5.53307891e-01
-5.36017895e-01 -6.06734753e-01 6.08153164e-01 3.83462936e-01
6.91873193e-01 -7.98664689e-01 5.28297186e-01 4.47510451e-01
-1.36283588e+00 2.42743641e-01 -5.22267938e-01 -6.79561719e-02
6.73627257e-01 8.06643665e-01 -5.33519924e-01 6.12629294e-01
6.31506562e-01 2.15897486e-01 -4.53734398e-01 1.28789043e+00
4.49473597e-03 -8.36125910e-02 8.36750567e-02 4.80328172e-01
-2.68265247e-01 -7.75146186e-01 5.15358329e-01 3.87046576e-01
5.82771599e-01 3.23096156e-01 1.23689920e-01 1.02262664e+00
5.00114977e-01 3.58202010e-01 -8.23021710e-01 8.05378854e-01
3.07208717e-01 9.19429481e-01 -8.37350309e-01 -3.23818386e-01
-2.66900033e-01 1.42312145e+00 -3.31111223e-01 5.10544300e-01
-8.16067755e-01 -4.25517201e-01 1.98653758e-01 6.18744493e-01
8.23143870e-02 -4.68181409e-02 -4.41372603e-01 -8.84409189e-01
4.48868632e-01 -6.62767351e-01 -4.34443094e-02 -1.16213000e+00
-4.48745728e-01 6.14711821e-01 -1.06579244e-01 -1.69570684e+00
1.01512395e-01 -1.00425594e-01 -6.28895581e-01 1.25036657e+00
-1.33505285e+00 -6.26474619e-01 -8.37689519e-01 3.80727917e-01
8.81029129e-01 4.10902858e-01 5.05726576e-01 3.86151910e-01
-2.96886176e-01 8.21723342e-02 -7.44481981e-02 -3.89711827e-01
5.18788159e-01 -7.23062277e-01 -7.77315050e-02 9.30440068e-01
-4.74215150e-01 3.27854961e-01 7.11798072e-01 -1.16541898e+00
-7.46688604e-01 -5.28945982e-01 6.76534176e-01 1.22242287e-01
-2.83898324e-01 -3.60435486e-01 -1.19615054e+00 2.24716142e-01
4.56641227e-01 -1.70588434e-01 3.53656381e-01 -7.06323862e-01
2.46603236e-01 3.06083797e-03 -1.59068465e+00 5.92227161e-01
8.40037107e-01 -2.93361902e-01 -1.52015060e-01 -1.83945656e-01
4.36694145e-01 -8.55308890e-01 -3.94193083e-01 5.61564207e-01
5.99914074e-01 -1.54981470e+00 6.31889999e-01 2.73642242e-01
3.94480973e-01 -8.77008975e-01 1.97086796e-01 -9.91638005e-01
2.95580737e-02 -6.21616006e-01 2.49015525e-01 1.33248472e+00
1.67979062e-01 -5.48858821e-01 1.06262112e+00 7.61499226e-01
-2.90824939e-02 -6.99522734e-01 -8.43430161e-01 -1.76573679e-01
-6.34904623e-01 1.72197148e-01 3.67466122e-01 7.52490640e-01
-1.72920972e-01 -6.30146265e-02 -2.65696734e-01 2.96540976e-01
6.36012852e-01 3.43367726e-01 8.10352683e-01 -1.19250464e+00
-1.00894585e-01 8.60721394e-02 -3.18822026e-01 -1.26714933e+00
-5.10940671e-01 -2.58990169e-01 4.59037013e-02 -1.89513290e+00
1.49773464e-01 -5.58334291e-01 3.57941359e-01 -3.38648289e-01
-4.19269711e-01 2.60897338e-01 7.39062652e-02 1.31414115e-01
5.39997369e-02 4.81977999e-01 1.80900538e+00 2.44570076e-01
-6.95506811e-01 2.82060523e-02 -2.99543887e-01 7.44904637e-01
7.58222938e-01 -6.95803901e-03 -9.41826999e-01 -3.84342790e-01
4.40895669e-02 5.47311962e-01 2.61089772e-01 -8.75487268e-01
3.35967727e-02 -2.10029259e-01 6.72429144e-01 -1.21297801e+00
6.40645146e-01 -9.55707550e-01 4.78003323e-01 3.15068692e-01
1.34248883e-01 3.02626520e-01 1.92521989e-01 5.11696875e-01
-2.07925186e-01 -2.54733324e-01 8.32046807e-01 -3.96480531e-01
-6.50326908e-01 1.20186031e-01 -4.26384270e-01 -6.10809028e-02
1.21073484e+00 -1.10762346e+00 -2.09443167e-01 -4.98937964e-01
-6.42327726e-01 -6.54346049e-02 8.96796942e-01 2.08189577e-01
1.20886898e+00 -1.15172851e+00 -2.51886755e-01 6.77580893e-01
-1.09011993e-01 3.59288514e-01 7.84133136e-01 7.38818526e-01
-1.04959965e+00 5.20173870e-02 -3.22819024e-01 -6.02388382e-01
-1.50372219e+00 6.25009179e-01 3.18175524e-01 7.61041360e-04
-8.79347324e-01 7.42839217e-01 8.91107738e-01 2.27394596e-01
2.44252205e-01 -5.19841611e-01 -2.88010269e-01 -1.71935499e-01
7.59263992e-01 5.12421429e-01 5.91435544e-02 -4.58044440e-01
-8.94501209e-02 8.69785488e-01 1.55817628e-01 -5.03193617e-01
7.07786381e-01 -7.08200216e-01 -1.16183870e-01 4.33464348e-01
9.10376728e-01 6.44567788e-01 -1.44610918e+00 1.66830525e-01
-5.57855248e-01 -1.18913341e+00 -7.87835568e-02 -5.11448443e-01
-1.15100574e+00 9.62321877e-01 9.19228733e-01 2.15401780e-03
1.25093520e+00 -5.29155731e-01 7.89470315e-01 -7.63438880e-01
6.25280082e-01 -1.06060421e+00 -2.77336687e-02 -2.09659323e-01
7.92184353e-01 -9.89771366e-01 1.74026355e-01 -1.26637030e+00
-4.97219443e-01 8.98124099e-01 1.10237110e+00 1.51828825e-01
5.99746108e-01 3.27231556e-01 1.33779719e-01 -1.19272985e-01
-4.54501867e-01 1.06380470e-01 -2.42613181e-02 6.80432558e-01
2.54920721e-01 -3.64948630e-01 -7.10672617e-01 1.33688703e-01
1.31776392e-01 -9.48137641e-02 8.33582520e-01 1.00377047e+00
-6.97828889e-01 -9.43828523e-01 -7.58651376e-01 1.45868316e-01
-4.00311708e-01 5.76137239e-03 -8.36014524e-02 6.46622598e-01
5.00030816e-01 1.09089303e+00 3.11856031e-01 2.27664188e-02
2.14025766e-01 -2.02642456e-01 6.19812548e-01 -6.51062250e-01
-2.68979669e-01 4.18017119e-01 -1.06899731e-01 -5.38200736e-01
-2.41406620e-01 -2.71555126e-01 -1.46246076e+00 1.41446684e-02
-4.16717768e-01 7.52922446e-02 4.29777831e-01 4.23315346e-01
2.06183091e-01 3.86448443e-01 6.48423135e-01 -8.42556596e-01
3.82866204e-01 -5.43016613e-01 -8.08171928e-01 3.29522669e-01
3.63225639e-01 -7.52657175e-01 -5.51728606e-01 4.18458953e-02] | [9.350672721862793, -2.448458194732666] |
b13395c8-fd48-4b63-b632-53e28842fd18 | leveraging-explanations-in-interactive | 2207.14526 | null | https://arxiv.org/abs/2207.14526v2 | https://arxiv.org/pdf/2207.14526v2.pdf | Leveraging Explanations in Interactive Machine Learning: An Overview | Explanations have gained an increasing level of interest in the AI and Machine Learning (ML) communities in order to improve model transparency and allow users to form a mental model of a trained ML model. However, explanations can go beyond this one way communication as a mechanism to elicit user control, because once users understand, they can then provide feedback. The goal of this paper is to present an overview of research where explanations are combined with interactive capabilities as a mean to learn new models from scratch and to edit and debug existing ones. To this end, we draw a conceptual map of the state-of-the-art, grouping relevant approaches based on their intended purpose and on how they structure the interaction, highlighting similarities and differences between them. We also discuss open research issues and outline possible directions forward, with the hope of spurring further research on this blooming research topic. | ['Elizabeth Daly', 'Wolfang Stammer', 'Öznur Alkan', 'Stefano Teso'] | 2022-07-29 | null | null | null | null | ['machine-learning', 'machine-learning'] | ['methodology', 'miscellaneous'] | [ 3.47807735e-01 9.41005111e-01 -5.70092499e-01 -5.89857280e-01
-1.72043536e-02 -6.59847319e-01 8.58728349e-01 3.90609115e-01
1.72172815e-01 5.05483687e-01 2.96786606e-01 -7.29443431e-01
7.39964098e-02 -4.76958364e-01 -3.88497591e-01 -5.16758338e-02
4.93163802e-02 4.71371353e-01 -1.71680495e-01 -7.87783116e-02
7.53428817e-01 4.81058836e-01 -1.73923850e+00 5.74685395e-01
1.16847241e+00 1.08605988e-01 6.88612089e-02 6.95133448e-01
-4.10469413e-01 9.91672575e-01 -6.16686046e-01 -3.98429006e-01
-3.54176491e-01 -6.81215286e-01 -1.18608594e+00 2.17071548e-01
3.59713972e-01 -1.97852671e-01 1.20887712e-01 6.83846414e-01
-1.24422787e-02 6.40508682e-02 5.37158728e-01 -1.49418938e+00
-8.58694494e-01 7.46887982e-01 1.14083037e-01 -2.24714339e-01
7.11517394e-01 4.08480406e-01 9.19999719e-01 -4.76325125e-01
6.27166629e-01 1.24173319e+00 2.88359642e-01 1.15562081e+00
-1.44228268e+00 -5.80295742e-01 7.07921147e-01 2.25944519e-01
-7.78578997e-01 -4.12947208e-01 5.31159759e-01 -6.62001610e-01
1.09236193e+00 6.92214370e-01 1.02642834e+00 9.64886010e-01
-1.88127775e-02 8.92133534e-01 1.11197758e+00 -7.61847019e-01
1.77580789e-01 8.19292068e-01 4.04602319e-01 8.71497214e-01
3.38792205e-01 1.97012544e-01 -7.15786278e-01 -2.18022838e-01
9.63456094e-01 -1.55944064e-01 -2.98252851e-01 -7.27999091e-01
-7.87437499e-01 9.09844577e-01 3.08668345e-01 5.24401248e-01
-1.33803591e-01 1.83086008e-01 -1.80851951e-01 3.96930486e-01
1.78009421e-01 1.18586683e+00 -5.71501672e-01 -4.17664677e-01
-6.38650298e-01 3.58423471e-01 1.21078014e+00 7.74330199e-01
6.37212634e-01 1.21081345e-01 -5.78639135e-02 4.66246158e-01
8.35787356e-01 -5.20823635e-02 1.86192736e-01 -1.10850191e+00
-9.95509475e-02 9.27921712e-01 2.97511429e-01 -8.13732803e-01
-1.43526211e-01 -3.70897502e-01 -2.81806767e-01 5.45529664e-01
2.52958179e-01 -1.41920537e-01 -6.61715984e-01 1.69907808e+00
4.88978513e-02 8.83509684e-03 -5.52588105e-02 6.56325400e-01
8.21158230e-01 4.30062354e-01 2.46113434e-01 -1.02194235e-01
1.09336555e+00 -9.96262133e-01 -6.33001685e-01 -6.24813676e-01
1.02425373e+00 -5.12942374e-01 1.35117245e+00 5.58898270e-01
-1.20872045e+00 -5.58794379e-01 -9.38078940e-01 -1.22025743e-01
-4.13288414e-01 2.39766872e-04 1.01938093e+00 7.05295861e-01
-1.03654778e+00 7.23336160e-01 -8.29839170e-01 -6.12021625e-01
2.84645855e-01 2.41548523e-01 1.43045992e-01 8.93565044e-02
-9.42312419e-01 1.33274209e+00 1.36536002e-01 -7.17488155e-02
-6.19493484e-01 -7.99912989e-01 -9.08677638e-01 -4.35805172e-02
3.02255869e-01 -9.79230881e-01 1.80331087e+00 -1.22887874e+00
-1.59234023e+00 8.00179660e-01 -2.95371473e-01 -3.05354118e-01
4.74819154e-01 -4.07049239e-01 -2.05556780e-01 -2.95475125e-01
-2.85862446e-01 8.93740952e-01 3.58038455e-01 -1.67618608e+00
-6.18144333e-01 -7.67672062e-02 6.58481836e-01 2.72412837e-01
-9.84070003e-02 8.21319744e-02 -2.36105248e-01 -2.05299020e-01
-4.41424213e-02 -9.36636925e-01 -4.59761828e-01 1.21749580e-01
-4.89116549e-01 -2.15618685e-01 8.05976093e-01 -3.76667857e-01
1.73503458e+00 -1.60862720e+00 3.32848877e-01 1.72215641e-01
6.26643658e-01 3.43984783e-01 1.24568701e-01 1.03756106e+00
-1.27830654e-01 6.81197703e-01 -4.82710749e-02 -4.39404398e-01
2.09831938e-01 3.27735208e-02 -1.83250383e-01 -9.14733484e-02
1.80404067e-01 1.00966048e+00 -8.18855226e-01 -1.20846406e-01
4.39481974e-01 4.97170269e-01 -5.27525008e-01 4.00318176e-01
-6.40964448e-01 6.84222281e-01 -4.33673322e-01 2.04527915e-01
5.34060970e-02 -5.17028868e-01 3.10560614e-01 3.14797312e-01
-5.30833602e-01 7.50182748e-01 -1.14805484e+00 1.40700781e+00
-4.94618744e-01 9.11777377e-01 -8.84465501e-02 -5.99636853e-01
7.17583418e-01 3.57584298e-01 -1.69811636e-01 -7.08268732e-02
-3.45563516e-02 -1.03408284e-01 4.16195273e-01 -6.07893348e-01
2.53811926e-01 5.51897958e-02 1.86637610e-01 8.73765528e-01
-4.97741252e-01 -3.96469682e-01 1.64580896e-01 5.21525443e-01
5.28893769e-01 2.52520561e-01 6.95302010e-01 -1.98817533e-02
5.00820100e-01 1.74820244e-01 -1.24392278e-01 8.97919655e-01
3.37902099e-01 5.77584729e-02 4.87069756e-01 -5.64813793e-01
-6.42659664e-01 -5.97948551e-01 1.27687231e-01 9.26948786e-01
-1.01165593e-01 -8.93934488e-01 -8.01664352e-01 -8.04015577e-01
-1.11234568e-01 1.50424075e+00 -7.77362347e-01 -1.11644067e-01
-1.96656629e-01 4.78070695e-04 1.42621562e-01 5.67329228e-01
2.85158187e-01 -1.19459021e+00 -1.08996582e+00 -1.38558000e-02
-4.33760285e-02 -3.91343445e-01 5.61574399e-02 -9.56501216e-02
-1.30419731e+00 -1.03762555e+00 -4.40501906e-02 -5.87248981e-01
9.34658170e-01 2.43845090e-01 1.23915040e+00 1.04598439e+00
1.15061603e-01 7.04483807e-01 -3.12352359e-01 -7.37689555e-01
-7.66078353e-01 3.51819992e-02 -2.94052035e-01 -5.14142036e-01
3.74775052e-01 -4.88921672e-01 -3.09975952e-01 2.19687060e-01
-8.79477143e-01 9.24304903e-01 4.45799649e-01 3.07952762e-01
-1.03490740e-01 -4.04419988e-01 1.58507377e-01 -1.47575188e+00
1.00370967e+00 -2.10023895e-01 -4.79909360e-01 1.92804977e-01
-1.26506805e+00 1.80315778e-01 2.54026532e-01 -3.30412984e-01
-1.00739050e+00 -1.86453387e-01 2.34723344e-01 1.40278429e-01
-4.25209492e-01 7.08883703e-01 -4.89496626e-02 -1.64321437e-01
6.77125275e-01 -4.61881422e-02 3.30255814e-02 -5.42477429e-01
8.41102004e-01 4.91724044e-01 4.68493700e-02 -3.69041175e-01
8.79443109e-01 5.02182245e-02 -4.17821676e-01 -5.95972538e-01
-7.28983760e-01 1.49154523e-02 -7.45828211e-01 -2.91451901e-01
5.25038064e-01 -2.06315711e-01 -6.93867505e-01 3.03050261e-02
-1.28545129e+00 -7.28640497e-01 -1.79601833e-01 2.45264292e-01
-4.67791200e-01 7.30661601e-02 -5.55779278e-01 -1.02513814e+00
5.90596534e-02 -1.17092013e+00 4.17899042e-01 6.15195632e-01
-1.23963785e+00 -1.34479332e+00 -1.24876462e-01 3.68542582e-01
5.14912844e-01 6.69192597e-02 1.21031749e+00 -8.43480587e-01
-7.78519273e-01 -2.02052638e-01 1.55605912e-01 -1.52624592e-01
-9.79348551e-03 4.10046011e-01 -1.07403600e+00 -4.85400558e-02
-2.15396792e-01 -9.46032405e-02 2.49133691e-01 2.19839528e-01
1.02989209e+00 -5.96752822e-01 -7.06954360e-01 9.72396582e-02
1.04202354e+00 2.28754878e-01 5.45621812e-01 3.56517106e-01
4.94316757e-01 9.18395281e-01 4.87009019e-01 2.23328814e-01
5.43743789e-01 4.68211919e-01 4.09556329e-01 -3.39226633e-01
-7.38881007e-02 -4.65052634e-01 7.22574368e-02 5.72941780e-01
-1.13848336e-01 -2.20825016e-01 -1.07276237e+00 1.00776210e-01
-1.93378246e+00 -8.59409451e-01 -4.30517703e-01 2.27979374e+00
6.56768322e-01 1.91428870e-01 -1.16078043e-02 -4.73256931e-02
3.78412098e-01 7.99084008e-02 -4.54121858e-01 -9.15447176e-01
4.20102209e-01 -4.19748500e-02 -2.70409077e-01 1.00521719e+00
-4.94130284e-01 1.17430556e+00 7.21801949e+00 -1.51215807e-01
-1.09718406e+00 -2.45925114e-01 4.71535474e-01 1.99613109e-01
-5.90218544e-01 4.81286138e-01 -5.30668080e-01 -3.61746550e-02
9.40931201e-01 -1.93377301e-01 5.22558212e-01 8.34002256e-01
5.00035286e-01 -3.82092446e-01 -1.69125366e+00 6.30434752e-01
3.52139361e-02 -1.56708777e+00 2.21175715e-01 1.50240839e-01
6.11571431e-01 -5.05958974e-01 4.15327922e-02 2.56464303e-01
5.79717278e-01 -1.24947703e+00 5.65306187e-01 5.44598818e-01
3.67495120e-01 -4.99059230e-01 2.11711407e-01 6.25017703e-01
-8.69398296e-01 1.47043699e-02 1.09376281e-01 -9.53195870e-01
1.12713397e-01 -4.67831679e-02 -9.16743577e-01 2.27674380e-01
2.93446988e-01 6.62819922e-01 -6.86617136e-01 1.12779844e+00
-8.39951873e-01 5.92736244e-01 1.52844906e-01 -3.47702384e-01
2.72640176e-02 -1.06217898e-01 4.48858589e-01 1.15078628e+00
-1.33728877e-01 4.52472866e-01 2.96308905e-01 1.35383284e+00
3.93046498e-01 -4.04985100e-02 -5.60826838e-01 -4.34024811e-01
7.26006448e-01 1.06923103e+00 -5.31417370e-01 -5.53116918e-01
-4.10160184e-01 8.43533218e-01 2.53823251e-01 4.32790130e-01
-6.17112517e-01 -1.68330416e-01 7.78129518e-01 3.65888357e-01
-3.31647724e-01 -3.11818391e-01 -7.61011779e-01 -1.17000532e+00
-5.26366949e-01 -1.32680953e+00 1.93950385e-02 -1.11932325e+00
-6.72913671e-01 4.53578949e-01 1.89512014e-01 -7.62786150e-01
-6.59814835e-01 -2.66337305e-01 -8.16488922e-01 8.88045847e-01
-1.09662926e+00 -9.63286102e-01 -5.03110588e-01 -1.18882693e-01
7.10957646e-01 3.97393852e-01 1.18961275e+00 -3.59012425e-01
-4.81037259e-01 1.71704516e-01 -4.04141843e-01 -3.55061710e-01
5.28932214e-01 -1.25361049e+00 7.87505031e-01 5.45513451e-01
4.39701706e-01 1.27697289e+00 1.18096209e+00 -8.28105688e-01
-1.40126050e+00 -4.39216167e-01 1.21882010e+00 -9.95197356e-01
3.70353609e-01 -2.78380781e-01 -1.09383643e+00 1.11806560e+00
5.72886586e-01 -8.74151051e-01 9.12569523e-01 4.37764615e-01
-1.24958917e-01 5.32178342e-01 -9.66891587e-01 1.08532143e+00
9.28642511e-01 -3.41711521e-01 -6.47412777e-01 9.64069664e-02
4.43434656e-01 -4.34240580e-01 -3.52753669e-01 -9.93238948e-03
7.34998465e-01 -1.37077749e+00 6.48853242e-01 -8.05534184e-01
4.83469874e-01 -2.36900926e-01 4.32233483e-01 -1.45829523e+00
-4.01599407e-01 -9.55266058e-01 -1.50347412e-01 1.27008832e+00
8.60392630e-01 -8.01707149e-01 7.09097862e-01 1.39361680e+00
-1.59659982e-01 -8.45917702e-01 6.60095662e-02 -1.40109375e-01
-8.14349055e-02 -8.33432794e-01 5.24609029e-01 1.09691870e+00
7.75985479e-01 5.60370088e-01 -2.30003312e-01 5.09564504e-02
2.01347351e-01 -8.61646701e-03 1.24439621e+00 -1.78466487e+00
-3.84321630e-01 -5.60461640e-01 7.56201446e-02 -1.40054643e+00
-1.59366801e-01 -7.90295124e-01 -1.88213035e-01 -2.01448417e+00
1.81924716e-01 -2.23154664e-01 1.59505293e-01 5.01776218e-01
-2.83227977e-03 -1.11017421e-01 5.82820594e-01 3.74580175e-01
-3.82293701e-01 -1.59353931e-02 1.23345792e+00 3.25509638e-01
-5.41167974e-01 1.10403568e-01 -1.34961581e+00 1.04238439e+00
9.55215394e-01 -4.35626864e-01 -8.13884318e-01 -5.95483005e-01
2.50832021e-01 -1.87118620e-01 4.10075754e-01 -8.86005342e-01
2.07085803e-01 -5.72257817e-01 4.26502526e-01 -1.90256387e-01
4.08938348e-01 -8.28056872e-01 2.63146937e-01 6.67130291e-01
-8.94196332e-01 1.24218635e-01 3.77924949e-01 2.74890929e-01
2.07305178e-01 -3.60579729e-01 5.44969082e-01 -1.97271898e-01
-5.88403523e-01 -2.17107013e-01 -8.94681752e-01 -3.70039850e-01
9.65689480e-01 -5.56465089e-01 -2.54095584e-01 -9.36228454e-01
-7.09246576e-01 4.22893316e-01 7.55227566e-01 5.79836190e-01
4.45571363e-01 -8.76845777e-01 -2.64005512e-01 3.50493163e-01
1.24986954e-01 -3.10171187e-01 -1.40619501e-01 5.49840808e-01
-2.59688079e-01 6.96883082e-01 5.86106116e-03 -3.54892522e-01
-1.49236286e+00 2.93138087e-01 4.04922366e-01 -3.91613878e-02
-4.92521584e-01 9.01559055e-01 2.38566712e-01 -4.61122125e-01
5.10685086e-01 -4.51602846e-01 -4.51243311e-01 -2.70402938e-01
7.11113930e-01 2.24116340e-01 -3.49777669e-01 -1.27149150e-01
-1.99280143e-01 2.53248721e-01 -1.13406762e-01 -3.71509582e-01
1.08255363e+00 -3.44614893e-01 -6.07643500e-02 8.61774504e-01
4.00239140e-01 -3.27169932e-02 -1.10488999e+00 4.24044728e-02
3.25667381e-01 -5.14970779e-01 -1.85234532e-01 -1.37718117e+00
-4.24477726e-01 1.10768461e+00 2.02646345e-01 4.78734255e-01
5.90366125e-01 2.93797135e-01 1.06867015e-01 3.15798253e-01
4.51150499e-02 -7.68405080e-01 1.50557563e-01 3.17052037e-01
1.06740952e+00 -1.01034105e+00 5.31264618e-02 -5.21525741e-01
-9.85520840e-01 1.40460873e+00 9.67785537e-01 2.22415358e-01
3.96067828e-01 9.83072072e-02 1.86454639e-01 -4.45717514e-01
-1.13479805e+00 -6.39892966e-02 4.44081932e-01 9.32933569e-01
1.19024014e+00 -2.49496568e-02 -3.13100070e-01 4.60559011e-01
-3.88724416e-01 3.57716650e-01 6.10150278e-01 1.05970180e+00
-7.48770177e-01 -1.55452228e+00 -2.44016975e-01 4.32410419e-01
8.08339939e-03 -8.80307183e-02 -1.16161573e+00 1.04683447e+00
-1.05963960e-01 1.31336915e+00 -2.45363042e-01 -3.47292095e-01
3.27650666e-01 2.99874276e-01 4.58263606e-01 -1.06773365e+00
-9.70467746e-01 -2.22202897e-01 4.34049308e-01 -5.69589078e-01
-1.09793268e-01 -6.04594409e-01 -1.23922765e+00 -4.70838606e-01
-3.57240081e-01 4.29012090e-01 6.49197996e-01 1.03757739e+00
3.49735141e-01 3.46069306e-01 -5.53462980e-03 -8.27024341e-01
-2.27041796e-01 -9.06334519e-01 6.39102906e-02 -1.57400258e-02
1.01106107e-01 -4.51903939e-01 -2.32993424e-01 1.84314251e-01] | [9.002471923828125, 6.221385478973389] |
00cd7b9c-8259-4849-a3cf-db19c018b908 | learning-delays-in-spiking-neural-networks | 2306.17670 | null | https://arxiv.org/abs/2306.17670v1 | https://arxiv.org/pdf/2306.17670v1.pdf | Learning Delays in Spiking Neural Networks using Dilated Convolutions with Learnable Spacings | Spiking Neural Networks (SNNs) are a promising research direction for building power-efficient information processing systems, especially for temporal tasks such as speech recognition. In SNNs, delays refer to the time needed for one spike to travel from one neuron to another. These delays matter because they influence the spike arrival times, and it is well-known that spiking neurons respond more strongly to coincident input spikes. More formally, it has been shown theoretically that plastic delays greatly increase the expressivity in SNNs. Yet, efficient algorithms to learn these delays have been lacking. Here, we propose a new discrete-time algorithm that addresses this issue in deep feedforward SNNs using backpropagation, in an offline manner. To simulate delays between consecutive layers, we use 1D convolutions across time. The kernels contain only a few non-zero weights - one per synapse - whose positions correspond to the delays. These positions are learned together with the weights using the recently proposed Dilated Convolution with Learnable Spacings (DCLS). We evaluated our method on the Spiking Heidelberg Dataset (SHD) and the Spiking Speech Commands (SSC) benchmarks, which require detecting temporal patterns. We used feedforward SNNs with two hidden fully connected layers. We showed that fixed random delays help, and that learning them helps even more. Furthermore, our method outperformed the state-of-the-art in both SHD and SSC without using recurrent connections and with substantially fewer parameters. Our work demonstrates the potential of delay learning in developing accurate and precise models for temporal data processing. Our code is based on PyTorch / SpikingJelly and available at: https://github.com/Thvnvtos/SNN-delays | ['Timothée Masquelier', 'Ismail Khalfaoui-Hassani', 'Ilyass Hammouamri'] | 2023-06-30 | null | null | null | null | ['speech-recognition'] | ['speech'] | [ 2.25126222e-01 -5.86888790e-01 3.07027459e-01 -2.56242961e-01
-7.68436417e-02 -6.88761950e-01 4.96084601e-01 1.96171641e-01
-9.03859377e-01 6.09597564e-01 -2.47760937e-01 -2.86852598e-01
-9.68577266e-02 -6.88009977e-01 -9.54641879e-01 -1.06542134e+00
-5.54617345e-01 1.44036382e-01 8.87324750e-01 -3.02829564e-01
4.16453958e-01 7.63766110e-01 -1.65215659e+00 5.01044869e-01
3.14386040e-01 1.10338521e+00 3.68118823e-01 6.95146084e-01
-2.42063046e-01 5.37811041e-01 -5.59021235e-01 2.25523621e-01
2.49865770e-01 -2.18177557e-01 -3.52764189e-01 -6.25119746e-01
-2.22565532e-01 7.41141066e-02 -7.31769860e-01 6.08226657e-01
5.29106855e-01 -8.92584845e-02 4.09212947e-01 -1.18784606e+00
-3.35405469e-01 9.17812347e-01 -2.33065531e-01 7.13904321e-01
-4.60201770e-01 3.59499842e-01 5.89107215e-01 -6.42653823e-01
4.54632312e-01 9.23537433e-01 8.22939754e-01 7.69008934e-01
-1.54168153e+00 -8.71484756e-01 1.72526091e-01 6.74952716e-02
-1.10663402e+00 -4.50849354e-01 4.37365532e-01 -4.95975494e-01
1.48799074e+00 -6.68349564e-02 9.14383352e-01 1.17627418e+00
5.84859669e-01 5.34007311e-01 1.13459063e+00 -1.56829789e-01
5.38894117e-01 -5.68769753e-01 4.16515619e-01 1.12190492e-01
7.13738948e-02 4.39921230e-01 -7.66464293e-01 1.17476553e-01
1.22079301e+00 1.68269828e-01 -2.20140368e-01 2.64402390e-01
-1.34869683e+00 3.28411907e-01 4.13809955e-01 6.48363888e-01
-2.87478745e-01 7.06907332e-01 4.38721240e-01 6.35867238e-01
1.12259351e-01 3.49157482e-01 -6.41789556e-01 -2.99360693e-01
-7.51414001e-01 3.19733977e-01 7.85542428e-01 7.16413856e-01
5.85751295e-01 5.94289601e-02 -2.82514364e-01 7.47722268e-01
-1.41965583e-01 4.01115626e-01 4.11893755e-01 -1.03942156e+00
1.34406686e-01 3.50475997e-01 -6.22693598e-02 -5.34722149e-01
-6.98242247e-01 -3.99729282e-01 -1.01823473e+00 3.42353106e-01
9.39546645e-01 -2.18684990e-02 -1.13800347e+00 1.93720984e+00
-4.60941166e-01 5.44911146e-01 -3.30458693e-02 7.24527359e-01
3.01848382e-01 7.88232267e-01 -1.29045993e-01 -1.41004428e-01
1.01934552e+00 -5.28326035e-01 -5.89681089e-01 -2.09488109e-01
4.72903579e-01 -5.55307209e-01 6.93617702e-01 4.54539061e-01
-1.18136561e+00 -2.04708457e-01 -9.33375239e-01 1.09106466e-01
-5.05938411e-01 -2.24973544e-01 8.03286612e-01 9.13185701e-02
-1.33692396e+00 1.17789781e+00 -1.44233060e+00 -3.21266234e-01
4.78282839e-01 7.56963253e-01 9.29662287e-02 4.27182496e-01
-1.12276721e+00 5.82214713e-01 2.95132212e-02 1.04762159e-01
-9.13314402e-01 -8.76022458e-01 -1.82184637e-01 -1.04236946e-01
-3.25787902e-01 -3.52150828e-01 1.45802248e+00 -7.42040336e-01
-1.52471936e+00 8.83190036e-01 -3.13382566e-01 -9.85927582e-01
2.91776806e-01 3.88867706e-01 -2.54203498e-01 -6.71990737e-02
-3.56858879e-01 8.93426239e-01 4.67518419e-01 -6.82306051e-01
-3.47289920e-01 -3.12723309e-01 -1.99195504e-01 -4.56086487e-01
-1.56759202e-01 7.23809078e-02 -2.98912466e-01 -6.69476211e-01
2.36104459e-01 -9.70916212e-01 -3.11968684e-01 3.12363595e-01
-7.09987283e-02 -1.00711867e-01 4.44500387e-01 -1.49549603e-01
1.02518451e+00 -2.31329489e+00 8.79331008e-02 1.54143542e-01
2.52990305e-01 3.23758543e-01 -2.75007993e-01 4.78770941e-01
-1.49438724e-01 -4.03086804e-02 -5.76615214e-01 -1.81360513e-01
-1.42858714e-01 4.30775434e-01 -5.50931156e-01 2.70358264e-01
3.45671207e-01 7.96402395e-01 -7.95136988e-01 2.73450375e-01
-1.68780833e-01 6.79639697e-01 -3.11026812e-01 -4.99821194e-02
-2.37410575e-01 5.06543219e-01 7.43861794e-02 3.22323710e-01
6.47840500e-01 -1.97053224e-01 -2.59648472e-01 9.41646174e-02
-8.53090167e-01 5.63049257e-01 -8.63145888e-01 1.51412129e+00
-1.87493756e-01 8.65672171e-01 -2.15888917e-01 -1.05818486e+00
1.08123565e+00 1.75515875e-01 5.65341294e-01 -1.02072501e+00
2.94851065e-01 6.45202577e-01 4.78544652e-01 -1.05926760e-01
-1.64655849e-01 1.60402477e-01 2.08089575e-01 4.59985524e-01
8.94986093e-03 7.74786100e-02 4.54168826e-01 4.31237015e-04
1.72704971e+00 -4.14712846e-01 -3.76194179e-01 -2.43236288e-01
-6.68335930e-02 -2.40672722e-01 6.26988769e-01 7.07271934e-01
-1.79315820e-01 4.79180425e-01 7.97613144e-01 -3.78818780e-01
-1.17742252e+00 -1.35953534e+00 -3.81739020e-01 6.81831598e-01
-1.70774758e-02 -3.06920540e-02 -6.82522893e-01 4.13615972e-01
1.21334262e-01 3.33977878e-01 -6.62790060e-01 -1.25043586e-01
-6.84326589e-01 -5.91234803e-01 1.12933910e+00 7.79763579e-01
2.93558151e-01 -1.42664361e+00 -9.13459420e-01 7.00284123e-01
3.17442179e-01 -1.06472719e+00 -3.29264641e-01 1.21144450e+00
-1.16165841e+00 -6.64172471e-01 -9.06366587e-01 -8.88189971e-01
5.55933774e-01 1.22422017e-01 6.55019283e-01 6.44734576e-02
-5.81902027e-01 -6.48309663e-02 -7.61348978e-02 -4.86087710e-01
2.13629170e-03 -2.45426353e-02 1.59225807e-01 -2.68318146e-01
4.80437607e-01 -1.18858135e+00 -5.96268475e-01 3.46329808e-01
-1.08324873e+00 1.53444722e-01 4.19970393e-01 8.26786280e-01
8.43089938e-01 -3.18798244e-01 5.35076559e-01 -6.08863711e-01
4.54245836e-01 -2.82115430e-01 -9.27110732e-01 -1.47015929e-01
-2.38579705e-01 4.50970978e-01 7.96661437e-01 -8.01593840e-01
-3.14592123e-01 1.15699522e-01 -1.92651436e-01 -3.33734781e-01
-1.45124227e-01 2.03206494e-01 6.00246608e-01 -2.97833294e-01
7.19007075e-01 3.92637908e-01 -2.58701388e-02 -1.87253565e-01
-1.78263843e-01 2.01607615e-01 3.82546157e-01 -5.17468333e-01
3.91480297e-01 6.83488369e-01 -7.29687139e-02 -7.38185823e-01
-7.28154257e-02 -3.89830738e-01 -5.00357628e-01 -8.60368460e-02
5.09103954e-01 -5.30837715e-01 -1.03942525e+00 1.20343280e+00
-1.55611765e+00 -1.06154239e+00 -2.09410697e-01 4.38212395e-01
-7.78744876e-01 -2.55825698e-01 -1.22766328e+00 -6.83529377e-01
-1.07857853e-01 -8.72346640e-01 5.76879561e-01 3.09177518e-01
-6.69562817e-02 -8.35262477e-01 1.72415167e-01 -7.46804237e-01
7.98724115e-01 -3.80704068e-02 8.39347422e-01 -4.28415447e-01
-6.45204842e-01 2.48250186e-01 -2.60358423e-01 1.33711204e-01
-1.72584623e-01 3.18002731e-01 -1.13412988e+00 -1.59852043e-01
-2.04875425e-01 -6.09307550e-02 1.30093861e+00 7.66512930e-01
1.31808972e+00 1.19013498e-02 -4.47467744e-01 7.95317352e-01
1.40447640e+00 4.26244140e-01 8.30150664e-01 2.67437249e-01
7.26978555e-02 4.98655587e-01 -1.78112999e-01 6.91427290e-01
-6.46712407e-02 3.49629313e-01 4.07241940e-01 4.38475143e-03
-5.59775494e-02 2.05259681e-01 3.88799697e-01 1.05002761e+00
-1.77300170e-01 -2.12902650e-01 -1.00198269e+00 6.04200661e-01
-1.83141911e+00 -1.03351665e+00 -2.99072325e-01 2.27561450e+00
1.06356657e+00 4.68236238e-01 -1.40349520e-03 4.51386124e-01
5.70205271e-01 -2.01707304e-01 -9.58402336e-01 -4.30406302e-01
-5.57409406e-01 7.06925333e-01 8.54379952e-01 2.68802911e-01
-6.06549799e-01 8.52589905e-01 6.02872562e+00 3.20081979e-01
-1.62686825e+00 -1.45329818e-01 3.24995667e-01 -3.47192913e-01
-3.22946876e-01 -1.36955515e-01 -1.00430894e+00 5.77288985e-01
1.36264646e+00 -2.45157704e-01 9.70750451e-01 7.59732872e-02
5.48339486e-01 -1.01511672e-01 -1.17782235e+00 1.02799439e+00
-7.98058212e-01 -1.68042946e+00 -3.43672693e-01 -1.15778074e-01
6.23971760e-01 5.56774020e-01 1.81285694e-01 2.72671860e-02
1.87799513e-01 -1.01684082e+00 7.43351042e-01 6.39784515e-01
4.76678103e-01 -5.89820623e-01 4.32962298e-01 2.75115937e-01
-1.29794753e+00 -1.28025129e-01 -5.57799578e-01 -3.25793982e-01
1.21592451e-02 9.56302881e-01 -5.44995129e-01 -2.82489538e-01
9.11358476e-01 8.96917284e-01 -1.87656149e-01 1.56463063e+00
8.16421434e-02 7.38477170e-01 -6.99568212e-01 -3.86791378e-01
2.76273698e-01 1.30906627e-01 2.29136795e-01 1.34740317e+00
6.68541849e-01 2.13463336e-01 -5.38818777e-01 1.20610833e+00
-1.21521585e-01 -3.60009998e-01 -6.09185636e-01 -3.56414109e-01
8.50570560e-01 8.77220511e-01 -1.11013639e+00 1.90748289e-01
-1.74848005e-01 8.58045876e-01 2.88908929e-01 6.04071915e-01
-6.78012013e-01 -6.76135361e-01 9.17713642e-01 2.57933825e-01
3.94376457e-01 -6.86235249e-01 -5.87419271e-01 -6.43538654e-01
4.25411053e-02 -3.88718843e-01 -1.02059826e-01 -5.33896863e-01
-1.26662707e+00 6.64542913e-01 -4.19432014e-01 -1.08858824e+00
8.56759213e-03 -9.67951596e-01 -6.32121146e-01 9.23738241e-01
-1.61023724e+00 -3.41948360e-01 -3.58839855e-02 6.90240681e-01
3.05762351e-01 3.30460072e-01 8.18289578e-01 4.29337919e-01
-3.86081070e-01 4.54204381e-01 1.46233603e-01 1.20496444e-01
6.47762239e-01 -9.29868817e-01 7.80898988e-01 6.20582998e-01
-6.44819364e-02 7.86417305e-01 7.10660040e-01 -2.34127849e-01
-1.33085287e+00 -9.11078334e-01 9.62523699e-01 1.47748202e-01
1.01961315e+00 -7.14616179e-01 -1.21223676e+00 3.99170280e-01
2.75798384e-02 2.46447265e-01 4.24814671e-01 -4.00519729e-01
-4.16861445e-01 -2.43744075e-01 -7.14609563e-01 6.39354944e-01
1.31290662e+00 -4.49805498e-01 -1.15452334e-01 3.86117697e-02
7.83945084e-01 -3.11083108e-01 -5.16642809e-01 1.38838619e-01
5.62379122e-01 -9.76541102e-01 6.23875022e-01 -2.16459021e-01
2.14255601e-01 -4.14563090e-01 -3.15158032e-02 -1.33159554e+00
-3.07655185e-01 -6.06259704e-01 1.11769989e-01 8.40344965e-01
6.71397090e-01 -9.91107643e-01 5.52228868e-01 3.05811465e-01
-4.77332592e-01 -8.59787285e-01 -1.07699597e+00 -1.16539943e+00
2.59242922e-01 -5.23045897e-01 1.98619276e-01 4.57481980e-01
-1.13816345e-02 -2.26674095e-01 2.34181941e-01 2.46661171e-01
4.35819983e-01 -3.54687780e-01 3.22139561e-02 -1.31572962e+00
-2.80411839e-01 -8.97619545e-01 -5.41622877e-01 -1.11387777e+00
7.45939463e-03 -8.64052832e-01 3.75144452e-01 -1.26170599e+00
-2.78984129e-01 -7.09989488e-01 -6.91030324e-01 7.40246236e-01
5.21586835e-01 -2.64615659e-02 -8.44107345e-02 2.74226874e-01
-1.08108521e-01 1.64827004e-01 9.58856463e-01 -8.57494548e-02
-2.19066769e-01 1.44327775e-01 -2.70631583e-03 4.47531223e-01
1.15193963e+00 -7.13861704e-01 -2.13824749e-01 -6.78623199e-01
2.32991621e-01 -3.46633121e-02 4.82507050e-01 -1.38250530e+00
8.66460621e-01 1.31107271e-01 1.21705264e-01 -5.96634865e-01
3.92030716e-01 -5.75454295e-01 1.20219573e-01 7.94465244e-01
-5.75955451e-01 2.52985805e-01 6.90332174e-01 3.58786136e-01
-4.03029807e-02 1.33294435e-02 8.85019660e-01 4.11295034e-02
-6.96114302e-01 4.50234413e-01 -8.56292307e-01 6.38555661e-02
8.51233006e-01 -1.25238106e-01 -5.68655133e-01 -2.01534498e-02
-6.15503907e-01 9.27701443e-02 3.10723305e-01 1.43725097e-01
7.55870342e-01 -1.01628661e+00 -4.24421281e-01 4.02789563e-01
3.28529030e-02 -2.25871980e-01 1.30737424e-01 1.05069780e+00
-5.60146332e-01 5.82798541e-01 -6.20349765e-01 -8.03347230e-01
-8.30742359e-01 3.28725547e-01 5.09571612e-01 2.79783886e-02
-4.72750545e-01 1.20016909e+00 -1.03122614e-01 -2.54765958e-01
6.58523083e-01 -1.05012202e+00 1.37604252e-01 -1.55228302e-01
7.05366433e-01 8.54080617e-02 3.55966598e-01 1.81422740e-01
-4.58716452e-01 6.07126236e-01 -8.92903134e-02 -4.06558573e-01
1.61608219e+00 3.71350139e-01 -3.12379777e-01 1.11689913e+00
1.04856324e+00 -5.73450744e-01 -1.69164920e+00 -1.97099254e-01
2.61922985e-01 1.46537051e-01 -1.90523252e-01 -7.16869950e-01
-1.11492538e+00 1.31755710e+00 6.89620733e-01 2.46425644e-01
1.14775181e+00 -1.40727594e-01 9.89246309e-01 5.61649978e-01
6.94527507e-01 -8.36015403e-01 1.16030306e-01 1.07970536e+00
6.80549264e-01 -5.49735308e-01 -9.38238621e-01 -1.70983955e-01
-1.57836944e-01 1.42813838e+00 4.40114439e-01 -4.16756541e-01
7.73518085e-01 1.19523239e+00 1.14728406e-01 1.90373272e-01
-1.37348890e+00 -1.65458545e-01 -4.37254876e-01 5.21628141e-01
5.79363048e-01 4.18874621e-02 -2.14615837e-01 5.21700501e-01
1.27555400e-01 3.65260899e-01 4.52246904e-01 9.73244727e-01
-5.90806544e-01 -1.18209696e+00 -6.51387274e-02 4.42829698e-01
-2.33369067e-01 -4.53843445e-01 -1.33679315e-01 2.75896251e-01
-4.24416363e-02 6.71776831e-01 5.55186152e-01 -5.88617206e-01
2.38528222e-01 -4.50061560e-02 7.42364347e-01 -4.50669169e-01
-8.00228417e-01 -1.20154291e-01 -4.03624505e-01 -6.21979535e-01
-2.51925170e-01 -5.92870295e-01 -2.18497586e+00 -5.64228296e-01
1.50915131e-01 -1.33088887e-01 7.59244978e-01 7.06659734e-01
4.62702423e-01 9.17315066e-01 4.92195725e-01 -8.75008225e-01
-2.73520559e-01 -4.91914690e-01 -6.57393277e-01 -6.85428306e-02
4.51717913e-01 -4.65408623e-01 -6.84308708e-01 7.29282871e-02] | [8.158945083618164, 2.5601913928985596] |
6d78fc16-700f-44cb-9042-eb7b3c192a16 | attention-based-end-to-end-speech-recognition | 1707.07167 | null | http://arxiv.org/abs/1707.07167v3 | http://arxiv.org/pdf/1707.07167v3.pdf | Attention-Based End-to-End Speech Recognition on Voice Search | Recently, there has been a growing interest in end-to-end speech recognition
that directly transcribes speech to text without any predefined alignments. In
this paper, we explore the use of attention-based encoder-decoder model for
Mandarin speech recognition on a voice search task. Previous attempts have
shown that applying attention-based encoder-decoder to Mandarin speech
recognition was quite difficult due to the logographic orthography of Mandarin,
the large vocabulary and the conditional dependency of the attention model. In
this paper, we use character embedding to deal with the large vocabulary.
Several tricks are used for effective model training, including L2
regularization, Gaussian weight noise and frame skipping. We compare two
attention mechanisms and use attention smoothing to cover long context in the
attention model. Taken together, these tricks allow us to finally achieve a
character error rate (CER) of 3.58% and a sentence error rate (SER) of 7.43% on
the MiTV voice search dataset. While together with a trigram language model,
CER and SER reach 2.81% and 5.77%, respectively. | ['Junbo Zhang', 'Yujun Wang', 'Changhao Shan', 'Lei Xie'] | 2017-07-22 | null | null | null | null | ['l2-regularization'] | ['methodology'] | [ 2.66000777e-01 9.19440091e-02 2.55337097e-02 -3.52320701e-01
-1.07800031e+00 -2.67311037e-01 3.92108649e-01 -2.45042577e-01
-6.56417370e-01 6.18213773e-01 4.93899196e-01 -7.35800207e-01
3.39157313e-01 -2.57779241e-01 -6.55778944e-01 -5.81829667e-01
4.90797430e-01 1.96386859e-01 1.04558103e-01 1.71408951e-02
1.67533264e-01 3.09538841e-01 -1.10235250e+00 9.26129986e-03
8.48635256e-01 7.21623898e-01 7.30044842e-01 1.04214168e+00
-2.31067821e-01 7.43611515e-01 -7.37748325e-01 -4.90188271e-01
-1.24088779e-01 -5.12808561e-01 -8.00095975e-01 1.46277487e-01
3.25669706e-01 -3.20268333e-01 -7.09683597e-01 9.77298856e-01
7.20284343e-01 2.75023133e-01 5.40961742e-01 -5.25327623e-01
-9.55557227e-01 6.12618685e-01 -2.36237288e-01 5.68956971e-01
7.76924193e-02 2.66095903e-02 1.03642893e+00 -1.21557558e+00
2.13897228e-01 1.30721855e+00 3.32238853e-01 7.86935031e-01
-7.75676489e-01 -3.07782471e-01 6.51042908e-02 4.95467991e-01
-1.47425067e+00 -8.99523735e-01 5.43869317e-01 -2.16407537e-01
1.65857422e+00 4.67354298e-01 2.82141060e-01 9.29319561e-01
1.84817284e-01 6.38359725e-01 7.54029334e-01 -1.00312364e+00
2.31826119e-02 2.12148085e-01 2.63469487e-01 3.48178834e-01
-1.45104617e-01 2.25961092e-03 -5.07308066e-01 2.94178575e-01
6.07906163e-01 -2.91946739e-01 -3.61376792e-01 4.24234241e-01
-8.50833535e-01 7.83205032e-01 9.63871852e-02 2.58710176e-01
-3.21910679e-01 1.44708887e-01 4.60718483e-01 2.03690454e-01
2.35643134e-01 8.33550319e-02 -3.78333986e-01 -3.73547524e-01
-8.67597342e-01 -4.83914793e-01 6.70529246e-01 1.18315220e+00
4.07299787e-01 5.65792501e-01 -1.64872468e-01 1.17238081e+00
4.61651951e-01 6.89837098e-01 7.15216696e-01 -4.82009619e-01
6.69444203e-01 -8.81532207e-02 -2.79704511e-01 -3.56180102e-01
2.11437598e-01 -2.57230848e-01 -6.74467802e-01 -2.23443389e-01
1.88722029e-01 -2.07559660e-01 -1.14001381e+00 1.61324096e+00
-7.15044420e-03 -6.84202239e-02 1.98785484e-01 8.72527957e-01
5.19078374e-01 9.37252164e-01 2.86570215e-03 -2.96604693e-01
1.34001899e+00 -1.25482249e+00 -1.21718180e+00 -4.37420189e-01
5.49603403e-01 -1.12038660e+00 1.34682763e+00 5.47597148e-02
-1.18225098e+00 -5.86896777e-01 -1.07050025e+00 -3.51909399e-01
-1.24013089e-01 3.48578304e-01 -1.46540284e-01 8.09735715e-01
-1.05148542e+00 1.98857591e-01 -7.71726429e-01 -3.45716417e-01
7.02023227e-03 4.04305488e-01 -7.59122372e-02 -6.43721148e-02
-1.16355479e+00 1.03573143e+00 1.56638354e-01 8.40917602e-02
-7.55582392e-01 -9.41414088e-02 -7.53178895e-01 2.93200701e-01
1.87344670e-01 -1.18153058e-01 1.24809444e+00 -7.73966014e-01
-1.82399893e+00 5.56164145e-01 -6.12349629e-01 -5.93328059e-01
-2.14373153e-02 -4.95658964e-01 -5.73374391e-01 2.78048925e-02
-4.64130223e-01 5.15214801e-01 9.16002452e-01 -4.96688098e-01
-3.85571718e-01 -1.21613614e-01 -2.98495263e-01 2.15690434e-01
-4.18308973e-01 6.28700197e-01 -6.69495583e-01 -8.28200400e-01
-1.35026336e-01 -8.97531688e-01 2.81624999e-02 -5.96699119e-01
-2.02719659e-01 -3.66773188e-01 8.23706865e-01 -1.27190840e+00
1.67213738e+00 -2.32294679e+00 2.02956349e-01 -1.73813671e-01
-3.72826159e-01 7.13388562e-01 -1.11623809e-01 3.12569141e-01
1.01646863e-01 3.21006715e-01 -1.50499806e-01 -4.97933030e-01
-1.66501135e-01 3.79194200e-01 -4.24669176e-01 2.35581994e-01
5.08303106e-01 9.59918261e-01 -3.86379600e-01 -3.49660814e-01
2.32434750e-01 7.58962393e-01 -5.52867591e-01 2.61298060e-01
4.56804149e-02 2.26920202e-01 3.70055065e-02 5.63088655e-01
3.29778969e-01 2.38996893e-01 1.82680354e-01 2.64119685e-01
-1.53220803e-01 1.06500709e+00 -6.64319932e-01 1.23711073e+00
-6.38144970e-01 9.45589066e-01 2.80575573e-01 -6.65448487e-01
8.42740357e-01 5.44025242e-01 -3.30919087e-01 -6.95913792e-01
1.83692530e-01 3.91724199e-01 3.13306421e-01 -4.48544025e-01
5.47079921e-01 -1.95314035e-01 2.59049267e-01 1.45481750e-01
2.34552875e-01 -8.27746559e-03 -1.44810975e-01 -2.29262397e-01
8.30540061e-01 -2.39236340e-01 2.20913682e-02 -2.52607107e-01
7.74667144e-01 -4.43453699e-01 4.84218568e-01 5.12157142e-01
-1.13562644e-01 8.59017074e-01 1.74059510e-01 -8.40876102e-02
-1.28249717e+00 -6.69810891e-01 -1.14977524e-01 9.65909719e-01
-3.92348915e-01 -6.35845482e-01 -9.45033133e-01 -4.83079880e-01
-4.40384597e-01 1.02422750e+00 3.93291190e-02 -2.44861081e-01
-1.06296396e+00 -3.18418860e-01 6.34674489e-01 5.79129398e-01
2.66427130e-01 -1.20801675e+00 -1.02323033e-01 3.11313361e-01
-1.28045499e-01 -1.33049595e+00 -1.06959963e+00 3.88215601e-01
-7.22950757e-01 -5.46627223e-01 -9.62816238e-01 -1.15102470e+00
5.40651560e-01 8.51259008e-02 7.13805199e-01 6.52148500e-02
1.07964978e-01 8.68882984e-02 -4.80540633e-01 -3.20085615e-01
-6.72902226e-01 2.01753318e-01 2.59782940e-01 -2.96386071e-02
5.55372000e-01 -1.64717197e-01 -1.57786071e-01 3.05290639e-01
-6.90973580e-01 -3.02839071e-01 6.23631835e-01 9.54469264e-01
5.37403286e-01 -4.39366758e-01 5.34186840e-01 -3.30634922e-01
5.54819047e-01 -1.29562229e-01 -6.60760045e-01 3.36136580e-01
-5.84981859e-01 1.58190638e-01 6.53717458e-01 -5.85887671e-01
-8.44372928e-01 6.20483533e-02 -7.18329966e-01 -5.80581665e-01
-1.25690490e-01 4.46594149e-01 -4.33288217e-01 2.15268090e-01
1.71932191e-01 5.00308871e-01 5.97807281e-02 -8.21688116e-01
1.32607266e-01 1.44536889e+00 2.71697521e-01 5.06924763e-02
2.53267258e-01 -3.23834568e-01 -6.30237401e-01 -1.50405312e+00
-5.17568111e-01 -4.19215441e-01 -4.57764596e-01 1.17735215e-01
1.00509834e+00 -8.95171285e-01 -3.19741994e-01 3.94537657e-01
-1.47935343e+00 -3.74060869e-01 -1.78665876e-01 8.65536392e-01
-4.06821728e-01 7.49968827e-01 -7.50981748e-01 -1.14313424e+00
-5.51981330e-01 -1.35129690e+00 8.09904516e-01 -9.67804044e-02
-3.12747866e-01 -7.47405410e-01 -2.78935313e-01 4.81050074e-01
6.76061034e-01 -8.98797750e-01 8.62315714e-01 -7.03769386e-01
-5.92386305e-01 6.08308725e-02 -2.06988454e-02 8.86344612e-01
1.24371767e-01 -1.31284565e-01 -1.13880658e+00 -2.61185378e-01
3.35801601e-01 -3.12251672e-02 7.73956180e-01 3.58028144e-01
9.83462572e-01 -4.78977293e-01 1.02303185e-01 4.49785441e-01
1.03571916e+00 6.87375605e-01 1.02067435e+00 -7.83344656e-02
6.73803151e-01 3.33688438e-01 3.77646118e-01 7.14404434e-02
1.01907969e-01 9.16584611e-01 -5.92505261e-02 3.49055707e-01
-6.41499758e-01 -2.59156734e-01 8.13153386e-01 1.78438091e+00
2.80530691e-01 -5.57544231e-01 -1.05969405e+00 6.74693763e-01
-1.50587797e+00 -6.70046091e-01 -1.51816040e-01 2.42125440e+00
9.52212691e-01 2.24536225e-01 -2.21873939e-01 4.78464551e-02
9.26457226e-01 3.85939926e-02 -1.65892497e-01 -9.47420359e-01
-2.44791567e-01 2.62495399e-01 5.13114572e-01 1.12283623e+00
-8.34439158e-01 1.30227447e+00 6.48180294e+00 1.11554635e+00
-1.23936117e+00 2.54524559e-01 4.90821302e-01 8.47634673e-02
-2.08156884e-01 2.36391816e-02 -1.30895019e+00 5.91662228e-01
1.57340991e+00 4.87286001e-02 5.76886058e-01 6.59382463e-01
1.22256927e-01 1.10762350e-01 -8.41935396e-01 1.18604493e+00
3.06635827e-01 -1.01118159e+00 7.97057450e-02 8.02312046e-02
3.70122612e-01 1.92656353e-01 -8.08561295e-02 3.93050879e-01
-1.94211096e-01 -1.18774414e+00 6.82709813e-01 2.32285723e-01
1.06430805e+00 -6.22411013e-01 7.70519078e-01 5.29235899e-01
-9.76077437e-01 2.10195914e-01 -4.62904602e-01 -8.07701871e-02
4.05834109e-01 3.20302784e-01 -1.07590580e+00 7.71987960e-02
2.72490501e-01 1.63174391e-01 -1.68510169e-01 9.32206392e-01
-3.43265623e-01 1.08370686e+00 -5.07967174e-01 -4.31771755e-01
2.79434115e-01 -1.34436637e-01 6.21489644e-01 1.42665076e+00
4.18484926e-01 -5.54554872e-02 -3.33435267e-01 5.84012747e-01
-2.09286273e-01 1.15221269e-01 -3.70005220e-01 -3.57755065e-01
4.63943809e-01 6.63249135e-01 -2.10559502e-01 -2.61451900e-01
-5.71290016e-01 1.21561968e+00 3.96494359e-01 3.80577087e-01
-8.68309498e-01 -6.16299927e-01 6.69985056e-01 4.13208501e-03
8.78671646e-01 -7.14837968e-01 -1.16447821e-01 -1.09521711e+00
2.67057568e-01 -9.13424075e-01 -3.04151982e-01 -4.90915895e-01
-9.20912862e-01 9.21988726e-01 -4.09311086e-01 -7.49395967e-01
-3.29367846e-01 -6.32588506e-01 -6.59460008e-01 1.28506303e+00
-1.60566640e+00 -6.90862715e-01 3.62084597e-01 1.47094563e-01
1.26823092e+00 -4.16280627e-01 9.19158936e-01 5.52225590e-01
-7.51732767e-01 8.03650200e-01 2.88927048e-01 2.43831351e-01
5.13499856e-01 -9.12760556e-01 8.00543249e-01 1.00428486e+00
4.67578560e-01 6.52700603e-01 3.67124647e-01 -5.96685112e-01
-1.42680359e+00 -9.14216578e-01 1.68527102e+00 -3.69531661e-01
5.33907712e-01 -5.77288985e-01 -1.09679902e+00 5.42221665e-01
5.98529339e-01 -2.40047365e-01 6.20916426e-01 -5.09817749e-02
5.38232736e-02 4.90036374e-03 -5.71611643e-01 7.03646898e-01
8.09256136e-01 -8.48420382e-01 -8.29377234e-01 1.57505706e-01
9.28859591e-01 -2.35582709e-01 -5.38233876e-01 7.43336603e-02
2.81574309e-01 -4.47540879e-01 6.30316257e-01 -5.28005421e-01
3.98624092e-02 -6.49950206e-02 -5.15414298e-01 -1.11721838e+00
-3.40497166e-01 -8.50323439e-01 -2.10306346e-01 1.48627281e+00
6.69734299e-01 -3.82621765e-01 3.87688100e-01 6.53258622e-01
-4.47397888e-01 -6.65485322e-01 -1.32701862e+00 -9.79480028e-01
2.25170657e-01 -4.41572279e-01 2.57034123e-01 3.70577216e-01
-4.82852347e-02 8.35558355e-01 -4.80216920e-01 1.62978515e-01
1.29070073e-01 -4.60608959e-01 2.97810942e-01 -6.07933521e-01
-2.88074583e-01 -1.82447523e-01 -1.23386852e-01 -1.70234823e+00
2.36930847e-01 -7.52718151e-01 1.92769542e-01 -1.24392807e+00
-2.01758057e-01 -2.33415827e-01 -2.17679322e-01 2.73278773e-01
-8.58889148e-02 -2.25994334e-01 3.60761493e-01 1.33118302e-01
-3.03080797e-01 8.96239400e-01 7.81520665e-01 -6.39713928e-02
-1.41225696e-01 -2.08470467e-02 -3.35540712e-01 5.69987774e-01
1.05353963e+00 -4.45234895e-01 -1.00048959e-01 -9.63095069e-01
-2.93684423e-01 8.36876556e-02 -8.12062025e-02 -8.82376909e-01
3.73470396e-01 2.23187774e-01 -1.86230727e-02 -5.73833644e-01
6.69959724e-01 -7.82613635e-01 -2.23564550e-01 2.94024497e-01
-2.42068917e-01 6.34558350e-02 3.21002483e-01 4.12543386e-01
-5.06227195e-01 -6.79792762e-01 6.81961954e-01 1.83501661e-01
-5.55516243e-01 -9.99045968e-02 -7.70532906e-01 1.14804320e-01
6.51040256e-01 -1.33463427e-01 3.19974199e-02 -4.91446942e-01
-6.96446359e-01 3.09823547e-02 3.76661010e-02 5.49156129e-01
8.06230605e-01 -1.21627128e+00 -7.83838987e-01 7.41309464e-01
-3.41502577e-01 -3.62874866e-01 4.18250076e-02 8.11915696e-01
-4.26199645e-01 8.45308602e-01 1.96458653e-01 -3.21408987e-01
-1.62306225e+00 4.40730751e-01 1.79610461e-01 1.28823340e-01
-3.75067502e-01 9.81801569e-01 -9.36280340e-02 -1.17606111e-01
5.40979862e-01 -3.83816928e-01 7.30224475e-02 -2.23663256e-01
6.05818868e-01 2.37789899e-01 2.02382267e-01 -8.33604932e-01
-5.61731040e-01 5.42261600e-01 -2.81898797e-01 -2.17360020e-01
1.03639650e+00 -3.84622157e-01 1.86967328e-01 3.31794471e-01
1.21745014e+00 3.98199946e-01 -1.06528914e+00 -1.24357842e-01
1.05359167e-01 -2.51008064e-01 3.22248071e-01 -5.40980816e-01
-6.58530056e-01 1.40061426e+00 3.44162077e-01 1.59722015e-01
8.90494764e-01 4.21212427e-02 1.05137146e+00 3.80229294e-01
-2.01350957e-01 -1.34214759e+00 -3.38411182e-01 1.29957962e+00
1.04825020e+00 -1.15266836e+00 -4.18330818e-01 -4.07208681e-01
-7.44355559e-01 1.14410913e+00 3.24063271e-01 1.44433558e-01
5.07364213e-01 3.50155383e-01 8.39011595e-02 4.34338152e-01
-8.69729817e-01 -3.27506512e-01 3.59378040e-01 4.49376881e-01
6.39319420e-01 -1.69039115e-01 -3.11811328e-01 6.46560192e-01
-2.19655469e-01 -2.92669952e-01 3.55652004e-01 6.55741513e-01
-6.99002147e-01 -1.22516167e+00 -4.48709190e-01 1.45247251e-01
-6.88442349e-01 -7.52209306e-01 -4.83672798e-01 2.75392294e-01
-3.91423583e-01 1.17186213e+00 2.97408640e-01 -3.65473121e-01
1.93409607e-01 6.12093389e-01 2.11315647e-01 -6.33408964e-01
-5.10783195e-01 5.42081654e-01 2.65914857e-01 -4.51502725e-02
3.15725766e-02 -4.33441192e-01 -1.12973356e+00 -1.14927493e-01
-7.56923199e-01 1.66414350e-01 8.15025151e-01 8.52209091e-01
5.22581160e-01 5.27491510e-01 4.29443777e-01 -3.63611907e-01
-9.65801179e-01 -1.29882872e+00 -4.02389050e-01 -9.63979885e-02
5.36927223e-01 -1.44918665e-01 -5.35631597e-01 7.06801936e-02] | [14.464195251464844, 6.900814533233643] |
e8a88af3-4987-4daa-bdfd-a8823b15a9f9 | predict-persian-reverse-dictionary | 2105.00309 | null | https://arxiv.org/abs/2105.00309v2 | https://arxiv.org/pdf/2105.00309v2.pdf | PREDICT: Persian Reverse Dictionary | Finding the appropriate words to convey concepts (i.e., lexical access) is essential for effective communication. Reverse dictionaries fulfill this need by helping individuals to find the word(s) which could relate to a specific concept or idea. To the best of our knowledge, this resource has not been available for the Persian language. In this paper, we compare four different architectures for implementing a Persian reverse dictionary (PREDICT). We evaluate our models using (phrase,word) tuples extracted from the only Persian dictionaries available online, namely Amid, Moein, and Dehkhoda where the phrase describes the word. Given the phrase, a model suggests the most relevant word(s) in terms of the ability to convey the concept. The model is considered to perform well if the correct word is one of its top suggestions. Our experiments show that a model consisting of Long Short-Term Memory (LSTM) units enhanced by an additive attention mechanism is enough to produce suggestions comparable to (or in some cases better than) the word in the original dictionary. The study also reveals that the model sometimes produces the synonyms of the word as its output which led us to introduce a new metric for the evaluation of reverse dictionaries called Synonym Accuracy accounting for the percentage of times the event of producing the word or a synonym of it occurs. The assessment of the best model using this new metric also indicates that at least 62% of the times, it produces an accurate result within the top 100 suggestions. | ['Ali Mohades', 'Amin Gheibi', 'Arman Malekzadeh'] | 2021-05-01 | null | null | null | null | ['reverse-dictionary'] | ['natural-language-processing'] | [-1.41663387e-01 4.67331745e-02 -1.56956211e-01 -1.36653945e-01
-3.31268072e-01 -5.69471598e-01 7.03364074e-01 8.60146165e-01
-8.55612874e-01 8.17916989e-01 3.62340868e-01 -3.47417116e-01
-1.93183333e-01 -8.17203879e-01 -2.78988868e-01 -3.16886574e-01
3.35189134e-01 8.58835042e-01 -1.79442242e-01 -6.36757493e-01
5.88996887e-01 3.80269706e-01 -1.34333444e+00 1.19991466e-01
8.64519775e-01 6.69034898e-01 6.16699994e-01 3.20693433e-01
-4.23553854e-01 1.89637899e-01 -1.17212212e+00 -7.62977600e-01
-5.92265241e-02 -4.12776649e-01 -9.33095455e-01 -3.84947747e-01
2.86973327e-01 -9.44916531e-02 1.03138529e-01 9.88976300e-01
5.26771545e-01 9.65506807e-02 6.01736844e-01 -5.36161542e-01
-8.95615995e-01 1.09658659e+00 2.91530564e-02 4.58162993e-01
6.22976184e-01 -2.90757269e-01 1.25696862e+00 -1.01827121e+00
5.03782511e-01 1.16613495e+00 3.84424359e-01 5.14728487e-01
-1.08319736e+00 -7.95764685e-01 -3.59889232e-02 6.99593350e-02
-1.64962077e+00 -1.09998249e-01 3.71738195e-01 -1.82149440e-01
1.42116392e+00 1.25308871e-01 8.98946464e-01 1.23161864e+00
5.02882302e-01 1.28630698e-01 1.06455398e+00 -6.63173735e-01
1.56186044e-01 4.53892767e-01 2.84721047e-01 3.64980072e-01
5.13709724e-01 4.55389880e-02 -7.64082909e-01 -1.39359310e-02
5.61655521e-01 -2.53297627e-01 -2.14134544e-01 4.79668438e-01
-1.07170820e+00 7.95496166e-01 4.77902651e-01 1.01730096e+00
-7.09557056e-01 -1.65283978e-02 2.01690421e-01 2.70988703e-01
1.16158031e-01 1.16722572e+00 -5.16902566e-01 -3.20834994e-01
-8.21189880e-01 2.04024032e-01 9.21364665e-01 5.76421976e-01
5.01241207e-01 3.63634154e-02 6.63986802e-02 9.41749394e-01
2.19938695e-01 5.89124024e-01 1.09870434e+00 -4.00122613e-01
2.69228876e-01 5.77860296e-01 1.41478077e-01 -1.24859381e+00
-5.47765493e-01 -7.71065891e-01 -6.41242027e-01 -4.36437607e-01
4.12492864e-02 1.03442753e-02 -5.04628062e-01 1.74255192e+00
-3.19976211e-01 -2.18459487e-01 3.23993832e-01 7.77395010e-01
1.07748306e+00 8.43893945e-01 3.41985494e-01 -2.79097557e-01
1.55553675e+00 -3.43292981e-01 -7.55694449e-01 -6.51325703e-01
5.34894824e-01 -8.41104329e-01 9.45037246e-01 4.43462163e-01
-8.07434857e-01 -7.91369796e-01 -1.32604420e+00 1.13885559e-01
-7.16091752e-01 -5.81258582e-03 3.20805460e-01 6.02486968e-01
-1.00305974e+00 6.58842385e-01 -2.33653903e-01 -8.30443084e-01
-3.70876163e-01 3.21871221e-01 -4.26797986e-01 2.80268818e-01
-1.60083735e+00 1.43719018e+00 8.69025290e-01 -1.79200977e-01
-4.08624887e-01 -2.19301596e-01 -6.97545826e-01 1.77640125e-01
-1.28591210e-01 -6.78433418e-01 8.83983076e-01 -1.23072398e+00
-1.00143075e+00 8.93242538e-01 -2.10541878e-02 -7.51963377e-01
-2.21542165e-01 -1.85820863e-01 -7.08287835e-01 -7.00420216e-02
3.43612731e-01 9.27659035e-01 6.82296932e-01 -1.06564689e+00
-7.87601829e-01 -2.11509824e-01 2.27695420e-01 2.35048890e-01
-5.71462452e-01 2.23407540e-02 -1.31479114e-01 -9.22782183e-01
3.15348571e-03 -7.54438102e-01 8.28676745e-02 -5.65756738e-01
-2.12782741e-01 -5.72521806e-01 2.51629859e-01 -7.53475964e-01
1.60273159e+00 -2.06930590e+00 6.91879615e-02 2.94532955e-01
2.07863837e-01 3.76614690e-01 -6.92393184e-02 8.24176550e-01
-2.72577703e-01 3.68469715e-01 4.40210663e-02 -1.62304193e-01
-2.95420717e-02 5.58482707e-01 -5.23907959e-01 1.46441683e-01
-4.81750220e-02 6.95174098e-01 -7.69460976e-01 -1.93542480e-01
-1.38970697e-03 4.91642565e-01 -1.56534314e-01 1.82178840e-02
-4.95069288e-03 -7.22542405e-02 -1.22210398e-01 2.66185045e-01
9.48705301e-02 -6.92978874e-02 2.58281380e-01 -6.07282668e-02
-3.86138521e-02 7.07087040e-01 -1.07700098e+00 1.18794930e+00
-7.10316777e-01 5.63485324e-01 -4.59686041e-01 -6.67820513e-01
1.32500625e+00 4.75698620e-01 -6.02910034e-02 -1.07372093e+00
9.26922038e-02 7.31785953e-01 3.51367354e-01 -1.86032161e-01
9.43013847e-01 -4.58010703e-01 -1.88746899e-01 5.48806369e-01
1.80850819e-01 9.71151441e-02 2.10191041e-01 1.05247125e-01
8.13129485e-01 -3.02640647e-01 9.66087699e-01 -4.13620979e-01
4.99779016e-01 -2.49134421e-01 2.87707895e-01 7.84993410e-01
1.35878280e-01 1.67141214e-01 4.05384079e-02 -4.09212381e-01
-7.61341035e-01 -7.20040262e-01 -1.04323946e-01 1.12184322e+00
-2.47282535e-02 -9.43436623e-01 -5.02810001e-01 -4.23177510e-01
-2.09651217e-01 1.41924977e+00 -4.63283777e-01 -2.46407717e-01
-4.74132836e-01 -4.02872056e-01 5.41868210e-01 4.72258627e-01
2.49376699e-01 -1.36654234e+00 -9.48050141e-01 3.92300963e-01
-3.72639298e-01 -7.33665764e-01 -2.70873666e-01 4.63595659e-01
-6.43083394e-01 -6.97647572e-01 -4.97161180e-01 -7.76980340e-01
4.61939603e-01 1.04341679e-03 1.24447882e+00 3.28921199e-01
3.16040933e-01 2.50225216e-01 -6.42710209e-01 -3.07341933e-01
-4.17795181e-01 2.59198118e-02 2.86826372e-01 -5.11445701e-01
6.90571725e-01 -2.79199511e-01 -1.24839708e-01 -4.24545929e-02
-8.12686324e-01 -2.11205229e-01 6.17844880e-01 8.35284531e-01
2.91787773e-01 1.71225533e-01 5.97353041e-01 -5.25006831e-01
1.21424389e+00 -4.48223501e-01 -1.24441862e-01 -1.44172907e-02
-9.78367090e-01 3.55546117e-01 7.62785912e-01 -3.23606014e-01
-4.29321498e-01 -3.69705409e-01 -5.82191885e-01 1.21843942e-01
-2.92204261e-01 8.35654199e-01 1.53506458e-01 9.65456590e-02
8.56060207e-01 3.74716550e-01 -4.28049535e-01 -4.50130850e-01
2.53907144e-01 6.39976740e-01 2.44450852e-01 -5.67829430e-01
3.05988163e-01 -2.03972384e-01 -1.67189687e-01 -9.67708409e-01
-7.04618454e-01 -3.71597499e-01 -5.45877159e-01 2.15813983e-02
6.71607316e-01 -5.15635014e-01 -6.12645149e-01 -2.00415105e-01
-1.44239509e+00 2.85623908e-01 -2.58159190e-01 5.02015650e-01
-2.58239776e-01 1.55602798e-01 -2.69504756e-01 -7.58870780e-01
-6.01389408e-01 -1.05186248e+00 7.39118218e-01 1.09261677e-01
-9.58806515e-01 -1.04222918e+00 -2.29245592e-02 -1.65218353e-01
4.35653329e-01 -4.25638109e-01 1.43932605e+00 -1.29435802e+00
3.10067385e-01 -2.43427783e-01 4.63590398e-02 1.94440514e-01
9.52373296e-02 -2.75440097e-01 -6.72342360e-01 -1.76761582e-01
1.09859079e-01 -5.65072987e-04 8.06261778e-01 1.51390344e-01
6.10882163e-01 -5.36559165e-01 -1.88545272e-01 5.89160547e-02
1.42710912e+00 5.29376984e-01 4.76245254e-01 4.51116413e-01
3.54118943e-01 6.38454556e-01 5.52170455e-01 3.61931145e-01
4.32649136e-01 9.54681516e-01 2.10149586e-01 3.49917442e-01
-4.65545896e-03 -4.11707669e-01 5.99505186e-01 8.16908836e-01
3.04964721e-01 -5.19469023e-01 -9.89003360e-01 6.64100289e-01
-1.32157934e+00 -7.88918674e-01 5.06994538e-02 2.51290774e+00
7.79261768e-01 3.29420179e-01 4.43922654e-02 3.77639651e-01
5.16036034e-01 6.01556078e-02 3.38586122e-02 -9.25024569e-01
-1.67552516e-01 6.72386765e-01 1.77562431e-01 7.67223656e-01
-7.46870697e-01 1.10673559e+00 6.16723061e+00 9.13244963e-01
-1.23878753e+00 -6.87678307e-02 3.58299494e-01 2.60643810e-01
-3.68656516e-01 -8.90684426e-02 -9.82146919e-01 5.08161545e-01
1.35608435e+00 -4.26455528e-01 4.10423309e-01 5.41481912e-01
7.45548680e-02 -1.69852778e-01 -1.13335812e+00 1.15687203e+00
5.74297071e-01 -9.38302100e-01 6.90922260e-01 -1.50241265e-02
2.22014207e-02 -4.44834054e-01 6.35973141e-02 2.80387580e-01
1.21489950e-02 -1.31570506e+00 8.70983422e-01 4.72749501e-01
4.60306287e-01 -9.64005589e-01 9.79405522e-01 5.20122409e-01
-1.00360644e+00 -8.67067799e-02 -4.74253446e-01 -1.94173515e-01
-5.34298792e-02 2.21387535e-01 -1.16290879e+00 2.34758601e-01
4.14166540e-01 6.66607797e-01 -8.26851606e-01 7.37987816e-01
-5.78532040e-01 5.74794173e-01 -2.88266271e-01 -5.23969173e-01
3.91953260e-01 3.34156156e-02 7.51944602e-01 1.31092560e+00
6.48709476e-01 1.37817174e-01 8.82887915e-02 8.16600800e-01
9.39625949e-02 6.92705929e-01 -6.82550371e-01 -3.46935958e-01
8.24220181e-01 9.80960310e-01 -6.11290216e-01 -3.01506430e-01
-1.33020505e-01 1.04691219e+00 3.54631513e-01 9.79404971e-02
-2.35391870e-01 -1.69894814e-01 3.13006759e-01 1.32783294e-01
7.99657851e-02 -8.08800906e-02 -4.07849103e-01 -4.30632740e-01
-2.10001692e-01 -8.91040742e-01 4.57116932e-01 -8.74058723e-01
-1.43210542e+00 9.63425815e-01 -1.06335469e-01 -8.95286739e-01
-5.10072768e-01 -8.00103128e-01 -2.38783702e-01 1.10807097e+00
-9.20339108e-01 -8.28304231e-01 1.99139059e-01 2.27559835e-01
5.41971385e-01 -3.25227767e-01 1.45031583e+00 2.36340210e-01
-2.26526767e-01 4.56710279e-01 -3.83821040e-01 -6.09756000e-02
8.37609112e-01 -1.42055011e+00 2.27598041e-01 7.04412103e-01
5.47375858e-01 1.16129243e+00 1.14630687e+00 -8.54317009e-01
-8.40822101e-01 -6.33617878e-01 1.78918993e+00 -2.63087958e-01
5.70460200e-01 1.58024326e-01 -8.25124383e-01 5.10210752e-01
4.86036628e-01 -9.29747462e-01 9.43644941e-01 1.29242271e-01
-1.59133628e-01 -8.76548216e-02 -9.80232775e-01 6.02969527e-01
4.68410581e-01 -3.46413463e-01 -1.15278125e+00 1.98653877e-01
6.28055513e-01 2.10672468e-02 -6.88166380e-01 -1.03073508e-01
4.90184069e-01 -7.72493720e-01 7.98196852e-01 -4.33650702e-01
4.86867547e-01 -1.16730236e-01 -4.64423746e-01 -1.75087559e+00
-3.98000300e-01 -1.64500639e-01 3.29822712e-02 1.19899821e+00
6.36370063e-01 -6.68174028e-01 8.53680819e-02 3.02419275e-01
-7.76147917e-02 -6.26479924e-01 -1.12711692e+00 -6.19124889e-01
8.42286050e-02 -4.74866152e-01 6.25554383e-01 9.44279552e-01
1.59200996e-01 7.57468045e-01 -2.66019106e-01 -2.63200074e-01
-3.93230654e-02 -6.61628507e-03 1.71623334e-01 -1.42664921e+00
-2.24561960e-01 -6.70078099e-01 -5.03536642e-01 -1.11431491e+00
3.03184092e-01 -1.09008777e+00 -8.38853195e-02 -1.63546085e+00
-8.59640017e-02 -3.36385876e-01 -3.59725416e-01 6.13499939e-01
-9.37033817e-02 7.59518445e-02 2.70842344e-01 1.14180006e-01
-8.69897380e-02 2.64295608e-01 6.93605423e-01 -5.43320812e-02
-2.12346882e-01 -1.23865657e-01 -1.14231253e+00 6.42155766e-01
7.88119435e-01 -6.22982383e-01 -1.44051746e-01 -3.28643888e-01
7.62882411e-01 -2.53307205e-02 3.50488007e-01 -9.90129590e-01
2.10539982e-01 3.95641811e-02 4.71372873e-01 -3.55440229e-01
5.76373816e-01 -1.08999288e+00 1.40736520e-01 7.95952439e-01
-4.25237298e-01 8.66258681e-01 2.90593863e-01 1.27990052e-01
-1.79531619e-01 -6.62902355e-01 6.28800035e-01 -1.96988806e-01
-9.49791729e-01 -3.56629401e-01 -6.34104192e-01 -2.15881914e-01
5.70119083e-01 -1.37840986e-01 -2.37861902e-01 -5.94282150e-01
-5.30311584e-01 -2.27183938e-01 1.92576259e-01 6.57860696e-01
9.53545690e-01 -1.28110695e+00 -7.02401936e-01 1.79591939e-01
3.23579967e-01 -9.53351021e-01 -3.77990186e-01 4.14841503e-01
-4.68749851e-01 9.12348926e-01 -2.88142830e-01 6.10439032e-02
-1.19919014e+00 5.51398456e-01 3.76998991e-01 -1.08693227e-01
-5.09705603e-01 8.97805393e-01 -1.48006231e-01 -1.28618270e-01
1.10530384e-01 -2.96308279e-01 -8.28735650e-01 4.55155015e-01
6.45800471e-01 1.57612935e-01 1.64044350e-01 -1.30365551e+00
-4.20057654e-01 3.52802426e-01 -9.01697483e-03 -4.24054384e-01
9.68794048e-01 2.65912861e-02 -4.73248214e-01 8.13195109e-01
5.54995179e-01 2.83013821e-01 1.09451108e-01 -1.85763717e-01
7.44716227e-02 -3.66639674e-01 -4.98062000e-03 -1.17061424e+00
-6.47189856e-01 8.27784717e-01 6.76791489e-01 2.56944925e-01
8.86733651e-01 -2.37515926e-01 7.30181694e-01 6.43237174e-01
5.37144780e-01 -1.01697242e+00 -8.95014927e-02 8.54981661e-01
1.07275665e+00 -7.93436944e-01 -2.24479496e-01 3.50755523e-03
-6.91406727e-01 1.39621663e+00 6.05541885e-01 -1.67384911e-02
5.54684401e-01 -1.20837554e-01 3.19295935e-02 -3.38819593e-01
-7.06347764e-01 -9.45138633e-02 4.39381063e-01 5.67535877e-01
8.01851869e-01 4.04094428e-01 -1.17224085e+00 1.07043815e+00
-9.18479085e-01 -5.63803792e-01 5.00913739e-01 3.78099740e-01
-7.85039246e-01 -1.11572456e+00 -3.35662246e-01 5.45946479e-01
-4.03587580e-01 -5.41502595e-01 -8.21083009e-01 5.74736059e-01
3.23857695e-01 1.16835320e+00 1.39814273e-01 -4.94959980e-01
3.08766454e-01 2.23222166e-01 1.64357290e-01 -9.48567033e-01
-1.02295518e+00 -9.65480432e-02 1.43680885e-01 -2.48122275e-01
-1.57241687e-01 -3.04531872e-01 -1.36769938e+00 -3.61001372e-01
-1.98983982e-01 4.64789778e-01 5.40689349e-01 1.27476406e+00
5.96960746e-02 3.47263247e-01 1.07523665e-01 -4.28821862e-01
-3.27647120e-01 -1.11424696e+00 -6.31369233e-01 1.35604516e-01
-9.36422944e-02 -7.21047640e-01 7.33009577e-02 -3.33802104e-01] | [10.498750686645508, 9.4298734664917] |
b18cb36f-6620-4bee-b7ad-6b7924b09f08 | brightness-restricted-adversarial-attack | 2307.00421 | null | https://arxiv.org/abs/2307.00421v1 | https://arxiv.org/pdf/2307.00421v1.pdf | Brightness-Restricted Adversarial Attack Patch | Adversarial attack patches have gained increasing attention due to their practical applicability in physical-world scenarios. However, the bright colors used in attack patches represent a significant drawback, as they can be easily identified by human observers. Moreover, even though these attacks have been highly successful in deceiving target networks, which specific features of the attack patch contribute to its success are still unknown. Our paper introduces a brightness-restricted patch (BrPatch) that uses optical characteristics to effectively reduce conspicuousness while preserving image independence. We also conducted an analysis of the impact of various image features (such as color, texture, noise, and size) on the effectiveness of an attack patch in physical-world deployment. Our experiments show that attack patches exhibit strong redundancy to brightness and are resistant to color transfer and noise. Based on our findings, we propose some additional methods to further reduce the conspicuousness of BrPatch. Our findings also explain the robustness of attack patches observed in physical-world scenarios. | ['Mingzhen Shao'] | 2023-07-01 | null | null | null | null | ['adversarial-attack'] | ['adversarial'] | [ 2.13123053e-01 -2.35775933e-01 1.56804025e-01 2.00677574e-01
-1.93449900e-01 -1.10174131e+00 6.26975179e-01 -4.48629260e-02
-2.45018497e-01 5.59000671e-01 -1.93790615e-01 -4.80544746e-01
-1.17061339e-01 -8.35541129e-01 -5.20522892e-01 -9.41978574e-01
-3.40080798e-01 -6.34024739e-01 6.02251112e-01 -3.26477081e-01
4.52526152e-01 1.07853186e+00 -1.00902593e+00 -6.58157021e-02
6.78322315e-01 7.90971816e-01 -2.04552501e-01 4.88279819e-01
3.72021556e-01 7.10024416e-01 -1.19116330e+00 -6.14162147e-01
7.76466191e-01 -3.61918241e-01 -1.17002554e-01 2.81831603e-02
3.54645193e-01 -7.33686924e-01 -7.24369884e-01 9.98272479e-01
5.72475076e-01 -1.44271359e-01 5.08822322e-01 -1.31615984e+00
-5.90079248e-01 4.71972264e-02 -8.48161936e-01 2.38675505e-01
2.20047578e-01 3.80699307e-01 4.83720392e-01 -2.91335434e-01
3.25681001e-01 1.06011474e+00 5.07308245e-01 4.78218734e-01
-1.03377450e+00 -9.66759980e-01 -2.55115796e-03 1.73377246e-01
-1.41740978e+00 -4.46091682e-01 9.49182630e-01 3.80857326e-02
3.68140072e-01 5.37456214e-01 3.54941070e-01 1.00750136e+00
5.54641664e-01 1.16110809e-01 1.31065500e+00 -3.67181718e-01
1.96808740e-01 3.10828447e-01 -5.64732432e-01 5.20454466e-01
5.66401780e-01 5.09310246e-01 -1.59274638e-01 -4.52807516e-01
9.00611281e-01 -4.55932654e-02 -4.52299386e-01 -4.70122218e-01
-6.31325424e-01 6.97314382e-01 8.26594770e-01 2.26531103e-02
-2.11501658e-01 2.01080933e-01 3.61939184e-02 1.76543728e-01
2.39415154e-01 7.64694393e-01 -2.86288373e-02 5.40109202e-02
-3.14178109e-01 -1.74152717e-01 5.24368584e-01 4.06725824e-01
5.30459702e-01 3.88508409e-01 7.71521628e-02 6.28850579e-01
-4.83790897e-02 9.06741798e-01 -1.01174727e-01 -7.79366195e-01
2.28239819e-01 2.79876143e-01 1.58026159e-01 -1.54553747e+00
-1.29020080e-01 -3.38818192e-01 -6.13147974e-01 7.18592286e-01
4.05273527e-01 -4.26671594e-01 -5.98472297e-01 1.57623386e+00
1.26149788e-01 1.12237468e-01 2.44875178e-02 8.49505305e-01
2.70104110e-01 4.44335818e-01 1.31782934e-01 -8.80615227e-03
9.54511642e-01 -4.61163193e-01 -3.53672653e-01 -1.83644429e-01
1.33953616e-01 -8.82732749e-01 8.78799975e-01 3.09660554e-01
-5.57457387e-01 -2.17865050e-01 -1.14145541e+00 9.75052178e-01
-4.18868870e-01 -1.11975327e-01 3.59300405e-01 1.37906420e+00
-7.93184221e-01 4.26490664e-01 -3.22996676e-01 -4.24553633e-01
3.45241576e-01 2.26541474e-01 -4.09685671e-01 7.92514309e-02
-1.00377178e+00 9.72491264e-01 -9.17109996e-02 8.12061504e-02
-9.67184424e-01 -3.52564067e-01 -3.95891607e-01 1.53650912e-02
5.44902742e-01 1.00001860e-02 5.41213989e-01 -1.01471043e+00
-1.54254520e+00 3.28355849e-01 3.99612784e-01 -3.73564035e-01
4.59450871e-01 5.43262102e-02 -7.56550252e-01 5.47093511e-01
-3.56035471e-01 2.90812671e-01 1.35246348e+00 -1.73979998e+00
-1.98787943e-01 -1.27824813e-01 5.52273333e-01 -9.13989022e-02
-8.14468205e-01 1.36007160e-01 2.48933006e-02 -9.00895417e-01
-1.93127543e-01 -1.06774592e+00 -1.44713223e-01 1.52859628e-01
-3.96875232e-01 5.55350602e-01 1.29812145e+00 -4.41346675e-01
9.44163263e-01 -2.40442467e+00 -4.13995892e-01 4.00688529e-01
2.35212311e-01 7.48840094e-01 -4.18273360e-01 7.84588814e-01
-5.01304539e-03 4.78465855e-01 -4.96047065e-02 4.66398388e-01
-3.33348811e-01 1.48788868e-02 -5.69117129e-01 7.94304013e-01
1.69870660e-01 4.57631648e-01 -5.74614406e-01 -1.78092912e-01
3.18542898e-01 4.70515519e-01 -2.09659219e-01 1.35673448e-01
2.68381089e-01 2.90160269e-01 -5.89239657e-01 6.60481334e-01
1.14464581e+00 4.16746616e-01 9.74429324e-02 -2.01870009e-01
-1.21040650e-01 -3.24054509e-01 -7.07116663e-01 4.83592272e-01
-2.55382091e-01 7.13931799e-01 1.81162700e-01 -6.57628655e-01
9.82771754e-01 9.66446623e-02 2.66822547e-01 -8.68956983e-01
9.64485183e-02 -1.60865158e-01 2.85514027e-01 -3.32282603e-01
4.51939315e-01 -1.37390301e-01 -1.76303685e-02 5.02874374e-01
-6.38926983e-01 -2.20089167e-01 -1.75237134e-01 3.47212285e-01
1.28418756e+00 -2.76605457e-01 1.01739075e-02 -1.88491136e-01
3.02280068e-01 -1.50670186e-01 4.71417069e-01 8.71664286e-01
-5.60928941e-01 4.77013171e-01 5.45541227e-01 -4.84701812e-01
-9.35265243e-01 -1.13432801e+00 8.78484547e-02 6.71768188e-01
8.71058464e-01 -2.64740944e-01 -6.21859670e-01 -9.36198354e-01
1.52805140e-02 4.27537084e-01 -5.76560616e-01 -5.03108561e-01
-3.93691331e-01 -7.09521830e-01 1.07789087e+00 3.06140542e-01
7.30817020e-01 -7.76222408e-01 -8.36797953e-01 -2.16210529e-01
2.64329523e-01 -1.17198646e+00 -3.06669891e-01 -2.08530173e-01
-4.38917458e-01 -1.28799152e+00 -4.34318423e-01 -7.56190568e-02
1.00908446e+00 8.86130989e-01 6.87023044e-01 4.89856243e-01
-6.13003016e-01 6.53725088e-01 -6.97710514e-01 -4.91112649e-01
-4.04786527e-01 -3.68045181e-01 4.99938279e-02 2.92783797e-01
-3.81176591e-01 -4.13921416e-01 -7.37038374e-01 7.44738102e-01
-1.12106264e+00 -6.93664551e-01 7.71152973e-01 4.31816906e-01
-7.80941024e-02 6.53131068e-01 2.98917770e-01 -6.43413961e-01
7.61020660e-01 -1.78016067e-01 -6.09218717e-01 3.38029712e-01
-1.33160979e-01 -2.79981494e-01 8.40502083e-01 -5.95878184e-01
-1.13475406e+00 -4.34337854e-01 2.66802847e-01 -2.12008446e-01
-2.53579348e-01 -7.61033781e-03 -1.47463337e-01 -1.04919446e+00
6.96873367e-01 -9.91655737e-02 1.10084802e-01 -4.33454476e-02
1.38242811e-01 2.22713187e-01 8.83691236e-02 -4.67066824e-01
1.61455822e+00 8.50480676e-01 2.84071445e-01 -1.18733156e+00
-2.95712501e-01 1.58607528e-01 -1.28552899e-01 -3.70117992e-01
4.43712026e-01 -4.58244026e-01 -7.31297016e-01 6.69522643e-01
-8.97501409e-01 -2.63170928e-01 1.26693875e-01 2.09692568e-01
2.57373035e-01 7.61750519e-01 -4.60777700e-01 -9.14524078e-01
-3.07509080e-02 -8.43701243e-01 4.62123781e-01 4.24077868e-01
1.77634731e-01 -8.82673740e-01 -1.81412086e-01 1.79535836e-01
8.28937769e-01 5.36058187e-01 9.51411128e-01 -2.04128757e-01
-6.83581352e-01 -5.03368616e-01 -3.96265000e-01 3.33295673e-01
3.90839070e-01 4.57610250e-01 -8.38072956e-01 -6.99032962e-01
-8.76958445e-02 -1.45988464e-01 4.74310488e-01 2.00116366e-01
9.80884671e-01 -2.77301371e-01 -3.33069831e-01 5.94879866e-01
1.34357083e+00 3.99773687e-01 9.77306247e-01 4.27644193e-01
3.87510419e-01 5.71460485e-01 6.00396633e-01 4.57539320e-01
-4.51394022e-01 7.39063680e-01 9.62561965e-01 -6.15531564e-01
-1.60893023e-01 -4.05178741e-02 5.52177310e-01 -9.11927894e-02
-1.16404794e-01 -4.78525400e-01 -5.15789568e-01 8.78925547e-02
-1.05120862e+00 -9.24462557e-01 1.47270411e-01 2.33674216e+00
9.47968066e-02 3.06421816e-01 2.23592982e-01 5.18047661e-02
8.62714589e-01 4.52785015e-01 -2.41240799e-01 -3.59564692e-01
-4.31175232e-01 -6.21338189e-02 1.10889733e+00 1.62461922e-01
-1.01548672e+00 7.11133182e-01 7.02985573e+00 9.62508261e-01
-1.45223641e+00 -2.39447206e-01 5.00561833e-01 6.85912967e-02
-1.60946012e-01 2.47460365e-01 -3.76675636e-01 4.85921979e-01
3.99466902e-01 -2.14288291e-02 3.73943806e-01 4.73501325e-01
2.46567115e-01 -3.09183836e-01 -3.40508878e-01 4.35118705e-01
3.04706186e-01 -9.38871443e-01 1.40519112e-01 3.10715824e-01
3.92014474e-01 -4.57880229e-01 6.73902750e-01 -1.54694498e-01
2.84039497e-01 -8.66387784e-01 5.31947970e-01 -1.36464506e-01
6.91765130e-01 -9.58357334e-01 6.23214424e-01 -4.03332636e-02
-9.71791863e-01 -3.45167309e-01 -4.74825591e-01 -1.33766323e-01
-4.37819436e-02 3.71080935e-01 -7.15611160e-01 4.88344163e-01
5.59901953e-01 5.38062863e-02 -9.82290447e-01 1.03100669e+00
-4.99606490e-01 5.27965546e-01 -1.87458053e-01 1.79317966e-02
2.56341189e-01 1.90737434e-02 7.79534340e-01 7.57584453e-01
1.98777199e-01 2.69828327e-02 1.59916013e-01 7.10834265e-01
2.43659690e-01 -2.43821308e-01 -9.09238935e-01 1.64665245e-02
7.01802135e-01 1.27024186e+00 -1.01261294e+00 3.13072234e-01
-3.36223006e-01 8.07340205e-01 -1.17476143e-01 6.28498077e-01
-1.04541266e+00 -6.53270781e-01 8.01842093e-01 3.86996984e-01
3.00495982e-01 -2.65253812e-01 -6.64571002e-02 -8.76438439e-01
-7.26626739e-02 -9.62829649e-01 1.22435965e-01 -5.41092813e-01
-1.18586791e+00 7.24251449e-01 3.83302830e-02 -1.29721940e+00
4.95219946e-01 -5.45735061e-01 -9.37386990e-01 4.22877401e-01
-1.26031935e+00 -1.16996503e+00 -3.84348452e-01 6.03202581e-01
9.76639912e-02 -3.91490489e-01 6.76248193e-01 8.78382996e-02
-7.30448604e-01 7.08164871e-01 8.26738924e-02 3.53620827e-01
8.16926301e-01 -7.44387746e-01 3.10328662e-01 1.29255223e+00
9.86286346e-03 5.57424128e-01 7.30397463e-01 -6.64393008e-01
-1.47443235e+00 -6.97044194e-01 -2.75879264e-01 -3.26586515e-01
6.25273347e-01 -2.50550002e-01 -5.03362596e-01 1.44137263e-01
3.54255319e-01 1.87748875e-02 4.68563795e-01 -2.63836950e-01
-6.14648283e-01 -3.56549859e-01 -1.43762302e+00 8.64748955e-01
5.10260463e-01 -4.40695226e-01 -3.50217372e-02 -1.11236818e-01
2.62888670e-01 6.34484366e-02 -3.80252242e-01 3.39217186e-01
4.67294902e-01 -1.20716095e+00 1.13439071e+00 -1.90786347e-01
5.27365543e-02 -4.39991087e-01 5.90926558e-02 -1.37075520e+00
-3.88884723e-01 -7.69528210e-01 4.46946084e-01 1.15551031e+00
2.28565618e-01 -9.79363143e-01 7.43142009e-01 4.19005364e-01
4.04619545e-01 -1.76940992e-01 -8.06586027e-01 -1.07850051e+00
-6.14219345e-02 2.59496689e-01 4.07779723e-01 7.82964408e-01
-8.72862339e-02 -1.62570238e-01 -7.46960759e-01 5.06838202e-01
7.69229472e-01 -1.42361090e-01 9.11655724e-01 -9.61345732e-01
-2.65840709e-01 -1.89442083e-01 -5.96411288e-01 -3.90526533e-01
5.02183326e-02 -9.50961262e-02 -2.01622501e-01 -7.33756661e-01
8.39917436e-02 -5.72592199e-01 -2.80471653e-01 4.42837596e-01
-2.32855871e-01 7.64223635e-01 5.59858620e-01 2.59460956e-01
-2.36778259e-01 3.00013512e-01 9.45381522e-01 -1.62151515e-01
2.96522509e-02 -5.13996892e-02 -6.14440680e-01 5.47389567e-01
9.92628336e-01 -5.24085701e-01 -3.88287276e-01 -4.77182984e-01
-1.02271676e-01 -1.75763205e-01 5.72163641e-01 -1.08535469e+00
-1.86453238e-01 -4.55104113e-01 3.73535246e-01 -7.54679460e-03
4.03074473e-01 -1.06635725e+00 4.32879254e-02 8.11749101e-01
1.06902733e-01 1.97438762e-01 4.54269022e-01 6.36845350e-01
-1.03482166e-02 -1.92929715e-01 1.07202435e+00 3.94162461e-02
-5.67054212e-01 -2.70624496e-02 -7.19052315e-01 -3.17784786e-01
1.54121518e+00 -4.05151129e-01 -9.01992321e-01 -7.44746208e-01
7.55630061e-02 -4.39508855e-01 9.07943368e-01 2.83308029e-01
6.63508415e-01 -9.69564080e-01 -6.22763693e-01 3.78216743e-01
-7.00200275e-02 -1.05661237e+00 2.44082138e-01 6.01019144e-01
-8.78695667e-01 1.16422243e-01 -7.09043562e-01 -2.35876158e-01
-1.41673183e+00 7.53191710e-01 2.89511859e-01 1.17846832e-01
-4.77497816e-01 6.23032629e-01 4.53655452e-01 1.06178626e-01
1.30300447e-01 4.90808278e-01 9.06428546e-02 -5.55191576e-01
4.15246546e-01 4.18341398e-01 -9.32424515e-02 -4.72786576e-01
-5.10726750e-01 6.20008588e-01 -1.27866760e-01 7.69446120e-02
9.98154581e-01 -9.05606076e-02 -8.81585926e-02 -5.23847222e-01
7.42022395e-01 6.71298683e-01 -1.36114812e+00 1.23646386e-01
-3.36524248e-01 -1.10028672e+00 -1.36115760e-01 -6.75265610e-01
-1.48091829e+00 6.22471035e-01 5.28882921e-01 5.85690498e-01
1.29611456e+00 -3.43555003e-01 5.13807833e-01 2.30578512e-01
3.94824773e-01 -7.24683285e-01 5.46691060e-01 1.10719591e-01
5.80089033e-01 -7.99263954e-01 2.41399467e-01 -7.86597013e-01
-5.92089593e-01 9.87358391e-01 6.68896496e-01 -1.98186502e-01
3.56297761e-01 4.42167968e-01 3.73136729e-01 4.69834879e-02
-3.53883624e-01 3.41100059e-03 -1.73303619e-01 1.12483740e+00
-2.52597034e-02 -3.43866870e-02 -1.53368399e-01 -1.68819875e-01
2.21421540e-01 -8.18229795e-01 8.33889842e-01 1.01024449e+00
-2.98167288e-01 -1.22363484e+00 -8.81024003e-01 -1.87424943e-03
-6.07845724e-01 2.26708859e-01 -9.24793005e-01 9.09918547e-01
-4.61107902e-02 1.29775429e+00 -2.77610093e-01 -6.35183692e-01
3.28222275e-01 -6.55829370e-01 4.27977353e-01 -1.62599906e-01
-4.60716248e-01 -8.01623911e-02 -5.60122961e-03 -3.60569686e-01
-1.36730522e-01 -1.95087835e-01 -5.23017764e-01 -7.31713891e-01
-4.74928200e-01 2.06186801e-01 5.73230505e-01 5.83561718e-01
2.85393864e-01 3.67326200e-01 1.07959259e+00 -5.71248293e-01
-4.52851951e-01 -5.84569931e-01 -7.28522420e-01 4.98652399e-01
1.20956898e-01 -7.35321105e-01 -7.14620829e-01 -2.44338721e-01] | [5.408068656921387, 7.916049957275391] |
70001a7e-3e9e-4afe-b16c-2f46d6705f0f | crossmodal-learning-for-audio-visual-speech | 2003.04358 | null | https://arxiv.org/abs/2003.04358v2 | https://arxiv.org/pdf/2003.04358v2.pdf | Cross modal video representations for weakly supervised active speaker localization | An objective understanding of media depictions, such as inclusive portrayals of how much someone is heard and seen on screen such as in film and television, requires the machines to discern automatically who, when, how, and where someone is talking, and not. Speaker activity can be automatically discerned from the rich multimodal information present in the media content. This is however a challenging problem due to the vast variety and contextual variability in the media content, and the lack of labeled data. In this work, we present a cross-modal neural network for learning visual representations, which have implicit information pertaining to the spatial location of a speaker in the visual frames. Avoiding the need for manual annotations for active speakers in visual frames, acquiring of which is very expensive, we present a weakly supervised system for the task of localizing active speakers in movie content. We use the learned cross-modal visual representations, and provide weak supervision from movie subtitles acting as a proxy for voice activity, thus requiring no manual annotations. We evaluate the performance of the proposed system on the AVA active speaker dataset and demonstrate the effectiveness of the cross-modal embeddings for localizing active speakers in comparison to fully supervised systems. We also demonstrate state-of-the-art performance for the task of voice activity detection in an audio-visual framework, especially when speech is accompanied by noise and music. | ['Krishna Somandepalli', 'Rahul Sharma', 'Shrikanth Narayanan'] | 2020-03-09 | null | null | null | null | ['active-speaker-localization'] | ['audio'] | [ 1.50424257e-01 -1.61774491e-03 -2.41867602e-01 -3.66882771e-01
-9.54548419e-01 -8.59148145e-01 8.78719151e-01 2.45665357e-01
-5.45185864e-01 2.20626801e-01 6.37313068e-01 4.83700773e-03
2.39443511e-01 -3.54456693e-01 -6.64755046e-01 -7.36150980e-01
-7.76239485e-02 2.27125600e-01 4.49376404e-02 1.00474127e-01
-1.19654620e-02 3.87635112e-01 -1.78247917e+00 4.25629824e-01
1.32242218e-01 1.10914052e+00 1.56800091e-01 8.42842817e-01
-2.92130381e-01 9.88122106e-01 -8.23494613e-01 -5.50144166e-02
-3.60964149e-01 -4.79267329e-01 -6.03361249e-01 4.88317281e-01
8.44262064e-01 -2.65277892e-01 -2.04339951e-01 8.87515724e-01
4.72645611e-01 1.04039505e-01 6.58357143e-01 -1.06296778e+00
-4.12249744e-01 4.02686447e-01 -4.37436730e-01 4.90452439e-01
6.22319937e-01 2.57493276e-02 1.20428157e+00 -1.08631194e+00
5.18645644e-01 1.21747696e+00 1.82291105e-01 4.04753536e-01
-1.15045166e+00 -3.78858119e-01 3.26030940e-01 1.81154251e-01
-1.55791771e+00 -1.05808270e+00 1.01814282e+00 -7.73124635e-01
5.69763839e-01 4.12277013e-01 5.49835563e-01 1.38771653e+00
-5.70737660e-01 8.70351791e-01 7.50653028e-01 -5.55779696e-01
2.06489399e-01 4.55859572e-01 1.19706981e-01 7.71139085e-01
-5.53647161e-01 -2.45395467e-01 -9.28276002e-01 -8.89340267e-02
5.66471338e-01 -1.61346436e-01 -4.62132126e-01 -4.58380640e-01
-1.34433675e+00 7.50013947e-01 3.18786442e-01 4.56698358e-01
-2.29443774e-01 4.48825993e-02 4.37461227e-01 -2.57621575e-02
7.00195432e-01 1.27205446e-01 1.61111057e-02 -3.38841468e-01
-1.05128205e+00 -2.58311033e-01 5.99176466e-01 3.88869017e-01
5.58070004e-01 2.39626035e-01 -9.46943536e-02 1.08604527e+00
3.80499393e-01 5.33229947e-01 4.03607219e-01 -9.24553037e-01
4.69136596e-01 5.06652176e-01 1.69784218e-01 -1.05112708e+00
-2.58707911e-01 -1.65798128e-01 -4.24415380e-01 1.48794070e-01
3.88840079e-01 -7.66920075e-02 -7.24420011e-01 1.85272813e+00
4.09123808e-01 2.25359589e-01 -1.46478862e-01 1.03582096e+00
1.07027245e+00 8.70638609e-01 1.11424424e-01 -4.15282577e-01
1.47890282e+00 -8.19176137e-01 -1.08664632e+00 -2.61816084e-01
2.85104126e-01 -5.40308893e-01 1.19267023e+00 8.55342969e-02
-9.26629007e-01 -5.63693225e-01 -1.04058254e+00 -7.10982829e-02
-4.26040381e-01 3.01453769e-01 2.34212339e-01 4.44591850e-01
-8.04971874e-01 1.56303141e-02 -8.60713661e-01 -3.77334177e-01
3.97160113e-01 2.38239672e-03 -6.32894576e-01 2.87433416e-01
-9.55756724e-01 5.70991457e-01 -9.04114097e-02 2.26058438e-01
-1.18316877e+00 -5.05167484e-01 -1.20528769e+00 8.72133896e-02
4.50283647e-01 -1.67259574e-02 1.30798447e+00 -1.32872176e+00
-1.42220891e+00 1.01196337e+00 -3.69490087e-01 -1.68053105e-01
4.81694639e-01 -1.11298509e-01 -7.12692618e-01 5.07582068e-01
-6.23885579e-02 6.02756858e-01 1.10911489e+00 -1.18328583e+00
-5.86301506e-01 -4.70438212e-01 3.61403853e-01 3.78389955e-01
-5.35493374e-01 2.21755579e-01 -6.54884040e-01 -4.66691971e-01
-1.86851565e-02 -7.25725412e-01 4.22052950e-01 3.41271430e-01
-3.04767370e-01 -3.17011148e-01 1.13355517e+00 -6.66195393e-01
1.13882709e+00 -2.66401005e+00 2.04894632e-01 -7.58406743e-02
1.79051623e-01 1.74411442e-02 -5.10766134e-02 2.94972330e-01
1.11677349e-01 -1.44094881e-02 -2.94891205e-02 -5.51372111e-01
6.48721159e-02 1.64038330e-01 -2.71881193e-01 7.40944028e-01
1.50720492e-01 5.59420705e-01 -8.03921461e-01 -6.07923090e-01
2.08413497e-01 7.87163079e-01 -1.66509792e-01 4.52412248e-01
-1.17467001e-01 5.15164614e-01 -6.32882118e-02 6.11225426e-01
9.80063379e-02 -1.75062060e-01 1.66821361e-01 -3.33999485e-01
-2.28298992e-01 4.02421296e-01 -1.13563871e+00 1.75976348e+00
-6.59029722e-01 1.16931665e+00 5.21924675e-01 -9.41100180e-01
5.46251178e-01 7.77011275e-01 1.83644831e-01 -7.05671012e-01
3.92206386e-02 -1.50034681e-01 -9.25813541e-02 -7.65473068e-01
2.72814661e-01 -2.93615535e-02 -2.52178490e-01 4.27133769e-01
1.24850407e-01 2.54487157e-01 1.23655610e-01 2.21050173e-01
8.15811574e-01 -1.68186441e-01 1.67895243e-01 7.43572265e-02
5.59569538e-01 -3.99137020e-01 2.63362646e-01 4.13352340e-01
-3.25881153e-01 5.61389804e-01 4.78774130e-01 -2.54364014e-01
-7.20253408e-01 -1.16294587e+00 -1.31264612e-01 1.67566264e+00
3.36327171e-03 -4.54961181e-01 -5.75988531e-01 -5.89674115e-01
-3.41489196e-01 3.71376485e-01 -7.32168972e-01 2.40156248e-01
-5.75138927e-01 -3.74273896e-01 4.20802027e-01 5.07735431e-01
8.83918330e-02 -9.94239748e-01 -5.31736135e-01 -1.17633805e-01
-3.44921499e-01 -1.42311656e+00 -6.51215136e-01 6.53044879e-02
-4.20407206e-01 -9.00289118e-01 -6.52882457e-01 -8.43794107e-01
5.93465626e-01 2.15838999e-01 1.02315950e+00 -3.17684263e-01
-2.71081477e-01 7.03391790e-01 -8.69789943e-02 -2.21437365e-01
-3.15005451e-01 -2.70131826e-01 -1.89091600e-02 6.53353512e-01
1.54010966e-01 -4.66318041e-01 -6.04786158e-01 3.18230897e-01
-6.82944834e-01 -9.86884162e-02 2.00472236e-01 5.44251740e-01
4.63706583e-01 -3.73647928e-01 3.06840360e-01 -7.27307141e-01
4.86430258e-01 -3.98073226e-01 -2.40484193e-01 4.79867086e-02
2.01370254e-01 -2.12136954e-01 1.93577811e-01 -7.27313995e-01
-8.03558767e-01 2.99782485e-01 1.22373793e-02 -5.73967159e-01
-4.09620225e-01 3.61613929e-01 -2.85292089e-01 1.86090335e-01
6.16105437e-01 3.69740613e-02 -2.58872043e-02 -5.46771407e-01
5.06169558e-01 9.88210857e-01 6.81353033e-01 -1.69860199e-01
4.57099080e-01 6.64033294e-01 -4.16172743e-01 -1.41719818e+00
-9.56671059e-01 -7.12997556e-01 -5.42027175e-01 -4.24910069e-01
1.08910477e+00 -1.14297032e+00 -6.68877423e-01 1.66021779e-01
-1.09014750e+00 -1.27052411e-01 -1.42452106e-01 4.33604181e-01
-4.44769472e-01 3.44904125e-01 -4.46839005e-01 -1.01337266e+00
4.83739525e-02 -1.07266462e+00 1.12788427e+00 -1.23392791e-01
-4.59813207e-01 -1.06557620e+00 1.10367537e-01 6.49871886e-01
1.49502635e-01 1.93980902e-01 7.52179086e-01 -6.29713953e-01
-3.92082423e-01 -2.84225672e-01 5.63666821e-02 3.18651229e-01
2.73356467e-01 8.30483623e-03 -1.54027128e+00 -1.67716444e-01
-2.30584145e-01 -4.19209957e-01 5.85187674e-01 2.43879840e-01
1.10285699e+00 -5.77311933e-01 -1.21207952e-01 3.26559484e-01
8.33490729e-01 5.28756529e-02 1.88010842e-01 -1.90083414e-01
8.33101928e-01 8.98334920e-01 3.69200557e-01 3.27694088e-01
1.80026338e-01 1.05118036e+00 5.60347378e-01 -2.97371566e-01
-1.99913755e-01 -2.31035337e-01 5.56315303e-01 8.57045293e-01
6.21491745e-02 -3.73418778e-01 -7.33010828e-01 7.46098757e-01
-1.63274586e+00 -1.15043795e+00 2.35653013e-01 2.07310724e+00
8.25346231e-01 -9.79519263e-02 3.20005685e-01 3.58940035e-01
8.65907907e-01 6.12358868e-01 -3.54758948e-01 -3.14108580e-01
-3.07063642e-03 -2.30531186e-01 -7.28220344e-02 6.40399516e-01
-1.46236277e+00 5.43019652e-01 6.07704306e+00 6.17398083e-01
-1.45670509e+00 3.84997368e-01 3.55481297e-01 -4.90257204e-01
-2.90212762e-02 -4.38028753e-01 -5.07439911e-01 4.94661540e-01
1.07278836e+00 3.65891397e-01 3.78975123e-01 6.93804204e-01
4.42176044e-01 1.19605988e-01 -1.39932680e+00 1.43107200e+00
6.00300670e-01 -1.14938438e+00 -1.77250266e-01 3.99803855e-02
2.47059822e-01 -1.87400788e-01 9.89780724e-02 1.15301833e-01
-2.03620031e-01 -9.83967483e-01 1.14504516e+00 2.85777271e-01
7.41900563e-01 -5.43594599e-01 2.63435751e-01 3.67932051e-01
-1.23758245e+00 -9.89582762e-02 2.53037453e-01 1.39339685e-01
2.50288457e-01 1.72615081e-01 -9.58841383e-01 -1.95052788e-01
7.11524367e-01 6.07988358e-01 -4.38005060e-01 6.29280508e-01
-3.08636516e-01 6.53014600e-01 -2.45408028e-01 1.26488507e-01
7.03458637e-02 1.62496671e-01 6.43405199e-01 1.41855550e+00
7.50427917e-02 -2.68744588e-01 2.22933993e-01 6.69527709e-01
-3.71146649e-01 2.05681130e-01 -6.95257485e-01 -4.36432451e-01
4.32527125e-01 1.34020019e+00 -4.62465197e-01 -2.13595614e-01
-5.34059465e-01 9.52439129e-01 3.48101109e-01 4.69472677e-01
-7.22764313e-01 -1.04098126e-01 6.30353630e-01 4.30394083e-01
3.18082690e-01 -3.06765854e-01 3.39901805e-01 -9.28343832e-01
1.72652811e-01 -8.38666379e-01 3.22612792e-01 -8.17459226e-01
-1.04254353e+00 6.29334152e-01 -1.35102928e-01 -1.29493964e+00
-4.75493819e-01 -4.94175941e-01 -5.48709452e-01 6.20833158e-01
-1.14827287e+00 -1.13825357e+00 -3.47322404e-01 6.03662670e-01
7.14144528e-01 -1.35285869e-01 8.05663586e-01 4.76903319e-01
-4.47848439e-01 5.54252207e-01 -2.22840190e-01 5.10228753e-01
7.38144696e-01 -1.12735546e+00 -1.74856141e-01 5.17161250e-01
8.81625235e-01 3.93389642e-01 7.36198008e-01 3.96628939e-02
-1.39493239e+00 -8.49097133e-01 7.82072902e-01 -5.47471046e-01
7.56912172e-01 -8.56940925e-01 -8.49789500e-01 5.71808755e-01
2.78468877e-01 3.31411719e-01 9.68711257e-01 3.96098793e-01
-4.90201175e-01 -1.72555655e-01 -8.02156627e-01 3.75472724e-01
7.33097017e-01 -1.23454380e+00 -6.25055909e-01 2.35291973e-01
5.01527131e-01 -2.00609669e-01 -4.84322667e-01 6.02458119e-02
5.24330854e-01 -6.95756793e-01 1.05104339e+00 -5.63827932e-01
1.56473845e-01 -3.80891412e-01 -3.08452457e-01 -1.09133959e+00
5.20198792e-02 -2.90717125e-01 -1.77520081e-01 1.55441260e+00
4.83378500e-01 -9.17156413e-02 4.42749888e-01 2.05604047e-01
-7.93561414e-02 -2.44744852e-01 -1.19819951e+00 -2.77254313e-01
-5.11469245e-01 -4.51829404e-01 4.46527153e-02 1.08480966e+00
3.82598862e-02 6.83756948e-01 -5.20902812e-01 3.81978899e-01
2.55255371e-01 1.26200989e-01 6.84591472e-01 -1.24090219e+00
-2.65164047e-01 -9.78440568e-02 -7.60318637e-01 -1.01040590e+00
4.25643295e-01 -6.59043610e-01 2.04371989e-01 -1.31006956e+00
1.84871763e-01 1.03073299e-01 -2.40768537e-01 4.18552250e-01
2.87921846e-01 5.90886831e-01 1.19410045e-01 2.72246540e-01
-7.90144444e-01 4.84852761e-01 1.01411819e+00 -5.58453441e-01
-1.78167835e-01 -2.46733278e-01 -3.73357266e-01 9.68031049e-01
3.49017739e-01 -2.24452049e-01 -4.28286701e-01 -4.21860814e-01
1.32432014e-01 2.56544113e-01 5.13783932e-01 -8.12122345e-01
1.72109261e-01 1.41394138e-01 2.87287176e-01 -4.58395004e-01
1.02577257e+00 -8.66718352e-01 -1.76125407e-01 -9.74503011e-02
-6.18120492e-01 -1.33108512e-01 1.31605253e-01 7.58465707e-01
-5.78025401e-01 -7.17076659e-02 7.15381444e-01 2.27260254e-02
-7.08950281e-01 6.75042719e-02 -5.35438597e-01 8.55524912e-02
7.54637122e-01 -9.73121896e-02 -3.02799702e-01 -6.88025355e-01
-1.08495116e+00 -8.62367749e-02 2.19606504e-01 6.62012637e-01
4.82421875e-01 -1.33443475e+00 -3.45728904e-01 7.73753822e-02
3.74343187e-01 -3.88553411e-01 3.77620071e-01 7.87111521e-01
-2.10745603e-01 2.07152039e-01 8.28174055e-02 -9.42862391e-01
-1.63277709e+00 5.34125328e-01 3.58764321e-01 4.11707193e-01
-5.18210948e-01 6.27336800e-01 5.52026331e-01 2.31641531e-02
7.16051817e-01 -2.06450477e-01 -5.65424442e-01 6.70789719e-01
8.17996621e-01 1.68863058e-01 1.53906783e-02 -1.14583302e+00
-4.02823120e-01 5.21863580e-01 1.00040980e-01 -4.34275061e-01
1.08872414e+00 -3.75962794e-01 1.15393475e-01 1.13114119e+00
1.36732006e+00 4.12869215e-01 -1.20719147e+00 -3.10850441e-01
-2.06864282e-01 -4.10189837e-01 3.31972569e-01 -5.20125091e-01
-1.01548779e+00 1.45385456e+00 9.50407386e-01 3.15484643e-01
7.84181774e-01 4.22418922e-01 2.66195893e-01 2.41903454e-01
-1.17674455e-01 -1.01442194e+00 3.64918083e-01 2.46894866e-01
1.00538206e+00 -1.45863008e+00 -2.70924807e-01 -1.58690229e-01
-7.66658902e-01 8.89377773e-01 2.94699728e-01 3.82387489e-01
5.12319028e-01 1.11684397e-01 5.11976898e-01 -2.71935612e-01
-6.39322042e-01 -3.39939624e-01 5.02321362e-01 4.60759580e-01
5.86490810e-01 4.16425914e-02 3.27513039e-01 3.25026959e-01
-5.36042824e-02 -6.29348636e-01 2.44617462e-01 8.52117717e-01
-4.84909117e-01 -5.76353610e-01 -3.96257639e-01 -1.66403994e-01
-7.12458491e-01 1.30396947e-01 -6.36397302e-01 5.40362179e-01
6.60075024e-02 1.17212081e+00 5.52474082e-01 8.24561622e-03
2.85172760e-01 3.92837733e-01 3.17567348e-01 -6.71046138e-01
-3.63323689e-01 4.13904577e-01 3.30347151e-01 -4.57867265e-01
-7.76015759e-01 -5.02130091e-01 -9.47872102e-01 2.14860126e-01
5.30290306e-02 2.35098600e-01 8.98389995e-01 8.71010780e-01
1.49488136e-01 5.26995778e-01 6.43020213e-01 -1.00076556e+00
-1.57189727e-01 -1.10697997e+00 -6.19406462e-01 6.15502238e-01
8.52094531e-01 -7.90365815e-01 -6.67837679e-01 3.44978541e-01] | [14.53526782989502, 5.036471366882324] |
749e6608-97df-4f4f-aa19-9dc9920b83b7 | neural-probabilistic-motor-primitives-for | 1811.11711 | null | http://arxiv.org/abs/1811.11711v2 | http://arxiv.org/pdf/1811.11711v2.pdf | Neural probabilistic motor primitives for humanoid control | We focus on the problem of learning a single motor module that can flexibly
express a range of behaviors for the control of high-dimensional physically
simulated humanoids. To do this, we propose a motor architecture that has the
general structure of an inverse model with a latent-variable bottleneck. We
show that it is possible to train this model entirely offline to compress
thousands of expert policies and learn a motor primitive embedding space. The
trained neural probabilistic motor primitive system can perform one-shot
imitation of whole-body humanoid behaviors, robustly mimicking unseen
trajectories. Additionally, we demonstrate that it is also straightforward to
train controllers to reuse the learned motor primitive space to solve tasks,
and the resulting movements are relatively naturalistic. To support the
training of our model, we compare two approaches for offline policy cloning,
including an experience efficient method which we call linear feedback policy
cloning. We encourage readers to view a supplementary video (
https://youtu.be/CaDEf-QcKwA ) summarizing our results. | ['Arun Ahuja', 'Nicolas Heess', 'Josh Merel', 'Alexandre Galashov', 'Yee Whye Teh', 'Vu Pham', 'Leonard Hasenclever', 'Greg Wayne'] | 2018-11-28 | neural-probabilistic-motor-primitives-for-1 | https://openreview.net/forum?id=BJl6TjRcY7 | https://openreview.net/pdf?id=BJl6TjRcY7 | iclr-2019-5 | ['humanoid-control'] | ['robots'] | [ 1.56483293e-01 5.75720608e-01 -2.07248271e-01 2.24508584e-01
-5.84616065e-01 -6.23077691e-01 5.48143148e-01 -6.30539417e-01
-4.92319375e-01 8.55696976e-01 2.48168528e-01 -2.17034474e-01
-7.02665448e-02 -2.77745157e-01 -1.24953234e+00 -7.45672047e-01
-2.54461169e-01 5.17844558e-01 1.42808229e-01 -8.10951516e-02
1.87240988e-01 5.95246315e-01 -1.52192163e+00 -1.85819473e-02
5.93109369e-01 3.54148328e-01 5.94560087e-01 1.03264809e+00
5.86257994e-01 7.99923420e-01 -2.24531546e-01 1.06291540e-01
4.12370205e-01 -5.47326624e-01 -8.48331392e-01 -3.38667855e-02
3.53153888e-03 -6.69663668e-01 -7.92731166e-01 7.27391362e-01
4.56684291e-01 5.38672447e-01 7.22011030e-01 -1.35911834e+00
-5.24661422e-01 5.51569223e-01 5.72242513e-02 -2.81239420e-01
1.85589120e-01 8.19498539e-01 8.02738547e-01 -3.18988949e-01
8.10224473e-01 1.35804820e+00 4.18538719e-01 1.17833102e+00
-1.23122370e+00 -5.15822113e-01 6.96499869e-02 -5.01895845e-02
-9.26336110e-01 -3.59472901e-01 3.21855307e-01 -4.82828587e-01
1.14361346e+00 -1.21494852e-01 9.64099646e-01 1.71366084e+00
4.21040118e-01 1.05857635e+00 8.48247290e-01 -1.58397868e-01
2.99702257e-01 -5.46932220e-01 -4.02345359e-01 8.67796183e-01
-2.00338452e-03 7.00887501e-01 -4.18337345e-01 -2.29981959e-01
1.16398406e+00 1.47874311e-01 -2.89838552e-01 -7.57876217e-01
-1.46231735e+00 6.00106061e-01 3.66914779e-01 -6.00182414e-02
-3.77900809e-01 1.01491463e+00 1.42545238e-01 2.32333168e-01
-2.98786998e-01 7.13212550e-01 -5.51009476e-01 -5.80159843e-01
-2.97864288e-01 7.36548126e-01 8.59728515e-01 1.09244895e+00
3.11419874e-01 2.34907076e-01 -4.88074347e-02 3.50465953e-01
1.28786802e-01 5.89420140e-01 4.58331972e-01 -1.86078000e+00
1.55862659e-01 1.48509160e-01 4.24000323e-01 -2.96445161e-01
-2.25959510e-01 1.99109674e-01 -3.83869290e-01 7.79075146e-01
2.97790766e-01 -4.75470573e-01 -9.46388543e-01 1.99605167e+00
3.31571698e-01 3.07005018e-01 3.45260836e-02 9.20737982e-01
-1.41300470e-01 5.43305814e-01 9.69747305e-02 3.49491656e-01
7.03710377e-01 -1.05921173e+00 -2.35858679e-01 -2.15522453e-01
7.06937850e-01 -1.19708158e-01 1.24859583e+00 3.05908889e-01
-1.18640494e+00 -3.15219760e-01 -9.36662614e-01 -9.20303315e-02
-5.77297881e-02 2.45631799e-01 4.72573906e-01 -1.67660266e-01
-1.03165424e+00 1.30829573e+00 -1.67826509e+00 -3.89159262e-01
1.77112043e-01 5.95813751e-01 -4.23876673e-01 2.88832307e-01
-9.82303679e-01 1.11660850e+00 4.40639943e-01 -2.90162474e-01
-1.65092921e+00 -4.65217501e-01 -7.11203277e-01 -4.36205193e-02
4.37424988e-01 -1.00395739e+00 1.68629682e+00 -8.23005319e-01
-2.24917865e+00 4.16272968e-01 9.82283503e-02 -3.73682201e-01
4.90743041e-01 -4.83481020e-01 2.96067417e-01 2.61968493e-01
-5.19319624e-02 9.78717446e-01 1.03645396e+00 -1.02585518e+00
-5.51011205e-01 -9.51470211e-02 2.38185078e-01 2.21138462e-01
-6.33689314e-02 -2.12539002e-01 -2.13667497e-01 -7.45128572e-01
-5.33938706e-01 -1.48229623e+00 -4.37023073e-01 4.83038187e-01
-2.20348507e-01 -1.23955317e-01 4.21021819e-01 -5.30986130e-01
7.54892826e-01 -2.06283307e+00 1.12410748e+00 -1.18943647e-01
1.94234967e-01 6.85969368e-02 -2.32147142e-01 6.07990086e-01
9.89775658e-02 -1.21786237e-01 -4.42526877e-01 -2.58160561e-01
3.24765474e-01 6.09925389e-01 -4.67568934e-01 4.49964792e-01
1.93899676e-01 1.16070557e+00 -1.11407614e+00 -2.43873447e-01
8.92086402e-02 4.03899819e-01 -7.80338466e-01 3.85409862e-01
-6.45648599e-01 9.61713076e-01 -7.06441224e-01 2.23703608e-01
-2.51901150e-01 -1.52546912e-01 2.76174426e-01 3.44026089e-01
-5.27424030e-02 2.93664485e-01 -9.35363472e-01 1.81720603e+00
-2.86133528e-01 3.94031793e-01 4.25607860e-01 -8.68597209e-01
2.39452675e-01 4.07262206e-01 5.61374009e-01 -1.17433548e-01
1.97604969e-01 1.70840859e-01 -8.70469287e-02 -4.81135488e-01
2.90634990e-01 -3.78511995e-02 -1.96348041e-01 6.62766039e-01
2.07719848e-01 -3.19013417e-01 -1.25877962e-01 9.47028995e-02
1.36216140e+00 9.66565549e-01 1.99422956e-01 1.83535293e-02
6.06320426e-02 2.38461480e-01 3.37689728e-01 6.19421184e-01
-1.76561281e-01 1.02946311e-01 4.19644743e-01 -1.08283674e-02
-1.36191666e+00 -1.36534333e+00 4.73122656e-01 1.11851621e+00
-6.33112118e-02 -4.59870577e-01 -8.47066402e-01 -2.43670553e-01
3.07732224e-01 6.50258839e-01 -5.99306107e-01 -3.34091187e-01
-9.98975277e-01 1.55270278e-01 6.62751913e-01 7.98860908e-01
-5.50258905e-02 -1.37742269e+00 -1.19904113e+00 1.87368691e-01
1.23588264e-01 -6.37764573e-01 -4.02985960e-01 2.90866017e-01
-8.49178553e-01 -1.04338539e+00 -9.30458963e-01 -7.53414214e-01
5.05021214e-01 -5.00339009e-02 6.72601044e-01 8.60219598e-02
-3.09193999e-01 8.15972924e-01 -2.56759644e-01 -9.30180252e-02
-7.64166355e-01 7.31986621e-03 4.77479190e-01 -7.30790317e-01
-1.42615631e-01 -7.92748272e-01 -5.74044049e-01 1.17953509e-01
-7.20742702e-01 4.37691629e-01 6.20020270e-01 9.79469478e-01
5.44390619e-01 -5.10984421e-01 -2.23565567e-03 -1.36686862e-01
6.43369496e-01 -4.80795175e-01 -6.34046495e-01 2.82827532e-04
-2.77039558e-01 4.92638528e-01 9.77127373e-01 -8.35811973e-01
-7.06689417e-01 4.99974340e-01 -9.35845673e-02 -7.68000364e-01
-3.02740186e-01 -3.02544385e-02 1.63143829e-01 -6.20949976e-02
5.43704212e-01 3.75845939e-01 4.82563853e-01 -5.05488157e-01
9.55497503e-01 4.54499334e-01 7.19883502e-01 -9.89782274e-01
6.93617880e-01 4.79438961e-01 -6.64122254e-02 -5.64721644e-01
-1.30830556e-01 -6.96080783e-03 -6.32307291e-01 -2.36848161e-01
9.65581119e-01 -7.96657324e-01 -1.16189790e+00 3.56310815e-01
-1.01361692e+00 -1.39903224e+00 -4.61869657e-01 6.51971877e-01
-1.68815124e+00 3.06855112e-01 -9.46929753e-01 -5.85012913e-01
2.75258850e-02 -1.23673153e+00 1.21949673e+00 -8.32774043e-02
-4.80086863e-01 -5.88264108e-01 4.96708333e-01 -3.55269551e-01
2.44578093e-01 2.68633366e-01 7.99013317e-01 -1.61221966e-01
-6.98861420e-01 2.79803187e-01 5.00648141e-01 1.05659336e-01
6.27378076e-02 7.91246444e-03 -4.61080968e-01 -5.33290207e-01
-2.38294080e-01 -6.18637562e-01 6.20704353e-01 3.03457499e-01
1.11773825e+00 -3.59027237e-01 -6.48225129e-01 6.73065782e-01
1.03151536e+00 -9.69455466e-02 3.43351185e-01 2.51131147e-01
6.60586536e-01 4.98959631e-01 4.79611278e-01 4.34647441e-01
2.54530191e-01 7.17870951e-01 4.00145292e-01 3.11149627e-01
3.17717642e-02 -6.51098490e-01 8.23997021e-01 6.67673767e-01
-2.40803584e-01 9.74241421e-02 -7.25416601e-01 5.09901583e-01
-2.06366110e+00 -1.06924284e+00 5.26673973e-01 1.78446805e+00
8.50892603e-01 -3.36373001e-02 4.62772608e-01 -2.84806788e-01
3.96889627e-01 -2.42866814e-01 -1.02573895e+00 -4.46901292e-01
5.26927471e-01 3.15175742e-01 7.75977790e-01 6.32387698e-01
-8.51670682e-01 1.20072651e+00 6.69109154e+00 4.94020492e-01
-1.02642775e+00 -6.28565624e-02 -1.26804039e-01 -4.57752675e-01
-1.24750808e-01 6.62819296e-02 -4.96476263e-01 4.33678538e-01
1.14460146e+00 -2.63770640e-01 9.80110943e-01 1.02573884e+00
4.17979151e-01 1.34062603e-01 -1.25346541e+00 6.12901628e-01
-4.13456380e-01 -1.17390525e+00 -1.09826460e-01 8.41647536e-02
5.68970442e-01 3.03630859e-01 4.88852486e-02 5.38363874e-01
1.07739079e+00 -1.08483493e+00 8.05476904e-01 5.52784801e-01
8.50958705e-01 -4.14672285e-01 -2.94510722e-01 8.29331040e-01
-1.01025736e+00 -4.63530302e-01 -2.22992212e-01 -2.66736954e-01
5.10440290e-01 -5.65562427e-01 -3.46383631e-01 6.15540855e-02
6.31873131e-01 8.21947515e-01 -1.41656036e-02 8.36583138e-01
-4.89313692e-01 3.43448937e-01 -4.87674266e-01 -2.34048843e-01
3.20842892e-01 -1.69032216e-01 7.09507763e-01 7.44511127e-01
3.83871645e-01 1.29523933e-01 5.65607697e-02 9.42067742e-01
1.87422320e-01 -4.14195001e-01 -9.02106464e-01 -3.91249925e-01
1.69338673e-01 8.16542208e-01 -3.05532485e-01 -3.07090491e-01
-1.43760785e-01 1.37382054e+00 7.53128648e-01 4.14021909e-01
-1.01606953e+00 -8.24725032e-02 1.11590350e+00 -1.60815403e-01
6.27118111e-01 -8.33730042e-01 2.71437824e-01 -1.15672255e+00
-2.06755958e-02 -1.05947065e+00 -1.25904724e-01 -8.83883774e-01
-9.77221668e-01 1.91215172e-01 1.64962471e-01 -1.25563574e+00
-8.97014380e-01 -8.59799564e-01 -4.58075941e-01 5.56605220e-01
-8.90934646e-01 -8.51697206e-01 1.07171379e-01 7.08152473e-01
4.28912699e-01 3.83761413e-02 9.61071253e-01 -1.70711592e-01
-2.93613493e-01 3.61689448e-01 4.12504762e-01 -6.51305765e-02
5.61980546e-01 -1.24944913e+00 7.00168252e-01 6.56789899e-01
-1.71573356e-01 8.20804954e-01 8.76794994e-01 -7.09416091e-01
-1.68243015e+00 -9.90564883e-01 7.20759779e-02 -5.61354518e-01
9.14942265e-01 -2.81442612e-01 -5.64444840e-01 1.21295106e+00
9.94766451e-05 -4.41397637e-01 2.66902328e-01 -4.60515648e-01
6.12692125e-02 5.54531932e-01 -8.12863529e-01 1.27074802e+00
1.39774489e+00 -4.39260572e-01 -9.08800006e-01 3.28723043e-01
9.14152980e-01 -5.23519099e-01 -7.60734558e-01 1.19438782e-01
8.45398843e-01 -2.95396209e-01 1.04877484e+00 -1.15255845e+00
5.04767776e-01 -1.99627131e-01 5.01345247e-02 -1.46035433e+00
-4.27619606e-01 -9.62438703e-01 -6.22798145e-01 2.75883228e-01
1.51328087e-01 -6.43817723e-01 7.51641154e-01 5.84321260e-01
-2.01442063e-01 -8.40523124e-01 -9.48289990e-01 -1.04219818e+00
5.85223258e-01 -1.79486632e-01 3.41714084e-01 3.82391989e-01
4.52632993e-01 -5.61942346e-02 -6.16084993e-01 4.44551371e-02
4.05923963e-01 7.60974064e-02 1.01167238e+00 -6.22583032e-01
-9.45857346e-01 -5.02999127e-01 -1.19152837e-01 -1.71932566e+00
5.50411820e-01 -9.05385613e-01 3.22649121e-01 -1.26576757e+00
6.18993957e-03 -1.17344774e-01 -8.08350742e-03 4.86261457e-01
4.55667637e-02 -4.26073402e-01 4.12393987e-01 3.89754683e-01
-3.88568908e-01 9.94171977e-01 1.53863096e+00 2.20157251e-01
-1.57488734e-01 2.07886621e-02 -3.32203031e-01 6.47941589e-01
1.12453794e+00 -6.23725593e-01 -4.80224699e-01 -5.57248831e-01
-7.45645761e-02 3.16847861e-01 6.14984035e-01 -9.98878658e-01
2.22751498e-01 -4.46327448e-01 1.72750890e-01 -8.64507258e-02
4.85065222e-01 -5.83099544e-01 1.44071579e-01 1.01907170e+00
-5.97752810e-01 2.69539416e-01 1.69292495e-01 8.12267363e-01
3.34998250e-01 5.09091793e-03 6.22362375e-01 -3.82051289e-01
-7.16630220e-01 3.91116053e-01 -9.64424133e-01 -7.55759999e-02
1.19979119e+00 -8.07439238e-02 -8.30721781e-02 -3.36800635e-01
-9.48592663e-01 4.84160691e-01 9.68467057e-01 3.96024823e-01
5.75229406e-01 -1.16896713e+00 -3.31184685e-01 8.09543058e-02
-8.63298550e-02 -3.22071314e-01 -5.95318861e-02 6.22128129e-01
-6.04572594e-01 3.24451745e-01 -6.62075579e-01 -3.02027553e-01
-7.87172377e-01 7.39184082e-01 5.50899565e-01 -1.11591583e-02
-1.10544753e+00 5.48299611e-01 4.85548042e-02 -7.03404605e-01
1.81637555e-01 -5.23981869e-01 2.45074809e-01 -9.67858791e-01
2.27451175e-01 3.16884696e-01 -8.95326078e-01 -4.25745338e-01
-8.56080726e-02 4.48448509e-01 4.14453030e-01 -6.15911603e-01
1.26590180e+00 8.20691288e-02 1.40621901e-01 3.03981066e-01
9.59244847e-01 -2.75242120e-01 -2.02118325e+00 2.22188160e-01
-1.98047429e-01 -1.10140018e-01 -5.09142041e-01 -4.40415740e-01
-4.43156868e-01 8.51787031e-01 3.48852336e-01 -3.80937904e-01
7.33431220e-01 -1.07145263e-02 8.92127097e-01 8.92873108e-01
7.12364376e-01 -1.23339474e+00 1.46076411e-01 6.70846522e-01
9.94803786e-01 -8.75558197e-01 -2.34290361e-01 2.46464789e-01
-6.06917560e-01 1.10359144e+00 4.93033767e-01 -6.84587240e-01
4.93204981e-01 5.64811110e-01 -3.16632599e-01 -7.94359669e-02
-1.08395493e+00 -3.00208300e-01 -1.33714139e-01 7.37824261e-01
-5.89464232e-03 1.99518085e-01 -1.41039565e-01 3.14836621e-01
-2.49211609e-01 3.78463149e-01 4.12247479e-01 1.15665925e+00
-5.80071032e-01 -1.24839008e+00 8.62396732e-02 2.34747574e-01
-1.99847877e-01 2.21216276e-01 -3.08917165e-01 9.25849855e-01
-3.68661255e-01 4.64491129e-01 -2.30692938e-01 -4.46565390e-01
7.62429684e-02 8.13882872e-02 8.98914635e-01 -7.20679462e-01
-2.16689721e-01 -2.43274555e-01 -1.39585033e-01 -1.26805127e+00
-1.14728190e-01 -5.93528509e-01 -1.52177310e+00 -3.71941298e-01
2.49678299e-01 -2.19906867e-01 4.33418989e-01 8.18502665e-01
4.85913336e-01 4.06895965e-01 1.30319282e-01 -1.41813564e+00
-1.16081178e+00 -6.94707632e-01 -4.17545766e-01 4.49328512e-01
5.12446821e-01 -1.01298940e+00 -4.09680754e-01 2.37392470e-01] | [4.519857406616211, 1.009150505065918] |
d5980ec6-4d94-40cc-9688-187daec97d32 | neural-intrinsic-embedding-for-non-rigid | 2303.01038 | null | https://arxiv.org/abs/2303.01038v1 | https://arxiv.org/pdf/2303.01038v1.pdf | Neural Intrinsic Embedding for Non-rigid Point Cloud Matching | As a primitive 3D data representation, point clouds are prevailing in 3D sensing, yet short of intrinsic structural information of the underlying objects. Such discrepancy poses great challenges on directly establishing correspondences between point clouds sampled from deformable shapes. In light of this, we propose Neural Intrinsic Embedding (NIE) to embed each vertex into a high-dimensional space in a way that respects the intrinsic structure. Based upon NIE, we further present a weakly-supervised learning framework for non-rigid point cloud registration. Unlike the prior works, we do not require expansive and sensitive off-line basis construction (e.g., eigen-decomposition of Laplacians), nor do we require ground-truth correspondence labels for supervision. We empirically show that our framework performs on par with or even better than the state-of-the-art baselines, which generally require more supervision and/or more structural geometric input. | ['Ruqi Huang', 'Mingze Sun', 'Puhua Jiang'] | 2023-03-02 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Jiang_Neural_Intrinsic_Embedding_for_Non-Rigid_Point_Cloud_Matching_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Jiang_Neural_Intrinsic_Embedding_for_Non-Rigid_Point_Cloud_Matching_CVPR_2023_paper.pdf | cvpr-2023-1 | ['point-cloud-registration'] | ['computer-vision'] | [ 1.77398175e-01 1.83174953e-01 -1.55124724e-01 -3.46109927e-01
-6.37233019e-01 -7.81552732e-01 7.88970470e-01 -3.19975503e-02
-1.75780728e-01 1.15945891e-01 8.32594484e-02 -1.44526765e-01
-1.39857844e-01 -7.56974876e-01 -9.44158494e-01 -5.79269767e-01
2.02186540e-01 8.03476214e-01 6.42528981e-02 -6.46880642e-02
2.01052427e-01 1.06981659e+00 -1.34227467e+00 -2.55571961e-01
7.22663760e-01 6.90043688e-01 -3.35328914e-02 1.73675418e-01
-1.42186806e-01 -4.31260727e-02 2.82611907e-01 -2.96734005e-01
5.26850939e-01 6.78014830e-02 -6.97947860e-01 3.24327201e-01
8.31343412e-01 -2.27530926e-01 -4.30044711e-01 1.10042381e+00
2.62340039e-01 3.04768048e-02 1.00642657e+00 -1.09867322e+00
-1.04591358e+00 -1.20595381e-01 -6.84732556e-01 -3.12447608e-01
3.01714659e-01 -8.23588967e-02 1.10098469e+00 -1.50178015e+00
6.72287822e-01 1.24005198e+00 8.83723795e-01 4.58242536e-01
-1.63237011e+00 -3.76254320e-01 1.30248621e-01 -3.12568843e-01
-1.59492946e+00 -5.67352772e-01 1.21635258e+00 -7.63524294e-01
6.99069858e-01 2.83820987e-01 4.93169844e-01 1.04910731e+00
-1.97803229e-01 3.73564452e-01 8.49783123e-01 -3.71432602e-01
1.41888201e-01 2.20680702e-02 8.82333796e-03 6.30045593e-01
3.97865385e-01 1.66065514e-01 -3.54294270e-01 -4.77923274e-01
1.22976696e+00 3.25059533e-01 -2.32814595e-01 -1.13157630e+00
-1.47050190e+00 7.31661618e-01 5.92976034e-01 2.16814741e-01
-3.09890985e-01 1.93561703e-01 -1.78542718e-01 8.14298093e-02
6.26470625e-01 1.37648448e-01 -2.80577004e-01 1.72290236e-01
-8.11624467e-01 1.21267036e-01 5.69181085e-01 1.17518544e+00
1.22792161e+00 -7.55398199e-02 3.99203867e-01 4.42589760e-01
6.46656334e-01 7.01027095e-01 -2.41045028e-01 -1.15734291e+00
3.47821444e-01 8.10990274e-01 1.47920147e-01 -1.29797113e+00
-1.59590751e-01 -1.70668319e-01 -9.61023569e-01 3.83924723e-01
2.52590477e-01 3.98905426e-01 -9.39946234e-01 1.56945217e+00
4.48717952e-01 6.54080987e-01 -2.07248747e-01 8.14400852e-01
5.62255383e-01 3.84438008e-01 -5.06991208e-01 1.46247119e-01
1.13343561e+00 -5.76903164e-01 -2.97390223e-01 -5.80055155e-02
2.14417547e-01 -6.43784642e-01 1.15960908e+00 -1.14886038e-01
-1.11258984e+00 -4.61907059e-01 -9.51708794e-01 -4.67985690e-01
-1.78041101e-01 -4.11822684e-02 5.52614748e-01 4.68822002e-01
-1.16098654e+00 5.10638654e-01 -1.34187412e+00 -3.02366644e-01
4.41613466e-01 4.01842952e-01 -7.26892114e-01 -1.71879549e-02
-4.59696859e-01 7.24187732e-01 -1.67904735e-01 1.51458517e-01
-5.54257870e-01 -8.84817898e-01 -1.07425213e+00 -2.97214687e-01
2.17069581e-01 -9.96154964e-01 8.05855334e-01 -4.26698357e-01
-1.48930788e+00 1.40703976e+00 -3.87316167e-01 2.63666194e-02
5.40163398e-01 -1.97937936e-01 1.08260944e-01 -1.33599099e-02
-9.15376237e-04 6.13002837e-01 8.96942973e-01 -1.80810308e+00
8.21311548e-02 -7.08446562e-01 2.07091153e-01 2.17083827e-01
-1.28401861e-01 -1.40485168e-01 -5.25158882e-01 -4.60710377e-01
8.72948170e-01 -1.31879950e+00 -3.74319941e-01 6.49628282e-01
-3.99501711e-01 -2.31762990e-01 9.52857673e-01 -1.88601360e-01
4.62296903e-01 -2.27123594e+00 3.35025877e-01 4.81569707e-01
5.21005809e-01 -1.17287464e-01 -1.62238970e-01 2.05130652e-01
-1.35779247e-01 1.52793303e-01 -5.38458645e-01 -6.14675462e-01
4.06341553e-01 4.03343439e-01 -4.24114525e-01 9.91158426e-01
4.18172687e-01 1.02745616e+00 -8.46015871e-01 -2.30290607e-01
3.66359472e-01 8.56199861e-01 -6.33937538e-01 7.34761283e-02
-4.71651033e-02 6.74360573e-01 -6.05537295e-01 6.92333519e-01
7.80208051e-01 -4.42320287e-01 -3.25852066e-01 -4.59734142e-01
-1.07479259e-01 4.87715513e-01 -1.39625418e+00 2.23433590e+00
-2.86438853e-01 2.23725080e-01 2.98206866e-01 -9.98202980e-01
1.01393425e+00 2.20118910e-01 8.19879770e-01 -2.26957574e-02
-1.30305029e-02 2.36268952e-01 -2.71639019e-01 3.22628506e-02
2.93958664e-01 -6.15697540e-02 1.99469909e-01 5.53813040e-01
-2.26574212e-01 -4.19639289e-01 -5.42627037e-01 5.52812070e-02
9.03374851e-01 3.72954309e-01 6.70856759e-02 -3.73669147e-01
4.15114760e-01 -1.78745329e-01 6.51529431e-01 3.70984763e-01
1.05836481e-01 9.46338952e-01 -5.94228879e-02 -5.00282586e-01
-1.17754054e+00 -1.54992414e+00 -4.31740016e-01 5.45239151e-01
4.07219380e-01 -2.15533271e-01 -5.00356257e-01 -4.43550497e-01
3.44028324e-01 3.28715682e-01 -4.27468598e-01 6.44977689e-02
-5.98633885e-01 -3.70653719e-01 2.81777591e-01 6.12632871e-01
7.41865560e-02 -5.71714044e-01 -2.59072274e-01 6.27072752e-02
3.00877631e-01 -1.36759210e+00 -6.64591730e-01 -3.05915792e-02
-1.16091573e+00 -8.57569873e-01 -4.97627169e-01 -7.71880209e-01
1.18834591e+00 5.63987195e-01 1.31657040e+00 -7.37229455e-03
4.75567579e-02 8.69126618e-01 -7.91515857e-02 -3.03054780e-01
-9.70008969e-02 5.01060970e-02 3.90046626e-01 4.97556068e-02
4.98452455e-01 -1.16834700e+00 -4.43746775e-01 4.47071791e-01
-8.08395326e-01 -2.08074227e-02 5.69376290e-01 4.74481046e-01
9.84741211e-01 -4.23575521e-01 6.74283281e-02 -7.99910247e-01
2.39612475e-01 -2.83448398e-01 -7.67920911e-01 6.11256994e-02
-4.65262562e-01 6.17604032e-02 1.80567712e-01 -2.11878330e-01
-5.19414544e-01 5.35584748e-01 -2.60217078e-02 -9.89357889e-01
-3.35883915e-01 1.68647826e-01 -2.97234148e-01 -5.17795146e-01
6.71632588e-01 1.18316390e-01 4.73301485e-02 -8.03969443e-01
6.58287942e-01 4.41524237e-01 7.21572876e-01 -8.67965698e-01
1.64582944e+00 1.02343369e+00 1.49834275e-01 -7.90866852e-01
-5.33452451e-01 -5.52700758e-01 -1.49590099e+00 2.59349436e-01
8.59615147e-01 -1.00228143e+00 -5.41681468e-01 1.00935325e-01
-1.42953813e+00 3.01941540e-02 -4.08146352e-01 3.72809917e-01
-7.22635746e-01 5.71545780e-01 -4.36702639e-01 -6.29945457e-01
-8.11594129e-02 -1.07559705e+00 1.49444342e+00 -3.50334346e-01
-5.47345281e-02 -9.94243324e-01 1.88757688e-01 2.41937488e-01
1.39872208e-01 5.69003105e-01 6.89308345e-01 -2.49569133e-01
-1.00920999e+00 -2.08183378e-01 -2.65668362e-01 2.08482876e-01
5.23653388e-01 2.24414002e-02 -1.06380403e+00 -4.48926747e-01
1.61663517e-01 -7.35317916e-02 4.42205369e-01 1.64165735e-01
1.14641798e+00 -1.37255356e-01 -4.29464191e-01 9.99567389e-01
1.28922796e+00 -4.28524822e-01 4.93248701e-01 2.05110803e-01
1.19300354e+00 5.52581966e-01 1.62836924e-01 1.67425588e-01
6.01356685e-01 7.34927475e-01 5.73986769e-01 -2.98494309e-01
1.43800884e-01 -4.16138083e-01 2.51121908e-01 1.05286038e+00
-5.15786469e-01 3.05562764e-01 -1.12912333e+00 5.44777274e-01
-1.62720716e+00 -7.31371343e-01 -1.59655735e-01 2.23407555e+00
7.47965813e-01 -3.84231322e-02 -1.21118680e-01 9.20854136e-02
5.55429578e-01 3.25102746e-01 -7.26304829e-01 7.63578415e-02
-7.98172057e-02 1.57085001e-01 4.84484553e-01 6.90497041e-01
-1.02783418e+00 9.20146644e-01 6.21457434e+00 2.06123918e-01
-9.36404228e-01 8.50914493e-02 8.66170526e-02 1.47657052e-01
-8.48549426e-01 1.22309968e-01 -5.79502404e-01 1.86470866e-01
3.92456859e-01 -9.35270712e-02 4.52021211e-01 7.42733121e-01
-8.27415381e-03 6.49748802e-01 -1.53327715e+00 1.30638409e+00
-1.07476395e-03 -1.47697222e+00 1.52270272e-01 4.96511489e-01
8.12589228e-01 4.07956988e-01 -1.03060203e-02 -3.47020358e-01
5.19879639e-01 -1.07661366e+00 6.11024797e-01 4.75933135e-01
9.49813068e-01 -3.90443981e-01 2.44517222e-01 4.29353982e-01
-1.29343557e+00 6.49995625e-01 -7.00343549e-01 -3.93750239e-03
2.00781167e-01 7.77598798e-01 -4.27006781e-01 6.83140635e-01
6.39564574e-01 9.61718202e-01 -1.76570848e-01 6.06357574e-01
-2.64890522e-01 3.53497863e-01 -6.98783576e-01 6.13864899e-01
5.70395105e-02 -6.50869250e-01 7.47193158e-01 8.82441759e-01
3.17867696e-01 3.18125784e-01 3.88470680e-01 1.33326483e+00
-3.68450940e-01 -1.37547880e-01 -1.05127776e+00 7.71449879e-02
5.25055110e-01 1.22062635e+00 -4.34820443e-01 -7.25703388e-02
-6.80036008e-01 1.02721238e+00 2.85917938e-01 3.89467329e-01
-4.95701402e-01 1.16820917e-01 1.03986704e+00 3.10619622e-01
1.93964750e-01 -7.96954870e-01 -5.33442259e-01 -1.34661770e+00
4.46029842e-01 -5.03614068e-01 -1.86649635e-01 -6.46621108e-01
-1.76719427e+00 4.44065362e-01 -1.14016578e-01 -1.52796078e+00
-9.34666172e-02 -6.83143973e-01 -6.13127172e-01 1.01377749e+00
-1.78801942e+00 -1.40440977e+00 -3.73110086e-01 7.49646366e-01
6.89854100e-02 1.78650513e-01 9.21363652e-01 2.48047754e-01
-5.40713267e-03 3.67208213e-01 5.54051101e-02 2.48881340e-01
4.70861018e-01 -1.22949862e+00 8.99646938e-01 7.20939577e-01
6.33475482e-01 9.37466383e-01 3.54319632e-01 -5.11744380e-01
-1.86726749e+00 -1.06014919e+00 6.75498188e-01 -1.11728776e+00
5.98025322e-01 -7.71986842e-01 -1.15370440e+00 9.83157158e-01
-1.57857880e-01 5.48427820e-01 5.82881212e-01 1.47654727e-01
-5.91917753e-01 1.03318602e-01 -1.09422266e+00 4.77753490e-01
1.63575590e+00 -1.01749492e+00 -8.78441870e-01 3.20787966e-01
7.62961805e-01 -6.59719944e-01 -9.53142822e-01 6.05474293e-01
3.07127833e-01 -6.16086900e-01 1.53260946e+00 -6.27919018e-01
9.68858898e-02 -6.71466172e-01 -4.07169789e-01 -9.84445632e-01
-7.15240955e-01 -6.20336413e-01 -4.18248065e-02 1.05856526e+00
2.13886559e-01 -5.00306606e-01 1.10878384e+00 8.49774957e-01
-3.53345513e-01 -5.03014207e-01 -1.00087416e+00 -9.06498492e-01
3.25865269e-01 -5.22290111e-01 6.88795507e-01 1.43926096e+00
-4.73310471e-01 1.93100438e-01 1.96209699e-02 7.34390020e-01
1.02694464e+00 3.84727806e-01 1.03354359e+00 -1.77059615e+00
5.01051694e-02 -2.42269009e-01 -7.24728465e-01 -1.29927242e+00
3.93534303e-01 -1.12522125e+00 -4.98736501e-02 -1.47528708e+00
-9.65076499e-03 -8.24954093e-01 -2.32501969e-01 5.36627531e-01
1.54466897e-01 4.69889402e-01 3.95232849e-02 6.39755964e-01
-2.13578418e-01 8.95891964e-01 1.13526475e+00 -2.14176908e-01
-1.38904557e-01 -1.44356862e-01 -5.69810212e-01 1.01190591e+00
4.45750326e-01 -4.97267604e-01 -3.84528399e-01 -8.90380204e-01
1.46322593e-01 -3.25866759e-01 6.50554955e-01 -6.94323778e-01
3.28278452e-01 -3.91691983e-01 1.52264729e-01 -7.22679913e-01
5.11661291e-01 -1.27655125e+00 3.61502171e-01 -5.75912185e-03
1.09498739e-01 2.73394138e-01 4.76969257e-02 6.69108808e-01
1.60122067e-02 -1.07614547e-02 5.64317703e-01 -6.69699609e-02
-2.80909628e-01 8.83612990e-01 3.55958879e-01 -8.51207674e-02
7.19627678e-01 -5.16463459e-01 4.52380255e-02 -1.16970018e-02
-4.93839949e-01 -1.41817987e-01 1.30804098e+00 4.41668481e-01
6.67281866e-01 -1.65245533e+00 -6.70321882e-01 4.22380865e-01
2.44100899e-01 7.08246768e-01 -2.60594159e-01 6.57356858e-01
-3.61916244e-01 2.08891064e-01 -4.06868719e-02 -1.06844556e+00
-8.38623881e-01 4.63433594e-01 2.64821261e-01 2.73534268e-01
-9.87199128e-01 6.12441540e-01 4.94382441e-01 -1.09371018e+00
5.73405027e-02 -5.04802823e-01 3.26824039e-01 -3.72121841e-01
1.36497304e-01 1.06553875e-01 1.20979004e-01 -9.00104284e-01
-5.60734272e-01 1.22899604e+00 1.90218106e-01 -1.80318803e-01
1.64547813e+00 1.17170718e-02 -2.41365790e-01 5.78605950e-01
1.17301857e+00 3.48274112e-01 -1.51012516e+00 -6.93847656e-01
1.34832844e-01 -7.44518340e-01 8.76957402e-02 -7.00392053e-02
-1.02183294e+00 1.04464626e+00 3.20213974e-01 3.40483896e-02
6.54526114e-01 3.10310632e-01 6.10424280e-01 5.41228771e-01
6.42778337e-01 -5.55743515e-01 -1.66135311e-01 3.93002748e-01
1.06133342e+00 -1.24573672e+00 2.17092961e-01 -7.52061844e-01
-4.06957269e-02 8.78543615e-01 2.47080252e-01 -5.99753797e-01
8.47805500e-01 2.71603823e-01 2.48028012e-03 -5.30549645e-01
-3.97663683e-01 -1.17214330e-01 6.76998913e-01 8.81062150e-01
2.46650741e-01 -9.80233699e-02 2.44068921e-01 1.06591858e-01
-2.40459219e-01 -2.08224311e-01 1.54310346e-01 8.30792010e-01
-2.32619286e-01 -1.26665974e+00 -4.12626475e-01 1.64275870e-01
3.53726074e-02 1.84562840e-02 -6.03325427e-01 6.24002695e-01
-1.16482243e-01 4.15849775e-01 2.18483716e-01 -3.28493655e-01
6.06991231e-01 -6.44771084e-02 5.21754146e-01 -8.28356862e-01
8.39925930e-02 3.90908234e-02 -4.82311755e-01 -6.71264052e-01
-7.38723159e-01 -9.16351318e-01 -1.25714099e+00 -3.18291366e-01
-2.26731375e-01 -1.56695396e-01 6.87489510e-01 7.43990183e-01
6.57293260e-01 -1.18646592e-01 6.95383310e-01 -1.39926827e+00
-5.08369207e-01 -5.54987192e-01 -4.82699960e-01 6.30290627e-01
4.75463837e-01 -9.13631082e-01 -5.37360549e-01 2.56511778e-01] | [8.24924087524414, -3.2735655307769775] |
9ff37d9e-5f95-4820-a3d9-6e3aefd56e04 | benchmarks-for-automated-commonsense | 2302.04752 | null | https://arxiv.org/abs/2302.04752v2 | https://arxiv.org/pdf/2302.04752v2.pdf | Benchmarks for Automated Commonsense Reasoning: A Survey | More than one hundred benchmarks have been developed to test the commonsense knowledge and commonsense reasoning abilities of artificial intelligence (AI) systems. However, these benchmarks are often flawed and many aspects of common sense remain untested. Consequently, we do not currently have any reliable way of measuring to what extent existing AI systems have achieved these abilities. This paper surveys the development and uses of AI commonsense benchmarks. We discuss the nature of common sense; the role of common sense in AI; the goals served by constructing commonsense benchmarks; and desirable features of commonsense benchmarks. We analyze the common flaws in benchmarks, and we argue that it is worthwhile to invest the work needed ensure that benchmark examples are consistently high quality. We survey the various methods of constructing commonsense benchmarks. We enumerate 139 commonsense benchmarks that have been developed: 102 text-based, 18 image-based, 12 video based, and 7 simulated physical environments. We discuss the gaps in the existing benchmarks and aspects of commonsense reasoning that are not addressed in any existing benchmark. We conclude with a number of recommendations for future development of commonsense AI benchmarks. | ['Ernest Davis'] | 2023-02-09 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [ 4.37077552e-01 1.12812571e-01 1.70796543e-01 -4.60508168e-01
-1.59142986e-01 -6.61975324e-01 8.69445145e-01 1.26183897e-01
-2.79417455e-01 9.16389704e-01 3.11942995e-01 -3.08769017e-01
-3.71088088e-01 -8.84603381e-01 -4.86520559e-01 -1.46054626e-01
3.88304919e-01 4.29240078e-01 2.86077708e-01 -8.64882648e-01
6.84666574e-01 2.86911130e-01 -1.90011394e+00 4.57080483e-01
1.16794634e+00 7.23575532e-01 -1.13797607e-02 7.49570191e-01
-1.18962571e-01 1.86769986e+00 -1.06605816e+00 -5.20168304e-01
-1.12577872e-02 -8.75637233e-01 -1.35135245e+00 -5.55737466e-02
5.78374445e-01 -2.41466224e-01 -4.24407333e-01 1.50900805e+00
2.22903654e-01 3.88706714e-01 5.54174125e-01 -1.86250460e+00
-1.31839287e+00 8.33733022e-01 1.81639269e-01 5.73496461e-01
8.32009315e-01 5.43700039e-01 1.00948179e+00 -9.59049463e-02
8.14021885e-01 1.36575341e+00 5.68258047e-01 7.61826694e-01
-9.86010611e-01 -7.19144881e-01 -3.38369370e-01 7.61569798e-01
-1.18813789e+00 -5.92479765e-01 8.26081455e-01 -4.27437752e-01
1.40040219e+00 5.29688835e-01 1.08685565e+00 1.10049593e+00
3.81085098e-01 5.65457582e-01 1.62735534e+00 -5.58706880e-01
4.25307363e-01 9.10837501e-02 4.52531099e-01 6.15202904e-01
6.10636890e-01 9.60669070e-02 -5.18991470e-01 -2.01786548e-01
5.58729529e-01 -4.12116736e-01 -3.54880914e-02 -1.17360249e-01
-1.66303504e+00 6.52882874e-01 3.19066644e-01 6.05323017e-01
-2.75822431e-01 5.01572847e-01 7.22619712e-01 4.87978488e-01
6.74154004e-03 1.04898298e+00 5.90691417e-02 -5.95441401e-01
-9.27436829e-01 7.35269129e-01 8.41188788e-01 1.06849599e+00
2.66033679e-01 2.66926229e-01 -3.00721135e-02 7.53906667e-01
3.11621949e-02 6.83673084e-01 5.35070360e-01 -1.58168495e+00
-1.21432848e-01 5.87683618e-01 1.30787820e-01 -1.60385990e+00
-8.89035910e-02 7.41004199e-02 -4.08957124e-01 4.39071864e-01
7.71965459e-02 2.19805270e-01 -5.58615327e-01 1.69325304e+00
-2.06962004e-01 1.94551140e-01 4.17881995e-01 7.32346714e-01
1.16449344e+00 3.96103114e-01 1.99796304e-01 -1.10616200e-02
1.16346717e+00 -5.65579891e-01 -9.36043441e-01 -4.28198427e-01
5.32621324e-01 -7.13907003e-01 1.52199435e+00 2.87732571e-01
-1.28810990e+00 -1.54905632e-01 -1.20515072e+00 -2.98511297e-01
-7.79180825e-01 -6.96875870e-01 9.62462664e-01 7.46382773e-01
-9.77105796e-01 5.69783390e-01 -2.87567616e-01 -4.96711999e-01
3.89806747e-01 -1.63302034e-01 9.03433487e-02 -1.78721577e-01
-1.56654751e+00 1.55177498e+00 6.72021747e-01 -4.81681764e-01
-7.44563401e-01 -3.62813175e-01 -1.07159531e+00 -2.46312961e-01
3.12552214e-01 -7.34625220e-01 1.27949119e+00 -1.10487890e+00
-1.11116278e+00 1.43414664e+00 4.06107306e-02 -6.25281274e-01
3.60721290e-01 2.77685151e-02 -8.23244333e-01 1.68361828e-01
4.65059578e-01 4.81309056e-01 -4.39491682e-02 -1.37982917e+00
-6.47321403e-01 5.71359973e-03 5.71779609e-01 1.29791200e-01
-1.22948119e-03 3.09890002e-01 2.91886479e-01 -6.59476995e-01
-2.50849605e-01 -8.31658900e-01 2.99448818e-01 1.36894897e-01
-3.06509525e-01 -5.82916103e-02 5.93213260e-01 -1.52554780e-01
1.13070118e+00 -2.02467775e+00 -3.75501961e-01 -1.00307010e-01
3.01987737e-01 -6.37798235e-02 5.78751080e-02 3.32748622e-01
4.78840359e-02 2.62335837e-01 -1.68277636e-01 4.90968883e-01
4.58315343e-01 5.42699337e-01 -5.54092526e-01 2.35148177e-01
-8.93220231e-02 1.03415596e+00 -1.58762372e+00 -1.02492452e+00
5.28070629e-01 9.70239341e-02 -3.85510653e-01 -2.44005039e-01
-3.10603559e-01 -4.16217059e-01 -1.91076636e-01 7.70184815e-01
2.76357889e-01 -3.81758720e-01 -7.91267604e-02 8.39850400e-03
1.64866075e-01 5.70971012e-01 -9.61452067e-01 1.42306471e+00
-9.84534696e-02 1.06589973e+00 -5.50783992e-01 -6.73968077e-01
7.59308934e-01 2.72007167e-01 1.61175609e-01 -6.38827384e-01
6.09687343e-02 2.99697429e-01 4.14024740e-01 -6.45764649e-01
7.67776549e-01 -8.47937047e-01 -1.30973712e-01 5.76898217e-01
-2.76962727e-01 -1.36297107e+00 5.38433790e-01 3.25351775e-01
1.27477515e+00 -1.39298201e-01 7.00369656e-01 -5.69920480e-01
2.68725604e-01 8.94889116e-01 3.29098612e-01 8.94825101e-01
-8.06567132e-01 1.95072219e-01 3.58402461e-01 -6.23666286e-01
-9.88167584e-01 -1.13928092e+00 5.33894524e-02 1.10197198e+00
6.20225549e-01 -3.37867677e-01 -6.72180116e-01 -8.73945802e-02
-1.66965485e-01 1.88700199e+00 -6.64989829e-01 -4.27473128e-01
-6.29005507e-02 -3.52295846e-01 1.09054053e+00 5.29429078e-01
7.27147698e-01 -1.75650263e+00 -1.40353060e+00 -5.92759922e-02
-4.90032911e-01 -1.45803714e+00 1.05575718e-01 -1.82720404e-02
-7.69484162e-01 -1.23609924e+00 1.71458900e-01 -4.43396449e-01
1.05756693e-01 5.13990521e-01 1.67768788e+00 3.23232144e-01
-2.69517690e-01 6.50968254e-01 -5.17095983e-01 -7.55575359e-01
-7.92381346e-01 -8.58248115e-01 -9.15803388e-02 -1.10418034e+00
9.38833416e-01 -4.03082788e-01 -7.57358270e-03 1.22744583e-01
-9.28935409e-01 1.26863092e-01 7.83099309e-02 8.18200350e-01
1.99194625e-01 3.43751609e-01 3.52703691e-01 -6.72167718e-01
1.13246477e+00 -4.25694853e-01 9.38671678e-02 2.36871034e-01
-5.30084908e-01 -2.23490119e-01 4.04784709e-01 -2.46609598e-01
-9.55391467e-01 -6.62153304e-01 3.68305266e-01 -1.26199663e-01
-3.29113603e-01 3.83693099e-01 1.64187327e-01 8.87527466e-02
1.25534379e+00 2.33863369e-01 -1.12909369e-01 5.05546033e-01
3.24550629e-01 5.98678172e-01 7.21699715e-01 -1.14229393e+00
3.80636156e-01 5.64450979e-01 -2.96266288e-01 -8.21449220e-01
-9.88215506e-01 -1.79324538e-01 -1.57650054e-01 -2.62523264e-01
8.35973620e-01 -5.25531828e-01 -5.11502624e-01 1.46346882e-01
-1.33537376e+00 -4.35334116e-01 -9.48267639e-01 3.36147308e-01
-1.28843498e+00 4.12586272e-01 -6.51584268e-01 -7.12772608e-01
-3.37115496e-01 -1.04843712e+00 6.20335340e-01 1.15872860e-01
-1.14441776e+00 -9.28448617e-01 -2.24405602e-02 6.15058959e-01
4.48013365e-01 6.27701759e-01 1.03717136e+00 -5.95359862e-01
1.05003417e-01 1.72006711e-02 -3.28029364e-01 3.22040915e-01
2.14827627e-01 3.13937604e-01 -9.11691010e-01 4.22345221e-01
6.85975626e-02 -8.79328728e-01 3.57052982e-01 3.45972091e-01
1.03539670e+00 -1.72193930e-01 -1.49633596e-03 1.35055184e-01
1.30619848e+00 5.10704339e-01 1.00386858e+00 8.48177731e-01
1.02886252e-01 3.81940663e-01 7.68724203e-01 4.38196063e-01
4.60724205e-01 8.17625374e-02 4.18936342e-01 3.90192866e-01
-1.49283499e-01 7.83950165e-02 2.88302571e-01 4.98927414e-01
-6.54940963e-01 -1.80855900e-01 -1.25913572e+00 9.15171325e-01
-1.66525877e+00 -1.86606169e+00 -1.17093235e-01 1.47470593e+00
1.18622625e+00 3.94977629e-01 -7.31141418e-02 3.99977565e-01
7.82964885e-01 1.74993008e-01 -4.71442699e-01 -9.44669366e-01
-4.57852483e-01 9.58281010e-02 -2.27901593e-01 5.07001996e-01
-7.71046162e-01 1.22589719e+00 7.66823626e+00 6.06624961e-01
-5.42243361e-01 -6.19855374e-02 1.54106244e-01 -1.38684481e-01
-5.07728636e-01 5.10775335e-02 2.46674698e-02 2.88426906e-01
7.62719393e-01 -7.32769072e-01 6.52370632e-01 1.04786348e+00
6.33186400e-02 -1.65002838e-01 -1.15335977e+00 1.13856506e+00
3.53537232e-01 -1.51547503e+00 2.58008838e-01 -3.00526261e-01
7.72682250e-01 1.72200520e-02 -2.42537469e-01 3.54395896e-01
9.01790023e-01 -1.27853096e+00 1.08616483e+00 4.62825119e-01
4.69153553e-01 -6.70204699e-01 6.43875003e-01 1.76449582e-01
-5.00110805e-01 1.34557396e-01 -5.52314460e-01 -6.11382008e-01
-1.15768380e-01 5.95400572e-01 -3.51293325e-01 -2.03539189e-02
8.23041260e-01 6.07215881e-01 -3.99161756e-01 6.41732991e-01
-2.28972793e-01 3.37162822e-01 -1.97984710e-01 -5.46666205e-01
3.14765960e-01 2.43969679e-01 6.81396425e-01 1.43281746e+00
-1.63017341e-03 6.01584733e-01 -6.37703091e-02 1.50983620e+00
1.51683107e-01 -3.21626633e-01 -8.72525930e-01 -4.95839626e-01
7.51476645e-01 6.23280346e-01 -7.75494754e-01 -8.15994918e-01
-1.34763688e-01 9.72338080e-01 -1.69370800e-01 -4.70318384e-02
-1.18493652e+00 -6.58452213e-01 8.92802775e-01 -1.43266380e-01
-5.30133188e-01 -1.04535051e-01 -8.07726562e-01 -1.10134768e+00
-4.75501604e-02 -1.46425641e+00 4.90804851e-01 -1.57437849e+00
-1.58531082e+00 4.34013665e-01 4.38817948e-01 -7.47380793e-01
-4.18088108e-01 -6.89573586e-01 -6.49004281e-01 4.53772455e-01
-1.06592882e+00 -9.98772085e-01 -7.87438571e-01 6.86110318e-01
5.12759209e-01 -3.97364162e-02 1.09007502e+00 -5.19254506e-01
1.57363564e-01 1.14596300e-01 -3.56084585e-01 5.13107739e-02
5.92207372e-01 -1.20281982e+00 3.59240115e-01 4.41691458e-01
-1.20308615e-01 9.74374056e-01 1.30268896e+00 -7.75666595e-01
-1.04796755e+00 -4.97431040e-01 7.61930645e-01 -7.19538987e-01
1.03494251e+00 2.94184566e-01 -6.50638759e-01 9.40639675e-01
4.70313817e-01 -6.77901879e-02 9.12637889e-01 -1.43089090e-02
-5.79315722e-01 3.88012975e-01 -1.63865876e+00 9.07784045e-01
1.33981037e+00 -6.99742913e-01 -1.52830839e+00 4.30108100e-01
6.41456366e-01 -2.53370881e-01 -7.21117496e-01 -3.48205231e-02
5.55736661e-01 -1.37033403e+00 9.41743791e-01 -8.29370022e-01
1.22416663e+00 -3.16617131e-01 -6.47361994e-01 -1.25891340e+00
-3.77877414e-01 -3.76727521e-01 -2.03183666e-02 9.05490160e-01
3.36707160e-02 -7.17509985e-01 3.37663144e-01 1.02012074e+00
-1.12509191e-01 -3.04416776e-01 -6.98849201e-01 -1.12241924e+00
3.09012324e-01 -8.83927107e-01 7.12294996e-01 1.63116562e+00
7.68923402e-01 3.73109907e-01 2.61163265e-01 -3.88508230e-01
7.77876318e-01 1.62261695e-01 4.86849457e-01 -1.06007802e+00
-1.31737307e-01 -9.23192859e-01 -1.02466953e+00 -4.01762038e-01
2.94847488e-01 -6.22297168e-01 -2.78256405e-02 -1.70779347e+00
6.86718464e-01 -1.06632710e-01 -1.52777478e-01 6.25804007e-01
-4.10942696e-02 2.15653390e-01 3.34512681e-01 5.26714586e-02
-7.26658881e-01 1.27129704e-01 1.40766954e+00 -3.59753847e-01
2.25211173e-01 -1.00624681e+00 -1.13274312e+00 1.05989790e+00
1.06886399e+00 -1.14594616e-01 -6.34699285e-01 -4.03587103e-01
3.89914542e-01 -4.85483617e-01 9.59086299e-01 -1.29255378e+00
2.22778082e-01 -1.08179820e+00 3.03349584e-01 -2.74552405e-01
3.05551618e-01 -7.78817415e-01 1.11201880e-05 4.61934239e-01
-5.37156701e-01 1.62822857e-01 4.13362533e-01 -1.20180636e-03
-1.86235696e-01 -4.13459480e-01 1.03890634e+00 -5.75578392e-01
-1.27443552e+00 -7.15216637e-01 -3.48952442e-01 7.47495294e-01
1.30292654e+00 -7.16025651e-01 -5.78641653e-01 -4.06197965e-01
-3.92377645e-01 -1.22490063e-01 8.94239068e-01 3.00218284e-01
7.04300046e-01 -1.25774908e+00 -6.54055119e-01 -5.85020363e-01
3.88785571e-01 -4.40793782e-01 -2.96318941e-02 4.15060699e-01
-9.54325914e-01 3.71505201e-01 -7.05001891e-01 -1.58392921e-01
-1.19497275e+00 7.60015666e-01 4.51113790e-01 2.56471366e-01
-7.44562089e-01 6.26720488e-01 -1.20819293e-01 -1.89035147e-01
-1.30026385e-01 -5.34919441e-01 6.14305213e-02 -6.00537479e-01
7.99772441e-01 7.00728714e-01 -3.76402467e-01 -6.55634403e-01
-6.77128732e-01 1.88732594e-01 2.52748817e-01 -2.35528514e-01
1.40295935e+00 9.14524570e-02 -5.65598845e-01 8.25707972e-01
4.91086662e-01 -2.64427125e-01 -2.80740231e-01 2.98925489e-01
-2.61366665e-01 -5.11231005e-01 -3.92536726e-03 -1.18800437e+00
-3.50192934e-01 7.20173240e-01 1.68578401e-01 4.85459656e-01
1.04298627e+00 6.21802919e-02 6.69193327e-01 7.38842368e-01
6.88891113e-01 -1.51143563e+00 -8.27225484e-03 8.12739074e-01
1.20993555e+00 -1.12028182e+00 2.38245919e-01 -3.71970206e-01
-8.68835926e-01 1.12002921e+00 6.53069198e-01 -3.55351299e-01
3.06188822e-01 3.72220635e-01 -2.47807670e-02 -8.12259614e-01
-6.26914024e-01 -1.38681859e-01 -4.14501131e-02 9.77054238e-01
7.90911615e-01 3.56165797e-01 -2.65593886e-01 2.87087083e-01
-8.76619041e-01 3.79781425e-01 8.63129854e-01 1.24835622e+00
-7.02882588e-01 -2.35427588e-01 -6.50428236e-01 3.93099934e-01
-1.75384328e-01 -3.08103204e-01 -1.14730704e+00 1.02315331e+00
2.09271118e-01 1.34497178e+00 -1.70018762e-01 -5.66086829e-01
3.27928364e-01 6.84349984e-02 8.27335358e-01 -6.18611097e-01
-4.54856217e-01 -9.96358156e-01 5.14848232e-01 -5.29578924e-01
-7.27822244e-01 -5.10242641e-01 -1.58506489e+00 -1.15108597e+00
-2.50311524e-01 -3.87482308e-02 1.69559285e-01 1.11272526e+00
-4.68963273e-02 4.90352571e-01 -4.62092221e-01 -3.16229433e-01
-5.64696610e-01 -6.40761912e-01 -5.25483668e-01 8.29059958e-01
6.29714951e-02 -8.12216520e-01 -3.94030660e-01 5.81489742e-01] | [9.954985618591309, 7.895907878875732] |
4d49111c-bbba-43bd-b62c-5b0f208ea375 | joint-edge-model-sparse-learning-is-provably | 2302.02922 | null | https://arxiv.org/abs/2302.02922v1 | https://arxiv.org/pdf/2302.02922v1.pdf | Joint Edge-Model Sparse Learning is Provably Efficient for Graph Neural Networks | Due to the significant computational challenge of training large-scale graph neural networks (GNNs), various sparse learning techniques have been exploited to reduce memory and storage costs. Examples include \textit{graph sparsification} that samples a subgraph to reduce the amount of data aggregation and \textit{model sparsification} that prunes the neural network to reduce the number of trainable weights. Despite the empirical successes in reducing the training cost while maintaining the test accuracy, the theoretical generalization analysis of sparse learning for GNNs remains elusive. To the best of our knowledge, this paper provides the first theoretical characterization of joint edge-model sparse learning from the perspective of sample complexity and convergence rate in achieving zero generalization error. It proves analytically that both sampling important nodes and pruning neurons with the lowest-magnitude can reduce the sample complexity and improve convergence without compromising the test accuracy. Although the analysis is centered on two-layer GNNs with structural constraints on data, the insights are applicable to more general setups and justified by both synthetic and practical citation datasets. | ['Miao Liu', 'Songtao Lu', 'Sijia Liu', 'Pin-Yu Chen', 'Meng Wang', 'Shuai Zhang'] | 2023-02-06 | null | null | null | null | ['sparse-learning'] | ['methodology'] | [ 5.44745445e-01 5.91826975e-01 -3.19493622e-01 -1.86518505e-01
-3.51469368e-01 -1.75259620e-01 8.81513506e-02 2.15670228e-01
-2.88100839e-01 8.72398973e-01 -2.72837192e-01 -3.68912965e-01
-5.40819466e-01 -8.92302334e-01 -9.89990950e-01 -8.08948219e-01
-2.34844297e-01 4.23064470e-01 -5.72971739e-02 2.58588493e-01
1.78722501e-01 5.36451221e-01 -1.46586728e+00 -9.15825367e-02
1.05069911e+00 1.13693821e+00 2.06789345e-01 3.01875293e-01
-7.39029571e-02 8.13301265e-01 -4.10453767e-01 -3.85261178e-01
4.88111734e-01 -3.75704855e-01 -4.47523743e-01 4.70058993e-02
6.01745725e-01 -8.63141716e-02 -7.04894722e-01 1.29986370e+00
5.19244313e-01 1.88312590e-01 3.45466852e-01 -1.18396139e+00
-5.97975075e-01 1.05358100e+00 -5.86564958e-01 3.48770082e-01
-3.89785349e-01 -2.40003452e-01 1.02240694e+00 -5.47686279e-01
5.90772331e-01 7.05826938e-01 7.71236718e-01 2.83645838e-01
-1.32180882e+00 -6.53245866e-01 2.27696776e-01 2.36370340e-01
-1.40920734e+00 -3.58335853e-01 8.52502823e-01 -3.25506628e-02
8.69836748e-01 2.41218179e-01 7.63376415e-01 8.91743481e-01
-1.80600733e-01 6.85475230e-01 8.11898232e-01 -4.30030495e-01
4.55238104e-01 2.11937912e-02 3.08989346e-01 9.16012466e-01
1.14971983e+00 -1.72828704e-01 -6.33837938e-01 -9.84518975e-02
1.03646123e+00 6.50310963e-02 -2.82578468e-01 -5.92431128e-01
-5.24101138e-01 8.71519148e-01 5.08302867e-01 3.01050574e-01
-3.61628175e-01 2.40855470e-01 2.36227974e-01 5.07962167e-01
4.60933298e-01 5.75635314e-01 -3.75512779e-01 -1.93720181e-02
-1.26603651e+00 -1.31518066e-01 1.03540599e+00 1.00379348e+00
9.26186025e-01 6.96608305e-01 2.31754497e-01 7.60853112e-01
-2.88106859e-01 4.18697268e-01 2.03323647e-01 -8.87898147e-01
5.82152545e-01 8.86742830e-01 -5.09885490e-01 -1.15785325e+00
-4.07832146e-01 -1.20164466e+00 -1.23395157e+00 -1.53861493e-01
4.77457106e-01 -3.68157893e-01 -8.07796061e-01 1.70052445e+00
9.10020024e-02 3.81731659e-01 -2.29801238e-01 7.11679459e-01
6.89190626e-01 3.93980265e-01 -1.97520137e-01 -1.58258364e-01
8.02026033e-01 -6.50939286e-01 -3.25027615e-01 -5.71035922e-01
1.00784159e+00 -2.28402793e-01 9.06362891e-01 4.62307572e-01
-1.18108130e+00 -2.42346630e-01 -1.11919129e+00 1.28414657e-03
-1.92263141e-01 2.03329280e-01 1.04980969e+00 7.05238998e-01
-9.97182488e-01 8.12377512e-01 -8.18989217e-01 -2.78314173e-01
8.50462914e-01 6.73233390e-01 -2.12636381e-01 -4.75753248e-01
-8.63897204e-01 5.77299237e-01 4.19916332e-01 1.69729277e-01
-7.09234655e-01 -7.76960790e-01 -6.42759681e-01 5.99686146e-01
6.09991312e-01 -6.68455303e-01 6.33261979e-01 -1.04030299e+00
-1.11436713e+00 4.54596937e-01 5.94070368e-03 -9.01903152e-01
5.64024858e-02 1.62419900e-01 8.57225806e-03 2.83060879e-01
-3.18353415e-01 4.16679800e-01 8.36366415e-01 -9.85443771e-01
-1.95299298e-01 -4.62064147e-01 -5.20726526e-03 1.11484110e-01
-8.84515107e-01 -4.88297999e-01 -3.58528644e-01 -7.62098491e-01
2.92166501e-01 -8.80033612e-01 -4.58478570e-01 -4.73587066e-01
-4.00751323e-01 -2.43919734e-02 7.53192425e-01 -4.79544133e-01
1.33874166e+00 -1.98575675e+00 3.87225360e-01 5.47842562e-01
5.44058204e-01 2.99303561e-01 -3.13564897e-01 3.05798531e-01
-4.92425114e-02 -5.32726236e-02 -3.14394712e-01 -3.37277442e-01
-3.16486776e-01 3.99371088e-01 -2.18186691e-01 5.49865305e-01
1.02993101e-01 1.03762424e+00 -5.26070058e-01 -3.57697546e-01
-8.17004517e-02 4.40490782e-01 -8.07784319e-01 -1.02924973e-01
-1.64300799e-01 -1.27615035e-01 -5.87661803e-01 5.64921498e-01
5.41813612e-01 -8.24066818e-01 5.56554615e-01 -9.57093239e-02
3.32037121e-01 2.74939835e-01 -1.35591066e+00 1.45573962e+00
-4.18534249e-01 5.76527774e-01 4.29177433e-01 -1.84496427e+00
7.63832510e-01 -1.97468735e-02 3.93409759e-01 -6.61074519e-01
1.91067308e-01 2.94328749e-01 1.24694675e-01 -2.54335642e-01
2.89033532e-01 1.01353019e-01 2.07099527e-01 5.55178046e-01
2.63915271e-01 2.53249228e-01 4.69570309e-01 4.07341421e-01
1.11631107e+00 -2.91995555e-01 9.60806087e-02 -4.89321619e-01
6.60398901e-02 -1.18512027e-01 4.23986107e-01 8.83477271e-01
3.52292776e-01 1.76297218e-01 9.01470482e-01 -4.02619690e-01
-1.07675695e+00 -7.42096424e-01 5.22963703e-02 1.04060125e+00
-1.10804133e-01 -4.07398850e-01 -8.00670207e-01 -4.61125791e-01
7.54513517e-02 5.47271252e-01 -6.70503259e-01 -3.29009444e-01
-6.21233165e-01 -8.73729348e-01 4.14951175e-01 4.45356905e-01
2.14889184e-01 -8.03453565e-01 -4.82902467e-01 -1.60820112e-02
2.88049787e-01 -1.10839307e+00 -1.40269101e-01 5.54118812e-01
-1.38534820e+00 -1.12678564e+00 -5.60982883e-01 -9.43570435e-01
1.17293239e+00 2.58151621e-01 9.81616795e-01 3.56409013e-01
-2.21541479e-01 2.37699047e-01 -6.61113709e-02 -4.03240293e-01
1.14246443e-01 5.04144132e-01 -1.20260321e-01 -1.59708276e-01
2.97003835e-01 -1.16962886e+00 -3.17843020e-01 -1.05103098e-01
-9.27393675e-01 1.31246790e-01 8.58123779e-01 7.97846377e-01
7.68253624e-01 2.94186860e-01 7.23783314e-01 -1.13569057e+00
5.99780023e-01 -6.01058006e-01 -6.68369472e-01 1.90944582e-01
-9.27130044e-01 3.42163563e-01 9.58222389e-01 -6.07777417e-01
-5.34616530e-01 -1.83073014e-01 4.08445179e-01 -5.77142358e-01
2.59066850e-01 8.95558298e-01 2.24761710e-01 -4.72138673e-01
5.81547797e-01 5.10856748e-01 4.25882638e-02 -3.93212199e-01
1.39083356e-01 4.65000086e-02 2.04235137e-01 -5.97087383e-01
8.26202810e-01 3.73795450e-01 6.07181370e-01 -9.74515975e-01
-1.08128560e+00 -6.02658577e-02 -3.14672947e-01 5.61333634e-03
1.90043166e-01 -8.90151262e-01 -5.58292389e-01 3.25475298e-02
-5.95053434e-01 -4.79766279e-01 -6.08478844e-01 3.53812844e-01
-1.92943946e-01 4.00075018e-01 -6.07984066e-01 -6.61771655e-01
-4.86848921e-01 -7.95746922e-01 5.98466098e-01 1.16744138e-01
1.66535407e-01 -1.01277113e+00 -4.44313288e-01 1.33708715e-01
4.30437773e-01 8.99738893e-02 1.00624454e+00 -7.78160393e-01
-8.15915763e-01 -3.60889941e-01 -3.89132828e-01 2.88329363e-01
-1.56305358e-01 -4.41246539e-01 -6.71026230e-01 -4.96807337e-01
-5.37474931e-04 -3.78351271e-01 1.03854656e+00 7.78401852e-01
1.38767385e+00 -6.26936197e-01 -2.91007489e-01 8.29175830e-01
1.50739002e+00 -7.15557858e-02 4.37087625e-01 2.06767917e-02
9.67236161e-01 4.61475462e-01 -1.71413511e-01 4.28208798e-01
8.68211165e-02 1.51188731e-01 4.71612781e-01 -1.01162940e-01
-2.24988088e-01 -2.76330262e-01 -9.41862017e-02 9.49216127e-01
-1.96044713e-01 -3.27491999e-01 -7.51379311e-01 5.12786686e-01
-1.69534039e+00 -8.12055051e-01 1.40097544e-01 2.24212742e+00
6.20138049e-01 2.29004681e-01 8.41257870e-02 2.62600511e-01
6.24862313e-01 2.99197584e-01 -7.56888390e-01 -1.63188845e-01
-2.80092180e-01 4.22955573e-01 7.26902068e-01 3.39868397e-01
-6.27197683e-01 6.82185352e-01 5.90999556e+00 9.48968530e-01
-1.11177588e+00 -1.41759422e-02 9.62959111e-01 -6.59844756e-01
-3.74530494e-01 -2.15931628e-02 -8.89099360e-01 1.41919836e-01
1.11400366e+00 -1.98125541e-01 9.68463719e-01 9.92749274e-01
-9.98822004e-02 5.04109003e-02 -1.00594258e+00 9.35225129e-01
1.46834627e-01 -1.67479074e+00 2.19296902e-01 2.14206874e-01
1.03819263e+00 2.99435437e-01 9.77813974e-02 3.04889798e-01
1.46559194e-01 -1.11619401e+00 3.78446907e-01 3.31425637e-01
8.78617346e-01 -8.88105214e-01 6.18953943e-01 4.19455767e-01
-1.06732118e+00 -3.19625080e-01 -6.65543377e-01 -2.63198435e-01
-3.82112823e-02 9.85903084e-01 -6.75204039e-01 3.96027893e-01
4.99104738e-01 6.34984076e-01 -6.30178809e-01 8.90694916e-01
9.17462632e-02 9.34897125e-01 -7.39359260e-01 -2.13748232e-01
2.34304368e-01 -5.46283126e-01 4.46903646e-01 8.48110139e-01
4.98340309e-01 4.25393041e-03 1.07367478e-01 8.21617603e-01
-3.66387874e-01 4.50070426e-02 -6.87876642e-01 -4.90792155e-01
6.05287850e-01 1.08033741e+00 -8.37290287e-01 -1.32305861e-01
-3.40914369e-01 4.10757124e-01 8.75230014e-01 5.84316790e-01
-4.41093236e-01 -3.41478229e-01 1.41745538e-01 5.73491335e-01
5.54933727e-01 -4.37128425e-01 -8.79713416e-01 -9.91766155e-01
1.55763403e-01 -7.31735587e-01 4.88384128e-01 -2.05545291e-01
-9.20076907e-01 4.10153389e-01 -1.63517743e-01 -6.73158526e-01
7.40779517e-03 -4.67965096e-01 -5.27387381e-01 5.48602104e-01
-1.31629682e+00 -9.14448142e-01 -2.23175973e-01 3.73668522e-01
1.49494648e-01 -3.22888762e-01 5.18447459e-01 3.17364603e-01
-9.35544312e-01 8.70511532e-01 1.87145218e-01 -2.24527270e-02
-3.22638154e-02 -8.43461812e-01 5.82872033e-02 1.01039076e+00
2.60962427e-01 8.16985488e-01 6.53245211e-01 -6.59915328e-01
-1.77254963e+00 -9.66401458e-01 6.14365637e-01 1.73059955e-01
7.85599768e-01 -4.57657903e-01 -9.24255669e-01 6.20179534e-01
-2.38747701e-01 1.81476086e-01 5.58799624e-01 4.93623108e-01
-3.53447795e-01 -2.61803776e-01 -9.17332709e-01 7.28325546e-01
1.31236041e+00 -4.16643143e-01 -7.65306801e-02 3.66749942e-01
4.84274507e-01 -2.96774805e-01 -1.03256583e+00 3.48269254e-01
1.88539356e-01 -6.85481668e-01 9.25988793e-01 -6.44766212e-01
3.71642858e-01 1.70177698e-01 -1.04118571e-01 -1.07334101e+00
-2.82568395e-01 -6.03951454e-01 -7.39578307e-01 8.04624438e-01
4.59991306e-01 -6.58485830e-01 1.36586952e+00 4.11844313e-01
-4.93901037e-02 -1.20939338e+00 -1.11496091e+00 -7.69139051e-01
-8.54381323e-02 -2.73352295e-01 2.63376385e-01 9.39871907e-01
-1.26173034e-01 3.58968168e-01 -4.73032266e-01 1.10418953e-01
8.85432541e-01 2.13253036e-01 4.85767335e-01 -1.48529994e+00
-4.55552399e-01 -4.34045643e-01 -2.78847545e-01 -1.02533698e+00
2.70328283e-01 -1.13637674e+00 -6.25871539e-01 -1.40346324e+00
2.26564467e-01 -5.16312659e-01 -3.69343013e-01 4.64213043e-01
-1.32074747e-02 1.16331100e-01 5.82747301e-03 9.39281937e-03
-6.11343622e-01 4.67512190e-01 1.16170537e+00 3.01971696e-02
-1.53326318e-01 -5.73041737e-02 -1.00045824e+00 4.73236233e-01
9.13587689e-01 -6.74688160e-01 -8.74754310e-01 -6.54774904e-01
6.62550807e-01 5.63302115e-02 2.79686630e-01 -1.01354897e+00
4.34263349e-01 -1.79806072e-02 2.14044064e-01 -2.54133195e-01
1.03417270e-01 -7.12149739e-01 6.60080463e-02 6.29617691e-01
-3.34717900e-01 -1.08531803e-01 2.15670884e-01 8.37471068e-01
1.62628125e-02 -3.96951675e-01 8.16456676e-01 -4.76175360e-02
-2.80524969e-01 6.40275419e-01 1.54021000e-02 2.21723408e-01
7.89786458e-01 -3.90235007e-01 -3.99489880e-01 -3.60049993e-01
-6.01115882e-01 9.55824479e-02 3.23408186e-01 -9.27905962e-02
4.16928113e-01 -1.06449592e+00 -4.44062948e-01 4.21043783e-01
-4.18823719e-01 2.88330138e-01 3.94641787e-01 9.74341393e-01
-3.59671235e-01 5.77918947e-01 -5.69301657e-02 -2.60015637e-01
-9.64811742e-01 6.57984614e-01 1.95746288e-01 -5.92672408e-01
-7.31396735e-01 1.02539957e+00 8.38623047e-02 -5.22474274e-02
5.37397742e-01 -1.16522752e-01 8.37360248e-02 -3.95170301e-02
1.96936563e-01 6.15402997e-01 1.49056166e-01 -3.10339164e-02
-7.84155056e-02 2.08209887e-01 -2.27414772e-01 4.96075690e-01
1.81571567e+00 7.06711113e-02 -1.74536735e-01 1.33420035e-01
1.06512296e+00 -1.98844373e-01 -1.28685558e+00 -3.91350836e-01
7.94606283e-02 -7.03506768e-02 3.08502972e-01 -3.94590437e-01
-1.54953313e+00 7.61638582e-01 1.59452304e-01 4.43798661e-01
1.20263934e+00 -9.33119655e-02 7.15573728e-01 9.23294842e-01
4.05121714e-01 -1.21764398e+00 8.10573925e-04 4.92407203e-01
5.82585216e-01 -8.90107512e-01 3.19091678e-01 -4.53812420e-01
-2.68024921e-01 9.19961691e-01 5.98113716e-01 -4.60939437e-01
6.87336504e-01 4.42771226e-01 -6.78826571e-01 -3.65032703e-01
-6.59717679e-01 2.26874053e-01 1.38929978e-01 4.58271474e-01
1.49967879e-01 -1.01976089e-01 -2.78568536e-01 7.34007835e-01
-1.39813989e-01 3.15569192e-02 4.37130600e-01 5.77664971e-01
-4.54088211e-01 -6.97826982e-01 1.78655744e-01 1.21323669e+00
-3.41095686e-01 -4.38304633e-01 -2.14241460e-01 7.31620848e-01
-2.06295490e-01 5.28358698e-01 6.71705082e-02 -1.77052602e-01
1.49166398e-02 -2.78050639e-02 6.37183666e-01 -6.05205834e-01
-3.16256702e-01 -2.54979938e-01 2.98751473e-01 -4.88213599e-01
-2.13762268e-01 -4.10799086e-01 -1.08609676e+00 -4.63314146e-01
-3.84486586e-01 3.13741952e-01 3.88907284e-01 9.40878987e-01
6.42906129e-01 5.83317339e-01 2.71629512e-01 -7.65762091e-01
-8.34604025e-01 -9.75837827e-01 -9.67446387e-01 1.21039152e-01
1.00595914e-01 -5.25311172e-01 -5.48158407e-01 -3.60052377e-01] | [8.388063430786133, 3.555940628051758] |
740a30c5-5b62-4068-ae92-ff8d979422fd | foveation-based-mechanisms-alleviate | 1511.06292 | null | http://arxiv.org/abs/1511.06292v3 | http://arxiv.org/pdf/1511.06292v3.pdf | Foveation-based Mechanisms Alleviate Adversarial Examples | We show that adversarial examples, i.e., the visually imperceptible
perturbations that result in Convolutional Neural Networks (CNNs) fail, can be
alleviated with a mechanism based on foveations---applying the CNN in different
image regions. To see this, first, we report results in ImageNet that lead to a
revision of the hypothesis that adversarial perturbations are a consequence of
CNNs acting as a linear classifier: CNNs act locally linearly to changes in the
image regions with objects recognized by the CNN, and in other regions the CNN
may act non-linearly. Then, we corroborate that when the neural responses are
linear, applying the foveation mechanism to the adversarial example tends to
significantly reduce the effect of the perturbation. This is because,
hypothetically, the CNNs for ImageNet are robust to changes of scale and
translation of the object produced by the foveation, but this property does not
generalize to transformations of the perturbation. As a result, the accuracy
after a foveation is almost the same as the accuracy of the CNN without the
adversarial perturbation, even if the adversarial perturbation is calculated
taking into account a foveation. | ['Xavier Boix', 'Gemma Roig', 'Yan Luo', 'Tomaso Poggio', 'Qi Zhao'] | 2015-11-19 | null | null | null | null | ['foveation'] | ['computer-vision'] | [ 3.64672482e-01 7.42277563e-01 6.47064745e-01 1.58707444e-02
4.34656590e-02 -1.10144615e+00 7.67353714e-01 -3.47090513e-01
-4.42107648e-01 4.39017385e-01 9.04799998e-02 -3.03385139e-01
3.37972730e-01 -8.55437994e-01 -1.43185925e+00 -8.91013265e-01
2.69719027e-02 -2.85558045e-01 4.21346962e-01 -5.47750890e-01
1.18988417e-01 9.00949836e-01 -1.37914443e+00 4.88414735e-01
1.71746954e-01 7.41545260e-01 -4.25650984e-01 9.02232170e-01
3.18645656e-01 9.45181966e-01 -1.03073764e+00 -3.36500674e-01
7.62273252e-01 -3.92570972e-01 -8.17054629e-01 -7.67418891e-02
9.78139281e-01 -3.22949648e-01 -7.43098855e-01 1.39471138e+00
2.39717886e-01 3.24362405e-02 6.61291957e-01 -1.33228171e+00
-1.08310950e+00 3.89061689e-01 -1.39948875e-01 1.32545708e-02
3.60799909e-01 5.94516873e-01 4.65260476e-01 -6.23150349e-01
6.96532786e-01 1.58825707e+00 8.65939379e-01 9.56309140e-01
-1.45462942e+00 -3.84731412e-01 3.41127157e-01 -9.54713598e-02
-1.12471652e+00 -3.73636484e-01 5.66318750e-01 -3.67122829e-01
8.61969233e-01 6.98290229e-01 3.72671455e-01 1.22024834e+00
5.18549681e-01 2.14968711e-01 9.87744272e-01 -4.05493438e-01
2.11394459e-01 2.61474341e-01 -5.32140315e-01 4.13924068e-01
2.97027469e-01 5.66754937e-01 1.36864811e-01 1.79717004e-01
7.95010805e-01 -1.42913461e-01 -5.33332586e-01 -1.65826932e-01
-1.09982026e+00 4.86841589e-01 1.18027806e+00 4.08815980e-01
-3.32075685e-01 4.89384562e-01 3.13874781e-01 7.89646983e-01
2.52227992e-01 1.10684192e+00 -3.77528667e-01 5.39824605e-01
-4.42282110e-01 2.70617992e-01 6.72091126e-01 6.71393156e-01
7.91886985e-01 3.76864344e-01 6.31262138e-02 3.28557938e-03
-1.07963562e-01 6.18836641e-01 5.04312813e-01 -9.58394945e-01
1.69337630e-01 4.80211198e-01 4.92957979e-02 -1.21879387e+00
-4.66725618e-01 -2.41952702e-01 -7.68576026e-01 1.21580243e+00
9.39205289e-01 -2.59780765e-01 -8.62115264e-01 1.98590291e+00
-5.11615090e-02 1.51188657e-01 3.17026913e-01 8.66411746e-01
6.45771027e-01 2.91973025e-01 -8.81450251e-02 9.62513015e-02
9.96705949e-01 -4.11037683e-01 -5.06762683e-01 -3.93885821e-01
5.26075423e-01 -7.29031444e-01 1.29165745e+00 1.24112375e-01
-9.78952944e-01 -6.63284183e-01 -1.29369271e+00 2.29121372e-01
-8.79300117e-01 -5.91599345e-01 4.45675142e-02 5.93063831e-01
-1.17797017e+00 8.21010053e-01 -6.38208091e-01 -5.41468978e-01
2.11703882e-01 3.73449713e-01 -6.98789179e-01 1.39062628e-01
-1.31786251e+00 1.29847360e+00 4.43122178e-01 8.20795298e-02
-9.28097367e-01 -6.07601464e-01 -6.54175282e-01 1.61763951e-01
3.42588127e-02 -4.90629017e-01 9.08726573e-01 -1.84069014e+00
-1.05677629e+00 9.49914873e-01 1.71932787e-01 -7.23840833e-01
9.06923234e-01 -1.26911134e-01 -2.89161652e-01 9.07775909e-02
-2.67955333e-01 7.80408859e-01 1.00820553e+00 -1.62837207e+00
-2.09977943e-02 -3.23566884e-01 5.55505991e-01 1.60257053e-03
-1.05061211e-01 -1.53172269e-01 1.99511856e-01 -6.51926756e-01
-1.51597913e-02 -1.01513875e+00 -2.85444915e-01 3.89472902e-01
-7.44140863e-01 4.79039282e-01 1.15698528e+00 -3.12044889e-01
5.06659985e-01 -2.57648826e+00 -1.30254760e-01 3.17527533e-01
3.74276787e-01 4.58457083e-01 -3.23348999e-01 1.54174089e-01
-8.79321277e-01 6.47001624e-01 -6.44167066e-02 1.39011338e-01
-1.26214534e-01 1.85097158e-01 -6.33479357e-01 8.64692092e-01
5.39510369e-01 1.03362632e+00 -7.77947605e-01 -2.19199844e-02
1.96708813e-01 3.38826984e-01 -4.41976786e-01 6.08130582e-02
-2.70629786e-02 3.09031785e-01 -1.08917579e-01 1.05113007e-01
8.42041850e-01 2.67724425e-01 -2.94867516e-01 -2.04026043e-01
-8.07024911e-02 -2.59686023e-01 -8.14973414e-01 8.82647812e-01
-1.19938254e-01 1.05647278e+00 -1.57303780e-01 -8.92415881e-01
6.62157953e-01 3.73062074e-01 -4.21587192e-02 -7.04806626e-01
1.26847997e-01 1.57283589e-01 4.73337054e-01 -4.93698537e-01
1.04871318e-01 -2.75052249e-01 8.90902281e-02 2.70558417e-01
-1.58196166e-01 -3.27333152e-01 -1.91404581e-01 1.68281779e-01
1.36117160e+00 -2.04691738e-02 2.44897351e-01 -1.69651881e-01
4.38656479e-01 2.18576305e-02 2.15515301e-01 1.12125564e+00
-2.57864565e-01 7.43167043e-01 7.35505641e-01 -6.88122809e-01
-1.35416937e+00 -1.12714839e+00 5.66798300e-02 5.34684122e-01
1.93528935e-01 2.49010548e-01 -1.14228463e+00 -7.88208783e-01
2.30779499e-01 6.20990515e-01 -1.21245456e+00 -8.99084270e-01
-5.63522637e-01 -2.87448734e-01 9.35749710e-01 4.44550276e-01
6.37555957e-01 -1.28985441e+00 -6.02669716e-01 -1.82404891e-01
3.17370534e-01 -9.74336565e-01 -1.98132321e-01 1.73395231e-01
-6.23342693e-01 -1.27993369e+00 -6.03711843e-01 -6.33837700e-01
1.05535340e+00 -2.88697593e-02 9.05463040e-01 4.04025614e-01
-1.23136804e-01 2.21713588e-01 -1.90583661e-01 -5.84263206e-01
-8.92420650e-01 -3.39486897e-01 2.43565813e-01 -1.17585689e-01
-2.70856060e-02 -6.82586551e-01 -5.18414319e-01 3.66044968e-01
-1.14439702e+00 -3.99814159e-01 5.23255110e-01 6.17281556e-01
1.11406386e-01 8.12278241e-02 3.30938280e-01 -9.38642263e-01
4.84304637e-01 -1.48723885e-01 -3.35363507e-01 4.67049405e-02
-2.25485831e-01 8.79696459e-02 1.17552710e+00 -9.63187873e-01
-6.51110470e-01 1.71207577e-01 3.07255071e-02 -6.02176905e-01
-5.42476833e-01 -2.02634763e-02 -3.35513383e-01 -6.02122784e-01
1.54449546e+00 5.65694906e-02 -7.33953044e-02 -4.16812077e-02
5.29961646e-01 1.93512738e-01 7.94816256e-01 -5.59108630e-02
1.50252748e+00 7.92898476e-01 2.07606003e-01 -6.31750464e-01
-4.04529124e-01 2.85197437e-01 -5.40210485e-01 -4.25537944e-01
8.88893545e-01 -4.66107726e-01 -7.04490721e-01 5.66396356e-01
-1.44057596e+00 -4.71144319e-01 -7.12383032e-01 1.66582957e-01
-6.67489827e-01 1.59194842e-01 -3.10075551e-01 -7.70131171e-01
2.62106895e-01 -1.12344897e+00 4.32187974e-01 1.46799237e-01
-2.64908165e-01 -9.75275159e-01 -3.84458959e-01 -5.15426397e-01
5.95719099e-01 8.16174209e-01 9.99830425e-01 -8.88539672e-01
-4.11834955e-01 -4.75042462e-01 -1.66229233e-01 6.80833042e-01
2.32955799e-01 1.26312539e-01 -1.39152753e+00 -2.90281385e-01
1.22612059e-01 -1.47286251e-01 6.99084580e-01 2.65684307e-01
9.20174897e-01 -8.26751173e-01 -1.40808135e-01 6.11577213e-01
1.44076478e+00 1.77634642e-01 1.13794100e+00 5.68129182e-01
6.63928807e-01 5.84648371e-01 1.87974907e-02 -2.17875943e-01
-6.27238631e-01 5.00264049e-01 1.15633404e+00 -6.80221438e-01
-3.50672364e-01 -2.04706565e-01 5.30830681e-01 -1.86656490e-01
-4.79069149e-06 -2.63604194e-01 -6.91983223e-01 3.95698100e-01
-1.60903037e+00 -1.12583613e+00 -2.46285379e-01 2.30737591e+00
6.52439833e-01 5.38127363e-01 -1.23187125e-01 2.09534928e-01
8.33445489e-01 1.09832652e-01 -7.05112517e-01 -7.19016969e-01
-4.79620934e-01 2.64185481e-02 9.58234072e-01 6.18278980e-01
-1.01530588e+00 1.06097353e+00 7.38475418e+00 3.43463361e-01
-1.47461092e+00 -1.95781052e-01 5.20670414e-01 -2.66327541e-02
-1.27146795e-01 -1.84327438e-01 1.08101852e-01 2.77037233e-01
6.90213561e-01 -2.29196489e-01 6.55986547e-01 8.18464875e-01
2.74869293e-01 1.64692760e-01 -1.34231472e+00 3.58466387e-01
-1.25152528e-01 -1.21123350e+00 2.99191356e-01 3.29301208e-02
7.66517818e-01 -2.27777362e-01 5.66177189e-01 6.52343929e-02
4.07841861e-01 -1.30018735e+00 8.77936244e-01 7.28304088e-01
6.50661945e-01 -7.11035430e-01 8.05436134e-01 3.23674440e-01
-5.74247837e-01 -5.11358939e-02 -4.68033671e-01 -2.51275659e-01
-3.55494678e-01 1.02118373e-01 -8.72624755e-01 -1.54511891e-02
5.11706233e-01 4.31500711e-02 -7.77086258e-01 8.25072706e-01
-4.17227507e-01 3.88647109e-01 -6.60628304e-02 2.88797766e-01
2.55165905e-01 3.17026556e-01 8.76757145e-01 1.05849183e+00
-8.56402889e-02 -1.85248420e-01 -4.00726050e-01 1.09370601e+00
-9.21332389e-02 -2.36114085e-01 -1.22706234e+00 1.67417973e-01
2.71800905e-01 8.42833102e-01 -4.94520277e-01 -2.25265145e-01
-3.81982476e-01 1.19098079e+00 1.28381386e-01 9.14650261e-01
-7.17019737e-01 -5.77280521e-01 9.38268423e-01 3.83815438e-01
1.28839582e-01 1.87867865e-01 -4.76914376e-01 -7.39742398e-01
9.73965228e-02 -1.01591587e+00 -3.23386312e-01 -1.25482905e+00
-9.99546468e-01 7.07425594e-01 -2.43707433e-01 -1.27671385e+00
-2.02313140e-01 -9.09632266e-01 -9.74552214e-01 1.05929911e+00
-9.75736558e-01 -8.87413383e-01 -1.88595414e-01 7.70253181e-01
2.02667210e-02 -2.64136288e-02 9.26445663e-01 -1.87026426e-01
-1.30000427e-01 8.49371850e-01 -2.68336922e-01 4.38463688e-01
7.90991366e-01 -1.43445528e+00 6.78590536e-01 1.24470246e+00
9.00582820e-02 8.08596253e-01 1.32842612e+00 -3.91543359e-01
-9.40626800e-01 -1.36260164e+00 5.89451253e-01 -6.22260511e-01
7.10327268e-01 -3.09353560e-01 -1.05561888e+00 1.02127337e+00
3.32778275e-01 2.34811276e-01 -2.12071445e-02 -3.28148842e-01
-6.70376003e-01 -2.17180289e-02 -1.48417580e+00 1.04628229e+00
8.87012005e-01 -7.46946692e-01 -6.65308177e-01 4.09563690e-01
8.78457844e-01 -4.32413816e-01 -4.34682190e-01 2.16514423e-01
5.29812753e-01 -1.06695998e+00 1.09439933e+00 -1.07939255e+00
5.67416549e-01 -4.64933276e-01 -6.39206097e-02 -1.59139442e+00
-2.18731552e-01 -4.62456822e-01 2.68915981e-01 8.26494694e-01
6.02858305e-01 -9.93755043e-01 4.71808285e-01 6.59358025e-01
-7.21881166e-02 -2.50597477e-01 -9.67460752e-01 -1.01321447e+00
4.10372525e-01 -2.99229622e-01 4.94551510e-01 1.05737913e+00
-4.07411903e-02 -2.38057226e-01 -1.98201567e-01 5.70523381e-01
1.33126915e-01 -7.25108504e-01 8.36352408e-01 -8.70297790e-01
-2.39759073e-01 -5.39690375e-01 -1.05157638e+00 -5.92713058e-01
2.26614669e-01 -4.57627475e-01 1.72966331e-01 -9.82079327e-01
-2.91435152e-01 1.82767194e-02 -1.39204472e-01 7.95239151e-01
-1.56820327e-01 5.86848199e-01 4.46917504e-01 1.97667256e-01
1.58524550e-02 -1.08919300e-01 1.50181699e+00 -2.68895179e-01
9.91812944e-02 6.03147689e-03 -7.99410105e-01 1.00934553e+00
7.64585197e-01 -6.51096404e-01 -1.45708397e-01 -3.19322795e-01
3.02234739e-01 -4.46094662e-01 8.86154592e-01 -1.03408921e+00
1.18797056e-01 -7.97250867e-03 6.42244279e-01 2.54470855e-01
2.32365847e-01 -1.01802301e+00 9.53533351e-02 8.68090808e-01
-5.09169161e-01 1.08044751e-01 5.51793456e-01 4.61667836e-01
-2.44188607e-02 -2.15028167e-01 1.10112083e+00 -3.11862916e-01
-5.13792038e-01 -7.68690482e-02 -7.30339825e-01 -8.65739807e-02
9.02320325e-01 -3.82368267e-01 -6.59215152e-01 -6.38572514e-01
-1.04274738e+00 -3.73107910e-01 8.50098848e-01 3.06522995e-01
4.68973279e-01 -1.35098314e+00 -6.11700714e-01 3.16434890e-01
-9.52003300e-02 -9.99038294e-02 7.49042816e-03 6.18696272e-01
-6.51732206e-01 1.06224708e-01 -4.84111369e-01 -3.00363302e-01
-1.18877554e+00 1.00159216e+00 1.00918210e+00 2.30648413e-01
-3.76836628e-01 8.06187391e-01 7.73358524e-01 -1.73652709e-01
1.33773491e-01 -3.94078642e-01 1.34670651e-02 -4.81505632e-01
5.01796782e-01 -1.16562121e-01 8.46684724e-02 -6.13588035e-01
-3.59092087e-01 3.92133534e-01 -3.27460393e-02 -4.19906341e-02
8.26560318e-01 1.73737228e-01 2.23380685e-01 3.41067314e-01
1.14796019e+00 9.68618318e-02 -1.36213589e+00 2.00036149e-02
-4.43276256e-01 -3.79020244e-01 -3.08728278e-01 -8.43885601e-01
-1.16364884e+00 8.14736962e-01 6.35114431e-01 7.01095879e-01
1.11269665e+00 -3.88614833e-02 -4.73136529e-02 4.59693104e-01
-1.16432466e-01 -5.89546621e-01 6.50195107e-02 4.68434274e-01
1.42042172e+00 -8.49056423e-01 -2.54299492e-01 -2.71385849e-01
-4.23459500e-01 1.15151560e+00 7.50759840e-01 -7.63530254e-01
4.08456028e-01 2.89800942e-01 3.66631210e-01 -1.09845683e-01
-4.99028087e-01 -1.18739724e-01 1.45386726e-01 9.71785784e-01
6.88634589e-02 -2.68284380e-01 7.47962445e-02 -4.62940484e-02
-4.75297689e-01 -4.94954824e-01 7.40635574e-01 6.14184558e-01
-3.38131726e-01 -5.42679012e-01 -7.39382505e-01 -1.09430335e-01
-3.45760018e-01 -1.72447592e-01 -1.02610278e+00 1.41626430e+00
4.37531084e-01 6.93829298e-01 2.08672374e-01 -5.63313603e-01
7.17727721e-01 -2.57829521e-02 4.51243967e-01 -4.48394537e-01
-8.45251977e-01 -4.58814502e-01 -2.73466915e-01 -7.43735433e-01
-1.35521367e-01 -4.20010895e-01 -1.09880757e+00 -4.84241664e-01
-1.76046640e-01 -3.24444354e-01 4.74680364e-01 1.00878501e+00
-6.17764667e-02 7.66059816e-01 6.91510916e-01 -9.30158198e-01
-6.51152432e-01 -1.05223656e+00 -5.16841829e-01 8.27445686e-01
7.32121348e-01 -3.07562768e-01 -1.08040822e+00 2.31316745e-01] | [5.647904396057129, 7.868929386138916] |
e36bdaf9-c8cf-4212-a8ec-912c83788e20 | mangngalapp-an-integrated-package-of | 2301.02893 | null | https://arxiv.org/abs/2301.02893v1 | https://arxiv.org/pdf/2301.02893v1.pdf | MangngalApp -- An integrated package of technology for COVID-19 response and rural development: Acceptability and usability using TAM | The COVID19 pandemic has challenged universities and organizations to devise mechanisms to uplift the well-being and welfare of people and communities. In response, the design and development of an integrated package of technologies, MangngalApp -- A web-based portal and mobile responsive application for rural development served as an opportunity. It showcases different packets of technologies that were outputs of R&D in the field of fisheries and aqua-culture, innovations that were IP-protected, and technologies that harness locally available resources for post-harvest development and aiding in sustaining growth and development in the communities. This paper focused on the usability and acceptability of the MangngalApp implementing a descriptive research design using the Technology Acceptance Model or TAM and ISO 25010 software quality standards. Constrained by government health restrictions due to COVID-19, a Google form-based questionnaire was forwarded to consented participants via an email with the attached consent and evaluation form. Results revealed that the MangngalApp was found to be very acceptable and usable, and compliant to ISO 25010 software quality characteristics to the higher extent. From the results, it is concluded that the developed MangngalApp will be a usable and responsive technology that aids to rural development especially among target users: fishers, gatherers, processors, traders, and farmers. Considering compatibility and usefulness, the MangngalApp is expected to provide greater social development in the community. | ['Jesty S. Agoto', 'James Karl A. Agpalza', 'Leo P. Paliuanan', 'Billy S. Javier'] | 2023-01-07 | null | null | null | null | ['culture'] | ['speech'] | [-2.97214717e-01 2.36252517e-01 3.12521905e-02 1.33501306e-01
4.96059749e-03 -8.36056411e-01 6.18322566e-02 3.04995030e-01
-3.63095134e-01 3.37729335e-01 4.70194340e-01 -8.23410511e-01
-2.98528284e-01 -7.47939825e-01 -4.06605661e-01 -5.46285808e-01
-2.00085230e-02 -3.69254440e-01 5.58957122e-02 -6.29979670e-01
6.43838644e-01 3.28123242e-01 -1.31409431e+00 -1.64472610e-01
7.60993302e-01 3.64602596e-01 5.44108748e-01 3.59528720e-01
-1.07095623e-02 3.47753376e-01 2.95280293e-02 -3.97709548e-01
2.99922109e-01 -1.16184250e-01 -5.61351240e-01 -6.07517958e-01
-5.12303889e-01 -1.07332587e+00 4.78093207e-01 7.23266780e-01
6.78193629e-01 -4.90919948e-01 2.92013526e-01 -9.83122170e-01
-7.30481625e-01 4.52850133e-01 -4.09719288e-01 -2.60877579e-01
4.42651868e-01 2.92878866e-01 2.05133125e-01 -1.00467253e+00
6.27012849e-01 1.04836202e+00 9.67220724e-01 1.72407195e-01
-8.52431536e-01 -1.05339730e+00 -6.72727704e-01 -3.25786948e-01
-1.19944882e+00 -4.99000132e-01 1.85710326e-01 -7.28951693e-01
1.05155730e+00 2.35318452e-01 1.07385254e+00 3.26943815e-01
7.22393036e-01 -2.39916563e-01 1.38565850e+00 -6.03190184e-01
3.98667634e-01 4.45562899e-01 -4.89141084e-02 1.43360645e-01
1.00341332e+00 -1.45770133e-01 -3.12700868e-01 -1.51630580e-01
6.19521976e-01 9.59161818e-02 4.75708432e-02 1.97834909e-01
-7.23615408e-01 7.15975046e-01 -4.57286239e-02 4.00640875e-01
-8.32049131e-01 2.93553472e-02 4.04225171e-01 3.23927402e-01
5.97969770e-01 -8.39823857e-02 -7.45993495e-01 -6.37024820e-01
-2.57794023e-01 -9.58079621e-02 9.71730769e-01 5.78395545e-01
6.39237881e-01 -5.97606227e-02 3.64291191e-01 2.74021924e-01
1.24429345e+00 1.13139093e+00 -1.21578895e-01 -1.02345359e+00
-1.34080648e-01 7.70640135e-01 4.76698756e-01 -1.20990789e+00
-2.10513175e-01 -6.83651306e-03 1.17125758e-03 1.09141991e-01
9.26562697e-02 -7.42442429e-01 -5.32613933e-01 1.31935096e+00
4.75008756e-01 -6.20911300e-01 3.05072457e-01 5.88130832e-01
8.17765117e-01 8.42613935e-01 6.24449372e-01 -2.21888602e-01
1.28632045e+00 2.24174663e-01 -1.12791669e+00 7.09310621e-02
5.82346618e-01 -6.43725932e-01 1.04419959e+00 -1.49288625e-01
-8.67440879e-01 3.79757918e-02 -8.30326796e-01 6.96455300e-01
-4.69696939e-01 -3.16508412e-01 5.89950085e-01 1.35685289e+00
-1.49165642e+00 2.44948551e-01 -8.37483943e-01 -1.29035044e+00
9.43709910e-02 2.52499402e-01 -6.30563319e-01 -4.08279561e-02
