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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
e2aca4bd-8f05-4269-a643-7307b55dcf04 | unsupervised-sentence-textual-similarity-with | 2210.02284 | null | https://arxiv.org/abs/2210.02284v1 | https://arxiv.org/pdf/2210.02284v1.pdf | Unsupervised Sentence Textual Similarity with Compositional Phrase Semantics | Measuring Sentence Textual Similarity (STS) is a classic task that can be applied to many downstream NLP applications such as text generation and retrieval. In this paper, we focus on unsupervised STS that works on various domains but only requires minimal data and computational resources. Theoretically, we propose a light-weighted Expectation-Correction (EC) formulation for STS computation. EC formulation unifies unsupervised STS approaches including the cosine similarity of Additively Composed (AC) sentence embeddings, Optimal Transport (OT), and Tree Kernels (TK). Moreover, we propose the Recursive Optimal Transport Similarity (ROTS) algorithm to capture the compositional phrase semantics by composing multiple recursive EC formulations. ROTS finishes in linear time and is faster than its predecessors. ROTS is empirically more effective and scalable than previous approaches. Extensive experiments on 29 STS tasks under various settings show the clear advantage of ROTS over existing approaches. Detailed ablation studies demonstrate the effectiveness of our approaches. | ['Yong Zhang', 'Jiaheng Dou', 'ZiHao Wang'] | 2022-10-05 | null | https://aclanthology.org/2022.coling-1.441 | https://aclanthology.org/2022.coling-1.441.pdf | coling-2022-10 | ['sentence-embeddings', 'sentence-embeddings'] | ['methodology', 'natural-language-processing'] | [ 6.35344088e-01 -2.67126888e-01 -8.73114392e-02 -3.66912693e-01
-1.01671422e+00 -6.64631367e-01 6.88981712e-01 5.86089313e-01
-6.27371788e-01 4.70669627e-01 7.29164600e-01 -5.22495568e-01
-2.84490347e-01 -4.26561803e-01 -4.17955637e-01 -5.39753914e-01
2.13874027e-01 3.86805207e-01 5.13945758e-01 -3.27546269e-01
7.20239222e-01 3.07938028e-02 -1.14345276e+00 2.57808238e-01
1.26734567e+00 8.33251655e-01 4.15558726e-01 6.72356844e-01
-4.54091907e-01 6.00247145e-01 -3.13621640e-01 -8.04235458e-01
6.15555048e-02 -3.85477155e-01 -9.63282585e-01 -3.08607101e-01
2.27969050e-01 1.19335353e-01 -1.39779583e-01 9.76974487e-01
6.12542808e-01 5.06643951e-01 8.71168137e-01 -1.24651611e+00
-7.59980202e-01 8.93921196e-01 -5.26826620e-01 3.64362866e-01
4.20134932e-01 -2.78812975e-01 1.55686414e+00 -1.20617664e+00
4.44633722e-01 1.13560855e+00 9.03353989e-01 9.71135721e-02
-1.19903195e+00 -3.47029448e-01 3.04959808e-02 2.10818723e-01
-1.25008976e+00 -1.97855636e-01 4.15928155e-01 -1.60556570e-01
1.21772325e+00 3.98421437e-01 2.16087461e-01 9.36848402e-01
7.23149255e-02 1.08975625e+00 9.61288214e-01 -6.69512451e-01
1.82810023e-01 -2.60661729e-03 2.86466688e-01 4.65478212e-01
1.91856503e-01 -4.52939898e-01 -7.61777163e-01 -4.55062389e-01
2.44011983e-01 -1.30379841e-01 -1.93915561e-01 -1.13326065e-01
-1.36843288e+00 9.82571363e-01 -8.21823850e-02 5.51773310e-01
-1.84066340e-01 9.29721519e-02 6.31088018e-01 3.99315596e-01
7.44127810e-01 4.16799426e-01 -5.81828415e-01 -2.96217322e-01
-1.00007296e+00 2.13759497e-01 7.41471589e-01 1.26120317e+00
3.84177685e-01 -2.55174309e-01 -5.86379349e-01 1.17707586e+00
5.49569018e-02 4.68453020e-01 7.47851849e-01 -8.02649558e-01
7.87754714e-01 5.07811069e-01 -7.60698542e-02 -8.87472630e-01
-5.02813570e-02 -1.87805489e-01 -6.70560300e-01 -8.59794855e-01
1.56869486e-01 6.06648587e-02 -4.73623157e-01 1.65157378e+00
4.58142877e-01 4.69256282e-01 1.03557967e-01 6.47735357e-01
5.60647249e-01 7.47748613e-01 8.00468922e-02 -2.83662230e-01
1.39544237e+00 -1.19026697e+00 -7.51093388e-01 -9.24801454e-02
1.14491093e+00 -1.10077620e+00 1.24391651e+00 4.44639064e-02
-1.04153121e+00 -1.04967773e-01 -7.76903152e-01 -3.60380799e-01
-2.59935111e-01 1.52567193e-01 6.27330482e-01 6.49227440e-01
-1.19982100e+00 9.17890072e-01 -6.50708377e-01 -7.55935252e-01
7.55653754e-02 2.81035244e-01 -1.43550381e-01 1.13653280e-01
-1.30831587e+00 9.14638579e-01 3.72135967e-01 -1.61431227e-02
-1.90991536e-01 -8.05330336e-01 -9.37594473e-01 1.15628541e-01
3.72363687e-01 -8.57857287e-01 1.58151495e+00 -5.14373541e-01
-1.75893486e+00 5.29785633e-01 -7.04326808e-01 -5.44252694e-01
1.56809390e-01 -3.64188313e-01 -2.35795513e-01 2.39517257e-01
2.30541155e-01 2.41218328e-01 6.82653189e-01 -5.58956027e-01
-6.02915227e-01 2.69061495e-02 -2.78978020e-01 4.09514099e-01
-1.02987897e+00 4.27460790e-01 -3.25682372e-01 -9.59433258e-01
1.61274478e-01 -8.43875825e-01 -2.35446110e-01 -1.41476497e-01
-2.98787624e-01 -7.59398758e-01 4.67327386e-01 -4.61961776e-01
1.73926079e+00 -1.98178327e+00 1.71934783e-01 -8.11029375e-02
5.72346859e-02 2.88165480e-01 -2.39455625e-01 1.05928433e+00
1.03876919e-01 1.85543105e-01 -6.63212717e-01 -4.86023903e-01
4.48996633e-01 1.02655865e-01 -5.76073110e-01 9.27570835e-02
1.43361256e-01 1.01739180e+00 -1.08036101e+00 -9.18905675e-01
-4.64428663e-02 -3.39815021e-03 -4.21205372e-01 1.16387976e-03
6.81710467e-02 -2.13953733e-01 -5.72941601e-01 2.59394854e-01
5.59213817e-01 -2.20565096e-01 2.77277559e-01 -2.05535173e-01
-1.75877213e-01 8.63351524e-01 -9.25928652e-01 1.82847452e+00
-5.17937958e-01 5.40938318e-01 -5.48114061e-01 -1.11763370e+00
8.32158029e-01 2.34205723e-01 3.12140375e-01 -4.47384506e-01
1.15816258e-01 3.33298922e-01 -3.00869912e-01 -5.32294452e-01
9.46926594e-01 -1.55381456e-01 -2.36378267e-01 9.45259094e-01
4.52017598e-02 -1.13206066e-01 4.29880351e-01 6.91800117e-01
1.34055459e+00 -2.55728210e-03 5.54441154e-01 -3.76914054e-01
6.53464854e-01 -1.01933688e-01 4.95564073e-01 7.04917729e-01
-2.34094813e-01 4.47927475e-01 4.03421313e-01 1.28882065e-01
-1.11303580e+00 -1.27301633e+00 9.17119384e-02 1.27526534e+00
1.77883938e-01 -8.48000705e-01 -6.42169714e-01 -7.90993452e-01
-1.59302920e-01 8.83397281e-01 -2.78009683e-01 -1.81305811e-01
-5.91875851e-01 -6.68094695e-01 7.36778438e-01 7.77592480e-01
1.99236199e-01 -8.37038100e-01 -1.55571565e-01 2.67599195e-01
-6.32025540e-01 -1.30030692e+00 -1.13738191e+00 -1.78314686e-01
-1.01445520e+00 -5.43012083e-01 -5.09593070e-01 -9.51881886e-01
4.21876162e-01 7.18558133e-01 8.47343743e-01 -1.39321312e-01
-9.38823745e-02 1.22272350e-01 -7.42149532e-01 -2.82622814e-01
-1.17007397e-01 1.56022981e-01 2.21606374e-01 -6.32219017e-05
4.69252735e-01 -5.28030396e-01 -4.48114365e-01 6.90583466e-03
-9.37705815e-01 -9.20710936e-02 6.38779402e-01 9.05172169e-01
3.95953804e-01 -1.09409831e-01 5.37112474e-01 -9.37371194e-01
1.25967574e+00 -4.36776638e-01 -2.43503615e-01 5.28271019e-01
-8.44794273e-01 3.71894211e-01 6.63475454e-01 -4.35065120e-01
-1.26164639e+00 -2.06679046e-01 1.02028459e-01 -6.51676133e-02
2.44210601e-01 6.24067545e-01 2.80419022e-01 2.50710607e-01
3.32465053e-01 4.89005953e-01 -3.68025780e-01 -4.35664177e-01
4.17711943e-01 8.92682850e-01 2.14805499e-01 -7.34017551e-01
8.15835953e-01 2.74175435e-01 -2.35115051e-01 -7.91025460e-01
-9.45507705e-01 -8.53526890e-01 -6.18175507e-01 2.05660760e-01
6.17193162e-01 -5.89541733e-01 -5.68773329e-01 1.39431536e-01
-1.27603340e+00 -2.05566711e-03 -7.06155524e-02 7.17242420e-01
-3.89698893e-01 9.92082298e-01 -7.69208491e-01 -9.09877598e-01
-8.37797344e-01 -6.19450986e-01 1.27477205e+00 -1.31719604e-01
-4.91524220e-01 -1.20257902e+00 3.46146703e-01 3.82204920e-01
2.94672996e-01 -2.26029441e-01 9.21135902e-01 -8.56613815e-01
-1.58139661e-01 6.02567829e-02 -2.81978995e-01 2.97364682e-01
9.93832350e-02 1.65879410e-02 -6.98856831e-01 -4.22044508e-02
-6.35690689e-02 -1.74789205e-01 9.80481923e-01 2.14507148e-01
9.55630124e-01 -3.42708111e-01 -2.72054046e-01 2.80432552e-01
1.30846024e+00 -1.69387519e-01 3.43974501e-01 1.36136398e-01
4.84055609e-01 6.11839771e-01 9.68994915e-01 4.93079066e-01
5.24303436e-01 5.89395761e-01 -3.13496679e-01 1.80848181e-01
1.78873599e-01 -3.56712967e-01 7.97999144e-01 1.75694847e+00
1.49809286e-01 -4.39544350e-01 -5.91968477e-01 6.72789335e-01
-2.19088912e+00 -1.11815643e+00 -5.73470652e-01 2.13067174e+00
9.99700546e-01 6.69441894e-02 -9.79727283e-02 2.21521541e-01
8.89826775e-01 1.19198263e-01 -1.83228418e-01 -9.61063206e-01
-1.55605018e-01 6.05223656e-01 4.47211832e-01 3.49377066e-01
-9.93652999e-01 1.19542015e+00 6.42353821e+00 1.21745157e+00
-5.03800392e-01 3.19537848e-01 2.23255977e-02 5.25884936e-03
-5.39517403e-01 2.66329765e-01 -6.55373812e-01 4.07863885e-01
9.83068228e-01 -6.14333630e-01 2.01898411e-01 6.31755412e-01
2.57508665e-01 -5.09614907e-02 -1.01623964e+00 6.33894145e-01
2.55119771e-01 -1.19492543e+00 1.02220006e-01 -3.92362446e-01
8.43082964e-01 -3.40598710e-02 -5.78538924e-02 1.79336354e-01
4.67710257e-01 -5.04453659e-01 5.91993809e-01 1.66630670e-01
5.40535152e-01 -6.20506585e-01 8.87573183e-01 3.28486085e-01
-1.69429815e+00 6.13654740e-02 -4.22687203e-01 -6.90550590e-03
3.45617533e-01 5.87482274e-01 -9.92335379e-01 9.03692722e-01
6.13848031e-01 7.64546871e-01 -3.61044317e-01 8.41696262e-01
-3.87661189e-01 7.13949740e-01 -4.07033682e-01 -6.40702784e-01
5.04385889e-01 -3.45575184e-01 5.84711015e-01 1.60168684e+00
4.42404360e-01 9.41410288e-02 -2.63387095e-02 5.24150431e-01
-1.22507952e-01 5.75408161e-01 -4.46579605e-01 -2.97487587e-01
9.75853443e-01 1.25594437e+00 -6.55304372e-01 -5.33041120e-01
-2.27895141e-01 1.36142051e+00 5.07376492e-01 1.49021819e-01
-8.18816662e-01 -8.36944580e-01 4.77698654e-01 -3.22940111e-01
5.95749438e-01 -2.38362774e-01 -2.26971507e-01 -1.13650894e+00
2.33533263e-01 -5.46101570e-01 5.40243208e-01 -6.95457160e-01
-1.63405323e+00 3.95057231e-01 4.66873348e-02 -1.37847388e+00
-6.94572106e-02 -4.26096439e-01 -7.46451139e-01 6.90997779e-01
-1.47532833e+00 -9.37807262e-01 9.54317078e-02 5.24241507e-01
8.30715835e-01 1.81927800e-01 7.86972761e-01 1.28813550e-01
-6.08030260e-01 7.79395103e-01 5.15411139e-01 -1.76699117e-01
9.17159855e-01 -1.37218797e+00 6.49869859e-01 1.00026786e+00
3.95747900e-01 8.62911761e-01 5.35924673e-01 -5.80263555e-01
-1.27194822e+00 -1.12055218e+00 1.78617728e+00 -2.47021124e-01
1.20362747e+00 -3.65827948e-01 -7.50470340e-01 4.75930810e-01
4.62327749e-01 -5.20470560e-01 9.15727377e-01 1.63281837e-03
-4.78398889e-01 -5.85661493e-02 -7.18591869e-01 9.13031518e-01
1.21348238e+00 -6.72552586e-01 -9.75636244e-01 5.76901495e-01
1.03727448e+00 -2.68422570e-02 -9.02843177e-01 2.68073946e-01
4.26831573e-01 -7.23507822e-01 8.17274392e-01 -5.88154018e-01
5.58296025e-01 -2.10662261e-01 -1.29245326e-01 -1.24038363e+00
-3.77555817e-01 -9.25375223e-01 6.19456619e-02 1.45984566e+00
5.09611785e-01 -8.16677332e-01 2.87773162e-01 2.69759893e-01
-3.51824790e-01 -8.51611853e-01 -7.89815605e-01 -1.10113394e+00
6.39151856e-02 -6.09570026e-01 5.90264380e-01 1.07427466e+00
4.82787848e-01 5.56765676e-01 -1.12414718e-01 8.69118795e-02
5.46147168e-01 2.81231701e-01 4.12165403e-01 -9.26134586e-01
-2.58267313e-01 -4.36168045e-01 -1.28067985e-01 -1.24536359e+00
2.54873395e-01 -1.32890141e+00 1.44150987e-01 -1.64271796e+00
3.84067655e-01 -2.88710326e-01 -3.90888095e-01 2.44449094e-01
-5.01819968e-01 1.50640411e-02 -1.43878721e-02 4.71188687e-02
-7.71065056e-01 8.11943054e-01 8.72349024e-01 2.97064573e-04
3.15508805e-02 -3.50907892e-01 -6.57355964e-01 6.35343134e-01
9.49808061e-01 -7.42444336e-01 -6.65652335e-01 -6.42604589e-01
1.77307799e-01 -3.26338798e-01 -1.78827476e-02 -4.75702912e-01
6.38184071e-01 -1.27742469e-01 -3.31655711e-01 -7.61965573e-01
1.67923972e-01 -3.34759325e-01 -2.39801124e-01 3.33681971e-01
-6.79264843e-01 3.32485288e-01 -2.49727309e-01 6.57288313e-01
-1.86034009e-01 -6.92777395e-01 2.88309008e-01 1.14659160e-01
-3.08225662e-01 2.24621668e-01 -4.50294256e-01 4.12467152e-01
7.80228138e-01 -2.23691583e-01 -1.50641143e-01 -1.53282121e-01
-1.08250678e-01 3.12441736e-01 4.74419212e-03 4.99347538e-01
9.11807716e-01 -1.44940937e+00 -7.85055161e-01 -2.74754435e-01
2.66045928e-01 -3.56633991e-01 6.76901564e-02 1.35839880e+00
-3.57291013e-01 5.66006780e-01 6.10096395e-01 -3.80685329e-01
-1.45879555e+00 3.74624878e-01 -4.27147716e-01 -5.69398105e-01
-6.45820439e-01 9.31563616e-01 1.43979490e-01 -5.28112352e-01
-9.05154366e-03 -3.56591463e-01 -5.15900850e-02 -1.63526148e-01
4.57712740e-01 5.15683711e-01 3.02653890e-02 -4.17915195e-01
-3.69201809e-01 6.26213491e-01 -2.52952099e-01 -3.40850264e-01
1.23152196e+00 -3.13152969e-01 -3.23674440e-01 5.66184044e-01
1.36835754e+00 2.74996217e-02 -3.76339227e-01 -6.03977501e-01
6.78448498e-01 -2.94328451e-01 -1.81366727e-01 -2.14171425e-01
-4.14203405e-01 8.56932402e-01 -5.91990054e-02 9.31788683e-02
1.03287256e+00 -1.96859345e-01 1.45111048e+00 6.15382195e-01
8.98791552e-02 -1.45226920e+00 1.41900346e-01 6.67222679e-01
5.65536320e-01 -8.34515095e-01 1.43140496e-03 -6.79462910e-01
-9.46853876e-01 1.06701028e+00 2.34218597e-01 -2.79483106e-02
4.50608164e-01 2.24121138e-01 -3.79305005e-01 1.02905273e-01
-1.19917572e+00 -2.99884647e-01 2.98472553e-01 1.39980391e-01
5.31282961e-01 -1.28307819e-01 -1.07103693e+00 5.58038294e-01
-2.63498753e-01 -9.81810316e-02 2.77760535e-01 1.02819812e+00
-4.87777978e-01 -1.38054812e+00 2.14904193e-02 5.18202424e-01
-4.28284168e-01 -8.17019463e-01 -6.16973281e-01 3.64094466e-01
-1.84264407e-01 1.12723827e+00 -1.13177650e-01 -4.84260827e-01
2.03436658e-01 1.55190066e-01 4.38981801e-01 -5.79901516e-01
-6.88046992e-01 4.46746908e-02 2.87620753e-01 -3.16015333e-01
-4.70596403e-01 -8.86199176e-01 -1.32141733e+00 -3.79553467e-01
-6.98534667e-01 3.68377239e-01 5.66720188e-01 1.18137825e+00
4.16197687e-01 1.85804710e-01 9.13275063e-01 -1.54864028e-01
-8.40538800e-01 -1.18045270e+00 -3.64340782e-01 3.80394369e-01
-2.07241774e-01 -4.72026587e-01 -5.04430711e-01 1.59475103e-01] | [11.047858238220215, 8.874863624572754] |
54f30c7d-6f62-46e3-a231-eeb259ed08b2 | 1st-place-solution-for-ava-kinetics-crossover | 2006.09116 | null | https://arxiv.org/abs/2006.09116v1 | https://arxiv.org/pdf/2006.09116v1.pdf | 1st place solution for AVA-Kinetics Crossover in AcitivityNet Challenge 2020 | This technical report introduces our winning solution to the spatio-temporal action localization track, AVA-Kinetics Crossover, in ActivityNet Challenge 2020. Our entry is mainly based on Actor-Context-Actor Relation Network. We describe technical details for the new AVA-Kinetics dataset, together with some experimental results. Without any bells and whistles, we achieved 39.62 mAP on the test set of AVA-Kinetics, which outperforms other entries by a large margin. Code will be available at: https://github.com/Siyu-C/ACAR-Net. | ['Manyuan Zhang', 'Ziyi Lin', 'Yu Liu', 'Siyu Chen', 'Junting Pan', 'Hao Shao', 'Hongsheng Li', 'Jing Shao', 'Guanglu Song'] | 2020-06-16 | null | null | null | null | ['spatio-temporal-action-localization'] | ['computer-vision'] | [-3.50812495e-01 -1.08495422e-01 -6.21139050e-01 3.05445455e-02
-7.07266510e-01 -7.57026136e-01 7.70830393e-01 -1.38632074e-01
-4.83388722e-01 9.21205938e-01 7.26270974e-01 -1.64434895e-01
-2.62352049e-01 -1.86002880e-01 -4.91861224e-01 -6.05029881e-01
-7.19837725e-01 5.26423901e-02 4.57096785e-01 -2.32834563e-01
-3.68568078e-02 2.92192400e-01 -9.61083472e-01 5.25748074e-01
3.88606973e-02 1.00255668e+00 -1.07914962e-01 8.08920979e-01
3.63719910e-01 1.45343184e+00 -6.90455616e-01 5.30510210e-02
1.99082300e-01 -4.52793002e-01 -8.82695615e-01 -5.37877262e-01
8.07027668e-02 -1.00761935e-01 -9.24299777e-01 4.70802695e-01
6.25476122e-01 2.83734053e-01 3.25701624e-01 -1.77222872e+00
-1.03108838e-01 1.01334882e+00 -5.89571893e-01 8.57277632e-01
6.73081696e-01 2.66312510e-01 1.08349288e+00 -9.77039218e-01
8.89641285e-01 7.79908955e-01 7.98462629e-01 4.52692658e-01
-9.77809191e-01 -5.98441720e-01 4.32982713e-01 4.44156587e-01
-1.97807407e+00 -5.88658869e-01 4.14159358e-01 -2.36838832e-01
1.35344160e+00 3.17913443e-01 1.02731574e+00 1.85092211e+00
1.86164603e-01 1.30185103e+00 7.87908196e-01 1.49087384e-01
2.96326041e-01 -5.00280261e-01 -6.57821214e-03 6.70280218e-01
-9.60948318e-02 8.83669928e-02 -1.16377878e+00 -1.02512635e-01
7.71870434e-01 -2.05919325e-01 -2.10473672e-01 -1.62089989e-01
-1.68018270e+00 2.80596465e-01 5.68606615e-01 5.71057975e-01
-4.25350368e-01 1.05485809e+00 5.73762238e-01 1.16784533e-03
5.03459632e-01 3.02156746e-01 -4.06468242e-01 -9.40704465e-01
-7.23670363e-01 4.16481912e-01 5.02042770e-01 1.00411201e+00
-1.36032160e-02 -1.48528600e-02 -4.36930865e-01 4.04707998e-01
9.75043923e-02 2.29084820e-01 1.76230893e-01 -1.25470340e+00
6.64950311e-01 3.01788688e-01 9.49992537e-02 -7.59723842e-01
-6.41120553e-01 -3.84696841e-01 -6.17201328e-01 -2.55780816e-02
4.77089912e-01 -4.32079077e-01 -4.94897187e-01 1.53311908e+00
1.82031155e-01 7.35319197e-01 -2.55036026e-01 8.29682112e-01
8.15613449e-01 5.80382407e-01 1.60718828e-01 -1.47197098e-01
1.09561265e+00 -1.36290717e+00 -8.38938773e-01 -9.12663341e-02
9.14221704e-01 -1.57255873e-01 1.01669621e+00 2.93351620e-01
-1.13955212e+00 -2.02675879e-01 -1.10493541e+00 1.33453891e-01
-4.58401650e-01 1.96056679e-01 7.96666384e-01 1.77746430e-01
-1.30703306e+00 4.11152810e-01 -1.31030738e+00 -8.32135618e-01
7.49522507e-01 4.89839129e-02 -4.68622953e-01 3.49957973e-01
-1.27311766e+00 8.58948112e-01 3.57351214e-01 6.60758391e-02
-1.45142400e+00 -6.25629783e-01 -6.16664886e-01 -2.32378095e-01
6.91916943e-01 -2.83518374e-01 1.48228872e+00 -2.74837852e-01
-1.18993449e+00 3.99251550e-01 -1.57100827e-01 -7.07605541e-01
7.99595714e-01 -3.43199998e-01 -7.06623197e-01 -4.74777352e-03
1.05574973e-01 5.38916051e-01 -9.79756378e-03 -6.51790619e-01
-6.74560905e-01 -8.77193362e-02 2.68207252e-01 3.92984390e-01
-4.12915684e-02 2.58181542e-01 -6.21190250e-01 -7.44905055e-01
-2.38528863e-01 -8.91852498e-01 -3.41655165e-01 -1.21695332e-01
-3.01597446e-01 -5.55751801e-01 7.28093028e-01 -3.94184291e-01
1.57426727e+00 -2.01979518e+00 -2.56659985e-02 1.56672210e-01
4.96648550e-01 3.24369743e-02 -2.99363405e-01 9.16648746e-01
-2.30584174e-01 1.40798539e-01 8.69871154e-02 -2.47402132e-01
-8.88218135e-02 -1.45929351e-01 3.35215069e-02 9.26781833e-01
-2.01729953e-01 1.45080388e+00 -1.19288492e+00 -4.19642359e-01
3.22276413e-01 3.07013422e-01 -3.84664461e-02 -2.39344686e-01
-2.25108087e-01 5.62725544e-01 -5.84962785e-01 8.63478541e-01
1.46056945e-02 -4.28461969e-01 3.03930402e-01 3.24810371e-02
-2.05326378e-01 4.56744820e-01 -8.21438074e-01 2.50487304e+00
-1.93653002e-01 1.01001740e+00 -1.29765138e-01 -7.58235455e-01
4.29577619e-01 5.21278501e-01 1.18566537e+00 -6.64102137e-01
4.61066216e-02 -1.98266618e-02 -1.88284162e-02 -3.49683464e-01
2.53545016e-01 7.00867593e-01 -4.01670873e-01 5.19510388e-01
1.37154236e-01 5.67301691e-01 6.19500518e-01 6.88003600e-01
2.12558842e+00 5.78573048e-01 3.26001376e-01 -3.90423208e-01
3.30674708e-01 7.33232945e-02 5.04947305e-01 8.45130622e-01
-7.31926322e-01 4.24647748e-01 6.90424144e-01 -5.85832536e-01
-6.38368666e-01 -1.14462876e+00 2.50138074e-01 1.14743888e+00
1.36102796e-01 -1.22851801e+00 -5.75586140e-01 -1.09097385e+00
-2.75485605e-01 5.80801427e-01 -8.70796561e-01 2.16072708e-01
-6.60435319e-01 -4.35373187e-01 1.14262140e+00 9.45108950e-01
6.28814638e-01 -1.20010233e+00 -4.39633518e-01 7.88630694e-02
-4.12505001e-01 -1.14262497e+00 -5.79865217e-01 2.08664402e-01
-5.49409509e-01 -1.17888534e+00 -5.91988623e-01 -3.02047133e-01
1.08046293e-01 8.77351761e-02 1.05927610e+00 -1.17429413e-01
-2.38366991e-01 3.99214059e-01 -4.79512155e-01 -7.24951476e-02
1.04531497e-01 5.22577822e-01 7.42371306e-02 -3.23116452e-01
3.69426548e-01 -9.23189640e-01 -8.43019068e-01 6.50728703e-01
-1.94111377e-01 6.16224818e-02 4.35116798e-01 2.08291367e-01
5.64426541e-01 -1.96968392e-01 4.80644643e-01 -5.02892792e-01
6.08752429e-01 -7.98730433e-01 -4.42633182e-01 1.73421577e-01
-4.93457764e-01 -2.69363642e-01 -5.62583953e-02 -2.81528026e-01
-7.75020480e-01 3.75401527e-01 -1.67023912e-01 -3.83413315e-01
-1.14047818e-01 7.25409865e-01 5.77899218e-02 2.16766804e-01
1.12729919e+00 1.18730552e-01 -4.48131323e-01 -3.06777954e-01
5.51671922e-01 3.93113583e-01 7.13678420e-01 -4.49101537e-01
3.93721372e-01 5.50699830e-01 -1.04699716e-01 -5.36945581e-01
-7.19123125e-01 -7.79435694e-01 -6.41247153e-01 -7.25820005e-01
9.32587266e-01 -1.18268597e+00 -9.76821661e-01 3.89987320e-01
-1.18657315e+00 -8.99597228e-01 -5.14282227e-01 4.76269126e-01
-6.54216886e-01 -1.19998068e-01 -7.12147772e-01 -5.57699323e-01
-6.68147579e-02 -6.67809367e-01 7.85606742e-01 -1.96919567e-03
-6.34325087e-01 -1.04659760e+00 4.47685361e-01 4.56740588e-01
4.91346419e-01 4.44185227e-01 -3.05558056e-01 -7.19652116e-01
-4.34179306e-01 -2.98980892e-01 2.20455043e-02 -1.04730576e-01
1.54184373e-02 -1.57097116e-01 -7.68518090e-01 -1.59943238e-01
-6.35622621e-01 -4.58832949e-01 8.39583516e-01 4.33444053e-01
1.25973117e+00 -1.19283095e-01 -6.74561381e-01 3.61747891e-01
9.43472564e-01 1.21848017e-01 8.50330234e-01 2.60299176e-01
6.52673483e-01 7.46927112e-02 6.77546263e-01 5.48825979e-01
4.35103327e-01 9.34345782e-01 5.17319083e-01 8.11145082e-02
-3.45370263e-01 -5.21360517e-01 6.82862699e-01 6.11724317e-01
-6.84365690e-01 -6.94397330e-01 -1.30653846e+00 8.10606360e-01
-2.47180319e+00 -1.24562359e+00 -4.51312184e-01 1.56714964e+00
4.64857459e-01 -1.97047349e-02 6.08208418e-01 -6.93761855e-02
4.53016698e-01 7.57657230e-01 -3.89301807e-01 2.34487072e-01
-2.19144002e-01 2.77982838e-02 8.02836120e-01 4.75448221e-01
-1.30482125e+00 1.11122298e+00 7.11354494e+00 1.24714041e+00
-4.78599966e-01 7.00350463e-01 1.87442869e-01 -7.35406995e-01
2.77762413e-01 1.60773285e-02 -5.97640276e-01 4.42701221e-01
1.20189273e+00 -2.87974209e-01 4.76639152e-01 6.08869255e-01
6.37865245e-01 -4.54857290e-01 -9.91135776e-01 1.04095554e+00
-5.23452908e-02 -1.80912316e+00 -6.83622062e-01 9.65010896e-02
7.60401666e-01 6.32405579e-01 -3.79537702e-01 2.89877325e-01
6.18262231e-01 -1.04115570e+00 6.95008039e-01 3.94170403e-01
7.26395786e-01 -4.17353600e-01 5.41163385e-01 1.92801565e-01
-1.64945400e+00 7.31633157e-02 1.76045358e-01 -2.70393968e-01
5.02488911e-01 2.54018962e-01 -4.17317510e-01 5.41669667e-01
9.16158915e-01 1.50340343e+00 -6.01915598e-01 1.03263998e+00
-7.68091917e-01 7.98687458e-01 -2.76051134e-01 -2.24313647e-01
3.37012678e-01 3.05577010e-01 6.82839870e-01 1.20030630e+00
1.50947258e-01 1.64021358e-01 7.43371844e-02 5.06941617e-01
-3.43595743e-01 -2.45367810e-01 -6.70176387e-01 -2.43357658e-01
5.12442470e-01 1.26129758e+00 -7.79411912e-01 -3.40046793e-01
-1.89051002e-01 9.48403418e-01 3.13948721e-01 5.08796573e-01
-1.45382392e+00 -9.91251096e-02 6.52141511e-01 3.58192146e-01
-1.22034654e-01 -3.48823547e-01 -1.37003154e-01 -9.98149097e-01
1.30044818e-01 -5.79631209e-01 5.98138869e-01 -9.89393950e-01
-7.96744049e-01 4.19400305e-01 3.31302375e-01 -1.19467640e+00
-5.98751493e-02 -2.02960417e-01 -5.48331738e-01 4.39072520e-01
-5.80801129e-01 -1.13433528e+00 -4.61112231e-01 7.64186442e-01
5.83223522e-01 -2.40492523e-01 6.02534115e-01 6.25571489e-01
-5.79757750e-01 4.24915969e-01 -2.35139459e-01 4.58609253e-01
4.79980975e-01 -1.03526866e+00 9.17036712e-01 8.26925099e-01
4.45788831e-01 1.16809070e-01 5.18107116e-01 -6.54797733e-01
-1.08868647e+00 -9.68434989e-01 8.71807456e-01 -1.08586669e+00
1.17581403e+00 -7.28864372e-01 -3.07751060e-01 1.16394472e+00
5.41028380e-01 4.71105456e-01 3.39508355e-01 3.08410853e-01
-1.49370208e-01 1.20219387e-01 -6.43216014e-01 6.43071890e-01
1.83596313e+00 -4.47002530e-01 -4.33982201e-02 7.71506667e-01
5.55549145e-01 -5.09005368e-01 -9.99520123e-01 2.78600633e-01
6.45953417e-01 -7.70459414e-01 8.04433465e-01 -6.94716334e-01
1.45894364e-01 -7.19733894e-01 -3.40541862e-02 -1.02608609e+00
-3.51746172e-01 -7.79338360e-01 -6.31141186e-01 8.29362869e-01
7.24706173e-01 -3.78191113e-01 1.11060560e+00 -1.13038169e-02
-1.98547229e-01 -8.63654673e-01 -1.30828512e+00 -1.20665157e+00
-4.29135412e-01 -8.35770965e-01 2.01730072e-01 9.21454251e-01
4.19283718e-01 3.41096133e-01 -6.97509944e-01 -1.37676284e-01
3.44091237e-01 -5.99693477e-01 7.55322754e-01 -5.56459963e-01
-2.70691931e-01 -4.63432878e-01 -5.40318012e-01 -1.15721500e+00
-1.68302745e-01 -9.40038979e-01 -9.60899964e-02 -1.86614060e+00
4.96155918e-02 -1.34904340e-01 -5.46074748e-01 9.62182164e-01
4.52396512e-01 5.04861414e-01 1.42562315e-01 3.18850309e-01
-1.42153466e+00 5.83016753e-01 9.58106399e-01 -1.40045345e-01
-2.71756738e-01 7.17203543e-02 -3.77379358e-01 6.09814167e-01
1.33715796e+00 -5.99880993e-01 -5.93621552e-01 -3.38125527e-01
3.87891650e-01 1.96838696e-02 4.36555505e-01 -1.28694201e+00
4.49054658e-01 -2.68884540e-01 2.31727958e-01 -7.74121761e-01
5.25314510e-01 -4.82636571e-01 4.66288626e-01 4.04003441e-01
-5.72353244e-01 3.08141768e-01 2.31858596e-01 6.79774702e-01
1.55187482e-02 3.84755075e-01 1.03903711e-01 -3.87036093e-02
-1.00341082e+00 4.34143603e-01 -6.97963119e-01 1.45483166e-01
1.40091848e+00 -6.55289814e-02 -9.55102921e-01 -5.55514276e-01
-1.07548034e+00 5.08086145e-01 7.52067193e-02 6.76468253e-01
3.49390388e-01 -1.67560315e+00 -7.47489572e-01 -5.75413823e-01
4.46102381e-01 -3.92909199e-01 3.73680234e-01 1.43650544e+00
-7.03697205e-01 5.76212347e-01 -1.31830886e-01 -4.13490027e-01
-1.14539886e+00 3.21694881e-01 5.78352749e-01 -3.94760489e-01
-6.33618236e-01 8.38448226e-01 -3.20247650e-01 -6.80944249e-02
4.46229219e-01 -1.49681792e-01 7.09777921e-02 -1.76797181e-01
6.14409328e-01 7.98776209e-01 -1.22071110e-01 -4.31111991e-01
-1.02797198e+00 -7.72772878e-02 3.69964331e-01 -4.96797413e-01
1.19308245e+00 -1.29449610e-02 1.08404852e-01 5.55436552e-01
8.69097412e-01 -3.91780101e-02 -1.26137888e+00 -9.88259763e-02
2.47124851e-01 -3.07115018e-01 7.20351487e-02 -1.25460100e+00
-1.18894100e+00 5.53495824e-01 6.39392972e-01 1.02243483e-01
8.42894375e-01 4.59506929e-01 3.87747496e-01 3.65875602e-01
6.12266183e-01 -1.21872389e+00 2.81959653e-01 3.11660528e-01
1.18311954e+00 -8.46481085e-01 2.55547315e-01 -9.68865529e-02
-6.75832510e-01 4.75034952e-01 7.70681739e-01 -1.43596783e-01
7.37567782e-01 5.07734537e-01 -2.47973111e-02 -5.90030074e-01
-1.06221879e+00 -2.98710823e-01 -9.63500366e-02 7.60429502e-01
3.92690718e-01 9.67914909e-02 -4.13368642e-01 4.74013031e-01
3.07539105e-01 1.27916753e-01 4.32471663e-01 1.13243818e+00
-4.89979349e-02 -8.97324085e-01 1.26518667e-01 1.11743018e-01
-3.49998891e-01 -1.53989822e-01 -6.81934595e-01 9.60031450e-01
6.34296285e-03 9.67033863e-01 -3.06380801e-02 -6.76076829e-01
5.63444614e-01 -1.01893567e-01 4.65290099e-01 -3.59354734e-01
-6.46349430e-01 2.42095124e-02 6.55778587e-01 -1.41841567e+00
-6.51004791e-01 -8.14661682e-01 -1.28647149e+00 -8.38037252e-01
2.19017357e-01 1.18871860e-01 4.85332787e-01 4.84699219e-01
6.21221483e-01 8.24361026e-01 9.86585915e-02 -6.74383342e-01
2.13443443e-01 -1.16771877e+00 -6.34584367e-01 -2.36433327e-01
-6.60034046e-02 -6.31005943e-01 -2.38835856e-01 -1.78943291e-01] | [8.388585090637207, 0.4816572070121765] |
e46c6049-3485-4957-a679-2c8955e4a7e5 | requirement-analysis-for-an-artificial | 2110.12464 | null | https://arxiv.org/abs/2110.12464v2 | https://arxiv.org/pdf/2110.12464v2.pdf | Requirement analysis for an artificial intelligence model for the diagnosis of the COVID-19 from chest X-ray data | There are multiple papers published about different AI models for the COVID-19 diagnosis with promising results. Unfortunately according to the reviews many of the papers do not reach the level of sophistication needed for a clinically usable model. In this paper I go through multiple review papers, guidelines, and other relevant material in order to generate more comprehensive requirements for the future papers proposing a AI based diagnosis of the COVID-19 from chest X-ray data (CXR). Main findings are that a clinically usable AI needs to have an extremely good documentation, comprehensive statistical analysis of the possible biases and performance, and an explainability module. | ['Tuomo Kalliokoski'] | 2021-10-24 | null | null | null | null | ['covid-19-detection'] | ['medical'] | [-1.60170779e-01 9.73949805e-02 -2.28052646e-01 -5.52596211e-01
-4.81955111e-01 -2.88718641e-01 1.72448292e-01 3.12183350e-01
5.35241328e-02 5.28565407e-01 2.32331529e-01 -9.71237421e-01
-9.83669996e-01 -3.41362834e-01 -1.11538142e-01 -4.29246724e-01
1.09945498e-01 1.05024087e+00 7.96426609e-02 2.40148142e-01
2.79852152e-01 5.27528942e-01 -1.10439312e+00 3.85903120e-01
6.99364305e-01 7.46757209e-01 1.72390908e-01 7.87718892e-01
2.26585031e-01 1.32590246e+00 -7.48137414e-01 -3.32775563e-01
5.58989830e-02 -7.74545431e-01 -7.72292912e-01 -2.57253140e-01
1.49894595e-01 -5.32914758e-01 7.93582350e-02 3.03727686e-01
5.54681897e-01 -4.64542985e-01 1.08236325e+00 -1.27452660e+00
-7.23387718e-01 5.35287559e-01 -5.36854519e-03 5.13308108e-01
5.22568762e-01 3.68621767e-01 5.49260914e-01 -6.32246256e-01
6.55919492e-01 8.68265808e-01 8.04053962e-01 5.17781138e-01
-6.30081892e-01 -7.07818747e-01 -1.16827644e-01 2.13868305e-01
-1.16317987e+00 4.12509382e-01 4.13637489e-01 -8.11573803e-01
1.28805923e+00 8.71026695e-01 1.22136366e+00 1.07151711e+00
8.24702919e-01 1.63015544e-01 1.35854328e+00 -5.24833798e-01
7.31991604e-02 6.18336499e-01 6.60316408e-01 4.39187616e-01
8.11369836e-01 7.88607895e-02 -5.10491394e-02 -3.61127466e-01
7.23494589e-01 8.28333125e-02 3.97851989e-02 -2.75020022e-02
-8.55348110e-01 1.02296674e+00 2.72584140e-01 5.12350500e-01
-5.90454817e-01 -1.24323510e-01 4.40695077e-01 1.79477721e-01
2.45829597e-01 7.81883538e-01 -2.42312521e-01 -2.31534287e-01
-9.14190948e-01 4.56827968e-01 7.26555049e-01 6.33199632e-01
-3.35575968e-01 1.46881238e-01 -1.22095138e-01 4.21321720e-01
5.15853763e-01 7.43332028e-01 4.12459552e-01 -8.58890116e-01
1.10091113e-01 8.09627771e-01 -1.14928991e-01 -1.07502592e+00
-8.21961403e-01 -2.85038978e-01 -7.13533878e-01 1.09463930e-01
2.58568898e-02 -2.64266521e-01 -1.09181273e+00 9.88568664e-01
-2.07083449e-01 -4.91106659e-01 -6.62160069e-02 8.42891514e-01
1.19783998e+00 1.70249462e-01 3.34160596e-01 -4.44781542e-01
1.43469369e+00 -5.80122709e-01 -1.14878047e+00 -2.28343368e-01
7.15575635e-01 -6.55322313e-01 9.35525596e-01 8.55274856e-01
-1.40764821e+00 -4.11669433e-01 -9.60647702e-01 2.05991536e-01
-1.78754076e-01 1.20892070e-01 8.12390625e-01 8.96848202e-01
-8.01626384e-01 2.52430201e-01 -7.87260234e-01 -4.16279405e-01
1.76531017e-01 3.59801888e-01 8.49385411e-02 4.81573083e-02
-1.12525821e+00 1.62520695e+00 1.32937267e-01 2.34108102e-02
-5.64271688e-01 -8.48860383e-01 -5.10537803e-01 -3.50430816e-01
6.26729280e-02 -1.32947564e+00 9.48929429e-01 -5.85056484e-01
-7.21751153e-01 9.03991461e-01 1.24634914e-02 -4.43319738e-01
4.91855413e-01 -1.19604670e-01 -4.07830536e-01 4.60284144e-01
-1.57458723e-01 8.78814161e-02 4.22606409e-01 -1.26004457e+00
-5.78511000e-01 -3.57753694e-01 -2.27510229e-01 2.76070996e-03
2.29100019e-01 5.84185481e-01 1.69822816e-02 -7.66210556e-01
-1.80620868e-02 -1.01077342e+00 -3.84372532e-01 -4.95315045e-01
-1.49215505e-01 -2.85302967e-01 4.14748430e-01 -8.46140563e-01
1.93901348e+00 -1.67692101e+00 -1.15321063e-01 3.53970408e-01
6.65091515e-01 3.20999414e-01 4.65708226e-01 7.16950476e-01
-2.84880668e-01 3.00691247e-01 -1.67413741e-01 2.21103206e-01
-1.86087966e-01 3.60240787e-01 -2.11473674e-01 3.89089972e-01
2.73005962e-01 8.85716856e-01 -6.08484566e-01 -5.39472938e-01
6.41099572e-01 5.27579367e-01 -3.67624432e-01 1.37527615e-01
2.54048020e-01 1.83345586e-01 -6.26952410e-01 6.48342907e-01
5.51170409e-01 -4.57694501e-01 4.96878754e-03 1.22580878e-01
2.41195560e-02 4.09049451e-01 -8.60753596e-01 7.42257357e-01
-1.98309243e-01 5.48430800e-01 -3.91379774e-01 -8.24406087e-01
8.10091972e-01 8.23476195e-01 7.45864213e-01 -3.62372398e-01
3.62439573e-01 5.49604177e-01 5.85696161e-01 -8.95514905e-01
-2.31134426e-03 -4.39299524e-01 2.38784462e-01 5.61794877e-01
-4.38125163e-01 -6.84168696e-01 -3.62965092e-02 1.57366708e-01
1.17397642e+00 -5.58859766e-01 5.80937624e-01 -4.61521834e-01
3.86094749e-01 4.66423452e-01 2.35985458e-01 9.85476375e-01
2.01551486e-02 9.62487340e-01 2.22998187e-01 -3.99830848e-01
-9.52735603e-01 -7.88202345e-01 -7.40895271e-01 1.47596031e-01
-4.63890970e-01 -3.64687920e-01 -7.73260295e-01 -4.79812592e-01
-8.87459069e-02 1.00015366e+00 -6.94458246e-01 4.67580520e-02
-3.43384564e-01 -7.27478862e-01 5.04204631e-01 9.27763760e-01
-1.74344361e-01 -1.10946655e+00 -1.19231355e+00 1.33045390e-01
1.42773107e-01 -6.05077684e-01 5.55435419e-01 1.33325741e-01
-1.05415809e+00 -1.20488489e+00 -6.98432028e-01 -6.05993450e-01
6.60035849e-01 -1.14844246e-02 1.04633069e+00 6.57167673e-01
-5.96925616e-01 7.26167083e-01 -4.23212141e-01 -1.08224833e+00
-6.56614780e-01 -2.78833240e-01 -6.72091097e-02 -1.14120913e+00
7.51245081e-01 -1.22657068e-01 -6.44709766e-01 2.04587623e-01
-8.69273007e-01 1.35121718e-01 5.94196975e-01 7.32658803e-01
3.33032131e-01 -3.59997191e-02 4.97304767e-01 -9.26569104e-01
9.35716987e-01 -7.55180657e-01 1.16409905e-01 1.91653118e-01
-1.32479548e+00 -5.18406808e-01 2.15909541e-01 -2.14843184e-01
-8.70623469e-01 -5.42604923e-01 -9.23063904e-02 -1.42447919e-01
-2.26776555e-01 5.98028481e-01 7.53536403e-01 1.69180930e-01
9.02451396e-01 -2.81130165e-01 -6.37404844e-02 -3.83879662e-01
-2.89066046e-01 1.09877920e+00 1.59293130e-01 -2.35104457e-01
5.61258912e-01 2.34797090e-01 4.91770953e-02 -4.55452442e-01
-6.49748206e-01 -4.95962203e-01 -5.11766911e-01 -1.24868453e-01
9.71956611e-01 -5.90272069e-01 -3.14951658e-01 8.69233813e-03
-7.62774110e-01 -1.22757711e-01 -4.26143229e-01 1.16875303e+00
-6.33981705e-01 -6.29159287e-02 -6.03062451e-01 -1.01903188e+00
-6.00525141e-01 -1.52945137e+00 4.10269558e-01 1.46218657e-01
-1.04297984e+00 -1.24584377e+00 3.11259717e-01 5.05119562e-01
5.77690065e-01 2.85522342e-01 1.20646358e+00 -9.82846856e-01
-7.40728602e-02 -4.91911680e-01 -1.30200207e-01 1.13312855e-01
2.30420277e-01 5.37077308e-01 -7.11810470e-01 -6.05403632e-02
5.31816483e-01 -2.31148213e-01 2.77767807e-01 8.34388077e-01
1.18247008e+00 -3.41149747e-01 -5.44795454e-01 1.60378531e-01
1.27731788e+00 9.52227473e-01 5.22254467e-01 2.59044349e-01
4.22178626e-01 6.70847535e-01 7.99556911e-01 3.07774007e-01
3.04881305e-01 2.97563255e-01 4.45647761e-02 -1.84508070e-01
-3.84111293e-02 2.77545601e-01 -2.37919182e-01 9.89023626e-01
-4.25940514e-01 -2.53585577e-01 -1.54373527e+00 6.05688155e-01
-1.40762496e+00 -8.21756184e-01 -7.38506913e-01 1.89182818e+00
3.45336884e-01 2.09249988e-01 1.29044726e-01 3.27027977e-01
2.48357914e-02 -3.02828312e-01 -9.86699983e-02 -1.09107733e+00
2.59260714e-01 2.43244946e-01 5.61556816e-02 3.24087173e-01
-5.09799480e-01 -8.75395257e-03 8.89443493e+00 1.75549120e-01
-8.05472910e-01 -1.44270718e-01 6.03664041e-01 7.55122339e-04
-6.83544338e-01 7.43795633e-02 -3.72003585e-01 3.41589212e-01
1.25732708e+00 -2.86086917e-01 -1.22383766e-01 6.11039042e-01
4.27282393e-01 -3.37512195e-01 -8.94597530e-01 6.64148211e-01
1.33707359e-01 -1.18496108e+00 3.37175392e-02 -9.08685010e-03
4.22421247e-01 -1.73518836e-01 1.32854149e-01 1.04053766e-01
-2.83857412e-03 -1.20844567e+00 4.09038991e-01 7.54835010e-01
7.65305519e-01 -2.98404694e-01 1.07962608e+00 6.84414990e-03
-4.37529743e-01 -1.78667262e-01 -3.95729184e-01 -4.15868580e-01
1.58987775e-01 2.13394493e-01 -1.21529651e+00 7.24794149e-01
6.89458370e-01 4.35682952e-01 -6.88227773e-01 9.51617002e-01
9.79402587e-02 1.06324160e+00 -2.84669161e-01 -2.38326669e-01
2.29980096e-01 -5.73289543e-02 5.04917920e-01 1.30670047e+00
1.92328259e-01 5.49459457e-01 -4.34330910e-01 6.80736244e-01
8.71384382e-01 2.82115310e-01 -8.91180098e-01 -1.29656851e-01
3.53915483e-01 8.73992562e-01 -6.71425104e-01 -4.95578200e-01
-6.91213787e-01 3.66729170e-01 -3.96154195e-01 3.40489373e-02
-7.19049633e-01 1.50905773e-01 1.88771650e-01 5.63321173e-01
-2.68508554e-01 1.84365496e-01 -9.50986207e-01 -5.33676028e-01
-1.17666602e-01 -1.25320745e+00 6.72898948e-01 -1.28968108e+00
-1.34807360e+00 5.06918609e-01 7.23721206e-01 -1.03932869e+00
-6.51146054e-01 -7.50604987e-01 -6.01366520e-01 8.63329828e-01
-5.72360933e-01 -9.86759305e-01 -3.39835405e-01 1.10369056e-01
3.16924930e-01 -2.53767490e-01 1.14633715e+00 -1.41047360e-02
-3.60547155e-01 2.23740876e-01 -2.31333911e-01 -2.50859618e-01
6.16258681e-01 -1.31713653e+00 -2.22045816e-02 3.23272049e-01
-3.58076394e-01 9.79698896e-01 8.52375567e-01 -1.10584199e+00
-1.04359818e+00 -3.27214211e-01 9.69607353e-01 -1.10688961e+00
3.95383745e-01 3.82669210e-01 -9.11780417e-01 7.59683430e-01
3.58756959e-01 -6.21919155e-01 1.01183915e+00 -1.45758867e-01
3.99969637e-01 1.67865604e-01 -1.22187150e+00 4.14325535e-01
5.76982260e-01 -2.43387930e-02 -1.09973073e+00 3.54991227e-01
3.65872979e-01 -3.72819781e-01 -1.24154747e+00 6.83472991e-01
6.15150213e-01 -9.08062577e-01 8.94710660e-01 -8.22352350e-01
3.62683028e-01 1.91830188e-01 2.48294309e-01 -8.97466362e-01
-3.97923082e-01 -3.11819434e-01 2.91928530e-01 7.79958785e-01
4.61093456e-01 -6.53515935e-01 1.85810953e-01 1.17277813e+00
-1.51257798e-01 -1.03322887e+00 -4.91962612e-01 -3.19695115e-01
2.37691194e-01 -8.09776366e-01 3.66839945e-01 9.60853040e-01
2.29873866e-01 1.49174273e-01 -1.09287985e-01 -7.21043646e-02
2.51060754e-01 -1.22984491e-01 4.45792884e-01 -1.23921645e+00
7.07384869e-02 -3.18775713e-01 -4.49100822e-01 5.05848937e-02
-7.03868628e-01 -3.87629241e-01 -6.50590301e-01 -2.18111777e+00
6.61086261e-01 -6.76416159e-01 -2.85130709e-01 2.14365348e-01
-2.95101106e-01 4.91903760e-02 7.40118772e-02 3.66826564e-01
9.76525247e-02 -3.99629503e-01 1.13620138e+00 3.03351432e-01
-1.68369189e-01 -2.98067946e-02 -8.50635886e-01 9.97344375e-01
1.04531193e+00 -8.84613991e-01 -7.68795252e-01 -1.57897890e-01
4.87528473e-01 3.88136774e-01 4.08152848e-01 -9.65234935e-01
-1.62441403e-01 -4.32774991e-01 5.55256665e-01 -9.95816410e-01
1.87219813e-01 -1.13010979e+00 7.23608673e-01 9.49362099e-01
-2.19261020e-01 6.93729281e-01 3.48652095e-01 7.94225782e-02
1.08742733e-02 -6.19909465e-01 3.92682701e-01 -2.03256413e-01
-6.75968602e-02 -8.22063088e-02 -8.48485768e-01 1.14414088e-01
1.35481656e+00 -2.80409396e-01 -1.84254408e-01 -4.82493162e-01
-5.75860083e-01 -1.26169667e-01 5.37649333e-01 2.49487057e-01
6.93752110e-01 -9.09509599e-01 -6.61422729e-01 2.34934449e-01
2.49249078e-02 -1.44842833e-01 2.32312098e-01 1.29887319e+00
-1.12547457e+00 9.91692424e-01 -4.21190441e-01 -5.08535564e-01
-1.41142356e+00 8.56723964e-01 3.61807078e-01 -4.02999580e-01
-8.63630891e-01 3.28792602e-01 1.53254777e-01 1.77334011e-01
2.25073844e-01 -5.81410825e-01 -3.79301488e-01 3.48473154e-02
5.77517450e-01 6.11765563e-01 6.64536878e-02 -3.37067902e-01
-6.72676086e-01 5.21708846e-01 -4.17474061e-02 -1.83830142e-01
1.19001698e+00 -1.94679350e-01 6.69612736e-02 7.74843097e-01
5.82538784e-01 -7.90835544e-02 -3.40421051e-01 4.60176349e-01
-2.27801353e-01 -3.91326308e-01 -1.13736857e-02 -1.22577548e+00
-5.65765142e-01 8.75634730e-01 6.33428574e-01 3.64164621e-01
1.14480436e+00 5.79903759e-02 9.05382931e-02 7.55076557e-02
-4.80349325e-02 -1.05707288e+00 -1.23263456e-01 -8.59850124e-02
1.24038470e+00 -1.00218034e+00 6.22367322e-01 -1.80719286e-01
-1.26047742e+00 1.01015651e+00 3.68992716e-01 -1.42657295e-01
8.59840810e-01 4.30803776e-01 5.37976742e-01 -8.35773528e-01
-8.07190001e-01 2.27462068e-01 5.78573644e-01 8.12633097e-01
6.73093140e-01 1.86354205e-01 -1.13356709e+00 9.17417407e-01
-1.39314756e-01 1.97859213e-01 5.45666575e-01 8.35451543e-01
-1.42719537e-01 -9.71272469e-01 -5.57845771e-01 9.17472780e-01
-8.47086966e-01 1.74912035e-01 -8.06284964e-01 1.39562631e+00
1.20812869e-02 9.68605399e-01 -1.19300038e-02 -2.99287081e-01
3.45072687e-01 7.80686513e-02 5.13254404e-01 -5.98580062e-01
-1.09088850e+00 4.30791527e-02 3.01664531e-01 -3.47878993e-01
-5.22722483e-01 -6.03379011e-01 -1.28665018e+00 -3.67172658e-01
-3.42682362e-01 6.31328374e-02 6.07161522e-01 7.12307394e-01
-3.80022712e-02 8.20958078e-01 1.39801189e-01 5.80900237e-02
-3.82205397e-01 -1.13748050e+00 -3.98360223e-01 3.04957300e-01
1.12976477e-01 -6.59490764e-01 -5.16396940e-01 -3.63012850e-01] | [8.469444274902344, 5.77777099609375] |
135b43c8-95d3-4994-97cb-ba2332c1a9e3 | a-monaural-speech-enhancement-method-for | 1906.08415 | null | https://arxiv.org/abs/1906.08415v1 | https://arxiv.org/pdf/1906.08415v1.pdf | A Monaural Speech Enhancement Method for Robust Small-Footprint Keyword Spotting | Robustness against noise is critical for keyword spotting (KWS) in real-world environments. To improve the robustness, a speech enhancement front-end is involved. Instead of treating the speech enhancement as a separated preprocessing before the KWS system, in this study, a pre-trained speech enhancement front-end and a convolutional neural networks (CNNs) based KWS system are concatenated, where a feature transformation block is used to transform the output from the enhancement front-end into the KWS system's input. The whole model is trained jointly, thus the linguistic and other useful information from the KWS system can be back-propagated to the enhancement front-end to improve its performance. To fit the small-footprint device, a novel convolution recurrent network is proposed, which needs fewer parameters and computation and does not degrade performance. Furthermore, by changing the input features from the power spectrogram to Mel-spectrogram, less computation and better performance are obtained. our experimental results demonstrate that the proposed method significantly improves the KWS system with respect to noise robustness. | ['HUI ZHANG', 'Zhihao Du', 'Yue Gu', 'Xueliang Zhang'] | 2019-06-20 | null | null | null | null | ['small-footprint-keyword-spotting'] | ['speech'] | [ 3.48007619e-01 4.15740535e-02 3.95866275e-01 -3.12485516e-01
-6.45649016e-01 -3.34236659e-02 2.07553521e-01 -1.90343615e-02
-7.24491835e-01 1.45124421e-01 3.48862261e-01 -6.37221634e-01
1.00348398e-01 -6.61060631e-01 -2.60021180e-01 -7.51287401e-01
3.99224371e-01 -8.11441541e-01 4.01554316e-01 -4.40844208e-01
-1.72030076e-01 3.10981810e-01 -1.21488941e+00 1.35962635e-01
9.46224153e-01 1.21803403e+00 6.29560769e-01 7.90263891e-01
1.36149555e-01 7.00145781e-01 -6.61383033e-01 -4.37190756e-02
9.64791030e-02 -3.54118407e-01 -2.39603013e-01 -9.28679183e-02
-2.10434884e-01 -5.38585603e-01 -8.80466223e-01 1.25589800e+00
1.06274545e+00 5.33308744e-01 2.33141422e-01 -7.94779062e-01
-6.57661378e-01 7.45144606e-01 -2.73125589e-01 1.09206401e-01
-9.14398208e-02 2.22257018e-01 5.52029729e-01 -1.00664318e+00
-2.70809350e-03 1.22577584e+00 7.41081059e-01 2.57317275e-01
-7.95007586e-01 -8.58942091e-01 1.43425658e-01 3.78154099e-01
-1.16796553e+00 -9.62154984e-01 9.62912858e-01 1.29492566e-01
1.11773205e+00 3.34117502e-01 5.21428883e-01 7.65576482e-01
-1.09192446e-01 8.62514079e-01 7.35400081e-01 -5.62359929e-01
1.02975532e-01 3.23301762e-01 2.97912031e-01 3.82784963e-01
-1.77691821e-02 -5.98576106e-03 -4.01503325e-01 1.87661156e-01
5.42769909e-01 7.54515976e-02 -5.76254427e-01 4.62096006e-01
-8.49355698e-01 3.77380192e-01 4.77216512e-01 5.10028780e-01
-4.85945791e-01 -3.81322950e-02 6.23261869e-01 5.20182133e-01
6.29730582e-01 5.86446859e-02 -5.45933187e-01 -2.50185192e-01
-1.18548262e+00 -1.99014813e-01 4.08351302e-01 7.37589478e-01
3.45739543e-01 5.05126357e-01 -2.89280146e-01 1.03568530e+00
3.08550507e-01 5.46602309e-01 6.95505679e-01 -2.36246213e-01
7.62387037e-01 4.69538271e-01 -3.04422155e-02 -9.12898123e-01
-3.88194531e-01 -7.77485311e-01 -1.12140477e+00 3.37866768e-02
-6.74979389e-02 -4.53988582e-01 -9.80863333e-01 1.54061341e+00
3.43401700e-01 2.20293656e-01 4.16372150e-01 1.05907190e+00
8.42936575e-01 1.13198268e+00 -3.02453898e-02 -4.08913344e-01
1.27881658e+00 -1.20537615e+00 -1.22033215e+00 -1.73470572e-01
4.15260643e-01 -1.15842521e+00 1.19509864e+00 8.83699358e-02
-1.03204453e+00 -8.22250128e-01 -1.27610517e+00 -2.60759711e-01
-5.24758160e-01 7.33976841e-01 -1.54376686e-01 4.88573372e-01
-1.07689512e+00 5.05775511e-01 -6.60745680e-01 4.87485044e-02
-5.90820238e-02 3.55076611e-01 -1.85335100e-01 4.63167876e-01
-1.74057388e+00 8.76389205e-01 5.44128776e-01 6.36489272e-01
-3.67029905e-01 -4.88013476e-01 -9.71479714e-01 5.45997143e-01
3.98035079e-01 -3.64630729e-01 1.43483377e+00 -7.83646524e-01
-2.11998963e+00 -1.30104870e-01 -2.04525009e-01 -2.12051287e-01
2.11954013e-01 -2.30768636e-01 -9.47110474e-01 5.60883246e-02
-4.35645163e-01 -7.66431075e-03 1.19268787e+00 -7.15398312e-01
-7.19153464e-01 -7.63394218e-03 -2.09769472e-01 3.58559608e-01
-9.94907677e-01 2.32133880e-01 -7.05340445e-01 -1.19691420e+00
1.60107344e-01 -5.75851738e-01 -2.36857176e-01 -3.52936983e-01
-5.41297019e-01 3.35975550e-02 1.32666790e+00 -1.33888423e+00
1.62922430e+00 -2.57237577e+00 -8.50156471e-02 4.39204365e-01
-1.01128973e-01 1.12536371e+00 -1.82125255e-01 2.57488579e-01
-1.82258114e-01 -1.59775913e-01 -9.98826474e-02 -5.52869201e-01
4.18386310e-02 -2.07392871e-01 -2.03600869e-01 3.58473659e-01
3.96711469e-01 7.72628903e-01 -5.02050519e-01 -2.18849614e-01
4.55366492e-01 7.76707768e-01 -3.00266117e-01 5.24478316e-01
3.03080350e-01 1.02676287e-01 -1.65778488e-01 1.26104891e-01
7.67691612e-01 2.11207584e-01 -1.58362806e-01 -3.25219005e-01
-3.39388609e-01 6.36788309e-01 -1.47242713e+00 1.30552161e+00
-8.32243860e-01 5.88509977e-01 4.29783076e-01 -7.91628897e-01
9.74362850e-01 6.49453819e-01 -9.81879830e-02 -9.15779293e-01
2.80734777e-01 1.25640333e-01 4.88977134e-02 -4.69450653e-01
6.71150088e-01 1.96586251e-01 1.70866594e-01 3.02177936e-01
7.37474859e-02 2.04946652e-01 -3.01389426e-01 5.41758724e-02
8.65740180e-01 -3.66547704e-01 9.77623165e-02 2.64676381e-02
8.88076127e-01 -5.43008685e-01 6.89977586e-01 3.38373005e-01
6.40087649e-02 3.10177535e-01 3.29592563e-02 -3.61823626e-02
-1.11792719e+00 -7.32115984e-01 1.95951909e-01 1.06138813e+00
2.03080714e-01 -4.65151846e-01 -1.04945588e+00 -2.37917066e-01
-4.59578782e-01 5.16708255e-01 -1.93407461e-01 -7.36291230e-01
-5.69395959e-01 -5.47613442e-01 8.41808379e-01 6.59601450e-01
8.86372864e-01 -9.77374852e-01 -2.54647613e-01 4.90070224e-01
-1.75446764e-01 -1.07958114e+00 -9.12684500e-01 2.94555813e-01
-5.05569816e-01 -4.91555393e-01 -9.43292677e-01 -9.98193204e-01
5.99324346e-01 4.62197423e-01 7.25920722e-02 4.33674827e-02
2.11593106e-01 -2.22339258e-01 -5.37270904e-01 -3.51606995e-01
-4.67419535e-01 2.32328311e-01 2.63178170e-01 3.12072903e-01
9.85936522e-02 -5.15327752e-01 -6.71749115e-01 1.42267242e-01
-8.81188273e-01 2.94817030e-01 8.55155706e-01 1.00672328e+00
2.45766491e-01 4.71284568e-01 5.90238512e-01 -3.99655938e-01
9.97915566e-01 -5.01456717e-03 -6.10186219e-01 2.71788567e-01
-4.58150357e-01 -1.45294471e-02 1.10421336e+00 -7.89698541e-01
-1.42551434e+00 1.56725552e-02 -5.37347436e-01 -3.75715047e-01
1.43709287e-01 5.04533708e-01 -5.50325811e-01 -7.64894560e-02
3.34415466e-01 6.22823954e-01 -1.34440795e-01 -8.90702724e-01
3.46642822e-01 1.51320004e+00 6.84277534e-01 1.84959412e-01
1.04588640e+00 -2.19952807e-01 -5.83056629e-01 -1.16957104e+00
-1.80423006e-01 -5.61813354e-01 -2.44916588e-01 -1.93766996e-01
4.53859270e-01 -1.00597346e+00 -7.62822509e-01 8.32011104e-01
-1.48574948e+00 -8.56529400e-02 -1.35550410e-01 7.84706414e-01
-6.68601617e-02 3.32752764e-01 -8.34627151e-01 -1.07974720e+00
-9.55276370e-01 -1.28759229e+00 8.55865359e-01 6.69596195e-01
2.08988994e-01 -4.78546113e-01 -4.57485646e-01 4.56881523e-03
9.27558362e-01 -5.12620032e-01 6.43267632e-01 -5.90990067e-01
-9.24153700e-02 -2.78236091e-01 -3.74426723e-01 9.37697470e-01
2.68757254e-01 -1.26544923e-01 -1.24227917e+00 -2.15901345e-01
4.43791449e-01 7.83414990e-02 8.84959102e-01 1.51285782e-01
1.15221894e+00 -5.19462049e-01 2.06179824e-02 7.27053761e-01
1.04229271e+00 7.32261896e-01 6.12400472e-01 1.96714967e-01
5.39898336e-01 3.91010761e-01 4.95798647e-01 2.46065289e-01
8.88561830e-03 5.44811547e-01 -8.16609561e-02 -7.67037094e-01
-3.64927262e-01 -2.49878496e-01 7.72360742e-01 1.60159123e+00
1.65508792e-01 -2.46633112e-01 -5.43343782e-01 3.70453238e-01
-1.82331061e+00 -7.71318614e-01 1.57956854e-01 1.92096913e+00
9.57948506e-01 1.47851884e-01 -2.67860562e-01 5.81510901e-01
1.02301145e+00 3.70093614e-01 -6.28970146e-01 -5.18361628e-01
1.23407446e-01 2.25386739e-01 2.82629281e-01 6.22873902e-01
-1.11153769e+00 9.12918985e-01 5.25080156e+00 1.34800911e+00
-1.40774679e+00 8.72541294e-02 5.02832234e-01 -6.71128854e-02
-1.06133766e-01 -4.06760871e-01 -5.74319720e-01 4.99632001e-01
1.11880279e+00 -6.86601549e-02 5.16521633e-01 8.54585290e-01
8.05146992e-01 1.68235257e-01 -4.37449992e-01 1.02266848e+00
-2.83671170e-01 -8.84189844e-01 -2.85817534e-01 -3.61587316e-01
1.82529226e-01 -3.26879501e-01 2.13603199e-01 3.13218087e-01
-8.78602564e-02 -5.96079230e-01 8.14482510e-01 4.09632504e-01
9.41913545e-01 -1.15322089e+00 8.61890912e-01 3.33440781e-01
-1.49229252e+00 3.70830297e-03 -2.75026649e-01 4.47444841e-02
1.41958386e-01 8.46823454e-01 -9.56292212e-01 5.81325769e-01
6.42442226e-01 1.71264902e-01 -2.44398460e-01 7.75760710e-01
-3.93073112e-01 8.10463428e-01 -3.05701256e-01 -1.76520482e-01
3.25467020e-01 -9.22812670e-02 4.89859641e-01 1.66135013e+00
2.67104000e-01 2.00583026e-01 -2.21513897e-01 5.19743741e-01
-1.06144376e-01 2.34841824e-01 -2.56942302e-01 -1.69359639e-01
3.33637625e-01 1.45648873e+00 -3.31514299e-01 -4.72181171e-01
-4.06404495e-01 1.14361143e+00 6.12156093e-02 6.76978827e-01
-9.00598168e-01 -1.44903624e+00 8.10224891e-01 -1.12652570e-01
2.87188083e-01 -1.51566505e-01 -9.87780392e-02 -1.06158316e+00
2.86625117e-01 -7.87009895e-01 -2.49211580e-01 -7.85070002e-01
-8.10601473e-01 9.67276990e-01 -7.02964187e-01 -1.21096158e+00
-1.23979062e-01 -3.60963553e-01 -6.14809513e-01 1.43960416e+00
-1.73058200e+00 -1.07420027e+00 -1.12802230e-01 6.59694672e-01
4.90448117e-01 -1.75963998e-01 6.40110910e-01 5.30938864e-01
-8.35271001e-01 9.70400691e-01 2.88960278e-01 3.27930599e-01
5.98686337e-01 -8.36839378e-01 6.38765275e-01 1.50341427e+00
-2.44038656e-01 8.05673122e-01 3.79685998e-01 -5.59220135e-01
-1.13406551e+00 -1.45303738e+00 9.09845233e-01 8.77770722e-01
4.87472594e-01 -5.57965755e-01 -9.90506351e-01 1.01334862e-01
4.18773234e-01 -2.58490909e-02 4.31257457e-01 -2.25092545e-01
-2.17354864e-01 -3.88785064e-01 -8.00136387e-01 9.38387334e-01
7.82194197e-01 -1.13881314e+00 -6.43718481e-01 -9.95615795e-02
1.28019595e+00 -5.67542076e-01 -5.90404928e-01 3.65065545e-01
3.28283846e-01 -5.00527084e-01 8.07428122e-01 -3.81415218e-01
1.14037365e-01 -6.60592794e-01 -2.60558575e-02 -1.65080845e+00
-3.20993125e-01 -8.88256848e-01 -7.47543499e-02 1.43927073e+00
6.72858417e-01 -6.36179328e-01 1.68258488e-01 4.57441032e-01
-5.14282107e-01 -6.91426814e-01 -7.95695364e-01 -6.86561465e-01
-6.51773155e-01 -5.88066995e-01 7.16551661e-01 6.68021262e-01
2.46677175e-01 2.85775095e-01 -5.71545184e-01 6.45472467e-01
-1.13326032e-02 -1.87459365e-01 5.18594384e-01 -4.21650261e-01
-3.53694975e-01 -1.91132590e-01 -1.44009233e-01 -1.19166517e+00
-1.70517251e-01 -6.95636988e-01 3.57100248e-01 -1.35210407e+00
-2.81090140e-01 -5.45229353e-02 -5.61192334e-01 5.56532621e-01
-6.15930438e-01 -1.06070809e-01 2.57152766e-01 -1.89962879e-01
-2.52755910e-01 9.80844378e-01 1.07593536e+00 -2.66927153e-01
-4.56428498e-01 1.70433328e-01 -5.85172057e-01 7.53220379e-01
8.73554647e-01 -3.61139685e-01 -4.82854098e-01 -4.15133804e-01
-3.50842506e-01 1.17236398e-01 4.81613949e-02 -1.14577115e+00
7.84669101e-01 2.35521808e-01 3.94518077e-01 -4.25232649e-01
4.83712316e-01 -9.79641676e-01 -7.95562863e-02 5.14431000e-01
-3.50690722e-01 -1.05605789e-01 3.63801539e-01 3.67979527e-01
-4.72384572e-01 -2.35473841e-01 8.41751575e-01 4.36338276e-01
-4.82392222e-01 1.81509897e-01 -5.06582260e-01 -5.09811521e-01
6.72071755e-01 -8.13544542e-02 -1.81318164e-01 -4.87759769e-01
-7.11392760e-01 2.14038603e-02 -1.22872032e-01 3.73343050e-01
8.15899074e-01 -1.33502877e+00 -3.68815362e-01 4.99271303e-01
-2.60988206e-01 -4.57216911e-02 5.16371667e-01 6.18889093e-01
-1.74098983e-01 3.10182750e-01 2.44415000e-01 -2.09457856e-02
-1.38697088e+00 3.76777768e-01 3.61046791e-01 -6.96409121e-02
-6.37942851e-01 8.33986759e-01 -1.19351625e-01 -2.38941967e-01
5.78043222e-01 -6.00880921e-01 -2.75659591e-01 -2.63973139e-02
9.71799791e-01 4.46144491e-01 3.68334472e-01 -4.99688447e-01
-1.92050770e-01 2.56796241e-01 -2.06709504e-01 -3.97692412e-01
1.53562760e+00 -3.47072065e-01 3.90229523e-02 1.24456950e-01
1.33070266e+00 -1.89775098e-02 -9.72723126e-01 -6.37584090e-01
-2.70757169e-01 -1.88420936e-01 6.06401861e-01 -7.18827128e-01
-1.27987158e+00 9.57246721e-01 6.44202113e-01 1.37900367e-01
1.73644352e+00 -5.53307414e-01 1.17472470e+00 3.41243595e-01
-2.48313934e-01 -1.65686977e+00 -1.03601381e-01 5.78190148e-01
9.43830073e-01 -8.34914267e-01 -5.26396155e-01 -3.95906180e-01
-5.72647512e-01 1.19930410e+00 4.62480545e-01 1.55654654e-01
7.03367054e-01 7.12920725e-01 1.34013936e-01 3.52839142e-01
-4.75982070e-01 -2.49695480e-01 2.08447427e-01 4.47685927e-01
1.13778427e-01 -2.26744711e-01 -1.76217452e-01 1.26698017e+00
-2.79959261e-01 -1.07464157e-01 2.08113715e-01 6.63775206e-01
-5.08083403e-01 -9.70769227e-01 -3.64041924e-01 2.93540657e-01
-3.01062286e-01 -4.09563571e-01 -9.60558355e-02 1.03617743e-01
1.80186003e-01 1.26172388e+00 -1.59203604e-01 -9.19869125e-01
8.30168843e-01 9.81134474e-02 -3.84462982e-01 -3.32419008e-01
-9.16312575e-01 3.97726625e-01 4.80605615e-03 -4.37805384e-01
-5.60866147e-02 -3.30792256e-02 -1.27214003e+00 -3.54932845e-01
-8.99791002e-01 2.69643337e-01 7.53635585e-01 9.34524775e-01
3.85588169e-01 1.03478909e+00 1.00904191e+00 -5.97598493e-01
-4.73919928e-01 -1.24113345e+00 -5.01153409e-01 1.39141336e-01
5.87582529e-01 -4.93385307e-02 -2.45633274e-01 5.82169332e-02] | [14.815231323242188, 5.955252170562744] |
9e52f603-8f56-4a7c-a7a2-65e254ae8c79 | spidr-sdf-based-neural-point-fields-for | 2210.08398 | null | https://arxiv.org/abs/2210.08398v3 | https://arxiv.org/pdf/2210.08398v3.pdf | SPIDR: SDF-based Neural Point Fields for Illumination and Deformation | Neural radiance fields (NeRFs) have recently emerged as a promising approach for 3D reconstruction and novel view synthesis. However, NeRF-based methods encode shape, reflectance, and illumination implicitly and this makes it challenging for users to manipulate these properties in the rendered images explicitly. Existing approaches only enable limited editing of the scene and deformation of the geometry. Furthermore, no existing work enables accurate scene illumination after object deformation. In this work, we introduce SPIDR, a new hybrid neural SDF representation. SPIDR combines point cloud and neural implicit representations to enable the reconstruction of higher quality object surfaces for geometry deformation and lighting estimation. meshes and surfaces for object deformation and lighting estimation. To more accurately capture environment illumination for scene relighting, we propose a novel neural implicit model to learn environment light. To enable more accurate illumination updates after deformation, we use the shadow mapping technique to approximate the light visibility updates caused by geometry editing. We demonstrate the effectiveness of SPIDR in enabling high quality geometry editing with more accurate updates to the illumination of the scene. | ['Yushi Guan', 'Nandita Vijaykumar', 'Chen Yang', 'Haoda Li', 'Jiahao Zhang', 'Ruofan Liang'] | 2022-10-15 | null | null | null | null | ['lighting-estimation'] | ['computer-vision'] | [ 5.04030764e-01 -4.02699262e-01 5.48421085e-01 -5.55068433e-01
-7.64216408e-02 -5.13127208e-01 4.03948516e-01 -2.94065565e-01
5.82608320e-02 6.40998721e-01 4.53579426e-02 3.90019491e-02
1.92030743e-01 -1.26255918e+00 -9.10935402e-01 -5.01345873e-01
4.14914906e-01 1.58848330e-01 1.88862979e-01 -1.81131989e-01
1.87317893e-01 1.29434419e+00 -1.88273001e+00 4.15146172e-01
1.03350270e+00 9.04231668e-01 3.74458343e-01 7.29812682e-01
-4.06195611e-01 6.71157241e-01 -5.50437629e-01 -1.26565710e-01
4.24130350e-01 -1.51417866e-01 -2.86404103e-01 2.02828288e-01
9.23488319e-01 -9.41518545e-01 -3.22171181e-01 7.32674122e-01
5.52711368e-01 3.69274527e-01 4.99234229e-01 -8.84294391e-01
-6.89758301e-01 -2.84366637e-01 -5.65144658e-01 -3.93363208e-01
4.65095282e-01 8.12786892e-02 3.24249655e-01 -1.12309885e+00
7.24343419e-01 1.37453222e+00 7.51688898e-01 5.55971324e-01
-1.40977728e+00 -5.20899475e-01 3.13607723e-01 -1.66868612e-01
-1.45648777e+00 -3.25727791e-01 1.15402877e+00 -2.30314538e-01
9.47341323e-01 6.57567084e-01 9.30546343e-01 8.01932812e-01
2.40961388e-01 2.80964047e-01 1.19436777e+00 -2.47185260e-01
2.57506937e-01 1.04774818e-01 -2.35925481e-01 6.31255090e-01
-6.40305728e-02 3.41850400e-01 -6.61704540e-01 -2.24877641e-01
1.55608141e+00 2.48166680e-01 -7.52622783e-01 -3.71391505e-01
-1.05675912e+00 4.04524177e-01 7.08375037e-01 -2.15161443e-01
-3.36247414e-01 6.32330775e-01 -7.59752020e-02 3.02741285e-02
8.28076124e-01 2.81684816e-01 -3.65309447e-01 1.57841489e-01
-7.56852269e-01 2.78137326e-01 4.46882069e-01 8.18265080e-01
1.17233682e+00 4.21376884e-01 -1.51688278e-01 8.75364721e-01
4.29555416e-01 9.44672227e-01 -2.76256710e-01 -1.55132878e+00
-2.21348200e-02 5.58386445e-01 3.65890563e-01 -1.16096961e+00
-1.52282178e-01 -3.10068309e-01 -7.99675405e-01 7.54786193e-01
5.24250902e-02 3.25125605e-01 -1.10059333e+00 1.29380059e+00
6.26338661e-01 7.23720729e-01 -1.58942521e-01 1.04947114e+00
9.19068694e-01 8.73448670e-01 -3.91779304e-01 -4.02224669e-03
9.39334691e-01 -5.83445251e-01 -7.75551915e-01 1.59865260e-01
3.41711819e-01 -7.65480638e-01 1.27175343e+00 5.17964065e-01
-1.21441734e+00 -4.15128529e-01 -8.13502431e-01 -5.96387625e-01
-8.23552087e-02 -1.26756892e-01 8.32430899e-01 4.18566734e-01
-1.16137350e+00 6.41647279e-01 -8.91650021e-01 1.57043636e-01
3.67513001e-01 3.93476516e-01 1.00709990e-01 -4.17493671e-01
-8.95347536e-01 5.76600254e-01 -2.88435668e-01 2.43911162e-01
-6.92715168e-01 -1.11281288e+00 -8.24424446e-01 -5.33469878e-02
4.29253094e-02 -1.08161294e+00 9.15593863e-01 -1.31269622e+00
-2.03759360e+00 6.87176645e-01 -2.34517321e-01 1.33838385e-01
2.95330524e-01 -3.19188714e-01 1.16535500e-01 -5.08218398e-03
-3.71877551e-01 5.45917213e-01 8.63644361e-01 -1.82443225e+00
-3.32665886e-03 -2.98035502e-01 3.67100924e-01 5.32697499e-01
3.47178765e-02 -2.44752765e-01 -4.25287306e-01 -6.57165945e-01
8.57458785e-02 -3.79806459e-01 -1.27982199e-01 8.56028974e-01
-1.39094755e-01 2.93063760e-01 1.07863915e+00 -7.12890387e-01
7.10822582e-01 -2.17403197e+00 1.22081392e-01 2.45429769e-01
1.73745438e-01 7.79412687e-02 -2.00275794e-01 1.14421092e-01
1.43114522e-01 -2.85131872e-01 -3.37481469e-01 -6.57532632e-01
-1.07539028e-01 6.32211268e-01 -5.11171520e-01 3.27467114e-01
1.30322233e-01 6.37324929e-01 -7.82324791e-01 -3.39138880e-02
6.45065367e-01 1.54322982e+00 -8.23408365e-01 3.35933805e-01
-5.71998775e-01 8.13577354e-01 -4.68733281e-01 4.62028027e-01
1.24792302e+00 -2.30232533e-02 -1.88123882e-01 -5.72108209e-01
-3.68180186e-01 1.19937606e-01 -1.30531120e+00 1.91338086e+00
-8.24577868e-01 4.32085693e-01 2.93646932e-01 -1.19025044e-01
9.38065648e-01 9.93920490e-02 3.62280667e-01 -8.83114755e-01
1.06912993e-01 -6.40752120e-03 -6.27104819e-01 7.40494579e-02
6.90333486e-01 -1.02288174e-02 6.88442111e-01 2.61113018e-01
-7.18895733e-01 -8.03181708e-01 -6.59997463e-01 5.76658845e-02
5.56985676e-01 8.04882586e-01 -3.54800582e-01 -5.40610217e-02
4.59389359e-01 -2.71390647e-01 5.45874000e-01 1.71646118e-01
6.79009914e-01 9.77109373e-01 -2.09292665e-01 -7.32198119e-01
-9.61314201e-01 -1.25138593e+00 -2.25242183e-01 7.24290848e-01
3.31581980e-01 -7.52219632e-02 -8.73157680e-01 -6.05899468e-02
1.07714720e-02 9.71066177e-01 -3.91909420e-01 1.08562335e-01
-8.81963551e-01 -4.42109108e-01 -8.30940157e-02 4.71150398e-01
5.60205460e-01 -6.94721878e-01 -7.38597393e-01 3.69848274e-02
-2.76710279e-02 -9.57761288e-01 -4.45448458e-01 -4.22785312e-01
-1.06462920e+00 -7.94087946e-01 -5.87397993e-01 -3.02476197e-01
1.01551259e+00 3.96629840e-01 1.28849351e+00 4.18588221e-01
-4.07131463e-01 5.53271472e-01 -6.71036318e-02 -2.73696423e-01
-3.95997435e-01 -5.86414993e-01 -2.10261345e-01 1.43921629e-01
-4.78013963e-01 -7.84168422e-01 -8.83825421e-01 4.89597619e-01
-1.16653812e+00 8.35767150e-01 -1.71728089e-01 3.94775629e-01
1.00690258e+00 -3.07609230e-01 -1.06017359e-01 -8.59225333e-01
2.99166068e-02 8.93031284e-02 -1.00799990e+00 8.10206607e-02
-2.56929308e-01 -1.17540419e-01 5.00723481e-01 -4.34182763e-01
-1.52711642e+00 4.73906174e-02 -2.79872149e-01 -9.04623747e-01
-4.52832691e-03 -1.13630993e-02 -1.81438714e-01 -3.20452392e-01
4.53863919e-01 3.83944660e-02 -3.51050645e-01 -7.00227141e-01
3.36807102e-01 2.43545130e-01 3.69990528e-01 -6.69257283e-01
8.22730482e-01 8.21335852e-01 2.50502318e-01 -9.83649015e-01
-7.36912072e-01 -3.90053429e-02 -4.71547902e-01 -3.85200560e-01
7.61694431e-01 -9.45879042e-01 -6.59184158e-01 7.33412445e-01
-1.50975430e+00 -8.90765011e-01 -5.75320899e-01 2.22322688e-01
-5.03285587e-01 2.97500014e-01 -4.24905539e-01 -7.65726686e-01
-3.38006467e-01 -1.04548621e+00 1.53133249e+00 1.40184894e-01
1.65360659e-01 -9.40665245e-01 1.02282297e-02 1.01225033e-01
6.01935744e-01 6.97189093e-01 7.47523129e-01 9.74214077e-01
-1.27955115e+00 3.66079092e-01 -4.18400705e-01 3.94498140e-01
3.65458131e-01 2.22197920e-01 -1.24703479e+00 -2.83106714e-01
5.93483597e-02 1.82950646e-01 4.07338262e-01 4.29722458e-01
1.57789731e+00 -1.16838790e-01 -1.08993351e-01 1.45844460e+00
1.78766906e+00 -4.15126383e-02 9.07798707e-01 -3.38847637e-02
1.26990616e+00 4.44056898e-01 4.81896609e-01 5.94610512e-01
4.49839592e-01 9.86572564e-01 7.26130605e-01 -5.02257824e-01
-7.01468229e-01 -8.36157054e-02 3.17314774e-01 7.67688632e-01
-5.34584343e-01 -3.49414855e-01 -5.50206065e-01 -6.05227053e-03
-1.39253771e+00 -6.28225625e-01 -5.60019672e-01 2.39571118e+00
7.89208531e-01 -4.97663081e-01 -5.59680462e-01 -1.95858628e-01
3.00987363e-01 4.34275530e-02 -6.87183201e-01 -6.29495919e-01
-2.36738473e-01 5.74244142e-01 4.12782937e-01 9.34518695e-01
-3.18722993e-01 8.82064521e-01 5.64011335e+00 3.96433622e-01
-1.37229705e+00 1.84487730e-01 5.17357349e-01 -1.88371807e-01
-1.25322187e+00 -2.01285612e-02 -6.03548646e-01 1.70179129e-01
4.88311827e-01 2.98999250e-01 8.66813898e-01 4.83269721e-01
5.08010507e-01 -2.71840811e-01 -9.04694498e-01 1.19467092e+00
6.12430312e-02 -1.36264360e+00 3.46034259e-01 3.77162322e-02
1.07272816e+00 -6.07720874e-02 -2.38909740e-02 -3.50319743e-01
1.72465026e-01 -8.87829185e-01 6.25286877e-01 1.11318910e+00
1.30278039e+00 -7.49191940e-01 2.19316915e-01 1.77327201e-01
-1.09427822e+00 4.28321630e-01 -3.99727583e-01 8.00182447e-02
4.11383450e-01 7.86812067e-01 -3.86079848e-01 5.62393427e-01
6.95780814e-01 7.09228992e-01 -2.26487577e-01 7.61710823e-01
-9.42503959e-02 8.67440701e-02 -4.72989172e-01 4.59340483e-01
-2.55938590e-01 -5.33186436e-01 5.48923850e-01 7.44766712e-01
5.03864229e-01 2.07246974e-01 1.58200741e-01 1.43862522e+00
-1.02172419e-01 3.06059178e-02 -5.34511685e-01 3.76948565e-01
1.09684452e-01 9.88079131e-01 -5.26604474e-01 -2.96003763e-02
-3.65977496e-01 1.40495670e+00 2.70604312e-01 7.59275734e-01
-8.41259837e-01 -1.27953827e-01 9.19216633e-01 5.97462535e-01
3.22478227e-02 -5.59528232e-01 -2.96563417e-01 -1.31649387e+00
-1.45155802e-01 -3.85562837e-01 -5.06410658e-01 -1.45847976e+00
-1.08018064e+00 4.26571339e-01 -9.31839049e-02 -9.27913129e-01
2.29668424e-01 -5.83924770e-01 -4.03403580e-01 1.14013600e+00
-1.92333663e+00 -1.26834583e+00 -7.27563500e-01 7.62958109e-01
4.57276285e-01 5.77769756e-01 9.30004537e-01 3.10699403e-01
-1.52429461e-01 9.91765633e-02 1.71801865e-01 -2.81113803e-01
5.60261428e-01 -9.86365497e-01 4.41100061e-01 6.03001118e-01
-2.02464432e-01 4.05577868e-01 3.78017128e-01 -6.90066278e-01
-1.80977607e+00 -1.18247807e+00 9.26404372e-02 -4.32190627e-01
-2.09402353e-01 -4.17883366e-01 -1.06689095e+00 5.44116795e-01
3.34114172e-02 2.11312681e-01 2.37793565e-01 -5.21881223e-01
-2.49559939e-01 -3.75312060e-01 -1.35387838e+00 5.89489996e-01
1.25741589e+00 -5.98739386e-01 5.50195463e-02 1.82284713e-01
8.91801357e-01 -8.25783312e-01 -8.29637945e-01 5.64062774e-01
5.49060166e-01 -1.13315213e+00 1.21103692e+00 2.81949669e-01
1.95532128e-01 -7.35661745e-01 -4.25229907e-01 -1.24740458e+00
-1.47649333e-01 -5.91119468e-01 -3.88602138e-01 1.05061686e+00
-2.56650716e-01 -8.10652018e-01 6.66968346e-01 6.53124034e-01
-5.23571730e-01 -6.81563079e-01 -6.18179083e-01 -2.27235690e-01
-2.25596771e-01 -5.83716273e-01 9.90106225e-01 1.05810177e+00
-1.16555667e+00 -2.88563430e-01 -1.57613233e-01 5.30669749e-01
8.15494120e-01 3.33597124e-01 9.27203476e-01 -1.26050818e+00
-3.06042016e-01 -3.92032489e-02 1.13674067e-02 -1.32827711e+00
6.80784732e-02 -6.60770595e-01 -2.50970013e-03 -1.88241100e+00
-1.37367651e-01 -8.76474202e-01 1.89907238e-01 9.06329602e-02
8.09411779e-02 5.22336602e-01 -4.65153754e-02 -3.81982774e-02
1.18453778e-01 6.89399958e-01 1.74947321e+00 2.52179533e-01
-3.74962419e-01 -1.58635974e-01 -2.59010375e-01 7.83961356e-01
5.54215193e-01 -1.95174918e-01 -5.93185186e-01 -1.22228479e+00
6.77614927e-01 -1.86461240e-01 6.36569619e-01 -8.18603754e-01
-2.33952656e-01 -2.58161366e-01 6.33086920e-01 -8.21935594e-01
8.21720600e-01 -1.05261004e+00 7.92115927e-01 2.67174859e-02
1.37257934e-01 -6.47717938e-02 3.86486411e-01 4.41180795e-01
1.52131751e-01 1.53665766e-01 8.75580311e-01 -2.23870844e-01
-3.77398044e-01 6.87158406e-01 2.75801378e-03 -2.25699797e-01
6.61723077e-01 -6.68162823e-01 -6.44486444e-03 -2.68801540e-01
-3.56818259e-01 -1.55224636e-01 1.20400965e+00 1.15000352e-01
1.18321192e+00 -1.32165098e+00 -4.59383130e-01 5.74488819e-01
-1.81438863e-01 7.49691784e-01 4.85496193e-01 4.10938323e-01
-1.24191165e+00 -3.45889986e-01 7.16378763e-02 -7.25521982e-01
-1.32846403e+00 1.09694019e-01 7.11130440e-01 2.33335510e-01
-8.60084236e-01 7.61468828e-01 7.61028647e-01 -6.55054331e-01
7.21922219e-02 -6.63181126e-01 2.53603399e-01 -5.86047113e-01
5.27034342e-01 4.45102066e-01 9.98885185e-02 -6.04625046e-01
1.56274885e-02 1.17071319e+00 4.26374525e-01 1.37504727e-01
1.46857417e+00 -2.46688873e-01 -3.62261474e-01 4.37977076e-01
9.61764991e-01 3.47751051e-01 -1.70207131e+00 -1.30328104e-01
-9.38590586e-01 -1.02968884e+00 5.96974671e-01 -8.39485288e-01
-1.19022799e+00 1.08611941e+00 5.49683332e-01 -3.64999861e-01
1.34601080e+00 -4.69280988e-01 1.04066384e+00 3.65767814e-02
7.65417576e-01 -8.13430488e-01 -1.60078079e-01 6.05375230e-01
1.15889776e+00 -6.30132854e-01 2.40185961e-01 -9.42190289e-01
-1.37602285e-01 1.20625293e+00 5.71508646e-01 -1.17395326e-01
6.24421895e-01 5.62883079e-01 -1.91053152e-02 -2.56188244e-01
-3.40222627e-01 2.78511018e-01 4.61959124e-01 6.94691718e-01
4.42907840e-01 -1.21516474e-01 3.50431710e-01 -3.58554125e-01
5.81511781e-02 6.55044466e-02 4.25390244e-01 6.40858889e-01
-7.85442442e-02 -1.11109638e+00 -4.91522223e-01 1.69028893e-01
-5.07627539e-02 -1.89199463e-01 -1.32146701e-01 2.78421968e-01
2.62520015e-01 3.57788295e-01 4.57383037e-01 1.58266559e-01
5.85226536e-01 -3.96007597e-01 9.25203562e-01 -7.98448980e-01
-6.19500697e-01 1.68237731e-01 9.00889188e-03 -1.02619851e+00
-5.84802032e-01 -3.37450951e-01 -1.44356298e+00 -4.25950676e-01
-3.91357392e-01 -4.39175397e-01 9.77458417e-01 4.07451034e-01
5.92639863e-01 6.52138352e-01 8.04146290e-01 -1.46020281e+00
1.91865131e-01 -2.76599765e-01 -5.94027281e-01 3.20942372e-01
3.32097828e-01 -7.62960434e-01 -3.24248761e-01 9.52770561e-02] | [9.634037971496582, -3.1195993423461914] |
99aadd74-41a7-4979-9068-0040ebbea69d | tell-me-why-explanations-support-learning-of | 2112.03753 | null | https://arxiv.org/abs/2112.03753v3 | https://arxiv.org/pdf/2112.03753v3.pdf | Tell me why! Explanations support learning relational and causal structure | Inferring the abstract relational and causal structure of the world is a major challenge for reinforcement-learning (RL) agents. For humans, language--particularly in the form of explanations--plays a considerable role in overcoming this challenge. Here, we show that language can play a similar role for deep RL agents in complex environments. While agents typically struggle to acquire relational and causal knowledge, augmenting their experience by training them to predict language descriptions and explanations can overcome these limitations. We show that language can help agents learn challenging relational tasks, and examine which aspects of language contribute to its benefits. We then show that explanations can help agents to infer not only relational but also causal structure. Language can shape the way that agents to generalize out-of-distribution from ambiguous, causally-confounded training, and explanations even allow agents to learn to perform experimental interventions to identify causal relationships. Our results suggest that language description and explanation may be powerful tools for improving agent learning and generalization. | ['Felix Hill', 'Jane X. Wang', 'Neil C. Rabinowitz', 'Adam Santoro', 'Chen Yan', 'James L. McClelland', 'Allison C. Tam', 'Stephanie C. Y. Chan', 'Ishita Dasgupta', 'Nicholas A. Roy', 'Andrew K. Lampinen'] | 2021-12-07 | null | null | null | null | ['odd-one-out'] | ['reasoning'] | [ 9.07360285e-04 3.86036843e-01 -4.23862308e-01 -2.83608526e-01
-2.99830973e-01 -5.99889159e-01 8.42435420e-01 4.15194988e-01
-3.85325909e-01 1.01782739e+00 5.82453370e-01 -4.69389856e-01
-2.95480579e-01 -8.42896461e-01 -9.15789545e-01 -3.22761148e-01
-3.73828590e-01 6.38979912e-01 -1.24913335e-01 -4.77321833e-01
3.93976867e-01 6.53382897e-01 -1.27049863e+00 1.52768448e-01
7.83644378e-01 -1.07334644e-01 4.61238652e-01 4.94530916e-01
-1.69081166e-01 1.49576414e+00 -4.71410871e-01 8.49875063e-02
-3.67222019e-02 -8.53251159e-01 -9.39582586e-01 -2.31144637e-01
-2.35451519e-01 -8.09449732e-01 -4.18582261e-01 6.09148622e-01
2.59016883e-02 2.44831085e-01 6.73751354e-01 -1.37293756e+00
-7.66803443e-01 1.05721176e+00 -1.98237047e-01 2.53138840e-01
6.74558222e-01 5.05101800e-01 1.02660263e+00 -2.01869115e-01
6.23915911e-01 1.68100882e+00 3.03646505e-01 9.95043933e-01
-1.58712578e+00 -7.14893341e-01 6.01961017e-01 2.33846009e-01
-5.97060382e-01 -2.39937723e-01 5.61572969e-01 -6.31087065e-01
1.15246642e+00 -2.21677214e-01 9.47971404e-01 1.28964055e+00
1.48566291e-01 7.30430305e-01 1.32528281e+00 -2.91179240e-01
1.21935472e-01 -1.87568411e-01 8.74620397e-03 9.62975085e-01
2.91961521e-01 1.06610835e+00 -1.03049684e+00 -8.33217353e-02
1.15521288e+00 -1.43557549e-01 -2.69015998e-01 -2.77261078e-01
-1.34309268e+00 1.05247986e+00 7.83379376e-01 1.35837033e-01
-4.59861994e-01 7.82283068e-01 3.42469573e-01 4.85689044e-01
-1.91134825e-01 1.31765687e+00 -4.57536340e-01 -1.80111036e-01
9.27520320e-02 5.38844645e-01 6.08457625e-01 4.99644250e-01
7.82816350e-01 3.12221527e-01 6.08985126e-02 5.83203435e-01
4.69249815e-01 4.92809772e-01 2.87982851e-01 -1.22922957e+00
1.26191184e-01 5.76916456e-01 1.25695050e-01 -6.20893061e-01
-6.77393019e-01 -2.02342331e-01 -1.33422762e-01 5.45889318e-01
6.42807782e-01 -3.66646290e-01 -5.98872185e-01 2.42582607e+00
1.10298924e-01 -3.19373012e-02 4.01685119e-01 9.95783746e-01
5.56816638e-01 5.10678351e-01 4.62932348e-01 -2.04491824e-01
1.05938601e+00 -5.89523792e-01 -5.31252801e-01 -7.33596027e-01
8.07100892e-01 -2.54258305e-01 1.26726592e+00 1.22753493e-01
-9.73205388e-01 -2.63380438e-01 -9.29128230e-01 -5.99279702e-02
-1.99478105e-01 -4.39207673e-01 1.29124665e+00 -3.39696780e-02
-9.40840244e-01 7.66873121e-01 -7.50349104e-01 -5.55102229e-01
5.57739437e-01 4.21912342e-01 -4.36743587e-01 -2.65202913e-02
-1.17743266e+00 1.27087319e+00 3.96795928e-01 -2.16716111e-01
-1.41210055e+00 -6.22074187e-01 -1.14408731e+00 2.90031880e-02
5.40060341e-01 -1.02065170e+00 1.56241906e+00 -8.48939717e-01
-1.49195445e+00 4.53009009e-01 -1.81629539e-01 -5.31737268e-01
-1.30949551e-02 -2.71134853e-01 3.33318114e-01 1.13017492e-01
3.22200358e-01 1.00806487e+00 3.90338510e-01 -1.51991129e+00
-3.75146449e-01 -3.86316091e-01 5.49428344e-01 4.16679859e-01
2.71400690e-01 -9.83875990e-02 3.96054476e-01 -3.84344369e-01
-2.16151848e-02 -8.72814953e-01 -5.23568153e-01 -1.94892734e-01
-9.73774642e-02 -5.12196004e-01 2.99590021e-01 -3.12251389e-01
2.92362452e-01 -1.84801280e+00 4.46393341e-01 -1.27281249e-01
3.98292840e-01 -1.34025365e-01 -3.86605233e-01 5.67774534e-01
-3.73998061e-02 3.82100761e-01 1.03977516e-01 -1.61357924e-01
3.14698070e-01 7.56320894e-01 -6.42074168e-01 1.71088859e-01
4.23537493e-01 1.12999785e+00 -1.17866707e+00 -8.54903758e-02
1.92150220e-01 3.94024193e-01 -6.43118382e-01 6.70830846e-01
-8.76800239e-01 1.03911042e+00 -6.53849542e-01 -3.85990478e-02
-7.45196044e-02 -4.67578135e-02 3.60185087e-01 5.91047764e-01
-2.76710303e-03 8.22446883e-01 -6.28972888e-01 1.38386071e+00
-6.71189308e-01 8.24508727e-01 -2.01458037e-01 -1.00787091e+00
6.11401737e-01 2.65997797e-01 -5.43692373e-02 -8.79525006e-01
8.67513418e-02 4.91871722e-02 5.91081977e-01 -7.53890514e-01
-5.40420786e-03 -5.90837717e-01 5.46572320e-02 8.67988884e-01
-1.71142891e-01 -5.87937117e-01 2.52605736e-01 5.06478131e-01
9.55924213e-01 4.10662085e-01 2.75021762e-01 -4.02882993e-02
7.10737333e-02 4.51315939e-01 4.83630508e-01 8.83248448e-01
1.21913090e-01 -1.92298383e-01 1.00153923e+00 -6.10065639e-01
-9.26501453e-01 -1.10677457e+00 3.62410247e-01 1.29947340e+00
4.24850546e-03 -2.06990644e-01 -4.09976035e-01 -5.56344628e-01
1.36045113e-01 1.36706793e+00 -7.20090926e-01 -4.98735070e-01
-6.54724300e-01 -3.37603599e-01 3.75913411e-01 8.30497801e-01
1.82214111e-01 -1.53776443e+00 -9.32090700e-01 6.14101067e-02
7.46341720e-02 -9.14401531e-01 -8.56426209e-02 5.77133477e-01
-8.60738158e-01 -1.15189493e+00 -4.86540198e-02 -5.76039970e-01
8.34326982e-01 2.67944604e-01 1.14758122e+00 5.45365870e-01
8.77456740e-02 7.19413757e-01 -1.06252789e-01 -4.61866111e-01
-7.90613830e-01 -3.78512949e-01 2.30299443e-01 -8.85908604e-01
-3.53311026e-03 -8.26805294e-01 -1.03060082e-01 1.39399841e-01
-7.07873046e-01 2.04614848e-01 3.79312485e-01 8.98956776e-01
2.03098133e-02 -2.47304626e-02 7.74921954e-01 -9.28487837e-01
1.09029734e+00 -4.37697172e-01 -8.45366180e-01 -5.63274100e-02
-5.31760633e-01 6.67647600e-01 7.75825202e-01 -5.39573431e-01
-1.12985170e+00 -7.06128851e-02 2.75521815e-01 1.78057492e-01
-3.76199752e-01 6.64273322e-01 6.73355982e-02 1.80418536e-01
8.16262007e-01 8.56477618e-02 2.16302544e-01 -1.28046587e-01
6.74442053e-01 -2.29520559e-01 2.03135550e-01 -1.11182725e+00
9.97476220e-01 1.27136692e-01 1.16096281e-01 -4.49788809e-01
-7.83411920e-01 1.16406910e-01 -4.23623532e-01 -6.99694315e-03
1.03532922e+00 -9.69653845e-01 -1.21497786e+00 -1.69344515e-01
-1.13921475e+00 -1.13842762e+00 -2.95088023e-01 8.25812697e-01
-8.47776294e-01 -3.33920151e-01 -6.01922095e-01 -7.36908257e-01
5.76234639e-01 -1.35862446e+00 5.52850246e-01 3.62834573e-01
-6.50497913e-01 -1.13296151e+00 7.37005174e-02 2.52631336e-01
3.08268189e-01 -1.03210760e-02 1.44199312e+00 -7.78252244e-01
-9.04758453e-01 5.21811843e-01 -1.13078572e-01 -3.36071283e-01
5.05499728e-02 -3.20669532e-01 -7.02445984e-01 1.83219060e-01
-2.16980278e-01 -9.61028576e-01 7.49867797e-01 3.15555871e-01
6.89896405e-01 -4.21314716e-01 -4.13996130e-01 1.06980577e-01
8.90635669e-01 3.68259341e-01 2.65501529e-01 2.30227724e-01
4.37045634e-01 9.11493421e-01 3.66562307e-01 1.72953427e-01
8.16968799e-01 4.94665265e-01 4.52294171e-01 5.50793856e-02
-1.47125632e-01 -6.96554542e-01 5.82967699e-01 1.49563551e-01
6.97991997e-02 1.06364161e-01 -9.04239953e-01 4.27495539e-01
-1.92476571e+00 -1.20802188e+00 -1.30186286e-02 1.79554272e+00
1.34992671e+00 3.35479304e-02 1.70560881e-01 -2.28351355e-01
1.37693092e-01 -7.77152777e-02 -8.04023027e-01 -5.72722316e-01
2.17025355e-02 -6.22116029e-02 9.61545855e-02 1.15266383e+00
-3.89221579e-01 1.29281068e+00 7.30981684e+00 3.56318280e-02
-8.13331068e-01 -4.12827194e-01 4.01080489e-01 2.63603806e-01
-7.24666595e-01 1.28526539e-01 -5.10843337e-01 -3.64158988e-01
7.43116260e-01 -5.73157668e-02 1.13459516e+00 5.38739502e-01
4.12926644e-01 -3.62184018e-01 -1.84036493e+00 4.82978523e-01
-2.13279679e-01 -1.32983792e+00 6.02999032e-02 1.80658594e-01
5.57977617e-01 -1.74261943e-01 1.24665409e-01 5.61467052e-01
1.32867634e+00 -1.64313269e+00 6.41381741e-01 2.14466736e-01
2.39856586e-01 -4.81154531e-01 4.99608904e-01 6.93728209e-01
-7.64708519e-01 -2.28575125e-01 -3.73620719e-01 -7.77468801e-01
-1.73798233e-01 -1.58247724e-01 -1.23800600e+00 -1.55783549e-01
1.38434991e-01 6.06721520e-01 -3.96318555e-01 4.11545813e-01
-9.75051284e-01 4.00880605e-01 -1.26676872e-01 -3.65665585e-01
9.52264816e-02 -7.35819489e-02 3.70819122e-01 6.63088977e-01
-9.97407883e-02 7.21690655e-01 3.35955918e-01 1.42392766e+00
1.99758261e-01 -3.67214590e-01 -1.05698633e+00 -4.63403344e-01
4.80386138e-01 5.55949152e-01 -6.00794494e-01 -2.18938470e-01
-1.57463834e-01 6.11222029e-01 7.77378321e-01 6.25982225e-01
-5.65299213e-01 3.36452276e-01 8.33421826e-01 -2.26576403e-01
-6.95267171e-02 -6.24285996e-01 -3.98802161e-01 -9.45570648e-01
-5.34592748e-01 -1.24581218e+00 1.41648948e-01 -1.10178256e+00
-1.02945316e+00 3.25964510e-01 2.32549682e-01 -3.64750922e-01
-7.43385315e-01 -5.64406514e-01 -4.09911305e-01 8.04055810e-01
-1.43338728e+00 -9.73940313e-01 5.38700745e-02 5.13543606e-01
6.60465777e-01 -2.38842636e-01 8.34580660e-01 -6.70648396e-01
-2.91880250e-01 -1.01381123e-01 -5.25310218e-01 1.05397902e-01
3.27026606e-01 -1.48166704e+00 2.07534611e-01 4.08691138e-01
4.68708068e-01 1.08141410e+00 1.00121713e+00 -7.06197083e-01
-1.52568293e+00 -5.47161102e-01 6.12551808e-01 -6.66945577e-01
6.77911401e-01 -2.80714035e-01 -7.81360865e-01 1.12189746e+00
4.25996184e-01 -3.93097669e-01 5.03661573e-01 5.48583269e-01
-8.17842007e-01 1.91648677e-01 -9.03914809e-01 1.22598565e+00
1.24613857e+00 -8.21458220e-01 -1.02020586e+00 3.47404003e-01
1.08437228e+00 -2.82148600e-01 -4.07320917e-01 3.34406346e-02
2.89876878e-01 -9.18044567e-01 9.57758248e-01 -1.17651188e+00
7.51308262e-01 -4.48918492e-01 -7.49335662e-02 -1.91724229e+00
-1.76952407e-01 -6.13571346e-01 3.93551975e-01 9.26718235e-01
6.84744298e-01 -7.86553204e-01 4.19184119e-01 6.52494311e-01
7.42446482e-02 -1.22737922e-01 -4.25252765e-01 -5.00884950e-01
4.15931314e-01 -5.46278775e-01 4.12403882e-01 9.29819047e-01
4.55272079e-01 7.42097259e-01 2.83631179e-02 3.47697824e-01
5.31473219e-01 1.12908185e-01 9.55648124e-01 -1.14263165e+00
-3.30169469e-01 -5.23963094e-01 1.55028209e-01 -8.93922985e-01
8.03185344e-01 -1.00558627e+00 1.61212653e-01 -1.76004028e+00
2.96360403e-01 -5.45429587e-01 3.49062562e-01 6.86536729e-01
-1.15532398e-01 -4.46444064e-01 4.49906677e-01 -2.32001916e-02
-4.79145050e-01 6.14816189e-01 1.53221333e+00 1.21430278e-01
-2.51847357e-01 -2.77673125e-01 -9.27551508e-01 9.76655066e-01
1.01045144e+00 -5.94188869e-01 -8.77554059e-01 -5.98996520e-01
6.19337380e-01 3.69601488e-01 7.05040932e-01 -5.13107896e-01
1.15255982e-01 -8.56403947e-01 3.79717350e-01 6.72394484e-02
1.71655536e-01 -7.45044529e-01 -1.42202228e-01 8.33751440e-01
-9.90379632e-01 2.68388957e-01 5.65838695e-01 5.71712017e-01
1.92113802e-01 -1.36838809e-01 4.34576631e-01 -5.04167497e-01
-5.52282810e-01 -3.26093525e-01 -9.34115887e-01 2.87637204e-01
7.32547700e-01 2.21741274e-01 -6.76192045e-01 -7.87939191e-01
-6.92707002e-01 5.31225026e-01 3.68024677e-01 2.00618371e-01
7.18361497e-01 -1.19362605e+00 -6.26030684e-01 -2.29576640e-02
-1.20728806e-01 -1.43428221e-01 -1.69702083e-01 4.83210891e-01
-2.60383248e-01 4.92044657e-01 -2.98065484e-01 -3.79963517e-01
-9.37146425e-01 6.41037524e-01 5.93423665e-01 -5.91369830e-02
-4.67213690e-01 1.10921180e+00 7.49137700e-01 -6.30344808e-01
3.41022909e-01 -4.87978578e-01 -4.20252860e-01 -1.68399543e-01
5.20349026e-01 -2.97199160e-01 -8.21104169e-01 -2.48823434e-01
-2.29411140e-01 4.57429528e-01 1.14189871e-02 -5.35845220e-01
1.48906207e+00 -3.35032418e-02 -1.68445975e-01 5.53884208e-01
5.15549719e-01 1.02659710e-01 -1.60799766e+00 -1.99619845e-01
4.72179353e-02 -2.93031335e-01 -2.02555671e-01 -1.17197680e+00
-5.90380430e-01 1.04334676e+00 -8.08274820e-02 2.78046668e-01
5.49665630e-01 5.64776242e-01 2.40517855e-01 7.09859550e-01
4.12449419e-01 -5.58572173e-01 7.56840289e-01 6.25768483e-01
1.25697696e+00 -1.34086394e+00 9.80653986e-02 -1.31841645e-01
-7.73815632e-01 1.19263065e+00 8.30199122e-01 -1.05103724e-01
1.04278624e-01 2.83814400e-01 2.00883690e-02 -4.25974190e-01
-1.28778875e+00 -5.26078463e-01 -1.05347611e-01 1.10208225e+00
5.26340723e-01 1.20816156e-01 1.27825886e-01 3.00087571e-01
-4.71117824e-01 -2.03145877e-01 8.48110914e-01 5.12075603e-01
-4.23008680e-01 -1.27992189e+00 -5.79832256e-01 -1.88730322e-02
-6.93268180e-02 -1.79031953e-01 -7.09204674e-01 9.89473820e-01
2.14259047e-02 1.10739172e+00 -1.13701418e-01 -2.44878188e-01
1.01238228e-01 4.44180481e-02 8.28415751e-01 -8.47353816e-01
-4.43511486e-01 -1.71336815e-01 2.51059502e-01 -7.41150856e-01
-3.65377545e-01 -8.90919566e-01 -1.96605361e+00 -4.06181186e-01
1.55196548e-01 1.56434625e-01 3.01431030e-01 1.22101736e+00
-7.54000768e-02 7.46871889e-01 1.82719126e-01 -5.12329161e-01
-5.58359444e-01 -6.53581917e-01 -1.53679714e-01 3.05960774e-01
7.93451130e-01 -9.08159912e-01 -5.05947292e-01 1.23577556e-02] | [4.257334232330322, 1.2208791971206665] |
5bc372e1-a178-483a-bde8-8da8357c9b6e | ocatari-object-centric-atari-2600 | 2306.08649 | null | https://arxiv.org/abs/2306.08649v1 | https://arxiv.org/pdf/2306.08649v1.pdf | OCAtari: Object-Centric Atari 2600 Reinforcement Learning Environments | Cognitive science and psychology suggest that object-centric representations of complex scenes are a promising step towards enabling efficient abstract reasoning from low-level perceptual features. Yet, most deep reinforcement learning approaches rely on only pixel-based representations that do not capture the compositional properties of natural scenes. For this, we need environments and datasets that allow us to work and evaluate object-centric approaches. We present OCAtari, a set of environment that provides object-centric state representations of Atari games, the most-used evaluation framework for deep RL approaches. OCAtari also allows for RAM state manipulations of the games to change and create specific or even novel situations. The code base for this work is available at github.com/k4ntz/OC_Atari. | ['Kristian Kersting', 'Sebastian Sztwiertnia', 'Bjarne Gregori', 'Jannis Blüml', 'Quentin Delfosse'] | 2023-06-14 | null | null | null | null | ['atari-games'] | ['playing-games'] | [-1.96670741e-01 -1.42768979e-01 -3.24713066e-02 -1.90516278e-01
-2.56178796e-01 -6.02005005e-01 8.51333916e-01 -1.53518561e-02
-5.35153925e-01 6.13076627e-01 2.17306033e-01 -2.10908115e-01
-7.43397176e-02 -8.73829603e-01 -8.09260428e-01 -4.34873819e-01
-1.11154400e-01 4.62613553e-01 3.52575570e-01 -6.74302101e-01
3.06992769e-01 6.14559352e-01 -2.00059533e+00 5.05099714e-01
3.04790199e-01 7.95538783e-01 4.26066995e-01 9.76815403e-01
-2.06553619e-02 1.22188079e+00 -5.31364560e-01 -1.31812185e-01
3.25157702e-01 -4.28712726e-01 -9.13209260e-01 -2.35048071e-01
3.34959060e-01 -6.31569266e-01 -6.53385878e-01 8.80352855e-01
4.17663097e-01 5.36904395e-01 3.41843456e-01 -1.34209323e+00
-5.75921178e-01 7.07648337e-01 -2.30289578e-01 5.51266193e-01
4.58080083e-01 8.26039135e-01 8.96540225e-01 -5.70653498e-01
7.65405476e-01 1.60953665e+00 2.51864269e-02 7.17231631e-01
-1.33301198e+00 -4.84518468e-01 3.31394047e-01 6.30677223e-01
-1.08950186e+00 -4.18087035e-01 7.36488879e-01 2.34792177e-02
1.44947076e+00 2.67727315e-01 1.15798616e+00 1.31209505e+00
1.87049583e-01 1.11499047e+00 1.36906922e+00 -2.97087044e-01
5.39096951e-01 -3.72784317e-01 -1.77131146e-02 7.98336029e-01
2.71142781e-01 5.15493929e-01 -6.05398476e-01 9.44494903e-02
1.25525570e+00 -5.28941676e-02 1.33020282e-01 -6.89869165e-01
-1.34607220e+00 7.09390581e-01 6.36582494e-01 2.77592033e-01
-5.24331748e-01 9.68923211e-01 4.82732415e-01 2.88803667e-01
-2.75274336e-01 8.14117253e-01 -3.65270257e-01 -5.39100051e-01
-3.96354318e-01 7.67373800e-01 4.12266225e-01 8.78800094e-01
8.40471745e-01 4.81144249e-01 -3.15510064e-01 4.55579370e-01
1.05240107e-01 4.29983944e-01 5.48740923e-01 -1.76482642e+00
2.15518344e-02 4.57799852e-01 2.31409878e-01 -6.80328310e-01
-5.23620129e-01 -2.76185721e-01 -2.42693871e-01 8.07255566e-01
2.92690605e-01 -2.89733242e-02 -9.77751732e-01 1.86011064e+00
2.16934010e-01 1.57597169e-01 2.73446530e-01 1.04377460e+00
9.48029339e-01 6.52550876e-01 3.45206261e-01 2.51120239e-01
1.30829215e+00 -1.03085053e+00 -4.71413523e-01 -4.01348293e-01
5.77465177e-01 -1.76892713e-01 1.48995996e+00 4.94269520e-01
-1.56002653e+00 -4.87925977e-01 -1.06391978e+00 -2.24562824e-01
-5.92681289e-01 -3.00551802e-01 1.05102813e+00 3.93720508e-01
-1.36483538e+00 4.58180279e-01 -9.45151925e-01 -2.99556434e-01
5.00425398e-01 2.75830537e-01 -2.50021935e-01 -1.80006370e-01
-1.05915320e+00 1.22848427e+00 5.82488716e-01 5.51694371e-02
-1.49258149e+00 -3.69759142e-01 -1.09038973e+00 1.53263122e-01
6.93582654e-01 -6.67041123e-01 1.71453762e+00 -1.01274228e+00
-1.78295588e+00 7.88676977e-01 1.45801574e-01 -5.16742945e-01
2.88339853e-01 -2.28756160e-01 -4.23497297e-02 3.57726574e-01
-2.08763838e-01 9.86498296e-01 5.70754051e-01 -1.31061709e+00
-3.25096756e-01 -2.39205986e-01 7.81308472e-01 5.22023201e-01
2.36087829e-01 -4.54897620e-02 -2.19292030e-01 -3.61516535e-01
-1.84141174e-01 -8.51589918e-01 -4.93549526e-01 5.71952425e-02
-7.03451922e-03 -1.19768925e-01 5.65780878e-01 -1.77307546e-01
4.60740864e-01 -2.04211760e+00 2.56356448e-01 -2.39104941e-01
1.66908711e-01 3.18842560e-01 -6.35290682e-01 6.15892470e-01
-6.30466789e-02 2.16654390e-02 2.02894270e-01 -9.97051895e-02
3.17172498e-01 4.15076643e-01 -2.30099663e-01 1.99876606e-01
2.14067310e-01 1.32473588e+00 -1.11727834e+00 -3.98901492e-01
7.37744510e-01 2.37870246e-01 -8.53774846e-01 9.30900946e-02
-7.18601584e-01 3.19075972e-01 -4.40064400e-01 3.01630527e-01
3.91841918e-01 6.36593774e-02 2.97094107e-01 2.75920946e-02
-9.95552018e-02 4.99779224e-01 -1.29144526e+00 2.11822200e+00
-4.67020035e-01 4.86578017e-01 -1.50672570e-01 -7.82312334e-01
5.37673175e-01 5.38545661e-02 3.25389445e-01 -1.07071686e+00
3.30723703e-01 -2.59235024e-01 1.90562069e-01 -2.82356083e-01
7.35572338e-01 9.24962014e-02 -2.07306534e-01 5.44935763e-01
2.04937190e-01 -5.87812245e-01 4.89843994e-01 2.56170452e-01
1.22453630e+00 7.01896012e-01 3.79754961e-01 -1.73873752e-01
1.57038942e-01 2.91624010e-01 5.68062663e-01 1.18222857e+00
-4.09610182e-01 3.49771857e-01 4.97516930e-01 -7.72807956e-01
-9.59208369e-01 -1.36379325e+00 3.48488778e-01 1.30697453e+00
3.21297497e-01 -4.53016609e-01 -5.87096095e-01 -2.11041361e-01
-2.66405463e-01 1.09293330e+00 -6.33985698e-01 -4.83545542e-01
-5.73521018e-01 -3.10179025e-01 5.57912171e-01 6.95336878e-01
6.51892304e-01 -1.77999628e+00 -1.39533603e+00 1.73635647e-01
-4.22406346e-02 -9.08266723e-01 2.67795324e-01 2.12737456e-01
-9.19378996e-01 -1.00564599e+00 -3.29447955e-01 -4.01031017e-01
2.89051086e-01 2.79822052e-01 1.35165012e+00 2.97464728e-01
-6.92154408e-01 6.64074481e-01 -4.72561806e-01 -3.94353986e-01
-4.37500298e-01 -2.67309666e-01 -5.18991128e-02 -7.28610039e-01
6.06266782e-02 -6.09778762e-01 -7.56336510e-01 1.31445564e-03
-1.09661055e+00 3.72816920e-01 5.20573139e-01 4.75578934e-01
4.19999897e-01 -5.73776290e-02 1.23666570e-01 -6.57614231e-01
6.79547966e-01 -1.13497488e-01 -7.61273146e-01 5.35398759e-02
-8.11355561e-02 2.31364608e-01 4.40596849e-01 -4.71772403e-01
-1.13455129e+00 -9.24948826e-02 -6.12938069e-02 -3.79330605e-01
-5.21932542e-01 4.62724119e-01 1.34201879e-02 3.78563970e-01
9.21233356e-01 3.61243933e-01 -2.77896374e-01 1.43643878e-02
7.88687468e-01 3.29760350e-02 4.43312496e-01 -1.08305335e+00
5.35380721e-01 4.81981248e-01 -1.54392749e-01 -5.64582288e-01
-4.60374087e-01 1.74533892e-02 -4.43339139e-01 -2.73677319e-01
7.93559194e-01 -8.03020358e-01 -1.17854953e+00 4.41827744e-01
-9.63551462e-01 -1.21450746e+00 -7.80438244e-01 2.67977625e-01
-1.28428102e+00 -9.20646638e-03 -7.84963846e-01 -8.07410657e-01
-4.84448113e-02 -1.31642592e+00 9.90283012e-01 3.58550876e-01
-2.61361808e-01 -5.44591010e-01 2.58456826e-01 2.55627126e-01
3.90564263e-01 1.71250895e-01 1.01473200e+00 -1.73139378e-01
-7.21341670e-01 1.12744600e-01 2.07697488e-02 -4.08908315e-02
-1.75693527e-01 7.63973594e-02 -1.03845048e+00 -1.24673545e-01
-3.14573050e-01 -9.30995941e-01 7.30895638e-01 3.83669138e-01
1.44615543e+00 -1.24307811e-01 1.23751163e-01 4.54503477e-01
1.45466781e+00 2.53499389e-01 1.07052684e+00 6.38352811e-01
3.10004711e-01 3.16980064e-01 6.55079365e-01 6.86200917e-01
5.79089105e-01 6.77242041e-01 8.50646317e-01 -1.50088547e-03
-1.44597545e-01 -2.04825327e-01 4.80000108e-01 9.43118855e-02
-3.31379712e-01 -6.33093789e-02 -8.86704326e-01 4.61025208e-01
-1.87440729e+00 -1.37355912e+00 2.34146461e-01 1.72585928e+00
8.01567018e-01 2.29272351e-01 2.23306939e-01 4.10674997e-02
2.67549008e-01 3.65458816e-01 -7.59556472e-01 -8.56831670e-01
1.69522032e-01 5.17340243e-01 1.03170380e-01 3.35327685e-01
-7.58295953e-01 1.67514205e+00 6.80832338e+00 4.90842432e-01
-1.02610314e+00 -5.23615181e-02 2.20957145e-01 -3.36478442e-01
-1.86514601e-01 3.29432525e-02 -2.10652858e-01 -1.38686687e-01
1.01141810e+00 -9.01826769e-02 8.97253454e-01 9.20190752e-01
4.82481003e-01 -4.85602140e-01 -1.12228525e+00 9.10799503e-01
-3.30420345e-01 -1.50760257e+00 6.47584796e-02 -1.29135028e-02
4.43070620e-01 1.00725666e-01 3.59569609e-01 7.03362226e-01
1.11215723e+00 -1.16600502e+00 9.66901600e-01 4.06289577e-01
5.02528727e-01 -6.73749208e-01 2.30120122e-01 1.85831159e-01
-8.40599120e-01 -1.83368638e-01 -4.93709534e-01 -4.56054002e-01
-3.04026514e-01 -1.37256175e-01 -6.05311632e-01 1.84984654e-01
9.68112409e-01 5.47503948e-01 -6.30164027e-01 7.93073058e-01
-3.84731740e-01 3.44385386e-01 -5.17372508e-03 -1.32143304e-01
3.25893044e-01 9.13991407e-02 2.89804608e-01 9.02913630e-01
-9.72817242e-02 2.17012227e-01 1.41827732e-01 1.09872282e+00
1.51975751e-01 -2.58467048e-01 -8.10645044e-01 -9.49934199e-02
4.76608455e-01 1.27060044e+00 -8.39691520e-01 -3.98173988e-01
-2.89113641e-01 9.73720312e-01 6.37140572e-01 4.44845766e-01
-1.03566599e+00 -5.50576746e-02 1.13644636e+00 -7.41073415e-02
4.30237204e-01 -7.05797255e-01 3.01709529e-02 -1.30344820e+00
-4.01000291e-01 -1.37472856e+00 2.89271235e-01 -1.39681971e+00
-5.94188750e-01 4.02175486e-01 2.55154073e-01 -6.84819520e-01
-4.18476522e-01 -7.98222065e-01 -6.87116742e-01 3.29723746e-01
-1.39096701e+00 -1.07055295e+00 -4.70267534e-01 9.64186668e-01
6.75005078e-01 3.01003903e-02 1.05256927e+00 -4.18460041e-01
-4.79971170e-01 1.86857134e-01 -2.20699221e-01 1.48343384e-01
2.15488523e-01 -1.22053504e+00 6.79642797e-01 6.75218582e-01
2.67602026e-01 5.48896432e-01 8.68640244e-01 -3.69102299e-01
-1.72992516e+00 -6.25961304e-01 -1.50271237e-01 -5.29863834e-01
5.43172717e-01 -4.54796612e-01 -6.05909646e-01 9.59551752e-01
3.61859590e-01 7.46802539e-02 2.44938672e-01 1.14287607e-01
-5.10072529e-01 -2.55127195e-02 -1.03713548e+00 1.28142858e+00
1.34784186e+00 -4.52052861e-01 -7.92516291e-01 4.26686592e-02
5.96595049e-01 -5.19882083e-01 -6.31918013e-01 4.58034240e-02
3.63821954e-01 -1.25191438e+00 1.21808398e+00 -7.65470564e-01
5.46301663e-01 -4.07503009e-01 -2.62722045e-01 -1.28245628e+00
-5.16305149e-01 -3.13358128e-01 -1.59880728e-01 5.53984940e-01
8.16173702e-02 -5.17295122e-01 8.11600387e-01 4.95151877e-01
-8.45266804e-02 -2.97134250e-01 -6.95229411e-01 -7.94810891e-01
2.05752432e-01 -7.11873055e-01 6.22854233e-01 5.06124437e-01
1.89896896e-02 -2.17066170e-03 7.31935948e-02 3.62600647e-02
3.11366737e-01 2.76187450e-01 9.59749639e-01 -8.72626066e-01
-3.18484545e-01 -5.97221255e-01 -5.23585260e-01 -5.59514999e-01
2.02202424e-01 -7.25783050e-01 1.48074523e-01 -1.83110976e+00
1.28262460e-01 -3.56873363e-01 -3.28031540e-01 7.77832925e-01
1.61358163e-01 1.69377059e-01 5.87905228e-01 3.63920070e-02
-9.75827396e-01 8.12467277e-01 1.48486733e+00 -1.58791736e-01
-1.54208586e-01 -4.47559178e-01 -7.33752906e-01 7.33241796e-01
1.20269608e+00 -2.23760307e-01 -6.88454807e-01 -6.01954341e-01
2.99026906e-01 -9.98748317e-02 7.32794523e-01 -1.40212166e+00
-6.55706897e-02 -8.04820359e-01 5.24952412e-01 -2.69066602e-01
8.49573314e-01 -6.10706866e-01 -1.12249419e-01 7.61339724e-01
-6.52301908e-01 3.02320123e-01 5.37538111e-01 2.68164784e-01
1.39564335e-01 -2.21907362e-01 7.05236912e-01 -8.18703294e-01
-1.27741456e+00 1.40788868e-01 -9.10364330e-01 1.82120293e-01
1.09818065e+00 -2.35761061e-01 -4.93836373e-01 -5.95549405e-01
-6.74276590e-01 3.05983461e-02 6.18499398e-01 5.39202631e-01
7.76136220e-01 -1.07377851e+00 -4.50153917e-01 -1.94030359e-01
1.25720814e-01 -6.35690019e-02 2.53610700e-01 2.50509024e-01
-9.09475446e-01 1.70087665e-01 -1.06176949e+00 -2.15931281e-01
-9.10826027e-01 8.33571017e-01 5.89876115e-01 -1.69906631e-01
-7.55566299e-01 6.11812174e-01 4.37537819e-01 -4.44513559e-01
-1.70981232e-02 -5.55748641e-01 -1.24415345e-01 -5.33011556e-01
6.09675467e-01 7.22833723e-02 -9.53297317e-02 -3.74532938e-01
-3.06989312e-01 -6.36451617e-02 -5.58068603e-02 -6.37520552e-01
1.45763206e+00 3.51945721e-02 9.93370414e-02 4.92389023e-01
5.01049280e-01 -5.98132908e-01 -1.48935795e+00 -7.49372169e-02
-1.45561010e-01 -3.69942486e-01 2.03686863e-01 -8.74708056e-01
-8.79501462e-01 9.32094336e-01 6.79874659e-01 -1.83950961e-01
9.59719658e-01 1.28162401e-02 3.04572642e-01 9.19235528e-01
5.79569876e-01 -1.19324815e+00 6.13510370e-01 6.59298182e-01
1.13239586e+00 -9.49201941e-01 2.89724134e-02 1.34067282e-01
-8.69230449e-01 8.48784566e-01 1.09221947e+00 -4.47315186e-01
2.32636020e-01 2.05173030e-01 -1.81203559e-02 -3.32288861e-01
-1.09307456e+00 -6.52997971e-01 -3.40221077e-01 9.73181665e-01
2.77871996e-01 1.10363990e-01 1.43472999e-01 2.66058058e-01
-3.74940544e-01 -1.22959474e-02 8.92339468e-01 1.26228976e+00
-5.26553512e-01 -1.01552796e+00 -3.03014696e-01 1.74055144e-01
-3.09275463e-02 -9.75775421e-02 -2.84740627e-01 9.41925228e-01
-1.56803221e-01 7.93614686e-01 1.24203965e-01 -1.91291541e-01
3.46243232e-01 -1.45108446e-01 1.01649368e+00 -7.03299284e-01
-6.04967237e-01 -2.80898660e-01 -2.19470682e-03 -1.11463308e+00
-3.22054356e-01 -5.33322811e-01 -1.65990794e+00 -5.63391626e-01
3.28247756e-01 -2.57419020e-01 5.84495187e-01 7.11361468e-01
3.35395753e-01 6.36417389e-01 3.08662299e-02 -1.20226574e+00
-2.55030036e-01 -5.80550492e-01 -4.76210862e-01 4.17422205e-01
1.03520662e-01 -9.05885696e-01 1.67245641e-01 -1.65425122e-01] | [4.1944074630737305, 1.1579225063323975] |
7e6a2a9d-7b56-4788-b2e6-3d80af10923a | unsupervised-deep-homography-a-fast-and | 1709.03966 | null | http://arxiv.org/abs/1709.03966v3 | http://arxiv.org/pdf/1709.03966v3.pdf | Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model | Homography estimation between multiple aerial images can provide relative
pose estimation for collaborative autonomous exploration and monitoring. The
usage on a robotic system requires a fast and robust homography estimation
algorithm. In this study, we propose an unsupervised learning algorithm that
trains a Deep Convolutional Neural Network to estimate planar homographies. We
compare the proposed algorithm to traditional feature-based and direct methods,
as well as a corresponding supervised learning algorithm. Our empirical results
demonstrate that compared to traditional approaches, the unsupervised algorithm
achieves faster inference speed, while maintaining comparable or better
accuracy and robustness to illumination variation. In addition, on both a
synthetic dataset and representative real-world aerial dataset, our
unsupervised method has superior adaptability and performance compared to the
supervised deep learning method. | ['Steven W. Chen', 'Ty Nguyen', 'Camillo J. Taylor', 'Vijay Kumar', 'Shreyas S. Shivakumar'] | 2017-09-12 | null | null | null | null | ['homography-estimation'] | ['computer-vision'] | [ 2.09485054e-01 -1.15305185e-01 -7.56547600e-02 -4.19510782e-01
-9.75687504e-02 -5.93257427e-01 4.66288149e-01 -2.55357713e-01
-2.93530673e-01 5.89323223e-01 -1.90509692e-01 -3.35796066e-02
-5.87810755e-01 -1.08485198e+00 -8.76835346e-01 -3.81709427e-01
-3.28372210e-01 5.29792666e-01 7.61548057e-02 -1.78591460e-01
1.91201344e-01 7.47248113e-01 -1.46711874e+00 -6.84503198e-01
8.43229413e-01 1.11934042e+00 3.89705420e-01 5.35776854e-01
4.63856280e-01 7.48107135e-01 -5.05265534e-01 7.43642896e-02
8.89712811e-01 8.16458240e-02 -7.86970556e-01 3.20471108e-01
9.63307858e-01 -9.23304737e-01 -8.06544721e-01 1.09063745e+00
2.21643627e-01 3.36411774e-01 5.63978136e-01 -1.08856952e+00
-4.96313959e-01 4.39343393e-01 -5.13300359e-01 -3.67950588e-01
2.18094528e-01 5.65105788e-02 8.67354512e-01 -6.19567633e-01
7.53664494e-01 9.27534342e-01 8.16075861e-01 -1.73503309e-01
-1.04578066e+00 -6.04264557e-01 -5.26636988e-02 -5.88955940e-04
-1.47604728e+00 -4.86364700e-02 8.07132363e-01 -4.05165136e-01
7.33048856e-01 -2.71009833e-01 8.14680099e-01 6.88030839e-01
4.18656468e-01 7.38056183e-01 8.36887658e-01 -2.62240410e-01
-1.99120175e-02 -2.57669359e-01 -2.71672577e-01 1.29374778e+00
3.30843151e-01 4.37144250e-01 -3.68333220e-01 3.16506885e-02
1.27242792e+00 1.93459272e-01 -5.53748786e-01 -8.92721415e-01
-1.45874393e+00 5.52983642e-01 8.79245698e-01 -2.05347836e-01
-4.34322745e-01 4.95045602e-01 -6.54550781e-03 2.89145589e-01
7.16328248e-02 8.02674174e-01 -4.22132283e-01 1.02053292e-01
-9.22607183e-01 8.73311087e-02 9.57652688e-01 1.55344796e+00
1.10199857e+00 1.93462655e-01 4.48065042e-01 5.07723033e-01
3.77494663e-01 7.90070117e-01 3.97858948e-01 -1.11467612e+00
1.09504037e-01 6.37745678e-01 1.93309516e-01 -1.65891600e+00
-5.01630068e-01 -5.22128344e-01 -9.26569283e-01 4.06562805e-01
1.47382975e-01 -2.49353275e-01 -7.53564835e-01 1.45203698e+00
1.13160133e-01 6.33559143e-03 1.56401858e-01 9.73381698e-01
6.97997332e-01 4.72669095e-01 -7.49265075e-01 9.63800997e-02
7.10684657e-01 -1.09564126e+00 -7.32131422e-01 -3.35141897e-01
3.12151760e-01 -8.18601310e-01 7.75479615e-01 5.64745963e-01
-6.43598616e-01 -6.89728022e-01 -1.55485225e+00 -8.30316320e-02
-4.32644635e-01 8.53612959e-01 9.99278784e-01 1.90896302e-01
-9.08407569e-01 8.30106378e-01 -7.93376207e-01 -6.35780990e-01
1.25102416e-01 7.34975815e-01 -5.40559411e-01 2.44517475e-01
-8.22592735e-01 6.34381533e-01 6.91046536e-01 1.37802020e-01
-9.61900711e-01 -2.17757702e-01 -9.99092758e-01 -1.02080986e-01
4.22604918e-01 -7.16922820e-01 1.31190181e+00 -6.99578822e-01
-2.17661667e+00 4.55602765e-01 2.34087363e-01 -7.94123828e-01
3.87905896e-01 -6.44442439e-01 1.38239428e-01 2.30520070e-01
-9.27711651e-02 9.52027440e-01 9.76416707e-01 -1.14966834e+00
-5.73718667e-01 -3.88085157e-01 3.80325854e-01 3.17494839e-01
-3.33278835e-01 -7.30995297e-01 -2.21051931e-01 -3.48993957e-01
8.31793368e-01 -1.21346700e+00 -1.34648338e-01 4.67378587e-01
-3.58536124e-01 1.70649901e-01 1.16727316e+00 -3.76806170e-01
4.81565922e-01 -1.68166447e+00 4.12647784e-01 2.34587893e-01
1.89395308e-01 4.72022071e-02 1.92461237e-01 3.96755755e-01
3.30532640e-01 -5.80853224e-01 -3.02420527e-01 5.18643819e-02
8.12755078e-02 2.93929368e-01 -3.58640432e-01 8.28234017e-01
-1.18885562e-01 7.44540095e-01 -1.04151726e+00 -2.59020030e-01
5.08503735e-01 2.77166724e-01 -2.79307067e-01 2.27361605e-01
-1.30232692e-01 4.77473229e-01 -2.94902861e-01 8.01114440e-01
7.24527359e-01 -3.79611462e-01 3.14437270e-01 -4.72600013e-01
-1.55715361e-01 -1.02062970e-01 -1.21261072e+00 1.87310958e+00
-5.12888432e-01 1.09044802e+00 8.03740881e-03 -9.75883901e-01
1.51891112e+00 -1.57732189e-01 3.89430851e-01 -2.56619334e-01
5.33000767e-01 1.94394797e-01 -1.65557429e-01 -3.30194592e-01
8.86323094e-01 5.24286568e-01 9.89087000e-02 1.27316326e-01
3.35857630e-01 -5.55827022e-01 -1.16935641e-01 -5.41324243e-02
6.93837345e-01 5.70703149e-01 4.16903645e-01 -4.69430059e-01
5.35917401e-01 7.76111186e-02 3.56466055e-01 5.16702414e-01
-8.49727988e-02 1.76922768e-01 -1.10705428e-01 -8.31656694e-01
-7.18212247e-01 -7.69346476e-01 -1.32234156e-01 4.64825302e-01
8.03190589e-01 -3.76932144e-01 -4.40752208e-01 -4.37212169e-01
3.01404387e-01 -2.99058948e-02 -3.77711296e-01 -3.49518359e-02
-3.50240171e-01 -2.91510791e-01 5.30333519e-01 7.09659874e-01
1.31292331e+00 -6.14778399e-01 -1.00541627e+00 -2.88681369e-02
-2.67731417e-02 -1.44301295e+00 7.57856071e-02 1.57115921e-01
-1.04762578e+00 -1.22838211e+00 -3.77669811e-01 -9.83896732e-01
8.47203732e-01 5.29543340e-01 8.53736103e-01 -2.30958387e-01
-2.51392901e-01 6.15648866e-01 -1.72528625e-01 -2.98586816e-01
1.92530692e-01 1.46478578e-01 4.55138326e-01 -2.47010812e-01
2.80295074e-01 -6.98037148e-01 -5.48707545e-01 4.07358676e-01
-5.20633936e-01 -9.03688967e-02 8.56608570e-01 8.98978651e-01
7.68303871e-01 6.65486082e-02 -2.23603889e-01 -3.53857487e-01
3.50750178e-01 2.20621447e-03 -1.46665931e+00 9.11756605e-02
-7.49028444e-01 7.45474473e-02 7.26632893e-01 -2.83749580e-01
-7.64028370e-01 6.90635085e-01 4.89832640e-01 -6.68012023e-01
-3.01622778e-01 5.41962743e-01 -2.73000728e-02 -9.17362988e-01
5.24412990e-01 4.05972093e-01 2.03965455e-01 -2.55172670e-01
3.52143496e-01 5.68981767e-01 9.89906192e-01 -3.23247015e-01
1.23611438e+00 6.54116154e-01 4.05883789e-01 -1.12532878e+00
-4.49786872e-01 -2.97960937e-01 -1.44868934e+00 -3.08402508e-01
6.47286952e-01 -1.01011014e+00 -7.57901967e-01 5.35981715e-01
-1.14364493e+00 -4.55198109e-01 4.76033837e-02 9.01209056e-01
-7.96865463e-01 6.42188847e-01 -2.93582171e-01 -5.99432468e-01
-2.67540336e-01 -1.26215851e+00 1.15480340e+00 2.95569330e-01
1.50996238e-01 -1.05279899e+00 1.14815190e-01 4.24232148e-02
1.89891592e-01 4.90549892e-01 9.17297900e-02 -1.93378046e-01
-1.10117495e+00 -4.46767509e-01 -2.64000684e-01 1.85658649e-01
3.26946139e-01 2.50920922e-01 -5.90142667e-01 -5.10910034e-01
-2.65374810e-01 -5.32388866e-01 6.89737558e-01 3.08294117e-01
1.07019377e+00 -2.56884217e-01 -2.38025665e-01 1.26618683e+00
1.53562999e+00 1.83513254e-01 5.10405123e-01 5.98613143e-01
8.31671357e-01 4.52483505e-01 9.20067549e-01 4.10685897e-01
2.08084002e-01 4.05644000e-01 7.90785491e-01 -7.92764351e-02
3.80225778e-01 -3.82208765e-01 1.32121980e-01 8.18568289e-01
-2.21225798e-01 -1.84498578e-02 -7.77323365e-01 3.11110646e-01
-2.05826449e+00 -5.82312763e-01 3.53651606e-02 2.20999432e+00
2.26995364e-01 -1.62556380e-01 -4.33570445e-01 -2.71587167e-02
4.29025114e-01 3.25314105e-01 -5.72611034e-01 3.58847380e-02
-8.66638348e-02 -1.00366451e-01 1.02405989e+00 4.28161740e-01
-1.57566786e+00 1.30873322e+00 7.17326975e+00 1.38181999e-01
-1.17059374e+00 -4.86837268e-01 -2.38605425e-01 3.66614670e-01
3.58293146e-01 1.43162653e-01 -6.48453891e-01 -9.67421830e-02
2.40186796e-01 -3.51174101e-02 4.59742606e-01 1.30667830e+00
-2.69488394e-01 -4.51701842e-02 -9.85230565e-01 1.18127298e+00
3.51170927e-01 -1.21621299e+00 1.70841098e-01 2.28238821e-01
8.07540357e-01 2.20732704e-01 -5.60717396e-02 -7.90115669e-02
4.65464294e-01 -7.51078427e-01 2.82744229e-01 5.65901160e-01
5.92055440e-01 -8.08391333e-01 1.01726270e+00 2.86835194e-01
-1.44588470e+00 -6.93332702e-02 -8.49534452e-01 -4.38486606e-01
-8.65343288e-02 1.70395419e-01 -6.91836536e-01 7.84576535e-01
7.73958206e-01 1.08884823e+00 -5.61985373e-01 1.03639603e+00
-5.48178554e-01 2.01491266e-02 -3.90813738e-01 -6.47393465e-02
5.38828731e-01 -4.68126327e-01 5.44217527e-01 7.56023586e-01
3.30240428e-01 6.76069483e-02 5.82542896e-01 8.44242573e-01
-1.27939716e-01 7.81004354e-02 -1.21334624e+00 -2.28970617e-01
4.73339200e-01 1.34993446e+00 -7.50382185e-01 -3.43445152e-01
-3.17972928e-01 1.06007135e+00 2.88799077e-01 1.34463102e-01
-6.42290056e-01 -8.94601405e-01 4.26287025e-01 -4.20937300e-01
2.87592888e-01 -8.80293310e-01 8.20727423e-02 -1.13386178e+00
-2.38531247e-01 -6.12454057e-01 -8.06755722e-02 -8.01499367e-01
-9.41427350e-01 6.35231674e-01 1.66555509e-01 -1.63081312e+00
-2.71231145e-01 -1.16435981e+00 -5.90276301e-01 3.22748244e-01
-1.53263605e+00 -1.30105603e+00 -1.23288667e+00 5.36626101e-01
5.14047265e-01 -7.54674911e-01 8.00071716e-01 -1.98633671e-01
-3.58260334e-01 2.91530728e-01 2.21644461e-01 2.82796323e-01
8.04143131e-01 -1.37184167e+00 2.76938438e-01 1.03464448e+00
2.23871410e-01 7.72253513e-01 5.66289425e-01 -6.40464008e-01
-1.84946346e+00 -1.05252206e+00 1.60333082e-01 -9.51255485e-02
6.11042798e-01 2.49590259e-02 -4.57588971e-01 1.02173138e+00
2.68437177e-01 -7.80265853e-02 4.24694270e-01 -1.97904572e-01
-5.12182526e-02 -1.72058076e-01 -9.93602812e-01 3.03632408e-01
1.09529793e+00 -4.91505861e-01 -4.48713422e-01 5.74197531e-01
6.19704902e-01 -7.36273468e-01 -1.12503779e+00 7.31256068e-01
9.09811854e-01 -9.59408879e-01 9.18287098e-01 -6.85039908e-02
5.16329885e-01 -4.98294711e-01 -3.65441144e-01 -1.14414418e+00
-5.42865992e-01 -8.22720349e-01 -2.91776776e-01 5.90456486e-01
-5.30410791e-03 -6.30599320e-01 7.73637235e-01 8.05701967e-03
-7.48877898e-02 -5.03245533e-01 -1.95000902e-01 -9.70329165e-01
-3.07772636e-01 1.09850451e-01 6.91265762e-01 9.60530341e-01
-2.59303451e-01 1.40594214e-01 -6.16525531e-01 8.98020923e-01
9.43422318e-01 6.49435341e-01 1.47476256e+00 -1.51004457e+00
-1.64579660e-01 -1.03360385e-01 -8.91021073e-01 -1.57485545e+00
3.89919817e-01 -6.09485865e-01 2.09543243e-01 -1.53309417e+00
-3.04235339e-01 1.17800847e-01 2.06149355e-01 3.20307732e-01
4.80480254e-01 3.79447401e-01 -1.68636486e-01 4.14754272e-01
-3.60904843e-01 8.55734885e-01 1.20759749e+00 -4.23217356e-01
-2.46763498e-01 -8.53511915e-02 1.34610116e-01 1.14146328e+00
9.42292809e-01 4.54627462e-02 -5.61593056e-01 -7.99946785e-01
-7.52263963e-02 -2.09690094e-01 4.51931357e-01 -1.48670006e+00
6.58264279e-01 -5.53380176e-02 6.52273893e-01 -1.00060451e+00
3.01221728e-01 -9.11637008e-01 -1.71602786e-01 7.85734177e-01
6.19087648e-03 1.35787994e-01 1.59352034e-01 6.67120934e-01
-2.89407432e-01 -1.17485970e-01 6.16829455e-01 -2.98111916e-01
-1.22747970e+00 5.56016386e-01 -2.46324822e-01 -5.31219065e-01
8.35896254e-01 -3.45401436e-01 -2.01081410e-01 -6.40013993e-01
-2.96529025e-01 2.37163067e-01 7.04999149e-01 3.29073757e-01
9.61369753e-01 -1.20384240e+00 -1.99756101e-01 4.64413941e-01
2.73088336e-01 2.34591052e-01 -3.54315966e-01 6.63071334e-01
-1.32555068e+00 4.94330317e-01 -7.49711990e-01 -8.07929873e-01
-1.01813769e+00 3.21718186e-01 3.87667745e-01 3.71016204e-01
-7.82129645e-01 7.21599817e-01 -1.07469812e-01 -9.13105190e-01
6.17980719e-01 -4.45542693e-01 -1.07590325e-01 -4.48090643e-01
1.91273883e-01 5.48514724e-01 -4.81429935e-01 -7.38990664e-01
-2.08589822e-01 9.06075060e-01 3.15315753e-01 8.78830478e-02
1.17979932e+00 -9.73040015e-02 -2.24152297e-01 2.29707584e-01
1.02250469e+00 -3.41397375e-01 -1.22894239e+00 -3.21260333e-01
-1.89314142e-01 -7.48710394e-01 4.42835301e-01 -2.99282134e-01
-1.16433430e+00 7.70948112e-01 7.78140783e-01 -8.79254416e-02
8.66450608e-01 -5.38232923e-01 7.53471553e-01 1.53774726e+00
8.16526353e-01 -1.15483594e+00 1.65845528e-01 7.53584385e-01
8.93228590e-01 -1.51685202e+00 6.11992359e-01 -4.77510691e-01
-2.63801306e-01 1.62773931e+00 8.81995857e-01 -6.16273880e-01
5.47339439e-01 1.68053895e-01 1.02884591e-01 -3.72833997e-01
-2.25895792e-01 -2.79310346e-01 3.36019903e-01 6.45249903e-01
1.63631409e-01 -1.41604275e-01 1.01839855e-01 -1.30038247e-01
-6.85590088e-01 7.89369270e-03 4.83592093e-01 1.14073825e+00
-6.03953481e-01 -7.32335985e-01 -1.37213111e-01 6.29723147e-02
2.30113462e-01 3.21227312e-01 -8.20695102e-01 1.14175725e+00
-7.89932385e-02 4.99997169e-01 2.29827389e-01 -7.71056712e-01
2.32589781e-01 -3.15342724e-01 7.31778324e-01 -1.61493853e-01
-7.21472315e-03 -1.00453235e-01 -1.60802945e-01 -8.74864280e-01
-5.89584827e-01 -9.08444300e-02 -9.78960633e-01 -2.50262231e-01
-6.82728171e-01 -1.41306475e-01 7.19662607e-01 8.72145057e-01
2.30313420e-01 3.33525151e-01 7.40453243e-01 -1.32977688e+00
-7.01225936e-01 -9.36046779e-01 -6.55385911e-01 -8.74181092e-02
2.29898080e-01 -1.02798164e+00 -2.34911546e-01 6.37603626e-02] | [7.8733649253845215, -2.0742275714874268] |
d09d3337-4189-4671-a575-d1290b27e9e3 | distributed-cooperative-control-and | 2208.13412 | null | https://arxiv.org/abs/2208.13412v3 | https://arxiv.org/pdf/2208.13412v3.pdf | Distributed Cooperative Control and Optimization of Connected Automated Vehicles Platoon Against Cut-in Behaviors of Social Drivers | Connected automated vehicles (CAVs) have brought new opportunities to improve traffic throughput and reduce energy consumption. However, the uncertain lane-change behaviors (LCBs) of surrounding vehicles (SVs) as an uncontrollable factor significantly threaten the driving safety and the consistent movement of a group of platoon CAVs. How to ensure safe, efficient, and fuel economic platoon control poses a key challenge faced by researchers in complex traffic environments. This study proposes a dynamic platoon management and cooperative driving framework for a mixed traffic flow consisting of multiple CAVs and possible human-driven vehicles (HDVs) as the SVs on unsignalized roads. In the proposed framework, the leader CAV of the platoon provides a high-level automatic driving decision to the follower CAVs by developing an optimal trajectory estimation of the HDVs while distributed observers and tracking controllers are properly implemented by the follower CAVs. Specifically, the proposed framework consists of three stages. At the observation stage, the cruising information of all the SVs will be collected by the leader CAV through the Cellular-Vehicle-to-X (C-V2X) infrastructure, while an automatic decision-making driving assistance system (ADMDSS) is constructed to determine the driving states of the platoon. When the HDVs approach the communication range of the platoon, in the prediction stage, the trajectories of the HDVs as the target vehicles will be estimated and the reference trajectory planning for the leader CAV and the cooperative controller design for the follower CAVs will be respectively activated by using C-V2X infrastructure. Simulation cases are presented to illustrate the effectiveness of the proposed approaches. | ['Rong Su', 'Bohui Wang'] | 2022-08-29 | null | null | null | null | ['trajectory-planning'] | ['robots'] | [-3.89366567e-01 4.47305083e-01 -1.39760509e-01 5.17475270e-02
2.60830913e-02 -5.28095603e-01 5.38200378e-01 -1.20326683e-01
-1.65520534e-01 9.36668873e-01 -5.42723596e-01 -6.27943337e-01
-3.81956518e-01 -9.85413074e-01 -4.51734185e-01 -1.08484519e+00
-2.47621611e-01 4.81425762e-01 7.77058721e-01 -4.24185932e-01
-1.96193717e-02 7.29804873e-01 -1.33676708e+00 -6.93238378e-01
9.36726749e-01 1.02324009e+00 4.87651676e-01 4.65791047e-01
9.71047357e-02 5.80205321e-01 -1.35108173e-01 2.00549304e-01
2.89636999e-01 -1.52269304e-01 1.09327473e-01 9.80405882e-02
-4.51476693e-01 -3.87403458e-01 -4.29415971e-01 8.97261918e-01
3.65448296e-02 4.03669447e-01 4.57865506e-01 -2.30537128e+00
3.39324474e-01 1.61251098e-01 -4.26279545e-01 2.97769858e-03
-4.77072716e-01 7.52126038e-01 3.38819236e-01 -3.74751538e-01
5.91308296e-01 1.26068258e+00 1.20873600e-01 3.06473404e-01
-8.39276075e-01 -9.80988979e-01 4.05902624e-01 7.37963915e-01
-1.72818935e+00 -5.35615027e-01 6.35711789e-01 -4.41560179e-01
1.94120899e-01 2.84722269e-01 8.96283925e-01 2.87051737e-01
6.33687973e-01 3.77473474e-01 6.19025946e-01 1.81930616e-01
3.11956048e-01 4.84582216e-01 6.92682862e-02 3.07812303e-01
5.71674168e-01 3.18709254e-01 2.01113984e-01 8.41131955e-02
3.37540768e-02 -1.26035541e-01 -9.77571011e-02 -2.12488368e-01
-8.44497800e-01 6.52010024e-01 3.05516779e-01 -1.37679996e-02
-9.30214167e-01 7.01887980e-02 3.64109278e-01 1.20953463e-01
4.51014889e-03 -2.92190492e-01 1.31413750e-02 7.66049698e-02
-6.79284751e-01 3.58534664e-01 4.53939766e-01 1.27916455e+00
8.25124323e-01 2.56884873e-01 -2.22224355e-01 -8.77227560e-02
4.77579057e-01 9.45189893e-01 -4.84282404e-01 -1.06517649e+00
3.60899627e-01 6.76591933e-01 5.00602245e-01 -1.29828024e+00
-3.91032428e-01 -3.90642703e-01 -7.23380387e-01 4.19135004e-01
-4.92408015e-02 -5.93381643e-01 -2.73253053e-01 1.41175997e+00
7.94196844e-01 2.12332100e-01 2.59710789e-01 8.77347171e-01
1.37749650e-02 1.21759558e+00 1.68625548e-01 -7.39540875e-01
1.16572595e+00 -6.28009677e-01 -1.05689454e+00 -7.45355934e-02
4.05902386e-01 -4.41338986e-01 -1.39806578e-02 -5.48642070e-04
-9.30763900e-01 -5.26106775e-01 -1.19947970e+00 6.07314944e-01
-1.92851141e-01 1.63371205e-01 -3.82516861e-01 1.06128454e-01
-8.43102634e-01 -1.00076281e-01 -6.64951622e-01 -2.58005887e-01
3.69461387e-01 3.08503538e-01 5.59388883e-02 -1.47722051e-01
-1.33571517e+00 1.01946759e+00 3.25892121e-01 4.96928841e-01
-1.51240897e+00 -6.75345421e-01 -5.61324179e-01 -8.27397108e-02
6.53339863e-01 -2.09678203e-01 6.86687112e-01 -7.15794325e-01
-1.00448227e+00 -9.56246105e-04 -4.08585742e-02 -4.95023817e-01
8.00095260e-01 5.22863925e-01 -6.05121851e-01 1.47927463e-01
3.10463041e-01 3.31901908e-01 2.67136216e-01 -1.30692029e+00
-1.32969391e+00 -1.38916168e-02 -3.07368904e-01 3.68066430e-01
1.74251527e-01 -2.93978274e-01 -2.09130824e-01 3.13113898e-01
-5.28431177e-01 -1.50076389e+00 -3.38133186e-01 9.17439461e-02
-5.96655011e-01 -6.08487666e-01 1.51921165e+00 -1.61491081e-01
1.15840578e+00 -2.34537029e+00 -1.78857908e-01 4.41814005e-01
1.60736620e-01 4.38616782e-01 3.27832922e-02 7.67624199e-01
4.87116039e-01 -3.85068089e-01 3.01528245e-01 1.22401230e-01
-8.30447599e-02 3.11280370e-01 -2.04167604e-01 6.33916080e-01
-7.56037980e-02 2.62587041e-01 -6.44660473e-01 -5.10909796e-01
4.30656493e-01 3.39846194e-01 -1.51742734e-02 4.14566129e-01
-8.06929544e-02 5.07879138e-01 -9.93090630e-01 1.16940491e-01
1.07389903e+00 4.28020954e-01 -3.00001055e-02 -3.84250060e-02
-1.04817963e+00 -3.89406681e-01 -1.26976299e+00 2.88986713e-01
-2.16101691e-01 8.09134781e-01 8.01780939e-01 -8.94112110e-01
1.01479816e+00 2.82413870e-01 6.02779150e-01 -3.40451896e-01
3.16429675e-01 2.58650094e-01 2.30096236e-01 -5.22950649e-01
2.22510114e-01 1.92198247e-01 1.64988190e-02 4.53274585e-02
-7.05685377e-01 1.23208717e-01 1.53176785e-01 2.57418126e-01
9.84151900e-01 -4.23107207e-01 -1.16891410e-04 -3.72112393e-01
1.18759406e+00 4.69663799e-01 9.72329795e-01 1.21524788e-01
-6.68811321e-01 -7.88757861e-01 3.60183716e-01 -8.87514576e-02
-8.49695504e-01 -7.18215048e-01 5.28673865e-02 1.94765672e-01
9.84162390e-01 3.59384090e-01 -4.98789042e-01 -1.60428181e-01
1.76871255e-01 1.13786328e+00 -1.94256663e-01 -3.34963918e-01
-5.05650103e-01 9.95533764e-02 2.15967558e-02 -7.55565688e-02
7.40710855e-01 -5.30075073e-01 -6.11435056e-01 5.03294408e-01
1.87879011e-01 -1.25140953e+00 -4.85067457e-01 -2.99414247e-01
8.37979093e-02 -1.37139273e+00 3.16797346e-02 -6.95885897e-01
7.86434114e-01 6.42949641e-01 -2.33307600e-01 1.62574336e-01
3.26897919e-01 4.74751778e-02 1.34658098e-01 -6.97262168e-01
-8.42244804e-01 -1.19810492e-01 1.99368939e-01 6.03049934e-01
2.61029065e-01 -1.06197536e-01 -6.70796871e-01 7.59929597e-01
-3.43044400e-01 3.57025206e-01 2.12704137e-01 8.01694542e-02
6.96129978e-01 6.65542781e-01 9.23412621e-01 -3.14411640e-01
3.03184152e-01 -8.80460739e-01 -1.22079611e+00 -1.92724898e-01
-6.60179794e-01 -4.78720367e-01 9.89250779e-01 -1.41825333e-01
-9.38121557e-01 1.93175182e-01 1.49295449e-01 -3.87890339e-01
-2.44362146e-01 -3.60591076e-02 -5.54181993e-01 -1.18438698e-01
-9.52618644e-02 4.12388265e-01 4.47907299e-01 1.82434097e-01
8.29524621e-02 9.33340490e-01 2.80431002e-01 1.78792372e-01
1.43080604e+00 4.79665279e-01 6.63330972e-01 -9.15933311e-01
2.07664758e-01 -1.37224272e-01 -1.98429495e-01 -1.05677259e+00
8.47455740e-01 -1.19232726e+00 -1.17284131e+00 3.22540969e-01
-1.21535659e+00 -2.30205655e-02 3.06612588e-02 6.17110193e-01
-3.70964468e-01 -5.32466620e-02 -1.30885169e-01 -1.27298295e+00
-3.57979536e-02 -1.41022158e+00 4.45666254e-01 4.86044973e-01
2.09087282e-01 -7.26837277e-01 -2.76160926e-01 1.96828023e-01
5.44436455e-01 5.38347125e-01 7.86993742e-01 -4.16714311e-01
-1.25778079e+00 -1.48615748e-01 2.81410422e-02 5.64459413e-02
-1.13548756e-01 2.40306884e-01 -2.91119903e-01 -4.91329223e-01
-2.80461550e-01 5.29339671e-01 -1.76661331e-02 3.04511428e-01
3.26510489e-01 -6.24799132e-01 -1.09195447e+00 2.35776126e-01
1.52711916e+00 1.06466353e+00 4.22684878e-01 3.31029296e-01
4.22202468e-01 9.54668045e-01 1.25373054e+00 1.97374418e-01
9.61563587e-01 6.62059903e-01 1.04762101e+00 6.96968362e-02
1.63383514e-01 -2.32451901e-01 5.78991294e-01 5.58014750e-01
2.46096328e-01 -3.86217058e-01 -4.83908892e-01 7.62904227e-01
-1.86950088e+00 -8.88008416e-01 -8.70083213e-01 1.99146760e+00
1.10504813e-01 1.12015076e-01 -1.88646931e-02 1.62057012e-01
1.19618094e+00 -4.98272702e-02 -6.19291425e-01 -4.28188443e-01
2.39628002e-01 -9.48885322e-01 9.60744619e-01 5.12931526e-01
-4.06309724e-01 5.03074348e-01 3.60709023e+00 1.04623342e+00
-9.38455641e-01 2.09778808e-02 2.27224633e-01 1.70882702e-01
-1.70812726e-01 4.67864007e-01 -1.07891166e+00 8.47523332e-01
1.12431419e+00 -7.02962816e-01 4.18595254e-01 8.07519853e-01
1.40790391e+00 -2.94882506e-01 -5.11995852e-01 3.61543745e-01
-5.61319828e-01 -1.30576098e+00 -4.86388445e-01 2.63013572e-01
5.45750320e-01 9.57835466e-02 -5.08154988e-01 6.40155077e-02
1.26278892e-01 -2.28047043e-01 7.88796127e-01 6.78130448e-01
3.48543912e-01 -1.24473166e+00 5.33219993e-01 8.95663679e-01
-1.63813269e+00 -5.24850547e-01 7.73684680e-02 1.32117569e-01
8.85963500e-01 3.28972727e-01 -5.79978466e-01 4.93269533e-01
2.05551907e-02 4.87685978e-01 -1.32316843e-01 8.89087737e-01
-1.01691790e-01 3.83096099e-01 -1.81487113e-01 -4.90651190e-01
5.83566904e-01 -5.59970677e-01 1.27482378e+00 5.19099176e-01
3.02838385e-01 3.01378071e-01 5.45496047e-01 7.11474001e-01
3.87561053e-01 -6.66013733e-02 -7.39004791e-01 3.66730362e-01
9.55968916e-01 1.41325986e+00 -4.26949769e-01 -4.24019814e-01
-2.35836938e-01 -4.13614176e-02 -2.55967438e-01 5.74590087e-01
-1.14286745e+00 -8.87632489e-01 1.12555134e+00 7.05383599e-01
1.05038002e-01 -4.09434766e-01 1.40780032e-01 -1.32632867e-01
-2.13876307e-01 -1.50068864e-01 -1.77695677e-01 -7.75059640e-01
-6.00364208e-01 5.07821739e-01 8.63030925e-02 -1.55938160e+00
2.00002685e-01 4.15848754e-02 -1.05014992e+00 8.14465940e-01
-1.90551758e+00 -8.42103779e-01 -4.32695031e-01 6.15322948e-01
2.92859286e-01 -3.12613487e-01 -2.39672691e-01 3.79066318e-01
-8.96207452e-01 -3.04091811e-01 2.54915059e-01 -3.32873672e-01
1.75862193e-01 -2.63649702e-01 -3.77707034e-01 1.08846319e+00
-1.15091956e+00 -5.61873093e-02 9.49772418e-01 -7.83226490e-01
-1.81414509e+00 -1.67402804e+00 8.29040408e-01 5.57580531e-01
7.25926042e-01 -5.42916358e-02 -5.18795252e-01 4.78218138e-01
3.33797216e-01 -8.16100463e-02 -2.57616341e-01 -1.24702442e+00
9.61751521e-01 -6.60290956e-01 -1.07706678e+00 7.80224681e-01
5.62573671e-01 9.66181457e-02 1.62858471e-01 2.61474609e-01
6.03419304e-01 -4.40663546e-02 -4.00734097e-01 2.07099006e-01
1.48215801e-01 -3.62303317e-01 2.78160602e-01 -8.82622004e-02
-4.96223092e-01 -1.02265310e+00 1.45318031e-01 -1.38466752e+00
-1.94990665e-01 -6.99756682e-01 3.09518844e-01 1.31859195e+00
-4.99570817e-02 -9.18572962e-01 4.10324097e-01 5.59533000e-01
-3.65577608e-01 -3.88150096e-01 -1.55195105e+00 -7.11865604e-01
-2.49827027e-01 -3.45936447e-01 5.57786226e-01 3.09017152e-01
-3.00413580e-03 2.03207791e-01 -1.63603395e-01 9.21612322e-01
9.08042431e-01 -1.88694105e-01 1.06510198e+00 -9.35703695e-01
5.50516903e-01 -2.01972172e-01 -3.35298479e-01 -7.72273958e-01
3.28792065e-01 -7.54924119e-01 3.65466475e-01 -1.57223725e+00
-3.75537753e-01 -6.08762383e-01 3.42718959e-01 1.25780091e-01
2.75151432e-01 -5.20605862e-01 2.34870046e-01 2.14907989e-01
-3.70605886e-01 6.81749880e-01 1.36872602e+00 -1.28988773e-01
-1.83234379e-01 4.72014129e-01 -9.04327780e-02 3.93747002e-01
7.55319297e-01 -4.02501225e-01 -9.04532611e-01 8.94219577e-02
-4.16101575e-01 8.42330575e-01 4.11903560e-01 -1.11245191e+00
7.74175048e-01 -6.14619613e-01 -5.14592648e-01 -1.03873098e+00
1.26962602e-01 -1.62065351e+00 8.84297550e-01 1.10749459e+00
-4.25340533e-02 -2.14619547e-01 1.09781355e-01 8.99895489e-01
-4.27870117e-02 3.04158688e-01 1.03791130e+00 5.25717199e-01
-6.39523387e-01 6.16415441e-01 -1.30070865e+00 -3.63803387e-01
2.30611777e+00 -2.18302414e-01 -4.82594132e-01 -4.01124090e-01
-2.63285786e-01 1.40683401e+00 5.13502024e-02 5.03287077e-01
4.27140355e-01 -1.14561355e+00 -6.36178553e-01 2.66189307e-01
-7.44986683e-02 -1.83828369e-01 6.52109265e-01 1.14819467e+00
-3.23032171e-01 6.00129187e-01 -2.27683499e-01 -5.37541568e-01
-1.10098183e+00 6.10376775e-01 5.91255128e-01 2.77892351e-01
-2.19714120e-01 -1.14661075e-01 1.03276901e-01 1.63307548e-01
-1.68450654e-01 1.51500359e-01 -4.97736812e-01 1.82271019e-01
2.81929970e-01 9.80076253e-01 -3.27984333e-01 -1.34574342e+00
-5.42970240e-01 3.73142064e-01 4.51883435e-01 1.08732529e-01
8.80292356e-01 -8.65281582e-01 3.29092592e-02 7.92413112e-03
1.04058707e+00 -1.87222451e-01 -1.45304799e+00 1.36925176e-01
-4.25036699e-01 -1.73401475e-01 3.33111197e-01 -3.17459434e-01
-1.24290860e+00 4.62104946e-01 4.37160343e-01 2.63847023e-01
8.26363027e-01 -3.06087941e-01 9.48997557e-01 -1.39356375e-01
7.31742084e-01 -1.19531476e+00 -7.27377653e-01 1.36346385e-01
5.51830053e-01 -6.06313467e-01 -3.46288562e-01 -5.09622455e-01
-8.05801749e-01 9.66586769e-01 1.01182520e+00 -1.17848665e-01
8.83807480e-01 2.63975561e-01 -7.67387226e-02 -1.32753670e-01
-1.16209650e+00 -1.48838550e-01 -4.68515068e-01 5.04772663e-01
-8.35912645e-01 2.02454925e-01 -6.20538175e-01 4.34703529e-01
2.78413862e-01 9.20861289e-02 9.55093265e-01 5.74225605e-01
-8.00682545e-01 -5.12663782e-01 -3.74209851e-01 1.54066056e-01
3.70629042e-01 7.14964628e-01 5.48840344e-01 9.53229904e-01
5.97289383e-01 1.38235998e+00 2.15963155e-01 -3.66213739e-01
5.26795089e-01 -5.00628293e-01 -6.65061116e-01 -6.82440847e-02
-6.65295199e-02 -1.76933572e-01 3.13070625e-01 -3.51129144e-01
-6.09633885e-02 -7.02340961e-01 -1.68208361e+00 -5.36559463e-01
-2.55282938e-01 7.41518140e-01 7.28478193e-01 1.04991817e+00
5.13881683e-01 2.90848702e-01 1.48343062e+00 -4.45988864e-01
-2.09571883e-01 -4.72698689e-01 -7.62079358e-01 -3.15615356e-01
4.65096563e-01 -6.56465769e-01 -6.69664979e-01 -3.93954307e-01] | [5.539227485656738, 1.6179099082946777] |
028152c5-613a-4900-91c0-9bed80305ca2 | comparison-of-genres-in-word-sense | null | null | https://aclanthology.org/2020.clib-1.17 | https://aclanthology.org/2020.clib-1.17.pdf | Comparison of Genres in Word Sense Disambiguation using Automatically Generated Text Collections | The best approaches in Word Sense Disambiguation (WSD) are supervised and rely on large amounts of hand-labelled data, which is not always available and costly to create. In our work we describe an approach that is used to create an automatically labelled collection based on the monosemous relatives (related unambiguous entries) for Russian. The main contribution of our work is that we extracted monosemous relatives that can be located at relatively long distances from a target ambiguous word and ranked them according to the similarity measure to the target sense. We evaluated word sense disambiguation models based on a nearest neighbour classification on BERT and ELMo embeddings and two text collections. Our work relies on the Russian wordnet RuWordNet. | ['Natalia Loukachevitch', 'Angelina Bolshina'] | null | null | null | null | clib-2020-9 | ['word-sense-disambiguation'] | ['natural-language-processing'] | [-3.14866640e-02 -9.99256968e-03 -1.64529398e-01 -2.78736174e-01
-5.96973360e-01 -1.01579618e+00 6.28564537e-01 1.00990534e+00
-1.07664216e+00 1.28001547e+00 4.47480530e-01 -4.23252493e-01
-5.47902465e-01 -8.15729856e-01 9.39711854e-02 -3.04649711e-01
1.14719406e-01 9.10574496e-01 4.22756344e-01 -1.09105837e+00
6.72923923e-01 4.79168534e-01 -1.41287291e+00 -1.46697834e-01
1.03627181e+00 3.12768310e-01 7.42351949e-01 4.41257954e-01
-6.44818008e-01 1.81862518e-01 -9.00894046e-01 -2.25443736e-01
7.49551058e-02 -4.56609577e-02 -1.36121726e+00 -4.25102323e-01
8.33731219e-02 6.05687797e-01 6.73058704e-02 1.21874750e+00
6.87862933e-01 4.79047239e-01 7.04364538e-01 -6.35796010e-01
-8.16681206e-01 1.08272076e+00 -1.02404214e-01 6.84542537e-01
6.28292263e-01 -5.98424792e-01 1.43035722e+00 -7.40657747e-01
1.05804431e+00 1.16268325e+00 5.37398100e-01 2.99777359e-01
-1.10629284e+00 -3.19422781e-01 6.51210696e-02 4.16508883e-01
-1.77447283e+00 -1.64660454e-01 4.05751020e-01 -3.07993770e-01
1.31245363e+00 2.20956966e-01 6.12303734e-01 8.14944983e-01
-2.19948784e-01 2.17122987e-01 1.28064346e+00 -1.03503180e+00
4.32280600e-01 6.13905005e-02 8.24045539e-01 2.91716993e-01
6.62110984e-01 -2.12143332e-01 -1.90407440e-01 -1.82089791e-01
4.32302147e-01 -4.55081314e-01 -1.41976789e-01 -1.43974304e-01
-1.13025558e+00 8.46909404e-01 2.11161137e-01 1.23783624e+00
-9.28309634e-02 -5.19389868e-01 3.28070760e-01 3.59441131e-01
4.49995160e-01 1.18995523e+00 -9.31876481e-01 -7.23963305e-02
-6.77126348e-01 1.95893839e-01 9.86500442e-01 8.47989023e-01
9.14826095e-01 -4.19543058e-01 2.85656452e-02 9.61761057e-01
2.26426393e-01 4.55184311e-01 1.04742098e+00 -1.57671094e-01
2.05440819e-01 8.00450206e-01 2.46411487e-01 -1.03590870e+00
-4.90706563e-01 -1.71614021e-01 -2.67121583e-01 -1.71770126e-01
1.22421294e-01 -6.88914442e-03 -9.70140040e-01 1.47882259e+00
2.15531290e-01 1.57366365e-01 6.28723562e-01 7.15205073e-01
1.07583189e+00 3.53789210e-01 3.28556955e-01 -2.66947865e-01
1.73078394e+00 -4.21801537e-01 -8.51241350e-01 -2.48134881e-01
1.03102815e+00 -1.11218965e+00 1.01872027e+00 -5.11647086e-04
-4.32326943e-01 -3.91068459e-01 -1.28390849e+00 -7.62684084e-03
-1.30716932e+00 -1.63597479e-01 5.33847988e-01 8.12812746e-01
-9.45248723e-01 7.63225377e-01 -2.49951437e-01 -8.05390656e-01
-4.31083858e-01 2.57946372e-01 -7.01498330e-01 2.00006932e-01
-1.94600165e+00 1.33069372e+00 1.34443033e+00 -2.93596327e-01
-8.11041296e-02 -2.99880892e-01 -1.10992074e+00 -2.76554644e-01
3.37904721e-01 -4.25368696e-01 8.74976277e-01 -6.92118883e-01
-1.01739657e+00 1.35550404e+00 -1.37399808e-02 -5.79521179e-01
-1.49025887e-01 -4.62615825e-02 -1.10914111e+00 -6.09344579e-02
5.96535802e-01 2.51773357e-01 2.43169099e-01 -1.11752272e+00
-6.87889218e-01 -5.27714550e-01 4.47897241e-02 3.62733752e-01
-1.87050611e-01 2.11981013e-01 1.39312923e-01 -9.50373352e-01
2.34863892e-01 -8.09335947e-01 -4.29599375e-01 -1.05380428e+00
-1.89186811e-01 -6.46528244e-01 5.54265678e-01 -6.73933566e-01
1.45293152e+00 -1.88623953e+00 -5.25116213e-02 2.71655023e-01
2.49923125e-01 8.39735568e-01 -1.15739316e-01 7.21482754e-01
-4.66532350e-01 5.17987490e-01 -6.16208836e-02 2.09407151e-01
1.12610847e-01 6.88561022e-01 -3.86225432e-01 7.67696351e-02
-1.23618208e-01 4.93760228e-01 -1.39696825e+00 -6.91145658e-01
3.65888447e-01 7.09803104e-02 9.57484543e-02 -1.36018768e-01
7.32344165e-02 -3.27067673e-02 -5.15982866e-01 4.09216553e-01
4.57758784e-01 2.69967914e-01 5.55552244e-01 -3.47077288e-02
-1.55421659e-01 7.11861908e-01 -1.50044990e+00 1.71052742e+00
-5.07389605e-01 3.21917623e-01 -5.98621070e-01 -9.05150712e-01
1.32706952e+00 4.14908737e-01 1.56852394e-01 -4.67240274e-01
2.32748818e-02 6.84245944e-01 -1.14398770e-01 -3.20772082e-01
1.22635520e+00 -2.05420151e-01 -4.80457455e-01 8.12676549e-02
5.13706028e-01 -2.02957690e-01 7.34831691e-01 1.18898734e-01
9.68310058e-01 -5.96177056e-02 1.33294821e+00 -8.09617341e-01
6.97283506e-01 3.13423008e-01 5.74985027e-01 7.06667781e-01
-2.48767927e-01 2.67243028e-01 1.74174830e-01 -5.82373202e-01
-8.62897992e-01 -1.10300589e+00 -4.20094132e-01 1.16090226e+00
3.93512398e-01 -1.05019176e+00 -4.58333343e-01 -7.95333683e-01
-1.55926093e-01 8.74597847e-01 -3.90443265e-01 1.54322147e-01
-4.27277386e-01 -5.22385836e-01 7.96122849e-01 3.27206343e-01
-3.86688188e-02 -1.01881266e+00 -1.18641108e-01 4.14696664e-01
-6.26467988e-02 -9.83392060e-01 -2.44818151e-01 5.88237107e-01
-6.31817639e-01 -9.83887017e-01 -4.41077828e-01 -1.05907774e+00
3.25745702e-01 1.17507219e-01 1.42917013e+00 -6.37882426e-02
-3.50878924e-01 -1.39182568e-01 -8.61368418e-01 -6.00837409e-01
-2.49052733e-01 4.58134651e-01 5.32733619e-01 -5.96930742e-01
1.05596805e+00 -3.95464987e-01 4.50779311e-02 2.98227787e-01
-1.04740381e+00 -8.59015584e-01 2.20914632e-01 7.67391443e-01
6.90778017e-01 -7.43978396e-02 5.40092528e-01 -9.54486430e-01
7.84374177e-01 -4.50518727e-01 -5.45005500e-01 3.25609833e-01
-6.78254366e-01 6.72789216e-01 2.47269258e-01 -2.17598960e-01
-7.55377769e-01 1.19860679e-01 -5.00035703e-01 2.32445240e-01
-3.08161229e-01 5.28636634e-01 -1.50496304e-01 -2.69713029e-02
1.04830873e+00 -2.06086919e-01 -6.88690484e-01 -6.07883155e-01
7.54798889e-01 9.91266251e-01 2.48002127e-01 -6.54830933e-01
5.63398600e-01 -5.58046699e-02 -1.94365427e-01 -1.10074866e+00
-9.99733806e-01 -1.04576540e+00 -1.11840630e+00 1.91796184e-01
9.55159903e-01 -6.01354957e-01 -2.91997492e-01 -2.00218707e-01
-1.13602495e+00 4.96635705e-01 -5.06249428e-01 6.63636267e-01
-2.87942886e-01 7.01110125e-01 -2.03844517e-01 -7.65893936e-01
-3.87926489e-01 -5.37955940e-01 9.45904016e-01 2.26569861e-01
-7.75268793e-01 -1.23818254e+00 4.91142273e-01 1.14605695e-01
1.71177074e-01 -7.55468979e-02 9.16439056e-01 -1.59426796e+00
2.26288006e-01 -1.46534845e-01 2.97737122e-02 5.51166683e-02
3.01597744e-01 -2.97288537e-01 -7.65401542e-01 1.25693902e-02
-2.50217378e-01 -1.39396027e-01 7.93604612e-01 -8.90474115e-03
3.65176708e-01 3.06960382e-02 -6.29446626e-01 6.37051165e-02
1.53672600e+00 2.09160432e-01 5.27883649e-01 6.74829245e-01
4.51963961e-01 4.06342715e-01 9.79832113e-01 7.69282058e-02
1.22536868e-01 5.41550338e-01 -1.40006378e-01 2.67863005e-01
1.06451094e-01 -3.21772248e-01 -1.54391736e-01 1.21952200e+00
-1.62084296e-01 -2.10102126e-01 -1.20108545e+00 1.05880284e+00
-1.68520844e+00 -7.78986394e-01 -2.06898779e-01 2.19219208e+00
9.58773315e-01 1.76332444e-01 -1.97495595e-01 3.06890547e-01
1.00762510e+00 2.75012642e-01 3.20391029e-01 -5.94363868e-01
-4.44792122e-01 7.78982580e-01 8.10826421e-01 8.58651459e-01
-1.12206542e+00 1.58964849e+00 6.75448322e+00 1.20623553e+00
-6.08014584e-01 1.48634642e-01 -1.05794109e-01 5.59101939e-01
-2.43001297e-01 2.17871845e-01 -1.04811394e+00 2.60394037e-01
8.58073950e-01 -3.17909300e-01 -7.80662820e-02 8.43473196e-01
-2.89930493e-01 -6.93586394e-02 -6.00589335e-01 1.03472662e+00
1.90263003e-01 -1.01320505e+00 4.84270789e-02 -2.41183862e-01
5.03162146e-01 6.27949908e-02 -6.56319320e-01 2.00626224e-01
9.18604732e-01 -1.05669725e+00 2.19882146e-01 4.20676470e-01
6.07358098e-01 -8.65493059e-01 1.17232645e+00 5.20024747e-02
-1.35513747e+00 3.18450809e-01 -7.84546733e-01 -1.91476241e-01
5.35367914e-02 6.76428556e-01 -1.15830362e+00 9.01962578e-01
4.89966452e-01 5.81019819e-01 -5.94999790e-01 8.58634830e-01
-5.37050366e-01 3.73888791e-01 -2.96383440e-01 -4.73172605e-01
2.28713378e-01 -1.85745925e-01 8.59783113e-01 1.45897734e+00
3.71558249e-01 9.26907733e-02 4.35314864e-01 3.18628103e-01
2.35378265e-01 7.38011181e-01 -7.62414277e-01 -2.11338580e-01
7.70226479e-01 1.18114662e+00 -9.66203690e-01 -5.40207207e-01
1.48782969e-01 9.31802750e-01 3.54274124e-01 -1.10720336e-01
-4.67264950e-02 -1.01380026e+00 1.01830781e+00 -1.94304645e-01
1.66488945e-01 -3.56015384e-01 -1.38247505e-01 -1.11866713e+00
-7.61409998e-02 -4.78795052e-01 9.38396752e-01 -7.23612487e-01
-1.72509968e+00 1.18172598e+00 -6.71994016e-02 -1.10220683e+00
-3.99901479e-01 -8.86219442e-01 4.88008372e-03 1.19117427e+00
-1.41854823e+00 -8.05073977e-01 1.65319279e-01 4.07650083e-01
2.82884538e-01 -3.36939514e-01 1.45883322e+00 2.26199329e-01
-4.50277422e-03 2.33233899e-01 1.15741521e-01 2.90418118e-01
9.10674214e-01 -1.68962753e+00 5.75899065e-01 6.71629608e-01
5.72296023e-01 9.33818579e-01 8.87754500e-01 -8.78505170e-01
-4.97420162e-01 -6.82513952e-01 2.00206614e+00 -6.33571625e-01
1.24788141e+00 -4.96758111e-02 -7.22675025e-01 5.53442955e-01
2.97558039e-01 -5.03905751e-02 9.47671056e-01 4.17141080e-01
-5.40228128e-01 1.26987159e-01 -1.09620917e+00 5.68689048e-01
1.29695463e+00 -7.11908102e-01 -1.78425264e+00 2.90696025e-01
7.88969874e-01 -1.21040568e-01 -8.51035774e-01 6.42113313e-02
-1.13023572e-01 -2.22193301e-01 1.06709814e+00 -9.49089348e-01
-2.37016201e-01 -7.65441597e-01 -3.88515592e-01 -1.81064272e+00
-2.50185281e-01 -5.35395801e-01 7.00055897e-01 1.31403208e+00
5.52473605e-01 -6.44162416e-01 1.31480828e-01 -7.03723803e-02
-7.30499029e-02 7.47124851e-02 -1.03235352e+00 -9.86066401e-01
8.74013752e-02 -3.82591605e-01 8.44349623e-01 1.50206375e+00
4.75763649e-01 8.86798143e-01 3.77412178e-02 1.07998475e-01
5.49270958e-02 3.54435444e-02 -5.03484160e-02 -1.81023383e+00
7.06752986e-02 -3.33092302e-01 -1.22469842e+00 -5.62013209e-01
4.74943757e-01 -1.11544883e+00 6.56350330e-02 -1.53959727e+00
-3.87805998e-01 -5.52017093e-01 -6.28711462e-01 4.71105695e-01
-3.79937649e-01 2.60935903e-01 2.57242694e-02 -1.31488562e-01
-7.30914354e-01 1.50096968e-01 5.25717795e-01 3.57760745e-03
-2.65675932e-01 -4.54923958e-01 -7.23000884e-01 8.24376822e-01
6.72684789e-01 -8.25908542e-01 -1.38797760e-01 -2.32194975e-01
6.42709613e-01 -3.48097205e-01 -1.84108973e-01 -8.20724547e-01
3.66544724e-02 -1.49122104e-01 1.03803068e-01 -3.45237523e-01
7.49449432e-02 -8.36960077e-01 -1.51254043e-01 2.56795496e-01
-2.58561134e-01 4.20047283e-01 2.37372518e-03 4.13889706e-01
-3.69239807e-01 -9.00886774e-01 4.88419414e-01 -2.98671603e-01
-1.38713288e+00 -2.69485325e-01 -3.52755040e-01 4.23745364e-01
1.03285372e+00 5.28618740e-03 -7.85279125e-02 1.73850134e-01
-9.39076722e-01 -1.98801249e-01 4.79332924e-01 6.95203483e-01
2.45727941e-01 -1.38420546e+00 -5.86161196e-01 -1.95186079e-01
7.24209607e-01 -6.51525736e-01 -2.48093098e-01 -9.79649182e-03
-6.28804922e-01 5.44243991e-01 -3.56968135e-01 -9.54625476e-03
-1.46262121e+00 8.81888926e-01 4.77256849e-02 -3.19788337e-01
-3.80350769e-01 6.76968336e-01 -7.13771582e-01 -9.17968512e-01
-1.50025934e-01 -2.76614994e-01 -1.02176774e+00 4.65925753e-01
7.14848161e-01 1.55561581e-01 4.08466339e-01 -1.07194006e+00
-5.73633075e-01 5.40181279e-01 2.22700257e-02 -2.90153384e-01
1.05386758e+00 -1.51652887e-01 -4.56582189e-01 6.94056392e-01
7.87500739e-01 4.70313281e-01 1.79741323e-01 -4.57922608e-01
8.01451564e-01 -5.17458439e-01 -2.04019889e-01 -6.73726320e-01
-3.51918578e-01 3.35744262e-01 7.01875687e-01 5.96783757e-01
6.51627779e-01 1.72116354e-01 6.58127367e-01 8.29427004e-01
5.52430689e-01 -1.51896369e+00 -7.53238022e-01 1.16813827e+00
3.51332545e-01 -1.04329085e+00 -2.46917412e-01 -5.85959196e-01
-5.84739387e-01 1.05928731e+00 1.11595541e-01 -1.56422928e-01
7.59015501e-01 1.73591629e-01 3.95687848e-01 -1.74403876e-01
-1.60914212e-01 -9.55020905e-01 4.09044445e-01 1.01050901e+00
7.14254498e-01 3.49410802e-01 -1.41515136e+00 7.26369441e-01
-5.84106505e-01 -3.67853522e-01 4.60644394e-01 9.17124033e-01
-6.46507561e-01 -1.64300478e+00 -4.01113808e-01 1.84560299e-01
-4.89524961e-01 -6.36561751e-01 -8.07395816e-01 7.11461306e-01
2.82230020e-01 1.15279758e+00 -6.47346675e-02 -3.46355796e-01
2.58808821e-01 3.25884968e-01 1.59785479e-01 -9.28103209e-01
-4.31959361e-01 -4.65247780e-01 5.20783484e-01 -1.42777413e-01
-5.33145368e-01 -3.69562358e-01 -1.34455848e+00 -1.14509292e-01
-3.73502970e-01 8.38492513e-01 6.65565789e-01 1.33348024e+00
6.40446842e-02 2.53390312e-01 2.16581225e-01 -3.88898492e-01
-2.67743826e-01 -1.09773147e+00 -1.03251600e+00 6.35913014e-01
-1.18023632e-02 -7.14124799e-01 -1.92782432e-01 -1.17048487e-01] | [10.241291999816895, 9.288840293884277] |
afa6c9cf-3e6b-4392-959d-50d4d9310771 | bidirectional-transformer-reranker-for | 2305.13000 | null | https://arxiv.org/abs/2305.13000v1 | https://arxiv.org/pdf/2305.13000v1.pdf | Bidirectional Transformer Reranker for Grammatical Error Correction | Pre-trained seq2seq models have achieved state-of-the-art results in the grammatical error correction task. However, these models still suffer from a prediction bias due to their unidirectional decoding. Thus, we propose a bidirectional Transformer reranker (BTR), that re-estimates the probability of each candidate sentence generated by the pre-trained seq2seq model. The BTR preserves the seq2seq-style Transformer architecture but utilizes a BERT-style self-attention mechanism in the decoder to compute the probability of each target token by using masked language modeling to capture bidirectional representations from the target context. For guiding the reranking, the BTR adopts negative sampling in the objective function to minimize the unlikelihood. During inference, the BTR gives final results after comparing the reranked top-1 results with the original ones by an acceptance threshold. Experimental results show that, in reranking candidates from a pre-trained seq2seq model, T5-base, the BTR on top of T5-base could yield 65.47 and 71.27 F0.5 scores on the CoNLL-14 and BEA test sets, respectively, and yield 59.52 GLEU score on the JFLEG corpus, with improvements of 0.36, 0.76 and 0.48 points compared with the original T5-base. Furthermore, when reranking candidates from T5-large, the BTR on top of T5-base improved the original T5-large by 0.26 points on the BEA test set. | ['Manabu Okumura', 'Hidetaka Kamigaito', 'Ying Zhang'] | 2023-05-22 | null | null | null | null | ['grammatical-error-correction'] | ['natural-language-processing'] | [ 3.07355195e-01 3.35786253e-01 1.19264098e-02 -4.93046284e-01
-1.16869271e+00 -2.90513188e-01 3.00458610e-01 1.30217791e-01
-6.95565879e-01 9.23362911e-01 2.97313869e-01 -4.02601063e-01
3.23778719e-01 -6.86561465e-01 -9.88919616e-01 -4.79261011e-01
2.64086962e-01 3.89697373e-01 1.35400578e-01 -4.33831692e-01
3.16215277e-01 -1.67467117e-01 -1.28672481e+00 7.22984254e-01
1.42078328e+00 7.09116340e-01 6.48458600e-01 8.23688149e-01
-7.69132841e-03 8.26917768e-01 -8.29324901e-01 -8.35856438e-01
-9.82270092e-02 -7.38724709e-01 -6.60725236e-01 -8.22035611e-01
4.52767193e-01 -3.80631387e-01 -4.12536323e-01 1.24654007e+00
6.67540014e-01 -9.82066840e-02 3.25870246e-01 -6.39842868e-01
-5.75702012e-01 1.33563113e+00 -4.03971106e-01 1.18325368e-01
3.29511702e-01 -3.54788229e-02 1.42954135e+00 -1.05668867e+00
4.65310484e-01 1.31076109e+00 5.82130194e-01 8.31871986e-01
-1.14990020e+00 -8.01159680e-01 2.29354337e-01 3.00015032e-01
-1.36442971e+00 -3.77139419e-01 3.07894528e-01 -6.85037524e-02
1.32841802e+00 4.29105371e-01 5.71549416e-01 9.66863513e-01
5.24687469e-01 1.09810007e+00 9.22101676e-01 -5.20110309e-01
5.13933375e-02 -4.04704332e-01 2.25183338e-01 7.37630546e-01
-1.44187210e-03 6.18662164e-02 -1.01361239e+00 9.15083960e-02
2.60028899e-01 -5.48872769e-01 -1.95422575e-01 6.01658404e-01
-1.14141786e+00 5.98006845e-01 2.18924642e-01 2.66895562e-01
-3.04616451e-01 3.50881279e-01 4.77177799e-01 3.11215699e-01
4.60389674e-01 4.51144785e-01 -6.81741059e-01 -5.58425903e-01
-9.21667933e-01 3.02781522e-01 4.31634724e-01 1.01082575e+00
3.94562572e-01 1.48659825e-01 -9.88100171e-01 9.89867985e-01
3.67948264e-01 7.72272587e-01 5.37883461e-01 -7.00859427e-01
7.91270137e-01 4.48747396e-01 -1.74108416e-01 -4.74588871e-01
-6.17571287e-02 -6.92069948e-01 -7.87071884e-01 -3.40553552e-01
3.37693751e-01 -8.07562494e-04 -1.02075040e+00 2.02260923e+00
-1.59035966e-01 -1.91772848e-01 2.07044929e-01 7.27954388e-01
7.06372559e-01 1.02609515e+00 5.77125624e-02 -6.82681054e-02
1.22254944e+00 -9.52176869e-01 -5.78953624e-01 -3.85947078e-01
9.08686280e-01 -7.42607117e-01 1.25169325e+00 6.21491849e-01
-1.33480990e+00 -5.46612620e-01 -9.27670658e-01 -1.94999948e-01
3.87656033e-01 4.78155911e-01 5.26677482e-02 3.66492271e-01
-1.03037620e+00 6.85252547e-01 -7.68965065e-01 -1.50925070e-01
9.13225561e-02 1.07745171e-01 -1.53087705e-01 -3.34379882e-01
-1.50001299e+00 9.31681275e-01 4.41261858e-01 2.50587702e-01
-1.05501604e+00 -7.96368003e-01 -6.88896835e-01 2.47612327e-01
-2.42634844e-02 -4.93454278e-01 1.41337335e+00 -7.77274489e-01
-1.66040492e+00 7.89567828e-01 -6.67860627e-01 -5.29613972e-01
5.61214507e-01 -5.85181713e-01 -2.83429176e-01 -2.23614931e-01
3.00394237e-01 6.49666607e-01 3.32051277e-01 -7.05864310e-01
-8.37212563e-01 -2.36075431e-01 -1.54555276e-01 1.72115386e-01
-6.30397648e-02 2.25683182e-01 -5.50716698e-01 -8.59078407e-01
-7.37546533e-02 -8.71933460e-01 2.99701653e-02 -6.26118839e-01
-7.82176375e-01 -3.56542975e-01 8.64626914e-02 -1.01411927e+00
1.66338217e+00 -2.05968523e+00 2.39340454e-01 -6.39626905e-02
-8.74056965e-02 5.24042249e-01 -5.68761945e-01 3.58782351e-01
1.33491429e-02 2.88053274e-01 -2.64889121e-01 -3.20567608e-01
-7.67383426e-02 1.39007285e-01 -2.29292437e-01 1.20357633e-01
3.09294999e-01 8.62105370e-01 -1.15177000e+00 -1.91276819e-01
-2.89535195e-01 1.64891988e-01 -9.41940904e-01 5.03056288e-01
-2.55048215e-01 3.21039617e-01 -5.26220314e-02 2.72092074e-01
5.44965148e-01 2.12996870e-01 4.37030613e-01 2.48305928e-02
-2.50506729e-01 1.03992784e+00 -7.52208829e-01 1.62562788e+00
-5.25025725e-01 4.69388008e-01 -2.61628121e-01 -6.34144247e-01
1.03454173e+00 2.35976294e-01 -1.81796327e-01 -1.00127530e+00
7.75149465e-03 5.73553562e-01 5.62856078e-01 -1.93583086e-01
8.20622146e-01 -8.50521252e-02 -2.98009455e-01 2.09000826e-01
1.19397879e-01 1.41802281e-01 4.13064659e-01 2.75084704e-01
1.24311459e+00 2.66653597e-01 1.88057378e-01 -2.01720387e-01
6.41152859e-01 -3.26189518e-01 1.05004013e+00 9.28574860e-01
1.62785888e-01 7.74949849e-01 6.79995060e-01 -1.71115190e-01
-9.59984124e-01 -1.04621387e+00 -2.18618773e-02 1.12891638e+00
-2.47753426e-01 -7.07312226e-01 -7.47570157e-01 -9.40526307e-01
-1.31552458e-01 1.42080712e+00 -4.05583680e-01 -3.93774211e-01
-8.89640212e-01 -8.38084698e-01 1.02813363e+00 4.19376433e-01
3.39563429e-01 -9.50877011e-01 -1.20728433e-01 3.31901878e-01
-6.26668453e-01 -7.98447430e-01 -6.58460319e-01 2.19591901e-01
-7.46133626e-01 -6.66978359e-01 -6.68439984e-01 -6.42449796e-01
7.10140586e-01 -3.08704644e-01 1.13050103e+00 3.25298190e-01
3.15201223e-01 -5.32702029e-01 -6.89841986e-01 -2.37112373e-01
-7.70916820e-01 2.43792728e-01 -9.75865871e-02 -1.83208063e-01
2.21048474e-01 -1.33346841e-01 -1.19422063e-01 1.72038957e-01
-4.19271320e-01 2.57455349e-01 5.79753458e-01 1.30982661e+00
5.93298376e-01 -5.64783931e-01 7.14489579e-01 -8.20025921e-01
3.17213356e-01 -2.15996459e-01 -5.17454863e-01 3.89690489e-01
-7.65585780e-01 3.00664783e-01 7.80815721e-01 -1.23018004e-01
-1.07164526e+00 -4.19759512e-01 -5.56127250e-01 3.38730626e-02
3.17682713e-01 5.92607498e-01 -4.14087474e-01 5.13708949e-01
2.40536168e-01 4.86690998e-01 -3.31880063e-01 -5.23026884e-01
2.16256425e-01 7.63321936e-01 3.67339522e-01 -5.66527903e-01
2.90433824e-01 -4.60003436e-01 -4.36641693e-01 -4.20784533e-01
-1.00046134e+00 1.20078757e-01 -3.06716055e-01 -3.69181260e-02
6.67439342e-01 -9.14839387e-01 -5.04683137e-01 7.17347503e-01
-1.64751554e+00 -3.74891937e-01 -1.56243756e-01 5.77545822e-01
-3.34180117e-01 2.07751319e-01 -6.87900901e-01 -6.61386132e-01
-6.32290483e-01 -1.21065676e+00 9.78363037e-01 -6.48487825e-03
-3.90966505e-01 -4.45615381e-01 1.85601898e-02 3.97848755e-01
2.17323929e-01 -5.33935189e-01 1.28737450e+00 -8.83039117e-01
-4.36839402e-01 -3.87815237e-02 -8.86866301e-02 6.93093896e-01
-2.02146620e-01 -1.23234890e-01 -9.03151512e-01 -2.12867826e-01
-3.63720000e-01 -1.21149421e-01 9.68111157e-01 1.15283035e-01
1.18101001e+00 -3.98031563e-01 -1.94021612e-02 5.78008652e-01
1.06883907e+00 2.78271049e-01 8.30181599e-01 1.36168927e-01
6.34503365e-01 2.53922641e-01 7.72290111e-01 2.67323911e-01
6.05779946e-01 7.02764690e-01 3.18341702e-01 2.74937421e-01
-2.91000187e-01 -7.80059338e-01 1.03529382e+00 1.41384006e+00
1.33649379e-01 -6.27288163e-01 -8.38363826e-01 6.43310666e-01
-1.60836625e+00 -8.43442738e-01 -3.99919689e-01 2.32775736e+00
1.33579063e+00 2.68172085e-01 -3.51122677e-01 9.21213776e-02
8.71553421e-01 -4.88249287e-02 -3.67005020e-01 -7.16187656e-01
-2.65427798e-01 2.63296276e-01 3.76279712e-01 7.89049447e-01
-4.81643915e-01 1.27956986e+00 5.83365202e+00 1.12466407e+00
-1.13680887e+00 1.04582883e-01 6.93781018e-01 -2.26552039e-01
-7.09977329e-01 6.31917268e-02 -1.27331471e+00 7.19761729e-01
1.17600501e+00 -1.98819950e-01 4.96163189e-01 3.95635784e-01
3.94675493e-01 -1.88966125e-01 -1.16983342e+00 7.02109039e-01
2.47365445e-01 -9.37096000e-01 2.56337672e-01 -1.95323393e-01
7.11668372e-01 1.22401953e-01 -1.83597967e-01 7.48022735e-01
3.90113264e-01 -9.73253846e-01 1.13274121e+00 5.88913023e-01
8.18452835e-01 -7.75227845e-01 1.02227592e+00 4.57838058e-01
-7.43485510e-01 4.36556339e-02 -3.80129158e-01 -1.78074405e-01
3.72842014e-01 9.64077771e-01 -9.60488200e-01 5.76323688e-01
5.92020869e-01 5.77374458e-01 -3.94908488e-01 7.52981961e-01
-8.22271228e-01 1.31664121e+00 -2.15950869e-02 -3.15134615e-01
-1.10138627e-02 -5.34598455e-02 7.54990697e-01 1.34050608e+00
5.92761874e-01 -4.11866419e-02 -2.18302876e-01 7.48057008e-01
-4.91186798e-01 1.44095212e-01 -1.38050094e-01 7.51582533e-02
7.06489682e-01 9.00782108e-01 2.68588569e-02 -4.83339697e-01
-2.34036576e-02 1.08672404e+00 7.58600831e-01 1.34889290e-01
-9.45826530e-01 -7.75956810e-01 5.73085904e-01 -9.08266660e-03
3.65583301e-01 1.38969705e-01 -2.98783004e-01 -1.08636737e+00
2.76404265e-02 -1.07590485e+00 3.27491872e-02 -8.66825938e-01
-1.14033103e+00 8.61099541e-01 -2.84742028e-01 -9.51494813e-01
-2.40438029e-01 -4.42350090e-01 -5.10485411e-01 1.16718364e+00
-1.46321964e+00 -9.04056251e-01 2.92088211e-01 -1.64459258e-01
5.81179857e-01 -2.59930909e-01 7.81104445e-01 5.55215359e-01
-7.18819082e-01 1.20977008e+00 2.01363951e-01 2.92662293e-01
7.16550827e-01 -1.36960948e+00 5.65481365e-01 1.13781130e+00
-1.17972396e-01 7.51515269e-01 4.66405869e-01 -7.74790883e-01
-1.25885844e+00 -1.31559348e+00 1.91528606e+00 -1.93029493e-01
3.81778896e-01 -3.92083645e-01 -9.05426383e-01 6.51372313e-01
1.50734231e-01 -3.23206961e-01 4.44153637e-01 1.64886326e-01
-2.76675880e-01 -3.29267919e-01 -9.72129226e-01 6.32595778e-01
1.04469144e+00 -4.74411964e-01 -7.67235637e-01 -3.72024328e-02
7.93244243e-01 -5.87596238e-01 -7.48813808e-01 3.58468682e-01
5.96582294e-01 -8.54840398e-01 4.61200178e-01 -5.43286800e-01
8.64844918e-01 -2.32114166e-01 -3.69466037e-01 -1.65175724e+00
-3.89771312e-01 -5.19166350e-01 1.69110268e-01 1.53681016e+00
8.46777320e-01 -4.63216543e-01 4.22571182e-01 9.89401489e-02
-6.44426823e-01 -8.75187516e-01 -1.20954013e+00 -8.49700093e-01
2.84745872e-01 -5.41192651e-01 5.88023186e-01 4.06702608e-01
4.20195225e-04 4.13748413e-01 -4.17704403e-01 2.52518468e-02
3.01561445e-01 -2.04211876e-01 4.01319802e-01 -7.53465652e-01
-3.23169470e-01 -2.67273515e-01 -6.36593252e-02 -1.50313950e+00
3.29165637e-01 -1.41967082e+00 6.20653749e-01 -1.45177436e+00
2.81923413e-01 -4.44737911e-01 -4.34445947e-01 6.13375604e-01
-7.11676776e-01 9.66822952e-02 4.80271637e-01 -1.91024616e-01
-3.77388060e-01 9.53045249e-01 1.18292975e+00 -1.87765837e-01
6.83983415e-02 -2.13297769e-01 -6.84041858e-01 5.72892368e-01
8.00187886e-01 -8.51243138e-01 6.73348457e-02 -9.70466018e-01
4.46390241e-01 3.92128341e-02 1.66971684e-02 -7.16405094e-01
-6.33531064e-02 1.05308577e-01 1.58813149e-01 -7.10019767e-01
4.23538983e-02 -1.36665662e-03 -2.35648483e-01 6.24823868e-01
-7.10214198e-01 3.60841930e-01 1.47467852e-01 2.48274371e-01
-2.87732065e-01 -5.49824297e-01 5.92525959e-01 -1.29351765e-02
-3.41953367e-01 -2.15488002e-01 -6.45419061e-01 5.01531482e-01
4.60607708e-01 9.46109742e-02 -2.20553666e-01 -1.12485602e-01
-4.07379687e-01 3.00602347e-01 1.27230018e-01 3.47768635e-01
5.93224943e-01 -1.23338008e+00 -1.25635147e+00 3.24000359e-01
3.30473781e-02 -1.49664907e-02 1.72862336e-01 7.61081338e-01
-3.02081198e-01 3.88275594e-01 1.61576897e-01 -3.83956850e-01
-1.35296369e+00 -2.42497116e-01 2.18088821e-01 -5.22532761e-01
-2.72008836e-01 1.17759013e+00 7.93289319e-02 -8.83456230e-01
1.16997793e-01 -5.28654933e-01 -3.88920680e-02 -1.06766544e-01
4.63906825e-01 5.48412502e-01 3.74633163e-01 -5.45425296e-01
-5.14310539e-01 2.68794715e-01 -1.85427696e-01 -2.38824919e-01
1.20384884e+00 5.58735020e-02 -3.78870666e-01 3.64309490e-01
9.72851157e-01 4.68781650e-01 -7.17321992e-01 3.31946090e-02
-3.40500399e-02 -3.17063510e-01 2.94182776e-03 -1.32458615e+00
-9.31104064e-01 1.01380754e+00 9.87756103e-02 -4.56963003e-01
8.61640930e-01 -2.56244838e-01 9.35073853e-01 2.62167364e-01
2.87112951e-01 -8.92327011e-01 -1.16138533e-01 1.40620673e+00
9.75139737e-01 -6.85310364e-01 -4.65078115e-01 -2.24049374e-01
-7.85855711e-01 9.50121105e-01 8.64941001e-01 -3.54802614e-04
1.01032376e-01 9.12273377e-02 -2.57757753e-02 3.52115571e-01
-1.10329008e+00 1.69002861e-01 1.85259178e-01 1.28271967e-01
9.00564790e-01 1.54851705e-01 -8.11786652e-01 1.01117897e+00
-4.89506245e-01 -1.11401342e-01 5.25950611e-01 5.37544906e-01
-3.97095144e-01 -1.33378243e+00 -2.15470195e-01 4.96557653e-01
-5.97980797e-01 -6.95762932e-01 -9.86002088e-02 2.73169428e-01
1.76595062e-01 9.69027519e-01 1.71262771e-01 -5.85145652e-01
4.66171563e-01 2.21935794e-01 5.06023586e-01 -7.12257266e-01
-9.51595426e-01 -4.21393663e-02 4.91440684e-01 -4.54217046e-01
2.87688732e-01 -7.41281271e-01 -1.46769321e+00 -1.66828141e-01
-5.66650271e-01 3.86825979e-01 4.10066068e-01 9.64119077e-01
4.88080502e-01 7.91846931e-01 5.06841481e-01 -3.38874400e-01
-8.87879729e-01 -1.39538360e+00 -3.14452797e-01 5.07917292e-02
1.73574522e-01 -1.74678400e-01 -9.41220522e-02 -2.49870107e-01] | [11.189090728759766, 10.523324966430664] |
17f2f628-944a-4791-a275-2586b913d6f8 | an-adaptive-volatility-method-for | 2303.01855 | null | https://arxiv.org/abs/2303.01855v1 | https://arxiv.org/pdf/2303.01855v1.pdf | An adaptive volatility method for probabilistic forecasting and its application to the M6 financial forecasting competition | In this note, we address the problem of probabilistic forecasting using an adaptive volatility method based on classical time-varying volatility models and stochastic optimization algorithms. These principles were successfully applied in the recent M6 financial forecasting competition for both probabilistic forecasting and investment decision-making under the team named AdaGaussMC. The key points of our strategy are: (a) apply a univariate time-varying volatility model, called AdaVol, (b) obtain probabilistic forecasts of future returns, and (c) optimize the competition metrics using stochastic gradient-based algorithms. We claim that the frugality of the methods implies its robustness and consistency. | ['Nicklas Werge', 'Joseph de Vilmarest'] | 2023-03-03 | null | null | null | null | ['stochastic-optimization'] | ['methodology'] | [-4.25067514e-01 -2.65630633e-01 3.11120510e-01 -5.30056596e-01
-6.78398192e-01 -8.89537394e-01 1.07360756e+00 -2.38106221e-01
-2.07562700e-01 7.51034617e-01 7.74472654e-02 -7.36648202e-01
-5.12727976e-01 -8.02635074e-01 -1.68271616e-01 -9.51801062e-01
-4.90552783e-02 7.76904106e-01 8.08919445e-02 -2.70739317e-01
8.24942112e-01 8.04387271e-01 -1.44839025e+00 -5.81630245e-02
7.15496540e-01 1.45149744e+00 -1.73101112e-01 9.17692840e-01
-5.10777712e-01 9.29947674e-01 -3.15958142e-01 -9.72935557e-01
4.92124945e-01 -9.21466574e-02 -4.04918462e-01 -3.09344143e-01
-3.34919780e-01 5.73969148e-02 3.78894694e-02 9.03515339e-01
2.11261690e-01 4.30938154e-01 9.66800988e-01 -1.20771396e+00
-2.17928126e-01 8.32533121e-01 -4.16184306e-01 6.23321176e-01
-2.35033445e-02 5.84214963e-02 9.04864371e-01 -7.84184277e-01
4.45875943e-01 1.27444410e+00 6.49433613e-01 3.77839625e-01
-1.00939655e+00 -2.82293648e-01 2.33692914e-01 7.63765946e-02
-1.21603084e+00 -2.72514910e-01 8.27705860e-01 -7.90930271e-01
6.18895888e-01 4.78763014e-01 5.08979976e-01 8.04214239e-01
8.52984905e-01 5.32628357e-01 1.36220181e+00 -2.42592007e-01
6.54453635e-01 2.76477963e-01 1.46290883e-01 5.02965413e-02
1.41879320e-01 6.87132835e-01 -4.16859984e-01 -4.73346829e-01
1.42502248e-01 1.18103571e-01 2.76319057e-01 -1.23071268e-01
-7.22258866e-01 1.26255262e+00 -1.80550292e-01 4.11985546e-01
-7.72759974e-01 3.41838449e-01 2.82635480e-01 4.86359984e-01
9.05510366e-01 4.15237337e-01 -4.44026321e-01 -4.40945059e-01
-1.10833299e+00 6.84322774e-01 1.08020866e+00 4.67485368e-01
1.51662016e-02 6.17143273e-01 -2.61370331e-01 1.35950387e-01
7.41983116e-01 7.25744486e-01 5.43930590e-01 -1.09726918e+00
3.25635582e-01 -3.29925954e-01 4.40083832e-01 -1.04002726e+00
-3.21535498e-01 -3.84876996e-01 -3.87928277e-01 5.95195234e-01
4.46116447e-01 -4.27545458e-01 -6.10162497e-01 1.35332167e+00
2.02755526e-01 1.19716980e-01 1.82987511e-01 3.67038786e-01
1.10631265e-01 7.31069148e-01 8.81149471e-02 -4.41717207e-01
7.70008564e-01 -3.42079163e-01 -6.68015361e-01 3.45721781e-01
4.95018899e-01 -9.15309012e-01 2.93255299e-01 5.69804430e-01
-9.91131723e-01 -6.54457435e-02 -9.85306740e-01 7.29673147e-01
-5.13447762e-01 -5.69917321e-01 8.35938036e-01 1.18612254e+00
-1.19024730e+00 9.12041605e-01 -8.24681401e-01 4.33225840e-01
-1.78257585e-01 4.40162374e-03 3.92189324e-01 4.74013299e-01
-1.18227983e+00 1.04061770e+00 3.36736739e-01 3.26011240e-01
-7.60087609e-01 -4.01852161e-01 -4.47488248e-01 1.07096173e-01
1.31637543e-01 -5.50677121e-01 1.20733213e+00 -6.59684300e-01
-1.94969308e+00 3.82472008e-01 -2.14102548e-02 -7.28077352e-01
9.84529078e-01 -1.46425273e-02 -5.40633202e-01 -2.78829068e-01
-2.78435409e-01 -2.84202963e-01 8.99938166e-01 -8.47889602e-01
-9.13396418e-01 -3.00678968e-01 -3.86816144e-01 -1.02141321e-01
2.21771911e-01 3.47270966e-01 3.29729587e-01 -7.75737882e-01
3.47069502e-01 -9.09639239e-01 -4.43304002e-01 -9.15640652e-01
-2.17320293e-01 -1.39358804e-01 4.57214087e-01 -5.94212770e-01
1.15546012e+00 -1.85014844e+00 -1.33643180e-01 7.98937082e-01
6.37539998e-02 -3.37097943e-01 3.47076565e-01 4.49492335e-01
-1.26570016e-01 2.98658341e-01 -4.06687915e-01 -3.67131919e-01
3.63364458e-01 -1.07953422e-01 -1.19858611e+00 3.50645959e-01
-3.28788385e-02 9.40458477e-01 -7.07943678e-01 -1.04138866e-01
2.73258001e-01 7.39871189e-02 -1.35426775e-01 -1.29468013e-02
-2.61737078e-01 1.78384259e-01 -5.53934038e-01 6.71052575e-01
6.62803352e-01 1.55876666e-01 2.01348051e-01 4.67966050e-01
-5.62703431e-01 1.81062728e-01 -1.36156309e+00 6.99918389e-01
-2.72452772e-01 4.77526218e-01 -2.08907470e-01 -9.14878786e-01
1.05776322e+00 2.57453322e-01 5.23884237e-01 -2.13300839e-01
1.46792177e-02 5.94608843e-01 -2.85243481e-01 1.47119418e-01
6.41356707e-01 -5.58522165e-01 -1.51437134e-01 7.30533481e-01
-6.56558350e-02 -6.99532509e-01 1.74197316e-01 1.12652473e-01
5.87084770e-01 1.97339490e-01 -3.24698985e-02 -6.67240381e-01
3.19094062e-01 -2.15891823e-01 5.66227853e-01 9.29548740e-01
-7.16777071e-02 4.67625350e-01 7.15261042e-01 -6.74458027e-01
-9.01454747e-01 -1.08562315e+00 -3.96523565e-01 7.95118332e-01
-5.93189061e-01 2.96872687e-02 -2.08407715e-01 -4.62293267e-01
3.84876460e-01 1.39412010e+00 -8.73654246e-01 2.71985501e-01
-1.74737111e-01 -1.17624843e+00 -8.33355933e-02 2.72046804e-01
-2.14699596e-01 -8.68074417e-01 -6.06319249e-01 5.12732863e-01
4.86463130e-01 -3.63314331e-01 -2.81841606e-01 3.61547559e-01
-9.28969741e-01 -5.91170073e-01 -6.95363164e-01 3.66276473e-01
-3.17959189e-02 -1.07785769e-01 1.25991273e+00 -3.57438475e-01
3.18497062e-01 6.48274839e-01 -1.40987277e-01 -8.65822196e-01
-3.98377478e-01 -3.56464446e-01 2.43155703e-01 3.11487913e-01
1.04628772e-01 -7.18069553e-01 -2.62089878e-01 1.64693132e-01
-7.07072496e-01 -7.20230460e-01 -1.38365394e-02 7.53739178e-01
5.49169123e-01 1.45043284e-01 4.71391827e-01 -9.30320740e-01
8.77205908e-01 -6.57884419e-01 -1.45429063e+00 5.00097156e-01
-1.39094043e+00 2.08419040e-01 1.61875650e-01 -1.76019982e-01
-1.22976649e+00 -3.00100625e-01 8.90902951e-02 -3.65883738e-01
5.36677897e-01 7.95716465e-01 3.65564227e-01 -1.38165638e-01
2.84110665e-01 1.72658190e-01 -2.58862913e-01 -2.85779089e-01
4.01879609e-01 1.99822590e-01 4.25617039e-01 -6.64305449e-01
8.21600139e-01 5.62336624e-01 2.72687852e-01 -2.31882796e-01
-6.88679874e-01 -8.52993578e-02 -2.72814363e-01 -3.68673116e-01
7.29520917e-01 -5.56229949e-01 -7.79615045e-01 5.54982781e-01
-1.00878501e+00 -8.04162957e-03 -4.51564014e-01 7.50161111e-01
-6.59577727e-01 1.49401560e-01 -3.80567104e-01 -1.77594864e+00
-2.56281883e-01 -1.02219296e+00 6.35565042e-01 2.99450219e-01
-8.40155780e-03 -1.51015270e+00 5.10284185e-01 -1.93049312e-01
8.77118468e-01 5.05053878e-01 5.96571386e-01 -1.14178622e+00
-4.49615538e-01 -6.83915436e-01 1.54824406e-01 2.34463185e-01
-2.40348533e-01 7.80775189e-01 -7.95789838e-01 1.83085188e-01
5.99354804e-01 1.25393391e-01 8.50349844e-01 8.01905930e-01
8.77311885e-01 -3.51867005e-02 -7.77536854e-02 8.55561495e-01
1.24249411e+00 5.78250408e-01 4.34809059e-01 5.71105957e-01
2.70328462e-01 1.00221062e+00 5.39325297e-01 8.41180742e-01
3.48821610e-01 4.79507983e-01 2.49857277e-01 7.28169918e-01
8.09076190e-01 -9.50853527e-02 4.82829779e-01 1.15348077e+00
-1.62248567e-01 -1.50954425e-01 -1.19381285e+00 2.65779257e-01
-1.94137132e+00 -1.45320356e+00 -1.60477042e-01 2.15293932e+00
3.58120710e-01 3.91934663e-01 1.42229721e-01 -3.69802386e-01
6.16937280e-01 3.11145872e-01 -3.08907986e-01 -8.50378990e-01
-4.77501988e-01 2.21125316e-02 7.26154387e-01 6.47467017e-01
-9.21001673e-01 6.02063179e-01 7.89247417e+00 1.02453232e+00
-1.07953990e+00 4.06281576e-02 1.28184462e+00 -1.69879168e-01
-1.05179799e+00 2.05487490e-01 -1.03773713e+00 8.85561585e-01
1.48002279e+00 -9.03812826e-01 3.55563998e-01 1.07955837e+00
-5.00278035e-03 -1.23026215e-01 -6.12994075e-01 7.38767803e-01
-3.00358444e-01 -1.59165728e+00 -1.19861245e-01 4.38062474e-02
9.29327071e-01 6.79687411e-02 4.41370875e-01 4.87608463e-01
9.40630913e-01 -1.12205446e+00 9.62308049e-01 1.39620829e+00
1.78680286e-01 -1.17527711e+00 9.89782035e-01 9.98715460e-02
-9.10490572e-01 -2.07544148e-01 -4.26320642e-01 1.04309924e-01
5.07374763e-01 1.16656470e+00 -2.57819831e-01 6.28748238e-01
6.82311893e-01 1.07993208e-01 1.19469492e-02 9.30378258e-01
-8.26468691e-02 7.33761549e-01 -5.99378288e-01 -3.24152619e-01
3.55159849e-01 -8.93466413e-01 8.35587740e-01 9.83782530e-01
6.75480962e-01 9.38312635e-02 -2.96553075e-01 8.88017595e-01
4.32450175e-01 -1.06505960e-01 -5.13187587e-01 -3.68244827e-01
3.24547976e-01 1.07645130e+00 -7.91119158e-01 -3.19296896e-01
-1.12705246e-01 2.35062614e-01 -2.50059932e-01 2.84699440e-01
-6.07470751e-01 -2.74138868e-01 3.92462164e-01 -2.47674957e-01
4.86181855e-01 -3.92425537e-01 -9.34375823e-01 -1.04434669e+00
-1.07617266e-01 -5.71350157e-01 4.85275030e-01 -4.43376243e-01
-1.48270893e+00 7.16320157e-01 9.04461294e-02 -1.10610807e+00
-9.80286658e-01 -6.73117161e-01 -1.08992052e+00 1.36861217e+00