-6.27504408e-01 1.11560214e+00 4.65419963e-02 3.89278471e-01
-4.10506308e-01 -5.52696586e-01 -5.58338165e-01 -1.88481271e-01
-3.68857384e-01 -2.70439178e-01 9.57218826e-01 -7.95798004e-01
-1.39498603e+00 6.13222241e-01 4.36389714e-01 1.82594940e-01
9.36479345e-02 -3.93308163e-01 -3.23112398e-01 2.85246074e-01
6.60114706e-01 2.71681935e-01 -2.90985942e-01 -1.01471281e+00
-6.66355610e-01 -6.14125252e-01 -3.91591609e-01 6.69068992e-02
-5.89745760e-01 7.52647519e-01 2.62703836e-01 -2.10459009e-01
2.06763163e-01 -5.26131094e-01 -2.34760821e-01 -1.65948689e-01
3.43125463e-01 8.86537954e-02 9.66514587e-01 -1.11230111e+00
1.00340366e+00 -2.06587601e+00 -1.06109560e+00 3.42073381e-01
-3.40227753e-01 4.04337049e-01 -8.12763348e-02 1.61667991e+00
6.38576746e-01 6.66195154e-01 1.82253614e-01 7.47934639e-01
-7.46301934e-02 2.13908032e-01 3.47512245e-01 4.54219490e-01
-3.70565206e-02 3.85804921e-01 -1.11417723e+00 -1.23831406e-01
1.02158435e-01 6.60032332e-01 -2.08319962e-01 -2.56505050e-02
5.25352716e-01 1.89866856e-01 -6.38857245e-01 9.89977539e-01
1.13768148e+00 4.24909711e-01 3.58210623e-01 3.32510412e-01
-9.28154826e-01 -1.00651182e-01 -8.19600344e-01 1.18700349e+00
-3.39197636e-01 5.10870695e-01 5.30185401e-01 -3.15035224e-01
1.14389443e+00 6.62289202e-01 5.91653645e-01 -8.32407117e-01
1.01356737e-01 6.38615310e-01 -2.07411662e-01 -1.36813283e+00
4.94121969e-01 -6.11552000e-02 3.37062508e-01 3.62111121e-01
-3.85297805e-01 2.66238272e-01 -3.11683953e-01 3.71018834e-02
8.83739352e-01 6.42234385e-01 5.18987060e-01 -7.45593488e-01
-1.80605069e-01 5.07917292e-02 3.80593777e-01 1.02374181e-01
-3.01283360e-01 -3.37822318e-01 9.52269658e-02 -8.21714699e-02
-6.44103467e-01 -6.04699075e-01 -2.65919089e-01 1.13691950e+00
7.76551217e-02 -1.24327265e-01 -7.54102826e-01 -7.74165690e-02
-2.02293560e-01 4.74855840e-01 -2.70205051e-01 3.46284091e-01
1.16660237e-01 -3.55200708e-01 4.70556736e-01 8.87948796e-02
7.08491147e-01 -1.02565181e+00 -1.52155292e+00 5.37461519e-01
1.55402571e-02 -6.41183138e-01 1.33498982e-01 -7.39307553e-02
-1.04159296e+00 -7.81532347e-01 -7.42879331e-01 -8.54927301e-01
5.94769597e-01 5.57223499e-01 4.24523175e-01 1.89775992e-02
1.63063914e-01 5.36683440e-01 -8.38655233e-01 -7.85206914e-01
-3.02626014e-01 -5.06860971e-01 -1.99376419e-01 -4.94904935e-01
6.87889159e-01 -7.50752032e-01 -9.09395218e-01 1.39469698e-01
-8.25344086e-01 -1.58975586e-01 6.52389228e-01 -3.84961516e-02
-9.23682153e-02 -3.07837337e-01 1.30295575e+00 -4.74161267e-01
5.34267545e-01 -1.11751914e+00 -1.52395830e-01 -6.18365407e-03
-7.75314331e-01 -8.59735787e-01 -1.88858181e-01 -1.76082686e-01
-9.99963760e-01 -6.19350262e-02 -4.00523067e-01 8.53336990e-01
-5.95856756e-02 1.14520371e+00 -3.36325437e-01 -3.05291623e-01
5.88991046e-01 -4.30069178e-01 5.50464451e-01 -3.67776752e-01
8.40012282e-02 1.29577398e+00 2.01935470e-01 -1.67050257e-01
3.53532523e-01 2.45940126e-02 -3.29157114e-01 -1.15463257e+00
5.08103907e-01 -5.57518840e-01 1.60346463e-01 -8.32878947e-01
7.38584638e-01 -1.11409366e+00 -7.12462842e-01 3.76024038e-01
-7.59737074e-01 -5.31656265e-01 2.04939932e-01 6.82106256e-01
1.41837656e-01 2.48328838e-02 -2.77518153e-01 -1.38279402e+00
-6.78632081e-01 -5.42972326e-01 2.58625865e-01 5.72266161e-01
7.40804430e-03 -7.17381239e-01 3.65567982e-01 1.45628452e-01
1.11013067e+00 8.29754293e-01 5.95784068e-01 -3.11291188e-01
-5.46052828e-02 -2.78276742e-01 -1.78058058e-01 1.14875875e-01
4.09485966e-01 3.20940375e-01 -6.25120699e-01 -2.00417653e-01
-1.15455300e-01 -2.71414546e-03 -2.91874975e-01 2.89258003e-01
-2.38525972e-01 -9.25261497e-01 -3.87647003e-02 -8.50025341e-02
1.96165168e+00 6.28425717e-01 9.23326492e-01 6.83789849e-01
-1.43240631e-01 1.17079902e+00 9.20982361e-01 7.12774694e-01
7.67398953e-01 6.86599389e-02 6.93116426e-01 -1.43944949e-01
3.64609808e-01 -2.01564044e-01 1.00866556e+00 6.88634396e-01
-3.96892339e-01 1.39005154e-01 -1.09580851e+00 8.45099449e-01
-1.39878583e+00 -7.17260838e-01 -8.14404249e-01 2.12630081e+00
5.11578321e-01 -5.82531810e-01 3.36185098e-01 -3.88753824e-02
3.91311437e-01 -6.72567606e-01 1.50801346e-01 -7.19578743e-01
3.17822903e-01 3.43262941e-01 8.28962862e-01 2.35510856e-01
-1.90032169e-01 7.02772558e-01 5.83161926e+00 -4.57407385e-02
-1.05058181e+00 2.85110891e-01 2.21737549e-01 8.64607334e-01
-7.66651869e-01 4.35064226e-01 -2.81254441e-01 2.15412408e-01
1.26859629e+00 -2.62088269e-01 -6.84863329e-02 4.98331279e-01
1.19092417e+00 -4.46434021e-01 -1.63134933e-01 -1.18509121e-01
-6.25963211e-01 -1.05139613e+00 -5.35562932e-01 3.73435795e-01
6.67494118e-01 5.16033620e-02 -8.63570124e-02 -1.92686036e-01
1.23111881e-01 -5.55586934e-01 7.80596018e-01 4.40010667e-01
8.18090320e-01 -4.92104739e-01 1.20217752e+00 3.10628325e-01
-1.08427477e+00 -3.15394215e-02 -1.85731247e-01 -6.62661552e-01
-1.78053994e-02 1.97643675e-02 -7.84915447e-01 2.68237323e-01
1.14108193e+00 1.15586042e-01 -4.05032307e-01 1.00981951e+00
2.80519605e-01 1.04620206e+00 -3.67560029e-01 -2.70433098e-01
1.01271421e-01 -5.37996769e-01 2.18458280e-01 1.44509399e+00
9.16576803e-01 4.76332396e-01 -4.89641935e-01 3.71641159e-01
7.74922907e-01 7.25704670e-01 -1.12785339e+00 -4.07793045e-01
1.01598239e+00 1.14780176e+00 -8.61745298e-01 3.61611336e-01
-4.88393247e-01 2.18336016e-01 -7.86372364e-01 5.29560566e-01
-1.99145615e-01 -6.13098145e-01 2.01537400e-01 4.65463012e-01
-8.39269627e-03 -5.91381453e-02 -4.00848687e-01 -1.54332116e-01
-2.36209184e-01 -7.60273457e-01 -2.09509239e-01 -5.89977503e-01
-7.12308645e-01 -2.90046800e-02 2.89105326e-01 -9.96604919e-01
5.00509217e-02 -2.05263942e-01 -4.40162122e-01 9.88345683e-01
-9.78353739e-01 -1.66972101e+00 -1.79198012e-01 5.85514866e-03
-7.84479082e-02 2.18197495e-01 9.10679519e-01 1.80509672e-01
-1.58353388e-01 -8.89447331e-02 4.85447764e-01 -5.77555180e-01
2.65120238e-01 -5.04035234e-01 -3.30089629e-01 9.14464712e-01
-1.11351681e+00 9.90899324e-01 8.18685055e-01 -1.15691650e+00
-1.76842606e+00 -7.84525812e-01 1.23613787e+00 8.92814025e-02
4.25736308e-01 9.00356025e-02 -5.16088367e-01 5.47659218e-01
7.39459932e-01 -6.47299588e-01 1.11946404e+00 -5.56462288e-01
4.14977133e-01 -2.74100900e-01 -1.74591696e+00 5.24210095e-01
4.25141335e-01 -3.24476033e-01 1.84121151e-02 -7.22328797e-02
5.62701225e-01 2.58998036e-01 -1.11778080e+00 2.02555135e-02
1.27952242e+00 -8.02776635e-01 3.37224931e-01 -1.83317792e-02
3.30607772e-01 -2.29074791e-01 -4.69388425e-01 -6.91687047e-01
6.88909646e-03 -9.59008753e-01 6.33877218e-01 1.83690858e+00
3.11318278e-01 -1.01994276e+00 3.12690258e-01 1.20403051e+00
-2.67254770e-01 -1.96286261e-01 -6.28141999e-01 -2.54465431e-01
1.47491038e-01 -2.51972824e-01 5.34696162e-01 9.34874833e-01
4.25330341e-01 -2.20398977e-01 -1.25228673e-01 3.28270346e-01
3.06044757e-01 -9.71287489e-01 8.28065574e-01 -1.02185297e+00
4.04006839e-01 1.13551497e-01 -1.29455954e-01 -4.65469137e-02
-8.16576898e-01 -2.65112698e-01 7.18745738e-02 -2.13824463e+00
2.03800604e-01 -3.20593089e-01 2.66077399e-01 7.55771339e-01
2.07143396e-01 1.08443052e-01 1.85760245e-01 -2.01082602e-02
3.06674987e-01 6.60060346e-02 8.12334061e-01 6.29310548e-01
-7.15609550e-01 -1.17890477e-01 -1.37699378e+00 2.91258961e-01
9.32823598e-01 -6.18135571e-01 -3.13781947e-01 -2.19601929e-01
8.34481001e-01 3.86115521e-01 2.19530210e-01 -5.02753496e-01
-7.83089548e-02 -7.34785914e-01 -2.98750520e-01 -5.17884374e-01
-4.16899115e-01 -1.36287200e+00 1.16719556e+00 1.12298691e+00
2.34289110e-01 3.95312995e-01 2.39702731e-01 -1.56636816e-02
4.66265559e-01 -2.83974946e-01 -3.61165567e-03 2.28894442e-01
-3.73728007e-01 -3.90981466e-01 -1.10714865e+00 -6.80761099e-01
1.19809735e+00 -9.72075164e-01 -6.90764546e-01 -2.86880493e-01
-9.05840546e-02 1.50678769e-01 8.11780512e-01 4.78336364e-02
7.07673550e-01 -7.76058018e-01 -7.49902844e-01 8.00889060e-02
-1.01220362e-01 -5.47668874e-01 1.93127990e-01 8.55665505e-01
-1.27174747e+00 4.61384319e-02 -7.90845394e-01 1.74170695e-02
-9.30296183e-01 -2.52344966e-01 -1.08585753e-01 5.87884486e-01
-3.31758440e-01 1.48909152e-01 -2.55772173e-01 -2.90200353e-01
-1.43067539e-01 -2.39351958e-01 -5.93686402e-01 7.68709481e-02
3.57118219e-01 9.78144765e-01 -2.93506414e-01 -7.08641291e-01
-6.02883995e-01 4.56763625e-01 6.00289643e-01 -5.41275024e-01
1.64113045e+00 -6.27103686e-01 -5.00504076e-01 3.42010081e-01
6.18456006e-01 3.20811212e-01 -1.03162551e+00 7.06780851e-01
9.26893651e-02 -6.99373841e-01 -2.21679762e-01 -1.14914525e+00
-7.52039492e-01 4.00104672e-01 1.17324686e+00 4.05035049e-01
1.19578111e+00 -4.66154993e-01 2.71941215e-01 1.37082547e-01
4.01980728e-01 -1.09673715e+00 -3.13521773e-01 1.90931689e-02
8.91215622e-01 -7.26754248e-01 -3.03118557e-01 -2.22976848e-01
-5.10343790e-01 1.06370747e+00 2.02123567e-01 3.56694311e-01
9.60438311e-01 3.91485363e-01 3.98359716e-01 -1.44979596e-01
-3.67645144e-01 -5.69071919e-02 -6.87562525e-01 1.35513401e+00
8.77221704e-01 3.35534930e-01 -1.41848207e+00 8.75880957e-01
2.56185353e-01 7.10818350e-01 7.27322757e-01 1.20591438e+00
-7.73671389e-01 -9.24069107e-01 -4.48298603e-01 3.79317135e-01
-9.40277517e-01 1.94143131e-01 -3.38917136e-01 1.01329243e+00
2.90013909e-01 1.67699015e+00 -3.78894120e-01 -4.68246520e-01
2.64442652e-01 -4.64962125e-01 -2.40221471e-01 -2.61866957e-01
-1.19465303e+00 3.11058402e-01 4.26746309e-01 -1.56948373e-01
-7.57323980e-01 -7.74605036e-01 -1.14282930e+00 -5.96689105e-01
-4.11911637e-01 3.94789040e-01 1.51536667e+00 7.66645312e-01
5.15700042e-01 -2.77533799e-01 8.39697301e-01 -4.26775992e-01
-3.41167241e-01 -1.23978531e+00 -7.58640289e-01 -3.76419246e-01
-1.07448474e-01 6.37466833e-02 1.22989364e-01 1.55746592e-02] | [9.085504531860352, 6.257633686065674] |
bd603f03-511d-491e-95b5-4f2f2fdf7874 | priberam-at-mesinesp-multi-label | 2105.05614 | null | https://arxiv.org/abs/2105.05614v1 | https://arxiv.org/pdf/2105.05614v1.pdf | Priberam at MESINESP Multi-label Classification of Medical Texts Task | Medical articles provide current state of the art treatments and diagnostics to many medical practitioners and professionals. Existing public databases such as MEDLINE contain over 27 million articles, making it difficult to extract relevant content without the use of efficient search engines. Information retrieval tools are crucial in order to navigate and provide meaningful recommendations for articles and treatments. Classifying these articles into broader medical topics can improve the retrieval of related articles. The set of medical labels considered for the MESINESP task is on the order of several thousands of labels (DeCS codes), which falls under the extreme multi-label classification problem. The heterogeneous and highly hierarchical structure of medical topics makes the task of manually classifying articles extremely laborious and costly. It is, therefore, crucial to automate the process of classification. Typical machine learning algorithms become computationally demanding with such a large number of labels and achieving better recall on such datasets becomes an unsolved problem. This work presents Priberam's participation at the BioASQ task Mesinesp. We address the large multi-label classification problem through the use of four different models: a Support Vector Machine (SVM), a customised search engine (Priberam Search), a BERT based classifier, and a SVM-rank ensemble of all the previous models. Results demonstrate that all three individual models perform well and the best performance is achieved by their ensemble, granting Priberam the 6th place in the present challenge and making it the 2nd best team. | ['Sebastião Miranda', 'Afonso Mendes', 'Zita Marinho', 'Ruben Cardoso'] | 2021-05-12 | null | null | null | null | ['extreme-multi-label-classification'] | ['methodology'] | [ 4.71662521e-01 4.69927937e-02 -6.72412932e-01 -2.28238896e-01
-1.19132376e+00 -6.34418726e-01 6.20309711e-01 9.34896171e-01
-6.72175527e-01 1.07140911e+00 3.59100066e-02 -3.79030049e-01
-5.78506827e-01 -4.51455832e-01 -1.50817588e-01 -6.71039522e-01
2.97129810e-01 1.14568675e+00 3.25134009e-01 -1.15959495e-02
4.14731026e-01 5.57696521e-01 -1.69865787e+00 4.62876350e-01
8.45046461e-01 9.88627195e-01 1.47453174e-01 4.97836202e-01
-3.65756243e-01 5.13674438e-01 -4.58291560e-01 -3.49463940e-01
-2.11457357e-01 -2.58262873e-01 -1.05243564e+00 -2.89392740e-01
2.56588966e-01 4.12216663e-01 4.07899618e-01 7.54706144e-01
5.67377925e-01 -4.25170176e-02 1.13218451e+00 -9.64020789e-01
9.08726677e-02 2.84013033e-01 -4.09337521e-01 3.62087250e-01
4.83817577e-01 -6.04084671e-01 1.17877030e+00 -7.13395774e-01
7.93400407e-01 8.75147521e-01 7.06190825e-01 2.89505839e-01
-1.02756333e+00 -3.69876772e-01 -3.00858736e-01 1.98452786e-01
-1.27218878e+00 -3.07797253e-01 2.51902014e-01 -7.91546106e-01
9.46018875e-01 6.83471918e-01 3.27557981e-01 9.86271203e-01
3.38474572e-01 3.20012718e-01 1.32921779e+00 -5.24539113e-01
2.74658352e-01 5.49632430e-01 3.11672479e-01 6.98783398e-01
4.31470543e-01 -3.04848164e-01 -3.24860275e-01 -7.07679987e-01
5.44618964e-02 6.77734688e-02 -1.06513694e-01 -4.92199175e-02
-1.19280159e+00 9.31395650e-01 2.36024633e-01 8.43917191e-01
-3.67914945e-01 -2.08357498e-01 5.00865579e-01 2.60377735e-01
4.81775522e-01 7.97466934e-01 -6.14419520e-01 7.97011778e-02
-1.21960211e+00 3.67235661e-01 8.84300232e-01 5.57756662e-01
3.04098219e-01 -8.92214477e-01 -7.75545314e-02 1.16185915e+00
2.00769365e-01 1.82553411e-01 7.26315200e-01 -6.31489158e-01
1.72226295e-01 6.68421268e-01 -1.02433027e-03 -1.04915726e+00
-9.51316476e-01 -6.82289362e-01 -8.79630387e-01 -9.20098349e-02
4.78979796e-01 1.54268458e-01 -6.98435724e-01 1.11428249e+00
3.78780812e-01 -3.24940056e-01 -7.96212256e-02 5.40423393e-01
1.01399076e+00 5.34692168e-01 4.34975266e-01 -5.83726764e-01
1.87462246e+00 -8.29771578e-01 -6.65946841e-01 -8.62984061e-02
9.50476468e-01 -9.29555297e-01 4.11904663e-01 6.14335477e-01
-9.29786205e-01 -1.92489117e-01 -8.96156430e-01 3.33440751e-02
-7.44460106e-01 5.38353473e-02 6.08518958e-01 4.37541902e-01
-9.04360414e-01 5.88878155e-01 -3.32008004e-01 -4.75604355e-01
3.55443209e-01 5.40056884e-01 -3.98170590e-01 -2.68693417e-01
-1.29354978e+00 1.39601099e+00 5.37016571e-01 -3.05318803e-01
-1.06390774e-01 -6.62506223e-01 -5.09290278e-01 -1.43120334e-01
4.54504251e-01 -7.63100684e-01 9.24197257e-01 -3.85494173e-01
-7.94006944e-01 1.30836594e+00 -5.14794439e-02 -2.09679648e-01
4.27270710e-01 2.89355427e-01 -3.94894660e-01 2.94011980e-01
3.83574814e-01 3.56500983e-01 4.61722016e-01 -8.30103159e-01
-9.68151450e-01 -5.09404719e-01 -4.25050378e-01 3.24693263e-01
-3.34738225e-01 2.86702186e-01 -2.51186579e-01 -5.06062925e-01
1.06190376e-01 -1.01010358e+00 -3.61329585e-01 -3.65524799e-01
-3.68192226e-01 -7.38075674e-01 3.30899239e-01 -5.53963900e-01
1.26250577e+00 -1.78213513e+00 3.02520245e-01 2.37129316e-01
5.39183199e-01 2.18894016e-02 2.77874917e-01 6.08704805e-01
-1.40025824e-01 2.25195259e-01 2.49619056e-02 -8.91002361e-03
-1.36669978e-01 4.70729321e-02 4.21343446e-02 4.84438866e-01
-1.84136808e-01 7.12666392e-01 -8.38425279e-01 -9.52246726e-01
-1.91876777e-02 2.59346932e-01 -9.06051919e-02 -1.85099274e-01
-1.76450029e-01 2.53658324e-01 -7.25748241e-01 7.45459378e-01
-7.37378150e-02 -7.53512084e-01 6.72719628e-02 -2.76097864e-01
1.23262061e-02 1.80537626e-01 -9.64648724e-01 1.31808519e+00
-2.87398249e-01 2.52119929e-01 -3.28470975e-01 -1.21243250e+00
6.92471981e-01 6.69978857e-01 9.93810415e-01 -4.69649136e-01
1.13773607e-01 7.80847490e-01 -4.19812314e-02 -5.35959244e-01
1.94037125e-01 -4.89282429e-01 -2.26866558e-01 2.45813012e-01
-2.14680079e-02 -7.18189850e-02 3.94051969e-01 1.18215360e-01
1.29289389e+00 -4.40208316e-01 7.69493937e-01 -3.82272273e-01
5.89430392e-01 2.82224059e-01 2.24906817e-01 6.46748245e-01
-6.22053863e-03 4.40403193e-01 5.42285681e-01 -4.90725875e-01
-7.61572957e-01 -5.44213951e-01 -7.39833117e-01 1.06015205e+00
-3.49133044e-01 -3.59614164e-01 -4.65383649e-01 -7.54299581e-01
1.51245683e-01 3.48274797e-01 -6.31080985e-01 1.76515996e-01
-2.07592487e-01 -1.08835864e+00 2.90627271e-01 -1.30506247e-01
-2.80538178e-03 -1.01230872e+00 -5.68388343e-01 3.32731932e-01
-3.06223303e-01 -9.36799049e-01 -1.74196791e-02 5.52968562e-01
-7.07172096e-01 -1.19177604e+00 -1.21126664e+00 -7.55721211e-01
3.92948776e-01 -2.43017107e-01 1.12024891e+00 6.27992377e-02
-6.33295894e-01 1.44153535e-01 -5.34947753e-01 -5.52420676e-01
-5.42539895e-01 5.36071122e-01 -1.44227415e-01 -2.80584574e-01
3.45310926e-01 -2.27884948e-01 -4.89987791e-01 1.50096565e-01
-6.90304697e-01 -5.94854495e-03 8.47526550e-01 8.26424003e-01
6.12579465e-01 3.08057759e-02 9.46744680e-01 -1.31287241e+00
5.98610580e-01 -6.02362514e-01 -3.73091191e-01 3.01455110e-01
-1.02882266e+00 4.33496460e-02 3.47007841e-01 -6.00011200e-02
-5.02788901e-01 -4.60183667e-03 -3.90182823e-01 2.08553463e-01
-3.27673346e-01 8.56617451e-01 3.11418444e-01 -2.13598087e-01
9.21531856e-01 -5.11305295e-02 -1.41758859e-01 -5.44141948e-01
6.10572193e-03 1.02352774e+00 1.01263195e-01 -1.29219592e-01
2.43017808e-01 5.23380451e-02 4.97751206e-01 -8.34375620e-01
-1.13983321e+00 -9.93732333e-01 -3.53778362e-01 -2.24093348e-01
8.25729609e-01 -5.84516287e-01 -6.61033630e-01 1.06332470e-02
-9.32478368e-01 2.76079655e-01 2.34531853e-02 3.35991710e-01
-3.16844612e-01 2.95095772e-01 -4.92183149e-01 -4.53264028e-01
-4.24351275e-01 -1.16603410e+00 1.10483110e+00 -3.91826034e-02
-6.92558169e-01 -1.02865398e+00 2.29633138e-01 8.82937908e-01
2.82568574e-01 4.78711337e-01 1.32421136e+00 -1.07916307e+00
2.92631369e-02 -5.65319419e-01 -1.06897771e-01 -2.64828026e-01
7.33180791e-02 -2.81394988e-01 -8.19527209e-01 -2.48515442e-01
-2.77372934e-02 -2.99928308e-01 1.01930404e+00 4.37262028e-01
9.95778859e-01 -8.72572605e-03 -9.42484200e-01 3.52800786e-02
1.42105985e+00 3.62641662e-01 2.02188790e-01 4.60310072e-01
4.24542814e-01 9.71964478e-01 6.62292182e-01 1.07425198e-01
2.65702069e-01 9.02128875e-01 2.46790797e-01 -8.93384498e-03
2.19584733e-01 4.13007796e-01 -3.84051293e-01 8.31478119e-01
-1.09372646e-01 -2.01998428e-01 -1.37624383e+00 4.49103296e-01
-1.63235593e+00 -6.84588194e-01 -2.88705647e-01 2.07307577e+00
1.10039043e+00 1.84644744e-01 6.37569204e-02 2.99251348e-01
3.35459054e-01 -2.00769514e-01 -2.09866196e-01 -3.55712056e-01
1.47839352e-01 2.40973473e-01 3.60325724e-01 3.13088685e-01
-1.30390072e+00 4.88517851e-01 6.09455776e+00 1.11476970e+00
-7.84475744e-01 2.00051785e-01 7.86168933e-01 3.50764766e-02
7.01019317e-02 -4.02695119e-01 -1.04337716e+00 6.26025379e-01
1.38774347e+00 -1.59355611e-01 -1.27381459e-01 6.50195599e-01
-1.06875310e-02 -5.43179214e-01 -1.03938210e+00 1.10873377e+00
1.50531724e-01 -1.33156574e+00 -1.45626038e-01 2.45365202e-01
6.19923651e-01 2.00654626e-01 -1.06625177e-01 1.37960643e-01
1.24086812e-02 -1.20332921e+00 1.21620178e-01 5.53927064e-01
8.35363686e-01 -3.90014559e-01 9.74057078e-01 4.32527274e-01
-7.02052951e-01 -7.57691413e-02 -1.12583719e-01 3.75441015e-01
6.64792433e-02 9.31472003e-01 -6.84331059e-01 6.79239273e-01
4.87262160e-01 7.25018084e-01 -5.19014537e-01 1.34424257e+00
3.64037365e-01 3.73201907e-01 -1.91595271e-01 -3.11343670e-01
2.27851063e-01 1.50071695e-01 3.34284902e-01 1.20968115e+00
2.91955292e-01 1.58799753e-01 2.86400229e-01 -2.55465843e-02
-8.84820074e-02 7.31824517e-01 -3.66765261e-01 -1.21906295e-01
1.43898889e-01 1.35020769e+00 -1.18348861e+00 -4.31603551e-01
-3.43469024e-01 6.85779154e-01 1.72244444e-01 -1.26313284e-01
-3.99575353e-01 -3.34750205e-01 4.74780537e-02 -4.97644814e-03
-1.75410032e-01 4.55362350e-01 -3.62844259e-01 -7.82704234e-01
-2.22895086e-01 -9.74068344e-01 7.66820490e-01 -4.40162629e-01
-1.35098624e+00 7.18122840e-01 -2.03843061e-02 -1.13497627e+00
-3.99263531e-01 -7.11594582e-01 2.75900513e-01 8.75476062e-01
-1.37775588e+00 -8.18011999e-01 4.16446943e-03 1.75420985e-01
3.80251080e-01 -1.61851197e-01 1.13686168e+00 5.88208914e-01
-2.44866818e-01 2.20761836e-01 4.09251004e-01 -4.58343774e-01
9.64270651e-01 -1.24674451e+00 -5.91551185e-01 -2.42456466e-01
1.15007214e-01 2.68101245e-01 7.22461224e-01 -5.68552852e-01
-7.59618521e-01 -8.58870387e-01 1.55038476e+00 -6.75764322e-01
5.52162588e-01 1.38478398e-01 -8.09135854e-01 1.26434892e-01
1.58742946e-02 -2.94945449e-01 1.29152548e+00 2.06072703e-01
-6.05188459e-02 4.53035645e-02 -1.02704430e+00 1.70196220e-01
4.78600919e-01 -2.97000527e-01 -6.11370981e-01 1.00069833e+00
2.68445283e-01 -2.83827454e-01 -1.21263182e+00 4.24790263e-01
6.07434213e-01 -5.93072295e-01 9.30256665e-01 -5.56559443e-01
5.91433585e-01 2.93465927e-02 9.93549302e-02 -1.14403880e+00
-2.25520745e-01 -2.29937173e-02 1.38942674e-01 8.43353570e-01
8.19998443e-01 -6.01059079e-01 6.30527616e-01 4.42117363e-01
1.03441164e-01 -1.34408760e+00 -9.90958393e-01 -4.50784713e-01
9.66277048e-02 -1.36871055e-01 -6.00102134e-02 1.22771406e+00
2.64385670e-01 5.90477645e-01 -2.31535677e-02 -2.46580064e-01
5.47333181e-01 2.44161516e-01 7.55183846e-02 -1.93767297e+00
-1.32237121e-01 -7.20794797e-01 -4.61767942e-01 -2.00021207e-01
-8.50186720e-02 -1.22126424e+00 -1.86148897e-01 -1.91850936e+00
4.94992673e-01 -5.35317659e-01 -4.79440421e-01 5.05341828e-01
-8.06651115e-02 4.38976228e-01 -1.11070432e-01 5.27322471e-01
-6.00918293e-01 -1.89980477e-01 8.50696802e-01 -2.67931074e-01
5.36897480e-02 2.46010005e-01 -7.96566486e-01 6.88971043e-01
5.62346816e-01 -7.98440218e-01 -1.28961563e-01 2.72289455e-01
6.79843068e-01 1.94850817e-01 1.05153145e-02 -9.14095998e-01
3.54534209e-01 -1.35709077e-01 3.71089458e-01 -5.43062508e-01
2.92903483e-01 -7.95352519e-01 3.50887746e-01 7.14971006e-01
-5.56006372e-01 -1.15198866e-01 6.45393655e-02 4.72141802e-01
-2.32683867e-01 -6.10887289e-01 6.91287696e-01 -2.85568535e-01
-3.31601948e-01 1.28152728e-01 -4.68888938e-01 9.12604406e-02
1.15739954e+00 9.35028866e-02 -2.11894974e-01 5.47909439e-02
-1.12442255e+00 1.94690749e-01 1.75696947e-02 2.88125187e-01
1.94030508e-01 -9.59450305e-01 -7.04646349e-01 -4.09639895e-01
3.45840186e-01 -2.93001533e-01 2.12738797e-01 1.08258343e+00
-5.26596248e-01 8.94090652e-01 1.31371822e-02 -3.97353858e-01
-1.67581892e+00 5.61151147e-01 1.37163147e-01 -8.99308622e-01
-3.19296718e-01 7.18199611e-01 -1.60438061e-01 -1.95846528e-01
2.00873911e-01 -3.53117995e-02 -9.17245328e-01 8.10616612e-01
5.17495632e-01 4.33299333e-01 4.41065490e-01 -5.89683115e-01
-4.70663607e-01 6.64031565e-01 -4.27398711e-01 -2.45602820e-02
1.31750631e+00 2.21765757e-01 -4.66756999e-01 5.99789917e-01
1.18252373e+00 -1.87983140e-01 2.13227253e-02 -1.10747978e-01
5.07010758e-01 6.44110218e-02 1.58018097e-01 -1.19472587e+00
-4.76637185e-01 6.36127353e-01 5.39330006e-01 4.53027248e-01
9.24002945e-01 3.02334577e-01 4.43589181e-01 4.32970524e-01
4.16693777e-01 -1.04485440e+00 -2.73388326e-01 1.44303143e-01
8.40162456e-01 -1.28780472e+00 2.76020110e-01 -6.17140770e-01
-3.54753286e-01 1.01104164e+00 -2.21504532e-02 3.82801592e-01
6.69758022e-01 7.71651343e-02 6.95245862e-02 -4.99849588e-01
-6.71433806e-01 -1.76651716e-01 6.62450671e-01 -6.20070659e-03
5.69478631e-01 -2.71798465e-02 -1.09478021e+00 5.08923113e-01
9.93727595e-02 1.13126673e-01 7.51144439e-02 8.05772781e-01
-5.37262559e-01 -1.38805985e+00 -3.30473602e-01 1.19325924e+00
-1.07624936e+00 -1.28534347e-01 -3.02730769e-01 4.28500324e-01
1.53235272e-01 8.55507493e-01 -4.35867399e-01 -8.79784450e-02
1.91188693e-01 2.90024787e-01 1.04704291e-01 -7.60192990e-01
-8.16400588e-01 1.95194483e-01 2.26022005e-01 -3.54173720e-01
-5.78474760e-01 -7.73419023e-01 -1.05767858e+00 1.38750240e-01
-4.55058575e-01 6.13294065e-01 9.19021845e-01 1.17232168e+00
1.85870245e-01 5.93344808e-01 2.80855864e-01 -5.15693486e-01
-5.23349166e-01 -8.49011660e-01 -5.83234191e-01 3.16920638e-01
2.28923738e-01 -8.29673171e-01 -2.89613456e-01 -2.38172729e-02] | [8.398963928222656, 8.588618278503418] |
54f30920-0cae-41b3-bfda-adabb5d7d522 | combining-hololens-with-instant-nerfs | 2304.14301 | null | https://arxiv.org/abs/2304.14301v2 | https://arxiv.org/pdf/2304.14301v2.pdf | Combining HoloLens with Instant-NeRFs: Advanced Real-Time 3D Mobile Mapping | This work represents a large step into modern ways of fast 3D reconstruction based on RGB camera images. Utilizing a Microsoft HoloLens 2 as a multisensor platform that includes an RGB camera and an inertial measurement unit for SLAM-based camera-pose determination, we train a Neural Radiance Field (NeRF) as a neural scene representation in real-time with the acquired data from the HoloLens. The HoloLens is connected via Wifi to a high-performance PC that is responsible for the training and 3D reconstruction. After the data stream ends, the training is stopped and the 3D reconstruction is initiated, which extracts a point cloud of the scene. With our specialized inference algorithm, five million scene points can be extracted within 1 second. In addition, the point cloud also includes radiometry per point. Our method of 3D reconstruction outperforms grid point sampling with NeRFs by multiple orders of magnitude and can be regarded as a complete real-time 3D reconstruction method in a mobile mapping setup. | ['Patrick Huebner', 'Miriam Jaeger', 'Markus Ulrich', 'Boris Jutzi', 'Dennis Haitz'] | 2023-04-27 | null | null | null | null | ['3d-reconstruction'] | ['computer-vision'] | [ 2.96582848e-01 -1.71050832e-01 1.66654065e-01 -4.01831895e-01
-7.27761209e-01 -4.87815768e-01 4.72863883e-01 -1.34001940e-01
-9.12686348e-01 5.78123152e-01 -3.33927393e-01 -3.47566277e-01
-5.12365736e-02 -1.31144500e+00 -1.26209629e+00 -5.64669430e-01
-1.25115039e-02 1.03799832e+00 5.14902174e-02 -1.17449537e-01
2.37432405e-01 1.17253232e+00 -1.88982153e+00 -6.02122903e-01
4.95322466e-01 1.51235032e+00 4.68246371e-01 8.89900506e-01
-1.41536862e-01 2.97672093e-01 -2.66164005e-01 2.39653885e-01
5.95602632e-01 2.77051061e-01 -3.15133035e-01 -1.22245289e-01
6.98557079e-01 -6.17268682e-01 -2.64235437e-01 8.54909480e-01
6.82083726e-01 3.24436098e-01 1.09153260e-02 -8.97418737e-01
3.75939846e-01 -2.19682410e-01 -2.75132120e-01 -4.08304602e-01
6.11371040e-01 -5.23999631e-02 1.41572386e-01 -1.19075716e+00
3.91370505e-01 7.81671286e-01 1.20205283e+00 -5.29611446e-02
-9.42965329e-01 -5.52110851e-01 -5.63853920e-01 -1.75940201e-01
-1.91359115e+00 -3.85680735e-01 6.05784714e-01 -1.07494488e-01
1.21110535e+00 4.42156106e-01 1.15601766e+00 7.37928808e-01
1.30156294e-01 3.17579098e-02 1.00512505e+00 -4.02615726e-01
6.25973940e-01 -1.10628761e-01 -2.28898108e-01 6.04287624e-01
3.18216741e-01 2.22899124e-01 -8.15976441e-01 -3.62583429e-01
1.29535961e+00 3.32507700e-01 -3.40954065e-01 -3.80822867e-01
-1.26905930e+00 6.29312158e-01 7.58493364e-01 -2.09468395e-01
-8.82863224e-01 5.99369109e-01 -1.70969546e-01 -1.20362729e-01
3.03992838e-01 5.23615442e-02 -3.66261512e-01 -6.02157772e-01
-1.07897639e+00 -8.51675645e-02 7.66983330e-01 1.02470684e+00
1.38293219e+00 -2.11919304e-02 9.29791033e-01 4.04123574e-01
5.70776224e-01 1.31986284e+00 3.27562064e-01 -1.31881833e+00
3.00664812e-01 6.21638000e-01 2.39383921e-01 -9.29477394e-01
-5.65249205e-01 -2.18534991e-01 -9.13715839e-01 4.20282990e-01
2.26417873e-02 -6.77558705e-02 -7.27130353e-01 8.54630888e-01
7.35353231e-01 6.53163314e-01 3.99515079e-03 1.07449329e+00
5.69852054e-01 6.75954938e-01 -7.81013310e-01 2.21905559e-01
1.22300160e+00 -2.73654193e-01 -2.38879055e-01 -4.90829945e-01
3.89621466e-01 -5.62897086e-01 8.28378141e-01 5.18729210e-01
-6.89647675e-01 -5.54375827e-01 -1.11366880e+00 -2.28462413e-01
-3.50883067e-01 2.59646531e-02 9.76431072e-01 4.98791218e-01
-1.25484228e+00 5.51473200e-01 -1.13056946e+00 -4.35024709e-01
1.39787272e-01 5.42800546e-01 -6.50431752e-01 -2.40728512e-01
-8.10741663e-01 1.00635946e+00 3.67710292e-01 4.32147115e-01
-6.92041278e-01 -6.34771466e-01 -1.00834382e+00 -1.92166939e-01
1.11217313e-01 -8.23729336e-01 8.98199797e-01 -5.01681387e-01
-1.70683289e+00 7.51080990e-01 -1.11119330e-01 -3.47231239e-01
3.21821094e-01 -4.65583265e-01 -7.95022771e-02 1.52486756e-01
4.13448103e-02 4.91182059e-01 6.10588491e-01 -1.17249727e+00
-5.93259335e-01 -7.74502277e-01 -1.15179218e-01 4.76992220e-01
3.06960851e-01 -5.67491353e-01 -6.70567691e-01 1.96044251e-01
7.88065970e-01 -1.22806585e+00 -4.46872860e-01 1.75132453e-01
-1.73509970e-01 4.76631045e-01 4.82339293e-01 -3.30484658e-01
3.51187199e-01 -2.06235552e+00 -1.64228380e-01 5.66632867e-01
-4.24357392e-02 -3.82633686e-01 3.13694626e-01 9.99754369e-02
3.20954502e-01 -4.00747597e-01 -3.27715755e-01 -7.65172899e-01
-1.62071675e-01 4.33027267e-01 -1.78039193e-01 9.30942059e-01
-5.93327343e-01 5.17232060e-01 -6.55505955e-01 -1.37978256e-01
8.26830566e-01 1.09174681e+00 -1.69976071e-01 -9.16794816e-04
8.08669534e-03 4.61288989e-01 -3.08454037e-01 6.75180376e-01
1.14851034e+00 -1.30118728e-01 -1.38993576e-01 -1.18768394e-01
-4.31048959e-01 3.95336300e-01 -1.50301003e+00 2.58271194e+00
-8.11733544e-01 7.27784276e-01 2.04260215e-01 -3.71268332e-01
1.16456139e+00 -1.56511649e-01 5.51754951e-01 -5.96693099e-01
1.87257722e-01 2.49100119e-01 -9.88701642e-01 -1.39705643e-01
1.14483047e+00 6.82215346e-03 -5.11751510e-02 2.84459025e-01
-1.63574904e-01 -5.65605462e-01 -6.26331747e-01 -4.62977998e-02
1.01209366e+00 6.54144645e-01 1.52373403e-01 1.86486673e-02
3.21550548e-01 3.30319643e-01 4.38370645e-01 7.42313683e-01
4.56857651e-01 6.62623703e-01 -5.01480579e-01 -6.39830768e-01
-9.62695301e-01 -9.82073367e-01 -2.21457183e-01 3.64738047e-01
4.92404401e-01 -4.00207967e-01 -3.89968246e-01 -1.10231340e-01
2.34954029e-01 6.07501149e-01 -2.68212676e-01 1.92518279e-01
-3.89927030e-01 -5.16654968e-01 5.77348948e-01 2.45095238e-01
7.83425570e-01 -4.64902461e-01 -1.33756649e+00 6.62495419e-02
-1.65578425e-01 -1.15720487e+00 4.12874371e-01 4.69377995e-01
-1.24588609e+00 -9.83114660e-01 -2.50654668e-01 -1.62774831e-01
7.13625669e-01 5.59478283e-01 9.82009947e-01 -2.29874223e-01
1.64169054e-02 5.49243569e-01 3.35954688e-02 -4.00981307e-01
1.31736591e-01 -2.36050263e-01 3.33394527e-01 -1.70364752e-01
4.82548922e-01 -9.15352345e-01 -5.30846357e-01 1.21242113e-01
-5.00266552e-01 2.30736434e-01 2.21647143e-01 5.93827851e-02
1.23575175e+00 -3.79532911e-02 -5.36491513e-01 -3.67289424e-01
-1.18272848e-01 -3.80235404e-01 -1.42745793e+00 -2.53662586e-01
-3.77463698e-01 -1.28977939e-01 2.01763958e-01 -4.87077422e-02
-8.11556041e-01 9.67071712e-01 -2.24675640e-01 -5.44242084e-01
-3.07705760e-01 5.77598631e-01 -1.12624615e-01 -5.49136519e-01
8.19022834e-01 4.58046228e-01 5.03319222e-03 -6.67408526e-01
2.58766085e-01 8.68314564e-01 8.26990902e-01 -2.99673796e-01
8.06442618e-01 8.98939252e-01 3.93355131e-01 -1.11006045e+00
-3.57246101e-01 -5.48546076e-01 -8.86499584e-01 -4.96203363e-01
6.64819181e-01 -1.35963261e+00 -8.87088239e-01 5.16102195e-01
-1.28467202e+00 -4.81829286e-01 -2.83465475e-01 8.88307095e-01
-6.44473195e-01 1.65492743e-01 -2.16842487e-01 -1.05437446e+00
-1.83672100e-01 -1.07111573e+00 1.51540601e+00 3.09471577e-01
1.71201035e-01 -6.13253772e-01 5.11580169e-01 2.99474001e-01
3.09089094e-01 5.11617184e-01 -1.65684700e-01 3.24688494e-01
-1.06230581e+00 -6.13266885e-01 -1.03555843e-02 -2.38973666e-02
-1.34923309e-01 -3.03599060e-01 -1.32495213e+00 -4.87235934e-02
3.69306803e-01 -6.61031231e-02 3.46072555e-01 4.54502434e-01
7.97517657e-01 1.09108932e-01 -3.39970827e-01 1.44053376e+00
2.09252405e+00 -9.89819020e-02 7.95414209e-01 7.16106176e-01
9.05851483e-01 1.45520000e-02 5.99156380e-01 6.06220841e-01
7.54877865e-01 6.73525631e-01 1.11266041e+00 -1.23498879e-01
2.17136756e-01 -2.22794756e-01 1.53368682e-01 6.60173476e-01
-4.90616530e-01 3.45902771e-01 -1.02394617e+00 2.02902481e-02
-1.61101139e+00 -5.49637318e-01 -4.18323427e-01 2.73781872e+00
3.65017235e-01 -1.57464445e-02 -3.57699305e-01 1.46894813e-01
6.00787401e-01 8.48873928e-02 -5.73685884e-01 5.81483170e-02
3.50613855e-02 1.83676854e-01 1.41629779e+00 7.40697742e-01
-6.84371769e-01 8.18868220e-01 5.38038778e+00 2.48407960e-01
-1.31000769e+00 -3.50654125e-04 -8.91878009e-02 -2.00415701e-01
-1.85088217e-01 2.53188640e-01 -1.00788391e+00 3.62818033e-01
1.39388919e+00 4.12683010e-01 9.69173193e-01 1.21319675e+00
1.57680884e-01 -7.28987694e-01 -6.43138647e-01 1.56279206e+00
1.24926060e-01 -1.52795339e+00 -5.11166334e-01 5.95071137e-01
3.01969677e-01 1.09871459e+00 -6.66092277e-01 -1.95525169e-01
1.22998491e-01 -8.05661380e-01 9.26305711e-01 8.63684356e-01
1.11397386e+00 -1.03680742e+00 5.06259859e-01 7.92541981e-01
-1.01096117e+00 3.84272963e-01 -6.64639115e-01 -3.72393072e-01
3.72112125e-01 1.12389410e+00 -9.24017072e-01 6.74561024e-01
9.82639670e-01 5.16975224e-01 -3.18618238e-01 1.24433339e+00
-2.35949636e-01 2.43477583e-01 -1.23052990e+00 1.07576311e-01
1.77249983e-02 -5.18721581e-01 4.23227638e-01 6.66345954e-01
8.63785446e-01 1.58559461e-03 8.40825681e-03 5.44617116e-01
3.75895798e-02 -5.00191689e-01 -9.47998941e-01 6.50589943e-01
7.35312819e-01 1.38338125e+00 -9.21436250e-01 -3.14171851e-01
-1.01946622e-01 1.10266936e+00 -1.84884928e-02 7.15880692e-02
-6.69652462e-01 -5.61062336e-01 5.48915029e-01 -1.33387017e-04
1.22140698e-01 -7.96947777e-01 -3.18110704e-01 -1.18899298e+00
1.37692504e-02 -3.92666191e-01 -3.15219432e-01 -1.32534361e+00
-4.11411583e-01 6.36597872e-01 -3.30135882e-01 -1.38524461e+00
-3.17103505e-01 -5.50652444e-01 -3.99509221e-02 9.71564353e-01
-1.46777678e+00 -8.23804975e-01 -1.09483385e+00 8.22807670e-01
-6.18805885e-02 2.88480580e-01 1.20105636e+00 2.47213349e-01
-1.47506043e-01 -4.74274665e-01 3.70496869e-01 -4.52075377e-02
1.89023957e-01 -1.15173209e+00 6.27706587e-01 8.98710191e-01
3.68586957e-01 4.95693296e-01 4.08702999e-01 -8.38377476e-01
-2.26903272e+00 -8.68055463e-01 7.06512749e-01 -6.38749003e-01
2.83307552e-01 -7.06299722e-01 -5.74170589e-01 8.14446390e-01
-4.21171486e-01 2.56737053e-01 6.39530182e-01 -3.45791019e-02
5.32899052e-02 -4.63706285e-01 -1.20535982e+00 4.35083807e-02
9.97940063e-01 -7.27663755e-01 -3.61496508e-01 4.86644745e-01
6.00286603e-01 -1.19619739e+00 -1.00350356e+00 2.83483714e-01
5.66081583e-01 -1.12784886e+00 1.18996680e+00 5.28432906e-01
-1.62823021e-01 -6.43545210e-01 -7.13422418e-01 -9.70198810e-01
2.87235491e-02 -4.96507049e-01 -1.59026831e-01 5.94263911e-01
-1.87759891e-01 -7.27875948e-01 9.72960532e-01 5.83128154e-01
-1.08567834e-01 -4.42527570e-02 -1.45171428e+00 -6.09236181e-01
-7.80374050e-01 -1.17945826e+00 1.00023472e+00 6.39007747e-01
-6.87560976e-01 1.72175869e-01 -1.06729753e-01 7.18814790e-01
1.00554168e+00 1.36698201e-01 1.27985549e+00 -1.45897603e+00
1.89132355e-02 1.15296401e-01 -5.49421310e-01 -1.23403180e+00
-1.98214412e-01 -6.65289640e-01 1.88338593e-01 -1.59258974e+00
-6.17890477e-01 -6.68730021e-01 1.52359203e-01 1.57874942e-01
4.91162241e-01 4.76792008e-01 -6.15456142e-02 4.57908899e-01
-4.11645263e-01 3.28827649e-01 5.29063404e-01 3.38712066e-01
-1.37743026e-01 8.51202458e-02 -1.13127708e-01 8.85550916e-01
4.95866179e-01 -5.81692636e-01 -8.18837285e-02 -8.72064531e-01
7.64471889e-01 2.60733992e-01 7.45736063e-01 -1.43770134e+00
6.09741390e-01 6.49856031e-02 8.16906929e-01 -1.17306149e+00
1.01727831e+00 -1.43670130e+00 7.39183903e-01 4.55892444e-01
4.76234555e-01 1.77423894e-01 1.56652123e-01 4.06007648e-01
1.55561969e-01 -6.87744394e-02 3.73884171e-01 -3.54472220e-01
-8.14379156e-01 3.95726681e-01 -1.81271970e-01 -6.30504310e-01
6.32789671e-01 -6.10791087e-01 7.24480525e-02 -3.72943431e-01
-1.99913904e-01 -2.06321806e-01 1.03939641e+00 -9.89079177e-02
8.18215787e-01 -1.46357667e+00 -2.92971879e-01 5.31782627e-01
1.85056683e-02 1.04348123e+00 1.15746804e-01 6.39034808e-01
-1.20579290e+00 2.63858169e-01 3.98006104e-02 -1.23052669e+00
-7.40717173e-01 7.39648065e-04 4.88226861e-01 4.34325308e-01
-9.41669524e-01 5.54602027e-01 -7.92306483e-01 -9.31644201e-01
1.84359461e-01 -3.38539213e-01 3.15617323e-01 -2.22136557e-01
6.32562160e-01 5.06571352e-01 4.15621579e-01 -8.61686766e-01
-6.97271228e-01 9.08993602e-01 1.04302454e+00 -4.36398298e-01
1.61652386e+00 -3.41766447e-01 -2.72547573e-01 7.33364105e-01
1.08920705e+00 1.36469752e-01 -1.44252264e+00 5.04746027e-02
-3.02070022e-01 -7.19969213e-01 6.08034790e-01 -5.48207581e-01
-8.50378871e-01 6.32916391e-01 7.73934603e-01 -6.63565770e-02
1.00654078e+00 -2.51304716e-01 5.91570079e-01 7.59646177e-01
1.19912219e+00 -9.15666938e-01 -8.46759140e-01 7.29279637e-01
5.58812618e-01 -1.11067867e+00 3.23036194e-01 7.01545253e-02
-1.15702800e-01 1.22279394e+00 4.78565469e-02 -3.21491688e-01
5.21155834e-01 3.35115105e-01 1.02129288e-01 -2.91791886e-01
1.11594163e-01 -3.63534912e-02 -1.29248083e-01 7.49943137e-01
-2.70051539e-01 -1.65204238e-02 6.73017859e-01 -1.87467277e-01
-6.36558354e-01 3.03771257e-01 4.97595966e-01 9.95094001e-01
-6.46700680e-01 -7.17717886e-01 -9.70095634e-01 1.34561896e-01
2.48695359e-01 -1.13583744e-01 2.29359288e-02 7.60045290e-01
7.04499856e-02 5.05476534e-01 5.21199703e-01 -5.21471083e-01
3.27043563e-01 -2.77204901e-01 6.39915228e-01 -1.85252696e-01
-3.52057964e-01 7.43033886e-02 -2.84172371e-02 -1.34922433e+00
-3.36038560e-01 -5.12241483e-01 -1.52801752e+00 -4.62226629e-01
-3.28214198e-01 9.99262929e-02 1.87241054e+00 6.33968890e-01
6.52559996e-01 -7.53242970e-02 6.83782816e-01 -1.70220137e+00
4.48817620e-03 -7.19504774e-01 -6.72764182e-01 -3.95573795e-01
3.96752536e-01 -4.33850378e-01 -3.69095892e-01 -3.37608427e-01] | [7.372773170471191, -2.2548844814300537] |
4fe5888c-356a-4985-94ad-a185f25eb57a | biked-a-dataset-and-machine-learning | 2103.05844 | null | https://arxiv.org/abs/2103.05844v3 | https://arxiv.org/pdf/2103.05844v3.pdf | BIKED: A Dataset for Computational Bicycle Design with Machine Learning Benchmarks | In this paper, we present "BIKED," a dataset comprised of 4500 individually designed bicycle models sourced from hundreds of designers. We expect BIKED to enable a variety of data-driven design applications for bicycles and support the development of data-driven design methods. The dataset is comprised of a variety of design information including assembly images, component images, numerical design parameters, and class labels. In this paper, we first discuss the processing of the dataset, then highlight some prominent research questions that BIKED can help address. Of these questions, we further explore the following in detail: 1) Are there prominent gaps in the current bicycle market and design space? We explore the design space using unsupervised dimensionality reduction methods. 2) How does one identify the class of a bicycle and what factors play a key role in defining it? We address the bicycle classification task by training a multitude of classifiers using different forms of design data and identifying parameters of particular significance through permutation-based interpretability analysis. 3) How does one synthesize new bicycles using different representation methods? We consider numerous machine learning methods to generate new bicycle models as well as interpolate between and extrapolate from existing models using Variational Autoencoders. The dataset and code are available at http://decode.mit.edu/projects/biked/. | ['Faez Ahmed', 'Brent Curry', 'Lyle Regenwetter'] | 2021-03-10 | null | null | null | null | ['design-synthesis'] | ['adversarial'] | [-1.12297215e-01 -2.53583580e-01 -4.12699282e-01 -1.13005817e-01
-3.99495572e-01 -8.26862514e-01 3.57965469e-01 -6.40705347e-01
3.06014836e-01 7.16170073e-01 5.95457852e-01 -6.10990882e-01
-6.47007883e-01 -7.22374976e-01 -7.08893359e-01 -6.10420823e-01
2.96143681e-01 5.34210920e-01 -3.93351436e-01 -4.30143178e-01
4.29533452e-01 4.38452065e-01 -1.82234108e+00 4.48344231e-01
4.90908593e-01 9.15130675e-01 2.71087974e-01 6.13484681e-01
4.70028549e-01 5.52591562e-01 -4.74451661e-01 -1.49551988e-01
1.54406890e-01 -6.88628495e-01 -6.18974864e-01 1.02910802e-01
-1.57278880e-01 -2.64497131e-01 -3.36690694e-01 5.19361615e-01
7.21198261e-01 1.08260684e-01 1.11942112e+00 -1.41849434e+00
-1.14904022e+00 5.75868845e-01 3.03470548e-02 -1.33348871e-02
2.62108058e-01 5.36752284e-01 1.17254305e+00 -1.34791780e+00
5.67813933e-01 1.11202049e+00 7.18454659e-01 2.92888403e-01
-1.57943368e+00 -6.32600129e-01 -4.47603434e-01 2.25906268e-01
-1.56858408e+00 -5.77995896e-01 1.13929093e+00 -9.52929437e-01
9.67066288e-01 3.02063048e-01 1.08144844e+00 1.26620591e+00
2.91274905e-01 6.35815382e-01 6.98622584e-01 -6.20376706e-01
4.75311786e-01 -1.09396977e-02 -2.67968550e-02 4.74899799e-01
3.12671125e-01 6.23791993e-01 -3.00274968e-01 -7.43419081e-02
1.01559603e+00 -6.66315317e-01 2.22700953e-01 -6.85488164e-01
-1.03569949e+00 1.26973927e+00 1.17762253e-01 3.18139307e-02
-1.10375606e-01 4.89544898e-01 3.76670957e-01 5.09469658e-02
-5.34314699e-02 1.07682467e+00 -5.75625122e-01 -2.69105494e-01
-5.55630982e-01 5.32373965e-01 6.59002364e-01 1.12498081e+00
7.32089639e-01 4.93962646e-01 2.24123523e-01 1.05873322e+00
3.82578790e-01 1.98936567e-01 3.13229173e-01 -1.49313998e+00
1.54942885e-01 1.63917124e-01 -7.43611716e-03 -1.21028233e+00
-5.14388919e-01 -8.58722031e-02 -5.20623326e-01 1.98627368e-01
2.65786231e-01 -3.52992564e-01 -5.93514562e-01 1.45172358e+00
9.51426327e-02 -5.01785159e-01 -2.63425171e-01 7.68046558e-01
1.08883774e+00 6.50409520e-01 -3.13500464e-01 1.11397408e-01
1.30886173e+00 -6.00910187e-01 -5.12598932e-01 5.10557443e-02
7.07746983e-01 -8.76685679e-01 1.11292946e+00 7.02503502e-01
-8.24917495e-01 -8.08409393e-01 -1.42940962e+00 -1.67155750e-02
-3.84641021e-01 5.59843719e-01 7.60110140e-01 1.07256949e+00
-8.50197732e-01 5.82221627e-01 -3.92489523e-01 -1.94256827e-01
4.24721271e-01 4.13318932e-01 8.14320296e-02 1.62567690e-01
-9.95123446e-01 9.63603914e-01 3.60949218e-01 7.27874562e-02
-7.08195686e-01 -7.57541656e-01 -1.12774611e+00 -3.03073168e-01
4.03883308e-01 -8.40707839e-01 1.25096607e+00 -6.79357231e-01
-1.70669603e+00 5.01988269e-02 1.07873529e-01 3.23568612e-01
-4.59235981e-02 1.92455262e-01 -6.28329515e-01 -3.15348536e-01
1.10803843e-01 6.79345250e-01 7.34052658e-01 -1.22241676e+00
-3.59097213e-01 5.51284738e-02 -2.96864808e-01 -2.05289349e-01
5.03076091e-02 1.43144233e-02 -2.94641435e-01 -1.02800989e+00
-9.25764441e-02 -1.23476946e+00 -1.12233974e-01 -1.11981191e-01
-5.18967211e-01 -2.29003847e-01 6.77913129e-01 -8.43601167e-01
1.46052432e+00 -1.86443996e+00 1.38716429e-01 2.32399225e-01
4.02874798e-02 -2.01038837e-01 -1.18243992e-01 6.56495690e-01
-3.37051213e-01 2.35463798e-01 5.94613552e-02 2.31123045e-01
2.38840923e-01 1.79463163e-01 -5.36780097e-02 3.45660508e-01
5.01479864e-01 1.26823640e+00 -5.01235425e-01 -6.52653500e-02
5.37730336e-01 1.93269998e-01 -1.11797011e+00 -2.83169389e-01
-1.81508198e-01 5.63534915e-01 -4.66408640e-01 6.64618671e-01
4.74431515e-01 -9.14853718e-03 2.63835877e-01 -6.51208043e-01
-3.43527734e-01 1.78041473e-01 -1.37156796e+00 1.43180370e+00
-4.33225214e-01 9.99370933e-01 -3.76448065e-01 -1.16887426e+00
1.11984849e+00 -2.02800781e-02 4.16521281e-01 -7.17329741e-01
3.63421261e-01 2.38474429e-01 1.81738317e-01 -7.62906253e-01
7.88996458e-01 -8.93236399e-02 -7.17712641e-01 5.49655437e-01
8.34787562e-02 -4.88069534e-01 3.17408979e-01 -4.21112120e-01
7.24075675e-01 3.31496149e-01 2.52157629e-01 -5.99919200e-01
4.76595201e-02 3.33137453e-01 7.55093932e-01 3.20914149e-01
1.25803262e-01 8.11077595e-01 6.55403852e-01 -6.29906118e-01
-1.66818583e+00 -9.30155635e-01 -7.57491961e-02 7.02072978e-01
-1.19507976e-01 -5.39235234e-01 -3.56567085e-01 -4.05018171e-03
2.02785745e-01 9.88631427e-01 -6.02167368e-01 -2.65579164e-01
-5.87484300e-01 -7.94317007e-01 5.09592295e-01 8.98219705e-01
-8.69758502e-02 -8.24607372e-01 -3.37327510e-01 1.29335776e-01
-4.44603153e-02 -5.09337723e-01 -4.91382301e-01 2.30361298e-01
-5.14945149e-01 -1.07299876e+00 -4.08433616e-01 -1.07285154e+00
3.60678315e-01 -1.79962486e-01 1.21298242e+00 -3.38401765e-01
-5.42604566e-01 3.96795906e-02 -1.62536666e-01 -2.28539690e-01
-7.52839029e-01 4.98079285e-02 2.59688258e-01 -4.95099425e-01
2.55578160e-01 -5.21475494e-01 -5.30520260e-01 8.98711979e-01
-3.84807527e-01 1.61888704e-01 4.95774031e-01 9.05775309e-01
3.65847558e-01 4.92955327e-01 9.82311070e-01 -2.89310753e-01
9.11133707e-01 -5.21886230e-01 -5.47936678e-01 2.75879037e-02
-6.13171458e-01 2.30174258e-01 5.80442667e-01 -3.21948022e-01
-9.26752687e-01 1.09383658e-01 -2.55302042e-01 -1.03947170e-01
-1.74245521e-01 5.53980649e-01 -6.44164741e-01 4.79847997e-01
7.44822681e-01 -2.16973498e-02 2.37262636e-01 -6.64025068e-01
7.78124392e-01 8.19090247e-01 2.50032187e-01 -9.67437327e-01
4.94136184e-01 -1.28194541e-02 -1.02087386e-01 -8.83478224e-01
2.50307739e-01 1.16806395e-01 -5.07347286e-01 -3.94674599e-01
7.29900360e-01 -8.20200086e-01 -9.02567863e-01 2.99987286e-01
-9.40789223e-01 -4.91701216e-01 -4.81096148e-01 4.40180510e-01
-9.92138922e-01 -6.67798892e-02 -4.23505545e-01 -5.26843011e-01
2.35060170e-01 -1.74418676e+00 8.23247612e-01 -3.72044975e-03
-1.06321609e+00 -7.73450196e-01 2.22377643e-01 3.71870667e-01
1.72704235e-01 1.19671993e-01 1.48756790e+00 -9.81947631e-02
-3.26195210e-01 6.62892759e-02 1.33145243e-01 3.31429839e-01
3.71023864e-01 2.85407841e-01 -6.00827157e-01 8.67347345e-02
-3.42026502e-01 -2.28678092e-01 2.86872149e-01 8.77501130e-01
1.10816360e+00 -2.67531812e-01 -2.54388690e-01 4.10250336e-01
1.18707073e+00 7.57694840e-01 6.52419984e-01 4.62297760e-02
7.80587971e-01 8.89785528e-01 2.58238584e-01 2.77258813e-01
3.76784384e-01 9.29057956e-01 -7.48989880e-02 3.92601669e-01
-3.32910299e-01 -6.17744446e-01 1.14216477e-01 1.04615724e+00
1.81339160e-01 -1.92889005e-01 -7.98096478e-01 5.54696083e-01
-1.84994757e+00 -7.38178253e-01 -2.73243994e-01 1.60832226e+00
5.86285591e-01 -1.28603354e-01 6.58662140e-01 3.76916260e-01
7.56852984e-01 -1.29890725e-01 -6.05379522e-01 -8.99261713e-01
-1.28436103e-01 1.89426333e-01 3.63257915e-01 1.15083046e-01
-9.24567401e-01 5.22058845e-01 7.43950272e+00 1.13399684e+00
-6.88241780e-01 -2.83820033e-01 8.89864504e-01 2.69936062e-02
-6.76802814e-01 2.25736067e-01 -5.75197518e-01 6.45727098e-01
8.41518283e-01 -1.21810883e-01 5.56830168e-01 9.32893276e-01
7.16042936e-01 -2.34546661e-02 -1.28143537e+00 1.02723503e+00
4.51359712e-03 -1.93180931e+00 -8.00848752e-02 3.00279319e-01
1.01010907e+00 -6.89112425e-01 2.74692833e-01 2.46336773e-01
3.91953409e-01 -1.27433729e+00 1.14386857e+00 5.63180566e-01
8.33959758e-01 -9.02756929e-01 8.07826966e-02 -1.22904867e-01
-1.39515924e+00 -4.60323423e-01 8.97171572e-02 -4.12753314e-01
2.58157551e-01 4.22238827e-01 -5.92996478e-01 4.48363036e-01
6.73137009e-01 8.21322262e-01 -5.32620430e-01 8.42237473e-01
-3.16946656e-02 4.56641316e-01 -2.17885956e-01 -4.57574785e-01
-2.83314615e-01 -3.43079299e-01 2.61189997e-01 6.57563627e-01
6.09498978e-01 -3.57809812e-01 -2.51626343e-01 1.38538909e+00
2.66383439e-01 -2.39006013e-01 -7.58637011e-01 -2.25201115e-01
6.16229355e-01 7.06339955e-01 -7.14777827e-01 1.83333680e-01
-2.40365997e-01 3.68542314e-01 -4.44137186e-01 4.31916922e-01
-8.42982292e-01 -4.42275882e-01 8.20881724e-01 1.90152660e-01
5.96132159e-01 -4.34506208e-01 -8.06522250e-01 -4.84485269e-01
-2.11399242e-01 -1.07468975e+00 1.02245510e-02 -1.19819808e+00
-1.12162316e+00 -3.51939276e-02 3.13784957e-01 -1.42285264e+00
-5.84754109e-01 -9.86799002e-01 -4.52895254e-01 5.75585425e-01
-2.37001076e-01 -9.93564129e-01 2.38791361e-01 -2.08110452e-01
8.11091065e-01 -5.88482499e-01 6.64978504e-01 4.46222186e-01
-7.45738924e-01 5.73629200e-01 4.49805796e-01 1.55503646e-01
2.68896997e-01 -7.47651279e-01 7.93400109e-01 2.33138129e-01
1.38008296e-02 4.68662471e-01 8.57536793e-01 -7.25763381e-01
-1.72420669e+00 -7.82418311e-01 3.92777681e-01 -7.65746653e-01
6.87071919e-01 -3.75924438e-01 -8.86279196e-02 4.25889522e-01
3.42717990e-02 -6.15517855e-01 8.71188521e-01 8.25765729e-02
1.46402434e-01 1.95598915e-01 -7.73175657e-01 8.91773403e-01
1.19917619e+00 -3.91116410e-01 -5.09386063e-01 -1.85981303e-01
6.03599668e-01 9.50780734e-02 -8.87939572e-01 2.54485250e-01
9.00805712e-01 -7.40528107e-01 1.08136821e+00 -5.92129529e-01
7.89113343e-01 -4.57237035e-01 -3.15817863e-01 -1.48167050e+00
-6.71078324e-01 -5.43289065e-01 2.43485365e-02 1.21832180e+00
6.17054820e-01 -3.83874983e-01 9.96257246e-01 7.33904004e-01
-3.84285986e-01 -8.70650709e-01 -9.17246103e-01 -7.86723435e-01
4.74393696e-01 -9.44808960e-01 9.49224770e-01 5.68022549e-01
1.38658345e-01 4.44811463e-01 -4.73681599e-01 -6.02864087e-01
2.84202129e-01 1.36018515e-01 9.80188072e-01 -9.93820727e-01
-3.23127121e-01 -6.94513679e-01 -4.00000691e-01 -1.02212453e+00
-7.37032155e-03 -8.69592905e-01 7.58224577e-02 -1.27914059e+00
9.27697867e-03 -5.84471285e-01 4.20264125e-01 9.87342894e-02
5.43986738e-01 2.71943301e-01 -1.53367981e-01 -1.14177607e-01
1.81666493e-01 7.42863894e-01 1.05852342e+00 -2.88152337e-01
-4.77136642e-01 -4.03745510e-02 -1.27147913e+00 5.70964158e-01
9.25383925e-01 -1.23452820e-01 -6.93170011e-01 -3.75364721e-01
6.55848503e-01 -1.33208916e-01 4.46443081e-01 -6.48247659e-01
-1.06132239e-01 -3.07767600e-01 5.96645951e-01 -7.16336429e-01
2.47538045e-01 -5.40242255e-01 7.84109950e-01 2.38326252e-01
-1.27532288e-01 3.27646673e-01 4.42579120e-01 4.50495303e-01
3.61900479e-01 -1.91880494e-01 3.22483987e-01 1.26523599e-01
-9.23212767e-01 -4.20632958e-01 -1.15549624e+00 -1.09866716e-01
7.49082685e-01 -4.54434246e-01 -1.41617537e-01 -3.83494109e-01
-7.66836226e-01 -4.77484576e-02 5.16639292e-01 8.61542225e-01
4.91020828e-01 -2.06274915e+00 -3.52438807e-01 5.04556119e-01
1.49359703e-01 -3.10619414e-01 2.71194190e-01 4.58060116e-01
-7.43786454e-01 5.83851695e-01 -3.06522131e-01 -4.74581301e-01
-6.76416099e-01 6.30601764e-01 1.54062763e-01 4.79057819e-01
-4.59878623e-01 4.69958454e-01 1.42087504e-01 -5.59355021e-01
-3.97964120e-01 -4.93372053e-01 -2.34600604e-02 2.81701982e-01
2.10349374e-02 6.63194418e-01 -3.81686278e-02 -9.95662928e-01
-4.38136637e-01 8.13351154e-01 5.23109913e-01 -2.00165078e-01
1.40206552e+00 4.80662994e-02 3.06508601e-01 4.01984543e-01
1.44389355e+00 -4.87169594e-01 -1.24968195e+00 4.32033896e-01
-6.58742413e-02 -3.21265548e-01 -4.17201198e-04 -4.83961999e-01
-1.07482314e+00 3.50938171e-01 6.18380606e-01 -5.60406223e-02
8.43826592e-01 1.94698498e-01 5.88652194e-01 1.00971043e-01
1.47180408e-01 -1.65742481e+00 3.83066952e-01 2.28906155e-01
1.16329241e+00 -8.42955410e-01 1.20496124e-01 -2.94748217e-01
-7.67729938e-01 9.72693026e-01 2.02735648e-01 -1.41174823e-01
9.79731858e-01 3.88103783e-01 -2.67312884e-01 -2.70753056e-01
-4.50848252e-01 6.06233627e-02 4.93679523e-01 7.28873134e-01
1.68710709e-01 1.55371502e-01 -1.60396889e-01 1.28987002e+00
-7.65888751e-01 2.02257425e-01 4.47639227e-01 9.93763030e-01
-2.26788878e-01 -1.23191750e+00 -6.16243482e-01 5.95940828e-01
3.77547383e-01 1.06605977e-01 -4.14420962e-01 8.09157073e-01
3.99309188e-01 1.24938881e+00 2.00183392e-01 -9.67374146e-01
4.66256022e-01 -5.74689358e-02 4.64159697e-01 -3.88839900e-01
-5.07376641e-02 1.69000730e-01 5.24547994e-01 -1.67436108e-01
-1.07955132e-02 -9.03723836e-01 -7.47986376e-01 -5.24813712e-01
-4.51558530e-01 -8.32070708e-02 6.60115957e-01 8.10988307e-01
5.21213055e-01 6.77040935e-01 6.52369857e-01 -1.23142922e+00
-1.45598203e-01 -6.25106096e-01 -5.19537628e-01 -1.15393788e-01
-7.28351995e-02 -1.10665143e+00 -1.37545124e-01 2.30016425e-01] | [5.8431267738342285, 3.254589557647705] |
aec077e4-fb96-453a-b627-deb9374e4886 | st-mvl-filling-missing-values-in-geo-sensory | null | null | https://www.microsoft.com/en-us/research/publication/st-mvl-filling-missing-values-in-geo-sensory-time-series-data/ | https://www.ijcai.org/Proceedings/16/Papers/384.pdf | ST-MVL: Filling Missing Values in Geo-Sensory Time Series Data | Many sensors have been deployed in the physical world, generating massive geo-tagged time series data. In reality, readings of sensors are usually lost at various unexpected moments because of sensor or communication errors. Those missing readings do not only affect real-time monitoring but also compromise the performance of further data analysis. In this paper, we propose a spatio-temporal multi-view-based learning (ST-MVL) method to collectively fill missing readings in a collection of geosensory time series data, considering 1) the temporal correlation between readings at different timestamps in the same series and 2) the spatial correlation between different time series. Our method combines empirical statistic models, consisting of Inverse Distance Weighting and Simple Exponential Smoothing, with data-driven algorithms, comprised of User-based and Item-based Collaborative Filtering. The former models handle general missing cases based on empirical assumptions derived from history data over a long period, standing for two global views from spatial and temporal perspectives respectively. The latter algorithms deal with special cases where empirical assumptions may not hold, based on recent contexts of data, denoting two local views from spatial and temporal perspectives respectively. The predictions of the four views are aggregated to a final value in a multi-view learning algorithm. We evaluate our method based on Beijing air quality and meteorological data, finding advantages to our model compared with ten baseline approaches. | ['Yu Zheng', 'Tianrui Li', 'Junbo Zhang', 'Xiuwen Yi'] | 2016-07-09 | null | null | null | ijcai-2016-2016-7 | ['multivariate-time-series-imputation'] | ['time-series'] | [ 8.85992795e-02 -5.02877235e-01 -8.45611989e-02 -5.12010276e-01
-1.05918562e+00 -5.05184948e-01 5.75215518e-01 7.00223565e-01
-2.92610228e-01 1.01335895e+00 6.10247612e-01 -1.46112755e-01
-6.96484506e-01 -1.13050795e+00 -6.92509294e-01 -7.74734855e-01
-3.26932877e-01 1.24788061e-01 2.42359579e-01 -1.97836310e-02
-1.70724794e-01 1.68572158e-01 -1.67094791e+00 2.94469833e-01
8.86120021e-01 1.16499949e+00 1.84748232e-01 5.57590604e-01
-1.06059320e-01 5.76350033e-01 -4.55343902e-01 8.51118863e-02
2.25828335e-01 -6.72338158e-02 -1.36577815e-01 -4.91027609e-02
1.00911118e-01 -1.18076913e-01 3.17998230e-01 6.74758315e-01
4.27568406e-01 2.79716998e-01 3.06307465e-01 -1.37244272e+00
-1.38523534e-01 4.80709344e-01 -6.43997848e-01 1.46880850e-01
4.37082350e-01 -4.41905081e-01 9.26010251e-01 -8.78572941e-01
1.06602021e-01 8.39241624e-01 1.09070075e+00 -2.32848555e-01
-7.96875834e-01 -2.83292919e-01 6.45744920e-01 2.26977989e-01
-1.24898434e+00 -3.23434412e-01 7.32181251e-01 -4.25875247e-01
6.12997353e-01 5.50523102e-01 6.16603553e-01 9.08827066e-01
3.87194127e-01 3.99854422e-01 1.11680210e+00 1.90259907e-02
5.44402957e-01 -1.11650536e-03 -2.67243564e-01 -1.34720787e-01
3.25878590e-01 1.51238546e-01 -6.29059434e-01 -6.00605369e-01
1.54664353e-01 9.34733987e-01 -7.58875087e-02 -1.39090523e-01
-1.21582997e+00 5.06834030e-01 1.89893275e-01 3.79323721e-01
-8.55438113e-01 -5.72845101e-01 2.37239525e-01 3.76229048e-01
1.11709964e+00 -1.72303647e-01 -9.86386001e-01 7.06137940e-02
-1.24551070e+00 1.34060532e-01 6.38552070e-01 6.60212517e-01
7.32842624e-01 -9.16414410e-02 1.01399943e-01 6.00177228e-01
3.23219955e-01 8.27434242e-01 3.23010892e-01 -4.40730274e-01
6.66907907e-01 6.35408819e-01 6.85315430e-01 -1.04483688e+00
-6.53984666e-01 -2.94512808e-01 -1.20129895e+00 -1.47238016e-01
4.63693291e-01 -5.55824399e-01 -3.81429225e-01 1.46414077e+00
7.69931912e-01 6.50711060e-01 -3.27436030e-02 5.35161436e-01
5.60982049e-01 7.37293243e-01 5.61063960e-02 -7.23288715e-01
1.04571331e+00 -3.82922530e-01 -8.22428107e-01 2.06246838e-01
4.97736514e-01 -5.32893777e-01 4.80768502e-01 6.01858079e-01
-8.54634464e-01 -7.87806809e-01 -7.80810237e-01 6.28046095e-01
-9.52009201e-01 2.15935364e-01 1.26072466e-01 4.77532506e-01
-7.97337413e-01 6.88342214e-01 -8.60494077e-01 -2.76688337e-01
1.12430409e-01 -9.86922681e-02 -1.67116687e-01 -2.96609886e-02
-1.17647362e+00 4.84299153e-01 1.41254842e-01 2.01630101e-01
-4.32521433e-01 -9.06156957e-01 -5.01115501e-01 -1.95311680e-01
4.14144218e-01 -5.09878874e-01 7.67567039e-01 -2.54209071e-01
-6.82819605e-01 1.37973413e-01 -5.87851763e-01 -4.28283960e-01
7.00672865e-01 -4.23362970e-01 -1.29162705e+00 -5.46689093e-01
2.71307081e-01 -3.56376916e-01 7.79036045e-01 -1.19635487e+00
-1.14027357e+00 -6.44342959e-01 -1.95531309e-01 6.80389181e-02
-2.82266945e-01 -3.94463956e-01 5.13169952e-02 -6.87367320e-01
2.52508789e-01 -5.54395139e-01 -4.14028972e-01 -2.40778372e-01
-4.26974110e-02 -3.36524218e-01 8.89305592e-01 -8.32291067e-01
1.65009868e+00 -2.04552841e+00 -2.05060467e-01 2.22106695e-01
-2.32733618e-02 1.54894188e-01 2.42344350e-01 1.02390814e+00
8.81974846e-02 -7.10535571e-02 -1.33155063e-01 -3.97114396e-01
-3.58143508e-01 3.73291850e-01 -5.29492497e-01 6.00121737e-01
-3.53451222e-01 4.17278647e-01 -1.15561855e+00 -1.17692061e-01
4.91653115e-01 2.22207353e-01 -2.39839945e-02 3.27964634e-01
-2.03825831e-01 6.77197516e-01 -4.64970231e-01 6.29481733e-01
8.79340708e-01 -1.87590107e-01 9.50065181e-02 -3.26325655e-01
-6.28511250e-01 1.18717633e-01 -1.65547156e+00 1.51224637e+00
-7.19526052e-01 -1.94240257e-01 1.79376863e-02 -9.92497742e-01
8.47531497e-01 6.51771426e-01 1.11431766e+00 -6.09908164e-01
-3.22732389e-01 1.36607200e-01 -7.02463984e-01 -7.62710452e-01
3.92473549e-01 2.40187142e-02 -7.82469437e-02 5.99247634e-01
-4.33891684e-01 3.36212993e-01 -1.00552663e-01 -2.59322554e-01
8.58251631e-01 2.27886736e-01 5.44794500e-01 -1.78566866e-03
6.31399512e-01 -2.22462997e-01 8.47974777e-01 8.79255891e-01
1.15547732e-01 4.88526076e-01 -1.76984742e-01 -8.83760333e-01
-1.03261876e+00 -1.08220208e+00 -3.95225994e-02 1.26479459e+00
-4.09264583e-03 -5.14583886e-01 9.63493884e-02 -6.33631825e-01
2.72207886e-01 7.04111695e-01 -5.90629041e-01 2.32750759e-01
-3.34029794e-01 -1.02484512e+00 7.36315995e-02 5.66401720e-01
2.89004475e-01 -5.26859283e-01 -7.84556925e-01 5.06267071e-01
-1.80982605e-01 -8.40781212e-01 -6.19707964e-02 4.54950258e-02
-1.02294147e+00 -1.04992998e+00 -4.67151076e-01 8.79020393e-02
4.06754911e-01 7.27171659e-01 1.02286327e+00 -2.52118617e-01
2.83469498e-01 5.38065374e-01 -5.75344563e-01 -8.29122901e-01
1.43230096e-01 -3.09928775e-01 3.44641477e-01 4.52846050e-01
4.09515977e-01 -9.84428227e-01 -4.85978991e-01 4.16182071e-01
-9.76142704e-01 -2.86184192e-01 1.97578177e-01 5.30583918e-01
8.02902877e-01 4.47477221e-01 9.57562387e-01 -7.05484152e-01
3.03313017e-01 -1.15889549e+00 -5.47252059e-01 4.46423978e-01
-5.68739176e-01 -3.65020812e-01 7.68576264e-01 -2.21602142e-01
-1.02027738e+00 -1.35660306e-01 1.67716574e-02 -2.49562353e-01
-3.82701576e-01 8.17871213e-01 -1.64699867e-01 5.51870286e-01
4.47268903e-01 4.29230481e-01 -3.93371105e-01 -8.85545611e-01
7.93347657e-02 6.41062617e-01 2.88098603e-01 -3.77261072e-01
7.65171170e-01 9.01705027e-01 -2.23904893e-01 -8.74988317e-01
-1.20013559e+00 -9.82736766e-01 -7.56209791e-01 -3.12976807e-01
5.16850471e-01 -1.41957605e+00 -2.30838776e-01 5.01071870e-01
-8.31739306e-01 2.40987375e-01 -4.36708629e-01 7.87070870e-01
-6.02926128e-02 5.77031299e-02 1.02208063e-01 -1.21455324e+00
-3.82055417e-02 -4.01313901e-01 1.21540916e+00 8.33222419e-02
-4.98454757e-02 -1.40576684e+00 3.48379642e-01 3.73986661e-02
4.83007938e-01 6.01592839e-01 3.32433641e-01 -8.39423180e-01
-2.58192718e-01 -2.48383388e-01 6.45560548e-02 1.67665467e-01
6.84618711e-01 -3.58912379e-01 -1.08047044e+00 -5.40214002e-01
3.13599527e-01 2.02651754e-01 5.96248627e-01 5.07395625e-01
1.06104934e+00 -4.44494009e-01 -4.91434008e-01 3.52703810e-01
1.58279848e+00 2.03250811e-01 2.32333124e-01 7.06533045e-02
5.48660398e-01 4.72670913e-01 7.89844215e-01 1.20499539e+00
7.52827704e-01 3.79690766e-01 5.99971771e-01 -8.52212012e-02
5.04248798e-01 -3.09569120e-01 1.74756318e-01 8.70534301e-01
-3.15669656e-01 -4.08882618e-01 -6.20796561e-01 9.83514428e-01
-2.26903081e+00 -1.10008156e+00 -3.82150203e-01 2.76890421e+00
2.28677675e-01 -7.44203776e-02 4.21602011e-01 2.35494033e-01
5.87293506e-01 6.68747783e-01 -6.21144116e-01 2.39007711e-01
-9.62698758e-02 -2.29917124e-01 7.91852057e-01 2.15740383e-01
-1.25380147e+00 -1.59347594e-01 5.56808281e+00 3.55035454e-01
-8.38579357e-01 4.45918918e-01 2.23723099e-01 -2.11246297e-01
-5.45437574e-01 -8.55904222e-02 -7.80614078e-01 7.93434560e-01
1.13064599e+00 -7.10415840e-02 1.27346843e-01 5.49746931e-01
8.48517835e-01 -2.64719754e-01 -8.61749589e-01 8.25328469e-01
-1.26171261e-01 -1.24741912e+00 -3.67376685e-01 1.99697480e-01
9.56515849e-01 4.19597954e-01 -2.91621506e-01 -1.70154825e-01
2.75483549e-01 -5.45361757e-01 5.52911580e-01 1.14922857e+00
3.09730649e-01 -5.52237988e-01 8.12650323e-01 8.02472949e-01
-1.74772799e+00 -3.19970995e-01 -2.40695193e-01 -2.79208034e-01
5.12274027e-01 1.38568556e+00 -2.68721521e-01 1.33123124e+00
1.00479734e+00 1.19772148e+00 -2.93344319e-01 1.22171462e+00
2.40121126e-01 7.20025122e-01 -6.46488667e-01 2.35029221e-01
4.44783568e-02 -3.84822428e-01 6.13148570e-01 7.06390858e-01
9.27130103e-01 1.96421742e-01 4.30252522e-01 3.31054777e-01
4.35256958e-01 -7.75440335e-02 -9.55305398e-01 5.75976312e-01
7.58910120e-01 9.80526030e-01 -2.51164973e-01 -4.48490500e-01
-9.72349286e-01 3.12715530e-01 -1.76980987e-01 3.57579768e-01
-7.45753825e-01 1.43977895e-01 4.89046961e-01 2.19308019e-01
4.58634973e-01 -3.88590276e-01 -2.57908255e-01 -1.25541580e+00
3.47917736e-01 -4.32438254e-01 6.21443808e-01 -4.95676458e-01
-1.71870458e+00 8.84182826e-02 2.12410003e-01 -1.91401458e+00
-1.31758630e-01 -1.01309977e-01 -8.27228904e-01 7.86923647e-01
-1.52528179e+00 -1.13566041e+00 -2.53109455e-01 7.76976287e-01
4.88784462e-01 5.94731495e-02 8.88880551e-01 5.88609219e-01
-1.64854273e-01 -2.25503519e-02 6.34614646e-01 -3.59614789e-01
8.25603247e-01 -1.08247554e+00 2.76162297e-01 7.92914927e-01
1.46883816e-01 2.61456639e-01 5.63172042e-01 -7.77195156e-01
-1.23753428e+00 -1.55066597e+00 1.39971912e+00 -4.85614628e-01
6.39319420e-01 -6.65537119e-02 -1.07213235e+00 5.58034837e-01
-1.20347977e-01 2.36410096e-01 1.06844115e+00 3.21825147e-01
-1.10234961e-01 -5.34864962e-01 -1.21655011e+00 -1.44138694e-01
8.36412132e-01 -4.65625137e-01 -5.40525734e-01 4.36916292e-01
3.15184742e-01 -1.95079878e-01 -1.20615566e+00 6.14347160e-01
5.86705267e-01 -1.13613522e+00 1.01422560e+00 -5.81066191e-01
-1.37870878e-01 -6.85498595e-01 -5.90524852e-01 -1.55153894e+00
-4.20213878e-01 -2.20135540e-01 -1.40141696e-01 1.31911993e+00
3.79677474e-01 -7.21484721e-01 3.67130339e-01 3.38194340e-01
1.08658135e-01 -3.72316092e-01 -1.08729291e+00 -7.14307547e-01
-2.62170762e-01 -8.49604309e-01 1.20050383e+00 1.06207073e+00
-1.82294443e-01 -1.21421166e-01 -1.04075181e+00 7.10743964e-01
5.28573275e-01 3.64945769e-01 6.23291194e-01 -1.83328962e+00
-6.30809590e-02 3.25700402e-01 1.06571831e-01 -7.12278008e-01
-4.45165962e-01 -2.21593782e-01 -3.11010718e-01 -1.58223999e+00
-2.09556922e-01 -5.88950276e-01 -6.86554193e-01 1.73651651e-01
-4.84144911e-02 -1.17757067e-01 -2.45428979e-01 1.83631212e-01
-6.32526636e-01 4.78437692e-01 6.45874798e-01 6.70205504e-02
-4.54676598e-01 7.76438177e-01 -2.32760891e-01 8.93858731e-01
8.68164361e-01 -3.63729775e-01 -6.32693231e-01 -4.52941626e-01
6.65866017e-01 3.73438567e-01 4.39131796e-01 -1.23937321e+00
3.91925275e-01 -5.74196696e-01 5.46904981e-01 -1.26906371e+00
1.90138340e-01 -1.46097136e+00 7.05246508e-01 1.66105404e-01
-4.96980734e-02 4.39434618e-01 -9.72079337e-02 1.09552157e+00
-2.65972614e-01 2.46936738e-01 2.14799479e-01 -3.50836456e-01
-6.33510709e-01 4.56400216e-01 -2.28225499e-01 -2.02990636e-01
8.22401464e-01 -2.33459160e-01 5.48783597e-03 -4.81464863e-01
-1.07330048e+00 4.50771719e-01 1.50635988e-01 5.61050296e-01
3.59123915e-01 -1.56809485e+00 -7.78306067e-01 3.03808659e-01
3.14860016e-01 -2.86566094e-02 6.73033059e-01 1.02959442e+00
4.97662008e-01 2.45552346e-01 1.03789270e-01 -6.04112446e-01
-1.03529048e+00 5.54493904e-01 5.50518185e-02 -3.93927842e-01
-4.62114334e-01 3.17429990e-01 -2.86772490e-01 -4.90939319e-01
1.25407159e-01 -5.23189604e-01 -4.66596037e-01 7.37018824e-01
7.58933306e-01 8.24573040e-01 4.62471277e-01 -6.00554824e-01
-3.60544324e-01 5.39015114e-01 4.47555482e-01 1.00747161e-01
1.74440598e+00 -6.80631518e-01 2.56934687e-02 1.21731722e+00
9.84475791e-01 1.97935089e-01 -1.05355334e+00 -7.77460337e-01
-5.20912092e-03 -5.04620552e-01 -1.50263697e-01 -7.90400207e-01
-8.28649342e-01 4.75659072e-01 7.52326488e-01 8.31738591e-01
1.18992674e+00 -1.46457165e-01 8.87285590e-01 -1.75830442e-02
6.64763987e-01 -1.36504424e+00 -6.13991559e-01 2.53478587e-01
5.12918711e-01 -1.36669898e+00 2.38428131e-01 3.72827984e-02
-3.54469687e-01 8.16719949e-01 1.42492190e-01 8.62115100e-02
1.37741649e+00 9.98882279e-02 -6.39097169e-02 1.00341374e-02
-9.28056359e-01 -1.93926290e-01 2.91325480e-01 7.13534236e-01
3.39780480e-01 3.92422587e-01 -2.39482284e-01 5.05871415e-01
5.83843440e-02 1.37320593e-01 1.01958625e-01 1.05184579e+00
-5.26590526e-01 -1.00038874e+00 -6.36438608e-01 7.03108609e-01
-3.91818732e-01 3.12632263e-01 2.52450943e-01 3.69946808e-01
6.63051307e-01 1.37245631e+00 1.60234019e-01 -3.67405742e-01
6.05741739e-01 1.94138765e-01 -2.26821989e-01 -1.83805957e-01
-3.26552778e-01 2.62284249e-01 1.43643856e-01 -6.03976130e-01
-1.02681196e+00 -1.21727479e+00 -6.65819764e-01 -2.36513704e-01
-6.59837127e-02 1.90533206e-01 5.77909112e-01 1.21735072e+00
5.49392819e-01 5.40895283e-01 1.13457572e+00 -8.85370910e-01
-5.31611025e-01 -9.10746157e-01 -8.61496627e-01 3.08163494e-01
7.99749553e-01 -4.75083292e-01 -3.36578220e-01 1.87266201e-01] | [6.778815746307373, 2.731411933898926] |
5e45668a-c3fd-47b0-9fcb-c74b1dec5b9d | atttrack-online-deep-attention-transfer-for | 2210.08648 | null | https://arxiv.org/abs/2210.08648v2 | https://arxiv.org/pdf/2210.08648v2.pdf | AttTrack: Online Deep Attention Transfer for Multi-object Tracking | Multi-object tracking (MOT) is a vital component of intelligent video analytics applications such as surveillance and autonomous driving. The time and storage complexity required to execute deep learning models for visual object tracking hinder their adoption on embedded devices with limited computing power. In this paper, we aim to accelerate MOT by transferring the knowledge from high-level features of a complex network (teacher) to a lightweight network (student) at both training and inference times. The proposed AttTrack framework has three key components: 1) cross-model feature learning to align intermediate representations from the teacher and student models, 2) interleaving the execution of the two models at inference time, and 3) incorporating the updated predictions from the teacher model as prior knowledge to assist the student model. Experiments on pedestrian tracking tasks are conducted on the MOT17 and MOT15 datasets using two different object detection backbones YOLOv5 and DLA34 show that AttTrack can significantly improve student model tracking performance while sacrificing only minor degradation of tracking speed. | ['Rong Zheng', 'Keivan Nalaie'] | 2022-10-16 | null | null | null | null | ['deep-attention', 'visual-object-tracking', 'deep-attention'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 4.66661230e-02 -4.87570986e-02 -2.24495918e-01 -2.60942757e-01
-2.49033898e-01 -3.71389270e-01 4.16187465e-01 -6.22297935e-02
-5.25754333e-01 4.59431499e-01 -4.04160649e-01 -5.60135543e-01
1.55968890e-01 -5.77515602e-01 -9.21740890e-01 -5.77236414e-01
1.70034319e-01 5.87048471e-01 1.04428303e+00 2.84857243e-01
-1.75074279e-01 5.02669454e-01 -1.91351807e+00 2.65348136e-01
7.45095193e-01 1.12613380e+00 3.46831113e-01 1.20028198e+00
-3.13903838e-02 1.19892645e+00 -4.19266462e-01 -3.88763189e-01
1.98586002e-01 1.53765172e-01 -4.18625742e-01 -6.38420274e-03
9.82918262e-01 -5.98698199e-01 -5.17895937e-01 1.01873612e+00
4.03852850e-01 1.26670778e-01 3.20017308e-01 -1.86654568e+00
-1.44821778e-01 4.04371858e-01 -6.24639094e-01 4.96633828e-01
-3.68165523e-01 4.89202976e-01 5.41351914e-01 -5.67087233e-01
2.24845618e-01 1.38615525e+00 9.68266547e-01 7.10769176e-01
-9.61697400e-01 -1.04261672e+00 4.57177520e-01 6.69844389e-01
-1.06083572e+00 -6.79043591e-01 4.17202443e-01 -6.55558944e-01
9.96313691e-01 2.85669100e-02 8.02186549e-01 8.94764006e-01
2.98060209e-01 1.07579494e+00 7.11651564e-01 -9.24341753e-02
7.37261251e-02 2.66062230e-01 5.27939200e-01 1.09386837e+00
4.63805705e-01 3.51548851e-01 -5.28437793e-01 -8.26342106e-02
4.62589234e-01 1.17272727e-01 2.52784610e-01 -4.06452835e-01
-7.46540248e-01 5.80264270e-01 4.40228909e-01 -7.11139888e-02
-4.28391784e-01 6.21023476e-01 4.24208581e-01 1.51460439e-01
2.54553370e-02 -4.15164620e-01 -5.36956966e-01 -6.76789209e-02
-1.00195086e+00 1.21888466e-01 6.27442002e-01 1.27402008e+00
7.69182622e-01 3.13442230e-01 -2.42972881e-01 1.73714966e-01
8.11451375e-01 7.72401333e-01 2.56735981e-01 -8.65634561e-01
1.90994278e-01 5.84741294e-01 1.67575970e-01 -7.41478801e-01
-3.83755714e-01 -4.08434272e-01 -3.65463644e-01 4.04702544e-01
5.66439509e-01 -2.34149814e-01 -1.04041660e+00 1.61085582e+00
8.81060660e-01 7.71008313e-01 -7.97408819e-02 6.63502634e-01
1.11292195e+00 5.50501406e-01 5.64136982e-01 2.13587224e-01
1.67257965e+00 -1.34524202e+00 -4.20940220e-01 -4.10100728e-01
6.85062349e-01 -5.81095755e-01 4.07263577e-01 2.54390985e-01
-1.07136154e+00 -1.12789464e+00 -1.00294268e+00 -1.45352393e-01
-3.41852397e-01 4.11701053e-01 5.03085494e-01 7.59101510e-01
-1.12396276e+00 4.32015210e-01 -1.41943014e+00 -3.13574374e-01
8.06365490e-01 7.14339256e-01 1.21633790e-01 4.61485647e-02
-5.33840060e-01 9.08633232e-01 3.35330963e-01 1.42232329e-01
-1.37054574e+00 -9.18739676e-01 -5.83061099e-01 1.55561164e-01
2.73449957e-01 -8.86348367e-01 1.37123823e+00 -8.34366918e-01
-1.46365237e+00 4.81854796e-01 -3.26384485e-01 -7.37736464e-01
6.63073421e-01 -5.86449146e-01 -3.99152726e-01 -1.01320175e-02
-7.22344294e-02 1.06079495e+00 1.14327896e+00 -9.24613059e-01
-1.23874366e+00 -1.64676666e-01 -7.19833076e-02 -7.23290071e-02
-3.65596324e-01 5.53777143e-02 -5.36687136e-01 -6.73327297e-02
-4.28544372e-01 -1.04752219e+00 -1.02695353e-01 4.79752243e-01
-5.51372170e-02 -5.01889467e-01 1.65927362e+00 -6.53748930e-01
8.33909452e-01 -1.94761157e+00 -1.69678316e-01 -1.05242901e-01
5.89828014e-01 7.31264472e-01 -1.36579722e-01 -2.68877894e-01
2.45251387e-01 -6.13030374e-01 5.15150905e-01 -4.07636970e-01
-2.66221110e-02 2.15020433e-01 -2.02947602e-01 4.82123226e-01
2.27758646e-01 1.06322050e+00 -8.17560911e-01 -8.61400783e-01
5.30994117e-01 4.94577438e-01 -6.20895445e-01 2.26116806e-01
-4.27998275e-01 3.64896715e-01 -4.61699784e-01 5.75973690e-01
6.23531103e-01 -6.04229569e-01 1.29406318e-01 -4.11145657e-01
-2.82132775e-01 1.77335650e-01 -1.17070067e+00 1.14189899e+00
-2.80392438e-01 9.88505304e-01 8.40436369e-02 -7.70840943e-01
6.29979491e-01 1.88783035e-01 4.66090769e-01 -5.88800728e-01
3.43078285e-01 -2.74821043e-01 1.59535185e-01 -3.47266257e-01
5.77736378e-01 3.99639398e-01 2.82801330e-01 3.42462361e-01
1.44776866e-01 7.20614016e-01 2.19150901e-01 5.56114912e-02
1.09393978e+00 2.96239913e-01 -1.10027052e-01 -8.73326287e-02
4.88685757e-01 2.85849661e-01 7.72130847e-01 8.86001229e-01
-6.52156234e-01 -2.93968439e-01 -7.15994388e-02 -8.89677227e-01
-8.66597712e-01 -1.05642760e+00 1.82917535e-01 1.53197432e+00
1.72334045e-01 -2.80668348e-01 -5.06480932e-01 -7.95173585e-01
1.88251868e-01 5.30461371e-01 -3.61043632e-01 -2.80714720e-01
-8.96520793e-01 -5.44443011e-01 5.89669168e-01 9.96307671e-01
5.00737071e-01 -8.85965228e-01 -1.43564987e+00 2.46589959e-01
7.58021623e-02 -1.44373703e+00 -3.24664354e-01 2.22386301e-01
-8.22102904e-01 -1.25184572e+00 -2.00303063e-01 -6.68635964e-01
6.64284110e-01 4.69239086e-01 8.98131669e-01 4.89432573e-01
-3.97917181e-01 4.33846802e-01 9.34143271e-03 -7.02527821e-01
-4.24668044e-01 3.58279236e-02 2.09302828e-01 -1.42686903e-01
5.17546237e-01 -2.39940345e-01 -5.04557133e-01 4.34140205e-01
-4.32765990e-01 3.45333010e-01 7.14981377e-01 6.06084406e-01
3.02091599e-01 -1.11363027e-02 2.67206490e-01 -4.66961712e-01
-2.19327480e-01 -2.20565230e-01 -1.28210866e+00 3.99240792e-01
-5.33651829e-01 -7.93025363e-03 5.32725453e-01 -8.28277290e-01
-9.69927013e-01 3.54274273e-01 -1.00460148e-03 -7.58710265e-01
-7.73879141e-02 -1.89098030e-01 3.53655331e-02 -3.04205328e-01
2.51179487e-01 6.37106821e-02 -1.02960169e-01 -3.64241332e-01
2.28630632e-01 3.77741754e-01 6.24024034e-01 -4.39205199e-01
1.00983834e+00 4.95390177e-01 1.53618917e-01 -8.72854710e-01
-8.25330079e-01 -3.84169102e-01 -7.56356180e-01 -6.50281906e-01
9.62973773e-01 -1.14165878e+00 -1.30845571e+00 4.89278346e-01
-1.11008883e+00 -5.58160305e-01 -9.14236009e-02 5.47367036e-01
-2.07361728e-01 1.02698579e-01 -4.38335687e-01 -7.01532066e-01
-4.95159626e-01 -1.23772252e+00 1.02953184e+00 7.86646605e-01
1.00984745e-01 -9.66438890e-01 -9.76844132e-02 3.98873746e-01
4.19524878e-01 -2.60717589e-02 5.73992372e-01 -6.07481420e-01
-1.10043728e+00 -4.95051928e-02 -4.43378687e-01 -4.89360541e-02
-1.60306841e-01 1.82015672e-01 -1.27685547e+00 -6.97310627e-01
-4.64353263e-01 -3.14826876e-01 7.32120156e-01 5.62578380e-01
9.89411473e-01 -1.66268736e-01 -8.84007156e-01 6.53438747e-01
1.12020242e+00 1.29489407e-01 1.55842483e-01 2.69064546e-01
9.29660976e-01 1.42051131e-01 5.89408398e-01 2.36462504e-01
6.47763252e-01 7.37780511e-01 5.58412731e-01 1.09981917e-01
-4.37249243e-01 -2.26684064e-01 6.50912344e-01 6.31478131e-01
2.27436706e-01 -3.09188422e-02 -9.83064413e-01 5.23717463e-01
-2.08202076e+00 -1.03247631e+00 -2.31326163e-01 1.91323972e+00
3.16805929e-01 3.30523401e-01 4.70230997e-01 -2.73463786e-01
6.44858241e-01 -8.94454196e-02 -8.77528667e-01 -2.52906159e-02
5.23128748e-01 -6.17951602e-02 5.49114645e-01 3.63608152e-01
-1.29803860e+00 1.11244333e+00 5.53170061e+00 4.90969300e-01
-1.14853072e+00 3.01415861e-01 8.46646875e-02 -2.51851827e-01
5.09039521e-01 2.77663134e-02 -1.55183876e+00 4.64114189e-01
1.39934337e+00 5.44452965e-02 1.76235706e-01 1.14304340e+00
-4.19986853e-03 7.57385567e-02 -1.24966776e+00 8.92669678e-01
-2.06880853e-01 -1.40527284e+00 -2.79445807e-03 6.05940968e-02
4.08375114e-01 3.76949131e-01 9.02717337e-02 6.41618252e-01
7.79437840e-01 -6.64038479e-01 9.70514476e-01 5.43806672e-01
2.70535409e-01 -5.85584819e-01 4.83426183e-01 4.52873498e-01
-1.76134515e+00 -2.31037840e-01 -4.85847086e-01 1.53111443e-01
-2.66292356e-02 5.95977604e-02 -1.02775002e+00 2.95731336e-01
1.07908046e+00 7.22469687e-01 -7.92885065e-01 1.16850102e+00
9.49339271e-02 6.56086564e-01 -4.08131301e-01 -2.97462922e-02
2.61399120e-01 2.35730395e-01 5.82915306e-01 1.04416752e+00
1.21878767e-02 -3.31547886e-01 5.92529118e-01 7.34933078e-01
7.17589855e-02 -7.14812040e-01 -3.28528762e-01 1.74953669e-01
6.27816617e-01 1.47539139e+00 -8.02757621e-01 -6.80601001e-01
-6.40886605e-01 4.44362670e-01 3.91558886e-01 -1.29792914e-02
-1.37943697e+00 4.85080853e-02 9.48427200e-01 -7.97865465e-02
9.79477167e-01 -2.03206509e-01 5.19037843e-02 -6.95305347e-01
-1.99662268e-01 -6.11791730e-01 4.39133376e-01 -6.26778722e-01
-7.61722386e-01 4.24100012e-01 2.62889434e-02 -1.08818722e+00
-1.74115166e-01 -8.48180175e-01 -6.55935168e-01 6.24873936e-01
-1.70827460e+00 -1.41145766e+00 -6.35462224e-01 6.32652760e-01
4.74252164e-01 -3.45443860e-02 3.31599027e-01 6.53534293e-01
-1.01177740e+00 7.89707601e-01 -9.34512466e-02 3.33815426e-01
2.91075110e-01 -9.48136151e-01 6.11828506e-01 9.01852965e-01
4.20080759e-02 2.07611606e-01 5.98132849e-01 -7.08196580e-01
-1.56773138e+00 -1.68213952e+00 4.94676769e-01 -7.32160568e-01
6.10273123e-01 -2.22680688e-01 -8.39538276e-01 1.05740142e+00
1.61739916e-01 4.05999094e-01 4.06172246e-01 -1.59323886e-01
-1.72863305e-01 -2.97477156e-01 -9.81574476e-01 5.58885992e-01
8.00987065e-01 -2.62119770e-01 -3.31693858e-01 7.74343386e-02
6.07717514e-01 -4.95403707e-01 -8.77197862e-01 2.03583404e-01
9.85881627e-01 -6.26945317e-01 1.03211021e+00 -7.98765063e-01
-3.37067872e-01 -5.92520297e-01 -5.47304973e-02 -7.82553434e-01
-3.55340749e-01 -5.76106250e-01 -8.49584281e-01 1.10596657e+00
1.34944487e-02 -3.39391619e-01 1.15514719e+00 5.02911091e-01
-2.08488509e-01 -5.42438567e-01 -8.81429613e-01 -8.05718601e-01
-3.07436585e-01 -3.17997694e-01 5.39363205e-01 4.32993323e-01
-7.30580866e-01 3.10973644e-01 -3.13036650e-01 6.96634114e-01
1.02690649e+00 1.00658104e-01 1.19489443e+00 -1.54764378e+00
-2.05821097e-01 -2.78028131e-01 -5.24026990e-01 -1.23289299e+00
6.96486086e-02 -6.61890626e-01 1.18413724e-01 -1.16535616e+00
3.16745669e-01 -5.64844847e-01 -2.43136242e-01 5.96101999e-01
-2.50463396e-01 4.09082370e-03 3.80427718e-01 1.22310862e-01
-1.10896289e+00 4.51840252e-01 8.90299439e-01 -2.47258916e-01
-5.81126176e-02 1.94643855e-01 -2.66576767e-01 9.13399279e-01
4.81330425e-01 -8.03726017e-01 -3.23644042e-01 -7.26001263e-01
-2.94082016e-01 -1.10299513e-01 8.99645507e-01 -1.51405227e+00
8.33390355e-01 1.05256215e-01 6.98607504e-01 -1.08871663e+00
4.67564285e-01 -1.18623066e+00 1.09260008e-01 1.03495741e+00
-3.82878482e-02 4.64135885e-01 6.97674394e-01 7.41317570e-01
4.13277119e-01 4.18716073e-02 1.02564263e+00 1.94989026e-01
-1.08803701e+00 4.62443918e-01 -3.43485087e-01 -1.09368429e-01
1.18815112e+00 -3.10977399e-01 -6.29607677e-01 1.76017761e-01
-6.28543198e-01 5.75176239e-01 2.53398895e-01 6.05583668e-01
5.42242765e-01 -1.17466927e+00 -3.78456861e-01 2.35652164e-01
-1.86133623e-01 -7.23427981e-02 1.72360539e-01 1.07880425e+00
-2.91314185e-01 5.42125881e-01 -4.01356995e-01 -1.22107828e+00
-1.80364954e+00 5.56401551e-01 4.22045112e-01 -4.09790725e-01
-8.25271904e-01 8.18878412e-01 1.55723065e-01 -2.48260304e-01
6.50479436e-01 -4.07191962e-01 -7.96528608e-02 -2.67587692e-01
7.03435779e-01 6.97937548e-01 -8.37691501e-02 -6.24425232e-01
-5.27227104e-01 4.24665511e-01 -5.91897368e-01 4.00497556e-01
1.16690242e+00 7.15380860e-03 4.59754676e-01 7.79604614e-02
8.21812689e-01 -7.57452011e-01 -1.76709843e+00 -3.73952061e-01
2.00036421e-01 -2.89186865e-01 3.02880198e-01 -4.81914103e-01
-1.18976462e+00 9.21293676e-01 1.09753644e+00 -2.32932180e-01
7.95473099e-01 -5.02141789e-02 9.53829765e-01 7.77091026e-01
2.77616054e-01 -9.69971001e-01 1.62910879e-01 3.58615994e-01
9.68470797e-02 -1.20063138e+00 1.68997683e-02 -1.31901251e-02
-3.68722260e-01 9.13170516e-01 1.21030235e+00 6.49643317e-03
6.16440892e-01 5.10616243e-01 6.67721480e-02 -2.62625396e-01
-1.16504490e+00 -1.97545275e-01 3.05294156e-01 7.24714577e-01
-3.66562679e-02 -2.63599277e-01 7.29753852e-01 9.92928520e-02
1.96292683e-01 7.85454437e-02 7.36724585e-02 1.15605950e+00
-6.63654745e-01 -6.53939128e-01 -5.04091918e-01 5.52577496e-01
-1.98285326e-01 1.74774811e-01 1.76290870e-02 8.94842029e-01
2.64575422e-01 9.56752837e-01 2.44033337e-01 -4.64581996e-01
1.85082987e-01 5.11390856e-03 5.67440033e-01 -2.31381536e-01
-7.09416032e-01 -1.71733633e-01 -9.79273543e-02 -6.13103449e-01
-4.13102627e-01 -7.74879277e-01 -1.16667902e+00 -4.95749295e-01
-5.13846517e-01 -3.29947621e-02 6.54472888e-01 1.10741401e+00
5.28079331e-01 9.84137356e-01 5.09088896e-02 -9.50286686e-01
-5.40095091e-01 -7.69954085e-01 8.26497972e-02 -2.26520255e-01
6.51565254e-01 -1.03030682e+00 1.75881192e-01 2.12337658e-01] | [6.324197292327881, -2.074770927429199] |
2768d521-b925-497f-8daf-18a061efacdd | lmpriors-pre-trained-language-models-as-task | 2210.12530 | null | https://arxiv.org/abs/2210.12530v1 | https://arxiv.org/pdf/2210.12530v1.pdf | LMPriors: Pre-Trained Language Models as Task-Specific Priors | Particularly in low-data regimes, an outstanding challenge in machine learning is developing principled techniques for augmenting our models with suitable priors. This is to encourage them to learn in ways that are compatible with our understanding of the world. But in contrast to generic priors such as shrinkage or sparsity, we draw inspiration from the recent successes of large-scale language models (LMs) to construct task-specific priors distilled from the rich knowledge of LMs. Our method, Language Model Priors (LMPriors), incorporates auxiliary natural language metadata about the task -- such as variable names and descriptions -- to encourage downstream model outputs to be consistent with the LM's common-sense reasoning based on the metadata. Empirically, we demonstrate that LMPriors improve model performance in settings where such natural language descriptions are available, and perform well on several tasks that benefit from such prior knowledge, such as feature selection, causal inference, and safe reinforcement learning. | ['Stefano Ermon', 'Sanjari Srivastava', 'Chris Cundy', 'Kristy Choi'] | 2022-10-22 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [ 2.54176855e-01 4.08652991e-01 -8.06613088e-01 -6.82596922e-01
-8.65095913e-01 -3.88625860e-01 1.08452928e+00 1.84916109e-01
-5.44389725e-01 9.12960351e-01 8.83898914e-01 -3.65063727e-01
-1.53120056e-01 -5.97786605e-01 -9.23438311e-01 -2.81050056e-01
-3.28349099e-02 4.60305184e-01 -2.79062539e-01 -5.44802062e-02
2.84967095e-01 9.91161764e-02 -1.22916603e+00 1.37679711e-01
7.44278610e-01 4.84669179e-01 4.10266429e-01 2.16892123e-01
-2.01745883e-01 1.13833570e+00 -2.86196060e-02 -3.20229828e-01
9.52312276e-02 -1.35085925e-01 -6.78745151e-01 -9.35143307e-02
9.81418490e-02 -1.56031519e-01 -5.53963125e-01 9.22912359e-01
4.30652834e-02 4.60055143e-01 9.06175792e-01 -1.17510808e+00
-8.44572067e-01 1.19273674e+00 -4.79779601e-01 2.44745091e-01
6.80061951e-02 5.75989366e-01 1.46008241e+00 -7.57490754e-01
7.50233769e-01 1.67048049e+00 5.92670143e-01 5.77673674e-01
-1.66703689e+00 -7.37659633e-01 7.78555155e-01 -1.73912734e-01
-1.08309746e+00 -7.63372004e-01 7.10478127e-01 -6.87984586e-01
9.58268344e-01 -4.75795753e-02 1.35094017e-01 1.57007217e+00
2.38527894e-01 7.78400600e-01 9.52634633e-01 -5.77525437e-01
3.60095143e-01 2.60529160e-01 3.42812389e-01 5.47876239e-01
2.75364816e-01 2.25842580e-01 -1.14040995e+00 -4.08231199e-01
7.79759347e-01 -2.62196243e-01 -4.28631045e-02 -2.92859048e-01
-1.27448833e+00 1.11335707e+00 8.67652297e-02 -7.01729134e-02
-6.19947672e-01 5.94556987e-01 1.29548624e-01 5.62240034e-02
6.30816579e-01 9.06908214e-01 -9.07024503e-01 -6.53380677e-02
-7.78488636e-01 2.42756814e-01 6.97333813e-01 1.06773293e+00
8.88364136e-01 -2.69062594e-02 -2.69191086e-01 7.52704680e-01
8.13890159e-01 5.26153147e-01 5.71996987e-01 -1.13081467e+00
4.90505666e-01 2.09489167e-01 2.24262998e-01 -6.54059291e-01
-2.70267904e-01 -4.76542354e-01 -3.86872470e-01 -2.30963334e-01
3.44843417e-01 -3.70939553e-01 -1.01629531e+00 2.52266312e+00
-1.35396356e-02 3.07619870e-01 1.08363666e-01 5.23936152e-01
4.87647623e-01 4.18076932e-01 8.00214350e-01 -1.44493058e-01
1.10951078e+00 -5.23044229e-01 -5.82354903e-01 -9.75237548e-01
7.06556797e-01 -4.38108981e-01 1.30081999e+00 1.76338315e-01
-9.61766958e-01 -3.32031399e-01 -5.57021141e-01 -1.89254537e-01
-8.02401379e-02 -1.15132686e-02 1.20171118e+00 1.79131806e-01
-9.23841298e-01 4.00462568e-01 -9.49573100e-01 -3.58100563e-01
5.39615452e-01 1.86509952e-01 -2.96094626e-01 8.68434906e-02
-1.40968823e+00 1.08817351e+00 4.34434980e-01 -5.17236777e-02
-1.31606066e+00 -8.91706288e-01 -1.09506929e+00 1.52668804e-01
6.36881292e-01 -1.02933407e+00 1.62870693e+00 -9.54558909e-01
-1.43495703e+00 6.44817829e-01 -4.55268204e-01 -5.63376844e-01
2.23159865e-01 -5.07079661e-01 4.49655838e-02 -4.36855525e-01
1.78934842e-01 8.92067134e-01 8.78441632e-01 -1.08596909e+00
-5.86631358e-01 -5.92813008e-02 1.17683239e-01 2.59739071e-01
-4.94718224e-01 4.75759096e-02 -3.50995690e-01 -6.20313883e-01
-1.82036266e-01 -7.13237822e-01 -7.55124390e-01 7.85776973e-02
-6.73579752e-01 -4.30044562e-01 2.30176404e-01 -4.45834547e-01
1.00616348e+00 -2.04121614e+00 5.98999038e-02 3.23589742e-01
2.25459173e-01 -1.55205920e-01 -2.46952564e-01 3.19341749e-01
1.10633830e-02 3.14402133e-01 -2.28090867e-01 -5.80415487e-01
2.81732708e-01 4.47423160e-01 -6.81814313e-01 3.83051574e-01
4.26094532e-01 7.69858122e-01 -1.06494200e+00 -3.96223277e-01
-2.70520840e-02 1.57637835e-01 -7.93649793e-01 4.05596793e-01
-8.63972545e-01 3.70110869e-01 -7.89847195e-01 1.07419312e-01
7.50208348e-02 -5.80994546e-01 3.24147582e-01 1.13379948e-01
4.08548713e-02 8.33751619e-01 -1.11661935e+00 1.74411094e+00
-6.28303945e-01 5.28423250e-01 1.55989677e-01 -9.61599112e-01
5.94060600e-01 1.90541014e-01 2.01186851e-01 -1.70810893e-01
-2.33138755e-01 -1.53800428e-01 -9.70913470e-02 -6.82166755e-01
2.09775686e-01 -4.14455622e-01 6.45081699e-02 6.23855412e-01
3.01024705e-01 -1.67842403e-01 1.49920732e-01 5.51991940e-01
9.99893248e-01 4.91965652e-01 5.55919707e-01 -4.61532742e-01
5.06510064e-02 -7.45749250e-02 8.17030728e-01 1.24112415e+00
1.47678331e-01 -9.77612883e-02 5.45773149e-01 -1.35557905e-01
-7.82887936e-01 -1.12069511e+00 -2.43172690e-01 1.44383514e+00
-5.32223821e-01 -6.46449149e-01 -1.67545035e-01 -5.93216360e-01
3.15493584e-01 1.20742762e+00 -5.65993905e-01 -2.49928266e-01
-1.96421355e-01 -8.65432084e-01 3.85745078e-01 7.24377692e-01
6.70983940e-02 -1.02655375e+00 -2.85259277e-01 3.13230813e-01
4.84764315e-02 -1.05518770e+00 -2.45968267e-01 5.87234318e-01
-8.87835860e-01 -8.96066129e-01 -1.44762516e-01 -3.05217534e-01
6.54325366e-01 -3.61218095e-01 1.42137992e+00 -3.87710333e-01
-1.28221273e-01 4.47058141e-01 7.18441978e-02 -4.71244812e-01
-4.38010514e-01 -1.47318825e-01 2.63276279e-01 -2.59026051e-01
5.06245971e-01 -6.63576424e-01 -6.51057586e-02 -2.40573809e-01
-7.09163964e-01 1.72416329e-01 7.14220881e-01 8.62499356e-01
3.00459176e-01 -2.27131739e-01 7.40380228e-01 -1.49520397e+00
6.20569050e-01 -8.50024402e-01 -6.76084340e-01 2.17573002e-01
-8.05514514e-01 5.45775533e-01 4.14000928e-01 -5.40194869e-01
-1.57681680e+00 -9.69447102e-03 1.83233395e-01 -1.50152475e-01
-2.86052912e-01 1.01263976e+00 -9.41184163e-02 5.81547856e-01
8.36099625e-01 -3.70747074e-02 -1.65013388e-01 -7.31042802e-01
8.82025301e-01 4.17816430e-01 4.50118661e-01 -1.17760575e+00
7.89123595e-01 2.03494236e-01 -3.89411189e-02 -6.78078115e-01
-1.38826680e+00 -2.26688311e-01 -5.08853376e-01 3.90340060e-01
6.80566430e-01 -1.31902003e+00 -3.89966011e-01 4.94306758e-02
-9.53548372e-01 -8.61947477e-01 -2.95791298e-01 8.28055859e-01
-5.74849844e-01 -2.27887575e-02 -6.30908489e-01 -1.00998950e+00
1.25556186e-01 -9.33558583e-01 7.31453180e-01 1.71062231e-01
-6.09354138e-01 -1.49408472e+00 2.41126101e-02 5.22857113e-03
4.38538253e-01 -2.15089396e-01 1.34017873e+00 -9.36470091e-01
-5.45497239e-01 3.65077198e-01 9.70506296e-02 2.55492091e-01
1.76973805e-01 1.09773770e-01 -1.13573515e+00 6.74779108e-03
-9.30165723e-02 -5.24257720e-01 1.11720645e+00 6.79577053e-01
1.32089746e+00 -5.20140350e-01 -5.64048171e-01 5.27373016e-01
1.16447163e+00 -2.68341333e-01 -1.67588562e-01 5.03167212e-02
4.98893172e-01 7.44091570e-01 3.47831875e-01 7.54335105e-01
4.18984622e-01 2.71987259e-01 4.57934365e-02 -4.07149382e-02
7.02095479e-02 -8.03962052e-01 4.75714356e-01 4.50238556e-01
4.09024686e-01 -1.78939477e-01 -1.02931917e+00 5.49809515e-01
-1.94671595e+00 -7.59763300e-01 1.43736064e-01 2.04732919e+00
1.50620699e+00 3.13203633e-01 -2.19172582e-01 -7.22680330e-01
5.25830328e-01 3.08246076e-01 -8.56950223e-01 -5.41624147e-03
-6.27298430e-02 1.74034402e-01 4.52137202e-01 7.51608491e-01
-1.13985026e+00 1.15516293e+00 7.26353264e+00 6.00868762e-01
-7.68827260e-01 6.80949688e-02 6.53867543e-01 -1.56935036e-01
-8.47467363e-01 4.89549935e-01 -1.16050541e+00 2.92298943e-01
1.11923587e+00 -3.62993866e-01 4.86088723e-01 7.74172008e-01
5.28200269e-01 -1.60844252e-01 -1.65805042e+00 4.59283739e-01
-1.42240822e-01 -1.40276551e+00 1.17956862e-01 -7.06920251e-02
7.98639476e-01 4.12014127e-01 2.08881855e-01 5.10444880e-01
1.53933871e+00 -1.07573545e+00 7.05591202e-01 6.05058193e-01
5.80955923e-01 -3.42205882e-01 2.28128776e-01 6.43415332e-01
-6.44569874e-01 -2.54666179e-01 -4.79072183e-01 -3.03675234e-01
7.59001672e-02 9.28354204e-01 -8.68175685e-01 2.70361807e-02
1.94286585e-01 1.09069824e+00 -4.33950394e-01 6.98452652e-01
-7.91578650e-01 1.17260420e+00 -2.92761475e-01 2.41464198e-01
2.36962467e-01 1.28451705e-01 3.60032767e-01 1.28874612e+00
-1.43122867e-01 -8.50890204e-03 5.55918813e-01 1.39429653e+00
-4.34579581e-01 -2.67219134e-02 -9.00483966e-01 -3.26693565e-01
7.98271954e-01 9.82340038e-01 -1.03242777e-01 -4.65277821e-01
-6.34332359e-01 1.40308484e-01 5.75670838e-01 7.40193903e-01
-5.15192866e-01 2.34204456e-01 8.57075751e-01 -2.38621518e-01
-7.61357024e-02 -2.56690532e-01 -3.78291965e-01 -1.25896537e+00
-3.22724581e-01 -9.38900411e-01 5.91562331e-01 -8.68798852e-01
-1.84285867e+00 -2.14041751e-02 3.58639032e-01 -4.35608476e-01
-4.60892648e-01 -5.93550205e-01 -5.00391603e-01 1.15191042e+00
-1.63144076e+00 -1.25921178e+00 5.40707290e-01 5.71944714e-01
5.77047944e-01 -1.21695653e-01 7.82819867e-01 -2.73741663e-01
-5.21832943e-01 3.13715756e-01 6.18239753e-02 5.93466721e-02
1.02340436e+00 -1.29055476e+00 3.87370229e-01 9.19166803e-01
4.50361192e-01 1.37975752e+00 9.05327260e-01 -9.41363096e-01
-1.42959917e+00 -1.01391757e+00 8.63891780e-01 -7.30135798e-01
1.03728843e+00 -4.17787969e-01 -6.74096286e-01 1.44913137e+00
1.18248843e-01 -6.34272248e-02 9.04573441e-01 1.05270159e+00
-7.10699916e-01 1.14112481e-01 -7.67084360e-01 6.25792325e-01
1.01833165e+00 -8.46811414e-01 -9.63935316e-01 6.74032748e-01
6.90653026e-01 -6.94011822e-02 -6.73882604e-01 9.68924090e-02
1.78856924e-01 -1.15879342e-01 8.07618976e-01 -1.50305271e+00
6.04404986e-01 -4.74503972e-02 -3.54462028e-01 -1.58765960e+00
-4.30248260e-01 -9.45509493e-01 -9.63124260e-02 1.33944416e+00
7.58160174e-01 -5.56595266e-01 4.80315506e-01 1.28393102e+00
-1.21736228e-01 -2.99889803e-01 -5.52096069e-01 -6.19523704e-01
3.37749124e-01 -6.85059071e-01 3.16252470e-01 1.14862955e+00
1.86101645e-01 5.20796955e-01 -3.12630415e-01 2.92821050e-01
6.31829202e-01 -2.23476253e-03 6.59955502e-01 -1.55040956e+00
-5.92812657e-01 -2.77414739e-01 2.57077217e-01 -1.25840974e+00
9.44572508e-01 -1.05841434e+00 3.44343513e-01 -1.35200727e+00
7.67558038e-01 -7.28627443e-01 -4.84496176e-01 9.42941904e-01
-5.67819238e-01 -4.14802790e-01 5.98395383e-03 1.44845575e-01
-6.14895701e-01 5.58518410e-01 9.36009705e-01 1.34538263e-01
-9.44136903e-02 -1.28196090e-01 -1.22441709e+00 1.19290423e+00
4.57301736e-01 -6.99250460e-01 -6.24562442e-01 -5.72763860e-01
4.17836875e-01 6.60777688e-02 3.18462402e-01 -2.65823424e-01
3.53997469e-01 -8.58021140e-01 3.11572373e-01 -6.33199066e-02
2.87369013e-01 -4.60166276e-01 -1.68812305e-01 3.42485905e-02
-1.41554916e+00 -1.40062034e-01 1.15674473e-01 8.77078950e-01
7.81389922e-02 -3.01484704e-01 4.90959078e-01 -4.14208114e-01
-9.08267140e-01 2.78234571e-01 -5.46502233e-01 6.32955432e-01
5.56649923e-01 6.41086459e-01 -4.91447777e-01 -6.35889530e-01
-8.29717517e-01 4.25231844e-01 2.77475268e-01 3.63041967e-01
3.70576948e-01 -9.82991755e-01 -8.00336659e-01 1.85934573e-01
1.49711445e-01 -1.27285405e-03 -2.93684959e-01 6.12829804e-01
4.92058814e-01 6.46365702e-01 1.56418443e-01 -2.96124697e-01
-5.62863529e-01 5.07885337e-01 1.06024683e-01 -3.31417233e-01
-5.51296473e-01 1.18575835e+00 7.18968868e-01 -3.68720621e-01
3.65852594e-01 -4.08937544e-01 9.35905846e-05 -2.40719780e-01
4.44602132e-01 -3.12440097e-01 -4.32920367e-01 1.23001076e-02
-2.17508376e-01 -1.21107005e-01 -2.16131076e-01 -5.11086702e-01
1.56593037e+00 -1.13772742e-01 -2.50713099e-02 7.29758501e-01
6.32141531e-01 2.26227731e-01 -1.80326629e+00 -7.42855966e-01
4.97019082e-01 -4.66207206e-01 3.13022077e-01 -9.12807763e-01
-5.26871562e-01 7.97469437e-01 -1.51081830e-01 -3.58108252e-01
5.83384454e-01 6.48625642e-02 1.31449610e-01 8.08222413e-01
3.44795614e-01 -9.83677268e-01 3.05038150e-02 6.26231611e-01
6.08822286e-01 -1.29569352e+00 6.51169792e-02 -1.19661510e-01
-7.79361784e-01 8.21905851e-01 6.76765501e-01 7.35311136e-02
8.60261321e-01 4.57060486e-01 -7.14137927e-02 -1.07898243e-01
-1.46567559e+00 -1.57514408e-01 2.32825428e-01 5.99530518e-01
7.56347477e-01 -2.12995280e-02 6.31029457e-02 7.38544822e-01
-3.20978789e-03 1.14813916e-01 4.17475849e-01 6.46806598e-01
-4.42193449e-01 -1.18022406e+00 2.27685682e-02 8.19338918e-01
-4.84316885e-01 -6.33322477e-01 -1.05212867e-01 6.76122487e-01
-1.53245449e-01 1.08144784e+00 -3.80184837e-02 9.23526734e-02
-2.68438756e-01 4.48479742e-01 1.62595972e-01 -1.31505466e+00
-3.63502890e-01 7.20283613e-02 4.11282539e-01 -6.57227993e-01
-2.77802050e-01 -7.44394243e-01 -1.16182959e+00 2.15164293e-02
-1.54634207e-01 1.50075495e-01 5.19292414e-01 1.22218037e+00
1.69677332e-01 3.43103588e-01 3.10244679e-01 -4.34079140e-01
-1.22885227e+00 -1.00678647e+00 -6.13148391e-01 5.74218094e-01
3.40394378e-01 -7.68053532e-01 -3.93308312e-01 4.32135403e-01] | [10.456806182861328, 8.0504150390625] |
eb83ac8f-3218-4090-88fe-b05ca043ec3d | joint-metrics-matter-a-better-standard-for | 2305.06292 | null | https://arxiv.org/abs/2305.06292v1 | https://arxiv.org/pdf/2305.06292v1.pdf | Joint Metrics Matter: A Better Standard for Trajectory Forecasting | Multi-modal trajectory forecasting methods commonly evaluate using single-agent metrics (marginal metrics), such as minimum Average Displacement Error (ADE) and Final Displacement Error (FDE), which fail to capture joint performance of multiple interacting agents. Only focusing on marginal metrics can lead to unnatural predictions, such as colliding trajectories or diverging trajectories for people who are clearly walking together as a group. Consequently, methods optimized for marginal metrics lead to overly-optimistic estimations of performance, which is detrimental to progress in trajectory forecasting research. In response to the limitations of marginal metrics, we present the first comprehensive evaluation of state-of-the-art (SOTA) trajectory forecasting methods with respect to multi-agent metrics (joint metrics): JADE, JFDE, and collision rate. We demonstrate the importance of joint metrics as opposed to marginal metrics with quantitative evidence and qualitative examples drawn from the ETH / UCY and Stanford Drone datasets. We introduce a new loss function incorporating joint metrics that, when applied to a SOTA trajectory forecasting method, achieves a 7% improvement in JADE / JFDE on the ETH / UCY datasets with respect to the previous SOTA. Our results also indicate that optimizing for joint metrics naturally leads to an improvement in interaction modeling, as evidenced by a 16% decrease in mean collision rate on the ETH / UCY datasets with respect to the previous SOTA. | ['Kris Kitani', 'Deva Ramanan', 'Hana Hoshino', 'Erica Weng'] | 2023-05-10 | null | null | null | null | ['trajectory-forecasting'] | ['computer-vision'] | [-3.50953281e-01 -2.13682115e-01 6.07452542e-02 -1.22126147e-01
-6.81867480e-01 -5.36058009e-01 1.16681767e+00 3.73337328e-01
-7.20817208e-01 1.05314910e+00 4.57558990e-01 -6.11396953e-02
-5.06273985e-01 -8.29141080e-01 -5.43521166e-01 -5.90052009e-01
-6.36449695e-01 6.02156460e-01 2.95792818e-01 -4.61328894e-01
8.56032688e-03 5.07363796e-01 -1.72252452e+00 -2.40610823e-01
8.47240031e-01 5.01072347e-01 -3.33751887e-01 7.09398031e-01
4.62785274e-01 5.53125024e-01 -4.97692138e-01 -5.51319420e-01
7.36863092e-02 -2.64459729e-01 -6.14604294e-01 -4.23747301e-01
4.98994112e-01 -4.55394298e-01 -2.96225399e-01 4.03465718e-01
3.92496735e-01 5.57959020e-01 1.18158364e+00 -2.00113297e+00
-1.62513584e-01 4.23833191e-01 -1.62787795e-01 2.42886767e-01
4.77857828e-01 4.26717967e-01 9.71734703e-01 -4.65407997e-01
7.15414226e-01 1.21779358e+00 1.42533791e+00 1.77331448e-01
-1.39763451e+00 -3.34854990e-01 2.18966797e-01 -5.79892145e-03
-1.49410808e+00 -3.24782133e-01 2.49059677e-01 -7.50100851e-01
1.15919077e+00 3.58738273e-01 4.92082685e-01 1.06179702e+00
3.31546307e-01 6.91286147e-01 5.67877054e-01 1.77161902e-01
2.03328684e-01 -1.92971885e-01 1.20398961e-01 3.78478080e-01
4.00840133e-01 6.37130618e-01 -1.80387005e-01 -3.53342265e-01
2.64366180e-01 -2.06825897e-01 7.00961724e-02 -1.83667883e-01
-1.46178365e+00 6.56753302e-01 3.78510714e-01 6.85304701e-02
-6.11842036e-01 4.07955736e-01 4.15537089e-01 9.73691326e-03
5.56729257e-01 2.97877043e-01 -7.83137754e-02 -8.56798053e-01
-9.59829569e-01 1.34952128e+00 7.14849889e-01 7.88301170e-01
5.59663296e-01 -8.55560973e-02 -2.77318418e-01 4.98644888e-01
2.33569175e-01 8.55571806e-01 -1.89905658e-01 -1.16828263e+00
5.63181639e-01 6.43197656e-01 7.13655174e-01 -1.43233621e+00
-9.59112585e-01 -1.90039143e-01 -5.37231207e-01 3.63512456e-01
7.49291122e-01 -4.46135551e-01 -4.53879148e-01 1.96773112e+00
3.14378709e-01 2.20066920e-01 4.95912768e-02 9.39695656e-01
6.42926753e-01 5.73350370e-01 3.57071400e-01 3.16692293e-02
7.80961871e-01 -7.86420643e-01 -4.04558659e-01 -2.87606139e-02
1.18856645e+00 -3.66032869e-01 9.71760571e-01 6.54869080e-02
-9.22267318e-01 -3.71856004e-01 -8.50611091e-01 2.89255947e-01
-5.92651486e-01 -6.88491687e-02 4.42086399e-01 6.19313836e-01
-9.07003999e-01 6.54309332e-01 -1.09926713e+00 -5.59015274e-01
2.92396575e-01 2.12558106e-01 -2.05697343e-01 2.63410151e-01
-1.03727615e+00 1.21123075e+00 -4.59672436e-02 -1.71474889e-01
-7.40282297e-01 -1.07159233e+00 -7.02613175e-01 -1.57108992e-01
-5.98759216e-04 -4.53900278e-01 9.66057122e-01 -1.21747799e-01
-1.13232899e+00 3.20162058e-01 4.40514907e-02 -7.02855945e-01
9.52086866e-01 -3.07461202e-01 -6.03709280e-01 -1.72931790e-01
2.16313556e-01 9.77173924e-01 5.65724671e-02 -1.20617318e+00
-8.90494764e-01 -8.57941881e-02 1.35760307e-01 2.70705253e-01
-1.75564326e-02 -2.59194285e-01 2.50867337e-01 -5.31889677e-01
-5.47509313e-01 -1.22431564e+00 -1.82148293e-01 -1.05073892e-01
-2.01521024e-01 -5.64931512e-01 8.17490757e-01 -4.06339586e-01
1.52063680e+00 -1.77933121e+00 1.57877818e-01 6.55591413e-02
9.11421627e-02 1.85285449e-01 -1.53034955e-01 8.58117700e-01
4.72415388e-01 2.94532944e-02 -5.11214375e-01 -5.07856071e-01
3.60449702e-01 1.71258107e-01 -3.45758975e-01 6.62219524e-01
-2.93099228e-02 8.58255446e-01 -1.09789073e+00 -2.64392853e-01
3.19094211e-01 4.01714891e-01 -5.67603946e-01 -1.15888752e-01
-9.94972810e-02 2.46209651e-01 -1.97002798e-01 4.39471245e-01
6.00609899e-01 5.87683581e-02 -2.95864493e-01 1.41843379e-01
-5.23049295e-01 2.36916728e-02 -9.45318103e-01 1.23405278e+00
-1.21473275e-01 8.43733728e-01 -3.33849818e-01 -4.92042303e-01
7.18383074e-01 1.40377328e-01 7.75503218e-01 -5.72051704e-01
-4.54462208e-02 2.05463335e-01 -4.60259281e-02 -2.44410619e-01
8.29377413e-01 1.23117641e-01 -3.37117285e-01 6.87756538e-01
-4.13638175e-01 1.70490649e-02 3.84208232e-01 2.94143677e-01
1.06687164e+00 2.66442716e-01 -1.87126711e-01 -4.84454632e-01
1.66348979e-01 5.95302939e-01 2.50302583e-01 6.62518799e-01
-4.28582072e-01 5.09922922e-01 3.90930593e-01 -6.18985534e-01
-1.26809812e+00 -1.17769623e+00 -1.74765870e-01 8.69464576e-01
4.26086336e-01 -6.38513207e-01 -7.30201662e-01 -7.04227090e-01
5.07787406e-01 9.41588402e-01 -6.87829614e-01 -1.34320363e-01
-6.90360129e-01 -9.54092801e-01 1.24813640e+00 6.50280535e-01
2.92598933e-01 -7.31885433e-01 -9.76321578e-01 3.10908169e-01
-6.08109906e-02 -1.02903271e+00 -3.86116445e-01 -5.01115918e-01
-5.43812156e-01 -9.14744258e-01 -8.25560033e-01 5.67918308e-02
3.10869426e-01 2.56164908e-01 1.05059147e+00 1.53306052e-01
7.99588934e-02 4.53508019e-01 -3.81559849e-01 -1.74136907e-01
-2.96332002e-01 1.15105726e-01 4.66566920e-01 -3.03200126e-01
3.36521655e-01 -5.47249377e-01 -7.66424954e-01 8.36217225e-01
-4.94951218e-01 -1.39718741e-01 4.51913513e-02 5.92155457e-01
5.53723536e-02 -2.56855607e-01 8.30812693e-01 -4.92202103e-01
7.89823234e-01 -8.62075865e-01 -4.64193702e-01 -2.85484996e-02
-9.19552624e-01 -1.64777115e-01 4.84705955e-01 -4.48621124e-01
-8.19060385e-01 -3.73010099e-01 -9.63465776e-03 -2.12277204e-01
-2.73254544e-01 6.12625957e-01 5.96690238e-01 1.45411953e-01
8.78991008e-01 -9.67935249e-02 1.81458458e-01 -1.15855135e-01
3.57542664e-01 5.00083923e-01 5.13543487e-01 -5.32611132e-01
5.04868448e-01 6.29447281e-01 2.02228129e-01 -8.28897119e-01
-2.35747620e-01 -1.75476849e-01 -6.56624615e-01 -5.62152326e-01
7.96365321e-01 -7.28385389e-01 -1.32658803e+00 6.16774678e-01
-8.05282354e-01 -7.43134558e-01 -8.43405947e-02 7.02247202e-01
-7.38223016e-01 2.25749061e-01 -2.33719200e-01 -9.81739461e-01
9.60258543e-02 -1.13336444e+00 1.07937729e+00 2.19625030e-02
-6.96062088e-01 -1.37004018e+00 5.62336981e-01 1.66572016e-02
6.08278930e-01 8.22222352e-01 7.08522081e-01 -3.92756224e-01
-1.96064517e-01 -3.33734989e-01 -2.05172136e-01 -2.25693390e-01
-9.32684839e-02 1.56846985e-01 -4.56188202e-01 -3.25322837e-01
-9.90367115e-01 -1.17350563e-01 6.74604058e-01 4.09228116e-01
2.73963004e-01 -2.27857828e-01 -7.77134538e-01 1.94462895e-01
1.06542385e+00 1.87317789e-01 4.54566807e-01 3.77167493e-01
6.41750693e-01 6.05157197e-01 7.94450223e-01 5.67611694e-01
1.11468792e+00 1.14300168e+00 3.69985372e-01 2.88883120e-01
-1.98262185e-01 -3.39368761e-01 5.02320170e-01 2.58974701e-01
-4.54996914e-01 -6.75351858e-01 -1.37453222e+00 8.65063787e-01
-2.27683139e+00 -1.17136836e+00 -4.32592005e-01 2.16083169e+00
2.29256600e-01 5.60674928e-02 9.68663514e-01 -7.86181614e-02
4.20445740e-01 2.03161374e-01 -4.79669780e-01 -2.94931978e-01
-4.35469113e-02 -6.77545786e-01 6.28169000e-01 7.35177755e-01
-1.10091090e+00 7.02356100e-01 7.08959961e+00 7.17191756e-01
-8.27537954e-01 -9.09913480e-02 2.26249591e-01 -2.89051175e-01
-2.12560624e-01 -5.31346500e-02 -8.55705440e-01 7.67458022e-01
1.35030508e+00 -2.95618445e-01 4.72927570e-01 5.44324338e-01
3.56416315e-01 -3.38683248e-01 -1.18202984e+00 7.70761967e-01
-2.06144646e-01 -1.30853403e+00 -1.58575535e-01 4.04601127e-01
6.43960238e-01 2.53523499e-01 -7.15118740e-03 4.48169798e-01
5.62223196e-01 -1.13144493e+00 9.68942165e-01 8.34547698e-01
3.49417508e-01 -9.64672208e-01 7.80217469e-01 5.11794209e-01
-1.21440351e+00 9.92858112e-02 -7.31781721e-02 -3.15541536e-01
7.48156428e-01 1.71232074e-01 -8.29847038e-01 6.32923901e-01
6.25603557e-01 9.07126725e-01 -1.74208596e-01 1.07330990e+00
4.33701962e-01 6.10419750e-01 -7.65636742e-01 -2.20343038e-01
7.09229469e-01 -3.43985260e-01 1.10001671e+00 1.35307539e+00
4.45715338e-01 1.05104357e-01 1.04592368e-01 7.97519624e-01
4.11885530e-01 -2.67710984e-01 -8.39238703e-01 -4.29455601e-02
7.02365994e-01 9.32986498e-01 -5.16314209e-01 -9.55530480e-02
-2.13676214e-01 5.09635687e-01 2.85499871e-01 2.81850308e-01
-1.10993814e+00 -2.33552322e-01 1.05652499e+00 3.21607441e-01
-6.56130239e-02 -5.77808797e-01 -2.52655417e-01 -5.79569757e-01
-1.14530072e-01 -4.19237912e-01 2.73986667e-01 -3.98427188e-01
-1.27705324e+00 6.50049210e-01 6.49479210e-01 -1.54910779e+00
-6.07637703e-01 -2.08126485e-01 -7.07796633e-01 5.79236984e-01
-9.13397372e-01 -1.17800903e+00 -2.98719168e-01 1.47900641e-01
1.52805880e-01 -2.44184762e-01 6.13692403e-01 5.76233983e-01
-4.86572266e-01 7.26232767e-01 8.80865306e-02 -2.49268949e-01
3.89713794e-01 -1.13908613e+00 7.18628645e-01 5.83288729e-01
-1.63492054e-01 3.88363093e-01 1.05847001e+00 -7.37462461e-01
-1.06729209e+00 -1.12670946e+00 8.45592618e-01 -1.05569315e+00
5.74524760e-01 -1.21133544e-01 -5.93193412e-01 7.90210903e-01
-1.24436326e-01 -2.14872777e-01 3.01051497e-01 1.99677348e-01
1.35470897e-01 1.40687793e-01 -1.20859075e+00 9.19251561e-01
1.27340424e+00 -1.20948195e-01 -1.47059932e-01 2.20231682e-01
5.04532039e-01 -1.80537239e-01 -1.20654011e+00 4.85077113e-01
1.00113177e+00 -9.74055946e-01 1.09454048e+00 -6.19996667e-01
2.05289289e-01 -3.56282622e-01 -2.26715729e-01 -1.64074767e+00
-3.14318746e-01 -4.73582357e-01 -9.12997797e-02 1.05941808e+00
5.78778386e-01 -7.39928842e-01 6.77471042e-01 9.22287703e-01
-3.56204838e-01 -9.42398369e-01 -1.08145654e+00 -1.16674972e+00
3.82107079e-01 -5.56086898e-01 9.18418348e-01 7.03122854e-01
6.18189834e-02 -1.78878322e-01 -5.78736544e-01 6.48746639e-02
7.29109526e-01 -2.38694742e-01 1.16370022e+00 -1.26927757e+00
3.14723015e-01 -7.63106823e-01 -5.92307806e-01 -9.28869903e-01
9.54043493e-02 -8.06014597e-01 4.65178713e-02 -1.55520809e+00
-1.68197706e-01 -9.16120887e-01 2.50600725e-01 2.40724221e-01
-7.07692560e-03 2.43257239e-01 2.11504832e-01 2.93539971e-01
-6.19518936e-01 8.15980434e-01 8.57400119e-01 5.95862195e-02
-2.80600578e-01 4.99796085e-02 -1.56884253e-01 7.32990205e-01
5.39453268e-01 -5.20269334e-01 -3.53262037e-01 -3.83460373e-01
2.76197553e-01 1.11303337e-01 7.32124805e-01 -1.25312185e+00
4.10101295e-01 -2.98691273e-01 8.32894817e-02 -8.48791897e-01
5.81146479e-01 -6.38583422e-01 6.03200972e-01 4.23728019e-01
-1.88474238e-01 4.68242735e-01 3.70711356e-01 5.80381155e-01
3.36731300e-02 2.60962248e-01 2.46579558e-01 3.68239611e-01
-5.79927683e-01 1.86886951e-01 -5.00606596e-01 -1.86899109e-04
1.25631678e+00 -4.41987574e-01 -7.26411343e-01 -5.83367229e-01
-6.14545703e-01 6.96103573e-01 4.78638619e-01 5.10469139e-01
4.01460886e-01 -1.49437726e+00 -1.00404584e+00 -2.24222854e-01
9.24548730e-02 -4.53502387e-01 3.52086365e-01 1.24477077e+00
-5.61220407e-01 5.82815707e-01 -3.55343491e-01 -5.86955130e-01
-1.05763781e+00 1.63347170e-01 2.81118661e-01 -1.86919346e-01
-7.17796743e-01 5.37544787e-01 -2.20435739e-01 -8.12261581e-01
6.74799457e-02 -3.38650048e-01 -1.89662259e-02 2.67504543e-01
3.43892932e-01 1.33727145e+00 -1.26777440e-01 -1.10850322e+00
-7.32782066e-01 4.64202762e-01 2.21270144e-01 -4.42041129e-01
1.26694298e+00 -2.97917187e-01 3.84796292e-01 3.79929006e-01
9.56890821e-01 -1.28633752e-01 -1.55477166e+00 2.65219420e-01
1.79954946e-01 -3.92707586e-01 -2.44331099e-02 -9.51135874e-01
-5.00706851e-01 5.04021347e-01 5.50701678e-01 3.92553329e-01
4.51024085e-01 -1.97436467e-01 1.15321648e+00 2.89816529e-01
6.18383467e-01 -8.72672260e-01 -3.81709188e-01 5.28449357e-01
9.11739707e-01 -1.17913640e+00 1.37196064e-01 7.80978948e-02
-1.04364598e+00 6.87658131e-01 5.69412053e-01 -3.32586616e-01
7.72729635e-01 1.64153844e-01 -1.05827548e-01 -2.32651189e-01
-8.49297106e-01 -2.52337575e-01 2.71048814e-01 5.81679404e-01
1.08514778e-01 4.49112743e-01 -3.77582073e-01 4.20171857e-01
-4.68395799e-01 -9.83297005e-02 4.05866176e-01 6.78018630e-01
-2.64698207e-01 -7.43352056e-01 -1.07427843e-01 3.85646552e-01
6.37608171e-02 4.04357493e-01 -1.60454288e-01 1.13906467e+00
-4.72480804e-02 1.04744506e+00 4.27290589e-01 -7.37515032e-01
5.75729072e-01 -2.06228465e-01 1.98605299e-01 6.53218329e-02
-6.43768132e-01 -5.16868114e-01 4.78798807e-01 -4.87530291e-01
-5.40935934e-01 -1.05442035e+00 -1.39949894e+00 -1.00590527e+00
5.66403680e-02 7.49274418e-02 3.35445732e-01 9.15444970e-01
5.56681097e-01 2.17305869e-01 2.79574245e-01 -1.22190917e+00
-2.97520190e-01 -9.16153848e-01 -2.85846025e-01 5.04182756e-01
4.97952729e-01 -1.26855350e+00 -3.60535949e-01 -3.53209704e-01] | [5.799829483032227, 0.9513121843338013] |
bae6a775-bab4-4962-9647-c2cc1977f7f2 | exploring-the-limits-of-a-base-bart-for-multi | null | null | https://aclanthology.org/2022.sdp-1.23 | https://aclanthology.org/2022.sdp-1.23.pdf | Exploring the limits of a base BART for multi-document summarization in the medical domain | This paper is a description of our participation in the Multi-document Summarization for Literature Review (MSLR) Shared Task, in which we explore summarization models to create an automatic review of scientific results. Rather than maximizing the metrics using expensive computational models, we placed ourselves in a situation of scarce computational resources and explore the limits of a base sequence to sequence models (thus with a limited input length) to the task. Although we explore methods to feed the abstractive model with salient sentences only (using a first extractive step), we find the results still need some improvements. | ['Horacio Saggion', 'Silvia Casola', 'Ishmael Obonyo'] | null | null | null | null | sdp-coling-2022-10 | ['document-summarization'] | ['natural-language-processing'] | [ 5.71268380e-01 6.66933954e-01 -5.96930087e-01 -9.91654322e-02
-1.33772755e+00 -6.83881044e-01 7.71252155e-01 4.57280487e-01
-4.95292127e-01 1.08737016e+00 8.16765726e-01 -6.84709430e-01
-1.13125429e-01 -1.77983910e-01 -4.88745630e-01 -2.59192288e-01
3.05277407e-01 3.25554401e-01 -2.64709797e-02 2.29198448e-02
1.20742309e+00 3.63466620e-01 -1.19512784e+00 4.83435422e-01
9.74009991e-01 -4.62271012e-02 3.26000243e-01 1.14089990e+00
-4.28257495e-01 6.38038218e-01 -9.14704263e-01 -2.68269330e-01
-1.59082413e-01 -7.99093306e-01 -1.28460431e+00 -1.00840487e-01
5.70428908e-01 6.47634594e-03 1.56326443e-01 9.23716247e-01
6.16067410e-01 1.80425286e-01 9.19056058e-01 -6.17859900e-01
-6.27987802e-01 9.64859962e-01 -8.93121123e-01 6.93036377e-01
7.17431009e-01 -8.47501010e-02 1.23633337e+00 -9.15510595e-01
1.19027972e+00 1.31843078e+00 3.36738229e-01 5.40733039e-01
-1.00791419e+00 -8.94211084e-02 3.63878340e-01 1.93272710e-01
-9.66571271e-01 -8.73884082e-01 6.65508091e-01 -2.82881021e-01
1.49078894e+00 4.32055146e-01 4.55805957e-01 1.02170074e+00
5.11333704e-01 6.63299799e-01 7.06980765e-01 -6.32857740e-01
3.56748909e-01 1.10183485e-01 4.80392158e-01 3.07077497e-01
5.09967327e-01 -6.91880941e-01 -6.83943450e-01 -3.96133929e-01
3.66029739e-02 -5.87355733e-01 -1.78696007e-01 4.00626481e-01
-1.31772292e+00 8.19808424e-01 -1.79734230e-01 4.33159620e-01
-6.03853941e-01 5.65531924e-02 5.94086885e-01 1.51604310e-01
8.20089400e-01 1.00382972e+00 -2.93567210e-01 -3.70276213e-01
-1.69761300e+00 5.28862834e-01 1.17931545e+00 9.24047768e-01
2.79351652e-01 -5.85247688e-02 -3.44801486e-01 8.23497772e-01
9.22159404e-02 2.34464332e-02 3.79875064e-01 -1.03419781e+00
5.52125454e-01 1.39444873e-01 1.06412873e-01 -9.39926803e-01
-4.68875378e-01 -4.50859278e-01 -3.65090638e-01 -2.59818494e-01
1.40076010e-02 -3.71814072e-01 -5.70873439e-01 1.24153721e+00
-1.41292319e-01 -9.84662250e-02 1.48112610e-01 5.65700352e-01
1.33023190e+00 9.79228377e-01 3.48716468e-01 -9.22358692e-01
1.46091294e+00 -1.09076750e+00 -9.32609797e-01 -2.87807405e-01
9.62854803e-01 -9.79592800e-01 6.56865895e-01 5.79247415e-01
-1.67299712e+00 -1.67264551e-01 -1.28476357e+00 -4.10046697e-01
-3.52280170e-01 1.35806441e-01 5.58312178e-01 2.88806617e-01
-1.39328706e+00 8.18700671e-01 -5.12643754e-01 -5.48544347e-01
3.98590773e-01 1.10384310e-02 -1.40422702e-01 2.11327881e-01
-9.67458963e-01 1.41208243e+00 3.44141930e-01 1.28604816e-02
-3.22453916e-01 -8.08497906e-01 -6.33157432e-01 3.21490783e-03
4.17843997e-01 -1.02506089e+00 1.21459234e+00 -6.79038703e-01
-1.39640391e+00 1.00579202e+00 -6.13565683e-01 -3.80614907e-01
2.76788622e-01 -1.77776694e-01 9.42143574e-02 4.29089040e-01
2.31492028e-01 7.07635283e-01 3.04136902e-01 -1.14512122e+00
-3.60817790e-01 -2.18294978e-01 -2.17888989e-02 3.59740019e-01
-7.83694237e-02 8.13707232e-01 -1.86742291e-01 -6.62170112e-01
-3.31013054e-01 -6.35033906e-01 -4.71377045e-01 -4.95111763e-01
-5.59910536e-01 -5.30038357e-01 2.13172078e-01 -1.01226437e+00
1.52916157e+00 -1.24756491e+00 5.60993373e-01 -1.94665447e-01
1.61107212e-01 2.03671053e-01 -2.80541182e-01 7.09823191e-01
-7.22786859e-02 8.09189618e-01 -4.87801254e-01 -5.61668992e-01
-2.81362444e-01 -2.80725211e-01 -5.97376585e-01 2.81170398e-01
3.54330420e-01 1.06367922e+00 -1.14680195e+00 -8.39965343e-01
-2.41959989e-01 -4.63191606e-02 -3.70352507e-01 -1.77449077e-01
-3.34535688e-01 -1.41270284e-03 -7.97857642e-01 4.62518930e-01
5.54272950e-01 -1.39492646e-01 1.37039749e-02 2.09058031e-01
-6.30625069e-01 6.86296046e-01 -8.33589852e-01 1.93054092e+00
-1.97553799e-01 7.76506305e-01 1.64197221e-01 -9.69970524e-01
6.45609021e-01 3.40050340e-01 3.96374911e-01 -2.19440714e-01
-1.35375738e-01 2.34970123e-01 1.08917594e-01 -4.68949080e-01
1.08588505e+00 -1.35790989e-01 4.09639291e-02 7.21325576e-01
9.82417762e-02 -5.09145260e-01 7.70806789e-01 7.70756304e-01
1.28580737e+00 3.58743578e-01 6.47504926e-01 -5.88264227e-01
3.19477558e-01 3.20564270e-01 2.29611352e-01 1.04523039e+00
-9.70397070e-02 7.67191470e-01 8.29224050e-01 -2.65329462e-02
-1.31832552e+00 -4.47288364e-01 -2.13506877e-01 8.85663867e-01
-4.57787663e-01 -9.26785111e-01 -6.09143019e-01 -4.62113321e-01
-3.66956919e-01 1.13613153e+00 -4.29100931e-01 1.28368810e-01
-7.08162546e-01 -9.28314626e-01 5.98345757e-01 1.78415596e-01
-2.65328616e-01 -1.18063724e+00 -6.31745756e-01 2.70664513e-01
6.02834038e-02 -7.80577898e-01 -3.87765199e-01 2.18099043e-01
-9.41176176e-01 -6.44987822e-01 -1.03974450e+00 -6.16235375e-01
2.83503771e-01 1.21224530e-01 1.14318490e+00 8.35150853e-02
-4.45677042e-01 3.45225155e-01 -3.98282260e-01 -9.12834167e-01
-5.24252057e-01 5.10457933e-01 -3.64153713e-01 -9.68011916e-01
3.66747051e-01 -3.73041123e-01 -4.52145636e-01 -7.37606645e-01
-9.13762212e-01 3.90959442e-01 7.08907247e-01 5.49239516e-01
2.36487702e-01 -6.18477702e-01 1.27862060e+00 -9.58282053e-01
1.34988999e+00 -6.32976592e-01 -1.14027217e-01 3.03213745e-01
-5.69378078e-01 4.39826511e-02 3.55409294e-01 -1.42119229e-01
-1.06796777e+00 -4.20359462e-01 -2.75629193e-01 2.65171736e-01
6.97043091e-02 1.04401159e+00 1.24385044e-01 2.74448454e-01
7.60125995e-01 6.23960122e-02 9.27553251e-02 -4.56483155e-01
5.89447439e-01 6.37515187e-01 1.19958654e-01 -5.03693223e-01
2.70876735e-01 -5.12880459e-02 1.29603848e-01 -1.20766640e+00
-9.05259132e-01 -5.36397398e-01 -6.09139144e-01 -2.56741010e-02
6.84974551e-01 -7.77436495e-01 -2.75207520e-01 -4.00023699e-01
-1.76542819e+00 1.03434563e-01 -3.76929879e-01 3.31715077e-01
-4.44953501e-01 8.26528966e-01 -7.06562221e-01 -8.03162515e-01
-7.64543295e-01 -9.88507032e-01 1.11263132e+00 3.11890334e-01
-8.89199734e-01 -1.03428590e+00 2.74248421e-01 1.75066411e-01
2.98462510e-01 1.74506560e-01 9.41510081e-01 -8.93734157e-01
-3.06200653e-01 7.20809493e-03 -1.14477761e-01 5.76025024e-02
-5.28731272e-02 3.92489463e-01 -7.97216237e-01 9.08391178e-02
7.10349157e-02 -2.87273467e-01 1.30766535e+00 6.68231905e-01
1.20536971e+00 -4.62970614e-01 -4.66690361e-01 9.49993432e-02
1.05724239e+00 2.30513796e-01 5.60990930e-01 4.59761202e-01
5.47565937e-01 9.50687885e-01 3.39913249e-01 3.59782457e-01
3.82076412e-01 3.33693802e-01 -2.59417623e-01 9.53539237e-02
-2.80426025e-01 1.44243687e-02 2.04802170e-01 1.26623452e+00
1.11088715e-02 -5.40008485e-01 -8.56558979e-01 8.47402692e-01
-1.81096351e+00 -1.06696188e+00 -1.71913370e-01 1.71378922e+00
1.10502207e+00 1.65098220e-01 1.58505216e-01 -3.22522491e-01
5.19336760e-01 5.83250523e-01 -3.46881807e-01 -1.28155625e+00
-2.18074605e-01 3.35403651e-01 4.23822433e-01 7.44953454e-01
-6.46767557e-01 8.89961958e-01 7.83995867e+00 8.43042016e-01
-7.60476172e-01 -2.30036035e-01 6.40295684e-01 -4.32360739e-01
-8.46418977e-01 3.59835148e-01 -7.86085188e-01 3.02888006e-01
1.49857855e+00 -8.25726807e-01 7.51046762e-02 4.99240279e-01
5.98052859e-01 -4.46301788e-01 -1.19152784e+00 4.02195662e-01
2.53045142e-01 -1.71410739e+00 4.44069624e-01 -7.51473233e-02
5.97375512e-01 -8.00915137e-02 -3.31317365e-01 8.26726556e-02
6.28073439e-02 -1.11236703e+00 4.19270456e-01 6.90634489e-01
4.55290824e-01 -7.90882289e-01 6.20368958e-01 4.10870314e-01
-5.24037600e-01 3.70883912e-01 -5.42904735e-01 -2.03925669e-01
3.27522516e-01 6.80278182e-01 -7.40457892e-01 8.92540753e-01
-1.04591399e-02 1.07390785e+00 -6.71405077e-01 1.07300079e+00
-1.57510430e-01 8.42344940e-01 1.46772131e-01 -5.91099858e-01
2.21867576e-01 -2.15453029e-01 1.12500799e+00 1.85891485e+00
2.51232415e-01 2.50047624e-01 -7.49385974e-04 8.41855645e-01
-2.92768896e-01 4.20160323e-01 -7.21006334e-01 -3.96180749e-01
2.75680542e-01 1.20297003e+00 -8.56069684e-01 -4.91465509e-01
-1.84019789e-01 7.55716205e-01 2.05228254e-01 3.95987719e-01
-1.79574549e-01 -7.54398286e-01 1.71690751e-02 -3.60430688e-01
9.57461372e-02 -6.64368495e-02 -9.00144219e-01 -1.16247952e+00
3.90290394e-02 -9.43233550e-01 3.40349287e-01 -9.59212601e-01
-8.71782839e-01 3.25905114e-01 3.06699693e-01 -6.30772591e-01
-5.24490774e-01 -2.28904113e-01 -9.10315275e-01 1.14443111e+00
-1.47677636e+00 -7.93118238e-01 6.55830443e-01 -4.09666538e-01
9.28785682e-01 7.72500113e-02 7.12917328e-01 -2.24707976e-01
-4.62617487e-01 1.95840612e-01 1.03529803e-01 -6.82293534e-01
8.03862989e-01 -1.36785579e+00 5.17443538e-01 8.55269194e-01
-9.61337462e-02 1.19414711e+00 1.27298582e+00 -9.19923306e-01
-1.48931646e+00 -5.08350313e-01 1.62930608e+00 -6.19010031e-01
8.21455956e-01 -1.58080403e-02 -9.30334687e-01 4.39148843e-01
9.20741200e-01 -9.82200503e-01 8.83097351e-01 2.92084605e-01
2.39595413e-01 5.68727553e-01 -8.12702477e-01 8.00704718e-01
9.73287523e-01 -2.56584525e-01 -1.25897491e+00 4.48935598e-01
8.77108455e-01 -1.93792477e-01 -7.67207026e-01 1.46561772e-01
4.43765551e-01 -6.31754994e-02 6.93793714e-01 -8.39485168e-01
1.23786533e+00 -2.59978890e-01 5.32594264e-01 -1.50408804e+00
-2.76352435e-01 -7.35407114e-01 -9.71396081e-03 1.27340031e+00
7.40860641e-01 -2.80013978e-01 5.01253247e-01 5.99759996e-01
-5.32566547e-01 -7.69747198e-01 -8.70520711e-01 -2.06439734e-01
5.94463408e-01 -2.11081699e-01 6.13982091e-03 6.83733284e-01
6.75568461e-01 7.55370319e-01 -1.63966194e-01 -6.25042021e-01
3.82032543e-01 1.36089191e-01 4.01259333e-01 -1.14760458e+00
3.02047236e-03 -1.02236676e+00 1.66645795e-01 -9.23142314e-01
4.25983667e-01 -1.13573980e+00 1.16864800e-01 -2.30964684e+00
8.69342804e-01 4.22536790e-01 -1.94154292e-01 2.76989967e-01
-5.65041244e-01 -2.12764800e-01 1.05850987e-01 1.71648279e-01
-9.41569269e-01 4.03516293e-01 1.05381346e+00 -1.58410117e-01
-2.19768032e-01 -3.13796490e-01 -1.55717731e+00 5.31356275e-01
5.16131878e-01 -4.84946221e-01 -3.05544168e-01 -2.69877672e-01
4.45611864e-01 1.50453746e-01 -5.64893633e-02 -3.11352700e-01
4.52809006e-01 -3.16343069e-01 3.18913013e-01 -7.87297964e-01
-1.37681752e-01 9.52880606e-02 -2.26037338e-01 2.50941962e-01
-1.11599338e+00 1.36891171e-01 5.35101235e-01 2.57237643e-01
-9.67004523e-03 -9.08758759e-01 2.12192297e-01 -4.92383212e-01
-1.55257463e-01 -2.54002988e-01 -9.20824468e-01 -7.98680857e-02
4.97276813e-01 2.01506820e-02 -5.88650525e-01 -3.30855668e-01
-3.91282827e-01 3.84389669e-01 4.83516365e-01 3.53573948e-01
6.11241996e-01 -5.92298329e-01 -1.14609647e+00 -6.91552639e-01
-3.40174176e-02 -1.09745495e-01 4.71264645e-02 5.26521862e-01
-4.51520503e-01 9.62522745e-01 4.80301902e-02 -3.27109359e-02
-1.23731220e+00 5.99390745e-01 -3.36415887e-01 -6.11418486e-01
-7.27413058e-01 5.70692301e-01 -1.32312626e-01 -1.89356487e-02
4.29310389e-02 1.67657584e-02 -7.11174786e-01 4.71858144e-01
7.72390187e-01 5.92820108e-01 1.96875438e-01 -4.12292272e-01
-3.88769090e-01 3.68422687e-01 -4.73162413e-01 -4.37045693e-01
1.58591366e+00 -8.47307816e-02 -3.82521123e-01 6.31100476e-01
1.06219673e+00 3.53917599e-01 -4.12869990e-01 1.32123735e-02
5.97576320e-01 1.00381345e-01 3.56095791e-01 -8.53749692e-01
-1.08691864e-01 6.47769809e-01 -2.15577692e-01 4.76429462e-02
6.48379505e-01 2.62550171e-02 4.22197491e-01 6.97210670e-01
-1.52117744e-01 -1.27227974e+00 -2.92898089e-01 5.45619905e-01
1.25509810e+00 -9.49045599e-01 9.95549977e-01 4.07983698e-02
-6.89496934e-01 1.36833549e+00 5.24636097e-02 -1.11173339e-01
2.91647762e-01 1.56004116e-01 -3.08645576e-01 -3.80588144e-01
-1.24127543e+00 -2.27919240e-02 4.94906157e-01 1.71996906e-01
9.19576764e-01 -2.91148424e-01 -1.41895330e+00 3.97304088e-01
-2.32740179e-01 1.75027907e-01 1.14393246e+00 1.14326811e+00
-6.47543550e-01 -1.08529663e+00 -1.00121528e-01 7.72973120e-01
-1.00728023e+00 -4.86470252e-01 -9.00398076e-01 5.41944623e-01
-6.07097566e-01 1.17426622e+00 -1.83782026e-01 1.45235389e-01
2.29543462e-01 1.73864573e-01 3.18413407e-01 -1.06801999e+00
-8.09797883e-01 2.30064169e-01 6.14831030e-01 -6.95457086e-02
-6.26477540e-01 -8.84517729e-01 -1.02429867e+00 -2.17963368e-01
-9.37366188e-02 5.63982189e-01 9.68823671e-01 9.89348173e-01
5.79277039e-01 7.96798646e-01 2.63404995e-01 -1.00229681e+00
-4.54575956e-01 -1.26822984e+00 -3.22454095e-01 -2.60368943e-01
3.35754335e-01 -1.18515663e-01 -4.28550124e-01 2.05792591e-01] | [12.340157508850098, 9.589061737060547] |
a5497826-5621-4920-b6c5-ee2465b81313 | learning-scene-flow-with-skeleton-guidance | 2306.13285 | null | https://arxiv.org/abs/2306.13285v1 | https://arxiv.org/pdf/2306.13285v1.pdf | Learning Scene Flow With Skeleton Guidance For 3D Action Recognition | Among the existing modalities for 3D action recognition, 3D flow has been poorly examined, although conveying rich motion information cues for human actions. Presumably, its susceptibility to noise renders it intractable, thus challenging the learning process within deep models. This work demonstrates the use of 3D flow sequence by a deep spatiotemporal model and further proposes an incremental two-level spatial attention mechanism, guided from skeleton domain, for emphasizing motion features close to the body joint areas and according to their informativeness. Towards this end, an extended deep skeleton model is also introduced to learn the most discriminant action motion dynamics, so as to estimate an informativeness score for each joint. Subsequently, a late fusion scheme is adopted between the two models for learning the high level cross-modal correlations. Experimental results on the currently largest and most challenging dataset NTU RGB+D, demonstrate the effectiveness of the proposed approach, achieving state-of-the-art results. | ['Athanasios Psaltis', 'Vasileios Magoulianitis'] | 2023-06-23 | null | null | null | null | ['action-recognition-in-videos', '3d-human-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 2.72776991e-01 -7.55441561e-02 -4.38934565e-01 -1.54851645e-01
-5.94705403e-01 -6.19604960e-02 7.32870877e-01 -2.17799976e-01
-3.46365780e-01 4.51262206e-01 6.66181445e-01 2.68553019e-01
-2.59818882e-01 -4.15531576e-01 -4.08832788e-01 -8.53686988e-01
-3.32173258e-01 1.41025394e-01 2.87590891e-01 1.74555760e-02
2.95730740e-01 5.19480288e-01 -1.68889499e+00 3.10652077e-01
7.10641980e-01 1.24277091e+00 1.77369609e-01 5.53015053e-01
2.68583093e-03 1.22043145e+00 -2.74095327e-01 2.23275907e-02
3.20646077e-01 -6.23521030e-01 -7.81279385e-01 3.75102967e-01
5.76968729e-01 -7.28443801e-01 -5.48956871e-01 7.31617451e-01
4.37925339e-01 4.99119282e-01 6.57999456e-01 -1.05999458e+00
-1.82636783e-01 5.05317636e-02 -6.19849026e-01 4.95877057e-01
6.09533727e-01 5.00248134e-01 1.05516815e+00 -7.45930791e-01
7.81612694e-01 1.30444658e+00 2.56808013e-01 5.18939376e-01
-9.68140781e-01 -3.04080635e-01 2.42190644e-01 5.24171233e-01
-9.09136176e-01 -2.86174923e-01 1.26575005e+00 -5.29165685e-01
8.50299239e-01 -1.13395721e-01 1.01698434e+00 1.41373658e+00
1.99814066e-01 1.22371483e+00 1.14794981e+00 -2.11212322e-01
2.64115900e-01 -6.06688738e-01 -1.20748825e-01 8.47431123e-01
-2.53337592e-01 9.40319002e-02 -9.54512239e-01 1.86659425e-01
9.27779078e-01 2.92337909e-02 -3.47597808e-01 -6.44952178e-01
-1.24703312e+00 5.03741264e-01 5.25882781e-01 5.52587569e-01
-7.48770714e-01 3.86305630e-01 3.35438758e-01 -1.99865445e-01
5.31423509e-01 -7.31298029e-02 -2.81144172e-01 -5.86510122e-01
-7.75008738e-01 3.19942713e-01 2.28892729e-01 4.31988806e-01
6.26434565e-01 1.43839158e-02 -3.91501069e-01 6.06598556e-01
5.53822160e-01 2.56207973e-01 3.49119544e-01 -1.33279884e+00
6.95132494e-01 8.16973269e-01 6.51457384e-02 -1.14132929e+00
-5.97453833e-01 -3.87564361e-01 -7.30910063e-01 4.56813455e-01
6.87017381e-01 2.15174675e-01 -7.05485344e-01 1.72681212e+00
6.33608043e-01 4.58073974e-01 -1.73616171e-01 1.31480014e+00
4.76366162e-01 3.32093447e-01 2.97393352e-01 -2.57046521e-02
1.25031292e+00 -1.00305462e+00 -6.93034232e-01 -1.80853218e-01
5.17984033e-01 -3.31853062e-01 9.49268401e-01 4.14436996e-01
-1.13370371e+00 -9.12497163e-01 -8.02170515e-01 -2.48373836e-01
-6.53700456e-02 1.67986661e-01 7.03572392e-01 3.50127637e-01
-6.63183749e-01 7.35979140e-01 -1.37485147e+00 -4.16869700e-01
8.01155686e-01 8.09734538e-02 -5.01005590e-01 -1.24075584e-01
-1.06737387e+00 8.60707939e-01 3.53415132e-01 2.50907540e-01
-9.72386599e-01 -6.09178305e-01 -1.03512073e+00 -1.56544641e-01
2.43572444e-01 -9.79278862e-01 8.50728691e-01 -8.22423995e-01
-1.80536866e+00 6.11641586e-01 -4.04180065e-02 -4.02549416e-01
7.87041187e-01 -5.80503583e-01 -6.42869845e-02 7.81443357e-01
7.61905462e-02 7.31486380e-01 9.41100895e-01 -1.06938004e+00
-7.46419132e-01 -6.74535930e-01 2.39574596e-01 5.44770479e-01
-3.54842454e-01 -2.78264910e-01 -5.70208371e-01 -7.34601676e-01
2.24483892e-01 -7.42778540e-01 -2.75268763e-01 3.12427372e-01
-1.00772493e-01 -3.01672429e-01 8.37370038e-01 -8.22375417e-01
1.22374022e+00 -1.95544207e+00 5.42930305e-01 1.19446985e-01
1.22691818e-01 1.97775394e-01 -3.96222547e-02 1.86500952e-01
4.78354916e-02 -3.71277869e-01 -3.49112421e-01 -4.68600541e-01
4.54412252e-02 3.08023512e-01 2.78489172e-01 7.37896442e-01
5.47654271e-01 9.55944479e-01 -9.98052239e-01 -5.60804784e-01
7.54802287e-01 7.26287663e-01 -5.52920103e-01 2.56791055e-01
-2.02600863e-02 1.00084877e+00 -8.07283521e-01 8.18319261e-01
3.61492962e-01 -1.73791721e-01 -1.03583738e-01 -4.11462426e-01
3.36280391e-02 5.43920435e-02 -1.12682784e+00 2.42140484e+00
-2.85035908e-01 3.99177998e-01 -7.83473104e-02 -1.18896532e+00
8.02389681e-01 1.52779847e-01 1.11757171e+00 -8.59555662e-01
2.47499585e-01 4.55117747e-02 -1.02792211e-01 -1.01499176e+00
2.64016122e-01 3.22665386e-02 -2.36815196e-02 2.08140641e-01
1.63844302e-01 1.28385559e-01 1.14193045e-01 -1.16323896e-01
1.21144044e+00 9.66303587e-01 3.38514000e-01 1.39126638e-02
8.63543093e-01 -2.59294569e-01 4.63102490e-01 5.19530773e-01
-7.44955182e-01 6.40619874e-01 3.92259240e-01 -4.47267085e-01
-7.16204703e-01 -9.36158538e-01 1.93727538e-01 6.79035783e-01
3.72530282e-01 -2.02603728e-01 -5.94210625e-01 -8.13323319e-01
-2.21257145e-03 3.59300971e-01 -8.45465004e-01 -2.28551656e-01
-8.02495778e-01 -3.49385768e-01 3.53002101e-01 7.15624034e-01
7.64027715e-01 -1.01836383e+00 -1.18537581e+00 2.56349623e-01
-4.79893386e-01 -1.24892497e+00 -1.88994914e-01 -2.06280723e-01
-1.08016241e+00 -1.19834328e+00 -1.01724541e+00 -2.41970181e-01
3.32639456e-01 2.15711087e-01 7.91348159e-01 -1.42614126e-01
-3.46584469e-01 6.48861408e-01 -4.96182650e-01 3.27119827e-01
-1.28635675e-01 -3.00244808e-01 -9.61249620e-02 4.32806313e-01
2.38524541e-01 -7.13245094e-01 -9.65898216e-01 1.35465086e-01
-9.18598652e-01 -1.63871914e-01 7.88651586e-01 8.39442611e-01
3.98306876e-01 -2.32732713e-01 3.77335012e-01 -1.71399504e-01
3.54396366e-02 -4.09071654e-01 -5.65220751e-02 -7.40486234e-02
-8.81829262e-02 -6.14317916e-02 3.39547575e-01 -2.80114114e-01
-1.34057748e+00 2.79834658e-01 -1.20944999e-01 -7.29549170e-01
-5.80984294e-01 3.34203392e-01 -2.57360548e-01 2.82091238e-02
3.18374783e-01 1.86308116e-01 2.60001928e-01 -4.56770718e-01
4.77484614e-01 2.41114184e-01 6.06024265e-01 -5.04670024e-01
5.28497398e-01 8.24935913e-01 3.01283032e-01 -8.90151560e-01
-7.98478544e-01 -6.22643471e-01 -1.15137434e+00 -6.51206732e-01
1.16976428e+00 -8.85359824e-01 -7.10095108e-01 6.97377801e-01
-1.02142477e+00 -2.06240609e-01 -3.29679042e-01 9.57701445e-01
-1.07022929e+00 7.85395384e-01 -5.72568953e-01 -9.33636308e-01
5.93367545e-03 -1.17177069e+00 1.38832545e+00 7.78960884e-02
-2.57388115e-01 -9.53532875e-01 2.66176224e-01 8.04433644e-01
-3.92954377e-03 6.18687630e-01 6.50307655e-01 -2.71999002e-01
-7.25231588e-01 -3.87098677e-02 -9.08937231e-02 2.98117608e-01
1.10948376e-01 -1.82959169e-01 -9.52424228e-01 7.79417902e-02
-8.58937353e-02 -4.49647129e-01 1.09958076e+00 5.69732964e-01
1.00396466e+00 1.31991312e-01 -1.55743718e-01 4.69001919e-01
1.07823193e+00 1.13224154e-02 6.67341053e-01 3.94530088e-01
8.10949922e-01 7.71437109e-01 9.60798681e-01 7.20366716e-01
3.20166469e-01 7.94477165e-01 6.74960434e-01 -3.68615091e-02
-2.89233595e-01 -1.36057839e-01 4.85305816e-01 5.42744875e-01
-6.87738121e-01 6.09368607e-02 -5.95983446e-01 4.08851147e-01
-1.96255994e+00 -1.12849641e+00 4.22535352e-02 1.87445879e+00
3.98058593e-01 1.35534868e-01 2.52896100e-01 2.88714319e-01
3.15414071e-01 5.92498720e-01 -5.70551515e-01 2.34117806e-01
-1.59754455e-01 2.09919751e-01 -1.99902686e-03 3.50577086e-01
-1.39064872e+00 7.92115629e-01 5.21014404e+00 8.16914499e-01
-9.05671716e-01 -1.07041590e-01 3.82595539e-01 -5.59675433e-02
1.59209594e-02 -1.40500395e-02 -3.64902169e-01 3.81870031e-01
3.67982894e-01 4.08742607e-01 -3.26210409e-02 5.57234347e-01
7.49031365e-01 -3.66185248e-01 -8.77359092e-01 8.73535752e-01
1.34195477e-01 -1.00199676e+00 -2.36151759e-02 7.74032101e-02
5.19310057e-01 -2.35084325e-01 -1.43484041e-01 3.21924463e-02
-2.03974009e-01 -6.96093321e-01 8.12760890e-01 9.14165974e-01
1.84097335e-01 -7.80398428e-01 4.33480442e-01 3.81861746e-01
-1.28950357e+00 -3.13958466e-01 -2.62963003e-04 -1.53479621e-01
5.24005413e-01 3.05212110e-01 -2.31688738e-01 8.92709434e-01
6.67635381e-01 1.43284059e+00 -5.90234816e-01 8.46422493e-01
-2.65728861e-01 3.95577252e-01 -1.85062021e-01 2.10037470e-01
5.16589046e-01 -2.52324075e-01 8.03614080e-01 1.01054931e+00
3.70016992e-01 2.12039709e-01 2.72776544e-01 6.71315312e-01
3.93094569e-01 -4.79100607e-02 -6.56174242e-01 1.72305539e-01
-1.30474672e-01 1.16239119e+00 -7.23383248e-01 -1.37612134e-01
-3.84319782e-01 1.18943679e+00 2.45629415e-01 4.58549142e-01
-8.60463321e-01 1.38576299e-01 7.74026752e-01 -1.39914349e-01
3.70645314e-01 -4.81062472e-01 1.26545681e-02 -1.23603404e+00
1.27172828e-01 -6.27133012e-01 6.04676902e-01 -6.84383810e-01
-1.10675776e+00 1.69500619e-01 8.05672035e-02 -1.72226846e+00
-4.40383911e-01 -6.40221834e-01 -3.31989199e-01 5.92392862e-01
-1.56496310e+00 -1.30347860e+00 -4.15256649e-01 8.44078064e-01
7.00022817e-01 1.56431571e-02 4.48319286e-01 3.00451547e-01
-5.47256172e-01 6.03140779e-02 -3.34618419e-01 -2.85470989e-02
4.18229699e-01 -1.08772695e+00 -3.75049263e-02 8.79001200e-01
8.75054523e-02 2.54405051e-01 4.62317377e-01 -5.87369740e-01
-1.51649082e+00 -8.20958912e-01 4.79302108e-01 -3.74412507e-01
6.51824355e-01 6.13651983e-02 -8.34183633e-01 2.58712202e-01
6.79871589e-02 2.56343603e-01 4.60312992e-01 -4.26722169e-01
-1.07809834e-01 -9.57690738e-03 -8.98476601e-01 4.30386066e-01
1.33087075e+00 -4.20248896e-01 -5.48676014e-01 1.76575202e-02
1.72604501e-01 -2.99786657e-01 -1.12296927e+00 4.77479726e-01
7.89567828e-01 -1.35956395e+00 1.21085775e+00 -6.32019639e-01
7.76488304e-01 -4.36825693e-01 -2.02430353e-01 -9.74896669e-01
-2.55295396e-01 -4.00522977e-01 -7.02625751e-01 1.05375087e+00
-2.84265250e-01 -1.10142536e-01 1.05853081e+00 2.75290936e-01
-2.30826661e-01 -7.46819675e-01 -1.25632823e+00 -5.52384257e-01
-3.08645219e-01 -7.21790731e-01 3.82153541e-02 8.05727065e-01
-1.71418980e-01 9.54360366e-02 -5.81632733e-01 3.69065180e-02
7.84897506e-01 1.29356042e-01 8.17794561e-01 -9.91850495e-01
-4.10661906e-01 -7.70724952e-01 -9.24587429e-01 -1.50681806e+00
3.70893329e-01 -6.42881513e-01 2.46535521e-02 -1.38245475e+00
-7.37887546e-02 -1.30853066e-02 -3.49486262e-01 2.19141394e-01
-3.51772398e-01 3.64959568e-01 2.70063877e-01 1.76324278e-01
-5.62036693e-01 1.00550020e+00 1.56230164e+00 -2.21604124e-01
-1.24524154e-01 3.42154838e-02 -1.37185052e-01 9.06692028e-01
4.19324905e-01 -2.72295978e-02 -5.12548149e-01 -1.72308400e-01
-4.43483233e-01 2.82785892e-01 7.87740409e-01 -1.15862775e+00
-8.72969721e-03 -1.95347458e-01 6.19626820e-01 -7.87352204e-01
6.70683086e-01 -9.40713942e-01 -1.38196811e-01 5.93743205e-01
-3.70715946e-01 -2.98208952e-01 -6.90998659e-02 8.87449205e-01
-3.89169574e-01 1.67601213e-01 7.07747579e-01 -1.83131218e-01
-1.11322343e+00 5.34397721e-01 -2.07092360e-01 2.29289308e-02
1.06475949e+00 -5.14527380e-01 1.16515629e-01 -2.26688221e-01
-9.33083057e-01 9.20199528e-02 2.82124311e-01 4.52002913e-01
7.25968897e-01 -1.47669566e+00 -5.23751676e-01 2.05582395e-01
2.37157300e-01 -1.52590945e-01 7.03301966e-01 1.23775101e+00
-3.62721086e-01 1.77667037e-01 -5.09896815e-01 -9.63581502e-01
-9.30234730e-01 3.10341269e-01 3.16092134e-01 -4.22368884e-01
-9.78183985e-01 4.69168574e-01 1.44967020e-01 6.14239089e-02
2.46027485e-01 -3.87652755e-01 -5.43595612e-01 2.56737590e-01
4.95820701e-01 7.03765392e-01 -1.16151445e-01 -9.95631576e-01
-4.67029601e-01 1.05890393e+00 4.70998973e-01 -2.33293191e-01
1.22933877e+00 -3.89687717e-01 2.20021307e-01 3.55264783e-01
1.27862203e+00 -4.50336218e-01 -1.96971619e+00 -1.38381347e-01
7.62393326e-02 -8.23392153e-01 -9.81431380e-02 -3.81783277e-01
-1.17557716e+00 1.23973608e+00 6.60132468e-01 -8.58989507e-02
1.37881958e+00 -6.56967908e-02 7.44875312e-01 1.13569871e-01
2.27170497e-01 -9.63601470e-01 4.92416829e-01 2.47359887e-01
7.78093219e-01 -1.34003735e+00 6.11824580e-02 -1.90468356e-01
-6.20692611e-01 1.25096393e+00 6.30862892e-01 -1.74403206e-01
6.33762717e-01 -2.02177897e-01 8.72007571e-03 -1.91460758e-01
-5.14993727e-01 -4.20680881e-01 4.77153569e-01 7.43094742e-01
2.46287644e-01 -3.86856437e-01 -3.83526415e-01 1.95773914e-01
4.31546062e-01 1.53296962e-01 9.46315825e-02 1.00511968e+00
-2.07561493e-01 -8.38073730e-01 -1.49028927e-01 3.20893563e-02
-4.46531713e-01 4.58086044e-01 -1.31665781e-01 9.78426814e-01
1.59487143e-01 6.58778071e-01 6.19838480e-03 -2.67341822e-01
4.26142603e-01 -2.35687830e-02 6.99243248e-01 -3.40553559e-02
-3.50230426e-01 2.76950926e-01 6.22218102e-02 -1.21995580e+00
-1.26786947e+00 -8.90154243e-01 -1.19019222e+00 2.21514642e-01
1.53454006e-01 -3.72938722e-01 4.01364714e-01 1.07908583e+00
2.75613874e-01 4.79714423e-01 6.31193459e-01 -1.33906794e+00
-4.76323158e-01 -8.15759718e-01 -5.55542231e-01 7.67560005e-01
4.31821078e-01 -1.19660211e+00 -2.04603493e-01 1.99949026e-01] | [7.882893085479736, 0.34209510684013367] |
d75fe57c-f5f5-46b3-9ee5-3d965f1d9752 | skeleton-aided-articulated-motion-generation | 1707.01058 | null | http://arxiv.org/abs/1707.01058v2 | http://arxiv.org/pdf/1707.01058v2.pdf | Skeleton-aided Articulated Motion Generation | This work make the first attempt to generate articulated human motion
sequence from a single image. On the one hand, we utilize paired inputs
including human skeleton information as motion embedding and a single human
image as appearance reference, to generate novel motion frames, based on the
conditional GAN infrastructure. On the other hand, a triplet loss is employed
to pursue appearance-smoothness between consecutive frames. As the proposed
framework is capable of jointly exploiting the image appearance space and
articulated/kinematic motion space, it generates realistic articulated motion
sequence, in contrast to most previous video generation methods which yield
blurred motion effects. We test our model on two human action datasets
including KTH and Human3.6M, and the proposed framework generates very
promising results on both datasets. | ['Bingbing Ni', 'Yichao Yan', 'Jingwei Xu', 'Xiaokang Yang'] | 2017-07-04 | null | null | null | null | ['gesture-to-gesture-translation'] | ['computer-vision'] | [ 4.88836437e-01 2.30677366e-01 -4.79242206e-02 6.16364852e-02
-4.61961806e-01 -4.11344707e-01 9.45308685e-01 -1.00559628e+00
-4.18126553e-01 9.68032002e-01 1.87669858e-01 1.70102447e-01
3.81586283e-01 -5.46753645e-01 -6.64191246e-01 -8.56285691e-01
2.61315972e-01 9.21874419e-02 1.79420143e-01 1.33889094e-02
8.46778415e-03 8.76273066e-02 -1.31300187e+00 3.78039479e-02
6.73222244e-01 7.46684611e-01 2.48843014e-01 8.52455676e-01
2.96818674e-01 1.00833583e+00 -4.81342405e-01 -5.48209548e-01
4.23550338e-01 -8.34169030e-01 -7.99285471e-01 7.13220298e-01
4.08277303e-01 -6.52713954e-01 -5.03001571e-01 8.44530165e-01
4.62136447e-01 3.17778885e-01 6.40102446e-01 -1.24422109e+00
-5.98272264e-01 2.38070518e-01 -8.87415409e-01 -4.79358360e-02
5.60077846e-01 7.22360551e-01 6.04609728e-01 -7.49075115e-01
1.00434315e+00 1.30099368e+00 2.92440325e-01 1.01122856e+00
-1.05379987e+00 -3.45053226e-01 2.52057970e-01 1.10827573e-01
-9.85023618e-01 -5.01089334e-01 9.88070011e-01 -4.55953687e-01
4.98237520e-01 4.16070856e-02 9.66621518e-01 1.61225045e+00
2.69226819e-01 1.09731483e+00 9.07339334e-01 -3.55223328e-01
2.75676064e-02 -4.09791648e-01 -4.40615535e-01 7.47912765e-01
7.29813501e-02 3.34035963e-01 -4.81992662e-01 7.77729228e-02
1.21955049e+00 -1.16118096e-01 -6.73110723e-01 -5.22143781e-01
-1.55095530e+00 5.23847759e-01 9.63920876e-02 1.77485660e-01
-7.21521974e-01 4.53479052e-01 1.23947531e-01 -8.72695670e-02
1.72518134e-01 -2.19757743e-02 1.74210861e-01 -3.33599925e-01
-1.06701922e+00 3.33539486e-01 4.01886225e-01 9.24207091e-01
6.35643125e-01 5.94511330e-01 -3.22414815e-01 5.48152924e-01
3.24841946e-01 5.95683396e-01 5.61705947e-01 -1.29781604e+00
5.84433436e-01 5.85426129e-02 4.44339752e-01 -9.86397743e-01
7.36311600e-02 -4.93105538e-02 -1.04357004e+00 2.81966388e-01
4.62835878e-01 -2.09530517e-01 -1.26810515e+00 1.73168695e+00
4.33066756e-01 5.14355719e-01 2.11624041e-01 1.12949765e+00
7.07459331e-01 5.40973365e-01 1.16785541e-01 -1.91486850e-01
1.02463222e+00 -1.46307993e+00 -1.00734687e+00 1.11529976e-02
1.25392571e-01 -7.53928602e-01 8.29534352e-01 4.30916518e-01
-1.69096112e+00 -7.84507811e-01 -9.40395951e-01 8.76536146e-02
2.85860151e-01 1.50564119e-01 3.48393351e-01 4.74323064e-01
-1.04531097e+00 4.22839195e-01 -9.81422246e-01 -1.59761116e-01
2.07562223e-01 3.34122702e-02 -4.02424186e-01 -2.38187894e-01
-1.15745640e+00 7.40722060e-01 5.16253173e-01 3.27203035e-01
-1.34536624e+00 -3.11163813e-01 -9.93926167e-01 -3.30004036e-01
1.57976612e-01 -1.35863030e+00 1.06395125e+00 -1.24075913e+00
-1.98497891e+00 5.41408360e-01 -1.42834306e-01 -3.71430784e-01
1.13234270e+00 -3.15039605e-01 -2.33147234e-01 5.40195346e-01
-1.61157891e-01 1.06654954e+00 1.25156653e+00 -1.36604881e+00
-5.39184034e-01 5.68303354e-02 3.75756854e-03 3.94646704e-01
-2.08071768e-02 -2.78263360e-01 -7.63856828e-01 -1.00793684e+00
-2.66271651e-01 -9.94423091e-01 -3.38932395e-01 -1.55191943e-01
-3.17993522e-01 3.10215235e-01 8.79708886e-01 -9.70997453e-01
1.12593961e+00 -1.82848835e+00 7.06835508e-01 -2.22863983e-02
3.00013367e-02 3.60352188e-01 -3.41805905e-01 1.68387145e-01
-3.94940488e-02 -2.24576429e-01 -5.43811500e-01 -5.51851630e-01
-1.47985250e-01 3.45588714e-01 -1.61124542e-01 3.71077299e-01
3.18597019e-01 1.28118694e+00 -1.05714297e+00 -6.03218794e-01
4.88184839e-01 6.49399877e-01 -4.21573788e-01 3.69580686e-01
-1.50055483e-01 1.01589620e+00 -3.94976109e-01 4.94251817e-01
5.78549922e-01 -2.20232010e-01 8.68356973e-02 -8.04547295e-02
1.91901878e-01 -3.61292243e-01 -1.05070639e+00 2.10783792e+00
-4.30111796e-01 3.19408983e-01 -4.92136925e-02 -6.43956482e-01
5.99362552e-01 6.90238118e-01 6.75686300e-01 -4.13544357e-01
5.14978962e-03 -8.21124241e-02 -1.97406784e-01 -4.67748016e-01
5.10097504e-01 -4.41417471e-02 2.42451772e-01 3.36064935e-01
2.50735153e-02 3.80337983e-02 2.09608719e-01 5.07404171e-02
1.05045056e+00 8.17309737e-01 1.38885543e-01 2.51787096e-01
7.14807868e-01 -1.90246984e-01 6.63846076e-01 4.17936683e-01
-2.10282847e-01 1.22363937e+00 2.57445797e-02 -4.53671336e-01
-1.25346911e+00 -1.33521748e+00 5.33648372e-01 3.07498664e-01
3.63128632e-01 -5.66910356e-02 -8.35542023e-01 -8.46094310e-01
-4.82653528e-01 5.60508490e-01 -5.05571425e-01 -2.61657797e-02
-9.00160611e-01 -4.70258296e-01 4.92382258e-01 5.33201456e-01
8.63809526e-01 -1.34344518e+00 -7.91558385e-01 2.40037039e-01
-6.37374997e-01 -1.28721809e+00 -7.82344997e-01 -5.57052612e-01
-6.28802657e-01 -1.10646176e+00 -1.42699850e+00 -6.35058045e-01
6.88460171e-01 1.34075195e-01 1.05095303e+00 -2.16806587e-03
-4.14492995e-01 6.31631017e-01 -3.50814819e-01 2.27286473e-01
-4.46816146e-01 -4.19820815e-01 -1.15673229e-01 3.50928962e-01
-2.60922045e-01 -6.05773032e-01 -1.02176559e+00 2.68920422e-01
-1.12601447e+00 4.35277164e-01 7.26158142e-01 1.11568630e+00
3.64412487e-01 -5.44430688e-02 2.25224033e-01 -4.83884454e-01
1.22890577e-01 -3.41667801e-01 -3.71546298e-01 3.40038985e-01
-1.99315503e-01 -1.13632366e-01 5.16263425e-01 -3.96218389e-01
-1.57157397e+00 3.96226466e-01 7.28310132e-03 -7.28206813e-01
-2.84936816e-01 -1.03464752e-01 -3.68245304e-01 1.72551483e-01
3.04621965e-01 4.98904884e-01 -9.97602195e-02 -2.68966168e-01
6.74695134e-01 2.23145202e-01 9.53010261e-01 -6.55386150e-01
1.01601386e+00 8.15905392e-01 -3.73372957e-02 -8.91700923e-01
-1.36178970e-01 -3.41882408e-01 -8.74587178e-01 -4.25285339e-01
1.36022925e+00 -9.29068685e-01 -3.83652210e-01 7.61060238e-01
-1.42851830e+00 -3.64434540e-01 -1.61894798e-01 6.85158908e-01
-1.03181744e+00 8.58915508e-01 -7.07611084e-01 -9.03056145e-01
-3.50720108e-01 -1.14382970e+00 1.27174687e+00 -3.86053622e-02
-3.14626604e-01 -1.01374221e+00 8.29777941e-02 4.79659885e-01
1.09520962e-03 8.68464708e-01 3.90985817e-01 2.87151992e-01
-8.14035356e-01 1.90385535e-01 4.66128811e-02 3.70222270e-01
4.22219396e-01 1.77386124e-02 -7.19836295e-01 -3.98805439e-01
-1.29807010e-01 -2.29993224e-01 9.93224144e-01 3.95312428e-01
6.85197890e-01 -2.49164939e-01 -9.01866332e-02 6.67925835e-01
1.11906052e+00 3.77616435e-01 1.01805997e+00 2.00711772e-01
1.00275123e+00 5.39708793e-01 6.70860171e-01 5.40427625e-01
2.18192950e-01 8.47571015e-01 2.71556705e-01 -2.77243644e-01
-5.06801426e-01 -4.54983830e-01 6.84425116e-01 6.84522808e-01
-5.93464196e-01 -4.08354998e-01 -5.62997818e-01 6.15544856e-01
-2.02117062e+00 -1.24561369e+00 9.99851836e-05 2.13276172e+00
5.39706588e-01 -1.56833574e-01 3.47220093e-01 -8.90758261e-02
7.75503755e-01 3.05313677e-01 -4.83211309e-01 1.61341697e-01
-2.31285512e-01 2.15802997e-01 2.84682214e-01 4.91571993e-01
-8.93884003e-01 8.59271407e-01 6.03420544e+00 5.86127937e-01
-9.11805451e-01 1.65349431e-02 5.29520035e-01 -2.34130859e-01
-4.39420342e-01 -4.83743846e-02 -3.76603037e-01 6.49613023e-01
4.81912762e-01 -3.91864032e-02 2.48406485e-01 4.91739333e-01
4.80789542e-01 1.41620962e-03 -8.54420900e-01 1.11227167e+00
4.32456210e-02 -1.04530013e+00 4.23997909e-01 8.17707852e-02
1.04167485e+00 -6.55695915e-01 2.60345131e-01 -1.84223220e-01
5.17476141e-01 -1.04734397e+00 8.91304016e-01 8.32222164e-01
7.27753401e-01 -8.18133712e-01 6.19052052e-01 2.07107678e-01
-1.24280202e+00 2.78928965e-01 -3.61151807e-02 7.23888725e-02
8.43928635e-01 3.51296037e-01 -4.79286283e-01 1.10146594e+00
5.21554351e-01 8.15302491e-01 -3.28640908e-01 7.96276152e-01
-3.57596695e-01 4.64072704e-01 -2.31760181e-02 6.58992469e-01
3.29667568e-01 -4.87078577e-01 6.22209668e-01 9.23308730e-01
5.45607865e-01 5.03425160e-03 2.20760599e-01 7.62741506e-01
2.45911524e-01 -2.33542189e-01 -7.45692611e-01 4.67946455e-02
3.42403688e-02 1.11255991e+00 -5.94212830e-01 -4.15908456e-01
-4.36252445e-01 1.63464165e+00 -4.20056842e-02 7.44975448e-01
-1.07695758e+00 1.24966025e-01 4.75610584e-01 -9.20263380e-02
4.96979028e-01 -3.67259324e-01 2.49772921e-01 -1.46057606e+00
1.06980063e-01 -9.40055132e-01 2.06655324e-01 -7.08411276e-01
-9.86005962e-01 4.82308060e-01 1.30324252e-02 -1.42007899e+00
-8.20854783e-01 -2.28442505e-01 -6.04904950e-01 7.48826504e-01
-1.25545776e+00 -1.49856234e+00 -4.04182315e-01 6.68252945e-01
8.04537892e-01 -7.69713670e-02 4.06647235e-01 3.37980896e-01
-4.68106449e-01 5.14313221e-01 -2.30988607e-01 1.29607916e-01
7.03021705e-01 -1.09354663e+00 6.65053904e-01 1.10935557e+00
2.23575443e-01 4.67533410e-01 5.87485850e-01 -7.54787982e-01
-1.21981466e+00 -1.15307820e+00 2.89901972e-01 -4.74806666e-01
1.88440084e-01 -1.90049149e-02 -6.44843757e-01 5.99220514e-01
6.54053628e-01 -9.90213081e-02 1.64877892e-01 -8.45428407e-01
2.38212436e-01 4.60817754e-01 -9.39179778e-01 7.96593726e-01
1.35962498e+00 -1.99896619e-02 -5.40125668e-01 -1.21377647e-01
4.45963323e-01 -4.32511628e-01 -8.46442759e-01 5.88278472e-01
6.81213379e-01 -9.36297834e-01 9.76301014e-01 -4.80207503e-01
7.42487013e-01 -6.36968553e-01 2.38198824e-02 -1.23920548e+00
-2.83998579e-01 -1.04992223e+00 -3.86455595e-01 1.29331028e+00
-7.59326434e-03 -1.44518524e-01 1.03375947e+00 4.83909607e-01
3.53607833e-02 -5.72801650e-01 -7.06222951e-01 -8.75147104e-01
-9.10900459e-02 -8.57558101e-02 4.32061762e-01 7.27480888e-01
-5.54049313e-01 1.98038563e-01 -1.15966225e+00 -1.77956909e-01
9.12652194e-01 1.06717674e-02 1.17925084e+00 -5.32267690e-01
-5.99785030e-01 -2.72480518e-01 -5.34489870e-01 -1.39456654e+00
3.17314088e-01 -4.33169812e-01 1.42698854e-01 -1.39907968e+00
8.88358951e-02 1.89632311e-01 -1.77395195e-01 1.41374871e-01
-4.49361086e-01 4.65611488e-01 4.12416309e-01 9.01090726e-02
-4.62973505e-01 9.90897298e-01 1.64882851e+00 -1.87750965e-01
-5.78399487e-02 -1.10263802e-01 -1.29919276e-01 9.33122277e-01
4.41603124e-01 5.34483045e-02 -6.74908042e-01 -4.67789620e-01
-4.43796873e-01 4.09024328e-01 7.53869236e-01 -1.02605140e+00
9.29360911e-02 -3.12400818e-01 6.84477687e-01 -6.11465633e-01
5.13852715e-01 -7.02946246e-01 6.14640415e-01 6.07657373e-01
-1.76467821e-01 2.58036375e-01 -9.26759541e-02 9.84299302e-01
-3.62829149e-01 7.48512894e-02 8.41931701e-01 -3.72047961e-01
-1.05283105e+00 4.50701237e-01 -2.18046159e-01 1.44628435e-01
1.09518003e+00 -4.47406262e-01 2.95086354e-02 -5.42284787e-01
-7.87022948e-01 1.12710312e-01 7.92031646e-01 5.46371281e-01
9.21214938e-01 -1.69586694e+00 -7.99339414e-01 6.55427342e-03
-1.79708615e-01 -2.28150305e-03 4.99106020e-01 6.70464993e-01
-6.79605007e-01 2.90426910e-01 -5.40957570e-01 -6.28203988e-01
-1.20388997e+00 7.12302566e-01 1.32083341e-01 -2.76828915e-01
-8.59593868e-01 5.71622610e-01 7.30624020e-01 1.41178817e-02
-4.93759895e-03 1.86382532e-02 -6.16851356e-03 -4.26886350e-01
4.56150323e-01 4.62436467e-01 -4.38414335e-01 -9.07646775e-01
-5.53270690e-02 7.31990635e-01 1.58049062e-01 -6.97315395e-01
9.59313869e-01 -4.09777641e-01 3.07071745e-01 2.65324295e-01
9.86594200e-01 -8.20495337e-02 -1.85002756e+00 3.90267745e-02
-1.45229861e-01 -7.67673969e-01 -4.79475796e-01 -5.04108965e-01
-1.30331218e+00 6.85251415e-01 4.31992441e-01 -4.58663583e-01
1.17091727e+00 -4.38990980e-01 1.12184882e+00 -1.19438015e-01
5.13748050e-01 -9.82602477e-01 5.54030955e-01 1.24047920e-01
7.54201293e-01 -1.14231837e+00 -7.31460825e-02 -3.38943064e-01
-9.13471758e-01 9.63793576e-01 8.47362518e-01 -4.47248369e-02
-2.45264228e-02 -1.19417123e-01 4.11672927e-02 1.99956536e-01
-6.82235897e-01 -2.63902038e-01 3.54032934e-01 8.27825904e-01
3.80118459e-01 -3.02815437e-01 -5.52633107e-01 1.06004089e-01
-1.67763252e-02 2.00053081e-01 5.75992525e-01 6.70848787e-01
4.65969704e-02 -1.19107378e+00 -4.19531703e-01 -1.34079084e-01
-4.57466483e-01 1.67507529e-01 -3.53516787e-01 9.09495354e-01
3.97304669e-02 9.12671089e-01 -2.05658481e-01 -2.27377087e-01
1.58274010e-01 5.21984100e-02 7.81599641e-01 -3.17288011e-01
-2.03833997e-01 1.21383682e-01 4.39127460e-02 -7.37090409e-01
-7.22404718e-01 -6.47888243e-01 -1.12008047e+00 -9.37251896e-02
2.10081339e-02 1.10400571e-02 1.51584953e-01 7.28064358e-01
1.11613892e-01 6.12969518e-01 5.26829898e-01 -1.14659929e+00
-3.41644794e-01 -7.14684963e-01 -4.97197449e-01 9.31374133e-01
3.71042341e-01 -7.14199245e-01 -2.37717837e-01 7.03037679e-01] | [10.86663818359375, -0.7326459884643555] |
d13765d6-0e7b-403c-be74-06f6c12fcef0 | breaking-on-device-training-memory-wall-a | 2306.10388 | null | https://arxiv.org/abs/2306.10388v1 | https://arxiv.org/pdf/2306.10388v1.pdf | Breaking On-device Training Memory Wall: A Systematic Survey | On-device training has become an increasingly popular approach to machine learning, enabling models to be trained directly on mobile and edge devices. However, a major challenge in this area is the limited memory available on these devices, which can severely restrict the size and complexity of the models that can be trained. In this systematic survey, we aim to explore the current state-of-the-art techniques for breaking on-device training memory walls, focusing on methods that can enable larger and more complex models to be trained on resource-constrained devices. Specifically, we first analyze the key factors that contribute to the phenomenon of memory walls encountered during on-device training. Then, we present a comprehensive literature review of on-device training, which addresses the issue of memory limitations. Finally, we summarize on-device training and highlight the open problems for future research. By providing a comprehensive overview of these techniques and their effectiveness in breaking memory walls, we hope to help researchers and practitioners in this field navigate the rapidly evolving landscape of on-device training. | ['Li Li', 'Rui Ma', 'Kahou Tam', 'Chunlin Tian', 'Shitian Li'] | 2023-06-17 | null | null | null | null | ['navigate'] | ['reasoning'] | [ 3.77762049e-01 6.76270872e-02 -1.12624240e+00 -1.01802073e-01
-5.62488973e-01 -3.93215179e-01 -1.63700193e-01 -1.72675103e-01
-2.46781990e-01 5.85559547e-01 -1.89059511e-01 -1.01250982e+00
-1.56171665e-01 -6.33765340e-01 -7.94778585e-01 -9.88487303e-02
2.09676608e-01 6.50015920e-02 -2.10226048e-02 3.26816291e-01
1.10881172e-01 3.06011736e-01 -1.65193617e+00 5.26646614e-01
7.20138192e-01 1.12732482e+00 2.92893857e-01 6.81300402e-01
-2.62080925e-03 6.31518424e-01 -6.41734838e-01 -2.55293995e-01
-2.19502673e-01 -1.56498358e-01 -7.05224335e-01 -3.65732946e-02
4.29943055e-01 -4.88477409e-01 -4.27698076e-01 5.14526665e-01
6.75073981e-01 1.55306071e-01 3.85679513e-01 -7.96921492e-01
-4.87528741e-01 3.36252242e-01 -3.46641615e-02 5.34367383e-01
2.09043324e-01 -1.57194048e-01 4.12613541e-01 -8.47380102e-01
3.09502393e-01 6.46522939e-01 9.84937966e-01 9.70636666e-01
-6.56958640e-01 -6.66791797e-01 3.23071957e-01 2.54582278e-02
-1.39345253e+00 -8.89643013e-01 7.91267335e-01 -4.00116622e-01
1.53386807e+00 4.56103772e-01 6.71976507e-01 1.00700378e+00
1.12320624e-01 9.85355794e-01 7.49631763e-01 -7.05307305e-01
2.29837343e-01 4.11790550e-01 2.90006965e-01 6.25705183e-01
3.55995297e-01 -1.14086764e-02 -7.66600847e-01 -1.57472625e-01
8.25425982e-01 1.51384458e-01 7.37100374e-03 -2.10956320e-01
-4.18730736e-01 6.51471198e-01 1.78527176e-01 6.04278982e-01
-6.42550588e-02 -1.23870447e-01 4.09896225e-01 3.89922597e-02
4.11975354e-01 4.52511787e-01 -7.62377799e-01 -6.54669523e-01
-1.14295185e+00 -3.75588566e-01 7.06703067e-01 1.01612830e+00
1.74577206e-01 1.26254544e-01 1.18551984e-01 8.88267756e-01
2.55225569e-01 5.77534974e-01 2.37990409e-01 -5.35809159e-01
7.20761061e-01 4.86728609e-01 1.23934902e-03 -6.61356866e-01
-3.58776689e-01 -4.55798209e-01 -7.71381259e-01 -3.43293399e-01
8.77991021e-02 -3.69143277e-01 -9.66416121e-01 1.34034324e+00
-1.11679129e-01 5.09421527e-01 -2.88959324e-01 3.27526093e-01
1.06765163e+00 2.32725158e-01 3.68732244e-01 -1.56889185e-01
1.12748098e+00 -1.25166678e+00 -7.80394256e-01 -6.49576545e-01
1.10610342e+00 -5.04355669e-01 1.34959352e+00 2.44437039e-01
-1.18659461e+00 -8.27442884e-01 -1.33489871e+00 -2.21171398e-02
-5.83706975e-01 4.31640029e-01 1.11861622e+00 1.46946776e+00
-7.60785162e-01 6.87937081e-01 -1.17692077e+00 -4.21584487e-01
7.46902883e-01 1.01088071e+00 1.27031907e-01 -1.32296592e-01
-9.48501468e-01 9.15005565e-01 1.93144798e-01 3.51078212e-02
-4.08832490e-01 -7.79352844e-01 -7.83769846e-01 -1.27517074e-01
2.80889690e-01 -7.28394151e-01 1.29128659e+00 -6.99392200e-01
-1.56881905e+00 9.34075356e-01 -3.36043775e-01 -1.67601168e-01
-1.79959089e-02 -5.10491014e-01 -9.12646353e-01 -3.81560624e-01
-3.01731676e-01 4.16995555e-01 5.07692814e-01 -9.41079140e-01
-5.50847173e-01 -2.26387113e-01 1.01394452e-01 1.28898054e-01
-9.56823230e-01 -1.18292019e-01 -9.91156042e-01 -4.40198392e-01
-2.57391572e-01 -1.13751054e+00 -5.13406843e-02 -2.40073219e-01
-3.37580681e-01 1.19341612e-01 8.97223055e-01 -5.05966783e-01
2.15152073e+00 -2.45552969e+00 -1.83943376e-01 3.23320739e-02
1.25587478e-01 7.75117517e-01 1.77238971e-01 6.81665540e-02
2.29582727e-01 4.57697690e-01 3.64801854e-01 -6.38755918e-01
-3.91039044e-01 3.89965415e-01 -1.32001847e-01 1.14554251e-02
-1.16983145e-01 1.15432823e+00 -5.89613378e-01 -3.16850781e-01
4.16504174e-01 6.23992026e-01 -4.59778816e-01 8.46371707e-03
1.32528111e-01 4.28579271e-01 -5.89569926e-01 8.29302728e-01
4.18547213e-01 -6.62503004e-01 3.41146767e-01 1.47555014e-02
1.29417911e-01 7.35439003e-01 -8.72202039e-01 1.67133951e+00
-9.09322441e-01 6.63426816e-01 -1.93390548e-02 -7.09083617e-01
4.69980240e-01 3.07125837e-01 3.16822410e-01 -8.96104157e-01
4.18554097e-01 4.58105773e-01 -2.19723627e-01 -3.78287435e-01
5.09220302e-01 1.72163740e-01 -3.68008651e-02 3.26131314e-01
-2.11827859e-01 2.97916561e-01 -8.68040770e-02 -3.41979355e-01
9.19573843e-01 -2.19657198e-01 7.30800182e-02 2.38354728e-02
2.30953395e-01 -4.40141350e-01 1.06236510e-01 9.84563112e-01
-4.24993671e-02 2.20298111e-01 -2.48018414e-01 -5.84887087e-01
-7.34947145e-01 -8.10143828e-01 -3.24450821e-01 1.24584138e+00
-2.53276322e-02 -6.62016928e-01 -7.74855852e-01 -8.14628303e-01
1.16989054e-02 3.45431268e-01 -5.72883308e-01 -3.03767383e-01
-4.99197066e-01 -8.02536130e-01 6.11528456e-01 9.93611991e-01
6.01510644e-01 -8.57107282e-01 -6.24123037e-01 -9.29828435e-02
-9.43074189e-03 -1.20003605e+00 -2.05083519e-01 4.09810692e-01
-1.69416106e+00 -7.69250572e-01 -5.05663335e-01 -8.86306167e-01
8.48213732e-01 4.42278564e-01 1.34328604e+00 5.12036264e-01
4.80335206e-02 4.34984624e-01 -1.75003976e-01 -4.85935867e-01
-1.61779076e-01 8.98432493e-01 1.83375657e-01 -6.11458719e-01
6.41835153e-01 -4.57191676e-01 -5.72954059e-01 3.89649600e-01
-6.32650793e-01 3.67004909e-02 7.52171874e-01 5.17313719e-01
4.82516170e-01 3.12261552e-01 3.56724292e-01 -1.25365829e+00
6.23524010e-01 -4.82294589e-01 1.29515845e-02 3.55017066e-01
-8.29369426e-01 -1.43715993e-01 3.40942234e-01 -9.83639598e-01
-6.30398989e-01 5.61646484e-02 -3.35595936e-01 -3.11694294e-01
7.07099736e-02 5.79059243e-01 -1.87761411e-01 -3.89889508e-01
6.51466608e-01 -5.07634096e-02 -5.50468028e-01 -7.50134349e-01
7.31751025e-02 1.05340981e+00 2.71989144e-02 -4.55130279e-01
2.37207383e-01 1.63683489e-01 -2.39253685e-01 -8.87780190e-01
-9.58058894e-01 -3.04134578e-01 -5.44521451e-01 2.19252668e-02
3.92627239e-01 -9.81711447e-01 -3.98837358e-01 3.74123722e-01
-7.69302309e-01 -7.67204940e-01 -2.34102249e-01 1.95121869e-01
-1.23593457e-01 -8.40447843e-02 -6.39139593e-01 -9.16523635e-01
-4.65821326e-01 -9.45501447e-01 1.03213322e+00 5.14598131e-01
-6.65744662e-01 -1.29966033e+00 -2.40613163e-01 6.64442480e-01
5.75518489e-01 -4.09294397e-01 9.22359467e-01 -2.84844249e-01
-3.28928977e-01 -5.29291034e-01 6.12395480e-02 2.32594028e-01
2.55617172e-01 -4.15365487e-01 -1.22170591e+00 -4.53520328e-01
-7.91170970e-02 -1.05471440e-01 5.09011567e-01 7.28532851e-01
1.42543113e+00 6.19957186e-02 -1.06085706e+00 5.83623469e-01
1.04566431e+00 4.63293254e-01 6.83646441e-01 2.70172238e-01
6.65677607e-01 -8.80701933e-03 5.66935599e-01 2.05923662e-01
1.96266904e-01 8.50412309e-01 -1.15946963e-01 -3.68279815e-01
-1.87476888e-01 -5.45334935e-01 8.23855922e-02 1.10933888e+00
-1.45074219e-01 -2.63302922e-01 -7.83212066e-01 3.44175339e-01
-1.66995072e+00 -4.56371307e-01 1.15070432e-01 2.40943623e+00
5.15441477e-01 5.59132040e-01 1.99949965e-01 4.49402601e-01
4.69717085e-01 -7.59160891e-02 -6.86102450e-01 -3.42299521e-01
1.66438103e-01 4.33743596e-01 5.63623071e-01 4.68531102e-01
-1.20264411e+00 9.13264930e-01 8.34734821e+00 8.02504182e-01
-1.41686571e+00 4.41566765e-01 8.26081574e-01 -3.90718341e-01
6.42418414e-02 -4.03207332e-01 -1.19141078e+00 7.30683625e-01
1.25904775e+00 3.96989226e-01 3.41408998e-01 1.25203001e+00
-2.47076042e-02 -2.90618271e-01 -1.18196785e+00 1.40063167e+00
-4.73350808e-02 -1.38288522e+00 -2.55269945e-01 2.67387331e-01
8.95699561e-01 -1.39121503e-01 6.07726991e-01 5.42241633e-01
-6.09940410e-01 -1.14804411e+00 1.32245645e-01 4.43662815e-02
1.33852196e+00 -5.01733720e-01 6.69772506e-01 3.16274703e-01
-1.21553111e+00 -2.71868825e-01 -2.11106502e-02 -5.57656169e-01
1.52877728e-02 7.13514686e-01 -4.34317529e-01 2.17738785e-02
7.50556648e-01 6.38889015e-01 -6.16449356e-01 8.65854800e-01
1.43645480e-01 7.16879308e-01 -4.36186850e-01 -2.29439482e-01
-3.80614489e-01 4.48747694e-01 -2.82970458e-01 1.18652987e+00
3.90198022e-01 1.35968775e-01 2.87914854e-02 3.59564334e-01
-1.66850597e-01 -1.14502281e-01 -6.32403016e-01 -3.37495595e-01
7.82851160e-01 7.11773932e-01 -8.96350741e-01 -3.77459466e-01
-5.59484124e-01 1.01208568e+00 1.80984676e-01 1.02964431e-01
-8.94137919e-01 -2.77111102e-02 4.13023204e-01 4.54589665e-01
1.42452583e-01 -4.96033221e-01 -7.72405982e-01 -1.02593052e+00
2.31660068e-01 -8.09609652e-01 2.64467597e-01 -4.83188927e-01
-7.91830599e-01 4.60201591e-01 -1.23568490e-01 -1.12465084e+00
2.98960824e-02 -7.18429685e-01 -3.37725788e-01 5.32928407e-01
-1.17284203e+00 -1.02661943e+00 -2.72755146e-01 3.24829996e-01
6.33436263e-01 -6.75410181e-02 1.21313643e+00 9.46386695e-01
-8.04690361e-01 1.02093875e+00 1.14483044e-01 -1.92578465e-01
2.53957689e-01 -3.49693447e-01 5.72679818e-01 3.64261836e-01
1.39845654e-01 1.04073882e+00 1.68779433e-01 -8.22737634e-01
-1.49089324e+00 -7.40212500e-01 8.75568271e-01 -9.12772477e-01
1.62067562e-01 -4.62963909e-01 -6.87738419e-01 7.31328547e-01
-4.54929709e-01 1.60147585e-02 1.25174320e+00 8.28418612e-01
7.01076491e-03 1.23117000e-01 -8.66163671e-01 5.96988201e-01
1.51104057e+00 -8.66802931e-01 1.44279122e-01 1.24207370e-01
1.95558861e-01 -7.98464358e-01 -8.19015622e-01 6.61723137e-01
8.97122145e-01 -6.46199167e-01 8.31592262e-01 -6.35969818e-01
7.60340989e-02 2.80080855e-01 2.54182350e-02 -7.32053339e-01
-2.01118395e-01 -6.51428819e-01 -8.88208747e-01 1.03974903e+00
7.80604124e-01 -3.38175952e-01 1.52723753e+00 1.01543677e+00
-1.68582480e-02 -1.32224274e+00 -7.40024149e-01 -8.03801239e-01
-4.86443043e-02 -8.43670309e-01 3.16827178e-01 7.30597198e-01
1.66726008e-01 5.37855148e-01 -6.47137761e-01 -3.31213951e-01
-1.54799834e-01 -4.54367489e-01 5.71272731e-01 -1.14845109e+00
-5.33758104e-01 -2.43788719e-01 -1.10209398e-01 -1.82604456e+00
-1.43197000e-01 -5.34526467e-01 -2.81996310e-01 -1.30264580e+00
2.87359059e-01 -8.88117373e-01 -3.67179245e-01 3.65431428e-01
-1.32109001e-01 4.85231698e-01 1.58921018e-01 2.37847120e-01
-8.74928713e-01 7.55786747e-02 1.06502378e+00 -1.80024542e-02
-7.51788437e-01 4.57801074e-01 -9.09309804e-01 6.81082726e-01
9.96040463e-01 -4.00819778e-01 -7.57073164e-01 -6.70120537e-01
3.69122863e-01 -1.78731740e-01 -1.88899264e-01 -1.29918718e+00
2.15538412e-01 1.44904152e-01 4.51689661e-01 -3.30656171e-01
6.09132111e-01 -9.65112388e-01 2.43221283e-01 2.86565185e-01
-1.78234447e-02 1.04187563e-01 8.05336416e-01 1.95278108e-01
1.22987494e-01 -2.50367939e-01 4.72164959e-01 2.43923560e-01
-5.35322249e-01 8.92009363e-02 -5.01963377e-01 -1.67472526e-01
9.15318131e-01 -5.74207604e-01 -1.29258305e-01 -3.33581120e-01
-7.80730367e-01 -2.10630670e-01 3.30096543e-01 7.00074077e-01
4.51017082e-01 -1.27829790e+00 4.01841611e-01 4.93854463e-01
-1.68583617e-01 -1.34439483e-01 6.20403588e-01 6.96371794e-01
-1.05939776e-01 8.50406229e-01 1.85894675e-03 -3.54669780e-01
-1.63011634e+00 6.01841033e-01 4.07859445e-01 -5.36605418e-01
-2.23796502e-01 7.78152823e-01 -3.33146304e-01 -2.81327754e-01
5.66276550e-01 -3.13602507e-01 -1.06984489e-02 -4.18685704e-01
5.72951734e-01 5.12599051e-01 4.40852314e-01 -2.94165641e-01
-5.32726645e-01 8.05788338e-01 -1.14244625e-01 1.53905571e-01
8.30660760e-01 1.05241304e-02 4.94794041e-01 5.39079666e-01
9.82606053e-01 6.90031983e-03 -9.62866545e-01 7.08378702e-02
-1.77292570e-01 -2.85818040e-01 1.94246858e-01 -8.44147146e-01
-1.01035440e+00 1.03290594e+00 9.95391846e-01 1.15993664e-01
1.23568654e+00 -2.30086353e-02 1.12460196e+00 2.42749706e-01
4.74813819e-01 -1.37606502e+00 1.38807133e-01 5.65569937e-01
1.22498788e-01 -1.27503967e+00 8.88697281e-02 -8.58873606e-01
-2.17453554e-01 7.52561927e-01 5.55606306e-01 3.31514508e-01
1.17552018e+00 6.30871117e-01 -9.06266794e-02 -7.78096989e-02
-3.15503269e-01 8.86702910e-02 5.14868796e-01 7.86152899e-01
7.54668474e-01 -3.57580162e-03 -5.98892160e-02 9.40996826e-01
-7.31737465e-02 6.62547529e-01 -2.72864997e-01 1.31538367e+00
-9.73023102e-02 -1.56147563e+00 -1.57514021e-01 7.78475225e-01
-6.35615766e-01 -2.52578467e-01 -3.79358113e-01 6.89450562e-01
5.16759992e-01 1.26333737e+00 3.01089473e-02 -9.81115460e-01
4.81054515e-01 1.44209251e-01 7.17702270e-01 -7.98623800e-01
-5.35860896e-01 -2.37289608e-01 2.95372903e-01 -1.71744794e-01
-2.66657859e-01 -3.28340232e-01 -9.28995848e-01 -6.31002128e-01
-6.61687851e-01 -7.75518715e-02 8.60081613e-01 1.02461505e+00
9.48116183e-01 8.25398445e-01 1.35794729e-01 -9.72028315e-01
1.18739873e-01 -8.39263022e-01 -3.27939719e-01 6.59257323e-02
8.92070606e-02 -8.03761840e-01 -1.22682288e-01 -7.19309226e-02] | [8.595466613769531, 3.1079325675964355] |
d57457e9-9045-4ab8-afb6-a4e261cc9b87 | large-scale-study-of-curiosity-driven | 1808.04355 | null | http://arxiv.org/abs/1808.04355v1 | http://arxiv.org/pdf/1808.04355v1.pdf | Large-Scale Study of Curiosity-Driven Learning | Reinforcement learning algorithms rely on carefully engineering environment
rewards that are extrinsic to the agent. However, annotating each environment
with hand-designed, dense rewards is not scalable, motivating the need for
developing reward functions that are intrinsic to the agent. Curiosity is a
type of intrinsic reward function which uses prediction error as reward signal.
In this paper: (a) We perform the first large-scale study of purely
curiosity-driven learning, i.e. without any extrinsic rewards, across 54
standard benchmark environments, including the Atari game suite. Our results
show surprisingly good performance, and a high degree of alignment between the
intrinsic curiosity objective and the hand-designed extrinsic rewards of many
game environments. (b) We investigate the effect of using different feature
spaces for computing prediction error and show that random features are
sufficient for many popular RL game benchmarks, but learned features appear to
generalize better (e.g. to novel game levels in Super Mario Bros.). (c) We
demonstrate limitations of the prediction-based rewards in stochastic setups.
Game-play videos and code are at
https://pathak22.github.io/large-scale-curiosity/ | ['Harri Edwards', 'Yuri Burda', 'Deepak Pathak', 'Amos Storkey', 'Alexei A. Efros', 'Trevor Darrell'] | 2018-08-13 | large-scale-study-of-curiosity-driven-1 | https://openreview.net/forum?id=rJNwDjAqYX | https://openreview.net/pdf?id=rJNwDjAqYX | iclr-2019-5 | ['snes-games'] | ['playing-games'] | [-2.25060329e-01 -1.51517153e-01 7.05160499e-02 -3.05006862e-01
-8.65423322e-01 -7.04992712e-01 5.64089000e-01 -3.07375193e-02
-9.05638337e-01 1.04427373e+00 2.17653781e-01 7.44801015e-02
-3.10215265e-01 -6.63703620e-01 -6.94975913e-01 -7.64982700e-01
-5.51504016e-01 3.41859996e-01 3.94598663e-01 -7.72160113e-01
4.30738419e-01 1.03084549e-01 -1.62902403e+00 -5.80968969e-02
6.38969123e-01 7.54820526e-01 4.72959816e-01 1.12621546e+00
4.73360717e-01 1.15757525e+00 -5.59991777e-01 -6.95439354e-02
5.82025468e-01 -6.21104002e-01 -6.97866380e-01 -5.16874671e-01
-1.07346691e-01 -3.62225622e-01 -2.18596324e-01 7.94862449e-01
6.46716118e-01 5.39408207e-01 4.09050971e-01 -1.41552389e+00
-5.01733005e-01 7.73937285e-01 -2.42761523e-01 3.58027846e-01
2.73538947e-01 8.05112422e-01 1.37028849e+00 -3.15544367e-01
7.55142331e-01 8.93310964e-01 5.44627905e-01 7.94281125e-01
-1.24182224e+00 -6.35066152e-01 4.24890183e-02 8.00603032e-02
-8.69239688e-01 -2.06298649e-01 5.36961734e-01 -5.53107381e-01
1.10618508e+00 2.66630054e-01 5.29943824e-01 1.33731461e+00
2.68658429e-01 7.58374035e-01 1.33708131e+00 6.39468571e-03
6.65186644e-01 -6.73310310e-02 -2.56247818e-01 4.52108204e-01
-1.03385091e-01 9.16084588e-01 -6.77741408e-01 1.11673884e-02
7.80570328e-01 -2.13393345e-01 3.72229703e-02 -6.95908666e-01
-1.21034026e+00 1.02214193e+00 6.25196576e-01 2.39682332e-01
-3.78141522e-01 7.56890357e-01 3.76129866e-01 4.79898810e-01
1.56082120e-02 1.25622928e+00 -6.48087859e-01 -9.93799329e-01
-3.73656780e-01 7.20953345e-01 4.73100960e-01 8.48195612e-01
8.19848657e-01 2.12201700e-01 -1.62476122e-01 6.94604397e-01
-6.69804215e-02 3.34704071e-01 7.41363525e-01 -1.40255153e+00
2.84907043e-01 1.28166005e-01 4.65076387e-01 -7.88054645e-01
-8.13174665e-01 -5.44087410e-01 -5.68761118e-02 8.34102988e-01
7.90751398e-01 -5.37155211e-01 -4.63552237e-01 2.19270658e+00
1.62762418e-01 -4.46840338e-02 1.77334905e-01 1.18478894e+00
7.26617813e-01 1.23304985e-01 4.57454324e-02 2.28353217e-01
8.76928031e-01 -1.07401145e+00 -1.83797896e-01 -5.58692098e-01
9.26019192e-01 -3.43726277e-01 1.66138768e+00 5.57620764e-01
-1.07736588e+00 -4.17612612e-01 -9.69728112e-01 1.58136249e-01
-3.17803204e-01 -2.85857707e-01 9.94610727e-01 5.22696793e-01
-1.08191335e+00 9.99751449e-01 -8.76766205e-01 -3.13178778e-01
1.74508810e-01 4.04540330e-01 -2.55970389e-01 3.52210075e-01
-1.18136191e+00 9.71388340e-01 3.94824415e-01 -4.21268910e-01
-1.32289362e+00 -3.48891973e-01 -6.98538125e-01 1.05944835e-01
7.32553720e-01 -1.88760608e-01 1.54038894e+00 -1.15775943e+00
-1.66162312e+00 5.00986278e-01 4.29461718e-01 -4.71486092e-01
5.16159356e-01 -3.64156812e-01 2.16435958e-02 -2.53638953e-01
2.00502038e-01 8.43472481e-01 5.82206309e-01 -1.21629322e+00
-7.12780654e-01 -9.75479558e-03 5.72732627e-01 5.81090808e-01
-5.48259318e-02 -1.14465863e-01 1.20486850e-02 -5.37254035e-01
-2.80904353e-01 -1.01121771e+00 -5.65468967e-01 -5.35915017e-01
-1.02645136e-01 -1.87428936e-01 2.18217671e-02 -1.04071505e-01
9.30971682e-01 -2.12280917e+00 1.68956250e-01 -1.41885420e-02
3.19563985e-01 -3.20139199e-01 -4.25635457e-01 3.92743737e-01
4.69424650e-02 -3.75300669e-03 5.86535521e-02 -1.38779029e-01
3.48936558e-01 2.38773540e-01 -2.36688145e-02 1.30271360e-01
1.41776145e-01 9.67386544e-01 -1.41708374e+00 -1.26286745e-01
1.15651831e-01 -5.36315255e-02 -1.06187868e+00 2.35635325e-01
-3.69615793e-01 6.57423615e-01 -5.44202685e-01 3.89261872e-01
6.80305138e-02 -9.49692428e-02 -8.42818469e-02 5.37737310e-01
-1.56691894e-01 5.80427170e-01 -1.06649780e+00 1.73746610e+00
-3.43805164e-01 4.17839080e-01 -1.21838555e-01 -5.85608602e-01
8.79246294e-01 -2.01812208e-01 6.20084584e-01 -9.62289512e-01
2.34920278e-01 3.10150683e-01 5.07852256e-01 -1.90319106e-01
9.23824966e-01 -1.58192948e-01 -4.20111775e-01 6.08519197e-01
2.71200478e-01 -3.61366153e-01 4.61241245e-01 1.89864308e-01
1.52140331e+00 5.63451529e-01 1.15440346e-01 -2.96379983e-01
-1.77540660e-01 2.47016639e-01 7.17756629e-01 1.23939598e+00
-5.06511450e-01 5.59073806e-01 9.26980376e-01 -3.93559575e-01
-9.79632020e-01 -1.02075064e+00 1.76632926e-01 1.73572302e+00
1.33059740e-01 -5.85781693e-01 -5.76060176e-01 -6.60579562e-01
4.51589972e-02 7.28311360e-01 -9.46190953e-01 -3.22284043e-01
-3.81279200e-01 -6.61872685e-01 4.93522912e-01 7.00037003e-01
1.12334125e-01 -1.50475955e+00 -1.09527123e+00 2.96645284e-01
1.17980123e-01 -7.44469464e-01 -4.09965724e-01 8.51828039e-01
-5.41478276e-01 -1.00366604e+00 -3.99360925e-01 -3.09253633e-01
6.91331998e-02 1.03653912e-02 1.47444749e+00 5.00367433e-02
-1.84222028e-01 5.28708220e-01 -6.44800007e-01 -2.38013640e-01
-8.23915452e-02 1.51162595e-01 3.02308619e-01 -6.84203744e-01
3.00431490e-01 -6.80631340e-01 -5.62631488e-01 4.46520001e-01
-4.50452030e-01 -1.06198624e-01 5.24210215e-01 1.05334675e+00
3.84211540e-01 -1.71337038e-01 5.82563877e-01 -6.31903648e-01
9.32921410e-01 -6.22856259e-01 -7.56267786e-01 -3.91755342e-01
-4.48594809e-01 3.22173953e-01 6.71920776e-01 -5.31303942e-01
-5.65715909e-01 -8.90518501e-02 -8.90184939e-02 -5.99910691e-02
-8.62068608e-02 3.33678454e-01 2.56721288e-01 5.91537058e-02
1.31644845e+00 -4.52583060e-02 -1.02516547e-01 -1.73915103e-01
3.02036315e-01 1.17168084e-01 2.74251699e-01 -1.05992389e+00
7.04953909e-01 -1.55871389e-02 -1.12728447e-01 -2.79004931e-01
-4.32287723e-01 -1.82653829e-01 -7.39324987e-02 -3.32954586e-01
6.72625065e-01 -7.97283947e-01 -1.24007750e+00 1.81071937e-01
-3.88768613e-01 -1.37780499e+00 -8.27701211e-01 7.43247330e-01
-1.26324153e+00 -1.93541288e-01 -5.45767903e-01 -8.10292065e-01
1.19603835e-01 -1.43574214e+00 8.16805422e-01 4.59138632e-01
-2.61134356e-01 -6.80610180e-01 4.67185199e-01 -1.30982464e-02
5.66176176e-01 3.40040356e-01 2.66800374e-01 -6.21454120e-01
-4.99334842e-01 3.41904163e-01 1.84639730e-02 8.15091580e-02
3.36379409e-02 -2.33213484e-01 -9.87905145e-01 -2.64424175e-01
-3.26785505e-01 -1.03199172e+00 7.41471112e-01 3.50669980e-01
8.37459445e-01 6.35612523e-03 3.40104222e-01 5.68896234e-01
1.34096360e+00 2.43063048e-01 5.32630682e-01 9.09415662e-01
3.65084738e-01 4.97510582e-01 9.77623284e-01 7.89223909e-01
5.47521591e-01 8.01245332e-01 7.28380084e-01 1.50258765e-01
1.66529760e-01 -2.95917094e-01 6.12214029e-01 2.03555062e-01
-3.09797347e-01 4.80654277e-02 -8.18802953e-01 4.83379275e-01
-2.07454562e+00 -1.11576784e+00 6.60501942e-02 2.21657538e+00
9.18323696e-01 5.65711439e-01 6.51186883e-01 -2.07273737e-01
1.39982238e-01 1.17288858e-01 -9.68393803e-01 -6.90693617e-01
-3.10072582e-03 2.36444488e-01 7.20128536e-01 6.81384921e-01
-9.45222318e-01 1.28957653e+00 6.16287374e+00 8.91463816e-01
-9.88516033e-01 9.57999304e-02 4.70090866e-01 -6.18616521e-01
-1.73057035e-01 6.75013736e-02 -4.90131825e-01 3.45245123e-01
7.98567116e-01 -1.53707743e-01 1.00652981e+00 1.27762604e+00
2.46972769e-01 -3.35246146e-01 -1.06129193e+00 7.43008494e-01
-5.44068515e-01 -9.56952035e-01 -9.02121484e-01 1.73436493e-01
6.56028092e-01 5.66234112e-01 2.88954735e-01 9.87638891e-01
1.26253128e+00 -1.19276130e+00 8.81812394e-01 3.87970895e-01
5.58525801e-01 -8.30250859e-01 6.34926021e-01 1.79858312e-01
-8.43512416e-01 -2.29876161e-01 -4.35046226e-01 -5.43011010e-01
-3.38233054e-01 2.86944900e-02 -7.67732620e-01 8.03748220e-02
9.53536034e-01 6.83709383e-01 -7.97494590e-01 8.33511293e-01
-2.86128074e-01 5.02897561e-01 -4.57448065e-01 -4.80693072e-01
7.11955428e-01 -7.13905022e-02 5.33034027e-01 8.44976902e-01
1.55989647e-01 -8.98192227e-02 1.68322891e-01 8.98405135e-01
1.32045612e-01 6.72159493e-02 -6.43432260e-01 -1.90058261e-01
1.76887050e-01 1.30641913e+00 -5.89158416e-01 5.09873964e-02
-8.97481218e-02 6.36762798e-01 6.54076099e-01 2.18981698e-01
-9.22332942e-01 -2.49929711e-01 9.76808071e-01 -1.46447554e-01
3.25354755e-01 -2.84751803e-01 -4.37895030e-01 -1.02980793e+00
-1.40152305e-01 -9.73849177e-01 2.31635243e-01 -8.13229203e-01
-1.12078023e+00 5.24860620e-01 -3.71111184e-01 -1.39044380e+00
-5.38799882e-01 -6.83667660e-01 -6.34921074e-01 7.14458823e-01
-1.31023467e+00 -5.42119205e-01 -2.49893695e-01 6.43153131e-01
3.63683105e-01 -3.33921403e-01 8.18810880e-01 -9.70437229e-02
-2.64637500e-01 5.93110859e-01 2.91884810e-01 -2.84470934e-02
8.54362428e-01 -1.84753788e+00 3.60311955e-01 4.15910184e-01
3.28785889e-02 4.60546076e-01 1.07836199e+00 -4.20386434e-01
-1.27771199e+00 -6.26684248e-01 1.09660171e-01 -7.24717319e-01
8.94221604e-01 -4.27528739e-01 -4.48254168e-01 5.37315547e-01
-2.77342647e-02 -4.11636829e-02 5.87269187e-01 6.42047822e-01
-1.90825030e-01 6.48700073e-02 -1.01850474e+00 9.12960947e-01
1.12440765e+00 -1.93830416e-01 -3.33535939e-01 1.94354221e-01
6.59056723e-01 -5.78272760e-01 -7.89622009e-01 8.37548897e-02
6.27220869e-01 -1.26599991e+00 7.47864723e-01 -8.17836106e-01
7.59628296e-01 -2.40306959e-01 -3.46993744e-01 -1.62209117e+00
-4.90821838e-01 -6.45145416e-01 2.02901557e-01 8.09118688e-01
5.49882293e-01 -5.54592431e-01 1.03777814e+00 6.74893618e-01
-1.18133068e-01 -8.22781086e-01 -7.60947168e-01 -1.12961090e+00
3.38481516e-01 -6.48862302e-01 5.34114957e-01 8.23818088e-01
4.11619425e-01 2.77028561e-01 -5.24461865e-01 -2.10605487e-01
3.58256400e-01 -9.78022143e-02 1.12973630e+00 -8.16554785e-01
-9.18419242e-01 -8.16272438e-01 -3.56214255e-01 -8.60098004e-01
-2.23575234e-01 -7.37304747e-01 3.69090885e-01 -1.00665951e+00
6.04811907e-02 -9.43510294e-01 -5.27855754e-01 5.16897917e-01
-2.78584540e-01 1.33480042e-01 4.18537736e-01 -3.21255289e-02
-1.04683304e+00 7.69703984e-01 1.30018377e+00 2.45443210e-01
-5.02346396e-01 -2.07657531e-01 -9.76419806e-01 6.59406304e-01
1.27152979e+00 -6.21020913e-01 -4.50399309e-01 -3.73154759e-01
5.48249722e-01 -2.67350189e-02 2.87854671e-01 -1.22693634e+00
3.87989767e-02 -6.55869603e-01 2.06281915e-01 1.36357024e-01
4.79904294e-01 -4.28622067e-01 -3.18740696e-01 4.14857298e-01
-6.19899213e-01 1.81691751e-01 1.68420658e-01 5.39614797e-01
1.29425272e-01 -3.60524535e-01 5.83520949e-01 -2.53761262e-01
-8.04298937e-01 2.91765835e-02 -3.67304593e-01 5.51178873e-01
9.88300860e-01 -2.06818447e-01 -3.28749031e-01 -6.95227504e-01
-7.53622472e-01 4.43612337e-01 6.55496657e-01 4.92608756e-01
3.60729694e-01 -1.21312177e+00 -6.79699361e-01 -1.42030418e-01
2.49817342e-01 -2.42268980e-01 7.85604119e-02 7.25719988e-01
-3.45416605e-01 -2.06270441e-02 -5.01582265e-01 -4.33595538e-01
-8.44699919e-01 3.98533016e-01 4.51035172e-01 -6.17991447e-01
-4.76807535e-01 1.06538498e+00 2.70021319e-01 -7.58774757e-01
1.81410268e-01 -3.27345550e-01 -8.37007314e-02 -3.28925461e-01
3.12042356e-01 9.41773206e-02 -1.86499104e-01 -2.53815800e-01
-2.92204022e-01 1.36861846e-01 -2.26408895e-03 -4.70619440e-01
1.53281569e+00 2.14843288e-01 6.10786080e-01 5.73975980e-01
6.12363756e-01 -9.07953978e-02 -1.87104702e+00 -5.44287264e-02
9.20851827e-02 -5.63608289e-01 1.05987247e-02 -9.63930190e-01
-8.64125431e-01 7.08191693e-01 6.71690762e-01 2.61034578e-01
8.97772431e-01 -1.49453998e-01 4.74782288e-01 6.54900193e-01
8.45406115e-01 -1.56460917e+00 5.13681054e-01 7.91289806e-01
9.16258097e-01 -1.42736292e+00 -1.97553039e-01 4.35029924e-01
-1.31110489e+00 7.25689769e-01 9.97666061e-01 -5.28136790e-01
1.94948852e-01 2.46356070e-01 -1.02278687e-01 -1.90747812e-01
-1.02388191e+00 -6.66319072e-01 -3.12477231e-01 7.30365098e-01
2.89828539e-01 3.35828215e-01 -2.43239090e-01 8.27382445e-01
-7.35850811e-01 -1.49356961e-01 7.19213188e-01 1.01239228e+00
-4.62622344e-01 -1.11294675e+00 -1.69168860e-01 5.95880806e-01
-3.51201773e-01 -1.03784092e-01 -3.58892798e-01 7.70074844e-01
-4.18460555e-02 1.02149284e+00 -1.35243297e-01 -7.63736844e-01
1.92678005e-01 -3.25229824e-01 4.40429926e-01 -6.58207476e-01
-9.49006915e-01 -1.75185591e-01 1.93709552e-01 -9.94810820e-01
2.32661795e-03 -7.01299369e-01 -1.28231132e+00 -3.70287299e-01
-1.26376733e-01 2.19457373e-01 5.42370319e-01 6.23793364e-01
2.42321312e-01 6.38511896e-01 6.65604413e-01 -9.35953498e-01
-5.71764588e-01 -9.06071603e-01 -7.33011365e-01 4.67498243e-01
2.26782650e-01 -1.08762562e+00 -4.80944008e-01 -4.65352923e-01] | [3.8789546489715576, 1.6108957529067993] |
e0f490ca-d290-4fc8-a2ce-4c9000244e33 | graph-neural-network-for-cell-tracking-in | 2202.04731 | null | https://arxiv.org/abs/2202.04731v2 | https://arxiv.org/pdf/2202.04731v2.pdf | Graph Neural Network for Cell Tracking in Microscopy Videos | We present a novel graph neural network (GNN) approach for cell tracking in high-throughput microscopy videos. By modeling the entire time-lapse sequence as a direct graph where cell instances are represented by its nodes and their associations by its edges, we extract the entire set of cell trajectories by looking for the maximal paths in the graph. This is accomplished by several key contributions incorporated into an end-to-end deep learning framework. We exploit a deep metric learning algorithm to extract cell feature vectors that distinguish between instances of different biological cells and assemble same cell instances. We introduce a new GNN block type which enables a mutual update of node and edge feature vectors, thus facilitating the underlying message passing process. The message passing concept, whose extent is determined by the number of GNN blocks, is of fundamental importance as it enables the `flow' of information between nodes and edges much behind their neighbors in consecutive frames. Finally, we solve an edge classification problem and use the identified active edges to construct the cells' tracks and lineage trees. We demonstrate the strengths of the proposed cell tracking approach by applying it to 2D and 3D datasets of different cell types, imaging setups, and experimental conditions. We show that our framework outperforms current state-of-the-art methods on most of the evaluated datasets. The code is available at our repository: https://github.com/talbenha/cell-tracker-gnn. | ['Tammy Riklin Raviv', 'Tal Ben-Haim'] | 2022-02-09 | null | null | null | null | ['3d-multi-object-tracking'] | ['computer-vision'] | [-6.99426457e-02 -3.35122585e-01 3.86007093e-02 2.05658991e-02
-3.43832254e-01 -6.38889015e-01 6.75921977e-01 4.95642245e-01
-7.57003605e-01 9.69592750e-01 -1.79734334e-01 -3.01183797e-02
-6.11222163e-02 -8.28385174e-01 -7.48610079e-01 -1.25166881e+00
-4.48515445e-01 4.85432804e-01 3.07188213e-01 1.44551158e-01
2.08878860e-01 8.91766667e-01 -1.03175700e+00 3.45322080e-02
1.17632650e-01 8.95112038e-01 9.29608643e-02 1.09180462e+00
-1.53317302e-01 8.25647831e-01 -2.28065833e-01 -1.99967157e-02
2.93706730e-02 -3.28190893e-01 -6.70296431e-01 1.25986142e-02
1.44116357e-01 -1.93933696e-01 -5.64839244e-01 8.35604012e-01
6.33404076e-01 -1.06799647e-01 6.93998694e-01 -1.26125360e+00
-2.82521456e-01 2.06141472e-01 -5.20572126e-01 7.07847536e-01
3.38803194e-02 1.60587892e-01 7.80204356e-01 -8.03600132e-01
1.44536602e+00 6.95055425e-01 6.93968475e-01 6.37866735e-01
-1.39020300e+00 -4.43949223e-01 -4.87816632e-02 5.95951453e-02
-1.31526065e+00 -4.79343772e-01 6.49696648e-01 -7.86917925e-01
7.58627117e-01 -6.60417676e-02 1.04558527e+00 9.44091916e-01
4.26776022e-01 6.40800178e-01 7.89840519e-01 -1.53669491e-01
2.62531668e-01 -4.75274712e-01 4.22164142e-01 9.42914844e-01
1.92674577e-01 -1.33128297e-02 -6.63461328e-01 -7.17224106e-02
9.40161705e-01 3.36967349e-01 -5.48481882e-01 -5.94552696e-01
-1.57383561e+00 5.14458120e-01 4.09624547e-01 5.91337800e-01
-2.46625781e-01 5.20054996e-01 5.28477132e-01 4.77307886e-02
3.32573742e-01 -7.88332745e-02 -4.45228547e-01 2.01594252e-02
-6.84921622e-01 1.22882664e-01 8.08770895e-01 6.38369858e-01
7.17443287e-01 -4.57230806e-01 -1.45088539e-01 3.97056371e-01
3.15658659e-01 -3.95201519e-02 2.20013037e-01 -9.11225915e-01
-2.87218004e-01 8.82082760e-01 -4.94314451e-03 -1.01565194e+00
-6.76289141e-01 -5.74154913e-01 -1.01146233e+00 2.72001415e-01
9.19330180e-01 -1.87989786e-01 -6.40766561e-01 1.77704060e+00
7.13904381e-01 5.77200532e-01 -2.32652158e-01 6.68196142e-01
8.89779031e-01 3.57313931e-01 -1.29132047e-01 -2.74076104e-01
1.29373133e+00 -5.65586329e-01 -7.35144198e-01 3.88956636e-01
1.05991173e+00 -2.65910745e-01 2.51242459e-01 -2.08320073e-03
-8.77533793e-01 -2.81628877e-01 -8.30161035e-01 -2.11396247e-01
-7.13869452e-01 7.02247992e-02 6.01695180e-01 -8.08028504e-02
-1.21178198e+00 8.75386536e-01 -1.13969386e+00 -4.91252601e-01
9.67615962e-01 5.00819206e-01 -6.17070913e-01 6.82948530e-02
-5.98108172e-01 2.63307214e-01 2.94230312e-01 2.77971953e-01
-1.08228934e+00 -7.09740758e-01 -8.08023930e-01 -8.07971731e-02
3.38513404e-02 -9.34946299e-01 6.82701111e-01 -3.55595559e-01
-9.96943653e-01 1.31409526e+00 -4.24787700e-01 -3.77196789e-01
5.72255552e-01 2.19012335e-01 -2.64110882e-02 2.51669824e-01
-9.84880701e-02 8.02189469e-01 1.33026093e-01 -1.25030494e+00
-8.10694218e-01 -5.95102370e-01 -1.87499538e-01 -1.50455207e-01
-1.41403586e-01 -2.36696988e-01 -7.63224542e-01 -3.57521653e-01
-6.91251829e-02 -8.73026311e-01 -7.45774433e-02 5.35040259e-01
-5.30516446e-01 -1.85705461e-02 8.32129478e-01 -3.83378327e-01
1.06132150e+00 -2.05852890e+00 5.04574835e-01 1.03574730e-01
9.95679021e-01 -3.26260589e-02 6.30090609e-02 5.34670353e-01
2.27873147e-01 -8.68756399e-02 -2.58003503e-01 -6.53275907e-01
-2.48073116e-01 1.26921251e-01 2.85167783e-01 9.61743057e-01
-7.32742473e-02 1.06535006e+00 -1.16634965e+00 -5.93535006e-01
1.24563813e-01 8.71062636e-01 -1.38626277e-01 5.62267844e-03
6.07583113e-02 8.17088902e-01 -2.44357631e-01 6.79068685e-01
5.03068447e-01 -7.04639554e-01 2.86539108e-01 -2.25540116e-01
-1.33954898e-01 -2.37465635e-01 -8.72139394e-01 1.59288180e+00
2.55265325e-01 9.69227433e-01 1.71387047e-01 -9.72374618e-01
6.68313026e-01 1.82967052e-01 8.65546107e-01 -2.03244984e-01
3.84939551e-01 2.11544380e-01 -1.10817045e-01 -3.09165239e-01
-3.30571756e-02 1.87631249e-01 2.17279032e-01 2.85277396e-01
3.17807674e-01 5.50788045e-01 5.21971107e-01 4.47239131e-01
1.48554873e+00 1.72812089e-01 2.18474686e-01 -4.33760822e-01
6.31981730e-01 -1.81783050e-01 6.98775887e-01 5.77085316e-01
-5.14966130e-01 4.04989660e-01 8.62198055e-01 -7.25759864e-01
-1.00729942e+00 -9.39705610e-01 -7.76055828e-02 7.22708642e-01
1.78689077e-01 -8.86002332e-02 -6.78606808e-01 -6.80555224e-01
6.01165704e-02 -1.58762574e-01 -1.07722783e+00 1.18139274e-01
-5.91153026e-01 -6.83768094e-01 4.61204767e-01 2.77578473e-01
1.36982143e-01 -9.32330370e-01 -7.35382020e-01 3.36317986e-01
7.02394843e-02 -1.06772137e+00 -3.33790809e-01 3.22272956e-01
-8.11511695e-01 -1.39179909e+00 -6.38493478e-01 -1.00479829e+00
1.00148749e+00 1.24142230e-01 9.83346045e-01 5.17101228e-01
-4.99076098e-01 2.44558856e-01 -9.48588401e-02 -8.99726525e-02
-2.99833238e-01 3.03741358e-03 -7.18861967e-02 1.57082766e-01
3.21103811e-01 -6.62527382e-01 -9.26444829e-01 3.76014449e-02
-7.50387967e-01 8.60554576e-02 5.69712929e-02 9.78997886e-01
1.10303295e+00 -1.45552441e-01 4.02795851e-01 -9.96932626e-01
2.19247267e-01 -6.05027854e-01 -8.31486523e-01 5.25588542e-02
-1.36054335e-02 -1.39746666e-01 5.58931828e-01 -2.44386166e-01
-3.38848174e-01 1.65729329e-01 -6.48757294e-02 -3.35500002e-01
-1.92053750e-01 3.87523383e-01 7.85908848e-02 -3.73264521e-01
2.16364324e-01 3.19979519e-01 1.59354120e-01 -1.67827129e-01
2.18338277e-02 8.48954022e-02 5.30397475e-01 -2.92294383e-01
4.05647427e-01 1.14878893e+00 6.17717862e-01 -7.88255334e-01
-4.02536482e-01 -6.57532692e-01 -8.72606516e-01 -6.69653475e-01
8.02759171e-01 -6.22617662e-01 -1.25161254e+00 9.21901464e-01
-1.24215794e+00 -6.17889225e-01 -2.51105666e-01 3.54036987e-01
-4.93291348e-01 3.35348397e-01 -1.20569134e+00 -5.67969620e-01
-2.52877086e-01 -1.02901435e+00 1.19187355e+00 2.65363365e-01
-1.58884794e-01 -1.51355278e+00 5.32799840e-01 -1.38559282e-01
8.85327980e-02 7.58757591e-01 9.64335084e-01 -7.66724586e-01
-8.64412248e-01 -9.81032923e-02 -1.83132514e-01 -3.30986828e-01
2.57211268e-01 3.12009066e-01 -7.50510335e-01 -6.07521594e-01
-3.55009109e-01 -6.83085760e-03 1.03841710e+00 8.38885188e-01
8.61082315e-01 -7.14743510e-02 -1.08721280e+00 9.84356165e-01
1.62997639e+00 1.61558151e-01 3.95272851e-01 3.02992433e-01
9.15949464e-01 5.72055519e-01 1.58689097e-01 3.12149525e-01
2.43037283e-01 4.62854892e-01 4.87774283e-01 -2.87555426e-01
-1.81519940e-01 1.06216609e-01 -2.26185787e-02 8.59591246e-01
5.37790032e-03 -5.94584525e-01 -9.68461812e-01 7.84779489e-01
-1.94035256e+00 -7.85114586e-01 -2.89245993e-01 1.99914515e+00
5.53229094e-01 2.11078171e-02 1.15987964e-01 1.36696160e-01
8.64893973e-01 3.02147232e-02 -7.47313857e-01 2.47628242e-01
-2.18994424e-01 -1.08560242e-01 2.87794709e-01 5.35014749e-01
-9.97090340e-01 6.99698210e-01 5.36755371e+00 5.64170420e-01
-9.50740278e-01 -8.73588622e-02 9.51011062e-01 -2.06715718e-01
-1.36404276e-01 -1.41292170e-01 -9.24753129e-01 5.18843532e-01
4.59015816e-01 -1.05968229e-01 2.21592039e-01 2.22553402e-01
2.14329824e-01 -3.94188501e-02 -1.44614470e+00 8.84086549e-01
-3.45164537e-01 -1.98719430e+00 -1.67528400e-03 5.03328919e-01
5.33286631e-01 2.64683694e-01 -3.17454845e-01 -2.18822047e-01
3.17125857e-01 -7.30419219e-01 3.66028100e-01 7.17835486e-01
7.18696654e-01 -4.64552104e-01 8.11378837e-01 1.60196617e-01
-1.34023023e+00 1.54368162e-01 -1.43561497e-01 3.69191691e-02
4.34175879e-01 8.01099956e-01 -6.26960218e-01 4.30922210e-01
5.22189379e-01 1.15271974e+00 -2.46476784e-01 1.19909132e+00
4.06874895e-01 3.35559428e-01 -2.98460305e-01 -6.54897913e-02
8.71172845e-02 -3.53161871e-01 7.39448071e-01 1.38427913e+00
4.05962855e-01 5.63016720e-02 -1.28451139e-01 8.85226846e-01
-5.23486972e-01 -2.44670466e-01 -7.84978569e-01 2.86734067e-02
6.61782265e-01 1.56108773e+00 -1.46875584e+00 -2.48748153e-01
-2.23288760e-01 7.97753036e-01 7.67372012e-01 4.41140920e-01
-6.82398081e-01 -3.38451803e-01 8.67034733e-01 1.32129654e-01
4.68795508e-01 -2.42089257e-01 9.91441831e-02 -8.02905262e-01
-2.73016930e-01 -3.45913768e-01 4.11669433e-01 -4.49404359e-01
-1.22671306e+00 3.64181250e-01 -5.93341291e-01 -1.08740735e+00
2.26440951e-01 -7.34468818e-01 -5.19108176e-01 3.16506565e-01
-1.52570117e+00 -9.19059277e-01 -3.45278442e-01 4.78429377e-01
1.91508368e-01 1.16648652e-01 5.79925835e-01 3.90601605e-01
-7.11429119e-01 1.94428742e-01 3.50383461e-01 5.50920427e-01
1.67314634e-01 -1.15089154e+00 4.04922932e-01 7.37063587e-01
2.47936547e-01 6.42772853e-01 4.33085144e-01 -6.94472909e-01
-1.40121233e+00 -1.10460508e+00 8.89370322e-01 -5.25542498e-01
7.19227552e-01 -6.15166485e-01 -9.18818474e-01 6.99615777e-01
-2.73241196e-02 5.53880095e-01 7.67622471e-01 -4.27838027e-01
4.58869077e-02 -1.19112572e-02 -9.64025497e-01 5.48768163e-01
1.36591721e+00 -4.81376469e-01 1.70328438e-01 2.94408530e-01
5.47611475e-01 -3.54741991e-01 -1.02554846e+00 2.17074126e-01
7.04552054e-01 -1.00489318e+00 7.98528850e-01 -4.98959750e-01
3.24285984e-01 -5.42332768e-01 1.46904126e-01 -1.02256179e+00
-4.45713818e-01 -5.25726199e-01 -4.26227659e-01 1.08932126e+00
2.54976064e-01 -6.43746436e-01 9.59050715e-01 -1.52089760e-01
-1.45436779e-01 -1.25949299e+00 -1.11038721e+00 -4.92895752e-01
-3.71300243e-02 3.09951566e-02 3.24082732e-01 9.46866274e-01
-2.49910317e-02 2.49053776e-01 -1.88011713e-02 -1.42145744e-02
9.30943429e-01 -3.55404392e-02 7.19723701e-01 -1.35560107e+00
-1.22616112e-01 -6.07307136e-01 -7.89832830e-01 -1.03175604e+00
1.35938600e-01 -9.73520517e-01 -8.79306272e-02 -1.50338972e+00
3.93068552e-01 -2.87524760e-01 -5.45004487e-01 1.77670658e-01
-2.74996907e-02 3.32971781e-01 -4.77860868e-02 2.41400838e-01
-9.15127575e-01 2.81889468e-01 1.29045630e+00 -1.64710611e-01
-3.26390043e-02 -5.48371613e-01 -2.15366289e-01 6.24477983e-01
7.15341926e-01 -5.40204763e-01 1.42479418e-02 -4.01548713e-01
2.80875683e-01 1.47120267e-01 6.23980105e-01 -1.24823141e+00
6.08526468e-01 3.60068351e-01 5.33398390e-01 -6.05576158e-01
3.17072690e-01 -7.77405441e-01 5.48896551e-01 7.37142205e-01
-3.29749107e-01 -4.55881059e-02 1.37784541e-01 7.64639378e-01
4.43286113e-02 2.77760625e-01 7.82821476e-01 -1.29715785e-01
-3.26590717e-01 7.60385871e-01 -5.04794717e-01 1.41585842e-02
1.19686222e+00 -5.14627159e-01 -4.78265017e-01 1.47846192e-02
-7.95214236e-01 3.40553373e-01 8.40172648e-01 -1.13149963e-01
5.55918694e-01 -1.36584103e+00 -5.87805092e-01 1.52965814e-01
1.78650036e-01 5.92871122e-02 2.89493471e-01 1.08132279e+00
-7.08707869e-01 1.84486181e-01 -3.62939239e-01 -9.81331587e-01
-1.29234636e+00 4.26442385e-01 5.90797305e-01 -4.80473220e-01
-5.10169983e-01 1.02292550e+00 4.06094313e-01 -2.36982793e-01
1.60930917e-01 -2.70291865e-01 -3.99898499e-01 3.72631475e-02
3.59292120e-01 4.78875577e-01 -2.20322326e-01 -8.11529458e-01
-3.86111736e-01 7.97143459e-01 -2.57068664e-01 3.01608145e-01
1.37761223e+00 -2.47931525e-01 -4.24238801e-01 6.62880123e-01
1.44707668e+00 -1.98284239e-01 -1.59625518e+00 -8.48897081e-03
-2.53125932e-02 -2.30582803e-01 -1.15035035e-01 -2.71159083e-01
-1.36918187e+00 6.92893624e-01 5.61256289e-01 6.62535056e-02
8.43848228e-01 1.79139718e-01 9.27393258e-01 1.89814731e-01
4.85018373e-01 -5.84436297e-01 -6.85740113e-02 4.48036373e-01
1.66962206e-01 -9.69807267e-01 -2.04676196e-01 -2.27370963e-01
1.51213974e-01 1.06203008e+00 2.90859848e-01 -7.17387274e-02
7.23328710e-01 5.71122646e-01 1.40975202e-02 -7.98077762e-01
-8.41187954e-01 -1.08705029e-01 -3.74467552e-01 5.76027334e-01
3.43694746e-01 -1.81559935e-01 -1.88730955e-01 1.90193951e-01
3.80208790e-01 4.88229126e-01 5.14175594e-01 8.38784993e-01
-3.01849872e-01 -8.36374342e-01 2.61222214e-01 5.02544522e-01
-5.26512504e-01 1.01863794e-01 -3.06844085e-01 8.45982134e-01
1.67240188e-01 3.52698177e-01 3.45595419e-01 -2.00613469e-01
-9.90352184e-02 -1.61594763e-01 4.90898728e-01 -3.13222945e-01
-2.35538751e-01 2.22735312e-02 -2.18750834e-01 -5.49196541e-01
-5.72856069e-01 -6.89444840e-01 -1.59325850e+00 -4.32886124e-01
-1.86438113e-01 -3.81772919e-03 2.37108558e-01 7.89199233e-01
6.36031866e-01 6.39440358e-01 3.29267532e-01 -1.00288343e+00
3.71769190e-01 -4.41149503e-01 -6.94411576e-01 4.73312944e-01
6.81639493e-01 -6.53505981e-01 -3.91493201e-01 3.35689127e-01] | [14.571443557739258, -3.1927428245544434] |
0574ab6a-e58c-4d93-b1e9-88c46daa4cbd | language-quantized-autoencoders-towards | 2302.00902 | null | https://arxiv.org/abs/2302.00902v2 | https://arxiv.org/pdf/2302.00902v2.pdf | Language Quantized AutoEncoders: Towards Unsupervised Text-Image Alignment | Recent progress in scaling up large language models has shown impressive capabilities in performing few-shot learning across a wide range of text-based tasks. However, a key limitation is that these language models fundamentally lack visual perception - a crucial attribute needed to extend these models to be able to interact with the real world and solve vision tasks, such as in visual-question answering and robotics. Prior works have largely connected image to text through pretraining and/or fine-tuning on curated image-text datasets, which can be a costly and expensive process. In order to resolve this limitation, we propose a simple yet effective approach called Language-Quantized AutoEncoder (LQAE), a modification of VQ-VAE that learns to align text-image data in an unsupervised manner by leveraging pretrained language models (e.g., BERT, RoBERTa). Our main idea is to encode image as sequences of text tokens by directly quantizing image embeddings using a pretrained language codebook. We then apply random masking followed by a BERT model, and have the decoder reconstruct the original image from BERT predicted text token embeddings. By doing so, LQAE learns to represent similar images with similar clusters of text tokens, thereby aligning these two modalities without the use of aligned text-image pairs. This enables few-shot image classification with large language models (e.g., GPT-3) as well as linear classification of images based on BERT text features. To the best of our knowledge, our work is the first work that uses unaligned images for multimodal tasks by leveraging the power of pretrained language models. | ['Pieter Abbeel', 'Wilson Yan', 'Hao liu'] | 2023-02-02 | null | null | null | null | ['few-shot-image-classification'] | ['computer-vision'] | [ 2.63396859e-01 2.64185909e-02 -3.65645736e-02 -2.97404587e-01
-6.72737181e-01 -3.98031205e-01 8.31976771e-01 -8.57993066e-02
-6.26592457e-01 1.61627844e-01 1.33547947e-01 -2.07385674e-01
3.14336389e-01 -6.73767626e-01 -8.82961631e-01 -4.59095508e-01
4.12948459e-01 5.60333788e-01 2.36035213e-01 -2.68619806e-01
4.23318706e-02 7.59844705e-02 -1.71721017e+00 4.99219447e-01
2.69407213e-01 8.46708059e-01 5.98726153e-01 8.67465794e-01
-6.04324818e-01 1.04428971e+00 -3.15313131e-01 -4.48558390e-01
2.30128467e-01 -5.89137137e-01 -7.38658011e-01 3.83051038e-01
4.50798333e-01 -4.70717549e-01 -4.93051350e-01 8.87085497e-01
3.95661056e-01 3.04447711e-01 7.86647439e-01 -1.28249419e+00
-1.02753103e+00 4.43094403e-01 -4.14696693e-01 -1.77466527e-01
2.25417331e-01 2.03645721e-01 1.01727426e+00 -1.04811573e+00
7.27430999e-01 1.35750449e+00 3.83232385e-01 7.28123963e-01
-1.43391466e+00 -4.35699373e-01 -2.19840050e-01 2.77227730e-01
-1.38369906e+00 -5.41188121e-01 6.01158559e-01 -4.40874517e-01
1.17687392e+00 -4.51926235e-03 5.46305358e-01 1.24606276e+00
2.04410538e-01 7.25067437e-01 1.02448320e+00 -7.58788884e-01
3.18379372e-01 3.87599498e-01 -3.22115123e-01 9.48313177e-01
-3.18553418e-01 -1.49594620e-01 -5.77484846e-01 1.73830524e-01
5.80471933e-01 3.15472931e-01 -2.65275966e-02 -6.62272692e-01
-1.36828578e+00 8.95516157e-01 5.05253434e-01 4.34921771e-01
-8.76377970e-02 3.48955423e-01 3.77174944e-01 4.63484943e-01
1.03951246e-01 3.76676440e-01 -2.03815639e-01 9.40459874e-03
-9.68144655e-01 -2.13672623e-01 6.90761507e-01 8.58407736e-01
1.05562294e+00 9.01890248e-02 -8.24115053e-02 8.71606350e-01
3.20697635e-01 6.74703956e-01 9.36225653e-01 -8.28982472e-01
2.67729551e-01 3.69538844e-01 -3.10589254e-01 -6.79903984e-01
-4.54470962e-02 3.45488429e-01 -7.83720374e-01 2.99325585e-01
1.96747482e-01 1.41657904e-01 -1.28896666e+00 1.55197823e+00
-6.18350655e-02 2.09155381e-01 3.09380352e-01 7.34179020e-01
7.41516650e-01 7.77546346e-01 9.46027189e-02 1.57866091e-01
1.45364499e+00 -1.08675134e+00 -5.02782643e-01 -4.16191220e-01
6.12564862e-01 -7.30933487e-01 1.43229735e+00 7.40391314e-02
-8.70479763e-01 -6.10600471e-01 -1.03730941e+00 -4.45492536e-01
-8.08773339e-01 -2.74626110e-02 2.98215896e-01 5.41510880e-01
-1.14533389e+00 2.23569572e-01 -8.09502125e-01 -9.38696563e-01
3.62094343e-01 2.38095015e-01 -6.09546244e-01 -2.86911666e-01
-9.17965531e-01 8.15980852e-01 5.21451294e-01 -4.13515776e-01
-9.26430821e-01 -3.04776311e-01 -1.33038425e+00 1.96040466e-01
4.77989167e-01 -6.70332909e-01 1.16988683e+00 -1.30126822e+00
-1.58299339e+00 1.10980427e+00 -2.73512304e-01 -7.17825770e-01
2.09215581e-01 1.78966343e-01 -1.25289977e-01 3.62335145e-01
5.80076016e-02 1.30102682e+00 1.32330418e+00 -1.13425159e+00
-2.22672299e-01 -3.53161603e-01 3.40627059e-02 1.99340954e-01
-6.43443465e-01 -4.53303270e-02 -5.54490268e-01 -4.27817434e-01
-9.77170989e-02 -7.47321188e-01 -4.10107039e-02 2.44817853e-01
-3.03901229e-02 -1.61949322e-01 1.00141203e+00 -4.65323955e-01
6.69573963e-01 -2.20100808e+00 1.44019127e-01 -1.58445045e-01
2.98494756e-01 2.89755076e-01 -5.77536345e-01 6.91214383e-01
-1.10634195e-03 1.02680892e-01 -1.09487690e-01 -7.40983844e-01
1.30751029e-01 5.20747304e-01 -6.06185973e-01 3.03909391e-01
1.95348606e-01 1.45599902e+00 -7.53151298e-01 -6.89644098e-01
7.02425122e-01 3.57502401e-01 -5.15042961e-01 3.74956399e-01
-5.39554775e-01 7.45738670e-02 -6.80757910e-02 4.55502629e-01
1.92056865e-01 -3.90215695e-01 -4.13908847e-02 -1.73976749e-01
6.73561543e-02 -1.94149718e-01 -7.55010843e-01 1.96368909e+00
-6.46250486e-01 1.02071548e+00 -1.29808977e-01 -1.41289413e+00
7.37701118e-01 3.56903315e-01 3.61022711e-01 -8.52305233e-01
3.45877081e-01 -4.65130806e-02 -3.33927751e-01 -7.85259008e-01
5.25942981e-01 -4.03224438e-01 -2.81703472e-01 6.10533535e-01
5.48871517e-01 -4.42348689e-01 1.20650142e-01 4.03346926e-01
8.20401907e-01 6.32090867e-02 3.55002761e-01 4.33625281e-01
2.22384885e-01 2.27030553e-02 -1.13994099e-01 8.96317780e-01
-2.18474805e-01 8.69758189e-01 2.67966032e-01 -2.56240815e-01
-1.49218976e+00 -1.05062628e+00 1.30839273e-01 1.32602954e+00
-1.62731037e-02 -4.55934227e-01 -6.53096616e-01 -4.83155459e-01
-1.22135893e-01 6.90833330e-01 -7.54177094e-01 -1.59980685e-01
-8.15618709e-02 -1.86659902e-01 6.62615478e-01 4.73255575e-01
4.52066332e-01 -1.07387185e+00 -7.71709204e-01 -3.04557607e-02
-6.72224537e-02 -1.41908741e+00 -3.27379733e-01 3.31422895e-01
-3.74333262e-01 -6.85843766e-01 -8.63421500e-01 -1.11169398e+00
5.46590626e-01 5.28976560e-01 8.18336368e-01 -8.82942155e-02
-5.32955289e-01 8.99949372e-01 -5.13634801e-01 -3.56759429e-01
-4.23568487e-01 -2.93575615e-01 -4.38904539e-02 1.05979621e-01
4.89877731e-01 -3.54284644e-01 -2.44382590e-01 5.72771244e-02
-1.27468944e+00 1.81270748e-01 7.53772974e-01 1.18726838e+00
5.99966645e-01 -2.96799064e-01 1.27249837e-01 -4.94293034e-01
4.05873299e-01 -3.33633989e-01 -4.57391620e-01 5.75256944e-01
-2.38284260e-01 1.69984505e-01 7.29440391e-01 -6.27218246e-01
-6.93719864e-01 1.91388652e-01 2.13633832e-02 -9.66319561e-01
-3.69274557e-01 3.91873628e-01 6.88663796e-02 -7.35509321e-02
6.62931800e-01 5.50496042e-01 1.15646400e-01 -7.12388009e-02
8.36783469e-01 1.09502971e+00 6.33178353e-01 -2.16869429e-01
7.54966378e-01 7.20955729e-01 -2.16583580e-01 -1.19474006e+00
-6.13917291e-01 -6.95901632e-01 -7.88134336e-01 6.07291237e-02
1.30853069e+00 -8.81796420e-01 -5.02302706e-01 2.90401995e-01
-1.06099939e+00 -5.25905311e-01 -4.50642556e-01 4.62538511e-01
-8.47933233e-01 6.27767801e-01 -5.00287712e-01 -7.09958255e-01
-1.28773943e-01 -9.74358916e-01 1.28973126e+00 9.46226493e-02
7.36488914e-03 -9.69902635e-01 6.00413745e-03 4.36609924e-01
2.79753625e-01 -2.49597803e-01 8.03518534e-01 -6.06753707e-01
-5.22865295e-01 -9.90530923e-02 -3.35634530e-01 3.31355661e-01
-4.44901139e-02 -2.93829501e-01 -9.71896172e-01 -2.90153474e-01
5.04311100e-02 -1.01599145e+00 9.29234445e-01 1.71864316e-01
1.03704059e+00 -3.94250415e-02 -1.54411182e-01 6.41958416e-01
1.51469541e+00 -1.12627745e-01 6.03947043e-01 1.33596212e-01
6.31267130e-01 5.59831142e-01 3.22463423e-01 2.18234569e-01
4.09676075e-01 5.47726631e-01 3.17504853e-01 -9.91210118e-02
-3.04123670e-01 -5.33582389e-01 4.73835319e-01 6.90666735e-01
4.39290524e-01 -3.06171536e-01 -1.08698738e+00 6.18372023e-01
-1.82270658e+00 -1.06437206e+00 4.75197732e-01 1.96711743e+00
5.13504565e-01 -1.36771277e-01 -2.31173813e-01 -2.47066349e-01
4.28978652e-01 2.04533115e-01 -5.34796655e-01 -4.67430055e-01
-5.88358156e-02 2.57201403e-01 4.07371879e-01 3.43861997e-01
-1.05831051e+00 1.42862904e+00 5.78466845e+00 7.47660637e-01
-1.35169232e+00 2.84490913e-01 3.55114520e-01 -7.56330192e-02
-9.36810225e-02 9.96962637e-02 -4.53421652e-01 2.17734531e-01
1.01216960e+00 -2.28667006e-01 7.51808941e-01 8.19781899e-01
9.93872993e-03 -1.90808579e-01 -1.08468056e+00 1.29675055e+00
7.53729999e-01 -1.37616825e+00 2.86650360e-01 -8.78023431e-02
7.45610476e-01 2.94609755e-01 7.80143738e-02 5.54603815e-01
2.97213644e-01 -1.30234098e+00 6.80685580e-01 4.86622185e-01
9.62890625e-01 -1.77332655e-01 5.56962192e-01 5.52441180e-01
-1.08556986e+00 -1.67945623e-01 -6.92313969e-01 2.18961593e-02
1.31944075e-01 4.38103825e-02 -9.42198753e-01 2.07848921e-01
7.43264556e-01 7.90904045e-01 -5.38223684e-01 5.75852334e-01
-2.12029144e-02 3.16168129e-01 -1.61904722e-01 -1.39341414e-01
3.30411226e-01 -4.34330245e-03 2.21981660e-01 1.02765536e+00
2.93413162e-01 -7.17524290e-02 3.06661457e-01 9.64465678e-01
-2.44220704e-01 1.83836803e-01 -9.52150643e-01 -5.40292561e-01
1.74929455e-01 1.20028162e+00 -6.80918336e-01 -6.33072495e-01
-8.38713586e-01 1.48245490e+00 4.07154530e-01 4.72073495e-01
-6.02353811e-01 -4.04408455e-01 2.21733481e-01 -1.11217462e-01
5.98866105e-01 -3.99845868e-01 -2.88998280e-02 -1.50002217e+00
-1.31476611e-01 -7.66694427e-01 1.29251987e-01 -1.26030624e+00
-1.25253654e+00 5.19505739e-01 -1.18522525e-01 -1.24352753e+00
-3.63767385e-01 -7.54951239e-01 -4.26847428e-01 6.87140524e-01
-1.42747450e+00 -1.63456357e+00 -3.63788873e-01 9.47220743e-01
8.03774297e-01 -3.38936746e-01 9.41122591e-01 -1.42909080e-01
-2.05301166e-01 4.04989362e-01 2.70291090e-01 3.83974046e-01
9.05983627e-01 -1.14589417e+00 2.91126370e-01 6.73795521e-01
6.93499625e-01 4.38123703e-01 4.87065822e-01 -3.58059049e-01
-1.51304293e+00 -1.02814031e+00 8.07538331e-01 -4.65709329e-01
9.62631524e-01 -7.79825866e-01 -8.71558905e-01 8.60769928e-01
4.49726790e-01 1.93109229e-01 6.56406641e-01 -3.21003288e-01
-5.89717925e-01 4.52970415e-02 -7.62612402e-01 7.76605487e-01
6.58408403e-01 -1.10740399e+00 -9.11690593e-01 3.33770394e-01
7.17963278e-01 -1.65127531e-01 -6.13388121e-01 -1.04022019e-01
5.16218007e-01 -8.17357421e-01 9.53822553e-01 -5.95690012e-01
5.08488536e-01 -8.11937824e-02 -4.37101245e-01 -1.01064408e+00
1.26962259e-01 -2.59948790e-01 4.83540371e-02 1.05650759e+00
5.64730801e-02 -3.95594776e-01 6.38875186e-01 3.76464844e-01
1.63732886e-01 -4.08431679e-01 -8.78074586e-01 -7.88026035e-01
1.25356689e-01 -5.36727846e-01 1.78188249e-01 8.64152968e-01
9.88566726e-02 5.58041930e-01 -5.84988654e-01 -1.45588726e-01
4.43200946e-01 2.72427276e-02 1.00778413e+00 -8.03179264e-01
-4.68870163e-01 -3.37679356e-01 -6.64790094e-01 -1.12873685e+00
4.45619613e-01 -1.06431437e+00 3.45987648e-01 -1.47438860e+00
3.72436255e-01 -4.21322025e-02 -4.03409302e-02 6.62743866e-01
7.70079345e-02 5.73655427e-01 4.10082400e-01 3.32197487e-01
-8.35491955e-01 7.65728533e-01 9.56760049e-01 -4.28358167e-01
8.81438702e-02 -5.89510024e-01 -2.86859334e-01 6.12528443e-01
6.95760548e-01 -3.32621664e-01 -4.73861009e-01 -5.12575030e-01
6.45512044e-02 3.54315564e-02 6.66191399e-01 -9.81294096e-01
4.37909901e-01 -1.01533540e-01 4.40894186e-01 -3.87344211e-01
5.99979341e-01 -8.71031582e-01 -3.16071540e-01 3.37045461e-01
-4.77945536e-01 -1.07090376e-01 8.70553255e-02 6.83689713e-01
-3.51988882e-01 -3.18243802e-01 7.06849337e-01 -3.39002609e-01
-1.16019046e+00 2.04142541e-01 -5.56509435e-01 -9.64312628e-02
1.18641794e+00 -2.46180370e-01 -1.68260738e-01 -5.78528106e-01
-5.77207267e-01 2.52201289e-01 8.24507713e-01 6.40198290e-01
7.86427796e-01 -1.09431601e+00 -3.61810833e-01 4.22307223e-01
5.88781655e-01 -3.37876529e-01 2.45632991e-01 7.03302503e-01
-3.19882065e-01 6.17725134e-01 -3.14194709e-01 -8.50763857e-01
-1.30039775e+00 9.53497052e-01 3.57319921e-01 -6.90934807e-02
-6.86778605e-01 6.09409153e-01 2.66350001e-01 -5.14507353e-01
2.34427691e-01 -1.66679367e-01 -1.02188522e-02 -3.39901485e-02
5.76851130e-01 -2.93007702e-01 -1.82880253e-01 -7.68306732e-01
-1.45121664e-01 7.71919250e-01 -1.12684369e-01 -5.04602194e-01
1.05352795e+00 -3.87603581e-01 -9.20195598e-03 6.85437202e-01
1.47074580e+00 -2.93610990e-01 -1.20271456e+00 -4.92591649e-01
-1.52177021e-01 -2.95852810e-01 4.45694327e-02 -4.28679675e-01
-6.34480953e-01 1.46815372e+00 6.88105822e-01 -2.18725670e-02
9.93106544e-01 4.00312185e-01 8.68884683e-01 7.56425738e-01
1.96496606e-01 -1.10418832e+00 7.48700976e-01 5.97983062e-01
5.99524140e-01 -1.57571352e+00 -4.13606584e-01 1.71497747e-01
-9.08807278e-01 1.18528664e+00 4.78476107e-01 4.44046827e-03
4.47633833e-01 9.55854058e-02 2.73701131e-01 -1.16049074e-01
-8.40654671e-01 -6.39014781e-01 2.21935838e-01 6.54791117e-01
1.45550057e-01 -1.27564549e-01 1.42582253e-01 1.80859566e-01
-7.57695138e-02 1.13688782e-01 4.87235487e-01 9.13476348e-01
-5.65460503e-01 -1.07207954e+00 -3.13979447e-01 3.63604367e-01
-1.24469325e-01 -3.36161762e-01 -3.91487896e-01 5.76991498e-01
8.64867568e-02 8.86092603e-01 3.80898386e-01 -3.88307810e-01
-8.40703323e-02 3.48418206e-01 5.26776910e-01 -9.29154158e-01
-1.66764379e-01 1.50127513e-02 -5.07867754e-01 -4.67035770e-01
-4.69226301e-01 -3.92206848e-01 -1.20654964e+00 -3.48879471e-02
-1.04150675e-01 -1.23082750e-01 7.51182020e-01 1.09393919e+00
2.53542393e-01 2.84323126e-01 3.69795233e-01 -9.44262266e-01
-4.96087939e-01 -7.78328001e-01 -5.04795969e-01 6.74409509e-01
4.23278153e-01 -3.80810440e-01 -2.08483249e-01 4.14633512e-01] | [10.496408462524414, 1.8802789449691772] |
805aa9a8-5bd2-4702-bab0-e6bd368a2e26 | learning-cross-lingual-ir-from-an-english-1 | null | null | https://openreview.net/forum?id=OXqk4rKn9LT | https://openreview.net/pdf?id=OXqk4rKn9LT | Learning Cross-Lingual IR from an English Retriever | We present a new cross-lingual information retrieval (CLIR) system trained using multi-stage knowledge distillation (KD). The teacher relies on a highly effective but expensive two-stage process consisting of query translation and monolingual IR, while the student executes a single CLIR step. We teach the student powerful multilingual encoding as well as CLIR by optimizing two corresponding KD objectives. Learning useful non-English representations from an English-only retriever is accomplished through a cross-lingual token alignment algorithm that relies on the representation capabilities of the underlying multilingual language model. In both in-domain and zero-shot evaluation, the proposed method demonstrates far superior accuracy over direct fine-tuning with labeled CLIR data. One of our systems is also the current best single-model system on the XOR-TyDi leaderboard. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['cross-lingual-information-retrieval'] | ['natural-language-processing'] | [-9.52550843e-02 -7.62368664e-02 -7.07552791e-01 -2.33021602e-01
-2.06815600e+00 -1.03928864e+00 6.58079028e-01 3.08295786e-01
-1.13008237e+00 8.11291814e-01 -3.29760052e-02 -5.61717868e-01
-1.28348231e-01 -4.17802334e-01 -7.17909694e-01 -3.67737204e-01
4.64387506e-01 1.26606560e+00 -7.58036552e-03 -6.76775157e-01
-1.41953602e-01 4.10609961e-01 -1.15682650e+00 4.83969271e-01
8.75697792e-01 7.24548399e-01 2.88139313e-01 8.03390920e-01
-3.67206603e-01 1.06313372e+00 -4.66781169e-01 -6.51077032e-01
1.29127562e-01 -1.05429791e-01 -1.07685590e+00 -7.76679933e-01
5.34765422e-01 -2.63282418e-01 -2.86241323e-01 8.10266316e-01
8.74493957e-01 3.20273101e-01 7.48941183e-01 -6.87820435e-01
-1.16952991e+00 1.12708664e+00 -3.47526252e-01 1.90247834e-01
4.40770566e-01 -3.56924266e-01 1.34160304e+00 -1.16309059e+00
8.40949178e-01 1.25948298e+00 3.06398213e-01 5.16105473e-01
-1.24598706e+00 -8.27208221e-01 -2.19653666e-01 3.81957829e-01
-1.76010919e+00 -5.45591116e-01 3.18095684e-01 -2.02770337e-01
1.34561980e+00 -2.45617200e-02 4.70180027e-02 8.86556089e-01
-4.93873246e-02 1.28155971e+00 9.67989564e-01 -9.64496911e-01
-2.30468109e-01 6.32916212e-01 3.59037578e-01 6.92082524e-01
-2.00793132e-01 4.91778068e-02 -6.77374601e-01 -2.50801351e-02
2.95981824e-01 -3.68419319e-01 2.72337254e-02 -3.03540111e-01
-1.19356573e+00 7.73712516e-01 4.64662425e-02 3.77717853e-01
-1.25217721e-01 -1.44411894e-02 4.31904495e-01 1.04940534e+00
5.51690161e-01 8.10091496e-01 -9.07782733e-01 -2.88033895e-02
-9.85800028e-01 -6.30344823e-03 7.52630234e-01 1.14680469e+00
9.44215178e-01 -1.88177586e-01 -3.92491668e-01 1.15221763e+00
1.38792768e-01 8.60597670e-01 6.89560115e-01 -6.77680194e-01
4.38145608e-01 2.14437813e-01 4.20083180e-02 -3.77558582e-02
9.88344774e-02 -1.86164275e-01 -3.23274344e-01 -1.04919620e-01
1.66274667e-01 -8.41433108e-02 -7.33795881e-01 1.61050057e+00
5.97495697e-02 -1.58961430e-01 7.07366168e-01 5.12747943e-01
8.19439650e-01 6.39178693e-01 2.74103343e-01 -2.40548968e-01
1.43004656e+00 -1.11719823e+00 -6.97568238e-01 -1.08490668e-01
1.03416431e+00 -1.23968458e+00 9.71214414e-01 4.13238525e-01
-1.48508930e+00 -5.13992250e-01 -8.95528257e-01 -8.86133850e-01
-8.28988075e-01 4.52133983e-01 5.36443055e-01 1.25100061e-01
-1.38545775e+00 4.88207974e-02 -4.75465208e-01 -3.28508705e-01
-4.12274361e-01 2.73132026e-01 -4.93506581e-01 -7.15362549e-01
-1.61310351e+00 1.44925082e+00 4.68498349e-01 -2.87130266e-01
-1.08872318e+00 -7.85769165e-01 -8.99759650e-01 -5.46486117e-02
3.21351111e-01 -4.17371690e-01 1.62180042e+00 -7.42336690e-01
-1.73169053e+00 1.23691618e+00 -1.25055420e-04 -1.04376063e-01
2.94183880e-01 -5.91278136e-01 -3.21758121e-01 6.43256456e-02
1.95587561e-01 6.72307253e-01 4.51347470e-01 -7.81147122e-01
-4.13531393e-01 -4.24376428e-01 -5.61073869e-02 7.33188450e-01
-2.21281067e-01 5.60969055e-01 -8.55118573e-01 -5.44117808e-01
-1.86873570e-01 -8.21165144e-01 1.34237230e-01 -5.27148426e-01
-1.61445051e-01 -8.68431509e-01 2.70210981e-01 -9.31394160e-01
1.33623338e+00 -1.87254786e+00 2.76341528e-01 2.31765062e-02
-3.15664113e-01 4.18138206e-01 -5.82433343e-01 6.86538041e-01
-1.46386802e-01 4.46995981e-02 5.23888767e-01 -4.81520087e-01
6.33023158e-02 9.14218649e-02 -4.76415902e-01 1.13107227e-01
2.41410255e-01 1.22891116e+00 -1.11269093e+00 -6.76042676e-01
-1.77298948e-01 4.98481750e-01 -2.87542880e-01 6.36936784e-01
-2.77522415e-01 1.19530492e-01 -6.63887203e-01 5.79740226e-01
4.81651947e-02 -2.81059891e-02 3.91099453e-01 6.56212717e-02
-2.31122643e-01 6.76975369e-01 -9.75279152e-01 2.51814556e+00
-8.58894408e-01 4.42160010e-01 5.04770763e-02 -9.43334401e-01
8.28790784e-01 6.11842990e-01 2.01534912e-01 -1.13124883e+00
-2.60985225e-01 5.11168480e-01 -4.94133085e-01 -3.13847601e-01
9.55124438e-01 1.04975916e-01 -3.97699893e-01 9.58279788e-01
5.13145387e-01 -8.35871026e-02 1.53104529e-01 6.36192262e-01
9.05540049e-01 4.73552585e-01 2.22864479e-01 -4.17733788e-01
3.99893701e-01 3.20037037e-01 6.83051050e-02 9.52115238e-01
2.31517509e-01 -9.95289907e-03 -2.92359382e-01 -2.82340974e-01
-9.13911521e-01 -1.17871094e+00 -2.42156312e-01 2.11052299e+00
-1.61608443e-01 -3.55632931e-01 -4.09324080e-01 -5.75413048e-01
-3.22639830e-02 9.37271118e-01 -1.64817795e-01 -2.29055241e-01
-6.17807031e-01 -2.87751943e-01 9.26073372e-01 4.05119836e-01
1.21163487e-01 -9.54878628e-01 7.00406730e-02 3.41343045e-01
-2.11800337e-01 -1.17302275e+00 -7.16087401e-01 6.30151749e-01
-4.18167949e-01 -6.23172879e-01 -7.27967381e-01 -1.19834507e+00
2.97360718e-01 3.11114311e-01 1.72842395e+00 -1.24523729e-01
-5.65375984e-01 5.92837453e-01 -1.54401451e-01 -2.16189712e-01
-5.14371574e-01 4.44643021e-01 1.95901185e-01 -6.75697565e-01
9.56116319e-01 -4.49923761e-02 5.43558644e-03 3.62839177e-02
-6.69596136e-01 -1.00753624e-02 1.02176213e+00 8.57498765e-01
8.50804389e-01 -2.76078224e-01 4.85389411e-01 -8.61858070e-01
8.00137103e-01 -3.55151564e-01 -7.71352053e-01 1.11882091e+00
-8.16032469e-01 7.02146351e-01 4.07100737e-01 -5.20739555e-01
-1.24106669e+00 -2.02184200e-01 2.29599997e-01 -5.51017225e-01
6.11029938e-02 6.54483557e-01 1.27721861e-01 1.23396978e-01
8.15584600e-01 3.98476928e-01 -5.24720430e-01 -7.85712183e-01
1.00212729e+00 8.20928037e-01 6.26836419e-01 -1.28447115e+00
6.51868939e-01 -4.55101103e-01 -7.19700515e-01 -6.48039877e-01
-1.14124179e+00 -8.78967643e-01 -8.11845183e-01 2.75833488e-01
9.60212469e-01 -1.67410719e+00 -6.65412068e-01 2.00493529e-01
-1.24889994e+00 -5.19040465e-01 -3.56411308e-01 6.05872452e-01
-3.48909646e-01 -2.14620531e-01 -9.95132685e-01 -6.05715990e-01
-5.94268739e-01 -1.20371163e+00 1.21968770e+00 -2.81552318e-02
1.10221304e-01 -1.07878888e+00 7.30366230e-01 6.11148238e-01
5.10199428e-01 -8.35560918e-01 1.18429339e+00 -1.17603362e+00
-6.57967269e-01 -1.33755490e-01 -2.11731553e-01 3.64371896e-01
-4.20711845e-01 -1.65331259e-01 -1.10739613e+00 -4.52739865e-01
-7.26756990e-01 -1.49885678e+00 8.00870121e-01 -2.34944433e-01
5.26582479e-01 -7.73930177e-02 -1.48661643e-01 4.49870765e-01
1.46246231e+00 -7.31796548e-02 5.50343469e-02 3.64551067e-01
6.94851458e-01 4.66844887e-01 6.59882486e-01 -2.80251950e-01
8.50259006e-01 9.24475253e-01 -4.23040837e-01 -2.89266873e-02
-4.01860952e-01 -4.36968267e-01 8.51319671e-01 1.45466650e+00
2.50887334e-01 1.06002949e-01 -1.01786602e+00 6.33237720e-01
-1.68906188e+00 -7.83181190e-01 5.51482618e-01 2.22815084e+00
1.65902376e+00 -4.28819805e-01 -3.75583857e-01 -7.24306703e-01
2.01688573e-01 -2.71763504e-01 -4.91595656e-01 -5.78660905e-01
-1.91693261e-01 7.87970185e-01 4.73208696e-01 9.27262485e-01
-8.27666342e-01 1.66230226e+00 6.34075022e+00 1.18225586e+00
-9.60391998e-01 3.70106697e-01 3.34042639e-01 2.48219315e-02
-4.19837654e-01 -2.46675551e-01 -1.37578726e+00 -3.47755075e-01
1.28730857e+00 -4.37772691e-01 8.25686812e-01 9.60558951e-01
-4.57645088e-01 8.40758532e-02 -1.28846240e+00 8.53266478e-01
3.39591146e-01 -9.11104441e-01 9.93878320e-02 -2.49765053e-01
9.52404559e-01 7.19933748e-01 3.68379951e-02 1.06105971e+00
1.19465303e+00 -9.14331615e-01 3.88719350e-01 4.67477441e-01
1.22400963e+00 -8.56877029e-01 4.81769234e-01 5.03314912e-01
-1.13328648e+00 2.74817169e-01 -5.08001149e-01 5.10982156e-01
-5.16850293e-01 -8.72344302e-04 -1.00663245e+00 6.52737617e-01
5.90051234e-01 5.13436019e-01 -5.01253963e-01 2.39852548e-01
-5.72795391e-01 2.45240748e-01 -2.69257843e-01 2.28582904e-01
2.03710720e-01 -2.92977512e-01 1.65509388e-01 1.56886840e+00
1.66410118e-01 -4.75919060e-02 6.35967970e-01 5.32617331e-01
-5.61145723e-01 5.06103575e-01 -8.58921945e-01 -2.40415156e-01
5.51161230e-01 1.36046255e+00 3.08294326e-01 -7.15272486e-01
-4.90291923e-01 1.12149024e+00 1.00670409e+00 5.42240858e-01
-5.86718023e-02 -4.17214811e-01 7.10326016e-01 -4.56571072e-01
7.60333554e-04 -2.32274979e-01 3.17980081e-01 -1.67743111e+00
-2.75823414e-01 -1.18031108e+00 7.64743745e-01 -8.73889148e-01
-1.33303130e+00 6.56540096e-01 2.03717407e-02 -7.36350060e-01
-9.59559023e-01 -4.30240661e-01 -1.01388350e-01 1.29311395e+00
-2.27227807e+00 -1.46148002e+00 3.34510922e-01 1.15224719e+00
6.21906638e-01 -4.81556684e-01 1.56057382e+00 5.66965759e-01
-4.69373345e-01 1.03612494e+00 5.39168894e-01 4.03450936e-01
1.42485535e+00 -1.40504420e+00 3.98703516e-02 5.46287358e-01
5.79014480e-01 9.85203981e-01 1.52400658e-01 -3.35934699e-01
-1.86045182e+00 -9.51333225e-01 1.61584616e+00 -7.19164670e-01
1.08442485e+00 -2.73416519e-01 -8.03868353e-01 1.01903510e+00
6.64183676e-01 -2.47413680e-01 8.17300200e-01 4.95632768e-01
-8.42729330e-01 -1.39009893e-01 -6.96544111e-01 3.15359145e-01
3.72848898e-01 -1.35407782e+00 -9.28725064e-01 6.57632470e-01
9.40725029e-01 -3.22138697e-01 -1.28450775e+00 6.46040738e-02
5.98934591e-01 -8.95228295e-04 1.17225671e+00 -1.14326465e+00
1.66074857e-01 -4.34087105e-02 -2.99503595e-01 -1.44153881e+00
-1.22874759e-01 -5.09174705e-01 1.07435547e-01 1.06020272e+00
6.52465463e-01 -1.61007077e-01 1.39882952e-01 5.15581250e-01
-5.31724170e-02 -2.56504208e-01 -5.52104115e-01 -7.39750564e-01
5.31195998e-01 -5.36000192e-01 -4.81457589e-03 1.33282435e+00
2.90488273e-01 9.78278518e-01 -1.38383657e-01 1.88685022e-02
7.40585387e-01 1.14876889e-01 7.31789172e-01 -9.87653613e-01
-5.08774757e-01 -2.17688352e-01 2.97193170e-01 -1.06782746e+00
6.11266375e-01 -1.44450486e+00 1.90401912e-01 -1.13366401e+00
4.05092031e-01 -6.40783072e-01 -8.41127992e-01 8.39471459e-01
-2.90240735e-01 1.35442078e-01 -9.71490368e-02 3.13738912e-01
-1.07663846e+00 1.74524948e-01 8.93357813e-01 -1.91617951e-01
-4.41945568e-02 -4.41367447e-01 -5.35989344e-01 2.86550015e-01
2.54847646e-01 -5.63816249e-01 -4.39864308e-01 -9.70610499e-01
4.51019078e-01 3.99191380e-01 -3.18702668e-01 -3.96110833e-01
6.27488673e-01 3.12865563e-02 2.12666601e-01 -3.70563716e-01
2.32221678e-01 -6.19458139e-01 -3.88632804e-01 -8.82963017e-02
-9.71517265e-01 4.75009412e-01 2.67737478e-01 1.97132081e-02
-4.24425393e-01 -9.47303101e-02 6.59488261e-01 -3.78620476e-01
-8.59239697e-01 2.32609436e-01 -7.73940831e-02 4.21903193e-01
5.94848275e-01 8.66772771e-01 -5.23113132e-01 -2.42782190e-01
-6.72550380e-01 4.23149675e-01 1.70057103e-01 5.83493888e-01
2.47794300e-01 -1.30868900e+00 -1.01702249e+00 2.69571185e-01
5.39223909e-01 -1.79133505e-01 -2.23546833e-01 5.42414129e-01
-1.52284056e-01 9.69306707e-01 1.25841469e-01 -3.72717440e-01
-1.07470334e+00 5.74583590e-01 4.50733379e-02 -1.12791705e+00
-8.11613798e-02 1.09041619e+00 1.35030271e-02 -9.52461302e-01
4.02306587e-01 1.58379763e-01 -1.43007934e-01 2.98739731e-01
7.90116787e-01 -5.31603768e-03 4.21872824e-01 -4.59917665e-01
-7.88574964e-02 6.25447571e-01 -7.21928537e-01 -4.80408072e-01
1.00846744e+00 -1.27154037e-01 -1.22206010e-01 6.24194682e-01
1.41730499e+00 -9.60961580e-02 -3.56890947e-01 -9.29869890e-01
3.70817423e-01 1.36374265e-01 4.25888598e-01 -1.29396260e+00
-5.41616797e-01 1.08837068e+00 5.06954551e-01 -5.36312103e-01
9.02344704e-01 1.21893294e-01 7.14726508e-01 1.37654543e+00
6.05692148e-01 -1.25278938e+00 3.32664996e-02 9.08344030e-01
5.42620897e-01 -1.53098083e+00 -1.35368064e-01 3.61094236e-01
-6.82329416e-01 9.80786264e-01 4.08980131e-01 1.78939775e-01
3.89184147e-01 2.15890139e-01 6.06950819e-01 -2.19061337e-02
-1.28970146e+00 -3.48888099e-01 6.84382856e-01 6.58815429e-02
8.73657048e-01 2.05302373e-01 -4.77923453e-02 4.07415926e-01
1.52371852e-02 -2.68537756e-02 -8.39248151e-02 9.23283160e-01
-4.21326160e-01 -1.58532000e+00 1.83309950e-02 -2.29756653e-01
-7.50347376e-01 -8.57473373e-01 -2.98601061e-01 8.50160539e-01
-4.47073877e-01 6.46834731e-01 -4.55685034e-02 -1.53389245e-01
2.64759749e-01 7.81443357e-01 4.27499652e-01 -8.54106426e-01
-9.77492154e-01 1.47830740e-01 2.18821749e-01 -7.30672896e-01
-1.71125859e-01 -3.29497367e-01 -9.20635998e-01 -2.44906135e-02
-3.49046886e-01 6.21267378e-01 9.91376579e-01 1.02056110e+00
2.21033841e-01 2.19759941e-01 4.83962029e-01 -3.11219364e-01
-9.31018651e-01 -1.09211087e+00 -3.64702761e-01 1.73703775e-01
1.21938847e-01 -2.20694914e-01 -6.47027418e-02 4.74577993e-02] | [11.352873802185059, 9.787575721740723] |
7375a625-8d17-467f-a781-ada04e0e8543 | an-overview-of-hierarchical-task-network | 1403.7426 | null | http://arxiv.org/abs/1403.7426v1 | http://arxiv.org/pdf/1403.7426v1.pdf | An Overview of Hierarchical Task Network Planning | Hierarchies are the most common structure used to understand the world
better. In galaxies, for instance, multiple-star systems are organised in a
hierarchical system. Then, governmental and company organisations are
structured using a hierarchy, while the Internet, which is used on a daily
basis, has a space of domain names arranged hierarchically. Since Artificial
Intelligence (AI) planning portrays information about the world and reasons to
solve some of world's problems, Hierarchical Task Network (HTN) planning has
been introduced almost 40 years ago to represent and deal with hierarchies. Its
requirement for rich domain knowledge to characterise the world enables HTN
planning to be very useful, but also to perform well. However, the history of
almost 40 years obfuscates the current understanding of HTN planning in terms
of accomplishments, planning models, similarities and differences among
hierarchical planners, and its current and objective image. On top of these
issues, attention attracts the ability of hierarchical planning to truly cope
with the requirements of applications from the real world. We propose a
framework-based approach to remedy this situation. First, we provide a basis
for defining different formal models of hierarchical planning, and define two
models that comprise a large portion of HTN planners. Second, we provide a set
of concepts that helps to interpret HTN planners from the aspect of their
search space. Then, we analyse and compare the planners based on a variety of
properties organised in five segments, namely domain authoring, expressiveness,
competence, performance and applicability. Furthermore, we select Web service
composition as a real-world and current application, and classify and compare
the approaches that employ HTN planning to solve the problem of service
composition. Finally, we conclude with our findings and present directions for
future work. | ['Ilche Georgievski', 'Marco Aiello'] | 2014-03-28 | null | null | null | null | ['service-composition'] | ['miscellaneous'] | [ 8.15126970e-02 7.06838965e-01 -2.23908007e-01 -1.90206707e-01
-1.13865912e-01 -7.50784814e-01 1.10906637e+00 4.38142605e-02
-6.58501610e-02 5.39372027e-01 6.37986004e-01 -3.58334810e-01
-8.81944478e-01 -1.07465398e+00 -2.87562166e-03 -3.87187928e-01
-1.04870059e-01 1.12761343e+00 7.23057151e-01 -6.47043467e-01
2.96155453e-01 6.03602588e-01 -1.96840572e+00 3.92171830e-01
8.07013690e-01 9.42212045e-01 3.12381566e-01 1.14795521e-01
-4.98488307e-01 8.95406604e-01 -6.16160393e-01 -3.88105363e-01
3.01208615e-01 -3.69790286e-01 -1.57325566e+00 5.88667512e-01
-5.04801512e-01 1.61607683e-01 -8.99484660e-03 8.25411439e-01
-4.82817367e-02 2.03870818e-01 5.28204799e-01 -1.66382551e+00
-4.22690101e-02 8.23253155e-01 2.30352923e-01 -4.20161486e-01
9.24518466e-01 -2.51889136e-02 1.29083383e+00 -9.21664387e-02
9.41144347e-01 1.22852480e+00 2.02616602e-01 6.35210335e-01
-9.24458385e-01 -1.73145935e-01 2.53215909e-01 4.57423180e-01
-1.21598411e+00 -2.76254743e-01 4.01097387e-01 -4.58977968e-01
1.24286759e+00 7.57795751e-01 9.11937118e-01 6.38784230e-01
6.64660335e-02 4.57715452e-01 1.02455866e+00 -5.98352969e-01
2.76785076e-01 3.58791091e-02 -9.01523009e-02 2.64136791e-01
2.20273212e-01 -1.37327418e-01 -4.17529464e-01 -4.29140359e-01
6.78200066e-01 -3.38957697e-01 -1.00901231e-01 -6.85611904e-01
-1.06790113e+00 8.15865815e-01 1.06052153e-01 8.62125874e-01
-3.09892446e-01 -3.61821443e-01 3.66867870e-01 9.39343870e-02
-2.87871510e-01 8.31146479e-01 -4.03542697e-01 -3.59092146e-01
-4.89493787e-01 3.39563519e-01 1.54214299e+00 9.75959063e-01
6.42264128e-01 -2.33256400e-01 5.71985424e-01 5.19228876e-01
5.29125988e-01 -1.27479389e-01 5.26469171e-01 -1.46538842e+00
-1.04494527e-01 1.16641009e+00 2.63874441e-01 -9.57276523e-01
-6.60463750e-01 -1.95747226e-01 -3.81872773e-01 1.04015872e-01
4.65761811e-01 5.19350767e-01 -4.94363904e-01 1.36411262e+00
3.15413982e-01 -4.70783979e-01 1.40327051e-01 8.74117970e-01
6.43024027e-01 6.01881981e-01 -1.37842178e-01 -4.38728005e-01
1.53448212e+00 -1.02953756e+00 -5.35641253e-01 -2.86175627e-02
5.43781042e-01 -4.36801076e-01 1.01944411e+00 6.37624264e-01
-1.17457879e+00 1.02380991e-01 -7.22370803e-01 1.17411703e-01
-6.37129784e-01 -7.33572423e-01 9.13043857e-01 6.41203463e-01
-1.17134309e+00 4.02488559e-01 -5.07846713e-01 -1.00173938e+00
-2.51788318e-01 5.79865932e-01 -1.35345608e-01 1.78482793e-02
-1.37286699e+00 1.04532325e+00 8.73652041e-01 -2.74271399e-01
-1.00848103e+00 1.31570473e-01 -6.94917381e-01 2.12338388e-01
8.54275942e-01 -8.54838729e-01 1.43529403e+00 -9.47761953e-01
-1.50593495e+00 5.36727726e-01 1.30807832e-01 -4.54832107e-01
3.81212413e-01 6.94111705e-01 -7.10777402e-01 4.40612659e-02
3.10568869e-01 3.48809808e-01 6.30028248e-02 -1.41806161e+00
-1.28170395e+00 -1.31606728e-01 9.51511264e-01 4.35311675e-01
-2.74505734e-01 3.95872056e-01 -5.02298892e-01 8.98354352e-02
4.25657868e-01 -5.90806484e-01 -6.09720767e-01 -7.75398314e-01
-1.01173528e-01 -7.17941821e-01 3.97032082e-01 -3.10709327e-01
1.56586432e+00 -1.78229034e+00 4.23964888e-01 6.15017354e-01
4.03914750e-01 -3.63374054e-01 8.09955895e-02 1.14114356e+00
2.01798335e-01 5.00546336e-01 -2.61327565e-01 5.04383326e-01
5.40918529e-01 8.95519793e-01 8.59357268e-02 -6.58821734e-03
-3.12202007e-01 3.72637063e-01 -8.48626673e-01 -6.38187289e-01
3.73961474e-03 1.13372400e-01 -6.17814064e-01 -8.09854046e-02
-5.99455476e-01 4.23801124e-01 -6.32070959e-01 6.85156822e-01
-1.04234613e-01 -2.76033252e-01 9.98437107e-01 1.56695217e-01
-5.03139198e-01 5.35147965e-01 -1.32805097e+00 1.56751955e+00
-3.48182946e-01 6.78973943e-02 1.58183739e-01 -9.26775992e-01
7.37377703e-01 8.13791931e-01 8.47681880e-01 -7.24771619e-01
-6.37438819e-02 5.42571783e-01 3.65471750e-01 -4.95930552e-01
3.20638806e-01 -1.32214755e-01 -4.44597930e-01 4.68324751e-01
-2.66387373e-01 -4.79484320e-01 8.39585423e-01 2.09828645e-01
1.40643370e+00 2.03573316e-01 7.53977716e-01 -3.99041086e-01
8.48884881e-01 4.39977914e-01 8.14010739e-01 4.87021446e-01
-7.22022057e-02 1.67499512e-01 8.92686963e-01 -9.51134026e-01
-1.06147850e+00 -5.85211992e-01 4.06210460e-02 1.17717087e+00
2.31090114e-01 -8.69728863e-01 -8.14902782e-01 -8.39427769e-01
-4.85053569e-01 8.82935405e-01 -2.17266113e-01 4.88696277e-01
-6.87113225e-01 -3.51716250e-01 2.05295250e-01 -8.95601660e-02
3.21443647e-01 -1.38923693e+00 -8.83814454e-01 3.90335560e-01
-5.31556129e-01 -1.26865113e+00 1.83731884e-01 -2.46079341e-02
-4.66807991e-01 -1.21290588e+00 -2.91599315e-02 -5.91371775e-01
3.69781911e-01 3.92467231e-02 1.53223407e+00 3.84663463e-01
1.93573192e-01 6.01664960e-01 -7.17536628e-01 -2.80153424e-01
-5.73018312e-01 2.88627088e-01 1.34972215e-01 -2.04024419e-01
1.43024459e-01 -8.68979931e-01 -2.29033992e-01 8.24536860e-01
-1.19145107e+00 2.62162507e-01 4.41530377e-01 3.39557797e-01
6.27160668e-02 7.75191784e-01 1.69890538e-01 -8.66121411e-01
8.21065784e-01 -3.26766551e-01 -6.16951108e-01 3.47695500e-01
-8.05785477e-01 -6.90650642e-02 5.07048130e-01 1.19217046e-01
-9.68606234e-01 -1.59145921e-01 -1.29768953e-01 6.05697513e-01
-4.85244364e-01 7.78217077e-01 -6.75272524e-01 5.54687344e-02
6.32148206e-01 1.42026782e-01 -2.32852221e-01 -3.17487538e-01
2.21350163e-01 6.09211564e-01 7.50184134e-02 -1.02705061e+00
5.65987706e-01 4.80492085e-01 -1.96580533e-02 -7.65954137e-01
-4.92415577e-01 -6.56783402e-01 -5.11663616e-01 -4.58606809e-01
7.10886180e-01 -5.38231134e-02 -1.19435668e+00 -2.12950513e-01
-1.20895731e+00 -2.78662413e-01 -4.99650925e-01 -5.70466332e-02
-1.04166079e+00 3.92993063e-01 -3.30438823e-01 -9.45911467e-01
1.32584319e-01 -1.35376418e+00 6.65387571e-01 7.87713006e-02
-4.74790126e-01 -8.91042590e-01 -2.42070109e-01 5.36392391e-01
4.68888998e-01 2.84880131e-01 1.33640766e+00 -9.64047313e-01
-6.48677230e-01 1.19277194e-01 -1.46653697e-01 -4.36581634e-02
5.46919815e-02 -3.17526370e-01 -2.64996022e-01 8.26128870e-02
1.51890844e-01 2.00936928e-01 -2.05191270e-01 -6.39218539e-02
7.61157990e-01 -6.62179232e-01 -3.28692645e-01 -8.46824721e-02
1.46768367e+00 8.78306150e-01 8.44696462e-01 1.08151627e+00
2.07574368e-01 1.36191380e+00 5.52206874e-01 5.24665534e-01
6.42302275e-01 1.00875056e+00 8.43534470e-01 2.78481007e-01
1.55814916e-01 4.71857339e-02 -2.87123560e-03 9.52345550e-01
-8.04314196e-01 -4.46793204e-03 -1.56749356e+00 4.18533802e-01
-2.10303640e+00 -8.51671219e-01 -1.07645236e-01 1.83877969e+00
6.60816193e-01 2.78231144e-01 5.01564503e-01 3.05185914e-01
5.22486448e-01 -1.42712116e-01 -1.97424501e-01 -4.60090160e-01
-4.04555611e-02 -2.21991941e-01 2.10141033e-01 5.21374285e-01
-5.85709214e-01 9.91212368e-01 6.46251869e+00 7.21838832e-01
-4.32124108e-01 1.77413315e-01 5.14110588e-02 5.21231234e-01
-5.35254180e-01 5.32249987e-01 -6.25427604e-01 1.70775980e-01
1.07062042e+00 -6.40843749e-01 8.71201277e-01 7.70646036e-01
2.50511080e-01 -1.48233902e-02 -1.08480382e+00 4.89747882e-01
-1.72757655e-01 -1.28389704e+00 2.60546118e-01 5.81046224e-01
4.58210289e-01 -2.50223637e-01 -4.80416000e-01 2.03288659e-01
6.86831236e-01 -1.02844346e+00 8.77364874e-01 1.81322843e-01
1.49267435e-01 -6.66612208e-01 9.18980539e-01 8.19833636e-01
-1.21551049e+00 -5.21482587e-01 -9.38002989e-02 -5.00899017e-01
3.40017885e-01 1.23448133e-01 -8.44952583e-01 1.14669430e+00
7.71717906e-01 9.49730575e-02 -1.14141457e-01 1.09597039e+00
-1.72935739e-01 8.18005484e-03 -2.53122807e-01 -1.62841737e-01
5.41231751e-01 -5.81430078e-01 5.59388995e-01 8.58341157e-01
2.92320043e-01 3.60299200e-01 5.50786138e-01 3.35817128e-01
5.89206815e-01 2.74135292e-01 -5.74537754e-01 -1.05901778e-01
3.67314428e-01 1.03158641e+00 -1.03320599e+00 -2.58466333e-01
-5.77713549e-01 2.73146629e-01 -2.49875531e-01 3.34174484e-01
-6.27968371e-01 -2.43273914e-01 5.26262045e-01 6.74908459e-01
-3.87252837e-01 -4.51728940e-01 2.67335381e-02 -8.11289191e-01
-4.21019226e-01 -1.49633825e+00 5.31524479e-01 -7.19436288e-01
-8.42161357e-01 9.84697342e-01 3.48310590e-01 -1.07144904e+00
-5.09499550e-01 -4.81367618e-01 1.19352579e-01 3.91262531e-01
-1.23804569e+00 -1.25263774e+00 -1.70898259e-01 4.94831413e-01
6.63141787e-01 -8.50867033e-02 1.13078773e+00 2.08625212e-01
-8.81634206e-02 -4.30929929e-01 -5.49486101e-01 -1.16323091e-01
9.82816890e-02 -1.28823864e+00 2.25862056e-01 6.62336111e-01
1.99102215e-03 5.97987056e-01 1.06600094e+00 -4.82359737e-01
-1.07284319e+00 -4.04653847e-01 1.40358853e+00 -4.16230947e-01
9.35454190e-01 -2.64944971e-01 -4.50630754e-01 7.08460510e-01
4.50957745e-01 -7.98323631e-01 5.19942343e-01 2.29746416e-01
-7.75261894e-02 -1.42299294e-01 -1.16272032e+00 7.38315105e-01
1.48969460e+00 -1.96016505e-01 -7.72411585e-01 6.47773266e-01
7.63681293e-01 -9.82174352e-02 -9.86611784e-01 4.04894710e-01
6.59953058e-01 -1.56022751e+00 6.86832547e-01 -5.65250456e-01
1.17599189e-01 -4.14971739e-01 -3.93772155e-01 -8.68126452e-01
-4.73642766e-01 -1.02382827e+00 3.32466781e-01 1.06508148e+00
4.78614151e-01 -1.02161431e+00 8.62918615e-01 8.71311903e-01
-4.15564269e-01 -5.45472085e-01 -9.02717471e-01 -9.74131107e-01
-2.76271522e-01 -6.02183223e-01 1.15456295e+00 9.10359025e-01
5.23154438e-01 2.57632077e-01 1.26367182e-01 -3.34122740e-02
4.36615497e-01 1.21163033e-01 5.29458642e-01 -1.80914116e+00
-1.56749800e-01 -6.25018954e-01 -4.71869379e-01 -7.57527590e-01
1.22387417e-01 -6.90869093e-01 -1.45631179e-01 -2.37883377e+00
-1.31182238e-01 -5.17691493e-01 1.15037607e-02 5.73298693e-01
1.06406558e+00 -1.61974967e-01 3.96536678e-01 5.44772983e-01
-8.07768047e-01 -1.24650180e-01 1.32382846e+00 1.58675343e-01
-1.26012534e-01 2.08955273e-01 -8.22401285e-01 1.04711974e+00
7.20329225e-01 -2.17459887e-01 -4.59967911e-01 -2.25461453e-01
8.11416388e-01 1.50580600e-01 -3.10573250e-01 -9.59367573e-01
5.44380307e-01 -7.90102422e-01 -4.81895596e-01 -1.36959031e-01
3.03127885e-01 -1.44451344e+00 6.87655091e-01 4.95148987e-01
3.03970501e-02 3.92043553e-02 -3.73162478e-01 -3.66402753e-02
-4.68047619e-01 -6.60076380e-01 2.56388575e-01 -6.57060027e-01
-9.61301804e-01 -1.94090884e-02 -7.67765522e-01 -2.48708710e-01
1.12782097e+00 -5.17395973e-01 -1.75279766e-01 -4.67069954e-01
-1.04230177e+00 3.68101329e-01 7.29436517e-01 1.81778684e-01
3.26642007e-01 -1.09906805e+00 -3.30087990e-01 1.00867897e-02
3.43343467e-02 -1.03007361e-01 -8.74040350e-02 6.50943816e-01
-7.29307652e-01 8.65389466e-01 -4.23235387e-01 -2.42739037e-01
-8.85763705e-01 6.42736793e-01 1.88934878e-01 -6.31887376e-01
-3.45987111e-01 2.57381648e-01 2.87540942e-01 -5.30213714e-01
3.38227212e-01 -2.32351646e-01 -8.02776277e-01 1.53393999e-01
2.41989866e-01 3.45177621e-01 -1.39944358e-02 -7.31121063e-01
-4.33730006e-01 4.91128445e-01 4.35098767e-01 -5.08168042e-01
1.33845484e+00 -2.47721925e-01 -8.14096808e-01 2.28022665e-01
1.96377501e-01 -2.17275545e-01 -5.50501406e-01 -6.28871322e-02
7.91122139e-01 -2.10657418e-01 -5.47812223e-01 -8.39688420e-01
-4.84162688e-01 2.80827522e-01 -9.16901454e-02 1.50575030e+00
1.32110000e+00 3.48485470e-01 4.39080328e-01 2.31066048e-01
1.04843438e+00 -1.17458069e+00 -1.94072291e-01 7.13288963e-01
1.08549333e+00 -4.53076094e-01 -1.00019105e-01 -8.60210538e-01
-6.53393507e-01 1.10761487e+00 3.47209007e-01 3.90850961e-01
3.61329913e-01 3.89151365e-01 -1.25603527e-01 -3.95270675e-01
-7.94441640e-01 -6.01309299e-01 -2.18857110e-01 8.86266291e-01
1.01195410e-01 1.55464515e-01 -9.22029316e-01 6.46778882e-01
-5.56321621e-01 -1.82113945e-01 6.12213731e-01 1.05780804e+00
-7.51494527e-01 -1.80763936e+00 -5.14032900e-01 -2.37661358e-02
-3.40060323e-01 1.75963148e-01 -6.28041267e-01 1.12054908e+00
4.17312831e-01 1.33634579e+00 -2.75837064e-01 -2.85567641e-01
7.97813416e-01 3.92970890e-02 2.98218220e-01 -8.13586235e-01
-8.43977034e-01 2.60187238e-01 7.04050422e-01 -6.71360373e-01
-6.87289894e-01 -7.19441712e-01 -1.43554068e+00 -3.40338945e-01
1.31782265e-02 9.32203293e-01 6.73758268e-01 1.09843206e+00
-1.57940283e-01 5.21320999e-01 4.06412244e-01 -7.61574030e-01
-2.98300594e-01 -5.40473580e-01 -7.83717215e-01 2.26792753e-01
-4.60776150e-01 -6.27572179e-01 -3.24350625e-01 1.86313670e-02] | [8.655872344970703, 6.895969390869141] |
679648ae-5a4a-41a1-8c3d-7681ef206ca9 | universal-weak-coreset | 2305.16890 | null | https://arxiv.org/abs/2305.16890v1 | https://arxiv.org/pdf/2305.16890v1.pdf | Universal Weak Coreset | Coresets for $k$-means and $k$-median problems yield a small summary of the data, which preserve the clustering cost with respect to any set of $k$ centers. Recently coresets have also been constructed for constrained $k$-means and $k$-median problems. However, the notion of coresets has the drawback that (i) they can only be applied in settings where the input points are allowed to have weights, and (ii) in general metric spaces, the size of the coresets can depend logarithmically on the number of points. The notion of weak coresets, which have less stringent requirements than coresets, has been studied in the context of classical $k$-means and $k$-median problems. A weak coreset is a pair $(J,S)$ of subsets of points, where $S$ acts as a summary of the point set and $J$ as a set of potential centers. This pair satisfies the properties that (i) $S$ is a good summary of the data as long as the $k$ centers are chosen from $J$ only, and (ii) there is a good choice of $k$ centers in $J$ with cost close to the optimal cost. We develop this framework, which we call universal weak coresets, for constrained clustering settings. In conjunction with recent coreset constructions for constrained settings, our designs give greater data compression, are conceptually simpler, and apply to a wide range of constrained $k$-median and $k$-means problems. | ['Amit Kumar', 'Ragesh Jaiswal'] | 2023-05-26 | null | null | null | null | ['data-compression'] | ['time-series'] | [-2.73670461e-02 -1.16156720e-01 -2.66600370e-01 -1.53309703e-01
-5.28898001e-01 -6.47343278e-01 -2.20101506e-01 4.80963051e-01
-4.03809458e-01 3.69849026e-01 -2.02813536e-01 -2.17156969e-02
-9.92784142e-01 -1.14816046e+00 -7.12952256e-01 -7.69937336e-01
-5.02558231e-01 9.05662060e-01 3.22478682e-01 -8.23023543e-02
5.08804917e-01 4.13553327e-01 -1.79364622e+00 -3.22536558e-01
8.16595078e-01 1.00199509e+00 3.91494423e-01 5.03207326e-01
-2.58943707e-01 -3.15103620e-01 -4.95492220e-01 -1.96069270e-01
6.14469886e-01 -3.09781224e-01 -9.71372962e-01 3.39124829e-01
2.97337443e-01 4.16514009e-01 -2.97538847e-01 1.07968783e+00
1.29939541e-01 2.78975815e-01 5.08270621e-01 -1.45964742e+00
-5.11108518e-01 7.16910243e-01 -8.92823815e-01 -1.66961819e-01
2.07437724e-01 -2.10163087e-01 1.19698250e+00 -6.39777660e-01
4.32417870e-01 8.79431903e-01 7.03601360e-01 3.39189410e-01
-1.36740470e+00 -4.83118147e-01 3.08898687e-02 -5.25632910e-02
-1.71885264e+00 -3.81931633e-01 5.35780907e-01 -1.46700844e-01
8.85695159e-01 6.51038885e-01 5.71474850e-01 -2.47240305e-01
-4.90029752e-01 4.00922298e-01 7.37309635e-01 -5.61186731e-01
6.78220451e-01 -4.34298515e-02 4.15733159e-01 4.98867899e-01
7.04502285e-01 -3.97722423e-01 -1.95586741e-01 -4.36084747e-01
4.68462229e-01 2.38791674e-01 -2.69136071e-01 -7.50871539e-01
-1.09344316e+00 9.64497209e-01 4.80858713e-01 5.02625763e-01
1.54029176e-01 3.69312823e-01 2.16407806e-01 1.19191408e-01
2.88290624e-02 6.10835910e-01 -5.49605966e-01 1.39560387e-01
-1.03506434e+00 3.73432964e-01 6.92398906e-01 1.60513854e+00
1.13228512e+00 -4.01700199e-01 2.99376190e-01 8.83632541e-01
-2.32597202e-01 7.74972916e-01 7.84235448e-02 -1.12583709e+00
6.51353240e-01 9.20280814e-01 1.62659392e-01 -1.20525193e+00
-3.55574310e-01 5.91107234e-02 -9.63331997e-01 -1.74593218e-02
3.47201437e-01 2.71908373e-01 -6.68788612e-01 2.03381157e+00
2.07191393e-01 -7.27221146e-02 -3.61953467e-01 5.10731816e-01
1.36778116e-01 4.70776558e-01 -4.07801449e-01 -5.41102648e-01
1.18431258e+00 -3.90038639e-01 -2.48890832e-01 -9.96968076e-02
7.45113194e-01 -7.21537948e-01 1.18059611e+00 2.12071419e-01
-1.64268422e+00 -3.60509694e-01 -9.01485741e-01 2.23167613e-01
-5.89531958e-01 -2.67303288e-01 5.98751366e-01 8.00958276e-01
-1.46410859e+00 4.37993228e-01 -5.59280217e-01 -5.39384246e-01
3.26080739e-01 6.53082728e-01 -2.49881983e-01 -4.12294686e-01
-6.26934409e-01 4.08711910e-01 3.37500751e-01 -1.95512101e-01
-1.90057099e-01 -5.18481612e-01 -7.55559981e-01 2.41986230e-01
5.40315986e-01 -5.70550501e-01 5.85235775e-01 -4.90415484e-01
-6.30578160e-01 8.40883374e-01 -3.60590756e-01 -8.37335140e-02
2.85963267e-01 4.31948423e-01 -9.95525904e-03 3.66219819e-01
4.29284662e-01 4.67835963e-01 4.71119821e-01 -1.02870011e+00
-7.95841813e-01 -6.75119221e-01 -7.51938149e-02 1.56110495e-01
-4.24688429e-01 -2.55637586e-01 -9.09645259e-01 -3.75703543e-01
5.71204305e-01 -1.10208571e+00 -6.96511149e-01 -1.11223891e-01
-4.85559464e-01 -4.14299399e-01 8.63129616e-01 3.30027521e-01
1.62540412e+00 -2.34282541e+00 7.57515654e-02 9.53643799e-01
4.14504439e-01 9.90861431e-02 -3.96282412e-02 6.36779785e-01
-1.66916415e-01 3.47810894e-01 -4.81043160e-01 -4.19898689e-01
3.13963413e-01 4.05510902e-01 -4.00824994e-02 5.77304304e-01
-4.56873983e-01 3.50360215e-01 -6.37567878e-01 -3.14945281e-01
1.93890974e-01 -2.75428712e-01 -9.16377068e-01 -1.73136070e-01
-1.65401399e-01 -6.44411385e-01 -3.78627121e-01 3.54271084e-01
9.67759490e-01 -4.58219916e-01 1.00360058e-01 1.38455331e-01
4.09435146e-02 -1.52262613e-01 -1.77035332e+00 1.55988753e+00
8.49481076e-02 6.04523122e-02 4.09713358e-01 -1.21878493e+00
8.94637883e-01 -1.02339732e-02 9.43582416e-01 -3.28078359e-01
-8.70215744e-02 4.79706615e-01 -3.63207400e-01 9.88331623e-03
7.80534446e-01 -2.88168818e-01 -6.13501430e-01 8.83178592e-01
-3.89292777e-01 -2.03290612e-01 6.67539418e-01 6.47183120e-01
1.41720700e+00 -7.77410388e-01 -1.19727686e-01 -7.22827554e-01
1.70689464e-01 1.60408854e-01 7.53571570e-01 8.38067412e-01
-7.00707361e-02 7.22916961e-01 3.94982100e-01 -2.36256663e-02
-1.16679001e+00 -1.18269718e+00 -2.16248944e-01 9.17772174e-01
5.59958816e-01 -5.96703827e-01 -9.86696839e-01 -2.32771710e-01
2.61287987e-01 2.11864322e-01 -5.14553785e-01 -8.10146704e-02
-5.19180417e-01 -6.11837387e-01 1.28566191e-01 5.76013744e-01
3.25503826e-01 -6.95793271e-01 -6.51902974e-01 1.84637368e-01
-1.53191715e-01 -8.38305116e-01 -1.09914804e+00 4.17624295e-01
-1.09975839e+00 -1.49709439e+00 -3.91149342e-01 -1.07559848e+00
1.05995870e+00 8.06458116e-01 1.00053239e+00 3.87823284e-01
-3.23225975e-01 4.82901275e-01 -5.51426113e-01 1.05344251e-01
2.64947236e-01 -1.81493368e-02 3.00765365e-01 -4.73251760e-01
6.24567866e-01 -5.06356776e-01 -9.05867696e-01 7.28141487e-01
-1.16661823e+00 -6.00287080e-01 2.73801744e-01 5.62750936e-01
8.21493149e-01 7.19910264e-01 6.30974293e-01 -1.01221144e+00
3.61724854e-01 -4.44827318e-01 -5.60413957e-01 9.22657996e-02
-8.35935414e-01 -4.53405194e-02 8.40075135e-01 -3.87585685e-02
-1.08865641e-01 -2.71800868e-02 1.26274854e-01 -5.05856574e-01
1.49939984e-01 2.79981554e-01 -3.82693768e-01 2.29488015e-01
5.35712540e-01 -2.80099967e-03 -4.59339097e-02 -4.56380993e-01
5.79951882e-01 6.88852429e-01 4.34138864e-01 -7.81982541e-01
6.28620088e-01 7.85913110e-01 1.93946093e-01 -7.45311439e-01
-1.11033216e-01 -8.47492874e-01 -5.38219988e-01 4.22467858e-01
3.53434861e-01 -4.57368702e-01 -1.01420331e+00 1.36563197e-01
-5.03303826e-01 -1.52030468e-01 -7.02298462e-01 1.03110150e-01
-9.69323337e-01 5.52789390e-01 -2.60402083e-01 -5.70421100e-01
-1.80070952e-01 -1.10650992e+00 6.20019078e-01 -8.99229571e-02
-2.85297811e-01 -8.01947773e-01 -1.34953946e-01 1.03744224e-01
1.23309098e-01 2.64203012e-01 1.24998140e+00 -5.17868102e-01
-6.19943976e-01 -4.66789335e-01 -2.52311289e-01 1.19734585e-01
1.14066400e-01 -9.55979824e-02 -3.20693493e-01 -6.98823094e-01
-2.34042689e-01 -1.13330437e-02 6.16608739e-01 6.92208350e-01
1.42733455e+00 -4.91313845e-01 -7.96954334e-01 5.66849291e-01
1.76442981e+00 2.51949131e-01 6.26190960e-01 -7.38389939e-02
3.56760889e-01 4.27830130e-01 3.67919385e-01 6.16619885e-01
3.35479677e-01 4.10285771e-01 2.66863555e-01 5.57541847e-02
3.06673974e-01 8.82728472e-02 -1.96393639e-01 1.05034006e+00
2.84890950e-01 -1.55093044e-01 -6.88294888e-01 1.04970145e+00
-1.81608462e+00 -9.09658074e-01 -6.96254447e-02 2.59327269e+00
6.89262986e-01 1.69130743e-01 5.55131793e-01 6.26987517e-01
1.08375418e+00 -8.40788037e-02 -6.34599686e-01 -5.63499093e-01
-7.56540447e-02 7.26682007e-01 8.79190266e-01 3.50101084e-01
-7.09632754e-01 3.94006401e-01 6.75694895e+00 1.13295329e+00
-4.62650836e-01 -1.08214833e-01 6.30609155e-01 -4.60551441e-01
-5.09380579e-01 7.05383793e-02 -6.66780770e-01 8.33330452e-01
6.44145429e-01 -2.19423681e-01 3.41828942e-01 1.06684625e+00
-5.44439666e-02 -3.74849260e-01 -1.43372989e+00 1.22126794e+00
-1.99281529e-01 -1.57295656e+00 -2.77119964e-01 4.21049595e-01
7.60263503e-01 -1.71794891e-01 2.37607196e-01 -7.69022852e-02
6.57130599e-01 -9.37023163e-01 4.13002580e-01 -1.17039561e-01
1.21750951e+00 -1.21352291e+00 1.68583646e-01 4.29465860e-01
-1.54336071e+00 -1.22754358e-01 -7.45457172e-01 1.96256116e-01
-1.75105873e-02 7.40374506e-01 -1.57310963e-01 3.53303015e-01
1.10958898e+00 2.81834126e-01 -7.48578906e-02 8.76849771e-01
7.45830655e-01 1.59266800e-01 -8.78702939e-01 -1.36543721e-01
3.42552841e-01 -3.91949713e-01 1.53207123e-01 1.19018579e+00
6.15343690e-01 4.52153593e-01 1.61036521e-01 5.07762849e-01
-1.87904507e-01 1.02875076e-01 -6.16267443e-01 5.37401810e-02
1.05061889e+00 9.61076379e-01 -1.08472884e+00 -3.24964762e-01
-1.59740493e-01 5.67801774e-01 2.16957137e-01 8.51979256e-02
-2.71678627e-01 -1.19769931e+00 9.78307605e-01 5.60884178e-01
5.75576127e-01 -3.39337111e-01 -6.58387899e-01 -8.32045555e-01
-2.92178579e-02 -4.44162607e-01 9.20938194e-01 -5.10921299e-01
-1.38641131e+00 3.00722659e-01 9.62485075e-02 -1.07221663e+00
8.07975680e-02 -5.43449521e-01 -5.05718112e-01 7.32875109e-01
-8.05736423e-01 -3.37516159e-01 1.08820274e-02 1.06978905e+00
1.11765079e-02 1.56698793e-01 6.70052409e-01 1.35948062e-01
-1.96462125e-01 6.49200559e-01 5.97444952e-01 1.67360378e-03
2.54739046e-01 -1.35818362e+00 1.46262127e-03 8.25099468e-01
-1.44603074e-01 1.00534427e+00 7.16038227e-01 -2.58402050e-01
-1.51908135e+00 -9.50088382e-01 7.83163488e-01 -3.41167390e-01
2.67015696e-01 -3.51364881e-01 -6.73880219e-01 5.68843842e-01
-1.92919016e-01 3.51329930e-02 1.03283215e+00 3.48115861e-01
-2.85499781e-01 -5.95891118e-01 -1.56903768e+00 4.32921976e-01
1.26819360e+00 -2.08172992e-01 -2.55181968e-01 2.37520784e-01
6.07760310e-01 9.35628787e-02 -9.37948406e-01 2.59593785e-01
2.84575373e-01 -1.03925347e+00 1.00934017e+00 -1.85901284e-01
-4.95457761e-02 -5.32884896e-01 -5.85202992e-01 -1.12881684e+00
-5.48176765e-01 -6.01095378e-01 3.16801101e-01 8.74064922e-01
2.94902891e-01 -5.61925292e-01 1.14321959e+00 8.41569126e-01
-1.09147750e-01 -6.78525925e-01 -1.14562345e+00 -8.85385156e-01
2.76933908e-01 -3.42951477e-01 9.71541107e-01 9.45717633e-01
6.17817521e-01 3.24143097e-02 -1.00655230e-02 -1.68470353e-01
8.72244537e-01 4.73715305e-01 7.12679565e-01 -1.09431958e+00
-3.79217327e-01 -7.45863259e-01 -3.20715845e-01 -1.26199687e+00
-2.92423040e-01 -9.18605208e-01 1.42193409e-02 -1.38555598e+00
2.91243613e-01 -1.47411251e+00 -1.47494435e-01 2.61948347e-01
3.55962552e-02 2.71544099e-01 1.65859908e-01 3.33324373e-01
-8.40421200e-01 9.56102610e-02 9.76738572e-01 2.37884611e-01
-2.91883439e-01 4.58073728e-02 -9.41908479e-01 5.24792314e-01
6.68851078e-01 -3.05746436e-01 -5.47524691e-01 -3.73277843e-01
2.99394965e-01 1.91965789e-01 -3.13965619e-01 -7.54263222e-01
3.57118756e-01 -4.18745428e-01 1.80321872e-01 -8.48780036e-01
2.96447426e-01 -1.12156582e+00 2.69707024e-01 4.69454676e-01
-1.90915272e-01 2.56878465e-01 -1.77580133e-01 6.81311607e-01
1.51563380e-02 -2.60619462e-01 1.04484665e+00 -3.24943513e-01
-3.42684001e-01 7.43445694e-01 -7.40911951e-03 2.08383292e-01
1.47399974e+00 -9.30625737e-01 -4.42474365e-01 -8.29535276e-02
-7.14388847e-01 5.78334689e-01 8.72389972e-01 -1.45790979e-01
6.22323513e-01 -1.33951628e+00 -2.62787759e-01 2.72876531e-01
3.59712541e-01 7.41924405e-01 2.01821044e-01 5.46933651e-01
-6.30268574e-01 3.62997532e-01 1.50688171e-01 -6.79431200e-01
-9.20254827e-01 9.94081616e-01 -9.27007198e-02 1.53176636e-01
-4.85090107e-01 8.99259090e-01 1.38688877e-01 -1.96514070e-01
2.48048469e-01 -3.30321640e-01 5.56717277e-01 -8.13880190e-02
4.17673767e-01 6.96217537e-01 3.55278030e-02 -5.73109448e-01
-5.25531292e-01 8.11346054e-01 -6.35342151e-02 -1.30831972e-01
1.36212444e+00 -3.11423779e-01 -4.51326907e-01 5.85143175e-03
1.31393814e+00 1.52121112e-01 -8.54918182e-01 -4.13142294e-01
7.16298670e-02 -8.24379325e-01 -5.91015399e-01 -1.74149737e-01
-1.19530141e+00 5.62917590e-01 4.02128175e-02 6.02782607e-01
1.47529387e+00 4.57662880e-01 9.22011554e-01 2.22607538e-01
7.93703914e-01 -1.55652928e+00 6.67218566e-02 2.43727684e-01
2.26894468e-01 -5.82262516e-01 -1.28202051e-01 -1.38499379e-01
-2.89255291e-01 9.73611236e-01 4.40951824e-01 -4.99124706e-01
1.04124987e+00 3.88974816e-01 -4.00590122e-01 -3.09937090e-01
-7.30691135e-01 -3.21675181e-01 -2.94501722e-01 5.43248355e-01
-1.12048723e-01 3.06203276e-01 -5.49741387e-01 5.85338414e-01
-4.23456848e-01 -1.74372405e-01 4.83569771e-01 1.02956700e+00
-1.14577138e+00 -1.10855031e+00 -6.15117431e-01 8.15574467e-01
-1.23709604e-01 1.75248846e-01 -7.72269517e-02 5.67800879e-01
3.37267101e-01 1.23243141e+00 5.62897801e-01 -3.55761707e-01
3.39044631e-01 -2.32577801e-01 3.55039299e-01 -7.71046042e-01
-2.05712050e-01 2.83418566e-01 -5.50193965e-01 -4.67977911e-01
-3.41994822e-01 -5.80015659e-01 -1.45036209e+00 -1.03792024e+00
-4.49379832e-01 7.74012744e-01 4.20805544e-01 6.01746082e-01
4.77280468e-01 -2.34592795e-01 9.80574429e-01 -2.72031873e-01
-3.42925310e-01 -5.98196149e-01 -1.37248480e+00 5.24191201e-01
-9.49652866e-03 -3.72189045e-01 -4.33835506e-01 -1.46999734e-03] | [6.790865898132324, 4.976040840148926] |
daee6f19-364d-4057-9bb0-56c44363271a | unicon-ictcas-ucas-submission-to-the-ava | 2206.10861 | null | https://arxiv.org/abs/2206.10861v1 | https://arxiv.org/pdf/2206.10861v1.pdf | UniCon+: ICTCAS-UCAS Submission to the AVA-ActiveSpeaker Task at ActivityNet Challenge 2022 | This report presents a brief description of our winning solution to the AVA Active Speaker Detection (ASD) task at ActivityNet Challenge 2022. Our underlying model UniCon+ continues to build on our previous work, the Unified Context Network (UniCon) and Extended UniCon which are designed for robust scene-level ASD. We augment the architecture with a simple GRU-based module that allows information of recurring identities to flow across scenes through read and update operations. We report a best result of 94.47% mAP on the AVA-ActiveSpeaker test set, which continues to rank first on this year's challenge leaderboard and significantly pushes the state-of-the-art. | ['Shiguang Shan', 'Shuang Yang', 'Susan Liang', 'Yuanhang Zhang'] | 2022-06-22 | null | null | null | null | ['audio-visual-active-speaker-detection'] | ['computer-vision'] | [ 8.31620842e-02 1.91329762e-01 -1.24359280e-01 -4.98402297e-01
-1.28792822e+00 -6.83100700e-01 1.10111225e+00 -3.24998915e-01
-3.10384959e-01 4.70467776e-01 7.51303852e-01 -2.63119996e-01
2.30095983e-01 -6.94512995e-03 -2.40812182e-01 -2.37280205e-01
-1.91629738e-01 4.39648300e-01 8.08997750e-01 -3.73908937e-01
3.91081274e-02 2.13441879e-01 -1.64898360e+00 7.49396563e-01
2.56284207e-01 9.04915571e-01 -5.50794043e-02 1.12756503e+00
-1.16585337e-01 1.16592944e+00 -9.36306953e-01 -7.35943988e-02
6.83627501e-02 -3.23810548e-01 -1.16371465e+00 -4.66270834e-01
1.50905657e+00 -2.19085589e-01 -4.77622926e-01 6.40498340e-01
1.00691187e+00 2.46871710e-01 8.14192668e-02 -1.36225903e+00
-3.79125208e-01 1.09824824e+00 -2.26570338e-01 1.03352153e+00
6.70848608e-01 2.76002675e-01 1.36612093e+00 -1.11741793e+00
7.05550373e-01 1.33005440e+00 9.14419532e-01 8.32652032e-01
-9.88737106e-01 -7.87259042e-01 6.26233220e-01 6.01997137e-01
-1.57655942e+00 -1.28054357e+00 3.00815284e-01 -1.26694292e-01
1.40261281e+00 6.47574067e-01 7.65391231e-01 1.73905730e+00
-5.62186837e-01 1.17917669e+00 6.75645113e-01 -1.20099753e-01
2.50908583e-01 -6.58634529e-02 3.69888902e-01 3.25421095e-01
-2.36637414e-01 -1.54805511e-01 -1.61020708e+00 2.59745866e-02
4.01233971e-01 -1.03138983e+00 -2.37116113e-01 -6.34379238e-02
-1.34991288e+00 6.56091869e-01 3.24809104e-01 2.01155916e-01
5.74392118e-02 3.08181345e-01 3.66073132e-01 1.35847569e-01
5.45948029e-01 5.16988277e-01 -7.99564794e-02 -5.97372055e-01
-1.25707257e+00 3.60200405e-01 1.04220831e+00 9.57979023e-01
-7.01311305e-02 3.21387976e-01 -5.26742578e-01 1.13488603e+00
4.77152586e-01 3.56678933e-01 2.59361833e-01 -9.81756985e-01
2.96612829e-01 -6.04608059e-02 -2.22184926e-01 -4.88308072e-01
-2.64105618e-01 -6.45652831e-01 -1.15646206e-01 9.71647203e-02
2.71965355e-01 -9.56115723e-02 -7.74119854e-01 1.71036732e+00
2.77134657e-01 8.69086206e-01 -1.54799938e-01 7.08807290e-01
1.44487178e+00 5.15042782e-01 8.01407173e-02 2.07315996e-01
1.23511648e+00 -1.46394801e+00 -6.32894754e-01 -5.88343918e-01
3.26642811e-01 -6.79179251e-01 8.24437499e-01 5.22713184e-01
-1.16153681e+00 -4.79191184e-01 -1.45588124e+00 -1.15631707e-01
-4.91233885e-01 -4.34595615e-01 5.96003890e-01 1.01196325e+00
-2.04099631e+00 7.29590431e-02 -6.40218794e-01 -8.27753961e-01
4.55789685e-01 2.56785870e-01 -1.73273087e-01 1.79250166e-01
-1.23370469e+00 8.54287922e-01 -2.53792971e-01 1.70029253e-02
-1.53467453e+00 -8.20013583e-01 -6.83709443e-01 -3.12594801e-01
5.16959190e-01 -5.59640169e-01 1.99747455e+00 -4.82049376e-01
-1.62525308e+00 9.05896604e-01 -7.25803524e-02 -7.66474903e-01
6.29745126e-01 -2.21201509e-01 -9.95839775e-01 9.74264368e-02
1.75947100e-01 1.00218248e+00 4.65869606e-01 -7.62183964e-01
-1.01588821e+00 -1.46454647e-01 7.79253021e-02 4.96390432e-01
-4.46163304e-03 7.92375863e-01 -5.35573781e-01 -5.92339694e-01
1.21170074e-01 -6.55583799e-01 1.94150001e-01 6.08104095e-02
-4.75754857e-01 -4.96255368e-01 1.12517953e+00 -6.92925215e-01
1.14823663e+00 -2.26906657e+00 -5.55743799e-02 -2.39262022e-02
5.23503661e-01 4.49709408e-02 -3.57747734e-01 1.74061671e-01
-5.79608977e-02 1.36487780e-03 -1.18908333e-02 -9.54903126e-01
2.25974932e-01 -8.13864768e-02 -1.69011980e-01 4.55579430e-01
-2.30739966e-01 6.76826775e-01 -8.66291523e-01 -3.62615496e-01
4.45103534e-02 4.20659572e-01 -4.12632287e-01 -1.96678519e-01
-1.68270141e-01 2.82554358e-01 1.00654855e-01 6.57525539e-01
5.54746628e-01 -5.54820895e-02 7.98110291e-02 2.29410127e-01
-6.14336312e-01 1.06982875e+00 -1.47753417e+00 2.24817920e+00
-3.39271575e-01 1.25589347e+00 7.30715156e-01 -4.25409585e-01
6.86658502e-01 3.18854511e-01 2.40545660e-01 -6.28191531e-01
-2.71098047e-01 8.27393532e-02 -8.57530385e-02 -7.88303092e-02
7.84180641e-01 5.83799422e-01 -8.21561217e-02 3.52454007e-01
2.98204124e-01 -3.98779213e-02 -3.34597472e-03 8.48952472e-01
1.52522528e+00 5.15662506e-02 2.71686483e-02 -7.54028738e-01
7.67641068e-01 -2.76591092e-01 2.65513450e-01 1.23314095e+00
-6.62282586e-01 7.39696801e-01 2.85727054e-01 -1.51771605e-01
-5.47656417e-01 -1.30563271e+00 -2.36207873e-01 1.74017489e+00
-7.39034861e-02 -8.77200544e-01 -5.99210680e-01 -7.75708377e-01
-3.99586499e-01 8.78901958e-01 -5.50109625e-01 4.36810136e-01
-4.98376280e-01 -3.73039335e-01 1.07031476e+00 6.97071970e-01
1.05317843e+00 -9.52868998e-01 -1.14827819e-01 1.70907393e-01
-3.54991078e-01 -1.22685099e+00 -7.32649505e-01 -1.05173001e-02
-1.92096248e-01 -8.02306890e-01 -5.23250461e-01 -7.75967121e-01
-4.85748500e-01 2.15773195e-01 1.41276109e+00 -4.39719349e-01
-3.34002823e-01 6.05345905e-01 -1.18845798e-01 -5.05267203e-01
-4.93387550e-01 3.88850987e-01 1.54889733e-01 -4.74285558e-02
4.15392458e-01 -5.57015777e-01 -6.55505598e-01 4.62108850e-01
1.06662422e-01 1.54385651e-02 2.32379623e-02 1.00597702e-01
2.17308238e-01 -7.42435992e-01 8.49259973e-01 -4.01081383e-01
4.45479751e-01 -3.21354747e-01 -4.14120376e-01 1.80571228e-01
-2.50547290e-01 -4.20885086e-01 -3.22688162e-01 -4.11454663e-02
-1.06117129e+00 1.33017257e-01 -4.73762631e-01 -1.50063157e-01
-1.52927056e-01 -1.68025717e-01 -2.97265053e-01 -1.52368098e-01
8.22991490e-01 3.00576061e-01 -2.60983706e-01 -4.99282628e-01
5.26855111e-01 1.01634371e+00 1.17402625e+00 -3.01544517e-01
1.83422044e-01 3.89191806e-01 -5.00113189e-01 -1.22459316e+00
-6.36260271e-01 -9.49336529e-01 -4.53980088e-01 -5.32904327e-01
1.09010708e+00 -1.45507240e+00 -6.83333397e-01 7.43695080e-01
-1.17193937e+00 -5.50975561e-01 -3.28178644e-01 2.03679740e-01
-1.84715241e-01 2.20869169e-01 -4.41704243e-01 -8.26924860e-01
-4.21053469e-01 -8.77419174e-01 1.01643550e+00 -4.77551036e-02
-6.74938500e-01 -7.88348138e-01 5.33673942e-01 5.80731988e-01
8.81729126e-01 -3.31129521e-01 -4.52705659e-02 -1.04436147e+00
-8.29041719e-01 2.85935938e-01 -1.45172298e-01 2.49186054e-01
-3.24516207e-01 -1.02924071e-01 -1.70993471e+00 -1.12699457e-01
-3.15606356e-01 -1.91428900e-01 1.04649508e+00 3.34176064e-01
7.99712598e-01 -4.74069752e-02 -4.00577575e-01 5.14249444e-01
5.88933289e-01 -5.33718476e-03 3.98836315e-01 1.47814795e-01
5.95940650e-01 4.07913327e-01 -6.31604269e-02 3.09881866e-01
6.84938312e-01 1.09937716e+00 5.75515747e-01 1.42831922e-01
-9.69272017e-01 -8.06124210e-02 6.95263326e-01 7.81072259e-01
1.28167585e-01 -4.12577599e-01 -1.09220874e+00 8.06810081e-01
-1.69027460e+00 -1.19523060e+00 -5.09597719e-01 2.02879453e+00
5.89296162e-01 7.39511847e-02 6.52623117e-01 -1.94829747e-01
8.22620273e-01 6.90622807e-01 -4.59051400e-01 -4.24146056e-01
-4.18742955e-01 2.32291743e-01 1.69517964e-01 8.05053651e-01
-1.30700719e+00 1.01576495e+00 8.20135593e+00 8.38775098e-01
-8.26517940e-01 6.67578816e-01 2.29427338e-01 -6.68385446e-01
8.48840401e-02 -4.11762953e-01 -1.25133276e+00 3.28894109e-01
1.39089394e+00 -1.36887833e-01 4.65276331e-01 8.46292198e-01
-2.57226061e-02 -1.47278681e-01 -1.30955529e+00 1.22832227e+00
6.63891673e-01 -1.59939504e+00 -3.54431957e-01 -5.61022572e-03
4.65784848e-01 1.04178119e+00 1.37202024e-01 5.88234961e-01
8.27964067e-01 -9.47712064e-01 1.02883995e+00 4.77618799e-02
8.13810468e-01 -3.67075980e-01 3.57373208e-01 -1.44665211e-01
-1.46556187e+00 -3.46550085e-02 1.81828588e-01 4.86628339e-02
2.25577027e-01 -6.96079433e-02 -1.07425630e+00 3.51760238e-01
9.69583571e-01 9.02073383e-01 -9.33324635e-01 1.16815662e+00
-2.75880963e-01 7.93173611e-01 -7.24642336e-01 -6.65917322e-02
3.20068985e-01 5.76495111e-01 1.13727820e+00 1.43358099e+00
-2.25415863e-02 -3.92270654e-01 -3.70612703e-02 5.42880177e-01
-4.19874191e-01 -4.43700323e-04 -4.75322902e-01 5.13160110e-01
6.79416716e-01 1.22272813e+00 -3.20765287e-01 -5.10438085e-01
-4.88647044e-01 9.76012349e-01 1.70633234e-02 7.43659660e-02
-9.95745897e-01 8.08987543e-02 8.83652508e-01 7.95081928e-02
1.89298406e-01 -7.24886870e-03 -1.66577324e-01 -8.90510380e-01
-3.12597424e-01 -8.35973442e-01 7.73470521e-01 -8.94117773e-01
-1.04134429e+00 8.95434201e-01 -1.44551264e-03 -8.78065825e-01
-1.36834651e-01 -2.31217846e-01 -9.77558255e-01 7.52265751e-01
-9.96365607e-01 -1.24054849e+00 -5.37073910e-01 7.64141440e-01
1.03141129e+00 -4.63780850e-01 7.59994030e-01 4.41397339e-01
-7.22999513e-01 9.93727386e-01 -1.47233069e-01 -8.41297442e-04
1.00501049e+00 -1.39279151e+00 1.14412582e+00 1.11951959e+00
6.02704048e-01 3.30707431e-01 5.94314814e-01 -3.75865817e-01
-8.44378829e-01 -7.86742926e-01 8.90937209e-01 -1.12281132e+00
9.84599590e-01 -1.03774619e+00 -3.88486415e-01 9.44557965e-01
9.07620728e-01 4.39763721e-03 5.84926128e-01 6.60735071e-01
-6.65530980e-01 -2.29927808e-01 -9.47108328e-01 4.77747411e-01
1.54836035e+00 -7.19919682e-01 -6.18121207e-01 4.24010754e-01
7.39384472e-01 -4.99097914e-01 -3.54209006e-01 9.46480408e-02
6.50123358e-01 -9.90799189e-01 9.78526294e-01 -3.36925894e-01
-3.25413346e-02 -2.63631999e-01 -4.58640993e-01 -1.07394278e+00
-3.14904660e-01 -1.11579287e+00 -2.18206570e-01 1.34636962e+00
6.58543289e-01 -5.38974643e-01 9.23934400e-01 3.23398978e-01
-7.06960380e-01 -1.24865286e-01 -1.68177295e+00 -8.78669024e-01
-4.16645586e-01 -7.96034992e-01 5.38659692e-01 9.84118462e-01
-7.09420964e-02 8.50111663e-01 -2.22832426e-01 1.30852103e-01
6.01885736e-01 -6.61659598e-01 9.18800175e-01 -1.18563271e+00
-4.92573172e-01 -6.26966178e-01 -5.86189449e-01 -1.40132797e+00
-2.28739277e-01 -1.04905307e+00 4.08386067e-02 -1.51324844e+00
7.51109421e-02 -1.02981620e-01 -3.41147482e-01 5.50396979e-01
3.33826691e-01 5.52970290e-01 3.16904664e-01 2.03856118e-02
-1.35978889e+00 3.30971211e-01 6.39583886e-01 -3.27856839e-01
-2.16441765e-01 9.92858186e-02 -8.50932837e-01 5.42904377e-01
5.29527545e-01 -1.52271405e-01 -4.80483860e-01 -5.62063694e-01
3.52915451e-02 -3.23708743e-01 4.71650571e-01 -1.43304837e+00
7.10200727e-01 4.40741062e-01 4.30618860e-02 -9.96272266e-01
7.23500252e-01 -1.61968514e-01 -1.51798561e-01 -4.63118330e-02
-5.22063911e-01 -1.42133916e-02 3.76817226e-01 5.75700998e-01
-8.40315688e-03 2.13902786e-01 5.95716834e-01 1.25138953e-01
-1.29560161e+00 1.76427960e-02 -5.22571802e-01 2.28441730e-01
7.63391197e-01 -2.37552211e-01 -8.72756004e-01 -7.13040054e-01
-1.10020709e+00 4.52096790e-01 -1.76835760e-01 8.31909060e-01
1.96359485e-01 -1.08234370e+00 -1.10190654e+00 6.34443015e-02
4.03205514e-01 -2.48381019e-01 5.87169528e-01 8.12934518e-01
-4.43843395e-01 5.23643255e-01 2.18382254e-01 -9.58293915e-01
-1.62866998e+00 -2.12721840e-01 6.72749698e-01 6.84401840e-02
-6.20539069e-01 1.60013986e+00 3.57930772e-02 -3.32794249e-01
6.88364625e-01 -4.09667082e-02 -2.64007956e-01 1.80162460e-01
9.96717632e-01 6.45202518e-01 2.92981297e-01 -6.25845671e-01
-8.84496033e-01 -1.07317559e-01 -3.49610984e-01 -7.11973846e-01
9.99993801e-01 -2.91271031e-01 1.82467580e-01 4.58814621e-01
7.35596001e-01 3.35525244e-01 -1.10934341e+00 -2.71192372e-01
-1.08075134e-01 -1.61370605e-01 4.36834961e-01 -1.41130161e+00
-7.77400017e-01 7.35627592e-01 8.60532761e-01 1.69671416e-01
5.02432764e-01 5.95789135e-01 3.51165146e-01 3.93108249e-01
1.02537155e-01 -1.20530045e+00 1.45055264e-01 6.26296759e-01
1.20699155e+00 -9.48208153e-01 -6.73801303e-02 -3.76952678e-01
-5.65308392e-01 6.11227214e-01 9.78959739e-01 3.94122362e-01
5.45618773e-01 4.22408670e-01 2.91693568e-01 -2.75785923e-01
-9.72122490e-01 -8.48599821e-02 7.21396925e-03 9.08814669e-01
3.22774887e-01 -1.29689174e-02 3.75493616e-01 4.71755832e-01
-4.15883034e-01 -5.04080713e-01 4.30995315e-01 6.81501031e-01
-4.33773130e-01 -8.93638015e-01 -2.08058119e-01 1.04881497e-02
-3.16296101e-01 -5.34155190e-01 -8.54578257e-01 7.26270854e-01
-1.06001548e-01 1.30805683e+00 4.32820499e-01 -3.83950949e-01
4.94990110e-01 2.91711092e-01 3.53802234e-01 -7.94339478e-01
-8.99566889e-01 3.05855437e-03 9.27300513e-01 -9.23124850e-01
-2.85648584e-01 -1.04349566e+00 -6.25475347e-01 -5.98908722e-01
-3.31149474e-02 -8.77136588e-02 8.87278676e-01 4.76842612e-01
5.60500383e-01 7.46917844e-01 3.01673949e-01 -6.97659492e-01
-2.36521155e-01 -1.27525377e+00 -5.02263904e-01 -3.95838857e-01
3.18337619e-01 -2.85431802e-01 -3.84923935e-01 -3.70288551e-01] | [14.405085563659668, 5.80491828918457] |
0b54e3e5-528c-4ff2-af62-c050e2169dfb | team-enigma-at-argmining-emnlp-2021 | 2110.12370 | null | https://arxiv.org/abs/2110.12370v1 | https://arxiv.org/pdf/2110.12370v1.pdf | Team Enigma at ArgMining-EMNLP 2021: Leveraging Pre-trained Language Models for Key Point Matching | We present the system description for our submission towards the Key Point Analysis Shared Task at ArgMining 2021. Track 1 of the shared task requires participants to develop methods to predict the match score between each pair of arguments and keypoints, provided they belong to the same topic under the same stance. We leveraged existing state of the art pre-trained language models along with incorporating additional data and features extracted from the inputs (topics, key points, and arguments) to improve performance. We were able to achieve mAP strict and mAP relaxed score of 0.872 and 0.966 respectively in the evaluation phase, securing 5th place on the leaderboard. In the post evaluation phase, we achieved a mAP strict and mAP relaxed score of 0.921 and 0.982 respectively. All the codes to generate reproducible results on our models are available on Github. | ['Abhilash Nandy', 'Varun Madhavan', 'Siba Smarak Panigrahi', 'Sohan Patnaik', 'Manav Nitin Kapadnis'] | 2021-10-24 | null | https://aclanthology.org/2021.argmining-1.21 | https://aclanthology.org/2021.argmining-1.21.pdf | emnlp-argmining-2021-11 | ['key-point-matching'] | ['natural-language-processing'] | [-2.35893950e-01 4.61719632e-01 -3.62756491e-01 -5.54198146e-01
-1.44051254e+00 -1.11550987e+00 1.26022530e+00 7.30756640e-01
-6.46651447e-01 5.39829910e-01 4.49548185e-01 -7.26513416e-02
-2.62474447e-01 -5.09547412e-01 -8.75281811e-01 2.29834821e-02
-1.96940467e-01 6.77970052e-01 4.40845758e-01 -2.99884319e-01
5.70160747e-01 -3.10173810e-01 -1.35073102e+00 1.01236880e+00
3.96827310e-01 1.16428709e+00 -1.97729126e-01 9.07178223e-01
7.13290945e-02 3.93119335e-01 -7.74063528e-01 -7.77320385e-01
4.63741869e-01 2.24305376e-01 -1.07570970e+00 -7.89001644e-01
8.31802309e-01 2.01990947e-01 2.28691533e-01 7.46693194e-01
4.60951954e-01 8.78833681e-02 5.82302272e-01 -1.25110495e+00
-1.21870346e-01 1.08577776e+00 -3.37535918e-01 2.29072332e-01
4.82546329e-01 -3.14290285e-01 1.60071385e+00 -1.35745847e+00
8.42904031e-01 1.05374491e+00 7.15186715e-01 1.07250169e-01
-9.56427038e-01 -6.64772034e-01 3.67019743e-01 2.24815577e-01
-1.06255853e+00 -4.97033447e-01 3.54240298e-01 -2.88868308e-01
1.17898798e+00 3.89705241e-01 4.65030432e-01 1.10369885e+00
5.09115830e-02 9.84383702e-01 1.06842554e+00 -4.63971525e-01
-6.30505532e-02 4.79189456e-01 4.18630540e-01 3.74044657e-01
-1.08602971e-01 -1.35514960e-01 -9.95066881e-01 -4.56450492e-01
-1.74352769e-02 -6.60771787e-01 2.53562123e-01 -1.26971304e-01
-1.60668302e+00 9.18257833e-01 5.07404327e-01 1.84235811e-01
-3.40560734e-01 -5.84638156e-02 5.23424566e-01 4.25239295e-01
5.54492354e-01 9.45960879e-01 -8.87731433e-01 -5.63871264e-01
-1.09348655e+00 1.04074442e+00 1.11773312e+00 9.23406422e-01
4.48826879e-01 -1.06814051e+00 -3.78617615e-01 6.84590340e-01
4.19630051e-01 3.69294494e-01 2.17935354e-01 -1.00018609e+00
9.50942636e-01 9.30525601e-01 3.68235290e-01 -8.61753643e-01
-3.62483442e-01 -5.95206320e-01 -5.86563610e-02 -9.55251455e-02
4.56838250e-01 -1.77897379e-01 -6.70169055e-01 1.70290303e+00
3.75021189e-01 -1.32704139e-01 1.13745578e-01 5.63193023e-01
1.03577733e+00 6.60598516e-01 -1.59751158e-02 1.51227236e-01
1.51031971e+00 -1.03855312e+00 -5.07652938e-01 -2.62211919e-01
9.23388004e-01 -1.40120673e+00 1.14652312e+00 6.09023094e-01
-1.38354850e+00 -4.65476215e-01 -1.10211027e+00 -7.71206021e-02
-6.21540010e-01 2.52521992e-01 2.70856321e-01 1.67621553e-01
-9.41760600e-01 4.59819227e-01 -5.64569831e-01 -1.70614913e-01
-8.97769406e-02 3.14102560e-01 -4.21364546e-01 3.66713285e-01
-1.57178211e+00 9.54465210e-01 1.87765107e-01 -3.26079696e-01
-5.90319157e-01 -1.06841648e+00 -5.44525564e-01 -2.77978539e-01
2.59063780e-01 -2.62864739e-01 1.65267241e+00 -2.05355123e-01
-1.12406993e+00 1.14675236e+00 -2.21190453e-01 -5.48970222e-01
9.20800090e-01 -1.01385963e+00 -1.48229167e-01 -1.95727721e-01
5.88954568e-01 9.29437220e-01 2.91153818e-01 -6.55947208e-01
-1.18454647e+00 -5.07359654e-02 3.44370782e-01 3.59005302e-01
-2.11420193e-01 3.83164912e-01 -4.45169032e-01 -4.83384281e-01
-1.89885590e-02 -8.67752433e-01 8.41336250e-02 -4.06809121e-01
-4.20124471e-01 -8.14187825e-01 5.11054337e-01 -7.28016496e-01
1.27118170e+00 -1.86317384e+00 8.80428702e-02 2.60831594e-01
4.80510145e-02 1.58009797e-01 -3.65444534e-02 6.21996582e-01
5.77024221e-02 1.25537083e-01 1.91339433e-01 -5.45232594e-01
2.58895248e-01 -4.59767610e-01 -5.70121527e-01 3.07022572e-01
-5.35471775e-02 8.85891438e-01 -8.73390436e-01 -3.66042227e-01
-2.51423642e-02 1.61893189e-01 -5.79903901e-01 3.32413495e-01
-3.79506022e-01 -2.35611480e-02 -3.52256566e-01 3.73495311e-01
3.49745542e-01 -2.63859272e-01 4.94399033e-02 -2.81283081e-01
-3.64859074e-01 1.12135625e+00 -1.32013488e+00 1.82119119e+00
-4.71917480e-01 6.12510800e-01 1.58134792e-02 -6.34794891e-01
9.35117602e-01 4.20075178e-01 3.93579751e-01 -7.19277918e-01
-8.84256214e-02 3.96139473e-01 -2.02480674e-01 -2.27080565e-02
8.22723746e-01 4.31168884e-01 -6.33428574e-01 7.28160679e-01
-4.71630432e-02 -4.25309390e-02 4.28551018e-01 5.56089640e-01
9.71389174e-01 2.12985799e-01 1.45049229e-01 -5.69124639e-01
5.14126897e-01 5.18361665e-02 2.43307844e-01 9.35363412e-01
5.87704517e-02 6.01500988e-01 5.20036578e-01 -5.88206947e-01
-1.10060191e+00 -8.14113677e-01 -1.54018894e-01 1.45606077e+00
-5.25885411e-02 -1.06931531e+00 -4.26030636e-01 -9.04784739e-01
-4.00773138e-02 6.84976399e-01 -8.42667043e-01 2.03884959e-01
-8.23444366e-01 -4.59710181e-01 7.69444585e-01 4.50661480e-01
2.95440167e-01 -7.75025070e-01 -6.32773221e-01 3.04914713e-01
-4.43533391e-01 -1.02153897e+00 -2.28133723e-01 2.25100517e-01
-3.26461524e-01 -1.12208140e+00 -5.18695652e-01 -5.64205348e-01
6.94750398e-02 -1.32493585e-01 1.40082037e+00 -4.82323132e-02
1.06717922e-01 -8.83585587e-02 -4.30667043e-01 -6.84822798e-01
-1.62892759e-01 6.46068275e-01 -3.00471811e-03 -4.30486500e-01
3.00740302e-01 -2.39176378e-01 -6.41838431e-01 5.20527482e-01
-4.78747070e-01 1.68828830e-01 3.94022763e-01 5.43551326e-01
2.77387619e-01 -6.90029144e-01 3.70357245e-01 -9.11543667e-01
7.68930972e-01 -5.17980993e-01 -4.86253917e-01 4.12505686e-01
-8.47671688e-01 1.03929304e-01 2.10025847e-01 -1.12863958e-01
-7.00252473e-01 -2.50276357e-01 -1.78578570e-02 1.86113298e-01
1.23235308e-01 6.72720730e-01 1.88052818e-01 3.74134392e-01
7.58746803e-01 -3.56256455e-01 -3.33027661e-01 -6.44563019e-01
6.65709972e-01 6.62449419e-01 3.66712093e-01 -8.71917188e-01
6.14979744e-01 2.24482298e-01 -3.02182108e-01 -2.96949625e-01
-1.39518607e+00 -9.09814954e-01 -5.50223351e-01 1.14260435e-01
4.81278688e-01 -1.12688088e+00 -7.13564873e-01 2.00196758e-01
-1.28851831e+00 -1.96573764e-01 -5.13513386e-02 4.73580122e-01
-4.56435978e-01 1.63642913e-02 -4.12130684e-01 -4.00009543e-01
-7.24233687e-01 -8.49479556e-01 1.18868220e+00 6.75774291e-02
-9.62771952e-01 -1.00226152e+00 5.52182078e-01 5.91676533e-01