-1.13426030e+00 -4.18261498e-01 5.64361930e-01 1.92409366e-01
1.75721392e-01 -7.33132541e-01 4.71545190e-01 -4.63978231e-01
-3.61961067e-01 1.99354485e-01 9.08035517e-01 -6.19904518e-01
1.01533577e-01 -1.57460344e+00 7.51498222e-01 9.18893278e-01
1.66062623e-01 6.10628724e-01 1.18113434e+00 -7.75811374e-01
-1.24971330e+00 -6.91728234e-01 1.10040283e+00 -9.47150826e-01
1.26986241e+00 -9.04757231e-02 -6.75737798e-01 3.06724846e-01
-1.80080056e-01 -2.81249642e-01 4.55928773e-01 1.64700314e-01
-1.51470602e-01 -1.65825084e-01 -1.23816156e+00 5.01826286e-01
1.32452771e-01 -4.17387605e-01 -7.12156773e-01 5.00433743e-01
6.29012883e-01 -8.50113407e-02 -7.35027671e-01 3.47766697e-01
6.86612248e-01 -1.00948989e+00 6.14050627e-01 -7.47568071e-01
2.23492205e-01 7.28697926e-02 -5.03330827e-01 -1.23956394e+00
-1.61039889e-01 -1.23672378e+00 -2.96992093e-01 1.08263588e+00
6.19121969e-01 -1.32175934e+00 4.64659125e-01 1.14487076e+00
1.47970572e-01 -7.08102345e-01 -1.39823771e+00 -1.05936444e+00
4.85532820e-01 -8.40727448e-01 7.50741065e-01 8.50238323e-01
-3.40732723e-01 -5.97004354e-01 -4.10878599e-01 -1.34280846e-01
7.57928789e-01 4.43728447e-01 3.90906245e-01 -1.36675251e+00
-5.50569415e-01 -8.67024004e-01 -2.18688458e-01 -3.78778934e-01
1.60146803e-01 -5.56218088e-01 -3.01422477e-02 -6.79155648e-01
-9.92225781e-02 -4.55844343e-01 -7.73143828e-01 -2.58374244e-01
-1.34788245e-01 -3.88062596e-01 2.77880609e-01 3.53926152e-01
-2.37462059e-01 6.20182335e-01 2.61193871e-01 1.19428545e-01
-1.21477857e-01 6.06865823e-01 -7.07662165e-01 5.18411577e-01
7.26580381e-01 -6.56692326e-01 -1.37052253e-01 1.80293977e-01
5.83564520e-01 3.52494955e-01 1.44379109e-01 -5.15893638e-01
8.09174478e-02 -6.01562142e-01 3.34606349e-01 -7.89936721e-01
1.61886707e-01 -3.83069783e-01 4.88897443e-01 4.82226282e-01
-1.11103542e-01 7.21955597e-01 1.55395538e-01 5.95095456e-01
-3.00812960e-01 -6.10942662e-01 5.30230641e-01 -6.25008866e-02
-4.51722383e-01 3.05425137e-01 -5.14010549e-01 -2.36485209e-02
1.00045002e+00 4.28842269e-02 -1.54523402e-01 -6.27486706e-01
-6.89658582e-01 1.46377951e-01 2.33882308e-01 1.85665283e-02
3.43133479e-01 -1.19164968e+00 -9.22827661e-01 -3.95481527e-01
-4.90118980e-01 -7.88065732e-01 2.53364325e-01 9.20307577e-01
-5.41637361e-01 5.71137190e-01 1.54432505e-01 -3.41225237e-01
-5.23966193e-01 4.98097330e-01 4.62556362e-01 -6.42644227e-01
-2.39388525e-01 1.24495769e+00 -1.49686225e-02 -2.59366542e-01
-1.70067549e-01 1.96403086e-01 -1.81539834e-01 3.79448950e-01
6.99223220e-01 5.31000853e-01 8.10371563e-02 -3.02259296e-01
-3.90491098e-01 2.76675880e-01 5.96187562e-02 -9.93855536e-01
1.54830289e+00 -2.58493163e-02 -1.02791987e-01 1.03668463e+00
5.63752532e-01 2.69575477e-01 -1.27578330e+00 6.36837780e-02
7.39453316e-01 -5.07734895e-01 2.50476420e-01 -6.08350992e-01
-9.06992614e-01 7.36559510e-01 5.79813898e-01 7.25398958e-01
7.18869328e-01 -5.28200209e-01 3.36904883e-01 3.32612485e-01
4.22197551e-01 -1.31643963e+00 -7.74772108e-01 5.97136974e-01
1.10975897e+00 -9.39672291e-01 1.73151672e-01 2.57169455e-01
-9.28264201e-01 1.41440511e+00 -2.81420797e-01 -1.70125529e-01
1.08800268e+00 2.60834575e-01 2.52637029e-01 -1.09759502e-01
-1.20502400e+00 4.86690700e-02 5.20011961e-01 4.07445341e-01
3.82069319e-01 3.95012438e-01 -5.25940299e-01 6.55380130e-01
-4.58441943e-01 -4.00818974e-01 4.69680786e-01 7.95460463e-01
-3.02830279e-01 -9.30161417e-01 -5.19474685e-01 5.24258852e-01
-6.44715190e-01 -2.63173252e-01 -1.82076737e-01 5.28768420e-01
-6.18634582e-01 9.56555665e-01 -1.12052768e-01 -3.02064300e-01
5.86302914e-02 3.46254557e-01 -9.58952084e-02 -1.75304145e-01
-6.54506385e-01 1.05736367e-01 -5.76440757e-03 -4.41284686e-01
1.32264540e-01 -1.47911489e+00 -5.32257438e-01 -6.85556829e-01
-1.11001156e-01 5.81149399e-01 1.24174058e+00 9.28687871e-01
1.83982074e-01 2.98016787e-01 1.21584141e+00 -9.10899460e-01
-1.51297426e+00 -7.81922340e-01 -1.01813388e+00 -8.20639431e-02
-1.80939779e-01 -7.61331141e-01 -9.99095201e-01 -4.73421842e-01] | [4.953858375549316, 4.051660537719727] |
c76339aa-dd3d-4366-bc0a-72919633e688 | greedy-transition-based-dependency-parsing-1 | 2007.04686 | null | https://arxiv.org/abs/2007.04686v1 | https://arxiv.org/pdf/2007.04686v1.pdf | Greedy Transition-Based Dependency Parsing with Discrete and Continuous Supertag Features | We study the effect of rich supertag features in greedy transition-based dependency parsing. While previous studies have shown that sparse boolean features representing the 1-best supertag of a word can improve parsing accuracy, we show that we can get further improvements by adding a continuous vector representation of the entire supertag distribution for a word. In this way, we achieve the best results for greedy transition-based parsing with supertag features with $88.6\%$ LAS and $90.9\%$ UASon the English Penn Treebank converted to Stanford Dependencies. | ['Ali Basirat', 'Joakim Nivre'] | 2020-07-09 | null | null | null | null | ['transition-based-dependency-parsing'] | ['natural-language-processing'] | [-2.23049790e-01 4.25291687e-01 -2.62755841e-01 -8.95407081e-01
-1.20737326e+00 -6.38208568e-01 1.72228009e-01 4.85025495e-01
-7.35638559e-01 7.90264428e-01 3.91412824e-01 -6.20753169e-01
1.81650713e-01 -9.48853970e-01 -4.55843836e-01 -6.03547752e-01
-4.89662111e-01 4.66332197e-01 4.82812375e-01 -6.50444388e-01
-3.87861840e-02 2.29969859e-01 -8.31946850e-01 3.20592254e-01
4.54131156e-01 6.00400865e-01 1.70857698e-01 6.59439445e-01
-5.44848382e-01 4.23620403e-01 -7.91517019e-01 -5.96213937e-01
1.11278802e-01 -4.06028420e-01 -9.61074591e-01 -4.48392719e-01
4.06728089e-01 -1.11333564e-01 -3.84913534e-01 1.08222294e+00
2.77956516e-01 9.03543383e-02 3.31897825e-01 -7.04498649e-01
-2.03534007e-01 1.36905885e+00 -3.84281516e-01 5.48963070e-01
3.04124206e-01 -3.77576947e-01 1.88966537e+00 -3.80796760e-01
6.73831820e-01 1.33634794e+00 5.66783369e-01 3.79144907e-01
-1.14274085e+00 -6.80430412e-01 4.34757441e-01 -1.84107661e-01
-8.45010221e-01 -1.74106196e-01 7.43553221e-01 -8.79798457e-02
1.48584974e+00 9.37684551e-02 4.06192809e-01 5.68212330e-01
5.06717503e-01 6.84306026e-01 9.07784462e-01 -6.20575547e-01
3.37622762e-02 -3.73827040e-01 1.04144716e+00 9.11446989e-01
1.59548327e-01 -2.38226712e-01 -4.55227137e-01 -4.08242792e-01
4.11069036e-01 -2.82426149e-01 1.82097569e-01 7.04639405e-02
-7.44429708e-01 1.46137083e+00 4.34113026e-01 5.89014709e-01
-2.20200077e-01 4.69423711e-01 5.34828901e-01 3.66728067e-01
4.99510497e-01 3.70174497e-01 -9.70353365e-01 -4.19417083e-01
-6.74874485e-01 2.14440659e-01 9.61719751e-01 9.69276249e-01
1.17069173e+00 1.48295179e-01 5.57931103e-02 9.03964281e-01
6.14612997e-02 4.31814700e-01 3.61312926e-01 -1.07137513e+00
7.43437946e-01 3.34397435e-01 2.33154446e-02 -5.94665587e-01
-6.88326240e-01 -2.33496860e-01 -8.08838308e-02 -3.14150274e-01
5.36061406e-01 -6.17272437e-01 -9.84638393e-01 1.85319781e+00
2.71857530e-01 -5.07381797e-01 1.80772960e-01 4.51115280e-01
7.55935729e-01 7.32228756e-01 4.52102661e-01 -2.93193340e-01
1.65612817e+00 -4.66488123e-01 -8.29842210e-01 -8.03564191e-01
1.27725899e+00 -6.80120170e-01 9.55994248e-01 9.65846553e-02
-9.75119710e-01 -2.18141705e-01 -8.46294641e-01 1.63463242e-02
-1.90376595e-01 -2.82924145e-01 1.44743276e+00 8.75855803e-01
-9.98558819e-01 6.68545246e-01 -1.21226776e+00 -2.12459296e-01
7.00404271e-02 5.74434638e-01 -6.35636449e-01 -3.29349756e-01
-1.29242885e+00 7.48196602e-01 4.15144920e-01 -9.91807356e-02
2.63790004e-02 -3.10600549e-01 -1.27400827e+00 5.70895849e-03
1.10638097e-01 -1.57196708e-02 1.39018214e+00 -3.57238710e-01
-1.16820097e+00 6.67638600e-01 -3.46105754e-01 -4.81694221e-01
-3.94437164e-01 -3.12939644e-01 -1.87444892e-02 -7.35820387e-04
3.16027761e-01 4.34591234e-01 1.72571391e-01 -7.46786237e-01
-8.34481597e-01 -5.08570731e-01 4.08060253e-01 2.81037949e-02
-2.90446579e-01 4.21388447e-01 -3.03678364e-01 -5.39353728e-01
7.42287040e-01 -9.36057568e-01 -7.42237985e-01 -1.02648795e+00
-5.15170991e-01 -5.47043502e-01 3.41417372e-01 -6.90063536e-01
1.26413786e+00 -2.12996101e+00 -5.30717857e-02 1.52856503e-02
-9.40120891e-02 4.39028218e-02 -1.65681288e-01 5.29474080e-01
1.44993039e-02 1.59516379e-01 -1.55572310e-01 -2.48903066e-01
-5.12022451e-02 7.66138315e-01 -9.29432549e-03 2.58547902e-01
2.15289861e-01 7.87148535e-01 -9.71857667e-01 -5.47923923e-01
-3.51096205e-02 1.83180407e-01 -7.94662714e-01 1.07167587e-01
-1.17814772e-01 -2.46423662e-01 -8.37485254e-01 5.94842315e-01
5.24884045e-01 1.81648716e-01 8.07212114e-01 2.64629163e-02
-2.09172353e-01 8.73318970e-01 -9.67756927e-01 1.62240887e+00
-5.77615380e-01 4.33706969e-01 3.35246861e-01 -9.01791275e-01
9.83789980e-01 5.76649755e-02 4.99254495e-01 -4.40575838e-01
1.50633603e-01 5.94308227e-02 3.35255712e-01 -2.20987629e-02
6.84283793e-01 -4.63802040e-01 -9.89267766e-01 4.46119845e-01
2.01275051e-01 -4.59044486e-01 3.01960766e-01 4.04194295e-01
1.63367617e+00 3.74620147e-02 2.39671096e-01 -5.77166855e-01
-1.13184072e-01 3.34228992e-01 8.56857836e-01 8.23121071e-01
-2.23473668e-01 4.30250674e-01 9.34103251e-01 -4.86022919e-01
-5.18429458e-01 -1.02673364e+00 -1.99477747e-01 1.74963772e+00
-3.50722730e-01 -7.34547734e-01 -6.13824248e-01 -1.00030565e+00
-7.87110031e-02 7.63461649e-01 -6.54364288e-01 1.49246424e-01
-1.30327499e+00 -1.21665764e+00 5.04615366e-01 8.61420333e-01
1.02046914e-01 -9.30832148e-01 -5.79208732e-01 6.83508396e-01
-1.57392517e-01 -1.32037163e+00 -2.13115290e-01 1.22353196e+00
-9.75066364e-01 -8.26166511e-01 -1.51720896e-01 -1.06838739e+00
2.49317914e-01 -1.03803106e-01 1.40380704e+00 1.30733929e-03
-4.23007235e-02 -2.07534119e-01 -9.11941648e-01 3.69028263e-02
-2.88480818e-01 3.42308223e-01 -2.71865726e-01 -7.90662408e-01
4.67667282e-01 -3.75484109e-01 -1.96002275e-01 -2.01707795e-01
-4.49366122e-01 -5.88456333e-01 3.86581689e-01 1.05772781e+00
3.39292020e-01 -2.93672711e-01 2.24639162e-01 -1.47788012e+00
2.13569909e-01 -2.95946002e-01 -3.68846059e-01 -1.58215955e-01
-1.53767899e-01 3.97745073e-01 5.42540669e-01 -7.27776438e-02
-8.64830971e-01 4.55738485e-01 -8.91146958e-01 3.36207479e-01
-9.71293226e-02 6.89949393e-01 -2.62187272e-01 2.57866770e-01
5.17156124e-01 -1.95652470e-01 -6.97368145e-01 -5.87949157e-01
4.66354966e-01 3.75755727e-01 1.82556942e-01 -6.58046484e-01
5.41573584e-01 1.93877697e-01 -9.22825262e-02 -8.16127360e-01
-9.36569035e-01 -4.10301834e-01 -7.26232886e-01 5.46257734e-01
9.83489156e-01 -7.36275733e-01 -2.25232139e-01 1.18646525e-01
-1.20594347e+00 -3.57064098e-01 -2.10374624e-01 5.32110214e-01
-2.61830568e-01 4.89070863e-01 -1.10484898e+00 -5.07699728e-01
-2.25540161e-01 -9.95951056e-01 1.03163278e+00 -8.34886357e-02
-3.45680833e-01 -9.15526807e-01 2.28377894e-01 -9.12021555e-04
1.47740409e-01 6.76419064e-02 1.27231765e+00 -1.02771831e+00
-4.04146314e-02 -3.67068619e-01 -2.61081811e-02 1.74133152e-01
2.56282002e-01 -2.40773037e-01 -6.84571326e-01 -1.48412153e-01
-2.36291617e-01 -7.78361410e-02 1.17712116e+00 5.55542946e-01
2.31731012e-01 -7.97492042e-02 -3.38625669e-01 4.10512060e-01
1.35827756e+00 3.16008389e-01 3.89993638e-01 3.44342828e-01
6.60099328e-01 4.96101350e-01 1.03277266e+00 4.05867279e-01
5.35419226e-01 4.84103978e-01 3.95458728e-01 1.27408668e-01
1.16655126e-01 -6.84534684e-02 4.96555537e-01 7.96509206e-01
2.22002417e-01 -2.79107213e-01 -1.08967674e+00 7.41638899e-01
-1.34917891e+00 -3.72117072e-01 -3.67708892e-01 1.65396440e+00
9.07310545e-01 5.27561307e-01 -4.17169556e-02 2.07777202e-01
6.56171739e-01 6.04368806e-01 1.50928393e-01 -9.31797802e-01
-5.26163727e-02 8.35557759e-01 9.50665951e-01 7.45947778e-01
-1.29744709e+00 1.77017212e+00 7.18302965e+00 5.18832207e-01
-7.82277226e-01 3.07133675e-01 4.42411155e-01 1.33116826e-01
-5.69572985e-01 1.55165240e-01 -1.27109325e+00 1.25856698e-01
1.12884665e+00 2.45455265e-01 -2.80069172e-01 8.92906129e-01
-2.30003625e-01 -2.45905936e-01 -8.61833394e-01 3.92221004e-01
-2.66086906e-01 -9.89716172e-01 -2.76050717e-01 -2.85378434e-02
3.42055649e-01 3.57442528e-01 -2.70697743e-01 4.24804747e-01
1.11721206e+00 -6.39063418e-01 4.49929208e-01 -5.90613723e-01
5.65083981e-01 -8.21874321e-01 1.01715910e+00 1.86911542e-02
-1.25880468e+00 2.95337476e-02 -6.67479932e-01 -2.96200067e-01
3.76004070e-01 7.71752238e-01 -8.40524912e-01 4.62146521e-01
7.58288860e-01 3.10221076e-01 -2.74103314e-01 3.49945068e-01
-5.41691899e-01 9.75593507e-01 -4.92382854e-01 -2.89542943e-01
5.60794353e-01 9.19469222e-02 2.89352953e-01 1.50059927e+00
8.00838470e-02 3.44269633e-01 4.48937923e-01 -1.08280100e-01
7.94091746e-02 3.46143216e-01 -4.82352436e-01 -3.43036801e-02
4.49266315e-01 9.55499649e-01 -9.48582709e-01 -1.59671530e-01
-6.13907278e-01 6.41943276e-01 7.22939610e-01 2.67106369e-02
-5.83857775e-01 -6.65514469e-01 8.14590335e-01 -6.56004176e-02
7.62738407e-01 -7.17037559e-01 -2.28030950e-01 -8.60887527e-01
-2.89559305e-01 -4.11343664e-01 9.83404458e-01 -2.59549886e-01
-9.88774896e-01 8.62437546e-01 4.12086323e-02 -4.46160018e-01
-6.21872067e-01 -7.28245080e-01 -5.86958885e-01 7.16982543e-01
-1.36753500e+00 -1.05380297e+00 4.43165720e-01 2.95023412e-01
5.62010527e-01 1.46651641e-01 1.20395029e+00 -1.66066274e-01
-4.47012126e-01 7.39835441e-01 -1.00341417e-01 5.60584009e-01
4.51734513e-01 -1.50946331e+00 8.77428114e-01 9.07541871e-01
3.78745556e-01 7.16375291e-01 9.14351463e-01 -6.98903203e-01
-1.47614455e+00 -7.09749877e-01 1.32848585e+00 -3.47334355e-01
8.21624994e-01 -5.28453171e-01 -8.01388144e-01 1.17064393e+00
1.82120413e-01 1.12012252e-01 6.68950498e-01 8.43012452e-01
-5.49162209e-01 1.28785416e-01 -1.05761373e+00 2.82099605e-01
9.84121501e-01 -2.77880281e-01 -7.79357791e-01 2.66581357e-01
8.38338673e-01 -3.18629980e-01 -1.16851497e+00 1.57110408e-01
2.94632733e-01 -7.40747988e-01 7.15984821e-01 -7.32102275e-01
1.70163170e-01 4.29382116e-01 -6.20651186e-01 -1.45545256e+00
-5.09085596e-01 -3.35386187e-01 3.57549876e-01 1.06809008e+00
6.85610235e-01 -7.08773553e-01 1.24647248e+00 6.76868618e-01
-5.84754586e-01 -5.09034514e-01 -1.53787315e+00 -5.78344405e-01
8.38028371e-01 -5.81526995e-01 4.60465819e-01 7.25449562e-01
5.86293340e-01 3.12676191e-01 -6.50809854e-02 1.88327223e-01
3.12559485e-01 3.03691506e-01 2.50212640e-01 -1.05818927e+00
-2.96646595e-01 -5.04260100e-02 -5.53031862e-01 -1.07026231e+00
5.03243864e-01 -6.74802125e-01 4.44965422e-01 -1.54989862e+00
8.97683576e-02 -7.47227490e-01 -4.98855829e-01 8.07090163e-01
-3.81010562e-01 2.42352173e-01 2.92887613e-02 -3.10444772e-01
-3.80279154e-01 5.21874651e-02 8.72401357e-01 3.01642977e-02
-1.11226469e-01 -2.56138265e-01 -7.63104618e-01 5.61282992e-01
9.34967995e-01 -9.06863272e-01 2.13545989e-02 -7.84041405e-01
2.46574730e-01 3.34555686e-01 -2.77520716e-01 -3.75548124e-01
-1.43650785e-01 -1.93523407e-01 -5.84503002e-02 -5.77556968e-01
6.63995564e-01 -3.46278638e-01 -5.04301012e-01 5.99304080e-01
-1.98379457e-01 3.47069770e-01 2.55928606e-01 3.60848010e-01
-2.99130708e-01 -4.70307678e-01 5.89323878e-01 -4.73164171e-01
-7.76730835e-01 -1.69007685e-02 -5.36580205e-01 2.96188116e-01
5.31065047e-01 8.54341388e-02 -1.93167493e-01 -2.26750121e-01
-9.49911892e-01 -7.50760734e-02 2.05095470e-01 1.24443941e-01
2.36476168e-01 -9.34610784e-01 -6.95954800e-01 2.17191413e-01
-2.64657408e-01 -3.34436625e-01 2.10239477e-02 1.72459319e-01
-4.57399786e-01 4.39849973e-01 -9.60728154e-02 -3.43603045e-01
-1.23954284e+00 7.49678072e-03 -4.90989834e-02 -4.44820434e-01
-5.52553236e-01 1.39693713e+00 -8.57083686e-03 -3.70049238e-01
-1.36744380e-01 -5.62088013e-01 -2.65824229e-01 2.22692207e-01
1.06667161e-01 -6.90634549e-03 3.29262912e-01 -7.55808353e-01
-8.33176196e-01 3.44131082e-01 -2.61968374e-01 -2.57074237e-01
1.52807593e+00 2.56088674e-01 -1.16277710e-01 3.22883427e-01
1.27107430e+00 4.15325463e-01 -7.95259655e-01 3.60624678e-02
2.16187388e-01 -3.74341488e-01 -7.52250571e-03 -4.05955493e-01
-1.08927298e+00 7.20403910e-01 2.34341294e-01 3.08284253e-01
6.37810588e-01 5.19888222e-01 9.72279251e-01 5.41553080e-01
7.96132505e-01 -9.03345346e-01 -3.90989870e-01 1.05707109e+00
1.80092886e-01 -1.01748049e+00 -1.38964765e-02 -8.83141637e-01
-5.54116666e-01 9.17826355e-01 2.15068504e-01 -4.86961871e-01
8.79028201e-01 6.91239536e-01 4.67722803e-01 -2.29143053e-01
-6.54619217e-01 -4.19685334e-01 -3.76508743e-01 4.95778680e-01
9.76410151e-01 2.97939271e-01 -6.57684207e-01 8.88615072e-01
-7.27260888e-01 -8.93612921e-01 6.45587981e-01 1.30490696e+00
-7.13696420e-01 -1.63202381e+00 -1.36295110e-01 6.52027071e-01
-1.03552341e+00 -5.15573561e-01 -5.54358773e-02 1.13868499e+00
-3.24293375e-01 1.18488228e+00 1.99142754e-01 -5.17795503e-01
3.69770944e-01 4.92371976e-01 4.86489177e-01 -1.16569865e+00
-9.48655486e-01 2.34512135e-01 8.33368301e-01 -5.74552596e-01
-2.18158904e-02 -1.10858202e+00 -1.67216265e+00 -9.41176713e-02
-3.96684408e-01 7.51909614e-01 6.88225567e-01 8.55248153e-01
-9.80597585e-02 3.41553807e-01 4.48257089e-01 -5.59354126e-01
-4.66890782e-01 -1.10920119e+00 -7.76490092e-01 2.06361189e-01
7.05935583e-02 -5.52082837e-01 -3.66209745e-01 -3.83093208e-01] | [10.344884872436523, 9.720023155212402] |
bbdfcb2c-9df0-40fb-b0b1-58961320c028 | domain-transfer-through-image-to-image | 2307.00479 | null | https://arxiv.org/abs/2307.00479v1 | https://arxiv.org/pdf/2307.00479v1.pdf | Domain Transfer Through Image-to-Image Translation for Uncertainty-Aware Prostate Cancer Classification | Prostate Cancer (PCa) is often diagnosed using High-resolution 3.0 Tesla(T) MRI, which has been widely established in clinics. However, there are still many medical centers that use 1.5T MRI units in the actual diagnostic process of PCa. In the past few years, deep learning-based models have been proven to be efficient on the PCa classification task and can be successfully used to support radiologists during the diagnostic process. However, training such models often requires a vast amount of data, and sometimes it is unobtainable in practice. Additionally, multi-source MRIs can pose challenges due to cross-domain distribution differences. In this paper, we have presented a novel approach for unpaired image-to-image translation of prostate mp-MRI for classifying clinically significant PCa, to be applied in data-constrained settings. First, we introduce domain transfer, a novel pipeline to translate unpaired 3.0T multi-parametric prostate MRIs to 1.5T, to increase the number of training data. Second, we estimate the uncertainty of our models through an evidential deep learning approach; and leverage the dataset filtering technique during the training process. Furthermore, we introduce a simple, yet efficient Evidential Focal Loss that incorporates the focal loss with evidential uncertainty to train our model. Our experiments demonstrate that the proposed method significantly improves the Area Under ROC Curve (AUC) by over 20% compared to the previous work (98.4% vs. 76.2%). We envision that providing prediction uncertainty to radiologists may help them focus more on uncertain cases and thus expedite the diagnostic process effectively. Our code is available at https://github.com/med-i-lab/DT_UE_PCa | ['Parvin Mousavi', 'Robert Siemens', 'Alexandre Menard', 'Jason Izard', 'Amoon Jamzad', 'Meng Zhou'] | 2023-07-02 | null | null | null | null | ['image-to-image-translation', 'image-to-image-translation'] | ['computer-vision', 'miscellaneous'] | [ 3.87363940e-01 3.22462529e-01 -2.14493766e-01 -6.10943079e-01
-1.48029721e+00 -6.87589824e-01 2.89779007e-01 2.52548575e-01
-4.14177597e-01 6.14023507e-01 -1.09143592e-01 -4.68119800e-01
-2.45966390e-01 -5.29303074e-01 -6.52394414e-01 -9.00593758e-01
-2.75141478e-01 8.99812937e-01 1.49154037e-01 5.01168370e-01
5.96835501e-02 5.21759927e-01 -5.82782924e-01 1.36709496e-01
1.25918436e+00 1.05773342e+00 4.51167375e-01 4.22076166e-01
1.18441880e-01 4.81880069e-01 -3.22084576e-01 -4.93233383e-01
2.97875673e-01 -9.92780253e-02 -5.54947913e-01 -1.46971464e-01
2.80070931e-01 -3.98341537e-01 -1.82630852e-01 1.07363462e+00
6.70716703e-01 -2.36758068e-01 6.79815710e-01 -7.33017027e-01
-3.68349791e-01 6.67909622e-01 -8.18846464e-01 2.89449453e-01
-2.56312937e-01 3.67657721e-01 2.83970356e-01 -6.59425318e-01
4.21982884e-01 7.10002840e-01 8.32313180e-01 5.42493522e-01
-1.18108833e+00 -8.00760686e-01 -4.07364011e-01 8.57848823e-02
-1.00226426e+00 -1.50209367e-01 6.99151218e-01 -3.50787014e-01
2.41959393e-01 4.02565598e-01 5.62330484e-01 1.10682869e+00
7.45807946e-01 7.28383899e-01 1.49185777e+00 -5.35064675e-02
2.46181622e-01 -9.95128043e-03 -2.02500492e-01 5.07962465e-01
2.64613926e-01 9.10594091e-02 2.20180373e-03 -2.43919596e-01
9.65181231e-01 1.10242091e-01 -1.56470388e-01 -4.32421446e-01
-9.78077292e-01 7.94177890e-01 8.53111565e-01 2.75685072e-01
-4.33989406e-01 7.20422193e-02 4.88747716e-01 -3.49284083e-01
2.92874962e-01 4.47547674e-01 -2.74204277e-02 -5.01362549e-04
-1.02021527e+00 -1.15645165e-03 3.51736456e-01 5.41087568e-01
7.57872164e-02 -4.45152223e-01 -2.76545584e-01 8.68088067e-01
1.15188211e-01 6.10443711e-01 6.47036612e-01 -9.98143792e-01
2.37342849e-01 2.17428714e-01 -1.11066103e-01 -5.98352790e-01
-5.64755559e-01 -7.84140110e-01 -9.81273770e-01 -2.10091263e-01
5.44638038e-01 8.80554095e-02 -1.20948064e+00 1.42293775e+00
3.32575262e-01 2.52373666e-01 -3.16653639e-01 1.10013366e+00
4.23751950e-01 5.01296595e-02 2.29342490e-01 -2.54144341e-01
1.53630817e+00 -7.66845167e-01 -6.23186946e-01 -3.83561611e-01
4.98111248e-01 -6.11493826e-01 9.97493684e-01 2.92591333e-01
-1.14855337e+00 -1.20156974e-01 -9.25852656e-01 1.07646145e-01
1.89419925e-01 2.60565907e-01 8.73258173e-01 9.31946814e-01
-7.24579811e-01 7.35920250e-01 -1.57868469e+00 -7.60292485e-02
9.43014979e-01 6.19251192e-01 -2.93300301e-01 -6.05712652e-01
-9.65461016e-01 8.89911354e-01 2.12814480e-01 3.22263300e-01
-1.18426895e+00 -1.15621889e+00 -7.78221607e-01 -1.31975710e-01
4.57067043e-01 -5.30955553e-01 1.31440139e+00 -3.60327035e-01
-1.24382293e+00 7.04925537e-01 -4.65639010e-02 -5.55811524e-01
8.66076052e-01 3.96362431e-02 -1.73164994e-01 4.09345627e-01
3.35198820e-01 7.13304460e-01 4.58156466e-01 -1.08614278e+00
-1.65496960e-01 -8.64787400e-01 -3.94876719e-01 2.09510978e-02
7.87308905e-03 -2.27338627e-01 -5.67131519e-01 -4.15831149e-01
2.84825832e-01 -1.29610777e+00 -7.07398832e-01 2.30495390e-02
-2.66882479e-01 2.91067660e-01 3.75830531e-01 -9.81639922e-01
6.48230612e-01 -2.10193658e+00 -2.27802604e-01 3.11161935e-01
3.69238824e-01 1.01149075e-01 2.94172734e-01 -4.58508641e-01
2.84312628e-02 1.30730644e-01 -7.15972364e-01 -2.41064429e-01
-2.05953285e-01 2.54034102e-01 1.42499194e-01 5.88243008e-01
2.57630348e-01 1.07573032e+00 -1.06608725e+00 -6.96855128e-01
2.65267819e-01 4.57602531e-01 -3.05179685e-01 8.45648274e-02
1.97095871e-01 1.12103271e+00 -6.18109643e-01 7.39955842e-01
1.02208710e+00 -4.16923940e-01 5.93816578e-01 -2.95517951e-01
1.34417415e-01 9.60545018e-02 -5.52437484e-01 1.83266068e+00
-4.20303553e-01 1.59263939e-01 1.23314507e-01 -8.53644550e-01
7.59470820e-01 5.49848415e-02 7.06289709e-01 -6.62183225e-01
7.98426643e-02 6.38735950e-01 3.98451239e-01 -3.52574050e-01
2.98739299e-02 -5.46159744e-01 -6.42606691e-02 1.00139409e-01
-9.29815695e-02 -3.28242153e-01 -1.32994071e-01 2.77428776e-02
1.23366153e+00 -7.34434128e-02 -6.73847422e-02 -1.89034462e-01
3.27800393e-01 7.08715543e-02 7.38833010e-01 8.11179042e-01
-6.29019141e-01 6.85271263e-01 4.53607440e-01 -1.00002475e-01
-8.99264991e-01 -1.36752295e+00 -7.01156676e-01 2.80141354e-01
4.81864251e-02 9.45678577e-02 -5.47800362e-01 -8.62958670e-01
5.37657505e-03 6.87678754e-01 -5.54062128e-01 -8.63626152e-02
-6.81966424e-01 -1.08395433e+00 6.59362733e-01 8.38393927e-01
3.71081740e-01 -4.71393108e-01 -4.41053540e-01 1.05883621e-01
-2.11849287e-01 -1.07519603e+00 -3.25824887e-01 3.81357849e-01
-1.18952310e+00 -9.28460240e-01 -1.13057637e+00 -3.51023674e-01
8.69468629e-01 -1.39435664e-01 7.42547214e-01 -4.15650785e-01
-4.49528426e-01 2.80231208e-01 4.00567837e-02 -3.24507326e-01
-3.37338120e-01 2.13457122e-02 -2.66535338e-02 -4.09612477e-01
2.99282908e-01 -4.88250643e-01 -9.22780573e-01 2.33247325e-01
-8.79060328e-01 -7.55287930e-02 1.25516546e+00 1.02213311e+00
1.01778352e+00 -8.23772550e-02 5.47587633e-01 -9.35433626e-01
3.50789458e-01 -4.72265065e-01 -5.09821415e-01 2.02636734e-01
-5.73179066e-01 7.76546746e-02 4.33797657e-01 -4.03572679e-01
-1.21956146e+00 2.23383486e-01 -1.89144120e-01 -5.40017426e-01
1.68405995e-02 6.67187035e-01 1.01510175e-01 -4.13713366e-01
6.52553141e-01 1.54133067e-01 2.12092280e-01 -2.21247345e-01
4.12451774e-02 4.66180652e-01 7.24030256e-01 -7.14315474e-01
5.96331894e-01 6.05654299e-01 3.29629451e-01 -1.77246317e-01
-7.10783005e-01 -3.79319936e-01 -4.99484241e-01 -2.22811475e-01
7.49224365e-01 -7.65586793e-01 -6.02388799e-01 2.29283750e-01
-6.03858888e-01 -4.14343655e-01 -9.35626402e-02 7.59509206e-01
-4.04997915e-01 6.67775154e-01 -7.89067328e-01 -5.45920908e-01
-5.93024909e-01 -1.60925436e+00 1.07826447e+00 3.46992493e-01
7.95069560e-02 -9.43781197e-01 -1.71642363e-01 6.95733845e-01
5.86627603e-01 3.78521621e-01 7.64587641e-01 -7.03641117e-01
-5.50610185e-01 -2.80868262e-01 -2.37086505e-01 1.57175854e-01
4.37827185e-02 -5.74412167e-01 -9.92234468e-01 -2.74316162e-01
3.61126810e-01 -3.27690363e-01 7.80072093e-01 9.58013892e-01
1.74678123e+00 3.19836974e-01 -5.17453551e-01 5.46301544e-01
1.15375423e+00 2.45185405e-01 6.81480110e-01 1.92147017e-01
4.65843141e-01 3.95018220e-01 8.17573786e-01 2.47778803e-01
3.80126268e-01 5.56218565e-01 3.83847594e-01 -1.79126095e-02
-2.29361031e-04 -1.02362409e-01 -4.89873812e-02 6.72690690e-01
9.70227867e-02 4.47282135e-01 -1.31331503e+00 5.85999429e-01
-1.53620183e+00 -3.83646160e-01 -1.07246321e-02 2.40089965e+00
1.03439128e+00 3.67843598e-01 -3.10824037e-01 -3.69507045e-01
7.36637771e-01 -2.82933950e-01 -8.26992333e-01 6.25189096e-02
3.70686918e-01 2.86442697e-01 9.36390281e-01 3.97744596e-01
-1.08207428e+00 4.10335451e-01 5.39104748e+00 9.21415687e-01
-1.51315629e+00 4.80859160e-01 1.16802049e+00 -1.26078248e-01
-2.60817975e-01 -1.77323952e-01 -3.65728915e-01 5.04483819e-01
1.08447230e+00 -2.08918557e-01 1.41628727e-01 8.50993156e-01
1.10647149e-01 -3.17977637e-01 -1.14471543e+00 8.61490071e-01
-1.44907311e-01 -1.16320872e+00 -6.57431841e-01 2.58217275e-01
5.22163391e-01 5.19555397e-02 4.47577596e-01 4.43755507e-01
-1.25657722e-01 -1.28472173e+00 -6.56662975e-03 3.76090884e-01
1.01389802e+00 -5.97659111e-01 1.02238083e+00 9.93774012e-02
-7.46560335e-01 1.36704341e-01 -2.41560355e-01 5.85622907e-01
3.97066697e-02 1.08471310e+00 -1.41199851e+00 7.42154837e-01
5.93732595e-01 2.10370898e-01 -6.49618924e-01 1.19144702e+00
-1.54167652e-01 5.75365603e-01 -3.60554665e-01 3.29837441e-01
2.65697725e-02 -1.38963126e-02 3.48349541e-01 1.03779459e+00
4.47598219e-01 -1.52724318e-03 1.05971567e-01 9.80253935e-01
-7.61283785e-02 -2.45468423e-01 -1.81154177e-01 3.67544182e-02
4.01779622e-01 1.58090258e+00 -8.34890544e-01 -7.95341879e-02
-3.80577073e-02 7.70250440e-01 8.18022415e-02 -9.97698754e-02
-1.06142318e+00 -2.44224556e-02 -1.41255781e-01 1.27085283e-01
8.23523775e-02 -4.64324020e-02 -5.02924860e-01 -9.68834400e-01
2.24354908e-01 -5.68620682e-01 3.89562190e-01 -5.40274501e-01
-1.51288342e+00 3.89061958e-01 -3.49182449e-02 -9.41525698e-01
-3.75764668e-01 -6.61892653e-01 -3.72401416e-01 8.70829642e-01
-1.53190410e+00 -1.17876101e+00 -2.74615616e-01 1.58992499e-01
1.96953446e-01 3.90986145e-01 8.19390237e-01 5.18838882e-01
-3.58272880e-01 6.18150592e-01 2.59470522e-01 7.75549635e-02
9.97586071e-01 -1.52118850e+00 -1.09024964e-01 5.75219810e-01
-6.61009848e-01 7.62403131e-01 4.42383111e-01 -8.87460351e-01
-1.56047559e+00 -1.11050093e+00 3.30047339e-01 -4.39427376e-01
7.97600746e-01 -1.25390813e-01 -8.57081413e-01 7.27656305e-01
-2.61233836e-01 4.48454291e-01 9.70216751e-01 -6.28726184e-02
1.50014162e-01 -2.00983882e-01 -1.71768796e+00 4.26127195e-01
5.47345161e-01 -2.31196433e-01 -4.04295266e-01 3.62987220e-01
4.11728382e-01 -1.07722592e+00 -1.42332125e+00 7.34109402e-01
6.30718350e-01 -5.76574385e-01 9.31991696e-01 -5.71400439e-03
4.40377653e-01 -1.34457111e-01 1.34077102e-01 -1.19915831e+00
-2.53342062e-01 -1.31751344e-01 -6.93387166e-02 8.71104717e-01
3.37209374e-01 -5.90730250e-01 9.99328434e-01 1.02339303e+00
-3.98521096e-01 -8.22784185e-01 -1.24170911e+00 -6.04511976e-01
3.35301220e-01 -4.76303279e-01 1.77929580e-01 8.75011683e-01
-1.88720658e-01 -3.88429821e-01 -5.31364866e-02 3.66978228e-01
9.73659933e-01 -8.16665869e-03 1.13735974e-01 -1.00531781e+00
-5.60083330e-01 -2.05287725e-01 -3.53039742e-01 -5.89213014e-01
-1.39786258e-01 -1.10111809e+00 -1.84774213e-02 -1.39328325e+00
6.29605293e-01 -8.57982516e-01 -5.21001339e-01 3.61455739e-01
-3.23131800e-01 1.85258701e-01 7.55262971e-02 2.64382243e-01
-2.98359722e-01 3.39312792e-01 1.71976149e+00 -2.68180430e-01
7.14830235e-02 1.46901995e-01 -7.61583447e-01 4.29563105e-01
8.21445048e-01 -5.53382337e-01 -2.80315250e-01 -2.19795331e-01
-2.67916054e-01 2.96323836e-01 4.50936288e-01 -1.02304947e+00
2.06153676e-01 -1.74010489e-02 9.32667851e-01 -5.58231771e-01
4.21413392e-01 -8.07038307e-01 2.73068100e-01 8.22301090e-01
-9.39762741e-02 -5.44440031e-01 4.86571401e-01 4.45590347e-01
-3.02920863e-02 -2.61923403e-01 8.30805182e-01 -2.78301895e-01
-2.41520777e-01 3.99551183e-01 -1.85663067e-02 -2.43354782e-01
8.84361207e-01 2.17011154e-01 -3.43545467e-01 -9.71469358e-02
-9.52715158e-01 3.91545355e-01 2.66979545e-01 -5.78421950e-02
5.14250338e-01 -1.24270487e+00 -5.61270297e-01 -2.12065160e-01
7.20176054e-03 2.65190005e-01 8.36160004e-01 1.63357651e+00
-4.79142219e-01 4.92803693e-01 -1.80219918e-01 -1.09430611e+00
-8.75181735e-01 3.54281992e-01 5.16383350e-01 -7.97090292e-01
-6.19545579e-01 6.71463251e-01 2.78641343e-01 -7.74798453e-01
-6.12084493e-02 -4.51253146e-01 2.11567909e-01 -5.09263754e-01
4.43103909e-01 -2.78730262e-02 2.65185446e-01 -1.49936453e-01
-5.98342955e-01 2.08377570e-01 -4.98830140e-01 -1.27667859e-01
1.35633004e+00 1.13109365e-01 5.03778495e-02 2.06425071e-01
8.24029565e-01 4.80109043e-02 -1.29815829e+00 -6.10092841e-02
-9.87025350e-03 -6.05081916e-01 2.97224015e-01 -1.21044147e+00
-1.02419293e+00 8.80248129e-01 1.15801013e+00 -2.13337585e-01
1.14648676e+00 1.07903659e-01 7.00313389e-01 -3.05472258e-02
4.13847655e-01 -6.83759689e-01 -7.34898224e-02 -4.09765504e-02
7.13317811e-01 -1.52112889e+00 1.50197819e-01 -5.21688521e-01
-8.94755006e-01 9.96269584e-01 5.44231057e-01 9.24751535e-02
2.21171603e-01 4.23472732e-01 7.56540820e-02 -2.31268048e-01
-3.36978108e-01 4.06641811e-01 1.92948386e-01 6.65240467e-01
5.43037355e-01 5.03541410e-01 -4.24997807e-01 1.00476289e+00
1.48186991e-02 1.88242838e-01 5.38281143e-01 8.31763148e-01
-5.04822880e-02 -1.20726073e+00 -5.30082107e-01 7.31432021e-01
-8.35159719e-01 1.10216439e-01 3.01024783e-02 5.49570501e-01
-6.72354177e-02 5.39838850e-01 -1.60348982e-01 -3.73339318e-02
-4.00595330e-02 -2.58946866e-01 7.54141152e-01 -4.11849082e-01
-3.60642731e-01 3.70401263e-01 -8.95179659e-02 -3.81690472e-01
-1.84593439e-01 -7.27561712e-01 -1.35290420e+00 -1.62422627e-01
-2.08984226e-01 9.32421833e-02 1.00237381e+00 7.48091638e-01
3.88457865e-01 8.02424252e-01 5.15963137e-01 -8.22746992e-01
-6.74005866e-01 -9.65060055e-01 -4.84712392e-01 2.21341830e-02
-1.97159946e-02 -7.21100211e-01 -1.40044987e-01 -3.84996265e-01] | [14.6195707321167, -2.3797786235809326] |
ce61cb6c-268f-4962-99b1-5ae48e0f7c2f | integrity-and-junkiness-failure-handling-for | 2304.09287 | null | https://arxiv.org/abs/2304.09287v1 | https://arxiv.org/pdf/2304.09287v1.pdf | Integrity and Junkiness Failure Handling for Embedding-based Retrieval: A Case Study in Social Network Search | Embedding based retrieval has seen its usage in a variety of search applications like e-commerce, social networking search etc. While the approach has demonstrated its efficacy in tasks like semantic matching and contextual search, it is plagued by the problem of uncontrollable relevance. In this paper, we conduct an analysis of embedding-based retrieval launched in early 2021 on our social network search engine, and define two main categories of failures introduced by it, integrity and junkiness. The former refers to issues such as hate speech and offensive content that can severely harm user experience, while the latter includes irrelevant results like fuzzy text matching or language mismatches. Efficient methods during model inference are further proposed to resolve the issue, including indexing treatments and targeted user cohort treatments, etc. Though being simple, we show the methods have good offline NDCG and online A/B tests metrics gain in practice. We analyze the reasons for the improvements, pointing out that our methods are only preliminary attempts to this important but challenging problem. We put forward potential future directions to explore. | ['Pramodh Karanth Prabhakar', 'Hao Fu', 'Guangdeng Liao', 'Shuai Ding', 'Chiyao Shen', 'Yunxi Guo', 'Wenping Wang'] | 2023-04-18 | null | null | null | null | ['text-matching'] | ['natural-language-processing'] | [-2.12754719e-02 1.84123844e-01 -2.77491957e-01 5.09381294e-02
-8.14673722e-01 -8.43981326e-01 9.95863080e-01 4.52159107e-01
-6.60395265e-01 4.76678401e-01 4.33530837e-01 -6.52895868e-01
-8.29432130e-01 -4.38669533e-01 -2.52502918e-01 -2.98594117e-01
2.79329177e-02 3.77954066e-01 4.13801819e-01 -6.49511456e-01
7.50585735e-01 3.92322659e-01 -1.53926301e+00 3.07776742e-02
8.62283945e-01 7.33365715e-01 -1.14647053e-01 5.73454499e-01
-2.35505223e-01 7.93163240e-01 -8.27208042e-01 -8.50093365e-01
1.87483281e-02 -6.15543276e-02 -9.21289086e-01 -6.31317973e-01
2.52510548e-01 -3.60330909e-01 -5.55727303e-01 1.21825814e+00
1.04382908e+00 6.41905069e-02 6.03454292e-01 -1.45727968e+00
-7.59033680e-01 3.42558831e-01 -2.16273412e-01 3.95223200e-01
5.83042800e-01 -1.07164972e-01 1.14601493e+00 -6.91530585e-01
6.87123358e-01 1.30479407e+00 6.64559484e-01 4.56748813e-01
-9.23147798e-01 -6.04877412e-01 -1.97874799e-01 3.67495269e-01
-1.25515378e+00 -4.95803535e-01 5.94963431e-01 -3.96818012e-01
8.84803832e-01 8.76553118e-01 1.61493123e-01 1.30244231e+00
8.80071372e-02 7.04192758e-01 9.44016635e-01 -6.92945480e-01
6.76409975e-02 7.06310213e-01 4.40643311e-01 4.05584663e-01
2.92857647e-01 1.25909224e-01 -2.88323432e-01 -7.88525581e-01
1.83278650e-01 -2.11696416e-01 -2.17385814e-01 -9.26287621e-02
-7.58172929e-01 1.07767582e+00 1.56343102e-01 7.18600214e-01
-1.65983260e-01 1.93899851e-02 5.30041695e-01 4.74986732e-01
6.81379795e-01 7.18750238e-01 -2.86674738e-01 -1.44387662e-01
-9.65513825e-01 4.03450012e-01 9.17797744e-01 7.37198114e-01
3.41476291e-01 -6.37473285e-01 -2.92727768e-01 1.01189435e+00
4.02387768e-01 3.73360962e-01 5.91017723e-01 -6.96651280e-01
3.83734673e-01 3.48890394e-01 1.96736947e-01 -1.58116567e+00
-1.84547395e-01 -3.77769113e-01 -3.60546798e-01 -1.75517678e-01
3.62845868e-01 -4.13640738e-02 -5.13239205e-01 1.46613431e+00
-5.05261831e-02 -1.37515306e-01 -4.77508336e-01 5.76874614e-01
6.57242477e-01 4.38406050e-01 2.79405743e-01 -2.90562898e-01
1.38523328e+00 -7.58454084e-01 -1.20504010e+00 -1.04797713e-01
9.07357454e-01 -1.23217511e+00 8.12970579e-01 1.63213536e-01
-1.07093740e+00 -5.03102206e-02 -8.43472362e-01 -3.26060541e-02
-7.92894959e-01 -3.45417947e-01 6.10163748e-01 1.22796893e+00
-1.16259682e+00 8.01984489e-01 -2.05788627e-01 -6.39024556e-01
4.07743044e-02 4.55171824e-01 -5.67657351e-02 1.68341979e-01
-1.73518419e+00 1.22963941e+00 3.82418409e-02 7.59172067e-02
-1.38435569e-02 -4.47832048e-01 -3.77440631e-01 1.77538633e-01
6.28007472e-01 -4.81116503e-01 1.04717636e+00 -4.57803130e-01
-1.06275940e+00 8.77011001e-01 3.12330630e-02 -1.62337825e-01
5.81186116e-01 -3.15094322e-01 -7.44749665e-01 -8.59967694e-02
4.00624722e-02 1.09675542e-01 6.13156855e-01 -8.75891447e-01
-2.63774902e-01 -4.67938423e-01 -2.46787518e-02 1.65477231e-01
-8.48804951e-01 8.08900714e-01 -5.58628201e-01 -7.18655884e-01
-1.88407511e-01 -1.03947186e+00 -6.90361708e-02 -1.43289819e-01
-4.57652807e-01 -3.99813354e-01 7.52170861e-01 -1.00669110e+00
1.93753827e+00 -1.97668517e+00 -8.67529809e-02 5.67142487e-01
2.58988649e-01 6.85377896e-01 -8.94787535e-03 9.83706117e-01
3.09197162e-03 8.47787738e-01 3.23086321e-01 -6.23422489e-02
2.55380154e-01 -2.64562130e-01 -3.85256827e-01 2.79689163e-01
-3.03685904e-01 1.12040734e+00 -8.53818595e-01 -6.22351944e-01
3.89511627e-03 4.21843052e-01 -4.45756942e-01 -1.00797728e-01
1.55780271e-01 -3.02110970e-01 -6.99772894e-01 7.40030348e-01
3.94822806e-01 -2.35390533e-02 7.58252442e-02 -3.34847942e-02
6.02857545e-02 4.72241700e-01 -1.00604558e+00 1.27154875e+00
-2.60842800e-01 7.61145175e-01 9.95960534e-02 -7.11213470e-01
5.14301896e-01 3.86200428e-01 3.08664113e-01 -8.52584958e-01
1.65990844e-01 3.76236260e-01 -1.26972035e-01 -8.29844296e-01
9.87827539e-01 1.78537384e-01 -2.59821825e-02 7.33357370e-01
-1.74887612e-01 2.32574999e-01 -1.20091632e-01 4.54968899e-01
1.22931397e+00 -2.55006462e-01 1.15066476e-01 -3.87158304e-01
3.34211677e-01 -1.87004447e-01 -8.59526396e-02 9.16064978e-01
-5.16274512e-01 1.90977156e-01 5.03316998e-01 5.68550602e-02
-9.23780918e-01 -4.95654464e-01 2.54251491e-02 1.02343953e+00
2.69714683e-01 -7.14401186e-01 -7.28262305e-01 -5.47129929e-01
-6.28941599e-03 7.19451427e-01 -3.95487964e-01 -4.84697908e-01
-3.80585581e-01 -6.71150684e-01 7.70013571e-01 8.42118487e-02
1.52164176e-01 -1.05519843e+00 -2.77766883e-01 1.76985517e-01
-3.05560201e-01 -7.58431494e-01 -5.26114643e-01 -1.36795789e-01
-7.57888734e-01 -9.94720697e-01 -7.84654438e-01 -5.98804057e-01
3.16995174e-01 3.90863508e-01 8.54893029e-01 4.68243599e-01
-5.28851032e-01 5.04019976e-01 -3.55084568e-01 -3.32670152e-01
-3.84695977e-01 1.10367410e-01 2.04800010e-01 -3.40819180e-01
6.77136362e-01 -3.15727741e-01 -7.78365016e-01 4.31164026e-01
-1.02407444e+00 -6.79317176e-01 2.53844380e-01 8.73085678e-01
-3.00605476e-01 1.65028647e-01 4.23940212e-01 -1.17709982e+00
1.49241793e+00 -6.21140122e-01 -1.59323633e-01 4.18291122e-01
-1.18014967e+00 3.55564281e-02 1.36049995e-02 -4.57810163e-01
-7.17703998e-01 -7.36735702e-01 -1.32108033e-01 -5.10378391e-04
2.77889132e-01 6.33918405e-01 1.82790965e-01 -1.46047026e-01
8.06403339e-01 -1.00204155e-01 2.69346684e-02 -7.56617546e-01
1.75973594e-01 1.17950094e+00 -3.15653961e-05 -2.47495845e-01
1.01470077e+00 -3.21546122e-02 -4.83413279e-01 -1.00968289e+00
-3.81045192e-01 -9.66019273e-01 4.39412221e-02 -2.61680305e-01
5.88156402e-01 -5.23002684e-01 -7.40051270e-01 1.73137188e-02
-1.15645218e+00 2.64354706e-01 2.28872642e-01 1.45840079e-01
-1.95170820e-01 7.09672868e-01 -7.22355902e-01 -1.17497838e+00
-4.90020394e-01 -8.43678534e-01 7.28352606e-01 8.34375545e-02
-6.63262606e-01 -1.12597489e+00 2.47853354e-01 6.61518872e-01
7.53967881e-01 -1.74238443e-01 1.22120762e+00 -1.08537018e+00
-3.00668269e-01 -5.66277206e-01 -1.58214554e-01 1.54189244e-01
-5.07813655e-02 9.65489522e-02 -9.47313845e-01 -4.06203806e-01
-1.01107411e-01 -2.22217485e-01 4.97541547e-01 -4.91134636e-02
1.04290032e+00 -4.17169422e-01 -6.45532668e-01 -6.46546856e-02
1.28268206e+00 3.43407810e-01 9.73970473e-01 6.53219759e-01
1.33471370e-01 9.21019375e-01 6.26100481e-01 1.83637097e-01
-2.55306304e-01 1.11233485e+00 2.48770937e-01 -3.55726071e-02
1.25266030e-01 -4.20507491e-01 1.43511385e-01 6.63511872e-01
2.12964159e-03 -5.89011788e-01 -6.63959205e-01 3.74902308e-01
-1.99698901e+00 -1.25288200e+00 -2.26443503e-02 2.36608028e+00
4.55612838e-01 3.22233468e-01 2.06111029e-01 2.39530951e-01
8.74652445e-01 4.33688574e-02 -4.01550271e-02 -5.78953743e-01
4.13548388e-02 2.15763718e-01 6.29923880e-01 6.65582418e-01
-7.74828613e-01 7.76897192e-01 6.64740849e+00 1.07833397e+00
-7.04418361e-01 3.33248138e-01 5.27261376e-01 1.18271247e-01
-6.23902798e-01 7.03647137e-02 -6.12078667e-01 6.98846996e-01
8.69676530e-01 -3.13660532e-01 5.23122489e-01 7.00016439e-01
1.50907025e-01 1.11365199e-01 -7.34253049e-01 8.10912788e-01
8.59269053e-02 -1.17058718e+00 -1.97349057e-01 2.85115033e-01
1.42719060e-01 -1.99674159e-01 1.98361903e-01 5.03955722e-01
-1.34761795e-01 -8.87213469e-01 4.32562917e-01 3.70925337e-01
5.38593948e-01 -6.59320951e-01 6.93310440e-01 4.21914637e-01
-4.04136449e-01 -9.75504965e-02 -2.15016156e-01 1.04422264e-01
-8.53605345e-02 3.82090122e-01 -5.89882731e-01 3.17708731e-01
4.36545908e-01 9.93306190e-02 -8.29071641e-01 1.27146864e+00
5.24305813e-02 5.38397372e-01 -2.25306496e-01 -5.78977525e-01
9.83605459e-02 -1.24219947e-01 8.14436018e-01 1.09934056e+00
2.50498682e-01 -1.37709394e-01 -4.42847133e-01 6.75248742e-01
1.27117708e-02 4.24302787e-01 -8.62113297e-01 -4.74957049e-01
5.89107811e-01 1.15554357e+00 -6.56236589e-01 -6.17592037e-02
-1.68210104e-01 1.05465579e+00 -4.16321307e-02 4.02331889e-01
-6.95873380e-01 -7.36965716e-01 4.76157635e-01 3.41515690e-01
-1.82575643e-01 5.86177520e-02 4.47282083e-02 -1.13867545e+00
1.80621907e-01 -9.76497471e-01 3.50162387e-01 -6.53415561e-01
-1.12858343e+00 3.28048289e-01 6.57040551e-02 -1.01672375e+00
-4.52031702e-01 -6.37580156e-01 -4.14541811e-01 9.14911687e-01
-1.26013339e+00 -5.95004439e-01 2.74315983e-01 2.49742761e-01
3.49051148e-01 -2.57515013e-01 9.31592286e-01 8.59196484e-01
-4.38545763e-01 9.34068501e-01 3.73842746e-01 -1.07471503e-01
1.00642002e+00 -9.20161188e-01 3.05619925e-01 6.13138914e-01
1.97965637e-01 1.06038260e+00 1.00659978e+00 -6.14083409e-01
-1.34205234e+00 -4.18884933e-01 1.77640057e+00 -6.66326940e-01
9.32583690e-01 -4.16088641e-01 -6.42393589e-01 3.25543702e-01
1.56581700e-01 -6.30363584e-01 7.85026073e-01 4.65791643e-01
-2.50465363e-01 2.17005849e-01 -9.76604939e-01 1.06432438e+00
9.35301781e-01 -8.26122463e-01 -5.01575351e-01 8.76607239e-01
7.65479982e-01 8.02809224e-02 -6.16721690e-01 1.08945683e-01
5.52428246e-01 -9.62911606e-01 1.21478152e+00 -7.92990446e-01
2.83334941e-01 2.40441442e-01 2.46177390e-02 -1.00038457e+00
-4.58856314e-01 -1.08028138e+00 2.17774715e-02 1.34852898e+00
5.78175187e-01 -7.33551621e-01 6.55680001e-01 1.08490634e+00
3.03870797e-01 -5.26395738e-01 -7.66534686e-01 -9.34555769e-01
1.18835215e-02 -2.09514976e-01 1.97750583e-01 1.15153277e+00
4.61397201e-01 5.16325831e-01 -5.44626415e-01 -2.24720493e-01
4.36087340e-01 -3.85553330e-01 3.89172137e-01 -1.29054999e+00
-2.45489314e-01 -7.14894116e-01 -3.10866147e-01 -6.67448103e-01
4.77775075e-02 -6.82226717e-01 -5.89692295e-01 -1.17295396e+00
3.40490103e-01 -4.71384466e-01 -4.49221879e-01 5.20379506e-02
-1.71450898e-01 1.97376549e-01 2.11888969e-01 3.32131833e-01
-5.18524051e-01 1.46829247e-01 7.58078575e-01 -2.43045971e-01
-1.15765668e-01 2.27106810e-01 -9.65499282e-01 2.86169201e-01
7.02255726e-01 -5.69082201e-01 -4.68620241e-01 -1.15060069e-01
8.34588230e-01 1.30803511e-01 2.45334730e-01 -3.54516566e-01
3.20956796e-01 6.45817369e-02 -2.41199866e-01 -1.75469339e-01
2.42355064e-01 -9.72314298e-01 3.68246548e-02 5.93855441e-01
-6.73380673e-01 2.56150514e-01 8.51756632e-02 6.95824146e-01
-8.70562419e-02 -7.95867324e-01 9.05014500e-02 1.94166712e-02
-3.47918242e-01 4.13103774e-02 -5.80302894e-01 1.37159582e-02
6.85665429e-01 -1.01849243e-01 -3.93826276e-01 -8.84350359e-01
-6.34823322e-01 2.00748071e-01 2.16506258e-01 7.97834575e-01
3.74007851e-01 -1.21920896e+00 -4.09943223e-01 -2.22853154e-01
1.99506640e-01 -1.09534538e+00 -2.89928392e-02 7.12299526e-01
-4.16245788e-01 7.50334918e-01 7.60410056e-02 -2.30264431e-03
-1.61829329e+00 7.52651334e-01 -7.44429827e-02 -4.07871604e-01
-2.17546180e-01 5.85330486e-01 -3.30432564e-01 -2.70211786e-01
6.05476856e-01 3.06186259e-01 -4.24655229e-01 2.97724903e-01
5.79096615e-01 7.26052105e-01 3.27616781e-01 -4.63734925e-01
-4.16150451e-01 2.46198192e-01 -3.28764796e-01 -3.38406503e-01
1.04551744e+00 -2.35955954e-01 -2.81174660e-01 2.01824620e-01
1.39762056e+00 1.80850133e-01 2.29834486e-02 -2.83018529e-01
5.40012836e-01 -5.18302262e-01 1.93017110e-01 -9.87184882e-01
-4.61020768e-01 7.71876752e-01 6.81635320e-01 7.17095077e-01
8.66806328e-01 -2.74184227e-01 8.40089738e-01 6.08121336e-01
2.58034945e-01 -1.37404621e+00 4.81545180e-03 4.12292510e-01
6.46621406e-01 -1.11229408e+00 1.36475414e-01 -3.81614149e-01
-3.01785558e-01 8.87212753e-01 3.85293067e-02 3.58145356e-01
6.55430317e-01 -1.69003353e-01 8.91789123e-02 -3.37903529e-01
-6.45104527e-01 -1.55680357e-02 3.21246356e-01 2.64340550e-01
6.25343621e-01 -1.79883048e-01 -9.90303397e-01 1.49027854e-01
-3.06836423e-02 -2.34376848e-01 3.19723099e-01 8.82362843e-01
-3.26547354e-01 -1.39024079e+00 -3.27841073e-01 5.03524542e-01
-9.28786635e-01 -3.38118941e-01 -7.72107899e-01 8.55075300e-01
-3.45506907e-01 1.06440783e+00 -3.67287576e-01 -6.38570786e-01
2.70933867e-01 3.84769976e-01 2.31879488e-01 -3.14011902e-01
-9.63680089e-01 8.92171934e-02 4.90889281e-01 -5.77172995e-01
-1.61017910e-01 -5.62789321e-01 -4.04120415e-01 -3.61546099e-01
-8.88171494e-01 4.04014587e-01 1.00182784e+00 8.01099360e-01
4.04730320e-01 6.11527301e-02 5.61268330e-01 -2.83662766e-01
-1.06857634e+00 -9.36143041e-01 -4.56416607e-01 5.76381981e-01
9.43396166e-02 -5.79386890e-01 -6.63715303e-01 -5.66532254e-01] | [11.380078315734863, 7.56084680557251] |
158c93cb-1fe9-4600-92e7-76e800139973 | pushing-the-limits-of-unsupervised-unit | 2306.08920 | null | https://arxiv.org/abs/2306.08920v1 | https://arxiv.org/pdf/2306.08920v1.pdf | Pushing the Limits of Unsupervised Unit Discovery for SSL Speech Representation | The excellent generalization ability of self-supervised learning (SSL) for speech foundation models has garnered significant attention. HuBERT is a successful example that utilizes offline clustering to convert speech features into discrete units for a masked language modeling pretext task. However, simply clustering features as targets by k-means does not fully inspire the model's performance. In this work, we present an unsupervised method to improve SSL targets. Two models are proposed, MonoBERT and PolyBERT, which leverage context-independent and context-dependent phoneme-based units for pre-training. Our models outperform other SSL models significantly on the LibriSpeech benchmark without the need for iterative re-clustering and re-training. Furthermore, our models equipped with context-dependent units even outperform target-improvement models that use labeled data during pre-training. How we progressively improve the unit discovery process is demonstrated through experiments. | ['Xie Chen', 'Chao Zhang', 'Yu Wang', 'Guanrou Yang', 'Zhisheng Zheng', 'Ziyang Ma'] | 2023-06-15 | null | null | null | null | ['clustering'] | ['methodology'] | [ 1.36233136e-01 2.24693090e-01 -2.67375499e-01 -8.55397344e-01
-1.09469557e+00 -7.21314788e-01 8.11462760e-01 2.64018834e-01
-6.36038601e-01 5.43633461e-01 3.42905432e-01 -7.18897283e-01
9.39069316e-02 -4.83173400e-01 -5.52161813e-01 -4.40577447e-01
-9.65791047e-02 5.42274177e-01 3.20579171e-01 -1.18266195e-01
-1.69757148e-03 3.73670429e-01 -1.50456727e+00 6.08867764e-01
9.02816176e-01 6.79504335e-01 3.75121832e-01 6.34929895e-01
-3.70671123e-01 7.76603401e-01 -5.81200182e-01 -2.06275918e-02
4.80124950e-02 -6.30035341e-01 -9.30121660e-01 1.64932281e-01
2.08535358e-01 2.02809930e-01 -3.21902931e-01 7.52290130e-01
4.72876728e-01 5.69196105e-01 7.06879497e-01 -9.00828600e-01
-4.57274079e-01 1.24389565e+00 -4.46566381e-04 1.27017081e-01
-3.38709019e-02 -1.07699968e-01 1.14556682e+00 -1.20808327e+00
1.50189370e-01 1.13720226e+00 6.13030672e-01 6.45473719e-01
-1.27541757e+00 -5.42014778e-01 1.59595475e-01 4.97312874e-01
-1.57367933e+00 -8.80546093e-01 6.85410023e-01 -3.19952726e-01
1.24961603e+00 3.55275571e-01 9.65169743e-02 8.31275046e-01
-5.48864603e-01 8.99628341e-01 1.14330733e+00 -9.87703204e-01
4.73164976e-01 5.76950967e-01 3.48638177e-01 6.73246205e-01
-2.66784340e-01 1.85905561e-01 -5.63065886e-01 -9.23496112e-02
3.52601588e-01 -2.53428549e-01 -3.05019859e-02 -1.71742842e-01
-1.08305013e+00 8.44417512e-01 3.29272330e-01 7.67331183e-01
-5.21956570e-02 -1.55371442e-01 1.93583846e-01 4.38319534e-01
5.65727592e-01 5.87711811e-01 -5.85248470e-01 -2.74156690e-01
-1.31262374e+00 -6.62305534e-01 7.39286184e-01 1.11018467e+00
1.08828259e+00 2.91330159e-01 -1.31746635e-01 1.11343133e+00
2.30725259e-01 8.70149434e-02 9.21862125e-01 -6.76465809e-01
2.96958953e-01 4.75344718e-01 -3.59042197e-01 -2.11585477e-01
-2.66787767e-01 -4.87244219e-01 -6.49263680e-01 -3.90253291e-02
1.52833432e-01 -2.83935845e-01 -1.22038770e+00 1.56199718e+00
4.24327031e-02 5.24464130e-01 2.62436032e-01 5.24915576e-01
7.47062564e-01 8.20851564e-01 2.86877099e-02 -4.05074328e-01
7.38028288e-01 -1.30495834e+00 -5.34159422e-01 -2.56658018e-01
9.77968872e-01 -6.62235796e-01 1.12794733e+00 3.48715544e-01
-9.75183785e-01 -8.40451479e-01 -1.10013461e+00 2.03425974e-01
-5.31322062e-01 1.31120443e-01 4.88448560e-01 1.13259852e+00
-1.18622851e+00 6.32103860e-01 -8.42578292e-01 -4.07635182e-01
8.44812691e-02 4.79104459e-01 -1.85341299e-01 2.80848444e-01
-1.15039361e+00 8.76489103e-01 5.66964984e-01 -4.25104588e-01
-9.53735530e-01 -6.24497712e-01 -6.56584978e-01 2.44724289e-01
3.22914004e-01 -9.37517434e-02 1.41322184e+00 -7.84860790e-01
-1.77641487e+00 6.23531342e-01 -4.59734857e-01 -8.33481729e-01
1.13956496e-01 1.88957006e-01 -4.97977793e-01 2.28507584e-03
-3.11902970e-01 6.15730107e-01 9.24397707e-01 -1.49581826e+00
-7.46257067e-01 1.53631056e-02 -3.09181303e-01 3.75041872e-01
-8.32608759e-01 1.85622387e-02 -4.43313509e-01 -5.76301515e-01
3.07682276e-01 -6.97143137e-01 -2.44018108e-01 -9.18129742e-01
-3.49037945e-01 -6.37382865e-01 7.17731118e-01 -3.77687365e-01
1.58492577e+00 -2.17872524e+00 -5.40879108e-02 4.70233470e-01
2.14709014e-01 5.59020042e-01 -1.14402749e-01 4.45289850e-01
-2.42114872e-01 3.62910330e-01 -2.56244957e-01 -9.16506052e-01
1.69070095e-01 4.15283114e-01 -2.88193315e-01 1.01926096e-01
1.60781384e-01 7.02056944e-01 -7.82682836e-01 -5.47279477e-01
2.85695046e-01 2.61111379e-01 -6.92715049e-01 3.56785864e-01
-1.30577818e-01 3.41045022e-01 2.61734307e-01 3.29049736e-01
2.55108267e-01 -1.21536903e-01 1.49621591e-01 1.18743688e-01
-2.87476897e-01 9.21078086e-01 -9.50658619e-01 1.48796117e+00
-6.88674748e-01 5.85772157e-01 -6.79607540e-02 -1.33355534e+00
1.06081104e+00 4.24047112e-01 3.52291554e-01 -3.27576578e-01
-9.34465379e-02 1.38449758e-01 2.45313287e-01 -3.74935567e-02
3.52181882e-01 -2.81498253e-01 1.14244014e-01 5.32283723e-01
4.06783879e-01 -1.48685575e-01 9.61195678e-02 2.29425341e-01
1.11712956e+00 -3.36504221e-01 8.51565674e-02 -1.23460241e-01
5.93021750e-01 -8.29521865e-02 4.06085283e-01 8.90690804e-01
-1.95200011e-01 7.46027052e-01 -4.15126346e-02 1.93374753e-01
-7.91975379e-01 -1.22669303e+00 5.19015938e-02 1.62244844e+00
-3.88144612e-01 -6.92447126e-01 -8.02978694e-01 -8.89153421e-01
-1.78502858e-01 1.02348554e+00 -3.97592664e-01 -3.04396808e-01
-4.89878505e-01 -4.80951130e-01 6.82614565e-01 5.13844430e-01
2.77989984e-01 -1.07477808e+00 1.14995182e-01 2.39919841e-01
-9.40965191e-02 -1.03929186e+00 -5.05347311e-01 9.77919817e-01
-7.28854179e-01 -3.32162380e-01 -5.88829935e-01 -1.25275159e+00
7.12252736e-01 3.19224060e-01 9.87061858e-01 1.04219839e-01
1.07636325e-01 5.19488931e-01 -7.18479931e-01 -2.28460073e-01
-9.45683897e-01 4.71428692e-01 5.54894626e-01 1.04474619e-01
6.45487547e-01 -6.55932963e-01 -9.53434780e-02 4.52729046e-01
-5.20188212e-01 -1.17848657e-01 6.45352781e-01 9.37895060e-01
4.21090364e-01 2.17180163e-01 7.96682835e-01 -7.60649085e-01
3.97656322e-01 -4.44595754e-01 -1.74559802e-01 2.53349334e-01
-9.00905967e-01 1.01838000e-01 6.17343962e-01 -4.27239567e-01
-9.87546027e-01 2.68227786e-01 -1.94808751e-01 -6.01447105e-01
-4.48731631e-01 5.40211737e-01 -1.89771071e-01 1.18389450e-01
8.85888159e-01 4.99260306e-01 -8.50895941e-02 -6.28153086e-01
4.69829381e-01 1.06930363e+00 4.96742278e-01 -5.48391163e-01
9.12595987e-01 2.63991151e-02 -8.55091751e-01 -1.20313156e+00
-7.66816318e-01 -9.64970708e-01 -8.92799199e-01 7.05482289e-02
6.66987598e-01 -8.01585495e-01 -2.08359316e-01 6.34553879e-02
-9.40392315e-01 -6.29771054e-01 -4.40501869e-01 6.92985356e-01
-5.57070673e-01 4.54309493e-01 -3.37264925e-01 -1.01617396e+00
-1.46774814e-01 -9.96703446e-01 4.07924145e-01 -9.77755040e-02
-3.12545687e-01 -1.05777550e+00 -6.04533777e-03 3.68015796e-01
4.85031694e-01 -8.30032289e-01 7.86616802e-01 -1.32206881e+00
-2.48360500e-01 -2.95262113e-02 9.27257761e-02 8.66142631e-01
5.66281021e-01 -2.72030175e-01 -1.45339394e+00 -3.96091849e-01
-4.78026122e-02 -3.73944432e-01 9.80771720e-01 1.25559047e-01
1.14905500e+00 -4.13278639e-01 -2.94691145e-01 4.94645059e-01
7.76525378e-01 4.40706313e-01 3.04934829e-01 8.23300779e-02
7.22504258e-01 6.29572749e-01 2.21957073e-01 2.66008079e-01
2.65413374e-01 4.07916874e-01 -9.61226299e-02 -2.86067069e-01
-2.76755005e-01 -3.25683445e-01 6.19724512e-01 1.55366206e+00
2.51581877e-01 -1.20953068e-01 -1.11093402e+00 7.01907754e-01
-1.58844292e+00 -9.76965189e-01 1.80183023e-01 2.25441456e+00
1.17002308e+00 4.75848317e-01 2.57137150e-01 3.74874681e-01
7.02882588e-01 -1.11227676e-01 -3.38859648e-01 -2.84894675e-01
-8.88468027e-02 4.74349737e-01 3.31393182e-01 7.97785938e-01
-1.17597401e+00 1.37690318e+00 6.49591446e+00 1.03793848e+00
-9.37831759e-01 3.17921042e-01 5.84930897e-01 9.11926478e-02
-2.11147428e-01 1.33804142e-01 -1.00912774e+00 3.27080488e-01
1.47762620e+00 -2.11647242e-01 4.49987680e-01 9.54994023e-01
1.66083783e-01 2.02481061e-01 -1.20954001e+00 8.26120555e-01
8.33419412e-02 -1.30439711e+00 -5.52464165e-02 -2.07163587e-01
7.68654764e-01 2.46415168e-01 1.23855114e-01 8.46502960e-01
5.76724052e-01 -9.69838262e-01 4.61608350e-01 -2.40404326e-02
5.80891728e-01 -8.53966713e-01 3.46011847e-01 5.21231651e-01
-1.04426849e+00 -2.43648365e-02 -4.54331934e-01 8.87276903e-02
-8.51058215e-02 3.68869007e-01 -1.67572200e+00 2.84371257e-01
4.17126805e-01 3.19229156e-01 -6.78253412e-01 1.03984272e+00
-1.16669402e-01 1.49336576e+00 -4.92518395e-01 -1.67154774e-01
4.81963545e-01 -7.81863928e-03 2.90860355e-01 1.69878340e+00
1.30675405e-01 1.22793438e-02 3.82889360e-01 5.44045925e-01
9.07406807e-02 2.60000646e-01 -3.22421581e-01 -1.80257022e-01
8.18327844e-01 1.04406917e+00 -6.69386268e-01 -6.65236592e-01
-4.56864029e-01 9.44899797e-01 3.85450542e-01 3.52658898e-01
-5.75713396e-01 -4.61561203e-01 5.73071122e-01 2.15464622e-01
4.45396304e-01 -4.89180624e-01 -4.54170763e-01 -8.88587892e-01
-4.11319494e-01 -8.17679405e-01 2.48810023e-01 -3.25411320e-01
-1.22137749e+00 8.38142455e-01 -8.48956555e-02 -1.22728443e+00
-5.37833154e-01 -4.74784136e-01 -7.67364681e-01 7.34304905e-01
-1.26842308e+00 -9.17012632e-01 3.62388082e-02 6.66671038e-01
8.20231259e-01 -6.80171490e-01 9.22506630e-01 2.42956728e-01
-5.14653563e-01 1.03350091e+00 2.59103656e-01 3.26273412e-01
9.56587851e-01 -1.54393339e+00 5.24357557e-01 8.23164165e-01
7.34417140e-01 9.13551152e-01 5.57583570e-01 -4.60647672e-01
-9.10725534e-01 -1.01898885e+00 1.04566205e+00 -5.62094688e-01
8.24856997e-01 -5.55517316e-01 -9.31649268e-01 7.91248739e-01
2.33615652e-01 -3.00041229e-01 9.77890909e-01 4.08258319e-01
-3.73204738e-01 -9.52598602e-02 -6.83682978e-01 4.89553332e-01
9.93316412e-01 -7.45673835e-01 -8.76313150e-01 4.10309732e-01
9.62946653e-01 -1.36746187e-02 -6.61948085e-01 2.25757912e-01
-1.02041066e-01 -9.29812968e-01 7.93698192e-01 -6.29028857e-01
-1.42349526e-01 -1.55698314e-01 -2.47049525e-01 -1.57777643e+00
-4.23167259e-01 -9.44252968e-01 -2.91605131e-03 1.49879611e+00
9.43415284e-01 -4.42931056e-01 9.03906107e-01 2.37922385e-01
-6.99069619e-01 -3.42863023e-01 -1.02627814e+00 -1.08115971e+00
2.08571702e-01 -7.68543482e-01 3.37200999e-01 1.16286147e+00
5.17326295e-01 6.63783371e-01 -1.43553887e-03 1.27188399e-01
4.43572819e-01 -1.86126083e-01 7.05251575e-01 -1.14999020e+00
-5.12598991e-01 -5.86474121e-01 -2.35874861e-01 -1.27217257e+00
3.61541927e-01 -1.34785342e+00 3.33331764e-01 -1.31249273e+00
-2.39447728e-01 -7.70975173e-01 -5.19238532e-01 8.54431331e-01
-1.01014018e-01 3.45381126e-02 1.03689760e-01 1.58349797e-01
-7.72676468e-01 5.11148572e-01 4.76273149e-01 -1.67936742e-01
-5.75936615e-01 3.48123789e-01 -5.05001009e-01 5.73335290e-01
1.10473990e+00 -4.39003706e-01 -6.91519797e-01 -1.97469592e-01
-4.48998392e-01 -4.18966562e-01 -2.47897387e-01 -1.30335855e+00
6.23703599e-01 1.63587689e-01 2.08350971e-01 -7.07569420e-01
4.84818399e-01 -6.91093802e-01 -4.47747171e-01 1.65491030e-01
-5.93812346e-01 -1.86296239e-01 2.06432819e-01 3.52274716e-01
-3.57517481e-01 -5.30300021e-01 8.63660991e-01 -1.02515802e-01
-8.71667206e-01 -3.29658464e-02 -8.20551932e-01 -4.63335998e-02
6.39532208e-01 -6.13554455e-02 4.24256548e-02 -5.61559856e-01
-1.16152215e+00 1.46619394e-01 5.69641851e-02 3.98596197e-01
5.91540277e-01 -1.24323332e+00 -6.64458632e-01 4.35192198e-01
8.00351501e-02 -1.63634256e-01 -1.20849237e-01 6.67809069e-01
6.25010505e-02 5.10800719e-01 4.29287523e-01 -7.36153901e-01
-1.25444603e+00 6.19285464e-01 1.51836425e-01 -1.40226200e-01
-2.09608048e-01 1.13573408e+00 -3.23122703e-02 -8.67292345e-01
7.59041011e-01 -3.68194938e-01 -1.22213000e-02 1.79270044e-01
4.82981473e-01 3.12391341e-01 4.51013803e-01 -3.42770070e-01
-3.41555387e-01 6.58358410e-02 -3.95398438e-01 -5.23326397e-01
1.21861589e+00 -2.67737895e-01 2.45544076e-01 7.95682847e-01
1.17254996e+00 9.54183340e-02 -9.79438901e-01 -5.92952907e-01
5.50642490e-01 2.37534434e-01 2.36245289e-01 -7.89498985e-01
-6.41725183e-01 1.01843083e+00 4.21243072e-01 4.23857808e-01
8.12818229e-01 1.86315224e-01 4.54760700e-01 8.01451325e-01
2.28544116e-01 -1.34272277e+00 2.18434766e-01 9.83725071e-01
5.44441760e-01 -1.24700403e+00 -3.41579080e-01 -2.73544312e-01
-6.69453740e-01 8.68193567e-01 7.09394634e-01 2.29023695e-01
8.52759778e-01 4.07227516e-01 3.32688421e-01 3.81289184e-01
-8.30675304e-01 -6.91569448e-01 3.40441823e-01 6.48181558e-01
6.04561806e-01 6.77567869e-02 3.37432884e-02 6.86159968e-01
-6.13465011e-01 -5.32851517e-01 2.63517261e-01 9.64504123e-01
-1.04611063e+00 -1.21224236e+00 -4.65574324e-01 3.11203182e-01
9.52082649e-02 -4.90977943e-01 -6.56800032e-01 4.20803756e-01
-1.28671899e-01 1.37731361e+00 2.06143752e-01 -8.36248636e-01
1.02739118e-01 6.81923687e-01 1.61573723e-01 -1.19315994e+00
-7.40113139e-01 3.68662477e-01 7.22889975e-02 -5.15384637e-02
-2.27120072e-01 -5.09042680e-01 -1.46583867e+00 2.70126350e-02
-4.81630206e-01 6.76454961e-01 6.52569354e-01 9.45356548e-01
2.98492223e-01 5.03725290e-01 1.11050844e+00 -8.05897176e-01
-6.36340499e-01 -1.34348190e+00 -2.84847617e-01 6.29463792e-02
2.38586515e-01 -4.40944105e-01 -7.38311052e-01 2.39648476e-01] | [14.428496360778809, 6.604561805725098] |
66feb5d4-3673-496c-96ce-338623fc4859 | improving-code-generation-by-training-with | 2303.16749 | null | https://arxiv.org/abs/2303.16749v1 | https://arxiv.org/pdf/2303.16749v1.pdf | Improving Code Generation by Training with Natural Language Feedback | The potential for pre-trained large language models (LLMs) to use natural language feedback at inference time has been an exciting recent development. We build upon this observation by formalizing an algorithm for learning from natural language feedback at training time instead, which we call Imitation learning from Language Feedback (ILF). ILF requires only a small amount of human-written feedback during training and does not require the same feedback at test time, making it both user-friendly and sample-efficient. We further show that ILF can be seen as a form of minimizing the KL divergence to the ground truth distribution and demonstrate a proof-of-concept on a neural program synthesis task. We use ILF to improve a Codegen-Mono 6.1B model's pass@1 rate by 38% relative (and 10% absolute) on the Mostly Basic Python Problems (MBPP) benchmark, outperforming both fine-tuning on MBPP and fine-tuning on repaired programs written by humans. Overall, our results suggest that learning from human-written natural language feedback is both more effective and sample-efficient than training exclusively on demonstrations for improving an LLM's performance on code generation tasks. | ['Ethan Perez', 'Kyunghyun Cho', 'Samuel R. Bowman', 'Jun Shern Chan', 'Jon Ander Campos', 'Tomasz Korbak', 'Jérémy Scheurer', 'Angelica Chen'] | 2023-03-28 | null | null | null | null | ['program-synthesis'] | ['computer-code'] | [ 3.43133174e-02 3.41899425e-01 -2.25396484e-01 -4.40988660e-01
-1.01240993e+00 -6.62079513e-01 6.05574846e-01 1.43000960e-01
-6.64702713e-01 6.68179512e-01 -4.77434322e-02 -8.59943748e-01
4.99339283e-01 -6.54416502e-01 -1.59377742e+00 -5.40758632e-02
-2.17731103e-01 2.66474336e-01 9.51370373e-02 -1.16910823e-01
3.70378405e-01 1.05654523e-01 -1.34425712e+00 3.83615792e-01
1.12442696e+00 3.44611615e-01 5.45616925e-01 1.17595327e+00
-3.81107740e-02 1.10505211e+00 -5.91193318e-01 -2.60651290e-01
2.43245602e-01 -4.59933132e-01 -9.75863218e-01 -3.32945347e-01
4.85322744e-01 -5.60075998e-01 2.90520731e-02 9.65323329e-01
3.75179082e-01 6.24560416e-02 4.29277152e-01 -1.24375999e+00
-7.34156191e-01 8.14233720e-01 -3.78197044e-01 -6.49390668e-02
5.04644573e-01 6.74034715e-01 9.36616480e-01 -8.57408047e-01
6.07198894e-01 1.36335719e+00 8.35276842e-01 7.73642540e-01
-1.60756791e+00 -5.26545942e-01 -7.46905729e-02 -3.48225564e-01
-1.03544247e+00 -4.25601751e-01 1.45713622e-02 -7.34148443e-01
1.62160587e+00 8.23738500e-02 2.36001864e-01 8.69868517e-01
1.97144598e-01 1.05225241e+00 9.15180922e-01 -7.30924308e-01
1.10106371e-01 3.28032553e-01 2.49417722e-02 1.23618507e+00
-3.22636007e-03 4.82807636e-01 -4.00985837e-01 -2.78625488e-01
6.94587827e-01 -3.90707225e-01 -3.95688504e-01 -5.05454302e-01
-1.25838649e+00 9.31437612e-01 4.38341379e-01 -5.96795529e-02
-1.02881929e-02 7.88113952e-01 6.12194300e-01 6.77840531e-01
4.68765832e-02 9.13279891e-01 -6.27716124e-01 -6.44758105e-01
-9.52168107e-01 4.36963171e-01 9.69099522e-01 1.10669827e+00
9.20137465e-01 1.39577016e-01 -1.10810466e-01 7.18680382e-01
3.76548648e-01 5.45147955e-01 8.22398007e-01 -1.24979162e+00
6.01583123e-01 5.04046679e-01 1.07718714e-01 -4.95692641e-01
-1.07227139e-01 -1.06570870e-01 -2.89212555e-01 5.90860665e-01
5.08035183e-01 -3.84791940e-01 -5.22019684e-01 2.03439903e+00
-1.51661038e-01 -2.34221518e-01 5.38888685e-02 4.00588065e-01
3.79248500e-01 8.59568834e-01 -1.61261544e-01 9.20401067e-02
6.17145658e-01 -1.42755258e+00 7.80798867e-02 -2.61591613e-01
1.17651141e+00 -5.40678680e-01 1.74822915e+00 4.94815111e-01
-1.25093424e+00 -6.56568408e-01 -1.10603487e+00 -1.36832997e-01
-3.27499621e-02 1.50271773e-01 6.79578066e-01 6.09312117e-01
-1.50751293e+00 9.90431190e-01 -1.10673940e+00 -2.54771829e-01
1.93843901e-01 3.67400646e-01 -3.23633343e-01 5.29151484e-02
-6.05633438e-01 8.89468491e-01 3.92740071e-01 -3.70904207e-01
-1.21370709e+00 -1.10815012e+00 -9.88340914e-01 4.98596057e-02
1.37329042e-01 -5.96373498e-01 1.97885287e+00 -1.15947402e+00
-1.85361707e+00 8.22378993e-01 -2.27663238e-02 -7.63259888e-01
7.44703948e-01 -3.46962273e-01 1.80497825e-01 -1.19803071e-01
6.78950101e-02 1.01825154e+00 6.57124519e-01 -9.15696859e-01
-3.41305703e-01 4.45777893e-01 3.17484945e-01 -2.37673193e-01
-1.48797810e-01 4.61114980e-02 -5.13931550e-03 -4.52299029e-01
-7.33354449e-01 -1.07879996e+00 -1.07446074e-01 3.23008895e-01
-6.20600097e-02 -4.61396933e-01 4.25326169e-01 -7.32627809e-01
9.16108608e-01 -2.02966976e+00 4.48196195e-02 -1.13985851e-01
8.20073560e-02 3.89003873e-01 -5.30094385e-01 5.44899344e-01
-9.24600568e-03 2.56344885e-01 -5.03262281e-01 -2.48552501e-01
3.30252618e-01 5.30205928e-02 -5.73803127e-01 2.99660355e-01
4.50736314e-01 1.20856524e+00 -1.26181984e+00 -1.67990193e-01
-9.19152871e-02 5.55339530e-02 -1.32359409e+00 4.99666303e-01
-7.05348372e-01 2.76863098e-01 1.61121860e-01 2.63737500e-01
2.07714424e-01 -4.02214170e-01 -1.38829157e-01 5.68401515e-01
-1.29708141e-01 5.86185038e-01 -7.84423947e-01 2.01245427e+00
-1.05303621e+00 9.39099848e-01 -5.15537746e-02 -8.01697493e-01
7.25125670e-01 2.12917298e-01 -2.77498454e-01 -6.83617473e-01
-3.49154770e-01 5.46181381e-01 2.83708155e-01 -5.58025777e-01
2.78527826e-01 5.50475866e-02 -2.17331038e-03 8.85888338e-01
2.80456215e-01 -5.63557565e-01 3.43541801e-01 3.20682317e-01
1.31360304e+00 5.98017633e-01 1.76266104e-01 -3.66893619e-01
3.90381575e-01 -4.12358008e-02 1.00328356e-01 1.21197665e+00
3.20958979e-02 2.65492499e-01 5.32226384e-01 -1.44339234e-01
-1.25258744e+00 -1.03024852e+00 2.23239079e-01 1.47389412e+00
-5.04016936e-01 -6.66222036e-01 -8.77662480e-01 -7.16272950e-01
1.44649163e-01 1.16860783e+00 -4.24304426e-01 -2.34154180e-01
-6.04056537e-01 -1.91502407e-01 8.17540944e-01 6.92105770e-01
3.03796619e-01 -1.36226082e+00 -8.83584738e-01 2.15541109e-01
9.73356515e-02 -5.61909199e-01 -7.46824026e-01 3.07962894e-01
-9.74290848e-01 -7.79126465e-01 -7.40851700e-01 -7.53554285e-01
7.91735947e-01 3.43043543e-02 1.53541732e+00 4.15621787e-01
-1.18878014e-01 4.23119217e-01 -1.61712468e-01 -3.50368440e-01
-1.23804569e+00 1.69577509e-01 -5.28127737e-02 -8.12109053e-01
-4.92464751e-02 -6.04115427e-01 -2.42243364e-01 1.04431883e-01
-6.24299228e-01 3.94496381e-01 7.20312536e-01 1.17376316e+00
-2.11361330e-03 -5.26161909e-01 5.32291114e-01 -8.36192131e-01
7.91139603e-01 -2.32168660e-01 -1.03455794e+00 2.01877087e-01
-6.26677334e-01 6.83501542e-01 9.79445755e-01 -4.49545324e-01
-7.98713565e-01 7.75653794e-02 -1.69189826e-01 -3.27642083e-01
1.18906349e-01 4.74862278e-01 3.86230260e-01 -2.86455065e-01
1.16172349e+00 4.28965956e-01 2.36377656e-01 -2.82686204e-01
5.78761458e-01 5.58089554e-01 5.07362723e-01 -1.09049201e+00
7.22980618e-01 -1.63452491e-01 -5.36162555e-01 -4.92744446e-01
-4.48997974e-01 -1.00641251e-01 -3.05465221e-01 1.57900780e-01
4.46300328e-01 -9.17344630e-01 -8.84653866e-01 3.75350058e-01
-1.32187295e+00 -1.28254199e+00 -2.20583990e-01 2.34517455e-01
-8.95369589e-01 2.70046145e-01 -7.68659234e-01 -6.04359806e-01
-4.98199135e-01 -1.39282310e+00 9.90799010e-01 -1.11223776e-02
-6.34741902e-01 -9.45359170e-01 1.80285856e-01 1.55839445e-02
6.62559092e-01 -1.18236154e-01 1.06078506e+00 -4.57785875e-01
-7.95358598e-01 -9.97028276e-02 -2.17234477e-01 7.55187988e-01
-2.31697634e-01 3.45945805e-02 -8.73402357e-01 -5.95880270e-01
-5.71816117e-02 -9.30260777e-01 7.00418293e-01 -1.15290828e-01
9.83043551e-01 -5.98309517e-01 -3.75904068e-02 5.46418667e-01
1.29238355e+00 -2.56628394e-01 2.28608891e-01 3.11602056e-01
5.39782822e-01 3.75307381e-01 3.69304508e-01 2.26543680e-01
1.72350690e-01 4.83038723e-01 2.52034575e-01 2.09680080e-01
-4.05664854e-02 -6.01666451e-01 9.84063506e-01 8.14310610e-01
3.62758547e-01 2.79710948e-01 -1.05060744e+00 6.66470110e-01
-1.78320897e+00 -7.10124791e-01 1.26930296e-01 2.29382515e+00
1.50970042e+00 3.28681588e-01 2.24843517e-01 -2.82965213e-01
2.58376688e-01 -2.94486046e-01 -5.56611896e-01 -9.17599499e-01
3.92022014e-01 4.87063050e-01 4.48330641e-01 8.13453853e-01
-6.71683133e-01 8.41037810e-01 6.65984344e+00 7.05561996e-01
-1.24976087e+00 8.42647851e-02 3.62523675e-01 -7.32115656e-03
-3.81249666e-01 9.49679911e-02 -5.62844396e-01 4.12691683e-01
1.43031800e+00 -5.09465039e-01 9.31677282e-01 1.33973885e+00
7.26579726e-02 -2.73916066e-01 -1.69853663e+00 7.72160769e-01
-2.29937315e-01 -1.45206416e+00 -1.85682967e-01 -1.91795573e-01
8.08659375e-01 5.05959868e-01 -2.21668094e-01 1.01117933e+00
7.65593588e-01 -1.29142046e+00 9.40087020e-01 1.53799236e-01
8.14972162e-01 -5.55693150e-01 3.61019611e-01 9.64221299e-01
-7.33080029e-01 4.01001163e-02 -3.82402867e-01 -4.92586821e-01
-1.92328990e-01 1.34266943e-01 -1.42568219e+00 -2.24833954e-02
3.65708143e-01 4.81138945e-01 -7.13375449e-01 9.49864924e-01
-5.83074331e-01 8.36592257e-01 -3.12722772e-01 -3.81541371e-01
2.55013227e-01 3.01210582e-01 1.95336238e-01 1.55742455e+00
2.10045904e-01 -5.19266307e-01 2.47856751e-01 1.47479558e+00
-2.91920871e-01 -5.19404300e-02 -6.29378378e-01 -3.93529326e-01
2.42803290e-01 9.24123406e-01 -1.73988938e-01 -4.51056868e-01
-4.06496167e-01 1.03763533e+00 6.07819021e-01 1.00919560e-01
-8.61267805e-01 -5.94160259e-01 3.30505371e-01 -3.48927081e-02
2.26454362e-01 -3.39247018e-01 -1.06820628e-01 -1.09844387e+00
9.76406112e-02 -1.21949685e+00 -1.99173853e-01 -8.65666926e-01
-9.07475352e-01 4.00799096e-01 -1.01528905e-01 -9.84567165e-01
-7.86077321e-01 -7.64715850e-01 -5.39841056e-01 1.14331806e+00
-1.30262816e+00 -6.47981942e-01 2.17722971e-02 1.47658065e-01
5.61057031e-01 -8.09509456e-02 1.09245956e+00 1.24483295e-01
-1.22232810e-01 9.77211952e-01 3.00670099e-02 9.93217379e-02
5.40070713e-01 -1.61614954e+00 9.35201466e-01 8.93025935e-01
1.35297671e-01 1.09557498e+00 8.27531159e-01 -3.51602733e-01
-1.59196353e+00 -9.85131145e-01 7.84173131e-01 -6.01504803e-01
8.78631592e-01 -5.36339521e-01 -8.86985481e-01 8.26110244e-01
3.19322765e-01 -8.97442177e-02 2.96564668e-01 -5.98701760e-02
-6.14607871e-01 3.11464101e-01 -9.36965585e-01 6.81251347e-01
8.94432664e-01 -8.15342247e-01 -7.02125728e-01 5.44544518e-01
1.07929254e+00 -6.74684763e-01 -7.14826107e-01 6.46078587e-02
3.61461490e-01 -8.39689076e-01 7.21635163e-01 -5.78895152e-01
9.16029990e-01 -2.46290684e-01 -9.92494076e-02 -1.41573012e+00
5.04782721e-02 -9.16725218e-01 -5.19458354e-02 1.04611957e+00
6.37728453e-01 -5.79406500e-01 5.63458145e-01 5.39395809e-01
-1.88203543e-01 -7.99291611e-01 -4.38324809e-01 -1.04717720e+00
5.92292547e-01 -5.38634539e-01 1.12790301e-01 5.61803162e-01
2.28360265e-01 3.76199245e-01 -1.14614248e-01 -2.10781440e-01
2.07830310e-01 1.46881104e-01 1.25180483e+00 -7.28094280e-01
-1.12376177e+00 -5.71094215e-01 3.61849703e-02 -1.38073790e+00
3.95367444e-01 -1.33486533e+00 5.13493717e-01 -1.09354365e+00
2.57137597e-01 -3.56280506e-01 1.78720176e-01 6.72550082e-01
-1.18072733e-01 -2.33182326e-01 4.44410443e-01 1.74012080e-01
-5.92288613e-01 4.22702342e-01 9.31174636e-01 -1.05845317e-01
-1.17947869e-01 -6.49388507e-02 -5.45066535e-01 6.95037961e-01
4.95458156e-01 -5.78752518e-01 -4.21364874e-01 -6.21730149e-01
4.42452550e-01 2.21459061e-01 3.79482806e-01 -1.03072476e+00
1.12631164e-01 -9.50760096e-02 -1.72572821e-01 -2.94619855e-02
-2.35438779e-01 -2.49932632e-01 -2.94396132e-01 8.74138713e-01
-7.78823555e-01 2.45647907e-01 6.40495300e-01 3.25312048e-01
2.70406939e-02 -6.83345973e-01 7.54714847e-01 -4.24897552e-01
-7.78670490e-01 -1.86473817e-01 -4.03911889e-01 3.97951484e-01
6.04548812e-01 1.08766086e-01 -3.68871629e-01 -3.51136684e-01
-1.51765227e-01 2.95599133e-01 8.00019801e-01 3.74726027e-01
3.09109390e-01 -9.40676868e-01 -6.95864618e-01 2.78722584e-01
1.88040420e-01 1.55826854e-02 -3.62395406e-01 8.14555228e-01
-9.00293350e-01 5.63513994e-01 -5.36953919e-02 -6.95885122e-01
-1.05495226e+00 5.46985745e-01 4.43379074e-01 -3.98817778e-01
-4.83100057e-01 1.35983121e+00 2.01871067e-01 -1.21211433e+00
3.08943182e-01 -7.34208226e-01 7.46650517e-01 -6.99279010e-01
5.23429513e-01 3.58482376e-02 3.86952702e-03 1.50174171e-01
-1.93846896e-01 5.25891148e-02 -3.30851302e-02 -4.23858404e-01
1.36559129e+00 5.49493551e-01 -2.00367600e-01 6.60808623e-01
1.34657431e+00 7.56642297e-02 -1.47940004e+00 -2.44815409e-01
2.86530644e-01 -2.79442221e-01 -3.70075464e-01 -1.06607020e+00
-5.25626600e-01 1.21905768e+00 3.60289276e-01 3.74165270e-03
6.10230207e-01 -3.06109607e-01 7.01817036e-01 9.98893142e-01
6.99053943e-01 -8.39088023e-01 4.94336903e-01 7.34500945e-01
1.21100807e+00 -1.28197885e+00 -3.61849517e-01 1.42237887e-01
-4.84418780e-01 1.23678112e+00 6.41932964e-01 -4.40216482e-01
2.63714403e-01 4.19226855e-01 -2.62244940e-01 3.63019198e-01
-1.07601440e+00 4.42174107e-01 6.75558448e-02 5.22599399e-01
8.51408124e-01 -2.61879936e-02 4.03130502e-02 2.77441859e-01
-5.14274418e-01 4.43172306e-01 6.53802454e-01 1.12369072e+00
-4.61417139e-01 -1.22369719e+00 -8.51065889e-02 3.50335896e-01
-2.54793137e-01 -4.52212065e-01 -1.03906572e-01 8.13831270e-01
-2.36220896e-01 6.28478229e-01 -2.21027136e-01 -2.22202718e-01
2.56859846e-02 2.46696025e-01 7.29710221e-01 -1.17569780e+00
-7.45734870e-01 -5.52409470e-01 1.00026801e-01 -6.83751047e-01
5.89578133e-03 -4.46319699e-01 -1.30637681e+00 -3.85779589e-01
-2.63917267e-01 6.22942820e-02 6.70277059e-01 7.75323510e-01
4.27821815e-01 4.58135813e-01 2.77123451e-01 -9.96641815e-01
-1.25022137e+00 -9.82696176e-01 3.21699455e-02 1.67615816e-01
4.95030850e-01 -1.88469693e-01 -3.63018811e-01 1.50510967e-01] | [8.057413101196289, 7.6296067237854] |
4e18952b-5609-46ec-b753-33c18e9c82b8 | ai-in-pursuit-of-happiness-finding-only | 1911.05187 | null | https://arxiv.org/abs/1911.05187v1 | https://arxiv.org/pdf/1911.05187v1.pdf | AI in Pursuit of Happiness, Finding Only Sadness: Multi-Modal Facial Emotion Recognition Challenge | The importance of automated Facial Emotion Recognition (FER) grows the more common human-machine interactions become, which will only continue to increase dramatically with time. A common method to describe human sentiment or feeling is the categorical model the `7 basic emotions', consisting of `Angry', `Disgust', `Fear', `Happiness', `Sadness', `Surprise' and `Neutral'. The `Emotion Recognition in the Wild' (EmotiW) competition is now in its 7th year and has become the standard benchmark for measuring FER performance. The focus of this paper is the EmotiW sub-challenge of classifying videos in the `Acted Facial Expression in the Wild' (AFEW) dataset, consisting of both visual and audio modalities, into one of the above classes. Machine learning has exploded as a research topic in recent years, with advancements in `Deep Learning' a key part of this. Although Deep Learning techniques have been widely applied to the FER task by entrants in previous years, this paper has two main contributions: (i) to apply the latest `state-of-the-art' visual and temporal networks and (ii) exploring various methods of fusing features extracted from the visual and audio elements to enrich the information available to the final model making the prediction. There are a number of complex issues that arise when trying to classify emotions for `in-the-wild' video sequences, which the above two approaches attempt to directly address. There are some positive findings when comparing the results of this paper to past submissions, indicating that further research into the proposed methods and fine-tuning of the models deployed, could result in another step forwards in the field of automated FER. | ['Carl Norman'] | 2019-10-24 | null | null | null | null | ['facial-emotion-recognition'] | ['computer-vision'] | [ 4.06455621e-02 -2.57379979e-01 1.73959285e-01 -6.62139654e-01
-3.73915583e-01 -3.06318402e-01 4.87997442e-01 -1.13820046e-01
-4.91032213e-01 5.55273592e-01 3.27986807e-01 3.05418670e-01
6.52569234e-02 -2.51241654e-01 -1.26257241e-01 -6.42011702e-01
-4.67720121e-01 -1.22988440e-01 -1.46611676e-01 -6.92395568e-01
-6.99323192e-02 5.70640326e-01 -2.20281291e+00 7.55858958e-01
-5.47322594e-02 1.82681310e+00 -4.68309224e-01 7.74311423e-01
3.79335098e-02 1.35695565e+00 -6.51054859e-01 -7.57404745e-01
-1.30592644e-01 -4.72570419e-01 -9.68127191e-01 -5.86989261e-02
1.03083462e-01 3.89433168e-02 1.69778671e-02 7.59553134e-01
8.29806149e-01 1.45642132e-01 4.51913089e-01 -1.66136432e+00
-2.11563542e-01 -4.92936857e-02 -6.44047558e-01 2.70268679e-01
7.38014817e-01 -1.29986897e-01 8.38958979e-01 -9.12941575e-01
7.86628008e-01 1.24566913e+00 1.09299004e+00 6.79582179e-01
-8.38719368e-01 -8.54730427e-01 -1.34049654e-01 5.09804785e-01
-1.27576065e+00 -6.59024239e-01 8.33101094e-01 -5.73749125e-01
1.15519941e+00 5.14185190e-01 7.62125909e-01 1.63798475e+00
-1.33558333e-01 8.86589944e-01 1.53260696e+00 -4.90488440e-01
-1.35923913e-02 4.58924741e-01 -2.42400169e-01 5.18222749e-01
-5.74654579e-01 3.10790271e-01 -8.60586762e-01 -1.41950354e-01
1.94192186e-01 -3.04133266e-01 -2.30636746e-01 -2.58361939e-02
-7.07182646e-01 9.47827101e-01 1.57076672e-01 7.02631533e-01
-6.04129076e-01 1.65976763e-01 1.09495676e+00 6.90190971e-01
9.03298199e-01 2.42467150e-01 -4.99589324e-01 -8.22649896e-01
-9.31991935e-01 2.66510040e-01 7.81653404e-01 2.71829247e-01
4.98992145e-01 2.95004934e-01 6.03520982e-02 1.16950977e+00
3.45420480e-01 1.01875499e-01 4.29028600e-01 -1.02633345e+00
-2.40107834e-01 3.55653107e-01 3.04945987e-02 -1.27826643e+00
-4.54392701e-01 1.03231952e-01 -8.70533407e-01 6.39585853e-01
8.92015994e-02 -5.63341200e-01 -6.58503234e-01 1.77647626e+00
1.52818352e-01 -4.44495715e-02 -4.00924981e-02 8.60391080e-01
1.18519402e+00 6.24122977e-01 2.65543193e-01 -3.69691581e-01
1.40978146e+00 -5.88324785e-01 -1.00779843e+00 1.35177717e-01
3.92770737e-01 -1.08486414e+00 5.37008643e-01 7.88945794e-01
-9.21204388e-01 -7.05467045e-01 -8.56911242e-01 3.26354772e-01
-7.98424542e-01 7.04480633e-02 8.49356234e-01 7.39481568e-01
-1.27056897e+00 7.23952532e-01 -4.88001227e-01 -5.35397351e-01
5.40204346e-01 4.28466529e-01 -9.18328643e-01 3.17437530e-01
-1.34555864e+00 1.18709934e+00 -8.14275146e-02 4.18080419e-01
-5.91133773e-01 -3.48258793e-01 -8.26431036e-01 -2.45143503e-01
1.08770564e-01 -6.54151887e-02 1.21587074e+00 -1.92116654e+00
-1.51401818e+00 1.43085122e+00 -5.05218953e-02 -7.75579661e-02
3.04688156e-01 -2.70914465e-01 -8.98349941e-01 9.40934494e-02
-3.75557154e-01 8.27145636e-01 8.95753264e-01 -1.26151764e+00
-6.12872481e-01 -2.82927573e-01 -2.45323889e-02 3.05966358e-03
-1.55860722e-01 9.26100492e-01 -3.99478190e-02 -5.75865388e-01
-5.88178217e-01 -8.68088305e-01 1.07127167e-01 -1.09009456e-03
4.53375652e-02 -4.62389588e-01 9.67353225e-01 -5.98911822e-01
1.24222994e+00 -2.41185355e+00 2.66291171e-01 1.07661225e-01
1.09526210e-01 4.72992212e-01 5.47716171e-02 6.37108862e-01
-6.88439548e-01 1.57238081e-01 1.65053546e-01 -4.82812852e-01
2.01968715e-01 1.32054374e-01 -1.02010816e-01 4.72665966e-01
3.34399819e-01 5.27750611e-01 -6.67414665e-01 -4.41245675e-01
1.18583247e-01 1.00067508e+00 -2.10938707e-01 3.86588097e-01
3.76344413e-01 3.48313600e-01 -2.91001685e-02 6.78798139e-01
4.29381877e-01 3.55132371e-01 -1.15453623e-01 -4.07370657e-01
-2.32839808e-01 -2.44724140e-01 -1.08034360e+00 1.24827838e+00
-2.40938365e-01 1.17452872e+00 3.94874930e-01 -1.29039478e+00
1.00645757e+00 9.48250234e-01 7.52759933e-01 -6.10017836e-01
6.60200477e-01 1.44218460e-01 -1.58598319e-01 -9.91029739e-01
2.14098826e-01 -5.66072702e-01 -1.72039509e-01 1.43669188e-01
4.96910304e-01 2.56349951e-01 -7.98644796e-02 -2.61077791e-01
9.55367148e-01 3.90480816e-01 2.57463843e-01 -7.22575886e-03
7.16232419e-01 -3.33432585e-01 4.49088007e-01 2.23855719e-01
-7.88895011e-01 4.24522221e-01 6.52282059e-01 -7.52180338e-01
-5.99226594e-01 -6.16509378e-01 4.16910984e-02 1.46666896e+00
-4.33902770e-01 -3.91099662e-01 -6.45726740e-01 -5.26026189e-01
-3.68967563e-01 1.50252864e-01 -1.07678878e+00 -2.56084204e-02
-1.30048051e-01 -5.49274385e-01 7.73989618e-01 5.34156978e-01
4.46330458e-01 -1.48238766e+00 -1.03334320e+00 7.17421025e-02
-4.17512447e-01 -9.46734250e-01 3.00187945e-01 5.29266894e-01
-2.17180908e-01 -8.15842807e-01 -8.46886456e-01 -5.40537119e-01
-1.59000214e-02 -2.10995138e-01 1.23325658e+00 6.32916242e-02
-5.57595849e-01 7.54629731e-01 -9.60426152e-01 -8.12825263e-01
-1.58712253e-01 -6.15162432e-01 4.97097485e-02 3.84434909e-01
7.42761254e-01 -5.82782924e-01 -4.85765904e-01 1.69343024e-01
-8.39411676e-01 -3.68833512e-01 1.65230155e-01 8.75512362e-01
1.19037755e-01 4.60392870e-02 6.03505075e-01 -5.70599854e-01
7.38713682e-01 -4.88153726e-01 2.04364017e-01 -1.03568956e-01
-1.22148298e-01 -4.97866631e-01 2.85235524e-01 -4.32615578e-01
-1.13988411e+00 -1.73394978e-01 -6.87812567e-01 -6.86587691e-01
-6.52748644e-01 5.43781817e-01 8.74651149e-02 -3.33118230e-01
6.30263090e-01 -2.83749133e-01 5.27228378e-02 -1.17851496e-01
8.05918872e-02 8.96916568e-01 4.57597524e-01 -3.53812724e-01
2.59819865e-01 4.24631625e-01 -5.22134453e-02 -9.13554788e-01
-6.17699623e-01 -6.58623517e-01 -5.07016778e-01 -9.79754031e-01
1.04308307e+00 -9.54432726e-01 -1.12021399e+00 6.03712261e-01
-1.06515813e+00 -1.05140395e-01 -3.21162581e-01 2.94756114e-01
-5.48446894e-01 2.59545922e-01 -6.68500960e-01 -1.29442132e+00
-3.07533473e-01 -1.03112602e+00 1.09342945e+00 2.11443126e-01
-7.46417820e-01 -9.15898561e-01 3.32837790e-01 2.35968843e-01
6.87822342e-01 8.82565737e-01 4.76179570e-01 -4.55271274e-01
6.34898961e-01 -4.42081839e-01 -2.14788258e-01 8.55406702e-01
-1.28472894e-01 3.08383971e-01 -1.59066010e+00 -9.85738337e-02
1.88250840e-01 -1.15277684e+00 5.28739691e-01 1.15280837e-01
9.92530465e-01 -1.93434227e-02 5.20326644e-02 3.11772048e-01
1.37173080e+00 4.26214427e-01 9.47223604e-01 3.54204208e-01
9.28258896e-02 9.67001855e-01 8.37188005e-01 7.51946449e-01
7.54501522e-02 7.66736805e-01 5.56961358e-01 -3.16375554e-01
1.74809560e-01 2.95114964e-01 6.88230872e-01 5.84329367e-01
-8.98988962e-01 1.04047172e-02 -6.45913184e-01 3.05166274e-01
-1.61063910e+00 -1.49476171e+00 8.13934579e-02 1.66221344e+00
5.46877742e-01 -2.81531036e-01 3.90534312e-01 5.77746332e-01
4.57888305e-01 3.16686660e-01 1.51098296e-01 -1.26519847e+00
-1.14784196e-01 6.83918834e-01 -3.21390897e-01 1.64586395e-01
-1.40586281e+00 7.27377117e-01 5.52582884e+00 7.32372522e-01
-1.63482594e+00 -2.69775558e-02 7.84829080e-01 3.91248688e-02
3.29590768e-01 -4.28348035e-01 -4.21343297e-01 3.47746015e-01
1.22744286e+00 3.45823824e-01 4.60066229e-01 1.00912213e+00
2.58005798e-01 -1.94470152e-01 -9.28150177e-01 1.43195963e+00
5.33237159e-01 -7.60927916e-01 -4.87231523e-01 -2.64458418e-01
2.45602608e-01 -6.84596375e-02 1.85597643e-01 6.13056123e-01
-1.81321859e-01 -1.24178600e+00 6.67613089e-01 5.90403974e-01
9.31799948e-01 -1.01537848e+00 1.00262630e+00 -1.20442756e-01
-1.11683023e+00 -8.06973651e-02 -1.76327482e-01 -2.57628173e-01
1.22880556e-01 1.96352869e-01 -2.99713731e-01 4.74205285e-01
1.55831146e+00 7.26276159e-01 -3.14099997e-01 7.19916701e-01
2.12903678e-01 4.10195261e-01 -1.50320470e-01 -1.75085142e-02
2.92243659e-01 1.29843131e-01 3.53147477e-01 1.70212936e+00
2.10012719e-01 8.27742964e-02 -2.49654010e-01 3.29472780e-01
1.95910975e-01 2.63389975e-01 -7.23303199e-01 -1.31144300e-01
-2.32221931e-01 1.66360748e+00 -6.67239726e-01 -8.22752416e-02
-4.05130655e-01 1.16249323e+00 1.51555479e-01 2.96524882e-01
-1.09507704e+00 -5.56927025e-01 7.98371732e-01 -3.12496781e-01
2.99378783e-01 3.09415162e-01 3.72005463e-01 -8.51795971e-01
-2.84475535e-01 -1.18952656e+00 4.27228957e-01 -1.24027014e+00
-1.32362020e+00 1.16079402e+00 -6.94986284e-02 -1.02922547e+00
-3.02391022e-01 -1.01481390e+00 -4.78825480e-01 6.13664627e-01
-1.30947161e+00 -1.09359431e+00 -4.66217697e-01 8.25301528e-01
2.60663450e-01 -1.12768106e-01 1.34532213e+00 6.61737204e-01
-2.15857983e-01 5.91786146e-01 -3.27364713e-01 1.13511138e-01
9.43463922e-01 -8.24228108e-01 -5.47974348e-01 2.59225816e-01
1.19769618e-01 1.25436127e-01 9.33984756e-01 5.87752946e-02
-1.00709271e+00 -5.57902873e-01 1.11193860e+00 -2.42608830e-01
5.51847041e-01 -3.54882151e-01 -6.08267903e-01 5.53895712e-01
6.88672960e-01 -7.34364539e-02 1.19131088e+00 1.02389812e-01
-4.16327924e-01 -6.11486994e-02 -1.28043258e+00 3.31032753e-01
5.66473067e-01 -6.51321590e-01 -4.21417266e-01 -4.99835946e-02
-8.16171542e-02 -1.54779926e-01 -1.17392540e+00 7.53611505e-01
9.51966226e-01 -1.60456359e+00 8.03735316e-01 -7.23963916e-01
6.20436668e-01 1.59378394e-01 -3.13352466e-01 -1.35378218e+00
-1.30431026e-01 -6.71620548e-01 1.42199874e-01 1.49401510e+00
4.71309870e-02 -2.43431717e-01 6.67734563e-01 3.98784339e-01
3.36349346e-02 -1.06894577e+00 -1.10729301e+00 -2.38254786e-01
-2.83924073e-01 -9.40126181e-01 1.26064524e-01 1.09990966e+00
2.53188640e-01 4.94176090e-01 -7.40451872e-01 -5.58613837e-01
-8.06809515e-02 -3.60568702e-01 6.73251510e-01 -1.38835526e+00
-2.41890065e-02 -6.15323007e-01 -9.48479056e-01 -2.30735257e-01
2.11331353e-01 -4.42826629e-01 2.80696694e-02 -9.55764234e-01
7.26871416e-02 3.65800709e-02 -3.59772921e-01 5.57081997e-01
2.35072240e-01 7.80653000e-01 3.19582254e-01 -3.28538507e-01
-6.97798669e-01 4.10179257e-01 7.91145563e-01 9.85527113e-02
1.53044075e-01 -1.87352404e-01 -5.35770595e-01 9.74094987e-01
5.66904724e-01 -7.98492804e-02 -1.50564268e-01 4.65016365e-02
5.72032154e-01 2.97362860e-02 3.17722350e-01 -8.96186829e-01
-2.44273767e-01 1.40983626e-01 5.36984682e-01 -2.04369918e-01
9.44499254e-01 -1.07095635e+00 2.89536685e-01 9.76264551e-02
-1.64054811e-01 2.53736496e-01 5.03293574e-01 1.98819250e-01
-7.73039579e-01 -8.13751966e-02 1.10806119e+00 -1.66231006e-01
-1.02017450e+00 1.89287029e-02 -6.94015861e-01 -1.44404292e-01
1.09182346e+00 -4.87481385e-01 1.33771315e-01 -9.32039976e-01
-1.21352351e+00 -2.47785866e-01 8.09854344e-02 5.44590771e-01
6.27347291e-01 -1.28828573e+00 -5.21405041e-01 -6.45524561e-02
1.52964562e-01 -7.82413661e-01 5.55765867e-01 1.23294652e+00
-3.21044743e-01 1.33812711e-01 -5.70506454e-01 -4.90254760e-01
-1.80563843e+00 4.29540992e-01 4.57724005e-01 -2.25781053e-01
-1.69104680e-01 1.12992895e+00 -1.58180475e-01 -1.35334611e-01
4.97742295e-01 3.07580650e-01 -8.43364894e-01 6.92961454e-01
5.89641511e-01 4.88719761e-01 1.37468157e-02 -1.48995864e+00
-3.90861422e-01 4.82298076e-01 2.12720111e-01 -8.61294493e-02
1.59914315e+00 -1.38223674e-02 -2.76053429e-01 8.24984789e-01
1.47224236e+00 -2.66127765e-01 -6.30736589e-01 4.06325936e-01
-1.29188687e-01 -2.83831984e-01 6.66149035e-02 -1.03012359e+00
-1.24952018e+00 1.26851857e+00 1.13692164e+00 3.28707337e-01
1.59374547e+00 -1.91294327e-01 4.49665517e-01 -5.29207215e-02
2.62076020e-01 -1.25209606e+00 1.61714494e-01 4.86436665e-01
1.13391531e+00 -1.25777376e+00 -2.80243605e-01 -1.69555485e-01
-1.08147097e+00 1.32802618e+00 4.46788341e-01 -2.13841666e-02
8.72961760e-01 2.83954173e-01 4.19461608e-01 -5.36385357e-01
-8.05401444e-01 -3.11601073e-01 2.15079799e-01 6.58618867e-01
9.29501355e-01 -2.01813757e-01 -1.66536823e-01 7.04926252e-01
-9.48390961e-02 2.97580510e-01 2.36510798e-01 8.61585557e-01
-1.21379167e-01 -1.08231997e+00 -3.56689513e-01 3.38478476e-01
-1.14749730e+00 2.80060500e-01 -6.93295717e-01 9.72246170e-01
7.25394964e-01 1.11796832e+00 -2.12037042e-01 -6.68827653e-01
4.17511642e-01 3.80779266e-01 4.64357734e-01 -1.73385590e-01
-9.79812682e-01 1.61659941e-01 4.77444470e-01 -8.68640542e-01
-1.15031278e+00 -6.32172346e-01 -7.18448579e-01 -4.27463889e-01
-2.48332307e-01 1.95629463e-01 5.92865586e-01 8.25163066e-01
1.21343225e-01 3.96622866e-01 6.25697374e-01 -1.33125496e+00
-4.78935875e-02 -9.55547154e-01 -7.16923773e-01 8.84903967e-01
1.99247837e-01 -9.21753943e-01 -5.12300551e-01 2.12332949e-01] | [13.558692932128906, 2.1156373023986816] |
a44c450d-d5cb-4ac6-a5aa-f87a3305a6ca | bootstrap-your-own-prior-towards-distribution | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Yang_Bootstrap_Your_Own_Prior_Towards_Distribution-Agnostic_Novel_Class_Discovery_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Yang_Bootstrap_Your_Own_Prior_Towards_Distribution-Agnostic_Novel_Class_Discovery_CVPR_2023_paper.pdf | Bootstrap Your Own Prior: Towards Distribution-Agnostic Novel Class Discovery | Novel Class Discovery (NCD) aims to discover unknown classes without any annotation, by exploiting the transferable knowledge already learned from a base set of known classes. Existing works hold an impractical assumption that the novel class distribution prior is uniform, yet neglect the imbalanced nature of real-world data. In this paper, we relax this assumption by proposing a new challenging task: distribution-agnostic NCD, which allows data drawn from arbitrary unknown class distributions and thus renders existing methods useless or even harmful. We tackle this challenge by proposing a new method, dubbed "Bootstrapping Your Own Prior (BYOP)", which iteratively estimates the class prior based on the model prediction itself. At each iteration, we devise a dynamic temperature technique that better estimates the class prior by encouraging sharper predictions for less-confident samples. Thus, BYOP obtains more accurate pseudo-labels for the novel samples, which are beneficial for the next training iteration. Extensive experiments show that existing methods suffer from imbalanced class distributions, while BYOP outperforms them by clear margins, demonstrating its effectiveness across various distribution scenarios. | ['Hanwang Zhang', 'Cheng Deng', 'Liancheng Wang', 'Muli Yang'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['novel-class-discovery', 'novel-class-discovery'] | ['computer-vision', 'methodology'] | [ 3.47595841e-01 1.81382939e-01 -6.53191388e-01 -4.87062126e-01
-8.53134096e-01 -5.14206767e-01 4.20609564e-01 2.88410485e-01
-6.41418919e-02 1.18607962e+00 -2.02450454e-01 -1.64111659e-01
-1.54031992e-01 -8.48644018e-01 -6.69850230e-01 -9.20869708e-01
3.41038495e-01 8.83779526e-01 3.31822187e-01 2.77436465e-01
2.84550220e-01 3.59293818e-01 -1.58641803e+00 1.17284164e-01
1.13017595e+00 9.05412078e-01 7.84040764e-02 1.10927135e-01
-8.75378251e-02 6.72394097e-01 -7.59796500e-01 -3.61434639e-01
2.36950532e-01 -3.62630367e-01 -5.90784073e-01 7.69517273e-02
1.06880642e-01 -3.01862597e-01 2.02673331e-01 1.22443187e+00
4.45494235e-01 1.04611009e-01 7.30399013e-01 -1.39144814e+00
-4.54441547e-01 5.87643445e-01 -9.85086322e-01 1.42365405e-02
-5.79344593e-02 -8.96153376e-02 7.74215519e-01 -6.42100036e-01
3.49858701e-01 1.11676455e+00 8.09332252e-01 6.59328043e-01
-1.34914255e+00 -1.16976893e+00 3.92626941e-01 2.91639715e-01
-1.71878815e+00 -3.60173970e-01 8.67734313e-01 -3.26000154e-01
2.11323082e-01 2.82291770e-01 4.49245811e-01 9.96700108e-01
-2.28177994e-01 9.98827875e-01 1.29118645e+00 -3.62938493e-01
7.41770446e-01 4.86022770e-01 3.71969561e-03 1.84161484e-01
5.33805788e-01 -1.61137894e-01 -4.37276155e-01 -4.57183361e-01
3.48966241e-01 1.86810240e-01 -4.38836634e-01 -6.31523728e-01
-7.56743968e-01 6.37504756e-01 6.35329708e-02 9.01276395e-02
-2.60656238e-01 -2.59888589e-01 1.59905702e-01 3.13207577e-03
8.34672451e-01 7.64388070e-02 -7.66255379e-01 -5.15937991e-02
-9.10992682e-01 3.11370701e-01 6.80996180e-01 9.28991675e-01
9.93232012e-01 -4.78154451e-01 -1.61562428e-01 9.35274959e-01
2.00073525e-01 4.97926593e-01 5.53567886e-01 -6.35790467e-01
3.12043607e-01 6.87373221e-01 4.85902250e-01 -9.67408538e-01
-9.48661044e-02 -7.87408888e-01 -8.32683146e-01 1.02361515e-02
5.32395661e-01 3.23527679e-02 -1.04585505e+00 1.77997923e+00
9.22344923e-01 4.99633282e-01 6.74626008e-02 5.69159210e-01
2.29306668e-01 5.71240962e-01 -4.86688828e-03 -4.86396432e-01
9.43878770e-01 -7.12340891e-01 -5.79662383e-01 -1.57318845e-01
3.44425768e-01 -4.66817528e-01 1.07450390e+00 6.86680973e-01
-5.83280146e-01 -2.63198227e-01 -9.68488038e-01 4.96220171e-01
-1.77828837e-02 2.71924865e-02 4.67509598e-01 8.91479075e-01
-4.79177684e-01 4.48468804e-01 -7.53936768e-01 3.77827995e-02
9.17770326e-01 1.53978616e-01 -5.99266700e-02 -3.93645138e-01
-7.81290710e-01 4.39016104e-01 6.90579116e-01 -2.12616120e-02
-8.99997652e-01 -1.05617332e+00 -3.80359560e-01 2.46304423e-02
6.85197413e-01 -4.16550934e-01 1.25707293e+00 -1.10498607e+00
-1.43888807e+00 6.66047394e-01 -3.36882651e-01 -3.63432527e-01
6.97375000e-01 -1.21941492e-01 -1.69467807e-01 1.71268247e-02
4.10598032e-02 4.84004676e-01 1.09527123e+00 -1.66943860e+00
-8.10535908e-01 -4.24739927e-01 1.30203813e-02 6.15939051e-02
-6.19271517e-01 -5.68340182e-01 -2.01551929e-01 -6.19065762e-01
3.32861632e-01 -8.84357274e-01 -1.69204623e-01 2.48561259e-02
-5.82716286e-01 -5.40855587e-01 8.45660329e-01 -3.33368570e-01
1.09782088e+00 -2.11052108e+00 -2.71635205e-01 3.46457303e-01
5.25587320e-01 2.05373377e-01 1.59728587e-01 -2.72438899e-02
4.65100035e-02 8.06202143e-02 -4.52968270e-01 -2.45144799e-01
4.84395996e-02 4.57209259e-01 -5.00300765e-01 4.07546043e-01
6.61217868e-02 3.65489691e-01 -1.12198114e+00 -4.24758166e-01
-2.19314441e-01 3.23877394e-01 -5.17116070e-01 1.93128064e-01
-4.44848508e-01 4.93139207e-01 -3.39636832e-01 7.35792398e-01
1.32027507e+00 -5.00378489e-01 4.55188692e-01 -1.10448465e-01
2.64501542e-01 -1.45667614e-02 -1.37427735e+00 1.14948142e+00
-1.43402740e-01 1.15575768e-01 -2.17191994e-01 -1.25542939e+00
1.17464888e+00 3.48816440e-02 3.26313823e-01 -3.95732284e-01
2.83966493e-02 3.14772308e-01 -1.09680600e-01 -3.06588978e-01
1.92398518e-01 -4.44747180e-01 2.08434224e-01 3.74299765e-01
-2.67127335e-01 1.50974765e-02 -6.45645484e-02 -1.46275042e-02
9.19547856e-01 6.79996237e-02 2.54264206e-01 -1.60795808e-01
2.09955096e-01 -2.66584605e-01 1.08153820e+00 8.97587776e-01
-3.14436316e-01 5.41334689e-01 4.88313526e-01 -3.56439859e-01
-8.29271138e-01 -1.15981901e+00 -3.87930036e-01 7.90223300e-01
3.64505470e-01 -7.08009675e-02 -6.28588378e-01 -1.06745708e+00
2.23840661e-02 7.30245709e-01 -6.99049950e-01 -2.17574760e-01
-1.50045916e-01 -1.19745362e+00 1.52001768e-01 3.93241405e-01
4.83224988e-01 -5.71930528e-01 -1.82529673e-01 1.16262250e-01
-3.89816672e-01 -8.45088959e-01 -5.00091724e-02 8.03870708e-02
-9.00747180e-01 -1.10212517e+00 -8.41211379e-01 -5.55406094e-01
8.81338358e-01 9.36979353e-02 9.41964984e-01 -1.48718685e-01
-1.62843000e-02 -3.80062088e-02 -4.68975008e-01 -7.06292629e-01
-3.27162236e-01 1.45022914e-01 5.44039831e-02 3.23112398e-01
5.81209898e-01 -7.67964482e-01 -6.19225204e-01 2.89628386e-01
-8.78574014e-01 1.01382248e-01 6.54050827e-01 8.59432876e-01
8.90576661e-01 5.81903398e-01 1.14209151e+00 -1.15144455e+00
2.79474616e-01 -8.23173165e-01 -5.41155100e-01 3.70936960e-01
-8.50693345e-01 -1.24992371e-01 6.84106946e-01 -6.30226493e-01
-1.25510132e+00 -1.55645728e-01 -1.79175828e-02 -4.98927087e-01
-2.57083088e-01 3.09325993e-01 -5.50953507e-01 2.22091988e-01
5.59619367e-01 2.87776262e-01 -2.41776124e-01 -6.28412664e-01
-3.35059762e-02 8.20003510e-01 5.32409310e-01 -8.57844472e-01
8.55624139e-01 6.00160062e-01 -1.88504711e-01 -5.02056062e-01
-1.33704281e+00 -4.45717692e-01 -5.88179767e-01 4.47632112e-02
1.05548322e-01 -9.83780503e-01 -5.26856661e-01 5.83199263e-01
-7.80266941e-01 -3.55966985e-01 -5.19819856e-01 5.10867298e-01
-1.73576772e-01 3.54514897e-01 -1.84791163e-01 -9.83302534e-01
-1.85708389e-01 -7.21311808e-01 8.17129016e-01 4.87340033e-01
1.15840575e-02 -7.62471616e-01 6.92506433e-02 4.06799167e-01
1.65810809e-01 2.48450056e-01 1.00715768e+00 -9.40151453e-01
-5.13218939e-01 -3.38678896e-01 -3.95023167e-01 4.88502413e-01
2.82488853e-01 -1.79763004e-01 -1.15654290e+00 -4.00043100e-01
-1.14693396e-01 -2.97802120e-01 7.40165532e-01 1.80119395e-01
1.70787442e+00 -3.26703936e-01 -5.64150572e-01 3.15439224e-01
1.40252757e+00 1.68444753e-01 5.60888052e-01 1.65663019e-01
5.21775961e-01 3.96592200e-01 8.46140265e-01 8.63999188e-01
5.06600261e-01 3.85335177e-01 5.35249889e-01 1.47660032e-01
1.24297284e-01 -3.46674830e-01 -4.54647578e-02 6.41900599e-01
2.47356176e-01 -4.95252311e-01 -8.28789592e-01 8.15359473e-01
-1.85618818e+00 -7.34942257e-01 2.95871884e-01 2.48918223e+00
1.42695498e+00 2.99382269e-01 3.61770689e-02 4.43981975e-01
9.39400554e-01 -3.69463712e-01 -1.04553998e+00 2.34719455e-01
1.18511487e-02 3.22596222e-01 1.99858844e-01 1.57070071e-01
-9.64361668e-01 4.84698474e-01 5.64602375e+00 1.38413930e+00
-8.83593798e-01 9.79441851e-02 8.55195463e-01 -3.22061628e-02
-6.01475775e-01 1.66476324e-01 -9.74532723e-01 7.62427807e-01
5.32826364e-01 -2.90818155e-01 2.71552484e-02 9.07562792e-01
-1.40863871e-02 -3.67639095e-01 -1.11214542e+00 8.99266601e-01
1.02614284e-01 -1.08491576e+00 -5.49399434e-03 -1.31755471e-01
9.91018176e-01 -3.23777705e-01 -1.03903404e-02 3.91562134e-01
4.79014874e-01 -5.18442214e-01 5.25210440e-01 6.08549178e-01
7.30738997e-01 -9.34176922e-01 7.89043665e-01 6.55113697e-01
-7.60374188e-01 -1.57729924e-01 -4.90330487e-01 -3.00629362e-02
-2.93580651e-01 1.38398242e+00 -1.12370574e+00 5.24507463e-01
7.47761190e-01 7.46387422e-01 -4.93157923e-01 1.39973581e+00
-1.26208201e-01 1.11807775e+00 -5.17107368e-01 1.02096729e-01
-2.59087980e-01 5.45739047e-02 4.16550785e-01 7.76243329e-01
2.97162145e-01 -1.25766052e-02 4.45592880e-01 6.72069848e-01
-4.40958232e-01 8.66227150e-02 -9.25555304e-02 2.26288021e-01
9.54868674e-01 1.02977920e+00 -9.83418047e-01 -4.65692163e-01
4.82431985e-02 6.44785285e-01 3.22123528e-01 2.54542410e-01
-8.85624290e-01 -4.04212952e-01 2.80967355e-01 4.77718078e-02
4.35192734e-01 2.58641034e-01 -3.42222661e-01 -1.23146510e+00
2.38502055e-01 -6.73276961e-01 5.04166901e-01 -4.99461442e-01
-1.66648829e+00 1.72032908e-01 7.32370168e-02 -1.38456178e+00
1.57696828e-01 -2.33908281e-01 -3.39785486e-01 5.83762884e-01
-1.91240573e+00 -9.69183743e-01 -3.52386743e-01 2.94192016e-01
3.59690338e-01 1.44194722e-01 6.30823970e-01 4.95011628e-01
-6.65512681e-01 9.85344231e-01 5.82363307e-01 -2.23248109e-01
9.47206199e-01 -1.16528094e+00 -2.77884811e-01 5.00132084e-01
-2.67874092e-01 2.05051869e-01 7.12386131e-01 -6.37353837e-01
-6.76909447e-01 -1.33698392e+00 5.59363127e-01 -3.52544457e-01
4.06136036e-01 -2.69620985e-01 -1.19191957e+00 4.58204836e-01
-4.36398119e-01 1.95274606e-01 1.01278293e+00 1.61444336e-01
-4.93244499e-01 -4.88701046e-01 -1.55323398e+00 2.41533861e-01
8.93045008e-01 -1.80182651e-01 -3.65313023e-01 5.55503786e-01
8.03438604e-01 -2.34919101e-01 -7.15750635e-01 5.84191024e-01
5.09659827e-01 -8.20406675e-01 7.80721784e-01 -3.97369593e-01
2.89251149e-01 -6.23519301e-01 -2.22941086e-01 -1.22832131e+00
-2.25957707e-02 -3.42803389e-01 -3.77513707e-01 1.60908401e+00
3.58325869e-01 -8.38520646e-01 9.55608130e-01 4.40871984e-01
5.33675849e-02 -8.66047025e-01 -9.96212661e-01 -9.34938133e-01
8.12972039e-02 -3.79057109e-01 9.91799176e-01 1.35748291e+00
-2.83360243e-01 6.41818419e-02 -4.06556666e-01 3.80153596e-01
8.28410566e-01 4.17985588e-01 8.69395077e-01 -1.51821887e+00
-2.98501402e-01 2.37060357e-02 -3.30967069e-01 -1.08136892e+00
1.83748305e-01 -7.21996546e-01 3.27058673e-01 -9.74185705e-01
6.52105391e-01 -1.17975402e+00 -4.85483199e-01 7.66050875e-01
-5.31212032e-01 4.54094529e-01 -1.99873671e-01 4.34231639e-01
-8.83349478e-01 7.14574337e-01 1.17744339e+00 -1.41601846e-01
-1.77921489e-01 2.73405164e-01 -1.03560448e+00 8.28808606e-01
8.41214120e-01 -8.06081891e-01 -6.24913216e-01 -1.37144133e-01
1.66780099e-01 -3.12130958e-01 3.14466178e-01 -1.05758238e+00
3.43765579e-02 -4.84576195e-01 5.22175908e-01 -8.17298710e-01
3.33595574e-02 -7.55197108e-01 7.89003670e-02 2.69251704e-01
-2.41637200e-01 -7.15655148e-01 9.75752622e-02 9.48799253e-01
-6.54200232e-03 -3.25781643e-01 8.77413511e-01 1.66635394e-01
-3.75863910e-01 4.26642060e-01 4.05325517e-02 2.87266374e-01
1.15379870e+00 -1.68817565e-01 -3.23232800e-01 -1.71997443e-01
-6.64786637e-01 2.12600365e-01 5.06951988e-01 1.44275323e-01
3.72792870e-01 -1.22749543e+00 -7.80427396e-01 1.07216187e-01
2.62814999e-01 6.53898835e-01 2.92951763e-01 6.50127232e-01
-4.01640944e-02 -1.50848702e-01 2.51141965e-01 -8.43719959e-01
-9.64799821e-01 5.71230769e-01 1.73056111e-01 -2.99376607e-01
-3.52004588e-01 9.47931826e-01 2.47462034e-01 -7.44215667e-01
1.37265533e-01 -2.39590555e-02 -6.09139204e-02 1.91340428e-02
5.73090196e-01 3.64340514e-01 1.55375168e-01 -5.59039451e-02
-2.06662461e-01 2.34588817e-01 -4.43865001e-01 4.76555079e-01
1.38110697e+00 -2.12277994e-01 -8.14994797e-02 5.94672024e-01
8.99484754e-01 1.51995838e-01 -1.53055334e+00 -6.62234783e-01
-6.35983944e-02 -9.05149579e-01 -1.41853929e-01 -1.00553834e+00
-8.68516624e-01 6.57606840e-01 7.97234178e-01 1.68121308e-01
1.43556225e+00 -2.36226991e-02 8.29535902e-01 2.06754565e-01
3.43834311e-01 -1.07248068e+00 2.16039103e-02 2.74507284e-01
3.72784883e-01 -1.31883335e+00 1.77919790e-01 -4.94713753e-01
-2.85413027e-01 9.33502734e-01 7.94103265e-01 2.66898692e-01
7.23533750e-01 1.00549437e-01 -1.70783162e-01 2.03516051e-01
-6.49536252e-01 7.89720863e-02 -5.51447971e-03 7.66594529e-01
-1.28449604e-01 1.77718982e-01 -1.33890450e-01 8.64459991e-01
3.52621973e-02 3.14861804e-01 4.44550782e-01 8.99063587e-01
-4.76748645e-01 -1.14455056e+00 -4.64690536e-01 7.63620794e-01
-2.87368625e-01 7.44699314e-02 -1.21668100e-01 3.86712343e-01
4.98056263e-01 7.24796474e-01 1.17218658e-01 -2.47123182e-01
-3.46975587e-02 -3.50139998e-02 4.18989211e-01 -5.45960784e-01
2.66632110e-01 -1.19777679e-01 -2.80950010e-01 8.40638652e-02
-5.45250118e-01 -6.85070932e-01 -1.07071638e+00 -3.61019254e-01
-8.46247673e-01 4.29387033e-01 3.78344238e-01 1.08426225e+00
5.06498098e-01 3.24879676e-01 1.02512395e+00 -5.64496458e-01
-7.64359117e-01 -7.78923750e-01 -6.58695698e-01 2.32530937e-01
2.97202617e-01 -9.28224862e-01 -5.68965197e-01 1.76214635e-01] | [9.395710945129395, 3.7038326263427734] |
001c34a4-3440-4e13-a598-4d2fff91e379 | comparing-abstractive-summaries-generated-by | 2303.17650 | null | https://arxiv.org/abs/2303.17650v2 | https://arxiv.org/pdf/2303.17650v2.pdf | Evaluating and Detecting ChatGPT's Responses on Abstractive Summarization | Large Language Models (LLMs) have gathered significant attention due to their impressive performance on a variety of tasks. ChatGPT, developed by OpenAI, is a recent addition to the family of language models and is being called a disruptive technology by a few, owing to its human-like text-generation capabilities. Although, many anecdotal examples across the internet have evaluated ChatGPT's strength and weakness, only a few systematic research studies exist. To contribute to the body of literature of systematic research on ChatGPT, we evaluate the performance of ChatGPT on Abstractive Summarization by the means of automated metrics and blinded human reviewers. We also build automatic text classifiers to detect ChatGPT generated summaries. We found that while text classification algorithms can distinguish between real and generated summaries, humans are unable to distinguish between real summaries and those produced by ChatGPT. | ['Vincent Wade', 'Mayank Soni'] | 2023-03-30 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [-8.04921612e-02 3.05178106e-01 -4.20699060e-01 3.44619416e-02
-1.06406212e+00 -5.52888751e-01 1.08372378e+00 6.19668782e-01
-1.32199720e-01 6.67967141e-01 7.28999555e-01 -5.24246871e-01
5.59114143e-02 -3.68607670e-01 -1.06888965e-01 -1.22090384e-01
1.35700017e-01 5.29148042e-01 2.23952562e-01 -1.35522597e-02
9.27388549e-01 -1.35184258e-01 -1.21344519e+00 5.55249155e-01
1.34604859e+00 4.02106076e-01 1.91189706e-01 9.53182459e-01
-8.13591659e-01 1.31397831e+00 -1.25948429e+00 -4.41242754e-01
-2.45817065e-01 -7.51320422e-01 -9.39926386e-01 -1.69456050e-01
3.44628453e-01 -2.31933355e-01 -2.88383752e-01 6.08758688e-01
6.19678438e-01 -1.59665421e-01 6.27473235e-01 -1.35976779e+00
-8.83214355e-01 9.94221628e-01 -3.40054631e-01 2.68064737e-01
8.58895898e-01 1.51948676e-01 1.12719035e+00 -5.80766976e-01
6.93132043e-01 1.32366633e+00 7.39405692e-01 3.31696630e-01
-1.12237346e+00 -6.45528257e-01 -2.87317038e-01 -2.05007687e-01
-9.68464673e-01 -4.77294445e-01 3.20690632e-01 -6.27314448e-01
1.26766312e+00 4.49579686e-01 6.98201120e-01 1.13281465e+00
6.74179435e-01 9.06082392e-01 8.55881333e-01 -6.58069551e-01
2.00587615e-01 4.33961630e-01 2.56815642e-01 3.78339320e-01
3.48722994e-01 -4.88804340e-01 -8.53791714e-01 -6.96396232e-01
2.66560167e-01 -2.83305109e-01 -1.61514491e-01 3.73699605e-01
-1.12612069e+00 8.77349675e-01 -1.51142687e-01 5.39714575e-01
-2.64308780e-01 -7.06317835e-03 7.58934319e-01 4.22071844e-01
9.48265195e-01 7.85365939e-01 -2.49421690e-02 -7.77439654e-01
-1.08931148e+00 3.55878592e-01 1.40142870e+00 1.09192860e+00
1.04936123e-01 -5.78326546e-03 -5.54799795e-01 1.00755918e+00
2.50355542e-01 3.55010659e-01 8.03408802e-01 -8.24343204e-01
7.17650712e-01 9.35483992e-01 -1.16183208e-02 -1.10397530e+00
-1.35219455e-01 1.80207957e-02 -4.88054156e-01 -3.84945929e-01
1.27453700e-01 -3.02890986e-01 -2.69927084e-01 1.00817823e+00
-4.06073809e-01 -4.15216506e-01 -8.93863104e-03 2.98744082e-01
1.17368937e+00 9.85551655e-01 2.03924268e-01 -3.75869334e-01
8.47474158e-01 -8.89629364e-01 -9.05419648e-01 -2.91982085e-01
1.09209776e+00 -1.13087022e+00 1.01010132e+00 3.08637530e-01
-1.14100444e+00 -4.55926806e-02 -7.99310625e-01 8.43582302e-02
-1.46031708e-01 -6.97600767e-02 6.17974281e-01 8.41185153e-01
-1.08067846e+00 5.22370279e-01 -4.95650321e-01 -8.81878734e-01
3.23429316e-01 -1.74029857e-01 -4.44183089e-02 1.64372340e-01
-8.64746094e-01 1.16628528e+00 6.47977442e-02 -5.58135688e-01
-2.97390014e-01 -6.34698510e-01 -5.69327116e-01 -4.07532090e-03
2.95647979e-01 -6.35654390e-01 1.73254430e+00 -4.91282970e-01
-1.28994024e+00 7.17155397e-01 -2.71750540e-01 -5.32029390e-01
4.62390065e-01 -2.52945453e-01 -2.65935212e-01 3.57562155e-01
3.80445093e-01 2.24417001e-01 3.57891232e-01 -9.06847060e-01
-6.16001844e-01 -6.82946518e-02 -2.52600014e-01 1.99285835e-01
-6.91686392e-01 6.92304134e-01 6.40638545e-02 -6.91637933e-01
-3.46166015e-01 -5.17852187e-01 -9.50763002e-02 -2.94630736e-01
-6.10883772e-01 -5.88272989e-01 8.61941397e-01 -7.61783540e-01
1.82431495e+00 -1.60164428e+00 -5.89607000e-01 -3.29433769e-01
5.14655232e-01 4.62478131e-01 5.87829240e-02 1.34229934e+00
6.12514138e-01 8.36459517e-01 7.54302964e-02 -3.56975228e-01
5.69499470e-02 -1.08279638e-01 -3.99631172e-01 1.54834660e-02
-1.74571261e-01 8.66946757e-01 -1.02451479e+00 -7.46499002e-01
8.10975656e-02 -2.12717969e-02 -1.82777837e-01 2.78905220e-02
-2.57853985e-01 -1.10594854e-01 -5.65671980e-01 3.86024505e-01
1.99115738e-01 -3.23684514e-01 6.97148666e-02 7.06286788e-01
-4.71304446e-01 8.19990635e-01 -5.94289601e-01 1.16016960e+00
-3.44252765e-01 1.18349159e+00 -5.61261363e-02 -5.34715474e-01
9.12576497e-01 8.18860114e-01 1.65691033e-01 -4.13492292e-01
2.34430298e-01 3.45651776e-01 3.50494720e-02 -6.61482930e-01
9.07688677e-01 5.39146401e-02 -1.88850909e-01 1.22908092e+00
-1.27355177e-02 -4.23130095e-01 3.98789674e-01 6.65336967e-01
1.32290840e+00 -3.53005767e-01 5.65296352e-01 -6.74378797e-02
5.93776926e-02 3.55024099e-01 5.01553081e-02 1.22227097e+00
-8.59022737e-02 5.49827337e-01 6.77001774e-01 -7.75194392e-02
-9.19039249e-01 -4.27290112e-01 6.44373894e-02 9.58559155e-01
-2.95275986e-01 -9.93912876e-01 -9.14301157e-01 -3.99043560e-01
-8.89401883e-02 1.10184395e+00 -1.12982288e-01 -1.01360343e-01
-2.33537003e-01 -5.19897223e-01 9.01435375e-01 3.09277385e-01
4.08257604e-01 -1.19862151e+00 -7.32969642e-01 2.83933491e-01
-4.33951855e-01 -1.05730677e+00 -7.18876064e-01 -1.99826717e-01
-9.55470562e-01 -9.85176325e-01 -5.89684963e-01 -7.87636638e-01
2.50551581e-01 4.60745752e-01 1.22545397e+00 1.68336511e-01
-1.55216232e-01 5.89317083e-01 -6.03305221e-01 -8.68668854e-01
-1.30490839e+00 2.87155420e-01 -8.06959644e-02 -6.65161669e-01
5.31060100e-01 -3.42554837e-01 -9.12724063e-02 1.64613854e-02
-7.43019581e-01 2.15195209e-01 5.83038151e-01 4.39270794e-01
-3.98168087e-01 -1.36584729e-01 1.05807364e+00 -8.73484015e-01
1.58499539e+00 -5.94314933e-01 4.82301004e-02 2.28658915e-01
-6.95328832e-01 -3.92242342e-01 4.54922736e-01 -4.66631025e-01
-9.67644095e-01 -9.19187248e-01 2.57648468e-01 9.23809782e-02
-5.51245585e-02 9.02953267e-01 3.98324341e-01 1.23129688e-01
8.82352889e-01 2.43892565e-01 1.68432623e-01 -3.79246205e-01
5.04649132e-02 1.19877994e+00 5.33813499e-02 -3.55883926e-01
5.58382869e-01 -5.46461456e-02 -7.03882754e-01 -1.21400666e+00
-5.09559929e-01 -5.36244571e-01 -2.26417705e-01 -3.74845415e-01
5.38506687e-01 -7.44525254e-01 -4.92337793e-01 4.09822553e-01
-1.34807765e+00 -3.08842957e-01 -1.08951420e-01 3.45703870e-01
-2.35673845e-01 5.79061091e-01 -8.31524312e-01 -1.20388842e+00
-9.56655860e-01 -6.09939039e-01 5.77497244e-01 2.63955146e-01
-1.16938496e+00 -1.15496719e+00 1.36196673e-01 5.17695785e-01
5.04098475e-01 1.00372307e-01 1.06875813e+00 -1.24388492e+00
-1.37324452e-01 -8.32433283e-01 1.50587335e-02 1.74948007e-01
1.46717727e-01 2.83895433e-01 -6.82345748e-01 5.56120276e-03
1.83664456e-01 -1.81129858e-01 2.89572656e-01 3.38694334e-01
4.88899231e-01 -8.49567950e-01 -4.67024386e-01 -3.88084292e-01
9.47379529e-01 3.93931627e-01 3.16627920e-01 3.61305445e-01
2.97941983e-01 5.87511182e-01 4.06974912e-01 5.81622422e-01
4.42635655e-01 2.97266006e-01 -1.94816381e-01 4.77979869e-01
-5.99443875e-02 -6.42431259e-01 5.36008894e-01 1.30417550e+00
2.37018943e-01 -6.71940625e-01 -1.10715961e+00 5.57287216e-01
-1.77422059e+00 -9.86535609e-01 -5.82529187e-01 1.99950492e+00
6.89235747e-01 4.85048562e-01 1.12967312e-01 3.19374911e-02
7.03388989e-01 1.93591118e-01 -1.51142329e-01 -8.14756870e-01
-2.06322316e-02 -7.39661977e-02 6.55130744e-02 -3.02271228e-02
-4.08214390e-01 5.80906689e-01 7.75500679e+00 7.38906264e-01
-8.46643984e-01 -3.00560854e-02 4.66644853e-01 -9.22856107e-02
-2.57538408e-01 1.78336307e-01 -5.16585946e-01 5.75438917e-01
1.27269983e+00 -1.06917953e+00 -1.91505224e-01 8.13376844e-01
6.27231359e-01 -4.80183184e-01 -9.85858321e-01 6.59924150e-01
4.08592194e-01 -1.47867417e+00 3.98516089e-01 1.79078914e-02
7.57345855e-01 -5.33855334e-02 -3.37026924e-01 5.14503598e-01
3.18612754e-01 -8.35376561e-01 7.42265940e-01 3.08392912e-01
3.54164511e-01 -4.77774471e-01 7.96070039e-01 8.74323487e-01
-7.04542577e-01 -5.74340560e-02 -1.55810714e-01 -6.39703333e-01
2.66686380e-01 4.11983043e-01 -1.21784234e+00 3.91717583e-01
3.99143934e-01 7.40220129e-01 -7.01749086e-01 1.24914038e+00
1.37096997e-02 1.05822349e+00 1.41208336e-01 -6.20860577e-01
1.77623570e-01 -1.95000291e-01 8.87120426e-01 1.37811708e+00
2.80245960e-01 -4.91197370e-02 2.37458557e-01 8.29628706e-01
-2.30931379e-02 3.98978025e-01 -9.56713140e-01 -7.21104741e-01
8.39066982e-01 1.03052998e+00 -7.58795142e-01 -4.42322642e-01
-5.05777180e-01 5.57070076e-01 -1.84253290e-01 6.98465258e-02
-2.83215731e-01 -7.34426856e-01 1.26790717e-01 3.46135169e-01
-3.46711516e-01 -1.05716936e-01 -7.17592120e-01 -8.61607313e-01
1.39187193e-02 -9.48045373e-01 1.27458140e-01 -1.08132291e+00
-1.26664066e+00 5.23774266e-01 6.50212690e-02 -1.09376740e+00
-4.45181727e-01 1.28864974e-03 -1.20456636e+00 7.76944041e-01
-6.18792295e-01 -8.21326256e-01 -1.45587489e-01 -1.40221894e-01
8.93527567e-01 -2.04078883e-01 8.88746023e-01 -1.21712245e-01
-4.35939342e-01 4.39561814e-01 1.81814522e-01 4.44739722e-02
8.32959712e-01 -1.13557649e+00 6.55356348e-01 5.35656154e-01
-2.57503092e-01 8.50198925e-01 9.76946950e-01 -1.03790760e+00
-1.12470579e+00 -6.53637707e-01 1.53499901e+00 -8.42567146e-01
8.86818469e-01 -2.27914840e-01 -1.00291932e+00 6.96123004e-01
7.43666351e-01 -9.45370495e-01 8.83390963e-01 9.57721751e-03
-2.69841887e-02 4.09576207e-01 -8.76234531e-01 8.00211668e-01
6.72925770e-01 -5.04924953e-01 -1.00624096e+00 4.86845613e-01
6.19865477e-01 -1.67470962e-01 -8.96990895e-01 -2.46645778e-01
4.67493415e-01 -7.89126456e-01 4.05887157e-01 -3.39889377e-01
8.60639870e-01 3.03440303e-01 3.31059873e-01 -1.38481748e+00
1.33160844e-01 -1.13526332e+00 -5.49317189e-02 1.80446911e+00
4.15146440e-01 -7.26798534e-01 4.89080340e-01 9.15161490e-01
-4.71304685e-01 -5.32579839e-01 -5.97030878e-01 -8.61533999e-01
2.76796579e-01 -2.71924555e-01 1.80289671e-01 8.75223160e-01
9.51857924e-01 6.80906773e-01 -3.25864047e-01 -6.69676602e-01
2.55215049e-01 -1.51678890e-01 9.42592680e-01 -1.55647182e+00
1.39167204e-01 -8.02566171e-01 -1.51136279e-01 -7.77340710e-01
-7.54107088e-02 -9.75365818e-01 -9.57127940e-03 -2.16059637e+00
6.38794780e-01 -2.75942888e-02 6.16827369e-01 2.93593347e-01
-2.64992472e-02 -2.32010394e-01 2.51441479e-01 5.19337654e-01
-7.45566189e-01 3.26908201e-01 1.08214545e+00 2.27135010e-02
-4.91887003e-01 1.11116491e-01 -1.24545443e+00 6.55288041e-01
8.45764518e-01 -6.26983821e-01 -3.98617625e-01 -1.76817939e-01
2.88399935e-01 2.79208481e-01 -7.30259195e-02 -1.02274382e+00
3.33907813e-01 -1.88633069e-01 -1.07203498e-01 -5.61523020e-01
-2.15057880e-01 -1.23574995e-01 -1.05552830e-01 2.66357630e-01
-8.15463603e-01 1.91252425e-01 2.58180290e-01 4.32633042e-01
-2.38062575e-01 -4.74657953e-01 3.07228267e-01 -3.63064289e-01
-1.04411416e-01 -1.49122775e-01 -1.38142920e+00 4.00323123e-01
7.31417120e-01 -4.81730789e-01 -6.86269224e-01 -9.42167461e-01
2.03762338e-01 4.08541381e-01 4.13303435e-01 6.08939469e-01
3.48515928e-01 -8.86565328e-01 -7.20930755e-01 -4.31614399e-01
1.39190540e-01 -2.53582299e-01 7.58083388e-02 9.72583950e-01
-5.02890289e-01 7.79346168e-01 1.26327187e-01 -3.30959976e-01
-1.29516196e+00 1.82698473e-01 -1.48394197e-01 -4.26926672e-01
-9.68089461e-01 2.28111878e-01 -2.07428873e-01 -4.05230403e-01
3.50840986e-01 -3.07446778e-01 -2.84830332e-01 1.91648602e-01
9.15000260e-01 7.65859187e-01 5.00888973e-02 -3.21659416e-01
4.82062176e-02 -1.48736224e-01 -2.21890628e-01 -1.91802785e-01
1.10600471e+00 -2.10016116e-01 -2.06147149e-01 9.94112492e-01
9.96627212e-01 -6.16548918e-02 -4.30270821e-01 -6.25271797e-02
4.38159615e-01 -2.93221384e-01 -2.46482953e-01 -7.90303349e-01
-2.19731405e-02 7.10175574e-01 -2.46155098e-01 9.18899596e-01
4.94751781e-01 -4.65200618e-02 9.42193925e-01 4.12992179e-01
2.01789856e-01 -1.20752668e+00 4.49985951e-01 7.95877039e-01
9.87964332e-01 -1.02589393e+00 5.78963682e-02 -8.77732784e-02
-8.65474880e-01 1.09506297e+00 4.18939680e-01 2.01462835e-01
3.12921673e-01 1.73327848e-01 1.95687279e-01 -2.28169173e-01
-1.17598033e+00 5.79207480e-01 1.39866779e-02 5.13888836e-01
8.73379827e-01 -1.04923204e-01 -7.25547910e-01 6.84439123e-01
-6.84935451e-01 1.82362005e-01 1.12492502e+00 1.13811314e+00
-7.46715784e-01 -9.25074756e-01 -3.41643870e-01 1.13876855e+00
-6.38726711e-01 -1.62996337e-01 -8.62125695e-01 7.74854600e-01
-6.96323514e-01 1.56705987e+00 -1.94800068e-02 -3.53319257e-01
1.10466771e-01 3.18645179e-01 -8.51857886e-02 -1.07415998e+00
-1.08330834e+00 -1.82023332e-01 5.37198901e-01 1.73544679e-02
-2.03622103e-01 -7.50796199e-01 -9.70394194e-01 -7.37571180e-01
-4.17492628e-01 6.89282954e-01 5.21008134e-01 9.89514053e-01
4.20317680e-01 2.80545175e-01 3.68119359e-01 -6.70550048e-01
-5.21660209e-01 -1.50915849e+00 -5.61883688e-01 2.76073590e-02
-1.21159004e-02 -4.15184610e-02 -5.11513114e-01 -6.89422339e-02] | [11.947612762451172, 9.04207992553711] |
4a0885f6-9be9-4e5f-97ed-a04cbde86cf8 | covsegnet-a-multi-encoder-decoder | 2012.01473 | null | https://arxiv.org/abs/2012.01473v1 | https://arxiv.org/pdf/2012.01473v1.pdf | CovSegNet: A Multi Encoder-Decoder Architecture for Improved Lesion Segmentation of COVID-19 Chest CT Scans | Automatic lung lesions segmentation of chest CT scans is considered a pivotal stage towards accurate diagnosis and severity measurement of COVID-19. Traditional U-shaped encoder-decoder architecture and its variants suffer from diminutions of contextual information in pooling/upsampling operations with increased semantic gaps among encoded and decoded feature maps as well as instigate vanishing gradient problems for its sequential gradient propagation that result in sub-optimal performance. Moreover, operating with 3D CT-volume poses further limitations due to the exponential increase of computational complexity making the optimization difficult. In this paper, an automated COVID-19 lesion segmentation scheme is proposed utilizing a highly efficient neural network architecture, namely CovSegNet, to overcome these limitations. Additionally, a two-phase training scheme is introduced where a deeper 2D-network is employed for generating ROI-enhanced CT-volume followed by a shallower 3D-network for further enhancement with more contextual information without increasing computational burden. Along with the traditional vertical expansion of Unet, we have introduced horizontal expansion with multi-stage encoder-decoder modules for achieving optimum performance. Additionally, multi-scale feature maps are integrated into the scale transition process to overcome the loss of contextual information. Moreover, a multi-scale fusion module is introduced with a pyramid fusion scheme to reduce the semantic gaps between subsequent encoder/decoder modules while facilitating the parallel optimization for efficient gradient propagation. Outstanding performances have been achieved in three publicly available datasets that largely outperform other state-of-the-art approaches. The proposed scheme can be easily extended for achieving optimum segmentation performances in a wide variety of applications. | ['Sun-Yuan Kung', 'Shaikh Anowarul Fattah', 'Md Awsafur Rahman', 'Tanvir Mahmud'] | 2020-12-02 | null | null | null | null | ['covid-19-image-segmentation'] | ['computer-vision'] | [ 5.01440585e-01 3.20431627e-02 -4.97449748e-02 -2.42186800e-01
-1.13819683e+00 -3.78229626e-04 1.42572954e-01 4.20253724e-01
-7.17029572e-01 4.74299431e-01 1.24830469e-01 -3.38196635e-01
-1.12750009e-01 -7.79694378e-01 -3.67741704e-01 -7.69227445e-01
-1.79702695e-02 1.53318748e-01 7.52810001e-01 2.06737947e-02
3.27417925e-02 5.05009711e-01 -1.12590170e+00 4.97568280e-01
9.24320459e-01 1.14289320e+00 7.89821804e-01 7.51905262e-01
-2.60052860e-01 7.43628681e-01 -3.12611997e-01 -4.39354748e-01
2.04859912e-01 -4.07545984e-01 -5.47266901e-01 6.01983890e-02
1.32663995e-01 -4.72873002e-01 -4.25828040e-01 9.32488263e-01
8.21450889e-01 -8.66874456e-02 5.92585921e-01 -4.94200230e-01
-1.69255584e-01 4.12248492e-01 -6.92018151e-01 4.69757259e-01
-1.06254309e-01 1.80441573e-01 6.53428495e-01 -9.32928145e-01
1.82821006e-01 6.81380033e-01 9.37726259e-01 2.87081987e-01
-8.60147059e-01 -2.18707398e-01 -1.72911540e-01 2.02919915e-01
-1.34993875e+00 1.59768090e-01 6.92480981e-01 -3.19694191e-01
1.09028268e+00 4.91127133e-01 7.60129809e-01 4.68189329e-01
5.40556490e-01 5.19231677e-01 7.18052030e-01 -2.96243221e-01
-1.13157360e-02 2.32452556e-01 -2.39462703e-01 1.02708030e+00
3.30451846e-01 -2.05680192e-01 5.03194239e-03 9.86002609e-02
1.00351655e+00 2.82758653e-01 -2.29913026e-01 -2.03634217e-01
-1.15101874e+00 8.13919127e-01 1.00863171e+00 6.03672683e-01
-6.87223017e-01 3.30392867e-02 6.07955337e-01 -2.55799085e-01
2.49712422e-01 3.10685903e-01 -4.64325994e-02 2.94855773e-01
-1.35149574e+00 -4.44942527e-02 2.84803957e-01 6.24031484e-01
3.91615242e-01 1.20462149e-01 -5.45215189e-01 8.76142561e-01
1.92227945e-01 7.69456178e-02 1.00558639e+00 -5.70346117e-01
8.25559974e-01 8.35239232e-01 -3.81425917e-01 -8.37543011e-01
-7.66768396e-01 -9.51884747e-01 -1.11025476e+00 6.03757538e-02
3.18360209e-01 -8.59615430e-02 -1.10636854e+00 1.31608510e+00
3.29879522e-01 1.29527792e-01 -1.87611077e-02 1.01610339e+00
8.27507138e-01 3.90415907e-01 2.87981391e-01 -8.83713961e-02
1.76837730e+00 -1.04625976e+00 -3.98559004e-01 -1.54277399e-01
6.77870333e-01 -8.38955939e-01 8.27949226e-01 9.86533146e-03
-1.39859879e+00 -6.04942322e-01 -1.19458103e+00 -1.09792598e-01
-2.47062176e-01 5.03083885e-01 4.06654537e-01 8.23651612e-01
-1.09950852e+00 4.80186582e-01 -1.04372644e+00 -9.38322246e-02
6.38654470e-01 6.06453776e-01 -4.57637496e-02 2.44404771e-03
-1.13739622e+00 1.03827190e+00 5.50803363e-01 2.18807191e-01
-5.68519771e-01 -7.48939276e-01 -7.90485203e-01 1.21384174e-01
3.29727918e-01 -9.22827899e-01 1.18469167e+00 -4.70143110e-01
-1.37659681e+00 5.80055475e-01 -4.29469394e-03 -6.12324357e-01
6.90706968e-01 2.45605186e-01 -2.38126829e-01 5.51538825e-01
-6.33755550e-02 7.33557761e-01 8.39457333e-01 -6.45961404e-01
-7.83231795e-01 -3.77040148e-01 -1.32819459e-01 6.55563653e-01
-4.78944361e-01 -1.01461083e-01 -5.63320696e-01 -8.49737823e-01
4.89674002e-01 -6.29770577e-01 -7.09007144e-01 2.55532637e-02
-1.91590920e-01 2.79860765e-01 7.95130193e-01 -8.70919704e-01
1.54683638e+00 -1.88576448e+00 -1.40491083e-01 7.00968280e-02
4.97850478e-01 4.16130483e-01 1.20599322e-01 -1.47009388e-01
-4.52113263e-02 -3.57886516e-02 -5.67646384e-01 -1.09382764e-01
-5.29115379e-01 3.61583941e-02 3.54209036e-01 3.83160084e-01
4.39595312e-01 9.79248405e-01 -5.54917514e-01 -9.12870944e-01
5.05182981e-01 7.12680995e-01 -7.67678797e-01 1.17339753e-01
1.02640942e-01 4.83656645e-01 -5.63148260e-01 4.88306910e-01
6.70345187e-01 -4.55055892e-01 -7.94412419e-02 -3.47917199e-01
-2.97899812e-01 2.58613557e-01 -1.04011130e+00 1.71782911e+00
-6.94147348e-01 1.09585769e-01 2.97705054e-01 -1.06701708e+00
6.32601380e-01 6.64981782e-01 8.55977416e-01 -5.67635596e-01
3.67019713e-01 4.67050493e-01 1.66117772e-01 -7.25110769e-01
5.77272058e-01 -4.27999735e-01 8.17900300e-02 3.50090228e-02
-1.67143568e-01 -4.04599637e-01 2.06663385e-01 -7.32359812e-02
1.01085746e+00 -2.41005212e-01 2.77129680e-01 -5.54123707e-02
1.05347800e+00 9.49938744e-02 4.52205390e-01 5.14521360e-01
-3.07621419e-01 9.17943716e-01 2.87772208e-01 -1.68288291e-01
-9.84634459e-01 -9.89540100e-01 -3.23306173e-01 5.04773080e-01
3.36272977e-02 -2.03966394e-01 -7.98949838e-01 -6.95278883e-01
-3.84522647e-01 4.05755997e-01 -2.62951195e-01 9.95400399e-02
-9.22750652e-01 -1.24152863e+00 6.76019132e-01 7.59750009e-01
7.94644892e-01 -8.00859034e-01 -1.10976303e+00 6.02965832e-01
-2.12945551e-01 -1.06179297e+00 -4.25514013e-01 3.46709490e-01
-1.26922357e+00 -8.51844788e-01 -1.12452924e+00 -8.56782615e-01
7.60578990e-01 2.05557808e-01 7.41602182e-01 3.34540009e-02
-7.86422491e-01 -1.83771580e-01 -4.91084754e-02 -1.46027505e-01
-3.95056337e-01 2.58750468e-01 -6.12288296e-01 -4.34937000e-01
2.91720182e-02 -3.70463938e-01 -1.04987204e+00 1.74451634e-01
-1.08890891e+00 3.64513874e-01 9.90935028e-01 9.84415054e-01
7.13800728e-01 2.68578380e-01 7.63396978e-01 -7.26014495e-01
4.65192735e-01 -4.07298177e-01 -4.20677125e-01 1.26823202e-01
-4.86255646e-01 -1.12788759e-01 8.01892459e-01 5.95553853e-02
-1.32530260e+00 4.45373133e-02 -6.73237145e-01 -2.83748526e-02
6.25739694e-02 3.20696801e-01 2.45406911e-01 -2.17933685e-01
6.73627913e-01 2.71773666e-01 1.51827671e-02 -1.81919500e-01
1.59882531e-01 6.20662153e-01 4.25832421e-01 -3.25530432e-02
5.38068056e-01 4.02535290e-01 1.67218238e-01 -6.42963111e-01
-6.15339041e-01 -5.88701487e-01 -7.77256370e-01 -1.85984328e-01
1.18863225e+00 -9.83636200e-01 -2.27138728e-01 4.37624872e-01
-9.38392699e-01 9.12176818e-02 -4.36507165e-01 8.23911428e-01
-4.00752693e-01 4.97227758e-01 -9.62960720e-01 -3.51943076e-01
-7.60356605e-01 -1.73628283e+00 8.45814526e-01 4.80832785e-01
4.90144379e-02 -8.95133913e-01 -3.78013700e-01 4.76528287e-01
8.54724944e-01 2.88081635e-02 9.65319514e-01 -4.95654464e-01
-5.77616811e-01 -3.60495627e-01 -5.93691707e-01 4.96309638e-01
1.81037620e-01 -4.39725190e-01 -7.10463285e-01 -2.36525699e-01
3.24072063e-01 -1.29491895e-01 7.77810633e-01 5.88653862e-01
1.42204642e+00 5.70729598e-02 -2.62226343e-01 7.60879278e-01
1.66157103e+00 3.05181563e-01 3.57573032e-01 2.47744203e-01
7.18083739e-01 1.79007456e-01 2.33048856e-01 4.44698960e-01
3.03523302e-01 3.75556767e-01 3.91062677e-01 -4.82727438e-01
-4.70541596e-01 6.25267103e-02 -1.28559396e-01 9.77850795e-01
-1.96168572e-02 -1.52771398e-01 -8.43140304e-01 5.09722412e-01
-1.24158144e+00 -5.76048672e-01 -3.14302981e-01 2.07991695e+00
7.86285996e-01 2.26542711e-01 -1.86032981e-01 1.94887906e-01
6.43581986e-01 3.47448260e-01 -4.63653266e-01 -3.44459683e-01
1.92760691e-01 4.51784551e-01 6.58962309e-01 2.53491908e-01
-1.18915927e+00 5.11504948e-01 5.57262659e+00 1.13540709e+00
-1.40483963e+00 4.11990672e-01 6.76652670e-01 -1.52018219e-01
-1.40455924e-02 -4.05981123e-01 -7.55205750e-01 3.34749252e-01
7.66802967e-01 1.82449341e-01 -7.71370903e-02 6.96069896e-01
2.34801188e-01 -3.41725826e-01 -5.52334070e-01 7.70820975e-01
-7.28560761e-02 -1.24732471e+00 -2.27682501e-01 -2.00763509e-01
6.41464531e-01 2.68080086e-01 4.67629880e-02 1.28115743e-01
-3.37921500e-01 -7.42355108e-01 3.32172096e-01 2.06289008e-01
1.06934714e+00 -6.37363315e-01 8.62997830e-01 2.99942166e-01
-1.37883222e+00 -1.10483214e-01 -4.13115263e-01 3.94622862e-01
3.73366505e-01 7.04801619e-01 -1.27165008e+00 9.36727166e-01
4.22325969e-01 2.29798615e-01 -6.53871596e-01 1.19434452e+00
-1.05721384e-01 4.27103937e-01 -4.92191821e-01 1.30000859e-01
4.62707698e-01 1.28334807e-02 5.94867468e-01 1.45319569e+00
5.64194560e-01 4.41239998e-02 3.66669707e-02 6.05276108e-01
-6.21649213e-02 3.33245873e-01 -2.68714696e-01 3.80423427e-01
4.94534569e-03 1.38284886e+00 -1.00792539e+00 -5.23665547e-01
-6.56663775e-01 7.77857900e-01 3.63995284e-02 6.95231929e-02
-1.03205848e+00 -2.80584097e-01 -2.50186529e-02 4.22078997e-01
4.25195932e-01 -1.51466718e-03 -6.08739734e-01 -9.75796282e-01
-6.61233999e-03 -4.95291054e-01 5.76320410e-01 -2.65070021e-01
-8.02550614e-01 9.12734389e-01 -1.72793016e-01 -1.30073524e+00
-2.59367973e-01 -4.02195632e-01 -6.92066729e-01 1.01578283e+00
-1.74597132e+00 -9.09210026e-01 -3.94643515e-01 5.25123835e-01
7.59914935e-01 2.46020816e-02 6.18931890e-01 5.44438958e-01
-7.18993545e-01 5.45247734e-01 -1.16127208e-02 -7.50256628e-02
4.05559927e-01 -1.10357642e+00 2.70965714e-02 1.00639462e+00
-4.27631468e-01 1.30149141e-01 1.67825565e-01 -5.88639498e-01
-9.38352764e-01 -1.34964669e+00 7.40511835e-01 1.57656342e-01
4.35436159e-01 7.54006356e-02 -9.82686341e-01 2.62043566e-01
6.56596664e-03 1.21549681e-01 5.78466594e-01 -7.67731965e-01
3.17426890e-01 -4.62677218e-02 -1.35265923e+00 5.37307143e-01
5.66393733e-01 -2.62297153e-01 -5.13925552e-01 1.88465729e-01
5.51864028e-01 -7.33404934e-01 -1.06949425e+00 7.23426700e-01
3.17824513e-01 -1.13298810e+00 1.12355471e+00 1.60919994e-01
5.77879310e-01 -3.12280506e-01 6.82351738e-02 -8.55244994e-01
-4.01210874e-01 -3.66568267e-02 1.91763803e-01 6.71465755e-01
5.28897583e-01 -6.83971882e-01 1.04992294e+00 4.80541617e-01
-4.44696814e-01 -1.37288058e+00 -1.05176055e+00 -3.32064658e-01
-2.02871040e-02 -3.61706704e-01 2.32839152e-01 3.24154913e-01
-2.67171681e-01 -3.09619550e-02 1.56910166e-01 9.82538313e-02
5.02153218e-01 -1.77651241e-01 1.01272296e-02 -6.45455182e-01
-3.89400601e-01 -6.45361781e-01 -3.93951029e-01 -1.08683980e+00
-6.34112895e-01 -1.17239881e+00 -5.28947003e-02 -1.65154564e+00
2.79940903e-01 -4.41885680e-01 -5.28058350e-01 1.10654406e-01
-3.90294850e-01 5.37196994e-01 1.35595500e-01 2.20895380e-01
-2.60552108e-01 3.97114933e-01 1.76871705e+00 1.36706531e-01
-2.06718057e-01 2.16306373e-01 -4.56191301e-01 8.53964984e-01
8.22286248e-01 -5.61473548e-01 -6.08592629e-01 -3.71168882e-01
-2.81135738e-01 5.29014587e-01 3.72441411e-01 -1.32306635e+00
3.25935572e-01 4.27232474e-01 6.95176542e-01 -1.00284564e+00
3.15863341e-01 -8.34820211e-01 -7.19265342e-02 8.37674618e-01
-1.84909418e-01 1.82859808e-01 2.58198708e-01 3.75934392e-01
-5.10695517e-01 -5.63187480e-01 1.07310045e+00 -3.42468202e-01
-4.44559634e-01 3.81803542e-01 -5.25769830e-01 -2.33154610e-01
1.20950985e+00 -5.20083547e-01 1.41006500e-01 1.63122460e-01
-7.44340420e-01 1.31214797e-01 9.34435725e-02 -3.25732045e-02
6.90597475e-01 -1.01121771e+00 -7.36585796e-01 3.53435636e-01
-3.45249116e-01 3.99283707e-01 8.05207551e-01 1.50721192e+00
-9.50264215e-01 6.83203757e-01 -5.69852591e-02 -7.09721804e-01
-1.11961579e+00 2.20546186e-01 5.82185268e-01 -1.14268816e+00
-8.52282763e-01 9.52317119e-01 4.00992662e-01 -6.52864948e-02
-7.04155955e-03 -5.85639119e-01 -2.50250250e-01 8.89710616e-03
3.04384440e-01 2.77003169e-01 5.02028286e-01 -5.34022570e-01
-3.28501016e-01 5.05667150e-01 -3.39557976e-01 6.15083724e-02
1.12434208e+00 -1.26005292e-01 3.34219113e-02 -1.94668025e-01
1.28480315e+00 -2.26037592e-01 -1.20753205e+00 -2.82145530e-01
-2.03654587e-01 -1.30481467e-01 3.86989713e-01 -7.32549548e-01
-1.25027299e+00 1.00202584e+00 8.44534874e-01 -6.64550215e-02
1.58463728e+00 -2.76705891e-01 1.15222490e+00 -1.75056577e-01
-5.48789017e-02 -1.02443790e+00 1.06257901e-01 1.11675404e-01
6.34035707e-01 -1.25191307e+00 1.12822473e-01 -5.18657863e-01
-7.10829973e-01 1.24207091e+00 5.58847010e-01 -8.76587257e-02
5.25398314e-01 4.98915821e-01 -4.37689051e-02 -2.03178957e-01
-3.46670002e-01 -2.00290427e-01 3.80753994e-01 1.95502639e-01
7.37857461e-01 7.02405721e-02 -6.55966640e-01 7.00735092e-01
-3.18181291e-02 1.19431373e-02 1.80036262e-01 9.22332108e-01
-5.32412946e-01 -9.76475775e-01 -4.97219384e-01 7.39623427e-01
-9.46965575e-01 -1.95457220e-01 4.30197388e-01 8.81051242e-01
3.50354612e-01 6.20911002e-01 -1.32122889e-01 -9.29452553e-02
3.28126937e-01 2.17108540e-02 4.17863101e-01 -5.35468280e-01
-1.08133805e+00 4.21504855e-01 -1.25789329e-01 -4.08346772e-01
-1.22681901e-01 -5.57021916e-01 -1.53851962e+00 2.08500251e-01
-3.98461819e-01 -8.51161629e-02 8.25227082e-01 7.80871868e-01
1.07260466e-01 1.12918055e+00 5.44954300e-01 -6.30937278e-01
-6.95964456e-01 -8.39964867e-01 -2.30363041e-01 -4.22881395e-02
1.62953705e-01 -3.73956561e-01 -5.58792800e-02 -1.28611043e-01] | [15.165205001831055, -2.1720025539398193] |
f601d1df-f1ec-4094-8691-a14f0c9f7c83 | cross-modal-compression-towards-human | 2209.02574 | null | https://arxiv.org/abs/2209.02574v1 | https://arxiv.org/pdf/2209.02574v1.pdf | Cross Modal Compression: Towards Human-comprehensible Semantic Compression | Traditional image/video compression aims to reduce the transmission/storage cost with signal fidelity as high as possible. However, with the increasing demand for machine analysis and semantic monitoring in recent years, semantic fidelity rather than signal fidelity is becoming another emerging concern in image/video compression. With the recent advances in cross modal translation and generation, in this paper, we propose the cross modal compression~(CMC), a semantic compression framework for visual data, to transform the high redundant visual data~(such as image, video, etc.) into a compact, human-comprehensible domain~(such as text, sketch, semantic map, attributions, etc.), while preserving the semantic. Specifically, we first formulate the CMC problem as a rate-distortion optimization problem. Secondly, we investigate the relationship with the traditional image/video compression and the recent feature compression frameworks, showing the difference between our CMC and these prior frameworks. Then we propose a novel paradigm for CMC to demonstrate its effectiveness. The qualitative and quantitative results show that our proposed CMC can achieve encouraging reconstructed results with an ultrahigh compression ratio, showing better compression performance than the widely used JPEG baseline. | ['Wen Gao', 'Siwei Ma', 'Xinfeng Zhang', 'Chuanmin Jia', 'Jiguo Li'] | 2022-09-06 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [ 8.74971986e-01 -2.66254216e-01 -1.50436893e-01 -2.40423337e-01
-7.36902237e-01 -2.76437432e-01 5.85699618e-01 -4.42807004e-02
-1.49433270e-01 5.41637957e-01 3.84078562e-01 5.85885122e-02
-3.11452180e-01 -7.57782102e-01 -7.84219384e-01 -3.50563198e-01
2.10329831e-01 -1.37818471e-01 1.49943382e-01 -4.63927723e-02
5.04687965e-01 2.60519654e-01 -1.78043008e+00 3.03556144e-01
9.53323364e-01 1.37046885e+00 6.31608605e-01 5.60183227e-01
-1.31429225e-01 7.56960869e-01 -4.53765929e-01 -7.65036345e-01
2.50239462e-01 -5.42285681e-01 -8.79882514e-01 2.65457243e-01
3.37871432e-01 -4.16246384e-01 -7.82562017e-01 1.58096361e+00
2.48327821e-01 -1.38379186e-01 5.92662036e-01 -1.42353642e+00
-8.75562489e-01 4.86229569e-01 -7.02238441e-01 -1.70087531e-01
6.88051999e-01 -1.12891547e-01 8.45403969e-01 -8.22777867e-01
9.10132647e-01 1.29429007e+00 4.45279747e-01 2.46019453e-01
-1.21772516e+00 -6.43348515e-01 -2.31444806e-01 6.35153294e-01
-1.77009344e+00 -6.09269738e-01 8.61282885e-01 -1.78662449e-01
4.96456683e-01 4.56540287e-01 6.33388579e-01 1.05214739e+00
1.31597251e-01 7.53158629e-01 8.87477458e-01 -3.58709216e-01
7.53090754e-02 1.55150518e-01 -5.15214622e-01 5.47371984e-01
2.25873247e-01 -1.89611033e-01 -8.40375364e-01 1.65931046e-01
8.68182719e-01 1.73758417e-01 -6.15341663e-01 -2.28365868e-01
-1.40618181e+00 5.21372080e-01 3.03808212e-01 2.35430628e-01
-4.02361780e-01 3.18020672e-01 3.43788087e-01 4.15896833e-01
2.70646185e-01 7.88197890e-02 1.35830641e-01 -4.28983152e-01
-1.30136883e+00 1.60703346e-01 5.49405098e-01 1.37093794e+00
4.50949728e-01 -3.73452194e-02 9.26505625e-02 9.42867398e-01
5.06338000e-01 6.54011726e-01 5.42251527e-01 -1.26890683e+00
8.01238418e-01 3.66914690e-01 -2.13849336e-01 -1.51197934e+00
3.54768723e-01 -2.83281654e-01 -1.27592480e+00 -1.54514745e-01
-3.58105123e-01 5.58652878e-01 -5.28484881e-01 1.85585511e+00
-1.17799193e-01 2.50801980e-01 -3.33754942e-02 8.25252652e-01
5.03919184e-01 8.35489035e-01 -4.42710370e-02 -5.14505923e-01
1.20160890e+00 -6.67115986e-01 -1.13013303e+00 4.07998711e-02
2.29141768e-02 -9.39878404e-01 1.02166677e+00 4.53964263e-01
-1.47927713e+00 -6.94737911e-01 -1.40764701e+00 -4.03566092e-01
-1.48677260e-01 -2.62284279e-01 1.78483590e-01 3.87828439e-01
-1.09526336e+00 7.19197571e-01 -4.91486251e-01 -3.56652200e-01
4.60875362e-01 1.03950635e-01 -5.75868845e-01 -5.40651917e-01
-1.22930348e+00 6.15266621e-01 4.89809215e-01 -3.04389805e-01
-6.89033210e-01 -5.82796395e-01 -8.07358205e-01 1.17032059e-01
2.18711376e-01 -8.28918993e-01 9.02197301e-01 -9.08339679e-01
-1.33062744e+00 6.25988483e-01 -1.61652058e-01 -3.62865657e-01
6.43639743e-01 -2.21413642e-01 -5.36108673e-01 6.95222557e-01
5.62897958e-02 9.68727350e-01 1.19136798e+00 -1.32007897e+00
-8.27335715e-01 -1.15536973e-01 -9.70028266e-02 2.67478973e-01
-5.48495829e-01 -1.39749423e-01 -9.07058656e-01 -9.98967111e-01
2.93204695e-01 -5.86855531e-01 2.73003072e-01 4.74234313e-01
-2.77422458e-01 1.81114495e-01 1.14892495e+00 -9.39972758e-01
1.55399776e+00 -2.40125394e+00 7.07021892e-01 2.49463916e-01
3.66286010e-01 -9.59307887e-03 -9.52779576e-02 5.02209783e-01
9.76975784e-02 1.78547055e-01 -7.35935867e-01 -3.91215801e-01
7.37646595e-02 2.65546769e-01 -3.99241567e-01 4.20997471e-01
1.43735915e-01 9.64156508e-01 -7.88845837e-01 -8.55515897e-01
2.97310680e-01 6.86567724e-01 -6.24139011e-01 3.29023153e-01
9.43776071e-02 1.91498637e-01 -4.61869299e-01 7.98549533e-01
8.37662160e-01 -3.05855095e-01 4.83043678e-02 -5.82941413e-01
1.78382188e-01 -2.52311647e-01 -1.10931933e+00 2.15509462e+00
-2.66536415e-01 7.75582612e-01 1.73683375e-01 -1.16004634e+00
7.15287507e-01 2.82576442e-01 5.80997050e-01 -8.29002261e-01
4.15366329e-03 3.17394525e-01 -5.43368936e-01 -6.00227177e-01
8.35589826e-01 -1.38165355e-01 5.99472504e-03 5.89445001e-03
-6.32873401e-02 -4.61075038e-01 6.11164467e-03 4.88788694e-01
9.13273394e-01 1.94541827e-01 2.38731414e-01 -2.74315327e-02
4.79440957e-01 -1.44174889e-01 2.54833728e-01 2.27916315e-01
-1.26584142e-01 8.61665785e-01 2.29721457e-01 2.90698320e-01
-1.39371955e+00 -1.10225964e+00 -5.92973903e-02 4.98000383e-01
6.42901003e-01 -6.14260375e-01 -9.96777296e-01 -1.09015562e-01
-1.10178716e-01 5.25493026e-01 -1.08194999e-01 -4.87804055e-01
-5.15354097e-01 -1.89306408e-01 6.72256947e-01 2.08975688e-01
1.00801408e+00 -6.84364557e-01 -5.40395796e-01 1.34690270e-01
-6.84080482e-01 -1.30672681e+00 -5.94811440e-01 -4.17181313e-01
-8.15658569e-01 -8.80512118e-01 -7.34251082e-01 -9.16669607e-01
4.33441818e-01 6.27224922e-01 9.82505858e-01 1.88004747e-01
-2.57731616e-01 4.88860101e-01 -7.00339258e-01 1.25195071e-01
-4.05438274e-01 -2.45713249e-01 -4.80497591e-02 1.73904151e-01
-2.09676325e-01 -8.11010599e-01 -9.54185009e-01 1.77571207e-01
-1.60862613e+00 6.62488580e-01 6.73637152e-01 6.08687758e-01
7.15758204e-01 2.51450151e-01 4.31858778e-01 -5.84940374e-01
6.42827392e-01 -6.32658601e-01 -5.98603524e-02 3.87426972e-01
-7.52766490e-01 1.18445762e-01 6.23970747e-01 -1.99501991e-01
-7.98453569e-01 -2.93937266e-01 -1.76685542e-01 -7.00002491e-01
2.26357341e-01 5.55097520e-01 -4.46586907e-01 -1.45484269e-01
2.14564264e-01 8.74477267e-01 1.32487401e-01 -5.57957411e-01
4.94497865e-01 8.17433417e-01 9.88105595e-01 -4.05730218e-01
9.90950644e-01 5.11341095e-01 1.22550517e-01 -7.81790078e-01
-2.72758245e-01 -2.39987075e-01 -4.27753836e-01 -2.78884500e-01
7.63916671e-01 -8.21764886e-01 -5.19617021e-01 2.72177786e-01
-1.06641424e+00 4.61087793e-01 -3.79027396e-01 4.11146283e-01
-9.68192935e-01 9.73273695e-01 -5.09861290e-01 -3.98327112e-01
-2.58178592e-01 -1.42957091e+00 1.15349233e+00 -1.34226739e-01
4.57920991e-02 -7.28926539e-01 -4.02718991e-01 4.42565590e-01
5.52010059e-01 2.64593810e-01 8.32657158e-01 1.03965692e-01
-8.32378089e-01 -1.29092291e-01 -6.31083608e-01 4.15471166e-01
-6.61327988e-02 -1.68371826e-01 -5.92649341e-01 -4.60392565e-01
2.76340187e-01 -2.88984567e-01 7.05568016e-01 2.89374758e-02
1.59112895e+00 -4.00219023e-01 -1.26234621e-01 9.82064188e-01
1.53473020e+00 1.25169560e-01 1.03220415e+00 1.04656622e-01
5.32600820e-01 5.09159446e-01 7.21080661e-01 4.92488176e-01
4.37794358e-01 7.68349290e-01 5.68796992e-01 2.23360911e-01
-3.62699240e-01 -6.73663199e-01 3.79235387e-01 1.45615184e+00
-1.40568554e-01 -4.34938520e-01 -3.74200672e-01 3.33348930e-01
-1.83456004e+00 -9.93951142e-01 2.61611372e-01 2.08504891e+00
7.98131585e-01 -1.97513759e-01 -3.83060694e-01 5.77842653e-01
7.99593806e-01 2.76725382e-01 -3.68416876e-01 -1.44502493e-02
-2.80311555e-01 1.12233929e-01 5.06150067e-01 2.77916402e-01
-8.06585729e-01 6.40194833e-01 6.18100023e+00 1.48435700e+00
-1.00393534e+00 1.48888856e-01 5.46594083e-01 3.25393192e-02
-6.35128140e-01 -1.44903660e-01 -1.63802087e-01 6.21950328e-01
1.09526610e+00 -5.49417496e-01 7.57620633e-01 5.07734418e-01
1.63564980e-01 7.58975297e-02 -1.01179004e+00 1.58979034e+00
3.91550422e-01 -1.47479808e+00 6.69358492e-01 2.32907236e-01
4.11596507e-01 -5.98549426e-01 2.16144383e-01 -2.48983479e-03
-4.32042003e-01 -9.95604217e-01 1.22142935e+00 6.74289107e-01
1.49841523e+00 -7.16400683e-01 3.90764743e-01 2.04954460e-01
-1.51050019e+00 8.27892944e-02 -4.44400847e-01 3.69631141e-01
4.52696562e-01 4.11261290e-01 5.94654083e-02 9.49408293e-01
6.70426130e-01 1.15270078e+00 -4.15801555e-01 8.15884054e-01
2.00986609e-01 1.76054180e-01 5.05796187e-02 3.88365358e-01
1.45558864e-02 -2.41480350e-01 5.23984909e-01 1.01858664e+00
7.56578982e-01 2.44759947e-01 1.54636763e-02 7.37778008e-01
-3.90243322e-01 1.06999166e-01 -7.18841672e-01 -2.64846265e-01
5.20467281e-01 7.95456231e-01 -4.08268929e-01 -4.13910359e-01
-4.47655022e-01 1.48454237e+00 1.51979225e-02 2.27947265e-01
-1.05408454e+00 -4.51685220e-01 5.00211418e-01 1.51250556e-01
1.04909368e-01 -2.66694069e-01 -7.40570277e-02 -1.35143030e+00
3.07600021e-01 -9.73510742e-01 -1.84777882e-02 -8.47960174e-01
-9.92158949e-01 4.97618020e-01 2.16368988e-01 -1.71769679e+00
-1.60787683e-02 -2.93529749e-01 2.61479653e-02 6.24717832e-01
-1.59817624e+00 -1.14502871e+00 -3.20477456e-01 8.79901588e-01
8.97261322e-01 -1.35793924e-01 5.61223507e-01 6.93434060e-01
-1.29854679e-01 6.03323340e-01 2.40486935e-01 -2.97342926e-01
5.67117155e-01 -7.71802783e-01 1.23494044e-01 9.23517466e-01
-5.52385747e-02 5.08538544e-01 5.26361763e-01 -6.54952288e-01
-1.99154365e+00 -1.07185864e+00 6.88632905e-01 2.22615346e-01
5.14094412e-01 -2.14313462e-01 -9.67084706e-01 2.60344893e-01
2.69217640e-01 -1.24717534e-01 4.54458088e-01 -9.18232679e-01
-4.17704314e-01 -1.31509885e-01 -1.48505616e+00 5.82159996e-01
1.43184757e+00 -8.07057023e-01 -6.01000488e-01 8.12058300e-02
1.29549921e+00 -1.57358810e-01 -1.15728462e+00 4.80694503e-01
6.54565752e-01 -9.35340106e-01 1.24219573e+00 -5.31442836e-02
1.02578378e+00 -3.98065269e-01 -8.82447720e-01 -9.21129346e-01
-2.87254363e-01 -5.51632345e-01 -3.05304497e-01 1.18615186e+00
-1.66785102e-02 -2.31227487e-01 4.05182987e-01 3.24667841e-01
-8.06562752e-02 -4.92685705e-01 -1.18201029e+00 -5.79537570e-01
-3.95704120e-01 -6.22398019e-01 5.39922059e-01 9.04609323e-01
1.29549587e-02 -3.03453896e-02 -8.42055142e-01 -4.25983518e-02
8.27915192e-01 1.91661119e-02 4.05729353e-01 -8.67000341e-01
-2.07521632e-01 -3.49633306e-01 -7.76778579e-01 -1.34240603e+00
-1.18115604e-01 -9.52530205e-01 -7.99546540e-02 -1.42553961e+00
5.25925159e-01 -4.52156886e-02 -2.52472818e-01 -1.54145077e-01
1.73203439e-01 5.91617942e-01 4.45027202e-01 6.59336746e-01
-6.75241470e-01 8.97943079e-01 1.40096223e+00 -2.68753827e-01
3.97404283e-01 -4.85559672e-01 -7.04114795e-01 4.51676130e-01
3.16013157e-01 -1.60250738e-01 -6.33164346e-01 -5.36450863e-01
1.87490195e-01 3.93637568e-01 2.63997436e-01 -1.15296197e+00
2.80038357e-01 -2.28027269e-01 1.49127021e-01 -4.58894670e-01
4.74261522e-01 -1.21476960e+00 5.20850837e-01 6.16891801e-01
-4.22885418e-01 1.37748778e-01 -1.76888987e-01 9.02175426e-01
-7.30652094e-01 -1.15580231e-01 8.62344742e-01 -1.53204119e-02
-6.70123756e-01 2.95861274e-01 4.38118819e-03 -1.53061181e-01
1.01455426e+00 -5.40354967e-01 -3.30418915e-01 -5.76072156e-01
-4.55266058e-01 -1.19211338e-01 6.99975669e-01 6.43014014e-01
1.25359619e+00 -1.50316298e+00 -7.34804988e-01 2.81577915e-01
2.17605636e-01 -2.57231027e-01 3.13714117e-01 6.39421165e-01
-8.82654309e-01 3.56020719e-01 -2.91639626e-01 -5.15746117e-01
-1.11679912e+00 6.30142272e-01 -3.14879924e-01 9.54147987e-03
-8.47867906e-01 3.79259109e-01 1.97057590e-01 2.53563076e-01
1.69114605e-01 -1.64615229e-01 -5.60081713e-02 -2.34981209e-01
5.66532791e-01 4.24234211e-01 -3.18454146e-01 -8.49442184e-01
-1.29102066e-01 7.19220579e-01 2.63378620e-01 -3.21393460e-01
1.28580129e+00 -5.92375517e-01 -2.01336771e-01 1.96908131e-01
1.49524140e+00 -2.13836148e-01 -1.11026573e+00 -2.61417896e-01
-7.03092068e-02 -1.01607227e+00 9.54385623e-02 -3.42496902e-01
-1.09447372e+00 8.54825139e-01 6.89646244e-01 2.14810744e-01
1.54300630e+00 -1.00847542e-01 1.12894607e+00 -1.41921878e-01
6.41666353e-01 -1.02855504e+00 1.86365023e-01 -1.48173030e-02
1.21444571e+00 -9.00616050e-01 2.29524061e-01 -6.10518098e-01
-6.05654418e-01 9.47361648e-01 8.75158533e-02 9.00830925e-02
4.94110703e-01 2.05641054e-02 -5.69737434e-01 5.96099673e-03
-6.15782499e-01 2.11202413e-01 3.84961605e-01 4.87122029e-01
1.87449932e-01 -2.49233041e-02 -4.69357133e-01 2.76683599e-01
-2.86443025e-01 8.14433992e-02 3.85564595e-01 1.12074852e+00
-6.11477137e-01 -1.11047816e+00 -3.61333162e-01 4.15342808e-01
-4.10280615e-01 -2.12639704e-01 -1.06835984e-01 4.71417487e-01
1.70728102e-01 1.03081489e+00 3.35217193e-02 -6.36774421e-01
2.94885993e-01 -7.12668151e-02 3.97927523e-01 -1.40899390e-01
5.22077903e-02 9.73283723e-02 -3.34763348e-01 -7.83376217e-01
-7.99337387e-01 -3.82415503e-01 -9.75599468e-01 -7.02412069e-01
-7.45671540e-02 -7.33793378e-02 9.33398187e-01 7.04568982e-01
3.92308503e-01 5.14320791e-01 7.80065179e-01 -6.66306317e-01
-2.54215568e-01 -6.48433983e-01 -6.23271108e-01 9.14542675e-01
9.30984393e-02 -4.48679447e-01 -2.44361579e-01 5.23149967e-01] | [11.2887544631958, -1.7411006689071655] |
1227845f-c392-41c9-8602-b15a2d648306 | pre-training-transformers-for-molecular | 2207.02724 | null | https://arxiv.org/abs/2207.02724v1 | https://arxiv.org/pdf/2207.02724v1.pdf | Pre-training Transformers for Molecular Property Prediction Using Reaction Prediction | Molecular property prediction is essential in chemistry, especially for drug discovery applications. However, available molecular property data is often limited, encouraging the transfer of information from related data. Transfer learning has had a tremendous impact in fields like Computer Vision and Natural Language Processing signaling for its potential in molecular property prediction. We present a pre-training procedure for molecular representation learning using reaction data and use it to pre-train a SMILES Transformer. We fine-tune and evaluate the pre-trained model on 12 molecular property prediction tasks from MoleculeNet within physical chemistry, biophysics, and physiology and show a statistically significant positive effect on 5 of the 12 tasks compared to a non-pre-trained baseline model. | ['Erik Ylipää', 'Maria Bånkestad', 'Johan Broberg'] | 2022-07-06 | null | null | null | null | ['molecular-property-prediction'] | ['miscellaneous'] | [ 4.84708279e-01 -1.98458359e-01 -6.64183259e-01 -3.44521970e-01
-7.88377106e-01 -5.03407240e-01 5.66055417e-01 1.03883195e+00
-5.67994952e-01 1.43046868e+00 3.38190705e-01 -7.34927833e-01
-3.37923802e-02 -7.69070864e-01 -1.00389373e+00 -8.52441013e-01
-3.65991831e-01 2.69320369e-01 1.82548061e-01 -1.09446257e-01
5.34304082e-01 7.71344304e-01 -8.32544029e-01 6.71091259e-01
8.89017463e-01 6.53586745e-01 8.09402838e-02 3.55965495e-01
-3.22295912e-02 8.97817612e-01 -2.14477435e-01 -2.16364801e-01
-2.44736493e-01 -3.99822146e-01 -1.03466558e+00 -8.98856163e-01
3.98329318e-01 3.09186101e-01 -2.86561936e-01 6.29607141e-01
7.54082143e-01 5.26732802e-01 1.01399493e+00 -3.51650834e-01
-5.86473048e-01 3.47859561e-01 -1.83523685e-01 2.61346877e-01
6.45233691e-01 2.52012461e-01 1.17491698e+00 -7.05871046e-01
8.51225138e-01 1.01388252e+00 5.54799139e-01 5.49502432e-01
-1.58864176e+00 -8.29246640e-01 -2.63279766e-01 4.73498434e-01
-1.00864565e+00 -2.29344919e-01 3.99310648e-01 -4.99214768e-01
1.69408703e+00 -1.48039296e-01 3.24300915e-01 1.23250866e+00
6.37959301e-01 4.62105900e-01 9.19994473e-01 -1.17356293e-01
3.49239200e-01 -3.90238352e-02 2.36890046e-03 4.75101501e-01
-9.00963992e-02 3.90712827e-01 -6.60427928e-01 -3.45239908e-01
4.01578128e-01 2.56607179e-02 -3.31595272e-01 -2.83446193e-01
-1.25958931e+00 8.02965581e-01 8.98481965e-01 2.12974295e-01
-5.64950645e-01 9.17791203e-02 5.38909435e-01 5.02138317e-01
6.27164602e-01 1.17430139e+00 -1.03076375e+00 -1.41703352e-01
-5.78450561e-01 1.58000812e-01 8.08562100e-01 3.35370898e-01
7.13137031e-01 -2.97716856e-01 -9.54854563e-02 5.92196226e-01
3.33863273e-02 1.10853888e-01 4.07340020e-01 -7.01324269e-02
3.34333211e-01 4.97616500e-01 -5.05374596e-02 -4.09263879e-01
-5.59940457e-01 -9.90702882e-02 -5.29879034e-01 2.51693398e-01
2.27833226e-01 1.21720113e-01 -9.88397062e-01 1.47812831e+00
2.69176871e-01 3.30957860e-01 2.79724836e-01 4.30783600e-01
9.90499616e-01 1.01096094e+00 1.05098367e+00 -2.57764101e-01
8.75295639e-01 -7.06228375e-01 -2.88852602e-01 3.86611491e-01
9.65148389e-01 -7.49907792e-01 7.78476059e-01 5.73660970e-01
-5.97986698e-01 -4.19748276e-01 -1.06725383e+00 -6.93100244e-02
-9.50889289e-01 -2.55868256e-01 1.16956556e+00 4.21990842e-01
-4.17332083e-01 1.45226419e+00 -5.68638742e-01 -1.62671864e-01
8.31497848e-01 7.73368418e-01 -8.59998941e-01 -1.39701188e-01
-1.20095336e+00 1.31057405e+00 9.32588220e-01 -6.31310105e-01
-8.68314445e-01 -1.45681489e+00 -6.50770426e-01 1.30916074e-01
1.21421896e-01 -7.92405963e-01 1.02025259e+00 -5.41901410e-01
-1.84559536e+00 5.34456491e-01 -9.05478895e-02 -4.69281644e-01
7.27218315e-02 2.59503555e-02 -4.73526210e-01 -9.72928554e-02
-1.97185561e-01 7.38298476e-01 4.11754787e-01 -6.15763247e-01
-2.12516263e-01 -1.00721575e-01 -2.50088006e-01 7.23034376e-04
-9.92917791e-02 -5.35960346e-02 4.46510971e-01 -4.71720248e-01
-4.82962340e-01 -5.60768068e-01 -3.99669141e-01 -3.04757029e-01
-2.30302557e-01 -4.62153912e-01 3.74277353e-01 -3.07037920e-01
6.57441258e-01 -1.71972454e+00 3.13771784e-01 3.99004668e-01
2.16323301e-01 4.62012410e-01 -4.40962583e-01 8.93419325e-01
-8.67109895e-01 1.74016967e-01 -8.74774605e-02 2.62207031e-01
-5.69747567e-01 -2.72760659e-01 -3.04221183e-01 5.16089976e-01
3.91576797e-01 9.79806125e-01 -1.15339148e+00 -2.80282684e-02
3.03516060e-01 5.85939825e-01 -5.32112539e-01 3.32089007e-01
-7.45682001e-01 8.88773084e-01 -6.77457631e-01 7.67543793e-01
4.10828531e-01 -4.05802935e-01 1.46563962e-01 -3.36669266e-01
-2.03529060e-01 9.24896359e-01 -2.68778443e-01 1.73853230e+00
-4.23101515e-01 1.79265544e-01 -8.66660595e-01 -9.20457184e-01
7.16748416e-01 4.88554478e-01 7.15430737e-01 -8.57974827e-01
3.99818197e-02 2.50277787e-01 4.23070490e-01 -3.25370014e-01
-2.38066930e-02 -5.94755411e-01 5.22625923e-01 3.07937086e-01
3.45777720e-01 -9.79563594e-02 2.15716124e-01 5.01621515e-02
1.20131505e+00 4.52832222e-01 6.18267775e-01 -9.26013514e-02
8.92749071e-01 -5.64640723e-02 9.63080674e-02 2.72715241e-01
1.62069157e-01 8.71069059e-02 3.07032138e-01 -6.66119158e-01
-9.74377513e-01 -7.64271140e-01 -4.04802114e-01 1.44286406e+00
-2.80732125e-01 -7.10476160e-01 -2.73792416e-01 -7.21652865e-01
2.06676006e-01 4.60636288e-01 -7.18566656e-01 -4.88291949e-01
-1.91470325e-01 -8.64190757e-01 4.18121010e-01 3.83597225e-01
1.20460086e-01 -1.60484099e+00 2.80736864e-01 4.59092051e-01
5.33045232e-01 -7.67303526e-01 -1.26764044e-01 6.59045458e-01
-9.48595464e-01 -1.12369847e+00 -4.49869961e-01 -5.40115595e-01
3.59753698e-01 -1.74611360e-02 1.02036095e+00 -2.42683634e-01
-5.23323357e-01 -2.40846202e-01 -3.35488766e-01 -7.71912813e-01
-6.19647443e-01 4.70925629e-01 4.86637615e-02 -3.18993151e-01
4.42517847e-01 -6.21019542e-01 -5.50754249e-01 -1.52575836e-01
-7.17256963e-01 -2.15307564e-01 8.35595548e-01 7.04802096e-01
6.30735934e-01 -5.69894314e-01 8.96825492e-01 -1.06702316e+00
6.08459890e-01 -5.92078209e-01 -4.42021877e-01 1.81839421e-01
-6.80836499e-01 4.60240424e-01 6.95269585e-01 -4.41828787e-01
-7.28502452e-01 2.54845172e-01 -3.31008315e-01 -3.57578956e-02
-1.58449680e-01 9.81593072e-01 -1.86690465e-01 -4.89672095e-01
7.90412545e-01 5.32180117e-03 -1.40615940e-01 -5.34974337e-01
2.79644907e-01 1.85042724e-01 6.00684658e-02 -7.35152423e-01
5.39635301e-01 -9.83536541e-02 5.35896242e-01 -8.24731410e-01
-7.31613576e-01 -6.09099269e-01 -6.15508616e-01 4.96490747e-01
7.98628449e-01 -6.72491491e-01 -1.27234447e+00 -3.38846594e-02
-1.15162778e+00 -3.05617034e-01 -7.13284835e-02 8.66199493e-01
-5.32756090e-01 3.69640440e-01 -5.02936721e-01 -1.48153767e-01
-5.79639792e-01 -9.96375680e-01 6.55385077e-01 -3.85002755e-02
-3.71525884e-01 -1.30681837e+00 6.51744723e-01 7.20892847e-02
3.61613035e-01 2.65425056e-01 1.60232389e+00 -9.89455640e-01
-3.81303400e-01 -2.29005933e-01 -1.24978043e-01 4.69314866e-02
5.73380053e-01 -1.32683262e-01 -1.08338237e+00 -1.12227350e-01
-8.05705667e-01 -7.17991650e-01 1.39016616e+00 2.43660852e-01
1.34024739e+00 6.04732558e-02 -5.23238659e-01 5.17887354e-01
1.10156322e+00 3.28438699e-01 7.92744696e-01 3.35302681e-01
7.41796911e-01 4.30024266e-01 3.10775578e-01 2.30530605e-01
-3.05501431e-01 4.41460043e-01 3.32441926e-01 -2.44412318e-01
1.43583506e-01 -5.39039195e-01 2.41349146e-01 1.47048771e-01
-5.20541549e-01 -2.36724943e-01 -6.72228158e-01 -3.42603996e-02
-1.61590302e+00 -1.13856852e+00 6.55812398e-02 2.36148548e+00
1.52264643e+00 -4.63229381e-02 1.17931314e-01 -1.32943526e-01
1.38985515e-01 -1.19851641e-01 -6.95276082e-01 -5.54821670e-01
-5.93532585e-02 9.82727289e-01 5.66141188e-01 5.36292553e-01
-1.19418418e+00 1.37025654e+00 6.89980221e+00 9.77294266e-01
-1.36388838e+00 -3.30068082e-01 5.99585056e-01 3.20418209e-01
-1.86195478e-01 -7.54502490e-02 -5.32062232e-01 1.30556062e-01
1.37873614e+00 -1.99261248e-01 1.58193141e-01 6.95067167e-01
3.22280735e-01 6.32585362e-02 -1.71216571e+00 8.34088087e-01
-5.15754104e-01 -2.04324365e+00 5.55178702e-01 1.52066186e-01
6.63988709e-01 2.19351500e-01 5.32784499e-02 3.40644807e-01
-3.64870764e-02 -1.71965659e+00 -1.55339077e-01 4.19838876e-01
7.42499769e-01 -8.28496933e-01 5.11628747e-01 6.70834482e-02
-7.46733844e-01 4.29765701e-01 -5.66639662e-01 -5.98895252e-02
-2.85996050e-01 4.74341899e-01 -1.42604291e+00 6.24149323e-01
7.31259882e-02 1.24153018e+00 -3.00827563e-01 1.17788064e+00
-1.76037893e-01 8.48100841e-01 -2.23063976e-02 -2.23803595e-01
2.43050501e-01 -3.75775695e-01 9.99875814e-02 1.44975483e+00
-1.55829802e-01 1.56598151e-01 1.78193241e-01 8.79541218e-01
-5.65346062e-01 5.80171824e-01 -5.27086794e-01 -6.55626893e-01
2.19122103e-05 1.15193605e+00 -4.39418435e-01 -2.89715707e-01
-3.96900088e-01 8.52673233e-01 2.55705208e-01 3.35363030e-01
-7.58863389e-01 -4.51829165e-01 8.61159861e-01 4.40663621e-02
2.81028241e-01 -1.44794896e-01 2.81693600e-02 -8.76626790e-01
-6.76216483e-01 -7.94556439e-01 2.53753841e-01 -3.90910625e-01
-1.47545540e+00 3.48953694e-01 -2.32884243e-01 -8.43959808e-01
8.43533576e-02 -1.20660603e+00 -6.65084004e-01 1.08442497e+00
-1.80793607e+00 -1.17663252e+00 2.23372921e-01 4.27060187e-01
1.79621279e-01 -3.46491605e-01 1.25756788e+00 4.08537745e-01
-3.63771647e-01 4.99809206e-01 2.35899061e-01 -3.56938034e-01
1.17359698e+00 -1.29883552e+00 4.12461549e-01 -3.32477018e-02
4.25744578e-02 9.44407940e-01 5.49362719e-01 -7.49897897e-01
-1.44111490e+00 -1.18727827e+00 8.44968081e-01 -5.38011074e-01
7.51175940e-01 -1.47484168e-01 -1.24299634e+00 3.42286140e-01
-1.20129120e-02 6.78579882e-02 1.30655670e+00 3.29734027e-01
-8.44605505e-01 2.80152529e-01 -1.07588363e+00 1.80894837e-01
7.05229700e-01 -5.96478283e-01 -4.61174995e-01 5.81696928e-01
6.64440274e-01 -1.98215067e-01 -1.27418709e+00 4.87079114e-01
5.72284579e-01 -2.80921072e-01 1.31892300e+00 -1.48493481e+00
5.73445022e-01 -5.75413890e-02 1.60163566e-01 -1.46861875e+00
-4.30131078e-01 -5.89981437e-01 4.37458977e-03 6.30487978e-01
1.07789540e+00 -6.60191476e-01 8.57200265e-01 4.22278643e-01
-3.17859575e-02 -6.87760055e-01 -6.79285049e-01 -6.89800680e-01
5.27003586e-01 -3.43738534e-02 1.60929859e-01 9.95988131e-01
3.64292383e-01 1.02451217e+00 -1.49132341e-01 -2.32526466e-01
4.01338264e-02 1.02702104e-01 6.41167462e-01 -1.38098395e+00
-4.78878826e-01 -4.08802986e-01 -5.23689210e-01 -8.39717090e-01
4.13098574e-01 -1.37803161e+00 -3.82122248e-01 -1.40211833e+00
4.95216191e-01 -1.71604395e-01 -8.51195574e-01 7.72934854e-01
-2.71461934e-01 -6.99259341e-02 -2.59074301e-01 7.09188059e-02
-5.72112620e-01 7.04282939e-01 1.35453129e+00 -4.69810545e-01
-4.66074228e-01 4.86404449e-02 -4.03577179e-01 1.25672802e-01
8.07736933e-01 -6.01885378e-01 -1.43391252e-01 4.84931231e-01
3.44113052e-01 -2.98048407e-01 -1.33988373e-02 -7.58810818e-01
-1.45781338e-01 -4.48087245e-01 6.85020387e-01 -2.40176961e-01
3.25004309e-01 -5.40604830e-01 -2.24305153e-01 6.58406675e-01
-5.41070580e-01 -5.38483083e-01 4.86967504e-01 6.96731567e-01
-1.02987356e-01 -2.38899495e-02 8.18013906e-01 -2.64984258e-02
-5.15864074e-01 7.63817310e-01 -4.88762110e-01 -4.56311136e-01
8.30139577e-01 -1.57005370e-01 -3.08819294e-01 -6.43446285e-04
-6.87239826e-01 -1.05984479e-01 1.07048430e-01 2.05704942e-01
5.75547993e-01 -8.03054810e-01 -5.94810128e-01 -1.41570136e-01
4.28296000e-01 -5.32396972e-01 -2.18577161e-01 6.56379521e-01
-5.15093267e-01 8.31551552e-01 -2.81596869e-01 -1.02436282e-01
-1.23856366e+00 7.31380701e-01 3.79104853e-01 -3.49652559e-01
-2.16147095e-01 5.39969385e-01 1.78112909e-01 -3.85570765e-01
7.54425749e-02 -4.96859282e-01 -2.88310468e-01 -9.26457569e-02
4.95056689e-01 2.40238324e-01 3.47667754e-01 -8.66653323e-02
-4.38368499e-01 5.09141088e-01 -5.48404455e-01 3.96014780e-01
1.72236300e+00 8.22124958e-01 -1.86828971e-01 2.90772438e-01
1.48366761e+00 -1.18218482e-01 -8.47948611e-01 -2.00738117e-01
3.24881673e-01 -3.14610638e-03 1.19651081e-02 -1.10981691e+00
-2.61007369e-01 9.90962267e-01 2.95298994e-01 -4.48321790e-01
5.15639246e-01 -3.22369230e-03 5.96668243e-01 1.23912656e+00
1.79458663e-01 -6.35215402e-01 1.98234454e-01 8.03008676e-01
6.62005842e-01 -1.43724120e+00 4.88009721e-01 -4.73106980e-01
-4.20113802e-01 1.41305077e+00 2.98246413e-01 -2.98452172e-02
6.52166486e-01 -2.03490660e-01 -4.12053525e-01 -3.13750356e-01
-8.15535307e-01 -1.53738454e-01 5.19801676e-01 6.73138618e-01
1.21788788e+00 -6.19049482e-02 -2.81011939e-01 2.61567503e-01
2.33402520e-01 1.94696352e-01 1.70182213e-02 9.99081492e-01
-5.82173288e-01 -1.79128170e+00 1.47036642e-01 2.01403886e-01
-7.45659292e-01 -5.91583371e-01 -9.77203131e-01 5.39563417e-01
-7.09966943e-02 5.22827506e-01 -4.44428980e-01 -3.10765058e-01
1.09701417e-01 1.86295360e-01 8.91887605e-01 -8.76961231e-01
-7.54122555e-01 -2.17068508e-01 9.83177647e-02 -4.17284012e-01
-6.11381173e-01 -1.49033695e-01 -1.57238758e+00 -3.23641717e-01
-3.26468915e-01 4.16035205e-01 7.43616939e-01 9.07953858e-01
6.17238581e-01 5.17517507e-01 5.13670206e-01 -8.34428430e-01
-3.31943870e-01 -1.05234134e+00 -8.77795517e-02 3.21537971e-01
2.40214944e-01 -5.15802383e-01 1.52659938e-01 1.55003175e-01] | [5.05210018157959, 5.8477301597595215] |
935f57d7-82b3-4f7d-9dd6-dba7abc0da4e | movieclip-visual-scene-recognition-in-movies | 2210.11065 | null | https://arxiv.org/abs/2210.11065v2 | https://arxiv.org/pdf/2210.11065v2.pdf | MovieCLIP: Visual Scene Recognition in Movies | Longform media such as movies have complex narrative structures, with events spanning a rich variety of ambient visual scenes. Domain specific challenges associated with visual scenes in movies include transitions, person coverage, and a wide array of real-life and fictional scenarios. Existing visual scene datasets in movies have limited taxonomies and don't consider the visual scene transition within movie clips. In this work, we address the problem of visual scene recognition in movies by first automatically curating a new and extensive movie-centric taxonomy of 179 scene labels derived from movie scripts and auxiliary web-based video datasets. Instead of manual annotations which can be expensive, we use CLIP to weakly label 1.12 million shots from 32K movie clips based on our proposed taxonomy. We provide baseline visual models trained on the weakly labeled dataset called MovieCLIP and evaluate them on an independent dataset verified by human raters. We show that leveraging features from models pretrained on MovieCLIP benefits downstream tasks such as multi-label scene and genre classification of web videos and movie trailers. | ['Shrikanth Narayanan', 'Huisheng Wang', 'Kree Cole-McLaughlin', 'Yin Cui', 'Haoyang Zhang', 'Krishna Somandepalli', 'Rajat Hebbar', 'Digbalay Bose'] | 2022-10-20 | null | null | null | null | ['scene-recognition', 'genre-classification'] | ['computer-vision', 'computer-vision'] | [ 1.83142185e-01 -6.51475549e-01 -4.32705164e-01 -5.66141069e-01
-7.27305472e-01 -1.37563694e+00 6.88840985e-01 7.85131902e-02
-3.21810842e-01 2.62382865e-01 7.26454318e-01 1.21823043e-01
3.55970442e-01 -1.54807195e-01 -7.58325815e-01 -1.69688806e-01
4.07271162e-02 -1.20942466e-01 5.06142974e-01 -2.57075205e-02
1.85952559e-01 8.19841102e-02 -1.63442540e+00 1.33874393e+00
-7.65676424e-02 1.34481525e+00 1.63783848e-01 1.00675070e+00
-2.27492955e-02 1.64493155e+00 -4.98550236e-01 -6.77415788e-01
1.65983394e-01 -4.18979585e-01 -9.81132269e-01 6.39836133e-01
1.08533084e+00 -3.58034283e-01 -6.05544448e-01 7.33909607e-01
9.77954715e-02 4.82729614e-01 6.87246859e-01 -1.33636856e+00
-5.76192677e-01 7.03931034e-01 -3.21688354e-01 6.89543247e-01
7.98007429e-01 1.68593213e-01 1.22231400e+00 -9.98516858e-01
1.40130270e+00 9.89717364e-01 6.39594972e-01 3.52903575e-01
-1.04064274e+00 -4.65848058e-01 4.02385771e-01 5.66517830e-01
-1.30477881e+00 -7.31610000e-01 8.37563634e-01 -1.11535835e+00
9.30135190e-01 4.72706169e-01 5.66768348e-01 1.92291343e+00
-3.30703229e-01 7.72608101e-01 8.82921875e-01 -1.72985010e-02
-1.84597317e-02 2.89016008e-01 -7.30221625e-03 5.26256144e-01
-5.36614180e-01 -5.51233649e-01 -8.56143296e-01 1.47440106e-01
6.06695414e-01 1.36253715e-01 -2.55418271e-01 -4.44700181e-01
-1.25304294e+00 6.87607944e-01 1.37691572e-01 1.25925764e-01