4.30502951e-01 1.55233517e-01 6.29911363e-01 -1.40967858e+00
-1.63310274e-01 -5.18260598e-01 -6.79690167e-02 7.05406535e-03
-2.19225883e-01 8.72798264e-02 -8.11095536e-01 -3.11160237e-01
-3.16060096e-01 -5.01301885e-01 8.38209927e-01 3.00039142e-01
4.49608237e-01 -4.09166425e-01 -5.53909063e-01 3.94813448e-01
7.73493230e-01 -3.94595683e-01 8.74886066e-02 1.00190711e+00
2.86284953e-01 9.16464210e-01 1.10532999e+00 3.46629053e-01
5.80083132e-01 1.13568711e+00 1.97764814e-01 -9.89110791e-04
-2.63700187e-02 -4.30176646e-01 3.56328815e-01 3.44178617e-01
9.58245546e-02 -3.22265595e-01 -1.08772695e+00 8.04187477e-01
-2.15779972e+00 -7.51614213e-01 -3.40439677e-01 2.14435673e+00
9.72344339e-01 6.65082693e-01 2.88238823e-01 -5.47146089e-02
3.93382877e-01 2.19998509e-01 4.55432124e-02 -5.73106408e-01
-2.01280087e-01 3.18297386e-01 3.08830410e-01 7.27347851e-01
-1.36795545e+00 1.06360495e+00 6.53751087e+00 6.50792181e-01
-8.74553502e-01 7.82520846e-02 4.19695199e-01 -4.84555602e-01
-1.35178223e-01 2.94095993e-01 -1.39963949e+00 3.10670942e-01
1.26673925e+00 -3.67800206e-01 -1.65350392e-01 7.78795958e-01
8.00275244e-04 1.96787238e-01 -1.20241749e+00 3.76405716e-01
-8.94926488e-03 -1.54758215e+00 -1.05468400e-01 1.26384860e-02
7.82744348e-01 3.81944805e-01 2.24162802e-01 4.45675045e-01
6.14803135e-01 -9.60361063e-01 9.45783913e-01 3.02121758e-01
4.30550814e-01 -5.92558861e-01 9.25275743e-01 4.30523455e-01
-1.05486906e+00 2.19855234e-01 -7.36217247e-03 -2.06561491e-01
1.54990107e-01 4.32583779e-01 -1.28927684e+00 5.97561061e-01
7.67147362e-01 8.84480417e-01 -6.98646426e-01 7.81369150e-01
-6.48976743e-01 6.18793607e-01 -6.76852107e-01 -1.80251002e-01
4.67466354e-01 5.82434177e-01 8.47764194e-01 1.28763616e+00
2.88971979e-02 -4.02306914e-01 2.25964338e-01 6.08078361e-01
-1.82880253e-01 3.22696149e-01 -1.41045421e-01 2.82771587e-01
6.55460656e-01 1.49020696e+00 -2.98152447e-01 -4.46642756e-01
-2.47755215e-01 3.84592235e-01 2.83383429e-01 -6.83262944e-02
-8.05146039e-01 -4.30492759e-01 5.00221729e-01 3.02389175e-01
1.68237761e-01 4.05892497e-03 -3.45084131e-01 -1.07980835e+00
1.50526226e-01 -9.47167099e-01 7.88173079e-01 -4.71373588e-01
-1.15370989e+00 6.26532555e-01 4.78807509e-01 -1.04256284e+00
-4.93203610e-01 -4.92722481e-01 -7.25545049e-01 1.17043984e+00
-1.22258067e+00 -1.57740998e+00 -8.27592313e-02 2.29473352e-01
4.46807683e-01 -2.48051152e-01 9.84433830e-01 1.73311219e-01
-1.40221370e-02 9.13770735e-01 -1.92471921e-01 9.88015831e-02
1.29124975e+00 -1.30382061e+00 1.00398397e+00 7.30598509e-01
3.92701566e-01 1.09935617e+00 9.59827423e-01 -6.39875174e-01
-8.14206243e-01 -7.48879552e-01 1.53108191e+00 -1.15158725e+00
8.21885943e-01 -5.63281000e-01 -6.80253386e-01 6.85728967e-01
3.88694167e-01 -3.07800025e-01 6.74000323e-01 8.79306197e-01
-4.66957092e-01 1.92481354e-01 -6.48508310e-01 9.38036814e-02
5.86695850e-01 -4.12305564e-01 -1.10486865e+00 5.24494290e-01
7.32870698e-01 -8.26285660e-01 -9.55010951e-01 6.15528762e-01
7.38626838e-01 -5.40264547e-01 1.02409697e+00 -7.33700871e-01
6.19128525e-01 -2.42608815e-01 -3.28804016e-01 -8.99783015e-01
-1.05958737e-01 -6.12334847e-01 -2.53860295e-01 1.09326684e+00
1.18675220e+00 -2.72217274e-01 8.47262561e-01 4.36705619e-01
-5.16944118e-02 -9.55807149e-01 -9.12490785e-01 -3.71713817e-01
2.92612255e-01 -3.83842885e-01 3.18277031e-01 7.36250699e-01
2.85453767e-01 6.20242596e-01 -2.35529721e-01 3.87423858e-02
5.22187114e-01 3.75393659e-01 9.72705662e-01 -1.33923340e+00
-2.10595548e-01 -4.59406167e-01 -3.86732370e-02 -9.41829741e-01
8.89084637e-02 -1.02109766e+00 -2.84624130e-01 -1.64785373e+00
2.09926412e-01 -6.86357141e-01 -6.06586277e-01 5.53535640e-01
-3.58830452e-01 2.48099238e-01 8.69848132e-02 5.04381478e-01
-9.62664306e-01 9.62810442e-02 4.03341383e-01 1.09914280e-01
-2.58953214e-01 3.53521883e-01 -8.54627669e-01 7.38352358e-01
7.17630804e-01 -4.91624206e-01 -6.25262558e-02 -3.76686573e-01
8.82171810e-01 -4.69988227e-01 2.16302156e-01 -6.70957625e-01
3.69490266e-01 3.25772315e-02 2.94345289e-01 -9.67713535e-01
3.69679034e-01 -1.89051107e-01 -1.71631828e-01 4.18849617e-01
-1.00567544e+00 4.02958184e-01 3.24028701e-01 1.92346647e-01
-2.60248840e-01 -6.24141563e-03 2.03808069e-01 -5.88976666e-02
-5.01749754e-01 -1.82920665e-01 3.42383236e-01 3.31160337e-01
9.49115515e-01 1.72600180e-01 -5.71472049e-01 -3.75399917e-01
-6.35229826e-01 5.25286794e-01 3.11976373e-01 8.26073825e-01
2.51570821e-01 -1.18634439e+00 -1.11584747e+00 4.59804125e-02
2.80500144e-01 -1.33444950e-01 -2.00088412e-01 7.94286966e-01
-2.66854644e-01 8.16063881e-01 -3.30788530e-02 -4.83806849e-01
-1.54310143e+00 -1.58817306e-01 1.75999268e-03 -8.13276947e-01
-3.47340554e-01 1.15999818e+00 -2.20803633e-01 -8.32521081e-01
3.13420504e-01 -2.74791837e-01 -3.04433137e-01 4.08043891e-01
6.93658769e-01 3.31764698e-01 4.78246242e-01 -4.49490249e-01
-6.64621294e-01 4.84773010e-01 -5.90550005e-01 -6.01312459e-01
1.59566391e+00 2.02873841e-01 4.50362451e-02 3.90193909e-01
1.23855603e+00 3.98732543e-01 -8.73338521e-01 -4.64202106e-01
3.64639550e-01 -1.34726569e-01 -1.08722515e-01 -1.22430313e+00
-2.37755567e-01 6.85746491e-01 3.03748995e-01 3.14085424e-01
3.61024171e-01 2.63102323e-01 6.68194711e-01 2.88654536e-01
8.81379545e-02 -1.18111658e+00 -8.13507065e-02 8.27490270e-01
9.35650527e-01 -1.18111408e+00 4.09335226e-01 -2.71075338e-01
-5.89150012e-01 9.41852748e-01 4.72431213e-01 -1.24582067e-01
5.22541225e-01 2.24158853e-01 2.35416397e-01 -4.39613044e-01
-1.43124104e+00 4.72616017e-01 8.19714367e-01 -2.60718353e-02
9.71988082e-01 8.83209631e-02 -5.86012602e-01 6.86635733e-01
-8.74810755e-01 -4.58086461e-01 2.01480463e-01 1.00844848e+00
-3.63288492e-01 -1.40419400e+00 5.37674539e-02 1.31509468e-01
-8.70009601e-01 -2.18308374e-01 -7.68279970e-01 6.68520868e-01
-1.18132249e-01 8.17302883e-01 -3.04624140e-02 -4.60855752e-01
7.07211435e-01 4.33113389e-02 -1.10077513e-02 -7.86055088e-01
-1.22178793e+00 -1.55604199e-01 5.09742916e-01 -7.57579982e-01
-3.77608925e-01 -6.69473410e-01 -1.10093713e+00 -3.44385237e-01
-2.44825736e-01 5.70434749e-01 9.27150428e-01 7.95885324e-01
5.62537909e-01 2.25090474e-01 2.37725928e-01 -5.06679714e-01
-7.77966559e-01 -1.28335738e+00 1.19864270e-01 5.30791044e-01
1.89296886e-01 -5.08495510e-01 -3.78059715e-01 -4.91549522e-02] | [10.179322242736816, 9.152390480041504] |
6fc66565-df8a-4deb-8035-d8226bab4165 | flowtransformer-a-transformer-framework-for | 2304.14746 | null | https://arxiv.org/abs/2304.14746v1 | https://arxiv.org/pdf/2304.14746v1.pdf | FlowTransformer: A Transformer Framework for Flow-based Network Intrusion Detection Systems | This paper presents the FlowTransformer framework, a novel approach for implementing transformer-based Network Intrusion Detection Systems (NIDSs). FlowTransformer leverages the strengths of transformer models in identifying the long-term behaviour and characteristics of networks, which are often overlooked by most existing NIDSs. By capturing these complex patterns in network traffic, FlowTransformer offers a flexible and efficient tool for researchers and practitioners in the cybersecurity community who are seeking to implement NIDSs using transformer-based models. FlowTransformer allows the direct substitution of various transformer components, including the input encoding, transformer, classification head, and the evaluation of these across any flow-based network dataset. To demonstrate the effectiveness and efficiency of the FlowTransformer framework, we utilise it to provide an extensive evaluation of various common transformer architectures, such as GPT 2.0 and BERT, on three commonly used public NIDS benchmark datasets. We provide results for accuracy, model size and speed. A key finding of our evaluation is that the choice of classification head has the most significant impact on the model performance. Surprisingly, Global Average Pooling, which is commonly used in text classification, performs very poorly in the context of NIDS. In addition, we show that model size can be reduced by over 50\%, and inference and training times improved, with no loss of accuracy, by making specific choices of input encoding and classification head instead of other commonly used alternatives. | ['Marius Portmann', 'Mohanad Sarhan', 'Gayan K. Kulatilleke', 'Wai Weng Lo', 'Siamak Layeghy', 'Liam Daly Manocchio'] | 2023-04-28 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 1.73152417e-01 -5.47084391e-01 -1.53672665e-01 -2.42500275e-01
-1.15196981e-01 -9.38712180e-01 8.29263985e-01 1.38335079e-01
-3.24791789e-01 4.03239071e-01 -2.35559896e-01 -1.07020926e+00
-5.38953304e-01 -9.51865673e-01 -1.04635537e-01 -2.94075072e-01
-2.33323723e-01 6.20475829e-01 6.72822833e-01 -3.71739686e-01
4.20200855e-01 9.90680933e-01 -1.12844968e+00 2.79777199e-01
2.89115399e-01 1.37118530e+00 -7.63926864e-01 5.18055856e-01
-2.71960109e-01 1.02791524e+00 -8.60147119e-01 -8.45781922e-01
3.96887958e-01 2.31766310e-02 -9.26227093e-01 -3.79329383e-01
3.23285580e-01 -3.66361648e-01 -4.66203809e-01 6.26442432e-01
2.72218317e-01 -1.36260003e-01 4.06075031e-01 -1.68171716e+00
2.59984192e-03 4.82833236e-01 -2.77196884e-01 8.75701368e-01
3.48261267e-01 4.57245380e-01 1.07690763e+00 -3.44076753e-01
5.41835487e-01 1.43329525e+00 1.03755665e+00 1.17143385e-01
-1.41568947e+00 -9.75514114e-01 -9.82675031e-02 3.12903017e-01
-1.11593056e+00 -6.43359900e-01 4.52440947e-01 -1.40777051e-01
1.42324722e+00 5.10871828e-01 2.58946121e-01 1.27825844e+00
4.44155425e-01 4.30515260e-01 1.04711962e+00 -2.61888474e-01
1.94155797e-01 1.01286620e-01 4.51201499e-01 6.64061904e-01
2.40617305e-01 1.91451892e-01 -2.87935793e-01 -6.17805004e-01
4.57946867e-01 1.35138005e-01 1.46706151e-02 -1.55570701e-01
-7.38144100e-01 9.22889948e-01 2.69172072e-01 3.81391555e-01
-3.43669057e-01 -1.11319358e-02 8.34920228e-01 7.58224487e-01
4.75567251e-01 4.28684473e-01 -5.46356797e-01 -4.68462586e-01
-8.88991594e-01 3.07637274e-01 1.31137514e+00 6.48465037e-01
3.96834642e-01 4.65615653e-02 -1.39921904e-01 5.57772100e-01
1.99489370e-01 2.54013419e-01 2.46595606e-01 -5.88246942e-01
4.83235151e-01 8.30073118e-01 -4.09963429e-01 -1.05268896e+00
-4.22556460e-01 -5.35578132e-01 -5.37147760e-01 -8.37708786e-02
4.71313179e-01 -6.37996867e-02 -9.65683281e-01 1.52183473e+00
2.13586494e-01 1.97817340e-01 -1.95279688e-01 1.03186823e-01
2.45147049e-01 2.52317518e-01 3.28211010e-01 9.86258611e-02
1.49484348e+00 -4.20496613e-01 -2.75161833e-01 -2.29698494e-01
7.88953900e-01 -7.85965919e-01 7.02353716e-01 3.07172745e-01
-7.13295162e-01 1.32010460e-01 -9.75945294e-01 2.98526973e-01
-8.27252746e-01 -6.61042273e-01 6.99103415e-01 1.16202414e+00
-9.95223284e-01 8.24306488e-01 -8.30092251e-01 -7.48610675e-01
5.99138319e-01 4.77528065e-01 -3.47889334e-01 -5.32196350e-02
-1.28016305e+00 9.88460183e-01 2.71636009e-01 -3.02092046e-01
-6.16069257e-01 -1.11787581e+00 -5.03367960e-01 4.20905948e-01
3.88532430e-01 -4.14465040e-01 1.39489758e+00 -3.82628649e-01
-1.13541353e+00 3.23844403e-01 2.57785264e-02 -8.50079656e-01
3.74782741e-01 2.82105327e-01 -6.00506544e-01 8.96993279e-02
-9.70588848e-02 -2.60857623e-02 6.85563684e-01 -5.74625850e-01
-5.79098463e-01 -3.68501872e-01 1.80598453e-01 -3.19640428e-01
-7.19721079e-01 4.94323730e-01 -8.50193202e-02 -3.92399073e-01
-3.94124657e-01 -5.63719034e-01 -1.16812855e-01 -1.85716301e-01
-5.44805527e-01 -2.30339006e-01 1.45895135e+00 -4.03730184e-01
1.36176896e+00 -1.92612672e+00 -4.31128979e-01 8.35976243e-01
3.32186222e-01 7.52889335e-01 -2.46642195e-02 7.60218680e-01
-1.78913429e-01 5.37420869e-01 -2.20323391e-02 -2.06266955e-01
2.00752318e-01 2.69374967e-01 -6.20827496e-01 2.34951720e-01
1.17311254e-01 7.51312435e-01 -5.94999015e-01 -2.30025470e-01
3.61624062e-01 5.20422339e-01 -4.99589980e-01 -1.80720538e-02
3.49198766e-02 -1.75866082e-01 -4.73501414e-01 7.66181469e-01
3.95841360e-01 -2.90239513e-01 2.34952882e-01 -2.47191608e-01
6.10860661e-02 8.36693704e-01 -8.48403513e-01 8.63619030e-01
-3.99523467e-01 6.22487307e-01 -2.93177329e-02 -9.30742860e-01
8.53471279e-01 4.56533641e-01 5.44447124e-01 -9.26085591e-01
3.19677949e-01 8.35751519e-02 1.45579532e-01 -4.50269952e-02
2.13587716e-01 -2.34435469e-01 1.11146264e-01 1.04620850e+00
1.64949164e-01 2.98062325e-01 4.79497939e-01 3.50386381e-01
1.94194186e+00 -6.80435717e-01 2.30725810e-01 -1.97714552e-01
5.22861898e-01 -3.25858369e-02 3.01173568e-01 8.12922478e-01
-3.21832865e-01 -1.11499108e-01 9.70137417e-01 -5.34787774e-01
-8.38312447e-01 -1.02520943e+00 -2.46902019e-01 9.88389611e-01
-5.21164060e-01 -8.67079318e-01 -6.62501872e-01 -1.14316320e+00
1.08052194e-01 6.06241286e-01 -5.30138731e-01 -3.48197669e-01
-4.97586101e-01 -1.08307028e+00 1.22324944e+00 3.78579021e-01
6.30159199e-01 -1.01071370e+00 -5.53389192e-01 3.01463842e-01
-8.42222050e-02 -1.40171373e+00 -1.34807810e-01 3.36730599e-01
-8.30525935e-01 -1.46798182e+00 2.20327735e-01 -1.66157544e-01
1.90380976e-01 7.36559480e-02 1.12967896e+00 3.13621253e-01
-5.12596846e-01 3.13072830e-01 -3.34812671e-01 -4.06596303e-01
-3.75041366e-01 4.78042901e-01 -1.13448992e-01 -1.61789060e-01
7.02704906e-01 -9.51093376e-01 -3.01505536e-01 4.99777764e-01
-9.99856353e-01 -7.12766826e-01 3.77824128e-01 5.51861227e-01
-2.70598859e-01 4.40445632e-01 4.47574854e-01 -1.16932631e+00
9.98956680e-01 -7.85637796e-01 -5.10635614e-01 1.46792561e-01
-1.01885223e+00 -7.09676594e-02 9.07155335e-01 -3.20914179e-01
-5.49315035e-01 -7.33379900e-01 -2.53282875e-01 -3.74095261e-01
-8.25490728e-02 4.37608689e-01 1.64596888e-03 -4.42242563e-01
7.60581732e-01 1.09990902e-01 1.37753800e-01 -4.67859268e-01
-2.70807475e-01 6.78053379e-01 8.28933716e-02 -5.29293895e-01
8.94229829e-01 3.12913418e-01 8.49548802e-02 -6.97156668e-01
-3.76748830e-01 -4.00078923e-01 -2.55345583e-01 1.75791755e-01
1.21770896e-01 -9.28844884e-02 -1.04314232e+00 5.23382306e-01
-1.01643240e+00 -1.75695151e-01 -2.07249463e-01 -7.91476518e-02
-9.76097658e-02 3.10374320e-01 -8.92075121e-01 -5.01844764e-01
-7.97928691e-01 -1.18508470e+00 5.03959417e-01 -1.46044508e-01
-3.87793303e-01 -1.28605056e+00 -5.02692945e-02 2.25126281e-01
9.75585520e-01 1.05831183e-01 1.20143700e+00 -1.34408605e+00
-3.84615004e-01 -5.25344133e-01 -4.74427789e-01 3.38656008e-01
1.19587712e-01 1.02778226e-01 -9.23858464e-01 -3.85372698e-01
-4.63136509e-02 1.60791893e-02 4.90133286e-01 -2.39455864e-01
9.58312809e-01 -4.88287836e-01 -3.78022224e-01 6.26179576e-01
1.34905052e+00 4.91276294e-01 7.43772626e-01 5.23555756e-01
4.81438190e-01 3.71890694e-01 -1.98422950e-02 3.70080084e-01
2.21516505e-01 6.42366171e-01 3.78693551e-01 9.33297575e-02
1.53470948e-01 -1.30236700e-01 3.71571392e-01 2.54224390e-01
3.08838755e-01 -3.62879813e-01 -1.14631557e+00 2.31126800e-01
-1.35577929e+00 -1.19436896e+00 1.83622092e-01 2.00276613e+00
5.48496187e-01 6.24479234e-01 3.75319332e-01 7.00506687e-01
3.51455092e-01 8.99125636e-02 -4.49520856e-01 -8.02063406e-01
3.62744659e-01 8.68271470e-01 7.59328306e-01 2.55027622e-01
-9.73470449e-01 8.95830035e-01 7.01839018e+00 8.78114641e-01
-1.21655715e+00 -6.50289506e-02 3.02951127e-01 3.22959796e-02
1.11333318e-01 9.28024426e-02 -8.12897563e-01 4.63734001e-01
1.72145164e+00 -3.79912764e-01 6.17238462e-01 4.69681770e-01
1.33961365e-01 3.62194270e-01 -1.08335102e+00 5.30632019e-01
-2.91670561e-01 -1.34023595e+00 1.74814522e-01 2.69025266e-01
-1.59279346e-01 1.43817663e-01 4.71542478e-02 3.13570976e-01
6.06490135e-01 -1.01422894e+00 3.26965243e-01 -3.41929332e-03
4.00878549e-01 -9.19184864e-01 8.34243000e-01 -4.94057685e-02
-1.10588503e+00 -3.98913324e-01 1.14686057e-01 7.48672858e-02
-1.16546400e-01 5.70526123e-01 -1.16916823e+00 4.99623924e-01
7.87854135e-01 7.03520298e-01 -7.16546834e-01 1.02149904e+00
5.81630394e-02 9.14830327e-01 -5.77076375e-01 2.03157574e-01
3.84388626e-01 2.65782833e-01 6.07198000e-01 1.47745788e+00
3.77843417e-02 -1.90767825e-01 4.19603661e-02 5.52504480e-01
-1.27310872e-01 -1.49586737e-01 -6.16470337e-01 -3.15142930e-01
8.71365726e-01 1.34585106e+00 -7.97557533e-01 -2.75549978e-01
-2.11896807e-01 3.83582979e-01 -1.71422902e-02 3.09065193e-01
-7.09527433e-01 -7.52027452e-01 1.05898356e+00 5.55320621e-01
4.30041313e-01 5.79339191e-02 -2.40239844e-01 -9.12959278e-01
-6.33021146e-02 -1.28243792e+00 7.48398364e-01 -2.05439925e-01
-1.39419436e+00 8.27911556e-01 2.29913771e-01 -7.14016438e-01
-4.95767713e-01 -8.31866324e-01 -9.50963140e-01 7.75067806e-01
-1.35544419e+00 -1.06933475e+00 4.39730845e-02 9.22486603e-01
8.01296383e-02 -2.22891361e-01 9.33429122e-01 6.27314091e-01
-9.36892867e-01 9.47734594e-01 -2.13542983e-01 4.30525541e-01
4.30049211e-01 -1.10307574e+00 9.16562259e-01 8.84289801e-01
1.41224768e-02 8.81430984e-01 6.64499223e-01 -3.95954669e-01
-1.15790260e+00 -1.05820656e+00 9.12603676e-01 -3.86012286e-01
1.15111732e+00 -6.09621763e-01 -7.63610125e-01 9.93437290e-01
-6.56161159e-02 4.00138125e-02 7.33811140e-01 2.95200676e-01
-9.65008974e-01 -3.68233085e-01 -1.67763364e+00 4.90336508e-01
1.05864358e+00 -7.84395933e-01 -3.59178454e-01 2.66544551e-01
4.66829062e-01 6.53558597e-02 -1.10083199e+00 2.70792127e-01
5.26992917e-01 -1.09799445e+00 1.13034368e+00 -8.19855750e-01
-1.14653766e-01 1.72695160e-01 -2.75753438e-02 -1.15897346e+00
-2.84960061e-01 -6.72893584e-01 -1.94608957e-01 1.36792243e+00
2.79304206e-01 -1.41908836e+00 8.76800120e-01 5.64053357e-01
3.08925718e-01 -6.67582810e-01 -9.88212407e-01 -7.80738831e-01
-8.82204250e-03 -6.70602620e-01 9.68678355e-01 9.36675608e-01
1.20675765e-01 3.16920698e-01 9.34694931e-02 -6.94043264e-02
7.50600100e-01 -3.29279423e-01 6.00570500e-01 -1.54390550e+00
-1.78995356e-01 -6.23247087e-01 -8.81770730e-01 -3.93879682e-01
6.24124892e-02 -8.48589063e-01 -7.23496199e-01 -1.04930401e+00
-2.24320710e-01 -7.55107045e-01 -5.17099261e-01 1.09318984e+00
4.10182297e-01 3.60403955e-01 3.43545049e-01 1.85312748e-01
-3.04901659e-01 2.53400020e-03 6.42706096e-01 -7.27091059e-02
8.25171731e-03 9.98129696e-02 -8.19513023e-01 5.32720149e-01
1.02053189e+00 -5.30068457e-01 -6.01476073e-01 -1.09675556e-01
-1.45308077e-01 -2.59004146e-01 4.73845929e-01 -1.04908764e+00
3.97575021e-01 -1.74760949e-02 3.05454493e-01 -1.24914102e-01
2.93983996e-01 -8.17688286e-01 1.77896693e-02 5.35127163e-01
-2.66818497e-02 6.85441911e-01 5.32469332e-01 2.26659611e-01
9.14784372e-02 3.95279713e-02 6.46458864e-01 3.40303704e-02
-3.31868768e-01 4.77583021e-01 -5.91568708e-01 1.63052768e-01
8.85676980e-01 -1.90728247e-01 -7.06320286e-01 -1.18971884e-01
-2.14023471e-01 2.05868725e-02 2.07668200e-01 5.04795015e-01
5.48572063e-01 -8.82539093e-01 -3.35293919e-01 6.60293341e-01
-1.09075472e-01 -7.75224924e-01 -2.06720963e-01 8.71742010e-01
-4.15361047e-01 7.10728347e-01 -3.81419033e-01 -3.46082777e-01
-1.31784022e+00 3.99903446e-01 5.47104359e-01 -8.31791162e-01
-5.25555551e-01 3.51998299e-01 -3.66316766e-01 -5.84670901e-01
1.25745684e-01 -1.20815665e-01 -1.83183011e-02 2.76555046e-02
7.57715404e-01 5.03212810e-01 6.10280275e-01 -3.16747129e-01
-6.48926377e-01 -3.15767415e-02 -6.97482228e-01 2.50898972e-02
1.27318978e+00 3.04480791e-01 -2.34381199e-01 -1.01790003e-01
1.24104238e+00 -3.75787944e-01 -4.70433146e-01 -2.61307478e-01
2.77311623e-01 -4.12045598e-01 2.25653410e-01 -9.62573946e-01
-1.25432706e+00 7.97372520e-01 4.02668297e-01 6.64866030e-01
1.14265013e+00 -5.65922916e-01 9.73415434e-01 3.66765529e-01
4.62034047e-01 -4.23897266e-01 -2.90076077e-01 7.92252421e-01
1.45018041e-01 -5.78877151e-01 -1.35422602e-01 -3.89653355e-01
-2.46189460e-02 1.09369361e+00 3.36810499e-01 -1.06811292e-01
8.64501119e-01 5.43785751e-01 -1.07959315e-01 -5.30636132e-01
-1.11611187e+00 2.10280523e-01 -2.04292819e-01 5.80473006e-01
1.76586509e-01 -2.27426752e-01 -1.34710120e-02 1.05220385e-01
-3.51790607e-01 -3.34371366e-02 3.67890447e-01 1.22544515e+00
-9.81571749e-02 -1.57916451e+00 -2.94261992e-01 8.57422888e-01
-7.32248843e-01 -2.07366228e-01 -4.61953521e-01 8.71451557e-01
-2.58870333e-01 1.16675580e+00 1.23594113e-01 -7.94875145e-01
5.41611552e-01 3.54584754e-01 2.09016755e-01 -3.47034365e-01
-1.23426163e+00 -6.59267604e-01 3.06245238e-01 -8.00837636e-01
1.93632483e-01 -5.14472723e-01 -7.60920227e-01 -1.29033017e+00
-2.25840151e-01 4.34204608e-01 5.38714290e-01 1.04610813e+00
5.47649443e-01 5.63383102e-01 5.98505139e-01 -3.40519756e-01
-8.49982142e-01 -8.31594169e-01 -3.25718433e-01 1.74446672e-01
1.96249530e-01 -8.06503892e-01 -6.57849669e-01 -5.24603307e-01] | [5.325231552124023, 7.3015947341918945] |
03a4a99e-6fe3-4857-84e7-998bf3768792 | a-method-for-studying-semantic-construal-in | 2305.18598 | null | https://arxiv.org/abs/2305.18598v1 | https://arxiv.org/pdf/2305.18598v1.pdf | A Method for Studying Semantic Construal in Grammatical Constructions with Interpretable Contextual Embedding Spaces | We study semantic construal in grammatical constructions using large language models. First, we project contextual word embeddings into three interpretable semantic spaces, each defined by a different set of psycholinguistic feature norms. We validate these interpretable spaces and then use them to automatically derive semantic characterizations of lexical items in two grammatical constructions: nouns in subject or object position within the same sentence, and the AANN construction (e.g., `a beautiful three days'). We show that a word in subject position is interpreted as more agentive than the very same word in object position, and that the nouns in the AANN construction are interpreted as more measurement-like than when in the canonical alternation. Our method can probe the distributional meaning of syntactic constructions at a templatic level, abstracted away from specific lexemes. | ['Katrin Erk', 'Kyle Mahowald', 'Gabriella Chronis'] | 2023-05-29 | null | null | null | null | ['word-embeddings'] | ['methodology'] | [-1.82646289e-01 3.57402295e-01 -3.63415759e-03 -8.40937376e-01
7.08924159e-02 -9.73742783e-01 5.94463646e-01 7.01560080e-01
-7.16035306e-01 4.95138735e-01 9.92066503e-01 -5.68671107e-01
-6.43810630e-02 -1.00230539e+00 -5.47827721e-01 -7.25041747e-01
3.57475691e-02 3.55083615e-01 -1.91261157e-01 -6.32010102e-01
1.91732585e-01 2.19139144e-01 -1.24433506e+00 2.04169616e-01
4.81267184e-01 6.34306669e-01 3.61100733e-01 1.75630525e-01
-4.05258268e-01 4.78556365e-01 -6.03295505e-01 -4.89459395e-01
4.94925156e-02 -4.89419192e-01 -9.60927069e-01 -1.03331925e-02
4.35892045e-01 2.94959784e-01 2.68504947e-01 1.24151349e+00
4.44873907e-02 9.22831893e-02 5.30315995e-01 -7.94635355e-01
-1.65120137e+00 1.13687062e+00 -4.60050181e-02 2.36682758e-01
5.78259468e-01 -2.25960925e-01 1.87483418e+00 -9.28385854e-01
6.90169036e-01 1.90429258e+00 3.60324681e-01 6.00164354e-01
-1.71608865e+00 -1.56565934e-01 4.38297957e-01 5.57149053e-02
-1.13395369e+00 -5.07707195e-03 7.09392130e-01 -4.52955663e-01
9.10420656e-01 2.67775893e-01 7.64043272e-01 1.12993252e+00
4.72586960e-01 1.77197948e-01 1.33442652e+00 -6.51410162e-01
4.02967751e-01 -5.46780322e-03 7.81233728e-01 2.98981130e-01
5.81527233e-01 1.44957542e-01 -1.17864564e-01 -1.24049075e-01
4.55541998e-01 -1.56375751e-01 -1.26892000e-01 -8.99348855e-02
-1.36252832e+00 1.21959317e+00 3.99279773e-01 8.12703907e-01
-3.71134937e-01 7.28367686e-01 5.07069290e-01 3.99304956e-01
4.08177733e-01 7.20719218e-01 -6.19198680e-01 1.98495328e-01
1.27555579e-01 5.97098768e-01 6.85551524e-01 7.16339171e-01
8.54769886e-01 -7.69289136e-02 1.66409507e-01 8.68489087e-01
5.81886172e-01 3.64017814e-01 4.33607936e-01 -9.75925863e-01
-5.19880354e-02 3.51173908e-01 7.61772841e-02 -9.07549322e-01
-5.30489862e-01 2.54905373e-01 4.06810045e-02 -6.92882165e-02
3.08331758e-01 -2.96958871e-02 -4.61501896e-01 2.48261809e+00
3.13398153e-01 -2.57620096e-01 3.22390169e-01 9.10493374e-01
7.71120667e-01 6.09681964e-01 1.02984393e+00 -1.67363510e-01
2.09386802e+00 -7.90791437e-02 -7.92573929e-01 -3.90578419e-01
1.14927638e+00 -5.74825764e-01 1.76317191e+00 -2.73874164e-01
-9.30342197e-01 -4.86032635e-01 -1.00062537e+00 -6.26743495e-01
-5.96934557e-01 -4.63214457e-01 8.97950113e-01 6.95559859e-01
-9.39219594e-01 4.40661669e-01 -4.52144355e-01 -4.83285427e-01
-3.75334136e-02 -1.54629216e-01 -5.76671243e-01 3.25971484e-01
-1.55570424e+00 1.16464388e+00 6.63267612e-01 1.02129765e-02
-4.29852396e-01 -5.39876223e-01 -1.36868799e+00 1.27534777e-01
3.78703028e-02 -2.87629336e-01 1.05921769e+00 -1.13657367e+00
-9.27518427e-01 1.56222928e+00 -2.64254678e-02 -3.27373177e-01
-4.59112287e-01 -1.98406175e-01 -7.51767099e-01 -4.57104355e-01
5.16469300e-01 4.18799758e-01 4.60739762e-01 -1.16998160e+00
-3.01714092e-01 -6.10675037e-01 6.71893120e-01 2.82099247e-01
-1.40327886e-01 5.25591135e-01 6.77055895e-01 -6.51851356e-01
5.06887972e-01 -8.25386345e-01 -3.86737958e-02 -1.39969185e-01
-2.98968762e-01 -9.57482994e-01 4.46134686e-01 -2.49409482e-01
1.08936977e+00 -2.44900823e+00 1.30004659e-01 -2.64365356e-02
3.05876285e-01 -3.92203867e-01 -5.53510562e-02 5.36043882e-01
-6.76852167e-01 4.77772206e-01 -1.52223796e-01 1.71864331e-01
6.37402236e-01 9.20159459e-01 -2.73478478e-01 7.82864273e-01
1.52982160e-01 9.60263789e-01 -1.10472023e+00 -2.96070367e-01
1.46250218e-01 -5.88024706e-02 -8.58334601e-01 -3.38359177e-02
-2.20617220e-01 -1.06260434e-01 -2.74566382e-01 1.96774870e-01
5.21251380e-01 2.63671935e-01 7.63478160e-01 -2.83625871e-01
-2.52549022e-01 1.00463593e+00 -1.03830993e+00 1.38819003e+00
-6.63099945e-01 4.02970791e-01 -1.41651243e-01 -9.79788005e-01
7.93188393e-01 3.51830125e-01 -4.59073246e-01 -7.29130626e-01
2.73333937e-01 2.62464225e-01 7.03096032e-01 -6.16519809e-01
6.27444148e-01 -9.68909025e-01 -8.39789152e-01 6.63290799e-01
7.85023626e-03 -4.40599144e-01 1.34012759e-01 7.53570423e-02
6.44622266e-01 1.25616029e-01 7.82105744e-01 -1.29012930e+00
4.41568702e-01 -6.50193334e-01 9.35501575e-01 2.23637789e-01
-2.60092735e-01 7.08142994e-03 9.56066728e-01 -7.83708453e-01
-1.09786928e+00 -1.87388325e+00 -7.80433357e-01 1.52356136e+00
2.55289018e-01 -5.76178074e-01 -6.36943698e-01 -6.55190349e-01
2.63857484e-01 1.59289193e+00 -1.04045415e+00 -2.35311300e-01
-8.25921297e-01 -5.05971193e-01 2.95549601e-01 7.97587752e-01
-3.33220750e-01 -1.36962640e+00 -8.61852288e-01 2.08840981e-01
1.09306432e-03 -8.71365368e-01 -3.49353850e-01 3.68506759e-01
-3.48438799e-01 -6.85561776e-01 5.49606383e-01 -8.31669867e-01
5.60907185e-01 -3.04250509e-01 1.56085885e+00 2.82585829e-01
-9.62669402e-02 2.05007732e-01 -4.42636192e-01 -9.15004969e-01
-7.00205207e-01 -6.60306394e-01 4.61297065e-01 -1.62305012e-01
9.87350464e-01 -5.18229842e-01 -2.08024040e-01 -1.39220595e-01
-9.02808130e-01 -4.11551893e-01 -3.50302011e-01 7.03711331e-01
4.74551082e-01 -6.61853194e-01 4.33850497e-01 -1.00021505e+00
9.33325887e-01 -7.28780329e-01 -2.77674288e-01 2.86795497e-02
-2.48208195e-01 4.75090116e-01 4.56875622e-01 -3.06875765e-01
-9.34558272e-01 -6.86367035e-01 -2.68988490e-01 3.95668030e-01
-2.52111167e-01 5.70652425e-01 -5.18171847e-01 7.76431262e-01
8.86119604e-01 -2.33870342e-01 -7.65912235e-02 -2.83962101e-01
8.80159080e-01 3.46189171e-01 3.56096029e-01 -1.26492441e+00
5.73392868e-01 4.03470397e-01 -8.20351914e-02 -9.08055902e-01
-1.24227703e+00 2.33464893e-02 -6.81624770e-01 2.11577073e-01
1.46655941e+00 -6.57216668e-01 -6.65132105e-01 -2.03756675e-01
-1.32345939e+00 2.06803396e-01 -7.99204588e-01 5.01737952e-01
-6.65980101e-01 2.37973243e-01 -5.50674081e-01 -4.37220216e-01
4.49654870e-02 -7.73729682e-01 1.03021336e+00 -3.00240427e-01
-1.00738978e+00 -1.45031166e+00 6.15868755e-02 -3.26237053e-01
-1.86054066e-01 2.87927091e-01 1.61949217e+00 -9.65898573e-01
-2.51470897e-02 2.80343086e-01 -9.95203108e-02 5.34123302e-01
5.21629274e-01 -1.80963039e-01 -7.85036147e-01 1.18401513e-01
5.77419698e-01 -3.17275554e-01 4.66199815e-01 2.11327091e-01
8.51140440e-01 -3.84133548e-01 1.84762701e-02 4.31518793e-01
1.58742642e+00 1.99594617e-01 7.25242317e-01 1.50780261e-01
5.91204047e-01 7.66228437e-01 4.95211303e-01 -1.08974926e-01
3.28274667e-01 3.37256163e-01 1.51428252e-01 4.13012236e-01
1.83238193e-01 -1.40028849e-01 4.11788672e-01 7.53512740e-01
1.52017072e-01 -1.73776835e-01 -7.63568044e-01 9.47455704e-01
-1.32406664e+00 -9.05251622e-01 -4.89988804e-01 1.83429050e+00
1.08110690e+00 -1.13275964e-02 -4.91425991e-01 8.85325018e-03
5.41955590e-01 6.86250210e-01 2.40285799e-01 -1.36922228e+00
-2.56689906e-01 7.74480462e-01 -8.94086212e-02 7.50109673e-01
-7.86176980e-01 1.35873854e+00 6.24903917e+00 2.78251886e-01
-5.14878392e-01 3.72744471e-01 1.86678469e-01 2.15255380e-01
-1.08332121e+00 4.01138306e-01 -5.10558248e-01 3.84136438e-01
1.16739345e+00 -3.04336131e-01 -4.47740257e-02 8.62781107e-01
2.61331081e-01 -1.13544352e-01 -1.95537221e+00 5.86392462e-01
-1.03224717e-01 -1.20913231e+00 5.52452095e-02 9.73083917e-03
1.48855627e-01 -5.54591656e-01 -1.94433570e-01 -2.13934146e-02
4.11530167e-01 -1.12751126e+00 1.37247407e+00 -4.70505618e-02
4.37432319e-01 -6.53958559e-01 5.57961166e-01 -1.98272526e-01
-9.67401862e-01 -3.03393640e-02 -7.63387680e-01 -7.38786221e-01
3.29509497e-01 3.87060195e-01 -1.71462476e-01 -2.55392920e-02
5.15097260e-01 3.84506494e-01 -9.71964002e-02 -1.60653591e-01
-8.13687503e-01 6.06833756e-01 4.54228856e-02 -4.42260921e-01
3.97604734e-01 -6.44415677e-01 6.40041173e-01 1.19384909e+00
-5.02556469e-03 5.08090079e-01 5.13066258e-03 1.45998693e+00
2.74784006e-02 3.55868697e-01 -7.49718070e-01 -6.63575297e-03
6.75825953e-01 1.05155110e+00 -4.46840703e-01 -3.43585640e-01
-4.59944606e-01 7.15846658e-01 6.56801105e-01 3.58716659e-02
-9.52265322e-01 -1.67041138e-01 1.50644386e+00 1.06911793e-01
-2.10224800e-02 -3.81468505e-01 -2.30102494e-01 -8.77425969e-01
9.17117298e-02 -2.83484519e-01 1.85732722e-01 -6.27672791e-01
-1.53529596e+00 2.82370925e-01 1.81694508e-01 -5.66664457e-01
-4.47920933e-02 -1.02835715e+00 -6.80124640e-01 1.10130513e+00
-7.70670891e-01 -8.50246549e-01 3.73925656e-01 1.03332162e-01
3.52157384e-01 4.92053628e-01 1.35632503e+00 -2.34301940e-01
-1.45468548e-01 3.77037108e-01 -2.02933952e-01 1.77236676e-01
2.47180715e-01 -1.62570536e+00 6.40350223e-01 5.97256303e-01
3.77382427e-01 1.19549024e+00 9.44274724e-01 -4.18492168e-01
-9.75843370e-01 -7.67872989e-01 1.58820260e+00 -1.00315237e+00
1.11782038e+00 -8.47636700e-01 -1.07234490e+00 1.20100236e+00
3.90316248e-01 1.05841927e-01 9.75784659e-01 7.03923106e-01
-8.30812633e-01 2.21403733e-01 -1.13475096e+00 9.46982622e-01
1.31290698e+00 -8.81782830e-01 -1.77923048e+00 4.47076738e-01
1.47804129e+00 -7.60785639e-02 -7.64437437e-01 -1.78040713e-01
3.48922491e-01 -8.61965835e-01 9.23312366e-01 -1.19305885e+00
5.00412166e-01 -1.94394261e-01 -6.55938685e-01 -1.60652590e+00
-6.27343476e-01 5.93209565e-02 5.57683766e-01 9.07209516e-01
3.71010184e-01 -9.17092204e-01 5.24544530e-02 5.85395336e-01
-4.40913707e-01 -3.57754201e-01 -1.12727964e+00 -9.65012133e-01
8.27080965e-01 -6.46393836e-01 8.82202864e-01 9.91159618e-01
4.22761858e-01 6.76060796e-01 3.66156965e-01 -9.84413922e-02
4.73033398e-01 2.60177255e-01 -1.55470192e-01 -1.41658485e+00
-9.25422013e-02 -1.43488511e-01 -7.56967545e-01 -5.62527001e-01
8.28590751e-01 -1.35566223e+00 -7.17269480e-02 -1.21497071e+00
1.50751676e-02 -3.80476594e-01 -2.20991850e-01 2.62886822e-01
-1.78257063e-01 -1.13466620e-01 2.33579159e-01 -4.84067589e-01
3.50903533e-02 5.79379022e-01 9.79807079e-01 2.28010386e-01
9.16391313e-02 -8.71218443e-01 -1.10328150e+00 1.18795681e+00
6.13117874e-01 -5.50534308e-01 -3.73756766e-01 -6.13695741e-01
5.78009129e-01 -3.49855930e-01 5.01873076e-01 -3.91113609e-01
-7.64151633e-01 -6.07324421e-01 1.16369069e-01 1.56553477e-01
1.19388970e-02 -8.10296297e-01 -1.41327024e-01 3.46926957e-01
-5.36905468e-01 4.71365899e-01 1.24112047e-01 3.41699839e-01
-6.39110208e-02 -4.00584489e-01 9.17946517e-01 -2.24875465e-01
-9.44858670e-01 -3.40813041e-01 -4.16015089e-01 5.97902060e-01
1.02136195e+00 -2.49410197e-01 -2.31515035e-01 1.93388239e-01
-8.63082886e-01 -1.75065383e-01 4.42223400e-01 7.72574186e-01
5.04779577e-01 -1.61083281e+00 -7.18731344e-01 8.68982598e-02
4.35894787e-01 -5.76653183e-01 -5.35285054e-03 4.33047622e-01
-5.48964679e-01 9.00049787e-03 -2.01823071e-01 -2.41109148e-01
-8.77499044e-01 9.08587098e-01 2.40206286e-01 5.55186450e-01
-5.97366214e-01 1.02053893e+00 8.96414697e-01 -7.55489767e-01
-6.31682754e-01 -9.33449686e-01 3.58593389e-02 1.43638179e-01
4.05901641e-01 -1.48663297e-01 -3.26160014e-01 -1.05533981e+00
-4.91177022e-01 4.26327288e-01 1.07410364e-01 -3.63651738e-02
1.10060620e+00 -2.22488850e-01 -8.93468320e-01 1.06427455e+00
1.52742386e+00 2.92600542e-01 -3.79212737e-01 5.14087118e-02
3.46314788e-01 -3.95447761e-01 -3.94833565e-01 -3.71424854e-01
-3.35486501e-01 5.14175296e-01 2.57692367e-01 5.73618710e-01
4.13659751e-01 8.18111181e-01 4.55278993e-01 1.68002754e-01
3.97013187e-01 -1.35352850e+00 -3.58250976e-01 8.15168202e-01
1.24194682e+00 -7.04286277e-01 -1.71913818e-01 -6.05612099e-01
-4.69742894e-01 8.61467361e-01 4.89774734e-01 -5.35563469e-01
6.97157383e-01 1.10202476e-01 1.15585998e-01 -6.72493815e-01
-7.13420153e-01 -1.43703848e-01 -2.83602369e-03 4.73391265e-01
1.08820856e+00 8.12239707e-01 -1.27757132e+00 6.06485069e-01
-8.77420247e-01 -9.46431398e-01 8.16454172e-01 7.43125618e-01
-4.74650294e-01 -1.13037324e+00 -2.62944669e-01 3.37060064e-01
-4.44148600e-01 -5.68459213e-01 -2.99653441e-01 1.05510366e+00
4.53464746e-01 7.19106495e-01 7.64394581e-01 -2.88052168e-02
3.57624352e-01 3.42981160e-01 7.39188254e-01 -1.16281438e+00
-4.18492675e-01 -2.74778515e-01 4.66123611e-01 -7.01274097e-01
-3.97427648e-01 -9.08864737e-01 -1.88894367e+00 -2.19621092e-01
-1.06457219e-01 2.58960962e-01 3.63996565e-01 1.08415163e+00
-2.02189058e-01 1.29220545e-01 2.55910367e-01 -3.21123064e-01
-5.96663713e-01 -8.68747950e-01 -1.06176066e+00 1.05457938e+00
3.75362299e-02 -7.77785718e-01 -5.87323844e-01 -1.21852100e-01] | [10.362661361694336, 8.978385925292969] |
5197da2a-e0e6-44e6-b544-bdfc49eb6185 | causality-between-sentiment-and | 2306.05803 | null | https://arxiv.org/abs/2306.05803v1 | https://arxiv.org/pdf/2306.05803v1.pdf | Causality between Sentiment and Cryptocurrency Prices | This study investigates the relationship between narratives conveyed through microblogging platforms, namely Twitter, and the value of crypto assets. Our study provides a unique technique to build narratives about cryptocurrency by combining topic modelling of short texts with sentiment analysis. First, we used an unsupervised machine learning algorithm to discover the latent topics within the massive and noisy textual data from Twitter, and then we revealed 4-5 cryptocurrency-related narratives, including financial investment, technological advancement related to crypto, financial and political regulations, crypto assets, and media coverage. In a number of situations, we noticed a strong link between our narratives and crypto prices. Our work connects the most recent innovation in economics, Narrative Economics, to a new area of study that combines topic modelling and sentiment analysis to relate consumer behaviour to narratives. | ['Abhijeet Chandra', 'Sarwesh P', 'Began Gowsik S', 'Abinandhan S', 'Udeshya Raj', 'Lubdhak Mondal'] | 2023-06-09 | null | null | null | null | ['sentiment-analysis'] | ['natural-language-processing'] | [-4.90949690e-01 1.18554339e-01 -4.68724191e-01 2.22577363e-01
-6.30842030e-01 -9.88895416e-01 1.39586663e+00 6.22371376e-01
-1.72263622e-01 4.64186251e-01 1.14965475e+00 -4.19031411e-01
1.54729575e-01 -1.05395150e+00 -3.51654947e-01 -6.74903393e-01
-3.66735101e-01 8.89173821e-02 -1.81644619e-01 -4.87774581e-01
8.99617434e-01 -3.19060057e-01 -1.04761302e+00 4.72604841e-01
1.20575815e-01 1.06684148e+00 -8.85771662e-02 3.59643698e-01
-3.82189065e-01 1.51075494e+00 -7.73060381e-01 -8.38395357e-01
-1.68584511e-01 -3.69370818e-01 -4.96928275e-01 1.10123716e-01
-7.24669933e-01 4.15274687e-02 -2.74935782e-01 1.05488348e+00
1.33992121e-01 -5.84132016e-01 5.08642197e-01 -9.75038052e-01
-3.69943947e-01 1.29402983e+00 -5.31165242e-01 4.69468057e-01
4.40602541e-01 -1.25007510e-01 1.14665282e+00 -5.95660090e-01
1.25330937e+00 9.40952122e-01 7.07570016e-01 -3.52084577e-01
-5.98599315e-01 -7.97455907e-01 -1.71998098e-01 -1.46196529e-01
-8.90188038e-01 1.62135720e-01 9.64312673e-01 -1.04173183e+00
6.73648179e-01 2.69882083e-01 1.30937409e+00 1.29905033e+00
8.17386627e-01 6.22108102e-01 1.52875423e+00 -2.99493879e-01
3.01006496e-01 7.19827533e-01 1.85831994e-01 -1.84007972e-01
3.39930803e-01 -9.19857249e-02 -1.03406513e+00 -5.07743180e-01
5.13559468e-02 6.14457428e-02 2.19994798e-01 7.50031590e-01
-1.29604757e+00 1.63034892e+00 -3.12084705e-01 7.48797119e-01
-7.58229315e-01 -3.42014842e-02 7.42964804e-01 4.61059302e-01
1.37855470e+00 6.50600731e-01 -3.63050401e-01 -6.83513761e-01
-8.52061689e-01 4.61725235e-01 1.34341633e+00 6.24209106e-01
3.40532690e-01 -3.34571302e-01 4.04876798e-01 3.03347260e-01
5.41749954e-01 2.42053300e-01 7.41667867e-01 -3.23736370e-01
7.07962394e-01 4.63188022e-01 1.03341945e-01 -1.48935556e+00
-1.25274375e-01 -3.93668085e-01 -1.23100780e-01 -3.62224907e-01
1.21790759e-01 -6.66047871e-01 5.96692376e-02 7.77842104e-01
5.04888222e-02 -3.12815875e-01 8.83069262e-02 6.82030559e-01
6.99390888e-01 8.43758523e-01 1.55475795e-01 -5.35391271e-01
1.85058284e+00 -3.06337416e-01 -1.33843482e+00 5.41083813e-02
7.19950795e-01 -1.25972223e+00 6.11716509e-01 7.18428731e-01
-9.79628086e-01 5.12096047e-01 -1.20560944e+00 2.96807259e-01
-8.40108752e-01 -7.89614379e-01 1.01586115e+00 8.23907375e-01
-5.25641680e-01 3.90255600e-01 -4.42987710e-01 -2.28324816e-01
4.67229843e-01 -3.75804424e-01 2.61851579e-01 5.01495183e-01
-1.54058111e+00 7.50410736e-01 4.64543492e-01 -4.08232868e-01
-4.83399630e-01 -5.40176749e-01 -7.74581552e-01 -2.03388765e-01
6.24223888e-01 9.94397774e-02 1.00463200e+00 -8.11793447e-01
-1.47433496e+00 8.67946744e-01 2.33368754e-01 -7.79514790e-01
3.02244902e-01 -2.69796610e-01 -5.88086665e-01 7.32356906e-02
4.43507992e-02 -2.66069651e-01 4.52692300e-01 -7.33970463e-01
-5.60089886e-01 -1.83782756e-01 1.32909119e-01 -3.20823997e-01
-7.00155735e-01 9.10317302e-01 3.93044829e-01 -1.14078081e+00
2.12080553e-01 -4.61332589e-01 -3.01917847e-02 -1.28094757e+00
-4.89252031e-01 -4.05228376e-01 7.69452870e-01 -6.99996412e-01
1.38150811e+00 -1.96420944e+00 -3.46438706e-01 3.99154723e-02
4.37502086e-01 -7.68628061e-01 9.41468179e-01 1.23338866e+00
1.68687120e-01 8.11275482e-01 1.60972163e-01 -9.99959838e-03
1.82928681e-01 -3.11720520e-01 -7.46495306e-01 8.46890211e-01
1.44650578e-01 8.53457868e-01 -9.36481833e-01 -4.00635570e-01
-2.04449058e-01 2.93686599e-01 -3.13664347e-01 -5.11760235e-01
-4.87090439e-01 2.51155403e-02 -6.12758279e-01 6.96270823e-01
4.62869108e-01 -3.87883633e-01 5.80751598e-01 4.01501983e-01
-8.84588242e-01 9.37139630e-01 -5.61819851e-01 1.15058482e+00
-1.56392202e-01 1.38907647e+00 -4.15036112e-01 -5.99461675e-01
8.71229827e-01 5.59046149e-01 7.13720441e-01 -6.59763038e-01
6.49185181e-01 4.40100674e-03 -4.30671096e-01 -4.08095449e-01
9.58053768e-01 -5.26379883e-01 -7.03139424e-01 1.08085346e+00
-1.82266444e-01 -3.91241252e-01 -1.53580112e-02 3.72862637e-01
6.87123060e-01 -1.77481264e-01 6.30527675e-01 -7.51115501e-01
-1.23632751e-01 4.50211853e-01 2.91592658e-01 7.29418844e-02
1.15480922e-01 1.43742085e-01 1.23605669e+00 -6.15732908e-01
-1.03901529e+00 -1.56762570e-01 -4.28362489e-01 6.63050830e-01
-5.85743934e-02 -9.88443255e-01 -7.12694466e-01 -2.77149141e-01
-6.78243861e-02 7.72234976e-01 -7.93499708e-01 7.07941949e-01
-4.96646523e-01 -1.44154251e+00 4.14890870e-02 -2.62647122e-01
2.78737485e-01 -7.92560935e-01 -9.23802853e-01 5.11968732e-01
-3.28635544e-01 -1.12960804e+00 -2.65799873e-02 1.43383309e-01
-6.52083874e-01 -1.01478887e+00 -4.37227786e-01 -1.91327959e-01
1.07498929e-01 1.13841994e-02 1.01984847e+00 -3.46949369e-01
1.63019910e-01 -2.47257873e-01 -5.52503347e-01 -1.12784302e+00
-5.57880044e-01 -3.41184586e-02 -2.00440332e-01 -1.35533130e-02
5.09090900e-01 -3.46802711e-01 -4.26196545e-01 -4.73386906e-02
-8.55621815e-01 -7.71555528e-02 2.55488101e-02 3.72331381e-01
-9.80722755e-02 6.85512304e-01 3.05843949e-01 -1.23553860e+00
8.93265963e-01 -1.54462838e+00 -4.63532954e-01 -4.73090947e-01
-6.43120825e-01 -4.95571136e-01 -8.02819952e-02 -4.12654519e-01
-8.78240705e-01 -8.44571590e-01 4.47522104e-01 5.61633229e-01
3.00006628e-01 1.38098109e+00 3.76410812e-01 5.46186984e-01
3.94704461e-01 -1.23643652e-01 -4.75871451e-02 -2.47878850e-01
3.04942667e-01 6.93811774e-01 -1.11601576e-01 6.07429110e-02
5.89837849e-01 1.02436972e+00 -5.65965354e-01 -7.68382192e-01
-6.43080592e-01 -4.86084878e-01 -5.38147762e-02 -6.87965453e-01
6.94443762e-01 -1.15592051e+00 -9.81444597e-01 6.91219926e-01
-1.15913153e+00 2.12934151e-01 -3.31463873e-01 7.68428802e-01
-3.54989827e-01 1.21265672e-01 -9.48952734e-01 -1.21188557e+00
-1.24130875e-01 -6.83177888e-01 3.08538347e-01 1.43888425e-02
-7.29060650e-01 -1.41048753e+00 4.35733885e-01 4.16520447e-01
5.60868382e-01 1.04559636e+00 5.44690609e-01 -1.06529212e+00
-5.40362418e-01 -2.83242881e-01 5.42532131e-02 -2.48139605e-01
-1.31235011e-02 -8.62090010e-03 -7.49213159e-01 -4.51291315e-02
1.08991373e+00 -2.75493592e-01 5.09220362e-01 2.61827558e-01
1.30343437e-01 -9.73693132e-01 -9.63298455e-02 3.92590202e-02
1.62345672e+00 4.88137424e-01 5.69378674e-01 1.30961251e+00
-2.34648615e-01 1.11713278e+00 6.95386171e-01 1.29240775e+00
6.25013232e-01 4.05860633e-01 2.18344659e-01 3.70087862e-01
5.30306935e-01 -3.29440534e-01 6.57156706e-01 1.47945642e+00
-2.23996848e-01 -7.29266033e-02 -1.11990774e+00 7.46573389e-01
-1.52541661e+00 -9.97184753e-01 -4.85132933e-01 1.45201838e+00
5.89752555e-01 6.09185815e-01 3.75706583e-01 2.84724414e-01
5.13663054e-01 5.34027636e-01 3.07312280e-01 -2.89907187e-01
-5.85323393e-01 -3.23099387e-03 8.90327573e-01 2.61659205e-01
-7.01248646e-01 6.23499274e-01 7.01022911e+00 7.85158277e-01
-1.05704594e+00 6.11703217e-01 9.18542862e-01 -2.76266754e-01
-8.34675908e-01 2.22148448e-01 -5.70909202e-01 8.41043293e-01
1.19975340e+00 -6.57384634e-01 -3.22115272e-01 8.07224572e-01
4.91663337e-01 -5.79993665e-01 -2.61242867e-01 3.69731009e-01
3.33904207e-01 -1.87970376e+00 -3.70492876e-01 6.53324962e-01
9.89776731e-01 -1.03273690e-01 3.21091712e-01 -3.32057357e-01
4.55657423e-01 -4.73199338e-01 1.30376172e+00 2.08980381e-01
5.12764789e-02 -7.52697110e-01 1.19745052e+00 1.88087255e-01
-4.85856682e-01 1.39526697e-02 -3.53957643e-03 -5.38341165e-01
5.68188012e-01 8.46257865e-01 -8.05080235e-01 3.67352158e-01
7.11631835e-01 9.35216606e-01 1.31150370e-03 3.94368947e-01
-1.79061651e-01 9.13096726e-01 -9.21575469e-04 -5.05178452e-01
6.53742254e-01 -3.95156413e-01 7.27416277e-01 1.21055710e+00
3.05462033e-01 9.50844437e-02 -5.67008317e-01 7.46580839e-01
6.38008043e-02 3.90690207e-01 -7.06169724e-01 -6.41022742e-01
2.21186593e-01 9.70891297e-01 -1.32829154e+00 -2.16474578e-01
-5.51783442e-01 1.37386203e-01 -4.99558717e-01 3.75811160e-02
-6.20437801e-01 -1.82903811e-01 2.95141399e-01 5.97651720e-01
5.34567498e-02 -3.02763462e-01 -9.38823521e-01 -1.12306952e+00
-1.07643172e-01 -7.49335706e-01 2.13875368e-01 -2.68998861e-01
-9.66249645e-01 4.26415861e-01 1.17846896e-04 -9.26791787e-01
-2.35590160e-01 -7.06690550e-02 -8.06947887e-01 3.87012392e-01
-1.56075394e+00 -7.74873018e-01 4.07540828e-01 4.25702892e-03
3.95153850e-01 -3.69117409e-01 3.04504156e-01 -1.01357609e-01
-4.63176891e-02 -3.27206701e-01 2.15463042e-01 -9.23627391e-02
2.82267034e-01 -1.09274530e+00 8.23543012e-01 4.67867792e-01
1.18571416e-01 4.36729163e-01 1.06440294e+00 -9.62018847e-01
-1.23998475e+00 -2.95955569e-01 1.65802217e+00 -8.21600795e-01
1.70360386e+00 -7.40172327e-01 -3.96439403e-01 4.30950254e-01
7.92328775e-01 -9.60021675e-01 1.12914705e+00 2.98196644e-01
-2.67477006e-01 5.43490708e-01 -9.19612706e-01 3.26044500e-01
2.29765251e-01 -6.59074903e-01 -8.55203807e-01 6.67660832e-01
1.04499221e+00 -5.56574427e-02 -1.15837514e+00 -4.06492054e-01
6.96272016e-01 -6.57062769e-01 4.55951869e-01 -4.18829411e-01
1.09155941e+00 3.79113972e-01 -2.91603327e-01 -9.65214252e-01
1.38902456e-01 -1.41717672e+00 2.35982239e-01 1.36054003e+00
5.51286519e-01 -7.48876452e-01 9.00891244e-01 3.46088737e-01
3.98898631e-01 -5.13059556e-01 -7.82755196e-01 -3.29004824e-01
3.22439760e-01 -8.58767569e-01 7.12863863e-01 1.63862932e+00
8.13373923e-01 -2.23423913e-01 -3.28994125e-01 -4.44024414e-01
4.88348454e-01 1.27038658e-01 4.61751491e-01 -1.07770741e+00
9.71024483e-03 -4.81633604e-01 -3.56315762e-01 -4.77906287e-01
-2.78212160e-01 -4.49220061e-01 -6.43602848e-01 -8.71472299e-01
3.90818328e-01 -1.81943193e-01 3.66142113e-03 -5.85561216e-01
4.69551772e-01 1.53324425e-01 4.85871673e-01 6.93298161e-01
-3.94182801e-01 5.61291054e-02 8.88988793e-01 1.27543688e-01
-3.95692557e-01 -1.56818464e-01 -1.28248632e+00 1.03369153e+00
6.52356327e-01 -7.99998581e-01 2.77021199e-01 2.72900045e-01
1.56531107e+00 3.64867717e-01 1.28628999e-01 -2.11050034e-01
3.57283920e-01 -2.03777090e-01 -1.87745824e-01 -1.05178893e+00
-1.02912411e-01 -5.71090519e-01 3.35961670e-01 4.72521096e-01
-1.85148269e-01 2.40707442e-01 5.93002625e-02 3.71900141e-01
-7.05507755e-01 -1.31889760e-01 6.91362098e-02 -3.32890004e-01
-9.23791900e-02 -2.48307273e-01 -1.16293657e+00 1.78349361e-01
1.05389059e+00 -1.33157119e-01 -5.93073726e-01 -6.67813480e-01
-6.46088362e-01 -1.26525611e-01 2.46371299e-01 2.97876120e-01
3.34280431e-01 -1.15871119e+00 -1.08820248e+00 -3.32277834e-01
-1.58094883e-01 -5.16458809e-01 -5.66370375e-02 8.32587659e-01
-6.47519708e-01 9.02590156e-01 8.76506865e-02 -1.04166307e-01
-8.29042256e-01 3.51979107e-01 -4.39362764e-01 -4.79157507e-01
-7.17034638e-01 5.46112657e-01 -5.39498627e-02 4.81609702e-01
-2.28181228e-01 -4.56522346e-01 -6.58524454e-01 1.31496906e+00
5.92473626e-01 4.10370469e-01 -3.54764491e-01 -7.22607613e-01
-8.96300450e-02 1.38476163e-01 -1.34621620e-01 -6.31885707e-01
1.73692107e+00 -4.56851691e-01 -3.73996496e-01 1.08101296e+00
1.25205386e+00 3.22820932e-01 -7.77321219e-01 -1.75495803e-01
9.52260971e-01 -5.36441803e-01 2.89375156e-01 -5.94065547e-01
-7.50621438e-01 2.58704513e-01 -3.33425969e-01 1.26741505e+00
6.03282273e-01 3.76930416e-01 7.92825580e-01 -2.46619180e-01
3.50607187e-01 -1.45605850e+00 7.23091960e-02 3.24731559e-01
8.79509866e-01 -1.03007948e+00 3.73812526e-01 -4.60981816e-01
-7.98193097e-01 1.05477023e+00 -4.64362144e-01 -1.96569443e-01
1.32399619e+00 6.38157547e-01 1.05933510e-01 -9.70306754e-01
-9.03673470e-01 3.76413882e-01 -1.58005327e-01 -5.65232895e-02
5.61477780e-01 4.11434025e-01 -8.57044876e-01 1.22360456e+00
-9.89449203e-01 -8.22969824e-02 1.18826318e+00 9.58713889e-01
-2.76917428e-01 -7.78134108e-01 -5.70003629e-01 1.15183912e-01
-1.72663522e+00 -3.19457650e-01 -4.50948983e-01 1.17802024e+00
9.87676010e-02 9.77208972e-01 4.26042914e-01 -4.33619559e-01
-1.66877270e-01 -1.30204290e-01 -3.33878756e-01 -3.38390589e-01
-1.27031732e+00 5.71236908e-01 4.77007091e-01 -3.63884836e-01
-8.66577029e-01 -1.52907276e+00 -6.33987725e-01 -7.53766179e-01
-6.58814669e-01 4.73358870e-01 1.37514603e+00 1.11596930e+00
-1.32114291e-01 3.23135614e-01 9.10585940e-01 -9.06611606e-03
1.90312386e-01 -1.11716402e+00 -9.52500284e-01 -2.41850782e-02
1.69286862e-01 -3.57854903e-01 -7.06744134e-01 1.26026228e-01] | [4.525018215179443, 4.425189018249512] |
7d0fd543-ec3c-4dc2-812c-15328c406c39 | audio-visual-recognition-of-overlapped-speech | 2001.01656 | null | https://arxiv.org/abs/2001.01656v1 | https://arxiv.org/pdf/2001.01656v1.pdf | Audio-visual Recognition of Overlapped speech for the LRS2 dataset | Automatic recognition of overlapped speech remains a highly challenging task to date. Motivated by the bimodal nature of human speech perception, this paper investigates the use of audio-visual technologies for overlapped speech recognition. Three issues associated with the construction of audio-visual speech recognition (AVSR) systems are addressed. First, the basic architecture designs i.e. end-to-end and hybrid of AVSR systems are investigated. Second, purposefully designed modality fusion gates are used to robustly integrate the audio and visual features. Third, in contrast to a traditional pipelined architecture containing explicit speech separation and recognition components, a streamlined and integrated AVSR system optimized consistently using the lattice-free MMI (LF-MMI) discriminative criterion is also proposed. The proposed LF-MMI time-delay neural network (TDNN) system establishes the state-of-the-art for the LRS2 dataset. Experiments on overlapped speech simulated from the LRS2 dataset suggest the proposed AVSR system outperformed the audio only baseline LF-MMI DNN system by up to 29.98\% absolute in word error rate (WER) reduction, and produced recognition performance comparable to a more complex pipelined system. Consistent performance improvements of 4.89\% absolute in WER reduction over the baseline AVSR system using feature fusion are also obtained. | ['Shansong Liu', 'Shi-Xiong Zhang', 'Shahram Ghorbani', 'Jian Wu', 'Helen Meng', 'Xunying Liu', 'Bo Wu', 'Shiyin Kang', 'Jianwei Yu', 'Dong Yu'] | 2020-01-06 | null | null | null | null | ['lipreading', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [ 4.39545661e-01 -4.95711342e-02 2.43525013e-01 -4.71780807e-01
-1.29010582e+00 -3.56719911e-01 6.38287425e-01 -2.16184050e-01
-3.79531264e-01 3.57898414e-01 2.69126296e-01 -5.13415515e-01
8.38385746e-02 1.12989284e-01 -3.75327498e-01 -8.28050017e-01
2.36143976e-01 2.95232832e-02 2.36204028e-01 -1.94412097e-01
1.33747071e-01 7.66892016e-01 -1.95390606e+00 4.47387755e-01
5.26069820e-01 1.16891372e+00 6.22378886e-01 1.10910606e+00
-2.24507660e-01 5.24941385e-01 -1.03601527e+00 2.69084811e-01
7.88618997e-02 -2.36402839e-01 -5.03566623e-01 7.99727738e-02
7.03268290e-01 -1.15151241e-01 -7.50192285e-01 7.18058765e-01
9.17498291e-01 3.45955998e-01 5.11341929e-01 -1.02273881e+00
-5.40009141e-01 5.43572187e-01 -4.40088302e-01 4.66620445e-01
4.07316834e-01 1.15368441e-01 9.86174405e-01 -1.56540430e+00
2.88288504e-01 1.42555892e+00 1.91376939e-01 5.65795183e-01
-1.22644448e+00 -6.56182408e-01 3.91056575e-02 2.93208510e-01
-1.70716572e+00 -1.21273494e+00 6.02429330e-01 -2.17678294e-01
1.84181416e+00 4.80003327e-01 3.86882275e-01 9.71438110e-01
-8.86545866e-04 1.03609312e+00 1.00533342e+00 -8.07984531e-01
3.42509598e-02 7.48409927e-02 4.19207633e-01 4.05881792e-01
-2.51232296e-01 4.63404596e-01 -1.07860017e+00 2.22965732e-01
2.61887401e-01 -5.77161431e-01 -3.21898222e-01 1.84557065e-01
-1.02316320e+00 3.76237452e-01 -1.84505228e-02 4.80964571e-01
-1.18650541e-01 8.39475822e-03 5.18696129e-01 4.82259721e-01
3.43667924e-01 -1.92691162e-01 -3.52354825e-01 -1.56372741e-01
-1.45152366e+00 -2.12511823e-01 4.28487003e-01 1.02810085e+00
1.83069378e-01 1.07862246e+00 -4.31088269e-01 1.23535502e+00
8.81383777e-01 7.42457092e-01 4.84312266e-01 -5.53496063e-01
4.85827893e-01 1.11571945e-01 -3.57458711e-01 -2.03742206e-01
-2.97901869e-01 -3.47589880e-01 -7.07890809e-01 5.44347346e-01
2.11807221e-01 3.06934029e-01 -1.56473827e+00 1.24314833e+00
-9.97798219e-02 -2.69973218e-01 5.87634206e-01 8.37648690e-01
1.30044186e+00 1.15220451e+00 -1.37505561e-01 -2.96974540e-01
1.36470735e+00 -9.33042586e-01 -1.07841504e+00 -2.77423739e-01
1.46461353e-01 -1.10428774e+00 8.48869503e-01 5.00780821e-01
-1.26243258e+00 -9.76301670e-01 -1.32585907e+00 -3.89981791e-02
-1.26759082e-01 5.18438220e-01 2.24822648e-02 9.99564171e-01
-1.61746764e+00 -1.17691755e-01 -5.85896313e-01 -2.52390951e-01
7.26199299e-02 6.37716353e-01 -4.37081665e-01 1.32716745e-01
-1.10893655e+00 7.69117057e-01 3.11299801e-01 4.42617804e-01
-1.09880638e+00 -2.15748206e-01 -9.03791606e-01 5.06461561e-02
5.10841534e-02 -1.42712727e-01 1.54451048e+00 -8.87672365e-01
-1.99067879e+00 6.83006346e-01 -6.58533394e-01 -6.11264467e-01
-1.22875027e-01 1.47062868e-01 -1.01873124e+00 2.94273347e-01
-3.23456496e-01 8.60050976e-01 1.11644220e+00 -9.94461358e-01
-7.17109680e-01 -2.06661880e-01 -6.15650952e-01 2.77841955e-01
-8.91274065e-02 3.59724462e-01 -3.16467464e-01 -6.03531957e-01
2.18749404e-01 -6.10238075e-01 2.43427411e-01 -2.10873127e-01
-2.16891840e-01 -3.61613214e-01 1.16947448e+00 -8.52873206e-01
1.11350822e+00 -2.62419748e+00 -1.95504993e-01 1.11652441e-01
-2.58757491e-02 9.35027778e-01 -2.73069888e-01 4.30254370e-01
-1.65023819e-01 -2.72595346e-01 -1.18449959e-03 -5.30922234e-01
-7.22207874e-02 1.57294329e-02 -3.66514236e-01 5.09707153e-01
1.01152033e-01 4.02346373e-01 -3.63260150e-01 -3.89132857e-01
5.23113728e-01 7.79877484e-01 -4.31584567e-02 3.16524059e-01
2.42763162e-01 3.37317944e-01 3.42634827e-01 1.01023018e+00
5.67171693e-01 3.63003492e-01 -1.26931757e-01 -2.73627520e-01
-4.69405502e-01 5.24843991e-01 -1.14297855e+00 1.64884543e+00
-3.97668570e-01 1.30412114e+00 5.16826749e-01 -7.82239795e-01
1.55558610e+00 9.46426809e-01 -7.80224577e-02 -9.85097051e-01
-1.87815763e-02 6.03354692e-01 9.21255723e-02 -2.38262504e-01
6.12707675e-01 9.08155367e-02 2.04127833e-01 -2.59562135e-01
4.54959363e-01 -5.99232987e-02 -8.38864148e-02 9.39227734e-03
8.05413604e-01 -2.01460287e-01 1.73561186e-01 -6.84083030e-02
1.00051045e+00 -1.88740864e-01 4.19460833e-01 7.68297493e-01
-4.87654805e-01 7.40113020e-01 -7.31263831e-02 5.82891926e-02
-9.35546994e-01 -1.41006434e+00 -3.22480410e-01 9.53721046e-01
-5.07847033e-02 -1.53550327e-01 -4.17828560e-01 -6.38263077e-02
-4.35229480e-01 7.91963339e-01 3.17389965e-01 1.96794406e-01
-5.19179344e-01 -1.00747414e-01 1.12432694e+00 6.18629932e-01
5.71573973e-01 -9.51914549e-01 -2.50710458e-01 2.74008989e-01
-3.44645195e-02 -1.19236767e+00 -4.94562030e-01 3.53221208e-01
-4.47581947e-01 -3.52942437e-01 -8.34354281e-01 -1.14147878e+00
2.04165906e-01 4.87191617e-01 5.50337791e-01 -5.59083343e-01
-3.77753586e-01 4.43517625e-01 -2.73241967e-01 -1.77066162e-01
-6.72590077e-01 -2.10852027e-01 1.98582530e-01 -2.03517973e-02
2.75811881e-01 -3.16062212e-01 -5.04517496e-01 1.40777871e-01
-5.07007003e-01 -1.94383282e-02 8.69796693e-01 1.12074506e+00
6.83035493e-01 -2.50726253e-01 7.75892735e-01 1.21252254e-01
5.00299752e-01 1.91469342e-01 -8.39298010e-01 3.17923069e-01
-6.82366014e-01 -9.56844911e-02 2.78808951e-01 -4.32529002e-01
-1.40257633e+00 4.63599503e-01 -5.86156666e-01 -5.64272463e-01
-3.95979732e-01 1.71450764e-01 -4.19187397e-01 -1.06580950e-01
3.88179153e-01 5.89815259e-01 2.16617808e-01 -4.69581246e-01
3.69666159e-01 1.62011337e+00 7.20074296e-01 6.55960366e-02
2.18193054e-01 3.05350926e-02 -3.75684530e-01 -1.67329252e+00
9.93167683e-02 -8.55647445e-01 -2.13975728e-01 -3.78157675e-01
7.54138410e-01 -1.33752048e+00 -6.55638635e-01 7.74635732e-01
-1.46271110e+00 -1.24727171e-02 -1.65018424e-01 7.19526052e-01
-3.91158879e-01 4.54997957e-01 -5.92602372e-01 -1.44194758e+00
-7.10149586e-01 -1.52654338e+00 1.12564886e+00 1.32288471e-01
-1.81788594e-01 -3.09678465e-01 -5.02604842e-02 6.20258927e-01
5.10263681e-01 -5.79259694e-01 4.34181750e-01 -9.20600832e-01
-4.26269740e-01 -5.05744256e-02 -2.96982884e-01 6.84620023e-01
-7.68893287e-02 1.65142536e-01 -1.66270196e+00 -3.41182381e-01
-2.38319226e-02 -9.24661383e-02 1.01038969e+00 5.07693887e-01
4.60565597e-01 -6.24564067e-02 -2.32806966e-01 4.02071714e-01
1.08244705e+00 1.00790322e+00 6.86560035e-01 -1.12819009e-01
4.73586649e-01 4.87005264e-01 5.55021763e-01 2.19056889e-01
-6.86361566e-02 8.82766902e-01 1.18503034e-01 -1.11941420e-01
-8.26317728e-01 4.12579952e-03 8.90894353e-01 1.14540958e+00
5.53483129e-01 -5.39897740e-01 -8.86213362e-01 6.67605698e-01
-1.32162416e+00 -8.05511236e-01 -2.08023295e-01 2.11855459e+00
4.88974154e-01 1.42565012e-01 8.55568573e-02 7.13056982e-01
9.34542179e-01 3.76395792e-01 -2.37128392e-01 -8.50521982e-01
-4.45000052e-01 5.11186659e-01 4.57274884e-01 8.11624944e-01
-9.49888110e-01 9.01071429e-01 6.44393492e+00 1.32833946e+00
-1.32719851e+00 2.11766407e-01 3.03014934e-01 -1.67457223e-01
1.90339936e-03 -2.83363640e-01 -1.16483366e+00 1.15443900e-01
1.62496483e+00 2.63625085e-01 3.40156198e-01 5.59031487e-01
3.43466103e-01 1.02396226e-02 -8.18836510e-01 1.27999175e+00
1.79883644e-01 -1.08625078e+00 -1.46652386e-01 -1.29138201e-01
1.05858184e-01 2.32394338e-01 2.22115695e-01 1.55786276e-01
-2.31896207e-01 -1.05179584e+00 1.08932137e+00 1.95607319e-01
1.36763561e+00 -7.94521630e-01 3.74500781e-01 -9.00022760e-02
-1.66560411e+00 -5.37811033e-02 7.90037587e-02 2.25682303e-01
2.35904083e-01 2.90114313e-01 -1.25718737e+00 3.66320550e-01
6.18894279e-01 1.54726565e-01 -1.87531948e-01 9.06850576e-01
-1.01834156e-01 8.01019549e-01 -3.77188712e-01 -8.04776698e-02
2.30697900e-01 3.72017443e-01 8.79536450e-01 1.66037345e+00
3.55553716e-01 -2.85708487e-01 -3.53439897e-01 4.82391119e-01
4.15333927e-01 -2.77741458e-02 -6.76869214e-01 -3.02645355e-01
7.25480735e-01 1.10245299e+00 -4.58145261e-01 -1.65422589e-01
-5.89546382e-01 8.61414611e-01 -1.70684978e-01 4.38378453e-01
-5.95355749e-01 -7.34903455e-01 6.16909981e-01 -1.94274485e-01
5.88559210e-01 -4.71421808e-01 -1.37406573e-01 -5.71181417e-01
5.35949990e-02 -9.15551782e-01 -1.40971359e-04 -8.99641037e-01
-7.65099227e-01 1.23748851e+00 -1.95459932e-01 -1.40976870e+00
-4.38811511e-01 -7.96580851e-01 -3.63848567e-01 1.23037505e+00
-1.44467473e+00 -1.26391935e+00 3.19639482e-02 6.45206749e-01
1.11136508e+00 -7.69950688e-01 8.17916930e-01 3.90347689e-01
-5.87586462e-01 8.67337167e-01 2.81055748e-01 -1.48448363e-01
5.02615869e-01 -8.52390051e-01 6.33957326e-01 1.14308894e+00
4.84411985e-01 2.48284280e-01 5.67888379e-01 -4.59634870e-01
-1.45447731e+00 -7.95130372e-01 1.07944858e+00 3.84702384e-01
3.92556340e-01 -5.34852982e-01 -8.09256375e-01 2.60675043e-01
5.30752063e-01 -3.09416428e-02 5.16965747e-01 -5.01865089e-01
-4.42461133e-01 -4.52331096e-01 -1.00117254e+00 4.12382662e-01
8.25292647e-01 -1.01732254e+00 -6.14329875e-01 3.15569043e-02
9.44006085e-01 -3.46190125e-01 -4.03487772e-01 2.78095663e-01
4.99074101e-01 -8.11415076e-01 1.02789283e+00 1.02405764e-01
-4.55861956e-01 -6.31520212e-01 -6.56031072e-01 -9.67300832e-01
3.61879021e-02 -8.49112332e-01 -2.31883582e-02 1.50148785e+00
5.79899848e-01 -6.51140869e-01 3.22199017e-01 6.00997880e-02
-7.37589955e-01 -2.82356411e-01 -1.57547307e+00 -1.09618986e+00
-5.73715270e-01 -6.94887161e-01 -3.93870398e-02 -1.64863449e-02
-4.36330289e-02 4.42757517e-01 -2.34672517e-01 5.02567053e-01
3.75690639e-01 -4.12882954e-01 1.98934317e-01 -8.16143930e-01
-2.43316397e-01 -4.44836706e-01 -6.45591140e-01 -1.26079869e+00
8.29656646e-02 -7.92016685e-01 4.08263087e-01 -1.53929472e+00
-5.44946015e-01 6.99023083e-02 -2.11251393e-01 2.98647612e-01
3.92306805e-01 7.36886486e-02 1.45533487e-01 1.63872436e-01
-2.86572129e-01 5.82919300e-01 6.29538536e-01 -4.52484816e-01
-3.74899715e-01 -7.56168291e-02 -2.76973117e-02 2.87712872e-01
6.00379288e-01 -1.89239472e-01 -6.15697563e-01 -2.69300967e-01
-5.58607042e-01 5.59376657e-01 2.20400080e-01 -1.20148337e+00
7.43468583e-01 3.79760742e-01 9.19284299e-02 -1.33917558e+00
1.01643884e+00 -6.93510473e-01 1.30518526e-01 2.44511306e-01
-3.08287501e-01 -5.89679293e-02 5.39006174e-01 4.75672990e-01
-7.13907778e-01 -8.33320916e-02 8.31673920e-01 2.72570193e-01
-9.53755498e-01 -3.27336162e-01 -1.08168197e+00 -2.93425888e-01
7.95564651e-01 -6.95777476e-01 -3.02432060e-01 -2.04461411e-01
-7.70321250e-01 -2.48707175e-01 -4.04649526e-01 3.82313401e-01
1.23558557e+00 -1.07097876e+00 -7.08788335e-01 6.58649087e-01
1.08944960e-01 -1.80519477e-01 4.65494126e-01 7.65192688e-01
-3.96739930e-01 8.96841824e-01 -3.17570791e-02 -8.32908392e-01
-1.81736970e+00 2.29301170e-01 3.69295567e-01 3.10478061e-01
-5.99210382e-01 1.14972353e+00 6.68378994e-02 -4.60630516e-03
8.69752347e-01 -2.04565406e-01 -2.13591740e-01 2.73616966e-02
5.98331273e-01 4.97405112e-01 4.19510037e-01 -9.13824081e-01
-7.08125174e-01 1.92180559e-01 -1.94030672e-01 -7.56127894e-01
1.02477467e+00 -4.33854669e-01 3.44615459e-01 6.45115674e-01
1.13399208e+00 -1.13250524e-01 -9.43242908e-01 -1.91586956e-01
-1.08245306e-01 -1.36207625e-01 4.67095971e-01 -9.61475432e-01
-9.91841197e-01 1.10944045e+00 1.16649830e+00 6.11462668e-02
1.26203036e+00 -1.03615612e-01 6.28171325e-01 2.42173478e-01
-6.13265075e-02 -1.11324739e+00 -1.81641772e-01 5.97464025e-01
1.14039350e+00 -9.50173259e-01 -5.20260513e-01 -2.47135431e-01
-7.02502370e-01 1.41385555e+00 4.27073926e-01 3.25387150e-01
4.53365445e-01 6.78036869e-01 2.96877801e-01 3.97913828e-02
-7.45195210e-01 -3.73176485e-01 4.53409225e-01 7.21916854e-01
5.81733823e-01 -5.58442175e-02 -2.14354321e-02 3.03700417e-01
8.97282660e-02 -3.45216900e-01 1.25676513e-01 7.49375224e-01
-5.51067710e-01 -8.86101484e-01 -6.67808831e-01 -2.10076924e-02
-3.20581228e-01 -3.34952354e-01 -2.88169086e-01 5.93131900e-01
-2.80960798e-01 1.51389396e+00 1.43411979e-01 -6.67776525e-01
4.06209290e-01 3.05829883e-01 2.63748795e-01 -2.54679203e-01
-6.51460290e-01 6.53677881e-01 3.11079353e-01 -2.88351625e-01
-2.36474186e-01 -5.65544665e-01 -1.38749373e+00 1.39758199e-01
-5.13769448e-01 -1.39390692e-01 1.10756946e+00 8.59300315e-01
4.19758439e-01 7.44695604e-01 6.36037230e-01 -7.78390050e-01
-4.80885863e-01 -1.03902936e+00 -5.86725533e-01 -4.76937503e-01
7.49081790e-01 -3.36276263e-01 -6.99616551e-01 -5.48707396e-02] | [14.481708526611328, 5.332518577575684] |
0408110f-08f3-48c4-a90a-b9c6e280799b | noisy-label-detection-for-speaker-recognition | 2212.00239 | null | https://arxiv.org/abs/2212.00239v2 | https://arxiv.org/pdf/2212.00239v2.pdf | Inconsistency Ranking-based Noisy Label Detection for High-quality Data | The success of deep learning requires high-quality annotated and massive data. However, the size and the quality of a dataset are usually a trade-off in practice, as data collection and cleaning are expensive and time-consuming. In real-world applications, especially those using crowdsourcing datasets, it is important to exclude noisy labels. To address this, this paper proposes an automatic noisy label detection (NLD) technique with inconsistency ranking for high-quality data. We apply this technique to the automatic speaker verification (ASV) task as a proof of concept. We investigate both inter-class and intra-class inconsistency ranking and compare several metric learning loss functions under different noise settings. Experimental results confirm that the proposed solution could increase both the efficient and effective cleaning of large-scale speaker recognition datasets. | ['Zhizheng Wu', 'Lei Zhang', 'Yushi Ye', 'Yifan He', 'Yi Wang', 'Hanzhi Yin', 'Ruibin Yuan'] | 2022-12-01 | null | null | null | null | ['speaker-recognition', 'speaker-verification'] | ['speech', 'speech'] | [-7.21651837e-02 -2.38329187e-01 2.75796443e-01 -8.78225088e-01
-1.26183784e+00 -5.59875786e-01 2.23925561e-01 3.53589386e-01
-6.73055768e-01 8.09573770e-01 5.75008169e-02 -2.80470289e-02
-2.29670331e-01 -4.59638983e-01 -5.70068240e-01 -8.37197185e-01
1.91482827e-01 3.49637151e-01 -1.84454001e-03 1.38227776e-01
1.47171080e-01 3.30009013e-01 -1.76371789e+00 -6.99402168e-02
1.16529870e+00 1.06350005e+00 3.75836529e-02 1.46414787e-01
-2.63210446e-01 3.67356151e-01 -9.87494588e-01 -6.95142984e-01
2.95498282e-01 -2.37426266e-01 -6.01354897e-01 1.99885622e-01
5.87362885e-01 6.51938841e-02 4.30427939e-01 1.31702840e+00
9.85153973e-01 1.58496201e-01 -2.90132333e-02 -1.68244839e+00
-2.68055290e-01 6.22618020e-01 -3.41468722e-01 8.20566714e-02
2.44799480e-02 -3.23556662e-02 8.96025658e-01 -8.63939643e-01
2.14853764e-01 1.02845895e+00 7.63249934e-01 7.00667381e-01
-1.01865447e+00 -9.47196603e-01 1.43137991e-01 3.86703879e-01
-1.51623821e+00 -8.90663028e-01 8.18382084e-01 -3.23032141e-01
4.54824954e-01 2.40286887e-01 2.02364445e-01 1.06125510e+00
-3.51257980e-01 4.72222120e-01 1.15503979e+00 -4.88268286e-01
6.30526900e-01 3.31471384e-01 3.26118261e-01 4.26647097e-01
3.70051891e-01 -2.49630302e-01 -8.31581473e-01 -2.20911115e-01
1.26978248e-01 -1.65542603e-01 -3.12358469e-01 -1.85778812e-01
-1.01912975e+00 7.00868607e-01 6.64343014e-02 3.52969468e-01
-2.01381475e-01 -5.91438673e-02 5.67112148e-01 2.50734329e-01
5.31066358e-01 2.02024430e-01 -4.19521958e-01 -2.11277694e-01
-9.95888114e-01 4.59916517e-02 7.41038561e-01 7.75141835e-01
5.15214443e-01 -2.77831703e-02 -2.03245848e-01 1.18062413e+00
2.97444761e-01 4.46250170e-01 5.39926529e-01 -7.86936045e-01
5.97019970e-01 5.16455173e-01 3.01191449e-01 -8.89272213e-01
-4.31902051e-01 -4.03288990e-01 -9.45990860e-01 3.34042013e-02
5.29301941e-01 -5.26810028e-02 -7.03700125e-01 1.95749807e+00
6.99671030e-01 2.40470231e-01 -1.91405326e-01 1.01125395e+00
7.90096819e-01 1.28953502e-01 2.67857224e-01 -5.38272977e-01
1.29845047e+00 -7.82370508e-01 -1.21043980e+00 -1.00362122e-01
5.77671111e-01 -7.97366321e-01 1.31402600e+00 4.59195793e-01
-7.35059202e-01 -4.31021810e-01 -8.89787436e-01 -1.14206292e-01
-2.31105432e-01 1.91917539e-01 2.79164076e-01 9.42341089e-01
-8.16658497e-01 3.22378337e-01 -6.35807276e-01 -1.37430608e-01
6.02849424e-01 3.98995668e-01 -5.37978828e-01 -6.44505545e-02
-1.11586976e+00 5.45459867e-01 7.73295835e-02 4.11471844e-01
-8.54939222e-01 -3.62590522e-01 -6.88215971e-01 4.86532450e-02
5.23538172e-01 -8.50917846e-02 1.26400065e+00 -7.98978448e-01
-1.30168128e+00 7.76563764e-01 -4.11559552e-01 -3.20207030e-01
4.97201860e-01 -1.60534203e-01 -5.11242151e-01 -3.17041010e-01
8.88744667e-02 2.79253304e-01 7.21532643e-01 -1.23798394e+00
-6.07452512e-01 -5.96928537e-01 -1.28047213e-01 1.37858628e-03
-4.61049259e-01 2.59430617e-01 -2.99495995e-01 -5.06859958e-01
1.78418517e-01 -8.51175964e-01 3.97817157e-02 -2.96271294e-02
-3.65103394e-01 -4.40963089e-01 5.89212596e-01 -8.56422126e-01
1.20266390e+00 -2.28083396e+00 -1.34210274e-01 1.67021126e-01
1.09295055e-01 4.30395633e-01 -2.09674627e-01 -1.26410395e-01
1.40358821e-01 1.93825692e-01 -3.49453479e-01 -8.15021873e-01
1.24698095e-01 1.76873490e-01 -8.60771630e-04 5.13416529e-01
1.17327899e-01 4.89682764e-01 -7.01366067e-01 -6.42383218e-01
-5.88588119e-02 4.87917155e-01 -1.29105285e-01 2.74284035e-01
-2.86901165e-02 4.64598566e-01 -1.07362844e-01 7.62060821e-01
9.44278836e-01 -3.84628400e-03 1.19690739e-01 -1.35801837e-01
1.62225246e-01 3.71720433e-01 -1.72302830e+00 1.55145848e+00
-5.12302816e-01 5.24947345e-01 4.46084678e-01 -9.78172183e-01
8.87698770e-01 3.26465309e-01 2.15754151e-01 -8.43253791e-01
2.54464000e-01 3.53340775e-01 -2.26252452e-01 -6.64482594e-01
4.00495917e-01 -2.07010806e-01 3.69123323e-03 3.87307614e-01
-1.65242165e-01 1.94190264e-01 6.98627159e-02 -6.79929629e-02
9.64114428e-01 -2.54528672e-01 1.23768263e-01 -1.67680606e-01
6.35375023e-01 -3.48075867e-01 1.13403583e+00 5.27241766e-01
-7.73112893e-01 5.21845758e-01 3.66849989e-01 -2.57893115e-01
-7.26442993e-01 -4.76000667e-01 -1.94626838e-01 9.84829545e-01
4.73498330e-02 -1.50673151e-01 -1.01317000e+00 -8.42728019e-01
-7.33726919e-02 4.51634735e-01 -3.54450792e-01 9.26004425e-02
-4.01038498e-01 -8.37177038e-01 7.21091747e-01 3.30140024e-01
5.18656313e-01 -9.29054141e-01 -1.68378815e-01 1.93688259e-01
-6.33309186e-01 -1.38446963e+00 -4.36285317e-01 2.47228742e-01
-4.95246232e-01 -1.12059700e+00 -2.05935046e-01 -8.00803304e-01
6.29564524e-01 3.18210751e-01 9.86378133e-01 2.18855917e-01
-8.41489807e-02 -1.04228057e-01 -5.15068173e-01 -4.64205235e-01
-4.31337059e-01 8.33975896e-03 4.49254125e-01 2.08155215e-01
7.60262072e-01 -2.83101410e-01 -2.94874579e-01 6.21656477e-01
-9.18200374e-01 -4.75658029e-01 1.61403343e-01 7.98583806e-01
6.75135314e-01 3.68374825e-01 1.01198459e+00 -4.64388609e-01
7.33028769e-01 -1.56127810e-01 -9.34623361e-01 4.96699780e-01
-6.19500399e-01 1.69127174e-02 4.16197002e-01 -4.69374865e-01
-9.16856647e-01 1.35221764e-01 -2.02985168e-01 -1.38133317e-01
-2.18197063e-01 1.92128077e-01 -7.59902298e-01 -5.19799702e-02
5.24032950e-01 -6.60847947e-02 -2.61515528e-01 -6.42711461e-01
5.07696234e-02 1.15759301e+00 2.35425338e-01 -2.90546358e-01
6.62833691e-01 3.79886806e-01 -1.53777748e-01 -6.70887887e-01
-9.28812027e-01 -5.24133444e-01 -5.25353312e-01 -7.25251883e-02
6.74203098e-01 -9.50269938e-01 -7.76268363e-01 6.40200377e-01
-1.07910669e+00 -1.32030740e-01 -1.16273105e-01 4.50424939e-01
1.19563326e-01 5.47126591e-01 -2.16359407e-01 -1.06121731e+00
-4.27649975e-01 -1.35881472e+00 1.22492504e+00 1.25891432e-01
8.48906636e-02 -4.85370487e-01 -6.20151907e-02 8.20210576e-01
3.17114949e-01 -1.02845721e-01 4.09786403e-01 -8.22392881e-01
-5.11746228e-01 -2.15125054e-01 -2.11635500e-01 6.70856535e-01
6.83666393e-02 -2.54901975e-01 -1.28590548e+00 -1.73353076e-01
2.41244003e-01 -4.80641603e-01 5.31547546e-01 1.94354191e-01
1.42009771e+00 -2.57509589e-01 1.38274327e-01 2.44129792e-01
1.17820513e+00 -1.20285973e-01 4.01833326e-01 2.64318436e-01
6.19781792e-01 7.98363388e-01 8.68006945e-01 5.11376679e-01
3.61237884e-01 8.99750292e-01 3.27467412e-01 2.25043729e-01
-2.95427199e-02 1.29601061e-01 2.38586798e-01 1.03260696e+00
3.17453176e-01 -3.92012298e-01 -9.23132062e-01 6.83957040e-01
-1.82681024e+00 -8.46662521e-01 -2.49711543e-01 2.49288225e+00
1.23546946e+00 -3.95473503e-02 3.09778266e-02 6.88634157e-01
1.01169074e+00 -1.05806917e-01 -3.04231405e-01 -9.18957889e-02
-1.79739252e-01 -1.21553443e-01 2.38707557e-01 5.63382924e-01
-1.08622909e+00 6.17006481e-01 5.11066675e+00 1.00193298e+00
-1.05847049e+00 6.46324933e-01 5.16442120e-01 -2.82741606e-01
-8.86471495e-02 -3.94580454e-01 -9.68931973e-01 6.59342945e-01
8.81887913e-01 3.73609722e-01 4.35383350e-01 8.18291366e-01
4.75952625e-01 -1.68296024e-01 -9.88767624e-01 1.25937903e+00
9.96550918e-02 -8.42500448e-01 -5.54580152e-01 1.88459177e-02
5.78632236e-01 7.98524916e-02 -3.09779078e-01 2.32326984e-01
5.02374917e-02 -7.94318497e-01 8.52907360e-01 1.33120477e-01
5.84804177e-01 -6.88011050e-01 1.03175354e+00 3.41236234e-01
-9.75982189e-01 -1.91497996e-01 -3.71821582e-01 2.40223527e-01
1.63315833e-01 1.21939898e+00 -5.82245827e-01 1.55492142e-01
9.22067881e-01 1.82566755e-02 -5.89136720e-01 1.25827575e+00
-4.37425792e-01 5.94955266e-01 -4.25557971e-01 -8.77268314e-02
-2.74165690e-01 -9.31415334e-02 3.86656076e-01 1.01365519e+00
1.88504472e-01 -1.33439481e-01 2.51255706e-02 4.32164878e-01
-5.10886312e-01 1.00157514e-01 -1.88201681e-01 1.64272755e-01
9.67543066e-01 1.31653881e+00 -5.39104283e-01 -2.18808398e-01
-2.13745162e-01 6.96920633e-01 3.95425349e-01 1.66318327e-01
-1.08554912e+00 -3.06761593e-01 7.30921209e-01 -4.47785221e-02
1.31605059e-01 -2.00890645e-01 -5.51218271e-01 -1.16562402e+00
7.10868001e-01 -1.09772944e+00 1.73469707e-01 -2.27185383e-01
-1.30103552e+00 5.37718594e-01 -4.64592218e-01 -1.24636281e+00
2.69695163e-01 -7.38366693e-02 -1.43605381e-01 5.82110941e-01
-1.73898208e+00 -8.85132194e-01 -7.72240222e-01 4.54931200e-01
5.04765272e-01 -5.64796850e-02 6.86835110e-01 9.01914418e-01
-8.70485187e-01 9.85563099e-01 1.81190446e-01 1.37536377e-01
8.90493512e-01 -9.64307249e-01 8.93972740e-02 9.54621494e-01
2.54078060e-01 4.32330906e-01 6.02447093e-01 -4.48778927e-01
-1.10462391e+00 -1.25977600e+00 1.11605680e+00 -3.39749128e-01
2.69736618e-01 -6.79557264e-01 -1.11121750e+00 1.86340213e-01
-1.49846092e-01 1.02917098e-01 9.71626937e-01 2.82860398e-01
-4.58745033e-01 -5.77143729e-01 -1.51626849e+00 1.23883307e-01
1.16781116e+00 -6.94847047e-01 -4.11979854e-01 4.63172525e-01
6.34209633e-01 -2.32511893e-01 -6.79311812e-01 3.99251580e-01
3.05745244e-01 -8.05121124e-01 5.19172013e-01 -2.84697652e-01
-2.87049055e-01 -5.32287717e-01 -3.05684090e-01 -1.24982727e+00
1.14951372e-01 -4.15860623e-01 1.81833699e-01 1.83884764e+00
4.15502846e-01 -3.91086847e-01 5.78406274e-01 1.08823967e+00
-4.41931970e-02 -3.39906871e-01 -1.46837056e+00 -8.52818072e-01
-4.05999929e-01 -7.21068501e-01 1.00981319e+00 1.10373175e+00
-3.03322852e-01 1.73969060e-01 -3.43035102e-01 5.50287902e-01
6.59065425e-01 -1.44157887e-01 7.31676936e-01 -1.39625132e+00
8.58602077e-02 -2.11792842e-01 -3.66320193e-01 -2.69826859e-01
2.62294263e-01 -4.88547146e-01 4.21437144e-01 -1.34316266e+00
7.52966478e-02 -7.86516666e-01 -4.22004938e-01 6.22032762e-01
-3.94297093e-01 2.74149030e-01 -9.46556684e-03 1.75124526e-01
-9.23551202e-01 6.98081672e-01 5.25032461e-01 -9.43073556e-02
-2.19351351e-01 3.53691988e-02 -5.14221072e-01 5.43296456e-01
9.53355789e-01 -8.83215547e-01 -2.36074731e-01 -6.91850305e-01
2.80420423e-01 -4.09822732e-01 2.57783175e-01 -9.78898704e-01
1.94757774e-01 -1.83570907e-01 -6.15582243e-02 -2.47062072e-01
2.81178594e-01 -1.09396470e+00 -8.91871378e-02 5.04442304e-02
-4.00040507e-01 -1.28755614e-01 -3.85166779e-02 6.62333667e-01
-4.35671121e-01 -2.59722173e-01 9.10791874e-01 8.39840546e-02
-4.27147567e-01 8.76406357e-02 1.86769575e-01 1.12938859e-01
9.93758023e-01 1.51575252e-01 -4.43249553e-01 -2.32077703e-01
-5.02730072e-01 2.85946846e-01 3.33751857e-01 4.87165660e-01
4.67819631e-01 -1.29993129e+00 -6.60305083e-01 2.06670240e-01
2.94869661e-01 2.53058523e-01 1.78949401e-01 8.78045380e-01
-8.09536129e-02 1.21353269e-01 1.41839042e-01 -6.12615049e-01
-1.68035567e+00 3.50138783e-01 3.80681723e-01 -1.20999657e-01
-6.55382425e-02 1.00988030e+00 -3.39259595e-01 -6.25439048e-01
1.01771009e+00 -2.36632153e-01 -1.09770902e-01 3.69616181e-01
7.31541097e-01 4.84745085e-01 6.26397848e-01 -6.17736399e-01
-6.22346938e-01 3.63765150e-01 2.55603880e-01 -9.21789929e-02
1.20529616e+00 -4.49580520e-01 -2.28008509e-01 2.18429238e-01
1.02927303e+00 1.72578156e-01 -9.32146549e-01 -3.43270302e-01
2.46617496e-01 -5.49746037e-01 1.60605818e-01 -7.71574974e-01
-1.12233686e+00 9.67386305e-01 9.70431507e-01 2.65138447e-01
1.13812566e+00 -2.09206045e-01 9.38495755e-01 4.65176910e-01
4.35149759e-01 -1.56738508e+00 -1.25299469e-01 6.90540373e-02
8.04854393e-01 -1.68134129e+00 -1.92936540e-01 -3.76170725e-01
-6.65163815e-01 5.39416432e-01 5.92579246e-01 5.39846122e-01
6.72441244e-01 1.36923537e-01 4.78811324e-01 -9.23118368e-02
-4.18666899e-01 -1.88653708e-01 1.79853290e-01 4.98651445e-01
3.24598700e-01 1.53682783e-01 -4.94933844e-01 7.49220133e-01
2.81410981e-02 6.90596104e-02 3.37837547e-01 8.51567209e-01
-3.93461168e-01 -1.27313864e+00 -5.95732510e-01 2.50130653e-01
-5.27907431e-01 7.61765465e-02 -3.24905664e-01 2.47180700e-01
5.84239364e-01 1.62341046e+00 -2.80430436e-01 -2.64345199e-01
4.03494835e-01 3.47343773e-01 -5.98487165e-03 -4.60171312e-01
-5.72791636e-01 -3.59429009e-02 2.88937807e-01 -4.82336760e-01
-6.99957192e-01 -4.40769017e-01 -1.14028633e+00 -1.91253528e-01
-8.87052536e-01 2.34969556e-01 1.32231522e+00 9.19176638e-01
5.54356158e-01 2.51380146e-01 7.83424020e-01 -3.05461615e-01
-9.23098505e-01 -1.19535899e+00 -5.35948098e-01 7.18403637e-01
4.54274774e-01 -8.39214861e-01 -6.07783079e-01 4.30508293e-02] | [9.428751945495605, 3.850180149078369] |
eb82c20f-8573-4ae0-8b99-fb9027c5fc35 | attack-agnostic-adversarial-detection-on | 2105.01959 | null | https://arxiv.org/abs/2105.01959v1 | https://arxiv.org/pdf/2105.01959v1.pdf | Attack-agnostic Adversarial Detection on Medical Data Using Explainable Machine Learning | Explainable machine learning has become increasingly prevalent, especially in healthcare where explainable models are vital for ethical and trusted automated decision making. Work on the susceptibility of deep learning models to adversarial attacks has shown the ease of designing samples to mislead a model into making incorrect predictions. In this work, we propose a model agnostic explainability-based method for the accurate detection of adversarial samples on two datasets with different complexity and properties: Electronic Health Record (EHR) and chest X-ray (CXR) data. On the MIMIC-III and Henan-Renmin EHR datasets, we report a detection accuracy of 77% against the Longitudinal Adversarial Attack. On the MIMIC-CXR dataset, we achieve an accuracy of 88%; significantly improving on the state of the art of adversarial detection in both datasets by over 10% in all settings. We propose an anomaly detection based method using explainability techniques to detect adversarial samples which is able to generalise to different attack methods without a need for retraining. | ['Noura Al Moubayed', 'Matthew Watson'] | 2021-05-05 | null | null | null | null | ['explainable-models'] | ['computer-vision'] | [ 4.12201971e-01 6.70703232e-01 2.20411703e-01 -4.43658829e-01
-9.76574600e-01 -8.66669595e-01 4.99632210e-01 3.18939209e-01
-1.67770311e-01 5.84203839e-01 -5.40784022e-05 -8.69271815e-01
-1.37588888e-01 -4.72141981e-01 -8.29462230e-01 -3.34747672e-01
-2.98293084e-01 6.77477479e-01 -3.39508057e-01 9.65606794e-02
-2.13119105e-01 4.75511968e-01 -6.56558633e-01 6.48672938e-01
5.48639476e-01 8.22413862e-01 -8.79440248e-01 1.24372602e+00
5.21395326e-01 1.25681973e+00 -6.89350486e-01 -7.43875623e-01
5.08226871e-01 -5.60208678e-01 -1.00337887e+00 -4.32103038e-01
5.28335333e-01 -6.06860697e-01 -5.79259992e-01 8.54803383e-01
5.87849617e-01 -5.01708150e-01 5.97010672e-01 -1.38202536e+00
-4.99117136e-01 8.89005482e-01 -5.09166904e-02 2.15142950e-01
3.66616368e-01 4.31396276e-01 7.90363312e-01 -1.37641683e-01
5.80584705e-01 1.09248102e+00 1.05400240e+00 1.22038233e+00
-1.35606527e+00 -8.64438534e-01 -2.52756804e-01 3.98406051e-02
-1.04254651e+00 -2.18860462e-01 4.16630477e-01 -5.06782115e-01
8.12231898e-01 8.48154545e-01 3.90794605e-01 1.58603179e+00
5.91958404e-01 4.90952492e-01 1.18102229e+00 -1.40086427e-01
4.46390688e-01 1.30468845e-01 1.31107390e-01 6.53635681e-01
4.03987139e-01 5.85882366e-01 -5.74683510e-02 -9.45109725e-01
3.92900169e-01 3.88857931e-01 -2.72566676e-01 7.67786950e-02
-9.52252090e-01 9.99236345e-01 3.56130481e-01 1.27276883e-01
-3.45718324e-01 5.37035346e-01 5.46944678e-01 5.08915424e-01
2.88899034e-01 1.09304607e+00 -6.54936254e-01 5.31958789e-02
-7.63476551e-01 4.07480150e-01 9.95654702e-01 6.73043787e-01
3.63930315e-02 1.24266475e-01 -1.31156906e-01 -1.07899800e-01
9.45467036e-03 4.36892807e-01 2.49187320e-01 -8.31317782e-01
2.59322792e-01 4.20186222e-01 -6.07970953e-02 -8.79840255e-01
-5.77914715e-01 -4.65281755e-01 -1.00682676e+00 4.04250026e-01
3.43804687e-01 -3.48609090e-01 -1.13125014e+00 1.58037853e+00
2.47828394e-01 4.05548364e-01 1.84871450e-01 5.51563680e-01
5.88211119e-01 3.32919694e-02 3.02329570e-01 1.33133441e-01
1.32193506e+00 -3.51449937e-01 -5.10773599e-01 -7.20672980e-02
8.85774970e-01 -4.60660636e-01 7.51556635e-01 7.09292233e-01
-8.74352932e-01 -9.98746157e-02 -1.14636219e+00 2.70952821e-01
-1.50154769e-01 -7.15360463e-01 6.77907467e-01 1.12610972e+00
-6.85612023e-01 8.72872055e-01 -1.10333240e+00 6.47229254e-02
8.66460741e-01 5.40859580e-01 -7.08091259e-01 -1.15261022e-02
-1.25257134e+00 9.77037787e-01 -1.47297792e-03 -3.98040444e-01
-1.26261389e+00 -1.13221705e+00 -6.79915547e-01 7.59707168e-02
9.55356732e-02 -8.95291269e-01 1.10498297e+00 -9.56934035e-01
-7.27687299e-01 9.02975082e-01 2.38579735e-01 -1.21542406e+00
1.03905940e+00 -3.29361320e-01 -5.61141253e-01 1.05843566e-01
-3.03688288e-01 1.45901918e-01 9.86354291e-01 -1.11983407e+00
-4.74180002e-03 -2.34330639e-01 1.14618883e-01 -5.51724851e-01
-3.79858017e-02 -2.47974042e-02 5.73946536e-01 -7.23532438e-01
-2.89340019e-01 -1.14620948e+00 -5.40684819e-01 9.89487208e-03
-1.08135748e+00 3.48638415e-01 6.48607433e-01 -7.60755479e-01
1.04624724e+00 -1.87090552e+00 -2.84934580e-01 3.56968820e-01
9.03575718e-01 4.37147707e-01 1.50724612e-02 2.38442302e-01
-5.43148458e-01 6.08605385e-01 -6.47123396e-01 -2.99362063e-01
-9.16396920e-03 3.39088708e-01 -5.39801419e-01 6.51748419e-01
4.50379461e-01 1.04258120e+00 -7.94778109e-01 -1.80797368e-01
5.40625490e-02 3.80307764e-01 -8.94618630e-01 3.32477003e-01
-3.37344185e-02 6.72734678e-01 -4.04150188e-01 6.21964633e-01
4.25978273e-01 -2.45953307e-01 1.05442047e-01 1.96232095e-01
8.54682386e-01 1.08417712e-01 -8.27806175e-01 1.12500989e+00
-1.91879421e-01 5.83230197e-01 -6.28843829e-02 -6.86479926e-01
6.10023975e-01 7.36340821e-01 2.86784291e-01 -2.21451968e-01
3.17429632e-01 1.68959886e-01 3.82061392e-01 -4.06585902e-01
-5.28656878e-02 -5.40413737e-01 -2.57363349e-01 7.67338634e-01
-3.52730244e-01 -1.45990811e-02 -7.34388113e-01 3.47603470e-01
1.92594647e+00 -4.65795994e-01 4.28149819e-01 -2.32984275e-01
3.88620377e-01 8.15392938e-03 4.16254461e-01 1.31155765e+00
-4.51951534e-01 8.64595592e-01 6.01068854e-01 -1.12921810e+00
-1.10554826e+00 -1.05271804e+00 -2.92048395e-01 3.44178677e-01
-4.22342479e-01 -3.04541856e-01 -1.00086761e+00 -1.28222680e+00
3.14964622e-01 9.81045783e-01 -1.17240560e+00 -5.91070354e-01
-5.65780401e-01 -6.24273419e-01 1.05707943e+00 5.20228982e-01
-1.22161664e-01 -1.07102549e+00 -4.99356687e-01 1.29778430e-01
1.34874865e-01 -1.08760977e+00 -2.31755465e-01 2.73639858e-01
-6.07198656e-01 -1.46052253e+00 -1.33286595e-01 4.67332304e-02
7.33219087e-01 -6.34364009e-01 1.40301406e+00 5.80148876e-01
-8.75592351e-01 5.07569194e-01 -2.04168350e-01 -1.02939045e+00
-1.41294324e+00 1.01887926e-01 2.89171904e-01 -1.82575017e-01
3.19319397e-01 -3.42306823e-01 -7.11190581e-01 7.55318105e-02
-1.14996624e+00 -2.41624177e-01 2.11330071e-01 7.45081186e-01
4.96286571e-01 -3.49178731e-01 5.68380535e-01 -1.62997687e+00
6.20202839e-01 -7.58680582e-01 -1.71424225e-01 -2.88031567e-02
-8.46747160e-01 1.36308327e-01 1.08576024e+00 -5.00182390e-01
-2.26204589e-01 5.61062619e-02 -3.78150016e-01 -6.63929224e-01
-4.10438657e-01 7.62998080e-03 1.35841712e-01 -2.28677616e-01
1.16720736e+00 -1.25119731e-01 7.77227953e-02 -3.98176491e-01
1.38541743e-01 5.39838910e-01 5.05623639e-01 -2.01512545e-01
1.16169894e+00 5.81328094e-01 3.56645823e-01 -2.04411075e-01
-8.74072731e-01 1.25564083e-01 -3.30346972e-01 2.57136524e-01
9.46963370e-01 -6.23686612e-01 -7.55927801e-01 1.20146491e-01
-1.00879610e+00 -2.60274738e-01 -5.28601766e-01 2.59042442e-01
-5.08735716e-01 3.27357501e-01 -7.40512669e-01 -7.77913630e-01
-8.02744627e-01 -1.11791563e+00 1.02062595e+00 -3.99987310e-01
-8.55529666e-01 -1.11955416e+00 1.89076573e-01 2.60460466e-01
5.22982299e-01 1.16276014e+00 1.06523967e+00 -1.63647187e+00
-3.53991896e-01 -8.04398835e-01 2.25232258e-01 4.44855422e-01
7.24526271e-02 -2.70498037e-01 -1.38026941e+00 -3.90868694e-01
2.33734474e-01 -4.45363998e-01 6.08098567e-01 1.65392965e-01
1.37029612e+00 -7.45425761e-01 -2.63111770e-01 7.61667967e-01
1.18499374e+00 3.25971060e-02 7.40163624e-01 2.08512396e-01
8.16760004e-01 2.91322768e-01 2.97865085e-02 3.34098458e-01
-9.40721557e-02 2.47091502e-01 7.86071777e-01 -5.01587629e-01
2.85234779e-01 -7.02016205e-02 2.21514832e-02 6.58895150e-02
1.73097476e-01 -5.09828068e-02 -1.12230623e+00 2.86402822e-01
-1.49352086e+00 -1.05214417e+00 -2.11265638e-01 2.01670575e+00
7.60389686e-01 1.84439838e-01 -3.48641202e-02 3.17519933e-01
3.78273368e-01 -1.86820313e-01 -7.66010463e-01 -8.19998801e-01
1.41335249e-01 5.83731830e-01 5.52899301e-01 6.16356730e-01
-1.25640452e+00 3.85173112e-01 7.07150745e+00 3.23182344e-01
-8.86156619e-01 1.22766212e-01 1.13381457e+00 -2.32301876e-01
-4.25948054e-01 -2.68477738e-01 -1.63647801e-01 4.27382350e-01
1.28534353e+00 -9.21547562e-02 4.92649317e-01 1.06660259e+00
-7.89472535e-02 5.76626182e-01 -1.55936491e+00 7.01344967e-01
-1.09954692e-01 -1.51383948e+00 -3.38176452e-02 1.96748406e-01
6.71875715e-01 -4.68156766e-04 1.75675735e-01 2.09271207e-01
5.43553054e-01 -1.72814250e+00 3.95523965e-01 5.42698383e-01
7.70626962e-01 -8.97120237e-01 1.03093565e+00 3.20861757e-01
-3.94577235e-01 1.19558405e-02 -1.26241535e-01 6.38616756e-02
-1.57444701e-01 2.10398510e-01 -1.28112614e+00 3.41566384e-01
4.96381164e-01 2.71058708e-01 -3.98921520e-01 3.92442405e-01
-1.55976847e-01 1.00294387e+00 -2.41511002e-01 3.72411400e-01
1.83957055e-01 5.90588212e-01 8.00468087e-01 1.03374469e+00
-6.18588440e-02 5.24953082e-02 -1.13932930e-01 9.98399794e-01
-3.06025326e-01 -3.34069878e-01 -8.70736837e-01 6.82949126e-02
3.07652593e-01 1.17171693e+00 -2.22515389e-01 -2.98464209e-01
1.68009222e-01 9.85686421e-01 7.42014591e-03 -5.64038269e-02
-9.83896494e-01 -1.79439217e-01 8.70552301e-01 4.12009388e-01
4.47428226e-02 4.39883679e-01 -3.95939201e-01 -9.66165006e-01
-6.04811870e-02 -1.47966230e+00 6.21129751e-01 -3.56096119e-01
-1.55088711e+00 9.63517487e-01 -2.49961153e-01 -1.04367924e+00
-7.38205612e-01 -6.33395374e-01 -7.56667852e-01 9.97322738e-01
-1.02685094e+00 -1.12544918e+00 -1.33602947e-01 5.32942891e-01
-3.00632380e-02 -3.25561583e-01 1.37629974e+00 2.75924932e-02
-3.29586774e-01 1.12448549e+00 -9.12655666e-02 4.58852440e-01
4.15203512e-01 -1.53932524e+00 1.01209748e+00 8.07318628e-01
1.43112078e-01 7.43384778e-01 8.96124184e-01 -5.68076193e-01
-1.32854664e+00 -1.46088278e+00 6.63049400e-01 -1.29555285e+00
5.38791358e-01 -5.06421089e-01 -1.14081085e+00 1.08043897e+00
-6.47468939e-02 3.35701883e-01 1.03451908e+00 -2.35082097e-02
-6.37079954e-01 1.54086769e-01 -1.80317783e+00 4.77440000e-01
8.09427321e-01 -6.40703857e-01 -4.24022347e-01 5.68649530e-01
8.66845787e-01 -5.04793108e-01 -1.17579556e+00 4.02137160e-01
5.45155644e-01 -1.05491364e+00 9.98403072e-01 -1.44840419e+00
4.76237595e-01 4.62670662e-02 3.36863799e-03 -1.00270414e+00
-2.63441473e-01 -1.21072483e+00 -5.39641023e-01 7.46966600e-01
5.17443657e-01 -9.49299216e-01 9.05317545e-01 1.16509759e+00
1.61934122e-01 -9.76986587e-01 -1.11462891e+00 -5.73012292e-01
2.74788231e-01 -7.34411359e-01 8.87173295e-01 1.23531055e+00
-2.66726226e-01 -1.43662304e-01 -6.11389399e-01 4.57385331e-01
6.85712576e-01 -2.88050801e-01 7.91761756e-01 -1.02629733e+00
-7.83801019e-01 -1.07235074e-01 -9.60552096e-01 -1.66492984e-02
-3.03444378e-02 -9.75850284e-01 -4.09357697e-01 -1.01669014e+00
3.23985159e-01 -4.91915017e-01 -4.26744759e-01 6.97513759e-01
-6.00144267e-01 3.86353880e-01 2.40561292e-01 2.49258518e-01
-3.27055961e-01 -9.34686810e-02 6.81740999e-01 -1.91708848e-01
2.17818260e-01 1.73744231e-01 -9.77065682e-01 7.09872127e-01
8.06090117e-01 -1.17623293e+00 -1.98824152e-01 -1.28519669e-01
1.67113006e-01 -4.27427515e-02 7.44254231e-01 -1.07583475e+00
-3.03702891e-01 2.80822933e-01 6.06292605e-01 -4.16231938e-02
-3.64794694e-02 -9.92629468e-01 5.10340512e-01 1.26786113e+00
-6.93163693e-01 3.16071361e-01 3.54544014e-01 8.60832632e-01
3.31712246e-01 7.18190596e-02 8.32602382e-01 -2.13082135e-01
2.57262588e-01 5.62622488e-01 -2.54321337e-01 3.66880715e-01
1.22215092e+00 1.62732258e-01 -3.61549675e-01 -6.19292080e-01
-9.34358478e-01 7.02883350e-03 5.36236286e-01 2.65164644e-01
6.48695529e-01 -9.66364443e-01 -1.03203523e+00 4.71357048e-01
-2.05736281e-03 -2.79852331e-01 1.32997915e-01 5.62277615e-01
-6.60968602e-01 1.93781391e-01 2.12834515e-02 -4.85220134e-01
-1.50765479e+00 7.95479655e-01 7.82730162e-01 -6.09432995e-01
-6.70007527e-01 7.31186867e-01 1.86363548e-01 -5.74516535e-01
1.33606642e-01 -2.72589922e-01 4.46663886e-01 -7.83220053e-01
6.67471051e-01 2.16187358e-01 7.35083371e-02 -2.97327459e-01
-5.88853121e-01 -1.17953829e-01 -4.20058727e-01 3.15240473e-01
1.28971577e+00 6.33793354e-01 1.56943545e-01 2.94203721e-02
1.24986422e+00 -3.27419117e-02 -8.92609298e-01 2.72075415e-01
-8.99013579e-02 -2.32625946e-01 -2.73552865e-01 -1.10836005e+00
-8.39192271e-01 9.79677796e-01 7.70058870e-01 7.04788864e-01
8.00072730e-01 1.08376127e-02 8.74040782e-01 2.80115038e-01
7.59924203e-02 -1.99317589e-01 -2.99383909e-01 -4.96597514e-02
8.11042845e-01 -1.24498570e+00 2.16246679e-01 -6.20857887e-02
-6.27420008e-01 9.44081843e-01 1.89460039e-01 -2.92304635e-01
5.93887091e-01 4.28275287e-01 3.74472886e-01 -3.88810277e-01
-6.87405944e-01 7.06484437e-01 3.39965820e-01 7.40435600e-01
1.51961237e-01 3.91789675e-01 3.88475060e-01 8.18577886e-01
-3.37342709e-01 -3.56199950e-01 5.77598810e-01 7.67478466e-01
1.34322405e-01 -1.11310065e+00 -3.20506066e-01 8.67372215e-01
-1.17151332e+00 -3.42683971e-01 -8.30047250e-01 9.66077507e-01
-1.19740248e-01 8.43816221e-01 -1.89822733e-01 -6.10478044e-01
3.15091044e-01 1.83535576e-01 2.02695370e-01 -4.14253175e-01
-1.32585347e+00 -6.36750877e-01 2.07199946e-01 -9.00417924e-01
1.65981680e-01 -4.77617055e-01 -1.06727505e+00 -5.80624104e-01
-1.11279376e-01 -4.22078930e-02 3.34308803e-01 9.79824364e-01
7.06003189e-01 6.32875144e-01 6.56075180e-01 -3.00878435e-01
-1.00224793e+00 -7.51417696e-01 -3.75815779e-01 9.90061462e-01
8.57331932e-01 -2.56588100e-03 -6.85522795e-01 -5.13805151e-02] | [5.844677448272705, 7.605072975158691] |
0419e254-f78c-4d04-8c74-291a24ba19e2 | span-identification-of-epistemic-stance | 2306.02038 | null | https://arxiv.org/abs/2306.02038v1 | https://arxiv.org/pdf/2306.02038v1.pdf | Span Identification of Epistemic Stance-Taking in Academic Written English | Responding to the increasing need for automated writing evaluation (AWE) systems to assess language use beyond lexis and grammar (Burstein et al., 2016), we introduce a new approach to identify rhetorical features of stance in academic English writing. Drawing on the discourse-analytic framework of engagement in the Appraisal analysis (Martin & White, 2005), we manually annotated 4,688 sentences (126,411 tokens) for eight rhetorical stance categories (e.g., PROCLAIM, ATTRIBUTION) and additional discourse elements. We then report an experiment to train machine learning models to identify and categorize the spans of these stance expressions. The best-performing model (RoBERTa + LSTM) achieved macro-averaged F1 of .7208 in the span identification of stance-taking expressions, slightly outperforming the intercoder reliability estimates before adjudication (F1 = .6629). | ['Kristopher Kyle', 'Masaki Eguchi'] | 2023-06-03 | null | null | null | null | ['automated-writing-evaluation'] | ['natural-language-processing'] | [ 9.19898227e-02 5.98314524e-01 -5.44675291e-01 -3.40560526e-01
-1.09415925e+00 -9.60896671e-01 9.94324744e-01 7.51080036e-01
-6.13514960e-01 8.62158239e-01 9.39932883e-01 -9.79426861e-01
-1.02670407e-02 -3.07161152e-01 -2.81147033e-01 -4.38482836e-02
4.65778857e-01 -3.12393475e-02 -3.71639132e-01 -1.43548876e-01
1.06015658e+00 1.05001358e-02 -1.03490162e+00 5.97346425e-01
1.12780595e+00 5.97129822e-01 -1.73766971e-01 9.03745830e-01
-1.82777554e-01 1.73813748e+00 -1.31815100e+00 -8.27573299e-01
-6.37140572e-01 -5.00988185e-01 -1.15081751e+00 -2.58769333e-01
6.35097206e-01 -3.90747160e-01 4.13119793e-03 7.03429163e-01
5.81373572e-01 -4.67848666e-02 8.62130940e-01 -4.70425278e-01
-1.04516435e+00 1.18330336e+00 -3.96688193e-01 6.22914791e-01
6.08234942e-01 5.07227182e-02 1.18420029e+00 -1.12654889e+00
7.87288189e-01 1.15494800e+00 6.86437547e-01 3.45035464e-01
-9.50722218e-01 -6.94098353e-01 -6.60735667e-02 2.74228007e-01
-7.54462957e-01 -7.54445791e-01 7.40191877e-01 -1.10616982e+00
8.62532854e-01 2.28743032e-01 6.74021959e-01 1.47162306e+00
2.10186109e-01 6.18128896e-01 1.79923499e+00 -7.65949845e-01
8.20583701e-02 3.19837093e-01 7.74315119e-01 4.69033808e-01
1.17005847e-01 -5.03761351e-01 -7.46697068e-01 -3.39347929e-01
-1.03843268e-02 -7.58875668e-01 1.45998836e-01 8.11871111e-01
-1.11220622e+00 9.91467535e-01 1.33695155e-02 4.79121178e-01
-3.20341080e-01 -1.11659840e-01 9.10901904e-01 2.82341689e-01
8.37919176e-01 9.18803811e-01 -1.07650079e-01 -9.55377400e-01
-1.14305544e+00 3.76778305e-01 8.43333721e-01 5.12912393e-01
-2.44343895e-02 2.39725456e-01 -7.69597352e-01 1.18605208e+00
1.77154556e-01 4.34968799e-01 7.17738152e-01 -1.09030843e+00
6.84829116e-01 5.00873446e-01 -3.26487832e-02 -9.29416478e-01
-3.44609201e-01 -4.01351392e-01 -3.48250061e-01 -3.20424974e-01
4.93400663e-01 -3.85706633e-01 -2.25888938e-01 1.66825402e+00
-3.53397369e-01 -7.15402484e-01 2.05866676e-02 3.70440006e-01
1.12240767e+00 4.32145327e-01 6.01006806e-01 -5.71232975e-01
1.56191552e+00 -6.76574290e-01 -1.02696133e+00 -2.25787133e-01
1.01350975e+00 -1.05147409e+00 1.42177045e+00 2.90141493e-01
-1.50354934e+00 -4.52527106e-01 -1.20672095e+00 -3.62750918e-01
-3.41671519e-02 3.09831142e-01 1.91397712e-01 6.78062499e-01
-7.95597196e-01 1.94725037e-01 -2.74767637e-01 1.04537696e-01
6.24716818e-01 -5.20438075e-01 2.25991923e-02 6.47371292e-01
-1.24614477e+00 1.32306540e+00 1.78081512e-01 -3.60094719e-02
-6.60801232e-01 -6.32585764e-01 -7.13764250e-01 -2.83471048e-01
1.18945032e-01 2.63224453e-01 1.55192792e+00 -9.16138530e-01
-1.39114678e+00 1.47763324e+00 -2.34559581e-01 -3.71108800e-01
2.97161669e-01 -5.04631519e-01 -2.16804653e-01 -4.85121682e-02
4.49400187e-01 3.34771611e-02 5.15206873e-01 -6.81633174e-01
-3.01284760e-01 -4.64979038e-02 1.95737720e-01 2.66770422e-01
-7.71153331e-01 7.90630400e-01 9.66172576e-01 -7.75350511e-01
-3.59385014e-01 -5.22182345e-01 5.53765595e-01 -4.86660033e-01
-5.19317269e-01 -9.81801331e-01 5.30295730e-01 -1.30924833e+00
1.79336941e+00 -1.72087479e+00 -1.34836048e-01 -6.07238114e-02
5.91290534e-01 3.23776573e-01 3.83523524e-01 5.36372781e-01
-1.81700557e-01 5.68301857e-01 -1.96948633e-01 -2.83198416e-01
2.10453734e-01 -1.63869143e-01 -5.21746337e-01 4.67312872e-01
5.34064062e-02 1.13442266e+00 -1.02518427e+00 -6.31039619e-01
-3.46717000e-01 4.56832647e-02 -1.51961043e-01 1.92163363e-01
-6.18088357e-02 1.88440427e-01 -2.72690486e-02 5.60400605e-01
1.97141450e-02 -1.06869690e-01 1.44755840e-01 4.38772887e-01
-6.16393447e-01 1.25276399e+00 -2.77084649e-01 1.00416052e+00
-5.13604581e-01 1.54831100e+00 -1.17577441e-01 -6.88720882e-01
1.09268391e+00 4.90428060e-01 -1.02090146e-02 -6.96384490e-01
2.18931839e-01 5.07420540e-01 6.20509863e-01 -6.74214005e-01
6.55896366e-01 -1.10559590e-01 -4.75760639e-01 1.00565445e+00
-1.80001080e-01 -5.32645471e-02 2.44633958e-01 8.32854062e-02
9.24005806e-01 -9.30720195e-02 7.27023244e-01 -6.98860586e-01
5.31376719e-01 -1.45399764e-01 1.88394651e-01 6.25899732e-01
-4.21036273e-01 7.75986984e-02 9.85394537e-01 -2.35264242e-01
-1.04114044e+00 -7.24375069e-01 -4.72334176e-01 1.37099397e+00
-8.68923008e-01 -5.99936962e-01 -9.96189237e-01 -5.81621826e-01
-2.71155417e-01 1.35193384e+00 -6.37052894e-01 -1.17630579e-01
-8.57330143e-01 -2.92710125e-01 9.95644510e-01 4.69278485e-01
4.71135527e-01 -1.39817727e+00 -1.00262356e+00 2.65032440e-01
-6.03238881e-01 -9.79431152e-01 -1.73088059e-01 -4.34926860e-02
-2.52859384e-01 -1.00339937e+00 -3.71663570e-01 -4.82957691e-01
-2.15877052e-02 -3.68012995e-01 1.14998436e+00 3.19265991e-01
1.75403535e-01 1.99749053e-01 -4.60014582e-01 -7.90065944e-01
-9.93623972e-01 3.10847014e-01 -2.22008944e-01 -7.88021803e-01
7.20117211e-01 -1.91379800e-01 -1.31070852e-01 -4.50862586e-01
-3.73865694e-01 3.40843424e-02 1.72337517e-01 9.01977181e-01
-1.84697881e-01 -8.78400326e-01 9.35744524e-01 -9.25437868e-01
1.51251221e+00 -5.84996700e-01 -1.51443705e-01 1.60085410e-01
-7.64886916e-01 -5.57528734e-01 3.97073001e-01 -4.74967629e-01
-9.54445541e-01 -1.27037144e+00 -3.17980289e-01 3.84596616e-01
6.97895512e-02 9.09155548e-01 5.09910643e-01 3.51672053e-01
1.04348552e+00 -1.77376211e-01 4.14170586e-02 1.36811910e-02
1.46614671e-01 1.16330481e+00 4.94321585e-01 -9.35198069e-01
3.59597981e-01 -4.63093817e-01 -3.89783829e-01 -9.35440660e-01
-1.59544420e+00 7.03922361e-02 -4.73947763e-01 -6.01228356e-01
9.73060668e-01 -8.83422434e-01 -8.82333636e-01 2.46324524e-01
-1.45426238e+00 -6.96136951e-01 -2.12723717e-01 3.59276146e-01
-6.05648756e-01 2.96672106e-01 -8.90022218e-01 -1.07580912e+00
-6.22213781e-01 -8.25720251e-01 6.45417154e-01 -7.97164030e-05
-1.37232852e+00 -1.01802123e+00 2.83824414e-01 5.42266369e-01
2.90390700e-01 5.91889977e-01 1.25095820e+00 -9.67745781e-01
5.25524497e-01 9.68016237e-02 -1.60736114e-01 4.18036073e-01
-2.39864402e-02 1.76024318e-01 -1.02181172e+00 7.24505782e-02
3.47618967e-01 -8.81219685e-01 4.08179015e-01 2.44506255e-01
1.20054114e+00 -9.19250667e-01 2.92696834e-01 -1.34664074e-01
7.34423876e-01 2.36731008e-01 3.69495183e-01 6.28042758e-01
4.38154846e-01 6.88863218e-01 5.35093606e-01 4.69878405e-01
4.95438784e-01 4.37530190e-01 -2.16128200e-01 3.40985775e-01
-1.05414778e-01 -2.10532039e-01 7.52089381e-01 9.83727396e-01
-1.48660302e-01 -7.67630264e-02 -1.41232431e+00 5.63230693e-01
-1.40995860e+00 -1.09245038e+00 -3.92237395e-01 1.73366880e+00
1.37374043e+00 5.33563852e-01 1.88742951e-01 3.12570423e-01
4.32267547e-01 5.45550048e-01 -8.76143500e-02 -1.30190527e+00
-2.15600416e-01 4.12499756e-01 6.67783618e-02 7.27475703e-01
-8.23676229e-01 8.33189189e-01 6.72620153e+00 6.82844698e-01
-7.49600828e-01 4.47625309e-01 8.30889940e-01 6.48148954e-02
-4.10155177e-01 -2.25929379e-01 -5.93636870e-01 7.57497013e-01
1.44169235e+00 -5.03521740e-01 1.07834890e-01 6.53624594e-01
2.02684879e-01 -2.18536794e-01 -9.91956830e-01 3.08498502e-01
4.11408365e-01 -1.22992039e+00 -4.14604843e-01 3.67804617e-02
9.21494782e-01 -3.68691802e-01 3.93804051e-02 5.47964275e-01
3.95780325e-01 -1.21131253e+00 1.14891434e+00 4.07368690e-01
1.05375826e+00 -3.66413474e-01 7.24239707e-01 5.55334747e-01
-2.07501486e-01 -2.17497468e-01 -1.68115273e-02 -9.01022792e-01
3.85294482e-02 4.33150619e-01 -1.02678406e+00 -1.66300207e-01
2.87606269e-01 6.22571409e-01 -6.94884777e-01 8.62069726e-02
-8.54029655e-01 1.34736383e+00 2.29711190e-01 -5.51167130e-01
2.61853814e-01 1.62136465e-01 5.39253294e-01 1.52175641e+00
1.57393143e-02 1.48621574e-01 -2.47566238e-01 7.97243536e-01
-4.26624745e-01 2.44757205e-01 -5.32411754e-01 -4.37123626e-01
8.95335376e-01 1.26162815e+00 -5.61379850e-01 -6.54372990e-01
-1.63407072e-01 2.53850013e-01 7.48274088e-01 1.81258172e-01
-6.34721696e-01 -1.52941912e-01 1.46287367e-01 1.41750082e-01
-3.53007257e-01 -3.13405871e-01 -9.87316251e-01 -5.86739719e-01
8.96815434e-02 -1.10653651e+00 1.46330491e-01 -6.89882696e-01
-1.17792869e+00 2.19084755e-01 1.91000566e-01 -6.31878376e-01
-5.89712262e-01 -6.84920788e-01 -8.16788852e-01 1.06121886e+00
-1.10473549e+00 -1.16980553e+00 -2.23237574e-01 -2.58691847e-01
6.91802919e-01 -2.67162174e-01 9.20201302e-01 2.00441247e-03
-2.98975587e-01 6.04613662e-01 -3.15030426e-01 5.03788471e-01
9.91315961e-01 -1.44838750e+00 2.02524781e-01 6.49587512e-01
-2.39380166e-01 9.32824254e-01 6.91733658e-01 -6.62075520e-01
-4.78058875e-01 -6.55534208e-01 1.67277050e+00 -8.89963567e-01
1.01325250e+00 -3.41777354e-01 -9.49364960e-01 8.50090742e-01
8.42232108e-01 -9.82488573e-01 1.16103959e+00 3.99482608e-01
-4.51311469e-01 5.72496712e-01 -7.48671651e-01 6.32973611e-01
9.46773589e-01 -1.07138419e+00 -1.20342588e+00 5.72080135e-01
5.59770405e-01 -6.37497365e-01 -1.14254391e+00 2.21346900e-01
5.86444974e-01 -7.69825101e-01 6.17204010e-01 -8.64606798e-01
1.50157082e+00 4.32941884e-01 6.18250333e-02 -1.13622975e+00
-2.91700780e-01 -6.03345394e-01 -5.09455621e-01 1.47552931e+00
5.02606094e-01 -3.65352422e-01 6.54897243e-02 6.01137877e-01
-5.54405570e-01 -9.60796893e-01 -1.03393328e+00 -3.03292483e-01
9.52503145e-01 -2.76047766e-01 2.13354588e-01 1.26645100e+00
6.46423876e-01 6.77778125e-01 -2.41568480e-02 -6.87735081e-01
7.58360624e-02 -2.17921406e-01 4.59577948e-01 -1.28923392e+00
-5.98139539e-02 -1.12278354e+00 1.56294912e-01 -4.50760126e-01
8.11668813e-01 -1.03907704e+00 -1.45652890e-01 -1.61950541e+00
4.58635807e-01 -5.89547306e-02 4.45847362e-02 5.06857038e-01
-4.30125564e-01 2.05241628e-02 2.22298324e-01 2.46792004e-01
-3.61815333e-01 3.86850268e-01 8.51266205e-01 -2.40790583e-02
-2.00614370e-02 -3.87171149e-01 -1.18591046e+00 8.26430261e-01
8.89635742e-01 -1.63990036e-01 -8.65143463e-02 -3.43858868e-01
5.43571353e-01 -6.09880611e-02 3.59051198e-01 -5.27676105e-01
-6.20072782e-02 -1.40684068e-01 3.05991352e-01 -5.02111197e-01
5.85783832e-02 7.34284669e-02 -5.34685671e-01 4.76480097e-01
-1.15447056e+00 5.91434538e-01 2.23717555e-01 -4.04856890e-01
-9.18582827e-02 -8.33784938e-01 5.65088928e-01 3.46432514e-02
-1.95498988e-02 -6.40058279e-01 -9.24002111e-01 7.06795812e-01
8.73666346e-01 -9.03766379e-02 -9.64158177e-01 -5.49207807e-01
-2.95732796e-01 5.59171960e-02 3.35531197e-02 4.20125246e-01
4.29128826e-01 -1.17716515e+00 -1.19084132e+00 -3.48606884e-01
6.82652146e-02 -2.26995856e-01 -2.30214402e-01 9.33938086e-01
-3.51309180e-01 4.14392978e-01 1.97486002e-02 -1.84875965e-01
-1.37416911e+00 1.64167639e-02 1.15330731e-02 -4.29356575e-01
-2.77842581e-01 7.30126798e-01 -3.82908821e-01 1.90479774e-02
-1.88461632e-01 -5.63880662e-03 -6.22536719e-01 6.15873575e-01
5.97784460e-01 7.44168282e-01 -1.07490700e-02 -8.29905033e-01
2.57467963e-02 1.07918745e-02 -1.57533795e-01 -4.51985359e-01
8.83207977e-01 1.15562499e-01 -5.08508921e-01 1.25475073e+00
1.03961360e+00 4.64193851e-01 -3.89114916e-01 -7.38166198e-02
5.48028529e-01 -1.66054815e-01 -1.07638545e-01 -9.78958309e-01
2.84551710e-01 1.02357996e+00 -4.06562015e-02 5.85175395e-01
3.44284356e-01 -1.06567331e-01 3.14615309e-01 2.61114508e-01
-3.24591815e-01 -1.68443418e+00 3.04536968e-01 1.11275578e+00
1.33248544e+00 -8.93860996e-01 2.02271819e-01 -1.24184927e-03
-6.68695331e-01 1.33231807e+00 7.29065955e-01 6.73614144e-02
3.14596444e-01 1.15118049e-01 4.08872999e-02 -5.35382092e-01
-7.73362875e-01 5.28408825e-01 3.74170005e-01 1.00445285e-01
1.42279088e+00 2.86920607e-01 -1.32489848e+00 7.35126853e-01
-1.07131124e+00 -4.03233588e-01 8.90401065e-01 5.35467923e-01
-4.33966964e-01 -5.90897083e-01 -3.05205733e-01 8.72892618e-01
-9.71019447e-01 -2.70752251e-01 -7.57557034e-01 7.92954743e-01
-3.69424403e-01 1.02559984e+00 3.06814224e-01 -7.60921165e-02
1.30585313e-01 5.44433713e-01 1.15430206e-01 -7.21155643e-01
-1.12003696e+00 -1.74205303e-01 8.22337091e-01 2.32549328e-02
-6.37529135e-01 -1.06611621e+00 -9.25769031e-01 -4.16688025e-01
-4.88592684e-02 1.38836458e-01 3.19472581e-01 1.32672036e+00
-1.42732650e-01 7.29173064e-01 4.67169881e-01 -6.17076233e-02
-7.49439538e-01 -1.82715726e+00 2.72107664e-02 2.67913878e-01
3.30003321e-01 -6.25673354e-01 -2.73665279e-01 -1.67515092e-02] | [11.204071998596191, 9.424884796142578] |
add9c0bb-ae4d-45f1-a6dd-53d0c031cd88 | building-fast-and-compact-convolutional | 1702.07975 | null | http://arxiv.org/abs/1702.07975v1 | http://arxiv.org/pdf/1702.07975v1.pdf | Building Fast and Compact Convolutional Neural Networks for Offline Handwritten Chinese Character Recognition | Like other problems in computer vision, offline handwritten Chinese character
recognition (HCCR) has achieved impressive results using convolutional neural
network (CNN)-based methods. However, larger and deeper networks are needed to
deliver state-of-the-art results in this domain. Such networks intuitively
appear to incur high computational cost, and require the storage of a large
number of parameters, which renders them unfeasible for deployment in portable
devices. To solve this problem, we propose a Global Supervised Low-rank
Expansion (GSLRE) method and an Adaptive Drop-weight (ADW) technique to solve
the problems of speed and storage capacity. We design a nine-layer CNN for HCCR
consisting of 3,755 classes, and devise an algorithm that can reduce the
networks computational cost by nine times and compress the network to 1/18 of
the original size of the baseline model, with only a 0.21% drop in accuracy. In
tests, the proposed algorithm surpassed the best single-network performance
reported thus far in the literature while requiring only 2.3 MB for storage.
Furthermore, when integrated with our effective forward implementation, the
recognition of an offline character image took only 9.7 ms on a CPU. Compared
with the state-of-the-art CNN model for HCCR, our approach is approximately 30
times faster, yet 10 times more cost efficient. | ['Yafeng Yang', 'Xuefeng Xiao', 'Weixin Yang', 'Lianwen Jin', 'Jun Sun', 'Tianhai Chang'] | 2017-02-26 | null | null | null | null | ['offline-handwritten-chinese-character', 'offline-handwritten-chinese-character'] | ['computer-vision', 'natural-language-processing'] | [ 2.12891027e-01 -3.58140469e-01 -5.84958168e-03 -3.00818145e-01
-5.22936404e-01 -2.68572956e-01 1.93675473e-01 -6.02785647e-02
-9.46231723e-01 4.28847075e-01 -3.11469883e-01 -7.11282313e-01
5.49447276e-02 -7.06203401e-01 -7.07507551e-01 -6.37799919e-01
-5.55859804e-02 2.50645459e-01 2.66167104e-01 -9.45634991e-02
3.55145037e-01 9.32635367e-01 -1.26296711e+00 2.81340361e-01
7.30519295e-01 1.29610968e+00 3.31662357e-01 8.56940866e-01
1.29343331e-01 1.13029635e+00 -7.28737354e-01 -5.07731318e-01
2.52441466e-01 -2.87388768e-02 -7.34131813e-01 -7.70630408e-03
7.74024129e-01 -7.13044345e-01 -9.66648638e-01 8.96491885e-01
6.01165175e-01 1.32040083e-01 6.09929711e-02 -6.48510277e-01
-7.74019182e-01 7.05937088e-01 -3.06566954e-01 1.35611802e-01
-2.71121293e-01 -2.50990063e-01 7.69046605e-01 -1.10726476e+00
4.13789332e-01 8.07666063e-01 9.22412992e-01 7.08443284e-01
-7.67436028e-01 -6.53837860e-01 6.22287616e-02 2.53986955e-01
-1.49882424e+00 -4.79171365e-01 5.33308387e-01 1.47891879e-01
1.34821153e+00 2.86711335e-01 7.02263951e-01 7.51832008e-01
-1.58682764e-01 9.92635250e-01 8.70794237e-01 -4.15535748e-01
1.78099141e-01 -2.63044298e-01 2.85961300e-01 8.94171357e-01
4.56034452e-01 -3.57441396e-01 -5.19471467e-01 1.59441426e-01
9.23378944e-01 1.13741718e-01 -1.22323789e-01 3.29790771e-01
-1.05695033e+00 7.21296370e-01 5.56062520e-01 3.30772012e-01
-2.30143189e-01 4.90598530e-01 5.21053851e-01 2.50227422e-01
3.54798228e-01 5.03479898e-01 -5.59763491e-01 -4.21829939e-01
-1.18775308e+00 2.87017133e-03 9.24936354e-01 9.87498283e-01
4.91715670e-01 4.37131435e-01 1.91553950e-01 1.02748549e+00
-9.59365740e-02 3.59644622e-01 4.46615934e-01 -8.97769868e-01
5.54472804e-01 5.91324151e-01 -4.07261521e-01 -1.19717467e+00
-2.55795062e-01 -5.80340028e-01 -1.39720726e+00 -1.42460465e-01
5.47290266e-01 -3.52380425e-02 -9.69433367e-01 1.03489840e+00
-2.99389452e-01 -2.07675211e-02 9.09363776e-02 9.06828880e-01
5.86574674e-01 7.43393183e-01 -3.48013937e-01 1.29035875e-01
1.11005139e+00 -1.30992591e+00 -4.28488404e-01 -3.07406843e-01
7.60722756e-01 -7.90932953e-01 9.00890291e-01 6.24147832e-01
-1.08866823e+00 -4.73771662e-01 -1.36195505e+00 -1.85372293e-01
-3.18249762e-01 7.66111434e-01 9.16371286e-01 6.51616275e-01
-1.15258038e+00 9.60475683e-01 -9.45291340e-01 -1.44140542e-01
6.38053119e-01 6.46880448e-01 -3.35914493e-01 -4.28950042e-01
-7.62520492e-01 8.95705283e-01 4.87943262e-01 6.29732728e-01
-8.15628588e-01 -6.13768637e-01 -4.88070518e-01 2.63118356e-01
2.61674285e-01 1.73522964e-01 1.09852278e+00 -9.47482824e-01
-1.70481896e+00 4.77587461e-01 4.55625076e-03 -7.47598946e-01
5.37923753e-01 -3.46957207e-01 -2.73823231e-01 3.62879038e-01
-7.14174330e-01 5.48451960e-01 6.23133659e-01 -6.61677539e-01
-4.11172926e-01 -2.20916137e-01 -1.65867824e-02 -4.86330837e-02
-9.25007939e-01 2.92867601e-01 -9.88112330e-01 -7.46102870e-01
2.76598424e-01 -1.13462555e+00 -4.09269243e-01 2.56653965e-01
-1.59517556e-01 -1.23095214e-01 9.87492800e-01 -9.36264455e-01
1.24537206e+00 -2.09520268e+00 -2.53807276e-01 1.14121601e-01
2.56476700e-01 1.00771809e+00 -2.70404249e-01 1.02179840e-01
1.28051728e-01 5.66425659e-02 -8.97862092e-02 -4.47975248e-01
-3.51354927e-01 2.18820214e-01 -1.63280517e-01 6.26690626e-01
1.32188201e-01 8.35190892e-01 -5.53940535e-01 -2.36454442e-01
-1.02330903e-02 6.65917873e-01 -5.19606113e-01 -1.18942838e-02
1.28774717e-01 -3.25218350e-01 -2.05872297e-01 9.39502597e-01
6.90685153e-01 -5.21565497e-01 3.94821435e-01 -2.95942456e-01
-1.25581324e-01 1.30993292e-01 -9.58543837e-01 1.43648088e+00
-2.36826494e-01 1.27200067e+00 -1.38557166e-01 -1.14748907e+00
1.27669239e+00 5.69930486e-02 6.49250820e-02 -6.32575035e-01
2.68510699e-01 5.87959409e-01 1.84673414e-01 -9.84915495e-02
7.86973536e-01 5.82144260e-01 2.72258967e-01 2.10418463e-01
-7.77912661e-02 2.97281653e-01 1.51956916e-01 4.17727008e-02
1.23264837e+00 -3.08018833e-01 -2.39251375e-01 -2.10215300e-01
5.32501757e-01 -6.24848381e-02 4.58562046e-01 9.39081132e-01
-6.95048496e-02 6.44182265e-01 4.08312619e-01 -9.26130235e-01
-1.35725033e+00 -3.56295198e-01 1.43124878e-01 9.28703964e-01
-2.14464217e-01 -3.49465668e-01 -7.86187530e-01 -2.34004706e-01
-2.43865475e-01 7.50083029e-02 -2.05038711e-01 7.29287565e-02
-1.10834086e+00 -8.15656006e-01 1.18612075e+00 8.99437189e-01
1.11979973e+00 -8.34927261e-01 -6.33284748e-01 2.30926126e-01
1.36356413e-01 -1.43654180e+00 -2.97890514e-01 2.04103932e-01
-1.19042253e+00 -8.49032819e-01 -1.04605269e+00 -1.03055763e+00
9.67680037e-01 2.24980667e-01 9.13479745e-01 5.10799229e-01
-4.54422355e-01 -2.03121290e-01 -5.27153730e-01 -5.07332087e-02
-1.47877559e-01 4.02439237e-01 -2.28491928e-02 -1.77949801e-01
2.77718693e-01 -1.00570939e-01 -5.84629059e-01 1.24464065e-01
-8.16381216e-01 1.00986831e-01 8.53471220e-01 1.12275386e+00
3.42749327e-01 5.56039475e-02 2.98315227e-01 -6.47652447e-01
5.46902001e-01 2.38909483e-01 -8.85549724e-01 3.93665552e-01
-8.17019761e-01 -1.03648603e-01 9.95683074e-01 -8.39689016e-01
-7.01469123e-01 2.37538964e-01 -1.77507624e-02 -6.60023272e-01
2.42691755e-01 7.69587576e-01 2.82365143e-01 -5.48625648e-01
4.72672701e-01 5.05115807e-01 -5.02311997e-02 -4.77866977e-01
8.71785879e-02 8.74366462e-01 6.63731396e-01 -4.87938315e-01
5.09451389e-01 2.82010436e-01 -4.09457320e-03 -1.07407701e+00
-6.34159327e-01 -2.22638026e-01 -4.72345471e-01 -1.18880816e-01
4.97900814e-01 -1.07491231e+00 -9.78901684e-01 1.22336495e+00
-1.30320287e+00 -5.63960016e-01 1.21921130e-01 4.64191437e-01
-2.76281741e-02 5.17352045e-01 -1.18230104e+00 -6.75878823e-01
-7.08150864e-01 -1.12780190e+00 5.48561215e-01 6.84580132e-02
3.40876400e-01 -6.68383121e-01 -6.86843932e-01 3.75531435e-01
7.86317110e-01 -1.37252256e-01 5.54948509e-01 -6.88443601e-01
-7.93230414e-01 -6.63856924e-01 -7.44235098e-01 8.67535293e-01
-3.04392993e-01 6.78317770e-02 -7.87940204e-01 -6.44132793e-01
-3.11346203e-01 -4.97729331e-01 1.09915721e+00 8.96044970e-02
1.69896889e+00 -3.61622900e-01 1.12902366e-01 7.70219088e-01
1.60667598e+00 3.05875242e-01 9.56941903e-01 2.94037014e-01
9.19652998e-01 1.58176407e-01 3.54493409e-01 6.03337526e-01
7.58885071e-02 3.41699928e-01 2.54981339e-01 -2.09499165e-01
-1.24896809e-01 -8.19630474e-02 2.52693653e-01 1.41685069e+00
-4.85171229e-01 -2.48968616e-01 -1.16349471e+00 4.58065152e-01
-1.73342049e+00 -6.03492975e-01 3.08544980e-03 2.03817630e+00
8.57253611e-01 1.74493000e-01 -4.35884714e-01 2.34450266e-01
5.86661339e-01 1.54978991e-01 -5.69432378e-01 -4.18040693e-01
-2.63310909e-01 3.24242234e-01 1.08746779e+00 2.00245067e-01
-1.02964640e+00 1.11592567e+00 6.48041630e+00 9.95078504e-01
-1.35680342e+00 -2.45626688e-01 9.59644437e-01 4.90939021e-02
3.21489990e-01 -9.83190387e-02 -1.04509866e+00 1.30876467e-01
1.14605653e+00 3.11329663e-01 7.05790520e-01 1.08464718e+00
-1.25733808e-01 6.49088621e-02 -9.68279481e-01 1.22803438e+00
4.13244396e-01 -1.77410102e+00 5.34529723e-02 1.78164020e-01
7.21729398e-01 2.30988741e-01 -5.39755262e-02 1.31607249e-01
-5.82176968e-02 -1.18108761e+00 5.46923637e-01 3.95017177e-01
1.09610426e+00 -9.06925857e-01 1.07582498e+00 2.62077034e-01
-1.13946033e+00 -2.86563277e-01 -8.33167613e-01 -2.18069449e-01
-3.62254202e-01 5.88063896e-01 -4.21337277e-01 8.40972811e-02
7.93996334e-01 5.90457261e-01 -6.93742812e-01 9.27757025e-01
5.29053062e-02 6.89459622e-01 -3.18019331e-01 -5.45940399e-01
4.52096343e-01 8.13375190e-02 -3.50714773e-02 1.33357716e+00
3.32875878e-01 2.13587657e-01 -1.48818344e-01 3.95786881e-01
-6.05099440e-01 -1.78011939e-01 -2.17074335e-01 -3.74394178e-01
6.25769317e-01 1.15429914e+00 -8.28252971e-01 -5.53373635e-01
-3.17091852e-01 1.21904695e+00 5.52922070e-01 9.21454951e-02
-8.08455944e-01 -9.03149724e-01 3.09716195e-01 -3.70088786e-01
5.20803809e-01 -5.50868988e-01 -2.37030327e-01 -1.28323448e+00
2.96399117e-01 -1.14964652e+00 -1.20392039e-01 -4.04726535e-01
-1.05777478e+00 7.77980268e-01 -5.55871665e-01 -1.07580698e+00
1.97037920e-01 -1.19966936e+00 -2.78324127e-01 6.07482255e-01
-1.61932790e+00 -9.42696929e-01 -5.60267866e-01 4.84327525e-01
6.47695482e-01 -4.08041716e-01 9.95971978e-01 6.77039802e-01
-8.64279807e-01 1.13469243e+00 2.77524710e-01 7.17440188e-01
3.30797672e-01 -8.23994637e-01 5.75263977e-01 9.44788933e-01
-1.11038640e-01 6.92486346e-01 1.36317506e-01 -3.15951526e-01
-1.90599668e+00 -1.08855999e+00 1.09312260e+00 2.42724642e-01
6.62661850e-01 -4.26653951e-01 -8.99223745e-01 3.10247570e-01
-7.99300969e-02 4.03373569e-01 4.47811604e-01 -5.82044870e-02
-6.52445316e-01 -3.19768906e-01 -7.69688547e-01 6.66347086e-01
8.81850004e-01 -5.92744231e-01 -8.40171576e-02 3.28155428e-01
7.04774082e-01 -6.56975746e-01 -9.39744771e-01 1.77109495e-01
7.31781006e-01 -5.96854270e-01 8.80882680e-01 -3.62363100e-01
6.64314866e-01 -6.74151257e-02 -3.53985757e-01 -5.93467712e-01
-2.80037731e-01 -6.63931131e-01 -3.13442886e-01 7.81758308e-01
5.13875782e-01 -5.64210832e-01 1.08985150e+00 7.44780898e-01
-2.53036380e-01 -1.13041818e+00 -7.97129810e-01 -9.53779399e-01
8.54298938e-03 -4.37819749e-01 2.60233283e-01 8.27635884e-01
-2.20701471e-01 -2.66947150e-02 -5.85705221e-01 -1.42307756e-02
4.68442649e-01 -6.98362738e-02 4.85044479e-01 -9.56201851e-01
-1.48015440e-01 -3.43376011e-01 -4.64362442e-01 -1.43785405e+00
3.93991470e-02 -7.03987360e-01 3.52517068e-02 -1.23183894e+00
2.13204101e-01 -6.28764570e-01 -1.68202147e-01 7.70213544e-01
2.66484112e-01 5.67511559e-01 6.23650551e-01 2.99331546e-01
-5.12115180e-01 3.86930943e-01 1.00537503e+00 -2.68318623e-01
6.26747385e-02 -3.34531307e-01 -4.07931358e-01 6.91734612e-01
8.61087263e-01 -1.97144985e-01 4.59223939e-03 -8.72667313e-01
1.46228999e-01 -3.78429294e-02 1.75223798e-01 -1.23506320e+00
8.82846177e-01 2.72806853e-01 6.48414314e-01 -7.21356750e-01
4.80274349e-01 -7.41673708e-01 -3.36499244e-01 7.39400506e-01
-4.49265987e-01 2.92968005e-01 2.02731818e-01 2.07378879e-01
-3.33715022e-01 -4.01299268e-01 8.43171597e-01 -6.80453032e-02
-6.57955229e-01 3.70610386e-01 -5.85283756e-01 -2.88259000e-01
5.75031698e-01 -3.65778625e-01 -5.53306699e-01 -6.39417544e-02
-2.66008824e-01 -1.91789106e-01 4.76882048e-02 1.95579767e-01
1.09388423e+00 -1.05602300e+00 -6.55574441e-01 1.29652262e-01
-3.56110334e-01 9.06160772e-02 2.55268365e-01 5.49349010e-01
-1.33033454e+00 6.75641537e-01 -6.43581226e-02 -3.96284938e-01
-1.41243017e+00 9.06196088e-02 2.46038899e-01 -4.86203343e-01
-7.55176306e-01 9.26175714e-01 -7.98970163e-01 -2.85050243e-01
6.79630816e-01 -2.60742366e-01 5.27282432e-02 -1.39875799e-01
7.15992928e-01 6.75136685e-01 3.28964770e-01 -2.38582805e-01
-2.42912844e-01 4.74241138e-01 -3.84954005e-01 2.54394740e-01
1.50588357e+00 5.08578300e-01 -3.05509567e-01 -9.27817151e-02
1.43824959e+00 -5.17161369e-01 -1.28202844e+00 -2.49509782e-01
-5.88442944e-02 -4.01450247e-01 2.90364802e-01 -5.48726618e-01
-1.44455600e+00 8.29895020e-01 5.78977942e-01 -1.16843589e-01
1.32878506e+00 -4.40741241e-01 1.06063008e+00 1.26861548e+00
3.48170817e-01 -1.40719104e+00 2.04032198e-01 1.06480455e+00
6.29156709e-01 -1.15126598e+00 3.50491524e-01 -2.62589276e-01
-4.03262168e-01 1.60897005e+00 6.49993360e-01 -5.03249347e-01
6.26919031e-01 4.56004173e-01 -1.12024277e-01 -2.10615192e-02
-8.02141368e-01 3.67083132e-01 1.12463444e-01 1.22073323e-01
3.42874974e-01 7.20884576e-02 -1.51853547e-01 2.72189766e-01
2.99173091e-02 1.28321186e-01 6.51233971e-01 1.08454180e+00
-2.42523462e-01 -8.03168297e-01 5.77997155e-02 5.51856995e-01
-7.17252076e-01 -3.61065894e-01 -1.40443206e-01 6.38991833e-01
-2.80830562e-01 7.71996379e-01 3.13138247e-01 -6.47138774e-01
1.70729011e-01 -3.16549033e-01 3.60431820e-01 -9.59380716e-02
-6.93419099e-01 -1.52528077e-01 3.81936394e-02 -6.15550280e-01
-3.83872837e-01 -2.61786729e-01 -1.04609048e+00 -7.16662347e-01
-6.58423364e-01 -3.43848079e-01 9.52334821e-01 7.51259089e-01
2.96357155e-01 4.53594416e-01 5.23488760e-01 -8.31049621e-01
-8.43343794e-01 -9.40546870e-01 -3.88576567e-01 -1.90912560e-01
1.04637012e-01 1.50463641e-01 -2.86141694e-01 -3.56335640e-02] | [11.673197746276855, 2.612318515777588] |
129bd941-d5c6-4548-95e4-e10eb8a10e69 | counterexample-guided-abstraction-refinement-1 | 2301.08687 | null | https://arxiv.org/abs/2301.08687v1 | https://arxiv.org/pdf/2301.08687v1.pdf | Counterexample Guided Abstraction Refinement with Non-Refined Abstractions for Multi-Agent Path Finding | Counterexample guided abstraction refinement (CEGAR) represents a powerful symbolic technique for various tasks such as model checking and reachability analysis. Recently, CEGAR combined with Boolean satisfiability (SAT) has been applied for multi-agent path finding (MAPF), a problem where the task is to navigate agents from their start positions to given individual goal positions so that the agents do not collide with each other. The recent CEGAR approach used the initial abstraction of the MAPF problem where collisions between agents were omitted and were eliminated in subsequent abstraction refinements. We propose in this work a novel CEGAR-style solver for MAPF based on SAT in which some abstractions are deliberately left non-refined. This adds the necessity to post-process the answers obtained from the underlying SAT solver as these answers slightly differ from the correct MAPF solutions. Non-refining however yields order-of-magnitude smaller SAT encodings than those of the previous approach and speeds up the overall solving process making the SAT-based solver for MAPF competitive again in relevant benchmarks. | ['Pavel Surynek'] | 2023-01-20 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [ 2.67584324e-01 6.11525178e-01 8.31535012e-02 5.87220900e-02
-6.82554543e-01 -6.80651486e-01 5.91491938e-01 6.95371687e-01
-2.86256611e-01 1.36572039e+00 -1.72817513e-01 -4.79447991e-01
-3.82742971e-01 -1.17345643e+00 -6.70500100e-01 -4.31528658e-01
-5.73862553e-01 1.19668770e+00 9.88019764e-01 -5.39134026e-01
3.44880261e-02 4.16115284e-01 -1.52205563e+00 2.41175815e-01
5.08423746e-01 4.60898072e-01 -2.42396846e-01 8.50571692e-01
-1.90966085e-01 6.74091280e-01 -5.38934886e-01 -5.80707975e-02
4.37093049e-01 -5.16787708e-01 -1.35040092e+00 -1.55309156e-01
5.96240573e-02 -8.85186270e-02 -6.35991618e-02 1.22386360e+00
-2.98678190e-01 -3.08098085e-03 2.04621479e-01 -2.18225098e+00
4.72595841e-01 9.31964219e-01 -4.21779990e-01 -1.16764076e-01
8.50061536e-01 1.70282006e-01 9.54047382e-01 2.04539463e-01
7.67269969e-01 1.42487121e+00 4.75040793e-01 5.12171805e-01
-1.35873556e+00 -4.51367438e-01 5.20757973e-01 4.11276877e-01
-1.43539214e+00 -9.62843448e-02 2.03505352e-01 -2.39981562e-01
1.54727256e+00 8.62697363e-01 8.95735979e-01 4.36934412e-01
3.21813971e-01 1.84620336e-01 1.21573031e+00 -4.84914750e-01
5.27638972e-01 -5.59716262e-02 3.66784811e-01 7.30957091e-01
6.98104501e-01 3.29323202e-01 -8.68909210e-02 -5.46084404e-01
5.33500910e-01 -2.85936534e-01 -7.11524412e-02 -4.80578005e-01
-1.30275047e+00 1.11228502e+00 2.61256427e-01 -7.90900812e-02
-2.15604603e-01 6.11742437e-01 5.74060202e-01 5.32206476e-01
-3.42987448e-01 7.64058292e-01 -4.22064811e-01 -7.09009692e-02
-7.03710854e-01 1.01684797e+00 1.10799992e+00 9.90974545e-01
7.86028445e-01 -1.74026817e-01 2.08210424e-02 -4.48617965e-01
1.94002673e-01 4.28639144e-01 -4.10417944e-01 -1.22251248e+00
4.37317461e-01 9.12819862e-01 4.77633774e-01 -7.36218750e-01
-4.66645449e-01 -1.53472140e-01 -4.47298199e-01 8.72500420e-01
6.90889657e-01 -9.97454002e-02 -7.65845478e-01 1.80385232e+00
6.96781933e-01 4.18313682e-01 3.06239575e-01 7.09775090e-01
3.07846934e-01 8.90511870e-01 -2.88442194e-01 -4.76291090e-01
1.32216740e+00 -9.12810028e-01 -3.88299942e-01 -2.83914894e-01
8.56970012e-01 -1.16425872e-01 1.97332054e-01 7.65114427e-01
-1.44254947e+00 3.05691868e-01 -1.21898842e+00 3.54012072e-01
-3.20255995e-01 -1.01930571e+00 8.75835299e-01 5.35487354e-01
-1.19694471e+00 1.72115013e-01 -9.20827091e-01 -4.93531227e-02
1.77674413e-01 1.01177621e+00 -4.82254952e-01 -3.27678919e-01
-1.09675062e+00 9.09754634e-01 5.75566471e-01 -5.52119613e-02
-9.31812525e-01 -2.67793328e-01 -1.11145246e+00 1.80907771e-01
1.35874474e+00 -6.41450047e-01 1.55585873e+00 -6.04856431e-01
-1.45214784e+00 3.11580151e-01 -4.36579078e-01 -9.22078252e-01
4.23238605e-01 4.47274178e-01 4.72885631e-02 3.52899753e-03
7.33424798e-02 3.62993240e-01 1.96305707e-01 -1.23569334e+00
-8.06162953e-01 -2.33152747e-01 8.30018401e-01 -2.53514051e-01
8.80610406e-01 1.52007371e-01 4.81991060e-02 3.84037346e-01
9.50504914e-02 -1.21562457e+00 -8.71635556e-01 -3.84816229e-01
-4.14760709e-01 -1.61933973e-01 5.75288892e-01 -8.35682824e-02
1.14812362e+00 -1.63716841e+00 6.91333413e-01 7.19681621e-01
4.19856966e-01 1.51409090e-01 -2.90222079e-01 9.01991189e-01
-9.88732558e-03 -6.84698001e-02 -2.05601975e-01 5.59060350e-02
2.82329708e-01 5.02959132e-01 -2.26120263e-01 5.30943573e-01
1.91836432e-01 8.62240434e-01 -1.08996725e+00 -4.87606972e-01
1.97069019e-01 -2.00792417e-01 -8.83977473e-01 -1.18325934e-01
-9.40511465e-01 1.58936620e-01 -3.03589284e-01 2.27922782e-01
8.37675393e-01 5.72175533e-02 2.97526360e-01 6.53036594e-01
-4.23244148e-01 5.34836769e-01 -1.69784677e+00 1.25491428e+00
-2.14745954e-01 2.09834158e-01 3.23046476e-01 -6.35394394e-01
5.66781163e-01 2.35680819e-01 1.42328188e-01 -6.88578904e-01
2.29072541e-01 2.69511282e-01 2.33099699e-01 8.49322230e-02
3.44939440e-01 -2.12357149e-01 -6.07739508e-01 3.76544893e-01
-5.58889270e-01 -3.54777724e-01 7.98718452e-01 3.07609081e-01
1.46927655e+00 -1.60744831e-01 6.16458952e-01 -2.88930863e-01
1.04124355e+00 7.82482207e-01 6.39906764e-01 8.37336242e-01
1.79515079e-01 8.69799033e-02 1.08379436e+00 -5.75801075e-01
-7.32086778e-01 -5.61775744e-01 5.37149370e-01 3.34735215e-01
5.08180976e-01 -1.11224425e+00 -8.26488972e-01 -5.93353033e-01
-1.19559862e-01 8.06198239e-01 -5.52028358e-01 -1.89780235e-01
-9.30736840e-01 2.44222302e-02 3.93503726e-01 3.16000134e-01
2.37488300e-01 -8.45302165e-01 -1.25829506e+00 4.82759476e-01
-1.45499527e-01 -8.87933671e-01 1.69744954e-01 2.50597149e-01
-4.63792443e-01 -1.37510943e+00 7.37294257e-02 -4.22770470e-01
6.88736677e-01 1.18916757e-01 7.51210093e-01 4.30856586e-01
1.18454278e-01 -1.63474277e-01 -2.90109307e-01 -4.13661033e-01
-5.63108504e-01 1.76148042e-01 -2.83596069e-01 -6.55306995e-01
6.69941008e-02 -2.28807598e-01 2.28716001e-01 2.78881788e-01
-5.64983130e-01 9.48466882e-02 5.36710173e-02 7.26162016e-01
5.84566414e-01 7.99850583e-01 1.78051502e-01 -8.55582178e-01
5.51536202e-01 -2.90229768e-01 -1.47382712e+00 4.09270674e-02
-1.69552013e-01 2.84078807e-01 8.44076335e-01 -2.28129491e-01
-3.60035270e-01 1.21398419e-01 1.71786800e-01 5.66328838e-02
-1.83628768e-01 6.94479644e-01 -8.48991275e-02 -2.70743698e-01
2.04523340e-01 -6.00267351e-02 -1.00026568e-02 1.48741990e-01
-1.64627075e-01 -1.10227667e-01 5.55809438e-01 -7.46245861e-01
9.64314103e-01 2.21360534e-01 9.19898868e-01 -7.89293647e-02
-2.74090379e-01 -8.53652209e-02 7.66727142e-03 1.09936856e-01
3.47934872e-01 -3.44940424e-01 -1.35958719e+00 1.06534064e-01
-1.29319668e+00 -7.16707528e-01 -5.26316643e-01 -4.15112963e-03
-6.13856792e-01 1.94182709e-01 -1.82229847e-01 -1.34276974e+00
2.37056807e-01 -1.46766531e+00 1.04309618e+00 -3.90667410e-04
-6.90252364e-01 -4.20363754e-01 4.66263831e-01 -1.12581775e-01
1.57030821e-01 7.83677757e-01 1.08776522e+00 -6.94354653e-01
-9.02317405e-01 -3.29760790e-01 -4.47992012e-02 -6.51968658e-01
-2.95733541e-01 -9.29868594e-02 -1.56190664e-01 -4.51685101e-01
-3.23258668e-01 2.12707862e-01 3.95532735e-02 2.22656995e-01
4.57642466e-01 -6.37618601e-01 -5.84769547e-01 4.18539234e-02
1.54681039e+00 2.67401576e-01 7.96063602e-01 8.85231793e-01
-4.01278883e-02 5.28564155e-01 7.21545696e-01 2.71917760e-01
5.01359701e-01 8.82263184e-01 7.52881765e-01 2.82267421e-01
2.66828597e-01 1.68641180e-01 2.50155538e-01 -3.19829136e-01
-2.02410415e-01 -4.44851428e-01 -1.11647999e+00 5.98444104e-01
-2.22046566e+00 -1.00667644e+00 -7.03229368e-01 2.23260045e+00
6.65320277e-01 5.65007925e-01 5.49083233e-01 5.80994666e-01
3.85806710e-01 -4.10035551e-01 -4.06762213e-02 -9.49338377e-01
3.19727838e-01 5.49640656e-01 8.11105430e-01 1.34015596e+00
-6.20119393e-01 8.66752326e-01 5.85505867e+00 2.04205707e-01
-7.45825589e-01 -2.48602517e-02 -2.86004543e-01 -2.14684587e-02
-4.70078252e-02 5.50847232e-01 -7.54216909e-01 6.50197193e-02
1.19547904e+00 -3.53533536e-01 7.91998088e-01 6.26241267e-01
-4.48278002e-02 -6.63564146e-01 -1.24392855e+00 2.35789359e-01
-4.00907218e-01 -1.26561093e+00 -3.89862180e-01 1.82494059e-01
6.03357911e-01 -4.26064759e-01 -4.02530611e-01 3.65543395e-01
6.99878633e-01 -1.19640815e+00 9.42238867e-01 2.48554791e-03
3.58290464e-01 -1.19683850e+00 1.14168382e+00 4.45638269e-01
-1.29772091e+00 -2.38859281e-01 1.13455288e-01 -7.55656779e-01
5.78501165e-01 -1.20347165e-01 -9.59085584e-01 8.33685994e-01
3.54558796e-01 -1.86001942e-01 -7.21082166e-02 1.37240386e+00
-2.43819997e-01 1.18491136e-01 -7.27809131e-01 -5.31585626e-02
7.30544448e-01 -6.40172809e-02 9.46802795e-01 8.13552976e-01
1.65534079e-01 2.82600641e-01 2.68776089e-01 8.25491428e-01
8.26935470e-01 -6.00407243e-01 -2.70566285e-01 1.76357180e-01
3.05122018e-01 6.39078796e-01 -9.93155062e-01 -4.06015038e-01
-2.31626183e-01 4.22428459e-01 -9.68445092e-03 2.27407560e-01
-1.03599417e+00 -4.40765053e-01 7.86892891e-01 4.74070698e-01
3.97498578e-01 -4.55624044e-01 -1.15390241e-01 -6.25540853e-01
-1.40818670e-01 -1.13445258e+00 5.05031943e-01 -5.53228676e-01
-3.48245680e-01 8.24913859e-01 7.06158936e-01 -6.30960405e-01
-6.98697984e-01 -3.31476450e-01 -5.68624377e-01 7.15252399e-01
-1.62298012e+00 -7.54273415e-01 -2.92219013e-01 6.28540754e-01
2.91030288e-01 4.13194776e-01 9.94618952e-01 -1.34812951e-01
-5.07028759e-01 1.73418894e-01 -6.59422815e-01 -5.61478376e-01
-1.99842200e-01 -1.15413547e+00 4.33952391e-01 1.14908767e+00
-4.49021041e-01 6.45127356e-01 1.16860747e+00 -8.85673761e-01
-1.69831765e+00 -9.04155433e-01 1.00285983e+00 -6.14342727e-02
6.33642673e-01 -3.23669583e-01 -6.33988738e-01 1.10927010e+00
2.40263253e-01 -3.36121470e-01 -1.07201777e-01 -1.98300943e-01
-1.64879739e-01 1.41621038e-01 -1.34174538e+00 6.97815537e-01
7.77890265e-01 1.21320523e-01 -4.84760433e-01 9.24510211e-02
6.80645764e-01 -8.02108705e-01 -2.81187445e-01 1.41077444e-01
1.92440897e-02 -8.44373882e-01 7.40176022e-01 -6.92406535e-01
5.19799534e-03 -1.13445759e+00 -5.12129664e-02 -1.08070993e+00
-3.47902954e-01 -1.13038683e+00 -3.25326234e-01 7.21031070e-01
3.31139833e-01 -1.15628803e+00 7.14215577e-01 7.13084221e-01
-1.26471490e-01 -6.22454643e-01 -1.29109395e+00 -9.46093738e-01
-2.41714835e-01 -2.26622924e-01 1.22408807e+00 4.41348404e-01
6.63194776e-01 1.73156098e-01 -5.15329540e-02 6.78789794e-01
5.64612627e-01 2.04796448e-01 9.05708909e-01 -1.49939668e+00
-4.64163870e-01 -3.56771231e-01 -5.07518530e-01 -3.70747507e-01
3.32130939e-01 -6.59922063e-01 1.04709141e-01 -1.71216941e+00
4.33357470e-02 -3.64550471e-01 3.09143215e-01 8.74814391e-01
5.25282085e-01 -7.19343200e-02 2.92079329e-01 -4.53319699e-01
-8.58539879e-01 -1.02081440e-01 1.10799122e+00 -2.24350348e-01
-2.25791901e-01 7.32643008e-02 -5.09574473e-01 4.34160262e-01
5.92025518e-01 -7.61012673e-01 -3.69371176e-01 4.63683233e-02
8.57187808e-01 6.70160711e-01 5.16133189e-01 -9.94906068e-01
3.55189204e-01 -7.58497059e-01 -6.26617551e-01 -3.60684007e-01
4.83399421e-01 -1.35816836e+00 9.70116675e-01 1.10188591e+00
-1.32029533e-01 2.86857933e-01 6.23211324e-01 2.10313722e-01
-1.64200842e-01 -5.33095062e-01 4.85557556e-01 -1.31071001e-01
-6.29052699e-01 -1.29722744e-01 -8.22432756e-01 -2.78376907e-01
1.61905825e+00 -4.43744838e-01 -3.84985954e-01 -4.41896051e-01
-6.93653345e-01 6.39366865e-01 6.48800135e-01 -4.33177590e-01
4.24280703e-01 -9.33060706e-01 -5.64890742e-01 1.59837112e-01
-1.05519257e-01 5.21859825e-01 -9.94169191e-02 1.12716603e+00
-6.97964191e-01 7.00477540e-01 -3.07281345e-01 -1.77241340e-01
-1.53889489e+00 9.12050843e-01 2.92657793e-01 -5.82966685e-01
-6.98751569e-01 6.30405426e-01 -7.33490884e-02 -9.75651592e-02
7.16169700e-02 -8.98663282e-01 9.22783017e-02 -5.23416281e-01
7.78171301e-01 6.13902986e-01 9.37878899e-03 -3.65684807e-01
-9.10432994e-01 3.47423613e-01 -1.09883688e-01 -2.81730980e-01
1.47340858e+00 6.14696406e-02 -4.16375577e-01 -1.27544597e-01
4.34415668e-01 1.15088895e-01 -7.24159122e-01 1.47642449e-01
3.22024105e-03 -3.99444014e-01 -3.72520350e-02 -7.32380688e-01
-7.53516853e-01 2.11731419e-01 -3.95309299e-01 6.08962715e-01
1.00339901e+00 8.43491182e-02 6.76711202e-01 5.08651912e-01
1.12349510e+00 -5.49446046e-01 -6.09138489e-01 7.27669358e-01
6.41416073e-01 -4.64378357e-01 1.77466989e-01 -9.75304782e-01
-3.72230172e-01 8.76016736e-01 6.32656336e-01 -3.88978422e-01
-4.25269663e-01 9.47732329e-01 -5.69435000e-01 -2.55776852e-01
-7.97885537e-01 -3.06385517e-01 -4.96103048e-01 6.29059792e-01
-6.22080088e-01 -2.42610779e-02 -3.11589658e-01 5.73870122e-01
-4.39729512e-01 1.22697487e-01 1.04613686e+00 1.32183194e+00
-4.42652971e-01 -1.45978105e+00 -7.83212483e-01 -5.22030070e-02
9.29494873e-02 2.54020602e-01 -6.28520131e-01 1.27946520e+00
5.36256991e-02 9.49224651e-01 -1.36010408e-01 -1.44689828e-01
3.59536231e-01 -1.40424162e-01 1.00057447e+00 -6.51728511e-01
-8.08360994e-01 -5.06835207e-02 7.68538833e-01 -9.20790255e-01
-1.58665672e-01 -8.02627325e-01 -1.79393756e+00 -6.16983652e-01
-5.65122306e-01 7.39859521e-01 1.01891033e-01 1.00675046e+00
-2.34413147e-02 6.74371958e-01 2.36102045e-02 -8.70379388e-01
-3.26803565e-01 -4.44493107e-02 -1.55709431e-01 -3.78889263e-01
5.77504277e-01 -8.33183587e-01 -2.38674089e-01 -5.37946522e-01] | [4.981617450714111, 1.9072625637054443] |
fa0bd870-0c45-4d0a-a6b1-05628fd12c45 | structural-restricted-boltzmann-machine-for | 2306.09628 | null | https://arxiv.org/abs/2306.09628v1 | https://arxiv.org/pdf/2306.09628v1.pdf | Structural Restricted Boltzmann Machine for image denoising and classification | Restricted Boltzmann Machines are generative models that consist of a layer of hidden variables connected to another layer of visible units, and they are used to model the distribution over visible variables. In order to gain a higher representability power, many hidden units are commonly used, which, in combination with a large number of visible units, leads to a high number of trainable parameters. In this work we introduce the Structural Restricted Boltzmann Machine model, which taking advantage of the structure of the data in hand, constrains connections of hidden units to subsets of visible units in order to reduce significantly the number of trainable parameters, without compromising performance. As a possible area of application, we focus on image modelling. Based on the nature of the images, the structure of the connections is given in terms of spatial neighbourhoods over the pixels of the image that constitute the visible variables of the model. We conduct extensive experiments on various image domains. Image denoising is evaluated with corrupted images from the MNIST dataset. The generative power of our models is compared to vanilla RBMs, as well as their classification performance, which is assessed with five different image domains. Results show that our proposed model has a faster and more stable training, while also obtaining better results compared to an RBM with no constrained connections between its visible and hidden units. | ['Santana Roberto', 'Pérez Aritz', 'Bidaurrazaga Arkaitz'] | 2023-06-16 | null | null | null | null | ['image-denoising', 'classification-1'] | ['computer-vision', 'methodology'] | [ 2.94122785e-01 2.21470118e-01 1.26508489e-01 -2.14898840e-01
-9.80813876e-02 -2.09222108e-01 7.79661834e-01 -3.07295114e-01
-5.75034261e-01 7.45696902e-01 -2.03036871e-02 -1.42297432e-01
8.39430094e-02 -1.04232299e+00 -7.11290777e-01 -1.36810756e+00
8.86428207e-02 4.64774966e-01 9.05659720e-02 3.87151004e-03
-2.04116106e-02 5.81132948e-01 -1.40924728e+00 2.37943426e-01
6.67002201e-01 9.54399765e-01 6.02614105e-01 4.58441228e-01
-1.17160782e-01 8.75567198e-01 -5.76810062e-01 -1.85297087e-01
-4.11945842e-02 -5.05796731e-01 -6.29659057e-01 1.31233662e-01
7.95379505e-02 -3.76311950e-02 -2.33007416e-01 1.01738858e+00
3.03529590e-01 4.13026869e-01 8.25686216e-01 -6.75805330e-01
-7.16738999e-01 2.75390565e-01 -1.92422211e-01 2.79634595e-01
-2.84969896e-01 1.41165424e-02 1.01552463e+00 -5.68530619e-01
4.23813790e-01 1.14479458e+00 2.48086423e-01 8.04169178e-01
-1.64327514e+00 -3.94999385e-01 1.91416815e-01 3.16094011e-01
-1.36033666e+00 -2.88489193e-01 8.35157573e-01 -4.49132264e-01
1.05691743e+00 3.89164299e-01 7.40926981e-01 1.07751405e+00
5.01914978e-01 4.28137273e-01 1.11699581e+00 -6.42583311e-01
6.43728852e-01 4.71757054e-01 1.40629783e-01 5.86000621e-01
9.56455097e-02 -2.82745101e-02 -1.60643190e-01 -2.22165897e-01
9.50763702e-01 9.36486199e-03 -4.07119453e-01 -4.91471738e-01
-8.18869412e-01 1.14674747e+00 7.99105585e-01 4.64086980e-01
-4.93393421e-01 2.94121712e-01 -2.11555064e-01 -1.62761569e-01
6.37438595e-01 1.86243013e-01 3.22477631e-02 4.79052424e-01
-9.47495341e-01 -2.81869799e-01 7.36449659e-01 3.62173945e-01
9.74141955e-01 3.51885915e-01 -6.77997097e-02 9.25687253e-01
4.58242118e-01 3.67960930e-01 7.23331392e-01 -5.80493271e-01
2.80942410e-01 5.58944404e-01 -2.21096635e-01 -9.14448142e-01
-1.63422957e-01 -5.46703637e-01 -1.17046940e+00 7.11441457e-01
5.87847531e-02 2.04460114e-01 -1.47769928e+00 1.74019730e+00
-8.84102937e-03 1.10511765e-01 6.65840879e-02 8.51520240e-01
6.35164082e-01 9.39835489e-01 1.15292393e-01 4.81314845e-02
1.13594818e+00 -7.55404472e-01 -5.65581381e-01 -5.23556054e-01
2.86198378e-01 -2.58701742e-01 7.88415790e-01 3.67163211e-01
-9.18369055e-01 -7.34127939e-01 -1.09419847e+00 -2.42529009e-02
-4.64543670e-01 5.67114614e-02 4.94275033e-01 6.20795548e-01
-1.28193188e+00 6.34481549e-01 -9.86172259e-01 2.36071702e-02
4.52502728e-01 5.21079481e-01 -3.70245844e-01 4.37820181e-02
-1.07689393e+00 1.12084770e+00 4.97377813e-01 5.59408247e-01
-1.03679430e+00 -9.62038562e-02 -8.74397755e-01 3.29262882e-01
-5.98633848e-02 -7.01015353e-01 4.24731076e-01 -1.04728651e+00
-1.35606098e+00 6.61491573e-01 -1.79389313e-01 -5.40598869e-01
3.80536199e-01 8.68335590e-02 -5.98010235e-02 1.06992282e-01
-4.84820485e-01 7.62485087e-01 1.14613736e+00 -1.48025763e+00
-2.73820966e-01 -3.16207886e-01 1.57533124e-01 9.19870511e-02
-3.60248953e-01 -3.76898497e-01 -4.43199664e-01 -4.54251170e-01
2.36233324e-01 -1.13088393e+00 -4.17440206e-01 -4.89095151e-01
-3.59283566e-01 2.71922909e-02 4.81389582e-01 -8.20686996e-01
8.87816668e-01 -2.37650895e+00 7.78443575e-01 4.78083134e-01
3.20093989e-01 1.08924791e-01 3.21961045e-02 4.40382063e-02
-2.25458562e-01 1.00873046e-01 -4.86667752e-01 -5.87027252e-01
-1.93178043e-01 7.70192981e-01 -3.95576805e-01 5.95933914e-01
4.56693992e-02 7.56407440e-01 -2.65380800e-01 -3.35391849e-01
4.80637252e-01 9.19292510e-01 -3.35987747e-01 2.27031767e-01
-2.22010523e-01 5.54057717e-01 -2.90726811e-01 -1.19225763e-01
4.94681150e-01 -1.36027977e-01 5.46602048e-02 1.66243762e-01
1.98926523e-01 1.29854426e-01 -9.60195005e-01 1.26781559e+00
-6.64360583e-01 5.45476079e-01 -8.33993629e-02 -1.12096286e+00
1.00849140e+00 3.89630705e-01 5.97319305e-02 -6.76280975e-01
1.02569580e-01 -1.28704950e-01 1.73207477e-01 -2.39283770e-01
1.69737503e-01 -2.90809810e-01 3.01071852e-01 3.32692474e-01
2.23833323e-01 -1.83365587e-02 1.06868923e-01 1.42840324e-02
8.08792412e-01 1.68378260e-02 6.20699301e-03 -1.35874733e-01
6.68133676e-01 -4.73630369e-01 4.49289680e-01 5.72933197e-01
3.92568976e-01 6.13045752e-01 4.64288920e-01 -5.55416465e-01
-9.18233573e-01 -1.03193402e+00 -1.48104578e-01 8.22101653e-01
6.17115647e-02 2.75127757e-02 -8.21306288e-01 -2.81019926e-01
-3.26679170e-01 9.49062586e-01 -1.14005244e+00 -3.07612747e-01
-6.06010497e-01 -1.09918487e+00 8.94470066e-02 3.71706367e-01
4.90607291e-01 -1.56820035e+00 -8.78469884e-01 6.27683662e-03
-6.46060929e-02 -8.33501697e-01 2.33739540e-01 7.21421540e-01
-1.17183948e+00 -6.84630990e-01 -7.99569011e-01 -6.64655387e-01
9.47916925e-01 -9.57458466e-02 1.09538555e+00 6.85874596e-02
-1.89286366e-01 1.03926063e-01 -2.37734601e-01 -2.32958347e-01
-4.87087548e-01 1.16596244e-01 -4.25400764e-01 2.67706156e-01
9.02670696e-02 -6.62719846e-01 -4.77147400e-01 -2.48493105e-02
-1.19500434e+00 2.09368408e-01 8.62048090e-01 9.70380247e-01
6.47028327e-01 6.58040196e-02 -1.22140780e-01 -1.02590060e+00
2.86133111e-01 -3.70650083e-01 -6.98874176e-01 7.93325901e-02
-4.05717731e-01 3.91118407e-01 4.99700546e-01 -5.67978978e-01
-1.08590698e+00 -2.96371486e-02 -1.35687009e-01 -2.95025438e-01
-2.44655639e-01 4.92772341e-01 -3.08946252e-01 -5.22338413e-02
4.86594707e-01 5.09230733e-01 -1.33162603e-01 -5.86354077e-01
2.63422310e-01 3.29339564e-01 2.04626128e-01 -1.34660766e-01
7.04631090e-01 5.36886811e-01 4.97015789e-02 -9.74241972e-01
-4.33967561e-01 -1.66215986e-01 -7.67273426e-01 -2.37998851e-02
1.02521896e+00 -6.15276098e-01 -1.49649113e-01 4.19895738e-01
-9.52946544e-01 -5.41846633e-01 -4.37944889e-01 5.79158604e-01
-5.87428153e-01 1.70782283e-01 -7.91537941e-01 -8.73391211e-01
-2.27125987e-01 -1.22865319e+00 4.96728390e-01 1.30269423e-01
-2.14235723e-01 -1.36209965e+00 8.22445676e-02 3.76775384e-01
4.29604560e-01 1.68813601e-01 1.29713356e+00 -4.22623366e-01
-6.21501207e-01 -1.89326629e-01 2.31683314e-01 8.14984560e-01
-7.56419376e-02 -3.30818087e-01 -1.19513988e+00 -4.37036484e-01
4.26983953e-01 7.03295227e-03 1.31284499e+00 6.22866571e-01
1.08980048e+00 -2.99856961e-01 -2.29111940e-01 6.26918137e-01
1.51978815e+00 2.76474923e-01 1.19879079e+00 5.33867538e-01
6.33996665e-01 4.89202589e-01 -1.28333002e-01 1.86365172e-01
-8.47134590e-02 5.36185861e-01 8.67835402e-01 -6.30571187e-01
-9.17144772e-03 2.81049371e-01 2.36851841e-01 7.04383671e-01
-4.37953085e-01 -3.97290736e-01 -5.68089485e-01 5.01721680e-01
-1.56837428e+00 -9.28087234e-01 9.13913697e-02 2.32313132e+00
5.53092480e-01 1.24362156e-01 -1.69123933e-01 2.71032244e-01
7.20025301e-01 3.66934568e-01 -3.12624812e-01 -6.02210045e-01
-5.43088876e-02 4.79600161e-01 3.30492675e-01 5.65852046e-01
-9.08274055e-01 6.33545637e-01 6.15982056e+00 7.47706652e-01
-1.33782196e+00 2.60705292e-01 7.19531238e-01 -2.62059093e-01
-2.59527385e-01 -1.09627880e-01 -5.79506099e-01 6.39622927e-01
8.41134906e-01 5.86444855e-01 6.17493033e-01 6.56224251e-01
-9.64135900e-02 -2.22956344e-01 -1.04255152e+00 8.05628657e-01
1.58938989e-01 -1.13554108e+00 2.41059974e-01 4.98613328e-01
5.76680303e-01 1.06370568e-01 3.90395314e-01 1.06776521e-01
1.42110869e-01 -1.49108899e+00 7.98220098e-01 5.47410786e-01
3.77422333e-01 -8.15741897e-01 9.75214422e-01 5.47532380e-01
-6.01608872e-01 -8.11729282e-02 -7.57465303e-01 -1.48054525e-01
-1.61327064e-01 5.41641057e-01 -5.91335833e-01 3.02357644e-01
8.06019843e-01 1.46995634e-01 -4.12552476e-01 8.39419186e-01
-5.70927203e-01 7.50273705e-01 -1.58113003e-01 -2.63242908e-02
2.02193365e-01 -5.23331583e-01 1.84916824e-01 1.12581182e+00
1.73860148e-01 -2.18560435e-02 -1.92849115e-01 1.31972504e+00
6.46549240e-02 -7.01001957e-02 -3.99725169e-01 2.83874124e-01
9.52622071e-02 1.34644675e+00 -8.44397962e-01 -2.68407136e-01
-2.03319192e-01 1.14461887e+00 4.03477013e-01 7.07292497e-01
-9.21624660e-01 -4.18985821e-02 3.65603238e-01 1.14776716e-01
6.32436752e-01 -2.71618664e-01 -2.20170498e-01 -8.73750567e-01
-1.03709780e-01 -4.76624995e-01 1.91088870e-01 -6.50420249e-01
-9.78955448e-01 1.00666034e+00 -5.81327118e-02 -7.09922016e-01
-3.00416678e-01 -5.52524805e-01 -6.22226775e-01 1.44972122e+00
-1.40341794e+00 -1.14911234e+00 -3.38380098e-01 5.68495214e-01
2.55420834e-01 -7.93141350e-02 1.05469823e+00 -6.37175888e-02
-6.28334820e-01 1.89050466e-01 3.23820889e-01 2.06437752e-01
4.67409119e-02 -1.21441758e+00 2.64775097e-01 7.31664181e-01
5.91682494e-01 7.03146636e-01 7.40477860e-01 -3.78687143e-01
-8.78844798e-01 -8.52395356e-01 5.71851552e-01 -2.97656178e-01
1.96285829e-01 -6.02806926e-01 -1.13294673e+00 8.30775499e-01
2.29386568e-01 -6.46270253e-03 6.63665354e-01 9.22205672e-02
-3.98200303e-02 6.83311652e-03 -9.88910675e-01 3.21309566e-01
3.68051022e-01 -4.49947238e-01 -7.13592231e-01 -3.43298540e-03
1.85529262e-01 -1.85767621e-01 -5.25506675e-01 9.90702808e-02
3.08157563e-01 -1.03817391e+00 1.01742387e+00 -1.45386338e-01
2.11171195e-01 -8.64336938e-02 -1.44016016e-02 -1.47280192e+00
-7.90406108e-01 1.89305484e-01 3.96290421e-03 1.07815206e+00
5.01472771e-01 -8.26157808e-01 9.06757772e-01 6.47658944e-01
2.27749541e-01 -7.53836095e-01 -1.15754652e+00 -6.17158592e-01
9.02885124e-02 -1.98923752e-01 1.65431708e-01 6.20011210e-01
-4.85029221e-01 4.96459424e-01 -3.66676629e-01 1.68912768e-01
4.98895556e-01 7.46699348e-02 4.18531477e-01 -1.18454838e+00
-5.32718718e-01 -2.21011832e-01 -6.30909920e-01 -8.39954197e-01
1.70276180e-01 -7.91846216e-01 1.40172243e-01 -1.75742602e+00
4.18568403e-01 -5.91956615e-01 -5.41722596e-01 6.28913403e-01
4.24628891e-02 6.17149055e-01 1.65232569e-01 3.25890958e-01
-1.07924037e-01 6.13966525e-01 9.64313924e-01 -3.03422779e-01
-2.71036774e-01 6.57873303e-02 -2.55293161e-01 8.42811525e-01
7.92063773e-01 -4.30091232e-01 -4.84773725e-01 -5.34730494e-01
-1.90108246e-03 -3.16110224e-01 5.22558451e-01 -1.05823803e+00
-5.35550192e-02 5.24500795e-02 9.10140872e-01 -3.03599745e-01
8.65856647e-01 -8.40589225e-01 4.65692490e-01 5.65523922e-01
-2.56859243e-01 -3.19695145e-01 6.95043057e-02 3.91766906e-01
-2.64504462e-01 -6.29493117e-01 8.87604237e-01 -3.56308043e-01
-6.01870954e-01 -2.03736536e-02 -6.03085339e-01 -5.46777606e-01
1.01390898e+00 -4.28408474e-01 -7.25026522e-03 -3.76771718e-01
-1.07113624e+00 -3.52462083e-01 5.00304520e-01 1.60738498e-01
7.24363744e-01 -9.50355113e-01 -4.53470379e-01 4.07229066e-01
-2.70049989e-01 1.39849603e-01 2.52522200e-01 6.73466563e-01
-4.42142576e-01 3.55253845e-01 -3.47468376e-01 -5.74159384e-01
-1.29491496e+00 6.35358989e-01 5.21573961e-01 -2.98210859e-01
-8.16295981e-01 8.57003868e-01 6.60364866e-01 -1.57765538e-01
1.70816705e-01 -3.88371199e-01 -5.37077069e-01 -8.12324509e-02
3.38830054e-01 1.83838964e-01 1.17459737e-01 -7.52194107e-01
-2.04771116e-01 3.22890639e-01 1.36523411e-01 -1.85091913e-01
1.46068037e+00 -5.76224066e-02 -3.16820443e-01 5.22558868e-01
9.44937348e-01 3.47727584e-03 -1.24248958e+00 -6.49661571e-02
-2.63889968e-01 -2.60398626e-01 3.78182530e-01 -6.48164690e-01
-1.26350093e+00 1.21433544e+00 7.30294287e-01 3.03584099e-01
1.15093696e+00 1.14484601e-01 3.81190032e-01 9.85996146e-03
3.00043344e-01 -9.23858523e-01 -1.50226653e-01 3.49166453e-01
7.58591890e-01 -1.03281534e+00 -1.10659420e-01 -1.98263466e-01
-5.48722684e-01 9.89676654e-01 2.57146925e-01 -2.77346790e-01
3.40534061e-01 2.48644412e-01 9.17855501e-02 -7.64005855e-02
-5.35915613e-01 -1.32086515e-01 3.40308905e-01 5.46108663e-01
3.26822728e-01 -5.28339762e-03 5.84641024e-02 2.97095150e-01
-6.84768558e-02 -3.01187932e-01 2.70275563e-01 5.90426683e-01
-5.34759581e-01 -9.52643812e-01 -6.43772244e-01 1.99087739e-01
-3.94252360e-01 -1.47144094e-01 -2.07289681e-01 7.39758670e-01
3.31886292e-01 8.46791029e-01 1.38124421e-01 -1.19481176e-01
-1.41088203e-01 5.04380502e-02 7.14870632e-01 -5.77154219e-01
-4.02033001e-01 3.56253199e-02 -3.41437697e-01 -2.71872252e-01
-3.60340267e-01 -2.32625261e-01 -1.19728959e+00 -1.22664779e-01
-4.50385571e-01 3.02607447e-01 8.47924411e-01 1.05001342e+00
-3.85880917e-02 7.80679226e-01 3.54997367e-01 -1.08067989e+00
-5.07535517e-01 -1.20610964e+00 -7.54292011e-01 4.59435970e-01
1.97647437e-01 -6.01397216e-01 -5.00237525e-01 9.48183686e-02] | [9.148859024047852, 2.6410422325134277] |
3a419869-f3d0-4a22-9af8-4a3eca63d40a | constant-memory-attention-block | 2306.12599 | null | https://arxiv.org/abs/2306.12599v1 | https://arxiv.org/pdf/2306.12599v1.pdf | Constant Memory Attention Block | Modern foundation model architectures rely on attention mechanisms to effectively capture context. However, these methods require linear or quadratic memory in terms of the number of inputs/datapoints, limiting their applicability in low-compute domains. In this work, we propose Constant Memory Attention Block (CMAB), a novel general-purpose attention block that computes its output in constant memory and performs updates in constant computation. Highlighting CMABs efficacy, we introduce methods for Neural Processes and Temporal Point Processes. Empirically, we show our proposed methods achieve results competitive with state-of-the-art while being significantly more memory efficient. | ['Mohamed Osama Ahmed', 'Yoshua Bengio', 'Hossein Hajimirsadeghi', 'Frederick Tung', 'Leo Feng'] | 2023-06-21 | null | null | null | null | ['point-processes'] | ['methodology'] | [-8.70702490e-02 -3.10752720e-01 -1.80549577e-01 -4.22964208e-02
-5.85817158e-01 -4.48722452e-01 8.58828127e-01 4.56638545e-01
-6.49732113e-01 4.22189295e-01 1.65198520e-01 -3.31785768e-01
9.31868628e-02 -9.31605577e-01 -9.52542305e-01 -4.03964579e-01
2.33898908e-02 4.59039092e-01 4.68457788e-01 1.13917485e-01
5.65578282e-01 6.11755490e-01 -1.53497159e+00 5.14816165e-01
4.87159640e-01 1.02335322e+00 1.42610699e-01 9.76692140e-01
-3.19513440e-01 1.11525905e+00 -4.57996696e-01 -3.37545902e-01
4.39189114e-02 -5.03472500e-02 -8.81875455e-01 -5.15564978e-01
5.90606928e-01 -3.06614459e-01 -5.59703231e-01 6.19525313e-01
4.14913327e-01 3.57699722e-01 4.93742466e-01 -8.27275574e-01
-1.11863852e+00 4.72787052e-01 -4.57769603e-01 8.72389972e-01
6.59395242e-03 4.97947574e-01 1.16961038e+00 -1.34567678e+00
3.15681696e-01 1.39842880e+00 6.72940195e-01 3.36200148e-01
-1.48199368e+00 -5.09081483e-01 7.19324768e-01 2.02000484e-01
-1.24737728e+00 -4.94694740e-01 2.83599228e-01 -3.06057632e-01
1.74188912e+00 1.64941818e-01 6.92650497e-01 8.96584272e-01
4.58101183e-01 9.22884703e-01 7.52643287e-01 -8.14203098e-02
5.54870963e-01 -8.76340747e-01 4.42104965e-01 3.82177621e-01
1.57754093e-01 1.16358986e-02 -9.34275031e-01 -5.65466918e-02
1.15654075e+00 3.20812970e-01 -6.79374561e-02 2.03524917e-01
-1.29291558e+00 5.83203137e-01 7.86782682e-01 1.89590737e-01
-8.21690142e-01 1.17653191e+00 2.41446912e-01 4.54095230e-02
4.91579115e-01 4.36769843e-01 -4.55577493e-01 -4.05814826e-01
-9.34071839e-01 2.35466897e-01 5.94005644e-01 1.09905672e+00
4.47537452e-01 2.26557516e-02 -7.48955488e-01 4.62507457e-01
4.55889590e-02 4.91336972e-01 4.25616235e-01 -8.34410787e-01
3.85541171e-01 5.99261522e-01 1.67079136e-01 -8.58019233e-01
-3.68997127e-01 -2.77454317e-01 -8.41766655e-01 -4.29392532e-02
-6.11818247e-02 1.52816623e-01 -1.15766895e+00 1.45879221e+00
5.91421835e-02 6.52566195e-01 -2.07579076e-01 8.50082695e-01
3.99648190e-01 8.97613406e-01 6.73968375e-01 8.28459635e-02
1.31997025e+00 -1.34112525e+00 -7.02058017e-01 -4.40091133e-01
4.71137352e-02 -6.22119129e-01 1.17946684e+00 1.96424589e-01
-1.71781802e+00 -7.53717363e-01 -7.50570774e-01 -7.62129068e-01
-2.86804885e-01 -3.86388786e-02 1.09791267e+00 1.06152140e-01
-1.34568846e+00 7.04318285e-01 -1.45575738e+00 -3.76582555e-02
6.06202543e-01 5.34801900e-01 2.08620906e-01 1.31875902e-01
-7.52500176e-01 7.12830365e-01 2.51581401e-01 1.74624830e-01
-1.08586419e+00 -9.00096655e-01 -4.62963045e-01 6.47834361e-01
1.91199094e-01 -9.96206343e-01 1.60344875e+00 -7.67087996e-01
-1.49470711e+00 4.95475888e-01 -5.79617500e-01 -7.68084824e-01
2.09769636e-01 -8.17886531e-01 -1.94180742e-01 2.09439084e-01
-2.89987922e-01 8.44629645e-01 9.41529214e-01 -6.56575501e-01
-5.86560726e-01 -2.91550010e-01 1.40543163e-01 7.12406710e-02
-1.23530015e-01 9.34570581e-02 -8.60856354e-01 -7.39656925e-01
1.61120653e-01 -8.77591670e-01 -3.67509902e-01 2.21467271e-01
1.86293170e-01 -3.83669704e-01 6.43863201e-01 -4.24902976e-01
1.38384736e+00 -2.21924758e+00 1.35418907e-01 -1.88819572e-01
3.62928331e-01 1.36602879e-01 -8.93014111e-03 3.17368060e-01
3.11667711e-01 2.45394126e-01 -2.38523096e-01 -6.91575587e-01
-2.51000337e-02 3.98989737e-01 -7.03472912e-01 2.31422201e-01
7.55798995e-01 1.30139494e+00 -1.02035439e+00 -2.88807064e-01
3.55172530e-02 4.96350288e-01 -7.97419667e-01 1.66351065e-01
-4.29773062e-01 7.38789588e-02 -2.45575175e-01 6.94812357e-01
3.34138602e-01 -5.77580094e-01 -5.61526306e-02 2.72635352e-02
-2.69293666e-01 5.74424088e-01 -7.62989640e-01 1.69999123e+00
-7.95128167e-01 8.47201884e-01 -4.46831323e-02 -2.65198708e-01
5.66264510e-01 3.14399362e-01 2.69437786e-02 -5.11919081e-01
-4.72979508e-02 -9.24490765e-03 3.09982616e-03 5.29852621e-02
9.56327677e-01 1.94121718e-01 1.16014689e-01 3.77017796e-01
3.14578749e-02 -2.35759951e-02 2.22147718e-01 -4.70107310e-02
1.26556396e+00 9.99726281e-02 2.32344046e-01 -3.58756423e-01
2.76934296e-01 -1.95091575e-01 3.15527797e-01 8.78175616e-01
-1.95107415e-01 4.07295913e-01 4.89099771e-01 -6.96102202e-01
-1.08216202e+00 -1.03344285e+00 1.96168184e-01 1.59468687e+00
7.88665935e-02 -5.42414665e-01 -5.67619145e-01 -1.26346931e-01
4.04852033e-02 6.53097093e-01 -8.36789489e-01 1.37411023e-03
-9.69085097e-01 -5.39043844e-01 4.90474015e-01 1.39285612e+00
3.81504267e-01 -1.22949290e+00 -1.04360497e+00 5.03725648e-01
3.10835451e-01 -7.37597823e-01 -5.46502590e-01 2.76489317e-01
-1.23460019e+00 -5.59907138e-01 -6.02312744e-01 -4.79887128e-01
5.53774416e-01 3.44818741e-01 1.39606893e+00 1.22239262e-01
-8.75267312e-02 5.27133383e-02 1.13198698e-01 -5.67977130e-01
2.75372207e-01 4.06101018e-01 -2.79918224e-01 -1.82243288e-01
4.03598368e-01 -5.27750313e-01 -9.68352914e-01 1.55631881e-02
-9.78996873e-01 2.04665646e-01 8.40196371e-01 7.92765260e-01
8.89216125e-01 -4.20922458e-01 1.67861462e-01 -7.00752079e-01
7.54499912e-01 -4.74062413e-01 -5.94942451e-01 -5.75304069e-02
-5.23425579e-01 4.54895794e-01 4.53330278e-01 -6.49578631e-01
-9.73193228e-01 -2.20117331e-01 1.01043552e-01 -6.37474775e-01
1.83762297e-01 4.26763982e-01 4.44846481e-01 1.90369844e-01
4.45519656e-01 1.47535682e-01 -5.20612001e-01 -3.14949214e-01
4.92363214e-01 -1.77916437e-02 6.53513312e-01 -8.88635039e-01
1.23991646e-01 7.59702206e-01 5.23921326e-02 -3.81713182e-01
-6.76631033e-01 -3.23307127e-01 -5.10749221e-01 1.62974015e-01
8.38708699e-01 -9.42069590e-01 -9.01411057e-01 3.91205847e-01
-1.50901866e+00 -6.20468616e-01 -2.49944046e-01 1.29865691e-01
-5.60684562e-01 -3.09275031e-01 -1.29753578e+00 -8.99388313e-01
-7.99670398e-01 -1.06003487e+00 1.38062632e+00 9.72125009e-02
-4.62366670e-01 -8.60308290e-01 -4.47216593e-02 -2.33258486e-01
8.68102789e-01 -1.85383692e-01 7.08217144e-01 -2.57320762e-01
-1.19354630e+00 8.72556716e-02 -4.68389094e-01 -1.76778018e-01
-3.81603092e-01 4.74953372e-03 -1.03601229e+00 -1.08678043e-01
-2.95334905e-01 9.68978107e-02 1.03736651e+00 4.09327477e-01
1.64389503e+00 -1.64081156e-01 -2.71430939e-01 6.39170170e-01
1.32808638e+00 5.82552999e-02 6.37643099e-01 2.97660083e-02
6.27277195e-01 8.30814019e-02 3.02664459e-01 4.92213607e-01
2.59139240e-01 4.02525604e-01 3.29349995e-01 -8.54423642e-02
-2.01362565e-01 -1.18342713e-01 2.65571773e-01 9.59678769e-01
-4.81445700e-01 -1.27621278e-01 -1.25772512e+00 8.84422302e-01
-2.10491705e+00 -1.08321393e+00 -2.42006451e-01 1.90041614e+00
8.10018778e-01 5.03897309e-01 -2.48282686e-01 -1.76041886e-01
5.05674779e-01 1.12608902e-01 -5.85874736e-01 -7.46671200e-01
7.20848963e-02 7.85202205e-01 6.71232760e-01 4.27138746e-01
-9.69297409e-01 1.14301419e+00 7.31032991e+00 5.30489326e-01
-1.19674122e+00 4.03854698e-01 6.34253502e-01 -7.53685594e-01
-5.82499951e-02 -1.34206787e-01 -8.52465272e-01 4.53879863e-01
1.42465770e+00 -1.11954540e-01 5.33925653e-01 1.00571382e+00
2.35161539e-02 -3.41353118e-02 -1.44924319e+00 9.88122582e-01
-1.56654328e-01 -1.51289880e+00 1.93834215e-01 -6.39880896e-02
9.31596518e-01 1.56273618e-01 3.15053135e-01 5.47343850e-01
4.31397498e-01 -1.06505835e+00 8.71508658e-01 8.41385543e-01
4.73037153e-01 -8.77771795e-01 5.56397378e-01 -1.02220595e-01
-1.31943643e+00 -2.40750998e-01 -4.28805083e-01 -4.16592836e-01
2.30163887e-01 3.47070545e-01 -3.77811134e-01 1.13691613e-02
8.76186013e-01 4.67340767e-01 -4.44160014e-01 8.80980730e-01
-2.69628018e-01 6.62804425e-01 -3.80718946e-01 -2.00096771e-01
6.47824109e-01 3.46444190e-01 1.10170901e-01 1.41607392e+00
3.98050517e-01 3.63050789e-01 5.05937263e-02 9.62154746e-01
-2.78844655e-01 -1.91099107e-01 -1.63298577e-01 -9.90536660e-02
5.18199921e-01 8.98186803e-01 -7.70156384e-01 -6.74595773e-01
-4.18588758e-01 1.22253633e+00 7.10517645e-01 3.75038534e-01
-9.73404109e-01 -1.28743142e-01 9.65162396e-01 2.10935548e-01
4.64576900e-01 -6.41256690e-01 -7.92536974e-01 -9.24460888e-01
-9.03802961e-02 -4.34064060e-01 2.91315019e-01 -9.45806921e-01
-1.23517478e+00 5.23683548e-01 -2.97362834e-01 -5.90540230e-01
1.01114504e-01 -5.55837870e-01 -6.56941831e-01 1.06860602e+00
-1.34163022e+00 -1.17630482e+00 -5.26615202e-01 4.73697811e-01
8.52473080e-01 3.34715843e-01 7.05353022e-01 2.38048837e-01
-4.46368039e-01 3.26025724e-01 -2.09244654e-01 -3.50768030e-01
4.85973924e-01 -1.29627502e+00 1.28463006e+00 7.66762555e-01
-5.39230928e-03 1.35314381e+00 4.14167494e-01 -6.15844309e-01
-1.63634396e+00 -1.05154049e+00 8.27487588e-01 -5.95082164e-01
9.26094472e-01 -2.54806727e-01 -1.12716877e+00 9.87911999e-01
3.00518334e-01 3.14872384e-01 4.93056685e-01 4.22989279e-01
-3.14803302e-01 4.57362048e-02 -6.55385554e-01 1.12016439e+00
1.21698809e+00 -6.26515567e-01 -6.36244416e-01 2.38344097e-03
1.04606164e+00 -5.22907853e-01 -8.21144998e-01 1.75758466e-01
4.77704257e-01 -4.36299026e-01 9.55264509e-01 -6.73937201e-01
6.69611752e-01 -2.97640413e-01 -1.01704942e-02 -9.05879855e-01
-8.02977026e-01 -5.43083608e-01 -1.09819925e+00 8.06533933e-01
2.30146736e-01 -4.76755649e-01 5.30651987e-01 8.00550342e-01
-2.94776589e-01 -1.03350401e+00 -7.23016918e-01 -4.75830972e-01
2.81928387e-02 -6.44049048e-01 7.41604507e-01 6.77347839e-01
-2.73932666e-01 4.22350496e-01 -4.72941715e-03 2.57757634e-01
1.22658171e-01 1.97636709e-01 4.75592881e-01 -1.03215146e+00
-5.34442604e-01 -7.69346118e-01 -1.97831839e-01 -1.39445627e+00
-1.44172788e-01 -5.61237693e-01 -2.93879509e-02 -1.42325234e+00
3.98290269e-02 -3.32301497e-01 -5.28435647e-01 5.29471338e-01
-4.92189705e-01 1.43952072e-01 2.82143861e-01 5.34937084e-01
-7.42429495e-01 6.28576100e-01 9.21763241e-01 -9.97281224e-02
-1.91389292e-01 -3.06159288e-01 -4.93960261e-01 6.28492475e-01
8.37450147e-01 -2.98690677e-01 -2.42384449e-01 -9.60255384e-01
3.14601302e-01 -2.75508225e-01 4.30299699e-01 -1.23387396e+00
6.84792280e-01 -2.43591845e-01 5.89148223e-01 -9.26141262e-01
5.89577436e-01 -6.14321470e-01 -2.68690567e-02 3.97397250e-01
-6.03010952e-01 8.16798925e-01 6.21630311e-01 7.51881242e-01
-7.95063600e-02 2.36654468e-02 5.66519260e-01 -2.58929729e-01
-7.61739075e-01 5.37195444e-01 -4.60240871e-01 -2.06928968e-01
1.02441335e+00 1.81936160e-01 -4.55946922e-01 -5.82212806e-02
-6.34973586e-01 4.63920943e-02 4.33776736e-01 3.61818403e-01
5.07699072e-01 -1.21083033e+00 -3.92282873e-01 3.13912742e-02
-8.70361328e-02 2.77481467e-01 2.60139167e-01 6.89379334e-01
-8.85237813e-01 7.28497803e-01 -9.69827175e-02 -5.07133603e-01
-8.91031981e-01 1.05096352e+00 -2.92501897e-02 -4.03467357e-01
-4.98434722e-01 8.07569385e-01 3.14763039e-01 1.90307006e-01
8.99610668e-02 -9.30740416e-01 3.80847394e-01 -1.95259675e-01
8.62982750e-01 4.55913395e-01 8.60939994e-02 4.70571928e-02
-1.43535450e-01 2.21803442e-01 -1.71097919e-01 -2.79565930e-01
1.16152036e+00 1.29440814e-01 -2.85326660e-01 6.03671968e-01
6.45720065e-01 -3.37943405e-01 -1.57427394e+00 -2.45149955e-01
2.95212120e-01 -2.99913585e-01 3.24089378e-01 -5.51463187e-01
-7.94509768e-01 1.14840662e+00 3.50700945e-01 5.41334506e-03
1.24044943e+00 -1.59955159e-01 8.36932838e-01 4.25662994e-01
3.43106508e-01 -1.11042535e+00 1.06406629e-01 8.76326501e-01
1.03754187e+00 -8.27736974e-01 -3.23023200e-02 -2.48980999e-01
-2.95076221e-01 8.84457290e-01 8.25071573e-01 -4.78698850e-01
5.36157131e-01 5.14272451e-01 -4.60958093e-01 -1.88082665e-01
-1.55500638e+00 -2.09046707e-01 2.66961604e-01 2.33362660e-01
6.62090480e-01 -5.19342124e-02 -2.54423261e-01 5.09221494e-01
8.24466199e-02 4.58926931e-02 8.24465454e-02 1.24943376e+00
-3.67339224e-01 -6.29206479e-01 -3.13146114e-01 3.30479294e-01
-7.34570026e-01 -7.08207309e-01 -1.12460040e-01 5.43470144e-01
-1.81294695e-01 5.65630674e-01 7.01337218e-01 2.53781646e-01
2.05620140e-01 1.13740325e-01 4.00311738e-01 -5.42742312e-01
-1.16082489e+00 -6.72182292e-02 -1.86544031e-01 -7.62144208e-01
-1.58778317e-02 -4.26902026e-01 -1.30052924e+00 -7.12161720e-01
1.17009833e-01 -5.78991711e-01 5.96031487e-01 5.34947336e-01
9.32098687e-01 9.38201904e-01 -2.73325056e-01 -1.06430757e+00
-3.84490192e-01 -1.04378021e+00 1.51790529e-01 3.05642009e-01
1.34188637e-01 -5.65720856e-01 1.73067898e-01 2.74929821e-01] | [10.748778343200684, 6.762882709503174] |
e1eeffa8-afa6-400d-a493-073452acc77c | personalized-keyphrase-detection-using | 2104.13970 | null | https://arxiv.org/abs/2104.13970v2 | https://arxiv.org/pdf/2104.13970v2.pdf | Personalized Keyphrase Detection using Speaker and Environment Information | In this paper, we introduce a streaming keyphrase detection system that can be easily customized to accurately detect any phrase composed of words from a large vocabulary. The system is implemented with an end-to-end trained automatic speech recognition (ASR) model and a text-independent speaker verification model. To address the challenge of detecting these keyphrases under various noisy conditions, a speaker separation model is added to the feature frontend of the speaker verification model, and an adaptive noise cancellation (ANC) algorithm is included to exploit cross-microphone noise coherence. Our experiments show that the text-independent speaker verification model largely reduces the false triggering rate of the keyphrase detection, while the speaker separation model and adaptive noise cancellation largely reduce false rejections. | ['Ian McGraw', 'Arun Narayanan', 'Huang', 'Yiteng', 'Ding Zhao', 'Yanzhang He', 'Qiao Liang', 'Quan Wang', 'Rajeev Rikhye'] | 2021-04-28 | null | null | null | null | ['text-independent-speaker-verification', 'speaker-separation'] | ['speech', 'speech'] | [ 3.58377963e-01 -3.58414024e-01 2.31966406e-01 -2.60501355e-01
-1.39277232e+00 -7.99229205e-01 6.80182397e-01 3.02545011e-01
-4.91782367e-01 1.05140634e-01 5.17378449e-01 -4.07237202e-01
2.70519704e-01 -2.32339531e-01 -4.76129442e-01 -6.26297712e-01
4.90835123e-02 -1.19047128e-01 2.99091637e-01 -1.97367057e-01
1.73349708e-01 3.22468966e-01 -1.55168998e+00 2.02870369e-01
4.18790221e-01 9.49698150e-01 2.65942395e-01 1.45814049e+00
-1.41561076e-01 7.67468393e-01 -9.21078026e-01 4.80443947e-02
1.74177751e-01 -8.22854713e-02 -3.16785038e-01 -1.36329187e-02
4.89166051e-01 -4.06696409e-01 -5.43606102e-01 1.07535064e+00
7.57615328e-01 1.58437893e-01 1.35125458e-01 -1.03302360e+00
1.49149159e-02 8.21607113e-01 -1.46952957e-01 4.92351413e-01
6.83628380e-01 -8.46110061e-02 1.06867301e+00 -1.46587563e+00
-7.02377269e-03 1.30464208e+00 3.45519185e-01 3.97018731e-01
-1.21652734e+00 -1.08042276e+00 2.03828231e-01 -7.50381500e-02
-1.78174794e+00 -1.04418921e+00 8.37838650e-01 -7.37252459e-03
1.02463770e+00 6.69955194e-01 3.93755674e-01 1.10626924e+00
3.86782885e-02 8.92327785e-01 5.81520677e-01 -5.56926191e-01
3.92586857e-01 3.24398994e-01 3.90558869e-01 3.38830531e-01
-5.66231571e-02 1.52959570e-01 -1.15065336e+00 -6.71575010e-01
2.41372973e-01 -3.16660821e-01 -4.93476629e-01 1.58378825e-01
-1.09367931e+00 6.19560063e-01 -2.95618862e-01 2.75345623e-01
-5.17365754e-01 -3.59741747e-02 4.08170342e-01 5.95674515e-01
3.90422404e-01 2.53248096e-01 -4.99120355e-01 -1.18831225e-01
-1.29859221e+00 1.46784157e-01 1.14618516e+00 1.01160300e+00
5.51354825e-01 3.18315953e-01 -2.77011752e-01 8.74772608e-01
5.07592380e-01 9.06867445e-01 6.21848941e-01 -3.11736226e-01
3.60177100e-01 1.70785204e-01 5.62259741e-02 -8.19426417e-01
-1.73383296e-01 -6.22746408e-01 -4.94956434e-01 -3.33096355e-01
-7.97426626e-02 -1.52946725e-01 -8.61033559e-01 1.29586327e+00
4.26209420e-01 3.91097069e-01 1.96432784e-01 5.88665545e-01
7.52235532e-01 8.78086686e-01 -2.07741082e-01 -3.85138035e-01
1.60502565e+00 -6.32702827e-01 -1.06335807e+00 -4.51981276e-01
2.61854082e-01 -1.22073126e+00 7.61869311e-01 4.20920491e-01
-9.33882415e-01 -4.34961379e-01 -1.22074687e+00 1.26514494e-01
-2.39920959e-01 -3.91068384e-02 -5.73203675e-02 9.10196543e-01
-1.09506929e+00 -2.95566142e-01 -7.00163245e-01 -1.42773807e-01
-3.41948330e-01 3.89822781e-01 -1.70266300e-01 6.54619858e-02
-1.31516290e+00 4.47970629e-01 5.71676083e-02 3.21797770e-03
-9.27124023e-01 -5.38887739e-01 -8.81445527e-01 1.40134498e-01
4.91852790e-01 -2.79896975e-01 1.59246886e+00 -8.40271592e-01
-1.64040923e+00 5.43541551e-01 -6.44069433e-01 -4.79043692e-01
7.20070526e-02 -2.59951115e-01 -9.84003544e-01 4.09707874e-01
-7.35453442e-02 -7.28502125e-02 1.77036643e+00 -9.64872479e-01
-8.62659276e-01 -1.16338804e-01 -5.63630641e-01 2.33660743e-01
-5.27709603e-01 6.22830093e-01 -5.14758229e-01 -1.13846123e+00
2.91256756e-01 -5.99118292e-01 -1.66930407e-02 -6.00902259e-01
-4.66450214e-01 1.46929175e-01 9.90353405e-01 -8.91879857e-01
1.53415167e+00 -2.63714576e+00 -3.22185934e-01 5.86429715e-01
-7.64804780e-02 2.81619579e-01 -3.00660700e-01 6.19436622e-01
-1.07877508e-01 -2.87514269e-01 -1.90472300e-03 -5.64870417e-01
-1.80997029e-01 -1.02079324e-01 -8.35047185e-01 4.58546549e-01
1.97550803e-01 4.49233919e-01 -6.40440822e-01 -1.21328466e-01
1.60975620e-01 4.63539749e-01 -2.66521692e-01 3.39537978e-01
3.45010450e-03 5.99078871e-02 -1.55570716e-01 6.06385291e-01
6.69025123e-01 2.31465057e-01 5.23551833e-04 6.53857961e-02
-2.04497457e-01 9.24343944e-01 -1.73469353e+00 1.03502429e+00
-2.36988679e-01 4.98247743e-01 9.51046824e-01 -5.71300745e-01
7.66564488e-01 8.40401709e-01 4.36330913e-03 -3.52355450e-01
1.39002189e-01 2.81795532e-01 -2.18907759e-01 -1.22625634e-01
8.66397798e-01 9.18670520e-02 -8.28519389e-02 3.53034079e-01
1.44253016e-01 -1.74936056e-01 -2.47480333e-01 4.40002143e-01
1.27369976e+00 -9.84811187e-01 2.70147681e-01 -2.78508723e-01
1.00851536e+00 -3.56720507e-01 2.59576082e-01 9.99182105e-01
-2.30971053e-01 5.81518769e-01 -2.02376872e-01 1.00386389e-01
-4.91136909e-01 -9.51786935e-01 6.62207976e-02 1.41075218e+00
-1.03815801e-01 -7.38349378e-01 -5.50376952e-01 -3.65799069e-01
-3.93205881e-02 6.66129410e-01 5.99861160e-05 -1.96233884e-01
-4.64171827e-01 -3.08203727e-01 8.49763632e-01 2.91699111e-01
7.86483735e-02 -6.01854861e-01 -3.45218033e-02 4.16399807e-01
-2.97847271e-01 -1.35458076e+00 -1.01249325e+00 5.12868941e-01
-2.73351610e-01 -6.98305607e-01 -5.28732717e-01 -9.88322616e-01
3.82191628e-01 7.62867332e-01 6.43727660e-01 -2.13106140e-01
-1.52630778e-02 8.81548882e-01 -4.78786498e-01 -5.37892163e-01
-9.17291939e-01 -1.55665874e-01 5.31169593e-01 4.66086000e-01
5.52570343e-01 -2.86976188e-01 -2.32167348e-01 3.12501460e-01
-8.99990261e-01 -4.17968661e-01 2.66755342e-01 7.64289558e-01
1.15385070e-01 3.78209323e-01 4.30679679e-01 -7.52572492e-02
8.55041027e-01 7.34651880e-03 -7.84965396e-01 1.10645786e-01
-3.55781347e-01 -7.19111785e-02 5.48214614e-01 -8.10477197e-01
-7.41810143e-01 1.85464501e-01 -2.96747893e-01 -2.61941493e-01
-3.93085107e-02 4.83059287e-01 -3.08111906e-01 -1.81567684e-01
5.16561568e-01 9.12208438e-01 -2.14538276e-01 -4.69711691e-01
2.68661380e-01 1.39954484e+00 6.28051996e-01 -5.19759022e-02
1.19837964e+00 2.74467975e-01 -6.65377676e-01 -1.64763165e+00
-4.34397250e-01 -1.22540832e+00 -2.44485468e-01 -4.86375391e-02
2.83050597e-01 -1.58412039e+00 -3.38085681e-01 7.66443968e-01
-1.08807278e+00 1.31786242e-01 -8.40267837e-02 6.83341146e-01
-1.45606855e-02 4.91981357e-01 -6.00332618e-01 -1.31101060e+00
-9.08908486e-01 -9.38176095e-01 1.32058835e+00 3.99336740e-02
-4.32505608e-01 -4.11117315e-01 -7.34741986e-02 2.73950487e-01
5.75224996e-01 -7.79665470e-01 3.70666444e-01 -1.28252053e+00
-3.43778580e-01 -6.19089484e-01 1.76957697e-01 5.95735133e-01
2.57455319e-01 -1.54012784e-01 -1.40570748e+00 -4.61015373e-01
4.19819653e-01 5.94928004e-02 7.24233985e-01 1.51721641e-01
5.37023902e-01 -4.37304288e-01 -1.16311081e-01 1.57051757e-01
8.54968727e-01 1.30804017e-01 1.73851967e-01 -1.19511239e-01
4.70229030e-01 3.32526177e-01 2.85092950e-01 4.64484185e-01
2.35136762e-01 5.68482041e-01 -2.12658808e-01 2.05091149e-01
9.47527587e-02 -2.23677948e-01 9.55663681e-01 1.24670434e+00
9.56422448e-01 -2.54405051e-01 -9.30305243e-01 6.00048542e-01
-1.37388933e+00 -9.07847524e-01 -1.07597388e-01 2.26276088e+00
9.57317114e-01 3.50763381e-01 5.92786185e-02 5.58710814e-01
8.00408721e-01 3.17153871e-01 -1.18591234e-01 -2.38270685e-01
-1.42498881e-01 3.50199252e-01 7.04652131e-01 8.59858632e-01
-1.17303383e+00 9.33134794e-01 6.86049557e+00 8.59701812e-01
-1.17636752e+00 -8.51402432e-02 -9.22397897e-02 -1.13709681e-01
-5.55363484e-03 -2.59193808e-01 -1.10158944e+00 1.86561704e-01
1.18993843e+00 -3.07396293e-01 5.16484082e-01 6.95398927e-01
3.81206095e-01 -1.02576286e-01 -1.01647055e+00 1.07950151e+00
4.04947490e-01 -7.15001285e-01 2.83525772e-02 -1.51371837e-01
-4.54830825e-02 1.50534824e-01 2.07599029e-02 2.28821054e-01
2.31042698e-01 -5.13681233e-01 8.88162613e-01 3.24119106e-02
4.92255241e-01 -7.60487139e-01 3.53224725e-01 5.52214742e-01
-1.47443914e+00 -2.36957476e-01 7.76524395e-02 1.53201092e-02
5.59880659e-02 7.64417052e-01 -1.31816006e+00 1.47779182e-01
4.72267658e-01 -1.69478450e-02 -3.12071234e-01 8.27863038e-01
-3.13375473e-01 1.01231742e+00 -6.70993805e-01 -1.06810234e-01
-1.05847649e-01 3.86852235e-01 1.15800464e+00 1.58955407e+00
9.98096094e-02 1.27157956e-01 1.63050398e-01 2.51764625e-01
-1.42083783e-02 2.27039114e-01 -2.04428434e-01 -2.12748721e-01
7.79921472e-01 1.29446900e+00 -6.28501773e-01 -4.66333002e-01
-2.62449414e-01 1.18597031e+00 -3.52435499e-01 5.43650985e-01
-4.36160535e-01 -5.58538258e-01 9.26398277e-01 -1.24391668e-01
6.35741472e-01 -3.88943344e-01 8.28155428e-02 -1.39491868e+00
6.27602115e-02 -1.28011048e+00 1.50071144e-01 -5.51872313e-01
-1.14947820e+00 2.45957360e-01 -4.56445992e-01 -9.88075137e-01
-1.72110066e-01 -2.76089281e-01 -6.18006527e-01 1.08702815e+00
-1.44819295e+00 -6.45152628e-01 1.24934055e-01 8.05527389e-01
6.66125894e-01 -2.33733714e-01 1.07132912e+00 2.31641799e-01
-4.62957412e-01 8.62877250e-01 8.32493827e-02 4.22304124e-01
8.34294677e-01 -1.03335345e+00 8.18151653e-01 1.35685408e+00
2.16849342e-01 1.00479126e+00 1.03672588e+00 -6.47269249e-01
-1.90964866e+00 -1.09814370e+00 1.12894738e+00 -3.40857029e-01
9.10600841e-01 -1.11862099e+00 -9.93103743e-01 4.64749545e-01
1.12275533e-01 -1.22812442e-01 9.41560984e-01 -4.84888703e-02
-6.94448233e-01 -1.73213005e-01 -7.76807129e-01 5.22185326e-01
2.70363867e-01 -1.20973015e+00 -9.11052704e-01 7.57365674e-02
1.03927553e+00 -2.70622790e-01 -3.78199011e-01 1.38808072e-01
5.55915952e-01 -2.83944577e-01 1.04706264e+00 -1.17360443e-01
-4.89020169e-01 -5.30630291e-01 -4.26180184e-01 -1.14780414e+00
-1.93110541e-01 -1.11210465e+00 -4.70001817e-01 1.49078429e+00
4.49139208e-01 -6.11521184e-01 3.70836407e-01 4.84077394e-01
3.45958680e-01 3.66203904e-01 -1.25494373e+00 -7.74319053e-01
-7.04814672e-01 -7.78326571e-01 2.81453907e-01 6.99872077e-01
2.93135375e-01 7.56812871e-01 -3.90239090e-01 9.40693736e-01
4.81432527e-01 -3.44956040e-01 7.77972758e-01 -9.67912853e-01
-4.59140271e-01 -1.95129141e-01 -1.67345598e-01 -1.29433000e+00
-1.46560326e-01 -4.61037606e-01 4.39501762e-01 -8.71138632e-01
-3.20815355e-01 2.30577633e-01 -3.79381895e-01 1.03941716e-01
-4.19221729e-01 -2.08247393e-01 6.04655361e-03 4.82302345e-02
-3.89810890e-01 4.22816753e-01 1.95139706e-01 -3.45964700e-01
-5.45521438e-01 4.01652485e-01 -5.99928260e-01 6.19123280e-01
5.35116494e-01 -6.83164418e-01 -2.08366647e-01 2.84525193e-02
1.49451628e-01 1.10132799e-01 1.41606525e-01 -8.51491928e-01
7.85248280e-01 3.99777234e-01 9.41802338e-02 -7.33890295e-01
4.63113695e-01 -9.12784994e-01 -3.56023341e-01 3.54600072e-01
-3.77078056e-01 -1.08672999e-01 4.42345202e-01 7.62980878e-01
-2.29697302e-01 6.93448037e-02 6.11511707e-01 2.78962821e-01
-2.72062421e-01 -1.65673435e-01 -1.23698378e+00 -2.27516964e-01
5.86686730e-01 7.24126175e-02 7.96905980e-02 -6.76021278e-01
-3.05569619e-01 2.45811358e-01 -8.72108489e-02 6.57445967e-01
8.05681765e-01 -8.28473687e-01 -8.15345526e-01 8.72637570e-01
3.36405843e-01 -2.51129508e-01 -1.12451039e-01 5.44342339e-01
4.14125621e-02 4.22964483e-01 7.78368771e-01 -4.49249774e-01
-1.78030479e+00 4.69706565e-01 2.54064143e-01 -9.05181374e-03
-3.25028896e-01 1.11573744e+00 -1.53656572e-01 -3.62369388e-01
8.78412485e-01 -4.71296728e-01 3.44561413e-02 2.16896206e-01
1.28238583e+00 1.25019684e-01 5.42846262e-01 -8.60453248e-01
-6.52912855e-01 1.13572113e-01 -4.01085675e-01 -6.84105396e-01
8.60120416e-01 -3.42795968e-01 6.51249886e-02 5.66730797e-01
1.05284429e+00 5.54815948e-01 -4.84060317e-01 -7.44217098e-01
5.02685718e-02 -8.18999782e-02 5.26163340e-01 -6.56396806e-01
-2.50508189e-01 5.78698754e-01 4.92822230e-01 4.47768927e-01
1.21097958e+00 -2.92498499e-01 9.57233429e-01 5.80857515e-01
2.06223786e-01 -1.20143414e+00 -2.05228686e-01 9.92718816e-01
7.82684505e-01 -1.10030246e+00 -1.97045594e-01 -2.87759840e-01
-2.45748118e-01 1.00944304e+00 -2.44740513e-04 1.57499075e-01
9.31373715e-01 7.94980764e-01 2.64026314e-01 6.49601072e-02
-9.25013959e-01 -1.78970709e-01 3.62385422e-01 2.28494018e-01
-4.57744673e-03 -8.91251341e-02 1.26507401e-01 8.22565854e-01
-2.64185786e-01 -4.29955900e-01 3.80186319e-01 9.95518327e-01
-7.71905124e-01 -9.25877213e-01 -9.95753109e-01 7.79382363e-02
-6.96218967e-01 -6.27990723e-01 -7.25282550e-01 -5.86266555e-02
-6.63106680e-01 1.70038199e+00 -3.97204459e-02 -7.62320817e-01
4.53133017e-01 4.93204862e-01 -2.06911251e-01 -6.86149061e-01
-9.80549276e-01 6.84037924e-01 3.99456397e-02 -4.30968344e-01
-2.77211033e-02 -5.41921318e-01 -1.06413925e+00 -6.54951632e-02
-8.80415618e-01 3.39215815e-01 9.18423176e-01 9.33701038e-01
3.78733426e-01 4.67788696e-01 1.01748610e+00 -4.64447707e-01
-8.60769928e-01 -1.06877661e+00 -6.88155711e-01 -1.44086555e-01
9.97921824e-01 9.55602750e-02 -7.59556353e-01 9.82889812e-03] | [14.513395309448242, 6.225064754486084] |
938a94ad-9c1f-4b50-8276-42d97026a9a7 | leveraging-sparsity-for-efficient-submodular | 1703.02690 | null | http://arxiv.org/abs/1703.02690v1 | http://arxiv.org/pdf/1703.02690v1.pdf | Leveraging Sparsity for Efficient Submodular Data Summarization | The facility location problem is widely used for summarizing large datasets
and has additional applications in sensor placement, image retrieval, and
clustering. One difficulty of this problem is that submodular optimization
algorithms require the calculation of pairwise benefits for all items in the
dataset. This is infeasible for large problems, so recent work proposed to only
calculate nearest neighbor benefits. One limitation is that several strong
assumptions were invoked to obtain provable approximation guarantees. In this
paper we establish that these extra assumptions are not necessary---solving the
sparsified problem will be almost optimal under the standard assumptions of the
problem. We then analyze a different method of sparsification that is a better
model for methods such as Locality Sensitive Hashing to accelerate the nearest
neighbor computations and extend the use of the problem to a broader family of
similarities. We validate our approach by demonstrating that it rapidly
generates interpretable summaries. | ['Alexandros G. Dimakis', 'Erik M. Lindgren', 'Shanshan Wu'] | 2017-03-08 | leveraging-sparsity-for-efficient-submodular-1 | http://papers.nips.cc/paper/6382-leveraging-sparsity-for-efficient-submodular-data-summarization | http://papers.nips.cc/paper/6382-leveraging-sparsity-for-efficient-submodular-data-summarization.pdf | neurips-2016-12 | ['data-summarization'] | ['miscellaneous'] | [ 3.52685928e-01 2.05433726e-01 -5.70104182e-01 -4.02637422e-01
-8.53258371e-01 -8.10829103e-01 2.35707611e-02 8.82352114e-01
-2.33265877e-01 9.42367792e-01 5.31275570e-01 -2.66798526e-01
-7.39622951e-01 -9.47105110e-01 -9.33779776e-01 -6.80330575e-01
-4.72191334e-01 6.05942667e-01 1.77277625e-01 -1.64450899e-01
6.03759766e-01 6.63253129e-01 -1.49655139e+00 7.86332488e-02
5.89815080e-01 8.61615479e-01 2.48481616e-01 4.92928624e-01
2.56862026e-02 2.69513994e-01 -5.10380328e-01 -4.90616597e-02
6.14407957e-01 -1.59554228e-01 -9.13570046e-01 1.78546593e-01
2.79075086e-01 -5.74242055e-01 -1.10257499e-01 8.92664731e-01
5.83204806e-01 2.27501288e-01 3.71839046e-01 -1.62464917e+00
-6.03704691e-01 9.70163584e-01 -7.25901842e-01 -7.74589255e-02
6.19350612e-01 -6.56533420e-01 1.18621171e+00 -4.47688371e-01
7.13160872e-01 9.65998650e-01 7.68644452e-01 1.81100637e-01
-1.10021007e+00 -2.30218485e-01 5.47965728e-02 6.84792995e-02
-1.56820202e+00 -3.35356176e-01 5.92372596e-01 2.44586214e-01
8.22104752e-01 7.70518899e-01 6.51032090e-01 2.44544759e-01
-5.22006415e-02 6.60430312e-01 8.12187970e-01 -1.94642305e-01
6.99743986e-01 2.33996347e-01 2.89270073e-01 4.25458908e-01
1.09372866e+00 -2.75915205e-01 -5.04998267e-01 -7.00683296e-01
4.77382332e-01 4.09606010e-01 -3.42360735e-01 -8.13159943e-01
-1.30499697e+00 1.05153894e+00 4.42039222e-01 3.17525893e-01
-2.44842753e-01 5.82555056e-01 1.40169129e-01 2.71333158e-01
1.86863899e-01 4.60740358e-01 -2.13455856e-01 2.59626638e-02
-1.03559566e+00 6.64060175e-01 1.19074368e+00 1.20058262e+00
8.07015836e-01 -5.40656865e-01 3.44337523e-02 4.06254321e-01
-3.60949896e-02 4.67952430e-01 -6.21073954e-02 -1.24898303e+00
5.59228241e-01 2.53460616e-01 3.41508508e-01 -1.45517516e+00
-4.44284111e-01 -1.72864467e-01 -5.67534983e-01 -2.25261912e-01
2.86185771e-01 1.12523645e-01 -4.62231159e-01 1.51760578e+00
5.10084331e-01 -1.61412269e-01 -9.17238817e-02 7.69880891e-01
4.12079483e-01 8.56887877e-01 -4.32411283e-01 -5.49662411e-01
1.02853990e+00 -7.21210301e-01 -7.44602859e-01 1.59606695e-01
9.18580353e-01 -4.64602470e-01 6.42601252e-01 3.01983267e-01
-1.34966564e+00 2.30905771e-01 -1.15521169e+00 -2.23241374e-01
-3.95061791e-01 -4.92532492e-01 1.00376809e+00 7.46121824e-01
-1.39265871e+00 4.23111856e-01 -4.88222241e-01 -7.57388890e-01
2.19885364e-01 6.58871353e-01 -2.83558667e-01 -2.98580498e-01
-6.11366391e-01 7.33947933e-01 3.97320598e-01 -3.78116667e-01
-5.22342503e-01 -5.86227238e-01 -9.20296013e-01 3.36112499e-01
4.49612647e-01 -6.98137820e-01 9.85387564e-01 -2.55090714e-01
-5.41218638e-01 5.78683555e-01 -4.81735855e-01 -5.53588450e-01
1.45462096e-01 1.77131504e-01 1.87469244e-01 2.59413570e-01
1.69158936e-01 6.94557905e-01 3.12467307e-01 -1.37603855e+00
-6.55392408e-01 -6.24106407e-01 4.31910306e-01 4.48074609e-01
-7.47703850e-01 -2.23249793e-01 -3.45721811e-01 -4.75504518e-01
3.88675064e-01 -9.82363641e-01 -7.35457003e-01 2.66810775e-01
-4.12355989e-01 -1.38250589e-01 6.05068266e-01 -2.35825375e-01
1.23635936e+00 -2.01928830e+00 6.45290986e-02 7.92758942e-01
1.56342849e-01 -4.76766258e-01 -1.62456423e-01 9.63404477e-01
3.72914046e-01 1.97975889e-01 -3.78417432e-01 -4.19010192e-01
2.23129585e-01 5.92721343e-01 -4.64192927e-01 7.48895466e-01
-5.55922389e-01 7.13163614e-01 -8.73768389e-01 -5.30182660e-01
-7.26513639e-02 1.63529173e-01 -8.01228404e-01 -4.70001966e-01
-1.24273092e-01 -2.88888216e-01 -4.11082834e-01 7.47433186e-01
9.34434295e-01 -4.05759692e-01 3.13681096e-01 1.12091891e-01
-1.48146246e-02 9.36212018e-02 -1.69379020e+00 1.86561739e+00
-1.14493862e-01 1.86129436e-01 2.33926058e-01 -1.09409297e+00
5.52421987e-01 2.06262097e-02 9.32753980e-01 -2.08349884e-01
-7.86311775e-02 3.39387625e-01 -5.97512841e-01 -2.65765160e-01
9.34300780e-01 -6.64224243e-03 -4.34409440e-01 8.16801667e-01
-6.05879724e-01 -3.24568629e-01 2.35872924e-01 4.37587768e-01
1.22129500e+00 -6.37387693e-01 4.64504391e-01 -6.18148088e-01
-5.58081269e-02 4.44183439e-01 5.26307285e-01 1.07351720e+00
1.95881501e-01 6.44811451e-01 3.05644304e-01 -4.35929656e-01
-1.03936207e+00 -9.74913299e-01 -2.62080163e-01 9.02776182e-01
6.13650382e-01 -5.24573207e-01 -5.90519130e-01 -4.99843091e-01
6.01680338e-01 4.48980480e-01 -5.02194405e-01 1.71128139e-01
-1.59054533e-01 -8.64210427e-01 2.19261080e-01 7.07315803e-01
-2.22744364e-02 -3.52165490e-01 -6.38438880e-01 9.99113321e-02
-3.02483916e-01 -8.07084084e-01 -4.80092227e-01 2.58324981e-01
-1.07203579e+00 -8.52976561e-01 -5.64123690e-01 -8.39939713e-01
1.03341925e+00 7.48092294e-01 7.51122117e-01 2.16984704e-01
-2.34015316e-01 6.07746303e-01 -2.33124137e-01 -3.65375757e-01
7.04274550e-02 7.57917091e-02 1.54372394e-01 -4.49322045e-01
2.54169345e-01 -5.38748801e-01 -4.93888557e-01 1.85060844e-01
-1.26693296e+00 -4.58939046e-01 4.61608618e-01 3.45778704e-01
9.21259880e-01 3.80792707e-01 5.00356972e-01 -8.39798510e-01
6.57288373e-01 -7.31486261e-01 -6.21013820e-01 2.82515168e-01
-5.67257583e-01 3.01428050e-01 3.67181957e-01 -8.54501203e-02
-4.15021211e-01 3.53188187e-01 1.40188396e-01 3.36355269e-02
2.43791372e-01 5.52504599e-01 -1.96928605e-01 -3.26729268e-01
3.51447761e-01 9.17690247e-02 -1.06437080e-01 -2.35311434e-01
3.15706730e-01 6.35375798e-01 2.77921706e-01 -6.55979276e-01
9.13144529e-01 1.13903344e+00 5.32825470e-01 -8.13457251e-01
-6.79256439e-01 -7.45233178e-01 -8.90965983e-02 4.40171659e-01
4.33089584e-01 -8.01582813e-01 -8.62270176e-01 -1.80573761e-01
-1.12843013e+00 7.11333230e-02 -6.38468146e-01 2.58059621e-01
-6.42250061e-01 5.86537540e-01 -2.46721968e-01 -8.08215976e-01
-4.02011760e-02 -9.74013925e-01 1.06364524e+00 -3.37881781e-02
-2.05978170e-01 -1.07517481e+00 2.10400537e-01 2.02603817e-01
4.44617212e-01 3.18531156e-01 7.12464809e-01 -7.31050074e-01
-9.37621713e-01 -1.41226903e-01 -1.64095506e-01 -2.25065768e-01
1.63496584e-01 -3.52741867e-01 -5.51012099e-01 -5.34621775e-01
-1.40507787e-01 -1.70762450e-01 7.63122916e-01 5.87950945e-01
1.53771579e+00 -7.11250067e-01 -6.10251546e-01 5.69476426e-01
1.74829221e+00 -1.91447005e-01 5.39966106e-01 2.36390203e-01
4.09514070e-01 6.78153038e-01 7.84745574e-01 9.51022089e-01
8.07883143e-01 5.74142694e-01 5.54817438e-01 1.19351037e-01
4.15299833e-01 -2.21152470e-01 7.70368278e-02 4.53128666e-01
1.86736792e-01 -4.15139735e-01 -5.73864043e-01 1.14462554e+00
-2.03503036e+00 -9.90340590e-01 -8.63895863e-02 2.43727732e+00
6.87618971e-01 -3.67571950e-01 3.90607178e-01 3.87717545e-01
7.55972624e-01 2.43605420e-01 -3.29078287e-01 -6.18722081e-01
-2.27171764e-01 -9.05306488e-02 1.15617287e+00 5.66904604e-01
-7.99243391e-01 1.78159818e-01 7.34095144e+00 4.79341179e-01
-5.29551148e-01 -6.97339475e-02 4.46184993e-01 -4.31899011e-01
-1.03082597e+00 2.20410347e-01 -9.26208079e-01 3.86885613e-01
5.96087694e-01 -4.57844198e-01 5.09751141e-01 9.77777362e-01
1.59382790e-01 -4.22527939e-01 -1.34663439e+00 9.62385952e-01
3.33397776e-01 -1.55817270e+00 1.56366855e-01 4.65281039e-01
1.04560316e+00 -3.11141372e-01 1.31559059e-01 -5.00294387e-01
2.40282357e-01 -8.20651531e-01 4.39474076e-01 -9.63208005e-02
6.40671849e-01 -8.70123029e-01 5.25545299e-01 2.34599292e-01
-1.14071453e+00 -4.67120886e-01 -7.96145499e-01 -1.55235976e-01
3.38418722e-01 9.20024991e-01 -8.35124969e-01 5.40131986e-01
7.91111946e-01 4.27586734e-01 -1.78228766e-01 1.41736460e+00
3.34119111e-01 1.13423297e-03 -9.92361784e-01 -2.69951373e-01
2.45459691e-01 2.62471009e-03 5.90215206e-01 9.76531744e-01
7.36860394e-01 4.58234251e-02 1.78108245e-01 2.90332377e-01
-3.54116375e-04 5.43415882e-02 -1.10504496e+00 8.06382447e-02
9.01628554e-01 1.07938445e+00 -8.04622769e-01 -1.07108213e-01
-2.93169677e-01 7.79622495e-01 8.12698305e-02 9.32108238e-02
-7.63616502e-01 -5.66386580e-01 5.32490730e-01 4.02283251e-01
5.74668229e-01 -3.50503415e-01 -6.86108768e-01 -8.48933220e-01
2.94290870e-01 -5.83868384e-01 7.36131907e-01 -3.37827086e-01
-9.94794548e-01 -2.40580425e-01 4.84375387e-01 -9.52975035e-01
-1.76662877e-01 -2.43531466e-01 -3.72069418e-01 3.51993412e-01
-1.60747516e+00 -9.02158260e-01 -2.51590282e-01 7.91289032e-01
-8.24252144e-02 5.55183172e-01 5.56742311e-01 1.82036296e-01
7.66790956e-02 5.82403123e-01 4.42088127e-01 -6.83369994e-01
4.99687940e-01 -1.26704741e+00 -9.46106240e-02 7.99481988e-01
-9.19387788e-02 7.24865615e-01 9.38813686e-01 -5.87725759e-01
-1.97439182e+00 -9.60092902e-01 1.39874482e+00 -3.87237370e-01
5.16075432e-01 -2.20670223e-01 -3.53309929e-01 6.67622387e-01
-7.54027665e-02 9.72968061e-03 8.52802455e-01 5.10158651e-02
-1.26465499e-01 -4.89303768e-01 -1.51711416e+00 3.55681688e-01
1.36972082e+00 -2.09229797e-01 -4.05757844e-01 7.32874334e-01
6.90979719e-01 -2.01164395e-01 -8.50345492e-01 2.15064853e-01
4.85764861e-01 -7.62429476e-01 9.96708095e-01 -1.85040861e-01
7.94933736e-02 -4.97388154e-01 -7.75315464e-01 -8.25677931e-01
-1.35695636e-01 -7.27055013e-01 -1.52362287e-01 1.06422985e+00
3.57938349e-01 -6.83525503e-01 8.90069008e-01 8.40385139e-01
1.14524633e-01 -7.12723196e-01 -8.73926580e-01 -8.84707212e-01
-2.14100257e-01 -9.68988240e-02 8.90441656e-01 9.82603908e-01
3.73423576e-01 -1.69669360e-01 -3.63592297e-01 5.42113662e-01
9.33971763e-01 5.29889166e-01 6.19715810e-01 -1.18312967e+00
-2.80157238e-01 1.31400272e-01 -3.98134083e-01 -1.26286602e+00
-2.40629777e-01 -8.39175701e-01 -1.07148871e-01 -1.89963210e+00
5.71185172e-01 -9.70752716e-01 -2.37233296e-01 5.61091840e-01
2.22837240e-01 2.55392432e-01 9.60797220e-02 1.41766042e-01
-9.96635318e-01 2.40230888e-01 8.98807406e-01 -3.25159021e-02
-2.84724444e-01 -4.65429723e-02 -1.14760423e+00 2.90248930e-01
7.50970542e-01 -6.17044985e-01 -6.55475259e-01 -6.81110203e-01
6.13322794e-01 6.25996441e-02 2.31489435e-01 -8.71855557e-01
5.00111282e-01 -3.16834986e-01 -5.29772900e-02 -9.21024919e-01
2.97385991e-01 -1.29373050e+00 3.57251555e-01 5.54033399e-01
-2.92888880e-01 4.37132984e-01 -9.23966765e-02 7.58053899e-01
7.28420764e-02 -4.14232194e-01 3.27308863e-01 -1.32396817e-01
-4.92717981e-01 3.31137300e-01 1.03514306e-02 -7.66347796e-02
1.41013420e+00 -5.08259356e-01 -3.64334494e-01 -7.39013553e-01
-3.16081673e-01 4.33181286e-01 1.13127625e+00 -1.28737420e-01
7.27440953e-01 -1.24301577e+00 -5.89629352e-01 -1.98329881e-01
1.46607921e-01 1.97144434e-01 3.47014889e-02 9.48370516e-01
-6.94490969e-01 5.55856526e-01 -4.88170469e-03 -4.94464397e-01
-1.18782234e+00 8.13854933e-01 -3.24511051e-01 -1.39577106e-01
-3.24971855e-01 5.87189972e-01 -4.21617739e-02 -2.81057000e-01
5.16017735e-01 -4.51185137e-01 3.22658330e-01 -2.72962488e-02
6.19270265e-01 5.69155991e-01 5.92047460e-02 -2.16023102e-01
-6.08446956e-01 5.61597705e-01 4.29571532e-02 -1.37600601e-01
1.72191787e+00 -3.94158304e-01 -4.42625374e-01 1.43372461e-01
1.14838135e+00 2.91192919e-01 -7.48108685e-01 -6.37305249e-03
-1.22373059e-01 -6.14054799e-01 -1.16702668e-01 -2.46970594e-01
-7.48283982e-01 2.68281847e-01 9.18078199e-02 5.15390635e-01
1.22115719e+00 2.15964571e-01 1.02794337e+00 1.03306782e+00
7.78623700e-01 -1.17020547e+00 -4.06696647e-01 5.65349832e-02
5.24875104e-01 -9.60273802e-01 5.43612301e-01 -4.98840243e-01
-3.06520373e-01 8.35932910e-01 -2.96933521e-02 -6.25201836e-02
6.12135470e-01 4.57978576e-01 -5.51471591e-01 -1.20076150e-01
-5.18584967e-01 -2.38666646e-02 -2.43530601e-01 4.37541634e-01
-8.42286199e-02 -1.33057430e-01 -7.43102491e-01 2.75516272e-01
-8.50194171e-02 -7.59238601e-02 7.44285822e-01 1.34012640e+00
-6.58247828e-01 -1.09708416e+00 -5.64460337e-01 5.34305871e-01
-3.99869502e-01 -4.66642007e-02 -3.18816066e-01 5.34857452e-01
-1.38783708e-01 1.11793697e+00 8.70954618e-02 -9.44679454e-02
-6.90864176e-02 -4.47851449e-01 5.47101259e-01 -5.17573297e-01
-2.81513274e-01 -3.18712771e-01 -8.48736335e-03 -7.04950333e-01
-6.48183107e-01 -7.54464447e-01 -1.32742274e+00 -7.50946283e-01
-3.25613022e-01 5.13916969e-01 9.31457698e-01 5.81632078e-01
5.30797839e-01 -1.62971377e-01 7.70356834e-01 -6.11306548e-01
-8.99592042e-01 -1.04033604e-01 -8.84204686e-01 2.70516872e-01
3.95621240e-01 -3.29913586e-01 -4.71061975e-01 -1.76221043e-01] | [6.632683753967285, 4.945424556732178] |
b734ce50-0b8a-4958-8316-253949834742 | quantum-recurrent-neural-networks-for | 2302.03244 | null | https://arxiv.org/abs/2302.03244v1 | https://arxiv.org/pdf/2302.03244v1.pdf | Quantum Recurrent Neural Networks for Sequential Learning | Quantum neural network (QNN) is one of the promising directions where the near-term noisy intermediate-scale quantum (NISQ) devices could find advantageous applications against classical resources. Recurrent neural networks are the most fundamental networks for sequential learning, but up to now there is still a lack of canonical model of quantum recurrent neural network (QRNN), which certainly restricts the research in the field of quantum deep learning. In the present work, we propose a new kind of QRNN which would be a good candidate as the canonical QRNN model, where, the quantum recurrent blocks (QRBs) are constructed in the hardware-efficient way, and the QRNN is built by stacking the QRBs in a staggered way that can greatly reduce the algorithm's requirement with regard to the coherent time of quantum devices. That is, our QRNN is much more accessible on NISQ devices. Furthermore, the performance of the present QRNN model is verified concretely using three different kinds of classical sequential data, i.e., meteorological indicators, stock price, and text categorization. The numerical experiments show that our QRNN achieves much better performance in prediction (classification) accuracy against the classical RNN and state-of-the-art QNN models for sequential learning, and can predict the changing details of temporal sequence data. The practical circuit structure and superior performance indicate that the present QRNN is a promising learning model to find quantum advantageous applications in the near term. | ['Yongjian Gu', 'Guoqiang Zhong', 'Haiyong Zheng', 'Ruimin Shang', 'Jiaxin Li', 'Shangshang Shi', 'Rongbing Han', 'Zhimin Wang', 'Yanan Li'] | 2023-02-07 | null | null | null | null | ['text-categorization'] | ['natural-language-processing'] | [ 2.63282135e-02 -4.38031077e-01 -1.01715796e-01 -1.19711962e-02
-2.82578856e-01 -9.79082361e-02 4.28483874e-01 -1.26020744e-01
-6.02588117e-01 6.40462101e-01 -2.04738468e-01 -5.67027450e-01
-2.28803381e-01 -1.19661522e+00 -6.76902413e-01 -1.25819242e+00
3.67579252e-01 -1.84812307e-01 4.67303813e-01 -8.72393012e-01
2.74120152e-01 2.79276907e-01 -1.55930781e+00 -5.74933141e-02
7.26098418e-01 1.16138792e+00 1.88104033e-01 3.41990799e-01
3.51150185e-02 7.19406605e-01 -3.84205967e-01 -3.38612020e-01
1.70921788e-01 -7.23338068e-01 -3.76833677e-01 -8.27012539e-01
-3.23892653e-01 -2.37210706e-01 -9.41661537e-01 1.41604054e+00
7.09449470e-01 2.27477461e-01 4.57711518e-01 -4.76201177e-01
-6.80539131e-01 9.49025750e-01 -8.03062320e-02 4.00247782e-01
-3.00641984e-01 1.91769674e-01 1.06759131e+00 -3.47474694e-01
2.71896213e-01 9.59765255e-01 5.56402862e-01 6.22618318e-01
-9.24783230e-01 -1.05130625e+00 -5.09809017e-01 5.29494762e-01
-1.63011515e+00 -3.06070030e-01 6.80617154e-01 6.57515377e-02
1.03639197e+00 4.18034866e-02 8.08489621e-01 8.80718946e-01
6.03303969e-01 3.88344109e-01 1.07588518e+00 -4.22961682e-01
4.43071932e-01 -1.28000587e-01 3.81232947e-01 8.76203716e-01
1.19963102e-01 5.80503404e-01 -6.73236847e-01 2.10787699e-01
7.63768196e-01 3.24032038e-01 -4.57845815e-02 3.51812571e-01
-1.08549166e+00 8.15852106e-01 1.02937782e+00 6.58502519e-01
-2.08764881e-01 4.35752034e-01 5.48225582e-01 2.97622591e-01
8.23740065e-02 8.67393836e-02 -1.14209883e-01 -2.26493031e-01
-5.73395193e-01 -3.88181303e-04 3.42942357e-01 6.28142059e-01
8.30525994e-01 3.48923594e-01 1.39525563e-01 5.02585292e-01
1.66597843e-01 9.56886113e-01 9.12161112e-01 -4.68731493e-01
2.82936305e-01 3.08553904e-01 -3.10079843e-01 -8.54035735e-01
-6.74885392e-01 -5.61653852e-01 -1.39822102e+00 -2.36636028e-01
1.08835004e-01 -1.38109490e-01 -5.56979656e-01 1.38819396e+00
-1.61220118e-01 3.21277350e-01 5.28426588e-01 8.62723529e-01
1.05630624e+00 1.34064853e+00 -4.44144130e-01 -5.13434350e-01
1.21377695e+00 -5.60166299e-01 -8.90202284e-01 2.06446782e-01
9.32968855e-01 -4.14815545e-01 7.60437429e-01 4.37964261e-01
-5.76695025e-01 -7.17391193e-01 -1.22242641e+00 2.68358644e-02
-4.88022149e-01 8.01243335e-02 8.38851392e-01 9.39372182e-01
-7.35363960e-01 9.66580153e-01 -9.02114451e-01 -1.28013089e-01
-1.01250291e-01 5.85580885e-01 3.20033506e-02 2.33833358e-01
-1.88085175e+00 7.15918183e-01 7.85646796e-01 7.26439893e-01
-5.25267422e-01 1.63439989e-01 -1.47987470e-01 3.76753844e-02
6.30186573e-02 -3.97765070e-01 1.17474592e+00 -4.78167653e-01
-2.16993546e+00 2.35778064e-01 -4.21479642e-02 -7.08173573e-01
-1.73283294e-01 2.83447534e-01 -6.61870837e-01 -8.01620856e-02
-3.75866205e-01 1.03684112e-01 6.40897274e-01 -1.28331438e-01
-4.41229433e-01 -4.29295123e-01 -5.72195426e-02 2.19949722e-01
-5.53797543e-01 -2.46207759e-01 7.64451921e-02 -1.44931197e-01
4.64596093e-01 -1.06983149e+00 -3.71220410e-01 -7.74560273e-01
-2.43867382e-01 -6.74799919e-01 5.96498251e-01 1.97849460e-02
1.50680411e+00 -2.19703746e+00 1.92246726e-03 1.69079795e-01
-9.71078500e-02 4.80463028e-01 2.56583035e-01 5.04897654e-01
2.44154856e-01 6.64667562e-02 8.62092152e-02 3.30880016e-01
-1.85372412e-01 1.79459542e-01 -4.66425806e-01 3.80204290e-01
-3.13668847e-01 1.16615665e+00 -6.66973472e-01 -1.99206024e-01
7.71749765e-02 2.13706329e-01 -3.23776722e-01 -2.03047022e-01
-1.66032419e-01 4.21873450e-01 -6.84242666e-01 2.52151042e-01
4.65506673e-01 -2.36015216e-01 -4.20147069e-02 -3.77789527e-01
-4.67117041e-01 4.44869787e-01 -1.18957722e+00 1.42723823e+00
-1.88038826e-01 4.62889493e-01 -6.10399961e-01 -1.02700722e+00
1.20328367e+00 2.08447561e-01 4.94222417e-02 -1.38025284e+00
3.06829304e-01 7.10611343e-01 2.40704298e-01 -4.16685194e-01
5.56254685e-01 -5.24312615e-01 -2.78178751e-01 4.38227206e-01
3.84460576e-02 -3.32199410e-02 7.06535764e-04 -4.26720567e-02
7.40973353e-01 -2.31424853e-01 1.75435781e-01 -1.85115248e-01
5.66049814e-01 -4.24185425e-01 6.65697873e-01 8.62022698e-01
-1.39862224e-01 7.59920180e-02 3.02683473e-01 -5.30985892e-01
-1.04449511e+00 -8.33000362e-01 -4.00734901e-01 7.34750807e-01
6.01344347e-01 -3.04087728e-01 -5.11023521e-01 3.73319052e-02
-4.96992171e-01 3.30247313e-01 -1.04767248e-01 -4.84769464e-01
-5.27691722e-01 -1.21316719e+00 9.55754459e-01 2.31503561e-01
1.05388367e+00 -1.16640365e+00 -4.26426589e-01 3.47045422e-01
1.31617770e-01 -8.23942840e-01 1.57600224e-01 4.66089517e-01
-1.24097919e+00 -4.87503409e-01 -4.41364229e-01 -6.73889279e-01
7.83657376e-03 2.07437053e-01 3.42722416e-01 -2.82847229e-03
2.40743175e-01 -5.19429147e-01 -5.14350593e-01 -1.52591825e-01
-4.69494373e-01 4.19260889e-01 5.11093259e-01 3.84477526e-02
3.34262252e-01 -5.88587403e-01 -6.54558778e-01 4.03631814e-02
-9.56429780e-01 9.28611681e-02 7.59522796e-01 1.03015327e+00
6.05565667e-01 5.08070588e-01 6.14859223e-01 -7.26159513e-01
4.40048009e-01 -1.18793719e-01 -9.74200904e-01 4.44803871e-02
-6.05241597e-01 4.08755630e-01 1.30567157e+00 -1.41211465e-01
-9.22352254e-01 -1.81601197e-01 -5.88525116e-01 1.38533413e-01
2.39811227e-01 6.51475728e-01 2.37292916e-01 -3.28400314e-01
7.63820469e-01 6.12217247e-01 -2.80561924e-01 -2.64877379e-01
2.71520764e-01 1.03170598e+00 1.71830326e-01 -1.10840760e-01
6.65793777e-01 2.80413806e-01 5.63976347e-01 -1.15856206e+00
-6.60348177e-01 -2.15017781e-01 -3.80206555e-01 1.41680036e-02
8.55058372e-01 -8.16032529e-01 -1.31566846e+00 7.97874629e-01
-1.26038420e+00 -1.10799208e-01 1.63145944e-01 9.22121286e-01
-2.84240037e-01 2.23190561e-01 -1.15902495e+00 -1.11900830e+00
-6.05693221e-01 -1.29641497e+00 6.91270232e-01 6.97744548e-01
6.48358405e-01 -7.14134216e-01 -9.99316424e-02 1.11772187e-01
3.81810337e-01 -2.87715912e-01 9.45614815e-01 -3.07954580e-01
-9.52720821e-01 -4.88816984e-02 -2.23172963e-01 3.19459647e-01
-4.11230683e-01 -3.48576456e-02 -9.59845483e-01 -1.52952343e-01
2.25462958e-01 -3.49274486e-01 9.72583473e-01 1.37587428e-01
8.88799489e-01 9.52637047e-02 -1.10980883e-01 6.24129415e-01
1.62336540e+00 7.46064186e-01 7.89816260e-01 1.31535009e-01
6.81866527e-01 -1.57185555e-01 2.71999419e-01 1.82627738e-01
1.89431190e-01 5.68871260e-01 3.65706146e-01 1.84319884e-01
5.02721846e-01 -3.03500682e-01 6.37256444e-01 1.81244063e+00
-3.38943183e-01 -3.03499531e-02 -8.48971725e-01 -9.11023691e-02
-1.81248951e+00 -1.26410878e+00 -3.94401282e-01 2.14951563e+00
5.68719268e-01 3.33116114e-01 -3.42780948e-01 2.42324650e-01
6.05509758e-01 3.52839828e-01 -6.83600247e-01 -5.16653419e-01
-2.60861069e-01 4.33586627e-01 6.71515644e-01 1.13415524e-01
-8.98242831e-01 9.26408589e-01 5.16810417e+00 1.33003986e+00
-1.37322652e+00 2.11685956e-01 4.24323887e-01 3.62669706e-01
-9.82958302e-02 1.22995608e-01 -1.16342831e+00 5.80146372e-01
1.63704407e+00 3.99851501e-02 6.08792484e-01 5.14237404e-01
2.33503297e-01 -4.05992866e-02 -6.19854033e-01 1.42644072e+00
-3.54444981e-01 -1.61440563e+00 -1.20387506e-02 -1.10433966e-01
7.37674415e-01 5.02484083e-01 1.52315348e-01 5.01763940e-01
-2.17517287e-01 -8.41477573e-01 5.08709192e-01 7.90421009e-01
7.37355232e-01 -8.96037221e-01 1.17535055e+00 7.39744008e-01
-1.16140330e+00 -3.65821749e-01 -8.28858554e-01 -4.21981454e-01
-1.26940936e-01 5.31091690e-01 -1.53780594e-01 7.11519897e-01
7.43469417e-01 9.27159905e-01 -5.01329124e-01 7.55807102e-01
-6.79871533e-03 7.47099698e-01 -3.62423092e-01 -1.16951919e+00
4.11245704e-01 -7.94771850e-01 6.74744993e-02 7.78101623e-01
4.84869063e-01 5.23973286e-01 -3.06832641e-01 6.74768269e-01
-1.76943734e-01 2.68938929e-01 -7.70180762e-01 -2.76103109e-01
4.31000412e-01 9.43452179e-01 -7.97726929e-01 -2.54840851e-01
-2.66156822e-01 8.13562989e-01 1.10936321e-01 2.08771676e-01
-7.62423515e-01 -8.11056674e-01 -1.64145157e-02 -4.11294580e-01
1.53291032e-01 -2.36334980e-01 -2.49762639e-01 -1.43142331e+00
-1.41101211e-01 -5.24706483e-01 -9.73882824e-02 -6.77964687e-01
-1.06042540e+00 6.45882964e-01 -6.56575143e-01 -1.40434110e+00
5.49668223e-02 -6.92975402e-01 -2.87168115e-01 8.38185906e-01
-1.35889804e+00 -5.38141966e-01 6.54768124e-02 4.55798060e-01
-9.17183608e-02 -2.11529225e-01 1.05323625e+00 3.71732086e-01
-9.28796947e-01 4.74876881e-01 1.13596058e+00 1.39647990e-01
1.59977213e-01 -6.80453062e-01 3.16128224e-01 7.33087718e-01
2.25485876e-01 9.14536119e-01 5.13275146e-01 -2.51798153e-01
-1.69099462e+00 -6.95977151e-01 6.22958541e-01 1.49147496e-01
8.49892616e-01 -4.35163140e-01 -7.54539490e-01 7.43847340e-02
-1.25969425e-01 -1.09385848e-01 4.17316377e-01 4.21560593e-02
-1.10565677e-01 -7.33998656e-01 -6.10678494e-01 6.25479698e-01
8.56037974e-01 -1.19220984e+00 -3.66335124e-01 4.45202351e-01
9.02837157e-01 -3.74805033e-01 -7.34517097e-01 3.06732088e-01
8.78509939e-01 -1.22447741e+00 4.90878582e-01 -1.71809897e-01
2.25231126e-01 -5.45170128e-01 -2.73233593e-01 -1.05696774e+00
-1.05248377e-01 -7.84214318e-01 1.01357676e-01 8.21838260e-01
3.46008450e-01 -9.77259874e-01 9.15068865e-01 -6.69002831e-02
-1.00824781e-01 -7.18497634e-01 -1.37819302e+00 -7.97075450e-01
1.76755860e-01 -3.79459739e-01 4.53039020e-01 4.33683962e-01
4.94463071e-02 6.96922183e-01 -3.84499878e-01 5.47181852e-02
2.91530520e-01 1.76966280e-01 2.28362992e-01 -1.16068161e+00
-2.38997221e-01 -5.07634997e-01 -6.36764348e-01 -1.49482501e+00
-1.65589992e-02 -9.73120809e-01 -2.14980450e-02 -1.02941811e+00
7.51998872e-02 -6.64867580e-01 -7.28209913e-01 -1.13952495e-01
1.28581956e-01 1.95893288e-01 4.84740622e-02 5.10191560e-01
-6.31241977e-01 9.45895433e-01 1.15184760e+00 -6.27291948e-02
-2.43231803e-01 3.40601623e-01 -1.50457144e-01 3.55739266e-01
7.80075192e-01 -5.75975537e-01 -2.53549904e-01 -2.38401130e-01
7.23217785e-01 4.83888268e-01 1.48090303e-01 -1.23996556e+00
7.07490146e-01 1.75760031e-01 8.19123443e-03 -6.34876370e-01
3.38763833e-01 -5.62916398e-01 1.77582111e-02 9.35986161e-01
-1.37502730e-01 1.54033288e-01 -2.99382985e-01 6.46421373e-01
-2.70441175e-01 -4.84609872e-01 7.14548230e-01 4.73383302e-03
-6.35330319e-01 4.15935308e-01 -3.68685007e-01 -3.22637767e-01
5.61625004e-01 -1.64536789e-01 -5.78319967e-01 -1.11323833e-01
-3.41160327e-01 -1.79708332e-01 4.85691167e-02 7.70700425e-02
5.38126945e-01 -1.23914433e+00 -5.79543896e-02 3.13871622e-01
-2.06813291e-01 -1.86552137e-01 6.40717387e-01 8.13253522e-01
-6.50022686e-01 1.01966155e+00 -1.68842599e-01 -5.82448661e-01
-6.03929937e-01 4.22658920e-01 6.13348246e-01 -2.63464838e-01
-4.97802734e-01 7.14642823e-01 -1.54824436e-01 -4.37774181e-01
3.56879495e-02 -3.95376146e-01 -3.66391420e-01 7.64378998e-03
3.85388196e-01 3.54292423e-01 2.91613758e-01 -6.56660140e-01
-1.58804268e-01 7.61245549e-01 1.20721713e-01 -1.79934740e-01
1.10185349e+00 3.42010111e-02 -4.27205950e-01 9.64899242e-01
1.23729753e+00 -3.95286918e-01 -6.24551296e-01 -5.05886257e-01
-1.08444460e-01 4.03542995e-01 1.18065059e-01 -1.98322505e-01
-6.74490750e-01 1.29543746e+00 1.04395413e+00 4.57169086e-01
1.06301880e+00 -3.27865988e-01 1.24301505e+00 1.13171124e+00
8.70345056e-01 -1.15645742e+00 -1.68118224e-01 9.98075962e-01
-4.87801693e-02 -9.71202552e-01 -1.31297454e-01 1.96022362e-01
-1.69311196e-01 1.49693334e+00 2.20860347e-01 -1.69536352e-01
8.65736187e-01 -2.68016517e-01 -1.25029296e-01 -1.33917958e-01
-6.25012696e-01 -1.47736847e-01 -8.15088004e-02 -1.27196714e-01
3.59266996e-01 3.69096071e-01 -5.90089738e-01 6.18450880e-01
-4.33653444e-01 1.50654623e-02 7.22365022e-01 5.59122860e-01
-5.91075361e-01 -9.98874843e-01 -8.14805478e-02 5.32886684e-01
-3.39795440e-01 -3.50393236e-01 2.82444000e-01 2.72882372e-01
3.24641943e-01 8.05539012e-01 -2.01587752e-01 -9.98836398e-01
2.82597929e-01 -4.40289862e-02 2.59054303e-01 -4.00653780e-01
-5.97173512e-01 -1.36715472e-01 -3.61042529e-01 -3.78077507e-01
-5.01362026e-01 -3.95468593e-01 -1.68294704e+00 -4.47185993e-01
-8.50695550e-01 4.82705653e-01 7.76529074e-01 1.23542809e+00
9.76936147e-02 4.41891968e-01 7.82357574e-01 -5.64078033e-01
-7.34987676e-01 -8.01273763e-01 -8.66970479e-01 -1.03115581e-01
3.08357060e-01 -2.91638494e-01 -1.70610070e-01 -5.21896541e-01] | [5.564509391784668, 4.971056938171387] |
f0ba841d-0906-408c-912a-ff7f3321d5a5 | cross-lingual-low-resource-set-to-description | 2005.08188 | null | https://arxiv.org/abs/2005.08188v1 | https://arxiv.org/pdf/2005.08188v1.pdf | Cross-Lingual Low-Resource Set-to-Description Retrieval for Global E-Commerce | With the prosperous of cross-border e-commerce, there is an urgent demand for designing intelligent approaches for assisting e-commerce sellers to offer local products for consumers from all over the world. In this paper, we explore a new task of cross-lingual information retrieval, i.e., cross-lingual set-to-description retrieval in cross-border e-commerce, which involves matching product attribute sets in the source language with persuasive product descriptions in the target language. We manually collect a new and high-quality paired dataset, where each pair contains an unordered product attribute set in the source language and an informative product description in the target language. As the dataset construction process is both time-consuming and costly, the new dataset only comprises of 13.5k pairs, which is a low-resource setting and can be viewed as a challenging testbed for model development and evaluation in cross-border e-commerce. To tackle this cross-lingual set-to-description retrieval task, we propose a novel cross-lingual matching network (CLMN) with the enhancement of context-dependent cross-lingual mapping upon the pre-trained monolingual BERT representations. Experimental results indicate that our proposed CLMN yields impressive results on the challenging task and the context-dependent cross-lingual mapping on BERT yields noticeable improvement over the pre-trained multi-lingual BERT model. | ['Xiaozhong Liu', 'Jian Wang', 'Hongsong Li', 'Chang Liu', 'Rui Yan', 'Dongyan Zhao', 'Lidong Bing', 'Juntao Li'] | 2020-05-17 | null | null | null | null | ['cross-lingual-information-retrieval'] | ['natural-language-processing'] | [-2.40382031e-01 -6.25503123e-01 -7.05801904e-01 -7.85579562e-01
-1.55499232e+00 -9.72452164e-01 6.58428311e-01 2.07301244e-01
-4.53772128e-01 3.81432742e-01 2.50003710e-02 -2.03555465e-01
-3.34981173e-01 -7.21240580e-01 -6.92236423e-01 -3.52175057e-01
8.94019380e-02 1.16965330e+00 -2.99351960e-01 -8.71756792e-01
-1.63153917e-01 1.75777242e-01 -1.24375892e+00 6.00385189e-01
8.64225984e-01 1.29924726e+00 2.45046094e-01 -1.79728158e-02
-3.14316273e-01 2.06917822e-01 -1.69691220e-01 -1.22900236e+00
7.09155023e-01 -1.31610081e-01 -7.52091587e-01 -1.86318427e-01
4.67139959e-01 -4.91826348e-02 1.17891051e-01 1.09673810e+00
5.15547872e-01 7.47493431e-02 4.77639556e-01 -1.29221129e+00
-1.16579688e+00 9.13662434e-01 -4.81547147e-01 -4.87255566e-02
2.84484744e-01 -2.22218558e-01 1.56617260e+00 -9.15070355e-01
7.53796756e-01 1.24116468e+00 6.17703199e-01 2.32552722e-01
-1.15612435e+00 -8.53457868e-01 2.07837164e-01 1.76982954e-01
-1.64370775e+00 -3.63265276e-01 9.14704323e-01 -9.86668915e-02
7.97839761e-01 1.63050413e-01 3.57537419e-01 1.02152419e+00
-4.92482893e-02 9.86090183e-01 8.42901587e-01 -3.77087653e-01
-3.55288833e-01 8.47712040e-01 2.75937356e-02 2.72119462e-01
1.61410555e-01 1.93905503e-01 -5.71748078e-01 -4.47488986e-02
5.97267151e-01 9.15964693e-03 1.01534121e-01 -4.93581086e-01
-9.83914018e-01 1.11587477e+00 6.35692596e-01 4.26167935e-01
-4.76818532e-01 -2.43410498e-01 7.37172067e-01 7.05555797e-01
6.49248064e-01 4.55603778e-01 -9.89659250e-01 1.69433445e-01
-5.80415964e-01 3.49476397e-01 8.71297121e-01 1.71145558e+00
7.06559479e-01 -4.75065380e-01 2.73281634e-01 1.32428384e+00
5.53335011e-01 8.55005920e-01 4.43182439e-01 -2.56190568e-01
7.73004889e-01 4.67576623e-01 1.88508064e-01 -9.68796372e-01
1.30979801e-02 -6.02022529e-01 -7.35989332e-01 -5.51496625e-01
2.55863667e-01 2.58871526e-01 -3.30280066e-01 1.75826311e+00
1.77067086e-01 -5.81920862e-01 4.02678639e-01 1.07737446e+00
1.04736054e+00 7.60189116e-01 2.77348518e-01 -1.01854421e-01
1.51794386e+00 -1.24998307e+00 -7.81906486e-01 -3.27108055e-01
9.03556287e-01 -1.24998581e+00 1.20592582e+00 5.59865236e-02
-9.12585139e-01 -7.16059208e-01 -1.10436428e+00 -3.34750593e-01
-9.52331841e-01 1.62510768e-01 9.38430965e-01 4.54537898e-01
-5.08882999e-01 1.80730090e-01 -7.71155506e-02 -2.92785227e-01
-9.67491418e-03 2.68116891e-01 -5.85756540e-01 -6.33241355e-01
-1.79154158e+00 9.58200336e-01 5.52078128e-01 3.54478389e-01
-3.28292996e-01 -7.06846893e-01 -1.17311633e+00 -6.76917881e-02
2.66833425e-01 -3.59096736e-01 1.09458101e+00 -1.04332590e+00
-1.03833222e+00 1.05857337e+00 1.68330520e-01 -4.41819578e-02
3.71436834e-01 -9.05176252e-02 -1.12301052e+00 -3.53900403e-01
6.14639699e-01 7.78269589e-01 3.51537287e-01 -1.37058151e+00
-8.26415002e-01 -4.83209759e-01 1.55856282e-01 3.93442303e-01
-5.81107959e-02 3.03768516e-01 -9.76860583e-01 -6.33017361e-01
6.39008209e-02 -8.43969643e-01 1.16287068e-01 -3.10333848e-01
-3.53802182e-02 -5.57771087e-01 4.05675799e-01 -7.11305022e-01
1.01158035e+00 -2.16114259e+00 -3.74434292e-01 1.21925354e-01
-3.80177021e-01 2.60852575e-01 -5.08969605e-01 6.50381863e-01
1.23879068e-01 5.56723997e-02 2.08506241e-01 -2.63177097e-01
6.03736162e-01 2.50570953e-01 -2.35464156e-01 9.39696282e-02
2.10150689e-01 1.25241423e+00 -1.03446472e+00 -7.02430487e-01
-6.71930760e-02 4.05972034e-01 -2.55962640e-01 1.29518554e-01
-6.60626814e-02 1.39179453e-01 -7.91028500e-01 1.02027905e+00
8.50505114e-01 -1.65957194e-02 3.37840140e-01 -5.21824419e-01
1.87610924e-01 6.16949916e-01 -8.25043917e-01 2.05988908e+00
-9.08052027e-01 6.31324854e-03 1.12782409e-02 -6.29034400e-01
1.22767806e+00 2.75373191e-01 5.89852929e-01 -1.48784924e+00
4.27508354e-02 8.46332312e-01 -2.74707437e-01 -3.38896960e-01
7.93569922e-01 -3.84281874e-01 -7.57007718e-01 4.66771811e-01
-8.21639411e-03 3.97161320e-02 4.35712487e-01 -2.72212159e-02
3.63567978e-01 3.39461476e-01 8.18619877e-02 -4.01381791e-01
5.10860503e-01 3.95008922e-01 7.73187041e-01 4.47288394e-01
-1.91271216e-01 1.43795341e-01 -1.67900860e-01 -6.39790595e-01
-1.04752421e+00 -1.03228831e+00 -5.13138354e-01 1.36242568e+00
4.69397038e-01 -4.93580163e-01 -2.90769428e-01 -1.01585090e+00
2.32739583e-01 5.11498511e-01 -2.27489114e-01 -1.50492072e-01
-6.10720217e-01 -5.48098147e-01 4.85145956e-01 3.79382372e-01
9.54570353e-01 -1.00086653e+00 2.95720458e-01 3.74596447e-01
-4.46780443e-01 -1.16343188e+00 -1.01348102e+00 1.97374433e-01
-4.02841777e-01 -6.80730760e-01 -4.53040987e-01 -1.39739203e+00
3.02830637e-01 3.57305735e-01 1.58283782e+00 -3.72974843e-01
1.97435170e-02 -2.30073556e-02 -5.43754041e-01 -3.71417999e-01
-4.60627288e-01 3.27572405e-01 -4.83029559e-02 -4.40717489e-03
1.22948062e+00 1.48847327e-02 -4.31205571e-01 8.31710875e-01
-8.63378763e-01 -1.42741308e-01 7.84701586e-01 1.06732547e+00
8.82907927e-01 1.84315115e-01 9.66058791e-01 -1.05825830e+00
7.12263703e-01 -6.56456769e-01 -5.69941342e-01 6.36011541e-01
-7.55365551e-01 -1.71720851e-02 6.99054360e-01 -5.07483900e-01
-1.09751296e+00 -7.96299055e-02 -1.68582603e-01 -1.09834358e-01
2.74189562e-01 9.91497934e-01 -5.50525010e-01 1.80510998e-01
2.90304929e-01 4.39352095e-02 -2.42915422e-01 -8.48278165e-01
5.83211601e-01 1.11466765e+00 4.02288646e-01 -6.48272932e-01
5.34475744e-01 -1.41339988e-01 -4.51088458e-01 -1.00623444e-01
-9.64773953e-01 -1.02484214e+00 -8.65176022e-01 1.52262390e-01
3.69048864e-01 -1.07426417e+00 -6.27612352e-01 7.66158625e-02
-1.07232511e+00 5.75246327e-02 -2.83066742e-02 5.04050493e-01
-4.10277784e-01 -1.52945638e-01 -6.81934774e-01 -3.15905064e-01
-5.85350573e-01 -1.04398048e+00 1.28533566e+00 -2.19973385e-01
-1.24432445e-02 -1.18627727e+00 -6.37951121e-02 8.52895200e-01
4.33133990e-01 -3.23775023e-01 1.16536069e+00 -9.59684670e-01
-4.22068536e-01 -6.00593030e-01 -4.63841975e-01 4.39728826e-01
2.59531021e-01 -6.87098563e-01 -6.76113486e-01 -4.82498705e-01
-1.68055415e-01 -6.07591033e-01 3.97103012e-01 -9.43226367e-02
4.49633479e-01 -3.40175480e-01 -1.84754714e-01 4.07234907e-01
1.57092702e+00 2.74291456e-01 2.62734979e-01 3.79358500e-01
5.06766319e-01 9.71698344e-01 1.39113486e+00 1.16457619e-01
8.17500830e-01 1.18266177e+00 -1.77866176e-01 -4.40616667e-01
-4.03360650e-02 -5.73666811e-01 1.69793352e-01 1.26512337e+00
4.30713892e-01 -1.62067741e-01 -5.42253494e-01 7.03446686e-01
-1.89586723e+00 -7.25798607e-01 1.17527530e-01 2.22492385e+00
9.85501349e-01 -8.65882337e-02 1.40075712e-02 -4.42552716e-01
5.61564028e-01 -1.64627180e-01 -6.83560014e-01 -4.01538223e-01
-3.52898836e-01 -1.07111990e-01 5.27204633e-01 3.41521531e-01
-1.26729333e+00 1.33586860e+00 5.70768642e+00 1.17019904e+00
-9.88015056e-01 2.94741511e-01 6.08022153e-01 2.53174841e-01
-5.91829598e-01 -3.22151572e-01 -1.07769716e+00 3.19090158e-01
7.97594428e-01 -3.81243825e-01 4.17772591e-01 8.58545125e-01
-7.28066191e-02 3.91869307e-01 -1.33936810e+00 1.17499506e+00
2.38356695e-01 -9.79793847e-01 2.98630834e-01 8.52197185e-02
6.15080953e-01 1.14652105e-01 1.59693420e-01 8.80102515e-01
4.82793212e-01 -7.55428195e-01 6.87187016e-01 -1.87673599e-01
9.99193072e-01 -8.45501244e-01 1.18628824e+00 2.69980937e-01
-1.56660092e+00 1.33208200e-01 -4.12804693e-01 6.43352866e-01
3.12943250e-01 1.13793902e-01 -6.25617325e-01 9.06321824e-01
7.56822348e-01 9.75567758e-01 -2.10808501e-01 5.44109881e-01
1.70102999e-01 -7.19570294e-02 -3.10023218e-01 -7.58918226e-02
6.51377916e-01 -5.69620788e-01 1.58609785e-02 1.26029694e+00
3.45018715e-01 -1.93048745e-01 3.80158275e-01 9.45259392e-01
-4.63539213e-01 6.61613584e-01 -8.32996905e-01 -2.23364174e-01
4.93192971e-01 1.33206236e+00 -1.49956688e-01 -1.85395479e-01
-5.83501041e-01 9.64955091e-01 2.59512156e-01 3.03419948e-01
-7.12841451e-01 -2.95527637e-01 7.32767403e-01 -7.87758380e-02
2.38625035e-01 4.48790081e-02 8.78714472e-02 -1.02207053e+00
5.05867779e-01 -1.01883006e+00 6.72306895e-01 -6.47389472e-01
-1.97597635e+00 9.15308058e-01 -6.95606694e-02 -1.48686409e+00
-4.98079419e-01 -5.60041189e-01 3.71835500e-01 1.09101903e+00
-2.05674434e+00 -1.91005027e+00 2.83494592e-01 5.37831962e-01
7.15431750e-01 -2.38749102e-01 1.04999459e+00 1.01675999e+00
-1.74432620e-01 9.51642573e-01 4.36576962e-01 2.80703783e-01
1.04408491e+00 -8.52452993e-01 3.38943541e-01 3.31225812e-01
2.87936509e-01 1.03908110e+00 1.92501783e-01 -5.93968868e-01
-1.75121319e+00 -1.17870712e+00 1.64081240e+00 -4.06203806e-01
7.37125099e-01 -7.92551219e-01 -6.63900375e-01 7.88140893e-01
5.31692952e-02 -9.38517526e-02 9.20031607e-01 5.02278864e-01
-7.64019132e-01 -7.09609866e-01 -1.27380288e+00 2.98500568e-01
9.50850427e-01 -8.42171669e-01 -3.68750185e-01 5.34316778e-01
7.26506412e-01 -1.41679496e-01 -1.33487844e+00 2.36368492e-01
6.85142219e-01 -3.39080185e-01 1.10315692e+00 -6.73471510e-01
1.34319644e-02 -5.00002876e-02 -5.82837343e-01 -1.29017317e+00
-4.05942470e-01 -3.87303621e-01 8.37581038e-01 1.64969981e+00
9.10334229e-01 -5.47964871e-01 4.88049418e-01 7.69817770e-01
-1.59630626e-01 -5.12982190e-01 -7.78125107e-01 -9.53090549e-01
1.91318691e-01 -3.85735333e-01 1.00473595e+00 1.06432259e+00
1.12300403e-01 6.72810674e-01 -2.89949656e-01 -8.42532814e-02
5.26244938e-01 6.09989524e-01 5.06103814e-01 -1.08101189e+00
-1.49399132e-01 -3.37295830e-01 -1.33479625e-01 -1.48139584e+00
2.54999965e-01 -1.30724204e+00 2.09995002e-01 -1.19014573e+00
2.27518454e-01 -1.22047031e+00 -4.92907077e-01 3.45483065e-01
1.26953855e-01 3.48056704e-01 1.83269661e-02 4.84718233e-01
-7.31951058e-01 4.97078240e-01 1.33529091e+00 -4.81023699e-01
-9.55139752e-03 -2.62676906e-02 -8.65969300e-01 6.59094527e-02
4.50332046e-01 -4.92860347e-01 -5.56164384e-01 -5.68650544e-01
4.10029471e-01 2.67369626e-03 -1.49876237e-01 1.07681407e-02
1.55323118e-01 2.71187238e-02 -7.73763210e-02 -5.59971035e-01
5.96820414e-01 -1.28488839e+00 2.20665336e-01 -1.01402514e-01
-5.83655417e-01 4.32027936e-01 9.81172398e-02 3.20343345e-01
-6.93562508e-01 -2.24597111e-01 3.05388480e-01 -2.32091293e-01
-9.59547400e-01 3.94477338e-01 2.45669186e-01 -1.29280575e-02
6.97016239e-01 1.05015114e-01 -3.77705514e-01 -2.57700831e-01
-3.31723303e-01 5.13929307e-01 1.80506900e-01 9.57785845e-01
2.48439372e-01 -1.87536573e+00 -9.60376859e-01 4.03976053e-01
8.56247663e-01 -3.77844900e-01 1.52809471e-01 5.19219637e-01
-1.82279721e-01 9.51612175e-01 -2.93637663e-01 -3.53804201e-01
-1.12124825e+00 7.45639503e-01 7.61054903e-02 -6.34437621e-01
-1.18303075e-01 6.89666212e-01 1.41400844e-01 -1.02384531e+00
7.08229020e-02 1.24986671e-01 -1.77915260e-01 7.57665783e-02
3.18458855e-01 -2.15083718e-01 3.66127670e-01 -1.18727744e+00
-2.87018090e-01 5.61078548e-01 -6.44335747e-01 -2.78934352e-02
1.34896386e+00 -4.51688379e-01 -1.67141348e-01 3.75200391e-01
1.73347831e+00 -3.10326636e-01 -6.32753909e-01 -8.07799459e-01
2.50526994e-01 -6.50353670e-01 1.70315742e-01 -1.18384755e+00
-1.05018997e+00 5.63159227e-01 3.93865258e-01 7.79979154e-02
1.05558467e+00 2.34982252e-01 1.14267170e+00 4.48575377e-01
9.40565705e-01 -1.36196971e+00 -2.85652727e-01 2.53860354e-01
1.07095206e+00 -1.73249793e+00 -3.96999776e-01 -7.54095972e-01
-9.11417961e-01 5.83500147e-01 4.28226441e-01 2.58494318e-01
7.04627514e-01 -1.71954647e-01 6.63506567e-01 -4.16727662e-01
-6.78285241e-01 -2.44931132e-01 5.15306771e-01 6.17762089e-01
6.64768815e-01 3.11752707e-01 -2.14075208e-01 7.61626065e-01
-2.09671900e-01 -2.29773641e-01 -3.81905854e-01 7.52023339e-01
2.99302608e-01 -1.65733802e+00 1.72752440e-01 1.13939263e-01
-1.85959592e-01 -3.89346719e-01 -4.33525652e-01 1.04228628e+00
9.61210355e-02 1.13355517e+00 -3.71214747e-02 -3.33977669e-01
5.02409518e-01 -5.61040230e-02 3.20242286e-01 -3.89486849e-01
-9.91240621e-01 3.52227181e-01 5.97995520e-01 -4.32227045e-01
-3.78903329e-01 -5.00584424e-01 -8.63493264e-01 -4.70660627e-01
-5.31154037e-01 4.96310920e-01 1.08275557e+00 6.40434980e-01
5.37688017e-01 -7.31794536e-02 7.87617207e-01 -3.09988350e-01
-7.93765128e-01 -1.02158999e+00 -9.95237887e-01 1.00331271e+00
-3.21001299e-02 -6.21446669e-01 2.01699883e-02 -5.98022193e-02] | [11.495587348937988, 9.922776222229004] |
153c4613-858c-47c1-ba32-c561e8357842 | towards-deep-observation-a-systematic-survey | 2201.07935 | null | https://arxiv.org/abs/2201.07935v2 | https://arxiv.org/pdf/2201.07935v2.pdf | Towards deep observation: A systematic survey on artificial intelligence techniques to monitor fetus via Ultrasound Images | Developing innovative informatics approaches aimed to enhance fetal monitoring is a burgeoning field of study in reproductive medicine. Several reviews have been conducted regarding Artificial intelligence (AI) techniques to improve pregnancy outcomes. They are limited by focusing on specific data such as mother's care during pregnancy. This systematic survey aims to explore how artificial intelligence (AI) can assist with fetal growth monitoring via Ultrasound (US) image. We used eight medical and computer science bibliographic databases, including PubMed, Embase, PsycINFO, ScienceDirect, IEEE explore, ACM Library, Google Scholar, and the Web of Science. We retrieved studies published between 2010 to 2021. Data extracted from studies were synthesized using a narrative approach. Out of 1269 retrieved studies, we included 107 distinct studies from queries that were relevant to the topic in the survey. We found that 2D ultrasound images were more popular (n=88) than 3D and 4D ultrasound images (n=19). Classification is the most used method (n=42), followed by segmentation (n=31), classification integrated with segmentation (n=16) and other miscellaneous such as object-detection, regression and reinforcement learning (n=18). The most common areas within the pregnancy domain were the fetus head (n=43), then fetus body (n=31), fetus heart (n=13), fetus abdomen (n=10), and lastly the fetus face (n=10). In the most recent studies, deep learning techniques were primarily used (n=81), followed by machine learning (n=16), artificial neural network (n=7), and reinforcement learning (n=2). AI techniques played a crucial role in predicting fetal diseases and identifying fetus anatomy structures during pregnancy. More research is required to validate this technology from a physician's perspective, such as pilot studies and randomized controlled trials on AI and its applications in a hospital setting. | ['Michel Makhlouf', 'Alaa Abd-Alrazaq', 'Mowafa Househ', 'Mohammed Anbar', 'Uzair Shah', 'Khaled A Althelaya', 'Khalid Alyafei', 'Marco Agus', 'Mahmood Alzubaidi'] | 2022-01-17 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [ 3.53961438e-01 7.59309947e-01 -8.67825627e-01 -1.30787060e-01
-2.07116812e-01 -5.77919543e-01 1.10084489e-01 6.47539318e-01
-3.13112408e-01 5.18347681e-01 6.39582798e-02 -8.70126188e-01
-4.10481900e-01 -7.81754434e-01 -7.00808585e-01 -4.17250037e-01
-2.67356694e-01 6.54560566e-01 -1.95810467e-01 3.93274099e-01
5.04165053e-01 7.08046257e-01 -1.25692403e+00 1.68716684e-01
9.75768983e-01 8.20046604e-01 -2.49729641e-02 6.07905567e-01
-5.39188623e-01 5.70447445e-01 -5.37562311e-01 -4.05646414e-01
3.78722958e-02 -9.34201717e-01 -6.67006612e-01 -5.50897241e-01
6.47063926e-02 -4.63078260e-01 8.62301663e-02 6.03384972e-01
8.90092432e-01 -2.94070482e-01 9.32641029e-01 -6.04800999e-01
-5.01373887e-01 5.87006748e-01 -5.99304020e-01 5.52343965e-01
4.02826697e-01 -2.18272265e-02 -1.01157740e-01 -6.41988695e-01
6.50281012e-01 1.09093106e+00 7.14630067e-01 7.51711428e-01
-6.50258482e-01 -1.17527246e+00 -4.30904329e-01 -3.29982489e-02
-8.38032365e-01 -3.58611047e-01 5.50282300e-01 -6.32277727e-01
8.01698327e-01 2.67333329e-01 1.20053041e+00 6.47279739e-01
4.29490596e-01 3.69144231e-01 7.77608037e-01 -7.94939995e-01
6.65198565e-02 2.04262033e-01 -1.84379026e-01 7.11517036e-01
8.13496649e-01 2.65966654e-01 2.42186096e-02 -2.01080665e-01
9.96508062e-01 -7.62810409e-02 2.90394574e-03 -8.02845508e-02
-6.37639642e-01 9.09956336e-01 1.13764130e-01 6.83843672e-01
-5.79149365e-01 -3.47704440e-01 6.31348729e-01 2.23793602e-03
3.63463342e-01 7.78321922e-01 -3.51854056e-01 -3.13102394e-01
-8.84921849e-01 -1.20115630e-01 1.01129854e+00 6.78583801e-01
3.41993362e-01 1.69020191e-01 1.54201686e-02 7.89974630e-01
6.04960322e-01 5.38305461e-01 5.69097281e-01 -1.20501387e+00
1.88592181e-01 5.20155787e-01 -1.94216713e-01 -9.96474266e-01
-9.29299593e-01 -2.59118408e-01 -1.01284289e+00 -2.03809932e-01
-1.79143660e-02 -8.69185209e-01 -1.00615406e+00 1.34175742e+00
3.50208312e-01 2.63127327e-01 6.40980676e-02 5.22227347e-01
1.68372238e+00 2.75906384e-01 3.13376486e-01 -7.09899843e-01
1.35229182e+00 -6.57206714e-01 -1.02843916e+00 -4.08550240e-02
7.22887695e-01 -4.40682739e-01 1.41708003e-02 4.42915589e-01
-1.67352343e+00 -2.24392220e-01 -8.54609966e-01 3.76490325e-01
-4.75946724e-01 -1.39956653e-01 7.01489985e-01 1.20945835e+00
-9.04157579e-01 3.53440106e-01 -1.24451602e+00 -7.50826955e-01
6.53770924e-01 3.69100869e-01 -1.50740802e-01 1.52492002e-01
-1.17559028e+00 1.66292882e+00 2.13918582e-01 -2.35249996e-01
-3.93477082e-01 -1.04466844e+00 -1.01912248e+00 -8.98379683e-02
2.76997298e-01 -7.99181819e-01 1.11225438e+00 -9.85698700e-01
-1.34912944e+00 1.13157594e+00 -1.58167988e-01 -4.74987119e-01
1.33674279e-01 3.13492306e-02 -2.36435249e-01 7.62661159e-01
-3.51434462e-02 5.83848834e-01 9.64662954e-02 -1.13232648e+00
-9.61919963e-01 -5.47724783e-01 -2.47011051e-01 1.16205342e-01
-3.43467921e-01 4.51948732e-01 -3.71318609e-01 -5.83611608e-01
3.72062266e-01 -7.97084987e-01 -2.70266622e-01 -9.97254476e-02
3.55784953e-01 -2.95972526e-01 3.32025170e-01 -9.09133852e-01
1.47955859e+00 -1.68059516e+00 -3.58164757e-01 2.83146024e-01
2.25174427e-01 8.22507381e-01 3.33291531e-01 3.43233496e-01
-2.89302677e-01 5.90203643e-01 -1.08804278e-01 2.79878080e-01
-7.12733865e-01 2.21196249e-01 7.94652224e-01 5.55131137e-01
6.80705905e-02 1.16844749e+00 -9.85328972e-01 -9.21115875e-01
6.74172103e-01 3.66300792e-01 -4.77436632e-01 1.23785138e-01
3.93801898e-01 4.37714666e-01 -5.39174020e-01 7.46083438e-01
6.28599524e-01 -1.14335574e-01 2.08281055e-01 6.69873729e-02
-2.45817363e-01 1.63466498e-01 -8.20287108e-01 1.50477815e+00
-4.90759552e-01 6.06755018e-01 1.51196733e-01 -1.23064911e+00
9.23861861e-01 8.93809617e-01 8.10657024e-01 -6.86419606e-01
3.93420160e-01 2.25271851e-01 4.69660610e-01 -1.23144221e+00
-4.37380463e-01 -3.72005582e-01 6.82673454e-01 3.78572166e-01
5.24090081e-02 -2.96322793e-01 1.99995637e-01 -2.55621165e-01
9.22755003e-01 2.99157165e-02 7.09640622e-01 -1.10842533e-01
2.78927296e-01 -2.37199605e-01 7.06294239e-01 1.15155232e+00
-1.82453275e-01 7.14577019e-01 5.09011745e-01 -5.82785606e-01
-5.67124903e-01 -5.96880138e-01 -5.61489284e-01 4.05653179e-01
-1.37717754e-01 4.12602067e-01 -7.44022071e-01 -3.97307962e-01
-2.55627781e-02 6.27330840e-01 -9.45127845e-01 -2.81789392e-01
-5.17983317e-01 -6.90083444e-01 7.94981360e-01 5.62685192e-01
3.52945209e-01 -1.45974481e+00 -1.11429477e+00 3.93880457e-01
1.53487613e-02 -6.68972194e-01 2.40405023e-01 -3.93264294e-02
-1.06426942e+00 -1.00896740e+00 -1.15569365e+00 -9.14965093e-01
5.93747616e-01 -4.51459676e-01 9.72063243e-01 3.27519298e-01
-5.04164517e-01 4.03937995e-01 -3.15434903e-01 -1.06202734e+00
-6.86635256e-01 -1.59424633e-01 -7.51085505e-02 -6.29139304e-01
3.81868154e-01 -4.20882285e-01 -9.72929358e-01 -1.90863580e-01
-7.93271303e-01 -1.51442870e-01 1.10298169e+00 6.00483000e-01
3.07057258e-02 -6.28636122e-01 8.22940230e-01 -1.10672629e+00
3.57162297e-01 -9.85696614e-01 -2.71657288e-01 1.70366883e-01
-9.22772765e-01 -8.25066686e-01 -5.97743951e-02 -5.67534924e-01
-9.66810882e-01 -6.29352570e-01 -1.25076815e-01 -4.22440797e-01
-6.14954233e-01 9.74317908e-01 5.18275142e-01 -1.29588097e-01
7.09969699e-01 4.35429066e-02 6.47617221e-01 -3.32551330e-01
-5.70139229e-01 8.54450822e-01 1.39118433e-01 -3.11311692e-01
-2.17096098e-02 1.78256873e-02 2.22837240e-01 -8.50883245e-01
-1.45563751e-01 -2.43481532e-01 -2.27608338e-01 -2.11046115e-01
9.12894130e-01 -5.51601171e-01 -8.29790831e-01 -1.59921944e-01
-9.10111070e-01 -3.86134796e-02 4.78557646e-02 1.17586803e+00
-8.07481110e-02 1.72764450e-01 -7.57798195e-01 -9.79608655e-01
-8.88409734e-01 -1.13128543e+00 5.35859227e-01 7.70673871e-01
-3.98812473e-01 -9.76280093e-01 1.63889080e-01 3.10020924e-01
6.50451720e-01 5.45503914e-01 1.18042386e+00 -1.00321329e+00
3.49160910e-01 -3.79145235e-01 -2.57478893e-01 3.09133474e-02
6.32137299e-01 2.28591591e-01 -4.40122664e-01 6.64497912e-02
-8.46528821e-03 2.29468629e-01 1.40953720e-01 1.17335141e+00
1.08670628e+00 -3.00678223e-01 -6.62745237e-01 4.42263424e-01
1.08113194e+00 1.46878254e+00 5.95534980e-01 1.64681688e-01
1.54213130e-01 7.91214466e-01 6.03478789e-01 3.52834672e-01
3.86450619e-01 1.03032636e-02 4.57570732e-01 -4.00859177e-01
5.31798787e-02 4.14178252e-01 -5.45226991e-01 2.73517549e-01
-6.40359461e-01 -1.36809975e-01 -1.34751368e+00 6.80815518e-01
-1.31792331e+00 -7.63325632e-01 1.17758967e-01 1.95512593e+00
5.53574204e-01 1.25706226e-01 6.44146577e-02 -1.68828174e-01
5.83825767e-01 -5.38991809e-01 -3.50699872e-01 -9.38093543e-01
4.84924704e-01 5.93464494e-01 2.23082557e-01 -9.64732654e-03
-8.65949452e-01 2.76406944e-01 6.47021341e+00 2.18112975e-01
-1.09741592e+00 -4.15445529e-02 9.46870029e-01 7.12056309e-02
-6.89907148e-02 -2.06366301e-01 -1.41297534e-01 6.69431150e-01
1.05189586e+00 -6.76656067e-02 -1.58858150e-01 5.76803744e-01
3.24432999e-01 -4.64855731e-01 -7.33154297e-01 6.45044684e-01
-6.10562973e-03 -1.67608118e+00 -2.80431032e-01 -2.22099379e-01
6.67550743e-01 -2.08137378e-01 -1.75727203e-01 1.81992158e-01
-4.58442718e-01 -1.15850472e+00 -1.37561053e-01 6.14585221e-01
1.08270705e+00 -5.23150802e-01 1.12419152e+00 2.32931271e-01
-4.63732898e-01 6.25073537e-02 2.53742207e-02 -9.88818333e-02
-1.65560514e-01 2.77751178e-01 -1.12175465e+00 7.60199487e-01
1.10534739e+00 6.24870539e-01 -2.70892084e-02 1.16214120e+00
2.34645039e-01 1.21627390e+00 -4.55239564e-01 -4.49157953e-01
1.86149210e-01 -2.84364223e-01 3.94051164e-01 1.42243314e+00
1.77168399e-01 7.23635674e-01 -6.13928795e-01 5.24496913e-01
1.26749873e-01 2.42074236e-01 -7.47500539e-01 -1.20029055e-01
3.45094055e-01 1.19569540e+00 -1.01445913e+00 -2.69196510e-01
-9.52821374e-01 -3.11861355e-02 -5.57791054e-01 4.41044688e-01
-4.61314082e-01 -4.63316441e-01 -5.90315498e-02 3.81799102e-01
-1.01781376e-01 7.87500262e-01 -5.77061653e-01 -1.91035345e-01
-3.98207337e-01 -7.70330608e-01 6.97607696e-01 -6.29315734e-01
-4.67059851e-01 1.20008647e-01 4.38156307e-01 -1.04604042e+00
-3.01924586e-01 -3.69865268e-01 -2.64838189e-01 9.51704443e-01
-1.04141319e+00 -7.65652835e-01 -4.89632711e-02 -1.74303412e-01
2.93587774e-01 -3.55187029e-01 1.08369505e+00 4.89563525e-01
-3.38947564e-01 5.52556336e-01 -1.37843966e-01 3.77318293e-01
3.68563175e-01 -9.95306134e-01 -3.56246382e-01 1.05700195e-01
-1.03445065e+00 8.25105309e-01 2.73625404e-01 -1.00165510e+00
-1.22758269e+00 -5.02045453e-01 8.82347405e-01 4.03211899e-02
-7.04558939e-02 6.61206007e-01 -7.57635593e-01 6.81442320e-01
4.21849638e-01 3.78902722e-03 9.84865904e-01 -2.98802793e-01
7.40914524e-01 5.61012141e-02 -1.72270715e+00 5.26240945e-01
5.87637901e-01 5.89229763e-01 -6.43346012e-01 4.16442975e-02
3.50787044e-01 -1.06484747e+00 -1.34197581e+00 1.31792641e+00
1.08998716e+00 -6.32883608e-01 8.77523363e-01 -6.50655389e-01
9.40946400e-01 4.62463111e-01 9.14916813e-01 -8.09578061e-01
-2.47910336e-01 -6.44902766e-01 -2.66198754e-01 9.57750738e-01
5.48186421e-01 -9.89751756e-01 1.00678313e+00 7.81604707e-01
-2.76906967e-01 -1.18748820e+00 -8.65682065e-01 2.25819126e-01
3.05097520e-01 -2.56242659e-02 5.62216938e-01 1.14398992e+00
1.83820844e-01 -1.92247599e-01 1.95276275e-01 -4.53797393e-02
1.15713440e-01 -7.99772963e-02 1.09514676e-01 -9.41430211e-01
3.00967008e-01 -7.54667699e-01 -4.92740124e-01 -1.75309300e-01
-4.10861313e-01 -6.08953416e-01 -4.25043672e-01 -2.20337057e+00
2.38626182e-01 -4.03596044e-01 -2.05340132e-01 5.13461113e-01
-1.86109707e-01 -2.24113017e-01 -2.35535815e-01 -3.91636074e-01
1.03845581e-01 -2.19843835e-01 1.59787261e+00 1.52599126e-01
-4.53174502e-01 6.32274806e-01 -9.83776093e-01 8.87242556e-01
8.45785201e-01 -5.53201795e-01 -2.20253587e-01 -1.39914036e-01
1.62258312e-01 8.60727072e-01 -4.58446652e-01 -5.28444469e-01
2.78892547e-01 -3.05430055e-01 7.32685983e-01 -6.73601687e-01
-1.61341637e-01 -1.02624440e+00 1.73298270e-01 9.79831278e-01
-2.23756999e-01 -4.52619940e-02 6.66899025e-01 -3.67214799e-01
-4.62438067e-04 -6.48013651e-01 7.30451286e-01 -2.38242730e-01
-1.80434401e-03 2.34986275e-01 -9.12476659e-01 2.76242495e-01
1.11961460e+00 -4.48526442e-01 1.11906663e-01 -6.24944806e-01
-8.60830486e-01 4.65140015e-01 -1.72670230e-01 4.07051533e-01
8.03704798e-01 -7.50727057e-01 -8.54008913e-01 3.91164236e-02
-2.73300201e-01 3.92905176e-01 1.99257344e-01 1.33190596e+00
-1.12289572e+00 5.81576109e-01 -1.73188105e-01 -4.30508584e-01
-1.15470231e+00 4.71474439e-01 3.58582705e-01 -3.48309427e-03
-5.27320921e-01 7.37194538e-01 -1.84769183e-01 -6.11453295e-01
7.18751848e-01 -3.05313289e-01 -7.88743317e-01 3.19003046e-01
4.33869243e-01 7.45055258e-01 1.25096709e-01 -2.57537633e-01
-6.88865125e-01 8.61907244e-01 2.27805942e-01 2.81201787e-02
1.36195767e+00 -2.00963113e-02 -3.35301399e-01 8.66493136e-02
1.11829925e+00 -6.58471346e-01 -1.40592903e-01 1.48663178e-01
-1.54501289e-01 2.33966056e-02 -3.41016263e-01 -8.32230270e-01
-1.25339293e+00 7.15391695e-01 8.28400016e-01 2.36362308e-01
1.37052488e+00 -1.29505724e-01 6.70252964e-02 6.88814325e-03
8.22350308e-02 -9.15886819e-01 -2.79309392e-01 2.66774036e-02
7.34564841e-01 -1.32173526e+00 8.73558670e-02 -9.91143361e-02
-7.14688778e-01 1.22344077e+00 5.57989180e-01 9.54681039e-02
1.16083562e+00 4.07214731e-01 2.88641751e-01 -1.81154221e-01
-2.78849363e-01 2.41114482e-01 3.50980759e-01 4.76506412e-01
8.38563144e-01 -2.99128354e-01 -9.98077035e-01 9.45888698e-01
1.51466370e-01 3.94378275e-01 2.45840997e-01 1.35318506e+00
-1.17279924e-01 -5.39608717e-01 -4.58582193e-01 1.24280572e+00
-1.23266125e+00 6.12478994e-04 -8.62121210e-02 8.19386959e-01
3.90705377e-01 9.81420696e-01 2.84674671e-02 3.99277300e-01
1.07151486e-01 -6.00254536e-02 5.12333930e-01 -8.12978625e-01
-8.90931547e-01 2.29791179e-01 8.81526470e-02 -2.85729766e-01
-6.89057946e-01 -8.49904776e-01 -1.45876360e+00 2.44368032e-01
-3.71805876e-01 2.66854614e-01 1.00512052e+00 9.90413487e-01
1.97664768e-01 6.53354287e-01 3.53421956e-01 -5.52260101e-01
3.72532487e-01 -1.18535411e+00 -2.67879218e-01 -3.37149799e-01
3.40925306e-01 -5.97501636e-01 -3.00520748e-01 -1.42894387e-01] | [14.05460262298584, -2.2413952350616455] |
9fa77937-c9a1-4938-9d0f-edcf0b0ec771 | discovery-of-2d-materials-using-transformer | 2301.05824 | null | https://arxiv.org/abs/2301.05824v1 | https://arxiv.org/pdf/2301.05824v1.pdf | Discovery of 2D materials using Transformer Network based Generative Design | Two-dimensional (2D) materials have wide applications in superconductors, quantum, and topological materials. However, their rational design is not well established, and currently less than 6,000 experimentally synthesized 2D materials have been reported. Recently, deep learning, data-mining, and density functional theory (DFT)-based high-throughput calculations are widely performed to discover potential new materials for diverse applications. Here we propose a generative material design pipeline, namely material transformer generator(MTG), for large-scale discovery of hypothetical 2D materials. We train two 2D materials composition generators using self-learning neural language models based on Transformers with and without transfer learning. The models are then used to generate a large number of candidate 2D compositions, which are fed to known 2D materials templates for crystal structure prediction. Next, we performed DFT computations to study their thermodynamic stability based on energy-above-hull and formation energy. We report four new DFT-verified stable 2D materials with zero e-above-hull energies, including NiCl$_4$, IrSBr, CuBr$_3$, and CoBrCl. Our work thus demonstrates the potential of our MTG generative materials design pipeline in the discovery of novel 2D materials and other functional materials. | ['Jianjun Hu', 'Edirisuriya M. D. Siriwardane', 'Yuqi Song', 'Rongzhi Dong'] | 2023-01-14 | null | null | null | null | ['formation-energy', 'self-learning'] | ['miscellaneous', 'natural-language-processing'] | [-1.35053933e-01 -1.54717907e-01 -2.05414444e-02 -2.05436647e-01
-8.42230976e-01 -3.38732332e-01 6.45085037e-01 7.47372508e-02
2.07351536e-01 1.09027863e+00 1.49193734e-01 -4.87395704e-01
4.79607694e-02 -1.09499192e+00 -8.09605539e-01 -1.07110214e+00
-2.51420975e-01 8.33987832e-01 2.68572718e-01 -4.42317247e-01
4.80372638e-01 5.04393041e-01 -1.77567446e+00 2.20356762e-01
1.15160632e+00 1.18425071e+00 5.46271026e-01 -8.11935067e-02
-8.72974917e-02 2.54257977e-01 -3.91264796e-01 -8.17865357e-02
2.30942592e-01 -3.81258696e-01 -5.34701943e-01 -4.16300684e-01
6.77660778e-02 -1.92219287e-01 -3.69164526e-01 6.57155216e-01
7.38249779e-01 -3.97551060e-02 1.23068547e+00 -5.90845644e-01
-1.18200612e+00 8.33523393e-01 -2.90183425e-01 -2.41713166e-01
3.71667445e-01 5.60500383e-01 9.97249544e-01 -1.10959184e+00
6.67898953e-01 8.06736290e-01 3.81548554e-01 9.48932886e-01
-1.15153193e+00 -9.46008861e-01 -4.42196816e-01 1.05190165e-01
-1.30443370e+00 -2.87541240e-01 1.18707395e+00 -6.03255749e-01
1.55370879e+00 -6.42592311e-02 9.44840968e-01 9.88105357e-01
3.16718221e-01 4.59324867e-01 1.18637156e+00 -4.24007475e-01
3.43973309e-01 -1.90221474e-01 -2.32757613e-01 7.36479640e-01
5.77367187e-01 1.97984800e-01 -8.51984441e-01 -1.27850935e-01
4.37698096e-01 -3.11762899e-01 1.62777796e-01 -2.14402840e-01
-7.88758516e-01 4.37242836e-01 6.70051157e-01 4.23459858e-01
-4.02987778e-01 2.00523466e-01 2.09286347e-01 -4.22080047e-02
4.96011615e-01 9.73288715e-01 -4.06809479e-01 -8.47025141e-02
-5.94093263e-01 6.83245420e-01 3.09182227e-01 8.74822438e-01
8.25539470e-01 2.40855426e-01 1.41875386e-01 6.29342377e-01
2.12837428e-01 4.76513624e-01 3.28180134e-01 -2.85333902e-01
1.61799401e-01 7.87813306e-01 -2.55115088e-02 -3.18815321e-01
-3.30102563e-01 -1.22188948e-01 -8.91384065e-01 -2.33311877e-01
-3.68979603e-01 2.12682575e-01 -8.63606989e-01 1.43841100e+00
3.89578938e-01 -3.00497532e-01 -4.56556343e-02 8.79223347e-01
1.14864969e+00 1.00939178e+00 -1.17173597e-01 -3.21010262e-01
7.66926348e-01 -3.76864612e-01 5.77371046e-02 3.63414198e-01
4.52725410e-01 -3.67560416e-01 1.05302489e+00 1.88035890e-01
-1.42831933e+00 -4.56471294e-01 -1.05593681e+00 -1.41897365e-01
-3.83124679e-01 -3.96361910e-02 9.95836616e-01 4.85340446e-01
-7.93785989e-01 9.08992708e-01 -7.26110756e-01 1.18270509e-01
4.72126096e-01 6.47250652e-01 -1.59771889e-01 1.66546702e-01
-1.15838504e+00 6.37201428e-01 5.83158493e-01 -1.96924940e-01
-1.26626766e+00 -9.23032284e-01 -4.16162521e-01 -2.85344809e-01
1.26471117e-01 -8.19114506e-01 9.86217856e-01 1.43342227e-01
-1.67118061e+00 9.94171143e-01 3.45344692e-02 -2.73472995e-01
-1.07688017e-01 2.02651083e-01 -4.49221164e-01 -9.09058750e-02
2.66282439e-01 5.46527922e-01 5.81115007e-01 -1.15955412e+00
4.73317597e-03 -1.56257376e-01 -4.05188084e-01 -1.02751017e-01
-2.49129951e-01 -2.98848510e-01 3.72546434e-01 -5.20339966e-01
3.39812547e-01 -6.62451863e-01 -1.64661735e-01 -5.91975451e-01
-7.32270479e-01 -6.83124602e-01 3.30222636e-01 -2.74439871e-01
9.68109131e-01 -1.70380068e+00 3.10337335e-01 4.13548678e-01
4.58563030e-01 5.11746369e-02 2.03402326e-01 6.93041086e-01
-1.21411964e-01 2.96151787e-01 -2.36280695e-01 3.41584176e-01
-4.49479111e-02 -6.09308779e-01 -3.49198803e-02 1.54143021e-01
3.14388752e-01 1.20362556e+00 -8.26845109e-01 -2.12305933e-02
1.38646439e-01 9.25252363e-02 -7.53124237e-01 2.63295710e-01
-9.36505616e-01 5.14470935e-01 -6.44755661e-01 8.73939335e-01
6.64267182e-01 -3.45752805e-01 1.44713163e-01 -4.03398514e-01
-6.09211147e-01 1.07850313e+00 -3.28477949e-01 1.59220970e+00
-1.63599804e-01 5.10292314e-02 -4.24429476e-01 -9.96507883e-01
1.44601822e+00 -9.28225815e-02 9.16114628e-01 -1.20605791e+00
-7.74986818e-02 7.14218438e-01 3.03902686e-01 -3.84591997e-01
5.84150791e-01 -6.37366712e-01 -2.81119704e-01 6.40571535e-01
5.14589474e-02 -9.25633252e-01 2.46309385e-01 2.13562891e-01
1.04614639e+00 1.76115721e-01 -1.33319616e-01 -5.94388723e-01
3.13081563e-01 3.74336913e-02 2.09729776e-01 5.51182151e-01
4.44683015e-01 2.77764320e-01 2.12282106e-01 -5.56111038e-01
-1.73616421e+00 -1.22686100e+00 -1.85625836e-01 4.95198667e-01
1.16244130e-01 -7.24224567e-01 -5.01051903e-01 -1.07686356e-01
1.24534696e-01 6.31101131e-01 -2.58906215e-01 -5.86335599e-01
-5.06466627e-01 -9.93792415e-01 1.47374168e-01 1.95548311e-01
4.60677445e-01 -9.39894021e-01 -3.27280648e-02 3.74119073e-01
2.57739246e-01 -4.28424150e-01 -1.45682722e-01 3.10195476e-01
-6.95677161e-01 -8.09962034e-01 -4.31585133e-01 -8.13202620e-01
7.47905433e-01 1.33122995e-01 9.81615007e-01 2.64204405e-02
-7.06965327e-01 -1.86118871e-01 -5.30368835e-02 -1.85107335e-01
-8.68588090e-01 3.25274736e-01 7.61364579e-01 -6.52016699e-01
6.64980114e-02 -9.20191586e-01 -6.40433788e-01 6.91525489e-02
-4.57426250e-01 5.35310268e-01 6.03248060e-01 7.58322299e-01
8.71950507e-01 3.18419039e-01 8.50555837e-01 -5.20016313e-01
6.60462141e-01 -5.47931433e-01 -6.63178742e-01 2.53616333e-01
-6.15917683e-01 5.09571135e-01 8.50120366e-01 -1.73938185e-01
-1.00183845e+00 -2.35717744e-01 -5.02771676e-01 -1.60284773e-01
-2.52895225e-02 3.74293953e-01 -4.18755233e-01 1.11997105e-01
6.98980510e-01 4.70623672e-01 -3.11606467e-01 -5.59701264e-01
1.89115331e-01 7.15758860e-01 2.54316509e-01 -1.32169545e+00
8.33486915e-01 -2.10811183e-01 2.98525155e-01 -9.58380878e-01
-6.38044059e-01 2.89032876e-01 -4.34957355e-01 -4.40999568e-02
7.54256845e-01 -8.31371784e-01 -1.06678319e+00 6.74969554e-01
-8.35477829e-01 -3.59571993e-01 -2.93750733e-01 1.60264015e-01
-5.18719554e-01 -2.85499860e-02 -5.87315023e-01 -6.71106756e-01
-8.71693313e-01 -1.45231807e+00 1.08102024e+00 9.60808694e-02
-1.00303955e-01 -4.75276202e-01 -9.43321958e-02 3.90945554e-01
5.28707206e-01 1.73515052e-01 1.84571004e+00 -3.44665289e-01
-9.94639933e-01 1.60791278e-01 -1.61584187e-02 -1.10242506e-02
5.28951168e-01 -1.03199162e-01 -4.30078208e-01 -7.54283071e-02
-3.22068572e-01 -4.31558967e-01 9.35676694e-01 3.56306016e-01
1.50653183e+00 -1.56659231e-01 -3.18924248e-01 2.92591840e-01
1.13977993e+00 4.42299008e-01 5.44647694e-01 9.35297459e-03
8.72306824e-01 1.67276651e-01 7.55679533e-02 4.56415743e-01
2.50289768e-01 4.94047165e-01 2.27434203e-01 2.91702986e-01
-6.81950375e-02 -6.17543042e-01 3.03334266e-01 1.28121912e+00
-4.07324135e-01 -1.82087585e-01 -1.05117631e+00 8.37012902e-02
-1.06229842e+00 -9.60780442e-01 -2.21888013e-02 2.15826988e+00
1.23268092e+00 6.07304871e-01 2.46543437e-01 -1.29878726e-02
5.75200200e-01 -1.15381002e-01 -1.24944663e+00 -8.39640424e-02
-2.66599655e-01 8.23792040e-01 1.64891288e-01 -1.53799132e-01
-5.59782505e-01 1.08387220e+00 6.48923111e+00 1.04172421e+00
-1.44391131e+00 -1.92370459e-01 7.70359337e-01 -2.01357797e-01
-9.65246499e-01 1.62198037e-01 -8.67277741e-01 7.57843137e-01
7.06323206e-01 -3.76442879e-01 5.37094295e-01 6.55707538e-01
9.04920995e-02 9.39875394e-02 -9.98215675e-01 1.04306197e+00
-4.46216911e-01 -2.27184081e+00 2.49425635e-01 2.04945028e-01
9.94406641e-01 8.17071050e-02 6.24353662e-02 5.87401865e-03
1.90164372e-01 -7.63726056e-01 7.78572977e-01 5.39082825e-01
1.29104638e+00 -6.95149481e-01 -1.62116915e-01 7.80733526e-02
-1.04080379e+00 1.43744498e-01 -3.15174371e-01 1.51206434e-01
4.40282598e-02 9.50905442e-01 -1.07765687e+00 5.26698291e-01
5.93735218e-01 8.39605808e-01 -1.13731161e-01 5.98844886e-01
-3.90849076e-02 5.63013434e-01 -2.77742386e-01 -6.83240950e-01
2.38322429e-02 -4.41925108e-01 3.89267892e-01 3.14786911e-01
5.65198779e-01 2.41494000e-01 9.86011103e-02 1.58396745e+00
-6.26113653e-01 -1.55850634e-01 -5.18490136e-01 -4.19058323e-01
6.07485652e-01 8.31718683e-01 -7.26553082e-01 -2.98861414e-01
5.60232103e-02 3.53484988e-01 1.71335682e-01 -1.99695051e-01
-5.49203634e-01 -2.22666100e-01 6.12018406e-01 7.85063624e-01
1.70484528e-01 -6.02763712e-01 1.79797877e-02 -1.09803283e+00
-1.50791816e-02 -6.18010163e-01 -2.54593045e-01 -8.29893351e-01
-1.69359124e+00 2.35429421e-01 1.37828872e-01 -8.66532445e-01
7.59302154e-02 -7.01082408e-01 -7.52094924e-01 7.06741691e-01
-8.90870631e-01 -9.34041739e-01 1.19382992e-01 -6.41555712e-02
3.01409304e-01 -4.23955262e-01 5.70277214e-01 1.95901349e-01
-6.02864265e-01 2.94359952e-01 4.83592153e-01 -5.39331496e-01
2.96372056e-01 -1.08015668e+00 8.32008421e-01 1.45435303e-01
-3.89007747e-01 5.22056758e-01 6.91397369e-01 -1.06966782e+00
-2.16040850e+00 -8.60390961e-01 4.45669562e-01 -1.17754415e-01
5.99173307e-01 -8.54287565e-01 -7.63305068e-01 -1.71588168e-01
-2.20535547e-01 -5.42375088e-01 6.52521610e-01 -1.24133535e-01
-1.39970740e-03 -1.29685506e-01 -1.25960064e+00 5.08709073e-01
1.52959943e+00 -5.28050601e-01 -1.91870838e-01 7.02174723e-01
1.03029335e+00 -3.12665790e-01 -1.01488459e+00 5.96475601e-01
4.05983865e-01 -8.37000668e-01 1.05403996e+00 -5.60272932e-01
8.37726653e-01 -1.71992201e-02 -2.06355959e-01 -1.08146870e+00
-3.93618792e-01 -6.96974337e-01 -1.10201895e-01 1.12037301e+00
5.68692684e-01 -3.57169479e-01 8.20300639e-01 5.88809967e-01
-7.89638042e-01 -1.02987647e+00 -1.06373572e+00 -7.47948945e-01
4.69250292e-01 -2.43257537e-01 1.26876068e+00 5.59148490e-01
-9.99333039e-02 4.52736080e-01 -2.32742429e-01 -3.86829197e-01
5.46272278e-01 6.13212883e-01 4.26693469e-01 -1.29471934e+00
-3.29077333e-01 -6.02846503e-01 -9.53967944e-02 -1.05445421e+00
4.09995377e-01 -1.35284507e+00 -1.31256655e-01 -1.49158800e+00
2.38507688e-01 -1.08372438e+00 -8.23915079e-02 3.57493103e-01
3.73666555e-01 -1.36981338e-01 -3.80412698e-01 3.01632077e-01
-3.72628748e-01 1.19650209e+00 1.33272886e+00 -3.06188434e-01
-3.62404287e-01 -1.08127825e-01 -7.43065417e-01 9.64048058e-02
7.83821762e-01 -2.19530448e-01 6.54089153e-02 -1.17617629e-01
6.77227795e-01 -1.68299824e-01 -2.36728955e-02 -9.40750659e-01
-3.55322361e-01 -3.04457247e-01 3.62514853e-01 -1.04126930e+00
2.87750751e-01 -1.92854777e-01 3.75702679e-01 5.05816936e-01
-1.00649036e-01 -3.28182459e-01 -6.48864433e-02 -5.71602397e-02
3.21491390e-01 -2.64872342e-01 6.20694637e-01 -5.02300084e-01
-3.59643668e-01 7.15078771e-01 -4.15480375e-01 -2.65672028e-01
9.18256164e-01 -6.58066273e-02 -2.75446087e-01 2.05285385e-01
-5.44858515e-01 -1.25321686e-01 6.79857552e-01 2.60265917e-01
7.94755816e-01 -1.39561236e+00 -4.30984080e-01 4.35298949e-01
8.00777301e-02 5.32384813e-01 3.65633905e-01 1.79282472e-01
-6.95398510e-01 4.60977644e-01 -2.26515278e-01 -5.32965660e-01
-4.59699601e-01 5.63976109e-01 3.42717737e-01 -5.39732985e-02
-3.49937886e-01 8.07827592e-01 -1.92933846e-02 -4.24962908e-01
-4.67466533e-01 -3.40302765e-01 5.72941303e-01 -3.59929502e-01
-1.08176596e-01 2.14326501e-01 3.28165501e-01 -2.77400404e-01
-3.14720035e-01 3.83856475e-01 -2.21711233e-01 2.90964395e-01
2.03826571e+00 4.25639004e-01 -4.20648068e-01 2.58002460e-01
1.09566104e+00 -1.59326717e-01 -8.54827404e-01 -1.38950437e-01
-3.53393368e-02 -6.50259629e-02 -3.06220084e-01 -4.36776161e-01
-8.27524960e-01 7.50529230e-01 2.87221551e-01 7.67476857e-02
9.87931013e-01 6.92361951e-01 1.05773878e+00 6.39657259e-01
6.41018391e-01 -1.03862393e+00 3.62771451e-01 4.05272633e-01
7.62993753e-01 -9.19898212e-01 2.14071944e-01 -1.62688959e-02
-1.69305131e-02 1.05090272e+00 9.33224320e-01 7.85557553e-02
7.37260699e-01 1.66664079e-01 -9.81001198e-01 -4.98711765e-01
-9.27466571e-01 8.22189823e-02 1.67859510e-01 4.41315711e-01
6.00185215e-01 2.50910610e-01 -1.66873068e-01 5.70345640e-01
-6.27589285e-01 -4.38799948e-01 1.71971112e-01 9.58909273e-01
-5.98708391e-01 -1.54265380e+00 9.03760642e-02 9.01278496e-01
2.61737388e-02 -4.06749487e-01 -5.30769944e-01 1.77429110e-01
3.69475007e-01 2.30901554e-01 1.11091644e-01 -5.58490753e-01
5.58646806e-02 -9.24428459e-03 9.74494755e-01 -6.48799717e-01
-3.07945162e-01 -1.45383134e-01 -2.14253832e-02 7.45167136e-02
-2.47540787e-01 -6.25237048e-01 -1.54223824e+00 -3.87931257e-01
-6.18079484e-01 3.59173864e-01 9.22666907e-01 8.36458385e-01
7.32410491e-01 1.35638088e-01 1.14088082e+00 -1.07441533e+00
-1.59799889e-01 -6.71072125e-01 -5.24065733e-01 1.65945396e-01
-1.62572727e-01 -8.96516442e-01 1.72943786e-01 -3.90536398e-01] | [5.1774492263793945, 5.371654033660889] |
1b2309cc-870e-40fb-adf3-f6895bb3040c | unbiased-scene-graph-generation-via-rich-and | 2002.00176 | null | https://arxiv.org/abs/2002.00176v1 | https://arxiv.org/pdf/2002.00176v1.pdf | Unbiased Scene Graph Generation via Rich and Fair Semantic Extraction | Extracting graph representation of visual scenes in image is a challenging task in computer vision. Although there has been encouraging progress of scene graph generation in the past decade, we surprisingly find that the performance of existing approaches is largely limited by the strong biases, which mainly stem from (1) unconsciously assuming relations with certain semantic properties such as symmetric and (2) imbalanced annotations over different relations. To alleviate the negative effects of these biases, we proposed a new and simple architecture named Rich and Fair semantic extraction network (RiFa for short), to not only capture rich semantic properties of the relations, but also fairly predict relations with different scale of annotations. Using pseudo-siamese networks, RiFa embeds the subject and object respectively to distinguish their semantic differences and meanwhile preserve their underlying semantic properties. Then, it further predicts subject-object relations based on both the visual and semantic features of entities under certain contextual area, and fairly ranks the relation predictions for those with a few annotations. Experiments on the popular Visual Genome dataset show that RiFa achieves state-of-the-art performance under several challenging settings of scene graph task. Especially, it performs significantly better on capturing different semantic properties of relations, and obtains the best overall per relation performance. | ['Xianglong Liu', 'Lei Huang', 'Jie Luo', 'Bin Wen'] | 2020-02-01 | null | null | null | null | ['unbiased-scene-graph-generation'] | ['computer-vision'] | [ 2.67557800e-01 2.98297256e-01 -4.33985323e-01 -4.69328940e-01
-1.41612411e-01 -3.47755790e-01 7.06549227e-01 1.43009856e-01
-9.98898596e-02 3.93249959e-01 4.54707026e-01 5.46932966e-03
-1.68635964e-01 -8.24836969e-01 -6.40334904e-01 -5.30066073e-01
1.07130744e-01 4.01148468e-01 4.82911944e-01 -3.17585170e-01
-1.02509903e-02 1.72860697e-01 -1.42376864e+00 4.07742351e-01
8.83880019e-01 1.19159615e+00 2.71393694e-02 1.13878764e-01
-2.66274869e-01 1.38594568e+00 -3.49768370e-01 -8.83167744e-01
1.51200905e-01 -5.66506863e-01 -1.00520492e+00 2.16346368e-01
5.14546096e-01 6.60614595e-02 -7.83297122e-01 1.50801075e+00
2.32746005e-01 -1.47352159e-01 7.35597074e-01 -1.55234826e+00
-9.80035067e-01 5.84953964e-01 -7.77616978e-01 2.57178128e-01
3.39412779e-01 2.26291139e-02 1.53383231e+00 -7.31876194e-01
8.23189378e-01 1.52877915e+00 3.95196885e-01 1.80750683e-01
-1.02217090e+00 -7.42459416e-01 4.11754221e-01 4.88809377e-01
-1.56331241e+00 -2.70820498e-01 8.86758685e-01 -3.99654716e-01
8.75115573e-01 2.66723871e-01 6.84583306e-01 1.05365086e+00
3.98367308e-02 8.05657923e-01 9.58127379e-01 -3.26422490e-02
-5.33136576e-02 -4.13090773e-02 2.82177925e-01 9.11656916e-01
3.67654800e-01 -5.48830211e-01 -4.91346896e-01 1.36416242e-01
5.69382012e-01 2.27902513e-02 -4.35503334e-01 -6.74696565e-01
-1.39747202e+00 5.86999953e-01 9.79912579e-01 1.39541343e-01
-1.56934410e-01 -1.32042766e-01 5.45902908e-01 4.65903878e-02
4.21567321e-01 3.20523292e-01 -3.11829090e-01 5.02200246e-01
-3.07458729e-01 2.05201268e-01 5.84182441e-01 1.34727776e+00
9.39354360e-01 -3.10470372e-01 -4.73644763e-01 9.13202643e-01
2.13974208e-01 2.97105730e-01 3.94451410e-01 -5.65081596e-01
8.04615438e-01 1.24564981e+00 -3.06017965e-01 -1.71885049e+00
-3.33637595e-01 -3.96124959e-01 -1.29003501e+00 -3.29106092e-01
2.08041266e-01 2.96784073e-01 -1.01361048e+00 1.81346667e+00
3.39332134e-01 1.11294001e-01 5.41181676e-02 9.75084424e-01
1.51918435e+00 3.69158149e-01 3.62365395e-01 -1.35048300e-01
1.60067046e+00 -1.24798071e+00 -8.88817966e-01 -5.24275005e-01
6.01051152e-01 -4.75422591e-01 1.05342710e+00 -1.74215242e-01
-7.54925549e-01 -5.63135862e-01 -1.05921197e+00 -3.71958017e-01
-5.31410038e-01 6.54014498e-02 1.02452278e+00 1.27105817e-01
-7.81899929e-01 2.90430397e-01 -3.41816247e-01 -6.44395590e-01
9.91121471e-01 6.55717552e-02 -5.68511307e-01 -2.73715705e-01
-1.24406970e+00 7.36188352e-01 5.95435739e-01 2.63870358e-01
-6.14522040e-01 -4.85368103e-01 -1.17056823e+00 1.09534293e-01
8.70238304e-01 -9.25197661e-01 5.74793220e-01 -9.66373265e-01
-7.52584934e-01 1.21107626e+00 -1.61782712e-01 -2.67591596e-01
5.32234907e-01 -1.42634124e-01 -3.78909886e-01 8.20690691e-02
4.22366649e-01 6.54132128e-01 6.56768501e-01 -1.26493371e+00
-3.75987917e-01 -5.25949836e-01 2.40782157e-01 4.94255066e-01
-2.95018822e-01 7.38369599e-02 -7.44417310e-01 -5.95989645e-01
4.77074623e-01 -8.58472586e-01 -9.68365148e-02 -3.17037851e-02
-8.69765282e-01 -5.68681419e-01 6.76964223e-01 -4.95422363e-01
1.11763191e+00 -1.99457395e+00 1.60219610e-01 -1.45717030e-02
6.68326020e-01 2.98810571e-01 -1.19133621e-01 2.92074680e-01
-3.18567246e-01 7.95726851e-02 -2.26147726e-01 -1.17400568e-02
-2.30660155e-01 3.29661608e-01 -4.62353081e-01 2.95731753e-01
2.80513495e-01 1.25995278e+00 -1.17121875e+00 -8.99695575e-01
3.25536821e-03 6.15443811e-02 -3.74243736e-01 3.20784867e-01
-1.41832054e-01 2.54920900e-01 -7.13705361e-01 7.03866124e-01
6.73357129e-01 -8.08794618e-01 3.92509282e-01 -8.05164516e-01
5.17285466e-01 3.14820498e-01 -1.12339222e+00 1.70604908e+00
4.08556014e-02 4.35672939e-01 -2.88123757e-01 -1.27506077e+00
8.89147103e-01 -6.20492622e-02 3.04622382e-01 -7.52179801e-01
2.20653430e-01 -1.31948963e-01 2.18155026e-01 -8.69517148e-01
2.38536641e-01 5.59884384e-02 5.13390340e-02 -6.52150437e-02
2.84130007e-01 -8.82784128e-02 1.59217864e-01 7.11465895e-01
1.02917135e+00 1.41524956e-01 4.95780230e-01 -2.89647669e-01
6.11907125e-01 2.20687445e-02 7.94994771e-01 5.01759410e-01
-4.81462777e-01 5.94538808e-01 8.72371197e-01 -4.65351731e-01
-7.40331709e-01 -9.94785666e-01 8.89199972e-02 9.81762528e-01
8.90921891e-01 -7.41414785e-01 -4.06345129e-01 -1.06014884e+00
-5.45723066e-02 2.98264891e-01 -6.63997114e-01 -3.05715889e-01
-4.15000051e-01 -8.91316414e-01 4.25692588e-01 5.78527093e-01
1.07123291e+00 -1.17492259e+00 1.35484979e-01 -2.80580729e-01
-4.21357423e-01 -1.73850453e+00 -2.10171923e-01 -1.31571919e-01
-3.98850411e-01 -1.41450667e+00 -3.36131275e-01 -1.03646839e+00
8.01494181e-01 6.60307467e-01 1.35774791e+00 1.92977309e-01
-1.59865469e-01 -2.87211817e-02 -3.90251398e-01 -3.36571634e-01
6.48732111e-02 -1.34154975e-01 -2.34472975e-01 2.85207242e-01
3.54601294e-01 -4.07917440e-01 -5.60311854e-01 2.66030252e-01
-6.67097151e-01 4.11436558e-01 5.19176424e-01 7.74865508e-01
5.67137539e-01 3.51612002e-01 3.48174870e-01 -1.31977499e+00
3.44334871e-01 -3.80890012e-01 -2.40613654e-01 4.99723077e-01
-6.41882956e-01 2.99990140e-02 5.76928139e-01 -1.83229342e-01
-1.10638547e+00 -1.50748298e-01 2.35030219e-01 -4.71495360e-01
-1.35787010e-01 2.68127799e-01 -6.25260353e-01 6.84138983e-02
4.65813488e-01 1.56797424e-01 -3.03114861e-01 -2.30886973e-02
4.98380661e-01 2.24472567e-01 5.26810527e-01 -4.61808980e-01
9.56352293e-01 6.71622455e-01 2.91071504e-01 -4.90455180e-01
-1.63058996e+00 -5.65752864e-01 -6.63566232e-01 -8.19349512e-02
1.15429342e+00 -1.21248472e+00 -6.06517136e-01 5.27637243e-01
-1.11019623e+00 3.78163275e-03 -4.89191264e-02 2.56072819e-01
-1.64518520e-01 4.54716712e-01 -5.42365134e-01 -3.87195498e-01
-1.33523196e-01 -1.05990589e+00 9.87180531e-01 4.00743514e-01
-1.99135691e-02 -8.11325490e-01 -2.16708958e-01 5.90950489e-01
-4.37046885e-02 2.80706078e-01 1.22520411e+00 -6.29916012e-01
-8.33376944e-01 9.41959992e-02 -1.02660823e+00 2.88630098e-01
2.51464427e-01 -8.00447240e-02 -1.00381196e+00 2.04462148e-02
-3.50646913e-01 -5.57410657e-01 1.07589960e+00 2.08520014e-02
1.26746619e+00 -2.99966097e-01 -5.57137489e-01 7.46333539e-01
1.34806287e+00 -1.91734672e-01 6.30056381e-01 1.94545403e-01
1.43886685e+00 8.89866829e-01 6.76271141e-01 -2.06441265e-02
9.14556921e-01 5.57308912e-01 6.86084628e-01 -2.84512490e-01
-4.91917014e-01 -6.61123633e-01 -3.91407460e-02 8.94436717e-01
-1.51867479e-01 -9.98656824e-02 -7.95312703e-01 6.42398834e-01
-2.15894198e+00 -9.00916815e-01 -4.41617727e-01 1.77657032e+00
7.95150578e-01 2.01817080e-01 2.57483497e-02 -2.63001114e-01
9.06669855e-01 7.10771203e-01 -4.69166040e-01 1.09049700e-01
-5.88882864e-01 -7.69307241e-02 4.75501418e-01 5.63958101e-02
-1.20688415e+00 1.37574911e+00 5.32524872e+00 8.98902833e-01
-6.72294974e-01 -8.79559219e-02 7.79518604e-01 4.39649463e-01
-2.04057917e-01 2.86797941e-01 -5.93258023e-01 2.66240239e-01
2.92335991e-02 -1.24074958e-01 2.53628492e-01 8.63327324e-01
-3.64583313e-01 2.76213372e-03 -9.94260132e-01 1.25308859e+00
3.87061715e-01 -1.14932704e+00 5.92628717e-01 -1.76703110e-01
8.07144165e-01 -1.50391579e-01 -2.31770888e-01 3.45414191e-01
3.55578125e-01 -1.11745119e+00 5.43613255e-01 3.64755541e-01
6.02871776e-01 -3.77781510e-01 8.70090425e-01 2.36839086e-01
-1.43598425e+00 6.48974925e-02 -5.89811921e-01 -2.65703559e-01
-2.37168804e-01 5.61247230e-01 -5.85996687e-01 8.40031862e-01
9.10747051e-01 1.23956025e+00 -1.01847279e+00 8.55013430e-01
-6.66663885e-01 4.77062374e-01 2.69957632e-02 -5.51441871e-03
1.22407965e-01 -4.03069496e-01 2.30256721e-01 1.10780573e+00
-1.66497812e-01 2.33458221e-01 2.30911538e-01 8.40008974e-01
-4.04280484e-01 2.47333452e-01 -7.58428872e-01 9.21536759e-02
4.33642566e-01 1.52180696e+00 -9.20538485e-01 -5.23647010e-01
-5.26282787e-01 9.02067542e-01 7.70686805e-01 5.17621219e-01
-8.53691876e-01 -2.70049632e-01 6.15912855e-01 -1.95006132e-01
1.25174478e-01 3.90363671e-02 -3.61854464e-01 -1.48940897e+00
1.96794137e-01 -7.19637692e-01 7.91120052e-01 -1.07926476e+00
-1.44565415e+00 5.60847878e-01 2.59367237e-03 -1.06468403e+00
4.31688279e-01 -6.25403047e-01 -4.69685614e-01 6.40606105e-01
-1.54861784e+00 -1.40958858e+00 -7.91708469e-01 7.64599562e-01
4.93902862e-01 -7.42101297e-03 4.55490768e-01 2.09149942e-01
-8.08900356e-01 4.76028830e-01 -6.08674347e-01 5.15432954e-01
7.53139436e-01 -1.22160637e+00 2.38775089e-01 7.62320161e-01
4.74731922e-01 4.71999377e-01 5.42379737e-01 -7.43911147e-01
-1.06524289e+00 -1.22966754e+00 1.06819427e+00 -5.24932027e-01
8.30550730e-01 -4.90306735e-01 -9.72042978e-01 8.61258149e-01
2.67965853e-01 3.24240148e-01 2.85983682e-01 2.81599969e-01
-8.10071766e-01 -2.15854257e-01 -6.73636913e-01 8.78829896e-01
1.72447288e+00 -7.22187102e-01 -7.16243625e-01 6.30981147e-01
8.20105016e-01 -3.95722538e-01 -5.08540094e-01 8.02138031e-01
1.10807516e-01 -1.12825835e+00 1.08373523e+00 -8.76202643e-01
7.36041069e-01 -4.57189053e-01 -5.33372723e-02 -1.06844318e+00
-6.28368556e-01 -1.65635079e-01 1.61197692e-01 1.58733988e+00
2.62970805e-01 -5.64959288e-01 6.45391107e-01 4.16705817e-01
1.85447633e-01 -8.09382021e-01 -4.21898991e-01 -5.99665761e-01
-4.34126109e-01 -2.24438682e-01 5.07056534e-01 1.34553182e+00
-3.13908637e-01 1.05993843e+00 -4.49044526e-01 2.76622713e-01
5.73165238e-01 4.30909425e-01 8.48240376e-01 -1.11024618e+00
6.41957521e-02 -3.51859748e-01 -8.45057726e-01 -1.05657709e+00
3.92963856e-01 -1.18897665e+00 -1.70443222e-01 -1.69339299e+00
8.14206898e-01 -3.49344581e-01 -4.15516615e-01 6.20836079e-01
-7.82867312e-01 3.25406641e-01 1.15882643e-01 3.42622846e-01
-1.09988081e+00 7.91262090e-01 1.61292613e+00 -2.52574265e-01
2.25876987e-01 -3.30426812e-01 -8.59027445e-01 1.06173968e+00
3.07483286e-01 -2.95571595e-01 -6.94759488e-01 -4.22277123e-01
5.47680855e-01 -2.90274113e-01 6.93281710e-01 -7.11437821e-01
1.89954638e-01 -2.24634573e-01 4.10418898e-01 -4.04501617e-01
6.13961406e-02 -7.47421801e-01 -1.48521855e-01 6.51012436e-02
-4.39921796e-01 -8.58423486e-02 -2.97598243e-01 8.32892716e-01
-4.49544936e-01 2.04735845e-01 5.27511477e-01 -1.54100075e-01
-1.07540631e+00 5.97954333e-01 4.78638619e-01 5.59822619e-01
1.00454688e+00 -7.02087507e-02 -6.87025249e-01 -2.69963384e-01
-4.70319360e-01 3.41652691e-01 3.02805156e-01 6.45293593e-01
4.49262172e-01 -1.48876810e+00 -5.78954995e-01 -6.84016272e-02
6.57236338e-01 3.22451621e-01 3.73097062e-01 7.11046040e-01
-4.34956789e-01 1.88371316e-01 -1.07010208e-01 -6.90340579e-01
-1.31657171e+00 8.00480962e-01 2.15832114e-01 -3.69316131e-01
-7.87776768e-01 1.15167606e+00 9.89469111e-01 -2.95782417e-01
-9.05834546e-04 -1.89758182e-01 -4.99593288e-01 1.70826212e-01
2.53188163e-01 5.75767830e-02 -8.82336125e-02 -1.04773057e+00
-5.68125129e-01 6.95514381e-01 -6.81322217e-02 5.83183467e-01
1.03106499e+00 -1.36910006e-01 -3.88522595e-01 2.87121087e-01
1.13927877e+00 -7.80718103e-02 -9.85846579e-01 -5.94085932e-01
-1.10732168e-01 -6.52638793e-01 -2.44713008e-01 -5.85968733e-01
-1.22574711e+00 7.93513238e-01 2.38781404e-02 3.56760204e-01
1.16627741e+00 3.95719647e-01 6.64075375e-01 2.08638355e-01
2.64221311e-01 -8.34499002e-01 2.14835644e-01 4.00320798e-01
7.51097798e-01 -1.44997644e+00 3.16373378e-01 -1.17751086e+00
-1.04206383e+00 7.06279576e-01 9.35570002e-01 -2.92790711e-01
5.15680373e-01 -2.84558505e-01 3.53492051e-02 -6.36308193e-01
-5.68435609e-01 -6.77533686e-01 6.33784235e-01 6.22680843e-01
3.12388718e-01 7.72423372e-02 -2.59027272e-01 4.31106091e-01
-1.47947356e-01 -4.69598532e-01 -1.31276355e-03 3.99299473e-01
-9.74701568e-02 -6.24207377e-01 2.06007451e-01 5.25839329e-01
-3.53247881e-01 -2.91220278e-01 -6.74588561e-01 8.68273795e-01
2.91123271e-01 8.60316038e-01 -1.86740741e-01 -3.71436477e-01
4.00852650e-01 -2.33423501e-01 3.10897589e-01 -5.66585898e-01
-3.92849952e-01 -2.25133389e-01 2.75426775e-01 -8.22440982e-01
-6.68796062e-01 -3.54511321e-01 -1.19967270e+00 1.60300449e-01
-2.03160346e-01 -1.56311929e-01 -6.24782313e-03 1.16549921e+00
2.85959423e-01 7.87014008e-01 4.73450571e-01 -2.43042693e-01
-2.41471320e-01 -8.27523112e-01 -7.13084877e-01 1.26687074e+00
-1.64747816e-02 -8.66183102e-01 -2.76884228e-01 1.40589699e-01] | [10.308211326599121, 1.6656508445739746] |
7580196f-e1f8-44bd-a873-85d7b902ef1d | spatio-temporal-outdoor-lighting-aggregation | 2202.09206 | null | https://arxiv.org/abs/2202.09206v1 | https://arxiv.org/pdf/2202.09206v1.pdf | Spatio-Temporal Outdoor Lighting Aggregation on Image Sequences using Transformer Networks | In this work, we focus on outdoor lighting estimation by aggregating individual noisy estimates from images, exploiting the rich image information from wide-angle cameras and/or temporal image sequences. Photographs inherently encode information about the scene's lighting in the form of shading and shadows. Recovering the lighting is an inverse rendering problem and as that ill-posed. Recent work based on deep neural networks has shown promising results for single image lighting estimation, but suffers from robustness. We tackle this problem by combining lighting estimates from several image views sampled in the angular and temporal domain of an image sequence. For this task, we introduce a transformer architecture that is trained in an end-2-end fashion without any statistical post-processing as required by previous work. Thereby, we propose a positional encoding that takes into account the camera calibration and ego-motion estimation to globally register the individual estimates when computing attention between visual words. We show that our method leads to improved lighting estimation while requiring less hyper-parameters compared to the state-of-the-art. | ['Carsten Rother', 'Jan Rexilius', 'Robert Herzog', 'Christian Homeyer', 'Haebom Lee'] | 2022-02-18 | null | null | null | null | ['lighting-estimation'] | ['computer-vision'] | [ 4.64551270e-01 -3.23231488e-01 3.48991215e-01 -7.44864881e-01
-6.01488471e-01 -5.89750350e-01 7.38581181e-01 -3.66394222e-01
-5.22937894e-01 4.51011866e-01 3.04948777e-01 6.57315850e-02
3.25134128e-01 -5.21639228e-01 -1.01822877e+00 -7.03908324e-01
4.51217830e-01 1.81584775e-01 1.16511136e-01 3.29114683e-02
3.24994087e-01 4.94259775e-01 -1.65898693e+00 2.79948086e-01
4.44947153e-01 1.09897089e+00 5.42128861e-01 1.07799435e+00
-7.00706765e-02 7.91372180e-01 -4.58210915e-01 -3.79568428e-01
3.17448884e-01 -3.86521876e-01 -3.24001402e-01 4.73267198e-01
1.13803411e+00 -6.34127259e-01 -7.40358010e-02 1.03347957e+00
6.30723238e-01 1.30093023e-01 3.49887013e-01 -8.95912588e-01
-4.98888224e-01 -1.35786355e-01 -5.19352734e-01 5.93915433e-02
4.87041265e-01 1.93624690e-01 7.12397814e-01 -1.03984272e+00
5.03042281e-01 1.08062828e+00 7.61124551e-01 2.46693015e-01
-1.27860129e+00 -3.14396322e-01 3.17473918e-01 3.49391073e-01
-1.25904250e+00 -7.31250584e-01 1.00103152e+00 -3.22737187e-01
9.23765719e-01 1.71078861e-01 8.45411658e-01 1.25640130e+00
-2.98905950e-02 5.01016259e-01 1.40891480e+00 -4.92527127e-01
2.71780640e-01 1.39741987e-01 -3.56283784e-01 6.21225297e-01
-9.77464914e-02 -9.50229238e-04 -7.24981725e-01 2.55550474e-01
8.69327843e-01 -5.14796227e-02 -6.45481825e-01 -6.13867044e-01
-1.20662725e+00 5.49860239e-01 3.87517750e-01 1.22112118e-01
-3.77555043e-01 4.21872735e-01 1.19395301e-01 -1.33633276e-03
8.07230949e-01 1.82360008e-01 -4.63692725e-01 6.07984178e-02
-1.20802391e+00 -4.44879197e-02 7.25965858e-01 7.49280155e-01
8.64185691e-01 8.24988484e-02 2.00407907e-01 7.68225253e-01
3.10260713e-01 5.65889418e-01 2.06821203e-01 -1.11054623e+00
3.77904952e-01 1.39307767e-01 3.75542611e-01 -9.56250787e-01
-3.70531976e-01 -2.22751677e-01 -7.18918800e-01 3.90730083e-01
4.89964396e-01 7.63070509e-02 -9.40941751e-01 1.84897065e+00
4.50917840e-01 5.18421590e-01 -1.12384304e-01 1.23272455e+00
5.28077602e-01 5.39700508e-01 -2.97295302e-01 -3.43879044e-01
1.40034652e+00 -1.04533541e+00 -9.36476231e-01 -4.62841958e-01
-2.71222051e-02 -1.04285693e+00 1.12964904e+00 7.77756453e-01
-1.21484375e+00 -7.53709853e-01 -1.11098480e+00 -5.96293390e-01
-3.82879436e-01 3.55305344e-01 2.86219776e-01 6.77863657e-01
-1.42185628e+00 3.25040817e-01 -8.37217808e-01 -3.76644015e-01
2.61423379e-01 1.18912950e-01 -3.47137958e-01 -2.91077673e-01
-8.77956688e-01 1.01104879e+00 -8.68745819e-02 4.01385903e-01
-9.72465873e-01 -5.86259544e-01 -1.06036520e+00 -5.12765124e-02
4.22548711e-01 -8.59597683e-01 9.99964058e-01 -1.16280758e+00
-2.02140546e+00 6.42540574e-01 -5.33397675e-01 -1.59796685e-01
7.19402611e-01 -6.06727362e-01 -3.23585346e-02 1.53180927e-01
-3.13710034e-01 6.36735022e-01 1.14021409e+00 -1.45272398e+00
-2.82408804e-01 -4.89958316e-01 1.93596214e-01 4.33507174e-01
-3.27359945e-01 -9.82984826e-02 -7.33713388e-01 -3.99395406e-01
-4.74638641e-02 -8.79033744e-01 -1.09316103e-01 2.73041725e-01
-2.45505333e-01 2.10295171e-01 8.05076659e-01 -7.04541802e-01
7.27792323e-01 -1.79549587e+00 3.64270657e-01 -2.05766410e-01
-5.17574735e-02 2.19708279e-01 -8.77900496e-02 2.06201911e-01
-9.11363214e-02 -4.00534153e-01 -2.89356023e-01 -9.85192835e-01
-4.30288650e-02 2.84100860e-01 -4.25039947e-01 9.55974936e-01
1.70801133e-01 5.95732212e-01 -8.14099371e-01 -2.93176144e-01
8.76859605e-01 1.20916188e+00 -6.04921579e-01 5.66755056e-01
-3.00229400e-01 7.33642817e-01 1.19079918e-01 3.31368834e-01
8.90431583e-01 -7.72230327e-02 2.71510482e-01 -7.05218732e-01
-3.00999910e-01 2.12397993e-01 -1.18885851e+00 2.35164905e+00
-1.08006835e+00 1.11046183e+00 1.20236948e-01 -7.02679932e-01
7.23731399e-01 2.18688503e-01 7.29702339e-02 -7.81713307e-01
2.09982187e-01 1.02507010e-01 -6.21358454e-01 -6.68868959e-01
6.13514125e-01 1.46043286e-01 2.77436137e-01 2.03497410e-01
1.93471704e-02 -5.70528388e-01 6.69202730e-02 -1.74185753e-01
6.20274603e-01 8.22021604e-01 3.64511222e-01 1.41600464e-02
7.73393393e-01 -5.22552729e-01 2.21601173e-01 4.41446394e-01
1.96641952e-01 1.07769585e+00 1.48558736e-01 -6.06585145e-01
-1.21997166e+00 -8.56328011e-01 -6.22136258e-02 8.48585308e-01
4.17997651e-02 -2.71035165e-01 -9.26449418e-01 -3.16366524e-01
-3.20345491e-01 7.32868850e-01 -6.13378048e-01 4.16417599e-01
-7.39394486e-01 -4.77297395e-01 6.72534481e-02 3.36644739e-01
4.97635841e-01 -6.71979904e-01 -1.11566961e+00 -1.73781924e-02
-4.05445367e-01 -1.53267932e+00 -5.36029160e-01 1.86278820e-01
-3.23505849e-01 -9.04128075e-01 -1.00080585e+00 -3.14606160e-01
6.13653362e-01 4.74360198e-01 1.33539152e+00 -2.90744364e-01
-4.44212019e-01 5.04114985e-01 -3.84341106e-02 -2.72089213e-01
5.47935069e-02 -2.40831956e-01 -8.64257812e-02 4.08714563e-01
1.09861262e-01 -8.06548476e-01 -9.62490082e-01 3.13516945e-01
-9.09115672e-01 2.37957463e-01 3.02110046e-01 6.83082998e-01
4.96974677e-01 -3.06193978e-01 -1.52447939e-01 -7.69944489e-01
1.34682089e-01 -1.45033583e-01 -9.98367369e-01 1.68722942e-01
-3.35108697e-01 4.78958972e-02 7.52667785e-01 -2.90342897e-01
-1.32303166e+00 3.56454998e-01 -6.87827542e-02 -7.25172877e-01
-3.32476616e-01 -1.60271391e-01 -2.03772441e-01 -1.10750593e-01
4.35961843e-01 1.62176356e-01 -2.24125624e-01 -4.62096035e-01
6.55588090e-01 4.89607573e-01 6.32373333e-01 -2.99076378e-01
5.52457273e-01 9.70259964e-01 1.97032139e-01 -8.88114035e-01
-1.09954333e+00 -5.28551221e-01 -8.75398457e-01 -4.14729536e-01
1.11500859e+00 -1.11960340e+00 -9.31590974e-01 5.57452261e-01
-1.47709608e+00 -4.62418050e-01 -4.86106500e-02 4.40777004e-01
-7.48832941e-01 4.48699236e-01 -2.27331057e-01 -9.14070189e-01
-3.97590958e-02 -1.39471841e+00 1.51009440e+00 1.10450327e-01
3.18400259e-03 -9.40214396e-01 1.73041001e-01 4.20120835e-01
5.46424925e-01 2.68983752e-01 2.51975417e-01 2.13551134e-01
-8.85183096e-01 1.69833839e-01 -3.84101331e-01 5.27630866e-01
2.68334290e-03 -1.65675566e-01 -1.56247520e+00 -1.18677139e-01
2.72158712e-01 -2.71696240e-01 8.89651060e-01 5.23282051e-01
1.18144691e+00 -2.04991624e-01 2.53466547e-01 1.06349492e+00
1.92257023e+00 -2.38259748e-01 6.20984077e-01 2.48149380e-01
1.05908906e+00 8.31995010e-01 2.70885319e-01 4.91142541e-01
6.74176574e-01 1.04736054e+00 7.93994427e-01 -3.38566869e-01
-4.55150425e-01 2.20881961e-02 3.80005151e-01 5.94035387e-01
-4.95617315e-02 -5.80559015e-01 -6.96103156e-01 5.93837082e-01
-1.64652550e+00 -8.29017878e-01 -1.30690068e-01 2.29328561e+00
7.49218106e-01 -2.17505574e-01 -3.20558578e-01 -2.44848765e-02
4.52733576e-01 6.14093840e-01 -5.79512537e-01 -5.27589262e-01
-1.48445398e-01 2.33496213e-03 6.37653530e-01 8.38889658e-01
-8.66040885e-01 7.87588596e-01 5.68652821e+00 3.32753718e-01
-1.16490960e+00 2.14146063e-01 6.13855243e-01 -3.52618933e-01
-2.99566418e-01 -3.89848650e-02 -5.53455591e-01 3.74330223e-01
8.44033003e-01 6.05250597e-01 8.09490085e-01 5.82273722e-01
3.22221965e-01 -4.96404827e-01 -1.21561885e+00 1.18853319e+00
7.69288540e-01 -9.31496918e-01 -2.66800255e-01 -4.29757833e-02
8.29951882e-01 1.27551958e-01 1.94453850e-01 -3.68152678e-01
1.33521587e-01 -8.99151981e-01 9.37271357e-01 8.42281222e-01
9.03608859e-01 -5.59204340e-01 5.12297213e-01 1.51931673e-01
-1.11615646e+00 7.11200610e-02 -3.56235713e-01 -1.41987223e-02
5.37608981e-01 7.92939186e-01 -6.74401402e-01 4.72742260e-01
7.92290151e-01 8.54493558e-01 -5.28487265e-01 7.11830676e-01
-4.95536655e-01 1.92591324e-01 -3.47262800e-01 3.09714109e-01
1.17005363e-01 -4.35984671e-01 3.17547292e-01 1.22508705e+00
4.11819756e-01 -7.80997723e-02 -8.32376257e-03 8.33607376e-01
7.75466114e-02 -2.56894499e-01 -5.75175822e-01 4.92734909e-01
1.82318576e-02 1.41576910e+00 -5.76410711e-01 -3.71790111e-01
-5.41841924e-01 1.41213155e+00 3.16831529e-01 5.01404881e-01
-8.18281353e-01 -1.04921222e-01 6.27065897e-01 -1.20261870e-02
5.49653828e-01 -4.72267598e-01 -1.09239511e-01 -1.29660404e+00
1.66515037e-01 -5.37004828e-01 -2.50160038e-01 -1.49786353e+00
-1.01487410e+00 7.55312979e-01 -1.55928835e-01 -1.03216422e+00
-3.01238865e-01 -6.70706987e-01 -2.49421611e-01 9.91616666e-01
-2.08474588e+00 -1.26431704e+00 -8.04875731e-01 5.52962661e-01
8.19255471e-01 3.74595672e-01 6.48846269e-01 3.17712009e-01
-3.05223286e-01 1.69816703e-01 2.74174139e-02 -2.36217737e-01
9.53314602e-01 -1.35058141e+00 5.24583697e-01 1.08476377e+00
2.56477505e-01 4.17957127e-01 9.79153991e-01 -1.19895130e-01
-1.49786913e+00 -9.05471385e-01 9.10398662e-01 -6.09153926e-01
3.55560809e-01 -7.39710331e-01 -6.57494068e-01 5.22267997e-01
6.63550496e-01 3.98167223e-01 3.19017678e-01 -1.74208939e-01
-4.36507016e-01 -2.69397497e-01 -8.72466743e-01 4.88872737e-01
1.01191080e+00 -8.11174035e-01 -1.87396407e-01 3.59266460e-01
5.36493719e-01 -6.61818564e-01 -5.61120629e-01 -3.53035145e-02
7.10727513e-01 -1.57795990e+00 1.24996281e+00 1.36975229e-01
5.32131493e-01 -4.67611492e-01 -4.08974797e-01 -1.35417128e+00
5.14597185e-02 -5.58540940e-01 9.03257914e-03 1.09994817e+00
-1.08106494e-01 -4.54316050e-01 4.95212018e-01 4.70090032e-01
-7.03128576e-02 -3.82572800e-01 -7.49727845e-01 -2.45446488e-01
-5.10931492e-01 -5.38060009e-01 4.15533543e-01 7.14379370e-01
-7.15374947e-01 3.12103927e-01 -8.38186264e-01 3.09540749e-01
8.94791543e-01 4.49713878e-02 1.03761828e+00 -9.70528841e-01
-2.89362907e-01 -8.17235112e-02 -2.23142818e-01 -1.18942857e+00
3.38054955e-01 -3.25587511e-01 3.10398012e-01 -1.41188157e+00
-7.61780143e-02 4.94847670e-02 4.81220596e-02 5.78283593e-02
-1.39528945e-01 6.89250231e-01 2.40939960e-01 -3.16218510e-02
-5.56906521e-01 5.00450671e-01 1.05475044e+00 1.52474359e-01
1.90236568e-01 -4.18131918e-01 -1.97164074e-01 8.42725933e-01
5.01344979e-01 -1.78511664e-01 -5.77609837e-01 -9.70737576e-01
5.94201803e-01 2.02388223e-02 6.22936487e-01 -8.69479895e-01
3.17581117e-01 1.33936316e-01 6.05599523e-01 -7.50453234e-01
9.22082543e-01 -1.14929116e+00 2.04233751e-01 2.76049860e-02
-4.18974072e-01 1.84242427e-01 -8.77195671e-02 8.39972734e-01
-7.87134096e-02 -3.19274403e-02 8.53150070e-01 -2.76997626e-01
-5.80214381e-01 5.55729009e-02 -1.21236168e-01 -1.44671515e-01
7.17666388e-01 -1.74107239e-01 -2.17959449e-01 -4.49264973e-01
-4.08909470e-01 -2.09985584e-01 8.14906657e-01 1.53248876e-01
5.73742807e-01 -1.00232768e+00 -5.73802352e-01 4.24962759e-01
-6.34032264e-02 7.59074911e-02 3.88933092e-01 9.05385673e-01
-7.00020313e-01 3.80054891e-01 -1.39541447e-01 -8.85257125e-01
-1.24655533e+00 5.38846493e-01 3.73367697e-01 -9.97476131e-02
-6.09812677e-01 9.19808567e-01 3.06964695e-01 -3.16734344e-01
1.88348219e-01 -4.57781345e-01 -1.95919380e-01 1.75896451e-01
6.06688917e-01 1.24953292e-01 2.43663073e-01 -7.74956465e-01
-2.97175586e-01 1.13298953e+00 1.72984332e-01 -3.42616528e-01
1.42387223e+00 -7.80375659e-01 -9.67520550e-02 6.61073923e-01
1.63444722e+00 8.90982300e-02 -1.73518491e+00 -3.61020684e-01
-4.82673973e-01 -8.94149542e-01 4.66628879e-01 -7.02134371e-01
-1.12525690e+00 1.16362560e+00 7.38445938e-01 -8.15588906e-02
1.32333398e+00 -3.65660220e-01 6.14798248e-01 1.94139063e-01
2.55259633e-01 -1.12054491e+00 8.03833157e-02 3.60047907e-01
8.49999964e-01 -1.55675423e+00 2.41480917e-01 -2.37936407e-01
-5.32580674e-01 1.30085099e+00 2.51973748e-01 -1.79138377e-01
4.02559936e-01 3.14049602e-01 2.73829877e-01 -4.30085622e-02
-6.97619438e-01 -2.99325347e-01 2.89628893e-01 5.31411707e-01
3.62207979e-01 -3.38108629e-01 1.18870609e-01 -1.92983836e-01
-1.07955188e-01 2.75167897e-02 5.58483303e-01 4.66271639e-01
-1.14171058e-01 -8.10804367e-01 -4.37383771e-01 -2.46845737e-01
-3.60695839e-01 -4.07217592e-01 -1.96820423e-01 5.02923906e-01
2.11936384e-01 8.97902846e-01 3.19315910e-01 9.36881229e-02
1.50568992e-01 -4.61382717e-02 7.16799200e-01 -1.71517625e-01
-3.89466137e-01 2.04909116e-01 3.40577960e-02 -9.70347583e-01
-9.72225487e-01 -6.33493245e-01 -4.94328916e-01 8.50134250e-03
-2.39259571e-01 -4.14964736e-01 1.35462129e+00 8.11873019e-01
1.07603081e-01 7.66426504e-01 6.84648573e-01 -1.63689196e+00
-1.02695867e-01 -8.17849696e-01 -2.10022241e-01 3.43883574e-01
8.41219664e-01 -5.99995017e-01 -5.11178195e-01 3.35011244e-01] | [9.775296211242676, -2.885542392730713] |
25a17fbb-95ae-4a45-9b79-36ece6262929 | cnn-bilstm-model-for-english-handwriting | 2307.00664 | null | https://arxiv.org/abs/2307.00664v1 | https://arxiv.org/pdf/2307.00664v1.pdf | CNN-BiLSTM model for English Handwriting Recognition: Comprehensive Evaluation on the IAM Dataset | We present a CNN-BiLSTM system for the problem of offline English handwriting recognition, with extensive evaluations on the public IAM dataset, including the effects of model size, data augmentation and the lexicon. Our best model achieves 3.59\% CER and 9.44\% WER using CNN-BiLSTM network with CTC layer. Test time augmentation with rotation and shear transformations applied to the input image, is proposed to increase recognition of difficult cases and found to reduce the word error rate by 2.5\% points. We also conduct an error analysis of our proposed method on IAM dataset, show hard cases of handwriting images and explore samples with erroneous labels. We provide our source code as public-domain, to foster further research to encourage scientific reproducibility. | ['Berrin Yanikoglu', 'Firat Kizilirmak'] | 2023-07-02 | null | null | null | null | ['handwriting-recognition'] | ['computer-vision'] | [ 3.18578094e-01 -3.55067812e-02 -2.15751261e-01 -4.47243810e-01
-1.04843223e+00 -6.67148113e-01 3.46125960e-01 -4.98459250e-01
-7.45243728e-01 5.43701649e-01 -1.71724670e-02 -7.91324198e-01
2.92184293e-01 -4.15522069e-01 -7.73253143e-01 -3.60931098e-01
2.91712105e-01 3.27562183e-01 3.51306796e-02 -1.82446931e-02
5.01155794e-01 3.73248041e-01 -6.77997589e-01 5.74864626e-01