-5.22009693e-02 1.02390364e-01 1.06508410e+00 1.96797416e-01
4.32362139e-01 5.15018344e-01 -1.70459852e-01 -4.43785399e-01
-1.05114233e+00 4.09915179e-01 4.62015629e-01 1.31794071e+00
3.22237104e-01 -1.01767890e-01 -5.37288845e-01 1.08814728e+00
-1.79922089e-01 2.24125728e-01 1.95138201e-01 -1.33911002e+00
7.73453355e-01 3.46193105e-01 1.66694283e-01 -1.39683771e+00
-3.52310300e-01 6.40080646e-02 -4.28744853e-01 -3.73586386e-01
4.58180904e-01 2.28880584e-01 -6.77974820e-01 1.56345141e+00
2.95292884e-01 1.13145642e-01 -5.26320577e-01 1.16490614e+00
1.42929339e+00 8.87854218e-01 1.19303040e-01 -2.16888189e-01
1.43953645e+00 -1.35422671e+00 -7.18288898e-01 -1.87694728e-01
3.07972640e-01 -7.05984533e-01 1.41365802e+00 6.33613467e-01
-1.04456460e+00 -7.38808751e-01 -7.28888333e-01 -4.25152928e-01
-3.02587897e-01 2.04406351e-01 5.04291594e-01 1.21352524e-01
-7.41310894e-01 3.56614143e-01 -3.85720730e-01 -5.00016809e-01
7.21743166e-01 -4.90936935e-01 -7.31164396e-01 -2.06395656e-01
-7.89221406e-01 4.33044791e-01 9.28819627e-02 -3.47291738e-01
-1.70911217e+00 -6.27358139e-01 -9.99455750e-01 -1.68225080e-01
6.39843225e-01 -5.60149178e-02 1.41244340e+00 -1.24752605e+00
-1.08095098e+00 1.33309472e+00 -8.52363110e-02 -6.63330406e-02
4.41155076e-01 -2.41297707e-01 -6.87962115e-01 7.54106462e-01
3.27385247e-01 3.89185905e-01 8.63506556e-01 -1.33531070e+00
-7.93191195e-01 2.15233177e-01 6.33008420e-01 1.77368984e-01
-7.40704000e-01 7.04161227e-01 -9.02891397e-01 -8.49218071e-01
-4.84533757e-01 -6.83149159e-01 2.04466097e-02 -3.61301936e-02
-4.77124512e-01 1.60144478e-01 5.40357053e-01 -7.06769288e-01
1.37053013e+00 -2.46180105e+00 2.71752149e-01 -4.32086259e-01
4.34252799e-01 -4.10851926e-01 -1.65989310e-01 6.47848189e-01
1.29695550e-01 -7.93641061e-02 3.26671988e-01 -6.22688293e-01
-1.52140677e-01 -8.95449817e-02 -5.48204660e-01 5.47836602e-01
-1.62835956e-01 6.49003267e-01 -1.02330995e+00 -9.25890923e-01
1.09857455e-01 -1.02723474e-02 -7.80102313e-01 5.55500150e-01
-3.51630390e-01 4.68420237e-01 -8.13985467e-02 8.60532105e-01
1.65539280e-01 -5.88427603e-01 2.02978760e-01 -4.06781167e-01
-6.33229613e-02 1.36228919e-01 -9.41774309e-01 2.13216829e+00
-3.24489057e-01 1.18013561e+00 -9.33378190e-03 -8.54127109e-01
3.92398655e-01 3.77623171e-01 4.90524232e-01 -4.67945576e-01
1.75010115e-01 -6.27659678e-01 -6.13067627e-01 -1.06927192e+00
8.09437692e-01 -2.27301996e-02 -7.03497767e-01 1.80840746e-01
6.64284825e-01 1.34181634e-01 5.99968195e-01 6.41239226e-01
1.20703197e+00 3.31603974e-01 1.76511139e-01 1.30543366e-01
8.34733024e-02 3.90696734e-01 5.97485244e-01 7.80264378e-01
-4.35986519e-01 9.50139999e-01 4.85417396e-01 -5.39797962e-01
-1.20079482e+00 -1.08911872e+00 -1.06657133e-01 1.86666501e+00
3.72261375e-01 -1.07631147e+00 -3.96563768e-01 -6.72577202e-01
-2.94446647e-01 2.26222426e-01 -6.95653617e-01 2.19127551e-01
-3.81985456e-01 -1.72700182e-01 5.27651012e-01 4.55235481e-01
4.69265245e-02 -9.86915827e-01 -4.36371833e-01 -1.31067783e-02
-5.39727032e-01 -2.04953432e+00 -6.48844063e-01 -2.67334372e-01
-9.22190249e-02 -1.23663962e+00 -2.23929510e-01 -7.56889641e-01
2.07871288e-01 5.18565834e-01 1.39868498e+00 -5.35704374e-01
-4.74008143e-01 6.35769546e-01 -8.15617085e-01 1.90605164e-01
-2.93339521e-01 -2.67905235e-01 1.76780954e-01 3.85043919e-01
1.21343210e-01 -4.78936344e-01 -4.24188137e-01 2.60551691e-01
-7.08027661e-01 4.29196894e-01 -3.27314645e-01 4.99129772e-01
4.59477156e-01 -1.92543209e-01 2.00901851e-01 -1.00285399e+00
1.63151845e-01 -8.25328767e-01 -4.66470756e-02 2.58057564e-01
2.74708927e-01 -8.31893921e-01 9.94286358e-01 -8.46417606e-01
-1.05541933e+00 1.16536811e-01 3.20366770e-01 -1.12912261e+00
-5.15531600e-01 1.44782007e-01 -1.49945552e-02 3.27072263e-01
8.31320286e-01 -7.76229426e-02 -9.29464459e-01 -6.05401754e-01
7.55997717e-01 6.37562573e-01 1.11570036e+00 -4.34311688e-01
4.83615994e-01 7.11692393e-01 -3.14927459e-01 -8.03669453e-01
-1.26491570e+00 -8.95141482e-01 -5.68482995e-01 -1.00034785e+00
1.24241281e+00 -1.37319684e+00 -7.26400495e-01 1.03529669e-01
-1.04218185e+00 -3.76244664e-01 -1.03199050e-01 3.01674098e-01
-7.09271669e-01 4.78842348e-01 -1.10914564e+00 -4.81288940e-01
1.25704259e-01 -8.60714138e-01 1.00432777e+00 -1.10989586e-01
-5.18831789e-01 -7.59843647e-01 6.01230785e-02 7.54772246e-01
-1.02548406e-01 7.78528631e-01 4.57487017e-01 -3.85182887e-01
-3.09299707e-01 -2.35581651e-01 -7.74940774e-02 9.13456976e-02
-3.73284072e-02 2.02872261e-01 -1.13572896e+00 -2.37853155e-01
-1.80748180e-01 -1.08304572e+00 8.34569991e-01 1.51090324e-01
1.42648995e+00 -5.00522017e-01 -1.76137298e-01 8.39936614e-01
1.42687893e+00 -2.02318523e-02 3.88520285e-02 3.13093394e-01
1.13645554e+00 7.45956123e-01 7.78134704e-01 7.95108020e-01
4.82435435e-01 9.15041208e-01 4.29002732e-01 5.07857315e-02
-2.52683610e-01 -6.30874634e-01 5.18962264e-01 7.94416487e-01
-1.35088906e-01 -4.59006935e-01 -7.31392980e-01 6.75003886e-01
-1.85394013e+00 -1.45782614e+00 -9.27348509e-02 1.52886069e+00
8.95568013e-01 1.99351251e-01 5.14120102e-01 -1.72220349e-01
7.87225306e-01 5.89712083e-01 -2.35503316e-01 -3.07367116e-01
-3.60293180e-01 -4.48587447e-01 2.72529483e-01 7.29153454e-02
-1.62948346e+00 1.01024830e+00 6.37890387e+00 8.02907646e-01
-8.65219414e-01 5.93948483e-01 5.10667443e-01 -9.61712003e-01
-5.01792766e-02 -3.57339494e-02 -6.06840074e-01 5.19541204e-01
7.76562989e-01 2.09203333e-01 5.51630974e-01 1.16371095e+00
2.06146777e-01 2.28596166e-01 -1.38546097e+00 1.42976248e+00
6.97411358e-01 -1.80104995e+00 1.05272129e-01 -3.72208297e-01
9.54148352e-01 -8.28649253e-02 -1.80427700e-01 3.82806659e-01
3.24312657e-01 -9.93946195e-01 1.48392141e+00 4.12344307e-01
1.30765545e+00 -2.97989368e-01 4.66712452e-02 9.79075581e-02
-1.29110360e+00 -3.35213661e-01 -1.89708859e-01 -3.03332031e-01
5.55505514e-01 3.91802564e-02 -4.18983877e-01 1.63919374e-01
1.20239401e+00 1.56790721e+00 -8.23786378e-01 7.78607249e-01
6.64029941e-02 6.02630913e-01 1.59611300e-01 8.23840052e-02
1.23606764e-01 1.66240141e-01 5.07564068e-01 1.74752271e+00
-1.50135368e-01 -3.97529341e-02 4.35354352e-01 4.55144912e-01
-2.62183130e-01 2.25053519e-01 -7.53274143e-01 -2.91994601e-01
4.90598887e-01 1.65811670e+00 -8.89550090e-01 -5.12074292e-01
-6.70675874e-01 1.17551291e+00 4.97976065e-01 4.23634410e-01
-1.19274461e+00 1.37862429e-01 6.31624818e-01 2.34076813e-01
3.43751580e-01 -1.18820384e-01 1.43425703e-01 -1.65704632e+00
-6.50913045e-02 -1.00520861e+00 5.99604905e-01 -1.05863559e+00
-1.44204485e+00 7.95472205e-01 2.09069327e-01 -1.50370347e+00
-1.53981224e-01 -5.99815965e-01 -3.54030818e-01 -1.26448125e-01
-6.35894477e-01 -1.35207641e+00 -5.28085709e-01 9.55790222e-01
1.12070632e+00 -2.03161255e-01 6.17757261e-01 5.85592628e-01
-5.77830434e-01 5.86039543e-01 5.56159504e-02 3.02441716e-01
1.03666127e+00 -9.83693302e-01 9.95274484e-02 1.02035689e+00
5.83480775e-01 7.95698613e-02 8.51893902e-01 -2.95369416e-01
-1.38031888e+00 -1.22862113e+00 4.95182723e-01 -1.11714518e+00
8.84841919e-01 -1.06027102e+00 -3.43091220e-01 8.34405482e-01
2.12490529e-01 7.59439394e-02 9.85104024e-01 2.52899289e-01
-9.04652238e-01 9.57672074e-02 -5.90973377e-01 6.04562223e-01
1.63828850e+00 -1.00968051e+00 -4.99027282e-01 7.10958600e-01
7.13641584e-01 -4.15010691e-01 -9.74139392e-01 2.25179911e-01
6.14390016e-01 -6.40755355e-01 1.11113870e+00 -1.28446245e+00
1.25415754e+00 -1.59756809e-01 -5.91967165e-01 -7.71310806e-01
-5.39548099e-01 -3.96494865e-01 -2.52448380e-01 1.68924975e+00
6.34990036e-02 6.20601177e-01 4.44270611e-01 3.85730743e-01
-2.63976812e-01 -2.46713698e-01 -6.81921721e-01 -5.82191050e-01
-4.34212089e-01 -8.06239605e-01 -8.54543410e-04 1.41746879e+00
4.19653535e-01 6.70430362e-01 -1.03881383e+00 -1.59508094e-01
3.23291659e-01 4.08789128e-01 1.04440653e+00 -7.49859273e-01
-5.61552525e-01 -2.62248605e-01 -4.97722596e-01 -9.22317863e-01
1.87380597e-01 -9.42020953e-01 9.76616237e-03 -1.29588675e+00
7.93380082e-01 -5.15003130e-02 -2.06780449e-01 2.24822924e-01
7.69522116e-02 8.30355585e-01 4.39501673e-01 4.05351430e-01
-1.68170249e+00 1.97937340e-01 1.00898147e+00 -1.38862103e-01
2.04131886e-01 -5.70775211e-01 -3.81963074e-01 8.84172201e-01
2.45599926e-01 -3.50896448e-01 -3.73828918e-01 -5.36521077e-01
4.50830251e-01 3.11813712e-01 3.73592198e-01 -9.60432172e-01
-6.21278258e-03 -5.28647184e-01 4.46599215e-01 -3.02038312e-01
7.44510889e-01 -4.38631266e-01 2.22766310e-01 -4.18682486e-01
-8.07645261e-01 7.93312117e-02 -4.04882394e-02 8.15559685e-01
-3.18839461e-01 1.09911688e-01 5.91973841e-01 -3.06124538e-01
-1.17712080e+00 3.82256716e-01 -6.55350924e-01 4.58717555e-01
1.23638332e+00 -1.53179467e-01 -6.07046843e-01 -6.20898366e-01
-9.29708183e-01 6.79034218e-02 8.91210496e-01 9.98526514e-01
5.16055703e-01 -1.38714743e+00 -5.42289257e-01 -4.57949638e-01
7.33064055e-01 -5.29465556e-01 7.59418905e-01 2.92398065e-01
-5.88141024e-01 1.45693347e-01 -3.81927669e-01 -6.28960431e-01
-1.37245536e+00 1.19206417e+00 -8.30574930e-02 1.80953413e-01
-7.34156668e-01 1.05489790e+00 5.45603871e-01 9.65748280e-02
3.65405083e-01 -1.59292385e-01 -6.23681068e-01 3.63736451e-01
7.19615400e-01 5.29003814e-02 -3.58112305e-01 -1.21230483e+00
-3.92911881e-01 3.73589844e-01 1.25850379e-01 -1.05723441e-01
1.21967006e+00 -4.64396089e-01 2.09101737e-01 9.21751499e-01
1.41163397e+00 1.64122269e-01 -1.54572189e+00 -1.95646763e-01
-1.87518001e-01 -7.69223571e-01 -3.25820208e-01 -4.61099565e-01
-8.58319342e-01 7.51639009e-01 1.75862461e-01 2.69399613e-01
1.15471184e+00 5.24609268e-01 6.52220964e-01 1.02105170e-01
3.54849130e-01 -1.27396798e+00 6.09491527e-01 1.90646216e-01
8.77964258e-01 -1.29884541e+00 -4.09062617e-02 -4.26279366e-01
-1.24078751e+00 8.08638096e-01 6.75126851e-01 -9.19007957e-02
4.09369648e-01 -1.95813626e-02 4.15232219e-02 -3.23110849e-01
-1.15845859e+00 -1.21045984e-01 3.93758267e-01 3.02021891e-01
2.06680730e-01 9.12000090e-02 2.54273638e-02 1.05866218e+00
-2.19641894e-01 -2.92092294e-01 6.77924633e-01 7.43231714e-01
-2.09178925e-01 -5.34148157e-01 -1.21661998e-01 3.21028233e-01
-8.83720517e-01 -6.82822317e-02 -4.96968836e-01 3.03679883e-01
3.53680044e-01 1.20024490e+00 3.13063979e-01 -7.64452934e-01
1.07441522e-01 -2.12052494e-01 5.01498699e-01 -7.81831801e-01
-6.73082411e-01 -9.30040479e-02 6.79516971e-01 -8.78075361e-01
-7.41426468e-01 -8.78882289e-01 -4.60642129e-01 -9.11711603e-02
3.39793473e-01 -1.69241309e-01 3.87556940e-01 8.07475150e-01
1.16559021e-01 3.44988912e-01 7.99519897e-01 -9.09337580e-01
2.61031419e-01 -8.83779228e-01 -5.23765385e-01 1.33000255e+00
2.20987603e-01 -7.17636824e-01 -2.53177673e-01 9.10278082e-01] | [10.303550720214844, 0.7997976541519165] |
bb953425-7179-44f2-b568-f1b99df719c1 | aerial-scene-understanding-in-the-wild-multi | 2104.11200 | null | https://arxiv.org/abs/2104.11200v1 | https://arxiv.org/pdf/2104.11200v1.pdf | Aerial Scene Understanding in The Wild: Multi-Scene Recognition via Prototype-based Memory Networks | Aerial scene recognition is a fundamental visual task and has attracted an increasing research interest in the last few years. Most of current researches mainly deploy efforts to categorize an aerial image into one scene-level label, while in real-world scenarios, there often exist multiple scenes in a single image. Therefore, in this paper, we propose to take a step forward to a more practical and challenging task, namely multi-scene recognition in single images. Moreover, we note that manually yielding annotations for such a task is extraordinarily time- and labor-consuming. To address this, we propose a prototype-based memory network to recognize multiple scenes in a single image by leveraging massive well-annotated single-scene images. The proposed network consists of three key components: 1) a prototype learning module, 2) a prototype-inhabiting external memory, and 3) a multi-head attention-based memory retrieval module. To be more specific, we first learn the prototype representation of each aerial scene from single-scene aerial image datasets and store it in an external memory. Afterwards, a multi-head attention-based memory retrieval module is devised to retrieve scene prototypes relevant to query multi-scene images for final predictions. Notably, only a limited number of annotated multi-scene images are needed in the training phase. To facilitate the progress of aerial scene recognition, we produce a new multi-scene aerial image (MAI) dataset. Experimental results on variant dataset configurations demonstrate the effectiveness of our network. Our dataset and codes are publicly available. | ['Xiao Xiang Zhu', 'Konrad Heidler', 'Jianzhe Lin', 'Lichao Moua', 'Yuansheng Hua'] | 2021-04-22 | null | null | null | null | ['scene-recognition'] | ['computer-vision'] | [ 7.26794124e-01 -5.59761584e-01 -2.22336501e-02 -4.98902410e-01
-6.03569925e-01 -5.71455359e-01 2.90206075e-01 1.97481781e-01
-3.70083183e-01 2.52451152e-01 -4.25501436e-01 -1.19879708e-01
-7.84482211e-02 -8.42086732e-01 -7.98526227e-01 -5.52240610e-01
3.03602636e-01 8.97333845e-02 3.70410174e-01 2.01509014e-01
1.37047842e-01 5.23554921e-01 -1.74544525e+00 2.54597068e-01
7.70090282e-01 1.36823285e+00 8.70380521e-01 4.61037904e-01
1.57079488e-01 8.02280366e-01 -5.17963052e-01 -1.77551374e-01
2.38046974e-01 -2.42567286e-01 -8.32938552e-01 7.12511063e-01
6.41362965e-01 -4.53530967e-01 -2.00362518e-01 1.02899027e+00
3.41845542e-01 4.10263270e-01 3.24135691e-01 -1.05830717e+00
-7.01157868e-01 2.41717935e-01 -4.61281747e-01 1.83159724e-01
2.79425085e-02 3.90666753e-01 9.75981474e-01 -9.68532503e-01
2.13125646e-01 7.59994984e-01 2.62078613e-01 6.70306161e-02
-8.84605765e-01 -4.97810841e-01 6.37893438e-01 2.70045012e-01
-1.90084231e+00 -3.82208437e-01 8.48017097e-01 -4.81100470e-01
7.75083661e-01 3.94803286e-01 7.05379844e-01 6.64471149e-01
-9.95892063e-02 8.80217195e-01 7.65009284e-01 -1.09711699e-01
1.29926845e-01 1.05605856e-01 4.92665619e-02 7.50868917e-01
7.48992264e-02 -3.44379932e-01 -1.69373024e-03 1.04346611e-01
7.52060175e-01 5.90299070e-01 -4.62626398e-01 -2.97846645e-01
-1.03830838e+00 6.09789908e-01 9.13686752e-01 3.36615264e-01
-5.55612683e-01 -1.07315980e-01 2.91168451e-01 1.42237142e-01
3.23827147e-01 3.51020426e-01 -2.09562197e-01 5.88605762e-01
-1.10017812e+00 -2.80159749e-02 2.27415547e-01 1.12769306e+00
8.57858658e-01 4.41597804e-04 -7.85273165e-02 9.72140133e-01
1.50988519e-01 5.63530803e-01 6.76295161e-01 -3.07549477e-01
4.37768996e-01 9.42948639e-01 1.15912966e-01 -1.33161867e+00
-3.68965089e-01 -4.92861271e-01 -1.21815062e+00 -4.23026174e-01
-2.47720942e-01 2.19158903e-01 -9.29221392e-01 1.36185920e+00
3.01788598e-01 3.18560034e-01 -2.67387722e-02 1.39871907e+00
8.57709169e-01 8.71691167e-01 2.65109956e-01 -1.31059900e-01
1.05050397e+00 -1.36675525e+00 -1.52594462e-01 -4.92682397e-01
3.87595892e-01 -7.20273077e-01 1.09453535e+00 2.90677696e-02
-9.32677209e-01 -1.04234827e+00 -1.05586779e+00 1.02863774e-01
-5.22117674e-01 8.12793672e-01 4.79363561e-01 1.85153872e-01
-7.64422297e-01 2.43139908e-01 -4.61642832e-01 -4.64377344e-01
3.42099518e-01 3.97164434e-01 -3.78383040e-01 -3.60336959e-01
-7.52656043e-01 4.77823824e-01 7.09429324e-01 5.17450929e-01
-1.15798497e+00 -2.16936782e-01 -9.35579598e-01 1.04877487e-01
5.20031929e-01 -7.40445852e-01 1.02299786e+00 -1.17333245e+00
-1.11570346e+00 8.55159521e-01 -2.15506628e-02 -3.41733769e-02
2.40328684e-02 -8.10003579e-02 -5.43348670e-01 1.52130634e-01
9.21756998e-02 5.94954967e-01 1.10105455e+00 -1.30904996e+00
-7.86333442e-01 -4.00654465e-01 3.86718005e-01 3.02931309e-01
-4.47100013e-01 -1.42430067e-01 -6.07147038e-01 -5.40284395e-01
1.02940463e-01 -8.47597599e-01 -2.77098536e-01 -1.48547009e-01
-5.21794438e-01 -7.68324956e-02 7.02064157e-01 -3.32772970e-01
1.28511226e+00 -2.17937064e+00 2.83504963e-01 -8.37128982e-02
-1.27246268e-02 5.62268138e-01 -4.39179957e-01 1.46624655e-01
-1.32111549e-01 -3.95251513e-02 -4.47320402e-01 -2.73704112e-01
-3.34662825e-01 -3.88628989e-02 -7.82873094e-01 3.27344477e-01
4.82562393e-01 9.19646800e-01 -8.26286972e-01 -4.99023527e-01
3.30204964e-01 1.52729720e-01 -3.35587025e-01 6.26113057e-01
-2.17710987e-01 3.02106678e-01 -5.79028428e-01 1.06093514e+00
6.27141416e-01 -5.96217632e-01 -1.14174569e-02 -4.69404548e-01
-3.19151431e-01 -3.92850786e-01 -1.07900083e+00 1.71511590e+00
-5.33053219e-01 1.93530381e-01 -1.04868054e-01 -1.07585835e+00
9.08521831e-01 1.18621103e-02 8.95830616e-02 -4.49741632e-01
2.01127470e-01 1.48587972e-01 -2.39750221e-01 -5.87540507e-01
9.38190281e-01 -1.95422135e-02 -8.64744782e-02 1.83642864e-01
1.56799987e-01 -2.36396864e-01 1.82493001e-01 -1.51186585e-01
8.49497914e-01 -7.78579116e-02 3.82228255e-01 1.13229670e-01
8.00069690e-01 2.67465800e-01 3.88820797e-01 5.90009570e-01
-1.89726219e-01 5.90872526e-01 -3.11055988e-01 -6.82380140e-01
-7.98514962e-01 -6.42446339e-01 -4.41197120e-02 9.76969063e-01
5.98928571e-01 -3.93674552e-01 -3.63187551e-01 -5.54796934e-01
-1.79577991e-01 2.40457818e-01 -3.67009163e-01 -1.97781697e-01
-2.94212759e-01 -7.13311732e-01 4.43183124e-01 5.49900532e-01
8.54135931e-01 -1.25994706e+00 -1.00265348e+00 4.41949889e-02
-1.44447222e-01 -1.12112379e+00 -5.32157123e-01 7.18353242e-02
-7.31007218e-01 -1.12215006e+00 -9.07674074e-01 -9.58903372e-01
9.21715140e-01 9.04719949e-01 8.18205416e-01 2.24312752e-01
-5.87979198e-01 4.45189446e-01 -3.85666221e-01 -9.45334509e-02
2.22489715e-01 2.49882400e-01 -2.45731827e-02 4.89142954e-01
1.13569178e-01 -3.99375677e-01 -5.53676784e-01 4.41917300e-01
-1.26807570e+00 2.70170659e-01 1.03222799e+00 8.98773968e-01
9.17893291e-01 2.35391170e-01 3.01719725e-01 -5.10964274e-01
3.28076690e-01 -5.56100488e-01 -8.63617718e-01 6.32035077e-01
-1.70387235e-02 -3.80859226e-01 1.02026868e+00 -5.58930218e-01
-6.97755694e-01 6.01013362e-01 6.75959140e-02 -8.19074988e-01
-5.70842326e-01 8.51830423e-01 -3.12852740e-01 -1.70461580e-01
2.92937249e-01 7.21633852e-01 -6.75252795e-01 -2.28336424e-01
2.76088506e-01 8.85652602e-01 5.58411598e-01 -2.89512277e-01
8.42441022e-01 1.30767450e-01 -1.99611783e-01 -1.03496444e+00
-1.22600615e+00 -7.23494709e-01 -9.25171673e-01 -3.48694444e-01
8.04105341e-01 -1.26138818e+00 -5.95559239e-01 5.07265210e-01
-1.09593284e+00 -5.11544108e-01 -1.71305481e-02 2.09930718e-01
-3.01903307e-01 2.44197831e-01 -2.14420974e-01 -7.50680506e-01
-4.00291413e-01 -1.16172493e+00 1.37266123e+00 4.31662500e-01
2.96148449e-01 -5.54082155e-01 -1.03585802e-01 3.43547553e-01
3.47867668e-01 -1.14420168e-02 7.20709443e-01 -3.18431884e-01
-1.03955233e+00 -3.15415710e-01 -5.69552124e-01 2.91776776e-01
1.04434744e-01 -1.78576961e-01 -1.02338362e+00 -4.06983137e-01
-1.87883571e-01 -7.38226533e-01 9.51512039e-01 3.68539281e-02
1.38842988e+00 -1.37089282e-01 -3.29546332e-01 6.35349810e-01
1.63368869e+00 3.32511336e-01 2.23624662e-01 5.80457374e-02
9.30907786e-01 3.93720597e-01 9.02621984e-01 5.13733208e-01
4.16116267e-01 6.08413339e-01 5.17356277e-01 -1.53798103e-01
2.03308582e-01 -2.94704735e-01 8.38172585e-02 9.95543540e-01
5.17086461e-02 -5.81727982e-01 -8.20358813e-01 6.13089025e-01
-1.78228283e+00 -8.25151384e-01 1.73567161e-01 2.09802294e+00
4.30422395e-01 -2.10497379e-01 -1.83004618e-01 1.33070322e-02
6.51289642e-01 4.76617783e-01 -8.59799862e-01 3.12537313e-01
-1.28661513e-01 -2.00172633e-01 2.13582560e-01 -5.73162474e-02
-1.58642375e+00 1.13837290e+00 5.15269756e+00 6.88898385e-01
-1.49848306e+00 -8.77800211e-02 6.26638055e-01 1.53164431e-01
2.20547453e-01 -3.84939909e-02 -8.12451005e-01 3.93323570e-01
6.22733593e-01 -3.74607667e-02 3.66085380e-01 1.17415869e+00
-1.55338153e-01 -8.62054005e-02 -8.39747727e-01 1.33017349e+00
4.83163506e-01 -1.05413377e+00 5.56809068e-01 -2.51229793e-01
5.43890834e-01 3.14464569e-02 4.34848964e-02 2.95121729e-01
-1.30163416e-01 -8.64783645e-01 7.25794673e-01 6.27682328e-01
8.92317772e-01 -5.58349490e-01 6.46171451e-01 5.74149489e-01
-1.81773257e+00 -3.72882694e-01 -8.04689705e-01 -1.08931037e-02
7.90857226e-02 3.97875309e-01 -3.71547669e-01 8.48086655e-01
6.30444944e-01 9.71238613e-01 -9.25961792e-01 1.41972208e+00
-2.44123302e-02 3.14339727e-01 -4.56899107e-02 1.17150322e-01
4.70757246e-01 -1.21564396e-01 1.14764012e-01 9.22472835e-01
4.49987590e-01 4.64251190e-01 7.27665424e-01 7.92854607e-01
-2.95812607e-01 2.19358534e-01 -8.56099784e-01 -2.28443652e-01
3.73182178e-01 1.63506866e+00 -8.05157363e-01 -4.68246967e-01
-3.85872751e-01 1.38404393e+00 4.57389206e-01 3.16536933e-01
-7.94530511e-01 -4.06306416e-01 2.97471374e-01 -2.80873656e-01
3.92800808e-01 -2.07266226e-01 3.06434631e-01 -1.36342704e+00
1.24930143e-01 -7.06687212e-01 3.79694760e-01 -1.11851299e+00
-1.17063403e+00 1.05439436e+00 -1.80506572e-01 -1.44961357e+00
1.34633765e-01 -5.87660074e-01 -5.76289833e-01 8.00076008e-01
-1.72863531e+00 -1.53354692e+00 -1.00356126e+00 6.79875851e-01
8.46958816e-01 -9.54661444e-02 9.53686833e-01 4.40071523e-01
-1.03102994e+00 2.95969665e-01 -3.10360134e-01 1.95337370e-01
5.77361941e-01 -7.04530180e-01 2.45473951e-01 1.01848376e+00
2.93710023e-01 5.75359523e-01 2.10853424e-02 -4.73093539e-01
-1.47238207e+00 -1.74514103e+00 6.25171602e-01 -1.23545803e-01
5.58644235e-01 -3.50157805e-02 -8.31594288e-01 7.56270051e-01
-8.40989873e-02 2.62827069e-01 7.65604198e-01 -4.01235074e-01
-1.12246946e-01 -1.83965847e-01 -6.15040898e-01 4.25266236e-01
1.01204419e+00 -7.03644693e-01 -3.69697660e-01 4.91382658e-01
6.74749792e-01 -3.20069194e-01 -7.97937214e-01 6.18572950e-01
2.69436330e-01 -6.95953786e-01 9.14175212e-01 -3.33749682e-01
3.41309369e-01 -7.09842443e-01 -5.47566891e-01 -1.18695748e+00
-4.27523494e-01 -3.12551968e-02 2.56126791e-01 1.13693523e+00
6.91386461e-02 -2.69339830e-01 3.48005623e-01 4.63391423e-01
-4.51931328e-01 -7.70273805e-01 -4.52346206e-01 -6.72783017e-01
-4.14198637e-01 -2.05797195e-01 6.06434524e-01 8.24895322e-01
-3.85172635e-01 5.47494173e-01 -5.75087428e-01 5.37962973e-01
3.47758651e-01 1.01221156e+00 9.11054850e-01 -1.19495738e+00
-2.56673545e-01 -1.84629634e-01 -4.60897833e-01 -1.52935970e+00
2.23915622e-01 -8.80243778e-01 2.29752377e-01 -1.41235983e+00
4.65409666e-01 -5.87108791e-01 -3.72721225e-01 5.67968428e-01
-2.20642313e-01 4.00193870e-01 3.99805129e-01 4.24032897e-01
-1.19233406e+00 8.27741086e-01 9.84957695e-01 -6.00024521e-01
-1.52571827e-01 -2.52247602e-02 -4.91596282e-01 6.24882877e-01
6.97469532e-01 -1.76513582e-01 -5.22688925e-01 -6.79430425e-01
-2.77247638e-01 4.08824757e-02 6.57132447e-01 -1.29444885e+00
5.83001196e-01 -2.70244688e-01 5.18335044e-01 -7.41343141e-01
5.63596010e-01 -9.56607759e-01 7.72918910e-02 2.00660795e-01
-2.00690299e-01 1.56918585e-01 2.89793432e-01 5.21818101e-01
-5.68688810e-01 -3.20137292e-01 6.01022542e-01 -3.37354541e-01
-1.26239824e+00 6.92601681e-01 -7.57988319e-02 -3.63814175e-01
1.13828659e+00 4.04500179e-02 -3.34664524e-01 1.00538939e-01
-4.44287181e-01 2.86155552e-01 4.76555258e-01 4.88282830e-01
9.53170300e-01 -1.16507328e+00 -3.58166724e-01 4.16813642e-01
5.01632035e-01 2.88861066e-01 7.87422299e-01 6.37953579e-01
-2.69234717e-01 6.89390004e-01 -1.87449291e-01 -6.28125191e-01
-1.33969724e+00 9.08847272e-01 1.98057994e-01 4.08587903e-02
-3.42595458e-01 7.85321653e-01 3.84014726e-01 -2.78250068e-01
1.51763588e-01 -3.62211198e-01 -2.48554990e-01 6.18606433e-03
6.29105628e-01 -1.16901383e-01 -1.06897451e-01 -1.15500426e+00
-2.47457072e-01 9.25175667e-01 1.29181027e-01 4.91803348e-01
1.18564630e+00 -1.97132975e-01 -1.29237741e-01 6.21386707e-01
1.09778106e+00 -5.79102933e-01 -1.06265831e+00 -3.08165669e-01
-1.11127548e-01 -4.93691206e-01 4.95213233e-02 -5.74223340e-01
-1.11777699e+00 9.61242080e-01 5.84304333e-01 1.17293566e-01
1.59181798e+00 -1.61334530e-01 8.17516446e-01 8.23151052e-01
5.50013661e-01 -6.46518350e-01 9.98838320e-02 3.22263509e-01
9.33485210e-01 -1.44410503e+00 -4.26529441e-03 -3.13622683e-01
-5.08248925e-01 1.01609898e+00 8.63290668e-01 1.61223765e-03
5.04605353e-01 -3.13923329e-01 4.48546037e-02 -1.62312716e-01
-6.18014932e-01 -5.17137945e-01 6.00869179e-01 2.36253202e-01
4.53758724e-02 1.06567942e-01 4.93233651e-01 7.51320541e-01
2.15897396e-01 -3.19013707e-02 1.65136009e-01 8.82702172e-01
-6.02971256e-01 -8.39477718e-01 -1.61806390e-01 3.96318227e-01
4.34274562e-02 -1.78587258e-01 -4.60520089e-01 5.34103036e-01
1.84154227e-01 9.90180910e-01 -2.29689702e-02 -7.53986716e-01
5.00533044e-01 -4.29130383e-02 7.68628046e-02 -6.60560310e-01
-4.84972090e-01 -1.56349421e-01 -4.80483204e-01 -5.58923483e-01
-6.61526382e-01 -1.96904749e-01 -7.66770244e-01 6.84981421e-02
-5.57424366e-01 -5.19587658e-02 3.93248230e-01 8.96144986e-01
4.38180238e-01 5.23189604e-01 5.84415793e-01 -1.18263662e+00
-2.14602590e-01 -8.09635878e-01 -5.09040952e-01 2.00649425e-01
2.02763170e-01 -5.86054027e-01 -8.43019933e-02 3.35297018e-01] | [9.784725189208984, 1.7184507846832275] |
6076925f-81a2-4421-814f-f603d13a386b | le-hgr-a-lightweight-and-efficient-rgb-based | 2001.05654 | null | https://arxiv.org/abs/2001.05654v1 | https://arxiv.org/pdf/2001.05654v1.pdf | LE-HGR: A Lightweight and Efficient RGB-based Online Gesture Recognition Network for Embedded AR Devices | Online hand gesture recognition (HGR) techniques are essential in augmented reality (AR) applications for enabling natural human-to-computer interaction and communication. In recent years, the consumer market for low-cost AR devices has been rapidly growing, while the technology maturity in this domain is still limited. Those devices are typical of low prices, limited memory, and resource-constrained computational units, which makes online HGR a challenging problem. To tackle this problem, we propose a lightweight and computationally efficient HGR framework, namely LE-HGR, to enable real-time gesture recognition on embedded devices with low computing power. We also show that the proposed method is of high accuracy and robustness, which is able to reach high-end performance in a variety of complicated interaction environments. To achieve our goal, we first propose a cascaded multi-task convolutional neural network (CNN) to simultaneously predict probabilities of hand detection and regress hand keypoint locations online. We show that, with the proposed cascaded architecture design, false-positive estimates can be largely eliminated. Additionally, an associated mapping approach is introduced to track the hand trace via the predicted locations, which addresses the interference of multi-handedness. Subsequently, we propose a trace sequence neural network (TraceSeqNN) to recognize the hand gesture by exploiting the motion features of the tracked trace. Finally, we provide a variety of experimental results to show that the proposed framework is able to achieve state-of-the-art accuracy with significantly reduced computational cost, which are the key properties for enabling real-time applications in low-cost commercial devices such as mobile devices and AR/VR headsets. | ['Jiafang Wang', 'Mingyang Li', 'Jian Gu', 'Hongwei Xie', 'Baitao Shao'] | 2020-01-16 | null | null | null | null | ['hand-detection', '3d-part-segmentation'] | ['computer-vision', 'computer-vision'] | [ 2.06854612e-01 -6.01207376e-01 -2.96162516e-01 3.57032046e-02
-7.49019444e-01 -3.28063667e-01 2.05611467e-01 -5.14494002e-01
-7.02777028e-01 3.56411934e-01 -2.23409042e-01 -3.80511791e-01
-2.06792757e-01 -6.61572516e-01 -7.03176081e-01 -3.81489694e-01
2.62663011e-02 1.05797797e-01 4.93819207e-01 -6.21317551e-02
1.67170018e-01 9.49275434e-01 -1.60211480e+00 -1.13647720e-02
3.05696160e-01 1.18722272e+00 5.02824903e-01 8.38523924e-01
2.23973662e-01 4.57535207e-01 -3.78901571e-01 -2.07907543e-01
1.81437999e-01 -1.18494242e-01 -2.53012359e-01 -3.70675296e-01
1.53467298e-01 -9.02315378e-01 -8.46810997e-01 8.16972196e-01
9.74709451e-01 1.93905324e-01 1.97028011e-01 -8.76548707e-01
-4.01949197e-01 2.44408280e-01 -7.56728828e-01 -4.20309678e-02
4.98975307e-01 2.59793639e-01 9.40100491e-01 -1.11750317e+00
3.73357356e-01 9.79710817e-01 6.49072886e-01 7.66035378e-01
-4.86079216e-01 -7.79361069e-01 3.68754268e-01 2.93665707e-01
-1.72492957e+00 -2.53062844e-01 6.01967335e-01 -1.24392770e-01
1.06604540e+00 5.02050042e-01 6.13667130e-01 1.30442607e+00
3.88532169e-02 1.13608527e+00 5.78940630e-01 -3.00368339e-01
4.83837873e-02 -3.61904442e-01 -6.15565106e-02 8.26745033e-01
6.81256875e-02 1.41137421e-01 -6.27375722e-01 1.60134420e-01
1.33037829e+00 7.20225096e-01 -4.20123786e-01 -1.22719500e-02
-1.31438744e+00 3.31609100e-01 5.96153200e-01 4.54331636e-01
-5.52103221e-01 3.36350411e-01 2.55203247e-01 -1.36070430e-01
2.41342606e-03 -3.34558994e-01 -3.87855917e-01 -6.03347719e-01
-5.74165881e-01 9.98541415e-02 5.29917002e-01 1.13119161e+00
5.76706342e-02 -1.18694834e-01 -1.59675568e-01 7.12306082e-01
5.70591986e-01 7.83755958e-01 4.30819094e-01 -3.64717335e-01
7.57342160e-01 4.91045505e-01 3.91699344e-01 -8.30617726e-01
-5.54702103e-01 -1.37552291e-01 -9.77215111e-01 2.74811149e-01
3.71520489e-01 7.17162937e-02 -1.00808597e+00 1.41389644e+00
1.89274266e-01 3.24148506e-01 -2.51199067e-01 1.31489170e+00
6.69554949e-01 4.09260333e-01 1.77356333e-03 -1.89275201e-02
1.55893922e+00 -7.60733724e-01 -8.26214433e-01 -1.01738706e-01
4.23850000e-01 -4.48323280e-01 1.39202642e+00 4.39532578e-01
-7.73789644e-01 -5.70978284e-01 -1.08783269e+00 -1.15326308e-01
-8.81684292e-03 7.32813895e-01 7.54616380e-01 9.14589226e-01
-6.29745364e-01 5.43721497e-01 -1.37383521e+00 -1.81181625e-01
3.64605159e-01 6.17558122e-01 -1.43037140e-01 -2.32547298e-02
-7.99757719e-01 5.63636422e-01 3.71183306e-02 8.68081748e-01
-4.55487430e-01 -3.38511288e-01 -6.63526952e-01 1.05187431e-01
3.86413634e-01 -2.99522072e-01 1.10504377e+00 -4.51951683e-01
-1.96826577e+00 3.55169624e-01 -3.87166679e-01 -4.11708467e-02
9.23833132e-01 -7.57832646e-01 -3.84257585e-01 -1.07132524e-01
-3.43718648e-01 1.40461296e-01 6.82496250e-01 -8.99625301e-01
-9.22040164e-01 -6.79792047e-01 -6.19432330e-02 1.64357141e-01
-4.65710878e-01 2.08591282e-01 -8.43059361e-01 -5.77384651e-01
1.91023812e-01 -9.86068904e-01 -1.62881657e-01 3.22526097e-02
-3.33334893e-01 -2.14707389e-01 8.17951024e-01 -7.81911492e-01
1.15727496e+00 -2.24854708e+00 8.13071951e-02 3.77747774e-01
1.80730253e-01 4.52076584e-01 -7.80885741e-02 1.26531208e-02
3.61724019e-01 -2.18785122e-01 9.34518650e-02 -3.63339067e-01
-9.20063481e-02 -6.43944070e-02 -4.57415581e-01 3.72470826e-01
6.87836409e-02 1.17785311e+00 -7.83457994e-01 1.27561828e-02
7.08295047e-01 8.88663352e-01 -1.49990261e-01 3.38434488e-01
-7.04663172e-02 7.20589578e-01 -5.77701569e-01 8.72261167e-01
4.04158890e-01 -2.43502438e-01 2.44507208e-01 -1.35761559e-01
-1.60478070e-01 5.89640066e-02 -1.40725684e+00 1.70595634e+00
-7.68772960e-01 5.08467317e-01 -1.23938963e-01 -2.50476360e-01
9.05924678e-01 2.29804784e-01 3.43299210e-01 -8.99129391e-01
3.37413728e-01 4.81604367e-01 -6.46331087e-02 -3.55971217e-01
7.10947573e-01 4.37884897e-01 1.37887686e-01 3.58211815e-01
-3.67925018e-01 6.89865649e-01 -5.13203919e-01 -2.33684286e-01
1.36575282e+00 3.11548650e-01 2.11305797e-01 5.12287974e-01
3.88484687e-01 -3.38366985e-01 3.62834901e-01 8.35817456e-01
-3.57313603e-02 5.31168699e-01 -3.11347216e-01 -4.69778538e-01
-7.56767452e-01 -9.55164969e-01 2.51423359e-01 1.06831419e+00
4.77172852e-01 -5.35006560e-02 -3.44498247e-01 -4.02602315e-01
-1.63362175e-01 1.37853384e-01 -1.50369421e-01 2.12174475e-01
-1.14373136e+00 -3.60666484e-01 6.21112645e-01 1.18209243e+00
6.41984224e-01 -1.22206604e+00 -1.14579725e+00 2.48776987e-01
1.40236365e-02 -1.31335557e+00 -4.57758904e-01 -2.03261912e-01
-7.56695867e-01 -8.53595734e-01 -1.33615947e+00 -7.66744673e-01
3.46634001e-01 4.65528995e-01 3.68509889e-01 1.77917898e-01
-3.58061045e-01 4.96518463e-01 -6.03407145e-01 -1.55069947e-01
2.87574589e-01 2.98194796e-01 1.49621442e-01 1.43093958e-01
4.78933901e-01 -4.58613992e-01 -9.56112862e-01 6.38550937e-01
-5.51454365e-01 -1.59396172e-01 7.40721285e-01 8.12808216e-01
6.63865864e-01 -3.42466801e-01 4.94118989e-01 -3.70973378e-01
4.35961634e-01 -9.47254058e-03 -7.18144059e-01 3.89474034e-01
-3.26599330e-01 -2.90791959e-01 4.53933150e-01 -8.25504363e-01
-9.77746606e-01 4.73048270e-01 -6.26258790e-01 -6.75324678e-01
-8.88747424e-02 2.38110512e-01 -3.11111003e-01 -3.14659774e-01
1.71677426e-01 4.44988817e-01 -3.57354909e-01 -5.97034097e-01
4.84241277e-01 1.11975336e+00 7.11038053e-01 -3.01197708e-01
4.56992865e-01 5.86572349e-01 -6.87658042e-02 -8.69907796e-01
-1.82229251e-01 -6.28261864e-01 -6.00287259e-01 -4.20806050e-01
6.36984169e-01 -9.92941201e-01 -1.38916755e+00 7.55041718e-01
-1.28088856e+00 -4.59341139e-01 7.88070560e-02 7.45955050e-01
-5.23481786e-01 2.47096807e-01 -6.41390622e-01 -1.47739279e+00
-5.53666532e-01 -1.24465168e+00 1.44382691e+00 3.66957843e-01
-4.95051742e-02 -3.91587228e-01 -3.22644830e-01 -1.40470508e-02
2.28039458e-01 -6.37600049e-02 3.91819268e-01 -3.77064705e-01
-1.06020939e+00 -4.84227598e-01 -5.16480207e-01 -2.09348917e-01
2.56070465e-01 -3.56180370e-01 -9.66345429e-01 -4.40700173e-01
-3.37100089e-01 -9.87830907e-02 5.81848264e-01 3.03443551e-01
1.31235623e+00 1.55378833e-01 -6.82798564e-01 5.63326120e-01
1.19894099e+00 5.59764326e-01 9.05667007e-01 2.48815864e-01
1.20753622e+00 2.59928733e-01 8.73375356e-01 6.00533366e-01
3.65188062e-01 1.10979319e+00 2.33995914e-01 3.22323060e-03
-3.29754651e-02 -3.05976301e-01 2.39479199e-01 7.96999276e-01
-6.27075374e-01 -3.35920393e-01 -6.99142754e-01 4.38832790e-01
-2.09535146e+00 -6.13517284e-01 -9.80542004e-02 2.31602025e+00
2.44911477e-01 -1.22767858e-01 2.70639837e-01 2.18648598e-01
7.02181101e-01 -9.92296729e-04 -7.19045758e-01 1.85792014e-01
2.30244219e-01 6.33556604e-01 5.43084264e-01 2.05241486e-01
-1.00901997e+00 1.05404639e+00 5.13362837e+00 7.82278240e-01
-1.27831805e+00 2.70518363e-01 3.14097330e-02 -1.83508322e-01
2.08684519e-01 -6.87628567e-01 -9.59618747e-01 3.60543609e-01
7.80483007e-01 5.35689235e-01 4.90495861e-01 9.90371823e-01
2.17536494e-01 2.21789002e-01 -1.10816109e+00 1.54137504e+00
-2.71283574e-02 -1.15756035e+00 -2.44477570e-01 1.55053690e-01
6.32554367e-02 2.98192725e-02 1.45639420e-01 7.10838139e-02
-9.61600319e-02 -9.57811773e-01 8.39047134e-01 4.07321721e-01
1.34383857e+00 -8.04957926e-01 7.01596141e-01 3.40839714e-01
-1.75268042e+00 -2.55450070e-01 -7.24812225e-02 -2.63199266e-02
4.72318143e-01 1.26652747e-01 -4.61576879e-01 4.94736910e-01
8.34664166e-01 3.12805027e-01 -1.90548990e-02 9.37927067e-01
-3.09847891e-01 1.66657284e-01 -4.16700989e-01 -5.32809258e-01
-1.96037382e-01 1.15977988e-01 4.38737810e-01 1.00578320e+00
6.14422679e-01 4.93649751e-01 -1.13147637e-02 6.00537479e-01
-1.08051181e-01 -7.52006322e-02 -5.09328365e-01 4.54701781e-02
5.79699516e-01 1.01415622e+00 -9.62830544e-01 4.46145833e-02
-3.59978616e-01 1.55262029e+00 1.61407933e-01 1.83891639e-01
-7.76335180e-01 -7.50437021e-01 6.61700666e-01 -1.85203671e-01
3.98102641e-01 -7.12869406e-01 -3.54721576e-01 -1.15227950e+00
5.87731600e-01 -6.94342792e-01 1.31637439e-01 -4.30115610e-01
-9.87434626e-01 6.04367793e-01 -6.00950599e-01 -1.45313299e+00
-3.07814300e-01 -1.05159128e+00 -3.10919523e-01 8.62801492e-01
-1.40797651e+00 -1.25245619e+00 -6.27543449e-01 6.69341385e-01
6.83292985e-01 -1.21968843e-01 9.19932783e-01 6.52828813e-01
-6.17014706e-01 1.10456300e+00 7.48706004e-03 4.47156042e-01
3.18770230e-01 -7.83076644e-01 8.34231317e-01 8.74333262e-01
3.15729380e-01 9.11960602e-01 9.01877135e-03 -8.69510889e-01
-2.06158400e+00 -9.43540096e-01 2.41667524e-01 -4.77179646e-01
3.79342258e-01 -5.35544574e-01 -7.58869290e-01 6.46372557e-01
-7.37147629e-01 2.71583200e-01 4.68006402e-01 1.45272985e-01
-3.28082472e-01 6.93276301e-02 -9.79722261e-01 8.08816135e-01
1.73289645e+00 -7.02849090e-01 -1.99697211e-01 5.06917499e-02
4.81863946e-01 -8.56395483e-01 -6.81350946e-01 5.18956661e-01
1.40481508e+00 -5.81308365e-01 9.69861209e-01 -2.59356976e-01
-6.25533611e-02 -3.44617665e-01 -3.47359359e-01 -6.74667060e-01
-6.47776946e-02 -5.72105706e-01 -6.92352593e-01 9.71039116e-01
9.44415033e-02 -6.65619373e-01 1.08918881e+00 7.01252580e-01
1.34937719e-01 -6.84397399e-01 -1.25066912e+00 -8.67212474e-01
-6.29451931e-01 -8.07630122e-01 6.00241065e-01 3.14064682e-01
-1.51667297e-01 5.59596857e-03 -6.22846603e-01 4.99408782e-01
4.29990739e-01 5.80810532e-02 8.55468214e-01 -1.03474891e+00
-3.57158184e-01 -3.30209225e-01 -6.56130135e-01 -1.80383563e+00
-1.01418793e-01 -2.69133866e-01 2.58168459e-01 -1.33222175e+00
-8.13590735e-02 -6.82139874e-01 -2.94977784e-01 3.45707387e-01
-1.20628402e-01 4.61883694e-01 2.62881875e-01 4.57999706e-01
-5.01174629e-01 5.50795972e-01 9.18753505e-01 -4.28176345e-03
-6.90626800e-01 3.22654545e-01 -1.44863352e-01 6.41680837e-01
3.78433645e-01 -5.16098775e-02 -1.32750943e-01 -5.20631433e-01
9.38125849e-02 1.43967673e-01 5.28009713e-01 -9.87542152e-01
3.43654513e-01 1.11852355e-01 4.38586116e-01 -7.70731032e-01
5.31425297e-01 -1.07031989e+00 4.92092483e-02 5.35073459e-01
-1.20644516e-03 3.95793281e-02 8.18414465e-02 6.84066296e-01
1.84292808e-01 2.49452159e-01 3.13207716e-01 2.63809264e-01
-8.92092168e-01 4.34589088e-01 -1.00328572e-01 -7.23836660e-01
9.55339730e-01 -3.90704900e-01 -6.69198558e-02 -2.07860172e-01
-4.28237140e-01 1.40885508e-03 -2.39419937e-03 7.64411330e-01
1.15823102e+00 -1.37639427e+00 -2.21917957e-01 2.85548568e-01
1.42216489e-01 1.00273155e-01 4.89721447e-01 5.88224709e-01
-5.05490363e-01 5.60548842e-01 -1.26890823e-01 -7.00740933e-01
-1.32340097e+00 2.37996206e-01 3.39685455e-02 -3.98203619e-02
-1.18821096e+00 6.56956255e-01 -2.77147919e-01 -3.01636666e-01
6.16173387e-01 -4.40136611e-01 -1.56976521e-01 -3.55515361e-01
1.14380765e+00 4.61339951e-01 1.95893958e-01 -5.59377193e-01
-4.87158895e-01 8.16338360e-01 -8.00806358e-02 -2.80769289e-01
1.23102617e+00 -3.81988995e-02 4.25258934e-01 4.13888872e-01
7.09962785e-01 -5.80444112e-02 -1.38998854e+00 -1.26880586e-01
-6.60224706e-02 -6.98147118e-01 -2.60503124e-02 -7.05927134e-01
-1.04405236e+00 1.02931130e+00 1.17320085e+00 -3.12787056e-01
9.43745732e-01 -1.63910374e-01 1.22630060e+00 5.83974838e-01
1.12350631e+00 -6.03080273e-01 -4.24529985e-02 4.61807430e-01
6.98044002e-01 -1.04537225e+00 -3.27593595e-01 -4.52748239e-01
-3.80055606e-01 1.04047227e+00 5.54865777e-01 -1.76635925e-02
4.60733593e-01 3.62137437e-01 -1.43014891e-02 -6.86665252e-02
1.04582630e-01 -4.05507773e-01 2.26609066e-01 7.55845010e-01
1.67497620e-01 2.91219711e-01 -2.08961770e-01 9.60206747e-01
9.22581777e-02 3.53095204e-01 -4.89212684e-02 9.75749910e-01
3.98712745e-03 -9.10003304e-01 -3.04907233e-01 4.71844852e-01
-4.32703853e-01 7.42407814e-02 -7.88966641e-02 8.09514463e-01
-2.93750107e-01 7.16127753e-01 4.52809781e-03 -1.04141998e+00
6.07533932e-01 -1.98545277e-01 7.36036897e-01 -2.72740871e-01
-4.38927174e-01 8.22983980e-02 -3.19205254e-01 -8.00872147e-01
-2.43887752e-01 -3.22940677e-01 -1.24616456e+00 -2.09827930e-01
-5.19053161e-01 -5.52889109e-01 8.15107405e-01 1.11115789e+00
4.56640810e-01 6.82575881e-01 3.34120721e-01 -1.29333103e+00
-6.16396785e-01 -9.03684378e-01 -5.76327503e-01 -7.71028176e-02
2.24687889e-01 -8.09487343e-01 2.52211809e-01 -4.16887552e-01] | [6.67405366897583, -0.11719825863838196] |
04dc29fc-cf3d-44fc-bb5b-4c55e29a7480 | weight-excitation-built-in-attention | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/7039_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123750086.pdf | Weight Excitation: Built-in Attention Mechanisms in Convolutional Neural Networks | We propose novel approaches for simultaneously identifying important weights of a convolutional neural network (ConvNet) and providing more attention to the important weights during training. More formally, we identify two characteristics of a weight, its magnitude and its location, which can be linked with the importance of the weight. By targeting these characteristics of a weight during training, we develop two separate weight excitation (WE) mechanisms via weight reparameterization-based backpropagation modifications. We demonstrate significant improvements over popular baseline ConvNets on multiple computer vision applications using WE (e.g. 1.3% accuracy improvement over ResNet50 baseline on ImageNet image classification, etc.). These improvements come at no extra computational cost or ConvNet structural change during inference. Additionally, including WE methods in a convolution block is straightforward, requiring few lines of extra code. Lastly, WE mechanisms can provide complementary benefits when used with external attention mechanisms such as the popular Squeeze-and-Excitation attention block. | ['Md Mafijul Islam Bhuiyan', 'Peng Dai', 'Juwei Lu', 'Wei Li', 'Niamul Quader'] | null | null | null | null | eccv-2020-8 | ['deep-attention', '3d-object-classification', '3d-classification', '3d-human-action-recognition', 'deep-attention', 'attention-score-prediction'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'natural-language-processing', 'time-series'] | [ 2.59822786e-01 1.65998936e-01 -6.03190251e-02 -3.31768245e-01
-3.65789026e-01 -3.39276940e-01 5.86241305e-01 1.20939188e-01
-1.02534199e+00 5.64810514e-01 3.96196425e-01 -3.18038434e-01
-8.56597349e-02 -6.76218510e-01 -1.03822267e+00 -7.10790455e-01
4.82248887e-02 -3.61516178e-02 4.15277600e-01 -1.71357259e-01
3.28766584e-01 6.37921691e-01 -1.29008508e+00 2.20667750e-01
5.29557884e-01 1.18132377e+00 4.35812861e-01 7.76136637e-01
-1.09984003e-01 1.01509225e+00 -5.77049315e-01 -2.88932204e-01
2.79613644e-01 1.28397167e-01 -9.59072173e-01 -1.52438536e-01
7.83114314e-01 -4.83223349e-01 -2.38148570e-01 1.06189406e+00
3.91642690e-01 1.61254689e-01 3.32193762e-01 -9.98308957e-01
-6.19017303e-01 9.57958698e-01 -7.58146882e-01 6.80804670e-01
-3.60673875e-01 1.54095814e-01 1.25868309e+00 -1.08197284e+00
3.59657794e-01 1.04137325e+00 9.24561024e-01 6.17556810e-01
-1.53994608e+00 -8.20979953e-01 4.14774746e-01 8.94601569e-02
-1.09271669e+00 -5.37005007e-01 5.77412128e-01 -4.03662294e-01
1.18850458e+00 7.82373473e-02 5.53793550e-01 9.19785738e-01
7.53922090e-02 8.10851634e-01 5.57434380e-01 -2.12734327e-01
9.06925499e-02 -7.56238922e-02 4.97281909e-01 6.86424077e-01
3.21437031e-01 -2.08987013e-01 -2.24829063e-01 1.08530231e-01
9.79665101e-01 2.44306922e-01 -2.84440190e-01 9.75844562e-02
-1.05140722e+00 6.57001376e-01 8.41018915e-01 6.49168193e-02
-4.73647207e-01 9.97851253e-01 3.28424633e-01 2.76216328e-01
4.52499628e-01 7.28402317e-01 -7.04012454e-01 -6.47225901e-02
-9.12605107e-01 1.71064168e-01 2.20776871e-01 7.71880925e-01
1.12930059e+00 2.97127098e-01 -4.14498657e-01 8.74250233e-01
1.35835499e-01 7.14456290e-02 4.68004972e-01 -7.73924351e-01
5.69953203e-01 4.97456193e-01 -2.29792818e-01 -7.58737504e-01
-5.24914920e-01 -7.08352447e-01 -6.73379421e-01 2.37232566e-01
2.31923610e-01 -5.14902115e-01 -1.05078661e+00 1.99672365e+00
-1.00132950e-01 6.14937782e-01 -8.37598667e-02 5.92475176e-01
7.69575357e-01 5.29533565e-01 4.26256925e-01 3.77788991e-01
1.54535520e+00 -1.16609907e+00 -2.72872895e-01 -6.38240516e-01
4.74175006e-01 -7.31310785e-01 1.11878681e+00 6.42326474e-02
-1.40414524e+00 -6.64373159e-01 -1.09665406e+00 -3.30628783e-01
-2.52786756e-01 2.02518523e-01 7.52796412e-01 2.16590434e-01
-1.33874357e+00 9.64389741e-01 -9.28692758e-01 -5.95391877e-02
7.00842738e-01 7.77095795e-01 -2.44694963e-01 2.92926133e-01
-9.24771369e-01 9.69836354e-01 4.33645159e-01 -1.41091973e-01
-6.78511143e-01 -1.32734096e+00 -9.11188364e-01 6.31718338e-01
1.34979352e-01 -6.48234665e-01 1.37086248e+00 -1.03899598e+00
-1.19435036e+00 5.45464396e-01 -1.75097853e-01 -7.55660772e-01
1.92582324e-01 -4.61005807e-01 -1.21274710e-01 2.20679909e-01
-2.90667176e-01 1.12140286e+00 9.41471517e-01 -9.28129792e-01
-6.12020552e-01 9.58122909e-02 3.98500562e-01 -3.70820984e-02
-8.23384821e-01 1.76266953e-01 -5.08851290e-01 -1.08239055e+00
-6.41835928e-02 -7.35304952e-01 -3.53763908e-01 3.01421911e-01
-2.33000651e-01 -2.51617402e-01 6.51545703e-01 -3.57844740e-01
1.21063066e+00 -2.15267062e+00 5.63385710e-02 1.04513161e-01
7.02417552e-01 4.53939319e-01 -5.37619233e-01 -4.87415353e-03
-4.30678308e-01 3.16407710e-01 -1.34738505e-01 -4.13353860e-01
-7.60709494e-02 9.83625352e-02 -4.94237065e-01 1.41713366e-01
8.57255757e-01 1.33670151e+00 -8.22110713e-01 2.29561552e-02
1.42125204e-01 6.11169100e-01 -1.02373636e+00 -1.34686351e-01
8.47916827e-02 -1.65277496e-01 -2.41201505e-01 1.21979974e-01
6.23620570e-01 -3.49705070e-01 -1.70414105e-01 -6.36538208e-01
-7.62458593e-02 5.64451396e-01 -9.79977012e-01 1.35489655e+00
-6.63294733e-01 8.80782783e-01 -5.69580356e-03 -8.53857875e-01
7.20546722e-01 6.03359826e-02 1.37736782e-01 -1.99325398e-01
3.37012827e-01 -9.22224224e-02 4.34404820e-01 -1.12026311e-01
7.90727675e-01 1.40849739e-01 4.98142093e-01 5.79610348e-01
4.50893134e-01 3.30428541e-01 1.66748360e-01 1.44357756e-01
1.24657857e+00 -2.08322003e-01 3.02309357e-02 -7.19901741e-01
1.81190550e-01 -4.38191921e-01 4.06827688e-01 6.17533505e-01
-2.37050112e-02 6.73932016e-01 5.52715421e-01 -4.96566683e-01
-9.47107971e-01 -6.92079663e-01 -2.14511678e-01 1.44078207e+00
-9.74504501e-02 -4.53928560e-01 -6.31233573e-01 -4.97490346e-01
1.40604109e-01 4.69778866e-01 -9.33981717e-01 -4.01648194e-01
-8.94490302e-01 -6.91158175e-01 6.97520554e-01 1.14893115e+00
6.23426616e-01 -1.02159023e+00 -6.93659723e-01 2.73929864e-01
3.41524601e-01 -1.08517075e+00 -6.11837089e-01 5.74899912e-01
-9.97769713e-01 -8.42234850e-01 -8.75434935e-01 -8.84936512e-01
1.04450858e+00 4.45540428e-01 1.20859575e+00 4.11995798e-01
-2.86010832e-01 1.53591335e-01 -4.86993557e-03 -3.83365721e-01
1.12814315e-01 4.74406242e-01 -1.26229703e-01 -7.32185021e-02
2.63738334e-01 -7.09135294e-01 -8.33191216e-01 2.29686499e-01
-8.24909627e-01 5.95337711e-02 5.41399240e-01 8.64918709e-01
2.33409464e-01 -2.36000136e-01 4.59015757e-01 -9.74118292e-01
6.57558918e-01 -4.57609832e-01 -4.55766499e-01 -4.25378233e-03
-3.61982346e-01 5.87172627e-01 5.65754294e-01 -6.25363052e-01
-8.34278584e-01 -1.38569638e-01 -4.28634226e-01 -4.18241829e-01
7.45338127e-02 3.57641995e-01 4.52851385e-01 -2.86603510e-01
6.44607186e-01 -9.85150933e-02 -2.23075002e-01 -4.23280478e-01
2.10253909e-01 5.40444888e-02 3.38060081e-01 -7.28747964e-01
8.91453743e-01 2.39373893e-01 -3.58419865e-01 -4.29036528e-01
-8.44744861e-01 -3.74435812e-01 -3.87513816e-01 1.77862272e-01
7.17148125e-01 -9.95975137e-01 -8.67302537e-01 4.20302480e-01
-1.05262268e+00 -5.86162269e-01 -4.52992827e-01 4.74194884e-01
-1.79300867e-02 -2.08511343e-03 -8.81557226e-01 -2.70224661e-01
-5.23003697e-01 -1.35372424e+00 7.79312074e-01 3.50739002e-01
-2.47987822e-01 -1.11727214e+00 -3.78188580e-01 -1.80361122e-01
8.97567809e-01 -7.00106770e-02 8.05258095e-01 -4.60691571e-01
-4.07590479e-01 -3.53600979e-02 -7.89511919e-01 4.41848874e-01
2.96625718e-02 3.73537429e-02 -1.24397230e+00 -3.72243315e-01
-2.59756595e-01 -5.13235688e-01 1.66055858e+00 4.20091748e-01
1.62521470e+00 -2.45447263e-01 -1.27651781e-01 9.10011113e-01
1.35039473e+00 -2.14727446e-01 6.10335112e-01 1.51976675e-01
8.53295565e-01 1.97061941e-01 -1.22694105e-01 3.38544637e-01
2.38449216e-01 3.71392339e-01 4.70656961e-01 -4.64090466e-01
-4.89679009e-01 -1.31036881e-02 2.78460115e-01 5.85645795e-01
-4.38414752e-01 2.19565675e-01 -7.73759425e-01 5.61067462e-01
-1.72099817e+00 -9.12663221e-01 1.22866586e-01 1.83950388e+00
1.03754306e+00 4.71828997e-01 -2.44733602e-01 1.40231755e-02
5.66266477e-01 3.84219229e-01 -7.54225433e-01 -4.42428946e-01
-4.98642039e-04 7.33455420e-01 9.94249105e-01 5.58859646e-01
-9.94846761e-01 9.90316093e-01 6.92646170e+00 6.82312489e-01
-1.32493234e+00 -4.01014611e-02 6.95365369e-01 -1.26617387e-01
-4.53917831e-01 -2.87313461e-01 -1.05258346e+00 1.71091914e-01
6.27163053e-01 1.61128670e-01 4.75618452e-01 7.92925894e-01
-1.49740964e-01 1.60688952e-01 -1.12183630e+00 7.74385273e-01
-1.34501517e-01 -1.68705916e+00 1.71266377e-01 -5.82745299e-03
8.47006917e-01 5.01119196e-01 2.17493787e-01 3.04013252e-01
6.11710250e-01 -9.02400076e-01 7.09683597e-01 2.72634357e-01
7.13671565e-01 -8.09406281e-01 6.40001357e-01 -6.23321123e-02
-1.31041229e+00 -2.67960995e-01 -8.38983297e-01 -2.45326445e-01
-2.62562275e-01 5.58202803e-01 -4.76789117e-01 -8.83694217e-02
5.75653911e-01 7.83836424e-01 -7.32585013e-01 9.66630220e-01
-5.33919752e-01 8.51668119e-01 -1.78762674e-01 7.28709623e-02
5.35066187e-01 3.84850144e-01 2.27972373e-01 1.30221200e+00
1.37979686e-01 -1.01047702e-01 -4.43990603e-02 1.07253826e+00
-7.70307899e-01 -3.45642716e-01 1.08687161e-02 1.73123911e-01
5.41965842e-01 1.53577971e+00 -7.04066753e-01 -4.75355655e-01
-3.80782336e-01 8.48489702e-01 7.04788446e-01 4.36941445e-01
-9.63375151e-01 -5.46520531e-01 1.15875685e+00 7.54538625e-02
6.15070045e-01 -1.26368046e-01 -3.34483534e-01 -1.12238657e+00
-1.43940851e-01 -4.46304798e-01 -1.69351585e-02 -6.65721118e-01
-1.01277781e+00 6.07505262e-01 -1.99067026e-01 -7.31673658e-01
2.68260121e-01 -9.39256549e-01 -1.03004968e+00 9.80999053e-01
-2.03271747e+00 -9.21747923e-01 -2.78083444e-01 6.00126803e-01
7.19956458e-01 -1.49341375e-01 8.03283870e-01 3.97599429e-01
-7.33903468e-01 1.00779808e+00 -2.40191028e-01 4.10620987e-01
6.06801450e-01 -1.34811413e+00 1.03764224e+00 8.89390051e-01
1.25327900e-01 1.10912633e+00 3.34619075e-01 -1.78930774e-01
-1.15443909e+00 -1.36417365e+00 7.20495760e-01 -1.23183496e-01
9.00446236e-01 -4.90589350e-01 -9.07992482e-01 9.19170380e-01
4.21560287e-01 3.44636232e-01 4.06938225e-01 4.55314130e-01
-7.28170693e-01 -1.71800330e-01 -8.75481844e-01 7.22885191e-01
1.02875292e+00 -3.62080514e-01 -5.79672217e-01 3.52080874e-02
9.55883682e-01 -2.84808904e-01 -6.66285098e-01 2.70161301e-01
4.83520836e-01 -6.85685992e-01 1.26917970e+00 -7.57539570e-01
7.48968184e-01 -1.23487808e-01 3.82703915e-02 -1.42090046e+00
-8.31090748e-01 -5.29023170e-01 -1.21227890e-01 1.02903116e+00
6.71572626e-01 -7.91318893e-01 7.70280242e-01 7.62486100e-01
-2.91537821e-01 -8.86533976e-01 -4.33224201e-01 -4.86508101e-01
8.53185728e-02 -6.34981930e-01 6.54027462e-01 8.74181986e-01
-2.73258537e-01 1.89992815e-01 -1.85357943e-01 1.11109748e-01
3.29903305e-01 -4.46478903e-01 4.55272973e-01 -1.14056063e+00
-4.02131289e-01 -9.24447715e-01 -3.84105265e-01 -1.30576110e+00
1.59694880e-01 -9.78280544e-01 -1.47527337e-01 -1.32461691e+00
1.91562459e-01 -5.04005611e-01 -9.43266869e-01 1.01364124e+00
-5.29092848e-01 6.34884417e-01 3.69208276e-01 1.15659580e-01
-3.05703461e-01 3.47729892e-01 1.03161454e+00 -2.22352430e-01
1.03685455e-02 -1.62709013e-01 -1.04619825e+00 1.02096117e+00
1.06247914e+00 -4.33925986e-01 -4.83789682e-01 -1.10857987e+00
3.55406940e-01 -6.40690863e-01 4.72383082e-01 -1.18085682e+00
3.58547896e-01 6.13022149e-02 5.05955875e-01 -2.26602092e-01
2.08383083e-01 -6.90850616e-01 -3.17425430e-01 6.56087935e-01
-7.09755123e-01 2.58688152e-01 7.05537796e-01 1.69747725e-01
-9.36346874e-02 -4.93939638e-01 9.56314504e-01 -5.03692515e-02
-7.84443915e-01 3.55619997e-01 -2.89991528e-01 8.45214054e-02
7.23102272e-01 -2.37523362e-01 -5.21331966e-01 1.94816440e-02
-6.09125555e-01 1.92611203e-01 -4.69830818e-02 4.05355453e-01
6.29438102e-01 -1.28953648e+00 -6.17328584e-01 3.12842101e-01
6.76029399e-02 -1.20723441e-01 -2.90118251e-03 8.01356018e-01
-5.29811203e-01 9.91585329e-02 -3.55685115e-01 -2.80448288e-01
-1.18912482e+00 2.03819215e-01 2.41837248e-01 -2.60698885e-01
-5.98911047e-01 1.38149810e+00 4.40806866e-01 -1.94819883e-01
3.72394651e-01 -9.37750816e-01 -1.88456878e-01 1.51220843e-01
9.29940343e-01 2.58804739e-01 1.03948228e-01 -1.67496547e-01
-2.40619346e-01 6.35545254e-01 -4.44685310e-01 2.32674271e-01
1.66495669e+00 2.32517540e-01 -1.79088712e-02 1.09950610e-01
1.20192385e+00 -3.58497165e-02 -1.66812801e+00 -4.84059423e-01
-1.10046610e-01 -1.02390811e-01 3.64817291e-01 -6.89399540e-01
-1.65371251e+00 1.11162305e+00 3.86242092e-01 4.15343195e-02
1.05349874e+00 -2.10687488e-01 8.09821010e-01 6.10947609e-01
-1.05309375e-01 -8.91872644e-01 5.45558743e-02 7.75526166e-01
8.73095572e-01 -1.18234432e+00 -4.84512523e-02 -3.12308893e-02
-5.17516136e-01 1.30172861e+00 6.41288340e-01 -6.43241227e-01
9.05452073e-01 5.51046729e-01 -1.40445471e-01 -1.33086547e-01
-8.94363225e-01 -3.10205042e-01 3.58092159e-01 2.31898502e-01
5.46163857e-01 -1.64557174e-01 -2.71682683e-02 4.85757709e-01
-1.78315714e-01 -2.48374358e-01 3.69025707e-01 6.89597189e-01
-6.43585384e-01 -8.28311563e-01 -5.89072064e-04 5.62913954e-01
-7.13845432e-01 -7.26509392e-01 2.55271047e-02 5.77309251e-01
1.30652830e-01 5.04627407e-01 4.24721807e-01 -4.75135535e-01
3.18689287e-01 -5.92366941e-02 1.91132918e-01 -7.35579610e-01
-9.92599308e-01 -7.65485540e-02 -3.09389830e-02 -4.70028907e-01
-3.78210366e-01 -1.24708854e-01 -1.35596371e+00 -4.26289886e-01
-5.35910130e-01 -1.74504712e-01 4.57244158e-01 7.92057037e-01
5.38812935e-01 1.06221664e+00 3.51345837e-01 -1.06326401e+00
-4.52496797e-01 -9.95537162e-01 -2.29509279e-01 3.54915142e-01
4.36003804e-01 -5.18482983e-01 -1.37241796e-01 5.76370321e-02] | [9.026493072509766, 2.369802236557007] |
251607b9-538a-4b4c-8c46-b4177a970b36 | contrastive-mean-teacher-for-domain-adaptive | 2305.03034 | null | https://arxiv.org/abs/2305.03034v1 | https://arxiv.org/pdf/2305.03034v1.pdf | Contrastive Mean Teacher for Domain Adaptive Object Detectors | Object detectors often suffer from the domain gap between training (source domain) and real-world applications (target domain). Mean-teacher self-training is a powerful paradigm in unsupervised domain adaptation for object detection, but it struggles with low-quality pseudo-labels. In this work, we identify the intriguing alignment and synergy between mean-teacher self-training and contrastive learning. Motivated by this, we propose Contrastive Mean Teacher (CMT) -- a unified, general-purpose framework with the two paradigms naturally integrated to maximize beneficial learning signals. Instead of using pseudo-labels solely for final predictions, our strategy extracts object-level features using pseudo-labels and optimizes them via contrastive learning, without requiring labels in the target domain. When combined with recent mean-teacher self-training methods, CMT leads to new state-of-the-art target-domain performance: 51.9% mAP on Foggy Cityscapes, outperforming the previously best by 2.1% mAP. Notably, CMT can stabilize performance and provide more significant gains as pseudo-label noise increases. | ['Yu-Xiong Wang', 'Liang-Yan Gui', 'Dhiraj Joshi', 'Shengcao Cao'] | 2023-05-04 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Cao_Contrastive_Mean_Teacher_for_Domain_Adaptive_Object_Detectors_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Cao_Contrastive_Mean_Teacher_for_Domain_Adaptive_Object_Detectors_CVPR_2023_paper.pdf | cvpr-2023-1 | ['pseudo-label'] | ['miscellaneous'] | [ 3.55599433e-01 2.42702752e-01 -3.29976678e-01 -4.52720970e-01
-1.11567318e+00 -6.17300868e-01 7.94172645e-01 7.65654584e-03
-4.48430777e-01 5.90784252e-01 -2.15892330e-01 5.51511347e-02
-1.12234443e-01 -4.79911327e-01 -8.51291358e-01 -7.69586802e-01
9.36708227e-02 7.58019865e-01 6.33318543e-01 -2.13735759e-01
-1.32585138e-01 8.98365304e-02 -1.65940499e+00 4.48579013e-01
1.13106728e+00 1.03439069e+00 3.07055622e-01 5.51679492e-01
-1.18240967e-01 7.69940197e-01 -5.34496427e-01 -4.55651611e-01
3.93129557e-01 -3.17383885e-01 -6.64016366e-01 2.37428769e-01
9.78010178e-01 -1.91856265e-01 -1.10314898e-01 1.04513884e+00
6.40120804e-01 -4.46018726e-02 9.59440172e-01 -1.23253739e+00
-9.07640815e-01 5.59965253e-01 -5.75903475e-01 3.61386865e-01
-3.32317710e-01 4.70904201e-01 1.09282625e+00 -1.02338314e+00
3.76977205e-01 1.15584052e+00 7.53074408e-01 7.43807375e-01
-1.54813015e+00 -8.57030272e-01 3.81402344e-01 1.64450884e-01
-1.38025606e+00 -5.47485232e-01 6.22780383e-01 -6.82879865e-01
9.90453482e-01 -1.27358600e-01 2.70319432e-01 1.35497487e+00
-4.12311971e-01 1.03245378e+00 1.19162464e+00 -5.40811002e-01
2.31448844e-01 7.24673867e-01 1.08455092e-01 4.93141264e-01
1.15020648e-01 3.36515903e-01 -5.21362305e-01 -1.63932666e-02
4.25963670e-01 -2.51693130e-01 1.07899085e-01 -5.99542439e-01
-1.01014149e+00 6.66813135e-01 5.45402408e-01 1.09347165e-01
-1.56081140e-01 -9.08913165e-02 1.78011164e-01 2.58747309e-01
7.85106063e-01 6.45614624e-01 -7.76006103e-01 1.58312649e-01
-9.92620051e-01 2.33627841e-01 4.74827349e-01 1.19839644e+00
9.15623426e-01 3.18568259e-01 -4.87523794e-01 9.85580504e-01
1.45693660e-01 7.60912836e-01 5.18566549e-01 -7.12548494e-01
2.46192843e-01 6.04654431e-01 1.75355867e-01 -2.99608678e-01
-4.54630554e-01 -1.06105399e+00 -4.63944614e-01 2.70451486e-01
6.09335065e-01 -5.42902499e-02 -1.13185036e+00 1.90246749e+00
5.46738982e-01 5.52050710e-01 2.05672279e-01 7.91715384e-01
9.92834747e-01 3.13556999e-01 4.47998941e-01 -1.10483374e-02
1.41623187e+00 -1.42060506e+00 -2.95656711e-01 -6.95861280e-01
4.26874608e-01 -3.40648532e-01 1.36188889e+00 3.11361432e-01
-7.39417970e-01 -7.76995301e-01 -1.01694846e+00 2.14876950e-01
-4.68810827e-01 3.15481573e-01 1.75994575e-01 6.31960154e-01
-9.84005928e-01 3.64702195e-01 -6.48662090e-01 -4.42824036e-01
7.13694394e-01 3.29658329e-01 -1.49992779e-01 1.27768710e-01
-8.79914165e-01 1.11774457e+00 4.79489625e-01 -6.65323913e-01
-1.33084965e+00 -1.08798027e+00 -3.92744929e-01 -1.64344199e-02
6.04326665e-01 -5.70564151e-01 1.86088252e+00 -1.06271458e+00
-1.52933538e+00 1.24526501e+00 3.45905311e-02 -7.97758698e-01
6.42382503e-01 -5.19410610e-01 -5.94535172e-01 5.09348325e-02
2.92930126e-01 1.11775053e+00 1.30578780e+00 -1.36540866e+00
-1.02334964e+00 -2.39763573e-01 -2.90033579e-01 1.71675771e-01
-6.17016613e-01 -2.57113993e-01 -3.66035759e-01 -4.86698478e-01
-8.48218948e-02 -8.72816086e-01 -1.54686362e-01 -8.79648924e-02
-2.01580212e-01 -4.92100567e-01 8.71192336e-01 -1.35723185e-02
8.79471421e-01 -2.32466912e+00 -2.68384725e-01 -2.16033608e-01
5.01955807e-01 6.78468406e-01 -3.84763747e-01 -1.36424229e-01
-6.03474900e-02 -2.08223864e-01 -2.79251069e-01 -3.93900454e-01
1.02703817e-01 1.88664824e-01 -4.39300746e-01 3.42769921e-01
8.79456937e-01 1.02441823e+00 -1.22470665e+00 -5.62921703e-01
2.04200104e-01 2.65685767e-01 -5.66478491e-01 3.91223222e-01
-6.15736365e-01 6.05604947e-01 -4.03447866e-01 6.30863965e-01
7.35661268e-01 -6.22250855e-01 -3.24010141e-02 1.03972845e-01
2.56257188e-02 3.18524837e-01 -9.33305264e-01 1.53490913e+00
-2.08535150e-01 6.60804987e-01 3.14993151e-02 -1.03079355e+00
1.04478467e+00 3.01398300e-02 4.30625170e-01 -9.61903214e-01
-1.14323264e-02 3.72827202e-01 -1.36822075e-01 -4.22874838e-01
2.67236590e-01 -2.35210788e-02 1.52929693e-01 1.92632169e-01
6.57311141e-01 -1.92638978e-01 1.43953979e-01 6.19672984e-02
1.06509793e+00 2.25865439e-01 3.28491926e-01 -3.58607382e-01
2.30829045e-01 2.75519520e-01 5.21263778e-01 1.09420025e+00
-5.47282994e-01 6.63531303e-01 1.96003899e-01 -3.87046546e-01
-9.92374837e-01 -1.35801184e+00 -3.89953136e-01 1.73201263e+00
7.20676780e-02 -1.65001795e-01 -8.18276644e-01 -1.13266551e+00
2.33030632e-01 7.20132768e-01 -6.69389129e-01 -3.48134965e-01
-4.23760533e-01 -8.80305290e-01 5.17753899e-01 4.74124342e-01
5.25033295e-01 -8.94667029e-01 -5.14195383e-01 3.54104340e-01
4.15664986e-02 -1.28393245e+00 -2.22060412e-01 6.41437054e-01
-8.37978661e-01 -8.15995455e-01 -8.53383839e-01 -7.23374844e-01
4.40589368e-01 4.50867653e-01 1.73005962e+00 -3.97135645e-01
-2.13182136e-01 3.79442453e-01 -3.99787456e-01 -7.53282785e-01
-5.24129093e-01 4.03978735e-01 1.82281390e-01 3.71059068e-02
6.84556603e-01 -7.56591916e-01 -5.93047976e-01 4.10997599e-01
-6.22516930e-01 -1.53441846e-01 8.50549161e-01 1.01143587e+00
5.95544994e-01 -2.69506276e-01 9.18193877e-01 -1.19494915e+00
1.73594967e-01 -6.81379914e-01 -8.15423012e-01 2.56714433e-01
-1.05533421e+00 1.57835320e-01 3.98179919e-01 -8.49309027e-01
-1.16107416e+00 3.53227466e-01 1.40537411e-01 -5.25428474e-01
-3.32721025e-01 -2.49260187e-01 -4.48127128e-02 -1.16300266e-02
1.60046935e+00 3.29814225e-01 -1.61763206e-01 -5.50515652e-01
5.86714804e-01 7.42301524e-01 8.95550132e-01 -5.66887021e-01
9.22863066e-01 4.40430194e-01 -3.35904002e-01 -4.77861673e-01
-1.53121829e+00 -7.78648674e-01 -7.85691857e-01 -5.10636978e-02
7.00708151e-01 -1.38196719e+00 -2.73521960e-01 4.36045259e-01
-7.68197000e-01 -6.37979031e-01 -6.00088179e-01 1.84878781e-01
-4.95606273e-01 -1.39574260e-01 -1.45232305e-01 -8.21270764e-01
-2.55956888e-01 -9.11495626e-01 1.24765706e+00 2.67661482e-01
1.01656444e-01 -7.10529506e-01 1.41597324e-04 1.05254628e-01
5.33948779e-01 -1.68999016e-01 2.75437802e-01 -1.07428408e+00
-5.07680237e-01 1.16004229e-01 -5.94972253e-01 5.79304039e-01
5.60654141e-02 -3.64824802e-01 -1.51856315e+00 -2.26531684e-01
-1.48508534e-01 -5.96267223e-01 1.07985973e+00 2.75338084e-01
8.38695407e-01 -2.17659950e-01 -4.18858707e-01 6.43744171e-01
1.19791842e+00 -2.07058072e-01 1.11242279e-01 6.54616714e-01
5.18218875e-01 5.11553109e-01 6.46188676e-01 5.51101863e-01
5.08332908e-01 7.58533657e-01 4.94627357e-01 -1.89092621e-01
-6.03572428e-01 -4.56660956e-01 3.16542685e-01 2.20934317e-01
4.05081928e-01 -5.71356826e-02 -1.03137577e+00 7.12654352e-01
-1.84172380e+00 -7.48166621e-01 -2.55301148e-01 2.02078319e+00
1.09064734e+00 3.56619865e-01 5.56451082e-01 -3.03723067e-01
5.02657950e-01 -9.63825062e-02 -9.81256127e-01 3.76042455e-01
-3.05470765e-01 2.07199436e-02 7.19602346e-01 1.32123098e-01
-1.65756726e+00 1.18943977e+00 6.74925518e+00 9.77798402e-01
-1.10291815e+00 6.99545085e-01 4.69098717e-01 -1.29921332e-01
2.07600743e-01 -1.70366317e-01 -1.21153963e+00 4.51014936e-01
9.89768207e-01 1.57115310e-01 2.33296379e-01 1.43647349e+00
-2.21607119e-01 8.16009641e-02 -1.05181968e+00 9.79031861e-01
-3.94496694e-02 -1.15713966e+00 -3.55920941e-01 -1.96073189e-01
1.08596432e+00 6.71049118e-01 4.69206601e-01 7.75513172e-01
7.72427619e-01 -6.08129740e-01 9.86880004e-01 6.85822591e-02
8.68422031e-01 -2.29306847e-01 4.01722401e-01 5.67439079e-01
-9.36099470e-01 -3.49327624e-01 -4.55504209e-01 1.52967840e-01
-1.72038272e-01 6.22646630e-01 -1.21221840e+00 9.68016386e-02
9.25094008e-01 7.37309456e-01 -9.33385670e-01 1.18272948e+00
-4.48645025e-01 1.03039515e+00 -3.98519546e-01 9.97403190e-02
2.83220917e-01 2.38611653e-01 6.94361866e-01 1.67688584e+00
2.43315343e-02 -1.60352141e-01 4.29144442e-01 9.90521491e-01
-4.22021300e-02 -8.59462842e-02 -4.17945325e-01 2.12906405e-01
6.18865669e-01 1.21130824e+00 -4.92607355e-01 -7.25122809e-01
-2.91377217e-01 8.41183066e-01 7.26659358e-01 3.30598831e-01
-8.72422278e-01 7.42268637e-02 7.67686725e-01 2.23520547e-01
4.48326826e-01 8.14396292e-02 -4.38224018e-01 -1.10040021e+00
-1.09971799e-01 -8.19511533e-01 4.41734642e-01 -5.58058798e-01
-1.64260304e+00 6.32211089e-01 7.27528408e-02 -1.61390662e+00
-2.21748054e-01 -7.80648470e-01 -3.43776047e-01 6.31206036e-01
-1.98651230e+00 -1.33403206e+00 -4.06067789e-01 4.52810049e-01
6.82410300e-01 -4.43424076e-01 7.80679345e-01 3.46433342e-01
-4.82754141e-01 1.01367700e+00 4.77632731e-01 -1.25503406e-01
1.09604502e+00 -1.57081544e+00 5.76434374e-01 6.68276310e-01
4.14507657e-01 9.87730771e-02 7.13570297e-01 -3.81373554e-01
-7.57115126e-01 -1.47235131e+00 5.28944731e-01 -9.72224236e-01
6.76328123e-01 -6.07519686e-01 -9.80399430e-01 5.37702262e-01
7.20845908e-03 5.26685715e-01 4.27204669e-01 2.76392072e-01
-8.97609472e-01 -3.45206976e-01 -1.16730189e+00 3.91747683e-01
1.11094749e+00 -3.88596863e-01 -6.01400435e-01 7.66906917e-01
1.06023788e+00 -4.84227300e-01 -5.29106915e-01 4.26039875e-01
2.79779136e-01 -9.52131033e-01 1.06637800e+00 -6.55069351e-01
7.27985501e-02 -3.67807418e-01 3.51249473e-03 -1.35744429e+00
-6.32839501e-01 -5.27366817e-01 -3.16759318e-01 1.36264014e+00
5.99614918e-01 -6.22628331e-01 9.95054483e-01 2.16053978e-01
-4.24694896e-01 -2.90280402e-01 -7.51989961e-01 -1.25351727e+00
1.31540224e-01 -4.98121470e-01 5.05711555e-01 9.54655409e-01
-2.79365808e-01 5.83350956e-01 -3.27081025e-01 3.49174798e-01
8.06529522e-01 -1.83159977e-01 9.13344264e-01 -1.47427857e+00
-5.10531187e-01 -6.16261005e-01 -6.75243065e-02 -1.29110742e+00
1.61516741e-01 -8.59324753e-01 4.57684278e-01 -1.11127090e+00
2.57523298e-01 -6.70457721e-01 -6.31048322e-01 7.23429918e-01
-4.15205956e-01 2.95598745e-01 1.38002306e-01 3.84603053e-01
-1.08026302e+00 5.67828000e-01 1.09304976e+00 -3.41934860e-01
-3.78878713e-01 1.34829924e-01 -7.81177044e-01 5.54113030e-01
6.80482090e-01 -8.76181901e-01 -4.57491785e-01 -4.24275249e-01
-3.58775735e-01 -6.48643732e-01 3.11808527e-01 -1.21463311e+00
1.07941560e-01 -1.34090081e-01 2.53472835e-01 -3.75748843e-01
3.05991143e-01 -7.75535643e-01 -4.49047029e-01 2.44771287e-01
-3.74742419e-01 -3.78611505e-01 3.54058772e-01 7.14166045e-01
-1.25579327e-01 -2.08242565e-01 1.14355338e+00 -1.03848971e-01
-1.01750147e+00 1.44726649e-01 -7.89599270e-02 4.26066965e-01
9.69088435e-01 -1.80245534e-01 -6.56261504e-01 2.24626847e-02
-7.54450679e-01 2.54680246e-01 1.74995258e-01 5.00638247e-01
2.15563461e-01 -1.15926051e+00 -1.02383578e+00 2.79252976e-01
6.23093784e-01 1.74121812e-01 2.17811957e-01 8.84519160e-01
2.14341640e-01 3.08434844e-01 -5.96718229e-02 -1.11039054e+00
-8.67068291e-01 6.43241048e-01 3.30867797e-01 -3.03850174e-01
-5.04931152e-01 1.23107529e+00 6.47347450e-01 -7.67959654e-01
3.19506347e-01 -5.48085012e-02 -1.09054223e-01 4.62794080e-02
5.96037209e-01 1.59722388e-01 1.57929063e-01 -4.37970728e-01
-2.20997885e-01 2.92494982e-01 -1.81378484e-01 -2.01647654e-02
1.36666763e+00 2.59839892e-02 4.31520969e-01 3.04524988e-01
9.21914577e-01 -3.39027792e-01 -1.88935769e+00 -7.90193021e-01
3.79983723e-01 -2.84937143e-01 -1.62832253e-02 -1.12321937e+00
-7.78309464e-01 8.25946867e-01 1.07781398e+00 1.43847242e-01
1.14694107e+00 3.33835006e-01 4.36980695e-01 4.79586244e-01
2.08039433e-01 -1.12249255e+00 6.23983204e-01 6.67893469e-01
4.96529847e-01 -1.79179633e+00 -2.64449775e-01 -2.31694639e-01
-8.24322641e-01 5.72928548e-01 1.11458111e+00 -2.42898971e-01
6.00781500e-01 1.42957643e-01 1.84528053e-01 -3.50365676e-02
-8.44189584e-01 -7.25847006e-01 6.26653373e-01 1.07872689e+00
1.54473171e-01 1.03774711e-01 3.10770988e-01 7.76524603e-01
-9.81138945e-02 -2.22920805e-01 -6.89989924e-02 7.21444845e-01
-7.70399213e-01 -9.54343915e-01 -4.25504684e-01 4.93418247e-01
-1.79294333e-01 -1.90257981e-01 -4.05488342e-01 7.83830822e-01
4.02285576e-01 9.34020460e-01 6.86625317e-02 -4.50873315e-01
5.46671629e-01 1.86636865e-01 2.00852722e-01 -8.44036102e-01
-6.28410041e-01 2.60830104e-01 -1.31709874e-01 -2.80805826e-01
-3.50328773e-01 -6.61860764e-01 -9.74848628e-01 2.45106772e-01
-5.05234122e-01 6.15583640e-03 4.41922337e-01 8.35886359e-01
5.90628624e-01 4.76375014e-01 4.88553166e-01 -7.91984737e-01
-1.00326383e+00 -1.26420164e+00 -4.01482284e-01 3.77934575e-01
5.87745070e-01 -9.55057263e-01 -2.22262383e-01 2.18674898e-01] | [9.366785049438477, 1.4798976182937622] |
95aeedbc-e2f3-4872-82cd-9183c770bbb3 | high-performance-automatic-categorization-and | 2202.08965 | null | https://arxiv.org/abs/2202.08965v1 | https://arxiv.org/pdf/2202.08965v1.pdf | High-performance automatic categorization and attribution of inventory catalogs | Techniques of machine learning for automatic text categorization are applied and adapted for the problem of inventory catalog data attribution, with different approaches explored and optimal solution addressing the tradeoff between accuracy and performance is selected. | ['Anton Kolonin'] | 2022-02-09 | null | null | null | null | ['text-categorization'] | ['natural-language-processing'] | [-1.73135065e-02 -1.63966373e-01 -1.02448750e+00 -6.95989966e-01
-6.43536031e-01 -6.32895589e-01 5.58812559e-01 7.53477097e-01
-5.55327594e-01 1.40706658e-01 4.23407614e-01 -2.06871033e-01
-5.86103201e-01 -3.62848401e-01 1.96897820e-01 -1.87647909e-01
2.75110066e-01 1.12097764e+00 5.35033122e-02 -1.25948757e-01
1.42903292e+00 5.55394948e-01 -1.39236474e+00 1.41650513e-01
2.26316974e-01 1.09420705e+00 1.05521753e-01 4.80297714e-01
-9.77922797e-01 1.12152612e+00 -7.79276252e-01 -7.35845745e-01
2.58617848e-01 4.89220172e-02 -1.13739109e+00 3.09056163e-01
3.93806845e-01 1.10141955e-01 -2.22102493e-01 1.07296014e+00
9.10846442e-02 -1.86691925e-01 1.28579843e+00 -1.52821803e+00
-6.64677739e-01 7.99154043e-01 -3.51641119e-01 4.60155100e-01
3.35552812e-01 -4.85213041e-01 1.40049195e+00 -6.15692019e-01
4.02884007e-01 1.42831516e+00 6.96697891e-01 2.01633926e-02
-1.30163729e+00 -7.23122716e-01 -1.38781518e-01 5.94334975e-02
-1.52769721e+00 -1.89032421e-01 5.62514007e-01 -9.61549759e-01
6.74303114e-01 3.30411017e-01 2.02971548e-01 6.45830929e-01
6.74221158e-01 5.69068730e-01 1.29665327e+00 -5.57312012e-01
1.50088504e-01 9.28480923e-01 8.17317843e-01 5.01931548e-01
4.15895373e-01 -2.05960482e-01 -6.20441437e-01 -4.93538558e-01
4.05889273e-01 -2.08481342e-01 4.54541683e-01 -7.13716149e-01
-7.65003383e-01 1.37725282e+00 4.25191782e-02 3.46427500e-01
-1.94347188e-01 -3.90967041e-01 7.42487133e-01 6.21445477e-01
4.53161806e-01 1.14590776e+00 -7.14790404e-01 2.64499992e-01
-9.14926052e-01 5.50621510e-01 8.96462560e-01 1.20475936e+00
4.33425725e-01 3.68567295e-02 6.09549284e-02 9.45345998e-01
3.42786640e-01 -8.73355642e-02 9.88693833e-01 -9.63888764e-01
2.45998189e-01 7.20834136e-01 1.39036849e-01 -1.15754640e+00
-5.99646211e-01 -2.39702463e-01 -1.02633581e-01 2.09113687e-01
2.10124820e-01 5.32794356e-01 -3.71559948e-01 7.76750326e-01
-9.27452445e-02 -1.24741066e+00 -9.92433876e-02 3.06982279e-01
4.24343884e-01 2.52914637e-01 5.24343908e-01 -2.73749262e-01
1.50756764e+00 -7.17201650e-01 -9.10165310e-01 -6.23029564e-03
8.07633221e-01 -1.04608333e+00 7.89571643e-01 3.17248255e-01
-6.61942601e-01 -3.84195745e-01 -1.03968120e+00 -1.73716575e-01
-1.04294968e+00 -3.24350037e-02 5.70851922e-01 7.12931037e-01
-3.43939066e-01 7.42371023e-01 2.15756651e-02 -4.91274774e-01
1.90131500e-01 5.50150990e-01 -2.87967920e-02 3.35057288e-01
-8.47305894e-01 1.21009755e+00 7.37291217e-01 -5.75253844e-01
4.37613092e-02 -3.45284134e-01 -4.56797212e-01 2.91426666e-02
-9.53048747e-03 -1.99800089e-01 1.26403165e+00 -9.46918309e-01
-9.80693996e-01 1.11506331e+00 3.16399097e-01 -4.03045654e-01
4.62610811e-01 9.51473564e-02 -5.64348578e-01 -2.86051065e-01
4.47140068e-01 5.92303991e-01 8.20437729e-01 -1.00978911e+00
-8.57368290e-01 -5.48426330e-01 -6.03189409e-01 1.35076717e-01
-7.60899723e-01 6.85937643e-01 3.27357411e-01 -1.05254197e+00
2.28978962e-01 -6.88400269e-01 2.99591690e-01 -1.84376508e-01
-2.72331145e-02 -8.71643841e-01 6.63628519e-01 -7.77812243e-01
1.12811220e+00 -1.84466136e+00 -1.38797611e-01 1.61907300e-01
-7.03045279e-02 -4.92817968e-01 2.90027916e-01 3.28893453e-01
-1.59424037e-01 3.01759481e-01 3.50501537e-01 -8.82757977e-02
2.56707728e-01 2.18316130e-02 -5.88423073e-01 2.53805995e-01
-3.27798188e-01 3.55551749e-01 -5.88891387e-01 -7.30158806e-01
6.69637620e-02 -5.11023760e-01 -9.90505964e-02 2.28971198e-01
-2.25074932e-01 -3.45536560e-01 -4.17785943e-01 9.10818875e-01
1.21786162e-01 -9.40291509e-02 4.81825531e-01 -1.96724758e-01
-2.55966127e-01 4.86215055e-01 -1.18915379e+00 9.25787747e-01
-1.61963865e-01 6.54298902e-01 1.17052853e-01 -7.69743323e-01
1.18299747e+00 1.23696312e-01 5.08311689e-01 -3.44555706e-01
5.20391643e-01 2.41800040e-01 -3.70971486e-02 -2.23812073e-01
8.77948761e-01 -1.68169484e-01 -5.34264207e-01 7.55803168e-01
4.31423753e-01 -1.65821046e-01 -8.75790566e-02 3.90278906e-01
5.45502245e-01 -1.72245771e-01 6.25891328e-01 -9.54496682e-01
2.59735882e-01 6.55242085e-01 1.34259241e-03 7.02269077e-01
-2.41121486e-01 1.47447348e-01 3.68594080e-01 -4.30064768e-01
-1.15756154e+00 -3.48775953e-01 -3.46560478e-01 1.85193670e+00
8.39312077e-02 -2.45480180e-01 -5.91397882e-01 -7.60112584e-01
6.37105703e-01 1.02008653e+00 -5.22663713e-01 8.70022625e-02
-3.76763314e-01 -6.58934116e-01 3.01838875e-01 5.65963209e-01
-2.36393064e-01 -8.43905270e-01 -2.61188895e-01 2.12913632e-01
1.06330678e-01 -9.32558835e-01 -6.08777463e-01 4.38569874e-01
-1.06678796e+00 -1.11710298e+00 -1.17043748e-01 -8.11006188e-01
3.43660116e-01 5.07927656e-01 1.05323577e+00 1.76817700e-01
-4.40059841e-01 5.11581004e-01 -1.01884268e-01 -7.08385766e-01
-8.00226390e-01 6.36593461e-01 1.70201346e-01 -3.76662403e-01
8.33246231e-01 5.23763150e-03 1.42673716e-01 4.94778246e-01
-6.15879893e-01 -3.80327404e-01 3.57290268e-01 7.31052160e-01
7.73243979e-02 4.11712885e-01 5.38872719e-01 -9.09124970e-01
8.86940300e-01 -4.40683514e-01 -6.78992569e-01 2.80044973e-01
-1.81768703e+00 1.94322348e-01 3.18875492e-01 -3.12974870e-01
-8.52119923e-01 -1.73610896e-01 2.07582027e-01 -1.56689927e-01
-1.91943467e-01 4.62287292e-02 -2.09234715e-01 -4.93810982e-01
4.43543732e-01 -9.29316059e-02 -5.77755347e-02 -8.07571769e-01
3.23703140e-01 9.74139035e-01 2.72161007e-01 -4.05018419e-01
3.00772071e-01 -9.75502282e-02 -2.17838496e-01 -4.00317580e-01
-9.22327161e-01 -6.83663487e-01 -1.06600356e+00 -2.39127472e-01
4.41064268e-01 -4.58406538e-01 -5.91569662e-01 2.03207299e-01
-6.85822845e-01 4.81193870e-01 -8.98855850e-02 4.17627484e-01
-7.47276902e-01 1.04041852e-01 -5.44808507e-01 -8.07353616e-01
-3.32565367e-01 -8.68657768e-01 5.57851791e-01 -2.61471063e-01
-8.47418606e-01 -9.80138183e-01 -5.57464967e-03 6.21395886e-01
2.82282293e-01 -4.17572826e-01 1.84807682e+00 -1.53541636e+00
-2.52895236e-01 -6.61743164e-01 -1.93336695e-01 5.79915754e-02
2.23326370e-01 -1.90703481e-01 -7.10159183e-01 -1.46155268e-01
8.79940614e-02 -3.46118063e-01 2.48850495e-01 1.79046750e-01
1.29423857e+00 -5.36510348e-01 -4.26919073e-01 -8.00164342e-02
1.46468854e+00 2.35421419e-01 -5.53999767e-02 9.80425060e-01
4.92350280e-01 1.18474555e+00 8.14926803e-01 8.67414698e-02
2.28550285e-01 6.44460380e-01 5.09425402e-02 4.41961288e-01
2.78846268e-02 -2.13304371e-01 -3.24110866e-01 8.05408776e-01
6.35292351e-01 -3.71416181e-01 -9.23646808e-01 4.54232037e-01
-1.65085351e+00 -7.08206236e-01 6.27792627e-02 1.87408161e+00
3.69076192e-01 6.38153732e-01 6.75586224e-01 4.54170704e-01
8.94196332e-01 -1.84943676e-02 -3.65684927e-01 -4.97673601e-01
-7.24173039e-02 -2.64548600e-01 1.19547296e+00 3.62173885e-01
-1.16312671e+00 7.52286434e-01 9.03137779e+00 8.32433999e-01
-4.39363569e-01 2.39648491e-01 4.59038734e-01 8.50562230e-02
-4.35112976e-02 -1.38843507e-01 -9.03407097e-01 4.97178853e-01
8.16475511e-01 -2.99168736e-01 2.60969520e-01 1.00461376e+00
-4.02772337e-01 6.03990741e-02 -1.04560697e+00 5.78882396e-01
2.42320493e-01 -1.01168823e+00 4.21516657e-01 9.49378610e-02
1.96852475e-01 -4.44758415e-01 -4.43734601e-02 3.50335449e-01
5.36890209e-01 -3.99863869e-01 1.00631118e+00 9.93503258e-02
5.87428100e-02 -7.09505975e-01 6.15618408e-01 1.77389637e-01
-6.51833892e-01 -5.35162210e-01 -1.46602124e-01 1.50349215e-01
-2.28590578e-01 -1.41704172e-01 -6.07614875e-01 2.80144662e-01
6.44790351e-01 3.51781070e-01 -8.11774135e-01 6.06975555e-01
5.50282717e-01 2.19581842e-01 2.70012707e-01 -3.91812652e-01
8.79600793e-02 -4.94866557e-02 5.50258994e-01 1.40134895e+00
-1.02254987e-01 -1.12123728e-01 5.43763816e-01 5.19936085e-01
3.79753718e-03 4.98240650e-01 -4.90586460e-01 -4.35911417e-02
6.55532181e-01 1.10374224e+00 -1.01145744e+00 -2.80118406e-01
-8.27548280e-02 5.29202461e-01 1.79432541e-01 -2.10237339e-01
-2.24582389e-01 -3.29663128e-01 5.09319544e-01 2.94422477e-01
6.80916160e-02 2.20866930e-02 -9.80636716e-01 -5.14745533e-01
-5.24413884e-01 -8.34316254e-01 1.04044962e+00 -6.18027031e-01
-1.71156156e+00 1.54311880e-01 2.30963230e-01 -9.13394928e-01
-1.66481048e-01 -1.08877146e+00 1.82946175e-02 6.55218124e-01
-7.35732317e-01 -7.77558446e-01 3.33520412e-01 3.47152958e-03
9.79845226e-01 -9.04363751e-01 7.88413644e-01 2.67082397e-02
-3.02330792e-01 5.29698730e-01 4.51097250e-01 3.63559760e-02
8.27911735e-01 -1.56445456e+00 2.22165778e-01 3.68246473e-02
-1.64113298e-01 2.87750810e-01 7.78850257e-01 -6.53187156e-01
-1.22350967e+00 -8.65719438e-01 9.10620511e-01 -6.22835040e-01
1.17827535e+00 -2.21405506e-01 -8.34848881e-01 5.65775454e-01
3.26212168e-01 -1.18541050e+00 8.24469447e-01 3.19899917e-01
-7.48975456e-01 -8.90106410e-02 -1.44962180e+00 2.48294860e-01
6.26118362e-01 -6.55272067e-01 -7.84557104e-01 6.35755718e-01
2.46175557e-01 -6.41198382e-02 -1.27370894e+00 -1.66355535e-01
4.76661384e-01 -1.15390345e-01 8.21045995e-01 -9.08007801e-01
1.77269764e-02 7.96002597e-02 -2.24698186e-01 -9.74025607e-01
-7.58437872e-01 -1.43496662e-01 1.09926187e-01 1.36001420e+00
3.54049385e-01 -5.17607152e-01 5.27541518e-01 6.98872924e-01
3.24534178e-01 -4.59919944e-02 -6.79305732e-01 -7.82983303e-01
3.50347519e-01 4.66149524e-02 4.17846292e-01 1.33133292e+00
2.20656589e-01 8.89592350e-01 -6.94230646e-02 -4.64946657e-01
9.69497323e-01 3.81734461e-01 3.39918256e-01 -1.69599462e+00
4.14246142e-01 -1.15837073e+00 -4.31573004e-01 -5.55016696e-01
6.23231530e-01 -8.12331796e-01 -7.83662423e-02 -8.32712770e-01
4.19923007e-01 -7.67260015e-01 -2.81712234e-01 1.39881000e-01
7.63882250e-02 -2.38156274e-01 3.51363957e-01 8.77765298e-01
-1.99586898e-01 -1.49156794e-01 2.19651058e-01 -5.82665980e-01
7.97489211e-02 -6.54396554e-03 -9.01344597e-01 6.56198502e-01
8.93129945e-01 -1.00357699e+00 -7.25989118e-02 -1.12355433e-01
-2.90994346e-02 5.72019536e-03 1.36957452e-01 -5.39521337e-01
1.66504338e-01 -3.03988278e-01 5.11094093e-01 -8.41676831e-01
-5.27719371e-02 -9.17081058e-01 3.10447030e-02 2.63065487e-01
-1.20901799e+00 6.38501644e-01 -2.14605287e-01 7.34942615e-01
1.28786266e-02 -1.03908956e+00 9.08704162e-01 -2.21558258e-01
-5.95133483e-01 -1.44706383e-01 -9.46822524e-01 -2.46068552e-01
6.81988955e-01 1.17024660e-01 -2.47194439e-01 -1.59890100e-01
-5.09598136e-01 8.22469220e-02 3.70438218e-01 9.08118427e-01
-1.14437744e-01 -1.12168980e+00 -4.11155790e-01 -3.21834534e-01
3.73028874e-01 -1.32837749e+00 -6.40225768e-01 4.35458198e-02
-1.75099105e-01 8.70527446e-01 -2.83402056e-01 -7.60176554e-02
-1.22829139e+00 1.05367565e+00 2.85237014e-01 -1.46212175e-01
-3.26448023e-01 2.55428165e-01 -1.55822918e-01 -3.85797679e-01
5.15981615e-01 1.89472407e-01 -6.43584549e-01 2.40538806e-01
3.87650102e-01 7.13423073e-01 3.73900086e-01 -6.84092164e-01
-1.59905821e-01 1.39640406e-01 -5.11651695e-01 -5.16226888e-03
6.54350162e-01 -3.44956815e-01 -2.50298828e-01 8.24364662e-01
1.23914468e+00 -4.52089071e-01 -3.19942236e-01 -5.56090117e-01
8.89476180e-01 -4.32829052e-01 4.79990661e-01 -9.48060811e-01
-5.22220731e-01 3.15080643e-01 9.59983528e-01 6.05043471e-01
6.21138752e-01 -1.13505386e-01 2.73382634e-01 4.91736084e-01
1.26593590e-01 -1.65411973e+00 -1.48277152e-02 3.25913459e-01
5.62708974e-01 -1.08321548e+00 6.36871397e-01 -2.36710042e-01
-5.94325483e-01 1.45475268e+00 5.03343165e-01 1.48636192e-01
6.80210769e-01 1.42985985e-01 5.20364419e-02 -4.90434229e-01
-3.41714203e-01 1.72864258e-01 2.74293363e-01 2.90446103e-01
6.56248927e-01 1.34666160e-01 -6.17623150e-01 3.29545408e-01
-2.34373719e-01 -6.12679839e-01 3.09851408e-01 1.18125844e+00
-5.44082820e-01 -1.03263736e+00 -6.38466537e-01 8.70304108e-01
-6.89359844e-01 4.64691296e-02 -8.22361946e-01 9.73049998e-01
-1.87220544e-01 8.87476802e-01 3.97809565e-01 -1.93935469e-01
2.91540325e-01 4.36163336e-01 2.86657393e-01 -4.04407978e-01
-1.03672659e+00 7.45415986e-02 1.40547067e-01 7.18851835e-02
-5.59763551e-01 -1.02019274e+00 -7.16930032e-01 -4.29394782e-01
-7.82733083e-01 5.53694606e-01 1.06556296e+00 7.02815413e-01
1.43636882e-01 2.23897696e-01 1.06654155e+00 -3.78611714e-01
-9.35189247e-01 -1.03233683e+00 -8.31101716e-01 2.97726482e-01
-1.05076253e-01 -8.42257679e-01 -3.49658132e-01 2.00344607e-01] | [9.995992660522461, 6.579092025756836] |
c2c357e4-8f63-48c7-a4b4-2599424077aa | unmasked-teacher-towards-training-efficient | 2303.16058 | null | https://arxiv.org/abs/2303.16058v1 | https://arxiv.org/pdf/2303.16058v1.pdf | Unmasked Teacher: Towards Training-Efficient Video Foundation Models | Video Foundation Models (VFMs) have received limited exploration due to high computational costs and data scarcity. Previous VFMs rely on Image Foundation Models (IFMs), which face challenges in transferring to the video domain. Although VideoMAE has trained a robust ViT from limited data, its low-level reconstruction poses convergence difficulties and conflicts with high-level cross-modal alignment. This paper proposes a training-efficient method for temporal-sensitive VFMs that integrates the benefits of existing methods. To increase data efficiency, we mask out most of the low-semantics video tokens, but selectively align the unmasked tokens with IFM, which serves as the UnMasked Teacher (UMT). By providing semantic guidance, our method enables faster convergence and multimodal friendliness. With a progressive pre-training framework, our model can handle various tasks including scene-related, temporal-related, and complex video-language understanding. Using only public sources for pre-training in 6 days on 32 A100 GPUs, our scratch-built ViT-L/16 achieves state-of-the-art performances on various video tasks. The code and models will be released at https://github.com/OpenGVLab/unmasked_teacher. | ['Yu Qiao', 'LiMin Wang', 'Yinan He', 'Yi Wang', 'Yizhuo Li', 'Yali Wang', 'Kunchang Li'] | 2023-03-28 | null | null | null | null | ['action-classification', 'video-question-answering', 'video-retrieval', 'spatio-temporal-action-localization'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 2.82036692e-01 -2.18818873e-01 -4.74603444e-01 -3.37367415e-01
-9.31786180e-01 -3.68858755e-01 4.72308606e-01 -4.13409740e-01
-4.06034201e-01 2.38775909e-01 3.12215060e-01 -3.30075920e-01
2.92346776e-01 -2.86465466e-01 -1.11403871e+00 -5.11226177e-01
1.84211314e-01 2.78849363e-01 3.35740507e-01 -1.67235360e-01
9.68562439e-02 -1.89915299e-01 -1.78514397e+00 7.95564294e-01
8.66494954e-01 8.35832596e-01 8.68308306e-01 7.76676297e-01
-2.33433202e-01 9.54657912e-01 -2.22687781e-01 -3.45417202e-01
1.87851667e-01 -3.58858883e-01 -9.57775950e-01 1.01557858e-01
8.46968174e-01 -6.18449688e-01 -5.82208812e-01 8.83895516e-01
4.94044185e-01 7.65312314e-02 2.25736752e-01 -1.57081842e+00
-3.51239979e-01 5.56775749e-01 -6.42275453e-01 1.34800524e-01
3.11270177e-01 3.06135267e-01 7.75188804e-01 -1.10101700e+00
7.84862936e-01 1.14053047e+00 5.35632253e-01 6.92808270e-01
-9.40903842e-01 -7.73980737e-01 2.85552591e-01 7.05756724e-01
-1.36271834e+00 -8.25015545e-01 3.15972388e-01 -3.18194211e-01
1.27791166e+00 2.34892517e-01 6.96351171e-01 1.33520126e+00
4.34573144e-02 1.13342905e+00 7.90476441e-01 -2.29518563e-01
-1.63630113e-01 -2.04847395e-01 -2.04794973e-01 8.07382226e-01
-3.59958589e-01 -8.82055536e-02 -9.95162368e-01 1.55618027e-01
6.40838623e-01 -2.49656200e-01 -4.01748121e-01 -2.99680352e-01
-1.47127724e+00 5.56837797e-01 1.00692809e-01 5.93112148e-02
-1.51885629e-01 4.71580029e-01 6.73652291e-01 2.63038248e-01
3.81081223e-01 -8.56623724e-02 -5.53097486e-01 -5.63053489e-01
-1.16129303e+00 -2.25864276e-02 4.65536743e-01 1.29241920e+00
8.44874978e-01 1.44483268e-01 -1.65570050e-01 8.47744584e-01
3.42055917e-01 4.63498324e-01 3.48862648e-01 -1.28759074e+00
5.97732544e-01 1.22505397e-01 -2.79176772e-01 -7.76678562e-01
-1.17561623e-01 -2.49074578e-01 -6.70277834e-01 -2.28638083e-01
2.48705521e-01 2.45891184e-01 -1.22270954e+00 1.81121349e+00
5.33497334e-01 8.39934707e-01 3.83688770e-02 1.05016327e+00
1.16792238e+00 9.68993545e-01 1.68108344e-01 -1.40275225e-01
1.35220528e+00 -1.56009817e+00 -6.29822373e-01 -4.03388500e-01
8.79304409e-01 -1.00271773e+00 1.34771240e+00 4.67337787e-01
-1.10867882e+00 -6.27856672e-01 -8.12264562e-01 -4.69591886e-01
1.01568550e-01 9.87331793e-02 7.39638686e-01 2.73277819e-01
-1.38020468e+00 3.61237764e-01 -1.01693511e+00 -4.56125081e-01
4.31605339e-01 2.75760680e-01 -4.12999064e-01 -4.42963123e-01
-9.26590681e-01 6.59232199e-01 3.37203145e-01 2.86191907e-02
-1.23428893e+00 -8.27169538e-01 -9.62423563e-01 -1.81668609e-01
5.60262680e-01 -8.55445266e-01 1.36766446e+00 -1.24851406e+00
-1.50964952e+00 9.00214195e-01 -4.55632687e-01 -2.59001613e-01
5.73882878e-01 -2.59131432e-01 -2.43537754e-01 3.86321723e-01
2.37550363e-01 1.22622252e+00 1.05776143e+00 -9.36177313e-01
-6.08622432e-01 1.77135020e-01 4.58168611e-02 4.57444310e-01
-3.54530632e-01 1.82708442e-01 -1.32854366e+00 -6.28487647e-01
-6.87692910e-02 -9.68098640e-01 -2.79199816e-02 -1.06107712e-01
-1.06928423e-01 -8.68192911e-02 9.40559387e-01 -7.40233362e-01
1.06160831e+00 -2.31821918e+00 4.40258861e-01 -2.02117279e-01
1.72182232e-01 3.89491647e-01 -5.18040061e-01 5.32353759e-01
9.32187680e-03 -1.79943562e-01 -8.42975155e-02 -7.04298735e-01
-1.10114381e-01 3.96119028e-01 -4.03824568e-01 3.89746666e-01
5.46618141e-02 9.43485737e-01 -9.41977143e-01 -7.99549997e-01
4.45559531e-01 5.22117198e-01 -9.08528328e-01 2.82977581e-01
-4.46531475e-01 4.65334117e-01 -2.47477964e-01 7.25365281e-01
7.75379539e-01 -4.79864895e-01 2.79965520e-01 -5.23828804e-01
-1.56035408e-01 2.67996430e-01 -8.88525426e-01 2.42477512e+00
-4.52301025e-01 7.59063065e-01 2.21253008e-01 -1.00427365e+00
3.60271037e-01 3.11538309e-01 5.96187949e-01 -8.28235209e-01
-1.37425074e-02 2.02735245e-01 -1.79785624e-01 -6.69349432e-01
7.68422663e-01 3.46604586e-01 2.30368033e-01 3.37212712e-01
2.70690084e-01 3.53509840e-03 2.71610916e-01 6.78711295e-01
9.45601523e-01 6.71764970e-01 -1.36708885e-01 -1.84521586e-01
2.04635113e-01 1.93817362e-01 6.15816891e-01 6.17445588e-01
-1.57961160e-01 6.53422713e-01 2.31881231e-01 -2.88815022e-01
-1.01776946e+00 -9.69016731e-01 -2.01899558e-02 1.48329806e+00
3.65885973e-01 -1.01972783e+00 -7.66782641e-01 -4.27404523e-01
-2.61614025e-01 4.18038845e-01 -2.81063735e-01 -9.90414321e-02
-5.91765285e-01 -5.27392745e-01 6.42585397e-01 4.34971154e-01
5.08255243e-01 -9.07165945e-01 -3.93402338e-01 1.88505292e-01
-7.71692038e-01 -1.45193458e+00 -7.04803050e-01 -2.31959164e-01
-6.55631423e-01 -9.23259616e-01 -5.48783839e-01 -8.91170144e-01
4.48303252e-01 7.47719586e-01 1.27281475e+00 3.18289608e-01
-1.72627628e-01 5.26749551e-01 -4.28049266e-01 -9.16148275e-02
-3.29904199e-01 9.35450792e-02 1.02232948e-01 -1.49171472e-01
1.28383785e-01 -3.79706234e-01 -6.49416089e-01 5.30189157e-01
-8.68206859e-01 7.85522997e-01 3.36697340e-01 8.40045571e-01
7.27568805e-01 -4.56324816e-01 2.94523239e-01 -6.03231668e-01
-1.46434113e-01 -5.52860975e-01 -4.05583203e-01 2.84776628e-01
-2.57186949e-01 -2.14538008e-01 3.91113341e-01 -7.23839819e-01
-1.05217123e+00 7.25979209e-02 -1.71165660e-01 -9.31991816e-01
1.08391576e-01 4.36971962e-01 -1.42196745e-01 -1.03763947e-02
3.65663886e-01 3.96134675e-01 4.95710522e-02 -3.06689411e-01
5.23512483e-01 5.84506333e-01 7.63397455e-01 -9.64950204e-01
4.81065512e-01 5.24122477e-01 -4.11144674e-01 -1.00285745e+00
-8.21430922e-01 -6.26318634e-01 -2.63554156e-01 -4.73217398e-01
8.94400477e-01 -1.47962737e+00 -6.05847776e-01 6.43336475e-01
-1.04301834e+00 -8.37047398e-01 5.13628796e-02 5.25391340e-01
-7.03041911e-01 6.87371016e-01 -7.42812157e-01 -2.59707034e-01
-3.33024859e-01 -1.48562181e+00 1.29780054e+00 2.53535062e-02
-2.19784081e-02 -7.25380480e-01 -1.88900590e-01 8.21254790e-01
2.89901793e-01 -2.75165439e-01 4.45230037e-01 -1.13905873e-02
-9.60430980e-01 4.72019166e-01 -3.82320642e-01 2.63970137e-01
-1.59187317e-01 1.16810530e-01 -9.29805875e-01 -5.73185146e-01
-2.12761045e-01 -7.77044535e-01 1.03985643e+00 4.33933616e-01
1.39353013e+00 -1.26167089e-01 -2.78942704e-01 1.33003330e+00
1.16786242e+00 -2.38055795e-01 7.34200239e-01 4.72617626e-01
1.13178480e+00 3.58924538e-01 9.37011242e-01 5.29405832e-01
9.06694174e-01 7.50076771e-01 6.74374163e-01 -2.03096062e-01
-3.20936620e-01 -3.33650261e-01 7.54503369e-01 1.26931715e+00
-9.06674564e-03 -4.39496338e-01 -8.24701309e-01 6.40215099e-01
-2.20913911e+00 -8.50405097e-01 -2.89555222e-01 1.76942217e+00
8.91158223e-01 -2.26167440e-01 -4.74508852e-04 -4.47767109e-01
5.56916118e-01 3.41377348e-01 -5.55060089e-01 -2.00559348e-01
-2.62207210e-01 4.85838614e-02 5.08309484e-01 3.80509824e-01
-1.03779125e+00 1.44495535e+00 5.44833803e+00 1.31828773e+00
-1.18400192e+00 3.96951258e-01 5.01799703e-01 -3.33084524e-01
-3.60628664e-01 1.98292375e-01 -7.19359696e-01 3.84303361e-01
8.83259773e-01 1.00859404e-01 6.21510983e-01 6.31111979e-01
2.69713163e-01 -9.84619409e-02 -1.10544133e+00 1.22386670e+00
1.33126900e-01 -1.57385969e+00 9.65873301e-02 -5.33399172e-03
7.48918235e-01 5.90640128e-01 3.91725861e-02 3.69576156e-01
1.02478169e-01 -9.88349140e-01 1.10288095e+00 2.78606266e-01
1.31410480e+00 -4.32651281e-01 3.86430502e-01 2.83798546e-01
-1.44979048e+00 2.36562192e-01 -4.11805958e-01 1.45535860e-02
2.83398986e-01 2.05583394e-01 -6.12291932e-01 5.80633581e-01
9.62107420e-01 1.10896122e+00 -3.50654781e-01 5.86774111e-01
-1.15827464e-01 7.49637961e-01 -5.46461880e-01 4.55417454e-01
3.75767380e-01 -1.62028491e-01 3.18690211e-01 1.33581924e+00
4.66983885e-01 1.03923827e-01 4.01973724e-01 3.84479403e-01
-4.45446596e-02 8.65945220e-02 -4.60604846e-01 3.02507393e-02
4.05160666e-01 1.20481491e+00 -6.40085340e-01 -4.59229112e-01
-6.11114860e-01 1.14164817e+00 3.16499889e-01 3.33030760e-01
-1.28011763e+00 3.04811716e-01 9.73119974e-01 1.99307948e-02
3.40138614e-01 -3.73790622e-01 3.53624709e-02 -1.70776677e+00
-1.89056266e-02 -1.22905982e+00 6.14934802e-01 -8.92715693e-01
-8.83109391e-01 4.36619818e-01 1.47189751e-01 -1.23121619e+00
-2.75502473e-01 -3.78731906e-01 -3.65104496e-01 3.94469589e-01
-1.48981047e+00 -1.53874564e+00 -4.08485949e-01 9.22192097e-01
9.33970392e-01 -5.42753004e-02 5.24117112e-01 6.29801512e-01
-5.94049335e-01 6.52699232e-01 -6.78400919e-02 -1.11205220e-01
9.52589631e-01 -7.93807387e-01 5.26932359e-01 9.50124204e-01
2.07653791e-01 3.88438523e-01 6.22359514e-01 -6.13516092e-01
-1.86456966e+00 -1.16015434e+00 6.58282399e-01 -2.74636626e-01
6.95950627e-01 -4.88834649e-01 -8.78545523e-01 8.52458954e-01
3.75796050e-01 1.01413086e-01 4.70092416e-01 -1.23564944e-01
-4.83148754e-01 8.14717114e-02 -5.60316563e-01 7.74497807e-01
1.53896153e+00 -6.22081518e-01 -1.98355719e-01 5.16917586e-01
8.42373252e-01 -8.47367644e-01 -7.06460893e-01 3.14637929e-01
5.39950609e-01 -9.71016586e-01 1.14481938e+00 -3.84070963e-01
6.26253426e-01 -4.13582563e-01 -3.64196807e-01 -8.99966419e-01
-2.63744555e-02 -7.58290768e-01 -1.28180444e-01 1.16894877e+00
2.73227468e-02 -2.90918231e-01 7.76728511e-01 2.93559760e-01
-4.90398526e-01 -6.68169022e-01 -8.69870603e-01 -7.28964090e-01
-3.12138796e-01 -1.05732632e+00 3.56786370e-01 1.07953143e+00
-1.60329312e-01 1.56612843e-01 -7.97817826e-01 1.49650693e-01
5.82166195e-01 1.40380085e-01 1.05399525e+00 -2.82233834e-01
-4.82261300e-01 -2.63895780e-01 -1.25298634e-01 -1.53586948e+00
2.93651938e-01 -9.61428165e-01 1.58123504e-02 -1.48369753e+00
3.10162246e-01 -4.85667169e-01 -7.42031410e-02 8.35131824e-01
-4.24095541e-02 5.21039963e-01 5.40621281e-01 3.19649488e-01
-9.23873186e-01 7.67198622e-01 1.24278879e+00 -2.02170849e-01
-1.76543556e-02 -5.04696310e-01 -3.19392800e-01 7.10209310e-01
7.42598176e-01 -4.04635787e-01 -6.39396071e-01 -1.01591635e+00
1.14929311e-01 1.56009465e-01 3.60830665e-01 -7.29194045e-01
2.45597631e-01 -2.79671818e-01 1.71491429e-02 -6.99983358e-01
6.09401941e-01 -4.48239535e-01 4.39901143e-01 3.01533133e-01
-4.96562347e-02 1.76620275e-01 3.81529212e-01 3.26010138e-01
-3.07280391e-01 -8.01232755e-02 5.57791591e-01 -4.51987535e-02
-1.28892529e+00 5.13058543e-01 -2.96720117e-01 1.75788671e-01
8.20011497e-01 -2.60790706e-01 -5.79121947e-01 -5.36764979e-01
-3.76745135e-01 5.49422979e-01 7.76194930e-01 6.65974975e-01
8.47648978e-01 -1.12666106e+00 -7.96383202e-01 5.98296598e-02
2.45810762e-01 1.30787671e-01 7.73855627e-01 1.03696370e+00
-6.74879611e-01 1.08419731e-01 -1.17834583e-01 -1.16429579e+00
-1.43969488e+00 4.62369263e-01 2.64287218e-02 3.14957500e-02
-8.10208857e-01 1.10067427e+00 6.65701151e-01 -3.03546846e-01
2.03358829e-01 -1.37571290e-01 2.16429710e-01 -2.63899677e-02
4.83955741e-01 1.26573235e-01 -2.80284099e-02 -8.45782757e-01
-5.06796122e-01 5.86576343e-01 -1.32206500e-01 -1.02724567e-01
1.20777845e+00 -2.08647609e-01 -7.69956931e-02 1.24852791e-01
1.21772873e+00 -2.97258824e-01 -1.54877591e+00 -3.23147357e-01
-4.32257473e-01 -5.53869069e-01 1.44182324e-01 -4.42621291e-01
-1.23358512e+00 8.76007318e-01 2.57350385e-01 -5.19608259e-01
1.32864070e+00 8.43200237e-02 1.00748146e+00 1.54697597e-01
4.50497270e-01 -1.12118661e+00 1.23238638e-01 6.64827585e-01
6.62953615e-01 -1.31534922e+00 2.52244286e-02 -5.80445766e-01
-8.58652592e-01 9.15049672e-01 8.06450009e-01 3.33504319e-01
2.56197780e-01 3.31459939e-01 1.54491335e-01 -2.70364806e-03
-1.24403858e+00 -2.80995667e-01 2.80622035e-01 5.09443939e-01
2.23048210e-01 -8.44266415e-02 -8.60000253e-02 2.56121248e-01
-1.24340922e-01 3.61269563e-02 4.96308923e-01 9.16387141e-01
-1.18730322e-01 -1.10543203e+00 -2.06510067e-01 1.88639104e-01
-3.39293957e-01 -4.65229630e-01 1.72608107e-01 7.11237073e-01
1.40041113e-01 8.59529734e-01 1.54678643e-01 -5.56465685e-01
-4.27289158e-02 -2.66933560e-01 6.52012348e-01 -4.16966230e-01
-4.42179710e-01 4.47148412e-01 2.96170861e-01 -1.05444992e+00
-6.83994889e-01 -7.43576884e-01 -1.27472091e+00 -6.68741107e-01
-9.56501216e-02 -1.13943510e-01 6.55585527e-01 9.89172220e-01
6.21277809e-01 4.07601237e-01 2.32274324e-01 -1.07094812e+00
-1.74290761e-01 -7.11818933e-01 4.80278805e-02 3.94736350e-01
1.67334110e-01 -4.84898269e-01 3.98865491e-02 3.57476532e-01] | [9.940741539001465, 0.8601009249687195] |
1ff3cce1-080a-4e25-96a5-c9e1fe48a2a0 | visualizing-evidence-for-alzheimers-disease | 1903.07317 | null | http://arxiv.org/abs/1903.07317v1 | http://arxiv.org/pdf/1903.07317v1.pdf | Visualizing evidence for Alzheimer's disease in deep neural networks trained on structural MRI data | Deep neural networks have led to state-of-the-art results in many medical
imaging tasks including Alzheimer's disease (AD) detection based on structural
magnetic resonance imaging (MRI) data. However, the network decisions are often
perceived as being highly non-transparent making it difficult to apply these
algorithms in clinical routine. In this study, we propose using layer-wise
relevance propagation (LRP) to visualize convolutional neural network decisions
for AD based on MRI data. Similarly to other visualization methods, LRP
produces a heatmap in the input space indicating the importance of each voxel
contributing to the final classification outcome. In contrast to susceptibility
maps produced by guided backpropagation ("Which change in voxels would change
the outcome most?"), the LRP method is able to directly highlight positive
contributions to the network classification in the input space. Thus, the
highlighted areas can be interpreted as the 'positive evidence' used by the
network for deciding whether an individual has AD. We find that this
LRP-evidence indeed fulfills those expectations that one would have towards AD
evidence: (1) it is very specific for individuals ("Why does this person have
AD?") with high inter-patient variability, (2) there is very little evidence
for AD in healthy controls and (3) areas that exhibit a lot of evidence
correlate well with what is known from literature. To quantify the latter, we
compute size-corrected metrics of the summed evidence per brain area, e.g. the
'evidence density' or 'evidence gain'. Although these metrics produce very
individual 'fingerprints' of relevance patterns for AD patients, a lot of
importance is put on areas in the temporal lobe including hippocampus and
amygdala. We conclude that LRP provides a powerful tool for assisting
clinicians in finding evidence for AD (and potentially other diseases) in
structural MRI data. | ['Moritz Böhle', 'Martin Weygandt', 'Kerstin Ritter', 'Fabian Eitel'] | 2019-03-18 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [ 1.42650619e-01 3.85642767e-01 -2.97317151e-02 -6.01175666e-01
-2.36995786e-01 -2.52167881e-01 6.36523843e-01 5.84788322e-01
-4.83303934e-01 7.74012804e-01 4.33921069e-01 -5.59943557e-01
-5.75707734e-01 -7.85053432e-01 -3.99129301e-01 -7.41826892e-01
-7.42554843e-01 3.86255771e-01 1.29282340e-01 2.09082793e-02
9.28460211e-02 6.48152530e-01 -1.10859323e+00 5.69675505e-01
6.22213244e-01 9.95237231e-01 1.11689016e-01 9.52703282e-02
-2.03937180e-02 7.49797881e-01 -5.13418555e-01 -2.56367415e-01
-1.74584016e-01 -3.52900803e-01 -9.43375528e-01 -4.20640022e-01
5.47177672e-01 -2.20056280e-01 9.01295021e-02 1.21182525e+00
5.25102615e-01 -3.49289805e-01 8.95307660e-01 -9.12814260e-01
-7.48902619e-01 5.50735116e-01 -5.16487360e-01 7.93998718e-01
5.67668229e-02 4.55157608e-01 8.89072239e-01 -6.82086289e-01
1.00096262e+00 1.39480329e+00 7.13631928e-01 3.10194731e-01
-1.43233156e+00 -4.75009382e-01 2.28984833e-01 4.49691772e-01
-8.22727978e-01 -9.92743298e-03 5.74502707e-01 -9.12247717e-01
8.86017859e-01 5.69077671e-01 1.18748963e+00 8.90727282e-01
6.56113267e-01 1.41917095e-01 1.60819793e+00 -1.85223922e-01
4.17671770e-01 -8.60569160e-03 2.55020767e-01 4.97434229e-01
4.01384175e-01 1.50810331e-01 -1.46127641e-01 -3.49240899e-01
9.10153866e-01 -5.77215850e-02 -2.05073074e-01 -5.24417758e-02
-1.51058578e+00 7.43837893e-01 1.09422314e+00 5.11117041e-01
-5.72843075e-01 2.01176018e-01 4.70211238e-01 2.07297862e-01
5.65879405e-01 3.47906083e-01 -2.85008341e-01 4.48243409e-01
-1.08877540e+00 4.45317656e-01 2.18480915e-01 -2.95140624e-01
2.13866815e-01 -6.06501848e-02 -2.44151816e-01 9.98317659e-01
6.28605783e-01 2.03715190e-01 5.31992674e-01 -8.83911371e-01
2.80513346e-01 8.00370574e-01 -1.75787956e-01 -1.03581321e+00
-9.30738032e-01 -6.56196237e-01 -1.05911422e+00 1.04548383e+00
4.78899062e-01 -1.10153943e-01 -9.78299022e-01 1.69128275e+00
3.24386433e-02 -6.05802000e-01 -4.53868330e-01 1.18575251e+00
8.26786637e-01 1.63305983e-01 5.10765612e-01 8.43424574e-02
1.78821599e+00 -3.11311156e-01 -5.98766267e-01 -5.15108228e-01
5.97767174e-01 -1.21709310e-01 1.09553301e+00 2.40783527e-01
-9.62476969e-01 -2.50188679e-01 -1.17264521e+00 1.30748600e-01
-5.30161679e-01 -7.35592395e-02 6.06482387e-01 4.08699811e-01
-1.41815281e+00 7.71321535e-01 -7.32228756e-01 -3.21183294e-01
9.90907013e-01 3.97212237e-01 -5.23624480e-01 1.15972579e-01
-1.21730626e+00 1.50268018e+00 2.15101302e-01 3.22129518e-01
-8.37444842e-01 -8.87067735e-01 -4.57732648e-01 1.19791955e-01
-1.52198136e-01 -8.38886559e-01 7.00873494e-01 -1.08961773e+00
-6.04522943e-01 1.10362601e+00 1.22225590e-01 -4.68587250e-01
6.48370266e-01 1.04276873e-01 -6.10689461e-01 3.65397453e-01
5.61492778e-02 1.09392846e+00 4.71380323e-01 -1.10228324e+00
-1.76044211e-01 -8.44286740e-01 1.50256315e-02 -3.45080271e-02
8.52103606e-02 4.90689427e-01 4.78026688e-01 -8.80249500e-01
2.98915118e-01 -5.59327722e-01 -5.93479633e-01 8.93200099e-01
-3.81513447e-01 -1.83509335e-01 6.54633582e-01 -1.00900066e+00
8.86321306e-01 -1.84461498e+00 -2.40955025e-01 4.35104072e-01
9.70403194e-01 -6.76350575e-03 2.96269655e-01 -1.59244269e-01
-6.20128691e-01 6.13437057e-01 -6.64642274e-01 3.38029802e-01
-1.36057856e-02 -7.84553960e-02 2.30620146e-01 6.15991533e-01
4.47487384e-01 9.99388158e-01 -7.61934757e-01 -4.71105754e-01
3.18011865e-02 8.39759409e-01 -2.93266088e-01 -2.94863999e-01
7.40130916e-02 4.91535008e-01 -1.36783838e-01 4.93211478e-01
5.60788333e-01 -5.12274325e-01 4.43645447e-01 -2.65133113e-01
-9.19184312e-02 4.86984581e-01 -7.50499547e-01 8.48959208e-01
1.06961749e-01 8.42276454e-01 1.58800691e-01 -9.31760967e-01
8.55876386e-01 3.20117682e-01 2.52056301e-01 -9.39807713e-01
-5.80815896e-02 2.68350184e-01 8.17208052e-01 -2.90721834e-01
-3.56112808e-01 -3.66486788e-01 4.60616857e-01 7.34336913e-01
-1.07949346e-01 5.05077422e-01 -1.52457263e-02 1.08083241e-01
1.18751931e+00 -1.37278706e-01 3.76812518e-01 -6.46523476e-01
3.01008821e-01 -4.30931114e-02 2.77770251e-01 4.41606015e-01
-3.13652277e-01 5.53562701e-01 9.47164416e-01 -6.40283227e-01
-9.77671444e-01 -1.13324273e+00 -4.99411464e-01 5.48056543e-01
-4.11164999e-01 -1.21965565e-01 -6.63770437e-01 -6.51453912e-01
-1.20574847e-01 5.09226441e-01 -1.18096030e+00 -3.25172693e-02
-5.65042078e-01 -1.12183285e+00 3.06911141e-01 5.96858740e-01
5.51930487e-01 -1.21566844e+00 -1.04198813e+00 4.70097400e-02
1.23744085e-01 -3.64431292e-01 1.19339325e-01 3.87835383e-01
-1.35412824e+00 -1.05777848e+00 -1.13064981e+00 -5.93806684e-01
9.82429445e-01 -2.50802219e-01 1.23745871e+00 1.43996924e-01
-5.55032313e-01 -7.29345903e-02 1.29460365e-01 -1.33121535e-01
-3.52823943e-01 -4.18769598e-01 -1.90001249e-01 -3.93217564e-01
1.62191600e-01 -7.57584870e-01 -1.23892057e+00 2.08482847e-01
-7.79813051e-01 -2.86476500e-02 7.13810444e-01 4.52015460e-01
6.13119543e-01 -8.33914578e-02 6.07250869e-01 -6.63525224e-01
9.55832601e-01 -7.06677437e-01 -2.89530330e-03 6.94683641e-02
-9.05761123e-01 -2.10741181e-02 1.85366660e-01 -2.92608649e-01
-7.50239968e-01 -5.05290031e-01 -7.95562714e-02 1.84587479e-01
-4.62025911e-01 7.33623147e-01 1.44852117e-01 8.58558342e-02
9.19775009e-01 -3.92982394e-01 2.42749006e-01 -7.24048078e-01
3.10053285e-02 1.99792370e-01 2.94885546e-01 -3.13944757e-01
6.26998246e-02 4.38827842e-01 7.73430392e-02 -3.78281802e-01
-2.33671129e-01 1.85024679e-01 -6.65691793e-01 -4.89021629e-01
9.87135291e-01 -3.63087207e-01 -5.14790595e-01 5.94716519e-02
-9.65753555e-01 -5.15471935e-01 -2.46553674e-01 4.98132855e-01
-3.42750810e-02 1.11682072e-01 -3.87297720e-01 -4.44914699e-01
-5.60544848e-01 -1.29574466e+00 4.43663597e-01 -1.04142867e-01
-8.19159567e-01 -1.33778608e+00 1.29481182e-01 -1.57913432e-01
6.26540542e-01 4.86314356e-01 1.72831059e+00 -6.02410853e-01
4.20762375e-02 4.25010435e-02 -6.60653174e-01 1.53040543e-01
8.29505548e-02 -1.76812515e-01 -7.66105890e-01 5.85237108e-02
3.59230563e-02 4.26125415e-02 8.06006670e-01 7.40624368e-01
1.15317786e+00 -5.44955611e-01 -4.78240877e-01 -1.40993834e-01
1.29828346e+00 1.65700406e-01 7.57018387e-01 5.51053286e-01
3.47044498e-01 9.96529043e-01 -4.63091135e-02 -6.65132701e-02
1.60162792e-01 7.63522983e-01 6.42318964e-01 -5.11733770e-01
-5.00831008e-01 3.18937123e-01 4.59180027e-01 -6.32492155e-02
-3.80407095e-01 3.39155436e-01 -1.21546209e+00 3.67795527e-01
-1.49928379e+00 -7.64925241e-01 -6.17755413e-01 2.21362638e+00
8.20769906e-01 3.13213646e-01 2.40218461e-01 1.72765851e-01
7.45687604e-01 2.31693700e-01 -4.80636269e-01 -5.79281509e-01
-1.30838886e-01 3.10040981e-01 1.82098135e-01 2.94341713e-01
-7.88625062e-01 9.26972181e-02 6.96955442e+00 2.69754499e-01
-1.42742956e+00 3.54366273e-01 1.11650002e+00 -2.19086602e-01
-4.26540285e-01 -1.48597628e-01 -1.86763272e-01 6.25933349e-01
1.04356790e+00 2.49498963e-01 -2.06984254e-03 5.11376321e-01
5.36297202e-01 -2.96094149e-01 -1.09330213e+00 5.67004442e-01
-4.03459370e-01 -1.27464926e+00 1.05154164e-01 3.10952902e-01
2.45537400e-01 1.49346560e-01 1.42995879e-01 -3.95734817e-01
2.43271187e-01 -1.33993649e+00 7.66695082e-01 7.49104321e-01
9.72529769e-01 -3.54545921e-01 7.20116019e-01 -2.73469031e-01
-5.71916819e-01 6.09296151e-02 -1.78932562e-01 -4.63015102e-02
1.62407756e-01 1.08436263e+00 -1.01317108e+00 -1.76383063e-01
1.09785163e+00 4.31819886e-01 -8.66755664e-01 1.24581397e+00
-4.84227866e-01 7.53334105e-01 -2.15852305e-01 1.84828386e-01
3.37908566e-01 -2.19502859e-02 6.17817283e-01 1.34282172e+00
2.17741445e-01 -2.07924098e-01 -4.48106050e-01 1.52153683e+00
2.49901190e-01 2.29307130e-01 -4.39060658e-01 -1.74905080e-02
-9.29710865e-02 1.30238318e+00 -1.22466600e+00 -4.14189577e-01
-1.20839782e-01 6.31136715e-01 1.30887121e-01 2.87073642e-01
-3.57699037e-01 -1.40751585e-01 2.77217150e-01 6.10580623e-01
-8.91335830e-02 -1.05827577e-01 -6.85081959e-01 -3.60646307e-01
1.43726856e-01 -6.45412385e-01 3.81197274e-01 -1.08486104e+00
-1.29885030e+00 6.70238137e-01 1.06088221e-02 -7.62604117e-01
-4.77330238e-02 -9.59627211e-01 -7.44732201e-01 1.14622724e+00
-1.13095295e+00 -8.53968322e-01 -6.40351251e-02 2.83001840e-01
-1.15693532e-01 1.80941597e-01 9.45346653e-01 2.93098330e-01
-3.30681384e-01 9.11410153e-02 -1.83747411e-01 5.16337976e-02
4.70704347e-01 -1.46079195e+00 2.88940609e-01 2.94304460e-01
-2.08158016e-01 7.96431482e-01 6.63626492e-01 -9.36588109e-01
-3.94794881e-01 -6.78850234e-01 7.29297578e-01 -1.96775883e-01
6.59577250e-01 1.57254577e-01 -1.15342331e+00 4.60770279e-01
1.50362283e-01 1.33349240e-01 7.98019230e-01 2.28422552e-01
-3.20029855e-01 1.48738384e-01 -1.45054722e+00 6.30002677e-01
8.07313383e-01 -1.97540611e-01 -7.46389925e-01 3.82840216e-01
2.15211436e-01 2.49819145e-01 -1.06856155e+00 4.89449382e-01
8.20768595e-01 -1.02269316e+00 1.07170904e+00 -7.96571493e-01
6.94743335e-01 -1.40039191e-01 9.69404504e-02 -1.43642306e+00
-6.17817223e-01 3.81409764e-01 5.59559390e-02 6.82893813e-01
7.34013319e-01 -8.37191343e-01 3.67444038e-01 8.19685102e-01
-1.04625657e-01 -1.06803143e+00 -1.13775206e+00 -3.42631787e-01
3.65131289e-01 -3.10273498e-01 4.20324951e-01 1.10523844e+00
5.45940921e-02 3.59609425e-02 3.21820915e-01 -1.01489693e-01
4.76548225e-01 -4.44480568e-01 -3.92396092e-01 -1.69002724e+00
9.15647745e-02 -9.31149006e-01 -5.21070659e-01 1.63829356e-01
-3.38112980e-01 -1.36984789e+00 -4.60027784e-01 -2.03105235e+00
3.79614592e-01 -6.37447476e-01 -7.25656807e-01 8.07661712e-01
4.42739911e-02 4.43081111e-01 9.77806598e-02 4.19123501e-01
-5.90371713e-02 -1.46838233e-01 1.22647166e+00 -3.11954170e-01
-3.76756638e-02 -3.90108734e-01 -8.82782042e-01 8.56956065e-01
9.62977111e-01 -4.53325838e-01 -1.46633580e-01 -2.90049881e-01
6.40686214e-01 -3.95351171e-01 1.16902149e+00 -1.03073251e+00
-2.84553289e-01 9.77838561e-02 1.06790805e+00 -2.73278713e-01
1.35521606e-01 -6.38732255e-01 2.66153038e-01 8.15869868e-01
-4.52094853e-01 1.40232101e-01 2.51000792e-01 5.32131307e-02
6.13065027e-02 -1.91098124e-01 7.21666157e-01 -5.29905915e-01
-3.65501583e-01 -2.65808795e-02 -6.03916287e-01 -2.40231007e-01
4.00680184e-01 -2.30392247e-01 -6.07076764e-01 -2.29376063e-01
-1.22219336e+00 -1.35165840e-01 1.29885226e-01 2.99278852e-02
6.68909132e-01 -1.41022456e+00 -7.79966295e-01 -2.39649579e-01
-2.23150194e-01 -4.12672281e-01 4.33310062e-01 1.38167298e+00
-5.75216711e-01 4.19016600e-01 -7.41597652e-01 -6.72635734e-01
-1.15688562e+00 1.61689758e-01 6.40199780e-01 -4.43432212e-01
-1.00868988e+00 6.76184237e-01 3.62117648e-01 -2.08021328e-01
1.20422465e-03 -5.01672089e-01 -5.39392591e-01 4.42183375e-01
8.50127697e-01 5.38017333e-01 2.66144246e-01 -5.29201329e-01
-5.01512885e-01 1.70391828e-01 -3.44086766e-01 -3.31097871e-01
1.54441559e+00 1.91174850e-01 -6.05731845e-01 4.56155717e-01
8.94973934e-01 -3.89807463e-01 -1.03518069e+00 1.58631265e-01
-1.29740583e-02 -7.22848065e-03 3.41616005e-01 -1.63956487e+00
-1.29719722e+00 1.25307190e+00 1.54686701e+00 2.81925112e-01
6.35195792e-01 2.36999452e-01 1.75994605e-01 5.13147488e-02
6.69454932e-02 -8.87060523e-01 -1.75286740e-01 -3.79213579e-02
1.44574308e+00 -9.96995568e-01 2.69623458e-01 -1.00897565e-01
-4.56425548e-01 1.16545963e+00 3.27799767e-01 3.52508179e-03
7.12517381e-01 4.80108969e-02 7.32737035e-02 -9.51572776e-01
-4.51396376e-01 1.68464676e-01 5.12688160e-01 7.45878756e-01
6.32140875e-01 3.77850920e-01 -6.57019854e-01 7.03869700e-01
-1.77861035e-01 9.17195156e-02 8.87485147e-02 7.14954972e-01
-5.71776927e-01 -8.52981925e-01 -5.78289330e-01 1.09939373e+00
-6.13616288e-01 -1.64174557e-01 -7.05373108e-01 6.86370432e-01
2.69549698e-01 3.67540091e-01 2.48507977e-01 -4.47501764e-02
-4.42466289e-02 2.87410736e-01 4.56066906e-01 -4.39229727e-01
-6.67057335e-01 -1.39031306e-01 3.05521876e-01 -4.73230779e-01
-4.29853052e-01 -5.39309919e-01 -1.43322492e+00 1.04673253e-02
1.25627831e-01 -1.84021086e-01 7.44569540e-01 8.92607510e-01
4.02858853e-01 9.10476446e-01 -2.73164928e-01 -8.28834772e-01
1.21990561e-01 -1.13192379e+00 -5.02430677e-01 1.22915141e-01
1.33357003e-01 -9.64369357e-01 -2.59576172e-01 -1.50338054e-01] | [14.168574333190918, -1.777555227279663] |
15c6ccec-6e21-499f-83bf-4ac2fa80302d | paying-attention-to-multiscale-feature-maps | 2103.11247 | null | https://arxiv.org/abs/2103.11247v1 | https://arxiv.org/pdf/2103.11247v1.pdf | Paying Attention to Multiscale Feature Maps in Multimodal Image Matching | We propose an attention-based approach for multimodal image patch matching using a Transformer encoder attending to the feature maps of a multiscale Siamese CNN. Our encoder is shown to efficiently aggregate multiscale image embeddings while emphasizing task-specific appearance-invariant image cues. We also introduce an attention-residual architecture, using a residual connection bypassing the encoder. This additional learning signal facilitates end-to-end training from scratch. Our approach is experimentally shown to achieve new state-of-the-art accuracy on both multimodal and single modality benchmarks, illustrating its general applicability. To the best of our knowledge, this is the first successful implementation of the Transformer encoder architecture to the multimodal image patch matching task. | ['Yosi Keller', 'Aviad Moreshet'] | 2021-03-20 | null | null | null | null | ['multimodal-patch-matching', 'patch-matching'] | ['computer-vision', 'computer-vision'] | [ 3.28332692e-01 1.04652122e-01 -1.78872898e-01 -3.85376930e-01
-1.51900482e+00 -2.45024949e-01 6.45415127e-01 -6.00982420e-02
-5.29036343e-01 2.17890024e-01 3.28207821e-01 3.77502926e-02
2.08907247e-01 -4.38179106e-01 -1.17057979e+00 -4.43369061e-01
1.41328394e-01 3.00481737e-01 1.15443781e-01 -3.44158053e-01
2.69419849e-01 3.77150476e-01 -1.61801875e+00 7.02936411e-01
3.95746976e-01 1.21497118e+00 1.57635987e-01 8.68850231e-01
3.27632189e-01 9.43026125e-01 -2.10425667e-02 -5.61349273e-01
2.14285821e-01 -1.99644178e-01 -1.01849699e+00 1.47870421e-01
1.27981079e+00 -4.61556792e-01 -7.07723677e-01 6.78571701e-01
5.30313611e-01 1.95431679e-01 4.28723156e-01 -1.09908521e+00
-1.28810167e+00 1.41497821e-01 -4.67975110e-01 2.30756581e-01
6.74439371e-01 1.52213007e-01 1.65351677e+00 -1.28545761e+00
9.33604062e-01 1.15665007e+00 6.54163063e-01 4.35108423e-01
-1.54752326e+00 -5.88058122e-02 5.55005297e-02 1.93346545e-01
-1.23670936e+00 -4.93758112e-01 8.83357704e-01 -2.12495878e-01
1.35180092e+00 -1.34423539e-01 4.93645817e-01 9.85008776e-01
2.42695302e-01 1.03197455e+00 8.14663947e-01 -2.86802381e-01
-1.10195369e-01 -9.04666409e-02 -1.49001583e-01 1.11764014e+00
-5.31837106e-01 1.61647841e-01 -6.79683805e-01 -8.93473104e-02
8.29038441e-01 1.84865654e-01 -1.71325207e-01 -7.35695422e-01
-1.41709411e+00 9.62796211e-01 1.03249586e+00 3.84994239e-01
-2.62789786e-01 5.38041174e-01 4.20267731e-01 4.96707320e-01
2.38928452e-01 3.44765753e-01 -3.01235527e-01 5.85873313e-02
-8.25591683e-01 -3.34849544e-02 3.52320433e-01 9.11341488e-01
8.59533668e-01 -1.53008938e-01 -1.43115968e-01 8.73806477e-01
2.23706126e-01 5.15943527e-01 3.99737597e-01 -1.18768084e+00
4.66293395e-01 5.90160012e-01 -1.52454346e-01 -8.33665967e-01
-2.67873019e-01 -2.84363329e-01 -5.56459844e-01 2.83844829e-01
2.19775319e-01 3.79462987e-01 -7.77557790e-01 1.92475402e+00
-1.10373065e-01 8.20591226e-02 -4.74252477e-02 1.11099911e+00
8.79990935e-01 4.06902581e-01 1.32944867e-01 6.16195858e-01
1.25755548e+00 -1.33963394e+00 -2.18659073e-01 -4.50025499e-02
3.59382838e-01 -7.86019683e-01 1.28157258e+00 -1.58366978e-01
-1.55087972e+00 -7.59732783e-01 -9.20848906e-01 -8.34431350e-01
-5.47762334e-01 2.16745600e-01 5.60670257e-01 7.56417289e-02
-1.38233435e+00 5.19571960e-01 -7.21224606e-01 -7.10674107e-01
5.95707536e-01 4.98884231e-01 -9.32126820e-01 -1.25521779e-01
-8.95103097e-01 9.73616242e-01 -2.74172843e-01 -1.39747430e-02
-8.30030024e-01 -7.66934395e-01 -1.18541372e+00 8.56167227e-02
-2.00520143e-01 -1.22100604e+00 1.32459474e+00 -1.20139015e+00
-1.48662579e+00 1.26691973e+00 -3.77391636e-01 -3.95810604e-01
1.00873202e-01 6.95850998e-02 -4.39995266e-02 7.11968005e-01
1.70257092e-01 1.37597299e+00 1.06120837e+00 -1.09959483e+00
-4.54183161e-01 -3.28259528e-01 2.89200008e-01 1.20145202e-01
-3.17931354e-01 -3.05707902e-02 -2.66225427e-01 -6.87267244e-01
-6.37587905e-02 -8.49462152e-01 -2.36687530e-02 4.81972843e-01
1.44478127e-01 -2.12345421e-01 6.58247888e-01 -7.53005028e-01
5.84284604e-01 -2.16199780e+00 6.61070645e-01 4.55334579e-04
1.23496786e-01 -3.82028550e-01 -7.46001065e-01 5.35453320e-01
-2.23364234e-01 -2.26491913e-01 -4.45054740e-01 -8.54249656e-01
1.61746278e-01 4.32326086e-02 -4.12051827e-01 5.86504817e-01
6.90382421e-01 1.52342319e+00 -7.70554185e-01 -3.27019036e-01
2.13268429e-01 8.50354195e-01 -8.54431570e-01 3.75307649e-01
-4.95438837e-02 2.25618884e-01 4.76849750e-02 9.36215699e-01
3.48069906e-01 -4.60422367e-01 -1.41772449e-01 -6.65454030e-01
9.76692662e-02 6.92166463e-02 -4.60688561e-01 2.59473276e+00
-8.24074984e-01 1.00582826e+00 1.81011334e-01 -1.25370407e+00
4.98553634e-01 2.42084727e-01 3.91789854e-01 -9.18792784e-01
1.22136101e-01 3.93277913e-01 -2.56380022e-01 -3.39400679e-01
6.15161538e-01 -1.37619242e-01 -9.27914605e-02 4.46741521e-01
7.53769159e-01 -7.96059892e-02 -6.29589520e-03 4.20624614e-01
1.01460338e+00 3.27409536e-01 4.75227507e-03 -2.22260624e-01
7.08104253e-01 -2.20936447e-01 -2.32632682e-01 5.58671057e-01
-5.77225834e-02 9.94423270e-01 2.84066409e-01 -5.94718158e-01
-1.35172307e+00 -1.22512269e+00 -1.42902985e-01 1.49100649e+00
5.43618016e-02 -1.31790832e-01 -4.54384625e-01 -6.46560907e-01
8.64315107e-02 2.79409681e-02 -1.12228239e+00 -8.64937156e-02
-4.79315370e-01 -9.04582813e-02 4.81657654e-01 7.92257190e-01
2.89178580e-01 -9.71396983e-01 -4.38178450e-01 1.62859902e-01
-2.03579694e-01 -1.24739766e+00 -8.74814987e-01 3.03452741e-02
-7.96142280e-01 -1.14921319e+00 -1.00778377e+00 -1.22413087e+00
8.24435771e-01 3.18343133e-01 1.28621590e+00 8.90363380e-03
-5.72601557e-01 9.89036202e-01 -8.03579092e-02 5.34666061e-01
-1.10460185e-01 3.41245949e-01 -5.24013698e-01 1.68578357e-01
8.48162100e-02 -6.06193006e-01 -8.09286177e-01 3.16818416e-01
-9.73717093e-01 -2.53752291e-01 5.50577581e-01 1.30613077e+00
5.82088649e-01 -9.48139906e-01 2.41088256e-01 -2.23849162e-01
4.05107319e-01 -2.19927862e-01 -4.23743755e-01 2.72587061e-01
-5.27173206e-02 2.49371991e-01 5.10633469e-01 -3.05507272e-01
-5.66390872e-01 3.23361665e-01 -2.38146186e-01 -6.60705209e-01
-2.11269230e-01 2.97396332e-01 1.45316809e-01 -6.25852585e-01
3.59551519e-01 2.57977664e-01 2.78680384e-01 -2.93311745e-01
6.82261884e-01 2.95519739e-01 9.11668837e-01 -6.37016654e-01
6.28124297e-01 7.51491904e-01 9.61427018e-02 -4.90191251e-01
-6.27461314e-01 -4.64876264e-01 -6.77257955e-01 -4.09171917e-02
9.90802824e-01 -1.07701504e+00 -9.20863628e-01 2.42184982e-01
-1.17758131e+00 -2.99473703e-01 -2.65734732e-01 2.53185540e-01
-9.92811561e-01 3.00498843e-01 -8.87899220e-01 -9.60945413e-02
-3.57221603e-01 -1.33244503e+00 1.76433134e+00 -1.00258991e-01
-5.76008707e-02 -1.23744428e+00 2.24815339e-01 4.76942092e-01
6.20033979e-01 2.07420997e-02 5.90655148e-01 -3.14980328e-01
-8.70161593e-01 -3.23335111e-01 -5.20355582e-01 9.16493014e-02
-2.82151252e-01 -8.26737881e-02 -1.11509395e+00 -4.83553350e-01
-3.59307081e-01 -7.94295788e-01 1.31161273e+00 1.34184927e-01
8.54961753e-01 1.16009146e-01 -7.14070350e-02 6.17055893e-01
1.57597208e+00 -5.03890932e-01 7.15262055e-01 2.43738696e-01
5.59737861e-01 4.22458202e-01 1.02442160e-01 -6.45483136e-02
6.69403076e-01 8.13595891e-01 5.42491317e-01 -5.25248051e-01
-4.00770873e-01 -2.87826896e-01 3.62070501e-01 5.73371649e-01
8.34685192e-02 1.25902995e-01 -6.99491501e-01 7.54991293e-01
-1.89048791e+00 -1.15933883e+00 4.11939770e-01 1.96724093e+00
4.92711097e-01 -3.03161502e-01 1.37789026e-01 -3.09625715e-01
4.30103362e-01 3.18603516e-01 -3.73595357e-01 -4.67299432e-01
-1.58318937e-01 4.77270275e-01 1.68322712e-01 8.90473604e-01
-1.36621308e+00 8.66120219e-01 6.82460880e+00 4.75143254e-01
-8.94459367e-01 3.88495356e-01 4.36009943e-01 -5.54580949e-02
-4.22524452e-01 -6.04800135e-02 -2.62500137e-01 1.33446231e-01
5.86257279e-01 3.40228587e-01 5.91322720e-01 6.11658275e-01
-4.40408707e-01 1.17202096e-01 -1.40050757e+00 9.59590971e-01
5.82376182e-01 -1.60676861e+00 2.18359560e-01 4.16470272e-03
7.61367381e-01 3.49929243e-01 3.86849076e-01 7.76817948e-02
-1.54009581e-01 -1.08473134e+00 6.76633835e-01 6.53860688e-01
7.83113956e-01 -5.92672586e-01 7.21461892e-01 -3.64207506e-01
-1.44493985e+00 -1.19909436e-01 -2.69965500e-01 1.59268469e-01
1.47658974e-01 -1.39467478e-01 -2.15352356e-01 4.17436987e-01
4.95557249e-01 1.08882606e+00 -7.96049178e-01 8.37585509e-01
9.12140906e-02 -5.09767830e-02 -1.90493852e-01 2.26834372e-01
5.69579303e-01 1.75552011e-01 3.52624714e-01 1.36931956e+00
7.95992687e-02 -3.86234730e-01 6.74825832e-02 9.30407524e-01
-3.60050082e-01 4.55917083e-02 -8.31838965e-01 7.91349709e-02
1.76899377e-02 1.27185297e+00 -2.25602999e-01 -2.99865425e-01
-8.38479459e-01 1.40200865e+00 8.30813766e-01 5.04092515e-01
-7.75154471e-01 -4.73800868e-01 5.63427448e-01 -3.26229185e-02
8.25373232e-01 -1.41086832e-01 -1.55810788e-01 -1.34371877e+00
1.22082680e-02 -8.42661917e-01 3.84819210e-01 -9.93926466e-01
-1.63702679e+00 6.59683347e-01 -4.13736373e-01 -1.14330399e+00
-6.89327940e-02 -8.91731679e-01 -4.86150205e-01 8.06796312e-01
-1.71967721e+00 -1.60506570e+00 -2.67489165e-01 8.96688640e-01
4.15496677e-01 -3.85808796e-01 1.04318762e+00 3.52342039e-01
-2.41255999e-01 9.61537361e-01 -7.01753944e-02 1.99481547e-01
7.80612111e-01 -1.14794660e+00 3.02875698e-01 4.99530017e-01
3.34897250e-01 7.92426646e-01 2.38124907e-01 6.74605891e-02
-1.80583680e+00 -7.37094939e-01 9.23719406e-01 -5.02940953e-01
9.95369852e-01 -3.95113349e-01 -7.41192818e-01 7.96518743e-01
9.12479341e-01 1.75054282e-01 7.19934464e-01 4.35264409e-02
-1.00996172e+00 -1.55682579e-01 -1.01969588e+00 5.64541221e-01
8.79368663e-01 -1.24765444e+00 -6.72065198e-01 2.86398619e-01
4.89755332e-01 -4.17356312e-01 -1.22833622e+00 2.55788505e-01
8.06107163e-01 -8.51599455e-01 1.24326706e+00 -6.44013703e-01
8.76480043e-01 -1.35669515e-01 -5.58122694e-01 -1.05587637e+00
-4.39963579e-01 -4.79431361e-01 -4.54494469e-02 7.36631215e-01
6.07331932e-01 -4.99532849e-01 6.04762971e-01 3.85372818e-01
-1.25672042e-01 -8.16457212e-01 -1.18993878e+00 -4.98897880e-01
2.38422170e-01 -7.94269890e-02 1.77745894e-01 7.37443686e-01
2.36775488e-01 3.86340767e-01 -2.77590364e-01 1.02912568e-01
7.37559915e-01 4.59534645e-01 6.79532051e-01 -7.29233146e-01
-3.84412467e-01 -8.23659599e-01 -6.60546124e-01 -1.24588811e+00
4.31692302e-01 -1.15963447e+00 -1.50873780e-01 -1.33407664e+00
5.28806388e-01 2.52145618e-01 -4.26531404e-01 4.45143282e-01
-3.52367721e-02 9.20582116e-01 2.97925383e-01 -4.23915684e-03
-8.93658340e-01 7.60351181e-01 1.10281372e+00 -4.74777609e-01
2.95999348e-01 -6.58628464e-01 -4.40770864e-01 1.59722507e-01
9.69214201e-01 -1.97942629e-01 -1.65904716e-01 -8.53317380e-01
1.75861642e-01 -1.22727476e-01 8.99252713e-01 -7.09900558e-01
3.98419261e-01 3.13158721e-01 1.96037784e-01 -2.52496868e-01
5.83051980e-01 -8.42427969e-01 -3.23049843e-01 2.75006294e-01
-7.36031890e-01 3.11897546e-01 3.54037344e-01 4.72887844e-01
-5.61347187e-01 1.41264841e-01 7.91961133e-01 -5.28585911e-02
-6.32522583e-01 4.71466392e-01 -3.32851589e-01 1.38371065e-01
6.16800725e-01 -8.08923319e-02 -3.82578105e-01 -4.32161003e-01
-8.53741050e-01 1.11479253e-01 7.62315989e-01 4.29798007e-01
8.23217750e-01 -1.67674303e+00 -7.04249382e-01 3.63978297e-01
4.70243216e-01 -6.96567833e-01 3.18637848e-01 1.09344161e+00
-4.11411554e-01 5.03158629e-01 -3.41161668e-01 -7.61272490e-01
-1.10322416e+00 7.17375159e-01 6.71457708e-01 -1.40195787e-01
-5.62885225e-01 9.65946555e-01 1.33497521e-01 -5.66028893e-01
3.13387424e-01 -2.88023740e-01 2.68964469e-01 -1.47527546e-01
4.33112711e-01 5.45642152e-02 -4.13025767e-02 -8.22573245e-01
-3.90046895e-01 1.01980233e+00 -2.99657248e-02 -2.66353428e-01
1.18817949e+00 -1.19471423e-01 -3.43250602e-01 2.36312538e-01
1.97739708e+00 -3.02899748e-01 -1.22573435e+00 -3.48994315e-01
-3.86129737e-01 -3.49859208e-01 -2.50049029e-02 -6.32896185e-01
-1.10524976e+00 1.31293476e+00 5.82613826e-01 -1.36662111e-01
1.12248743e+00 3.35556418e-01 9.50268865e-01 5.70021391e-01
1.73754111e-01 -8.38941813e-01 5.39718091e-01 3.41918230e-01
1.24861026e+00 -1.62411523e+00 -3.20726305e-01 1.79102853e-01
-5.58208108e-01 1.20817518e+00 3.99723232e-01 -6.13427520e-01
5.15840113e-01 1.09867245e-01 -6.47329018e-02 -2.24086165e-01
-9.55216587e-01 -4.38217729e-01 6.43811882e-01 4.16244239e-01
5.32350421e-01 -2.81346709e-01 1.03398569e-01 -1.09021708e-01
3.03717464e-01 -1.92226410e-01 -4.12450656e-02 9.24762905e-01
-2.68251956e-01 -1.03374982e+00 -2.66137198e-02 -6.16659895e-02
-3.16443801e-01 -3.69086266e-01 -4.64533597e-01 7.29281127e-01
-1.68498874e-01 6.17772758e-01 2.41957381e-01 -9.86010879e-02
4.05942619e-01 1.43562421e-01 1.01388597e+00 -2.20708206e-01
-8.14626634e-01 -7.18054920e-03 -1.69877216e-01 -9.99688983e-01
-6.80051625e-01 -4.43914294e-01 -9.67236519e-01 -1.64810926e-01
1.99855343e-01 -2.43223161e-01 4.44983691e-01 4.90986705e-01
6.75817311e-01 3.08936417e-01 6.68287754e-01 -1.28521264e+00
-2.64808118e-01 -6.89083397e-01 -1.75459206e-01 7.72957444e-01
7.15781391e-01 -5.24493098e-01 -1.89101294e-01 1.05910450e-01] | [10.728009223937988, 1.3421988487243652] |
94eb3162-fbb6-442e-a829-2f43fd314fb1 | diverse-difficult-and-odd-instances-d2o-a-new | 2301.12527 | null | https://arxiv.org/abs/2301.12527v1 | https://arxiv.org/pdf/2301.12527v1.pdf | Diverse, Difficult, and Odd Instances (D2O): A New Test Set for Object Classification | Test sets are an integral part of evaluating models and gauging progress in object recognition, and more broadly in computer vision and AI. Existing test sets for object recognition, however, suffer from shortcomings such as bias towards the ImageNet characteristics and idiosyncrasies (e.g., ImageNet-V2), being limited to certain types of stimuli (e.g., indoor scenes in ObjectNet), and underestimating the model performance (e.g., ImageNet-A). To mitigate these problems, we introduce a new test set, called D2O, which is sufficiently different from existing test sets. Images are a mix of generated images as well as images crawled from the web. They are diverse, unmodified, and representative of real-world scenarios and cause state-of-the-art models to misclassify them with high confidence. To emphasize generalization, our dataset by design does not come paired with a training set. It contains 8,060 images spread across 36 categories, out of which 29 appear in ImageNet. The best Top-1 accuracy on our dataset is around 60% which is much lower than 91% best Top-1 accuracy on ImageNet. We find that popular vision APIs perform very poorly in detecting objects over D2O categories such as ``faces'', ``cars'', and ``cats''. Our dataset also comes with a ``miscellaneous'' category, over which we test the image tagging models. Overall, our investigations demonstrate that the D2O test set contain a mix of images with varied levels of difficulty and is predictive of the average-case performance of models. It can challenge object recognition models for years to come and can spur more research in this fundamental area. | ['Ali Borji'] | 2023-01-29 | null | null | null | null | ['object-recognition', 'miscellaneous'] | ['computer-vision', 'miscellaneous'] | [ 2.31311873e-01 -1.89354256e-01 -1.28237903e-01 -4.67748761e-01
-5.97098291e-01 -8.62494051e-01 9.21390772e-01 -9.23743024e-02
-5.53889930e-01 3.46146077e-01 -1.55028939e-01 -4.82891649e-01
-1.31894842e-01 -7.01674104e-01 -8.57389271e-01 -4.97970670e-01
-4.96761352e-02 4.26780313e-01 4.99176115e-01 -1.51760340e-01
2.23400101e-01 4.60944593e-01 -1.82787108e+00 5.21456838e-01
5.34308016e-01 1.21526909e+00 2.85034299e-01 5.41467369e-01
-4.88921767e-03 7.75526464e-01 -8.65947843e-01 -6.93213701e-01
2.67299831e-01 -2.57657468e-01 -7.01354146e-01 1.30471051e-01
1.15049613e+00 -3.31152320e-01 -3.44092160e-01 1.22286367e+00
3.79179686e-01 -2.87560642e-01 7.53886104e-01 -1.53782129e+00
-9.27370727e-01 2.83686727e-01 -3.11919630e-01 4.22023952e-01
1.74006388e-01 5.80724001e-01 1.01111603e+00 -9.91012275e-01
7.21489847e-01 1.19480276e+00 8.78470123e-01 6.22229338e-01
-1.38138354e+00 -8.13409686e-01 1.79648504e-01 1.38994470e-01
-1.38401425e+00 -5.51861107e-01 2.85025090e-01 -7.11024225e-01
8.12075734e-01 5.58742285e-01 3.97738755e-01 1.23880982e+00
2.44785249e-02 6.59621239e-01 1.23158550e+00 -2.29756296e-01
1.21130198e-01 3.52989167e-01 1.78974435e-01 4.92689908e-01
5.17601192e-01 2.29949087e-01 -1.99169338e-01 -7.29570910e-02
5.38064539e-01 -2.51279864e-02 -2.26495385e-01 -3.80869299e-01
-1.24340427e+00 5.47079802e-01 5.25313199e-01 2.18863323e-01
-1.14000164e-01 8.28346983e-02 2.46331617e-01 2.94787586e-01
2.06327692e-01 6.21279716e-01 -4.54059154e-01 -5.58398031e-02
-5.23656189e-01 3.83232802e-01 7.96498060e-01 1.02036059e+00
7.65248835e-01 1.04754753e-01 -3.27254646e-02 1.18346488e+00
3.64771523e-02 5.35366893e-01 5.18431783e-01 -9.46757913e-01
4.86684740e-01 4.80025530e-01 1.48659915e-01 -1.01560843e+00
-2.29926705e-01 -4.53536272e-01 -4.53991115e-01 3.90113711e-01
7.06047177e-01 1.52311400e-01 -1.37450552e+00 1.82696891e+00
-8.90507028e-02 9.33168232e-02 -1.34944543e-01 1.00219417e+00
1.11101449e+00 4.28648263e-01 3.30442607e-01 3.88640940e-01
1.43169689e+00 -8.00424397e-01 -1.51542518e-02 -5.49438894e-01
5.45805216e-01 -7.82845795e-01 1.20490277e+00 2.87953526e-01
-9.78395760e-01 -7.88681090e-01 -9.06699479e-01 4.37608242e-01
-8.19902658e-01 -1.11648053e-01 7.11666703e-01 8.42062175e-01
-1.05351698e+00 2.53094107e-01 -2.46956542e-01 -6.78157330e-01
5.86118281e-01 1.79638177e-01 -5.78112304e-01 -4.35079277e-01
-8.46402586e-01 1.19304729e+00 4.00323153e-01 -2.21916601e-01
-1.26447058e+00 -6.94363117e-01 -6.81908369e-01 -1.45274088e-01
3.74995261e-01 -2.27878556e-01 1.17488337e+00 -1.12538064e+00
-4.47769672e-01 1.30383670e+00 1.74866959e-01 -4.16418821e-01
4.86465573e-01 1.23426504e-01 -8.05481017e-01 3.32775414e-02
2.75325984e-01 1.02215660e+00 6.55596972e-01 -1.67268610e+00
-7.48012125e-01 -8.90456215e-02 4.25669074e-01 -4.70074303e-02
-2.18319982e-01 1.72347248e-01 -5.29633403e-01 -5.45615435e-01
-3.41407806e-02 -1.02839398e+00 -8.05832595e-02 -1.12539669e-02
-1.32660687e-01 -1.29273683e-01 7.95654118e-01 -1.51886627e-01
9.60186541e-01 -2.19630504e+00 -5.64393103e-01 6.98151737e-02
2.33693138e-01 4.98387724e-01 -5.45969427e-01 2.89562702e-01
-3.33138764e-01 6.44223392e-01 -2.83581056e-02 8.32538903e-02
2.78782360e-02 3.06788951e-01 -2.68864334e-01 3.84061664e-01
5.37542403e-01 8.29731703e-01 -9.82032835e-01 -4.13057446e-01
3.19439381e-01 7.77545720e-02 -3.99752200e-01 -2.83812415e-02
-1.97529942e-01 6.04670420e-02 -2.27752477e-01 9.15806115e-01
6.70345366e-01 -2.88948029e-01 -1.67680070e-01 -1.85502410e-01
-9.80208963e-02 1.26315042e-01 -8.81314337e-01 1.12877202e+00
-1.03022598e-01 7.99857616e-01 -1.36911064e-01 -9.47233856e-01
7.59906828e-01 6.87138289e-02 2.99810618e-01 -8.33380103e-01
-2.77542155e-02 2.06870213e-01 5.02823293e-01 -5.16642332e-01
2.91278481e-01 6.66652992e-02 8.06667283e-02 2.67405868e-01
1.93670973e-01 -2.58443654e-01 3.48054081e-01 9.05976892e-02
1.37345171e+00 -1.68025970e-01 1.84879396e-02 -2.36141518e-01
1.15422480e-01 3.76209468e-01 4.86365080e-01 1.32233465e+00
-5.33067524e-01 8.83714139e-01 3.47942680e-01 -4.88696575e-01
-1.29724848e+00 -1.20164835e+00 -6.30695999e-01 1.00084960e+00
1.77967817e-01 -3.17054987e-01 -5.88074923e-01 -7.40244508e-01
2.60884255e-01 5.64532697e-01 -6.20861173e-01 -2.24706843e-01
-2.25057751e-01 -8.60350013e-01 8.04998398e-01 5.90477228e-01
6.57310069e-01 -1.14842486e+00 -2.95838892e-01 4.54602987e-02
-8.22887346e-02 -1.19962668e+00 -1.48063049e-01 2.17831686e-01
-3.82193029e-01 -1.43013084e+00 -5.17296970e-01 -8.29600155e-01
7.28950083e-01 6.52300000e-01 1.78809369e+00 3.39851111e-01
-3.02088946e-01 6.37598276e-01 -4.39804167e-01 -7.03224778e-01
-4.99570221e-01 -2.21925095e-01 -4.90248613e-02 -9.26542655e-03
6.92459404e-01 -2.07140639e-01 -4.88421232e-01 9.60740507e-01
-1.05477607e+00 -1.89298213e-01 8.78146768e-01 9.00609553e-01
2.92999297e-01 2.62417197e-01 4.61105883e-01 -9.88587499e-01
2.74373710e-01 -8.93559992e-01 -3.63565594e-01 2.75039732e-01
-4.46294904e-01 -5.21555901e-01 5.48680007e-01 -8.62349808e-01
-6.71318054e-01 -2.87107587e-01 2.56356113e-02 -4.97076809e-01
-5.51731884e-01 4.30048376e-01 -1.02601215e-01 -2.41314754e-01
9.67359900e-01 9.03239995e-02 -1.21171564e-01 -4.05095696e-01
-6.41694069e-02 6.32887840e-01 6.29081309e-01 -7.05202818e-01
9.57904339e-01 3.00097913e-01 -4.56984490e-01 -9.05545890e-01
-7.60037482e-01 -4.79494333e-01 -2.26500317e-01 -3.85998130e-01
7.07766056e-01 -9.93939757e-01 -5.47322273e-01 5.91910005e-01
-8.22041750e-01 -3.84353250e-01 1.94702353e-02 4.94390547e-01
-2.45539114e-01 -6.29713088e-02 -4.28565651e-01 -6.83227360e-01
3.61003667e-01 -1.27141881e+00 9.50906873e-01 7.85761774e-02
-2.04404682e-01 -7.20693171e-01 -4.51427460e-01 5.33432066e-01
3.97176266e-01 2.80899405e-01 8.01030040e-01 -7.76495695e-01
-6.78553045e-01 -2.82714874e-01 -6.06232703e-01 5.78032255e-01
-7.24186376e-03 2.13636741e-01 -1.18245685e+00 -3.74964744e-01
-2.17189059e-01 -7.20320880e-01 8.24750245e-01 -6.97589219e-02
1.27558613e+00 -2.53001601e-01 -3.49226117e-01 2.57894248e-01
1.49866092e+00 4.23123300e-01 7.94925988e-01 4.03117508e-01
4.95092422e-01 5.34955740e-01 6.35923684e-01 7.36211613e-02
3.02519947e-01 7.37933815e-01 6.88835442e-01 -1.06598854e-01
-3.06598008e-01 -1.63602516e-01 1.95488468e-01 3.89127284e-01
-9.55660362e-03 -3.84878099e-01 -1.37669897e+00 7.60832667e-01
-1.51376081e+00 -1.07845306e+00 -1.09320454e-01 2.14691710e+00
6.64754331e-01 5.99944532e-01 1.95735812e-01 -4.97798659e-02
8.66162002e-01 9.17033777e-02 -4.81351465e-01 -1.68284759e-01
-1.52608573e-01 8.29104781e-02 5.99591434e-01 1.03147112e-01
-1.27997303e+00 8.15824151e-01 7.09870338e+00 7.78524101e-01
-1.16431081e+00 -3.61340493e-02 6.47422850e-01 1.79226816e-01
-1.15004480e-01 -2.40511484e-02 -8.41268182e-01 8.33238006e-01
8.96595955e-01 2.22791165e-01 2.95561433e-01 1.08104813e+00
-3.15940470e-01 -7.48048723e-02 -1.07811213e+00 1.03390026e+00
3.11905950e-01 -1.17599678e+00 1.15911618e-01 1.25444606e-01
6.27533317e-01 4.51260775e-01 1.46186084e-01 4.84464198e-01
6.46280706e-01 -1.28993964e+00 1.02832699e+00 1.37191251e-01
9.70066667e-01 -3.42310697e-01 6.99915946e-01 2.88725495e-01
-8.90553653e-01 -1.95599928e-01 -6.25218332e-01 4.88943513e-03
-4.60657597e-01 3.30437988e-01 -7.24222481e-01 1.44770918e-02
1.12179053e+00 5.59805810e-01 -1.09410775e+00 1.22532618e+00
1.62803710e-01 7.32748508e-01 -3.36729348e-01 -5.95955588e-02
4.95864332e-01 1.06005393e-01 2.85230190e-01 1.23228800e+00
1.03952698e-02 5.11920638e-02 2.43904069e-01 7.18617380e-01
-2.00319320e-01 -3.76658082e-01 -1.02965796e+00 -5.10924384e-02
5.21249115e-01 1.11314702e+00 -7.98104048e-01 -3.59410644e-01
-7.79293001e-01 4.53630447e-01 2.20961526e-01 4.92479324e-01
-1.00642109e+00 -6.49576634e-02 8.66519928e-01 1.05619006e-01
1.92979515e-01 -2.75386032e-02 -6.10289276e-02 -1.09348452e+00
1.54312983e-01 -1.23484993e+00 1.66024223e-01 -9.27060544e-01
-1.53736091e+00 6.92052066e-01 1.93514928e-01 -1.31277049e+00
9.52940732e-02 -1.08187509e+00 -4.39341158e-01 5.71706474e-01
-1.22856939e+00 -1.05694568e+00 -7.07599282e-01 4.60509568e-01
3.32139283e-01 -2.73414135e-01 7.18298852e-01 5.62786222e-01
-4.24084693e-01 5.72914720e-01 -1.45082863e-03 5.10341525e-01
7.04867780e-01 -1.18451202e+00 5.53561211e-01 7.13892102e-01
3.53841543e-01 6.14095867e-01 6.02970362e-01 -3.49747449e-01
-1.20949388e+00 -1.23944628e+00 5.92766225e-01 -9.88994539e-01
7.28777587e-01 -3.62528116e-01 -8.75424683e-01 7.30026484e-01
-2.31730744e-01 3.05779845e-01 5.09031415e-01 8.06986466e-02
-7.10074723e-01 -2.87080079e-01 -1.17806756e+00 6.80534124e-01
1.31055629e+00 -6.53856754e-01 -5.72021425e-01 3.55279505e-01
3.44233185e-01 -2.73401171e-01 -8.82823467e-01 6.04286909e-01
7.60430694e-01 -1.14459217e+00 1.16039145e+00 -9.41298306e-01
4.92083281e-01 -2.07408488e-01 -3.90572071e-01 -1.20177400e+00
-5.40162563e-01 7.19511062e-02 3.47733080e-01 1.43421531e+00
5.01449525e-01 -7.46238232e-01 8.48018229e-01 7.79770970e-01
-2.43748903e-01 -6.38848245e-01 -6.30488992e-01 -1.09931779e+00
2.29002833e-02 -7.15667307e-01 5.81251442e-01 1.08343303e+00
-6.59461617e-01 1.51241317e-01 5.31903207e-02 1.20002283e-02
5.22259891e-01 -1.80770233e-01 1.02023590e+00 -1.18136179e+00
-1.55061111e-01 -7.17487812e-01 -8.94200563e-01 -8.93694699e-01
-1.28505528e-01 -8.49217951e-01 2.54147518e-02 -1.18262279e+00
2.90698320e-01 -1.10593534e+00 -4.55828041e-01 5.71464002e-01
-5.13365753e-02 8.37155938e-01 3.29412550e-01 3.94770712e-01
-4.40898687e-01 -1.09262824e-01 1.17771590e+00 -5.45173228e-01
4.95055884e-01 -2.32628569e-01 -8.66133332e-01 7.97075510e-01
6.39048517e-01 -4.18988436e-01 -4.42839473e-01 -6.24792576e-01
-5.70356734e-02 -3.25151145e-01 8.42260838e-01 -1.31145358e+00
-6.73367158e-02 -3.40019643e-01 6.42098665e-01 -3.38715285e-01
4.08469319e-01 -9.77700770e-01 2.46939838e-01 3.51052612e-01
-2.38806218e-01 8.96016732e-02 2.29922503e-01 2.52464503e-01
-2.82697737e-01 -3.67680311e-01 9.26165581e-01 -4.59865242e-01
-1.22298884e+00 3.21542293e-01 -1.06836475e-01 3.45586360e-01
1.01858413e+00 -5.17138898e-01 -8.75578880e-01 -1.94793835e-01
-3.51881504e-01 4.74823937e-02 7.08734930e-01 7.01122642e-01
4.17036027e-01 -1.30637455e+00 -6.91175103e-01 2.32072696e-01
7.19368696e-01 -2.61195272e-01 -1.66140452e-01 4.93030101e-01
-4.73426193e-01 4.47275639e-01 -3.52155656e-01 -8.71790588e-01
-1.02056277e+00 4.46585953e-01 3.02076221e-01 2.34330654e-01
-3.96345824e-01 7.65916288e-01 5.08801937e-01 -3.33619773e-01
2.50700861e-01 -1.93420455e-01 -1.61417462e-02 -1.80468019e-02
4.38425928e-01 1.44342892e-02 2.16331091e-02 -7.79178858e-01
-4.89038199e-01 2.65132695e-01 -1.34272560e-01 2.88552701e-01
1.08855712e+00 3.34252924e-01 5.73873743e-02 3.48695755e-01
1.20342660e+00 -2.43802696e-01 -1.05397725e+00 -5.68901328e-03
1.07339248e-01 -7.83957779e-01 -3.45860183e-01 -1.08642268e+00
-9.04275894e-01 5.13904274e-01 7.11180210e-01 5.68185806e-01
8.59801710e-01 1.94341511e-01 4.12162453e-01 4.25955981e-01
6.56413794e-01 -8.60273778e-01 1.55316114e-01 3.71416628e-01
8.40706587e-01 -1.63974047e+00 -1.73805296e-01 -2.74381846e-01
-4.28019315e-01 6.43493116e-01 1.12446690e+00 3.06067038e-02
5.30437887e-01 1.70454860e-01 2.07646042e-01 -1.61255971e-01
-7.95621097e-01 -4.17941868e-01 2.52005696e-01 8.02051246e-01
2.41901815e-01 9.19977948e-02 -4.29800898e-02 2.46525392e-01
-2.07393318e-01 -1.55170828e-01 3.14510643e-01 7.84317493e-01
-4.61816162e-01 -8.79473746e-01 -5.55067003e-01 8.33223879e-01
-4.82465357e-01 -8.57305080e-02 -6.10959113e-01 1.21061110e+00
5.68790615e-01 9.60367024e-01 1.84782818e-01 -6.76022351e-01
4.69297707e-01 -5.03025651e-02 4.48212713e-01 -6.50178075e-01
-5.44589758e-01 -5.34425199e-01 3.60804915e-01 -4.28703368e-01
-3.60162467e-01 -4.39127505e-01 -4.77709621e-01 -4.92410183e-01
-3.05193007e-01 -2.71261930e-02 6.23153210e-01 7.10224271e-01
6.57839179e-02 2.16256514e-01 3.84623855e-01 -8.31614494e-01
-3.96719456e-01 -1.00301909e+00 -3.68361890e-01 1.07500422e+00
1.11675993e-01 -9.75264728e-01 -3.50528777e-01 2.85778362e-02] | [9.734122276306152, 2.23158597946167] |
a4824820-0ac3-4f74-8c44-3e1200816bc3 | detrex-benchmarking-detection-transformers | 2306.07265 | null | https://arxiv.org/abs/2306.07265v2 | https://arxiv.org/pdf/2306.07265v2.pdf | detrex: Benchmarking Detection Transformers | The DEtection TRansformer (DETR) algorithm has received considerable attention in the research community and is gradually emerging as a mainstream approach for object detection and other perception tasks. However, the current field lacks a unified and comprehensive benchmark specifically tailored for DETR-based models. To address this issue, we develop a unified, highly modular, and lightweight codebase called detrex, which supports a majority of the mainstream DETR-based instance recognition algorithms, covering various fundamental tasks, including object detection, segmentation, and pose estimation. We conduct extensive experiments under detrex and perform a comprehensive benchmark for DETR-based models. Moreover, we enhance the performance of detection transformers through the refinement of training hyper-parameters, providing strong baselines for supported algorithms.We hope that detrex could offer research communities a standardized and unified platform to evaluate and compare different DETR-based models while fostering a deeper understanding and driving advancements in DETR-based instance recognition. Our code is available at https://github.com/IDEA-Research/detrex. The project is currently being actively developed. We encourage the community to use detrex codebase for further development and contributions. | ['Lei Zhang', 'Jianwei Yang', 'Yuhui Yuan', 'Xianbiao Qi', 'Zhaoyang Zeng', 'Jianan Wang', 'He Cao', 'Hongyang Li', 'Ding Jia', 'Xingyu Liao', 'Jie Yang', 'Ailing Zeng', 'Hao Zhang', 'Feng Li', 'Shilong Liu', 'Tianhe Ren'] | 2023-06-12 | null | null | null | null | ['pose-estimation'] | ['computer-vision'] | [ 1.23080038e-01 -3.51232767e-01 -3.77791315e-01 -4.17043924e-01
-7.93518603e-01 -6.79478467e-01 6.42479122e-01 5.70640974e-02
-1.95536226e-01 3.28299738e-02 -1.72212258e-01 -1.45982638e-01
3.11108768e-01 -6.56465113e-01 -6.34739339e-01 -4.93745208e-01
2.11134315e-01 3.62818301e-01 6.07001483e-01 1.63693234e-01
2.22301796e-01 4.28579003e-01 -1.95327282e+00 4.95590419e-01
5.65766633e-01 1.08832312e+00 4.44551677e-01 5.28003931e-01
4.94909696e-02 5.11764407e-01 -3.22029650e-01 -4.54250664e-01
2.20836133e-01 -1.14130020e-01 -6.87146544e-01 1.52469352e-01
8.30068111e-01 -4.16605175e-02 -9.77571756e-02 8.81405115e-01
5.86192846e-01 -1.24841556e-01 4.27666634e-01 -1.30789399e+00
-6.56047463e-01 5.26175618e-01 -8.25398147e-01 2.24050879e-01
1.70674145e-01 4.35945064e-01 1.15866899e+00 -1.17392850e+00
5.43511987e-01 1.05825937e+00 8.53748858e-01 4.83885437e-01
-1.17881119e+00 -6.45669401e-01 4.24365401e-01 4.04777169e-01
-1.30286205e+00 -6.01878881e-01 5.68047404e-01 -5.16964138e-01
8.68681014e-01 5.70296586e-01 6.14296317e-01 1.27447045e+00
-3.42305243e-01 1.47993600e+00 1.09827018e+00 -2.95041293e-01
8.76124501e-02 2.85988569e-01 2.28744641e-01 7.13667095e-01
3.62263381e-01 -2.39757504e-02 -4.92359370e-01 8.08624551e-02
5.40319502e-01 3.67788114e-02 -1.55831784e-01 -6.08350277e-01
-1.26967335e+00 4.45198178e-01 3.67190182e-01 2.52394319e-01
-1.57054961e-01 1.10599093e-01 3.34540427e-01 -5.70342615e-02
3.91509563e-01 3.73837769e-01 -3.56696427e-01 -3.14033002e-01
-9.69295084e-01 4.64835048e-01 7.20054507e-01 9.20455754e-01
4.31185156e-01 -3.01997632e-01 -3.20219398e-01 9.90896821e-01
3.44944865e-01 3.90942842e-01 2.38048509e-01 -1.02046764e+00
2.85025597e-01 8.17231297e-01 1.79594323e-01 -8.35004270e-01
-1.79352313e-01 -8.31710458e-01 -4.18767899e-01 1.34008989e-01
4.33479488e-01 1.87499315e-01 -6.09089971e-01 1.44680202e+00
5.97810090e-01 4.30003256e-01 -4.37715471e-01 1.09306240e+00
9.88797665e-01 4.17821556e-01 -9.42903906e-02 3.25612664e-01
1.55921149e+00 -1.26912272e+00 -4.43897732e-02 -5.54818511e-01
5.49580216e-01 -9.50135171e-01 1.23680925e+00 5.14784336e-01
-1.02567196e+00 -6.35218859e-01 -1.03356552e+00 -1.47432581e-01
-3.29779804e-01 4.55116451e-01 7.44212568e-01 7.03243196e-01
-7.42938876e-01 2.65984625e-01 -1.17707670e+00 -6.95677519e-01
6.30656183e-01 7.30558336e-02 -2.09039003e-01 -1.74804658e-01
-5.68127096e-01 6.35609984e-01 1.58048570e-01 1.62326381e-01
-8.69475484e-01 -6.64477587e-01 -7.20468998e-01 -2.65257150e-01
5.89440525e-01 -6.55959845e-01 1.53217578e+00 -5.20489097e-01
-1.03379703e+00 1.25008309e+00 1.08477809e-02 -4.76992190e-01
5.78128934e-01 -3.70503634e-01 -1.03818431e-01 -2.01773435e-01
9.89262238e-02 7.55829275e-01 6.86613560e-01 -1.13539660e+00
-7.88944006e-01 -4.94400680e-01 2.06042342e-02 1.00262292e-01
-4.43977565e-01 1.61377594e-01 -9.40035641e-01 -5.36261678e-01
3.67758423e-02 -1.05515110e+00 -4.83005717e-02 3.49301457e-01
-5.27775168e-01 -2.60089755e-01 9.40921903e-01 -1.36758387e-01
1.03277099e+00 -2.26409459e+00 1.19200861e-03 -3.67401570e-01
3.68558615e-01 5.31683564e-01 -2.19053328e-01 2.82868057e-01
1.89133689e-01 -1.28589362e-01 -2.53642857e-01 -6.70402646e-01
1.93172008e-01 -8.39309618e-02 -3.56693655e-01 5.41395307e-01
1.72766209e-01 9.98222888e-01 -8.66172731e-01 -2.91613668e-01
4.79231238e-01 4.95131552e-01 -4.93332565e-01 1.76369175e-01
-5.03089786e-01 3.02334875e-01 -5.70402980e-01 9.08466220e-01
8.82588089e-01 -5.07556140e-01 3.65471356e-02 -2.82126546e-01
-3.62655967e-01 2.01480389e-01 -1.34955418e+00 1.52835429e+00
-2.34478295e-01 8.37297678e-01 2.39119694e-01 -9.65201318e-01
6.71767950e-01 -1.00876339e-01 4.77216631e-01 -4.61649895e-01
5.01188561e-02 2.32073054e-01 -2.12273374e-01 -4.80802745e-01
6.36582494e-01 2.37460911e-01 1.79359332e-01 5.03233552e-01
-1.51806965e-01 -7.12829977e-02 4.13414776e-01 1.56458005e-01
9.75370049e-01 4.52108175e-01 1.68886408e-01 1.06282130e-01
3.28203529e-01 9.82332230e-02 3.18540871e-01 8.32238972e-01
-4.75843012e-01 8.38796735e-01 1.61088556e-01 -4.62230116e-01
-6.38113379e-01 -9.66466308e-01 -5.91274083e-01 1.17913485e+00
1.84250548e-01 -6.36721194e-01 -6.65599287e-01 -7.06131220e-01
2.62315214e-01 2.69361287e-01 -5.56920469e-01 2.67879158e-01
-3.27538997e-01 -1.09735036e+00 5.82988977e-01 7.42924809e-01
7.02274561e-01 -9.44097698e-01 -5.89438677e-01 1.55384257e-01
-3.70399863e-01 -1.20105994e+00 -4.22506362e-01 -5.17994389e-02
-7.18218029e-01 -1.21170735e+00 -6.69507504e-01 -7.20506072e-01
2.47459039e-01 7.04566956e-01 1.15853834e+00 1.35069683e-01
-6.53607130e-01 6.41928494e-01 -3.62168610e-01 -7.01660156e-01
3.09856180e-02 3.17460567e-01 -1.95474118e-01 3.96456048e-02
4.87668931e-01 -3.20768267e-01 -9.60818172e-01 7.86980331e-01
-7.75336504e-01 1.55117288e-01 3.27305496e-01 3.10448885e-01
6.72875464e-01 -5.46454370e-01 3.54342997e-01 -8.84705305e-01
2.71649659e-01 -4.31328595e-01 -9.03730869e-01 3.53150219e-02
-6.44465864e-01 -2.18712911e-01 -1.86232403e-02 -2.89990336e-01
-9.15570080e-01 6.59914017e-02 -4.27188814e-01 -9.96874645e-02
-2.70794660e-01 3.98223341e-01 3.91393378e-02 -2.33338952e-01
8.59571338e-01 3.01580369e-01 -1.80814356e-01 -8.27945292e-01
3.41220140e-01 7.41704106e-01 5.07432640e-01 -5.90061843e-01
5.80496311e-01 5.98726451e-01 -3.52351397e-01 -7.62419164e-01
-1.16717505e+00 -8.33444417e-01 -2.41072983e-01 -2.22758234e-01
6.02499366e-01 -1.01152635e+00 -6.23676300e-01 6.84531093e-01
-9.71056998e-01 -4.51135933e-01 -1.33861348e-01 2.66439050e-01
-3.62117022e-01 3.33454400e-01 -6.29762530e-01 -7.60220289e-01
-9.02505592e-02 -1.25301707e+00 1.39326787e+00 2.34641850e-01
-1.95149332e-02 -7.98758090e-01 2.30242476e-01 8.65151525e-01
4.53733653e-01 1.04025707e-01 2.75301069e-01 -3.77275407e-01
-9.81194973e-01 -3.30467016e-01 -4.24550354e-01 3.13393325e-01
-1.87174007e-01 3.55079263e-01 -1.26215601e+00 -2.87816644e-01
-1.83330074e-01 -3.64689410e-01 1.18486524e+00 4.37141031e-01
1.33671987e+00 2.00124547e-01 -7.87708938e-01 6.09321058e-01
1.29126430e+00 -2.69874871e-01 4.59036946e-01 5.76527953e-01
7.12071598e-01 4.15860742e-01 6.69658005e-01 4.29417342e-01
7.77486145e-01 1.01524854e+00 6.44295216e-01 -1.77555054e-01
-4.09488291e-01 1.23762503e-01 3.28762382e-01 5.31543672e-01
-2.09894434e-01 -1.87791780e-01 -1.06156850e+00 6.03459597e-01
-1.95275354e+00 -1.01310778e+00 -4.69470859e-01 2.19139910e+00
7.08464503e-01 1.23870529e-01 4.02455926e-01 -5.79059012e-02
7.08950281e-01 2.18442768e-01 -5.05937099e-01 1.57303423e-01
9.30303037e-02 -7.79470578e-02 1.60897419e-01 9.81059074e-02
-1.38969648e+00 9.13435400e-01 6.14060593e+00 7.71442652e-01
-1.32945156e+00 6.37259558e-02 5.93260169e-01 -1.58098221e-01
1.54118523e-01 -1.49452493e-01 -1.19992220e+00 3.40262979e-01
5.69367886e-01 -1.12308096e-03 1.90399304e-01 1.18551719e+00
1.05221964e-01 -8.75675455e-02 -1.24701309e+00 9.84626114e-01
-3.93954962e-02 -1.49957144e+00 -3.16869378e-01 -7.63344066e-03
6.57752156e-01 5.98804712e-01 3.62877876e-01 5.57767093e-01
1.01240203e-01 -7.13052332e-01 7.91695058e-01 1.73062786e-01
5.80605865e-01 -1.98534772e-01 5.24768114e-01 3.46425354e-01
-1.40588927e+00 3.00307870e-02 -3.24482828e-01 -4.62761484e-02
-5.69147393e-02 6.79577351e-01 -6.99480474e-01 3.74493629e-01
8.44349027e-01 9.87605631e-01 -8.78416657e-01 1.53891838e+00
7.38967676e-03 8.56560290e-01 -2.73229867e-01 1.48003966e-01
8.41198787e-02 7.55232796e-02 6.94122314e-01 1.48066843e+00
8.58589932e-02 -3.36110622e-01 5.10542274e-01 9.18081343e-01
-1.96800366e-01 6.95241839e-02 -3.53346407e-01 -5.46773374e-02
4.19835120e-01 1.51573396e+00 -9.27603781e-01 -1.97317258e-01
-6.39198244e-01 7.34389186e-01 2.31309503e-01 1.35707468e-01
-1.05726540e+00 -1.75553843e-01 8.43491018e-01 3.66437942e-01
4.73784685e-01 -2.83841491e-01 -3.46072376e-01 -1.35572577e+00
2.28707835e-01 -1.06358480e+00 4.04924750e-01 -6.83281481e-01
-1.32529330e+00 5.05086720e-01 9.98518541e-02 -1.17952752e+00
1.17070749e-01 -8.84495378e-01 -5.32647908e-01 4.52278852e-01
-1.66089463e+00 -1.20607746e+00 -6.10977948e-01 2.76848882e-01
6.50162160e-01 1.36698604e-01 6.63592696e-01 5.16753018e-01
-1.17794931e+00 4.90326285e-01 -1.04344562e-01 9.45617408e-02
8.84407222e-01 -1.19151413e+00 6.62564337e-01 7.66537786e-01
4.97600436e-01 7.88873076e-01 7.96279013e-01 -3.06955665e-01
-1.62296748e+00 -1.26701891e+00 4.20080394e-01 -9.52415884e-01
6.46227658e-01 -4.66151416e-01 -7.57414222e-01 8.76915812e-01
-1.60422083e-02 3.23770404e-01 5.79154313e-01 4.27279890e-01
-4.41007584e-01 -2.78365791e-01 -8.67422640e-01 4.90802377e-01
1.02810049e+00 -4.80511636e-01 -3.09006512e-01 4.87927526e-01
4.15957510e-01 -5.52917242e-01 -6.25164211e-01 3.83205533e-01
8.03855896e-01 -1.25407863e+00 1.11207592e+00 2.38668136e-02
3.00412238e-01 -4.29915726e-01 -8.15785676e-02 -8.21316421e-01
-3.04902494e-01 -3.95689905e-01 -4.66630042e-01 1.25100112e+00
4.16497737e-01 -6.99532330e-01 1.09670806e+00 3.77317667e-01
-1.52294308e-01 -1.01362407e+00 -5.15128255e-01 -8.70409787e-01
-2.80675232e-01 -1.03682303e+00 5.15218675e-01 6.23244703e-01
-2.38099515e-01 2.68951714e-01 -5.76040670e-02 1.60273910e-01
7.67562807e-01 5.05599558e-01 1.06330788e+00 -1.11403811e+00
-5.10589063e-01 -6.22321725e-01 -5.20716488e-01 -1.38674188e+00
-3.91784221e-01 -1.11253273e+00 1.25607830e-02 -1.65728557e+00
5.31968236e-01 -4.73622710e-01 -3.54443878e-01 6.24462903e-01
-1.69104487e-01 8.73889208e-01 3.09467494e-01 3.16558123e-01
-1.06001794e+00 3.86282533e-01 1.19682050e+00 -1.01742573e-01
2.34323163e-02 3.07808429e-01 -9.29514170e-01 7.41435230e-01
8.40912879e-01 -4.57491904e-01 -1.96582958e-01 -4.52250540e-01
-1.78924203e-02 -5.76399207e-01 6.77539051e-01 -1.26854336e+00
1.51688412e-01 -9.26344693e-02 2.04895020e-01 -6.19082212e-01
3.92295092e-01 -5.50065935e-01 -3.68116051e-02 2.34209254e-01
-6.07105531e-02 -1.86836809e-01 3.93330634e-01 3.19250464e-01
-1.78506553e-01 6.12146268e-03 6.96162105e-01 -1.68294281e-01
-9.31980908e-01 5.02144694e-01 -3.80980968e-02 1.13808267e-01
1.11071062e+00 -4.33954567e-01 -4.69398558e-01 -7.19127664e-03
-6.48986280e-01 2.31853172e-01 6.05511427e-01 6.20966136e-01
4.21268880e-01 -1.04425788e+00 -8.24646890e-01 1.56570166e-01
5.35604000e-01 -1.41267091e-01 8.93044248e-02 1.01093662e+00
-3.80394816e-01 3.60984355e-01 -8.89223292e-02 -1.01449990e+00
-1.38622880e+00 4.04799879e-01 2.69193202e-01 -1.26338810e-01
-7.46037424e-01 1.00001490e+00 3.17013919e-01 -4.54684824e-01
3.81158173e-01 -3.73810858e-01 2.89027561e-02 3.79279559e-03
7.54430413e-01 5.32469273e-01 2.64541596e-01 -3.90314311e-01
-6.28617644e-01 4.29762095e-01 -2.27061465e-01 9.48329940e-02
1.40367544e+00 4.69138334e-03 8.55184495e-02 3.81435066e-01
7.38419056e-01 -2.94380300e-02 -1.49147248e+00 -1.68532088e-01
1.82787720e-02 -4.79199708e-01 8.21037889e-02 -7.92916179e-01
-9.99978185e-01 8.45549762e-01 7.24198043e-01 3.62365603e-01
9.97013032e-01 2.23563477e-01 6.56500041e-01 2.27501765e-01
6.12166941e-01 -6.87094867e-01 1.66224211e-01 4.62649047e-01
9.40704584e-01 -1.32062244e+00 8.62306952e-02 -6.05994642e-01
-3.94664675e-01 7.58093059e-01 7.76407301e-01 6.56492636e-02
3.40779483e-01 5.45885623e-01 9.29145291e-02 -2.11712182e-01
-7.84078777e-01 -3.72431099e-01 5.25419056e-01 6.40993178e-01
9.03761506e-01 8.64918679e-02 -2.28342675e-02 3.73263985e-01
1.27182275e-01 9.69596058e-02 2.19498053e-01 9.57064390e-01
-5.19197583e-01 -1.46212113e+00 -4.55076993e-01 4.42467779e-01
-3.07950437e-01 -1.67755350e-01 -3.41138929e-01 6.92327917e-01
2.74603486e-01 8.29572082e-01 -2.31202736e-01 -2.11652651e-01
4.81908530e-01 -1.24865510e-01 7.89903045e-01 -9.91689086e-01
-5.94521463e-01 -1.25235215e-01 -2.53149215e-02 -7.76040494e-01
-2.98398197e-01 -9.73673046e-01 -8.31086397e-01 -2.09368676e-01
-3.31722915e-01 -2.19409928e-01 7.46307254e-01 6.63957357e-01
6.57801151e-01 3.56831521e-01 8.52110982e-02 -1.11722875e+00
-4.10417944e-01 -7.80249238e-01 -2.31083110e-01 1.07640237e-01
1.53683648e-01 -8.53919625e-01 2.39108466e-02 7.05924556e-02] | [8.971108436584473, 0.09494529664516449] |
18993254-c755-4a62-b49d-f8de867f3106 | general-place-recognition-survey-towards-the | 2209.04497 | null | https://arxiv.org/abs/2209.04497v1 | https://arxiv.org/pdf/2209.04497v1.pdf | General Place Recognition Survey: Towards the Real-world Autonomy Age | Place recognition is the fundamental module that can assist Simultaneous Localization and Mapping (SLAM) in loop-closure detection and re-localization for long-term navigation. The place recognition community has made astonishing progress over the last $20$ years, and this has attracted widespread research interest and application in multiple fields such as computer vision and robotics. However, few methods have shown promising place recognition performance in complex real-world scenarios, where long-term and large-scale appearance changes usually result in failures. Additionally, there is a lack of an integrated framework amongst the state-of-the-art methods that can handle all of the challenges in place recognition, which include appearance changes, viewpoint differences, robustness to unknown areas, and efficiency in real-world applications. In this work, we survey the state-of-the-art methods that target long-term localization and discuss future directions and opportunities. We start by investigating the formulation of place recognition in long-term autonomy and the major challenges in real-world environments. We then review the recent works in place recognition for different sensor modalities and current strategies for dealing with various place recognition challenges. Finally, we review the existing datasets for long-term localization and introduce our datasets and evaluation API for different approaches. This paper can be a tutorial for researchers new to the place recognition community and those who care about long-term robotics autonomy. We also provide our opinion on the frequently asked question in robotics: Do robots need accurate localization for long-term autonomy? A summary of this work and our datasets and evaluation API is publicly available to the robotics community at: https://github.com/MetaSLAM/GPRS. | ['Sebastian Scherer', 'Howie Choset', 'Changliu Liu', 'Micheal Milford', 'Guoquan Huang', 'Abulikemu Abuduweili', 'Ivan Cisneros', 'Shiqi Zhao', 'Peng Yin'] | 2022-09-09 | null | null | null | null | ['simultaneous-localization-and-mapping', 'loop-closure-detection'] | ['computer-vision', 'computer-vision'] | [-1.45110562e-01 -3.99742007e-01 -1.84090823e-01 -5.67108393e-01
-6.05547428e-01 -7.35482156e-01 6.05420172e-01 1.57496408e-01
-7.47637630e-01 7.36015320e-01 -2.55934387e-01 -5.40436767e-02
-1.89953804e-01 -5.45120835e-01 -8.48353505e-01 -5.80952287e-01
-5.15469968e-01 3.94550681e-01 3.67589027e-01 -3.93628657e-01
5.95122039e-01 9.69309151e-01 -1.89624667e+00 -5.38331628e-01
8.22127640e-01 7.33541727e-01 5.13247609e-01 5.14312983e-01
2.62136221e-01 2.49540269e-01 -4.03527349e-01 3.16121966e-01
3.07941645e-01 1.55100664e-02 -7.11426139e-01 -1.46020904e-01
5.22292793e-01 -7.94052258e-02 -3.73056233e-01 1.03421462e+00
8.03049922e-01 2.41862059e-01 1.85707584e-01 -1.50389206e+00
-3.06134343e-01 -1.40230373e-01 -3.26196849e-01 9.53851268e-02
8.32517564e-01 4.41093892e-02 3.77318919e-01 -9.87906158e-01
5.68657994e-01 9.27334785e-01 1.04015028e+00 2.63063669e-01
-8.91542673e-01 -5.74546814e-01 2.94577897e-01 3.16305399e-01
-1.72269523e+00 -7.88519979e-01 4.60148603e-01 -2.82061666e-01
1.09227264e+00 7.60990232e-02 5.64421475e-01 9.31360602e-01
4.08800423e-01 6.88196242e-01 8.77945960e-01 -3.89110833e-01
2.47775525e-01 -2.01676846e-01 -2.46009514e-01 6.66015267e-01
4.13003594e-01 1.55601203e-01 -8.10415685e-01 -9.22654644e-02
8.99954259e-01 2.66544610e-01 -1.40924871e-01 -9.92095232e-01
-1.91776741e+00 4.69753802e-01 9.53891277e-01 3.55315566e-01
-2.38302276e-01 3.79063547e-01 -5.70250535e-03 3.27538282e-01
-4.55437116e-02 6.50288582e-01 -3.47415924e-01 -3.35672766e-01
-5.35534263e-01 2.81699717e-01 8.32686067e-01 1.48015749e+00
1.13909030e+00 -2.02561229e-01 6.55068219e-01 9.51447546e-01
5.86559772e-01 9.04540658e-01 5.74865282e-01 -8.61053169e-01
2.88867086e-01 3.75808388e-01 4.70057786e-01 -1.00944221e+00
-6.57320440e-01 -2.56814957e-01 -4.68412608e-01 3.09569627e-01
1.58103853e-01 -1.03813305e-01 -8.68753850e-01 1.45820403e+00
3.82893682e-01 1.49855644e-01 3.00510406e-01 8.17402124e-01
7.17226744e-01 3.11843455e-01 -3.36289376e-01 4.09243405e-01
1.13273382e+00 -1.17795634e+00 -6.26289904e-01 -1.17805099e+00
6.47955775e-01 -9.29433882e-01 4.58592713e-01 -9.18118656e-02
-2.55482167e-01 -2.31542632e-01 -1.44179380e+00 -7.02532008e-02
-8.45624745e-01 1.61629736e-01 8.32605064e-01 2.29147404e-01
-1.24609935e+00 3.05130959e-01 -1.08182847e+00 -1.43230593e+00
-3.58171389e-02 5.38698912e-01 -7.67160714e-01 -3.27181637e-01
-9.30641711e-01 1.38702857e+00 1.69588163e-01 3.55290979e-01
-6.65937006e-01 4.45228629e-02 -1.31012285e+00 -8.10645223e-01
1.31754503e-01 -4.20168966e-01 1.32359195e+00 -3.66296321e-01
-1.26272666e+00 9.66071248e-01 -4.63389993e-01 -3.74873310e-01
3.49954247e-01 -3.01569134e-01 -5.77883720e-01 -4.52832341e-01
6.51467443e-01 9.22747374e-01 9.98309180e-02 -1.23089027e+00
-1.07422042e+00 -5.27396023e-01 1.33915991e-01 6.49375498e-01
5.30281246e-01 -3.34476441e-01 -4.86560553e-01 -1.89496621e-01
1.06268179e+00 -1.43322909e+00 -5.14726818e-01 3.76417488e-01
-9.07332171e-03 1.34142950e-01 6.99746132e-01 1.66141577e-02
4.89797384e-01 -2.28318381e+00 -3.35206926e-01 -9.13018957e-02
-4.19860899e-01 -3.01312387e-01 -4.89587300e-02 7.34349012e-01
3.82307351e-01 -3.07739705e-01 -6.03410006e-02 -4.72917408e-01
5.87118641e-02 5.90758264e-01 -2.78742760e-01 1.10797894e+00
-4.39247429e-01 7.28652716e-01 -1.29281104e+00 9.05501842e-03
7.06966341e-01 3.75401199e-01 1.16545549e-02 -1.90822799e-02
3.77138823e-01 6.67131364e-01 -3.14011961e-01 1.10527456e+00
7.70141780e-01 1.04472138e-01 -1.53092131e-01 3.09639871e-01
-7.16344059e-01 2.95926750e-01 -1.30908477e+00 2.31754923e+00
-4.15603131e-01 7.69962728e-01 1.91145346e-01 -5.77170491e-01
1.10853112e+00 -1.16956808e-01 4.59474355e-01 -7.77010322e-01
-1.50300354e-01 9.94900882e-01 -3.34719688e-01 -1.21500768e-01
9.60329294e-01 3.29570949e-01 -4.11546141e-01 -1.68310031e-01
-1.68150783e-01 -3.71570677e-01 -3.71610187e-02 -3.56653392e-01
1.12994397e+00 2.84690410e-01 6.05325043e-01 -2.70593375e-01
4.37075227e-01 3.64761531e-01 4.94692266e-01 9.18619394e-01
-7.29600608e-01 6.94089890e-01 -5.20913303e-01 -6.31466746e-01
-6.05428457e-01 -8.80167782e-01 -1.98839784e-01 9.70332682e-01
1.15657270e+00 -1.13918789e-01 1.53117049e-02 -2.62026131e-01
2.85959512e-01 8.64022076e-02 -4.30480719e-01 1.18779033e-01
-4.70071346e-01 -4.31244433e-01 6.62408292e-01 4.21002388e-01
8.17923725e-01 -9.46402609e-01 -8.64171565e-01 2.28862241e-01
-4.96238023e-01 -1.10440946e+00 -2.46243682e-02 4.53847140e-01
-8.71414125e-01 -1.03947830e+00 -6.57788932e-01 -1.18819368e+00
7.75943696e-01 1.00124407e+00 5.53943813e-01 -1.02122732e-01
-5.53940237e-02 6.75723195e-01 -4.99259889e-01 -4.94222164e-01
3.26231152e-01 1.48157120e-01 5.92010617e-01 -4.75124419e-01
4.43508208e-01 -6.24001324e-01 -5.82364619e-01 9.19413388e-01
-3.85487139e-01 -3.13306332e-01 6.78995013e-01 6.03140473e-01
8.90348792e-01 -5.52535355e-01 6.24961443e-02 -9.55340266e-02
3.80102582e-02 -5.63142896e-01 -8.01297903e-01 8.04726034e-02
-5.87265611e-01 -6.74350113e-02 -1.45480096e-01 -1.38034850e-01
-4.41242576e-01 5.38587034e-01 -1.24479413e-01 -1.37047125e-02
-4.51139212e-01 4.88832980e-01 7.86074325e-02 -9.51102257e-01
8.55149627e-01 4.17731851e-01 -1.32447546e-02 -4.26061958e-01
2.48718441e-01 6.92333400e-01 7.30245471e-01 -3.91041338e-01
6.89578414e-01 9.23449099e-01 2.58785374e-02 -9.06000137e-01
-4.26210076e-01 -1.18019319e+00 -8.67227435e-01 -1.21026449e-01
4.56630349e-01 -1.18621409e+00 -3.14530939e-01 8.07482779e-01
-1.08298600e+00 -4.03698027e-01 -1.48661092e-01 6.80553734e-01
-8.11825037e-01 2.04577118e-01 -7.45009035e-02 -7.12576151e-01
5.57375811e-02 -1.23303163e+00 1.20841694e+00 5.65375388e-01
5.97384125e-02 -9.41505313e-01 5.39602160e-01 -7.89958164e-02
5.30444384e-01 3.19332212e-01 -4.20276821e-01 -2.52478063e-01
-8.30365658e-01 -5.50377429e-01 1.08017318e-01 -4.81356263e-01
1.98630810e-01 -6.40554488e-01 -9.02013242e-01 -6.07279480e-01
-4.12475854e-01 -1.84961498e-01 5.21610379e-01 2.00230196e-01
3.05132926e-01 7.84203932e-02 -8.95669460e-01 9.12267089e-01
1.38520765e+00 2.48463497e-01 6.57665908e-01 1.16669154e+00
4.93848503e-01 4.91619259e-01 1.30780029e+00 2.48455063e-01
8.69434357e-01 7.97060430e-01 7.56465971e-01 -3.41295712e-02
1.35408089e-01 -4.62535590e-01 3.27928334e-01 4.07295287e-01
-1.78546309e-02 3.46433856e-02 -1.20078719e+00 8.83306861e-01
-2.31777382e+00 -6.26153946e-01 -1.51455194e-01 2.32919955e+00
2.40322929e-02 -3.71032268e-01 -5.60555279e-01 -2.40334973e-01
7.64771521e-01 3.48623991e-01 -5.77948213e-01 1.48567989e-01
-2.45551988e-01 -5.86091995e-01 1.20225167e+00 6.79687262e-01
-1.62416470e+00 1.14122021e+00 6.38647556e+00 2.40237683e-01
-1.45246422e+00 -6.49150759e-02 -2.96426266e-01 4.24075484e-01
2.99231827e-01 3.65472019e-01 -9.36031461e-01 6.51746541e-02
5.80752134e-01 7.67156854e-02 3.30273390e-01 1.41109216e+00
-1.20458305e-01 -7.88150907e-01 -1.08923125e+00 1.45356905e+00
3.09887439e-01 -1.04630399e+00 -7.23891735e-01 3.08050841e-01
7.73317039e-01 1.15448809e+00 -1.06746502e-01 2.11298928e-01
1.37590140e-01 -6.71569943e-01 1.09500480e+00 2.20760986e-01
8.23374987e-01 -2.85090685e-01 1.02318668e+00 5.26618779e-01
-1.59977829e+00 -9.04068574e-02 -4.91295516e-01 -3.88879716e-01
2.38260955e-01 3.54965121e-01 -9.00252640e-01 5.62250495e-01
8.66142571e-01 1.10905790e+00 -7.51369059e-01 1.71921253e+00
-3.31360012e-01 -4.09450859e-01 -8.59635949e-01 -8.91978443e-02
3.44377667e-01 -2.25895997e-02 6.14011943e-01 9.91492689e-01
7.01059997e-01 -2.12055534e-01 4.35037106e-01 3.00012171e-01
3.76942694e-01 -2.10807502e-01 -1.07282591e+00 3.75159442e-01
8.42398822e-01 1.01323032e+00 -8.06234062e-01 9.53493863e-02
-2.66168088e-01 1.22816622e+00 1.68204576e-01 2.73414433e-01
-3.95162761e-01 -6.01491272e-01 9.26790655e-01 -2.95983199e-02
-2.79981364e-02 -8.98734808e-01 -2.14876994e-01 -1.22503591e+00
1.82179525e-01 -2.75142044e-01 2.09930196e-01 -7.63173759e-01
-8.27518523e-01 3.76688391e-01 -1.89066514e-01 -1.77651489e+00
-2.78254449e-01 -4.20912087e-01 -4.26852554e-02 7.03410208e-01
-1.94892561e+00 -1.29080105e+00 -7.10443974e-01 5.15284181e-01
6.25155926e-01 2.00911984e-01 1.15060520e+00 3.28819811e-01
3.54501717e-02 1.48241773e-01 4.79517013e-01 2.00593360e-02
1.14300466e+00 -9.88526106e-01 6.71121895e-01 1.01794958e+00
7.91621953e-02 8.77672732e-01 8.53875279e-01 -7.03863442e-01
-1.80374944e+00 -9.87800777e-01 1.22587752e+00 -5.91455102e-01
4.35178787e-01 -3.72661412e-01 -2.42943108e-01 1.05096054e+00
-2.71554798e-01 3.65207016e-01 1.95426464e-01 1.26616165e-01
-1.53231561e-01 -2.05238596e-01 -1.20080674e+00 4.89934176e-01
1.25817585e+00 -4.71014559e-01 -4.77373272e-01 3.18441153e-01
8.02254081e-02 -1.00872707e+00 -4.48137075e-01 5.78270018e-01
8.39831591e-01 -8.17862868e-01 9.73546803e-01 5.83981037e-01
-5.52718997e-01 -8.81829679e-01 -5.21840453e-01 -1.18036556e+00
-3.84357989e-01 -3.45444620e-01 2.65110284e-01 7.72487581e-01
1.99934319e-01 -1.12857091e+00 5.54525495e-01 3.12927037e-01
-3.27043235e-01 -2.37495542e-01 -1.23789203e+00 -1.02924037e+00
-4.72824007e-01 -3.77009690e-01 4.24726695e-01 8.22761416e-01
8.79326314e-02 -5.34030423e-02 -3.46611261e-01 7.14643300e-01
5.73098779e-01 6.97315335e-02 1.25068629e+00 -1.21553397e+00
5.68807304e-01 -8.06481615e-02 -1.23163044e+00 -1.56234372e+00
-1.19619064e-01 -5.31727374e-01 8.73094678e-01 -2.15391660e+00
-3.83275151e-01 -8.72694373e-01 -1.58473819e-01 5.52045643e-01
4.05282170e-01 3.38422298e-01 -2.21075788e-02 7.15517402e-01
-1.09287381e+00 5.39957106e-01 6.94810808e-01 -8.79898965e-02
-1.89174235e-01 1.04898043e-01 -3.36550891e-01 6.53587699e-01
7.83940315e-01 -3.33155930e-01 -1.50364742e-01 -7.86795795e-01
1.90936893e-01 -3.73318881e-01 3.71073872e-01 -1.74479914e+00
9.11828279e-01 -3.17214727e-01 2.29086399e-01 -8.52181256e-01
5.98485291e-01 -1.03228664e+00 3.43255311e-01 6.04204476e-01
3.59879047e-01 3.12307328e-01 1.59636691e-01 6.87420726e-01
-3.62960488e-01 2.68905554e-02 6.52105689e-01 -2.54258156e-01
-1.73013473e+00 2.88105398e-01 -4.17227924e-01 -5.52454233e-01
1.24558520e+00 -6.41857982e-01 -2.50806779e-01 -3.76447916e-01
-3.56946200e-01 5.19110382e-01 1.22406745e+00 8.11964750e-01
6.32587254e-01 -1.43116868e+00 -2.68313557e-01 4.11631346e-01
8.39829147e-01 2.42058024e-01 3.53036374e-02 1.00689352e+00
-9.63802576e-01 5.33580363e-01 -2.86646545e-01 -9.67186272e-01
-9.77398276e-01 1.25225708e-01 4.60257620e-01 4.01266426e-01
-4.72723693e-01 7.69888759e-01 -1.27924070e-01 -1.09767663e+00
3.44878048e-01 -2.85072088e-01 -4.59911451e-02 -2.90644050e-01
4.25019473e-01 4.24480706e-01 1.56060442e-01 -1.18100035e+00
-1.14623034e+00 1.00996542e+00 4.59409654e-01 -2.03017920e-01
1.25892079e+00 -8.06131244e-01 -1.71449706e-01 6.76553726e-01
8.54825258e-01 -5.22753745e-02 -1.23069596e+00 -9.82187241e-02
1.85752839e-01 -5.25598943e-01 -2.09109887e-01 -5.87811410e-01
-4.11086798e-01 5.13248801e-01 1.01590621e+00 -3.21467876e-01
4.56007510e-01 1.40895858e-01 4.00083840e-01 9.85110819e-01
1.56808448e+00 -1.10312092e+00 -3.02390575e-01 1.31874204e+00
1.05888236e+00 -1.60753894e+00 5.81535399e-02 -2.58975744e-01
-2.49039948e-01 8.41557324e-01 5.06806672e-01 -2.30868101e-01
5.18135965e-01 9.75993872e-02 5.33292055e-01 -8.86759460e-02
1.17909662e-01 -3.11540663e-01 -1.21250592e-01 1.00419891e+00
1.30032375e-01 8.59429762e-02 2.84434762e-02 -1.52837649e-01
-3.78397912e-01 -1.98084429e-01 3.02892983e-01 1.65880990e+00
-8.00218940e-01 -9.95591164e-01 -7.58518696e-01 -7.07472786e-02
1.99127987e-01 2.12564066e-01 -3.48295987e-01 6.25083745e-01
2.48152256e-01 1.02003479e+00 -9.56505388e-02 -3.90269756e-01
5.10959029e-01 -1.60801679e-01 4.69190359e-01 -5.08903325e-01
5.77918068e-02 -2.27943063e-01 8.17543268e-02 -1.00777173e+00
-6.14094377e-01 -8.78545582e-01 -1.37545145e+00 -3.20031881e-01
-3.57336015e-01 1.07761234e-01 1.34917760e+00 7.48703599e-01
6.51565194e-01 -9.08767991e-03 2.78357834e-01 -1.57322824e+00
8.70953873e-03 -8.31115246e-01 -6.03395402e-01 -1.75164193e-01
7.91283190e-01 -1.02252936e+00 -3.58687222e-01 -3.02534372e-01] | [7.307878494262695, -2.048964023590088] |
667aed3f-7812-461e-ad25-074980824b8c | tdiot-target-driven-inference-for-deep-video | 2103.11017 | null | https://arxiv.org/abs/2103.11017v2 | https://arxiv.org/pdf/2103.11017v2.pdf | TDIOT: Target-driven Inference for Deep Video Object Tracking | Recent tracking-by-detection approaches use deep object detectors as target detection baseline, because of their high performance on still images. For effective video object tracking, object detection is integrated with a data association step performed by either a custom design inference architecture or an end-to-end joint training for tracking purpose. In this work, we adopt the former approach and use the pre-trained Mask R-CNN deep object detector as the baseline. We introduce a novel inference architecture placed on top of FPN-ResNet101 backbone of Mask R-CNN to jointly perform detection and tracking, without requiring additional training for tracking purpose. The proposed single object tracker, TDIOT, applies an appearance similarity-based temporal matching for data association. In order to tackle tracking discontinuities, we incorporate a local search and matching module into the inference head layer that exploits SiamFC for short term tracking. Moreover, in order to improve robustness to scale changes, we introduce a scale adaptive region proposal network that enables to search the target at an adaptively enlarged spatial neighborhood specified by the trace of the target. In order to meet long term tracking requirements, a low cost verification layer is incorporated into the inference architecture to monitor presence of the target based on its LBP histogram model. Performance evaluation on videos from VOT2016, VOT2018 and VOT-LT2018 datasets demonstrate that TDIOT achieves higher accuracy compared to the state-of-the-art short-term trackers while it provides comparable performance in long term tracking. | ['Bilge Gunsel', 'Ozgun Cirakman', 'Llukman Cerkezi', 'Filiz Gurkan'] | 2021-03-19 | null | null | null | null | ['video-object-tracking'] | ['computer-vision'] | [-2.11764961e-01 -3.83144945e-01 -2.63609171e-01 -4.29438539e-02
-4.32683796e-01 -3.76158446e-01 5.44989884e-01 -1.38102495e-03
-7.60480225e-01 2.77004331e-01 -3.18964392e-01 7.71187320e-02
1.19856037e-01 -4.67577159e-01 -8.40180099e-01 -7.35486567e-01
7.50653353e-03 3.69931281e-01 9.81073022e-01 1.19030721e-01
-3.23879123e-02 6.47093177e-01 -1.49066162e+00 9.14250091e-02
5.51828206e-01 1.47825158e+00 3.76236558e-01 4.99284714e-01
5.04458174e-02 5.39834976e-01 -5.88143110e-01 -2.68553406e-01
6.19045734e-01 3.49198729e-02 5.50141558e-02 7.84054771e-03
9.34968233e-01 -4.89833295e-01 -4.49683815e-01 1.14190388e+00
6.28116429e-01 2.62503419e-02 3.34274143e-01 -1.13067484e+00
-2.61301279e-01 3.80592167e-01 -5.87030590e-01 5.24272084e-01
6.40482977e-02 5.92512369e-01 5.66895306e-01 -9.44005191e-01
4.61531013e-01 1.12779570e+00 9.92644548e-01 5.62242866e-01
-1.12828982e+00 -1.00704622e+00 3.23850811e-01 2.10833132e-01
-1.57086122e+00 -3.94390941e-01 6.72495186e-01 -4.52511102e-01
6.56316400e-01 -5.59113063e-02 6.84951544e-01 1.08380091e+00
3.38742673e-01 7.13451922e-01 6.95125341e-01 -9.97048095e-02
5.42828552e-02 2.36438796e-01 3.89406458e-02 6.70620441e-01
2.96968937e-01 5.31186581e-01 -2.02690601e-01 1.17570162e-01
8.23786676e-01 8.52365419e-02 -1.37024879e-01 -5.06470680e-01
-1.21524966e+00 4.86815304e-01 9.05583560e-01 5.05141020e-01
-4.37432885e-01 3.96522045e-01 5.04832804e-01 3.94671671e-02
2.04013020e-01 -5.59804030e-02 -2.61821806e-01 3.47330242e-01
-1.28991187e+00 1.79861456e-01 2.35906780e-01 8.56982350e-01
3.67363483e-01 3.77302766e-01 -8.50215614e-01 3.12556982e-01
7.15252697e-01 6.23515129e-01 4.21989709e-01 -3.64066303e-01
2.26882637e-01 5.60386360e-01 3.16338003e-01 -8.72334599e-01
-3.96325201e-01 -1.17553139e+00 -5.98622680e-01 5.38370609e-01
3.90963733e-01 1.24641366e-01 -1.15137780e+00 1.73329425e+00
7.02767372e-01 7.10725427e-01 -8.68406072e-02 1.11830509e+00
7.75499642e-01 2.72015631e-01 3.14554095e-01 -6.86154440e-02
1.68293107e+00 -9.35903370e-01 -7.46779859e-01 2.36012191e-02
4.09836322e-01 -8.08619916e-01 4.38620329e-01 8.06192681e-02
-9.23420548e-01 -1.26345468e+00 -1.05722034e+00 2.76395172e-01
-3.55415076e-01 7.68756986e-01 2.35845204e-02 6.80035651e-01
-8.68672609e-01 3.30787569e-01 -1.05686760e+00 -3.28502744e-01
4.63761300e-01 5.03498912e-01 -2.11225543e-03 4.04822171e-01
-9.36352491e-01 9.57158983e-01 6.76640511e-01 4.54916120e-01
-1.07207930e+00 -7.75217891e-01 -8.33768189e-01 -4.20302637e-02
5.17308176e-01 -7.27379084e-01 9.47359324e-01 -7.11598635e-01
-1.47701085e+00 6.43140495e-01 1.31862126e-02 -9.98521566e-01
8.41355741e-01 -3.17288637e-01 -4.72135067e-01 5.99050634e-02
-3.94336358e-02 7.81871498e-01 1.25091386e+00 -1.05021060e+00
-9.18987274e-01 -2.92861134e-01 -1.41197637e-01 -8.84454325e-02
-1.65496826e-01 1.97458267e-01 -8.63474488e-01 -8.82562697e-01
-1.59399718e-01 -1.00487876e+00 -5.85919507e-02 3.71084690e-01
-2.18395501e-01 -4.18483257e-01 1.53314519e+00 -5.17876089e-01
1.24900901e+00 -2.18660498e+00 -1.15353443e-01 6.71158507e-02
1.06449649e-01 8.26239228e-01 -1.65895328e-01 -2.06789181e-01
1.25120267e-01 -6.22674227e-01 2.74018615e-01 -6.32661521e-01
-6.94161933e-03 -1.61293671e-01 -1.46971717e-01 8.58530641e-01
2.15101883e-01 1.00920713e+00 -6.18440568e-01 -6.92620933e-01
7.16720223e-01 6.51276648e-01 -2.82744586e-01 1.88885212e-01
-4.58411336e-01 5.02616405e-01 -4.33941096e-01 6.78960621e-01
8.47137213e-01 -1.69024572e-01 -2.14132473e-01 -6.04413450e-01
-5.96168160e-01 -2.07431972e-01 -1.29994774e+00 1.60533464e+00
-2.28364512e-01 4.17515695e-01 3.11816573e-01 -6.42345667e-01
9.74597394e-01 3.63023818e-01 5.28970182e-01 -6.14660919e-01
4.80619818e-01 1.62149400e-01 1.51562080e-01 -1.80587068e-01
3.55990648e-01 2.40463287e-01 2.51924574e-01 -1.19331211e-01
-6.79402426e-03 6.29391670e-01 2.21026346e-01 -1.07000761e-01
8.49139273e-01 4.31307107e-01 -1.27018571e-01 -1.51863813e-01
9.34619308e-01 3.57148387e-02 7.45850027e-01 7.85282671e-01
-4.20299202e-01 3.56335819e-01 -3.74137759e-01 -5.38315117e-01
-7.03493059e-01 -9.68834341e-01 -3.40252459e-01 1.03058314e+00
4.46816385e-01 -5.01268581e-02 -4.78896230e-01 -7.83700585e-01
2.22664312e-01 2.62403816e-01 -6.73904777e-01 -5.64850010e-02
-9.66771007e-01 -3.83711249e-01 5.86794198e-01 7.99071074e-01
7.41278470e-01 -1.00260806e+00 -1.05377960e+00 4.98290718e-01
1.89505950e-01 -1.37348342e+00 -6.78641856e-01 9.35001597e-02
-6.23669267e-01 -9.77309704e-01 -8.28335941e-01 -6.92031503e-01
4.03927177e-01 1.53529689e-01 4.72299457e-01 2.66290590e-04
-4.02343154e-01 2.38043904e-01 -1.47449285e-01 -1.80937499e-01
-2.85062551e-01 -6.39509931e-02 1.93574026e-01 2.62995631e-01
2.81820893e-01 -1.45178482e-01 -9.36947227e-01 6.19803727e-01
-6.45398974e-01 -2.49577165e-01 7.82568872e-01 6.20542407e-01
5.13410389e-01 -1.34830087e-01 3.38820577e-01 3.89173650e-03
-1.77721545e-01 -1.51202738e-01 -1.32004440e+00 3.11002851e-01
-3.41392785e-01 -4.03631218e-02 5.09421289e-01 -7.95214713e-01
-7.68218458e-01 4.57196623e-01 -1.25794798e-01 -1.16962576e+00
4.19675000e-02 -4.08776104e-03 -7.76099935e-02 -4.96320546e-01
3.81492138e-01 3.39511782e-01 8.13984945e-02 -5.65209091e-01
2.05985412e-01 3.31839085e-01 7.90097356e-01 -6.30439669e-02
1.16518939e+00 5.96476078e-01 1.44656515e-03 -4.92952555e-01
-6.81976736e-01 -6.65666163e-01 -6.46248877e-01 -6.27369642e-01
1.23825693e+00 -1.28142142e+00 -1.02642381e+00 4.03817832e-01
-1.14861608e+00 -2.35355109e-01 -1.58928633e-01 7.91850626e-01
-2.20711336e-01 2.19024658e-01 -4.39804345e-01 -7.60985136e-01
-6.18340015e-01 -1.11005139e+00 1.32407987e+00 3.98573250e-01
2.57290930e-01 -8.10216129e-01 -7.11624175e-02 1.28473878e-01
7.15824187e-01 3.19755107e-01 -4.66324564e-04 -6.28115654e-01
-1.00970554e+00 -3.23135674e-01 -4.34393287e-01 3.05711359e-01
-5.87342083e-02 -8.48376751e-02 -7.35821068e-01 -8.03375065e-01
-1.92327216e-01 4.77881171e-02 1.15497315e+00 6.29190922e-01
8.43498290e-01 4.00932021e-02 -6.80632055e-01 7.10605741e-01
1.51523328e+00 1.37624219e-01 2.85180092e-01 3.93224925e-01
7.68208444e-01 1.46087669e-02 9.05998945e-01 3.74959737e-01
1.15738444e-01 1.22779667e+00 5.54975867e-01 8.79302695e-02
-6.83606029e-01 2.66479254e-02 6.72798336e-01 2.02858657e-01
2.13307530e-01 -3.62809114e-02 -5.36996305e-01 4.33098465e-01
-1.98651206e+00 -1.05964768e+00 -6.55109584e-02 2.32058334e+00
3.64202112e-01 5.17036796e-01 3.82096112e-01 -2.89479107e-01
9.78923082e-01 8.41872916e-02 -4.95239973e-01 2.94195980e-01
4.01001796e-02 -3.44546698e-03 7.63359427e-01 2.04012573e-01
-1.43131781e+00 1.00373733e+00 4.75748110e+00 1.05295312e+00
-1.57272851e+00 2.88243413e-01 -1.17304891e-01 -6.39796779e-02
5.86682379e-01 -2.46153772e-01 -1.47978342e+00 5.66974699e-01
6.39300466e-01 3.39098781e-01 -2.42651343e-01 8.32667232e-01
3.83915484e-01 9.30548981e-02 -9.01490152e-01 8.44361484e-01
-9.92633551e-02 -1.24637234e+00 -2.54808396e-01 -3.45644988e-02
2.92492390e-01 1.94004759e-01 1.54395923e-01 4.93283093e-01
-1.07595325e-01 -5.52223802e-01 1.03346872e+00 5.26694536e-01
5.04328907e-01 -4.69008625e-01 7.46222019e-01 2.71288633e-01
-1.98684537e+00 -2.71398753e-01 -3.46030086e-01 3.55403781e-01
2.74088770e-01 2.72892952e-01 -8.27682734e-01 6.48018658e-01
7.87612081e-01 7.19218135e-01 -7.62878358e-01 1.54726052e+00
6.38495684e-02 3.55054975e-01 -4.84078854e-01 2.94434965e-01
3.60228181e-01 2.21847683e-01 9.31233227e-01 1.40427971e+00
2.74033159e-01 -4.45480555e-01 7.81278610e-01 7.83918560e-01
1.36578664e-01 -2.32830003e-01 -2.74823666e-01 4.19073433e-01
4.99940783e-01 1.47706437e+00 -7.95696259e-01 -4.17161316e-01
-2.40927607e-01 6.64056003e-01 8.91847759e-02 8.11928660e-02
-1.46929753e+00 -1.19440556e-01 4.99123722e-01 1.50030285e-01
1.17750978e+00 -1.18533440e-01 2.89566964e-01 -9.25780535e-01
3.21883708e-02 -6.46291018e-01 2.93740571e-01 -4.45665747e-01
-9.54744339e-01 7.13996053e-01 -1.09230869e-01 -1.61921012e+00
-8.51097777e-02 -5.39864421e-01 -6.41071916e-01 6.50970876e-01
-1.53184307e+00 -1.57716787e+00 -4.60822642e-01 6.55901372e-01
4.30075407e-01 -1.20322853e-01 1.16787449e-01 7.23554790e-01
-8.25459599e-01 8.14745724e-01 -2.39686713e-01 4.98781174e-01
7.29522884e-01 -9.78663564e-01 2.15267196e-01 1.01771784e+00
1.53448861e-02 5.98920882e-01 7.19011009e-01 -9.65547264e-01
-1.47106314e+00 -1.72412550e+00 1.83705509e-01 -3.38654488e-01
5.94108462e-01 -3.77819628e-01 -9.84871209e-01 6.59080327e-01
1.61385089e-01 6.41383708e-01 5.21398224e-02 -4.67965454e-01
-3.02527100e-01 -3.80073667e-01 -9.90836024e-01 2.50401944e-01
8.72299850e-01 -1.41622677e-01 -4.96763855e-01 6.25138283e-02
7.37087846e-01 -5.61912835e-01 -8.36597323e-01 8.46844971e-01
7.28057384e-01 -6.31857693e-01 1.08308434e+00 1.64292995e-02
-5.90370178e-01 -1.12951219e+00 1.35690048e-01 -5.86048067e-01
-4.38209772e-01 -4.86297399e-01 -4.25681829e-01 1.07254505e+00
-1.27609912e-02 -3.91143143e-01 1.04590476e+00 -6.30480349e-02
-2.43881524e-01 -6.06144488e-01 -1.21697664e+00 -1.25729215e+00
-4.99113172e-01 -2.76435018e-01 2.31606320e-01 4.01081115e-01
-8.32856178e-01 1.15047090e-01 -4.27849233e-01 6.00722909e-01
1.12464631e+00 2.09314749e-01 9.05420661e-01 -1.10901880e+00
-2.67303526e-01 -5.36631703e-01 -6.88946009e-01 -1.20604539e+00
5.05523989e-03 -7.05770373e-01 2.40601242e-01 -1.03308392e+00
1.38406930e-02 -4.24409658e-01 -5.74107945e-01 3.02534193e-01
-9.18541327e-02 6.55617833e-01 4.84428495e-01 1.39311641e-01
-9.67967927e-01 6.98935807e-01 1.03079212e+00 -1.04981601e-01
-2.80508906e-01 2.93724269e-01 1.12827301e-01 4.89827663e-01
4.31502372e-01 -5.70084810e-01 1.71722081e-02 -2.36947596e-01
-4.17776883e-01 1.40077174e-02 9.56142306e-01 -1.66775763e+00
7.55679309e-01 3.76061082e-01 6.62231982e-01 -1.25704372e+00
4.50775445e-01 -1.20530105e+00 2.55566359e-01 9.82730687e-01
-1.00226402e-01 -4.05827630e-03 4.75994319e-01 7.14673638e-01
-1.17300138e-01 5.02992347e-02 1.12979090e+00 3.49117339e-01
-7.86815882e-01 5.21796823e-01 -1.56282768e-01 -4.94382292e-01
1.26974905e+00 -4.25980479e-01 -1.85838819e-01 3.24627638e-01
-7.93673098e-01 4.28295106e-01 3.21278185e-01 6.18117869e-01
5.59135914e-01 -1.37352669e+00 -6.38266802e-01 7.05117285e-02
1.21881582e-01 -2.57073075e-01 2.70498484e-01 1.33174264e+00
-1.41508058e-01 5.48559666e-01 -1.96934447e-01 -9.95438814e-01
-1.53831148e+00 8.04766357e-01 5.08555055e-01 -3.40785474e-01
-8.36947501e-01 6.74621046e-01 2.13307485e-01 7.16777667e-02
6.62646890e-01 -5.39816260e-01 -3.17809135e-01 -7.84981847e-02
6.67309761e-01 1.85411006e-01 -2.26000343e-02 -8.19186509e-01
-5.99726319e-01 7.94743657e-01 -1.83515206e-01 2.15866640e-01
7.73037255e-01 -5.43294437e-02 2.39205807e-01 8.33329558e-03
9.06588316e-01 -3.90325263e-02 -1.61669433e+00 -4.99462128e-01
9.07237008e-02 -2.78924733e-01 2.29051247e-01 -6.17136836e-01
-1.31093752e+00 5.26524425e-01 1.28516233e+00 -1.65877759e-01
9.25867736e-01 -1.69128209e-01 8.36373031e-01 9.18914080e-02
2.25946441e-01 -7.58027971e-01 1.60214797e-01 4.12432790e-01
7.24863052e-01 -1.29550564e+00 -7.73028880e-02 -1.04268797e-01
-1.57176405e-01 1.00889027e+00 9.69105363e-01 -3.30297261e-01
5.62842250e-01 2.72855729e-01 1.19918147e-02 -9.25241262e-02
-5.75540662e-01 -6.51177168e-01 7.17103660e-01 3.22055638e-01
2.98894495e-02 -3.70675266e-01 -7.65022188e-02 1.75668836e-01
3.56325686e-01 1.08145513e-01 -2.65138507e-01 7.86977112e-01
-6.37885928e-01 -8.61633599e-01 -6.82240248e-01 3.32721099e-02
-4.72498804e-01 2.41197482e-01 9.85587314e-02 9.70420539e-01
5.95375061e-01 6.08375192e-01 1.79022521e-01 -3.75550002e-01
2.57086873e-01 -1.86143667e-01 5.00880420e-01 -3.14543009e-01
-1.07175124e+00 4.21352804e-01 -3.20237398e-01 -7.27154076e-01
-5.66257536e-01 -6.43508852e-01 -1.21693015e+00 9.80748385e-02
-7.14578032e-01 -2.02246513e-02 7.30417550e-01 9.09823716e-01
3.44259620e-01 8.15982580e-01 2.68818855e-01 -1.02280915e+00
-5.77401221e-01 -1.04631937e+00 -1.45255834e-01 6.38080835e-02
5.64472795e-01 -1.00621951e+00 5.94526790e-02 -1.46842092e-01] | [6.358063697814941, -2.1259827613830566] |
b9e0dc2d-41ee-4d80-8836-5e088964377a | differentially-private-learning-does-not | 2010.12112 | null | https://arxiv.org/abs/2010.12112v4 | https://arxiv.org/pdf/2010.12112v4.pdf | Investigating Membership Inference Attacks under Data Dependencies | Training machine learning models on privacy-sensitive data has become a popular practice, driving innovation in ever-expanding fields. This has opened the door to new attacks that can have serious privacy implications. One such attack, the Membership Inference Attack (MIA), exposes whether or not a particular data point was used to train a model. A growing body of literature uses Differentially Private (DP) training algorithms as a defence against such attacks. However, these works evaluate the defence under the restrictive assumption that all members of the training set, as well as non-members, are independent and identically distributed. This assumption does not hold for many real-world use cases in the literature. Motivated by this, we evaluate membership inference with statistical dependencies among samples and explain why DP does not provide meaningful protection (the privacy parameter $\epsilon$ scales with the training set size $n$) in this more general case. We conduct a series of empirical evaluations with off-the-shelf MIAs using training sets built from real-world data showing different types of dependencies among samples. Our results reveal that training set dependencies can severely increase the performance of MIAs, and therefore assuming that data samples are statistically independent can significantly underestimate the performance of MIAs. | ['Matthew Rafuse', 'Simon Oya', 'Urs Hengartner', 'Florian Kerschbaum', 'Ian Goldberg', 'Lindsey Tulloch', 'Thomas Humphries'] | 2020-10-23 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 2.18132019e-01 1.34540871e-01 -5.08145802e-02 -6.30541027e-01
-5.38476825e-01 -1.08140433e+00 4.22891796e-01 3.32410485e-01
-5.96380949e-01 7.63616979e-01 -2.93695539e-01 -6.98537230e-01
-6.72282875e-02 -9.79163587e-01 -8.10713470e-01 -9.01460052e-01
-3.37989956e-01 1.67813733e-01 2.58503295e-02 9.40068886e-02
1.03780456e-01 6.96101129e-01 -1.32738769e+00 3.26957256e-01
6.95676088e-01 9.55102205e-01 -7.88589537e-01 3.85208219e-01
3.38974684e-01 1.05771735e-01 -8.93252671e-01 -9.93233979e-01
8.90371084e-01 -2.69555569e-01 -7.16729820e-01 -3.53248984e-01
3.49503100e-01 -3.11040908e-01 -2.27910280e-01 1.28520823e+00
3.23201239e-01 -2.68404216e-01 4.31237221e-01 -1.61560535e+00
-1.32658929e-01 5.33211470e-01 -3.21681917e-01 5.40850684e-02
-1.46457702e-02 1.19780980e-01 9.11712348e-01 -1.10279784e-01
3.45744967e-01 8.67483199e-01 6.22814775e-01 5.47581255e-01
-1.39487433e+00 -1.06756544e+00 -1.77040353e-01 -1.19002433e-02
-1.60349774e+00 -3.64897877e-01 4.28008050e-01 -2.03942299e-01
3.65339726e-01 8.85130048e-01 1.71606600e-01 1.18269074e+00
2.22856253e-01 4.23301607e-01 1.27959919e+00 -2.50526577e-01
6.68458819e-01 6.87627792e-01 1.80351734e-01 9.36143249e-02
8.98353517e-01 2.56039441e-01 -2.11115658e-01 -1.05244088e+00
2.60546297e-01 3.74423265e-02 -3.41655225e-01 -5.78403354e-01
-5.56654096e-01 1.02463818e+00 6.28742650e-02 1.36050642e-01
-1.61699299e-02 -2.02358231e-01 5.27066052e-01 6.24131382e-01
3.14688683e-01 4.40731704e-01 -5.16887784e-01 8.94202739e-02
-6.97445929e-01 4.03657347e-01 1.30668926e+00 8.45434189e-01
6.87904894e-01 -6.15435719e-01 8.93094614e-02 3.31579626e-01
5.93028180e-02 2.62615561e-01 2.70775288e-01 -6.77609503e-01
4.62698519e-01 4.71648723e-01 1.96976244e-01 -1.12413335e+00
4.67127077e-02 -4.26362246e-01 -7.43480265e-01 4.15720016e-01
9.40760016e-01 -5.59497297e-01 -1.83937863e-01 1.91871357e+00
3.67783755e-01 3.86705175e-02 1.66316122e-01 4.63868469e-01
1.08893313e-01 2.80336261e-01 2.81945672e-02 -1.84715942e-01
1.19118083e+00 -4.82503399e-02 -3.98207754e-01 -9.95331183e-02
8.55479836e-01 -3.17393154e-01 7.65895545e-01 4.55406755e-01
-7.40970254e-01 -8.67847074e-03 -1.19103050e+00 2.68155515e-01
-6.09034121e-01 -3.99302602e-01 7.90639937e-01 1.40805662e+00
-7.84225404e-01 6.66545153e-01 -7.03752279e-01 -2.68934935e-01
9.37943876e-01 7.74374187e-01 -7.15181291e-01 -6.32241368e-02
-1.39461899e+00 5.39977193e-01 2.56522238e-01 -1.98009968e-01
-5.12137592e-01 -6.43170059e-01 -5.52506030e-01 1.80619299e-01
4.00572151e-01 -3.14659506e-01 8.99358034e-01 -7.52095520e-01
-8.74284804e-01 1.18672276e+00 1.79715008e-01 -1.04537451e+00
9.23038423e-01 1.25155936e-03 -5.67971826e-01 -1.28292561e-01
-4.10639614e-01 2.38143951e-02 7.30132282e-01 -1.23264205e+00
-5.85393548e-01 -8.18852127e-01 2.62546241e-01 -2.90986806e-01
-5.70323765e-01 9.43610817e-02 6.62778988e-02 -4.00535345e-01
-2.10447058e-01 -1.03303778e+00 -2.86714464e-01 2.57996798e-01
-5.65469563e-01 2.04209730e-01 1.15614820e+00 -2.53893495e-01
1.35112250e+00 -2.53997922e+00 -5.02742290e-01 7.22829401e-01
1.02379881e-01 5.82794607e-01 3.56816202e-01 5.77242672e-01
-8.57708789e-03 5.31172097e-01 -5.57247341e-01 -4.01554257e-01
1.56840622e-01 4.51508433e-01 -5.79109848e-01 9.53106046e-01
-2.76298791e-01 3.94060016e-01 -4.47926968e-01 -1.42475992e-01
4.01719846e-02 3.84088010e-01 -6.47958100e-01 2.48535182e-02
-1.79973066e-01 2.95340896e-01 -4.16434705e-01 1.50982305e-01
1.14453554e+00 -3.16227466e-01 3.72975737e-01 2.85854936e-01
1.56201050e-01 5.24151437e-02 -1.28657949e+00 1.04562140e+00
-8.01472086e-03 4.69372809e-01 1.45800263e-01 -7.99997389e-01
8.06408644e-01 2.61556119e-01 2.20586568e-01 -1.74637690e-01
2.38265365e-01 1.62323743e-01 2.57511407e-01 -2.20588863e-01
-2.24811919e-02 -1.86071172e-01 -1.20925188e-01 5.16533136e-01
-4.59398687e-01 4.02222961e-01 -3.29988897e-01 -7.62707070e-02
1.34625387e+00 -5.69908857e-01 3.32022130e-01 -4.51922625e-01
6.01768434e-01 -4.21360105e-01 7.52562404e-01 8.29800010e-01
-3.48876655e-01 5.11260629e-01 6.93405926e-01 -2.87433863e-01
-8.02160978e-01 -9.12407756e-01 -5.27832508e-01 6.87484205e-01
2.27921724e-01 -4.84067351e-01 -1.01889813e+00 -1.11149013e+00
4.86120194e-01 8.18064094e-01 -7.72508860e-01 -2.36604035e-01
-1.20133981e-01 -9.15376484e-01 8.45043957e-01 2.25354079e-02
6.37394667e-01 -5.78611672e-01 -7.86278605e-01 -2.26805404e-01
2.14272514e-01 -9.87005413e-01 -2.16786072e-01 9.72449780e-02
-6.60907507e-01 -1.27697992e+00 -1.85559213e-01 -2.06891090e-01
8.22754264e-01 5.97175062e-02 7.56349027e-01 1.02022633e-01
-2.51788825e-01 2.36703992e-01 -1.67277887e-01 -7.63244510e-01
-6.67091906e-01 5.01602255e-02 3.12823020e-02 4.80573028e-01
7.82106817e-01 -7.07981646e-01 -5.70336580e-01 4.23190534e-01
-1.20064104e+00 -5.25336742e-01 2.93113858e-01 3.35519880e-01
3.12437117e-01 4.43975747e-01 3.66293907e-01 -1.62897050e+00
5.85902035e-01 -6.42995298e-01 -8.38252604e-01 2.32707694e-01
-4.81449991e-01 2.18627248e-02 8.20219874e-01 -3.00204009e-01
-7.04919934e-01 9.42643639e-03 -6.05511069e-02 -3.78386170e-01
-4.69946831e-01 2.74120748e-01 -5.72467387e-01 -2.74358571e-01
8.17542970e-01 2.84320135e-02 3.28325778e-01 -5.26029289e-01
-2.01452766e-02 8.61573279e-01 2.69160539e-01 -4.23923373e-01
8.85043383e-01 7.80571222e-01 4.16877121e-01 -7.37761378e-01
-4.05735850e-01 -2.27113739e-01 -3.49584341e-01 3.97415102e-01
3.77293706e-01 -6.01620257e-01 -1.02797866e+00 4.11283344e-01
-6.97703898e-01 -2.20920667e-02 -2.62491733e-01 3.14657271e-01
-2.98216462e-01 5.91886818e-01 -3.08257073e-01 -9.79748011e-01
-2.21586317e-01 -1.01275992e+00 4.12397116e-01 -1.45295793e-02
-3.87279212e-01 -1.02261531e+00 -2.40727276e-01 1.90842643e-01
3.84322137e-01 7.13558018e-01 8.43565643e-01 -1.32600605e+00
-3.41443807e-01 -8.44133496e-01 2.14552253e-01 5.45752823e-01
7.64026344e-02 -2.00836882e-02 -1.30711615e+00 -5.61334193e-01
4.62259769e-01 1.17342789e-02 4.71803397e-01 3.45674902e-02
1.56792653e+00 -8.31572890e-01 -3.45172465e-01 6.07129335e-01
1.36317921e+00 9.48395878e-02 8.13949049e-01 8.74158591e-02
2.37397268e-01 7.27083147e-01 4.07695532e-01 8.18927109e-01
-1.06650561e-01 5.38793981e-01 6.28541708e-01 -2.91318689e-02
8.37177753e-01 -1.86902002e-01 1.57301143e-01 -4.31116283e-01
4.01220918e-01 -3.00838947e-01 -7.09376097e-01 1.54224545e-01
-1.56874180e+00 -1.04950249e+00 -4.83641326e-02 3.01618648e+00
8.73481214e-01 1.62266284e-01 2.33198047e-01 3.21876973e-01
6.04958236e-01 3.82594368e-03 -5.41002870e-01 -5.07003665e-01
-2.07313865e-01 2.85177112e-01 1.03507459e+00 2.63042837e-01
-1.21824920e+00 3.87870342e-01 5.72254324e+00 6.41902685e-01
-1.18714833e+00 -1.22474983e-01 9.79596734e-01 -4.73314300e-02
-2.98439026e-01 1.91305012e-01 -6.95573270e-01 7.54747152e-01
1.11688447e+00 -6.06063187e-01 1.63880616e-01 8.87745142e-01
-1.64167032e-01 -1.40462052e-02 -1.56988573e+00 9.19177592e-01
-1.88556448e-01 -1.12422383e+00 -2.41644204e-01 9.01482761e-01
3.73861730e-01 -1.54145673e-01 4.74509716e-01 3.36522572e-02
3.59727651e-01 -1.00611281e+00 1.43020928e-01 1.02478988e-01
7.45701373e-01 -1.09542167e+00 8.09968472e-01 6.53173327e-01
-4.31834906e-01 -1.58283353e-01 -3.84346843e-01 -5.39486930e-02
-3.93220484e-01 6.50166810e-01 -7.06542611e-01 5.16775250e-01
6.80091262e-01 1.69682994e-01 -6.19010568e-01 1.01132262e+00
2.70142462e-02 8.02181900e-01 -7.85484791e-01 2.06067875e-01
7.93314949e-02 -1.47428244e-01 4.31044549e-01 9.13738906e-01
8.33897218e-02 1.02607012e-01 -2.44524926e-01 8.08567405e-01
-2.43726045e-01 5.70766516e-02 -8.08039665e-01 -2.30931714e-02
6.57968700e-01 8.05102646e-01 -2.24517792e-01 1.66130215e-02
-4.75822240e-01 8.91430020e-01 -1.12514099e-04 1.85672313e-01
-6.07172132e-01 -2.09958673e-01 1.20772576e+00 3.51880699e-01
2.25981727e-01 -2.47141551e-02 -2.78174907e-01 -1.00514317e+00
1.36800438e-01 -1.00523281e+00 9.37597156e-01 2.75776714e-01
-1.56435859e+00 1.32483035e-01 4.91163209e-02 -1.35419083e+00
-1.76316410e-01 -3.66560727e-01 -5.38750529e-01 1.04670203e+00
-1.19358671e+00 -6.87308252e-01 1.68825015e-01 6.07938707e-01
-4.60690796e-01 8.48725289e-02 1.26585412e+00 1.70717612e-01
-6.27944946e-01 1.30595851e+00 4.09223139e-01 2.52098441e-01
5.78399658e-01 -8.63886654e-01 4.86160219e-01 8.78940165e-01
2.02267513e-01 1.10789680e+00 8.83691907e-01 -3.66704881e-01
-1.33151948e+00 -9.98837352e-01 8.11979115e-01 -5.31268120e-01
1.58480063e-01 -7.23113239e-01 -1.18405426e+00 7.80686677e-01
-1.90605789e-01 4.81179178e-01 1.40776789e+00 1.23012044e-01
-5.91235518e-01 -2.37556756e-01 -2.00112510e+00 4.93519306e-01
6.30818069e-01 -5.66754937e-01 -7.06218416e-03 1.67753369e-01
3.51287365e-01 -2.34856993e-01 -9.90914404e-01 3.33999485e-01
5.39176285e-01 -1.37175417e+00 7.25358903e-01 -7.03119040e-01
-1.49222240e-01 -1.26388133e-01 -1.47834450e-01 -6.90130651e-01
2.33947963e-01 -9.58374500e-01 -2.77992427e-01 1.23952222e+00
4.08189088e-01 -1.18457103e+00 1.23131454e+00 1.58412182e+00
7.05433786e-01 -6.69047952e-01 -1.15699339e+00 -8.40069652e-01
3.78591537e-01 -3.09877664e-01 8.96589935e-01 1.26797688e+00
9.42631587e-02 -2.74527580e-01 -2.59993523e-01 5.40882468e-01
7.19348729e-01 -8.80663767e-02 1.08842862e+00 -1.40370190e+00
-5.28303385e-01 -1.01847850e-01 -6.45766318e-01 -5.96928000e-01
8.09353888e-02 -7.30199099e-01 -6.37305737e-01 -6.38625562e-01
-1.68085128e-01 -7.69801795e-01 -3.10584962e-01 5.71295738e-01
5.74637018e-03 6.33368175e-03 5.69238588e-02 1.56192288e-01
6.04504021e-03 9.37395021e-02 5.35793841e-01 1.49837956e-01
-7.28263557e-02 7.12349355e-01 -9.02815104e-01 5.66355109e-01
9.13370490e-01 -5.10353863e-01 -5.32725930e-01 1.62732795e-01
2.00100243e-01 -1.55506358e-01 3.48162681e-01 -1.17654228e+00
1.66375428e-01 -1.14959605e-01 2.06803218e-01 -1.88612163e-01
1.00313284e-01 -1.51206529e+00 5.45743942e-01 4.93742824e-01
-3.89885932e-01 -1.66226417e-01 2.97306508e-01 6.59091175e-01
3.38274874e-02 -2.26143375e-01 8.66886318e-01 4.03297767e-02
-1.27994314e-01 5.56567073e-01 9.56700146e-02 3.84022593e-02
1.47195101e+00 -4.11373198e-01 -3.88477713e-01 -3.40757251e-01
-4.93478984e-01 1.43937379e-01 9.05051291e-01 1.49898440e-01
2.30505809e-01 -7.25024343e-01 -5.37355185e-01 6.48788095e-01
2.85220325e-01 -6.97184727e-02 1.18237033e-01 4.00791079e-01
-3.08587313e-01 9.82242599e-02 -4.73728590e-02 -3.22087944e-01
-1.72208583e+00 8.08473468e-01 2.95096248e-01 -2.85594491e-03
-5.49549580e-01 8.54311824e-01 2.02309206e-01 -1.03709683e-01
3.97453994e-01 -1.01473391e-01 2.66990155e-01 -2.17862681e-01
6.57511532e-01 2.95209676e-01 1.72329947e-01 -5.48452079e-01
-3.63976210e-01 -2.19324470e-01 -4.32100177e-01 -1.58835799e-02
1.01759458e+00 1.21463366e-01 -3.81047390e-02 1.40702695e-01
1.55469918e+00 2.47046933e-01 -1.19772267e+00 -2.66371876e-01
-1.20647021e-01 -9.68483746e-01 -4.28178817e-01 -5.50380230e-01
-1.04174709e+00 8.50451648e-01 5.95502257e-01 6.30356014e-01
9.85050142e-01 -2.80921966e-01 5.89827240e-01 2.05674320e-01
7.18551934e-01 -7.51029789e-01 -8.01939785e-01 4.52168509e-02
4.75926995e-01 -1.19270110e+00 1.72241271e-01 -6.68948889e-01
-5.08620739e-01 6.38777733e-01 3.58269751e-01 4.16223891e-02
9.20640528e-01 2.91222751e-01 -4.61770734e-03 1.82411492e-01
-5.32254517e-01 4.76951033e-01 -1.79969504e-01 5.79507768e-01
4.13017757e-02 1.07683457e-01 -4.57403719e-01 7.07713366e-01
-2.70687491e-01 -5.57634123e-02 4.46903616e-01 1.20416057e+00
1.60664737e-01 -1.54843080e+00 -3.03146243e-01 6.24766886e-01
-9.79700565e-01 2.04302534e-01 -5.55378020e-01 7.63890982e-01
1.02581844e-01 8.18606675e-01 1.10899461e-02 -3.97733837e-01
2.82059669e-01 4.63751843e-03 1.30528748e-01 -5.93481302e-01
-1.02354372e+00 -6.59932196e-01 -5.67647517e-02 -4.30708617e-01
-4.37751114e-02 -8.41039419e-01 -1.04246461e+00 -8.13703060e-01
-2.01222375e-01 3.18375111e-01 5.68546474e-01 6.55680299e-01
6.67032659e-01 -3.43379021e-01 9.19745326e-01 -8.22977498e-02
-1.11965930e+00 -4.87365901e-01 -8.08521867e-01 7.51220107e-01
3.28008771e-01 -1.99169680e-01 -7.97474802e-01 -2.90036738e-01] | [5.949230670928955, 7.056524753570557] |
3ac55f96-3b58-444e-bd9c-66d72fa95935 | confess-a-framework-for-single-source-cross | null | null | https://openreview.net/forum?id=zRJu6mU2BaE | https://openreview.net/pdf?id=zRJu6mU2BaE | ConFeSS: A Framework for Single Source Cross-Domain Few-Shot Learning | Most current few-shot learning methods train a model from abundantly labeled base category data and then transfer and adapt the model to sparsely labeled novel category data. These methods mostly generalize well on novel categories from the same domain as the base categories but perform poorly for distant domain categories. In this paper, we propose a framework for few-shot learning coined as ConFeSS (Contrastive Learning and Feature Selection System) that tackles large domain shift between base and novel categories. The first step of our framework trains a feature extracting backbone with the contrastive loss on the base category data. Since the contrastive loss does not use supervision, the features can generalize better to distant target domains. For the second step, we train a masking module to select relevant features that are more suited to target domain classification. Finally, a classifier is fine-tuned along with the backbone such that the backbone produces features similar to the relevant ones. To evaluate our framework, we tested it on a recently introduced cross-domain few-shot learning benchmark. Experimental results demonstrate that our framework outperforms all meta-learning approaches and produces competitive results against recent cross-domain methods. Additional analyses are also performed to better understand our framework. | ['Fatih Porikli', 'Sungrack Yun', 'Debasmit Das'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['cross-domain-few-shot', 'cross-domain-few-shot-learning'] | ['computer-vision', 'computer-vision'] | [ 5.08107364e-01 -1.05153047e-01 -4.83560354e-01 -6.68528914e-01
-9.78575885e-01 -5.02252340e-01 8.55952144e-01 2.57121861e-01
-4.51635003e-01 8.33102643e-01 7.97865465e-02 3.14462274e-01
-1.83900177e-01 -9.55106497e-01 -6.54432476e-01 -6.44398570e-01
9.40530598e-02 5.61862826e-01 8.15270662e-01 -3.21258098e-01
3.09004694e-01 1.14466175e-01 -1.76272595e+00 5.55355132e-01
8.22302818e-01 9.80362356e-01 2.78599650e-01 1.39422759e-01
-2.57300436e-01 5.14016926e-01 -4.34232146e-01 -2.34773353e-01
3.25888097e-01 -5.95386803e-01 -7.78249800e-01 1.19633824e-01
4.09672171e-01 -4.29322645e-02 -2.66218614e-02 8.69638264e-01
4.43246752e-01 6.05314076e-01 1.04385543e+00 -1.13872612e+00
-5.85971713e-01 4.51338828e-01 -2.15569928e-01 2.56144524e-01
2.86659002e-01 1.45141110e-01 1.07510018e+00 -1.27433932e+00
8.73407662e-01 9.91436958e-01 8.17651391e-01 8.23828578e-01
-1.31633818e+00 -6.35523140e-01 3.24902475e-01 4.58273530e-01
-1.35095847e+00 -4.75724787e-01 8.61549973e-01 -3.75811577e-01
9.16500866e-01 -1.58247739e-01 2.34912917e-01 1.32785118e+00
-7.44449869e-02 5.93930423e-01 1.06867313e+00 -5.19633174e-01
8.04763317e-01 5.85789442e-01 4.41847205e-01 2.42609337e-01
-1.60679407e-02 2.65778482e-01 -5.07113338e-01 -2.07472548e-01
2.55044103e-02 2.87554413e-01 -1.65454801e-02 -9.90850687e-01
-9.67623472e-01 1.08192873e+00 5.76242268e-01 5.14361620e-01
-2.87846506e-01 -4.31324422e-01 6.07817471e-01 6.34015262e-01
6.66855097e-01 5.19149721e-01 -5.88478267e-01 1.02933787e-01
-9.22165334e-01 2.37296522e-01 7.11890936e-01 1.11910844e+00
1.04234254e+00 -2.73129165e-01 -4.37512904e-01 1.25694895e+00
-1.40670300e-01 5.99661060e-02 8.47012877e-01 -4.58569348e-01
3.56426775e-01 5.73646903e-01 -1.23400204e-01 -3.97361845e-01
-2.81282872e-01 -4.52880889e-01 -5.35671294e-01 1.88611448e-02
1.79622859e-01 -8.35876092e-02 -1.09588480e+00 1.62799323e+00
3.26198906e-01 3.34726453e-01 2.40194112e-01 6.56631052e-01
8.13274205e-01 6.76413894e-01 1.95603728e-01 -7.73007870e-02
1.18651867e+00 -9.44691777e-01 -2.11501747e-01 -4.51998353e-01
7.48944521e-01 -3.89200270e-01 1.16506052e+00 8.49825516e-02
-7.11667657e-01 -7.32728302e-01 -1.13770211e+00 2.01898769e-01
-7.83820331e-01 -3.88826668e-01 2.45946988e-01 4.02839690e-01
-5.43376088e-01 8.65694821e-01 -3.14320952e-01 -7.70575941e-01
6.67476416e-01 6.78245798e-02 -3.07308644e-01 -5.14042497e-01
-1.49293888e+00 1.05158079e+00 7.78721035e-01 -8.61774743e-01
-1.01258063e+00 -1.07835615e+00 -1.01811099e+00 1.61815375e-01
3.46832961e-01 -6.24650478e-01 1.21027625e+00 -9.93403256e-01
-1.39331591e+00 1.01493669e+00 5.97595498e-02 -6.86633348e-01
3.01378429e-01 -3.43780033e-02 -4.77056503e-01 -6.41266815e-04
4.22439247e-01 5.95603347e-01 9.92496073e-01 -1.07734752e+00
-1.06756520e+00 -3.02660435e-01 -1.52292192e-01 1.07379533e-01
-4.55495775e-01 -2.16363102e-01 3.14646773e-02 -6.87093437e-01
-1.15281217e-01 -7.22956657e-01 -5.85759655e-02 -3.59966494e-02
-2.19620597e-02 -4.70406532e-01 8.26403022e-01 -3.22285108e-02
1.02170181e+00 -2.30198431e+00 2.04757765e-01 6.06635995e-02
-1.29259810e-01 3.10700804e-01 -3.49946260e-01 4.62993562e-01
-2.42888600e-01 -4.81272221e-01 -3.77486199e-01 -7.95043856e-02
4.01749313e-02 8.46276507e-02 -5.38561463e-01 2.76979297e-01
3.47962111e-01 7.46886969e-01 -9.45586145e-01 -2.90241033e-01
2.30429783e-01 1.41950306e-02 -5.90905309e-01 2.60605872e-01
-2.18562722e-01 1.19173296e-01 -1.96695223e-01 6.74481153e-01
6.75983369e-01 -1.33942902e-01 -9.21772141e-03 1.56629514e-02
5.72852232e-02 1.52672052e-01 -9.72030342e-01 1.83525872e+00
-6.51463747e-01 2.19363689e-01 -5.29529631e-01 -1.45234287e+00
1.24615633e+00 -4.11233790e-02 1.70593008e-01 -7.25124478e-01
-2.66841631e-02 3.09113234e-01 -1.57987654e-01 -2.84174114e-01
1.16232991e-01 -7.34874308e-01 -4.56224471e-01 3.10388118e-01
7.65951276e-01 -2.74939060e-01 1.66858405e-01 5.03705516e-02
1.09820712e+00 1.51750088e-01 7.48764873e-01 -1.30896479e-01
4.79300588e-01 2.10918471e-01 6.90522969e-01 8.89672756e-01
-4.42661405e-01 5.41297436e-01 2.02058062e-01 -5.01549304e-01
-1.03912032e+00 -1.37531829e+00 -1.96680352e-01 1.75421047e+00
1.79869860e-01 -1.18350647e-01 -3.61156166e-01 -1.25145304e+00
2.00929254e-01 1.06289458e+00 -8.64590108e-01 -7.25849628e-01
-1.38247758e-01 -5.35616755e-01 9.77747440e-02 4.89593387e-01
3.08629006e-01 -1.09541130e+00 -4.84583437e-01 2.67602623e-01
1.18391819e-01 -7.84275532e-01 -2.23551795e-01 7.32574761e-01
-8.13027084e-01 -1.06598234e+00 -1.02236927e+00 -1.21705198e+00
3.22138786e-01 3.40001196e-01 1.04370224e+00 -5.56847036e-01
-2.86612421e-01 1.81519166e-01 -7.36880839e-01 -3.25129479e-01
-2.30671704e-01 2.44255066e-01 2.31923312e-02 1.02669857e-01
9.67825234e-01 -6.11462831e-01 -1.60157219e-01 3.51538211e-01
-6.44105434e-01 -4.74529803e-01 5.96842766e-01 1.32680261e+00
5.52932441e-01 -1.81611195e-01 1.01064432e+00 -1.15270972e+00
3.95673037e-01 -9.45329070e-01 -3.09926063e-01 3.00779492e-01
-5.12180388e-01 9.01109129e-02 7.88831174e-01 -6.82301998e-01
-1.22049427e+00 1.23004884e-01 1.79855630e-01 -2.99757332e-01
-4.34221447e-01 2.66461819e-01 -2.41262376e-01 1.02635600e-01
1.10034013e+00 4.29076314e-01 -1.99655816e-01 -6.17316663e-01
4.95120883e-01 7.82942057e-01 2.78724879e-01 -3.39540213e-01
8.55936468e-01 5.34497917e-01 -3.93775463e-01 -7.40673244e-01
-1.26637495e+00 -8.08305919e-01 -9.34782147e-01 1.34922236e-01
4.42646235e-01 -8.93155932e-01 2.78559059e-01 2.44723409e-01
-7.12584257e-01 -2.29090199e-01 -7.72975087e-01 4.71750528e-01
-8.96163464e-01 7.08220676e-02 -1.63078621e-01 -4.17936325e-01
-1.33080885e-01 -5.59077859e-01 9.51437593e-01 2.09449828e-01
-3.40008110e-01 -1.05626738e+00 4.04249072e-01 -9.40205604e-02
4.58900988e-01 8.77082869e-02 9.57833648e-01 -1.33315289e+00
-4.89382818e-03 -1.45442650e-01 -1.95938051e-01 3.59233499e-01
2.22558409e-01 -6.41761482e-01 -1.11131191e+00 -3.32554549e-01
-5.90956472e-02 -6.64043725e-01 1.40343988e+00 1.67420760e-01
8.78211796e-01 6.75013810e-02 -5.53352058e-01 6.08969271e-01
1.45878422e+00 1.81282520e-01 4.18781340e-01 5.62144041e-01
1.79886714e-01 6.91473424e-01 1.04481304e+00 5.20943284e-01
3.62031572e-02 7.55267143e-01 2.91103441e-02 2.84744471e-01
-2.78852731e-01 -2.35853344e-01 3.54940414e-01 4.44196105e-01
3.94881725e-01 2.91487485e-01 -7.43420124e-01 9.50932920e-01
-1.72493958e+00 -1.21495283e+00 5.63161016e-01 2.08962965e+00
9.42777574e-01 4.12744850e-01 3.86681825e-01 4.46517318e-02
9.15075779e-01 -1.49764791e-02 -8.26461196e-01 -4.04437900e-01
5.04698642e-02 5.80685139e-01 9.47211832e-02 1.53256968e-01
-1.41908896e+00 1.07387733e+00 5.33195496e+00 9.89884496e-01
-9.38469589e-01 3.19258034e-01 2.08460182e-01 -2.38540977e-01
3.26610310e-03 3.67481858e-02 -9.47983742e-01 4.44929898e-01
7.96349287e-01 -4.92741287e-01 5.62841631e-02 1.27050614e+00
-3.19050491e-01 1.61419630e-01 -1.33485448e+00 7.51364172e-01
2.50742167e-01 -1.17448056e+00 1.69043671e-02 -2.24602312e-01
8.40079427e-01 1.04041211e-01 8.66081119e-02 1.02594197e+00
3.53185624e-01 -5.76585174e-01 3.92964691e-01 3.85252565e-01
7.81445265e-01 -8.57253611e-01 7.09235609e-01 4.09417391e-01
-1.03564155e+00 -5.29779077e-01 -8.90362501e-01 -8.10406506e-02
-7.76016116e-02 5.29182911e-01 -9.77301478e-01 5.00234604e-01
6.06779337e-01 1.15589345e+00 -6.50625706e-01 1.25284588e+00
-8.73085782e-02 3.09030354e-01 4.27630320e-02 -4.10508662e-02
3.41861784e-01 2.20507547e-01 5.63357413e-01 1.16822636e+00
3.68636698e-01 -3.45154926e-02 3.29203993e-01 7.35006630e-01
-5.51776178e-02 1.57347247e-01 -9.52797651e-01 2.54222542e-01
5.64488471e-01 1.08676672e+00 -3.79423648e-01 -7.00643063e-01
-6.58694565e-01 1.07067573e+00 5.35396636e-01 2.38749340e-01
-6.57178462e-01 -7.94578195e-01 5.83551705e-01 1.03733733e-01
7.12294102e-01 4.98911500e-01 -1.13620207e-01 -1.30398190e+00
-2.35125825e-01 -5.60369670e-01 8.79726529e-01 -3.41554552e-01
-2.02676463e+00 4.71598446e-01 8.54067355e-02 -1.65741670e+00
-3.32082719e-01 -4.45955515e-01 -7.40381539e-01 6.86095953e-01
-1.77754581e+00 -1.18405938e+00 -1.86906770e-01 8.22172701e-01
9.19065893e-01 -5.88340998e-01 1.08399582e+00 1.62863076e-01
-2.00810954e-01 7.38678217e-01 4.47806388e-01 -5.79847246e-02
1.15500402e+00 -1.15966976e+00 3.11628282e-01 5.53176463e-01
-4.41394188e-02 4.88483042e-01 5.99074066e-01 -6.14243865e-01
-7.33040690e-01 -1.57740092e+00 8.90216470e-01 -3.26640517e-01
6.90746784e-01 -4.60985988e-01 -1.19618726e+00 6.14125907e-01
5.32262679e-03 3.52199942e-01 8.92416298e-01 3.25116903e-01
-7.57921159e-01 -1.81634173e-01 -1.28577673e+00 1.78169787e-01
1.05945289e+00 -3.61312300e-01 -1.31361151e+00 3.29949558e-01
6.60109758e-01 2.41097450e-01 -6.79380953e-01 3.39074224e-01
4.44580823e-01 -8.94603670e-01 9.42748129e-01 -9.66176569e-01
3.19806039e-01 -4.59509492e-02 -3.64010096e-01 -1.82165146e+00
-6.04574561e-01 -1.14390485e-01 -9.26018655e-02 1.28035367e+00
3.93976152e-01 -5.25464535e-01 6.73144519e-01 2.37590019e-02
-1.30305663e-01 -4.24012303e-01 -9.05287087e-01 -1.37572432e+00
4.06665772e-01 -1.98660418e-02 3.60083729e-01 1.11319315e+00
2.68642694e-01 5.43605924e-01 -2.52479076e-01 -3.02293032e-01
8.81162643e-01 5.22988141e-01 6.75925732e-01 -1.61442173e+00
-2.22531110e-01 -2.63568252e-01 -5.13721764e-01 -6.26137733e-01
3.69713426e-01 -1.21444356e+00 1.03275336e-01 -1.19037199e+00
5.80638349e-01 -2.80074477e-01 -6.60343289e-01 4.83486325e-01
-1.74360409e-01 1.14429601e-01 2.53255039e-01 -8.91506299e-03
-9.67901766e-01 7.65101969e-01 7.63489902e-01 -2.91365027e-01
-4.05247658e-01 2.22968385e-01 -7.58955479e-01 6.36637568e-01
7.42857695e-01 -6.83153391e-01 -3.91329885e-01 1.78689033e-01
-4.17997390e-01 -4.82387871e-01 2.61280388e-01 -1.24490905e+00
2.13808939e-01 -2.40637660e-01 5.12413263e-01 -3.94658208e-01
3.02178293e-01 -7.47459590e-01 -3.61967981e-01 5.43002009e-01
-5.36808372e-01 -7.24541664e-01 -6.34628832e-02 6.83231413e-01
-2.60075241e-01 -4.28081781e-01 1.36242914e+00 -1.65646076e-01
-1.37063885e+00 2.34506339e-01 -2.46071313e-02 4.22793448e-01
1.46235704e+00 -2.75886446e-01 -1.45962253e-01 -6.42054901e-02
-1.13793766e+00 9.98109505e-02 4.97389495e-01 7.22871900e-01
6.19255960e-01 -1.50126004e+00 -7.43588150e-01 3.34039003e-01
9.79591072e-01 -3.52979362e-01 2.90534824e-01 4.15681124e-01
2.13714123e-01 3.68787229e-01 -5.24936140e-01 -5.74363410e-01
-9.83278036e-01 1.10780120e+00 2.04247847e-01 -1.90722764e-01
-6.33976102e-01 8.72043073e-01 3.37913036e-01 -7.56528616e-01
2.03772649e-01 1.25370808e-02 -3.67954731e-01 5.03023863e-01
8.12581539e-01 3.87861520e-01 1.41424254e-01 -4.35412169e-01
-5.38667917e-01 5.41588664e-01 -3.52215111e-01 1.36108339e-01
1.69636178e+00 -1.68705165e-01 3.93763065e-01 7.39028215e-01
1.32997274e+00 -5.47009945e-01 -1.38408923e+00 -8.36376250e-01
4.27451253e-01 -4.21606988e-01 -3.59686464e-01 -8.90408337e-01
-5.25891185e-01 9.58452165e-01 7.69170463e-01 -5.87044924e-04
1.05808425e+00 2.56736249e-01 5.95271647e-01 2.82064736e-01
4.70252782e-01 -1.48234081e+00 2.59158760e-01 8.05467606e-01
5.99978983e-01 -1.37781417e+00 -1.75421789e-01 -1.66183248e-01
-8.18849146e-01 1.01056790e+00 7.88598537e-01 -4.28470165e-01
7.66845584e-01 -1.90106347e-01 -2.74293333e-01 -7.71378949e-02
-9.79857385e-01 -4.87340897e-01 3.12374949e-01 1.10533094e+00
9.22962874e-02 -1.01782955e-01 -1.83120713e-01 9.47229564e-01
1.12282038e-01 1.67701915e-01 2.49010995e-01 9.35098529e-01
-9.40460443e-01 -1.02119434e+00 -1.92120224e-01 5.98254025e-01
-2.37078331e-02 2.87345727e-03 -3.99155736e-01 6.33407295e-01
2.92044938e-01 8.45448136e-01 1.15394428e-01 -3.63128036e-01
7.38700926e-01 6.17531121e-01 2.99749434e-01 -1.29455006e+00
-4.69688207e-01 -2.59973943e-01 -8.87613222e-02 -5.02598226e-01
-2.03357518e-01 -6.53413475e-01 -8.36190045e-01 1.34957209e-01
-2.00998262e-01 1.87510103e-01 3.22329164e-01 1.00078118e+00
3.55327219e-01 4.10883158e-01 8.70719552e-01 -7.16844559e-01
-9.48600054e-01 -9.92693067e-01 -8.02991688e-01 6.99018776e-01
3.14775318e-01 -1.07908607e+00 -4.86047447e-01 -5.02887880e-03] | [10.050186157226562, 3.037947177886963] |
00883efc-8471-4a25-8b62-de8580e40edd | bootstrapped-edge-count-tests-for | 2304.13848 | null | https://arxiv.org/abs/2304.13848v1 | https://arxiv.org/pdf/2304.13848v1.pdf | Bootstrapped Edge Count Tests for Nonparametric Two-Sample Inference Under Heterogeneity | Nonparametric two-sample testing is a classical problem in inferential statistics. While modern two-sample tests, such as the edge count test and its variants, can handle multivariate and non-Euclidean data, contemporary gargantuan datasets often exhibit heterogeneity due to the presence of latent subpopulations. Direct application of these tests, without regulating for such heterogeneity, may lead to incorrect statistical decisions. We develop a new nonparametric testing procedure that accurately detects differences between the two samples in the presence of unknown heterogeneity in the data generation process. Our framework handles this latent heterogeneity through a composite null that entertains the possibility that the two samples arise from a mixture distribution with identical component distributions but with possibly different mixing weights. In this regime, we study the asymptotic behavior of weighted edge count test statistic and show that it can be effectively re-calibrated to detect arbitrary deviations from the composite null. For practical implementation we propose a Bootstrapped Weighted Edge Count test which involves a bootstrap-based calibration procedure that can be easily implemented across a wide range of heterogeneous regimes. A comprehensive simulation study and an application to detecting aberrant user behaviors in online games demonstrates the excellent non-asymptotic performance of the proposed test. | ['Gourab Mukherjee', 'Bhaswar B. Bhattacharya', 'Trambak Banerjee'] | 2023-04-26 | null | null | null | null | ['hypothesis-testing', 'hypothesis-testing'] | ['methodology', 'miscellaneous'] | [ 2.20918715e-01 -2.17607155e-01 -2.69680917e-01 -2.45559901e-01
-5.95697284e-01 -4.70508099e-01 3.94286752e-01 2.59282082e-01
-3.41593891e-01 1.23324609e+00 -2.40796313e-01 -7.73440361e-01
-4.10642684e-01 -9.52625871e-01 -5.09036362e-01 -6.78101778e-01
-1.81516364e-01 7.54532814e-01 1.33536488e-01 3.31420183e-01
3.14989746e-01 1.66318312e-01 -1.58095062e+00 -3.46673995e-01
1.44459414e+00 5.43061674e-01 -3.38114798e-01 8.09011579e-01
1.55887082e-01 3.04735363e-01 -8.69755805e-01 -2.78653294e-01
2.95365125e-01 -6.37140751e-01 -1.49373114e-01 1.40002355e-01
1.00440212e-01 -4.56083268e-01 3.12002599e-01 1.27410471e+00
5.96663654e-01 1.10769302e-01 1.07871819e+00 -1.66316676e+00
-3.88257414e-01 4.32335734e-01 -1.03051102e+00 1.15216032e-01
1.47114515e-01 -9.53029096e-02 7.53696799e-01 -5.55840313e-01
8.96587521e-02 1.11987197e+00 7.12342501e-01 2.54965033e-02
-1.58149421e+00 -7.27394342e-01 -3.32888544e-01 1.58691779e-02
-1.58229423e+00 -1.63352340e-01 3.42796326e-01 -5.33373654e-01
2.73670815e-02 4.76048797e-01 5.02998054e-01 1.12313867e+00
3.08118969e-01 4.67319906e-01 1.31956804e+00 -3.01666737e-01
7.87501097e-01 2.18739346e-01 2.58144051e-01 2.91053236e-01
1.11035526e+00 2.01861486e-01 -1.16347842e-01 -8.63230646e-01
8.85607481e-01 4.23735678e-02 -1.47477865e-01 -4.97631878e-01
-8.10090005e-01 1.06353045e+00 -5.19530952e-01 -3.55900750e-02
-4.09827173e-01 -1.64982364e-01 2.90479928e-01 3.32411259e-01
6.97533131e-01 -9.78756920e-02 -2.66007334e-01 -2.96778679e-01
-9.83226836e-01 2.62264609e-01 1.08317554e+00 7.09890902e-01
3.28437030e-01 4.94103022e-02 -5.02078831e-01 5.23187816e-01
-4.44740467e-02 5.67221761e-01 5.53813577e-01 -8.54328930e-01
1.30160913e-01 2.42751926e-01 5.25288284e-01 -7.12763309e-01
-1.84441715e-01 -4.88072515e-01 -9.88123655e-01 2.51374930e-01
8.75930727e-01 -4.59227592e-01 -4.11156803e-01 1.77726495e+00
5.04629076e-01 5.78948915e-01 -2.28037253e-01 3.56410563e-01
1.16986416e-01 1.19681530e-01 5.59785180e-02 -5.03889024e-01
9.86508191e-01 -1.83977678e-01 -5.88581860e-01 2.26800546e-01
4.29799646e-01 -2.88032442e-01 1.08485198e+00 4.30006981e-01
-1.01875150e+00 -1.52016789e-01 -8.39455962e-01 4.97915238e-01
-3.21982279e-02 -2.80135185e-01 4.21811551e-01 1.10674548e+00
-9.31494594e-01 4.73110616e-01 -5.16115367e-01 -5.11234879e-01
2.61197567e-01 3.06371123e-01 -2.18653888e-01 -1.20851081e-02
-9.54424202e-01 3.33620489e-01 1.77974507e-01 3.40404138e-02
-3.49701643e-01 -6.49202406e-01 -6.41257286e-01 3.44551861e-01
5.45830011e-01 -6.08138621e-01 1.03157985e+00 -9.01646793e-01
-1.31565213e+00 4.58680481e-01 -2.13053480e-01 -2.15091169e-01
9.81606960e-01 4.48698878e-01 -3.35689813e-01 -1.18217848e-01
4.68046367e-01 -3.65503252e-01 7.68598795e-01 -8.66351128e-01
-6.50871575e-01 -4.53444004e-01 -4.07027036e-01 6.74299002e-02
-1.03285396e-02 -2.40132421e-01 8.17388967e-02 -6.27438903e-01
-6.74606711e-02 -6.76433086e-01 -1.26476586e-01 -3.31543118e-01
-5.25630057e-01 -2.68081147e-02 1.83188409e-01 -3.77205163e-01
1.11859345e+00 -2.18264532e+00 -3.57692510e-01 7.86114693e-01
2.56148696e-01 -4.12469119e-01 -3.85891385e-02 2.90973306e-01
9.31121930e-02 1.31592795e-01 -3.51163268e-01 1.12427166e-02
2.01587677e-01 1.94142070e-02 1.32476091e-02 1.01131845e+00
-8.61651599e-02 4.20471907e-01 -7.95630276e-01 -2.55743951e-01
-1.66456569e-02 2.85738688e-02 -5.78031361e-01 1.36536658e-01
3.52161258e-01 3.33230555e-01 -1.96956143e-01 4.70548391e-01
9.68466878e-01 -2.63076216e-01 4.49199557e-01 5.40954411e-01
-9.69629064e-02 -2.28311136e-01 -1.30694866e+00 3.72043043e-01
-4.55671027e-02 4.93318349e-01 9.12942365e-02 -1.33105075e+00
7.91701376e-01 1.45179749e-01 2.07482651e-01 -3.57564539e-01
3.31416517e-01 4.23020422e-01 3.13154787e-01 -3.51904839e-01
3.41753036e-01 -4.28001940e-01 -1.68796569e-01 4.46565539e-01
-2.39522278e-01 1.64824978e-01 1.59052104e-01 -2.58078337e-01
1.25951886e+00 -2.64202952e-01 7.20646441e-01 -6.03285551e-01
3.25084329e-01 -4.94599164e-01 7.02828705e-01 1.39985538e+00
-2.67902523e-01 4.01130110e-01 1.15600038e+00 2.32853085e-01
-8.17583144e-01 -1.58805573e+00 -5.27128100e-01 6.97742283e-01
9.11753774e-02 3.34107220e-01 -4.96153206e-01 -3.93040240e-01
4.43911195e-01 9.13867533e-01 -9.02967691e-01 -2.77083248e-01
2.70629078e-01 -1.08452892e+00 4.20573354e-01 2.98604757e-01
3.36827666e-01 -5.37828386e-01 -4.37375307e-01 6.52436614e-02
-7.13457987e-02 -6.77642822e-01 -2.34169170e-01 -7.39919301e-03
-9.25075710e-01 -1.42001724e+00 -5.91678858e-01 -3.83564532e-01
6.89855993e-01 1.24720439e-01 8.69057775e-01 -1.32311538e-01
-1.11556843e-01 4.47706372e-01 -1.89529657e-01 -3.23314130e-01
-5.16932845e-01 -4.52210069e-01 1.27924919e-01 2.23459378e-01
7.73667276e-01 -4.81314689e-01 -4.67445582e-01 4.38782841e-01
-9.80552256e-01 -4.60958660e-01 4.75056440e-01 8.76497746e-01
2.55839765e-01 3.52922618e-01 1.00062609e+00 -9.95314300e-01
1.03823769e+00 -1.11084712e+00 -6.93767309e-01 2.84308553e-01
-5.23732901e-01 -1.42594585e-02 4.46996957e-01 -7.85284281e-01
-8.26283276e-01 -7.99597859e-01 4.48724598e-01 -1.29370674e-01
-1.87097624e-01 7.69945979e-01 -2.59387434e-01 2.73959339e-01
4.92682487e-01 4.96170856e-02 3.67826730e-01 -2.72836089e-01
-9.59692225e-02 8.29063952e-01 5.50536811e-01 -6.60029113e-01
6.70217752e-01 4.09443378e-01 3.81492019e-01 -1.04983878e+00
-2.51683295e-01 -4.18449014e-01 -2.54781812e-01 -1.43479789e-02
4.74816054e-01 -7.56996870e-01 -1.02114332e+00 5.18390179e-01
-4.68333453e-01 -4.60742891e-01 -3.99184734e-01 7.51997709e-01
-5.99030614e-01 5.55111825e-01 -3.12192202e-01 -1.28253353e+00
2.29571894e-01 -8.48071754e-01 7.13466525e-01 1.42828852e-01
-2.73266852e-01 -1.30607224e+00 2.08111659e-01 -5.58343120e-02
1.18484348e-01 5.12278438e-01 1.02012122e+00 -1.08708417e+00
9.97703746e-02 -5.83240747e-01 -7.70569593e-02 4.07375917e-02
4.25636411e-01 3.19857657e-01 -5.45532227e-01 -4.78556573e-01
6.37927130e-02 1.52117938e-01 3.37829590e-01 8.37661445e-01
1.10723996e+00 -2.55069435e-01 -1.28756568e-01 1.66848987e-01
1.17153692e+00 1.19105808e-01 5.16423047e-01 1.47108257e-01
1.24076955e-01 7.48331428e-01 5.11136234e-01 1.01691508e+00
2.88675874e-01 3.60092014e-01 1.01882862e-02 4.85050417e-02
6.76420033e-01 -1.40323147e-01 3.53668958e-01 3.81544203e-01
1.29732952e-01 -3.27870369e-01 -6.13983631e-01 4.49908674e-01
-1.85011041e+00 -1.24166775e+00 -4.25102293e-01 3.13416719e+00
7.44151711e-01 1.35234922e-01 6.20923638e-01 1.83768868e-01
1.33932579e+00 -5.97377539e-01 -8.00328434e-01 -4.19649631e-01
-1.71479821e-01 3.16637009e-01 6.03534281e-01 4.00168389e-01
-5.52650213e-01 9.78782922e-02 6.90774918e+00 8.61569643e-01
-6.62327409e-01 -1.21199796e-02 8.04102719e-01 1.05059654e-01
-4.25438374e-01 5.98294847e-02 -3.10477436e-01 7.40689278e-01
1.10292947e+00 -8.05013657e-01 -3.85850891e-02 5.25142431e-01
5.73916137e-01 -6.53246582e-01 -9.73548114e-01 7.34202504e-01
-1.30035862e-01 -5.83802164e-01 -2.53538013e-01 5.47753632e-01
7.16309667e-01 -5.48295617e-01 2.36860737e-01 3.42472494e-01
4.46514696e-01 -8.17148209e-01 5.90781033e-01 4.44318801e-01
9.26150739e-01 -1.00313067e+00 9.75408852e-01 4.29029226e-01
-7.08859682e-01 -1.24819152e-01 -4.79274660e-01 -4.33095485e-01
-2.37084970e-01 1.03638518e+00 -6.23352945e-01 2.38226578e-01
3.98672909e-01 1.68105140e-01 -4.91209060e-01 1.38677120e+00
5.38927056e-02 8.83357406e-01 -2.91857660e-01 -3.68418694e-02
-3.12716186e-01 -8.10941398e-01 5.53316832e-01 5.72595656e-01
9.36659634e-01 -1.96887091e-01 -1.62651584e-01 9.24506187e-01
1.12808496e-01 2.00374067e-01 -5.73292911e-01 -4.22337107e-05
7.25017726e-01 8.39221835e-01 -9.28867042e-01 -4.19407308e-01
-6.60075247e-01 6.76527739e-01 5.21866977e-02 6.24692440e-01
-9.55158353e-01 -3.67741764e-01 5.39576054e-01 2.96285063e-01
1.96710214e-01 1.02911284e-02 -4.45963621e-01 -1.29569745e+00
-9.42190215e-02 -6.85033560e-01 5.20753026e-01 -2.74395466e-01
-1.72524571e+00 -1.70026958e-01 3.32668304e-01 -1.16131592e+00
-4.59306777e-01 -4.19342309e-01 -9.12957907e-01 8.09637547e-01
-9.78242278e-01 -2.48485625e-01 -1.83955327e-01 5.84826767e-01
-2.69088540e-02 -4.81606368e-03 4.38635826e-01 2.96759848e-02
-6.74680054e-01 6.86856329e-01 7.22252011e-01 -1.85707033e-01
8.26008737e-01 -1.41889238e+00 -1.48572903e-02 8.27789068e-01
-4.76396412e-01 5.13593614e-01 9.51096117e-01 -1.01202250e+00
-7.83964753e-01 -8.71650636e-01 3.62466991e-01 -7.45592341e-02
9.68098283e-01 -4.02709752e-01 -8.06079507e-01 5.36881268e-01
-2.20497504e-01 -2.13797733e-01 1.20976627e+00 2.29925424e-01
-5.43238595e-02 1.99976146e-01 -1.43834901e+00 8.87627602e-01
6.51592672e-01 -1.58538312e-01 -3.26948196e-01 -2.49853358e-02
4.87760007e-02 3.10761809e-01 -7.35349059e-01 4.09234911e-01
6.21451735e-01 -1.34324074e+00 3.70190799e-01 -5.94246864e-01
-1.94290355e-02 -1.43367454e-01 -8.46741162e-03 -1.09609485e+00
-1.92587242e-01 -7.00569868e-01 8.47757459e-02 1.01113749e+00
1.88963011e-01 -1.23545849e+00 5.72411358e-01 7.37283707e-01
5.89164138e-01 -1.61994621e-01 -1.20973372e+00 -9.88373518e-01
2.23850638e-01 -2.27196336e-01 5.14641106e-01 9.34140205e-01
2.50602633e-01 1.39995636e-02 -2.40652040e-01 1.57187790e-01
1.16733086e+00 7.73678795e-02 8.49707365e-01 -1.75625002e+00
-5.54854631e-01 -4.46382731e-01 -5.89492202e-01 -6.39168262e-01
3.93111557e-01 -3.79025966e-01 8.41934085e-02 -8.33081365e-01
6.17978334e-01 -1.23901978e-01 -3.10967006e-02 -9.88264568e-03
-4.42310929e-01 2.54244059e-01 -6.42513692e-01 -1.44374907e-01
-1.22119054e-01 5.62381327e-01 7.27223694e-01 3.02514851e-01
-3.46560419e-01 4.35087472e-01 -8.56860042e-01 6.10016465e-01
8.77462029e-01 -4.16989386e-01 -5.45647562e-01 5.49004257e-01
1.33456901e-01 3.50272775e-01 4.43206847e-01 -7.50212848e-01
-1.32651217e-02 -2.70861238e-01 3.06765586e-01 -2.80547291e-01
-2.52844661e-01 -5.68183362e-01 3.43907595e-01 4.38735306e-01
-1.49373010e-01 7.18231201e-02 1.66845739e-01 8.01773131e-01
-2.25808583e-02 -2.87466288e-01 6.48033977e-01 3.03534865e-01
6.94001168e-02 1.21210977e-01 -8.38009417e-01 2.48282179e-01
1.13784158e+00 -4.02736902e-01 -3.79347414e-01 -7.20655739e-01
-5.45528710e-01 1.69705898e-01 5.46515524e-01 -9.70818847e-02
4.22502071e-01 -1.23119617e+00 -8.93643439e-01 2.95611858e-01
-7.00001046e-02 -4.96670246e-01 3.01960796e-01 1.33596015e+00
-2.84705907e-01 -1.97054327e-01 -1.43001527e-01 -4.15485024e-01
-1.04562461e+00 4.32196379e-01 2.12602913e-01 5.68470359e-02
-1.88694507e-01 1.48773417e-01 4.33899909e-01 -3.37421566e-01
-2.21877143e-01 -2.09653318e-01 5.25765717e-02 2.60896757e-02
4.75110888e-01 8.69860768e-01 -3.15943539e-01 -3.92500162e-01
-1.04009777e-01 1.04768343e-01 2.98753589e-01 -3.53472024e-01
9.54680324e-01 -3.17481279e-01 -3.17554444e-01 9.62379277e-01
8.30885291e-01 3.85209590e-01 -1.07167292e+00 1.24696210e-01
3.07190772e-02 -7.00700164e-01 -3.67063612e-01 -2.95486331e-01
-4.85708117e-01 3.47920805e-01 2.45633796e-01 7.59420156e-01
9.71734345e-01 -3.38123649e-01 -9.01427120e-02 -1.25791490e-01
3.09897661e-01 -8.81638229e-01 -3.41758996e-01 4.92053740e-02
3.11157227e-01 -1.15277886e+00 -3.54985259e-02 -2.95082182e-01
-3.01262140e-01 9.63327527e-01 3.77818793e-01 -1.93864971e-01
7.20426500e-01 2.71794111e-01 -3.12147796e-01 2.12912887e-01
-5.66187680e-01 -5.03266230e-02 2.60277510e-01 6.53478503e-01
3.51489216e-01 3.57201934e-01 -8.66059840e-01 6.37587070e-01
-1.58905774e-01 -5.83114699e-02 1.17408514e+00 6.22082055e-01
-2.23485231e-01 -8.89889956e-01 -4.68394697e-01 9.57904279e-01
-5.39996028e-01 -9.82956775e-03 -1.39426380e-01 1.07217491e+00
-4.07746732e-01 1.12912524e+00 5.81659675e-01 5.34756407e-02
1.51909575e-01 -1.92058296e-03 3.78314972e-01 -4.39901471e-01
4.21246514e-02 2.91152745e-01 -1.97126538e-01 -2.10973173e-01
-3.45269918e-01 -1.00531387e+00 -6.51262462e-01 -8.83589506e-01
-4.33427930e-01 6.03638768e-01 1.44382626e-01 9.03076947e-01
8.87489691e-02 1.52688459e-01 6.28473401e-01 -4.68594998e-01
-1.00345683e+00 -9.92141306e-01 -1.44866109e+00 1.89003125e-01
1.82581469e-01 -8.27196181e-01 -9.30373251e-01 -5.03287375e-01] | [7.277248859405518, 4.389183521270752] |
ebd329aa-10e0-40e9-8769-edb8cebc9232 | mssrnet-manipulating-sequential-style | 2306.07994 | null | https://arxiv.org/abs/2306.07994v1 | https://arxiv.org/pdf/2306.07994v1.pdf | MSSRNet: Manipulating Sequential Style Representation for Unsupervised Text Style Transfer | Unsupervised text style transfer task aims to rewrite a text into target style while preserving its main content. Traditional methods rely on the use of a fixed-sized vector to regulate text style, which is difficult to accurately convey the style strength for each individual token. In fact, each token of a text contains different style intensity and makes different contribution to the overall style. Our proposed method addresses this issue by assigning individual style vector to each token in a text, allowing for fine-grained control and manipulation of the style strength. Additionally, an adversarial training framework integrated with teacher-student learning is introduced to enhance training stability and reduce the complexity of high-dimensional optimization. The results of our experiments demonstrate the efficacy of our method in terms of clearly improved style transfer accuracy and content preservation in both two-style transfer and multi-style transfer settings. | ['Qi Liu', 'Zhou Zhao', 'Yazheng Yang'] | 2023-06-12 | null | null | null | null | ['style-transfer', 'text-style-transfoer'] | ['computer-vision', 'natural-language-processing'] | [ 4.76646185e-01 -2.12269530e-01 -1.03001647e-01 -3.48026752e-01
-5.55647075e-01 -9.34598148e-01 5.23448706e-01 -3.38671193e-03
-3.85646313e-01 7.73665190e-01 2.27606729e-01 -5.73862232e-02
2.01145604e-01 -7.97252238e-01 -5.92041075e-01 -7.18684733e-01
6.73469007e-01 1.73757836e-01 3.76492068e-02 -4.67414409e-01
2.76815295e-01 3.32530528e-01 -9.01189864e-01 -6.32183924e-02
1.19707477e+00 8.31266403e-01 3.07623178e-01 6.74475491e-01
-4.18568909e-01 4.01168108e-01 -9.98584926e-01 -4.83350813e-01
2.88634121e-01 -6.42511070e-01 -5.37447810e-01 4.09791013e-03
6.18335009e-01 -3.00847083e-01 -4.67511684e-01 1.25470138e+00
4.84630436e-01 2.94096738e-01 8.69442344e-01 -9.44761217e-01
-7.70010710e-01 3.52078974e-01 -5.99147499e-01 -4.16243449e-02
2.32765973e-01 1.03242435e-01 9.29849505e-01 -5.34363508e-01
4.16840017e-01 1.19522011e+00 4.26800609e-01 6.83687747e-01
-1.56710255e+00 -8.94596338e-01 2.79348016e-01 -2.58692235e-01
-1.07956183e+00 -2.89076030e-01 1.21193659e+00 -2.94153512e-01
2.32440159e-01 1.95884570e-01 6.75174654e-01 1.16584063e+00
4.22194719e-01 7.62711585e-01 1.13871014e+00 -4.44523841e-01
2.42206872e-01 2.43089572e-01 -1.41533166e-01 5.55037677e-01
-5.77191301e-02 -1.26960441e-01 -5.42351425e-01 1.57647640e-01
1.06886017e+00 -4.04392965e-02 -2.27549687e-01 -3.07686716e-01
-1.13832843e+00 7.39877582e-01 4.13484842e-01 1.67714685e-01
9.14825350e-02 5.79452738e-02 6.42982423e-01 5.21501422e-01
4.75030422e-01 5.66853344e-01 -3.75627160e-01 -2.87807435e-01
-8.04152250e-01 3.79709005e-01 6.77723885e-01 1.19620216e+00
8.91171813e-01 2.81326443e-01 -6.39755726e-01 9.81302202e-01
-2.39415485e-02 6.04821205e-01 6.16610289e-01 -8.00775290e-01
6.87616467e-01 6.08376741e-01 1.24858260e-01 -1.07964849e+00
5.46312965e-02 -3.34194273e-01 -9.70035613e-01 4.70233321e-01
5.45511365e-01 -3.61391276e-01 -8.65269601e-01 2.06174183e+00
2.76624948e-01 -4.14798371e-02 -1.97411552e-01 7.01980591e-01
3.94029528e-01 6.23831689e-01 1.60073265e-01 7.22349063e-02
1.26153004e+00 -8.85972798e-01 -8.76830041e-01 -2.03281656e-01
3.67460310e-01 -9.16161776e-01 1.69500303e+00 1.09786302e-01
-1.02024758e+00 -5.70618927e-01 -1.12127650e+00 -2.19671845e-01
-2.69998372e-01 2.69812290e-02 3.38764876e-01 6.64553940e-01
-7.06966341e-01 5.68196774e-01 -5.44755399e-01 -2.92019881e-02
6.08043790e-01 4.41987902e-01 -2.62617230e-01 1.79691598e-01
-1.17023909e+00 5.67058086e-01 2.38812581e-01 -1.66605368e-01
-4.72965568e-01 -1.06907570e+00 -7.73503244e-01 2.79058695e-01
8.97360519e-02 -6.63952649e-01 1.15084147e+00 -1.29407358e+00
-2.08681035e+00 6.98654711e-01 -8.80952552e-02 2.50298958e-02
7.49480188e-01 -3.48391324e-01 -1.45626187e-01 1.87743884e-02
-5.30424975e-02 7.06581414e-01 1.33618200e+00 -1.12852859e+00
-4.45023179e-01 -2.65888810e-01 1.41601428e-01 4.84792233e-01
-9.45285320e-01 -2.10438430e-01 -5.33901393e-01 -1.40797365e+00
-2.85236210e-01 -9.16123807e-01 -5.39841130e-02 1.52685001e-01
-3.50018740e-01 8.88920203e-02 1.00802016e+00 -4.43109721e-01
1.24176288e+00 -1.99658561e+00 5.38116932e-01 -5.47039416e-03
3.33382517e-01 3.89724195e-01 -2.25678489e-01 1.43057451e-01
1.86767787e-01 -1.03108529e-02 -2.15245515e-01 -4.69730407e-01
8.66029598e-03 1.98542982e-01 -6.19294643e-01 1.46360889e-01
3.45653832e-01 8.73437464e-01 -9.77643251e-01 -4.97278750e-01
6.81826919e-02 4.53372657e-01 -8.75246465e-01 5.04841208e-01
-2.76926279e-01 6.74452484e-01 -8.19539070e-01 2.19289288e-01
7.05786049e-01 3.35374735e-02 1.29855499e-01 -2.59365767e-01
1.20366633e-01 9.56080854e-02 -1.02672184e+00 1.86352682e+00
-6.98191285e-01 5.51814437e-01 7.49396682e-02 -6.58595800e-01
9.31943476e-01 6.39047474e-02 2.30644137e-01 -6.44899368e-01
2.40278319e-01 -2.22191766e-01 -3.02049279e-01 -6.10156804e-02
7.15036392e-01 -4.30063188e-01 -2.96851784e-01 6.92415833e-01
-3.03964503e-02 -5.04741132e-01 -1.74050294e-02 1.44158363e-01
7.09105074e-01 1.39364570e-01 -2.24110391e-02 -4.75717038e-01
5.49373031e-01 -5.22552013e-01 6.73833787e-01 6.48775637e-01
-1.44234106e-01 6.29594803e-01 5.37876487e-01 -1.15369856e-01
-1.15559208e+00 -9.73789752e-01 3.51915769e-02 1.40773666e+00
2.46576756e-01 -2.45889604e-01 -9.57323849e-01 -8.86413217e-01
3.45477425e-02 5.61983109e-01 -7.33440042e-01 -3.01940620e-01
-7.49591649e-01 -3.97048503e-01 6.67863131e-01 5.19685447e-01
5.19053698e-01 -1.05250573e+00 -2.24696860e-01 1.32174492e-01
1.57901943e-02 -8.55181515e-01 -1.36868000e+00 -3.98284644e-02
-8.88956845e-01 -7.29587793e-01 -8.46350849e-01 -8.94946933e-01
8.81311715e-01 9.42740664e-02 8.76660407e-01 -8.59186798e-02
6.50061220e-02 1.44532129e-01 -1.78507850e-01 -2.93677419e-01
-5.58911026e-01 4.59635705e-01 2.54492965e-02 1.29494086e-01
-4.37959358e-02 -7.51796126e-01 -6.54889524e-01 1.97710723e-01
-1.24340558e+00 1.52855888e-01 3.66795868e-01 1.17118919e+00
3.57628554e-01 -1.64780229e-01 6.88037455e-01 -1.03356242e+00
1.02941358e+00 -4.22905721e-02 -6.78498387e-01 2.01994315e-01
-6.90210223e-01 3.40762198e-01 1.14604068e+00 -7.96590686e-01
-1.28191888e+00 -6.20275885e-02 5.44576459e-02 -4.78813261e-01
-6.51237071e-02 2.25623604e-02 -3.73887777e-01 -1.61775067e-01
5.41977406e-01 3.86250377e-01 1.75081596e-01 -4.26729858e-01
6.28684521e-01 4.38803762e-01 3.89787108e-01 -9.75759268e-01
1.11926591e+00 2.14539766e-01 -7.80585688e-03 -6.19912505e-01
-9.15335298e-01 7.32048079e-02 -7.37599909e-01 8.86201337e-02
6.64132774e-01 -7.54581630e-01 -6.27753258e-01 7.57366359e-01
-8.57841253e-01 -6.94246352e-01 -3.91618073e-01 6.82553053e-02
-4.90570426e-01 4.15298730e-01 -7.22424388e-01 -3.11292738e-01
-6.21355116e-01 -1.22141898e+00 9.48943555e-01 3.14068735e-01
-2.42049739e-01 -1.35634780e+00 9.98132601e-02 1.25059828e-01
5.80654025e-01 1.18704692e-01 1.01882303e+00 -1.56623840e-01
-1.26219928e-01 -1.58658072e-01 -1.76337853e-01 4.84140068e-01
6.30701125e-01 -1.46801621e-01 -7.52421677e-01 -6.51878417e-01
3.70047130e-02 -4.74420846e-01 5.70244312e-01 -4.59655859e-02
1.19990075e+00 -4.26435560e-01 -1.03428690e-02 7.84072399e-01
1.21702230e+00 6.31097937e-03 5.14134169e-01 2.24260211e-01
1.13136232e+00 3.62513304e-01 4.14395183e-01 4.16100681e-01
6.85843453e-02 8.48099768e-01 -1.76920637e-01 -1.48060545e-01
-2.25392431e-01 -3.83287251e-01 3.20664674e-01 6.73993349e-01
3.17087620e-01 -2.36605152e-01 -3.81526262e-01 2.72061467e-01
-1.42064691e+00 -7.43792653e-01 4.63018835e-01 2.16563344e+00
1.36927223e+00 3.30125898e-01 9.63500440e-02 1.01967514e-01
6.34495199e-01 1.86589003e-01 -7.84514010e-01 -3.51385862e-01
3.74951325e-02 2.74534911e-01 3.41587692e-01 5.70696533e-01
-9.41200197e-01 1.14626753e+00 6.79041910e+00 1.06313705e+00
-1.50722539e+00 -2.53838539e-01 6.14180267e-01 -2.52867967e-01
-5.80011666e-01 -3.69571894e-01 -7.72709489e-01 7.10088432e-01
4.59361881e-01 -3.97898018e-01 6.60054803e-01 6.16190612e-01
2.51491189e-01 2.29440987e-01 -1.00553703e+00 7.87499070e-01
-3.44302207e-02 -1.20529044e+00 3.90870571e-01 -1.54405281e-01
8.60313654e-01 -4.93994713e-01 4.01602715e-01 3.25881809e-01
2.09410757e-01 -8.90526474e-01 7.43888676e-01 2.56024510e-01
1.28281939e+00 -9.46249306e-01 1.24349184e-01 9.96932015e-03
-9.98440921e-01 1.47200406e-01 -3.38158637e-01 -3.37494053e-02
-1.66408122e-01 3.82244498e-01 -4.91874933e-01 2.14318037e-01
3.21465880e-01 6.20584965e-01 -5.52942038e-01 3.47282827e-01
-2.95512319e-01 6.73275530e-01 -8.85624215e-02 -3.32636088e-02
1.25728160e-01 -6.01111114e-01 6.90402031e-01 1.24533117e+00
7.03257136e-03 3.58603150e-02 1.75559849e-01 9.60625648e-01
-4.32789594e-01 1.24199770e-01 -5.20315886e-01 -4.99254419e-03
5.35399556e-01 1.29676890e+00 -3.92300129e-01 -2.80571282e-01
-2.62578875e-01 1.23684549e+00 5.97847283e-01 4.98482704e-01
-6.14458621e-01 -7.81057596e-01 1.02053058e+00 -9.06395912e-02
4.28707451e-01 -1.96447074e-01 -8.30474317e-01 -1.34548450e+00
-8.22496880e-03 -9.93612051e-01 1.50357798e-01 -4.49878365e-01
-1.29370654e+00 4.76204336e-01 -2.64745474e-01 -1.31471384e+00
-2.75191129e-03 -4.78980958e-01 -8.85784805e-01 1.13302588e+00
-1.29447246e+00 -1.04124880e+00 -3.23233008e-01 6.28650844e-01
5.97175300e-01 -3.55854094e-01 7.78722644e-01 1.65052071e-01
-6.74715757e-01 1.36055839e+00 3.93799990e-01 2.38849714e-01
1.13966525e+00 -1.48927748e+00 3.41797501e-01 6.81396782e-01
-2.90100008e-01 7.82225907e-01 8.68127286e-01 -5.29151320e-01
-1.52559960e+00 -1.30176377e+00 4.41499293e-01 -4.16770875e-01
6.12444520e-01 -6.31267488e-01 -1.06528282e+00 5.97400844e-01
2.23345742e-01 -2.48236030e-01 6.83916330e-01 4.01153676e-02
-4.03466940e-01 -3.46576869e-01 -1.25909853e+00 9.84113514e-01
8.06738675e-01 -5.93281567e-01 -5.80280900e-01 7.15026259e-02
7.39047527e-01 -5.87978244e-01 -9.82336462e-01 -3.21331881e-02
6.32873237e-01 -5.98890007e-01 7.80834854e-01 -7.29251862e-01
7.54666805e-01 -2.32499748e-01 7.39537328e-02 -1.68523967e+00
-4.18627053e-01 -7.68153071e-01 3.95055972e-02 1.46806538e+00
9.60417315e-02 -5.27126670e-01 9.84784186e-01 6.35512471e-01
-3.84199694e-02 -6.75358355e-01 -5.69165826e-01 -5.81633270e-01
5.93679130e-01 5.84494807e-02 7.21808255e-01 8.98059309e-01
-1.01187147e-01 5.42660952e-01 -4.34121132e-01 -1.35325521e-01
6.10948026e-01 1.16199628e-01 9.36805487e-01 -8.29776287e-01
-3.63330185e-01 -7.26291001e-01 -1.96052387e-01 -1.23743749e+00
5.14343262e-01 -7.89259672e-01 2.11217254e-02 -9.54503059e-01
2.92416774e-02 -7.24448144e-01 -3.26725066e-01 3.41475874e-01
-6.97378695e-01 3.81085008e-01 2.94219464e-01 5.64773269e-02
-2.83972621e-01 9.47364390e-01 1.86328959e+00 -2.92272747e-01
-3.31563950e-01 -2.92424206e-03 -8.88002634e-01 3.86495948e-01
7.67952740e-01 -3.62287998e-01 -6.53368831e-01 -5.97722232e-01
-1.26402810e-01 -1.56848654e-01 -1.32718667e-01 -6.87679648e-01
-1.17860623e-01 -3.29856187e-01 5.70663929e-01 -1.25245377e-01
1.53257310e-01 -6.49501264e-01 -3.28223228e-01 2.78529227e-01
-5.72632909e-01 -3.21765617e-02 3.63503665e-01 6.17531598e-01
-8.44411477e-02 -5.60438298e-02 1.07979548e+00 1.11227766e-01
-2.18000963e-01 3.29472154e-01 -1.27120391e-01 2.90228724e-01
8.30763936e-01 -1.37810156e-01 2.75140461e-02 -2.93505967e-01
-2.39371955e-01 1.78100839e-01 7.56860375e-01 5.20200372e-01
3.69259745e-01 -1.66407025e+00 -5.71797550e-01 4.83684927e-01
5.84160723e-02 7.49709904e-02 1.94763675e-01 2.16244668e-01
-3.43055367e-01 6.63753524e-02 -4.98701751e-01 -3.55252177e-01
-1.14119315e+00 4.93639797e-01 2.24698722e-01 -3.47857863e-01
-6.98969662e-01 7.20420480e-01 6.52642667e-01 -3.91068280e-01
2.10117504e-01 -1.34169027e-01 -8.65827203e-02 -1.01289041e-02
6.78528786e-01 3.34143788e-01 -1.20348960e-01 -3.66365761e-01
1.59675181e-01 7.72066712e-01 -4.89872575e-01 -1.33958355e-01
1.04240429e+00 -2.08211094e-01 4.21741791e-02 3.85269493e-01
1.25488913e+00 2.13920519e-01 -1.81534028e+00 -3.35889012e-01
-5.15449584e-01 -5.38160563e-01 1.45138964e-01 -6.62098885e-01
-1.09211540e+00 7.19932616e-01 4.96995062e-01 -1.52277678e-01
1.14123666e+00 -6.13029063e-01 1.02086878e+00 3.06192189e-01
8.87116194e-02 -1.08821428e+00 4.30660546e-01 6.28130496e-01
9.41457331e-01 -1.10416603e+00 -2.82133698e-01 -1.72580883e-01
-7.28560209e-01 1.11233449e+00 7.30450034e-01 -1.77836820e-01
3.52974921e-01 2.51227558e-01 3.50436360e-01 3.49114537e-01
-3.22126389e-01 4.64991331e-01 4.86371726e-01 3.92163962e-01
5.20490050e-01 -2.33559916e-03 -1.18582450e-01 3.42725456e-01
-2.60128796e-01 -2.18222439e-01 3.13526392e-01 9.47933197e-01
-2.73358464e-01 -1.46985483e+00 -3.64983588e-01 4.26797450e-01
-4.79706883e-01 -5.50619997e-02 -2.84431905e-01 3.55600804e-01
-2.45625421e-01 6.59460664e-01 4.62352410e-02 -3.39493155e-01
4.36400235e-01 1.03150621e-01 6.69593215e-01 -5.64580381e-01
-7.68619180e-01 4.77811992e-02 -5.30686080e-01 -2.02989206e-01
2.18640938e-01 -4.30681348e-01 -1.06973994e+00 -5.44964492e-01
-8.77339765e-02 5.36227822e-02 2.92868882e-01 8.81359756e-01
3.44870895e-01 6.55385554e-01 9.90223467e-01 -9.02833819e-01
-7.73801208e-01 -9.71484840e-01 -7.30139494e-01 8.36156607e-01
3.72053981e-01 -5.68615913e-01 -1.82554185e-01 2.93420792e-01] | [11.498167991638184, -0.5032269954681396] |
25e7c302-2b39-47a8-8035-0e3981d87ac6 | frozen-pretrained-transformers-for-neural | null | null | https://aclanthology.org/2021.mtsummit-at4ssl.10/ | https://aclanthology.org/2021.mtsummit-at4ssl.10.pdf | Frozen Pretrained Transformers for Neural Sign Language Translation | One of the major challenges in sign language translation from a sign language to a spoken language is the lack of parallel corpora. Recent works have achieved promising results on the RWTH-PHOENIX-Weather 2014T dataset, which consists of over eight thousand parallel sentences between German sign language and German. However, from the perspective of neural machine translation, this is still a tiny dataset. To improve the performance of models trained on small datasets, transfer learning can be used. While this has been previously applied in sign language translation for feature extraction, to the best of our knowledge, pretrained language models have not yet been investigated. We use pretrained BERT-base and mBART-50 models to initialize our sign language video to spoken language text translation model. To mitigate overfitting, we apply the frozen pretrained transformer technique: we freeze the majority of parameters during training. Using a pretrained BERT model, we outperform a baseline trained from scratch by 1 to 2 BLEU-4. Our results show that pretrained language models can be used to improve sign language translation performance and that the self-attention patterns in BERT transfer in zero-shot to the encoder and decoder of sign language translation models. | ['Joni Dambre', 'Mieke Van Herreweghe', 'Severine Verlinden', 'Paloma Rabaey', 'Marija Pizurica', "Karel D'Oosterlinck", 'Mathieu De Coster'] | 2021-08-20 | null | null | null | international-workshop-on-automatic | ['sign-language-translation'] | ['computer-vision'] | [ 2.50780255e-01 -4.25773859e-02 -2.00194433e-01 -4.11669880e-01
-1.17499292e+00 -4.06479090e-01 7.77487576e-01 -9.70632792e-01
-6.94231927e-01 6.22244418e-01 7.24330366e-01 -3.70810777e-01
4.72735196e-01 -2.93596953e-01 -9.48103070e-01 -3.90121400e-01
3.18869770e-01 8.09462667e-01 3.73162895e-01 -5.06066263e-01
-2.29677901e-01 1.23879910e-01 -1.32586622e+00 7.53630877e-01
6.72340512e-01 4.91971880e-01 -5.34592494e-02 7.73286641e-01
-2.01751992e-01 9.49502587e-01 -5.26019335e-01 -4.35911238e-01
5.32657802e-01 -1.05258095e+00 -8.92202616e-01 -2.50802428e-01
5.29193223e-01 -7.44860947e-01 -5.15893221e-01 5.47080219e-01
9.61264849e-01 -2.06866667e-01 5.50832510e-01 -1.05466354e+00
-5.19345403e-01 8.41739714e-01 -7.48500675e-02 -1.72485724e-01
2.19923601e-01 5.39763987e-01 8.58484507e-01 -1.04680574e+00
1.03055632e+00 1.21184003e+00 5.15137017e-01 9.44768846e-01
-7.96051025e-01 -7.49939680e-01 -1.23251788e-01 2.69460261e-01
-1.00187612e+00 -6.28686607e-01 3.51494610e-01 -2.24038020e-01
1.42207599e+00 -6.80185333e-02 8.47543538e-01 1.35758924e+00
4.38398607e-02 1.03423464e+00 1.26728010e+00 -6.43732786e-01
-1.17600247e-01 -8.38065296e-02 -2.65905112e-01 7.75360465e-01
-3.73519421e-01 2.54048616e-01 -9.15793240e-01 1.32043436e-01
5.84275782e-01 -4.52263266e-01 -3.10447067e-01 -1.69200182e-01
-1.37209094e+00 6.24966085e-01 4.01686668e-01 2.43632749e-01
-2.13614672e-01 3.97241056e-01 5.60752332e-01 8.27776790e-01
2.94791222e-01 2.55952924e-01 -4.65814650e-01 -5.34074664e-01
-9.79659379e-01 9.92571861e-02 8.32212448e-01 1.04760087e+00
3.10875952e-01 -1.42482109e-02 -3.01457375e-01 7.22613633e-01
2.41855487e-01 9.43828940e-01 7.08726704e-01 -6.81920409e-01
8.61715317e-01 3.14306587e-01 -2.07283780e-01 -3.45088616e-02
-8.12246501e-02 -7.19044507e-02 -3.80098879e-01 3.89409930e-01
6.05925858e-01 -3.86451840e-01 -1.46467185e+00 1.43020666e+00
-1.66100636e-01 1.88026011e-01 2.60978311e-01 1.25152767e+00
8.00992012e-01 5.45221567e-01 -1.14943899e-01 2.30447039e-01
1.02419925e+00 -1.21043527e+00 -4.90084201e-01 -3.43950540e-01
7.32758641e-01 -1.02111125e+00 1.32107949e+00 1.23971783e-01
-1.11269557e+00 -1.58329770e-01 -7.83788562e-01 -1.65453270e-01
-1.30261764e-01 3.15052480e-01 2.54449666e-01 3.64065707e-01
-1.17350793e+00 4.40716237e-01 -1.18442547e+00 -9.30127323e-01
3.14622521e-01 3.76753837e-01 -2.77278900e-01 -3.15417618e-01
-1.17538309e+00 1.40608609e+00 -9.99853387e-03 2.03147784e-01
-7.05766976e-01 -4.36878771e-01 -4.66577679e-01 -2.92362720e-01
6.09897040e-02 -9.52654243e-01 1.48112631e+00 -1.15498328e+00
-2.25325274e+00 8.72880399e-01 -3.21314961e-01 -5.18585980e-01
9.30053413e-01 -2.53236413e-01 -9.41507965e-02 -4.89439815e-02
-3.02935272e-01 8.37009072e-01 7.20088065e-01 -7.60255277e-01
-5.08370638e-01 3.47064026e-02 -1.42426074e-01 2.37960055e-01
-7.89841488e-02 6.04207456e-01 -6.26124442e-01 -4.91780490e-01
-9.13038254e-02 -1.21074224e+00 -3.34802680e-02 2.46743839e-02
1.65031254e-01 1.63528979e-01 5.63968122e-01 -1.00027716e+00
8.66150916e-01 -1.72315013e+00 4.20477837e-01 1.19934127e-01
-3.13219905e-01 5.61164081e-01 -6.27133965e-01 6.54313922e-01
2.46632606e-01 -1.29747748e-01 -3.99052292e-01 -4.67475027e-01
7.41010439e-03 4.76512343e-01 -4.31654304e-01 3.26979131e-01
3.08044374e-01 1.45481467e+00 -7.48426378e-01 -3.27395797e-01
7.34685436e-02 6.05951846e-01 -7.23627090e-01 2.67775208e-01
-2.57860363e-01 6.89248502e-01 -1.17745057e-01 5.78415692e-01
1.93424702e-01 1.81556977e-02 3.87420431e-02 1.18061751e-01
1.97233297e-02 7.78771102e-01 -4.83634323e-01 1.88654327e+00
-7.36324549e-01 9.26716208e-01 -1.88035652e-01 -7.37117052e-01
7.19434321e-01 5.00940561e-01 3.87290150e-01 -8.12122226e-01
3.06976646e-01 6.98175967e-01 3.19113791e-01 -7.35449731e-01
1.24590077e-01 -4.39947903e-01 5.16645275e-02 7.81647742e-01
2.60967612e-01 -3.70715469e-01 1.65057316e-01 -2.27967173e-01
1.27301657e+00 5.00369012e-01 -1.25033572e-01 2.72131145e-01
1.98051766e-01 2.58269012e-01 2.52358258e-01 4.37007278e-01
-8.11358839e-02 8.82311761e-01 1.74291998e-01 -2.16044903e-01
-1.16203284e+00 -8.81903470e-01 2.91449487e-01 1.21847510e+00
-5.24199784e-01 -2.46339902e-01 -7.69581914e-01 -6.60504758e-01
-2.85245448e-01 6.00438297e-01 -2.04965800e-01 -1.35628223e-01
-1.18389177e+00 -4.98069704e-01 1.07131350e+00 6.41253173e-01
4.86942172e-01 -1.29824102e+00 -6.33927524e-01 2.11179867e-01
-4.11849350e-01 -1.28809953e+00 -6.85953677e-01 1.70636307e-02
-8.60556543e-01 -7.12700903e-01 -1.23620081e+00 -9.71081674e-01
5.84715962e-01 -2.35007539e-01 9.92506087e-01 -7.40278438e-02
6.70395717e-02 2.20430464e-01 -6.17665172e-01 -4.77616131e-01
-9.03362393e-01 2.49984443e-01 -1.22521065e-01 -1.82688057e-01
5.98757148e-01 -3.16884011e-01 -2.46482000e-01 4.02297854e-01
-6.33721292e-01 2.64063865e-01 9.03131843e-01 1.13671899e+00
1.74712583e-01 -1.02553380e+00 1.45566717e-01 -3.59621525e-01
5.20405173e-01 1.63897470e-01 -5.26142120e-01 2.93627083e-01
-4.40755218e-01 4.92281318e-01 5.55000484e-01 -6.06784701e-01
-7.23005056e-01 2.17628673e-01 -2.76353985e-01 -5.09803712e-01
1.37273893e-01 6.11075699e-01 8.21439400e-02 -1.61294952e-01
7.08087564e-01 2.82129556e-01 2.82817900e-01 -3.67347687e-01
4.98377532e-01 9.84150171e-01 4.23523515e-01 -3.53305548e-01
6.74671948e-01 2.66758770e-01 -2.19622836e-01 -5.21815181e-01
-4.50927049e-01 -3.25740874e-01 -7.43234932e-01 -2.39715710e-01
7.20147550e-01 -8.96174550e-01 -1.54682383e-01 6.28366113e-01
-1.22547364e+00 -9.15650785e-01 -4.10232484e-01 7.27902591e-01
-8.84948373e-01 1.98403418e-01 -7.30599582e-01 -3.42397630e-01
-4.40042496e-01 -1.22804892e+00 1.07877123e+00 -3.58814210e-01
-3.89403015e-01 -6.23311341e-01 2.99556166e-01 3.83425862e-01
7.88368464e-01 -7.66214356e-02 5.70847154e-01 -3.07795346e-01
-6.85667157e-01 -2.80601978e-01 3.89184020e-02 5.33352554e-01
4.29921858e-02 -2.50123709e-01 -8.10540915e-01 -3.47911298e-01
-3.53130519e-01 -5.99807501e-01 1.13182914e+00 1.34011209e-01
2.76502997e-01 -2.77303100e-01 -6.40060008e-02 6.50376439e-01
9.16192174e-01 -1.25959724e-01 7.64071405e-01 3.19913685e-01
5.35714746e-01 4.01871771e-01 3.69860291e-01 4.37291060e-03
4.98398453e-01 8.50911379e-01 -7.44913965e-02 -1.88971549e-01
-6.98294222e-01 -5.60503840e-01 9.83314693e-01 1.14831209e+00
-4.94656086e-01 -1.74552053e-01 -1.13762856e+00 6.07661128e-01
-1.84285295e+00 -9.57652688e-01 1.50367454e-01 2.12994599e+00
8.73674273e-01 -9.07431319e-02 2.36543134e-01 -1.00142390e-01
1.86686754e-01 -2.28020661e-02 -4.06326294e-01 -3.63869667e-01
-1.87857717e-01 5.20692766e-01 7.58806109e-01 6.68657839e-01
-7.00179100e-01 1.51439536e+00 6.08062315e+00 2.81432211e-01
-1.73760879e+00 4.09602791e-01 -5.96327074e-02 -5.13607025e-01
-1.82627931e-01 2.38442913e-01 -6.76950157e-01 2.27200255e-01
1.26907599e+00 -1.59390140e-02 7.32103109e-01 4.05933946e-01
3.46813947e-01 4.26436603e-01 -1.19013989e+00 9.38020647e-01
3.71263683e-01 -1.02395439e+00 1.36818141e-01 8.72509927e-02
7.79683828e-01 8.61416519e-01 -8.97795483e-02 5.44840395e-01
4.86741930e-01 -9.99112010e-01 9.25202191e-01 4.70797390e-01
1.09104681e+00 -2.83685654e-01 8.14781070e-01 3.38907242e-01
-8.71311426e-01 1.30153254e-01 -2.11479634e-01 -2.43368149e-02
7.18363643e-01 1.73167661e-02 -1.12438655e+00 2.32736051e-01
4.09298748e-01 6.20696783e-01 -1.97582796e-01 9.55306590e-01
-5.54748654e-01 1.03770220e+00 -6.09761238e-01 -1.73619017e-01
3.93693626e-01 2.62595806e-03 4.39842492e-01 1.21615267e+00
6.22444153e-01 -1.03384994e-01 -2.57974993e-02 6.35601699e-01
-3.82996723e-02 2.56210774e-01 -6.63462400e-01 -1.67671815e-01
-2.28378996e-01 4.64762807e-01 -1.60635725e-01 -4.02047575e-01
-5.98734915e-01 1.42873383e+00 1.00898646e-01 5.84116876e-01
-8.36632490e-01 -1.55744907e-02 5.69646478e-01 3.25402260e-01
4.35142785e-01 -3.58145654e-01 -1.56322390e-01 -1.48351312e+00
1.60215795e-01 -1.17669904e+00 6.32888004e-02 -6.48467839e-01
-1.03216946e+00 7.35837400e-01 -1.26805916e-01 -1.61301947e+00
-8.10879111e-01 -7.58372247e-01 -4.29020047e-01 9.64976013e-01
-1.47903645e+00 -1.62007308e+00 -8.57234299e-02 5.30960262e-01
4.98166710e-01 -3.84515107e-01 8.56056511e-01 3.79932225e-01
-3.55078802e-02 8.26498747e-01 2.83068884e-02 3.94600332e-01
1.12045443e+00 -7.63552964e-01 8.89604449e-01 9.21115875e-01
3.35689127e-01 3.77463132e-01 5.62666178e-01 -5.51003098e-01
-1.33610702e+00 -9.58438218e-01 1.43652701e+00 -6.07365847e-01
6.05351210e-01 -3.83030355e-01 -6.01074219e-01 7.92703748e-01
3.79656404e-01 -1.19064348e-02 3.46499056e-01 -2.69165844e-01
-3.45295131e-01 1.21250741e-01 -6.76474392e-01 8.07036221e-01
1.33328140e+00 -7.74897397e-01 -5.75451970e-01 3.03967178e-01
3.58789831e-01 -7.67800868e-01 -6.52706265e-01 2.97620088e-01
9.98601258e-01 -4.70904142e-01 6.10904932e-01 -7.95953751e-01
6.72556698e-01 -2.96385288e-01 -1.16160601e-01 -1.45143735e+00
2.48778984e-01 -7.48706818e-01 1.31576106e-01 8.60785067e-01
7.84504712e-01 -6.22791529e-01 7.50151634e-01 5.23913324e-01
-1.95524722e-01 -5.63458562e-01 -1.19267714e+00 -9.32177663e-01
4.94972616e-01 -4.41489697e-01 5.06829798e-01 5.24728894e-01
1.65937960e-01 3.71948928e-01 -7.38219321e-01 -3.67872745e-01
1.95604295e-01 3.66009325e-02 1.24081695e+00 -6.05243802e-01
-5.67374051e-01 -4.97321099e-01 -2.91304410e-01 -1.39468050e+00
2.21841276e-01 -1.35185349e+00 3.95117164e-01 -1.77363312e+00
-1.44520961e-02 2.69103721e-02 -4.62801270e-02 8.87230158e-01
7.60158598e-02 2.41516829e-01 6.12451673e-01 3.10407668e-01
-1.39426023e-01 6.92618191e-01 1.32960176e+00 -2.59570271e-01
-2.15658814e-01 -9.83651504e-02 -5.62270507e-02 3.81766707e-01
7.03273594e-01 -4.75220144e-01 -2.96628568e-02 -9.12158906e-01
-1.09016895e-01 -1.07550286e-01 3.12846482e-01 -9.30961847e-01
3.08388382e-01 6.38579503e-02 -2.73545593e-01 -2.70882100e-01
2.78900802e-01 -6.91394925e-01 -8.18074197e-02 6.49840951e-01
-4.22357947e-01 1.91734713e-02 2.23034903e-01 2.18634773e-02
-4.10603851e-01 1.61180809e-01 7.37984598e-01 -1.63553938e-01
-5.77795208e-01 6.04707636e-02 -5.57972372e-01 6.09443411e-02
5.06211698e-01 -2.52391100e-02 -2.12365717e-01 -6.53783798e-01
-3.45138848e-01 1.43745229e-01 4.52282369e-01 6.27516925e-01
6.15533590e-01 -1.32401419e+00 -1.10546434e+00 3.34507644e-01
1.86492115e-01 -2.27737993e-01 -3.27283800e-01 1.06720352e+00
-7.03686953e-01 5.57055593e-01 -4.20509458e-01 -7.00267017e-01
-1.29038739e+00 -6.78715631e-02 5.04509747e-01 -2.56425351e-01
-8.35785806e-01 7.13020384e-01 -4.35757697e-01 -7.43825734e-01
1.53605804e-01 -7.27155030e-01 3.01694244e-01 -3.89152884e-01
3.41537356e-01 -2.45801713e-02 9.27615687e-02 -8.68114531e-01
-2.60976225e-01 9.28701043e-01 6.70104427e-03 -7.49507070e-01
1.33244491e+00 1.95896566e-01 1.17324702e-01 3.33871067e-01
1.13196719e+00 -2.07042009e-01 -1.13191843e+00 -3.19646090e-01
-2.61046253e-02 -2.37337455e-01 -1.22834615e-01 -1.02910376e+00
-9.54343259e-01 1.01682901e+00 6.62022531e-01 -1.00385916e+00
1.10620904e+00 1.15860822e-02 1.16942489e+00 6.80850327e-01
5.94952404e-01 -1.04706311e+00 -1.39789999e-01 1.04449224e+00
1.20147574e+00 -1.26729918e+00 -5.29092968e-01 3.41228805e-02
-8.07118714e-01 1.02738070e+00 2.56438315e-01 -1.50288194e-01
3.84722143e-01 5.29445589e-01 7.46702909e-01 1.90737516e-01
-7.59088397e-01 -5.04113257e-01 3.52162033e-01 3.39072615e-01
5.91152966e-01 -1.47523224e-01 -5.30884147e-01 1.55916631e-01
-5.74979007e-01 7.80903637e-01 3.41678590e-01 9.89518523e-01
-1.68249644e-02 -1.54749548e+00 -1.79620564e-01 2.69337475e-01
-2.44906262e-01 -2.16448143e-01 -6.85388327e-01 6.59342766e-01
-1.56751573e-01 6.73871994e-01 -2.74928987e-01 -5.08970439e-01
6.87964678e-01 5.20098865e-01 7.82915413e-01 -4.44289327e-01
-7.14312851e-01 -8.87412671e-03 3.37820441e-01 -5.65617144e-01
-5.49186945e-01 -7.68757761e-01 -1.35917640e+00 -2.14043900e-01
-1.87063426e-01 -3.02012265e-01 6.78596377e-01 1.21967924e+00
2.13090152e-01 2.96247602e-01 -1.08348075e-02 -8.37837219e-01
-8.83086264e-01 -1.23558021e+00 1.06955990e-01 3.27828020e-01
2.19403744e-01 -2.46135011e-01 -2.59812206e-01 2.68699199e-01] | [9.217089653015137, -6.543366432189941] |
ce8b2a46-6112-4c50-b0c8-928e8380b978 | inversion-by-direct-iteration-an-alternative | 2303.11435 | null | https://arxiv.org/abs/2303.11435v3 | https://arxiv.org/pdf/2303.11435v3.pdf | Inversion by Direct Iteration: An Alternative to Denoising Diffusion for Image Restoration | Inversion by Direct Iteration (InDI) is a new formulation for supervised image restoration that avoids the so-called ``regression to the mean'' effect and produces more realistic and detailed images than existing regression-based methods. It does this by gradually improving image quality in small steps, similar to generative denoising diffusion models. Image restoration is an ill-posed problem where multiple high-quality images are plausible reconstructions of a given low-quality input. Therefore, the outcome of a single step regression model is typically an aggregate of all possible explanations, therefore lacking details and realism. The main advantage of InDI is that it does not try to predict the clean target image in a single step but instead gradually improves the image in small steps, resulting in better perceptual quality. While generative denoising diffusion models also work in small steps, our formulation is distinct in that it does not require knowledge of any analytic form of the degradation process. Instead, we directly learn an iterative restoration process from low-quality and high-quality paired examples. InDI can be applied to virtually any image degradation, given paired training data. In conditional denoising diffusion image restoration the denoising network generates the restored image by repeatedly denoising an initial image of pure noise, conditioned on the degraded input. Contrary to conditional denoising formulations, InDI directly proceeds by iteratively restoring the input low-quality image, producing high-quality results on a variety of image restoration tasks, including motion and out-of-focus deblurring, super-resolution, compression artifact removal, and denoising. | ['Peyman Milanfar', 'Mauricio Delbracio'] | 2023-03-20 | null | null | null | null | ['deblurring'] | ['computer-vision'] | [ 7.22226441e-01 8.89497921e-02 3.40028971e-01 -1.88072488e-01
-9.98035073e-01 -2.11316451e-01 7.20142126e-01 -4.87319291e-01
-1.94433734e-01 8.66240144e-01 3.93553734e-01 -7.48579651e-02
-3.84771377e-01 -6.60006583e-01 -8.06822300e-01 -1.33643198e+00
2.82468349e-01 2.60658771e-01 -9.89253446e-02 -2.70208776e-01
1.69256702e-01 2.61734307e-01 -1.36984324e+00 2.68408835e-01
9.53414381e-01 6.01638019e-01 5.18905103e-01 8.99479330e-01
2.64890432e-01 9.62541342e-01 -4.83833432e-01 3.80492449e-04
4.72091585e-02 -1.12264919e+00 -6.45430267e-01 6.36325240e-01
2.72534162e-01 -4.95073169e-01 -3.60805571e-01 1.34474969e+00
3.41667265e-01 1.13539591e-01 6.92168951e-01 -7.39033699e-01
-1.07998431e+00 2.14276612e-01 -6.48740590e-01 1.36855826e-01
4.70653057e-01 2.86004424e-01 5.23568153e-01 -9.34014142e-01
5.91227293e-01 1.33498967e+00 7.42981434e-01 4.97202694e-01
-1.84289312e+00 -2.20177099e-01 -1.83573991e-01 5.31981746e-03
-9.99418437e-01 -6.29322588e-01 7.51484513e-01 -2.28238001e-01
6.01345062e-01 2.98402131e-01 4.92084175e-01 9.73304689e-01
2.76683331e-01 4.16189551e-01 1.49206460e+00 -5.52333891e-01
1.55169144e-01 -2.00048462e-01 -6.63971007e-02 3.14445645e-01
4.88562584e-02 4.15151477e-01 -3.62264305e-01 2.86802221e-02
1.00103772e+00 2.41292734e-02 -8.47732902e-01 -1.92747653e-01
-1.38519835e+00 5.59356689e-01 4.25915301e-01 4.88702863e-01
-7.27948904e-01 3.22029501e-01 -6.26739934e-02 5.55744231e-01
6.46188438e-01 2.12585852e-01 -4.43696454e-02 1.98176727e-01
-1.26639736e+00 1.27109468e-01 4.37480867e-01 3.48162293e-01
8.70990992e-01 4.44019824e-01 1.93781137e-01 8.22321892e-01
3.56018901e-01 7.61434376e-01 3.72203588e-01 -1.50348568e+00
2.16413885e-02 -1.43404588e-01 2.88974226e-01 -7.07412899e-01
-4.79790382e-03 -5.43650985e-01 -1.37016511e+00 8.26064646e-01
4.59939629e-01 8.90475959e-02 -1.20837438e+00 1.74067235e+00
-7.68483877e-02 2.00675130e-01 1.46039620e-01 1.09132302e+00
5.91996670e-01 9.70601022e-01 -3.09922308e-01 -7.39450932e-01
8.48001122e-01 -7.76466429e-01 -1.07626617e+00 -4.03354913e-01
-7.19305575e-02 -9.30983961e-01 9.16157842e-01 7.26423621e-01
-1.56967664e+00 -7.12332070e-01 -9.81824338e-01 -1.87844381e-01
2.77177274e-01 -2.56725252e-01 1.35048538e-01 4.46954519e-01
-1.45144606e+00 1.01671863e+00 -6.92435205e-01 -1.19742505e-01
3.55158776e-01 2.51040570e-02 -4.58079368e-01 -6.93511248e-01
-8.04529548e-01 1.03888035e+00 -1.71691895e-01 3.96998882e-01
-1.14305854e+00 -7.83890724e-01 -7.42196739e-01 -1.79906234e-01
1.69259220e-01 -8.54627132e-01 9.25073385e-01 -1.38945854e+00
-1.49830687e+00 6.62989378e-01 -5.89804471e-01 -2.03142568e-01
5.74634016e-01 -2.45185494e-01 -3.27821493e-01 2.36912414e-01
4.51387689e-02 4.34541315e-01 1.47325897e+00 -1.90575039e+00
6.30406290e-03 -1.89388037e-01 -3.36150408e-01 2.45070979e-01
4.77236897e-01 -2.58484542e-01 -2.15662286e-01 -6.91294849e-01
6.61945939e-01 -6.23736382e-01 -4.36831951e-01 1.30499378e-01
-3.37199926e-01 6.04507387e-01 7.00584352e-01 -9.94623244e-01
6.65511310e-01 -2.22739315e+00 5.84164262e-01 3.53817940e-02
3.68003100e-01 3.53907645e-02 -4.01418090e-01 3.22702974e-01
-3.97948176e-01 5.93837835e-02 -8.29462826e-01 -4.99253064e-01
-3.91599655e-01 3.87027740e-01 -4.02642280e-01 7.58377075e-01
2.28205666e-01 9.00992334e-01 -1.09633076e+00 -1.64574906e-01
4.95102346e-01 1.01167643e+00 -3.42770487e-01 1.99005365e-01
8.94678757e-02 9.11706805e-01 1.89680979e-01 3.06441218e-01
8.68523598e-01 -1.86527073e-01 1.40042575e-02 -4.71780449e-01
-3.51197273e-02 -8.74813050e-02 -1.11071157e+00 1.64167142e+00
-5.90060592e-01 8.37808132e-01 4.27660584e-01 -1.22357678e+00
7.20150054e-01 4.96114492e-01 3.79417449e-01 -7.88347721e-01
-1.45804107e-01 3.15866649e-01 -9.14445221e-02 -4.87390250e-01
1.89242452e-01 -8.24454963e-01 5.84332466e-01 5.52394807e-01
1.58380382e-02 -5.03868401e-01 -8.41999520e-03 2.15803087e-01
1.02608836e+00 3.96594793e-01 7.12896809e-02 -1.22568235e-01
4.36429054e-01 -1.61092579e-01 3.35768729e-01 8.67781937e-01
8.61777365e-02 1.35735059e+00 1.57288611e-01 -7.24619403e-02
-1.31658232e+00 -1.40438604e+00 -1.05415933e-01 2.58013546e-01
2.37716272e-01 -2.41648275e-02 -7.33082294e-01 -9.15462747e-02
-4.48255092e-01 8.60985577e-01 -5.96221089e-01 -4.30160798e-02
-5.56047916e-01 -1.00667477e+00 8.53131413e-02 -7.68095851e-02
5.82132757e-01 -1.10365260e+00 -2.14367285e-01 3.02980334e-01
-7.44148254e-01 -8.79072428e-01 -3.54083359e-01 2.67038673e-01
-1.16858673e+00 -9.81786013e-01 -1.02474701e+00 -6.72593534e-01
9.59241331e-01 6.06431723e-01 1.22977185e+00 2.64317960e-01
-2.05642745e-01 3.01037639e-01 -9.96019039e-03 3.91402483e-01
-9.27524507e-01 -1.03281474e+00 -8.26762915e-02 1.44998372e-01
-3.91406506e-01 -9.36104238e-01 -6.38240218e-01 2.38729447e-01
-1.19031906e+00 1.13375992e-01 7.11830795e-01 1.19571936e+00
8.63194048e-01 5.90703070e-01 3.82032841e-01 -6.54060900e-01
7.23872423e-01 -3.49974126e-01 -6.11808263e-02 7.77629390e-02
-7.45795548e-01 6.51572496e-02 6.68424428e-01 -6.28988862e-01
-1.44239509e+00 -2.04961419e-01 -1.77510649e-01 -4.83923674e-01
-2.48918936e-01 5.33106387e-01 -4.96069640e-02 6.35943636e-02
8.84654403e-01 7.14099407e-01 4.27958369e-01 -6.65793538e-01
5.09417236e-01 7.91021511e-02 1.01046884e+00 -2.78378516e-01
1.03510976e+00 6.49649143e-01 1.37653708e-01 -1.04150271e+00
-6.44204438e-01 -1.02147751e-01 -5.45056045e-01 -3.61285746e-01
7.66479194e-01 -8.28606606e-01 -3.96321148e-01 6.85205936e-01
-1.07849729e+00 -5.37514508e-01 -7.08436668e-01 4.12846684e-01
-7.92555928e-01 7.18489826e-01 -7.09227502e-01 -7.93475688e-01
3.43015790e-02 -1.22458696e+00 7.62081802e-01 -1.95804611e-02
-2.66176388e-02 -1.09641969e+00 -2.27201760e-01 4.10885781e-01
6.23561859e-01 2.24367544e-01 8.95071924e-01 3.49153996e-01
-6.45516753e-01 -1.33093558e-02 -2.24159896e-01 9.48605776e-01
4.01014358e-01 -3.47204775e-01 -8.52918565e-01 -4.29758519e-01
7.98721671e-01 -2.42148504e-01 9.90389287e-01 9.13600922e-01
7.39285648e-01 -2.41653517e-01 1.28319189e-01 7.02789187e-01
1.64399838e+00 -3.39626856e-02 1.24148512e+00 1.94786429e-01
5.60555458e-01 6.29769325e-01 2.30391607e-01 -1.68142617e-01
-1.65257320e-01 3.25529069e-01 4.90238547e-01 -5.77210188e-01
-8.32454145e-01 1.60077848e-02 5.09636641e-01 5.67486525e-01
-3.65488648e-01 -2.02015787e-01 -4.72393781e-01 5.59094548e-01
-1.62418866e+00 -1.34671378e+00 -5.07519841e-01 2.33745360e+00
7.86592722e-01 -5.39053865e-02 -3.26761633e-01 4.34554815e-01
6.09653413e-01 1.26197308e-01 -5.08107483e-01 1.18553890e-02
-5.09284377e-01 1.62890017e-01 2.36025259e-01 1.09799922e+00
-5.54830432e-01 4.42178041e-01 6.79533768e+00 6.32469416e-01
-8.81749332e-01 3.02321404e-01 8.75832260e-01 1.28444592e-02
-5.52094936e-01 1.28520876e-01 1.87771227e-02 3.46469343e-01
7.50590086e-01 -1.31744882e-02 8.72821093e-01 5.40285781e-02
6.80464029e-01 -5.14316916e-01 -9.55228627e-01 9.46577311e-01
1.23219684e-01 -1.18742537e+00 2.97005083e-02 1.38225526e-01
8.64703715e-01 -1.49919689e-01 2.42319942e-01 -3.98034453e-01
3.01363975e-01 -1.25205946e+00 7.71959364e-01 9.00037289e-01
7.67491519e-01 -4.97000515e-01 6.33471131e-01 5.62013447e-01
-4.30403292e-01 1.16422422e-01 -2.59931356e-01 5.42051159e-02
7.20863640e-01 1.07893097e+00 1.38056632e-02 4.95389432e-01
7.58198738e-01 7.47632146e-01 -1.67879567e-01 9.07224119e-01
-4.91731435e-01 6.90644085e-01 -1.80636108e-01 8.50727558e-01
-5.70577756e-02 -7.40266681e-01 7.74475694e-01 1.00231874e+00
5.07422507e-01 3.12847942e-01 -2.37969205e-01 1.04506958e+00
2.37848535e-01 -4.14487332e-01 -4.32775557e-01 3.85442972e-01
-1.78191811e-01 9.37144041e-01 -6.23348594e-01 -3.13968778e-01
-2.17193380e-01 1.45553577e+00 -2.57420659e-01 9.71552908e-01
-5.11907816e-01 1.00694120e-01 3.83669376e-01 3.16454291e-01
2.83724368e-01 -2.22280264e-01 -4.07055795e-01 -1.07681262e+00
-8.63250419e-02 -1.14070845e+00 -1.67286992e-01 -1.37462378e+00
-1.23718929e+00 6.46148860e-01 -1.19874761e-01 -1.21608031e+00
-3.00652176e-01 -9.83068869e-02 -5.58302343e-01 1.34274507e+00
-1.69203913e+00 -1.00184894e+00 -4.05320019e-01 6.12635195e-01
5.92961788e-01 3.72129083e-01 6.28197014e-01 1.21030636e-01
-1.24742113e-01 -2.18480483e-01 4.97867703e-01 -4.83612925e-01
7.37947762e-01 -1.23798358e+00 1.47143275e-01 1.23087096e+00
-1.15877412e-01 5.61328053e-01 1.33704460e+00 -7.16343760e-01
-1.16444886e+00 -7.53374755e-01 7.84562945e-01 -1.67270362e-01
4.95827347e-01 1.57003656e-01 -1.33780372e+00 4.42547828e-01
4.19721544e-01 1.50218919e-01 1.15013525e-01 -4.06889081e-01
-1.41370058e-01 -7.02072456e-02 -1.23782670e+00 5.24711311e-01
8.96264493e-01 -4.82337624e-01 -4.54753757e-01 3.46436024e-01
3.53336662e-01 -2.31980979e-01 -6.87170386e-01 1.18112884e-01
2.18587115e-01 -1.30307734e+00 1.38512826e+00 -7.18051270e-02
8.87369573e-01 -5.59942067e-01 -1.92881748e-01 -1.62566006e+00
-5.79015434e-01 -6.66955173e-01 2.79288404e-02 9.61776793e-01
3.23927820e-01 -4.75496560e-01 3.78644258e-01 2.95957506e-01
-1.78992212e-01 -3.19579840e-01 -7.63288617e-01 -7.96600163e-01
8.28437209e-02 -4.36306179e-01 6.53684735e-02 8.15540254e-01
-4.84358191e-01 2.03030884e-01 -7.89231420e-01 2.70007402e-01
1.40946770e+00 1.77429002e-02 5.06401360e-01 -8.95969629e-01
-5.41980684e-01 -2.51635581e-01 5.23385927e-02 -1.28994155e+00
-1.96281597e-01 -6.36792064e-01 4.50306416e-01 -1.91118622e+00
1.68026850e-01 -1.70420229e-01 3.73831838e-02 1.41505063e-01
-1.94056615e-01 6.82439923e-01 -4.46114913e-02 6.27886355e-01
2.33001336e-02 4.85682815e-01 1.72003269e+00 -1.45496339e-01
-1.09506309e-01 -6.43839836e-02 -7.48134375e-01 6.17118180e-01
4.64258790e-01 -5.58564365e-01 -5.67482650e-01 -4.80655819e-01
1.57211334e-01 3.99977982e-01 8.15015376e-01 -7.89964736e-01
1.02589794e-01 -1.54527903e-01 5.91292739e-01 -2.56192952e-01
4.60051924e-01 -6.93978667e-01 6.66443765e-01 4.51005489e-01
-2.35147029e-01 -3.11725706e-01 -1.53242633e-01 7.00925589e-01
-3.77081126e-01 -3.73015612e-01 1.25132585e+00 -3.52379829e-01
-3.98698598e-01 -1.59510985e-01 -6.10973060e-01 -3.38961899e-01
4.43508744e-01 -5.56788802e-01 -2.03158408e-01 -9.39133823e-01
-1.19633985e+00 -3.41032922e-01 6.95776284e-01 -1.70281231e-02
8.34537089e-01 -1.13338637e+00 -1.04114962e+00 1.35853007e-01
-5.05092442e-01 2.12588012e-02 4.77263898e-01 1.00571024e+00
-4.10878539e-01 -2.50963330e-01 5.19565307e-02 -7.23419070e-01
-1.02381766e+00 6.10001445e-01 6.30591035e-01 -8.75640512e-02
-1.03395867e+00 6.67347431e-01 4.44587171e-01 -4.86270227e-02
-1.70378745e-01 3.16708647e-02 9.36236531e-02 -4.24302369e-01
7.64262080e-01 2.96024442e-01 -2.71227490e-02 -8.56049895e-01
1.37769669e-01 7.00702965e-01 2.30287135e-01 -4.86713767e-01
1.60918415e+00 -6.86441123e-01 -4.77435231e-01 4.05497462e-01
1.11664438e+00 -8.94927233e-02 -1.62109149e+00 -3.06072831e-01
-4.98537570e-01 -6.02141798e-01 5.45970857e-01 -1.02632725e+00
-1.32294965e+00 7.99643815e-01 6.54926419e-01 2.23302662e-01
1.55219078e+00 -9.97676402e-02 4.87127125e-01 -8.86245370e-02
-4.95411344e-02 -6.56984687e-01 3.25248152e-01 1.00398913e-01
1.37158179e+00 -1.21528792e+00 2.93430537e-01 -3.40167254e-01
-3.05592090e-01 1.12292182e+00 9.45892036e-02 -1.52445331e-01
5.89840412e-01 3.37518424e-01 2.53370374e-01 -2.59977371e-01
-4.71581727e-01 -5.27737886e-02 2.44869441e-01 9.39947248e-01
3.44754606e-01 -3.76568824e-01 -1.15800209e-01 -5.08380122e-02
1.48496181e-01 7.31926486e-02 7.82938242e-01 6.03609085e-01
-4.32569593e-01 -1.08121538e+00 -9.19728696e-01 2.32202262e-01
-2.99127102e-01 -3.29827338e-01 1.10004701e-01 4.33305085e-01
1.27330655e-02 1.30267251e+00 -1.63899735e-01 4.43223454e-02
1.82721093e-01 -1.82972878e-01 7.18169153e-01 -2.58169413e-01
-1.73959151e-01 5.85437417e-01 -1.54185176e-01 -5.16683042e-01
-7.68638670e-01 -7.47918367e-01 -1.03240108e+00 -4.68057871e-01
-2.59704590e-01 -2.20777746e-02 6.53058052e-01 1.03527200e+00
-5.53954393e-02 6.85465991e-01 5.33050597e-01 -1.40294909e+00
-2.35179633e-01 -8.19565296e-01 -6.31208837e-01 7.16184080e-01
9.54771161e-01 -2.41676077e-01 -9.38538313e-01 4.57085639e-01] | [11.641335487365723, -2.344101905822754] |
30c3f8b1-a707-430d-809e-f795522e50d2 | optical-flow-distillation-towards-efficient | 2007.05146 | null | https://arxiv.org/abs/2007.05146v2 | https://arxiv.org/pdf/2007.05146v2.pdf | Optical Flow Distillation: Towards Efficient and Stable Video Style Transfer | Video style transfer techniques inspire many exciting applications on mobile devices. However, their efficiency and stability are still far from satisfactory. To boost the transfer stability across frames, optical flow is widely adopted, despite its high computational complexity, e.g. occupying over 97% inference time. This paper proposes to learn a lightweight video style transfer network via knowledge distillation paradigm. We adopt two teacher networks, one of which takes optical flow during inference while the other does not. The output difference between these two teacher networks highlights the improvements made by optical flow, which is then adopted to distill the target student network. Furthermore, a low-rank distillation loss is employed to stabilize the output of student network by mimicking the rank of input videos. Extensive experiments demonstrate that our student network without an optical flow module is still able to generate stable video and runs much faster than the teacher network. | ['Chunjing Xu', 'Yiman Zhang', 'Yunhe Wang', 'Xinghao Chen', 'Han Shu', 'Chang Xu'] | 2020-07-10 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/13_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123510613.pdf | eccv-2020-8 | ['video-style-transfer'] | ['computer-vision'] | [ 3.70912641e-01 1.90155907e-03 -2.62016594e-01 -1.91863596e-01
-3.21160227e-01 -5.42419374e-01 4.64865863e-01 -4.28756744e-01
-4.45966631e-01 1.09691024e+00 -1.18493237e-01 -3.21502835e-01
3.22789736e-02 -7.99124241e-01 -1.05540645e+00 -7.92174459e-01
1.41209066e-01 -2.02321745e-02 3.34593415e-01 -4.31505106e-02
1.96642339e-01 1.90925464e-01 -1.57655883e+00 8.05983841e-02
1.14415157e+00 1.04270661e+00 2.67854333e-01 7.54422188e-01
-1.21210143e-01 1.23467970e+00 -5.16886830e-01 -5.58600128e-01
1.64502636e-01 -6.28712952e-01 -7.89758444e-01 -1.26290336e-01
9.70283747e-01 -6.17631257e-01 -8.38598728e-01 1.08991754e+00
4.47913051e-01 5.92408240e-01 3.96542430e-01 -1.34321761e+00
-8.24358284e-01 4.52833891e-01 -3.10487360e-01 3.82209003e-01
4.18915898e-01 1.60001069e-01 9.06143367e-01 -8.07561636e-01
4.64738458e-01 1.24434996e+00 4.49859083e-01 7.24635303e-01
-9.96063471e-01 -9.64121997e-01 1.29904479e-01 4.18610543e-01
-1.29334426e+00 -4.26985919e-01 7.50947714e-01 -2.25989699e-01
4.25738007e-01 1.93750486e-02 7.09993184e-01 1.12388623e+00
-7.27088898e-02 9.56619084e-01 9.01249528e-01 -1.84257269e-01
-4.44049314e-02 2.65572459e-01 -3.63648653e-01 9.71666276e-01
1.14901543e-01 -1.08088262e-01 -7.96748161e-01 1.92098618e-01
1.17494500e+00 4.68428656e-02 -7.83954680e-01 -3.81304294e-01
-1.03049564e+00 5.35693169e-01 5.79776704e-01 5.33562116e-02
9.51164216e-02 2.54440427e-01 2.99769878e-01 5.51630378e-01
3.20818365e-01 -2.47695837e-02 -1.56454295e-01 -4.06717837e-01
-1.07879245e+00 7.13963434e-02 6.68714106e-01 1.05649686e+00
9.07407284e-01 2.42415786e-01 -2.74439514e-01 6.21909320e-01
7.27129057e-02 5.57625294e-01 4.95811462e-01 -1.27149642e+00
5.78900993e-01 3.86286616e-01 -1.03446834e-01 -1.15951872e+00
2.17235103e-01 -2.59747922e-01 -9.74045873e-01 1.10224344e-01
5.04620671e-01 -2.60101855e-01 -5.07292747e-01 1.86164284e+00
1.57509938e-01 1.02450120e+00 -1.93947330e-01 8.94829452e-01
5.53826213e-01 6.62623286e-01 -1.29712507e-01 -1.00067623e-01
8.53534281e-01 -1.28241086e+00 -5.26044309e-01 1.62470564e-01
4.86628890e-01 -7.16968477e-01 1.07068133e+00 4.13710475e-01
-1.44010615e+00 -8.76709163e-01 -1.04707384e+00 -3.07835251e-01
8.85071531e-02 4.29721884e-02 5.26866317e-01 5.96984863e-01
-1.26570582e+00 8.81753802e-01 -7.30495036e-01 -6.22440502e-02
6.72289968e-01 5.27562261e-01 -2.86452472e-01 -5.77557124e-02
-1.13990831e+00 5.37832797e-01 4.18746114e-01 1.47431465e-02
-6.69703722e-01 -9.56584096e-01 -7.92959273e-01 2.95614868e-01
2.39019841e-01 -1.02547348e+00 1.08628452e+00 -1.33041394e+00
-2.21239948e+00 4.37769234e-01 -9.69821513e-02 -2.94125199e-01
9.25585270e-01 -4.47961330e-01 -1.06898077e-01 5.21122932e-01
-8.32685456e-02 1.00486887e+00 1.22643447e+00 -7.88415372e-01
-9.88643467e-01 1.53840259e-01 3.46492261e-01 3.37609023e-01
-8.39211702e-01 -4.68959898e-01 -6.69376016e-01 -7.42138505e-01
-2.93205261e-01 -1.11185789e+00 1.91653818e-01 5.11569321e-01
-3.17374989e-02 -3.04156601e-01 1.09305990e+00 -3.86375099e-01
1.34315228e+00 -2.10340476e+00 2.22298637e-01 1.69122219e-02
2.67639041e-01 6.78647161e-01 -2.36779064e-01 -1.63285568e-01
-7.41794705e-02 -1.99983940e-01 -8.85991603e-02 -9.38157067e-02
-3.88988912e-01 3.49669933e-01 -4.64569390e-01 2.58978367e-01
2.65671790e-01 8.51294219e-01 -1.25933790e+00 -3.69855583e-01
2.66873896e-01 7.88196921e-01 -1.04688907e+00 3.30755979e-01
1.73589289e-02 6.45384789e-01 -3.07280689e-01 1.44819185e-01
6.33068383e-01 -3.47794592e-01 1.72380418e-01 -2.42656767e-01
7.12597817e-02 3.58942062e-01 -1.10427392e+00 1.92279220e+00
-4.57673311e-01 9.19166744e-01 6.59973174e-02 -1.15158856e+00
6.48297310e-01 1.96406424e-01 4.26858813e-01 -5.79167068e-01
1.69307798e-01 1.37645170e-01 -9.75333005e-02 -4.89598513e-01
5.17817199e-01 -6.24657087e-02 5.41865647e-01 4.04707044e-01
1.12198636e-01 -1.72660518e-02 2.76563764e-01 2.92693406e-01
8.80057037e-01 5.49452364e-01 -2.83399373e-01 -2.83357084e-01
9.85032558e-01 -5.84687650e-01 6.71664774e-01 5.45056581e-01
-2.81235129e-01 5.76301336e-01 3.46964508e-01 -4.70230937e-01
-8.25723290e-01 -1.22002363e+00 2.74261117e-01 1.15223491e+00
4.47992414e-01 -2.56934702e-01 -7.43878186e-01 -8.27380717e-01
-9.89787728e-02 1.83802113e-01 -2.93427318e-01 -3.60392690e-01
-9.28729475e-01 -6.69433922e-02 7.68124461e-01 5.78456819e-01
8.76943886e-01 -7.63954699e-01 -5.77890933e-01 7.32359588e-02
-4.41996813e-01 -1.18456221e+00 -8.65742326e-01 -5.99274755e-01
-1.03939235e+00 -1.09447265e+00 -9.05444443e-01 -9.35327768e-01
7.03179061e-01 6.16539419e-01 9.49996948e-01 3.33564848e-01
1.66202653e-02 3.79587889e-01 -4.21033688e-02 1.27501622e-01
-2.36148044e-01 2.04161376e-01 2.90663302e-01 8.03763866e-02
-2.72228420e-02 -8.46462429e-01 -7.96923757e-01 2.48825133e-01
-1.01585042e+00 1.56099722e-01 4.78376567e-01 9.63863373e-01
1.42432377e-01 -2.26341225e-02 3.03929061e-01 -7.50338316e-01
3.80301058e-01 -2.79559195e-01 -4.39822525e-01 1.61782399e-01
-4.49600428e-01 2.02260405e-01 8.94450366e-01 -6.91501737e-01
-1.22589600e+00 -1.53013766e-01 4.50609904e-03 -8.00434530e-01
1.14193685e-01 9.09278989e-02 1.56054109e-01 -2.46232137e-01
3.03137451e-01 2.16072887e-01 2.32722297e-01 -1.68611154e-01
3.28024030e-01 4.41008091e-01 7.52736747e-01 -7.19603837e-01
1.01532614e+00 4.87368703e-01 -6.23975955e-02 -9.18688297e-01
-8.61401021e-01 -1.41008630e-01 -3.68315518e-01 -2.11671367e-01
5.46493530e-01 -1.08808649e+00 -1.16084719e+00 5.78745484e-01
-1.12086678e+00 -3.58967423e-01 -1.80874914e-01 6.05843842e-01
-5.84824920e-01 4.01833355e-01 -7.84346461e-01 -3.71205181e-01
-2.11035520e-01 -1.15035880e+00 5.98915339e-01 4.41609949e-01
6.33010045e-02 -1.20511317e+00 -1.76035494e-01 3.51086855e-01
4.98779565e-01 -7.86319152e-02 5.93690336e-01 7.85095394e-02
-8.42431724e-01 3.67554694e-01 -4.42709684e-01 4.35019761e-01
4.54579324e-01 2.12971002e-01 -1.00920999e+00 -6.31918490e-01
-1.69163421e-01 -3.16486120e-01 9.05719757e-01 7.95885473e-02
1.43412304e+00 -4.39720422e-01 -1.26992568e-01 7.43466437e-01
1.00694919e+00 4.65105437e-02 7.76423872e-01 1.02040038e-01
9.67973053e-01 3.40334058e-01 2.30383128e-01 1.37036562e-01
4.44569021e-01 5.10413289e-01 2.14169696e-01 1.03082575e-01
-2.46289760e-01 -3.49977672e-01 8.05593371e-01 1.19096208e+00
-4.08787876e-01 -9.11459625e-02 -3.81819665e-01 2.34952107e-01
-1.83413827e+00 -1.18983364e+00 3.05869371e-01 2.12992358e+00
1.06784952e+00 2.87090689e-01 7.95486346e-02 6.37488514e-02
7.50986755e-01 1.88275039e-01 -5.00679851e-01 -1.32779673e-01
1.25441089e-01 4.27343160e-01 2.64298767e-01 5.78729033e-01
-9.58442926e-01 8.43692183e-01 5.82539797e+00 9.57697749e-01
-1.46010792e+00 -3.78342047e-02 5.67726314e-01 -1.70905009e-01
-2.73699492e-01 -1.87533081e-01 -5.81385255e-01 7.26997912e-01
7.94032097e-01 -2.53790736e-01 5.43260038e-01 5.99408269e-01
2.27326035e-01 5.32110520e-02 -1.02355146e+00 1.17047691e+00
1.01442888e-01 -1.46767235e+00 5.20595312e-01 -2.38203764e-01
8.33800912e-01 -3.41234058e-01 2.13266790e-01 3.93867999e-01
7.41892532e-02 -8.23235750e-01 5.58953524e-01 4.01151359e-01
1.00380063e+00 -7.94792712e-01 3.34668338e-01 2.29955658e-01
-1.20498204e+00 -2.05320329e-03 -3.00609589e-01 -3.47921759e-01
-4.83726114e-02 4.49416846e-01 -4.92325664e-01 5.74002981e-01
7.36018896e-01 9.78876054e-01 -3.18125784e-01 8.97243917e-01
-2.36617342e-01 6.60302758e-01 -1.08829647e-01 2.52112955e-01
2.22524390e-01 -4.44844306e-01 5.13256073e-01 1.04830539e+00
5.01732528e-01 1.54411076e-02 2.40324363e-01 5.74348092e-01
-6.08065963e-01 -1.84209824e-01 -6.11771643e-01 1.71243504e-01
3.79093766e-01 1.17451787e+00 -4.13001418e-01 -5.55045545e-01
-5.11686265e-01 1.35175216e+00 4.16883975e-01 3.22730601e-01
-1.05480063e+00 -5.15059173e-01 9.67978179e-01 3.89269996e-03
3.21614474e-01 -2.26901874e-01 1.88155279e-01 -1.50899255e+00
3.17893503e-03 -6.05558991e-01 1.93646193e-01 -7.39257574e-01
-9.83752608e-01 3.04493994e-01 -2.04736412e-01 -1.36123002e+00
-1.87733546e-01 -7.67132640e-01 -6.62430465e-01 6.97762907e-01
-1.86367190e+00 -7.42424786e-01 -5.64765275e-01 7.50288486e-01
6.73656523e-01 1.94635596e-02 4.06139046e-01 6.95647299e-01
-4.94815409e-01 9.87665117e-01 4.57778722e-02 1.85902610e-01
1.05805588e+00 -1.19741738e+00 -3.01747266e-02 6.94717348e-01
-2.31156833e-02 7.61705041e-01 3.08601469e-01 -2.83080757e-01
-1.35794199e+00 -1.19078362e+00 5.92628837e-01 -2.30286434e-01
6.47196710e-01 9.67727229e-02 -9.98806417e-01 5.50874949e-01
4.21596527e-01 1.81082577e-01 5.09789884e-01 -4.15215820e-01
-4.71952707e-01 -3.94074321e-01 -9.08852279e-01 7.86476314e-01
1.10492027e+00 -6.19688392e-01 -4.08574224e-01 -7.16593191e-02
6.99138939e-01 -5.84051371e-01 -7.17816234e-01 3.07010353e-01
7.71254480e-01 -1.18719637e+00 1.01620471e+00 -5.48625767e-01
7.38974512e-01 -3.46958041e-01 2.84054130e-01 -1.29126477e+00
-2.80868709e-01 -9.55251396e-01 -4.91362840e-01 1.15594792e+00
7.16440603e-02 -4.62403655e-01 1.13330793e+00 3.07541996e-01
2.45332662e-02 -5.73655725e-01 -7.13728607e-01 -8.86285305e-01
1.00882851e-01 -1.38643052e-04 1.99544713e-01 1.04108357e+00
-1.78950146e-01 5.12954712e-01 -5.25618613e-01 -1.57337353e-01
4.87842888e-01 -7.42974579e-02 7.38390446e-01 -1.08227944e+00
-3.73348653e-01 -4.90516394e-01 -3.12387198e-01 -1.65204310e+00
3.55354428e-01 -8.56435120e-01 -1.61428794e-01 -8.60600352e-01
-2.00973954e-02 -3.78874630e-01 -3.95436704e-01 2.66406745e-01
-5.17931223e-01 6.85389221e-01 2.62058884e-01 1.47423789e-01
-5.23366034e-01 7.22699821e-01 1.62884355e+00 -2.65609562e-01
-2.83028603e-01 7.83506706e-02 -4.90485072e-01 8.73033285e-01
8.55524123e-01 -1.27591312e-01 -8.08767617e-01 -6.42550647e-01
6.57760650e-02 -1.22555532e-01 2.87188947e-01 -1.09471440e+00
3.78534406e-01 -4.76082042e-02 3.49831253e-01 -1.46061733e-01
2.55410671e-01 -7.51355469e-01 -2.13737682e-01 6.34391248e-01
-2.41770640e-01 1.45031363e-01 6.41905591e-02 6.38401866e-01
-2.36035451e-01 -1.07151583e-01 1.00681555e+00 -6.88203052e-03
-6.18251145e-01 4.42697048e-01 -3.01545382e-01 2.51140177e-01
8.27646792e-01 -2.78520256e-01 -1.78802446e-01 -4.63651836e-01
-4.03430104e-01 8.52348357e-02 2.19138801e-01 4.12145704e-01
7.39490747e-01 -1.54276431e+00 -4.50135767e-01 2.54113078e-01
-3.41789365e-01 1.40217841e-01 1.55010790e-01 8.25728714e-01
-6.81184828e-01 1.94885671e-01 -4.08245772e-01 -6.49411380e-01
-1.17253876e+00 3.03454816e-01 9.71250609e-02 -1.50709167e-01
-6.57542408e-01 9.72176790e-01 2.41892949e-01 -2.33392030e-01
4.09621924e-01 -1.90035522e-01 -6.22234344e-02 4.26476188e-02
6.78208351e-01 8.88191164e-01 -2.33670041e-01 -3.40231776e-01
-7.70559013e-02 6.08811796e-01 -1.19016796e-01 1.40651971e-01
1.06391656e+00 -2.65872329e-01 -3.56633291e-02 1.41864121e-01
1.26098847e+00 -5.88132553e-02 -1.67476964e+00 -2.06496432e-01
-2.39136994e-01 -5.87326109e-01 -1.49793252e-01 -7.45218247e-02
-1.38372207e+00 1.01978457e+00 3.89165580e-01 -3.76713388e-02
1.15224504e+00 -7.27393746e-01 1.23222518e+00 6.40740395e-01
2.52641797e-01 -1.02122045e+00 5.21393120e-01 7.20253646e-01
3.87635797e-01 -1.21862543e+00 -2.29585782e-01 -5.54762781e-01
-1.92315489e-01 1.22715247e+00 9.49110270e-01 -3.28812063e-01
4.52024758e-01 1.15707606e-01 -1.01296172e-01 4.80067790e-01
-6.52421713e-01 7.13217407e-02 1.56987026e-01 3.77159417e-01
4.36666876e-01 -2.84303069e-01 -1.69283256e-01 -5.39002940e-02
-1.64541408e-01 3.64853799e-01 5.78714132e-01 7.73125052e-01
-3.32953930e-01 -1.06685805e+00 -1.04411609e-01 2.55766749e-01
-5.22387803e-01 -1.74097810e-02 2.01674715e-01 5.60210764e-01
1.29866198e-01 7.54873157e-01 1.61050603e-01 -3.23608577e-01
2.05652282e-01 -8.91347229e-02 7.51733243e-01 -3.29360545e-01
-4.13872868e-01 -2.31976911e-01 -3.95081490e-01 -6.61501288e-01
-7.33229756e-01 -2.56834328e-01 -1.09056759e+00 -6.89324796e-01
-2.62697697e-01 3.71671408e-01 6.56818748e-02 7.91123807e-01
3.84163231e-01 4.53521550e-01 8.50494981e-01 -9.57996130e-01
-3.77945125e-01 -5.65382659e-01 -1.65282249e-01 4.22334641e-01
5.44314742e-01 -8.23208213e-01 -3.67938310e-01 5.55475175e-01] | [10.816245079040527, -1.1866010427474976] |
08f5ab1e-3719-43c5-9fe2-ea8cbe49ab57 | generalized-content-preserving-warps-for | 1809.06783 | null | http://arxiv.org/abs/1809.06783v1 | http://arxiv.org/pdf/1809.06783v1.pdf | Generalized Content-Preserving Warps for Image Stitching | Local misalignment caused by global homography is a common issue in image
stitching task. Content-Preserving Warping (CPW) is a typical method to deal
with this issue, in which geometric and photometric constraints are imposed to
guide the warping process. One of its essential condition however, is colour
consistency, and an elusive goal in real world applications. In this paper, we
propose a Generalized Content-Preserving Warping (GCPW) method to alleviate
this problem. GCPW extends the original CPW by applying a colour model that
expresses the colour transformation between images locally, thus meeting the
photometric constraint requirements for effective image stitching. We combine
the photometric and geometric constraints and jointly estimate the colour
transformation and the warped mesh vertexes, simultaneously. We align images
locally with an optimal grid mesh generated by our GCPW method. Experiments on
both synthetic and real images demonstrate that our new method is robust to
colour variations, outperforming other state-of-the-art CPW-based image
stitching methods. | ['Kai Chen', 'Jian Yao', 'Jingmin Tu'] | 2018-09-18 | null | null | null | null | ['image-stitching'] | ['computer-vision'] | [ 7.50354767e-01 -4.68386680e-01 1.15782894e-01 1.87875882e-01
-3.11836243e-01 -7.69807756e-01 6.84656739e-01 -4.13719684e-01
-9.79675204e-02 4.11332756e-01 9.26441997e-02 3.03831846e-02
-4.90425043e-02 -5.07377088e-01 -5.00868559e-01 -1.06859326e+00
4.29422528e-01 1.84285700e-01 4.87519622e-01 -3.42683792e-01
6.62410438e-01 4.36494023e-01 -1.23807502e+00 -1.23632662e-01
1.13160455e+00 7.94176459e-01 1.77792862e-01 4.07597750e-01
7.18395109e-04 -1.06485253e-02 -4.13610637e-01 -3.77890199e-01
6.33042157e-01 -5.22287667e-01 -5.51748812e-01 6.34432495e-01
4.84293073e-01 -7.94456899e-02 -8.32032636e-02 1.27835858e+00
2.57311136e-01 1.60430282e-01 4.14255947e-01 -1.46787727e+00
-3.87522906e-01 -1.27757773e-01 -1.23941028e+00 -3.34227681e-01
3.21569353e-01 1.02800399e-01 5.38033783e-01 -6.60150528e-01
7.04613268e-01 9.99058604e-01 5.98039091e-01 6.62825480e-02
-1.47321844e+00 -7.79265940e-01 -2.30999172e-01 -8.92344564e-02
-1.42031217e+00 -3.02877724e-01 1.39463735e+00 -4.64337736e-01
1.68808877e-01 5.42462647e-01 6.62849545e-01 6.57387316e-01
6.55924678e-01 1.70009151e-01 1.34145522e+00 -6.79796994e-01
8.39409456e-02 -3.68808568e-01 -3.97663087e-01 4.36655879e-01
1.94302648e-01 3.30497473e-01 -5.54893255e-01 -4.30702835e-01
1.21436369e+00 1.39484271e-01 -6.40543520e-01 -8.19810510e-01
-1.58941305e+00 5.12979150e-01 1.90570280e-01 3.33629698e-01
-6.35771314e-03 6.81409240e-02 1.48830146e-01 2.84747630e-01
6.35042608e-01 4.04611796e-01 5.85904606e-02 7.36868754e-02
-9.51424479e-01 1.46650881e-01 4.96717453e-01 1.00155127e+00
8.20238233e-01 7.60802347e-03 1.96841523e-01 9.65050817e-01
1.49504125e-01 5.13206899e-01 3.27890128e-01 -1.07741928e+00
3.87618959e-01 3.67529511e-01 1.77752614e-01 -1.49271190e+00
7.37764686e-03 2.04589292e-02 -1.07617712e+00 5.33167958e-01
2.68093467e-01 1.15051217e-01 -8.44685078e-01 1.32717001e+00
4.75301802e-01 4.50651020e-01 -2.33459324e-01 8.03881645e-01
3.11004430e-01 7.00212836e-01 -4.64244276e-01 -3.99039745e-01
1.38319480e+00 -8.38803351e-01 -8.31456959e-01 -3.92073870e-01
1.85664639e-01 -1.48811936e+00 6.83965385e-01 3.67892981e-01
-8.73474002e-01 -4.59452659e-01 -1.17235851e+00 5.88829927e-02
1.47766545e-01 -1.50916055e-01 1.70692950e-01 6.16819859e-01
-9.31751907e-01 4.96015847e-01 -6.03812397e-01 -2.28517875e-01
-2.23771900e-01 3.32373291e-01 -5.29258490e-01 -2.61127412e-01
-8.20490777e-01 7.96883106e-01 3.98862511e-01 1.57288298e-01
-4.76061404e-02 -5.21099925e-01 -8.54510903e-01 -1.66632563e-01
5.45963109e-01 -4.89251107e-01 9.00176287e-01 -1.03961873e+00
-1.78651774e+00 9.28785622e-01 -1.15154020e-01 8.27304795e-02
6.11912191e-01 5.77284992e-02 -2.65770167e-01 4.68901396e-02
-1.56518549e-01 4.44863886e-01 1.45143545e+00 -1.82053447e+00
-2.53884882e-01 -1.12483829e-01 -5.29871881e-01 2.57137656e-01
-1.43090084e-01 -4.36670594e-02 -7.95721173e-01 -9.97452021e-01
4.21136081e-01 -1.25564063e+00 -1.25033990e-01 6.34616539e-02
-3.32533062e-01 3.08695912e-01 1.48116136e+00 -7.87538886e-01
1.07294583e+00 -2.28167892e+00 3.96673054e-01 5.03686070e-01
1.22858256e-01 3.10744643e-01 -2.70451516e-01 8.56817305e-01
-1.37112305e-01 -4.15625393e-01 -6.60494328e-01 -4.87421870e-01
-3.80989224e-01 4.38476861e-01 -4.13582146e-01 7.77217031e-01
-8.55191946e-02 6.30369484e-01 -8.55619788e-01 -5.74042737e-01
4.76269066e-01 6.32872999e-01 -2.45136797e-01 1.20451823e-01
1.40520018e-02 6.97516501e-01 -1.18778497e-01 5.24616897e-01
1.23166347e+00 2.12812439e-01 2.31642917e-01 -4.63927150e-01
-4.17527169e-01 -4.67458367e-01 -1.21612263e+00 1.71023488e+00
-2.21901700e-01 6.30199790e-01 2.68307894e-01 -7.36546636e-01
1.21939242e+00 2.51650631e-01 7.29779124e-01 -4.82276589e-01
3.21069986e-01 3.75017673e-01 -2.21133888e-01 -3.04493308e-01
7.22938299e-01 -1.68772459e-01 1.09367684e-01 3.97787273e-01
-4.19752568e-01 -7.61907816e-01 -8.98294970e-02 -9.44446325e-02
6.02069318e-01 2.13106334e-01 3.22670698e-01 -5.14411390e-01
7.15811849e-01 -1.28164351e-01 8.19321930e-01 2.48367205e-01
-2.15552822e-02 1.18863368e+00 3.19704860e-01 -4.16259706e-01
-1.35373354e+00 -5.44404328e-01 7.50206485e-02 2.81835794e-01
8.23478162e-01 -2.73792475e-01 -9.32301819e-01 -1.66043952e-01
-1.65837497e-01 2.86714226e-01 -4.83595133e-01 -6.48127645e-02
-8.62238169e-01 -3.95075709e-01 3.00799264e-04 8.58701319e-02
8.97790372e-01 -6.14274919e-01 -7.45824337e-01 2.85132200e-01
-3.62066835e-01 -1.21576214e+00 -1.13072157e+00 -5.56613922e-01
-8.79615188e-01 -1.18146467e+00 -1.02929473e+00 -9.47502315e-01
9.83618021e-01 9.52537417e-01 7.14644015e-01 4.45092022e-01
-3.03980112e-01 1.86867222e-01 -4.94059682e-01 1.65936098e-01
-4.70700800e-01 -2.94038475e-01 -2.41897985e-01 4.39800858e-01
-3.91529948e-01 -5.41128814e-01 -7.61979699e-01 9.53033984e-01
-1.54960918e+00 5.86323142e-01 3.65602642e-01 1.02623558e+00
4.50407982e-01 3.06278914e-01 -3.04648280e-01 -6.29474521e-01
3.84557456e-01 2.41516590e-01 -8.32990587e-01 2.98992157e-01
-4.72858608e-01 1.29754690e-03 3.39514703e-01 -6.32556677e-01
-1.12946939e+00 2.77891338e-01 1.37252644e-01 -5.13582885e-01
2.66294301e-01 2.08262101e-01 -1.80806965e-01 -8.63275170e-01
1.29804417e-01 4.00469571e-01 4.17283922e-01 -3.34269255e-01
2.66474813e-01 4.58899975e-01 9.11044180e-01 -5.66635191e-01
1.41363585e+00 8.36857855e-01 2.84714878e-01 -9.62956548e-01
-1.73846394e-01 -5.26866198e-01 -8.30609560e-01 -3.89592081e-01
6.26745641e-01 -3.23488384e-01 -6.33819878e-01 9.53223050e-01
-1.26622045e+00 -1.13942891e-01 1.31448314e-01 2.33343214e-01
-5.89047074e-01 1.02075231e+00 -2.48930231e-01 -5.25398135e-01
-3.06990176e-01 -1.34486032e+00 1.15576911e+00 -2.29590982e-02
-5.37804067e-02 -9.70857978e-01 2.89859414e-01 1.97917446e-01
4.90398854e-01 7.94916630e-01 6.40971661e-01 4.76962864e-01
-5.26863873e-01 -1.48625046e-01 -1.94616467e-01 1.20876074e-01
5.76256037e-01 2.86677420e-01 -4.59205002e-01 -5.51992416e-01
2.97212303e-01 4.23797280e-01 5.56926191e-01 2.70118028e-01
6.84841096e-01 -2.26954132e-01 -3.60751569e-01 7.85367548e-01
1.65414310e+00 3.41033936e-01 9.07199383e-01 5.27499497e-01
8.52652967e-01 6.70848012e-01 7.83501804e-01 2.55669743e-01
-1.17931478e-01 1.22572696e+00 4.42660898e-01 -5.02251863e-01
-6.93451092e-02 -1.17742717e-01 1.50273353e-01 7.42507637e-01
-2.56877631e-01 -5.63561209e-02 -7.39677489e-01 2.45583743e-01
-2.04849291e+00 -6.85083389e-01 -2.14651331e-01 2.53857946e+00
7.79618979e-01 -2.79511660e-01 -1.99438572e-01 4.05299634e-01
1.12544787e+00 4.71943974e-01 -1.21606424e-01 -2.45894894e-01
-9.22083855e-02 4.16196231e-03 8.56673241e-01 7.56275892e-01
-9.27699745e-01 8.42527688e-01 5.77464437e+00 9.87144351e-01
-1.13322878e+00 -9.12617669e-02 2.43056938e-01 4.52753961e-01
-3.87976676e-01 3.50496918e-01 -1.73630118e-01 5.75698853e-01
-4.47505387e-03 -7.87060782e-02 4.30729151e-01 3.06086481e-01
1.52929395e-01 -2.88487494e-01 -4.82461929e-01 1.20320630e+00
3.73788685e-01 -1.13430512e+00 -2.10784838e-01 2.75503397e-01
9.52050388e-01 -6.82501495e-01 2.06304371e-01 -7.25982368e-01
3.55310887e-02 -6.00180566e-01 6.60858750e-01 1.39632702e-01
9.50715005e-01 -8.86365950e-01 5.91920197e-01 -2.91207489e-02
-1.41575277e+00 3.07824969e-01 -1.94118366e-01 3.65869939e-01
5.08698523e-01 6.70764625e-01 -2.57686913e-01 7.51396358e-01
3.32317740e-01 6.33223951e-01 -2.56389350e-01 1.17804694e+00
-2.04229251e-01 1.00436851e-01 -2.07780093e-01 6.59386456e-01
1.65931404e-01 -9.25841928e-01 8.12078536e-01 7.26961553e-01
6.05990589e-01 2.28266597e-01 9.87296775e-02 6.88858390e-01
2.20036373e-01 -6.80100769e-02 -5.66261590e-01 2.56074816e-01
4.01788473e-01 1.22827399e+00 -1.06753457e+00 -1.38187781e-01
-2.98482984e-01 1.55074596e+00 -4.00270820e-01 3.81135941e-01
-7.16092467e-01 -3.95092100e-01 7.11270869e-01 1.30523592e-01
1.86612830e-01 -7.07359910e-01 -2.59523243e-01 -1.18173134e+00
-2.13901848e-02 -1.00540590e+00 8.37360546e-02 -7.77366579e-01
-9.11629081e-01 4.19361085e-01 8.64651799e-02 -1.98927331e+00
8.53703544e-02 -3.73459488e-01 -7.74974346e-01 7.69219041e-01
-1.64989185e+00 -1.42198563e+00 -5.58659911e-01 6.69813693e-01
5.21684110e-01 3.38318825e-01 4.02575523e-01 -1.43016689e-02
-3.27600151e-01 2.20656797e-01 3.05814952e-01 -1.92633882e-01
1.08399403e+00 -7.45970666e-01 5.08869648e-01 1.18565977e+00
-6.64721280e-02 5.75531006e-01 9.82073307e-01 -7.53180087e-01
-1.61008728e+00 -7.21465826e-01 5.94428122e-01 1.37264505e-01
4.25054461e-01 -1.17312387e-01 -1.07397747e+00 3.97577494e-01
6.15342081e-01 -1.56768009e-01 2.72066921e-01 -7.35072672e-01
-4.14828628e-01 -8.77821371e-02 -1.19966412e+00 7.04158485e-01
8.55805159e-01 -2.21507892e-01 -4.25667793e-01 7.24280626e-02
5.72296143e-01 -7.29142845e-01 -7.57617831e-01 2.82331079e-01
6.40243590e-01 -1.11452222e+00 1.09433270e+00 4.61961240e-01
3.83867830e-01 -7.60966778e-01 2.82194376e-01 -1.54453933e+00
-3.79109174e-01 -1.38101470e+00 3.43129635e-01 1.31888509e+00
-3.63061726e-01 -1.02263951e+00 5.98263562e-01 6.24123752e-01
-4.85672466e-02 -8.65883157e-02 -9.53513920e-01 -9.94146585e-01
-2.37892851e-01 8.65235999e-02 6.71751022e-01 1.20269632e+00
2.44905576e-02 -2.40732700e-01 -8.82110000e-01 2.08261266e-01
9.60444987e-01 3.04342002e-01 9.05609548e-01 -1.04750013e+00
-1.99541003e-01 -3.75935763e-01 -4.94825214e-01 -7.34702587e-01
-7.78210461e-02 -1.97684184e-01 1.92209333e-01 -1.05870867e+00
-2.08174083e-02 -3.06680411e-01 3.69913161e-01 2.89197564e-01
-1.49322361e-01 6.01164043e-01 3.77475351e-01 5.96964300e-01
3.46232265e-01 5.71165323e-01 1.49839175e+00 -1.07718669e-02
-2.92996794e-01 -2.84707457e-01 -2.65190423e-01 4.42827761e-01
7.15031028e-01 -2.25478083e-01 -3.39209884e-01 -5.65189838e-01
1.43956451e-03 1.54066458e-01 2.00036287e-01 -9.10309911e-01
3.14049840e-01 -5.21090329e-01 -1.72085632e-02 -5.68648934e-01
4.04575169e-01 -1.26452756e+00 7.16114581e-01 6.80433154e-01
1.11137278e-01 1.80033967e-01 7.12132454e-02 3.86263043e-01
-3.63844514e-01 -1.15876980e-01 1.10642636e+00 2.56231219e-01
-5.12166142e-01 2.61238962e-01 1.18388580e-02 -3.78776729e-01
1.13414538e+00 -6.63364112e-01 -3.87621135e-01 -3.62909049e-01
-2.53966711e-02 -1.68266725e-02 1.10530436e+00 3.41769814e-01
8.16879392e-01 -1.53332806e+00 -5.64904630e-01 6.24210775e-01
2.14948192e-01 3.87397297e-02 1.36328369e-01 9.48708832e-01
-9.12792504e-01 5.53323738e-02 -3.57994407e-01 -6.19175792e-01
-1.76004362e+00 6.22186124e-01 1.10510938e-01 -6.16915412e-02
-8.29093575e-01 4.37908739e-01 2.81210184e-01 -7.73552060e-02
-2.38742158e-01 -1.40122175e-01 2.01673254e-01 -2.95154870e-01
4.75929409e-01 3.07796985e-01 5.20495325e-02 -9.14351165e-01
-1.43325955e-01 1.48689902e+00 6.36624619e-02 -4.67541277e-01
1.22480989e+00 -3.59249860e-01 -3.54929715e-01 -1.90553054e-01
1.22860372e+00 1.60134673e-01 -1.48760271e+00 -4.46460783e-01
-3.07009131e-01 -1.23829675e+00 2.72107963e-02 -7.50167817e-02
-1.38783884e+00 5.60585082e-01 3.73936534e-01 2.07632452e-01
1.34309268e+00 -4.58237410e-01 1.17931998e+00 -3.68938327e-01
3.68957013e-01 -1.04377460e+00 6.35244399e-02 1.61990151e-01
1.10655951e+00 -8.90215576e-01 2.72712290e-01 -8.83955240e-01
-5.19663811e-01 1.33952165e+00 2.17838362e-01 -2.54833400e-01
2.65928328e-01 2.29960069e-01 5.71419112e-02 2.36735679e-02
-6.17695600e-02 1.38043642e-01 4.06056345e-01 5.57999432e-01
9.30292010e-02 -1.99648261e-01 -5.93421519e-01 -5.29922545e-01
8.12569410e-02 -2.02415824e-01 4.73679692e-01 1.22677827e+00
-3.23421597e-01 -1.69261706e+00 -1.16199970e+00 -2.96290189e-01
3.25610377e-02 4.87666801e-02 -4.74105686e-01 7.05773652e-01
-8.86131823e-02 8.75912011e-01 -9.53066945e-02 -5.31789303e-01
2.23144427e-01 -2.81715184e-01 6.10180914e-01 -5.38342670e-02
-3.96831900e-01 5.75383663e-01 -2.66777217e-01 -5.57803690e-01
-7.19363332e-01 -4.94707525e-01 -8.32875609e-01 -4.92863864e-01
-5.01846313e-01 -1.22284777e-01 7.10486948e-01 7.31210530e-01
2.02916920e-01 4.22925828e-03 1.08026135e+00 -1.29875946e+00
-1.03431433e-01 -4.06466275e-01 -7.10416973e-01 7.49581873e-01
4.91889149e-01 -7.40595937e-01 -5.81917226e-01 4.54404294e-01] | [9.397333145141602, -2.3349509239196777] |
594a4407-dbd0-4f30-b0e4-2808aca112a0 | towards-suicide-prevention-from-bipolar | 2307.00995 | null | https://arxiv.org/abs/2307.00995v1 | https://arxiv.org/pdf/2307.00995v1.pdf | Towards Suicide Prevention from Bipolar Disorder with Temporal Symptom-Aware Multitask Learning | Bipolar disorder (BD) is closely associated with an increased risk of suicide. However, while the prior work has revealed valuable insight into understanding the behavior of BD patients on social media, little attention has been paid to developing a model that can predict the future suicidality of a BD patient. Therefore, this study proposes a multi-task learning model for predicting the future suicidality of BD patients by jointly learning current symptoms. We build a novel BD dataset clinically validated by psychiatrists, including 14 years of posts on bipolar-related subreddits written by 818 BD patients, along with the annotations of future suicidality and BD symptoms. We also suggest a temporal symptom-aware attention mechanism to determine which symptoms are the most influential for predicting future suicidality over time through a sequence of BD posts. Our experiments demonstrate that the proposed model outperforms the state-of-the-art models in both BD symptom identification and future suicidality prediction tasks. In addition, the proposed temporal symptom-aware attention provides interpretable attention weights, helping clinicians to apprehend BD patients more comprehensively and to provide timely intervention by tracking mental state progression. | ['Jinyoung Han', 'Seungbae Kim', 'Hyolim Jeon', 'Sejung Son', 'Daeun Lee'] | 2023-07-03 | null | null | null | null | ['multi-task-learning'] | ['methodology'] | [-1.74349815e-01 1.90802589e-01 -2.39686057e-01 -7.94516563e-01
-7.01718271e-01 1.79960385e-01 2.51272291e-01 7.42809713e-01
-4.59416717e-01 5.22771180e-01 7.16456473e-01 2.85924882e-01
-3.28286350e-01 -5.09889781e-01 5.87971628e-01 -4.10358250e-01
-4.79092449e-01 8.62922907e-01 -3.87709290e-01 -2.54833430e-01
4.46888626e-01 4.06322807e-01 -5.16894460e-01 6.70558751e-01
7.76303351e-01 1.05759418e+00 7.11658821e-02 1.60744667e-01
4.40058827e-01 7.32287943e-01 -4.62465376e-01 -7.29507089e-01
-3.08895737e-01 -2.01313078e-01 -1.02528036e+00 2.73106277e-01
-3.00250471e-01 -7.24313676e-01 -4.79194462e-01 7.62775064e-01
7.99327433e-01 -2.29894400e-01 9.07322109e-01 -1.31543446e+00
-1.00831115e+00 7.75195599e-01 -5.99274755e-01 5.79938650e-01
5.62763572e-01 2.87235320e-01 9.80599999e-01 -6.83318973e-01
2.93623030e-01 1.22782922e+00 9.39574003e-01 1.05523729e+00
-1.25520945e+00 -8.45266998e-01 1.87177494e-01 7.38093436e-01
-1.13521576e+00 -5.75254321e-01 7.72176623e-01 -5.46056271e-01
1.39935529e+00 -3.78448814e-01 1.26350057e+00 1.57665277e+00
4.46108639e-01 5.60511112e-01 6.10416353e-01 2.80356884e-01
1.60935849e-01 -1.49335369e-01 6.52013898e-01 4.81951803e-01
1.34404562e-02 -6.82445467e-01 -5.29239774e-01 -7.56918252e-01
3.53484303e-01 5.38863063e-01 2.07232218e-02 3.51605773e-01
-7.54428267e-01 8.16017926e-01 4.34903175e-01 2.21493855e-01
-6.96019530e-01 -2.46801272e-01 9.34947491e-01 3.56354304e-02
8.67363930e-01 1.07962810e-01 -6.05247080e-01 -4.52803671e-01
-7.93529987e-01 2.17473716e-01 3.90214115e-01 1.92004293e-01
5.02727404e-02 -1.43284544e-01 -4.04030114e-01 1.27099907e+00
5.30884445e-01 2.50746489e-01 1.06673288e+00 -6.18521690e-01
5.79117060e-01 1.13035500e+00 -6.62185550e-02 -1.22806048e+00
-1.06892920e+00 -2.04843014e-01 -1.07297194e+00 -5.21318913e-01
-1.69275597e-01 -4.26015824e-01 -3.52127820e-01 1.64341152e+00
-7.78000504e-02 9.71095785e-02 -7.10986778e-02 4.50186551e-01
6.05903149e-01 2.84059867e-02 2.22327143e-01 -6.01976871e-01
1.43046999e+00 -5.00170231e-01 -7.20560431e-01 -7.20189333e-01
8.07516515e-01 9.03590843e-02 7.05996037e-01 9.43947792e-01
-1.34745073e+00 7.16839656e-02 -5.17996848e-01 -4.93434258e-03
1.64303094e-01 1.84775770e-01 8.83003891e-01 5.31552911e-01
-1.05085802e+00 7.38596678e-01 -1.19542933e+00 -7.30172455e-01
1.03137243e+00 5.47816634e-01 -1.99815810e-01 1.14041813e-01
-1.18307066e+00 1.01970518e+00 8.33725482e-02 6.66651502e-02
-9.06293452e-01 -6.80809617e-01 -5.88787735e-01 1.13937639e-01
-4.44111258e-01 -8.38278770e-01 1.31785476e+00 -7.22784340e-01
-9.85378325e-01 1.06063914e+00 -2.80954331e-01 -5.77853441e-01
3.24541599e-01 -1.45346016e-01 -6.48705184e-01 4.97874081e-01
1.96044490e-01 5.50224125e-01 8.07191014e-01 -4.41368490e-01
-3.68769288e-01 -1.31316769e+00 -3.26395392e-01 2.86116362e-01
-1.26335573e+00 2.18469620e-01 1.42057985e-01 -4.33583379e-01
1.15440078e-01 -6.10076368e-01 -4.07274038e-01 1.29493311e-01
-4.62806135e-01 -5.18174708e-01 4.88204092e-01 -5.39837778e-01
1.58622777e+00 -1.93226945e+00 -1.68202147e-01 -4.03800160e-01
8.45066786e-01 2.36920908e-01 8.73583630e-02 7.03627646e-01
-3.66405606e-01 1.53293967e-01 6.03753403e-02 -7.65737951e-01
-2.58370489e-01 -2.08313659e-01 -1.85533062e-01 7.20267832e-01
4.08417195e-01 7.38332331e-01 -1.00280964e+00 -2.33020246e-01
-1.94437787e-01 4.29212868e-01 -8.41111839e-01 2.66718388e-01
2.22127154e-01 2.78071910e-01 -7.84815848e-01 5.16109347e-01
5.89916468e-01 -8.01764846e-01 2.38584742e-01 1.04409166e-01
4.74739730e-01 5.91686070e-01 -4.06787336e-01 1.08057559e+00
1.49619319e-02 1.80844545e-01 -1.25355527e-01 -9.76814032e-01
9.65128839e-01 5.83909154e-01 9.84942675e-01 -4.23599839e-01
2.78012872e-01 -1.30878553e-01 2.11530738e-02 -8.08207750e-01
1.15521245e-01 -5.34540415e-01 -3.53316590e-02 6.66396201e-01
-2.92399168e-01 3.24419320e-01 3.83575703e-03 3.17720354e-01
1.52200365e+00 -8.63701820e-01 7.25027502e-01 1.97642073e-01
4.28056002e-01 -1.86647654e-01 8.81197333e-01 3.81588370e-01
-7.65665412e-01 4.25673693e-01 6.94030643e-01 -5.02019525e-01
-5.75491130e-01 -8.01682115e-01 -4.52531546e-01 4.55731273e-01
-4.14160937e-01 -7.11252689e-01 -4.69829947e-01 -5.91497838e-01
5.93215637e-02 7.48877168e-01 -7.22277641e-01 -8.60978246e-01
6.07942939e-02 -1.22872555e+00 7.11363614e-01 7.29380965e-01
4.76114392e-01 -1.35605550e+00 -1.29170015e-01 3.70857567e-01
-4.99997824e-01 -6.14048004e-01 -3.04637372e-01 -4.75593954e-02
-1.22888529e+00 -1.17585731e+00 -7.96833813e-01 -7.03965545e-01
5.62051177e-01 7.23658800e-02 8.74504864e-01 -1.74501911e-02
-2.75962859e-01 4.30907607e-01 -3.89042348e-01 -1.28879085e-01
-1.69473756e-02 -2.73647271e-02 5.34251988e-01 1.77514762e-01
9.32034254e-01 -9.60348189e-01 -1.07998991e+00 -2.14605093e-01
-4.33447540e-01 -8.04527178e-02 2.34657392e-01 6.70712113e-01
4.87884656e-02 -1.17575146e-01 1.03320682e+00 -6.95391059e-01
1.12055182e+00 -1.19039333e+00 3.86963725e-01 -1.91594899e-01
-1.10343504e+00 -5.18874109e-01 4.63152051e-01 -5.43539286e-01
-9.68839765e-01 -6.69968054e-02 -4.64787453e-01 7.65815377e-02
-4.00658309e-01 6.15584970e-01 2.43266732e-01 5.25833547e-01
4.35007334e-01 3.48795444e-01 -2.69926578e-01 -5.26207447e-01
-3.78262103e-01 1.11643469e+00 -9.87166911e-02 7.80071765e-02
-2.12332577e-01 4.86941665e-01 -1.71475500e-01 -8.11478615e-01
-9.99128044e-01 -7.89979577e-01 -5.54476261e-01 7.30553940e-02
8.46987128e-01 -8.16302836e-01 -1.34726918e+00 8.29873741e-01
-1.23003340e+00 -1.60839543e-01 4.07431006e-01 1.90743566e-01
-7.89895594e-01 4.39275444e-01 -1.03917563e+00 -1.08535910e+00
-9.46353972e-01 -8.01956058e-01 1.01863360e+00 -3.38523060e-01
-8.74613285e-01 -1.16866291e+00 4.96343523e-01 4.22205955e-01
-1.19527690e-01 -3.29020917e-01 1.24214506e+00 -9.84459043e-01
5.22097111e-01 -2.29140371e-01 -5.20320714e-01 -6.46910816e-02
3.85231763e-01 -4.42628771e-01 -6.29441261e-01 -2.59779125e-01
3.47609043e-01 -7.91981936e-01 9.11547303e-01 5.05449891e-01
1.02439106e+00 -4.97016907e-01 -6.00239694e-01 2.51520544e-01
9.20414329e-01 3.37412149e-01 5.49327731e-01 2.78727770e-01
4.14578408e-01 4.22106355e-01 2.89378285e-01 1.56052792e+00
9.62945759e-01 5.52544594e-01 4.80988383e-01 4.04631644e-01
3.93881679e-01 7.49553517e-02 6.42813146e-01 5.59423566e-01
3.05965453e-01 -2.43617103e-01 -1.25716972e+00 7.67318487e-01
-1.84023154e+00 -1.00196874e+00 -5.32269657e-01 1.63419032e+00
9.10663188e-01 -7.05835447e-02 4.29509014e-01 2.13215396e-01
5.07771015e-01 -5.10186404e-02 -1.01567268e+00 -4.33304995e-01
1.72915906e-01 -2.40182117e-01 -3.22465360e-01 -6.60101837e-03
-8.51661921e-01 7.09288478e-01 6.83949995e+00 3.06507319e-01
-1.04311848e+00 2.70550758e-01 8.77849042e-01 -5.05876064e-01
4.60413564e-03 -5.25339067e-01 -9.39225018e-01 4.99489307e-01
9.61256742e-01 -3.14927548e-01 3.16273659e-01 6.38972521e-01
9.65566695e-01 2.12206617e-01 -1.05918133e+00 1.15945983e+00
3.08278531e-01 -6.90041900e-01 1.21948369e-01 5.34611866e-02
4.23922479e-01 9.45652574e-02 1.99793696e-01 2.18892485e-01
2.33171344e-01 -8.14160287e-01 1.73557296e-01 1.04976690e+00
4.93104219e-01 -5.75280547e-01 6.74082160e-01 4.65576947e-01
-7.81372607e-01 -7.07138598e-01 -2.35090882e-01 -3.26156288e-01
4.13835406e-01 6.75870836e-01 -1.31545568e+00 -1.41127497e-01
6.50730312e-01 1.61493540e+00 -6.91756964e-01 9.76612985e-01
-9.41974595e-02 6.00571752e-01 3.19210857e-01 8.40843543e-02
2.30534032e-01 -8.87264870e-03 1.74746633e-01 9.13160086e-01
5.26197493e-01 5.30188322e-01 4.15034741e-02 8.37346375e-01
1.61406249e-01 2.68445045e-01 -5.43643355e-01 -6.78709447e-01
-2.26727743e-02 1.32216763e+00 -4.70750839e-01 -1.77254930e-01
-4.14043039e-01 1.12456846e+00 7.94661582e-01 -2.31858306e-02
-3.98709416e-01 2.75392234e-01 7.39946008e-01 3.88852537e-01
2.35975794e-02 1.15526728e-01 -6.03188038e-01 -1.23748755e+00
-4.79584992e-01 -5.34032881e-01 8.70315373e-01 -9.73560810e-01
-2.03020644e+00 6.50597215e-01 -6.34965241e-01 -1.06194556e+00
-2.36628428e-01 -3.70141059e-01 -8.94980609e-01 6.03297055e-01
-1.03452957e+00 -1.04249716e+00 -1.65662949e-03 8.15797389e-01
6.30082905e-01 -1.68509334e-01 1.17933667e+00 1.64172307e-01
-1.01894283e+00 2.17873186e-01 -1.78164229e-01 2.00398844e-02
9.89883542e-01 -1.01046574e+00 7.86455423e-02 -1.76980644e-01
-7.39833355e-01 6.73614860e-01 2.17422739e-01 -1.33688271e+00
-8.97608459e-01 -1.20640600e+00 1.52970266e+00 -3.69418561e-01
1.02790225e+00 8.01063608e-03 -8.57927501e-01 6.90098226e-01
-7.08252117e-02 -5.32160759e-01 1.15103793e+00 5.09562314e-01
9.98366177e-02 -5.32628782e-02 -1.46043074e+00 8.20243061e-01
1.05942988e+00 -4.69620854e-01 -7.29242384e-01 6.87171400e-01
2.08096012e-01 4.23433930e-01 -1.13757586e+00 9.60089713e-02
5.63169420e-01 -9.99195695e-01 9.19829428e-01 -1.05525136e+00
9.43906486e-01 9.79561925e-01 3.03842843e-01 -1.37620628e+00
-8.08292627e-01 -1.42306700e-01 -1.60213172e-01 1.14421916e+00
3.37342292e-01 -4.26184744e-01 8.35113943e-01 1.09753299e+00
2.90069617e-02 -9.24202204e-01 -6.88474536e-01 -2.16470852e-01
4.22141589e-02 -7.57735491e-01 1.76224589e-01 1.12324607e+00
1.06913400e+00 4.22474444e-01 -3.62779915e-01 1.00562833e-01
2.16841370e-01 -3.98020566e-01 -8.07967782e-02 -1.47114456e+00
4.02701497e-02 -5.98310590e-01 -4.49370831e-01 -5.29279232e-01
2.70907104e-01 -1.15881371e+00 -5.86571395e-01 -1.79078221e+00
9.76926565e-01 -5.93921021e-02 -4.65234160e-01 8.93163443e-01
-2.29150906e-01 1.89807057e-01 -3.27977121e-01 1.56695560e-01
-8.72076213e-01 9.48612392e-01 1.13710499e+00 -3.09488863e-01
-3.80707860e-01 2.80255139e-01 -1.03327930e+00 1.14980543e+00
9.73702252e-01 -7.33919740e-01 -4.60252196e-01 -1.99422374e-01
6.21930242e-01 5.55508316e-01 4.29469556e-01 -5.74920654e-01
2.44664028e-01 -1.96990907e-01 5.38273931e-01 -7.26681411e-01
8.37825119e-01 -4.58443850e-01 -5.48200965e-01 6.32249236e-01
-5.18099904e-01 1.81989446e-01 4.18959409e-02 4.85235184e-01
2.61631936e-01 -4.23963755e-01 5.66591203e-01 -2.70473868e-01
-1.38358146e-01 7.87256002e-01 -1.02301836e+00 1.46953583e-01
7.97010183e-01 -1.26349777e-02 -2.00869031e-02 -6.52454615e-01
-1.42312491e+00 2.10097477e-01 -1.24043137e-01 9.49224010e-02
1.10373247e+00 -1.25388551e+00 -5.32554030e-01 2.86742955e-01
1.76824063e-01 -8.36584389e-01 7.62492239e-01 1.32157815e+00
2.14947969e-01 4.56034303e-01 -3.03873599e-01 -1.06339887e-01
-1.20765531e+00 6.26043022e-01 1.07250825e-01 -4.61568296e-01
-6.57539785e-01 6.22896850e-01 -1.09539665e-01 7.48597309e-02
3.22749764e-01 -6.70061409e-02 -7.61009693e-01 6.78824961e-01
9.84891772e-01 6.32281244e-01 2.21830353e-01 -4.86815482e-01
-3.91978234e-01 -9.03894827e-02 -6.62130117e-01 -2.85929721e-02
1.99216497e+00 -4.09261972e-01 -3.31639618e-01 5.20082951e-01
8.96706283e-01 -5.58334410e-01 -8.94470215e-01 -8.87821242e-02
-5.09081222e-02 -1.92421108e-01 2.92810738e-01 -9.79540348e-01
-1.09631932e+00 1.14899051e+00 5.39463639e-01 4.05487388e-01
9.89120603e-01 4.12918739e-02 1.20462561e+00 4.87553775e-01
3.05059135e-01 -8.53389680e-01 6.18345857e-01 6.41202450e-01
9.93781090e-01 -1.33642566e+00 -2.30872869e-01 6.20539859e-02
-1.11069906e+00 1.03651249e+00 6.91022217e-01 -2.13788420e-01
9.02594864e-01 -7.24212127e-03 -1.34498775e-01 -5.92449665e-01
-1.14512229e+00 1.58779368e-01 1.09198622e-01 8.57469082e-01
4.14754510e-01 -4.85951453e-02 -4.41762447e-01 1.64115798e+00
-1.60458788e-01 2.38736704e-01 4.36384767e-01 4.32767123e-01
-3.93810809e-01 -1.15160584e+00 -3.96765620e-02 1.13918877e+00
-6.50909841e-01 -5.31611800e-01 -1.02359855e+00 -8.79033208e-02
-3.06085050e-02 1.13479567e+00 8.50666761e-02 -3.44010323e-01
3.12511809e-02 3.96331310e-01 3.00517917e-01 -1.07758784e+00
-8.21128845e-01 -8.49134624e-02 2.70467788e-01 -3.63114148e-01
-1.17831424e-01 -1.11953032e+00 -1.37025774e+00 -3.59962493e-01
2.01339185e-01 -3.48087341e-01 -5.53192534e-02 1.18248892e+00
5.43705761e-01 5.23454964e-01 6.52006626e-01 -6.43869638e-01
-6.67353630e-01 -1.27022505e+00 -1.16247380e+00 3.71547639e-01
3.07627589e-01 -6.21534526e-01 -7.67163560e-02 -1.69281602e-01] | [8.770743370056152, 10.129104614257812] |
5f703ae0-567d-4574-9c1f-ba8163d0bb05 | agent-centric-risk-assessment-accident | 1705.06560 | null | http://arxiv.org/abs/1705.06560v1 | http://arxiv.org/pdf/1705.06560v1.pdf | Agent-Centric Risk Assessment: Accident Anticipation and Risky Region Localization | For survival, a living agent must have the ability to assess risk (1) by
temporally anticipating accidents before they occur, and (2) by spatially
localizing risky regions in the environment to move away from threats. In this
paper, we take an agent-centric approach to study the accident anticipation and
risky region localization tasks. We propose a novel soft-attention Recurrent
Neural Network (RNN) which explicitly models both spatial and appearance-wise
non-linear interaction between the agent triggering the event and another agent
or static-region involved. In order to test our proposed method, we introduce
the Epic Fail (EF) dataset consisting of 3000 viral videos capturing various
accidents. In the experiments, we evaluate the risk assessment accuracy both in
the temporal domain (accident anticipation) and spatial domain (risky region
localization) on our EF dataset and the Street Accident (SA) dataset. Our
method consistently outperforms other baselines on both datasets. | ['Fu-Hsiang Chan', 'Shih-Han Chou', 'Kuo-Hao Zeng', 'Min Sun', 'Juan Carlos Niebles'] | 2017-05-18 | agent-centric-risk-assessment-accident-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Zeng_Agent-Centric_Risk_Assessment_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Zeng_Agent-Centric_Risk_Assessment_CVPR_2017_paper.pdf | cvpr-2017-7 | ['accident-anticipation'] | ['computer-vision'] | [ 3.50343257e-01 -4.14094150e-01 8.13190192e-02 -1.74206883e-01
-7.23631620e-01 -4.47546661e-01 1.02813423e+00 2.18661383e-01
-7.23661602e-01 6.46709383e-01 6.38930321e-01 -2.63289332e-01
2.19932012e-02 -5.77206314e-01 -5.74135840e-01 -9.66065586e-01
-6.56059206e-01 2.43519515e-01 2.99979001e-01 -3.84181067e-02
6.81860000e-02 4.97759402e-01 -9.78516757e-01 3.37096930e-01
3.04247826e-01 6.38826191e-01 5.09659722e-02 9.34012651e-01
7.47298479e-01 1.65166473e+00 -6.83009207e-01 -1.04967952e-01
4.61895019e-02 -4.10904974e-01 -6.95188820e-01 -3.45593512e-01
-1.03811517e-01 -6.36395097e-01 -6.68904006e-01 6.19933724e-01
5.89465499e-01 2.75999606e-01 1.09567070e+00 -1.29277456e+00
-5.51316023e-01 3.49820286e-01 -6.42950594e-01 8.65414321e-01
4.06545997e-01 5.81634700e-01 4.88784701e-01 -4.82145876e-01
7.57481515e-01 1.27009165e+00 5.74009597e-01 6.09758854e-01
-6.94086909e-01 -3.58707994e-01 5.21548390e-01 4.94548231e-01
-1.15793645e+00 -2.14417532e-01 9.05208826e-01 -7.63972759e-01
1.35266232e+00 1.27482176e-01 3.72997820e-01 1.98994994e+00
6.78671718e-01 6.12197876e-01 8.65168810e-01 2.69833773e-01
3.26471776e-01 -4.88763243e-01 2.73455810e-02 4.70516562e-01
-1.38347641e-01 6.20919228e-01 -2.50971138e-01 -4.12081927e-01
6.60968363e-01 3.32701564e-01 -8.97205770e-02 2.58751661e-01
-1.41498470e+00 6.22287512e-01 7.54288614e-01 2.72576869e-01
-9.26270664e-01 5.04516006e-01 7.73095310e-01 -1.12467520e-01
6.39537156e-01 5.65547496e-03 -1.36488244e-01 -1.67031705e-01
-6.05456889e-01 3.12564909e-01 9.34273452e-02 1.84306838e-02
4.29280549e-02 3.19187529e-02 -6.02917075e-01 5.15667617e-01
1.57157719e-01 3.84649187e-01 1.48032039e-01 -7.01318443e-01
4.67932850e-01 4.76625860e-01 1.70800507e-01 -1.13889253e+00
-6.61430359e-01 -3.82100731e-01 -1.08660841e+00 3.83183390e-01
3.02080572e-01 -4.03119564e-01 -9.14740801e-01 2.21607089e+00
4.97269809e-01 8.29966784e-01 1.19653285e-01 8.35100532e-01
6.80742681e-01 1.03953421e+00 6.69282734e-01 -2.26435557e-01
1.40081596e+00 -8.47295582e-01 -5.06198406e-01 -2.67212063e-01
7.23630548e-01 -1.68582186e-01 6.86406553e-01 1.56719282e-01
-9.32390749e-01 -2.86213815e-01 -5.82242012e-01 1.49586573e-01
-4.84309584e-01 -3.51466268e-01 1.32513881e-01 -2.71059703e-02
-1.06443858e+00 1.31240562e-01 -5.87664425e-01 -4.79174435e-01
3.66923034e-01 -1.05344526e-01 -3.23592067e-01 1.22688763e-01
-1.56457520e+00 8.02486837e-01 1.74990073e-01 1.40278876e-01
-1.61035204e+00 -4.87843424e-01 -8.64735842e-01 -6.51127398e-02
2.25195706e-01 -9.19406295e-01 7.34801412e-01 -8.20333838e-01
-6.71754122e-01 7.45561242e-01 -7.93663114e-02 -6.79062843e-01
4.99140650e-01 -3.47787559e-01 -3.33789170e-01 2.66627520e-01
1.25795797e-01 5.51942050e-01 6.71325088e-01 -1.17664838e+00
-4.86383557e-01 -3.86969268e-01 9.35476795e-02 2.57529676e-01
3.31323184e-02 6.97605312e-01 3.11864708e-02 -1.22568333e+00
-8.42501998e-01 -8.17230225e-01 -3.88350576e-01 -2.72531211e-01
-5.38039505e-01 -4.15524006e-01 8.38215113e-01 -9.36695516e-01
1.31276000e+00 -2.26729059e+00 2.89889991e-01 -8.65152627e-02
2.02532023e-01 2.60153383e-01 -2.91869104e-01 5.84028721e-01
-3.46275747e-01 -8.57926235e-02 -4.98861879e-01 -5.43998659e-01
-2.53902972e-01 3.41977254e-02 -5.64125001e-01 6.19872332e-01
3.52600098e-01 9.43094313e-01 -1.16029644e+00 -5.18410206e-01
-6.44470900e-02 8.86702538e-01 -2.69602180e-01 3.20257485e-01
-9.67604741e-02 5.88713646e-01 -4.18468684e-01 7.54670680e-01
2.59569168e-01 -1.42981231e-01 -2.42614627e-01 3.53465140e-01
1.60765313e-02 1.21739507e-01 -5.59351683e-01 1.04683471e+00
-1.06282450e-01 5.85363209e-01 -2.46382318e-02 -6.48698688e-01
3.15623820e-01 6.12087369e-01 5.24524093e-01 -6.68132544e-01
8.20623115e-02 -3.03105891e-01 -2.10277513e-01 -5.86516738e-01
1.09081611e-01 8.17211941e-02 -3.56766850e-01 9.09849465e-01
-5.56279123e-01 8.00448656e-01 -1.13368683e-01 4.12001401e-01
1.75153935e+00 -1.53048173e-01 1.91643447e-01 2.87164226e-02
3.81123513e-01 -2.49250025e-01 6.91605985e-01 6.27811730e-01
-8.61541092e-01 2.50864297e-01 6.77977443e-01 -9.65784431e-01
-6.31645918e-01 -1.40555000e+00 2.82567203e-01 1.15644789e+00
1.52061402e-03 -1.26601523e-02 -8.23801100e-01 -1.10279548e+00
-3.47183585e-01 8.82336855e-01 -1.11222517e+00 -4.12626475e-01
-9.84759390e-01 -9.61527586e-01 5.38679123e-01 8.47335517e-01
4.30366367e-01 -1.59117091e+00 -1.02384198e+00 2.07874224e-01
-3.79681110e-01 -7.85079122e-01 -7.01521695e-01 -1.96229875e-01
-7.64508769e-02 -1.03788292e+00 -8.02478731e-01 -5.33342719e-01
5.43342590e-01 8.41794908e-02 1.02347898e+00 2.64174044e-01
-3.81663740e-01 3.18655491e-01 -2.53190070e-01 -3.48679513e-01
-1.38828352e-01 -2.64291853e-01 2.42496923e-01 1.07867219e-01
1.40204027e-01 -3.77849013e-01 -8.70037973e-01 4.41151649e-01
-9.42126155e-01 -1.61057562e-01 2.98157096e-01 5.58784425e-01
3.71755242e-01 2.14322761e-01 9.33144927e-01 -2.72078931e-01
6.97303772e-01 -1.09613478e+00 -3.19288313e-01 2.96599299e-01
1.21759966e-01 -3.95227402e-01 4.70651031e-01 -5.38364828e-01
-1.15943491e+00 5.42332679e-02 -5.42825833e-02 -2.53454983e-01
-3.54989469e-01 3.36966336e-01 2.05863431e-01 4.24596190e-01
4.81720954e-01 1.65275231e-01 -5.52964747e-01 1.82019770e-02
2.27917228e-02 3.31224591e-01 5.83775043e-01 -2.95334339e-01
6.31166756e-01 6.38443828e-01 1.57405287e-02 -4.52123344e-01
-5.34472287e-01 -2.10824862e-01 -3.80017877e-01 -7.79783368e-01
1.41467726e+00 -7.83405483e-01 -8.77568424e-01 6.54448330e-01
-1.45528400e+00 -7.61104643e-01 1.55625318e-03 3.22687358e-01
-4.76487994e-01 2.04910323e-01 -8.04008543e-01 -1.23653030e+00
-2.70583630e-01 -1.03443110e+00 1.35817909e+00 -8.65317136e-02
-4.30766374e-01 -1.03131557e+00 7.63022423e-01 1.98625281e-01
2.58469790e-01 9.86576140e-01 8.47173750e-01 -5.33106625e-01
-6.11826360e-01 -2.42982715e-01 -2.65101880e-01 -3.06873918e-01
1.54697463e-01 3.14970687e-02 -8.99394035e-01 -2.36721322e-01
5.23329936e-02 -3.06595176e-01 1.21671236e+00 5.98515928e-01
8.75153482e-01 -4.90418911e-01 -6.03756547e-01 3.28874379e-01
8.33571732e-01 5.09795070e-01 7.36222446e-01 3.54684502e-01
6.27836406e-01 1.05977690e+00 5.79563141e-01 4.39929664e-01
5.76072097e-01 7.55463302e-01 6.03781044e-01 -3.31117302e-01
6.68941960e-02 -2.79559940e-02 8.38318527e-01 1.09147213e-01
-1.83861956e-01 -8.28610122e-01 -1.15068066e+00 8.83338928e-01
-2.04715586e+00 -1.63115096e+00 1.00407451e-01 1.82598782e+00
5.52246690e-01 -8.06140900e-02 6.64867401e-01 -7.59277046e-02
6.43514931e-01 5.11371553e-01 -5.72872639e-01 -5.11935711e-01
-1.76773414e-01 -6.32401764e-01 2.65639387e-02 5.61542749e-01
-1.49213684e+00 7.26576149e-01 6.61587334e+00 6.85078502e-01
-8.96845400e-01 1.76786497e-01 1.30801475e+00 -4.85793322e-01
-7.15645552e-02 -5.48428595e-01 -3.92702222e-01 6.81535423e-01
1.09805524e+00 2.57515371e-01 4.02369797e-01 3.69847000e-01
3.93244237e-01 -5.00666499e-02 -8.82392883e-01 5.77526093e-01
5.38864098e-02 -1.07265997e+00 -1.91312343e-01 -5.93910590e-02
4.87031072e-01 1.56432673e-01 3.25673461e-01 1.63458019e-01
6.09736741e-01 -1.27138841e+00 5.70681691e-01 9.22867537e-01
4.96131778e-01 -7.61503518e-01 6.66049540e-01 3.36140603e-01
-1.35332584e+00 -2.15458035e-01 2.96873122e-01 7.01106479e-03
7.10987151e-01 2.83438951e-01 -8.24706852e-01 1.71250865e-01
9.33515131e-01 8.38772714e-01 -5.46623707e-01 5.06610274e-01
-2.75095791e-01 5.86706698e-01 -8.17760900e-02 1.45420283e-01
4.18338180e-01 1.09436683e-01 8.78838718e-01 1.28364444e+00
2.06215158e-02 3.29400718e-01 -4.72210012e-02 8.04282963e-01
2.32308045e-01 -4.82258826e-01 -1.00324428e+00 6.28925264e-02
1.02954105e-01 9.50512648e-01 -6.51222944e-01 -2.82609612e-01
-1.22706950e-01 9.71159816e-01 3.89225900e-01 6.75380766e-01
-1.26369941e+00 -1.92740560e-01 9.33820844e-01 -5.72718456e-02
-2.29882002e-02 -2.43022874e-01 5.08813672e-02 -7.49694824e-01
-1.28846630e-01 -5.18992543e-01 8.82918179e-01 -9.03600276e-01
-1.39469612e+00 1.01524723e+00 1.68608084e-01 -8.34743738e-01
-6.95220768e-01 -2.47996479e-01 -1.06265974e+00 6.26162410e-01
-1.36364758e+00 -1.09545088e+00 6.69922456e-02 7.82226741e-01
5.71415126e-01 -7.01984614e-02 4.00294274e-01 2.38747761e-01
-1.11474884e+00 3.10132444e-01 -3.40760976e-01 1.89580634e-01
5.37870347e-01 -8.88481200e-01 6.89266384e-01 1.07061458e+00
-2.42052659e-01 5.01304150e-01 5.04556954e-01 -1.21560788e+00
-8.54471385e-01 -1.56815290e+00 1.16774428e+00 -9.69476521e-01
6.42955601e-01 -4.86319929e-01 -8.40937853e-01 7.48391986e-01
3.64476562e-01 -1.54000461e-01 5.16007841e-01 -4.18709695e-01
-3.64174634e-01 4.30517972e-01 -1.39135444e+00 1.05243778e+00
1.22255850e+00 -6.11084700e-01 -4.21679795e-01 5.38859367e-01
8.77213657e-01 2.52819985e-01 -4.16299462e-01 5.64751685e-01
4.26939577e-01 -1.06326592e+00 1.41794586e+00 -8.23293388e-01
5.27942479e-01 -2.80650705e-01 2.17456389e-02 -1.16825068e+00
-7.38943875e-01 -5.56315184e-01 -2.17154622e-01 1.01590872e+00
1.64501742e-01 -7.26044118e-01 2.66145140e-01 3.18115532e-01
1.00953737e-02 -9.97679710e-01 -1.24631298e+00 -6.68456137e-01
-2.54725348e-02 -3.74756604e-01 6.09876275e-01 8.23829651e-01
-2.82762945e-01 2.97157615e-02 -7.11739540e-01 4.86709267e-01
5.39999962e-01 -2.06455231e-01 2.77065426e-01 -7.12797284e-01
-2.05426678e-01 -5.70123613e-01 -2.19046876e-01 -3.13927263e-01
2.37763286e-01 -4.53567028e-01 9.66278464e-02 -1.45123935e+00
4.25256610e-01 1.19123518e-01 -7.52433598e-01 4.92694944e-01
-5.01960039e-01 1.62975371e-01 -9.29131806e-02 7.45594949e-02
-8.13867927e-01 5.09834886e-01 6.18853569e-01 -1.04508743e-01
5.41585162e-02 -3.22322428e-01 -2.57407218e-01 8.55513394e-01
7.39255726e-01 -5.43090463e-01 -4.09677058e-01 -4.78398293e-01
3.69024724e-01 2.51189768e-01 1.00157523e+00 -7.92119205e-01
7.47670531e-02 -5.32848597e-01 3.79109532e-01 -7.38697886e-01
5.60540318e-01 -6.63433969e-01 8.74816924e-02 9.20699775e-01
-4.86709803e-01 5.76447248e-01 9.85018834e-02 9.77458715e-01
1.95983514e-01 3.31764460e-01 5.90611160e-01 9.17814821e-02
-5.63509941e-01 4.92182195e-01 -9.84197438e-01 6.10337965e-02
1.53806424e+00 3.63025963e-02 -7.56654143e-01 -6.45263851e-01
-4.44506794e-01 4.33445305e-01 4.54887867e-01 3.49500537e-01
9.99415398e-01 -1.39733565e+00 -9.00986016e-01 6.22659959e-02
1.70866475e-01 -5.27934492e-01 7.23772109e-01 9.63755369e-01
-3.35432619e-01 3.90240371e-01 -4.33472684e-03 -2.29386702e-01
-1.16486633e+00 1.21248281e+00 2.23317146e-01 -5.72698712e-01
-5.25360286e-01 1.21445560e+00 9.05583501e-01 -1.29410937e-01
4.82515544e-01 -1.08913176e-01 -5.90039909e-01 -1.00941882e-01
1.02890730e+00 6.13474190e-01 -6.11109495e-01 -1.08222187e+00
-7.59250164e-01 4.66357291e-01 7.54233301e-02 3.28757986e-02
1.27490175e+00 -3.94769795e-02 -1.47456222e-03 2.53764242e-01
1.07195139e+00 -3.21898073e-01 -1.43767631e+00 7.60605484e-02
-8.77204817e-03 -1.87122926e-01 3.41144949e-02 -1.02966118e+00
-9.90888894e-01 7.96926439e-01 6.80897892e-01 2.90608734e-01
1.34228218e+00 2.13485554e-01 1.18957710e+00 -1.14057377e-01
2.07737967e-01 -7.19577968e-01 1.82891831e-01 3.24832559e-01
1.26815903e+00 -1.19710135e+00 -4.40831393e-01 -2.52165459e-02
-9.35543478e-01 5.41787982e-01 4.16160285e-01 -1.56559497e-01
5.27595758e-01 2.75720477e-01 -1.45447487e-02 -5.87720037e-01
-1.29049027e+00 -3.12944665e-03 3.24385434e-01 7.42121577e-01
1.76801533e-01 4.22430150e-02 2.57136613e-01 3.29680294e-01
4.59589869e-01 -2.27882311e-01 1.26849696e-01 8.03758025e-01
-2.61036664e-01 -6.12564445e-01 -4.72929120e-01 9.56013277e-02
-4.51272070e-01 -1.19050816e-01 -9.33537245e-01 3.26732844e-01
2.37230062e-01 1.10707736e+00 1.95209339e-01 -4.27184194e-01
-3.62809077e-02 -2.29774401e-01 -8.92624035e-02 -1.45771578e-01
-8.79869819e-01 -7.10637718e-02 2.23884985e-01 -8.07986796e-01
-2.72366792e-01 -9.33321059e-01 -1.15122688e+00 -1.79881275e-01
4.46325630e-01 -3.26569915e-01 9.04138982e-02 8.64230871e-01
4.42292482e-01 8.10061932e-01 7.48356760e-01 -9.14900899e-01
4.71188203e-02 -6.24316871e-01 -2.41302729e-01 4.06868726e-01
7.66393304e-01 -7.43229985e-01 -3.90674889e-01 -8.26071352e-02] | [8.118729591369629, 0.5520936846733093] |
a5383262-8cfa-4ec2-bd07-b69901388d7e | a-universal-unbiased-method-for | 2306.11343 | null | https://arxiv.org/abs/2306.11343v2 | https://arxiv.org/pdf/2306.11343v2.pdf | A Universal Unbiased Method for Classification from Aggregate Observations | In conventional supervised classification, true labels are required for individual instances. However, it could be prohibitive to collect the true labels for individual instances, due to privacy concerns or unaffordable annotation costs. This motivates the study on classification from aggregate observations (CFAO), where the supervision is provided to groups of instances, instead of individual instances. CFAO is a generalized learning framework that contains various learning problems, such as multiple-instance learning and learning from label proportions. The goal of this paper is to present a novel universal method of CFAO, which holds an unbiased estimator of the classification risk for arbitrary losses -- previous research failed to achieve this goal. Practically, our method works by weighing the importance of each label for each instance in the group, which provides purified supervision for the classifier to learn. Theoretically, our proposed method not only guarantees the risk consistency due to the unbiased risk estimator but also can be compatible with arbitrary losses. Extensive experiments on various problems of CFAO demonstrate the superiority of our proposed method. | ['Heng Tao Shen', 'Xiaofeng Zhu', 'Gang Niu', 'Tongliang Liu', 'Bo Han', 'Lei Feng', 'Zixi Wei'] | 2023-06-20 | null | null | null | null | ['classification-1', 'multiple-instance-learning'] | ['methodology', 'methodology'] | [ 2.96082228e-01 5.02910674e-01 -3.15257788e-01 -7.99205899e-01