8.15275311e-01 8.46647143e-01 2.58654445e-01 9.82297838e-01
2.17233703e-01 8.76735389e-01 -1.22059751e+00 -1.10092664e+00
1.90175027e-01 -2.49932706e-01 -9.57315385e-01 2.83853054e-01
6.84749484e-01 -4.81792390e-01 -5.32970548e-01 9.35882568e-01
7.15575755e-01 5.01896925e-02 5.44880867e-01 -7.85353959e-01
-1.18973291e+00 6.56598508e-01 -3.61504227e-01 1.02050193e-01
-2.18637362e-01 4.63087231e-01 7.64784217e-01 -1.05290198e+00
5.76670527e-01 8.53045523e-01 7.46277392e-01 7.49359071e-01
-6.35434330e-01 -7.57456779e-01 -1.42946690e-01 3.20759296e-01
-1.05649054e+00 -2.69793391e-01 3.46194565e-01 -5.39982133e-02
1.58303535e+00 1.67478055e-01 2.66225278e-01 1.42558181e+00
3.50051910e-01 9.68855619e-01 1.28077507e+00 -5.97109258e-01
-3.37973624e-01 -9.13262814e-02 5.72810709e-01 9.99271512e-01
5.91404848e-02 -1.40153095e-01 -6.13283336e-01 3.26311439e-01
8.27962458e-01 -4.33911204e-01 -4.26533110e-02 6.08449578e-01
-1.02336609e+00 5.55334389e-01 9.02428478e-02 3.51457357e-01
-2.78495979e-02 2.54350603e-01 6.28126681e-01 4.44307625e-01
3.10424000e-01 6.27076387e-01 -5.97964227e-01 -6.56532705e-01
-9.08170462e-01 -1.85252234e-01 6.29513383e-01 1.41813672e+00
-2.68632807e-02 4.85166639e-01 -5.20077169e-01 1.12737787e+00
1.51635677e-01 6.16228580e-01 7.55569994e-01 -5.57756007e-01
9.17147338e-01 6.08696878e-01 -1.62811860e-01 -4.65990126e-01
-2.81496137e-01 -3.96207273e-01 -8.49587858e-01 -1.37218326e-01
6.48501337e-01 -2.04574436e-01 -1.50257969e+00 1.38202059e+00
-7.13142276e-01 -1.40287101e-01 1.28299817e-01 8.65621328e-01
6.78445399e-01 4.43373710e-01 -2.64289938e-02 1.96896821e-01
1.37808740e+00 -1.21789587e+00 -1.09406424e+00 -5.86591884e-02
7.67055511e-01 -1.16617119e+00 1.46532524e+00 7.08149433e-01
-1.16497433e+00 -7.57557988e-01 -1.23021805e+00 -3.07220608e-01
-3.57250214e-01 1.25956368e+00 2.56103903e-01 8.95922542e-01
-7.82020807e-01 5.49366951e-01 -7.36382663e-01 -3.53810728e-01
6.33722126e-01 4.65380937e-01 -2.30271310e-01 -1.40886799e-01
-9.85191524e-01 1.24603939e+00 3.10123801e-01 4.98782218e-01
-1.15387201e+00 -2.61368930e-01 -6.76484644e-01 -2.81390280e-01
1.43198743e-01 1.81716383e-01 1.19762444e+00 -4.06349599e-01
-1.49698329e+00 9.74941850e-01 -1.82837650e-01 -7.06392169e-01
7.30029881e-01 -2.80375838e-01 -6.14994526e-01 -3.64108264e-01
-2.67185450e-01 3.74944359e-01 6.57789052e-01 -7.02087462e-01
-3.88301849e-01 -1.94325954e-01 -3.21262300e-01 -1.52134940e-01
-4.10758913e-01 1.02745175e-01 -7.37277269e-01 -9.43835318e-01
-5.65325059e-02 -1.01848614e+00 4.19732183e-02 -6.67309761e-01
-5.69796681e-01 -2.75115728e-01 7.29676187e-01 -1.37393820e+00
1.15826583e+00 -2.04285169e+00 -3.15558314e-01 7.74100721e-02
-2.07204208e-01 6.90155208e-01 -5.38842082e-01 4.71842885e-02
4.03873801e-01 4.22696829e-01 -1.45374134e-01 -4.96700436e-01
-2.53830049e-02 2.24425957e-01 -4.54629451e-01 2.57747412e-01
6.74969494e-01 1.02554190e+00 -4.33484435e-01 -1.98078811e-01
1.58311084e-01 9.72056761e-02 -1.34564742e-01 1.84923097e-01
-7.24426880e-02 -7.73591548e-03 -3.09455395e-02 9.20361519e-01
5.22217333e-01 -3.37721288e-01 8.66562873e-02 -8.72030407e-02
-7.01517332e-03 3.19424301e-01 -8.83175015e-01 1.53837597e+00
-5.73781371e-01 1.02718222e+00 -3.82466614e-01 -4.08279896e-01
1.27575815e+00 3.63720685e-01 -4.19517756e-01 -7.39149511e-01
3.65376264e-01 4.63968039e-01 3.95288289e-01 -4.03036654e-01
7.89234161e-01 4.49448913e-01 7.62318447e-02 1.89780056e-01
4.01709527e-01 -5.59372120e-02 1.21569209e-01 -1.04894616e-01
9.39325571e-01 9.73364487e-02 -3.63010854e-01 -2.05300555e-01
3.04021031e-01 1.92284003e-01 4.13245082e-01 1.02196348e+00
-3.23946595e-01 6.29661262e-01 4.24080670e-01 -4.43141311e-01
-1.44235086e+00 -7.08667874e-01 -1.71347991e-01 8.02375019e-01
-2.70526439e-01 -1.30256817e-01 -8.27590525e-01 -5.65608203e-01
-2.12265253e-01 5.76423645e-01 -5.35190165e-01 -2.58321706e-02
-7.51322031e-01 -1.02727115e+00 1.28696084e+00 1.00190377e+00
9.39458787e-01 -1.18182814e+00 -2.47285604e-01 2.22841039e-01
-3.23881060e-02 -1.27521706e+00 -7.42070854e-01 4.14215386e-01
-8.09409618e-01 -1.22767878e+00 -8.34681273e-01 -1.26773965e+00
5.38413107e-01 -2.90838003e-01 8.25595737e-01 1.47878855e-01
-3.98924500e-01 -1.53869316e-01 -3.91909540e-01 -4.52728868e-01
-4.52644348e-01 4.17943656e-01 -5.74955828e-02 -3.54829341e-01
4.67332214e-01 2.28087351e-01 -1.69568598e-01 4.88409579e-01
-4.50571150e-01 2.30729342e-01 7.74896741e-01 1.36332762e+00
3.14840794e-01 -3.97477210e-01 5.16979754e-01 -8.49859774e-01
7.31729567e-01 1.51806355e-01 -6.23089254e-01 5.83741724e-01
-8.68600905e-01 3.24796103e-02 5.89215219e-01 -6.10412478e-01
-8.87846828e-01 -2.93771267e-01 -2.94874877e-01 -3.74708951e-01
9.99757424e-02 2.41134450e-01 -8.86179358e-02 -9.63443890e-02
4.31697756e-01 4.93670434e-01 -1.63829297e-01 -5.30535758e-01
1.45305127e-01 1.15394330e+00 7.50890493e-01 -6.01824284e-01
1.98854610e-01 1.02744140e-02 -3.54055196e-01 -6.54156148e-01
-5.68437994e-01 1.82929018e-03 -6.65564001e-01 -2.31252864e-01
7.68506169e-01 -7.40304232e-01 -1.00568640e+00 1.09192240e+00
-1.31372178e+00 -7.29584455e-01 1.80302918e-01 4.92508143e-01
-4.26956087e-01 3.44272524e-01 -1.23878229e+00 -7.15367913e-01
-5.84036529e-01 -1.60152054e+00 8.05563390e-01 8.81520808e-02
-9.82734114e-02 -7.39124179e-01 -6.34071529e-01 7.02901125e-01
4.40285265e-01 -1.77468166e-01 1.01495385e+00 -9.72183049e-01
-3.83857936e-01 -2.86514908e-01 -3.45173120e-01 7.94608951e-01
-3.43493856e-02 1.60661966e-01 -1.15930641e+00 -4.07564104e-01
-3.20535213e-01 -5.49803257e-01 1.02182472e+00 3.17894191e-01
1.39860070e+00 -1.59397691e-01 5.59205525e-02 5.14765322e-01
1.13703406e+00 5.75702190e-01 8.69220078e-01 3.70637119e-01
7.51375973e-01 2.60177910e-01 5.57254910e-01 3.58441651e-01
1.01741580e-02 2.76139826e-01 -3.10145002e-02 -9.46452171e-02
-5.01260340e-01 -2.58349448e-01 3.80147010e-01 1.23627651e+00
-2.95748889e-01 -5.54717362e-01 -1.16608882e+00 5.61303139e-01
-1.41050518e+00 -3.55933249e-01 -4.85089928e-01 1.90957236e+00
9.25759137e-01 4.60756749e-01 -1.21518478e-01 2.75320023e-01
7.32545197e-01 -1.06170908e-01 -3.60772491e-01 -6.37519479e-01
-3.38981241e-01 3.74776632e-01 9.66347516e-01 5.98811924e-01
-1.07798290e+00 1.49102581e+00 7.07358408e+00 1.06622982e+00
-1.10494363e+00 1.97920159e-01 9.85886216e-01 1.25661463e-01
2.57245779e-01 -2.97872841e-01 -1.17144227e+00 4.87912506e-01
1.01498413e+00 3.02842289e-01 3.53107989e-01 6.28487289e-01
3.74394581e-02 1.37453571e-01 -9.67117608e-01 9.54381526e-01
3.10076356e-01 -1.19963872e+00 6.44648522e-02 8.69485438e-02
6.59200430e-01 7.29881600e-02 2.14998022e-01 4.37825173e-01
4.17724460e-01 -1.32762003e+00 5.43150723e-01 4.24700528e-01
1.21234739e+00 -7.28300512e-01 1.19574952e+00 2.85260640e-02
-7.49818802e-01 9.31593329e-02 -4.42794412e-01 1.38889672e-02
-2.96275079e-01 1.54007701e-02 -1.14875233e+00 2.51588196e-01
5.65233648e-01 5.61844647e-01 -9.98933434e-01 5.58402061e-01
-1.90848663e-01 1.11685860e+00 -7.60820415e-03 -4.60820526e-01
4.86795217e-01 -1.01835974e-01 1.27662897e-01 1.64281166e+00
2.03912094e-01 -2.43619308e-01 -2.49638945e-01 7.87063777e-01
-4.26409990e-01 -5.15109263e-02 -3.48385453e-01 -2.75516689e-01
4.26750183e-01 9.18259144e-01 -4.49980676e-01 -4.59359795e-01
-3.25069815e-01 1.26830864e+00 2.88100570e-01 3.41856897e-01
-9.40010846e-01 -7.30587840e-01 1.30790681e-01 -4.11170363e-01
2.54920535e-02 -3.81084532e-01 -8.11214030e-01 -1.01178098e+00
4.30612028e-01 -1.16581059e+00 2.73972481e-01 -7.25733757e-01
-1.18978858e+00 7.23420858e-01 -7.35658050e-01 -1.05674851e+00
-9.88994073e-03 -1.28207695e+00 -4.76817250e-01 1.12048495e+00
-1.26580691e+00 -1.41389191e+00 -1.32789686e-01 3.08522642e-01
9.04779553e-01 -9.31950271e-01 8.07906926e-01 3.91280234e-01
-6.84849560e-01 1.31581569e+00 3.29183429e-01 9.23450649e-01
8.49722624e-01 -1.30490088e+00 8.88206899e-01 9.46922779e-01
6.68724775e-02 5.46410561e-01 7.41100162e-02 -6.25310540e-01
-1.45878780e+00 -1.26397455e+00 9.86486793e-01 -7.57734179e-01
5.64593792e-01 -5.31952798e-01 -7.46554971e-01 8.21964145e-01
1.79786131e-01 -3.75410855e-01 6.04237497e-01 7.55043402e-02
-3.82894307e-01 1.18645892e-01 -9.98221934e-01 7.06943333e-01
7.90652692e-01 -6.05923772e-01 -4.46806401e-01 4.10812438e-01
5.82005858e-01 -6.63508892e-01 -1.24420631e+00 3.90081286e-01
5.87335825e-01 -1.94722638e-01 4.28212196e-01 -8.91581237e-01
5.72421312e-01 -1.48803024e-02 -2.44188145e-01 -9.90566730e-01
-6.84417784e-02 -3.75297546e-01 2.54562050e-02 1.28606868e+00
8.60686779e-01 -4.40051287e-01 7.37016261e-01 5.28150678e-01
-4.07018811e-01 -9.35019553e-01 -1.01291823e+00 -1.07795131e+00
3.71217281e-01 -8.39576066e-01 2.13417634e-01 8.64120901e-01
-2.23431200e-01 7.33124018e-02 -6.85975373e-01 1.05395369e-01
1.89960584e-01 -3.29497814e-01 5.74270666e-01 -3.54178101e-01
-1.35666609e-01 -5.03328443e-01 -4.08671051e-01 -1.01166666e+00
1.09458946e-01 -8.36520851e-01 2.76015718e-02 -1.23333251e+00
3.39307696e-01 -7.13711977e-02 -1.19969353e-01 9.82950866e-01
1.30375167e-02 7.14430392e-01 3.90813738e-01 2.01534957e-01
-4.57035154e-01 2.68140465e-01 1.38914585e+00 -6.61431909e-01
1.12457938e-01 -3.08778077e-01 -1.88451007e-01 3.66507351e-01
8.68938923e-01 -1.49261028e-01 1.59137845e-01 -9.63538170e-01
-2.43176699e-01 -8.02594200e-02 1.31917804e-01 -9.47295964e-01
6.44720718e-02 9.39064696e-02 7.84476638e-01 -6.14802837e-01
2.45772734e-01 -3.99047166e-01 -5.42268097e-01 7.57717907e-01
-6.52965486e-01 1.28633112e-01 6.67334437e-01 -5.20780198e-02
-1.87893078e-01 -5.29253721e-01 7.39872634e-01 1.67686060e-01
-7.09111094e-01 8.21483806e-02 -4.73641723e-01 -9.23421904e-02
5.60750008e-01 -1.18753985e-01 -5.17663836e-01 -1.68798745e-01
-7.52304971e-01 3.78348529e-01 1.42496452e-01 7.23352432e-01
6.41494334e-01 -1.23317826e+00 -1.01331019e+00 2.41579250e-01
1.31824061e-01 -3.82652044e-01 -2.17078533e-02 5.23988724e-01
-7.50779986e-01 6.92109704e-01 -1.46988407e-01 -4.41951543e-01
-1.53712213e+00 4.74463357e-03 4.64059681e-01 -3.34925056e-01
-3.56285840e-01 8.37263823e-01 -5.73826611e-01 -5.69670379e-01
6.44124448e-01 -6.52295828e-01 1.31466553e-01 -3.50645989e-01
3.00602257e-01 4.53996271e-01 4.18386906e-01 -2.26337329e-01
-2.69142359e-01 4.58399653e-01 -3.51024687e-01 -1.85284331e-01
9.84925091e-01 5.90264559e-01 7.01027662e-02 4.53160614e-01
1.08827114e+00 -2.44100764e-01 -1.03534830e+00 -7.21354857e-02
2.56715238e-01 -3.65192622e-01 -2.22337529e-01 -1.43050945e+00
-9.29335356e-01 1.03759360e+00 9.49613988e-01 -3.98306906e-01
1.05865669e+00 -3.14086288e-01 7.46837318e-01 8.42391968e-01
1.27035111e-01 -1.57239997e+00 6.31100982e-02 1.33071494e+00
8.83231044e-01 -1.43099046e+00 -3.87687474e-01 1.18224509e-01
-8.14469516e-01 1.37797368e+00 9.20896292e-01 -1.01548962e-01
8.39545503e-02 5.94942033e-01 3.95768911e-01 1.27608031e-01
-6.79205000e-01 2.21858904e-01 2.97904223e-01 2.65773505e-01
8.69590342e-01 1.10313684e-01 -5.22256255e-01 9.70075190e-01
-2.52968997e-01 -7.05509558e-02 5.04585385e-01 6.72017097e-01
1.00659154e-01 -9.85024989e-01 -2.70318359e-01 7.66413867e-01
-7.17779100e-01 -3.12409073e-01 -5.84910989e-01 1.00391698e+00
-7.62915285e-03 8.36491823e-01 1.97093755e-01 -7.53480494e-01
5.27045786e-01 3.63385946e-01 6.62950575e-01 -2.85489798e-01
-7.91777968e-01 -6.90375119e-02 6.39649257e-02 -3.26409601e-02
1.20483913e-01 -3.06536466e-01 -1.19816709e+00 -2.25836217e-01
-6.21750474e-01 -2.01634094e-01 8.19772005e-01 7.99394071e-01
2.49987677e-01 7.88819790e-01 3.39019120e-01 -1.86659396e-01
-9.47091639e-01 -1.87474513e+00 -4.36539233e-01 4.51422274e-01
1.23876497e-01 -1.43466905e-01 5.24763577e-02 2.70932972e-01] | [11.877700805664062, 2.4863877296447754] |
03f6cbe5-b82c-4388-8dc9-c01d99f8809d | a-novel-approach-for-dimensionality-reduction | 2210.13901 | null | https://arxiv.org/abs/2210.13901v1 | https://arxiv.org/pdf/2210.13901v1.pdf | A Novel Approach for Dimensionality Reduction and Classification of Hyperspectral Images based on Normalized Synergy | During the last decade, hyperspectral images have attracted increasing interest from researchers worldwide. They provide more detailed information about an observed area and allow an accurate target detection and precise discrimination of objects compared to classical RGB and multispectral images. Despite the great potentialities of hyperspectral technology, the analysis and exploitation of the large volume data remain a challenging task. The existence of irrelevant redundant and noisy images decreases the classification accuracy. As a result, dimensionality reduction is a mandatory step in order to select a minimal and effective images subset. In this paper, a new filter approach normalized mutual synergy (NMS) is proposed in order to detect relevant bands that are complementary in the class prediction better than the original hyperspectral cube data. The algorithm consists of two steps: images selection through normalized synergy information and pixel classification. The proposed approach measures the discriminative power of the selected bands based on a combination of their maximal normalized synergic information, minimum redundancy and maximal mutual information with the ground truth. A comparative study using the support vector machine (SVM) and k-nearest neighbor (KNN) classifiers is conducted to evaluate the proposed approach compared to the state of art band selection methods. Experimental results on three benchmark hyperspectral images proposed by the NASA "Aviris Indiana Pine", "Salinas" and "Pavia University" demonstrated the robustness, effectiveness and the discriminative power of the proposed approach over the literature approaches. Keywords: Hyperspectral images; target detection; pixel classification; dimensionality reduction; band selection; information theory; mutual information; normalized synergy | ['Nacir Chafik', 'Ahmed Hammouch', 'Elkebir Sarhrouni', 'Hasna Nhaila', 'Asma Elmaizi'] | 2022-10-25 | null | null | null | null | ['classification-of-hyperspectral-images'] | ['computer-vision'] | [ 8.66098225e-01 -6.59113169e-01 -9.54044312e-02 -4.38338146e-02
-4.13558275e-01 -6.84076846e-01 4.13670391e-01 2.82776088e-01
-3.71966094e-01 1.01970863e+00 -2.07629219e-01 -1.34117668e-02
-1.08537221e+00 -9.41421926e-01 -1.30253628e-01 -1.25873029e+00
-1.27431661e-01 6.83040693e-02 1.54719232e-02 -1.95577055e-01
5.32356381e-01 8.93033922e-01 -2.16120982e+00 2.20205173e-01
1.18579197e+00 1.24141145e+00 5.19283891e-01 2.49976486e-01
2.00399145e-01 1.37131423e-01 -1.35039479e-01 4.84789789e-01
5.62144876e-01 -3.46267968e-01 -5.55353999e-01 2.56623060e-01
1.03288218e-01 6.31819069e-02 2.30394870e-01 1.43692088e+00
5.24004757e-01 3.94413114e-01 7.70041287e-01 -1.00158775e+00
-1.72632009e-01 3.80028993e-01 -7.84272015e-01 3.47553223e-01
-1.21915542e-01 6.45634085e-02 1.00528610e+00 -6.42548919e-01
3.62486482e-01 6.96573496e-01 2.96481252e-01 -2.36391023e-01
-1.25682926e+00 -6.60447121e-01 -2.33445406e-01 5.88129759e-01
-1.57713223e+00 2.53244732e-02 7.65484989e-01 -6.42182410e-01
7.47413754e-01 7.75755763e-01 8.74613225e-01 1.86528414e-01
-3.24216159e-03 1.66098371e-01 1.49388337e+00 -4.89475012e-01
1.26756936e-01 4.41004723e-01 3.76996398e-01 3.01305383e-01
6.16703629e-01 3.86954576e-01 -2.93564588e-01 -9.58806425e-02
1.68644160e-01 7.85302445e-02 -7.55029321e-01 -3.99307996e-01
-8.95464242e-01 8.10736895e-01 7.24846184e-01 6.25617981e-01
-7.22236097e-01 -7.26340234e-01 9.54448879e-02 4.67143543e-02
3.83434206e-01 4.40684795e-01 -2.54369885e-01 3.57146174e-01
-9.34287786e-01 -1.45332918e-01 3.68386894e-01 2.49278054e-01
1.02345073e+00 3.73789072e-02 2.22209051e-01 7.64972985e-01
2.23641038e-01 6.80572510e-01 5.56751370e-01 -2.85983771e-01
1.90793678e-01 1.15275884e+00 1.75824627e-01 -1.33188450e+00
-5.34184098e-01 -8.29927206e-01 -7.67031908e-01 4.33160573e-01
-9.47336704e-02 3.82040292e-02 -6.42535925e-01 1.04492688e+00
2.86584198e-01 -1.23646677e-01 3.63740623e-01 1.03927946e+00
5.89992881e-01 8.01353991e-01 6.33296138e-03 -4.58820909e-01
9.72168684e-01 -3.47195983e-01 -3.68324041e-01 -3.51665393e-02
3.17311794e-01 -7.91220903e-01 6.66229129e-01 5.08427322e-01
-3.11291069e-01 -3.23068291e-01 -1.27583325e+00 9.27461684e-01
-6.97711110e-01 3.94878477e-01 6.92712247e-01 8.21128368e-01
-3.93813759e-01 6.25057220e-01 -3.60093772e-01 -3.33587646e-01
3.42657238e-01 5.24850488e-01 -6.01513147e-01 1.86171550e-02
-8.75428736e-01 8.71915460e-01 9.95452166e-01 1.97248891e-01
-3.06979299e-01 -4.44586098e-01 -4.03885394e-01 8.91845524e-02
2.05051512e-01 4.94276173e-02 2.79889286e-01 -1.26659882e+00
-1.01089835e+00 5.83161175e-01 1.25706360e-01 -3.45905334e-01
2.91043613e-02 5.91714382e-02 -4.07956719e-01 3.85831833e-01
-3.83541780e-03 1.84544265e-01 2.15982080e-01 -1.26563036e+00
-8.98990214e-01 -9.39889431e-01 -3.21417660e-01 4.94626909e-01
-6.09877646e-01 -2.25716516e-01 5.88401854e-01 -3.00119638e-01
7.18421280e-01 -9.40169990e-01 6.34171814e-02 -4.87308979e-01
-4.04528260e-01 2.91548938e-01 1.12907803e+00 -6.59983516e-01
8.84955108e-01 -2.06997108e+00 1.79075882e-01 7.00773954e-01
-3.98985475e-01 4.15163040e-01 3.93882543e-02 3.74958575e-01
-4.97810215e-01 1.07009865e-01 -4.94273365e-01 7.13707447e-01
-5.46865404e-01 -1.57751277e-01 1.34836078e-01 7.59218335e-01
-1.37768136e-02 3.19313332e-02 -5.76321185e-01 -1.92231581e-01
6.17675841e-01 4.55850452e-01 7.76843131e-02 -1.37188882e-01
8.10897648e-02 4.14187074e-01 -3.85039300e-01 9.34853137e-01
7.43880749e-01 4.08663787e-02 5.05269356e-02 -8.42568398e-01
-5.49953103e-01 -5.38497627e-01 -1.43121028e+00 1.00829077e+00
1.37637863e-02 4.80733871e-01 -6.79719001e-02 -1.16600752e+00
1.13645053e+00 3.48032981e-01 6.99502110e-01 -4.95028526e-01
3.18185210e-01 4.75990355e-01 1.25619799e-01 -5.40247202e-01
2.71377832e-01 -2.77539223e-01 6.41483366e-01 3.65283433e-03
-1.88515455e-01 -1.31690934e-01 2.79532522e-01 -3.62368673e-01
3.05083245e-01 -1.10801600e-01 6.58159971e-01 -4.26963955e-01
8.36228669e-01 4.31265026e-01 3.62074137e-01 2.25186467e-01
-2.00147361e-01 2.98967701e-03 -1.60718575e-01 -1.43730700e-01
-7.55668819e-01 -5.86736917e-01 -4.80089515e-01 4.30661678e-01
3.42356533e-01 4.55754906e-01 -2.94455647e-01 -1.92155853e-01
1.84228048e-02 6.98459506e-01 -4.79809731e-01 -1.22363247e-01
1.87308222e-01 -1.28256679e+00 1.30353495e-01 -9.42347497e-02
9.24519658e-01 -7.45794237e-01 -7.27542877e-01 -2.19732639e-03
-5.29619716e-02 -5.57740629e-01 4.31891352e-01 3.20387334e-01
-1.03410923e+00 -1.28143799e+00 -2.96903074e-01 -3.78355324e-01
4.49834794e-01 7.60709286e-01 2.90919781e-01 -2.77978778e-01
-7.16728151e-01 -4.33899127e-02 -6.99347138e-01 -5.82469940e-01
-1.24459460e-01 -1.95722267e-01 -2.20985368e-01 2.76945710e-01
4.07017767e-01 -5.53951085e-01 -5.91198564e-01 2.59882897e-01
-1.00067639e+00 -1.76052660e-01 9.69197690e-01 9.15910423e-01
7.28383362e-01 8.33317995e-01 3.33487213e-01 -4.56572682e-01
3.72900009e-01 -3.86154175e-01 -1.04026902e+00 2.97683507e-01
-6.81179345e-01 -2.25705713e-01 3.52485418e-01 9.77271469e-04
-1.07470179e+00 3.09018254e-01 5.20730317e-01 3.98191325e-02
-2.38335148e-01 1.05346239e+00 -2.94141024e-01 -6.02527499e-01
9.32974219e-01 5.34149945e-01 1.81100562e-01 -3.11334252e-01
-6.93707690e-02 7.46438503e-01 2.52431989e-01 -7.42323846e-02
7.08365917e-01 4.56672519e-01 5.09461820e-01 -1.24730313e+00
-5.07431030e-01 -9.58465397e-01 -7.02060878e-01 -3.31692547e-01
7.71138489e-01 -7.25569427e-01 -6.31515503e-01 3.30183476e-01
-7.38454103e-01 4.84613568e-01 5.57778291e-02 9.08414066e-01
-6.48867711e-02 4.75307077e-01 8.99156630e-02 -1.20830703e+00
-5.27436793e-01 -1.26527762e+00 5.24263680e-01 4.51446384e-01
3.22228432e-01 -6.28522396e-01 -3.13761741e-01 5.08836567e-01
3.32048625e-01 5.48900783e-01 9.19458747e-01 -5.91803551e-01
-5.34904480e-01 -4.20687467e-01 -4.32575911e-01 7.31347680e-01
5.88469148e-01 1.45909011e-01 -7.91402400e-01 -1.54712737e-01
-2.01269379e-03 -6.77041039e-02 8.46289337e-01 5.27105272e-01
7.33205318e-01 -2.02961937e-01 -3.49838376e-01 4.24016178e-01
2.22691584e+00 7.94804573e-01 5.22891760e-01 7.54633665e-01
3.48830432e-01 7.39293814e-01 9.38491106e-01 7.27184057e-01
-5.33616304e-01 2.31977180e-01 1.01693952e+00 -4.20832895e-02
4.37341422e-01 4.97050375e-01 -1.85340255e-01 1.65651813e-01
-5.16613901e-01 -5.32873124e-02 -8.51140857e-01 3.20665121e-01
-1.52173686e+00 -1.35591567e+00 -4.44617331e-01 2.48692060e+00
3.60534966e-01 -2.54282773e-01 -9.28713009e-02 8.78382087e-01
9.23454523e-01 -6.45694211e-02 -3.43132019e-01 2.31539786e-01
-7.00843275e-01 2.31391102e-01 9.65605497e-01 3.96760434e-01
-1.44266462e+00 5.28194666e-01 4.45533133e+00 7.35310555e-01
-1.41834116e+00 -2.21133813e-01 4.25335050e-01 7.35065863e-02
6.78328052e-03 6.04582578e-02 -4.91899490e-01 1.46370962e-01
4.65204775e-01 -2.76117623e-01 4.34915930e-01 6.24910474e-01
5.03014743e-01 -6.63545787e-01 -2.19699636e-01 9.55093265e-01
-3.24630807e-03 -1.09800935e+00 2.84582108e-01 2.16457188e-01
8.95403087e-01 -3.95292193e-02 -2.36228704e-02 -5.71888506e-01
-2.38187507e-01 -8.16870570e-01 3.57032388e-01 7.56010592e-01
1.89545020e-01 -9.90021348e-01 9.66699362e-01 3.62809002e-01
-1.19723046e+00 -5.01361072e-01 -4.49958950e-01 3.99015658e-02
-3.31577092e-01 7.95223117e-01 -7.81050026e-01 1.01401067e+00
6.00763261e-01 5.71877420e-01 -5.40892363e-01 1.36337602e+00
5.02884053e-02 3.89311105e-01 -3.02844226e-01 -1.91772714e-01
2.64349461e-01 -7.92071998e-01 7.86666274e-01 7.60533750e-01
6.05715096e-01 3.93295556e-01 6.72932938e-02 6.34666204e-01
5.82134187e-01 8.19233775e-01 -6.55721545e-01 -3.48520637e-01
5.85496604e-01 1.22048128e+00 -8.62778902e-01 -1.30692407e-01
-2.36793861e-01 6.18334234e-01 -4.78760451e-01 2.01340824e-01
-5.02030432e-01 -4.32203948e-01 4.55209762e-01 -1.44869119e-01
1.38588071e-01 -8.97506997e-03 -4.01190370e-01 -6.14979982e-01
-1.70488060e-01 -8.49433839e-01 5.98674655e-01 -6.99997425e-01
-1.02053869e+00 6.76420212e-01 1.12491280e-01 -1.55023909e+00
3.30623597e-01 -8.37380707e-01 -7.65881389e-02 1.25570118e+00
-1.40499473e+00 -1.01124537e+00 -8.05529833e-01 5.37780285e-01
2.55000144e-01 -4.00869876e-01 9.33091998e-01 2.01026127e-02
-5.64053357e-01 -2.70789891e-01 6.22637987e-01 -3.42846304e-01
2.77593374e-01 -9.15726900e-01 -1.03829825e+00 9.51306999e-01
-3.23763728e-01 1.86846092e-01 7.47622609e-01 -7.80944586e-01
-1.17287052e+00 -9.03936625e-01 3.32464486e-01 6.12916410e-01
5.87666750e-01 6.61392570e-01 -6.53301656e-01 1.97714508e-01
1.09038532e-01 -2.57684588e-01 1.08619976e+00 -4.01286006e-01
-3.95755805e-02 -5.32504499e-01 -1.26304340e+00 3.22275847e-01
3.00880194e-01 -1.47127852e-01 -1.69171944e-01 5.61489224e-01
4.44062464e-02 3.26414347e-01 -9.58143055e-01 7.76347220e-01
5.89275420e-01 -1.22691512e+00 9.30822849e-01 -2.22650841e-01
1.07651234e-01 -6.27918541e-01 -4.87372816e-01 -1.22544491e+00
-4.23761100e-01 2.11300269e-01 8.25326324e-01 8.68894100e-01
5.73672116e-01 -7.10225999e-01 6.76018834e-01 2.32150838e-01
-2.67806742e-02 -3.57056201e-01 -7.91087270e-01 -8.02399695e-01
-5.17201662e-01 1.78950597e-02 4.24260318e-01 1.06253827e+00
-9.78713483e-03 5.93396723e-02 -2.00787023e-01 6.41143680e-01
9.32814360e-01 4.11454350e-01 2.67928571e-01 -1.83721471e+00
-1.44034609e-01 -7.35215008e-01 -7.94163704e-01 1.83514789e-01
-1.91533431e-01 -8.21994662e-01 -2.36770436e-01 -1.57770228e+00
2.45937467e-01 -2.02107608e-01 -3.95012915e-01 4.47876573e-01
-5.99967986e-02 2.69897550e-01 5.25963120e-02 3.69735807e-01
3.04629952e-01 3.56318921e-01 1.00695837e+00 -4.78906661e-01
-4.90994811e-01 9.86508280e-02 -5.17469406e-01 4.59117651e-01
9.66255009e-01 -2.80939847e-01 -5.20815194e-01 1.92889452e-01
6.90471134e-05 5.26849413e-03 3.49888265e-01 -1.36235189e+00
4.29866500e-02 -5.49749315e-01 5.00655830e-01 -7.12320507e-01
3.75573426e-01 -1.14137173e+00 5.91757476e-01 6.38953984e-01
6.47483468e-02 -4.26640034e-01 3.15491766e-01 3.56557786e-01
-4.82746094e-01 -4.60771948e-01 1.00276339e+00 -2.85811603e-01
-1.07900667e+00 2.70617455e-02 -2.45011225e-01 -1.02545726e+00
1.35427403e+00 -6.34226382e-01 -4.43394929e-01 -3.45699415e-02
-7.58446693e-01 -3.83855253e-01 1.99705660e-01 -6.15115203e-02
5.42479455e-01 -8.61159205e-01 -8.43292415e-01 -3.85644324e-02
3.06314558e-01 -5.63704550e-01 4.97820646e-01 9.17131722e-01
-6.99741960e-01 4.39361334e-01 -7.07640827e-01 -5.59637487e-01
-1.77017510e+00 2.89670020e-01 4.07379538e-01 6.49330840e-02
1.68449596e-01 6.91593647e-01 -1.21568963e-01 1.18741192e-01
-1.67477965e-01 9.86937582e-02 -8.66420150e-01 4.01393205e-01
3.05375367e-01 6.54533327e-01 3.65998507e-01 -1.02543628e+00
-3.66606027e-01 6.10855579e-01 5.28468907e-01 5.56191604e-04
1.50415432e+00 -2.48202472e-03 -5.26310146e-01 2.48822808e-01
1.00674975e+00 -2.59922683e-01 -5.23741663e-01 1.79091096e-02
6.44597039e-03 -7.58866489e-01 5.39639950e-01 -1.09030020e+00
-8.63521397e-01 6.97440922e-01 1.23704863e+00 3.28784317e-01
1.64891744e+00 -5.20185113e-01 -1.81971699e-01 4.43373829e-01
1.60996035e-01 -1.09726167e+00 -2.65355200e-01 8.67051631e-02
1.00217867e+00 -1.33965480e+00 3.49650115e-01 -8.39278460e-01
-4.31698769e-01 1.35324419e+00 4.23313022e-01 2.35442668e-01
7.50658035e-01 -2.94275701e-01 -1.01069748e-01 -1.50734216e-01
-1.36778289e-02 -6.67062163e-01 4.03407723e-01 5.16997993e-01
5.27254045e-01 3.41870010e-01 -8.52788806e-01 2.48399451e-01
5.42605408e-02 -2.00080454e-01 2.57063538e-01 9.84261096e-01
-1.05282009e+00 -8.26183915e-01 -9.01186466e-01 7.24156082e-01
-1.05773702e-01 -7.42058828e-02 -4.96265173e-01 8.48835051e-01
2.25083947e-01 1.16204226e+00 -5.14488041e-01 -5.38392544e-01
2.95297712e-01 1.75241046e-02 3.31964493e-01 -3.78700852e-01
-3.42491090e-01 1.84963942e-01 -6.34326264e-02 -7.24447817e-02
-8.86003256e-01 -7.03302741e-01 -9.68625307e-01 1.77483797e-01
-8.52735639e-01 4.56782877e-01 1.18490994e+00 9.78334010e-01
7.38996193e-02 2.48783633e-01 8.54277372e-01 -9.71381664e-01
-5.48161566e-01 -1.10092783e+00 -1.07113433e+00 2.77880758e-01
-1.68000236e-01 -8.59563708e-01 -6.47896528e-01 -1.68007668e-02] | [9.74095630645752, -1.8030469417572021] |
2f24ba0f-e1ea-46df-852d-54e40098be12 | fcc-fusing-conversation-history-and-candidate | 2304.00180 | null | https://arxiv.org/abs/2304.00180v1 | https://arxiv.org/pdf/2304.00180v1.pdf | FCC: Fusing Conversation History and Candidate Provenance for Contextual Response Ranking in Dialogue Systems | Response ranking in dialogues plays a crucial role in retrieval-based conversational systems. In a multi-turn dialogue, to capture the gist of a conversation, contextual information serves as essential knowledge to achieve this goal. In this paper, we present a flexible neural framework that can integrate contextual information from multiple channels. Specifically for the current task, our approach is to provide two information channels in parallel, Fusing Conversation history and domain knowledge extracted from Candidate provenance (FCC), where candidate responses are curated, as contextual information to improve the performance of multi-turn dialogue response ranking. The proposed approach can be generalized as a module to incorporate miscellaneous contextual features for other context-oriented tasks. We evaluate our model on the MSDialog dataset widely used for evaluating conversational response ranking tasks. Our experimental results show that our framework significantly outperforms the previous state-of-the-art models, improving Recall@1 by 7% and MAP by 4%. Furthermore, we conduct ablation studies to evaluate the contributions of each information channel, and of the framework components, to the overall ranking performance, providing additional insights and directions for further improvements. | ['Jinho Choi', 'Eugene Agichtein', 'ZiHao Wang'] | 2023-03-31 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [ 2.47497588e-01 4.69724089e-02 -2.24487141e-01 -6.95762992e-01
-1.16906488e+00 -5.72650254e-01 1.03959942e+00 1.17275789e-01
-6.13824487e-01 8.18483353e-01 8.67226541e-01 -7.93209746e-02
-8.74381065e-02 -5.95269799e-01 -8.82773474e-02 -3.20339441e-01
1.55951008e-01 6.26577854e-01 4.18156564e-01 -9.18868303e-01
6.24469340e-01 -5.07142656e-02 -1.19595432e+00 1.11454690e+00
9.41612005e-01 8.31989408e-01 2.84229457e-01 8.05531323e-01
-2.50034779e-01 1.21201956e+00 -7.99949825e-01 -4.20898974e-01
-3.61417651e-01 -5.75013101e-01 -1.41541660e+00 -3.51180106e-01
3.77978794e-02 -5.58865666e-01 -3.58709157e-01 3.85450006e-01
6.83340132e-01 4.92215574e-01 5.04641652e-01 -7.00374067e-01
-3.80917132e-01 8.86224926e-01 -1.03206355e-02 9.83736590e-02
7.68486559e-01 -6.86611012e-02 1.17018759e+00 -8.85276377e-01
5.71066380e-01 1.50982499e+00 2.15432465e-01 8.05882275e-01
-8.54899466e-01 -1.90667152e-01 3.93606246e-01 3.08283627e-01
-7.34415412e-01 -6.93686247e-01 8.19161057e-01 -1.77525684e-01
9.55257773e-01 5.31944513e-01 2.41390079e-01 1.21123815e+00
-2.01448172e-01 1.23828495e+00 1.06503284e+00 -5.02795041e-01
-3.89045514e-02 2.73999572e-01 5.89535534e-01 3.26020479e-01
-7.24551618e-01 -2.40476221e-01 -8.31657588e-01 -4.10979182e-01
2.08943665e-01 -1.66626036e-01 -4.55844194e-01 3.04838777e-01
-9.48519468e-01 9.47244883e-01 5.35378337e-01 2.78227776e-01
-2.19468638e-01 -9.93060172e-02 5.57337224e-01 5.61685979e-01
7.19028592e-01 6.58471882e-01 -3.41674030e-01 -6.09114110e-01
-4.31686133e-01 4.02689248e-01 1.19291580e+00 6.32397115e-01
6.21267974e-01 -5.63954830e-01 -8.66598547e-01 1.56120765e+00
1.29380777e-01 1.22302398e-01 4.01888281e-01 -9.35611963e-01
8.28590751e-01 8.78917933e-01 3.24380249e-01 -7.44792521e-01
-5.69838405e-01 -1.69074163e-01 -4.80367571e-01 -7.75665402e-01
3.78233492e-01 -2.77374148e-01 -3.57377231e-01 1.74519563e+00
2.92753875e-01 -1.45054281e-01 1.91074699e-01 1.10067928e+00
1.26103842e+00 6.89451337e-01 -7.78646767e-02 -1.55276239e-01
1.32546699e+00 -1.26089346e+00 -6.33717358e-01 -2.73904622e-01
6.11597538e-01 -9.58543956e-01 1.27449501e+00 7.80533627e-02
-8.66828322e-01 -3.15567046e-01 -7.11412251e-01 -9.01343226e-02
-4.84852232e-02 1.44472301e-01 6.20406449e-01 2.57859766e-01
-1.03404796e+00 2.63732076e-01 -2.48417661e-01 -3.79953772e-01
-2.34426513e-01 1.45075992e-01 4.30047885e-02 -9.93627161e-02
-1.86285031e+00 1.14236462e+00 3.22777480e-02 3.19336981e-01
-7.25262165e-01 -2.12469161e-01 -5.36094725e-01 9.31466147e-02
4.40859705e-01 -5.18979728e-01 1.70198059e+00 -4.84953195e-01
-1.91987884e+00 5.47600508e-01 -4.90753949e-01 -3.36406708e-01
4.49288189e-01 -3.40088576e-01 -1.44825339e-01 1.78788632e-01
-1.80208400e-01 4.34158653e-01 1.89257458e-01 -1.10156608e+00
-7.37159371e-01 -2.18300223e-01 7.43544519e-01 7.21576035e-01
-4.18175757e-01 1.54210746e-01 -6.34077489e-01 -1.15423866e-01
-1.08752698e-01 -1.11922407e+00 -5.49532436e-02 -8.91758204e-01
-3.63132417e-01 -7.98937976e-01 4.83931929e-01 -5.85895360e-01
1.42296648e+00 -1.69218338e+00 2.25234807e-01 -8.06142986e-02
1.29253387e-01 2.12223813e-01 -2.44589061e-01 1.02066958e+00
6.13022864e-01 -2.03794404e-03 -5.73244989e-02 -2.72144675e-01
4.50371727e-02 -2.39154130e-01 -2.65709162e-01 -1.15516640e-01
2.03065827e-01 9.96861279e-01 -1.02375388e+00 -1.22740582e-01
-7.14669675e-02 3.71428400e-01 -4.18372989e-01 4.67975438e-01
-2.39220902e-01 6.69400811e-01 -7.23491132e-01 2.51442343e-01
1.51914105e-01 -2.19502822e-01 3.96462172e-01 1.11806199e-01
1.39504120e-01 1.17316437e+00 -5.82445800e-01 1.69144571e+00
-8.77467096e-01 6.74534559e-01 6.42223284e-02 -6.24000072e-01
9.12249327e-01 3.94479066e-01 1.57082260e-01 -9.44437623e-01
-6.05859347e-02 -5.24423905e-02 9.19816419e-02 -4.47504103e-01
9.19111311e-01 3.07311088e-01 -3.61990511e-01 9.06887829e-01
-2.23182470e-01 6.33983761e-02 1.26362979e-01 6.80306256e-01
1.19702494e+00 -3.08339417e-01 1.00087032e-01 7.26976097e-02
8.05761218e-01 -9.00968388e-02 3.02994698e-01 1.01457143e+00
-8.98949653e-02 3.33694071e-01 6.99532568e-01 -3.46829802e-01
-5.74490249e-01 -3.57166260e-01 9.04060081e-02 1.70088577e+00
1.65582150e-01 -3.68150115e-01 -6.80705130e-01 -8.36318791e-01
-3.44791412e-01 6.37720525e-01 -4.87054437e-01 -1.34715751e-01
-8.01596403e-01 -7.92108357e-01 5.20828962e-01 3.47478926e-01
6.34664774e-01 -1.27146888e+00 -2.43244067e-01 3.70539427e-01
-1.03941739e+00 -1.07600272e+00 -5.20955503e-01 1.02342460e-02
-5.56740701e-01 -1.00583947e+00 -6.11115813e-01 -5.51686227e-01
9.29440930e-02 5.42841733e-01 1.35655570e+00 3.26755553e-01
3.15802664e-01 4.06878352e-01 -7.77956724e-01 -3.17986235e-02
-5.01432300e-01 4.90649521e-01 -2.43832931e-01 -5.15973493e-02
3.85669768e-01 -2.57955432e-01 -7.84799576e-01 6.67447925e-01
-5.62291920e-01 4.31809947e-02 3.86036158e-01 1.17589271e+00
-1.72075912e-01 -6.45992815e-01 1.09756553e+00 -1.29631031e+00
1.52234018e+00 -4.52930778e-01 4.89618555e-02 4.05974507e-01
-3.80867898e-01 -5.05115725e-02 4.40298140e-01 -2.57396519e-01
-1.40881741e+00 -3.88944238e-01 -2.65802115e-01 4.59051758e-01
4.02615145e-02 7.98965454e-01 -1.46725867e-02 2.47053578e-01
8.42454374e-01 2.38327816e-01 -2.35394910e-01 -4.02204335e-01
4.96544003e-01 1.22262025e+00 9.80740413e-02 -8.42682540e-01
5.39809205e-02 2.19257735e-02 -7.34942198e-01 -6.66799188e-01
-8.54190767e-01 -8.61583233e-01 -4.00411934e-01 -5.34988523e-01
3.88112813e-01 -9.03584719e-01 -9.74413514e-01 3.50561857e-01
-1.41552174e+00 -4.09383416e-01 4.89657015e-01 2.81560719e-01
-4.24983859e-01 2.53110617e-01 -1.07701302e+00 -1.08411694e+00
-5.88806510e-01 -1.21846604e+00 8.47858429e-01 2.14810938e-01
-3.02505255e-01 -9.12703693e-01 2.60173917e-01 9.00334775e-01
5.46408117e-01 -4.24025893e-01 7.93370485e-01 -1.11428690e+00
-3.74957651e-01 -2.16020003e-01 -2.12521672e-01 9.01153907e-02
-9.29291472e-02 -4.66043025e-01 -1.33368683e+00 -1.68602735e-01
3.48141864e-02 -8.78780007e-01 1.21946025e+00 -6.43778294e-02
8.02173078e-01 -5.35935581e-01 -2.30277792e-01 -2.66593933e-01
7.10168123e-01 2.87721664e-01 6.11371934e-01 3.52971315e-01
3.61547798e-01 1.14601123e+00 8.69911313e-01 4.86196309e-01
8.70218098e-01 9.80998993e-01 2.10804582e-01 8.71896744e-02
6.41862005e-02 -6.51495978e-02 4.46993470e-01 1.09088600e+00
3.16554159e-02 -5.21693826e-01 -7.56249547e-01 4.68825459e-01
-2.26220417e+00 -9.60405290e-01 2.44372841e-02 2.07587957e+00
1.10322154e+00 1.17089991e-02 2.44975820e-01 -2.87286431e-01
7.85844326e-01 3.60997885e-01 -4.07199532e-01 -6.03726149e-01
1.75347924e-01 -2.05974728e-01 -2.91855633e-01 8.31690311e-01
-9.60560501e-01 1.09442282e+00 6.16118431e+00 6.01469398e-01
-9.10543919e-01 9.28513184e-02 6.90544009e-01 -1.08029926e-02
-3.54503900e-01 -9.96275470e-02 -8.33435774e-01 3.27866971e-01
9.91215229e-01 -1.17924921e-01 6.33653164e-01 6.29880846e-01
3.23034614e-01 -1.41264021e-01 -1.26336884e+00 4.99909848e-01
1.50965169e-01 -1.12522757e+00 -7.43556991e-02 -2.29707032e-01
4.17496711e-01 4.93677557e-02 -7.80445784e-02 7.33708084e-01
5.69526494e-01 -6.95074260e-01 9.04210582e-02 4.37909424e-01
4.12375063e-01 -7.13265359e-01 1.04897523e+00 5.09010434e-01
-7.81178892e-01 -6.76885545e-02 -2.72195607e-01 -1.98149502e-01
5.07133417e-02 3.63824904e-01 -1.45400143e+00 5.09065449e-01
3.29085857e-01 4.99476671e-01 -3.24191213e-01 6.99278474e-01
-3.54356825e-01 5.77255785e-01 -9.61489826e-02 -4.96473104e-01
4.29708093e-01 5.33260331e-02 4.11095619e-01 1.44675434e+00
-2.31957242e-01 3.84384692e-01 3.92190665e-01 4.91146743e-01
-3.92105639e-01 3.31971943e-01 -4.32897002e-01 1.62322178e-01
9.27223980e-01 1.28194511e+00 -2.20494837e-01 -2.73566723e-01
-3.68919760e-01 8.12997401e-01 7.21267939e-01 4.11878794e-01
-5.33064783e-01 -3.81645322e-01 5.26682496e-01 -2.45126888e-01
-1.82823315e-01 1.14462540e-01 9.18269679e-02 -1.03733623e+00
8.61062482e-02 -9.42246020e-01 4.59444702e-01 -4.20884877e-01
-1.24805975e+00 8.65368545e-01 -2.01581672e-01 -1.07453966e+00
-8.51030946e-01 -2.02216968e-01 -6.32574856e-01 9.80055034e-01
-1.64863956e+00 -1.14585948e+00 -2.67097414e-01 4.93529677e-01
9.73261356e-01 -1.24465093e-01 9.30954814e-01 2.14976802e-01
-4.78219658e-01 7.26671696e-01 1.04483351e-01 2.97386944e-01
1.15305722e+00 -1.13723063e+00 1.84165359e-01 4.33462858e-01
6.49398984e-03 9.64568973e-01 3.93077463e-01 -3.82844537e-01
-1.25244141e+00 -7.75718689e-01 1.06927824e+00 -5.91596961e-01
4.34188724e-01 -3.32591951e-01 -8.81232023e-01 3.82252008e-01
4.33077812e-01 -7.73632288e-01 8.63131225e-01 9.21088398e-01
-2.42708594e-01 -2.06591226e-02 -7.65389144e-01 5.65967441e-01
7.62682080e-01 -8.36987734e-01 -7.29647040e-01 3.08897972e-01
8.72990310e-01 -5.34475088e-01 -7.60624111e-01 2.30673417e-01
6.11004353e-01 -9.46954787e-01 8.39692354e-01 -6.19967997e-01
6.86780572e-01 1.97284639e-01 -3.56655940e-02 -1.62156022e+00
-4.00635675e-02 -6.51250064e-01 -3.37940156e-02 1.29239237e+00
6.47865832e-01 -4.65411723e-01 4.06310409e-01 7.17006028e-01
-2.84486711e-01 -6.86815321e-01 -7.94035196e-01 -5.03062867e-02
8.82404745e-02 -2.59502679e-01 5.10741830e-01 8.93235981e-01
6.69179857e-01 1.11822295e+00 -6.40536606e-01 -3.87557089e-01
-1.32701606e-01 3.11409980e-01 9.91034746e-01 -1.16590798e+00
-4.46500033e-02 -5.24762273e-01 3.15746188e-01 -1.68949115e+00
3.10923219e-01 -8.14979970e-01 3.19960237e-01 -1.62465942e+00
4.87706363e-01 -4.87135530e-01 -2.88188010e-01 3.30018550e-01
-6.17219448e-01 -4.02607657e-02 2.75859833e-01 3.55043948e-01
-1.09093571e+00 7.35483527e-01 1.26655865e+00 -2.66507596e-01
-5.55191636e-01 2.20887005e-01 -9.50241268e-01 4.29587603e-01
5.98538220e-01 -2.42380142e-01 -5.82017422e-01 -5.74386060e-01
1.74071372e-01 6.26415253e-01 -6.05557226e-02 -4.68412280e-01
6.31277502e-01 -1.39267445e-01 -8.59395564e-02 -6.08652830e-01
7.09257841e-01 -3.17488939e-01 -6.14791214e-01 -3.14749964e-02
-1.12568402e+00 -8.13025311e-02 -1.34144783e-01 6.14857733e-01
-4.50573742e-01 -1.37628332e-01 2.88794011e-01 -2.27543950e-01
-5.83038330e-01 -1.12663805e-01 -5.00988364e-01 2.35948235e-01
2.84416199e-01 2.63114423e-01 -7.92500317e-01 -1.01138401e+00
-6.34080172e-01 6.60681367e-01 7.72744715e-02 7.22816706e-01
7.15997517e-01 -1.08077991e+00 -9.45083499e-01 -3.50906998e-01
4.01205599e-01 -3.10272753e-01 3.71050209e-01 7.40402341e-01
3.33001837e-02 8.71929884e-01 1.03232838e-01 -5.50211787e-01
-1.34375131e+00 -9.65427011e-02 2.16438860e-01 -6.70752347e-01
-3.47312957e-01 7.90604174e-01 1.49340525e-01 -9.21192050e-01
3.86092126e-01 5.18922731e-02 -8.33170474e-01 2.96892792e-01
9.25267577e-01 2.95989215e-01 2.68388093e-01 -3.21409583e-01
-2.36940190e-01 1.89527161e-02 -5.43938518e-01 -4.51259434e-01
1.25008106e+00 -4.11014289e-01 -2.31135204e-01 5.32824576e-01
9.41306174e-01 -5.19672185e-02 -8.26220274e-01 -7.51331449e-01
1.88112527e-01 -2.82936364e-01 -1.01398520e-01 -1.38597846e+00
-5.81588984e-01 8.89764726e-01 1.35166690e-01 2.75863975e-01
8.16448748e-01 -1.21935727e-02 6.59790933e-01 8.71040881e-01
3.47646892e-01 -1.11225057e+00 3.35689187e-01 1.02207947e+00
1.07298636e+00 -1.47702706e+00 -2.97189832e-01 -2.63578981e-01
-1.15131545e+00 9.98178363e-01 8.17687452e-01 3.10657412e-01
1.58488065e-01 -2.65021920e-01 2.78616279e-01 -1.93975911e-01
-1.31066394e+00 -2.36103907e-01 4.13999826e-01 1.61590114e-01
9.16042387e-01 3.46545354e-02 -6.98157728e-01 6.89118981e-01
1.93679053e-02 -3.37003648e-01 4.09226656e-01 7.89211631e-01
-5.36612630e-01 -1.37705183e+00 -3.69754136e-02 4.31811690e-01
-4.10657614e-01 -3.79158944e-01 -1.04848146e+00 5.13262331e-01
-8.08661163e-01 1.63560641e+00 -2.71354765e-01 -7.30658293e-01
4.03542161e-01 3.46362621e-01 2.39532981e-02 -7.01485097e-01
-1.09280622e+00 1.30062047e-02 8.74402344e-01 -4.99096513e-01
-6.06066465e-01 -3.97598714e-01 -8.70925426e-01 -3.68895769e-01
-5.70441961e-01 5.40932119e-01 5.05256057e-01 1.04866374e+00
3.89254332e-01 5.44946611e-01 1.03817546e+00 -4.70164061e-01
-6.48742139e-01 -1.59828830e+00 6.62343279e-02 3.37907344e-01
1.99924842e-01 -5.47373593e-01 -2.19351396e-01 -5.16344607e-01] | [12.426114082336426, 7.887401580810547] |
ef5f5c75-a10f-4541-aca2-0907876e27e8 | fast-3d-registration-with-accurate | 2112.03053 | null | https://arxiv.org/abs/2112.03053v1 | https://arxiv.org/pdf/2112.03053v1.pdf | Fast 3D registration with accurate optimisation and little learning for Learn2Reg 2021 | Current approaches for deformable medical image registration often struggle to fulfill all of the following criteria: versatile applicability, small computation or training times, and the being able to estimate large deformations. Furthermore, end-to-end networks for supervised training of registration often become overly complex and difficult to train. For the Learn2Reg2021 challenge, we aim to address these issues by decoupling feature learning and geometric alignment. First, we introduce a new very fast and accurate optimisation method. By using discretised displacements and a coupled convex optimisation procedure, we are able to robustly cope with large deformations. With the help of an Adam-based instance optimisation, we achieve very accurate registration performances and by using regularisation, we obtain smooth and plausible deformation fields. Second, to be versatile for different registration tasks, we extract hand-crafted features that are modality and contrast invariant and complement them with semantic features from a task-specific segmentation U-Net. With our results we were able to achieve the overall Learn2Reg2021 challenge's second place, winning Task 1 and being second and third in the other two tasks. | ['Mattias P. Heinrich', 'Lasse Hansen', 'Hanna Siebert'] | 2021-12-06 | null | null | null | null | ['deformable-medical-image-registration'] | ['medical'] | [ 3.67603362e-01 2.17313379e-01 2.53976941e-01 -4.15892810e-01
-1.31777537e+00 -4.51210618e-01 7.45941579e-01 4.44095105e-01
-7.92939246e-01 6.47736132e-01 1.07194282e-01 1.42052621e-01
-3.55189115e-01 -4.14896965e-01 -6.17019594e-01 -6.19525850e-01
-1.34216011e-01 8.12422812e-01 4.08705413e-01 -4.82330054e-01
1.47788391e-01 5.79552293e-01 -1.27045739e+00 -1.47887349e-01
7.73445725e-01 1.02243459e+00 1.65707320e-01 4.57162917e-01
3.83665055e-01 2.46792868e-01 7.98212141e-02 -3.43050867e-01
4.05090690e-01 -3.60268623e-01 -1.21797359e+00 -1.31206840e-01
5.11324525e-01 -6.66519068e-03 1.89306244e-01 8.27703297e-01
9.08797801e-01 2.81522840e-01 5.88815928e-01 -7.53169298e-01
-1.50238499e-02 3.23221922e-01 -2.88042724e-01 5.21379411e-02
3.82768840e-01 1.84569299e-01 7.61439502e-01 -7.00797856e-01
1.00881600e+00 8.60563278e-01 1.00359809e+00 6.53862596e-01
-1.45826483e+00 1.28130436e-01 -1.10362925e-01 -1.99629992e-01
-1.03708506e+00 -6.58799946e-01 6.02910519e-01 -5.58196902e-01
8.69526327e-01 5.63173890e-01 4.56319839e-01 9.13318515e-01
7.26415813e-02 6.24582946e-01 1.18442321e+00 -2.36070216e-01
3.01366635e-02 -1.91261649e-01 -3.46352220e-01 6.03933036e-01
-2.28050843e-01 1.24377966e-01 8.78460109e-02 -2.11966038e-02
7.83636451e-01 -7.25926235e-02 -4.84870344e-01 -5.00854850e-01
-1.79646337e+00 6.18697524e-01 6.89602494e-01 6.14278257e-01
-4.36870933e-01 2.12781549e-01 5.09739280e-01 1.53091267e-01
5.98813057e-01 6.83024168e-01 -4.66181606e-01 -1.66085437e-01
-1.12839544e+00 3.77916455e-01 5.79275131e-01 3.45466822e-01
5.48801720e-01 -2.65242219e-01 -3.74325454e-01 7.63076663e-01
1.77418053e-01 2.12547064e-01 5.53230941e-01 -8.20583284e-01
3.13300252e-01 3.44933629e-01 -7.61881769e-02 -8.93679440e-01
-7.54303336e-01 -3.64508897e-01 -1.06436551e+00 3.89148384e-01
6.44320309e-01 1.06811345e-01 -1.04443884e+00 1.64892721e+00
5.24206996e-01 1.04110345e-01 -7.67984986e-02 9.93620634e-01
6.83207393e-01 5.87381073e-04 1.85237899e-01 -1.01224259e-01
1.30243671e+00 -7.22586751e-01 -2.56299525e-01 -1.15946181e-01
8.79428566e-01 -9.27982569e-01 9.67428803e-01 3.09218615e-01
-1.52874219e+00 -4.17942226e-01 -7.89957523e-01 -9.89303440e-02
-1.16301343e-01 -1.52138308e-01 3.76929909e-01 2.71528780e-01
-1.23757315e+00 1.28902924e+00 -1.16303349e+00 -1.55350596e-01
5.80666125e-01 7.10382640e-01 -7.43006229e-01 1.54319778e-02
-8.70603621e-01 1.24142623e+00 3.14970732e-01 4.03165102e-01
-3.90084177e-01 -7.76605785e-01 -1.01273370e+00 -3.62044692e-01
1.59220263e-01 -8.19396853e-01 7.63096333e-01 -8.97946656e-01
-1.54530883e+00 1.30252755e+00 3.29970330e-01 -3.92963082e-01
1.14880240e+00 -1.05944194e-01 3.98065448e-02 6.21413514e-02
1.47611713e-02 6.31060898e-01 5.80675244e-01 -1.08397913e+00
-5.66348759e-03 -3.42323512e-01 -4.76228833e-01 9.25382078e-02
-1.43270269e-01 1.61031261e-01 -3.03222597e-01 -7.93293238e-01
1.28371224e-01 -1.03032613e+00 -6.52894557e-01 -2.27331892e-02
-2.99605399e-01 3.59489396e-02 4.43344265e-01 -9.03720021e-01
5.58242738e-01 -2.03190613e+00 6.12759769e-01 3.74109566e-01
4.32106018e-01 4.59373295e-01 -1.92221105e-01 2.18648575e-02
-7.66744539e-02 -1.04839087e-01 -6.71016753e-01 -6.38338089e-01
9.55583826e-02 1.84604257e-01 1.65367395e-01 7.86727309e-01
3.88208508e-01 1.30423641e+00 -9.15429890e-01 -4.21320498e-01
3.15429986e-01 7.62510836e-01 -5.30285358e-01 2.53205985e-01
-2.20400617e-01 1.25628626e+00 -4.15553749e-01 2.57758796e-01
5.17112255e-01 -4.28323485e-02 -9.88844708e-02 -3.09613585e-01
-1.24323908e-02 5.06413057e-02 -1.25356054e+00 2.13586164e+00
-5.38385153e-01 2.28407368e-01 2.55166262e-01 -1.25306320e+00
8.25898230e-01 5.34729958e-01 8.74022484e-01 -6.98076487e-01
4.12759215e-01 7.58991122e-01 -3.36284310e-01 -4.67998296e-01
1.18215904e-02 -3.53145242e-01 6.65574521e-02 3.33453625e-01
2.34151930e-01 -4.09754336e-01 -3.00223716e-02 -2.75350720e-01
1.01343894e+00 4.58880544e-01 4.70938459e-02 -3.91441345e-01
6.89285576e-01 -3.45339417e-01 3.08605462e-01 2.92592824e-01
-1.50240567e-02 1.14996576e+00 2.15385139e-01 -7.04511821e-01
-9.96807039e-01 -8.65982473e-01 -2.33986303e-01 7.99671531e-01
-6.17418475e-02 4.63564927e-03 -7.23002195e-01 -6.90850735e-01
-2.72267938e-01 4.06541489e-02 -6.23788238e-01 3.58533822e-02
-9.90381360e-01 -8.16263318e-01 3.51897925e-01 4.48657155e-01
1.88850135e-01 -1.16970754e+00 -6.38942957e-01 5.26249766e-01
-1.24923505e-01 -1.21158016e+00 -4.71996546e-01 6.16479181e-02
-1.01524115e+00 -9.02408719e-01 -1.21384609e+00 -8.26618016e-01
7.17903495e-01 -4.44828629e-01 1.22808325e+00 4.12881494e-01
-5.97702682e-01 3.32153916e-01 -1.42866969e-01 -1.27362534e-01
-4.03409570e-01 3.64866108e-01 -2.15033054e-01 4.86894324e-02
-4.94068325e-01 -7.76041806e-01 -7.23091185e-01 2.93750912e-01
-9.65017796e-01 -1.38772950e-01 5.21775126e-01 7.70372391e-01
7.33384728e-01 -5.72201729e-01 2.01925740e-01 -9.48066235e-01
4.06962037e-01 -5.15565909e-02 -4.46181536e-01 2.06167176e-01
-4.77167487e-01 2.78196752e-01 2.95041084e-01 -3.88192147e-01
-5.83293021e-01 4.75866765e-01 -6.62169337e-01 -1.41565472e-01
-2.70718187e-01 3.54293495e-01 4.18037921e-02 -5.22210300e-01
8.78304899e-01 7.84233212e-02 3.48535836e-01 -5.22433162e-01
4.20432448e-01 1.29061744e-01 6.95058346e-01 -6.08616292e-01
8.72138798e-01 5.64941764e-01 3.46299559e-01 -5.38384676e-01
-6.79822266e-01 -3.86388123e-01 -1.09184098e+00 -1.67168081e-01
1.06446099e+00 -6.42648399e-01 -5.81559062e-01 5.36038637e-01
-1.00147557e+00 -5.28009355e-01 -5.69293261e-01 5.75191259e-01
-8.83844018e-01 4.85746294e-01 -4.06879067e-01 -2.74219066e-01
-7.05851853e-01 -1.41002476e+00 1.45693433e+00 -1.27829790e-01
-2.63501257e-01 -1.37540579e+00 3.29776704e-01 4.46275026e-01
8.36798966e-01 1.15192807e+00 4.33846742e-01 -6.91456318e-01
-3.94855767e-01 -4.70845342e-01 -5.50609231e-02 2.68401921e-01
7.17348307e-02 -3.15171838e-01 -8.83578420e-01 -4.79396135e-01
4.89055328e-02 -3.45382780e-01 9.47248340e-01 3.05724293e-01
9.91221070e-01 -1.89151376e-01 -2.06714138e-01 8.47691894e-01
1.28784442e+00 -6.08146727e-01 6.09021127e-01 2.87498444e-01
8.53922069e-01 7.00187504e-01 3.35249096e-01 -6.12730347e-02
3.71198595e-01 1.09166527e+00 4.99865711e-01 -4.62689966e-01
-1.80404827e-01 3.02269191e-01 -3.92906629e-02 7.22284317e-01
-7.92444289e-01 3.70887011e-01 -1.03106952e+00 6.95508301e-01
-1.94722819e+00 -6.27322316e-01 -3.24975252e-01 2.41111279e+00
1.13115454e+00 2.20028296e-01 2.33994499e-01 4.79651764e-02
2.67360330e-01 -1.17429234e-02 -1.17517829e-01 3.15130986e-02
1.58856705e-01 5.80324709e-01 4.12090361e-01 6.58475935e-01
-1.27460873e+00 6.55199468e-01 5.54327869e+00 6.49783909e-01
-1.38185871e+00 2.40322962e-01 5.39794266e-01 3.21869440e-02
-3.95734936e-01 -9.17073339e-02 -3.19972783e-01 3.01517397e-01
7.95083642e-01 4.59955812e-01 4.18949068e-01 1.97447211e-01
2.69153398e-02 8.68751034e-02 -1.20391405e+00 7.53533721e-01
-8.27267915e-02 -1.40825868e+00 -3.15596044e-01 5.11680394e-02
6.82169557e-01 3.29591006e-01 -6.95404038e-02 -8.40977356e-02
6.01419024e-02 -1.41428101e+00 5.55978656e-01 7.67289162e-01
7.26119459e-01 -4.50237036e-01 7.73261368e-01 2.40631700e-01
-8.86440873e-01 4.27869081e-01 -4.48892415e-02 3.43022108e-01
4.25064564e-01 5.97744226e-01 -5.03053427e-01 8.92199874e-01
3.91795337e-01 7.14475393e-01 -3.14226866e-01 1.16487265e+00
-1.23205576e-02 2.83308867e-02 -4.75672573e-01 3.96516174e-01
2.60040343e-01 -8.31514150e-02 5.73334098e-01 1.17350078e+00
2.67807990e-01 -1.69787571e-01 1.31248236e-01 7.01623142e-01
-9.12787169e-02 2.02451825e-01 -3.34486961e-01 4.28647876e-01
-4.06457037e-01 1.54467940e+00 -9.32992578e-01 6.29140139e-02
6.03061765e-02 1.08123970e+00 3.77970964e-01 -9.62712616e-02
-4.89667684e-01 -1.11829877e-01 2.47212783e-01 2.50961006e-01
1.80833831e-01 -3.54828537e-01 -1.37781501e-01 -1.23403764e+00
2.37157211e-01 -5.93888164e-01 2.11483732e-01 -3.38290125e-01
-1.23424196e+00 9.55639660e-01 -8.40585157e-02 -9.38863993e-01
-3.99152160e-01 -5.34374893e-01 -6.63641751e-01 9.57849979e-01
-1.66373277e+00 -1.57695806e+00 -3.58693421e-01 6.60242081e-01
7.55470991e-02 1.34357572e-01 1.05413485e+00 6.21092677e-01
-2.00548440e-01 4.82757360e-01 -9.07613188e-02 1.38359159e-01
8.02122533e-01 -1.47875512e+00 4.86888230e-01 4.95250136e-01
1.82335675e-01 3.25350851e-01 6.63791239e-01 -3.08753103e-01
-1.18819988e+00 -9.30247605e-01 1.04825580e+00 -8.01182687e-01
5.11396348e-01 -3.65386158e-01 -9.90820289e-01 5.26268005e-01
-1.44350201e-01 7.67664254e-01 3.21838051e-01 -1.96795821e-01
-3.99993137e-02 1.09167837e-01 -1.26346338e+00 2.45301127e-01
9.09807742e-01 -3.22764605e-01 -4.55496669e-01 6.52363539e-01
2.28455663e-01 -9.84261453e-01 -1.20528233e+00 6.18077636e-01
3.95668715e-01 -7.55711913e-01 1.23001373e+00 -5.69322884e-01
2.99584657e-01 -7.03768283e-02 1.90731913e-01 -1.14694822e+00
4.74606231e-02 -9.23165262e-01 1.91354632e-01 1.09637940e+00
4.32739317e-01 -7.29388893e-01 4.72037703e-01 7.34596074e-01
-4.03464511e-02 -9.42783475e-01 -1.24285102e+00 -6.78499222e-01
2.85464197e-01 -1.71754003e-01 3.59891981e-01 1.04353571e+00
-3.09730321e-01 3.05037759e-02 -3.93928468e-01 -1.87965766e-01
8.01713347e-01 -7.73799513e-03 5.71993411e-01 -1.36130893e+00
-3.76024604e-01 -6.56991541e-01 -6.79152012e-01 -5.82931221e-01
1.44982170e-02 -1.15127802e+00 2.00628638e-01 -1.47490406e+00
-6.41853437e-02 -6.15648031e-01 -3.59178245e-01 5.70241213e-01
-2.48652741e-01 8.01714897e-01 8.32694843e-02 3.16789538e-01
-4.23115939e-01 3.85048509e-01 1.38185430e+00 1.25497088e-01
-1.16336301e-01 2.39719376e-01 -4.52017486e-01 5.17312586e-01
5.63072085e-01 -4.11194384e-01 5.42538725e-02 -4.93301302e-01
2.88123220e-01 -7.40421377e-03 7.86971390e-01 -8.87073100e-01
1.51563868e-01 1.66487068e-01 2.25894690e-01 4.48554382e-02
2.35882387e-01 -8.50498915e-01 1.49902239e-01 4.39039588e-01
-3.36166352e-01 -2.17908174e-01 1.97273970e-01 2.01770365e-01
-1.59048215e-01 -6.59020841e-02 9.78448629e-01 -1.62934363e-01
-2.45441839e-01 5.64465821e-01 2.10639074e-01 3.24223012e-01
1.02539611e+00 -1.02270193e-01 3.63415629e-01 -3.67940255e-02
-1.15362847e+00 2.13592857e-01 4.67940956e-01 3.37731898e-01
4.13206249e-01 -1.29106176e+00 -1.10380006e+00 2.45718971e-01
-1.45272627e-01 4.53854084e-01 3.40039402e-01 1.52266228e+00
-6.53255880e-01 -1.20531498e-02 -3.97901475e-01 -7.35776246e-01
-1.22670722e+00 1.80896595e-01 7.16440260e-01 -6.77889585e-01
-7.97508776e-01 8.02295268e-01 -3.21945608e-01 -7.83223987e-01
-5.93274310e-02 -3.21203798e-01 -2.49168739e-01 2.43037986e-03
2.00714350e-01 8.05121809e-02 5.61173260e-01 -8.24036360e-01
-6.18202329e-01 1.02394867e+00 1.58631280e-01 -9.60300714e-02
1.98242521e+00 1.44251972e-01 -1.87103018e-01 -5.06188571e-02
1.35985005e+00 -1.28452420e-01 -1.55114639e+00 -3.13216358e-01
1.40552640e-01 -3.09049368e-01 1.31342456e-01 -6.50658131e-01
-1.24954009e+00 7.91923106e-01 6.77846611e-01 2.35073894e-01
9.73590672e-01 8.56768042e-02 1.03142273e+00 -8.96766409e-02
2.29782000e-01 -8.35214555e-01 -2.74480939e-01 3.10076088e-01
1.01898766e+00 -1.43431914e+00 8.38454589e-02 -2.54815727e-01
-4.18763191e-01 1.25309360e+00 -4.09017988e-02 -2.52295583e-01
4.52372551e-01 2.15029866e-01 2.16009185e-01 -3.56608242e-01
-1.39250591e-01 -1.78790838e-01 9.71408665e-01 5.64758956e-01
5.68226635e-01 -1.80474371e-01 -2.69597411e-01 1.41887292e-01
-2.00256750e-01 -1.32018358e-01 -8.13089237e-02 8.50284696e-01
-1.38027683e-01 -1.48784077e+00 -1.18146062e-01 2.58001655e-01
-7.01678813e-01 1.29792899e-01 -6.31468818e-02 7.00431764e-01
-1.28172576e-01 2.78821617e-01 -2.24324852e-01 -5.54824919e-02
5.90951324e-01 -1.78824127e-01 8.31966221e-01 -5.47469676e-01
-8.89518917e-01 7.69727826e-02 -1.38378695e-01 -9.78555918e-01
-6.72668338e-01 -7.99738526e-01 -1.14788687e+00 -5.35090342e-02
-1.65050641e-01 -1.71983391e-01 7.90642202e-01 1.35782588e+00
2.56344736e-01 4.97546881e-01 4.37157989e-01 -1.42450166e+00
-7.16982305e-01 -7.40908146e-01 -2.81643003e-01 8.22554529e-01
5.33063114e-01 -4.15349245e-01 -5.77739961e-02 1.26620168e-02] | [14.099178314208984, -2.542720317840576] |
a0963169-6975-437f-8cce-3f46d03b6d02 | generating-lead-sheets-with-affect-a-novel | 2104.13056 | null | https://arxiv.org/abs/2104.13056v1 | https://arxiv.org/pdf/2104.13056v1.pdf | Generating Lead Sheets with Affect: A Novel Conditional seq2seq Framework | The field of automatic music composition has seen great progress in the last few years, much of which can be attributed to advances in deep neural networks. There are numerous studies that present different strategies for generating sheet music from scratch. The inclusion of high-level musical characteristics (e.g., perceived emotional qualities), however, as conditions for controlling the generation output remains a challenge. In this paper, we present a novel approach for calculating the valence (the positivity or negativity of the perceived emotion) of a chord progression within a lead sheet, using pre-defined mood tags proposed by music experts. Based on this approach, we propose a novel strategy for conditional lead sheet generation that allows us to steer the music generation in terms of valence, phrasing, and time signature. Our approach is similar to a Neural Machine Translation (NMT) problem, as we include high-level conditions in the encoder part of the sequence-to-sequence architectures used (i.e., long-short term memory networks, and a Transformer network). We conducted experiments to thoroughly analyze these two architectures. The results show that the proposed strategy is able to generate lead sheets in a controllable manner, resulting in distributions of musical attributes similar to those of the training dataset. We also verified through a subjective listening test that our approach is effective in controlling the valence of a generated chord progression. | ['Dorien Herremans', 'Kat R. Agres', 'Dimos Makris'] | 2021-04-27 | null | null | null | null | ['music-generation', 'music-generation'] | ['audio', 'music'] | [ 4.30661738e-01 -1.12356201e-01 7.31708556e-02 -2.33944014e-01
-5.72761655e-01 -9.09429789e-01 6.63627267e-01 -1.94352582e-01
-1.81913108e-01 5.38567305e-01 2.00254962e-01 1.02837101e-01
-6.20059073e-02 -9.91160274e-01 -6.54643238e-01 -6.67080045e-01
1.69602513e-01 3.34401250e-01 -1.28637984e-01 -5.93669891e-01
4.40782398e-01 2.45764166e-01 -1.65320468e+00 5.47911584e-01
6.13047421e-01 1.07477224e+00 2.14681521e-01 6.06432319e-01
-3.74299958e-02 6.61088645e-01 -8.37640405e-01 -5.36077738e-01
2.20927313e-01 -8.91597271e-01 -6.42848372e-01 -2.21226186e-01
-1.37175024e-01 1.42371491e-01 2.57908851e-01 9.78692889e-01
8.01493585e-01 1.52924657e-01 6.85656965e-01 -8.62199366e-01
-5.08390546e-01 1.08791173e+00 -1.95361059e-02 -4.15742040e-01
4.77220774e-01 1.03562929e-01 1.17087185e+00 -8.36402953e-01
6.04192436e-01 8.82180750e-01 7.14959145e-01 5.67266703e-01
-1.31416714e+00 -7.34861791e-01 -3.31334800e-01 1.60019279e-01
-1.36852777e+00 -2.91320860e-01 1.15990627e+00 -5.48151076e-01
6.56755805e-01 3.42083037e-01 1.01518643e+00 1.22743702e+00
1.69924513e-01 3.87147218e-01 1.29313421e+00 -8.01142216e-01
3.75561863e-01 2.27387175e-01 -4.86458510e-01 1.47802472e-01
-4.54937220e-01 2.13006243e-01 -7.86033273e-01 9.67896432e-02
8.07334483e-01 -8.01774025e-01 -2.07268730e-01 -2.91612208e-01
-1.31925547e+00 7.29718387e-01 2.45561063e-01 6.06620550e-01
-5.18420756e-01 7.20444024e-02 5.61860919e-01 2.61421174e-01
9.75597352e-02 9.10799801e-01 -2.24882871e-01 -3.39358062e-01
-1.06275380e+00 3.59438658e-01 8.87096584e-01 6.19351029e-01
4.53872800e-01 2.68994212e-01 -5.49042284e-01 8.67523849e-01
-9.54558477e-02 4.09868270e-01 7.29820728e-01 -8.76701057e-01
2.01003864e-01 9.61026624e-02 1.86463505e-01 -1.09100342e+00
-2.05972031e-01 -5.48443854e-01 -7.66192079e-01 3.28730226e-01
2.36660242e-01 -1.84736013e-01 -5.66193700e-01 2.18723750e+00
-3.99153978e-02 -1.07769422e-01 1.00355171e-01 9.63801742e-01
3.79368544e-01 7.63642788e-01 -1.58118621e-01 -4.21485722e-01
1.29028606e+00 -4.17044848e-01 -8.86817336e-01 1.33778974e-01
1.86833069e-01 -9.36045706e-01 1.43791234e+00 6.36219323e-01
-1.31167352e+00 -7.49105275e-01 -1.16427100e+00 2.09886432e-01
-1.18843079e-01 2.43226334e-01 4.01707977e-01 6.03077829e-01
-9.86334980e-01 9.27190542e-01 -3.37534189e-01 -3.84408459e-02
-1.55054167e-01 3.98945123e-01 4.50174920e-02 8.12461555e-01
-1.58801305e+00 9.06573117e-01 5.49319744e-01 2.46349186e-01
-5.74026763e-01 -5.29655695e-01 -4.59205478e-01 2.23238945e-01
-6.08707927e-02 -6.53123200e-01 1.36318290e+00 -1.65811646e+00
-2.24560165e+00 9.33593333e-01 1.89762458e-01 -3.12504321e-01
3.44612002e-01 1.56561192e-02 -2.37784684e-01 -5.22926897e-02
-1.06845997e-01 7.70093203e-01 9.09281075e-01 -1.13533294e+00
-2.52036780e-01 -3.14009935e-02 -7.25224540e-02 2.28591159e-01
-6.02470756e-01 1.20414510e-01 -1.73715308e-01 -1.08349252e+00
-1.88905895e-01 -1.21892607e+00 6.19463474e-02 -4.37869191e-01
-4.09005076e-01 -7.35875592e-02 -1.23804830e-01 -5.13975143e-01
1.32292509e+00 -2.13625693e+00 4.34700698e-01 4.74894345e-01
-5.30004919e-01 1.28428057e-01 -8.48461762e-02 5.42136312e-01
-3.20443988e-01 -3.90978381e-02 -3.26828271e-01 -2.43644249e-02
2.66303331e-01 -1.87270671e-01 -6.43644691e-01 -5.06653152e-02
7.52653703e-02 6.59026742e-01 -7.36866653e-01 -2.33836621e-01
-1.51131317e-01 7.26379275e-01 -5.58896959e-01 4.54831988e-01
-4.56346124e-01 5.27589440e-01 -5.73816486e-02 1.65078312e-01
1.98566973e-01 3.00767630e-01 4.00700867e-01 -3.04494470e-01
-3.58368367e-01 6.50509596e-01 -1.03922665e+00 1.82084262e+00
-7.03849316e-01 5.10977983e-01 -2.24510044e-01 -6.51354849e-01
1.47079289e+00 6.73681736e-01 3.23938876e-01 -8.45833302e-01
3.26487124e-01 3.73152792e-01 2.71456242e-01 -2.41751000e-01
5.61920702e-01 -6.05827332e-01 -3.29024583e-01 6.47592604e-01
-3.01848389e-02 -4.85612005e-01 1.77602917e-01 -4.55524147e-01
5.96358538e-01 4.53188300e-01 2.25753933e-01 -9.86249670e-02
5.58169127e-01 -2.31565833e-01 4.85131949e-01 3.50490600e-01
2.70626396e-01 6.61469758e-01 5.16106129e-01 -1.40287623e-01
-1.16279197e+00 -1.00132763e+00 2.05401361e-01 9.75146770e-01
-1.92406639e-01 -4.61435318e-01 -1.03394282e+00 1.07718289e-01
-5.37011445e-01 9.31825042e-01 -5.16699016e-01 -3.36683869e-01
-8.23923409e-01 -4.14690167e-01 7.90522158e-01 4.11878645e-01
2.22784892e-01 -1.63908875e+00 -6.64818585e-01 3.27939242e-01
-4.73425925e-01 -7.30631053e-01 -4.27039921e-01 2.55558044e-01
-6.94205284e-01 -4.10701722e-01 -7.04882443e-01 -8.67525220e-01
1.09779544e-01 -4.77722555e-01 1.17116630e+00 -3.15463394e-01
2.31261272e-02 -5.15721664e-02 -3.56881976e-01 -4.06853408e-01
-6.08239651e-01 2.13982671e-01 4.52558957e-02 2.17583150e-01
1.22993715e-01 -9.83227313e-01 -4.70241517e-01 2.43411317e-01
-8.92252147e-01 3.14679861e-01 6.25632703e-01 6.95139766e-01
7.06523597e-01 -2.58130208e-02 8.49454999e-01 -5.61095655e-01
1.05494249e+00 -1.33933708e-01 -4.04020429e-01 -6.45822892e-03
-5.82774520e-01 3.60240452e-02 8.73962700e-01 -6.80691361e-01
-9.78160858e-01 1.04691960e-01 -3.35998893e-01 -3.99594158e-01
-6.79926053e-02 5.68795085e-01 -3.03190321e-01 2.72542834e-01
7.25849569e-01 2.35947117e-01 -1.38664469e-01 -1.34946391e-01
4.33514923e-01 7.80703008e-01 6.37401581e-01 -7.08139002e-01
7.60271430e-01 3.60025652e-02 -4.92915027e-02 -4.35584724e-01
-7.16432452e-01 2.03791246e-01 -6.27845466e-01 -4.74784672e-01
7.16197968e-01 -7.12096930e-01 -8.19611788e-01 4.39039320e-01
-1.17312121e+00 -4.58787441e-01 -3.49240839e-01 4.13015574e-01
-1.11471021e+00 -9.75116566e-02 -6.22272611e-01 -8.86875093e-01
-6.08887434e-01 -8.46294224e-01 7.77910233e-01 2.58933417e-02
-8.87738407e-01 -6.89250410e-01 3.29216301e-01 5.67961559e-02
5.07259309e-01 4.14902836e-01 1.11283576e+00 -2.53348678e-01
-3.30047429e-01 -3.16827483e-02 3.23528230e-01 5.00437677e-01
6.50556535e-02 1.00738117e-02 -1.05054939e+00 8.08266476e-02
2.78696686e-01 -2.34908596e-01 5.78156710e-01 1.97162554e-01
9.71420348e-01 -2.21241355e-01 3.38565230e-01 6.03294551e-01
1.19875205e+00 4.86539871e-01 9.42933440e-01 3.53693396e-01
3.93336475e-01 6.97159767e-01 7.13270068e-01 5.51061332e-01
5.37789799e-02 1.08055139e+00 1.73132077e-01 4.02180590e-02
-1.52286083e-01 -4.46889877e-01 8.67859721e-01 1.00728476e+00
-3.50693673e-01 -1.09002843e-01 -5.50416827e-01 4.69574243e-01
-1.54567838e+00 -1.27848244e+00 9.58780870e-02 2.32731247e+00
1.26679218e+00 1.50697589e-01 2.62980998e-01 5.75761437e-01
7.36048162e-01 -5.96870929e-02 -2.58538663e-01 -9.65323687e-01
-1.62524700e-01 6.86823905e-01 -9.66383144e-02 2.76371241e-01
-8.18359494e-01 8.42863023e-01 6.10733795e+00 7.48643160e-01
-1.55978751e+00 -2.76899815e-01 2.48064831e-01 -1.37886465e-01
-5.22864521e-01 -1.79828659e-01 -3.06430519e-01 5.33480704e-01
9.02527332e-01 -1.46750629e-01 8.53091240e-01 4.65881616e-01
4.49486256e-01 3.44401330e-01 -1.23176873e+00 9.74174201e-01
2.65899181e-01 -1.09948444e+00 1.31200328e-01 -3.16938937e-01
6.34572268e-01 -5.43001950e-01 3.96486789e-01 2.24828199e-01
-3.05380493e-01 -1.10652530e+00 1.26082885e+00 7.10927665e-01
1.02181232e+00 -9.56202626e-01 4.74398017e-01 1.55945703e-01
-1.06256318e+00 -1.06105302e-03 9.65669677e-02 -3.25436711e-01
1.39420614e-01 6.07718885e-01 -1.00110221e+00 3.99860859e-01
3.76908422e-01 3.00692022e-01 -1.73935264e-01 7.27669477e-01
-5.14884949e-01 8.14635694e-01 -3.13107371e-02 -2.56911963e-01
-1.62484765e-01 -3.49591464e-01 4.95511383e-01 1.12571263e+00
7.89957821e-01 -1.63331017e-01 -2.15566635e-01 1.21683395e+00
-6.64725751e-02 4.86342937e-01 -4.06377614e-01 -1.70391753e-01
3.60971510e-01 1.23131490e+00 -6.96450591e-01 -1.22426994e-01
1.36540651e-01 1.09650207e+00 7.67119229e-02 1.16129793e-01
-9.34056461e-01 -5.06472588e-01 3.43721807e-01 2.67844088e-02
4.53602493e-01 -5.85775897e-02 -6.16516948e-01 -9.24793422e-01
8.17871019e-02 -1.12748516e+00 -8.00789148e-02 -1.12779164e+00
-1.05714858e+00 9.29308832e-01 -3.48032475e-01 -1.41258955e+00
-7.58406997e-01 -2.72688836e-01 -6.87420666e-01 1.00727677e+00
-1.07138956e+00 -9.59262669e-01 2.99733644e-03 5.35710692e-01
2.07484797e-01 -1.10856585e-01 1.24467576e+00 3.50671649e-01
-1.03238046e-01 4.99511510e-01 -3.75816017e-01 5.84167577e-02
8.29405844e-01 -1.19428635e+00 1.36616513e-01 5.22151530e-01
3.64242285e-01 6.11295760e-01 9.61587310e-01 -3.38131607e-01
-1.15365458e+00 -8.57351601e-01 1.10433257e+00 -1.41805217e-01
5.34018219e-01 -5.95113277e-01 -5.98656952e-01 2.70322412e-01
4.86721426e-01 -8.41261744e-01 8.51842225e-01 1.73837796e-01
-2.82876015e-01 -2.52920449e-01 -7.72896528e-01 7.70878375e-01
7.40660608e-01 -5.82553208e-01 -6.45984054e-01 -1.12391107e-01
3.52852881e-01 -3.43454897e-01 -8.06305230e-01 4.45575416e-01
8.61724079e-01 -1.13675308e+00 6.88745439e-01 -1.66121036e-01
6.42409444e-01 -5.41899860e-01 -9.01012570e-02 -1.58017802e+00
-3.48199040e-01 -8.11746538e-01 2.62720436e-01 1.41295683e+00
4.93929416e-01 -7.91057199e-02 4.64270651e-01 1.53648078e-01
-1.59935668e-01 -6.27604663e-01 -6.03383422e-01 -5.95604599e-01
4.61638533e-02 -4.04921979e-01 5.89274704e-01 7.74836838e-01
2.34328806e-01 6.32722378e-01 -5.49815238e-01 -2.17558980e-01
8.50687698e-02 6.63618207e-01 6.99018240e-01 -1.08171904e+00
-6.35630310e-01 -6.58398926e-01 -1.86844140e-01 -5.35314798e-01
2.61849344e-01 -1.08132803e+00 3.56311381e-01 -1.12067533e+00
-1.44216223e-02 -3.77358198e-01 -5.01266837e-01 3.66041929e-01
1.01572253e-01 4.52939689e-01 4.67114598e-01 -8.35884213e-02
1.52246252e-01 6.46459460e-01 1.09403086e+00 -4.20859568e-02
-3.57688010e-01 1.70648590e-01 -5.21853626e-01 5.39827764e-01
1.06200373e+00 -3.57598871e-01 -3.82468313e-01 -1.33584470e-01
7.73434758e-01 2.62630016e-01 1.83253974e-01 -1.11309063e+00
-8.30320343e-02 3.99332717e-02 2.24292189e-01 -4.13343519e-01
5.54850757e-01 -4.83221054e-01 4.36504871e-01 3.58531564e-01
-7.46360481e-01 7.81639889e-02 6.94595352e-02 -1.31806716e-01
-5.06276071e-01 -2.07781523e-01 7.68508255e-01 -8.50382913e-03
-2.68024087e-01 -2.29081661e-01 -4.33765948e-01 -2.02476397e-01
6.36233389e-01 -7.77758807e-02 3.45188349e-01 -6.03839934e-01
-6.75638616e-01 -4.09051716e-01 3.66713345e-01 5.35217285e-01
3.25408280e-01 -1.61681771e+00 -7.06328154e-01 1.80765346e-01
-3.87413353e-02 -6.14344418e-01 9.04713050e-02 6.61480010e-01
-1.55981123e-01 2.69864798e-01 -5.34059465e-01 -3.76713514e-01
-1.13945758e+00 5.66804945e-01 3.41352969e-01 -1.53376684e-01
-2.97762662e-01 5.49109936e-01 -9.52112079e-02 -3.35714877e-01
1.66176781e-01 -2.69098967e-01 -3.86519164e-01 3.95016700e-01
3.05416077e-01 1.33408848e-02 9.08584893e-02 -5.35387456e-01
-1.13314174e-01 7.25542128e-01 6.30720258e-01 -6.94944501e-01
1.29052269e+00 2.00250283e-01 -3.21500927e-01 9.48335171e-01
7.71872818e-01 3.27561349e-01 -7.07207739e-01 2.90413380e-01
-6.76945373e-02 -1.44207165e-01 -1.92346707e-01 -1.07588410e+00
-8.85090649e-01 8.13106596e-01 4.80548561e-01 2.16280833e-01
1.40143085e+00 -4.51045215e-01 8.00430477e-01 1.75994426e-01
4.41659659e-01 -1.36335886e+00 -4.91428014e-04 6.15491390e-01
1.15265286e+00 -5.04222393e-01 -6.13405824e-01 -8.32354650e-02
-7.26969063e-01 1.29070163e+00 2.77572274e-01 -1.29102692e-01
2.77681202e-01 3.02396595e-01 2.45083481e-01 1.34836763e-01
-7.94506431e-01 -3.62994820e-02 2.40086436e-01 4.10287738e-01
9.05158639e-01 2.83816397e-01 -5.80520451e-01 9.28818822e-01
-9.71604884e-01 2.19242707e-01 5.60974658e-01 2.75540084e-01
-1.47414163e-01 -1.38525462e+00 -5.27658284e-01 -2.10534278e-02
-4.34838533e-01 -2.10183024e-01 -6.32599473e-01 3.19268733e-01
4.63329107e-01 7.85797894e-01 5.15713654e-02 -6.78655982e-01
4.02607828e-01 3.38300198e-01 7.07483053e-01 -5.07495284e-01
-1.00027859e+00 2.24625349e-01 2.17785880e-01 -3.91399741e-01
-4.53951865e-01 -6.23242140e-01 -1.26014972e+00 1.48797601e-01
-1.55675039e-01 2.87958026e-01 7.44320095e-01 6.76487327e-01
1.27413064e-01 7.35269964e-01 9.54427898e-01 -1.02942860e+00
-4.92720842e-01 -1.02020586e+00 -6.11160040e-01 4.78646606e-01
-1.22132286e-01 -3.88564110e-01 -1.12920351e-01 2.72482991e-01] | [15.990113258361816, 5.539989948272705] |
6fcf8b09-f47b-4ad3-942b-a5e9e7611812 | bleurt-has-universal-translations-an-analysis | 2307.03131 | null | https://arxiv.org/abs/2307.03131v2 | https://arxiv.org/pdf/2307.03131v2.pdf | BLEURT Has Universal Translations: An Analysis of Automatic Metrics by Minimum Risk Training | Automatic metrics play a crucial role in machine translation. Despite the widespread use of n-gram-based metrics, there has been a recent surge in the development of pre-trained model-based metrics that focus on measuring sentence semantics. However, these neural metrics, while achieving higher correlations with human evaluations, are often considered to be black boxes with potential biases that are difficult to detect. In this study, we systematically analyze and compare various mainstream and cutting-edge automatic metrics from the perspective of their guidance for training machine translation systems. Through Minimum Risk Training (MRT), we find that certain metrics exhibit robustness defects, such as the presence of universal adversarial translations in BLEURT and BARTScore. In-depth analysis suggests two main causes of these robustness deficits: distribution biases in the training datasets, and the tendency of the metric paradigm. By incorporating token-level constraints, we enhance the robustness of evaluation metrics, which in turn leads to an improvement in the performance of machine translation systems. Codes are available at \url{https://github.com/powerpuffpomelo/fairseq_mrt}. | ['Mingxuan Wang', 'Jiajun Chen', 'ShuJian Huang', 'Chengqi Zhao', 'Tao Wang', 'Yiming Yan'] | 2023-07-06 | null | null | null | null | ['machine-translation'] | ['natural-language-processing'] | [ 2.13626876e-01 -2.08402276e-01 -3.48780513e-01 -5.88038266e-01
-1.09180224e+00 -6.53398097e-01 7.42926776e-01 3.91461849e-01
-4.72662002e-01 7.11915374e-01 5.03840268e-01 -6.54191196e-01
-1.25015993e-02 -5.70809364e-01 -5.40358961e-01 -4.25100476e-01
4.31856215e-01 2.52951831e-01 -1.43643141e-01 -5.57809174e-01
4.52415586e-01 1.59544021e-01 -1.18253875e+00 2.43996754e-01
1.09329784e+00 6.33777380e-01 -8.41081291e-02 3.11804175e-01
7.32717365e-02 4.27615285e-01 -6.24903917e-01 -9.67760384e-01
2.05653101e-01 -6.85624003e-01 -5.38386464e-01 -4.75323766e-01
3.78523320e-01 -5.76711707e-02 -2.28058428e-01 1.32860482e+00
6.55533373e-01 -6.21465668e-02 6.04448617e-01 -8.50216568e-01
-9.13803756e-01 7.48419583e-01 -2.55758107e-01 3.82312208e-01
3.42564732e-01 3.50241333e-01 1.38026845e+00 -9.73161817e-01
4.87514585e-01 1.11807394e+00 5.84087014e-01 6.05565906e-01
-1.07214952e+00 -5.14174938e-01 -3.21302742e-01 1.58137292e-01
-9.89846230e-01 -6.58751249e-01 4.16853160e-01 -4.25737530e-01
8.11872065e-01 4.31502461e-01 6.18939735e-02 1.21829104e+00
5.46763897e-01 5.73630869e-01 1.36719251e+00 -6.06335402e-01
-4.38279323e-02 3.39451611e-01 -3.58853973e-02 3.77425492e-01
3.13224047e-01 2.47882083e-01 -6.30001783e-01 -5.07787317e-02
4.58912879e-01 -3.25144112e-01 -2.68055767e-01 -6.86273798e-02
-1.40074289e+00 8.70454550e-01 3.13785106e-01 5.86586416e-01
-1.79091290e-01 -5.97347803e-02 6.46021903e-01 6.13763809e-01
7.40013897e-01 8.26851964e-01 -4.20051187e-01 -4.68584567e-01
-9.19640541e-01 7.42478669e-02 4.97934043e-01 5.84340036e-01
5.10562003e-01 -7.07197636e-02 -4.86138135e-01 9.33990061e-01
9.98561606e-02 4.90221441e-01 6.51592493e-01 -6.39469624e-01
8.33013058e-01 6.25754356e-01 1.87388584e-02 -1.01438510e+00
-3.82995531e-02 -6.82663321e-01 -5.66094995e-01 -7.47405812e-02
5.87545156e-01 9.20192972e-02 -5.85703135e-01 1.81326306e+00
1.59651891e-03 -6.61199510e-01 -1.30290821e-01 1.13952160e+00
3.53103310e-01 3.94243985e-01 -5.06013036e-02 -3.56288515e-02
1.03014755e+00 -8.27073932e-01 -6.19676709e-01 -2.05278143e-01
8.51863265e-01 -1.02564096e+00 1.61671662e+00 1.26015425e-01
-9.64561880e-01 -3.60947490e-01 -1.08277464e+00 -5.33320941e-02
-4.20686513e-01 -5.24238832e-02 5.41832685e-01 8.74764621e-01
-8.49066675e-01 9.67195988e-01 -6.24846756e-01 -4.62391347e-01
2.11243749e-01 2.65905380e-01 -1.35752723e-01 -8.92602354e-02
-1.40835345e+00 1.37174726e+00 -2.76739057e-03 2.34814778e-01
-5.97792387e-01 -3.27654421e-01 -5.38525581e-01 -8.39939192e-02
-1.60595458e-02 -4.32435930e-01 1.13364673e+00 -1.25105321e+00
-1.35954583e+00 8.37768435e-01 -1.42344519e-01 -2.64613628e-01
6.04675055e-01 -2.42383078e-01 -2.85719484e-01 -1.77331835e-01
2.03788549e-01 4.35160458e-01 4.70583290e-01 -8.83945763e-01
-3.39915067e-01 -3.74060750e-01 3.57198715e-02 2.36221507e-01
-5.63869417e-01 5.10996163e-01 -5.02427593e-02 -8.01043332e-01
-1.18074097e-01 -8.54278147e-01 -5.53180724e-02 -3.67938548e-01
-2.29458064e-01 -2.94976443e-01 4.39769700e-02 -8.09614122e-01
1.46545804e+00 -2.02476931e+00 5.95436804e-02 1.01707458e-01
-1.59380361e-01 3.43950599e-01 -1.96245000e-01 5.47984004e-01
1.84920013e-01 4.58415449e-01 -3.22225392e-01 -1.60265446e-01
1.81417540e-02 -4.07399721e-02 -3.42414111e-01 4.24435943e-01
3.10474187e-01 1.00771534e+00 -9.54051852e-01 -3.25184494e-01
4.36481833e-02 3.51472795e-01 -4.16013241e-01 -1.55878225e-02
-1.27096817e-01 5.24130464e-01 -2.96496898e-01 6.83308363e-01
2.39521772e-01 -3.10949516e-02 4.56436584e-03 1.46546006e-01
-4.88137342e-02 9.41481948e-01 -4.67267871e-01 1.57723355e+00
-3.02609324e-01 8.67600322e-01 -3.09519738e-01 -6.78618073e-01
9.43637073e-01 2.30774939e-01 9.94420126e-02 -1.03348529e+00
3.52824062e-01 7.00945914e-01 5.12677491e-01 -2.46719301e-01
6.44021988e-01 -1.28101796e-01 4.78837155e-02 5.52562177e-01
-1.21005856e-01 -6.36872649e-02 3.26861620e-01 -5.47606274e-02
1.04018784e+00 2.07176164e-01 3.47479135e-02 -3.75076830e-01
3.42805654e-01 7.49502629e-02 6.43925965e-01 4.76155460e-01
-2.79180676e-01 6.98192179e-01 4.59334195e-01 -1.88508928e-01
-1.29288721e+00 -8.10592711e-01 -3.50447953e-01 1.10773659e+00
-2.11223170e-01 -5.09279370e-01 -8.38480830e-01 -6.35825276e-01
-3.62024941e-02 8.16502273e-01 -5.70383072e-01 -2.74347007e-01
-5.14813781e-01 -9.58999813e-01 7.50287712e-01 2.60775030e-01
1.38901860e-01 -9.13105249e-01 -4.93285924e-01 8.42763484e-02
-3.73653859e-01 -7.71560907e-01 -4.43810850e-01 1.14277050e-01
-1.14482450e+00 -6.11438990e-01 -8.04137409e-01 -4.53427225e-01
5.67408979e-01 1.24488913e-01 1.19293261e+00 1.80777773e-01
1.38323680e-01 -5.51584102e-02 -6.12799466e-01 -4.05534089e-01
-7.23635852e-01 3.17256987e-01 1.99516535e-01 -2.20535234e-01
6.98865235e-01 -4.57642764e-01 -6.51683211e-01 5.27373075e-01
-7.28410900e-01 -1.49403527e-01 7.42924571e-01 7.36643434e-01
1.99744090e-01 -5.77194870e-01 7.10143030e-01 -6.98136449e-01
9.98105288e-01 -4.52389210e-01 -3.03535193e-01 2.48577565e-01
-1.12235355e+00 5.75599037e-02 5.78100204e-01 -2.58474380e-01
-6.20880067e-01 -7.15232432e-01 -1.23564631e-01 -9.18560177e-02
4.16437685e-02 6.28597021e-01 -3.39117920e-04 1.44576775e-02
1.00315750e+00 1.20389484e-01 1.21712588e-01 -4.45255816e-01
1.20948710e-01 9.29597914e-01 7.13559240e-02 -6.96968377e-01
7.19280303e-01 -6.93021268e-02 -3.81788701e-01 -4.99686688e-01
-8.24843347e-01 -1.52709186e-01 -4.18194324e-01 -1.58989936e-01
5.13045549e-01 -7.07892597e-01 -1.34521767e-01 2.25708723e-01
-9.97980714e-01 -3.71407598e-01 -4.21799021e-03 6.11256540e-01
-3.72258097e-01 3.27266276e-01 -6.77194953e-01 -4.79392678e-01
-5.27749360e-01 -1.30841935e+00 7.24102676e-01 4.57615554e-02
-6.71139419e-01 -9.83752012e-01 1.68255568e-01 6.80879295e-01
7.43522763e-01 1.61594655e-02 1.08827150e+00 -8.63752365e-01
-2.88124174e-01 -2.18606681e-01 -9.47860852e-02 6.96394742e-01
1.69924185e-01 1.27502875e-02 -9.33168888e-01 -2.38737330e-01
-1.11956615e-03 -1.57696575e-01 6.18767142e-01 7.42783695e-02
6.65513933e-01 -1.85401410e-01 1.05952986e-01 4.35299903e-01
1.26042914e+00 1.11020185e-01 6.66008353e-01 6.57962024e-01
3.75292152e-01 6.93311274e-01 5.92748761e-01 4.65412438e-02
2.25538835e-01 7.65187860e-01 2.22778678e-01 2.05912851e-02
-8.72840360e-02 -2.66081929e-01 7.05649376e-01 1.15611899e+00
-1.86091930e-01 -1.78092137e-01 -1.11572623e+00 4.61550832e-01
-1.67357194e+00 -7.53343582e-01 -1.68735161e-01 2.29255962e+00
1.13128889e+00 3.59914750e-01 -4.79621030e-02 1.87416712e-03
7.21358418e-01 6.61504492e-02 -3.40939939e-01 -8.20762813e-01
-3.40009898e-01 1.21324755e-01 5.01420736e-01 4.11115110e-01
-6.59670591e-01 9.43537951e-01 6.06686592e+00 7.25807488e-01
-1.22204196e+00 2.07166329e-01 7.75885284e-01 -1.05864935e-01
-6.17981911e-01 3.63345295e-02 -5.60703695e-01 5.81416011e-01
1.11282623e+00 -2.45737851e-01 5.02398551e-01 4.03195322e-01
3.91812354e-01 -2.50616781e-02 -1.16793501e+00 6.14997625e-01
7.51178265e-02 -9.15227592e-01 8.26542825e-02 9.10005122e-02
7.94296741e-01 3.84349793e-01 2.79639304e-01 2.30294898e-01
9.80800614e-02 -9.72780049e-01 8.50266099e-01 1.63889989e-01
7.53698230e-01 -5.75067401e-01 9.65039730e-01 1.82550535e-01
-3.88791978e-01 8.34392235e-02 -4.83513474e-01 -2.04681799e-01
-1.31274924e-01 8.49155068e-01 -6.50039256e-01 4.72598583e-01
4.76323783e-01 5.42652845e-01 -6.71824932e-01 8.86667430e-01
-4.62354302e-01 9.39673245e-01 8.73691365e-02 -2.21853644e-01
2.58186549e-01 -3.32176387e-01 5.02140701e-01 1.15975225e+00
3.44079405e-01 -3.31532210e-01 -4.32590663e-01 7.12714195e-01
-2.00345814e-01 4.19764727e-01 -6.48170948e-01 -4.24405515e-01
3.57934266e-01 1.12146783e+00 -5.32224834e-01 -7.46939192e-03
-5.51883399e-01 8.47795963e-01 3.18175226e-01 2.46740222e-01
-8.95826221e-01 -3.90717030e-01 8.29992414e-01 1.56320810e-01
-1.45963952e-01 -3.50023478e-01 -7.97298551e-01 -1.22142339e+00
3.63790929e-01 -1.40426683e+00 4.33940515e-02 -4.00259644e-01
-1.12470222e+00 4.99213755e-01 -4.13805246e-01 -1.25028849e+00
-6.38279095e-02 -6.58489168e-01 -5.29781461e-01 1.03485239e+00
-1.26587081e+00 -6.79613650e-01 2.16932610e-01 2.46523276e-01
5.44234991e-01 -2.23935813e-01 8.36909175e-01 4.55514610e-01
-5.13025641e-01 9.58601236e-01 4.07789379e-01 3.02570224e-01
1.02335560e+00 -8.63283157e-01 6.66079223e-01 9.51237261e-01
2.22961292e-01 1.02045214e+00 8.80523086e-01 -5.00720859e-01
-1.29108191e+00 -7.35115290e-01 1.28986251e+00 -7.83362925e-01
7.56569803e-01 -4.06077325e-01 -8.21239650e-01 4.50156331e-01
2.90638536e-01 -6.48607194e-01 7.52649367e-01 3.13462436e-01
-4.80372995e-01 -6.30918443e-02 -9.11331534e-01 7.54036784e-01
9.41431344e-01 -7.55916655e-01 -5.25607169e-01 3.23861986e-01
5.62901080e-01 -1.39560342e-01 -8.88166368e-01 4.66991812e-01
5.77058852e-01 -8.87763739e-01 5.15050232e-01 -6.72587276e-01
8.19198668e-01 3.35837435e-03 -2.68898398e-01 -1.46636498e+00
-1.94350079e-01 -6.56510532e-01 3.51572305e-01 1.33035493e+00
9.93643999e-01 -8.24481547e-01 4.08349246e-01 6.46795094e-01
-1.33077964e-01 -1.02795863e+00 -9.34316218e-01 -8.48208189e-01
5.82388341e-01 -3.50338221e-01 5.02849877e-01 1.03951001e+00
2.69270808e-01 2.91712821e-01 -1.15021348e-01 -3.47494483e-01
3.39857221e-01 -3.05126309e-01 5.34209967e-01 -8.93849373e-01
-1.96113840e-01 -7.93657899e-01 -2.76902825e-01 -6.42624617e-01
-6.94525838e-02 -1.11647177e+00 9.71529260e-02 -1.43523467e+00
2.63714463e-01 -3.84962201e-01 -4.73111689e-01 4.00214255e-01
-3.21036667e-01 3.00023466e-01 1.79097548e-01 4.15813178e-01
-3.04659486e-01 4.74684566e-01 1.14012563e+00 1.93897039e-02
6.86112344e-02 -2.24698503e-02 -8.25818360e-01 4.04074401e-01
1.29057407e+00 -6.18421376e-01 -6.42181709e-02 -9.56421912e-01
7.34091341e-01 -3.41900289e-01 1.08105570e-01 -8.80290449e-01
-1.16637245e-01 -1.97264835e-01 5.43197207e-02 -2.28891000e-02
-1.94800794e-02 -2.79040426e-01 -2.96972394e-01 4.19694036e-01
-5.38065672e-01 6.25184059e-01 7.81998113e-02 -5.35827503e-02
-3.47610533e-01 -2.47915938e-01 5.70757866e-01 -8.60048980e-02
-1.72022372e-01 -1.12983519e-02 -2.19415635e-01 1.74485907e-01
4.66207147e-01 -1.25791490e-01 -5.45471013e-01 -3.27630013e-01
-8.83568600e-02 6.94420710e-02 6.41479373e-01 7.93954909e-01
2.80093759e-01 -1.24119008e+00 -1.02193403e+00 -1.79986462e-01
1.99254647e-01 -4.95335251e-01 -1.95565686e-01 1.17212486e+00
-4.16771621e-01 6.33584559e-01 -2.19072521e-01 -4.06217426e-01
-9.46807325e-01 2.29704455e-01 2.71469980e-01 -2.33910561e-01
-1.18645273e-01 6.25612557e-01 -1.54504135e-01 -4.56792772e-01
8.36263746e-02 -2.25090325e-01 1.13849044e-01 1.43171728e-01
2.61147112e-01 4.30367023e-01 2.82731056e-01 -6.11236811e-01
-3.75923753e-01 2.40770385e-01 -1.23882949e-01 -3.41575354e-01
1.14420629e+00 -5.57048246e-02 -1.75897285e-01 6.02205753e-01
9.63361084e-01 1.64501593e-01 -7.01387107e-01 -1.82164744e-01
3.45981807e-01 -5.76326549e-01 -2.21502990e-01 -1.04383826e+00
-6.47441983e-01 1.08469808e+00 4.74804759e-01 5.23398351e-03
7.86651492e-01 -2.20677286e-01 7.90552855e-01 1.84760228e-01
3.78386617e-01 -1.15768039e+00 -8.38900134e-02 7.51129389e-01
8.23602498e-01 -1.20525801e+00 -1.37187928e-01 -6.69244453e-02
-5.80600500e-01 8.86352718e-01 3.56360883e-01 1.45882934e-01
2.78251059e-02 -3.80750634e-02 4.32037324e-01 1.44842312e-01
-6.49632037e-01 -4.37353551e-02 2.73789406e-01 2.31211111e-01
9.66770291e-01 1.83277279e-01 -9.31440055e-01 3.73830616e-01
-5.23263037e-01 -1.92771673e-01 3.21540207e-01 6.23789847e-01
-3.50783110e-01 -1.44618416e+00 -2.25303158e-01 4.52220380e-01
-8.58839452e-01 -5.03794372e-01 -6.29992008e-01 5.41045249e-01
-2.57172465e-01 1.02154028e+00 -2.59858280e-01 -5.92021704e-01
2.44175062e-01 3.76306772e-01 4.64010656e-01 -6.34465873e-01
-1.01894844e+00 -4.73847359e-01 2.50304818e-01 -3.65127981e-01
-1.60087228e-01 -7.69274533e-01 -7.12571502e-01 -4.54473138e-01
-4.33144450e-01 3.36237788e-01 9.10625875e-01 8.85708392e-01
3.86372119e-01 1.08594991e-01 6.17926419e-01 -3.58094662e-01
-9.64272201e-01 -1.37780297e+00 -3.10920402e-02 5.34550071e-01
4.40662727e-02 -3.44569385e-01 -5.24096906e-01 -3.13190013e-01] | [11.522455215454102, 10.178117752075195] |
ac378d77-5e41-4c43-a745-02ccc1c59875 | look-into-object-self-supervised-structure | 2003.14142 | null | https://arxiv.org/abs/2003.14142v1 | https://arxiv.org/pdf/2003.14142v1.pdf | Look-into-Object: Self-supervised Structure Modeling for Object Recognition | Most object recognition approaches predominantly focus on learning discriminative visual patterns while overlooking the holistic object structure. Though important, structure modeling usually requires significant manual annotations and therefore is labor-intensive. In this paper, we propose to "look into object" (explicitly yet intrinsically model the object structure) through incorporating self-supervisions into the traditional framework. We show the recognition backbone can be substantially enhanced for more robust representation learning, without any cost of extra annotation and inference speed. Specifically, we first propose an object-extent learning module for localizing the object according to the visual patterns shared among the instances in the same category. We then design a spatial context learning module for modeling the internal structures of the object, through predicting the relative positions within the extent. These two modules can be easily plugged into any backbone networks during training and detached at inference time. Extensive experiments show that our look-into-object approach (LIO) achieves large performance gain on a number of benchmarks, including generic object recognition (ImageNet) and fine-grained object recognition tasks (CUB, Cars, Aircraft). We also show that this learning paradigm is highly generalizable to other tasks such as object detection and segmentation (MS COCO). Project page: https://github.com/JDAI-CV/LIO. | ['Wei zhang', 'Yalong Bai', 'Mohan Zhou', 'Tiejun Zhao', 'Tao Mei'] | 2020-03-31 | look-into-object-self-supervised-structure-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Zhou_Look-Into-Object_Self-Supervised_Structure_Modeling_for_Object_Recognition_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhou_Look-Into-Object_Self-Supervised_Structure_Modeling_for_Object_Recognition_CVPR_2020_paper.pdf | cvpr-2020-6 | ['image-recognition'] | ['computer-vision'] | [ 1.05302021e-01 4.28707339e-02 -4.35397983e-01 -6.67362094e-01
-5.16643941e-01 -7.77321637e-01 4.41540092e-01 1.12675771e-01
-9.61026996e-02 7.44896084e-02 -1.32615298e-01 -2.11480573e-01
-1.10082023e-01 -6.43117309e-01 -1.05389166e+00 -5.20234883e-01
3.97438705e-02 5.90449393e-01 6.44720256e-01 2.94107288e-01
3.48524839e-01 9.12427247e-01 -1.48354197e+00 5.36640525e-01
3.76059771e-01 1.25388134e+00 4.94324774e-01 4.56464440e-01
-5.85816428e-02 8.68536353e-01 -2.60535628e-01 -1.20746195e-01
2.49647796e-01 -6.32959092e-03 -1.07703829e+00 5.03771126e-01
8.42830539e-01 -1.94894925e-01 -1.66064113e-01 9.60628569e-01
3.73758972e-02 8.26721564e-02 6.05328977e-01 -1.20455492e+00
-6.69004858e-01 3.30890119e-01 -7.34723985e-01 2.11899370e-01
-1.71541795e-01 2.54104942e-01 1.14368069e+00 -1.13062692e+00
6.16431415e-01 1.24504554e+00 5.65578222e-01 4.04024333e-01
-1.32665277e+00 -3.61921221e-01 7.07179606e-01 3.36973369e-01
-1.41157854e+00 -3.37721646e-01 8.44940305e-01 -6.11387908e-01
7.69107938e-01 2.57122487e-01 4.50138092e-01 6.89983904e-01
-2.00825095e-01 1.10137165e+00 8.98627102e-01 -2.65754431e-01
1.42678201e-01 9.67890248e-02 5.02158642e-01 9.91889417e-01
1.96401358e-01 -8.47015381e-02 -2.29646906e-01 2.66146481e-01
8.10426950e-01 4.04680759e-01 1.04915444e-02 -7.71635234e-01
-1.14045429e+00 5.48094571e-01 1.04091549e+00 2.14345381e-01
-2.26132855e-01 5.05194724e-01 2.35044450e-01 -7.92672411e-02
2.11402074e-01 2.10367456e-01 -5.84617555e-01 2.96921194e-01
-9.04844701e-01 2.07321346e-01 5.38325131e-01 1.10676849e+00
1.16802633e+00 -2.16505244e-01 -1.67810798e-01 8.51401567e-01
3.96290958e-01 9.05139521e-02 1.76288128e-01 -9.52996969e-01
2.94054419e-01 8.22655261e-01 -5.77361370e-03 -8.97398174e-01
-5.27428091e-01 -5.78733325e-01 -5.89523137e-01 3.34253222e-01
4.20154691e-01 2.02431247e-01 -1.24828601e+00 1.42735302e+00
5.21603465e-01 2.90273845e-01 -4.11895573e-01 1.03231597e+00
7.62224615e-01 5.46756208e-01 1.07463963e-01 4.25392240e-01
1.55946088e+00 -1.47640562e+00 -1.36288255e-01 -7.06914008e-01
7.38005817e-01 -5.06460488e-01 1.16223288e+00 1.66382730e-01
-9.12833929e-01 -6.96061373e-01 -8.35959852e-01 -3.95247340e-01
-3.98424923e-01 4.18301761e-01 8.26072752e-01 1.95149988e-01
-8.11197698e-01 4.36963350e-01 -9.95579422e-01 -2.96223313e-01
8.72714937e-01 3.29579741e-01 -4.02600527e-01 -2.17952535e-01
-2.56503373e-01 6.54008031e-01 5.80422580e-01 1.69078588e-01
-1.19287622e+00 -7.17570245e-01 -9.25101757e-01 1.29119575e-01
5.66726625e-01 -6.84294641e-01 1.17164779e+00 -1.21320832e+00
-9.85513270e-01 9.58408177e-01 -2.85178244e-01 -3.83379310e-01
2.17608318e-01 -2.53653377e-01 1.16891071e-01 1.03275821e-01
1.77331269e-01 1.03052974e+00 7.90894747e-01 -1.52819884e+00
-7.14437187e-01 -5.10605991e-01 2.18480051e-01 -8.67349356e-02
-5.43888323e-02 1.01000808e-01 -8.83544564e-01 -7.37046361e-01
2.79451966e-01 -9.91728961e-01 -3.58362436e-01 4.23474133e-01
-5.14561713e-01 -4.16249245e-01 1.04723382e+00 -4.28478032e-01
9.70365942e-01 -2.24764132e+00 4.75442037e-02 2.14769095e-01
1.11288279e-01 1.94012538e-01 -2.55588025e-01 4.20433693e-02
-4.64068428e-02 -5.62734939e-02 -2.69285053e-01 -4.11994487e-01
2.75821611e-02 3.44379991e-01 -2.99819320e-01 3.94560128e-01
3.16884071e-01 1.27485383e+00 -5.61988175e-01 -4.20530856e-01
2.63667941e-01 1.78361863e-01 -7.39497781e-01 2.00199112e-01
-4.83609885e-01 3.63690495e-01 -3.53064716e-01 9.43540275e-01
5.48799038e-01 -6.76235437e-01 1.04244776e-01 -3.96609038e-01
4.56821695e-02 3.37767631e-01 -1.21408212e+00 1.64442015e+00
-4.48636621e-01 4.94590610e-01 1.10957853e-01 -1.17149317e+00
7.86883712e-01 -2.09170446e-01 6.24015965e-02 -5.74846566e-01
-1.04670875e-01 3.83574814e-02 3.26144472e-02 -4.60099250e-01
1.79005563e-01 8.15410614e-02 3.60048376e-02 2.03011230e-01
1.21115513e-01 1.98187321e-01 1.34161159e-01 1.19017123e-03
7.13258445e-01 2.28524089e-01 2.90856034e-01 -2.76830614e-01
2.52781570e-01 6.99664503e-02 6.11205935e-01 7.08428085e-01
-1.27297729e-01 7.12647200e-01 4.27461714e-01 -5.98579288e-01
-8.63291204e-01 -9.71623003e-01 -2.78741837e-01 1.44535315e+00
3.89281541e-01 -3.33946377e-01 -6.29081249e-01 -9.29579735e-01
3.30726355e-01 4.08514708e-01 -7.13944793e-01 6.78516999e-02
-7.11340010e-01 -4.52212274e-01 3.23854566e-01 1.22268927e+00
2.33215228e-01 -1.10285115e+00 -5.71276784e-01 -1.93362380e-03
1.70901492e-01 -1.20441639e+00 -5.31793237e-01 3.04882854e-01
-9.32967901e-01 -9.45276856e-01 -4.16821033e-01 -1.02603042e+00
8.94236088e-01 3.78012508e-01 8.94904494e-01 1.64016500e-01
-6.60813272e-01 3.50798249e-01 -9.45914090e-02 -3.46330613e-01
1.33948535e-01 1.16789930e-01 -2.97298908e-01 2.99883097e-01
1.17474012e-01 -4.50837255e-01 -8.29804361e-01 5.11829138e-01
-7.38840759e-01 1.77964449e-01 7.44853616e-01 7.33655930e-01
1.00061309e+00 -2.78333575e-01 4.42688346e-01 -9.74087954e-01
-2.22375080e-01 -4.13362980e-01 -7.28274703e-01 3.37881297e-01
-3.80892247e-01 4.14457098e-02 3.89924705e-01 -3.57637227e-01
-9.12161767e-01 3.67368251e-01 -5.41584529e-02 -7.19946563e-01
-5.65481842e-01 3.54694605e-01 -4.04887259e-01 -6.34631887e-02
3.44202042e-01 2.59521753e-01 -1.49086744e-01 -8.81981313e-01
5.39425910e-01 3.90711755e-01 5.35947621e-01 -6.31937385e-01
6.72655284e-01 7.23979592e-01 -6.41918555e-02 -5.48834026e-01
-1.20240831e+00 -6.65318727e-01 -1.03918874e+00 4.66309711e-02
7.98651993e-01 -9.59643304e-01 -6.08590662e-01 1.48860246e-01
-1.09764874e+00 -6.05719984e-01 -3.01533610e-01 2.52557367e-01
-4.88664836e-01 2.02484041e-01 -5.00137031e-01 -3.04789096e-01
-9.66468826e-02 -1.06875193e+00 1.33713830e+00 1.43792033e-01
-7.79836345e-03 -9.24606740e-01 -2.56860167e-01 5.07461011e-01
1.33663401e-01 1.42557062e-02 9.82667387e-01 -5.46278298e-01
-1.16299295e+00 -1.94808304e-01 -6.28572285e-01 3.61515641e-01
9.47451591e-02 -3.88150513e-02 -9.87575829e-01 -2.67409205e-01
-4.46038306e-01 -2.92755455e-01 1.10754240e+00 2.65942156e-01
1.72633362e+00 -4.98383015e-01 -7.46599376e-01 9.05530989e-01
1.47887540e+00 -2.71668397e-02 3.44046146e-01 2.48696804e-01
1.19383669e+00 7.60965407e-01 6.41792953e-01 2.73811877e-01
3.70778024e-01 9.13444698e-01 6.39963925e-01 -1.95425197e-01
-4.20426130e-01 -2.00180948e-01 1.24169514e-01 2.71093398e-01
5.18901795e-02 -2.49251095e-03 -1.06704307e+00 8.23600769e-01
-1.88675952e+00 -7.85644889e-01 -8.26735348e-02 1.83375144e+00
5.89495301e-01 3.08802035e-02 2.48972341e-01 -3.14249218e-01
5.55187762e-01 1.68760687e-01 -6.50847256e-01 -1.23103440e-01
1.20704390e-01 1.02071226e-01 4.67054278e-01 3.57106417e-01
-1.34272754e+00 1.12619376e+00 5.91418743e+00 8.28926802e-01
-1.21421075e+00 1.29361063e-01 7.40639210e-01 -7.52719119e-02
-5.14038047e-03 2.06049889e-01 -1.10951126e+00 2.68188268e-01
4.70228344e-01 2.53727645e-01 1.53723985e-01 1.26501679e+00
-3.58688571e-02 1.20528631e-01 -1.47570324e+00 8.06476951e-01
-5.68398982e-02 -1.54765773e+00 2.33975425e-01 1.78696681e-02
6.45337880e-01 2.60079712e-01 -3.35189328e-02 1.42785430e-01
1.39925024e-02 -1.00369120e+00 1.04676080e+00 4.50171292e-01
5.89331448e-01 -4.80372965e-01 3.61707926e-01 4.75953519e-01
-1.36150098e+00 -2.76676059e-01 -4.58557338e-01 1.49291620e-01
-8.35914463e-02 4.47319537e-01 -7.79991150e-01 2.99835593e-01
8.17902207e-01 8.45043361e-01 -9.33097601e-01 1.19254100e+00
-2.22364128e-01 6.68518066e-01 -2.88058847e-01 2.47385636e-01
4.60229665e-01 4.58135270e-02 3.27514172e-01 1.42159152e+00
-1.07475862e-01 -1.36245534e-01 5.46242595e-01 1.06531787e+00
-1.73617691e-01 -9.67365652e-02 -5.18878937e-01 1.63906425e-01
3.32285553e-01 1.42305803e+00 -9.64568317e-01 -3.95746529e-01
-5.72123170e-01 9.39511418e-01 6.84549272e-01 2.74084300e-01
-8.03777218e-01 -2.42938519e-01 6.15561843e-01 4.12600070e-01
1.03159821e+00 -2.23330528e-01 -3.54147404e-01 -9.67388034e-01
2.92198837e-01 -6.49025261e-01 3.86663109e-01 -6.55436337e-01
-1.30135369e+00 3.55374515e-01 3.15642357e-02 -1.08227468e+00
1.56104937e-01 -1.03502643e+00 -6.29377723e-01 5.47646940e-01
-1.51171529e+00 -1.47599220e+00 -3.79995346e-01 5.05446017e-01
6.13033235e-01 2.33404696e-01 4.40739214e-01 3.24155718e-01
-7.40126908e-01 5.71683526e-01 -3.61623093e-02 3.76174629e-01
3.74040425e-01 -1.23253894e+00 3.65749389e-01 5.95825493e-01
5.43721735e-01 7.58577526e-01 9.99686271e-02 -3.96012038e-01
-1.20704949e+00 -1.60868585e+00 5.00119209e-01 -6.32363498e-01
6.46343887e-01 -6.51825428e-01 -1.07182014e+00 1.13290715e+00
-6.14220612e-02 6.81436896e-01 4.38689619e-01 2.56870210e-01
-8.19140375e-01 -2.59900421e-01 -7.55153835e-01 3.42942506e-01
1.25782406e+00 -5.46223700e-01 -5.11604249e-01 4.56283092e-01
6.79593027e-01 -3.16625953e-01 -7.43135154e-01 5.11598110e-01
4.81255680e-01 -7.35342324e-01 1.09541845e+00 -8.84408534e-01
2.24576741e-01 -6.59857631e-01 -2.65826702e-01 -6.65561974e-01
-4.30473655e-01 -1.28251538e-01 -2.73095220e-01 1.26174557e+00
3.74135435e-01 -3.40161771e-01 8.60985935e-01 4.48399097e-01
-2.01181471e-01 -1.06424856e+00 -8.01488638e-01 -9.62113261e-01
-1.28408358e-01 -5.26493192e-01 4.14427936e-01 7.75021970e-01
-4.65616733e-01 4.11027133e-01 -8.28882307e-02 5.69168627e-01
5.25799930e-01 7.52243757e-01 7.60980606e-01 -1.08972025e+00
-6.06421828e-01 -5.80255568e-01 -6.27603829e-01 -1.49194336e+00
7.08429962e-02 -1.18510175e+00 1.39195412e-01 -1.60994565e+00
3.22135240e-01 -6.93647861e-01 -3.83454680e-01 9.06978250e-01
-1.56759545e-01 3.94904554e-01 3.08230102e-01 4.25150037e-01
-9.41926897e-01 3.91480178e-01 1.17215955e+00 -2.36514464e-01
6.95953444e-02 1.33131444e-01 -6.66357517e-01 1.01929796e+00
5.97872853e-01 -4.45988923e-01 -2.26654634e-01 -6.55073464e-01
-1.74362883e-01 -3.23430151e-01 9.33713675e-01 -1.02144146e+00
2.39572525e-01 -1.34282097e-01 4.03054655e-01 -6.25286758e-01
2.07270935e-01 -9.28867459e-01 -8.21872279e-02 2.31795982e-01
-3.14444453e-01 -1.82803303e-01 3.35982710e-01 6.53405249e-01
-2.79623479e-01 -3.94798338e-01 8.87895882e-01 -1.40245274e-01
-1.03825688e+00 5.35612941e-01 4.05451544e-02 -1.10530995e-01
1.20916259e+00 -2.19787672e-01 -2.05955893e-01 2.26152167e-01
-9.92249250e-01 2.46102154e-01 4.67539251e-01 4.15624529e-01
5.10261238e-01 -1.14962459e+00 -2.98130691e-01 1.14054784e-01
4.32997495e-01 3.50294083e-01 2.17119351e-01 8.21109056e-01
-4.63338077e-01 5.07270455e-01 2.90182512e-02 -9.49073315e-01
-1.31625903e+00 8.55987549e-01 4.44881737e-01 -5.00433631e-02
-7.52605677e-01 9.61708069e-01 9.43672955e-01 -5.04082322e-01
4.50926960e-01 -6.11997128e-01 1.30397350e-01 -5.10647930e-02
4.39909011e-01 1.97132751e-01 2.15359777e-02 -5.65488458e-01
-5.70962131e-01 8.00765336e-01 -4.10462976e-01 3.93392801e-01
1.35343969e+00 1.25396885e-02 -2.25957826e-01 3.38139862e-01
1.24410498e+00 -2.28902847e-01 -1.61532152e+00 -4.24578100e-01
4.65280324e-01 -3.99485946e-01 -1.00842319e-01 -8.15194964e-01
-1.23808503e+00 1.14871895e+00 5.63326836e-01 -7.93039724e-02
9.53749537e-01 5.90872169e-01 3.90343666e-01 6.26462281e-01
4.05877203e-01 -7.48259664e-01 2.72340864e-01 3.68078530e-01
9.44046855e-01 -1.24083948e+00 -8.28366727e-02 -7.12273121e-01
-5.89638591e-01 9.46625948e-01 8.49271297e-01 -3.00319731e-01
7.93678463e-01 3.77613246e-01 -9.16693807e-02 -4.18131143e-01
-7.50777423e-01 -3.49606007e-01 6.64390624e-01 4.25943255e-01
3.12937766e-01 8.31399113e-02 2.45823070e-01 5.53959787e-01
3.16530436e-01 -2.67949402e-01 4.15613912e-02 1.02424669e+00
-6.02980494e-01 -9.53234196e-01 -1.83612436e-01 3.00698131e-01
-3.26704443e-01 -8.92291144e-02 -3.59884858e-01 7.75412381e-01
3.67149144e-01 3.50067496e-01 3.84822249e-01 -3.69072482e-02
3.63232434e-01 5.29364171e-03 5.49240470e-01 -9.06871736e-01
-4.32798952e-01 1.58925101e-01 -1.46329790e-01 -8.45453084e-01
-3.81357402e-01 -6.68168426e-01 -1.39301634e+00 2.15861797e-01
-2.57927924e-01 -2.49024406e-01 4.56471920e-01 8.67971957e-01
6.73818171e-01 4.69336420e-01 4.18121874e-01 -9.50656474e-01
-2.42379293e-01 -7.23324180e-01 -4.60825175e-01 3.02573204e-01
3.55949879e-01 -6.95272028e-01 -1.54208979e-02 2.99979120e-01] | [9.500125885009766, 1.1366032361984253] |
6c9e0c91-178b-4b2d-a6b3-a337f97a76cc | a-robust-semantic-frame-parsing-pipeline-on-a | 2212.08987 | null | https://arxiv.org/abs/2212.08987v1 | https://arxiv.org/pdf/2212.08987v1.pdf | A Robust Semantic Frame Parsing Pipeline on a New Complex Twitter Dataset | Most recent semantic frame parsing systems for spoken language understanding (SLU) are designed based on recurrent neural networks. These systems display decent performance on benchmark SLU datasets such as ATIS or SNIPS, which contain short utterances with relatively simple patterns. However, the current semantic frame parsing models lack a mechanism to handle out-of-distribution (\emph{OOD}) patterns and out-of-vocabulary (\emph{OOV}) tokens. In this paper, we introduce a robust semantic frame parsing pipeline that can handle both \emph{OOD} patterns and \emph{OOV} tokens in conjunction with a new complex Twitter dataset that contains long tweets with more \emph{OOD} patterns and \emph{OOV} tokens. The new pipeline demonstrates much better results in comparison to state-of-the-art baseline SLU models on both the SNIPS dataset and the new Twitter dataset (Our new Twitter dataset can be downloaded from https://1drv.ms/u/s!AroHb-W6_OAlavK4begsDsMALfE?e=c8f2XX ). Finally, we also build an E2E application to demo the feasibility of our algorithm and show why it is useful in real application. | ['Hongxia Jin', 'Yu Wang'] | 2022-12-18 | null | null | null | null | ['spoken-language-understanding', 'spoken-language-understanding'] | ['natural-language-processing', 'speech'] | [-4.78625484e-02 3.55715960e-01 -2.04530403e-01 -6.06455564e-01
-9.28623915e-01 -7.68123984e-01 6.42581522e-01 2.10976806e-02
-3.56777996e-01 7.15745509e-01 6.21369123e-01 -8.01025987e-01
1.07290626e-01 -8.36839557e-01 -7.02449322e-01 -2.66372859e-01
3.55408698e-01 5.63106298e-01 3.59071136e-01 -7.88678050e-01
-1.64271295e-01 -2.36140385e-01 -1.73090613e+00 6.04714811e-01
1.43126741e-01 8.80968511e-01 2.67084181e-01 7.04787612e-01
-5.86459041e-01 8.30460310e-01 -7.30040789e-01 -1.51694730e-01
1.69881172e-02 -4.13797706e-01 -1.18625963e+00 -3.19886118e-01
4.08773363e-01 -2.74455041e-01 -2.60412216e-01 7.30150580e-01
5.66205263e-01 2.33930334e-01 3.49820346e-01 -1.05116308e+00
-4.00505573e-01 1.01977527e+00 -1.16324306e-01 5.71080804e-01
7.33070970e-01 -1.93608090e-01 1.21925843e+00 -9.86969471e-01
8.70368659e-01 1.58939660e+00 6.03294611e-01 7.88397014e-01
-7.82641947e-01 -8.84698093e-01 3.40749741e-01 1.58978641e-01
-9.10224676e-01 -6.42457068e-01 5.15656054e-01 6.21921234e-02
1.25672472e+00 4.41423506e-01 2.62089133e-01 1.43324685e+00
-2.03263715e-01 1.34049428e+00 1.05320811e+00 -5.48815727e-01
2.45479792e-01 -2.41316110e-01 9.46682572e-01 3.75878602e-01
-2.95668930e-01 -2.73643970e-01 -7.80361056e-01 1.21203177e-01
2.88351446e-01 -1.43279448e-01 -2.63343364e-01 7.50073195e-01
-1.10837448e+00 1.09482944e+00 -1.24262825e-01 5.55458307e-01
2.04860806e-01 2.37400413e-01 4.77803618e-01 3.39328974e-01
5.72751641e-01 8.87382850e-02 -6.60732150e-01 -6.33667767e-01
-6.21831954e-01 5.65128207e-01 9.45372581e-01 1.25573182e+00
6.09999657e-01 1.73994094e-01 -5.04642203e-02 1.09728789e+00
3.10790598e-01 6.49087787e-01 6.08325839e-01 -1.02253759e+00
6.17554486e-01 2.95023084e-01 7.88673759e-02 -5.13525605e-01
-6.93825483e-01 1.29369916e-02 -2.41932660e-01 -3.67086977e-01
8.50429416e-01 -4.51906502e-01 -8.20588291e-01 1.88722980e+00
2.56594092e-01 1.94253270e-02 4.24723625e-01 6.76835716e-01
1.61840439e+00 1.00761461e+00 2.36154154e-01 -4.34245765e-02
1.59926462e+00 -8.70077372e-01 -1.08093488e+00 -4.12019342e-01
8.90365064e-01 -9.65616524e-01 1.51792240e+00 3.39809805e-02
-1.11690617e+00 -3.48246247e-01 -6.28736556e-01 -4.35220838e-01
-6.09503508e-01 -2.79124945e-01 4.99023736e-01 6.33574963e-01
-1.13319397e+00 1.63066342e-01 -5.71463704e-01 -6.31484985e-01
2.91336514e-02 1.06312186e-01 -4.90101464e-02 5.66631593e-02
-1.73750746e+00 4.64248836e-01 3.41590136e-01 -8.18189234e-02
-4.84997839e-01 -7.21423566e-01 -1.21622419e+00 -1.53118908e-01
4.74899590e-01 -6.49111196e-02 1.78229320e+00 -7.70017684e-01
-1.51392329e+00 7.97892928e-01 -6.60803735e-01 -5.03774107e-01
2.93136120e-01 -1.74899459e-01 -5.45519471e-01 -2.61439625e-02
1.69925600e-01 7.62327790e-01 4.94708329e-01 -8.97087455e-01
-8.90509427e-01 -1.48038611e-01 3.31474781e-01 1.82207957e-01
-2.66608335e-02 4.56381291e-01 -5.53082637e-02 -6.22367382e-01
1.04396246e-01 -8.44161153e-01 4.20534670e-01 -8.36271584e-01
-3.27488810e-01 -7.02680349e-01 1.07409894e+00 -4.98957962e-01
1.48394907e+00 -2.00722146e+00 -3.12984228e-01 -1.49729833e-01
-1.04122318e-01 3.40595990e-01 3.50361764e-02 6.17307127e-01
-5.20236976e-02 3.25333744e-01 -3.23576853e-02 -4.79780406e-01
2.99327850e-01 5.90966225e-01 -5.68940461e-01 -4.01072130e-02
-2.73350757e-02 8.02218616e-01 -1.00911677e+00 -5.91781661e-02
1.84290245e-01 2.82159925e-01 -5.70352614e-01 6.75453022e-02
-5.70421636e-01 3.39539319e-01 -3.22345257e-01 6.63442194e-01
3.28866333e-01 -8.97447094e-02 2.23887786e-01 1.33025974e-01
-4.46220577e-01 7.85535574e-01 -1.11172259e+00 1.51988053e+00
-3.30667734e-01 6.39722049e-01 6.07789541e-03 -5.88297009e-01
7.78073192e-01 5.32713771e-01 2.61025816e-01 -5.59221745e-01
3.98553044e-01 3.01851094e-01 -1.42747551e-01 -1.41192272e-01
7.55866230e-01 -1.15508027e-01 -6.17497325e-01 6.01712227e-01
7.20584691e-02 -3.88748586e-01 3.17723244e-01 1.30789563e-01
1.20060194e+00 -7.83454031e-02 1.62666544e-01 -4.87499118e-01
3.02065074e-01 -1.24115825e-01 4.83067989e-01 8.38028312e-01
-4.01891530e-01 7.32963204e-01 6.23784959e-01 -5.18203497e-01
-8.55238080e-01 -8.09918404e-01 -3.80612075e-01 1.49654341e+00
3.36351618e-02 -8.12784851e-01 -1.01965261e+00 -6.34097338e-01
-2.90239841e-01 1.25082529e+00 -3.93421143e-01 4.21652406e-01
-6.55460477e-01 -6.49765730e-01 9.37589407e-01 5.28358757e-01
5.17937124e-01 -1.41621077e+00 -4.91031587e-01 3.83799285e-01
-5.81309199e-01 -1.30434513e+00 -3.93412381e-01 2.93544143e-01
-1.50221244e-01 -8.13399792e-01 -4.25147086e-01 -8.06338549e-01
1.62891731e-01 1.82977065e-01 1.09084618e+00 -1.67817101e-02
2.18372285e-01 4.89670724e-01 -8.19990754e-01 -1.04877830e+00
-5.30600727e-01 2.35699311e-01 3.26138325e-02 -2.60636121e-01
7.74702132e-01 -7.94264898e-02 -2.13834509e-01 3.96340787e-01
-9.84497070e-01 2.37666667e-02 -5.56941628e-01 8.90357018e-01
3.80106300e-01 -3.42998207e-01 7.29330182e-01 -1.04183793e+00
5.20195544e-01 -5.81781983e-01 -3.40048730e-01 -6.25458732e-02
-2.18019057e-02 -2.37655461e-01 5.47766507e-01 -6.76524565e-02
-1.32336438e+00 -3.59156787e-01 -9.51472282e-01 -1.45917118e-01
-4.80650753e-01 5.26146352e-01 -1.32200971e-01 7.91763425e-01
4.86671209e-01 1.46894217e-01 -1.16164610e-01 -4.97004837e-01
2.51826406e-01 9.29304898e-01 3.23581129e-01 -4.40781116e-01
2.94741660e-01 4.74894136e-01 -6.85432911e-01 -1.38734186e+00
-1.14606535e+00 -3.80736947e-01 -3.17073584e-01 -1.00003548e-01
1.06164229e+00 -1.06959534e+00 -5.91317236e-01 7.61155307e-01
-1.19190788e+00 -7.95322478e-01 -3.45241815e-01 2.22885013e-01
-5.29800296e-01 3.48623335e-01 -8.58429432e-01 -6.97825432e-01
-3.21083874e-01 -1.02055025e+00 1.14703178e+00 2.61582613e-01
-7.35397995e-01 -1.09310269e+00 -3.55042666e-01 5.32356858e-01
4.40202475e-01 -3.47803198e-02 7.75619864e-01 -1.16397989e+00
-1.27388254e-01 2.24361509e-01 1.93517096e-02 1.34896830e-01
-1.16351694e-02 -6.30286857e-02 -1.15944934e+00 -4.39665765e-02
6.36062818e-03 -5.78179538e-01 6.67733192e-01 3.46917152e-01
8.88767302e-01 -3.26165676e-01 8.70506242e-02 3.39353293e-01
6.10246241e-01 3.81273508e-01 4.40496922e-01 3.45607758e-01
5.24541318e-01 5.34293532e-01 5.47059715e-01 5.96915960e-01
8.65254521e-01 5.67432523e-01 2.66268283e-01 2.39238575e-01
-2.83119947e-01 -2.79909760e-01 6.87667191e-01 9.49917793e-01
3.42301518e-01 -5.76010287e-01 -1.09783494e+00 6.54133976e-01
-1.95271027e+00 -8.96755695e-01 -3.80234003e-01 1.60851276e+00
8.57841969e-01 2.94695795e-01 -3.62329036e-02 1.68899283e-01
6.41521573e-01 6.60845459e-01 4.92935814e-02 -7.51545787e-01
-3.86849225e-01 4.33547169e-01 1.38150945e-01 8.52162242e-01
-1.09992015e+00 1.50100672e+00 5.67853594e+00 9.57353175e-01
-1.08708489e+00 3.73908222e-01 5.97332060e-01 -8.77900571e-02
-7.47310638e-01 -8.76930654e-02 -1.53132403e+00 7.03802466e-01
1.30542386e+00 1.50678456e-01 1.52653381e-01 6.67212427e-01
2.94567972e-01 -1.13113537e-01 -8.83064747e-01 8.52593482e-01
-1.09601505e-02 -1.63289154e+00 -1.83359385e-01 -2.89335996e-01
3.35256726e-01 3.30162078e-01 -8.88094828e-02 6.60994530e-01
5.18784702e-01 -8.18502784e-01 1.12060964e+00 -9.23130140e-02
8.47280324e-01 -5.58240473e-01 7.52451003e-01 3.71036410e-01
-1.13769209e+00 1.29475430e-01 -3.16169262e-01 -4.37326342e-01
3.97354811e-01 3.33179802e-01 -9.44896400e-01 2.99664199e-01
9.79789376e-01 6.71652675e-01 -1.47363842e-01 1.13008492e-01
-4.42841649e-01 1.00376010e+00 -8.04091156e-01 -3.60201061e-01
6.92397714e-01 7.67224878e-02 8.03118527e-01 1.40371239e+00
2.80213624e-01 5.21768868e-01 1.20392725e-01 5.62674463e-01
-8.35565925e-02 8.06220770e-02 -6.64020181e-01 7.43038878e-02
5.86215734e-01 8.73834074e-01 -9.67577636e-01 -3.27449173e-01
-5.55881500e-01 3.49037796e-01 1.95582345e-01 2.94147789e-01
-7.82625139e-01 -1.82507217e-01 8.61803353e-01 1.52624384e-01
3.90282512e-01 -2.35297754e-01 5.41639961e-02 -1.23273909e+00
-2.40868673e-01 -1.02213180e+00 6.51158631e-01 -6.31382823e-01
-1.02522981e+00 7.47253239e-01 2.33483538e-01 -7.45109200e-01
-5.87275207e-01 -6.45911157e-01 -5.87038815e-01 5.03725290e-01
-1.52003014e+00 -1.34398830e+00 -1.14348512e-02 7.35889554e-01
1.17029166e+00 -2.16226339e-01 9.68762279e-01 2.87396520e-01
-5.38177490e-01 5.47735751e-01 -4.82447911e-03 3.70057583e-01
6.28715038e-01 -1.23483062e+00 5.71365774e-01 7.58902013e-01
9.26492736e-02 3.15714121e-01 8.25676978e-01 -6.20465338e-01
-1.19114339e+00 -9.37104106e-01 1.33938062e+00 -7.88866699e-01
8.49339545e-01 -6.73260927e-01 -7.64362693e-01 1.14063728e+00
4.01538283e-01 -6.71681762e-02 7.00150371e-01 2.18532443e-01
-2.43038580e-01 2.73537397e-01 -9.30277646e-01 6.59756422e-01
1.13739455e+00 -4.68291789e-01 -9.68287051e-01 3.55476141e-01
9.75865006e-01 -8.63647103e-01 -6.26257420e-01 3.68203640e-01
4.58883137e-01 -1.03502381e+00 5.60098410e-01 -6.22308493e-01
2.66047686e-01 -1.28792793e-01 -5.42372406e-01 -1.02999508e+00
1.77727059e-01 -8.46993327e-01 7.83433467e-02 1.52025819e+00
6.41699731e-01 -7.06187904e-01 5.17337739e-01 5.53861082e-01
-5.04353344e-01 -4.00556266e-01 -1.25114489e+00 -3.91746461e-01
1.52293846e-01 -1.07022142e+00 5.32940865e-01 9.82655048e-01
2.48646826e-01 5.00092030e-01 -1.30668923e-01 -1.44028664e-01
1.50101572e-01 3.59738730e-02 5.78927875e-01 -7.19381154e-01
-3.39815617e-02 -3.42233717e-01 4.75431867e-02 -1.00371706e+00
5.80570459e-01 -8.59089255e-01 -1.39651164e-01 -1.32752740e+00
-4.18439299e-01 -4.98596191e-01 7.08254203e-02 9.01637316e-01
5.29967360e-02 1.98435523e-02 4.49940205e-01 2.03803014e-02
-5.85493267e-01 3.64733249e-01 7.16982782e-01 2.69941092e-02
-2.30045348e-01 7.11910129e-02 -5.50510466e-01 1.07203901e+00
6.54758394e-01 -2.83527553e-01 -4.00010705e-01 -6.11020744e-01
7.21430555e-02 2.63286382e-02 -2.56033707e-02 -6.25938892e-01
3.42026390e-02 1.29410818e-01 -4.05735187e-02 -7.82301128e-01
4.00044292e-01 -4.61141407e-01 -4.21689861e-02 2.11121023e-01
-2.51759291e-01 3.15460563e-01 4.48512465e-01 1.49243608e-01
-4.25026059e-01 -3.00000459e-01 5.16285121e-01 -3.84271771e-01
-1.07954228e+00 -1.50774792e-01 -8.84479165e-01 4.60346162e-01
7.42860734e-01 -4.22690399e-02 -7.44939864e-01 -8.65829170e-01
-7.68207908e-01 4.45665240e-01 -1.91876702e-02 8.51949453e-01
3.47801775e-01 -1.01252806e+00 -6.00939751e-01 4.16287631e-01
6.07107058e-02 3.54587615e-01 2.99132854e-01 4.67350513e-01
-5.80875099e-01 4.00929689e-01 8.68802890e-03 -6.49197459e-01
-1.15345466e+00 -1.76897403e-02 9.64093581e-02 -2.65906274e-01
-5.00232518e-01 1.17572308e+00 1.80731863e-01 -8.81064951e-01
2.93893844e-01 -7.10788071e-01 -2.58483082e-01 6.52978003e-01
6.83110178e-01 2.16799140e-01 8.65379721e-02 -8.40426683e-01
-4.76592720e-01 9.59345549e-02 -1.08048178e-01 -1.34801835e-01
1.35798621e+00 -3.83376569e-01 1.72111139e-01 6.51095390e-01
1.16330230e+00 9.55005139e-02 -9.04927433e-01 -2.20437363e-01
1.16512433e-01 -1.03817999e-01 -1.73564568e-01 -6.04941547e-01
-5.24330616e-01 7.95243680e-01 2.89804995e-01 3.44915509e-01
6.10236645e-01 2.95981795e-01 1.15543187e+00 3.15630138e-01
3.69069487e-01 -1.30347633e+00 -1.07265055e-01 1.43507266e+00
7.79782474e-01 -1.26036692e+00 -4.27398533e-01 -4.29116517e-01
-7.73939669e-01 1.03248763e+00 4.19683129e-01 2.34998763e-01
9.40220356e-01 3.39617461e-01 5.32583356e-01 -2.64418066e-01
-6.67712808e-01 -3.37429583e-01 -2.11351648e-01 1.99942246e-01
5.45855463e-01 1.92707911e-01 -5.18672407e-01 7.51209736e-01
-8.04376900e-01 -3.34773898e-01 8.28517735e-01 1.13414168e+00
-3.88979882e-01 -1.21203530e+00 -3.18642139e-01 1.82336181e-01
-7.87947357e-01 -2.73254126e-01 -2.69280106e-01 1.12343121e+00
-6.82173967e-02 1.21782196e+00 3.70368212e-01 -1.28067911e-01
3.49054158e-01 6.28049910e-01 1.11138290e-02 -8.67792070e-01
-7.65181243e-01 5.64969122e-01 6.00043476e-01 -4.11546081e-01
-5.13992131e-01 -8.64016056e-01 -1.75700128e+00 -3.93895179e-01
8.94781053e-02 5.33694774e-02 4.16003078e-01 1.14078581e+00
1.49131835e-01 4.07672077e-01 2.35047579e-01 -6.90139294e-01
-1.72603294e-01 -9.78192508e-01 -5.39029598e-01 4.56471235e-01
3.30541313e-01 -4.90536898e-01 -5.73693633e-01 -1.78405121e-01] | [13.951425552368164, 7.066309452056885] |
e5e208b1-13ef-4307-9967-2e8ef3f85572 | binocular-tone-mapping-with-improved-overall | 1809.06036 | null | http://arxiv.org/abs/1809.06036v1 | http://arxiv.org/pdf/1809.06036v1.pdf | Binocular Tone Mapping with Improved Overall Contrast and Local Details | Tone mapping is a commonly used technique that maps the set of colors in
high-dynamic-range (HDR) images to another set of colors in low-dynamic-range
(LDR) images, to fit the need for print-outs, LCD monitors and projectors.
Unfortunately, during the compression of dynamic range, the overall contrast
and local details generally cannot be preserved simultaneously. Recently, with
the increased use of stereoscopic devices, the notion of binocular tone mapping
has been proposed in the existing research study. However, the existing
research lacks the binocular perception study and is unable to generate the
optimal binocular pair that presents the most visual content. In this paper, we
propose a novel perception-based binocular tone mapping method, that can
generate an optimal binocular image pair (generating left and right images
simultaneously) from an HDR image that presents the most visual content by
designing a binocular perception metric. Our method outperforms the existing
method in terms of both visual and time performance. | ['Tien-Tsin Wong', 'Xinghong Hu', 'Zhuming Zhang', 'Xueting Liu'] | 2018-09-17 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 4.33794409e-01 -6.06489241e-01 -1.32945143e-02 -1.25474976e-02
8.08580592e-02 -5.59139788e-01 3.03818524e-01 -4.26794231e-01
-4.24776524e-02 6.39492810e-01 -3.65647450e-02 -3.36951256e-01
-9.34605896e-02 -9.86964345e-01 -4.20132309e-01 -4.58297461e-01
4.51963991e-01 -3.86385232e-01 5.39158165e-01 -3.83609354e-01
5.49066484e-01 3.94478887e-01 -2.00052571e+00 1.63712781e-02
9.08160508e-01 9.11460221e-01 9.28448915e-01 7.41996109e-01
2.00238064e-04 6.75740480e-01 -4.53261435e-01 -1.06498376e-01
6.88163877e-01 -8.77846599e-01 -2.35490918e-01 8.62426311e-02
2.85497308e-01 -6.18362069e-01 -3.83561224e-01 1.34369922e+00
6.12604916e-01 4.30814782e-03 2.07291603e-01 -1.00833237e+00
-1.07727313e+00 -7.26437494e-02 -9.49147582e-01 1.85457274e-01
6.25912368e-01 1.33168429e-01 4.09871966e-01 -6.63081467e-01
5.19041419e-01 1.10717475e+00 3.31147723e-02 5.29355943e-01
-1.18582666e+00 -7.84355879e-01 -3.40893418e-01 2.25083426e-01
-1.52880800e+00 -2.33266011e-01 9.07132387e-01 -1.78086743e-01
5.90250075e-01 5.42210937e-01 9.38272774e-01 3.76884460e-01
3.33518028e-01 7.28484010e-03 1.96984112e+00 -5.71587265e-01
6.24723658e-02 3.83943886e-01 -5.32371461e-01 2.00103000e-01
1.64458275e-01 4.52017128e-01 -6.03734672e-01 3.81626368e-01
1.27897775e+00 5.54986112e-02 -5.94644845e-01 -1.71763882e-01
-1.00076592e+00 2.60575771e-01 2.27508336e-01 4.95479435e-01
-1.49276718e-01 -3.60717595e-01 -1.03675708e-01 5.80578983e-01
1.70935795e-01 4.00041640e-01 8.50814730e-02 -2.14955181e-01
-7.83036411e-01 -4.07774411e-02 4.58815902e-01 1.00522423e+00
7.01603413e-01 -1.15151398e-01 1.09841665e-02 8.61134291e-01
2.79700667e-01 6.40172601e-01 3.88303906e-01 -9.83843863e-01
3.38249624e-01 4.92880285e-01 2.40966678e-01 -1.06485534e+00
6.32748306e-02 -1.87328216e-02 -7.07804918e-01 7.26644635e-01
2.20164303e-02 1.06933191e-01 -6.84231877e-01 1.44912302e+00
1.50131106e-01 -3.77716757e-02 -1.04493732e-02 1.38239539e+00
6.64570570e-01 8.44727337e-01 -4.85164613e-01 -4.70986992e-01
1.16274273e+00 -5.91283798e-01 -1.00043905e+00 -1.10600553e-01
-3.75043869e-01 -1.33148777e+00 1.56836486e+00 5.30056715e-01
-1.24648559e+00 -8.83016288e-01 -1.37025750e+00 -3.21547300e-01
-2.45147303e-01 -1.39130577e-01 1.61148384e-01 8.71992707e-01
-1.34074092e+00 1.26803309e-01 -5.60417473e-02 -3.15656275e-01
-3.99201423e-01 1.20018478e-02 -1.21435501e-01 -3.59202325e-01
-1.10971236e+00 1.04878044e+00 1.90600455e-01 3.29270065e-02
-4.31758165e-01 -5.15313029e-01 -4.89455879e-01 -3.50448862e-02
2.27608606e-01 -4.50766802e-01 8.63407135e-01 -9.95663345e-01
-1.93310332e+00 8.52865219e-01 -8.19819868e-02 1.86467737e-01
5.13810337e-01 6.79231361e-02 -5.50971270e-01 2.21339419e-01
-2.69309193e-01 6.44986153e-01 6.58246815e-01 -1.41506302e+00
-6.11672580e-01 -3.49166036e-01 1.93129092e-01 6.56675637e-01
-1.72386259e-01 2.02942833e-01 -6.96809173e-01 -4.33106542e-01
2.64881343e-01 -7.25712299e-01 2.48013690e-01 2.21668724e-02
-3.72895390e-01 2.11647496e-01 1.11130655e+00 -5.62439442e-01
1.42813528e+00 -2.41609144e+00 -1.20863505e-01 4.96973582e-02
1.15686424e-01 1.98689505e-01 2.04151616e-01 3.67863685e-01
-3.11154965e-02 -1.00212410e-01 9.85012725e-02 1.15506344e-01
-2.18929157e-01 -1.60719335e-01 -3.16929966e-01 3.98893207e-01
-2.56784290e-01 3.86131734e-01 -6.73288167e-01 -6.02211416e-01
6.21891916e-01 4.58253562e-01 -2.27217674e-01 3.99995804e-01
1.89878449e-01 5.36920369e-01 -6.46189898e-02 6.41700804e-01
1.22792184e+00 1.05175667e-01 9.15926099e-02 -1.86714053e-01
-7.27468371e-01 -1.60886258e-01 -1.06773305e+00 1.29455686e+00
-4.86231208e-01 9.58543181e-01 -1.14111222e-01 -1.29779160e-01
1.48387170e+00 8.18381608e-02 2.16889679e-01 -1.56030798e+00
-2.95817032e-02 4.10516262e-01 -7.05694258e-02 -6.21576607e-01
8.43541026e-01 -2.77740777e-01 1.87629476e-01 2.93533027e-01
-7.51994967e-01 -2.40966663e-01 1.53025657e-01 -1.90574631e-01
4.47985172e-01 1.90968305e-01 3.75493407e-01 -2.31490865e-01
5.17719269e-01 -2.67891467e-01 2.87963361e-01 4.10344571e-01
-1.69549227e-01 9.62050080e-01 2.25810945e-01 -2.59842873e-01
-1.54730105e+00 -1.20256019e+00 -3.70512605e-01 6.84525371e-01
1.06333828e+00 2.09183916e-02 -5.70567548e-01 3.47241580e-01
-2.99645245e-01 4.41385239e-01 -1.96512431e-01 -3.42252180e-02
-4.04887319e-01 -1.05555646e-01 2.18460351e-01 9.71829668e-02
9.36703622e-01 -1.08274901e+00 -9.83514428e-01 -9.17268991e-02
-2.45142907e-01 -8.65256846e-01 -6.32180452e-01 -3.60149533e-01
-6.24292195e-01 -9.87585008e-01 -9.74908292e-01 -9.07035708e-01
6.87692106e-01 8.91433656e-01 7.83761322e-01 5.32133989e-02
-3.27319741e-01 2.36715768e-02 -4.45438802e-01 -1.61732748e-01
-2.75209397e-01 -4.78954524e-01 -2.10943833e-01 -3.81438397e-02
3.31962973e-01 -4.34727669e-01 -9.78595912e-01 5.97350001e-01
-1.08101332e+00 6.17351115e-01 5.84964395e-01 4.30330604e-01
8.47088099e-01 3.38960201e-01 2.26305723e-01 -3.83839548e-01
7.39885569e-01 7.92186055e-03 -8.43452632e-01 3.40216398e-01
-6.67747319e-01 -3.15501183e-01 7.71783650e-01 -3.74582261e-01
-1.19339168e+00 -5.43752491e-01 3.35188061e-01 -3.75138849e-01
3.80812362e-02 -1.02062404e-01 -2.62978941e-01 -3.16281527e-01
4.81414229e-01 6.76559389e-01 -2.25380082e-02 -3.96880686e-01
1.60088435e-01 1.03290713e+00 6.62740648e-01 -1.60063967e-01
7.41435468e-01 3.76658648e-01 1.18660301e-01 -6.29280269e-01
-1.13989167e-01 -2.85417527e-01 -1.40336022e-01 -5.11168063e-01
8.94928277e-01 -8.49221230e-01 -7.94835448e-01 6.46019697e-01
-8.69838059e-01 -2.69729458e-02 3.27038132e-02 6.56672120e-01
-6.19382024e-01 3.17703068e-01 -3.99550229e-01 -8.06863010e-01
-1.55615166e-01 -1.13840616e+00 9.15769517e-01 7.11541593e-01
2.62026340e-01 -4.22422022e-01 7.76410401e-02 4.60925885e-02
7.20469117e-01 1.06529564e-01 7.55240917e-01 5.92699826e-01
-9.79061723e-01 1.93114817e-01 -6.13835454e-01 2.04877019e-01
5.45096993e-01 3.31587456e-02 -8.58942568e-01 -3.15584868e-01
1.46593004e-01 3.03024612e-03 4.42890406e-01 4.83360648e-01
1.23390651e+00 -6.04891740e-02 1.22861762e-03 7.71298647e-01
1.97000551e+00 8.92372966e-01 1.41035867e+00 5.19071639e-01
3.49752456e-01 6.10960007e-01 8.04976285e-01 2.97928929e-01
3.52659971e-01 9.51482236e-01 2.14898810e-01 -5.26380837e-01
-3.99329454e-01 -4.18425530e-01 1.69447094e-01 5.70513129e-01
-2.14162782e-01 -2.70828962e-01 -4.18541253e-01 2.08792344e-01
-1.22558331e+00 -1.00479972e+00 3.96135673e-02 2.51872277e+00
9.44614112e-01 -2.98641939e-02 -1.06938235e-01 1.38793617e-01
1.20640230e+00 -7.58631229e-02 -5.49221814e-01 -6.42524123e-01
-2.56822973e-01 -4.35656533e-02 4.14873391e-01 4.09221888e-01
-5.30944645e-01 5.42190373e-01 6.53996181e+00 6.29220247e-01
-1.51029909e+00 -2.48014599e-01 6.03411555e-01 -9.30114463e-02
-5.95798433e-01 1.19971685e-01 -4.49112713e-01 8.70518923e-01
3.95425618e-01 -3.45240057e-01 8.50323141e-01 4.91957843e-01
5.59419692e-01 -4.92028922e-01 -7.09869862e-01 1.46484160e+00
1.63138956e-01 -7.53840744e-01 -1.18370458e-01 2.79441208e-01
8.26240361e-01 -8.40760648e-01 6.05230927e-01 -2.62044013e-01
-3.01637083e-01 -9.15639699e-01 6.97793782e-01 5.95588744e-01
1.24976194e+00 -7.79752970e-01 4.66613203e-01 1.28499968e-02
-1.19988906e+00 1.07582230e-02 -5.14258623e-01 -4.74122204e-02
7.49158785e-02 3.82226497e-01 -4.90464211e-01 3.42501760e-01
9.88907695e-01 3.38556170e-01 -5.47818422e-01 1.10820770e+00
9.46897641e-02 2.39084624e-02 1.69958957e-02 -2.32158769e-02
-2.11114466e-01 -5.13990760e-01 4.41492677e-01 6.61989391e-01
6.86512291e-01 2.05591857e-01 -1.62481703e-02 1.15569317e+00
1.33789286e-01 1.76929161e-01 -8.99762869e-01 2.62988955e-01
7.55891860e-01 9.20282125e-01 -7.33742118e-01 -1.35366425e-01
-4.39216256e-01 9.33890879e-01 -4.46343929e-01 3.43494058e-01
-7.72679389e-01 -7.39011168e-01 1.93118602e-01 3.44406158e-01
1.19842321e-01 -1.72969684e-01 -6.13120198e-01 -8.09592247e-01
1.89046115e-01 -9.25045133e-01 -1.74950182e-01 -1.33884454e+00
-8.23717654e-01 5.02916098e-01 -3.01324967e-02 -1.81562304e+00
9.58864763e-02 -2.53365874e-01 -4.50986475e-01 1.21519506e+00
-1.70065904e+00 -8.48680973e-01 -6.72532201e-01 8.32089067e-01
4.27819371e-01 3.13758813e-02 3.54036152e-01 4.63736683e-01
-8.76564309e-02 3.89477670e-01 1.36385992e-01 -4.14176375e-01
9.24994946e-01 -1.07775605e+00 -3.45723927e-02 1.05915749e+00
-5.75948417e-01 6.71236575e-01 7.84016490e-01 -5.35666525e-01
-1.40462685e+00 -5.01499593e-01 7.62330949e-01 3.01651716e-01
1.10458154e-02 -2.06598669e-01 -7.67877996e-01 8.35551769e-02
4.67331171e-01 -3.84423763e-01 5.37137151e-01 -4.99484122e-01
-2.18451992e-01 -5.79431355e-01 -1.38021636e+00 8.62433016e-01
9.54279900e-01 -7.64797866e-01 -2.24729791e-01 -2.71291167e-01
7.43687868e-01 -4.96671677e-01 -6.94261849e-01 2.59752184e-01
9.12367165e-01 -1.49893677e+00 7.55497813e-01 5.77251971e-01
6.26945674e-01 -8.88648868e-01 -2.45275795e-01 -1.02224433e+00
-2.01507151e-01 -5.84235489e-01 4.20240462e-01 1.13607037e+00
1.09817468e-01 -7.61710286e-01 1.42129019e-01 3.40600491e-01
1.12991955e-03 -4.15122986e-01 -6.30335927e-01 -7.26767540e-01
-5.00680149e-01 2.11249277e-01 7.71325052e-01 6.66407168e-01
-6.06020428e-02 -1.14818633e-01 -7.41594851e-01 4.39060330e-02
5.14799714e-01 5.35441041e-01 7.24405408e-01 -8.15001845e-01
-3.21848243e-01 -3.02144766e-01 -4.80455637e-01 -1.03007925e+00
-6.65649235e-01 -2.42614821e-01 7.26432055e-02 -1.49037349e+00
5.00956714e-01 -3.73267621e-01 -1.97605729e-01 -8.12535658e-02
-1.26557425e-01 5.88338494e-01 5.01377285e-01 3.87566179e-01
-1.93344653e-01 2.72514105e-01 1.96140826e+00 1.13584653e-01
-4.51451331e-01 -3.57401162e-01 -8.71139348e-01 1.42560214e-01
6.74289405e-01 -2.41412353e-02 -7.69090056e-01 -3.55537295e-01
2.94466525e-01 4.16866511e-01 3.87311667e-01 -1.11813402e+00
2.80354440e-01 -4.31507409e-01 6.07740521e-01 -6.61688328e-01
2.18577281e-01 -8.54952693e-01 6.24566317e-01 4.87111628e-01
-1.32584527e-01 3.86044860e-01 -8.41405541e-02 2.65300781e-01
-3.35594147e-01 3.86576317e-02 1.07216716e+00 -1.24731228e-01
-9.50075567e-01 2.45919246e-02 -8.18113461e-02 -4.74138886e-01
1.28124774e+00 -9.40113485e-01 -6.34752929e-01 -5.00332654e-01
-1.72729548e-02 -2.95022488e-01 1.12004101e+00 5.91230750e-01
9.45451975e-01 -1.40698922e+00 -3.76923949e-01 5.22249997e-01
-3.23364548e-02 -3.43642414e-01 6.30517840e-01 5.55425584e-01
-1.00522590e+00 2.37616345e-01 -7.52398014e-01 -5.10462344e-01
-1.30162823e+00 6.85497820e-01 2.70377934e-01 1.41859978e-01
-7.50970840e-01 2.02679828e-01 4.04128730e-01 2.60397375e-01
-1.69735506e-01 -1.12498730e-01 -3.92461568e-01 -4.85549480e-01
7.15258598e-01 4.60469246e-01 -2.57903755e-01 -5.01393318e-01
-1.14343017e-01 1.15063083e+00 2.81277120e-01 -3.81716460e-01
7.29440451e-01 -8.61926198e-01 -2.52890229e-01 3.53008747e-01
1.19456363e+00 2.82747775e-01 -1.18683171e+00 1.44314170e-01
-7.03490913e-01 -1.33947897e+00 -1.91007107e-02 -6.81364596e-01
-9.82341111e-01 7.83550858e-01 1.21512222e+00 5.08191884e-01
1.81451023e+00 -2.96728969e-01 7.81448066e-01 -2.58980244e-01
6.45907402e-01 -1.25106180e+00 2.80059189e-01 3.94740105e-02
7.10414946e-01 -9.92526412e-01 -1.87245935e-01 -4.62375969e-01
-5.36168754e-01 1.13007581e+00 8.62936378e-01 1.10065632e-01
3.25656444e-01 3.06330889e-01 1.50440603e-01 -4.49979305e-02
-4.69013929e-01 -2.87081033e-01 8.42258930e-02 7.85749733e-01
3.13879013e-01 4.12377305e-02 -6.65895462e-01 -1.60538450e-01
-3.77702951e-01 1.59869179e-01 7.53876984e-01 7.39050150e-01
-7.08449185e-01 -9.44498777e-01 -6.38533294e-01 7.22644627e-02
-3.14748287e-01 -9.60577801e-02 -1.89390302e-01 5.35827756e-01
3.69652420e-01 1.21627414e+00 2.94497669e-01 -5.52025020e-01
3.06462646e-01 -6.06334031e-01 6.39454126e-01 -2.03051165e-01
9.12765265e-02 1.30197242e-01 -3.59774649e-01 -3.44535410e-01
-5.17677605e-01 -7.43522048e-02 -9.27159131e-01 -7.48378873e-01
-2.61483192e-01 -3.20480764e-01 7.16540098e-01 5.09095848e-01
2.51268655e-01 4.07249689e-01 1.12314558e+00 -4.13555592e-01
3.32679898e-02 -6.62183762e-01 -1.01437354e+00 3.75934124e-01
3.57221901e-01 -5.05664170e-01 -2.73012221e-01 1.25722528e-01] | [10.867596626281738, -2.4523656368255615] |
eedaabd1-5c80-4bc6-a404-5112ccc544e9 | end-to-end-context-aided-unicity-matching-for | 2210.12008 | null | https://arxiv.org/abs/2210.12008v2 | https://arxiv.org/pdf/2210.12008v2.pdf | End-to-End Context-Aided Unicity Matching for Person Re-identification | Most existing person re-identification methods compute the matching relations between person images across camera views based on the ranking of the pairwise similarities. This matching strategy with the lack of the global viewpoint and the context's consideration inevitably leads to ambiguous matching results and sub-optimal performance. Based on a natural assumption that images belonging to the same person identity should not match with images belonging to multiple different person identities across views, called the unicity of person matching on the identity level, we propose an end-to-end person unicity matching architecture for learning and refining the person matching relations. First, we adopt the image samples' contextual information in feature space to generate the initial soft matching results by using graph neural networks. Secondly, we utilize the samples' global context relationship to refine the soft matching results and reach the matching unicity through bipartite graph matching. Given full consideration to real-world person re-identification applications, we achieve the unicity matching in both one-shot and multi-shot settings of person re-identification and further develop a fast version of the unicity matching without losing the performance. The proposed method is evaluated on five public benchmarks, including four multi-shot datasets MSMT17, DukeMTMC, Market1501, CUHK03, and a one-shot dataset VIPeR. Experimental results show the superiority of the proposed method on performance and efficiency. | ['Qi Tian', 'Junchi Yan', 'Chen Chen', 'Cong Ding', 'Min Cao'] | 2022-10-20 | null | null | null | null | ['graph-matching'] | ['graphs'] | [ 1.52123436e-01 -4.24756795e-01 -1.16675282e-02 -4.86640066e-01
-2.38223016e-01 -3.64265710e-01 5.68184376e-01 -6.72406470e-03
-5.64824700e-01 2.93276489e-01 4.51210946e-01 5.10453999e-01
-3.43228132e-01 -7.47610271e-01 -2.72865057e-01 -4.67410654e-01
2.51902461e-01 4.42671925e-01 7.60134310e-02 -1.72490537e-01
6.91390485e-02 1.77589543e-02 -1.68810546e+00 1.33856595e-01
7.95809448e-01 7.05675185e-01 -9.51173082e-02 3.12890917e-01
2.18029335e-01 2.08527565e-01 -3.16968054e-01 -1.18411243e+00
6.60076439e-01 -3.51672471e-01 -8.33505750e-01 3.06713611e-01
7.87450612e-01 -2.20999077e-01 -6.22998595e-01 1.57901812e+00
6.70037508e-01 4.76195782e-01 2.13029310e-01 -1.36637151e+00
-1.05280912e+00 3.22139353e-01 -7.34459698e-01 3.58746380e-01
9.71689224e-01 1.69037953e-01 9.21789944e-01 -8.32397223e-01
7.31496334e-01 1.44604993e+00 9.63028014e-01 6.05936050e-01
-1.00188720e+00 -7.46174872e-01 3.20742965e-01 6.27328694e-01
-1.79991269e+00 -4.15294915e-01 5.63986599e-01 -3.35480779e-01
6.17521822e-01 3.45988423e-01 6.51674390e-01 8.11025739e-01
-3.81740570e-01 4.85058695e-01 9.29975986e-01 -2.57579148e-01
-3.53659719e-01 9.61254910e-02 3.93379807e-01 6.61790550e-01
2.95363724e-01 1.90105498e-01 -5.08675337e-01 -2.38412306e-01
5.81590593e-01 3.90870810e-01 -4.84181464e-01 -1.82176322e-01
-1.36445248e+00 4.38685536e-01 5.81062555e-01 2.81215668e-01
-5.43606691e-02 -4.26054627e-01 4.27521020e-01 3.70891184e-01
9.63186100e-03 9.62923914e-02 2.12796018e-01 3.23528707e-01
-6.51082456e-01 2.14819893e-01 4.67371434e-01 1.15930593e+00
8.70195627e-01 -5.42733073e-01 -3.52323383e-01 1.00860155e+00
1.94682091e-01 3.74662876e-01 7.00747967e-01 -5.31030834e-01
6.05215371e-01 9.11019385e-01 2.22263634e-01 -1.63763094e+00
-2.56806225e-01 -2.89519489e-01 -1.20488381e+00 -4.80467409e-01
5.50277948e-01 1.14922166e-01 -6.54852629e-01 1.80413926e+00
4.30460453e-01 7.26192772e-01 5.03737889e-02 1.29057753e+00
1.26155090e+00 3.76398802e-01 2.15354748e-02 -7.34779462e-02
1.57042086e+00 -8.96381259e-01 -5.50172567e-01 -6.64749593e-02
3.64484906e-01 -6.42086267e-01 6.40873373e-01 -1.62986830e-01
-7.58596957e-01 -1.00834870e+00 -1.00074458e+00 1.45669207e-01
-3.92515600e-01 5.87993674e-02 4.81270969e-01 8.34930539e-01
-9.21881497e-01 5.68031132e-01 -3.19458619e-02 -7.70160556e-01
1.21251002e-01 5.14240265e-01 -8.73750687e-01 -3.13560665e-01
-1.51422119e+00 6.43608034e-01 5.74822366e-01 3.06648344e-01
-1.48558870e-01 -5.91715753e-01 -9.41463292e-01 -8.78352765e-03
2.69256055e-01 -9.03389096e-01 5.28187275e-01 -1.04669261e+00
-9.93620157e-01 1.30559099e+00 -1.35328040e-01 -1.57571882e-01
5.99800348e-01 2.22073331e-01 -8.86203587e-01 3.33605222e-02
2.76680678e-01 4.73745108e-01 4.24156547e-01 -1.13275957e+00
-9.50333595e-01 -6.47269428e-01 1.42295808e-01 5.38134575e-01
-4.47585851e-01 2.10155591e-01 -9.26835597e-01 -5.60999095e-01
1.23534784e-01 -9.30249810e-01 -1.81323253e-02 -2.91351646e-01
-4.10961181e-01 -3.52505147e-01 3.23680401e-01 -8.45375001e-01
1.03645647e+00 -1.96324575e+00 2.57292390e-01 5.17425418e-01
1.60832837e-01 2.19461247e-01 -2.70677835e-01 2.06554666e-01
-3.21779162e-01 -2.19986513e-01 2.31366772e-02 -4.17362571e-01
-1.31405503e-01 -1.95647374e-01 2.44036809e-01 6.42771542e-01
-3.65779638e-01 9.45877492e-01 -1.00750470e+00 -6.74043179e-01
3.06951165e-01 1.80688858e-01 -3.15218240e-01 2.34796926e-01
7.75170982e-01 3.19095790e-01 -8.12840685e-02 6.81474268e-01
8.09591711e-01 -2.36981973e-01 2.95341790e-01 -6.72003329e-01
3.14740866e-01 -5.85153461e-01 -1.77982712e+00 1.61497068e+00
7.27740154e-02 8.78258049e-02 -2.95331091e-01 -9.05529201e-01
9.14129496e-01 1.61869839e-01 6.06614649e-01 -7.78959870e-01
1.13445908e-01 -1.29520103e-01 -2.14776024e-01 -4.14641947e-01
6.07035995e-01 -1.12908572e-01 -1.10356472e-01 3.56849998e-01
2.50527024e-01 7.89280891e-01 2.92398095e-01 2.60925323e-01
3.49257380e-01 -1.11295633e-01 3.56756300e-01 -2.05952853e-01
1.03162301e+00 -4.23617542e-01 9.02852535e-01 7.51198888e-01
-5.70824087e-01 7.26134121e-01 -5.40966913e-02 -5.98735213e-01
-9.89381671e-01 -9.53919232e-01 1.02922924e-01 9.74469125e-01
1.03288639e+00 -5.38233221e-01 -6.86573565e-01 -6.12055004e-01
1.39267799e-02 1.30989075e-01 -6.77756727e-01 -3.06765020e-01
-4.13357764e-01 -9.69746292e-01 5.60699463e-01 2.80192047e-01
1.00085890e+00 -7.00585842e-01 2.19207093e-01 -1.17992483e-01
-6.09288335e-01 -1.32911253e+00 -1.26855958e+00 -9.83631492e-01
-1.95174530e-01 -1.37587142e+00 -1.02629864e+00 -1.12572634e+00
8.62159252e-01 7.39144921e-01 9.00339067e-01 5.11474192e-01
-3.54242623e-01 6.98998690e-01 -1.69601142e-01 2.73037493e-01
1.05319461e-02 -2.16295347e-01 5.44428706e-01 7.37260938e-01
7.84417748e-01 -3.72377306e-01 -7.60144353e-01 6.85293734e-01
-4.98155683e-01 7.87150711e-02 1.60049662e-01 9.09679949e-01
5.03875315e-01 2.26525724e-01 5.54667830e-01 -4.99937028e-01
5.86560369e-01 -2.71795601e-01 -3.27841908e-01 7.44683504e-01
-6.41156197e-01 -3.51974398e-01 4.59971100e-01 -6.25874221e-01
-1.16669881e+00 3.11218258e-02 7.82358795e-02 -3.72596264e-01
-9.81123745e-02 1.37919813e-01 -4.46891636e-01 -3.26358944e-01
3.92189175e-01 4.47330981e-01 -1.48656055e-01 -2.51543224e-01
2.33252376e-01 6.42937899e-01 1.04830289e+00 -4.59887356e-01
9.73733604e-01 5.20693004e-01 -2.54268914e-01 -3.87599558e-01
-4.95634258e-01 -9.56887066e-01 -8.28750014e-01 -4.87988055e-01
8.64674270e-01 -1.12634468e+00 -1.17330432e+00 7.61410475e-01
-8.89438093e-01 4.89669412e-01 1.03107952e-01 4.10398901e-01
-1.83952242e-01 1.13009810e+00 -5.67948520e-01 -5.13011694e-01
-6.02404892e-01 -9.40781891e-01 7.71645248e-01 7.72250593e-01
1.55352363e-02 -7.82635272e-01 7.00078830e-02 6.36183321e-01
-1.19546466e-01 1.05191201e-01 4.21283573e-01 -5.57572424e-01
-3.11772108e-01 -4.85086828e-01 -5.81847906e-01 -1.70285434e-01
3.39350611e-01 -4.44709986e-01 -7.45045304e-01 -5.58972776e-01
-4.44410712e-01 4.12621535e-02 7.73114443e-01 -4.35255989e-02
7.51869977e-01 -2.18978420e-01 -4.44357306e-01 8.59461308e-01
1.48780560e+00 -7.02340975e-02 5.99416673e-01 4.27481949e-01
1.13820958e+00 8.04267704e-01 4.89589125e-01 4.00429070e-01
8.11667502e-01 9.08612847e-01 -3.91183533e-02 1.07973956e-01
-1.11783259e-01 -4.83368307e-01 2.08566114e-02 7.27831244e-01
-4.32381779e-01 1.77820064e-02 -6.25432312e-01 6.04430199e-01
-2.17085457e+00 -1.50010753e+00 -8.78822431e-02 2.54908681e+00
5.75350881e-01 -2.59799033e-01 4.36003625e-01 -1.91214472e-01
1.52241480e+00 -3.99213731e-02 -3.52401435e-01 1.82420358e-01
-2.78070241e-01 -4.76900131e-01 3.98198754e-01 4.00428295e-01
-1.25435126e+00 8.16530645e-01 5.27708006e+00 6.89950109e-01
-6.08127713e-01 7.47175217e-02 5.79481363e-01 7.91903883e-02
-1.60853609e-01 -1.79434419e-01 -9.44952965e-01 7.69991040e-01
2.69876331e-01 -4.37975645e-01 6.08862758e-01 4.53101695e-01
-2.53270388e-01 1.58022791e-01 -1.07496119e+00 1.79065001e+00
4.83293444e-01 -1.09740734e+00 2.74312943e-01 -2.13265359e-01
7.29609370e-01 -5.47990859e-01 -9.75439996e-02 2.46765643e-01
1.39223576e-01 -8.56635869e-01 4.55726862e-01 7.64106572e-01
7.92673051e-01 -9.28107679e-01 1.01832521e+00 2.00360298e-01
-1.76221502e+00 -1.37469456e-01 -5.75064003e-01 8.12888294e-02
1.39726430e-01 6.72863331e-03 -4.53713834e-01 1.14465845e+00
9.64382529e-01 9.41298485e-01 -9.11915958e-01 1.13575113e+00
3.23192090e-01 -1.85174957e-01 -1.19125009e-01 2.96125889e-01
-1.90809324e-01 -4.56493706e-01 4.81681526e-01 1.16000664e+00
2.20987335e-01 3.53525043e-01 3.48692030e-01 6.41046226e-01
-5.12040406e-02 2.68403262e-01 -3.40472758e-01 4.87794966e-01
3.99811834e-01 1.15041411e+00 -5.42512417e-01 -4.07583743e-01
-6.14330769e-01 1.31978631e+00 3.21135044e-01 4.10234153e-01
-7.56829143e-01 -2.79873699e-01 6.95451319e-01 -1.89143911e-01
-1.14934973e-01 2.42029190e-01 8.48779231e-02 -1.60018671e+00
1.35904118e-01 -9.67148900e-01 1.06484818e+00 -5.10691524e-01
-1.88432086e+00 4.91176605e-01 3.96887437e-02 -1.36076450e+00
1.08240824e-03 -3.09144437e-01 -5.90293407e-01 1.00716341e+00
-1.13585401e+00 -1.48994637e+00 -8.35397899e-01 1.13355291e+00
2.23267108e-01 -5.11389494e-01 7.09977806e-01 6.74534261e-01
-8.09467256e-01 1.19817662e+00 -2.70101756e-01 6.53894603e-01
9.98869658e-01 -9.50319588e-01 5.36377013e-01 1.18386483e+00
6.73926622e-02 9.48843658e-01 4.82346624e-01 -9.25329387e-01
-1.34835982e+00 -1.01917124e+00 9.02805865e-01 -3.45015436e-01
2.59055495e-01 -1.10293843e-01 -8.84399533e-01 5.60823083e-01
6.06034175e-02 -6.13174587e-02 7.90300727e-01 2.65078753e-01
-5.64717710e-01 -2.59916604e-01 -1.27698612e+00 7.05954492e-01
1.59660184e+00 -6.38572693e-01 -7.65249550e-01 2.10458100e-01
3.35502058e-01 -2.46416077e-01 -1.04462004e+00 3.99516404e-01
7.34959900e-01 -9.13473427e-01 1.39442921e+00 -7.14793801e-01
-1.27967298e-01 -5.40459454e-01 -1.10801950e-01 -9.88299668e-01
-8.14078927e-01 -4.49965328e-01 4.19797093e-01 1.60986078e+00
-2.16688320e-01 -6.87946558e-01 7.39995778e-01 1.04562747e+00
4.01674539e-01 -1.75448671e-01 -9.37209606e-01 -7.37860262e-01
-5.37503004e-01 1.79009423e-01 1.05492723e+00 1.26330209e+00
1.15410231e-01 1.59762099e-01 -8.16332757e-01 4.71265644e-01
1.16785848e+00 3.60274643e-01 8.63253415e-01 -1.23009562e+00
-2.75590926e-01 -4.12715882e-01 -8.81115139e-01 -7.94489622e-01
2.18426660e-01 -1.07875705e+00 -2.75928497e-01 -1.37084889e+00
9.61732507e-01 -3.03232729e-01 -5.45704901e-01 2.10589916e-01
-6.67043984e-01 4.59752798e-01 3.87422085e-01 5.67934394e-01
-8.35017443e-01 4.30524915e-01 1.03898430e+00 -4.67761010e-01
-1.21493638e-01 3.07764998e-03 -7.28329897e-01 5.17673135e-01
3.78166616e-01 -6.11675605e-02 -3.18389088e-01 -3.19512337e-01
-1.21100945e-02 -1.33697197e-01 6.38764858e-01 -1.03491926e+00
7.01002479e-01 -1.44025654e-01 6.78036451e-01 -2.44059369e-01
3.17485094e-01 -7.42512226e-01 5.19616783e-01 3.42020094e-01
-2.28524178e-01 1.20815761e-01 -2.02190414e-01 8.71015429e-01
-1.33082435e-01 -1.21500455e-01 8.19554687e-01 -2.76821613e-01
-1.34566891e+00 8.29373896e-01 4.33499724e-01 3.02485246e-02
1.09142184e+00 -7.79443741e-01 -1.98806882e-01 -3.28541934e-01
-7.75026321e-01 5.15661716e-01 6.61780000e-01 7.76700020e-01
7.09391296e-01 -1.81144965e+00 -9.83316958e-01 2.50371456e-01
5.42743146e-01 -7.12868631e-01 7.45879889e-01 6.01538837e-01
-6.34768009e-02 4.77190390e-02 -5.38204491e-01 -5.74970722e-01
-1.63428950e+00 8.36693227e-01 7.74243057e-01 -1.72893424e-02
-6.86593354e-01 8.17831278e-01 3.38164270e-01 -5.46425402e-01
1.55616730e-01 4.92542058e-01 -6.00178182e-01 6.78911135e-02
8.71789277e-01 3.73797655e-01 -2.63428479e-01 -1.33064723e+00
-6.08931363e-01 9.23496068e-01 -9.11853537e-02 2.56898582e-01
7.95847595e-01 -4.93258506e-01 -2.02489451e-01 -3.49406153e-02
1.13002336e+00 -2.79841036e-01 -7.78366685e-01 -7.34786570e-01
-1.19306475e-01 -1.05377734e+00 -5.74279547e-01 -3.15466464e-01
-1.16098607e+00 2.52122790e-01 9.35412288e-01 -6.32139742e-02
9.82867002e-01 -2.48182401e-01 9.54071701e-01 1.83785245e-01
6.50911987e-01 -1.23001909e+00 -3.93225439e-02 6.59189969e-02
4.89299715e-01 -1.55802000e+00 1.41909182e-01 -5.39071977e-01
-7.37862110e-01 8.69812012e-01 8.46016347e-01 -3.74782947e-03
4.76829320e-01 -4.64953303e-01 -1.90733761e-01 1.01162806e-01
-4.89777587e-02 -3.84249330e-01 6.86471283e-01 8.20145845e-01
-6.56406507e-02 1.56521603e-01 -1.94454595e-01 8.22761774e-01
-1.08803749e-01 -1.19256817e-01 4.45432886e-02 2.35833123e-01
-1.81304052e-01 -9.12971199e-01 -4.98377949e-01 3.68443489e-01
-1.92684531e-01 4.50151935e-02 -3.87923092e-01 4.40277308e-01
4.66220260e-01 1.16215980e+00 6.67855702e-03 -7.07058489e-01
4.60050166e-01 -1.70915931e-01 4.56199855e-01 -2.79225320e-01
-7.38417268e-01 -4.26156491e-01 3.23399529e-02 -4.15787965e-01
-5.45057058e-01 -6.66493893e-01 -8.92602205e-01 -7.47389197e-01
-2.99483687e-01 1.78064778e-02 -1.26781821e-01 9.35007334e-01
3.75564128e-01 1.91313699e-02 5.36831856e-01 -6.34987772e-01
-4.31230605e-01 -6.79814517e-01 -5.19791722e-01 1.33282471e+00
-1.11607157e-01 -6.35178983e-01 -8.57910886e-02 1.92607373e-01] | [14.73765754699707, 0.9596536159515381] |
d6a7ace9-8f3b-4b0b-bb6f-3c4ddb1634dd | smart-data-collection-system-for-brownfield | null | null | https://doi.org/10.1016/j.procir.2022.04.022 | https://doi.org/10.1016/j.procir.2022.04.022 | Smart Data Collection System for Brownfield CNC Milling Machines: A New Benchmark Dataset for Data-Driven Machine Monitoring | Manufacturing processes have undergone tremendous technological progress in recent decades. To meet the agile philosophy in industry, data-driven algorithms need to handle growing complexity, particularly in Computer Numerical Control machining. To enhance the scalability of machine learning in real-world applications, this paper presents a benchmark dataset for process monitoring of brownfield milling machines based on acceleration data. The data is collected from a real-world production plant using a smart data collection system over a two-years period. In this work, the edge-to-cloud setup is presented followed by an extensive description of the different normal and abnormal processes. An analysis of the dataset highlights the challenges of machine learning in industry caused by the environmental and industrial factors. The new dataset is published with this paper and available at: https://github.com/boschresearch/CNC_Machining. | ['Klaus Diepold', 'Michael Feil', 'Mohamed-Ali Tnani'] | 2022-06-29 | null | null | null | procedia-cirp-2022-6 | ['time-series-clustering'] | ['time-series'] | [ 8.78402442e-02 -4.53566819e-01 1.43784538e-01 -2.68769950e-01
2.29305878e-01 -1.76129401e-01 1.82647794e-01 5.04543364e-01
2.48461053e-01 4.10848528e-01 -5.99106729e-01 -1.07927531e-01
-5.58601797e-01 -8.56359303e-01 -3.33681107e-01 -6.62734985e-01
-1.93768561e-01 1.00338626e+00 -2.33708903e-01 -3.19660813e-01
6.37888312e-01 9.77033019e-01 -1.63998711e+00 7.77996242e-01
4.30016905e-01 1.31904638e+00 5.18737972e-01 9.25288379e-01
5.15637733e-02 8.74773800e-01 -5.61942756e-01 1.49247587e-01
6.78547323e-01 -1.90495670e-01 -5.19084692e-01 4.90771472e-01
-1.09635353e-01 5.00472896e-02 -1.25305995e-01 7.45612919e-01
3.26717317e-01 2.96817839e-01 9.80109274e-02 -1.60972106e+00
-4.54316258e-01 2.19070718e-01 -2.94682950e-01 2.68198371e-01
-7.58767268e-03 5.25112152e-01 6.89909577e-01 -9.05503094e-01
5.78705847e-01 1.00900018e+00 2.47758001e-01 9.20030624e-02
-8.67142141e-01 -4.37166601e-01 -1.64596006e-01 4.86470193e-01
-7.18491971e-01 8.99292156e-03 6.37623608e-01 -5.06525099e-01
1.10604286e+00 3.78816485e-01 9.89312232e-01 7.14152575e-01
1.07380331e+00 6.11008525e-01 1.17888820e+00 -1.97933286e-01
5.94697058e-01 -1.32581130e-01 2.62818877e-02 2.03869998e-01
4.30696756e-01 3.83928478e-01 -6.16335571e-01 2.02895075e-01
1.02000439e+00 6.26240492e-01 2.95187503e-01 -4.53530282e-01
-1.12517309e+00 7.20034182e-01 -1.04268365e-01 2.19424903e-01
-7.91022360e-01 -1.40903845e-01 9.17599738e-01 8.38705003e-01
3.21847647e-01 9.64403868e-01 -1.39305985e+00 -6.51367247e-01
-5.14892638e-01 2.53705263e-01 1.01564026e+00 1.18481708e+00
3.67184043e-01 1.95351556e-01 2.34595686e-01 5.95873415e-01
-2.13835817e-02 1.87943559e-02 2.98649043e-01 -9.34182942e-01
2.49068007e-01 5.96759737e-01 -8.49619508e-03 -6.28712237e-01
-4.45745826e-01 -2.64979899e-01 -9.83899355e-01 5.70904553e-01
9.46555138e-02 -3.35343063e-01 -8.17890227e-01 1.78619087e-01
3.26661587e-01 -8.58779922e-02 -6.40798286e-02 8.14431965e-01
3.40072781e-01 6.05755985e-01 -3.63049686e-01 -3.85609329e-01
1.21638370e+00 -1.23441494e+00 -1.29694521e+00 -2.64928024e-02
6.28714800e-01 -1.06469142e+00 8.37642610e-01 1.30446744e+00
-1.15464830e+00 -8.69388342e-01 -1.05074418e+00 3.05942029e-01
-7.01008201e-01 -9.50067490e-02 9.37966287e-01 2.51311898e-01
-3.82173896e-01 1.10274935e+00 -1.16644478e+00 -5.47877431e-01
4.31649894e-01 2.04365313e-01 -3.03409219e-01 -1.63562328e-01
-5.69019556e-01 1.04124308e+00 6.36518300e-01 4.02185440e-01
-9.63551462e-01 -1.05254257e+00 -4.06229109e-01 -1.82621226e-01
6.02053046e-01 -4.23083782e-01 1.65413976e+00 -2.85438418e-01
-1.60485947e+00 3.91963512e-01 5.33754766e-01 -1.70435056e-01
5.37769735e-01 -7.25287437e-01 -8.22834730e-01 -1.88226372e-01
-4.21414882e-01 -2.65719682e-01 5.77922463e-01 -1.03892720e+00
-1.30175567e+00 -6.59169436e-01 -5.02368689e-01 -2.32357323e-01
1.77595094e-01 1.72161058e-01 2.62883063e-02 -3.87920648e-01
1.68061674e-01 -9.14544880e-01 -5.95823824e-01 -4.76846993e-01
-2.10896969e-01 6.16743229e-03 1.33883059e+00 -4.60904896e-01
6.43140018e-01 -2.11328816e+00 -1.63123697e-01 1.02901570e-01
-1.92090973e-01 2.12609228e-02 1.22385979e-01 1.08090913e+00
-4.22716171e-01 -2.99608737e-01 2.77291030e-01 1.34814858e-01
-2.05556691e-01 5.98811388e-01 -6.47123605e-02 5.29191911e-01
3.90095145e-01 3.24447185e-01 -8.93596113e-01 1.61802530e-01
6.83756351e-01 -1.90119967e-01 -1.06527537e-01 2.31369451e-01
-2.03288227e-01 6.92974448e-01 -2.42138997e-01 1.23097801e+00
7.75662422e-01 1.79126158e-01 5.80616929e-02 -3.21182370e-01
-3.96888912e-01 -4.89860505e-01 -1.41565883e+00 1.61467934e+00
-6.89579546e-01 1.80907354e-01 6.43593907e-01 -1.00171041e+00
1.22250164e+00 5.30623615e-01 9.99554098e-01 -4.60770607e-01
3.23497683e-01 3.76345456e-01 3.03118110e-01 -7.14435399e-01
5.54952979e-01 7.76310265e-02 2.91172892e-01 -8.30086693e-02
-4.22323234e-02 -9.18495953e-01 8.10075700e-01 -4.00178462e-01
1.31784809e+00 5.04979910e-03 1.97787762e-01 -3.30642343e-01
2.78297782e-01 4.74769503e-01 6.66265368e-01 1.08721644e-01
-2.37643331e-01 1.16708763e-01 2.33841196e-01 -9.76709008e-01
-1.20722616e+00 -8.04270983e-01 -7.73869529e-02 7.04416811e-01
-8.60272869e-02 -4.34797883e-01 -1.60578489e-02 -1.96655273e-01
4.56934333e-01 8.08880866e-01 -2.74825931e-01 -1.48319766e-01
-2.85077125e-01 -6.10792756e-01 -4.45512086e-01 5.15521646e-01
1.99880943e-01 -1.34321523e+00 -8.56253982e-01 5.03623545e-01
7.50785649e-01 -1.14397252e+00 2.14791119e-01 5.87011516e-01
-1.45872045e+00 -1.11656797e+00 3.91185403e-01 -3.90459359e-01
4.39201355e-01 -5.41757382e-02 1.19830513e+00 -1.51406884e-01
-9.97775495e-01 -9.64742899e-03 -5.88540673e-01 -1.26176083e+00
-6.71134412e-01 -3.41199040e-02 2.35086694e-01 -4.94005203e-01
5.38189650e-01 -3.30164790e-01 -3.62100899e-01 2.36170068e-01
-7.25895643e-01 -9.77948979e-02 8.95399153e-01 8.00114393e-01
6.45819247e-01 9.46569443e-01 5.12893856e-01 -8.54352057e-01
5.22997081e-01 -6.06699765e-01 -1.07536960e+00 -2.06613109e-01
-1.09557414e+00 -6.59066796e-01 7.53445685e-01 -1.60127684e-01
-9.72701609e-01 2.74545372e-01 2.43501842e-01 -7.60626197e-01
-7.57246196e-01 6.00759566e-01 4.56942767e-02 5.51015496e-01
9.80165228e-02 -2.86968529e-01 3.70550841e-01 -7.81192601e-01
8.04029778e-02 7.14738369e-01 3.94029409e-01 -4.32002097e-01
6.53095841e-01 2.91745365e-01 3.33188564e-01 -8.03398669e-01
-3.51242930e-01 -7.32527673e-01 -7.47287393e-01 -5.65468073e-01
5.47256887e-01 -5.23856401e-01 -9.11370516e-01 5.98886311e-01
-7.92939186e-01 -3.90866727e-01 -7.69944191e-01 7.91002631e-01
-8.33292246e-01 -3.47643107e-01 -1.18584394e+00 -7.85147429e-01
-5.36988974e-01 -9.22105432e-01 9.05295908e-01 1.59942880e-02
-1.84542134e-01 -5.34427285e-01 -1.11767575e-01 5.39738476e-01
4.63542968e-01 6.22784138e-01 7.08043039e-01 -7.99555540e-01
-5.73096633e-01 -6.06611967e-01 2.38865420e-01 6.56088769e-01
8.46751988e-01 6.45732641e-01 -6.37177885e-01 -6.11012399e-01
5.08878946e-01 4.05444987e-02 3.41815129e-02 3.27270240e-01
1.28231192e+00 1.93631127e-01 -1.13943785e-01 3.17322135e-01
1.46116710e+00 6.57233298e-01 3.09473336e-01 5.49691200e-01
5.76971710e-01 6.52481616e-01 1.66584516e+00 1.15840518e+00
-6.56144321e-01 5.07189780e-02 9.98292565e-01 -4.00483273e-02
4.61768270e-01 4.08245265e-01 1.35416687e-01 1.17257023e+00
-4.08944517e-01 1.18710667e-01 -1.03516889e+00 4.73748296e-01
-1.77409410e+00 -8.42994094e-01 -7.39150107e-01 1.79297554e+00
4.43513513e-01 1.70391411e-01 -2.55770266e-01 4.33242440e-01
5.17519653e-01 -4.77221996e-01 -6.30253851e-01 -1.16514671e+00
3.94891262e-01 2.80094326e-01 6.94658637e-01 -1.75358385e-01
-1.10783803e+00 5.87646067e-01 5.62353086e+00 3.14018369e-01
-1.05959427e+00 2.77612228e-02 3.71548653e-01 -5.63452661e-01
8.61366451e-01 -3.22829783e-01 -6.05901539e-01 5.33519149e-01
1.32643569e+00 -1.95751920e-01 3.86909038e-01 1.38264918e+00
5.78889132e-01 -4.39347662e-02 -1.47461176e+00 9.49792325e-01
-6.14025831e-01 -1.51000643e+00 -5.45982778e-01 3.61473590e-01
1.06731617e+00 5.59252918e-01 -1.78617284e-01 1.77951917e-01
5.17690778e-01 -7.42013335e-01 5.87902844e-01 4.73412603e-01
3.23100299e-01 -1.07392645e+00 1.28213644e+00 2.36355558e-01
-1.18209946e+00 -7.30429053e-01 -2.67443061e-01 -5.13805807e-01
2.38374382e-01 1.30962646e+00 -9.14273441e-01 1.17695498e+00
1.02925479e+00 8.30205917e-01 -6.60594404e-02 9.33334351e-01
3.58966619e-01 2.82819092e-01 3.03004906e-02 3.36250335e-01
-3.23501304e-02 -7.18710542e-01 1.93332627e-01 9.39873099e-01
5.78823447e-01 -4.40294951e-01 4.33321923e-01 4.41884995e-01
3.96818757e-01 4.91525792e-02 -6.54771805e-01 -3.54492545e-01
4.33573633e-01 1.65183985e+00 -6.39950573e-01 -8.77065957e-02
-5.16492546e-01 8.05895865e-01 -3.05575162e-01 -1.97441876e-01
-6.06864274e-01 -4.79773611e-01 1.08978546e+00 2.88777035e-02
2.04470828e-01 -7.03704119e-01 -5.52489698e-01 -3.37176055e-01
-7.15183541e-02 -1.10217273e+00 3.24246377e-01 -6.16556108e-01
-1.76531947e+00 1.50198579e-01 -2.11041391e-01 -1.34249914e+00
-3.97465937e-02 -1.20127261e+00 -4.64415461e-01 6.21813595e-01
-1.18140686e+00 -8.44270170e-01 -5.83737016e-01 2.39209503e-01
1.32764626e+00 -4.57067370e-01 8.11320901e-01 4.07814205e-01
-9.48171437e-01 -6.02270424e-01 4.93474662e-01 -4.02312487e-01
7.99633741e-01 -1.39939582e+00 4.62229550e-01 5.95204115e-01
-4.06706363e-01 2.01727092e-01 9.69501495e-01 -1.04216480e+00
-2.19507623e+00 -1.46605909e+00 2.90970922e-01 -3.47026616e-01
1.02753222e+00 -1.50526270e-01 -6.55921936e-01 7.96823502e-01
4.92398083e-01 2.25772291e-01 6.17958784e-01 -9.97323096e-02
6.31630063e-01 -4.69646931e-01 -1.34656346e+00 1.67243287e-01
7.58407176e-01 3.79905626e-02 -2.35989004e-01 9.52503026e-01
2.64141589e-01 -6.40121877e-01 -1.69333947e+00 4.74395066e-01
-8.28979537e-03 -5.82626343e-01 3.65591645e-01 -7.25457966e-01
7.37037122e-01 -1.03555225e-01 -4.37939763e-02 -1.60479200e+00
-3.77802938e-01 -7.78479040e-01 -5.16447842e-01 1.07170522e+00
-8.98569152e-02 -7.06437826e-01 8.40677440e-01 2.84237951e-01
-6.15116775e-01 -1.08483100e+00 -6.58354819e-01 -1.20380640e+00
-2.02166185e-01 -2.31647104e-01 8.83472443e-01 1.20124090e+00
6.30969107e-02 -5.37712239e-02 2.80939508e-02 2.85399795e-01
3.36243719e-01 1.42733946e-01 9.01312053e-01 -1.59559834e+00
-2.91430671e-02 -2.83305123e-02 -5.58640122e-01 -1.12752765e-01
-6.16989374e-01 -5.66292703e-01 2.74768248e-02 -1.66885996e+00
-3.00808489e-01 -2.25068524e-01 -3.43031496e-01 2.27447122e-01
4.97505009e-01 -3.14165801e-01 3.97874743e-01 1.87441349e-01
9.66096446e-02 3.01995724e-01 1.65632868e+00 -2.10551806e-02
-3.25502194e-02 2.35653207e-01 -3.13044130e-03 5.10719657e-01
1.52154338e+00 -1.62753686e-01 -3.58804315e-01 -1.72221988e-01
2.02281341e-01 -1.25631973e-01 -2.06930220e-01 -1.11220312e+00
2.81632304e-01 -5.56438267e-01 4.16283518e-01 -1.01538301e+00
3.40095401e-01 -1.58813965e+00 6.68293536e-01 1.05238616e+00
1.71625227e-01 8.18565190e-01 2.74161130e-01 3.94481152e-01
-4.02688384e-01 -1.62547544e-01 6.39225066e-01 -4.03983027e-01
-8.48243773e-01 4.11945164e-01 -4.72320825e-01 -6.98927522e-01
1.54124057e+00 -1.78839639e-01 -3.83632302e-01 3.13137800e-01
-8.25342000e-01 2.60763049e-01 2.28635967e-01 6.75398588e-01
4.16956037e-01 -1.01337063e+00 -9.17603254e-01 4.93027091e-01
3.33201773e-02 2.38431469e-01 2.78914809e-01 1.01273143e+00
-9.83188868e-01 2.94511884e-01 -8.90167058e-01 -5.80687284e-01
-1.14671803e+00 1.06535184e+00 1.31147802e-01 -2.79310018e-01
-5.88478148e-01 4.29129243e-01 -7.32345939e-01 -4.33453351e-01
-2.40204558e-01 -8.49488854e-01 2.59870052e-01 6.92107752e-02
3.38880509e-01 1.03233480e+00 9.83836889e-01 3.45840454e-02
-1.28687760e-02 -3.83417271e-02 -1.70175940e-01 5.30326605e-01
1.71603978e+00 1.05085164e-01 -3.70499402e-01 8.85507762e-01
7.76123226e-01 -3.60689372e-01 -1.11017621e+00 2.78077900e-01
4.20049787e-01 -9.28755939e-01 2.43617490e-01 -8.37674320e-01
-1.59815633e+00 4.91691679e-01 8.46226931e-01 5.74969471e-01
1.05107403e+00 -3.45020324e-01 6.83609903e-01 5.17594576e-01
4.05909359e-01 -1.73819053e+00 9.33095962e-02 7.39038706e-01
1.14508986e+00 -1.28648305e+00 1.59505174e-01 -6.74891710e-01
-6.82881832e-01 1.40654933e+00 7.93994904e-01 3.67823988e-02
7.56870091e-01 9.51854169e-01 1.78327367e-01 -5.88220656e-01
-1.04798198e+00 3.54203999e-01 -8.15231144e-01 5.78656018e-01
3.57015252e-01 1.00129955e-01 1.03410505e-01 1.13749690e-01
-3.57207268e-01 5.27318776e-01 6.25352323e-01 1.73625803e+00
-3.94324176e-02 -1.26150322e+00 -5.01954854e-01 8.29801142e-01
-4.61562842e-01 3.61831754e-01 5.97837083e-02 8.75589967e-01
3.00502032e-01 1.27991450e+00 6.12165391e-01 -2.90398151e-01
9.31835711e-01 -4.16793004e-02 3.45784009e-01 -8.62324476e-01
-8.40544105e-01 -1.29816905e-01 1.75845206e-01 -7.57668257e-01
5.79749905e-02 -9.36884284e-01 -1.42899907e+00 -6.13658309e-01
-3.80721658e-01 -8.63920078e-02 1.34644997e+00 3.51213247e-01
5.06174147e-01 1.38299739e+00 8.14547539e-01 -1.04802406e+00
-8.49152625e-01 -1.17137551e+00 -1.27718711e+00 3.66128236e-01
-1.39056578e-01 -6.02877498e-01 -2.05273733e-01 2.87327588e-01] | [6.985742568969727, 2.2975776195526123] |
4679655f-3037-4d78-ada4-2d4a3515046a | body-part-based-representation-learning-for | 2211.03679 | null | https://arxiv.org/abs/2211.03679v1 | https://arxiv.org/pdf/2211.03679v1.pdf | Body Part-Based Representation Learning for Occluded Person Re-Identification | Occluded person re-identification (ReID) is a person retrieval task which aims at matching occluded person images with holistic ones. For addressing occluded ReID, part-based methods have been shown beneficial as they offer fine-grained information and are well suited to represent partially visible human bodies. However, training a part-based model is a challenging task for two reasons. Firstly, individual body part appearance is not as discriminative as global appearance (two distinct IDs might have the same local appearance), this means standard ReID training objectives using identity labels are not adapted to local feature learning. Secondly, ReID datasets are not provided with human topographical annotations. In this work, we propose BPBreID, a body part-based ReID model for solving the above issues. We first design two modules for predicting body part attention maps and producing body part-based features of the ReID target. We then propose GiLt, a novel training scheme for learning part-based representations that is robust to occlusions and non-discriminative local appearance. Extensive experiments on popular holistic and occluded datasets show the effectiveness of our proposed method, which outperforms state-of-the-art methods by 0.7% mAP and 5.6% rank-1 accuracy on the challenging Occluded-Duke dataset. Our code is available at https://github.com/VlSomers/bpbreid. | ['Alexandre Alahi', 'Christophe De Vleeschouwer', 'Vladimir Somers'] | 2022-11-07 | null | null | null | null | ['person-retrieval', 'human-parsing', 'part-based-representation-learning'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-3.60892043e-02 5.59075586e-02 -2.01010913e-01 -3.89596403e-01
-5.44174552e-01 -3.67480546e-01 5.97894609e-01 -1.02637433e-01
-1.68852925e-01 6.33356392e-01 3.68832797e-01 3.98775667e-01
3.86307910e-02 -5.88931203e-01 -8.65206540e-01 -6.20687485e-01
7.87682980e-02 7.41489232e-01 6.57860637e-02 -8.91661420e-02
-1.35242552e-01 6.20688975e-01 -1.84152818e+00 1.81911141e-01
6.88125610e-01 1.01890254e+00 -2.09333003e-01 2.79038042e-01
1.02193095e-01 2.22010314e-01 -5.83599031e-01 -8.45744610e-01
5.02475023e-01 -3.58417332e-01 -8.20536911e-01 7.98673779e-02
1.25455403e+00 -3.81089091e-01 -4.98905778e-01 7.59251058e-01
6.97145045e-01 1.57191738e-01 8.89867425e-01 -1.30569208e+00
-8.09595466e-01 6.42989203e-02 -8.89325261e-01 3.46058518e-01
6.26430154e-01 -1.33798644e-01 5.51602364e-01 -8.52571845e-01
6.76894605e-01 1.57657087e+00 7.97848165e-01 9.09887433e-01
-1.15586543e+00 -7.11563647e-01 4.20177996e-01 4.21297312e-01
-1.78078294e+00 -5.64458072e-01 6.10029578e-01 -4.79755610e-01
5.75994670e-01 4.42463756e-01 7.25626886e-01 1.13682175e+00
1.50327772e-01 9.73838151e-01 1.11111069e+00 -2.81284481e-01
-3.43069613e-01 2.02722788e-01 5.94575591e-02 7.26141334e-01
4.73773688e-01 3.19768451e-02 -5.65340757e-01 -1.04392782e-01
8.41360569e-01 1.91899955e-01 -3.14583123e-01 -7.43812919e-01
-9.13775504e-01 4.16426659e-01 9.05227780e-01 1.60516545e-01
-3.20101589e-01 1.37207046e-01 3.08700562e-01 2.08883453e-02
5.45439482e-01 -9.95358527e-02 -6.78452477e-02 4.19042885e-01
-7.29859173e-01 4.42141443e-01 4.20245707e-01 9.14078176e-01
4.35535043e-01 -2.68999845e-01 -5.37134945e-01 1.07382429e+00
3.26891810e-01 7.34982193e-01 3.73047262e-01 -4.30754781e-01
2.45460287e-01 6.35089219e-01 2.35793352e-01 -9.38382268e-01
-5.71408927e-01 -6.31817043e-01 -9.03995216e-01 2.66434383e-02
7.06809223e-01 3.75501812e-01 -1.01361740e+00 1.82361174e+00
7.91326344e-01 -1.38500512e-01 -2.91689724e-01 1.20499563e+00
1.21468651e+00 1.20691024e-01 4.42744434e-01 3.23287636e-01
1.80643404e+00 -1.06932247e+00 -5.05420804e-01 -3.30694944e-01
2.40313441e-01 -5.81167519e-01 5.73378325e-01 -2.67725587e-02
-1.04848862e+00 -9.10913706e-01 -7.45679975e-01 -1.71895504e-01
-6.23586535e-01 3.92612010e-01 4.58562195e-01 8.34223330e-01
-1.10830295e+00 3.35812122e-01 -5.10024846e-01 -7.11237252e-01
5.86394131e-01 5.15467823e-01 -7.83326983e-01 -2.58382797e-01
-9.06548321e-01 8.74642015e-01 7.91869387e-02 3.07515144e-01
-6.77620947e-01 -3.90387923e-01 -1.03912950e+00 -1.88764352e-02
1.73063930e-02 -1.16194463e+00 8.50829005e-01 -9.24622893e-01
-1.03553271e+00 1.47162211e+00 -5.45000255e-01 -1.76794022e-01
7.59358585e-01 -2.91279495e-01 -4.56935048e-01 1.78348988e-01
2.49175504e-01 7.85195291e-01 9.82007205e-01 -1.47118628e+00
-3.69185627e-01 -8.03220689e-01 -1.69492781e-01 3.69668782e-01
-1.39439702e-01 1.69648051e-01 -7.54081726e-01 -6.70131564e-01
2.48479843e-01 -9.53976750e-01 2.44316429e-01 2.76658833e-01
-3.62413257e-01 -4.64637309e-01 4.46736157e-01 -1.15311682e+00
7.24066794e-01 -1.88659227e+00 7.01624826e-02 1.61073297e-01
2.09529459e-01 2.46303111e-01 -4.58786525e-02 2.31857017e-01
-5.78686036e-02 -1.16922371e-02 1.90026209e-01 -7.82805741e-01
8.93690512e-02 -2.17650130e-01 1.98362529e-01 8.41525078e-01
-1.81238889e-03 1.17460406e+00 -6.38627410e-01 -8.18167627e-01
3.03045988e-01 6.83299839e-01 -7.81380460e-02 1.78605646e-01
5.20753980e-01 7.42512286e-01 -3.06442440e-01 1.01508784e+00
8.29082668e-01 -5.09089194e-02 -1.09601766e-01 -3.42324018e-01
1.83998242e-01 -2.76619732e-01 -9.61803377e-01 1.70158494e+00
-1.74085736e-01 2.38097146e-01 2.54163817e-02 -6.93468094e-01
8.59993815e-01 2.71859467e-01 3.69110376e-01 -9.38104689e-01
2.22177163e-01 4.69899327e-02 -3.29720229e-01 -2.93064952e-01
2.68282861e-01 1.26990452e-01 4.30004559e-02 8.82190187e-03
-8.10248554e-02 5.36391675e-01 1.42706007e-01 2.38176752e-02
7.59075820e-01 5.62674642e-01 2.93698639e-01 -1.44516081e-01
6.67603731e-01 -2.05529794e-01 6.54925346e-01 7.63487816e-01
-5.23413122e-01 8.93390298e-01 -1.00098222e-01 -7.37725496e-01
-8.22013855e-01 -1.14918935e+00 -4.86079276e-01 1.21070683e+00
6.54029191e-01 -2.43611544e-01 -7.41290748e-01 -6.92815900e-01
2.60057300e-01 8.63644257e-02 -1.00760436e+00 1.99667886e-02
-5.99871874e-01 -5.67632377e-01 4.27673995e-01 7.13159382e-01
6.41244888e-01 -1.01130366e+00 -2.04654217e-01 -1.07752673e-01
-3.72940332e-01 -9.26781058e-01 -6.61195099e-01 -5.33413410e-01
-5.45831323e-01 -1.22075284e+00 -1.45440125e+00 -8.93549681e-01
1.07820821e+00 4.60340321e-01 1.15678918e+00 2.85191536e-01
-6.14881337e-01 6.16069973e-01 -2.41452128e-01 -2.67502010e-01
1.16611756e-01 -5.30048832e-02 2.41377160e-01 3.03922743e-01
3.74980688e-01 -3.09015989e-01 -1.12707031e+00 6.68275356e-01
-3.93668234e-01 9.70588103e-02 8.04697812e-01 7.36280560e-01
8.15385342e-01 -3.83377582e-01 5.26831031e-01 -6.90957785e-01
8.82770792e-02 -3.24565828e-01 -1.99683949e-01 5.91920853e-01
-2.83483326e-01 -4.24824834e-01 3.11331928e-01 -4.05271500e-01
-1.10912120e+00 1.74753830e-01 -3.86510119e-02 -4.33022052e-01
-5.31079829e-01 -2.43907765e-01 -3.32083791e-01 -3.59054059e-01
5.24181962e-01 2.64722407e-01 -8.83441046e-02 -6.75021172e-01
2.46894449e-01 6.30192518e-01 7.98745632e-01 -7.09999323e-01
8.98682833e-01 7.08825648e-01 -4.72986959e-02 -4.76876080e-01
-7.23766923e-01 -9.83217299e-01 -9.83350873e-01 -3.86455595e-01
8.49798143e-01 -1.22473550e+00 -6.67273939e-01 4.91456687e-01
-8.85020733e-01 -1.02345094e-01 -1.17255360e-01 8.07957053e-02
-3.94456685e-01 5.76189041e-01 -3.70305181e-01 -8.29900205e-01
-7.81357348e-01 -7.01747835e-01 1.51275241e+00 5.34790695e-01
-3.05967689e-01 -5.70992827e-01 1.42495949e-02 7.11737573e-01
3.28285336e-01 4.76928443e-01 3.40874672e-01 -4.52115834e-01
-3.27901125e-01 -7.30061293e-01 -5.58820546e-01 -1.60570860e-01
7.46251419e-02 -4.87234414e-01 -1.13183975e+00 -6.77787006e-01
-5.83950520e-01 -3.14124048e-01 8.11554313e-01 4.09597367e-01
1.10452986e+00 -2.72063822e-01 -8.32177162e-01 6.79186702e-01
1.18131793e+00 -2.62941808e-01 5.30569375e-01 3.73940408e-01
9.11673069e-01 8.31066072e-01 5.36365688e-01 3.02789986e-01
6.87236011e-01 1.13149762e+00 2.42136642e-01 -3.94409060e-01
-7.10416257e-01 -4.63312417e-01 1.12559155e-01 2.54223626e-02
-4.11694944e-01 -9.73710120e-02 -7.28739679e-01 7.50576198e-01
-1.89018011e+00 -9.59688485e-01 -2.20513612e-01 2.34234953e+00
4.98665631e-01 -3.31818700e-01 5.39699912e-01 -1.61702007e-01
9.73766625e-01 -1.89675733e-01 -4.55959409e-01 -1.67869218e-02
-1.39730394e-01 -3.50489430e-02 4.75249648e-01 -3.80596779e-02
-1.52930701e+00 8.72144938e-01 4.99695253e+00 7.35968351e-01
-6.00695491e-01 3.72596353e-01 4.47587907e-01 -2.85463519e-02
1.79195240e-01 -3.66832942e-01 -1.03118908e+00 4.71884459e-01
4.11710829e-01 1.88737214e-01 4.04184610e-02 9.44911599e-01
-1.43142611e-01 1.69017864e-03 -1.04502237e+00 1.33128488e+00
4.45766777e-01 -6.86965585e-01 7.18552843e-02 1.45003304e-01
7.55528212e-01 -3.29303473e-01 9.12592113e-02 2.81318843e-01
-1.64454415e-01 -1.06253564e+00 9.29451406e-01 7.56488442e-01
8.42926264e-01 -5.81863761e-01 9.14285064e-01 -4.89530452e-02
-1.59495997e+00 1.56004682e-01 -6.16174877e-01 1.84565753e-01
1.11907057e-01 3.20702828e-02 -4.50551420e-01 7.70693243e-01
1.12312257e+00 5.21372259e-01 -9.85131621e-01 1.61942112e+00
-1.61203131e-01 -2.36053877e-02 -2.74406165e-01 4.27242100e-01
-4.14172798e-01 1.90471783e-01 3.13157082e-01 1.19006038e+00
6.73759580e-02 4.22611982e-02 2.84177095e-01 7.67479300e-01
-7.99935982e-02 3.73497337e-01 -4.93058234e-01 5.43043613e-01
2.17033014e-01 1.27542901e+00 -6.74104750e-01 -3.09473455e-01
-4.13323015e-01 1.38249624e+00 4.69130367e-01 4.41550910e-01
-7.25446701e-01 -4.19703797e-02 6.27349317e-01 4.32987422e-01
2.55638927e-01 2.45249629e-01 6.70312867e-02 -1.13016486e+00
1.62038535e-01 -7.24165320e-01 7.09245682e-01 -6.14060223e-01
-1.65384269e+00 5.67823768e-01 9.62876752e-02 -1.21143019e+00
-2.38129739e-02 -4.83479977e-01 -4.45199817e-01 9.98076975e-01
-1.35571992e+00 -1.93005943e+00 -8.50666046e-01 6.99269652e-01
4.03043836e-01 9.19030011e-02 8.80528569e-01 3.93479973e-01
-7.90600538e-01 1.02591228e+00 2.49587279e-02 3.07361513e-01
9.75393891e-01 -1.09346652e+00 4.76701170e-01 8.25669825e-01
3.92498821e-02 7.70795882e-01 5.46396494e-01 -8.27796161e-01
-1.26717126e+00 -1.07939291e+00 8.13170016e-01 -7.35032856e-01
-7.95187876e-02 -4.27354187e-01 -8.31264853e-01 5.89670479e-01
-1.59839690e-01 2.68910050e-01 6.37565255e-01 2.43373454e-01
-5.04587770e-01 -2.22868547e-01 -1.33135962e+00 3.76229346e-01
1.50619507e+00 -3.00847590e-01 -3.83413464e-01 3.63430172e-01
-2.07467116e-02 -5.38310826e-01 -9.60328043e-01 5.39647758e-01
9.24114823e-01 -9.45894301e-01 1.35457957e+00 -3.08965176e-01
-1.66109487e-01 -3.95462811e-01 1.41344488e-01 -8.32108915e-01
-5.80141068e-01 -3.39848250e-02 -2.79608041e-01 1.37307501e+00
-1.86136723e-01 -7.07034767e-01 8.19313347e-01 8.08880985e-01
7.25423321e-02 -5.87186158e-01 -9.42926705e-01 -9.88064766e-01
-1.06043480e-01 2.38318786e-01 6.00158691e-01 5.83065569e-01
-3.06325734e-01 -5.19506745e-02 -5.58129251e-01 3.68957072e-01
9.19310927e-01 3.99092734e-01 1.12371504e+00 -1.33406854e+00
-6.61731465e-03 -4.99989063e-01 -8.49153280e-01 -8.41262579e-01
1.93299547e-01 -9.12435114e-01 -9.87194628e-02 -1.78000367e+00
6.52717173e-01 -5.41425884e-01 -3.98183048e-01 7.47731864e-01
-3.31721425e-01 8.49270403e-01 1.41206354e-01 5.39052904e-01
-6.83453083e-01 5.87707102e-01 1.09468973e+00 -4.17591453e-01
6.99920356e-02 1.20091259e-01 -5.72663188e-01 3.80971074e-01
7.86281645e-01 -2.22960740e-01 -1.48224011e-01 -2.17093632e-01
-3.83153558e-01 -4.29140061e-01 9.05755997e-01 -1.21161222e+00
6.26789927e-02 3.61804336e-01 1.15843892e+00 -6.43043876e-01
7.40462720e-01 -6.36284411e-01 4.52028215e-01 3.89486998e-01
8.89344439e-02 -1.67349070e-01 2.91856229e-01 6.08532131e-01
1.16457418e-02 1.28765656e-02 6.87628686e-01 -1.53276384e-01
-8.67668033e-01 5.72640061e-01 1.83776364e-01 -2.23894894e-01
9.40188229e-01 -4.52127188e-01 -3.17475796e-01 -2.16707423e-01
-7.23735631e-01 2.37722814e-01 8.35108817e-01 6.70357108e-01
5.97402036e-01 -1.35252357e+00 -8.25162649e-01 1.56426966e-01
4.47395951e-01 -2.96920776e-01 6.95089638e-01 9.17268634e-01
-2.67831177e-01 5.64041972e-01 -3.96060139e-01 -5.89517593e-01
-1.67484164e+00 7.05526948e-01 4.47745115e-01 -7.18590021e-02
-8.54416072e-01 9.83794749e-01 8.41927588e-01 -3.96148205e-01
3.31633955e-01 2.55206287e-01 -4.21566725e-01 -5.39929196e-02
6.67724967e-01 3.48391891e-01 -5.10827899e-02 -1.39833558e+00
-6.65143549e-01 1.04064775e+00 -1.19916253e-01 3.62899542e-01
8.44169497e-01 -4.12387848e-01 4.88425158e-02 -5.75242564e-02
9.33909655e-01 -1.87384695e-01 -1.10170484e+00 -3.52366358e-01
-3.80669981e-01 -7.30936348e-01 -3.34924668e-01 -8.29009891e-01
-9.70427811e-01 6.40079379e-01 1.15410447e+00 -2.10272744e-01
9.54251468e-01 3.25805277e-01 6.84989214e-01 -1.11835785e-01
7.00837016e-01 -7.79446900e-01 -2.40126684e-01 -1.50063008e-01
1.13039410e+00 -1.40632498e+00 1.95145875e-01 -6.14757061e-01
-4.32363063e-01 6.95601761e-01 9.51584101e-01 1.37135461e-01
1.95892781e-01 -5.37101984e-01 1.39706016e-01 -3.04688156e-01
-3.96204628e-02 -6.82176530e-01 9.45902944e-01 9.17808414e-01
4.27906096e-01 1.57172576e-01 -3.85867059e-01 5.04276037e-01
-2.16033712e-01 -3.81617665e-01 -3.27755570e-01 6.44582748e-01
-8.16410184e-02 -1.15077877e+00 -8.29939127e-01 3.80602241e-01
-4.31414127e-01 3.64604712e-01 -6.13593578e-01 8.48117232e-01
3.29920769e-01 7.72999465e-01 -8.04463774e-02 -5.02256379e-02
3.46193284e-01 2.08248645e-01 7.93284237e-01 -4.54930514e-01
-4.37546164e-01 -6.78676963e-02 3.96008156e-02 -6.56639457e-01
-4.93586779e-01 -7.28318036e-01 -8.76031935e-01 -3.52196246e-01
-2.48331547e-01 -1.29853442e-01 5.57934880e-01 7.23521888e-01
4.11004603e-01 3.34021032e-01 1.58170253e-01 -1.26804912e+00
-9.30146500e-02 -9.66839790e-01 -5.12459397e-01 8.48320842e-01
8.58271345e-02 -1.10705996e+00 1.11429751e-01 -4.21131514e-02] | [14.6673583984375, 0.913096010684967] |
731818e0-8501-4c62-bfcc-c58752de2d86 | enhanced-blind-face-restoration-with-multi | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Li_Enhanced_Blind_Face_Restoration_With_Multi-Exemplar_Images_and_Adaptive_Spatial_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Li_Enhanced_Blind_Face_Restoration_With_Multi-Exemplar_Images_and_Adaptive_Spatial_CVPR_2020_paper.pdf | Enhanced Blind Face Restoration With Multi-Exemplar Images and Adaptive Spatial Feature Fusion | In many real-world face restoration applications, e.g., smartphone photo albums and old films, multiple high-quality (HQ) images of the same person usually are available for a given degraded low-quality (LQ) observation. However, most existing guided face restoration methods are based on single HQ exemplar image, and are limited in properly exploiting guidance for improving the generalization ability to unknown degradation process. To address these issues, this paper suggests to enhance blind face restoration performance by utilizing multi-exemplar images and adaptive fusion of features from guidance and degraded images. First, given a degraded observation, we select the optimal guidance based on the weighted affine distance on landmark sets, where the landmark weights are learned to make the guidance image optimized to HQ image reconstruction. Second, moving least-square and adaptive instance normalization are leveraged for spatial alignment and illumination translation of guidance image in the feature space. Finally, for better feature fusion, multiple adaptive spatial feature fusion (ASFF) layers are introduced to incorporate guidance features in an adaptive and progressive manner, resulting in our ASFFNet. Experiments show that our ASFFNet performs favorably in terms of quantitative and qualitative evaluation, and is effective in generating photo-realistic results on real-world LQ images. The source code and models are available at https://github.com/csxmli2016/ASFFNet.