-8.12507093e-01 -1.86258614e-01 -2.63529811e-02 5.06053329e-01
-3.23825508e-01 1.14180422e+00 -3.51599902e-01 -1.04203716e-01
-2.83982933e-01 -9.09402966e-01 -5.30882895e-01 -1.05159175e+00
3.98387089e-02 3.39173704e-01 -1.69737071e-01 4.05019015e-01
6.10796623e-02 2.53610045e-01 -1.46287966e+00 9.19879749e-02
1.13267159e+00 1.35594642e+00 -2.78833926e-01 -7.54644871e-02
4.23091911e-02 7.41433382e-01 -7.02554941e-01 -8.28366458e-01
4.39204961e-01 -4.64321792e-01 -7.10603416e-01 2.67566085e-01
3.88905257e-01 -5.38663983e-01 3.17619145e-01 1.32216537e+00
3.49878699e-01 2.82699359e-04 1.03491747e+00 -1.79518259e+00
-5.60083091e-01 6.93232119e-01 -7.03532994e-01 -2.19996855e-01
-3.69181782e-02 -4.25008148e-01 1.48776639e+00 -3.79940838e-01
7.59297758e-02 9.60069537e-01 6.77109420e-01 6.52025580e-01
-1.09733629e+00 -9.05242503e-01 1.99137181e-01 -2.25840345e-01
-1.40261674e+00 -2.30220228e-01 5.04772902e-01 -2.82683730e-01
1.94005534e-01 3.84053558e-01 1.32582873e-01 6.92047834e-01
-1.08689982e-02 8.91819537e-01 1.31607389e+00 -4.74731654e-01
3.71046573e-01 6.87834084e-01 4.78608310e-01 4.87644285e-01
6.37614965e-01 4.22990769e-02 -3.97182375e-01 -5.11014700e-01
3.42861742e-01 1.83481500e-01 -2.15179220e-01 -4.86759126e-01
-7.48175859e-01 8.05955112e-01 1.85810328e-01 -1.79971784e-01
-1.95373014e-01 -6.23496510e-02 4.33056921e-01 3.71327937e-01
8.15257907e-01 -5.75208627e-02 -5.73023617e-01 4.14354175e-01
-6.62891865e-01 1.78334907e-01 7.70434856e-01 1.11501813e+00
6.56133115e-01 -3.66122156e-01 -1.83019072e-01 7.83103764e-01
3.12826812e-01 9.60570276e-02 3.14490974e-01 -7.67098606e-01
6.40270591e-01 5.34537137e-01 3.44624907e-01 -9.12585437e-01
-1.10269792e-01 -4.83582199e-01 -8.78646433e-01 2.50149965e-01
7.47549236e-01 -1.44919395e-01 -1.91650063e-01 1.92124808e+00
5.88172495e-01 6.98267668e-03 9.55832228e-02 7.03982234e-01
4.02742922e-01 3.52558583e-01 2.89685726e-01 -6.14464700e-01
1.17741823e+00 -7.50300407e-01 -9.52665925e-01 8.36164504e-02
9.61022854e-01 -2.60201722e-01 8.50178301e-01 3.83563280e-01
-7.18079507e-01 -1.30940855e-01 -8.73644054e-01 2.31759340e-01
-2.36461252e-01 3.28857481e-01 7.53765881e-01 1.03285980e+00
-3.31495732e-01 5.09885907e-01 -4.69244778e-01 -1.81525007e-01
7.44964182e-01 4.05185133e-01 -4.51679021e-01 9.58468318e-02
-1.09186530e+00 4.63098794e-01 5.37689865e-01 1.23660430e-01
-6.58879220e-01 -5.39055467e-01 -5.54100275e-01 3.15210402e-01
5.42681694e-01 -3.71561319e-01 1.13922250e+00 -1.14922869e+00
-9.15856242e-01 8.32811296e-01 -5.55694699e-02 -5.83571851e-01
8.66195738e-01 -1.22715808e-01 -4.32248525e-02 -1.85221985e-01
2.83359766e-01 2.46654645e-01 6.85596943e-01 -1.28039420e+00
-1.09414220e+00 -5.71409762e-01 1.53148890e-01 1.62235871e-01
-5.08097768e-01 -1.27299592e-01 3.32969785e-01 -5.90359986e-01
4.23259400e-02 -6.30284905e-01 -1.47288278e-01 1.26097828e-01
-4.78662103e-01 -6.35910571e-01 5.51037908e-01 -5.47289908e-01
1.19409978e+00 -2.42278528e+00 -5.31923831e-01 1.52834743e-01
5.86308241e-02 -1.93614945e-01 3.50559562e-01 9.69121084e-02
-4.20911126e-02 2.49041423e-01 -4.20405656e-01 -3.64457875e-01
1.01588488e-01 1.54538155e-02 -3.31611514e-01 6.44396782e-01
1.24727353e-01 2.92118132e-01 -9.39320445e-01 -8.36729646e-01
-1.07810713e-01 -9.83080715e-02 -4.80479956e-01 2.35001236e-01
-3.71668749e-02 1.97692201e-01 -6.75409317e-01 7.46756613e-01
8.16856861e-01 -1.99578583e-01 3.59903753e-01 -6.83649853e-02
3.77495885e-01 -1.68821126e-01 -1.30260491e+00 9.34532940e-01
-3.57501566e-01 -9.82245654e-02 -3.43706273e-02 -1.46114910e+00
1.06555724e+00 4.85070348e-01 5.23048937e-01 -9.95752066e-02
2.45572314e-01 4.77271289e-01 -3.74334872e-01 -3.95446330e-01
-6.05240203e-02 -5.15915334e-01 -2.62467891e-01 4.40799981e-01
-5.24524637e-02 3.79977822e-01 -2.10407153e-01 9.34028439e-03
6.06137395e-01 -1.31270140e-01 7.88313448e-01 -2.97835499e-01
5.79508901e-01 -3.76961589e-01 9.46502328e-01 8.22325826e-01
-4.40192252e-01 3.55216503e-01 6.05128706e-01 -2.81402647e-01
-7.48951316e-01 -1.02638602e+00 -5.42388260e-01 9.29335415e-01
2.24924073e-01 8.61700922e-02 -7.12144852e-01 -1.31477821e+00
1.90770969e-01 6.96544707e-01 -6.08347237e-01 -1.26917705e-01
5.58192432e-02 -1.18504691e+00 3.40806901e-01 4.28895712e-01
5.14328420e-01 -8.04817319e-01 -2.27574959e-01 -2.35587247e-02
-2.20520258e-01 -7.71187305e-01 -4.80817407e-01 2.14694485e-01
-9.79983568e-01 -1.08598936e+00 -4.80230391e-01 -7.39382267e-01
1.00526357e+00 8.26560110e-02 7.82306314e-01 1.09117277e-01
-4.97587547e-02 9.67752412e-02 -3.41580153e-01 -7.02649415e-01
-1.48903370e-01 4.82924469e-02 1.39191955e-01 7.19883323e-01
5.64060092e-01 -5.11882007e-01 -3.36296260e-01 4.85297650e-01
-8.42216730e-01 -4.82688934e-01 3.96005809e-01 1.07771373e+00
6.62757635e-01 3.60690713e-01 1.30522501e+00 -1.48489690e+00
6.00299835e-01 -6.84098125e-01 -7.20698714e-01 6.39829695e-01
-9.01508868e-01 -6.11665063e-02 7.40395367e-01 -1.83618039e-01
-1.26657498e+00 2.18331814e-01 3.61481160e-01 -2.73596030e-03
-2.04868883e-01 1.47567734e-01 -6.70708358e-01 1.02137275e-01
4.37286794e-01 -1.45358488e-01 5.95509298e-02 -5.47502100e-01
7.78164491e-02 1.07868445e+00 1.91472292e-01 -7.64610231e-01
4.69990313e-01 4.66280729e-01 2.96971172e-01 -3.04860234e-01
-1.47662234e+00 -3.51096451e-01 -6.17618680e-01 -4.29619513e-02
5.54429770e-01 -8.03833604e-01 -1.03645420e+00 3.30198795e-01
-7.73906767e-01 1.61471143e-01 -2.44747534e-01 5.71658731e-01
-4.72258687e-01 5.35302639e-01 -3.53913873e-01 -1.36638451e+00
-2.98858047e-01 -8.14668834e-01 9.11215365e-01 1.77917331e-01
1.58912856e-02 -1.08200502e+00 -3.49864513e-01 7.74548829e-01
-1.83823466e-01 4.22716826e-01 8.32384288e-01 -1.02051210e+00
-4.41107661e-01 -5.94507873e-01 -2.16266185e-01 7.60031044e-01
3.76238912e-01 -2.93396562e-01 -1.09571314e+00 -4.78195250e-01
2.25181803e-01 -7.36560524e-01 6.22203052e-01 2.55852014e-01
1.70670247e+00 -5.44578969e-01 -2.24976659e-01 3.10235500e-01
1.55176663e+00 1.57416373e-01 3.16901356e-01 2.91140944e-01
2.65296161e-01 9.23650324e-01 1.16395152e+00 6.28092229e-01
1.40231758e-01 4.48194504e-01 5.01513183e-01 1.98334590e-01
4.77572441e-01 -1.89778373e-01 1.29260525e-01 5.58107615e-01
2.52227075e-02 -2.64345795e-01 -5.38904727e-01 5.32482326e-01
-1.96706200e+00 -7.79792547e-01 -1.43107414e-01 2.63725901e+00
8.64052594e-01 -6.38370365e-02 4.81054373e-02 2.99629778e-01
1.02622104e+00 -1.56829283e-01 -4.79964852e-01 -2.69689798e-01
1.25082046e-01 -1.62428647e-01 6.39473200e-01 4.12451178e-01
-1.15860915e+00 2.08972096e-01 6.50300360e+00 1.16930079e+00
-5.96455336e-01 2.83062756e-01 1.12493396e+00 -1.94137022e-01
-1.86997820e-02 2.42380518e-02 -1.01603985e+00 6.42972648e-01
5.55015504e-01 -3.04101765e-01 1.26479298e-01 1.14906394e+00
-4.49399278e-03 -1.09999672e-01 -1.26786590e+00 7.17667460e-01
-8.92185420e-02 -5.73234439e-01 -2.86052246e-02 2.12676093e-01
7.44704604e-01 -7.63396919e-01 3.25794443e-02 1.94564983e-01
2.07416043e-01 -9.66196895e-01 5.94562531e-01 3.64185423e-01
7.47105539e-01 -1.09374082e+00 1.18815398e+00 7.31886864e-01
-8.08137894e-01 -3.39687794e-01 -6.30868196e-01 -1.25469163e-01
-6.68551475e-02 1.02722538e+00 -6.82337463e-01 8.24459076e-01
5.39513528e-01 5.54775000e-01 -3.39647263e-01 9.03413951e-01
-1.85227189e-02 6.12869382e-01 -3.16510290e-01 -5.88026084e-02
-2.21198555e-02 -3.08534414e-01 -2.30354983e-02 6.85207486e-01
3.75130773e-01 6.11199103e-02 3.66003662e-01 5.57452083e-01
-3.44129413e-01 5.31678379e-01 -5.12104630e-01 3.22781801e-01
6.72120333e-01 1.15760350e+00 -4.15064484e-01 -1.81940898e-01
-5.60800135e-01 6.32440627e-01 6.63370550e-01 6.90004453e-02
-8.27103615e-01 -4.84161526e-01 1.86707869e-01 -3.12793627e-02
-6.68223724e-02 5.78337312e-01 -4.41690505e-01 -1.07756817e+00
2.71051705e-01 -6.67802691e-01 8.30435038e-01 -2.36988515e-01
-1.85948527e+00 6.40042797e-02 1.57612488e-01 -1.38512516e+00
2.00485021e-01 -5.11758506e-01 -5.49332500e-01 7.64759362e-01
-1.25563180e+00 -1.06007957e+00 -2.05710813e-01 4.65818554e-01
7.48754591e-02 -1.18404813e-01 6.46355510e-01 4.52051729e-01
-7.03792572e-01 1.09604049e+00 3.91720623e-01 1.34783089e-01
8.35864365e-01 -1.28160179e+00 -5.38368046e-01 4.58241105e-01
-1.31658107e-01 3.40902716e-01 5.50466418e-01 -4.80386257e-01
-4.48680490e-01 -1.21995318e+00 1.12304103e+00 -4.56100523e-01
4.43957686e-01 -2.22361058e-01 -8.63678277e-01 8.85889769e-01
-3.44490409e-01 3.85213882e-01 1.07663143e+00 2.98173845e-01
-3.12895328e-01 -7.06555367e-01 -1.67126548e+00 1.85903728e-01
8.79821062e-01 -3.46987337e-01 -5.12823045e-01 5.62161744e-01
5.43657839e-01 5.32731451e-02 -1.04224586e+00 3.73823971e-01
5.45701563e-01 -9.20946598e-01 6.83583975e-01 -7.59755731e-01
3.01932096e-01 -2.08426282e-01 -2.18653530e-01 -1.03023732e+00
3.77205573e-02 -6.50497377e-02 3.57910097e-02 1.81584132e+00
4.10313606e-01 -9.79847610e-01 7.02526987e-01 8.94604445e-01
3.43398392e-01 -7.80765891e-01 -1.11999691e+00 -1.00772595e+00
8.77650157e-02 -1.05521806e-01 8.90843034e-01 1.17421353e+00
1.02776233e-02 2.11494882e-02 -6.80515289e-01 4.74164277e-01
1.17714977e+00 1.34021133e-01 4.87855524e-01 -1.45017409e+00
-3.36277813e-01 1.07785858e-01 -4.20799851e-01 -5.27815461e-01
2.45174944e-01 -9.68011677e-01 7.04725534e-02 -1.01158118e+00
5.54226339e-01 -8.61357987e-01 -6.50447726e-01 5.72861612e-01
-3.08157116e-01 1.83888227e-02 -1.36674363e-02 2.77313054e-01
-6.00960970e-01 4.87061650e-01 1.03236735e+00 -4.87445965e-02
1.76694155e-01 4.18986857e-01 -8.78140092e-01 6.56944156e-01
8.95583987e-01 -8.53368402e-01 -6.20998502e-01 -5.55459969e-02
-1.86717156e-02 1.25558421e-01 2.89704829e-01 -7.58670211e-01
-8.98325071e-02 -2.69367009e-01 1.17284236e-02 -3.33826840e-01
-9.55697000e-02 -1.14102685e+00 1.91139147e-01 3.65412682e-01
-6.75337374e-01 -6.70342326e-01 -5.51117957e-01 9.61062372e-01
-2.81327546e-01 -8.50579679e-01 8.62511098e-01 -1.53905869e-01
-8.75218511e-02 4.25671190e-01 1.89700693e-01 5.02698496e-02
1.41053307e+00 5.56707792e-02 -1.79872036e-01 -2.09970742e-01
-7.93346703e-01 3.32331389e-01 2.22088158e-01 -1.35819972e-01
1.08091712e-01 -1.53842592e+00 -6.08016908e-01 -1.80227444e-01
2.85807937e-01 1.32777989e-01 1.54614091e-01 6.58265710e-01
-1.37418836e-01 3.07352245e-01 4.83898483e-02 -3.14943403e-01
-1.33294606e+00 8.25486481e-01 3.10327977e-01 -3.62683773e-01
-2.47366428e-01 7.02302635e-01 5.80353618e-01 -5.71397007e-01
4.98375744e-01 -6.06663227e-02 -1.40250623e-01 2.10252896e-01
4.69054699e-01 4.15597588e-01 -8.18881914e-02 -2.96037138e-01
-1.86068088e-01 2.21915126e-01 -1.69551834e-01 3.29732329e-01
1.07163703e+00 -3.10199529e-01 -3.20042193e-01 6.02737963e-01
1.25896597e+00 -7.79912099e-02 -1.06174886e+00 -2.02788502e-01
3.47216683e-03 -7.79903829e-01 -2.30389133e-01 -6.62483037e-01
-1.11025453e+00 8.77827168e-01 5.41629493e-01 4.66262311e-01
1.09797812e+00 -1.72549516e-01 4.08427000e-01 3.14457536e-01
8.91452789e-01 -1.06371999e+00 -3.06115806e-01 -1.75878406e-01
3.74558002e-01 -1.50206578e+00 8.18788558e-02 -6.94742560e-01
-5.67207873e-01 8.04012597e-01 7.78925478e-01 4.02721018e-02
7.75952339e-01 -1.68263569e-01 -5.22312969e-02 1.52912542e-01
-6.21207774e-01 3.53958100e-01 1.03094965e-01 5.80202878e-01
4.38135743e-01 3.46907914e-01 -6.79651380e-01 1.06257355e+00
1.48554027e-01 6.10203855e-02 3.69087815e-01 7.66969979e-01
-3.48676980e-01 -1.10440755e+00 -4.21013325e-01 9.17327642e-01
-8.31882477e-01 1.99964032e-01 -1.42092675e-01 5.33485174e-01
3.86915773e-01 9.61863041e-01 -1.43069103e-01 3.73589285e-02
1.29474998e-01 2.43176594e-01 1.23633370e-01 -4.97584581e-01
-3.43850881e-01 -2.64652461e-01 6.97800294e-02 -2.38721490e-01
-6.10834479e-01 -8.72053206e-01 -9.26478803e-01 -2.32276037e-01
-7.76571155e-01 4.52290088e-01 1.14776038e-01 9.19138372e-01
-1.57006651e-01 1.00488693e-01 1.10173213e+00 -1.06072105e-01
-1.14275622e+00 -7.82098591e-01 -1.33089375e+00 5.78404307e-01
2.44693995e-01 -7.41368711e-01 -8.98525953e-01 -5.50423982e-03] | [9.096138000488281, 4.1873884201049805] |
4e79c89c-42a3-429d-9497-77e61069427f | vtlayout-fusion-of-visual-and-text-features | 2108.13297 | null | https://arxiv.org/abs/2108.13297v1 | https://arxiv.org/pdf/2108.13297v1.pdf | VTLayout: Fusion of Visual and Text Features for Document Layout Analysis | Documents often contain complex physical structures, which make the Document Layout Analysis (DLA) task challenging. As a pre-processing step for content extraction, DLA has the potential to capture rich information in historical or scientific documents on a large scale. Although many deep-learning-based methods from computer vision have already achieved excellent performance in detecting \emph{Figure} from documents, they are still unsatisfactory in recognizing the \emph{List}, \emph{Table}, \emph{Text} and \emph{Title} category blocks in DLA. This paper proposes a VTLayout model fusing the documents' deep visual, shallow visual, and text features to localize and identify different category blocks. The model mainly includes two stages, and the three feature extractors are built in the second stage. In the first stage, the Cascade Mask R-CNN model is applied directly to localize all category blocks of the documents. In the second stage, the deep visual, shallow visual, and text features are extracted for fusion to identify the category blocks of documents. As a result, we strengthen the classification power of different category blocks based on the existing localization technique. The experimental results show that the identification capability of the VTLayout is superior to the most advanced method of DLA based on the PubLayNet dataset, and the F1 score is as high as 0.9599. | ['Qing Wang', 'Lin Shi', 'Jun Hu', 'Shuaiqun Pan', 'Xuyan Ma', 'Shoubin Li'] | 2021-08-12 | null | null | null | null | ['document-layout-analysis'] | ['computer-vision'] | [-2.32558772e-01 -4.64230955e-01 3.28783467e-02 -5.34052588e-02
-4.95822459e-01 -5.42631686e-01 8.54846001e-01 2.93642670e-01
-7.23583251e-02 1.91542298e-01 5.33145927e-02 -4.65738505e-01
-2.26499364e-01 -8.39281440e-01 -4.22964364e-01 -9.04017568e-01
2.44012758e-01 7.01101720e-02 2.69557506e-01 2.11447570e-02
5.94463527e-01 8.21291149e-01 -1.41716003e+00 7.77902186e-01
7.60196805e-01 1.42921948e+00 5.59076548e-01 5.62448680e-01
-7.95213640e-01 9.07553017e-01 -9.32753325e-01 1.85186546e-02
-2.46406779e-01 -1.79389253e-01 -4.05475020e-01 2.06115961e-01
3.35310102e-01 -5.53977668e-01 -4.36454892e-01 9.97594237e-01
4.87977386e-01 -1.45105794e-01 7.16588795e-01 -9.52937543e-01
-8.57388437e-01 6.02520704e-01 -1.00213921e+00 3.64300668e-01
2.56988406e-01 -1.11471057e-01 1.02442992e+00 -1.40966928e+00
5.15361428e-01 1.21351039e+00 5.52529931e-01 -5.17459004e-04
-5.68302572e-01 -5.81061065e-01 2.93303847e-01 4.07740951e-01
-1.54665589e+00 -2.56913364e-01 1.03982353e+00 -5.87647676e-01
9.17999387e-01 2.29552492e-01 4.49746221e-01 9.41781640e-01
4.07124788e-01 1.30258179e+00 8.61966670e-01 -5.09929657e-01
-6.31493106e-02 1.81700870e-01 5.89878440e-01 8.78253818e-01
1.96681663e-01 -5.10188580e-01 -3.28728974e-01 3.54553014e-01
6.70747459e-01 2.27392346e-01 -2.14628816e-01 -4.97465581e-02
-9.47344542e-01 6.67988777e-01 5.88313162e-01 7.66062438e-01
-2.85024852e-01 -2.53431171e-01 3.90299946e-01 -2.76181340e-01
1.92129076e-01 2.45334297e-01 -2.06674054e-01 1.86378941e-01
-1.07697487e+00 1.12530231e-01 3.51618916e-01 9.64146852e-01
5.34048378e-01 1.05505511e-01 -5.90689003e-01 1.00254500e+00
3.67484629e-01 5.45302391e-01 5.25600433e-01 -1.51820391e-01
6.84399307e-01 1.07373524e+00 -2.43029356e-01 -1.43413603e+00
-5.41939318e-01 -6.54374182e-01 -1.09296024e+00 -4.09133285e-02
1.40117019e-01 1.74166366e-01 -1.19260275e+00 7.74468005e-01
-1.47434641e-02 -6.09608352e-01 -1.73079625e-01 7.02624142e-01
1.39803803e+00 1.14991891e+00 -2.31956258e-01 -4.12180945e-02
1.57130945e+00 -1.09408891e+00 -9.52650428e-01 -1.90011382e-01
2.70216763e-01 -1.04278970e+00 9.44991410e-01 5.82704663e-01
-8.34223866e-01 -1.02938938e+00 -1.33500254e+00 -3.84318382e-01
-7.78014123e-01 9.26375270e-01 4.94621336e-01 3.65738720e-01
-6.84827745e-01 1.55050591e-01 -3.38165283e-01 -2.74350733e-01
7.44350970e-01 1.59683943e-01 -1.83934569e-01 -1.90897107e-01
-7.78668761e-01 4.23995167e-01 3.69128078e-01 6.44902229e-01
-7.05187678e-01 -3.28172833e-01 -7.65505493e-01 4.06355381e-01
3.87954473e-01 -1.85820665e-02 9.00162041e-01 -4.39550638e-01
-9.91562426e-01 5.28003275e-01 -5.95610738e-02 6.91540167e-02
2.59413481e-01 -1.53310910e-01 -6.60447359e-01 2.54501224e-01
1.66065320e-01 3.55308473e-01 8.73759031e-01 -1.21568906e+00
-9.53856766e-01 -4.67343807e-01 -9.92929935e-02 1.84552316e-02
-6.96388543e-01 -9.48486291e-03 -9.82540548e-01 -6.60159945e-01
2.89616823e-01 -2.70480931e-01 4.57479089e-01 -9.36200842e-02
-7.64378667e-01 -5.12783706e-01 1.35583317e+00 -8.97336602e-01
1.42547429e+00 -2.22314978e+00 -1.11337565e-01 4.48596478e-01
3.21755975e-01 4.01688457e-01 -7.29425848e-02 3.80205899e-01
1.28629446e-01 1.35809571e-01 3.01547021e-01 -3.69920135e-02
-1.83266234e-02 -3.37698132e-01 -4.09791291e-01 3.21728170e-01
3.06335110e-02 8.63591075e-01 -4.79149908e-01 -7.13115275e-01
4.27596569e-01 3.64561886e-01 4.64795120e-02 8.18604529e-02
-8.11943635e-02 -3.59158516e-02 -6.62750840e-01 9.29646969e-01
8.76844108e-01 -3.70779425e-01 6.06672931e-03 -5.88181734e-01
-4.03823078e-01 4.54225466e-02 -1.12253511e+00 1.30761969e+00
-2.02985555e-01 1.03922069e+00 1.54619619e-01 -9.10582423e-01
1.14900255e+00 6.89341426e-02 3.67526859e-01 -9.61540520e-01
4.65737373e-01 8.67282152e-02 -8.38453975e-03 -6.99323952e-01
6.26382947e-01 4.55576748e-01 -1.17998913e-01 9.58566144e-02
-1.40369525e-02 1.42112076e-01 2.72456199e-01 3.47013116e-01
9.24655139e-01 -6.19707517e-02 1.35183223e-02 -1.79825261e-01
9.12333965e-01 -4.58992720e-02 2.67427564e-01 7.52475977e-01
1.50919229e-01 5.19029021e-01 5.54514647e-01 -3.51647735e-01
-1.10833418e+00 -8.37521553e-01 -3.04317385e-01 7.12823570e-01
2.25563288e-01 -4.82066453e-01 -6.23639464e-01 -6.50817454e-01
2.89128236e-02 3.21365684e-01 -4.92924541e-01 -9.58299339e-02
-5.33611000e-01 -5.32902062e-01 5.69603860e-01 7.58247912e-01
9.31880832e-01 -1.18495333e+00 -1.43478692e-01 8.15717876e-03
1.51364990e-02 -1.08087242e+00 -2.87472486e-01 9.35655013e-02
-3.65403205e-01 -9.56556380e-01 -7.81940758e-01 -1.09140229e+00
7.21170604e-01 4.26137894e-01 4.28684235e-01 2.15849817e-01
-4.49149281e-01 -3.89911532e-02 -3.64739984e-01 -5.11636376e-01
2.08173343e-03 3.50894108e-02 -3.20404649e-01 1.14876434e-01
4.27988321e-01 5.21639399e-02 -4.60079312e-01 -6.01106361e-02
-7.06372857e-01 1.26352251e-01 9.28656220e-01 6.47058189e-01
4.74762172e-01 6.87353611e-01 2.18919322e-01 -6.68080688e-01
8.22388113e-01 -1.66743085e-01 -5.49629748e-01 2.97351629e-01
-5.41806221e-01 -3.90361279e-01 1.08011949e+00 -3.37464631e-01
-1.12748277e+00 -2.21531630e-01 -1.25532910e-01 -3.12735409e-01
-2.50740618e-01 7.51387179e-01 -6.65095747e-01 2.28862539e-01
2.08498657e-01 5.14275253e-01 -4.46732461e-01 -7.98870742e-01
1.60505518e-01 1.03229547e+00 6.53324902e-01 -2.09929317e-01
8.58608544e-01 4.27706569e-01 -1.05176166e-01 -1.09232378e+00
-7.86120594e-01 -5.66426277e-01 -7.39552438e-01 -4.64110881e-01
9.02849972e-01 -6.96120739e-01 -7.97118843e-01 6.99626625e-01
-1.26226699e+00 2.01020807e-01 3.26595567e-02 2.87192106e-01
1.12767644e-01 6.05227590e-01 -6.32463157e-01 -5.87640166e-01
-5.33491611e-01 -1.14229369e+00 1.18582344e+00 6.66944385e-01
-6.89121569e-03 -6.58461034e-01 -5.44347584e-01 4.25900102e-01
6.02717474e-02 3.63547243e-02 1.17745101e+00 -5.55560112e-01
-6.33758664e-01 -5.68305314e-01 -7.25177228e-01 2.44204551e-01
1.83299989e-01 4.84965295e-01 -1.03510833e+00 -2.35084862e-01
-2.30434656e-01 4.07104865e-02 1.02129996e+00 4.35595095e-01
1.55083370e+00 -1.36284605e-01 -5.89959145e-01 4.70376253e-01
1.26800132e+00 7.05149949e-01 5.93014061e-01 5.04234791e-01
1.18908286e+00 3.40920806e-01 5.86475432e-01 3.77471477e-01
4.34773862e-02 3.29036057e-01 2.78633386e-01 -3.35342854e-01
-2.99386710e-01 -2.16783434e-01 2.21586660e-01 9.61216390e-01
2.15369150e-01 -5.61968982e-01 -1.11741424e+00 6.10544741e-01
-1.60456622e+00 -9.69704568e-01 -4.55917299e-01 1.47483397e+00
4.57488358e-01 4.41611141e-01 -3.25716823e-01 5.37541807e-01
7.91758060e-01 4.45697725e-01 -3.28287601e-01 -4.08351779e-01
-3.21597099e-01 2.12610792e-03 2.63306320e-01 -5.47154434e-02
-1.30043340e+00 7.51310110e-01 5.27332830e+00 1.19452453e+00
-1.05021477e+00 -3.64085168e-01 3.83097798e-01 1.79526597e-01
-1.03555664e-01 -2.51717180e-01 -1.09193420e+00 6.14222646e-01
4.23881680e-01 3.05443197e-01 7.56662386e-03 9.11273241e-01
1.20207980e-01 -1.48001105e-01 -8.97667706e-01 9.54307795e-01
3.86709958e-01 -1.58197415e+00 2.57152200e-01 2.24313959e-01
6.16013110e-01 -5.56569397e-01 2.12400854e-01 2.90888816e-01
-1.30443275e-01 -9.85185802e-01 9.08840179e-01 6.05354846e-01
7.08139002e-01 -9.13838148e-01 9.04036164e-01 4.01374161e-01
-1.48020780e+00 -4.38165814e-01 -4.83982950e-01 1.87990353e-01
-1.52487069e-01 8.38510990e-01 -7.39109993e-01 6.85410440e-01
9.13512886e-01 9.00902092e-01 -9.02221382e-01 9.89209175e-01
-3.22412372e-01 3.74577314e-01 8.46657380e-02 -4.69569951e-01
5.48361897e-01 -1.40724018e-01 2.02228099e-01 1.35845113e+00
3.81135941e-01 -1.29358530e-01 2.82859113e-02 8.43376279e-01
-3.04012656e-01 2.80265659e-01 -4.40431744e-01 -2.75259793e-01
3.88117403e-01 1.59146714e+00 -1.11796224e+00 -4.47959155e-01
-4.68066722e-01 6.76580787e-01 5.51452711e-02 2.65590101e-01
-6.94037020e-01 -1.07816899e+00 -1.22349940e-01 -8.58540609e-02
7.57409513e-01 -1.70856580e-01 -4.87371117e-01 -8.61066997e-01
3.97317231e-01 -8.71913016e-01 1.78877339e-01 -1.13218534e+00
-1.02358091e+00 3.82143021e-01 -3.61926913e-01 -1.15166664e+00
2.24677086e-01 -9.87683296e-01 -7.13078320e-01 9.19746816e-01
-1.08249986e+00 -1.45939827e+00 -6.50880933e-01 4.51689750e-01
7.73224771e-01 -5.50280333e-01 4.46118325e-01 3.29546005e-01
-1.01357591e+00 3.91952455e-01 5.50463676e-01 6.15657449e-01
4.10317898e-01 -1.12639678e+00 9.32596922e-02 1.03480220e+00
1.81522474e-01 6.76789224e-01 1.57561705e-01 -6.69677436e-01
-1.40254819e+00 -9.97467816e-01 7.22739100e-01 -2.10913479e-01
4.24132019e-01 -7.08292186e-01 -9.59771514e-01 3.93138021e-01
2.62758106e-01 -3.95524621e-01 2.36897469e-01 -1.93620156e-02
-2.18850628e-01 -3.51480961e-01 -7.15959251e-01 4.73146975e-01
4.30134475e-01 -6.81272328e-01 -5.70747316e-01 3.13029915e-01
3.66973788e-01 -2.86827087e-01 -6.87711418e-01 2.96007514e-01
5.95399320e-01 -7.12556958e-01 1.09534776e+00 -9.44207907e-02
6.98619843e-01 -4.75190580e-01 7.15819374e-02 -6.60233080e-01
-6.45255864e-01 1.05638981e-01 -3.21308523e-01 1.83890557e+00
2.36812428e-01 -1.28793921e-02 5.17829359e-01 -1.93164170e-01
-2.84873784e-01 -6.77670896e-01 -5.43361366e-01 -5.56250334e-01
-1.79806083e-01 -2.62523919e-01 6.45317495e-01 7.97097564e-01
-2.60080814e-01 6.51964664e-01 -1.90642357e-01 8.38198289e-02
3.16029727e-01 4.71399367e-01 5.97153783e-01 -1.32552803e+00
1.86369300e-01 -7.89988160e-01 -1.09750859e-01 -1.05779052e+00
-7.91349858e-02 -8.43354881e-01 2.82003712e-02 -2.08632326e+00
3.59863549e-01 -1.06192701e-01 -3.73711169e-01 4.84840184e-01
1.87385865e-02 -4.77049127e-02 5.79055361e-02 3.71384084e-01
-4.64850932e-01 4.14590597e-01 1.35120237e+00 -7.94594884e-01
-1.22477207e-02 -2.02815920e-01 -6.07780576e-01 8.27559829e-01
5.91617167e-01 3.94654125e-02 -2.34373853e-01 -3.95572126e-01
1.74730010e-02 -2.82814145e-01 2.19796881e-01 -9.61363792e-01
4.53865856e-01 1.30535364e-01 1.31435812e+00 -1.60262072e+00
1.71140030e-01 -7.19412148e-01 -5.81224203e-01 3.24306011e-01
-1.13807879e-01 -1.07900426e-01 3.60230237e-01 3.24110240e-01
-2.18728498e-01 -3.93477082e-01 5.41992068e-01 9.58401710e-02
-1.01894939e+00 9.86764207e-02 -6.66345775e-01 -4.67578948e-01
8.53575408e-01 -3.28881055e-01 -7.33488679e-01 -6.53087050e-02
-3.02268088e-01 2.16634065e-01 1.22406922e-01 6.54403627e-01
9.86080468e-01 -1.25920725e+00 -3.43161285e-01 3.65858316e-01
1.83967292e-01 1.29994243e-01 4.37751621e-01 5.67368269e-01
-7.76605785e-01 8.31299901e-01 -2.84218550e-01 -6.19072914e-01
-1.32719243e+00 8.38279188e-01 -8.94207954e-02 -2.99896717e-01
-6.33992553e-01 7.91036129e-01 2.23053679e-01 6.03535995e-02
5.06769717e-01 -2.36858591e-01 -7.63229311e-01 2.94147551e-01
7.01381207e-01 3.74970347e-01 1.44606218e-01 -6.86596394e-01
-5.42390287e-01 7.60301054e-01 -5.51364303e-01 2.13220760e-01
1.29767108e+00 -1.08397752e-02 -2.60776758e-01 3.17965984e-01
1.30787301e+00 2.20652416e-01 -8.03135812e-01 -1.53844699e-01
-7.16967136e-02 -1.87322423e-01 3.14455181e-01 -7.92278826e-01
-1.07569504e+00 1.36307085e+00 6.05049193e-01 4.75111663e-01
1.21157587e+00 -9.15131718e-02 6.87757969e-01 3.47325414e-01
-1.72261104e-01 -1.39296162e+00 3.60810906e-01 4.52344775e-01
9.79011297e-01 -8.85161340e-01 4.90151882e-01 -1.91785142e-01
-2.95230955e-01 1.35076821e+00 7.99473524e-01 4.62356992e-02
5.61236203e-01 3.46988231e-01 -1.30852699e-01 -4.57310766e-01
-2.41643682e-01 -5.68237863e-02 6.74006104e-01 4.28949386e-01
3.35012645e-01 -1.91970423e-01 -1.80414721e-01 8.54752064e-01
-8.00851509e-02 -4.64390576e-01 4.21359420e-01 1.03128028e+00
-5.99241316e-01 -7.91698456e-01 -6.26162529e-01 6.32147133e-01
-4.82409716e-01 -2.13253126e-02 -6.36478424e-01 9.03595269e-01
3.66390496e-01 9.91554022e-01 2.95785964e-01 -5.22110522e-01
2.61207849e-01 -1.72655791e-01 2.18642101e-01 -3.82018685e-01
-5.07967174e-01 5.78756571e-01 -1.89034656e-01 -2.72681087e-01
-1.19021155e-01 -5.03136218e-01 -1.46432817e+00 -2.66767442e-01
-4.04699147e-01 -1.89301111e-02 6.80774033e-01 1.00118387e+00
5.59156835e-02 1.08448029e+00 6.95924401e-01 -6.28143549e-01
-1.05585316e-02 -1.08268094e+00 -6.84053957e-01 1.67532459e-01
3.83976668e-01 -5.39589345e-01 -8.52321908e-02 5.41141219e-02] | [11.745874404907227, 2.4149482250213623] |
1fdaeb3e-70dd-45f3-a54e-744d248a1cbe | detecting-everything-in-the-open-world | 2303.11749 | null | https://arxiv.org/abs/2303.11749v2 | https://arxiv.org/pdf/2303.11749v2.pdf | Detecting Everything in the Open World: Towards Universal Object Detection | In this paper, we formally address universal object detection, which aims to detect every scene and predict every category. The dependence on human annotations, the limited visual information, and the novel categories in the open world severely restrict the universality of traditional detectors. We propose UniDetector, a universal object detector that has the ability to recognize enormous categories in the open world. The critical points for the universality of UniDetector are: 1) it leverages images of multiple sources and heterogeneous label spaces for training through the alignment of image and text spaces, which guarantees sufficient information for universal representations. 2) it generalizes to the open world easily while keeping the balance between seen and unseen classes, thanks to abundant information from both vision and language modalities. 3) it further promotes the generalization ability to novel categories through our proposed decoupling training manner and probability calibration. These contributions allow UniDetector to detect over 7k categories, the largest measurable category size so far, with only about 500 classes participating in training. Our UniDetector behaves the strong zero-shot generalization ability on large-vocabulary datasets like LVIS, ImageNetBoxes, and VisualGenome - it surpasses the traditional supervised baselines by more than 4\% on average without seeing any corresponding images. On 13 public detection datasets with various scenes, UniDetector also achieves state-of-the-art performance with only a 3\% amount of training data. | ['Shengjin Wang', 'Hengshuang Zhao', 'Antonio Torralba', 'Ser-Nam Lim', 'Xi Chen', 'YaLi Li', 'Zhenyu Wang'] | 2023-03-21 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Detecting_Everything_in_the_Open_World_Towards_Universal_Object_Detection_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Detecting_Everything_in_the_Open_World_Towards_Universal_Object_Detection_CVPR_2023_paper.pdf | cvpr-2023-1 | ['open-world-object-detection'] | ['computer-vision'] | [-3.89308408e-02 2.72382461e-02 -2.71693230e-01 -1.40832826e-01
-7.67171144e-01 -8.01581144e-01 8.27011287e-01 8.82954821e-02
-4.32651669e-01 3.51273805e-01 -4.32982855e-02 -1.57465741e-01
2.79747277e-01 -6.38754308e-01 -7.38547027e-01 -6.27198875e-01
1.37653068e-01 3.54012519e-01 8.52048516e-01 -1.24886245e-01
-1.22696027e-01 2.12342888e-01 -1.91508615e+00 2.10322946e-01
5.70528865e-01 1.18306231e+00 4.23669070e-01 7.16771662e-01
5.77407479e-02 5.85765660e-01 -1.11630842e-01 -4.41002250e-01
4.60407734e-01 9.16519985e-02 -6.51394248e-01 2.07442969e-01
1.13377953e+00 -5.33378005e-01 -6.66333497e-01 1.09106445e+00
4.87801254e-01 -5.22827357e-02 9.49497640e-01 -1.17103136e+00
-1.23815036e+00 4.00628775e-01 -6.42190695e-01 4.56274211e-01
-1.02230310e-02 5.42819560e-01 1.45371640e+00 -1.25332713e+00
6.38058364e-01 1.22998703e+00 6.04708910e-01 6.97620749e-01
-1.36049986e+00 -8.53741884e-01 1.52075887e-01 1.58618063e-01
-1.70425797e+00 -4.03306514e-01 1.50164545e-01 -7.56315887e-01
1.00725770e+00 1.40719220e-01 2.03963384e-01 1.28266501e+00
-9.03069228e-02 8.78146470e-01 9.16921198e-01 -3.50462109e-01
4.39084955e-02 4.97916579e-01 4.67167944e-01 6.54892564e-01
3.85164648e-01 1.57617971e-01 -3.28067511e-01 1.76876247e-01
6.45376563e-01 3.14500868e-01 -7.88753852e-02 -7.32536256e-01
-1.39270377e+00 8.06860268e-01 7.75123477e-01 2.09745355e-02
2.76167542e-01 5.32728359e-02 5.90881348e-01 1.01784505e-01
3.65189403e-01 3.94561112e-01 -4.19630557e-01 4.19056177e-01
-4.85780299e-01 4.36274655e-04 5.74660599e-01 1.39058316e+00
8.10813963e-01 7.81381037e-03 -3.11656237e-01 7.93669939e-01
1.66723013e-01 9.06553805e-01 5.81046999e-01 -5.21989822e-01
4.85143006e-01 7.10358799e-01 5.58883278e-03 -7.82633722e-01
-2.99005538e-01 -6.54740334e-01 -8.16164672e-01 1.70556698e-02
2.35217512e-01 1.39458358e-01 -1.14419508e+00 1.72067511e+00
3.19881350e-01 1.11218423e-01 7.01381043e-02 7.56896913e-01
1.19880104e+00 4.82801080e-01 1.28388941e-01 2.27441207e-01
1.53846347e+00 -9.38686371e-01 -1.77294955e-01 -5.44459403e-01
4.52696055e-01 -5.73146701e-01 1.26905441e+00 1.06827036e-01
-4.40790981e-01 -7.61516392e-01 -1.13851452e+00 -1.56750828e-01
-6.05544806e-01 1.39665514e-01 6.34061217e-01 5.55154443e-01
-8.75382841e-01 8.40777829e-02 -4.82238203e-01 -8.25506270e-01
5.62908471e-01 7.75903761e-02 -5.35550416e-01 -2.34903306e-01
-7.36793518e-01 8.40013325e-01 7.52309561e-01 -5.72050333e-01
-1.28147888e+00 -4.70320165e-01 -9.04396534e-01 2.07855865e-01
5.50172389e-01 -5.52546918e-01 1.11965156e+00 -6.12758040e-01
-8.70880961e-01 1.21645367e+00 1.18955128e-01 -5.61257303e-01
4.47367460e-01 -1.18327871e-01 -5.40378034e-01 6.73777536e-02
4.73919511e-01 8.90528917e-01 8.02363098e-01 -1.21884394e+00
-9.75109160e-01 -3.64753515e-01 7.35776275e-02 2.38799825e-02
-7.00614154e-01 -1.04398526e-01 -7.43057132e-01 -6.54324591e-01
2.83549279e-01 -8.88266385e-01 8.26415315e-04 1.01535894e-01
-5.54530799e-01 -5.52423477e-01 8.91685426e-01 -2.34715283e-01
7.87241280e-01 -2.46116471e+00 -6.34166747e-02 -3.12531888e-01
6.57404304e-01 1.07848652e-01 -1.92755595e-01 4.44515906e-02
1.23299159e-01 -1.00064777e-01 8.33035782e-02 -3.22097570e-01
1.38707533e-01 3.71074259e-01 -7.52569735e-01 5.16672730e-01
1.22657746e-01 1.07290614e+00 -9.87278402e-01 -5.39074123e-01
3.08013916e-01 3.68194990e-02 -5.17798722e-01 1.56990528e-01
-4.19540145e-02 7.78873218e-03 -9.25700366e-02 7.40955830e-01
5.55957973e-01 -6.37541354e-01 -4.95407693e-02 -7.26400688e-02
6.87030554e-02 -1.37285724e-01 -1.35136795e+00 1.40468860e+00
-1.83107734e-01 6.37579381e-01 -2.51791447e-01 -8.79431307e-01
8.20448935e-01 9.42735374e-02 -1.71185527e-02 -5.25258958e-01
1.24469094e-01 2.50936002e-01 -1.53903082e-01 -2.79578656e-01
3.34854484e-01 -2.66942475e-02 -3.18749547e-01 3.00188482e-01
7.98713207e-01 3.80536728e-02 3.42222989e-01 5.89513958e-01
8.58899891e-01 -2.88372278e-01 4.47926253e-01 -3.75453889e-01
1.74854457e-01 -1.82302088e-01 6.39847934e-01 1.38476539e+00
-4.08652753e-01 5.77074289e-01 9.92941633e-02 -4.86345172e-01
-1.00043631e+00 -1.50481761e+00 -5.83704114e-01 1.58401394e+00
6.04779541e-01 -3.69626790e-01 -2.25546494e-01 -7.81068385e-01
3.51804465e-01 4.74913925e-01 -7.06417561e-01 -2.66525149e-01
1.96444944e-01 -5.42926192e-01 5.26238441e-01 7.34856129e-01
4.03551787e-01 -7.99175739e-01 -4.44826186e-01 -1.33133128e-01
-7.47687817e-02 -1.48260617e+00 -3.70225698e-01 4.21951145e-01
-5.51237404e-01 -1.28869867e+00 -5.66743731e-01 -8.22641075e-01
3.57465953e-01 8.68758917e-01 1.17531013e+00 -2.73166746e-01
-7.07712471e-01 5.43832242e-01 -2.40146354e-01 -5.24890840e-01
-3.42450142e-01 2.48517492e-03 4.90686476e-01 1.17300697e-01
6.23435259e-01 -3.82453561e-01 -5.09338498e-01 4.45737064e-01
-5.59344590e-01 -4.60744239e-02 6.11519158e-01 8.68209481e-01
4.83241737e-01 -2.31010601e-01 4.90881801e-01 -6.79081559e-01
-4.85446081e-02 -6.40466273e-01 -6.08691037e-01 2.77398884e-01
-5.42777598e-01 -1.58348799e-01 3.97585273e-01 -6.62995100e-01
-8.20594668e-01 4.36781952e-03 3.70719939e-01 -4.91832376e-01
-3.16802353e-01 -2.00654835e-01 -6.98391944e-02 -2.05224082e-02
1.21032941e+00 2.92094439e-01 -3.96406978e-01 -4.96940643e-01
7.79491425e-01 8.59705925e-01 9.48173344e-01 -3.45598638e-01
9.66624737e-01 7.27954626e-01 -5.04434168e-01 -9.05207574e-01
-1.16914380e+00 -1.16422427e+00 -8.15571368e-01 4.52582687e-02
8.60046148e-01 -1.43054152e+00 -5.04656255e-01 3.57107282e-01
-9.71972406e-01 -8.97869526e-04 -4.19651926e-01 3.04506242e-01
-2.85930008e-01 4.46025133e-01 -6.49810553e-01 -6.42954350e-01
-3.17973584e-01 -9.80845749e-01 1.21809018e+00 1.45161539e-01
1.28767133e-01 -5.76559603e-01 -2.36176744e-01 3.22254121e-01
7.70731568e-02 -1.25354873e-02 5.23713529e-01 -8.20506155e-01
-7.55991817e-01 -4.14227307e-01 -8.04182529e-01 5.98725498e-01
-5.95332608e-02 -2.79153943e-01 -1.33768320e+00 -5.10412276e-01
-4.77725387e-01 -9.13749814e-01 1.37176096e+00 -4.96099740e-02
1.03893793e+00 -2.06120834e-02 -5.64332783e-01 7.15667427e-01
1.42388594e+00 -2.74768591e-01 1.90301329e-01 2.47922197e-01
9.23381567e-01 1.79533809e-01 2.99225181e-01 4.52654451e-01
4.41389829e-01 6.39853060e-01 5.98773420e-01 -1.37070775e-01
-3.67140949e-01 -2.99533933e-01 2.00264588e-01 5.44492066e-01
1.67906389e-01 -1.37013659e-01 -1.01500273e+00 9.07275975e-01
-1.79257047e+00 -9.19178069e-01 1.48230856e-02 2.20413947e+00
6.87120080e-01 3.20631534e-01 7.98858479e-02 -2.33336642e-01
7.25356340e-01 -1.66844558e-02 -7.36337304e-01 1.16674282e-01
-4.02128845e-01 -1.75340608e-01 8.36503863e-01 8.59495660e-04
-1.64842725e+00 1.08863544e+00 6.55378103e+00 9.16983306e-01
-8.90509367e-01 4.41667140e-01 3.34479779e-01 -1.34630799e-01
1.64277866e-01 -1.39654368e-01 -1.29195583e+00 3.17805558e-01
5.59056997e-01 -1.87600717e-01 1.06864557e-01 1.28054500e+00
-5.19983530e-01 1.27521724e-01 -1.20892668e+00 1.23180258e+00
2.78140724e-01 -1.38454390e+00 1.93440378e-01 9.04981121e-02
9.19669807e-01 7.79421628e-01 2.40737811e-01 6.61667109e-01
7.72746265e-01 -8.19969833e-01 9.66729522e-01 6.46338835e-02
1.11621976e+00 -3.33639145e-01 6.39560103e-01 6.07796609e-01
-1.17348194e+00 -5.07783651e-01 -8.14855754e-01 -4.22275951e-03
-2.36184254e-01 3.39154303e-01 -8.92345488e-01 2.58056760e-01
1.08695269e+00 8.31764102e-01 -1.02318323e+00 1.10782921e+00
-2.44791761e-01 6.31682098e-01 -4.93040144e-01 3.24307591e-01
2.82629460e-01 2.27049917e-01 5.44079304e-01 1.41831732e+00
4.38818969e-02 -6.87222332e-02 7.48002231e-01 6.71613455e-01
-3.39476228e-01 4.82267812e-02 -8.06829333e-01 2.82073498e-01
4.15460348e-01 1.20220649e+00 -7.07178116e-01 -4.91366684e-01
-7.14756608e-01 1.05948722e+00 5.35469472e-01 3.60247105e-01
-8.26762557e-01 -3.09324384e-01 5.78726649e-01 -1.78050175e-02
4.92745906e-01 -5.26138097e-02 -1.64236054e-01 -1.45377183e+00
-2.48333663e-02 -5.65704167e-01 7.66522706e-01 -5.75161397e-01
-1.65792274e+00 5.61239123e-01 -1.61478490e-01 -1.23896766e+00
1.05880551e-01 -9.99680459e-01 -4.07661915e-01 5.86082160e-01
-1.40420580e+00 -1.49055946e+00 -4.52469349e-01 6.61039650e-01
6.57078385e-01 -4.80468065e-01 7.81127095e-01 2.94690788e-01
-6.58881247e-01 7.91368306e-01 4.11774218e-01 3.35930556e-01
9.24071431e-01 -1.37479401e+00 4.88190114e-01 1.08186805e+00
5.72886646e-01 4.80171084e-01 5.46134770e-01 -3.01198453e-01
-1.14060402e+00 -1.38197720e+00 5.18496633e-01 -1.03712201e+00
9.86043513e-01 -7.78210580e-01 -8.74772549e-01 9.26916659e-01
-2.26886019e-01 6.51539505e-01 6.05029345e-01 4.35070366e-01
-1.32903826e+00 -1.60700426e-01 -7.36183763e-01 4.28637862e-01
1.32443583e+00 -8.11006486e-01 -8.30729842e-01 6.26215518e-01
9.92452741e-01 -3.32545877e-01 -4.78986263e-01 3.87300879e-01
4.12685990e-01 -7.85865009e-01 1.26195300e+00 -7.79883146e-01
2.52509505e-01 -2.95486987e-01 -6.94052517e-01 -8.48881483e-01
-7.53413439e-01 -9.37312692e-02 -3.86854857e-01 1.23059821e+00
2.68713921e-01 -8.37194443e-01 3.13638806e-01 2.29781866e-01
4.91761020e-05 -3.87684494e-01 -8.89191508e-01 -1.25925469e+00
1.69418324e-02 -5.18097401e-01 2.90327281e-01 7.90625632e-01
-1.92358598e-01 6.50872231e-01 -5.19432366e-01 3.39074850e-01
9.02609527e-01 3.31310660e-01 1.00832713e+00 -1.43344975e+00
-5.28893054e-01 -2.26899058e-01 -8.50803673e-01 -1.26837647e+00
2.01245807e-02 -1.23127449e+00 6.29175082e-02 -1.29295003e+00
9.24404979e-01 -5.23178637e-01 -6.83524668e-01 6.76916778e-01
-1.82372913e-01 7.79314995e-01 4.67059165e-01 5.16521215e-01
-1.19849420e+00 5.38085997e-01 8.72562408e-01 -4.92622018e-01
3.86161953e-02 -3.13306838e-01 -9.95796800e-01 8.11043203e-01
3.27864319e-01 -2.39275634e-01 -2.99636871e-01 -5.50722063e-01
-2.15926170e-01 -6.57198370e-01 7.60915399e-01 -1.20720649e+00
1.48303106e-01 -4.78805155e-02 4.35675025e-01 -4.87101465e-01
1.92629546e-01 -6.14579260e-01 -4.31069493e-01 2.86272109e-01
-8.17086250e-02 -4.82047468e-01 2.22084999e-01 1.10823202e+00
-2.74763871e-02 2.29313672e-02 9.83797371e-01 -1.13600485e-01
-1.56571651e+00 5.00643313e-01 6.66347519e-02 3.20694685e-01
1.21359956e+00 -2.38977268e-01 -5.51868737e-01 4.02725376e-02
-6.83247507e-01 3.28601480e-01 4.22822922e-01 8.84579897e-01
3.93908203e-01 -1.17267954e+00 -6.37604177e-01 1.67881519e-01
9.87768888e-01 -3.87458503e-02 2.65803009e-01 4.34997857e-01
-1.25932083e-01 5.26486814e-01 -1.17427111e-02 -1.21328342e+00
-1.11410940e+00 1.07347643e+00 2.86706835e-01 8.99569392e-02
-9.70853925e-01 1.03967810e+00 1.04943335e+00 -5.50990045e-01
3.33445132e-01 5.93781397e-02 -5.94260655e-02 4.65702154e-02
7.60864079e-01 1.20179228e-01 -6.24249727e-02 -8.46844614e-01
-3.85105371e-01 4.20001686e-01 -2.89768159e-01 4.14450318e-01
1.11190724e+00 -2.94734389e-01 1.60576448e-01 5.32004178e-01
1.14705825e+00 -3.41162682e-01 -1.38859355e+00 -7.34512568e-01
-1.40933827e-01 -5.30806422e-01 2.29720809e-02 -6.72354400e-01
-6.33702517e-01 9.59990323e-01 9.52983737e-01 1.40142739e-01
5.62910140e-01 7.25744843e-01 6.60920858e-01 5.41268826e-01
6.73707664e-01 -9.70377862e-01 3.24679196e-01 7.06547558e-01
6.70794964e-01 -1.75486612e+00 -1.76055878e-01 -4.61003125e-01
-6.53401971e-01 7.27220774e-01 9.20504451e-01 1.16599523e-01
4.91777152e-01 -6.79656193e-02 1.99487116e-02 -2.14023873e-01
-6.04086399e-01 -6.33002460e-01 5.78234315e-01 7.46913791e-01
-8.69327933e-02 3.61854583e-01 4.79138643e-01 6.67581201e-01
1.01966210e-01 -3.96941334e-01 2.74977893e-01 4.10518348e-01
-8.26220334e-01 -4.02660638e-01 -3.64931643e-01 6.08434796e-01
-7.44308159e-02 -1.32195294e-01 -1.24678180e-01 7.73833871e-01
4.38579708e-01 8.79107833e-01 1.15204044e-01 -2.72285014e-01
3.61738056e-01 -2.83139627e-02 2.21966147e-01 -1.05712068e+00
-1.12250762e-03 -2.40567744e-01 -3.56895477e-01 -5.15577376e-01
1.13076553e-01 -4.17005718e-01 -9.01435196e-01 -1.06199257e-01
-5.54306090e-01 -4.06621754e-01 2.17917785e-01 8.28925371e-01
4.96084183e-01 3.09297204e-01 3.97394240e-01 -7.87781715e-01
-9.72207487e-01 -8.87586772e-01 -7.77581573e-01 7.09001362e-01
4.25409585e-01 -1.04211235e+00 -5.15231550e-01 1.17555730e-01] | [9.549468040466309, 1.5091547966003418] |
a4dafa29-e704-4bd0-bd33-89d27ae8f173 | synthesizing-adversarial-negative-responses | 2106.05894 | null | https://arxiv.org/abs/2106.05894v1 | https://arxiv.org/pdf/2106.05894v1.pdf | Synthesizing Adversarial Negative Responses for Robust Response Ranking and Evaluation | Open-domain neural dialogue models have achieved high performance in response ranking and evaluation tasks. These tasks are formulated as a binary classification of responses given in a dialogue context, and models generally learn to make predictions based on context-response content similarity. However, over-reliance on content similarity makes the models less sensitive to the presence of inconsistencies, incorrect time expressions and other factors important for response appropriateness and coherence. We propose approaches for automatically creating adversarial negative training data to help ranking and evaluation models learn features beyond content similarity. We propose mask-and-fill and keyword-guided approaches that generate negative examples for training more robust dialogue systems. These generated adversarial responses have high content similarity with the contexts but are either incoherent, inappropriate or not fluent. Our approaches are fully data-driven and can be easily incorporated in existing models and datasets. Experiments on classification, ranking and evaluation tasks across multiple datasets demonstrate that our approaches outperform strong baselines in providing informative negative examples for training dialogue systems. | ['Jeffrey P. Bigham', 'Yulia Tsvetkov', 'Prakhar Gupta'] | 2021-06-10 | null | https://aclanthology.org/2021.findings-acl.338 | https://aclanthology.org/2021.findings-acl.338.pdf | findings-acl-2021-8 | ['dialogue-evaluation'] | ['natural-language-processing'] | [ 3.74946237e-01 5.07726490e-01 -5.71327694e-02 -1.09615624e+00
-1.11592162e+00 -8.14874172e-01 1.10383844e+00 1.86786667e-01
-4.81393754e-01 1.00884056e+00 7.39435375e-01 -1.77516535e-01
1.29221439e-01 -6.08625352e-01 8.90646782e-03 6.58138609e-03
2.67223001e-01 9.90420759e-01 -2.57554967e-02 -9.70310628e-01
3.93095613e-01 -6.47791252e-02 -9.99401093e-01 1.00669920e+00
7.27292001e-01 6.83444083e-01 -4.01178032e-01 1.08996415e+00
-2.47125059e-01 1.52935612e+00 -1.23066831e+00 -6.03956223e-01
-1.80741679e-02 -7.12823153e-01 -1.43626857e+00 -2.80438572e-01
1.53187662e-01 -6.94759130e-01 -1.86618254e-01 4.80019420e-01
7.65297115e-01 4.95421648e-01 9.08510923e-01 -1.09321046e+00
-9.05249655e-01 7.64451861e-01 1.54406443e-01 1.14202835e-01
1.04869962e+00 3.68261278e-01 1.01640308e+00 -8.21026921e-01
6.94949329e-01 1.44899690e+00 6.25591636e-01 1.28570461e+00
-1.44460857e+00 -4.39784974e-01 1.36594519e-01 7.96136111e-02
-5.55513799e-01 -7.99407482e-01 6.85373306e-01 -3.25201333e-01
1.21086752e+00 7.17075825e-01 7.98280984e-02 1.66254270e+00
-2.80549675e-01 8.59381437e-01 1.16316926e+00 -3.99756670e-01
4.25521806e-02 3.86378616e-01 2.41315559e-01 1.80759951e-01
-7.81455159e-01 4.04248014e-02 -5.45496702e-01 -6.04675233e-01
1.47056311e-01 -3.69155377e-01 -3.29759300e-01 2.62399077e-01
-1.07516754e+00 1.17948604e+00 2.44234994e-01 2.66599879e-02
-1.89888611e-01 -2.81711489e-01 7.49104381e-01 8.37911606e-01
7.87701607e-01 1.29582465e+00 -6.93959653e-01 -3.83342594e-01
-4.17404950e-01 8.33769679e-01 1.40227103e+00 7.17590153e-01
4.55492169e-01 8.33166465e-02 -9.34431732e-01 1.54273331e+00
-1.59628257e-01 2.76429057e-01 7.64432192e-01 -1.28838825e+00
5.51584184e-01 5.81537127e-01 3.62560838e-01 -9.50065553e-01
-5.07781863e-01 1.70323536e-01 -5.34245729e-01 -2.00871840e-01
4.62512374e-01 -3.90254796e-01 -4.22672421e-01 1.93782854e+00
1.54155388e-01 -4.25842822e-01 5.24315178e-01 8.81037712e-01
1.39601803e+00 5.30533731e-01 1.51623160e-01 -2.14972571e-01
9.72273350e-01 -9.63272035e-01 -7.78841853e-01 -3.38754714e-01
9.88621235e-01 -8.17326128e-01 1.47532988e+00 1.56488791e-01
-1.10843205e+00 -2.83540547e-01 -6.67426765e-01 -1.31540000e-01
-2.06192836e-01 -2.54451871e-01 4.67639476e-01 2.87936479e-01
-1.05965531e+00 5.70461571e-01 9.48166996e-02 -6.07871637e-02
-2.20380962e-01 3.50170195e-01 -3.11909556e-01 8.76284689e-02
-1.70238185e+00 1.36224973e+00 -4.51112073e-03 -1.42979622e-01
-5.99543691e-01 -6.02353036e-01 -9.66605663e-01 -2.95887589e-01
1.42846227e-01 -4.52133268e-01 2.17810035e+00 -1.18955469e+00
-1.74553525e+00 8.76073182e-01 1.91618744e-02 -4.01275098e-01
6.69751048e-01 -3.38925757e-02 -4.24138099e-01 5.02564060e-03
1.86783046e-01 8.31840932e-01 3.50082010e-01 -1.05414069e+00
-2.61382759e-01 -1.72170937e-01 4.01655346e-01 5.75627446e-01
-3.29795659e-01 3.47151101e-01 2.46260539e-01 -4.95873034e-01
-2.04515010e-01 -7.86097944e-01 -3.10518414e-01 -5.20327747e-01
-5.49641490e-01 -6.76811516e-01 5.28401792e-01 -5.47795832e-01
1.11904442e+00 -1.45790482e+00 -1.15521185e-01 -8.70939568e-02
1.40603676e-01 2.60319889e-01 -3.95644516e-01 6.80624187e-01
5.80693856e-02 3.39430243e-01 4.25120108e-02 -1.55069962e-01
1.83105469e-01 -1.64098796e-02 -5.53551972e-01 -2.40745291e-01
5.63591659e-01 8.27605426e-01 -1.21484685e+00 -3.22752744e-01
1.32351033e-02 -4.33446467e-02 -6.70999646e-01 8.39105904e-01
-6.33654296e-01 5.97962737e-01 -4.36273009e-01 2.91708052e-01
7.96022341e-02 6.71245903e-02 2.13633060e-01 1.72008917e-01
3.52679223e-01 1.04633355e+00 -4.85235631e-01 1.19293237e+00
-7.16822386e-01 6.39859319e-01 6.57485612e-03 -6.85615838e-01
1.13616121e+00 5.11104167e-01 6.28703013e-02 -9.56068516e-01
-9.43028033e-02 7.87311122e-02 1.22922629e-01 -6.39510572e-01
1.05508649e+00 -6.52294010e-02 -6.50558412e-01 1.08151054e+00
-5.23826517e-02 -6.58533633e-01 1.39753759e-01 6.44305229e-01
1.16549218e+00 -9.00759473e-02 1.52974591e-01 9.41359922e-02
4.37915117e-01 1.83411062e-01 4.42358285e-01 1.05233777e+00
-2.82433540e-01 4.86092687e-01 7.77571321e-01 -4.97360319e-01
-9.32156503e-01 -5.43594539e-01 1.28447726e-01 1.69703650e+00
-3.39647442e-01 -4.76978481e-01 -6.52539074e-01 -9.24860299e-01
-1.38434440e-01 9.38529909e-01 -6.30523860e-01 -3.97963494e-01
-5.39559782e-01 -3.89667749e-01 9.54951584e-01 4.69020665e-01
2.11413085e-01 -1.47028697e+00 -1.51095882e-01 4.45625275e-01
-7.18580723e-01 -8.61109138e-01 -4.69081104e-01 2.40641430e-01
-4.21108663e-01 -9.90912974e-01 -3.38713080e-01 -6.58259809e-01
4.56499040e-01 -1.78640950e-02 1.70227218e+00 2.57312745e-01
1.27858698e-01 5.49091041e-01 -5.15545964e-01 -3.97260040e-01
-1.18318784e+00 -1.47494212e-01 7.27250725e-02 -4.90079343e-01
6.54331684e-01 -2.17100233e-01 -3.77683550e-01 6.01477444e-01
-5.06759048e-01 4.70751189e-02 -1.18549690e-01 1.47408986e+00
-2.36988161e-02 -9.35996890e-01 1.16830087e+00 -1.42470872e+00
1.70195949e+00 -5.98520339e-01 1.78421885e-01 2.64790148e-01
-6.80822492e-01 1.80099327e-02 9.36079025e-01 -6.29736125e-01
-1.23342693e+00 -3.28197986e-01 -2.54097521e-01 5.64959720e-02
-3.34294051e-01 2.99191386e-01 2.37316728e-01 2.16093168e-01
1.34179199e+00 -1.06807977e-01 5.87802827e-02 -6.97621852e-02
5.28528035e-01 9.83133376e-01 5.48418522e-01 -9.59794164e-01
2.75907665e-01 -2.28931159e-01 -7.99116075e-01 -4.40842927e-01
-8.60898674e-01 -2.29895860e-01 -2.21667424e-01 -3.08335334e-01
4.09129918e-01 -7.08850861e-01 -5.86942792e-01 2.31036797e-01
-1.35356367e+00 -7.08818614e-01 -1.34385839e-01 -1.75880846e-02
-6.62669301e-01 3.03227633e-01 -1.02253139e+00 -9.62862074e-01
-6.62760198e-01 -8.86848032e-01 6.32629514e-01 1.12684287e-01
-1.18225861e+00 -1.11843562e+00 2.31065452e-01 5.11833847e-01
6.86207056e-01 8.42709094e-02 9.52386975e-01 -1.57921982e+00
1.92556590e-01 -3.13178569e-01 1.01858176e-01 2.95914233e-01
-2.68085692e-02 1.89685058e-02 -1.11312902e+00 1.20469920e-01
7.71999061e-02 -1.52833867e+00 5.18424213e-01 -2.46724039e-01
1.02150798e+00 -1.11719704e+00 2.01661617e-01 -2.29290053e-02
4.80334818e-01 1.37039408e-01 3.07655483e-01 1.50757045e-01
2.42552921e-01 1.32259226e+00 9.69569623e-01 6.19708300e-01
5.75591862e-01 6.90999568e-01 1.39691532e-01 -1.17195532e-01
3.49568039e-01 -2.63745040e-01 3.78804743e-01 3.85137945e-01
6.15306556e-01 -3.33925396e-01 -8.52777004e-01 5.78458488e-01
-1.79267311e+00 -1.22809756e+00 -1.64616123e-01 1.91330492e+00
1.65090621e+00 6.78475052e-02 1.88725844e-01 -2.01369897e-01
5.03076077e-01 2.16329336e-01 -4.97994572e-01 -1.20982444e+00
-1.79519966e-01 2.20957458e-01 -2.35814735e-01 7.67856181e-01
-8.42212319e-01 1.03560746e+00 6.83021975e+00 3.54680121e-01
-7.93993771e-01 -2.44964492e-02 9.41362441e-01 -1.92898452e-01
-7.04537570e-01 -1.99714541e-01 -4.78971601e-01 2.15312839e-01
9.93205011e-01 -1.38393119e-01 2.96722710e-01 7.83637583e-01
1.49366722e-01 4.62118164e-02 -1.56529534e+00 3.70566666e-01
1.42311091e-02 -1.09695923e+00 -4.57517542e-02 -4.90479678e-01
6.45269036e-01 -1.61618680e-01 -1.66967162e-03 8.69551301e-01
1.14567018e+00 -1.23481762e+00 3.71675849e-01 2.76041061e-01
7.13803530e-01 -5.12356699e-01 7.89464891e-01 4.55035567e-01
-1.67500287e-01 -8.81054252e-02 -2.27395296e-01 -4.00725543e-01
-1.22408941e-01 -1.80614114e-01 -1.47186387e+00 9.61220264e-03
3.71775568e-01 3.25790793e-01 -4.01866287e-01 1.11877181e-01
-3.45109254e-01 5.08973598e-01 -6.82441220e-02 -5.25047481e-01
2.11078227e-01 2.07823828e-01 3.61051887e-01 1.39918125e+00
-4.44145024e-01 3.74964982e-01 3.93501133e-01 7.92686284e-01
-1.01284251e-01 3.34656298e-01 -8.24226558e-01 -1.01303540e-01
8.08032691e-01 1.16187000e+00 2.72833370e-02 -4.98746485e-01
-1.87523276e-01 7.88254917e-01 5.43485224e-01 3.11577082e-01
-4.87677276e-01 -2.91237712e-01 6.09024823e-01 -2.79370219e-01
-4.77399677e-01 4.47549969e-01 -3.68801028e-01 -9.76624668e-01
-1.87649708e-02 -1.52242386e+00 5.94463825e-01 -6.98095739e-01
-1.73629797e+00 8.19931090e-01 -2.11183175e-01 -1.03249276e+00
-1.30484176e+00 -3.45713794e-01 -8.44938338e-01 1.09332049e+00
-1.08859646e+00 -8.54886770e-01 -8.54129568e-02 6.51966929e-01
9.36913669e-01 -3.61337095e-01 1.52655959e+00 -1.98709264e-01
-3.07294071e-01 9.54541028e-01 -1.86893448e-01 4.17473763e-01
1.38462496e+00 -1.47569418e+00 4.54688221e-01 1.82386503e-01
-2.14357495e-01 8.30188870e-01 8.11320961e-01 -5.34205139e-01
-6.37918532e-01 -7.38546550e-01 1.16494071e+00 -8.77464592e-01
5.77803016e-01 -3.47001493e-01 -1.02333891e+00 4.17597562e-01
5.37764430e-01 -4.99336869e-01 1.22343481e+00 5.04133463e-01
-6.60144448e-01 2.54088908e-01 -1.25479436e+00 7.21883237e-01
6.07507706e-01 -8.57979178e-01 -9.49411809e-01 8.28564465e-01
6.82962775e-01 -6.16497815e-01 -7.13911951e-01 2.96592802e-01
4.36164349e-01 -9.35234368e-01 7.50694573e-01 -1.44325054e+00
9.14013267e-01 3.37308675e-01 9.15712211e-03 -1.65512252e+00
-1.12379886e-01 -9.38773930e-01 5.95348962e-02 1.24224675e+00
7.52620757e-01 -2.08893880e-01 4.46050823e-01 1.41292953e+00
-1.82999775e-01 -6.77819192e-01 -5.39896071e-01 -1.74401388e-01
3.33622158e-01 -1.74201280e-01 5.15750110e-01 1.36192071e+00
6.43450558e-01 9.11684334e-01 -5.33570230e-01 -5.92525482e-01
-1.83821440e-01 -1.49033275e-02 1.07702887e+00 -1.07062840e+00
-3.49747479e-01 -4.79918629e-01 1.87448427e-01 -9.32853043e-01
5.73385417e-01 -6.98006332e-01 2.72287130e-01 -1.44317853e+00
5.94736030e-03 -5.89836538e-01 2.24766314e-01 7.30149150e-01
-5.19276738e-01 1.30055889e-01 -4.55930084e-02 2.03481197e-01
-7.38347650e-01 4.61241573e-01 9.08885717e-01 -2.78022885e-01
-4.07379061e-01 2.39410579e-01 -9.13756430e-01 5.59550762e-01
8.45866442e-01 -3.55434835e-01 -5.17182648e-01 -5.15862107e-01
1.81785166e-01 4.97462928e-01 8.74796286e-02 -3.45554680e-01
2.32024416e-01 -4.03052330e-01 3.30011427e-01 -2.04808682e-01
5.72291315e-01 -1.90751404e-01 -4.99068052e-01 6.85057137e-03
-1.44420433e+00 3.03849041e-01 8.54708254e-02 3.51732373e-01
-2.14412481e-01 -3.31899583e-01 5.31513095e-01 -5.02425134e-01
-4.78274852e-01 -2.25568950e-01 -6.26785994e-01 6.19665742e-01
5.46415865e-01 -1.12138353e-01 -6.88960493e-01 -1.24518073e+00
-6.15213573e-01 6.19420648e-01 2.98737228e-01 7.65193164e-01
7.71879673e-01 -1.14120293e+00 -1.07145023e+00 -9.54616293e-02
4.78738487e-01 -2.27548510e-01 1.64694220e-01 2.93517619e-01
-2.51406208e-02 4.33092386e-01 -1.39622465e-01 -3.13811243e-01
-1.36554933e+00 1.28360525e-01 5.66468537e-01 -5.22681832e-01
-1.29193198e-02 1.09808362e+00 5.25512025e-02 -1.24801326e+00
3.96114826e-01 1.84926569e-01 -5.00150919e-01 3.40096280e-02
6.61806941e-01 -7.06498325e-02 7.86135048e-02 -4.47444737e-01
3.66831683e-02 -3.54969889e-01 -4.70425278e-01 -6.21747017e-01
9.23190951e-01 6.46243319e-02 -7.29339570e-02 4.54483718e-01
9.51307893e-01 -3.79532240e-02 -8.87870431e-01 -6.07442677e-01
1.30505264e-01 -3.25431824e-01 -4.63084280e-01 -1.47847283e+00
-3.24539334e-01 7.87436604e-01 -7.95611143e-02 4.32377964e-01
6.57035172e-01 -2.31776267e-01 5.24677038e-01 8.95580888e-01
-1.20591903e-02 -1.39056182e+00 6.25081241e-01 1.10043240e+00
1.25164294e+00 -1.47965991e+00 -2.77264744e-01 -4.78398614e-02
-1.32946849e+00 1.13303220e+00 1.52351415e+00 2.32124940e-01
-1.93079889e-01 7.92875066e-02 7.19140530e-01 1.00348806e-02
-1.48629260e+00 2.48943508e-01 8.11296776e-02 8.69154632e-01
1.09505415e+00 1.82514284e-02 -4.40453678e-01 8.44711661e-01
-3.97376835e-01 -6.21896923e-01 7.55747974e-01 7.24927962e-01
-9.95931402e-02 -1.32324278e+00 -6.57927915e-02 6.85503006e-01
-5.21372318e-01 -2.45221466e-01 -1.48075676e+00 4.65295672e-01
-7.60796607e-01 1.50121737e+00 -1.64007276e-01 -6.90159917e-01
5.10606825e-01 4.36851621e-01 7.07233846e-02 -1.03368211e+00
-1.40279150e+00 -4.26714808e-01 1.14860308e+00 -4.29828048e-01
-7.42008537e-02 -3.48216176e-01 -1.07321060e+00 -3.29457730e-01
-4.83003765e-01 5.97705126e-01 2.43925929e-01 8.04133117e-01
3.14138561e-01 2.47094944e-01 1.01633918e+00 -4.79023546e-01
-1.44514346e+00 -1.42440653e+00 1.22303843e-01 9.65861261e-01
2.15457663e-01 -2.86327928e-01 -2.88088083e-01 -3.50316137e-01] | [12.729866981506348, 8.145323753356934] |
dc08d566-80cc-4ed5-8f9d-5afdf45cfe13 | exploiting-unsupervised-pre-training-and | 1906.09325 | null | https://arxiv.org/abs/1906.09325v1 | https://arxiv.org/pdf/1906.09325v1.pdf | Exploiting Unsupervised Pre-training and Automated Feature Engineering for Low-resource Hate Speech Detection in Polish | This paper presents our contribution to PolEval 2019 Task 6: Hate speech and bullying detection. We describe three parallel approaches that we followed: fine-tuning a pre-trained ULMFiT model to our classification task, fine-tuning a pre-trained BERT model to our classification task, and using the TPOT library to find the optimal pipeline. We present results achieved by these three tools and review their advantages and disadvantages in terms of user experience. Our team placed second in subtask 2 with a shallow model found by TPOT: a~logistic regression classifier with non-trivial feature engineering. | ['Marcin Możejko', 'Przemysław Sadownik', 'Rafał Rolczyński', 'Tomasz Korbak', 'Renard Korzeniowski'] | 2019-06-17 | null | null | null | null | ['automated-feature-engineering'] | ['methodology'] | [-3.37049216e-01 9.28445086e-02 5.14186285e-02 -6.47960007e-01
-7.69702375e-01 -5.70111394e-01 2.00071126e-01 -2.14202940e-01
-3.59833390e-01 4.00248408e-01 5.71811870e-02 -4.20695305e-01
-2.44951248e-02 -1.73924882e-02 -6.96930960e-02 -2.20243439e-01
8.34436640e-02 6.53194308e-01 4.78084356e-01 -4.00227994e-01
7.96727896e-01 6.44595385e-01 -1.17894447e+00 7.02113390e-01
6.56596839e-01 6.20737195e-01 -3.18573356e-01 1.10520065e+00
6.59144968e-02 1.00205016e+00 -9.98673916e-01 -7.60060012e-01
-1.83369722e-02 -9.64890197e-02 -1.27139747e+00 -2.88091093e-01
6.01205647e-01 -3.18674415e-01 -1.38046727e-01 8.77622902e-01
4.32776153e-01 2.46384763e-03 6.87155604e-01 -1.51111436e+00
-5.19604206e-01 3.69384915e-01 -2.93526143e-01 7.94068038e-01
5.25310814e-01 3.17268431e-01 8.20825279e-01 -6.66746676e-01
4.06009704e-01 1.47758615e+00 1.20069456e+00 7.36968338e-01
-1.53282332e+00 -9.73724782e-01 -4.72777754e-01 2.24326223e-01
-1.43515933e+00 -9.88074541e-01 5.94682574e-01 -7.94210315e-01
1.87086082e+00 2.45263651e-01 3.37456018e-01 1.17574954e+00
9.98170748e-02 6.33856356e-01 1.35663009e+00 -3.38751376e-01
-1.16830051e-01 7.69497156e-01 7.15011001e-01 7.91805863e-01
-2.24098384e-01 -1.99088767e-01 -5.65664232e-01 -6.55899644e-01
4.50104117e-01 -4.15378720e-01 1.10334285e-01 2.64163047e-01
-4.29788291e-01 1.01595867e+00 2.47585714e-01 4.43296283e-01
9.15765949e-03 7.62230307e-02 5.11961937e-01 5.53432226e-01
8.54973376e-01 8.75761211e-01 -6.45155966e-01 -3.00352633e-01
-1.10731947e+00 3.51837277e-01 1.06198061e+00 7.88315177e-01
5.21828890e-01 -3.45659852e-02 -7.12307245e-02 9.14534569e-01
2.69159526e-01 1.09402284e-01 3.28170031e-01 -8.51574898e-01
5.23425817e-01 9.95446071e-02 1.97726026e-01 -8.43802094e-01
-6.03150487e-01 1.52924269e-01 1.46123067e-01 3.61113220e-01
3.33071738e-01 -4.01742131e-01 -6.73207700e-01 1.04400647e+00
-1.08675808e-01 -1.70116611e-02 -3.31186444e-01 3.60397846e-01
7.72912383e-01 3.39323431e-01 3.71268570e-01 -1.28009051e-01
1.07699096e+00 -1.12559342e+00 -7.25733697e-01 -9.82803777e-02
1.06308067e+00 -8.96211803e-01 9.31230724e-01 6.95568800e-01
-1.00770092e+00 -1.91094130e-01 -8.33617806e-01 -2.68206805e-01
-5.21718562e-01 -1.07795946e-01 4.65926766e-01 1.19912148e+00
-1.14518499e+00 8.57622743e-01 -6.13036633e-01 -5.46488345e-01
3.87773812e-01 5.75935781e-01 -3.75558376e-01 3.60356539e-01
-8.77374530e-01 1.42976761e+00 4.33586538e-01 -3.58603388e-01
-6.61758125e-01 -4.48182523e-01 -5.72040081e-01 6.65023923e-02
1.86697438e-01 -2.73193449e-01 1.67286897e+00 -6.62228584e-01
-1.40323353e+00 9.16413486e-01 -3.11233968e-01 -2.16810912e-01
1.67325839e-01 -6.02292478e-01 -5.40729225e-01 4.99362573e-02
-6.54843673e-02 2.72374481e-01 1.00995886e+00 -7.07587481e-01
-4.82602239e-01 -1.16833232e-01 -5.54885752e-02 -5.45575060e-02
-3.43016148e-01 1.29451203e+00 1.56903714e-01 -2.68819451e-01
-6.07044518e-01 -5.39622664e-01 1.36725302e-03 -5.65230787e-01
-4.23687309e-01 -4.92453337e-01 9.37694132e-01 -1.04160690e+00
1.88031328e+00 -2.22788644e+00 -8.13681409e-02 1.02326453e-01
3.88349414e-01 8.50976348e-01 1.10278144e-01 6.05886340e-01
-4.07814324e-01 5.33135295e-01 1.60682052e-02 -6.62570000e-01
1.87912121e-01 1.09965727e-02 -2.43455112e-01 3.99914056e-01
5.53853691e-01 5.89329779e-01 -8.60219538e-01 -6.92609310e-01
2.02301621e-01 3.15396219e-01 -6.73231363e-01 5.09991050e-01
1.74170017e-01 -1.22835748e-01 -1.05452880e-01 6.83084011e-01
4.50634331e-01 4.03646052e-01 -3.22617143e-01 3.47636610e-01
-1.43489093e-01 6.19374216e-01 -6.93798542e-01 1.02661455e+00
-1.27550319e-01 7.98195958e-01 3.28575253e-01 -6.05064809e-01
1.12022328e+00 4.32453632e-01 -1.05739847e-01 4.09205966e-02
3.27095777e-01 2.22024471e-01 -1.85900494e-01 -1.11550808e+00
3.35315496e-01 -4.83731598e-01 -6.15466051e-02 1.56274632e-01
5.96067607e-01 -4.63820100e-01 -2.62870193e-01 -3.89005244e-02
1.46962821e+00 5.19766062e-02 4.53401506e-01 -3.54940206e-01
4.73701537e-01 2.50000566e-01 3.21346641e-01 5.81415892e-01
-8.42614770e-01 4.94317889e-01 9.94058192e-01 -6.70324564e-01
-9.70420718e-01 -5.78963816e-01 -1.54274821e-01 1.50566757e+00
-8.68948340e-01 -6.73241317e-01 -1.00276232e+00 -1.07375956e+00
-1.86202884e-01 1.15426004e+00 -5.76465726e-01 7.03927651e-02
-4.71191347e-01 -8.33835244e-01 1.13026738e+00 2.23202035e-01
1.00732125e-01 -1.31117868e+00 -5.25709987e-01 7.56274089e-02
1.72310680e-01 -8.06653023e-01 -9.64423418e-02 6.79671109e-01
-4.63500351e-01 -8.71767104e-01 -7.17998892e-02 -6.22478724e-01
1.90478444e-01 -1.10708356e-01 1.04158807e+00 2.91241497e-01
-3.08994621e-01 -1.24813825e-01 -4.23098177e-01 -5.54878533e-01
-4.08961475e-01 3.52617025e-01 7.11607784e-02 -5.69415092e-01
1.07845986e+00 -2.90942192e-01 8.64218862e-04 5.58058433e-02
-3.95797580e-01 -2.70928651e-01 1.39150083e-01 7.45165765e-01
-6.14698470e-01 4.98665087e-02 4.82942671e-01 -1.26169300e+00
9.27819610e-01 -6.60011351e-01 -1.84884995e-01 3.20390984e-02
-3.52506876e-01 -2.58389145e-01 4.56194609e-01 -2.69923478e-01
-8.45220566e-01 1.55182913e-01 -8.91109705e-01 -5.67122161e-01
-4.64545399e-01 -5.30257868e-03 3.30895036e-02 -1.35316223e-01
1.00078142e+00 -5.84170699e-01 -6.00178391e-02 -8.09678853e-01
1.77708551e-01 1.08681464e+00 3.16693693e-01 -2.97633857e-01
8.54773045e-01 -3.82258475e-01 -5.11131585e-01 -9.29900229e-01
-9.57655072e-01 -8.20750177e-01 -1.08306217e+00 1.18380882e-01
1.00897896e+00 -3.97755563e-01 -5.77900589e-01 5.05421519e-01
-1.39567864e+00 -3.86558980e-01 1.80272356e-01 1.62703618e-02
-3.69592547e-01 2.56443232e-01 -8.88971865e-01 -1.21886420e+00
-6.53877676e-01 -1.06684387e+00 9.29592073e-01 -1.51011273e-01
-8.10489416e-01 -1.10508811e+00 4.68061149e-01 5.03795087e-01
4.52778935e-01 -8.53558332e-02 9.42825317e-01 -1.42776215e+00
4.68784094e-01 -3.08325380e-01 -4.99998599e-01 5.59614182e-01
-4.31697428e-01 2.79663444e-01 -1.53102779e+00 -2.18205765e-01
-6.70656860e-02 -5.75598538e-01 6.44841611e-01 -1.07291460e-01
9.96296346e-01 -1.63529441e-01 -2.62432933e-01 5.70030034e-01
1.13418853e+00 1.82982981e-01 4.23325330e-01 3.71491522e-01
7.38658667e-01 7.89650857e-01 6.20689213e-01 8.66773874e-02
5.93084283e-02 6.70459151e-01 9.51179340e-02 2.07470313e-01
1.46977246e-01 -1.47677749e-01 7.73842156e-01 5.58548212e-01
3.63571346e-02 1.19538099e-01 -1.32767260e+00 4.73903805e-01
-1.76319778e+00 -1.14842761e+00 -4.08907503e-01 1.59785926e+00
6.46021008e-01 3.04253191e-01 6.30239487e-01 3.69566828e-02
4.63408649e-01 1.00244209e-02 7.63697252e-02 -1.39327669e+00
5.88085651e-01 4.92369235e-01 2.55517870e-01 8.44897389e-01
-1.25919366e+00 1.52433074e+00 9.00310421e+00 9.08601582e-01
-8.25053930e-01 3.94062102e-01 4.55426097e-01 -2.53754079e-01
1.74529433e-01 -1.55949280e-01 -1.11727583e+00 2.42433161e-01
1.72337723e+00 -3.25640999e-02 5.51589549e-01 1.08847821e+00
6.76502585e-02 6.60811961e-02 -9.60202575e-01 7.43928254e-01
5.03834367e-01 -8.00634086e-01 -6.00157082e-01 -1.08532488e-01
-2.08815094e-02 3.34443182e-01 -1.55493945e-01 8.45860779e-01
5.73670506e-01 -1.30769241e+00 4.10085052e-01 3.69355232e-01
4.81977820e-01 -9.64497268e-01 8.80182505e-01 5.51653743e-01
-4.95372117e-01 -3.00390631e-01 -4.23254371e-01 -3.53504151e-01
-3.05686057e-01 1.06557973e-01 -1.45849884e+00 -1.90956239e-02
1.00262570e+00 4.03565139e-01 -7.95910835e-01 9.86180782e-01
-2.70200461e-01 9.67319250e-01 -1.26735926e-01 4.44425317e-03
9.34021026e-02 1.67351216e-01 4.70674962e-01 1.97683692e+00
-1.78893611e-01 8.33379999e-02 4.54017445e-02 5.85091889e-01
4.76496041e-01 2.56193243e-02 -7.61680365e-01 8.54413435e-02
3.31428349e-01 1.38650715e+00 -3.10042650e-01 -2.68661767e-01
-3.07458162e-01 8.43274236e-01 7.70697474e-01 -4.70196791e-02
-7.10552335e-01 -8.73994052e-01 7.30930626e-01 1.58186868e-01
-1.35288164e-02 -2.27499288e-02 -4.38636452e-01 -1.11830389e+00
-5.89603543e-01 -9.18395638e-01 3.15185130e-01 -9.13514555e-01
-1.39575148e+00 8.75401676e-01 4.17887390e-01 -3.15520287e-01
-5.79761028e-01 -9.81973052e-01 -8.39440823e-01 1.26197183e+00
-1.05304527e+00 -1.15946722e+00 3.30447465e-01 3.47735018e-01
5.83915234e-01 -2.84470886e-01 1.06016183e+00 1.52859762e-01
-1.00232971e+00 5.99218845e-01 -4.14872557e-01 9.32321325e-02
9.58352089e-01 -1.50978446e+00 5.93511164e-01 7.59993076e-01
-4.37481076e-01 8.66583884e-01 8.48052084e-01 -6.78700745e-01
-6.82697594e-01 -7.34815955e-01 1.49962783e+00 -1.24807763e+00
1.23385823e+00 -6.87592685e-01 -1.17312431e+00 9.91414905e-01
4.35754865e-01 -3.62300724e-01 9.92412806e-01 6.11264348e-01
-5.07611930e-01 3.05601090e-01 -1.51091981e+00 2.15701535e-01
7.93265462e-01 -8.57157350e-01 -1.10576797e+00 2.97065258e-01
6.88677907e-01 -2.09709674e-01 -9.11855817e-01 -5.90560548e-02
6.06615901e-01 -9.81605113e-01 5.53277433e-01 -9.50243652e-01
4.19837028e-01 1.64464220e-01 1.50381476e-01 -1.35650814e+00
-6.07177615e-01 -9.49785948e-01 3.66456807e-02 1.37272644e+00
3.99954826e-01 -5.47429085e-01 5.51310420e-01 8.16669047e-01
-1.35126755e-01 -6.54275000e-01 -8.07428181e-01 -3.58572245e-01
3.31141889e-01 -5.54299831e-01 2.40029708e-01 1.05065095e+00
5.67989171e-01 6.78756177e-01 -6.05044901e-01 3.72554064e-02
1.31401852e-01 -7.05581486e-01 8.16528320e-01 -1.20069873e+00
-3.38321328e-01 -3.87259960e-01 -3.36324245e-01 -4.85182732e-01
2.94521809e-01 -6.51184797e-01 7.37779960e-02 -9.55962479e-01
2.65712291e-01 -2.56126672e-01 -3.04379970e-01 1.02440095e+00
-2.42887959e-01 3.29348028e-01 3.25815916e-01 2.03610614e-01
-5.40458322e-01 -1.46669462e-01 3.06238919e-01 3.50372165e-01
-1.69812322e-01 3.59924100e-02 -8.30597460e-01 9.58314836e-01
9.56081629e-01 -8.25803936e-01 -2.52425492e-01 -2.55903393e-01
7.88888112e-02 -6.47288188e-02 2.71980077e-01 -9.09872174e-01
-3.02192420e-02 -1.60534456e-01 4.33225423e-01 -6.06345296e-01
5.52696466e-01 -6.76578164e-01 -3.05386871e-01 3.82835448e-01
-2.06743628e-01 3.04536343e-01 3.90856236e-01 -1.87706307e-01
1.54061690e-01 -9.97917712e-01 8.31156552e-01 -5.42813800e-02
-5.90976238e-01 -2.97497749e-01 -7.93563843e-01 -2.69250691e-01
9.10818577e-01 1.02852158e-01 -7.14519858e-01 -1.56608567e-01
-8.47526550e-01 1.31767794e-01 4.01067048e-01 3.41612905e-01
3.92083079e-01 -5.45715213e-01 -5.55800498e-01 4.05815750e-01
2.51303450e-03 -7.11544693e-01 -3.54318798e-01 7.20278442e-01
-6.20601475e-01 4.54144865e-01 -5.69701433e-01 -2.19695792e-01
-1.77571166e+00 6.75633788e-01 3.71771336e-01 -2.90459752e-01
-3.25731158e-01 1.06428742e+00 -2.55783290e-01 -6.79447353e-01
1.04561269e-01 1.36440173e-01 -5.19106448e-01 -7.15509662e-03
8.69962037e-01 5.57267427e-01 1.43642843e-01 -7.45118201e-01
-4.48963493e-01 -4.09554457e-03 -4.40758854e-01 -2.85917372e-01
1.44475543e+00 1.85326338e-01 -2.33280614e-01 5.55182338e-01
1.15577066e+00 4.18490432e-02 -6.91931784e-01 1.75733417e-01
3.92680556e-01 -5.65951109e-01 1.33835599e-01 -8.95467520e-01
-3.06379676e-01 9.99892056e-01 3.11477661e-01 4.53459442e-01
8.78222227e-01 -1.45656243e-01 4.56114769e-01 5.10399878e-01
2.49117255e-01 -9.62479293e-01 1.60055496e-02 8.58619809e-01
7.60468721e-01 -8.57838869e-01 1.85326681e-01 -6.12109780e-01
-8.54144037e-01 1.17999220e+00 8.01544309e-01 -4.30829763e-01
6.92657351e-01 2.21752763e-01 3.17228734e-01 -5.06959319e-01
-1.06789362e+00 -7.97233917e-03 1.25769526e-01 9.13849115e-01
7.34276235e-01 2.46258210e-02 -9.25192311e-02 5.21879017e-01
-7.09428668e-01 -7.61558637e-02 3.38548869e-01 9.04646695e-01
-6.46041989e-01 -9.76093829e-01 -5.40058017e-01 4.32943076e-01
-7.41510510e-01 -1.70240104e-01 -1.00171125e+00 8.25310051e-01
3.09960514e-01 1.17820764e+00 -2.55602628e-01 -8.65847170e-01
4.36516792e-01 5.29534578e-01 1.39692307e-01 -1.13667202e+00
-1.49386179e+00 1.52872592e-01 9.02767122e-01 -4.47557479e-01
2.68474162e-01 -9.50407445e-01 -5.69096744e-01 -7.00687289e-01
-4.36187565e-01 2.20829740e-01 5.79146087e-01 1.00932336e+00
3.58317308e-02 1.97739691e-01 5.54744601e-01 -9.49860096e-01
-6.25645339e-01 -1.35448134e+00 -3.54359478e-01 4.76299003e-02
3.73179376e-01 -4.26695794e-01 -4.69058424e-01 -1.74380332e-01] | [8.823739051818848, 10.601326942443848] |
7ceb6d11-a504-4469-a9cd-1ce849f7a13c | ecg-based-blood-pressure-estimation-using | 2008.10099 | null | https://arxiv.org/abs/2008.10099v1 | https://arxiv.org/pdf/2008.10099v1.pdf | ECG-Based Blood Pressure Estimation Using Mechano-Electric Coupling Concept | The Electrocardiograph signal represents the heart's electrical activity while blood pressure results from the heart's mechanical activity. Previous studies have investigated how the heart's electrical and mechanical activities are related and have referred to their relationship as the Mechano-Electric Coupling term. A new method to estimate the blood pressure including is proposed which uses only the Electrocardiograph signal. In spite of studies performed on feature extraction based on the signals' physiological parameters (Parameter-based), in this work, the feature vectors are formed with samples of the Electrocardiograph signal in a particular time frame (Whole-based) and these vectors are input into Adaptive Boosting Regression to estimate blood pressure. The nonlinear relationship which correlates blood pressure with the Electrocardiograph signal is concluded by the results of this study. According to the results, the used algorithms, for estimating both diastolic blood pressures and mean arterial pressure, are in compliance with the standards of the Association for the Advancement of Medical Instrumentation. Also, according to the British Hypertension Society standard, estimating diastolic blood pressures and mean arterial pressure with the proposed method attain an A grade while it achieves B for systolic blood pressure. The results indicate that using the introduced method, blood pressure can be estimated continuously, noninvasively, without cuff, calibration-free and by using only the Electrocardiograph signal. | ['Yadollah Ghorbani', 'Maryam Moghadam', 'Mohammad Hemmati', 'Mohammad Firouzmand', 'Mostafa Charmi', 'Seyedeh Somayyeh Mousavi'] | 2020-08-23 | null | null | null | null | ['blood-pressure-estimation'] | ['medical'] | [ 1.47930071e-01 -1.66876853e-01 -1.08463168e-01 -3.70104402e-01
1.18928835e-01 -2.47411355e-01 -3.65774706e-02 2.17545837e-01
-6.38383985e-01 1.01308143e+00 -4.69189063e-02 -3.04462224e-01
-2.03601345e-01 -8.37443352e-01 -1.33936450e-01 -6.00070298e-01
-1.97097316e-01 2.47813642e-01 -1.18110493e-01 2.23578960e-02
2.41537720e-01 6.01815224e-01 -1.27627516e+00 -3.83574992e-01
9.32107747e-01 1.03366590e+00 -3.58430266e-01 9.87596631e-01
1.23981051e-01 3.15911502e-01 -8.11208785e-01 1.45711094e-01
2.58205742e-01 -9.46939588e-01 -2.52550602e-01 -1.90661460e-01
7.47053474e-02 -3.36558074e-01 -2.42442880e-02 7.02827275e-01
7.96809375e-01 -2.50761032e-01 6.57092929e-01 -7.28629291e-01
-6.86952323e-02 3.80276024e-01 -2.77745217e-01 4.31129426e-01
2.59392738e-01 -1.35057062e-01 2.38577694e-01 -4.55871761e-01
2.54417092e-01 6.64089322e-01 9.71640289e-01 2.11093917e-01
-1.66254532e+00 -3.95590395e-01 -3.23164076e-01 1.00794747e-01
-1.40160608e+00 -3.09559017e-01 1.00017071e+00 -5.31183243e-01
6.03655636e-01 7.59014487e-01 1.25406802e+00 4.00663912e-01
7.56958961e-01 -1.99616682e-02 1.44181526e+00 -7.62610912e-01
1.40549541e-01 7.02444017e-01 5.00350296e-01 4.21349645e-01
6.17114365e-01 1.90146953e-01 7.29655623e-02 -3.41297954e-01
8.54931712e-01 -2.36625671e-01 -5.77008843e-01 -4.77115400e-02
-7.77443230e-01 6.02466762e-01 -2.57375330e-01 9.37039912e-01
-7.18459785e-01 -6.84898496e-02 5.31352699e-01 3.69506001e-01
1.62286282e-01 5.02001643e-01 -4.48975772e-01 -4.06486481e-01
-9.57421780e-01 2.95520127e-01 1.15871394e+00 2.51230180e-01
3.23290259e-01 1.55753523e-01 -5.80581352e-02 5.64533710e-01
4.24607843e-01 7.04816759e-01 5.08523166e-01 -7.95916736e-01
-2.14951439e-03 4.94306028e-01 2.44461820e-01 -1.33279264e+00
-7.11219072e-01 -4.83034045e-01 -8.59921277e-01 1.62898675e-01
7.44822919e-01 -4.00520593e-01 -3.81582499e-01 1.43823004e+00
4.01578039e-01 -5.84589504e-02 2.47395914e-02 8.73319447e-01
7.47779548e-01 2.69103587e-01 2.51084268e-01 -9.35126662e-01
1.43544769e+00 -9.58105028e-02 -1.08296752e+00 3.40375811e-01
2.39411592e-01 -5.82526088e-01 4.47122216e-01 6.03602827e-01
-9.30141628e-01 -9.23285186e-01 -1.03180432e+00 3.78827572e-01
-1.41486570e-01 1.06141515e-01 2.87196189e-01 1.10677814e+00
-5.73364496e-01 8.35473478e-01 -7.31590986e-01 -2.31016368e-01
-2.15403020e-01 3.08280960e-02 6.00830540e-02 7.86308885e-01
-1.43023312e+00 1.32979798e+00 1.04680592e-02 4.40411091e-01
1.16071984e-01 -7.97939420e-01 -4.27890629e-01 -1.73743531e-01
-3.83756667e-01 -6.59029305e-01 6.20089948e-01 -4.62715715e-01
-1.68469739e+00 7.02396929e-01 -1.40419453e-01 -2.89695263e-01
6.88082814e-01 -7.11399540e-02 -6.61796391e-01 4.68222320e-01
-2.00520515e-01 -1.36561021e-01 4.95448619e-01 -6.39525294e-01
-1.89054370e-01 -6.10128701e-01 -6.16136372e-01 -1.83458582e-01
1.13100104e-01 -1.75376952e-01 1.11145355e-01 -2.13399217e-01
3.70125949e-01 -5.99910855e-01 -4.03870851e-01 -3.00928146e-01
-4.21764106e-02 -9.68490448e-03 3.89182955e-01 -9.03674960e-01
1.42507672e+00 -2.02964950e+00 -9.15435851e-02 8.27250183e-01
1.97439432e-01 3.53059471e-01 6.78384840e-01 5.07200122e-01
-2.10489213e-01 3.54145244e-02 -2.17806369e-01 5.71053922e-01
-1.49418429e-01 6.63812160e-02 -2.23167948e-02 7.89773524e-01
-1.99256897e-01 6.64391279e-01 -6.09214365e-01 -6.49937749e-01
6.59025252e-01 5.82002044e-01 -9.29394290e-02 1.95567369e-01
6.33141398e-01 6.89922214e-01 -3.58385950e-01 1.07679784e-01
6.92480564e-01 3.80314708e-01 5.51545918e-01 -5.86207032e-01
-2.81437486e-01 -9.28787235e-03 -1.20442522e+00 1.26886475e+00
-1.43820211e-01 4.79091376e-01 -7.23209977e-02 -1.28514457e+00
1.75636959e+00 9.00848329e-01 7.93854058e-01 -7.76650906e-01
3.28675956e-01 3.02340925e-01 4.27547574e-01 -9.60375607e-01
-2.51168400e-01 -4.88887459e-01 4.17402983e-01 1.47318587e-01
-1.48278236e-01 -1.44567207e-01 3.66847128e-01 -2.68250555e-01
6.03713274e-01 4.05541480e-01 8.52998614e-01 -5.92787087e-01
8.87321711e-01 -3.14390272e-01 4.68018323e-01 6.84561670e-01
-4.12861496e-01 3.42744514e-02 5.99891067e-01 -7.83132970e-01
-9.18325365e-01 -7.86650538e-01 -1.10404694e+00 4.61538434e-02
-1.24979042e-01 -2.11238489e-01 -6.54213846e-01 1.27773181e-01
4.06713784e-01 4.14459348e-01 -5.02576292e-01 -4.69758883e-02
-5.99777699e-01 -1.00628269e+00 4.65525955e-01 2.50142068e-01
2.75690049e-01 -6.62393987e-01 -1.24038839e+00 8.36416662e-01
-9.83297527e-02 -7.39939749e-01 3.32450956e-01 7.08309785e-02
-1.38649929e+00 -1.16433787e+00 -7.22289085e-01 -5.32393642e-02
3.92364234e-01 -6.18605435e-01 9.54985857e-01 5.47035411e-02
-8.08750033e-01 4.08887148e-01 -1.01093933e-01 -6.45887852e-01
-5.05347610e-01 -2.78630704e-01 1.58559214e-02 1.43131569e-01
7.06020117e-01 -1.10300279e+00 -8.52611423e-01 2.15406254e-01
-1.10514536e-01 -2.76410401e-01 4.43882972e-01 2.02094153e-01
3.44338208e-01 -3.87970001e-01 9.04948771e-01 -8.73402417e-01
5.93628287e-01 -2.36688614e-01 -5.00745893e-01 -3.01255494e-01
-1.07590485e+00 -1.60146490e-01 6.39372587e-01 -1.88757688e-01
-6.05044007e-01 -2.69956648e-01 -1.89264148e-01 2.14750737e-01
-4.18026149e-01 2.52755940e-01 1.64129004e-01 6.42595291e-02
1.01423049e+00 4.35270637e-01 4.33826745e-01 -4.81192350e-01
-1.51655465e-01 6.28881514e-01 6.29218936e-01 -5.95319033e-01
2.79712856e-01 5.71576394e-02 5.49998105e-01 -1.05903041e+00
-2.87945926e-01 -3.67501765e-01 -6.82999611e-01 -6.34045839e-01
8.07480395e-01 -4.36833411e-01 -1.21048284e+00 2.94151247e-01
-8.44910860e-01 4.45007324e-01 -2.32070938e-01 1.20488667e+00
-4.98651564e-01 7.16336071e-01 -4.92432237e-01 -1.37870884e+00
-7.39840746e-01 -4.84299362e-01 2.02431217e-01 1.33255169e-01
-6.01238966e-01 -9.86285210e-01 2.98311502e-01 -2.63199918e-02
6.68525875e-01 8.79716992e-01 7.33757377e-01 -2.13889331e-01
1.26276210e-01 -6.87262833e-01 1.53079733e-01 6.01843297e-01
2.25076497e-01 4.52309884e-02 -7.39030957e-01 9.08091590e-02
5.96892178e-01 3.09564173e-01 3.78836572e-01 8.46517265e-01
6.50318027e-01 5.47680035e-02 -2.37202704e-01 3.72290045e-01
1.68328154e+00 6.61682189e-01 9.07557786e-01 8.85718614e-02
8.62127170e-02 3.24340463e-01 2.45994240e-01 6.46866500e-01
-2.45801955e-02 5.35509348e-01 -9.35134441e-02 -3.38480860e-01
1.41179159e-01 1.66708335e-01 -5.58400787e-02 8.68648887e-01
-8.92411053e-01 5.16283512e-01 -3.71000022e-01 6.58258125e-02
-1.55149746e+00 -1.08010602e+00 -8.88339996e-01 2.49650335e+00
1.03139496e+00 2.22945452e-01 2.92218119e-01 6.74594581e-01
5.52393675e-01 -1.98408023e-01 -3.61476690e-01 -7.93939114e-01
2.06973985e-01 5.59361279e-01 5.17649233e-01 6.10518336e-01
-6.88217759e-01 -2.27036834e-01 7.27047968e+00 -2.55471259e-01
-1.44800019e+00 -1.53539017e-01 3.78045619e-01 2.82535881e-01
3.30919288e-02 4.86124419e-02 -5.53077519e-01 8.58304918e-01
1.31366718e+00 -4.23359782e-01 4.32303399e-02 6.27257228e-01
6.14868879e-01 -5.37492037e-01 -9.13712084e-01 9.18185174e-01
-1.36029154e-01 -7.84214556e-01 -6.71542108e-01 -4.14013192e-02
-5.65403774e-02 -6.86818242e-01 -5.40092528e-01 1.32697970e-01
-9.47383344e-01 -6.41961575e-01 1.46312386e-01 1.25284553e+00
6.31404221e-01 -4.57687557e-01 8.31307530e-01 1.74279228e-01
-1.01760137e+00 1.70785129e-01 -4.12667364e-01 -4.65408474e-01
2.58764505e-01 1.19495678e+00 -6.99199080e-01 5.51357627e-01
1.32988125e-01 4.86865342e-01 -3.26562285e-01 1.09682024e+00
-1.03988007e-01 9.06953275e-01 -4.85899448e-01 -2.01981321e-01
-3.16675872e-01 -7.50416100e-01 4.62538123e-01 1.05117166e+00
1.28608108e-01 1.53601944e-01 -1.99998379e-01 1.26910055e+00
8.65886092e-01 6.05422199e-01 -4.56946641e-01 3.31477791e-01
2.59347022e-01 1.07945073e+00 -4.23341811e-01 -6.62229478e-01
-3.17809522e-01 2.66300082e-01 -5.26048124e-01 2.93759108e-01
-8.25762331e-01 -7.48801410e-01 4.70093191e-02 3.82095158e-01
-3.11861992e-01 5.56774028e-02 -5.95396876e-01 -7.59718060e-01
2.07090005e-01 -4.19104367e-01 1.93834491e-02 -2.52785802e-01
-6.18530154e-01 3.85272861e-01 2.22771093e-01 -1.08327544e+00
-2.84092396e-01 -4.18521583e-01 -4.75857079e-01 1.66879463e+00
-1.06251502e+00 -2.63753533e-01 -2.28627414e-01 2.45892167e-01
-2.22982794e-01 1.33432388e-01 1.25052190e+00 4.85424638e-01
-2.65949190e-01 1.35395244e-01 -2.31375158e-01 2.51121446e-02
4.79714304e-01 -1.37915480e+00 -1.75870731e-01 5.02213240e-01
-6.06388688e-01 5.86645365e-01 1.02042711e+00 -4.57332909e-01
-1.10601223e+00 -3.99613023e-01 1.37272537e+00 -2.61512637e-01
8.00361112e-02 8.23405161e-02 -7.97887087e-01 -1.24908369e-02
-6.04947060e-02 6.71641976e-02 9.89610255e-01 -8.68328288e-02
2.49269426e-01 -6.46646142e-01 -1.27245545e+00 -2.63899583e-02
1.89271569e-01 -2.21407816e-01 -9.46672916e-01 8.79134536e-02
-1.80148259e-01 -2.37577662e-01 -1.72842014e+00 4.79379684e-01
1.23903894e+00 -1.07768440e+00 7.32703626e-01 -4.99962360e-01
1.51138529e-02 -2.63713181e-01 3.27957869e-01 -9.39328194e-01
-4.72706318e-01 -5.36970913e-01 5.75156212e-02 9.21855867e-01
2.96127379e-01 -1.01955867e+00 3.75402689e-01 8.45484614e-01
2.21741736e-01 -6.97675943e-01 -9.08630788e-01 -4.88997310e-01
-3.65646899e-01 1.00049958e-01 1.50909036e-01 7.82710195e-01
4.46553111e-01 4.39873070e-01 -4.75268394e-01 -2.26105645e-01
8.00252140e-01 1.18486263e-01 5.96206546e-01 -1.62367880e+00
-3.68980795e-01 -1.64005414e-01 -7.62299955e-01 -4.95229781e-01
-5.23352146e-01 -4.30381000e-01 -4.57074255e-01 -1.34035194e+00
-1.71970442e-01 -3.50630581e-01 -5.67706943e-01 -3.09821162e-02
-1.96591437e-01 1.69731006e-01 -5.80232292e-02 -1.04742274e-01
6.27803445e-01 -2.30189621e-01 1.21008813e+00 3.40244293e-01
-7.61021316e-01 3.03244859e-01 -2.96247572e-01 4.07124966e-01
6.84334338e-01 -4.29010689e-01 -2.92638391e-01 6.05517507e-01
1.71450660e-01 5.20104647e-01 4.34829220e-02 -1.15668571e+00
-3.46439362e-01 7.61460513e-02 7.95268774e-01 -4.74512249e-01
1.91055760e-01 -1.10327733e+00 7.55607307e-01 1.08881736e+00
-9.12408680e-02 -6.47335052e-02 -1.37458354e-01 7.77566433e-02
-1.63129121e-01 -4.44792092e-01 9.62882817e-01 -3.06708813e-01
5.95318861e-02 -2.58422494e-01 -7.54138768e-01 -2.44939730e-01
9.67083454e-01 -6.81401730e-01 -2.73406319e-02 -1.70251787e-01
-1.09659052e+00 -3.80360126e-01 -1.08197220e-01 -1.92342311e-01
2.29035676e-01 -1.17756271e+00 -8.59976828e-01 4.60855305e-01
-3.38231802e-01 -6.58016860e-01 4.39453304e-01 1.65065467e+00
-7.66450286e-01 4.86899287e-01 -5.62405825e-01 -6.85249984e-01
-1.15205657e+00 4.45266873e-01 7.82725632e-01 1.36650667e-01
-7.73411989e-01 1.46236524e-01 -5.19551694e-01 3.64664078e-01
1.12838268e-01 -4.36254114e-01 -6.83567941e-01 1.16502605e-01
5.28428197e-01 7.39583910e-01 -2.88902987e-02 -1.04586691e-01
-3.02794397e-01 7.69065320e-01 5.23569643e-01 5.88915683e-02
9.37965691e-01 -1.65603742e-01 -3.59354973e-01 9.25215304e-01
9.71462429e-01 1.86718285e-01 -3.29867512e-01 1.46935567e-01
-3.15754175e-01 -3.85812998e-01 -1.09500185e-01 -8.41724217e-01
-7.41351604e-01 7.00580418e-01 1.06598592e+00 5.65049469e-01
1.17710030e+00 -7.56948292e-01 3.30926716e-01 8.81675482e-02
8.40300024e-02 -1.20690513e+00 -7.62036264e-01 -3.41073513e-01
7.71086216e-01 -4.30257112e-01 3.78043711e-01 -4.57325399e-01
-1.89162016e-01 1.49444294e+00 2.02742130e-01 -3.20008874e-01
9.57658648e-01 2.50844926e-01 3.78242642e-01 2.40144745e-01
-3.66291672e-01 1.99014127e-01 1.37190282e-01 6.46311104e-01
9.87526357e-01 3.07469368e-01 -1.60569525e+00 3.89012337e-01
-1.68089077e-01 6.98413193e-01 5.88351250e-01 7.02326953e-01
-7.24027932e-01 -1.07468188e+00 -6.63379669e-01 6.38559520e-01
-6.57317340e-01 3.05225790e-01 -4.99687903e-02 6.80763900e-01
2.24096060e-01 9.34404612e-01 -5.01425304e-02 7.52712339e-02
7.36239552e-01 6.46850526e-01 7.57967770e-01 -1.70386601e-02
-5.30842900e-01 2.53934205e-01 1.87426582e-01 -2.87046850e-01
-6.65149808e-01 -3.50244433e-01 -1.01668155e+00 -1.64133348e-02
-3.16717565e-01 6.43804848e-01 1.07430863e+00 6.85521841e-01
5.76075204e-02 5.12919664e-01 7.96458066e-01 -4.61262316e-01
-7.86167562e-01 -1.41372907e+00 -1.19089103e+00 3.61417383e-01
1.85442612e-01 -3.82958591e-01 -4.42804396e-01 1.04564376e-01] | [14.082152366638184, 2.993879556655884] |
2907fad8-56a3-4583-9a8e-cc88e2939cc8 | pyramid-architecture-for-multi-scale | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Nie_Pyramid_Architecture_for_Multi-Scale_Processing_in_Point_Cloud_Segmentation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Nie_Pyramid_Architecture_for_Multi-Scale_Processing_in_Point_Cloud_Segmentation_CVPR_2022_paper.pdf | Pyramid Architecture for Multi-Scale Processing in Point Cloud Segmentation | Semantic segmentation of point cloud data is a critical task for autonomous driving and other applications. Recent advances of point cloud segmentation are mainly driven by new designs of local aggregation operators and point sampling methods. Unlike image segmentation, few efforts have been made to understand the fundamental issue of scale and how scales should interact and be fused. In this work, we investigate how to efficiently and effectively integrate features at varying scales and varying stages in a point cloud segmentation network. In particular, we open up the commonly used encoder-decoder architecture, and design scale pyramid architectures that allow information to flow more freely and systematically, both laterally and upward/downward in scale. Moreover, a cross-scale attention feature learning block has been designed to enhance the multi-scale feature fusion which occurs everywhere in the network. Such a design of multi-scale processing and fusion gains large improvements in accuracy without adding much additional computation. When built on top of the popular KPConv network, we see consistent improvements on a wide range of datasets, including achieving state-of-the-art performance on NPM3D and S3DIS. Moreover, the pyramid architecture is generic and can be applied to other network designs: we show an example of similar improvements over RandLANet. | ['Xiaofeng Ren', 'Ling Wang', 'Rui Lan', 'Dong Nie'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['point-cloud-segmentation'] | ['computer-vision'] | [-5.28827943e-02 -1.31838545e-02 -9.09362361e-02 -5.12210250e-01
-6.37162924e-01 -5.47133088e-01 4.97213036e-01 1.89947724e-01
-3.45707148e-01 2.41424978e-01 -1.85031474e-01 -2.85299867e-01
-1.19469419e-01 -9.82706904e-01 -9.55981374e-01 -3.39249581e-01
-1.68591648e-01 5.74216545e-01 7.95675755e-01 -4.30283338e-01
1.87925071e-01 1.10438299e+00 -1.63838732e+00 1.97759464e-01
8.99725378e-01 1.21710730e+00 1.93388075e-01 7.02264130e-01
-4.13127065e-01 2.86060452e-01 -5.19114256e-01 -3.37604582e-01
6.63050652e-01 9.37221125e-02 -7.96253085e-01 1.71042293e-01
6.80457592e-01 -2.81340718e-01 -2.12072819e-01 9.97187138e-01
4.19027328e-01 3.14435847e-02 4.66142803e-01 -1.36580324e+00
-3.45166057e-01 3.38904619e-01 -8.03921402e-01 2.12953880e-01
-3.10605735e-01 2.71951765e-01 8.72041285e-01 -7.73600161e-01
4.20868754e-01 1.36059320e+00 8.65767539e-01 9.78506878e-02
-1.07077074e+00 -7.20679224e-01 2.88632393e-01 1.69408441e-01
-1.33081913e+00 -7.60197788e-02 6.13217056e-01 -4.35262114e-01
1.06600714e+00 1.03308715e-01 9.05941844e-01 3.50179225e-01
1.76564321e-01 8.49750042e-01 7.98486233e-01 -8.56038649e-03
2.43045628e-01 -3.93760242e-02 5.63080609e-02 4.39651132e-01
2.24077404e-01 -1.21617056e-01 -4.37664360e-01 9.57176015e-02
1.10548174e+00 1.25691146e-01 1.13071389e-01 -4.74572062e-01
-1.11805606e+00 9.75509107e-01 1.00953889e+00 1.39449432e-01
-3.08743447e-01 4.97690767e-01 2.98356682e-01 1.81877539e-01
6.71835542e-01 5.10469377e-01 -7.16017306e-01 -3.68038565e-03
-1.31179702e+00 3.64841104e-01 5.60184658e-01 1.22114027e+00
1.10370290e+00 -2.23736599e-01 -9.85268280e-02 6.50551915e-01
2.22860202e-01 5.17253637e-01 -2.98834350e-02 -1.35451889e+00
5.36000013e-01 7.92299628e-01 -1.34030238e-01 -6.93835437e-01
-6.26133382e-01 -4.30437684e-01 -6.02541268e-01 6.17315233e-01
1.67172164e-01 -1.80481285e-01 -1.24404383e+00 1.18289173e+00
5.27819395e-01 3.94886255e-01 -1.20517902e-01 8.38144302e-01
7.24037230e-01 5.88988125e-01 -2.66454294e-02 5.64785123e-01
1.54355192e+00 -1.05330408e+00 -2.44054094e-01 -4.64146733e-01
4.84538376e-01 -7.18651712e-01 8.65290582e-01 2.27972254e-01
-1.23537755e+00 -6.98746026e-01 -1.20619714e+00 -4.97960180e-01
-6.97953939e-01 -1.48035020e-01 8.99939835e-01 3.65104556e-01
-1.31577218e+00 7.89521158e-01 -1.06561780e+00 -3.28086764e-01
8.71936858e-01 6.67691648e-01 -1.33838430e-01 7.72504285e-02
-9.25938725e-01 7.48082042e-01 2.71986395e-01 6.33452833e-02
-3.41161281e-01 -1.01562226e+00 -6.79126740e-01 2.59976815e-02
2.28844643e-01 -9.46712732e-01 1.30770683e+00 -6.20650709e-01
-1.33435428e+00 5.77882767e-01 -1.22314513e-01 -6.70392036e-01
3.87976348e-01 -2.41157696e-01 -4.15567383e-02 3.24906349e-01
3.00328791e-01 1.60606813e+00 6.79014862e-01 -1.08827174e+00
-1.15365243e+00 -5.25539577e-01 2.10578173e-01 4.81295794e-01
-6.54347911e-02 -1.06187247e-01 -9.01925206e-01 -3.54048193e-01
1.71696976e-01 -9.37766731e-01 -5.01187444e-01 2.85678983e-01
-1.00103490e-01 -2.52391636e-01 1.08692455e+00 -3.19495022e-01
6.10645235e-01 -2.17798424e+00 1.45136580e-01 1.68293655e-01
3.21760148e-01 1.35033518e-01 -3.15117836e-02 1.89783663e-01
2.88529307e-01 2.29069665e-01 -3.43121707e-01 -4.84137714e-01
1.01655103e-01 2.91080892e-01 -7.54157081e-02 3.79459679e-01
4.69394535e-01 1.15207171e+00 -5.32923341e-01 -4.14827287e-01
5.48294723e-01 6.37067199e-01 -7.88757324e-01 -2.06102759e-01
-2.86502600e-01 4.12708908e-01 -5.17014027e-01 7.38159776e-01
9.53294039e-01 -4.14095700e-01 -6.18369877e-01 -1.73293918e-01
-4.06249583e-01 1.29991591e-01 -9.72777665e-01 1.91627216e+00
-4.55319017e-01 8.13573718e-01 3.05441618e-01 -6.05324268e-01
7.80705869e-01 -8.80888402e-02 5.77287376e-01 -4.60486054e-01
2.83878803e-01 2.08653554e-01 -3.81035395e-02 -1.15061656e-01
8.58884454e-01 3.17464881e-02 -1.51147889e-02 -2.68901557e-01
-3.16704735e-02 -8.34379315e-01 2.71503240e-01 1.16112515e-01
9.87303555e-01 -1.10230654e-01 -2.04408750e-01 -2.88793385e-01
3.03159893e-01 4.54394966e-01 2.90371776e-01 6.37508869e-01
-1.90999597e-01 7.67315507e-01 5.72475851e-01 -2.29611441e-01
-1.12300289e+00 -8.27036917e-01 -4.81416464e-01 1.04315114e+00
5.54199874e-01 -2.26557061e-01 -7.98898697e-01 -4.66407865e-01
5.52486897e-01 4.49259698e-01 -3.92446607e-01 1.27580389e-01
-4.51334327e-01 -5.22489786e-01 5.44100285e-01 9.48697865e-01
7.86870241e-01 -8.88281703e-01 -6.30896091e-01 9.48028192e-02
2.17381552e-01 -1.39438391e+00 -2.81264395e-01 4.83689576e-01
-1.06755650e+00 -7.82298028e-01 -4.76065695e-01 -7.20692337e-01
4.44453716e-01 4.16737497e-01 1.12925851e+00 -1.54090360e-01
-1.87723145e-01 1.23897091e-01 -2.98057586e-01 -7.09329128e-01
4.56286222e-02 5.38119256e-01 -3.97184193e-01 -3.58516425e-01
4.16160256e-01 -4.69948590e-01 -8.36895704e-01 3.37389857e-01
-8.86343122e-01 4.52692732e-02 6.97790086e-01 3.15284401e-01
6.67635262e-01 7.57512078e-02 3.11308920e-01 -6.71809196e-01
3.90058160e-01 -3.91995609e-01 -8.13677371e-01 -2.15871856e-01
-3.69752645e-01 -1.19143404e-01 2.95771718e-01 9.20486003e-02
-6.14009500e-01 1.15058422e-01 -5.35706937e-01 -6.40971303e-01
-2.30897859e-01 6.92698658e-02 -3.34385931e-02 -4.77184117e-01
3.70492160e-01 -2.72479624e-01 1.72945354e-02 -2.64639437e-01
8.35148811e-01 6.57207191e-01 4.35687453e-01 -4.33200777e-01
7.39769459e-01 9.05987620e-01 7.32439682e-02 -1.01072276e+00
-5.03304183e-01 -6.89723790e-01 -9.67668056e-01 -1.10017233e-01
1.26212239e+00 -1.10132730e+00 -6.46991670e-01 4.87886041e-01
-1.21272147e+00 -4.34114069e-01 -6.51026726e-01 1.87035501e-01
-6.00244224e-01 -1.57661587e-02 -6.26505971e-01 -1.70009226e-01
-2.46148661e-01 -1.59129250e+00 1.66369939e+00 4.19511110e-01
1.26740217e-01 -9.55917895e-01 -4.49924320e-01 3.48571926e-01
6.09591603e-01 1.20856211e-01 5.98725498e-01 -3.78389329e-01
-1.02128422e+00 -7.74457529e-02 -6.98509514e-01 4.15500313e-01
-6.69602975e-02 2.12617461e-02 -9.62955475e-01 -1.85067728e-01
-1.89403340e-01 -3.85039374e-02 1.00136971e+00 6.69439137e-01
1.27632785e+00 3.54922205e-01 -4.48774338e-01 1.10201573e+00
1.26958513e+00 6.16875216e-02 5.21984696e-01 3.87089431e-01
9.36988533e-01 5.48432648e-01 4.45538759e-01 6.69414178e-02
6.69437051e-01 4.75646466e-01 6.57993674e-01 -5.29837489e-01
-2.77246177e-01 1.14485677e-02 -5.32410294e-02 5.34663379e-01
1.57363713e-01 -4.35593389e-02 -9.96975005e-01 5.48716545e-01
-1.75296307e+00 -5.08570671e-01 -2.78716564e-01 1.71879733e+00
5.40971935e-01 3.33352566e-01 -1.32966163e-02 -2.24365026e-01
5.32485127e-01 3.19043815e-01 -7.66723871e-01 -4.44563299e-01
-4.44518067e-02 3.54709119e-01 1.27703524e+00 4.29264694e-01
-1.29408491e+00 1.28722167e+00 6.82800722e+00 1.00609624e+00
-1.29641962e+00 9.58545581e-02 6.26523316e-01 -2.30831608e-01
-2.83730030e-01 -1.64492782e-02 -1.08267093e+00 1.99102715e-01
7.35663772e-01 2.00562835e-01 2.41089448e-01 9.46168244e-01
1.25156119e-01 -1.20201021e-01 -9.78208601e-01 8.53650928e-01
-2.60436743e-01 -1.59526169e+00 -2.12802753e-01 1.72190577e-01
7.95246899e-01 8.63519847e-01 -3.31451930e-02 2.28862479e-01
3.79310817e-01 -7.92340159e-01 7.91205108e-01 2.05811530e-01
6.78758264e-01 -8.20767343e-01 5.69170833e-01 1.67797014e-01
-1.50434983e+00 -6.28886074e-02 -5.44040680e-01 1.03064040e-02
3.38406235e-01 6.05879366e-01 -6.06737375e-01 4.53886360e-01
9.71394718e-01 7.48652756e-01 -6.77981734e-01 1.12346768e+00
-1.11706078e-01 2.69076914e-01 -6.90963507e-01 1.47124171e-01
6.37025535e-01 -1.81191355e-01 3.44053388e-01 1.21865141e+00
2.90945023e-01 -2.04431415e-01 1.58240587e-01 8.50582123e-01
-1.23328649e-01 -1.26141697e-01 -6.16088271e-01 2.57853538e-01
4.91302222e-01 1.37833810e+00 -1.09451020e+00 -5.13997555e-01
-7.18733966e-01 6.97333932e-01 2.69289672e-01 2.37452447e-01
-9.18156624e-01 -3.99673879e-01 1.24113297e+00 3.20622653e-01
8.30936670e-01 -5.41863501e-01 -8.08486640e-01 -7.38353670e-01
-1.81090429e-01 -3.77760112e-01 -3.96014974e-02 -7.79286563e-01
-1.17687118e+00 5.27471721e-01 1.30737334e-01 -1.11094987e+00
3.19002837e-01 -6.53250337e-01 -5.88521481e-01 8.41916502e-01
-1.90979838e+00 -1.09264851e+00 -4.26412165e-01 5.17859161e-01
7.65390277e-01 1.95274279e-01 5.59238717e-02 4.88249063e-01
-3.47431093e-01 2.09235385e-01 -3.06152292e-02 -7.70425424e-02
4.24171716e-01 -1.34385955e+00 9.46407795e-01 8.31438422e-01
-1.11773945e-01 4.10266310e-01 3.54312152e-01 -7.25663662e-01
-1.23847985e+00 -1.46732605e+00 3.31856728e-01 -5.13280630e-01
7.18952358e-01 -4.99482304e-01 -8.48672092e-01 6.34065211e-01
2.89394289e-01 2.04116851e-01 1.87495798e-01 -4.41642776e-02
-4.86680381e-02 -2.25465372e-01 -1.09439886e+00 4.96948302e-01
1.13724113e+00 -3.63818556e-01 -4.21176970e-01 3.22362989e-01
1.25482690e+00 -7.74001658e-01 -9.62634325e-01 7.38127410e-01
1.66677698e-01 -1.04627669e+00 1.11225927e+00 -1.84510455e-01
3.73956323e-01 -5.27878821e-01 -9.97715890e-02 -1.18673396e+00
-5.55417001e-01 -2.94295400e-01 2.18240917e-01 9.97038186e-01
4.18028533e-01 -7.48226285e-01 1.08110130e+00 4.82846349e-01
-6.89872265e-01 -9.35557008e-01 -9.31447387e-01 -6.75491273e-01
3.04925114e-01 -7.41445661e-01 1.02165306e+00 5.49770236e-01
-5.30698597e-01 1.30629361e-01 2.28495494e-01 5.74947834e-01
5.23438871e-01 1.48596466e-01 9.79406536e-01 -1.17727363e+00
6.78997934e-02 -8.08569133e-01 -6.96054220e-01 -1.56190240e+00
-1.99679554e-01 -1.04609406e+00 4.50995043e-02 -1.94031823e+00
-3.25946599e-01 -6.20130062e-01 5.40898256e-02 3.76522511e-01
-5.50525226e-02 4.65372801e-01 4.14987713e-01 3.13137561e-01
-5.65843940e-01 4.99909580e-01 1.43342805e+00 -5.84172234e-02
-2.62625843e-01 -7.98399299e-02 -7.16239452e-01 8.76318276e-01
8.23850095e-01 -3.01267207e-01 -3.91748071e-01 -7.12545633e-01
4.86054830e-02 -1.93611979e-01 3.83027703e-01 -1.25984776e+00
4.91688132e-01 -2.22051106e-02 4.10708994e-01 -1.10168183e+00
6.09741509e-01 -9.18313861e-01 -6.95046829e-03 1.31226972e-01
1.04853898e-01 2.08221793e-01 5.03236592e-01 4.27079707e-01
-3.45606893e-01 4.99849282e-02 8.07207048e-01 -8.85776356e-02
-9.69720900e-01 5.81912935e-01 6.39209012e-03 -6.00267425e-02
1.25609279e+00 -4.89409924e-01 -3.56493473e-01 1.19940393e-01
-5.50859690e-01 7.30435908e-01 7.11420238e-01 5.58790863e-01
3.60485464e-01 -1.18381619e+00 -4.74928141e-01 3.88366282e-01
2.36209258e-02 8.47746074e-01 2.30611369e-01 9.55418348e-01
-1.03496015e+00 3.74596775e-01 -8.75494480e-02 -1.14145601e+00
-7.65939772e-01 2.54940957e-01 4.08012420e-01 -2.71429997e-02
-8.49563181e-01 1.08312476e+00 2.10416541e-01 -4.54812050e-01
1.02732904e-01 -9.37022746e-01 -2.81225685e-02 1.62802860e-01
1.18746579e-01 2.91176707e-01 3.08767885e-01 -6.26940429e-01
-3.33503932e-01 9.88875568e-01 3.89533639e-02 1.26886144e-01
1.32178867e+00 -1.13463707e-01 -2.00895712e-01 2.43368611e-01
1.28839278e+00 -2.36168668e-01 -1.71965969e+00 -2.21158341e-02
-1.99961632e-01 -3.79130065e-01 2.47003660e-01 -3.68845522e-01
-1.45121086e+00 8.99627090e-01 4.97483253e-01 4.27127808e-01
1.00574565e+00 2.98772663e-01 9.95106816e-01 3.66819687e-02
3.80024850e-01 -1.12194312e+00 -3.32913041e-01 6.23575211e-01
6.60553813e-01 -1.33652282e+00 4.16987278e-02 -6.87806785e-01
-5.63345909e-01 9.17181671e-01 5.53757071e-01 -4.39529002e-01
8.04033279e-01 6.46966815e-01 4.94243093e-02 -6.05839372e-01
-5.26404142e-01 -4.39829260e-01 2.31176808e-01 4.42426801e-01
9.42074135e-02 1.02938473e-01 6.77005500e-02 1.54422104e-01
-4.73176330e-01 -5.67772798e-02 1.68585926e-01 9.03477669e-01
-6.88519537e-01 -9.79418278e-01 -4.35883075e-01 6.77618325e-01
-1.46869868e-01 -1.27275882e-03 -8.07479545e-02 8.91296268e-01
4.97714072e-01 7.52758861e-01 4.17081833e-01 -3.41640711e-01
5.38238168e-01 -1.21398062e-01 2.11365789e-01 -5.67695081e-01
-6.72814906e-01 3.35317850e-02 -2.82703459e-01 -7.37882555e-01
-3.30571979e-01 -6.74177468e-01 -1.65844774e+00 -3.06560636e-01
-3.50964665e-01 -4.10547890e-02 1.10218751e+00 8.08810711e-01
7.21300125e-01 9.26227987e-01 3.90770674e-01 -1.47171247e+00
-1.60768926e-01 -6.41519547e-01 -5.76543212e-01 3.22215445e-02
3.57754081e-01 -5.81477344e-01 -2.82893568e-01 -2.32963160e-01] | [7.993488311767578, -3.240955352783203] |
c00762b9-43b4-4183-9417-5e936a6d68f4 | smart-sentiment-analysis-based-search-engine | 2306.09777 | null | https://arxiv.org/abs/2306.09777v1 | https://arxiv.org/pdf/2306.09777v1.pdf | Smart Sentiment Analysis-based Search Engine Classification Intelligence | Search engines are widely used for finding information on the internet. However, there are limitations in the current search approach, such as providing popular but not necessarily relevant results. This research addresses the issue of polysemy in search results by implementing a search function that determines the sentimentality of the retrieved information. The study utilizes a web crawler to collect data from the British Broadcasting Corporation (BBC) news site, and the sentimentality of the news articles is determined using the Sentistrength program. The results demonstrate that the proposed search function improves recall value while accurately retrieving nonpolysemous news. Furthermore, Sentistrength outperforms deep learning and clustering methods in classifying search results. The methodology presented in this article can be applied to analyze the sentimentality and reputation of entities on the internet. | ['Mike Nkongolo'] | 2023-06-16 | null | null | null | null | ['clustering', 'classification-1', 'sentiment-analysis'] | ['methodology', 'methodology', 'natural-language-processing'] | [-8.17947865e-01 -3.73300195e-01 -5.44677377e-01 1.18584700e-01
-4.43496227e-01 -7.50474513e-01 6.66109264e-01 5.10603845e-01
-7.68736064e-01 3.93034726e-01 1.84882358e-01 -2.22543344e-01
-6.25749171e-01 -9.92857575e-01 -3.06421548e-01 -3.66501153e-01
2.08768263e-01 5.35071611e-01 -3.59624438e-02 -6.34323120e-01
1.00098944e+00 1.37709320e-01 -1.82491028e+00 5.33594012e-01
9.33899701e-01 1.31488037e+00 4.01644081e-01 1.81879416e-01
-4.19439703e-01 6.17224753e-01 -9.99332130e-01 -7.77614236e-01
1.13717187e-02 -4.63331752e-02 -8.87837648e-01 -4.28284079e-01
1.46449834e-01 2.32043769e-02 -1.07358597e-01 1.20814347e+00
4.22587484e-01 -3.52117582e-04 6.88083649e-01 -8.71921718e-01
-9.50181842e-01 6.43691659e-01 -3.94876570e-01 7.51482546e-01
3.90730888e-01 -9.38564301e-01 1.71615577e+00 -7.61739075e-01
5.33701718e-01 9.51249003e-01 4.07673627e-01 -1.71260536e-01
-3.77239853e-01 -8.11591685e-01 -3.41723502e-01 3.93235236e-01
-1.37668836e+00 3.78440768e-02 7.94537485e-01 -5.00031114e-01
1.16481543e+00 1.65524006e-01 8.50536346e-01 8.83108974e-01
5.16608238e-01 5.67256868e-01 1.07555318e+00 -5.96001983e-01
3.79470825e-01 4.71671969e-01 4.62660223e-01 3.50638360e-01
5.98376632e-01 -2.76512295e-01 -5.58036685e-01 -3.98768276e-01
8.87016952e-02 8.16877559e-02 1.64822862e-01 2.15293974e-01
-7.58624315e-01 1.22788560e+00 4.85440850e-01 7.90254235e-01
-6.20433450e-01 -4.67076212e-01 5.22160113e-01 5.64513803e-01
7.41710186e-01 1.10777807e+00 -4.96490777e-01 -1.75634488e-01
-1.04430532e+00 1.31834611e-01 1.17791259e+00 7.27502108e-01
7.17147768e-01 -3.95885676e-01 2.23763555e-01 9.02896643e-01
4.56323653e-01 5.83469391e-01 1.23598850e+00 -8.48601937e-01
4.27611321e-01 9.22315776e-01 1.68947607e-01 -1.89255345e+00
-4.51540917e-01 -6.91267490e-01 -1.12266980e-01 -5.79880416e-01
-2.74572730e-01 -1.95650414e-01 -4.72345352e-01 9.15590286e-01
-1.43953100e-01 -5.16143322e-01 -1.83374081e-02 7.45940030e-01
1.31476271e+00 5.22630155e-01 7.00475127e-02 1.56477336e-02
1.37583172e+00 -7.26541758e-01 -8.24092507e-01 -8.18419456e-02
4.29295093e-01 -1.01374245e+00 8.79705608e-01 3.76270652e-01
-8.75753641e-01 -2.76708364e-01 -8.77603948e-01 1.08242095e-01
-1.13624489e+00 5.15714660e-02 8.07956696e-01 7.33469725e-01
-1.14437735e+00 4.71636891e-01 -3.20221424e-01 -3.73847574e-01
-1.82053283e-01 2.21935719e-01 3.51384692e-02 4.39555049e-01
-1.74318719e+00 1.11074376e+00 5.55637538e-01 -3.10158402e-01
-1.18332833e-01 -1.94493055e-01 -5.03417552e-01 4.29731935e-01
-1.33544870e-03 -2.57387847e-01 1.02615583e+00 -1.12087047e+00
-1.02675533e+00 9.03300285e-01 1.20996654e-01 -4.47244197e-01
-4.97238860e-02 4.96083684e-02 -8.75359178e-01 4.25884277e-01
5.57030797e-01 -1.57687403e-02 2.27865532e-01 -9.31386650e-01
-1.04875815e+00 -3.26564252e-01 1.05700403e-01 4.33540970e-01
-9.60691154e-01 3.32790613e-01 -6.93935215e-01 -3.92718464e-01
1.14791788e-01 -6.24096692e-01 3.06741208e-01 -1.19158113e+00
-1.88917935e-01 -6.91733956e-01 2.58754760e-01 -5.59343874e-01
1.60455775e+00 -1.88374817e+00 -3.90829176e-01 6.15411639e-01
3.49713147e-01 1.48499414e-01 1.81181848e-01 7.11351693e-01
3.64702672e-01 5.19017875e-01 8.65440726e-01 2.22235397e-01
1.34898856e-01 -2.24229306e-01 -7.29318112e-02 1.50580734e-01
-7.34402359e-01 8.98847103e-01 -9.43778098e-01 -8.07404876e-01
-2.68175840e-01 1.13708533e-01 -2.38577515e-01 -6.08124375e-01
1.25560984e-01 -5.10555506e-01 -9.12960649e-01 1.04244828e+00
3.11247587e-01 -6.48592055e-01 1.72115892e-01 -1.33054420e-01
-2.19549835e-01 5.94940484e-01 -4.35613215e-01 1.04976428e+00
-5.56881249e-01 7.59309351e-01 -3.19342017e-01 -8.99624169e-01
9.07104135e-01 1.44642651e-01 5.88686526e-01 -1.29125404e+00
5.08600533e-01 5.44368148e-01 -1.80687502e-01 -6.48640156e-01
1.04392791e+00 2.36205488e-01 -2.64838129e-01 4.78010088e-01
-1.47708565e-01 1.84366629e-01 4.73681092e-01 3.35605919e-01
9.23577607e-01 -5.39630115e-01 2.10678101e-01 -4.08962667e-01
2.56488383e-01 5.29705882e-01 2.04519778e-01 8.00590754e-01
-1.02881212e-02 1.56344891e-01 1.80701420e-01 -1.71474069e-01
-7.99117088e-01 -4.88571256e-01 -3.56843889e-01 1.27187228e+00
3.50865871e-01 -3.61453086e-01 -7.66616464e-01 -5.00210702e-01
1.89592019e-01 6.72638893e-01 -5.75480700e-01 -7.72691369e-02
1.10081643e-01 -6.96133137e-01 3.77531618e-01 -2.40693800e-02
6.33348167e-01 -1.09666264e+00 -5.44195175e-01 4.39089686e-02
-6.57082617e-01 -8.64377201e-01 -1.49266303e-01 -5.85595891e-02
-5.98737895e-01 -9.06605959e-01 -4.87132668e-01 -1.05410075e+00
3.58042568e-01 5.55852771e-01 1.35445869e+00 3.94485623e-01
1.99149519e-01 4.52039838e-01 -7.89315760e-01 -3.39878500e-01
-2.03592002e-01 6.03201091e-01 -1.21666938e-01 -3.82869422e-01
1.33740497e+00 -1.36583121e-02 -1.00619113e+00 3.33698779e-01
-6.88870192e-01 -7.94949293e-01 5.13928056e-01 5.67435384e-01
1.81607351e-01 1.01464593e+00 9.37666714e-01 -6.80632234e-01
1.53410709e+00 -1.06879699e+00 -4.55319226e-01 2.92881757e-01
-1.52422476e+00 -4.21712190e-01 3.91414315e-01 -1.90230474e-01
-9.26589966e-01 -6.99182510e-01 1.91438884e-01 1.57804742e-01
3.31453025e-01 1.29663408e+00 5.08554399e-01 -2.07660645e-01
7.36243784e-01 8.79158005e-02 -1.18603818e-01 -5.12662530e-01
-1.41850084e-01 1.31726742e+00 -9.37652364e-02 -9.77318361e-02
1.71410009e-01 4.44355637e-01 -5.37712336e-01 -6.12304986e-01
-1.02242589e+00 -1.05505574e+00 -6.83560669e-02 -4.19597566e-01
5.80398440e-01 -1.01558256e+00 -8.27781916e-01 1.10251874e-01
-7.17773795e-01 7.06996083e-01 2.33373299e-01 7.47685671e-01
4.79989685e-02 3.68584752e-01 -6.82482302e-01 -5.71909487e-01
-6.97966039e-01 -9.07375097e-01 6.52438164e-01 2.68955082e-01
-3.65539044e-01 -1.20105326e+00 1.97501704e-01 9.21269417e-01
5.78970551e-01 -3.32716823e-01 8.52831244e-01 -1.50375068e+00
2.28497237e-02 -8.69072974e-01 1.83398381e-03 2.47892633e-01
4.44136672e-02 -6.53485358e-02 -5.55599034e-01 1.82308070e-02
2.54880011e-01 -1.31273657e-01 7.11452961e-01 3.83848667e-01
7.00942457e-01 -4.94073331e-01 -4.25254494e-01 -3.07092424e-02
1.53194249e+00 4.47452903e-01 3.94565374e-01 1.32965064e+00
1.30935431e-01 7.01043248e-01 6.76221967e-01 4.46727782e-01
7.20708013e-01 1.92952529e-01 1.26391232e-01 3.68586481e-01
3.33182573e-01 -2.67174721e-01 2.42267493e-02 1.04133308e+00
-9.65454355e-02 -4.38304365e-01 -1.10011530e+00 6.37335002e-01
-1.42334759e+00 -1.26570725e+00 1.67774841e-01 1.82351172e+00
7.48497605e-01 2.99898349e-02 -2.57860571e-02 8.87260213e-02
8.56511176e-01 -1.31552398e-01 -2.55220056e-01 -4.10879105e-01
-1.98731422e-01 -5.16041815e-02 6.69835448e-01 1.59832180e-01
-9.14221883e-01 8.18569779e-01 6.64338255e+00 9.07166779e-01
-8.83662462e-01 -9.35896020e-03 5.01899779e-01 5.74702732e-02
-5.97324371e-01 -2.00312331e-01 -7.36065328e-01 9.14466977e-01
8.67871761e-01 -6.67636871e-01 2.73970544e-01 1.17036963e+00
1.36231482e-01 -2.22402543e-01 -4.58443791e-01 8.26236486e-01
5.65743864e-01 -1.38846719e+00 1.06036291e-02 2.15438530e-01
8.71437907e-01 2.68294334e-01 3.11449230e-01 4.02741522e-01
3.12522918e-01 -5.35192192e-01 4.21976507e-01 4.42116976e-01
-3.40501107e-02 -7.59589136e-01 1.09276378e+00 4.04389888e-01
-5.37224948e-01 -3.64429981e-01 -4.36285645e-01 5.98340901e-03
-3.32811594e-01 6.29848897e-01 -8.58065426e-01 2.04072177e-01
1.21409023e+00 7.05913126e-01 -7.67609894e-01 1.19923353e+00
2.23561376e-01 4.53910768e-01 -1.70797884e-01 -1.00256693e+00
3.49457234e-01 -3.55776280e-01 3.57620656e-01 1.18173480e+00
5.05398333e-01 -3.41485977e-01 -2.57319868e-01 6.00776196e-01
-2.57235259e-01 6.76619232e-01 -5.23562789e-01 -3.28949124e-01
8.23924005e-01 1.28398478e+00 -1.14685774e+00 -3.21754634e-01
-5.97676396e-01 5.41003704e-01 -1.23499192e-01 3.48939776e-01
-4.28592563e-01 -6.00935221e-01 -1.57669976e-01 4.98046875e-02
2.93354779e-01 1.90793619e-01 -4.30343926e-01 -1.00378418e+00
-1.92486204e-03 -9.28231180e-01 6.57295108e-01 -7.94954598e-01
-1.54111576e+00 6.61058009e-01 -3.55023034e-02 -1.24055040e+00
-9.41339582e-02 -3.85241687e-01 -2.27184653e-01 7.18226075e-01
-1.56908631e+00 -6.39913261e-01 -2.14249790e-01 4.71377879e-01
3.18138301e-01 -8.59554410e-01 5.01954257e-01 5.08040249e-01
-1.17940366e-01 1.32393330e-01 6.21931136e-01 1.40840679e-01
3.47474664e-01 -1.07762396e+00 4.36683558e-02 3.31916630e-01
7.36851171e-02 1.04219592e+00 5.84294319e-01 -9.52318013e-01
-1.11958408e+00 -7.27497041e-02 1.42612231e+00 -5.03530204e-01
1.16316867e+00 2.90302515e-01 -2.39041150e-01 2.13927418e-01
6.42241716e-01 -9.22813296e-01 1.19839978e+00 3.78275454e-01
-1.87051490e-01 -5.35338745e-02 -1.16128707e+00 5.31070769e-01
1.43414676e-01 -5.48453689e-01 -9.59941208e-01 5.31909227e-01
4.91230458e-01 9.30548981e-02 -9.24863815e-01 3.41901369e-02
6.72049522e-01 -7.84362376e-01 1.05002081e+00 -3.60730439e-01
4.67969805e-01 1.90812007e-01 -1.99901029e-01 -1.34288013e+00
-3.16990882e-01 1.10004619e-02 -5.18461540e-02 1.00531256e+00
9.83705103e-01 -8.52795362e-01 7.69797623e-01 8.28132570e-01
3.01612884e-01 -3.94855440e-01 -8.08177292e-01 -4.87805009e-01
3.09715897e-01 -1.58870704e-02 5.46443701e-01 9.70479250e-01
5.58325350e-01 4.15694952e-01 2.76656687e-01 -2.76398808e-01
2.43632779e-01 4.13463473e-01 7.60718957e-02 -1.68996751e+00
2.90625811e-01 -7.77392268e-01 -2.93785095e-01 -6.09718204e-01
2.69657105e-01 -1.03873765e+00 -3.96603048e-01 -1.64181495e+00
3.39860529e-01 -2.85085559e-01 -7.60469794e-01 -1.52091496e-02
1.95644259e-01 1.65890958e-02 -1.39765903e-01 9.00390625e-01
-9.87607002e-01 1.31317973e-01 1.00952113e+00 -1.90795064e-01
4.05049175e-02 2.77365416e-01 -1.21903265e+00 7.23441601e-01
9.97194290e-01 -4.93799567e-01 -3.11729133e-01 -4.80839789e-01
1.38504338e+00 -3.23851734e-01 -9.21738297e-02 -6.64706230e-01
6.13091290e-01 9.84480008e-02 2.99781740e-01 -6.72942281e-01
1.12140454e-01 -1.03266025e+00 -1.39376402e-01 4.30204198e-02
-6.12510741e-01 6.09224319e-01 -3.13040078e-01 5.16543865e-01
-6.52745843e-01 -7.52892077e-01 1.47045851e-01 -3.81043702e-01
-6.07081652e-01 -4.56078239e-02 -6.14641964e-01 1.78424254e-01
7.20117807e-01 -1.98434606e-01 -4.65408236e-01 -6.17933214e-01
-2.06803337e-01 1.25957534e-01 3.12900424e-01 5.97168565e-01
5.80316246e-01 -1.08973038e+00 -4.48179454e-01 -2.21368015e-01
-5.64651713e-02 -7.41700947e-01 -1.52031824e-01 4.57627624e-01
-7.53674269e-01 7.99942195e-01 -3.10279950e-02 1.45847453e-02
-9.78120387e-01 5.10537028e-01 8.69218558e-02 -7.43215950e-03
-1.67030111e-01 6.25569344e-01 -6.11570477e-01 -2.43513495e-01
3.33319455e-01 3.17357033e-01 -1.18846488e+00 9.63046730e-01
3.58435810e-01 6.27074599e-01 3.55465442e-01 -8.48983109e-01
-2.20876306e-01 4.86932039e-01 -2.81040251e-01 -2.03930259e-01
1.18780482e+00 -5.68704665e-01 -5.58084190e-01 1.79127872e-01
1.41426384e+00 1.29669592e-01 4.64319251e-02 -1.84659109e-01
4.85143065e-01 -3.65566194e-01 5.38745046e-01 -8.89895618e-01
-1.15397251e+00 1.98500931e-01 4.10644054e-01 1.13887227e+00
9.25378859e-01 9.04812664e-02 7.72482097e-01 9.78458941e-01
3.29153180e-01 -2.12717700e+00 -4.34844047e-02 6.25869811e-01
4.19040263e-01 -1.22837365e+00 9.54383910e-02 3.15694898e-01
-9.03895020e-01 1.08265495e+00 2.09656566e-01 1.77709207e-01
1.16704226e+00 -1.64723247e-01 3.50463122e-01 -9.71374035e-01
-5.15141368e-01 -2.02056408e-01 4.43002015e-01 6.30929247e-02
6.22011840e-01 -8.94881412e-02 -1.28257632e+00 5.37064552e-01
-7.72896469e-01 -2.82453209e-01 4.17155862e-01 8.34239721e-01
-5.94070435e-01 -5.10466695e-01 -1.46065295e-01 8.39104116e-01
-1.26313353e+00 -5.15510380e-01 -6.58968031e-01 5.50533473e-01
-2.67024100e-01 1.62500262e+00 8.25057626e-02 -6.71471238e-01
-4.14997004e-02 -7.51942545e-02 -6.17373288e-01 -3.38011414e-01
-8.90471458e-01 1.94131747e-01 2.06954062e-01 -1.82257041e-01
-7.51840532e-01 -6.13543689e-01 -9.08241987e-01 -2.96296090e-01
-8.84116590e-01 1.05315793e+00 1.28812051e+00 8.73742163e-01
6.48518860e-01 6.19886704e-02 9.12244737e-01 -1.19333550e-01
-3.10784370e-01 -9.84352350e-01 -8.08607221e-01 4.67618912e-01
-1.79785967e-03 -7.04540074e-01 -9.01947975e-01 -4.56501633e-01] | [10.6874418258667, 6.913955211639404] |
9587fecb-c34f-4879-966a-cdde28fed08f | an-empirical-evaluation-of-word-embedding | null | null | https://ieeexplore.ieee.org/document/9392437 | https://ieeexplore.ieee.org/document/9392437 | An Empirical Evaluation of Word Embedding Models for Subjectivity Analysis Tasks | It is a clearly established fact that good categorization results are heavily dependent on representation techniques. Text representation is a necessity that must be fulfilled before working on any text analysis task since it creates a baseline which even advanced machine learning models fail to compensate. This paper aims to comprehensively analyze and quantitatively evaluate the various models to represent text in order to perform Subjectivity Analysis. We implement a diverse array of models on the Cornell Subjectivity Dataset. It is worth noting that the
BERT Language Model gives much better results than any other model but is significantly computationally expensive than the
other approaches. We obtained state-of-the-art results on the subjectivity task by fine-tuning the BERT Language Model. This
can open up a lot of new avenues and potentially lead to a specialized model inspired by BERT dedicated to subjectivity analysis. | ['Shashank Shekhar', 'Priya Kamath', 'Geetha Maiya', 'Ritika Nandi'] | 2021-04-06 | null | null | null | ieee-international-conference-on-advances-in | ['subjectivity-analysis'] | ['natural-language-processing'] | [-6.31621620e-03 2.59769917e-01 -4.20843333e-01 -5.52055359e-01
-5.85998476e-01 -4.32631701e-01 1.10888577e+00 6.85833395e-01
-3.22316945e-01 3.04197341e-01 6.00638211e-01 -4.60665196e-01
-3.94724101e-01 -8.36145937e-01 1.30489677e-01 -3.19699466e-01
1.50243700e-01 8.47980499e-01 -9.28434804e-02 -7.83628643e-01
7.34145641e-01 1.47409901e-01 -1.88322639e+00 5.21302581e-01
7.46752918e-01 8.56546283e-01 2.50172932e-02 4.12611425e-01
-3.43271285e-01 1.10757244e+00 -7.38880754e-01 -6.47191286e-01
-1.17728375e-01 -3.60421091e-01 -1.21314776e+00 1.36567399e-01
4.59392905e-01 2.32965142e-01 3.54509577e-02 1.00020647e+00
5.12889326e-01 2.49180973e-01 1.14744580e+00 -9.52983260e-01
-7.66176641e-01 7.83160210e-01 -3.09626341e-01 4.35005248e-01
5.37534177e-01 -5.85935831e-01 1.11912847e+00 -6.62232280e-01
6.28877938e-01 1.37376165e+00 5.15722871e-01 4.87375110e-01
-1.01255083e+00 -1.88208148e-01 3.54912549e-01 4.27125782e-01
-1.13815916e+00 -3.71549994e-01 9.54306126e-01 -5.76552629e-01
8.38939428e-01 5.11192203e-01 1.95752069e-01 1.33778441e+00
-1.33384407e-01 7.16934800e-01 1.40136468e+00 -8.47954869e-01
8.44640806e-02 7.80061364e-01 8.99278224e-01 9.25824165e-01
9.78998914e-02 -2.56502330e-01 -3.69328201e-01 -1.13910072e-01
1.61363915e-01 -6.41688630e-02 -9.71400663e-02 -3.59620810e-01
-9.56868231e-01 1.20862937e+00 3.95945311e-01 7.95082986e-01
-7.34697655e-02 -2.63542444e-01 6.77295327e-01 7.66338646e-01
7.11476743e-01 6.34685338e-01 -3.09209317e-01 -1.06915951e-01
-1.01869273e+00 2.55804896e-01 8.10410202e-01 4.55413997e-01
5.71601748e-01 -1.73526391e-01 -3.62264156e-01 1.34240913e+00
2.36138031e-01 1.02619089e-01 8.58722866e-01 -7.42300570e-01
4.72417116e-01 1.15092421e+00 -1.75158113e-01 -1.04897714e+00
-7.89740384e-01 -8.48539710e-01 -9.03792024e-01 3.09342057e-01
5.73669970e-01 3.27709258e-01 -7.17785001e-01 1.29523790e+00
-1.45461962e-01 -7.55284190e-01 -1.45701364e-01 7.23865390e-01
1.17891490e+00 4.73640949e-01 1.54332176e-01 -2.10683927e-01
1.55119598e+00 -9.12910581e-01 -8.27816784e-01 3.94790812e-04
9.24582839e-01 -8.88281226e-01 1.30890810e+00 6.38450325e-01
-1.13292146e+00 -4.39813524e-01 -9.36500847e-01 -2.70536661e-01
-9.15856957e-01 2.43296608e-01 8.88040781e-01 1.25134873e+00
-1.06012762e+00 4.54331458e-01 -3.79263073e-01 -6.80977464e-01
3.67710888e-01 2.62463689e-01 -2.17696980e-01 2.53363132e-01
-1.09147823e+00 1.16067100e+00 1.83850750e-01 -2.49459237e-01
-5.45967996e-01 -2.15371683e-01 -6.67945623e-01 1.79092690e-01
1.74379125e-01 -6.56785369e-01 1.10864961e+00 -1.27646017e+00
-1.43692088e+00 1.48829424e+00 -3.26127857e-01 -2.80044526e-01
7.46261060e-01 3.29815418e-01 -4.38260853e-01 1.95387341e-02
4.40724418e-02 1.85998574e-01 7.11434782e-01 -1.37067997e+00
-4.03897136e-01 -4.68932182e-01 2.38337085e-01 -8.62478465e-02
-9.66169298e-01 3.58173519e-01 -1.78292826e-01 -8.34443629e-01
2.12823465e-01 -7.22627342e-01 7.87777081e-02 -2.39938542e-01
-7.03372732e-02 -7.01175988e-01 4.41834360e-01 -3.52725297e-01
1.46731174e+00 -1.80086315e+00 2.06875220e-01 2.46355489e-01
4.39396441e-01 3.39918941e-01 -8.28168541e-02 6.41970813e-01
-2.34728187e-01 4.28354055e-01 8.50800797e-02 -5.59012473e-01
3.85058761e-01 -1.24831170e-01 -5.25459647e-01 3.81388694e-01
-2.31450930e-01 7.23120213e-01 -6.47685230e-01 -5.60094595e-01
2.13447824e-01 3.84126455e-01 -5.51116943e-01 -1.40814215e-01
-7.35862702e-02 -1.09690959e-02 -4.92694288e-01 5.69472551e-01
6.05123520e-01 -4.65521663e-01 2.51621217e-01 2.63127148e-01
-1.61470398e-01 5.77156186e-01 -7.03332484e-01 1.46712720e+00
-6.25598311e-01 6.74610376e-01 -1.47469059e-01 -1.41164792e+00
1.17438006e+00 3.87038916e-01 2.97501028e-01 -8.56390595e-01
5.94668746e-01 2.10010365e-01 9.30652581e-03 -5.65508664e-01
8.11971188e-01 -3.17175984e-01 -3.19083571e-01 7.01119006e-01
1.19109556e-01 4.23093699e-03 6.26584172e-01 4.00839329e-01
8.19493711e-01 -3.20745677e-01 4.10254180e-01 -7.61083961e-01
1.05992234e+00 -4.38689813e-02 -5.04144728e-02 8.63733470e-01
-3.41087937e-01 6.70971334e-01 7.48606086e-01 -6.79173708e-01
-7.01591074e-01 -8.92203391e-01 -4.51457620e-01 1.65477538e+00
-1.82381153e-01 -6.07882023e-01 -4.88564640e-01 -1.03682268e+00
-2.22078264e-01 6.40626848e-01 -1.00464642e+00 -4.80958493e-03
-1.77192718e-01 -1.11473942e+00 3.15873891e-01 2.68578261e-01
4.72860068e-01 -8.08973074e-01 3.06054601e-03 5.99742075e-03
-4.45472956e-01 -6.34321272e-01 1.85083359e-01 2.42321894e-01
-9.57748294e-01 -8.80766332e-01 -7.85776675e-01 -1.03375232e+00
6.32320523e-01 3.53110403e-01 1.38812399e+00 4.96947229e-01
1.16169430e-01 1.88858792e-01 -7.49572277e-01 -6.42154038e-01
-6.76070988e-01 6.56802654e-01 -1.87602147e-01 -3.41465294e-01
6.31028295e-01 -4.03720587e-01 -3.14106047e-01 4.68865782e-01
-7.97730625e-01 -2.78388709e-01 1.89793691e-01 7.69374490e-01
-1.60180986e-01 2.48001173e-01 8.96970630e-01 -1.03903496e+00
8.24205637e-01 -4.56985325e-01 -2.12449342e-01 3.35045010e-01
-6.26874328e-01 9.02092010e-02 7.26378858e-01 -1.41116723e-01
-1.19102740e+00 -3.11844200e-01 -3.55197251e-01 5.29199839e-01
-5.51626049e-02 5.87751150e-01 7.25846440e-02 2.93024648e-02
1.17220771e+00 -1.39471993e-01 -2.46189281e-01 -6.58807576e-01
1.42626867e-01 1.23936808e+00 2.45088376e-02 -6.28278136e-01
5.59196770e-01 6.47802114e-01 -2.71031767e-01 -1.01667929e+00
-1.28935468e+00 -7.67320454e-01 -5.74329019e-01 -1.59184888e-01
3.82832229e-01 -7.83482015e-01 -7.75547743e-01 -1.99980587e-02
-9.53605592e-01 -7.86830857e-02 3.53710055e-02 8.74226168e-02
-5.22267103e-01 5.75468004e-01 -4.08071905e-01 -9.25303578e-01
-2.04176560e-01 -6.75778449e-01 8.39738131e-01 -1.85540989e-01
-6.85220599e-01 -1.38934100e+00 1.10121943e-01 1.00910830e+00
4.87482846e-01 -3.20952535e-01 1.19894755e+00 -5.26871622e-01
1.10109456e-01 -2.62771100e-01 -2.69394219e-01 2.44549751e-01
-1.91157445e-01 -1.05298728e-01 -1.22687018e+00 -1.97010890e-01
1.54617727e-01 -5.34374893e-01 1.48185337e+00 8.09114501e-02
1.28592885e+00 -1.90870792e-01 -1.90953404e-01 1.92573115e-01
1.17104340e+00 -3.22261423e-01 7.49670923e-01 7.84132242e-01
2.47303098e-01 8.31945240e-01 3.15942883e-01 3.99156481e-01
6.51872277e-01 7.39291131e-01 2.58159488e-01 -5.91035858e-02
-2.06726715e-01 -1.08524092e-01 1.84077978e-01 9.08336639e-01
-3.56311858e-01 -5.18871307e-01 -9.50485349e-01 5.05455971e-01
-1.84152734e+00 -1.10988116e+00 -5.28383732e-01 2.04681110e+00
6.35504127e-01 8.84870365e-02 2.95528293e-01 8.86571348e-01
5.86632073e-01 2.59384632e-01 3.55231494e-01 -8.46343994e-01
-2.74890810e-01 6.24623261e-02 -7.86300898e-02 5.90839624e-01
-1.27707040e+00 9.16700542e-01 7.15818834e+00 7.24101961e-01
-1.03710127e+00 3.16781521e-01 3.96832287e-01 -8.63155574e-02
-1.58748329e-01 -4.25479591e-01 -7.19377160e-01 2.56974578e-01
8.23934555e-01 -3.14546138e-01 8.51763040e-02 8.30252826e-01
1.97291449e-01 -1.52957112e-01 -7.95271635e-01 9.35225308e-01
4.74553436e-01 -9.52252090e-01 7.60736540e-02 -3.64188626e-02
9.85814512e-01 -2.01371118e-01 2.82561511e-01 7.30867147e-01
2.06532121e-01 -9.51576769e-01 6.30879462e-01 2.41471440e-01
2.51102656e-01 -4.76736158e-01 9.03392434e-01 3.49827468e-01
-6.39883757e-01 -3.98376942e-01 -6.07304692e-01 -6.01974130e-01
-1.04273200e-01 6.26250863e-01 -4.83174562e-01 4.50194299e-01
5.33222318e-01 6.42309189e-01 -1.11860013e+00 8.34120750e-01
-9.55301747e-02 6.06888771e-01 5.84526770e-02 -3.07231247e-01
1.58577263e-01 1.89752012e-01 2.80635208e-01 1.19614077e+00
1.75701171e-01 -2.59176522e-01 -5.12459166e-02 5.13213396e-01
-1.94178075e-01 5.21014214e-01 -3.36340100e-01 -1.40003771e-01
-2.07943350e-01 1.32821560e+00 -9.81249511e-01 -6.84183180e-01
-4.07648116e-01 7.37802982e-01 6.33021772e-01 1.78940147e-01
-5.01059711e-01 -3.24095219e-01 1.63691625e-01 2.40119189e-01
-2.90644821e-02 -2.44589418e-01 -7.50211060e-01 -1.43869245e+00
-4.11189534e-02 -8.90180767e-01 7.52369225e-01 -4.81160581e-01
-1.52558184e+00 6.32324874e-01 -1.52877241e-01 -1.11650467e+00
-2.27250531e-01 -8.32999468e-01 -3.07674527e-01 5.98969877e-01
-1.63118303e+00 -1.02858484e+00 -5.13108194e-01 4.20886517e-01
5.59554040e-01 -4.30900902e-01 1.10425270e+00 3.54588628e-01
-4.77025360e-01 5.20094097e-01 8.74538049e-02 1.17609926e-01
8.12553704e-01 -1.22487104e+00 -8.06520432e-02 4.19700384e-01
2.24010214e-01 7.09465742e-01 9.86580014e-01 -2.40652099e-01
-8.04533422e-01 -8.19157541e-01 1.58594990e+00 -9.99067903e-01
7.49555528e-01 -3.48459154e-01 -8.76422465e-01 5.13601899e-01
2.37993211e-01 -5.40108681e-01 8.63286138e-01 7.35469401e-01
-5.46299100e-01 3.94359119e-02 -8.15404177e-01 4.90700424e-01
1.01659727e+00 -5.00610054e-01 -1.21567786e+00 6.46538258e-01
-3.34549472e-02 1.71886206e-01 -5.15285909e-01 3.91202308e-02
3.27790886e-01 -1.25180984e+00 9.02709782e-01 -6.14202619e-01
8.05793881e-01 9.05541927e-02 4.45862785e-02 -1.19293141e+00
-4.41595048e-01 -3.82222921e-01 1.13781877e-01 1.37942135e+00
5.09554148e-01 -6.72442973e-01 8.16914976e-01 4.68122184e-01
1.85088105e-02 -2.10924834e-01 -6.69900894e-01 -9.63663280e-01
4.88613486e-01 -4.32733893e-01 3.30873847e-01 1.31650209e+00
7.42628276e-01 7.39178658e-01 -1.73343405e-01 -4.77983743e-01
4.04968262e-01 1.99409351e-01 5.77000618e-01 -1.70391166e+00
2.10915521e-01 -1.07187557e+00 -5.33116758e-01 -8.02931547e-01
5.56474566e-01 -1.52401352e+00 -3.96600425e-01 -1.82667863e+00
3.33552271e-01 -3.97094905e-01 -3.19174647e-01 4.12335932e-01
-1.64885268e-01 7.14918494e-01 -7.53577873e-02 1.97628945e-01
-8.38173985e-01 5.65969229e-01 9.73389387e-01 -2.97854155e-01
-2.31710866e-01 6.77114725e-02 -1.18105769e+00 8.35304141e-01
1.07992458e+00 -4.65726852e-01 -3.16920340e-01 -2.78580934e-01
6.61849141e-01 -3.47154975e-01 2.68308014e-01 -8.40309858e-01
6.38592914e-02 2.06316218e-01 2.31215164e-01 -3.33706915e-01
2.61852115e-01 -6.69622779e-01 -6.72242224e-01 2.93223053e-01
-6.79362893e-01 -2.53025815e-02 -8.17606598e-02 1.21283248e-01
-3.76753181e-01 -5.74648380e-01 7.46370375e-01 -3.39377433e-01
-5.19603729e-01 -2.11603016e-01 -7.59331524e-01 7.45143518e-02
5.07854104e-01 5.60813919e-02 -6.14485145e-01 -4.08316135e-01
-8.28645647e-01 -2.58490536e-02 4.66907978e-01 7.09065735e-01
-2.23272238e-02 -1.06586599e+00 -6.99488759e-01 7.55347833e-02
5.36828935e-01 -1.03365278e+00 2.77859390e-01 7.36657619e-01
-3.55168253e-01 7.60806680e-01 -2.21058279e-01 -4.17887241e-01
-1.45306206e+00 7.93242455e-01 9.68881324e-02 -5.20082295e-01
-3.21139663e-01 4.82905895e-01 -8.98014158e-02 -6.11713827e-01
1.94513962e-01 5.04704043e-02 -9.26513314e-01 5.01811743e-01
7.21946180e-01 7.92401373e-01 4.84596580e-01 -7.66592622e-01
-3.31769437e-01 6.10850036e-01 -1.80389434e-01 -2.49397561e-01
1.18598962e+00 -3.91863078e-01 -2.33089194e-01 5.37238657e-01
1.21520853e+00 6.15500547e-02 -3.26954931e-01 -1.19252212e-01
1.53546721e-01 -5.37230730e-01 4.60553288e-01 -1.14670372e+00
-7.43915379e-01 1.19458985e+00 5.41533709e-01 9.24290597e-01
9.13473427e-01 -1.03608677e-02 5.00791259e-02 6.10699952e-01
2.16263562e-01 -1.50076556e+00 1.86140284e-01 7.37814009e-01
1.06202841e+00 -1.60398817e+00 2.51281530e-01 -5.75182498e-01
-7.00790763e-01 1.15164864e+00 9.30492654e-02 -1.02108821e-01
6.46923542e-01 -1.73980385e-01 5.30686975e-01 -3.58443439e-01
-6.91960752e-01 -6.22500360e-01 5.40127754e-01 6.03849590e-01
1.07424021e+00 -1.68360993e-02 -8.91367316e-01 5.00749230e-01
-5.67075372e-01 -5.60200447e-03 6.14582241e-01 5.56989253e-01
-5.54651082e-01 -1.45667410e+00 -3.82112771e-01 3.91508847e-01
-6.68474257e-01 9.24434289e-02 -7.47089624e-01 7.49554813e-01
-8.68580788e-02 1.47478735e+00 -1.20105512e-01 -2.17127025e-01
2.81786054e-01 2.52276480e-01 4.00790602e-01 -6.71516061e-01
-1.06809866e+00 -2.66008079e-01 3.60505611e-01 -2.13896796e-01
-5.00825822e-01 -6.01349235e-01 -8.23069692e-01 -4.89061058e-01
-6.20073490e-02 3.79788339e-01 9.05263245e-01 7.57127583e-01
-4.06010263e-02 5.32135129e-01 6.04147553e-01 -6.70922279e-01
-5.13596654e-01 -1.26732755e+00 -6.96436942e-01 5.23485005e-01
1.73370183e-01 -7.13319898e-01 -7.21231520e-01 -6.62165880e-02] | [10.371785163879395, 8.763903617858887] |
e4940bcd-ac7a-4568-a07d-72a7321bdc65 | semi-supervised-formality-style-transfer | 2010.05090 | null | https://arxiv.org/abs/2010.05090v1 | https://arxiv.org/pdf/2010.05090v1.pdf | Semi-supervised Formality Style Transfer using Language Model Discriminator and Mutual Information Maximization | Formality style transfer is the task of converting informal sentences to grammatically-correct formal sentences, which can be used to improve performance of many downstream NLP tasks. In this work, we propose a semi-supervised formality style transfer model that utilizes a language model-based discriminator to maximize the likelihood of the output sentence being formal, which allows us to use maximization of token-level conditional probabilities for training. We further propose to maximize mutual information between source and target styles as our training objective instead of maximizing the regular likelihood that often leads to repetitive and trivial generated responses. Experiments showed that our model outperformed previous state-of-the-art baselines significantly in terms of both automated metrics and human judgement. We further generalized our model to unsupervised text style transfer task, and achieved significant improvements on two benchmark sentiment style transfer datasets. | ['Diyi Yang', 'Kunal Chawla'] | 2020-10-10 | null | https://aclanthology.org/2020.findings-emnlp.212 | https://aclanthology.org/2020.findings-emnlp.212.pdf | findings-of-the-association-for-computational | ['formality-style-transfer', 'semi-supervised-formality-style-transfer'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.88487828e-01 4.10358638e-01 -1.95785761e-02 -9.91990328e-01
-1.06619298e+00 -7.89120376e-01 6.91342592e-01 -1.19893827e-01
-5.74693680e-01 1.06734121e+00 2.40483657e-01 -4.22665805e-01
5.41861475e-01 -8.19625020e-01 -7.24550486e-01 -2.21756771e-01
5.34912229e-01 5.63657761e-01 -1.22329779e-01 -3.50084424e-01
2.76287913e-01 -7.28678033e-02 -7.98493624e-01 7.70054758e-01
1.22527063e+00 6.23556256e-01 1.94930911e-01 6.02036417e-01
-4.17346209e-01 8.36662531e-01 -7.25731134e-01 -8.19336474e-01
-3.93637829e-03 -7.22340524e-01 -1.10489690e+00 -1.01984501e-01
5.18602312e-01 -1.05794974e-01 2.40847185e-01 1.08864546e+00
2.69294411e-01 8.35693628e-02 9.67608035e-01 -9.48766112e-01
-1.00746953e+00 7.36688256e-01 -2.97826558e-01 -1.58817500e-01
4.05168474e-01 1.31198913e-01 1.44014513e+00 -1.03013706e+00
5.34767568e-01 1.65183961e+00 5.71488261e-01 9.97794688e-01
-1.74357688e+00 -7.66496181e-01 2.22756878e-01 -3.54557842e-01
-8.69992673e-01 -3.44679385e-01 7.97367990e-01 -2.92389780e-01
9.30334628e-01 -8.75568204e-03 1.52828708e-01 1.33859360e+00
1.59956843e-01 1.12385285e+00 1.59270406e+00 -6.96924806e-01
7.24821910e-02 5.27226925e-01 6.86997175e-02 6.25570834e-01
-6.18946701e-02 -1.49874642e-01 -5.04751921e-01 2.34387238e-02
4.95448142e-01 -4.51131791e-01 -4.56841849e-02 1.41406849e-01
-1.16724396e+00 1.07346237e+00 3.89037818e-01 2.34249637e-01
-7.35402852e-02 -1.05940243e-02 2.73881584e-01 5.73612511e-01
7.27767110e-01 9.00661230e-01 -6.70259893e-01 -1.74368173e-01
-7.84842134e-01 1.73971951e-01 1.10411775e+00 1.07582963e+00
9.23380375e-01 -1.34236172e-01 -5.25130928e-01 1.06593263e+00
2.21329778e-01 6.79569304e-01 4.40524995e-01 -9.19982195e-01
6.88098609e-01 7.76724100e-01 1.02310866e-01 -5.19809783e-01
4.46913205e-02 -1.74535215e-01 -6.83625281e-01 -1.16958067e-01
4.34823513e-01 -4.66584504e-01 -7.28212595e-01 2.01957679e+00
-1.63751647e-01 -4.85987693e-01 2.17333242e-01 5.65326631e-01
5.81340373e-01 8.35892737e-01 2.96537638e-01 -3.54623124e-02
1.14773357e+00 -1.04162669e+00 -5.03813446e-01 -4.89522636e-01
9.26141024e-01 -8.63455117e-01 1.90494585e+00 2.89616823e-01
-9.85802114e-01 -5.46952128e-01 -7.28549242e-01 -1.23176932e-01
-6.39190823e-02 1.83985472e-01 6.04104877e-01 6.05626702e-01
-1.10354626e+00 7.18698382e-01 -5.09028494e-01 -3.48434776e-01
4.19291198e-01 1.44855857e-01 -2.18493670e-01 3.39927860e-02
-1.25471485e+00 7.76324570e-01 1.72636971e-01 -2.45216727e-01
-5.17583072e-01 -8.25932264e-01 -9.74927843e-01 6.72721444e-03
2.01751161e-02 -8.83934498e-01 1.60320425e+00 -1.49719775e+00
-1.96636128e+00 1.20929027e+00 -4.81524140e-01 -1.47142157e-01
4.34986502e-01 -5.55974007e-01 -1.44269854e-01 -1.13255814e-01
3.13011974e-01 1.07373786e+00 6.34553552e-01 -1.08192348e+00
-5.33838689e-01 1.17686152e-01 2.28199884e-01 3.18070054e-01
-6.89393520e-01 5.05883582e-02 -9.89185870e-02 -8.17723215e-01
-4.08215761e-01 -1.06453598e+00 -1.70339257e-01 -2.14262307e-01
-2.50663131e-01 -7.21910119e-01 2.88808465e-01 -5.03752112e-01
1.02181888e+00 -1.94975078e+00 3.10423821e-01 -3.99984093e-03
-2.20078140e-01 1.94143504e-01 -3.37860405e-01 3.15750927e-01
2.44214192e-01 3.41783524e-01 -4.45144743e-01 -5.61481357e-01
1.82186618e-01 2.38048390e-01 -6.43629551e-01 -2.81200022e-01
6.41709924e-01 1.05386901e+00 -1.10175765e+00 -4.55906987e-01
-3.13241452e-01 7.11682066e-02 -8.50921154e-01 6.46006584e-01
-4.77391034e-01 4.91335005e-01 -2.94447452e-01 1.13407746e-01
3.10887933e-01 -3.18867534e-01 1.86821565e-01 2.20727667e-01
2.79653966e-01 9.05341685e-01 -5.71684182e-01 1.77266896e+00
-1.01251340e+00 3.79989177e-01 -3.38951439e-01 -6.16343141e-01
1.04912972e+00 1.69950783e-01 -2.96784282e-01 -5.42928159e-01
8.36983509e-03 2.95406610e-01 -1.24782130e-01 -1.71578419e-03
5.46293795e-01 -5.77258527e-01 -5.91050208e-01 7.73566961e-01
1.65721223e-01 -6.95171535e-01 2.48482540e-01 3.86604041e-01
7.29356110e-01 4.41457987e-01 1.02000035e-01 -6.49860024e-01
6.85636580e-01 -2.41307274e-01 6.33115411e-01 7.35110879e-01
9.54541340e-02 5.55513561e-01 7.96218514e-01 -4.95950729e-02
-7.22189128e-01 -1.22090089e+00 1.83470353e-01 1.39835143e+00
-2.13400349e-01 -3.45920533e-01 -9.57341373e-01 -1.23594999e+00
-1.27206728e-01 1.11772668e+00 -6.06538832e-01 -1.67687744e-01
-6.13992631e-01 -3.88560504e-01 5.89006662e-01 6.67205811e-01
3.08035553e-01 -1.35290360e+00 1.72166284e-02 1.47927403e-01
-4.13465381e-01 -1.09649789e+00 -9.23196077e-01 5.28386347e-02
-8.87579918e-01 -5.24071276e-01 -7.27745652e-01 -9.37007368e-01
8.65318537e-01 -2.04933777e-01 1.49850607e+00 -6.55265599e-02
2.70071954e-01 7.44992346e-02 -3.68357062e-01 -6.23726904e-01
-8.10696185e-01 3.72471362e-01 1.27491895e-02 4.65672128e-02
5.68970978e-01 -3.37461680e-01 -3.84510279e-01 1.96593329e-01
-8.73816311e-01 2.84384131e-01 6.43632352e-01 1.02961528e+00
2.68320680e-01 -7.38927364e-01 8.58801782e-01 -1.43044090e+00
9.81202781e-01 -1.77486725e-02 -3.19616050e-01 3.20003390e-01
-6.47869289e-01 4.78613466e-01 9.61746693e-01 -3.46060842e-01
-1.38605368e+00 -8.79034922e-02 -1.77666306e-01 8.03647041e-02
-3.33854616e-01 3.74790400e-01 -1.63954020e-01 2.71171927e-01
6.81027710e-01 1.40144199e-01 -1.96026027e-01 -3.46952409e-01
4.41369236e-01 9.77869630e-01 2.58447468e-01 -1.10986054e+00
7.90837765e-01 6.43427856e-03 -2.63417840e-01 -6.91975594e-01
-1.55690956e+00 -1.06998950e-01 -6.36206090e-01 9.45853516e-02
7.94624448e-01 -8.70125115e-01 -3.92759472e-01 4.41091508e-01
-1.48447978e+00 -7.58493483e-01 -2.47817859e-01 2.28395626e-01
-5.27962923e-01 3.78522635e-01 -9.04693902e-01 -6.03699088e-01
-4.98790413e-01 -8.81176293e-01 1.13074231e+00 1.99449614e-01
-8.91593814e-01 -1.32264864e+00 1.58976167e-01 3.38045597e-01
5.25563002e-01 -1.50654510e-01 1.06839859e+00 -5.80683053e-01
-1.18960217e-01 7.23027438e-02 -2.88634896e-01 8.15797985e-01
2.24405929e-01 -2.21290514e-01 -1.02824044e+00 -2.42991835e-01
-1.98362499e-01 -9.23967540e-01 8.30972910e-01 -1.09368511e-01
9.01116729e-01 -4.30942327e-01 7.10758492e-02 5.37959933e-01
1.08780897e+00 -1.69427291e-01 6.41401768e-01 6.39539957e-02
6.87492132e-01 8.47384214e-01 6.72881961e-01 1.13895759e-02
4.78005946e-01 4.80411559e-01 -4.01748657e-01 -1.86524436e-01
-7.56239891e-02 -6.14913464e-01 7.97444344e-01 8.97635043e-01
3.00997555e-01 -2.43821636e-01 -5.79702914e-01 5.87410927e-01
-1.64268255e+00 -7.59648800e-01 8.30379489e-04 1.87632954e+00
1.55331147e+00 3.54516923e-01 -1.53930355e-02 -3.18316847e-01
6.34954035e-01 -7.94030055e-02 -1.77187443e-01 -9.32934761e-01
4.00745943e-02 6.47133172e-01 4.87159146e-03 7.58337617e-01
-9.82976437e-01 1.55776846e+00 6.22421932e+00 6.90383494e-01
-1.05701149e+00 4.90986370e-02 7.34338462e-01 3.01283039e-02
-6.74216151e-01 8.76819156e-03 -9.58013058e-01 4.40244228e-01
9.16817725e-01 -1.70212895e-01 3.92217517e-01 7.10336745e-01
7.17332065e-02 1.13495722e-01 -1.60168028e+00 5.10798097e-01
1.06783263e-01 -8.00355256e-01 3.71356934e-01 -2.01445565e-01
1.05842185e+00 -2.77220249e-01 5.76731339e-02 7.16825426e-01
7.80077577e-01 -1.00881898e+00 7.42467046e-01 2.81276286e-01
9.71463382e-01 -5.95361173e-01 5.85897326e-01 3.29344302e-01
-8.23563635e-01 2.98002303e-01 -2.91433543e-01 -4.03213561e-01
2.51480609e-01 4.48381245e-01 -9.79207397e-01 3.79181981e-01
4.78758335e-01 7.49758363e-01 -5.42289674e-01 1.98853120e-01
-9.53253210e-01 9.82277513e-01 -1.37704566e-01 -4.50437188e-01
2.98072308e-01 -2.16390401e-01 2.52547950e-01 1.56202400e+00
1.31456867e-01 -1.13610737e-01 2.26577103e-01 1.17471743e+00
-4.61733401e-01 2.48318642e-01 -6.11127794e-01 -1.30444244e-01
2.16963008e-01 1.36181223e+00 -4.60849285e-01 -4.68600214e-01
-4.87864941e-01 1.37471366e+00 8.10997605e-01 4.18524057e-01
-4.80285704e-01 -7.32214391e-01 4.79883999e-01 -1.45250872e-01
2.02415392e-01 -2.13032607e-02 -6.91421032e-01 -1.46356177e+00
1.59846261e-01 -8.85422766e-01 3.36075783e-01 -6.45483911e-01
-1.77375877e+00 6.29566550e-01 -2.05430880e-01 -1.00852001e+00
-3.33557218e-01 -7.64831901e-01 -1.00825071e+00 1.29590642e+00
-1.67378247e+00 -1.25505996e+00 1.12685241e-01 3.63207430e-01
6.68575048e-01 -2.26796478e-01 1.02058601e+00 -8.52886736e-02
-2.43246719e-01 1.04763222e+00 -3.39360893e-01 3.39544088e-01
1.37150443e+00 -1.62884176e+00 6.10483170e-01 8.26523304e-01
2.37604789e-02 8.79641950e-01 7.01640010e-01 -5.89759648e-01
-8.46578538e-01 -1.27245390e+00 1.62625432e+00 -6.77115381e-01
7.39596486e-01 -5.71593881e-01 -7.73516059e-01 7.53279209e-01
5.37880957e-01 -5.21760106e-01 8.76438498e-01 3.66033643e-01
-5.03418088e-01 -2.46585105e-02 -9.39998448e-01 7.71441400e-01
1.03443253e+00 -6.88961387e-01 -1.01049733e+00 2.65794933e-01
8.41680646e-01 -1.54212385e-01 -6.11330271e-01 2.49038383e-01
3.62323791e-01 -5.47264218e-01 4.78700995e-01 -8.44642162e-01
9.86674130e-01 4.65567932e-02 1.20382262e-02 -1.78849518e+00
-4.32845533e-01 -6.78462625e-01 3.14315706e-01 1.57738769e+00
8.97489667e-01 -5.33039093e-01 6.13194466e-01 6.40146375e-01
-1.14634022e-01 -6.07947886e-01 -4.51772839e-01 -7.50866890e-01
6.12261772e-01 -1.00613192e-01 3.64061266e-01 7.78141737e-01
2.38221720e-01 9.76008236e-01 -2.14192092e-01 -2.74165303e-01
4.37805116e-01 3.49021763e-01 8.02413583e-01 -1.06738544e+00
-6.54026568e-01 -4.99219000e-01 1.66491717e-01 -1.37953317e+00
7.31325507e-01 -1.35086858e+00 3.28879505e-01 -1.38710916e+00
2.69861251e-01 -3.43253046e-01 -2.20163465e-01 5.51257074e-01
-6.81209743e-01 3.80708128e-01 1.64492995e-01 -1.80206835e-01
-7.03884125e-01 6.52542591e-01 1.29850233e+00 -2.12246418e-01
-2.83081293e-01 4.60180864e-02 -1.12615013e+00 6.41732872e-01
8.11349571e-01 -5.42062938e-01 -3.37382376e-01 -7.07567096e-01
4.18185264e-01 -2.26014271e-01 1.78648625e-02 -4.51129079e-01
-2.62658954e-01 -2.81518698e-01 1.46020457e-01 -8.70550498e-02
-3.70821096e-02 -3.76877725e-01 -6.86242342e-01 2.76638299e-01
-9.59959686e-01 -3.96223925e-02 1.22735403e-01 2.75722355e-01
-2.31193542e-01 -2.76884913e-01 8.04624856e-01 -1.27836719e-01
-3.34591508e-01 1.42196817e-02 -4.06185180e-01 5.31621695e-01
6.88945234e-01 2.60124922e-01 -1.44331172e-01 -4.57229555e-01
-3.04945230e-01 3.32841784e-01 6.03493869e-01 6.05763018e-01
4.03944105e-01 -1.46023560e+00 -9.47668672e-01 8.68312493e-02
2.85811484e-01 -5.51648214e-02 -2.87336320e-01 4.60116088e-01
-2.82527030e-01 4.02336806e-01 -3.08392774e-02 -5.58359921e-01
-1.05222905e+00 6.45206496e-02 1.21955544e-01 -6.30666673e-01
-1.73233852e-01 1.06108999e+00 5.56129336e-01 -1.02638507e+00
1.11647760e-02 -3.35282058e-01 8.98748487e-02 -2.83825725e-01
4.77480471e-01 -8.09448659e-02 -2.28766687e-02 -2.87112027e-01
-1.67627409e-01 4.21761245e-01 -3.23039949e-01 -4.36983645e-01
1.21304715e+00 -8.65682811e-02 -2.39768952e-01 5.95654368e-01
1.27460408e+00 4.29816768e-02 -1.26178670e+00 -5.10394394e-01
1.73173100e-01 -1.73382714e-01 -3.19175035e-01 -1.03917956e+00
-4.22204226e-01 1.20397151e+00 -1.12497553e-01 -8.54670480e-02
7.82071888e-01 -4.79719555e-03 9.67038929e-01 6.60957932e-01
2.74854064e-01 -1.20883107e+00 2.64251798e-01 1.08172822e+00
7.82037437e-01 -1.34456849e+00 -5.22957444e-01 -4.07525063e-01
-1.01858866e+00 1.05734289e+00 9.46465671e-01 -2.06800520e-01
2.44102955e-01 2.43731499e-01 3.51949334e-01 3.07396382e-01
-8.65497828e-01 1.50703296e-01 5.03580034e-01 3.17523062e-01
1.12242198e+00 1.97603434e-01 -4.47173744e-01 7.85567939e-01
-4.81433481e-01 5.53819612e-02 2.73054957e-01 7.89969742e-01
-3.04145128e-01 -1.57295668e+00 6.40831143e-02 3.97137374e-01
-5.37802994e-01 -4.72894758e-01 -7.40788400e-01 2.53480464e-01
-3.63942981e-01 1.01641226e+00 4.70185354e-02 -2.18457222e-01
4.20995057e-01 3.81230831e-01 6.79433584e-01 -1.16216207e+00
-7.49505877e-01 -1.42858282e-01 3.11966032e-01 -3.50636333e-01
-1.90464333e-01 -5.76125264e-01 -1.15265524e+00 -1.01968125e-01
-6.58410490e-02 2.68284261e-01 2.45709747e-01 1.18669987e+00
1.68495148e-01 4.11926180e-01 8.13307524e-01 -7.80565679e-01
-9.59069252e-01 -1.26129568e+00 -2.42889792e-01 9.22958493e-01
2.66183559e-02 -7.99286962e-02 -3.44149739e-01 3.83016407e-01] | [11.527729034423828, 9.47980785369873] |
24dd2821-2bab-4489-a1d3-4ab9b8b5fa3d | vibration-control-of-a-rotating-cantilever | 2004.11703 | null | http://arxiv.org/abs/2004.11703v1 | http://arxiv.org/pdf/2004.11703v1.pdf | Vibration control of a rotating cantilever beam using piezoelectric actuator and feedback linearization method | The vibration of various structures such as blades of turbines, helicopters,
and all kinds of rotating robot arms can damage the structures and disrupt
their performance and balance. Thus, investigation of the reduction and control
of vibration of these structures is significant. In this paper, the coupled and
nonlinear vibration of a rotating beam in the presence of the piezoelectric
layer is considered. The Von Karman strain displacements are used for deriving
the potential equation. The nonlinear equations of motion are obtained based on
the Hamilton principle, and two ordinary differential equations are derived by
the Assume Mode method. Besides, a feedback linearization control strategy is
proposed for suppressing the vibration of the rotating beam. According to the
equation of motion, flexural and torsional vibrations are coupled. Therefore,
control of flexural mode can result in control of the whole system. | [] | 2020-03-23 | null | null | null | null | ['cantilever-beam'] | ['miscellaneous'] | [-1.15966298e-01 3.29393059e-01 2.88299322e-01 5.69451690e-01
5.56614518e-01 -6.01118267e-01 -1.32683322e-01 -9.08125818e-01
1.19040301e-02 7.05980957e-01 -1.98369905e-01 6.59968331e-03
-2.92226940e-01 -5.19731581e-01 -3.46752286e-01 -1.42902350e+00
1.36891112e-01 -5.49695909e-01 1.83397427e-01 -5.08533478e-01
4.06416267e-01 4.04513687e-01 -1.30900288e+00 -3.68477374e-01
5.63841641e-01 6.82965755e-01 5.46577752e-01 2.98938423e-01
6.05034947e-01 2.51271188e-01 -6.22639894e-01 3.70307863e-01
-9.14367586e-02 -3.97571713e-01 -6.16903245e-01 3.00974548e-01
-6.53223753e-01 -7.18172729e-01 -6.91673830e-02 9.68366981e-01
7.81086981e-01 5.05512476e-01 8.22733760e-01 -8.49829912e-01
-6.96438968e-01 4.51015115e-01 -8.09970677e-01 -1.09808989e-01
6.65180758e-02 -1.91009551e-01 7.10335851e-01 -1.05455351e+00
3.78531575e-01 9.71271098e-01 5.21581471e-01 6.67521477e-01
-7.73521185e-01 -2.00728938e-01 -3.92426819e-01 1.43219277e-01
-1.27762198e+00 -7.48823630e-04 9.69184816e-01 -5.77160716e-01
6.91196322e-01 3.57081532e-01 5.46098769e-01 3.24828535e-01
8.22306037e-01 7.70672485e-02 3.06431204e-01 -5.23369551e-01
-6.66173324e-02 -2.88494945e-01 -1.80570066e-01 5.01451075e-01
6.93776965e-01 2.55996704e-01 -1.76725034e-02 8.69205594e-02
9.10704732e-01 -5.55931441e-02 -6.82823479e-01 -6.76008984e-02
-7.03181267e-01 4.50817585e-01 2.72045910e-01 6.77858829e-01
-2.01128542e-01 -9.83910635e-02 3.22788116e-03 -2.88451109e-02
7.32606277e-02 4.39737618e-01 2.78241914e-02 1.41927391e-01
-2.23029032e-01 2.16537967e-01 6.00041628e-01 4.16905344e-01
1.66861504e-01 5.37269354e-01 2.76237667e-01 4.32688802e-01
4.56044286e-01 9.66261327e-01 3.77785593e-01 -1.02764523e+00
2.10329741e-02 2.00350329e-01 4.75357413e-01 -1.12696862e+00
-6.60175323e-01 -3.19483548e-01 -9.92866576e-01 2.18686327e-01
4.21704389e-02 -9.52869952e-01 -2.99974293e-01 1.33064783e+00
6.78349614e-01 -2.89828062e-01 1.60671562e-01 1.29056883e+00
6.44649684e-01 9.14655268e-01 -8.09090734e-01 -9.95165944e-01
1.08798707e+00 -4.40849811e-01 -1.26193523e+00 2.39710808e-01
4.81271833e-01 -8.29478502e-01 6.43721104e-01 3.60646337e-01
-1.29327834e+00 -4.87441957e-01 -9.07218397e-01 3.11674118e-01
2.08967596e-01 1.98235810e-01 -4.48666185e-01 1.58875227e-01
-7.59791791e-01 6.22799337e-01 -9.97941792e-01 3.39621633e-01
-8.09422076e-01 1.37780145e-01 9.03621595e-03 5.12486756e-01
-1.14851701e+00 1.19414914e+00 1.26332432e-01 9.62226093e-01
-3.67202945e-02 -4.02925909e-01 -2.93481648e-01 6.00465126e-02
4.46693748e-01 -6.80011928e-01 1.17776608e+00 3.01721264e-02
-1.93628299e+00 4.86120284e-02 -3.68318744e-02 4.80590254e-01
2.46251985e-01 -3.11199486e-01 3.00019290e-02 3.93415689e-01
-9.61907022e-03 -3.84152323e-01 9.09057796e-01 -1.00462365e+00
-1.10950895e-01 -7.82283470e-02 -4.49609667e-01 2.26573393e-01
-6.96225524e-01 -2.87304282e-01 5.10777950e-01 -4.93135363e-01
4.72156137e-01 -1.01197219e+00 -2.06965968e-01 -9.03918266e-01
-6.67468786e-01 -1.67654693e-01 1.13026857e+00 -8.11514556e-01
1.35623443e+00 -2.22645783e+00 8.09703588e-01 2.62633502e-01
-3.76998968e-02 4.73877601e-02 4.26866442e-01 8.62094462e-01
1.78387612e-01 -2.59996187e-02 -2.01060757e-01 6.19460225e-01
-4.80684698e-01 1.76772222e-01 -2.71519750e-01 4.26894218e-01
1.87213540e-01 3.14113677e-01 -5.23113012e-01 -2.33540624e-01
-3.30060363e-01 3.15345854e-01 -5.45360446e-01 2.68767834e-01
6.22987926e-01 5.37667632e-01 -9.70211446e-01 3.49434555e-01
6.92686677e-01 5.24643600e-01 1.58903256e-01 -4.84426975e-01
-8.51586103e-01 -2.08350256e-01 -1.11649358e+00 5.26652277e-01
-2.42584512e-01 -2.57849377e-02 7.82319605e-01 -1.00948644e+00
9.96402919e-01 9.38073099e-01 5.61249971e-01 -1.73944026e-01
3.53332430e-01 3.45191389e-01 3.18567991e-01 -1.20630336e+00
3.41950625e-01 -3.80060494e-01 1.19235463e-01 -1.40777817e-02
-4.70618635e-01 -7.17977643e-01 1.43224865e-01 -2.54887104e-01
6.16188884e-01 6.80246502e-02 -2.35987827e-01 -3.94733578e-01
7.05704212e-01 -3.42624541e-03 9.63842034e-01 -1.93744943e-01
2.94831812e-01 1.47514984e-01 4.22037661e-01 -2.51898617e-02
-9.86503839e-01 -5.12074471e-01 -2.87182122e-01 4.63819504e-01
4.74114716e-01 2.35234529e-01 -7.66190112e-01 5.98218143e-01
5.66768870e-02 5.36341548e-01 3.14743780e-02 -7.54270732e-01
-9.34527040e-01 -8.68818641e-01 1.64561756e-02 5.03951550e-01
3.26628804e-01 -8.87169480e-01 -1.03231442e+00 2.64377624e-01
-2.65196204e-01 -8.26364696e-01 -2.51868457e-01 -2.37110808e-01
-1.03283787e+00 -1.13460743e+00 -6.61459088e-01 -9.10875976e-01
7.06920683e-01 2.71411240e-01 2.62067884e-01 4.05543625e-01
-3.16811889e-01 1.87748864e-01 -2.77307093e-01 -3.09183866e-01
-3.80559474e-01 -1.57418534e-01 5.66787004e-01 -2.68483669e-01
-7.62409389e-01 -3.22584271e-01 -5.14683306e-01 7.24958181e-01
-8.19215357e-01 -2.73794770e-01 4.26896572e-01 8.20897520e-01
2.88347542e-01 1.02503800e+00 8.06236684e-01 -1.88896775e-01
7.05499470e-01 -7.12454394e-02 -8.40845644e-01 -3.90167594e-01
-3.96626413e-01 -3.34821492e-01 7.91310787e-01 -5.66908240e-01
-1.44194996e+00 -1.96475700e-01 4.91955783e-04 -3.62823904e-01
2.38480821e-01 8.48824143e-01 -1.92445546e-01 -4.94671203e-02
2.10307822e-01 1.69078320e-01 2.83305317e-01 -6.47138774e-01
-2.51382720e-02 5.01008093e-01 5.62468350e-01 -2.09646076e-01
1.25275934e+00 4.98184785e-02 9.09903765e-01 -1.34639657e+00
-2.85720408e-01 -2.00548485e-01 -4.61829215e-01 -5.74359953e-01
8.96052718e-01 -1.84805050e-01 -1.23446286e+00 8.83875430e-01
-1.17647612e+00 -1.19568333e-01 -3.50754783e-02 8.47373366e-01
-2.17250541e-01 6.73536241e-01 -9.64727759e-01 -1.33120728e+00
-4.53634739e-01 -1.04168832e+00 5.59307218e-01 4.76902723e-01
2.25319996e-01 -8.19283426e-01 4.69143279e-02 4.37091999e-02
3.75087857e-01 6.80088401e-01 7.15927482e-01 7.43411899e-01
-2.38148779e-01 -2.91802138e-01 6.63035274e-01 3.24710131e-01
2.14726537e-01 4.29040015e-01 -4.08933789e-01 -5.41182280e-01
8.85598779e-01 7.40509480e-02 4.14695501e-01 7.04073608e-01
2.60435641e-01 -6.42745674e-01 -4.37081039e-01 2.16148540e-01
1.48886478e+00 5.58043778e-01 3.55760574e-01 -3.08812410e-02
8.20289195e-01 9.59299147e-01 7.11966932e-01 3.61123413e-01
-2.41329953e-01 5.69579601e-01 8.84913683e-01 5.21815270e-02
5.84299266e-01 5.17758012e-01 6.66125238e-01 1.31299818e+00
-9.72112298e-01 -2.71603698e-03 -5.49981177e-01 4.11991984e-01
-1.34711671e+00 -1.00784528e+00 -1.04121828e+00 2.09657288e+00
7.01171756e-01 -1.89644694e-01 -1.58338919e-01 5.05403996e-01
1.01341498e+00 -4.01807666e-01 -3.91080260e-01 -5.80927134e-01
2.61735246e-02 -2.39948243e-01 5.20186305e-01 7.66916335e-01
-4.64374393e-01 2.04380572e-01 6.56881285e+00 3.04255128e-01
-1.49851978e+00 -3.75061452e-01 -2.13522151e-01 1.40879110e-01
-6.32440746e-02 -9.86501053e-02 -6.44437850e-01 5.44240713e-01
3.96298647e-01 -3.44048679e-01 3.22982848e-01 4.84881461e-01
5.80440283e-01 -2.28753433e-01 -2.26410553e-01 3.38591516e-01
-3.44304919e-01 -7.06713378e-01 -4.82160896e-01 1.16035022e-01
6.19569719e-01 -6.86236501e-01 -4.43709120e-02 -4.12477076e-01
-6.25379622e-01 -3.94869953e-01 6.82546973e-01 9.20082688e-01
3.91568094e-01 -7.75021613e-01 9.72819269e-01 9.15845633e-01
-1.08755434e+00 -2.72279829e-01 -6.49566174e-01 -6.19750738e-01
8.44385624e-01 9.77502465e-01 -4.49676245e-01 9.60977912e-01
6.28538251e-01 4.45410103e-01 1.66661575e-01 5.57711899e-01
-3.65578562e-01 6.06759965e-01 -3.43145430e-01 -5.11660129e-02
-2.73241550e-01 -8.96310031e-01 1.04935360e+00 1.70844316e-01
6.81245327e-01 5.50000310e-01 -1.44816697e-01 9.23148036e-01
5.51324844e-01 1.52470469e-02 -5.12053430e-01 1.04099736e-01
2.23520964e-01 1.27172363e+00 -2.95384914e-01 1.99450031e-01
-9.61798355e-02 1.81976974e-01 -6.65419400e-01 4.21327859e-01
-7.24733412e-01 -8.10060799e-01 1.97044343e-01 2.53160983e-01
2.20720932e-01 -5.30784011e-01 -2.95355469e-01 -9.17698443e-01
1.97898120e-01 2.14972988e-01 -8.45562518e-02 -8.18076789e-01
-8.55265141e-01 2.22130075e-01 3.59000564e-01 -1.25879133e+00
-4.59605962e-01 -6.22108221e-01 -1.23137832e+00 9.76670682e-01
-1.06816232e+00 -2.91858643e-01 6.26746044e-02 2.62847841e-01
-1.69590846e-01 2.28290051e-01 4.08319265e-01 3.44255306e-02
-1.03700221e+00 -1.64979547e-01 4.91084248e-01 -8.33395571e-02
4.70506340e-01 -8.99468005e-01 -4.40122157e-01 7.72467256e-01
-1.26517248e+00 4.09898609e-01 1.11470878e+00 -6.20401621e-01
-1.72726643e+00 -4.78648335e-01 7.02510655e-01 4.49937493e-01
4.71773326e-01 1.40727386e-01 -1.14066434e+00 5.70070893e-02
3.67136508e-01 -4.75451112e-01 1.83944672e-01 -6.17962658e-01
8.37167263e-01 -1.85469864e-03 -8.79055977e-01 4.85897452e-01
5.84876120e-01 -1.72553100e-02 -7.21570313e-01 1.93172842e-01
6.05007172e-01 -4.89291757e-01 -1.08301246e+00 7.09956765e-01
6.09907746e-01 -4.14221942e-01 9.88938928e-01 -3.68361682e-01
5.75727522e-01 -4.07011747e-01 4.21667188e-01 -1.33717775e+00
-7.46052504e-01 -6.88066423e-01 -2.10280009e-02 1.32937503e+00
6.51558712e-02 -9.34044659e-01 2.34370932e-01 3.82401794e-01
-7.00465739e-01 -7.75806606e-01 -8.63541484e-01 -8.34170163e-01
1.10386245e-01 7.66270280e-01 -7.50834867e-02 8.40359032e-01
6.90503776e-01 5.66354752e-01 -2.45185226e-01 6.12582922e-01
2.42712125e-01 -8.61161277e-02 3.01283777e-01 -1.12441909e+00
-1.11853868e-01 -1.85799584e-01 3.12319100e-02 -5.34452915e-01
-1.22422479e-01 -1.45093948e-01 2.46076480e-01 -1.66467631e+00
-5.51147640e-01 4.14618015e-01 4.77594733e-02 1.87030837e-01
-2.77385324e-01 3.47135030e-02 -1.41919300e-01 5.99375427e-01
7.77815759e-01 7.46186316e-01 2.06743264e+00 1.71361461e-01
-4.51589495e-01 3.64879549e-01 4.32974994e-02 6.57523990e-01
8.54790330e-01 -2.40454003e-01 -2.79627264e-01 -4.12851244e-01
3.36814463e-01 5.19692540e-01 3.97454463e-02 -7.24722087e-01
8.51478726e-02 -5.32147944e-01 -9.68435630e-02 -6.70812070e-01
1.05095282e-01 -7.24906981e-01 1.80707783e-01 1.14400339e+00
3.74723494e-01 1.01380989e-01 -1.93615213e-01 1.70323059e-01
-3.62522393e-01 -5.22605717e-01 8.80417705e-01 2.25482911e-01
2.13033080e-01 -5.67426622e-01 -1.02569437e+00 -3.07645172e-01
8.67496371e-01 -1.58030048e-01 -2.61992007e-01 -1.25566557e-01
-5.89530885e-01 2.42248997e-01 -5.58000393e-02 -1.98009416e-01
4.79905009e-01 -1.09111202e+00 -5.14217973e-01 2.43066400e-01
-9.39069331e-01 5.00304878e-01 5.07461071e-01 1.22661924e+00
-6.87232256e-01 4.36225474e-01 -3.16861600e-01 -4.73687649e-01
-1.12751412e+00 3.55573773e-01 4.17689741e-01 1.08651526e-01
-1.75791845e-01 5.33659875e-01 2.93566614e-01 1.64702371e-01
-4.76723880e-01 -6.04202390e-01 -6.62669599e-01 6.41589090e-02
8.68541822e-02 1.01231074e+00 1.22785300e-01 -8.36465657e-01
-2.65675455e-01 1.07797503e+00 7.12125182e-01 -7.99064264e-02
1.22371960e+00 -3.05330366e-01 -5.03333747e-01 1.61359355e-01
7.44740546e-01 3.95674169e-01 -9.79312778e-01 3.56847346e-01
-5.07822096e-01 -2.35097945e-01 1.89315408e-01 -6.65017813e-02
-1.06890023e+00 6.59203887e-01 -2.65120625e-01 8.46879423e-01
1.37811959e+00 -4.33469176e-01 7.67570794e-01 3.60107988e-01
1.44126285e-02 -1.23370433e+00 1.48404106e-01 4.66770321e-01
1.11422908e+00 -2.49706313e-01 3.04145608e-02 -9.66015697e-01
-2.22229674e-01 1.46681547e+00 7.68373311e-01 -3.45099390e-01
5.84852934e-01 3.32412869e-01 4.89182305e-03 2.06575572e-01
-5.63961148e-01 2.12971836e-01 3.32713455e-01 4.10012364e-01
8.24017525e-01 -2.56773472e-01 -1.18832135e+00 7.24842250e-01
-1.19757511e-01 -1.97094217e-01 9.00866866e-01 1.20836866e+00
-7.87893772e-01 -7.31979907e-01 -1.11007214e+00 -9.80275348e-02
-5.57250261e-01 5.36445200e-01 -2.72605658e-01 7.44278014e-01
4.57452498e-02 1.04479253e+00 -1.36689439e-01 -2.65822202e-01
8.55250716e-01 5.31631187e-02 7.38259703e-02 -3.87615561e-01
-1.94863543e-01 6.93806887e-01 -1.93416178e-01 -8.45561028e-02
-4.09754425e-01 -5.53218484e-01 -1.70723844e+00 -2.23429427e-02
-1.29366851e+00 5.53917289e-01 4.91396636e-01 8.61565590e-01
9.98213217e-02 7.79362142e-01 1.03388071e+00 -9.76271212e-01
-1.02368891e+00 -1.13428128e+00 -1.07178533e+00 3.85938510e-02
1.55249611e-01 -1.07234335e+00 -7.35310793e-01 1.86572447e-01] | [5.597653388977051, 2.7145426273345947] |
950cb202-c7ac-45d2-88fa-165defdb0c81 | headposr-end-to-end-trainable-head-pose | 2202.03548 | null | https://arxiv.org/abs/2202.03548v1 | https://arxiv.org/pdf/2202.03548v1.pdf | HeadPosr: End-to-end Trainable Head Pose Estimation using Transformer Encoders | In this paper, HeadPosr is proposed to predict the head poses using a single RGB image. \textit{HeadPosr} uses a novel architecture which includes a transformer encoder. In concrete, it consists of: (1) backbone; (2) connector; (3) transformer encoder; (4) prediction head. The significance of using a transformer encoder for HPE is studied. An extensive ablation study is performed on varying the (1) number of encoders; (2) number of heads; (3) different position embeddings; (4) different activations; (5) input channel size, in a transformer used in HeadPosr. Further studies on using: (1) different backbones, (2) using different learning rates are also shown. The elaborated experiments and ablations studies are conducted using three different open-source widely used datasets for HPE, i.e., 300W-LP, AFLW2000, and BIWI datasets. Experiments illustrate that \textit{HeadPosr} outperforms all the state-of-art methods including both the landmark-free and the others based on using landmark or depth estimation on the AFLW2000 dataset and BIWI datasets when trained with 300W-LP. It also outperforms when averaging the results from the compared datasets, hence setting a benchmark for the problem of HPE, also demonstrating the effectiveness of using transformers over the state-of-the-art. | ['Naina Dhingra'] | 2022-02-07 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-6.26852289e-02 5.34210503e-01 1.37130126e-01 -1.84323221e-01
-8.25537026e-01 2.64855381e-02 6.68794274e-01 6.39713416e-03
-7.74501562e-01 6.62644804e-01 3.77066761e-01 1.54100582e-02
-1.58403113e-01 -7.21790493e-01 -1.16255796e+00 -8.41938317e-01
-3.60537350e-01 5.85723579e-01 5.33989549e-01 -3.46606851e-01
7.47621730e-02 5.50273418e-01 -1.97714472e+00 -2.32553959e-01
2.90169418e-01 1.38098025e+00 3.24615747e-01 4.11197931e-01
2.25693986e-01 6.86873019e-01 -4.09506053e-01 -4.52171355e-01
3.06719661e-01 -1.16979539e-01 -5.30062258e-01 -1.10707402e-01
4.63644713e-01 -3.27164829e-01 -3.78053606e-01 4.52276707e-01
1.03933454e+00 -4.28140089e-02 7.10955381e-01 -1.27748895e+00
-4.24881071e-01 4.48114842e-01 -6.35997415e-01 8.46158192e-02
4.39699233e-01 1.77637294e-01 8.90359521e-01 -9.77853119e-01
5.06904423e-01 9.50137794e-01 8.38644445e-01 4.40582991e-01
-8.25096428e-01 -7.40554035e-01 -6.95418417e-02 4.46402699e-01
-1.60718274e+00 -2.84095109e-01 5.51242590e-01 -2.05716491e-01
1.10205698e+00 3.19451094e-02 7.49982178e-01 1.25281930e+00
3.23167920e-01 8.16613257e-01 1.20075071e+00 -4.64202583e-01
2.97347367e-01 -3.17286663e-02 6.90561384e-02 8.47684741e-01
5.41090667e-02 2.86777556e-01 -8.58366668e-01 2.24731341e-01
5.29661834e-01 -3.28818887e-01 -5.29670238e-01 -2.19337538e-01
-9.05006528e-01 9.86494064e-01 6.94645286e-01 3.10274333e-01
-3.35648298e-01 1.91188142e-01 4.68778282e-01 5.77268042e-02
3.39165062e-01 2.30110869e-01 -5.33789098e-01 -1.32004218e-02
-8.73471439e-01 1.79470435e-01 7.76127994e-01 1.18623841e+00
7.35062480e-01 -3.31775062e-02 -2.14961216e-01 5.65225363e-01
1.79393113e-01 4.96453851e-01 8.74184012e-01 -4.22227472e-01
4.96378124e-01 3.53095472e-01 -8.49421844e-02 -5.03252923e-01
-9.04866695e-01 -2.82349557e-01 -6.46532059e-01 1.76092982e-01
2.17516989e-01 -1.07663102e-01 -1.29537618e+00 1.66758883e+00
5.71625195e-02 2.48570308e-01 -4.20938209e-02 7.98067570e-01
1.25582302e+00 6.35689974e-01 -4.56480645e-02 1.45469993e-01
1.34960485e+00 -1.21368945e+00 -7.15530574e-01 -3.05319756e-01
5.84150314e-01 -5.14675796e-01 9.73112643e-01 5.35663664e-01
-1.19083893e+00 -5.90063572e-01 -1.20349491e+00 -4.12572503e-01
-8.57889473e-01 4.11487669e-01 5.46887696e-01 5.18129230e-01
-1.54301727e+00 4.76405680e-01 -6.56161129e-01 -6.09370768e-01
-1.67601407e-02 8.31632972e-01 -7.15533078e-01 1.21391751e-01
-1.13247335e+00 1.24714792e+00 3.63323957e-01 3.77807945e-01
-8.51914167e-01 -6.37452424e-01 -1.14878213e+00 5.49364761e-02
4.08596210e-02 -6.20298803e-01 9.40589309e-01 -5.02459228e-01
-1.66924858e+00 9.65799928e-01 -1.24317103e-05 -5.70322394e-01
7.13249445e-01 -5.01928270e-01 -1.64389051e-02 5.63536696e-02
9.52817574e-02 1.14686358e+00 7.57926166e-01 -1.08599436e+00
-5.97324789e-01 -5.62821031e-01 -1.07460670e-01 1.26023248e-01
-2.26076439e-01 -2.44528070e-01 -6.85498655e-01 -3.28628957e-01
6.48879483e-02 -1.07113051e+00 1.67165682e-01 -5.63704483e-02
-6.62697792e-01 -3.42202306e-01 8.02316129e-01 -7.64770389e-01
1.15756750e+00 -2.19364548e+00 4.06809568e-01 9.46879610e-02
1.56497099e-02 2.53765851e-01 -5.23958914e-02 4.89555836e-01
-3.29321027e-01 -1.98997647e-01 -8.34943876e-02 -8.42648864e-01
1.07820988e-01 2.81569749e-01 1.26250656e-02 6.92585647e-01
-7.01651722e-02 7.48875499e-01 -4.52935576e-01 -5.83525240e-01
5.92909276e-01 8.89189661e-01 -4.05008316e-01 2.19292074e-01
2.89289892e-01 2.43919447e-01 -1.71983331e-01 4.87637132e-01
6.41665936e-01 8.06987435e-02 -3.13297689e-01 -6.82883322e-01
-3.94733369e-01 1.39066353e-01 -1.16598439e+00 1.73092413e+00
-6.83340967e-01 6.75664544e-01 -2.60997355e-01 -6.99934006e-01
9.84182894e-01 4.68743831e-01 2.74224401e-01 -9.52877760e-01
2.56701142e-01 4.05768543e-01 -1.55085355e-01 -4.74768847e-01
4.11723107e-01 -6.21400476e-02 9.42583606e-02 -9.61824656e-02
5.65720499e-01 7.31037185e-02 2.43286058e-01 -2.62351513e-01
1.21078408e+00 3.28285187e-01 1.19622029e-01 -2.63349593e-01
5.10132194e-01 -7.72877812e-01 3.20341766e-01 4.17559832e-01
-1.23029590e-01 9.38255191e-01 4.10118461e-01 -4.07767028e-01
-9.31837797e-01 -1.15750825e+00 -4.22347963e-01 9.33811188e-01
-5.40941916e-02 -5.67948580e-01 -7.06779659e-01 -3.79096001e-01
-4.22480665e-02 6.27799153e-01 -1.14501083e+00 -1.25010714e-01
-6.05656564e-01 -8.06352079e-01 7.45047987e-01 9.00655329e-01
7.54026532e-01 -1.08983159e+00 -1.12661362e+00 -3.01915277e-02
-1.68165147e-01 -1.30502510e+00 -3.82694341e-02 8.42981100e-01
-5.16951442e-01 -9.09586847e-01 -9.12774682e-01 -7.43130147e-01
3.53263408e-01 -4.27385062e-01 1.06473470e+00 -6.98083639e-02
-1.17361091e-01 5.85525274e-01 -6.11547887e-01 -6.27600133e-01
4.20225352e-01 3.63175213e-01 -1.55286625e-01 -1.65121391e-01
1.36675462e-01 -5.59727490e-01 -6.55420661e-01 2.67063797e-01
-7.73920119e-01 -4.72635403e-02 8.40218842e-01 7.14360356e-01
4.91457820e-01 -1.05214223e-01 2.16142550e-01 -7.02002704e-01
1.70241594e-01 -3.20930451e-01 -3.90038431e-01 1.66918769e-01
-4.07885224e-01 2.18800902e-01 5.59439123e-01 -4.84394990e-02
-7.21713305e-01 2.35696677e-02 -6.66837096e-01 -4.58543688e-01
-6.13140315e-02 3.82699929e-02 -1.20541617e-01 -4.21279036e-02
3.43465298e-01 3.48158896e-01 -3.17757934e-01 -4.69735503e-01
1.34273797e-01 4.06135947e-01 4.38139409e-01 -3.54071319e-01
5.71859062e-01 3.90076578e-01 2.65255988e-01 -1.04373360e+00
-5.11911392e-01 -2.32327059e-01 -8.39057446e-01 -3.03823538e-02
1.09478033e+00 -8.27071071e-01 -5.64091504e-01 6.48942471e-01
-1.00421333e+00 -4.22573388e-01 -2.24723950e-01 4.68040556e-01
-6.36989474e-01 -6.35963604e-02 -6.68854892e-01 -6.45606458e-01
-4.40203249e-01 -1.41782284e+00 1.59586930e+00 1.66208893e-01
-2.92334408e-02 -9.82817471e-01 -4.28580567e-02 2.23783195e-01
3.51521730e-01 4.02315706e-01 6.46107733e-01 -5.42514324e-01
-3.49014223e-01 -1.53679624e-01 -7.79375508e-02 2.62522936e-01
-2.03742668e-01 1.02971748e-01 -1.29703951e+00 -4.43523765e-01
-3.63863140e-01 -4.75660563e-01 8.16624224e-01 4.19129282e-01
1.04181838e+00 -1.03505448e-01 -2.83625960e-01 8.09874833e-01
1.49847174e+00 1.48007110e-01 1.09863150e+00 7.47417510e-01
5.66720247e-01 5.52313030e-01 3.94921988e-01 4.88168865e-01
5.46225846e-01 8.95026982e-01 7.84721792e-01 -3.30250889e-01
-1.45661458e-01 -3.15258056e-01 3.70327502e-01 6.23463690e-01
-2.69006878e-01 -2.45464489e-01 -7.97404230e-01 5.06015420e-01
-1.60541189e+00 -3.26795042e-01 2.66447328e-02 2.22119069e+00
3.49611223e-01 1.22164130e-01 1.77159399e-01 5.09162307e-01
3.40418011e-01 2.76915044e-01 -3.50251019e-01 -5.03694117e-01
-9.48652849e-02 8.87354970e-01 7.07710266e-01 2.07583666e-01
-1.16866231e+00 7.15732396e-01 5.49102736e+00 6.64171338e-01
-1.42368996e+00 2.32068032e-01 3.21403265e-01 -2.93628946e-02
1.61873594e-01 -4.00250703e-01 -1.13138592e+00 6.13027871e-01
9.00253892e-01 3.69883627e-01 8.88469145e-02 7.60278821e-01
-2.28981927e-01 -2.79334158e-01 -1.28142571e+00 1.02971995e+00
4.57863182e-01 -8.12549770e-01 -1.11212961e-01 4.44504619e-02
1.21154808e-01 4.19740021e-01 1.54604346e-01 3.93111140e-01
-1.97408870e-01 -1.22878087e+00 1.03744829e+00 2.20133424e-01
9.12748396e-01 -6.35622203e-01 1.15444124e+00 1.08202495e-01
-1.44092190e+00 -2.13276595e-01 -2.33064964e-01 1.89484179e-01
2.92780511e-02 3.42632025e-01 -4.85484302e-01 6.99694932e-01
1.09193909e+00 6.31392121e-01 -9.26153660e-01 9.67712998e-01
-6.71875596e-01 5.21069467e-01 -7.83544719e-01 1.27733573e-01
3.83719176e-01 1.16308466e-01 1.64749563e-01 1.37815821e+00
4.64700848e-01 -1.72927409e-01 -3.40516299e-01 4.09203202e-01
1.38790399e-01 -1.40024796e-02 -5.84769845e-01 5.93322873e-01
3.64612281e-01 1.21060872e+00 -6.57276928e-01 7.27346689e-02
-3.93069804e-01 1.01544774e+00 3.04155141e-01 2.20582679e-01
-1.12063730e+00 -3.54458481e-01 2.92951405e-01 4.46921855e-01
6.34862483e-01 2.28463840e-02 1.03531994e-01 -7.07986474e-01
3.35694440e-02 -2.98443675e-01 3.41466486e-01 -1.18722188e+00
-9.44438994e-01 9.41580892e-01 1.91336557e-01 -1.00914884e+00
-2.53327906e-01 -9.59756434e-01 -5.64378977e-01 6.86949313e-01
-1.83145809e+00 -1.48169720e+00 -4.58470643e-01 7.71793306e-01
3.81271660e-01 9.69232842e-02 9.20287549e-01 5.33408761e-01
-8.05505455e-01 9.47269142e-01 -3.39494854e-01 1.13213077e-01
4.64151829e-01 -1.33892465e+00 5.97998202e-02 4.69312817e-01
-3.48413698e-02 4.49914694e-01 7.17209458e-01 -8.99795517e-02
-1.33685172e+00 -9.20098960e-01 9.27233458e-01 -1.35916814e-01
3.42318207e-01 -5.57970345e-01 -5.34342289e-01 1.21340311e+00
4.43955511e-01 1.21103264e-01 5.41413307e-01 5.82748652e-03
-7.50971809e-02 -3.36314797e-01 -1.36741340e+00 2.61141658e-01
9.08440411e-01 -2.55203426e-01 -6.10789657e-01 1.43558457e-01
4.94021356e-01 -7.41743863e-01 -9.55975056e-01 5.29796958e-01
6.42740130e-01 -1.44654834e+00 9.34565783e-01 1.00938551e-01
5.82404792e-01 -8.42295885e-02 -1.99824676e-01 -1.26433969e+00
-2.90968448e-01 -2.18586206e-01 -8.97679999e-02 1.12271667e+00
2.76288956e-01 -7.86861539e-01 6.99631512e-01 4.04250264e-01
-4.31288600e-01 -1.34214938e+00 -1.33927846e+00 -5.35448313e-01
1.15845762e-01 -2.48372898e-01 4.06301707e-01 3.17148805e-01
-4.28282529e-01 3.86837602e-01 -2.85949022e-01 1.29988983e-01
3.61375749e-01 -3.36168051e-01 7.82841086e-01 -1.04276323e+00
-1.45650342e-01 -5.89993224e-02 -8.77938390e-01 -9.62720335e-01
9.87275764e-02 -4.94600922e-01 -6.95666671e-02 -1.67337275e+00
-7.93387294e-02 -2.95520782e-01 -8.87573212e-02 8.42355132e-01
2.32580692e-01 4.18258637e-01 3.61506045e-01 -1.18826486e-01
-3.18018287e-01 7.58444726e-01 9.58107352e-01 2.21483618e-01
2.09628701e-01 -2.83484489e-01 -1.95531681e-01 8.21729422e-01
4.94560987e-01 -1.08999543e-01 -4.24672216e-01 -4.02823716e-01
1.50842965e-01 -1.02611735e-01 4.20598686e-01 -1.58671772e+00
2.67042369e-01 7.97154546e-01 5.62259018e-01 -7.65637040e-01
8.58636737e-01 -9.13747191e-01 -9.33740847e-03 4.35928285e-01
8.38380828e-02 5.37483454e-01 3.34503859e-01 1.40409499e-01
-1.76601395e-01 -2.43258893e-01 8.56679499e-01 -5.12292050e-02
-9.65209365e-01 1.06176570e-01 -4.79674190e-02 -6.76677898e-02
1.08782613e+00 -4.37632680e-01 -1.61549866e-01 -2.42137477e-01
-8.00218284e-01 1.95270836e-01 2.02117339e-01 3.43396991e-01
5.93279541e-01 -1.05519414e+00 -5.59558272e-01 4.69458431e-01
1.47958562e-01 1.79368824e-01 4.71557081e-02 9.67286587e-01
-6.14081681e-01 5.16932607e-01 -3.29697162e-01 -5.79792738e-01
-9.83847737e-01 3.40635031e-01 4.27429408e-01 -5.09607375e-01
-8.23598623e-01 1.01907933e+00 2.98711717e-01 -4.43182945e-01
6.21573329e-01 -5.22146583e-01 -2.64459550e-01 3.27902615e-01
1.69767082e-01 4.16869670e-01 4.55130458e-01 -8.58681798e-01
-4.65134561e-01 8.29103470e-01 7.84335136e-02 -6.67699203e-02
1.48157573e+00 2.17277948e-02 1.25402182e-01 4.09520715e-01
1.36496127e+00 -2.71974474e-01 -1.13366508e+00 2.08690792e-01
-2.09247723e-01 -1.28578827e-01 1.42273441e-01 -7.76983380e-01
-1.47824454e+00 9.99630272e-01 1.01422286e+00 -2.53383011e-01
1.35977912e+00 -3.83816585e-02 8.35716605e-01 1.01806466e-02
8.13404441e-01 -9.13955450e-01 -4.26208377e-02 4.16822106e-01
1.06299984e+00 -9.01198745e-01 4.61052405e-03 -1.20705403e-01
-5.09286761e-01 1.14345479e+00 7.00900137e-01 -2.33999684e-01
8.15902472e-01 5.95305502e-01 -2.17189550e-01 -3.36174309e-01
-3.97698462e-01 -4.66951013e-01 1.27373904e-01 5.33435106e-01
5.66668987e-01 -1.73496693e-01 -4.44092989e-01 6.00739479e-01
-6.87620640e-01 1.41914561e-01 2.55474895e-01 9.54967439e-01
5.65787479e-02 -8.79091501e-01 -1.88055143e-01 3.72723997e-01
-3.01065892e-01 -2.57460736e-02 -9.31093022e-02 1.36951756e+00
7.02627003e-01 5.79633176e-01 7.97097385e-02 -4.66098219e-01
6.68506384e-01 -4.73320186e-02 6.40298188e-01 -3.53345931e-01
-8.17966342e-01 -2.92340100e-01 -4.42071036e-02 -7.34906197e-01
-3.44345391e-01 -5.39879680e-01 -1.21769071e+00 -1.66928336e-01
-2.71787256e-01 6.25459924e-02 6.14904583e-01 8.71170640e-01
1.07696220e-01 7.72892416e-01 2.48851165e-01 -1.31515062e+00
-2.46126875e-01 -1.29532826e+00 -8.94538879e-01 2.70967811e-01
2.44368374e-01 -1.19700611e+00 -5.21300852e-01 -3.55926514e-01] | [13.683554649353027, 0.28112515807151794] |
4c709c42-df50-49b3-be5c-fe737c825c06 | unsupervised-anomaly-detection-via-nonlinear | 2306.09441 | null | https://arxiv.org/abs/2306.09441v1 | https://arxiv.org/pdf/2306.09441v1.pdf | Unsupervised Anomaly Detection via Nonlinear Manifold Learning | Anomalies are samples that significantly deviate from the rest of the data and their detection plays a major role in building machine learning models that can be reliably used in applications such as data-driven design and novelty detection. The majority of existing anomaly detection methods either are exclusively developed for (semi) supervised settings, or provide poor performance in unsupervised applications where there is no training data with labeled anomalous samples. To bridge this research gap, we introduce a robust, efficient, and interpretable methodology based on nonlinear manifold learning to detect anomalies in unsupervised settings. The essence of our approach is to learn a low-dimensional and interpretable latent representation (aka manifold) for all the data points such that normal samples are automatically clustered together and hence can be easily and robustly identified. We learn this low-dimensional manifold by designing a learning algorithm that leverages either a latent map Gaussian process (LMGP) or a deep autoencoder (AE). Our LMGP-based approach, in particular, provides a probabilistic perspective on the learning task and is ideal for high-dimensional applications with scarce data. We demonstrate the superior performance of our approach over existing technologies via multiple analytic examples and real-world datasets. | ['Ramin Bostanabad', 'Zahra Zanjani Foumani', 'Mehdi Shishehbor', 'Amin Yousefpour'] | 2023-06-15 | null | null | null | null | ['anomaly-detection', 'unsupervised-anomaly-detection'] | ['methodology', 'methodology'] | [-3.8378332e-02 5.2028079e-02 4.2194940e-02 -3.6297888e-01
-6.8875837e-01 -4.5896593e-01 8.8180280e-01 2.5971356e-01
1.9973171e-01 7.8931510e-02 1.0455915e-01 -3.9930162e-01
-3.8310912e-01 -6.5318984e-01 -5.3857225e-01 -9.8464018e-01
-3.0640244e-01 5.8892262e-01 -5.2048061e-02 2.6132631e-01
3.4068105e-01 6.4637023e-01 -1.6098350e+00 -7.2505005e-02
8.3591825e-01 9.0718418e-01 -4.7329953e-01 5.0191629e-01
-2.3394650e-01 4.6627793e-01 -3.5603523e-01 -8.1281878e-02
3.1294569e-01 -2.1670613e-01 -5.2246612e-01 2.9670531e-01
1.5088262e-01 -2.7763936e-01 -1.2838124e-01 1.0561552e+00
1.9055642e-01 5.1250465e-02 1.1662595e+00 -1.4951094e+00
-5.9653538e-01 1.3048957e-01 -5.8859700e-01 2.8784201e-01
6.9665439e-02 7.1262352e-02 9.5385975e-01 -1.1315352e+00
1.0390166e-01 1.1953973e+00 5.7758373e-01 3.5604438e-01
-1.4626360e+00 -2.5502017e-01 1.9744074e-01 8.8155955e-02
-1.2844442e+00 -3.8632151e-01 9.6916932e-01 -8.1283146e-01
5.6125838e-01 1.7298447e-01 3.4486315e-01 1.3200790e+00
2.7316535e-01 8.7427622e-01 5.2287292e-01 -3.6252171e-01
7.8597403e-01 -2.1069333e-02 6.8420484e-03 4.5287850e-01
2.4450776e-01 -1.9363744e-02 -5.8390695e-01 -7.1654820e-01
6.0755366e-01 6.2673962e-01 5.6815550e-02 -7.6058745e-01
-1.1411439e+00 9.7815198e-01 -2.0592497e-03 1.8568595e-01
-6.7624450e-01 1.4807134e-02 3.1919548e-01 3.9021245e-01
8.3146393e-01 4.4033867e-01 -3.3957642e-01 -1.9420011e-01
-1.0177933e+00 8.2238123e-02 6.6969299e-01 6.6047853e-01
6.6080976e-01 2.2086702e-01 -1.3667578e-02 6.3346481e-01
8.6345351e-01 4.1846368e-01 6.4000034e-01 -6.0848397e-01
1.4566177e-01 9.2962313e-01 3.9558884e-02 -1.1445533e+00
-3.5169908e-01 -2.2602618e-01 -8.0243659e-01 2.4670456e-01
2.5457051e-01 -5.2612640e-02 -8.1663644e-01 1.4962150e+00
4.5144212e-01 6.1314946e-01 3.4232312e-01 5.2150446e-01
1.9987918e-01 6.4626235e-01 -1.5856685e-01 2.0169493e-02
9.6214139e-01 -4.6380672e-01 -6.5951920e-01 1.6423557e-02
8.9873755e-01 -5.0383753e-01 1.0337738e+00 4.0761730e-01
-5.5765402e-01 -7.6013908e-02 -1.0560528e+00 4.0495941e-01
-4.3219095e-01 -6.3127771e-02 3.4215024e-01 6.2453920e-01
-8.6649352e-01 6.3115555e-01 -1.2107462e+00 -3.7410253e-01
5.8768213e-01 9.2101686e-02 -3.0489564e-01 -4.6995368e-02
-6.2733692e-01 4.6342775e-01 4.2025784e-01 1.8789232e-01
-9.8431861e-01 -4.7939396e-01 -8.9014131e-01 -7.4441917e-02
2.9290402e-01 -3.3471492e-01 9.6920949e-01 -6.1820465e-01
-1.1989666e+00 7.2480071e-01 -1.1805307e-01 -6.1683267e-01
3.2882512e-01 -5.0460976e-01 -7.0164901e-01 1.1977234e-01
4.8633575e-02 2.3063611e-02 1.5372487e+00 -1.1944631e+00
-6.3208866e-01 -6.2380660e-01 -8.2378846e-01 -2.2351412e-01
-6.5477765e-01 -6.7380331e-03 -2.0063298e-02 -8.3042735e-01
5.8868396e-01 -9.5223242e-01 -1.1789752e-01 -9.0411022e-02
-6.6986340e-01 -3.6182317e-01 1.4246451e+00 -5.4523295e-01
1.1898265e+00 -2.2976003e+00 7.4068390e-02 5.8196187e-01
5.0342327e-01 2.4940526e-01 2.9284528e-01 6.1861968e-01
-1.1771615e-01 -9.3464829e-02 -5.9980804e-01 -5.9470761e-01
1.8861133e-01 2.9882690e-01 -6.8813294e-01 7.6497501e-01
6.2643534e-01 6.7820978e-01 -9.9135578e-01 9.3925476e-02
2.2940695e-01 1.2598741e-01 -4.7062773e-01 5.0751078e-01
-1.6992959e-01 6.0389078e-01 -4.5945525e-01 9.5668805e-01
4.3449894e-01 -2.6254785e-01 -3.2438979e-01 2.1975389e-01
4.6402417e-02 -6.8690069e-02 -1.3416697e+00 1.4495240e+00
-4.4892229e-02 6.3813943e-01 -2.9592016e-01 -1.2894481e+00
1.1410308e+00 3.6952275e-01 7.0353872e-01 -1.1826548e-01
-2.3320517e-01 3.3674449e-01 -2.2674295e-01 -5.7909989e-01
1.9785474e-01 3.3262555e-02 4.7694847e-02 7.6719296e-01
1.0693305e-01 2.9411528e-01 -2.7066800e-01 1.0213726e-01
1.4397734e+00 -1.0332936e-01 1.5949756e-01 -3.0353689e-01
3.7406921e-01 -2.9790908e-01 6.2466967e-01 8.3227485e-01
-1.9969977e-01 7.1159184e-01 6.6411269e-01 -6.6446620e-01
-1.0350727e+00 -1.4446354e+00 -3.4642145e-01 8.1660861e-01
-2.2304918e-01 -3.7693858e-01 -5.4158014e-01 -9.0500927e-01
1.9508095e-01 8.6794412e-01 -5.6136525e-01 -6.7655492e-01
-2.4646282e-01 -1.0165904e+00 4.4376299e-01 4.8762897e-01
-3.8634308e-02 -1.1063043e+00 -2.8725722e-01 1.7327605e-01
2.5538394e-01 -8.9745700e-01 -1.3482060e-02 1.4477590e-01
-1.0507482e+00 -1.1288933e+00 -1.8705383e-01 -3.6638066e-01
8.6098921e-01 -2.6565600e-02 8.9673686e-01 -2.6083910e-01
-2.7935842e-01 7.5641668e-01 -3.1164831e-01 -6.8268073e-01
-4.5698357e-01 -2.5368115e-01 6.7949867e-01 7.0200562e-01
8.7907040e-01 -7.7189535e-01 -4.6444574e-01 2.1660824e-01
-1.1295401e+00 -6.1271048e-01 6.4327836e-01 8.6908579e-01
4.9256662e-01 3.0315802e-01 9.2165536e-01 -7.1027607e-01
7.0629036e-01 -1.1202235e+00 -5.8535284e-01 -3.8752317e-02
-8.4606510e-01 2.9567492e-01 5.3087229e-01 -3.0037153e-01
-7.0123571e-01 -4.9215768e-02 6.1235219e-02 -7.1937096e-01
-5.4827797e-01 4.7801784e-01 -2.5731468e-01 2.0242421e-01
7.5661451e-01 3.5820639e-01 2.1990693e-01 -6.9611877e-01
3.1426200e-01 7.6449460e-01 4.8654476e-01 -5.7812476e-01
1.0144320e+00 7.6933050e-01 9.9680446e-02 -9.3200856e-01
-6.0837495e-01 -7.0307583e-01 -9.3964505e-01 -1.5642915e-02
6.7637122e-01 -7.9554576e-01 -2.1701199e-01 5.6520611e-01
-6.5096265e-01 -2.0172705e-03 -5.2966416e-01 5.0829351e-01
-5.4743540e-01 4.4548708e-01 -3.1360498e-01 -1.0699339e+00
-2.7322516e-01 -8.7364775e-01 1.3397460e+00 -8.8283762e-02
-3.8552147e-01 -1.3909186e+00 2.0811558e-01 -1.3884680e-01
3.0534956e-01 5.6701356e-01 9.8331982e-01 -1.4361418e+00
-3.4204352e-01 -6.7980957e-01 8.3439574e-02 6.2659127e-01
4.6005920e-01 1.2178689e-01 -1.0850832e+00 -3.6932480e-01
3.5636243e-01 5.2038185e-02 6.4226288e-01 2.3223218e-01
1.2033204e+00 -4.1876724e-01 -3.8174382e-01 4.3835399e-01
9.9744153e-01 -1.3063811e-01 3.1249297e-01 3.5294560e-01
9.2939049e-01 7.3862052e-01 3.6946630e-01 5.8957916e-01
2.7592149e-01 4.3684426e-01 4.9336186e-01 9.3379490e-02
6.2833422e-01 -2.6104742e-01 5.6519109e-01 8.0319446e-01
3.6714482e-01 6.0597911e-02 -1.2876036e+00 6.5996426e-01
-2.2409501e+00 -8.4319413e-01 -1.3572937e-01 2.4848053e+00
3.8678688e-01 1.6362695e-01 1.8370128e-01 3.8468325e-01
6.1264497e-01 -9.1027915e-02 -7.6519239e-01 -1.7091596e-01
2.3101302e-02 -2.3717560e-01 1.3426066e-02 -5.1732514e-02
-1.4437410e+00 6.1416340e-01 6.3624630e+00 4.7716641e-01
-8.9168113e-01 -2.4698071e-01 4.4742414e-01 -3.2808271e-03
-1.4600120e-01 2.5974572e-02 -4.9811947e-01 6.7241359e-01
1.2228943e+00 4.5340668e-02 3.1453192e-02 1.0540543e+00
4.1384313e-01 2.0468675e-01 -1.4790907e+00 1.0695703e+00
9.5121013e-03 -1.0553277e+00 7.1748890e-02 4.0763432e-01
6.5481472e-01 -2.3088397e-02 3.9691859e-01 1.9525532e-01
1.8268803e-01 -9.9619371e-01 3.6690524e-01 8.6925399e-01
1.0551366e-01 -7.8128922e-01 7.0814675e-01 5.0855792e-01
-5.1062381e-01 -3.5389480e-01 -2.9823366e-01 9.6717812e-02
5.1489223e-02 1.0431529e+00 -9.8173368e-01 4.8518789e-01
8.1058061e-01 1.0550067e+00 -5.2562284e-01 1.0386337e+00
-4.4666994e-02 1.0985698e+00 -4.4274634e-01 4.3273669e-01
2.3000531e-01 -4.7440922e-01 1.1084980e+00 8.6423892e-01
5.4530102e-01 -5.2487075e-01 2.5853494e-01 1.1028777e+00
2.0417994e-01 7.0478849e-02 -9.5462924e-01 -2.7183342e-01
4.5111501e-01 1.2082946e+00 -7.6802814e-01 -7.6146476e-02
-5.3899175e-01 7.7659392e-01 1.2497286e-01 3.5087961e-01
-2.4936730e-01 -1.3393477e-01 8.7116194e-01 1.8287975e-02
3.2830906e-01 -3.7678143e-01 -1.9787946e-01 -1.2960659e+00
2.5699487e-01 -7.3885471e-01 6.3264096e-01 -1.8608084e-01
-1.9336476e+00 2.6048890e-01 -9.2005767e-02 -1.5915755e+00
-8.7043399e-01 -6.6504979e-01 -9.7078788e-01 6.7355835e-01
-1.1966290e+00 -1.0090164e+00 -6.1667237e-02 5.5575842e-01
2.1695639e-01 -6.6145438e-01 1.0149400e+00 6.8326354e-02
-8.5099006e-01 3.5961086e-01 7.3696595e-01 1.2441595e-01
6.1737609e-01 -1.7259471e+00 5.8377779e-01 1.2308699e+00
4.0274188e-01 7.3022765e-01 6.7360044e-01 -5.8601338e-01
-1.3487804e+00 -1.3927301e+00 2.0792465e-01 -9.5990902e-01
8.5009521e-01 -5.5118591e-01 -1.3977371e+00 8.4459054e-01
-4.4903746e-01 3.2362381e-01 1.1472758e+00 2.6935965e-01
-2.8573731e-01 1.3594857e-01 -1.1991121e+00 5.8050376e-01
6.0813451e-01 -5.8653522e-01 -7.3961061e-01 3.6830509e-01
3.9390194e-01 -4.5000765e-02 -8.7105697e-01 3.5275581e-01
3.2557655e-02 -9.1664553e-01 7.5462437e-01 -9.5139033e-01
6.9808915e-02 -5.4363745e-01 -2.0845377e-01 -1.3957723e+00
-1.3335948e-01 -6.5270478e-01 -1.0685894e+00 1.3668450e+00
2.1426620e-01 -8.6428648e-01 7.3267436e-01 6.8597925e-01
-2.1569598e-01 -7.8273380e-01 -9.4767100e-01 -8.3001697e-01
-2.2611244e-02 -7.2142768e-01 6.2288421e-01 1.1008967e+00
-1.7991409e-01 -7.5484350e-02 -3.7228054e-01 6.5847856e-01
8.7858921e-01 4.5857076e-02 8.9423645e-01 -1.7664033e+00
-3.5497330e-02 -3.1505591e-01 -9.9293476e-01 -6.3509870e-01
1.7021300e-01 -8.7919575e-01 -1.8508552e-02 -1.0425860e+00
-1.8810588e-01 -4.1219145e-01 -3.9221671e-01 2.9461113e-01
-1.9081925e-01 -1.5451317e-01 -5.4845262e-01 6.2393492e-01
-5.5472976e-01 8.1314218e-01 2.7139741e-01 2.0434976e-01
-4.1471276e-01 3.7187031e-01 -6.9083542e-01 9.8293656e-01
6.9430029e-01 -6.0594285e-01 -3.6150837e-01 -1.2404960e-02
2.9328970e-02 -6.8513107e-01 5.3328615e-01 -1.0463344e+00
1.7050883e-01 9.0284020e-02 5.2968395e-01 -5.7111442e-01
4.7099255e-03 -8.5491586e-01 -3.2398686e-01 1.0853981e-01
-1.2415108e-01 1.2909088e-01 -5.5225454e-02 1.1500554e+00
-2.2403364e-01 -1.7994972e-01 3.7468192e-01 2.6805115e-01
-5.0794792e-01 4.4790214e-01 -4.6053782e-01 3.5069179e-02
1.1921116e+00 1.7488970e-03 2.0269297e-02 -4.6274164e-01
-6.9732070e-01 2.5744909e-01 5.0298184e-01 7.4614465e-01
7.5192094e-01 -1.4922984e+00 -6.8170971e-01 6.0398948e-01
5.7241631e-01 3.8077462e-01 9.0429857e-02 8.5694098e-01
-2.1783230e-01 2.3845844e-01 1.3266447e-01 -1.1249408e+00
-5.7670563e-01 6.7537475e-01 2.0643826e-01 1.3289759e-01
-1.1677324e+00 3.3244061e-01 1.7222729e-02 -5.6969476e-01
1.7830247e-01 -1.3130504e-01 -1.2584141e-01 4.3100826e-02
7.3616821e-01 4.4786391e-01 9.5743991e-02 -7.5398451e-01
-1.4646807e-01 1.8792459e-01 -2.1306977e-01 6.7746215e-02
1.5166324e+00 -1.4098288e-01 -1.7444479e-01 1.0282778e+00
1.0540760e+00 -2.0233959e-01 -1.4585388e+00 -5.9888947e-01
7.0285749e-01 -4.9530029e-01 1.9719254e-02 -1.3319837e-03
-6.0953987e-01 8.4106243e-01 8.0539644e-01 6.8200392e-01
9.1204101e-01 2.1370308e-01 5.2088088e-01 5.0935179e-01
-7.5232938e-02 -1.1406677e+00 5.0048727e-01 2.6463556e-01
7.3334914e-01 -1.4562391e+00 -2.6335546e-01 1.1904686e-01
-5.6959862e-01 1.1809621e+00 3.3634067e-01 -2.3820071e-01
9.4266438e-01 -1.0813454e-01 1.5941642e-01 -6.2042975e-01
-5.5720133e-01 8.5181817e-02 6.2335998e-01 6.6562623e-01
-8.4211506e-02 -1.0842736e-01 4.5909351e-01 6.3498104e-01
-7.3383093e-02 -6.7246109e-01 4.7858655e-01 1.1511128e+00
-4.4427949e-01 -8.2751334e-01 -5.3654569e-01 8.5020083e-01
-5.7929015e-01 1.8921889e-01 -2.1023040e-01 3.2602492e-01
-2.6454389e-01 8.6180568e-01 2.9982409e-01 -2.3143901e-01
1.4566188e-01 6.5479749e-01 -3.0720225e-01 -6.6446549e-01
2.8852016e-01 -4.5503024e-02 -5.5770755e-01 -5.8678156e-01
-1.5797935e-01 -1.0939504e+00 -1.0955153e+00 -1.9000690e-02
-1.9084859e-01 5.2249048e-02 7.1652442e-01 1.3271710e+00
7.5898635e-01 3.2992932e-01 8.7087214e-01 -7.4424481e-01
-6.8069589e-01 -8.0074108e-01 -7.9066229e-01 6.0386282e-01
6.6146100e-01 -8.9806777e-01 -7.4436665e-01 1.0563379e-02] | [7.7035393714904785, 2.3330531120300293] |
66a6da17-8a76-4cde-a928-30f1423fe9c1 | toward-knowledge-driven-speech-based-models | 2210.02527 | null | https://arxiv.org/abs/2210.02527v1 | https://arxiv.org/pdf/2210.02527v1.pdf | Toward Knowledge-Driven Speech-Based Models of Depression: Leveraging Spectrotemporal Variations in Speech Vowels | Psychomotor retardation associated with depression has been linked with tangible differences in vowel production. This paper investigates a knowledge-driven machine learning (ML) method that integrates spectrotemporal information of speech at the vowel-level to identify the depression. Low-level speech descriptors are learned by a convolutional neural network (CNN) that is trained for vowel classification. The temporal evolution of those low-level descriptors is modeled at the high-level within and across utterances via a long short-term memory (LSTM) model that takes the final depression decision. A modified version of the Local Interpretable Model-agnostic Explanations (LIME) is further used to identify the impact of the low-level spectrotemporal vowel variation on the decisions and observe the high-level temporal change of the depression likelihood. The proposed method outperforms baselines that model the spectrotemporal information in speech without integrating the vowel-based information, as well as ML models trained with conventional prosodic and spectrotemporal features. The conducted explainability analysis indicates that spectrotemporal information corresponding to non-vowel segments less important than the vowel-based information. Explainability of the high-level information capturing the segment-by-segment decisions is further inspected for participants with and without depression. The findings from this work can provide the foundation toward knowledge-driven interpretable decision-support systems that can assist clinicians to better understand fine-grain temporal changes in speech data, ultimately augmenting mental health diagnosis and care. | ['Theodora Chaspari', 'Kexin Feng'] | 2022-10-05 | null | null | null | null | ['vowel-classification'] | ['audio'] | [ 2.70248353e-01 3.34878236e-01 -3.31621498e-01 -8.36257756e-01
-6.48276508e-01 -1.89687267e-01 2.28741154e-01 5.74418128e-01
-1.66946128e-01 3.74901175e-01 8.45378816e-01 -2.31614590e-01
-4.72920984e-01 -3.82457495e-01 -2.60577053e-01 -5.97682416e-01
-1.17259659e-01 1.52586073e-01 -3.99442464e-01 -2.95517981e-01
2.50912160e-01 4.06481594e-01 -1.83749485e+00 9.72650528e-01
1.00179398e+00 8.16334248e-01 3.55165899e-01 6.60132587e-01
-3.04738879e-01 4.69749272e-01 -4.59961265e-01 8.49711746e-02
-5.71867943e-01 -6.01921558e-01 -5.98884583e-01 -3.74389754e-04
7.79040679e-02 -2.62682915e-01 -1.02601141e-01 8.81287336e-01
5.97851932e-01 5.96515574e-02 7.44604886e-01 -7.92288482e-01
-1.04521501e+00 7.83895135e-01 -2.56165028e-01 4.82440770e-01
2.58768588e-01 7.08993003e-02 6.76735878e-01 -8.23531270e-01
2.59299219e-01 1.52448666e+00 8.18573952e-01 4.76684362e-01
-1.01025438e+00 -3.25430274e-01 3.57428044e-01 6.64225042e-01
-1.02347469e+00 -4.67109174e-01 7.00258017e-01 -9.04284179e-01
1.59819317e+00 -5.59937151e-04 6.69867456e-01 1.19324768e+00
5.37799895e-01 2.33500227e-01 8.52068007e-01 -4.70562428e-01
-5.67197315e-02 -6.84849396e-02 5.15586853e-01 8.57046247e-01
-3.26013148e-01 3.30909759e-01 -9.68069851e-01 1.60012439e-01
1.66669667e-01 4.71292548e-02 1.75422952e-01 6.38867140e-01
-7.43251622e-01 8.49327028e-01 2.97621250e-01 5.24962723e-01
-6.61569655e-01 9.49808881e-02 6.67767823e-01 3.24557334e-01
8.39437544e-01 -6.10040724e-02 -9.35214162e-01 -4.10522610e-01
-9.78181064e-01 -4.98763733e-02 1.23826869e-01 2.86571950e-01
6.07741177e-01 3.86373937e-01 -5.97613990e-01 1.10438704e+00
5.88200986e-01 4.41489875e-01 8.66998017e-01 -9.58348870e-01
3.83128077e-01 5.08898854e-01 -4.04146045e-01 -9.74304616e-01
-1.06467426e+00 -2.45161861e-01 -5.24266124e-01 -1.62242055e-02
6.11598939e-02 -1.84144631e-01 -1.02450132e+00 1.92792404e+00
5.02204448e-02 -1.80213496e-01 1.03144147e-01 5.77658474e-01
1.01172829e+00 3.91131222e-01 3.74347717e-02 -6.25526726e-01
1.57713091e+00 -7.88261771e-01 -1.32813990e+00 -2.54926533e-01
7.78994441e-01 -4.64890003e-01 9.85533416e-01 2.53003240e-01
-8.72631371e-01 -7.60519207e-01 -8.26463878e-01 -2.85756618e-01
-4.65887576e-01 3.41775864e-01 3.61196399e-01 6.09720647e-01
-9.40355539e-01 6.31703496e-01 -9.45066452e-01 -4.93402332e-01
2.60956496e-01 4.45909083e-01 -2.65531182e-01 5.57802856e-01
-1.28297782e+00 9.44443107e-01 1.11361623e-01 3.99672896e-01
-6.47670507e-01 -7.82660306e-01 -1.01740432e+00 5.09109646e-02
-3.84375393e-01 -2.95682460e-01 1.16092503e+00 -1.05210996e+00
-1.55500579e+00 7.38302886e-01 -6.72879517e-01 -5.12272716e-01
4.91200238e-02 1.26554608e-01 -5.43127596e-01 2.96115190e-01
1.35045454e-01 5.20520806e-01 7.60158479e-01 -6.60924375e-01
-4.55195606e-01 -8.16708565e-01 -5.19577563e-01 6.45155162e-02
-3.91424358e-01 9.78300646e-02 1.97674707e-01 -6.68480635e-01
3.60107183e-01 -7.49054074e-01 3.54951262e-01 -1.75912216e-01
-5.88564873e-01 -4.82563257e-01 6.61731005e-01 -1.13344920e+00
1.39244616e+00 -2.22010708e+00 -2.14196682e-01 -1.96313381e-01
2.19354182e-01 1.92144975e-01 -1.15016565e-01 5.08530974e-01
-3.61060560e-01 2.72254646e-01 -3.26351583e-01 -5.86591542e-01
2.46385057e-02 1.24322921e-01 5.32138199e-02 4.56848353e-01
5.28601229e-01 6.71702683e-01 -6.45427883e-01 -5.01214378e-02
1.37610614e-01 6.56898916e-01 -6.58713460e-01 -2.44141538e-02
5.91722988e-02 6.49950087e-01 -7.54923150e-02 5.45922697e-01
3.76925170e-01 4.89580631e-01 -5.14947474e-02 -1.32107243e-01
-4.11926001e-01 6.59666240e-01 -3.77602190e-01 1.53644037e+00
-3.80025864e-01 1.09112692e+00 -1.26646519e-01 -1.21339893e+00
1.01116312e+00 5.10396540e-01 2.07789555e-01 -9.39575613e-01
3.18525061e-02 3.09336632e-01 5.42896032e-01 -1.31208062e+00
-7.20273703e-02 -2.77638763e-01 6.42674565e-02 4.22762670e-02
-1.38736948e-01 6.79231957e-02 -3.10224652e-01 -4.78988171e-01
9.99013603e-01 4.06978391e-02 1.33455232e-01 -1.63556933e-01
4.85379428e-01 -2.72532672e-01 7.36576378e-01 3.99273843e-01
-5.37568927e-01 6.61780715e-01 6.52573287e-01 -3.47068489e-01
-7.80448079e-01 -8.29307914e-01 -5.41860282e-01 1.34003365e+00
-4.31432605e-01 -4.71648909e-02 -8.75870764e-01 -1.96676984e-01
1.47739366e-01 8.80275428e-01 -1.12193537e+00 -6.86752558e-01
-1.24801002e-01 -5.34102142e-01 8.06506276e-01 5.83902836e-01
2.08866477e-01 -1.56657565e+00 -4.88755792e-01 2.71003157e-01
-2.48846516e-01 -7.94155955e-01 -3.44026148e-01 6.63662195e-01
-7.23336518e-01 -5.49617171e-01 -2.34683633e-01 -7.81953037e-01
2.49110207e-01 -3.33502114e-01 3.79155308e-01 -1.92085207e-01
-3.74296099e-01 1.28640473e-01 -2.88022757e-01 -7.82871425e-01
-5.13990104e-01 -1.76203549e-01 3.27816308e-01 5.66390269e-02
6.00826085e-01 -6.41173065e-01 -4.94165242e-01 -2.72020906e-01
-5.57695031e-01 -2.30488181e-01 2.78742135e-01 8.25916767e-01
7.44397700e-01 -8.83657672e-03 1.16815698e+00 -4.89406824e-01
8.67420733e-01 -7.45259464e-01 1.83835790e-01 -7.36770108e-02
-6.83785975e-01 1.66464701e-01 3.04279894e-01 -4.97208595e-01
-1.19377017e+00 -3.12723577e-01 -4.94957417e-01 -2.60978818e-01
-6.01749778e-01 8.25148582e-01 1.07644461e-01 7.67468989e-01
5.27653694e-01 1.15886308e-01 1.64601699e-01 -5.47166705e-01
1.48235308e-02 1.17495060e+00 4.30265903e-01 -4.46270183e-02
-9.30311456e-02 3.02474737e-01 -3.08551341e-01 -1.13258040e+00
-9.05383885e-01 -1.95667595e-01 -8.23246956e-01 -5.33429198e-02
1.38206017e+00 -6.04020298e-01 -6.47532225e-01 5.11584938e-01
-1.42085135e+00 -3.75284523e-01 -1.41596943e-01 7.10835099e-01
-7.79761910e-01 -2.07155928e-01 -5.59692919e-01 -1.20105302e+00
-3.50695610e-01 -1.19389153e+00 1.16015267e+00 1.59082651e-01
-6.47999048e-01 -1.09495378e+00 2.13798821e-01 6.70475245e-01
1.20661236e-01 2.75776595e-01 1.58508241e+00 -9.65325475e-01
3.43837023e-01 2.37788618e-01 4.90363315e-03 2.93452173e-01
4.22583222e-01 3.91825140e-01 -1.38059688e+00 3.30496907e-01
9.65909883e-02 -1.89162701e-01 9.94886220e-01 1.17455566e+00
1.07010257e+00 -1.98860914e-01 3.10845319e-02 4.63870347e-01
8.84090245e-01 7.37497747e-01 2.13377565e-01 1.55131891e-01
5.58551610e-01 1.11773169e+00 4.57775712e-01 4.88589376e-01
6.96126163e-01 6.17174327e-01 3.11838359e-01 -1.95155799e-01
-3.95048827e-01 -1.38238519e-02 5.68027616e-01 9.48144853e-01
8.82147402e-02 8.23910236e-02 -8.91124904e-01 8.32204878e-01
-1.75772405e+00 -1.03321981e+00 -3.73643816e-01 1.72841108e+00
8.13923359e-01 -1.53753022e-02 6.27947822e-02 3.51533294e-01
7.14752913e-01 -7.40603283e-02 -9.33326185e-01 -1.42230916e+00
-8.78435448e-02 1.22562319e-01 -3.20674062e-01 5.05782723e-01
-7.16791034e-01 7.85218120e-01 6.27443457e+00 4.33317035e-01
-1.30312717e+00 2.27376074e-01 6.16649747e-01 -4.20467556e-02
-3.87231976e-01 -6.71707749e-01 -3.93477350e-01 4.76765662e-01
1.54769027e+00 1.36789262e-01 4.43140775e-01 3.79638314e-01
1.50030196e+00 -6.56558871e-02 -1.04356217e+00 6.09967709e-01
-1.15421019e-01 -9.71601963e-01 -2.49165565e-01 5.38139939e-02
4.57489699e-01 1.44109964e-01 4.81058717e-01 9.25129578e-02
-5.73024988e-01 -1.22584248e+00 9.04442310e-01 1.08528984e+00
8.26062679e-01 -9.90559816e-01 6.92054749e-01 2.18003988e-01
-1.15404916e+00 -5.20912230e-01 -2.24956572e-01 -3.12204897e-01
-3.08719963e-01 3.99578094e-01 -1.22755933e+00 4.47203219e-02
7.20054150e-01 7.86853850e-01 -1.76711082e-01 4.25936043e-01
-1.44573271e-01 8.32792819e-01 2.11030796e-01 2.13508233e-01
1.35070220e-01 -7.90374540e-03 3.17535341e-01 1.31109726e+00
3.88938308e-01 1.02850243e-01 -4.19960767e-01 1.05701935e+00
2.77477324e-01 1.21839896e-01 -7.15898693e-01 -4.35379386e-01
6.19510531e-01 7.59902477e-01 -5.69161117e-01 4.56675030e-02
-4.71692771e-01 8.65347445e-01 2.92949796e-01 3.58619332e-01
-5.22628963e-01 -1.96506888e-01 1.10141242e+00 -9.66142938e-02
2.10439235e-01 6.00096770e-02 -6.65778399e-01 -3.36121649e-01
-1.98785156e-01 -8.44034493e-01 2.63324410e-01 -8.88669968e-01
-1.06692159e+00 6.55367732e-01 -1.89515874e-01 -6.81045771e-01
-5.75237870e-01 -2.12298915e-01 -7.50443399e-01 1.17610729e+00
-1.55470181e+00 -1.06542361e+00 6.95497543e-03 4.22720969e-01
1.12723470e+00 -1.58114791e-01 1.05826235e+00 -7.54127279e-02
-8.67477179e-01 6.21253014e-01 1.35144606e-01 -2.51313541e-02
5.60318768e-01 -1.10456860e+00 -6.09523356e-02 5.04547119e-01
-6.96603954e-02 6.32350922e-01 6.21955633e-01 -7.62154460e-01
-5.97865105e-01 -1.06954360e+00 1.44332767e+00 -2.62835711e-01
5.32555580e-01 7.58947358e-02 -1.05822146e+00 3.89641762e-01
1.01020329e-01 -1.86118394e-01 9.53411698e-01 1.51494876e-01
-2.80017126e-02 -2.01429889e-01 -1.29017079e+00 1.24401622e-01
9.05074835e-01 -1.16365445e+00 -8.67885053e-01 2.08488926e-01
1.12950754e+00 5.70574682e-03 -8.63888144e-01 3.08763593e-01
8.43804061e-01 -1.18314385e+00 5.66340566e-01 -7.38861501e-01
7.06141591e-01 2.02925071e-01 -2.33480066e-01 -1.41739357e+00
-2.56821573e-01 -2.08274305e-01 8.67917314e-02 1.28928339e+00
6.80279136e-01 -5.67444682e-01 3.22449297e-01 4.70000178e-01
-7.14649439e-01 -9.11774099e-01 -1.02676535e+00 -4.70941842e-01
1.61792710e-01 -9.04513776e-01 5.43397784e-01 6.57106280e-01
2.69561708e-01 5.99641912e-02 -6.40181359e-03 4.90379214e-01
9.99709032e-03 -2.40443617e-01 -3.55142444e-01 -1.25663590e+00
1.25961438e-01 -5.18206656e-01 -5.20109713e-01 -2.21360810e-02
4.28065568e-01 -9.87262011e-01 2.42407769e-01 -1.77943647e+00
1.14989705e-01 5.77971302e-02 -2.91178286e-01 6.40324891e-01
-1.50493294e-01 -1.22416802e-02 -2.19433770e-01 -1.41385481e-01
2.29889557e-01 6.85410440e-01 1.15913832e+00 -1.69948325e-01
-6.88904345e-01 2.47707479e-02 -6.55618906e-01 1.14904511e+00
9.06308174e-01 -5.28584182e-01 -7.29235470e-01 -4.03734535e-01
-1.93874657e-01 4.15641218e-01 9.03616697e-02 -7.47756898e-01
-1.13813683e-01 -2.56667197e-01 2.83170909e-01 -7.48279095e-01
4.71521825e-01 -3.58574092e-01 -3.41926932e-01 5.75160801e-01
-8.54982734e-01 1.86296552e-02 3.74329656e-01 3.96602005e-01
-1.53178707e-01 -1.47987723e-01 6.65415645e-01 1.95432171e-01
-3.78860533e-01 1.70847252e-01 -1.01669061e+00 -2.25009829e-01
6.02213621e-01 -3.04818928e-01 -2.39577830e-01 -4.28854018e-01
-1.47557247e+00 -2.61563361e-01 -3.53648216e-01 5.27076483e-01
6.95277035e-01 -1.25561690e+00 -6.10387266e-01 4.76538748e-01
6.81355745e-02 -6.93660557e-01 6.59443319e-01 1.19666314e+00
-1.35534108e-01 7.35954881e-01 -1.20248705e-01 -5.95979273e-01
-1.36532950e+00 2.40842938e-01 6.99513972e-01 1.75711975e-01
-5.47421575e-01 1.19575155e+00 1.77878171e-01 -4.68019843e-01
4.86775756e-01 -9.64771092e-01 -7.44702995e-01 6.65509582e-01
5.45889854e-01 4.87729132e-01 2.09450260e-01 -8.73879850e-01
-6.47016585e-01 4.59771872e-01 3.05179000e-01 -1.91945106e-01
1.66957831e+00 -7.09238946e-01 -1.44730255e-01 1.07203579e+00
1.21873140e+00 -2.19519094e-01 -1.27340472e+00 6.39419258e-02
1.06535040e-01 3.16233128e-01 2.38004848e-01 -9.87012684e-01
-8.77449632e-01 1.25510073e+00 1.21243930e+00 4.21782702e-01
1.10514259e+00 -6.21068850e-03 5.68093359e-01 2.32013881e-01
-3.82428914e-01 -1.34171474e+00 1.94185063e-01 4.79907393e-01
1.08621335e+00 -1.14844298e+00 -5.84250927e-01 -2.97779981e-02
-7.45322824e-01 1.41907978e+00 2.65947610e-01 3.18028331e-01
8.41359258e-01 3.07430804e-01 4.77032185e-01 -3.44740570e-01
-8.44904184e-01 -5.39753616e-01 5.11064351e-01 1.00680578e+00
6.63169682e-01 2.29718059e-01 -3.38674963e-01 1.18592918e+00
-4.39266443e-01 -3.17172825e-01 3.93762559e-01 2.88169026e-01
-6.50837302e-01 -9.00558531e-01 -2.66775817e-01 5.85265040e-01
-4.66808796e-01 -4.04316783e-01 -3.82720351e-01 3.07879448e-01
8.53234708e-01 1.66280532e+00 4.67419356e-01 -4.53463227e-01
3.13273609e-01 6.21689558e-01 9.55727920e-02 -9.43834960e-01
-6.15404546e-01 1.53978929e-01 3.08902770e-01 -4.69490021e-01
-4.60872978e-01 -1.02397394e+00 -1.98429048e+00 4.55299355e-02
-1.17003836e-03 -3.26406449e-01 8.78156960e-01 1.60513186e+00
4.35653061e-01 1.06315660e+00 6.36789739e-01 -5.75131536e-01
-5.26704602e-02 -1.13473558e+00 -6.90928459e-01 6.21811487e-02
1.13245940e+00 -6.48252010e-01 -3.57615858e-01 2.14335591e-01] | [13.868674278259277, 5.647998809814453] |
eab0a0d3-1979-42ea-8a24-bf1350fb24df | target-driven-one-shot-unsupervised-domain | 2305.04628 | null | https://arxiv.org/abs/2305.04628v1 | https://arxiv.org/pdf/2305.04628v1.pdf | Target-driven One-Shot Unsupervised Domain Adaptation | In this paper, we introduce a novel framework for the challenging problem of One-Shot Unsupervised Domain Adaptation (OSUDA), which aims to adapt to a target domain with only a single unlabeled target sample. Unlike existing approaches that rely on large labeled source and unlabeled target data, our Target-driven One-Shot UDA (TOS-UDA) approach employs a learnable augmentation strategy guided by the target sample's style to align the source distribution with the target distribution. Our method consists of three modules: an augmentation module, a style alignment module, and a classifier. Unlike existing methods, our augmentation module allows for strong transformations of the source samples, and the style of the single target sample available is exploited to guide the augmentation by ensuring perceptual similarity. Furthermore, our approach integrates augmentation with style alignment, eliminating the need for separate pre-training on additional datasets. Our method outperforms or performs comparably to existing OS-UDA methods on the Digits and DomainNet benchmarks. | ['Vittorio Murino', 'Alessio Del Bue', 'Pietro Morerio', 'Suvarna Kishorkumar Kadam', 'Julio Ivan Davila Carrazco'] | 2023-05-08 | null | null | null | null | ['unsupervised-domain-adaptation'] | ['methodology'] | [ 4.84745294e-01 2.07814336e-01 -3.01998764e-01 -6.76840663e-01
-9.59999681e-01 -6.50481462e-01 8.33297729e-01 -2.07220986e-02
-3.24111789e-01 7.65217662e-01 1.94000974e-01 -2.62530651e-02
3.89613152e-01 -5.80945551e-01 -5.69248974e-01 -5.12887836e-01
6.14569068e-01 8.32372248e-01 4.74458545e-01 -3.04076523e-01
-7.56597221e-02 2.25483134e-01 -1.22124159e+00 5.48389331e-02
8.89216542e-01 1.01323497e+00 1.81956589e-01 4.84121770e-01
-4.49115247e-01 5.83167017e-01 -4.97880220e-01 -2.60813087e-01
4.33361530e-01 -7.74390042e-01 -8.71469080e-01 3.96072090e-01
3.70383918e-01 -2.72645593e-01 -1.47028938e-01 9.86913800e-01
5.93463540e-01 4.64232087e-01 1.00212526e+00 -1.20228767e+00
-8.31469476e-01 3.47857416e-01 -5.08378446e-01 -5.04446402e-02
2.00700358e-01 2.17058375e-01 6.91309452e-01 -9.72605824e-01
9.41270530e-01 1.16293275e+00 5.42584538e-01 1.02545440e+00
-1.52304602e+00 -5.51343262e-01 3.12368423e-01 3.87482494e-02
-1.09359276e+00 -6.26226544e-01 1.01858819e+00 -6.13489926e-01
4.93667752e-01 -4.09308851e-01 2.34930754e-01 1.46265233e+00
-5.07039487e-01 7.84556210e-01 1.14095688e+00 -7.99596786e-01
8.48249316e-01 3.93085897e-01 1.18548691e-01 2.30887786e-01
-1.35946423e-01 1.54323190e-01 -3.46492678e-01 -1.32951111e-01
6.90167367e-01 -3.15607816e-01 1.30270407e-01 -9.91608739e-01
-1.06432164e+00 8.32690895e-01 2.57515430e-01 6.37122104e-03
-2.22465694e-01 -4.39685166e-01 5.19697368e-01 4.39036161e-01
5.66254914e-01 4.06263798e-01 -5.69605052e-01 -2.78278310e-02
-7.91665614e-01 9.56095457e-02 7.04594374e-01 1.44904506e+00
9.25311387e-01 3.44013184e-01 -3.18112105e-01 1.11204052e+00
1.84981957e-01 5.52693367e-01 7.61944294e-01 -7.75444269e-01
3.39709967e-01 6.75700724e-01 1.75729617e-01 -1.15948848e-01
9.07847956e-02 -1.79140046e-01 -6.17248118e-01 4.71253395e-01
7.06438184e-01 -2.66616821e-01 -1.35315061e+00 2.01849079e+00
5.51870108e-01 1.02508619e-01 4.00787175e-01 6.26336157e-01
5.74730754e-01 3.97123009e-01 1.97888479e-01 -5.06623499e-02
9.55424428e-01 -1.13623154e+00 -4.78651553e-01 -5.88136792e-01
4.72120583e-01 -8.00348461e-01 1.45988321e+00 6.58144504e-02
-8.48280311e-01 -8.38590860e-01 -1.17613733e+00 3.38947587e-02
-4.81682777e-01 1.45306960e-01 2.11686164e-01 6.09684408e-01
-7.63823211e-01 4.45838690e-01 -4.74952936e-01 -7.27640569e-01
6.42954350e-01 1.92334414e-01 -3.45266372e-01 -2.76168793e-01
-9.12941515e-01 7.77151942e-01 5.97304702e-01 -7.06581175e-01
-9.27389920e-01 -7.44193792e-01 -1.04215264e+00 -2.55928576e-01
3.58829111e-01 -5.91560662e-01 1.63490093e+00 -1.54911804e+00
-2.03494143e+00 1.01726794e+00 -1.82890594e-01 -4.79846060e-01
5.30296624e-01 -1.26307413e-01 -3.43928486e-01 -1.23911768e-01
2.91196555e-01 8.24175894e-01 1.06078053e+00 -1.30669105e+00
-4.54779357e-01 -3.27617913e-01 -2.04196677e-01 3.18966329e-01
-4.29010332e-01 -1.34193018e-01 -4.78626996e-01 -9.27698016e-01
-1.97436824e-01 -8.04819107e-01 -3.92196804e-01 7.46881440e-02
-1.76523119e-01 3.11458688e-02 9.06501532e-01 -3.51403743e-01
8.40898395e-01 -2.47837090e+00 2.09050432e-01 1.23417757e-01
-5.44328019e-02 3.56776625e-01 -5.37006676e-01 1.01665035e-01
-1.20296732e-01 -4.78419602e-01 -5.38162827e-01 -3.89034092e-01
3.56292687e-02 1.61732107e-01 -4.44535762e-01 2.00231269e-01
4.31893498e-01 6.11134171e-01 -1.15725613e+00 -4.09328431e-01
8.98646712e-02 2.83406917e-02 -5.81521213e-01 5.89675248e-01
-6.25208199e-01 7.21665382e-01 -1.87590554e-01 5.50080478e-01
6.36046112e-01 -2.14375019e-01 2.70136118e-01 4.36183363e-02
2.52005786e-01 1.88702866e-01 -1.23704958e+00 2.04039288e+00
-3.71717006e-01 3.74263883e-01 -1.46350011e-01 -9.11584258e-01
1.39895010e+00 2.48890758e-01 3.24455023e-01 -6.15596831e-01
3.10356796e-01 2.49283537e-01 -8.65142494e-02 -1.25778159e-02
2.54148334e-01 -3.14886391e-01 -2.49462515e-01 5.81546605e-01
6.71295166e-01 -4.14439589e-02 1.93177700e-01 3.01157326e-01
9.16773021e-01 6.52207196e-01 6.00897908e-01 -2.96563804e-01
5.24702311e-01 2.80424863e-01 7.38698959e-01 5.90335667e-01
-4.00008678e-01 8.23159933e-01 1.73081547e-01 -2.82492250e-01
-1.44998646e+00 -1.30771148e+00 1.74949721e-01 1.28162217e+00
2.28064612e-01 5.10124937e-02 -8.63238454e-01 -1.00893688e+00
-1.29029840e-01 8.22376668e-01 -8.08357239e-01 -3.00800741e-01
-3.08430403e-01 -2.42351905e-01 3.36607039e-01 8.56424749e-01
5.26063323e-01 -1.04738700e+00 -2.16008931e-01 2.77646750e-01
-6.12384975e-02 -1.19025624e+00 -8.17153752e-01 2.75269628e-01
-8.49587798e-01 -8.90505552e-01 -8.52112412e-01 -1.01303732e+00
8.86777222e-01 3.85369547e-02 1.06361699e+00 -7.54713178e-01
9.44313221e-03 4.89616305e-01 -4.01545584e-01 -5.83495975e-01
-7.48050511e-01 1.76865622e-01 1.53411314e-01 1.65259466e-01
6.01621568e-01 -6.78636134e-01 -1.98815569e-01 3.89724255e-01
-7.35921562e-01 5.36435843e-02 6.27294421e-01 9.06542540e-01
7.40557611e-01 -4.80786026e-01 8.20805192e-01 -1.24598253e+00
3.73796493e-01 -4.87524867e-01 -5.09300113e-01 1.53569087e-01
-5.60582578e-01 1.39123634e-01 8.70916486e-01 -1.00970364e+00
-1.41272485e+00 5.38638592e-01 1.51506007e-01 -7.73206830e-01
-6.60223007e-01 1.13153213e-03 -6.33473814e-01 9.20675844e-02
1.22008812e+00 2.32971430e-01 1.12362333e-01 -6.90098941e-01
6.90267563e-01 6.91293657e-01 8.29119742e-01 -7.30389833e-01
1.09544373e+00 4.91919547e-01 -4.40688282e-01 -6.72257006e-01
-8.88044715e-01 -6.10980153e-01 -1.27800953e+00 -5.51469019e-03
5.53619683e-01 -9.46359634e-01 2.95371860e-01 6.37772858e-01
-9.87313688e-01 -6.18012071e-01 -9.96698081e-01 3.51944149e-01
-8.60415220e-01 2.18832225e-01 -2.50073045e-01 -6.07667804e-01
-2.61668175e-01 -8.17332745e-01 7.63915718e-01 2.80737966e-01
-4.85307366e-01 -1.13232350e+00 3.84436488e-01 2.63830777e-02
3.65133375e-01 5.91099486e-02 7.49918401e-01 -1.29685283e+00
-1.03728831e-01 -1.57283261e-01 -1.01891376e-01 5.30287147e-01
4.93865281e-01 -4.29999739e-01 -1.14954126e+00 -3.88589799e-01
-1.13282457e-01 -6.12867951e-01 4.77031201e-01 2.13417247e-01
6.86450601e-01 -9.88965929e-02 -1.11925758e-01 6.10605240e-01
1.30778027e+00 4.35864836e-01 4.81395334e-01 4.38938797e-01
6.45479858e-01 4.47972655e-01 7.22537339e-01 3.66192609e-01
1.57234982e-01 4.76670384e-01 -3.59812640e-02 -2.28048503e-01
-4.10819530e-01 -4.21711922e-01 3.71297330e-01 4.67984736e-01
2.46306717e-01 3.08511686e-02 -7.62227476e-01 9.03419018e-01
-1.68160534e+00 -7.85504103e-01 4.58274066e-01 2.31906056e+00
1.06560874e+00 2.02346057e-01 5.72613597e-01 -7.78650045e-02
8.12261403e-01 -6.16745576e-02 -1.13548601e+00 -2.36488879e-01
-3.74327861e-02 3.73629719e-01 3.53441417e-01 3.58066380e-01
-1.22200429e+00 1.13028967e+00 6.61559439e+00 7.63531208e-01
-9.48524237e-01 1.51392102e-01 4.53832388e-01 4.19680476e-02
-1.94089010e-01 -2.39620358e-02 -7.29445815e-01 3.59968394e-01
7.62306273e-01 -3.25378388e-01 3.88821542e-01 1.34835827e+00
-1.34357870e-01 1.94665864e-01 -1.28905940e+00 6.15823984e-01
1.35597736e-01 -1.06127000e+00 7.84914643e-02 -1.38441473e-01
9.12601352e-01 -1.10681601e-01 -4.14928086e-02 5.01208127e-01
9.54555035e-01 -4.85154927e-01 6.14858389e-01 1.93521291e-01
1.13695824e+00 -6.00896835e-01 4.42602307e-01 1.51063085e-01
-9.83384430e-01 4.79232408e-02 -3.35239917e-01 6.23940267e-02
-3.88696231e-02 2.32257649e-01 -9.34315741e-01 4.45403159e-01
3.83374542e-01 7.80961871e-01 -4.95988399e-01 7.87561715e-01
-2.50885934e-01 5.88226974e-01 -9.21377093e-02 3.14093590e-01
7.27598518e-02 -8.27016532e-02 6.78877831e-01 1.02299821e+00
1.61214337e-01 -9.84448120e-02 4.86181706e-01 7.53509879e-01
-2.15300277e-01 3.10291797e-01 -6.59859180e-01 -5.18904217e-02
6.49692178e-01 1.00107956e+00 -5.47271132e-01 -7.66676903e-01
-5.23314714e-01 1.31513679e+00 3.92081410e-01 4.03893828e-01
-4.67952579e-01 -4.29489702e-01 6.89685106e-01 7.20682368e-02
5.00423908e-01 1.12807676e-01 -5.57983100e-01 -1.22880232e+00
-1.60356328e-01 -9.66629744e-01 5.43801069e-01 -6.26253068e-01
-1.71049142e+00 7.31305540e-01 -9.23020914e-02 -1.64738202e+00
-4.27360535e-01 -3.75498831e-01 -7.68090665e-01 1.05780900e+00
-1.42277479e+00 -1.37589729e+00 -2.34642029e-01 8.24064493e-01
9.78883684e-01 -7.02017665e-01 9.16789889e-01 1.20073758e-01
-4.18435991e-01 7.59459496e-01 2.43539602e-01 2.52266496e-01
1.18446255e+00 -1.38162661e+00 7.00993836e-01 9.39862967e-01
-1.16543435e-01 3.84028494e-01 7.10328102e-01 -7.17072904e-01
-8.12821686e-01 -1.28688979e+00 4.51931655e-01 -3.87145519e-01
6.98933482e-01 -3.47945541e-01 -1.14674473e+00 8.94324899e-01
2.21677959e-01 8.82845372e-02 9.20820355e-01 1.45155296e-01
-8.56045067e-01 -2.93226708e-02 -1.29721189e+00 5.13301909e-01
9.63226140e-01 -3.09732348e-01 -8.32288563e-01 3.55287152e-03
6.67834878e-01 -3.90761375e-01 -6.66505218e-01 2.21631721e-01
3.19733858e-01 -5.64430058e-01 8.29558372e-01 -8.95780027e-01
4.58606482e-01 -3.21737707e-01 -2.31024578e-01 -1.64718056e+00
-5.34893751e-01 -3.92719537e-01 -1.19644701e-01 1.49379075e+00
2.97345847e-01 -5.28732657e-01 7.57866621e-01 5.42566717e-01
-1.06059201e-01 -2.26526812e-01 -6.09158695e-01 -1.11819363e+00
2.64432967e-01 -1.12042069e-01 6.57093883e-01 1.12165487e+00
-1.72138557e-01 7.06815839e-01 -3.44645292e-01 8.67298096e-02
8.70094657e-01 6.48785233e-02 1.09693468e+00 -1.36242461e+00
-4.84308332e-01 -2.96205491e-01 -2.40713254e-01 -9.55601692e-01
1.40406460e-01 -8.21998537e-01 2.04337373e-01 -1.22559130e+00
1.17897496e-01 -4.20513391e-01 -3.49728912e-01 5.62684476e-01
-2.17376441e-01 1.82571068e-01 1.19464278e-01 2.74805248e-01
-5.86722493e-01 7.52076566e-01 1.06851900e+00 -2.37216294e-01
-6.31029785e-01 2.31071357e-02 -8.89570892e-01 8.41261983e-01
8.03263366e-01 -2.93173015e-01 -7.04942822e-01 -1.99518472e-01
-6.97932363e-01 -3.32578868e-01 3.67370509e-02 -1.10394621e+00
-2.12412216e-02 -3.45938474e-01 5.51532924e-01 -2.36485854e-01
2.72424102e-01 -6.73269928e-01 -2.47191668e-01 1.05586901e-01
-4.75389540e-01 -4.30328071e-01 3.53090644e-01 7.17827857e-01
-6.58409446e-02 -2.32749358e-01 1.49019718e+00 -6.38100430e-02
-1.09163022e+00 3.71622771e-01 -2.62437582e-01 4.37970400e-01
1.18527424e+00 -3.10536653e-01 -8.97069424e-02 -1.60062045e-01
-9.03046370e-01 1.84574470e-01 8.52192283e-01 5.11234701e-01
3.24395448e-01 -1.61191308e+00 -5.66016316e-01 3.61781657e-01
5.60091913e-01 2.07691193e-01 1.08433075e-01 1.86128020e-01
8.31055269e-02 -7.84870386e-02 -5.07898271e-01 -4.42326754e-01
-1.09504628e+00 8.60488892e-01 1.69266343e-01 -1.91506729e-01
-6.46495938e-01 7.40022779e-01 4.14167076e-01 -8.14464867e-01
2.43421763e-01 2.11719975e-01 -9.93330181e-02 -7.07663745e-02
7.08130360e-01 1.76491484e-01 -1.12042777e-01 -5.98508596e-01
-4.95447256e-02 3.26455176e-01 -3.94183487e-01 -3.01387906e-01
1.25057161e+00 -1.44771129e-01 4.43637192e-01 7.06965506e-01
8.83972347e-01 -2.30339199e-01 -1.64515209e+00 -9.46447194e-01
1.36293799e-01 -3.91404390e-01 -4.19738919e-01 -9.65064824e-01
-6.90994501e-01 8.14184487e-01 5.84899127e-01 -3.60541642e-01
1.20305049e+00 7.03253299e-02 6.49741590e-01 2.37244114e-01
2.15706795e-01 -1.35118151e+00 4.08691168e-01 5.82953095e-01
6.31656945e-01 -1.32720065e+00 -3.66179079e-01 -2.87209123e-01
-1.04953921e+00 9.01367962e-01 1.06525481e+00 -1.32321432e-01
4.70009476e-01 1.36997208e-01 4.18380588e-01 3.39274943e-01
-5.53353965e-01 -4.52039510e-01 3.31978530e-01 1.19941866e+00
9.78344008e-02 -1.52868599e-01 1.65157229e-01 6.39464855e-01
-2.99103628e-03 1.11672625e-01 2.88951069e-01 1.02853405e+00
-5.26204288e-01 -1.51424432e+00 -4.21562731e-01 2.90753514e-01
2.92462092e-02 5.58310002e-02 -5.63109756e-01 5.80623686e-01
-5.77570610e-02 5.89549959e-01 1.35105923e-01 -3.77061486e-01
5.88192761e-01 5.41432023e-01 4.49449629e-01 -1.03160608e+00
-1.44433036e-01 1.95967555e-01 -1.41340598e-01 -2.12260649e-01
-3.18192154e-01 -7.26078093e-01 -9.80703354e-01 1.25424802e-01
3.90591659e-02 -3.59452404e-02 3.36500853e-01 1.08564568e+00
3.51497382e-01 3.96108776e-01 7.85142541e-01 -8.14843357e-01
-6.47392273e-01 -1.08913505e+00 -5.55831969e-01 7.35897183e-01
2.59725630e-01 -8.02816808e-01 -6.78306147e-02 5.54686904e-01] | [10.127700805664062, 2.7854702472686768] |
547dbb40-b8c0-4137-983e-742b29cb5bd1 | label-cleaning-multiple-instance-learning | 2109.10778 | null | https://arxiv.org/abs/2109.10778v2 | https://arxiv.org/pdf/2109.10778v2.pdf | Label Cleaning Multiple Instance Learning: Refining Coarse Annotations on Single Whole-Slide Images | Annotating cancerous regions in whole-slide images (WSIs) of pathology samples plays a critical role in clinical diagnosis, biomedical research, and machine learning algorithms development. However, generating exhaustive and accurate annotations is labor-intensive, challenging, and costly. Drawing only coarse and approximate annotations is a much easier task, less costly, and it alleviates pathologists' workload. In this paper, we study the problem of refining these approximate annotations in digital pathology to obtain more accurate ones. Some previous works have explored obtaining machine learning models from these inaccurate annotations, but few of them tackle the refinement problem where the mislabeled regions should be explicitly identified and corrected, and all of them require a -- often very large -- number of training samples. We present a method, named Label Cleaning Multiple Instance Learning (LC-MIL), to refine coarse annotations on a single WSI without the need of external training data. Patches cropped from a WSI with inaccurate labels are processed jointly within a multiple instance learning framework, mitigating their impact on the predictive model and refining the segmentation. Our experiments on a heterogeneous WSI set with breast cancer lymph node metastasis, liver cancer, and colorectal cancer samples show that LC-MIL significantly refines the coarse annotations, outperforming state-of-the-art alternatives, even while learning from a single slide. Moreover, we demonstrate how real annotations drawn by pathologists can be efficiently refined and improved by the proposed approach. All these results demonstrate that LC-MIL is a promising, light-weight tool to provide fine-grained annotations from coarsely annotated pathology sets. | ['Aleksander S. Popel', 'Aaron W. James', 'Sintawat Wangsiricharoen', 'Carla Saoud', 'Jeremias Sulam', 'Zhenzhen Wang'] | 2021-09-22 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 6.53053939e-01 5.12647212e-01 -3.96834284e-01 -3.44635397e-01
-1.52158332e+00 -4.83864546e-01 -2.40288582e-02 7.02998161e-01
-5.13127744e-01 8.67842615e-01 -2.71024168e-01 -2.76926607e-01
-4.01366726e-02 -5.20132959e-01 -6.70832694e-01 -1.18418503e+00
3.33467454e-01 7.82359123e-01 4.61707860e-01 2.90777624e-01
2.35626139e-02 5.37802041e-01 -1.00282538e+00 3.70076537e-01
1.01002789e+00 7.92194426e-01 2.35674128e-01 6.76692903e-01
-2.94224888e-01 4.89423931e-01 -6.25489235e-01 -4.98573244e-01
7.44839236e-02 -1.98493391e-01 -9.58218873e-01 3.46395969e-01
2.83359468e-01 1.05060026e-01 4.75122064e-01 1.12716854e+00
1.15926974e-01 -4.36859608e-01 6.25646949e-01 -8.58791113e-01
-1.19465970e-01 5.95286548e-01 -1.01566434e+00 -7.15461597e-02
-3.67846161e-01 3.68463136e-02 7.94158816e-01 -6.07479036e-01
7.54925728e-01 6.58164740e-01 9.88035619e-01 6.49474204e-01
-1.25533366e+00 -7.26895213e-01 8.69832337e-02 -7.79505894e-02
-1.57607818e+00 -2.02499717e-01 3.65847886e-01 -3.91259104e-01
3.32963645e-01 5.55239618e-01 5.78011632e-01 4.93280739e-01
6.56459481e-02 8.73714149e-01 1.09412897e+00 -5.81024945e-01
1.35073498e-01 3.97999167e-01 3.06013465e-01 9.75580871e-01
4.67532128e-01 -5.27568102e-01 -7.88891837e-02 -2.54124582e-01
6.13774002e-01 3.66203338e-01 -4.66158152e-01 -1.67304128e-01
-1.19278371e+00 5.01014471e-01 3.41912359e-01 4.80455041e-01
-1.91892281e-01 -1.62336022e-01 3.89801085e-01 -7.09180683e-02
8.11398745e-01 5.16479015e-01 -5.91903150e-01 2.55161405e-01
-1.17960727e+00 -2.93334007e-01 6.00325823e-01 7.82630861e-01
9.13495064e-01 -6.99055910e-01 -2.17038855e-01 7.23903954e-01
-7.17283115e-02 3.34039301e-01 4.11255151e-01 -4.49704677e-01
1.41747193e-02 9.80301261e-01 1.54982448e-01 -6.16370142e-01
-6.58846259e-01 -7.80280650e-01 -1.08418941e+00 -3.02196145e-02
6.90403283e-01 2.62060086e-03 -1.19529903e+00 1.24376535e+00
6.40058219e-01 5.04276991e-01 -2.79198617e-01 7.72747636e-01
4.44481760e-01 1.23008318e-01 3.74959677e-01 -3.88249457e-01
1.49268818e+00 -1.00823736e+00 -6.51682258e-01 -2.49643698e-02
1.05534899e+00 -7.01426983e-01 8.62090766e-01 3.98896873e-01
-9.31609273e-01 -1.83997214e-01 -7.81251907e-01 -5.76482154e-02
-2.37745315e-01 6.33938074e-01 5.67751467e-01 4.83226120e-01
-9.87049103e-01 4.52708542e-01 -1.07705140e+00 -3.43968898e-01
8.84260416e-01 5.53613544e-01 -4.34603691e-01 -5.22225350e-02
-5.40103555e-01 7.46618629e-01 2.12118417e-01 1.83603168e-01
-5.47143459e-01 -1.25067484e+00 -6.73666716e-01 -6.26083761e-02
6.23531997e-01 -3.43854606e-01 1.26059830e+00 -9.09815967e-01
-1.01946557e+00 1.17691290e+00 -4.50033665e-01 -2.76425391e-01
4.75207567e-01 1.31793216e-01 -1.45557463e-01 1.45339072e-01
6.68741912e-02 6.20759606e-01 6.78172231e-01 -1.34057260e+00
-9.54673350e-01 -5.00644088e-01 -3.64078403e-01 -1.10310517e-01
-2.10977882e-01 -4.58188295e-01 -7.24969447e-01 -5.77999055e-01
1.04084544e-01 -9.66015160e-01 -6.62236452e-01 1.79590791e-01
-5.97904563e-01 -1.67604268e-01 7.46259809e-01 -8.40178609e-01
1.13124657e+00 -2.14919353e+00 -4.35094163e-02 5.29778898e-01
3.26066762e-01 5.65424025e-01 3.87696587e-02 -2.63858974e-01
2.13138059e-01 4.45752084e-01 -4.79582906e-01 -4.53438252e-01
-2.19465747e-01 3.33225369e-01 1.39377937e-01 5.47389030e-01
3.17235678e-01 7.84703255e-01 -9.97847974e-01 -9.28310573e-01
3.55876759e-02 3.31756204e-01 -4.17673349e-01 1.52877614e-01
-1.67843372e-01 6.94930315e-01 -3.24520379e-01 9.44168150e-01
6.70880497e-01 -6.91606164e-01 3.24908495e-01 -5.14712036e-01
2.38107443e-01 -3.53608340e-01 -1.07561707e+00 1.51743543e+00
-3.94100845e-01 2.03502297e-01 3.71285170e-01 -9.69100356e-01
6.98293984e-01 2.91690826e-01 5.91991842e-01 -2.56202191e-01
1.14711985e-01 4.54520881e-01 -3.19468766e-01 -6.31179929e-01
1.05182633e-01 -2.67566472e-01 4.15345542e-02 4.37577516e-01
-5.58050014e-02 -1.23220190e-01 3.82583112e-01 1.94783229e-02
1.24249291e+00 -1.81332022e-01 6.79980516e-01 -1.62539601e-01
7.49843121e-01 3.55016291e-01 1.00976098e+00 7.10682869e-01
-3.33387077e-01 5.13964176e-01 4.79707271e-01 -4.68026042e-01
-1.02540195e+00 -7.02277303e-01 -3.63200456e-01 8.55149090e-01
1.64185047e-01 -1.20968722e-01 -1.01708722e+00 -1.22744584e+00
-1.66505218e-01 1.62583098e-01 -6.99844480e-01 2.92337626e-01
-6.54765189e-01 -1.10985482e+00 4.24594641e-01 3.75611395e-01
1.40162989e-01 -7.26297140e-01 -3.30076456e-01 1.78299785e-01
-2.98189491e-01 -9.61620510e-01 -4.45828706e-01 2.29679987e-01
-9.03987408e-01 -1.50426495e+00 -7.85373330e-01 -9.89100635e-01
1.54502034e+00 1.16297536e-01 1.03274190e+00 5.35971224e-01
-7.97694266e-01 -1.26625866e-01 -4.01835948e-01 -5.02363861e-01
-5.58617294e-01 3.20254624e-01 -4.78220671e-01 1.69202521e-01
3.64511192e-01 -3.48013081e-02 -5.77077568e-01 4.00258183e-01
-1.11812949e+00 2.74671197e-01 1.11648118e+00 1.18971491e+00
1.33310139e+00 5.90574890e-02 5.74494362e-01 -1.77449787e+00
1.03623262e-02 -3.26245159e-01 -4.87946957e-01 5.69458961e-01
-3.79917026e-01 7.84976408e-02 4.65703458e-01 -3.40228111e-01
-1.14633918e+00 4.00211453e-01 -2.70225465e-01 -8.96630734e-02
-3.71446639e-01 3.36786419e-01 1.09647378e-01 -3.65763605e-01
5.92989326e-01 -1.75134584e-01 -2.75231004e-02 -3.88143122e-01
2.98334677e-02 6.05729163e-01 5.75597703e-01 -2.90607214e-01
5.60450256e-01 7.72742450e-01 8.06927457e-02 -4.75348502e-01
-1.20775032e+00 -9.69432473e-01 -7.63162255e-01 2.31263228e-02
6.73340917e-01 -5.74791372e-01 -3.35336059e-01 3.17504913e-01
-7.57964313e-01 -5.48473418e-01 -3.21725368e-01 2.16806054e-01
-1.09118879e-01 2.18953148e-01 -8.61492991e-01 -3.03793907e-01
-5.26753306e-01 -1.20402813e+00 1.53711176e+00 4.53875363e-01
-3.55620831e-01 -1.04202294e+00 -3.22793587e-03 5.80774903e-01
2.06990033e-01 4.19989288e-01 1.06252766e+00 -7.05103219e-01
-5.17829239e-01 -4.56898570e-01 -3.62488240e-01 -7.44447717e-03
3.70476216e-01 1.04407609e-01 -9.60816205e-01 -3.20002735e-01
-2.70877451e-01 -1.70229509e-01 8.42517495e-01 4.57048327e-01
1.42689526e+00 -2.79046237e-01 -8.98380041e-01 6.51641250e-01
1.51401389e+00 -5.46198562e-02 1.82449996e-01 1.39037505e-01
5.48431396e-01 5.18071175e-01 9.92927611e-01 3.14089864e-01
9.19026807e-02 2.20683172e-01 2.51984060e-01 -7.13027596e-01
-1.67733595e-01 2.53827244e-01 -3.23662251e-01 5.19781947e-01
8.54788348e-03 -1.18524628e-02 -9.38167810e-01 9.31879401e-01
-1.78748083e+00 -4.54939008e-01 -2.02414081e-01 1.84850645e+00
1.44112229e+00 -1.01582617e-01 -2.24244148e-01 3.05697590e-01
8.11833203e-01 -4.20893222e-01 -5.33748567e-01 7.12500960e-02
3.15744251e-01 5.14450073e-01 5.39015055e-01 5.84787071e-01
-1.21527565e+00 7.55454123e-01 5.56530905e+00 1.06631005e+00
-1.09758902e+00 3.81423205e-01 1.18117726e+00 -6.80048168e-02
-4.80366088e-02 -4.15780962e-01 -9.98576283e-01 2.46865138e-01
5.43193102e-01 2.91442722e-01 -2.48527646e-01 8.61226737e-01
1.00788459e-01 -4.70139921e-01 -1.05871785e+00 6.61134064e-01
-1.23712365e-02 -1.52254260e+00 -1.10599026e-01 4.99539897e-02
1.00337577e+00 -3.28348190e-01 -2.54090101e-01 6.44329265e-02
3.07540417e-01 -7.98437893e-01 2.13205948e-01 5.08697152e-01
9.99846578e-01 -6.83572710e-01 1.48703694e+00 4.90233332e-01
-1.06071603e+00 1.31484002e-01 -2.65611082e-01 6.32275224e-01
8.40453058e-03 1.16085267e+00 -1.35818172e+00 4.19984519e-01
5.86586714e-01 3.35240185e-01 -7.92632759e-01 1.18396127e+00
-1.36604890e-01 6.87534332e-01 -1.11818731e-01 1.09606698e-01
5.17296651e-03 2.66505927e-01 1.67515203e-02 1.59500563e+00
3.71959984e-01 2.53636867e-01 2.79781014e-01 5.90072572e-01
-8.25459957e-02 5.75714465e-03 1.79098204e-01 2.13447124e-01
3.05534869e-01 1.90141451e+00 -1.23379958e+00 -4.77799863e-01
-8.38335454e-02 7.06127703e-01 4.15191323e-01 1.96434543e-01
-8.86085033e-01 -1.63115442e-01 3.00590813e-01 1.59285143e-01
8.64857584e-02 4.03492630e-01 -6.17863476e-01 -7.85753012e-01
-1.40371323e-01 -6.32750392e-01 5.21813989e-01 -1.92321822e-01
-1.27812588e+00 5.45480132e-01 -3.24899822e-01 -1.13567662e+00
5.44626340e-02 -5.53425670e-01 -3.52050722e-01 5.57733893e-01
-1.90568638e+00 -1.30855322e+00 -6.68072879e-01 3.72539759e-01
4.76432353e-01 5.36002636e-01 7.98054457e-01 3.00900728e-01
-6.55477285e-01 8.44771445e-01 -9.75204855e-02 1.58873230e-01
9.86627877e-01 -1.37384009e+00 -2.16593251e-01 5.40143967e-01
-1.77981928e-01 2.96388209e-01 4.55932051e-01 -6.12414062e-01
-8.68683517e-01 -1.54706669e+00 8.93645048e-01 -1.60118774e-01
4.54575151e-01 -6.73179403e-02 -1.16937196e+00 5.35414398e-01
-1.84426934e-01 5.10421872e-01 1.08124363e+00 -2.99585517e-02
1.11613676e-01 -2.43597969e-01 -1.53649056e+00 3.71311367e-01
6.02266431e-01 -1.65685266e-01 -1.17723949e-01 5.23144484e-01
3.82528573e-01 -7.20502973e-01 -1.08398974e+00 4.91600662e-01
1.96734056e-01 -5.82252920e-01 7.32005656e-01 -1.12178981e-01
1.61169413e-02 -4.47603106e-01 6.25559092e-01 -1.26863408e+00
-2.34332860e-01 -2.11958095e-01 2.03672066e-01 1.21879578e+00
6.30624175e-01 -3.79962146e-01 1.14019191e+00 6.11219168e-01
-3.91848803e-01 -1.04726195e+00 -7.86703885e-01 -2.43153125e-01
-2.00762510e-01 1.65155698e-02 5.64565539e-01 8.91704679e-01
4.65615932e-03 -2.41552293e-01 9.49348956e-02 3.68367553e-01
7.87507713e-01 -1.14922365e-02 6.16416276e-01 -1.27551031e+00
-2.51547545e-01 -3.17103654e-01 -2.40779608e-01 -4.21054512e-01
-2.78568510e-02 -7.79656112e-01 4.05187428e-01 -1.41310918e+00
5.18337965e-01 -1.02229714e+00 -2.67778218e-01 9.00381684e-01
-6.53678000e-01 7.39813626e-01 -2.93241203e-01 3.41971189e-01
-7.33724296e-01 -2.81829298e-01 1.44949412e+00 -3.54016632e-01
6.39313012e-02 7.63723105e-02 -6.40705466e-01 9.88851786e-01
6.58254087e-01 -5.80325365e-01 -1.82135999e-02 -3.99749950e-02
3.73831987e-02 4.65473626e-03 2.69425541e-01 -9.05041337e-01
3.96347195e-01 -1.62242681e-01 5.08512199e-01 -5.94585180e-01
-4.16765781e-03 -9.12785470e-01 1.41325876e-01 5.41791260e-01
-3.01113576e-01 -5.72826087e-01 1.36764660e-01 6.76772535e-01
-2.67022669e-01 -4.00332958e-01 1.04634488e+00 -3.35542381e-01
-5.34242749e-01 4.07216996e-01 -1.24588415e-01 -1.56661898e-01
1.35091901e+00 -2.03797683e-01 -4.06415224e-01 3.99225235e-01
-9.99616683e-01 2.59359092e-01 3.80778611e-01 -2.34027237e-01
1.57098830e-01 -8.10733736e-01 -7.08101273e-01 2.10533261e-01
1.58184513e-01 7.73109853e-01 4.23341453e-01 1.23707783e+00
-6.52088225e-01 3.46276045e-01 2.27065384e-01 -9.15239155e-01
-1.49054408e+00 2.97730774e-01 2.04153866e-01 -1.12162316e+00
-4.22957569e-01 1.18647182e+00 3.27816308e-01 -4.41654563e-01
2.85439879e-01 -6.93270802e-01 -8.00878406e-02 2.27601007e-02
5.40145516e-01 2.21852273e-01 5.12157977e-01 -1.80297360e-01
-1.82775661e-01 5.56812704e-01 -5.45626998e-01 4.71301913e-01
1.18327379e+00 -8.18425268e-02 -2.32056737e-01 1.00259848e-01
8.26863170e-01 -7.85987172e-03 -1.25551152e+00 -4.10636544e-01
3.13513488e-01 -2.77068228e-01 -7.26252422e-02 -8.62265050e-01
-1.17827952e+00 6.15185678e-01 3.60648662e-01 1.77225899e-02
1.34898460e+00 1.70807958e-01 8.73195589e-01 3.62996832e-02
3.72506231e-01 -1.00116551e+00 -6.32225871e-02 1.23286145e-02
3.92139167e-01 -1.38990200e+00 7.20517114e-02 -8.59731078e-01
-4.19285148e-01 1.01037085e+00 6.55525446e-01 4.45166118e-02
4.87508923e-01 7.66212642e-01 2.12730691e-01 -6.97216392e-02
-5.32734215e-01 1.01465564e-02 1.53402269e-01 3.97146851e-01
3.33449244e-01 1.98645189e-01 -1.60772681e-01 6.58508241e-01
2.46922210e-01 1.52047098e-01 3.13522875e-01 9.02076304e-01
-5.80350637e-01 -1.14017653e+00 -5.00564933e-01 6.70934618e-01
-6.79233193e-01 7.86583424e-02 -2.98941463e-01 8.87650788e-01
4.95244801e-01 5.54292679e-01 4.72671315e-02 1.15072362e-01
2.21649520e-02 -6.38745651e-02 2.82057315e-01 -7.77846098e-01
-7.29317367e-01 2.00588480e-01 -1.52707025e-01 -3.68758917e-01
-5.74392021e-01 -5.23174882e-01 -1.60348046e+00 -8.39482620e-03
-6.91295147e-01 2.88383096e-01 4.75221187e-01 1.19923639e+00
2.38649473e-01 7.50603974e-01 3.96969229e-01 -7.24206567e-01
-2.96175480e-01 -8.62158835e-01 -5.75510800e-01 4.19991732e-01
5.17342210e-01 -3.53841633e-01 -3.77262115e-01 5.21215796e-01] | [14.990372657775879, -2.931941270828247] |
119e5b98-fdc0-49ec-80f0-5c85289b9a6f | inverse-preference-learning-preference-based | 2305.15363 | null | https://arxiv.org/abs/2305.15363v1 | https://arxiv.org/pdf/2305.15363v1.pdf | Inverse Preference Learning: Preference-based RL without a Reward Function | Reward functions are difficult to design and often hard to align with human intent. Preference-based Reinforcement Learning (RL) algorithms address these problems by learning reward functions from human feedback. However, the majority of preference-based RL methods na\"ively combine supervised reward models with off-the-shelf RL algorithms. Contemporary approaches have sought to improve performance and query complexity by using larger and more complex reward architectures such as transformers. Instead of using highly complex architectures, we develop a new and parameter-efficient algorithm, Inverse Preference Learning (IPL), specifically designed for learning from offline preference data. Our key insight is that for a fixed policy, the $Q$-function encodes all information about the reward function, effectively making them interchangeable. Using this insight, we completely eliminate the need for a learned reward function. Our resulting algorithm is simpler and more parameter-efficient. Across a suite of continuous control and robotics benchmarks, IPL attains competitive performance compared to more complex approaches that leverage transformer-based and non-Markovian reward functions while having fewer algorithmic hyperparameters and learned network parameters. Our code is publicly released. | ['Dorsa Sadigh', 'Joey Hejna'] | 2023-05-24 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [-7.31220022e-02 1.88527387e-02 -7.25177705e-01 -2.93864310e-01
-8.04864883e-01 -9.40403044e-01 3.69657218e-01 7.05105364e-02
-7.35681951e-01 1.00546408e+00 3.47209513e-01 -3.76736671e-01
-4.02800560e-01 -4.84067738e-01 -5.85050225e-01 -5.17195284e-01
-2.57606983e-01 6.47663116e-01 1.20996669e-01 -3.68660331e-01
6.02928638e-01 2.66672939e-01 -1.37451553e+00 -2.02386096e-01
6.73587620e-01 1.12839186e+00 2.63011187e-01 6.17011666e-01
3.89921516e-01 8.25949907e-01 8.15236289e-03 3.72535852e-03
5.16255498e-01 -2.82460183e-01 -1.09064281e+00 -4.11269873e-01
-6.77793240e-03 -6.24884546e-01 -3.79426539e-01 7.66156197e-01
6.47448182e-01 4.30615485e-01 4.82877761e-01 -1.36098051e+00
-5.70629239e-01 7.34204948e-01 -1.16092376e-01 -4.54515629e-02
4.19075489e-01 4.48046774e-01 1.69918644e+00 -4.50500160e-01
5.70912421e-01 1.23560262e+00 4.42451745e-01 7.22835720e-01
-1.53875792e+00 -5.69107473e-01 2.81283617e-01 4.04851973e-01
-9.05304611e-01 -2.58026361e-01 8.12476099e-01 -1.91537350e-01
8.98662806e-01 7.58218169e-02 8.30695093e-01 1.08971179e+00
8.37912187e-02 1.20734799e+00 1.27562237e+00 -4.11291346e-02
5.54874420e-01 -1.39759496e-01 -2.10478589e-01 6.48295879e-01
-6.91634193e-02 7.22865283e-01 -4.90323752e-01 -2.43039101e-01
1.00076175e+00 2.64576823e-01 -1.75132394e-01 -1.24423575e+00
-1.31101692e+00 1.05298197e+00 6.40113354e-01 -1.21701658e-01
-3.79253298e-01 7.24050939e-01 2.77872890e-01 5.28696895e-01
-1.64598167e-01 9.58729029e-01 -6.61773086e-01 -6.02474928e-01
-6.32559001e-01 6.26050234e-01 9.53649700e-01 9.76613224e-01
9.41370845e-01 -7.98243657e-02 -4.41771597e-01 6.30987823e-01
3.47570926e-01 4.96192366e-01 5.10771573e-01 -1.63454008e+00
2.43244484e-01 5.03026724e-01 7.09823072e-01 -6.49130821e-01
-5.24992764e-01 -3.27993184e-01 -1.09199077e-01 3.35062593e-01
3.97591054e-01 -1.39149725e-01 -8.78098547e-01 1.84279776e+00
7.59053007e-02 -2.16957077e-01 1.10247716e-01 1.15584302e+00
1.58993781e-01 2.68403113e-01 -4.15601917e-02 -8.83311033e-03
7.12307274e-01 -9.96169329e-01 -2.18559042e-01 -3.36402923e-01
5.05896389e-01 -3.88476580e-01 1.46674669e+00 3.99614334e-01
-9.38706458e-01 3.60449217e-02 -9.78315234e-01 5.39201424e-02
-5.50772361e-02 -2.28801310e-01 1.04246581e+00 3.94559473e-01
-1.04964375e+00 1.05192208e+00 -7.12511063e-01 -3.04572046e-01
3.61821473e-01 7.16412783e-01 -4.66543511e-02 -2.19517305e-01
-1.03451848e+00 1.04861236e+00 1.28748551e-01 -9.94478464e-02
-1.27414811e+00 -5.62098742e-01 -8.25888157e-01 -1.71359144e-02
7.90172100e-01 -4.70736206e-01 1.98418355e+00 -7.80447006e-01
-2.10688519e+00 2.44967088e-01 3.01078081e-01 -6.31650567e-01
4.19080257e-01 -4.65364695e-01 9.19234157e-02 7.47854412e-02
4.33431715e-02 9.41189110e-01 7.55800605e-01 -1.09872758e+00
-5.33774972e-01 -8.50815512e-03 2.91418552e-01 4.57489163e-01
-8.67874920e-02 -2.09923372e-01 -3.94409187e-02 -3.15607429e-01
-2.01283365e-01 -1.25437224e+00 -7.14090109e-01 1.54018417e-01
7.68841524e-03 -2.11623400e-01 5.05759656e-01 7.35600293e-03
9.49707389e-01 -1.93107975e+00 1.95965976e-01 4.29363608e-01
1.14174463e-01 -9.15424898e-02 -4.46091294e-01 5.41937649e-01
3.34243089e-01 -1.50378391e-01 -1.64938971e-01 4.44569662e-02
3.71733755e-01 7.30557084e-01 -4.18382525e-01 3.90971661e-01
1.32144941e-02 1.18201280e+00 -1.54091609e+00 -4.84038889e-01
2.18412966e-01 -5.70725389e-02 -1.26808882e+00 3.35785657e-01
-6.19981527e-01 4.92678344e-01 -6.33328497e-01 6.53770387e-01
-1.30504191e-01 -1.71717644e-01 4.30390477e-01 3.13912183e-01
-1.65320069e-01 5.97029626e-01 -1.16887689e+00 1.91856480e+00
-4.83503491e-01 2.15925515e-01 4.99122664e-02 -8.00034046e-01
9.31279898e-01 2.63461798e-01 8.38530302e-01 -8.84709835e-01
2.53020942e-01 5.17067373e-01 -3.27960961e-02 -2.24441931e-01
7.06981659e-01 -4.97367047e-03 -3.98079753e-01 6.74902618e-01
4.69674878e-02 -3.62856090e-01 1.48352623e-01 -1.76964507e-01
1.35170448e+00 6.16268337e-01 2.91438341e-01 -2.60006845e-01
3.54706682e-02 1.95294037e-01 7.09477723e-01 9.26124096e-01
-4.41584468e-01 1.37991011e-01 4.54906702e-01 -4.69188005e-01
-8.60432506e-01 -1.25316918e+00 3.82825196e-01 1.49902010e+00
1.34450808e-01 -2.83327222e-01 -2.01923519e-01 -6.13633871e-01
2.84960717e-01 6.68045759e-01 -4.89767671e-01 -2.13018924e-01
-6.47987545e-01 -2.21520886e-01 2.70102024e-01 6.30753279e-01
2.71437198e-01 -1.30870450e+00 -1.09075677e+00 4.29656923e-01
-2.70166378e-02 -5.76277137e-01 -8.73085976e-01 6.64762437e-01
-1.06583834e+00 -1.02795708e+00 -5.68284094e-01 -5.03482044e-01
5.05314231e-01 1.39897928e-01 1.25633502e+00 -2.79817343e-01
-7.88644180e-02 8.30919802e-01 -3.60151380e-01 -2.26033196e-01
7.34573603e-02 5.10731749e-02 2.07388088e-01 -4.25940603e-01
5.79331294e-02 -7.26307929e-01 -8.71107042e-01 3.69788319e-01
-5.24129093e-01 -4.65197831e-01 9.67508614e-01 1.13615382e+00
6.39643490e-01 -5.07710814e-01 5.06811321e-01 -5.65403998e-01
1.01088858e+00 -3.11233103e-01 -8.11898291e-01 1.27437608e-02
-1.08800459e+00 8.66522372e-01 6.05150938e-01 -5.79318702e-01
-6.20913148e-01 4.30268884e-01 2.87406832e-01 -6.23288333e-01
1.56946421e-01 4.18968499e-01 2.35668689e-01 -1.22909434e-01
7.10726976e-01 1.07290275e-01 1.72668830e-01 -3.06992918e-01
6.94800854e-01 1.99760810e-01 5.11575639e-01 -1.04291391e+00
5.36404490e-01 1.47424728e-01 -1.78865623e-02 -2.59866454e-02
-7.82495975e-01 -5.39847016e-01 -2.21086532e-01 -1.19957626e-01
3.78735602e-01 -4.77772117e-01 -1.25918579e+00 -5.69150895e-02
-6.45754576e-01 -8.47089589e-01 -6.59234524e-01 6.53961480e-01
-1.29714119e+00 7.82243907e-02 -4.45291162e-01 -9.21415210e-01
-1.67172790e-01 -1.18311357e+00 7.99680173e-01 5.84421530e-02
-4.16303426e-01 -7.14401424e-01 3.88934195e-01 -8.40780810e-02
6.22020543e-01 -2.73935169e-01 7.57844388e-01 -5.59222579e-01
-6.99413657e-01 1.35419190e-01 -7.77634680e-02 4.93163206e-02
-7.63335899e-02 -4.38490063e-01 -5.68662226e-01 -4.02231872e-01
-3.20820272e-01 -8.98035169e-01 6.67971730e-01 2.60954559e-01
1.02490175e+00 -5.21080434e-01 7.14056892e-03 3.69197279e-01
1.25845671e+00 1.36651382e-01 2.10191056e-01 5.04200041e-01
3.65157217e-01 4.40221488e-01 9.91691947e-01 6.58581138e-01
6.84140086e-01 6.56499863e-01 8.42607617e-01 3.55472058e-01
4.81498092e-01 -6.84576213e-01 5.93181908e-01 3.52233201e-01
-1.95877239e-01 1.85460329e-01 -6.81162179e-01 5.69194913e-01
-2.35828710e+00 -1.00790679e+00 5.94576299e-01 2.29992247e+00
1.18707514e+00 5.04334942e-02 4.23792779e-01 -2.55887322e-02
1.52703404e-01 -1.84135709e-03 -1.12366140e+00 -5.52041829e-01
4.13136750e-01 3.43003482e-01 9.68967974e-01 4.69527453e-01
-9.73653257e-01 1.08988750e+00 7.42603636e+00 4.18123633e-01
-1.12919974e+00 -2.75207758e-01 1.35127112e-01 -4.15229619e-01
-5.03658593e-01 1.88035339e-01 -5.86329937e-01 1.22163100e-02
8.48018885e-01 -1.11660525e-01 1.17019451e+00 1.20194244e+00
2.66478688e-01 -1.15590259e-01 -1.54665041e+00 1.06663716e+00
-4.12369370e-01 -1.10802710e+00 -4.03949618e-01 4.12087031e-02
4.24218774e-01 3.03815722e-01 2.98646510e-01 7.18590677e-01
1.24535632e+00 -1.11135793e+00 9.77969527e-01 4.41417634e-01
5.98390818e-01 -7.12007344e-01 3.55708629e-01 2.12381005e-01
-1.02777553e+00 -6.38338745e-01 -2.25604847e-01 -3.92812997e-01
-1.24948360e-01 1.08751347e-02 -9.14097190e-01 1.75416663e-01
7.04035044e-01 9.12953258e-01 -2.53001601e-01 1.12211943e+00
-3.97099823e-01 3.04333925e-01 -3.19194078e-01 -4.95368719e-01
6.03754163e-01 -7.24309534e-02 3.33832890e-01 7.32329667e-01
1.94528922e-01 -1.90337062e-01 5.06757677e-01 8.50916266e-01
6.26757592e-02 -1.80336565e-01 -6.65843964e-01 -2.65558869e-01
6.03963315e-01 1.08562243e+00 -3.73677045e-01 1.16029404e-01
-1.32729828e-01 8.16737354e-01 6.30675614e-01 2.24133551e-01
-7.97070086e-01 -1.83267996e-01 8.56710672e-01 -2.21501932e-01
5.45878053e-01 -4.66893196e-01 7.59799359e-03 -8.14775705e-01
-1.48328677e-01 -9.59245324e-01 5.13570726e-01 -5.27587295e-01
-1.22734582e+00 2.08170816e-01 8.69532600e-02 -1.38158882e+00
-7.64903724e-01 -6.68065667e-01 -1.77812770e-01 3.83883059e-01
-1.73770046e+00 -7.69109070e-01 2.86619037e-01 8.77541304e-01
2.37656683e-01 8.71248692e-02 9.62487698e-01 1.87196527e-02
-2.20384672e-01 4.83842015e-01 8.11959729e-02 -1.97655126e-01
7.95689225e-01 -1.51621079e+00 -1.07956804e-01 2.88663477e-01
-3.82507890e-02 5.28376639e-01 7.64395535e-01 -2.68807322e-01
-1.80607879e+00 -7.90747166e-01 5.33386767e-01 -4.83934850e-01
8.24811816e-01 -1.61262512e-01 -1.42080382e-01 5.89433432e-01
2.66902428e-02 1.49940655e-01 4.45556045e-01 2.62736470e-01
-4.80586499e-01 -2.21688673e-01 -1.08757186e+00 8.09090078e-01
1.01803100e+00 -5.42801261e-01 -5.89491129e-01 -6.80316165e-02
6.67333007e-01 -2.74006456e-01 -8.02988052e-01 2.43488684e-01
7.79502094e-01 -7.27815270e-01 9.52007353e-01 -6.91465914e-01
2.12214768e-01 -3.13561320e-01 -2.88122505e-01 -1.53594053e+00
-3.52569044e-01 -9.96845543e-01 -2.46132001e-01 4.36020494e-01
4.19408172e-01 -6.45253301e-01 8.48117113e-01 8.82233739e-01
-1.89117998e-01 -9.57845747e-01 -8.20167482e-01 -9.58651483e-01
1.44243136e-03 -3.17549765e-01 5.43055475e-01 6.07058108e-01
4.22244608e-01 3.70872408e-01 -6.56737983e-01 -3.34666193e-01
5.15978694e-01 4.82041270e-01 5.92819929e-01 -1.15142751e+00
-5.06941617e-01 -6.49870992e-01 1.40116811e-01 -1.45052576e+00
-5.70695801e-03 -8.10461342e-01 5.98886073e-01 -1.37652409e+00
-4.41979915e-02 -8.90479207e-01 -5.90777695e-01 8.70023668e-01
2.42032528e-01 -1.45157203e-01 1.86187908e-01 2.36003399e-01
-1.03746200e+00 8.41610670e-01 1.60976613e+00 2.26876363e-02
-5.12674332e-01 -1.38253206e-02 -7.67318010e-01 5.88158011e-01
1.05390394e+00 -5.45455337e-01 -7.54017353e-01 -3.23317170e-01
3.69152695e-01 2.97721565e-01 4.21585381e-01 -7.72508502e-01
3.62911105e-01 -6.97354078e-01 3.62299643e-02 -2.98784763e-01
3.95744681e-01 -9.04475629e-01 -2.97760785e-01 5.20928144e-01
-6.62580669e-01 3.13486308e-01 -5.00469133e-02 7.70773470e-01
1.19901016e-01 -8.51635337e-02 5.77337205e-01 -3.44623744e-01
-8.22852969e-01 4.77445215e-01 -4.65083629e-01 1.09040022e-01
8.72084141e-01 -1.02020195e-02 -2.06286937e-01 -5.79119384e-01
-5.04852653e-01 5.68578422e-01 4.53032225e-01 4.03611541e-01
7.05937386e-01 -1.31431031e+00 -1.67713448e-01 -2.63305195e-02
1.61744356e-01 -9.64844897e-02 -3.97264063e-01 7.99132109e-01
-1.09727019e-02 6.27892196e-01 -5.69810688e-01 -2.62949288e-01
-7.17932820e-01 5.83491981e-01 3.39949459e-01 -5.79088628e-01
-3.25022221e-01 5.91848791e-01 -3.44734490e-01 -9.29127753e-01
4.88349736e-01 -5.46210051e-01 -7.56520703e-02 -3.07826698e-01
4.01096344e-02 3.92196983e-01 -4.24487621e-01 -2.75199059e-02
-2.56641865e-01 1.73624933e-01 4.09840904e-02 -5.20922840e-01
1.44481659e+00 1.11915536e-01 2.70307839e-01 3.19514275e-01
6.95090413e-01 -2.42281869e-01 -1.81334019e+00 -2.89839417e-01
2.40354732e-01 -4.63752419e-01 9.82146710e-02 -9.00155723e-01
-8.99969578e-01 4.95368421e-01 4.22670752e-01 -1.67121783e-01
8.90496790e-01 -2.43746832e-01 6.51939750e-01 1.04202235e+00
8.64545107e-01 -1.45974207e+00 4.81788069e-01 8.01762760e-01
7.62507796e-01 -1.19980407e+00 -5.97149841e-02 2.48632133e-01
-8.79604399e-01 9.48800027e-01 6.48072362e-01 -3.50150615e-01
4.59870607e-01 -5.67096099e-02 -1.54775381e-01 -1.29342685e-02
-9.74434733e-01 -4.32607740e-01 4.30061445e-02 6.01157963e-01
2.85755873e-01 1.93569064e-01 -1.69131070e-01 5.94739139e-01
-3.13467622e-01 2.43207723e-01 3.19245785e-01 1.34215462e+00
-5.51836491e-01 -1.63004565e+00 -3.57337818e-02 5.22019565e-01
-4.43666503e-02 -9.52319354e-02 -2.72632718e-01 5.72726965e-01
-4.93502945e-01 9.67050374e-01 -1.60393462e-01 -5.32855630e-01
2.86160737e-01 -1.42741591e-01 7.48230815e-01 -5.69248796e-01
-5.18013239e-01 -1.97068036e-01 9.06992331e-02 -1.31392634e+00
-4.49568957e-01 -7.20708966e-01 -1.37898421e+00 -2.84651577e-01
5.67848459e-02 2.28361085e-01 4.00784761e-01 7.67648518e-01
4.31409061e-01 1.67797074e-01 6.94946766e-01 -1.08145881e+00
-1.28821075e+00 -5.08410394e-01 -4.45532471e-01 1.05336435e-01
4.30527538e-01 -1.03810918e+00 -1.01936026e-03 -4.80874836e-01] | [4.044167518615723, 1.7925009727478027] |
2b0119c7-9937-4566-95db-dcdd14ee1110 | integrated-multi-omics-analysis-using | 1908.06278 | null | https://arxiv.org/abs/1908.06278v1 | https://arxiv.org/pdf/1908.06278v1.pdf | Integrated Multi-omics Analysis Using Variational Autoencoders: Application to Pan-cancer Classification | Different aspects of a clinical sample can be revealed by multiple types of omics data. Integrated analysis of multi-omics data provides a comprehensive view of patients, which has the potential to facilitate more accurate clinical decision making. However, omics data are normally high dimensional with large number of molecular features and relatively small number of available samples with clinical labels. The "dimensionality curse" makes it challenging to train a machine learning model using high dimensional omics data like DNA methylation and gene expression profiles. Here we propose an end-to-end deep learning model called OmiVAE to extract low dimensional features and classify samples from multi-omics data. OmiVAE combines the basic structure of variational autoencoders with a classification network to achieve task-oriented feature extraction and multi-class classification. The training procedure of OmiVAE is comprised of an unsupervised phase without the classifier and a supervised phase with the classifier. During the unsupervised phase, a hierarchical cluster structure of samples can be automatically formed without the need for labels. And in the supervised phase, OmiVAE achieved an average classification accuracy of 97.49% after 10-fold cross-validation among 33 tumour types and normal samples, which shows better performance than other existing methods. The OmiVAE model learned from multi-omics data outperformed that using only one type of omics data, which indicates that the complementary information from different omics datatypes provides useful insights for biomedical tasks like cancer classification. | ['Xiao-Yu Zhang', 'Yike Guo', 'Xian Yang', 'Jingqing Zhang', 'Chengliang Dai', 'Kai Sun'] | 2019-08-17 | null | null | null | null | ['tumour-classification'] | ['medical'] | [-6.14642957e-03 -1.75928846e-01 -3.56042653e-01 -3.38905305e-01
-5.52809000e-01 -2.12270632e-01 3.48912895e-01 4.78981555e-01
-2.66437024e-01 4.88451213e-01 1.10307336e-01 -1.09198891e-01
-2.97878712e-01 -8.84182394e-01 -2.52353370e-01 -1.30408716e+00
1.72415346e-01 7.25325465e-01 -4.62404579e-01 4.09437388e-01
-1.46329075e-01 4.74489897e-01 -1.42057753e+00 8.04165602e-01
8.24526608e-01 9.96835291e-01 1.29921719e-01 6.22773588e-01
-5.31132042e-01 1.33781567e-01 -3.17080081e-01 2.22096499e-02
-1.16593517e-01 -5.14731705e-01 -3.48939985e-01 9.84277576e-02
-1.68993562e-01 1.17700957e-01 -1.48878872e-01 9.38380063e-01
7.85896301e-01 -1.68316931e-01 9.44453895e-01 -1.08042526e+00
-5.98028600e-01 2.97274254e-02 -1.91020250e-01 -8.22475627e-02
-1.16194151e-01 2.48120740e-01 9.00892377e-01 -7.64937401e-01
5.83972812e-01 1.10023010e+00 2.51787394e-01 7.17324853e-01
-1.09547985e+00 -5.20705760e-01 -1.23016991e-01 1.59104869e-01
-1.12848413e+00 -2.66835153e-01 4.18622643e-01 -7.96831429e-01
8.48578453e-01 4.85561669e-01 1.00379574e+00 1.08801115e+00
6.51782870e-01 8.70490611e-01 8.86580527e-01 -1.14191376e-01
4.56531852e-01 1.32696792e-01 6.01067483e-01 6.63082123e-01
2.57034957e-01 1.29303157e-01 -2.51922488e-01 -4.35851693e-01
1.36168584e-01 8.21523070e-01 -1.21670805e-01 -1.57089576e-01
-1.14953327e+00 8.70908558e-01 4.22279060e-01 4.96126235e-01
-4.05281872e-01 -1.58368096e-01 4.33116138e-01 1.86739177e-01
5.99850833e-01 4.63866055e-01 -4.99164224e-01 3.12838018e-01
-6.70043945e-01 -2.22029760e-01 4.04337794e-01 3.96640331e-01
5.71330011e-01 -2.11271286e-01 -4.56591286e-02 9.70392168e-01
5.95096886e-01 4.96325165e-01 1.15508676e+00 -5.24968147e-01
-1.72396496e-01 1.00650907e+00 -2.76645333e-01 -6.71336651e-01
-8.08157146e-01 -6.22470140e-01 -1.11023819e+00 -1.72616079e-01
1.71239972e-01 -1.77618787e-01 -1.04179919e+00 1.34235370e+00
5.81520855e-01 1.72619343e-01 2.69616872e-01 8.21664393e-01
1.06153774e+00 7.65224159e-01 1.00170068e-01 -2.87320554e-01
1.61794269e+00 -5.41097403e-01 -1.02767432e+00 2.09197298e-01
9.12207961e-01 -2.37818062e-01 7.54953265e-01 4.60924268e-01
-3.85835856e-01 -1.99568927e-01 -1.05717409e+00 -2.52932310e-02
-7.42033780e-01 1.43048406e-01 7.86935151e-01 5.06023228e-01
-4.82884496e-01 5.82678199e-01 -1.05725765e+00 -1.30888298e-01
7.32347608e-01 4.90604103e-01 -5.63317299e-01 -2.23208100e-01
-9.35547888e-01 3.74532193e-01 5.03586113e-01 1.95370421e-01
-8.86101186e-01 -8.06686401e-01 -9.14438426e-01 1.33341178e-01
-2.19480917e-01 -8.94248426e-01 5.73236406e-01 -7.90005863e-01
-1.60843241e+00 8.11739564e-01 -4.12838727e-01 1.67171836e-01
9.32684094e-02 -1.87681958e-01 -4.72935557e-01 7.37108514e-02
-6.71745688e-02 3.33347529e-01 5.93861818e-01 -7.76780069e-01
-2.95865148e-01 -9.26301301e-01 -7.62939274e-01 -6.08294681e-02
-5.68813801e-01 -3.31897736e-01 7.14623183e-03 1.12071903e-02
3.48288208e-01 -8.65871310e-01 -8.90966654e-02 -1.63847134e-02
-5.22548735e-01 -3.13276142e-01 7.33903706e-01 -3.45490932e-01
7.58789659e-01 -2.27708387e+00 3.60314548e-01 -1.62193224e-01
5.22946000e-01 2.57871062e-01 -1.59629703e-01 3.63869607e-01
-3.21194708e-01 2.49079019e-01 7.88573083e-03 -1.74828842e-01
-6.94609582e-02 2.75504943e-02 1.21253513e-01 8.23679805e-01
3.67805660e-01 1.00327146e+00 -9.52496231e-01 -3.73725772e-01
2.62622446e-01 4.32854593e-01 -4.66317743e-01 4.23294961e-01
-2.66839445e-01 4.71203923e-01 -5.97356856e-01 1.02715290e+00
5.94248474e-01 -7.45227158e-01 2.72347182e-01 -8.78852308e-02
2.79230267e-01 -1.10278293e-01 -6.36438847e-01 1.51371384e+00
-1.44945249e-01 5.45466304e-01 -2.62530476e-01 -1.12849510e+00
6.93357408e-01 5.44794559e-01 7.67378390e-01 -4.94092733e-01
4.47317958e-01 2.04548284e-01 3.05768877e-01 -9.82079864e-01
-3.79903257e-01 -6.77747726e-01 -4.90444014e-04 1.58847019e-01
1.09613411e-01 4.97632861e-01 -1.24178350e-01 -2.08522871e-01
9.18298423e-01 -3.38072270e-01 2.97788322e-01 -2.14571163e-01
5.10150850e-01 1.13127325e-02 1.05618191e+00 2.44820759e-01
-2.51805156e-01 3.87514889e-01 5.25432825e-01 -5.20991385e-01
-8.01201761e-01 -8.87966752e-01 -6.73246384e-01 7.29456067e-01
-1.05427809e-01 -1.22501984e-01 -2.92973876e-01 -5.77402174e-01
3.01921427e-01 3.02021801e-01 -9.58734870e-01 -2.85859436e-01
7.21352771e-02 -1.71682382e+00 3.51183921e-01 3.15489650e-01
2.55679846e-01 -7.31600046e-01 -2.02258885e-01 1.27725318e-01
6.95871711e-02 -8.14724863e-01 -6.43791631e-03 3.84913027e-01
-1.13917232e+00 -1.44527805e+00 -6.67704701e-01 -4.81380671e-01
7.40481496e-01 -1.09151294e-02 4.25111771e-01 -7.85524920e-02
-5.18438101e-01 -2.52140313e-01 -2.49204397e-01 -8.03622723e-01
-3.64822626e-01 -1.81972571e-02 2.62777030e-01 3.66144836e-01
8.88592124e-01 -3.32323849e-01 -7.09758997e-01 -3.05078924e-02
-9.04459894e-01 9.98329073e-02 7.96008408e-01 1.32078755e+00
7.58022308e-01 -1.58577994e-01 8.42839956e-01 -9.05740798e-01
2.03705296e-01 -8.81765783e-01 -3.21689278e-01 1.72103830e-02
-4.65048075e-01 6.61152229e-02 7.57928193e-01 -2.68922240e-01
-6.84862196e-01 -5.89901619e-02 -3.42103899e-01 -3.09619278e-01
-3.37356061e-01 7.31453717e-01 -5.46126485e-01 4.65126991e-01
3.87390614e-01 3.12981725e-01 2.89371967e-01 -5.14094114e-01
-2.08641008e-01 1.21550643e+00 -9.58401710e-02 -2.93375272e-02
-2.10551471e-02 5.21743119e-01 2.38855734e-01 -9.94998217e-01
-8.37187350e-01 -6.65897667e-01 -7.69919932e-01 1.40691236e-01
1.17752564e+00 -1.02649677e+00 -1.02490628e+00 5.54024279e-01
-7.08492577e-01 -9.66473743e-02 6.73329309e-02 8.69049191e-01
-3.00463200e-01 1.95953652e-01 -9.35570240e-01 -6.21232390e-01
-5.40991187e-01 -1.14437854e+00 1.20637023e+00 9.56469551e-02
-1.58249930e-01 -1.00050890e+00 2.59718955e-01 4.98221278e-01
-2.03768268e-01 5.24347365e-01 1.45746601e+00 -1.00189126e+00
-3.74816954e-01 -5.35606265e-01 -6.44687489e-02 1.87898904e-01
4.80554610e-01 -1.33210257e-01 -1.32089353e+00 -2.79234946e-01
7.23175779e-02 -1.72048748e-01 9.59591627e-01 5.18934965e-01
1.29008245e+00 1.20837400e-02 -8.13543022e-01 9.17743325e-01
1.53370166e+00 4.87339705e-01 3.87095779e-01 4.37885299e-02
6.82621241e-01 5.29906869e-01 4.07399118e-01 2.67328441e-01
-9.12760198e-02 2.85485983e-01 3.97565633e-01 1.47473188e-02
3.88437927e-01 1.55887067e-01 2.16756344e-01 1.07591343e+00
5.03959179e-01 -3.81055236e-01 -8.82241666e-01 4.75082755e-01
-1.76098824e+00 -7.10274518e-01 -3.08615297e-01 1.85269880e+00
5.91881037e-01 -4.96825129e-01 2.71364432e-02 5.46554625e-02
5.48936725e-01 -2.04393059e-01 -1.02443182e+00 -2.73888201e-01
-1.79777145e-01 -2.01694354e-01 -3.76634300e-02 1.92951500e-01
-1.00813043e+00 4.42734838e-01 6.36389303e+00 4.17947382e-01
-1.32321405e+00 7.11393356e-02 7.19329774e-01 -4.66138005e-01
-1.89028174e-01 -4.80535984e-01 -8.33140492e-01 7.13221252e-01
1.27721846e+00 6.70300201e-02 -2.24014387e-01 5.45511901e-01
4.46415126e-01 1.80934500e-02 -1.25985646e+00 1.03390777e+00
-1.54040471e-01 -1.58931363e+00 6.41303584e-02 6.46567881e-01
5.17590046e-01 3.59576762e-01 -7.75870010e-02 1.04559816e-01
2.90877342e-01 -1.18562078e+00 -1.78757668e-01 7.02505231e-01
7.58767486e-01 -6.74313962e-01 1.03534091e+00 7.70870924e-01
-6.25211477e-01 -2.88527101e-01 -4.75414366e-01 1.35592520e-01
-2.27577344e-01 1.19321501e+00 -8.22344244e-01 7.15416372e-01
4.58327770e-01 1.08489311e+00 -2.35267982e-01 9.30738211e-01
2.44903952e-01 3.97781760e-01 -1.63880914e-01 -3.49452138e-01
1.27994761e-01 -3.08622599e-01 1.79374501e-01 1.13723350e+00
3.88242871e-01 1.89886004e-01 8.95259380e-02 7.18627453e-01
4.93521094e-02 3.34573150e-01 -3.94856811e-01 -7.13623643e-01
1.74297631e-01 1.44109046e+00 -5.76607287e-01 -4.95962173e-01
-2.09848672e-01 7.70486176e-01 2.22745940e-01 1.77907810e-01
-5.36345720e-01 -4.23854619e-01 9.71072853e-01 -1.22163326e-01
2.37756763e-02 9.51673929e-03 -3.05278301e-01 -1.56948960e+00
-4.20335710e-01 -7.78023303e-01 6.31100297e-01 -4.51351374e-01
-1.50684941e+00 3.75417829e-01 -4.89046216e-01 -1.14087403e+00
-1.34714484e-01 -1.04822314e+00 -5.39201438e-01 8.88467431e-01
-1.34823084e+00 -8.49550784e-01 -4.69717920e-01 1.75651729e-01
1.41895458e-01 -4.18680072e-01 1.23975611e+00 2.38835096e-01
-1.13302255e+00 3.34022045e-01 8.97792995e-01 1.94831103e-01
5.89106262e-01 -1.21795356e+00 -3.64399642e-01 1.91876203e-01
-2.32504562e-01 6.05563283e-01 3.79126817e-01 -7.55027115e-01
-1.63213181e+00 -1.24426162e+00 6.51649058e-01 -2.03241065e-01
3.17241818e-01 -4.63208020e-01 -9.76317346e-01 5.54775834e-01
-9.75183025e-02 1.74066111e-01 1.91702855e+00 2.90539831e-01
-6.96832240e-02 -6.35778764e-03 -1.08951545e+00 3.15151423e-01
5.35017729e-01 -4.19806927e-01 -4.30874586e-01 6.09935164e-01
4.61980015e-01 -5.92584647e-02 -1.31589043e+00 2.52406120e-01
6.38324320e-01 -5.11098564e-01 7.03509390e-01 -1.09666288e+00
5.13014019e-01 -3.05510014e-01 -3.58592033e-01 -1.42778182e+00
-5.13038695e-01 1.33059770e-01 -1.50883049e-01 5.92097163e-01
2.07973719e-01 -9.27691936e-01 9.49285686e-01 2.78102249e-01
-1.49731964e-01 -9.72466290e-01 -7.97853470e-01 -5.31731844e-01
3.06239516e-01 -1.46505058e-01 6.52300298e-01 1.15591002e+00
3.84943187e-01 4.04029936e-01 -1.04433209e-01 1.02180958e-01
5.27047813e-01 5.23031235e-01 5.56446970e-01 -1.64269137e+00
-3.05703282e-01 -2.74406761e-01 -6.19628191e-01 -4.28191990e-01
2.65931219e-01 -1.44907379e+00 -2.82956749e-01 -1.57860565e+00
5.54836571e-01 -2.40265980e-01 -6.01362348e-01 5.64925373e-01
-3.80974025e-01 -2.63784956e-02 -1.88078552e-01 2.40712285e-01
-2.37474963e-01 8.29424620e-01 1.20185006e+00 -4.50637966e-01
-3.44758540e-01 -2.40931422e-01 -8.38506043e-01 4.13639486e-01
6.54847026e-01 -6.08171284e-01 -1.63147449e-01 -1.13946363e-01
4.56294417e-02 8.17255825e-02 2.18570486e-01 -6.92374170e-01
1.20367870e-01 -3.35937291e-01 1.18631256e+00 -6.79585278e-01
4.58993673e-01 -5.21802306e-01 4.58777070e-01 8.75469744e-01
-1.91624656e-01 -5.32236695e-01 2.30262235e-01 7.10152090e-01
-1.74040601e-01 -2.03699440e-01 7.06750154e-01 -1.20244220e-01
-3.41726780e-01 5.26168048e-01 -6.95337534e-01 -4.92427409e-01
1.06950700e+00 -1.67374134e-01 -2.77377695e-01 2.22369120e-01
-1.14544654e+00 3.36118311e-01 1.78212106e-01 2.36319512e-01
7.16492474e-01 -1.45912409e+00 -6.40601218e-01 5.01734197e-01
3.59585047e-01 8.75462145e-02 6.10037982e-01 9.13044453e-01
-4.20183033e-01 5.97428739e-01 -2.13135436e-01 -1.09493160e+00
-1.26115274e+00 6.68667674e-01 5.65353751e-01 -1.53992772e-01
-6.73312306e-01 3.59633833e-01 4.28276777e-01 -6.09950244e-01
-3.50204080e-01 1.72645953e-02 -2.57777393e-01 5.29879034e-01
7.21967816e-01 1.46200106e-01 1.30757734e-01 -3.32446486e-01
-4.17547464e-01 4.46327031e-01 -2.34117612e-01 2.44402036e-01
1.68715870e+00 1.62750587e-01 -2.46978343e-01 8.42932045e-01
1.74126804e+00 -4.81060892e-01 -7.61119187e-01 2.88743675e-02
-4.62900437e-02 -2.95832962e-01 3.11005414e-01 -7.25990713e-01
-9.64283645e-01 1.31816447e+00 7.86592364e-01 -1.31395891e-01
8.86646330e-01 -1.16446063e-01 8.02526355e-01 5.10475636e-01
-1.44387558e-01 -1.00661957e+00 2.48470698e-02 2.15158418e-01
5.02765536e-01 -1.31907547e+00 -1.45984918e-01 -1.79010808e-01
-2.82830715e-01 1.23344731e+00 3.84703904e-01 8.45888853e-02
9.73975360e-01 7.85579160e-02 2.54245162e-01 -5.49071968e-01
-9.95224178e-01 4.14613597e-02 3.51027191e-01 4.52656776e-01
5.64856887e-01 2.38610923e-01 -1.80128291e-02 1.02580345e+00
9.12386477e-02 2.25802898e-01 2.31906831e-01 6.37282968e-01
-5.25537193e-01 -9.95971680e-01 -2.65377879e-01 7.82746732e-01
-5.56443751e-01 1.90518647e-01 -3.38544697e-01 5.82547486e-01
1.20538019e-01 8.70642483e-01 2.88946480e-01 -4.27203685e-01
-1.69407532e-01 5.99495292e-01 1.42812207e-01 -7.79173672e-01
-4.03074890e-01 4.30906147e-01 -2.03040376e-01 -4.52650160e-01
-1.74125135e-01 -7.15736508e-01 -1.34608352e+00 -5.59589081e-02
-4.53169495e-01 1.79413259e-01 7.83057272e-01 9.79161024e-01
8.36927116e-01 9.38525617e-01 6.64544582e-01 -4.93901193e-01
-2.10448578e-01 -9.34134245e-01 -8.16743076e-01 2.83732682e-01
5.12494504e-01 -5.19852579e-01 -2.50942320e-01 -2.02008262e-01] | [6.030148983001709, 5.692865371704102] |
9198c37d-5c38-4661-89c7-a1c57d4e65bd | deep-high-resolution-representation-learning | 1902.09212 | null | http://arxiv.org/abs/1902.09212v1 | http://arxiv.org/pdf/1902.09212v1.pdf | Deep High-Resolution Representation Learning for Human Pose Estimation | This is an official pytorch implementation of Deep High-Resolution
Representation Learning for Human Pose Estimation. In this work, we are
interested in the human pose estimation problem with a focus on learning
reliable high-resolution representations. Most existing methods recover
high-resolution representations from low-resolution representations produced by
a high-to-low resolution network. Instead, our proposed network maintains
high-resolution representations through the whole process. We start from a
high-resolution subnetwork as the first stage, gradually add high-to-low
resolution subnetworks one by one to form more stages, and connect the
mutli-resolution subnetworks in parallel. We conduct repeated multi-scale
fusions such that each of the high-to-low resolution representations receives
information from other parallel representations over and over, leading to rich
high-resolution representations. As a result, the predicted keypoint heatmap is
potentially more accurate and spatially more precise. We empirically
demonstrate the effectiveness of our network through the superior pose
estimation results over two benchmark datasets: the COCO keypoint detection
dataset and the MPII Human Pose dataset. The code and models have been publicly
available at
\url{https://github.com/leoxiaobin/deep-high-resolution-net.pytorch}. | ['Ke Sun', 'Bin Xiao', 'Jingdong Wang', 'Dong Liu'] | 2019-02-25 | deep-high-resolution-representation-learning-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Sun_Deep_High-Resolution_Representation_Learning_for_Human_Pose_Estimation_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Sun_Deep_High-Resolution_Representation_Learning_for_Human_Pose_Estimation_CVPR_2019_paper.pdf | cvpr-2019-6 | ['2d-human-pose-estimation'] | ['computer-vision'] | [-1.12776786e-01 7.03886449e-02 -3.67276222e-02 -2.91988015e-01
-1.21488619e+00 -1.72926068e-01 3.75365436e-01 6.91737533e-02
-4.28536952e-01 7.89619386e-01 4.56824362e-01 5.91497898e-01
-1.23255759e-01 -8.16792130e-01 -9.02385235e-01 -4.64554012e-01
-2.39179403e-01 7.58636355e-01 4.44818616e-01 -4.27233249e-01
-2.49908328e-01 4.82324094e-01 -1.47782087e+00 3.60789716e-01
3.99658293e-01 7.50324249e-01 4.72051688e-02 7.17094719e-01
5.79656065e-01 3.29431623e-01 -4.87714320e-01 -8.69629756e-02
2.13289678e-01 -7.98660815e-02 -8.43290091e-01 -3.88156891e-01
5.56524813e-01 -2.65555590e-01 -5.35756707e-01 8.62737417e-01
7.41437078e-01 3.00763696e-01 4.02229577e-01 -9.20153618e-01
-3.17741074e-02 1.77741244e-01 -1.14098740e+00 2.63744146e-01
5.86696327e-01 -4.34383601e-02 8.40822101e-01 -9.65488732e-01
8.54674935e-01 1.57352829e+00 8.59676778e-01 3.85528624e-01
-1.33421266e+00 -8.06127131e-01 2.25531310e-01 -1.41072303e-01
-1.62906945e+00 -2.21431911e-01 7.44093776e-01 -2.50554621e-01
6.81501329e-01 1.34554595e-01 7.73399472e-01 1.27744234e+00
3.71490002e-01 7.33864069e-01 9.15160537e-01 3.99552882e-02
-2.58879364e-01 -4.47175354e-01 8.91554505e-02 8.27204108e-01
4.92645442e-01 1.20268978e-01 -7.09483385e-01 -1.94251746e-01
1.27292168e+00 1.53678432e-01 -2.73577482e-01 -4.61607426e-01
-1.45227242e+00 6.43265009e-01 1.18942392e+00 1.92112282e-01
-7.46762455e-01 4.77779746e-01 1.82743132e-01 -2.28682205e-01
4.90351975e-01 4.90312010e-01 -4.41201866e-01 -3.49939279e-02
-8.90778959e-01 6.52859092e-01 2.77099758e-01 6.26375318e-01
7.30692148e-01 -3.73628765e-01 -1.88498706e-01 8.85749400e-01
2.65517533e-01 2.21131012e-01 4.24043894e-01 -9.06119347e-01
5.56194901e-01 4.13542241e-01 2.14544728e-01 -1.24950564e+00
-9.40905273e-01 -7.09904730e-01 -1.07213843e+00 2.36374125e-01
3.98944139e-01 -1.81934237e-01 -1.08278537e+00 1.70139527e+00
5.26397824e-01 1.53841034e-01 -2.20791265e-01 1.42055380e+00
9.57720459e-01 7.43958235e-01 1.39663696e-01 3.67347032e-01
1.54479098e+00 -9.79662299e-01 -4.53858882e-01 -3.80630910e-01
3.24235588e-01 -5.04901946e-01 7.29528487e-01 1.85400113e-01
-1.02047837e+00 -8.25008333e-01 -1.12006700e+00 -2.70287603e-01
-2.31146589e-01 3.21343899e-01 5.59657633e-01 -1.48477659e-01
-8.59943986e-01 6.56412840e-01 -8.80035520e-01 -3.01703483e-01
4.39577311e-01 3.28423649e-01 -7.21921444e-01 -1.20505348e-01
-1.38351583e+00 7.01841593e-01 3.68012756e-01 4.09434080e-01
-5.98174870e-01 -5.35966337e-01 -9.77931082e-01 -1.73654273e-01
4.91630226e-01 -9.20353234e-01 1.08785892e+00 -3.99823040e-01
-1.03568411e+00 6.89236939e-01 -7.00322315e-02 -3.69622529e-01
7.54705131e-01 -7.52792716e-01 -3.93088162e-03 2.72766232e-01
2.69307077e-01 1.11006999e+00 5.59001088e-01 -1.28776872e+00
-5.35728693e-01 -6.35560274e-01 -2.61487588e-02 4.22794789e-01
2.42141098e-01 -2.77085543e-01 -8.26617956e-01 -8.41310501e-01
4.67999905e-01 -1.07120585e+00 -4.49888974e-01 1.29497454e-01
-5.04356086e-01 -7.00136796e-02 4.65337187e-01 -8.00216377e-01
1.05688953e+00 -1.79467642e+00 4.85898852e-01 3.29503506e-01
6.05708420e-01 6.31634146e-02 -1.39833540e-01 -2.68430319e-02
-2.16145322e-01 -2.22460777e-01 -3.16069461e-02 -4.77502018e-01
-3.60279918e-01 -1.28841832e-01 7.59148076e-02 7.21968949e-01
8.77302513e-02 1.20449781e+00 -9.19019878e-01 -5.46154141e-01
3.69426727e-01 1.03882146e+00 -4.32754576e-01 7.51324594e-02
1.48619905e-01 5.82642138e-01 -4.98992980e-01 6.08113408e-01
5.13728261e-01 -5.12474477e-01 -1.18301220e-01 -5.45963109e-01
8.94809887e-02 4.32241522e-02 -1.27660608e+00 2.02565360e+00
-2.43465781e-01 4.55331922e-01 -2.00111382e-02 -5.04030704e-01
8.90978873e-01 7.98599795e-02 6.64734960e-01 -6.14690125e-01
1.60239816e-01 -1.01438366e-01 -3.11006516e-01 1.43716544e-01
7.91050613e-01 1.93022504e-01 -4.00490552e-01 8.72812346e-02
-4.09047268e-02 2.62816131e-01 -1.28395021e-01 1.93393335e-01
6.87153041e-01 4.99975383e-01 4.46022153e-01 7.85602331e-02
1.81874067e-01 1.66328289e-02 5.88824332e-01 3.78675967e-01
-1.90462917e-01 1.01842844e+00 4.95251447e-01 -5.15611172e-01
-1.02547586e+00 -1.10687327e+00 -8.74750540e-02 1.20112240e+00
2.64070630e-01 -6.68209791e-01 -6.99632227e-01 -3.04749906e-01
1.49722368e-01 -8.61529857e-02 -9.27842200e-01 -5.40284552e-02
-8.43305945e-01 -5.36050141e-01 6.00169182e-01 8.06783795e-01
5.77032745e-01 -9.35918331e-01 -8.27660561e-01 3.94541994e-02
-4.54317331e-01 -8.90185058e-01 -3.74028474e-01 2.51154928e-03
-7.26741850e-01 -1.10070622e+00 -1.25928462e+00 -6.11553967e-01
6.49072826e-01 9.37188417e-02 9.02578354e-01 -9.60989371e-02
-4.46966171e-01 9.92214829e-02 -1.73352271e-01 -2.78247744e-02
2.98297018e-01 3.26560587e-01 2.36054808e-01 -3.65407258e-01
1.60357542e-02 -3.79515767e-01 -7.59980440e-01 3.14456999e-01
-3.12412679e-01 2.38686755e-01 7.21665442e-01 7.42389441e-01
1.09756327e+00 -1.80869699e-01 3.46394479e-01 -6.73547089e-01
5.13664901e-01 -3.20176035e-01 -5.40768266e-01 -7.66127706e-02
-2.70948149e-02 1.09873764e-01 2.21012756e-01 -3.60091269e-01
-8.80475104e-01 5.29078066e-01 -2.66987324e-01 -6.86331451e-01
-2.46886536e-01 2.83225268e-01 -7.90653080e-02 9.99076590e-02
8.71267200e-01 -1.07217766e-01 -9.32992622e-03 -6.10973001e-01
3.39360148e-01 1.37092501e-01 8.49416018e-01 -5.95369339e-01
9.20869112e-01 3.41875374e-01 1.11711651e-01 -6.63734913e-01
-9.02014792e-01 -4.83003199e-01 -1.01699603e+00 -2.78421402e-01
9.93203282e-01 -1.22955692e+00 -6.54053867e-01 4.81100708e-01
-9.55629230e-01 -1.83925822e-01 -2.40624949e-01 5.21950483e-01
-4.45120603e-01 5.60276881e-02 -6.72609210e-01 -5.80645442e-01
-4.22292113e-01 -1.08790922e+00 1.42239106e+00 3.86076242e-01
-5.47780395e-01 -6.11361742e-01 3.58001292e-01 2.72990048e-01
8.54311883e-02 8.25249910e-01 1.06370091e-01 -2.45061636e-01
-4.82541829e-01 -4.25018847e-01 -2.58744538e-01 -2.30999395e-01
-1.47992671e-01 -3.37618858e-01 -9.18620408e-01 -5.29156983e-01
-6.46457732e-01 -5.47377646e-01 1.06166863e+00 6.74139917e-01
1.20529115e+00 -2.22422093e-01 -5.94840705e-01 9.44601893e-01
1.00328481e+00 -5.49659193e-01 5.16309559e-01 4.03605461e-01
1.03370273e+00 5.94124556e-01 8.79119873e-01 3.39785516e-01
5.01365542e-01 1.03044879e+00 2.06373915e-01 -4.38496113e-01
-1.49179935e-01 -6.42967880e-01 -6.34424388e-02 2.12741762e-01
-4.60031420e-01 2.91945279e-01 -8.90921056e-01 3.03421050e-01
-2.04891276e+00 -8.36933017e-01 4.78089452e-02 2.03065777e+00
6.26966894e-01 1.79980889e-01 5.35439491e-01 -1.41200468e-01
8.21652830e-01 5.07485628e-01 -5.42618394e-01 3.71091038e-01
1.07359469e-01 7.49118626e-02 4.65967774e-01 3.39243323e-01
-1.37240720e+00 9.73525047e-01 5.70076656e+00 6.05683088e-01
-9.41880167e-01 -5.24527021e-02 6.35770500e-01 -5.08289278e-01
1.31690606e-01 -4.60147619e-01 -9.30207908e-01 2.42433622e-01
5.94704628e-01 2.65186913e-02 8.46989229e-02 8.77916396e-01
-4.75066267e-02 -2.00740904e-01 -9.59515214e-01 1.17894733e+00
-1.67539150e-01 -1.32836282e+00 -5.08103939e-03 5.60170822e-02
5.08046925e-01 9.02307406e-02 1.65889282e-02 2.15748847e-01
1.95523977e-01 -1.17993140e+00 6.00444555e-01 6.37123048e-01
9.99314964e-01 -1.17467034e+00 6.61480486e-01 2.86770195e-01
-1.91282177e+00 9.71511826e-02 -4.39723969e-01 1.43535152e-01
1.32370502e-01 4.29242820e-01 -4.64694530e-01 5.19548237e-01
1.05215406e+00 7.93311775e-01 -6.89085484e-01 1.03379035e+00
-3.00061315e-01 -6.17558397e-02 -3.60664815e-01 1.83441728e-01
3.85992341e-02 1.83034003e-01 4.78256404e-01 1.27441716e+00
6.61074966e-02 2.71612585e-01 4.01851118e-01 7.25055873e-01
-2.28047550e-01 -1.85603321e-01 -4.39820051e-01 4.29826766e-01
4.64027941e-01 1.45214987e+00 -7.13913918e-01 -2.68866807e-01
6.34971485e-02 9.89551425e-01 5.58504164e-01 2.60666072e-01
-8.78147781e-01 -3.83370489e-01 9.57590342e-01 4.18017119e-01
1.09782524e-01 -2.57359713e-01 -1.94299623e-01 -1.07918048e+00
4.40216735e-02 -6.29538476e-01 6.31553590e-01 -8.26938033e-01
-1.02211642e+00 7.10895419e-01 3.28711897e-01 -1.18025506e+00
-4.79501516e-01 -2.89827168e-01 -2.71310598e-01 1.06672943e+00
-1.19399118e+00 -1.27510476e+00 -5.32306790e-01 6.60842359e-01
5.01609206e-01 2.10016519e-01 6.68006599e-01 1.21502385e-01
-4.57114607e-01 7.52900481e-01 -3.71430010e-01 5.98448038e-01
7.90062845e-01 -1.01882768e+00 5.73452234e-01 6.56281531e-01
-5.29788435e-02 7.62400389e-01 5.96683443e-01 -8.66257489e-01
-9.99426067e-01 -1.17134809e+00 5.07310987e-01 -5.35884976e-01
2.44077682e-01 -6.02976441e-01 -9.64459062e-01 6.51354492e-01
-4.21330571e-01 2.49704927e-01 1.97198331e-01 3.85871232e-01
-3.25517029e-01 -1.77234318e-02 -1.14147246e+00 3.58270705e-01
1.06605721e+00 -3.89057785e-01 -6.75696671e-01 1.56859294e-01
6.08904541e-01 -9.16008353e-01 -1.02265728e+00 5.42515099e-01
8.96532714e-01 -6.99438512e-01 1.51783156e+00 -2.79983699e-01
4.79123116e-01 -4.14155900e-01 1.18756205e-01 -1.23624551e+00
-7.08020210e-01 -2.82683194e-01 -2.21498266e-01 6.20665073e-01
3.19054514e-01 -4.31516469e-01 9.03720677e-01 1.69653535e-01
1.97768778e-01 -1.04626501e+00 -9.55440223e-01 -4.15049732e-01
-2.27203071e-02 -8.43489990e-02 4.13264751e-01 6.78474367e-01
-2.46116936e-01 3.79574329e-01 -6.22173607e-01 3.43802422e-01
9.12378788e-01 1.33061826e-01 1.05372036e+00 -1.43546295e+00
-2.09303096e-01 -2.09577695e-01 -5.28797865e-01 -1.24005926e+00
-4.77731489e-02 -5.91996133e-01 2.80704618e-01 -1.69515789e+00
3.67493629e-01 -1.76194414e-01 -2.42191404e-01 6.14131272e-01
-2.41246358e-01 6.79180980e-01 2.37255350e-01 4.36443508e-01
-7.15294361e-01 4.03184146e-01 1.36847889e+00 1.28527135e-01
-3.31829101e-01 -3.90999168e-02 -6.92862988e-01 8.87461066e-01
7.81156123e-01 -4.03492928e-01 -1.33420050e-01 -2.67236419e-02
-4.54475805e-02 1.52075678e-01 7.28449881e-01 -1.40818000e+00
2.32462287e-01 1.65029258e-01 1.26208234e+00 -9.29005265e-01
9.11058843e-01 -4.84662920e-01 1.56727612e-01 6.43815517e-01
-4.22847748e-01 1.95786893e-01 1.94812730e-01 5.54474711e-01
-9.15236473e-02 5.80208123e-01 1.03501344e+00 -2.81994909e-01
-8.49250197e-01 6.39039278e-01 1.32113263e-01 1.81511103e-04
9.36764538e-01 -2.59417593e-01 -3.59194010e-01 -3.48575264e-01
-9.67828751e-01 3.12074989e-01 5.31969488e-01 7.42721140e-01
9.47933853e-01 -1.43199670e+00 -7.19261467e-01 1.78006917e-01
2.47020513e-01 4.50803608e-01 3.49449962e-01 7.69665182e-01
-4.19203401e-01 3.86058450e-01 -5.88085771e-01 -7.92142928e-01
-1.34112108e+00 2.27200419e-01 5.33599138e-01 -4.22440112e-01
-1.07327211e+00 8.59516859e-01 3.59046012e-01 -4.36859280e-01
2.46777534e-01 -2.17574954e-01 -3.64891559e-01 1.08097181e-01
5.88648677e-01 5.20169377e-01 -1.81088269e-01 -1.22237515e+00
-7.02757657e-01 9.40339208e-01 -1.33472905e-01 -4.36322801e-02
1.40417409e+00 8.63079950e-02 2.33282775e-01 2.69919068e-01
1.34528553e+00 -2.80483395e-01 -1.45784807e+00 -1.95814133e-01
-3.51530671e-01 -5.88130534e-01 -9.95064080e-02 -5.97766340e-01
-1.14182496e+00 7.13089406e-01 6.70368195e-01 -6.09721005e-01
9.85212982e-01 3.60656351e-01 7.15493202e-01 3.84714544e-01
4.75235105e-01 -1.13541496e+00 3.61020803e-01 4.94055480e-01
1.22519040e+00 -1.06318247e+00 4.00593430e-01 -2.86443710e-01
-6.30184650e-01 8.63492072e-01 1.04342544e+00 -4.57637787e-01
5.10829747e-01 2.67936498e-01 -1.28429219e-01 -4.45863843e-01
-5.24778605e-01 -2.86129564e-01 6.51689053e-01 6.16264224e-01
5.88105500e-01 1.59204945e-01 7.21986815e-02 6.11552835e-01
-3.94799322e-01 -1.73238125e-02 8.47774968e-02 8.18899632e-01
-6.01355374e-01 -6.46976173e-01 -7.12375045e-01 5.48964441e-01
-1.51875854e-01 2.80650347e-01 -4.41800982e-01 8.75593424e-01
1.42507227e-02 4.70072150e-01 1.65439829e-01 -6.05854928e-01
6.43058300e-01 -3.35128218e-01 3.44369262e-01 -4.91374999e-01
-4.15893674e-01 2.42270276e-01 -4.26197797e-02 -1.17271793e+00
-2.28424013e-01 -5.47886074e-01 -1.55687523e+00 -3.84847045e-01
9.98409092e-02 -2.81135906e-02 3.02989691e-01 4.80443120e-01
4.15055782e-01 8.23731005e-01 9.82428715e-02 -1.56211460e+00
-1.14148892e-01 -1.00922036e+00 -2.95834213e-01 3.03156048e-01
5.02963424e-01 -9.85083282e-01 6.42824173e-02 -4.46599245e-01] | [7.091564178466797, -0.8728415966033936] |
609291dc-bc04-49ae-9247-b261b5d2945e | cache-transition-systems-for-graph-parsing | null | null | https://aclanthology.org/J18-1004 | https://aclanthology.org/J18-1004.pdf | Cache Transition Systems for Graph Parsing | Motivated by the task of semantic parsing, we describe a transition system that generalizes standard transition-based dependency parsing techniques to generate a graph rather than a tree. Our system includes a cache with fixed size m, and we characterize the relationship between the parameter m and the class of graphs that can be produced through the graph-theoretic concept of tree decomposition. We find empirically that small cache sizes cover a high percentage of sentences in existing semantic corpora. | ['Xiaochang Peng', 'Giorgio Satta', 'Daniel Gildea'] | 2018-03-01 | null | null | null | cl-2018-3 | ['tree-decomposition', 'transition-based-dependency-parsing'] | ['graphs', 'natural-language-processing'] | [ 2.80005366e-01 8.11077356e-01 -3.81433189e-01 -6.71205580e-01
-9.39962327e-01 -7.75960267e-01 3.94636869e-01 5.93048632e-01
-9.60708410e-02 5.93107462e-01 3.41424882e-01 -8.71563315e-01
3.64000112e-01 -1.29978859e+00 -7.72231877e-01 3.43072116e-02
-2.23208025e-01 7.78711796e-01 9.69242871e-01 -5.78336865e-02
1.73006535e-01 2.53552705e-01 -1.21445155e+00 3.92725974e-01
6.68544412e-01 4.91680831e-01 1.70851871e-01 1.16833842e+00
-7.03951418e-01 7.02442229e-01 -7.43012667e-01 -5.85751951e-01
-1.03832036e-01 -6.70228004e-01 -1.51214397e+00 1.97106317e-01
2.10180074e-01 1.36878103e-01 -4.97145921e-01 1.12963104e+00
-1.44128114e-01 2.38509670e-01 3.35269630e-01 -1.26153469e+00
-4.52225059e-01 1.33032310e+00 -1.56438738e-01 6.22846246e-01
5.39342582e-01 -3.38995218e-01 1.22696412e+00 -8.97316635e-02
9.67420280e-01 1.62406719e+00 5.15448093e-01 7.27087259e-01
-1.22656369e+00 -2.85968512e-01 2.77394056e-01 -2.76275843e-01
-7.46586442e-01 -2.58477658e-01 5.04259050e-01 -1.97807357e-01
1.51221514e+00 7.13243242e-03 4.61352885e-01 8.13951612e-01
4.73667830e-01 4.53604162e-01 1.00132477e+00 -8.04744482e-01
4.66447443e-01 -6.26292229e-01 9.06106055e-01 1.01323259e+00
6.07045352e-01 -1.94904789e-01 -5.84326923e-01 -5.80697715e-01
7.70054102e-01 -8.03903222e-01 8.35719034e-02 -2.59198934e-01
-6.02665782e-01 1.23696303e+00 9.51040834e-02 1.96465626e-01
2.06856787e-01 4.65586603e-01 7.16840088e-01 3.55433494e-01
3.46374691e-01 1.47167593e-01 -6.33404613e-01 -1.92696929e-01
-5.48303962e-01 6.68227747e-02 1.32864249e+00 1.37179983e+00
5.43244064e-01 -5.29304408e-02 3.82549614e-01 3.41616571e-01
2.17561409e-01 1.86832115e-01 7.63665810e-02 -1.18692231e+00
7.46672750e-01 6.03981614e-01 -1.04467422e-01 -3.27936590e-01
-4.83924598e-01 1.46685898e-01 9.03429314e-02 -2.56686240e-01
7.06207514e-01 -2.08067492e-01 -1.03408444e+00 2.06513953e+00
4.33474660e-01 -2.10722268e-01 3.34898114e-01 3.74541640e-01
4.77663577e-01 4.81112719e-01 5.81899703e-01 -3.15583318e-01
1.53373516e+00 -8.60032201e-01 -4.63830560e-01 -5.67979753e-01
1.20934987e+00 -5.11718512e-01 7.42409766e-01 -8.73562172e-02
-1.26174474e+00 -1.94413453e-01 -1.12157106e+00 -1.72988251e-01
-2.01081067e-01 -7.58027673e-01 1.23353899e+00 1.01152778e+00
-1.50760889e+00 6.38699174e-01 -1.11795461e+00 -5.94070196e-01
-6.55756444e-02 3.28997642e-01 -3.80900830e-01 4.56767082e-02
-9.85278845e-01 8.34304154e-01 8.53641450e-01 -4.48542535e-01
-4.12674576e-01 -1.07761815e-01 -1.12281597e+00 1.22879751e-01
2.80775487e-01 -8.65638077e-01 1.66181302e+00 -8.48182738e-01
-1.15798080e+00 9.82505441e-01 -3.45640004e-01 -6.00167036e-01
-1.99626625e-01 -7.85994530e-02 -2.96786189e-01 4.97407645e-01
2.46176377e-01 5.89749455e-01 5.25117636e-01 -1.00715280e+00
-7.09107399e-01 -4.44145620e-01 4.69788432e-01 2.65586376e-01
9.00832489e-02 3.97767544e-01 -4.26674664e-01 -2.80288219e-01
4.19253290e-01 -9.37205851e-01 -5.06587923e-01 -5.10495067e-01
-4.21794057e-01 -3.95821631e-01 4.64782119e-01 -3.81725520e-01
1.07689774e+00 -1.98659921e+00 6.90832064e-02 1.07715435e-01
2.74240762e-01 -1.70433491e-01 -2.42762610e-01 5.18968821e-01
-4.00351733e-02 4.34094012e-01 -4.83985603e-01 1.22627340e-01
-1.61546350e-01 6.36955142e-01 -1.37203768e-01 -2.94207074e-02
2.92159561e-02 9.03308272e-01 -1.02553725e+00 -9.01004851e-01
-3.19985509e-01 -2.02550083e-01 -6.76836967e-01 2.06674621e-01
-6.46661222e-01 -9.53000486e-02 -8.79915714e-01 3.72844517e-01
4.47289437e-01 -4.81562138e-01 1.09651303e+00 4.77503359e-01
2.42767692e-01 9.51255381e-01 -7.26157963e-01 1.88905346e+00
-3.25351298e-01 2.77787328e-01 2.99587756e-01 -7.91345716e-01
9.31493521e-01 1.50406793e-01 1.28484992e-02 -2.12687895e-01
9.12303627e-02 1.94910005e-01 -1.31705333e-03 1.95346903e-02
5.99717736e-01 -1.70604244e-01 -5.90309083e-01 5.89420795e-01
3.14603671e-02 -2.25308403e-01 3.94904912e-01 7.25895941e-01
1.73363209e+00 9.15228873e-02 2.17626616e-01 -6.48757517e-01
1.46243304e-01 7.88940489e-01 5.11842072e-01 7.35444367e-01
-2.08082259e-01 1.15676284e-01 9.95403826e-01 -3.29577237e-01
-1.04740775e+00 -1.04556513e+00 -5.45938388e-02 1.20340741e+00
2.59237796e-01 -8.09305251e-01 -1.32510555e+00 -9.85014319e-01
-3.09575856e-01 6.97487056e-01 -2.28145853e-01 -3.55062820e-02
-1.08051264e+00 -7.59567320e-01 6.95107460e-01 1.00362289e+00
1.85363114e-01 -9.62095618e-01 -5.53407907e-01 5.03718495e-01
-2.15861574e-01 -1.67104328e+00 -3.19578171e-01 6.11659169e-01
-1.48323810e+00 -1.30920208e+00 3.63340050e-01 -1.19728363e+00
7.88477600e-01 9.98591557e-02 1.65682745e+00 3.75050515e-01
1.03312962e-01 4.20913428e-01 -5.54683447e-01 1.93283498e-01
-1.12879145e+00 1.54545352e-01 -4.67368066e-01 -9.84398067e-01
3.79664183e-01 -5.61436176e-01 -7.17852116e-02 -2.54650742e-01
-7.80587912e-01 9.20437798e-02 1.89638902e-02 7.09743977e-01
3.22768152e-01 1.32810861e-01 1.90292463e-01 -1.56190014e+00
6.92911565e-01 -3.68016481e-01 -6.42385244e-01 3.09219837e-01
-3.68468463e-01 5.46782553e-01 5.94624817e-01 2.51097279e-03
-1.18716049e+00 2.93908894e-01 -1.92687184e-01 3.15396041e-01
-1.13739610e-01 3.51509392e-01 -2.19121471e-01 1.52276725e-01
4.84623373e-01 -3.12083364e-01 -2.51192123e-01 -3.75086099e-01
7.13716269e-01 2.79927224e-01 5.28767586e-01 -1.16574502e+00
3.84873271e-01 1.45335779e-01 2.93166846e-01 -4.38917488e-01
-7.51412630e-01 -5.25738783e-02 -6.81631029e-01 4.00752187e-01
1.00263739e+00 -7.51559258e-01 -3.49864453e-01 -4.44155782e-02
-1.52241433e+00 -6.13108337e-01 -2.03966886e-01 1.42688036e-01
-7.28480339e-01 6.79708898e-01 -1.30165303e+00 -5.93581975e-01
-3.88564974e-01 -8.42260063e-01 1.11456633e+00 3.26268449e-02
-3.36505949e-01 -1.08745778e+00 3.24609637e-01 2.36559078e-01
-1.81837678e-01 7.28543699e-02 1.78750157e+00 -8.02229822e-01
-6.79859102e-01 1.71085745e-01 -2.94879973e-01 -1.76552817e-01
-2.30084121e-01 -1.29248783e-01 -4.84160781e-01 -5.90168498e-03
-1.24675483e-01 -1.38517007e-01 5.60915470e-01 3.05299580e-01
6.60784483e-01 -7.83162117e-02 -6.59593701e-01 2.95228451e-01
1.59838581e+00 4.38853383e-01 6.21263742e-01 2.42696166e-01
5.74907064e-01 5.92084050e-01 3.89991969e-01 -1.13427185e-01
4.48115617e-01 2.87924439e-01 8.73514190e-02 4.41449642e-01
-2.77016073e-01 -6.76820278e-01 2.81478286e-01 1.06518769e+00
1.76661327e-01 -3.65902245e-01 -9.98699129e-01 7.28338599e-01
-1.77192140e+00 -4.04944599e-01 -4.50581908e-01 1.81081116e+00
7.45120466e-01 5.71157217e-01 2.13709816e-01 1.69300046e-02
1.21878922e+00 1.22117072e-01 -8.93991888e-02 -1.19477642e+00
2.12448537e-01 8.63257170e-01 8.58892500e-01 9.33818221e-01
-7.56664038e-01 1.57327437e+00 8.78223705e+00 3.87527883e-01
-3.60669106e-01 3.23133290e-01 4.53154445e-01 5.27316511e-01
-6.61748827e-01 7.20110595e-01 -8.88306737e-01 1.35946423e-01
1.81060016e+00 -2.80406982e-01 1.25751317e-01 1.07603788e+00
-3.92967492e-01 -7.77877793e-02 -1.08822763e+00 2.79303730e-01
-4.27291870e-01 -1.21242857e+00 1.02434814e-01 -1.46561772e-01
2.53375322e-01 -1.00965947e-02 -8.30865622e-01 1.70835182e-01
1.19075024e+00 -6.64684951e-01 4.84293967e-01 -3.26669931e-01
6.19951427e-01 -7.01935172e-01 2.64024973e-01 1.04280166e-01
-1.55519414e+00 6.38560876e-02 -5.33212125e-01 -7.81005695e-02
3.41303676e-01 5.34677625e-01 -5.22641599e-01 4.59825248e-01
3.52426052e-01 1.05612211e-01 -3.85906190e-01 4.35649782e-01
-5.28267264e-01 6.90465093e-01 -3.43231291e-01 -5.57677113e-02
2.11834043e-01 -2.22867548e-01 3.37413669e-01 1.28116214e+00
1.39589041e-01 3.98348391e-01 3.88540447e-01 3.81680250e-01
-2.02422202e-01 4.99396548e-02 -7.10268080e-01 -1.61962599e-01
7.93481290e-01 9.40139771e-01 -1.35157764e+00 -5.35734832e-01
-6.02949202e-01 7.98136890e-01 6.54502809e-01 -4.97188233e-02
-5.30984759e-01 -2.56092608e-01 3.91790181e-01 -3.18658948e-02
4.94575173e-01 -7.65918314e-01 -3.28390628e-01 -1.01735699e+00
8.86329729e-03 -6.61059678e-01 7.91946411e-01 -7.98805296e-01
-8.99327397e-01 9.72577095e-01 2.61913717e-01 -1.14299677e-01
-4.85594779e-01 -6.80691302e-01 -6.09780729e-01 4.60086405e-01
-1.09548283e+00 -8.02070796e-01 1.93477526e-01 3.85001808e-01
5.25216520e-01 4.16803837e-01 1.13126254e+00 -3.26115876e-01
-3.77700478e-01 3.01703811e-01 -6.01304114e-01 2.79260516e-01
-1.33669212e-01 -1.44974232e+00 1.41065788e+00 1.19999051e+00
2.35364273e-01 5.82846522e-01 7.64195085e-01 -9.79941308e-01
-1.59727263e+00 -8.51819515e-01 1.29043436e+00 -4.50774044e-01
1.00683153e+00 -4.25826132e-01 -8.68706286e-01 1.16337097e+00
1.30398393e-01 6.09474070e-02 2.85608679e-01 2.91792035e-01
-8.36716354e-01 4.23025668e-01 -1.03031063e+00 4.80177134e-01
1.77876401e+00 -7.48776734e-01 -9.08632755e-01 3.83007467e-01
1.17501998e+00 -6.08742356e-01 -1.05683255e+00 1.32566065e-01
1.07803494e-01 -6.12678766e-01 5.66617787e-01 -9.87577140e-01
3.88393879e-01 2.20874876e-01 -2.14687690e-01 -1.05326021e+00
-2.67494500e-01 -8.96729529e-01 -1.97978333e-01 8.91934633e-01
6.31434619e-01 -6.82214081e-01 1.14487386e+00 8.93557310e-01
-2.91219920e-01 -2.88123995e-01 -1.05685413e+00 -9.02106702e-01
4.71739620e-01 -3.44807655e-01 5.65281510e-01 5.37296832e-01
5.10784805e-01 8.93446028e-01 3.28077286e-01 2.30294377e-01
6.64928973e-01 5.29294610e-01 3.05631846e-01 -1.26821947e+00
-5.25947750e-01 -7.13056186e-03 -3.60958189e-01 -1.25645161e+00
5.04078567e-01 -1.03529155e+00 8.79708529e-02 -1.51772594e+00
3.85041922e-01 -5.83855569e-01 1.21667884e-01 5.17435312e-01
-4.81187142e-02 -3.21634710e-01 1.69191211e-02 -4.68751276e-03
-5.43789744e-01 -1.38549060e-01 9.92577016e-01 2.56665260e-01
-1.07540764e-01 -3.99948329e-01 -9.03274298e-01 7.91886389e-01
8.19226086e-01 -9.43212152e-01 -6.31908953e-01 -6.26948118e-01
2.51419276e-01 9.15992737e-01 -2.53538013e-01 -4.88319993e-01
8.16501211e-03 -2.80965716e-01 -3.30268443e-01 -2.54177064e-01
4.47173230e-02 -3.18762273e-01 1.43379360e-01 4.38340515e-01
-6.34032845e-01 6.02716148e-01 9.44014788e-02 6.52165711e-01
-9.75389481e-02 -5.09558797e-01 7.00832725e-01 -4.52048659e-01
-7.77137518e-01 -7.76263326e-03 -5.40693998e-01 5.06212771e-01
7.12404907e-01 -1.54746801e-01 -7.47068942e-01 -1.80328131e-01
-8.49335670e-01 -9.88336802e-02 6.07106268e-01 3.92790675e-01
3.31359893e-01 -8.25240076e-01 -2.75996864e-01 -1.36508137e-01
-1.24831721e-01 -1.04415283e-01 -3.78774315e-01 1.14652105e-01
-7.33554423e-01 2.92079091e-01 -9.26871002e-02 -1.04533322e-01
-1.31197643e+00 6.23776853e-01 1.05434656e-02 -4.84283954e-01
-1.12054944e+00 6.72670424e-01 2.06578568e-01 -1.24741636e-01
-3.03708315e-01 -3.91151249e-01 1.53142691e-01 -6.58196330e-01
1.66250914e-01 -1.84372216e-02 -1.40226316e-02 -4.54019129e-01
-5.17818928e-01 2.26858735e-01 -9.06496495e-02 -3.11842501e-01
1.08767426e+00 -1.06334519e-02 -3.70646834e-01 7.10649788e-02
9.71481264e-01 -3.76023278e-02 -7.75671899e-01 -8.45940411e-02
6.36283636e-01 -1.97046176e-01 -2.23846629e-01 -2.18835324e-01
-7.67235458e-01 4.27156329e-01 -1.72563307e-02 5.81117272e-01
8.80027950e-01 4.67378229e-01 1.34575725e+00 4.30456281e-01
9.79643404e-01 -1.09709775e+00 -2.11421624e-01 6.64720893e-01
9.66301262e-02 -6.07792675e-01 -6.90555573e-02 -1.34575200e+00
-3.74852926e-01 1.03292346e+00 4.03592348e-01 -2.77809918e-01
5.68033099e-01 9.20761466e-01 -1.33296251e-01 -2.27099866e-01
-1.06549060e+00 -3.70130867e-01 -5.44062912e-01 8.04896057e-01
4.45114523e-01 4.39388692e-01 -8.33506107e-01 6.15070879e-01
-5.03492892e-01 -3.67175967e-01 8.72666240e-01 1.43027067e+00
-9.35558259e-01 -1.82193148e+00 -6.58634901e-02 1.51848152e-01
-6.49717033e-01 -3.32181931e-01 -5.84906697e-01 8.73686552e-01
-5.53953707e-01 1.31324136e+00 2.67904311e-01 -2.63753682e-01
1.48244603e-02 4.78829622e-01 1.01617992e+00 -1.09873438e+00
-5.71460783e-01 -1.74940526e-01 1.13117969e+00 -5.63351691e-01
-4.54422265e-01 -5.43236315e-01 -1.89391363e+00 -5.75797737e-01
-4.31048661e-01 5.77136517e-01 6.27612054e-01 9.40916896e-01
2.02352136e-01 2.11894229e-01 2.97249973e-01 6.39137328e-02
-3.06684166e-01 -5.53233862e-01 -6.35242403e-01 3.04145515e-01
-4.91972983e-01 -4.50562030e-01 -2.86717683e-01 4.36553471e-02] | [10.338287353515625, 9.560135841369629] |
a1505763-d44a-44f9-9045-d28325679356 | unsupervised-segmentation-incorporating-shape | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Kim_Unsupervised_Segmentation_Incorporating_Shape_Prior_via_Generative_Adversarial_Networks_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Kim_Unsupervised_Segmentation_Incorporating_Shape_Prior_via_Generative_Adversarial_Networks_ICCV_2021_paper.pdf | Unsupervised Segmentation Incorporating Shape Prior via Generative Adversarial Networks | We present an image segmentation algorithm that is developed in an unsupervised deep learning framework. The delineation of object boundaries often fails due to the nuisance factors such as illumination changes and occlusions. Thus, we initially propose an unsupervised image decomposition algorithm to obtain an intrinsic representation that is robust with respect to undesirable bias fields based on a multiplicative image model. The obtained intrinsic image is subsequently provided to an unsupervised segmentation procedure that is developed based on a piecewise smooth model. The segmentation model is further designed to incorporate a geometric constraint imposed in the generative adversarial network framework where the discrepancy between the distribution of partitioning functions and the distribution of prior shapes is minimized. We demonstrate the effectiveness and robustness of the proposed algorithm in particular with bias fields and occlusions using simple yet illustrative synthetic examples and a benchmark dataset for image segmentation. | ['Byung-Woo Hong', 'Dahye Kim'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['unsupervised-image-decomposition'] | ['computer-vision'] | [ 6.98523521e-01 3.98769855e-01 3.41788352e-01 -4.24398661e-01
-4.51799899e-01 -5.49736202e-01 4.18679684e-01 1.59094781e-01
-4.01213765e-01 7.08630383e-01 -2.98631936e-01 1.02588855e-01
-1.68354824e-01 -9.46917653e-01 -8.98028493e-01 -9.71448123e-01
3.31282824e-01 3.96356255e-01 2.37265602e-02 6.24680072e-02
3.23042929e-01 7.40036845e-01 -1.22869074e+00 -4.51599687e-01
1.30293512e+00 6.73903584e-01 -9.06225741e-02 6.10436976e-01
-4.62733060e-02 3.59990388e-01 -6.79083705e-01 -1.70700282e-01
6.96743131e-01 -4.22051787e-01 -7.18033969e-01 9.53679323e-01
2.30422258e-01 -1.48596883e-01 7.34342411e-02 1.40888917e+00
5.06415963e-02 4.46710169e-01 1.18745148e+00 -1.15355134e+00
-2.76044875e-01 1.14241585e-01 -6.56658173e-01 -2.61032909e-01
-1.96791410e-01 2.01741636e-01 5.87869525e-01 -5.53096235e-01
6.54347658e-01 9.14845347e-01 4.44369525e-01 2.02500597e-01
-1.75274217e+00 -1.53299421e-01 -4.07186076e-02 -5.36439598e-01
-1.38057995e+00 -1.78626359e-01 1.24980700e+00 -6.43384874e-01
1.97865795e-02 1.18416272e-01 3.31275314e-01 6.75750911e-01
1.63196445e-01 4.73370075e-01 1.07348466e+00 -6.11718833e-01
5.12558520e-01 3.69485170e-01 1.07091568e-01 4.53367591e-01
3.45305502e-01 9.83769596e-02 1.89677149e-01 1.93060208e-02
1.00243783e+00 -1.17792815e-01 -2.10972149e-02 -8.25247765e-01
-5.45649946e-01 8.07729959e-01 5.99239230e-01 3.69116187e-01
-4.70947027e-01 -3.31252553e-02 3.16096805e-02 -8.10309350e-02
4.90991265e-01 4.46609259e-01 -1.58461686e-02 6.15725636e-01
-1.24017823e+00 1.75516054e-01 6.97835028e-01 9.52438831e-01
1.06786835e+00 3.93505454e-01 2.77892500e-02 8.61754537e-01
4.24360424e-01 2.99391478e-01 6.94441125e-02 -1.06635177e+00
2.32095048e-02 7.82463253e-01 3.64218712e-01 -1.40606308e+00
-1.39765769e-01 -7.07553148e-01 -9.12026763e-01 5.95769703e-01
5.89995742e-01 -9.06681791e-02 -1.23422706e+00 1.69832671e+00
5.91508567e-01 -1.31088167e-01 8.48506242e-02 7.68350065e-01
4.83737022e-01 4.74377424e-01 8.78476277e-02 -2.09811449e-01
7.78145373e-01 -7.13683963e-01 -6.56294763e-01 -6.43660501e-02
1.95478305e-01 -6.60701036e-01 8.94042253e-01 2.98344374e-01
-1.15323710e+00 -6.83482409e-01 -1.17166710e+00 7.98577592e-02
-2.47519895e-01 1.08055338e-01 5.88450953e-02 8.63152027e-01
-9.23510671e-01 5.02178311e-01 -9.60304260e-01 -3.95776153e-01
5.45859754e-01 5.02358198e-01 -8.78503174e-02 3.09423774e-01
-6.69845700e-01 5.87051213e-01 5.32786787e-01 2.90073246e-01
-8.21339846e-01 -4.82846200e-01 -9.00642753e-01 -7.35608414e-02
1.18737824e-01 -6.94421053e-01 5.89046299e-01 -1.57112908e+00
-1.61797059e+00 8.41215909e-01 6.36363626e-02 -6.72117531e-01
8.76700401e-01 1.13741249e-01 1.63232401e-01 3.43830436e-01
-7.64843076e-02 7.64758587e-01 1.22569191e+00 -1.81342423e+00
-1.71206146e-01 -2.42986277e-01 -5.77603206e-02 1.60889044e-01
-1.39399484e-01 -3.88227820e-01 -2.81225562e-01 -8.52342904e-01
3.48185897e-01 -7.47786283e-01 -5.78902602e-01 -5.93574345e-02
-7.36609995e-01 4.69284981e-01 9.59383070e-01 -5.61332405e-01
6.09761894e-01 -1.97141886e+00 2.57138699e-01 7.06042767e-01
2.78157257e-02 -9.79873445e-03 1.20660037e-01 1.16447218e-01
-1.68163806e-01 1.54932052e-01 -1.07980490e+00 -3.39964807e-01
-2.02323049e-01 2.98007488e-01 -6.37239739e-02 7.74360955e-01
2.56355405e-01 5.77111602e-01 -5.23196638e-01 -4.51729715e-01
4.62473720e-01 5.71448922e-01 -5.12436867e-01 3.53091180e-01
-3.06816608e-01 9.86700118e-01 -4.23186332e-01 2.97922313e-01
9.57236290e-01 1.26530990e-01 -6.89736903e-02 -2.39206553e-01
-4.27791476e-02 -2.62503654e-01 -1.21808791e+00 1.46373248e+00
-1.89605832e-01 3.05665910e-01 3.29294533e-01 -1.39594066e+00
1.19835639e+00 5.16285561e-02 6.95914030e-01 -1.26040459e-01
4.98341411e-01 1.28932625e-01 -3.97515595e-02 -1.96028158e-01
2.39225388e-01 -1.48870870e-01 1.79086924e-01 2.26682618e-01
3.07860337e-02 -5.39120495e-01 2.23429099e-01 1.69890404e-01
5.78953445e-01 2.13837504e-01 -8.32140446e-03 -6.29810631e-01
8.01039636e-01 6.16562814e-02 5.80294609e-01 5.03273129e-01
-3.13795879e-02 9.17999625e-01 4.71798688e-01 -3.18448663e-01
-1.25167823e+00 -9.41150010e-01 -3.62163365e-01 3.08846682e-01
3.07442188e-01 4.27710086e-01 -1.36978543e+00 -6.17254674e-01
-2.61631489e-01 7.02297926e-01 -7.56970465e-01 -8.72683823e-02
-3.19600910e-01 -8.18816960e-01 2.84930974e-01 3.79822612e-01
7.27628410e-01 -1.01436520e+00 -6.01438582e-01 2.98086017e-01
1.01120889e-01 -1.11064696e+00 -2.99006164e-01 2.00992867e-01
-9.18691754e-01 -1.17247868e+00 -8.59258771e-01 -9.08395529e-01
1.36283720e+00 -3.64802092e-01 8.41438115e-01 1.08294025e-01
-2.86758751e-01 5.59586883e-01 1.22978864e-02 -1.55747205e-01
-7.46293306e-01 -3.65841985e-02 -2.35174969e-01 5.38002551e-01
-1.00471623e-01 -7.63805807e-01 -7.20274270e-01 2.98893988e-01
-1.37556052e+00 2.39835009e-02 1.78833678e-01 7.83932388e-01
6.70509338e-01 4.78182107e-01 4.99800891e-01 -1.04367435e+00
2.84864783e-01 -1.67774647e-01 -9.58393216e-01 -7.56198838e-02
-5.00028729e-01 -6.33654594e-02 6.02120578e-01 -2.90443748e-01
-1.39690053e+00 5.16906440e-01 -1.37866158e-02 -3.15441340e-01
-7.00005114e-01 1.07796177e-01 -5.47702909e-01 -3.69241595e-01
5.33478677e-01 4.01777118e-01 1.55534074e-01 -2.57042468e-01
3.83516222e-01 2.92408258e-01 6.67253792e-01 -7.77382195e-01
1.22645855e+00 7.16303408e-01 1.98267907e-01 -1.00395691e+00
-4.44467783e-01 -6.30405918e-02 -9.78616595e-01 -3.13865155e-01
9.74813223e-01 -3.47806334e-01 -2.33258918e-01 7.63471961e-01
-1.09549677e+00 -3.05988997e-01 -4.19193178e-01 1.53721809e-01
-8.46475005e-01 5.85068464e-01 -4.95476604e-01 -9.08118188e-01
-2.96181798e-01 -1.40185821e+00 6.01684034e-01 4.51684147e-01
-1.76139288e-02 -1.23322189e+00 -1.08389132e-01 3.35450232e-01
1.10771805e-01 8.62823367e-01 1.01163816e+00 -5.26816487e-01
-7.27194667e-01 -4.13090326e-02 -1.78308487e-02 9.24632549e-01
3.54117304e-01 3.91102314e-01 -7.65585065e-01 -1.69345364e-01
3.77178639e-01 -1.33502662e-01 6.83558226e-01 6.93204701e-01
9.30892110e-01 -2.95604110e-01 -1.20589957e-01 8.63238811e-01
1.79490387e+00 4.55157131e-01 8.42722178e-01 3.73859137e-01
8.10756564e-01 8.41620207e-01 4.44496989e-01 1.36003166e-01
-1.06235944e-01 2.28514552e-01 5.31509221e-01 -4.83347744e-01
1.08887918e-01 2.89746951e-02 -2.88983136e-01 4.34499890e-01
4.46445197e-02 -1.63849592e-01 -8.20469618e-01 7.64555156e-01
-1.62141228e+00 -3.97250563e-01 -2.15206400e-01 2.30963349e+00
6.45649910e-01 3.64434719e-01 1.13738021e-02 2.64962643e-01
9.63466525e-01 3.50675844e-02 -5.69658041e-01 -5.32038033e-01
-7.85656422e-02 2.00014770e-01 5.69136083e-01 6.59505308e-01
-1.22808599e+00 9.16912615e-01 6.02418518e+00 6.25381589e-01
-1.04716754e+00 -2.86295593e-01 1.00727057e+00 7.06997633e-01
-4.14078593e-01 9.52049159e-03 -3.53134423e-01 3.83682728e-01
1.13533616e-01 1.33180737e-01 1.41690254e-01 7.10129499e-01
2.67091393e-01 -2.02185676e-01 -7.76959181e-01 5.20009577e-01
-6.88025355e-02 -9.74365592e-01 1.36790857e-01 1.89397223e-02
1.16710043e+00 -5.46212077e-01 2.32588425e-01 -2.95314640e-01
1.45288214e-01 -1.01138067e+00 5.25308430e-01 4.43551511e-01
3.75739217e-01 -1.01458418e+00 4.16698933e-01 3.84926319e-01
-7.40496039e-01 2.43067771e-01 -2.30476558e-01 2.33262017e-01
3.81555036e-02 5.19887388e-01 -6.96538150e-01 3.90828490e-01
1.19995117e-01 3.88700366e-01 -3.04823011e-01 9.90004838e-01
-2.84125388e-01 5.56405425e-01 -4.77948070e-01 4.26100105e-01
3.89496446e-01 -8.35232913e-01 9.13061440e-01 9.11955297e-01
-5.10931239e-02 5.18713593e-02 1.24761559e-01 1.43515193e+00
-1.10483229e-01 3.83937001e-01 -6.28013372e-01 1.64931625e-01
1.04293242e-01 1.25036991e+00 -1.31674993e+00 -1.53902918e-01
-1.03316583e-01 8.57587576e-01 1.66761894e-02 8.35769117e-01
-7.93973982e-01 -1.83402449e-01 1.48382723e-01 2.18548283e-01
3.40757102e-01 -2.33868018e-01 -7.05560565e-01 -8.25217664e-01
-1.02329023e-01 -7.14703023e-01 -5.41811110e-03 -4.63726044e-01
-1.06561649e+00 6.10577941e-01 2.37602182e-02 -1.10203195e+00
-7.84284920e-02 -3.04593384e-01 -8.87878597e-01 7.91227043e-01
-1.39816010e+00 -1.05854356e+00 -2.22183317e-01 5.06529808e-01
5.58450937e-01 -8.20738301e-02 5.20683646e-01 -5.31272069e-02
-5.71423113e-01 3.25045884e-01 2.54311860e-01 2.09954843e-01
3.38510156e-01 -1.28104413e+00 -9.99488980e-02 1.23657084e+00
-1.94638133e-01 5.75525761e-01 1.00988877e+00 -6.22541487e-01
-7.42394149e-01 -1.22285700e+00 2.27665976e-01 -1.42128440e-02
1.64234444e-01 -2.91560203e-01 -1.02243900e+00 6.79610372e-01
2.52217412e-01 4.04757522e-02 3.28209490e-01 -6.03932142e-01
1.16634868e-01 -2.02431440e-01 -1.60778451e+00 5.21259904e-01
3.46089333e-01 -9.27437842e-02 -5.45766294e-01 1.69544354e-01
2.94050157e-01 -3.61546457e-01 -7.68127918e-01 5.85989714e-01
2.39636704e-01 -8.80701602e-01 9.05653596e-01 -3.86009902e-01
3.85561109e-01 -3.25638294e-01 1.27851456e-01 -1.12388635e+00
-9.89163890e-02 -7.57175982e-01 4.15494919e-01 1.46301162e+00
1.87390730e-01 -6.11752212e-01 8.72281969e-01 8.58358264e-01
2.06752479e-01 -5.89540005e-01 -5.68756342e-01 -5.55176616e-01
5.64814024e-02 -8.45394582e-02 1.77673161e-01 8.28105569e-01
-7.91922808e-01 -4.46643606e-02 -6.08316064e-02 4.28595603e-01
1.11939669e+00 1.61278009e-01 8.43668878e-01 -1.03716075e+00
-1.16963454e-01 -1.17067106e-01 -3.18457425e-01 -8.22781086e-01
1.50112331e-01 -4.35351491e-01 3.34566176e-01 -1.20927405e+00
-8.17018300e-02 -5.26091874e-01 -2.90287912e-01 -2.81065349e-02
5.02119809e-02 3.46671999e-01 3.76962475e-04 5.37439771e-02
1.10108837e-01 6.13812447e-01 1.31941688e+00 -2.60467261e-01
-5.28822601e-01 1.50372416e-01 -4.76018399e-01 1.20339131e+00
8.00725222e-01 -5.80704331e-01 -5.95956922e-01 -2.44490907e-01
-2.90746599e-01 -4.48144227e-03 4.17882323e-01 -9.53737020e-01
8.34072530e-02 2.78044436e-02 4.55272198e-01 -5.42738736e-01
1.15244955e-01 -1.10235858e+00 2.81594515e-01 2.62122095e-01
-3.77473325e-01 -5.64822614e-01 1.58496812e-01 5.40494859e-01
-1.66314676e-01 -4.51400280e-01 1.33134830e+00 -5.30885197e-02
-2.14787781e-01 1.15483008e-01 -3.26930195e-01 -2.47891322e-02
1.09896636e+00 -5.49727559e-01 3.32813770e-01 -4.05428916e-01
-9.44987774e-01 -6.40866384e-02 7.23862648e-01 -2.21837386e-02
3.62145990e-01 -1.06889367e+00 -5.07349610e-01 3.72306764e-01
-3.51410419e-01 3.46017152e-01 7.67337084e-02 6.72454119e-01
-8.94152761e-01 -4.44246456e-02 -3.86636674e-01 -7.18228042e-01
-9.34691072e-01 6.13824368e-01 5.46240807e-01 -1.71688005e-01
-5.24652660e-01 5.23712695e-01 5.54572999e-01 -2.65696824e-01
2.48978525e-01 -4.33863014e-01 -1.73745811e-01 -2.21813112e-01
-5.09431809e-02 2.33828202e-01 -2.34678224e-01 -9.09618080e-01
-1.34020895e-01 7.70703554e-01 1.44520506e-01 -7.54835978e-02
1.18247318e+00 -2.84940839e-01 -5.86600006e-02 -2.98702773e-02
1.07527936e+00 4.38348651e-02 -1.62306023e+00 1.25608534e-01
-1.47760525e-01 -3.80453825e-01 6.35024309e-02 -4.13620085e-01
-1.35542035e+00 6.77104771e-01 5.87322652e-01 3.44795853e-01
1.26720834e+00 -5.73855221e-01 6.97172046e-01 -1.19831018e-01
7.28831068e-02 -1.11037326e+00 -2.39403937e-02 1.80796936e-01
7.24161327e-01 -1.05334830e+00 -8.76548961e-02 -6.90930128e-01
-2.54312932e-01 1.05692124e+00 4.52106416e-01 -6.86581671e-01
6.47380650e-01 3.65834683e-01 1.55512363e-01 4.28966098e-02
2.25305095e-01 -2.48643503e-01 3.47833455e-01 5.71430206e-01
6.70596883e-02 -2.37448573e-01 -6.82071567e-01 3.67871225e-01
1.73696697e-01 -9.02225599e-02 3.22768420e-01 5.98097801e-01
-2.32909456e-01 -1.06193995e+00 -5.58326900e-01 -1.83454916e-01
-5.75040579e-01 2.24903345e-01 -2.90379882e-01 9.52826321e-01
3.53046328e-01 7.38306344e-01 9.81507450e-02 1.69932321e-01
1.39250055e-01 -4.28999774e-03 4.29978997e-01 -4.78007048e-01
-4.44428146e-01 5.89709699e-01 -4.22730327e-01 -5.45457117e-02
-5.55299044e-01 -3.62628192e-01 -1.37253213e+00 4.09369111e-01
-2.55462736e-01 -3.43548879e-03 6.39795184e-01 9.64406252e-01
-1.17567107e-01 5.94055712e-01 7.31325686e-01 -8.63161802e-01
-1.72642305e-01 -8.09484422e-01 -5.82359970e-01 6.41847849e-01
1.45461708e-01 -6.30744278e-01 -4.92777765e-01 5.90129435e-01] | [10.277364730834961, 0.1702558696269989] |
13bcfa8d-455d-4e1e-841d-4ce92c6a20f8 | derived-metrics-for-the-game-of-go-intrinsic | 2009.01606 | null | https://arxiv.org/abs/2009.01606v3 | https://arxiv.org/pdf/2009.01606v3.pdf | Derived metrics for the game of Go -- intrinsic network strength assessment and cheat-detection | The widespread availability of superhuman AI engines is changing how we play the ancient game of Go. The open-source software packages developed after the AlphaGo series shifted focus from producing strong playing entities to providing tools for analyzing games. Here we describe two ways of how the innovations of the second generation engines (e.g.~score estimates, variable komi) can be used for defining new metrics that help deepen our understanding of the game. First, we study how much information the search component contributes in addition to the raw neural network policy output. This gives an intrinsic strength measurement for the neural network. Second, we define the effect of a move by the difference in score estimates. This gives a fine-grained, move-by-move performance evaluation of a player. We use this in combating the new challenge of detecting online cheating. | ['Antti Törmänen', 'Attila Egri-Nagy'] | 2020-09-03 | null | null | null | null | ['game-of-go'] | ['playing-games'] | [-2.54270673e-01 -4.21273917e-01 -1.73107237e-01 -8.98076445e-02
-3.19177151e-01 -9.07960236e-01 4.37214613e-01 -5.61401844e-02
-1.00135159e+00 6.31141424e-01 1.87137038e-01 -3.05257440e-01
-3.46571267e-01 -9.44162369e-01 -2.76382923e-01 -5.65125763e-01
-2.15950623e-01 3.26056778e-01 5.18667340e-01 -6.68573439e-01
5.47252059e-01 1.13970257e-01 -1.61813164e+00 1.81328490e-01
5.25091112e-01 8.11624527e-01 -4.92402762e-02 1.10108864e+00
1.50880724e-01 1.06372190e+00 -1.08164382e+00 -7.03102231e-01
3.44751686e-01 -5.94904542e-01 -8.50309312e-01 -9.36562598e-01
7.86961764e-02 -4.70836401e-01 -2.65766919e-01 1.15338504e+00
4.80899513e-01 3.15839261e-01 2.64422476e-01 -1.19543672e+00
-2.15922624e-01 9.86850798e-01 -1.74129441e-01 7.53417969e-01
5.86582780e-01 5.65542161e-01 1.47801208e+00 6.57211170e-02
7.76590705e-01 8.38730574e-01 6.72674477e-01 4.46969897e-01
-1.00734854e+00 -6.01376414e-01 -4.38542664e-01 4.02194858e-01
-1.10823250e+00 -9.52788442e-02 7.64071047e-01 -6.03124022e-01
9.24823046e-01 6.06847942e-01 8.23245823e-01 1.17426574e+00
-5.18749319e-02 7.44779885e-01 9.43288803e-01 -2.39322484e-01
1.55110464e-01 -1.10685870e-01 3.36344421e-01 6.58252358e-01
3.63637924e-01 4.55062777e-01 -7.69462347e-01 -3.22301954e-01
7.33447015e-01 -3.69460315e-01 5.25891967e-02 8.12050402e-02
-7.22025633e-01 1.09792304e+00 3.52666438e-01 6.33236945e-01
-3.89912516e-01 4.90334839e-01 4.44890112e-01 4.09561545e-01
2.80947655e-01 1.20379245e+00 -3.75888050e-01 -1.21337378e+00
-9.55767572e-01 8.08018684e-01 1.14363754e+00 8.87361839e-02
2.90415853e-01 -1.00527704e-01 -1.88616589e-01 5.89862883e-01
2.56827418e-02 -3.45279239e-02 7.35174060e-01 -1.35615420e+00
3.35865051e-01 4.79350984e-01 1.06677741e-01 -1.23859632e+00
-6.53758407e-01 -5.28351426e-01 3.45648080e-02 5.21839559e-01
8.16970468e-01 -3.60003293e-01 -3.53559166e-01 1.75555491e+00
-9.39634442e-02 -2.22805247e-01 -6.14619434e-01 8.70414436e-01
4.73895729e-01 1.45098343e-01 -1.64184545e-03 3.91889721e-01
1.14165068e+00 -8.46453071e-01 -2.93211341e-01 -1.80620268e-01
9.04177427e-01 -2.77754247e-01 1.12161946e+00 7.52371907e-01
-1.23751104e+00 -2.33607039e-01 -1.11699164e+00 2.28189453e-01
-7.54857242e-01 -5.68561971e-01 8.01873863e-01 1.03404438e+00
-1.02935743e+00 1.00330281e+00 -7.07135022e-01 -7.33679160e-02
2.52819926e-01 2.71609753e-01 -3.65089066e-02 5.44727027e-01
-1.49294043e+00 1.12236989e+00 5.99088252e-01 -5.59985936e-01
-4.57916468e-01 -4.50481951e-01 -3.57238233e-01 4.24426913e-01
8.39707732e-01 -3.07605237e-01 1.51870883e+00 -8.94523203e-01
-1.55784988e+00 8.89214396e-01 5.45033157e-01 -5.12603581e-01
6.78110123e-01 -1.83200151e-01 -1.47165880e-02 -2.35694706e-01
1.08468987e-01 1.40744582e-01 4.90492851e-01 -6.66854143e-01
-1.12166691e+00 -3.61879617e-01 3.78972322e-01 7.09431842e-02
-2.42391780e-01 2.81120628e-01 -4.90921289e-01 -4.29533422e-01
-4.65394646e-01 -8.63094628e-01 -8.42743274e-03 -5.52873731e-01
-2.39661947e-01 -4.34890658e-01 7.48872906e-02 -7.46278346e-01
1.98695087e+00 -1.89803040e+00 3.28619182e-01 4.55843836e-01
7.13634968e-01 4.23672348e-01 -6.88760281e-02 3.66265148e-01
1.06344230e-01 4.53884602e-01 3.16671193e-01 1.37334555e-01
3.50151032e-01 -1.60798639e-01 3.61181915e-01 1.27127796e-01
-2.91711032e-01 1.09022570e+00 -9.71582115e-01 -1.50646478e-01
-1.25426397e-01 -8.36190432e-02 -9.20651317e-01 -4.57481146e-05
1.80031955e-02 -8.41156840e-02 -4.86514062e-01 4.80363935e-01
5.82109112e-03 -1.64741740e-01 -7.54169598e-02 5.23774862e-01
-3.34745586e-01 4.96601522e-01 -8.83740127e-01 1.44210851e+00
4.40550558e-02 8.95694911e-01 2.29334477e-02 -6.58191860e-01
3.89962435e-01 -1.99358150e-01 4.99880880e-01 -9.30165827e-01
5.85264921e-01 1.32518411e-01 6.28538787e-01 -3.75549316e-01
7.55363166e-01 3.06434602e-01 -3.77162606e-01 6.15267038e-01
8.94065797e-02 1.62884742e-01 4.91167396e-01 1.73510462e-01
1.66545045e+00 8.87088925e-02 3.69863778e-01 -1.87433183e-01
1.85502261e-01 2.40243018e-01 3.43104780e-01 1.26926959e+00
-4.63366121e-01 1.39301538e-01 9.80614960e-01 -4.76030171e-01
-1.12536538e+00 -8.35309148e-01 3.32893491e-01 1.87280452e+00
-1.41220331e-01 -8.25325847e-01 -1.05698502e+00 -3.88026565e-01
2.21259937e-01 5.53952992e-01 -7.22241104e-01 -4.52930957e-01
-4.87293780e-01 -4.86016601e-01 1.15245330e+00 3.33205104e-01
5.05364299e-01 -1.17330647e+00 -1.07570934e+00 2.75192678e-01
-1.90286011e-01 -5.53087890e-01 -2.57449478e-01 3.62889975e-01
-4.57525969e-01 -1.13833570e+00 -3.82662296e-01 -1.43002853e-01
-3.23720068e-01 -1.43395271e-03 1.58786190e+00 5.07459402e-01
-4.18428898e-01 1.86009496e-01 -4.73329723e-01 -4.01691496e-01
-2.30270565e-01 4.80252922e-01 1.21076517e-01 -6.65710866e-01
7.16767311e-01 -5.65277636e-01 -4.60643291e-01 1.93700403e-01
-6.75711632e-01 -3.31773728e-01 4.10812080e-01 6.87662005e-01
-2.63779759e-01 1.69497058e-01 1.15996033e-01 -7.85933912e-01
1.67796540e+00 -5.77886760e-01 -5.40049911e-01 -1.02253266e-01
-7.58395612e-01 2.23026261e-01 2.53219634e-01 -4.97814476e-01
-5.10768235e-01 -4.40758675e-01 -1.15315035e-01 -2.30055135e-02
1.37148440e-01 7.66662657e-01 3.82647663e-01 -2.34539807e-03
1.15840554e+00 -1.46687523e-01 -3.58318500e-02 -5.22687376e-01
3.57890964e-01 5.55924177e-01 6.65157855e-01 -7.33814597e-01
7.25417495e-01 -8.35395008e-02 -2.82928556e-01 -5.27091861e-01
-3.55351388e-01 -6.00624323e-01 -3.19325000e-01 -4.39276725e-01
9.70124066e-01 -1.21613301e-01 -1.68474710e+00 6.30039811e-01
-7.90655077e-01 -5.75798333e-01 -2.89933234e-01 2.12991446e-01
-5.94665349e-01 1.40949667e-01 -5.43003619e-01 -9.17308629e-01
-1.99923381e-01 -9.65542316e-01 3.56250882e-01 6.47934675e-01
-6.47263706e-01 -8.77676845e-01 6.41435027e-01 4.04929906e-01
6.83375418e-01 1.59160048e-02 6.28117919e-01 -9.84172821e-01
-2.49355048e-01 -3.90017748e-01 4.69630100e-02 2.76162088e-01
-4.19565380e-01 -2.44727312e-03 -7.40115106e-01 1.10641435e-01
-1.44755274e-01 -1.23238690e-01 6.39790833e-01 4.23452467e-01
8.10309291e-01 -2.16432735e-01 -1.24928758e-01 4.88706380e-01
1.11460996e+00 2.80068547e-01 5.14490664e-01 8.72948527e-01
4.24232095e-01 5.00018537e-01 2.61547238e-01 4.13259387e-01
7.31953979e-02 9.74567950e-01 2.96527535e-01 3.83570760e-01
2.41832912e-01 -2.63169974e-01 4.18740809e-01 3.45404387e-01
-7.16753066e-01 -2.62827426e-01 -1.04357612e+00 2.01740548e-01
-1.96686316e+00 -1.42420971e+00 2.75652707e-02 2.15029645e+00
7.41485655e-01 5.66043556e-01 9.35479283e-01 1.19044632e-01
4.37765658e-01 2.25331083e-01 -3.83569360e-01 -7.27772534e-01
2.67910719e-01 4.10349190e-01 8.10230792e-01 5.20846069e-01
-7.98679590e-01 1.03265738e+00 7.27273321e+00 1.18667591e+00
-8.08115780e-01 1.53857872e-01 2.55505592e-01 -4.93408203e-01
1.80165395e-01 -2.21540973e-01 -5.06373882e-01 7.86617815e-01
1.14188528e+00 -3.62222046e-01 1.01961994e+00 1.03421068e+00
1.53678849e-01 -5.23727596e-01 -9.01327133e-01 1.00500369e+00
-8.88207257e-02 -1.16426837e+00 -5.96303225e-01 4.55663294e-01
1.90759599e-01 3.61670941e-01 7.40260780e-02 5.89242816e-01
1.04293811e+00 -1.04802418e+00 7.17184901e-01 7.32380152e-01
3.46117705e-01 -6.31543577e-01 7.10031152e-01 5.35231233e-01
-8.18810523e-01 -3.30099434e-01 -8.55741650e-02 -6.30199730e-01
-1.28551483e-01 -6.85157925e-02 -6.37718022e-01 -9.00559500e-03
9.02064323e-01 1.27157837e-01 -7.20781684e-01 1.24130738e+00
-2.26473719e-01 7.36932099e-01 -5.00102460e-01 -5.86244285e-01
4.44674224e-01 -1.35052547e-01 7.51970589e-01 9.22746599e-01
-1.19000040e-01 2.21270528e-02 -1.17633089e-01 1.29645395e+00
7.22449720e-02 -1.48782209e-01 -4.63484317e-01 -5.37436366e-01
3.63264561e-01 1.20040500e+00 -5.66641390e-01 -9.48467702e-02
-1.48820221e-01 8.26750278e-01 3.89584005e-01 -3.75246406e-02
-8.73582602e-01 -7.07751393e-01 1.08129394e+00 7.36780912e-02
2.45442744e-02 -3.17873567e-01 -3.69908750e-01 -9.28133070e-01
-2.10969016e-01 -1.05641568e+00 4.01884884e-01 -6.61520720e-01
-1.09274805e+00 2.49408707e-01 -5.26407920e-02 -6.28639221e-01
-7.48896539e-01 -8.30817223e-01 -9.33383107e-01 8.48700047e-01
-7.96983778e-01 -3.79903793e-01 -2.21580267e-01 2.74595529e-01
3.03782642e-01 -3.49822640e-01 5.43470621e-01 3.27772170e-01
-5.47888756e-01 7.73055792e-01 1.29874155e-01 4.99749422e-01
4.85617280e-01 -1.37662542e+00 5.76607823e-01 6.07224643e-01
2.71770507e-01 6.02816701e-01 7.66362369e-01 -5.09908497e-01
-8.84810269e-01 1.61190733e-01 4.40148324e-01 -1.06095397e+00
1.00973392e+00 -1.97721630e-01 -4.28487360e-01 2.60744005e-01
-2.68484801e-01 -7.06111550e-01 6.02687597e-01 6.42800152e-01
-2.36412227e-01 2.76718676e-01 -8.82480502e-01 5.23738027e-01
1.22531939e+00 -5.79995155e-01 -4.93713826e-01 -2.51934022e-01
4.20211434e-01 -2.62200892e-01 -6.50902569e-01 -4.79747094e-02
1.12793458e+00 -1.28998768e+00 8.36641192e-01 -1.19213414e+00
5.22912681e-01 2.62806058e-01 8.67689624e-02 -1.31299317e+00
-7.40207553e-01 -7.28549659e-01 5.41268103e-02 6.77180231e-01
5.82092464e-01 -5.64561009e-01 1.19574857e+00 8.65358055e-01
2.66121298e-01 -6.82413816e-01 -8.40008974e-01 -7.56351292e-01
1.56009778e-01 -6.39317751e-01 4.60063905e-01 9.62243557e-01
5.22529006e-01 8.96245912e-02 -3.39332044e-01 -4.93039966e-01
2.92411834e-01 -6.68906152e-01 8.07123601e-01 -1.54764593e+00
-7.89498389e-01 -1.24771845e+00 -8.70017231e-01 -8.23014617e-01
-3.75119150e-01 -6.24244928e-01 -1.98294550e-01 -1.01697540e+00
3.35035771e-01 -1.45696998e-01 -4.41020072e-01 3.76960188e-01
-1.95494682e-01 2.61165142e-01 5.38039029e-01 2.23870516e-01
-7.30935574e-01 -2.81107962e-01 7.97044337e-01 1.68212399e-01
-3.33485484e-01 7.62597769e-02 -7.12448478e-01 8.23017180e-01
1.02522886e+00 -6.05426192e-01 -6.42514676e-02 -1.83597356e-01
1.00116277e+00 -2.75129646e-01 3.13387096e-01 -1.26674020e+00
3.88530582e-01 -3.30924839e-01 1.59644522e-02 2.21279655e-02
1.12642169e-01 -5.22658527e-01 -1.70670062e-01 5.95640242e-01
-4.32501018e-01 1.51608989e-01 3.18394601e-02 2.68676221e-01
4.37743664e-02 -4.39182967e-01 4.78323072e-01 -3.03196937e-01
-5.70652425e-01 -1.13472350e-01 -4.90498543e-01 2.29326487e-01
8.45803738e-01 -4.74260598e-01 -5.48382878e-01 -7.82716870e-01
-7.46492326e-01 1.86791524e-01 3.37657124e-01 5.00611961e-01
-1.42610386e-01 -9.82548356e-01 -4.91490096e-01 6.95649311e-02
-1.92111790e-01 -8.72908413e-01 -2.00586189e-02 7.12349355e-01
-7.90733635e-01 5.00066578e-01 -7.70788729e-01 -1.19431645e-01
-1.55642331e+00 2.43725106e-01 6.48502469e-01 -6.32374406e-01
-1.46077886e-01 1.11305761e+00 -3.02741379e-01 -2.17396230e-01
2.67796725e-01 2.10771691e-02 -2.41018042e-01 1.02698401e-01
4.66682136e-01 7.34246373e-01 -1.65134162e-01 -2.98404694e-01
-2.53212512e-01 -5.91023192e-02 -9.62312296e-02 -5.41799843e-01
1.42963183e+00 3.73018563e-01 1.57337755e-01 5.03868520e-01
8.60535920e-01 8.67518261e-02 -8.04334521e-01 5.93488961e-02
1.79386079e-01 -5.32464147e-01 1.99122339e-01 -1.08130038e+00
-7.15898633e-01 5.10584295e-01 4.72838730e-01 8.18519175e-01
6.20879769e-01 -4.53792810e-02 5.90775013e-01 6.45359814e-01
5.26501298e-01 -1.52484167e+00 6.60241544e-02 7.31431305e-01
2.17724949e-01 -9.43241298e-01 -8.64899904e-02 2.84837395e-01
-5.50191164e-01 9.31890249e-01 5.85605621e-01 -1.92746028e-01
4.00470018e-01 6.28058687e-02 -1.51260316e-01 -6.25741780e-01
-5.23343742e-01 -4.76717770e-01 1.45320535e-01 3.91343832e-01
4.09979135e-01 1.62441686e-01 -7.57334471e-01 1.01374590e+00
-9.83249485e-01 4.58514467e-02 4.21125025e-01 9.18679833e-01
-7.30384409e-01 -1.06553781e+00 -2.38003895e-01 7.62077630e-01
-6.66234732e-01 9.57859773e-03 -1.08527303e+00 9.74216521e-01
1.25479445e-01 7.86758006e-01 -6.12346567e-02 -1.19039524e+00
3.86335641e-01 2.22835138e-01 2.85379916e-01 -4.99128759e-01
-1.06400061e+00 -5.86965561e-01 3.80395740e-01 -1.04702830e+00
1.40438288e-01 -4.73459512e-01 -6.72442138e-01 -9.57935393e-01
-5.40613592e-01 3.26689094e-01 7.85208285e-01 7.41108477e-01
1.23935029e-01 4.67978299e-01 2.79601604e-01 -9.18114364e-01
-6.85320258e-01 -1.02708900e+00 -6.73990965e-01 3.73374194e-01
-1.69696614e-01 -6.92594767e-01 -6.24263346e-01 -5.67090154e-01] | [3.4663586616516113, 1.4218956232070923] |
d0943b7b-5812-46eb-8f1a-f57c80b7dea3 | case-based-reasoning-for-better-1 | 2110.08470 | null | https://arxiv.org/abs/2110.08470v3 | https://arxiv.org/pdf/2110.08470v3.pdf | Case-based Reasoning for Better Generalization in Textual Reinforcement Learning | Text-based games (TBG) have emerged as promising environments for driving research in grounded language understanding and studying problems like generalization and sample efficiency. Several deep reinforcement learning (RL) methods with varying architectures and learning schemes have been proposed for TBGs. However, these methods fail to generalize efficiently, especially under distributional shifts. In a departure from deep RL approaches, in this paper, we propose a general method inspired by case-based reasoning to train agents and generalize out of the training distribution. The case-based reasoner collects instances of positive experiences from the agent's interaction with the world in the past and later reuses the collected experiences to act efficiently. The method can be applied in conjunction with any existing on-policy neural agent in the literature for TBGs. Our experiments show that the proposed approach consistently improves existing methods, obtains good out-of-distribution generalization, and achieves new state-of-the-art results on widely used environments. | ['Mrinmaya Sachan', 'Keerthiram Murugesan', 'Shehzaad Dhuliawala', 'Mattia Atzeni'] | 2021-10-16 | case-based-reasoning-for-better | https://openreview.net/forum?id=ZDaSIkWT-AP | https://openreview.net/pdf?id=ZDaSIkWT-AP | iclr-2022-4 | ['text-based-games'] | ['playing-games'] | [-1.70707196e-01 2.82590330e-01 -2.45681658e-01 -3.65002811e-01
-7.43470848e-01 -4.65110749e-01 8.97471845e-01 1.80670351e-01
-8.54097605e-01 1.01074123e+00 3.39030206e-01 -4.35478777e-01
-3.44637066e-01 -1.14380109e+00 -6.02559805e-01 -7.97124505e-01
-1.60468712e-01 1.03509390e+00 9.33591053e-02 -7.30715871e-01
3.85222882e-01 3.89377087e-01 -1.31260610e+00 2.81473458e-01
1.08615494e+00 7.68540919e-01 2.62757599e-01 5.50919354e-01
-2.61682421e-01 1.39064908e+00 -6.51176035e-01 -4.29597884e-01
3.08448821e-01 -5.60638726e-01 -1.03016412e+00 -1.87303677e-01
-6.37454838e-02 -5.98811805e-01 -6.95123315e-01 1.02886438e+00
5.98855793e-01 8.33440781e-01 5.10545790e-01 -1.32636321e+00
-9.83865559e-01 1.00035131e+00 -2.17935279e-01 2.52008528e-01
4.09742326e-01 2.40793824e-01 8.92413676e-01 -3.34654689e-01
6.21265888e-01 1.50299418e+00 3.77017677e-01 1.34026504e+00
-7.89003372e-01 -6.15122437e-01 6.24779403e-01 3.02749157e-01
-8.23179126e-01 -1.31440908e-01 5.69591820e-01 4.71949093e-02
1.45859861e+00 -1.91775799e-01 8.94777596e-01 1.40556121e+00
9.73689631e-02 1.50386226e+00 1.14725697e+00 -4.64247108e-01
1.04669273e+00 -1.86138928e-01 -1.71892747e-01 5.65392733e-01
1.18098639e-01 5.09298384e-01 -6.38622224e-01 -2.01167718e-01
5.59892178e-01 1.90275982e-02 6.02000356e-02 -4.05706137e-01
-8.39121103e-01 1.10649276e+00 6.60407603e-01 3.47047836e-01
-6.08997285e-01 4.65594918e-01 4.98101741e-01 3.40175182e-01
6.19707584e-01 6.68328464e-01 -3.69431645e-01 -3.20611000e-01
-7.55488575e-01 9.45757926e-01 8.56026590e-01 1.00824642e+00
5.92367589e-01 2.26494938e-01 -2.56631881e-01 6.82774901e-01
2.70366162e-01 4.53808248e-01 7.70207405e-01 -9.54061866e-01
5.65163434e-01 4.80754316e-01 1.74311250e-01 -3.60306352e-01
-5.43645024e-01 -2.92907387e-01 -4.44362670e-01 1.25946581e-01
2.05515444e-01 -3.98629159e-01 -9.48905170e-01 1.93387949e+00
1.45353615e-01 2.35902533e-01 8.02759349e-01 6.92563415e-01
6.42103195e-01 7.61841774e-01 5.34844577e-01 -8.28580633e-02
7.95879602e-01 -9.96232808e-01 -4.66453105e-01 -6.10527813e-01
7.81449556e-01 2.48279627e-02 1.15386879e+00 5.99167585e-01
-1.17555237e+00 -2.44063109e-01 -8.61528397e-01 6.06330521e-02
-5.48560798e-01 -3.58253211e-01 9.50334609e-01 5.34168601e-01
-1.40061963e+00 6.43900514e-01 -8.95273805e-01 -7.04377472e-01
6.93579972e-01 1.71873018e-01 1.52138993e-01 -2.32771724e-01
-1.35839546e+00 1.05699706e+00 8.55179727e-01 -1.13496199e-01
-1.39214265e+00 -4.75701064e-01 -8.84321988e-01 1.11119464e-01
7.05739379e-01 -7.95580268e-01 1.76258206e+00 -6.98644400e-01
-2.10101438e+00 5.71276546e-01 1.98404148e-01 -9.71713841e-01
5.42402089e-01 -3.60881418e-01 -2.27882206e-01 -1.07803494e-01
-2.69494514e-04 8.04069102e-01 5.06791770e-01 -1.18870425e+00
-9.67989087e-01 -4.44219202e-01 6.13794804e-01 5.99235117e-01
-6.59856722e-02 -4.93944913e-01 6.60065114e-02 -4.75052118e-01
-1.73402339e-01 -8.72214019e-01 -6.02463663e-01 -6.05942428e-01
-2.82016327e-03 -8.18253994e-01 3.64844531e-01 -1.12894878e-01
8.37746739e-01 -1.93293381e+00 1.37211666e-01 -1.29995003e-01
-5.22650182e-02 3.37273449e-01 -2.19032750e-01 8.27198386e-01
5.40767074e-01 -5.29626645e-02 -4.81226889e-04 -3.44345182e-01
4.87346679e-01 5.85071385e-01 -7.87889421e-01 6.21543154e-02
-3.24383602e-02 1.00729871e+00 -1.52961850e+00 -3.14453132e-02
2.23016143e-01 -1.47643745e-01 -9.38234925e-01 4.70435530e-01
-8.18102658e-01 3.05989742e-01 -7.47405469e-01 9.91398990e-02
4.38680142e-01 -3.05434112e-02 3.14902395e-01 5.79232752e-01
1.22039497e-01 4.89458799e-01 -7.72745132e-01 2.05035925e+00
-6.64029956e-01 2.23029017e-01 -4.98315930e-01 -1.20709944e+00
8.42978656e-01 2.34819323e-01 6.29307255e-02 -1.10780847e+00
7.82177076e-02 1.75169468e-01 1.27532974e-01 -5.82057953e-01
8.10316980e-01 -5.23516476e-01 -1.42365456e-01 5.41760921e-01
3.67915183e-01 -3.59469414e-01 3.69637907e-01 2.68601239e-01
1.01957870e+00 4.83073413e-01 4.86859053e-01 -3.40259612e-01
2.85732359e-01 7.54352063e-02 4.07801092e-01 1.28057182e+00
-1.81153178e-01 -1.91479027e-01 3.48806262e-01 -7.57784128e-01
-8.70178223e-01 -1.09986615e+00 5.66740453e-01 1.65505314e+00
2.92809039e-01 -2.79839367e-01 -6.76577628e-01 -7.36562908e-01
-2.18660027e-01 1.46501291e+00 -6.18727982e-01 -3.91813099e-01
-5.21673560e-01 -6.20089471e-01 7.06646919e-01 9.11697447e-01
9.72157240e-01 -1.67261088e+00 -9.24826801e-01 5.56679666e-01
-8.30398053e-02 -9.29555237e-01 1.92284033e-01 2.46336862e-01
-8.34633231e-01 -6.77735329e-01 -4.59063530e-01 -5.14998853e-01
1.99773401e-01 -1.02150209e-01 1.18149054e+00 -8.10343176e-02
2.43643180e-01 5.87596774e-01 -6.35015845e-01 -8.51396620e-01
-5.62061608e-01 2.10643202e-01 2.44389310e-01 -4.62997556e-01
4.71566707e-01 -3.79799485e-01 -5.30016720e-01 -2.01433465e-01
-1.03820693e+00 -1.07508518e-01 3.14334661e-01 9.97098923e-01
1.56367540e-01 1.63644224e-01 9.63502884e-01 -9.80766118e-01
1.25664330e+00 -5.34574687e-01 -6.36343300e-01 3.39928746e-01
-5.63095152e-01 3.69907647e-01 8.63979816e-01 -3.67670804e-01
-1.59539449e+00 -6.36902153e-01 -1.81305245e-01 -1.40669435e-01
-2.87626743e-01 5.87238848e-01 7.18318596e-02 4.45345789e-01
7.27741122e-01 5.05052865e-01 -2.69527555e-01 -6.48317561e-02
5.50153136e-01 4.91670430e-01 3.20054024e-01 -1.14707518e+00
2.38261089e-01 4.28306192e-01 -3.37578952e-01 -6.07406080e-01
-9.95409846e-01 -3.02891910e-01 -7.41136074e-02 -8.70435610e-02
7.77140558e-01 -7.57151484e-01 -9.67736900e-01 4.38336283e-01
-7.34538794e-01 -1.15659869e+00 -5.16271949e-01 5.76684654e-01
-1.17885065e+00 -1.36055321e-01 -6.95765197e-01 -1.30410123e+00
-2.21552670e-01 -1.01566756e+00 9.58584964e-01 6.79150701e-01
1.98874203e-03 -1.31671369e+00 3.26630712e-01 5.76817021e-02
5.59712648e-01 -1.38668180e-01 9.02981758e-01 -1.08287311e+00
-3.13677579e-01 2.55794913e-01 3.04230452e-01 2.56720856e-02
-7.73295835e-02 -5.31956792e-01 -1.00962949e+00 -5.06617129e-01
4.84830653e-03 -8.91304135e-01 8.86074901e-01 4.95338798e-01
1.10481822e+00 -4.78095978e-01 -2.49453843e-01 3.76810730e-01
1.29559147e+00 6.62203908e-01 7.36485302e-01 7.45053470e-01
1.72365412e-01 3.45243067e-01 6.86842024e-01 8.36925149e-01
4.82773542e-01 2.46261790e-01 4.14013952e-01 3.04980278e-01
3.87686104e-01 -6.15319371e-01 4.47731614e-01 2.37846017e-01
-1.97858855e-01 -6.88263118e-01 -7.55516171e-01 5.29816628e-01
-2.20079494e+00 -1.30439126e+00 9.15415645e-01 1.84713948e+00
7.06769705e-01 1.16870418e-01 2.65344650e-01 -2.57832438e-01
2.77161658e-01 1.84828237e-01 -9.42140043e-01 -6.83842957e-01
1.55879885e-01 5.25153995e-01 2.77252108e-01 4.76581633e-01
-8.68610620e-01 1.50525439e+00 6.33177519e+00 7.98285902e-01
-8.70948911e-01 -4.80440520e-02 5.82461357e-01 -8.70490372e-02
-2.63069838e-01 -2.58518130e-01 -7.67475665e-01 -7.75942132e-02
9.12374079e-01 -3.15243036e-01 8.77933025e-01 1.07591379e+00
-1.15250722e-01 -5.33580780e-02 -1.33706653e+00 7.68653333e-01
-5.05903959e-02 -1.32783043e+00 2.12239161e-01 -1.53824184e-02
8.75988185e-01 1.92649722e-01 2.93748498e-01 1.10904014e+00
1.11864448e+00 -9.70759869e-01 8.98548186e-01 2.79579014e-01
2.19618246e-01 -9.27232683e-01 7.72010267e-01 7.51898706e-01
-7.70229518e-01 -4.89984304e-01 -6.19782388e-01 -4.26790535e-01
-9.07724798e-02 -2.04526976e-01 -1.13401759e+00 6.66440487e-01
7.03439534e-01 6.51404738e-01 -1.32665977e-01 6.85714662e-01
-4.44703937e-01 5.72202504e-01 -6.80770352e-02 -7.51214802e-01
9.94658947e-01 -1.16313368e-01 2.98131078e-01 9.23476756e-01
2.78641433e-01 4.92070317e-01 4.96163875e-01 9.81366158e-01
-1.25004098e-01 -3.39570232e-02 -8.05265963e-01 -1.51305422e-01
5.02813697e-01 7.42375493e-01 -5.99280655e-01 -6.30610704e-01
-2.28712037e-01 7.26470649e-01 7.51083136e-01 5.67534924e-01
-6.59298062e-01 -1.16925552e-01 4.93677646e-01 -3.73655558e-01
4.66727287e-01 -7.88939819e-02 2.51432478e-01 -1.19191611e+00
-2.45926410e-01 -1.13697636e+00 8.27890098e-01 -6.99322581e-01
-1.65967536e+00 7.45958567e-01 3.59471500e-01 -9.37188029e-01
-8.93757939e-01 -8.23436201e-01 -4.34264034e-01 4.45482403e-01
-1.49136734e+00 -9.84372973e-01 4.64979410e-02 7.45582223e-01
8.78616273e-01 -5.77761769e-01 9.23334062e-01 -3.49247009e-01
-1.76706448e-01 3.93173128e-01 6.88270703e-02 6.13446012e-02
2.03210264e-01 -1.41079283e+00 5.34948528e-01 5.59407592e-01
1.85704857e-01 6.45602345e-01 7.20637918e-01 -5.62225342e-01
-1.40846276e+00 -9.21125054e-01 2.87844628e-01 -1.92295730e-01
7.00413823e-01 -2.19426781e-01 -6.81552231e-01 1.01481032e+00
4.84553754e-01 -1.76942497e-01 4.28389013e-01 2.36566454e-01
-2.92497814e-01 -3.36078480e-02 -1.29289472e+00 1.07748795e+00
1.26397812e+00 -2.81323701e-01 -1.04776430e+00 2.42813900e-01
5.32597423e-01 -7.09771276e-01 -3.89428258e-01 5.64481206e-02
1.89131543e-01 -1.05508769e+00 7.48226166e-01 -1.11913741e+00
3.51223886e-01 6.73068687e-03 -3.05448890e-01 -1.94331443e+00
-3.02501500e-01 -7.70670056e-01 4.35874686e-02 7.82346904e-01
1.99561790e-01 -9.69260871e-01 7.34313786e-01 6.71863914e-01
-3.34232837e-01 -7.98648655e-01 -9.33387578e-01 -8.95318985e-01
5.80834687e-01 -4.79602724e-01 8.56348395e-01 5.17670214e-01
3.76212686e-01 2.05190495e-01 -1.51406392e-01 5.37437834e-02
5.24635673e-01 1.89308181e-01 6.30353510e-01 -9.56489861e-01
-3.38403374e-01 -5.46989441e-01 -1.45867154e-01 -1.20883524e+00
6.29383862e-01 -9.91956770e-01 3.78656864e-01 -1.75149012e+00
9.48427841e-02 -5.43381095e-01 -3.78389359e-01 5.29716432e-01
-1.09269865e-01 -3.89703304e-01 3.12471956e-01 -2.51929253e-01
-9.21594739e-01 8.91186953e-01 1.25863707e+00 -2.73769826e-01
-3.43468547e-01 6.59035370e-02 -7.99866498e-01 8.67682040e-01
1.11086178e+00 -3.28013629e-01 -9.63587224e-01 -3.47279161e-01
4.37120706e-01 1.13245018e-01 9.63944942e-02 -1.07715571e+00
3.36844176e-01 -5.88783622e-01 3.32595021e-01 -3.35150301e-01
2.26808921e-01 -5.74078083e-01 -4.21006352e-01 5.79675615e-01
-8.71119201e-01 2.21571475e-01 3.32345724e-01 7.19534218e-01
8.22386704e-03 -4.21020269e-01 5.43006599e-01 -6.56560540e-01
-1.04539371e+00 3.92319679e-01 -5.53245246e-01 5.72854638e-01
8.22353840e-01 1.13965653e-01 -5.24063528e-01 -6.32606804e-01
-5.95904171e-01 4.06738251e-01 1.25983566e-01 4.46498692e-01
7.64213741e-01 -1.25782430e+00 -6.91628397e-01 -8.09586197e-02
1.98075056e-01 9.29456502e-02 3.22199106e-01 2.19268024e-01
-3.53023857e-01 4.74583328e-01 -3.81165415e-01 -2.74145454e-01
-4.22172576e-01 8.11008275e-01 5.22307217e-01 -6.62617266e-01
-6.53989673e-01 8.06759059e-01 2.89892316e-01 -6.96073711e-01
3.09846282e-01 -6.40816510e-01 -1.58091575e-01 -2.73697644e-01
6.80823445e-01 3.06839705e-01 -2.29334328e-02 1.39073143e-02
-1.38158634e-01 -9.57978293e-02 -3.27446461e-01 -5.31555593e-01
1.68564081e+00 1.36552244e-01 3.49700689e-01 6.12281680e-01
6.23735249e-01 -4.36887592e-01 -1.27965152e+00 -3.80924672e-01
6.17048331e-02 -2.71558493e-01 -2.37068921e-01 -9.48614657e-01
-8.69032681e-01 7.49921560e-01 2.41263151e-01 4.99905527e-01
9.28244293e-01 1.15042970e-01 5.22568703e-01 9.12058949e-01
9.45115387e-01 -1.43321383e+00 2.91179329e-01 8.69477510e-01
8.13660800e-01 -1.08007777e+00 -1.56616569e-01 3.50698292e-01
-1.07862091e+00 1.01757610e+00 8.59279156e-01 -4.20203447e-01
3.23532909e-01 8.32728222e-02 6.08111685e-03 -1.34056315e-01
-1.08062983e+00 -3.49772513e-01 -1.32620543e-01 9.75145698e-01
2.07310498e-01 2.28325605e-01 1.63913630e-02 4.55220640e-01
-4.26810861e-01 1.30951181e-01 4.80283499e-01 1.01907921e+00
-4.40151244e-01 -9.28452909e-01 1.00610822e-01 2.57206112e-01
-1.49086565e-01 -1.98260978e-01 -1.72251731e-01 1.03856254e+00
-2.26422250e-01 1.05999386e+00 5.78846224e-03 -1.52314320e-01
2.79180855e-01 1.06017649e-01 8.86001408e-01 -6.00439489e-01
-5.37808657e-01 -3.22466910e-01 1.26669735e-01 -7.10434735e-01
-5.48203230e-01 -5.97462654e-01 -1.69341958e+00 -2.10843295e-01
5.85157657e-04 2.32990503e-01 2.25444853e-01 1.32925963e+00
2.03617692e-01 4.64858800e-01 2.51950860e-01 -7.68354952e-01
-9.97720182e-01 -1.07341397e+00 -6.75896704e-01 4.78118330e-01
3.55238676e-01 -7.52296031e-01 -9.58976969e-02 -3.96685809e-01] | [3.891289472579956, 1.5166534185409546] |
0dc10e5b-e2bd-4142-93b1-78be27d961c9 | deep-contextual-attention-for-human-object | 1910.07721 | null | https://arxiv.org/abs/1910.07721v1 | https://arxiv.org/pdf/1910.07721v1.pdf | Deep Contextual Attention for Human-Object Interaction Detection | Human-object interaction detection is an important and relatively new class of visual relationship detection tasks, essential for deeper scene understanding. Most existing approaches decompose the problem into object localization and interaction recognition. Despite showing progress, these approaches only rely on the appearances of humans and objects and overlook the available context information, crucial for capturing subtle interactions between them. We propose a contextual attention framework for human-object interaction detection. Our approach leverages context by learning contextually-aware appearance features for human and object instances. The proposed attention module then adaptively selects relevant instance-centric context information to highlight image regions likely to contain human-object interactions. Experiments are performed on three benchmarks: V-COCO, HICO-DET and HCVRD. Our approach outperforms the state-of-the-art on all datasets. On the V-COCO dataset, our method achieves a relative gain of 4.4% in terms of role mean average precision ($mAP_{role}$), compared to the existing best approach. | ['Muhammad Haris Khan', 'Jorma Laaksonen', 'Fahad Shahbaz Khan', 'Rao Muhammad Anwer', 'Yanwei Pang', 'Tiancai Wang', 'Ling Shao'] | 2019-10-17 | deep-contextual-attention-for-human-object-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Wang_Deep_Contextual_Attention_for_Human-Object_Interaction_Detection_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Wang_Deep_Contextual_Attention_for_Human-Object_Interaction_Detection_ICCV_2019_paper.pdf | iccv-2019-10 | ['visual-relationship-detection'] | ['computer-vision'] | [ 2.67692983e-01 -1.84514940e-01 -1.92119583e-01 -2.02394709e-01
-3.46357703e-01 -3.80292118e-01 8.29206824e-01 4.69252944e-01
-4.16468501e-01 4.41898227e-01 1.68341994e-01 5.43981828e-02
8.60205591e-02 -2.94409424e-01 -6.26107335e-01 -5.05606592e-01
-1.44090012e-01 5.21311164e-01 7.65097678e-01 -6.62928075e-02
3.72394770e-01 5.01812577e-01 -1.99441803e+00 5.76271236e-01
4.49261785e-01 1.05637944e+00 3.08490366e-01 7.03354299e-01