| [' Wangmeng Zuo', ' Meng Wang', ' Hongzhi Zhang', ' Dongwei Ren', ' Wenyu Li', 'Xiaoming Li'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['blind-face-restoration'] | ['computer-vision'] | [ 2.51064986e-01 -4.14150953e-01 1.14101224e-01 -4.55748975e-01
-8.48268747e-01 -4.28768918e-02 5.15943289e-01 -4.38333869e-01
4.55956021e-03 5.07085443e-01 6.65235579e-01 1.45757973e-01
-4.21402127e-01 -4.44695830e-01 -5.58269083e-01 -9.68537331e-01
1.51847899e-01 -1.59299895e-01 -3.12398493e-01 -1.67513013e-01
3.31240237e-01 6.37806177e-01 -1.48508573e+00 1.90724164e-01
1.15175760e+00 9.34974611e-01 2.96157956e-01 3.37592602e-01
1.85485438e-01 2.27487549e-01 -5.18812895e-01 -3.58783573e-01
5.87869942e-01 -4.42390919e-01 -3.25715631e-01 5.02699316e-01
9.45247889e-01 -5.89006901e-01 -4.49942291e-01 1.18212187e+00
8.40459645e-01 3.41522962e-01 4.83161181e-01 -1.25940323e+00
-1.10941780e+00 -1.75166696e-01 -6.56460047e-01 2.59681851e-01
7.91619897e-01 3.96866351e-01 5.86809874e-01 -1.41446269e+00
3.50538641e-01 1.53375554e+00 5.52619874e-01 5.89906275e-01
-1.27426386e+00 -7.75617003e-01 8.94281268e-02 4.25094992e-01
-1.36109424e+00 -1.17718911e+00 9.33777273e-01 -1.30144507e-01
5.83953977e-01 1.98142931e-01 6.03470922e-01 7.88827240e-01
1.24233410e-01 5.17183423e-01 1.16000688e+00 -3.62814158e-01
4.72145304e-02 -1.64519608e-01 -2.51108259e-01 7.55642414e-01
4.05167975e-02 3.26248467e-01 -7.23102212e-01 -1.96296349e-01
9.18907762e-01 3.41351479e-01 -8.04224849e-01 -3.25901210e-01
-9.36682761e-01 5.81904173e-01 6.63144946e-01 7.63220936e-02
-4.60196376e-01 -2.22133130e-01 -6.90483749e-02 2.69159257e-01
5.99372506e-01 9.75914821e-02 -1.60618588e-01 3.85565281e-01
-9.23166692e-01 2.51151770e-01 2.29163364e-01 5.86691201e-01
9.30802166e-01 1.87399864e-01 -2.68374354e-01 1.33493447e+00
6.17908478e-01 8.03043365e-01 4.11214054e-01 -1.42651188e+00
3.80775928e-01 4.64582145e-01 1.29573792e-01 -1.26666200e+00
-2.70842880e-01 -4.55437183e-01 -9.41523552e-01 3.18710387e-01
2.60081947e-01 2.27117985e-01 -1.20266879e+00 1.68366444e+00
4.50335354e-01 4.57497180e-01 -3.58641148e-02 1.20621920e+00
8.75725389e-01 5.86364448e-01 -2.33248442e-01 -5.69531620e-01
1.13366055e+00 -8.85096014e-01 -6.94367349e-01 -3.37158322e-01
5.66202588e-02 -8.84102106e-01 1.27337146e+00 2.82019854e-01
-1.07406032e+00 -7.01699793e-01 -9.06673908e-01 -3.63030396e-02
1.13891549e-01 1.73950255e-01 2.96925932e-01 5.38402498e-01
-1.28092933e+00 4.76909339e-01 -5.05591750e-01 -4.92269903e-01
6.34907305e-01 2.68101245e-01 -5.69871962e-01 -8.01122725e-01
-8.12537551e-01 4.31740016e-01 -3.07525486e-01 4.72133547e-01
-9.61112261e-01 -6.41854107e-01 -8.58759046e-01 -2.64541954e-01
1.34911254e-01 -6.71007216e-01 8.57090294e-01 -9.62625504e-01
-1.42240036e+00 5.96811414e-01 -4.07692343e-01 9.83123183e-02
1.36261716e-01 -1.68081835e-01 -7.58202314e-01 4.34061497e-01
3.22716117e-01 5.35748184e-01 1.34272885e+00 -1.67057145e+00
-4.02420759e-01 -6.97568953e-01 -3.13527547e-02 4.84375447e-01
-4.48074818e-01 1.43680617e-01 -5.40881515e-01 -7.75802732e-01
6.39134049e-01 -6.26959801e-01 1.01711825e-02 2.46213436e-01
-1.98304296e-01 7.66785666e-02 7.55893469e-01 -1.04546750e+00
8.65152419e-01 -2.36086893e+00 1.08603045e-01 2.20748216e-01
1.00452855e-01 2.31203035e-01 -5.25193691e-01 9.48089063e-02
-5.99747971e-02 -1.15744516e-01 -3.49670887e-01 -6.51679099e-01
-1.73945844e-01 1.31805182e-01 -1.56751156e-01 8.72270107e-01
-5.56521751e-02 6.49770200e-01 -6.83068514e-01 -3.98404181e-01
2.60103524e-01 8.19586515e-01 -5.39824784e-01 3.51326436e-01
2.96733409e-01 7.93833196e-01 -3.01820904e-01 9.74941075e-01
1.02927518e+00 -5.29860221e-02 -2.43718520e-01 -4.89347517e-01
2.55135834e-01 -1.43078223e-01 -1.28934252e+00 1.70315862e+00
-4.54372764e-01 2.72147119e-01 2.46985734e-01 -6.91543519e-01
8.14997196e-01 2.78034985e-01 5.68013728e-01 -9.91238415e-01
-1.34907171e-01 1.86436534e-01 -2.05469996e-01 -4.10235196e-01
2.59616524e-01 4.24975418e-02 6.22777045e-01 3.92533600e-01
-6.69975579e-02 3.08453850e-02 -2.86553008e-03 1.32241398e-01
7.07049370e-01 1.13496415e-01 7.94384629e-02 2.01251041e-02
8.37318778e-01 -6.08039200e-01 8.33373725e-01 2.94739783e-01
-4.66935188e-01 1.09723437e+00 -1.14081122e-01 -2.73695648e-01
-8.22410524e-01 -1.14825809e+00 -2.84493029e-01 8.93782318e-01
3.29519272e-01 -2.94695497e-01 -8.12826276e-01 -4.77425069e-01
-5.65363392e-02 3.02730531e-01 -3.34437579e-01 -1.85099319e-01
-6.60158575e-01 -8.08591723e-01 6.97644949e-02 1.73300892e-01
7.34918892e-01 -1.14694989e+00 -1.13221025e-02 -3.62600051e-02
-2.08453372e-01 -8.86667371e-01 -8.58332336e-01 -6.69020414e-01
-8.03264737e-01 -1.17528760e+00 -7.93026626e-01 -7.19394505e-01
1.16480947e+00 8.55777264e-01 7.96515584e-01 4.47446525e-01
-1.68539330e-01 7.15647101e-01 -3.11127365e-01 3.07450771e-01
-4.35476974e-02 -6.50391042e-01 4.41180468e-01 5.02642155e-01
-3.43457982e-02 -6.97170615e-01 -9.60895300e-01 4.40405875e-01
-8.42999041e-01 -1.18975930e-01 5.46415925e-01 1.06460595e+00
6.36611640e-01 7.80171603e-02 6.07170701e-01 -2.13623986e-01
6.76845610e-01 -2.12073907e-01 -3.04547518e-01 2.53503442e-01
-7.46074617e-01 -3.66231084e-01 5.94332337e-01 -3.30801427e-01
-1.27643919e+00 -2.06913128e-01 -7.58928210e-02 -5.03332138e-01
-8.27446878e-02 1.34736314e-01 -6.34956419e-01 -4.66404915e-01
6.63977325e-01 3.14769775e-01 3.02291602e-01 -5.97722232e-01
3.09597433e-01 7.12696910e-01 6.36800349e-01 -5.22912979e-01
8.69751573e-01 5.26327014e-01 -1.12292893e-01 -7.07743466e-01
-5.88749111e-01 -3.56888711e-01 -3.62930566e-01 -3.70407820e-01
4.44532454e-01 -9.52903569e-01 -5.27189136e-01 6.41463161e-01
-7.52651155e-01 -2.66677082e-01 -1.46350160e-01 3.37824821e-01
-4.90069330e-01 5.91692209e-01 -4.70520467e-01 -6.48890376e-01
-4.08414960e-01 -1.26391768e+00 1.17274880e+00 6.35821104e-01
3.90662342e-01 -7.73052871e-01 -2.71742374e-01 6.81093156e-01
3.88347387e-01 -2.15297103e-01 4.60148811e-01 2.31253798e-03
-5.55209637e-01 -1.24041371e-01 -3.30971837e-01 6.62488103e-01
4.45063174e-01 -2.36886367e-01 -9.09209311e-01 -8.75715315e-01
1.24516644e-01 -4.89957482e-02 7.23597884e-01 6.19964004e-01
1.08174813e+00 -4.47182655e-01 -1.53559312e-01 1.03091538e+00
1.26562560e+00 4.61804867e-02 7.80972898e-01 3.29957634e-01
7.59757161e-01 5.45538127e-01 7.43233263e-01 2.83872128e-01
4.50445950e-01 8.34658384e-01 4.96838838e-01 -1.00051738e-01
-6.49060845e-01 -1.86465800e-01 5.88793635e-01 5.60873926e-01
-2.13542320e-02 4.83734161e-02 -5.89816928e-01 4.84432131e-01
-1.47526562e+00 -8.62975955e-01 2.49175563e-01 2.41797471e+00
5.52516520e-01 -4.34096813e-01 -1.28138095e-01 1.12024724e-01
7.13240623e-01 1.85573876e-01 -6.49619579e-01 2.07062811e-01
-4.27816391e-01 -1.40220210e-01 9.73262489e-02 6.37718379e-01
-8.79453719e-01 7.31193781e-01 5.58736658e+00 8.07928622e-01
-8.94331694e-01 3.28608334e-01 7.20347285e-01 -1.16176784e-01
-2.49668926e-01 -1.58064086e-02 -5.96711993e-01 4.92581189e-01
4.40749675e-01 -9.79495719e-02 8.53799284e-01 2.94819385e-01
5.18308461e-01 -7.84467757e-02 -6.04628146e-01 1.26364541e+00
5.40601492e-01 -8.71080637e-01 7.43498802e-02 2.35531881e-01
9.51352596e-01 -2.65247881e-01 4.05883193e-01 -1.31178781e-01
1.11451782e-02 -9.98460293e-01 3.83654952e-01 8.63311112e-01
7.23173916e-01 -6.77514374e-01 5.00801802e-01 -5.53301573e-02
-9.63202715e-01 -4.69462842e-01 -4.60108548e-01 3.69695574e-01
2.46766999e-01 6.01071715e-01 -4.29032713e-01 6.38144970e-01
9.08111036e-01 9.17923272e-01 -8.69054139e-01 1.15268683e+00
-3.42689723e-01 3.62223208e-01 -5.53021021e-02 8.02714705e-01
-2.88407624e-01 -4.47700620e-01 7.68604398e-01 5.76299489e-01
5.82156539e-01 2.69686222e-01 2.50540644e-01 5.20227790e-01
2.91993888e-03 2.01118976e-01 -3.09682310e-01 4.49751496e-01
4.62864369e-01 1.40917325e+00 -2.79837966e-01 -8.33186582e-02
-4.35119957e-01 1.13196671e+00 2.40033139e-02 7.37909973e-01
-4.72182244e-01 7.71723092e-02 9.28366363e-01 2.26756677e-01
-1.21299878e-01 -1.59573421e-01 -8.38722065e-02 -1.10805261e+00
2.02683002e-01 -1.08047009e+00 2.59814024e-01 -1.04447651e+00
-1.50335228e+00 7.31069326e-01 -2.20408410e-01 -1.38542795e+00
-1.02358945e-01 -4.01887178e-01 -5.40643096e-01 9.97567713e-01
-1.56633770e+00 -1.34164727e+00 -5.73032975e-01 9.80531752e-01
5.32428741e-01 -5.14737189e-01 5.04725039e-01 4.47615623e-01
-7.45831430e-01 7.46626616e-01 1.45355910e-01 -2.50115663e-01
1.03931320e+00 -9.06364560e-01 1.40852064e-01 1.20467043e+00
-8.89038667e-03 7.47559369e-01 6.44715667e-01 -6.12232029e-01
-1.51642072e+00 -1.11395407e+00 3.94073427e-01 -1.47565171e-01
1.36643186e-01 4.95859459e-02 -1.07156134e+00 3.42893839e-01
-7.40099996e-02 4.98804301e-01 4.68273729e-01 -2.22570539e-01
-3.47122341e-01 -5.41533828e-01 -1.53977847e+00 7.03146398e-01
1.32013285e+00 -6.15158558e-01 -2.02836841e-01 5.03384769e-01
4.89404976e-01 -2.84120262e-01 -9.35076356e-01 4.28370118e-01
5.09925902e-01 -1.10588241e+00 1.28923380e+00 -1.13414124e-01
1.51376367e-01 -7.58622825e-01 -4.90148455e-01 -1.48236513e+00
-4.28615630e-01 -3.90143096e-01 -2.41407052e-01 1.35944831e+00
-1.56704605e-01 -8.27066779e-01 5.48382044e-01 3.79715681e-01
-2.33014807e-01 -7.39219844e-01 -1.00920510e+00 -6.63388968e-01
-4.62636858e-01 -1.65242732e-01 9.42568541e-01 6.64411843e-01
-5.12016833e-01 -6.39100000e-02 -5.03137052e-01 5.58696330e-01
1.02668750e+00 4.59542461e-02 8.19408238e-01 -1.01820993e+00
-4.81288992e-02 -2.63970077e-01 -4.51964349e-01 -9.37201440e-01
1.99897185e-01 -7.65268207e-01 4.12032232e-02 -1.70242023e+00
2.85634220e-01 -3.49703431e-01 -3.83723676e-01 4.80468035e-01
-4.04659510e-01 8.70563388e-01 2.36230835e-01 5.54319859e-01
-3.87071401e-01 8.55127156e-01 1.49315035e+00 -1.16129451e-01
-2.35422060e-01 -6.04852363e-02 -9.01367188e-01 4.20396119e-01
6.39600813e-01 -7.65911266e-02 -3.64510447e-01 -5.69993377e-01
-1.74666792e-01 3.50920185e-02 6.48493230e-01 -1.03043723e+00
2.09094197e-01 -1.33161947e-01 7.11556733e-01 -1.49185896e-01
6.43596649e-01 -7.12654769e-01 1.10697351e-01 6.50812760e-02
1.17553577e-01 -1.33757010e-01 -6.53481930e-02 5.75503528e-01
-2.13652357e-01 3.02660950e-02 9.13727880e-01 1.36769727e-01
-6.66751087e-01 7.64608860e-01 2.78422922e-01 -4.45070386e-01
5.83169103e-01 -6.40090406e-01 -4.21406478e-01 -6.58329606e-01
-5.44207275e-01 1.61151513e-01 8.19389522e-01 4.98761535e-01
1.12751472e+00 -1.52766252e+00 -9.36116517e-01 6.52219951e-01
6.62092417e-02 -2.27543339e-01 7.52234519e-01 8.85394454e-01
-1.59055933e-01 6.04273900e-02 -3.03677708e-01 -6.12321436e-01
-1.33072031e+00 3.89994144e-01 5.45133233e-01 2.72515148e-01
-6.95989788e-01 6.91939116e-01 2.98592538e-01 -5.06651759e-01
1.33042186e-01 3.42241406e-01 -3.00562203e-01 -2.07405314e-01
9.04506266e-01 3.79599005e-01 1.90483689e-01 -1.24239945e+00
-2.77968973e-01 8.90019834e-01 1.42250163e-03 -1.58270478e-01
1.33630788e+00 -6.52756393e-01 -3.81045252e-01 -2.03855634e-01
1.16316307e+00 1.94099694e-01 -1.66722202e+00 -4.30884957e-01
-4.30377752e-01 -1.14674675e+00 1.51242703e-01 -7.19625592e-01
-1.52176762e+00 4.77871567e-01 1.16626501e+00 -2.12322429e-01
1.76761532e+00 -1.99079677e-01 5.42580187e-01 -1.03503488e-01
3.04259241e-01 -6.17742658e-01 3.54683936e-01 6.33423403e-02
1.29315507e+00 -1.26214242e+00 3.34350675e-01 -4.24309313e-01
-3.51017594e-01 8.36001039e-01 7.06608653e-01 2.52422653e-02
4.56855625e-01 -2.84420788e-01 2.49163061e-01 -1.68601856e-01
-3.06244165e-01 4.37546382e-03 4.39811617e-01 8.02410483e-01
1.52181342e-01 -2.17513051e-02 -1.73156396e-01 3.13551873e-01
-1.05939351e-01 -1.98335335e-01 3.40781152e-01 5.94615519e-01
-3.40798169e-01 -9.51313019e-01 -8.78946960e-01 5.36674857e-01
-2.62608886e-01 -3.01253200e-02 1.33124217e-01 1.85246244e-01
1.92343205e-01 1.38153744e+00 -1.31472126e-01 -3.05147767e-01
3.44406158e-01 -2.73715407e-01 5.67370892e-01 -4.67905462e-01
-1.15036048e-01 1.85442343e-01 -3.44360262e-01 -7.45514810e-01
-4.32351083e-01 -8.30814779e-01 -1.05116177e+00 -3.40778857e-01
-1.61000162e-01 -7.55788237e-02 5.80740869e-01 8.34915400e-01
4.36828941e-01 2.47658208e-01 1.04961944e+00 -1.27276051e+00
-2.90575296e-01 -1.01926029e+00 -8.17784548e-01 5.04880667e-01
6.14152610e-01 -8.76046419e-01 -6.01527929e-01 5.20697050e-02] | [12.846181869506836, -0.05700518563389778] |
d776cf4e-89a1-469e-8392-9782f0b9156d | dense-interspecies-face-embedding | null | null | https://openreview.net/forum?id=m67FNFdgLO9 | https://openreview.net/pdf?id=m67FNFdgLO9 | Dense Interspecies Face Embedding | Dense Interspecies Face Embedding (DIFE) is a new direction for understanding faces of various animals by extracting common features among animal faces including human face. There are three main obstacles for interspecies face understanding: (1) lack of animal data compared to human, (2) ambiguous connection between faces of various animals, and (3) extreme shape and style variance. To cope with the lack of data, we utilize multi-teacher knowledge distillation of CSE and StyleGAN2 requiring no additional data or label. Then we synthesize pseudo pair images through the latent space exploration of StyleGAN2 to find implicit associations between different animal faces. Finally, we introduce the semantic matching loss to overcome the problem of extreme shape differences between species. To quantitatively evaluate our method over possible previous methodologies like unsupervised keypoint detection, we perform interspecies facial keypoint transfer on MAFL and AP-10K. Furthermore, the results of other applications like interspecies face image manipulation and dense keypoint transfer are provided. The code is available at https://github.com/kingsj0405/dife. | ['Seon Joo Kim', 'Seonghyeon Nam', 'Subin Jeon', 'Sejong Yang'] | 2022-11-28 | null | null | null | neruips-2022-11 | ['interspecies-facial-keypoint-transfer', 'keypoint-detection', 'image-manipulation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.39265656e-01 -1.14456378e-01 1.66784927e-01 -5.56481242e-01
-2.92933106e-01 -6.91273570e-01 4.39230651e-01 -6.15817010e-01
2.24140957e-02 5.38176596e-01 1.01588473e-01 2.06753090e-01
-3.97521369e-02 -5.66734731e-01 -7.08584607e-01 -5.17161071e-01
-1.38239861e-02 1.99557111e-01 -2.37960309e-01 -2.55183250e-01
1.17068425e-01 6.95738673e-01 -1.58468640e+00 2.76174158e-01
5.65971434e-01 8.91663194e-01 3.88406254e-02 4.24415708e-01
-2.69155711e-01 2.40829453e-01 -3.23734730e-01 -6.25648737e-01
5.29589772e-01 -5.07588387e-01 -6.37631893e-01 -4.37247902e-02
9.36701894e-01 -5.03781438e-01 -2.50805497e-01 1.22935438e+00
4.35554743e-01 -9.34542120e-02 7.40674973e-01 -1.86912060e+00
-1.00258720e+00 2.23927900e-01 -9.92794633e-01 7.35446997e-03
3.96804154e-01 2.31792465e-01 6.37375116e-01 -1.12294900e+00
5.83351016e-01 1.84067178e+00 6.28148854e-01 8.78146231e-01
-1.23221910e+00 -1.48800778e+00 -1.14182040e-01 2.90498972e-01
-1.77416778e+00 -8.45446706e-01 9.90464032e-01 -5.32684028e-01
3.36159796e-01 1.65990472e-01 6.30783439e-01 1.18916202e+00
-1.25314267e-02 5.31184614e-01 1.21263921e+00 -2.22417459e-01
-2.00103059e-01 2.61849582e-01 -2.24472582e-01 1.10609245e+00
-2.15241499e-02 4.09215540e-01 -8.76738906e-01 -2.99294740e-01
1.03896320e+00 1.84272937e-02 -1.87072083e-01 -3.92587751e-01
-1.09064496e+00 7.96465933e-01 5.04826069e-01 -2.65581477e-02
9.11012292e-03 1.69466183e-01 9.45541859e-02 4.88235414e-01
3.05682838e-01 3.16003323e-01 -5.50284982e-01 3.70723605e-01
-7.42489934e-01 7.03794286e-02 6.51059508e-01 1.23697960e+00
1.17420018e+00 1.22369222e-01 3.63825530e-01 8.50778103e-01
6.68195605e-01 7.32224882e-01 1.87317625e-01 -1.09391713e+00
5.46726994e-02 3.71338487e-01 -2.48271525e-01 -1.47469044e+00
1.11725703e-01 2.13625446e-01 -8.22129905e-01 3.16231310e-01
2.24608749e-01 2.90068332e-02 -9.68046904e-01 1.95243752e+00
4.96659249e-01 5.72255373e-01 -3.02538816e-02 6.55331612e-01
1.30316007e+00 5.75339198e-01 3.74444425e-02 3.19153210e-03
1.65084052e+00 -7.52302468e-01 -6.67097688e-01 -1.48058966e-01
2.03991815e-01 -9.88098025e-01 8.94064128e-01 -3.02727688e-02
-7.03239202e-01 -6.26616240e-01 -8.40764403e-01 -1.47310868e-01
-5.64945996e-01 1.04411095e-01 7.36390650e-01 5.47370911e-01
-1.19396257e+00 3.08833778e-01 -5.03975093e-01 -5.09681880e-01
7.32863665e-01 5.48509955e-01 -9.42356467e-01 1.37105798e-02
-1.03908050e+00 6.05982780e-01 1.66898236e-01 2.67218888e-01
-8.79009664e-01 -1.17971146e+00 -1.07485962e+00 -1.57779545e-01
2.62505740e-01 -6.73258424e-01 7.78306901e-01 -1.20523751e+00
-1.66814494e+00 1.23327482e+00 -3.58521603e-02 1.45925000e-01
2.74914563e-01 -2.59595722e-01 -4.07341182e-01 3.13620478e-01
-1.59845818e-02 1.40873778e+00 1.21327651e+00 -1.40027130e+00
-1.88339949e-01 -5.09860694e-01 -1.03538744e-01 -7.91262463e-02
-1.73724845e-01 2.94980645e-01 -2.91263700e-01 -9.04117286e-01
-3.19688357e-02 -1.15134013e+00 3.03284168e-01 8.54328692e-01
-1.91385865e-01 -9.08285081e-02 1.12458050e+00 -6.36812031e-01
5.36987722e-01 -2.33098102e+00 2.32284799e-01 1.50047734e-01
3.13208133e-01 2.06164911e-01 -7.01491714e-01 2.28666142e-01
-3.78086060e-01 7.17471391e-02 -2.52735883e-01 -8.92173499e-02
-1.48573518e-01 1.57162935e-01 -3.33991647e-01 5.57206392e-01
4.41276997e-01 9.65949297e-01 -7.63244390e-01 -6.73104227e-01
3.43707204e-02 6.39821768e-01 -7.38424063e-01 2.91351646e-01
-6.68818951e-02 6.49893165e-01 -3.80774826e-01 8.45529377e-01
1.07234848e+00 -9.36515555e-02 -1.05311780e-03 -8.05940032e-01
1.31040305e-01 -3.51268858e-01 -1.18269897e+00 1.68515086e+00
-1.83250278e-01 4.92003858e-01 3.39721203e-01 -4.75921571e-01
7.62437999e-01 2.36607611e-01 3.49375993e-01 -2.34004021e-01
2.36573517e-01 -8.96420144e-03 -8.77651200e-03 -2.96838284e-01
2.11791590e-01 -2.65420973e-01 2.49638021e-01 4.24475670e-01
5.73788822e-01 -3.52275252e-01 -2.24332452e-01 2.37184286e-01
4.85104620e-01 1.12586856e-01 1.46297738e-01 -5.90062261e-01
4.50521886e-01 -5.69016397e-01 6.40999675e-01 4.53595109e-02
-3.34092110e-01 4.90260273e-01 2.45095730e-01 -4.24049288e-01
-7.13408768e-01 -1.23934042e+00 -1.96395531e-01 1.01294458e+00
2.07818761e-01 -4.77963388e-01 -6.20834291e-01 -6.34940326e-01
2.42196873e-01 8.09377655e-02 -9.48340952e-01 -1.56684905e-01
-3.36213082e-01 -2.55232453e-01 7.44545817e-01 4.10124838e-01
6.96604073e-01 -1.09963405e+00 -8.47550705e-02 -3.78557235e-01
-6.81087673e-02 -1.10679364e+00 -8.05115163e-01 -3.70877475e-01
-4.21292454e-01 -1.31035769e+00 -6.30710423e-01 -9.97915864e-01
9.97291088e-01 3.63193274e-01 9.12912667e-01 2.68092066e-01
-6.85361385e-01 7.85308182e-01 -1.94361851e-01 -3.35447162e-01
-1.64056897e-01 -4.69567299e-01 5.04238188e-01 4.79274467e-02
5.36761463e-01 -6.40383363e-01 -5.64926863e-01 6.20161176e-01
-8.44188035e-01 4.11725137e-03 3.29643160e-01 8.08052421e-01
4.15129393e-01 -1.41063586e-01 4.04665947e-01 -7.41709948e-01
9.57221016e-02 -3.28184932e-01 -7.60778308e-01 2.46580511e-01
-9.40334573e-02 -6.69435039e-02 1.58895046e-01 -5.60018718e-01
-1.03178382e+00 1.16509996e-01 5.85335195e-02 -7.10110307e-01
-2.62521982e-01 -1.46016344e-01 -4.88233000e-01 -5.93366146e-01
4.15279180e-01 6.52821213e-02 4.15094525e-01 -3.90345156e-01
5.82406521e-01 5.29346526e-01 5.68238497e-01 -7.01377928e-01
1.24488616e+00 6.16878033e-01 -1.33640617e-01 -1.12830615e+00
-4.96025354e-01 6.40824379e-04 -8.52837324e-01 -1.93567872e-01
1.08978999e+00 -1.02485538e+00 -8.67001474e-01 7.39017785e-01
-1.20301223e+00 -1.50065869e-01 8.34361315e-02 3.89646053e-01
-5.00808060e-01 4.35415596e-01 -7.06721365e-01 -2.52375871e-01
-4.11106288e-01 -1.01294887e+00 1.02354741e+00 3.86431217e-01
-2.61350513e-01 -7.96218157e-01 -5.15526533e-02 4.35504496e-01
2.33970344e-01 1.02072582e-01 8.24027717e-01 -2.08605915e-01
-5.69190383e-01 2.96359658e-01 -3.76660079e-01 3.83361489e-01
5.86313665e-01 3.71148854e-01 -1.01626182e+00 -6.03871465e-01
-1.71987832e-01 -5.68074405e-01 4.74482626e-01 1.41624520e-02
1.20593631e+00 -3.48928928e-01 -3.51208746e-01 9.95884597e-01
1.09932375e+00 2.31815688e-02 5.34798980e-01 -2.43654743e-01
9.23496425e-01 9.89804864e-01 3.71295244e-01 2.25219056e-01
3.56559873e-01 5.84236920e-01 1.19255431e-01 -1.44843683e-01
-2.16633439e-01 -5.48875451e-01 4.41997200e-01 9.59879577e-01
1.17783673e-01 8.44288766e-02 -6.83544219e-01 4.50412661e-01
-1.15068555e+00 -8.45787466e-01 3.65206182e-01 1.68179440e+00
8.05141389e-01 -6.07655108e-01 -1.53069114e-02 -3.57605040e-01
8.09845626e-01 -3.57596651e-02 -5.84111333e-01 -7.49219730e-02
-9.67026427e-02 1.82923645e-01 2.14271039e-01 4.28947002e-01
-9.14152741e-01 1.32772446e+00 6.28543758e+00 7.08858371e-01
-1.12127614e+00 -6.77984431e-02 4.20624316e-01 1.17615968e-01
-2.72272915e-01 -1.21691808e-01 -8.38225663e-01 4.27607626e-01
3.13082337e-01 -1.68555558e-01 5.65312386e-01 5.25583684e-01
-4.70602125e-01 3.16921741e-01 -1.31821692e+00 1.28927398e+00
3.01242173e-01 -1.00460267e+00 2.85973042e-01 4.38682027e-02
5.85991621e-01 -1.82228789e-01 3.85490328e-01 3.50681506e-02
3.55499029e-01 -1.16560674e+00 4.61527109e-01 2.97372818e-01
1.16961646e+00 -6.03052378e-01 2.13919595e-01 -3.55611108e-02
-1.36173165e+00 2.35446975e-01 -5.24740875e-01 3.61877680e-01
-2.35595465e-01 -4.09982391e-02 -4.26768810e-01 3.03946614e-01
9.11929846e-01 8.44113827e-01 -5.61799943e-01 5.39330602e-01
-2.63336092e-01 2.40478620e-01 -5.25413096e-01 5.04972637e-01
-1.62356496e-01 -2.85502523e-01 5.50761580e-01 1.00560832e+00
3.94900769e-01 3.61171365e-01 -1.36723295e-01 1.15981042e+00
-3.39491725e-01 4.80468944e-02 -8.85465145e-01 -1.62160784e-01
7.00372934e-01 1.33319628e+00 -6.09060466e-01 -1.15113854e-01
-4.51036334e-01 1.15880442e+00 1.20734908e-01 2.95325875e-01
-7.18381405e-01 -2.11815551e-01 1.16963315e+00 -6.99089793e-03
2.47695923e-01 -1.13324761e-01 2.86100268e-01 -1.27940428e+00
-2.70278931e-01 -1.15531766e+00 3.30173165e-01 -9.05889630e-01
-1.63488054e+00 4.96940553e-01 2.44542837e-01 -1.08136570e+00
1.13392770e-01 -7.81717718e-01 -5.17336726e-01 6.86795056e-01
-1.43365622e+00 -1.43449187e+00 -3.67317319e-01 8.74692202e-01
4.93359089e-01 -5.12047231e-01 1.02220166e+00 4.31005806e-01
-2.84167558e-01 9.73964572e-01 -3.25887442e-01 3.04353952e-01
1.01781058e+00 -8.28404069e-01 4.27011639e-01 4.72445577e-01
1.47151470e-01 9.49362934e-01 4.17235732e-01 -6.78565383e-01
-1.64314079e+00 -9.59109366e-01 2.58486837e-01 -4.99904186e-01
5.92663646e-01 -6.80129945e-01 -1.04784179e+00 8.88551652e-01
2.63259053e-01 2.62530893e-01 1.13346469e+00 -1.59987621e-02
-7.80180216e-01 -1.95350096e-01 -1.39944494e+00 7.60895431e-01
1.20254076e+00 -8.08165014e-01 -4.59799677e-01 6.14812262e-02
6.81559980e-01 -1.33071288e-01 -9.52350736e-01 5.08503854e-01
9.01633620e-01 -6.13084018e-01 1.08129239e+00 -5.92871070e-01
4.46863711e-01 -5.99596858e-01 -2.92499989e-01 -1.35979247e+00
-4.37270612e-01 -6.52854443e-01 1.51446775e-01 1.56393015e+00
-2.36775242e-02 -6.74610794e-01 7.26597905e-01 4.21218216e-01
4.61894363e-01 -2.77154893e-01 -7.54403174e-01 -6.68010712e-01
2.35119537e-01 1.03585228e-01 8.99849236e-01 1.36902082e+00
-4.12195861e-01 2.43803069e-01 -6.24182284e-01 3.77675623e-01
8.82284522e-01 1.58343002e-01 1.03334296e+00 -1.11426139e+00
-2.10465640e-01 -2.07199827e-01 -6.51918888e-01 -8.98312628e-01
6.31725907e-01 -9.60418224e-01 -1.30828470e-01 -6.89062297e-01
2.68747061e-01 -3.26148629e-01 -2.27162123e-01 7.22299039e-01
4.00329158e-02 6.61623120e-01 3.97819430e-01 2.45131142e-02
-2.34144554e-01 8.26249659e-01 1.43543124e+00 -1.02692001e-01
2.07890078e-01 -5.77476144e-01 -7.25935400e-01 9.16369617e-01
6.73758149e-01 -5.20961404e-01 -4.49488372e-01 -4.86280084e-01
-2.84775030e-02 -4.32205983e-02 6.05728209e-01 -6.60835624e-01
1.52212217e-01 -2.38561705e-01 7.00704157e-01 -3.63529593e-01
4.42157865e-01 -1.00632310e+00 4.57604229e-01 3.13894957e-01
-7.00848177e-02 7.05540106e-02 5.37820995e-01 5.84977448e-01
-1.16981171e-01 1.90168470e-01 1.10519171e+00 -1.97319403e-01
-7.63342381e-01 6.61015391e-01 -9.91794318e-02 -9.64873210e-02
1.02473342e+00 -2.01125517e-01 -2.36021757e-01 -2.51662195e-01
-5.03949583e-01 3.25565726e-01 6.29498005e-01 7.94104874e-01
7.99962223e-01 -1.54491186e+00 -8.32459152e-01 8.43344629e-01
3.43603402e-01 -3.36571306e-01 4.60135579e-01 4.42133695e-01
-5.10178328e-01 -9.53491312e-03 -8.36599469e-01 -5.39930761e-01
-1.75273538e+00 5.15699506e-01 3.53205860e-01 5.48710406e-01
-4.68576908e-01 1.26518989e+00 9.08341765e-01 -8.17080438e-01
2.46503130e-02 1.27731293e-01 -8.10765326e-02 9.16152000e-02
6.11015618e-01 7.99927115e-02 -4.74375606e-01 -1.09847915e+00
-6.29015088e-01 1.13267100e+00 -2.33715951e-01 2.37219930e-01
1.34766877e+00 -1.00783899e-01 -4.68940973e-01 7.13220239e-02
1.53100228e+00 9.67681929e-02 -1.19208598e+00 -3.22210848e-01
-4.49100614e-01 -8.08737755e-01 -2.65804946e-01 -5.82238078e-01
-1.25351548e+00 9.17061508e-01 9.22009706e-01 -5.63471794e-01
1.07918930e+00 3.25475000e-02 6.84046566e-01 1.86829418e-01
3.67887467e-01 -7.24743962e-01 1.74954042e-01 2.40087718e-01
1.21323490e+00 -1.40759504e+00 1.03028983e-01 -6.86473072e-01
-4.65556055e-01 9.46340442e-01 8.80712688e-01 8.08181614e-02
1.03093338e+00 2.70303190e-01 3.31419975e-01 -4.27481651e-01
-6.91598952e-01 3.76075739e-03 4.02115613e-01 6.78017974e-01
3.32063168e-01 -7.52111077e-02 1.70201302e-01 2.40423411e-01
-4.95584756e-01 -9.53045499e-04 1.48245022e-01 8.19572389e-01
2.37523876e-02 -1.14732778e+00 -3.17325056e-01 1.62025914e-01
-4.58573788e-01 -6.78552166e-02 -5.49082220e-01 5.77983260e-01
2.01878905e-01 6.98103070e-01 7.49118328e-02 -4.41337883e-01
3.74948941e-02 -5.10504097e-02 8.61808479e-01 -6.20611250e-01
-1.95357516e-01 -1.42857909e-01 -3.96823913e-01 -5.71860194e-01
-4.75193888e-01 -3.96165609e-01 -8.78286839e-01 -5.80851912e-01
-3.80568445e-01 3.78290154e-02 5.18046081e-01 5.77692509e-01
5.11702836e-01 4.70556580e-02 5.31690657e-01 -9.38151658e-01
-1.72003999e-01 -8.01445365e-01 -6.87431276e-01 4.18386132e-01
4.01771575e-01 -9.63587761e-01 -4.32852864e-01 4.91157770e-01] | [12.865028381347656, 0.02118346467614174] |
61d18c2f-e4f8-487b-80d0-d0830647cf7c | incorporating-expert-opinion-on-observable | 2302.06391 | null | https://arxiv.org/abs/2302.06391v1 | https://arxiv.org/pdf/2302.06391v1.pdf | Incorporating Expert Opinion on Observable Quantities into Statistical Models -- A General Framework | This article describes an approach to incorporate expert opinion on observable quantities through the use of a loss function which updates a prior belief as opposed to specifying parameters on the priors. Eliciting information on observable quantities allows experts to provide meaningful information on a quantity familiar to them, in contrast to elicitation on model parameters, which may be subject to interactions with other parameters or non-linear transformations before obtaining an observable quantity. The approach to incorporating expert opinion described in this paper is distinctive in that we do not specify a prior to match an expert's opinion on observed quantity, rather we obtain a posterior by updating the model parameters through a loss function. This loss function contains the observable quantity, expressed a function of the parameters, and is related to the expert's opinion which is typically operationalized as a statistical distribution. Parameters which generate observable quantities which are further from the expert's opinion incur a higher loss, allowing for the model parameters to be estimated based on their fidelity to both the data and expert opinion, with the relative strength determined by the number of observations and precision of the elicited belief. Including expert opinion in this fashion allows for a flexible specification of the opinion and in many situations is straightforward to implement with commonly used probabilistic programming software. We highlight this using three worked examples of varying model complexity including survival models, a multivariate normal distribution and a regression problem. | ['Arthur White', 'Philip Cooney'] | 2023-02-10 | null | null | null | null | ['probabilistic-programming'] | ['methodology'] | [ 2.81134665e-01 6.24806583e-01 -1.35469884e-01 -6.43873334e-01
-1.08143282e+00 -7.40765750e-01 5.22493899e-01 6.65954828e-01
-6.61946118e-01 1.06789780e+00 1.08257741e-01 -2.38046736e-01
-4.05168682e-01 -7.94224262e-01 -4.77480620e-01 -8.02179694e-01
1.30933106e-01 6.69739842e-01 2.32504427e-01 1.35748178e-01
4.60904092e-01 4.88233328e-01 -1.24616575e+00 -4.49404158e-02
7.41863370e-01 8.62288952e-01 -2.44011268e-01 5.83010554e-01
1.34440407e-01 6.12327397e-01 -8.96075845e-01 -4.50580657e-01
1.46556884e-01 -2.81888038e-01 -5.89441955e-01 2.66692489e-01
-9.68581960e-02 -5.90497404e-02 4.44660425e-01 1.00727773e+00
2.81884462e-01 9.41350609e-02 1.36243844e+00 -1.20002913e+00
-4.51817095e-01 2.80540198e-01 -1.35792762e-01 7.65896514e-02
8.27899933e-01 1.72167838e-01 1.04138970e+00 -5.48117399e-01
2.64525741e-01 1.37333906e+00 6.90567017e-01 -6.57052919e-02
-1.67239046e+00 -8.71118084e-02 3.16273451e-01 -2.12473124e-01
-1.46685433e+00 -2.19700560e-01 4.11745727e-01 -7.72505283e-01
6.00048184e-01 3.26244682e-01 6.19216919e-01 6.50645316e-01
5.42602479e-01 3.54756087e-01 1.21556318e+00 -4.53998446e-01
7.54448414e-01 7.55316019e-01 8.60511661e-02 4.06619012e-01
2.10301176e-01 3.88651013e-01 -2.67492175e-01 -7.06639528e-01
7.08644629e-01 -1.42952234e-01 -3.35326821e-01 -4.42513019e-01
-8.63135159e-01 9.78398025e-01 4.50044461e-02 -7.21750632e-02
-6.69335842e-01 2.78031856e-01 7.43538216e-02 4.82830226e-01
5.10981977e-01 5.06800473e-01 -5.96218407e-01 6.72977418e-02
-6.10245109e-01 3.12641203e-01 1.19736433e+00 5.15073955e-01
1.05457890e+00 -2.65279651e-01 -2.24580646e-01 3.26821446e-01
8.25461566e-01 5.00111401e-01 1.79564394e-02 -1.06951928e+00
1.05804093e-01 5.30230045e-01 7.16153622e-01 -8.64189029e-01
-9.80833471e-02 -2.34819025e-01 -2.27480605e-01 6.02078021e-01
5.79254270e-01 -4.31691140e-01 -9.07748520e-01 1.77024829e+00
3.97057265e-01 -3.16703975e-01 -1.82715803e-02 6.47873700e-01
2.90692329e-01 6.02072537e-01 2.03063712e-01 -4.14149135e-01
1.26200891e+00 -1.65943593e-01 -6.60798848e-01 -7.24171177e-02
3.32037657e-01 -6.35738194e-01 7.15913892e-01 5.14771461e-01
-1.09880769e+00 -3.13047469e-01 -8.91290367e-01 2.52946287e-01
-3.10736448e-01 -2.82190233e-01 5.96720517e-01 8.85906994e-01
-1.18338668e+00 5.88672578e-01 -7.68276036e-01 -1.23378471e-01
-8.86936262e-02 7.10185409e-01 -1.43373325e-01 3.64160210e-01
-1.27947092e+00 1.51784265e+00 4.00914401e-01 5.30988276e-02
-6.88910127e-01 -7.33862579e-01 -9.56061780e-01 2.97668874e-01
4.03066516e-01 -8.62551332e-01 1.27920556e+00 -8.72132778e-01
-1.62468612e+00 6.21370375e-01 -8.21525231e-02 -2.10061312e-01
5.19466639e-01 2.44825527e-01 -3.54814947e-01 -7.07222968e-02
5.56962471e-03 5.67350090e-01 8.59401226e-01 -1.28114522e+00
-5.15003264e-01 -9.05725807e-02 3.99093777e-01 2.90823489e-01
3.27491641e-01 -2.90857162e-02 -6.47983700e-02 -2.78777391e-01
4.99873012e-02 -8.05568039e-01 -5.16358197e-01 2.96368361e-01
-2.19292581e-01 -2.22092301e-01 3.74732614e-01 -3.96048099e-01
1.16936064e+00 -1.90894341e+00 2.70680077e-02 7.50683844e-01
8.93140435e-02 -3.64485055e-01 2.77186334e-01 7.16029525e-01
-1.74422517e-01 1.31213605e-01 -6.80737972e-01 -9.80916619e-02
3.01719576e-01 2.77817607e-01 -1.43088341e-01 6.93403065e-01
3.45892519e-01 4.40996945e-01 -8.52022648e-01 -3.55224848e-01
2.79057086e-01 4.17273641e-01 -4.60449904e-01 9.42889825e-02
-1.31652206e-01 2.59090483e-01 -5.49941778e-01 2.62393892e-01
2.10628852e-01 -2.19817474e-01 -4.77340706e-02 1.92416273e-02
-1.31336860e-02 2.13434875e-01 -1.67266846e+00 7.87811041e-01
-3.43589306e-01 1.75283223e-01 3.94620523e-02 -1.10033751e+00
9.53236759e-01 7.16985166e-01 3.45011979e-01 2.23056197e-01
1.59325168e-01 4.76222746e-02 -5.98568618e-02 -2.00840026e-01
2.11067453e-01 -6.74255490e-01 -2.14718655e-01 7.06768811e-01
-1.23232901e-02 -7.50026703e-01 4.16234657e-02 -3.67735401e-02
8.17286015e-01 -5.53743765e-02 7.89506972e-01 -5.05859375e-01
3.90325427e-01 -2.13314667e-01 3.68298769e-01 8.58285785e-01
3.43191717e-03 3.41582447e-01 9.03465927e-01 -1.12155944e-01
-7.51198053e-01 -1.16110778e+00 -4.56046611e-01 6.06777608e-01
-1.77071691e-01 7.25696832e-02 -4.51316476e-01 -6.54047251e-01
1.53185919e-01 1.05948353e+00 -8.92072678e-01 -1.05512440e-01
1.18799508e-01 -7.30326235e-01 -1.72781557e-01 4.06154960e-01
-1.40335247e-01 -1.00252450e+00 -4.14733946e-01 3.00934613e-01
2.16784120e-01 -4.15752798e-01 -5.13548374e-01 4.86589700e-01
-8.81038785e-01 -8.30406845e-01 -5.37100017e-01 -7.23291487e-02
1.14267325e+00 -5.95418572e-01 1.16018510e+00 -1.43004939e-01
6.95276186e-02 7.40679026e-01 3.92236970e-02 -8.08692932e-01
-4.80645508e-01 -4.62702036e-01 -8.67492426e-03 -7.40666315e-02
3.35165203e-01 -3.98978680e-01 -4.00462776e-01 2.18475029e-01
-1.06838131e+00 -6.12101972e-01 4.48762327e-01 7.36409366e-01
4.20685947e-01 3.58971715e-01 4.92716461e-01 -9.82365787e-01
8.22394669e-01 -6.40467584e-01 -9.11004543e-01 2.41984844e-01
-6.90285563e-01 3.52944613e-01 1.75866753e-01 -5.41504204e-01
-1.22714794e+00 1.62673444e-02 2.17347905e-01 1.86887175e-01
-2.65266657e-01 9.95938599e-01 -1.01349704e-01 1.55875444e-01
7.45893598e-01 -3.95444095e-01 -1.63638309e-01 -2.14807227e-01
2.25116953e-01 5.12911201e-01 2.77869105e-01 -8.47573757e-01
5.74731946e-01 2.47439027e-01 3.06113195e-02 -2.45215192e-01
-9.68756258e-01 -3.21390092e-01 -4.92979944e-01 -6.04865551e-02
7.31824100e-01 -5.64713895e-01 -7.12119222e-01 1.72811896e-01
-1.01215720e+00 -1.09808780e-01 -7.77557373e-01 8.29701662e-01
-6.22001290e-01 7.71189854e-02 -3.62234004e-02 -1.26872909e+00
1.73841983e-01 -1.30145359e+00 1.04815960e+00 8.14189315e-02
-7.50319898e-01 -1.73200893e+00 1.78820968e-01 -7.26747364e-02
1.83068916e-01 2.15233043e-01 9.78780210e-01 -8.54690313e-01
-4.90993232e-01 -7.37587988e-01 -6.48167878e-02 4.10917789e-01
1.99478790e-01 1.27563044e-01 -9.36944067e-01 -3.09764415e-01
4.31349069e-01 -1.43050194e-01 4.39687550e-01 6.60766244e-01
8.53753686e-01 -3.52809846e-01 -2.11511075e-01 -7.83172324e-02
1.37705529e+00 3.14333975e-01 4.83227342e-01 -2.39941888e-02
5.32691460e-03 7.62595594e-01 3.93839300e-01 5.56818306e-01
8.31773952e-02 4.04467523e-01 1.86847940e-01 -8.11561272e-02
6.90427661e-01 1.33818770e-02 2.73285538e-01 2.82084435e-01
-1.39194187e-02 -4.59443301e-01 -6.99168622e-01 5.34814954e-01
-1.68423486e+00 -8.67741883e-01 1.23477288e-01 2.50080633e+00
1.10537243e+00 3.86277914e-01 -3.21064070e-02 1.71450097e-02
6.45091534e-01 -1.69219509e-01 -6.50422215e-01 -5.75010896e-01
3.79112720e-01 1.24498144e-01 6.11468971e-01 1.12976813e+00
-6.54846728e-01 1.83845878e-01 8.00705528e+00 6.47300005e-01
-6.89537704e-01 -1.71040758e-01 5.86957455e-01 2.97599018e-01
-8.60240459e-01 5.98612964e-01 -7.18622446e-01 5.17150402e-01
8.87166917e-01 -4.44411457e-01 2.50235319e-01 5.17485023e-01
3.61608893e-01 -7.54536211e-01 -1.58773243e+00 2.05150321e-01
-4.22463179e-01 -7.92566180e-01 -1.52911693e-01 3.14864099e-01
6.35639668e-01 -5.60635269e-01 -1.98784787e-02 1.26266032e-01
9.39209104e-01 -1.24010777e+00 7.45393813e-01 1.05681098e+00
5.11623085e-01 -6.82910800e-01 1.10574043e+00 4.20074195e-01
-6.82088196e-01 2.06299320e-01 -1.92239121e-01 -4.23161983e-01
2.07715034e-01 8.32163036e-01 -1.01448202e+00 1.88183799e-01
3.31232786e-01 1.11253433e-01 -8.20780545e-02 1.00321102e+00
-3.03503096e-01 6.68529272e-01 -7.17377543e-01 -7.77452216e-02
4.49948795e-02 -2.90414631e-01 4.93899286e-01 9.87018943e-01
2.06057459e-01 2.26952419e-01 2.95834929e-01 1.08110094e+00
4.63092536e-01 8.01892858e-03 -3.06236118e-01 4.61049490e-02
6.33516014e-01 8.92064571e-01 -6.14673376e-01 -3.69318277e-01
-4.86571878e-01 3.16187233e-01 4.30087000e-02 6.19980335e-01
-5.08044899e-01 -2.71119863e-01 2.14216515e-01 5.54875433e-02
2.25223839e-01 3.31891388e-01 -2.07745388e-01 -7.60410786e-01
-5.00084013e-02 -9.08362567e-01 3.84912550e-01 -8.70768368e-01
-1.60782933e+00 2.66490549e-01 7.67838120e-01 -9.66558099e-01
-6.48155749e-01 -6.41742706e-01 -6.83248341e-01 1.48289621e+00
-8.98798287e-01 -5.86238205e-01 3.05959016e-01 1.74406171e-01
-1.10413730e-01 7.36824647e-02 9.97812390e-01 -4.20821846e-01
-1.75514951e-01 1.70126826e-01 -1.76832601e-01 -2.33516768e-01
5.52284718e-01 -1.59642768e+00 -9.34074968e-02 3.25753450e-01
-3.70167166e-01 8.20527673e-01 1.27617693e+00 -6.76839471e-01
-5.70241928e-01 -3.98419827e-01 8.07640553e-01 -7.00735092e-01
8.80298138e-01 9.16866586e-02 -8.84781897e-01 8.80725205e-01
1.32243142e-01 -2.15850547e-01 8.41963291e-01 2.75856256e-01
1.87695306e-02 8.23071748e-02 -1.68647873e+00 4.04147387e-01
2.14847505e-01 -5.46460569e-01 -7.61869907e-01 2.16764703e-01
3.38673681e-01 -1.06719494e-01 -1.24917293e+00 4.25983369e-01
4.02455270e-01 -8.85371029e-01 7.08333194e-01 -5.70905685e-01
-8.54277462e-02 -4.38805640e-01 -1.71092704e-01 -1.55646527e+00
-3.27575117e-01 -5.19330442e-01 1.53430730e-01 1.00143230e+00
7.71458566e-01 -1.04906893e+00 5.55622876e-01 1.37896597e+00
2.87445962e-01 -7.48905063e-01 -8.30182791e-01 -4.07608747e-01
6.34199306e-02 -4.45852548e-01 5.53740740e-01 7.83073127e-01
4.44563404e-02 1.45971775e-01 2.84712147e-02 4.36977893e-01
6.30332232e-01 -1.77754402e-01 3.98674428e-01 -1.46306252e+00
-5.63147604e-01 -2.47713029e-01 -5.46945572e-01 -8.84986043e-01
-3.08232196e-02 -4.38560337e-01 2.15601981e-01 -1.36828923e+00
2.81839103e-01 -3.15110713e-01 -2.26658598e-01 3.95319700e-01
-2.18459934e-01 -2.05770642e-01 -3.22178274e-01 -8.09892938e-02
-2.18865171e-01 3.13986361e-01 1.11411130e+00 1.95150882e-01
-2.99340010e-01 4.00471002e-01 -9.27330315e-01 1.29346573e+00
4.07241881e-01 -7.39883304e-01 -5.05460501e-01 6.65924251e-02
3.90549362e-01 7.37393200e-01 5.50330460e-01 -4.22414780e-01
2.14480549e-01 -5.00948548e-01 5.47887444e-01 -4.93872017e-01
3.94064456e-01 -9.34446216e-01 5.14985442e-01 2.17600316e-01
-5.53814292e-01 -3.19330990e-02 5.99905178e-02 6.36931837e-01
-1.62695765e-01 -8.81871998e-01 6.95976079e-01 -9.69929025e-02
-2.81309366e-01 -4.53941524e-02 -7.46869624e-01 -7.84184113e-02
8.54804099e-01 -3.73575449e-01 3.09381243e-02 -9.13971484e-01
-1.20161533e+00 2.58279324e-01 5.17638922e-01 -2.39894167e-01
3.51942629e-01 -1.00030565e+00 -7.78253794e-01 -4.70759161e-02
3.39909941e-02 -1.48369178e-01 -1.39585048e-01 7.36235023e-01
-1.66981995e-01 3.46738100e-01 3.56448919e-01 -4.58311796e-01
-8.01506042e-01 3.93503755e-01 5.60537994e-01 -3.29920202e-01
7.71711469e-02 7.40553916e-01 2.15770692e-01 -6.06048048e-01
2.81009406e-01 -4.58949089e-01 -5.26507683e-02 1.39737830e-01
1.56818360e-01 3.97761106e-01 -1.72286257e-01 -2.86334217e-01
-2.99945295e-01 3.70025873e-01 2.77802814e-02 -5.07550359e-01
1.02270639e+00 -3.34233314e-01 -1.71354026e-01 7.57557809e-01
9.50699866e-01 2.34317929e-01 -1.48021913e+00 -2.39951074e-01
1.45139053e-01 -2.63984382e-01 2.74825662e-01 -1.11153889e+00
-5.30945957e-01 3.37515146e-01 3.67362112e-01 4.39988583e-01
8.71745944e-01 1.72340900e-01 -2.24748254e-02 4.27484125e-01
4.03679639e-01 -9.25073147e-01 -8.85714367e-02 7.83131272e-02
9.84590888e-01 -1.37211788e+00 5.07290125e-01 -3.36416423e-01
-6.68440580e-01 7.97295094e-01 1.54459894e-01 -1.47661656e-01
1.06658185e+00 4.08717096e-01 3.00984889e-01 -2.50404865e-01
-6.70399487e-01 1.93210412e-02 7.05976486e-01 5.88216841e-01
3.56879383e-01 1.38265863e-02 -2.23392159e-01 4.64865088e-01
-8.71159434e-02 2.92582717e-02 4.22927976e-01 8.25708628e-01
-5.79837680e-01 -1.13172185e+00 -7.41023123e-01 5.18683136e-01
-6.32399261e-01 -1.59262773e-02 -2.01790303e-01 7.46888995e-01
1.80187244e-02 1.10185921e+00 1.07920669e-01 2.90631294e-01
2.85591751e-01 -1.84373427e-02 3.87205303e-01 -9.96311307e-01
-2.78823704e-01 9.92198810e-02 2.92340279e-01 -2.56580412e-01
-4.38167363e-01 -9.28579748e-01 -1.00375772e+00 -4.51698042e-02
-6.42151117e-01 5.06405950e-01 5.16521096e-01 1.22599578e+00
-2.87012607e-01 2.92964220e-01 4.56982195e-01 -8.18829775e-01
-9.40434813e-01 -8.05730164e-01 -8.62205029e-01 9.19577479e-02
3.76231074e-01 -7.03647614e-01 -9.05752659e-01 1.05873048e-01] | [6.497837543487549, 3.96376633644104] |
26b74409-548d-447b-bbca-9db2c947ebcd | simvp-simpler-yet-better-video-prediction-1 | 2206.05099 | null | https://arxiv.org/abs/2206.05099v1 | https://arxiv.org/pdf/2206.05099v1.pdf | SimVP: Simpler yet Better Video Prediction | From CNN, RNN, to ViT, we have witnessed remarkable advancements in video prediction, incorporating auxiliary inputs, elaborate neural architectures, and sophisticated training strategies. We admire these progresses but are confused about the necessity: is there a simple method that can perform comparably well? This paper proposes SimVP, a simple video prediction model that is completely built upon CNN and trained by MSE loss in an end-to-end fashion. Without introducing any additional tricks and complicated strategies, we can achieve state-of-the-art performance on five benchmark datasets. Through extended experiments, we demonstrate that SimVP has strong generalization and extensibility on real-world datasets. The significant reduction of training cost makes it easier to scale to complex scenarios. We believe SimVP can serve as a solid baseline to stimulate the further development of video prediction. The code is available at \href{https://github.com/gaozhangyang/SimVP-Simpler-yet-Better-Video-Prediction}{Github}. | ['Stan Z. Li', 'Lirong Wu', 'Cheng Tan', 'Zhangyang Gao'] | 2022-06-09 | simvp-simpler-yet-better-video-prediction | http://openaccess.thecvf.com//content/CVPR2022/html/Gao_SimVP_Simpler_Yet_Better_Video_Prediction_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Gao_SimVP_Simpler_Yet_Better_Video_Prediction_CVPR_2022_paper.pdf | cvpr-2022-1 | ['video-prediction'] | ['computer-vision'] | [-6.89903274e-03 -2.09399927e-02 -4.72273618e-01 -4.74834859e-01
-6.36760414e-01 -1.82692498e-01 2.39120841e-01 -5.10633588e-01
-1.33305922e-01 5.84803283e-01 2.15872779e-01 -5.14601052e-01
2.90838242e-01 -5.99391639e-01 -8.59749913e-01 -6.81225061e-01
-6.38643727e-02 -3.91975529e-02 2.27601066e-01 -2.83370107e-01
-8.29006433e-02 -2.58268006e-02 -1.35421956e+00 3.91577005e-01
5.08246660e-01 1.17305875e+00 3.59846473e-01 5.09139538e-01
3.86220396e-01 1.30030954e+00 3.66701484e-02 -7.78937578e-01
4.65357065e-01 -3.80005062e-01 -7.63794422e-01 -1.49414405e-01
2.09200740e-01 -4.76065665e-01 -9.14620519e-01 7.57333159e-01
5.28203785e-01 8.83845985e-02 2.72511244e-01 -1.23238814e+00
-7.44858086e-01 4.48999077e-01 -4.12833244e-01 2.62727708e-01
8.99760053e-02 2.48763084e-01 9.86693859e-01 -9.14680362e-01
3.55730951e-01 8.49254668e-01 8.85650754e-01 8.29705358e-01
-8.01778913e-01 -7.57439911e-01 2.59733677e-01 6.11814082e-01
-1.39113700e+00 -5.41537285e-01 5.18822491e-01 -2.18380004e-01
9.28914368e-01 2.35555828e-01 5.31703413e-01 1.54397571e+00
2.42273416e-02 1.23197877e+00 5.36465347e-01 -9.87702757e-02
-2.47535303e-01 5.11482619e-02 -1.87074736e-01 8.76989841e-01
-2.04105556e-01 1.15973793e-01 -3.92397314e-01 2.93011516e-01
7.97581136e-01 2.67219603e-01 -7.82316029e-01 -5.54887205e-02
-1.18929017e+00 7.24609494e-01 5.07288575e-01 1.32921681e-01
-3.73694211e-01 3.24591607e-01 5.44051647e-01 3.36566567e-01
4.67411608e-01 1.18740819e-01 -5.99580348e-01 -5.85407734e-01
-1.00474060e+00 2.82989085e-01 5.86090207e-01 1.05940807e+00
3.70768487e-01 2.46560633e-01 -1.41268641e-01 8.45513046e-01
7.73698315e-02 2.37312526e-01 3.60094965e-01 -1.30862510e+00
6.48502648e-01 1.56604305e-01 -1.79333668e-02 -9.57557082e-01
-9.93247852e-02 -5.19680023e-01 -1.13688576e+00 -1.86930999e-01
1.98545247e-01 -4.05922860e-01 -7.87499070e-01 1.81642473e+00
-1.59807242e-02 5.14797151e-01 -1.90163553e-01 1.10019219e+00
1.05443168e+00 1.08636940e+00 3.38558480e-02 -5.84782511e-02
8.71040761e-01 -1.58378625e+00 -2.63418287e-01 3.81038152e-02
6.19245291e-01 -5.54641306e-01 1.02921939e+00 5.64861357e-01
-1.21529830e+00 -7.33563244e-01 -9.28980708e-01 -1.47927910e-01
1.06130645e-01 2.37692595e-01 6.83475733e-01 3.07298452e-01
-1.24711311e+00 9.22160327e-01 -1.02756512e+00 -3.48235667e-01
6.20648026e-01 3.98720831e-01 -3.75798464e-01 -1.39137238e-01
-1.10987866e+00 7.61259258e-01 2.56243974e-01 2.04384819e-01
-1.01773334e+00 -6.12968683e-01 -7.45707691e-01 1.76323280e-01
6.41197562e-01 -7.97574162e-01 1.38773048e+00 -1.20753253e+00
-1.55847740e+00 5.25524855e-01 -3.35082591e-01 -5.16917884e-01
6.46097124e-01 -3.83453429e-01 -3.58696043e-01 1.82161983e-02
-2.05992147e-01 7.46771097e-01 6.47645354e-01 -9.66423333e-01
-5.47940135e-01 1.41731948e-01 -2.46008243e-02 3.86200882e-02
-3.89903665e-01 1.05115540e-01 -1.05957198e+00 -8.80414426e-01
-2.21711472e-01 -9.91169095e-01 -2.85408646e-01 1.02252260e-01
-2.80595332e-01 -2.57911474e-01 6.17165625e-01 -8.21966946e-01
1.42346680e+00 -2.18063450e+00 2.27444902e-01 -2.36276299e-01
1.98165923e-01 7.52297938e-01 -2.65846878e-01 4.60256606e-01
-7.57427141e-02 2.26950541e-01 -4.74254042e-02 -4.49688524e-01
-3.64635140e-02 6.67008236e-02 -3.08303297e-01 2.31308609e-01
2.46504143e-01 1.12207174e+00 -7.31971860e-01 -2.64617592e-01
1.57019988e-01 8.50615144e-01 -6.42070830e-01 3.48300546e-01
-1.53295800e-01 4.64222461e-01 -3.30576420e-01 7.35318720e-01
4.50601548e-01 -6.75680757e-01 5.04281893e-02 -1.87117159e-01
1.22233950e-01 4.08593088e-01 -7.37722278e-01 1.49798501e+00
-1.24564998e-01 8.37586462e-01 -9.21439752e-02 -1.31782532e+00
6.93919659e-01 5.17427266e-01 3.69198501e-01 -6.75041676e-01
3.62789959e-01 2.90336963e-02 -1.71042174e-01 -6.25307739e-01
3.47469538e-01 6.37226477e-02 3.27978015e-01 8.12359825e-02
1.96399480e-01 4.25767660e-01 2.05826744e-01 7.40553364e-02
1.02018332e+00 4.62356210e-01 1.77978486e-01 2.45794170e-02
3.70157957e-01 -1.35988191e-01 8.63357842e-01 5.76715589e-01
-3.63676041e-01 7.87053823e-01 4.21495616e-01 -6.69682860e-01
-1.21345198e+00 -9.65485871e-01 2.15140283e-01 1.15760541e+00
5.96761331e-02 -6.57740057e-01 -4.60371822e-01 -5.79934835e-01
-2.35693648e-01 3.88678759e-01 -5.29538155e-01 -3.45400791e-03
-5.82521975e-01 -5.20850599e-01 5.90992451e-01 7.95750439e-01
6.67831838e-01 -8.77086282e-01 -2.27847874e-01 1.20255291e-01
-4.58987355e-01 -1.28558433e+00 -3.93203855e-01 -8.75225365e-02
-9.26336944e-01 -7.52261639e-01 -1.03421223e+00 -8.71147871e-01
3.34051281e-01 5.19577801e-01 1.23609221e+00 4.26778466e-01
1.31358758e-01 8.88970569e-02 -5.67154706e-01 -2.46201426e-01
-2.05957875e-01 9.53860581e-02 -5.21684997e-02 -2.78066128e-01
3.04101646e-01 -7.18441606e-01 -8.13712239e-01 4.86699671e-01
-5.83110034e-01 4.95651662e-01 4.86049950e-01 8.59263599e-01
5.29791355e-01 -1.21510610e-01 3.56080353e-01 -6.62995219e-01
1.07113943e-01 -7.19092906e-01 -3.84224147e-01 1.78860351e-01
-4.38704371e-01 -3.10175747e-01 7.92120755e-01 -3.67160439e-01
-8.93013954e-01 -1.52808785e-01 -6.22332275e-01 -7.22372591e-01
-1.19198009e-01 6.83998883e-01 9.85263009e-03 1.08848289e-02
3.71068120e-01 5.05751491e-01 8.61010626e-02 -5.31172574e-01
1.32790223e-01 5.92356265e-01 4.80008870e-01 -2.86459953e-01
5.77068210e-01 3.33297312e-01 -2.56631166e-01 -6.90322816e-01
-8.42713654e-01 -2.62572199e-01 -4.11971182e-01 -1.91953659e-01
6.87962949e-01 -1.43790174e+00 -6.84249282e-01 4.87112790e-01
-8.97558987e-01 -7.91848063e-01 1.30826101e-01 4.14276689e-01
-5.94907999e-01 5.88738441e-01 -1.13578165e+00 -4.34631974e-01
-4.97505128e-01 -1.06778014e+00 5.83659530e-01 2.39797011e-01
4.01138552e-02 -8.79842639e-01 -2.27462336e-01 6.88837349e-01
6.03434682e-01 8.83203000e-03 5.30778468e-01 -3.55383068e-01
-8.01332057e-01 -1.27720118e-01 -2.30440542e-01 7.94447839e-01
-1.41184792e-01 2.48831198e-01 -8.43988359e-01 -3.82699460e-01
-2.62128413e-01 -5.10798275e-01 1.14190042e+00 2.98763126e-01
1.79085076e+00 -3.45037520e-01 -2.39531785e-01 1.02439213e+00
1.51663375e+00 2.34213054e-01 9.08454120e-01 2.80912578e-01
7.31235743e-01 9.01108161e-02 4.56211060e-01 4.55110371e-01
7.72631466e-01 7.40563393e-01 4.19210255e-01 -1.98994547e-01
-5.33498451e-02 -3.52483630e-01 5.47489107e-01 9.90554392e-01
-6.99055970e-01 -6.21630669e-01 -7.41700947e-01 4.91350502e-01
-2.02621508e+00 -1.27410936e+00 -4.57829051e-02 1.96254861e+00
5.98962843e-01 1.21865287e-01 3.29086244e-01 -1.08080700e-01
6.05476499e-01 3.48340601e-01 -3.18117201e-01 -1.85570315e-01
-1.77096203e-02 3.82266082e-02 3.46510828e-01 1.37636170e-01
-1.34717846e+00 1.05506015e+00 5.96080494e+00 9.02999163e-01
-1.55459762e+00 2.37400174e-01 1.01977813e+00 -2.61786163e-01
-1.72391366e-02 -2.53026128e-01 -7.19123900e-01 6.76183760e-01
1.05871046e+00 1.15549102e-01 4.66884553e-01 9.40659583e-01
2.64472365e-01 3.18794996e-01 -1.01852334e+00 1.23462868e+00
3.07041705e-02 -1.68203902e+00 3.82811502e-02 -1.83086932e-01
6.31574214e-01 6.00012183e-01 1.79340944e-01 6.68946683e-01
-5.48369847e-02 -1.03787076e+00 6.15176499e-01 3.94381911e-01
7.48606920e-01 -4.18301702e-01 8.87474060e-01 2.81781912e-01
-1.24177349e+00 -2.26547956e-01 -5.47250271e-01 -3.04524034e-01
3.28594506e-01 1.09986447e-01 -4.74459887e-01 5.86238086e-01
8.50925446e-01 1.16680658e+00 -3.51002902e-01 1.26386046e+00
-2.21829832e-01 1.04344893e+00 -8.71502310e-02 -4.66011688e-02
3.07520807e-01 -2.96417996e-02 1.71190977e-01 1.32827139e+00
4.74600881e-01 3.73647511e-01 1.03244938e-01 5.04483283e-01
-2.75494933e-01 9.66398492e-02 -4.99939770e-01 -1.02132440e-01
3.64329606e-01 9.66505349e-01 -4.26980495e-01 -4.31162566e-01
-8.41610312e-01 1.01782668e+00 4.18785661e-01 5.00898659e-01
-1.27394307e+00 -7.49598965e-02 6.88002229e-01 1.58039764e-01
6.66054606e-01 -1.44575268e-01 -2.14203056e-02 -1.48848999e+00
2.67203182e-01 -1.05386233e+00 2.02896670e-01 -8.06394041e-01
-1.23591602e+00 8.17530811e-01 -2.85561711e-01 -1.41203821e+00
-1.76472217e-01 -6.76602066e-01 -7.46327281e-01 4.66227859e-01
-1.51353264e+00 -1.08109283e+00 -3.79556477e-01 7.02883542e-01
7.46787012e-01 -1.93313956e-01 6.44834697e-01 6.08659923e-01
-7.98000395e-01 9.81174946e-01 2.58822232e-01 4.24706221e-01
5.33177078e-01 -6.69648647e-01 4.33481753e-01 8.27719212e-01
3.78458225e-03 1.21764131e-01 5.91123164e-01 -2.19764858e-01
-1.37531006e+00 -1.26264679e+00 7.79891968e-01 -4.29201245e-01
6.46127403e-01 -2.29474843e-01 -8.85532618e-01 9.97872531e-01
3.87112051e-01 1.98651198e-02 6.90403283e-01 3.59252095e-02
-3.18833053e-01 -1.98590368e-01 -5.51462770e-01 7.18100369e-01
1.11297297e+00 -3.99833620e-01 -2.74215311e-01 2.87892610e-01
9.07586992e-01 -4.77526277e-01 -9.27083433e-01 6.55477285e-01
6.83274925e-01 -1.09684706e+00 9.40744102e-01 -6.47210002e-01
9.92202699e-01 -4.50067334e-02 -3.00709039e-01 -8.61485302e-01
-6.29990220e-01 -6.76562965e-01 -3.61309111e-01 9.50006604e-01
6.87941551e-01 -5.17518222e-01 1.04625309e+00 5.92997909e-01
-3.02532196e-01 -1.53318119e+00 -7.05235660e-01 -8.45569968e-01
1.33643463e-01 -6.69016719e-01 3.76112729e-01 9.05376554e-01
1.31510377e-01 1.87142476e-01 -1.04987180e+00 4.39512879e-02
3.76187563e-01 -7.36928359e-02 8.29690874e-01 -7.07913101e-01
-6.06312573e-01 -4.74847764e-01 -6.09270930e-01 -1.58530068e+00
-8.87624994e-02 -5.79307854e-01 -1.87312871e-01 -1.41575563e+00
4.38952744e-01 -2.01107025e-01 -5.06072164e-01 6.62527561e-01
-1.83942273e-01 5.71300268e-01 7.01816440e-01 1.92811027e-01
-1.01414204e+00 8.12091887e-01 1.19482362e+00 -4.73078080e-02
1.18959337e-01 9.24230069e-02 -8.35028112e-01 6.51574194e-01
8.85576069e-01 -1.20076500e-01 -3.28936428e-01 -7.58783996e-01
3.76478769e-02 3.08055520e-01 3.39363277e-01 -1.17406976e+00
1.24063537e-01 4.12498526e-02 3.96106929e-01 -3.99159133e-01
6.05192482e-01 -5.64485252e-01 2.44137749e-01 3.61786187e-01
-2.35120445e-01 1.61267892e-01 1.72796041e-01 3.00823659e-01
-3.77413481e-01 3.14866602e-02 6.54034317e-01 -2.88646042e-01
-9.81471062e-01 8.29390705e-01 -2.04892173e-01 7.63381971e-03
1.03831553e+00 -1.95109636e-01 -3.49565268e-01 -7.82917738e-01
-8.52141201e-01 3.45721871e-01 5.42128921e-01 5.70623577e-01
6.37164474e-01 -1.24534869e+00 -9.13583696e-01 -9.02605802e-02
-7.53295496e-02 -2.69547850e-01 4.35112387e-01 9.71083760e-01
-6.42069101e-01 5.35534739e-01 -2.12967157e-01 -5.85487604e-01
-1.40845013e+00 6.51157737e-01 2.98190176e-01 -4.69269715e-02
-6.57022715e-01 1.22692168e+00 1.59612551e-01 -2.76151121e-01
5.08286119e-01 -7.09725842e-02 1.57174900e-01 -4.37605143e-01
6.31931663e-01 1.97459325e-01 -2.86664307e-01 -5.77369213e-01
-1.61287248e-01 3.15568864e-01 -4.19669971e-02 5.40480793e-01
1.69177103e+00 -1.83862895e-01 3.11579645e-01 1.35537311e-01
1.29915142e+00 -5.08532166e-01 -1.61558414e+00 -2.30028063e-01
-5.26293933e-01 -5.59139431e-01 -9.02978033e-02 -6.20336592e-01
-1.52867448e+00 1.02413130e+00 5.10169864e-01 3.44786681e-02
1.38594627e+00 -3.53837907e-02 9.70818400e-01 3.85747075e-01
4.22933400e-01 -7.61580348e-01 1.67292804e-02 7.02860415e-01
8.24980557e-01 -1.50908649e+00 -1.15796342e-01 -3.14540446e-01
-9.33122516e-01 9.56954837e-01 7.46770024e-01 -2.10216314e-01
8.27734590e-01 7.13493749e-02 7.60546625e-02 1.12954780e-01
-1.28855729e+00 -5.36609851e-02 1.84871733e-01 3.95941645e-01
6.90479755e-01 1.79048646e-02 -4.16030437e-02 5.44404864e-01
-9.21475068e-02 3.55647415e-01 3.48034233e-01 6.04164600e-01
-2.58660585e-01 -1.09241366e+00 1.27916411e-01 6.17368519e-01
-6.74112976e-01 -4.32556033e-01 1.40399009e-01 9.07528996e-01
4.49729450e-02 7.92171001e-01 -1.01183288e-01 -9.22277629e-01
1.17750347e-01 -1.14659324e-01 4.01338726e-01 -2.67395228e-01
-3.02146494e-01 -5.74985184e-02 1.50838450e-01 -8.24669003e-01
-4.37933356e-01 -4.80205238e-01 -9.05086279e-01 -8.84338677e-01
-1.09652646e-01 -2.83621397e-04 2.52676636e-01 8.12105417e-01
6.61685586e-01 2.23381713e-01 5.25646508e-01 -1.10701084e+00
-3.67143154e-01 -8.97880971e-01 -1.82021171e-01 2.35506549e-01
2.87467957e-01 -4.39246416e-01 -1.01624883e-03 1.24997884e-01] | [9.084601402282715, 0.43628621101379395] |
184fe19b-a748-4ac0-870d-d9ef518ec92a | scenegenie-scene-graph-guided-diffusion | 2304.14573 | null | https://arxiv.org/abs/2304.14573v1 | https://arxiv.org/pdf/2304.14573v1.pdf | SceneGenie: Scene Graph Guided Diffusion Models for Image Synthesis | Text-conditioned image generation has made significant progress in recent years with generative adversarial networks and more recently, diffusion models. While diffusion models conditioned on text prompts have produced impressive and high-quality images, accurately representing complex text prompts such as the number of instances of a specific object remains challenging. To address this limitation, we propose a novel guidance approach for the sampling process in the diffusion model that leverages bounding box and segmentation map information at inference time without additional training data. Through a novel loss in the sampling process, our approach guides the model with semantic features from CLIP embeddings and enforces geometric constraints, leading to high-resolution images that accurately represent the scene. To obtain bounding box and segmentation map information, we structure the text prompt as a scene graph and enrich the nodes with CLIP embeddings. Our proposed model achieves state-of-the-art performance on two public benchmarks for image generation from scene graphs, surpassing both scene graph to image and text-based diffusion models in various metrics. Our results demonstrate the effectiveness of incorporating bounding box and segmentation map guidance in the diffusion model sampling process for more accurate text-to-image generation. | ['Nassir Navab', 'Björn Ommer', 'Chengzhi Shen', 'Yu Chi', 'Yousef Yeganeh', 'Azade Farshad'] | 2023-04-28 | null | null | null | null | ['image-generation-from-scene-graphs'] | ['computer-vision'] | [ 5.92629015e-01 2.07535610e-01 -1.28678503e-02 -5.51212072e-01
-6.77987456e-01 -6.38457358e-01 1.07746017e+00 6.97813109e-02
-2.21302658e-01 4.29268479e-01 4.12841082e-01 -8.96344483e-02
1.62144601e-01 -1.29275560e+00 -1.04759765e+00 -4.83491331e-01
3.17607164e-01 6.85284972e-01 2.68975019e-01 -1.66715264e-01
1.30575716e-01 5.59401631e-01 -1.12051511e+00 1.75792307e-01
1.03728676e+00 1.03767955e+00 1.93586096e-01 9.27605569e-01
-4.06023979e-01 9.97116625e-01 -6.99418783e-01 -5.59806049e-01
3.93690675e-01 -2.44077116e-01 -5.00178277e-01 3.56591552e-01
8.86703551e-01 -8.27369571e-01 -8.83733392e-01 9.79896605e-01
4.68860894e-01 1.11783318e-01 9.24541712e-01 -1.43436134e+00
-1.18644953e+00 6.43790603e-01 -6.42364562e-01 -1.32151380e-01
3.73810798e-01 3.99201453e-01 9.84500647e-01 -8.87988269e-01
1.05641568e+00 1.33566964e+00 3.45369667e-01 5.14065266e-01
-1.51804245e+00 -7.27909088e-01 4.09500748e-01 -7.19732344e-02
-1.16945493e+00 -1.77009478e-01 1.04730344e+00 -5.82247913e-01
8.15416574e-01 1.10315703e-01 7.31702328e-01 1.33534849e+00
5.89102879e-02 9.62961376e-01 8.63620996e-01 -7.77258426e-02
3.30443203e-01 -4.13535722e-02 -4.90776718e-01 8.40019345e-01
1.02161668e-01 1.20955139e-01 -4.38632190e-01 6.04800694e-02
1.22131371e+00 9.15739760e-02 -6.01450764e-02 -5.31815231e-01
-1.10472786e+00 1.06997502e+00 7.89035022e-01 -2.78400689e-01
-5.13481021e-01 5.20121396e-01 8.30134675e-02 -2.29439944e-01
6.97825909e-01 4.76781458e-01 7.37760812e-02 7.43743107e-02
-1.31815207e+00 7.21722841e-01 7.73752153e-01 1.35772681e+00
7.24089921e-01 3.47893953e-01 -7.59285510e-01 6.56921506e-01
4.33396585e-02 9.37453389e-01 -4.60182950e-02 -9.84921694e-01
4.88821983e-01 6.42574489e-01 6.29215734e-03 -1.27919686e+00
4.37959097e-02 -2.27839217e-01 -9.83026326e-01 1.77667931e-01
3.31126958e-01 -7.29415491e-02 -1.43679762e+00 1.57631373e+00
3.80375475e-01 2.80493647e-01 -2.24629194e-01 8.74280989e-01
7.48066008e-01 9.56272840e-01 1.90812156e-01 3.61771852e-01
1.07588601e+00 -1.17256558e+00 -6.32251382e-01 -3.61398816e-01
1.23337939e-01 -7.75775135e-01 9.86976147e-01 1.57156616e-01
-1.10446858e+00 -4.72521216e-01 -8.87316763e-01 -4.59588230e-01
-3.62947702e-01 -1.40184298e-01 7.32497931e-01 3.11806291e-01
-9.71354127e-01 3.54113251e-01 -6.85375392e-01 -1.95037082e-01
7.83486307e-01 -2.35758466e-03 -1.09268270e-01 -6.72748029e-01
-8.97391438e-01 3.61576945e-01 8.91983956e-02 -3.12324554e-01
-1.29351652e+00 -1.06529975e+00 -1.06953704e+00 -1.52669018e-02
3.81571144e-01 -1.05824900e+00 1.00552487e+00 -5.05380154e-01
-1.36554646e+00 6.23864770e-01 -2.27193665e-02 -6.43438697e-01
9.07959402e-01 -1.94402069e-01 3.38949710e-02 3.83109838e-01
2.98785776e-01 1.33206832e+00 1.27386010e+00 -1.41204762e+00
-5.39370894e-01 -1.17749922e-01 3.57320130e-01 1.34350404e-01
-2.58525670e-01 -3.84025842e-01 -7.95361161e-01 -1.03041303e+00
-1.57549173e-01 -8.03150058e-01 -5.38374126e-01 4.96107757e-01
-6.83873951e-01 -3.23060118e-02 1.17468059e+00 -5.79011321e-01
1.01451826e+00 -1.92367220e+00 2.09835753e-01 -1.02113903e-01
5.52880883e-01 -4.53961007e-02 -3.67876559e-01 4.15418297e-01
3.10994983e-01 2.81150609e-01 -3.06156218e-01 -7.51278162e-01
3.05299491e-01 4.21434134e-01 -7.73416877e-01 2.13803723e-01
5.51524341e-01 1.34262669e+00 -1.14071882e+00 -6.26668572e-01
6.25795007e-01 7.67303050e-01 -9.54786897e-01 4.81819719e-01
-7.66141355e-01 3.07240754e-01 -5.23558736e-01 5.42839229e-01
7.09492266e-01 -3.49634945e-01 -3.69165510e-01 -3.43687415e-01
3.03096890e-01 -5.45297451e-02 -7.91773021e-01 1.97201145e+00
-3.98535639e-01 6.66067123e-01 -6.20971546e-02 -6.49710894e-01
8.65717947e-01 -5.34903035e-02 5.03538728e-01 -7.70586491e-01
-1.59595385e-01 -2.42823794e-01 -5.12691557e-01 -2.33264625e-01
9.37644839e-01 2.26829633e-01 -1.98236138e-01 3.67783040e-01
3.35130766e-02 -1.11404634e+00 3.89547348e-01 9.30398464e-01
1.02153945e+00 3.92603129e-01 -2.42065996e-01 8.73863623e-02
-5.10467999e-02 1.91163719e-01 2.32197508e-01 9.28555965e-01
1.58785522e-01 9.80459094e-01 5.30459285e-01 -3.13919276e-01
-1.49921954e+00 -1.31986547e+00 6.27590120e-02 7.71111310e-01
1.47657424e-01 -4.80959326e-01 -9.65637565e-01 -7.65729249e-01
1.30214840e-01 8.24086368e-01 -8.79877567e-01 -5.22486754e-02
-4.82645035e-01 -4.35909629e-01 5.16440153e-01 6.99075341e-01
5.22346973e-01 -8.82272542e-01 -2.33363837e-01 2.31619507e-01
-8.99239555e-02 -1.56549382e+00 -9.38288331e-01 -2.09126577e-01
-5.60883284e-01 -9.29953992e-01 -1.01999247e+00 -5.96761882e-01
1.07320857e+00 1.15354255e-01 1.22989881e+00 -2.17732519e-01
-6.50788486e-01 5.75639129e-01 -3.82652670e-01 -3.47479075e-01
-4.21305150e-01 9.53205861e-03 -4.49278921e-01 9.03685987e-02
-1.66285515e-01 -4.76271540e-01 -8.36966276e-01 1.13920018e-01
-1.29233909e+00 4.79829848e-01 5.61528683e-01 7.73919046e-01
8.16507876e-01 -8.12576562e-02 3.10191244e-01 -1.06930363e+00
6.04811966e-01 -4.55044538e-01 -6.68252349e-01 -1.33252725e-01
-5.91054380e-01 5.53235039e-02 8.35550964e-01 -5.16734660e-01
-1.00699568e+00 4.18694131e-02 -5.14254645e-02 -7.61881888e-01
-5.13039678e-02 1.60111785e-01 9.97560248e-02 -6.92511303e-03
6.96617544e-01 4.49834436e-01 -2.32713759e-01 -7.18450695e-02
9.60097492e-01 2.16244385e-01 7.59684205e-01 -7.97107935e-01
1.19477475e+00 8.37916434e-01 1.29350545e-02 -7.15674877e-01
-9.61193502e-01 -1.91131845e-01 -4.19953346e-01 -2.04692170e-01
1.04896712e+00 -9.16861236e-01 -2.14169309e-01 5.11850834e-01
-1.31514001e+00 -6.68471217e-01 -6.50665939e-01 -4.21562605e-02
-6.40077174e-01 1.76546678e-01 -7.02898204e-01 -4.17117625e-01
-3.60930622e-01 -1.18754315e+00 1.56646085e+00 9.49595198e-02
2.15654285e-03 -1.13480639e+00 -2.94094026e-01 3.09023827e-01
5.79473853e-01 6.97728097e-01 8.62808347e-01 -1.72164619e-01
-1.30138421e+00 -4.10504371e-01 -5.49982309e-01 3.21368575e-01
-4.01762202e-02 4.89643440e-02 -7.91189015e-01 -1.23441234e-01
-3.85671318e-01 -3.71386856e-01 8.97418976e-01 5.42350888e-01
1.31380618e+00 -4.31656599e-01 -3.36718172e-01 1.03328204e+00
1.50242329e+00 -1.70504183e-01 7.41278350e-01 -1.63696334e-02
1.13590169e+00 3.26369196e-01 4.61929530e-01 5.75008988e-01
5.85494637e-01 5.34161627e-01 6.57383204e-01 -4.13244188e-01
-6.79216504e-01 -9.80907977e-01 2.30999533e-02 4.64428961e-01
3.85548055e-01 -6.00263536e-01 -7.38806784e-01 6.83499634e-01
-1.60967731e+00 -8.02286744e-01 8.64301473e-02 1.78319633e+00
7.95925140e-01 1.71340585e-01 -2.07577586e-01 -2.10355327e-01
4.14046586e-01 6.56848669e-01 -7.93914676e-01 -8.21313262e-02
-2.19599083e-01 8.02421793e-02 6.26878142e-01 6.07457936e-01
-9.75768507e-01 1.40936363e+00 6.26820326e+00 1.01090789e+00
-1.14516687e+00 -6.96733668e-02 7.19140589e-01 -1.44354388e-01
-7.13756442e-01 -3.73239294e-02 -8.08863044e-01 3.07045370e-01
3.13337535e-01 -3.71760339e-01 6.06274068e-01 9.87108350e-01
2.13697121e-01 -3.11888587e-02 -1.20449936e+00 8.62613142e-01
2.14191332e-01 -1.74459410e+00 7.00800002e-01 1.36907652e-01
1.28941858e+00 -8.48771632e-02 3.67442757e-01 2.34854221e-01
8.44340622e-01 -1.19903302e+00 9.66451287e-01 4.64241862e-01
1.19673049e+00 -5.11176348e-01 9.14971232e-02 2.35729709e-01
-1.22142291e+00 1.91850692e-01 -4.19658780e-01 2.57823437e-01
5.08878887e-01 7.50455797e-01 -1.04994106e+00 4.43977565e-01
2.65335411e-01 8.18503141e-01 -5.48586786e-01 6.91613257e-01
-3.65351111e-01 4.73804146e-01 -2.33709395e-01 1.38385827e-02
5.02906084e-01 -6.65117279e-02 5.89788675e-01 1.19619298e+00
2.28333324e-01 -7.82403424e-02 4.79097813e-01 1.54602885e+00
-4.13136989e-01 -6.50047138e-02 -8.09724629e-01 -3.47834617e-01
5.60996592e-01 1.25151110e+00 -6.87029481e-01 -4.86376733e-01
-1.45162880e-01 1.30935311e+00 2.92506993e-01 6.84330642e-01
-1.10410142e+00 -3.52091998e-01 5.70952952e-01 4.29914683e-01
3.95177543e-01 -4.39711452e-01 -3.72129172e-01 -9.37056422e-01
-7.46852681e-02 -6.19221449e-01 -1.08507492e-01 -1.03252196e+00
-1.17836416e+00 6.34230673e-01 4.05847281e-02 -7.59333372e-01
-2.42681131e-01 -4.18040961e-01 -4.74651873e-01 8.44030201e-01
-1.57139003e+00 -1.54524016e+00 -6.91786170e-01 6.52814150e-01
8.34155977e-01 1.30788416e-01 6.33184791e-01 1.54500216e-01
-3.20756972e-01 5.13318658e-01 -1.71339571e-01 2.82787263e-01
6.51495457e-01 -1.44290245e+00 9.42967653e-01 8.78618240e-01
2.34596446e-01 3.62682402e-01 5.82965791e-01 -8.13481867e-01
-1.40016043e+00 -1.56330466e+00 3.17122698e-01 -5.58096528e-01
5.14476061e-01 -6.70730829e-01 -4.74964142e-01 6.13162577e-01
1.71584293e-01 1.40714467e-01 6.39550760e-03 -4.35261905e-01
-4.80617613e-01 3.24289501e-02 -1.02728570e+00 9.70006227e-01
1.17889905e+00 -5.86502254e-01 -5.38027147e-03 5.13109028e-01
1.18499362e+00 -7.84883320e-01 -8.10669184e-01 2.13996887e-01
2.20514804e-01 -7.47320235e-01 1.16930699e+00 -3.06927770e-01
1.00734508e+00 -1.05762549e-01 -6.01410866e-02 -1.39038932e+00
-2.47143313e-01 -7.01198459e-01 -2.42650211e-01 1.07567251e+00
2.07423061e-01 -1.57083198e-01 1.03860855e+00 6.89123094e-01
-1.31625757e-01 -8.88813853e-01 -4.36687857e-01 -4.93190646e-01
7.13537335e-02 -5.74385822e-01 6.84995532e-01 7.70756364e-01
-7.08894551e-01 1.68314248e-01 -4.45830435e-01 3.10216891e-03
8.71172965e-01 -3.82031314e-02 1.21241868e+00 -7.54445195e-01
-1.29745528e-01 -4.16280210e-01 -3.02097410e-01 -1.67741477e+00
1.58315599e-01 -9.38538730e-01 8.92982818e-03 -2.07419300e+00
1.32510647e-01 -5.75120747e-01 3.39891881e-01 1.58555284e-01
-2.22403586e-01 2.81802505e-01 3.03717643e-01 -4.24379855e-02
-4.83470678e-01 7.02618539e-01 1.84395909e+00 -5.46037316e-01
5.18230870e-02 -5.67003191e-01 -6.00892425e-01 5.00205636e-01
3.73903096e-01 -4.37079936e-01 -6.40006244e-01 -8.65803182e-01
1.23199068e-01 -2.45043896e-02 6.57620132e-01 -9.66997623e-01
5.11713684e-01 -2.28752732e-01 4.96794105e-01 -6.09585643e-01
6.36910498e-01 -6.86463773e-01 -7.55372792e-02 5.92445619e-02
-4.53107744e-01 -2.30793655e-01 1.29629552e-01 9.75794315e-01
-3.63643980e-03 2.68354505e-01 6.70416713e-01 -2.31060144e-02
-6.18457675e-01 8.90771031e-01 6.94514289e-02 3.28818738e-01
1.06341434e+00 -2.42396086e-01 -4.42438245e-01 -7.46433675e-01
-4.18610245e-01 2.87295043e-01 7.63059676e-01 7.23342061e-01
8.77924681e-01 -1.37278914e+00 -8.14763844e-01 3.56698781e-01
1.20453313e-01 5.64011276e-01 3.83265138e-01 3.41011882e-01
-7.47175753e-01 2.19850004e-01 7.35397115e-02 -7.09911168e-01
-8.13211977e-01 5.70886254e-01 1.45901516e-01 -2.45731130e-01
-7.75597394e-01 1.01661897e+00 6.55437708e-01 -2.10154384e-01
2.76153356e-01 -4.24827039e-01 3.56128961e-01 -2.82459974e-01
3.84776413e-01 6.90102279e-02 -5.23559153e-01 -5.68115890e-01
2.13703498e-01 5.41251659e-01 -2.93489277e-01 -2.69327879e-01
1.07537222e+00 4.28596176e-02 1.82763264e-01 -1.13779036e-02
1.10591125e+00 1.40050992e-01 -1.86896503e+00 -3.91289145e-01
-6.02652490e-01 -7.09455132e-01 2.89655715e-01 -8.51792037e-01
-1.21901357e+00 9.87345219e-01 9.28795785e-02 8.06301460e-02
7.98326254e-01 -5.59008941e-02 1.13403106e+00 2.62796674e-02
2.82586813e-01 -1.02288866e+00 6.00781739e-01 3.50119889e-01
1.10427725e+00 -1.13673377e+00 -1.05941318e-01 -6.17902696e-01
-6.62917852e-01 7.78628111e-01 6.15760386e-01 -3.83721441e-01
4.46090370e-01 5.17926037e-01 -7.15790316e-02 -2.38119915e-01
-6.67630494e-01 1.07782282e-01 3.88886571e-01 8.56196046e-01
6.88261539e-02 1.84718251e-01 2.12886974e-01 2.33967096e-01
-2.41598010e-01 -2.31593139e-02 4.71058667e-01 6.67092621e-01
-2.84815878e-01 -9.11518335e-01 -1.91139892e-01 5.83335459e-01
-2.35847786e-01 -3.86222482e-01 -3.92641753e-01 6.53450191e-01
-1.51133522e-01 7.85022497e-01 2.09129006e-01 -3.14949483e-01
2.96544760e-01 -3.82726997e-01 5.09582579e-01 -7.34286427e-01
-1.04026169e-01 5.17048687e-02 -1.48948520e-01 -7.33672917e-01
1.35986041e-02 -3.31158608e-01 -1.19985211e+00 -5.06971776e-01
-1.18488923e-01 -1.52546868e-01 7.76143610e-01 5.23211360e-01
4.82617378e-01 9.10530865e-01 4.66532350e-01 -1.17950106e+00
-4.28666294e-01 -8.40722263e-01 -4.50563610e-01 7.35379279e-01
1.28516465e-01 -4.16695505e-01 -2.28556499e-01 4.13052261e-01] | [11.30673885345459, -0.2583547532558441] |
c3ad034b-26e8-4511-88a6-acfbfde7bb3e | moving-avatars-and-agents-in-social-extended | 2306.14484 | null | https://arxiv.org/abs/2306.14484v1 | https://arxiv.org/pdf/2306.14484v1.pdf | Moving Avatars and Agents in Social Extended Reality Environments | Natural interaction between multiple users within a shared virtual environment (VE) relies on each other's awareness of the current position of the interaction partners. This, however, cannot be warranted when users employ noncontinuous locomotion techniques, such as teleportation, which may cause confusion among bystanders. In this paper, we pursue two approaches to create a pleasant experience for both the moving user and the bystanders observing that movement. First, we will introduce a Smart Avatar system that delivers continuous full-body human representations for noncontinuous locomotion in shared virtual reality (VR) spaces. Smart Avatars imitate their assigned user's real-world movements when close-by and autonomously navigate to their user when the distance between them exceeds a certain threshold, i.e., after the user teleports. As part of the Smart Avatar system, we implemented four avatar transition techniques and compared them to conventional avatar locomotion in a user study, revealing significant positive effects on the observer's spatial awareness, as well as pragmatic and hedonic quality scores. Second, we introduce the concept of Stuttered Locomotion, which can be applied to any continuous locomotion method. By converting a continuous movement into short-interval teleport steps, we provide the merits of non-continuous locomotion for the moving user while observers can easily keep track of their path. Thus, while the experience for observers is similarly positive as with continuous motion, a user study confirmed that Stuttered Locomotion can significantly reduce the occurrence of cybersickness symptoms for the moving user, making it an attractive choice for shared VEs. We will discuss the potential of Smart Avatars and Stuttered Locomotion for shared VR experiences, both when applied individually and in combination. | ['Frank Steinicke', 'Bernhard E. Riecke', 'Susanne Schmidt', 'Jann Philipp Freiwald'] | 2023-06-26 | null | null | null | null | ['navigate'] | ['reasoning'] | [-2.63080150e-01 3.78009439e-01 7.27637634e-02 8.79876986e-02
-3.01565558e-01 -7.05962360e-01 3.48167479e-01 1.05770670e-01
-5.31992197e-01 5.61320543e-01 2.49318123e-01 -1.73115190e-02
2.25423016e-02 -7.30420947e-01 -2.25731179e-01 -3.73281837e-01
-3.09657574e-01 4.34654802e-02 1.53716356e-01 -7.52562106e-01
2.04073071e-01 4.09752965e-01 -1.78347206e+00 -5.76169463e-03
7.83215821e-01 3.84240091e-01 4.08186525e-01 7.04159856e-01
3.24949861e-01 7.53858924e-01 -4.60752398e-01 -4.18150499e-02
-1.48032710e-01 -6.94746494e-01 -6.16399348e-01 -3.56873274e-02
-1.65448546e-01 -6.43640876e-01 -7.19565451e-02 2.90481359e-01
6.46935165e-01 9.08055425e-01 2.02561304e-01 -1.55834687e+00
-3.82247955e-01 1.96894392e-01 -1.96595922e-01 -1.42472401e-01
1.26243210e+00 2.15422317e-01 7.59491563e-01 -4.89953816e-01
1.01347196e+00 7.32169032e-01 7.33567357e-01 5.69270194e-01
-1.33567965e+00 -5.56179643e-01 1.05042793e-01 1.71972767e-01
-1.56001222e+00 -4.41538453e-01 5.82161903e-01 -4.40453649e-01
1.06803572e+00 7.70465851e-01 1.27987397e+00 1.37668872e+00
1.11852735e-01 1.98468834e-01 6.02773905e-01 -4.99762408e-02
6.37786627e-01 6.26362860e-01 -2.26124406e-01 7.13803545e-02
1.57913566e-01 2.43593100e-02 -5.06867528e-01 -2.62937009e-01
9.14258659e-01 -2.77843893e-01 -5.93639195e-01 -7.48652518e-01
-1.04629660e+00 6.52039111e-01 2.01804042e-01 6.50623858e-01
-7.34618247e-01 7.74197653e-02 4.67847317e-01 2.48736456e-01
3.31072882e-03 5.54839194e-01 -8.95214677e-02 -1.04695916e+00
-4.22417581e-01 2.97035903e-01 7.33694077e-01 5.43126345e-01
1.10152863e-01 -1.67091295e-01 -2.35953182e-02 4.47233617e-01
2.83911884e-01 1.95165247e-01 3.71316195e-01 -1.01743698e+00
-1.33656517e-01 4.15185124e-01 6.39881849e-01 -1.32760561e+00
-5.20478308e-01 8.47705975e-02 -1.35397643e-01 6.36963010e-01
1.26265258e-01 -2.11240575e-01 -1.64981201e-01 1.92050004e+00
6.85011864e-01 8.74225870e-02 4.78999726e-02 1.17030311e+00
9.09481525e-01 3.46176505e-01 3.43540609e-01 -4.43218023e-01
1.20169187e+00 -4.58609998e-01 -1.18981898e+00 -5.21452725e-02
1.17288780e+00 -3.91442537e-01 1.47595894e+00 4.04798180e-01
-9.09110427e-01 -2.71482438e-01 -1.06609440e+00 3.19396317e-01
-8.35295767e-02 -3.65448087e-01 4.77634549e-01 1.16552699e+00
-1.00935590e+00 8.57107401e-01 -1.04734206e+00 -9.23819363e-01
-2.81120896e-01 3.31810623e-01 -6.62562430e-01 5.39389729e-01
-1.07413828e+00 1.13895762e+00 -2.27614149e-01 1.12085804e-01
-3.16276819e-01 -3.30610842e-01 -9.47157204e-01 -8.77373740e-02
1.10879332e-01 -5.70407033e-01 1.04750621e+00 -8.16611409e-01
-1.77426684e+00 8.45489860e-01 2.76997894e-01 4.76940721e-02
8.81384373e-01 -1.94586024e-01 -5.86928070e-01 6.10624999e-02
3.08848768e-01 2.93110549e-01 2.24965334e-01 -1.44306231e+00
-4.14710604e-02 -1.86090216e-01 1.26817852e-01 6.85856700e-01
-2.26146162e-01 -1.56551763e-01 2.81413138e-01 3.49875689e-02
2.84836292e-01 -9.78233159e-01 -1.95242420e-01 2.81263798e-01
-5.60499318e-02 2.66640037e-01 9.51009989e-01 -3.34814221e-01
1.02868795e+00 -2.46668458e+00 8.11386555e-02 1.83133021e-01
2.57234842e-01 -1.71975140e-02 1.38081431e-01 9.68467832e-01
2.95143062e-03 7.17762411e-02 1.83800906e-01 -3.53485584e-01
-1.64455380e-02 1.44459739e-01 1.42717749e-01 7.22311378e-01
-5.98784506e-01 5.94387829e-01 -1.05921841e+00 -2.85104305e-01
5.34212410e-01 8.64941955e-01 -4.24280524e-01 2.64072418e-01
6.24852836e-01 8.72444153e-01 -5.50475856e-03 9.97129083e-02
5.68668783e-01 1.19860992e-01 4.06008303e-01 3.91440660e-01
-3.18933398e-01 1.44946247e-01 -1.32562923e+00 1.44175637e+00
-6.73189938e-01 6.63184464e-01 1.37948334e-01 -3.66668217e-02
8.40822935e-01 6.60678804e-01 5.61923087e-01 -1.01708245e+00
2.37650365e-01 1.34527162e-01 -3.09682842e-02 -8.87202680e-01
9.45454359e-01 -3.02048206e-01 -1.07578181e-01 5.19305468e-01
-6.09669745e-01 -8.22508112e-02 -3.40906650e-01 2.09352374e-01
1.10099661e+00 2.47164473e-01 4.60949093e-01 5.61895929e-02
-7.24756345e-02 -1.00401424e-01 2.67926157e-01 5.67915738e-01
-4.85186309e-01 5.58287203e-01 2.64543831e-01 3.93547900e-02
-6.74014449e-01 -1.13656211e+00 2.31766582e-01 1.09445632e+00
8.28233361e-01 -5.91121674e-01 -4.76162285e-01 2.49335580e-02
-2.12425649e-01 1.14833665e+00 -7.24136591e-01 -4.04063284e-01
-3.49489629e-01 6.22643754e-02 4.38407362e-01 5.07507861e-01
3.22569668e-01 -1.16483116e+00 -1.70936406e+00 2.14884147e-01
-6.51670694e-01 -8.08082223e-01 -2.19795927e-01 -2.34226346e-01
-5.59384525e-01 -5.29345036e-01 -5.35976410e-01 -2.91481614e-01
2.80764043e-01 5.59110701e-01 3.63817126e-01 1.66396394e-01
-9.27405134e-02 8.41190398e-01 -8.21783364e-01 2.73989856e-01
-1.00593843e-01 -5.65246999e-01 2.62750864e-01 -1.34193882e-01
-3.78037602e-01 -8.61319840e-01 -7.57411778e-01 5.11475384e-01
-2.45927736e-01 1.69442222e-01 -1.96782902e-01 3.67962867e-01
4.90907691e-02 -5.14505446e-01 4.23363894e-01 -3.22802573e-01
8.35850954e-01 -5.11114895e-01 3.80898625e-01 -3.26818198e-01
-7.66182989e-02 -8.09318066e-01 2.17043638e-01 -8.74208033e-01
-1.05625010e+00 -1.93388730e-01 -1.90825626e-01 -1.08389258e-01
-1.23598441e-01 3.87344748e-01 -2.14982286e-01 -2.54154146e-01
8.27764571e-01 -2.86831975e-01 8.29904377e-02 1.16231814e-01
3.20071638e-01 8.41197371e-01 3.01030695e-01 -1.78108551e-02
4.94375765e-01 5.84830225e-01 -5.22304654e-01 -1.24417162e+00
3.42005432e-01 -3.70865971e-01 -4.73303080e-01 -7.51505435e-01
9.95519400e-01 -6.50765181e-01 -1.41085792e+00 1.60906747e-01
-7.76132166e-01 -7.46010482e-01 -5.39542496e-01 8.26732218e-01
-8.33119214e-01 3.31946790e-01 -2.66445726e-01 -1.21121919e+00
1.83364257e-01 -9.35147464e-01 7.93630779e-01 3.00404459e-01
-1.33708310e+00 -9.59179521e-01 3.06132615e-01 3.35321993e-01
4.95765835e-01 8.68651390e-01 2.62860000e-01 -4.96094525e-01
-7.25696906e-02 -5.12106299e-01 3.85821372e-01 -3.70393127e-01
3.19864005e-01 -1.67509779e-01 -8.18276286e-01 -3.33176643e-01
3.73046473e-02 -2.34064966e-01 -3.28505397e-01 1.23044886e-01
3.56715806e-02 -7.82496333e-02 -4.30881411e-01 1.69055894e-01
1.07060051e+00 6.17904067e-01 1.09646022e+00 6.10945463e-01
4.84071314e-01 6.24852300e-01 1.09264255e+00 6.87792778e-01
3.47814441e-01 9.74989057e-01 5.27402163e-01 -1.48712143e-01
2.81267613e-01 -3.13125640e-01 3.12885851e-01 3.71259719e-01
-3.79667401e-01 -3.00579041e-01 -4.79903609e-01 4.87938523e-01
-1.67869413e+00 -1.02881896e+00 -5.90805888e-01 2.54711652e+00
-1.66074887e-01 -1.41134441e-01 1.42038211e-01 2.08641812e-01
5.79080820e-01 7.83544779e-02 -3.24589193e-01 -9.85895455e-01
2.13954583e-01 -1.45066440e-01 1.72974929e-01 4.83391255e-01
-3.63982469e-01 4.97294545e-01 5.73866463e+00 -2.60695238e-02
-9.35332656e-01 3.03687930e-01 -6.57668486e-02 -2.36860916e-01
-3.70846242e-01 2.43192673e-01 2.51545817e-01 1.78764388e-01
8.15217733e-01 -3.50218296e-01 2.53051817e-01 7.20054269e-01
9.64072227e-01 -7.23857462e-01 -9.92174566e-01 9.23526347e-01
-6.90147430e-02 -7.17963696e-01 -6.52591467e-01 3.09899330e-01
1.08872250e-01 -5.64055145e-01 -2.05337480e-02 1.32726520e-01
-2.80849457e-01 -9.64829743e-01 6.51682317e-01 2.35308588e-01
7.22492397e-01 -7.16149688e-01 5.09039342e-01 3.10379237e-01
-1.16689420e+00 2.38888860e-02 4.10867959e-01 -6.03367031e-01
7.80027986e-01 -4.92211729e-01 -5.63840568e-01 2.70160705e-01
6.69602036e-01 -4.88773510e-02 9.41764330e-04 1.07409966e+00
6.21188097e-02 4.88916077e-02 -3.09341490e-01 -2.72764027e-01
-7.52915516e-02 -3.24125022e-01 9.22696114e-01 5.24765730e-01
3.77410352e-01 4.90786999e-01 -2.43414745e-01 6.77016973e-01
6.49316728e-01 4.40743953e-01 -1.01211774e+00 6.33316711e-02
4.08350080e-01 9.78962302e-01 -1.08962977e+00 -6.13325182e-03
1.97819434e-02 1.47243488e+00 -9.27701369e-02 2.82018483e-01
-1.12014163e+00 -6.54928625e-01 7.86065757e-01 3.19390148e-01
5.40544353e-02 -3.09635252e-01 -3.12093139e-01 -7.68925428e-01
1.83272049e-01 -4.11459863e-01 -2.67185736e-02 -1.18832493e+00
-3.77423644e-01 5.62241137e-01 -4.87806499e-02 -1.67008436e+00
-2.87622869e-01 4.39642191e-01 -7.80942023e-01 5.34430623e-01
-2.82749772e-01 -8.99544358e-01 -8.31221163e-01 2.29968593e-01
2.39161670e-01 7.51335800e-01 9.99436498e-01 1.51650190e-01
-2.31266946e-01 4.71676528e-01 -1.38474762e-01 -5.49354732e-01
6.49876356e-01 -6.34858489e-01 2.63201654e-01 3.72993439e-01
-2.65595257e-01 7.71412313e-01 1.17762411e+00 -7.55310774e-01
-1.31295073e+00 -3.70637387e-01 6.15596890e-01 -4.24334079e-01
3.83988261e-01 -3.51760089e-01 -9.71097827e-01 8.37859333e-01
2.03759149e-02 -2.32844383e-01 9.46058393e-01 2.00915858e-01
1.55362472e-01 5.05655348e-01 -1.67229211e+00 1.04690421e+00
1.30222070e+00 -3.85475576e-01 -4.08403575e-01 2.98292991e-02
6.03179991e-01 -5.14247060e-01 -8.22250247e-01 6.50189724e-03
1.00842011e+00 -1.29457855e+00 6.79672003e-01 -1.11789517e-01
9.09195468e-02 -1.55039877e-01 -8.05505440e-02 -1.36316061e+00
-2.34472632e-01 -8.04517746e-01 5.20464420e-01 1.06097186e+00
1.12012684e-01 -9.85195577e-01 4.70619082e-01 1.13852298e+00
-6.34632483e-02 -3.60085636e-01 -1.09416568e+00 -8.22917104e-01
-3.55329424e-01 -4.44269568e-01 2.51227468e-01 1.10315764e+00
1.11557710e+00 2.81782206e-02 -6.06290221e-01 2.36381114e-01
1.67712256e-01 -4.46726799e-01 1.01384532e+00 -1.15611923e+00
-2.49472544e-01 1.56724174e-02 -9.05127585e-01 -5.90708375e-01
-3.64142179e-01 -4.19087529e-01 9.81587172e-02 -1.74731410e+00
-1.37506694e-01 -2.06980377e-01 4.31244731e-01 9.72191319e-02
2.77099341e-01 2.81822890e-01 2.57168233e-01 1.48786092e-02
-3.00711960e-01 6.24032259e-01 1.34005988e+00 7.62730122e-01
-1.26813543e+00 -1.73617542e-01 -6.38855517e-01 4.62215483e-01
6.65474951e-01 -1.82530776e-01 -7.38937616e-01 2.20482334e-01
3.44381034e-01 6.72648609e-01 5.64666271e-01 -9.43207145e-01
1.97898164e-01 -2.87365228e-01 -7.52384141e-02 -1.58753902e-01
9.08639908e-01 -7.37837493e-01 9.60020244e-01 6.36988580e-01
5.31827696e-02 -1.23884372e-01 3.88125718e-01 3.77498060e-01
3.64335686e-01 1.67595625e-01 4.89701420e-01 3.12468279e-02
-5.43780327e-01 -5.49377024e-01 -1.21536076e+00 -3.04714292e-01
1.60560024e+00 -1.06564975e+00 1.44824246e-02 -1.07389283e+00
-1.31469083e+00 1.75597295e-01 9.97893691e-01 4.50492084e-01
8.90559793e-01 -1.17649806e+00 -8.98448899e-02 -7.99605399e-02
1.51215732e-01 -6.53543830e-01 9.63527381e-01 1.21384907e+00
-6.31457031e-01 -6.96156323e-02 -5.26127219e-01 -5.13583899e-01
-1.70590043e+00 3.03397834e-01 4.13560390e-01 5.54284155e-01
-1.22971737e+00 4.58378047e-01 7.47145712e-02 -3.52260768e-01
2.93680802e-02 -9.67365317e-03 -3.78562421e-01 1.29689023e-01
3.85016590e-01 6.44564927e-01 -3.33028324e-02 -9.18741405e-01
-5.40070176e-01 3.85876805e-01 2.86510736e-01 -5.70926726e-01
1.00852299e+00 -6.78905129e-01 3.25271964e-01 9.85404193e-01
9.26825464e-01 1.81392685e-01 -8.27390850e-01 6.86597824e-01
-5.51029861e-01 -1.02657521e+00 -3.27377290e-01 -5.46478331e-01
-4.28357095e-01 2.66106844e-01 9.74183023e-01 2.34318584e-01
8.72181714e-01 -1.41943162e-02 7.20936596e-01 2.03672826e-01
8.89866531e-01 -9.68156457e-01 1.67914987e-01 -1.55643612e-01
7.75768876e-01 -7.50348568e-01 -2.86236107e-01 -7.29463577e-01
-1.21700108e+00 4.56599981e-01 7.35829949e-01 2.88764164e-02
2.10463688e-01 1.19164139e-01 2.70331234e-01 -3.38292867e-01
-6.49720788e-01 -1.33938745e-01 -4.93488252e-01 9.91018534e-01
4.23116475e-01 3.18014294e-01 -7.13010788e-01 5.59562206e-01
-5.79064906e-01 -1.68279354e-02 1.39480376e+00 1.23890924e+00
-4.19741273e-01 -5.83672702e-01 -5.00172377e-01 2.31455192e-02
2.53854901e-01 6.02994919e-01 -4.32245731e-01 1.05704498e+00
-1.75741140e-03 1.35947192e+00 2.13978350e-01 -6.86131537e-01
8.50696266e-01 -2.52870858e-01 3.81833583e-01 -5.26255429e-01
-7.87829399e-01 -1.50224522e-01 5.14640927e-01 -8.13048422e-01
-1.34549230e-01 -9.59174633e-01 -1.65809333e+00 -5.55402756e-01
-4.23378617e-01 2.06690267e-01 6.88897669e-01 7.10434258e-01
2.38575295e-01 4.26142126e-01 6.79592729e-01 -1.12228644e+00
8.51837471e-02 -7.35625327e-01 -9.42501009e-01 4.44187641e-01
2.70451635e-01 -8.29094410e-01 -5.75242937e-01 -4.75378484e-01] | [12.455071449279785, -0.11397939175367355] |
a89b9c52-6195-4b73-a010-50ce17eb8c26 | deep-learning-based-f0-synthesis-for-speaker | 2306.16860 | null | https://arxiv.org/abs/2306.16860v1 | https://arxiv.org/pdf/2306.16860v1.pdf | Deep Learning-based F0 Synthesis for Speaker Anonymization | Voice conversion for speaker anonymization is an emerging concept for privacy protection. In a deep learning setting, this is achieved by extracting multiple features from speech, altering the speaker identity, and waveform synthesis. However, many existing systems do not modify fundamental frequency (F0) trajectories, which convey prosody information and can reveal speaker identity. Moreover, mismatch between F0 and other features can degrade speech quality and intelligibility. In this paper, we formally introduce a method that synthesizes F0 trajectories from other speech features and evaluate its reconstructional capabilities. Then we test our approach within a speaker anonymization framework, comparing it to a baseline and a state-of-the-art F0 modification that utilizes speaker information. The results show that our method improves both speaker anonymity, measured by the equal error rate, and utility, measured by the word error rate. | ['Nils Peters', 'Ünal Ege Gaznepoglu'] | 2023-06-29 | null | null | null | null | ['voice-conversion', 'voice-conversion'] | ['audio', 'speech'] | [ 1.32483348e-01 3.02793920e-01 1.04980670e-01 -4.34120566e-01
-8.40515316e-01 -1.05381274e+00 5.37160516e-01 8.78057554e-02
-3.93125802e-01 5.58612287e-01 7.58812547e-01 -1.50177702e-01
2.28519499e-01 -5.96046686e-01 -5.86784244e-01 -6.82096720e-01
6.07267767e-02 -3.59902412e-01 -4.06733334e-01 -1.37221292e-01
-1.68801263e-01 7.16011167e-01 -1.37639749e+00 1.71306297e-01
6.64710939e-01 8.17988396e-01 -6.76913857e-01 5.48004389e-01
6.46583587e-02 1.74465507e-01 -1.17301309e+00 -9.09847736e-01
5.03710926e-01 -5.28844118e-01 -6.09605849e-01 -3.84088844e-01
4.94682491e-01 -4.62916285e-01 -6.38209820e-01 1.27463198e+00
9.61054802e-01 7.08559826e-02 2.95560181e-01 -1.39892125e+00
-6.26109719e-01 7.87497640e-01 1.34791583e-01 1.00813834e-02
4.62512940e-01 8.13352242e-02 7.06946313e-01 -5.30705452e-01
2.91388750e-01 1.48277080e+00 9.82461452e-01 8.54294419e-01
-1.58036494e+00 -9.98819888e-01 -7.57868886e-02 1.91655178e-02
-1.48478270e+00 -1.19210231e+00 9.70803142e-01 -2.02704638e-01
5.65574825e-01 6.25416934e-01 4.06398177e-01 1.54032111e+00
-1.58970907e-01 5.72854280e-01 6.64540827e-01 -2.64250636e-01
2.37358272e-01 5.48873901e-01 2.77013448e-03 2.88608491e-01
-7.33370483e-02 6.21908903e-01 -8.81272316e-01 -4.34434831e-01
1.98569685e-01 -4.15939152e-01 -6.52799129e-01 -2.41895411e-02
-9.90248144e-01 7.30103076e-01 -4.93298471e-02 7.22400323e-02
-3.54747884e-02 -4.67581954e-03 5.33627689e-01 7.01998830e-01
3.74341041e-01 4.35526282e-01 -4.47281450e-02 -7.70278051e-02
-9.16003644e-01 2.08678648e-01 1.03910649e+00 8.28945458e-01
4.23156291e-01 5.26521742e-01 -4.77649361e-01 7.41530955e-01
7.14425296e-02 6.49887323e-01 7.51568913e-01 -8.39385629e-01
3.23991805e-01 -2.21218616e-01 7.68593140e-03 -9.81396437e-01
-2.41527140e-01 -3.49703401e-01 -6.88861907e-01 1.17377862e-02
3.82509530e-01 -4.71911788e-01 -4.82089758e-01 2.18744755e+00
2.71522820e-01 2.51373976e-01 3.24262887e-01 7.28913486e-01
8.93274188e-01 4.95110840e-01 -1.60163268e-01 -2.41802946e-01
1.07037103e+00 -7.52957046e-01 -1.20362473e+00 1.67786211e-01
2.87161678e-01 -8.36348236e-01 1.18928552e+00 2.31693417e-01
-8.62425804e-01 -4.80751246e-01 -1.07746172e+00 -5.43311052e-02
-2.37657934e-01 -1.51327074e-01 1.97122604e-01 1.83625042e+00
-1.16113746e+00 5.55305779e-01 -4.82589364e-01 -7.81357959e-02
3.56334120e-01 5.14216840e-01 -5.87935328e-01 5.16075194e-01
-1.54612374e+00 5.35351992e-01 -2.32777297e-02 -3.19424748e-01
-7.76516438e-01 -1.19919717e+00 -9.17983949e-01 3.73050272e-01
-1.68363303e-01 -5.47445714e-01 1.26786041e+00 -7.94147730e-01
-2.15706444e+00 7.36203611e-01 -2.31017008e-01 -8.41526210e-01
8.30680370e-01 -2.22004399e-01 -1.04055595e+00 -3.08313556e-02
-3.38077575e-01 5.01811743e-01 1.30838466e+00 -1.00941944e+00
-4.40646738e-01 -2.98230618e-01 -2.08794966e-01 3.37912291e-02
-8.89143646e-01 1.23384865e-02 1.94255188e-01 -9.72790241e-01
-2.53711492e-01 -8.07271004e-01 1.35589108e-01 -1.16848825e-02
-7.48757422e-01 3.76259595e-01 7.28144526e-01 -1.08082855e+00
1.23894918e+00 -2.72221637e+00 -2.85553962e-01 2.57716835e-01
2.53331587e-02 3.35449517e-01 -5.20078391e-02 3.70451599e-01
-2.77148724e-01 3.84215385e-01 -4.07098770e-01 -7.45527148e-01
3.73444140e-01 -2.90492564e-01 -7.75665641e-01 7.94231892e-01
-3.89734730e-02 6.69467211e-01 -5.26690960e-01 7.52046183e-02
2.11654946e-01 7.68236876e-01 -7.76711881e-01 9.77567881e-02
3.52464259e-01 5.67394793e-01 2.64661878e-01 3.65747273e-01
1.04324830e+00 8.12435508e-01 3.83816436e-02 2.01979512e-03
-2.99782434e-04 6.89632595e-01 -1.10439992e+00 1.59794676e+00
-3.62263024e-01 8.77051711e-01 2.66906917e-01 -5.53734064e-01
1.09992421e+00 6.96382344e-01 3.64318728e-01 -3.22262079e-01
9.28679332e-02 -4.83128205e-02 -9.88093615e-02 -2.34074682e-01
4.13813293e-01 -1.58335418e-01 -8.66494924e-02 4.38265055e-01
-2.39544734e-02 -4.30537760e-02 -6.09657526e-01 -2.51630515e-01
9.42173779e-01 -6.26105726e-01 3.05036634e-01 -2.83785641e-01
6.45923793e-01 -8.57609630e-01 5.44945180e-01 6.79764032e-01
-5.47406077e-01 6.73249900e-01 3.96744490e-01 -3.62533219e-02
-9.03052866e-01 -1.45799911e+00 -3.16894829e-01 7.86089420e-01
-1.57346830e-01 -3.23132515e-01 -1.30177391e+00 -5.66240966e-01
3.72465819e-01 1.14413881e+00 -3.95915687e-01 -6.65955186e-01
-5.25138259e-01 -3.95201713e-01 1.41863346e+00 8.16880018e-02
2.54868299e-01 -7.70873070e-01 9.98404622e-02 2.14469463e-01
-2.39426106e-01 -1.15238488e+00 -1.11427796e+00 -2.84394592e-01
-4.06499594e-01 -2.67555535e-01 -4.68961239e-01 -4.94949520e-01
1.66447431e-01 1.19350934e-02 5.46092391e-01 -3.50132376e-01
1.12817781e-02 2.77399331e-01 1.53600788e-02 -6.26218438e-01
-9.84380662e-01 3.20190340e-01 5.41853011e-01 6.32477164e-01
2.85892904e-01 -8.89214635e-01 -3.08175862e-01 7.37593696e-02
-8.92922282e-01 -6.28351927e-01 -4.96824458e-02 5.33118248e-01
7.65032247e-02 5.59711233e-02 1.00070441e+00 -7.13701487e-01
8.31883609e-01 -2.60870039e-01 -3.60596955e-01 -5.74661419e-02
-4.99383569e-01 -5.36974967e-02 8.72415364e-01 -5.43642282e-01
-9.85812485e-01 1.72663420e-01 -4.35075223e-01 -4.56904680e-01
-3.63081962e-01 -4.47759666e-02 -8.89891148e-01 -1.77055806e-01
8.18323731e-01 4.28162009e-01 2.58465052e-01 -5.77990413e-01
7.14511931e-01 1.07913899e+00 9.66876090e-01 -2.19064772e-01
9.50329125e-01 4.89128321e-01 -3.96026820e-01 -1.12890065e+00
-2.06763566e-01 -1.33601919e-01 -5.41821718e-01 3.38238738e-02
3.67246956e-01 -9.41280484e-01 -1.01018953e+00 6.63191259e-01
-1.24720836e+00 2.37614587e-01 -3.89914155e-01 5.15855670e-01
-4.83187586e-01 5.22629321e-01 -4.41710055e-01 -9.35382128e-01
-5.40473521e-01 -9.87340927e-01 8.71361375e-01 -1.18337177e-01
-4.37400848e-01 -8.32423031e-01 -6.80401698e-02 2.68588394e-01
5.71266532e-01 2.60910958e-01 7.78996706e-01 -9.76401627e-01
-1.04771607e-01 -2.96679467e-01 3.16487789e-01 4.97247964e-01
5.34340262e-01 -3.11730981e-01 -1.79806483e+00 -5.11513770e-01
4.67066646e-01 3.42177600e-01 6.29255235e-01 1.64168641e-01
1.13549328e+00 -1.03121519e+00 -1.53347319e-02 1.09272921e+00
7.74694383e-01 2.08851010e-01 6.67077363e-01 -6.39870539e-02
7.73094118e-01 8.54173362e-01 -2.66639411e-01 5.11529744e-01
3.21383514e-02 7.56969810e-01 8.72940198e-02 9.96242464e-02
-2.78109074e-01 -4.10980940e-01 7.57051706e-01 6.31391406e-01
5.70422947e-01 -3.07211727e-01 -5.25771618e-01 6.17228508e-01
-1.31858397e+00 -1.18184018e+00 2.35169426e-01 2.48470211e+00
1.01107180e+00 -2.76763827e-01 4.18849975e-01 5.11288702e-01
1.10654891e+00 2.98575044e-01 -6.75512791e-01 -5.96910775e-01
-2.51804978e-01 2.56766468e-01 5.83598077e-01 7.93220282e-01
-1.06213427e+00 9.15903807e-01 6.56733322e+00 5.43150902e-01
-1.49355960e+00 9.10282210e-02 3.81221145e-01 -6.12570405e-01
-5.80880404e-01 -5.66024780e-01 -5.15779972e-01 5.68194628e-01
1.36905932e+00 -7.78868318e-01 8.48065317e-01 6.02481067e-01
2.69711673e-01 9.31534410e-01 -1.48779726e+00 1.11334324e+00
2.63488382e-01 -1.14554083e+00 1.57105252e-01 2.77865887e-01
2.74012536e-01 -4.40378577e-01 5.35489082e-01 -4.42382656e-02
9.77829918e-02 -1.15966511e+00 8.71172607e-01 2.96359867e-01
9.43687141e-01 -1.20410597e+00 3.30298066e-01 -5.97865433e-02
-8.15828621e-01 -1.18398577e-01 -5.43053299e-02 2.72981048e-01
1.78037450e-01 6.63705289e-01 -9.85314846e-01 5.38952947e-01
6.31717563e-01 2.94092119e-01 -1.51935622e-01 7.54521012e-01
-8.48435089e-02 7.01014400e-01 -2.98295170e-01 2.21094787e-01
-3.08852017e-01 -1.72805674e-02 1.07634091e+00 1.15782249e+00
4.80070353e-01 -1.08125545e-01 -3.53528798e-01 9.70221877e-01
-5.16665936e-01 1.35589644e-01 -7.08038092e-01 -9.24310833e-02
1.08333910e+00 7.30723202e-01 6.28933087e-02 5.73089235e-02
-8.24604779e-02 1.21751273e+00 -5.10597602e-02 3.91668856e-01
-7.78681219e-01 -6.67537868e-01 1.46580756e+00 4.34124880e-02
1.28103957e-01 1.35146111e-01 -4.57330912e-01 -1.12749839e+00
3.31819385e-01 -1.07020164e+00 1.15548737e-01 -4.59437743e-02
-1.23540819e+00 6.60398126e-01 -4.48158354e-01 -1.15896595e+00
-3.67236704e-01 -1.83158051e-02 -4.55172271e-01 1.07233632e+00
-1.32052970e+00 -8.60272884e-01 1.79051816e-01 6.55237913e-01
1.05802864e-01 -5.09257257e-01 1.21078682e+00 5.43973386e-01
-5.81925094e-01 1.62297225e+00 3.73841465e-01 2.52589673e-01
8.67141545e-01 -1.05712736e+00 1.19234455e+00 9.55234587e-01
3.20826352e-01 6.84055746e-01 6.83931589e-01 -2.26822212e-01
-1.24416041e+00 -1.30491388e+00 1.00869715e+00 -4.84494120e-01
3.28179717e-01 -7.99824357e-01 -9.70914900e-01 5.14035642e-01
1.30992174e-01 -4.10558879e-02 1.18160415e+00 -1.61090121e-01
-5.20651996e-01 -2.05313638e-01 -1.79307103e+00 6.76723301e-01
1.12093246e+00 -1.18526673e+00 -4.80313987e-01 -6.79448172e-02
1.30257583e+00 -2.68826395e-01 -9.15096760e-01 -9.59401652e-02
7.40184247e-01 -1.04046273e+00 1.15980983e+00 -6.54513478e-01
-1.96343616e-01 -1.70961544e-02 -3.76743108e-01 -1.59276426e+00
-1.69403657e-01 -1.46168637e+00 -1.25857666e-01 1.87865579e+00
5.38076043e-01 -9.79134798e-01 7.13732004e-01 6.85479939e-01
-6.53110370e-02 1.84185073e-01 -1.26131201e+00 -1.15337741e+00
2.82228857e-01 -3.55317622e-01 1.46655691e+00 1.27803838e+00
-3.62383351e-02 1.22025624e-01 -5.04627228e-01 6.98402941e-01
6.02893412e-01 -2.87766725e-01 8.83273065e-01 -1.14534593e+00
-1.13061629e-01 -4.82242882e-01 -5.06832421e-01 -5.09260297e-01
6.18723989e-01 -9.93524790e-01 -1.84491709e-01 -5.62101305e-01
-5.97791612e-01 1.10293142e-01 -2.38232478e-01 7.40426928e-02
-7.30816275e-02 -1.16150975e-01 3.39343876e-01 -1.43355414e-01
5.35330415e-01 8.73010993e-01 5.58375776e-01 -1.74855664e-01
-5.30284882e-01 2.30279818e-01 -8.20468545e-01 5.16997516e-01
9.66090500e-01 -6.72937810e-01 -4.26793754e-01 -2.91987747e-01
-5.32368422e-01 2.30196826e-02 2.62697726e-01 -1.13289928e+00
1.08546011e-01 2.03661442e-01 8.22807327e-02 -4.99841869e-02
5.33862710e-01 -8.23810577e-01 1.87768951e-01 4.15661335e-01
-6.59815788e-01 -2.55008221e-01 3.83266479e-01 5.09904802e-01
-1.75513536e-01 4.12983410e-02 7.89474428e-01 3.75592619e-01
3.49855237e-02 3.06185633e-01 -3.88893694e-01 -4.65679206e-02
6.68758750e-01 -1.19179124e-02 -3.45806479e-01 -6.67289853e-01
-6.57701969e-01 -4.34291691e-01 4.21557337e-01 6.98067486e-01
3.94203305e-01 -1.47985876e+00 -8.76686692e-01 6.94701493e-01
3.86014171e-02 -6.13477528e-01 1.58294737e-01 3.15798908e-01
-6.66210800e-02 3.15793723e-01 -4.26378697e-02 -2.88204879e-01
-1.39490199e+00 7.21305966e-01 5.04280210e-01 5.88691771e-01
-6.22390270e-01 8.26536298e-01 2.71670431e-01 -8.46732318e-01
5.26108801e-01 -2.62963474e-01 4.04219143e-02 1.36644527e-01
6.92211926e-01 4.77236569e-01 3.56777072e-01 -7.61243880e-01
-5.22715688e-01 9.51173231e-02 2.17616990e-01 -5.94301462e-01
9.73446608e-01 -4.34111565e-01 1.33410513e-01 2.18943253e-01
1.65235007e+00 6.17966294e-01 -9.93148386e-01 -3.19367707e-01
-1.89720616e-01 -6.36120081e-01 1.47876114e-01 -5.96991003e-01
-1.14168286e+00 8.61736834e-01 8.65779996e-01 5.53389549e-01
9.10091221e-01 -5.04211783e-01 1.13382185e+00 3.36298496e-01
-2.27599256e-02 -7.96601057e-01 -4.56359863e-01 2.39617720e-01
8.91350865e-01 -9.34961379e-01 -5.60221910e-01 -6.29244149e-01
-6.59815788e-01 7.66356707e-01 4.82534580e-02 3.65439862e-01
8.44390452e-01 3.33853811e-01 5.02058327e-01 5.18844366e-01
-2.70274192e-01 1.74040496e-01 1.40422866e-01 9.61530745e-01
2.37096220e-01 1.99886128e-01 1.17474705e-01 9.83569741e-01
-1.29680407e+00 -5.32812238e-01 5.30254006e-01 2.46446803e-01
6.72532152e-03 -1.14159703e+00 -6.16080821e-01 6.13215230e-02
-6.99322402e-01 -2.77073950e-01 -7.70495176e-01 2.53807276e-01
-1.96081832e-01 1.36534595e+00 1.15519747e-01 -6.28395915e-01
5.57700992e-01 3.31622362e-01 -1.88482292e-02 -3.06769103e-01
-9.27139103e-01 -2.91394740e-01 2.64865994e-01 -4.58452612e-01
9.02803913e-02 -9.03958917e-01 -1.07094300e+00 -9.41969693e-01
-9.38896686e-02 9.51734111e-02 8.88897300e-01 6.53746247e-01
6.10143244e-01 4.75965470e-01 1.26233137e+00 -3.43621016e-01
-9.06323016e-01 -7.01456130e-01 -6.72988713e-01 3.47103715e-01
1.06265473e+00 -2.67246943e-02 -6.41050279e-01 2.06346944e-01] | [14.01090145111084, 5.872395992279053] |
b696a128-6fe0-48ce-89f9-632008bdbf0d | bodega-benchmark-for-adversarial-example | 2303.08032 | null | https://arxiv.org/abs/2303.08032v1 | https://arxiv.org/pdf/2303.08032v1.pdf | BODEGA: Benchmark for Adversarial Example Generation in Credibility Assessment | Text classification methods have been widely investigated as a way to detect content of low credibility: fake news, social media bots, propaganda, etc. Quite accurate models (likely based on deep neural networks) help in moderating public electronic platforms and often cause content creators to face rejection of their submissions or removal of already published texts. Having the incentive to evade further detection, content creators try to come up with a slightly modified version of the text (known as an attack with an adversarial example) that exploit the weaknesses of classifiers and result in a different output. Here we introduce BODEGA: a benchmark for testing both victim models and attack methods on four misinformation detection tasks in an evaluation framework designed to simulate real use-cases of content moderation. We also systematically test the robustness of popular text classifiers against available attacking techniques and discover that, indeed, in some cases barely significant changes in input text can mislead the models. We openly share the BODEGA code and data in hope of enhancing the comparability and replicability of further research in this area. | ['Horacio Saggion', 'Alexander Shvets', 'Piotr Przybyła'] | 2023-03-14 | null | null | null | null | ['misinformation'] | ['miscellaneous'] | [ 9.06054154e-02 4.10442621e-01 -9.90890637e-02 -1.20589145e-01
-4.44330633e-01 -1.00830233e+00 1.20471179e+00 5.56897223e-01
-2.68025070e-01 6.77689612e-01 1.37826189e-01 -4.91915047e-01
3.66477698e-01 -8.44351470e-01 -7.86453247e-01 -4.07555282e-01
3.25630372e-03 4.28596646e-01 3.00357193e-01 -5.40931761e-01
5.86152017e-01 1.13298036e-01 -1.12797081e+00 8.00634027e-01
6.48781419e-01 5.93394458e-01 -6.11744344e-01 6.20887578e-01
8.31990093e-02 9.44568813e-01 -1.31950843e+00 -1.13204980e+00
3.03086370e-01 -1.96161911e-01 -9.23364460e-01 -3.41967404e-01
7.29673028e-01 -3.06519926e-01 -5.16018331e-01 1.42694306e+00
4.14132774e-01 -4.87903327e-01 5.74982941e-01 -1.33787024e+00
-4.78914052e-01 1.33549178e+00 -4.57995772e-01 3.73819083e-01
4.05771971e-01 3.76344502e-01 8.24369550e-01 -5.79129934e-01
9.67496157e-01 1.39720058e+00 8.74399126e-01 5.47923088e-01
-1.09720433e+00 -7.22563028e-01 -2.12918043e-01 3.40933576e-02
-7.83980906e-01 -5.49603760e-01 6.07017457e-01 -4.52561349e-01
3.60195100e-01 5.35286844e-01 2.29060411e-01 1.90929127e+00
3.92994523e-01 7.13815689e-01 1.33877087e+00 -8.40820372e-02
1.25335082e-01 5.69520533e-01 2.62720942e-01 6.03769183e-01
5.82885206e-01 8.57978314e-02 -3.93001050e-01 -9.17663157e-01
-3.23251098e-01 -4.57144171e-01 -2.74392903e-01 2.49527767e-01
-9.82390761e-01 1.29064310e+00 5.71432948e-01 5.11179030e-01
-2.61221588e-01 1.08959891e-01 9.54901218e-01 7.27814138e-01
6.97617352e-01 9.22182143e-01 -2.24128261e-01 -7.13975653e-02
-8.05984020e-01 5.89712203e-01 1.29938400e+00 2.41777524e-01
2.23699048e-01 -6.57459423e-02 -2.52631634e-01 4.42391396e-01
-1.54291570e-01 4.13045943e-01 7.07907319e-01 -4.70892698e-01
6.66159093e-01 3.89047354e-01 2.21988335e-01 -1.64940429e+00
-2.50334471e-01 -5.66965342e-01 -7.11305916e-01 2.56114364e-01
8.71724844e-01 -4.05035108e-01 -3.04700434e-01 1.19593489e+00
1.76865861e-01 -1.48234099e-01 -1.98018089e-01 7.88072586e-01
5.60570955e-01 5.16850829e-01 -1.02340817e-01 1.29208639e-01
1.28056574e+00 -8.57614040e-01 -5.49680769e-01 -3.05427372e-01
8.27043712e-01 -8.26865315e-01 7.68576980e-01 6.58023596e-01
-8.79595399e-01 -7.06269518e-02 -1.18387461e+00 3.13791871e-01
-8.46786201e-01 -4.84091550e-01 2.57751226e-01 1.05091012e+00
-5.05546868e-01 1.08261836e+00 -2.53651440e-01 5.06994762e-02
7.86061466e-01 -1.85540706e-01 -3.01131338e-01 2.58969426e-01
-1.75346053e+00 1.11504245e+00 1.97086126e-01 -4.68145274e-02
-1.01705670e+00 -6.16385162e-01 -1.96694210e-01 -9.55467746e-02
4.61996883e-01 -1.12618200e-01 1.30641210e+00 -1.41165304e+00
-1.06890452e+00 1.14235950e+00 5.16268015e-01 -8.39477539e-01
1.37409878e+00 -2.12100714e-01 -3.81402373e-01 -3.39645613e-03
3.82117890e-02 -2.69791801e-02 1.55100751e+00 -9.83557463e-01
-2.58573502e-01 -3.34677011e-01 1.06801055e-01 -4.50887173e-01
-5.47667861e-01 3.95240426e-01 6.33422017e-01 -8.05749297e-01
-3.76168728e-01 -7.55293906e-01 1.30507529e-01 -1.07870981e-01
-9.48433101e-01 6.22388385e-02 1.16365957e+00 -8.06329012e-01
1.18930221e+00 -1.71838951e+00 -5.77837490e-02 3.65707874e-01
7.04899430e-01 7.23581314e-01 -5.46904746e-03 5.77266097e-01
4.25463170e-02 7.64814436e-01 -9.83308256e-03 -1.57931045e-01
-1.96905877e-03 -2.95236766e-01 -6.68877184e-01 9.19142842e-01
-4.44573574e-02 8.85815024e-01 -9.62345541e-01 -5.68364561e-02
-3.61555040e-01 -5.83220199e-02 -3.03740501e-01 -1.72129944e-01
-4.95707333e-01 2.71078423e-02 -4.91287708e-01 3.82342637e-01
5.34504533e-01 -3.14992875e-01 7.53073841e-02 1.84461072e-01
1.57238573e-01 7.77535260e-01 -7.98645616e-01 5.94816029e-01
-1.55973360e-01 1.19094551e+00 2.46244922e-01 -9.12860096e-01
4.96084750e-01 2.18546852e-01 -2.44214341e-01 -3.54787618e-01
6.05701327e-01 4.18840826e-01 2.01646239e-01 -4.10034329e-01
6.40959978e-01 2.67070830e-01 -1.45983726e-01 1.00647223e+00
-3.93712133e-01 2.40493327e-01 -1.09134950e-01 6.07953131e-01
1.25721002e+00 -4.80768979e-01 1.47571161e-01 -6.46577626e-02
4.52869713e-01 7.60029554e-02 -2.01764241e-01 1.49881220e+00
-3.37483555e-01 2.03516573e-01 9.64416027e-01 -6.86002970e-01
-1.34812427e+00 -3.56272101e-01 -1.10449493e-01 1.10646045e+00
-1.30887270e-01 -4.66996223e-01 -9.06175911e-01 -1.42384851e+00
3.64519984e-01 8.52991104e-01 -9.78147447e-01 -3.28746557e-01
-4.99349505e-01 -7.71527946e-01 1.26358557e+00 -1.44224212e-01
4.56016600e-01 -9.84827101e-01 -5.01111388e-01 3.70420307e-01
-2.69655108e-01 -8.54004443e-01 -1.67315945e-01 -5.52252047e-02
-4.73038852e-01 -1.37541401e+00 -3.87461066e-01 5.09239398e-02
4.04564351e-01 1.70081571e-01 9.73934829e-01 7.41689324e-01
-3.75414155e-02 -1.41490594e-01 -5.54874241e-01 -4.24121261e-01
-1.60548115e+00 2.82332212e-01 2.79582348e-02 7.93600157e-02
1.94118679e-01 -2.34395117e-01 -2.53838480e-01 4.43108708e-01
-1.21004546e+00 -2.04875126e-01 9.94226858e-02 9.37839687e-01
-7.31140733e-01 2.91809328e-02 5.92522562e-01 -1.56018484e+00
1.14134657e+00 -9.12304997e-01 -5.76571047e-01 -1.25171587e-01
-7.84394264e-01 -2.86460668e-02 9.18064773e-01 -8.21454823e-01
-5.68072677e-01 -7.12479889e-01 3.23689394e-02 -2.84184944e-02
-2.03259103e-02 5.85580766e-01 4.02630240e-01 -2.64021456e-01
1.64101470e+00 4.65473756e-02 1.81149885e-01 -4.68118310e-01
3.24985147e-01 9.20164108e-01 1.10993430e-01 -2.44402513e-01
1.28508008e+00 4.87395495e-01 -4.13374960e-01 -5.23330748e-01
-9.02504146e-01 -3.71188447e-02 -1.15739338e-01 -2.34024748e-01
9.52656120e-02 -5.01707733e-01 -6.44825995e-01 7.50968874e-01
-1.44691014e+00 -1.31172389e-01 4.63007689e-01 -2.97583252e-01
1.82565808e-01 6.15049362e-01 -9.08787608e-01 -5.72581708e-01
-3.58456939e-01 -9.72166002e-01 4.19822693e-01 -1.97759748e-01
-4.35782015e-01 -9.06154096e-01 -9.13433060e-02 5.43791592e-01
7.84962475e-01 3.60017627e-01 8.40059876e-01 -1.42668712e+00
1.95120554e-02 -8.46733987e-01 -1.32747635e-01 3.42999488e-01
-2.53954947e-01 3.23401719e-01 -1.28325427e+00 -3.73011172e-01
-1.26960492e-02 -3.85593951e-01 7.73588479e-01 -3.23864698e-01
9.63949323e-01 -1.10299671e+00 -4.32003736e-01 -1.30417019e-01
8.78365636e-01 -3.81556362e-01 6.53360188e-01 7.27446795e-01
4.65687931e-01 7.90529132e-01 8.96024629e-02 4.33714956e-01
-2.08906278e-01 6.83172822e-01 6.79100573e-01 2.27703974e-01
2.55192757e-01 -3.12848389e-01 5.26567280e-01 3.00002515e-01
2.95235783e-01 -5.27141809e-01 -7.37058342e-01 1.15664281e-01
-1.60407293e+00 -1.20802748e+00 -6.07863307e-01 1.96347666e+00
9.96908128e-01 6.79796159e-01 2.56521761e-01 3.90653312e-01
9.40569520e-01 3.72016549e-01 -2.62256861e-01 -6.70434237e-01
-1.28074795e-01 -1.25902414e-01 7.27602184e-01 4.71511483e-01
-1.23421741e+00 8.70408356e-01 5.84197712e+00 7.54776835e-01
-1.18845892e+00 3.97431403e-01 8.39136600e-01 1.12555824e-01
-2.06861094e-01 -1.58926398e-02 -5.88914037e-01 9.09317315e-01
1.15573394e+00 -2.49182463e-01 4.32189256e-01 1.01100981e+00
2.63703972e-01 -7.27156550e-02 -9.48258281e-01 3.69079322e-01
2.44102895e-01 -1.76066661e+00 -1.06222726e-01 1.77859768e-01
6.73169911e-01 2.00544760e-01 1.03370354e-01 3.26432228e-01
4.81955230e-01 -9.87408996e-01 9.79675353e-01 2.18532104e-02
2.26809397e-01 -3.69204134e-01 9.49250638e-01 5.77331483e-01
7.75626749e-02 -2.15794489e-01 -2.13246271e-01 -2.23730177e-01
-1.62005231e-01 6.16012573e-01 -1.12384498e+00 1.67393625e-01
3.60346764e-01 3.38383913e-01 -9.64341462e-01 8.13778520e-01
-3.23139369e-01 1.07084203e+00 -2.46106133e-01 -5.14531612e-01
4.87996429e-01 4.36581463e-01 1.09788799e+00 1.22187054e+00
-3.33368927e-01 -4.37080979e-01 -1.89564884e-01 1.01524770e+00
-5.18889546e-01 7.76568577e-02 -7.24170327e-01 -4.10917610e-01
4.14965659e-01 1.12556076e+00 -5.55242836e-01 -5.08395433e-01
-9.06841382e-02 8.38920712e-01 2.86157727e-01 5.82060516e-02
-7.40332484e-01 -2.50033617e-01 2.70219982e-01 5.42087972e-01
3.02863540e-03 1.62087530e-01 -1.81898922e-01 -1.22237504e+00
-2.34408546e-02 -1.61577046e+00 2.61126786e-01 -6.50874138e-01
-1.56585062e+00 6.57048345e-01 -2.31830016e-01 -9.14743662e-01
-1.61013424e-01 -5.17363191e-01 -8.07416081e-01 6.43643916e-01
-1.05683196e+00 -6.89975560e-01 -1.08988374e-01 1.79684341e-01
3.41285020e-01 -3.61875802e-01 4.72586036e-01 1.08500406e-01
-3.60540301e-01 6.38365984e-01 1.65583655e-01 5.61534941e-01
8.04298759e-01 -9.88330185e-01 7.43464470e-01 6.58048451e-01
1.21843807e-01 5.31956375e-01 1.26681817e+00 -9.70615983e-01
-1.00362885e+00 -8.99038434e-01 8.74517500e-01 -9.08172607e-01
1.37926805e+00 -6.34338677e-01 -1.25013554e+00 5.05068660e-01
2.82800645e-01 -2.51516134e-01 2.97484994e-01 -6.48474619e-02
-8.99244428e-01 4.17604268e-01 -1.35159004e+00 7.81464398e-01
4.60649669e-01 -4.31280553e-01 -5.24959326e-01 7.93317199e-01
4.41744477e-01 -3.18778396e-01 -2.37938985e-01 -1.95411995e-01
5.31983256e-01 -1.12782657e+00 4.59442198e-01 -1.12666750e+00
9.33982909e-01 9.65311155e-02 3.23505580e-01 -1.32954860e+00
4.69065234e-02 -8.74952853e-01 -2.42867231e-01 9.31129932e-01
5.97324789e-01 -9.50477719e-01 5.98413885e-01 2.35350564e-01
4.58365917e-01 -2.63483673e-01 -8.47276032e-01 -5.01081526e-01
3.22864771e-01 -3.06278139e-01 2.64415443e-01 1.47356749e+00
2.14821264e-01 2.35554844e-01 -5.76440394e-01 8.90388936e-02
7.39059865e-01 -3.05714458e-01 9.33855355e-01 -1.33844817e+00
-2.68610805e-01 -7.60618448e-01 -3.05964053e-01 -4.88285363e-01
1.74088955e-01 -9.48653162e-01 -2.51062214e-01 -5.16277671e-01
9.85224769e-02 -3.09986979e-01 2.94269085e-01 4.48079079e-01
-2.32244492e-01 5.20229220e-01 1.93545550e-01 4.23482597e-01
-3.39417607e-01 -5.23818545e-02 9.87031698e-01 -5.72630346e-01
1.87730566e-01 2.82843024e-01 -7.53747821e-01 7.20106602e-01
8.27989399e-01 -1.21194303e+00 2.60199130e-01 9.04884264e-02
8.72738540e-01 -1.24174580e-01 6.91841304e-01 -5.74433386e-01
-3.35618444e-02 7.82770216e-02 1.53996989e-01 -8.71416628e-02
-2.26394206e-01 -5.38135111e-01 -2.97570318e-01 7.71124363e-01
-8.59600186e-01 2.55389828e-02 -8.36539492e-02 7.52224922e-01
2.04520136e-01 -7.23319113e-01 8.84330153e-01 -2.46078908e-01
1.88114807e-01 -4.23688889e-02 -1.00373054e+00 1.60870284e-01
6.77534938e-01 1.28931314e-01 -1.02345526e+00 -8.58613491e-01
-5.70163846e-01 -7.14611337e-02 4.23780918e-01 5.86231887e-01
1.64357442e-02 -8.09845269e-01 -1.18455076e+00 -1.25799492e-01
8.19617882e-03 -8.89644206e-01 -2.89806753e-01 7.31549025e-01
-6.36063516e-01 5.38416067e-03 -3.90754752e-02 -1.48301110e-01
-1.24197745e+00 5.81609845e-01 5.85286736e-01 -2.69640118e-01
-6.06036127e-01 4.67212707e-01 -4.74211037e-01 -2.28717253e-01
1.62076503e-01 2.09062710e-01 -5.43782264e-02 9.95804891e-02
8.45574260e-01 4.66112047e-01 3.27669501e-01 -5.67080736e-01
-8.81883949e-02 -7.13789165e-01 -5.31177461e-01 6.37731478e-02
1.03027737e+00 2.23562807e-01 -2.14998886e-01 1.87149256e-01
1.06450903e+00 5.22826135e-01 -5.91721296e-01 -3.36160958e-01
1.99609116e-01 -6.43372536e-01 2.55682971e-02 -1.24277961e+00
-8.44256341e-01 6.63586318e-01 2.09929734e-01 1.17714202e+00
1.02530345e-01 -1.76517278e-01 6.37442946e-01 4.63323087e-01
1.54029772e-01 -9.62130845e-01 3.11272025e-01 4.40438360e-01
1.03543758e+00 -1.30974567e+00 2.43731320e-01 -1.07133865e-01
-4.88605261e-01 1.29122925e+00 3.89329374e-01 -1.09340273e-01
3.92112046e-01 1.29563361e-01 1.63953602e-01 -3.81307453e-01
-8.41384411e-01 6.69557989e-01 4.68049161e-02 3.68782282e-01
2.48231009e-01 -1.48522696e-02 -6.14863634e-01 3.53894681e-01
-2.72770733e-01 -6.01363719e-01 1.12722898e+00 5.93540967e-01
-4.18947190e-01 -8.61019909e-01 -6.69632256e-01 6.52536154e-01
-1.00651276e+00 -2.38666207e-01 -1.10544431e+00 6.67809844e-01
-2.31439948e-01 1.06987393e+00 -2.95699179e-01 -4.02181596e-01
-1.28428355e-01 1.58195421e-01 -5.14445901e-02 -3.04294616e-01
-1.33714747e+00 -4.69093025e-01 6.70893192e-01 -4.34757233e-01
6.83147758e-02 -5.39265096e-01 -5.34541786e-01 -9.55481708e-01
-6.12219453e-01 2.46401370e-01 9.66549873e-01 1.07243335e+00
2.93060780e-01 -5.98690752e-03 7.86281526e-01 -7.32791543e-01
-1.15443575e+00 -1.36440825e+00 -1.77063957e-01 8.10799122e-01
3.29950124e-01 -5.37543058e-01 -9.98445928e-01 -2.01130092e-01] | [8.129610061645508, 10.22708511352539] |
3e9f829c-2d51-48b0-aaca-91e1e51f3623 | mcl-3d-a-database-for-stereoscopic-image | 1405.1403 | null | http://arxiv.org/abs/1405.1403v1 | http://arxiv.org/pdf/1405.1403v1.pdf | MCL-3D: a database for stereoscopic image quality assessment using 2D-image-plus-depth source | A new stereoscopic image quality assessment database rendered using the
2D-image-plus-depth source, called MCL-3D, is described and the performance
benchmarking of several known 2D and 3D image quality metrics using the MCL-3D
database is presented in this work. Nine image-plus-depth sources are first
selected, and a depth image-based rendering (DIBR) technique is used to render
stereoscopic image pairs. Distortions applied to either the texture image or
the depth image before stereoscopic image rendering include: Gaussian blur,
additive white noise, down-sampling blur, JPEG and JPEG-2000 (JP2K) compression
and transmission error. Furthermore, the distortion caused by imperfect
rendering is also examined. The MCL-3D database contains 693 stereoscopic image
pairs, where one third of them are of resolution 1024x728 and two thirds are of
resolution 1920x1080. The pair-wise comparison was adopted in the subjective
test for user friendliness, and the Mean Opinion Score (MOS) can be computed
accordingly. Finally, we evaluate the performance of several 2D and 3D image
quality metrics applied to MCL-3D. All texture images, depth images, rendered
image pairs in MCL-3D and their MOS values obtained in the subjective test are
available to the public (http://mcl.usc.edu/mcl-3d-database/) for future
research and development. | ['C. -C. Jay Kuo', 'Rui Song', 'Hyunsuk Ko'] | 2014-03-23 | null | null | null | null | ['stereoscopic-image-quality-assessment'] | ['computer-vision'] | [ 3.45414013e-01 -4.55882907e-01 2.77829021e-01 -2.71104336e-01
-1.01030898e+00 -4.45040137e-01 3.09549451e-01 -1.04997130e-02
-2.57415920e-01 5.91287076e-01 2.78126538e-01 -1.99188128e-01
-1.45712510e-01 -7.12827265e-01 -2.50301898e-01 -7.08585680e-01
-2.35982582e-01 -2.43525967e-01 5.64355135e-01 -3.14005464e-02
6.83632970e-01 3.86016428e-01 -1.86957359e+00 3.08394819e-01
9.30256724e-01 1.59792554e+00 4.60459679e-01 1.03915381e+00
3.57219219e-01 6.60520852e-01 -7.72031426e-01 -3.90444100e-01
5.40423632e-01 -3.11232090e-01 -5.59446514e-01 2.80062973e-01
5.76248407e-01 -8.39867115e-01 -3.13043296e-01 1.14244795e+00
1.10607255e+00 -1.20902032e-01 5.19890785e-01 -1.12788320e+00
-4.09587026e-01 -7.47842491e-01 -6.03091002e-01 4.87140298e-01
9.40597296e-01 4.61103052e-01 2.10609302e-01 -6.52685881e-01
6.36727691e-01 1.36303067e+00 1.59107789e-01 6.13440312e-02
-1.02210891e+00 -6.09549820e-01 -7.06578374e-01 4.32627499e-01
-1.45626235e+00 -4.17316318e-01 5.47232568e-01 -4.16028231e-01
7.15598404e-01 4.93260473e-01 5.96381545e-01 6.85720921e-01
6.58538997e-01 2.01420262e-01 1.57675195e+00 -5.16068339e-01
3.51659983e-01 1.06629044e-01 -2.53900081e-01 5.03316700e-01
7.78893679e-02 3.69551271e-01 -5.97603679e-01 -1.72501370e-01
1.13095820e+00 -4.09475982e-01 -5.85810602e-01 -1.26050249e-01
-9.94375706e-01 3.19241256e-01 1.62701130e-01 -5.93494102e-02
-2.22269014e-01 -3.59911680e-01 5.31202793e-01 6.98687196e-01
5.85492790e-01 1.02689929e-01 -8.50362107e-02 -4.55599725e-01
-5.60640574e-01 3.86495255e-02 4.11364347e-01 9.65133190e-01
6.53465986e-01 -1.76078796e-01 -6.14316538e-02 1.09832537e+00
1.28846437e-01 8.34429026e-01 5.60836494e-01 -1.69294119e+00
5.39028406e-01 2.53510982e-01 3.04919183e-01 -1.17649913e+00
6.20115399e-02 2.86848605e-01 -8.55287194e-01 9.15390491e-01
-3.19079831e-02 1.59507051e-01 -6.41507208e-01 9.49446976e-01
2.95209080e-01 -2.74783492e-01 2.12528005e-01 9.85156417e-01
1.00207925e+00 5.91514349e-01 -6.35827541e-01 -2.73481011e-01
1.35516524e+00 -6.07895672e-01 -7.21397400e-01 2.51964152e-01
-1.66295897e-02 -1.25606191e+00 1.15737808e+00 6.63345575e-01
-1.55801070e+00 -9.40249562e-01 -1.09599364e+00 -3.01462829e-01
2.72345524e-02 -1.51148066e-01 -5.16166836e-02 9.07437742e-01
-1.35476291e+00 2.81861335e-01 -2.63418585e-01 -1.98467880e-01
2.67163038e-01 1.46175325e-01 -4.27359551e-01 -4.76479530e-01
-1.10447991e+00 8.78233731e-01 -5.74360453e-02 -5.33578694e-01
-8.06765974e-01 -5.01605868e-01 -8.56096506e-01 -3.54470074e-01
-1.63818672e-01 -6.37660801e-01 1.03042710e+00 -8.04101348e-01
-1.77900624e+00 1.27819455e+00 -4.24013674e-01 -2.53735278e-02
5.74449003e-01 1.90312073e-01 -7.21296966e-01 7.93139279e-01
1.88360840e-01 4.40361142e-01 7.99042881e-01 -1.40865064e+00
-7.61480510e-01 -4.21603084e-01 2.38451540e-01 6.79082155e-01
2.35596031e-01 2.25622267e-01 -9.38353598e-01 -4.10518527e-01
1.78484008e-01 -3.85251522e-01 1.74909249e-01 2.99436957e-01
-3.12041819e-01 3.74321073e-01 6.92149520e-01 -7.19849169e-01
1.24208462e+00 -2.47587776e+00 -3.44888151e-01 3.35249752e-02
1.08611465e-01 2.55878925e-01 -1.70051828e-01 3.88724923e-01
4.27071825e-02 -6.57176301e-02 5.25156818e-02 -2.87000716e-01
-3.67801368e-01 -1.04771473e-01 1.29484221e-01 5.71493983e-01
-3.21757495e-01 3.47264111e-01 -6.23278379e-01 -6.23869538e-01
7.06556976e-01 5.32195210e-01 -3.05056810e-01 3.95318896e-01
5.93490422e-01 5.97117007e-01 -4.00061570e-02 8.68588865e-01
1.26368785e+00 -3.09241917e-02 -4.07610774e-01 -3.76785249e-01
-2.66317725e-01 1.45346835e-01 -1.04859567e+00 1.45310116e+00
-5.12677968e-01 9.36092496e-01 1.75646078e-02 -4.52547967e-02
8.70523810e-01 4.89535540e-01 2.44017556e-01 -1.50070810e+00
-4.65826020e-02 4.03364748e-01 -2.36873075e-01 -1.02622652e+00
6.32704794e-01 8.77296999e-02 4.68662620e-01 1.89722985e-01
-2.61142582e-01 -7.77987540e-01 2.37043098e-01 1.63729787e-01
7.51415014e-01 -3.97828937e-01 2.25891426e-01 -1.57355368e-01
6.87925220e-01 -3.41514826e-01 1.92318499e-01 4.99004751e-01
-4.47359324e-01 1.18405628e+00 5.30545533e-01 -1.13911875e-01
-1.48100770e+00 -1.31616247e+00 -5.29987633e-01 2.55262852e-01
8.20729136e-01 -1.93437919e-01 -6.35088325e-01 2.68614497e-02
-2.80892193e-01 4.01608080e-01 -2.88395822e-01 2.05040857e-01
7.51077011e-02 -3.31885666e-01 3.39369506e-01 -1.46536961e-01
1.07992363e+00 -8.06644440e-01 -6.63555086e-01 -1.39797196e-01
-3.83325487e-01 -1.09990990e+00 -3.62467468e-01 -5.19577980e-01
-9.19084311e-01 -1.37831676e+00 -1.19046342e+00 -6.40748084e-01
4.89587903e-01 5.59629083e-01 1.12718832e+00 -6.60712197e-02
-3.57622087e-01 5.10285735e-01 -2.89985567e-01 7.11294962e-03
-4.12821621e-01 -7.90662229e-01 -5.72756268e-02 -1.29887074e-01
-2.03658026e-02 -5.77927411e-01 -1.24001884e+00 7.19923139e-01
-1.07573295e+00 2.07333773e-01 1.40791029e-01 4.60400343e-01
6.84351981e-01 5.51247716e-01 -1.60690576e-01 -2.97054648e-01
6.70741558e-01 -2.65126582e-02 -7.12114871e-01 -1.87904179e-01
-4.11367893e-01 -7.05902278e-01 3.10317189e-01 -8.48428234e-02
-1.17393136e+00 -6.77982211e-01 -3.04712385e-01 -2.87473202e-01
-2.33546793e-01 1.03943795e-01 -3.88223648e-01 -2.51432985e-01
6.06134653e-01 3.05613488e-01 2.10549012e-01 -4.65491563e-01
-2.90384918e-01 1.11450863e+00 7.93590724e-01 -2.30724216e-01
4.58733052e-01 7.86216736e-01 -9.31770951e-02 -9.80659008e-01
-2.41128668e-01 -5.19012451e-01 -1.45228311e-01 -4.43864942e-01
8.02352250e-01 -1.17302370e+00 -7.91451454e-01 1.20143759e+00
-9.32355583e-01 -3.34180057e-01 1.20870182e-02 5.49096227e-01
-7.49918163e-01 5.58170199e-01 -7.34138012e-01 -7.27125406e-01
-2.00287908e-01 -1.67609322e+00 1.20691264e+00 3.35878700e-01
1.35055408e-01 -7.85115302e-01 -1.38232559e-01 6.05818689e-01
5.41553199e-01 1.69040337e-01 9.01745617e-01 4.72379595e-01
-5.60647726e-01 -1.95200413e-01 -7.12502658e-01 6.62491739e-01
4.68675077e-01 -7.72414282e-02 -9.75416422e-01 -4.23459262e-01
4.36225712e-01 -2.05566227e-01 2.20881641e-01 6.19548678e-01
1.23513985e+00 -2.64330371e-03 2.81177640e-01 7.65311360e-01
1.80143487e+00 7.14098990e-01 1.34793162e+00 5.10014474e-01
1.95593998e-01 4.62379783e-01 7.91434944e-01 4.78344858e-01
3.26246768e-01 8.62304091e-01 4.99371678e-01 -3.43107194e-01
-3.58615160e-01 8.91537145e-02 2.56551623e-01 8.16454947e-01
-3.03829312e-01 -4.15781498e-01 -5.73214412e-01 1.51810363e-01
-8.59514832e-01 -8.62907290e-01 -4.31418180e-01 2.55368710e+00
7.16943979e-01 4.93965410e-02 -1.35097340e-01 4.52514499e-01
7.57031202e-01 1.36172339e-01 -5.07340372e-01 -3.55576038e-01
-3.70250195e-01 4.73361462e-02 4.24930513e-01 7.70340025e-01
-6.78332984e-01 1.90682933e-01 6.40376139e+00 1.08350992e+00
-1.19086456e+00 -9.17496253e-03 9.67786729e-01 -7.33208954e-02
-1.39262751e-01 -2.16009289e-01 -3.58475864e-01 1.01933968e+00
7.21098363e-01 -2.15482295e-01 3.62975985e-01 3.72541130e-01
7.26236820e-01 -1.12387657e+00 -6.06767476e-01 1.72312415e+00
8.55178684e-02 -1.00956130e+00 4.34325784e-02 3.07316124e-01
9.06437814e-01 -4.07436118e-02 4.75358844e-01 -5.08620977e-01
-2.41510808e-01 -7.74480224e-01 6.35973871e-01 4.69368219e-01
1.52599335e+00 -6.86989188e-01 8.46673787e-01 -7.94529095e-02
-9.24763978e-01 2.20134258e-01 -5.07260084e-01 1.46412611e-01
2.09910408e-01 7.06220925e-01 3.22634913e-02 6.75016940e-01
1.23240578e+00 5.26953220e-01 -6.67269766e-01 1.30418575e+00
4.94913124e-02 8.94463584e-02 -3.76266316e-02 5.05846858e-01
-1.19307593e-01 -2.39228576e-01 6.19462132e-01 7.47706711e-01
6.85272753e-01 2.39740714e-01 -5.97885072e-01 3.54340613e-01
3.61024171e-01 -1.07037969e-01 -5.86494088e-01 6.01241648e-01
4.53588277e-01 7.68601716e-01 -3.61233860e-01 -5.60439706e-01
-3.83790612e-01 1.36527705e+00 -6.62623584e-01 5.74437976e-01
-6.36231005e-01 -6.16632879e-01 8.33292663e-01 1.65049180e-01
2.15779897e-02 -5.88947795e-02 -2.02990398e-01 -1.04304016e+00
1.08264245e-01 -1.14443099e+00 2.62068752e-02 -1.63615000e+00
-1.19752431e+00 6.63308740e-01 4.25082371e-02 -1.72463000e+00
2.48387866e-02 -5.51930368e-01 -3.43613595e-01 1.22536075e+00
-1.58128881e+00 -1.83636725e-01 -7.73688138e-01 8.69305134e-01
4.37455505e-01 -1.57406420e-01 6.38645887e-01 5.63841999e-01
-9.46678072e-02 3.52277875e-01 5.15061915e-01 -3.05288345e-01
8.76318872e-01 -9.67727482e-01 1.90472528e-01 4.61307228e-01
-7.92189360e-01 1.44488528e-01 6.22964680e-01 -3.47160459e-01
-1.11598313e+00 -6.63731158e-01 6.83736145e-01 -1.36262849e-01
9.25224945e-02 3.17222401e-02 -6.95751905e-01 -1.67781204e-01
4.42556113e-01 -6.12540916e-02 4.74393398e-01 -8.07291985e-01
-2.40372983e-03 -2.75339216e-01 -1.66946077e+00 5.74182630e-01
7.90177464e-01 -7.94452727e-01 1.99718252e-02 1.09129734e-02
5.06725073e-01 -7.91339397e-01 -1.13801122e+00 3.06178838e-01
6.01746142e-01 -1.92743289e+00 1.03932405e+00 6.57266796e-01
7.65947282e-01 -4.68797505e-01 -5.18414676e-01 -1.12871146e+00
1.42382130e-01 -4.60880548e-01 4.24901843e-01 9.90815520e-01
-7.41693154e-02 -8.53344679e-01 2.88916916e-01 3.44995767e-01
-5.76801365e-03 -4.22110885e-01 -1.02898514e+00 -8.29437435e-01
-3.48651648e-01 -3.38169336e-01 5.47115266e-01 4.97688025e-01
-1.99287772e-01 -1.12020753e-01 -1.76525235e-01 1.48809329e-01
8.68052423e-01 -8.77516642e-02 8.09270740e-01 -5.80920458e-01
-1.55356720e-01 -2.18909472e-01 -8.09054077e-01 -1.21978796e+00
-8.15410912e-01 -1.48577556e-01 -4.36820656e-01 -1.43224955e+00
1.49456367e-01 -1.20564155e-01 1.37308091e-01 -3.85434687e-01
-3.80187929e-02 7.41473675e-01 9.81254727e-02 4.63895977e-01
-3.11606377e-01 4.13178712e-01 1.71479559e+00 2.88466960e-02
-3.77586968e-02 -9.09937546e-03 -2.08638728e-01 4.78800297e-01
6.69629872e-01 -6.71829656e-02 -5.10305166e-01 -5.19941747e-01
-6.41111881e-02 5.86504281e-01 4.32163090e-01 -1.16404867e+00
3.68087292e-02 9.51461196e-02 4.40512449e-01 -7.27399886e-01
7.51581132e-01 -8.44028950e-01 2.22880378e-01 3.52920979e-01
9.05275345e-02 1.34845957e-01 8.55497122e-02 2.44551137e-01
-7.52776980e-01 -1.67306215e-01 1.20551801e+00 -8.53861272e-02
-7.27085471e-01 1.50084943e-01 -6.14486396e-01 -1.08713716e-01
8.77210736e-01 -8.25226426e-01 -5.05721450e-01 -7.97392070e-01
-3.33072305e-01 -1.73182815e-01 1.08160353e+00 -3.45070078e-03
1.01884747e+00 -1.41906261e+00 -5.36241591e-01 5.61186492e-01
2.74798334e-01 -2.08367646e-01 7.44886816e-01 8.73503029e-01
-1.19611061e+00 1.75315365e-01 -5.49530983e-01 -7.81202853e-01
-1.62008059e+00 2.15694487e-01 3.63082290e-01 9.47692543e-02
-4.47490841e-01 6.20396435e-01 3.70538205e-01 1.38561875e-01
2.21633628e-01 -1.68936834e-01 -3.85885835e-02 -3.47972900e-01
7.84201205e-01 6.74016654e-01 1.06015705e-01 -7.47494340e-01
-1.44494355e-01 8.36522102e-01 3.48911554e-01 -5.32701671e-01
9.66804028e-01 -9.68974948e-01 -1.53674990e-01 3.46172869e-01
1.61829996e+00 2.47704815e-02 -1.20012617e+00 1.63330622e-02
-5.57153940e-01 -1.29438758e+00 3.57216485e-02 -7.95683920e-01
-1.00064027e+00 9.52033520e-01 1.28998041e+00 3.47994179e-01
1.78633249e+00 -2.76455104e-01 6.93836927e-01 -3.69262338e-01
6.00662887e-01 -8.46985996e-01 2.30812684e-01 3.97030056e-01
9.12207961e-01 -1.36096656e+00 1.42236520e-02 -4.52946991e-01
-6.45008266e-01 8.94187987e-01 3.39295954e-01 7.54147321e-02
4.33111578e-01 2.17390031e-01 5.20852804e-01 -6.99945837e-02
-5.62835991e-01 4.67603887e-03 1.65928289e-01 8.81115973e-01
4.17015404e-01 -3.14076304e-01 -1.52953699e-01 -2.04284176e-01
-1.43022344e-01 -4.14104853e-03 6.57120407e-01 7.91230321e-01
-2.46924773e-01 -8.22672248e-01 -7.89362550e-01 2.54013360e-01
-5.62334239e-01 -1.97915286e-01 6.72743917e-02 4.96142268e-01
2.55851358e-01 1.32117093e+00 2.39174277e-01 -2.98128605e-01
3.68296504e-01 -6.56215191e-01 5.03455400e-01 -1.08001135e-01
-2.77365744e-01 1.15891986e-01 -5.44822700e-02 -8.59401107e-01
-5.13795376e-01 -2.38121971e-01 -7.16557503e-01 -7.13416278e-01
-4.13046107e-02 -1.60010353e-01 8.49043250e-01 1.59878686e-01
2.86525160e-01 2.09796727e-01 8.86215150e-01 -1.02619863e+00
2.22777069e-01 -7.76157141e-01 -9.92500246e-01 4.48715776e-01
6.16461754e-01 -4.06256378e-01 -6.64684176e-01 2.61584222e-01] | [11.75365924835205, -1.9155763387680054] |
538ed936-cd74-46a7-96c9-c0a178684658 | an-extension-to-hough-transform-based-on | 1510.04863 | null | http://arxiv.org/abs/1510.04863v1 | http://arxiv.org/pdf/1510.04863v1.pdf | An Extension to Hough Transform Based on Gradient Orientation | The Hough transform is one of the most common methods for line detection. In
this paper we propose a novel extension of the regular Hough transform. The
proposed extension combines the extension of the accumulator space and the
local gradient orientation resulting in clutter reduction and yielding more
prominent peaks, thus enabling better line identification. We demonstrate
benefits in applications such as visual quality inspection and rectangle
detection. | ['Sven Lončarić', 'Tomislav Petković'] | 2015-10-16 | null | null | null | null | ['line-detection'] | ['computer-vision'] | [-9.38268676e-02 -4.33897167e-01 -8.66452511e-03 3.04524660e-01
-3.46932381e-01 -4.41858232e-01 2.23626062e-01 3.67442459e-01
-6.37768209e-01 7.46397138e-01 -2.91597575e-01 -2.97509909e-01
1.80550441e-02 -9.19138610e-01 -3.02161545e-01 -5.44687331e-01
-2.52575219e-01 -2.59082377e-01 9.05849159e-01 -2.67371505e-01
4.96476829e-01 1.02738273e+00 -1.18855417e+00 -9.19296965e-02
1.00407004e+00 8.95179093e-01 -1.13950677e-01 6.97578013e-01
1.20723471e-01 1.90344125e-01 -7.66528368e-01 -8.84220079e-02
2.98836470e-01 -3.90772998e-01 -4.01559733e-02 4.00119603e-01
1.57780156e-01 -2.89860845e-01 -1.89274728e-01 9.46635783e-01
4.68281776e-01 1.55538335e-01 5.81904471e-01 -9.96454120e-01
-1.81564569e-01 1.93939164e-01 -1.32184982e+00 3.87654871e-01
6.24791682e-01 -3.19933593e-01 4.27358091e-01 -1.07593894e+00
3.83153558e-01 7.48073161e-01 8.53958309e-01 -3.10810149e-01
-1.09004831e+00 -3.70624870e-01 -6.33698285e-01 4.39305931e-01
-1.81859767e+00 -2.07823515e-01 8.12049448e-01 -1.23908944e-01
6.40571594e-01 4.65334237e-01 6.73298597e-01 -5.04063033e-02
3.20199847e-01 7.99948037e-01 1.05826139e+00 -1.01138687e+00
2.53571123e-02 3.48887257e-02 -5.74594475e-02 7.35391200e-01
5.69746137e-01 9.02544558e-02 -2.96166599e-01 -1.18076645e-01
1.24965084e+00 -1.60507828e-01 -5.10926843e-01 -6.51782811e-01
-1.15791881e+00 7.28983164e-01 4.82930213e-01 4.98217642e-01
-3.85895103e-01 -1.66725725e-01 2.20205307e-01 -2.62489378e-01
5.23951128e-02 3.24376971e-01 4.65726227e-01 -1.87052950e-01
-1.17512155e+00 1.31682903e-01 4.80596691e-01 7.82142758e-01
4.08916026e-01 1.70482367e-01 -1.23509899e-01 6.20010793e-01
3.34167480e-01 5.48273742e-01 2.94549614e-01 -3.57796729e-01
3.83447528e-01 4.93163764e-01 2.95599461e-01 -9.64071929e-01
-6.68019235e-01 -5.26872635e-01 -5.14489710e-01 5.10070086e-01
5.73475003e-01 -1.66286945e-01 -5.46045721e-01 6.92883611e-01
2.46523544e-01 3.37637588e-02 -4.62106653e-02 7.28256106e-01
3.45635206e-01 5.46236515e-01 -4.46457058e-01 -1.90645188e-01
1.44480491e+00 -7.87534714e-01 -8.20057273e-01 2.92469910e-03
4.07985300e-01 -1.25781238e+00 6.79164767e-01 6.12597525e-01
-9.54741061e-01 -5.41680336e-01 -1.63590920e+00 1.24622121e-01
-3.83256376e-01 4.36685413e-01 3.36722553e-01 9.24463391e-01
-6.33489668e-01 3.40682805e-01 -4.79633719e-01 -3.72368395e-01
-8.12937915e-02 1.71713591e-01 -4.47119683e-01 2.33281165e-01
-6.86922610e-01 9.28700447e-01 6.18633211e-01 2.25828856e-01
5.96250772e-01 -7.82529488e-02 -9.29150701e-01 1.53258204e-01
3.78507853e-01 5.53119071e-02 9.61342275e-01 2.16590464e-01
-1.29686403e+00 5.83099544e-01 -1.11276381e-01 -6.63302779e-01
6.99133754e-01 -4.38730299e-01 -6.39382005e-01 2.92545557e-01
-1.09979384e-01 1.00106686e-01 7.29431808e-01 -6.97411060e-01
-8.41383040e-01 -2.70150393e-01 -6.45561337e-01 1.61220178e-01
-1.22072712e-01 -6.63376227e-02 -3.30029428e-01 -6.99113846e-01
2.94399410e-01 -5.18272161e-01 -1.09169401e-01 -2.54051685e-01
-3.94160688e-01 -1.34195432e-01 1.09482586e+00 -5.87980509e-01
1.63420749e+00 -2.39521050e+00 -7.25296259e-01 7.05387175e-01
1.96857437e-01 3.05758506e-01 3.39833051e-01 5.18443406e-01
-6.08130470e-02 -3.11264813e-01 1.37450114e-01 3.61529082e-01
-3.09184790e-01 -2.16224477e-01 -6.67014420e-02 8.09309959e-01
1.53468356e-01 5.80392182e-01 -6.44598424e-01 -6.52021348e-01
5.69220364e-01 3.31350923e-01 1.74345925e-01 -2.11219281e-01
8.18792999e-01 -1.05090506e-01 -1.46472722e-01 4.43889678e-01
9.16082919e-01 2.15376049e-01 -2.95738697e-01 -4.23588187e-01
-6.51718020e-01 -1.91787273e-01 -1.55990791e+00 8.07154655e-01
9.04253200e-02 1.00860763e+00 -4.26882774e-01 -5.25270522e-01
1.57375824e+00 1.81768388e-01 4.48913574e-01 -9.34662819e-01
8.33785832e-02 2.50237942e-01 2.92758569e-02 -1.61698371e-01
8.76513779e-01 -1.34076523e-02 5.79451546e-02 5.18113840e-03
-4.01252419e-01 -2.65374243e-01 5.69288611e-01 -1.32835582e-01
4.79132712e-01 1.45131804e-03 1.25362420e+00 -2.42985114e-01
6.38938308e-01 -6.70900196e-02 2.58253276e-01 5.02501190e-01
-4.10801917e-01 4.92283762e-01 2.79982597e-01 -2.66731739e-01
-1.19648314e+00 -1.07777214e+00 -4.75673199e-01 4.34054941e-01
3.66392940e-01 -2.76622027e-01 -5.67817986e-01 -2.06516981e-01
9.59874094e-02 6.62668169e-01 -2.78754175e-01 2.95272917e-02
-7.93811202e-01 -4.86730337e-01 5.13214231e-01 7.12077022e-01
8.38781774e-01 -5.93059659e-01 -1.10240149e+00 3.21863800e-01
1.59133062e-01 -1.08960235e+00 -6.00143313e-01 1.27465695e-01
-9.99943256e-01 -1.08503652e+00 -9.54726756e-01 -7.47416854e-01
5.26983976e-01 6.98872149e-01 6.55245066e-01 -5.38168438e-02
-6.06853783e-01 1.87304258e-01 -3.13774049e-01 -4.55822468e-01
-8.64412710e-02 -3.56856674e-01 -2.91874796e-01 -1.20499119e-01
4.43492204e-01 8.41070488e-02 -5.86871266e-01 5.52861035e-01
-5.03973246e-01 -2.95336455e-01 5.91175675e-01 8.21835935e-01
2.74435401e-01 3.21102738e-01 2.85612866e-02 -5.06881297e-01
9.31751847e-01 2.66889930e-01 -1.06149292e+00 1.32995099e-01
-5.81535161e-01 1.07729863e-02 4.06683654e-01 -1.29060313e-01
-8.83238971e-01 6.39269203e-02 -5.07210232e-02 2.31359869e-01
-6.23561367e-02 2.05736846e-01 1.70623034e-01 -7.10077405e-01
9.21149790e-01 3.23689014e-01 -2.09839359e-01 -2.30419770e-01
1.96158975e-01 4.26571578e-01 9.38343823e-01 -1.38825953e-01
9.27332580e-01 3.61032963e-01 3.79545897e-01 -1.49742973e+00
1.44030645e-01 -9.07503963e-01 -7.64510155e-01 -4.41573411e-01
5.51020324e-01 -3.67657453e-01 -7.36906767e-01 2.72264332e-01
-1.12298322e+00 3.81743163e-01 -3.88292730e-01 8.13334405e-01
-4.56771225e-01 9.17545021e-01 -5.32635689e-01 -1.18293059e+00
-3.68663847e-01 -8.11643183e-01 7.24064887e-01 8.89741361e-01
-1.28966169e-02 -7.76470184e-01 1.39229791e-02 -4.43354279e-01
1.26333952e-01 4.77305442e-01 5.46682775e-01 -4.68609750e-01
-1.97768062e-01 -9.46807086e-01 -4.27153558e-01 -1.22215495e-01
2.16729920e-02 2.43495703e-01 -6.31812930e-01 -1.00569934e-01
-1.58916295e-01 4.34486181e-01 6.42070770e-01 3.61260593e-01
4.71561641e-01 3.13581049e-01 -4.75394309e-01 3.82450670e-01
1.52041411e+00 7.62365282e-01 1.20231807e+00 7.64338970e-01
1.31514385e-01 1.31792650e-01 9.76639211e-01 5.79102039e-01
-1.53587222e-01 8.68802905e-01 -1.84316099e-01 -6.57525361e-01
9.66939554e-02 -1.74899057e-01 -7.92391822e-02 4.52057570e-01
-4.04243886e-01 6.18648203e-03 -8.39393735e-01 3.68160784e-01
-1.56487858e+00 -9.72155035e-01 -6.13665760e-01 2.63492846e+00
3.01442087e-01 4.62351978e-01 4.29840028e-01 7.43613183e-01