2.33217418e-01 1.06309867e+00 -5.36395550e-01 -1.46541536e-01
8.97423401e-02 -2.14055493e-01 -8.40808988e-01 3.06473762e-01
7.51605332e-01 -2.25671291e-01 -3.31236452e-01 8.07243705e-01
3.75675589e-01 2.42868543e-01 5.65041661e-01 -1.47951198e+00
-7.60074317e-01 1.89541072e-01 -1.00375199e+00 6.80522740e-01
6.27659678e-01 3.99049580e-01 1.17077494e+00 -1.40561211e+00
7.55596936e-01 1.37491798e+00 1.42388776e-01 2.27641374e-01
-1.21821404e+00 -4.51996207e-01 6.14773512e-01 6.77646518e-01
-1.42442667e+00 -3.79471898e-01 6.49022639e-01 -6.30410910e-01
1.16066885e+00 5.24258137e-01 6.49574935e-01 9.46024120e-01
-2.37739488e-01 1.08417332e+00 8.30804527e-01 -6.05086565e-01
5.10131344e-02 2.74725437e-01 4.18303668e-01 6.25360012e-01
4.17722434e-01 -7.39617199e-02 -6.26737118e-01 -1.20028138e-01
6.05837703e-01 2.30701163e-01 -3.25929433e-01 -5.02797365e-01
-1.30948210e+00 2.92008489e-01 5.84674358e-01 3.31349403e-01
-3.86735827e-01 2.97406346e-01 2.01289892e-01 -1.55276477e-01
2.85405785e-01 3.30262035e-01 -2.36847535e-01 -3.12508713e-03
-3.98012847e-01 3.24372560e-01 3.69200736e-01 1.08417571e+00
4.19350088e-01 -5.13705432e-01 -6.95165217e-01 7.36493409e-01
9.03199017e-02 2.73294061e-01 -1.21986352e-01 -6.70965731e-01
4.17530447e-01 1.13412178e+00 5.07842660e-01 -1.22265375e+00
-1.89753398e-01 -2.83020616e-01 -4.55904156e-01 1.40948847e-01
3.71359408e-01 3.37668031e-01 -9.40749168e-01 1.32923925e+00
6.75026417e-01 1.59626871e-01 -4.43448573e-01 1.11478651e+00
1.03131604e+00 3.83081198e-01 3.27337652e-01 5.38795777e-02
1.62339044e+00 -1.41981578e+00 -4.95287031e-01 -5.56331158e-01
1.85532853e-01 -7.42751837e-01 1.22371328e+00 2.19092935e-01
-1.06750476e+00 -6.92409813e-01 -9.37781215e-01 -1.80141374e-01
-4.70706612e-01 3.63261729e-01 7.78418422e-01 1.79664835e-01
-7.21198618e-01 1.86147019e-01 -5.09149432e-01 -6.75495028e-01
7.66045451e-01 3.67607236e-01 -4.51107532e-01 -1.74360454e-01
-6.37289524e-01 5.85749686e-01 2.37526119e-01 9.79888588e-02
-1.17145133e+00 -5.61834753e-01 -6.25226021e-01 1.85259935e-02
7.79945850e-01 -6.20600581e-01 1.05638456e+00 -9.05088842e-01
-7.78783739e-01 1.07158720e+00 -2.98009157e-01 -2.05742121e-01
5.40815115e-01 -6.21512711e-01 -2.17115805e-01 2.96333134e-01
1.47201151e-01 7.50006974e-01 8.22962046e-01 -1.74279022e+00
-9.89445090e-01 -2.46818677e-01 3.81690949e-01 3.17038447e-01
-1.36534691e-01 5.35286009e-01 -1.06287682e+00 -4.50239092e-01
-4.14721072e-02 -9.14307833e-01 -1.14842601e-01 3.39331925e-01
-4.07416880e-01 -5.12165546e-01 1.15149486e+00 -2.99405932e-01
1.15238309e+00 -1.96882463e+00 2.70943511e-02 -8.68948400e-02
4.06418055e-01 3.81200790e-01 -1.01046897e-02 4.09083337e-01
-1.58887640e-01 -1.42409369e-01 -6.09198306e-03 -4.63480175e-01
-1.65514439e-01 -1.08120717e-01 -4.28558663e-02 5.90179682e-01
1.65643886e-01 1.13740265e+00 -1.09134459e+00 -6.63322866e-01
4.71632391e-01 5.71894467e-01 -3.23785603e-01 4.80776072e-01
-3.63808125e-01 5.36807835e-01 -2.28020400e-01 1.13753343e+00
5.05823493e-01 -5.78596234e-01 2.39432558e-01 -2.51741081e-01
1.34307861e-01 -1.39552161e-01 -1.13701510e+00 1.30957735e+00
-1.47596896e-01 7.15758383e-01 2.03598049e-02 -6.72960639e-01
5.03798246e-01 7.35767484e-02 4.01598513e-01 -5.96683502e-01
1.55191526e-01 -2.00873956e-01 1.13284819e-01 -5.52266419e-01
4.94413674e-01 7.35547185e-01 1.35481879e-01 1.08383931e-01
-6.70192912e-02 4.89610910e-01 1.76250741e-01 4.35239047e-01
1.19399643e+00 3.45611721e-02 6.17574215e-01 -2.13471264e-01
6.14346027e-01 -1.05684116e-01 2.74671018e-01 8.45423222e-01
-4.87750351e-01 5.74668169e-01 5.31766713e-01 -5.42775750e-01
-7.26844847e-01 -8.85453403e-01 1.30950123e-01 1.55728877e+00
6.89604461e-01 -5.77071786e-01 -4.53950316e-01 -1.02441430e+00
1.49763852e-01 5.36715329e-01 -1.10787034e+00 1.34522945e-01
-6.59140825e-01 -5.25021374e-01 -2.24600568e-01 8.32069337e-01
3.20589602e-01 -1.42598295e+00 -9.41888511e-01 -1.14451423e-01
-5.07405140e-02 -1.41229057e+00 -6.20988071e-01 -5.07369190e-02
-3.99548441e-01 -1.43729448e+00 -5.57184279e-01 -7.36268997e-01
8.90427589e-01 6.82799399e-01 1.43501294e+00 4.06306565e-01
-9.69313323e-01 7.20431983e-01 -3.10598046e-01 -4.54445004e-01
2.10351542e-01 -2.73914874e-01 -1.95386544e-01 4.05037075e-01
4.22771841e-01 -4.18265373e-01 -1.00792336e+00 6.54308736e-01
-4.95721489e-01 3.89459692e-02 6.81346416e-01 8.42783153e-01
6.88804805e-01 -2.20281586e-01 3.94551456e-02 -9.73659337e-01
1.70204565e-01 -4.37730014e-01 -5.33066750e-01 4.65669632e-01
-4.09303159e-01 -3.55951935e-01 1.05880864e-01 -4.64570463e-01
-1.01701391e+00 2.47057602e-01 6.90696836e-01 -6.21098816e-01
-4.66047943e-01 -1.58646539e-01 -4.44280207e-01 -7.73790851e-02
5.91684699e-01 -3.86071950e-02 -8.18342984e-01 -3.16831291e-01
3.88676733e-01 4.04203475e-01 6.43510222e-01 -5.68620861e-01
5.67211211e-01 7.56737232e-01 -1.30365878e-01 -8.24180663e-01
-8.37348580e-01 -1.04176033e+00 -8.00546825e-01 -4.46693629e-01
9.97717798e-01 -8.26888978e-01 -9.60073888e-01 1.98778644e-01
-1.32945240e+00 -2.69117445e-01 -6.57215789e-02 4.61994112e-02
-1.83213770e-01 1.53796941e-01 -2.16652006e-01 -1.15356076e+00
-2.07861617e-01 -1.01970994e+00 1.50454783e+00 2.16244191e-01
-1.58759668e-01 -5.97769201e-01 -2.67156690e-01 5.34641623e-01
1.43083200e-01 4.30591553e-01 6.28323197e-01 -4.06605154e-01
-9.50312316e-01 -3.29034299e-01 -6.85106814e-01 -2.14756921e-01
6.52640536e-02 -2.89267898e-02 -1.17654848e+00 -1.81476057e-01
-6.28779888e-01 -2.24877506e-01 9.50891912e-01 1.94151908e-01
1.41037321e+00 -7.99309462e-02 -9.73548234e-01 8.03355649e-02
1.25262034e+00 2.46709660e-01 5.50091803e-01 6.70194104e-02
1.01514864e+00 6.32392228e-01 9.45541024e-01 4.34524536e-01
3.74878079e-01 1.12024462e+00 7.02708066e-01 -3.25713784e-01
-4.12280053e-01 2.02160850e-02 8.76941606e-02 -5.72670847e-02
-4.67629611e-01 -2.88575351e-01 -1.13605547e+00 9.36688542e-01
-2.30503464e+00 -9.53592479e-01 -2.66475648e-01 2.08764124e+00
3.82304817e-01 2.67607898e-01 3.85813028e-01 1.91872101e-02
8.85351777e-01 1.38316229e-01 -4.08091009e-01 1.31258771e-01
1.16776727e-01 2.22696271e-02 1.20264634e-01 2.12586477e-01
-1.56720042e+00 9.35870767e-01 6.02303410e+00 5.12520552e-01
-6.01179600e-01 1.91982284e-01 5.49339414e-01 -3.31420690e-01
4.00496393e-01 -1.13928609e-01 -8.88883233e-01 3.24965477e-01
3.10696095e-01 2.02722177e-01 2.02888414e-01 1.08398211e+00
-5.36865704e-02 -6.60055161e-01 -1.54562819e+00 1.18192947e+00
2.68081009e-01 -1.00938976e+00 -1.94397300e-01 -4.04062495e-02
7.03913808e-01 -3.71121109e-01 1.15768053e-01 1.53524488e-01
2.35030338e-01 -1.16376626e+00 7.79682517e-01 5.11496842e-01
6.10002577e-01 -6.39374673e-01 7.36446381e-01 -6.57975441e-03
-1.69398546e+00 -2.46279910e-01 -9.26271304e-02 -1.03524990e-01
-2.55566016e-02 4.25655037e-01 -9.38727558e-01 3.43807191e-01
1.10037088e+00 8.68263066e-01 -9.58963454e-01 1.01637375e+00
-3.03406239e-01 3.79743457e-01 -5.24479561e-02 -1.20465772e-03
-4.62615974e-02 2.98853606e-01 6.36518240e-01 1.39170694e+00
-3.48836899e-01 2.74388641e-01 5.70871174e-01 7.44416833e-01
-1.04766682e-01 4.12949100e-02 -5.29009879e-01 1.08407967e-01
4.78206247e-01 1.44678354e+00 -1.03199065e+00 -4.74252850e-01
-5.49029231e-01 1.26401448e+00 5.22084534e-01 2.63156801e-01
-1.04658341e+00 -2.32817188e-01 5.66819847e-01 4.42937523e-01
6.46789432e-01 -8.92554224e-02 -4.38033678e-02 -8.87179852e-01
3.28673691e-01 -8.93549800e-01 5.96276700e-01 -7.77930856e-01
-1.17070544e+00 5.38820565e-01 2.76999146e-01 -1.15405011e+00
1.68217853e-01 -6.21907353e-01 -5.73720098e-01 5.49154997e-01
-1.11336827e+00 -1.59094691e+00 -9.37700450e-01 5.59460342e-01
7.87388086e-01 6.10636435e-02 5.73027134e-01 3.30227762e-01
-6.01466358e-01 6.21383011e-01 -5.17339945e-01 1.53404966e-01
6.32429659e-01 -1.27074456e+00 3.37149143e-01 8.38239431e-01
3.02732170e-01 6.55480564e-01 6.46609366e-01 -5.96556246e-01
-1.29440069e+00 -1.11099267e+00 7.20298707e-01 -9.84953225e-01
4.17288899e-01 -7.63332367e-01 -7.73540854e-01 5.69049120e-01
3.92306387e-01 6.86934710e-01 3.75896662e-01 2.38366470e-01
-5.53643465e-01 -5.49713057e-03 -9.41127837e-01 7.80199885e-01
1.66332614e+00 -4.92325634e-01 -4.70644146e-01 3.75645280e-01
6.67132258e-01 -3.32755059e-01 -4.59093779e-01 4.75849241e-01
6.91920400e-01 -1.07682133e+00 1.28560829e+00 -5.87082267e-01
3.00377369e-01 -6.02284014e-01 -2.71670848e-01 -6.54180765e-01
-4.27932113e-01 -4.92364734e-01 -6.13833010e-01 1.13312161e+00
1.18351720e-01 -9.69115719e-02 6.54968262e-01 5.17548800e-01
1.64922148e-01 -8.90356302e-01 -5.20651698e-01 -6.16995871e-01
-6.97964728e-01 -2.56029099e-01 3.25713843e-01 8.26565862e-01
-1.84646353e-01 5.53431094e-01 -3.79272103e-01 2.67749131e-01
6.63296163e-01 3.47389191e-01 1.16300702e+00 -1.16644406e+00
-5.38655758e-01 -4.52753514e-01 -6.31927490e-01 -7.88836360e-01
-1.56050473e-01 -5.40596724e-01 6.98769540e-02 -1.57740319e+00
7.50114083e-01 -1.89292654e-01 -5.27088463e-01 6.56892002e-01
-6.52565062e-01 4.41018373e-01 2.94758499e-01 1.93239331e-01
-1.34858298e+00 2.35666633e-01 1.05907488e+00 -3.17096591e-01
-2.37831950e-01 -3.35102677e-01 -5.47387838e-01 8.04327846e-01
4.15536702e-01 -3.79908293e-01 -3.91501099e-01 -1.71335503e-01
-2.24033922e-01 -3.78521085e-01 9.61031616e-01 -1.21923041e+00
1.90332741e-01 -3.66554052e-01 6.28628373e-01 -9.20944095e-01
5.13942838e-01 -8.48235786e-01 3.43205594e-02 2.60092944e-01
-2.85878003e-01 -1.26251578e-02 2.41111830e-01 8.72454941e-01
1.22802518e-02 2.73470879e-01 7.21248448e-01 -1.25811636e-01
-1.08099806e+00 2.04325721e-01 -3.26879509e-02 1.39067713e-02
1.50749338e+00 -2.15705559e-02 -3.45074207e-01 -2.38625914e-01
-6.45440578e-01 2.42304578e-01 3.82389903e-01 6.10962152e-01
7.00778365e-01 -1.22195005e+00 -5.56061447e-01 -1.29411854e-02
7.85890460e-01 -3.40170935e-02 2.15964824e-01 8.17000926e-01
-9.44623277e-02 4.65897113e-01 1.31605510e-02 -9.53811109e-01
-1.86075199e+00 8.25740695e-01 2.07787365e-01 -2.02044025e-01
-6.95290506e-01 1.09990430e+00 7.18986928e-01 1.65805966e-01
5.84187388e-01 -3.25292289e-01 -2.36724049e-01 -1.22592002e-01
7.40770996e-01 3.74220103e-01 -5.47952689e-02 -8.18574607e-01
-7.51933396e-01 5.30779421e-01 -2.54439890e-01 1.29027098e-01
1.10739529e+00 -1.83296611e-03 -2.02039089e-02 3.67416650e-01
9.66375589e-01 -3.21175233e-02 -1.28093481e+00 -3.03279012e-01
1.56584740e-01 -8.95826280e-01 -1.92869782e-01 -1.08708167e+00
-9.54733908e-01 9.03069735e-01 8.13042223e-01 7.95254484e-03
1.12530291e+00 6.03368104e-01 2.59173274e-01 2.57713109e-01
5.00260532e-01 -7.78701782e-01 6.65339053e-01 1.52288079e-01
1.27829182e+00 -1.54551661e+00 3.61760527e-01 -7.49065399e-01
-7.35058010e-01 5.28824985e-01 1.19013035e+00 -8.82131606e-02
5.22956610e-01 2.25133207e-02 -2.47040719e-01 -5.57475150e-01
-8.68367791e-01 -5.85608065e-01 7.38395751e-01 6.14147842e-01
4.79634762e-01 2.82064192e-02 -9.72251743e-02 4.75012690e-01
6.14884913e-01 -2.98825055e-01 -9.61906388e-02 1.03558779e+00
-3.52834702e-01 -9.19795871e-01 -5.06056249e-01 3.15737695e-01
-3.32326561e-01 -5.81570417e-02 -8.52507830e-01 9.45672512e-01
3.73040766e-01 1.05697215e+00 1.60996914e-01 -2.18483254e-01
5.43464839e-01 -2.15209901e-01 5.85982382e-01 -7.14336038e-01
-6.84229910e-01 1.72187667e-02 1.18342377e-01 -9.77872014e-01
-5.25346935e-01 -6.40100598e-01 -1.07332742e+00 6.16149716e-02
-2.87326127e-01 -3.41109335e-01 1.52669564e-01 5.86729527e-01
5.48862219e-01 6.63346827e-01 3.21518898e-01 -1.14475429e+00
2.92063445e-01 -7.46740162e-01 -2.01315209e-01 7.96469152e-01
2.69047081e-01 -1.23653281e+00 -4.63509858e-02 1.08788110e-01] | [9.806012153625488, 1.438041090965271] |
168027f6-6612-489c-a282-bfcb9faba14c | my-boli-code-mixed-marathi-english-corpora | 2306.14030 | null | https://arxiv.org/abs/2306.14030v1 | https://arxiv.org/pdf/2306.14030v1.pdf | My Boli: Code-mixed Marathi-English Corpora, Pretrained Language Models and Evaluation Benchmarks | The research on code-mixed data is limited due to the unavailability of dedicated code-mixed datasets and pre-trained language models. In this work, we focus on the low-resource Indian language Marathi which lacks any prior work in code-mixing. We present L3Cube-MeCorpus, a large code-mixed Marathi-English (Mr-En) corpus with 5 million tweets for pretraining. We also release L3Cube-MeBERT and MeRoBERTa, code-mixed BERT-based transformer models pre-trained on MeCorpus. Furthermore, for benchmarking, we present three supervised datasets MeHate, MeSent, and MeLID for downstream tasks like code-mixed Mr-En hate speech detection, sentiment analysis, and language identification respectively. These evaluation datasets individually consist of manually annotated \url{~}12,000 Marathi-English code-mixed tweets. Ablations show that the models trained on this novel corpus significantly outperform the existing state-of-the-art BERT models. This is the first work that presents artifacts for code-mixed Marathi research. All datasets and models are publicly released at https://github.com/l3cube-pune/MarathiNLP . | ['Raviraj Joshi', 'Shantanu Patankar', 'Aditya Kane', 'Omkar Gokhale', 'Tanmay Chavan'] | 2023-06-24 | null | null | null | null | ['language-identification', 'benchmarking', 'hate-speech-detection', 'sentiment-analysis', 'language-identification', 'benchmarking'] | ['audio', 'miscellaneous', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'robots'] | [-2.89751410e-01 -1.83112681e-01 -3.09558250e-02 -1.58279076e-01
-1.52377510e+00 -6.79254651e-01 4.96053040e-01 -1.53371707e-01
-4.52610552e-01 4.36705798e-01 4.43857163e-02 -7.56718874e-01
3.05081367e-01 -2.01100931e-01 -7.41735935e-01 -3.87460470e-01
-4.41186838e-02 5.73264122e-01 5.22242440e-03 -6.05617523e-01
2.21888855e-01 -1.61138013e-01 -9.68137980e-01 8.01065922e-01
9.87786889e-01 3.68373990e-01 4.91103321e-01 1.14684606e+00
1.19360030e-01 1.60441625e+00 -4.21461165e-01 -8.86558115e-01
2.58609708e-02 -5.01665950e-01 -9.06406760e-01 -5.14466286e-01
3.38100463e-01 1.20834941e-02 -4.25704747e-01 1.01671851e+00
3.01155448e-01 -3.41151923e-01 4.62046921e-01 -1.54922891e+00
-8.03209424e-01 1.21093702e+00 -5.42017102e-01 4.10652012e-01
1.57445252e-01 1.86244413e-01 1.00996184e+00 -9.95082021e-01
7.88325667e-01 9.47706521e-01 1.07864535e+00 6.01764739e-01
-9.80018198e-01 -1.03470361e+00 -3.97311956e-01 2.57907599e-01
-1.40894532e+00 -4.54092830e-01 5.39127588e-01 -8.65145504e-01
1.32180130e+00 1.40973359e-01 3.20435554e-01 1.45621109e+00
1.35867536e-01 1.07161081e+00 1.10790801e+00 -3.67384255e-01
-3.33639294e-01 5.82831204e-01 8.19182768e-02 8.72904181e-01
-4.09353167e-01 -1.17365636e-01 -5.26238263e-01 -3.77533622e-02
-1.50812864e-01 -3.66721928e-01 -1.40864134e-01 2.30630174e-01
-1.35271299e+00 1.06977487e+00 4.53706048e-02 4.77538943e-01
2.04163969e-01 2.74587959e-01 8.59318197e-01 5.26375473e-01
7.71011591e-01 2.40432128e-01 -5.45486510e-01 -1.03239870e+00
-1.13316524e+00 1.78946659e-01 7.85227716e-01 1.51416254e+00
5.71250260e-01 2.62785524e-01 1.04813978e-01 1.21492505e+00
2.39430040e-01 6.96600914e-01 4.72515434e-01 -5.91621637e-01
1.04758108e+00 2.76776671e-01 -4.25203025e-01 -4.58085626e-01
-3.11354369e-01 -1.69152319e-01 -3.38423252e-01 -4.23975997e-02
3.46262306e-01 -3.86498690e-01 -6.44577742e-01 1.41784060e+00
-2.47553557e-01 -1.72516063e-01 2.29163572e-01 4.34245676e-01
1.18405688e+00 8.71329963e-01 -2.79003549e-02 5.29772580e-01
1.34675634e+00 -1.58591604e+00 -3.53266537e-01 -4.63105947e-01
1.11127412e+00 -1.26987422e+00 1.30496848e+00 3.48807544e-01
-7.53647506e-01 -2.38770962e-01 -7.39417553e-01 -1.88737378e-01
-4.34911996e-01 5.23742914e-01 4.17999148e-01 9.89297628e-01
-1.02361310e+00 9.39159840e-02 -6.02924466e-01 -4.66455638e-01
4.62893546e-01 -4.50503603e-02 -4.48652774e-01 -2.49441396e-02
-1.00498486e+00 9.07326996e-01 7.21687302e-02 -1.96686104e-01
-1.47162426e+00 -8.14243317e-01 -9.10639763e-01 -4.11604673e-01
4.37262654e-02 2.89496660e-01 1.64879024e+00 -9.24091339e-01
-1.18018091e+00 1.32389390e+00 6.55721202e-02 -6.51186407e-01
5.27513325e-01 -2.54021078e-01 -6.67285860e-01 -3.61104719e-02
3.45528215e-01 5.60552120e-01 5.66844702e-01 -1.03062570e+00
-5.28082252e-01 2.54826814e-01 3.95450406e-02 -2.51338065e-01
-2.63767898e-01 7.35584855e-01 -3.05045277e-01 -6.75098896e-01
-8.54066312e-01 -1.28854930e+00 2.08940089e-01 -6.37216628e-01
-5.48678458e-01 8.54069665e-02 7.80118048e-01 -1.05376387e+00
1.37757003e+00 -2.55019784e+00 -1.15690522e-01 -2.84343809e-01
1.98093861e-01 3.84013563e-01 -6.61525607e-01 8.66663396e-01
-1.68747902e-01 2.78691113e-01 -3.40943843e-01 -7.82799900e-01
-4.16528396e-02 -8.41577444e-03 -3.73731315e-01 6.77430868e-01
2.09815189e-01 7.95580685e-01 -9.01903212e-01 -4.50479358e-01
-4.81913202e-02 4.93900865e-01 -9.96733546e-01 1.61428392e-01
-1.11613713e-01 3.85047853e-01 1.62426364e-02 7.83793449e-01
5.20317793e-01 2.03894079e-01 -3.52141410e-01 3.38970482e-01
-5.03094792e-01 6.13388538e-01 -2.20512733e-01 1.86545503e+00
-1.03558135e+00 1.15392423e+00 3.09085906e-01 -7.47869372e-01
7.59785056e-01 2.27302924e-01 1.17347524e-01 -5.51652253e-01
3.94222558e-01 5.57338297e-01 3.70249711e-02 -7.33296990e-01
8.26530755e-01 -7.36555876e-03 -5.32593846e-01 4.15959448e-01
2.96722323e-01 -3.20599407e-01 5.33805430e-01 5.20898461e-01
1.35661983e+00 1.02940090e-01 -1.76303133e-01 -3.28967541e-01
6.51122749e-01 4.31763589e-01 3.64818275e-01 5.90139210e-01
-2.79811472e-01 8.25672865e-01 7.41977692e-01 7.35482052e-02
-1.18917680e+00 -7.50028431e-01 -3.66424322e-01 1.43362427e+00
-5.11433840e-01 -7.69384921e-01 -9.55649555e-01 -8.09786677e-01
-2.95673281e-01 1.09592569e+00 -5.90018392e-01 1.18743747e-01
-5.92604756e-01 -8.15851808e-01 1.31416142e+00 1.35230035e-01
3.86090636e-01 -9.45970118e-01 -1.16533421e-01 1.39489293e-01
-6.61393702e-01 -1.14815772e+00 -7.78035998e-01 3.20072114e-01
-1.44191340e-01 -1.06229544e+00 -4.34517771e-01 -1.01681221e+00
1.75849825e-01 2.82146689e-02 1.34240246e+00 1.08661316e-02
-2.77516544e-01 2.51118779e-01 -8.97984266e-01 -4.09255177e-01
-1.22545874e+00 4.39055949e-01 -3.85672480e-01 -2.78723210e-01
5.66467404e-01 -3.43903214e-01 1.10496335e-01 1.02963947e-01
-5.38508296e-01 3.39313418e-01 4.26770866e-01 7.46661127e-01
-2.24181846e-01 -4.56665456e-01 4.07246232e-01 -1.22605932e+00
3.63277197e-01 -1.06921434e+00 -4.05019164e-01 8.31868798e-02
-6.48692250e-02 -2.54470199e-01 6.59696698e-01 -5.43307722e-01
-1.01718974e+00 -2.00583458e-01 -6.80773973e-01 -1.77869871e-01
-5.64243831e-02 6.27945542e-01 3.26934785e-01 -6.50867894e-02
8.74428093e-01 2.57723719e-01 -1.63568482e-01 -5.42324483e-01
4.16896313e-01 1.43126917e+00 5.67924678e-01 -6.37853026e-01
7.94231951e-01 1.89041868e-01 -8.88231635e-01 -1.04808915e+00
-5.66373169e-01 -5.92768073e-01 -5.64511955e-01 -2.33900726e-01
1.07158744e+00 -1.39570928e+00 -2.35450670e-01 8.33594024e-01
-1.29267812e+00 -8.29828799e-01 9.99919847e-02 4.41691041e-01
-4.65732783e-01 3.33129048e-01 -1.16523683e+00 -6.78040266e-01
-3.68070811e-01 -1.43670785e+00 9.56886590e-01 -5.12910843e-01
-3.11718881e-01 -8.80242944e-01 5.30358791e-01 8.25268149e-01
4.58261490e-01 -3.99898663e-02 7.95252860e-01 -9.14293110e-01
-1.87902391e-01 -1.58233121e-01 -2.30298504e-01 5.16650021e-01
-2.88446337e-01 5.13619721e-01 -1.00577569e+00 -3.75279903e-01
-1.57256767e-01 -7.97746837e-01 6.13944590e-01 -1.73953071e-01
6.07120454e-01 -4.06257868e-01 4.18934263e-02 6.60210669e-01
1.14424896e+00 7.54926428e-02 5.86879373e-01 3.99912179e-01
8.12939107e-01 3.28558236e-01 4.48080063e-01 5.97255230e-01
8.44151378e-01 2.86208749e-01 4.15169388e-01 8.19069222e-02
-3.36034536e-01 -2.10725039e-01 9.55107570e-01 1.75319386e+00
2.36855716e-01 -1.88034415e-01 -1.52300811e+00 1.17153168e+00
-1.42460370e+00 -9.57966566e-01 -5.87306261e-01 1.69665468e+00
1.06597960e+00 -4.93627377e-02 2.98602402e-01 -2.37719581e-01
6.45556450e-01 -2.57517733e-02 1.85873151e-01 -5.23381591e-01
-2.01089662e-02 8.71214923e-03 3.66500467e-01 5.26964247e-01
-1.06390953e+00 1.07299304e+00 5.37140608e+00 1.25698888e+00
-1.00937366e+00 8.51445615e-01 3.61145228e-01 -2.92559326e-01
-2.66525388e-01 2.54343241e-01 -8.23924303e-01 6.85772955e-01
1.49411309e+00 1.28292805e-02 8.46751213e-01 1.12991166e+00
-1.16553299e-01 3.39568183e-02 -9.38187659e-01 1.00155711e+00
3.38955492e-01 -1.13434470e+00 -6.13123477e-01 -1.17921568e-01
7.49271393e-01 1.18690860e+00 1.84879571e-01 1.10167706e+00
7.00452209e-01 -8.31733167e-01 1.48194969e+00 -1.23872638e-01
1.00493944e+00 -7.52774656e-01 7.52561629e-01 4.19392377e-01
-1.04625869e+00 -2.46315762e-01 -2.40277886e-01 1.19192794e-01
-1.11748964e-01 4.32957262e-01 -6.71624124e-01 2.00406671e-01
8.44997704e-01 9.79667604e-01 -1.03868949e+00 7.44378984e-01
-1.26261890e-01 1.24791074e+00 -2.15500861e-01 -1.72115207e-01
5.44897318e-01 -3.01879942e-02 5.61055899e-01 1.99541724e+00
4.19477731e-01 -4.57244635e-01 -6.74415752e-02 9.65181530e-01
-1.98819607e-01 2.81129599e-01 -6.39443338e-01 -4.52920824e-01
3.78131926e-01 1.56107736e+00 -4.77168947e-01 -2.59318680e-01
-1.08671701e+00 7.67303169e-01 5.37613630e-01 8.91537443e-02
-1.30935812e+00 -6.62357271e-01 4.86794621e-01 4.21379171e-02
1.11051179e-01 -2.60528147e-01 4.80788052e-02 -1.40582287e+00
-3.13445300e-01 -1.39951503e+00 3.48237842e-01 -7.95892239e-01
-1.28225172e+00 1.12016249e+00 -1.30699612e-02 -1.38621676e+00
-1.09926984e-01 -5.56331336e-01 -6.93818331e-01 6.67193294e-01
-1.37264800e+00 -1.59458447e+00 -1.15148485e-01 5.68393707e-01
8.54847312e-01 -7.01565802e-01 5.48394203e-01 8.98033798e-01
-7.89407313e-01 7.56457686e-01 4.42760706e-01 6.42177880e-01
8.00771713e-01 -1.09134328e+00 7.28867173e-01 9.70553875e-01
1.44378096e-01 4.09253925e-01 5.78241944e-01 -6.52285039e-01
-1.30941916e+00 -1.25787699e+00 1.01852167e+00 -9.16242778e-01
1.41543293e+00 -1.08989573e+00 -6.40605807e-01 8.30646336e-01
6.35379255e-01 -1.97110012e-01 7.56900489e-01 3.37513275e-02
-7.06122220e-01 2.95328110e-01 -1.03421509e+00 4.62590545e-01
7.04114556e-01 -1.05313277e+00 -4.11695600e-01 4.51862425e-01
6.55345917e-01 -3.61025423e-01 -8.57447326e-01 -1.28518552e-01
1.22886598e-01 -9.13052976e-01 3.96608323e-01 -2.34684855e-01
8.98931742e-01 -7.38841482e-03 -2.81711310e-01 -1.29364669e+00
1.14506772e-02 -7.82749593e-01 5.65780878e-01 1.66874254e+00
7.97882557e-01 -2.96513557e-01 2.57048339e-01 2.29548782e-01
-4.75540251e-01 -2.79605538e-01 -1.02478361e+00 -7.40534961e-01
6.71534419e-01 -9.56105113e-01 1.92915946e-01 1.23229575e+00
3.80231947e-01 2.40816131e-01 -4.84784126e-01 -2.29787827e-01
3.29638302e-01 -2.12616920e-01 9.62099195e-01 -5.19264758e-01
-4.28363591e-01 -3.80798012e-01 -1.13665968e-01 -6.15390182e-01
4.89288568e-01 -1.64644623e+00 3.30969900e-01 -1.18576682e+00
4.86313969e-01 -6.33558512e-01 3.59758049e-01 8.00295115e-01
2.17020199e-01 6.40200377e-01 3.15592259e-01 1.78932637e-01
-5.38923979e-01 4.74938244e-01 6.19865835e-01 -2.90248215e-01
1.60779104e-01 -1.96435928e-01 -3.55514586e-01 5.20821691e-01
1.05310047e+00 -1.04291356e+00 -2.23983556e-01 -5.75431287e-01
5.52581906e-01 -1.46922972e-02 2.11813837e-01 -1.07172620e+00
-1.62116781e-01 2.60759652e-01 -4.74846631e-01 -5.96722305e-01
3.47772352e-02 -3.72597337e-01 7.58444071e-02 3.85608703e-01
-3.72120678e-01 2.59492248e-01 5.22359490e-01 1.32411895e-02
-1.90208465e-01 -7.60721803e-01 9.75661695e-01 -2.11669892e-01
-4.52770174e-01 1.88423432e-02 -1.22308397e+00 8.42452645e-01
7.94693112e-01 3.48131329e-01 -7.07229733e-01 -2.83752799e-01
-1.06968433e-01 6.26500770e-02 4.53062505e-01 8.16308439e-01
2.37371907e-01 -1.03703868e+00 -1.24176848e+00 6.99362233e-02
5.04297197e-01 -6.15732670e-01 2.14642912e-01 1.03207564e+00
-1.04340684e+00 2.38618627e-01 -1.53989777e-01 -3.61791432e-01
-1.32822418e+00 6.11871839e-01 5.73462248e-02 -1.78187862e-01
-4.19088930e-01 1.05758739e+00 -1.93839505e-01 -1.20378911e+00
-1.47225767e-01 -2.24602059e-01 4.32599075e-02 3.91816422e-02
3.76673281e-01 3.51158947e-01 5.15003987e-02 -1.14342153e+00
-4.30580199e-01 1.70114875e-01 -1.61401525e-01 -1.71545982e-01
1.40606141e+00 1.93550382e-04 -5.40664554e-01 4.59865898e-01
1.63109076e+00 5.11810899e-01 -7.02108800e-01 6.83785230e-03
-1.91528630e-02 -4.62806553e-01 -1.31202146e-01 -6.97992861e-01
-1.06480563e+00 1.09789550e+00 3.20157766e-01 -1.50801823e-01
6.77382588e-01 1.12743549e-01 7.45856225e-01 3.01983833e-01
4.11741704e-01 -1.05643940e+00 -2.92348079e-02 9.37212765e-01
8.84640396e-01 -1.08296430e+00 -4.17872429e-01 -1.06461473e-01
-1.05187178e+00 9.22210217e-01 7.26100206e-01 1.67785212e-01
7.18141735e-01 7.41526246e-01 4.16464061e-01 -1.34864464e-01
-8.58362615e-01 -1.48720458e-01 -9.11825374e-02 5.85005522e-01
8.78864944e-01 -7.54238293e-02 -5.35404086e-02 7.54525244e-01
-7.36814559e-01 -3.18762839e-01 1.07282960e+00 9.01882172e-01
-1.82243697e-02 -1.02560174e+00 -2.58326054e-01 3.95131886e-01
-8.11728120e-01 -7.63751686e-01 -2.28468224e-01 8.89364362e-01
1.95305385e-02 1.11988652e+00 -1.73924133e-01 -6.43552840e-01
-1.54787060e-02 3.96854663e-03 9.80282649e-02 -7.10325181e-01
-1.18584132e+00 -1.03983015e-01 4.20985550e-01 -1.72383666e-01
-1.19792603e-01 -6.57392323e-01 -1.12850964e+00 -7.72697031e-01
-3.11862409e-01 4.36079830e-01 7.83541143e-01 4.74565506e-01
1.72947943e-01 4.99109805e-01 2.46860266e-01 -6.95606172e-01
3.56774102e-03 -1.18016863e+00 -4.01999474e-01 3.97685051e-01
4.22469854e-01 -1.17652029e-01 -5.37273228e-01 1.19821988e-01] | [9.217996597290039, 10.498470306396484] |
b357ebbf-6561-4784-8d0c-fbdfcf85bfcc | dit-self-supervised-pre-training-for-document | 2203.02378 | null | https://arxiv.org/abs/2203.02378v3 | https://arxiv.org/pdf/2203.02378v3.pdf | DiT: Self-supervised Pre-training for Document Image Transformer | Image Transformer has recently achieved significant progress for natural image understanding, either using supervised (ViT, DeiT, etc.) or self-supervised (BEiT, MAE, etc.) pre-training techniques. In this paper, we propose \textbf{DiT}, a self-supervised pre-trained \textbf{D}ocument \textbf{I}mage \textbf{T}ransformer model using large-scale unlabeled text images for Document AI tasks, which is essential since no supervised counterparts ever exist due to the lack of human-labeled document images. We leverage DiT as the backbone network in a variety of vision-based Document AI tasks, including document image classification, document layout analysis, table detection as well as text detection for OCR. Experiment results have illustrated that the self-supervised pre-trained DiT model achieves new state-of-the-art results on these downstream tasks, e.g. document image classification (91.11 $\rightarrow$ 92.69), document layout analysis (91.0 $\rightarrow$ 94.9), table detection (94.23 $\rightarrow$ 96.55) and text detection for OCR (93.07 $\rightarrow$ 94.29). The code and pre-trained models are publicly available at \url{https://aka.ms/msdit}. | ['Furu Wei', 'Cha Zhang', 'Lei Cui', 'Tengchao Lv', 'Yiheng Xu', 'Junlong Li'] | 2022-03-04 | null | null | null | null | ['document-image-classification', 'document-layout-analysis', 'table-detection', 'document-ai'] | ['computer-vision', 'computer-vision', 'miscellaneous', 'natural-language-processing'] | [ 5.51623046e-01 -2.84260392e-01 -9.98851061e-02 -3.98332119e-01
-8.04125428e-01 -9.56221700e-01 7.65132129e-01 1.84317362e-02
-1.98844865e-01 2.69516200e-01 -1.23936355e-01 -4.99901861e-01
1.27348602e-01 -4.41417187e-01 -7.52485156e-01 -6.83707356e-01
1.89644024e-01 3.70105624e-01 -5.67177795e-02 -7.98858851e-02
7.18917966e-01 5.10761976e-01 -1.30625319e+00 9.05269980e-01
8.55873168e-01 1.16674578e+00 2.29666054e-01 1.29595411e+00
-1.78214759e-01 1.01046872e+00 -8.18062544e-01 -4.63007659e-01
4.36039940e-02 -4.17512655e-01 -7.28881180e-01 3.74861091e-01
9.58991766e-01 -5.87992787e-01 -4.30550098e-01 1.13200498e+00
2.92902410e-01 1.24482233e-02 1.15080988e+00 -1.11006486e+00
-1.10449910e+00 6.14449322e-01 -1.01206982e+00 2.75480509e-01
2.78787822e-01 4.37505320e-02 1.06954193e+00 -1.16566896e+00
7.19670713e-01 1.06948221e+00 3.64873528e-01 3.54439229e-01
-1.01061988e+00 -6.71045005e-01 1.03752077e-01 5.52530810e-02
-1.22649741e+00 -4.11540747e-01 5.93433976e-01 -4.21481013e-01
1.04401767e+00 3.56414407e-01 1.80927858e-01 1.17402518e+00
2.58342564e-01 1.48326063e+00 1.14512122e+00 -9.57360864e-01
-9.17374417e-02 1.26320466e-01 4.14163321e-01 1.10002613e+00
1.03190534e-01 -2.86185414e-01 -6.09897912e-01 4.51067418e-01
6.53997362e-01 8.32347199e-02 -1.74775988e-01 9.71339494e-02
-1.05283034e+00 5.81136286e-01 4.59609985e-01 4.17439520e-01
1.92693457e-01 1.27808690e-01 2.98559308e-01 2.37556875e-01
3.99751961e-01 3.38439673e-01 -2.13308498e-01 -1.71157196e-02
-1.20828736e+00 -2.00029314e-01 5.03997266e-01 1.33845270e+00
6.02004588e-01 3.54529798e-01 -2.90382266e-01 1.19285488e+00
3.21184874e-01 9.73263025e-01 3.84029478e-01 -6.13219917e-01
7.70222604e-01 6.81981921e-01 -2.12305680e-01 -9.37180281e-01
-2.46879369e-01 -3.62212746e-03 -1.23618245e+00 2.22683877e-01
5.17821133e-01 8.03596713e-03 -1.50145102e+00 9.48263884e-01
-3.12739521e-01 -8.16871285e-01 -1.41716927e-01 5.75053811e-01
7.55941391e-01 1.04268289e+00 -3.87790710e-01 8.72029737e-02
1.41140258e+00 -1.30487418e+00 -6.49827957e-01 -3.89013916e-01
7.08424449e-01 -1.30406559e+00 1.26092148e+00 9.59963381e-01
-9.03042078e-01 -6.20525897e-01 -1.05205786e+00 -3.33145708e-01
-7.52003968e-01 8.96316111e-01 6.08708262e-01 9.02609348e-01
-1.00373244e+00 6.75698742e-02 -5.63793719e-01 -5.46893418e-01
6.64652228e-01 3.01987231e-01 -1.31230429e-01 -5.62263668e-01
-4.09155756e-01 3.45771670e-01 2.87534505e-01 2.46583641e-01
-8.98021221e-01 -1.50052253e-02 -6.69248044e-01 1.12031656e-03
4.12851512e-01 6.25490993e-02 1.15741587e+00 -1.03730869e+00
-1.42829847e+00 1.22833300e+00 -6.14618957e-02 -1.54272512e-01
4.22851294e-01 -5.05632579e-01 -5.34536123e-01 4.22880024e-01
5.99898249e-02 8.01307559e-01 1.21850097e+00 -1.39622402e+00
-5.19908071e-01 -7.14041531e-01 -3.70321751e-01 6.42279685e-02
-6.28548324e-01 2.95288861e-01 -7.30999947e-01 -1.00067627e+00
1.18056722e-01 -9.29295421e-01 3.61846924e-01 1.02914259e-01
-1.01072240e+00 -1.84214145e-01 1.17824829e+00 -7.03251779e-01
1.06984210e+00 -2.04495859e+00 -3.03346008e-01 2.14101791e-01
2.33545229e-02 5.17495155e-01 -3.07344675e-01 5.84306538e-01
-5.32689579e-02 1.98955625e-01 -1.78214252e-01 -3.77014011e-01
-2.74669868e-03 -2.17970386e-01 -5.43496668e-01 4.30449665e-01
1.28777787e-01 9.10034835e-01 -4.11761403e-01 -5.61413586e-01
3.20101112e-01 2.70027429e-01 -1.06666125e-01 5.19044474e-02
-4.51163024e-01 -1.73181087e-01 -3.76281142e-01 1.25027823e+00
6.69025600e-01 -3.77524853e-01 9.94591713e-02 -3.42046291e-01
-2.99765795e-01 -2.75618255e-01 -1.10387170e+00 1.62787139e+00
-8.47356394e-02 1.26192439e+00 1.33635774e-01 -1.01884449e+00
1.07613027e+00 -1.34078205e-01 1.63688034e-01 -8.80444467e-01
4.83250260e-01 1.09860681e-01 -1.48373455e-01 -3.34312081e-01
6.85447991e-01 5.12389123e-01 7.73096904e-02 7.36420095e-01
1.40026301e-01 -4.17414725e-01 6.53569758e-01 5.00795424e-01
9.90762234e-01 1.74926631e-02 -6.57097548e-02 -3.75814348e-01
7.34624088e-01 1.95394501e-01 -4.59754676e-01 1.10124254e+00
1.81738622e-02 8.04681778e-01 3.26120794e-01 -8.40623528e-02
-1.06681454e+00 -8.73938918e-01 -2.20036060e-01 1.47112429e+00
1.79051124e-02 -3.56103241e-01 -9.17952776e-01 -6.52711749e-01
-1.21261887e-01 5.08075714e-01 -4.42510575e-01 8.39769319e-02
-6.21349990e-01 -6.58821285e-01 8.39791358e-01 8.62718701e-01
7.12424755e-01 -1.22035944e+00 -2.77815223e-01 -3.05233806e-01
1.67668000e-01 -1.11241400e+00 -7.60641813e-01 6.66553974e-01
-6.97014987e-01 -8.38471532e-01 -9.39722121e-01 -1.18186295e+00
1.05914497e+00 3.95381689e-01 6.64610565e-01 1.44409791e-01
-7.62567341e-01 5.33670545e-01 -4.25832748e-01 -4.72962886e-01
-2.46376619e-01 -3.31307314e-02 -2.30491370e-01 -5.92561401e-02
2.57448614e-01 1.76121488e-01 -5.63349009e-01 3.24829251e-01
-1.11162150e+00 -7.31420219e-02 7.55957425e-01 1.05918300e+00
5.40829360e-01 4.33789454e-02 -1.01640731e-01 -8.28012288e-01
6.56674802e-01 3.88329506e-01 -6.77817941e-01 4.39223945e-01
-7.63832986e-01 -5.75867482e-02 9.23030376e-01 -2.91314542e-01
-1.21628106e+00 -7.47780874e-02 1.51894048e-01 -1.03498839e-01
-5.01633704e-01 2.10089028e-01 -1.18448451e-01 1.56480119e-01
7.40613401e-01 4.44759458e-01 -4.83592868e-01 -4.14701015e-01
4.82083857e-01 1.20023739e+00 7.23968744e-01 -4.10714626e-01
8.64245892e-01 6.17574215e-01 -2.62165099e-01 -1.12403142e+00
-1.00598717e+00 -5.32442629e-01 -8.59814465e-01 -1.31753534e-01
9.11280990e-01 -6.69941604e-01 -7.77342439e-01 8.72328579e-01
-9.67694938e-01 -5.77404022e-01 7.33509883e-02 1.60852119e-01
-2.99009293e-01 6.23178840e-01 -7.94788718e-01 -1.02525723e+00
-7.62830317e-01 -9.63057816e-01 1.33109057e+00 3.09714675e-01
-3.43266986e-02 -6.82798326e-01 -3.71289521e-01 7.59745240e-01
8.11690763e-02 -1.05541274e-01 9.72970068e-01 -6.24628425e-01
-7.17098236e-01 -3.61988634e-01 -8.89922261e-01 6.70535386e-01
6.73488453e-02 4.41009998e-01 -1.07060826e+00 -3.78397882e-01
-5.13411343e-01 -6.31330371e-01 1.16180003e+00 4.14340049e-01
1.30480742e+00 -5.27010709e-02 -2.99524277e-01 6.68301642e-01
1.27706182e+00 5.81896365e-01 7.12879658e-01 4.30809945e-01
9.06446815e-01 2.95912534e-01 5.16145170e-01 5.08571029e-01
-9.76416916e-02 2.65550017e-01 1.40844807e-01 -3.02043796e-01
-3.26565802e-01 -7.61743337e-02 5.09388268e-01 4.51891452e-01
4.22576107e-02 -1.08826542e+00 -1.28419888e+00 1.56162873e-01
-1.50584340e+00 -7.32136369e-01 -3.45332265e-01 1.78099573e+00
7.15935528e-01 4.55915838e-01 -2.10347563e-01 4.19078261e-01
7.98173130e-01 2.67012924e-01 -5.49503386e-01 -7.54360855e-01
-3.79388332e-01 3.33811700e-01 5.03290057e-01 1.04540609e-01
-1.34273875e+00 1.25716364e+00 5.19224024e+00 8.33777845e-01
-1.04558146e+00 -4.51665789e-01 9.19439077e-01 2.65097052e-01
2.44928241e-01 -3.15485537e-01 -9.35602307e-01 1.95752308e-01
4.87188816e-01 2.23701581e-01 4.30762500e-01 8.52950811e-01
-2.31596664e-01 -5.04803240e-01 -1.12260914e+00 1.19587398e+00
6.52520955e-01 -1.24523818e+00 1.89023331e-01 -1.21848267e-02
8.40367496e-01 -1.83556974e-01 4.29570019e-01 8.06937590e-02
3.60837817e-01 -1.10541260e+00 7.09720671e-01 1.29615933e-01
1.28562438e+00 -6.26403928e-01 4.37892586e-01 1.23760857e-01
-1.15751994e+00 -1.40272692e-01 -1.43824756e-01 3.60415220e-01
-2.73025781e-01 6.30740941e-01 -7.66489148e-01 2.44163141e-01
9.08670127e-01 8.23066890e-01 -8.17497313e-01 5.16039014e-01
-3.77032667e-01 6.30855918e-01 -2.67943650e-01 -2.46201083e-01
4.06251550e-01 -1.87320858e-01 2.03024313e-01 1.58805656e+00
3.66242826e-02 -1.10532761e-01 1.52296707e-01 7.52039969e-01
-4.05039728e-01 1.80673823e-01 -4.18175668e-01 -4.64232773e-01
-8.69871080e-02 1.25404155e+00 -1.33591115e+00 -4.21846837e-01
-2.04196796e-01 1.41210854e+00 -4.64863218e-02 5.12062132e-01
-5.68420887e-01 -9.18887317e-01 -1.32675022e-01 -2.76157200e-01
5.33696294e-01 -3.56063366e-01 -4.08359706e-01 -1.14046443e+00
8.31031129e-02 -1.02102435e+00 4.46206182e-01 -1.10278630e+00
-1.09142327e+00 5.30137241e-01 -3.13863993e-01 -8.85814607e-01
-1.03439882e-01 -1.35919774e+00 -5.09691775e-01 4.56126779e-01
-1.32364607e+00 -1.25460029e+00 -5.66524029e-01 6.92288935e-01
8.75092387e-01 -4.49838251e-01 6.93723917e-01 -2.18613595e-02
-9.08815801e-01 9.40800726e-01 7.29556918e-01 6.88340187e-01
9.87554073e-01 -1.46924865e+00 1.83182374e-01 1.02048576e+00
5.17346621e-01 5.89161873e-01 2.23156795e-01 -5.89491844e-01
-1.87082136e+00 -7.35739112e-01 3.60793591e-01 -5.30732989e-01
6.36675477e-01 -6.88307524e-01 -5.91827869e-01 6.31764829e-01
4.50308084e-01 -8.27767029e-02 4.66934413e-01 -2.54636407e-01
-5.86971760e-01 -2.53091633e-01 -9.38083827e-01 7.02371001e-01
8.97083342e-01 -6.56930149e-01 -1.96047634e-01 6.64548934e-01
1.19251356e-01 -3.18721473e-01 -5.75539887e-01 5.19066527e-02
7.15584815e-01 -9.44785476e-01 8.41046393e-01 -1.70988560e-01
7.41474211e-01 -7.96737671e-02 -2.12890550e-01 -5.68722665e-01
-3.98286283e-02 -5.78265190e-01 -1.70917898e-01 1.35824275e+00
5.07013619e-01 -1.30885631e-01 9.16757107e-01 2.61982143e-01
-2.20844328e-01 -3.91013891e-01 -4.24892515e-01 -7.37616420e-01
-4.85600643e-02 -3.48949313e-01 -1.53106868e-01 6.89620316e-01
-1.01871684e-01 4.50997978e-01 -2.80235410e-01 -5.91728538e-02
4.79067057e-01 3.89285862e-01 8.14234436e-01 -9.50696111e-01
-2.41906364e-02 -5.95681071e-01 -3.02715376e-02 -1.46340561e+00
-1.75918400e-01 -9.14073050e-01 1.99213818e-01 -1.69112778e+00
3.23481411e-01 -9.18074418e-03 -1.69918031e-01 5.24353266e-01
1.29515707e-01 4.14701492e-01 3.34237158e-01 2.59863049e-01
-7.37886548e-01 1.60728320e-01 1.22477126e+00 -6.69598281e-01
-5.05829528e-02 -3.49520862e-01 -4.27910119e-01 6.86630249e-01
9.48251963e-01 -2.77308345e-01 -1.99602932e-01 -5.83670914e-01
-3.91859785e-02 -3.08021307e-01 1.66197658e-01 -8.26697826e-01
3.51858139e-01 7.75414556e-02 1.07091522e+00 -1.12995565e+00
2.17845261e-01 -5.37425935e-01 -9.21726286e-01 3.52861762e-01
-4.65182334e-01 1.40388757e-01 3.45698655e-01 4.43828136e-01
-2.21113041e-01 -5.01397729e-01 8.19600999e-01 -1.08870827e-01
-7.46316791e-01 -6.99077174e-02 -7.01871276e-01 3.88667956e-02
6.71557069e-01 -5.49341321e-01 -7.83701003e-01 -2.65448987e-01
-3.10288429e-01 3.85789163e-02 2.38482356e-01 3.86469990e-01
8.12223732e-01 -6.69869661e-01 -4.87992346e-01 3.59998286e-01
3.35841566e-01 -1.05518904e-02 1.40557336e-02 5.19550383e-01
-5.99116266e-01 7.14099944e-01 -1.46093160e-01 -7.49059498e-01
-1.40406978e+00 4.59119380e-01 -1.21464081e-01 -2.87227124e-01
-4.86948013e-01 1.02322888e+00 1.11203134e-01 -1.11939386e-01
5.84572434e-01 -4.86781359e-01 1.44426644e-01 1.04471728e-01
5.83038211e-01 3.85714442e-01 1.79149568e-01 -3.21049750e-01
-2.90881932e-01 7.65767634e-01 -6.99606717e-01 -3.70314978e-02
1.16778719e+00 -2.62829494e-02 -7.06116706e-02 8.64014998e-02
1.27995920e+00 -8.43951628e-02 -9.01143253e-01 -3.50360870e-02
6.71876073e-02 -4.21462238e-01 1.85108587e-01 -1.20892799e+00
-9.29095566e-01 1.14544535e+00 6.87378585e-01 1.29606456e-01
1.28131509e+00 -1.57651305e-01 6.16832793e-01 1.02030122e+00
-4.57252935e-02 -1.48384070e+00 7.56646633e-01 6.54814601e-01
9.51415718e-01 -1.42888582e+00 3.71132165e-01 -2.03944027e-01
-6.67165458e-01 1.52173066e+00 7.70894945e-01 5.95945604e-02
3.95958424e-01 5.80111027e-01 1.08297743e-01 -2.65804023e-01
-4.15014237e-01 -3.09336305e-01 4.79236275e-01 5.52236199e-01
7.40168512e-01 -2.63123780e-01 1.29876181e-01 8.35879073e-02
-6.57049045e-02 -2.45384946e-01 5.90722144e-01 1.41467285e+00
-3.71423036e-01 -9.95873928e-01 -6.43493474e-01 6.03576899e-01
-2.84539908e-01 -3.31797779e-01 -9.23931062e-01 6.81691229e-01
-1.99030519e-01 1.01704478e+00 -4.51787747e-02 -9.10988152e-02
1.10856473e-01 1.80558562e-01 5.84351003e-01 -5.69474936e-01
-6.16715610e-01 4.51540291e-01 -5.35448380e-02 -2.80365124e-02
-1.85151920e-01 -4.01707202e-01 -1.33288884e+00 -6.76158220e-02
-6.09985054e-01 -3.60994041e-01 7.34888077e-01 5.50572276e-01
1.14444546e-01 4.36185956e-01 4.82072383e-01 -7.13613153e-01
-3.20589960e-01 -9.59374607e-01 -6.90357625e-01 2.57889837e-01
2.50230819e-01 -1.97639942e-01 -2.34632090e-01 6.62214160e-01] | [11.711808204650879, 2.289491653442383] |
40141dca-a873-4b44-b391-8a91be2eb329 | goal-conditioned-video-prediction | null | null | https://openreview.net/forum?id=B1g79grKPr | https://openreview.net/pdf?id=B1g79grKPr | Goal-Conditioned Video Prediction | Many processes can be concisely represented as a sequence of events leading from a starting state to an end state. Given raw ingredients, and a finished cake, an experienced chef can surmise the recipe. Building upon this intuition, we propose a new class of visual generative models: goal-conditioned predictors (GCP). Prior work on video generation largely focuses on prediction models that only observe frames from the beginning of the video. GCP instead treats videos as start-goal transformations, making video generation easier by conditioning on the more informative context provided by the first and final frames. Not only do existing forward prediction approaches synthesize better and longer videos when modified to become goal-conditioned, but GCP models can also utilize structures that are not linear in time, to accomplish hierarchical prediction. To this end, we study both auto-regressive GCP models and novel tree-structured GCP models that generate frames recursively, splitting the video iteratively into finer and finer segments delineated by subgoals. In experiments across simulated and real datasets, our GCP methods generate high-quality sequences over long horizons. Tree-structured GCPs are also substantially easier to parallelize than auto-regressive GCPs, making training and inference very efficient, and allowing the model to train on sequences that are thousands of frames in length.Finally, we demonstrate the utility of GCP approaches for imitation learning in the setting without access to expert actions. Videos are on the supplementary website: https://sites.google.com/view/video-gcp | ['Sergey Levine', 'Chelsea Finn', 'Dinesh Jayaraman', 'Frederik Ebert', 'Karl Pertsch', 'Oleh Rybkin'] | 2019-09-25 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 3.26537400e-01 2.13675410e-01 -2.76283622e-01 -1.23843282e-01
-6.19616747e-01 -6.80090010e-01 9.21040475e-01 -2.40388960e-01
1.50457844e-01 7.19067454e-01 5.75154841e-01 -1.32147640e-01
1.75412938e-01 -8.20126593e-01 -1.26539242e+00 -6.61845744e-01
-1.95953071e-01 3.22392941e-01 -1.05100982e-01 8.58374536e-02
7.67623931e-02 9.34584215e-02 -1.37589407e+00 5.15802741e-01
6.70604527e-01 5.19046366e-01 5.19488633e-01 1.15517604e+00
3.33300620e-01 1.29322147e+00 -1.11091129e-01 -2.80621737e-01
5.08317471e-01 -9.36249733e-01 -4.91569012e-01 6.20202720e-01
1.67145222e-01 -7.07609892e-01 -5.39333045e-01 7.67125070e-01
-7.67243328e-03 5.84767103e-01 5.73003948e-01 -1.45817900e+00
-8.36171329e-01 7.29311585e-01 -2.81198233e-01 -8.13783407e-02
7.94538796e-01 7.03281760e-01 1.04356360e+00 -7.17788637e-01
8.89467299e-01 1.32438672e+00 4.24261957e-01 6.37729287e-01
-1.32678914e+00 -5.16696751e-01 5.37797034e-01 2.80801654e-01
-8.73671055e-01 -4.38065797e-01 6.50061250e-01 -6.73689187e-01
9.44730341e-01 1.12036638e-01 9.72842991e-01 1.56684387e+00
2.99801618e-01 9.90048945e-01 9.88355517e-01 -2.13904217e-01
2.10755706e-01 -4.91298646e-01 -4.30483103e-01 8.88641238e-01
-3.70651633e-02 4.06284213e-01 -6.98257685e-01 9.11901444e-02
1.26574981e+00 2.58253843e-01 -3.64061654e-01 -4.88946140e-01
-1.54357755e+00 9.74510610e-01 1.14819162e-01 -1.29170761e-01
-7.95422435e-01 4.18266088e-01 1.08417623e-01 2.85472870e-01
2.27067605e-01 4.70219821e-01 -2.28419006e-01 -3.11550766e-01
-9.42186058e-01 4.35375869e-01 8.49059463e-01 1.22314298e+00
7.47834265e-01 2.77349979e-01 -3.17039907e-01 2.14716345e-01
1.79712072e-01 4.09227937e-01 3.08494419e-01 -1.64782190e+00
2.99535722e-01 7.44196847e-02 4.94726658e-01 -8.50285470e-01
3.56272012e-02 4.39292565e-03 -7.15117335e-01 1.47055164e-01
5.05779624e-01 -4.59693968e-01 -1.12628734e+00 1.77630436e+00
2.14725330e-01 6.91894352e-01 -3.76363099e-02 9.21371877e-01
2.62538254e-01 1.29212201e+00 1.53112352e-01 -4.11913902e-01
9.30778503e-01 -1.38002264e+00 -5.13526678e-01 -1.22542717e-01
3.37249309e-01 -5.14678478e-01 9.63058531e-01 5.68906546e-01
-1.37053657e+00 -7.35711098e-01 -6.90752387e-01 -1.99279673e-02
1.87130228e-01 8.15208536e-03 6.01870179e-01 -2.72646528e-02
-1.11979401e+00 9.17720020e-01 -1.34579504e+00 -2.86589742e-01
3.07123721e-01 -3.34266946e-02 -3.01080018e-01 -1.47737995e-01
-7.72489190e-01 5.67258894e-01 5.20898223e-01 1.34901768e-02
-1.70602620e+00 -7.52117217e-01 -1.02539825e+00 1.20938137e-01
7.05430031e-01 -9.76233363e-01 1.54071724e+00 -1.52213812e+00
-1.81389213e+00 3.09037387e-01 -3.36705178e-01 -6.13307893e-01
5.99815130e-01 -5.14370680e-01 -5.28765321e-02 3.36981624e-01
3.03432774e-02 8.82248461e-01 1.18633103e+00 -1.27489746e+00
-7.44185269e-01 1.95740655e-01 3.17615122e-01 4.29424256e-01
3.39918733e-01 -1.02300726e-01 -4.47901189e-01 -6.61630452e-01
-3.24989140e-01 -1.05232644e+00 -4.47814643e-01 -7.55219311e-02
-3.74920785e-01 1.93947423e-02 4.54507947e-01 -1.01163995e+00
9.68930066e-01 -1.93205130e+00 6.70271337e-01 -1.39738709e-01
4.32163961e-02 -7.98478797e-02 -3.46711069e-01 6.35740280e-01
-6.26749769e-02 2.37676799e-02 -3.48136458e-03 -3.04692328e-01
8.07675645e-02 3.82738352e-01 -4.48938280e-01 1.96354166e-01
2.06824690e-01 1.02274442e+00 -1.25361741e+00 -3.03451926e-01
3.48168224e-01 3.99600476e-01 -8.27912629e-01 4.78183448e-01
-7.32971609e-01 7.88574100e-01 -3.49635780e-01 3.20794284e-01
-1.70696918e-02 -5.51541388e-01 3.65638524e-01 -2.25073565e-02
-1.58931494e-01 2.65197873e-01 -9.45450246e-01 1.78820288e+00
-3.18826497e-01 5.83939135e-01 -1.12459786e-01 -9.26960230e-01
5.34126878e-01 5.72576940e-01 5.12884021e-01 -4.56088960e-01
-6.13465793e-02 -3.16504925e-01 -2.05592334e-01 -5.72315156e-01
5.32962084e-01 -2.65466243e-01 1.05882101e-01 5.15481770e-01
2.65724242e-01 -1.17498517e-01 6.50202572e-01 3.61879081e-01
8.56259882e-01 1.03704607e+00 3.87920707e-01 2.27547601e-01
8.91383961e-02 7.82316625e-02 8.14744234e-01 8.13260198e-01
1.73410010e-02 6.78344309e-01 6.40287161e-01 -3.06925327e-01
-1.49704385e+00 -1.19613326e+00 6.44000828e-01 1.19550431e+00
9.52795446e-02 -6.47826731e-01 -6.43038452e-01 -4.04150665e-01
-2.11614221e-01 9.79147732e-01 -6.65373564e-01 7.06700832e-02
-8.31105173e-01 -2.89671004e-01 -3.15534603e-03 7.66382754e-01
2.87092924e-01 -1.24659884e+00 -6.90558553e-01 4.10176724e-01
-4.72049028e-01 -9.99377728e-01 -7.17295289e-01 -1.53819472e-01
-9.03999329e-01 -1.01373315e+00 -7.81347632e-01 -6.32527769e-01
5.93972802e-01 2.10039034e-01 1.29881263e+00 -9.60008353e-02
1.56575255e-02 7.66055346e-01 -3.84746104e-01 -1.32194459e-01
-8.19210947e-01 -3.14245850e-01 -6.42488599e-02 -1.20639680e-02
-8.72796550e-02 -6.41931295e-01 -6.55685663e-01 1.76334172e-01
-6.59629107e-01 9.35037434e-01 4.83554035e-01 8.97194743e-01
8.06142509e-01 -1.54349387e-01 2.43925497e-01 -6.71833694e-01
1.62105620e-01 -6.35269880e-01 -5.87853551e-01 2.40732014e-01
-2.51421690e-01 9.71158445e-02 7.59505033e-01 -7.16868520e-01
-1.32641339e+00 2.32924640e-01 2.55424768e-01 -9.27836835e-01
-2.08988428e-01 4.37970877e-01 1.78200349e-01 6.18818462e-01
4.56839472e-01 4.81615365e-01 8.31927080e-03 -2.75970906e-01
6.46383822e-01 -2.20232904e-01 6.93713129e-01 -7.14645565e-01
8.25029194e-01 2.78154939e-01 -8.81850645e-02 -6.42112851e-01
-6.37412250e-01 -1.24559641e-01 -4.98648465e-01 -4.59872067e-01
1.01683152e+00 -1.16229951e+00 -6.99763477e-01 2.87843436e-01
-8.94006133e-01 -1.06836021e+00 -3.86822730e-01 6.26467586e-01
-1.15268934e+00 4.29553539e-01 -9.74794626e-01 -7.06166148e-01
2.83430666e-02 -9.44353700e-01 8.47734332e-01 3.03392112e-01
-3.27345133e-01 -1.01473284e+00 5.50947385e-03 6.53996244e-02
7.81640876e-03 4.67814773e-01 6.61005914e-01 -2.27118343e-01
-1.01929152e+00 2.44025752e-01 2.49548599e-01 2.56590873e-01
1.35062501e-01 2.77067691e-01 -4.10475075e-01 -3.95284206e-01
-1.69731066e-01 -3.27663898e-01 6.54986084e-01 6.54715240e-01
8.90848458e-01 -6.21660590e-01 -2.71708637e-01 5.98927915e-01
1.27767897e+00 3.69692594e-01 5.58854342e-01 1.76574856e-01
7.80876100e-01 3.92800242e-01 7.96183825e-01 5.84812522e-01
3.54741573e-01 4.81235504e-01 3.49919111e-01 1.44052833e-01
-2.28358552e-01 -9.64299500e-01 9.06787217e-01 7.33600974e-01
-6.60722256e-01 -4.44652081e-01 -4.99254316e-01 5.57799995e-01
-2.11151171e+00 -1.47807896e+00 1.94125772e-02 2.00395656e+00
7.39325464e-01 -1.23336740e-01 3.70703638e-01 -3.01296771e-01
6.84191942e-01 2.58303821e-01 -5.50948083e-01 -8.55525956e-02
1.84944645e-01 6.28004968e-02 2.90756226e-01 6.74235940e-01
-1.11309791e+00 9.33512747e-01 6.06800222e+00 4.18601334e-01
-9.30033028e-01 -5.49836569e-02 7.27995694e-01 -4.42792982e-01
-3.76243532e-01 2.86720216e-01 -5.53707540e-01 5.23184657e-01
9.82649267e-01 -3.44407678e-01 8.33929121e-01 8.15565288e-01
6.04753971e-01 -8.37591738e-02 -1.55442774e+00 8.21768939e-01
-2.03455016e-02 -1.40531325e+00 2.13792071e-01 -1.11541785e-01
9.60632145e-01 -1.77263990e-01 -4.84361909e-02 4.24513549e-01
8.96088541e-01 -9.68010724e-01 1.05798900e+00 6.28525555e-01
5.52026868e-01 -4.49937522e-01 -4.64809388e-02 4.08986062e-01
-1.25046277e+00 -2.26452664e-01 -3.10079902e-01 -3.11554879e-01
6.06659651e-01 -6.57422468e-02 -7.50688672e-01 5.46885312e-01
4.37494457e-01 1.11600757e+00 -1.44032240e-01 8.40969086e-01
-6.01920307e-01 8.62756670e-01 -1.60492599e-01 1.83368474e-01
3.92469645e-01 -5.56369960e-01 5.36920488e-01 1.09905779e+00
6.54506683e-01 2.94281542e-01 4.33082759e-01 1.02078724e+00
1.31772131e-01 -3.26111227e-01 -5.75604916e-01 -1.92769080e-01
1.92208946e-01 1.05035460e+00 -6.71126306e-01 -5.90435088e-01
-5.59789956e-01 1.17260182e+00 9.26638618e-02 7.60725319e-01
-1.18356454e+00 3.16625595e-01 5.58888316e-01 7.75668994e-02
6.78263068e-01 -4.01367188e-01 2.09998935e-01 -1.49311316e+00
-2.91201651e-01 -1.22578967e+00 3.23393047e-01 -1.13198376e+00
-1.03879583e+00 2.40111127e-01 1.71629742e-01 -1.23691094e+00
-7.96464324e-01 -3.51575613e-01 -5.57842135e-01 5.10404825e-01
-1.20923376e+00 -1.11806929e+00 -2.30791017e-01 6.41392112e-01
1.03739142e+00 2.79148728e-01 6.60261214e-01 -1.66282162e-01
-4.57166880e-01 -3.59705165e-02 -2.25604977e-02 7.47481063e-02
5.28988123e-01 -1.26535296e+00 4.71785367e-01 1.20980358e+00
4.72740501e-01 3.95056635e-01 9.44731772e-01 -8.70605469e-01
-1.44873726e+00 -1.08952761e+00 4.34571177e-01 -4.92119044e-01
6.24501526e-01 -2.76348650e-01 -7.50019610e-01 1.41456640e+00
3.71318907e-01 -3.46208215e-01 3.78280640e-01 -7.94825926e-02
-1.69529736e-01 3.43359441e-01 -5.07146597e-01 8.91909361e-01
1.18729019e+00 -3.20585787e-01 -5.76677144e-01 4.12251860e-01
6.33958578e-01 -5.88618338e-01 -8.61161709e-01 2.51207538e-02
5.30620933e-01 -9.64545786e-01 1.00965166e+00 -6.97425902e-01
1.02490473e+00 -2.64115721e-01 4.90038991e-02 -1.44000483e+00
-6.79197192e-01 -1.03572893e+00 -5.18012762e-01 8.92587900e-01
2.92170763e-01 -1.89988911e-01 7.55567670e-01 6.68396175e-01
-2.67008096e-01 -5.55882573e-01 -2.75148183e-01 -7.76088357e-01
-1.00923084e-01 -2.80933946e-01 4.39479828e-01 6.81592166e-01
3.51572633e-02 2.28861675e-01 -9.36941326e-01 2.06692636e-01
4.72763509e-01 4.92014140e-01 9.73008811e-01 -7.01595128e-01
-9.62195039e-01 -2.27534086e-01 3.09432037e-02 -1.46394563e+00
-2.07536872e-02 -6.29765153e-01 3.34847838e-01 -1.58451879e+00
4.29083586e-01 -2.00340748e-02 3.30628939e-02 5.34473538e-01
-3.79253417e-01 -5.67298420e-02 7.11544275e-01 2.77024329e-01
-7.12261915e-01 5.34782529e-01 1.64641476e+00 5.61430827e-02
-2.42069334e-01 -1.36015385e-01 -4.06077057e-01 7.16714799e-01
5.98817229e-01 -2.10008860e-01 -7.71422565e-01 -3.83134067e-01
2.24871069e-01 6.64237797e-01 6.30865395e-01 -8.07500005e-01
-4.93351649e-03 -6.10291004e-01 5.45882821e-01 -2.91671932e-01
4.35043484e-01 -4.57209587e-01 7.77248859e-01 4.56816643e-01
-5.17425537e-01 1.08270518e-01 -1.34234414e-01 8.07954609e-01
-4.61507626e-02 -4.89607267e-02 4.34897482e-01 -4.83112603e-01
-8.39821815e-01 5.74672759e-01 -6.25302374e-01 -1.17811248e-01
1.17353141e+00 -2.51323670e-01 -8.03512484e-02 -9.45551753e-01
-1.07354736e+00 2.41433427e-01 6.96331859e-01 4.23816621e-01
5.65317452e-01 -1.23777211e+00 -7.02236354e-01 -3.51206376e-03
-3.59896272e-01 1.81411002e-02 3.92644137e-01 7.08225608e-01
-4.79021370e-01 2.52655953e-01 -3.23777080e-01 -5.38046241e-01
-1.11954677e+00 1.10413563e+00 9.89184156e-02 -3.63679081e-01
-9.57133710e-01 6.84227884e-01 7.99256861e-01 7.02829435e-02
-2.16982085e-02 -5.12562096e-01 4.03348207e-02 -2.82261729e-01
5.13947010e-01 3.71481419e-01 -7.62979150e-01 -4.80157584e-01
3.34962934e-01 1.89548612e-01 9.04237702e-02 -1.91492364e-01
1.33737421e+00 -2.62280792e-01 3.34477812e-01 5.68753541e-01
7.65968323e-01 -2.69297749e-01 -2.29054642e+00 3.96885946e-02
-3.67175490e-01 -4.31881875e-01 -3.40688318e-01 -6.98414385e-01
-9.73735094e-01 6.98463500e-01 -4.60325815e-02 2.80645564e-02
1.07699478e+00 -1.66048989e-01 6.98010743e-01 2.30767518e-01
4.17689711e-01 -9.10947680e-01 4.60176945e-01 2.95351952e-01
9.34411764e-01 -9.19918478e-01 -1.25592783e-01 -3.70337844e-01
-1.02802360e+00 1.01095068e+00 4.83640522e-01 -4.16419566e-01
1.95180878e-01 1.23492405e-01 -3.60071480e-01 5.33099063e-02
-1.21142852e+00 4.87117842e-02 3.57352905e-02 6.19876087e-01
2.66296715e-01 1.99457318e-01 4.02961206e-03 4.22755927e-01
-1.56722248e-01 4.59100783e-01 7.26030171e-01 8.78171265e-01
-2.86032081e-01 -1.00748992e+00 -2.03367785e-01 3.66706371e-01
-2.76630044e-01 -6.35966361e-02 2.04027176e-01 7.36590207e-01
-1.23493269e-01 8.75094891e-01 1.89863235e-01 -1.24202073e-01
-8.10429081e-02 1.70490354e-01 8.06110442e-01 -5.87295413e-01
-3.18336308e-01 3.13483775e-01 2.41832748e-01 -9.56214368e-01
-5.72457314e-01 -1.01869810e+00 -1.10791993e+00 -4.71951187e-01
5.02960719e-02 1.01062562e-02 9.29935798e-02 7.74967670e-01
3.20678860e-01 5.39515853e-01 5.24195790e-01 -1.35356259e+00
-4.07569110e-01 -8.34732890e-01 -3.15576196e-01 7.34405756e-01
2.69422621e-01 -5.24053037e-01 -1.20995373e-01 1.01489067e+00] | [10.64317798614502, -0.5460270643234253] |
2076e6b2-6b27-4fcc-b101-1da271cd1b64 | deepsentipers-novel-deep-learning-models | 2004.05328 | null | https://arxiv.org/abs/2004.05328v1 | https://arxiv.org/pdf/2004.05328v1.pdf | DeepSentiPers: Novel Deep Learning Models Trained Over Proposed Augmented Persian Sentiment Corpus | This paper focuses on how to extract opinions over each Persian sentence-level text. Deep learning models provided a new way to boost the quality of the output. However, these architectures need to feed on big annotated data as well as an accurate design. To best of our knowledge, we do not merely suffer from lack of well-annotated Persian sentiment corpus, but also a novel model to classify the Persian opinions in terms of both multiple and binary classification. So in this work, first we propose two novel deep learning architectures comprises of bidirectional LSTM and CNN. They are a part of a deep hierarchy designed precisely and also able to classify sentences in both cases. Second, we suggested three data augmentation techniques for the low-resources Persian sentiment corpus. Our comprehensive experiments on three baselines and two different neural word embedding methods show that our data augmentation methods and intended models successfully address the aims of the research. | ['Parsa Abbasi Sarabestani', 'Seyed Abolghasem Mirroshandel', 'Javad PourMostafa Roshan Sharami'] | 2020-04-11 | null | null | null | null | ['persian-sentiment-anlysis'] | ['natural-language-processing'] | [-9.56690311e-02 3.77822429e-01 -2.14769378e-01 -7.34821558e-01
-5.59827626e-01 -4.04014677e-01 4.42443669e-01 7.52656907e-02
-6.88245893e-01 1.02249444e+00 3.44036549e-01 -4.12630647e-01
3.08721721e-01 -1.05268359e+00 -3.87258440e-01 -4.05099660e-01
2.94626653e-01 6.39507711e-01 8.11668485e-03 -7.81342685e-01
4.35397655e-01 3.37379545e-01 -1.04862750e+00 3.92826945e-01
9.51200902e-01 8.95266235e-01 9.95557308e-02 5.42453289e-01
-6.06294990e-01 1.11848652e+00 -8.99164855e-01 -1.12008011e+00
3.26731503e-02 -6.34106481e-03 -1.20847738e+00 -4.60302055e-01
3.65243793e-01 -2.64563680e-01 -5.94374090e-02 9.08601344e-01
7.32582271e-01 -1.21872768e-01 5.57455420e-01 -9.61068273e-01
-1.25337100e+00 1.21662676e+00 -3.00296813e-01 2.66011003e-02
-7.13812858e-02 -2.83564270e-01 1.34153485e+00 -6.38773322e-01
4.15305853e-01 1.19486356e+00 7.54807532e-01 7.51096129e-01
-6.50953233e-01 -5.42377949e-01 3.08055371e-01 4.22294199e-01
-5.42285860e-01 -1.77095085e-01 1.03189552e+00 -2.92057246e-01
1.15379703e+00 5.57280295e-02 6.67125106e-01 1.63578045e+00
3.67109954e-01 9.07577932e-01 1.18032002e+00 -5.01734138e-01
1.87029894e-02 3.15091789e-01 8.23356628e-01 5.66285849e-01
3.82631719e-01 -3.43518198e-01 -4.05355424e-01 4.25953558e-03
1.34892195e-01 1.80148035e-02 -1.08987622e-01 1.95588619e-02
-9.08813298e-01 9.42024887e-01 4.92769688e-01 7.62511551e-01
-3.16571921e-01 -1.32530198e-01 6.38017416e-01 1.23103186e-01
7.54060507e-01 7.02290893e-01 -1.18849087e+00 -1.24320991e-01
-6.26450717e-01 -1.08131647e-01 9.75148141e-01 7.33520865e-01
6.59556329e-01 3.46089184e-01 -1.43727109e-01 9.69150782e-01
1.56610250e-01 5.34334779e-01 9.72580612e-01 -4.52209860e-01
7.84554183e-01 8.99666607e-01 -9.88818780e-02 -1.25516748e+00
-5.70284724e-01 -3.16609770e-01 -9.93393123e-01 3.55840251e-02
7.34596327e-02 -4.02051598e-01 -9.86317039e-01 1.71672952e+00
-1.63637787e-01 -5.72640777e-01 4.14619267e-01 4.79132414e-01
1.00036633e+00 8.37672114e-01 1.99861690e-01 1.14202894e-01
1.54724681e+00 -1.24921453e+00 -9.16254044e-01 -6.60216987e-01
7.20349431e-01 -6.12606645e-01 1.24777722e+00 4.83413041e-01
-9.67976630e-01 -7.00867534e-01 -1.39644420e+00 -3.92670423e-01
-9.62593734e-01 4.62684810e-01 6.69957042e-01 6.66866481e-01
-1.00310528e+00 6.72481894e-01 -8.12340677e-01 -4.62012589e-01
2.81083882e-01 3.64734262e-01 -4.45176005e-01 1.99878171e-01
-1.41010487e+00 1.26062131e+00 4.20735985e-01 3.62392664e-01
-2.81122714e-01 -1.65658474e-01 -1.02199841e+00 1.23004541e-01
-1.36264861e-01 -4.74860221e-01 1.15899217e+00 -1.12853289e+00
-1.44357860e+00 9.33137119e-01 -1.48274228e-01 -6.93661213e-01
-7.26754740e-02 -5.50263882e-01 -4.76041883e-01 -2.86026925e-01
6.37949705e-02 8.15687835e-01 5.52344501e-01 -1.01086605e+00
-5.72315037e-01 -5.33426762e-01 2.55499572e-01 -1.73732519e-01
-9.38911378e-01 -1.04705155e-01 5.26454151e-02 -6.00350916e-01
-2.09464282e-01 -6.54415429e-01 -3.20269763e-01 -7.37738907e-01
-4.53348130e-01 -5.00362337e-01 8.56384695e-01 -9.09004271e-01
1.01541567e+00 -1.97938073e+00 2.54451819e-02 -2.97673285e-01
1.07666738e-01 6.55156732e-01 -3.81877154e-01 4.39071745e-01
-8.14008191e-02 4.55132186e-01 -3.67974162e-01 -5.52951455e-01
3.59799683e-01 4.91821378e-01 -6.40481055e-01 1.32913873e-01
6.96599782e-01 9.49551642e-01 -5.82211137e-01 -4.62438703e-01
-6.21918328e-02 5.16162515e-01 -4.61283118e-01 3.75100523e-01
-1.12334169e-01 2.29382534e-02 -2.94548810e-01 6.10108316e-01
7.24349320e-01 1.41164482e-01 1.35163933e-01 -3.35398972e-01
-6.46447390e-02 6.66164637e-01 -5.99177063e-01 1.62969601e+00
-5.79061449e-01 5.04082382e-01 3.44404764e-03 -1.23438561e+00
1.29546916e+00 1.45822674e-01 1.95769101e-01 -7.81217575e-01
3.92006904e-01 3.33867639e-01 9.17263478e-02 -4.64014351e-01
9.42088068e-01 -2.73630977e-01 -2.95987099e-01 2.39196315e-01
4.78749305e-01 2.61684856e-03 4.43781108e-01 -6.71015680e-02
7.73254216e-01 9.93988849e-03 2.58185893e-01 -3.66890669e-01
7.88223922e-01 3.71764004e-02 9.73565638e-01 3.08886498e-01
-1.43283978e-01 6.11637712e-01 7.48891234e-01 -7.88472295e-01
-9.13339198e-01 -4.19779748e-01 -7.58801326e-02 1.02555895e+00
-3.22062582e-01 -5.78263819e-01 -7.16071725e-01 -1.16627109e+00
-3.96230876e-01 8.70848536e-01 -6.98192537e-01 9.52194184e-02
-8.81944001e-01 -1.00765705e+00 6.82196081e-01 8.29109788e-01
6.63151801e-01 -1.55603290e+00 -4.62352842e-01 1.60954669e-01
-2.93550104e-01 -1.14561331e+00 -8.17097872e-02 6.75260127e-01
-8.41469944e-01 -1.06201541e+00 -3.95912081e-01 -1.17728937e+00
4.73994881e-01 -2.58892924e-01 1.43417490e+00 6.29083589e-02
2.22580746e-01 -4.00462635e-02 -6.68125391e-01 -7.08563864e-01
-9.59298387e-02 6.41695380e-01 -9.57452953e-02 -3.42293501e-01
9.05857563e-01 -5.96570194e-01 -3.07148010e-01 -2.73245037e-01
-7.63052166e-01 -2.45390728e-01 9.22132969e-01 1.01961672e+00
1.93162754e-01 -1.94922939e-01 4.84146684e-01 -9.63589132e-01
1.02898180e+00 -1.69218570e-01 -3.87676895e-01 6.25902340e-02
-7.31348276e-01 3.49786103e-01 9.08211112e-01 -1.63673446e-01
-1.03379917e+00 -1.64454982e-01 -8.38929057e-01 3.29688609e-01
-3.35435480e-01 3.85273308e-01 -3.04427177e-01 4.64240044e-01
4.42565173e-01 -5.26126735e-02 -3.09578806e-01 -5.93854845e-01
5.71206689e-01 9.50072765e-01 3.52396846e-01 -5.86016893e-01
3.65329504e-01 2.53646076e-01 -2.74355561e-01 -7.17166245e-01
-9.90591943e-01 2.37979437e-03 -9.32618618e-01 5.29130936e-01
1.26620102e+00 -6.38784766e-01 -4.27524120e-01 4.41007286e-01
-1.58671200e+00 -2.01057330e-01 -3.55767131e-01 2.25945890e-01
-1.36691824e-01 5.11061847e-01 -1.01064241e+00 -7.28481650e-01
-1.03939438e+00 -9.43750143e-01 8.57299328e-01 6.23585358e-02
-3.71433437e-01 -1.14096677e+00 5.13662577e-01 4.42948043e-01
4.43158269e-01 -2.71497488e-01 1.08484089e+00 -1.19178355e+00
1.95852742e-01 -3.18684697e-01 -2.78263837e-01 8.99591386e-01
-8.21164921e-02 8.83558914e-02 -1.26515543e+00 -1.28187820e-01
3.00526589e-01 -6.54270113e-01 1.07221806e+00 1.35175690e-01
1.09028935e+00 -4.94943440e-01 2.03810204e-02 3.56717438e-01
1.14532936e+00 3.94647181e-01 9.22160447e-01 6.51462317e-01
7.94514179e-01 7.57802367e-01 4.08375859e-01 -3.00546307e-02
6.87153220e-01 1.90412909e-01 4.99723613e-01 -2.90273309e-01
-4.08747746e-03 -7.64369890e-02 5.54348409e-01 1.40600634e+00
9.52374935e-03 -3.18120927e-01 -7.81214535e-01 7.30780423e-01
-1.74964046e+00 -8.30981731e-01 -3.19614619e-01 1.53437436e+00
8.19094539e-01 1.79229900e-01 -2.58348376e-01 3.53118062e-01
3.26420546e-01 4.08081681e-01 -2.50138342e-01 -1.07355464e+00
-3.53195578e-01 6.90554261e-01 3.38572592e-01 5.08978367e-01
-1.27365148e+00 1.19927120e+00 6.37603140e+00 5.39545298e-01
-1.24559033e+00 2.06468254e-01 5.91722071e-01 2.73811102e-01
-3.11794609e-01 -1.04747966e-01 -1.01476216e+00 3.64115089e-01
1.13554180e+00 6.00183547e-01 -1.23457767e-01 9.54595625e-01
-3.51199031e-01 3.54392767e-01 -8.17564189e-01 7.64454365e-01
4.98609006e-01 -1.05653918e+00 1.29422307e-01 -4.95063998e-02
4.13434446e-01 1.68030202e-01 -1.66703850e-01 8.42940807e-01
4.07589614e-01 -1.07714367e+00 5.16334176e-01 2.41742626e-01
1.38595358e-01 -8.76733720e-01 1.37077105e+00 2.24866986e-01
-7.20809519e-01 -2.50049055e-01 -5.96194088e-01 -4.26691025e-01
4.47296575e-02 5.87219715e-01 -6.08928859e-01 5.87174296e-01
9.37515438e-01 1.01653600e+00 -7.19339788e-01 2.27199554e-01
-6.97979331e-01 6.94969237e-01 1.98466197e-01 -4.65183020e-01
4.07376736e-01 -2.85303414e-01 -4.61628847e-02 1.51352024e+00
2.26570860e-01 -1.44311950e-01 -1.25135168e-01 7.82438695e-01
-6.67510107e-02 3.55234325e-01 -7.52965927e-01 -3.26822370e-01
4.82928231e-02 1.79590809e+00 -2.95258045e-01 -3.63547713e-01
-4.00749445e-01 1.03612542e+00 7.54320800e-01 1.90773845e-01
-4.46489662e-01 -6.07130647e-01 7.21176505e-01 -3.32819819e-01
2.43860245e-01 -3.42401117e-01 -7.55451381e-01 -1.52008092e+00
8.89894217e-02 -8.72488499e-01 3.03787231e-01 -7.87504613e-01
-1.60995150e+00 1.09814167e+00 -5.83357632e-01 -8.75981867e-01
-1.73686817e-01 -1.36565745e+00 -6.21841729e-01 7.10692883e-01
-1.74303603e+00 -1.46908224e+00 -6.29608938e-03 2.75016606e-01
6.94737077e-01 -5.02683580e-01 1.22608793e+00 3.87813956e-01
-7.94410527e-01 3.01691145e-01 -2.97575980e-01 5.30400872e-01
9.26428735e-01 -1.54853058e+00 4.92358088e-01 8.36915135e-01
1.51635915e-01 6.58210278e-01 4.27196532e-01 -3.73499751e-01
-1.10131466e+00 -8.42328191e-01 1.54622054e+00 -5.77287495e-01
7.36216009e-01 -3.92214984e-01 -8.07210743e-01 7.39429474e-01
1.01243341e+00 -6.34840071e-01 8.62465739e-01 4.21064883e-01
-1.76061958e-01 -3.55738670e-01 -9.66169596e-01 6.56664252e-01
5.50638795e-01 -3.01544577e-01 -1.11213076e+00 8.17411616e-02
7.91037679e-01 4.68107089e-02 -5.31243920e-01 6.33321702e-01
2.03321084e-01 -9.75731552e-01 7.80382991e-01 -9.29156840e-01
9.51985240e-01 -2.12589633e-02 -3.15496504e-01 -1.54689240e+00
-3.17744017e-01 1.13515146e-01 1.06497712e-01 1.57476509e+00
6.69009566e-01 -7.61955261e-01 6.51298165e-01 3.86602491e-01
-3.10401887e-01 -7.53087640e-01 -5.10242045e-01 -2.12706208e-01
6.45761251e-01 -4.38468903e-01 5.95566571e-01 1.15555751e+00
1.23013899e-01 1.00423753e+00 -4.79713649e-01 -1.15410328e-01
1.43020824e-01 8.66819248e-02 5.23219407e-01 -1.31845820e+00
-2.65048016e-02 -5.31472385e-01 -3.57701212e-01 -7.74372458e-01
6.47863150e-01 -8.24967563e-01 -9.79156494e-02 -1.96201992e+00
1.61007077e-01 -1.03609808e-01 -3.41494799e-01 8.40069830e-01
-1.23520061e-01 1.86717451e-01 4.79303449e-02 -1.38275206e-01
-2.15303421e-01 7.14158833e-01 9.00729716e-01 -6.34072840e-01
8.91768187e-02 -3.41790080e-01 -1.21903872e+00 9.77899313e-01
1.30114543e+00 -3.61033112e-01 -2.72521943e-01 -8.49208832e-01
6.15655422e-01 -6.58071697e-01 -1.02135427e-01 -8.51750731e-01
-1.01083800e-01 -1.46556264e-02 4.44970369e-01 -8.13276112e-01
4.39252108e-01 -7.96199977e-01 -7.77502477e-01 3.79732639e-01
-1.94464892e-01 3.49021494e-01 2.36862347e-01 2.47829482e-02
-4.25882846e-01 -4.18491930e-01 5.68132579e-01 -2.78213888e-01
-7.83032477e-01 -4.46713008e-02 -4.97541755e-01 -1.59147590e-01
5.44554532e-01 2.45958135e-01 -5.70823550e-01 -1.22143559e-01
-3.19805920e-01 2.54079700e-01 7.59759173e-02 5.35472810e-01
4.38131422e-01 -1.21740985e+00 -6.47126973e-01 2.86661327e-01
-1.22983374e-01 -1.13178782e-01 -1.04405902e-01 6.35660231e-01
-5.80488980e-01 6.66404963e-01 -4.58284825e-01 -1.31630898e-01
-1.05670536e+00 6.42184794e-01 1.45628914e-01 -6.28291368e-01
-2.61546105e-01 6.69761300e-01 -1.73908010e-01 -1.16624081e+00
2.15171024e-01 -5.65479338e-01 -9.13609385e-01 4.75405872e-01
4.32801455e-01 1.59561485e-01 2.85890341e-01 -6.36228085e-01
-2.62579679e-01 4.73121464e-01 -1.84240177e-01 -3.80722322e-02
1.56133437e+00 1.72113225e-01 -7.36788392e-01 5.10732353e-01
9.99767482e-01 2.61478007e-01 -4.40853029e-01 2.30902135e-01
1.96570963e-01 7.62460083e-02 -2.90960133e-01 -9.37535644e-01
-1.08832061e+00 1.37984455e+00 4.46486801e-01 3.72834504e-01
1.00837016e+00 -3.48346472e-01 1.15350270e+00 8.01492095e-01
9.24347043e-02 -1.25904298e+00 -3.45691182e-02 1.15527928e+00
7.93300807e-01 -1.22850633e+00 -3.44719559e-01 9.63528380e-02
-7.78587580e-01 1.42871320e+00 1.04812157e+00 -1.93806559e-01
5.61683476e-01 3.67361814e-01 6.14997625e-01 -1.14039525e-01
-7.52131939e-01 -2.52095670e-01 1.70428902e-01 6.14405572e-01
9.84310985e-01 -1.45884126e-01 -9.14140344e-01 1.18178928e+00
-6.87999189e-01 -2.26368695e-01 4.68367964e-01 9.84550118e-01
-5.26326060e-01 -1.58657467e+00 -1.33446306e-01 1.06260203e-01
-7.13837445e-01 -3.47699463e-01 -9.79672492e-01 5.98808348e-01
2.52060592e-01 1.03961790e+00 -1.80981848e-02 -6.62822366e-01
6.17897451e-01 3.64021540e-01 1.25919119e-01 -5.48491418e-01
-7.57495522e-01 -4.07580584e-01 4.60421979e-01 -2.73539543e-01
-4.63773936e-01 1.73953585e-02 -1.17564023e+00 -1.62328497e-01
-6.58777505e-02 3.07727307e-01 6.99540854e-01 1.03658414e+00
3.70881200e-01 5.73681951e-01 2.98125327e-01 -8.18211913e-01
-7.37468541e-01 -1.55471253e+00 -6.17442012e-01 4.60375547e-01
1.48476645e-01 -9.41113904e-02 -1.86238766e-01 -5.80513589e-02] | [11.379756927490234, 6.875241756439209] |
3825fc75-0e07-4a96-b4c5-61ad24e9c817 | consac-robust-multi-model-fitting-by | 2001.02643 | null | https://arxiv.org/abs/2001.02643v3 | https://arxiv.org/pdf/2001.02643v3.pdf | CONSAC: Robust Multi-Model Fitting by Conditional Sample Consensus | We present a robust estimator for fitting multiple parametric models of the same form to noisy measurements. Applications include finding multiple vanishing points in man-made scenes, fitting planes to architectural imagery, or estimating multiple rigid motions within the same sequence. In contrast to previous works, which resorted to hand-crafted search strategies for multiple model detection, we learn the search strategy from data. A neural network conditioned on previously detected models guides a RANSAC estimator to different subsets of all measurements, thereby finding model instances one after another. We train our method supervised as well as self-supervised. For supervised training of the search strategy, we contribute a new dataset for vanishing point estimation. Leveraging this dataset, the proposed algorithm is superior with respect to other robust estimators as well as to designated vanishing point estimation algorithms. For self-supervised learning of the search, we evaluate the proposed algorithm on multi-homography estimation and demonstrate an accuracy that is superior to state-of-the-art methods. | ['Florian Kluger', 'Michael Ying Yang', 'Eric Brachmann', 'Carsten Rother', 'Hanno Ackermann', 'Bodo Rosenhahn'] | 2020-01-08 | consac-robust-multi-model-fitting-by-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Kluger_CONSAC_Robust_Multi-Model_Fitting_by_Conditional_Sample_Consensus_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Kluger_CONSAC_Robust_Multi-Model_Fitting_by_Conditional_Sample_Consensus_CVPR_2020_paper.pdf | cvpr-2020-6 | ['homography-estimation'] | ['computer-vision'] | [ 3.80141914e-01 -3.37289393e-01 -1.84304714e-01 -2.81548709e-01
-1.03347278e+00 -6.71892941e-01 5.78058779e-01 -2.30679154e-01
-3.74474198e-01 2.76868612e-01 -2.76040226e-01 7.29898587e-02
-3.03239107e-01 -3.09928060e-01 -9.40458059e-01 -5.25153756e-01
1.99392825e-01 1.05805445e+00 3.18126470e-01 -4.43571471e-02
6.44825697e-01 6.71674252e-01 -1.30533767e+00 -3.81558776e-01
4.46129560e-01 9.84725058e-01 3.93202364e-01 6.08690143e-01
4.06896770e-01 3.01573455e-01 -1.43829778e-01 -3.28245431e-01
5.61048508e-01 -9.86320525e-02 -4.67523247e-01 5.98963439e-01
9.94834960e-01 -3.26628029e-01 -2.58619905e-01 9.65736687e-01
2.87569463e-01 9.78964046e-02 7.17091620e-01 -8.25828433e-01
-1.15679570e-01 6.95270598e-02 -6.60071731e-01 -1.72472432e-01
6.40620708e-01 -1.96656078e-01 9.25228238e-01 -1.13598609e+00
1.03959632e+00 9.56344962e-01 8.91649485e-01 3.05841386e-01
-1.37489593e+00 -3.83556426e-01 -1.59724101e-01 1.45526618e-01
-1.46579182e+00 -6.91135108e-01 1.10858262e+00 -5.14746487e-01
5.84111094e-01 1.69108212e-01 5.00613868e-01 9.43365514e-01
1.26578674e-01 8.64793539e-01 8.48212361e-01 -6.48988783e-01
-9.83712031e-04 -7.45593309e-02 -9.65216011e-02 9.85180855e-01
6.79508597e-02 2.39830434e-01 -5.11406958e-01 -3.61089736e-01
1.33335817e+00 -9.24478173e-02 -2.70523757e-01 -1.15733516e+00
-1.51158321e+00 6.70248032e-01 4.22148593e-02 6.99814409e-02
-3.97339284e-01 2.27705032e-01 1.49164230e-01 5.28670847e-02
9.14023444e-02 6.18270338e-01 -1.24302253e-01 -7.16341194e-03
-1.43910384e+00 1.20971039e-01 9.41707969e-01 1.19686949e+00
1.00171149e+00 1.39576510e-01 5.86368859e-01 7.01726615e-01
2.66339451e-01 5.63360333e-01 4.08417821e-01 -1.05499268e+00
3.30462813e-01 3.01611155e-01 3.81604612e-01 -1.31476867e+00
-4.62597579e-01 -5.59610128e-01 -8.61263633e-01 -9.36188847e-02
4.15888637e-01 1.06214218e-01 -6.40101016e-01 1.23334289e+00
5.14468372e-01 3.22766483e-01 -1.57411486e-01 1.04436648e+00
2.88554400e-01 2.26098001e-01 -9.03772533e-01 -3.50897759e-01
8.42708290e-01 -1.13325620e+00 -4.71006304e-01 -2.34614715e-01
3.83555472e-01 -1.22319448e+00 4.65087742e-01 7.57544458e-01
-9.21240270e-01 -5.72364748e-01 -1.12652087e+00 1.43655807e-01
2.36919314e-01 4.96402562e-01 4.33788747e-01 3.41364503e-01
-8.42189491e-01 9.03224230e-01 -9.02747154e-01 -5.26653588e-01
-1.21867307e-01 3.92868489e-01 -6.10035360e-01 2.12638900e-01
-4.48877931e-01 9.20176744e-01 2.70050496e-01 2.69062549e-01
-9.48944390e-01 -3.45038027e-01 -9.03190136e-01 -4.13636625e-01
6.66049659e-01 -7.73144901e-01 1.15141690e+00 -8.30781162e-01
-1.70973456e+00 9.88799751e-01 -4.11158413e-01 -5.74339807e-01
9.70356405e-01 -4.40236837e-01 -1.29138753e-01 2.91682094e-01
6.25640750e-02 4.08149898e-01 1.32119060e+00 -1.47587514e+00
-4.70802575e-01 -2.65190303e-01 -2.39476740e-01 4.40992005e-02
-4.20347042e-03 -9.39022750e-02 -6.12772524e-01 -4.62546349e-01
1.01739764e+00 -1.07403171e+00 -4.06528175e-01 2.41604373e-02
-5.04930615e-01 2.72613019e-01 9.43394184e-01 -6.25854015e-01
9.22170222e-01 -1.75482666e+00 5.84687114e-01 5.27247548e-01
5.96994758e-02 -2.45357156e-01 4.00408693e-02 3.59324515e-01
1.16701722e-01 -2.67213702e-01 -5.79581922e-03 -7.59044528e-01
-1.04772724e-01 2.40562513e-01 -1.54428795e-01 1.13938832e+00
-2.27997586e-01 4.26172405e-01 -8.22158277e-01 -5.00804484e-01
6.44462883e-01 1.40780941e-01 -3.34128648e-01 2.46180743e-01
1.22822069e-01 6.82774127e-01 -3.29311997e-01 7.42883742e-01
7.05692589e-01 -2.02186361e-01 7.45146349e-02 -4.47233886e-01
-1.18700475e-01 -2.20450297e-01 -1.91835427e+00 1.69740474e+00
-5.01662135e-01 5.26811063e-01 3.67758051e-02 -1.05344844e+00
1.28531587e+00 2.54305840e-01 7.09070563e-01 -6.11297674e-02
1.76379442e-01 5.47299981e-01 -3.53219867e-01 -4.23630327e-01
7.04355896e-01 1.78096592e-02 1.41953856e-01 4.10561003e-02
3.67824614e-01 -5.00425279e-01 3.46335657e-02 -2.65081674e-01
7.95931995e-01 5.46956658e-01 7.10936248e-01 -1.33252084e-01
8.07787418e-01 -1.14180215e-01 5.24677694e-01 7.84139097e-01
-5.85130006e-02 8.80319238e-01 -3.80663909e-02 -7.86378205e-01
-1.43969178e+00 -7.17347741e-01 -2.30639741e-01 4.54246253e-01
4.33333367e-01 -3.27320814e-01 -5.69086671e-01 -1.70209318e-01
7.61519745e-03 3.22585672e-01 -3.16600382e-01 2.99132884e-01
-7.63302147e-01 -8.96091461e-02 2.45002553e-01 3.15393329e-01
4.39685374e-01 -5.75759947e-01 -8.44752192e-01 3.62537444e-01
-3.81177999e-02 -1.52931070e+00 -6.19187713e-01 1.55647933e-01
-1.06655931e+00 -1.31953657e+00 -6.76887572e-01 -7.92823613e-01
7.54550517e-01 3.34910482e-01 9.92797494e-01 -1.05770685e-01
-7.39039332e-02 7.50739336e-01 -8.85096192e-02 -4.36034575e-02
-4.59744483e-01 -3.79632153e-02 4.78890270e-01 2.65864313e-01
-3.42971943e-02 -5.45064509e-01 -3.67823929e-01 9.27012861e-01
-4.33596075e-01 2.36059297e-02 4.80722576e-01 9.59737182e-01
9.06508386e-01 -3.15290570e-01 -1.95280790e-01 -4.99081343e-01
2.19875053e-01 -7.77283683e-02 -1.27356231e+00 2.43289337e-01
-3.81042063e-01 1.81224957e-01 6.48870289e-01 -6.72583342e-01
-7.44321585e-01 8.68587613e-01 1.69348925e-01 -1.05517220e+00
-2.86975235e-01 2.40992561e-01 2.15366647e-01 -7.65507042e-01
6.06904984e-01 4.28679079e-01 -8.96470901e-03 -5.59659779e-01
2.19253734e-01 1.83507025e-01 8.61782968e-01 -5.66582859e-01
1.19149530e+00 5.93083739e-01 5.04098415e-01 -1.22832966e+00
-5.16633570e-01 -1.13745427e+00 -1.35243797e+00 -3.46585840e-01
6.84834361e-01 -6.30870044e-01 -6.09761238e-01 4.70733404e-01
-1.39376938e+00 2.20851809e-01 1.28326282e-01 7.21968710e-01
-1.08415580e+00 9.38728213e-01 -2.70995587e-01 -1.04454207e+00
-1.26755714e-01 -1.35306704e+00 1.41436386e+00 -1.28976673e-01
-1.97427869e-01 -1.06241882e+00 3.22149813e-01 3.85278821e-01
-1.51969418e-01 3.71430904e-01 1.30098164e-01 -5.61117828e-01
-8.24163675e-01 -6.47386611e-01 3.25847715e-01 1.87780291e-01
-1.71379775e-01 1.21415161e-01 -7.29006529e-01 -5.07997215e-01
2.71846324e-01 -3.42485547e-01 6.08983994e-01 3.53440255e-01
8.11966956e-01 -1.57430008e-01 -5.01533687e-01 1.04181993e+00
1.59145415e+00 -6.08053766e-02 3.48798752e-01 8.58961701e-01
7.91446030e-01 4.46479172e-01 6.23358548e-01 6.03276193e-01
1.92195684e-01 1.06242836e+00 6.34050786e-01 1.40870810e-01
5.01722574e-01 -2.98501015e-01 4.22077291e-02 8.13526690e-01
-2.84679741e-01 2.26632416e-01 -8.80065322e-01 4.22252357e-01
-2.00319386e+00 -7.43609548e-01 -1.00697771e-01 2.49843669e+00
3.50975305e-01 1.72717586e-01 5.30696055e-03 1.10097840e-01
6.36999071e-01 2.05991253e-01 -5.15849292e-01 -5.21097556e-02
-2.02213228e-01 -1.65148363e-01 7.87655830e-01 6.43325806e-01
-1.43880594e+00 9.84126091e-01 6.34843206e+00 6.95248842e-01
-1.18206322e+00 -1.63949803e-01 -1.44948065e-01 3.54655474e-01
7.30458498e-02 3.03988963e-01 -1.03586924e+00 -6.91498145e-02
5.22479355e-01 1.33261472e-01 5.55498719e-01 1.07512748e+00
1.38502404e-01 -8.92472789e-02 -1.42922962e+00 1.36627162e+00
4.80218858e-01 -1.40148163e+00 -5.94248734e-02 6.96797296e-02
9.62292910e-01 -6.80419952e-02 -2.37600699e-01 -2.88278967e-01
-8.88155252e-02 -6.47106290e-01 8.09569061e-01 7.12002993e-01
2.11385146e-01 -3.77530366e-01 7.46747017e-01 7.42278159e-01
-1.25306821e+00 8.75966847e-02 -5.48752010e-01 9.22187269e-02
4.00217026e-01 3.46557856e-01 -1.00145793e+00 8.34934115e-01
2.95500159e-01 9.03891265e-01 -4.36395645e-01 1.39977705e+00
-8.68630409e-02 2.50043601e-01 -5.96015036e-01 1.04475118e-01
3.55668336e-01 -6.57949448e-01 9.03542042e-01 8.99981320e-01
5.24678051e-01 -3.13131452e-01 5.70446134e-01 6.80437624e-01
2.25878373e-01 2.17150822e-01 -8.32726002e-01 3.60942811e-01
3.42310905e-01 1.32418025e+00 -6.90325260e-01 -2.35108018e-01
-2.99387574e-01 1.08967698e+00 1.89751014e-01 2.13763863e-01
-4.71946150e-01 4.83567379e-02 -2.46911645e-02 2.15701059e-01
5.04266977e-01 -6.65858448e-01 -1.24449804e-01 -1.38907826e+00
1.77137449e-01 -8.59733760e-01 1.65858284e-01 -8.23547363e-01
-9.97827291e-01 4.80010867e-01 1.34484395e-01 -1.72132194e+00
-7.08992600e-01 -7.21690595e-01 -4.64014381e-01 4.85907614e-01
-1.20298767e+00 -1.40514171e+00 -2.14326710e-01 4.07054007e-01
7.64566243e-01 -4.26087111e-01 7.42404044e-01 -4.42298464e-02
-1.44989684e-01 3.00528258e-01 4.34594929e-01 3.95081155e-02
8.15121830e-01 -1.11154401e+00 1.38077781e-01 1.07131636e+00
5.68401754e-01 8.21145296e-01 8.80809784e-01 -5.02048910e-01
-1.76945412e+00 -4.92873877e-01 5.25616348e-01 -4.97646242e-01
8.55079770e-01 -2.48146310e-01 -7.31392264e-01 8.69701743e-01
-8.73251855e-02 1.03317857e-01 2.15774655e-01 -8.24503601e-02
-6.07802421e-02 2.20970009e-02 -9.07323480e-01 4.06230152e-01
8.55751395e-01 -4.30206031e-01 -4.90625978e-01 4.04511005e-01
1.53849661e-01 -8.81887436e-01 -8.35911214e-01 4.41344112e-01
7.67583013e-01 -1.03472209e+00 1.15919256e+00 -2.81606525e-01
2.30143487e-01 -3.31569105e-01 -3.56891125e-01 -1.18722892e+00
-3.03431034e-01 -9.75810170e-01 -2.78829575e-01 7.15835869e-01
2.48395931e-02 -2.91265070e-01 9.89898622e-01 6.20021932e-02
-1.37605652e-01 -6.09320045e-01 -1.16429889e+00 -8.25310767e-01
-2.88227588e-01 -3.68228704e-01 1.75874144e-01 8.19037974e-01
-5.52977562e-01 2.05915287e-01 -9.56350327e-01 6.21722579e-01
1.34342396e+00 3.47616106e-01 1.37686634e+00 -1.33262777e+00
-4.96574432e-01 -2.94054836e-01 -7.55556643e-01 -1.50815904e+00
3.41895133e-01 -5.76086164e-01 1.55399755e-01 -1.02601075e+00
-7.83163235e-02 -4.38983627e-02 2.20909119e-01 -9.68799554e-03
2.49060363e-01 2.40014687e-01 1.48954958e-01 5.13839960e-01
-5.92948616e-01 2.36749560e-01 1.24774396e+00 2.65908744e-02
-1.03164829e-01 3.54003757e-01 1.52084261e-01 1.22641683e+00
3.26992601e-01 -1.29785821e-01 -2.52042338e-02 -4.29145008e-01
1.95439067e-02 3.17898303e-01 6.21208131e-01 -1.24991453e+00
6.60708845e-01 -1.07326955e-01 4.07924056e-01 -1.08384287e+00
4.82892126e-01 -1.01565742e+00 3.63927960e-01 3.08317304e-01
-5.36833480e-02 1.31007165e-01 -1.39405131e-01 5.55606723e-01
-2.13999346e-01 -6.56444550e-01 7.70322382e-01 -3.73833269e-01
-9.53866303e-01 2.40957975e-01 -8.55271444e-02 -2.88444966e-01
7.22944379e-01 -7.81384647e-01 1.95220634e-01 -5.92411160e-01
-7.16422379e-01 7.79988095e-02 6.69546366e-01 2.29536846e-01
9.30386364e-01 -1.17850184e+00 -4.29043114e-01 3.40541810e-01
1.89879298e-01 7.79300705e-02 -9.52881053e-02 9.26441371e-01
-7.39703894e-01 3.98651123e-01 -1.03895947e-01 -1.14738572e+00
-1.29977214e+00 5.64196467e-01 5.31792700e-01 -1.19408593e-01
-4.75220501e-01 4.48028266e-01 -2.22163022e-01 -7.53629148e-01
2.83658981e-01 -2.27931231e-01 -8.81269341e-04 -2.07356483e-01
4.48376946e-02 6.31481946e-01 -8.47487822e-02 -1.07406521e+00
-2.81648070e-01 1.40053129e+00 1.95585966e-01 -7.50904009e-02
1.25771821e+00 -1.87008798e-01 -1.11970410e-01 5.81651747e-01
1.26692235e+00 -5.34859374e-02 -1.20831597e+00 -4.32245374e-01
1.96742877e-01 -7.53419816e-01 -8.94430131e-02 -1.71646371e-01
-7.17047930e-01 6.28985286e-01 5.05092740e-01 -1.69291556e-01
6.50210142e-01 -2.36998215e-01 3.98375362e-01 8.78946424e-01
8.69431794e-01 -1.12694144e+00 2.58196890e-01 5.57687938e-01
9.76919174e-01 -1.46686924e+00 3.35138530e-01 -4.75823641e-01
-2.67093003e-01 1.66220593e+00 4.23139960e-01 -6.38418138e-01
5.29875457e-01 2.12793127e-01 -1.17045850e-03 6.09938102e-03
-2.53389150e-01 1.34506300e-01 6.16738915e-01 5.93589604e-01
3.26621085e-02 -2.36836687e-01 1.31590329e-02 -1.67567819e-01
-3.36871505e-01 -3.10801137e-02 5.50652206e-01 7.93469846e-01
-4.14888352e-01 -1.14211702e+00 -9.46055830e-01 1.90612450e-01
1.45039763e-02 1.97429642e-01 -2.63028324e-01 9.55669761e-01
-2.28660896e-01 6.06532753e-01 -9.92966145e-02 -3.88869047e-01
4.50009912e-01 -1.17797486e-01 6.51465714e-01 -2.11384505e-01
-2.70003706e-01 7.10270584e-01 7.23964199e-02 -6.76512897e-01
-5.78004599e-01 -9.01839495e-01 -6.31190240e-01 1.64806179e-03
-6.23564422e-01 -1.28710970e-01 6.43354714e-01 1.02811706e+00
-9.43947807e-02 -1.33641548e-02 7.54769802e-01 -1.45365965e+00
-1.07023299e+00 -8.73634636e-01 -5.81863940e-01 3.82949531e-01
5.52577317e-01 -9.04722035e-01 -3.97848547e-01 1.69781327e-01] | [7.838505744934082, -2.3000361919403076] |
aa9ec860-0347-4c69-82f3-5f88939dc0e1 | v-net-fully-convolutional-neural-networks-for | 1606.04797 | null | http://arxiv.org/abs/1606.04797v1 | http://arxiv.org/pdf/1606.04797v1.pdf | V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation | Convolutional Neural Networks (CNNs) have been recently employed to solve
problems from both the computer vision and medical image analysis fields.
Despite their popularity, most approaches are only able to process 2D images
while most medical data used in clinical practice consists of 3D volumes. In
this work we propose an approach to 3D image segmentation based on a
volumetric, fully convolutional, neural network. Our CNN is trained end-to-end
on MRI volumes depicting prostate, and learns to predict segmentation for the
whole volume at once. We introduce a novel objective function, that we optimise
during training, based on Dice coefficient. In this way we can deal with
situations where there is a strong imbalance between the number of foreground
and background voxels. To cope with the limited number of annotated volumes
available for training, we augment the data applying random non-linear
transformations and histogram matching. We show in our experimental evaluation
that our approach achieves good performances on challenging test data while
requiring only a fraction of the processing time needed by other previous
methods. | ['Seyed-Ahmad Ahmadi', 'Fausto Milletari', 'Nassir Navab'] | 2016-06-15 | null | null | null | null | ['volumetric-medical-image-segmentation'] | ['medical'] | [ 4.95201558e-01 4.56402063e-01 2.44882330e-02 -6.15349233e-01
-6.13777876e-01 -4.46537435e-01 5.16252518e-01 7.43849456e-01
-9.19981122e-01 5.51874042e-01 -1.93695992e-01 -3.76678199e-01
4.44014296e-02 -7.45893836e-01 -5.79687357e-01 -4.82276291e-01
-3.97301704e-01 9.33952212e-01 2.70759791e-01 1.14379831e-01
4.54572402e-02 8.29646409e-01 -9.73498344e-01 2.38036305e-01
6.80038393e-01 1.03876591e+00 9.81114358e-02 7.34644353e-01
-2.44594976e-01 5.08359373e-01 -3.81167024e-01 -2.08308384e-01
4.91616517e-01 -3.94946516e-01 -9.34646249e-01 1.29358321e-01
4.13956046e-01 -2.01776251e-01 -1.50609598e-01 9.59106326e-01
6.38529539e-01 -8.30013976e-02 7.30994046e-01 -7.89016187e-01
-3.63538526e-02 5.57485342e-01 -4.79899675e-01 6.41051352e-01
-1.90777615e-01 1.16420820e-01 3.52034122e-01 -4.68542784e-01
7.00072110e-01 1.02990496e+00 9.21018124e-01 5.75596094e-01
-1.34528100e+00 -2.17448905e-01 -2.76433170e-01 -1.53501675e-01
-9.63498175e-01 -5.25941700e-02 6.15354538e-01 -5.67376316e-01
6.84678257e-01 3.98057640e-01 7.82466412e-01 4.62139398e-01
4.16325659e-01 6.69836521e-01 1.12029946e+00 -2.06585959e-01
2.90916413e-01 -2.11950973e-01 -1.53574675e-01 6.54455364e-01
1.23829424e-01 -1.76823497e-01 2.44014531e-01 -3.20966318e-02
9.92226303e-01 1.51466383e-02 -3.67735587e-02 -8.49735141e-01
-1.18612790e+00 9.62405205e-01 8.75846744e-01 6.56639516e-01
-4.60165352e-01 2.17699349e-01 8.06343555e-01 -1.18638743e-02
7.53154516e-01 3.10437113e-01 -3.08753788e-01 2.07312420e-01
-1.35281289e+00 2.06209823e-01 7.23567009e-01 2.74168879e-01
4.35513854e-01 -3.35529119e-01 -3.18963438e-01 7.13938415e-01
1.48787960e-01 1.24131121e-01 6.63443625e-01 -6.11855626e-01
1.95757166e-01 7.46605039e-01 -2.39002779e-01 -8.35970879e-01
-9.69899774e-01 -4.44316626e-01 -1.14396620e+00 3.06024730e-01
6.94293618e-01 4.49189283e-02 -1.36153615e+00 1.30754423e+00
5.55733800e-01 1.91606097e-02 -2.43596539e-01 9.23434794e-01
6.67894304e-01 1.65737629e-01 1.77470222e-01 -2.63930321e-01
1.27871597e+00 -7.12905049e-01 -5.26588500e-01 -1.99281260e-01
7.89123476e-01 -5.94365895e-01 6.08175039e-01 2.59029984e-01
-1.33195221e+00 -2.43863136e-01 -9.07299995e-01 -7.35300705e-02
-3.98738652e-01 -6.62316978e-02 5.97845256e-01 7.80070484e-01
-1.26901555e+00 9.86589730e-01 -1.19569969e+00 -2.60141999e-01
9.02326822e-01 7.96777964e-01 -3.88520032e-01 -3.72025408e-02
-6.78326011e-01 1.04678071e+00 6.02667987e-01 3.45627576e-01
-7.15515733e-01 -7.57715940e-01 -9.28483188e-01 1.14207249e-02
3.08858871e-01 -5.80173016e-01 1.22491288e+00 -8.53959918e-01
-1.17982125e+00 1.33281910e+00 3.75247926e-01 -7.68448472e-01
1.06247652e+00 1.99220777e-01 2.86439508e-01 1.63977355e-01
-1.96925253e-01 7.45087326e-01 4.95161086e-01 -1.21522808e+00
-1.63339257e-01 -8.11657548e-01 -2.93264836e-01 5.78658693e-02
3.30005860e-04 -9.08408239e-02 -4.44377720e-01 -4.03735459e-01
1.68455452e-01 -8.94652009e-01 -8.99755001e-01 2.53736019e-01
-3.96246403e-01 5.15027121e-02 7.48576939e-01 -6.93637252e-01
5.36891937e-01 -1.88311231e+00 2.90708840e-01 4.38546687e-01
5.41380048e-01 4.70926106e-01 1.10007398e-01 -1.86223119e-01
-9.29434001e-02 -2.95386799e-02 -9.11499023e-01 -5.27056098e-01
-2.08898127e-01 4.21693653e-01 3.86982501e-01 8.36297214e-01
1.20718122e-01 1.02452755e+00 -8.93492997e-01 -7.80871987e-01
5.36827028e-01 6.70109391e-01 -4.33809966e-01 2.13419646e-01
-2.57269740e-02 6.90830827e-01 -1.96702063e-01 3.18325251e-01
8.75411451e-01 -2.69430488e-01 3.29426825e-01 -5.35256267e-02
4.77843918e-02 -1.67896509e-01 -8.00271749e-01 1.91098654e+00
-6.07989013e-01 5.05625844e-01 3.59353244e-01 -1.39096439e+00
8.29217196e-01 1.82666451e-01 9.47418928e-01 -7.21338272e-01
5.82565606e-01 4.71686482e-01 2.28971258e-01 -3.13719332e-01
1.37150377e-01 -3.75596315e-01 2.02227607e-02 3.14734668e-01
1.40649036e-01 -4.97019082e-01 3.30413312e-01 -1.40271276e-01
1.33108020e+00 -1.10175706e-01 2.90019780e-01 -3.35671157e-01
6.66907728e-01 6.09578267e-02 2.33325019e-01 3.97579879e-01
-3.65023345e-01 7.80272424e-01 7.74308622e-01 -8.54561329e-01
-1.32718468e+00 -8.69889975e-01 -2.13570744e-01 5.40021241e-01
-1.24508254e-02 3.38886827e-02 -1.02593791e+00 -8.69341910e-01
-9.89125744e-02 1.53316721e-01 -1.03387272e+00 1.16717942e-01
-9.10609007e-01 -1.04584718e+00 3.88891160e-01 4.44064707e-01
2.92270988e-01 -1.22045314e+00 -1.16477740e+00 4.66528118e-01
1.83727637e-01 -1.15259588e+00 -1.48073494e-01 3.15744877e-01
-1.33533847e+00 -1.19209564e+00 -1.16106677e+00 -6.55664265e-01
8.51087570e-01 -3.64732862e-01 1.26039255e+00 3.22000504e-01
-8.62792850e-01 1.57842770e-01 7.03549013e-04 -3.40531647e-01
-6.47463977e-01 2.52634317e-01 -5.24422169e-01 -3.02911699e-01
4.06043567e-02 -4.97467786e-01 -7.88813889e-01 -3.79171744e-02
-1.25136495e+00 1.15254465e-02 7.50080228e-01 7.73455620e-01
7.55021155e-01 -3.11666578e-01 7.44995624e-02 -1.17362809e+00
5.65537333e-01 -1.35668397e-01 -5.85355580e-01 7.41699571e-03
-4.19384331e-01 1.11212172e-01 4.91686642e-01 -2.78506488e-01
-5.13110220e-01 5.46536565e-01 -4.28167880e-01 -4.19506341e-01
-2.91441977e-01 4.44799304e-01 2.39589244e-01 -4.42991644e-01
5.10917544e-01 4.07932959e-02 2.51093984e-01 -4.02867705e-01
2.03472525e-01 2.09955812e-01 5.52373052e-01 -6.61009029e-02
5.45483649e-01 5.83958030e-01 4.33549196e-01 -4.50142086e-01
-4.39053148e-01 -3.58345032e-01 -1.12546670e+00 -3.25435072e-01
1.05912149e+00 -4.14175272e-01 -7.12020278e-01 2.78046012e-01
-1.10974157e+00 -6.21821284e-01 -4.95175928e-01 3.85717303e-01
-6.88127816e-01 3.72735173e-01 -6.45412564e-01 -5.85743248e-01
-6.26339555e-01 -1.32862520e+00 1.13803613e+00 8.56585130e-02
4.29145470e-02 -1.16917837e+00 2.09776983e-01 1.07428022e-01
5.73739946e-01 8.34814668e-01 9.89575982e-01 -8.33843291e-01
-1.56564832e-01 -2.86402971e-01 -4.76105541e-01 2.39127412e-01
-7.68531188e-02 -4.76818591e-01 -7.48128414e-01 -2.28020445e-01
-5.53638451e-02 -1.75959453e-01 9.75650311e-01 6.58776104e-01
1.65518355e+00 8.46649185e-02 -3.35209489e-01 5.37301898e-01
1.62773585e+00 9.13424939e-02 6.17496073e-01 1.53162718e-01
7.41039932e-01 5.68608999e-01 1.60206228e-01 2.61238426e-01
1.96395725e-01 4.71984923e-01 7.90417194e-01 -6.48101807e-01
-1.29027128e-01 4.27625895e-01 -4.44372803e-01 5.71542025e-01
-1.58068225e-01 -5.06722257e-02 -1.23937905e+00 7.44002938e-01
-1.64952612e+00 -3.88765126e-01 -1.84475109e-01 2.24560833e+00
8.42596352e-01 3.93418640e-01 1.73577502e-01 1.32607788e-01
5.61741352e-01 -7.01376945e-02 -5.02473772e-01 -5.55465400e-01
4.42205489e-01 5.91834962e-01 8.47160578e-01 5.16246855e-01
-1.37516081e+00 4.71322775e-01 5.94358349e+00 4.52213556e-01
-1.31426644e+00 2.42061093e-01 1.09657860e+00 -7.19506294e-02
2.73229610e-02 -4.71043617e-01 -1.54238462e-01 2.79362798e-01
9.18123126e-01 2.14651763e-01 1.06399311e-02 6.30257130e-01
6.61171898e-02 -3.18444729e-01 -1.12370896e+00 9.14168417e-01
9.57288891e-02 -1.26113725e+00 -4.29031581e-01 2.21699402e-01
5.82649946e-01 1.37260228e-01 -4.47774604e-02 2.10509419e-01
-1.72940157e-02 -1.46199298e+00 2.21504658e-01 3.39138120e-01
5.45592725e-01 -7.06200540e-01 1.09010947e+00 2.61402488e-01
-8.44231844e-01 2.28176266e-01 -1.90196067e-01 3.93926799e-01
1.13083512e-01 7.31224895e-01 -1.15454268e+00 4.82415795e-01
4.13466930e-01 2.19168991e-01 -6.67185009e-01 1.42067623e+00
2.51378566e-01 1.37788817e-01 -4.29706603e-01 8.67226273e-02
5.09658575e-01 -3.03994981e-03 1.85861707e-01 1.36839199e+00
2.63371736e-01 -8.06584582e-02 2.57049382e-01 7.61048973e-01
-2.13157728e-01 4.47686017e-01 -4.61438596e-01 2.73541331e-01
-4.40951556e-01 1.46386051e+00 -1.56123996e+00 -3.50992501e-01
-6.72716871e-02 9.92964745e-01 2.09361047e-01 -2.69822836e-01
-6.08772993e-01 -1.25013798e-01 -7.27622062e-02 2.09043697e-01
2.56363153e-01 -2.88373232e-01 -4.13342327e-01 -7.59968340e-01
-6.21222630e-02 -4.16360497e-01 2.92928219e-01 -2.66153812e-01
-1.15299559e+00 7.73761690e-01 8.54831636e-02 -7.33452737e-01
-3.55941564e-01 -8.43373775e-01 -5.12609720e-01 7.43368089e-01
-1.57340336e+00 -1.07286000e+00 -4.05235440e-01 3.56409132e-01
4.92041439e-01 3.79957259e-01 7.61434197e-01 5.43547928e-01
-2.00969934e-01 2.17182875e-01 -7.71120042e-02 3.41563940e-01
4.50222820e-01 -1.65192771e+00 3.07310432e-01 4.82533842e-01
-1.98490009e-01 4.70945053e-02 6.98330522e-01 -5.45687258e-01
-1.10703123e+00 -9.62505460e-01 7.86768496e-01 1.92412436e-02
3.07730049e-01 -3.59112859e-01 -9.36468661e-01 4.53336656e-01
1.87934637e-01 6.24287903e-01 5.47959268e-01 -2.71865159e-01
2.04525679e-01 2.12351322e-01 -1.52153337e+00 1.96976095e-01
6.26411676e-01 5.88465594e-02 -2.60404885e-01 5.26813209e-01
2.46537119e-01 -9.76179361e-01 -1.10764313e+00 6.19173646e-01
1.56551838e-01 -1.04229224e+00 9.75198328e-01 -3.36353749e-01
5.09449601e-01 5.19549325e-02 3.53001714e-01 -1.29481053e+00
2.20130697e-01 -1.06268190e-01 -7.51494197e-04 4.58250344e-01
2.31400266e-01 -2.05029145e-01 1.09341073e+00 6.12495303e-01
-2.02844411e-01 -9.90832508e-01 -1.26445413e+00 -4.09471333e-01
3.44325840e-01 -2.36654699e-01 2.48047650e-01 8.10768783e-01
-4.43364263e-01 -9.03720707e-02 -3.54897752e-02 -2.71206647e-01
7.87547708e-01 7.70678744e-02 4.09473360e-01 -1.45254564e+00
-3.94228064e-02 -6.37843013e-01 -8.14880729e-01 -3.94673735e-01
-7.34561980e-02 -1.13128591e+00 2.15785369e-01 -1.65521443e+00
3.43377084e-01 -4.80832666e-01 -2.05487564e-01 5.08960664e-01
7.76028261e-02 7.59629309e-01 2.52468973e-01 -1.82207227e-01
-4.01041955e-01 1.93511084e-01 1.55983639e+00 -2.53809839e-01
-2.75770128e-01 -2.63146106e-02 -7.98179433e-02 6.69878900e-01
8.38747382e-01 -4.89308596e-01 -9.75503400e-02 -2.88188010e-01
-5.12865148e-02 1.46722183e-01 5.50993860e-01 -1.10175252e+00
1.04000978e-01 3.18766117e-01 8.94355893e-01 -6.78621769e-01
1.59781113e-01 -9.08104956e-01 -6.81327134e-02 9.51101840e-01
-3.96663070e-01 -4.13597794e-03 4.32556599e-01 1.62236795e-01
-5.58726192e-02 -4.17011231e-01 1.05746710e+00 -5.88960171e-01
-2.66750067e-01 6.46872580e-01 -2.71718442e-01 -1.22524597e-01
1.06011176e+00 -6.31853864e-02 2.71327287e-01 -6.04589768e-02
-1.18254852e+00 1.29394040e-01 3.22998554e-01 1.20145254e-01
5.38124204e-01 -1.16597342e+00 -7.10409999e-01 7.42365569e-02
-2.58860797e-01 4.03471798e-01 4.37443167e-01 1.13079774e+00
-1.08873928e+00 3.29039127e-01 -4.99082953e-01 -8.89715672e-01
-1.35439920e+00 5.52363396e-01 6.51414156e-01 -9.44864750e-01
-7.71846890e-01 5.90043306e-01 -1.08190939e-01 -6.01584375e-01
1.31924123e-01 -5.79357564e-01 -3.22645515e-01 -5.42144030e-02
4.29102212e-01 -2.24632826e-02 4.37281430e-01 -6.25046670e-01
-3.56479466e-01 3.69180322e-01 -5.39580025e-02 -4.54297289e-02
1.60470259e+00 1.58646435e-01 -2.73677319e-01 2.56831855e-01
1.40051460e+00 -4.10321176e-01 -1.11181092e+00 -8.11213776e-02
1.97952896e-01 -4.64005262e-01 3.17513347e-01 -6.10274911e-01
-1.39466131e+00 1.11693418e+00 1.27355838e+00 4.02408153e-01
1.04072189e+00 8.33829492e-03 8.58121693e-01 -8.72867182e-03
3.32930572e-02 -8.27674448e-01 -1.46169513e-01 3.47007096e-01
6.10972941e-01 -1.44136000e+00 1.47476882e-01 -3.62887502e-01
-3.85966122e-01 1.32233405e+00 4.05684382e-01 -3.87885422e-01
5.41381240e-01 4.36193019e-01 9.89754051e-02 -3.32655579e-01
-1.47771001e-01 -1.47253260e-01 3.77438664e-01 4.87229079e-01
5.71393430e-01 -2.82201394e-02 -5.29137194e-01 3.01004685e-02
-1.85506288e-02 5.11561297e-02 4.64085847e-01 9.58532929e-01
-2.60690063e-01 -1.16329825e+00 -3.50124896e-01 6.34006798e-01
-8.95554841e-01 1.51388064e-01 -3.07513684e-01 9.43603575e-01
2.48822495e-01 2.25027218e-01 3.19223404e-01 2.07414091e-01
2.24123523e-01 -1.53951898e-01 8.77882361e-01 -5.99877059e-01
-9.13252413e-01 2.77848572e-01 -3.47092628e-01 -6.11653566e-01
-5.96573830e-01 -5.63664258e-01 -1.34233415e+00 -5.19284047e-02
5.99991716e-02 -1.92958921e-01 1.08128881e+00 1.03346992e+00
-1.70664400e-01 9.22391653e-01 3.70903075e-01 -1.35303402e+00
-3.08351785e-01 -9.07286346e-01 -4.89138603e-01 4.31087941e-01
3.16682786e-01 -4.78255332e-01 1.65590867e-01 -5.31740971e-02] | [14.436193466186523, -2.5249786376953125] |
0c4849b7-4a3b-48d1-b54d-2fb754a3934e | from-speech-to-speech-translation-to | 2001.06785 | null | https://arxiv.org/abs/2001.06785v3 | https://arxiv.org/pdf/2001.06785v3.pdf | From Speech-to-Speech Translation to Automatic Dubbing | We present enhancements to a speech-to-speech translation pipeline in order to perform automatic dubbing. Our architecture features neural machine translation generating output of preferred length, prosodic alignment of the translation with the original speech segments, neural text-to-speech with fine tuning of the duration of each utterance, and, finally, audio rendering to enriches text-to-speech output with background noise and reverberation extracted from the original audio. We report on a subjective evaluation of automatic dubbing of excerpts of TED Talks from English into Italian, which measures the perceived naturalness of automatic dubbing and the relative importance of each proposed enhancement. | ['Roberto Barra-Chicote', 'Robert Enyedi', 'Umut Isik', 'Ritwik Giri', 'Marcello Federico', 'Hassan Sawaf', 'Arvindh Krishnaswamy'] | 2020-01-19 | from-speech-to-speech-translation-to-1 | https://aclanthology.org/2020.iwslt-1.31 | https://aclanthology.org/2020.iwslt-1.31.pdf | ws-2020-7 | ['speech-to-speech-translation'] | ['speech'] | [ 5.78635156e-01 6.70355618e-01 3.80423754e-01 -4.66175199e-01
-1.55280328e+00 -6.52746022e-01 2.94658571e-01 4.21993807e-02
1.48881171e-02 5.21753907e-01 8.41107547e-01 -3.64352405e-01
5.08912504e-01 -2.26628944e-01 -6.90829694e-01 -3.99062634e-01
4.21755284e-01 4.40739065e-01 3.92291648e-03 -3.42062891e-01
-2.94143289e-01 1.39880404e-01 -7.95033097e-01 7.03405201e-01
4.72211599e-01 9.96449947e-01 2.98827499e-01 1.07859910e+00
2.04375505e-01 1.89056158e-01 -1.05445266e+00 -3.49502146e-01
2.18456581e-01 -7.71962285e-01 -7.76439190e-01 2.55935788e-01
2.46429682e-01 -3.70732874e-01 -1.57926783e-01 6.66094363e-01
7.44666159e-01 7.12738978e-03 3.38304907e-01 -6.46071255e-01
-5.62218189e-01 1.21354365e+00 1.36120841e-01 2.43960515e-01
5.01066685e-01 2.75823832e-01 8.47385466e-01 -9.86265242e-01
6.60126984e-01 1.19046998e+00 4.79793668e-01 4.64569837e-01
-1.67793477e+00 -3.45091194e-01 -6.26499295e-01 -2.66862392e-01
-1.02104986e+00 -1.00118470e+00 6.18268430e-01 -2.32792914e-01
1.38684523e+00 3.77795398e-01 3.15769047e-01 1.66741014e+00
-1.22006245e-01 3.25806111e-01 6.51074350e-01 -7.27810204e-01
1.67240396e-01 3.27884942e-01 -2.53646106e-01 -8.11667666e-02
-1.03427947e+00 4.16625679e-01 -8.44491661e-01 -3.48276831e-02
1.43312037e-01 -1.06491017e+00 -2.56768346e-01 5.71801424e-01
-1.27685916e+00 5.13024032e-01 9.89904441e-03 2.44814053e-01
-4.69382375e-01 7.32221678e-02 8.22349787e-01 5.14229000e-01
5.71660399e-01 5.53867579e-01 -4.82543260e-01 -4.63490188e-01
-1.06625831e+00 1.49991363e-01 6.97656393e-01 1.23348510e+00
3.79173219e-01 5.56224227e-01 -4.90804493e-01 1.07499945e+00
-2.12984756e-01 7.72810280e-01 5.85684299e-01 -1.16273022e+00
7.80881166e-01 -2.37838835e-01 9.38392580e-02 -6.11848772e-01
-2.63417363e-01 -3.96109164e-01 -5.06914258e-01 -3.54331017e-01
-1.25576138e-01 -6.69736207e-01 -4.56151992e-01 1.64073133e+00
2.18947697e-03 -4.36199605e-01 3.05534959e-01 7.72314966e-01
4.22476262e-01 1.23333156e+00 -3.80856395e-01 -6.12361431e-01
1.09642768e+00 -1.32332432e+00 -1.15403140e+00 -2.21653581e-01
1.98292479e-01 -1.22606778e+00 1.39972794e+00 3.65927607e-01
-1.40404212e+00 -8.77015352e-01 -9.91092145e-01 -1.14029154e-01
7.15081692e-02 4.14922953e-01 -4.76393849e-01 6.75147474e-01
-1.18426979e+00 5.45043588e-01 -4.88046885e-01 -1.25285029e-01
-4.93664294e-01 1.48955420e-01 -2.17841983e-01 8.53119969e-01
-1.30090427e+00 6.75041139e-01 3.52549255e-01 -1.01143859e-01
-7.40368545e-01 -4.91899908e-01 -8.39321136e-01 2.70847261e-01
1.10095575e-01 -3.70267272e-01 1.72615612e+00 -1.20381427e+00
-2.03621340e+00 5.12332559e-01 -1.39264151e-01 -9.21507955e-01
2.99405634e-01 -3.10139686e-01 -5.77801347e-01 4.58010197e-01
1.89272724e-02 7.04433858e-01 1.13984132e+00 -7.43297338e-01
-4.40141737e-01 7.01330677e-02 -6.77439749e-01 4.21020567e-01
-3.34769517e-01 4.12143618e-01 -9.55541208e-02 -1.05224264e+00
-1.33573815e-01 -7.87078381e-01 6.01129234e-02 -6.68377757e-01
-5.65485477e-01 2.33997107e-01 8.83443713e-01 -1.26962233e+00
1.22708678e+00 -2.46485066e+00 2.57748604e-01 -1.16692811e-01
-4.86195028e-01 4.93095033e-02 -5.25612772e-01 5.59402347e-01
-1.76747851e-02 5.60246445e-02 -4.11242247e-02 -7.13764846e-01
1.53907567e-01 -2.07379431e-01 -8.72347116e-01 -3.86741646e-02
3.98385137e-01 6.91459179e-01 -6.25497699e-01 -1.76537991e-01
1.50444418e-01 4.93765265e-01 -3.75597268e-01 5.75331092e-01
-3.36205542e-01 5.82878530e-01 2.74438232e-01 3.86713207e-01
9.86666381e-02 7.53101468e-01 -1.84510633e-01 -2.72406846e-01
-1.90924093e-01 1.11660135e+00 -4.41997230e-01 1.44947004e+00
-5.33818662e-01 8.93947780e-01 4.44979429e-01 -3.35199714e-01
1.11491990e+00 9.08733785e-01 7.81569853e-02 -6.84176505e-01
1.47085935e-01 4.63098377e-01 -1.73762575e-01 -4.71842229e-01
8.32533956e-01 -6.08720630e-02 -2.09991083e-01 4.05098766e-01
3.95002753e-01 -9.27026987e-01 1.00970112e-01 2.83241160e-02
7.78029919e-01 1.77021220e-01 -2.28548958e-03 -2.15073396e-02
1.61332726e-01 -1.44719779e-01 8.76641423e-02 2.78720200e-01
-3.53155732e-02 1.17891943e+00 4.04073119e-01 -8.27304833e-03
-1.62707281e+00 -1.03214574e+00 2.35901028e-01 1.24820566e+00
-5.94491422e-01 -4.53782260e-01 -1.53182197e+00 -4.53724921e-01
-7.44958639e-01 1.18633151e+00 -1.23483583e-01 -2.10476354e-01
-8.51810515e-01 -2.79879034e-01 1.08402979e+00 2.15865269e-01
-7.82747045e-02 -1.36259162e+00 -1.69314936e-01 6.00329340e-01
-8.20519984e-01 -1.52752316e+00 -1.32533514e+00 4.17104036e-01
-5.81004620e-01 -1.28078699e-01 -7.74620473e-01 -1.02173316e+00
1.35117183e-02 -3.10537964e-01 8.73581052e-01 -6.97517812e-01
2.10342348e-01 1.24916881e-02 -6.32900596e-01 -2.38517717e-01
-1.46700656e+00 1.98587880e-01 2.98157185e-01 5.88768311e-02
-4.81754877e-02 -6.90630138e-01 -1.19514555e-01 4.92427558e-01
-6.97540522e-01 3.63436580e-01 3.95764321e-01 8.58793199e-01
4.32817757e-01 -3.81209195e-01 6.81795716e-01 -3.53864506e-02
1.12268257e+00 3.48532014e-02 -3.21015686e-01 -1.11366838e-01
-7.25203902e-02 -9.60523561e-02 8.08819473e-01 -5.92645168e-01
-1.03936374e+00 -2.94720381e-02 -5.27953207e-01 -4.06885266e-01
-1.27684250e-01 2.72782058e-01 -2.01750770e-01 5.12266576e-01
1.11339390e+00 2.09067389e-01 -1.02075428e-01 -3.55776727e-01
4.72974926e-01 1.36889696e+00 1.01618695e+00 -4.39575285e-01
3.29865247e-01 -3.26327443e-01 -6.52428865e-01 -1.01042533e+00
-3.87385368e-01 -1.56653807e-01 -4.75216508e-01 -1.98484845e-02
7.36865878e-01 -5.24312258e-01 -2.17358708e-01 2.53948480e-01
-1.62065387e+00 -5.53890228e-01 -4.14741725e-01 3.79461437e-01
-8.81462097e-01 1.74281374e-01 -1.03179812e+00 -8.06244671e-01
-8.48539889e-01 -1.18822515e+00 1.20931005e+00 -1.60259157e-01
-7.68779814e-01 -2.64207870e-01 6.70929179e-02 4.86879408e-01
6.63745821e-01 -1.52438387e-01 8.75692248e-01 -6.96738064e-01
2.79295202e-02 -1.25671159e-02 1.12329006e-01 9.42226887e-01
1.58182457e-01 3.57366055e-02 -1.24054027e+00 5.49406856e-02
2.66994953e-01 -3.32810819e-01 2.54111141e-01 5.21652699e-01
2.50043511e-01 -8.30612123e-01 2.66853303e-01 2.43843958e-01
5.49989462e-01 4.16597039e-01 4.59428221e-01 -1.19312830e-01
4.91892658e-02 8.31473291e-01 5.60796618e-01 2.76415586e-01
-1.98493645e-01 1.06497049e+00 1.16666257e-01 -1.54515386e-01
-4.66246903e-01 -3.45143437e-01 9.88837779e-01 1.31382811e+00
5.56831777e-01 -3.22177500e-01 -6.80711746e-01 6.67524755e-01
-1.06076622e+00 -4.82540190e-01 -3.15829888e-02 1.94308889e+00
1.21462631e+00 3.18590879e-01 2.88752884e-01 1.31689787e-01
9.18731749e-01 -8.61289054e-02 9.14054271e-03 -1.21619189e+00
-2.66934276e-01 1.90461338e-01 3.22744936e-01 1.00269294e+00
-6.51624978e-01 1.02813935e+00 6.73804379e+00 8.82977605e-01
-1.25524008e+00 3.56504202e-01 6.35138988e-01 -1.91248193e-01
-2.12214708e-01 -2.46696413e-01 -4.61193293e-01 3.15904796e-01
1.74871802e+00 -1.59101412e-01 8.01458240e-01 4.59276468e-01
8.83924067e-01 2.40489662e-01 -1.29154134e+00 4.29575980e-01
-9.77681056e-02 -1.05875731e+00 2.95537971e-02 -3.31676811e-01
5.01982152e-01 1.57601446e-01 2.35361755e-01 1.44338861e-01
-3.28634143e-01 -8.39671373e-01 1.30475736e+00 -5.06908521e-02
1.08301640e+00 -4.73528832e-01 5.62488317e-01 2.25525051e-01
-8.14005852e-01 2.32245177e-01 4.22328115e-02 2.24813238e-01
5.06548047e-01 3.23319167e-01 -1.60607827e+00 3.12762916e-01
3.59084457e-01 -3.58000994e-02 -1.07512586e-01 1.19782746e-01
-2.38737121e-01 1.09707248e+00 -4.31942135e-01 1.13144629e-02
1.33754760e-01 1.65077820e-01 1.10873580e+00 1.65708923e+00
6.59936666e-01 -2.30662674e-01 -2.20543921e-01 1.03763509e+00
-9.61868316e-02 3.05194944e-01 -3.83265525e-01 -5.12831807e-01
6.09227419e-01 9.06516850e-01 -3.34991693e-01 -1.08279839e-01
1.04703277e-01 1.34649122e+00 -3.43596667e-01 5.77081442e-01
-6.64899290e-01 -5.77796400e-01 2.00461984e-01 1.28328502e-01
7.69216791e-02 -8.89025629e-02 -3.60892206e-01 -5.47935545e-01
2.73905545e-01 -1.12743533e+00 -1.98665127e-01 -1.23072040e+00
-7.35685229e-01 1.27957273e+00 -3.49877268e-01 -1.06014848e+00
-6.94546103e-01 -8.71850774e-02 -3.99837226e-01 1.34700334e+00
-5.60259044e-01 -1.02976727e+00 3.03747177e-01 -8.02119225e-02
1.22947800e+00 -1.60021693e-01 1.12149346e+00 1.13240890e-01
2.30199061e-02 5.53226948e-01 -1.59626067e-01 3.92023996e-02
9.79314327e-01 -1.02189374e+00 1.06133008e+00 8.55352283e-01
1.22749940e-01 1.28866255e-01 1.14187145e+00 -5.50195932e-01
-7.18979239e-01 -1.07282460e+00 1.21946442e+00 -5.69335632e-02
6.08601689e-01 -5.82981706e-01 -6.64076388e-01 6.13887131e-01
7.65346944e-01 -6.21132076e-01 3.76621187e-01 -2.98113585e-01
-1.02554642e-01 -2.47308865e-01 -8.73476326e-01 6.49588168e-01
3.49460155e-01 -7.90920377e-01 -8.08988631e-01 3.68041694e-02
1.59617507e+00 -6.24550641e-01 -5.57573259e-01 1.46252051e-01
2.83811092e-01 -8.05864871e-01 5.03013551e-01 -8.28958005e-02
5.10072291e-01 -1.36611342e-01 -3.40862185e-01 -1.67725396e+00
2.19667897e-01 -1.46442187e+00 4.12030756e-01 1.42677534e+00
9.02158678e-01 -3.33030522e-03 1.87774435e-01 1.70338497e-01
-6.72802091e-01 -1.07301958e-01 -1.07741189e+00 -6.05086684e-01
-1.39924839e-01 -5.15959978e-01 3.07208568e-01 3.91074002e-01
5.11777461e-01 8.43509018e-01 -7.21251070e-01 1.25770047e-01
1.05409488e-01 -4.23905194e-01 5.05612671e-01 -2.89267570e-01
-6.04366302e-01 -3.03481072e-01 1.77651405e-01 -9.69804704e-01
-1.40465721e-01 -5.23585439e-01 5.38767099e-01 -8.54744732e-01
-3.57526988e-01 3.33459765e-01 3.46558154e-01 2.74291724e-01
1.66098848e-01 2.56855100e-01 2.53146917e-01 -1.24737896e-01
3.39207463e-02 7.03796089e-01 7.57969856e-01 -2.82320499e-01
-6.77423358e-01 1.24986969e-01 -1.82186365e-01 3.80449772e-01
7.22285092e-01 -5.55330813e-01 -9.21358466e-02 -4.89941657e-01
-2.11070338e-03 7.14981437e-01 2.36471873e-02 -7.99690664e-01
-2.19754741e-01 2.55409449e-01 -2.93713093e-01 -6.00954950e-01
6.17160261e-01 -4.75892752e-01 2.57964730e-01 2.13231206e-01
-9.77448881e-01 -1.89595800e-02 4.34103340e-01 1.04649350e-01
-5.11745274e-01 -2.45167509e-01 8.12269092e-01 2.27609247e-01
3.21047127e-01 -4.63160604e-01 -8.02641749e-01 -1.44332305e-01
3.38672996e-01 8.49324167e-02 8.81725401e-02 -8.70694280e-01
-1.06754291e+00 -2.72672772e-01 9.40091088e-02 4.20117021e-01
3.71596009e-01 -1.15621781e+00 -1.06903195e+00 2.76405662e-01
-2.58738488e-01 -3.82796675e-01 -1.45152649e-02 9.22067046e-01
-3.89944226e-01 5.14460087e-01 -8.64438564e-02 -4.03115064e-01
-1.47028339e+00 2.80469328e-01 3.72889757e-01 7.50491023e-02
-4.34061646e-01 6.45369172e-01 -2.88464548e-03 -1.57260537e-01
4.92089063e-01 -5.65703452e-01 3.19639236e-01 -7.88154900e-02
3.35168004e-01 3.58412564e-01 5.43897927e-01 -5.64961672e-01
1.21560499e-01 -6.81929290e-02 2.57222205e-01 -9.85203385e-01
8.92735839e-01 -6.95162952e-01 -4.51749079e-02 5.90034544e-01
1.22883534e+00 3.77124280e-01 -1.03168738e+00 -1.29338130e-01
2.63585355e-02 2.75306106e-01 -1.91237643e-01 -9.84732151e-01
-3.94312382e-01 8.98385525e-01 2.53373623e-01 4.20085073e-01
1.15755737e+00 -1.48195907e-01 1.23449969e+00 3.93749326e-01
2.09331373e-03 -1.48023605e+00 1.81547180e-01 8.09207320e-01
1.45286417e+00 -6.17508233e-01 -7.00728774e-01 -2.05570191e-01
-8.71219575e-01 1.32342994e+00 4.69240807e-02 2.64607102e-01
1.66604400e-01 6.72384024e-01 4.26227719e-01 5.56233406e-01
-7.44580984e-01 2.79980034e-01 2.48521999e-01 6.42111421e-01
2.52511352e-01 6.35074675e-02 -1.28378600e-01 9.16669905e-01
-9.25941765e-01 -3.89559567e-01 6.19359910e-01 1.21735655e-01
-3.74651134e-01 -8.33534062e-01 -6.23891711e-01 -3.34252775e-01
-7.68663228e-01 -6.21748626e-01 -9.06750739e-01 7.91759565e-02
-3.27733278e-01 1.36533892e+00 -4.79076467e-02 -4.58391666e-01
4.86388683e-01 3.61392170e-01 -2.55174432e-02 -5.71370125e-01
-1.03332925e+00 1.09050906e+00 6.09091818e-01 -3.63820076e-01
1.17445312e-01 -3.55865002e-01 -1.18296480e+00 2.45070711e-01
-3.42607737e-01 3.38426650e-01 8.86878788e-01 7.96037555e-01
2.93598413e-01 6.36411071e-01 8.47924232e-01 -1.03289759e+00
-6.98298216e-01 -1.62253344e+00 -3.37370962e-01 2.14085802e-01
7.01029181e-01 5.01072049e-01 -3.71945083e-01 5.47960460e-01] | [14.617754936218262, 6.958740711212158] |
1e0311e3-9ec3-4211-b71e-f7f9c9293a5a | spatial-temporal-sequential-hypergraph-1 | 2201.02435 | null | https://arxiv.org/abs/2201.02435v2 | https://arxiv.org/pdf/2201.02435v2.pdf | Spatial-Temporal Sequential Hypergraph Network for Crime Prediction with Dynamic Multiplex Relation Learning | Crime prediction is crucial for public safety and resource optimization, yet is very challenging due to two aspects: i) the dynamics of criminal patterns across time and space, crime events are distributed unevenly on both spatial and temporal domains; ii) time-evolving dependencies between different types of crimes (e.g., Theft, Robbery, Assault, Damage) which reveal fine-grained semantics of crimes. To tackle these challenges, we propose Spatial-Temporal Sequential Hypergraph Network (ST-SHN) to collectively encode complex crime spatial-temporal patterns as well as the underlying category-wise crime semantic relationships. In specific, to handle spatial-temporal dynamics under the long-range and global context, we design a graph-structured message passing architecture with the integration of the hypergraph learning paradigm. To capture category-wise crime heterogeneous relations in a dynamic environment, we introduce a multi-channel routing mechanism to learn the time-evolving structural dependency across crime types. We conduct extensive experiments on two real-world datasets, showing that our proposed ST-SHN framework can significantly improve the prediction performance as compared to various state-of-the-art baselines. The source code is available at: https://github.com/akaxlh/ST-SHN. | ['Tianyi Chen', 'Xiyue Zhang', 'Liefeng Bo', 'Peng Dai', 'Yong Xu', 'Chao Huang', 'Lianghao Xia'] | 2022-01-07 | spatial-temporal-sequential-hypergraph | https://www.ijcai.org/proceedings/2021/225 | https://www.ijcai.org/proceedings/2021/0225.pdf | ijcai-2021-8 | ['crime-prediction'] | ['miscellaneous'] | [-4.11375239e-02 -4.35641319e-01 -2.46766493e-01 -4.19114798e-01
-1.73548058e-01 -3.27474952e-01 6.02895081e-01 7.54760861e-01
-3.60522956e-01 5.10927141e-01 6.42235279e-01 -3.00929368e-01
-7.18056560e-01 -1.36016774e+00 -3.56992006e-01 -2.40425766e-01
-6.24984086e-01 4.46575969e-01 5.86253583e-01 -3.57497752e-01
4.58968073e-01 3.97029459e-01 -9.78948236e-01 2.93399036e-01
6.52171791e-01 3.55722010e-01 4.05686460e-02 4.12934572e-01
-2.29242265e-01 1.00183320e+00 -2.63805598e-01 -5.96699655e-01
-1.20473549e-01 -7.36076534e-02 -9.42267418e-01 -2.27351174e-01
-1.64740324e-01 -1.05659835e-01 -9.90617394e-01 1.00975382e+00
2.33182326e-01 2.18522146e-01 6.19849145e-01 -1.30417955e+00
-6.10289931e-01 5.58805883e-01 -8.47248018e-01 8.24056506e-01
3.71365339e-01 1.93195611e-01 8.76247883e-01 -3.69435042e-01
6.97522879e-01 1.38202047e+00 6.39582515e-01 1.99564651e-01
-1.00340581e+00 -7.19495654e-01 5.85338771e-01 8.18539679e-01
-1.45997882e+00 -1.86862610e-02 9.53709960e-01 -6.96724832e-01
1.18522537e+00 1.72547206e-01 5.72061777e-01 1.20347130e+00
3.00406963e-01 5.27275264e-01 7.89402843e-01 2.19760939e-01
8.63873661e-02 -5.62702596e-01 5.40450096e-01 6.03038728e-01
1.85079977e-01 -1.09271698e-01 -5.41955173e-01 -3.06739837e-01
5.23128450e-01 5.93194187e-01 1.95395812e-01 3.26315016e-01
-6.72189593e-01 8.85162115e-01 6.81845367e-01 4.56016928e-01
-3.26254159e-01 1.52832136e-01 7.63141096e-01 1.19100705e-01
7.95499861e-01 -1.09883778e-01 -2.87497759e-01 -3.56436014e-01
-6.26047671e-01 1.88190848e-01 4.90537822e-01 5.84561408e-01
7.21495211e-01 -3.09561431e-01 -2.79448450e-01 9.28250968e-01
2.78928727e-01 2.72621125e-01 -9.61007848e-02 -2.76795566e-01
1.29905581e+00 9.37980235e-01 -4.84070212e-01 -1.85300767e+00
-7.88374543e-01 -1.95681617e-01 -1.28670442e+00 -5.06736279e-01
2.51711428e-01 -1.66118577e-01 -8.17380190e-01 1.73345339e+00
4.90729779e-01 9.70504820e-01 -4.41388845e-01 7.00712204e-01
7.60545850e-01 7.61145055e-01 4.78583544e-01 -1.93009637e-02
1.21656668e+00 -4.19270873e-01 -5.86654246e-01 -3.42791855e-01
5.51535726e-01 -2.54827410e-01 6.08437002e-01 -2.33232066e-01
-7.79954076e-01 -8.17084238e-02 -3.67071569e-01 7.44493753e-02
-6.02370560e-01 -6.83258653e-01 6.43992126e-01 3.97044867e-01
-7.23171592e-01 4.74439174e-01 -9.62120593e-01 -4.05528188e-01
7.38583148e-01 6.40926063e-02 -2.63570696e-01 -4.12726909e-01
-1.55936635e+00 5.40867805e-01 4.33175772e-01 2.63110787e-01
-6.00696623e-01 -6.60465837e-01 -9.81089890e-01 1.35522157e-01
2.89468050e-01 -1.83716401e-01 3.03441674e-01 -3.18634771e-02
-5.69424808e-01 6.13771498e-01 -3.08174461e-01 -4.89595421e-02
2.67783940e-01 1.82148397e-01 -8.23148549e-01 -9.02324566e-04
4.23074812e-01 -3.01110327e-01 2.10632354e-01 -8.71881545e-01
-7.05650091e-01 -7.54520774e-01 2.32643679e-01 1.83241606e-01
-6.97651803e-01 3.03421319e-01 -5.38347125e-01 -7.49216199e-01
7.78749073e-03 -5.92141628e-01 -4.24700379e-01 -5.45036852e-01
-6.75989985e-01 -3.48560661e-01 7.89463162e-01 -5.78015208e-01
1.99064887e+00 -2.04376626e+00 1.02460682e-01 4.25838262e-01
2.89684802e-01 2.72918373e-01 -2.96001703e-01 1.10817516e+00
1.17785297e-01 1.68950647e-01 -5.31329513e-01 -1.99780196e-01
-1.16770744e-01 2.66926259e-01 -1.65256798e-01 5.34217179e-01
2.59305716e-01 8.63139391e-01 -1.33520532e+00 -5.41749418e-01
1.77891836e-01 5.21356642e-01 -3.54078680e-01 -1.43485181e-02
2.15921700e-01 5.16744971e-01 -9.29177642e-01 5.92659593e-01
6.62933350e-01 -4.13779259e-01 1.87624738e-01 3.71236533e-01
5.02527803e-02 2.10058704e-01 -1.12571287e+00 1.44076371e+00
-2.79673427e-01 4.05378520e-01 2.99513955e-02 -1.27659428e+00
7.39963412e-01 1.70797318e-01 4.93680090e-01 -1.15727437e+00
-2.45667949e-01 -7.65267992e-03 -2.37549916e-01 -7.32583404e-01
3.49871844e-01 1.93205193e-01 -3.72131050e-01 6.86381280e-01
-3.06592703e-01 3.67371410e-01 3.92411262e-01 4.76141244e-01
1.78080010e+00 -5.79294622e-01 -4.71174344e-02 -1.41365469e-01
4.11909729e-01 -1.32808805e-01 9.28013325e-01 6.97425306e-01
-1.48918614e-01 3.02736431e-01 8.59144092e-01 -9.49486792e-01
-7.27949321e-01 -9.43213820e-01 -1.18147083e-01 1.08767867e+00
4.09503907e-01 -5.23618877e-01 -4.80475992e-01 -5.67804813e-01
5.09790331e-02 2.18074813e-01 -6.22264326e-01 -9.98924896e-02
-9.84640121e-01 -1.36773539e+00 5.36216021e-01 3.70762974e-01
5.09445012e-01 -1.15857565e+00 -2.16763869e-01 4.67876315e-01
-3.04084629e-01 -1.09933436e+00 -5.00002146e-01 -4.56170887e-01
-5.32923639e-01 -1.32436335e+00 -1.26368046e-01 -5.05743802e-01
4.51695085e-01 1.98573768e-01 1.06796873e+00 5.28949499e-01
-5.84792435e-01 1.67178109e-01 -6.33014143e-01 7.88155943e-02
2.73535371e-01 2.08301619e-01 -2.56021023e-01 3.33774805e-01
5.45337558e-01 -1.10443676e+00 -8.74992907e-01 -1.33226365e-02
-1.05354774e+00 4.19088453e-02 -1.15817422e-02 5.27135253e-01
3.76802623e-01 5.68980098e-01 5.09528816e-01 -1.24585235e+00
9.45895433e-01 -1.30065775e+00 -4.38213080e-01 2.92478859e-01
-3.77321243e-01 -4.32150722e-01 5.65534472e-01 -2.10657269e-01
-9.58832979e-01 -3.35611761e-01 -1.04994364e-01 8.63419175e-02
-2.95182824e-01 7.99747705e-01 1.17701188e-01 4.27005351e-01
2.42514506e-01 2.15493008e-01 -8.96596491e-01 -3.13192189e-01
6.74862862e-02 5.88309288e-01 2.46088475e-01 -8.08325112e-01
7.53710032e-01 5.90993226e-01 2.54641950e-01 -6.51725590e-01
-6.21662021e-01 -6.28762722e-01 -7.16335058e-01 -2.20049456e-01
1.11793256e+00 -6.06072247e-01 -6.62788212e-01 8.72328818e-01
-1.52798462e+00 -5.98920345e-01 4.66243893e-01 2.23776728e-01
8.02363232e-02 2.67249525e-01 -1.07492578e+00 -8.10697258e-01
-3.04339677e-01 -7.81114340e-01 1.00934470e+00 1.25455081e-01
1.14468746e-02 -1.50107145e+00 4.03701127e-01 3.49134952e-01
2.24857107e-01 6.37671411e-01 1.27828455e+00 -4.10560608e-01
-5.60361445e-01 -3.33535932e-02 -6.46360040e-01 -4.17820036e-01
2.08380848e-01 -2.83201456e-01 -4.31748837e-01 -1.44989505e-01
-5.76568484e-01 -4.00088541e-02 8.86715949e-01 1.68975651e-01
1.41718221e+00 -5.15846252e-01 -5.14527977e-01 6.67020261e-01
1.48564982e+00 1.66638628e-01 6.45376801e-01 2.75649905e-01
1.13859713e+00 8.78395021e-01 4.31989998e-01 7.99164712e-01
1.08694625e+00 7.71065831e-01 4.66047317e-01 -3.25026363e-02
1.17942505e-01 -4.98070657e-01 -1.09345920e-01 7.64964461e-01
-2.25511730e-01 -3.30480665e-01 -1.48173964e+00 9.57955897e-01
-2.23299837e+00 -1.42686105e+00 -6.09883904e-01 1.70931983e+00
5.16150296e-01 8.88112858e-02 1.08134151e-01 -1.94226811e-03
9.48433399e-01 6.09026670e-01 -4.73817199e-01 -4.36416864e-01
1.74121894e-02 1.82867691e-01 4.44646448e-01 6.76679969e-01
-1.12917769e+00 1.11463058e+00 5.23888969e+00 1.10570395e+00
-8.82723629e-01 4.27998871e-01 8.33630145e-01 -1.20451517e-01
-3.39780867e-01 -6.64623082e-02 -5.10010123e-01 9.50226545e-01
7.01150000e-01 -2.05257252e-01 6.02723300e-01 5.29457033e-02
3.19804251e-01 2.08070159e-01 -5.96395075e-01 8.85198176e-01
-2.65461147e-01 -1.43854952e+00 -1.76677674e-01 4.62828092e-02
5.65933406e-01 2.37502366e-01 -6.15853071e-02 1.50195509e-01
6.63473845e-01 -1.19260907e+00 2.97878921e-01 6.00733221e-01
5.82576811e-01 -8.76758099e-01 3.85998875e-01 4.70400095e-01
-1.92638958e+00 -3.54193091e-01 -1.34127185e-01 -3.21476966e-01
5.43350041e-01 8.37314069e-01 -3.21548820e-01 7.77704418e-01
8.46414745e-01 1.21930444e+00 -4.39697057e-01 7.98775077e-01
-3.43050331e-01 7.67674327e-01 -1.04336411e-01 8.34890753e-02
4.64687169e-01 -1.98332340e-01 4.08071190e-01 1.60298204e+00
1.08762212e-01 7.72623181e-01 2.75086075e-01 7.21769154e-01
7.52866268e-02 -1.82665944e-01 -7.54224777e-01 1.53268874e-01
8.43365252e-01 1.08346939e+00 -8.22720289e-01 -3.98875173e-08
-4.56332415e-01 6.61487281e-01 8.03758860e-01 4.77307558e-01
-6.95900738e-01 -5.12539208e-01 8.14745843e-01 3.21587861e-01
-2.95906700e-02 -2.33842522e-01 -2.41535023e-01 -1.12987888e+00
1.94423124e-01 -2.72027373e-01 9.07337904e-01 -1.75838605e-01
-1.73620701e+00 7.67371655e-01 1.04868293e-01 -8.57026875e-01
1.15953192e-01 -2.23883286e-01 -1.10905540e+00 6.73774302e-01
-1.58390129e+00 -1.31912398e+00 -7.47399852e-02 1.11555088e+00
4.15278256e-01 -1.29935429e-01 5.62372804e-01 7.40739942e-01
-8.57845008e-01 2.83336729e-01 -2.27697879e-01 6.04201555e-01
4.15566824e-02 -8.27481925e-01 6.47948623e-01 1.03977704e+00
6.29088888e-03 3.95760357e-01 2.21444070e-01 -9.73987281e-01
-1.15591788e+00 -1.60840428e+00 1.15774751e+00 -4.20463026e-01
1.04353893e+00 -6.03538513e-01 -1.07426071e+00 5.28161347e-01
-1.41268179e-01 2.01986432e-01 5.28104126e-01 5.57345808e-01
-5.55702209e-01 -1.01030774e-01 -1.03757048e+00 4.54438925e-01
1.55984139e+00 -7.01019883e-01 -2.41352111e-01 5.98211348e-01
6.01028025e-01 -2.10815743e-01 -8.62260103e-01 2.93015063e-01
8.89208466e-02 -7.92650104e-01 9.76083279e-01 -7.81171441e-01
7.44453430e-01 1.41156882e-01 2.01794934e-02 -1.16375387e+00
-6.94007337e-01 -2.27847502e-01 -8.75409469e-02 1.22490191e+00
2.57063806e-01 -7.91407466e-01 6.17850244e-01 6.66841507e-01
1.57002643e-01 -9.66676652e-01 -1.38268948e+00 -5.51532328e-01
9.45449919e-02 -6.57188058e-01 1.15451205e+00 1.39325690e+00
3.37544590e-01 3.20572764e-01 -5.97951949e-01 4.50200856e-01
7.35793173e-01 1.10053740e-01 3.21862280e-01 -1.19275117e+00
-4.31706011e-02 -5.89149475e-01 -8.57277095e-01 -4.55395013e-01
3.42925489e-01 -9.17227924e-01 -5.84638298e-01 -1.70949221e+00
6.07757449e-01 -8.21851492e-01 -4.50987339e-01 5.71125627e-01
-3.62697244e-01 1.57775749e-02 7.63982758e-02 3.95236090e-02
-9.00878668e-01 3.76565218e-01 9.96206522e-01 -3.43377411e-01
-1.78353176e-01 -1.16650090e-01 -3.42257261e-01 5.59268296e-01
8.14270973e-01 -7.30482221e-01 -6.68744206e-01 -1.16196644e+00
5.50716400e-01 4.77386922e-01 4.75329816e-01 -6.60743833e-01
6.69603944e-01 -7.08485305e-01 -5.51186241e-02 -3.95897031e-01
2.81834185e-01 -6.50707006e-01 1.59345075e-01 4.60307002e-01
-1.65671751e-01 3.15274388e-01 1.17717266e-01 8.67154956e-01
-2.99405247e-01 2.53654689e-01 2.61073053e-01 -9.34067741e-03
-7.28362024e-01 1.15177941e+00 -1.35570228e-01 3.28630358e-01
9.69558060e-01 4.36303467e-02 -7.77204275e-01 -1.70215532e-01
-5.98593652e-01 6.87484920e-01 -5.06906630e-03 6.66851580e-01
8.35903406e-01 -1.31030953e+00 -9.80953455e-01 2.33451519e-02
8.14211220e-02 -2.51243357e-02 8.47646058e-01 6.87987030e-01
-3.17259461e-01 3.42246026e-01 -1.01478636e-01 -2.57269800e-01
-1.11475074e+00 3.97432536e-01 1.46669894e-03 -7.76264250e-01
-7.58182287e-01 7.81143308e-01 4.46665362e-02 -6.83925927e-01
-2.67691407e-02 3.28916192e-01 -6.36588693e-01 -7.76844323e-02
5.53300619e-01 6.09956086e-01 -2.49979228e-01 -9.00518119e-01
-5.69565296e-01 5.85686564e-01 5.77463210e-02 1.41864508e-01
1.96372700e+00 -5.11964113e-02 -5.18444002e-01 1.35466322e-01
1.13527668e+00 -4.53306437e-01 -8.57892573e-01 -4.35788006e-01
2.52427548e-01 -9.21477735e-01 -2.05706194e-01 -4.27738190e-01
-1.44597304e+00 8.77151430e-01 1.81327939e-01 4.57127154e-01
1.21613145e+00 1.79068729e-01 1.08342946e+00 2.83949692e-02
6.58955038e-01 -1.09139359e+00 -7.35712498e-02 7.54600823e-01
6.28593266e-01 -1.01405525e+00 -1.76232234e-01 -6.39024496e-01
-4.53032851e-01 7.16256440e-01 6.12442553e-01 -2.75836796e-01
1.07150555e+00 1.08815774e-01 -4.77129161e-01 -7.74602532e-01
-8.42557907e-01 -1.45427302e-01 3.27916592e-02 6.48079872e-01
2.23054335e-01 3.68045390e-01 -4.42128181e-01 5.70418537e-01
1.52224645e-01 -3.16570342e-01 1.57754853e-01 7.57272720e-01
-1.20819859e-01 -1.24946976e+00 -5.92897460e-02 6.55120730e-01
-4.82174993e-01 -1.72822610e-01 -2.16436952e-01 4.34476048e-01
3.11308146e-01 1.19070053e+00 1.45831347e-01 -5.53537011e-01
4.63350177e-01 -5.68087041e-01 9.48486105e-02 -7.01244295e-01
-7.08659112e-01 -4.86958146e-01 -4.44904715e-03 -8.03016961e-01
-6.47130609e-02 -7.90481985e-01 -1.26243496e+00 -7.67386079e-01
3.59126061e-01 -9.35299769e-02 2.14766577e-01 1.05184162e+00
4.54717606e-01 5.24458528e-01 7.41099894e-01 -3.53287607e-01
1.91386685e-01 -8.21823537e-01 -7.98917234e-01 8.94306302e-01
2.89240658e-01 -5.71563303e-01 1.99945364e-02 -3.64400238e-01] | [6.682809829711914, 2.0773050785064697] |
6b369295-1b2e-4664-b830-f4ea568cdef7 | layoutdiffusion-improving-graphic-layout | 2303.11589 | null | https://arxiv.org/abs/2303.11589v1 | https://arxiv.org/pdf/2303.11589v1.pdf | LayoutDiffusion: Improving Graphic Layout Generation by Discrete Diffusion Probabilistic Models | Creating graphic layouts is a fundamental step in graphic designs. In this work, we present a novel generative model named LayoutDiffusion for automatic layout generation. As layout is typically represented as a sequence of discrete tokens, LayoutDiffusion models layout generation as a discrete denoising diffusion process. It learns to reverse a mild forward process, in which layouts become increasingly chaotic with the growth of forward steps and layouts in the neighboring steps do not differ too much. Designing such a mild forward process is however very challenging as layout has both categorical attributes and ordinal attributes. To tackle the challenge, we summarize three critical factors for achieving a mild forward process for the layout, i.e., legality, coordinate proximity and type disruption. Based on the factors, we propose a block-wise transition matrix coupled with a piece-wise linear noise schedule. Experiments on RICO and PubLayNet datasets show that LayoutDiffusion outperforms state-of-the-art approaches significantly. Moreover, it enables two conditional layout generation tasks in a plug-and-play manner without re-training and achieves better performance than existing methods. | ['Dongmei Zhang', 'Jian-Guang Lou', 'Shizhao Sun', 'Jiaqi Guo', 'Junyi Zhang'] | 2023-03-21 | null | null | null | null | ['layout-design'] | ['computer-vision'] | [-6.34243935e-02 -2.15354972e-02 6.77881464e-02 -3.43985379e-01
-6.00041330e-01 -7.09928215e-01 7.07058668e-01 2.59973351e-02
-5.69283068e-02 6.33810759e-01 3.82589340e-01 -3.48830283e-01
-9.89070982e-02 -8.86927247e-01 -1.04645848e+00 -8.11499834e-01
9.53630656e-02 5.44609129e-01 -4.37997244e-02 2.23463234e-02
3.43246877e-01 2.06264839e-01 -1.20579171e+00 5.14826477e-01
1.12134862e+00 7.13146508e-01 1.79284737e-01 6.14710212e-01
-3.03401142e-01 8.49187911e-01 -6.91004574e-01 -5.13498485e-01
1.33496851e-01 -5.81059456e-01 -4.39932197e-01 1.91218570e-01
3.66870701e-01 -3.49123240e-01 -2.47688442e-01 8.89159441e-01
5.04738986e-01 9.23945233e-02 1.04105115e+00 -1.22801888e+00
-1.35270000e+00 1.06794107e+00 -7.67530262e-01 -4.43867356e-01
1.98985517e-01 3.38132620e-01 1.17440343e+00 -9.60279882e-01
7.45277524e-01 1.20588434e+00 6.18979216e-01 2.44872242e-01
-1.51796472e+00 -5.03925383e-01 7.34659314e-01 -4.24436256e-02
-1.40868902e+00 3.61129572e-03 1.04646266e+00 -5.38988471e-01
4.54249173e-01 2.28325367e-01 6.38017774e-01 1.44769013e+00
9.58905891e-02 9.10131454e-01 8.80078256e-01 -4.32224236e-02
5.62969267e-01 -2.77126610e-01 -4.62625206e-01 8.09362888e-01
2.27619737e-01 -5.31908333e-01 -4.26083446e-01 -7.33418539e-02
9.30021822e-01 -8.00761506e-02 2.13095516e-01 -4.45197046e-01
-1.08451235e+00 7.35247314e-01 6.27135694e-01 1.50107771e-01
-3.73353958e-01 4.49992746e-01 1.71207428e-01 6.09905273e-02
5.13011396e-01 4.83676642e-01 -2.57855151e-02 -1.57429174e-01
-1.05050290e+00 5.71883798e-01 7.39302397e-01 1.15955424e+00
6.20275617e-01 -6.42960146e-03 -5.80973148e-01 7.69495368e-01
2.69811064e-01 4.32920098e-01 1.39792692e-02 -7.10506797e-01
7.09320009e-01 4.41489786e-01 -2.03390121e-02 -1.08403122e+00
-1.36184573e-01 -4.78125453e-01 -1.32986784e+00 2.64843013e-02
3.20009589e-01 -4.22988534e-01 -1.19678175e+00 1.67561400e+00
2.12117642e-01 1.82817042e-01 -4.27620620e-01 5.32588840e-01
7.60953903e-01 8.49500239e-01 -3.78977284e-02 7.34487474e-02
1.16862524e+00 -1.17585886e+00 -8.16555738e-01 -6.15238100e-02
4.82481241e-01 -8.20993662e-01 1.26521373e+00 5.28404355e-01
-1.21646380e+00 -5.69828093e-01 -8.93551111e-01 -2.75085181e-01
-2.44274363e-01 3.27353925e-01 7.17241228e-01 4.73501146e-01
-1.07162189e+00 5.13805747e-01 -6.54609621e-01 1.40852546e-02
6.17925882e-01 -9.85848904e-02 5.44892475e-02 -4.09882516e-02
-7.49986351e-01 1.86110482e-01 1.15019068e-01 3.34828794e-01
-6.79430127e-01 -8.18165660e-01 -7.69825459e-01 2.04604864e-01
4.66952503e-01 -1.08311439e+00 9.69020486e-01 -3.22690964e-01
-1.56385863e+00 3.03515345e-01 -1.59150898e-01 -3.13887298e-01
1.07545996e+00 -2.79386014e-01 -9.64139998e-02 -2.48258427e-01
1.71166360e-01 7.22793877e-01 9.82010663e-01 -1.57170928e+00
-5.54647326e-01 8.44187140e-02 -1.18699528e-01 3.29019874e-01
-2.47773007e-01 -5.71423948e-01 -7.83207595e-01 -8.92977893e-01
-3.85836768e-03 -8.96420836e-01 -4.03948516e-01 -5.35866152e-03
-9.06428814e-01 -2.18282014e-01 5.51861048e-01 -5.86317241e-01
1.53571165e+00 -2.12398958e+00 2.31592298e-01 4.80173528e-01
3.76731068e-01 -4.57331352e-02 -2.78496832e-01 6.28822684e-01
3.07921618e-01 4.78333443e-01 -2.25911647e-01 -7.12569296e-01
4.17544365e-01 1.44650593e-01 -3.17228466e-01 4.53630555e-03
2.46400863e-01 1.13252747e+00 -9.96264756e-01 -3.19085866e-01
6.61379844e-02 5.24813712e-01 -9.62954700e-01 3.08010787e-01
-5.36432207e-01 4.53835934e-01 -4.47172076e-01 4.82498795e-01
8.54836583e-01 -2.42410511e-01 2.21294180e-01 -8.32076594e-02
-2.79664546e-01 1.81313083e-01 -1.19778562e+00 1.95808315e+00
-3.40736598e-01 3.57486844e-01 -1.18660681e-01 -6.47072077e-01
1.13681996e+00 -1.26755774e-01 2.00732678e-01 -7.43119717e-01
1.07554477e-02 5.12238853e-02 -1.50841892e-01 -2.83721298e-01
7.50511229e-01 3.47057074e-01 -3.66680413e-01 4.48807389e-01
-3.38558346e-01 -2.14005798e-01 4.51638609e-01 3.89167130e-01
1.24812722e+00 2.46330053e-01 -3.12186718e-01 -1.48013622e-01
-2.78446712e-02 -5.01697958e-01 5.07360935e-01 1.12913918e+00
3.49796534e-01 9.80561912e-01 1.17831945e+00 -2.65372276e-01
-1.19316602e+00 -1.30684698e+00 4.10380810e-01 8.28505695e-01
-4.56131212e-02 -7.50387013e-01 -1.04209888e+00 -5.67207098e-01
2.28600521e-02 6.51976585e-01 -9.80575085e-01 -2.95149647e-02
-7.00406730e-01 -5.27875066e-01 4.06685948e-01 6.74999475e-01
4.48975861e-01 -1.19955683e+00 -7.62675852e-02 1.84448451e-01
-2.36954633e-02 -8.56961489e-01 -7.05816388e-01 5.48609234e-02
-4.97817367e-01 -5.22330046e-01 -1.05175877e+00 -9.95116353e-01
1.14303422e+00 -5.69550991e-02 1.08139110e+00 1.61265165e-01
-1.24594189e-01 -1.84393004e-01 -3.18073869e-01 -6.57066107e-02
-1.63111359e-01 6.22879922e-01 -4.50008273e-01 1.55551881e-01
-3.59351546e-01 -8.26927781e-01 -8.92534018e-01 -6.13944829e-02
-9.90931451e-01 4.19642031e-01 8.78617465e-01 8.84216309e-01
6.00712538e-01 1.89681560e-01 3.01306665e-01 -1.32574439e+00
1.13856435e+00 -5.27385950e-01 -4.53478694e-01 2.27858052e-01
-4.23826098e-01 2.77897835e-01 9.54154134e-01 -5.21171510e-01
-1.16873848e+00 5.09497859e-02 -2.99630076e-01 -3.89394969e-01
-1.60706609e-01 7.19163418e-01 -3.20390373e-01 5.19928873e-01
3.53539169e-01 1.15773603e-01 -4.57103193e-01 -7.05110312e-01
9.21028495e-01 1.45602539e-01 5.25016546e-01 -9.64711070e-01
1.18775773e+00 3.75772685e-01 -5.75138703e-02 -6.63292527e-01
-5.05911827e-01 9.27704200e-02 -6.70151055e-01 -1.67437047e-01
7.51510978e-01 -6.59736454e-01 -7.74348497e-01 6.84356093e-01
-1.17378426e+00 -7.62246251e-01 -2.97579080e-01 -2.77496278e-01
-3.63002390e-01 1.89218923e-01 -7.02319384e-01 -7.09458292e-01
-1.54164121e-01 -1.15028119e+00 1.13220620e+00 1.64657742e-01
-2.10006103e-01 -9.40580368e-01 3.67069687e-03 -2.16702133e-01
3.20843458e-01 4.32922542e-01 1.22876728e+00 -6.08661026e-02
-8.81454766e-01 9.28853601e-02 -2.96113223e-01 9.15020034e-02
2.19348788e-01 4.76722419e-01 -4.40601319e-01 9.30066127e-03
-3.87430638e-01 -1.04780896e-02 1.00570846e+00 4.16567355e-01
1.52382112e+00 -6.52222931e-01 -1.44278795e-01 9.86048520e-01
1.15498567e+00 3.04365963e-01 8.16661656e-01 2.00970978e-01
1.30877626e+00 5.39591610e-01 2.58039594e-01 6.52133167e-01
6.48408771e-01 3.16508174e-01 2.29180381e-01 -4.09522653e-01
-1.62771493e-01 -8.82856131e-01 1.01889940e-02 9.76969838e-01
1.68226764e-01 -5.53336382e-01 -8.00835490e-01 4.28290248e-01
-2.18251061e+00 -8.33428442e-01 -2.91945279e-01 1.87931609e+00
7.89817214e-01 3.56205374e-01 8.85312855e-02 -1.14016436e-01
5.43073356e-01 4.70593631e-01 -4.69296962e-01 -4.45297718e-01
-1.70933992e-01 1.56558920e-02 3.31452012e-01 3.99248153e-01
-8.95660758e-01 1.12232697e+00 5.86844158e+00 9.69987988e-01
-8.83090198e-01 -2.99569875e-01 1.03733277e+00 3.17517370e-02
-8.51538062e-01 2.47166716e-02 -5.23779929e-01 9.47140396e-01
2.55556017e-01 8.75092074e-02 6.30929649e-01 7.53666282e-01
1.44042566e-01 1.48787796e-02 -1.06243873e+00 9.77681279e-01
-2.26095635e-02 -1.48346555e+00 3.10472161e-01 1.60856508e-02
1.03511512e+00 -5.49193561e-01 6.92071199e-01 2.11136192e-01
6.42278433e-01 -1.05058634e+00 1.13739216e+00 5.96434414e-01
6.77569509e-01 -9.68075097e-01 1.94056407e-01 2.40293533e-01
-1.38498116e+00 1.38761953e-01 -2.28347197e-01 1.43774003e-01
4.87118810e-01 7.25466788e-01 -4.79441434e-01 5.77375293e-01
5.77049136e-01 7.36854613e-01 -6.89629793e-01 1.07088089e+00
-6.59946978e-01 5.55328250e-01 -1.07987314e-01 -1.66912854e-01
4.75292385e-01 -5.93348503e-01 2.47399867e-01 1.25264442e+00
5.47556937e-01 -1.77377433e-01 8.71257335e-02 1.37138128e+00
-3.62498283e-01 -9.36433971e-02 -4.72135186e-01 -9.13884416e-02
6.16583765e-01 1.17343819e+00 -9.58115101e-01 -3.29846144e-01
-1.31383874e-02 1.12527347e+00 4.43416297e-01 7.15166748e-01
-1.06783772e+00 -4.64343607e-01 4.74417418e-01 1.45381317e-01
7.10612118e-01 -4.25597996e-01 -7.67991364e-01 -1.00876832e+00
2.91361213e-01 -7.70059943e-01 -2.51828078e-02 -7.70661533e-01
-1.42397642e+00 6.33413851e-01 -1.89983934e-01 -1.08756006e+00
-5.71927167e-02 -2.32133582e-01 -8.67595434e-01 7.51707196e-01
-1.12754226e+00 -1.35112500e+00 -3.41615230e-01 1.65309697e-01
4.74696189e-01 2.06889465e-01 4.08754706e-01 3.46139014e-01
-6.93289518e-01 7.17195928e-01 2.93196887e-01 2.83495963e-01
8.01517427e-01 -1.47010708e+00 1.27136815e+00 8.52933168e-01
2.94098377e-01 8.83400619e-01 6.15716994e-01 -1.04620111e+00
-1.27875090e+00 -1.13018823e+00 7.48259544e-01 -2.97207594e-01
6.47682548e-01 -1.06811202e+00 -8.35471749e-01 4.50640738e-01
2.98060536e-01 -4.79442894e-01 3.43622118e-01 3.69463950e-01
-3.89501452e-01 5.95632456e-02 -6.71930909e-01 1.05141330e+00
1.39813495e+00 -3.83387595e-01 -1.15409046e-01 2.28810191e-01
5.86625755e-01 -6.50250196e-01 -4.27211791e-01 2.69024298e-02
5.61291814e-01 -9.40709174e-01 8.42882931e-01 -2.07204074e-01
7.57109463e-01 -3.44607890e-01 4.27560031e-01 -1.53580558e+00
-6.03672504e-01 -1.03586829e+00 -1.82698518e-01 1.63826954e+00
5.39978445e-01 -3.74141127e-01 9.19718385e-01 4.63399917e-01
-2.12582052e-01 -8.89496565e-01 -3.37583661e-01 -6.97025120e-01
6.43435121e-02 -2.71772891e-01 9.45355713e-01 6.66630149e-01
-4.64918375e-01 4.05726105e-01 -6.73977554e-01 -1.33362383e-01
6.22150362e-01 5.29130772e-02 9.99343455e-01 -1.05617201e+00
-2.89791703e-01 -6.60872400e-01 1.31551117e-01 -1.49193776e+00
-7.30539635e-02 -6.64519846e-01 2.60139942e-01 -2.00768018e+00
8.46130401e-02 -7.03520536e-01 6.73290202e-03 2.73864090e-01
-3.54478061e-01 -8.03442746e-02 2.79593438e-01 6.98505864e-02
-7.67298400e-01 7.89608896e-01 1.43694520e+00 -1.86170757e-01
-3.31486344e-01 -1.75606132e-01 -8.89897466e-01 5.51562488e-01
5.89482069e-01 -5.06717682e-01 -6.25179112e-01 -8.26759398e-01
6.24302804e-01 -2.95841575e-01 -1.57227612e-03 -7.96897113e-01
4.86068070e-01 -3.18473801e-02 5.09320498e-01 -9.84285951e-01
2.39955336e-01 -3.65239084e-01 2.76622027e-01 1.76682353e-01
-5.24664879e-01 3.87472510e-01 -1.22913979e-01 6.52856171e-01
8.11250061e-02 1.20257713e-01 2.62923449e-01 3.93038196e-03
-2.63603777e-01 3.35734844e-01 -4.62456554e-01 5.84669709e-02
7.38821566e-01 -1.38917670e-01 -4.07217473e-01 -3.53361011e-01
-5.70583403e-01 2.62081295e-01 5.51855683e-01 5.68705440e-01
5.30104220e-01 -1.61085200e+00 -6.08751118e-01 1.90823168e-01
-1.37306958e-01 5.22590816e-01 5.39117157e-01 5.48905790e-01
-6.85969055e-01 9.65826958e-02 -5.44899553e-02 -3.50174367e-01
-9.77766931e-01 5.10350108e-01 -2.96512038e-01 -4.92836177e-01
-6.84672356e-01 1.01207435e+00 5.05572259e-01 -1.72723904e-01
3.96840304e-01 -7.55536914e-01 1.22245722e-01 2.12494940e-01
3.59031856e-01 1.91167638e-01 -1.28540531e-01 -8.69275331e-02
2.91392077e-02 4.10537720e-01 -3.44893992e-01 -2.78758943e-01
1.41683185e+00 3.14160064e-02 -8.86821300e-02 5.06519675e-01
1.01692104e+00 4.42776643e-02 -1.74824071e+00 7.06641227e-02
-1.00282714e-01 -5.12158632e-01 -3.58197451e-01 -7.52590418e-01
-1.01892316e+00 8.15018058e-01 2.49408092e-02 4.60508108e-01
8.78050089e-01 -1.03430167e-01 8.62309098e-01 2.11008772e-01
7.19053224e-02 -1.34474456e+00 4.24332678e-01 5.05395532e-01
9.78234470e-01 -7.07834125e-01 -1.70731977e-01 -4.32695836e-01
-7.89831817e-01 7.82256782e-01 4.96182889e-01 -3.14071566e-01
4.01967138e-01 5.79521477e-01 -1.36484995e-01 -1.00088879e-01
-7.48964548e-01 -5.31479083e-02 4.18574154e-01 6.29342914e-01
4.41286087e-01 1.34712592e-01 -2.57037729e-01 5.97076297e-01
-3.82744491e-01 -2.83473045e-01 2.57458866e-01 8.62713873e-01
1.22320931e-02 -1.34321427e+00 -1.58249035e-01 5.16205013e-01
-1.27881989e-01 -3.38177711e-01 -5.72408676e-01 5.97682059e-01
1.72969908e-01 6.00807667e-01 1.89079106e-01 -3.48944485e-01
4.07312691e-01 -2.93533951e-01 2.43586540e-01 -4.97756541e-01
-6.49035692e-01 3.66029412e-01 -8.85819346e-02 -3.41222227e-01
2.42424801e-01 -6.35459781e-01 -9.85469818e-01 -6.14164412e-01
1.11394562e-01 2.29473971e-02 6.16841197e-01 5.42237163e-01
5.94341576e-01 1.06448388e+00 5.82441151e-01 -9.02556956e-01
-3.31818938e-01 -6.65158808e-01 -5.12134135e-01 5.59786379e-01
7.87937865e-02 -4.19603795e-01 -1.65655598e-01 2.12979823e-01] | [11.32269287109375, -0.21739383041858673] |
9ac5fb1f-b040-475a-8c01-715a56f00262 | audioldm-text-to-audio-generation-with-latent | 2301.12503 | null | https://arxiv.org/abs/2301.12503v2 | https://arxiv.org/pdf/2301.12503v2.pdf | AudioLDM: Text-to-Audio Generation with Latent Diffusion Models | Text-to-audio (TTA) system has recently gained attention for its ability to synthesize general audio based on text descriptions. However, previous studies in TTA have limited generation quality with high computational costs. In this study, we propose AudioLDM, a TTA system that is built on a latent space to learn the continuous audio representations from contrastive language-audio pretraining (CLAP) latents. The pretrained CLAP models enable us to train LDMs with audio embedding while providing text embedding as a condition during sampling. By learning the latent representations of audio signals and their compositions without modeling the cross-modal relationship, AudioLDM is advantageous in both generation quality and computational efficiency. Trained on AudioCaps with a single GPU, AudioLDM achieves state-of-the-art TTA performance measured by both objective and subjective metrics (e.g., frechet distance). Moreover, AudioLDM is the first TTA system that enables various text-guided audio manipulations (e.g., style transfer) in a zero-shot fashion. Our implementation and demos are available at https://audioldm.github.io. | ['Mark D. Plumbley', 'Wenwu Wang', 'Danilo Mandic', 'Xubo Liu', 'Xinhao Mei', 'Yi Yuan', 'Zehua Chen', 'Haohe Liu'] | 2023-01-29 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 1.77062616e-01 -2.35574573e-01 1.43250227e-01 -1.45128548e-01
-1.29276955e+00 -5.93357444e-01 5.63052595e-01 -2.23033905e-01
1.28746152e-01 2.90203482e-01 5.18564284e-01 -1.01417176e-01
1.60277024e-01 -7.69664228e-01 -7.40164518e-01 -7.24363863e-01
5.78660443e-02 2.25323230e-01 -3.79016176e-02 -5.16333431e-02
-1.23357080e-01 6.55466244e-02 -1.72310936e+00 5.43318212e-01
7.96322882e-01 1.04128397e+00 3.22895259e-01 8.34834278e-01
-2.30346382e-01 5.94620526e-01 -7.22096145e-01 -2.67618626e-01
3.64518240e-02 -5.14408290e-01 -5.70418298e-01 -1.73169121e-01
5.14971197e-01 -4.22153234e-01 -4.54285949e-01 7.34259129e-01
9.21954453e-01 2.02142373e-01 6.87308431e-01 -1.49097967e+00
-6.71825290e-01 8.82636905e-01 -1.24825455e-01 -2.68567353e-01
5.12830377e-01 3.73336613e-01 1.23204565e+00 -1.29132974e+00
1.74084559e-01 1.51011169e+00 5.20009875e-01 5.41726232e-01
-1.51137948e+00 -9.89409745e-01 -1.37777656e-01 2.39565432e-01
-1.52038288e+00 -8.10597897e-01 9.25941169e-01 -4.73411918e-01
8.12218666e-01 4.12957877e-01 6.31018400e-01 1.42741430e+00
8.90028924e-02 9.66977954e-01 8.54864180e-01 -5.45464516e-01
2.65091777e-01 -2.45492101e-01 -3.57761294e-01 4.52665657e-01
-5.00934124e-01 -3.43328193e-02 -1.15220177e+00 -1.67932555e-01
8.09666812e-01 -3.57490659e-01 -2.78810889e-01 -4.36530858e-02
-1.24698663e+00 7.37918496e-01 -1.05291940e-02 1.82647314e-02
-2.99616724e-01 5.05761325e-01 6.78043783e-01 5.01761556e-01
4.44977701e-01 2.35649481e-01 3.07127628e-02 -5.81871331e-01
-1.08442700e+00 1.43126890e-01 4.73680168e-01 1.15518749e+00
3.92918050e-01 7.43161500e-01 -4.50780481e-01 1.03290856e+00
3.99017394e-01 6.85132861e-01 7.52011359e-01 -1.01637888e+00
5.29987037e-01 1.21695198e-01 -1.10713042e-01 -8.79582107e-01
7.31633678e-02 -3.58530462e-01 -9.08266723e-01 2.14331448e-01
6.06256947e-02 -2.84203682e-02 -7.81162143e-01 1.76059556e+00
1.31182969e-01 4.57415551e-01 7.87361525e-03 7.78599918e-01
7.50250578e-01 1.21986043e+00 -2.31160402e-01 -1.65857702e-01
1.17546952e+00 -1.21584404e+00 -9.63104248e-01 1.26311645e-01
4.87688750e-01 -1.32396281e+00 1.84111357e+00 6.35494113e-01
-1.31016946e+00 -9.07670557e-01 -1.00306165e+00 -3.13163638e-01
-2.98200175e-03 2.85122901e-01 2.45618626e-01 5.48838913e-01
-1.09740138e+00 5.37020087e-01 -9.36269224e-01 -7.34914094e-02
7.16296658e-02 1.58369571e-01 -3.41874287e-02 1.64516956e-01
-1.13477504e+00 7.93376416e-02 1.24027766e-01 -5.59027977e-02
-1.39555013e+00 -9.12343383e-01 -8.12877953e-01 2.17409074e-01
3.68396163e-01 -7.15078712e-01 1.46198833e+00 -6.43410206e-01
-2.31590343e+00 3.90316606e-01 -2.42199656e-02 -4.47468013e-01
5.03837466e-01 -6.50731504e-01 -5.14376402e-01 2.29096651e-01
5.36608435e-02 8.68780732e-01 1.32171810e+00 -8.48834455e-01
-3.53288144e-01 2.73421407e-01 -1.47080123e-01 1.64848447e-01
-6.44797385e-01 -2.98193227e-02 -4.18824762e-01 -1.20709753e+00
-1.02492124e-01 -9.32590842e-01 3.28092635e-01 2.12752640e-01
-5.11947751e-01 -2.02704623e-01 9.29320991e-01 -5.31498313e-01
1.50141716e+00 -2.51250267e+00 3.35721493e-01 -8.04956481e-02
3.46750468e-02 2.29118899e-01 -3.33723396e-01 7.99153268e-01
2.19814871e-02 -1.59721973e-03 -3.74260321e-02 -6.96854115e-01
4.75559086e-01 -8.83706734e-02 -8.90353262e-01 6.35290593e-02
3.79153229e-02 7.32570171e-01 -8.72006476e-01 -5.10348141e-01
2.50560105e-01 7.48242736e-01 -9.23515499e-01 4.88659769e-01
-3.10082883e-01 4.97932047e-01 -4.46650684e-02 4.08689529e-01
2.75062919e-01 2.85206493e-02 -1.23006932e-01 -3.01183015e-01
-2.32900977e-01 7.16308713e-01 -1.16225696e+00 2.18767452e+00
-9.47417855e-01 7.69855857e-01 -9.46963131e-02 -4.14006352e-01
9.93736029e-01 7.79647171e-01 2.98176825e-01 -3.29692125e-01
3.23183686e-02 3.34606707e-01 -2.51036376e-01 -3.14943343e-01
5.93648195e-01 -8.79297557e-04 -1.82351172e-01 5.10831237e-01
3.15907747e-01 -3.19841355e-01 2.73995008e-02 1.75526991e-01
8.67469370e-01 2.84624219e-01 -1.19583711e-01 -5.32645807e-02
2.83671618e-01 -5.27449369e-01 2.70599067e-01 5.02134383e-01
3.03580552e-01 7.69168973e-01 1.17670447e-01 -6.62319660e-02
-1.11151516e+00 -1.41222918e+00 8.21877122e-02 1.29522836e+00
-3.24971527e-01 -9.63114321e-01 -7.94683397e-01 -1.03430249e-01
-2.78712600e-01 8.78722668e-01 -2.13354096e-01 -3.94436598e-01
-4.09402043e-01 1.29576130e-02 9.71712887e-01 4.08514917e-01
4.27771449e-01 -1.02769804e+00 -1.82384536e-01 4.30352151e-01
-4.20455247e-01 -8.47042859e-01 -1.13619661e+00 -1.15181208e-01
-7.09680796e-01 -3.41877759e-01 -8.74485791e-01 -8.04967105e-01
8.00506845e-02 4.03808728e-02 8.45025063e-01 -5.18031240e-01
-1.66448295e-01 2.80113667e-01 -3.70096117e-01 -2.07341477e-01
-5.79409242e-01 2.56830871e-01 4.78836536e-01 3.12653154e-01
-3.98653708e-02 -1.11209035e+00 -5.85540473e-01 3.03113997e-01
-9.93819833e-01 5.28258502e-01 3.54938209e-01 6.80248022e-01
7.14558601e-01 2.44931225e-02 6.63576245e-01 -3.52457166e-01
8.42405558e-01 -2.48415172e-01 -4.59710568e-01 2.73205992e-02
-2.17886329e-01 -4.05969210e-02 9.21218693e-01 -8.00914526e-01
-8.67173254e-01 -2.26298407e-01 -3.02609950e-01 -8.76817226e-01
-8.13064873e-02 4.11183417e-01 -3.75900120e-01 5.06035089e-01
6.58630013e-01 3.82410675e-01 -2.50603110e-01 -7.40837514e-01
5.09865165e-01 9.73934352e-01 6.82555377e-01 -8.34891081e-01
8.74831378e-01 1.30658790e-01 -2.48481244e-01 -9.74300563e-01
-7.30259538e-01 -2.65058368e-01 -3.61708254e-01 -3.80211830e-01
6.19058669e-01 -1.03092206e+00 -6.01207495e-01 6.10606432e-01
-9.66090322e-01 -6.38370454e-01 -4.84031796e-01 7.42282271e-01
-9.31263447e-01 1.71463728e-01 -7.92896092e-01 -7.64806092e-01
-5.32926559e-01 -1.15048599e+00 1.25355947e+00 -3.11921507e-01
-4.62112665e-01 -6.67831779e-01 1.25358403e-01 1.29227743e-01
4.71832961e-01 8.84288251e-02 8.29888344e-01 1.22781098e-02
-6.11011088e-01 2.08482683e-01 2.72619635e-01 3.74233961e-01
3.87001663e-01 7.07740523e-03 -1.35952568e+00 -4.86951202e-01
-1.30244210e-01 -5.11167288e-01 5.50447285e-01 3.34838867e-01
1.24900877e+00 -5.93210638e-01 2.39072368e-01 8.99271071e-01
9.25163686e-01 1.57169163e-01 3.92079741e-01 -7.03388751e-02
7.34193265e-01 2.03730375e-01 5.32181025e-01 6.93147898e-01
-3.29704629e-03 1.08198798e+00 4.18479703e-02 1.17607936e-01
-5.31061947e-01 -7.40981221e-01 9.96065497e-01 1.86159945e+00
2.28088722e-01 -3.13382119e-01 -7.86851525e-01 4.98711973e-01
-1.60091901e+00 -7.44877219e-01 1.41935989e-01 2.05140209e+00
1.34213960e+00 2.45405361e-01 2.22648934e-01 5.53431511e-01
6.87286258e-01 1.67311877e-01 -5.16088426e-01 -3.92865181e-01
1.78212389e-01 2.74821073e-01 -2.10663944e-01 5.25321960e-01
-8.17378819e-01 8.77841294e-01 5.41806555e+00 1.48051715e+00
-1.27435863e+00 2.54013389e-01 1.51244208e-01 -4.08000916e-01
-5.20660639e-01 -2.26757213e-01 -6.06396496e-01 5.27894497e-01
1.31491959e+00 -3.33424091e-01 7.74711609e-01 7.56355643e-01
3.82410020e-01 5.68719506e-01 -1.30925977e+00 1.18245018e+00
-1.50989115e-01 -1.24322450e+00 5.06130278e-01 -1.09771915e-01
6.63525939e-01 -2.96313763e-01 5.11235595e-01 5.52137017e-01
4.41481061e-02 -9.17951167e-01 1.15593588e+00 5.02861142e-01
1.50258470e+00 -8.40817273e-01 2.31214374e-01 1.43891796e-01
-1.52373660e+00 2.31559381e-01 -1.66652024e-01 -1.51435295e-02
3.14426392e-01 4.19783682e-01 -7.13617742e-01 4.74378675e-01
6.27391577e-01 6.05787039e-01 -1.80388525e-01 8.68862689e-01
-2.80307978e-01 1.22096062e+00 -3.25626463e-01 3.03419083e-01
2.21335426e-01 -8.92124027e-02 7.67984569e-01 1.29679883e+00
7.74008930e-01 -3.00776184e-01 2.08312452e-01 9.95808363e-01
-2.46883780e-01 3.70279431e-01 -4.53678340e-01 -4.17292595e-01
7.33376503e-01 9.55038249e-01 -3.05051327e-01 -2.24690780e-01
-1.23225190e-01 1.05036366e+00 -1.70508176e-01 2.76830554e-01
-1.03313172e+00 -7.92951703e-01 7.11688936e-01 1.64684579e-01
1.32989898e-01 -3.47867101e-01 1.88049935e-02 -1.13589609e+00
-4.39832285e-02 -9.44003105e-01 4.51641493e-02 -1.06418180e+00
-1.22146606e+00 8.29248726e-01 5.70860058e-02 -1.71169853e+00
-3.99069875e-01 -1.66255757e-01 -4.76589292e-01 8.73312652e-01
-1.24631560e+00 -1.23558187e+00 -2.29005650e-01 7.55941212e-01
8.38941991e-01 -3.66258860e-01 1.02745330e+00 5.50090194e-01
-2.59014785e-01 1.00090086e+00 1.57900214e-01 8.96666292e-03
1.02034819e+00 -1.12404358e+00 6.26587391e-01 5.39301991e-01
3.54641795e-01 6.21884644e-01 6.59085870e-01 -3.20735782e-01
-1.28272069e+00 -1.28677738e+00 5.71382105e-01 -1.97732132e-02
6.91351712e-01 -8.74434352e-01 -9.14456129e-01 5.62777817e-01
3.56299698e-01 -3.50390226e-01 8.24034393e-01 7.24642444e-03
-4.28275496e-01 -3.05243850e-01 -6.30671561e-01 8.38485718e-01
9.27434802e-01 -1.12142193e+00 -4.49469775e-01 1.35270394e-02
1.31418002e+00 -4.70829040e-01 -8.97657394e-01 1.51967509e-02
5.73237002e-01 -6.32154405e-01 9.19175029e-01 -1.20545194e-01
4.82297540e-01 -4.88274157e-01 -4.02291596e-01 -1.35169065e+00
-1.74753234e-01 -1.25924885e+00 -2.67473966e-01 1.75981402e+00
1.41641721e-01 -3.75033319e-01 3.12932551e-01 -1.44207314e-01
-5.18285751e-01 -4.83879566e-01 -9.53606486e-01 -1.07268989e+00
-8.60262141e-02 -7.82618880e-01 8.06824744e-01 7.51472592e-01
-6.63499534e-02 5.32441974e-01 -7.49290705e-01 2.20279247e-01
5.10169089e-01 2.39547729e-01 9.24232304e-01 -8.23837638e-01
-6.64257586e-01 -2.78373361e-01 -2.22863719e-01 -1.28060389e+00
-7.99336284e-03 -1.15713108e+00 6.87256455e-02 -1.08324540e+00
-1.97425127e-01 -2.93007970e-01 -1.99797958e-01 4.18553919e-01
2.37839162e-01 3.71209443e-01 4.11388487e-01 1.89424902e-01
-3.80243540e-01 1.11648226e+00 1.25681591e+00 -2.18132839e-01
-4.32974011e-01 -1.04087465e-01 -1.95101872e-01 6.00724936e-01
1.05521595e+00 -5.28595090e-01 -6.72860622e-01 -5.80664158e-01
1.12156965e-01 2.52650917e-01 1.99962139e-01 -1.38987303e+00
1.62069991e-01 -2.36240253e-02 -5.78889921e-02 -6.50469005e-01
8.25987756e-01 -5.09523273e-01 3.16066355e-01 2.36782134e-01
-8.16290736e-01 -8.54428485e-02 3.39384913e-01 3.55683833e-01
-5.02529383e-01 -7.03155696e-02 5.91984272e-01 3.47761095e-01
-1.83621749e-01 3.60793501e-01 -5.33434510e-01 1.46241039e-01
3.80465865e-01 1.52534112e-01 -7.44483471e-02 -6.81860030e-01
-7.55832016e-01 -1.81631610e-01 1.72950208e-01 4.31735903e-01
6.98868036e-01 -1.99031663e+00 -7.89541662e-01 3.79190415e-01
1.09412216e-01 4.19846289e-02 4.01394665e-01 4.51480478e-01
-2.92177320e-01 3.67111117e-01 -2.61016220e-01 -7.48855591e-01
-1.31091714e+00 4.24521923e-01 -5.56134917e-02 -4.59619015e-02
-7.05840051e-01 6.98489845e-01 5.15494108e-01 -3.35521609e-01
6.66437685e-01 -4.89071250e-01 2.85543352e-01 -4.71653137e-03
6.21650994e-01 3.88004452e-01 -1.09923400e-01 -3.77200961e-01
4.27581444e-02 4.76959020e-01 2.40162238e-01 -6.06046200e-01
1.24608123e+00 -9.65655074e-02 2.36183181e-01 1.04507875e+00
1.30450058e+00 3.59065622e-01 -1.34276712e+00 -3.10401708e-01
-4.83968824e-01 -5.75966716e-01 2.10279956e-01 -4.34749126e-01
-8.23959947e-01 1.48690057e+00 6.29337013e-01 5.22027984e-02
1.23110557e+00 -2.92258263e-01 1.15764248e+00 1.95597664e-01
2.16456205e-01 -9.34198201e-01 6.67780161e-01 4.57898647e-01
1.30227745e+00 -5.53222299e-01 -5.05133450e-01 -1.39395505e-01
-6.23254180e-01 1.13453591e+00 3.43415231e-01 6.79592118e-02
4.86891359e-01 3.85915846e-01 1.05159514e-01 2.06634983e-01
-1.05186200e+00 7.46844485e-02 3.52124482e-01 4.37817276e-01
5.19749343e-01 1.57404155e-01 1.77443296e-01 5.73599696e-01
-7.68488109e-01 -4.20916788e-02 3.30248743e-01 5.94220638e-01
-2.08359241e-01 -1.25512099e+00 -4.47526634e-01 5.54929711e-02
-3.81725162e-01 -3.06227684e-01 2.66558584e-02 2.60879576e-01
-1.32440478e-01 9.35161352e-01 2.47319154e-02 -6.21662617e-01
3.14378321e-01 2.78530091e-01 3.20477813e-01 -6.25372350e-01
-4.38906103e-01 6.48893237e-01 -2.58689560e-02 -4.84518647e-01
-8.91485810e-02 -5.95057487e-01 -1.17997456e+00 -3.67258042e-01
-2.43708923e-01 2.21281901e-01 5.83296597e-01 4.47418660e-01
3.83771598e-01 9.51451719e-01 8.47112536e-01 -9.72564816e-01
-4.52289104e-01 -1.14010918e+00 -6.95195138e-01 2.87524849e-01
3.84505540e-01 -4.00377750e-01 -2.18858615e-01 3.22657019e-01] | [15.445962905883789, 5.655033111572266] |
f0e48b52-d8b6-4955-b43e-74cce507536b | recommending-on-graphs-a-comprehensive-review | 2212.12230 | null | https://arxiv.org/abs/2212.12230v2 | https://arxiv.org/pdf/2212.12230v2.pdf | Recommending on graphs: a comprehensive review from a data perspective | Recent advances in graph-based learning approaches have demonstrated their effectiveness in modelling users' preferences and items' characteristics for Recommender Systems (RSS). Most of the data in RSS can be organized into graphs where various objects (e.g., users, items, and attributes) are explicitly or implicitly connected and influence each other via various relations. Such a graph-based organization brings benefits to exploiting potential properties in graph learning (e.g., random walk and network embedding) techniques to enrich the representations of the user and item nodes, which is an essential factor for successful recommendations. In this paper, we provide a comprehensive survey of Graph Learning-based Recommender Systems (GLRSs). Specifically, we start from a data-driven perspective to systematically categorize various graphs in GLRSs and analyze their characteristics. Then, we discuss the state-of-the-art frameworks with a focus on the graph learning module and how they address practical recommendation challenges such as scalability, fairness, diversity, explainability and so on. Finally, we share some potential research directions in this rapidly growing area. | ['Jon Atle Gulla', 'Peng Liu', 'Lemei Zhang'] | 2022-12-23 | null | null | null | null | ['network-embedding'] | ['methodology'] | [-1.99801654e-01 2.46333942e-01 -8.31349075e-01 -4.63381737e-01
1.91583801e-02 -4.68181670e-01 2.51564324e-01 8.03110898e-01
3.19600105e-01 4.85216588e-01 5.17425954e-01 -5.42403579e-01
-7.20464826e-01 -1.08791602e+00 -3.48758072e-01 -4.00756270e-01
-5.98950982e-01 3.52975279e-01 1.03087472e-02 -4.65135008e-01
1.51574090e-01 3.63358349e-01 -1.40976810e+00 2.76043236e-01
9.45333779e-01 9.16815460e-01 -1.48654744e-01 3.83528411e-01
-3.36482942e-01 9.41965461e-01 -2.17607141e-01 -7.75216818e-01
1.11977331e-01 -2.80964017e-01 -6.01558924e-01 -1.83072109e-02
2.40255922e-01 1.05655601e-03 -7.27706313e-01 7.59818852e-01
3.33127558e-01 5.23055494e-01 8.73426497e-01 -1.59054625e+00
-1.22024500e+00 1.13277137e+00 -4.08569604e-01 -1.52820840e-01
5.06217778e-01 -5.63852072e-01 1.85873961e+00 -6.02623582e-01
3.49922359e-01 1.23768330e+00 5.11211872e-01 1.76757112e-01
-1.01630771e+00 -3.92463446e-01 7.97053754e-01 3.57446939e-01
-1.11528301e+00 6.37292415e-02 8.71030092e-01 -3.16629976e-01
4.25380439e-01 7.03597009e-01 9.07530487e-01 5.78527451e-01
-2.44658981e-02 9.39932942e-01 5.55705190e-01 -4.00109589e-01
2.84713238e-01 3.23160648e-01 7.15181112e-01 7.16139436e-01
7.04409957e-01 -1.88364923e-01 -4.11048025e-01 -3.65622073e-01
6.28953815e-01 4.98385668e-01 -6.94084615e-02 -9.67337370e-01
-6.18722618e-01 1.15351331e+00 8.93003464e-01 1.03524879e-01
-3.07684332e-01 -1.33244023e-01 2.60160323e-02 4.33913738e-01
5.06024718e-01 4.86003071e-01 -2.92505175e-01 7.05451667e-01
-5.13927341e-01 -3.26287970e-02 1.08027196e+00 1.06307268e+00
7.92373300e-01 1.00141592e-01 -2.01221928e-01 7.95840323e-01
9.52087939e-01 1.65122584e-01 -7.61774555e-02 -2.44293854e-01
1.84464559e-01 9.01857138e-01 -2.05969721e-01 -1.51403141e+00
-4.75804329e-01 -8.01782787e-01 -8.99473488e-01 -3.35427463e-01
-5.32457083e-02 -4.70237918e-02 -4.37094510e-01 1.30159450e+00
3.48140121e-01 2.52937019e-01 -2.50142485e-01 7.88429201e-01
1.40691519e+00 4.94464040e-01 2.02023163e-01 -2.52885640e-01
1.01847565e+00 -1.10633314e+00 -6.49729729e-01 9.47985202e-02
6.16392255e-01 -3.73047322e-01 1.00284207e+00 1.62320390e-01
-8.44361126e-01 -4.86092120e-01 -8.43354583e-01 1.91840127e-01
-5.85892081e-01 -3.92464586e-02 1.19587672e+00 7.73917079e-01
-1.15565991e+00 5.51212132e-01 -3.49800617e-01 -3.40388447e-01
4.00708050e-01 6.30542994e-01 -2.82568671e-02 -1.76171973e-01
-1.42834496e+00 4.81341600e-01 -4.99973893e-02 -3.31450790e-01
-2.54604310e-01 -6.48031950e-01 -7.57702291e-01 4.12702739e-01
4.61480469e-01 -8.50700617e-01 6.29682302e-01 -6.60557508e-01
-1.15377140e+00 2.77435213e-01 1.88112706e-01 -3.84732693e-01
-2.05904305e-01 6.48384318e-02 -8.80693793e-01 -4.66533393e-01
-4.80430126e-01 -5.40412962e-02 4.64051247e-01 -1.35880327e+00
-7.16278672e-01 -3.15656304e-01 6.63294792e-01 3.26328516e-01
-7.48721004e-01 -1.59861013e-01 -4.76464868e-01 -6.09747171e-01
-1.40112013e-01 -9.00230706e-01 -5.43390989e-01 -3.17310661e-01
-4.67035621e-01 -4.50659007e-01 3.88908744e-01 -1.55068278e-01
2.03077149e+00 -1.85264707e+00 1.73109949e-01 9.08913732e-01
7.21170425e-01 3.28012258e-01 -3.32309842e-01 9.64915156e-01
1.58235252e-01 3.29257458e-01 5.22024691e-01 2.97694914e-02
2.31514215e-01 1.69442430e-01 6.46022856e-02 4.25469905e-01
-5.53915918e-01 1.07268047e+00 -1.05701935e+00 -1.91651493e-01
3.74913692e-01 7.10107803e-01 -7.32616842e-01 2.29863405e-01
-9.97536853e-02 2.15951905e-01 -8.21087599e-01 4.14178431e-01
4.20234174e-01 -7.07793355e-01 7.98821092e-01 -4.73259568e-01
2.41510734e-01 5.81003368e-01 -1.30557895e+00 1.08968794e+00
-5.54516733e-01 1.83114275e-01 -2.58699626e-01 -1.06349838e+00
1.10548019e+00 -7.49788731e-02 7.04594672e-01 -4.63427663e-01
1.09398356e-02 -8.27244893e-02 1.56924292e-01 -9.80896130e-02
6.20372057e-01 4.82911378e-01 1.40746787e-01 7.85343230e-01
-1.03362966e-02 5.30505896e-01 1.63527757e-01 7.12253332e-01
8.26994956e-01 -3.88836831e-01 6.81071997e-01 -2.20288962e-01
6.18624628e-01 -5.22722781e-01 -1.01876212e-02 7.12271094e-01
3.68198514e-01 1.00521501e-02 3.83035153e-01 -5.06496847e-01
-4.82643992e-01 -7.42926300e-01 2.95926481e-01 1.67514992e+00
2.72300512e-01 -9.96452928e-01 -1.72406673e-01 -1.03382206e+00
4.37402964e-01 6.64456725e-01 -6.96603715e-01 -3.51469904e-01
1.55197093e-02 -5.92519879e-01 -2.62390584e-01 4.36774701e-01
-3.30068171e-01 -7.94599175e-01 3.68996471e-01 1.00863308e-01
2.38932461e-01 -5.81685007e-01 -5.57271779e-01 -1.79248556e-01
-1.03539681e+00 -1.18337667e+00 -3.41769934e-01 -7.34356105e-01
8.73542488e-01 9.21674907e-01 1.57419288e+00 5.23170829e-01
8.85214210e-02 7.84710109e-01 -8.96713734e-01 -2.92748868e-01
-1.58636309e-02 2.66852885e-01 9.84790623e-02 2.63320625e-01
2.07672864e-01 -7.49490440e-01 -8.66234124e-01 5.83542764e-01
-6.51504457e-01 -2.07096770e-01 5.25868833e-01 5.14837921e-01
7.07573116e-01 5.25657870e-02 6.71307325e-01 -1.68207324e+00
1.04868662e+00 -9.52545702e-01 -3.11379582e-01 7.49006927e-01
-1.16583502e+00 -6.39317334e-02 6.35710239e-01 -3.41734231e-01
-5.56705058e-01 -3.61378640e-01 3.31184082e-02 7.37954602e-02
2.35443547e-01 1.08198762e+00 -2.12923869e-01 -3.71783912e-01
6.03849113e-01 -2.56958604e-01 -3.06596071e-01 -5.62227964e-01
9.98183608e-01 7.47873425e-01 -2.19066218e-01 -3.25245708e-01
6.91356421e-01 3.74921151e-02 1.35796726e-01 -6.56289935e-01
-1.07205534e+00 -8.24945807e-01 -5.02727747e-01 -4.18789148e-01
2.21565917e-01 -7.68169165e-01 -8.22029054e-01 -3.91229808e-01
-3.57129127e-01 -5.69176041e-02 -3.65322083e-01 6.94635391e-01
-1.20034963e-01 2.60286778e-01 -4.51709867e-01 -9.31678712e-01
-4.64856833e-01 -6.83810174e-01 3.94888490e-01 4.94088799e-01
-2.08696024e-03 -1.64786911e+00 1.24267697e-01 3.13564658e-01
6.11496806e-01 -1.11655734e-01 1.07261539e+00 -1.25516069e+00
-4.52441812e-01 -3.30955267e-01 -2.85410404e-01 -1.15955263e-01
3.77712846e-01 -1.49865627e-01 -3.52756798e-01 -5.97652674e-01
-7.78198600e-01 4.88557331e-02 6.00539446e-01 4.88887817e-01
1.16678476e+00 -5.78742921e-01 -5.28132677e-01 4.83197302e-01
1.29833293e+00 -1.03145009e-02 4.13802356e-01 -1.06427353e-02
1.09029341e+00 6.08983994e-01 6.92949593e-01 6.46844923e-01
8.32568347e-01 7.27195144e-01 6.35453284e-01 -3.03265393e-01
-1.39731407e-01 -5.69163024e-01 2.24598020e-01 1.22469139e+00
-4.49877024e-01 -8.45061541e-01 -3.66224915e-01 3.79872252e-03
-2.19433141e+00 -6.51367605e-01 -5.45522094e-01 2.39284015e+00
2.05306336e-01 -3.35237265e-01 6.00217044e-01 2.02262495e-02
8.76381159e-01 3.88573408e-01 -5.45163095e-01 -3.61866474e-01
1.67641625e-01 9.29357260e-02 4.59792465e-01 4.10305738e-01
-9.40465987e-01 8.73673320e-01 6.47800303e+00 5.87341487e-01
-6.52278244e-01 -3.46771866e-01 4.62255150e-01 9.71166939e-02
-9.16547000e-01 -4.92203832e-02 -5.13858140e-01 1.50934637e-01
8.75122368e-01 -4.44132596e-01 7.67006099e-01 9.24320042e-01
5.96200936e-02 5.45003712e-01 -9.75187421e-01 8.77220333e-01
1.73601940e-01 -1.51999342e+00 4.25984502e-01 4.03550863e-01
8.69725645e-01 -1.31723598e-01 2.32528582e-01 3.94412369e-01
6.79303348e-01 -9.16813135e-01 -6.39654603e-03 5.94809413e-01
5.10085106e-01 -7.76079535e-01 6.78995550e-01 1.16682118e-02
-1.58362401e+00 -1.64381653e-01 -7.76100457e-01 -3.08879539e-02
2.59093698e-02 7.23015308e-01 -4.33100671e-01 1.17431116e+00
4.07762408e-01 1.32563901e+00 -6.28663778e-01 1.20957458e+00
-3.57716173e-01 8.29686642e-01 1.08217850e-01 -4.49495971e-01
-5.12392595e-02 -6.35495901e-01 1.44150376e-01 1.09960353e+00
3.50098342e-01 3.03652465e-01 3.35997224e-01 2.40741581e-01
-3.67250860e-01 8.67899418e-01 -5.09151638e-01 -1.32424310e-01
6.06319785e-01 1.58209836e+00 -5.55727422e-01 -1.56955183e-01
-8.70314658e-01 1.70332968e-01 4.44412857e-01 3.65940183e-01
-5.85313499e-01 -1.18075661e-01 6.76889658e-01 4.00368035e-01
2.44515985e-01 7.31845992e-03 8.46106783e-02 -1.19087291e+00