8.47914696e-01 -2.06757858e-01 7.33578354e-02 -5.32799482e-01
-3.45400631e-01 2.53249586e-01 1.12233043e+00 4.31848764e-01
-1.17787218e+00 5.10706663e-01 7.33012056e+00 7.87103653e-01
-1.11975658e+00 -4.74017113e-01 1.85946014e-03 8.02695096e-01
4.10720587e-01 -1.31433174e-01 -8.71340036e-01 2.38632306e-01
1.10047780e-01 -5.45043588e-01 -2.94551730e-01 6.56776965e-01
1.98339567e-01 -7.82328308e-01 -5.01930237e-01 1.23547113e+00
8.21053386e-02 -8.21121812e-01 -3.94375622e-01 1.34308442e-01
2.32664600e-01 -7.34340847e-01 -3.24164927e-02 -1.73787534e-01
-3.29886466e-01 -5.70515633e-01 5.03389299e-01 2.84302443e-01
5.02214313e-01 -8.61601830e-01 7.86424279e-01 -5.30381463e-02
-1.47781157e+00 1.13035180e-01 -4.30462122e-01 1.25048041e-01
3.76905233e-01 5.79618812e-01 -1.50681829e+00 5.37698627e-01
2.83160478e-01 6.95831552e-02 -6.67340577e-01 2.18943095e+00
-2.05441743e-01 3.54608864e-01 -6.13620818e-01 -1.05979137e-01
1.28111050e-01 -5.05829453e-01 7.10240901e-01 1.54422355e+00
6.77278042e-01 -1.42525835e-02 2.41247758e-01 5.87579370e-01
6.02605999e-01 6.08860195e-01 -6.09684706e-01 1.93445534e-01
4.80304331e-01 1.10137594e+00 -1.22129202e+00 -2.01722756e-01
-5.90677559e-01 9.65851009e-01 -2.77191877e-01 2.29478151e-01
-6.39996469e-01 -1.14257729e+00 -2.06975251e-01 2.28858650e-01
5.00965059e-01 -6.52964652e-01 -5.05914152e-01 -6.73325002e-01
-9.80157703e-02 -4.18021381e-01 5.79221666e-01 -6.03869975e-01
-4.81851906e-01 3.93169165e-01 -3.72535214e-02 -1.64818823e+00
-3.23996574e-01 -7.90171921e-01 -8.05275083e-01 9.52474236e-01
-1.15989864e+00 -8.00827801e-01 -3.89013678e-01 5.17085075e-01
3.60895604e-01 6.86225994e-03 4.93090898e-01 2.41954848e-01
-3.85690391e-01 7.40727425e-01 1.85206294e-01 2.27565110e-01
7.67608941e-01 -1.38530946e+00 4.38599467e-01 1.24068129e+00
1.05780713e-01 4.10335630e-01 1.01094425e+00 -6.46921933e-01
-6.91340625e-01 -1.74261495e-01 5.77203333e-01 4.40926969e-01
5.34504950e-01 -4.89525832e-02 -8.65537286e-01 2.12801248e-01
6.56443983e-02 -1.44920915e-01 6.70809329e-01 -2.65059054e-01
8.08547251e-03 -1.57809570e-01 -1.17021418e+00 6.44599736e-01
2.29302526e-01 -2.44233280e-01 -4.38141197e-01 -1.53920978e-01
-5.48506528e-02 -4.75795537e-01 -8.93483877e-01 3.57633591e-01
7.51863599e-01 -1.26640165e+00 1.06658721e+00 2.52071619e-01
-3.97634029e-01 -7.35021770e-01 2.83622056e-01 -1.16300654e+00
-7.07289159e-01 -5.68826735e-01 2.63483286e-01 9.69570637e-01
3.26457769e-01 -7.14403987e-01 8.12073410e-01 2.07197487e-01
6.03108108e-02 -3.17946523e-01 -7.84811378e-01 -7.98748612e-01
-7.03132033e-01 5.37042804e-02 2.99332887e-01 3.82649690e-01
6.31330371e-01 1.78590238e-01 -4.20098513e-01 2.89531369e-02
8.82140517e-01 -5.97921014e-02 7.24744618e-01 -1.11421049e+00
-2.16409191e-01 -5.93101680e-01 -9.17520642e-01 -1.10043180e+00
-8.66266131e-01 -1.49285078e-01 8.83435011e-02 -1.35198843e+00
-1.94019943e-01 3.21549997e-02 -3.55896764e-02 6.98562860e-02
-3.14360440e-01 5.72078288e-01 2.65322298e-01 -5.95107935e-02
-2.37644985e-01 9.40743461e-02 1.11990082e+00 2.35612676e-01
-4.34869528e-01 1.61740884e-01 -1.27571672e-01 6.90197110e-01
8.81525695e-01 6.59570843e-02 4.91063185e-02 1.12786502e-01
-1.64858311e-01 -4.45968695e-02 3.14800054e-01 -1.31313062e+00
4.37549204e-01 7.10378885e-02 1.03719783e+00 -1.08549511e+00
1.55532286e-01 -7.80260503e-01 -2.33776554e-01 8.24154973e-01
2.80042022e-01 3.96812230e-01 4.74691927e-01 2.53655076e-01
-3.38941783e-01 -4.78817821e-01 8.55457008e-01 2.16700718e-01
-1.21576333e+00 -4.55052018e-01 -3.10454577e-01 -6.39082313e-01
1.39156425e+00 -9.62211490e-01 -4.76924598e-01 -3.31161857e-01
-4.44562167e-01 -1.37586430e-01 4.78162915e-01 3.62716690e-02
9.38410223e-01 -1.36673403e+00 -6.86981678e-01 4.45269078e-01
2.09828183e-01 -5.78670382e-01 -8.83890223e-03 1.07937467e+00
-1.20427251e+00 6.62937760e-01 -5.71927607e-01 -5.35635352e-01
-1.48982227e+00 5.77175200e-01 2.56578654e-01 -1.90795854e-01
-9.29819524e-01 4.74074334e-01 -6.62796944e-02 7.04361618e-01
-6.05148897e-02 -2.23876581e-01 -5.93181193e-01 -1.38483629e-01
7.73191333e-01 1.14238322e+00 8.35231394e-02 -5.48464179e-01
-2.76372761e-01 9.57708478e-01 1.18193172e-01 -1.19718574e-01
6.39525771e-01 -1.73218220e-01 8.30829963e-02 2.68073887e-01
6.65014803e-01 8.20921957e-01 -8.81592095e-01 7.32773868e-03
3.48903179e-01 -8.99446130e-01 -1.44356430e-01 -3.04861754e-01
-3.59307617e-01 7.51376033e-01 8.98719907e-01 6.84274614e-01
1.16112614e+00 -4.52411056e-01 6.67360961e-01 -3.76474150e-02
3.97718489e-01 -1.08684278e+00 -3.39019299e-01 6.64040595e-02
7.22029507e-01 -7.54459262e-01 2.73558944e-01 -8.63342106e-01
-3.04256439e-01 1.71607172e+00 4.27746117e-01 -2.77538925e-01
1.63254619e-01 6.04468346e-01 3.33124511e-02 9.66939554e-02
1.11366391e-01 -5.28143585e-01 5.23814738e-01 7.49298871e-01
6.84751570e-01 1.88093066e-01 -9.95293260e-01 1.37676552e-01
4.34344932e-02 1.42851323e-02 7.26635695e-01 1.10691237e+00
-9.31067526e-01 -9.69587564e-01 -1.25495207e+00 3.23861986e-01
-1.91099703e-01 1.83291048e-01 -1.77920997e-01 9.84570920e-01
-6.85960576e-02 8.87624979e-01 5.39647602e-03 -1.26510024e-01
8.50339651e-01 -1.34108230e-01 7.20334947e-01 2.96291262e-02
5.22013083e-02 6.66083813e-01 3.97782959e-02 -2.74007618e-01
-7.17031658e-02 -4.92201537e-01 -1.40301979e+00 -1.52767405e-01
-6.73811138e-01 4.56692517e-01 5.76048732e-01 5.41733742e-01
-2.01087758e-01 3.33646476e-01 4.93140936e-01 -5.12971520e-01
-3.90662074e-01 -7.24450231e-01 -1.09757447e+00 1.57124102e-01
2.21973106e-01 -7.74341285e-01 2.23356206e-02 -1.02551088e-01] | [8.480840682983398, -1.615005373954773] |
d83fef4c-eea0-49b0-bd7a-2229eb981da2 | background-click-supervision-for-temporal | 2111.12449 | null | https://arxiv.org/abs/2111.12449v1 | https://arxiv.org/pdf/2111.12449v1.pdf | Background-Click Supervision for Temporal Action Localization | Weakly supervised temporal action localization aims at learning the instance-level action pattern from the video-level labels, where a significant challenge is action-context confusion. To overcome this challenge, one recent work builds an action-click supervision framework. It requires similar annotation costs but can steadily improve the localization performance when compared to the conventional weakly supervised methods. In this paper, by revealing that the performance bottleneck of the existing approaches mainly comes from the background errors, we find that a stronger action localizer can be trained with labels on the background video frames rather than those on the action frames. To this end, we convert the action-click supervision to the background-click supervision and develop a novel method, called BackTAL. Specifically, BackTAL implements two-fold modeling on the background video frames, i.e. the position modeling and the feature modeling. In position modeling, we not only conduct supervised learning on the annotated video frames but also design a score separation module to enlarge the score differences between the potential action frames and backgrounds. In feature modeling, we propose an affinity module to measure frame-specific similarities among neighboring frames and dynamically attend to informative neighbors when calculating temporal convolution. Extensive experiments on three benchmarks are conducted, which demonstrate the high performance of the established BackTAL and the rationality of the proposed background-click supervision. Code is available at https://github.com/VividLe/BackTAL. | ['Jianxin Chen', 'Dingwen Zhang', 'Tianwei Lin', 'Tao Zhao', 'Junwei Han', 'Le Yang'] | 2021-11-24 | null | null | null | null | ['weakly-supervised-temporal-action', 'action-localization'] | ['computer-vision', 'computer-vision'] | [ 3.79270494e-01 -3.42973620e-01 -4.46538180e-01 -4.59380209e-01
-7.85836041e-01 -2.49397680e-01 4.21453059e-01 -2.74536639e-01
-4.65954870e-01 5.78445256e-01 3.06536704e-01 8.97695497e-02
2.90879220e-01 -2.45392695e-01 -6.89663827e-01 -1.02686608e+00
6.68707564e-02 -1.25002757e-01 9.14985418e-01 2.28718311e-01
1.74325481e-01 1.02531500e-02 -1.49682403e+00 7.95338273e-01
5.31513453e-01 1.12755823e+00 3.55273306e-01 4.63081032e-01
-1.35637745e-01 1.42989206e+00 -4.36889052e-01 1.30362242e-01
1.61565587e-01 -6.50314450e-01 -6.37697279e-01 2.20047817e-01
4.60920423e-01 -5.98347247e-01 -3.42684716e-01 1.10111749e+00
4.21175689e-01 2.94580430e-01 7.37207457e-02 -1.33718431e+00
-2.78311729e-01 4.65505540e-01 -8.19956183e-01 6.70939028e-01
3.84711534e-01 3.21332484e-01 8.60493183e-01 -1.24943447e+00
4.62816149e-01 1.18462527e+00 6.14438355e-01 6.15257084e-01
-8.39445531e-01 -6.94581330e-01 7.32253075e-01 6.54217422e-01
-1.30542588e+00 -5.42306542e-01 8.17625225e-01 -4.72258836e-01
5.38870454e-01 1.90280989e-01 4.91421789e-01 1.23956060e+00
-1.97827205e-01 1.23272789e+00 1.02717578e+00 -2.38908038e-01
2.84630507e-01 -1.57440990e-01 9.49685201e-02 8.18506300e-01
-3.62853199e-01 -9.59989205e-02 -7.38450885e-01 -4.48577218e-02
8.32349539e-01 3.66170913e-01 -4.88121629e-01 -3.05898160e-01
-1.36531460e+00 4.07688707e-01 2.74736822e-01 4.99468923e-01
-1.88867122e-01 4.54012781e-01 4.96227235e-01 -1.26446888e-01
5.43374896e-01 -3.00485373e-01 -5.11764169e-01 -4.34111208e-01
-9.05578315e-01 -3.22420239e-01 3.76326501e-01 8.37972283e-01
7.52481699e-01 -2.01604143e-01 -5.50720513e-01 5.48938751e-01
3.03752959e-01 1.96358830e-01 5.77855110e-01 -1.11123455e+00
5.73983490e-01 5.43611944e-01 1.22759439e-01 -8.20176005e-01
-7.43369088e-02 -3.64947110e-01 -4.22791928e-01 1.40073150e-01
4.69773829e-01 -1.03549987e-01 -8.96041751e-01 1.76514161e+00
5.02733052e-01 8.91838133e-01 -2.11958140e-01 9.88138139e-01
6.53576434e-01 5.92679977e-01 3.55963081e-01 -3.34862232e-01
1.00790107e+00 -1.59093249e+00 -9.51245487e-01 -1.92719907e-01
8.76716495e-01 -6.60150588e-01 1.17989075e+00 9.26575288e-02
-9.50207472e-01 -8.54256272e-01 -8.30075085e-01 6.48158640e-02
-1.61100477e-01 7.22828746e-01 5.97638428e-01 1.94698930e-01
-7.83035994e-01 5.87195039e-01 -1.28849268e+00 -3.02081257e-01
5.63343048e-01 1.76012963e-01 -2.89405465e-01 -7.68685760e-03
-1.08232164e+00 4.27580386e-01 3.27412546e-01 2.51072586e-01
-1.24717963e+00 -5.04317403e-01 -8.01628470e-01 -6.17427081e-02
7.82267094e-01 -1.76143348e-01 1.24363267e+00 -1.50881648e+00
-1.45687854e+00 5.72123289e-01 -2.86424786e-01 -2.55818665e-01
6.83378339e-01 -3.79687607e-01 -2.38003075e-01 3.77452224e-01
2.47392833e-01 5.15908957e-01 7.93652117e-01 -9.37671363e-01
-1.05061948e+00 -1.80862829e-01 3.44175249e-01 3.05533975e-01
-4.25889701e-01 2.40135431e-01 -9.68913257e-01 -7.09779322e-01
3.12360544e-02 -8.17535639e-01 -1.70869723e-01 1.82708904e-01
-2.83133797e-02 -3.94773573e-01 1.00716138e+00 -6.43384516e-01
1.34397328e+00 -2.50896287e+00 -1.60827711e-01 -1.23080418e-01
9.37488526e-02 2.91251510e-01 -7.39966705e-02 7.63561651e-02
-2.68946469e-01 -2.01293454e-01 -4.72487696e-02 -3.48330557e-01
-2.23174557e-01 1.37049168e-01 -2.21903180e-03 5.13576269e-01
1.36944905e-01 7.19246805e-01 -1.13424921e+00 -7.26094365e-01
2.96420366e-01 2.85605401e-01 -5.06299078e-01 1.33181751e-01
-1.66005030e-01 5.85434794e-01 -6.27230525e-01 8.29956234e-01
5.03133416e-01 -5.34017861e-01 9.26400535e-03 -3.70038271e-01
-1.79866597e-01 1.81739613e-01 -1.28251755e+00 1.96541977e+00
-1.16332702e-01 4.76637632e-01 -1.70210730e-02 -1.05138922e+00
4.35293406e-01 2.30810791e-01 7.02901900e-01 -4.79069203e-01
3.51755060e-02 2.41667032e-02 -1.15522064e-01 -6.77858174e-01
-4.21004593e-02 1.96339354e-01 2.87572473e-01 1.81610271e-01
2.28836596e-01 7.56851494e-01 2.39466161e-01 2.84656316e-01
1.34028375e+00 7.40888774e-01 5.56373559e-02 -2.93877404e-02
9.08931494e-01 -1.60895884e-01 1.06110883e+00 5.54479718e-01
-6.52609229e-01 4.82255459e-01 4.85127389e-01 -3.20339650e-01
-4.28661644e-01 -8.38293016e-01 1.78992212e-01 1.41424561e+00
4.57943797e-01 -6.08449340e-01 -8.81297708e-01 -1.17167509e+00
-3.87698352e-01 2.56134003e-01 -8.17115426e-01 -2.30650708e-01
-7.47067034e-01 -5.16774595e-01 3.42613876e-01 9.31004763e-01
8.14758420e-01 -9.16454852e-01 -3.83362055e-01 -7.26581812e-02
-4.40606922e-01 -1.22312236e+00 -8.44859183e-01 2.13103130e-01
-7.52815485e-01 -1.04653275e+00 -6.65198922e-01 -8.74908566e-01
7.08844543e-01 4.47759628e-01 6.64351940e-01 1.18966751e-01
-7.06214011e-02 4.38544601e-01 -5.85498214e-01 8.02609921e-02
7.26780146e-02 -2.50048131e-01 4.36607227e-02 4.68039989e-01
5.54162860e-01 -4.50323045e-01 -8.76751184e-01 7.20012069e-01
-7.05607831e-01 1.57438561e-01 7.87866354e-01 7.78849602e-01
9.35389578e-01 4.31404039e-02 3.56733084e-01 -6.16705894e-01
-7.49225914e-02 -3.08746010e-01 -4.11109895e-01 3.68054330e-01
-1.39256537e-01 -1.98820189e-01 4.11500573e-01 -6.41784608e-01
-1.19000566e+00 5.37521899e-01 2.99300216e-02 -8.44260275e-01
-2.21425489e-01 2.62708634e-01 -3.32024693e-01 -5.48069365e-02
3.47253919e-01 3.35652053e-01 -2.84944266e-01 -5.37149787e-01
2.53503948e-01 4.78297293e-01 5.47772586e-01 -5.03639877e-01
6.75371587e-01 6.95140839e-01 -3.90673012e-01 -4.83578652e-01
-1.15527868e+00 -7.81403959e-01 -7.50843704e-01 -5.29509127e-01
1.17284632e+00 -1.05636513e+00 -4.65367466e-01 3.89096916e-01
-9.30015922e-01 -5.69931805e-01 -3.11391205e-01 7.12338746e-01
-5.39667845e-01 6.12601221e-01 -5.23208201e-01 -7.06386387e-01
1.56107709e-01 -1.10744917e+00 1.14354444e+00 2.65096933e-01
1.07018463e-01 -9.00032282e-01 1.21733844e-02 4.34131175e-01
4.13176790e-02 1.35092333e-01 3.09376746e-01 -6.73403323e-01
-7.64930546e-01 -1.71267852e-01 -2.44607612e-01 5.65048575e-01
3.23626965e-01 -1.76953450e-01 -1.00976264e+00 -1.39109612e-01
9.16209593e-02 -3.27224880e-01 9.81861234e-01 3.09510857e-01
1.28589535e+00 8.97401385e-03 -4.25133288e-01 5.53919435e-01
1.08176458e+00 3.34607333e-01 4.88078177e-01 2.34266475e-01
8.21621954e-01 3.22059780e-01 1.17088211e+00 4.39943522e-01
1.01259306e-01 8.63225758e-01 2.91501850e-01 -2.00793713e-01
-3.17693949e-01 -3.56317073e-01 8.80023062e-01 6.62879825e-01
-2.90887564e-01 1.39890179e-01 -6.30169868e-01 3.27533454e-01
-2.27764320e+00 -1.20944667e+00 -5.59943952e-02 2.04693913e+00
8.94290984e-01 3.13346744e-01 1.05476804e-01 1.19825583e-02
9.51943219e-01 2.71452904e-01 -6.61181867e-01 4.65895981e-01
3.95279669e-04 -1.04739465e-01 3.13289255e-01 4.00639564e-01
-1.53560030e+00 1.06278980e+00 5.29404783e+00 1.02634454e+00
-9.52295899e-01 4.39031065e-01 6.76810324e-01 -3.83939475e-01
3.67543459e-01 1.85149506e-01 -9.22836423e-01 7.54140973e-01
5.08468449e-01 1.73460841e-01 1.07955761e-01 1.12943876e+00
6.18518531e-01 -1.90845594e-01 -1.25166655e+00 1.07853377e+00
2.33550921e-01 -1.17969406e+00 -1.13026641e-01 -3.27851683e-01
5.82158566e-01 -1.16095319e-01 -1.77719250e-01 3.63356680e-01
-1.02860726e-01 -5.37751615e-01 8.35293174e-01 5.31508386e-01
4.83760327e-01 -3.76883030e-01 7.51322091e-01 3.52073401e-01
-1.50228596e+00 -2.41194531e-01 -2.53487259e-01 -9.94005501e-02
8.81681666e-02 3.64195108e-01 -2.54906416e-01 3.25521439e-01
9.13461566e-01 1.22101986e+00 -6.18324995e-01 1.03269196e+00
-4.80468005e-01 7.92308450e-01 -1.21075183e-01 2.07504839e-01
3.60873729e-01 -5.93042225e-02 2.55007237e-01 1.20255029e+00
5.07707261e-02 1.76791221e-01 6.99324906e-01 6.26401782e-01
1.30937085e-01 3.02727502e-02 -1.05692446e-01 4.46421728e-02
2.09613860e-01 1.27322340e+00 -7.79557347e-01 -4.73940700e-01
-7.68092573e-01 1.28365219e+00 1.49516210e-01 5.04896045e-01
-1.40372562e+00 -2.11141884e-01 4.60783094e-01 7.88845345e-02
4.26223844e-01 -7.22878948e-02 1.64672449e-01 -1.33820641e+00
3.99548024e-01 -7.70654321e-01 5.24899125e-01 -8.10169041e-01
-9.20448124e-01 3.34835857e-01 -1.08784653e-01 -1.47751141e+00
2.26334468e-01 -4.64502782e-01 -7.21080959e-01 4.29488063e-01
-1.36379313e+00 -1.03223443e+00 -6.04376853e-01 9.56445456e-01
9.41898286e-01 1.67855650e-01 3.08017164e-01 6.91935182e-01
-9.53063309e-01 5.30326366e-01 -1.40064299e-01 3.38718653e-01
7.87176073e-01 -1.10126626e+00 -1.19201422e-01 1.01006436e+00
2.71300703e-01 4.57314283e-01 3.32452834e-01 -6.22417629e-01
-9.99285102e-01 -1.19230485e+00 5.84461212e-01 -4.77786928e-01
8.61290872e-01 -4.13174987e-01 -7.84147561e-01 7.44571805e-01
-3.17319622e-03 5.23224115e-01 6.81143165e-01 -2.97760308e-01
-7.74958581e-02 -1.81078106e-01 -6.89211667e-01 4.47261274e-01
1.38341331e+00 -5.03047407e-01 -5.80354571e-01 5.09244800e-01
7.00621665e-01 -3.18065524e-01 -4.03460056e-01 4.63448972e-01
4.41374958e-01 -1.07250214e+00 8.52217615e-01 -5.03509283e-01
4.13080812e-01 -8.73236299e-01 -1.61864534e-01 -6.96284354e-01
-3.41951281e-01 -6.45228744e-01 -2.90333778e-01 1.45078385e+00
1.95395753e-01 -2.38692492e-01 9.25870895e-01 4.26337898e-01
-2.01861635e-01 -8.49328637e-01 -9.85196650e-01 -8.40379715e-01
-4.93990958e-01 -3.12853098e-01 1.13053985e-01 9.11982298e-01
-3.86474244e-02 2.40960285e-01 -5.41395366e-01 2.34112352e-01
4.76769775e-01 -1.18798213e-02 5.81773758e-01 -7.20153809e-01
-5.66157520e-01 -1.59167871e-01 -4.89740223e-01 -1.44528425e+00
1.02503367e-01 -6.74237788e-01 3.13250929e-01 -1.25499547e+00
4.27972913e-01 -1.32744178e-01 -8.10117662e-01 7.83360481e-01
-4.28038597e-01 2.72934645e-01 2.34685466e-02 3.63056093e-01
-1.33600259e+00 6.19049847e-01 1.11205471e+00 -8.10732692e-02
-1.32424027e-01 7.34522343e-02 -3.08846921e-01 1.20730531e+00
7.56405473e-01 -5.15591204e-01 -5.51062286e-01 -3.82606000e-01
-2.28096962e-01 -1.07950129e-01 5.11432469e-01 -1.29150689e+00
3.63984495e-01 -2.62979984e-01 4.93244737e-01 -6.40609026e-01
2.78209448e-01 -7.99805343e-01 -2.05036819e-01 3.96813571e-01
-5.16823292e-01 -2.40510851e-01 -4.35095839e-02 8.78409624e-01
-4.07743067e-01 -1.52145177e-01 7.83521354e-01 -1.56280234e-01
-1.11671495e+00 4.76408362e-01 -2.31605634e-01 -3.00814752e-02
1.28842914e+00 -2.80598611e-01 -2.06332043e-01 -1.83294550e-01
-1.02569818e+00 3.22397470e-01 3.04758459e-01 3.07232529e-01
4.85825986e-01 -1.53589547e+00 -3.30699027e-01 -3.82937975e-02
1.66480616e-01 -3.08115542e-01 3.66193771e-01 1.39115512e+00
-1.85770497e-01 1.83573306e-01 1.30007073e-01 -6.94543660e-01
-1.52281284e+00 5.32589734e-01 3.98365140e-01 -1.75213620e-01
-6.15814567e-01 1.09239972e+00 6.14998043e-01 1.72921702e-01
6.65620148e-01 -3.50275993e-01 -2.26443842e-01 -5.19396253e-02
8.57842207e-01 3.63009840e-01 -3.65980983e-01 -5.60369611e-01
-6.08587921e-01 5.56068182e-01 1.36919012e-02 3.40844728e-02
1.08078074e+00 -1.64114341e-01 -4.58748545e-03 5.43771386e-01
1.11710083e+00 -1.24677151e-01 -1.81158125e+00 -3.38672549e-01
1.71574071e-01 -7.18178034e-01 4.38383259e-02 -6.98394537e-01
-1.32827628e+00 9.32836235e-01 9.40326512e-01 -1.37137324e-01
1.29769278e+00 1.14230804e-01 7.37539887e-01 3.51186454e-01
2.31729820e-01 -1.34504068e+00 5.38201094e-01 3.24051440e-01
5.65806150e-01 -1.26077235e+00 -2.07798064e-01 -4.32373852e-01
-6.66671693e-01 8.90902042e-01 9.56955492e-01 -2.90188454e-02
6.06692135e-01 2.79538155e-01 8.64497200e-02 -1.31445611e-02
-7.23562419e-01 -4.60752785e-01 5.79968691e-02 4.17367548e-01
3.51456374e-01 -4.63094443e-01 -3.79034013e-01 7.71525145e-01
6.45195782e-01 3.17724049e-01 1.27309352e-01 1.05424070e+00
-5.34466624e-01 -1.00406289e+00 -1.12591013e-01 1.55677795e-01
-5.64670682e-01 -1.78032562e-01 -3.46520633e-01 4.94655579e-01
5.62747896e-01 9.24408793e-01 -1.72294855e-01 -5.28636396e-01
2.25381941e-01 1.62657812e-01 2.19684348e-01 -5.92614532e-01
-3.07661206e-01 3.61340225e-01 4.93237637e-02 -1.13135731e+00
-9.10345018e-01 -8.53954852e-01 -1.42692471e+00 3.40894163e-01
-4.44638431e-01 2.35876307e-01 2.40570381e-01 9.46236908e-01
2.88205773e-01 6.07560694e-01 6.94643676e-01 -8.52973700e-01
-4.04986709e-01 -9.12476003e-01 -4.88725185e-01 5.49580634e-01
2.94981092e-01 -9.43023920e-01 -6.16839647e-01 3.69858474e-01] | [8.494546890258789, 0.5865764617919922] |
e6054796-62f8-441c-b271-ff39b4eee51a | adversarial-audio-synthesis-with-complex | 2206.06811 | null | https://arxiv.org/abs/2206.06811v2 | https://arxiv.org/pdf/2206.06811v2.pdf | Adversarial Audio Synthesis with Complex-valued Polynomial Networks | Time-frequency (TF) representations in audio synthesis have been increasingly modeled with real-valued networks. However, overlooking the complex-valued nature of TF representations can result in suboptimal performance and require additional modules (e.g., for modeling the phase). To this end, we introduce complex-valued polynomial networks, called APOLLO, that integrate such complex-valued representations in a natural way. Concretely, APOLLO captures high-order correlations of the input elements using high-order tensors as scaling parameters. By leveraging standard tensor decompositions, we derive different architectures and enable modeling richer correlations. We outline such architectures and showcase their performance in audio generation across four benchmarks. As a highlight, APOLLO results in $17.5\%$ improvement over adversarial methods and $8.2\%$ over the state-of-the-art diffusion models on SC09 dataset in audio generation. Our models can encourage the systematic design of other efficient architectures on the complex field. | ['Volkan Cevher', 'Grigorios G Chrysos', 'Yongtao Wu'] | 2022-06-14 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 1.30870761e-02 2.59106010e-01 5.34771942e-02 -2.09158778e-01
-6.36687636e-01 -7.07510889e-01 5.92559338e-01 -3.20723593e-01
7.98582956e-02 3.82003576e-01 5.37867010e-01 -2.64299333e-01
-1.98718712e-01 -8.67468655e-01 -7.09771991e-01 -5.97886622e-01
-6.15168154e-01 1.65639490e-01 -1.06696360e-01 -6.28081977e-01
-1.70037389e-01 4.23705548e-01 -9.05367136e-01 5.54329574e-01
5.07825553e-01 1.22521818e+00 -4.32463944e-01 7.70366788e-01
9.36310813e-02 1.00021279e+00 -8.77792060e-01 -6.84246600e-01
2.42097065e-01 -2.86634207e-01 -6.52746081e-01 -2.05206964e-02
4.66071427e-01 -3.17157865e-01 -7.58427382e-01 9.35829580e-01
5.75101078e-01 7.90970847e-02 7.70615935e-01 -1.33294010e+00
-9.40948188e-01 1.31546640e+00 -3.58399034e-01 1.36861712e-01
5.10325395e-02 3.71935636e-01 1.33333111e+00 -1.02013314e+00
4.83477682e-01 1.50288165e+00 8.36456180e-01 4.87428308e-01
-1.48690915e+00 -7.68709362e-01 1.56585395e-01 1.95219424e-02
-1.20709491e+00 -5.15639961e-01 1.13013184e+00 -4.78567213e-01
8.79243135e-01 1.95705175e-01 5.82253635e-01 1.42482901e+00
1.41107991e-01 7.69630551e-01 8.96537423e-01 -3.98527011e-02
6.40416592e-02 -4.17707920e-01 -1.54950857e-01 6.01789355e-01
-1.35860965e-01 2.26069331e-01 -8.82452905e-01 -1.78848520e-01
7.67589450e-01 -1.51523262e-01 -3.11537743e-01 -1.34237885e-01
-1.50575829e+00 7.55485773e-01 7.67080367e-01 1.82893738e-01
-3.54663700e-01 9.88745213e-01 4.17763710e-01 3.74375582e-01
4.71163034e-01 7.79842615e-01 -2.06686005e-01 -2.63299435e-01
-9.57989156e-01 5.06006896e-01 6.19113743e-01 9.98145461e-01
3.87008786e-01 9.07132924e-01 -4.01516110e-01 8.32018971e-01
2.52700508e-01 4.75589633e-01 2.74863780e-01 -1.37049508e+00
6.25828326e-01 1.74470916e-01 -1.44851550e-01 -1.12457120e+00
-3.48323673e-01 -8.67683053e-01 -1.24374640e+00 1.31737113e-01
4.13832128e-01 -2.55944252e-01 -8.11166406e-01 1.77851820e+00
-1.50534719e-01 3.27377617e-01 -2.06568409e-02 8.90850782e-01
5.58054864e-01 6.68982625e-01 -3.27115178e-01 3.22400570e-01
1.32089221e+00 -1.04036510e+00 -7.93433189e-01 2.60176789e-02
1.98514879e-01 -9.83038008e-01 9.47855413e-01 6.66136980e-01
-1.32173526e+00 -4.61333156e-01 -1.23590136e+00 -1.63160767e-02
-2.37457901e-01 -1.33668957e-02 9.53402221e-01 6.57385826e-01
-1.15118670e+00 1.13916647e+00 -8.25932980e-01 4.76722926e-01
3.89707595e-01 3.31434309e-01 7.53056481e-02 1.89071447e-01
-1.44236028e+00 1.90784708e-01 -2.86266834e-01 3.12153816e-01
-1.15861261e+00 -1.16859841e+00 -6.15939379e-01 2.37483010e-01
9.28742290e-02 -7.69406676e-01 1.41846502e+00 -4.72047925e-01
-1.59458435e+00 1.00288309e-01 3.33325207e-01 -8.17564309e-01
3.37148279e-01 -2.47925848e-01 -6.23779178e-01 3.31887931e-01
-2.06848755e-01 8.38943183e-01 1.33765543e+00 -9.44675088e-01
-3.09746470e-02 1.46803752e-01 3.88409853e-01 -3.53190452e-01
-6.19222581e-01 -2.12936729e-01 -4.67047049e-03 -1.50828302e+00
3.14274244e-02 -1.11396551e+00 -4.31693226e-01 2.78264582e-01
-4.31848645e-01 -3.95365953e-02 5.27789533e-01 -5.01715839e-01
1.33771181e+00 -2.21507239e+00 4.35108304e-01 1.02811828e-01
7.47546017e-01 8.17559138e-02 -2.80029565e-01 5.60485184e-01
-1.58352599e-01 4.29666698e-01 -1.07643515e-01 -5.40085733e-01
3.48287255e-01 1.24060094e-01 -8.85616839e-01 1.39547557e-01
7.19707966e-01 8.81518483e-01 -9.89565194e-01 -8.53938162e-02
-1.90446213e-01 8.49817932e-01 -1.03009689e+00 -1.41541749e-01
-4.39177424e-01 3.48197937e-01 -3.49463373e-01 6.13198757e-01
5.12762189e-01 -3.79885554e-01 -1.20151993e-02 -4.68599796e-01
2.47505426e-01 5.58688581e-01 -9.91823792e-01 1.82587230e+00
-5.66526234e-01 8.78211141e-01 7.87208527e-02 -6.32709682e-01
9.24386322e-01 5.55894971e-01 5.76882243e-01 -3.04834485e-01
-2.58039404e-02 3.33440900e-01 3.58809233e-01 7.45706931e-02
7.17554331e-01 -2.79969838e-03 -1.41024336e-01 5.26286602e-01
3.04222554e-01 -5.36969900e-01 6.46231920e-02 4.33161616e-01
1.22016823e+00 3.49768810e-02 -3.86930734e-01 -1.83077767e-01
1.56192333e-01 -4.99708384e-01 3.41322958e-01 5.75768888e-01
-1.19929230e-02 7.66104519e-01 7.84569919e-01 -3.78634363e-01
-1.10363507e+00 -1.17522967e+00 6.22652024e-02 8.90801013e-01
-5.20997226e-01 -8.13730478e-01 -6.70531809e-01 -3.96085650e-01
6.84093237e-02 4.34840113e-01 -7.36998916e-01 -2.43946478e-01
-6.73774779e-01 -5.37912846e-01 1.16230798e+00 7.13765919e-01
2.52594650e-01 -7.33735621e-01 -2.60674786e-02 3.16747874e-01
-4.19963486e-02 -1.06727636e+00 -7.12631106e-01 2.37709671e-01
-7.58589506e-01 -4.51132298e-01 -1.00623357e+00 -3.16961259e-01
1.76554367e-01 1.22933701e-01 1.27479041e+00 -2.12788552e-01
2.23675948e-02 2.47480825e-01 -3.53269964e-01 -2.25370720e-01
-4.42275643e-01 2.19936758e-01 1.90789595e-01 2.07604960e-01
-2.58377761e-01 -1.20082760e+00 -8.70635092e-01 1.80780128e-01
-1.06631923e+00 -2.00402349e-01 4.42956567e-01 1.05527329e+00
2.57176310e-01 -8.53633508e-02 6.91716492e-01 -6.50010347e-01
9.88520622e-01 -4.40963000e-01 -2.85226136e-01 -1.37766525e-01
-5.24800420e-01 3.72670144e-01 1.00123680e+00 -8.75098050e-01
-6.07561171e-01 -4.37737852e-01 -1.20355733e-01 -7.92199731e-01
6.83333457e-01 5.02962649e-01 3.27386200e-01 -1.42120674e-01
1.00996041e+00 -1.24761686e-01 -2.77685642e-01 -4.20559078e-01
6.87227726e-01 3.37476343e-01 3.79565865e-01 -9.50735867e-01
1.20004058e+00 4.33082312e-01 2.07405716e-01 -5.06448150e-01
-7.58541048e-01 2.47692809e-01 -2.75123954e-01 -1.18738897e-01
4.96506929e-01 -1.00618148e+00 -8.18301141e-01 4.86899167e-01
-1.18515456e+00 -3.43410224e-01 -6.08014762e-01 5.15506506e-01
-4.92734194e-01 1.26530886e-01 -1.01309073e+00 -5.60987711e-01
-4.38180804e-01 -1.21646035e+00 1.06659675e+00 -1.85755134e-01
-3.50572854e-01 -8.59963715e-01 -1.62315398e-01 2.79168278e-01
7.84984410e-01 2.49874413e-01 9.63655293e-01 -9.70504880e-02
-8.77982438e-01 -1.73341506e-03 -4.20792922e-02 6.32930517e-01
-2.21240297e-01 1.24426112e-01 -1.20215929e+00 -8.52999017e-02
-1.55412614e-01 -3.43192130e-01 9.57682908e-01 9.22022760e-03
1.26856363e+00 -3.70334893e-01 1.77227095e-01 9.21040833e-01
7.94862151e-01 -1.11259408e-01 5.12768269e-01 -1.63537964e-01
8.20625961e-01 3.20516109e-01 1.83246702e-01 6.18923187e-01
2.12225541e-01 6.56036735e-01 4.52420473e-01 -3.50818634e-02
-6.48263633e-01 -4.81706262e-01 5.44192731e-01 1.37964237e+00
-2.55496800e-01 -1.03591375e-01 -7.74763465e-01 4.32579219e-01
-1.44618118e+00 -9.42172825e-01 -3.63570936e-02 1.68047118e+00
1.09437883e+00 3.47997397e-01 8.04157108e-02 4.61725026e-01
2.40358993e-01 7.40941346e-01 -5.34731865e-01 -3.74004215e-01
-1.89495698e-01 7.18277335e-01 3.60845000e-01 3.46229911e-01
-8.17419708e-01 8.22193563e-01 6.51666021e+00 9.89176035e-01
-1.32584095e+00 1.16742909e-01 5.99308968e-01 -2.50999391e-01
-7.70713151e-01 -1.69280976e-01 -4.85759944e-01 1.77013710e-01
1.08713853e+00 -1.48438200e-01 9.59986687e-01 6.09957516e-01
-5.56544065e-02 8.10421407e-01 -1.10549736e+00 7.77328432e-01
-2.07677826e-01 -1.57529485e+00 1.13378875e-01 1.66044489e-01
7.93996692e-01 -2.92802826e-02 6.95019782e-01 3.99181515e-01
4.16703999e-01 -1.21503830e+00 9.64787006e-01 5.27350843e-01
8.63199830e-01 -6.12767339e-01 5.75579032e-02 -2.23082170e-01
-1.33349478e+00 -8.32328722e-02 -3.85190219e-01 -3.46704051e-02
2.46668100e-01 6.95469379e-01 -7.63803780e-01 4.53490317e-01
6.69483960e-01 8.32745373e-01 -4.24660176e-01 5.44530272e-01
-2.18541175e-01 9.18852448e-01 -1.90265685e-01 -4.43336032e-02
4.35041994e-01 -5.75098097e-02 6.53231442e-01 1.04884148e+00
4.94966149e-01 -2.79886127e-01 -2.14318410e-01 1.12500131e+00
-4.42717642e-01 -3.45130473e-01 -4.45220023e-01 -6.31829500e-01
3.48870546e-01 1.21093154e+00 -2.13649780e-01 -2.10543185e-01
-3.06240708e-01 7.58085012e-01 2.34770492e-01 5.69786012e-01
-1.05784273e+00 -4.81313169e-01 9.56619561e-01 1.60966665e-01
3.65271270e-01 -5.77787578e-01 -3.01106840e-01 -1.17302775e+00
1.39391258e-01 -1.23810148e+00 -1.83140516e-01 -8.37302268e-01
-1.41201758e+00 1.01988780e+00 -2.23585337e-01 -1.46068275e+00
-5.75687647e-01 -6.81655288e-01 -2.80797392e-01 8.10823858e-01
-1.25644553e+00 -1.21454644e+00 5.95756955e-02 5.82363844e-01
3.27384233e-01 -2.28447556e-01 8.19918871e-01 5.49825490e-01
-2.82713652e-01 9.22523439e-01 -6.18883632e-02 2.12352738e-01
7.12450266e-01 -1.30879223e+00 9.86266196e-01 6.20996475e-01
4.68712002e-01 8.79347444e-01 5.33086717e-01 -2.60559171e-01
-1.58451796e+00 -1.08391726e+00 6.32142603e-01 -3.16001743e-01
1.38622987e+00 -5.87455809e-01 -6.91147447e-01 5.38303375e-01
3.16889495e-01 2.46056601e-01 5.64318836e-01 5.11498414e-02
-9.92200077e-01 -3.83827329e-01 -7.32886314e-01 1.06602335e+00
1.27750695e+00 -9.23341691e-01 -9.91326571e-02 3.37301970e-01
1.38627017e+00 -3.95805210e-01 -1.51446640e+00 3.29087675e-01
6.35166645e-01 -7.54779100e-01 1.32911694e+00 -7.34652400e-01
7.79147148e-01 -2.38184571e-01 -2.91913152e-01 -1.51784742e+00
-4.48291600e-01 -1.46745920e+00 -5.99938571e-01 1.14607441e+00
6.20227575e-01 -4.29270655e-01 6.00116074e-01 2.34638691e-01
-3.68779391e-01 -1.05094934e+00 -8.27401757e-01 -7.34061480e-01
3.88607442e-01 -7.06466138e-01 8.45962524e-01 9.64250982e-01
-1.28121495e-01 4.28278148e-01 -7.44161904e-01 9.86989494e-03
4.17250514e-01 -1.42424643e-01 6.07512772e-01 -8.62338901e-01
-7.72161603e-01 -5.91836452e-01 -3.48066539e-01 -1.34214282e+00
1.60103828e-01 -8.60023379e-01 -4.28623527e-01 -9.59645391e-01
-5.31537712e-01 -7.09044158e-01 -3.05180788e-01 2.84185946e-01
8.37273672e-02 4.94481981e-01 4.99103874e-01 4.55899135e-04
-9.68466699e-02 8.21416616e-01 1.68812299e+00 -5.54772735e-01
2.57415324e-01 6.63632061e-03 -7.50434220e-01 5.32989502e-01
8.77403975e-01 -4.17683810e-01 -6.25705302e-01 -7.66737580e-01
5.87392390e-01 3.47182512e-01 3.60806286e-01 -1.18940926e+00
1.31945074e-01 4.61870767e-02 2.30977580e-01 -2.94648409e-01
1.05000663e+00 -5.83203495e-01 1.48822814e-01 2.50023663e-01
-5.14848292e-01 2.97046185e-01 -1.24435378e-02 5.59327483e-01
-3.84105712e-01 -4.26753424e-02 3.62502635e-01 5.90720885e-02
-3.06840595e-02 4.91041452e-01 -2.94121951e-01 1.99901402e-01
3.48010749e-01 3.10599953e-01 -6.52962685e-01 -7.89375246e-01
-6.62493527e-01 -2.03078553e-01 3.08173876e-02 5.69488227e-01
6.05566263e-01 -1.62274230e+00 -6.42230034e-01 1.49732828e-01
-1.75662696e-01 -1.36911362e-01 2.63318032e-01 6.78297639e-01
-4.25413191e-01 3.38314384e-01 -8.02587811e-03 -4.04174358e-01
-5.68959355e-01 4.28599507e-01 4.19502556e-01 -4.90390509e-01
-3.33349615e-01 9.47275281e-01 4.50801812e-02 -3.51306051e-01
2.70749629e-01 -7.20508039e-01 2.37977222e-01 -5.59313409e-02
2.93715537e-01 3.80854785e-01 -8.76510218e-02 -4.03811812e-01
-1.73466519e-01 4.16878849e-01 1.01992562e-01 -5.30733347e-01
1.30240703e+00 3.83953661e-01 -8.15631747e-02 4.76116210e-01
1.21832657e+00 3.78038049e-01 -1.43213642e+00 -2.61472583e-01
-4.17448014e-01 -1.21637195e-01 7.17116967e-02 -6.29614592e-01
-1.48483169e+00 1.44336200e+00 3.03100199e-01 3.32376748e-01
1.04207861e+00 -3.57080251e-01 9.05926526e-01 4.36337680e-01
3.13319594e-01 -7.06744075e-01 5.91742158e-01 5.51129341e-01
1.26109815e+00 -6.00642264e-01 -6.05730228e-02 -5.63166380e-01
-5.28598189e-01 1.21786308e+00 1.51815936e-01 -4.67808366e-01
8.98226082e-01 4.30657476e-01 6.61100075e-02 -1.02116615e-01
-1.06024837e+00 1.85989156e-01 5.59307218e-01 6.23806715e-01
7.11098731e-01 2.85119385e-01 7.03158081e-02 6.28813088e-01
-7.02316523e-01 -2.59503812e-01 4.60290700e-01 5.65925598e-01
2.39242047e-01 -1.35231185e+00 -3.47659349e-01 2.79545158e-01
-7.29434848e-01 -4.71055090e-01 -2.32256100e-01 4.78271008e-01
-1.35534182e-01 1.07339609e+00 -3.44315283e-02 -8.31954837e-01
1.58192307e-01 -2.00271428e-01 3.92778069e-01 -2.50341237e-01
-1.06142676e+00 1.71818897e-01 2.17347249e-01 -5.65768242e-01
-1.06538646e-01 -2.46865779e-01 -9.66155708e-01 -4.93997246e-01
1.85412452e-01 -6.47527054e-02 5.63424885e-01 2.24567786e-01
6.19026244e-01 1.05785143e+00 7.42616594e-01 -1.00690651e+00
-1.06811786e+00 -1.07941246e+00 -3.90962034e-01 2.96961784e-01
4.90481466e-01 -5.44517815e-01 -4.02234524e-01 -8.89980607e-03] | [15.602115631103516, 5.926140308380127] |
9bc3ec13-2225-4d22-be13-dca447cbf4a0 | face-hallucination-using-split-attention-in | 2010.11575 | null | https://arxiv.org/abs/2010.11575v3 | https://arxiv.org/pdf/2010.11575v3.pdf | Face Hallucination via Split-Attention in Split-Attention Network | Recently, convolutional neural networks (CNNs) have been widely employed to promote the face hallucination due to the ability to predict high-frequency details from a large number of samples. However, most of them fail to take into account the overall facial profile and fine texture details simultaneously, resulting in reduced naturalness and fidelity of the reconstructed face, and further impairing the performance of downstream tasks (e.g., face detection, facial recognition). To tackle this issue, we propose a novel external-internal split attention group (ESAG), which encompasses two paths responsible for facial structure information and facial texture details, respectively. By fusing the features from these two paths, the consistency of facial structure and the fidelity of facial details are strengthened at the same time. Then, we propose a split-attention in split-attention network (SISN) to reconstruct photorealistic high-resolution facial images by cascading several ESAGs. Experimental results on face hallucination and face recognition unveil that the proposed method not only significantly improves the clarity of hallucinated faces, but also encourages the subsequent face recognition performance substantially. Codes have been released at https://github.com/mdswyz/SISN-Face-Hallucination. | ['Junjun Jiang', 'Yanduo Zhang', 'Zhongyuan Wang', 'Wei Liu', 'Yu Wang', 'Tao Lu', 'Yuanzhi Wang'] | 2020-10-22 | null | null | null | null | ['face-hallucination'] | ['computer-vision'] | [ 5.02828173e-02 6.32437617e-02 1.97940052e-01 -3.56533438e-01
-3.16882998e-01 3.00054485e-03 4.68419582e-01 -5.56499839e-01
1.45779833e-01 5.44325352e-01 4.05332714e-01 3.19340557e-01
-1.38157560e-02 -8.16380799e-01 -6.26520574e-01 -8.60320032e-01
4.95223761e-01 -2.09126115e-01 -2.97768325e-01 -1.37080148e-01
-6.44869218e-03 8.29904139e-01 -1.85815418e+00 3.12457234e-01
8.12085211e-01 1.18773961e+00 7.34449849e-02 -4.76586670e-02
-8.03633593e-03 6.62804425e-01 -1.39766037e-01 -4.62345958e-01
5.08748889e-02 -3.94460320e-01 -4.18739855e-01 3.41279477e-01
3.61554533e-01 -4.51419264e-01 -6.52399242e-01 1.24812841e+00
6.10261381e-01 8.21979418e-02 4.34828520e-01 -1.12098467e+00
-1.05096364e+00 -9.99310091e-02 -7.33677208e-01 -5.47651052e-02
2.74900913e-01 3.35839450e-01 4.24697697e-01 -1.35106266e+00
4.00230616e-01 1.62066603e+00 4.57702190e-01 6.97224140e-01
-1.16632903e+00 -1.08688641e+00 -9.11849290e-02 1.92918494e-01
-1.48247552e+00 -9.81863260e-01 9.02324378e-01 -2.50359505e-01
4.61715698e-01 2.45263740e-01 5.29908001e-01 1.04871011e+00
-4.33152281e-02 5.04200339e-01 1.01128471e+00 -2.02732198e-02
-2.03945488e-01 2.84616612e-02 -4.66510743e-01 8.41629863e-01
2.57651061e-02 2.92075008e-01 -4.00925756e-01 6.17930964e-02
1.09678519e+00 3.03389251e-01 -6.82387292e-01 8.02628398e-02
-9.23463225e-01 4.96348888e-01 5.55332959e-01 2.48147115e-01
-5.46161950e-01 -2.73371607e-01 1.88730896e-01 1.98796373e-02
6.04917169e-01 1.49496302e-01 9.00298432e-02 3.93271625e-01
-6.47655785e-01 -1.12833725e-02 1.40522361e-01 6.63858116e-01
7.35194743e-01 3.05222929e-01 -3.22557896e-01 1.22059548e+00
3.29611182e-01 4.15134460e-01 4.24707174e-01 -9.36845481e-01
1.98342115e-01 5.88013411e-01 1.91886909e-02 -1.32071722e+00
-2.21045211e-01 -3.57215703e-01 -1.39654350e+00 3.66755068e-01
1.30862249e-02 1.59054250e-01 -8.66488695e-01 1.81505895e+00
4.31923270e-01 2.39298880e-01 -5.86439595e-02 1.12012756e+00
1.11250818e+00 7.60964453e-01 1.46319568e-01 -3.80622864e-01
1.34714031e+00 -8.30963492e-01 -8.76277804e-01 -7.01320823e-03
-1.79641843e-01 -8.78492236e-01 8.98441672e-01 1.96927756e-01
-1.22952282e+00 -1.01542664e+00 -8.04306209e-01 -3.09609175e-01
1.11128919e-01 3.84986937e-01 3.23002607e-01 2.20739573e-01
-1.04504848e+00 5.57835519e-01 -4.90569651e-01 -5.69312125e-02
9.19211328e-01 3.60734284e-01 -7.42814541e-01 -3.42405558e-01
-1.05926895e+00 5.44937253e-01 1.82174221e-02 7.12689102e-01
-8.33705187e-01 -5.75123191e-01 -8.70570362e-01 3.61307055e-01
1.05468184e-01 -5.32891273e-01 7.59249687e-01 -1.07410538e+00
-1.46362853e+00 6.87281191e-01 -3.35691482e-01 3.45772117e-01
4.00820404e-01 2.16297526e-02 -6.60499752e-01 1.39630228e-01
7.96899125e-02 8.74752820e-01 1.02561557e+00 -1.31904447e+00
-7.46699274e-02 -7.10786223e-01 -3.75022709e-01 2.74337173e-01
-4.67244446e-01 1.04520367e-02 -3.67567837e-01 -6.91529989e-01
4.08012979e-02 -4.26677436e-01 1.84523404e-01 4.67323273e-01
-4.94395941e-01 -5.08433096e-02 8.28051090e-01 -9.00121570e-01
9.42468643e-01 -2.44397640e+00 1.64943099e-01 -9.70381573e-02
4.68638331e-01 5.43903649e-01 -4.46816951e-01 1.30548969e-01
-3.69788587e-01 2.64841560e-02 -1.57875121e-01 -4.58593607e-01
-3.96263957e-01 -7.96961486e-02 -3.41885120e-01 4.47644919e-01
6.06073201e-01 1.04352117e+00 -5.81736505e-01 -3.27795714e-01
2.70601213e-01 1.03394747e+00 -4.46849078e-01 4.24258292e-01
5.91228530e-03 7.67428398e-01 -3.85835141e-01 7.04438090e-01
1.15860415e+00 -2.14289650e-01 3.87438573e-02 -4.40869033e-01
-1.72262453e-02 -2.33569458e-01 -8.11031401e-01 1.21107984e+00
-4.35801297e-01 3.04473788e-01 2.15039834e-01 -7.31170893e-01
1.13492513e+00 5.06942272e-01 2.59263456e-01 -9.60830450e-01
1.82718113e-01 2.27273256e-01 -1.60620168e-01 -6.32715642e-01
-2.52100192e-02 -4.74055916e-01 5.38756371e-01 1.41124040e-01
7.87061602e-02 2.87380189e-01 -3.76782238e-01 -2.42511213e-01
3.89495194e-01 -2.50591338e-02 3.07535857e-01 2.89067510e-04
8.01731288e-01 -7.20835745e-01 6.24455214e-01 -1.26870975e-01
-1.11946248e-01 8.93816948e-01 4.32187408e-01 -4.32932675e-01
-1.12206602e+00 -8.46440971e-01 -2.22880155e-01 6.40689015e-01
1.94842041e-01 -4.99123223e-02 -6.90509975e-01 -3.64348263e-01
-7.85073861e-02 2.68038988e-01 -8.59813273e-01 -4.50450391e-01
-2.62972146e-01 -5.38468719e-01 4.85480011e-01 3.68742615e-01
7.50697672e-01 -1.49420989e+00 5.40363118e-02 3.73635776e-02
-3.26194346e-01 -9.84962285e-01 -7.04892874e-01 -6.72503710e-01
-6.33037746e-01 -9.71012414e-01 -9.49969411e-01 -7.00903952e-01
8.54215980e-01 4.83121783e-01 5.64339757e-01 4.67293531e-01
-3.48218411e-01 -1.62374333e-01 -4.67378832e-02 4.61104326e-02
-1.66045174e-01 -5.31964898e-01 4.88817655e-02 7.76243389e-01
9.94075611e-02 -8.00157070e-01 -6.81742668e-01 3.68871063e-01
-1.03903091e+00 3.22687894e-01 7.21128285e-01 1.05382311e+00
4.85185832e-01 3.86117189e-03 6.63065970e-01 -5.78445911e-01
5.07920206e-01 -5.61961114e-01 -2.99478590e-01 1.53159469e-01
-3.23138833e-01 -1.92820057e-01 8.89213860e-01 -3.94576639e-01
-1.34679019e+00 -2.26885870e-01 -4.41662461e-01 -8.55005324e-01
-2.81437695e-01 2.45722920e-01 -6.78945839e-01 -1.61281675e-01
2.71819651e-01 6.00291312e-01 5.94519496e-01 -5.67493975e-01
-8.82310048e-02 6.03313565e-01 5.78699052e-01 -2.82146364e-01
6.10497773e-01 5.08330464e-01 1.35448603e-02 -7.75797427e-01
-7.27741659e-01 -4.39811684e-03 -2.80324012e-01 -2.48439893e-01
8.03239882e-01 -1.00746930e+00 -9.88647044e-01 8.22739184e-01
-1.21092522e+00 6.48803590e-03 5.90291619e-02 2.69962400e-01
-3.56454134e-01 4.54371840e-01 -5.57043254e-01 -8.52393568e-01
-4.16520894e-01 -1.19894993e+00 1.08010745e+00 4.80379969e-01
2.09026486e-01 -5.45850277e-01 -3.13758433e-01 4.71496761e-01
5.30446589e-01 3.24158430e-01 8.97150695e-01 5.15843257e-02
-6.29177868e-01 1.10678211e-01 -6.87473774e-01 4.71935898e-01
4.29150105e-01 -1.23402756e-02 -1.31486619e+00 -3.02406132e-01
3.95598523e-02 -4.13855612e-01 7.81357765e-01 3.10095757e-01
1.38363314e+00 -4.74377453e-01 -1.56816524e-02 7.52622724e-01
1.22837794e+00 1.70070350e-01 1.08588219e+00 -2.35728666e-01
7.95827448e-01 9.50862825e-01 2.96941787e-01 5.10205865e-01
1.14711538e-01 7.78267145e-01 4.98006582e-01 -3.98464501e-01
-4.96204585e-01 -4.99452740e-01 3.39518458e-01 5.60630798e-01
-2.85192996e-01 1.78219546e-02 -4.68658924e-01 3.90714854e-01
-1.48910391e+00 -9.83240545e-01 1.67296976e-01 1.99495339e+00
6.85499787e-01 -4.11995590e-01 -2.11536527e-01 1.66369732e-02
9.85777080e-01 2.46058196e-01 -7.32246697e-01 -1.62816588e-02
-1.79493457e-01 2.28038654e-01 -2.81366378e-01 4.94581759e-01
-6.00947082e-01 7.76316583e-01 4.82199144e+00 9.64078307e-01
-1.23066866e+00 5.98116294e-02 1.01690400e+00 -2.38189269e-02
-4.42833364e-01 -3.53638798e-01 -4.94723111e-01 6.35585368e-01
4.12457198e-01 4.62031849e-02 7.36698091e-01 5.06732583e-01
3.47305447e-01 3.61919433e-01 -5.50903916e-01 1.12627280e+00
1.34371564e-01 -1.19226885e+00 5.42522371e-01 1.09936856e-01
5.79947472e-01 -5.01000404e-01 3.65591168e-01 -6.67666970e-03
-2.58976430e-01 -1.55861783e+00 5.81504881e-01 8.15815508e-01
1.44747174e+00 -8.92825127e-01 5.74087799e-01 8.31311122e-02
-1.17477477e+00 -2.17786089e-01 -5.55667162e-01 2.12828293e-01
-3.49589661e-02 5.43223083e-01 -2.83471286e-01 5.78792334e-01
6.81102216e-01 7.20401466e-01 -3.68983895e-01 7.32604623e-01
-1.64835587e-01 1.15049466e-01 1.06861480e-02 4.66581762e-01
-4.69136722e-02 -1.74621880e-01 4.61674005e-01 5.95021307e-01
5.66598952e-01 4.29286927e-01 -2.14124784e-01 1.29737103e+00
-3.42024803e-01 1.27465740e-01 -4.84916836e-01 -7.90662393e-02
4.79778260e-01 1.39682055e+00 -1.71309650e-01 -1.27259687e-01
-3.90786409e-01 8.77908945e-01 4.29115355e-01 4.63774860e-01
-6.02177143e-01 -2.52381861e-01 8.60044301e-01 3.53596836e-01
1.51319131e-01 1.76553234e-01 -8.72927904e-02 -1.19191074e+00
2.74498284e-01 -8.41977596e-01 2.55360384e-03 -1.04279637e+00
-1.24207091e+00 9.66829658e-01 -4.68032122e-01 -1.22093749e+00
2.60973722e-01 -3.91295522e-01 -6.89065695e-01 1.24354887e+00
-1.61956167e+00 -1.21354115e+00 -6.82093084e-01 8.13741744e-01
3.74177903e-01 -2.02386782e-01 7.60822475e-01 5.13607621e-01
-8.45006883e-01 6.88453972e-01 -2.04717830e-01 7.79224783e-02
7.51600385e-01 -4.61362541e-01 2.11662143e-01 5.49414933e-01
-2.84304321e-01 6.78679407e-01 3.06149274e-01 -6.61105514e-01
-1.09778845e+00 -1.33473551e+00 7.59530246e-01 8.36566612e-02
4.02917340e-02 -1.70760766e-01 -1.24959314e+00 4.10379261e-01
4.67264466e-02 4.22793001e-01 4.54859227e-01 -3.45758557e-01
-6.00072205e-01 -3.15891087e-01 -1.27265799e+00 5.79773426e-01
1.03569818e+00 -6.07063353e-01 -5.81844524e-02 5.57986461e-02
4.62804645e-01 -1.15818955e-01 -8.91805649e-01 6.06608570e-01
6.62583053e-01 -1.31072724e+00 8.72490704e-01 -4.37791228e-01
9.32355523e-01 -3.64278913e-01 7.87546672e-03 -1.14497018e+00
-5.63221455e-01 -4.13083225e-01 1.41514232e-02 1.33694327e+00
-2.29710955e-02 -8.29466581e-01 4.71987993e-01 3.25182796e-01
-1.22460321e-01 -8.98771524e-01 -8.57198656e-01 -4.18055385e-01
-2.01101527e-01 1.96121246e-01 8.40450823e-01 9.44495261e-01
-2.29283944e-01 2.92465210e-01 -7.06872761e-01 2.70226836e-01
6.30521417e-01 1.72191754e-01 4.55898076e-01 -1.15481424e+00
-4.08278890e-02 -4.46212649e-01 -3.16421002e-01 -7.71620512e-01
2.70009190e-01 -8.29962611e-01 -2.19078630e-01 -1.23313105e+00
4.27204251e-01 -1.63073301e-01 -8.79808739e-02 6.03355646e-01
-2.50861794e-01 6.60920739e-01 4.32782359e-02 3.46953005e-01
-2.32721388e-01 1.16655922e+00 1.81121778e+00 1.27310202e-01
6.91234693e-02 -2.22133636e-01 -1.04145360e+00 4.49887305e-01
7.36845970e-01 -4.69873706e-03 -3.60155761e-01 -3.14114958e-01
-2.98435777e-01 4.51411873e-01 7.48286009e-01 -9.72909570e-01
6.34141490e-02 -1.29160717e-01 9.58250642e-01 -1.22750483e-01
5.54120004e-01 -5.77278495e-01 4.50614691e-01 2.69719005e-01
-1.85150519e-01 -2.39861682e-01 2.78505385e-01 5.33843458e-01
-5.61867893e-01 2.73655802e-01 1.26003575e+00 -7.98511654e-02
-4.66842234e-01 8.78704548e-01 -9.45615843e-02 -3.78528148e-01
9.17333901e-01 -2.42777452e-01 -3.06692898e-01 -4.19082046e-01
-7.80650079e-01 -2.27034260e-02 4.65143085e-01 5.82396388e-01
1.00178194e+00 -1.72810638e+00 -8.52323472e-01 8.42759132e-01
8.28925446e-02 -5.84178604e-02 1.02637684e+00 9.28311110e-01
-1.94058970e-01 2.44258985e-01 -7.35083640e-01 -2.38668978e-01
-1.20418787e+00 6.46924257e-01 4.30732608e-01 7.75843039e-02
-6.88875020e-01 6.82589889e-01 6.96464300e-01 -3.29564542e-01
5.70408888e-02 2.34821066e-01 -4.72490668e-01 -9.78713483e-02
8.47878993e-01 1.69864595e-01 -1.12659819e-01 -1.02585876e+00
-1.45234451e-01 7.34908044e-01 -1.87978312e-01 3.15355808e-01
1.41477358e+00 -2.67691910e-01 -4.05984133e-01 -1.79680333e-01
1.32690001e+00 -7.07663372e-02 -1.47192740e+00 -2.43774429e-01
-6.25324070e-01 -8.26827049e-01 1.66178308e-02 -5.34170806e-01
-1.62610936e+00 1.09270608e+00 4.92934167e-01 -1.09075621e-01
1.47834122e+00 -2.25295305e-01 8.35263491e-01 -2.61431545e-01
1.56330600e-01 -4.30472046e-01 2.41779864e-01 1.10599957e-01
1.38911641e+00 -1.09679365e+00 -2.21214473e-01 -5.84050298e-01
-6.05693042e-01 1.23984158e+00 8.35929215e-01 -3.31417359e-02
5.86591840e-01 3.70242409e-02 -1.50118515e-01 -2.75772095e-01
-6.85793757e-01 -1.28986482e-02 3.47490340e-01 5.56788385e-01
3.73377234e-01 -1.11249380e-01 3.90931852e-02 7.51153350e-01
9.19720978e-02 1.65978521e-01 2.09554106e-01 2.07141578e-01
-2.45640457e-01 -7.47220159e-01 -4.87770200e-01 3.74943942e-01
-3.93724591e-01 -1.14690952e-01 -2.59340048e-01 4.43739265e-01
2.88505435e-01 8.11270118e-01 5.88858873e-03 -4.23451811e-01
3.23711962e-01 -1.57711953e-01 4.33869064e-01 -4.77794290e-01
-6.51388541e-02 2.01142743e-01 -2.13870734e-01 -6.55188501e-01
-2.36177608e-01 -3.37036580e-01 -8.44682336e-01 -6.00261807e-01
-6.91654161e-02 -6.46217093e-02 1.28719583e-01 5.81392646e-01
6.97133124e-01 4.10159886e-01 8.52617264e-01 -9.70270097e-01
-4.01932716e-01 -1.09535515e+00 -8.25515091e-01 4.32546824e-01
4.84473944e-01 -8.46323848e-01 -3.04719448e-01 -1.27682567e-01] | [12.840513229370117, -0.01770983263850212] |
051bc4d6-5e11-4766-b179-60b9d957e8ad | mathematical-modeling-in-human-evaluation-and | 2110.13909 | null | https://arxiv.org/abs/2110.13909v1 | https://arxiv.org/pdf/2110.13909v1.pdf | Mathematical Modeling, In-Human Evaluation and Analysis of Volume Kinetics and Kidney Function after Burn Injury and Resuscitation | Existing burn resuscitation protocols exhibit large variability in treatment efficacy. Hence, they must be further optimized based on comprehensive knowledge of burn pathophysiology. A physics-based mathematical model that can replicate physiological responses in diverse burn patients can serve as an attractive basis to perform non-clinical testing of burn resuscitation protocols and to expand knowledge on burn pathophysiology. We intend to develop, optimize, validate, and analyze a mathematical model to replicate physiological responses in burn patients. Using clinical datasets collected from 233 burn patients receiving burn resuscitation, we developed and validated a mathematical model applicable to computer-aided in-human burn resuscitation trial and knowledge expansion. Using the validated mathematical model, we examined possible physiological mechanisms responsible for the cohort-dependent differences in burn pathophysiology between younger versus older patients, female versus male patients, and patients with versus without inhalational injury. We demonstrated that the mathematical model could replicate physiological responses in burn patients associated with wide demographic characteristics and injury severity and that an increased inflammatory response to injury may be a key contributing factor in increasing the mortality risk of older patients and patients with inhalation injury via an increase in the fluid retention. The mathematical model may provide an attractive platform to conduct non-clinical testing of burn resuscitation protocols and test new hypotheses on burn pathophysiology. | ['Jose Salinas', 'Jin-OH Hahn', 'George C. Kramer', 'Chris Meador', 'Ali Tivay', 'Ghazal ArabiDarrehDor'] | 2021-10-24 | null | null | null | null | ['kidney-function'] | ['medical'] | [ 2.26547867e-01 -7.98193693e-01 6.89976066e-02 1.45229383e-03
-2.16413617e-01 8.31846818e-02 -2.53269166e-01 9.21256304e-01
-7.06352293e-01 5.48770189e-01 5.14591098e-01 -5.62401652e-01
-3.79566282e-01 -5.85378528e-01 -4.00652230e-01 -8.20559978e-01
-4.05870110e-01 5.89987874e-01 -2.03443378e-01 2.54193209e-02
3.09323549e-01 9.49917376e-01 -1.13380146e+00 6.91756308e-01
7.19273150e-01 7.29812980e-01 2.15021774e-01 8.63962591e-01
2.49049112e-01 5.80182493e-01 -2.92459875e-01 4.70501542e-01
1.78246573e-01 -8.76250744e-01 -3.15022796e-01 -6.93484485e-01
3.56712379e-02 -5.11455238e-01 -2.64113903e-01 -1.20540537e-01
1.27436912e+00 2.63298094e-01 8.12302172e-01 -8.93436134e-01
1.30973443e-01 6.35565162e-01 -1.07083008e-01 1.40334070e-01
2.22399771e-01 7.44928002e-01 -9.70940292e-02 -5.84166884e-01
-5.28323464e-02 9.22192633e-01 8.94531250e-01 9.93227899e-01
-1.19482636e+00 -6.07560813e-01 -3.68697912e-01 3.45824689e-01
-1.31093073e+00 -1.03866234e-01 2.80343533e-01 -5.61204255e-01
7.54331768e-01 6.68520272e-01 1.31726766e+00 7.71643519e-01
6.10067725e-01 -1.02711707e-01 1.16664708e+00 -3.40148807e-01
2.20064387e-01 -1.61944896e-01 1.67162083e-02 2.07008794e-01
6.20398939e-01 3.99282247e-01 -5.57071507e-01 -3.27761263e-01
7.95647025e-01 9.41034630e-02 -8.47698927e-01 2.54497267e-02
-9.56362486e-01 4.67128009e-01 -3.43099385e-02 1.06997415e-02
-1.11196101e+00 3.82128745e-01 8.09372842e-01 -3.41215968e-01
8.28688685e-03 5.55397868e-01 -6.06634617e-01 -2.48644218e-01
-3.66424084e-01 4.26360555e-02 9.11920011e-01 1.39680728e-01
5.07467240e-02 -1.58412635e-01 -2.24302381e-01 9.18855011e-01
1.27910823e-01 4.15729076e-01 3.15847874e-01 -8.73317778e-01
3.84738483e-02 2.90654063e-01 -2.51586497e-01 -3.92131180e-01
-5.84810197e-01 -6.44415468e-02 -9.71618176e-01 3.46551836e-01
1.82975277e-01 -1.56305015e-01 -3.04674715e-01 1.13291347e+00
2.65410811e-01 9.46852658e-03 2.88399998e-02 1.07277596e+00
5.54134607e-01 5.17448187e-02 9.33795333e-01 -5.20349145e-01
1.37249315e+00 -2.61746794e-01 -1.35527238e-01 2.66848266e-01
5.89997232e-01 -4.78166103e-01 8.35326254e-01 4.16902840e-01
-1.22007799e+00 -1.04313098e-01 -3.96368623e-01 6.84344351e-01
4.09215599e-01 -3.22316974e-01 1.80511624e-01 6.08658969e-01
-6.70208812e-01 8.14161003e-01 -7.91623175e-01 -1.00115275e+00
2.47665390e-01 1.33470237e-01 -1.75374106e-01 -5.38927555e-01
-1.07230115e+00 1.43380499e+00 2.74849802e-01 1.75408378e-01
-6.08174026e-01 -1.57373118e+00 -6.06616914e-01 -2.74453225e-04
-2.31793299e-01 -1.77548587e+00 8.00671458e-01 1.16836771e-01
-1.16975880e+00 3.55107546e-01 -1.89836115e-01 -2.95582771e-01
5.11222005e-01 -3.36443007e-01 2.32789263e-01 8.26575696e-01
-4.63096052e-01 -1.15839534e-01 -7.63186663e-02 -1.57792890e+00
6.26811832e-02 -2.83050239e-01 -5.60491443e-01 3.78538430e-01
2.06842810e-01 3.48384708e-01 6.22512341e-01 -2.03708902e-01
1.58867434e-01 -5.03918946e-01 -6.73650026e-01 -1.63301881e-02
2.65836060e-01 3.06098223e-01 3.42448920e-01 -5.45681119e-01
1.02366912e+00 -1.60052061e+00 -1.06272213e-01 2.74372995e-01
9.42865089e-02 2.03452706e-01 -4.20108214e-02 1.17534089e+00
-2.66049415e-01 8.06601167e-01 -1.79362252e-01 2.41551220e-01
-4.27437782e-01 1.94046702e-02 1.31703719e-01 7.98254132e-01
4.63294275e-02 7.40763962e-01 -8.67581725e-01 -6.31205499e-01
8.16375375e-01 6.02010190e-01 -5.07744908e-01 6.33395553e-01
1.90820992e-01 8.03955376e-01 -2.76125044e-01 4.48240578e-01
7.83404112e-01 5.07255733e-01 2.17202082e-01 -3.21501344e-01
-4.37529683e-01 -9.05086175e-02 -5.95750272e-01 9.20367181e-01
-4.98443604e-01 1.27840877e-01 3.93266290e-01 -6.55815065e-01
6.77365661e-01 7.78087199e-01 1.37996268e+00 -3.63323212e-01
5.96974611e-01 -6.57891408e-02 -4.00972739e-03 -1.21070993e+00
-2.66316980e-01 -1.62977314e+00 6.62602603e-01 5.58686972e-01
-7.17326164e-01 -7.49601305e-01 -1.67319074e-01 -3.35071445e-01
9.44954932e-01 -4.14177179e-01 8.15512463e-02 -4.95955855e-01
4.21263456e-01 1.39292121e-01 2.41674155e-01 6.16222501e-01
-2.79173881e-01 6.82636201e-01 2.38395140e-01 -3.26326102e-01
-7.50717640e-01 -7.46702850e-01 -5.00704467e-01 3.74292493e-01
1.06002897e-01 -8.67376775e-02 -5.45777261e-01 2.39610955e-01
2.56652534e-01 1.08481109e+00 -3.29618663e-01 -4.78283197e-01
-7.03538239e-01 -1.09796190e+00 7.54224718e-01 5.68946660e-01
-1.90151185e-01 -8.33524227e-01 -9.95269120e-01 3.78185749e-01
-4.27518457e-01 -6.58492446e-01 9.85218287e-02 1.95109025e-01
-1.05666471e+00 -1.62614405e+00 -6.05701864e-01 -5.27443111e-01
5.53762674e-01 3.06152701e-01 7.84308374e-01 9.53048527e-01
-9.85283971e-01 7.57269442e-01 -5.14369249e-01 -9.65712309e-01
-7.58142471e-01 -4.74521697e-01 4.90928352e-01 -5.43218732e-01
2.49239095e-02 -4.70158160e-01 -1.34880054e+00 4.21093911e-01
-1.17296314e+00 -2.38111764e-01 4.88963246e-01 4.63271469e-01
2.59109914e-01 6.24370724e-02 5.15970767e-01 -3.57145578e-01
8.68050992e-01 -8.23271632e-01 4.18667048e-01 4.91473377e-02
-9.91747618e-01 -3.76718223e-01 6.22019947e-01 -4.69906569e-01
-8.08201909e-01 -5.19997597e-01 -2.31312037e-01 -1.66574404e-01
-4.35805917e-01 6.93879664e-01 3.40524316e-01 5.80018796e-02
8.68867099e-01 -9.93900746e-02 6.53402507e-01 -5.27885735e-01
-4.53772157e-01 3.30584973e-01 1.99179277e-01 -1.21753943e+00
5.30624628e-01 2.65314937e-01 5.73428512e-01 -8.89653027e-01
3.90050337e-02 -9.42209780e-01 -6.99116766e-01 -5.56584597e-01
9.36809838e-01 -7.11766303e-01 -1.32901716e+00 6.68772340e-01
-9.58287299e-01 -1.02229178e+00 -5.16342163e-01 1.29699850e+00
-4.69744325e-01 7.04403520e-01 -8.11762094e-01 -1.17495060e+00
-7.36934125e-01 -9.00556386e-01 4.21726346e-01 4.22903076e-02
-5.52536130e-01 -1.23194051e+00 6.08906984e-01 3.82315218e-01
7.34337032e-01 8.10509324e-01 1.70567608e+00 -1.82037055e-01
-7.00944588e-02 -4.08506304e-01 1.64200246e-01 6.91917479e-01
8.13992985e-04 4.47351128e-01 -1.95181429e-01 -1.68092832e-01
1.90192744e-01 -3.70874703e-01 7.29004264e-01 5.43841600e-01
1.11593425e+00 -8.14062264e-03 -2.37376615e-01 6.07508242e-01
1.67783761e+00 2.68297583e-01 8.90193164e-01 -3.00328247e-02
3.09750021e-01 9.39840555e-01 2.68366158e-01 7.93700278e-01
7.54984468e-02 -1.16068475e-01 3.87363523e-01 -3.16843510e-01
-1.06653124e-01 5.48980199e-02 9.19341668e-02 6.36779904e-01
-8.09774995e-01 -3.36399853e-01 -1.16582477e+00 2.31340855e-01
-1.00034201e+00 -1.16008246e+00 -8.67531657e-01 2.31713915e+00
9.14316356e-01 -6.17001593e-01 3.53270508e-02 -7.36115351e-02
6.60870314e-01 -6.11277044e-01 -2.10937560e-01 -5.47715247e-01
1.68120965e-01 9.82287586e-01 3.98735166e-01 3.63868147e-01
-4.25815396e-02 2.31441095e-01 8.02016926e+00 -3.27843189e-01
-1.14653766e+00 -2.45269939e-01 1.79403663e-01 -2.13595182e-01
-1.29793510e-01 2.57379889e-01 -1.45839617e-01 8.50996841e-03
1.19462895e+00 -1.75720960e-01 2.68947035e-01 2.07133010e-01
9.63766992e-01 -6.80466712e-01 -1.14291024e+00 3.45575094e-01
2.83713136e-02 -1.06289005e+00 4.74976450e-02 2.10389048e-02
-1.58753488e-02 -2.87508160e-01 -3.92671496e-01 -3.21813039e-02
-3.19905818e-01 -1.26114905e+00 4.40558016e-01 8.91511917e-01
1.13291967e+00 -2.66090959e-01 1.18153977e+00 3.78762335e-01
-8.45965505e-01 4.08403482e-03 -1.03747338e-01 -2.68666714e-01
6.09408855e-01 4.68153864e-01 -1.26010847e+00 4.90419507e-01
3.85994822e-01 -3.61865461e-01 -4.13731225e-02 1.57904923e+00
2.07699835e-01 7.06541359e-01 -3.36736470e-01 6.25554249e-02
-5.64594030e-01 -2.49415800e-01 2.95756340e-01 1.01157391e+00
1.07214816e-01 1.04582703e+00 -7.41529241e-02 7.48763800e-01
7.39503205e-01 5.48515379e-01 -2.73073137e-01 1.48316160e-01
2.10291237e-01 6.25929296e-01 -4.42468524e-01 -3.69999893e-02
-1.06757060e-01 3.57023478e-01 -3.02315354e-01 5.30849993e-01
-6.17517591e-01 8.77058972e-03 8.39488149e-01 9.97945011e-01
-4.37301338e-01 -3.94597918e-01 -7.61733592e-01 -5.77145100e-01
-5.55862308e-01 -8.79525423e-01 4.28263605e-01 -6.86504245e-01
-1.35883033e+00 1.73922896e-01 6.08325899e-01 -6.69850051e-01
2.52543598e-01 -4.80711102e-01 -1.03209746e+00 1.53249824e+00
-1.37613356e+00 -5.57898164e-01 -9.93936419e-01 1.51719600e-01
9.39244553e-02 4.73152518e-01 1.15255845e+00 -1.06356964e-01
-7.44738758e-01 1.29357621e-01 -2.62932003e-01 -3.06080937e-01
7.20990539e-01 -5.96601665e-01 -6.93547785e-01 1.01656981e-01
-1.50703096e+00 9.30271029e-01 7.91925073e-01 -1.09306502e+00
-1.45113623e+00 -7.55285859e-01 4.09753472e-01 -3.92037660e-01
1.56674758e-01 4.18846339e-01 -9.28834617e-01 -2.23289117e-01
1.20461486e-01 -5.17572105e-01 1.54769218e+00 -3.02628726e-01
1.73159912e-01 1.07435323e-02 -1.38001609e+00 4.37874466e-01
7.86439955e-01 -1.24740198e-01 -4.00349945e-01 2.39353165e-01
3.16649973e-01 -1.14043355e-01 -1.59106946e+00 9.96135890e-01
7.52420604e-01 -8.07148218e-01 1.05783737e+00 -1.02033341e+00
4.75965470e-01 1.10293046e-01 1.32354453e-01 -1.05934727e+00
-4.79948908e-01 -4.83733833e-01 5.71752012e-01 6.35262489e-01
-1.07136883e-01 -8.20794225e-01 3.09197396e-01 1.48336840e+00
-2.26952434e-01 -1.24277234e+00 -6.66739345e-01 -5.66341043e-01
7.58993983e-01 -2.56383538e-01 4.24116105e-01 3.82316440e-01
4.46443051e-01 -4.03286397e-01 2.97986656e-01 -6.78920671e-02
4.26827967e-01 -6.06750607e-01 4.74347383e-01 -1.01526821e+00
1.86237678e-01 -5.02478004e-01 2.60325726e-02 6.58835843e-03
1.45958029e-02 -7.14609742e-01 2.05301657e-01 -2.11851144e+00
2.89850026e-01 -9.44228292e-01 -4.60246831e-01 1.73332080e-01
-2.52033383e-01 -1.94683418e-01 1.31690100e-01 9.37806293e-02
7.26292551e-01 4.77419496e-01 1.11762702e+00 3.91581833e-01
-3.11363161e-01 -1.10013828e-01 -5.90813279e-01 1.19673394e-01
9.95965242e-01 -6.20808423e-01 -1.40864447e-01 2.41027057e-01
-2.07821906e-01 4.83174890e-01 6.75501049e-01 -1.04812038e+00
1.25006571e-01 -1.12150586e+00 4.83455807e-01 -3.42911750e-01
-1.24470420e-01 -9.50909495e-01 5.98128557e-01 1.15891027e+00
-1.30972132e-01 1.12315670e-01 4.17545974e-01 2.16494799e-01
2.70927489e-01 -5.51829517e-01 9.92510915e-01 -3.91919076e-01
1.44302964e-01 3.68864894e-01 -1.12479591e+00 1.72187090e-01
1.29635990e+00 -3.27531070e-01 -3.99297327e-01 -1.32484779e-01
-4.55424041e-01 1.41871974e-01 9.20133173e-01 -1.64923757e-01
8.93512666e-01 -9.11028385e-01 -1.13461733e+00 2.40966771e-02
-2.10382883e-02 1.43382713e-01 8.01426828e-01 1.57182884e+00
-1.54904568e+00 2.45626569e-01 -2.18466982e-01 -3.10986578e-01
-1.25823653e+00 6.42012656e-01 7.26772130e-01 1.64059147e-01
-5.09543836e-01 7.83256829e-01 2.45369896e-02 2.94684798e-01
2.24034619e-02 -3.05027902e-01 1.78537041e-01 -2.17592835e-01
2.65289724e-01 8.08077574e-01 -2.29349434e-01 -2.31153831e-01
-5.69400251e-01 6.02161705e-01 6.60674691e-01 1.13563038e-01
1.11567771e+00 6.37165159e-02 -5.58507621e-01 4.16907102e-01
8.79370332e-01 -2.99702631e-03 -7.96279669e-01 5.01153171e-01
-8.62042785e-01 -6.49085939e-01 -3.41126055e-01 -8.78731072e-01
-6.88685000e-01 8.15776348e-01 3.83712828e-01 -2.28675932e-01
1.36302316e+00 -5.15288785e-02 8.67458105e-01 -2.45572016e-01
2.98368573e-01 -7.51397848e-01 1.28713444e-01 -4.00730185e-02
1.26494074e+00 -6.12674475e-01 3.08612227e-01 -6.29482508e-01
-7.34122574e-01 1.37088931e+00 5.98857701e-01 2.04088390e-01
8.75979543e-01 3.64024222e-01 4.66030687e-01 -1.99112713e-01
-8.19214106e-01 1.02810726e-01 -1.09163642e-01 9.04663801e-01
7.33358920e-01 2.21632764e-01 -9.58097041e-01 5.06445944e-01
5.56783825e-02 1.30224124e-01 5.60410261e-01 8.69514048e-01
-3.41226846e-01 -1.36852276e+00 -5.86529255e-01 5.49171865e-01
-3.08365285e-01 -2.63964325e-01 -2.80710250e-01 7.13818848e-01
1.89864427e-01 1.03320599e+00 -3.45798016e-01 -2.82823652e-01
7.13809490e-01 7.19364941e-01 7.79914200e-01 -7.04393864e-01
-1.14205980e+00 -3.34934920e-01 1.53338425e-02 -8.25964659e-02
-2.72201359e-01 -7.77280331e-01 -1.51687658e+00 -5.12028456e-01
-1.78607985e-01 2.41906032e-01 9.21615362e-01 5.00581741e-01
1.78063527e-01 4.85277325e-01 4.82371241e-01 -5.81263006e-01
-7.46478319e-01 -1.15117049e+00 -7.51283705e-01 3.63278508e-01
1.06556704e-02 -4.71516550e-01 -7.26184964e-01 -4.96241786e-02] | [8.06595230102539, 6.141534328460693] |
8894719e-512a-457e-bb64-8cb508f16786 | hyperspectral-unmixing-via-nonnegative-matrix | 2010.04611 | null | https://arxiv.org/abs/2010.04611v1 | https://arxiv.org/pdf/2010.04611v1.pdf | Hyperspectral Unmixing via Nonnegative Matrix Factorization with Handcrafted and Learnt Priors | Nowadays, nonnegative matrix factorization (NMF) based methods have been widely applied to blind spectral unmixing. Introducing proper regularizers to NMF is crucial for mathematically constraining the solutions and physically exploiting spectral and spatial properties of images. Generally, properly handcrafting regularizers and solving the associated complex optimization problem are non-trivial tasks. In our work, we propose an NMF based unmixing framework which jointly uses a handcrafting regularizer and a learnt regularizer from data. we plug learnt priors of abundances where the associated subproblem can be addressed using various image denoisers, and we consider an l_2,1-norm regularizer to the abundance matrix to promote sparse unmixing results. The proposed framework is flexible and extendable. Both synthetic data and real airborne data are conducted to confirm the effectiveness of our method. | ['Wei Chen', 'Jie Chen', 'Tiande Gao', 'Min Zhao'] | 2020-10-09 | null | null | null | null | ['hyperspectral-unmixing'] | ['computer-vision'] | [ 4.50423837e-01 -5.67113757e-01 -2.26971194e-01 -3.25999819e-02
-4.45354491e-01 -4.55161750e-01 4.90006089e-01 -3.00322622e-01
-3.25191796e-01 7.00651228e-01 3.99243325e-01 8.85571390e-02
-4.58577782e-01 -4.76791412e-01 -5.38620234e-01 -1.02985322e+00
2.15252206e-01 2.71028966e-01 -4.33057368e-01 -9.21102464e-02
-8.28646198e-02 4.55420405e-01 -1.60281706e+00 -1.78752899e-01
1.29100263e+00 6.92330062e-01 4.94125336e-01 5.04539311e-01
-5.18692508e-02 5.74142635e-01 -2.02228129e-01 8.49855319e-02
5.99628985e-01 -3.15607816e-01 -3.06223840e-01 7.64045238e-01
5.73038518e-01 3.67492959e-02 -2.11335674e-01 1.42228270e+00
2.21598074e-01 5.72764456e-01 7.91369438e-01 -1.23968232e+00
-6.87965035e-01 3.34345013e-01 -9.95113611e-01 1.89423934e-02
-5.55746071e-02 -4.08663563e-02 8.70799541e-01 -8.93693507e-01
1.39625981e-01 1.18788838e+00 6.22001767e-01 2.69874781e-01
-1.38469672e+00 -6.98146462e-01 2.72250533e-01 3.28341983e-02
-1.48693991e+00 -4.76045012e-01 9.90501523e-01 -7.11081803e-01
3.42657328e-01 5.39237857e-01 6.87329829e-01 7.01612055e-01
-2.25253075e-01 5.10599256e-01 1.26007843e+00 -4.75709587e-01
1.02190845e-01 -1.98510915e-01 -6.97909966e-02 4.04376477e-01
5.33106744e-01 -8.89326259e-02 -4.98813838e-01 -3.12852353e-01
9.39019084e-01 3.95527959e-01 -4.95575905e-01 -3.54533553e-01
-1.48265767e+00 7.46262193e-01 3.86633068e-01 1.14252023e-01
-7.27821648e-01 5.64575307e-02 -2.30267420e-01 -8.32485780e-02
6.36267900e-01 3.47641766e-01 7.87836686e-03 4.76160198e-01
-1.18167531e+00 2.70745546e-01 1.82328627e-01 6.92049980e-01
1.08111262e+00 3.70948315e-01 1.50504466e-02 1.19903421e+00
7.33788788e-01 1.10274673e+00 5.84991038e-01 -1.03668404e+00
3.94839227e-01 3.41760486e-01 4.55949277e-01 -1.24242067e+00
-2.61846870e-01 -4.90236819e-01 -1.18365109e+00 -6.87250048e-02
9.87561122e-02 -1.90821692e-01 -9.75355983e-01 1.62785780e+00
4.80068535e-01 8.82250130e-01 7.39961630e-03 1.24360263e+00
4.94294822e-01 9.34493661e-01 -4.97224480e-02 -6.92984641e-01
1.23472452e+00 -8.82860482e-01 -1.00608552e+00 -4.87580657e-01
-1.00092120e-01 -1.09435368e+00 9.71616387e-01 4.52403098e-01
-7.14086413e-01 -5.55578113e-01 -1.18503451e+00 1.41797587e-01
1.64748713e-01 3.51803243e-01 7.73367226e-01 3.76587778e-01
-6.01766944e-01 2.33910456e-01 -8.59703124e-01 -1.46032304e-01
2.72893091e-03 3.40730011e-01 -4.32068050e-01 -2.06866190e-01
-1.00774467e+00 4.93139774e-01 3.16812485e-01 5.62770128e-01
-9.44388092e-01 -5.44118166e-01 -7.49407947e-01 -2.98755527e-01
4.37203556e-01 -8.38150084e-01 7.29350030e-01 -1.17153704e+00
-1.47890103e+00 7.40772247e-01 -3.02125335e-01 -1.61888242e-01
7.61009082e-02 -4.01276320e-01 -6.86561882e-01 -7.59305656e-02
4.20407146e-01 1.65353343e-01 1.66639125e+00 -1.44130361e+00
-2.45414883e-01 -1.65582031e-01 -2.30657563e-01 3.95567834e-01
-5.07077754e-01 -2.60179669e-01 -2.96071112e-01 -1.18602955e+00
6.00682974e-01 -9.23235893e-01 -4.26586837e-01 -2.50394911e-01
-3.69610339e-01 4.53824639e-01 6.61851406e-01 -8.28029931e-01
1.27047992e+00 -2.31335902e+00 6.45275593e-01 3.21708143e-01
1.67031720e-01 1.91151366e-01 -1.89919993e-01 2.52998024e-01
-2.84618944e-01 -3.08219254e-01 -7.29305267e-01 -2.72789925e-01
-1.08244412e-01 3.88894141e-01 -4.13707882e-01 8.18969190e-01
5.15248906e-03 4.58022386e-01 -9.55222845e-01 -1.94320947e-01
4.80740845e-01 6.78766668e-01 -5.66788316e-01 4.97485608e-01
-3.47707421e-01 7.64345109e-01 -2.61621922e-01 4.39746082e-01
9.39052880e-01 -1.06422536e-01 2.83569366e-01 -6.14975512e-01
-2.98610121e-01 -3.22260767e-01 -1.81044674e+00 1.93249404e+00
-5.09370387e-01 1.95760146e-01 7.11708724e-01 -1.07096553e+00
7.30425298e-01 1.92999482e-01 5.57587981e-01 -1.11178286e-01
2.12124959e-02 2.53170609e-01 -1.64613441e-01 -6.56952083e-01
4.58434194e-01 -7.23268509e-01 4.54531282e-01 4.74081367e-01
1.08913161e-01 -2.62072086e-01 4.28309431e-03 -1.04704676e-02
4.89672720e-01 -4.38325033e-02 4.45356756e-01 -6.33173883e-01
7.66339660e-01 -1.00838818e-01 6.86111867e-01 3.52014601e-01
2.11967424e-01 6.55873954e-01 -1.94097832e-01 -2.79360086e-01
-7.86467969e-01 -8.80829036e-01 -1.94794163e-01 8.63571107e-01
1.91584349e-01 -3.17701668e-01 -5.10586619e-01 -1.84254214e-01
5.88680468e-02 2.71202236e-01 -4.27006483e-01 1.57708615e-01
-3.77995163e-01 -1.38370287e+00 8.64672661e-02 8.21106359e-02
5.41613042e-01 -4.53639954e-01 -2.00875346e-02 1.73833415e-01
-3.29909950e-01 -9.26782668e-01 -5.72817266e-01 2.11240411e-01
-8.58307123e-01 -8.98261309e-01 -8.78321350e-01 -5.20778477e-01
9.26148772e-01 8.04916143e-01 7.42671967e-01 -6.85985312e-02
-7.29673356e-02 2.60307312e-01 -1.24422848e-01 -1.14127636e-01
-9.33365449e-02 -2.45547831e-01 5.57059050e-01 8.16245973e-01
-2.30103806e-01 -9.50536489e-01 -4.96613592e-01 3.40774834e-01
-1.38143420e+00 1.26528338e-01 4.94698405e-01 7.55493879e-01
8.29734564e-01 2.37232640e-01 2.21901491e-01 -7.40845144e-01
5.89914262e-01 -4.10258383e-01 -7.15128601e-01 2.18898863e-01
-3.91123444e-01 3.30418974e-01 4.65430200e-01 -4.86045718e-01
-1.10043979e+00 3.54333699e-01 1.01373278e-01 -8.20963562e-01
6.87374249e-02 6.94095612e-01 -7.36547172e-01 -2.37703636e-01
6.88180745e-01 1.19097836e-01 9.38186720e-02 -7.23674238e-01
8.03195953e-01 6.56198382e-01 8.16097140e-01 -8.97742033e-01
1.26879883e+00 7.49575794e-01 1.47368595e-01 -1.11571038e+00
-1.01828575e+00 -6.67788684e-01 -4.97437298e-01 -1.68367684e-01
8.25803399e-01 -1.20586407e+00 -3.67060184e-01 4.63801742e-01
-7.48402536e-01 -1.48664683e-01 -3.24221134e-01 6.44692779e-01
-2.41123527e-01 7.77407229e-01 -2.18130007e-01 -7.80152082e-01
-5.44627607e-02 -1.03145671e+00 9.04417396e-01 6.60062805e-02
7.18758255e-02 -1.01228225e+00 3.46654236e-01 5.67625999e-01
2.92595595e-01 3.55716765e-01 5.82273662e-01 6.29989579e-02
-3.34352702e-01 9.06530470e-02 -1.70098796e-01 5.33421278e-01
5.98192632e-01 -4.62390366e-04 -1.11694157e+00 -3.19237202e-01
4.83580500e-01 9.78523493e-02 1.05840206e+00 6.09536171e-01
8.37855399e-01 -4.47435230e-01 -1.15047179e-01 8.87418807e-01
1.28596771e+00 -1.71275094e-01 2.79383987e-01 2.20378503e-01
1.20032668e+00 4.97163951e-01 4.55976158e-01 6.37033880e-01
4.12346274e-02 6.04652643e-01 5.07659972e-01 -8.90999362e-02
5.79244830e-02 -9.54458714e-02 2.54885584e-01 1.23819983e+00
-3.47797334e-01 -3.48558240e-02 -7.51391411e-01 3.99209827e-01
-2.11520386e+00 -8.07060361e-01 -2.20003799e-01 2.23457456e+00
7.02851534e-01 -5.23175597e-01 -3.11493184e-02 2.11288154e-01
9.75181103e-01 6.03252649e-01 -5.39580703e-01 4.61530685e-01
-3.26159716e-01 4.77464460e-02 5.02298534e-01 9.60412621e-01
-1.29551232e+00 7.05479383e-01 5.90567780e+00 9.29650247e-01
-1.21628201e+00 2.17376485e-01 2.44443789e-02 -4.13151532e-02
-7.24925101e-01 7.56125003e-02 -3.32401901e-01 3.23826373e-01
5.32699645e-01 1.10799909e-01 1.06177700e+00 3.48852485e-01
3.76403540e-01 6.80838302e-02 -5.18147171e-01 1.46477807e+00
1.63160503e-01 -1.14272118e+00 1.60783842e-01 -1.73902772e-02
9.95357215e-01 -1.63400486e-01 1.88391939e-01 -3.99286062e-01
3.14541698e-01 -1.01803398e+00 6.81218326e-01 6.74606919e-01
6.38231277e-01 -6.62331283e-01 3.33935708e-01 2.90402621e-01
-1.20750368e+00 4.34299894e-02 -4.28239703e-01 -2.60951459e-01
2.83358514e-01 1.24599326e+00 -2.54944384e-01 7.64931619e-01
3.04038882e-01 1.20668733e+00 -2.76200265e-01 8.02956223e-01
-5.04081249e-01 5.27395427e-01 -4.33440506e-01 7.54102290e-01
9.62283313e-02 -9.00464714e-01 7.22929597e-01 7.97681510e-01
4.65507299e-01 1.40261203e-01 3.21075082e-01 7.92680502e-01
3.39173786e-02 1.51931241e-01 -3.28909427e-01 -3.28962088e-01
1.61520779e-01 1.55018020e+00 -7.71863580e-01 -1.95287690e-01
-3.56465489e-01 1.02495408e+00 2.72359159e-02 6.66102409e-01
-6.90274417e-01 3.25564712e-01 1.02662897e+00 1.40097767e-01
2.52023265e-02 -4.39700454e-01 -2.57185753e-02 -1.85006762e+00
-7.84072354e-02 -1.17414320e+00 2.38718212e-01 -7.15939522e-01
-1.55475080e+00 3.65795434e-01 -8.10966566e-02 -1.33454132e+00
2.03337789e-01 -6.20528519e-01 -4.67772394e-01 1.07182479e+00
-1.53670132e+00 -1.15829635e+00 -5.42075872e-01 8.61001432e-01
1.70185536e-01 -1.94323510e-01 6.04510605e-01 5.16374528e-01
-9.63610590e-01 -2.19302982e-01 2.82193393e-01 -2.99236834e-01
7.76168585e-01 -1.06497705e+00 -1.85168236e-01 1.32948673e+00
4.24653053e-01 9.13701952e-01 9.31266725e-01 -6.51288629e-01
-1.60860634e+00 -1.18203056e+00 2.11051151e-01 1.18759675e-02
9.10015583e-01 -1.29119992e-01 -7.28546619e-01 6.20050073e-01
2.09910721e-01 3.52778703e-01 8.41895461e-01 -2.10420080e-02
-2.64265537e-01 -2.23695636e-01 -9.23828840e-01 2.72402644e-01
8.64436388e-01 -6.76174343e-01 -2.51727730e-01 5.59723973e-01
4.81348783e-01 -3.11736763e-01 -7.37754345e-01 4.45875943e-01
3.84951472e-01 -7.27837086e-01 1.27350676e+00 -5.60629308e-01
3.98835093e-02 -1.03346968e+00 -5.54511726e-01 -1.46592426e+00
-5.61698020e-01 -8.55930448e-01 -1.89878896e-01 1.23862660e+00
-3.36205401e-02 -6.41042173e-01 5.20456314e-01 2.06278160e-01
-2.85465792e-02 -1.16054818e-01 -7.10661530e-01 -6.99833453e-01
-2.69968182e-01 -4.86796409e-01 5.43060660e-01 1.39735007e+00
-3.68886471e-01 3.98283809e-01 -1.06205535e+00 7.13117421e-01
9.87195015e-01 2.17283323e-01 6.70652270e-01 -1.22037947e+00
-6.15679145e-01 -2.46377159e-02 -7.48633817e-02 -9.93861973e-01
4.22422737e-01 -8.56697500e-01 8.56754184e-02 -1.25736690e+00
2.02758312e-01 -3.23558271e-01 -2.56715298e-01 2.99517334e-01
-4.89127994e-01 2.71189630e-01 1.35880545e-01 5.95159769e-01
-1.72536582e-01 6.34507418e-01 1.16719532e+00 -4.65948224e-01
-1.71169385e-01 -2.46747434e-01 -6.77939773e-01 7.88625658e-01
5.56298554e-01 -4.05910134e-01 -6.00400746e-01 -6.24984682e-01
5.03122747e-01 -7.80920163e-02 2.57464856e-01 -9.05008554e-01
6.29252046e-02 -5.91676652e-01 1.54547542e-01 -1.90400347e-01
3.23551983e-01 -1.10230219e+00 8.44036341e-01 6.37824386e-02
1.19567476e-01 -1.50378466e-01 1.78303227e-01 6.75974131e-01
-3.82656425e-01 -3.59743118e-01 8.21921229e-01 -1.27517238e-01
-5.70867717e-01 4.01827693e-01 -2.01417178e-01 -5.96072488e-02
6.43427670e-01 1.59146190e-01 1.56196788e-01 -2.74194419e-01
-8.86386573e-01 -7.51950429e-04 5.05926192e-01 2.46970635e-02
4.23476905e-01 -1.55127776e+00 -5.96512616e-01 5.27418852e-01
1.37542915e-02 1.48897752e-01 4.58195627e-01 9.06134367e-01
-5.79386771e-01 2.14552898e-02 -1.88536465e-01 -4.65554208e-01
-1.08389652e+00 7.59426415e-01 3.26389164e-01 1.27260890e-02
-2.15738326e-01 9.12265539e-01 2.97988027e-01 -5.24617195e-01
-8.79453197e-02 -3.39747399e-01 -1.71973482e-02 3.20359856e-01
5.13183296e-01 4.99190360e-01 -1.16779722e-01 -1.00574720e+00
-2.73940206e-01 8.52778554e-01 5.67237616e-01 -1.87672734e-01
1.33133280e+00 -2.80210823e-01 -7.33055532e-01 3.98738325e-01
8.94416153e-01 3.87466699e-01 -1.27391112e+00 -3.85288537e-01
-2.26777866e-01 -5.11788130e-01 4.11112398e-01 -2.50465512e-01
-1.27533138e+00 8.14050794e-01 3.82306129e-01 1.16398543e-01
1.31110191e+00 -4.65531439e-01 5.96544981e-01 9.56739560e-02
1.52963087e-01 -6.36034727e-01 -8.33882168e-02 1.21198565e-01
9.67222810e-01 -1.14215362e+00 3.57123554e-01 -6.61705077e-01
-3.11365932e-01 8.69008422e-01 2.67250031e-01 -1.76972196e-01
7.65365362e-01 1.94422483e-01 3.50909561e-01 -1.20968312e-01
-1.00652575e-02 -3.89692873e-01 5.77115178e-01 3.34195465e-01
3.30501884e-01 3.48908454e-01 -8.71523395e-02 5.60313523e-01
1.51525356e-03 -1.74011052e-01 1.43595248e-01 4.57853436e-01
-4.41894799e-01 -1.25042343e+00 -1.07009423e+00 2.46041566e-01
-1.74927101e-01 -2.29852602e-01 -1.61777496e-01 2.42785513e-01
4.71130669e-01 9.53732371e-01 -2.53022343e-01 -3.17012608e-01
-3.01899714e-03 -3.03683989e-02 3.89224589e-01 -8.03118825e-01
-1.35480091e-01 5.81107020e-01 -2.42695138e-01 -2.69288033e-01
-1.08991385e+00 -8.08008850e-01 -9.39873159e-01 -1.04614891e-01
-4.74323481e-01 3.70243579e-01 2.31853783e-01 9.33224320e-01
9.08087939e-02 3.64368320e-01 5.39574146e-01 -1.17882490e+00
-1.72475398e-01 -9.55202520e-01 -1.01661432e+00 3.62033755e-01
4.92268115e-01 -9.68056798e-01 -5.69499791e-01 3.22246611e-01] | [10.074566841125488, -2.0694520473480225] |
afc2a55b-0598-4398-832a-101edfe24eae | cross-lingual-argument-mining-in-the-medical | 2301.10527 | null | https://arxiv.org/abs/2301.10527v1 | https://arxiv.org/pdf/2301.10527v1.pdf | Cross-lingual Argument Mining in the Medical Domain | Nowadays the medical domain is receiving more and more attention in applications involving Artificial Intelligence. Clinicians have to deal with an enormous amount of unstructured textual data to make a conclusion about patients' health in their everyday life. Argument mining helps to provide a structure to such data by detecting argumentative components in the text and classifying the relations between them. However, as it is the case for many tasks in Natural Language Processing in general and in medical text processing in particular, the large majority of the work on computational argumentation has been done only for English. This is also the case with the only dataset available for argumentation in the medical domain, namely, the annotated medical data of abstracts of Randomized Controlled Trials (RCT) from the MEDLINE database. In order to mitigate the lack of annotated data for other languages, we empirically investigate several strategies to perform argument mining and classification in medical texts for a language for which no annotated data is available. This project shows that automatically translating and project annotations from English to a target language (Spanish) is an effective way to generate annotated data without manual intervention. Furthermore, our experiments demonstrate that the translation and projection approach outperforms zero-shot cross-lingual approaches using a large masked multilingual language model. Finally, we show how the automatically generated data in Spanish can also be used to improve results in the original English evaluation setting. | ['Rodrigo Agerri', 'Anar Yeginbergenova'] | 2023-01-25 | null | null | null | null | ['argument-mining'] | ['natural-language-processing'] | [ 5.06326735e-01 8.09471130e-01 -4.88714337e-01 -2.36135498e-01
-8.94594133e-01 -3.96007389e-01 6.65016294e-01 1.05380630e+00
-7.74633527e-01 1.05164397e+00 6.79190278e-01 -9.10576582e-01
-1.83834314e-01 -6.87581122e-01 -3.80254000e-01 -3.87606949e-01
4.35212433e-01 8.42938244e-01 9.71129257e-03 -4.44536000e-01
3.89405005e-02 2.06570536e-01 -1.21609688e+00 9.62930858e-01
1.01361609e+00 4.80030715e-01 1.17982335e-01 2.84690946e-01
-5.43315232e-01 9.66077805e-01 -5.89434206e-01 -6.99168205e-01
-9.24611762e-02 -6.65939748e-01 -1.15054584e+00 3.95431602e-03
-2.79007852e-01 2.62506217e-01 5.30949473e-01 1.06733692e+00
6.55594051e-01 -4.54522759e-01 4.51455742e-01 -8.10769856e-01
6.52772412e-02 1.17736471e+00 -2.83854455e-01 -9.68935415e-02
7.13091910e-01 -3.76131445e-01 7.46007025e-01 -5.90051532e-01
1.25447321e+00 1.23632479e+00 3.43976706e-01 3.46332580e-01
-9.41807210e-01 -3.25100929e-01 -1.14069492e-01 3.38203497e-02
-8.18049848e-01 -2.04979554e-01 6.67274058e-01 -4.95245546e-01
8.58019054e-01 3.17526877e-01 6.48426354e-01 1.38186955e+00
1.45498112e-01 5.81481040e-01 1.23504877e+00 -1.15068734e+00
1.52119786e-01 5.51568747e-01 1.06380552e-01 5.50994277e-01
4.02973741e-01 -3.26810628e-01 -3.66470575e-01 -4.30518329e-01
-6.47314861e-02 -4.36043620e-01 -2.19094038e-01 6.00467063e-02
-1.39922595e+00 1.18501925e+00 -4.65405099e-02 7.20140636e-01
-4.77134645e-01 -6.37470663e-01 9.74759579e-01 3.82853299e-01
6.58591390e-01 4.34634656e-01 -6.27176225e-01 -7.92884976e-02
-7.44682372e-01 7.51440227e-02 1.24069774e+00 5.83020747e-01
1.24551266e-01 -5.53590119e-01 9.43906680e-02 8.75371158e-01
2.18597278e-01 3.13187242e-01 7.52652347e-01 -6.88854814e-01
9.17852521e-01 1.08457196e+00 -8.89580995e-02 -8.76746356e-01
-4.63879287e-01 -9.52275321e-02 -6.14510894e-01 2.59487238e-02
6.73184276e-01 -3.61613750e-01 -2.80027002e-01 1.49270713e+00
6.04914308e-01 -6.35157883e-01 6.81694865e-01 5.59112430e-01
1.05124331e+00 2.38055199e-01 2.64925599e-01 -5.96686184e-01
1.83234215e+00 -5.48899829e-01 -1.03277528e+00 -1.03339739e-01
1.40578902e+00 -1.18537366e+00 9.06554937e-01 4.74030346e-01
-9.31915998e-01 4.20151427e-02 -8.91281128e-01 -1.20634027e-01
-6.38057351e-01 1.65949240e-01 5.12386799e-01 6.17039859e-01
-4.03325230e-01 2.94138640e-01 -6.05584085e-01 -5.57492375e-01
4.46446627e-01 9.84986573e-02 -5.77493787e-01 -6.55522943e-02
-1.49383092e+00 1.13761795e+00 6.61078513e-01 -7.94767886e-02
7.29696006e-02 -4.25089985e-01 -9.20517623e-01 -2.72952169e-01
7.19176233e-01 -5.76865017e-01 1.03822625e+00 -1.09009933e+00
-9.67537642e-01 1.33539236e+00 -2.99995230e-03 -7.53612399e-01
8.30193460e-01 3.87806259e-02 -3.82094145e-01 1.11948170e-01
4.41911936e-01 4.02294993e-01 2.06394136e-01 -7.29124069e-01
-6.66526079e-01 -5.74520707e-01 2.12661996e-02 1.13749214e-01
-3.14628005e-01 5.02424419e-01 -2.97191888e-02 -8.22421074e-01
-2.59988189e-01 -1.00450230e+00 -4.18151617e-01 -2.08893061e-01
-4.94944602e-01 -4.04624194e-01 4.66283828e-01 -7.69445419e-01
1.42627788e+00 -1.83556247e+00 1.07146360e-01 2.32325375e-01
7.45049790e-02 2.31746227e-01 2.85003692e-01 6.84177041e-01
-1.96049407e-01 3.18603337e-01 -3.71127635e-01 2.17475280e-01
-2.93602884e-01 5.38562536e-01 -2.79709220e-01 2.65299201e-01
6.52171448e-02 7.35710144e-01 -9.19846892e-01 -1.21167219e+00
-5.72713837e-02 3.84332448e-01 -5.45448840e-01 -2.90996820e-01
-2.20310271e-01 5.06876290e-01 -6.82004392e-01 4.19715464e-01
1.54121881e-02 -2.23025665e-01 5.45824826e-01 -8.67120698e-02
-7.12739080e-02 6.29959762e-01 -1.00696218e+00 1.64917660e+00
-5.73604405e-01 4.67766970e-01 -4.39712889e-02 -1.19512856e+00
6.18604004e-01 8.21438372e-01 7.55097628e-01 -6.82999551e-01
2.75288850e-01 7.33605742e-01 3.62124711e-01 -7.65536427e-01
1.72355056e-01 -3.82502288e-01 -1.58512697e-01 5.29956162e-01
-4.74763542e-01 -1.10220104e-01 7.28079200e-01 2.51034111e-01
9.73074377e-01 -2.88155880e-02 7.39614427e-01 -2.95189053e-01
9.65866387e-01 7.14520216e-01 4.23595518e-01 1.16875537e-01
2.62441903e-01 2.86608309e-01 6.90671444e-01 -5.23893178e-01
-9.95428562e-01 -1.71039879e-01 -5.77683330e-01 4.72580850e-01
-4.95728344e-01 -7.03508794e-01 -9.76886630e-01 -8.69741261e-01
-3.62722784e-01 8.53261828e-01 -7.15659082e-01 2.63069630e-01
-5.75019658e-01 -9.64050770e-01 5.91980696e-01 6.29738346e-02
2.12213323e-01 -1.42479038e+00 -1.20518219e+00 5.07678151e-01
-6.69897854e-01 -1.30530381e+00 -2.17605960e-02 2.76554286e-01
-7.92212009e-01 -1.58355975e+00 -4.18081224e-01 -6.83439314e-01
5.57424247e-01 -6.30332708e-01 1.12904429e+00 6.69086492e-03
-5.28266370e-01 8.78209099e-02 -5.25032401e-01 -9.93028641e-01
-1.28673625e+00 4.11551110e-02 -3.38099808e-01 -4.11366850e-01
6.04384005e-01 8.20472538e-02 -3.04700643e-01 1.43332630e-01
-1.09980500e+00 1.98389366e-01 4.30314869e-01 9.43914711e-01
4.58654970e-01 -1.78856682e-02 5.30036867e-01 -1.46715331e+00
1.13216901e+00 -4.43171978e-01 -1.40104741e-01 2.05048293e-01
-5.92081487e-01 2.67541260e-01 4.98308271e-01 -3.54000539e-01
-9.83280241e-01 -1.68068811e-01 -4.35205162e-01 4.47475880e-01
-2.95255274e-01 1.08139634e+00 4.22253758e-02 6.37533724e-01
9.48399127e-01 -5.20905256e-01 2.60752022e-01 -3.60905468e-01
1.16356596e-01 9.29122865e-01 1.84233375e-02 -4.96398747e-01
3.65776867e-02 4.49982256e-01 6.99011013e-02 -8.46048713e-01
-9.89559114e-01 -3.16566736e-01 -5.31879425e-01 4.10254262e-02
9.53823447e-01 -6.51415884e-01 -4.57810313e-01 -3.26216429e-01
-1.21332169e+00 -1.51002795e-01 -6.33781970e-01 8.08412731e-01
-4.35615033e-01 2.93305814e-01 -4.22126025e-01 -6.36229753e-01
-4.97466594e-01 -1.26593935e+00 9.54093814e-01 -3.50996852e-01
-8.73861015e-01 -1.07749176e+00 2.08495706e-01 5.77905416e-01
-8.51056129e-02 4.99022186e-01 1.37189889e+00 -1.02879095e+00
4.60695952e-01 -2.70600051e-01 9.10913274e-02 1.02894187e-01
2.56571978e-01 -1.51352495e-01 -4.71941620e-01 8.90352353e-02
3.21146429e-01 -4.80376154e-01 3.81968975e-01 3.13443869e-01
6.52196407e-01 -4.48705256e-01 -4.02864277e-01 -3.20277154e-01
1.07244992e+00 1.14378586e-01 4.56982166e-01 6.28825188e-01
3.45939159e-01 1.43192959e+00 8.70839119e-01 2.40045160e-01
3.21135044e-01 7.64841259e-01 5.09497561e-02 -2.30608672e-01
-4.61052805e-02 4.34360318e-02 2.24030405e-01 7.20668554e-01
-1.21619008e-01 -9.92685631e-02 -1.13952732e+00 5.66829860e-01
-1.88742006e+00 -8.84538054e-01 -5.25022149e-01 2.12152433e+00
1.27891338e+00 2.28286088e-01 -1.38048930e-02 5.81277251e-01
4.94018108e-01 -5.04509747e-01 1.34264231e-02 -6.87745333e-01
-2.11161330e-01 2.84321189e-01 2.85896301e-01 3.99795920e-01
-8.99233460e-01 5.59501529e-01 5.02818060e+00 6.25403583e-01
-8.85776460e-01 1.57067835e-01 7.06158996e-01 1.65356383e-01
-1.90803215e-01 6.33208454e-02 -5.47148526e-01 4.50440973e-01
1.18138146e+00 -1.67015374e-01 -2.50042915e-01 6.29954278e-01
6.20717168e-01 -4.08109486e-01 -1.06650579e+00 7.73959994e-01
3.01017286e-03 -1.31025589e+00 -8.05427209e-02 3.41558367e-01
4.33139712e-01 -1.03948727e-01 -4.54880804e-01 -4.98255342e-02
1.60040110e-01 -9.88744140e-01 5.37189662e-01 1.29339442e-01
5.94469905e-01 -5.41903615e-01 1.17662001e+00 6.36651456e-01
-5.90157270e-01 7.92009532e-02 -1.23543970e-01 1.76049933e-01
3.18917394e-01 7.13445783e-01 -1.21569324e+00 7.37159431e-01
5.42460740e-01 4.36588943e-01 -5.27525842e-01 5.84901512e-01
-4.26151723e-01 5.78525186e-01 -2.73250520e-01 -1.25172213e-01
3.03909183e-01 -2.15446472e-01 5.24289548e-01 1.32603610e+00
3.22875619e-01 8.46634433e-02 1.89483702e-01 3.55217755e-01
-4.11315486e-02 9.58202481e-01 -9.03712213e-01 -2.27619782e-01
-1.67637974e-01 1.08963168e+00 -1.02927899e+00 -4.35334235e-01
-4.79778856e-01 5.40665209e-01 -3.06522418e-02 -1.12666160e-01
-4.86093014e-01 -1.13788173e-01 -1.39743418e-01 4.30568069e-01
-2.32532218e-01 3.26427579e-01 -1.65150583e-01 -9.11786616e-01
5.21340519e-02 -1.44829679e+00 8.30410063e-01 -5.35077333e-01
-1.09959447e+00 9.03841078e-01 3.07387531e-01 -1.26843810e+00
-7.81589687e-01 -7.23960221e-01 1.42443940e-01 9.18128431e-01
-1.14515066e+00 -1.16635025e+00 1.85454980e-01 6.19977653e-01
6.99344397e-01 -1.63849056e-01 1.15435016e+00 3.56373399e-01
-1.47804320e-01 6.80726692e-02 -2.54204929e-01 2.46097043e-01
1.00055265e+00 -1.09635830e+00 -2.81762123e-01 3.83870572e-01
2.45452538e-01 6.13117635e-01 8.99145484e-01 -8.50071609e-01
-9.75133419e-01 -5.65608084e-01 1.49936068e+00 -4.29967374e-01
8.20256650e-01 6.04633503e-02 -8.92747819e-01 2.71733493e-01
5.00308096e-01 -4.51046199e-01 1.01435721e+00 -3.03639434e-02
3.88249918e-03 4.33083355e-01 -1.09854352e+00 7.40231216e-01
6.03995442e-01 -2.60640174e-01 -1.02587032e+00 6.68588817e-01
3.43931884e-01 -4.52356011e-01 -9.25963879e-01 3.70197177e-01
3.85981768e-01 -4.63058650e-01 6.41479254e-01 -8.59270275e-01
8.62521410e-01 -1.76626652e-01 9.76663083e-03 -1.12628150e+00
6.90891623e-01 -3.65166426e-01 4.46490139e-01 9.03884232e-01
1.01089883e+00 -6.83325648e-01 5.85141301e-01 5.78771889e-01
8.33908692e-02 -7.78644085e-01 -8.87600303e-01 -1.82238892e-01
2.00632229e-01 -6.06261730e-01 1.47013560e-01 1.16466260e+00
3.98847342e-01 6.48406327e-01 1.20310150e-01 -4.63026911e-01
2.74606824e-01 2.84040689e-01 4.50837851e-01 -1.40494406e+00
-9.93606001e-02 -3.24688226e-01 -2.27994144e-01 2.17190366e-02
2.06877172e-01 -1.19148433e+00 -2.61619240e-01 -1.96203363e+00
2.16558531e-01 -4.67048407e-01 2.52836078e-01 5.55908680e-01
1.22965723e-01 9.96922553e-02 -1.55502960e-01 1.94513068e-01
-1.19125202e-01 -3.94712873e-02 1.02182841e+00 -2.07268760e-01
-2.77122855e-01 -5.36395311e-02 -7.88380921e-01 1.03483260e+00
6.20843828e-01 -8.88692439e-01 -4.38220650e-01 1.76974125e-02
6.48920178e-01 8.72833133e-02 8.73422157e-03 -3.66502434e-01
1.12541959e-01 -6.78273337e-03 -1.11958325e-01 -5.48088729e-01
-2.27476522e-01 -1.00489581e+00 1.58211395e-01 7.82720625e-01
-5.39238214e-01 2.12644026e-01 2.62129992e-01 3.43509912e-01
-5.20980775e-01 -6.12372100e-01 4.35855031e-01 -3.09325933e-01
-8.29256624e-02 -3.84287745e-01 -4.81208205e-01 3.71801198e-01
9.37659442e-01 -1.22051507e-01 -3.34356427e-01 -8.95583034e-02
-9.67662215e-01 1.72631979e-01 2.54547268e-01 4.15081441e-01
3.11374247e-01 -9.60039675e-01 -1.15003347e+00 -2.15499505e-01
2.70271659e-01 -9.96149033e-02 -6.17041700e-02 1.23320282e+00
-7.67522693e-01 6.02738380e-01 -3.88777303e-03 -3.67775381e-01
-1.66230762e+00 7.09194601e-01 -9.11735520e-02 -9.44528997e-01
-8.22111249e-01 9.17943269e-02 -1.43850163e-01 -3.21412206e-01
1.02134474e-01 -3.59463423e-01 -6.98716104e-01 5.92530012e-01
4.69059467e-01 1.51440268e-02 4.89324003e-01 -7.87865341e-01
-3.16750526e-01 1.73398852e-01 -9.83094573e-02 -4.79474425e-01
1.52784729e+00 1.32825002e-01 -4.52930301e-01 5.23378730e-01
8.79565597e-01 5.31783402e-01 -9.59460661e-02 8.54234025e-02
5.41903734e-01 1.47072464e-01 -2.71601319e-01 -9.12828326e-01
-5.04017770e-01 7.15462089e-01 2.47581482e-01 3.30657393e-01
8.74486923e-01 1.41313165e-01 1.80147976e-01 5.18354297e-01
2.61923701e-01 -1.37193811e+00 -2.84181625e-01 2.16559678e-01
1.17110956e+00 -1.32483709e+00 1.96778283e-01 -6.51920199e-01
-9.09294844e-01 1.20157826e+00 -1.66479245e-01 5.67040622e-01
6.83027446e-01 5.09181798e-01 2.68255115e-01 -5.23395836e-01
-5.85760772e-01 -6.57410845e-02 9.35583636e-02 2.12139249e-01
1.03270054e+00 4.01682453e-03 -1.28635538e+00 6.16735637e-01
-3.53643090e-01 1.57623410e-01 6.26653552e-01 1.07051575e+00
1.96554571e-01 -1.85866952e+00 -5.84860682e-01 5.06037593e-01
-1.21839845e+00 -3.49091023e-01 -7.48607159e-01 9.60768342e-01
1.15716793e-01 1.09031415e+00 -3.93780202e-01 4.58808869e-01
4.51388568e-01 3.32012266e-01 3.15196574e-01 -8.78285110e-01
-9.17060673e-01 2.35635921e-01 7.83080339e-01 -2.72233307e-01
-1.06996667e+00 -6.19602025e-01 -1.38971448e+00 1.82454824e-01
-2.55901039e-01 6.03503704e-01 8.48111153e-01 1.18252361e+00
4.86722700e-02 6.58361316e-01 -9.71921608e-02 -5.61932437e-02
-2.58308679e-01 -8.45278144e-01 -2.35076603e-02 5.74959338e-01
-2.19050154e-01 -3.47742826e-01 -1.66672692e-01 3.14533234e-01] | [8.572619438171387, 8.77623462677002] |
7749b4d5-7f21-46e0-8d65-ed0f62d0982f | completing-partial-point-clouds-with-outliers | 2203.09772 | null | https://arxiv.org/abs/2203.09772v1 | https://arxiv.org/pdf/2203.09772v1.pdf | Completing Partial Point Clouds with Outliers by Collaborative Completion and Segmentation | Most existing point cloud completion methods are only applicable to partial point clouds without any noises and outliers, which does not always hold in practice. We propose in this paper an end-to-end network, named CS-Net, to complete the point clouds contaminated by noises or containing outliers. In our CS-Net, the completion and segmentation modules work collaboratively to promote each other, benefited from our specifically designed cascaded structure. With the help of segmentation, more clean point cloud is fed into the completion module. We design a novel completion decoder which harnesses the labels obtained by segmentation together with FPS to purify the point cloud and leverages KNN-grouping for better generation. The completion and segmentation modules work alternately share the useful information from each other to gradually improve the quality of prediction. To train our network, we build a dataset to simulate the real case where incomplete point clouds contain outliers. Our comprehensive experiments and comparisons against state-of-the-art completion methods demonstrate our superiority. We also compare with the scheme of segmentation followed by completion and their end-to-end fusion, which also proves our efficacy. | ['Yanwen Guo', 'Chongjun Wang', 'Jie Guo', 'Yang Yang', 'Changfeng Ma'] | 2022-03-18 | null | null | null | null | ['point-cloud-completion'] | ['computer-vision'] | [ 1.29610032e-01 -5.03371730e-02 3.72765034e-01 -3.61057341e-01
-9.02363777e-01 -4.97226119e-01 2.98611045e-01 4.79565822e-02
-1.20182067e-01 3.14452320e-01 -1.65305659e-01 5.15993536e-02
1.17524505e-01 -6.25701785e-01 -1.13934648e+00 -5.96357584e-01
1.30776972e-01 5.14406681e-01 3.11057031e-01 -5.92441596e-02
2.36459896e-01 4.38090175e-01 -1.38507652e+00 5.79115376e-02
1.22872353e+00 9.21506763e-01 6.39678240e-01 4.50592041e-01
-1.88927665e-01 5.24331927e-01 -3.22667181e-01 -3.24401855e-01
6.44140065e-01 1.55601069e-01 -2.76898563e-01 4.38872457e-01
4.55813974e-01 -4.56329584e-01 -3.26934010e-01 9.40108478e-01
3.86346489e-01 7.56869763e-02 3.17844391e-01 -1.21361780e+00
-3.63816857e-01 3.06045413e-01 -8.43542933e-01 -4.51669663e-01
9.82979760e-02 3.39758754e-01 7.71080673e-01 -1.19335949e+00
3.56746227e-01 1.13201487e+00 9.88230169e-01 1.66585803e-01
-9.48001206e-01 -8.24679613e-01 4.15124059e-01 -2.71756589e-01
-1.25764728e+00 -3.02318871e-01 9.71778154e-01 -4.18231636e-01
4.11294103e-01 -1.48850784e-03 6.11111104e-01 7.13329971e-01
-1.13183171e-01 8.91277373e-01 5.56258082e-01 2.21129823e-02
2.42134765e-01 -3.27271074e-01 -5.29963337e-02 5.00379443e-01
1.72488168e-01 1.06065661e-01 -1.86254442e-01 -2.75079846e-01
6.67490482e-01 5.88633776e-01 -3.09176773e-01 -5.66166818e-01
-1.29854035e+00 5.20474553e-01 7.62497604e-01 -2.59197354e-01
-7.44877458e-01 2.81580001e-01 1.61896393e-01 2.32709795e-02
5.31697094e-01 6.77639842e-02 -4.11220551e-01 2.94231415e-01
-1.29690039e+00 3.35966378e-01 5.02420485e-01 1.41210377e+00
1.04102957e+00 -1.84491754e-01 -8.81345123e-02 7.11747527e-01
5.41804552e-01 4.80538338e-01 -9.84745026e-02 -1.01821864e+00
6.50894284e-01 6.82947040e-01 3.02566856e-01 -8.14961076e-01
-2.61916131e-01 -7.65536606e-01 -8.34150195e-01 2.29830563e-01
5.46602979e-02 -2.63783395e-01 -1.12450910e+00 1.39559305e+00
5.22709310e-01 9.28312480e-01 -4.29347046e-02 1.03965712e+00
5.07982075e-01 5.52220881e-01 -9.69550461e-02 9.42941457e-02
7.99423873e-01 -1.06660593e+00 -2.70085543e-01 -1.62991911e-01
5.70329845e-01 -8.84566545e-01 8.32061529e-01 4.44679588e-01
-9.23940539e-01 -7.27675319e-01 -9.12607491e-01 -9.12518725e-02
5.30547686e-02 3.12056631e-01 4.40825671e-01 -1.99097022e-02
-8.42071176e-01 9.02953088e-01 -1.16086936e+00 -5.83570637e-02
7.60109901e-01 3.93853039e-01 -1.93509594e-01 -3.03665340e-01
-4.07925308e-01 2.57314980e-01 2.34926820e-01 5.42757630e-01
-1.07512689e+00 -7.65077710e-01 -6.74072742e-01 3.02979257e-02
4.87460196e-01 -1.00776076e+00 1.26242018e+00 -7.49046445e-01
-1.10861731e+00 3.43812734e-01 -3.27410638e-01 -3.16971302e-01
7.36471474e-01 -4.98878390e-01 1.51494339e-01 -1.34597095e-02
4.21112537e-01 9.44380105e-01 7.70942152e-01 -1.74442863e+00
-9.36052144e-01 -4.43714678e-01 -1.66597337e-01 8.12115222e-02
1.75978214e-01 -3.78090084e-01 -1.08777845e+00 -5.58432221e-01
5.49059331e-01 -9.72905219e-01 -6.76532090e-01 5.23973145e-02
-7.13892817e-01 -1.89999220e-04 9.05147076e-01 -5.31855941e-01
8.44989121e-01 -2.36868453e+00 -2.03338526e-02 4.37344790e-01
3.83683175e-01 8.70771110e-02 -2.65476555e-01 5.78279972e-01
-1.25974581e-01 -1.28041282e-01 -5.47974348e-01 -1.21325219e+00
9.44102090e-03 3.55122477e-01 -6.32595539e-01 6.79319441e-01
4.66986060e-01 6.81411743e-01 -1.05443656e+00 -2.90160507e-01
4.21291679e-01 4.27150756e-01 -5.93800128e-01 2.95283943e-01
-4.17501152e-01 7.23561049e-01 -6.18878901e-01 8.41715753e-01
1.29930878e+00 -3.15818846e-01 -4.17540759e-01 2.36130543e-02
-1.80251971e-01 -1.30878165e-01 -1.21490848e+00 2.06509137e+00
-2.48376757e-01 1.30940050e-01 3.35609704e-01 -5.86301863e-01
1.07812309e+00 1.64692044e-01 5.39494514e-01 -1.58243313e-01
-1.29277870e-01 3.32836896e-01 -2.78515428e-01 -3.04103822e-01
5.19849598e-01 2.92191710e-02 1.61874145e-01 -7.42346272e-02
-1.76335156e-01 -3.16953421e-01 -1.00417875e-01 3.98975581e-01
1.16323674e+00 4.88979042e-01 -3.23477358e-01 2.65210867e-01
4.20540780e-01 1.03672363e-01 8.60283971e-01 6.55140281e-01
2.43403576e-02 1.27245641e+00 3.15964878e-01 -1.85852513e-01
-1.04993665e+00 -9.48230684e-01 7.38836229e-02 5.31627715e-01
5.06181002e-01 -4.10669684e-01 -5.12626588e-01 -6.53757036e-01
1.43314227e-01 5.50989568e-01 -3.17781687e-01 1.65651917e-01
-4.27916676e-01 -4.06835079e-01 1.72576383e-01 5.87110102e-01
3.29494715e-01 -8.60986769e-01 -2.50723243e-01 3.79523486e-01
-4.99588214e-02 -1.25004947e+00 -3.86256903e-01 5.03756627e-02
-1.05182886e+00 -1.04694164e+00 -7.43640363e-01 -6.92529619e-01
8.58549833e-01 7.06957102e-01 1.07065034e+00 3.69933486e-01
2.53221154e-01 1.75709888e-01 -5.11113703e-01 -4.70920026e-01
-1.95977762e-01 1.90658346e-01 -2.00430244e-01 1.12421803e-01
6.74546659e-02 -9.44169760e-01 -8.23589385e-01 2.47040838e-01
-1.11828971e+00 1.29348323e-01 7.29878128e-01 5.37533164e-01
9.02734995e-01 -8.68977159e-02 3.33715200e-01 -7.51967728e-01
1.58718258e-01 -6.16202354e-01 -8.89304042e-01 -1.69762131e-02
-1.11446284e-01 -2.45617241e-01 7.34492600e-01 1.23220325e-01
-7.76253879e-01 7.58882463e-01 -2.69560426e-01 -1.26210093e+00
-2.05326036e-01 3.83126140e-01 -2.53282934e-01 -1.07976682e-01
2.91536152e-01 1.52420983e-01 -1.28847584e-01 -7.33319759e-01
5.05753160e-01 5.42067826e-01 8.34103882e-01 -3.31769973e-01
1.13919389e+00 7.85375893e-01 -1.19200610e-01 -4.96473342e-01
-6.67475224e-01 -7.92378902e-01 -5.82551658e-01 -1.74170882e-01
5.69609106e-01 -1.34198534e+00 -6.21035933e-01 4.95894998e-01
-1.49561882e+00 8.18582345e-03 -2.54019499e-01 3.79030526e-01
-4.28300500e-01 6.41509295e-01 -5.00049233e-01 -8.06324124e-01
-3.16794008e-01 -1.24894691e+00 1.66401517e+00 5.48926629e-02
4.52895850e-01 -4.94231045e-01 -4.56256382e-02 1.29508972e-01
7.99996220e-03 3.91528755e-01 2.59446144e-01 -5.60253084e-01
-1.13122201e+00 -3.13583106e-01 -3.42022330e-01 4.84153092e-01
-9.66869518e-02 2.15274543e-01 -1.02701306e+00 -3.97815585e-01
1.03883155e-01 -3.18240821e-02 1.05583978e+00 2.09662139e-01
1.39246726e+00 6.42703399e-02 -4.75964844e-01 9.04013276e-01
1.59946454e+00 -1.85256898e-01 6.25055909e-01 5.09244110e-03
1.08889651e+00 4.69612181e-01 7.47155011e-01 7.16902196e-01
5.17165065e-01 3.02080125e-01 1.02131987e+00 -2.64535457e-01
1.51869684e-01 -4.21913385e-01 2.23973900e-01 9.33878183e-01
1.38966739e-01 -2.38756612e-01 -1.00267255e+00 6.36338055e-01
-2.08184576e+00 -5.44771135e-01 -4.60505515e-01 2.15797663e+00
3.21766615e-01 1.13064095e-01 -1.08345196e-01 -2.22131401e-01
8.05810332e-01 1.21474579e-01 -5.47415972e-01 4.61151689e-01
4.97044697e-02 8.60492885e-03 6.41165257e-01 3.57431650e-01
-1.08028603e+00 9.75713849e-01 5.02008390e+00 8.79974544e-01
-1.04619503e+00 -3.56196724e-02 4.61470246e-01 -1.12973914e-01
-3.82313967e-01 4.41705465e-01 -5.32785892e-01 6.00603342e-01
4.06048328e-01 5.06452024e-01 2.52455860e-01 9.42262411e-01
3.92225236e-01 -4.29090522e-02 -1.04285121e+00 9.68920350e-01
-2.73696303e-01 -1.19831121e+00 -1.21064976e-01 -6.44030645e-02
8.22742939e-01 6.39843702e-01 -2.40651652e-01 1.64275825e-01
4.64593858e-01 -6.18988514e-01 7.56238461e-01 7.27800131e-01
3.27617437e-01 -7.10364521e-01 7.54339635e-01 5.59904993e-01
-1.25040317e+00 4.04904783e-02 -5.79907715e-01 9.01798904e-02
5.37887096e-01 1.02332711e+00 -8.04489315e-01 1.15479374e+00
4.66528118e-01 1.02679944e+00 -4.90854889e-01 1.61939263e+00
-2.71632642e-01 5.97670078e-01 -6.88563526e-01 6.11036360e-01
4.24923837e-01 -5.08431733e-01 7.57388592e-01 9.80195940e-01
5.57701051e-01 9.89544205e-03 4.95941222e-01 1.03265047e+00
-9.12656933e-02 -1.99640542e-01 -4.81483132e-01 3.09183180e-01
6.34233832e-01 1.38389325e+00 -7.99175024e-01 -4.51891780e-01
-5.96989036e-01 9.48448062e-01 2.91904330e-01 3.59294564e-01
-9.25677538e-01 -3.08640420e-01 5.42809248e-01 -3.98013974e-03
6.70708358e-01 -3.34758759e-01 -4.96481776e-01 -1.07749403e+00
3.61454248e-01 -5.42146325e-01 -1.10620081e-01 -8.46937060e-01
-1.62143338e+00 5.21017551e-01 -3.51355195e-01 -1.78441465e+00
2.29034185e-01 -1.31861880e-01 -8.58030379e-01 9.31326687e-01
-1.68687391e+00 -1.31868422e+00 -7.43417442e-01 4.37677711e-01
4.52707499e-01 3.07623744e-01 2.01076865e-01 4.05027628e-01
-6.78137660e-01 1.51941836e-01 2.06152260e-01 -5.91588533e-03
6.39771163e-01 -1.17540431e+00 8.07362556e-01 1.10959661e+00
-1.90685108e-01 4.89025712e-01 4.43131804e-01 -1.01902914e+00
-1.33969891e+00 -1.60962176e+00 4.71942991e-01 -2.31826946e-01
3.23915303e-01 -4.76477504e-01 -9.84003723e-01 6.21522725e-01
-6.83433786e-02 1.88281894e-01 1.65686294e-01 -6.09733015e-02
-1.07258983e-01 -2.38064587e-01 -8.70196581e-01 5.08654118e-01
1.13423181e+00 -1.82492569e-01 -3.96700531e-01 4.99669701e-01
1.17395604e+00 -6.12047851e-01 -5.04569709e-01 5.63064456e-01
3.63415740e-02 -1.05596411e+00 8.47215354e-01 -2.02080444e-01
6.03235483e-01 -8.44847679e-01 -1.84584588e-01 -1.18475986e+00
-1.81021422e-01 -6.55473828e-01 1.86437637e-01 1.32070279e+00
2.81673551e-01 -3.84242922e-01 9.26817775e-01 5.95281482e-01
-8.13755751e-01 -6.96626902e-01 -7.44172871e-01 -6.61503196e-01
-3.19189280e-01 -8.31120253e-01 9.47944403e-01 7.06614554e-01
-5.59352875e-01 2.81922042e-01 -2.66740143e-01 7.74912715e-01
6.73276842e-01 3.27133656e-01 1.30067623e+00 -1.23043561e+00
-1.94200188e-01 -1.15393989e-01 -2.59642303e-01 -1.50194359e+00
-6.64269030e-02 -7.81982362e-01 3.76521736e-01 -1.64186072e+00
3.65861459e-03 -8.50329936e-01 -2.11889930e-02 4.34186667e-01
-4.39387143e-01 1.03234105e-01 5.31287551e-01 5.42360604e-01
-8.30103338e-01 9.23751831e-01 1.31009150e+00 6.99809045e-02
-4.29210395e-01 1.58558533e-01 -6.21709347e-01 7.78827190e-01
5.28577030e-01 -6.76686525e-01 -3.58233660e-01 -7.36862123e-01
9.60829556e-02 2.52621442e-01 4.38723713e-01 -1.44182837e+00
3.82746667e-01 2.15203658e-01 3.87929678e-01 -1.19253981e+00
4.39933062e-01 -1.21347308e+00 2.19870880e-01 1.21516965e-01
9.33237076e-02 -9.24915075e-03 -2.85029057e-02 9.15663421e-01
-2.80496061e-01 -1.07146427e-01 4.66715366e-01 -6.23573251e-02
-3.78546387e-01 7.98000574e-01 3.36521477e-01 -3.98427606e-01
1.02235723e+00 -2.14528441e-01 3.27300057e-02 -3.28738719e-01
-7.06022978e-01 8.25654387e-01 8.15310180e-01 2.81829596e-01
6.60027683e-01 -1.04956961e+00 -7.94182301e-01 2.30090082e-01
1.53663501e-01 1.08106208e+00 4.09306020e-01 9.52931166e-01
-5.74172437e-01 -2.84664780e-02 3.29492539e-01 -1.05155385e+00
-7.86999226e-01 7.34150946e-01 5.81627265e-02 -9.48671624e-02
-7.31814384e-01 7.36319780e-01 3.85459721e-01 -6.57513618e-01
3.29309762e-01 -6.98718965e-01 2.78457016e-01 -2.26405263e-01
1.70910865e-01 2.35107377e-01 2.26814076e-01 -3.87847066e-01
-1.59882635e-01 5.23425281e-01 -6.89381137e-02 5.79104498e-02
1.41394150e+00 -5.70162572e-02 -1.71243072e-01 1.97922856e-01
1.03893423e+00 7.94446021e-02 -1.66729438e+00 -2.87116945e-01
-2.11905524e-01 -5.96789539e-01 -9.24970582e-02 -4.06701207e-01
-1.39105582e+00 8.65059614e-01 3.54298770e-01 -6.68250769e-02
1.18446457e+00 -1.52889535e-01 1.04308450e+00 1.99821889e-01
3.40330422e-01 -8.10968757e-01 -2.33977020e-01 4.64667499e-01
7.32279003e-01 -1.32763803e+00 -1.58089057e-01 -7.05168188e-01
-5.42863965e-01 9.53092456e-01 6.40715837e-01 -4.93647456e-01
5.30917585e-01 2.30318740e-01 -4.11932580e-02 -2.31905654e-01
-5.50246835e-01 -3.07876974e-01 -1.35210812e-01 3.63023728e-01
-1.12331256e-01 -3.92735638e-02 1.00287423e-01 6.07170105e-01
-1.42270178e-01 3.06007415e-01 2.60286719e-01 9.55425203e-01
-4.63378042e-01 -1.02658439e+00 -6.83397412e-01 4.94006842e-01
-4.52347882e-02 -2.81140730e-02 -1.84679583e-01 4.76562619e-01
3.65798891e-01 9.57214117e-01 1.69667423e-01 -4.95692283e-01
4.14017051e-01 -2.70875961e-01 -1.32696286e-01 -6.09007835e-01
-6.32274747e-01 4.34016407e-01 -3.00686091e-01 -7.60914028e-01
-3.13055158e-01 -6.93920553e-01 -1.42375875e+00 -3.94004285e-01
-5.49226165e-01 6.59077615e-02 5.97424567e-01 7.48255610e-01
7.42423356e-01 5.46040475e-01 8.96481693e-01 -1.39415884e+00
-4.86291856e-01 -9.27538633e-01 -4.44358379e-01 3.35859895e-01
6.76798105e-01 -3.61896545e-01 -3.02269429e-01 -1.21157095e-01] | [8.275812149047852, -3.504476547241211] |
2ad75b3f-ba55-4243-bada-4438fb5984a5 | medvit-a-robust-vision-transformer-for | 2302.09462 | null | https://arxiv.org/abs/2302.09462v1 | https://arxiv.org/pdf/2302.09462v1.pdf | MedViT: A Robust Vision Transformer for Generalized Medical Image Classification | Convolutional Neural Networks (CNNs) have advanced existing medical systems for automatic disease diagnosis. However, there are still concerns about the reliability of deep medical diagnosis systems against the potential threats of adversarial attacks since inaccurate diagnosis could lead to disastrous consequences in the safety realm. In this study, we propose a highly robust yet efficient CNN-Transformer hybrid model which is equipped with the locality of CNNs as well as the global connectivity of vision Transformers. To mitigate the high quadratic complexity of the self-attention mechanism while jointly attending to information in various representation subspaces, we construct our attention mechanism by means of an efficient convolution operation. Moreover, to alleviate the fragility of our Transformer model against adversarial attacks, we attempt to learn smoother decision boundaries. To this end, we augment the shape information of an image in the high-level feature space by permuting the feature mean and variance within mini-batches. With less computational complexity, our proposed hybrid model demonstrates its high robustness and generalization ability compared to the state-of-the-art studies on a large-scale collection of standardized MedMNIST-2D datasets. | ['Ahmad Ayatollahi', 'Shahriar B. Shokouhi', 'Hossein Kashiani', 'Hamid Ahmadabadi', 'Omid Nejati Manzari'] | 2023-02-19 | null | null | null | null | ['medical-diagnosis'] | ['medical'] | [ 5.26333507e-03 2.40878224e-01 3.98601592e-01 -1.66815564e-01
-5.65953016e-01 -4.83876705e-01 2.85885960e-01 3.63859236e-02
-5.23964584e-01 3.61911893e-01 1.57891154e-01 -4.52783406e-01
-3.21883500e-01 -7.61280656e-01 -7.29743779e-01 -8.62043619e-01
-5.75089864e-02 1.05980419e-01 -4.48812656e-02 -1.22627035e-01
-1.67962074e-01 7.49445319e-01 -9.63859856e-01 1.84215710e-01
1.07996559e+00 1.22055125e+00 -2.35701934e-01 2.96531379e-01
5.23972869e-01 9.62033272e-01 -6.40003502e-01 -7.71916211e-01
2.53976941e-01 -7.99208805e-02 -8.45640242e-01 6.72552064e-02
4.66266960e-01 -5.28815925e-01 -8.11374903e-01 1.50529134e+00
7.12263584e-01 -1.48755223e-01 4.42511916e-01 -1.06424475e+00
-1.04347122e+00 2.79175460e-01 -2.73781091e-01 3.42430115e-01
-1.86767340e-01 4.67580229e-01 6.23655379e-01 -4.78263944e-01
3.91392350e-01 1.05907476e+00 7.77390420e-01 6.82081819e-01
-9.55272079e-01 -7.71071374e-01 2.98898518e-02 2.69108146e-01
-1.33477497e+00 -2.25014791e-01 8.66658330e-01 -3.72358054e-01
7.29710162e-01 2.60414720e-01 4.52086598e-01 1.41930306e+00
5.83990216e-01 6.21265054e-01 7.69478083e-01 2.51503557e-01
8.47794265e-02 1.46265343e-01 8.59218836e-02 8.78744781e-01
2.00723067e-01 2.13791713e-01 -2.12578960e-02 -4.05505985e-01
8.20551813e-01 3.12319547e-01 -4.92924333e-01 -1.93182871e-01
-1.15834153e+00 9.70952749e-01 1.06781673e+00 4.04368013e-01
-3.86142880e-01 9.90488082e-02 5.62816679e-01 2.07392320e-01
3.85423571e-01 6.13829911e-01 -4.01542574e-01 4.88570571e-01
-4.93226856e-01 -5.22804447e-02 2.63253182e-01 4.70128655e-01
-5.48081025e-02 8.57295245e-02 -3.95212680e-01 3.93984914e-01
3.16764377e-02 2.46230602e-01 5.65025806e-01 -4.97924149e-01
5.29636800e-01 5.91181219e-01 -1.85985982e-01 -1.32490599e+00
-6.19779229e-01 -1.02950084e+00 -1.62646675e+00 1.99521221e-02
2.58639008e-01 -9.78131592e-02 -8.38624239e-01 1.76353645e+00
4.27333981e-01 3.79393756e-01 1.47000000e-01 9.66941118e-01
7.17922211e-01 1.61547408e-01 1.29666105e-01 1.61955118e-01
1.58695853e+00 -8.00412178e-01 -6.78185225e-01 7.40981176e-02
4.90047365e-01 -4.00771350e-01 8.00724328e-01 2.70499051e-01
-9.10584033e-01 -4.98934388e-01 -9.84248340e-01 -2.44745612e-01
-3.06256920e-01 1.25617385e-01 5.65603077e-01 6.11119390e-01
-9.01552320e-01 7.47356355e-01 -1.16858137e+00 -3.17119481e-03
9.51420426e-01 5.03350854e-01 -5.75637639e-01 9.96813923e-03
-1.52141082e+00 9.12984490e-01 1.13264263e-01 3.71827602e-01
-1.04639590e+00 -1.01158738e+00 -9.79925990e-01 2.79577762e-01
2.04555586e-01 -9.36114609e-01 7.24169016e-01 -7.65301585e-01
-1.19338763e+00 6.82690859e-01 4.59713846e-01 -7.27283895e-01
6.82131827e-01 -3.06498021e-01 -2.17722535e-01 3.99076819e-01
-1.33676454e-01 5.29064655e-01 8.99191916e-01 -7.42266715e-01
-2.05509141e-01 -5.79568386e-01 6.45414293e-02 -1.25319004e-01
-8.26318264e-01 -9.66262296e-02 -1.06426850e-01 -1.04431915e+00
-1.19138092e-01 -9.35287535e-01 -5.89326322e-01 4.02477860e-01
-7.14306116e-01 6.28876612e-02 7.57450342e-01 -7.07652569e-01
9.15974379e-01 -2.57722068e+00 6.30679503e-02 1.51398584e-01
5.69337726e-01 5.43648422e-01 -7.83214942e-02 -1.42629221e-01
-1.92202657e-01 9.79812294e-02 -1.98112592e-01 -2.71445841e-01
-3.20370317e-01 1.21155821e-01 -3.24810743e-01 9.26359236e-01
4.58197773e-01 1.00398231e+00 -9.25049901e-01 -3.44299495e-01
2.16732547e-01 8.51601720e-01 -7.10404217e-01 2.01276571e-01
2.46892020e-01 6.14875495e-01 -7.51617253e-01 6.44848764e-01
6.73410535e-01 -3.87826800e-01 -1.02635086e-01 -4.36747223e-01
4.66815799e-01 1.53488934e-01 -5.98700643e-01 1.55567801e+00
-1.76181272e-01 2.87918121e-01 7.88757652e-02 -1.21395159e+00
5.92230201e-01 5.80332994e-01 5.73061168e-01 -5.08694232e-01
5.68297684e-01 -7.03458935e-02 1.24819219e-01 -6.72868133e-01
-1.62844788e-02 -1.37666509e-01 -1.51131511e-01 7.84403160e-02
1.12601638e-01 2.86298364e-01 -5.15867710e-01 1.00651324e-01
1.27572119e+00 -4.51187700e-01 4.79412340e-02 -4.08474892e-01
5.31041384e-01 -2.34579384e-01 5.82286894e-01 5.92764139e-01
-4.52525198e-01 5.49391329e-01 5.57403505e-01 -7.20178962e-01
-8.06135654e-01 -1.10333455e+00 -3.48049104e-01 4.22796011e-01
1.54016688e-02 5.91502413e-02 -8.53714943e-01 -1.06001008e+00
8.05489719e-02 4.33797091e-01 -1.08191907e+00 -8.66484165e-01
-3.89011025e-01 -9.28142369e-01 1.05066991e+00 7.46521354e-01
6.56389594e-01 -9.82168257e-01 -6.20984077e-01 2.94600297e-02
-1.42114952e-01 -1.08490562e+00 -7.23780870e-01 7.18144998e-02
-7.90293038e-01 -1.20542979e+00 -6.95349574e-01 -7.09247887e-01
1.00093138e+00 2.55137999e-02 8.12952757e-01 1.14983663e-01
-6.34322643e-01 4.43271399e-02 -1.29088640e-01 -2.98284233e-01
-2.84912705e-01 8.89465958e-03 8.24553221e-02 2.45580912e-01
7.03340769e-02 -4.67064917e-01 -8.02957058e-01 1.54734701e-01
-1.00504267e+00 -1.80358291e-01 5.24961054e-01 1.15711236e+00
4.46523517e-01 1.58789709e-01 5.58372617e-01 -8.98871601e-01
5.38438916e-01 -6.69126809e-01 -4.09400731e-01 2.04824150e-01
-3.63949925e-01 6.91533387e-02 8.90827775e-01 -5.66188574e-01
-6.96883619e-01 8.50784928e-02 -1.55507028e-01 -9.76890862e-01
-6.26507252e-02 3.33715409e-01 -1.90858111e-01 -1.82258770e-01
6.01801753e-01 1.15634799e-01 1.43422738e-01 -4.32421744e-01
7.72106498e-02 3.04021180e-01 6.55332029e-01 -1.71716586e-01
8.94405663e-01 5.96224070e-01 1.69221357e-01 -4.51749027e-01
-1.04556549e+00 -9.48797613e-02 -3.29703391e-01 1.69592246e-01
1.05588925e+00 -8.97787571e-01 -9.23300743e-01 7.32897818e-01
-1.04053509e+00 1.63805231e-01 -1.48053601e-01 3.15941155e-01
-1.29176781e-01 4.02009070e-01 -9.81041551e-01 -3.02696079e-01
-8.49209607e-01 -1.41522598e+00 8.29978585e-01 -2.16615181e-02
1.59465261e-02 -9.54093158e-01 -1.29060730e-01 3.94655883e-01
5.19420207e-01 5.12945652e-01 9.60811377e-01 -7.87354052e-01
-2.65879780e-01 -2.93263197e-01 -2.98803836e-01 4.90764230e-01
3.13743830e-01 -3.59751523e-01 -1.07010067e+00 -6.54811382e-01
2.84218132e-01 -3.42490673e-01 9.12902772e-01 3.85495394e-01
1.61260271e+00 -5.99499643e-01 -2.25813717e-01 1.01606321e+00
1.16281927e+00 -2.83932593e-02 6.54878199e-01 2.05671459e-01
9.02464509e-01 5.05427361e-01 9.38770398e-02 2.50054121e-01
1.90081909e-01 3.09700727e-01 7.90651560e-01 -4.60968822e-01
1.62610918e-01 3.78839336e-02 1.40522987e-01 4.52858210e-01
1.81678027e-01 -1.64133534e-01 -7.85804987e-01 6.95098877e-01
-1.66776991e+00 -7.40283251e-01 2.83465564e-01 2.01562142e+00
7.25217402e-01 1.20447725e-01 -2.69667506e-01 9.95219797e-02
5.62450409e-01 -2.23436318e-02 -6.72243237e-01 -1.06323861e-01
6.02468699e-02 8.42785612e-02 5.45248628e-01 1.85928926e-01
-1.43612194e+00 5.88241875e-01 5.68796968e+00 7.97953248e-01
-1.37987113e+00 1.08011700e-01 9.64471042e-01 -1.79974571e-01
-5.64865731e-02 -7.31534243e-01 -2.85109639e-01 4.61921513e-01
7.40158737e-01 1.82492152e-01 1.45902202e-01 8.26764345e-01
-4.15436551e-03 5.98322332e-01 -9.11323786e-01 8.14101160e-01
2.73395628e-02 -1.35074377e+00 1.19160004e-01 1.11286357e-01
5.44478714e-01 5.04695289e-02 5.21434069e-01 -1.99867170e-02
2.46542931e-01 -1.19759619e+00 5.00467479e-01 3.53274167e-01
8.13872457e-01 -9.57362652e-01 1.06605995e+00 1.77778274e-01
-8.32798481e-01 -2.33674660e-01 -3.34193975e-01 4.28521037e-01
-9.76025611e-02 5.21486998e-01 -6.80248082e-01 6.72410905e-01
8.70395720e-01 6.33765042e-01 -7.76872516e-01 8.04248452e-01
-8.66408795e-02 4.45466101e-01 -1.79645017e-01 3.87235343e-01
5.10243773e-01 1.83661222e-01 4.76121962e-01 9.49700832e-01
2.55868971e-01 5.63184731e-03 -1.62185058e-02 8.37206781e-01
-2.20437095e-01 -2.27159396e-01 -7.08537281e-01 7.52392784e-02
1.42367378e-01 1.20938623e+00 -4.53354865e-01 -1.69722423e-01
-1.36303112e-01 1.01743567e+00 4.39774364e-01 1.36783600e-01
-8.85276139e-01 -3.15419257e-01 8.45001996e-01 -3.52582661e-03
3.07930499e-01 1.84867859e-01 -4.17325646e-01 -1.29789305e+00
-7.27332663e-03 -9.59015548e-01 6.16541207e-01 -3.94097835e-01
-1.50715375e+00 1.00836444e+00 -4.80933160e-01 -1.31701493e+00
3.68455760e-02 -6.92351818e-01 -6.49397910e-01 6.23772502e-01
-1.39853191e+00 -1.28039718e+00 -7.26348162e-02 1.22185683e+00
1.43500984e-01 -2.04861715e-01 9.39699888e-01 4.76377100e-01
-7.89742291e-01 1.09204853e+00 1.60259437e-02 6.00869417e-01
6.22747064e-01 -9.93056715e-01 3.19533825e-01 8.15616190e-01
-1.41705170e-01 7.01878130e-01 4.38034475e-01 -3.94353718e-01
-1.25596213e+00 -1.44179034e+00 6.77468002e-01 -3.21507573e-01
7.81781495e-01 -2.46783271e-01 -1.11496150e+00 5.89815795e-01
-5.69081232e-02 5.13891160e-01 5.56351244e-01 -2.80274361e-01
-6.49540722e-01 -1.24618456e-01 -1.55267262e+00 5.54334700e-01
7.08096802e-01 -6.47788107e-01 -5.44131517e-01 2.57307231e-01
8.83825541e-01 -3.21528554e-01 -1.00270319e+00 5.51141918e-01
3.28849822e-01 -6.15215957e-01 1.22665679e+00 -8.06460977e-01
4.88112688e-01 -1.28117412e-01 1.51875541e-01 -1.22074676e+00
-5.58153033e-01 -5.12806177e-01 -7.83727691e-02 8.62870157e-01
2.73180664e-01 -7.85999238e-01 5.36813438e-01 5.10104239e-01
-2.08613142e-01 -9.69019771e-01 -1.26514518e+00 -3.98800462e-01
1.90409660e-01 -1.15456238e-01 4.17602599e-01 1.19386160e+00
-5.98006658e-02 -6.50638575e-03 -5.73709726e-01 6.97331667e-01
8.17458630e-01 -2.03303769e-01 1.62532285e-01 -9.65347290e-01
-3.09416771e-01 -3.11855942e-01 -6.80599630e-01 -4.62605655e-01
5.41623542e-03 -7.75333226e-01 -1.90351188e-01 -9.64476764e-01
2.40250990e-01 -2.23853782e-01 -8.10444355e-01 7.57230878e-01
-4.36856270e-01 4.64971334e-01 1.60339609e-01 1.04467787e-01
-3.89507711e-01 6.27690911e-01 1.49131739e+00 -4.55003947e-01
1.56774595e-01 -6.23790100e-02 -8.56876075e-01 6.24022126e-01
7.10191250e-01 -6.19173288e-01 -4.36761886e-01 -6.12915039e-01
-1.71512738e-01 7.24978969e-02 7.34981596e-01 -8.41473877e-01
3.78628701e-01 1.25911683e-01 4.17212605e-01 -3.39383543e-01
1.58763155e-01 -1.05418742e+00 9.64661762e-02 1.01184881e+00
-3.51564795e-01 -1.12202369e-01 3.11240703e-01 5.42288303e-01
-2.10537449e-01 2.04813346e-01 1.03277051e+00 -6.04830161e-02
1.04539119e-01 7.52730310e-01 -3.46578002e-01 -1.94558278e-01
1.04589927e+00 2.25542530e-01 -3.66948187e-01 6.54873550e-02
-7.04348981e-01 3.71043503e-01 1.04155496e-01 4.98501360e-01
6.63249254e-01 -1.24927068e+00 -6.50462627e-01 4.19550508e-01
-2.20397543e-02 1.44624799e-01 6.71319306e-01 9.30469036e-01
-4.40973938e-01 3.56017470e-01 -3.82071555e-01 -3.89352888e-01
-1.08855140e+00 1.12974942e+00 6.93205714e-01 -5.43999076e-01
-9.77648675e-01 8.80998969e-01 6.61870122e-01 -2.15500861e-01
3.89006674e-01 -3.03816289e-01 -1.68313086e-01 -1.76389232e-01
6.91892743e-01 7.59509206e-02 1.03830151e-01 -3.86058033e-01
-5.53188205e-01 2.80016780e-01 -5.10722995e-01 4.33741719e-01
1.33947372e+00 1.48924962e-01 -2.02459805e-02 -2.37884372e-01
1.24274635e+00 -2.72067517e-01 -1.34281898e+00 -1.97394609e-01
-4.14666593e-01 -1.65876597e-01 3.26563239e-01 -6.81558132e-01
-1.71237469e+00 9.85541880e-01 7.21570432e-01 1.42596319e-01
1.19740474e+00 -1.32721603e-01 1.04899263e+00 2.83383906e-01
1.20155878e-01 -5.03684759e-01 3.98125239e-02 2.40419969e-01
8.54741693e-01 -1.14897335e+00 -2.63731569e-01 -2.28834510e-01
-5.78257263e-01 8.18452477e-01 6.06297493e-01 -4.32766646e-01
8.86097908e-01 2.80617088e-01 2.16109112e-01 -4.01722342e-01
-5.41331291e-01 3.23854446e-01 3.94985557e-01 5.82090676e-01
-2.66363937e-03 -4.61357385e-02 1.32644638e-01 9.61657703e-01
-5.90512604e-02 -1.67947143e-01 2.15629742e-01 5.78425825e-01
8.51511583e-02 -5.42775929e-01 -3.12415391e-01 2.71330267e-01
-1.03795147e+00 -2.37667173e-01 -1.37308165e-01 5.45631766e-01
1.16730839e-01 6.83037758e-01 -1.89021323e-02 -4.86956447e-01
3.12701315e-01 -3.27938795e-01 1.94960266e-01 -2.22218961e-01
-9.59740281e-01 -6.87017338e-03 -3.00975889e-01 -6.90424919e-01
-3.24974135e-02 -4.22129452e-01 -8.81861448e-01 -3.46906066e-01
-2.69013107e-01 -6.36749268e-02 2.14645460e-01 9.52472150e-01
5.92455566e-01 9.40475643e-01 6.82571471e-01 -5.21238267e-01
-1.00802350e+00 -8.62390816e-01 -4.56580192e-01 6.91947818e-01
6.94871306e-01 -5.27382314e-01 -3.29293698e-01 -1.43604130e-01] | [5.6604461669921875, 7.810205936431885] |
f34fd0b7-fdb6-4717-ab35-4eacd7dd1b74 | fake-it-till-you-make-it-face-analysis-in-the | 2109.15102 | null | https://arxiv.org/abs/2109.15102v2 | https://arxiv.org/pdf/2109.15102v2.pdf | Fake It Till You Make It: Face analysis in the wild using synthetic data alone | We demonstrate that it is possible to perform face-related computer vision in the wild using synthetic data alone. The community has long enjoyed the benefits of synthesizing training data with graphics, but the domain gap between real and synthetic data has remained a problem, especially for human faces. Researchers have tried to bridge this gap with data mixing, domain adaptation, and domain-adversarial training, but we show that it is possible to synthesize data with minimal domain gap, so that models trained on synthetic data generalize to real in-the-wild datasets. We describe how to combine a procedurally-generated parametric 3D face model with a comprehensive library of hand-crafted assets to render training images with unprecedented realism and diversity. We train machine learning systems for face-related tasks such as landmark localization and face parsing, showing that synthetic data can both match real data in accuracy as well as open up new approaches where manual labelling would be impossible. | ['Jamie Shotton', 'Thomas J. Cashman', 'Virginia Estellers', 'Matthew Johnson', 'Sebastian Dziadzio', 'Charlie Hewitt', 'Tadas Baltrušaitis', 'Erroll Wood'] | 2021-09-30 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Wood_Fake_It_Till_You_Make_It_Face_Analysis_in_the_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Wood_Fake_It_Till_You_Make_It_Face_Analysis_in_the_ICCV_2021_paper.pdf | iccv-2021-1 | ['face-alignment', 'face-model', 'face-parsing'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 5.01232207e-01 7.59111404e-01 2.72762567e-01 -6.33420765e-01
-1.01749027e+00 -8.43019009e-01 7.74520576e-01 -5.08823812e-01
-4.33004014e-02 5.24524033e-01 -1.09421477e-01 -3.15691650e-01
3.14544111e-01 -6.97949708e-01 -7.60435224e-01 -3.43985438e-01
1.61680549e-01 8.90637875e-01 6.47844970e-02 -2.58621693e-01
-4.02810313e-02 8.93795609e-01 -1.83368444e+00 3.42317075e-01
7.22723603e-01 6.92659378e-01 -1.45534784e-01 6.66176319e-01
-2.80322522e-01 2.50541538e-01 -6.02822006e-01 -7.71805882e-01
7.06490993e-01 -5.26695669e-01 -8.08455765e-01 4.19302434e-01
8.67405295e-01 -2.84618109e-01 4.56818230e-02 7.98651934e-01
4.66860592e-01 -3.70662212e-01 5.73249280e-01 -1.64186287e+00
-5.36552906e-01 -1.42696142e-01 -5.37026584e-01 -6.00203156e-01
4.74864513e-01 4.02555615e-01 2.77775526e-01 -7.46477544e-01
1.05558741e+00 1.54127514e+00 9.56424356e-01 1.17349863e+00
-1.60283816e+00 -7.73165762e-01 -4.19311732e-01 -6.76737010e-01
-1.16026223e+00 -8.87232304e-01 8.11964631e-01 -6.10090494e-01
4.64355797e-01 1.08263999e-01 6.09060407e-01 1.42990768e+00
-2.50912607e-01 3.13863367e-01 1.22581387e+00 -6.44235730e-01
2.08373740e-01 1.09203428e-01 -7.35420763e-01 9.67368305e-01
2.94208974e-02 4.04538035e-01 -3.17281097e-01 -1.87460735e-01
1.11499679e+00 -4.55144703e-01 -2.25660596e-02 -7.23404884e-01
-1.03589869e+00 7.70623386e-01 1.67479843e-01 -2.07881019e-01
6.29494190e-02 1.10233910e-01 2.49686509e-01 3.12416255e-01
5.40839732e-01 6.50137842e-01 -5.74679136e-01 2.08414495e-01
-1.01322854e+00 5.47256291e-01 8.54585767e-01 1.07328272e+00
7.59346664e-01 2.93961167e-01 2.67883718e-01 9.03849363e-01
1.89055666e-01 5.78423440e-01 1.68260589e-01 -1.65325785e+00
1.97554603e-01 5.19794285e-01 1.67784736e-01 -8.61462653e-01
-1.94157049e-01 1.76062897e-01 -4.35405344e-01 1.00645518e+00
9.70035672e-01 -3.12256843e-01 -1.10425699e+00 1.99466455e+00
6.57127202e-01 1.60290077e-01 2.66542584e-02 3.78109187e-01
7.55659401e-01 2.32391164e-01 2.31991261e-01 1.27040699e-01
1.13189435e+00 -5.37731528e-01 -1.54785559e-01 -4.45418119e-01
5.42052627e-01 -7.52889991e-01 1.21537483e+00 1.86125427e-01
-1.16229880e+00 -5.30502141e-01 -1.04243553e+00 -1.65519074e-01
-3.30907822e-01 -2.28380859e-01 8.04464936e-01 1.15873206e+00
-1.28966558e+00 6.75917864e-01 -8.18507493e-01 -5.09939134e-01
9.68822360e-01 4.43237633e-01 -9.28574026e-01 -6.44989535e-02
-6.93761826e-01 7.88619995e-01 1.75836369e-01 -2.47030199e-01
-9.76007700e-01 -1.04790568e+00 -1.04521453e+00 -4.85893488e-01
1.89160481e-01 -9.18042779e-01 1.37287307e+00 -1.23797083e+00
-1.53166652e+00 1.54505372e+00 1.05877109e-02 -7.94801265e-02
7.72835314e-01 3.45019221e-01 -4.45995003e-01 8.74853134e-02
1.75996274e-01 1.30677795e+00 1.05927074e+00 -1.63370037e+00
-3.27810496e-01 -5.68710864e-01 3.18628550e-02 -1.52976245e-01
8.53095800e-02 5.40708080e-02 -2.76185960e-01 -4.98459101e-01
8.00900087e-02 -9.10092592e-01 -1.84245318e-01 7.80530453e-01
-1.25302359e-01 2.14348257e-01 9.52089965e-01 -6.81172848e-01
6.92261383e-02 -2.04336762e+00 -2.72791505e-01 3.02812923e-03
5.85985370e-02 4.49571013e-01 -4.06242639e-01 1.75904199e-01
-1.59025550e-01 2.61992216e-01 -5.76945305e-01 -5.21919727e-01
-1.36326075e-01 3.18629414e-01 -4.60982889e-01 3.27637911e-01
5.95924795e-01 1.03652048e+00 -8.81962419e-01 -6.05785131e-01
1.44041330e-01 6.23864651e-01 -8.26936245e-01 1.84138790e-01
-6.65754080e-01 1.03766978e+00 -2.28021339e-01 6.09545887e-01
9.71284091e-01 2.94122901e-02 2.53662497e-01 2.21463650e-01
2.92209506e-01 -3.83569859e-02 -1.04765320e+00 2.02407002e+00
-5.92810273e-01 4.76264805e-01 4.56618339e-01 -5.61510324e-01
8.32564175e-01 2.66955614e-01 1.91922322e-01 -6.51498973e-01
-1.29493937e-01 1.91822976e-01 -3.36331129e-01 -3.96939099e-01
2.23795399e-01 -5.74854016e-01 3.32995463e-04 5.07522881e-01
9.15125012e-02 -1.05893588e+00 -2.97170669e-01 -6.99241161e-02
8.97949874e-01 8.11216831e-01 -3.24712135e-02 -1.70796439e-01
1.47058219e-01 2.72804409e-01 4.52674955e-01 2.93692082e-01
-5.96565306e-02 1.08903348e+00 6.29303217e-01 -5.20021915e-01
-1.50590789e+00 -1.17311203e+00 -2.04249069e-01 9.26355779e-01
-2.27863476e-01 8.11774805e-02 -1.19192743e+00 -8.42653215e-01
1.38456762e-01 7.14370728e-01 -7.81337857e-01 -4.91770208e-02
-5.90657592e-01 -2.78733075e-01 9.30762351e-01 3.89753759e-01
4.67820048e-01 -1.14291739e+00 -5.54019034e-01 7.78300315e-03
3.77217412e-01 -1.18948257e+00 -1.91494897e-01 -8.82753879e-02
-7.40377963e-01 -1.11295331e+00 -5.56558132e-01 -8.83798361e-01
8.65509927e-01 1.91508811e-02 1.57060933e+00 9.13313329e-02
-5.19175828e-01 4.59626764e-01 -2.84094363e-03 -3.72978300e-01
-9.93953347e-01 -3.57565254e-01 -7.96310827e-02 -1.79869741e-01
-3.95996012e-02 -9.09634650e-01 -4.96336430e-01 4.35797513e-01
-9.49427128e-01 1.46552101e-01 1.28000662e-01 7.66095042e-01
4.15293336e-01 -4.25015271e-01 6.12255394e-01 -1.26790047e+00
3.75805497e-01 -4.97694053e-02 -8.56121361e-01 2.43658245e-01
-1.26140028e-01 1.24257438e-01 6.13736510e-01 -2.91423917e-01
-1.33477879e+00 3.84229273e-01 -2.95284867e-01 -2.97926962e-01
-5.03328085e-01 -3.58214527e-01 -6.85192525e-01 -4.97201622e-01
1.13301873e+00 -2.05112845e-01 4.04084206e-01 -3.14756066e-01
8.52873385e-01 4.45581585e-01 7.80577064e-01 -9.58498001e-01
9.14410532e-01 8.57128084e-01 2.31837690e-01 -8.46510708e-01
-4.93178517e-01 4.59623933e-01 -8.32116365e-01 1.04466341e-01
7.80038595e-01 -8.07068348e-01 -5.64202666e-01 4.13890809e-01
-1.27365470e+00 -6.10999107e-01 -5.84751427e-01 -1.64519727e-01
-8.88035476e-01 1.73039272e-01 -1.93379879e-01 -6.13057435e-01
2.43251607e-01 -1.08373094e+00 1.47463834e+00 1.15848653e-01
-3.48754644e-01 -9.07948732e-01 1.23083945e-02 2.98845559e-01
2.43717596e-01 8.68409574e-01 8.72310936e-01 -4.22805935e-01
-5.18554091e-01 -8.63784775e-02 -2.46280327e-01 2.49202371e-01
8.95741805e-02 2.15676084e-01 -1.39317882e+00 -9.00996476e-02
-5.61840758e-02 -6.89761043e-01 4.55357045e-01 -7.19433725e-02
1.19180632e+00 -1.72040179e-01 -4.42317069e-01 8.40041041e-01
1.15873432e+00 2.06805944e-01 7.94685125e-01 -1.64305076e-01
6.63182795e-01 1.06537259e+00 2.33419418e-01 -1.86360404e-01
1.47868440e-01 6.83540881e-01 2.96227872e-01 -1.84960350e-01
-4.31285411e-01 -7.13823080e-01 -1.33646265e-01 -1.02659166e-01
2.62840658e-01 3.93218808e-02 -1.05099678e+00 4.40532327e-01
-1.21231961e+00 -7.89130270e-01 1.30108207e-01 2.15045810e+00
8.72035325e-01 5.89620955e-02 8.26944560e-02 8.94025937e-02
5.75334787e-01 -1.05851457e-01 -4.82860178e-01 -5.10645151e-01
7.79407332e-03 6.25923395e-01 3.04009020e-01 5.75650752e-01
-1.04773295e+00 1.16571641e+00 6.95264769e+00 6.47665024e-01
-1.10441327e+00 1.17035754e-01 7.36047149e-01 2.51965940e-01
-5.50686240e-01 7.91200772e-02 -4.75195646e-01 1.96631700e-01
9.20173883e-01 1.71856567e-01 3.38741660e-01 9.46102738e-01
-1.54672638e-01 6.11028969e-02 -1.38880455e+00 9.81840670e-01
4.55787145e-02 -1.36269724e+00 7.03277364e-02 1.69854850e-01
7.74243295e-01 -3.34770173e-01 2.70266384e-01 8.95166397e-02
7.53529429e-01 -1.53658259e+00 5.45602620e-01 2.51008660e-01
1.40118539e+00 -6.29133821e-01 -8.13124888e-03 4.16125178e-01
-6.85453534e-01 3.96422774e-01 -2.41608918e-01 1.75618857e-01
-1.17938621e-02 2.61752725e-01 -1.18158889e+00 2.11245373e-01
2.71166801e-01 2.11277813e-01 -6.67144060e-01 6.52072370e-01
-7.15090930e-02 6.25263229e-02 -4.04967129e-01 6.58219039e-01
-2.56243020e-01 -1.97650522e-01 1.42155424e-01 7.20840156e-01
3.68881553e-01 -7.16025308e-02 -2.41780207e-01 9.74987447e-01
-2.38705486e-01 -1.82632491e-01 -1.26946533e+00 -1.39355689e-01
4.27379698e-01 1.09268475e+00 -8.27695489e-01 2.84611154e-02
-2.18203381e-01 1.15826845e+00 3.63081515e-01 1.33460164e-01
-5.39929509e-01 -2.24089727e-01 7.47265458e-01 3.82820040e-01
1.12185933e-01 -2.86843479e-01 -5.29504061e-01 -9.34593678e-01
-2.15558205e-02 -9.81438100e-01 -7.27652386e-02 -9.56746340e-01
-1.15380108e+00 8.78527045e-01 -2.11460903e-01 -1.15434599e+00
-6.79214299e-01 -6.99354649e-01 -3.26552987e-01 8.85828674e-01
-1.03114104e+00 -1.61046708e+00 -2.71527976e-01 4.24218416e-01
2.65468061e-01 -2.66788810e-01 1.21845150e+00 6.45955503e-02
1.95894837e-02 8.28239858e-01 -1.53357774e-01 1.86418235e-01
6.77571237e-01 -1.02699435e+00 1.12127399e+00 4.67541188e-01
3.92318070e-01 4.77458507e-01 7.52045989e-01 -4.99500513e-01
-1.26143265e+00 -1.02814293e+00 3.81410986e-01 -1.11272073e+00
9.40766782e-02 -8.00472558e-01 -8.82727265e-01 7.70379484e-01
-4.86838594e-02 4.89044279e-01 4.32997197e-01 -2.13980824e-01
-9.31385934e-01 1.42315328e-02 -1.91271079e+00 7.96521485e-01
1.43115425e+00 -8.17238271e-01 -3.65794063e-01 3.68410707e-01
7.57958651e-01 -4.67378318e-01 -5.58629274e-01 5.13823569e-01
6.83533490e-01 -1.11018729e+00 1.26023996e+00 -8.34345460e-01
4.20610756e-01 -2.91191012e-01 -1.19388498e-01 -1.39421165e+00
3.59295100e-01 -8.22204530e-01 3.37706506e-01 1.42956233e+00
3.89707297e-01 -5.17869651e-01 1.21731186e+00 9.71643865e-01
1.05572954e-01 -2.12700829e-01 -9.50190663e-01 -6.39024913e-01
4.96050328e-01 -4.74066883e-01 9.04480159e-01 9.41594958e-01
-4.83040422e-01 2.12705597e-01 -2.98499539e-02 6.27026483e-02
6.58607543e-01 1.12968059e-02 1.25464928e+00 -1.44530737e+00
-1.54076964e-01 -2.56445557e-01 -4.92674828e-01 -7.87727296e-01
5.53963423e-01 -9.96751189e-01 -1.39932707e-01 -1.25139451e+00
-2.52524853e-01 -7.55756438e-01 7.45005488e-01 3.58601660e-01
2.43116677e-01 6.81225359e-01 -6.90160180e-03 -5.00814803e-02
-4.79195677e-02 2.61242986e-01 1.66743314e+00 1.75130263e-01
2.41673011e-02 -1.90629736e-01 -9.72009659e-01 9.42940295e-01
5.01484036e-01 -3.93546522e-01 -5.85085750e-01 -6.14655316e-01
-1.01151258e-01 2.30450317e-01 5.46277642e-01 -9.90571499e-01
-3.75910737e-02 -1.25662521e-01 7.40953147e-01 4.29342054e-02
6.32520914e-01 -8.37967873e-01 3.93288493e-01 2.57164910e-02
-6.76676407e-02 -3.93124461e-01 5.42527378e-01 4.71423835e-01
2.19670743e-01 5.22334408e-03 1.07736909e+00 -3.00178796e-01
-6.09506488e-01 3.31733346e-01 4.08246487e-01 4.32151198e-01
1.13531125e+00 -4.23129439e-01 -2.83891201e-01 -4.12704080e-01
-7.50867486e-01 -1.94073722e-01 1.13104975e+00 3.15848410e-01
4.32706982e-01 -1.22184098e+00 -7.23537624e-01 7.74506390e-01
1.17657349e-01 2.79674530e-01 8.82816091e-02 -4.40921038e-02
-9.16891932e-01 1.62871741e-02 -4.76389021e-01 -7.31452405e-01
-1.05937529e+00 7.19918072e-01 5.63959181e-01 1.85914308e-01
-4.23267126e-01 1.00948858e+00 3.78033519e-01 -1.02793097e+00
-3.68882269e-02 7.81037062e-02 4.38000888e-01 -1.93873912e-01
4.68970478e-01 -1.22647315e-01 1.41129136e-01 -6.17978454e-01
-2.64586449e-01 7.21722007e-01 3.39800149e-01 -4.26871479e-01
1.27145135e+00 3.06983739e-01 2.86494400e-02 -1.42849654e-01
1.09749508e+00 2.35917777e-01 -1.82435083e+00 1.11660056e-01
-9.48457196e-02 -7.62183666e-01 -4.95398343e-01 -8.59320045e-01
-9.26871836e-01 1.08709502e+00 5.19201994e-01 1.31131321e-01
9.48863566e-01 5.53446226e-02 5.12656868e-01 3.24089341e-02
7.83024013e-01 -6.49989307e-01 2.82303207e-02 1.78720340e-01
9.85116482e-01 -1.33678079e+00 -2.09861428e-01 -7.46288598e-01
-6.06020093e-01 8.06576610e-01 5.75981081e-01 -2.08947569e-01
6.79433823e-01 5.06837368e-01 1.84206828e-01 -1.28371287e-02
-3.97235096e-01 -6.79387152e-02 -8.10189471e-02 1.41202462e+00
2.18338028e-01 -4.50784415e-02 2.88759202e-01 1.63445339e-01
-5.34168422e-01 1.43182889e-01 2.49632925e-01 8.73964489e-01
-4.90984274e-03 -1.73345339e+00 -4.82458502e-01 1.69072926e-01
-3.86494756e-01 2.25607842e-01 -4.85322744e-01 1.14533198e+00
2.70622939e-01 7.90628672e-01 1.69007152e-01 -1.57024086e-01
2.50958443e-01 4.82284278e-01 9.69904363e-01 -7.92086959e-01
-2.36800179e-01 -2.97842115e-01 1.29464418e-01 -4.98912781e-01
-2.49917567e-01 -4.65706199e-01 -1.06832922e+00 -3.22861642e-01
2.24637449e-01 -1.09496772e-01 8.63850176e-01 6.93073809e-01
5.56088448e-01 1.44007683e-01 4.97628391e-01 -1.03272927e+00
-3.23347300e-01 -6.21636868e-01 -4.40084040e-01 6.06703520e-01
3.33379537e-01 -6.48637354e-01 -2.69330233e-01 3.07793617e-01] | [12.413986206054688, -0.3238179683685303] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.