-5.99617958e-01 -9.02001262e-01 8.23647082e-01 -3.84371698e-01
-1.81515491e+00 4.26619619e-01 -3.37974817e-01 -1.38620675e+00
1.82392001e-01 -5.49200773e-01 -6.85464382e-01 5.12281060e-01
-1.50002086e+00 -1.30189288e+00 -2.02944860e-01 9.62775350e-01
-4.10363227e-02 -4.82897699e-01 9.04472113e-01 4.86438245e-01
-2.71678388e-01 8.96122217e-01 3.57798636e-01 -8.87541696e-02
4.96539086e-01 -1.29615498e+00 3.49829763e-01 2.37333611e-01
9.21336949e-01 8.92034471e-01 3.39974850e-01 -4.71817762e-01
-1.78075540e+00 -1.17939460e+00 7.11020052e-01 -2.53371716e-01
7.97090113e-01 -3.62475991e-01 -8.25450361e-01 7.13527620e-01
-7.31633753e-02 2.07783863e-01 1.34601092e+00 1.01355481e+00
-4.30007458e-01 -2.47549683e-01 -8.85455370e-01 6.47787511e-01
1.39381206e+00 -2.39328593e-01 -1.78220496e-01 4.88475442e-01
5.12332082e-01 -3.64923701e-02 -1.06978917e+00 1.85876325e-01
7.11055696e-01 -9.80999410e-01 1.31584978e+00 -1.20528638e+00
-1.10275097e-01 -8.17459077e-02 -8.71352702e-02 -1.56870365e+00
-9.77989376e-01 -7.47409344e-01 -7.61270165e-01 1.19850981e+00
5.96156478e-01 -6.48597419e-01 8.34418535e-01 5.50640941e-01
1.90767013e-02 -9.81655478e-01 2.17692889e-02 -4.42271411e-01
-3.74292940e-01 -1.83440909e-01 9.40231025e-01 1.03497827e+00
4.35143143e-01 7.40281343e-01 -8.09106588e-01 2.38292459e-02
5.11830926e-01 4.72772300e-01 8.21528494e-01 -1.76923442e+00
-2.29301989e-01 -4.13311094e-01 -4.98903155e-01 -1.21422625e+00
-6.26626313e-02 -1.28910971e+00 -8.10518205e-01 -2.26454091e+00
1.94160312e-01 -8.99584115e-01 -1.03751743e+00 2.35583857e-01
-3.14054221e-01 -5.84573206e-03 2.11449668e-01 2.38905996e-01
-1.20312309e+00 2.70263046e-01 1.28190672e+00 -1.03176273e-02
-4.90756571e-01 7.05150843e-01 -1.31782174e+00 4.85366017e-01
9.00670409e-01 -2.97393054e-01 -9.29672837e-01 -1.78888813e-01
7.76684880e-01 -2.22214274e-02 -2.69447565e-01 -3.66557151e-01
3.64598542e-01 -3.09554845e-01 -4.71309945e-02 -2.56572306e-01
-6.78626895e-02 -8.77329350e-01 2.48982280e-01 1.34723961e-01
-8.21307421e-01 -6.20496757e-02 -3.92133355e-01 1.00538445e+00
-6.35195076e-02 1.16447762e-01 2.36879364e-01 1.89946413e-01
-5.40583670e-01 1.00419331e+00 -1.34702057e-01 -1.48963630e-01
7.63528049e-01 1.45885095e-01 -2.00103089e-01 -1.00888622e+00
-9.21194911e-01 3.97278398e-01 7.44733512e-02 6.05910540e-01
6.63115919e-01 -1.54528451e+00 -6.07240558e-01 -9.67062116e-02
3.72588158e-01 -7.87717998e-01 1.95295528e-01 6.05439126e-01
1.21948570e-01 3.76239955e-01 -6.83068335e-02 -4.11999896e-02
-1.41355848e+00 8.25051725e-01 -9.31642577e-02 -6.29859865e-01
-2.67196596e-01 7.56800890e-01 1.81070045e-01 -5.86782217e-01
4.23812240e-01 -1.33434462e-03 -1.05204892e+00 1.63057387e-01
2.86211371e-01 5.91338277e-01 -2.38848865e-01 -5.29929280e-01
-3.90183955e-01 4.60505277e-01 -6.73147887e-02 6.29161835e-01
1.43124282e+00 -3.84189874e-01 -1.17223941e-01 2.72523761e-01
8.86146605e-01 4.91702259e-01 -5.42644262e-01 -5.07391989e-01
5.76710217e-02 -5.70569694e-01 1.29873976e-01 -7.60861039e-01
-1.45414102e+00 6.91770613e-01 1.73267320e-01 9.31450188e-01
8.35169375e-01 3.22096974e-01 5.68360209e-01 3.17932934e-01
5.77959716e-01 -7.35979855e-01 -1.32710263e-02 2.10974276e-01
5.83324730e-01 -1.13454747e+00 5.14593482e-01 -7.23833263e-01
-8.21318090e-01 1.14755774e+00 2.23371416e-01 -2.30867267e-01
1.26080656e+00 -3.56995195e-01 -2.14408666e-01 -2.58181453e-01
-6.81610525e-01 -5.02322316e-01 1.12226069e+00 8.91566336e-01
9.17746305e-01 4.16726977e-01 -4.10823107e-01 9.56549466e-01
-5.77304363e-02 -3.82487923e-01 2.37773418e-01 4.08410013e-01
-4.74097282e-01 -1.55441105e+00 3.02718759e-01 9.55667377e-01
-1.74358368e-01 -1.63481385e-01 -6.18061602e-01 6.14065886e-01
-1.55356154e-01 1.37425780e+00 -3.79334182e-01 -8.86055946e-01
4.97593731e-01 -5.09092331e-01 2.63649702e-01 -8.40406001e-01
-6.31651580e-01 -1.10346757e-01 1.92891717e-01 -4.79139715e-01
-4.90339845e-01 -2.63805747e-01 -9.67894793e-01 -5.26644468e-01
-5.94847381e-01 6.09015048e-01 5.49166918e-01 6.20097101e-01
5.80200553e-01 7.60485828e-01 9.93065894e-01 -4.44262713e-01
-1.16566852e-01 -4.94513869e-01 -1.03391874e+00 2.93479115e-01
-8.26267973e-02 -6.51903689e-01 -2.66279608e-01 -4.33283120e-01] | [10.162166595458984, 5.670096397399902] |
5cfd2fe0-2ef0-40e0-8e07-cc982725cb7f | bp-net-efficient-deep-learning-for-continuous | 2111.14558 | null | https://arxiv.org/abs/2111.14558v1 | https://arxiv.org/pdf/2111.14558v1.pdf | BP-Net: Efficient Deep Learning for Continuous Arterial Blood Pressure Estimation using Photoplethysmogram | Blood pressure (BP) is one of the most influential bio-markers for cardiovascular diseases and stroke; therefore, it needs to be regularly monitored to diagnose and prevent any advent of medical complications. Current cuffless approaches to continuous BP monitoring, though non-invasive and unobtrusive, involve explicit feature engineering surrounding fingertip Photoplethysmogram (PPG) signals. To circumvent this, we present an end-to-end deep learning solution, BP-Net, that uses PPG waveform to estimate Systolic BP (SBP), Mean Average Pressure (MAP), and Diastolic BP (DBP) through intermediate continuous Arterial BP (ABP) waveform. Under the terms of the British Hypertension Society (BHS) standard, BP-Net achieves Grade A for DBP and MAP estimation and Grade B for SBP estimation. BP-Net also satisfies Advancement of Medical Instrumentation (AAMI) criteria for DBP and MAP estimation and achieves Mean Absolute Error (MAE) of 5.16 mmHg and 2.89 mmHg for SBP and DBP, respectively. Further, we establish the ubiquitous potential of our approach by deploying BP-Net on a Raspberry Pi 4 device and achieve 4.25 ms inference time for our model to translate the PPG waveform to ABP waveform. | ['Vineeth Vijayaraghavan', 'Nitish Kumar M', 'Abhishek K', 'Poojah G', 'Vedanth S', 'Rishi Vardhan K'] | 2021-11-29 | null | null | null | null | ['blood-pressure-estimation'] | ['medical'] | [-8.87255296e-02 1.45406321e-01 2.30090767e-01 -5.54270744e-01
-4.72743064e-01 -4.65633124e-01 -2.08712265e-01 1.64123178e-01
-2.31274143e-01 1.10561478e+00 1.37215406e-01 -6.45160139e-01
1.00222550e-01 -8.83126736e-01 -3.60469252e-01 -4.77828920e-01
-3.68126452e-01 2.44358238e-02 -8.21093097e-02 2.96043634e-01
2.62108501e-02 4.57945824e-01 -8.42860281e-01 -1.82405487e-01
1.22503448e+00 1.46437275e+00 -7.65031755e-01 1.02766168e+00
1.41264200e-01 1.81613088e-01 -6.93547904e-01 -2.40267664e-01
5.87711513e-01 -3.84265661e-01 -1.19150385e-01 -6.55784786e-01
8.61298859e-01 -7.01796353e-01 -4.10863340e-01 7.05800593e-01
1.26784813e+00 -2.73789436e-01 3.52505714e-01 -7.75776148e-01
-6.00800812e-01 5.26406169e-01 -2.98291713e-01 3.44750196e-01
-1.24929190e-01 4.56000626e-01 7.16307834e-02 -5.64333558e-01
-3.43277454e-01 7.96673059e-01 1.17799532e+00 4.74500358e-01
-1.55463052e+00 -3.52325320e-01 -4.23858076e-01 -2.12626830e-01
-1.33286655e+00 -6.20165169e-01 6.24591231e-01 -5.08526087e-01
3.29826474e-01 7.02009797e-01 1.03711045e+00 7.74773419e-01
6.57425523e-01 -9.18425471e-02 1.37866247e+00 -3.48553151e-01
3.97628605e-01 4.63620991e-01 3.52747083e-01 2.61059314e-01
4.93521631e-01 1.57329828e-01 -3.57363224e-02 -3.75865370e-01
1.27256417e+00 -3.78092825e-02 -4.19351578e-01 1.12746313e-01
-5.70933461e-01 2.79814273e-01 1.09737672e-01 2.37888824e-02
-5.24308205e-01 1.08752988e-01 2.25701943e-01 2.86604196e-01
8.28701332e-02 5.32012105e-01 -4.13393050e-01 -5.42425275e-01
-8.81629407e-01 3.08036089e-01 1.12293780e+00 6.14648521e-01
1.67688861e-01 2.64420193e-02 -6.78392887e-01 8.79400849e-01
5.65976977e-01 9.64236140e-01 1.54935494e-01 -1.37625051e+00
1.42123729e-01 3.20435047e-01 7.00447619e-01 -1.22070074e+00
-6.93924546e-01 -5.17364621e-01 -1.01364112e+00 8.38953182e-02
8.53474915e-01 -6.06366992e-01 -6.95927322e-01 1.41262674e+00
2.13574931e-01 2.74174869e-01 -3.74075770e-01 1.03651297e+00
1.15655887e+00 6.26221439e-03 3.82336944e-01 -4.01783973e-01
1.38487327e+00 -3.21685493e-01 -5.18645287e-01 -2.70161420e-01
-1.72060132e-01 -9.27971825e-02 1.01606786e+00 1.89761460e-01
-1.11396718e+00 -8.64350677e-01 -8.76766562e-01 -1.09165274e-01
-1.43371671e-01 -6.25509918e-02 2.19067141e-01 1.30393660e+00
-8.90711427e-01 1.02276790e+00 -7.55606413e-01 7.09897205e-02
4.72475916e-01 4.24831510e-02 1.22636721e-01 6.07637286e-01
-1.54510820e+00 9.63576972e-01 -4.41978991e-01 7.37290502e-01
-8.35256130e-02 -1.56886959e+00 -4.57279027e-01 9.76686329e-02
-5.01476288e-01 -6.64210021e-01 7.11977899e-01 3.16254556e-01
-2.04751897e+00 6.46345556e-01 -9.04304832e-02 -4.24531370e-01
7.44708776e-01 -6.26785755e-01 -6.66203678e-01 3.20860207e-01
-2.86201477e-01 2.54466515e-02 6.03701234e-01 -4.64895189e-01
1.10785797e-01 -7.64331520e-01 -5.06307065e-01 -3.49868923e-01
-1.68551803e-01 -2.50559717e-01 5.24500869e-02 8.45810771e-02
2.63191402e-01 -5.61075747e-01 -5.43359816e-01 5.28961778e-01
-2.09484741e-01 4.21167612e-01 4.92391065e-02 -1.15581334e+00
1.32115638e+00 -1.79109848e+00 -6.31352305e-01 6.31324887e-01
7.79361606e-01 8.35574508e-01 5.28098524e-01 -5.19315302e-02
1.19327411e-01 1.72536582e-01 -1.45966500e-01 2.67287850e-01
8.54127854e-02 -1.54718459e-01 -1.96567491e-01 6.44049466e-01
-1.14938423e-01 1.25944865e+00 -7.10333467e-01 -3.14602315e-01
6.43325210e-01 6.31541431e-01 -1.88738182e-01 1.80603683e-01
3.47020417e-01 8.50429773e-01 -1.36601195e-01 5.02346694e-01
8.77570391e-01 3.67341302e-02 2.55810879e-02 -4.88318026e-01
-4.44345683e-01 2.12089986e-01 -1.04787266e+00 1.10557902e+00
-2.49903604e-01 5.65528095e-01 -6.12390265e-02 -6.93263590e-01
1.66288435e+00 6.07577145e-01 5.59064567e-01 -1.03825557e+00
2.29490429e-01 3.08673769e-01 1.97839901e-01 -8.89006734e-01
-4.92130309e-01 -3.00855219e-01 4.48568851e-01 2.47900739e-01
-5.02332568e-01 1.67627975e-01 -1.92276195e-01 -4.63937521e-01
8.42490911e-01 -7.21452311e-02 2.52831727e-01 -6.58635914e-01
6.54419422e-01 -6.97486818e-01 4.60074067e-01 1.15121937e+00
-1.02527821e+00 6.07861936e-01 5.31750739e-01 -1.06423569e+00
-1.01529968e+00 -1.46487927e+00 -9.61386204e-01 3.01134497e-01
-2.19369158e-01 1.18660122e-01 -2.57796109e-01 2.24964827e-01
7.75536299e-01 3.37527543e-01 -3.63183022e-01 -6.20582178e-02
-6.41940832e-01 -9.07305002e-01 7.65169024e-01 6.23467386e-01
6.64868951e-01 -6.22979522e-01 -7.96216369e-01 7.53108203e-01
3.01929959e-03 -8.87108862e-01 2.78471783e-02 -3.89093280e-01
-9.48252499e-01 -8.06742728e-01 -1.00182223e+00 -5.73654622e-02
1.32694662e-01 -6.86154783e-01 9.37973976e-01 -5.06377578e-01
-6.50575757e-01 2.01844350e-01 2.46912107e-01 -6.66274786e-01
-1.78320557e-01 -3.99343282e-01 2.06292123e-01 2.91870665e-02
8.19167554e-01 -1.36380982e+00 -1.55510724e+00 1.28233150e-01
3.88586074e-01 -3.85204375e-01 2.28383973e-01 -1.17807262e-01
4.09362018e-01 -8.99936736e-01 9.78433192e-01 -4.08315152e-01
8.21816504e-01 -5.07636555e-02 -7.27532029e-01 -6.27680798e-04
-8.80069911e-01 -6.76421642e-01 6.09606028e-01 -2.31832922e-01
-4.81249303e-01 -3.89476985e-01 -3.03347021e-01 -3.11736725e-02
-2.27265909e-01 2.22247466e-01 2.96755672e-01 -2.11682543e-01
1.29772663e+00 4.19945687e-01 4.31042403e-01 -5.09543121e-01
1.06521323e-01 1.11118019e+00 1.03753018e+00 -7.37767637e-01
2.50788599e-01 1.90218255e-01 4.32563692e-01 -1.03426576e+00
-5.50749600e-01 -6.31278455e-02 -6.07181907e-01 -4.11731809e-01
4.93924916e-01 -7.77256429e-01 -1.57817519e+00 4.80064332e-01
-6.61372900e-01 -1.11186586e-01 -8.68244916e-02 5.11640787e-01
-4.32588756e-01 4.44739074e-01 -7.45644987e-01 -1.07325935e+00
-1.25108171e+00 -2.65991211e-01 3.98169100e-01 4.10189629e-01
-4.40090418e-01 -8.96792471e-01 1.16505243e-01 -8.73044506e-03
1.29832828e+00 1.11116469e+00 5.37782669e-01 -1.48227215e-01
1.99203476e-01 -7.98071623e-01 -5.12715995e-01 6.07976496e-01
4.28544313e-01 -2.04565108e-01 -1.00621045e+00 4.75376993e-02
2.62694657e-01 2.66453803e-01 1.55151367e-01 9.69599307e-01
1.14477909e+00 -2.69439459e-01 -1.06123924e-01 6.10949159e-01
1.36202204e+00 2.74707705e-01 1.16325831e+00 6.39816970e-02
5.25512099e-01 -4.91168313e-02 -4.64453638e-01 7.66738057e-01
3.33647788e-01 2.96039194e-01 -1.76297560e-01 -1.53871194e-01
3.78790237e-02 1.45253554e-01 -9.48760584e-02 1.72911108e-01
-5.30128837e-01 6.92206264e-01 -8.70939970e-01 -3.16057801e-02
-1.37162864e+00 -8.11279297e-01 -6.07816041e-01 2.44159269e+00
1.35922468e+00 6.55533448e-02 3.26303691e-01 -1.06837451e-01
5.67390084e-01 -3.84201497e-01 -7.78811514e-01 -5.55083632e-01
3.05201620e-01 5.35627723e-01 4.60710764e-01 4.02192593e-01
-1.05192649e+00 8.54134932e-02 6.61957216e+00 -7.17633009e-01
-1.46029985e+00 -2.55296230e-01 8.11904311e-01 4.16942723e-02
3.63907933e-01 -4.00073916e-01 -7.30102599e-01 1.05659914e+00
1.48741901e+00 -5.62046468e-01 3.30241650e-01 8.47649753e-01
7.56543159e-01 3.80075201e-02 -1.10524571e+00 1.02945638e+00
-5.10697365e-01 -1.31110704e+00 -8.02449346e-01 -3.16061080e-01
1.36873601e-02 -1.36902809e-01 -2.24216968e-01 3.97790462e-01
-4.38952953e-01 -9.29842830e-01 -1.37424737e-01 1.12752247e+00
1.13393199e+00 4.65776585e-03 5.66415846e-01 -5.88847734e-02
-8.13086450e-01 5.37176952e-02 -6.32264376e-01 -4.51981843e-01
1.89536408e-01 1.22559869e+00 -5.47057688e-01 -3.34356502e-02
5.05079985e-01 4.01382357e-01 -3.79405916e-01 1.55938077e+00
-2.93711182e-02 7.91372955e-01 -5.38280487e-01 4.55480590e-02
-3.34924221e-01 -2.23203555e-01 3.09431076e-01 1.02306235e+00
1.60123751e-01 3.21927756e-01 -3.30076665e-02 1.57089508e+00
4.30476099e-01 -3.69745828e-02 1.17051927e-02 4.18049365e-01
6.94633484e-01 9.17774141e-01 6.32102042e-02 -4.33851808e-01
-2.60116071e-01 1.96415618e-01 -2.70559222e-01 5.48258841e-01
-8.14233959e-01 -8.76494586e-01 6.47212446e-01 5.38019478e-01
-3.80705059e-01 1.49936313e-02 -1.11998355e+00 -8.96941662e-01
5.34944892e-01 -3.47292304e-01 1.44662082e-01 -2.33354911e-01
-1.29135120e+00 9.72120613e-02 -3.57957244e-01 -1.04372895e+00
3.72332819e-02 -5.69313824e-01 -8.71400237e-01 1.87049031e+00
-1.54483140e+00 -3.74553174e-01 -7.18950033e-01 1.89005941e-01
-4.19969469e-01 3.36779714e-01 1.15542197e+00 5.87300718e-01
-5.37602246e-01 6.68059826e-01 -2.28131518e-01 5.11712968e-01
6.57163858e-01 -1.32954705e+00 6.14948273e-01 4.34547067e-01
-1.13896096e+00 9.99891758e-01 7.95770764e-01 -4.87267137e-01
-1.52496433e+00 -1.17103779e+00 1.04982162e+00 -5.22860944e-01
3.56477946e-01 -1.01071797e-01 -7.96368718e-01 3.13790262e-01
-4.38395292e-01 6.33405805e-01 8.51626515e-01 -2.44302414e-02
-1.60507247e-01 -6.30022407e-01 -1.42945635e+00 6.10112131e-01
3.57633889e-01 -2.56056696e-01 -4.84315217e-01 2.31046930e-01
2.65633821e-01 -7.26519406e-01 -1.72678745e+00 5.71396232e-01
1.44803154e+00 -7.68291056e-01 1.17001009e+00 -6.26239657e-01
3.98531824e-01 -9.49658677e-02 2.72970676e-01 -1.06523383e+00
-5.67594647e-01 -1.00657856e+00 -5.02025604e-01 9.16211247e-01
2.73713410e-01 -1.13097847e+00 7.17919290e-01 1.81007254e+00
-1.86676472e-01 -8.30510437e-01 -9.50578153e-01 -7.07521975e-01
3.12595159e-01 -7.83766713e-03 4.94083792e-01 4.69572484e-01
6.21392906e-01 -4.36861999e-02 -5.68352878e-01 2.39248350e-01
9.65685368e-01 2.47177720e-01 7.27244735e-01 -1.41116881e+00
-2.02578992e-01 -4.09936994e-01 -4.19170439e-01 -8.12856555e-01
-8.84292483e-01 -3.36253494e-01 -1.38194039e-01 -1.55988371e+00
-3.92280638e-01 -3.34862679e-01 -6.99148118e-01 3.35734010e-01
-3.61114204e-01 2.74250060e-01 -2.91296601e-01 -2.70056367e-01
2.21189573e-01 -2.10130364e-02 1.14987612e+00 2.77761638e-01
-9.78367627e-01 3.42244953e-01 -7.30600059e-01 1.82581261e-01
9.66497719e-01 -1.60900384e-01 -4.86417189e-02 1.44286081e-01
-1.21516868e-01 4.00857508e-01 5.02041817e-01 -1.19825435e+00
5.79129085e-02 -2.47353390e-02 1.06874585e+00 -1.01659089e-01
3.44895907e-02 -6.41341209e-01 4.20457497e-02 8.27188730e-01
-2.80875862e-01 -6.19500101e-01 2.77203530e-01 3.26399654e-02
2.54954487e-01 3.11410397e-01 1.01250088e+00 -3.71678531e-01
1.15471475e-01 3.61382365e-01 -3.74984622e-01 7.89657757e-02
5.16694188e-01 -3.80507886e-01 -6.39600158e-01 -2.17112780e-01
-1.11000025e+00 2.30390862e-01 -4.05066967e-01 1.07032016e-01
5.58188260e-01 -1.18225968e+00 -1.04929721e+00 5.31132400e-01
-7.22905844e-02 -4.83016849e-01 4.99675304e-01 1.36941385e+00
-6.20362043e-01 4.38509732e-01 -3.30927521e-01 -5.51215470e-01
-7.38261342e-01 4.48446302e-03 1.00685561e+00 4.91592765e-01
-1.22616124e+00 5.23664713e-01 -7.44516313e-01 1.87394973e-02
5.32748342e-01 -2.95941859e-01 -9.63326320e-02 -4.69961941e-01
1.18902230e+00 7.39837229e-01 5.73523156e-02 4.90326047e-01
-2.29062751e-01 4.01544034e-01 2.86086321e-01 3.34887415e-01
1.20755768e+00 -2.62927741e-01 -2.70143658e-01 3.29789907e-01
7.77502835e-01 -3.27509552e-01 -9.38253760e-01 -1.71976075e-01
-3.38824838e-01 -4.72818643e-01 -1.27794459e-01 -1.27768540e+00
-8.16121638e-01 9.32526946e-01 1.13173473e+00 3.87485057e-01
7.24984348e-01 -5.36930501e-01 9.48639154e-01 1.74121231e-01
-4.70913947e-02 -9.27079022e-01 -5.87010026e-01 7.14555010e-02
8.87009442e-01 -7.57479310e-01 1.19914874e-01 -3.90423238e-01
-3.22976671e-02 1.36991704e+00 6.30139351e-01 -3.10825557e-01
1.20260835e+00 1.79032117e-01 6.21963680e-01 7.58249536e-02
-8.33517835e-02 3.25451642e-01 2.64578283e-01 6.43139064e-01
7.24742711e-01 3.79572719e-01 -8.29328060e-01 7.27671206e-01
-1.98590711e-01 7.09277987e-01 3.43790799e-01 6.15803421e-01
-7.80387700e-01 -3.95561576e-01 -9.67822671e-02 9.18922961e-01
-4.86624658e-01 -7.35116974e-02 2.20899656e-01 1.98771581e-01
-2.38632500e-01 8.85516047e-01 6.56053275e-02 -2.16216221e-02
7.00008750e-01 5.24229884e-01 4.12981063e-01 -2.06822857e-01
-3.23787540e-01 -7.77376294e-02 -7.04517215e-02 -5.22157252e-01
5.02852947e-02 -1.60588652e-01 -9.12521243e-01 -2.89784163e-01
2.86140651e-01 -1.34261623e-01 7.40537167e-01 7.37565696e-01
7.36627400e-01 2.78656989e-01 5.63339651e-01 -3.67929876e-01
-9.83304620e-01 -1.19047403e+00 -7.75420904e-01 2.17091680e-01
5.26202321e-01 -2.85286784e-01 -2.79113144e-01 -2.22869497e-02] | [14.087676048278809, 2.948641300201416] |
817ca623-95db-4995-b799-d78feb741bbc | improving-dataset-distillation | 1910.02551 | null | https://arxiv.org/abs/1910.02551v3 | https://arxiv.org/pdf/1910.02551v3.pdf | Soft-Label Dataset Distillation and Text Dataset Distillation | Dataset distillation is a method for reducing dataset sizes by learning a small number of synthetic samples containing all the information of a large dataset. This has several benefits like speeding up model training, reducing energy consumption, and reducing required storage space. Currently, each synthetic sample is assigned a single `hard' label, and also, dataset distillation can currently only be used with image data. We propose to simultaneously distill both images and their labels, thus assigning each synthetic sample a `soft' label (a distribution of labels). Our algorithm increases accuracy by 2-4% over the original algorithm for several image classification tasks. Using `soft' labels also enables distilled datasets to consist of fewer samples than there are classes as each sample can encode information for multiple classes. For example, training a LeNet model with 10 distilled images (one per class) results in over 96% accuracy on MNIST, and almost 92% accuracy when trained on just 5 distilled images. We also extend the dataset distillation algorithm to distill sequential datasets including texts. We demonstrate that text distillation outperforms other methods across multiple datasets. For example, models attain almost their original accuracy on the IMDB sentiment analysis task using just 20 distilled sentences. Our code can be found at $\href{https://github.com/ilia10000/dataset-distillation}{\text{https://github.com/ilia10000/dataset-distillation}}$. | ['Matthias Schonlau', 'Ilia Sucholutsky'] | 2019-10-06 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [ 4.32141602e-01 8.07818249e-02 -4.79950339e-01 -7.00027466e-01
-1.08150053e+00 -7.10623741e-01 3.64400297e-01 2.49226078e-01
-7.77074337e-01 7.69096911e-01 -5.47512248e-02 -2.45402277e-01
4.85400587e-01 -8.16663861e-01 -9.63165581e-01 -5.98373890e-01
3.23140264e-01 5.32641113e-01 -1.62454341e-02 2.39994422e-01
3.95209976e-02 9.92488116e-02 -1.52872634e+00 7.08554566e-01
6.70647979e-01 1.21166062e+00 3.80231440e-01 7.71537125e-01
-4.56556492e-03 8.75689685e-01 -7.07808852e-01 -3.92325699e-01
4.82188970e-01 -2.65768290e-01 -7.77036369e-01 -5.75484931e-02
8.01550329e-01 -5.06742895e-01 -4.01346326e-01 1.08051980e+00
5.27211547e-01 -3.66466716e-02 4.57915217e-01 -1.25824046e+00
-6.37246132e-01 7.63250411e-01 -7.09083140e-01 1.72651783e-01
-4.95435148e-02 2.17264503e-01 1.04098928e+00 -9.89615083e-01
4.97524470e-01 9.24417198e-01 3.29293787e-01 7.88528264e-01
-1.38270605e+00 -1.19871604e+00 7.44129196e-02 6.62544593e-02
-1.31082726e+00 -4.68361586e-01 5.76960921e-01 -2.86964893e-01
7.26352096e-01 4.53204632e-01 4.81475830e-01 1.04049528e+00
-1.15799367e-01 1.27273417e+00 1.22161829e+00 -4.70526725e-01
3.18686336e-01 4.09627527e-01 3.24602932e-01 6.38183177e-01
3.70840698e-01 -2.75167644e-01 -3.87496710e-01 8.28859881e-02
3.99288893e-01 1.93008110e-01 -1.73995152e-01 2.38317139e-02
-1.28659177e+00 8.43920887e-01 5.62767684e-01 -3.87417502e-03
5.01222424e-02 4.13343668e-01 6.18458092e-01 3.93493652e-01
5.53637028e-01 4.24596637e-01 -6.98086619e-01 -1.27988800e-01
-1.00653708e+00 2.44260058e-01 6.34232461e-01 1.06416166e+00
1.06195772e+00 1.50155080e-02 -1.21028244e-01 1.15721977e+00
-2.15873495e-01 8.44170928e-01 7.79048920e-01 -1.08712232e+00
6.47423029e-01 4.74320650e-01 -9.08777565e-02 -8.62058759e-01
-2.95881152e-01 -2.03803852e-01 -1.12656343e+00 -3.17339450e-02
4.50620830e-01 -3.98666084e-01 -9.94137168e-01 1.93768227e+00
1.71139374e-01 -4.38779406e-02 -5.81055135e-02 4.71872240e-01
9.20333862e-01 8.99934232e-01 -5.44788428e-02 8.10828060e-02
1.47027540e+00 -1.15905607e+00 -3.53361785e-01 -5.80490351e-01
1.01336884e+00 -6.13868177e-01 1.45019841e+00 6.75143182e-01
-8.66699815e-01 -6.94386363e-01 -9.55307126e-01 -2.38678470e-01
-5.30475318e-01 4.46332574e-01 4.39001560e-01 5.91868222e-01
-1.03526139e+00 4.17080402e-01 -8.44976127e-01 1.97621970e-03
8.84264767e-01 4.43724245e-01 -2.47530237e-01 -5.11824369e-01
-9.49500084e-01 3.91991109e-01 5.60775280e-01 -2.70905465e-01
-7.81117797e-01 -7.12491512e-01 -9.06949937e-01 -1.24147847e-01
1.70433998e-01 -2.56088823e-01 1.39628851e+00 -9.69639838e-01
-9.33520019e-01 1.04142952e+00 -2.17951566e-01 -5.57975113e-01
2.74466693e-01 -9.29575488e-02 -2.58930713e-01 2.02794641e-01
-6.67389296e-03 1.37583077e+00 6.53650522e-01 -1.01003385e+00
-6.23017550e-01 -2.01075807e-01 2.57700793e-02 -5.73284691e-03
-8.04825306e-01 -2.72563547e-01 -5.94416440e-01 -9.13084149e-01
-1.71873555e-01 -1.26516247e+00 -4.99027111e-02 1.81472659e-01
-6.25380397e-01 -2.75740288e-02 8.30857992e-01 -4.49736863e-01
1.08055961e+00 -2.27618623e+00 -2.88205057e-01 -9.08310935e-02
3.39965731e-01 3.44944566e-01 -3.60425711e-01 -8.59012082e-02
-2.57366389e-01 2.71207690e-01 -2.11629897e-01 -5.86766601e-01
7.17100352e-02 2.78681993e-01 -2.13666260e-01 1.85616553e-01
-8.51499196e-03 9.75371540e-01 -7.56905019e-01 -3.63545954e-01
1.31205291e-01 3.42223495e-01 -6.54974401e-01 -1.51242211e-01
-2.87386477e-01 4.76553701e-02 -1.42002970e-01 4.04523402e-01
9.25092340e-01 -5.99807143e-01 1.00423947e-01 -3.35584879e-01
4.45323825e-01 2.11262420e-01 -1.15784395e+00 1.57656884e+00
-5.29740036e-01 9.55821395e-01 -1.00623928e-01 -9.85664308e-01
7.50030100e-01 1.67493280e-02 4.54999357e-01 -8.25156808e-01
1.10773787e-01 2.80916393e-01 -1.95153520e-01 -1.37421787e-01
4.57084417e-01 -1.24658737e-02 -3.93255413e-01 8.66961777e-01
-1.35932088e-01 -3.58264297e-01 4.52068955e-01 4.20621216e-01
1.10049117e+00 -3.25165004e-01 -1.84582859e-01 -4.12475467e-02
9.81967673e-02 1.99558493e-03 5.04049957e-01 7.09287703e-01
-5.14771193e-02 6.12773955e-01 5.17921865e-01 -5.44389307e-01
-1.25506747e+00 -9.19209957e-01 -3.77931029e-01 1.08905029e+00
-1.04196094e-01 -4.82765704e-01 -8.98387492e-01 -8.26783240e-01
9.17603672e-02 8.17705691e-01 -6.98036075e-01 -2.34322697e-01
-3.96554053e-01 -9.57713664e-01 5.12165308e-01 5.77510059e-01
7.42777646e-01 -9.49902713e-01 -5.86689591e-01 -2.83861041e-01
-4.41228360e-01 -1.10969627e+00 -6.05631292e-01 5.16857147e-01
-7.17053831e-01 -8.92996490e-01 -5.53650916e-01 -7.72191942e-01
1.02145350e+00 2.58320540e-01 1.09260428e+00 1.35586128e-01
-4.46372837e-01 -1.64271635e-03 -3.82345766e-01 -5.46980441e-01
-4.37418967e-01 2.51471490e-01 1.40913669e-02 -2.22422644e-01
5.56681454e-01 -2.65617460e-01 -7.84763753e-01 1.75964072e-01
-1.17188430e+00 3.98303181e-01 4.67111260e-01 8.90802979e-01
7.13340402e-01 7.11933896e-02 6.36645138e-01 -1.37742925e+00
3.86739612e-01 -4.91848648e-01 -4.29726690e-01 -4.36917208e-02
-6.33615315e-01 4.90054898e-02 8.55676293e-01 -6.45142674e-01
-5.44623375e-01 5.32178953e-02 -1.91590518e-01 -4.42914397e-01
-1.07800931e-01 2.13161156e-01 5.72829582e-02 3.39695424e-01
7.72940338e-01 2.47520447e-01 1.58105597e-01 -3.86224836e-01
4.26108569e-01 9.26284313e-01 3.70554805e-01 -6.09807193e-01
5.28051555e-01 3.93055737e-01 -4.07604605e-01 -4.21661466e-01
-1.03977692e+00 -1.77025586e-01 -3.89682293e-01 1.42569214e-01
5.76737821e-01 -1.11622012e+00 -4.58596110e-01 6.19276881e-01
-6.53362572e-01 -9.53896701e-01 -6.90360665e-01 3.91658276e-01
-1.96311355e-01 1.73560917e-01 -9.22364652e-01 -1.49043277e-01
-4.98591363e-01 -1.32475483e+00 9.42300916e-01 2.41814405e-02
-2.41457388e-01 -8.88872385e-01 -2.93631166e-01 4.60002631e-01
1.93299472e-01 1.80307940e-01 7.96883643e-01 -7.83063114e-01
-2.97852784e-01 -5.60670078e-01 -3.81975591e-01 8.94300282e-01
2.43096411e-01 -1.28249213e-01 -1.11788964e+00 -6.89897001e-01
-3.14052641e-01 -9.46100235e-01 1.02621710e+00 2.78394997e-01
1.82882321e+00 -5.45892715e-01 -3.84244204e-01 4.12578106e-01
1.36848509e+00 1.32963017e-01 5.10848939e-01 1.11736506e-01
7.03580379e-01 2.51836210e-01 7.13523924e-01 6.36515737e-01
3.59009147e-01 4.62313741e-01 2.90403068e-01 -2.93620050e-01
-2.83738226e-01 -2.38565542e-02 2.42585957e-01 7.48143077e-01
6.15672350e-01 -4.37786162e-01 -8.96686971e-01 5.97409844e-01
-1.36243665e+00 -7.96989679e-01 -5.37821017e-02 2.35878229e+00
1.33239114e+00 3.29794258e-01 1.06236190e-01 2.23593682e-01
4.64208692e-01 1.54856935e-01 -8.69665623e-01 -3.56344104e-01
-7.08111450e-02 4.56950665e-01 7.32417226e-01 5.24899006e-01
-1.10468125e+00 8.39670479e-01 5.22573566e+00 1.05307972e+00
-1.39910460e+00 1.11722365e-01 1.27726293e+00 -8.27135921e-01
-4.02546048e-01 -3.03334773e-01 -1.06212497e+00 9.33358073e-01
1.12440348e+00 -6.76787347e-02 3.24013084e-01 9.26718414e-01
-1.63142141e-02 -3.02822739e-01 -1.27276921e+00 1.16889644e+00
1.18245795e-01 -1.39802682e+00 1.01844937e-01 1.59490824e-01
9.35037553e-01 3.93481493e-01 4.12896395e-01 4.49709594e-01
5.75482309e-01 -8.86668622e-01 7.57493854e-01 1.23347938e-01
1.33930719e+00 -6.77870333e-01 5.60454130e-01 4.70503181e-01
-8.96999896e-01 -2.10542351e-01 -5.00201344e-01 1.67470142e-01
-2.89922059e-01 1.01234388e+00 -8.88873756e-01 -3.42316031e-02
9.54353988e-01 6.42575443e-01 -9.40373957e-01 6.16532564e-01
-5.43964095e-02 7.49952435e-01 -6.26825631e-01 -1.00373395e-01
4.19248156e-02 1.75088227e-01 -1.12606913e-01 1.31028640e+00
1.81826472e-01 1.23940274e-01 2.05431953e-01 3.87828767e-01
-6.43366277e-01 1.18935518e-02 -5.63084602e-01 1.64629236e-01
8.02552462e-01 1.21600413e+00 -8.46316516e-01 -8.40377331e-01
-3.08457553e-01 9.67967749e-01 3.14486742e-01 1.62986115e-01
-8.93079877e-01 -7.42043197e-01 4.13606614e-01 4.19771001e-02
2.08807722e-01 7.03288289e-03 -4.30867642e-01 -1.29024696e+00
1.67908862e-01 -1.04861546e+00 4.46765244e-01 -8.98840308e-01
-1.00933623e+00 6.04403973e-01 -1.12460405e-02 -1.26386714e+00
-1.33615389e-01 -6.54576063e-01 -4.36310396e-02 6.86426044e-01
-1.27370858e+00 -8.53461564e-01 -6.34144247e-01 4.15126771e-01
7.04280496e-01 1.11181503e-02 7.83652902e-01 6.15840912e-01
-6.87244236e-01 1.07872009e+00 4.69542235e-01 4.24263597e-01
1.05335402e+00 -1.24484503e+00 4.63148832e-01 5.42359233e-01
1.59083903e-01 3.80985737e-01 3.84289414e-01 -4.84238058e-01
-1.02335048e+00 -1.38097668e+00 5.16745687e-01 -4.26704496e-01
3.74396622e-01 -6.96054757e-01 -7.42222607e-01 9.73489463e-01
1.65980935e-01 2.98285633e-01 8.60349774e-01 -3.44543785e-01
-4.80034649e-01 -5.91972470e-01 -1.32290721e+00 4.85136449e-01
6.76660061e-01 -4.58931029e-01 4.72446866e-02 7.49459386e-01
7.57261038e-01 -4.58404601e-01 -8.53973508e-01 2.69436210e-01
4.25799310e-01 -5.76713145e-01 8.23639452e-01 -1.36693373e-01
6.43928647e-01 -7.68741407e-03 -3.51231396e-01 -1.16557181e+00
-8.23185965e-02 -1.59735963e-01 -2.00839743e-01 1.09087980e+00
6.85935438e-01 -6.99637532e-01 1.12380981e+00 8.03838253e-01
1.47965536e-01 -1.05898321e+00 -5.59134603e-01 -6.70083761e-01
3.67977172e-01 -5.72583795e-01 5.46441615e-01 1.12363517e+00
-1.94871739e-01 1.74143866e-01 -1.90061823e-01 -2.33106747e-01
6.39017224e-01 8.51669442e-03 7.57641852e-01 -9.05348182e-01
-3.88771832e-01 -3.38748425e-01 1.49283502e-02 -1.40570796e+00
-1.13947894e-02 -1.19598091e+00 -1.41149342e-01 -1.41889930e+00
4.25765306e-01 -9.60499823e-01 -1.45729110e-01 1.11808085e+00
-3.57029624e-02 8.63185346e-01 2.96321392e-01 2.73534894e-01
-4.74391997e-01 2.42760003e-01 1.19019008e+00 -3.83476883e-01
9.67878997e-02 -1.87245429e-01 -8.87473524e-01 4.38329965e-01
9.32265520e-01 -6.56276882e-01 -6.20665789e-01 -7.02598691e-01
2.93790221e-01 -2.63221949e-01 1.84898868e-01 -9.77257133e-01
1.77820742e-01 -1.40018836e-02 5.58276594e-01 -6.49201870e-01
4.44344014e-01 -5.81070065e-01 3.27821746e-02 5.78658044e-01
-5.77966809e-01 -1.01523243e-01 5.15978634e-01 3.29868525e-01
-1.05006412e-01 -3.31100255e-01 1.11223221e+00 -3.13038230e-01
-5.69875121e-01 2.24867985e-01 -1.83314294e-01 3.10828805e-01
1.12869155e+00 -2.42501676e-01 -5.14948308e-01 -2.01535016e-01
-7.45129168e-01 3.16890270e-01 5.12700796e-01 3.11866015e-01
5.43537974e-01 -1.30833972e+00 -6.52689040e-01 3.37568849e-01
6.77869990e-02 5.38151681e-01 3.03459406e-01 5.76714814e-01
-7.09672213e-01 3.07997793e-01 -5.86976148e-02 -6.54753745e-01
-1.38200963e+00 4.85588819e-01 1.80900455e-01 -2.85922348e-01
-3.61171037e-01 1.16426516e+00 3.07330519e-01 -5.93685925e-01
2.12714404e-01 -5.40984988e-01 2.74761349e-01 7.32606947e-02
6.46570921e-01 2.57655203e-01 1.19031891e-01 -1.99875355e-01
-1.60318553e-01 4.18701410e-01 -6.61611736e-01 5.71227409e-02
1.33913255e+00 2.89004426e-02 -2.41140398e-04 3.80223244e-01
1.65680587e+00 -8.70110914e-02 -1.26153135e+00 -4.43274170e-01
-4.52986032e-01 -4.73559916e-01 7.33421324e-03 -9.14463580e-01
-1.42539394e+00 9.88434017e-01 7.13484704e-01 1.31151289e-01
1.43183374e+00 1.74663626e-02 1.04292440e+00 5.13893068e-01
3.65305334e-01 -1.03861153e+00 5.08434355e-01 2.33453095e-01
4.90416974e-01 -1.44879901e+00 1.45736963e-01 -1.25854105e-01
-7.41683543e-01 8.22264969e-01 7.51451433e-01 -7.42875636e-02
6.88506484e-01 5.71198404e-01 -1.08961180e-01 -3.21315192e-02
-8.49187732e-01 3.40041608e-01 -3.72481495e-02 1.86435804e-01
2.35903144e-01 2.50190645e-01 3.86867151e-02 3.20802093e-01
-3.54852319e-01 7.79213682e-02 6.07587934e-01 9.96106327e-01
-3.25388759e-01 -1.25833440e+00 -8.14756677e-02 9.98275638e-01
-4.07110423e-01 -3.05954278e-01 -1.15927428e-01 6.22822344e-01
2.36987382e-01 7.37164080e-01 4.10424620e-01 -5.42101264e-01
9.77752209e-02 -3.33361863e-03 2.45423049e-01 -7.95109868e-01
-2.36952215e-01 -1.37198061e-01 -6.21826537e-02 -2.90294826e-01
-1.18289344e-01 -6.83844090e-01 -1.29637623e+00 -6.20628774e-01
-1.79324284e-01 1.46751121e-01 6.94254339e-01 4.86261219e-01
4.31088954e-01 3.80113333e-01 8.22672129e-01 -6.61616683e-01
-4.71399128e-01 -9.76982236e-01 -4.95845675e-01 3.52600247e-01
2.62551606e-01 -2.19315365e-01 -4.50086623e-01 4.06429470e-01] | [9.29942798614502, 2.5974016189575195] |
345f4b8a-6694-4064-a689-f5368b77dcb4 | graph-based-semi-supervised-learning-approach | null | null | https://aclanthology.org/L18-1624 | https://aclanthology.org/L18-1624.pdf | Graph Based Semi-Supervised Learning Approach for Tamil POS tagging | null | ['Uthayasanker Thayasivam', 'Surangika Ranathunga', 'Mokanarangan Thayaparan'] | 2018-05-01 | graph-based-semi-supervised-learning-approach-1 | https://aclanthology.org/L18-1624 | https://aclanthology.org/L18-1624.pdf | lrec-2018-5 | ['graph-similarity'] | ['graphs'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.21641731262207, 3.6164743900299072] |
56b2a15e-2426-40d3-aedb-2eb7698621c2 | dynamic-compressive-sensing-based-on-rls-for | 2304.11838 | null | https://arxiv.org/abs/2304.11838v2 | https://arxiv.org/pdf/2304.11838v2.pdf | Dynamic Compressive Sensing based on RLS for Underwater Acoustic Communications | Sparse structures are widely recognized and utilized in channel estimation. Two typical mechanisms, namely proportionate updating (PU) and zero-attracting (ZA) techniques, achieve better performance, but their computational complexity are higher than non-sparse counterparts. In this paper, we propose a DCS technique based on the recursive least squares (RLS) algorithm which can simultaneously achieve improved performance and reduced computational complexity. Specifically, we develop the sparse adaptive subspace pursuit-improved RLS (SpAdSP-IRLS) algorithm by updating only the sparse structure in the IRLS to track significant coefficients. The complexity of the SpAdSP-IRLS algorithm is successfully reduced to $\mathcal{O}(L^2+2L(s+1)+10s)$, compared with the order of $\mathcal{O}(3L^2+4L)$ for the standard RLS. Here, $L$ represents the length of the channel, and $s$ represents the size of the support set. Our experiments on both synthetic and real data show the superiority of the proposed SpAdSP-IRLS, even though only $s$ elements are updated in the channel estimation. | ['Zhen Qin'] | 2023-04-24 | null | null | null | null | ['compressive-sensing'] | ['computer-vision'] | [ 4.53307480e-01 -4.87711042e-01 -4.16173749e-02 2.14285012e-02
-8.68499637e-01 -1.95279032e-01 -5.78976944e-02 1.14966959e-01
-4.83525187e-01 7.61955261e-01 1.24971710e-01 -3.47270787e-01
-1.61261603e-01 -4.43332165e-01 -4.65269744e-01 -8.58604312e-01
-4.43871170e-01 -3.85886073e-01 2.42324010e-01 -1.74708828e-01
3.71957570e-01 3.67277622e-01 -1.02793050e+00 -3.28448087e-01
1.00693750e+00 1.48512304e+00 2.12515607e-01 2.79549003e-01
-1.11525603e-01 6.27127469e-01 -1.33876577e-01 6.06297143e-02
3.34557354e-01 -5.19442916e-01 1.65571302e-01 6.62421435e-02
-3.72614227e-02 -1.22422092e-01 -6.42087102e-01 1.28948331e+00
6.04638338e-01 2.38761399e-02 4.23501819e-01 -9.62488770e-01
8.60711839e-03 4.99104321e-01 -1.26418197e+00 1.98450997e-01
4.54170406e-01 -3.89781356e-01 5.30175030e-01 -1.01289058e+00
1.71733037e-01 1.23235428e+00 7.19091654e-01 -2.77659059e-01
-9.03200686e-01 -1.24105859e+00 1.30899265e-01 1.35543078e-01
-2.02040052e+00 -8.08349490e-01 8.53054583e-01 -2.58781224e-01
4.16891813e-01 2.69365281e-01 4.75654066e-01 3.33208084e-01
-1.78926662e-01 6.96488500e-01 1.12292850e+00 -7.21231222e-01
2.55780786e-01 3.07101160e-02 2.34845921e-01 7.63916910e-01
6.62291169e-01 1.88294705e-02 -5.39704263e-01 -4.43421155e-01
1.09242785e+00 -9.76075605e-03 -5.33997834e-01 -1.84261024e-01
-1.09995985e+00 8.61686587e-01 -2.34450400e-02 3.01685482e-01
-4.33744252e-01 9.85404626e-02 2.14502275e-01 1.88604951e-01
4.56453040e-02 -6.99773580e-02 -9.37150717e-02 -9.11909342e-02
-1.15035605e+00 9.44669917e-02 5.20951509e-01 1.30652249e+00
7.04911768e-01 5.96567631e-01 -5.36965616e-02 8.06816101e-01
5.09097159e-01 9.72457170e-01 1.60884738e-01 -9.47841227e-01
6.37187421e-01 2.92621315e-01 1.87113121e-01 -1.43134797e+00
-3.06827724e-01 -8.40949416e-01 -1.33072579e+00 -3.56960595e-01
2.22026393e-01 -4.89558160e-01 -2.15014249e-01 1.43787372e+00
-2.94124894e-02 4.69881266e-01 1.86884493e-01 7.67337501e-01
4.91031945e-01 9.45853055e-01 -2.17692077e-01 -9.13112581e-01
1.01322138e+00 -6.23894632e-01 -7.33008504e-01 -3.86480063e-01
4.27407473e-01 -1.01789474e+00 5.51944792e-01 4.34653729e-01
-9.77551222e-01 -4.07411516e-01 -1.07902408e+00 7.36719966e-01
5.44318020e-01 5.10354221e-01 6.03084326e-01 5.97517550e-01
-5.99661350e-01 5.43023236e-02 -4.27797824e-01 2.69170910e-01
2.48637825e-01 3.61396104e-01 -1.53247923e-01 -2.22827241e-01
-1.12361205e+00 4.06937376e-02 2.35584497e-01 2.38204032e-01
-4.36834931e-01 -2.74972767e-01 -9.03416812e-01 2.27332875e-01
5.90467155e-01 2.90284920e-02 7.03767836e-01 -5.08892775e-01
-1.14527595e+00 7.86991268e-02 -6.86937690e-01 -4.63818967e-01
-1.32635489e-01 1.52988926e-01 -6.53648078e-01 1.51001409e-01
1.16108961e-01 -3.51606533e-02 9.84157026e-01 -9.96521115e-01
-6.16135776e-01 -3.37884486e-01 -3.28602195e-01 1.92611352e-01
-2.91967362e-01 2.83807497e-02 -7.02321291e-01 -9.25234616e-01
9.28774238e-01 -1.06925118e+00 -6.69487000e-01 -1.40018165e-01
-3.03801805e-01 3.33904356e-01 5.47209859e-01 -6.73177242e-01
1.83582592e+00 -2.49084711e+00 -3.89767773e-02 8.36036205e-01
2.51008552e-02 2.96784759e-01 9.93733760e-03 3.57595265e-01
1.86827362e-01 -5.59416771e-01 -3.25682491e-01 1.30466655e-01
-3.47085863e-01 -1.11150861e-01 -1.86141685e-01 6.71931088e-01
-5.07771909e-01 1.33268952e-01 -6.33398592e-01 -3.20866555e-01
-4.84265573e-02 4.60139781e-01 -5.98104894e-01 -2.31312439e-01
4.06909704e-01 4.53026891e-01 -6.48407459e-01 7.63215363e-01
9.53171730e-01 -3.50610763e-01 2.77369827e-01 -5.05323946e-01
-5.54732442e-01 -8.44682902e-02 -1.92423463e+00 1.41448700e+00
-3.25379312e-01 4.19157296e-01 4.76616025e-01 -1.03874171e+00
1.02127361e+00 2.21379042e-01 5.72363019e-01 -9.06477034e-01
7.91723654e-02 6.18698001e-01 -9.49305519e-02 -1.51543066e-01
1.80378541e-01 -8.75594318e-02 -9.19887200e-02 2.49045759e-01
-4.29386884e-01 3.80174041e-01 2.28885904e-01 3.26525271e-01
1.07966983e+00 -4.59204108e-01 6.69437766e-01 -4.34294343e-01
1.16158640e+00 -2.90845394e-01 1.01283276e+00 6.69892669e-01
-2.42348135e-01 -2.72406861e-02 3.90043825e-01 1.55939469e-02
-5.97755730e-01 -6.97298110e-01 -2.70963460e-01 6.19071782e-01
6.78776860e-01 -3.69694978e-01 -3.93800646e-01 -3.33483815e-02
-3.67257930e-02 5.92369080e-01 2.35293403e-01 1.74941391e-01
-6.18673861e-01 -8.49477947e-01 4.22161162e-01 2.66567916e-01
7.80384779e-01 -6.03010476e-01 -2.77390182e-01 4.67472643e-01
-7.84898475e-02 -1.17663372e+00 -4.67726618e-01 3.62483063e-03
-7.32599199e-01 -7.33018577e-01 -8.89130116e-01 -7.85808980e-01
1.00817645e+00 7.79927552e-01 2.38663569e-01 -2.02342704e-01
8.73377919e-02 2.09188461e-01 -5.25338113e-01 -1.84741825e-01
2.40237594e-01 -3.64870638e-01 2.73692966e-01 2.98001766e-01
1.10755749e-01 -6.64994657e-01 -7.74164975e-01 1.73784941e-01
-4.93671447e-01 -1.08673321e-02 8.61075282e-01 7.17982411e-01
8.46093774e-01 3.29311818e-01 3.83006275e-01 -8.40842605e-01
5.70602357e-01 -2.53327906e-01 -9.59301054e-01 -7.74756819e-02
-6.49911225e-01 3.32351215e-02 6.43953919e-01 -1.14712127e-01
-8.54410529e-01 3.01938534e-01 -1.33426204e-01 -4.61579055e-01
3.61200958e-01 8.17386210e-01 3.05022262e-02 -8.15682828e-01
3.81139845e-01 7.54942477e-01 -1.77827962e-02 -4.67840165e-01
-3.97213027e-02 7.66196430e-01 3.71937811e-01 -3.31340402e-01
9.78901744e-01 5.05349517e-01 2.37828836e-01 -1.05786932e+00
-5.83064497e-01 -5.71658611e-01 -1.37792408e-01 2.10053489e-01
-8.74454528e-02 -1.37102866e+00 -6.91475809e-01 4.37311947e-01
-7.34578609e-01 1.11042768e-01 1.26921192e-01 8.95818353e-01
-2.03266039e-01 8.48778009e-01 -4.30876255e-01 -1.17210567e+00
-5.53983748e-01 -1.10622621e+00 6.90165639e-01 3.88065763e-02
-1.79028232e-02 -4.23977941e-01 -4.09526020e-01 -7.33102784e-02
5.99919915e-01 3.65198939e-03 8.07581723e-01 -8.59808847e-02
-6.40132189e-01 -2.62875795e-01 -5.74789762e-01 1.71177924e-01
-2.39791628e-02 -7.45536983e-01 -1.91202253e-01 -7.42212772e-01
1.77627280e-01 2.43382171e-01 5.58984697e-01 5.05859494e-01
1.01093614e+00 -5.81003547e-01 -4.72980618e-01 6.28815293e-01
1.58849502e+00 6.69202030e-01 6.38930798e-01 1.23801880e-01
4.30941701e-01 4.86157238e-02 8.95063043e-01 1.14114845e+00
1.49818018e-01 6.49092019e-01 1.53607100e-01 -1.59533352e-01
2.41271049e-01 7.12007582e-02 4.84127790e-01 1.26777399e+00
1.19553193e-01 1.76967680e-01 -5.12925446e-01 2.43268028e-01
-1.65948510e+00 -8.60955119e-01 -3.88501644e-01 2.43032646e+00
5.81461489e-01 2.18940794e-01 -2.65612632e-01 4.82684910e-01
6.91070914e-01 3.34411561e-01 -4.37975138e-01 -9.77473482e-02
-2.14987829e-01 3.62309277e-01 7.85441518e-01 4.91063595e-01
-8.49188089e-01 6.39820695e-01 5.05737495e+00 1.26727104e+00
-1.10047579e+00 -1.29263937e-01 2.63701349e-01 -4.66445414e-03
-1.95807904e-01 2.86185592e-01 -9.91898537e-01 8.11820745e-01
6.18681014e-01 -2.84577608e-01 5.82251728e-01 6.67970300e-01
3.43762308e-01 -3.56062353e-01 -3.88044059e-01 1.76303303e+00
1.77608982e-01 -1.15444887e+00 -3.12221404e-02 -5.19169606e-02
6.61107481e-01 -3.11812192e-01 4.50827591e-02 6.84834793e-02
-3.08936059e-01 -4.15229410e-01 5.00303388e-01 2.10564747e-01
1.05803072e+00 -7.04832971e-01 5.58488786e-01 4.86150771e-01
-1.64862216e+00 -2.63810903e-01 -3.24659854e-01 -1.13374041e-02
3.76238137e-01 8.65801990e-01 -3.96972358e-01 5.22026658e-01
5.23854434e-01 6.32350087e-01 2.47823671e-02 1.05127907e+00
1.94702763e-02 5.21865785e-01 -6.07242525e-01 2.53157318e-02
3.52688015e-01 -5.15438795e-01 7.87544787e-01 1.03774941e+00
9.27946568e-01 6.09897912e-01 4.15486515e-01 1.15224682e-01
1.53778657e-01 3.28421295e-01 -1.54644787e-01 2.82744016e-03
1.01133370e+00 8.62858117e-01 -2.83243209e-01 -3.14910680e-01
-5.89790821e-01 5.70662856e-01 -3.79598826e-01 3.27786118e-01
-6.29456162e-01 -7.99960315e-01 3.61511081e-01 3.64390731e-01
5.43237865e-01 -5.25677025e-01 -1.02803953e-01 -1.10617912e+00
-1.01008885e-01 -1.01993549e+00 2.10330650e-01 -4.03421193e-01
-7.12400198e-01 3.88088286e-01 -4.15430009e-01 -1.62124228e+00
-2.62937974e-02 -1.40882149e-01 -1.13410272e-01 8.97697747e-01
-1.65356123e+00 -4.96468812e-01 -3.11057031e-01 9.17204678e-01
3.07309628e-01 -4.84288126e-01 7.20990896e-01 6.20295405e-01
-6.91192269e-01 7.57816136e-01 5.14507055e-01 -1.31493464e-01
3.42792064e-01 -2.53408760e-01 -3.39438021e-01 9.31777120e-01
-2.07363069e-01 7.94514120e-01 7.59194791e-01 -4.55634385e-01
-1.61159050e+00 -8.51521611e-01 6.81925535e-01 8.95643711e-01
4.63787317e-01 1.73314080e-01 -6.74582720e-01 3.71744871e-01
-1.95570588e-01 -9.95645598e-02 8.82844567e-01 -1.97530940e-01
-2.41811290e-01 -5.19840837e-01 -9.66079712e-01 5.98536730e-01
8.22305143e-01 -3.28093797e-01 1.87556237e-01 1.43480480e-01
1.61710292e-01 -5.82039654e-01 -8.40093851e-01 7.31315136e-01
5.46435237e-01 -9.04601753e-01 8.75037134e-01 5.56652606e-01
-2.16301426e-01 -7.19032645e-01 -6.36996806e-01 -6.88637316e-01
-4.31452692e-01 -7.69642591e-01 -2.71950781e-01 6.58222258e-01
2.47924984e-01 -7.41267085e-01 8.02282631e-01 1.30405977e-01
-1.16096526e-01 -7.48536229e-01 -1.08434355e+00 -6.03286624e-01
-7.86970973e-01 -3.38421613e-01 1.51353955e-01 9.31733847e-01
4.50081564e-02 2.88340777e-01 -5.22834361e-01 4.20817763e-01
1.03130579e+00 2.40653127e-01 7.04091907e-01 -1.08959174e+00
-6.94793344e-01 2.71486696e-02 -3.56933057e-01 -1.52929020e+00
-2.53400177e-01 -4.79267746e-01 -3.02263409e-01 -9.59906280e-01
1.10379390e-01 -1.11390710e+00 -4.57095593e-01 2.23431721e-01
7.83129856e-02 4.26799022e-02 1.94492534e-01 4.25410777e-01
-6.34165466e-01 5.32971919e-01 9.75186706e-01 5.79104684e-02
-3.47940028e-01 9.18631777e-02 -8.78355205e-01 8.82577956e-01
4.82442021e-01 -2.58401841e-01 -3.47796381e-01 -4.48457211e-01
1.93089783e-01 6.88505709e-01 -9.30250958e-02 -1.17630935e+00
4.18215632e-01 1.22250646e-01 2.47564688e-01 -7.82856822e-01
5.81969321e-01 -8.07376146e-01 3.72179389e-01 7.64575481e-01
7.42079839e-02 -3.30978960e-01 2.39374250e-01 7.82333851e-01
-4.62222248e-01 -1.68154672e-01 6.62587106e-01 -1.00355577e-02
-9.05585885e-01 3.18168521e-01 -2.76510507e-01 -4.25497264e-01
9.11754310e-01 -3.66420805e-01 6.56479374e-02 -5.62755704e-01
-5.31091213e-01 2.15790182e-01 -3.82169299e-02 -2.43753180e-01
8.37398350e-01 -1.35894704e+00 -5.14370084e-01 5.28936267e-01
-1.71278998e-01 -1.37899116e-01 6.26252711e-01 1.11718869e+00
-5.11886716e-01 7.96296179e-01 1.12637892e-01 -5.51841974e-01
-1.22953808e+00 1.73565060e-01 -2.09302053e-01 -2.13883668e-01
-4.24845010e-01 9.81532097e-01 5.09256944e-02 1.61495328e-01
3.10459316e-01 3.34960699e-01 -1.05224863e-01 -1.24828272e-01
4.97032881e-01 6.13970160e-01 -2.43237287e-01 -9.12962914e-01
-3.26370001e-01 7.74267435e-01 -1.35658681e-01 -6.47714213e-02
1.16525972e+00 -3.67811143e-01 -3.54693532e-01 -6.44653291e-02
1.22539079e+00 6.21999085e-01 -1.06605804e+00 -6.52361751e-01
-2.77500033e-01 -5.77513576e-01 2.84171939e-01 -1.13588855e-01
-9.88343179e-01 5.70807815e-01 7.37573206e-01 -1.92917898e-01
1.37730515e+00 -4.66766149e-01 1.16391432e+00 2.26425424e-01
1.06691384e+00 -7.69786656e-01 -1.36761248e-01 5.11636257e-01
6.11282706e-01 -8.43840480e-01 4.20015544e-01 -9.26529944e-01
-4.97107327e-01 8.48214328e-01 1.88786879e-01 -2.19752207e-01
7.88987458e-01 1.65625155e-01 -4.31698143e-01 1.21068187e-01
-1.91671491e-01 -8.93146768e-02 -8.75352323e-02 1.07104175e-01
4.00046170e-01 1.86267646e-03 -6.95138276e-01 8.14162016e-01
1.42101198e-01 -3.61540914e-02 4.17135477e-01 9.89516973e-01
-7.45037198e-01 -1.16634011e+00 -6.98272824e-01 6.16565764e-01
-3.80587071e-01 -2.71723658e-01 4.50076252e-01 2.99081802e-01
-7.92020559e-02 8.59999239e-01 -1.99406013e-01 -3.02064329e-01
3.14864933e-01 -5.09246290e-01 3.49112660e-01 -3.54383528e-01
2.02913225e-01 8.47817540e-01 -6.79651201e-02 -5.39851248e-01
-3.07592243e-01 -7.08365381e-01 -1.45549941e+00 -2.94275552e-01
-3.76113951e-01 5.93856514e-01 4.40939426e-01 6.09218121e-01
4.57768738e-01 1.99654028e-01 1.01645756e+00 -5.60166776e-01
-4.73348022e-01 -5.06016016e-01 -1.08343685e+00 -1.60876200e-01
1.89022601e-01 -6.26372337e-01 -3.88191700e-01 -1.87244266e-01] | [6.454378604888916, 1.3502943515777588] |
8821d851-f5ab-4689-a2b2-25f1dbc44bad | towards-a-practical-lip-to-speech-conversion | 2104.14467 | null | https://arxiv.org/abs/2104.14467v1 | https://arxiv.org/pdf/2104.14467v1.pdf | Towards a practical lip-to-speech conversion system using deep neural networks and mobile application frontend | Articulatory-to-acoustic (forward) mapping is a technique to predict speech using various articulatory acquisition techniques as input (e.g. ultrasound tongue imaging, MRI, lip video). The advantage of lip video is that it is easily available and affordable: most modern smartphones have a front camera. There are already a few solutions for lip-to-speech synthesis, but they mostly concentrate on offline training and inference. In this paper, we propose a system built from a backend for deep neural network training and inference and a fronted as a form of a mobile application. Our initial evaluation shows that the scenario is feasible: a top-5 classification accuracy of 74% is combined with feedback from the mobile application user, making sure that the speaking impaired might be able to communicate with this solution. | ['Tamás Gábor Csapó', 'Frigyes Viktor Arthur'] | 2021-04-29 | null | null | null | null | ['lip-to-speech-synthesis'] | ['computer-vision'] | [-1.75601505e-02 3.10607761e-01 -5.95125020e-01 -3.27787876e-01
-9.99603391e-01 -1.48927480e-01 2.66807377e-01 -6.64515078e-01
-1.57789096e-01 5.77299297e-01 4.39581484e-01 -4.94896114e-01
2.84616888e-01 -2.30612472e-01 -6.45030022e-01 -4.91848558e-01
3.58233064e-01 2.81011939e-01 5.75429387e-02 1.33681595e-01
3.27208452e-02 2.81576842e-01 -1.91437399e+00 4.89676535e-01
7.59317160e-01 1.11808848e+00 7.62118697e-01 9.60717440e-01
-4.38869894e-01 5.81081510e-01 -4.50156480e-01 -3.59696329e-01
-2.01733425e-01 -2.24231854e-01 -4.60083902e-01 -2.29551777e-01
1.60072699e-01 -8.06405127e-01 -5.28472178e-02 7.73664951e-01
1.12981761e+00 -1.88370213e-01 4.11506653e-01 -1.22023714e+00
-2.91636318e-01 6.11475587e-01 7.16679990e-02 -3.50461245e-01
7.36949205e-01 2.44105846e-01 4.79038417e-01 -1.14216912e+00
5.31884909e-01 9.68824446e-01 8.85274589e-01 1.06996644e+00
-6.45483613e-01 -5.08233547e-01 -2.50362158e-01 2.89407253e-01
-1.07237029e+00 -1.39769173e+00 7.28806198e-01 -2.50541657e-01
1.01801920e+00 2.13553175e-01 6.04486048e-01 1.28289771e+00
-1.86858937e-01 1.11379921e+00 8.94151151e-01 -7.33115792e-01
1.73328370e-01 6.08295262e-01 -3.47207725e-01 4.16151285e-01
-2.96272814e-01 3.39765772e-02 -8.79147947e-01 9.68296900e-02
2.52900332e-01 -3.33800137e-01 -5.45470834e-01 -1.11882202e-02
-7.33540535e-01 5.27567863e-01 3.70711684e-02 2.99200684e-01
-5.93334317e-01 8.92558619e-02 5.41812837e-01 2.14583740e-01
4.86915439e-01 -2.20193684e-01 -4.70193535e-01 -4.48004276e-01
-1.24193954e+00 -1.34166211e-01 9.90293562e-01 7.16754258e-01
2.34082937e-01 1.67244598e-01 1.29051819e-01 1.13684642e+00
1.09505212e+00 6.87712133e-01 1.09583855e+00 -9.01211083e-01
3.75097573e-01 -7.84503147e-02 -2.06336737e-01 -1.97402954e-01
-3.03303331e-01 1.93740264e-01 -1.26275614e-01 5.23048282e-01
3.69264305e-01 -6.75648451e-01 -9.25243199e-01 1.56350183e+00
1.44202352e-01 3.88666809e-01 3.81137105e-03 8.13622057e-01
1.07203138e+00 5.11927605e-01 1.41250089e-01 -3.47292423e-01
1.27929521e+00 -1.03749812e+00 -1.14328790e+00 -3.36063862e-01
5.89803398e-01 -1.01984215e+00 1.12783110e+00 3.76677662e-01
-1.45712495e+00 -4.95908558e-01 -7.19336271e-01 -1.55264437e-01
-3.74115646e-01 3.66139382e-01 6.11421347e-01 1.25100374e+00
-1.57285476e+00 4.12423939e-01 -6.07556045e-01 -4.66382414e-01
1.07064076e-01 8.19009900e-01 -5.00327110e-01 6.47114888e-02
-1.05932152e+00 1.01247191e+00 -7.93995187e-02 5.79284355e-02
-5.78489244e-01 -4.32626128e-01 -8.84970307e-01 4.61166315e-02
-2.96997912e-02 -8.42362583e-01 1.46806061e+00 -9.88621950e-01
-2.57186007e+00 9.08704638e-01 -5.96807599e-01 -4.05983061e-01
6.90196514e-01 -2.07354724e-01 -3.97806317e-01 3.15651268e-01
-2.63211608e-01 8.80900681e-01 1.22099447e+00 -9.94140446e-01
-6.27840817e-01 -3.14395577e-01 -2.41076604e-01 3.78112555e-01
-3.87984961e-01 3.45425069e-01 -3.84664893e-01 -1.62810192e-01
-1.23247892e-01 -7.87826896e-01 4.04721886e-01 2.84547806e-01
-2.18811780e-01 -1.96672782e-01 1.06773472e+00 -1.15217400e+00
9.02573943e-01 -1.98950386e+00 -3.02599877e-01 -9.98053700e-02
-2.88113713e-01 7.05638885e-01 1.70656651e-01 2.07932726e-01
-2.47366980e-01 2.27116078e-01 4.03639786e-02 -9.49189365e-01
8.93899724e-02 1.57936402e-02 -1.21702775e-01 2.37722814e-01
-1.64565146e-01 8.82063806e-01 -4.45962399e-01 -5.64314961e-01
6.41497672e-01 8.97191703e-01 -5.22582710e-01 3.96764696e-01
-1.00649253e-01 2.61697888e-01 9.14252698e-02 8.07975054e-01
6.11295104e-01 3.16190779e-01 -4.14479040e-02 7.70260170e-02
-1.82702467e-01 7.15514123e-01 -8.22692394e-01 1.75341058e+00
-9.65996802e-01 9.08188820e-01 8.79159212e-01 -7.02431202e-01
5.24083674e-01 1.09970355e+00 3.73382181e-01 -1.80069059e-01
3.05896103e-01 4.95082498e-01 -2.53000677e-01 -1.24103773e+00
1.85621992e-01 -1.89974502e-01 6.92429543e-01 2.83494353e-01
1.69069111e-01 -1.42642573e-01 -4.54970658e-01 -3.57118964e-01
5.46234012e-01 5.15348375e-01 -7.09832609e-02 7.65032843e-02
6.01272285e-01 -5.74413180e-01 -6.64937869e-02 3.13442141e-01
-2.88797230e-01 6.66208148e-01 1.32652074e-01 2.43164346e-01
-6.58029258e-01 -7.47798800e-01 -3.87382647e-03 1.24073827e+00
-4.62905526e-01 -2.54564971e-01 -1.30434155e+00 -3.28549355e-01
-3.13723534e-01 6.64293349e-01 -1.32474944e-01 2.42328137e-01
-4.18677807e-01 -5.65007366e-02 5.82524240e-01 4.34125215e-01
1.19557776e-01 -1.40810502e+00 -2.91325361e-01 1.57238469e-01
-3.20675999e-01 -1.16135216e+00 -5.65729022e-01 -8.87290749e-04
-8.00522327e-01 -3.35949153e-01 -1.19430828e+00 -1.14455235e+00
2.36137807e-01 -1.66385081e-02 3.79550874e-01 -1.41209871e-01
2.02205613e-01 2.88124889e-01 4.07564044e-02 -5.63998997e-01
-7.91907728e-01 3.40809748e-02 4.14373338e-01 -8.43932759e-03
1.31516114e-01 -5.85140765e-01 -6.46993339e-01 -8.77602845e-02
-2.16555640e-01 2.78319847e-02 6.40332997e-01 6.20146871e-01
1.96972221e-01 -5.69425464e-01 8.27928483e-01 -3.39083523e-01
9.31399465e-01 -3.63904089e-01 -2.74095446e-01 1.56124420e-02
-6.29723430e-01 -4.38314527e-01 3.47234875e-01 -5.66542506e-01
-9.81018245e-01 3.67408305e-01 -1.19543397e+00 -6.17683530e-01
-4.38986242e-01 5.06691635e-01 -4.83009905e-01 4.26349305e-02
2.48608619e-01 1.25989273e-01 6.71421349e-01 -5.47756374e-01
3.63048345e-01 1.68385971e+00 5.09365201e-01 -7.64817372e-02
3.47423479e-02 1.54812708e-01 -6.32980824e-01 -1.14834237e+00
4.19658981e-02 -3.71411055e-01 -2.41450921e-01 -6.81292832e-01
7.11162388e-01 -8.08342338e-01 -1.36017454e+00 6.97596967e-01
-1.07538962e+00 -5.64861000e-01 -1.48651034e-01 8.28533292e-01
-8.16855669e-01 7.06179440e-02 -4.92737293e-01 -1.35309970e+00
-5.69349587e-01 -1.43301415e+00 1.07137966e+00 4.68281060e-01
-2.79390842e-01 -1.01367998e+00 -4.31602716e-01 8.61978650e-01
8.02862585e-01 -4.12028283e-01 5.28732836e-01 -3.75453353e-01
-1.86962262e-01 -1.78012028e-01 3.70839238e-02 4.71300781e-01
2.85704881e-01 8.62929448e-02 -1.76795518e+00 6.69576752e-04
2.58124202e-01 -2.92111516e-01 5.37591755e-01 9.94218946e-01
1.09016395e+00 -4.34331089e-01 -4.37194020e-01 4.72521722e-01
8.70463610e-01 1.74575120e-01 6.90461576e-01 -2.94763386e-01
3.49022239e-01 8.02865863e-01 2.00365737e-01 -1.79439932e-01
5.13342023e-01 7.84282327e-01 2.93890148e-01 -7.04431832e-02
-8.02000284e-01 -5.18121004e-01 9.65298712e-01 1.06345141e+00
1.41311139e-01 -1.91253439e-01 -7.65444517e-01 6.34305298e-01
-1.41258895e+00 -9.06015813e-01 1.10408455e-01 2.24953151e+00
6.81319773e-01 1.33070290e-01 2.16267958e-01 3.45589012e-01
9.55182493e-01 -2.63925731e-01 -3.87136936e-01 -7.63283432e-01
4.01058763e-01 3.45543593e-01 1.82240114e-01 1.00365925e+00
-6.92965627e-01 9.20141816e-01 6.83309507e+00 8.29553008e-01
-1.87344027e+00 3.03148955e-01 3.94211620e-01 -8.46752748e-02
-1.51033223e-01 -5.42026520e-01 -7.35854745e-01 6.98202670e-01
1.44328737e+00 1.51626140e-01 5.01976311e-01 9.08246040e-01
5.71944714e-01 -1.05836004e-01 -8.63924563e-01 9.55591738e-01
2.52195150e-01 -1.26971960e+00 -4.03289139e-01 6.30205944e-02
-1.94043498e-02 3.87931168e-01 3.04499537e-01 2.57739395e-01
-4.81282115e-01 -1.06491446e+00 6.86723650e-01 4.87536818e-01
1.46406400e+00 -4.30481970e-01 6.51600182e-01 4.94810998e-01
-8.45967710e-01 -2.45190691e-02 3.19730699e-01 7.32022673e-02
4.38821465e-01 2.75542326e-02 -1.43988669e+00 -1.23189976e-02
5.50437629e-01 9.52222720e-02 1.23943955e-01 1.07459998e+00
-3.26051980e-01 8.06275606e-01 -5.23133695e-01 -3.76666665e-01
-2.31664151e-01 2.98733890e-01 4.65700567e-01 1.20057774e+00
6.03629112e-01 -2.39840806e-01 -3.88961673e-01 4.81461734e-01
-9.27081481e-02 3.66926253e-01 -8.36063027e-01 1.88581288e-01
2.53158092e-01 1.00640464e+00 -2.23042354e-01 -1.79600324e-02
-4.23690110e-01 8.57771337e-01 -2.55176902e-01 4.50611860e-02
-3.70319158e-01 -3.78530651e-01 7.07390189e-01 4.12595153e-01
1.11892968e-01 1.00919023e-01 -3.55931818e-01 -8.77496481e-01
-4.56184521e-02 -8.39886904e-01 -2.77313232e-01 -1.44468141e+00
-7.74196088e-01 5.33530653e-01 -4.74886537e-01 -9.47047055e-01
-8.64628315e-01 -9.57850218e-01 -5.92419863e-01 1.09482396e+00
-1.70577705e+00 -1.38075912e+00 -8.14308003e-02 6.61996961e-01
8.55861008e-01 -2.05684200e-01 1.17049468e+00 6.23390734e-01
-1.59110308e-01 7.56222308e-01 -1.26689613e-01 3.62484343e-03
6.69207811e-01 -9.25480604e-01 1.03973038e-01 6.02568388e-01
-6.90157637e-02 6.89036608e-01 5.58387399e-01 -3.46031070e-01
-1.39485705e+00 -5.15676737e-01 1.22163963e+00 -2.21446112e-01
4.03249502e-01 -3.92544448e-01 -6.41061783e-01 3.83065075e-01
4.42680627e-01 -1.23592354e-01 7.98984885e-01 -9.49556455e-02
3.92812818e-01 -1.58329010e-01 -1.46582687e+00 6.64666831e-01
8.06897223e-01 -9.83544350e-01 -3.57363224e-01 3.61355245e-01
5.88234067e-01 -6.67468369e-01 -6.69732809e-01 1.87779874e-01
9.98744965e-01 -8.16364110e-01 7.37139285e-01 -8.53185877e-02
1.57889456e-01 4.18892652e-02 6.25972524e-02 -1.24417973e+00
6.62931859e-01 -1.07685697e+00 -3.58584672e-01 1.20504677e+00
6.49946809e-01 -8.15091789e-01 1.04350221e+00 7.63199389e-01
-2.69651413e-01 -7.37349510e-01 -1.19488883e+00 -3.46733242e-01
-1.31609783e-01 -9.04946983e-01 4.62058544e-01 4.79177743e-01
3.93786818e-01 3.06737334e-01 -4.03631896e-01 1.00140557e-01
7.13691637e-02 -4.09099013e-01 6.08601689e-01 -1.03570271e+00
-8.74296501e-02 -4.70940292e-01 -1.11627206e-01 -1.11976528e+00
4.06917095e-01 -9.14598405e-01 1.75073504e-01 -1.72901917e+00
-4.06180501e-01 -4.44181442e-01 1.56716675e-01 4.23385561e-01
2.67113477e-01 9.66756493e-02 1.38073534e-01 3.97399552e-02
2.63872445e-01 1.95591390e-01 1.12414873e+00 1.22698665e-01
-3.13555330e-01 7.08396077e-01 -4.50693280e-01 9.26345289e-01
8.07152271e-01 -2.80406654e-01 -3.60878378e-01 -4.22590584e-01
-1.70050234e-01 6.30447865e-01 2.33841874e-02 -9.10750628e-01
5.77524066e-01 4.27060664e-01 1.33442590e-02 -4.90897119e-01
1.03340852e+00 -8.52226555e-01 -9.29850787e-02 4.73969251e-01
-2.02457145e-01 -3.90802622e-01 2.43708238e-01 -5.76628111e-02
-1.78972542e-01 -6.32816434e-01 6.32318974e-01 1.18047625e-01
-4.20673937e-01 -3.46369259e-02 -6.79816544e-01 -4.35469419e-01
6.43585384e-01 -4.11292672e-01 -1.64244473e-01 -9.00809705e-01
-1.06740594e+00 -2.23042980e-01 1.01265803e-01 3.57983410e-01
6.72619164e-01 -9.53680992e-01 -5.48649788e-01 5.67733288e-01
-3.30816805e-01 -3.81955355e-01 2.18406878e-02 9.69297826e-01
-5.38645506e-01 5.37092924e-01 -9.21573117e-02 -6.09513342e-01
-1.57654512e+00 1.70466915e-01 5.72215199e-01 2.70671248e-01
-6.03501081e-01 9.34267759e-01 -4.52028751e-01 -5.51904917e-01
5.38576841e-01 -3.78425062e-01 -3.18381935e-01 1.82860374e-01
6.63352966e-01 2.00079218e-01 3.06510687e-01 -7.75703728e-01
-4.19357151e-01 5.85490406e-01 3.30530405e-01 -6.16827369e-01
1.27565026e+00 -3.11622471e-01 1.61075577e-01 3.37492555e-01
1.16390347e+00 4.64861572e-01 -9.62374449e-01 3.33839089e-01
-2.73368597e-01 -5.52363060e-02 3.72867316e-01 -9.92997527e-01
-1.00600779e+00 1.34595621e+00 9.06007886e-01 2.42176548e-01
1.17557418e+00 -6.25749677e-02 8.81483376e-01 2.24367559e-01
1.95943505e-01 -1.21768177e+00 -3.34446192e-01 2.57334083e-01
9.56398010e-01 -1.19872308e+00 -4.98543143e-01 -4.16117042e-01
-6.09645545e-01 1.12180769e+00 1.53000116e-01 3.42334300e-01
9.05952275e-01 7.47470140e-01 4.68155563e-01 3.58473063e-01
-6.08587980e-01 -3.77520084e-01 6.41350597e-02 1.05341017e+00
5.11422992e-01 1.36131674e-01 -2.42287293e-01 4.98721004e-01
-4.33835149e-01 6.05744660e-01 2.41184115e-01 3.83810133e-01
-3.36851805e-01 -1.13036931e+00 -4.16254640e-01 3.04667324e-01
-7.88791537e-01 -2.32167959e-01 5.50849140e-02 3.78540307e-01
2.79607832e-01 1.23910952e+00 -2.03657329e-01 -2.64646173e-01
9.85568762e-02 6.30925357e-01 3.19926560e-01 -4.37405705e-01
-5.27112663e-01 2.05250993e-01 3.40038478e-01 -4.20893788e-01
-1.39949173e-01 -8.31485391e-01 -1.13891280e+00 -1.70349225e-01
-6.00139737e-01 -1.61853820e-01 1.64321303e+00 8.97533298e-01
4.18586373e-01 3.99879098e-01 4.25719559e-01 -1.18278182e+00
-2.41081789e-01 -1.23347771e+00 -2.69872069e-01 -2.91358113e-01
6.93249464e-01 -3.23796332e-01 -4.69204366e-01 2.01635405e-01] | [14.297337532043457, 5.013840675354004] |
91856649-2ae2-4929-9d84-3f6b40977f98 | adaptive-model-predictive-control-by-learning | 2203.06783 | null | https://arxiv.org/abs/2203.06783v2 | https://arxiv.org/pdf/2203.06783v2.pdf | Adaptive Model Predictive Control by Learning Classifiers | Stochastic model predictive control has been a successful and robust control framework for many robotics tasks where the system dynamics model is slightly inaccurate or in the presence of environment disturbances. Despite the successes, it is still unclear how to best adjust control parameters to the current task in the presence of model parameter uncertainty and heteroscedastic noise. In this paper, we propose an adaptive MPC variant that automatically estimates control and model parameters by leveraging ideas from Bayesian optimisation (BO) and the classical expected improvement acquisition function. We leverage recent results showing that BO can be reformulated via density ratio estimation, which can be efficiently approximated by simply learning a classifier. This is then integrated into a model predictive path integral control framework yielding robust controllers for a variety of challenging robotics tasks. We demonstrate the approach on classical control problems under model uncertainty and robotics manipulation tasks. | ['Fabio Ramos', 'Rafael Oliveira', 'Rel Guzman'] | 2022-03-13 | null | null | null | null | ['density-ratio-estimation', 'bayesian-optimisation'] | ['methodology', 'methodology'] | [ 5.38766444e-01 2.34339774e-01 -2.62077481e-01 1.16439201e-01
-6.94140613e-01 -5.39352298e-01 7.41093278e-01 -6.51105791e-02
-4.31216389e-01 1.00462866e+00 -3.41841042e-01 -1.96126521e-01
-8.31678212e-01 -3.08154225e-01 -8.81349683e-01 -9.58819926e-01
6.98447227e-02 7.16802061e-01 2.96312809e-01 6.90264627e-02
4.56939876e-01 4.62133884e-01 -1.37608767e+00 -7.52893388e-01
8.75062108e-01 9.59539175e-01 4.98057932e-01 8.95761251e-01
5.68314493e-01 5.30700684e-01 -1.81829721e-01 2.06755683e-01
3.66780937e-01 3.96395065e-02 -3.32828134e-01 1.97408304e-01
-1.70068577e-01 -1.84563071e-01 -5.66649809e-02 1.12174273e+00
4.14459586e-01 6.65080011e-01 9.05970693e-01 -1.25168848e+00
-1.87727451e-01 1.10684879e-01 -2.36299068e-01 -3.04683805e-01
-8.81756172e-02 3.52570325e-01 5.56423306e-01 -4.21606064e-01
4.00121957e-01 1.40963638e+00 5.98055243e-01 1.97542220e-01
-1.61657643e+00 -3.26819748e-01 3.01539898e-01 -8.42738971e-02
-1.14738250e+00 -1.59837514e-01 3.09456229e-01 -9.32729602e-01
6.08842373e-01 -3.08588117e-01 3.30843806e-01 9.47699547e-01
7.89045751e-01 4.57116604e-01 1.07358551e+00 -3.62461925e-01
5.73640823e-01 4.21114638e-02 -2.03986123e-01 3.81914735e-01
3.18910301e-01 7.04156935e-01 -6.02521040e-02 -1.70947790e-01
7.17132568e-01 -1.02687605e-01 -3.20856005e-01 -1.01325822e+00
-1.19144404e+00 9.90612745e-01 1.58986822e-01 -2.78478086e-01
-5.65301955e-01 4.92558658e-01 1.38986707e-01 1.61680296e-01
1.97031483e-01 7.70354211e-01 -6.15814865e-01 -3.13192636e-01
-2.55191028e-01 8.03295910e-01 1.07389843e+00 1.28676176e+00
4.21181589e-01 1.98467568e-01 -7.03047886e-02 6.91525996e-01
4.97872353e-01 6.39654398e-01 1.81563795e-01 -1.62255716e+00
1.02887310e-01 4.91462066e-04 8.84887695e-01 -6.48994327e-01
-3.42195243e-01 -3.91201407e-01 -4.41466749e-01 6.51399970e-01
5.20425737e-01 -2.62693703e-01 -8.84265125e-01 1.63801837e+00
3.89342666e-01 -5.21836318e-02 -7.37139881e-02 7.04538286e-01
-6.15597069e-01 4.91380781e-01 -2.31480017e-01 -2.86840975e-01
7.41233766e-01 -5.71512640e-01 -9.02255595e-01 -3.89066607e-01
2.39346847e-01 -5.87386429e-01 7.07063615e-01 9.07862425e-01
-8.67339909e-01 -2.72233695e-01 -1.13701093e+00 3.65314275e-01
-1.44167662e-01 -8.17516074e-03 7.17628524e-02 4.08507377e-01
-6.75878167e-01 1.06499970e+00 -1.03453779e+00 -2.90753841e-01
1.98550493e-01 5.01425445e-01 1.67286918e-02 -9.16591063e-02
-6.76431179e-01 1.36528504e+00 7.35140741e-01 2.25645259e-01
-1.11794007e+00 -6.25056446e-01 -7.52006888e-01 -3.35995376e-01
1.13605690e+00 -7.16257334e-01 1.66816390e+00 -2.41132364e-01
-2.26154804e+00 -1.30402759e-01 2.25648075e-01 -6.69001937e-01
6.15437031e-01 -6.97320879e-01 4.13775176e-01 -8.99205953e-02
-1.72658309e-01 3.44399363e-01 1.35224235e+00 -1.04505265e+00
-5.42718589e-01 -3.84849638e-01 -1.04718439e-01 1.24987528e-01
3.20190877e-01 -2.76770353e-01 5.15499786e-02 -4.15029824e-01
3.77183519e-02 -1.49056911e+00 -7.80985415e-01 1.09741300e-01
-2.45609030e-01 -7.57852197e-02 6.17644727e-01 -4.45228904e-01
6.49412572e-01 -1.58947182e+00 6.42996132e-01 3.25582027e-01
-2.96340406e-01 -2.96875965e-02 1.62786081e-01 4.58928168e-01
2.20161214e-01 -1.36678308e-01 -5.21266639e-01 1.55266359e-01
3.95073384e-01 4.44318265e-01 -6.54587805e-01 7.49053061e-01
4.86064345e-01 5.33403337e-01 -1.03039110e+00 -5.12014590e-02
3.65558863e-01 3.00067753e-01 -5.40811121e-01 1.63736746e-01
-5.11599302e-01 7.62908041e-01 -6.96439683e-01 2.55345404e-01
2.10046545e-01 3.00923288e-01 1.33404462e-02 2.80616134e-01
-2.95520127e-01 -1.34612560e-01 -1.57731879e+00 1.40399861e+00
-4.76697356e-01 3.49038363e-01 5.68241298e-01 -1.18127227e+00
9.60323870e-01 1.08617231e-01 3.29261214e-01 1.98854044e-01
5.74289262e-01 2.77461380e-01 -1.43668473e-01 -4.13395137e-01
4.08104539e-01 -3.83940369e-01 -9.81372073e-02 -5.00856973e-02
1.53558314e-01 -1.14471614e+00 -5.16952053e-02 -3.75505865e-01
1.06196797e+00 8.09271693e-01 5.52847326e-01 -3.37988406e-01
3.09564471e-01 1.75720975e-01 5.00108480e-01 9.12048638e-01
-3.32087487e-01 4.06050950e-01 5.98177791e-01 3.05525959e-01
-9.09000814e-01 -1.04221272e+00 -2.30532616e-01 8.13025177e-01
-4.89825159e-02 5.68927713e-02 -4.23314393e-01 -3.53648961e-02
2.68205822e-01 9.02917743e-01 -4.25098389e-01 -4.89666313e-01
-4.28777546e-01 -5.91473520e-01 -1.60597116e-01 4.98060852e-01
-2.34933406e-01 -4.12480325e-01 -6.21179819e-01 5.79555392e-01
4.99847770e-01 -9.98877525e-01 -1.97833359e-01 5.15991867e-01
-8.62973988e-01 -1.05213428e+00 -6.41016543e-01 -1.74540639e-01
3.07133853e-01 -3.53153765e-01 5.06458282e-01 -6.90651417e-01
-3.14453632e-01 8.91903341e-01 4.17255946e-02 -8.96910965e-01
-5.65011263e-01 -1.23068430e-01 4.40794528e-01 -1.29652143e-01
-4.57966506e-01 -2.95740783e-01 -8.70892555e-02 3.74496132e-01
-5.69552183e-01 -3.90932232e-01 3.72814864e-01 1.05808616e+00
7.60978937e-01 2.76799768e-01 6.05558693e-01 -5.09414196e-01
6.88196540e-01 -5.08596003e-01 -1.39790940e+00 -1.14814803e-01
-7.76761532e-01 2.09092930e-01 4.16818142e-01 -7.30572641e-01
-9.57106352e-01 6.29517257e-01 4.58937585e-01 -6.52714491e-01
-5.01896590e-02 4.66936350e-01 8.12029839e-03 -6.77278936e-02
2.94752419e-01 -1.65709123e-01 5.79466105e-01 -3.83168578e-01
4.16153222e-01 5.09662390e-01 6.67955279e-01 -8.71510148e-01
7.28878856e-01 2.46322215e-01 5.67011476e-01 -6.80443048e-01
-6.02918863e-01 -3.63611817e-01 -7.59493351e-01 -2.85245448e-01
7.56456971e-01 -8.24324727e-01 -8.26990664e-01 3.80004197e-01
-9.32242036e-01 -7.14121461e-01 -4.55182195e-01 7.78073609e-01
-1.46561241e+00 -7.65423232e-04 -1.90611944e-01 -1.53933859e+00
3.30033422e-01 -1.18572581e+00 9.75894868e-01 1.71851218e-01
-1.06661871e-01 -9.52106237e-01 9.64820236e-02 -1.74588755e-01
3.84018004e-01 4.15073723e-01 5.82770646e-01 -4.31562781e-01
-5.30640364e-01 -5.72374582e-01 2.04402208e-01 5.04956067e-01
-1.60776898e-01 7.29009211e-02 -6.69314384e-01 -3.02385718e-01
3.48606855e-01 -1.84250638e-01 6.36665046e-01 8.47796202e-01
7.25020468e-01 -1.86711535e-01 -4.51893151e-01 1.67417809e-01
1.34832394e+00 4.24417973e-01 2.53706146e-02 2.92315423e-01
4.44185942e-01 9.93502319e-01 1.07433665e+00 4.72973436e-01
5.04200980e-02 7.83404768e-01 5.90216458e-01 9.36389506e-01
6.37482226e-01 -1.92855790e-01 4.59522009e-01 1.90270036e-01
2.00031072e-01 9.86487791e-02 -8.97240639e-01 5.88334441e-01
-2.14635396e+00 -6.55992568e-01 -8.49495828e-03 2.46887279e+00
6.08748317e-01 1.76657379e-01 7.20564425e-02 -1.37518987e-01
7.97060013e-01 -4.87853497e-01 -8.30199122e-01 -3.20319802e-01
4.03237671e-01 -5.94580062e-02 1.02898932e+00 7.08392739e-01
-1.16670907e+00 4.28717345e-01 6.59327793e+00 7.97259688e-01
-6.25289381e-01 -5.36017567e-02 1.23186693e-01 -3.03979069e-02
4.29648608e-01 1.17983960e-01 -8.44472945e-01 3.61234576e-01
1.08098686e+00 -4.45079148e-01 6.47800863e-01 1.11398494e+00
4.77682143e-01 -5.48713386e-01 -1.07308161e+00 5.91166079e-01
-2.91698337e-01 -6.51822507e-01 -5.90247154e-01 1.53708637e-01
8.21355879e-01 -4.35322337e-03 3.04673195e-01 4.75545794e-01
7.75915265e-01 -8.30758929e-01 8.35361898e-01 9.15431976e-01
1.06968477e-01 -6.47338033e-01 5.99587977e-01 7.05888748e-01
-6.99860215e-01 -6.84696436e-01 -3.17608625e-01 -2.00766116e-01
4.24414158e-01 4.12612081e-01 -1.01298189e+00 4.63578403e-01
3.42785269e-01 7.21879244e-01 -4.28899191e-02 1.29908776e+00
-2.08037555e-01 3.02702159e-01 -6.43172145e-01 -1.85918540e-01
7.73700476e-02 -4.78717834e-01 1.07076073e+00 6.76923513e-01
5.15505195e-01 -3.01043063e-01 4.02028799e-01 1.12000108e+00
7.65105128e-01 -3.14937919e-01 -8.99505973e-01 -6.59264699e-02
3.14051688e-01 9.52822089e-01 -4.48324233e-01 6.90498129e-02
1.49299353e-01 4.01015282e-01 1.10430703e-01 5.01471698e-01
-5.33180714e-01 -4.77845371e-01 4.62143213e-01 -3.14280748e-01
6.08724475e-01 -6.12689018e-01 -1.11369357e-01 -6.93388522e-01
-7.55117238e-02 -4.51755196e-01 3.84031609e-02 -7.43874252e-01
-1.20135558e+00 -1.59198299e-01 5.70338726e-01 -1.23706853e+00
-8.22745860e-01 -1.05206752e+00 -2.97105253e-01 7.49436140e-01
-1.30461407e+00 -5.33356428e-01 3.68587196e-01 6.59683049e-02
4.56606209e-01 2.05096919e-02 4.08129126e-01 -4.08910453e-01
-6.75812423e-01 -2.86969870e-01 5.88177264e-01 -6.42802238e-01
7.87039638e-01 -1.64686120e+00 -4.18081373e-01 7.30949879e-01
-7.73152769e-01 5.64727604e-01 1.28007436e+00 -7.53216386e-01
-1.59767640e+00 -1.30286443e+00 8.39014426e-02 -5.76954246e-01
1.24071205e+00 -9.85232741e-02 -8.31694305e-01 6.29408956e-01
-2.61957079e-01 2.87373569e-02 -2.44786710e-01 -2.96979815e-01
2.18548298e-01 2.54126880e-02 -1.08305621e+00 5.76405704e-01
5.47936141e-01 -1.08450621e-01 -6.52995944e-01 2.33333036e-01
8.33829224e-01 -5.54984868e-01 -1.10112906e+00 6.43856764e-01
5.21047235e-01 -1.94201425e-01 7.76484251e-01 -4.63850111e-01
-7.37865046e-02 -3.14670414e-01 -3.44192922e-01 -1.63819385e+00
3.68720898e-03 -1.27331603e+00 -2.45653033e-01 8.87231588e-01
8.30101147e-02 -7.50942528e-01 1.95092842e-01 7.48225570e-01
-2.89299101e-01 -6.72729790e-01 -1.22668386e+00 -1.26528835e+00
3.64077449e-01 -7.14722335e-01 -7.00985715e-02 1.55792192e-01
-3.81341800e-02 2.05049798e-01 -2.93162555e-01 3.49649638e-01
7.95222461e-01 -2.18759730e-01 8.00458550e-01 -1.21171474e+00
-4.77734596e-01 -4.51033264e-01 -3.11219484e-01 -8.95176888e-01
4.75294143e-01 -3.40464145e-01 1.11964262e+00 -1.10939837e+00
-3.06105375e-01 -2.76699841e-01 -6.79002106e-02 -5.46646230e-02
-2.05192775e-01 -4.10213172e-01 1.39759585e-01 -1.05798319e-01
-3.65390748e-01 8.53997409e-01 1.04159498e+00 -3.59334275e-02
-4.27268535e-01 5.71370244e-01 -1.45274609e-01 8.32832992e-01
8.22540820e-01 -4.58299845e-01 -6.23326540e-01 1.11670554e-01
1.35199994e-01 1.59541920e-01 5.11217415e-01 -1.07638717e+00
2.16114044e-01 -5.14657319e-01 -1.41181976e-01 -3.95304412e-01
4.08334017e-01 -1.11037600e+00 2.10990131e-01 6.50029600e-01
-3.91588449e-01 -2.05736265e-01 2.53635257e-01 1.43225026e+00
1.77837849e-01 -5.37774801e-01 9.36930954e-01 1.38789281e-01
-3.89599919e-01 1.41408116e-01 -7.82509089e-01 -1.82534978e-01
1.08084011e+00 7.67330602e-02 1.80609822e-01 -4.42019165e-01
-9.78438139e-01 3.75386298e-01 3.27981532e-01 3.63093793e-01
3.87210310e-01 -8.79407585e-01 -4.10168529e-01 -3.93386707e-02
-9.23807770e-02 8.68737027e-02 -3.12913693e-02 1.13454676e+00
3.66635360e-02 4.88808900e-01 7.33676478e-02 -8.99214864e-01
-5.79487503e-01 6.55279398e-01 4.95213181e-01 -1.21825442e-01
-3.74394625e-01 4.24925745e-01 -9.95997563e-02 -5.99198699e-01
2.69860953e-01 -7.15671360e-01 1.00235827e-01 -1.10085733e-01
3.18119228e-01 6.27892137e-01 -2.08896041e-01 -3.24829519e-01
1.14654385e-01 5.75581253e-01 3.47519636e-01 -5.12425601e-01
1.16367209e+00 -4.86365229e-01 2.18646660e-01 9.86013472e-01
6.82359338e-01 -5.52268863e-01 -2.10749006e+00 -1.04023926e-01
4.86557007e-01 -2.53923804e-01 2.11607248e-01 -5.44883847e-01
-3.82642925e-01 7.27461576e-01 5.76793313e-01 5.08244745e-02
5.70403576e-01 -3.10646147e-01 1.07457146e-01 6.78113937e-01
5.84467173e-01 -1.25587034e+00 -1.22404061e-01 8.68398547e-01
1.13575184e+00 -9.97304380e-01 5.74802533e-02 -2.47063547e-01
-5.50021887e-01 1.13659322e+00 4.06487435e-01 -6.34725690e-01
9.07564878e-01 5.01515985e-01 -5.55680871e-01 3.80059779e-01
-9.36457276e-01 -2.39956245e-01 1.33036703e-01 9.32390749e-01
-2.11634144e-01 -8.81327540e-02 -1.76463593e-02 5.15815079e-01
1.22715168e-01 3.19338232e-01 5.66343725e-01 1.17715752e+00
-7.54306197e-01 -7.60192275e-01 -5.67624390e-01 3.25540364e-01
-2.94495046e-01 4.23539907e-01 1.63428605e-01 8.63610446e-01
-4.04118061e-01 1.08162367e+00 -1.43207684e-01 -6.67701736e-02
4.83877569e-01 2.58481085e-01 6.97294891e-01 -7.32322454e-01
7.97971524e-03 3.13265711e-01 -6.27205372e-02 -7.15506136e-01
-3.13683480e-01 -8.99738073e-01 -9.50580001e-01 3.95184100e-01
-7.12093771e-01 -1.94480971e-01 7.38314390e-01 1.07800734e+00
2.08522767e-01 6.16372406e-01 3.38084906e-01 -1.40159738e+00
-1.52010727e+00 -1.04864848e+00 -6.37907624e-01 -5.35533905e-01
4.75461215e-01 -1.38142645e+00 -5.07727623e-01 1.60190731e-01] | [4.87252140045166, 2.3393614292144775] |
97aae79e-ece8-4d95-963d-26ee5bdf8310 | cocktail-mixing-multi-modality-controls-for | 2306.00964 | null | https://arxiv.org/abs/2306.00964v1 | https://arxiv.org/pdf/2306.00964v1.pdf | Cocktail: Mixing Multi-Modality Controls for Text-Conditional Image Generation | Text-conditional diffusion models are able to generate high-fidelity images with diverse contents. However, linguistic representations frequently exhibit ambiguous descriptions of the envisioned objective imagery, requiring the incorporation of additional control signals to bolster the efficacy of text-guided diffusion models. In this work, we propose Cocktail, a pipeline to mix various modalities into one embedding, amalgamated with a generalized ControlNet (gControlNet), a controllable normalisation (ControlNorm), and a spatial guidance sampling method, to actualize multi-modal and spatially-refined control for text-conditional diffusion models. Specifically, we introduce a hyper-network gControlNet, dedicated to the alignment and infusion of the control signals from disparate modalities into the pre-trained diffusion model. gControlNet is capable of accepting flexible modality signals, encompassing the simultaneous reception of any combination of modality signals, or the supplementary fusion of multiple modality signals. The control signals are then fused and injected into the backbone model according to our proposed ControlNorm. Furthermore, our advanced spatial guidance sampling methodology proficiently incorporates the control signal into the designated region, thereby circumventing the manifestation of undesired objects within the generated image. We demonstrate the results of our method in controlling various modalities, proving high-quality synthesis and fidelity to multiple external signals. | ['Tat-Jen Cham', 'DaCheng Tao', 'Chaoyue Wang', 'Chuanxia Zheng', 'Daqing Liu', 'Jianbin Zheng', 'Minghui Hu'] | 2023-06-01 | null | null | null | null | ['conditional-image-generation'] | ['computer-vision'] | [ 5.51487148e-01 2.00783357e-01 9.73056257e-02 -3.77495331e-03
-3.91899258e-01 -6.69908047e-01 1.32413495e+00 -2.21990287e-01
-2.04197600e-01 5.19688785e-01 4.48191881e-01 1.36098385e-01
-2.35791162e-01 -8.59234393e-01 -6.38139784e-01 -1.18248105e+00
3.22822601e-01 1.74487889e-01 1.55478045e-01 -4.01506782e-01
-2.46370137e-02 6.35010839e-01 -1.50412107e+00 2.46557906e-01
9.55684841e-01 1.00772190e+00 4.58892673e-01 4.95908290e-01
1.10821063e-02 6.93355739e-01 -4.84475225e-01 2.34660879e-02
7.82300755e-02 -5.31255722e-01 -1.36670187e-01 2.79567748e-01
3.79785389e-01 -1.78806245e-01 -3.42859477e-01 8.36082697e-01
6.27310097e-01 8.63925219e-02 7.62600482e-01 -1.06454778e+00
-9.14297104e-01 5.96124947e-01 -3.23892295e-01 -2.49197856e-01
4.62515265e-01 6.94234371e-01 7.87566543e-01 -7.74420679e-01
1.11056399e+00 9.51826334e-01 1.25563756e-01 7.23213375e-01
-1.71811056e+00 -3.82050514e-01 1.77647129e-01 -2.83302963e-01
-1.03261375e+00 -6.63045704e-01 1.01035976e+00 -7.09983945e-01
5.06002009e-01 1.80227920e-01 9.44626331e-01 1.53347242e+00
6.61204085e-02 6.17915452e-01 1.19190156e+00 -3.64347786e-01
3.58050406e-01 2.99089521e-01 -5.13698220e-01 5.02077579e-01
-2.63416708e-01 5.69648862e-01 -6.80457354e-01 2.80549586e-01
1.00109816e+00 -3.02349240e-01 -6.33724391e-01 -5.25040269e-01
-1.63337445e+00 5.53375900e-01 7.38751769e-01 5.83485126e-01
-7.25974977e-01 4.65853140e-02 -7.17207417e-02 1.32654682e-02
1.61927491e-01 5.33198953e-01 5.18043153e-02 3.49743605e-01
-1.03542912e+00 7.54159018e-02 2.83083588e-01 8.64299119e-01
6.82343423e-01 4.45791364e-01 -6.67301357e-01 5.96752286e-01
2.72945374e-01 6.49268925e-01 3.33298117e-01 -9.98151779e-01
2.87018776e-01 6.12658799e-01 1.87632479e-02 -9.87243056e-01
-4.51939672e-01 -6.45197928e-01 -1.12901831e+00 2.99534827e-01
1.81590393e-01 -1.64194569e-01 -8.90923262e-01 2.02557540e+00
3.59409690e-01 -5.68737127e-02 6.65327683e-02 1.15130365e+00
5.17489254e-01 6.70070112e-01 1.59257829e-01 -2.23424435e-01
9.83671904e-01 -7.09501743e-01 -7.96582222e-01 -2.18063071e-02
3.24500442e-01 -3.58898938e-01 1.14578342e+00 3.52811784e-01
-1.13305318e+00 -6.54455245e-01 -1.12421989e+00 -2.48959148e-03
-2.69317567e-01 2.15826288e-01 3.65591377e-01 2.85893083e-01
-1.08796859e+00 2.13221714e-01 -4.79825050e-01 -3.80860381e-02
3.12886745e-01 2.08926767e-01 -5.82534015e-01 -5.43746240e-02
-1.13796103e+00 6.25605166e-01 2.85122216e-01 2.80899525e-01
-1.15153968e+00 -7.69380927e-01 -8.93232882e-01 8.59139953e-03
2.33460754e-01 -1.00790596e+00 3.32480103e-01 -9.42881584e-01
-1.81184506e+00 5.80665529e-01 2.98255950e-01 -8.05336460e-02
6.10072732e-01 2.77426213e-01 -4.68517900e-01 3.36050361e-01
1.31981224e-01 1.16227829e+00 1.37836730e+00 -1.52298582e+00
-1.68358132e-01 -2.02134460e-01 8.01715404e-02 1.53757334e-01
-4.05755967e-01 -5.45697451e-01 -4.26503718e-01 -8.90788436e-01
7.17059374e-02 -7.13360369e-01 -1.29170075e-01 3.12352389e-01
-5.82285881e-01 3.56473625e-01 7.45747864e-01 -2.25974455e-01
1.06119502e+00 -2.58545017e+00 8.74594092e-01 3.06446791e-01
2.61589110e-01 2.07514953e-04 -5.95106602e-01 3.89675766e-01
-1.69961154e-01 -1.28430471e-01 -5.51663935e-01 -3.49029034e-01
1.30828813e-01 1.41891301e-01 -2.37041518e-01 4.18472201e-01
3.97856444e-01 9.48908031e-01 -8.76449943e-01 -3.04974198e-01
5.22148430e-01 7.97460914e-01 -6.31015003e-01 1.52744234e-01
-6.67390883e-01 1.02574515e+00 -3.36557329e-01 5.24480283e-01
5.28439403e-01 -8.27701539e-02 1.20617211e-01 -6.19023263e-01
-2.36234501e-01 -1.81762040e-01 -1.14179349e+00 2.02695131e+00
-5.54323792e-01 3.22514176e-01 5.14482915e-01 -5.01979768e-01
9.14690435e-01 1.85167149e-01 4.38019365e-01 -9.19292569e-01
1.76413506e-01 2.33853817e-01 -2.60452837e-01 -4.32038695e-01
4.63508695e-01 -2.83362061e-01 2.75950339e-02 2.81227887e-01
2.59819865e-01 -6.13936007e-01 2.10587934e-01 3.10376108e-01
6.28056347e-01 3.63878876e-01 -3.45554918e-01 -1.14229493e-01
7.62435079e-01 -1.84878543e-01 1.00160040e-01 6.33963585e-01
5.91993891e-02 7.29212165e-01 5.06763160e-01 1.19916476e-01
-9.38256025e-01 -1.02412570e+00 -3.24298232e-03 7.95668900e-01
1.93488613e-01 -9.55770761e-02 -5.31600296e-01 -6.69845939e-01
-1.89378649e-01 7.61240184e-01 -6.46561623e-01 -1.09254077e-01
-2.51183063e-01 -4.89617109e-01 4.95868266e-01 3.88722345e-02
5.49398541e-01 -9.23816442e-01 -4.01342541e-01 2.36294895e-01
-1.18094087e-01 -1.03548157e+00 -4.54983383e-01 1.30480453e-01
-4.20710802e-01 -7.63586760e-01 -8.92991245e-01 -4.28370655e-01
6.66466832e-01 -6.88659251e-02 5.76380491e-01 -1.67923629e-01
3.67098339e-02 5.19160926e-01 -5.30118719e-02 3.01441431e-01
-5.77877104e-01 -8.31103977e-03 -7.46510252e-02 5.26063621e-01
-5.57013690e-01 -7.34342158e-01 -7.57496834e-01 2.90532500e-01
-1.40160000e+00 3.08509409e-01 4.46216583e-01 1.00380898e+00
4.94801670e-01 -2.34245449e-01 5.15902579e-01 -3.16340923e-01
6.24231040e-01 -3.25318873e-01 -5.83762467e-01 1.94935873e-01
-2.00959742e-01 1.95193067e-01 5.94180644e-01 -8.10072064e-01
-1.00504196e+00 1.52078420e-01 -1.77518770e-01 -5.47869563e-01
-1.00584060e-01 4.36339200e-01 -4.00010735e-01 -1.84681416e-01
9.54717398e-01 6.08257353e-01 2.97141850e-01 4.48503308e-02
9.76590216e-01 2.76506931e-01 4.83310938e-01 -6.27924621e-01
7.14039862e-01 6.52405977e-01 6.50754338e-03 -8.91835272e-01
-2.49080241e-01 3.01027209e-01 -5.02874792e-01 -6.48421764e-01
1.07394874e+00 -7.69295812e-01 -4.31432158e-01 5.70460796e-01
-9.52939987e-01 -6.60538316e-01 -6.26722395e-01 4.38789308e-01
-6.63170993e-01 1.39597088e-01 -3.43730450e-01 -4.00505871e-01
1.06838815e-01 -1.52527988e+00 1.21461415e+00 8.00741464e-02
-3.19454484e-02 -1.04928124e+00 -1.37395442e-01 3.07374388e-01
6.69064522e-01 4.93808985e-01 8.78054500e-01 -9.72162634e-02
-8.40741098e-01 -3.13909128e-02 -1.32746428e-01 3.06523412e-01
2.20473204e-02 4.75262217e-02 -9.32080567e-01 -1.41890228e-01
-5.94189204e-02 -3.92081916e-01 8.78992081e-01 3.73934209e-01
6.75766170e-01 -1.58926889e-01 -2.02955544e-01 6.12818778e-01
1.32458472e+00 -8.70636851e-02 5.79578221e-01 2.06314668e-01
6.49846017e-01 7.50352085e-01 2.47508615e-01 3.94169122e-01
1.27201810e-01 8.45714450e-01 5.96957862e-01 -2.00060725e-01
-4.83650595e-01 -3.03794205e-01 4.65019822e-01 5.15972972e-01
5.60222752e-02 -4.25800800e-01 -5.11859834e-01 3.32834691e-01
-1.46010709e+00 -8.32629621e-01 -1.09583944e-01 1.88857234e+00
7.05480516e-01 1.08657606e-01 -1.32404000e-01 -2.61558387e-02
6.04632258e-01 5.24545789e-01 -5.52947581e-01 4.56740670e-02
-5.58967948e-01 -2.33885765e-01 4.91891764e-02 7.09190786e-01
-8.05615425e-01 8.46907973e-01 5.92413998e+00 8.05106282e-01
-1.67051756e+00 5.37174754e-02 3.12783867e-01 -1.90988958e-01
-9.88727152e-01 -2.04187393e-01 -3.93038541e-01 3.84127915e-01
3.28651011e-01 2.89156586e-01 5.87202311e-01 3.31179649e-01
3.67362082e-01 -9.31592509e-02 -9.40145850e-01 5.33770382e-01
-1.17877834e-02 -1.56328356e+00 5.88499784e-01 1.27711490e-01
7.56888628e-01 -2.76302457e-01 4.04660940e-01 -6.02187924e-02
1.85990240e-03 -7.19912469e-01 1.14774644e+00 9.05143201e-01
1.05020034e+00 -2.58958161e-01 1.03085116e-01 3.29124838e-01
-6.83936954e-01 -1.70342922e-01 2.54504502e-01 4.12225038e-01
5.23630619e-01 7.34721005e-01 -2.78696626e-01 7.47171938e-01
2.26123035e-02 7.17544317e-01 -2.58031815e-01 3.32347244e-01
-3.86314571e-01 1.56758979e-01 -4.16220635e-01 3.76724005e-01
3.94165099e-01 -3.80508184e-01 9.28450465e-01 8.68603230e-01
4.06436682e-01 -8.23470131e-02 -1.24666989e-01 1.39665735e+00
2.18628570e-01 -1.77466407e-01 -7.41680264e-01 -1.99360371e-01
1.94872066e-01 1.22273040e+00 -3.88030201e-01 -2.22757518e-01
-1.27890751e-01 1.03257024e+00 1.46287516e-01 6.66492045e-01
-9.40920472e-01 -1.00588566e-03 2.85165876e-01 1.33145913e-01
2.77304649e-01 -4.14728194e-01 -1.22838914e-01 -1.24373639e+00
-5.69572300e-02 -8.18896770e-01 -6.99078664e-02 -1.21096957e+00
-1.17192411e+00 7.37325907e-01 -1.99533388e-01 -1.12422299e+00
-2.89321020e-02 -4.05380458e-01 -3.69466811e-01 1.00853384e+00
-1.42660260e+00 -1.42802548e+00 -4.30048198e-01 9.09652352e-01
1.89331353e-01 -1.53433755e-01 6.38859093e-01 3.81610036e-01
-5.64345837e-01 1.95115134e-01 -2.77971715e-01 -3.62488925e-01
6.59119129e-01 -1.01227617e+00 -4.00167763e-01 8.30621719e-01
-1.84189469e-01 5.33142567e-01 4.85998601e-01 -6.66368127e-01
-1.44398391e+00 -9.45515513e-01 3.52073729e-01 -9.79331508e-02
7.52399504e-01 -3.63648236e-01 -7.13889360e-01 3.11079830e-01
2.69000798e-01 -8.58043134e-03 2.19028562e-01 -5.41067719e-01
-1.82151064e-01 -3.18233132e-01 -1.04550743e+00 7.73962915e-01
9.21573520e-01 -6.13501430e-01 -2.46882811e-01 6.25289828e-02
6.75349116e-01 -3.46166968e-01 -9.22293127e-01 3.28763694e-01
4.80797946e-01 -1.09139824e+00 8.67055357e-01 8.40661116e-03
6.71125531e-01 -6.29546165e-01 -2.42065892e-01 -1.39070880e+00
-3.04584533e-01 -4.41339403e-01 -2.16032937e-02 1.34278810e+00
4.85840619e-01 -4.81972575e-01 3.95697147e-01 3.35715622e-01
-1.63811296e-01 -3.84479582e-01 -9.24232244e-01 -3.86791587e-01
-1.33639202e-01 -3.54556203e-01 5.82048297e-01 1.14103603e+00
-1.27076074e-01 2.43720889e-01 -4.21676606e-01 1.46101460e-01
4.34282213e-01 -2.53750920e-01 5.40861130e-01 -6.21564746e-01
-3.72198939e-01 -7.17619419e-01 -1.65236250e-01 -1.29060698e+00
-3.63703705e-02 -9.32050228e-01 8.23367015e-03 -1.42613578e+00
-4.88752037e-01 -5.60435772e-01 2.99827363e-02 4.04426992e-01
1.98161796e-01 3.42946947e-01 4.03703749e-01 2.83861607e-01
6.60946369e-02 1.06292844e+00 2.00182748e+00 -4.62275267e-01
-3.40888500e-01 -4.54995841e-01 -5.30762911e-01 2.54230201e-01
3.98890018e-01 -1.43624991e-01 -6.79833949e-01 -4.76563424e-01
2.83542931e-01 3.20385516e-01 6.26268148e-01 -1.00510859e+00
4.40583110e-01 -7.37224594e-02 2.86497176e-01 -1.23631023e-01
5.15686572e-01 -1.03290987e+00 4.91527975e-01 1.49212301e-01
-4.08329815e-01 -4.69121486e-01 5.97697049e-02 5.73831320e-01
-3.06248158e-01 1.93465278e-01 8.01514864e-01 1.64438754e-01
-4.20883030e-01 1.35422781e-01 -4.57404464e-01 -4.27085429e-01
1.03248250e+00 -2.80915082e-01 -4.54533190e-01 -2.14646310e-01
-1.21071661e+00 1.92517996e-01 5.25938869e-01 2.73867279e-01
5.94569623e-01 -1.33472979e+00 -4.45580810e-01 7.23363876e-01
1.45327196e-01 -1.23798460e-01 5.25793791e-01 1.22517371e+00
-3.02291512e-01 -3.87231298e-02 -2.94980288e-01 -8.84518445e-01
-4.12334114e-01 6.36283040e-01 6.06792510e-01 -9.67198685e-02
-6.19410694e-01 4.40315545e-01 2.56936073e-01 -4.88227934e-01
2.74577625e-02 -2.90828526e-01 -2.07034156e-01 3.64000320e-01
2.08459586e-01 -5.00729941e-02 -2.24473193e-01 -5.64426839e-01
-5.12926579e-02 6.35961592e-01 3.04826707e-01 -5.29708385e-01
1.22781503e+00 -2.36550018e-01 -1.35677159e-01 2.35522673e-01
7.91406035e-01 1.35005906e-01 -1.67597532e+00 -3.54720629e-03
-5.21308422e-01 -9.91731808e-02 1.89504325e-01 -1.02861333e+00
-1.20586586e+00 7.12474883e-01 4.65276778e-01 2.18829468e-01
1.33853567e+00 -1.12393819e-01 1.91929832e-01 -2.59285063e-01
2.17787057e-01 -7.95756280e-01 3.54173958e-01 2.50672668e-01
1.31550050e+00 -7.73101926e-01 -3.14551622e-01 -2.96944886e-01
-7.88206041e-01 9.45446253e-01 4.96662378e-01 1.83044419e-01
5.07477343e-01 2.22442880e-01 4.80207689e-02 -3.88442814e-01
-5.92232466e-01 -1.25147581e-01 1.97794929e-01 7.59309411e-01
-1.16635785e-02 -2.01863453e-01 -8.06192458e-02 2.70047277e-01
1.55292571e-01 5.43661639e-02 3.20720166e-01 5.50697327e-01
-2.99041066e-02 -8.31186593e-01 -3.25170040e-01 -1.43329352e-01
2.79813200e-01 -8.97607431e-02 -2.22094730e-01 7.55998790e-01
5.09501040e-01 7.32723534e-01 4.28133681e-02 -3.09908509e-01
5.07906854e-01 -8.69643837e-02 4.82057899e-01 -3.01739752e-01
-5.85436463e-01 3.36482882e-01 -5.36981300e-02 -5.53064167e-01
-7.19604671e-01 -3.48418236e-01 -8.98719788e-01 -7.10703433e-03
-1.92879722e-01 -2.08302602e-01 7.29810059e-01 8.37077558e-01
5.40881932e-01 6.90869153e-01 5.82331002e-01 -1.31325078e+00
-6.11236803e-02 -8.38281751e-01 -5.85811019e-01 3.70167494e-01
5.81507325e-01 -7.50575423e-01 -5.21881700e-01 9.72290486e-02] | [11.613719940185547, -0.4872269630432129] |
a41ab0b4-181b-44be-9be5-2a4ed29133e7 | a-stochastic-metapopulation-state-space | 2106.07919 | null | https://arxiv.org/abs/2106.07919v1 | https://arxiv.org/pdf/2106.07919v1.pdf | A stochastic metapopulation state-space approach to modeling and estimating Covid-19 spread | Mathematical models are widely recognized as an important tool for analyzing and understanding the dynamics of infectious disease outbreaks, predict their future trends, and evaluate public health intervention measures for disease control and elimination. We propose a novel stochastic metapopulation state-space model for COVID-19 transmission, based on a discrete-time spatio-temporal susceptible/exposed/infected/recovered/deceased (SEIRD) model. The proposed framework allows the hidden SEIRD states and unknown transmission parameters to be estimated from noisy, incomplete time series of reported epidemiological data, by application of unscented Kalman filtering (UKF), maximum-likelihood adaptive filtering, and metaheuristic optimization. Experiments using both synthetic data and real data from the Fall 2020 Covid-19 wave in the state of Texas demonstrate the effectiveness of the proposed model. | ['Ulisses Braga-Neto', 'Martial Ndeffo-Mbah', 'Durward Cator III', 'Yukun Tan'] | 2021-06-15 | null | null | null | null | ['metaheuristic-optimization'] | ['methodology'] | [ 2.17204124e-01 -6.51403546e-01 8.77636597e-02 9.37080979e-02
-2.95861393e-01 -1.83521986e-01 6.17029905e-01 3.07686538e-01
-3.42950404e-01 9.12773013e-01 3.05387557e-01 -3.15942496e-01
-8.63126457e-01 -6.71986401e-01 -1.72986791e-01 -9.62445915e-01
-8.42521548e-01 7.41560459e-01 -2.26966456e-01 -1.83706172e-03
-2.68327564e-01 3.68453473e-01 -9.62391555e-01 -4.55228567e-01
1.11066115e+00 1.90518826e-01 4.02796209e-01 8.00581217e-01
3.37382615e-01 1.83332890e-01 -8.21495414e-01 1.35376081e-01
-2.86033731e-02 -6.38440624e-02 2.77806148e-02 -5.07597178e-02
-9.35681880e-01 -3.34898233e-01 6.13191444e-03 6.68969691e-01
2.49801219e-01 -8.78522545e-02 9.82738972e-01 -1.30888665e+00
-1.51716068e-01 -9.33249295e-02 -5.94521105e-01 3.45790714e-01
8.08234289e-02 3.67878556e-01 -7.33854622e-02 -1.92153692e-01
5.53622007e-01 1.42131865e+00 7.05719531e-01 4.58079368e-01
-1.29693878e+00 -5.72322547e-01 1.17670618e-01 -2.62810439e-01
-1.40807784e+00 2.05890629e-02 2.50788391e-01 -7.11285412e-01
9.94285941e-01 2.23985091e-01 1.00119567e+00 1.15826380e+00
9.38232780e-01 4.42051858e-01 8.66777539e-01 1.33749187e-01
4.16808158e-01 -4.18993145e-01 3.13047558e-01 2.96674699e-01
8.55359137e-01 4.41963375e-01 9.03547555e-03 -1.10472035e+00
5.20639718e-01 8.17094028e-01 -1.94440886e-01 -4.10792641e-02
-9.86171246e-01 9.81806874e-01 -1.31367132e-01 -3.59011553e-02
-1.19659257e+00 -1.47314936e-01 4.69357111e-02 8.61359611e-02
1.14869261e+00 -3.04986443e-02 -5.14126301e-01 2.89877892e-01
-9.22867000e-01 4.03676718e-01 4.88141328e-01 3.77828956e-01
3.25877964e-01 2.63718098e-01 -2.51634270e-01 3.35200191e-01
7.07455635e-01 1.89337027e+00 -2.51033515e-01 -6.27122521e-01
6.91691712e-02 3.47809851e-01 6.50396943e-01 -1.02522218e+00
-7.11725593e-01 -5.47899365e-01 -1.36264348e+00 -3.26372623e-01
5.27197793e-02 -7.32932329e-01 -8.96141768e-01 1.67847085e+00
6.26427889e-01 9.32987869e-01 1.64297134e-01 4.20796543e-01
-1.11930430e-01 1.33251715e+00 2.24105790e-01 -1.18289506e+00
1.07418966e+00 1.01051562e-01 -9.59213138e-01 9.04183164e-02
2.43954450e-01 -4.53729153e-01 -2.32702658e-01 2.41588548e-01
-9.51825380e-01 1.15056343e-01 -3.60363781e-01 1.18986416e+00
-1.82296917e-01 -3.63495275e-02 2.36711681e-01 3.65760326e-01
-7.69067883e-01 8.64392444e-02 -1.35337985e+00 -5.30435443e-01
1.61538765e-01 2.22154900e-01 1.37541592e-01 -7.47179389e-02
-1.14921272e+00 4.73312289e-01 1.45666704e-01 5.07764935e-01
-1.22302556e+00 -9.16849256e-01 -5.70368528e-01 -8.03255010e-03
2.70623356e-01 -1.05555069e+00 6.05112791e-01 -3.17435861e-02
-7.48341501e-01 1.12566322e-01 -4.78288829e-01 -7.33689487e-01
3.03007036e-01 -5.24723604e-02 -7.07505107e-01 5.13768271e-02
1.51879992e-02 -2.94519484e-01 8.95499647e-01 -1.09686375e+00
-4.93945420e-01 -8.45957994e-01 -8.41930747e-01 -1.85398668e-01
-1.24068283e-01 2.40574539e-01 2.98499405e-01 -9.56805170e-01
-1.91644013e-01 -1.12847352e+00 -4.40065980e-01 -5.58722258e-01
-1.02011822e-01 -1.68940052e-01 9.25443053e-01 -1.11845827e+00
1.36197472e+00 -1.65730882e+00 2.79157668e-01 3.82555246e-01
-9.37554985e-02 6.27610624e-01 -1.77731827e-01 1.06836307e+00
5.14481723e-01 -2.28475645e-01 -3.19673896e-01 1.70436520e-02
-5.63436091e-01 3.11138958e-01 -1.77263081e-01 6.21290505e-01
1.97471619e-01 3.10600549e-01 -1.07725286e+00 3.78154442e-02
2.75151312e-01 8.78189802e-01 -2.07951516e-01 4.13962752e-01
-2.33294040e-01 6.12238050e-01 -8.88865769e-01 1.48238614e-01
1.05765212e+00 -3.48262787e-01 3.94513011e-01 1.51399657e-01
-2.64075071e-01 -4.96202052e-01 -1.17362523e+00 5.05765498e-01
-3.82426530e-01 8.92211497e-02 3.95109773e-01 -8.92979681e-01
5.52344978e-01 6.59289598e-01 7.46414483e-01 -1.06788449e-01
8.73365998e-02 -1.75939009e-01 -1.45741835e-01 -7.29247153e-01
1.50588408e-01 7.84104243e-02 2.32176576e-02 8.14149439e-01
-3.13564867e-01 2.54059076e-01 2.83799112e-01 1.94158047e-01
1.23121202e+00 -4.65322435e-01 6.40166849e-02 -2.28981465e-01
2.60821402e-01 2.62735307e-01 8.22091281e-01 7.04081714e-01
1.42567396e-01 -2.00942859e-01 5.99429496e-02 -3.38594347e-01
-8.21312308e-01 -1.42496812e+00 1.98975969e-02 1.81227863e-01
-1.50305793e-01 9.90088359e-02 -5.41627109e-01 1.41881956e-02
-7.62896091e-02 6.87756896e-01 -5.03266335e-01 -2.92354643e-01
-5.27911842e-01 -1.60718775e+00 1.33531943e-01 -2.27036551e-01
1.95531711e-01 -7.54749835e-01 -5.31562090e-01 6.86367512e-01
-2.18923494e-01 -6.18801594e-01 1.72174182e-02 -2.83599049e-01
-9.34417188e-01 -1.17830420e+00 -9.36030507e-01 -3.33887964e-01
8.49133253e-01 3.62843245e-01 6.77119911e-01 -8.64166208e-03
-4.55633849e-01 6.64576828e-01 1.49873048e-01 -3.95180494e-01
-6.83581710e-01 -4.38513309e-01 5.02461433e-01 6.64080828e-02
2.04020947e-01 2.76277643e-02 -8.71165931e-01 1.90727338e-01
-1.09561813e+00 -2.67045498e-01 2.51028657e-01 7.31969357e-01
3.01163703e-01 4.05331969e-01 8.43300521e-01 -3.72912437e-01
8.97598147e-01 -8.83634627e-01 -9.22620237e-01 6.43170416e-01
-3.59783828e-01 -2.06962422e-01 1.12568103e-01 -4.95199263e-01
-1.30753732e+00 -6.18410558e-02 2.56036431e-01 -2.46452391e-02
-8.52824971e-02 7.63645172e-01 5.04081190e-01 6.26872480e-01
-9.43902954e-02 5.33514977e-01 4.44827855e-01 -4.60113615e-01
-3.55520725e-01 1.02533484e+00 1.81753904e-01 -1.03487574e-01
7.68984199e-01 8.03006351e-01 9.49492306e-02 -1.25358987e+00
-2.60742903e-01 -6.86225414e-01 6.36536554e-02 -2.54079491e-01
8.62548292e-01 -1.18486929e+00 -1.20355999e+00 1.16071630e+00
-1.08945858e+00 -1.57101944e-01 2.73791790e-01 9.66485679e-01
-2.50155598e-01 3.74394618e-02 -6.38554811e-01 -1.51698244e+00
-5.17087638e-01 -8.86446953e-01 9.28618014e-01 2.52818707e-02
-1.08456962e-01 -1.49372828e+00 7.26855040e-01 1.08080842e-01
7.20079124e-01 5.22611678e-01 9.65971231e-01 -2.32219741e-01
-3.73347878e-01 -2.37066731e-01 1.49154454e-01 1.07566647e-01
5.11866927e-01 3.41831565e-01 1.32501438e-01 -8.09445143e-01
1.25605008e-03 2.94524729e-01 5.01145244e-01 1.06999671e+00
3.83325964e-02 -7.18804896e-01 -8.78668249e-01 1.07633136e-01
1.50130451e+00 4.90750998e-01 1.89727306e-01 -2.33985111e-01
-8.75429288e-02 6.60195529e-01 6.52739942e-01 1.03208768e+00
4.25762087e-01 3.38292241e-01 4.09197927e-01 8.55786651e-02
6.93477631e-01 5.32123335e-02 3.41969818e-01 7.62032568e-01
-6.21931255e-02 -6.72202885e-01 -1.16374540e+00 8.82034183e-01
-1.78321743e+00 -1.40368056e+00 -5.48314989e-01 2.49110556e+00
3.19879204e-01 -2.67820925e-01 4.57802624e-01 -2.51299262e-01
7.90834844e-01 -2.90432245e-01 -3.59478861e-01 1.61583070e-02
-1.54707357e-01 -1.58795759e-01 6.58970714e-01 3.96777868e-01
-7.96993136e-01 1.97257906e-01 6.39614725e+00 6.23918831e-01
-8.86829257e-01 -7.41609139e-03 3.57037425e-01 2.19502196e-01
-9.96211916e-02 -2.12425366e-01 -6.66705072e-01 6.70743704e-01
1.19251657e+00 -2.21407726e-01 5.19811094e-01 -2.01911092e-01
1.21502411e+00 -3.24617893e-01 -5.06941527e-02 3.20318133e-01
-2.88205266e-01 -1.22365165e+00 -7.01660961e-02 2.49869421e-01
9.40805852e-01 1.93765476e-01 4.85584103e-02 -2.79609203e-01
6.89773381e-01 -3.41067940e-01 -2.47513980e-01 9.88535047e-01
5.07999621e-02 -8.60163271e-01 6.01871669e-01 7.63851762e-01
-9.64422107e-01 -2.81115592e-01 6.08999096e-02 2.64567107e-01
7.90382385e-01 6.34662628e-01 -6.82249844e-01 3.92264664e-01
5.98673403e-01 5.84350407e-01 9.86903906e-02 1.26272333e+00
1.48503497e-01 8.92627418e-01 -5.29546499e-01 -1.14288025e-01
2.97467392e-02 -4.09651607e-01 1.08326495e+00 5.76070011e-01
6.85664654e-01 1.95498392e-01 1.17820680e-01 5.17246664e-01
6.84983194e-01 -4.92485195e-01 -4.48765129e-01 -1.73021555e-01
4.37827170e-01 8.23416948e-01 -5.16661108e-01 -2.73278177e-01
8.83679241e-02 3.10974807e-01 -6.10601485e-01 8.18546832e-01
-7.29551077e-01 -9.72534940e-02 8.45410705e-01 2.38945410e-01
3.73423964e-01 -5.22472799e-01 6.34525716e-01 -1.03728926e+00
-5.43965757e-01 -6.74205422e-01 4.53534573e-01 -4.96973306e-01
-1.29332864e+00 4.32092726e-01 4.42981064e-01 -9.85983014e-01
-5.73397517e-01 -6.79474473e-02 -6.89167559e-01 7.27634132e-01
-1.26472068e+00 -7.32471287e-01 3.46869946e-01 6.61580861e-01
3.34919333e-01 7.81842694e-02 6.27821326e-01 -9.27990004e-02
-7.42258310e-01 -4.38066751e-01 1.07999527e+00 -2.54970789e-01
-1.12448342e-01 -3.83741379e-01 2.14520663e-01 5.86300731e-01
-8.33933771e-01 8.44064474e-01 1.09144056e+00 -1.36786389e+00
-1.56282246e+00 -1.35426235e+00 7.27348626e-01 3.22202861e-01
6.43298805e-01 -2.36281440e-01 -7.87810922e-01 3.14112276e-01
2.47066826e-01 -6.25226557e-01 3.39124918e-01 -4.19451237e-01
3.78855675e-01 -1.53093599e-02 -1.46319699e+00 4.94992048e-01
4.10888404e-01 2.27416486e-01 -3.38465065e-01 5.74468553e-01
5.94763756e-01 4.10632998e-01 -8.88218939e-01 5.53618848e-01
2.83616900e-01 -2.02128902e-01 1.15427566e+00 -5.46733737e-01
-5.12143314e-01 -1.13553382e-01 2.91133016e-01 -1.74859846e+00
-2.25520253e-01 -9.26528573e-01 5.43996170e-02 9.44087327e-01
5.32075949e-02 -8.46541047e-01 3.29487205e-01 7.21779242e-02
5.41804612e-01 -2.60529071e-01 -1.16745162e+00 -1.06025183e+00
-3.56144130e-01 -1.66302714e-02 6.38251901e-01 4.89645600e-01
-5.44730604e-01 -2.22102143e-02 -8.93475413e-01 8.26295257e-01
1.20670187e+00 -1.48287684e-01 5.61299264e-01 -1.35277236e+00
-9.72923115e-02 3.61212581e-01 1.16818976e-02 -3.98246169e-01
4.26809955e-03 5.32255992e-02 -2.36896072e-02 -1.68925917e+00
4.28543180e-01 -2.58449376e-01 -9.22568142e-02 5.43320403e-02
-1.85054779e-01 -2.03585759e-01 -2.54461735e-01 7.34643862e-02
-3.41714054e-01 7.07211494e-01 6.24802649e-01 -1.46127298e-01
-3.50394547e-01 7.18035638e-01 2.12078795e-01 4.74298745e-01
7.22612202e-01 -9.52235520e-01 -4.54100728e-01 -4.01924044e-01
1.71105310e-01 7.91041851e-01 6.62971556e-01 -6.57896042e-01
1.72370020e-03 -7.08582699e-01 -1.05568141e-01 -1.16387177e+00
5.10751605e-01 -1.05941403e+00 1.03804159e+00 1.59403002e+00
1.53116509e-01 3.69932592e-01 1.97458431e-01 1.11021435e+00
1.99620277e-01 2.59578556e-01 3.52558613e-01 2.80921996e-01
1.22689560e-01 4.04000431e-01 -1.23694324e+00 3.14915329e-02
1.18898010e+00 2.46471405e-01 -6.97508395e-01 -4.06932265e-01
-7.57519960e-01 6.85819209e-01 -4.47337963e-02 3.19362581e-01
7.58086503e-01 -7.28286386e-01 -1.20340538e+00 2.76535332e-01
-4.03255552e-01 -6.36232495e-01 8.00849736e-01 1.04570007e+00
-3.12349081e-01 4.27733809e-01 1.09616235e-01 -8.10015976e-01
-1.37597632e+00 6.83186471e-01 1.43782839e-01 -6.30964518e-01
-1.47860572e-02 -1.32038277e-02 -1.81738079e-01 -3.84964436e-01
-1.80909365e-01 1.09839916e-01 2.86244564e-02 4.02331129e-02
6.28819466e-01 9.50247228e-01 -3.08481634e-01 -7.69620836e-01
-5.18229902e-01 4.16507661e-01 2.10604265e-01 -6.42722398e-02
1.79447281e+00 -4.75038886e-01 -1.95653781e-01 2.53956497e-01
7.40691900e-01 -5.25129199e-01 -1.10316873e+00 -2.21948579e-01
-1.46553833e-02 -1.13982409e-01 1.02135122e-01 -6.46520674e-01
-7.86668360e-01 5.67724109e-01 8.07776213e-01 4.83552784e-01
1.08558118e+00 -2.63014108e-01 9.74371910e-01 5.62753645e-04
3.02452594e-01 -5.54745197e-01 -4.64372724e-01 9.16687623e-02
6.01290762e-01 -7.39196599e-01 8.49372819e-02 -1.44217461e-01
-1.78156883e-01 4.98499393e-01 -1.24672107e-01 -1.53959736e-01
1.16985011e+00 4.12105471e-01 -1.32578313e-01 -1.14536405e-01
-1.24525177e+00 -1.56821221e-01 -8.12023357e-02 8.58156979e-01
-2.52978444e-01 2.07513109e-01 -4.69218552e-01 6.11979887e-02
8.56088579e-01 3.99948388e-01 3.80013198e-01 1.05478966e+00
-3.48759234e-01 -7.80435026e-01 -7.75757670e-01 5.69361866e-01
-5.71421385e-01 -6.59845844e-02 1.66677639e-01 5.31492472e-01
-3.49876463e-01 1.23884821e+00 2.16379449e-01 2.96580464e-01
2.48238221e-02 -4.44735080e-01 1.12580858e-01 -2.41594970e-01
-5.35734773e-01 4.47034687e-01 -4.62956965e-01 1.16377108e-01
-5.38457930e-01 -8.93505573e-01 -8.83215725e-01 -4.19929504e-01
-4.95595455e-01 5.41761100e-01 7.97254682e-01 9.37626898e-01
7.64421165e-01 3.31781626e-01 1.08899713e+00 -4.97564882e-01
-8.42259645e-01 -7.32260287e-01 -4.95092541e-01 -1.15835465e-01
6.12074912e-01 -5.58185697e-01 -5.53850710e-01 -2.86965638e-01] | [5.997599124908447, 4.3704118728637695] |
ebc2076a-7ed6-4e1c-b54e-36a7c2410e47 | optimal-scalarizations-for-sublinear | 2307.03288 | null | https://arxiv.org/abs/2307.03288v1 | https://arxiv.org/pdf/2307.03288v1.pdf | Optimal Scalarizations for Sublinear Hypervolume Regret | Scalarization is a general technique that can be deployed in any multiobjective setting to reduce multiple objectives into one, such as recently in RLHF for training reward models that align human preferences. Yet some have dismissed this classical approach because linear scalarizations are known to miss concave regions of the Pareto frontier. To that end, we aim to find simple non-linear scalarizations that can explore a diverse set of $k$ objectives on the Pareto frontier, as measured by the dominated hypervolume. We show that hypervolume scalarizations with uniformly random weights are surprisingly optimal for provably minimizing the hypervolume regret, achieving an optimal sublinear regret bound of $O(T^{-1/k})$, with matching lower bounds that preclude any algorithm from doing better asymptotically. As a theoretical case study, we consider the multiobjective stochastic linear bandits problem and demonstrate that by exploiting the sublinear regret bounds of the hypervolume scalarizations, we can derive a novel non-Euclidean analysis that produces improved hypervolume regret bounds of $\tilde{O}( d T^{-1/2} + T^{-1/k})$. We support our theory with strong empirical performance of using simple hypervolume scalarizations that consistently outperforms both the linear and Chebyshev scalarizations, as well as standard multiobjective algorithms in bayesian optimization, such as EHVI. | ['Qiuyi Zhang'] | 2023-07-06 | null | null | null | null | ['bayesian-optimization'] | ['methodology'] | [ 4.55202349e-03 4.35752302e-01 -4.88646805e-01 -4.30052161e-01
-1.51354396e+00 -1.07230413e+00 -6.00407980e-02 1.53688997e-01
-6.95310950e-01 1.14248300e+00 1.20871194e-01 -4.10613060e-01
-1.20739543e+00 -5.86593747e-01 -8.83219004e-01 -9.86317933e-01
-2.35360980e-01 9.03865635e-01 -3.50017250e-01 -1.09393850e-01
5.23901105e-01 2.38243669e-01 -1.23190856e+00 -1.66636303e-01
1.03977406e+00 1.20963919e+00 -2.98380792e-01 7.39373505e-01
1.26699015e-01 1.19970791e-01 -5.49853206e-01 -8.24791908e-01
6.12701654e-01 -2.15568766e-01 -9.03597653e-01 -1.04949184e-01
4.37445819e-01 -2.06852838e-01 1.16476446e-01 1.12504911e+00
5.08825481e-01 2.43568912e-01 6.27331734e-01 -1.13921690e+00
-4.43642944e-01 1.11102068e+00 -9.20552731e-01 1.01668581e-01
1.11615220e-02 -2.14037709e-02 1.73460257e+00 -1.22024737e-01
2.92132258e-01 1.38851225e+00 2.99722016e-01 4.22286898e-01
-1.48503482e+00 -5.56181312e-01 3.01861227e-01 -2.15256870e-01
-1.17555070e+00 -2.03592554e-01 5.76974213e-01 -6.88574016e-02
5.60799062e-01 1.00142515e+00 4.43913728e-01 7.41156638e-01
-1.40063137e-01 9.17710483e-01 1.35323727e+00 -3.46486390e-01
4.37744588e-01 1.32043391e-01 3.18116605e-01 6.11986578e-01
7.66100228e-01 2.21828386e-01 -4.55517739e-01 -3.70415181e-01
6.03453815e-01 -2.66050577e-01 -2.41599441e-01 -6.71976328e-01
-1.11276853e+00 1.13091981e+00 2.38734990e-01 5.15730977e-02
-2.25750089e-01 7.56400347e-01 1.38636708e-01 2.93045908e-01
6.31405115e-01 9.25882041e-01 -5.09892285e-01 -1.50924250e-01
-1.04997003e+00 6.38405204e-01 7.43799746e-01 1.09835148e+00
4.07633007e-01 -2.10668400e-01 -5.03805697e-01 5.82672596e-01
2.53178000e-01 7.94659972e-01 -1.40232041e-01 -1.41965699e+00
8.62901032e-01 1.79113880e-01 7.87524223e-01 -6.85099006e-01
-4.92028236e-01 -9.28458095e-01 -3.87840152e-01 9.77268666e-02
7.14239180e-01 -4.21139598e-01 -3.37572902e-01 1.90462554e+00
3.83430421e-01 -5.22292495e-01 -1.58493906e-01 1.04931056e+00
-3.70501988e-02 3.94097656e-01 -2.72512287e-01 -6.17073476e-01
1.01481843e+00 -6.55543745e-01 -3.88706416e-01 -1.76683605e-01
7.22190797e-01 -6.53161228e-01 1.13618052e+00 5.27062774e-01
-1.59453046e+00 3.77457380e-01 -1.00895166e+00 3.98289651e-01
1.40837550e-01 -4.17271376e-01 1.04048240e+00 1.28338575e+00
-8.49731863e-01 7.19900489e-01 -3.91096354e-01 1.11599512e-01
5.90115786e-01 5.80674410e-01 2.60620594e-01 4.26108688e-02
-6.31128371e-01 6.44951463e-01 4.33561057e-01 -1.04365796e-01
-7.41558015e-01 -8.57101679e-01 -3.34355384e-01 2.63949633e-01
9.90026951e-01 -6.37681305e-01 1.25370252e+00 -7.38630831e-01
-1.42782152e+00 5.82072496e-01 1.13909036e-01 -4.53649610e-01
8.22989881e-01 -1.53000310e-01 1.95425659e-01 -1.03738263e-01
-1.54058309e-02 1.27535149e-01 6.02899611e-01 -1.19429255e+00
-7.24181056e-01 -7.52063930e-01 5.84180951e-01 4.67535049e-01
-3.86593431e-01 -1.87893193e-02 4.80636880e-02 -2.25796461e-01
8.51756632e-02 -9.66816545e-01 -5.36832094e-01 -4.70282882e-01
-6.11486912e-01 -9.03261974e-02 -2.45226279e-01 -1.46771953e-01
1.17911100e+00 -1.74555635e+00 4.03194755e-01 7.70959377e-01
2.25192741e-01 -3.31760436e-01 -6.18143715e-02 2.07191817e-02
4.01109636e-01 4.30736035e-01 -7.03121349e-02 -1.88550334e-02
5.67125082e-01 3.40808518e-02 -2.51761049e-01 8.70419860e-01
-5.15202761e-01 8.31498981e-01 -8.55722547e-01 -3.08842421e-01
-2.33542010e-01 -2.15911016e-01 -8.92954230e-01 -1.86139390e-01
-4.05023664e-01 -9.30632651e-02 -7.20094323e-01 6.72604680e-01
7.30528295e-01 -4.75952089e-01 3.55315775e-01 1.11269757e-01
5.09489663e-02 -1.12951621e-01 -1.45406568e+00 1.46229577e+00
-2.26039633e-01 1.37219071e-01 4.11786348e-01 -1.16765666e+00
3.94329190e-01 3.01691685e-02 7.36099839e-01 -4.04719174e-01
3.95197302e-01 4.82181430e-01 -1.21843643e-01 -7.22455531e-02
4.33021247e-01 -4.82513309e-01 -2.92277694e-01 6.00805223e-01
-2.24644393e-01 -1.14376247e-01 7.65714794e-02 -1.58630192e-01
9.18561935e-01 -1.16875023e-01 2.11898796e-02 -7.09802806e-01
1.34317636e-01 1.96789838e-02 5.64422250e-01 1.17537379e+00
-1.80799738e-01 2.84575820e-01 8.04077446e-01 -8.92543569e-02
-1.15536928e+00 -1.12542856e+00 -2.56917983e-01 1.58003688e+00
3.61109614e-01 5.11935502e-02 -7.36253738e-01 -6.60267472e-01
4.71511453e-01 1.09804285e+00 -8.04906905e-01 7.20017105e-02
-3.29237312e-01 -1.32063711e+00 5.34256876e-01 3.02800447e-01
1.06242508e-01 -2.30246812e-01 -7.79934049e-01 2.25936487e-01
-2.10976489e-02 -6.11020207e-01 -6.91966295e-01 5.25789320e-01
-9.31398630e-01 -8.27896476e-01 -9.94492471e-01 7.82956332e-02
5.19940913e-01 1.10850148e-01 9.30301785e-01 -5.62954307e-01
-8.93708318e-02 5.22827268e-01 -3.56639288e-02 -6.00987017e-01
1.17839001e-01 -1.04538808e-02 1.06675915e-01 -2.08486408e-01
1.22304879e-01 -3.97622645e-01 -9.00391281e-01 4.12210613e-01
-6.70014501e-01 -3.87697488e-01 5.41839719e-01 7.28226244e-01
6.42617106e-01 -1.91011891e-01 4.64001149e-01 -7.72687793e-01
8.29033911e-01 -3.82178307e-01 -1.11140060e+00 5.84522605e-01
-1.01218593e+00 5.75948656e-01 4.07950073e-01 -4.10452783e-01
-8.36445868e-01 -5.22469461e-01 3.69326115e-01 -1.95214152e-01
4.82979447e-01 4.18533444e-01 1.26745567e-01 -2.37110063e-01
7.74925053e-01 -1.80714298e-02 -2.55226314e-01 -3.92074615e-01
7.17078805e-01 4.51896936e-01 1.29395202e-01 -1.12443507e+00
6.52133703e-01 4.03571635e-01 4.48754579e-01 -3.45878601e-01
-1.41678798e+00 -2.42399499e-01 6.97475076e-02 -8.19987655e-02
5.17341018e-01 -3.34625185e-01 -1.50324667e+00 -6.85379386e-01
-6.43040717e-01 1.75399482e-02 -5.59970319e-01 7.12270498e-01
-1.22658336e+00 1.07983619e-01 -9.97025240e-03 -1.40951025e+00
-2.18474969e-01 -1.00409460e+00 8.24005127e-01 1.42806575e-01
-2.16328148e-02 -8.23600590e-01 3.52576119e-03 7.58849442e-01
4.57870752e-01 3.22413355e-01 1.05494249e+00 -7.14878082e-01
-7.24768698e-01 1.11513220e-01 -1.74693212e-01 6.17910214e-02
-4.08000410e-01 -4.12005901e-01 -5.84499538e-01 -5.73521793e-01
-1.18832678e-01 -5.68419099e-01 7.65948892e-01 8.42200279e-01
1.27276397e+00 -8.23858202e-01 -2.64593482e-01 8.07445586e-01
1.56856847e+00 1.96324587e-01 2.01954078e-02 5.97199142e-01
1.47672057e-01 4.45084155e-01 6.55018032e-01 8.56333733e-01
1.18929282e-01 4.90312636e-01 7.24765241e-01 3.06818277e-01
7.12038875e-01 1.25338137e-01 3.64146531e-02 9.48942900e-02
-3.40929031e-01 -2.74765223e-01 -5.91432631e-01 5.05811274e-01
-2.03304553e+00 -8.92795682e-01 3.24484944e-01 2.71395278e+00
8.54173005e-01 2.91704386e-01 4.21113938e-01 -4.83923033e-02
6.01344347e-01 1.11300699e-01 -8.20265830e-01 -7.58634329e-01
-2.57337570e-01 2.15630904e-01 1.27460718e+00 6.03152156e-01
-8.62830579e-01 6.24671519e-01 6.76124620e+00 1.05078518e+00
-5.24992943e-01 1.52892262e-01 8.45490754e-01 -1.20784318e+00
-8.43227208e-01 -7.55986944e-02 -7.85363436e-01 3.06150019e-01
9.51251388e-01 -6.44916654e-01 9.20458376e-01 1.00114858e+00
9.90112722e-02 -4.43563983e-02 -1.30773365e+00 9.72267330e-01
-2.00567693e-01 -1.46177983e+00 -2.68687993e-01 5.52620888e-01
9.81329024e-01 -3.19165677e-01 4.86694276e-01 1.90183431e-01
8.45425248e-01 -1.31838715e+00 6.71338737e-01 1.76018357e-01
6.38237000e-01 -1.37856662e+00 5.13312340e-01 3.24187487e-01
-4.98245776e-01 -4.73155499e-01 -5.48011839e-01 2.91102707e-01
2.07606927e-01 6.46933019e-01 -6.77130759e-01 5.35204649e-01
6.30404532e-01 -2.55360246e-01 1.14501670e-01 1.25133121e+00
3.56319129e-01 2.90658981e-01 -6.42022192e-01 -3.90181899e-01
6.93696380e-01 -5.26100576e-01 7.65082121e-01 8.77834439e-01
4.21495765e-01 2.17159465e-01 -1.12101614e-01 8.74440014e-01
-3.14379454e-01 4.26800698e-01 -2.42956057e-01 -2.67762810e-01
2.69821733e-01 8.21328998e-01 -5.80266476e-01 8.59546140e-02
4.83636521e-02 5.96823394e-01 4.05569375e-01 4.50288653e-01
-1.16936231e+00 -3.33162576e-01 8.35370779e-01 -2.25953370e-01
4.16616350e-01 -1.35073289e-01 -7.17701554e-01 -9.37585354e-01
2.09590942e-01 -5.75739563e-01 5.67172885e-01 -1.53905720e-01
-1.17603886e+00 1.83353469e-01 2.86407888e-01 -7.11074650e-01
-6.71099722e-02 -5.49171686e-01 1.13988772e-01 6.30379915e-01
-1.24479640e+00 -6.46142602e-01 3.67451310e-01 3.84962708e-01
3.39282118e-02 -5.81300408e-02 4.91539240e-01 -8.28860849e-02
-3.76550704e-01 9.24352229e-01 8.95094156e-01 -6.29351079e-01
5.17216504e-01 -1.48616874e+00 -4.96711642e-01 4.32379931e-01
-6.26546890e-02 5.32947481e-01 1.11664653e+00 -2.40332410e-01
-1.65567613e+00 -6.18001938e-01 4.12757099e-01 -3.41382504e-01
7.58766055e-01 -5.39396144e-02 -5.97832501e-02 7.34200358e-01
-8.26267749e-02 -2.52939135e-01 8.84094238e-01 5.41779101e-01
-2.92844594e-01 -3.76725882e-01 -1.48054981e+00 8.17285180e-01
1.31046939e+00 -5.35468338e-03 -2.97686398e-01 6.80015743e-01
7.24324644e-01 -2.95334458e-01 -1.10394371e+00 4.62461203e-01
7.27811992e-01 -9.57667887e-01 1.06036961e+00 -9.44759727e-01
2.49391168e-01 3.58613849e-01 -7.36247778e-01 -1.28500509e+00
-3.01853865e-01 -1.12588763e+00 -2.20358849e-01 8.27737093e-01
6.32550657e-01 -6.44546151e-01 1.10066998e+00 8.10895681e-01
2.78037071e-01 -1.09331858e+00 -1.41481376e+00 -1.04047430e+00
5.22171676e-01 -4.24082369e-01 5.77623367e-01 5.89460909e-01
1.35757774e-01 4.33290154e-02 -4.78616297e-01 2.13741399e-02
1.27047253e+00 4.85786945e-01 3.20183992e-01 -1.10729742e+00
-7.97930419e-01 -9.43409145e-01 1.19541533e-01 -1.19374728e+00
1.35088693e-02 -8.39600086e-01 -2.12148368e-01 -1.22397161e+00
4.45057929e-01 -6.98153138e-01 -5.12833774e-01 1.46904379e-01
5.91469556e-02 -1.96426570e-01 1.86300844e-01 -7.73224831e-02
-8.10620666e-01 5.40509999e-01 1.36805499e+00 -2.02269152e-01
-2.35577002e-01 2.71759331e-01 -1.49473083e+00 5.67716658e-01
6.35691047e-01 -5.75085819e-01 -4.65761244e-01 -4.49898630e-01
7.25078940e-01 5.01775324e-01 -1.40219614e-01 -3.86030436e-01
-1.44034117e-01 -9.18359280e-01 1.77722335e-01 -3.38766873e-01
3.68258387e-01 -7.73501337e-01 -2.15423942e-01 3.08353424e-01
-8.17893684e-01 -1.69872403e-01 -1.51909247e-01 7.43221402e-01
4.87664044e-01 -3.88150185e-01 7.89254546e-01 -2.00674579e-01
2.06589967e-01 3.07541937e-01 1.33391663e-01 4.11507726e-01
1.36785805e+00 -9.72909629e-02 -4.54648346e-01 -5.35016775e-01
-5.26733577e-01 5.73575914e-01 1.45273671e-01 -1.54773176e-01
3.06832343e-01 -1.11536586e+00 -7.38832176e-01 -4.95852649e-01
-4.83989529e-02 -1.26632780e-01 7.02322274e-02 9.49938118e-01
-2.89245546e-01 8.16067934e-01 -1.93988923e-02 -4.38453227e-01
-1.14484441e+00 5.76205134e-01 3.60460401e-01 -5.22605360e-01
-2.07299888e-01 1.11585355e+00 8.00007135e-02 -1.40550673e-01
4.47771251e-01 -1.49573743e-01 3.57454985e-01 1.01102032e-01
1.62898004e-01 7.95517266e-01 -2.32595101e-01 -5.63546754e-02
-3.71439606e-01 4.46104705e-01 -1.23347968e-01 -7.21540213e-01
1.50766540e+00 -1.81488842e-01 1.02891073e-01 2.01960221e-01
1.08846319e+00 1.71406791e-01 -1.25146985e+00 -1.06127203e-01
2.64018327e-01 -8.09027970e-01 1.15779459e-01 -9.64256227e-01
-7.12773860e-01 3.28541458e-01 4.97764826e-01 3.33355248e-01
8.67815018e-01 1.08391479e-01 4.54547584e-01 6.72397554e-01
6.58236146e-01 -1.38273537e+00 -4.12771255e-02 1.86838746e-01
8.01793337e-01 -1.06945539e+00 2.52377778e-01 -1.24085516e-01
-6.32353604e-01 8.57730210e-01 9.99335125e-02 -1.52039304e-01
3.64420921e-01 6.56128675e-02 -5.06008387e-01 -1.34781808e-01
-5.85594177e-01 -2.29685336e-01 3.02823007e-01 1.55912861e-01
3.24579149e-01 5.34046412e-01 -8.33527565e-01 6.87529862e-01
-4.85662490e-01 -3.61067027e-01 3.67624849e-01 7.50555277e-01
-7.28598058e-01 -1.04060769e+00 -5.69286823e-01 7.06688821e-01
-7.50337660e-01 5.68041541e-02 -1.32082105e-01 6.07896566e-01
-4.27404404e-01 1.09195805e+00 -2.32514262e-01 -9.05126184e-02
2.28332624e-01 -6.27974048e-02 9.31129336e-01 -1.73641086e-01
-5.38627625e-01 3.32519859e-01 2.66803294e-01 -6.53441072e-01
-1.56412423e-01 -5.77175319e-01 -8.98613811e-01 -6.03340685e-01
-3.30220431e-01 4.35494751e-01 6.81216300e-01 9.06751573e-01
9.37923416e-03 2.69834548e-01 7.38530517e-01 -5.31288445e-01
-1.56041849e+00 -6.23868704e-01 -9.24918532e-01 1.48113802e-01
3.99230778e-01 -7.16895819e-01 -5.71949899e-01 -8.70838046e-01] | [4.552755832672119, 3.3419384956359863] |
22478d34-32b1-43d5-9676-19567a6f4bca | blended-diffusion-for-text-driven-editing-of | 2111.14818 | null | https://arxiv.org/abs/2111.14818v2 | https://arxiv.org/pdf/2111.14818v2.pdf | Blended Diffusion for Text-driven Editing of Natural Images | Natural language offers a highly intuitive interface for image editing. In this paper, we introduce the first solution for performing local (region-based) edits in generic natural images, based on a natural language description along with an ROI mask. We achieve our goal by leveraging and combining a pretrained language-image model (CLIP), to steer the edit towards a user-provided text prompt, with a denoising diffusion probabilistic model (DDPM) to generate natural-looking results. To seamlessly fuse the edited region with the unchanged parts of the image, we spatially blend noised versions of the input image with the local text-guided diffusion latent at a progression of noise levels. In addition, we show that adding augmentations to the diffusion process mitigates adversarial results. We compare against several baselines and related methods, both qualitatively and quantitatively, and show that our method outperforms these solutions in terms of overall realism, ability to preserve the background and matching the text. Finally, we show several text-driven editing applications, including adding a new object to an image, removing/replacing/altering existing objects, background replacement, and image extrapolation. Code is available at: https://omriavrahami.com/blended-diffusion-page/ | ['Ohad Fried', 'Dani Lischinski', 'Omri Avrahami'] | 2021-11-29 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Avrahami_Blended_Diffusion_for_Text-Driven_Editing_of_Natural_Images_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Avrahami_Blended_Diffusion_for_Text-Driven_Editing_of_Natural_Images_CVPR_2022_paper.pdf | cvpr-2022-1 | ['text-guided-image-editing', 'zero-shot-text-to-image-generation'] | ['computer-vision', 'natural-language-processing'] | [ 5.35450518e-01 8.00957903e-02 3.54487032e-01 -2.53985524e-01
-5.70300221e-01 -7.72607446e-01 8.90514433e-01 -3.67889404e-02
-3.87190700e-01 4.69787478e-01 4.49431717e-01 -8.39549452e-02
2.88330913e-01 -6.30898774e-01 -7.80161202e-01 -5.65248847e-01
4.20825481e-01 1.10256476e-02 3.64035189e-01 -2.39598647e-01
2.42135346e-01 7.98028588e-01 -1.02795267e+00 3.62068415e-01
9.88025069e-01 6.80816174e-01 3.69148523e-01 9.90740538e-01
-2.15912014e-01 1.00666237e+00 -5.47946215e-01 -5.79910040e-01
5.99351883e-01 -4.72155482e-01 -6.48045659e-01 3.36212397e-01
7.15636373e-01 -6.90859437e-01 -5.29673517e-01 1.08303893e+00
4.50357884e-01 1.72946483e-01 6.54884517e-01 -1.14684379e+00
-1.03712881e+00 2.37499639e-01 -1.06515658e+00 -4.17526402e-02
4.46097285e-01 4.76630837e-01 3.13089579e-01 -8.95649135e-01
1.11848938e+00 1.39851069e+00 5.28986990e-01 7.55242169e-01
-1.57056499e+00 -3.77540082e-01 1.96384445e-01 -3.76851827e-01
-1.26086426e+00 -6.00091636e-01 8.62814844e-01 -5.37892640e-01
5.09459674e-01 3.80233049e-01 3.57882619e-01 1.17760086e+00
4.17137444e-01 5.64308405e-01 1.21797013e+00 -6.36401415e-01
2.75095496e-02 3.79035681e-01 -4.68527228e-01 6.56921387e-01
-2.59302706e-01 1.84383184e-01 -4.18506742e-01 -1.36643231e-01
9.57886100e-01 4.98245992e-02 -4.38955158e-01 -4.51311946e-01
-1.22101498e+00 4.79397327e-01 4.01589602e-01 4.47534248e-02
-4.92148727e-01 3.45661432e-01 2.33188078e-01 2.26365536e-01
8.17830384e-01 1.64412677e-01 -7.82226119e-03 1.82447553e-01
-1.27510166e+00 3.05356860e-01 4.83181536e-01 9.94339287e-01
7.94968247e-01 4.96115647e-02 -6.28267050e-01 8.44402492e-01
6.43685833e-02 5.49636185e-01 2.40278132e-02 -1.51197255e+00
1.50301859e-01 1.59224451e-01 2.33470008e-01 -1.03568029e+00
2.15179101e-01 -1.38863742e-01 -9.08214033e-01 9.75228548e-01
2.22454593e-01 -2.26904154e-01 -1.05121160e+00 1.69688606e+00
3.69846374e-01 -1.30865753e-01 -2.03783691e-01 7.78497219e-01
3.91329736e-01 8.12301636e-01 1.62747756e-01 -2.61092428e-02
1.08210349e+00 -1.07425368e+00 -7.89681315e-01 -2.72053242e-01
8.34875628e-02 -1.03874588e+00 1.25573385e+00 3.96009624e-01
-1.41371179e+00 -4.73792404e-01 -8.13182175e-01 -4.94102150e-01
-2.64964253e-01 3.83743308e-02 -1.19193438e-02 4.50810999e-01
-1.39049900e+00 5.37005901e-01 -6.09156966e-01 -4.44285959e-01
4.14693028e-01 -1.36283353e-01 -3.53264242e-01 -1.85402125e-01
-8.09681833e-01 9.82162893e-01 -9.14394408e-02 -2.14550510e-01
-9.97317612e-01 -9.36141431e-01 -7.50374198e-01 -3.44103575e-01
2.89308071e-01 -9.72720027e-01 1.04269350e+00 -1.31474805e+00
-1.55816555e+00 1.03963244e+00 -1.68373913e-01 -3.92837822e-01
1.24255800e+00 -2.32676432e-01 -9.37841535e-02 1.85078695e-01
5.32399379e-02 1.30736637e+00 1.21834743e+00 -1.69175649e+00
-2.86047280e-01 4.28949036e-02 1.03170626e-01 2.82824755e-01
-1.50741667e-01 2.57568657e-01 -7.75212526e-01 -1.40638912e+00
-4.30836231e-01 -7.49417484e-01 -3.76709491e-01 1.05080998e+00
-6.08434916e-01 6.74276531e-01 1.22687662e+00 -1.11673224e+00
1.07937634e+00 -2.41867042e+00 1.60401046e-01 2.38164678e-01
4.37464029e-01 6.49346858e-02 -3.62872303e-01 5.50717950e-01
-4.95539159e-02 1.63506731e-01 -7.13065982e-01 -9.30003107e-01
-1.88323289e-01 5.07546738e-02 -5.52646756e-01 4.14295405e-01
1.58006072e-01 9.49820399e-01 -8.00356388e-01 -4.68594044e-01
4.94379669e-01 8.54462504e-01 -5.75473189e-01 1.56828672e-01
-4.31502670e-01 6.33616388e-01 1.11991234e-01 4.23150510e-01
1.00096989e+00 2.50532389e-01 -2.73358852e-01 -3.48566145e-01
-1.09742537e-01 -3.24868500e-01 -1.13014948e+00 1.78833032e+00
-5.61376810e-01 8.52704048e-01 4.49195176e-01 -1.99553028e-01
6.89675629e-01 -2.61701606e-02 2.30291635e-01 -5.78694403e-01
-1.02138042e-01 -2.46552862e-02 -4.75558639e-01 -2.60746241e-01
6.64669394e-01 -2.84133758e-02 3.16734076e-01 6.37156785e-01
-2.07181960e-01 -5.13381898e-01 1.05618715e-01 6.64041281e-01
9.97025311e-01 3.44057232e-01 1.52905593e-02 -1.89135805e-01
2.65585244e-01 -1.40169874e-01 1.21433549e-01 7.98123300e-01
-8.91256928e-02 1.25468171e+00 3.53231937e-01 -1.27992183e-01
-1.23788679e+00 -1.29695034e+00 2.22350851e-01 7.89510667e-01
1.97534651e-01 -2.45157048e-01 -1.12366986e+00 -6.02911711e-01
-6.95831031e-02 1.04609025e+00 -6.99308932e-01 -1.26979262e-01
-3.72946054e-01 -1.71044111e-01 5.93168259e-01 3.34493130e-01
6.41467810e-01 -9.73052859e-01 -4.26289350e-01 3.61309536e-02
-1.97768807e-01 -1.01710367e+00 -1.23549116e+00 -1.43710375e-01
-6.49157047e-01 -4.93281633e-01 -1.18369555e+00 -6.36298656e-01
1.00473869e+00 4.19504911e-01 8.54071081e-01 1.87440395e-01
-4.14030522e-01 6.14508212e-01 -6.89708367e-02 -1.97129279e-01
-9.88792539e-01 -4.85552579e-01 -1.91073850e-01 1.76718518e-01
-5.55457711e-01 -5.00902653e-01 -8.12269211e-01 1.40180275e-01
-1.47423697e+00 4.84406948e-01 3.39709342e-01 5.07972896e-01
4.43583816e-01 -5.06939627e-02 8.06919858e-03 -7.75103211e-01
8.27789724e-01 -4.90355715e-02 -4.48278517e-01 2.37974763e-01
-3.45243245e-01 4.90252189e-02 4.87494081e-01 -6.07543290e-01
-1.43896389e+00 1.14998199e-01 1.39053455e-02 -7.41258621e-01
-4.23980765e-02 -4.12953496e-02 -8.73760507e-02 -2.85339952e-01
9.15821373e-01 1.67674735e-01 2.53425032e-01 -3.05852741e-01
1.08707893e+00 4.36151236e-01 7.98870623e-01 -5.33858955e-01
1.02725327e+00 9.05488729e-01 -3.94647747e-01 -5.75705290e-01
-3.94587189e-01 1.75863635e-02 -6.96733415e-01 -3.08933914e-01
8.45256388e-01 -7.96177626e-01 -3.59658413e-02 7.07717896e-01
-1.37511778e+00 -9.15254712e-01 -7.57067263e-01 -7.96924680e-02
-6.52841389e-01 5.84891498e-01 -7.38999665e-01 -5.40546417e-01
-2.56669879e-01 -1.20233274e+00 1.10842693e+00 8.26995522e-02
-3.35902125e-01 -1.10807621e+00 -1.82091780e-02 7.30357990e-02
8.67155135e-01 4.64284599e-01 7.22787082e-01 5.30086234e-02
-6.26155615e-01 1.53056309e-02 -3.79274100e-01 6.72443330e-01
2.52155691e-01 3.40012282e-01 -9.68892872e-01 -1.79411292e-01
-6.79302737e-02 -3.03920042e-02 8.50442231e-01 4.21922266e-01
1.10184371e+00 -3.63721848e-01 -2.54495770e-01 6.81825757e-01
1.30000138e+00 -3.88868675e-02 9.39458668e-01 2.06971034e-01
7.00015306e-01 6.06650591e-01 3.58711451e-01 3.57528567e-01
1.32200092e-01 5.55638969e-01 2.92195588e-01 -6.13460600e-01
-8.01424801e-01 -3.69842678e-01 4.69017118e-01 5.68081066e-02
1.74458310e-01 -5.09555399e-01 -5.46088815e-01 4.88262355e-01
-1.56376624e+00 -8.86671603e-01 2.72634141e-02 2.17565179e+00
1.04127479e+00 1.05148712e-02 -1.89717766e-02 -3.67234886e-01
8.28183413e-01 4.14035141e-01 -5.82067490e-01 -4.46608365e-01
-2.37581015e-01 -2.83258297e-02 6.19413137e-01 1.04645753e+00
-9.52942550e-01 1.09365129e+00 6.42830324e+00 8.64032030e-01
-1.16459739e+00 2.74182260e-01 9.23346519e-01 -2.80751705e-01
-6.08729720e-01 -2.26998720e-02 -1.61560223e-01 3.11371684e-01
4.13369089e-01 -2.55419284e-01 7.39178479e-01 3.69526625e-01
7.03913808e-01 -1.71466723e-01 -8.36333275e-01 8.20683181e-01
2.35973075e-01 -1.52719164e+00 3.55073392e-01 -1.09628914e-02
1.01562262e+00 -2.15458453e-01 2.71948576e-01 -2.15244114e-01
5.58642507e-01 -8.19697499e-01 1.20760131e+00 9.57178116e-01
1.01729274e+00 -3.91473800e-01 1.61393374e-01 2.06924126e-01
-8.30209851e-01 2.10231170e-01 -3.36319134e-02 3.33774477e-01
4.74496454e-01 7.57009685e-01 -3.26675504e-01 3.05586904e-01
7.16001868e-01 5.07187307e-01 -6.35291696e-01 8.02484989e-01
-2.55923748e-01 1.07094318e-01 -2.52485871e-01 6.59621418e-01
-7.73609057e-02 -3.68866920e-01 7.25547910e-01 1.47647452e+00
2.47462317e-01 -7.84768164e-03 -6.70687035e-02 1.46970332e+00
-2.47110859e-01 -2.71926895e-02 -7.18203187e-01 2.14807972e-01
2.34335929e-01 1.25949752e+00 -6.89730287e-01 -4.69886333e-01
-2.26943254e-01 1.90337658e+00 7.81770051e-03 7.14988291e-01
-9.80483830e-01 -4.68739957e-01 6.32961094e-01 2.99927384e-01
9.01585519e-02 -2.07155406e-01 -4.28387642e-01 -9.62826371e-01
1.73509791e-01 -9.41678584e-01 -1.06801108e-01 -1.43609452e+00
-1.17785585e+00 5.12243152e-01 -1.04973009e-02 -1.09737170e+00
2.02541500e-01 -1.33528844e-01 -8.53735805e-01 1.26128197e+00
-1.25517511e+00 -1.20763636e+00 -5.97464859e-01 5.88508666e-01
6.85326397e-01 1.70802310e-01 3.61905664e-01 3.34165841e-01
-2.00960532e-01 4.51014072e-01 -2.22935396e-05 -4.63222302e-02
1.23016262e+00 -1.21704030e+00 7.45082915e-01 1.16536200e+00
-1.46919459e-01 6.08101845e-01 8.53727579e-01 -8.78176570e-01
-1.01013899e+00 -1.29057503e+00 4.31854755e-01 -5.06827414e-01
4.49622184e-01 -4.00397390e-01 -9.44987237e-01 7.48531222e-01
7.33354628e-01 4.03630398e-02 3.67799355e-03 -8.36114407e-01
-2.55259216e-01 -2.65998412e-02 -1.45404923e+00 1.15410757e+00
1.02478015e+00 -6.78659797e-01 -1.32612869e-01 3.07590693e-01
8.79906058e-01 -6.99438810e-01 -6.48363531e-01 1.41818244e-02
4.31540132e-01 -1.00745821e+00 1.08142412e+00 -1.13556549e-01
5.58602750e-01 -6.42294586e-01 -3.04298406e-03 -1.50593448e+00
-2.31041223e-01 -9.99991477e-01 1.75312713e-01 1.45468557e+00
2.09754407e-01 -4.43077892e-01 4.52055275e-01 8.04229200e-01
8.68402943e-02 -4.00208473e-01 -5.23492217e-01 -5.51464319e-01
9.66474935e-02 -3.02027851e-01 1.66812494e-01 8.71692121e-01
-4.80920881e-01 -2.12083489e-01 -6.32805765e-01 2.16047913e-01
6.57331586e-01 -4.32103187e-01 9.46040750e-01 -4.03588265e-01
-2.09312871e-01 -5.45972586e-01 -1.69337288e-01 -1.16965234e+00
-2.34913975e-01 -7.13309228e-01 7.53337890e-02 -1.68692482e+00
1.40592813e-01 -2.19066188e-01 2.45094359e-01 3.82602394e-01
-1.23793498e-01 3.80736470e-01 4.44146007e-01 1.90970615e-01
-1.02631100e-01 5.42405307e-01 1.38911092e+00 -3.35918844e-01
-2.19863147e-01 -3.42344582e-01 -7.97469378e-01 6.86912715e-01
6.65988684e-01 -4.97033685e-01 -3.75026017e-01 -6.88798249e-01
-1.25871733e-01 -1.53342783e-01 6.45402610e-01 -8.49224508e-01
3.32798988e-01 -2.16631457e-01 4.20371026e-01 -2.09342495e-01
3.41702104e-01 -7.51174033e-01 3.45169067e-01 3.82208735e-01
-5.77754438e-01 1.09157726e-01 4.03501511e-01 6.52132154e-01
2.96573192e-02 -1.00693919e-01 1.17015409e+00 5.42721376e-02
-4.75003719e-01 2.13468179e-01 -3.98611158e-01 -8.82129297e-02
1.13545430e+00 -3.41643393e-01 -3.60425353e-01 -8.36392283e-01
-8.76559198e-01 3.85413282e-02 1.13996863e+00 5.22705257e-01
6.14198864e-01 -1.10171831e+00 -8.01669478e-01 1.95060819e-01
-1.25088617e-01 -1.99339896e-01 4.22136545e-01 7.76583672e-01
-1.02671182e+00 -3.97044808e-01 -1.33303508e-01 -4.58124191e-01
-1.21764207e+00 6.45589411e-01 6.78300917e-01 -2.64899377e-02
-8.57963383e-01 6.83574319e-01 5.78975201e-01 -4.35713083e-01
4.54813391e-01 -3.56205076e-01 5.52636206e-01 -3.80480766e-01
6.85387135e-01 2.17381477e-01 -1.50541604e-01 -3.43731582e-01
-1.43415019e-01 6.19193733e-01 -1.58064246e-01 -6.55746400e-01
9.76780951e-01 -6.26685381e-01 -1.26495153e-01 1.03594758e-01
9.96043742e-01 6.25946224e-01 -1.79924762e+00 -3.48568082e-01
-4.76839185e-01 -6.25933170e-01 1.23706162e-01 -1.08913100e+00
-1.29992008e+00 7.39180684e-01 7.30983138e-01 -7.69328028e-02
1.23047662e+00 -1.54362157e-01 6.54855013e-01 -1.48919970e-01
-4.15049307e-02 -8.60942185e-01 2.13182271e-01 1.69481009e-01
1.47729564e+00 -1.03411460e+00 1.67922482e-01 -5.14443517e-01
-9.74974275e-01 9.32584703e-01 4.10509020e-01 -1.41618997e-01
6.55371189e-01 6.39806986e-01 4.58223283e-01 6.04970306e-02
-5.49422860e-01 1.77696735e-01 8.85770768e-02 8.57193291e-01
2.07109883e-01 -2.62618959e-01 1.22851729e-01 -1.36526942e-01
5.40075563e-02 1.02396850e-02 7.31233835e-01 8.85467410e-01
-2.06015497e-01 -1.07218421e+00 -6.26757681e-01 1.72102213e-01
-1.90991953e-01 -4.09737974e-01 -5.84385276e-01 5.79385638e-01
4.46642712e-02 7.95281887e-01 -6.95146620e-02 -2.88921427e-02
3.49700689e-01 -1.84639394e-01 4.27669764e-01 -4.02948350e-01
-5.24277151e-01 1.78404555e-01 -1.10530809e-01 -6.50197148e-01
-2.32442707e-01 -5.79422534e-01 -1.02459002e+00 -5.37343204e-01
8.35621133e-02 -4.28324401e-01 6.39881492e-01 5.37352681e-01
5.94861686e-01 5.41814446e-01 4.69142884e-01 -1.26289296e+00
-2.06518605e-01 -6.00732565e-01 -4.53434497e-01 5.29418826e-01
3.68918389e-01 -2.23150000e-01 -4.59443331e-01 6.43526316e-01] | [11.405402183532715, -0.5176979899406433] |
0821e528-5652-4ef9-8874-30bb8232653d | crowd-robot-interaction-crowd-aware-robot | 1809.08835 | null | http://arxiv.org/abs/1809.08835v2 | http://arxiv.org/pdf/1809.08835v2.pdf | Crowd-Robot Interaction: Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning | Mobility in an effective and socially-compliant manner is an essential yet
challenging task for robots operating in crowded spaces. Recent works have
shown the power of deep reinforcement learning techniques to learn socially
cooperative policies. However, their cooperation ability deteriorates as the
crowd grows since they typically relax the problem as a one-way Human-Robot
interaction problem. In this work, we want to go beyond first-order Human-Robot
interaction and more explicitly model Crowd-Robot Interaction (CRI). We propose
to (i) rethink pairwise interactions with a self-attention mechanism, and (ii)
jointly model Human-Robot as well as Human-Human interactions in the deep
reinforcement learning framework. Our model captures the Human-Human
interactions occurring in dense crowds that indirectly affects the robot's
anticipation capability. Our proposed attentive pooling mechanism learns the
collective importance of neighboring humans with respect to their future
states. Various experiments demonstrate that our model can anticipate human
dynamics and navigate in crowds with time efficiency, outperforming
state-of-the-art methods. | ['Changan Chen', 'Alexandre Alahi', 'Yuejiang Liu', 'Sven Kreiss'] | 2018-09-24 | null | null | null | null | ['human-dynamics'] | ['computer-vision'] | [-4.91021663e-01 5.08904397e-01 3.38763535e-01 -1.57179907e-01
-7.29595823e-03 1.34610087e-02 5.93343198e-01 -1.03095397e-01
-9.22707558e-01 1.18998468e+00 4.36761707e-01 2.60785908e-01
-2.06617117e-01 -6.18418455e-01 -7.33536661e-01 -9.04549599e-01
-7.32880831e-01 7.39971638e-01 4.75550234e-01 -9.08996880e-01
-5.04541732e-02 2.36153007e-01 -1.58348036e+00 -2.94157773e-01
1.04818010e+00 2.95829415e-01 5.98424852e-01 1.02416050e+00
4.76420671e-01 1.47660303e+00 -4.79427725e-01 -2.27898657e-01
1.60765991e-01 -1.94106534e-01 -8.80182266e-01 -1.70505345e-02
-6.91121519e-01 -4.31112349e-01 -7.63458252e-01 6.09138548e-01
7.00035870e-01 5.80024958e-01 6.85246050e-01 -1.71816146e+00
-7.31297374e-01 7.14067817e-01 -5.10947227e-01 -5.53041473e-02
2.93334991e-01 5.63021004e-01 7.02369094e-01 -4.13488775e-01
5.89937925e-01 1.68272507e+00 4.43090379e-01 7.33671904e-01
-4.14745867e-01 -3.93614262e-01 4.47348893e-01 2.71578133e-01
-9.22065198e-01 -1.15936480e-01 4.06635016e-01 -4.80259836e-01
1.22130573e+00 -3.33783716e-01 9.61429298e-01 1.02339244e+00
4.58146900e-01 1.05093181e+00 5.46457946e-01 1.63147449e-02
4.38889563e-01 -5.95670283e-01 -2.92862564e-01 5.65578520e-01
2.83612579e-01 -1.32628918e-01 -6.45031035e-01 -2.37789378e-01
6.89896643e-01 3.95205878e-02 2.42283242e-03 -4.18069720e-01
-1.33185101e+00 7.98417687e-01 9.81221497e-01 2.65541226e-01
-1.00721276e+00 6.46519244e-01 2.07454517e-01 1.83677033e-01
9.09037068e-02 4.80733812e-01 -3.29263121e-01 -3.24562013e-01
-3.24157141e-02 7.38026381e-01 9.84358728e-01 8.59833002e-01
7.45671034e-01 -2.24425033e-01 -2.96557751e-02 4.99575734e-01
4.16869998e-01 5.56540191e-01 3.09456527e-01 -1.47241676e+00
3.85582238e-01 4.79543209e-01 8.33415568e-01 -1.10600722e+00
-8.66060555e-01 1.58452094e-02 -8.52804124e-01 2.29567498e-01
3.02797437e-01 -7.32029796e-01 -4.18258816e-01 1.85791540e+00
4.30592269e-01 -3.92755605e-02 5.68459809e-01 1.09684968e+00
5.87132037e-01 4.53339905e-01 2.82520235e-01 6.62014857e-02
1.12127399e+00 -1.50169647e+00 -5.80437481e-01 -5.52485526e-01
7.21654236e-01 3.34183909e-02 6.47287250e-01 -2.69868691e-02
-1.15537012e+00 -1.97477028e-01 -5.18878400e-01 5.96021190e-02
-1.64029658e-01 -2.41232395e-01 6.67977154e-01 -1.42186895e-01
-1.47602999e+00 5.92763722e-01 -1.10624778e+00 -8.84531498e-01
4.85678434e-01 7.97133803e-01 -4.42059994e-01 6.80856034e-02
-1.42192566e+00 1.01865506e+00 5.08240014e-02 2.00457916e-01
-9.41412687e-01 5.02947867e-02 -9.01354551e-01 6.56021237e-02
4.18385595e-01 -9.41785634e-01 1.40437305e+00 -7.69487560e-01
-1.79894674e+00 5.82345426e-01 -2.08535895e-01 -6.84803128e-01
6.88606620e-01 -4.28391546e-01 3.26848775e-01 1.59003153e-01
4.01173264e-01 1.14219189e+00 2.67096579e-01 -1.47942042e+00
-8.64527166e-01 -7.57055283e-02 3.78278166e-01 8.91550720e-01
-6.83269799e-02 -1.87277034e-01 -2.26205438e-01 -1.38544247e-01
-4.93453532e-01 -1.39686692e+00 -8.47089350e-01 -1.49158970e-01
-1.39352024e-01 -5.44420123e-01 5.12455285e-01 -2.65735149e-01
5.09256542e-01 -1.79168677e+00 4.61201996e-01 -2.17504352e-01
5.02221465e-01 2.90000916e-01 -3.63150865e-01 8.46700251e-01
8.31000030e-01 -2.03734696e-01 -2.17049092e-01 -6.68860614e-01
3.32891077e-01 6.06040299e-01 1.30951166e-01 5.11643708e-01
2.17539355e-01 1.27327561e+00 -1.57767117e+00 -4.84570533e-01
-1.27759069e-01 3.87943655e-01 -7.67649412e-01 4.99764860e-01
2.59830952e-02 1.01734352e+00 -7.79992223e-01 4.45913076e-01
4.40911770e-01 -4.03206468e-01 3.04816663e-01 8.27774048e-01
-7.60771409e-02 -2.09067002e-01 -7.05253065e-01 1.17957711e+00
-2.47941077e-01 5.74284136e-01 5.24058223e-01 -7.71480024e-01
7.14112937e-01 1.18142016e-01 6.29471421e-01 -4.64646399e-01
2.00662717e-01 1.31989926e-01 2.48224363e-01 -9.13031757e-01
7.84658909e-01 1.18898675e-01 -3.29156429e-01 3.79858911e-01
-2.03389511e-01 4.47319560e-02 -2.11883858e-02 2.16645882e-01
1.41504169e+00 7.75551051e-02 3.75941724e-01 -2.83328712e-01
2.91729093e-01 1.61295924e-02 6.15733147e-01 1.06774294e+00
-8.20712745e-01 2.57774871e-02 5.38926899e-01 -4.45848554e-01
-7.41888881e-01 -8.38570833e-01 7.98506796e-01 1.60506213e+00
7.47288287e-01 3.00336778e-01 -7.80510604e-01 -3.39735121e-01
2.11880341e-01 2.64741838e-01 -8.77923727e-01 -3.22473515e-03
-9.44775045e-01 -6.92430019e-01 4.05948609e-01 4.16819572e-01
7.52352774e-01 -1.77004254e+00 -1.18691707e+00 3.53841394e-01
-4.61270571e-01 -1.06458783e+00 -2.79312521e-01 1.65113837e-01
-1.67582646e-01 -9.75679696e-01 -1.01613331e+00 -1.07697678e+00
6.04305148e-01 4.52980995e-01 9.56512749e-01 3.90443712e-01
1.25164866e-01 6.93288088e-01 -5.55901825e-01 -3.90659302e-01
-7.41225407e-02 1.91002220e-01 4.97854322e-01 -1.99289590e-01
3.21717411e-01 -5.75397193e-01 -9.63869393e-01 4.60304469e-01
-5.66157162e-01 -3.17350149e-01 4.83660817e-01 8.54360759e-01
-1.83603674e-01 -6.46189079e-02 7.75450587e-01 -9.20533985e-02
1.06787169e+00 -1.00644863e+00 -1.49715826e-01 1.92225054e-02
2.19562456e-01 -3.42877582e-02 4.91928697e-01 -4.31661755e-01
-1.07382703e+00 2.81730354e-01 3.41194868e-01 6.14344627e-02
-1.62886664e-01 2.49085933e-01 4.46121767e-02 -5.89639395e-02
3.67639393e-01 2.66352519e-02 1.75653622e-01 2.00386569e-01
2.52668798e-01 7.46438444e-01 4.26183730e-01 -5.72746217e-01
4.62275624e-01 7.87926674e-01 3.13317627e-02 -1.00952590e+00
-1.49171486e-01 -3.63998771e-01 -6.74869478e-01 -5.01804292e-01
1.12242377e+00 -1.28303277e+00 -1.76326144e+00 8.24059069e-01
-1.47517550e+00 -1.04164040e+00 -9.66676101e-02 4.30120438e-01
-8.10495794e-01 4.20979083e-01 -8.05801988e-01 -1.40166914e+00
1.24797277e-01 -1.15948236e+00 1.19039452e+00 5.85304677e-01
-6.94482476e-02 -8.79006207e-01 2.81775713e-01 1.66641429e-01
5.46435773e-01 9.97943580e-02 -1.31567633e-02 -4.35000330e-01
-5.98340571e-01 1.70920938e-01 -7.17724189e-02 -3.83825392e-01
-2.08429724e-01 -3.05130512e-01 -5.49146235e-01 -1.94092736e-01
-4.07356113e-01 -6.29889071e-01 8.19383025e-01 6.18137836e-01
2.48451620e-01 -3.03602904e-01 -6.11530542e-01 -4.93448973e-02
6.25531733e-01 1.40395880e-01 5.87408602e-01 4.63945150e-01
5.29210746e-01 1.47962177e+00 7.66786575e-01 8.52506220e-01
1.51160717e+00 4.01909739e-01 9.20662701e-01 9.18386802e-02
5.66573560e-01 -2.42208570e-01 4.88972485e-01 6.56020820e-01
-3.93788517e-01 -7.96480596e-01 -1.12101161e+00 7.97259212e-01
-2.78072047e+00 -1.08196294e+00 -1.60452634e-01 1.46586943e+00
2.49702379e-01 -4.07276273e-01 4.18356687e-01 -5.35926342e-01
8.45226467e-01 -4.65125851e-02 -7.56308138e-01 -1.14991993e-01
-2.05506757e-01 -7.53278732e-01 5.66500008e-01 6.89987123e-01
-1.10942543e+00 1.47315371e+00 5.92267799e+00 1.70002237e-01
-5.96482217e-01 3.94092984e-02 4.14338827e-01 1.44896125e-02
2.23767817e-01 -3.42124134e-01 -5.32122433e-01 3.03541660e-01
4.82389659e-01 1.23293586e-02 7.25247920e-01 7.93020070e-01
4.36060727e-01 -5.66623390e-01 -7.44104147e-01 7.14815676e-01
-2.22177923e-01 -7.48614848e-01 -4.84221548e-01 4.11326379e-01
8.84810090e-01 2.97627658e-01 -1.67149365e-01 6.54790759e-01
1.23265314e+00 -1.20980871e+00 7.61210442e-01 8.81829858e-01
-2.37277880e-01 -6.29297972e-01 1.03896821e+00 8.21647286e-01
-1.29912972e+00 -6.77230477e-01 -4.64382857e-01 -8.89307499e-01
6.21988058e-01 1.12798810e-01 -7.17608631e-01 2.74678260e-01
9.38923717e-01 7.53952324e-01 -1.32549077e-01 8.83278668e-01
-1.98673427e-01 -1.30373895e-01 -2.35255867e-01 -8.71118844e-01
7.08508313e-01 -1.22924402e-01 6.15432739e-01 8.69386792e-01
3.41944277e-01 4.37249571e-01 4.30862486e-01 4.28393632e-01
2.19120644e-02 -1.49836630e-01 -8.22957695e-01 4.24363725e-02
4.89950418e-01 7.74756312e-01 -7.12899148e-01 -2.36979887e-01
-2.32940372e-02 1.25265479e+00 8.36223066e-01 5.34813941e-01
-7.28781462e-01 -3.20448786e-01 1.10952914e+00 -1.86168119e-01
4.43147570e-01 -4.86305743e-01 2.39616692e-01 -9.21242177e-01
2.48363987e-02 -3.46674353e-01 -1.16183288e-01 -7.83204138e-01
-1.35912943e+00 6.47815287e-01 -3.01906228e-01 -1.03062439e+00
-3.41910183e-01 -3.47453356e-01 -6.91878498e-01 3.82127017e-01
-1.58157527e+00 -1.18845534e+00 -4.01658744e-01 5.08391321e-01
2.63044000e-01 -2.37673253e-01 4.28153008e-01 -2.68611193e-01
-2.58720994e-01 7.46457875e-02 5.68256825e-02 7.35999122e-02
5.90769172e-01 -1.01146591e+00 4.17230040e-01 3.57698470e-01
-9.03708041e-01 4.35388595e-01 8.11343372e-01 -9.34845567e-01
-1.18882513e+00 -1.00610662e+00 1.03885996e+00 -4.83797163e-01
6.93759024e-01 -2.44525969e-01 -5.70169449e-01 5.22074819e-01
3.80207896e-01 1.47084102e-01 3.19484234e-01 -2.73921758e-01
3.59228283e-01 5.23967206e-01 -1.12292933e+00 1.03196084e+00
1.45229316e+00 -5.50805032e-02 -5.50736070e-01 2.92274356e-01
1.00295341e+00 -2.33236805e-01 -3.46495211e-01 1.23858146e-01
5.04001319e-01 -9.94922578e-01 8.48380804e-01 -6.58178926e-01
4.60025668e-01 -2.36036986e-01 -1.39418393e-01 -1.48564899e+00
-4.07527268e-01 -9.71948147e-01 -1.01982933e-02 5.00456810e-01
2.81346869e-02 -8.48623514e-01 7.73603618e-01 6.26848936e-01
6.75411522e-02 -7.01741219e-01 -1.14094341e+00 -7.39594221e-01
2.70386994e-01 2.08080262e-01 7.14630187e-01 6.36249781e-01
5.11043072e-01 1.42204121e-01 -8.40443492e-01 3.76820087e-01
5.17297387e-01 -6.04437172e-01 1.14559960e+00 -1.23911476e+00
-2.43334904e-01 -3.75261873e-01 -3.49341989e-01 -1.30317938e+00
5.84995508e-01 -3.52998346e-01 9.15215850e-01 -1.81765938e+00
1.65782318e-01 -3.58978778e-01 2.45779052e-01 2.37633318e-01
-3.00932199e-01 -1.92025587e-01 4.11077291e-01 3.33557308e-01
-1.46728277e+00 1.18020153e+00 1.40354991e+00 9.61978436e-02
-3.99416327e-01 -1.34830952e-01 -6.09196007e-01 7.79308081e-01
8.89451385e-01 -3.07151526e-01 -1.18494183e-01 -6.22624397e-01
3.08124006e-01 3.59898627e-01 4.89575088e-01 -1.18121672e+00
7.77416646e-01 -3.88957441e-01 -1.04087777e-01 -3.75385545e-02
4.73500729e-01 -5.34916639e-01 -4.26539421e-01 9.72176790e-01
-3.83250386e-01 9.27390307e-02 -3.28359902e-01 1.14726579e+00
2.83809025e-02 1.06019989e-01 4.60592508e-01 -3.00271094e-01
-5.80106139e-01 2.46543795e-01 -1.11546993e+00 7.38118365e-02
1.37627971e+00 6.77713901e-02 -3.74268442e-01 -9.57379282e-01
-7.44898319e-01 1.24649858e+00 6.93306029e-01 3.40931475e-01
3.73188883e-01 -9.49548185e-01 -7.85012245e-01 -4.68337685e-01
4.16759625e-02 7.93119594e-02 5.34776151e-01 7.93504596e-01
-5.68671525e-01 2.72865742e-01 -3.85776192e-01 -3.18061560e-01
-8.01979005e-01 2.95984507e-01 3.89348239e-01 -6.19071424e-01
-3.75559062e-01 9.97738361e-01 3.95830721e-01 -1.04397440e+00
4.21135634e-01 -3.28799561e-02 -5.67643106e-01 -2.01775163e-01
3.49034488e-01 6.08011961e-01 -8.57541919e-01 -1.03575540e+00
-5.67070723e-01 4.04935420e-01 4.00668770e-01 -2.15720490e-01
1.48773468e+00 -5.76680183e-01 -9.72544253e-02 1.11069210e-01
7.43397176e-01 -4.65763986e-01 -1.75370896e+00 -1.39803827e-01
-3.10181230e-02 -4.41510347e-04 -5.47176242e-01 -3.09983939e-01
-5.47383249e-01 7.65261352e-01 1.77565679e-01 2.84173667e-01
5.16686797e-01 2.34678492e-01 9.77667868e-01 1.10999978e+00
7.81071186e-01 -1.30511212e+00 5.31306148e-01 1.12001085e+00
9.66144681e-01 -1.72569227e+00 -4.84830767e-01 4.06233780e-02
-1.37451506e+00 8.00273776e-01 8.04683864e-01 -5.38767815e-01
6.79096401e-01 6.70102239e-02 -9.20749679e-02 -1.93646908e-01
-8.81814420e-01 -7.20894217e-01 -5.10020256e-01 1.15845430e+00
4.14639525e-02 3.05667102e-01 -4.63905334e-02 7.53001750e-01
-2.32556775e-01 1.70069244e-02 7.06129432e-01 1.17143643e+00
-9.09214318e-01 -6.22645438e-01 -3.50340843e-01 7.66998455e-02
1.37711480e-01 4.28301871e-01 -5.73818982e-01 6.99535131e-01
9.15909931e-02 1.43917203e+00 1.95288509e-01 -3.51680964e-01
2.89482653e-01 -4.94749963e-01 -3.73790339e-02 -3.55910927e-01
-6.71766222e-01 -4.58979577e-01 -7.73083493e-02 -6.76638067e-01
-7.08575428e-01 -7.73011148e-01 -1.71683872e+00 -4.21823114e-01
2.68781811e-01 1.91987976e-01 2.86360651e-01 1.15224266e+00
5.58300614e-01 4.69138414e-01 5.20333946e-01 -1.82932425e+00
-3.44905972e-01 -9.34978724e-01 -5.14876664e-01 2.01704636e-01
6.11703038e-01 -9.27283585e-01 -3.92131180e-01 -2.89408118e-01] | [4.794475078582764, 0.9858051538467407] |
a81e2c7b-ec4d-46f2-b0b9-5bb873ee6e41 | a-systematic-study-on-object-recognition | 2305.02085 | null | https://arxiv.org/abs/2305.02085v1 | https://arxiv.org/pdf/2305.02085v1.pdf | A Systematic Study on Object Recognition Using Millimeter-wave Radar | Due to its light and weather-independent sensing, millimeter-wave (MMW) radar is essential in smart environments. Intelligent vehicle systems and industry-grade MMW radars have integrated such capabilities. Industry-grade MMW radars are expensive and hard to get for community-purpose smart environment applications. However, commercially available MMW radars have hidden underpinning challenges that need to be investigated for tasks like recognizing objects and activities, real-time person tracking, object localization, etc. Image and video data are straightforward to gather, understand, and annotate for such jobs. Image and video data are light and weather-dependent, susceptible to the occlusion effect, and present privacy problems. To eliminate dependence and ensure privacy, commercial MMW radars should be tested. MMW radar's practicality and performance in varied operating settings must be addressed before promoting it. To address the problems, we collected a dataset using Texas Instruments' Automotive mmWave Radar (AWR2944) and reported the best experimental settings for object recognition performance using different deep learning algorithms. Our extensive data gathering technique allows us to systematically explore and identify object identification task problems under cross-ambience conditions. We investigated several solutions and published detailed experimental data. | ['Nirmalya Roy', 'Biplab Pal', 'Marc Conn', 'Zahid Hasan', 'Emon Dey', 'Mohammad Saeid Anwar', 'Avijoy Chakma', 'Maloy Kumar Devnath'] | 2023-05-03 | null | null | null | null | ['object-recognition', 'object-localization'] | ['computer-vision', 'computer-vision'] | [ 2.42180377e-01 -4.32081431e-01 7.84523785e-02 -8.26703608e-01
-4.44788158e-01 -4.01561856e-01 4.40169990e-01 -4.94185477e-01
-4.91906762e-01 9.17192578e-01 -2.34462455e-01 -4.80852395e-01
-2.63482243e-01 -8.51945460e-01 -1.79577351e-01 -8.94177139e-01
-2.75553539e-02 2.52109706e-01 -4.22092192e-02 5.71264103e-02
1.36800006e-01 8.01799059e-01 -1.55200696e+00 -1.90313324e-01
6.13838732e-01 1.15741360e+00 1.68850645e-01 8.25188518e-01
7.04352260e-02 2.99144506e-01 -8.60118210e-01 -2.39773840e-01
9.13128018e-01 1.67315423e-01 2.53117114e-01 -2.55691022e-01
7.85070658e-01 -4.37180817e-01 -4.69613522e-01 9.30725455e-01
9.57156777e-01 -1.91168487e-01 4.39698786e-01 -1.68183601e+00
-7.80005932e-01 -1.48466691e-01 -5.69663286e-01 3.48132282e-01
9.75192264e-02 5.87155819e-01 2.75390744e-01 -4.08311218e-01
4.86706942e-01 7.33236611e-01 6.63329542e-01 6.88939631e-01
-8.14402163e-01 -1.52025950e+00 -1.96624234e-01 4.91706163e-01
-1.37828660e+00 -7.88575828e-01 5.40428877e-01 -2.92356789e-01
6.88732624e-01 4.06338334e-01 5.09251475e-01 1.22256637e+00
6.23103559e-01 2.82982081e-01 1.43936825e+00 -1.69372588e-01
2.99748003e-01 3.38529795e-01 4.77928668e-01 5.94387352e-01
9.02861476e-01 5.03334224e-01 -7.18693674e-01 6.74050748e-02
1.77337274e-01 3.29155266e-01 -3.48428309e-01 -2.40015581e-01
-8.65143836e-01 5.86314559e-01 1.12366658e-02 4.12165999e-01
-4.05139148e-01 2.85378247e-01 -1.97918296e-01 5.20647109e-01
2.17622310e-01 3.88829410e-01 -4.14319813e-01 -2.10290134e-01
-7.77923942e-01 4.91346084e-02 6.73122585e-01 1.18156552e+00
8.47828627e-01 2.87206084e-01 -8.47626328e-02 3.35175723e-01
5.38214207e-01 1.43177330e+00 8.88395905e-02 -5.26890874e-01
4.92908746e-01 2.24012747e-01 1.52001753e-01 -1.12824297e+00
-6.54308558e-01 -6.69626296e-01 -9.52591717e-01 5.27213037e-01
2.62684494e-01 -5.79561770e-01 -1.28940976e+00 1.28962731e+00
8.94697383e-02 3.14432651e-01 2.39667714e-01 9.71313000e-01
1.28359568e+00 4.57138479e-01 1.00762099e-01 -3.18181932e-01
1.61652398e+00 -2.19673365e-01 -9.75852370e-01 -6.37868822e-01
3.04770231e-01 -6.53511286e-01 4.28866893e-01 2.06392601e-01
-4.64996725e-01 -4.73287404e-01 -1.33616281e+00 6.53940916e-01
-5.39131761e-01 3.58403809e-02 9.37011242e-01 1.67326736e+00
-6.18807912e-01 -2.71385342e-01 -6.42465174e-01 -5.56176484e-01
5.90691686e-01 4.09755439e-01 -2.60556877e-01 -3.36139917e-01
-1.09864151e+00 8.92988145e-01 -2.80892253e-01 4.26873475e-01
-5.63598275e-01 -7.84289062e-01 -6.18571639e-01 -5.17834544e-01
7.56566152e-02 -5.22004843e-01 7.87257969e-01 -3.15985262e-01
-1.15466845e+00 8.05335522e-01 -1.26090109e-01 -5.94281137e-01
1.65876806e-01 -2.96974145e-02 -1.30038595e+00 -1.77840590e-02
-9.58971679e-02 2.61729807e-01 9.76782084e-01 -1.06480253e+00
-7.34558403e-01 -7.08646476e-01 -4.03389424e-01 -3.54354680e-01
-3.11269648e-02 6.90754578e-02 1.63715303e-01 -1.98116347e-01
1.80179074e-01 -8.03809285e-01 -1.18472517e-01 -1.20286025e-01
-9.32242200e-02 1.97550461e-01 1.33196998e+00 -4.16726083e-01
7.24295437e-01 -2.19575953e+00 -1.16733944e+00 2.79365718e-01
1.68073982e-01 5.08288503e-01 -2.03281909e-01 1.89879045e-01
2.32673019e-01 -3.37315977e-01 1.30395308e-01 -5.97905405e-02
2.02633709e-01 4.24976610e-02 -1.43537089e-01 7.41319180e-01
-2.24098951e-01 9.76475120e-01 -1.48637563e-01 -1.02012195e-01
1.88703910e-01 2.91235954e-01 -2.33924575e-02 6.93925768e-02
1.70609653e-01 1.40625402e-01 -5.39914310e-01 1.27354431e+00
1.16156030e+00 4.14770752e-01 -5.16628146e-01 -4.67263132e-01
-2.09267125e-01 -3.92068446e-01 -1.26175523e+00 9.77364421e-01
-3.96744937e-01 1.15691292e+00 3.91925603e-01 -1.08942783e+00
1.54213738e+00 1.07160047e-01 2.92989731e-01 -1.17728829e+00
8.10948834e-02 -4.25256453e-02 -2.34849676e-02 -1.06836867e+00
3.86261731e-01 -1.89783573e-01 4.26839525e-03 6.04565740e-01
-3.51867110e-01 9.94970202e-02 -2.98104852e-01 -2.06323370e-01
1.52061486e+00 -3.12661797e-01 1.02829598e-01 -1.83822423e-01
1.46975875e-01 5.10349497e-03 5.93577027e-01 1.05305874e+00
-6.38275683e-01 2.01294050e-01 -7.50591040e-01 -8.18988681e-01
-3.38988096e-01 -1.13621056e+00 -3.41479570e-01 7.39056349e-01
3.23821843e-01 2.19334900e-01 -4.65945303e-02 -3.64095539e-01
2.34928608e-01 6.20005965e-01 -2.35040616e-02 -9.85934809e-02
-2.89034814e-01 -1.05237484e+00 5.57937860e-01 2.33013019e-01
9.38197851e-01 -7.98945546e-01 -1.14881337e+00 1.63465545e-01
7.04681128e-02 -1.35591936e+00 -8.31366628e-02 2.98636466e-01
-3.85450453e-01 -1.04359305e+00 -2.71087468e-01 -5.20823181e-01
6.00989580e-01 7.18021810e-01 8.24704826e-01 -1.85154498e-01
-1.00971544e+00 7.85669267e-01 -1.43638209e-01 -1.18060255e+00
3.44390541e-01 -3.84720445e-01 3.63895714e-01 1.72501028e-01
1.19107330e+00 -5.79896629e-01 -7.29970396e-01 3.70627165e-01
-3.56430799e-01 -4.17009532e-01 1.03430986e+00 3.35081875e-01
-5.69283497e-04 3.12956005e-01 7.49907196e-01 -6.28978312e-01
4.55504566e-01 -8.57545957e-02 -8.56410503e-01 1.71027854e-01
-5.58351934e-01 -2.90663183e-01 1.31124277e-02 -1.99849442e-01
-9.00942564e-01 9.08877030e-02 1.38131693e-01 1.28375590e-01
-6.56872153e-01 2.00478844e-02 -5.22065580e-01 -6.05104864e-01
6.33951843e-01 2.41050318e-01 -8.03342611e-02 1.18751200e-02
4.82936800e-02 1.34608364e+00 7.44344652e-01 -9.06860009e-02
1.36764920e+00 5.93796134e-01 1.41826291e-02 -1.45188677e+00
-5.43397725e-01 -6.47057176e-01 -1.56230301e-01 -4.66813922e-01
8.82720947e-01 -1.05463183e+00 -9.66328263e-01 3.68150175e-01
-1.00120521e+00 1.80910200e-01 1.25907078e-01 8.48270595e-01
-9.02937576e-02 -6.54868502e-03 1.68223709e-01 -1.41498220e+00
-4.69207764e-01 -5.73276758e-01 8.60254824e-01 4.36001539e-01
-7.95351341e-02 -4.22763258e-01 -1.11418255e-01 8.39376152e-01
8.83060813e-01 1.90178365e-01 3.71298909e-01 -3.97547334e-01
-1.30075121e+00 -5.61944604e-01 -4.71431613e-01 -1.79232195e-01
2.77648896e-01 -5.71994185e-01 -1.17614222e+00 -3.34643543e-01
2.58259267e-01 1.06058970e-01 8.50006640e-01 6.84088409e-01
7.69557118e-01 -1.41658232e-01 -6.62159681e-01 1.01614726e+00
1.41165400e+00 7.36507058e-01 7.86758780e-01 3.57466906e-01
2.54161149e-01 4.28137392e-01 6.56128287e-01 3.46394151e-01
3.39479521e-02 3.83426726e-01 5.01624048e-01 1.01791322e-02
1.86878890e-01 4.10993606e-01 3.49153757e-01 1.05465598e-01
-2.46205539e-01 -2.10971430e-01 -6.20731115e-01 2.69085169e-01
-1.49815226e+00 -1.40551865e+00 -2.48970389e-01 1.94447374e+00
1.74389705e-01 1.73623949e-01 -2.91883916e-01 -1.36545211e-01
3.85892868e-01 2.25270525e-01 -4.53333706e-01 -1.06553972e-01
-3.34104081e-03 3.01648587e-01 1.03691161e+00 1.64970696e-01
-1.22442973e+00 3.75016302e-01 6.54982948e+00 4.06772286e-01
-1.06953144e+00 8.94633755e-02 1.60428822e-01 -3.05885762e-01
-2.32739672e-01 -4.35295969e-01 -1.28276503e+00 4.12718177e-01
8.55703652e-01 3.47081535e-02 3.44447084e-02 7.65426993e-01
1.94646046e-01 -4.39617217e-01 -8.90639424e-01 1.59239924e+00
2.56470323e-01 -1.30961013e+00 -6.89935386e-01 2.65322894e-01
3.95849138e-01 1.78526536e-01 7.76280537e-02 3.86338592e-01
3.42913479e-01 -1.11759567e+00 1.77094281e-01 5.00821888e-01
4.73624080e-01 -7.48470128e-01 9.60069954e-01 2.03914851e-01
-1.08813107e+00 -1.05123781e-01 -6.00823879e-01 -2.54909515e-01
3.87370251e-02 9.41385567e-01 -7.32783616e-01 3.64838392e-01
8.15144658e-01 2.62149066e-01 -4.29653287e-01 1.03407288e+00
7.58433938e-02 4.40518618e-01 -4.37819123e-01 -5.51757336e-01
4.77445349e-02 -3.38675886e-01 6.34155571e-01 1.16461480e+00
5.76565504e-01 4.11652952e-01 1.37596682e-01 5.99619091e-01
3.46072584e-01 -5.25452435e-01 -1.12509942e+00 1.81899313e-02
5.86627901e-01 1.44093621e+00 -4.72987026e-01 2.76350230e-01
-4.95965093e-01 3.07642698e-01 -5.75939178e-01 5.77906072e-01
-8.32831800e-01 -4.74143624e-01 1.26971602e+00 2.62121290e-01
2.68103063e-01 -7.69275367e-01 -3.82460564e-01 -5.22592545e-01
-4.74279560e-02 -4.76476252e-01 2.25758672e-01 -7.34447122e-01
-1.20135760e+00 4.87887800e-01 -2.33813941e-01 -1.32430553e+00
1.04451515e-01 -8.10363948e-01 -5.38698971e-01 6.92250013e-01
-1.74322653e+00 -1.18523276e+00 -8.43258917e-01 6.04944110e-01
2.24509716e-01 -8.86985481e-01 8.36635947e-01 3.70617419e-01
-4.60846901e-01 6.54757857e-01 -1.33108282e-02 1.08711138e-01
8.31453264e-01 -5.27651250e-01 9.13897008e-02 1.01903427e+00
2.74438620e-01 3.88080508e-01 7.86861181e-01 -6.61707222e-01
-2.20853734e+00 -1.31472707e+00 6.49275422e-01 -4.99296069e-01
3.80089104e-01 -6.41729772e-01 -1.50944203e-01 6.26521707e-01
1.64905608e-01 -3.96834426e-02 9.96349573e-01 1.87082604e-01
-2.14604944e-01 -7.64929950e-01 -1.43351364e+00 4.34974939e-01
1.07649362e+00 -3.10015291e-01 -4.90852624e-01 4.08659935e-01
3.50193590e-01 7.14370832e-02 -2.61592299e-01 3.84810060e-01
6.68813288e-01 -9.27472532e-01 1.02387774e+00 -1.73057735e-01
-2.66858190e-01 -4.21749741e-01 -3.94890338e-01 -1.04938793e+00
-3.44481051e-01 -7.63212025e-01 9.94712636e-02 1.11250246e+00
2.19597265e-01 -1.15433002e+00 1.19377303e+00 7.65287936e-01
1.17290229e-01 -1.33808449e-01 -9.89542663e-01 -1.20329285e+00
-8.68909657e-01 -8.77297282e-01 6.67595148e-01 6.83302999e-01
-4.00128931e-01 1.41802728e-01 -6.06733143e-01 8.25759232e-01
1.20772696e+00 4.99014586e-01 1.15737510e+00 -1.37525523e+00
4.83653601e-03 1.17033288e-01 -8.37222695e-01 -6.55481935e-01
-4.82182875e-02 -4.94833082e-01 2.43463293e-01 -1.56687391e+00
-4.33683768e-02 -5.26404083e-01 2.49581761e-03 4.39733416e-01
3.40744734e-01 7.67481148e-01 4.21069488e-02 -9.24814120e-02
-4.23549712e-01 2.07901999e-01 8.31798136e-01 -6.37790680e-01
5.57399765e-02 3.65502149e-01 -7.18984187e-01 5.32791018e-01
9.72450137e-01 -3.94599259e-01 -2.74925351e-01 -4.50963885e-01
-6.96653873e-02 -2.58132249e-01 5.14649272e-01 -1.52455461e+00
8.18117142e-01 -4.66011465e-01 6.82386518e-01 -1.05452871e+00
6.34268224e-01 -1.41127241e+00 5.16604781e-01 6.28443956e-01
5.66826999e-01 -4.59154099e-01 4.43198383e-02 5.55663347e-01
1.48413315e-01 -2.02000484e-01 7.84918249e-01 7.28246644e-02
-1.13731182e+00 5.68155468e-01 -6.65094793e-01 -3.15957248e-01
1.57122087e+00 -8.21304739e-01 -6.61190510e-01 -5.96912742e-01
-1.23598635e-01 2.55307019e-01 4.14751470e-03 4.03962731e-01
7.66300499e-01 -1.30742621e+00 -7.67819166e-01 8.09557438e-01
4.35992211e-01 -6.32183552e-01 1.85468242e-01 5.60494959e-01
-2.72441268e-01 6.02795660e-01 -4.12875563e-01 -5.48817337e-01
-1.64995348e+00 2.35707179e-01 5.37146367e-02 2.75257111e-01
-6.40133739e-01 7.26135254e-01 -1.33872882e-01 -1.90534189e-01
4.15973544e-01 -1.69459492e-01 -1.33450667e-03 -1.09102979e-01
9.13105607e-01 2.74867505e-01 1.07175842e-01 -7.64419213e-02
-8.20658088e-01 7.20651329e-01 1.50798723e-01 1.37356386e-01
1.39502275e+00 -2.01251388e-01 3.29611838e-01 -1.02953523e-01
7.74445474e-01 1.86844561e-02 -9.00724232e-01 4.48352210e-02
-2.05387875e-01 -6.12612307e-01 2.08734199e-01 -8.13769519e-01
-1.22043502e+00 5.73383629e-01 1.18948436e+00 2.59503424e-01
1.24541569e+00 -2.25345299e-01 8.04918349e-01 8.57440174e-01
8.88559401e-01 -1.12289417e+00 -3.59545529e-01 2.34187812e-01
4.19319957e-01 -1.38567901e+00 2.86818892e-01 -2.77174294e-01
-1.11125208e-01 8.84236932e-01 5.28933227e-01 2.91528016e-01
9.78644192e-01 1.06621325e+00 5.55353105e-01 -4.22610611e-01
-5.64659715e-01 -3.57106328e-01 -2.07355451e-02 1.47075188e+00
-7.67501891e-02 9.68280733e-02 1.70013364e-02 6.15174174e-01
-3.98360819e-01 -5.20848706e-02 2.63060778e-01 9.62702215e-01
-8.63494277e-01 -7.07217097e-01 -7.56614327e-01 8.18884194e-01
-9.23784897e-02 2.84361750e-01 -2.15008527e-01 6.73575044e-01
2.99504519e-01 1.29106855e+00 -6.15272578e-03 -3.55403811e-01
3.39057237e-01 9.17406753e-03 3.45897794e-01 -1.41735017e-01
-1.11066543e-01 -5.63705623e-01 2.22225085e-01 -4.67082202e-01
-6.74512327e-01 -5.51683247e-01 -6.93891585e-01 -3.28480780e-01
-2.35306531e-01 9.11945768e-04 1.10542214e+00 1.03118360e+00
3.27441812e-01 4.11075473e-01 5.93237102e-01 -3.59334767e-01
-2.85757244e-01 -7.26696908e-01 -8.68520975e-01 1.89562272e-02
3.10362190e-01 -5.63550651e-01 -3.67100865e-01 -6.61575701e-03] | [7.800446510314941, -1.2803525924682617] |
656e3edb-2a2e-4a87-80b5-753bcc8531c1 | multi-class-graph-clustering-via-approximated | 2306.08617 | null | https://arxiv.org/abs/2306.08617v1 | https://arxiv.org/pdf/2306.08617v1.pdf | Multi-class Graph Clustering via Approximated Effective $p$-Resistance | This paper develops an approximation to the (effective) $p$-resistance and applies it to multi-class clustering. Spectral methods based on the graph Laplacian and its generalization to the graph $p$-Laplacian have been a backbone of non-euclidean clustering techniques. The advantage of the $p$-Laplacian is that the parameter $p$ induces a controllable bias on cluster structure. The drawback of $p$-Laplacian eigenvector based methods is that the third and higher eigenvectors are difficult to compute. Thus, instead, we are motivated to use the $p$-resistance induced by the $p$-Laplacian for clustering. For $p$-resistance, small $p$ biases towards clusters with high internal connectivity while large $p$ biases towards clusters of small ``extent,'' that is a preference for smaller shortest-path distances between vertices in the cluster. However, the $p$-resistance is expensive to compute. We overcome this by developing an approximation to the $p$-resistance. We prove upper and lower bounds on this approximation and observe that it is exact when the graph is a tree. We also provide theoretical justification for the use of $p$-resistance for clustering. Finally, we provide experiments comparing our approximated $p$-resistance clustering to other $p$-Laplacian based methods. | ['Mark Herbster', 'Shota Saito'] | 2023-06-14 | null | null | null | null | ['graph-clustering', 'clustering'] | ['graphs', 'methodology'] | [-1.54930770e-01 2.54334003e-01 -1.80263877e-01 -1.87343992e-02
-5.21185875e-01 -6.43882751e-01 -1.49778128e-01 2.56342083e-01
-2.91820586e-01 1.88613862e-01 -2.73207545e-01 -3.95737320e-01
-5.75379014e-01 -1.09542310e+00 -6.04463696e-01 -8.20164084e-01
-8.90131116e-01 3.62299591e-01 2.59867996e-01 -5.31007573e-02
3.47368002e-01 6.07960284e-01 -1.07997012e+00 -2.39036933e-01
6.58714116e-01 6.58271790e-01 -1.51830226e-01 5.91224313e-01
-1.30826429e-01 -2.13329401e-02 -2.18265131e-01 -3.02525431e-01
5.48016965e-01 -5.65165937e-01 -9.20182824e-01 2.01392099e-01
5.36868684e-02 5.50611317e-01 -2.39036247e-01 1.20870852e+00
3.16982865e-01 1.82635710e-01 1.03317523e+00 -1.61189973e+00
-3.55839252e-01 7.63717651e-01 -1.30365670e+00 1.41892403e-01
3.12718719e-01 -2.25047961e-01 1.23930347e+00 -6.89576149e-01
6.44640505e-01 1.11173880e+00 1.09991241e+00 2.37837136e-01
-1.52451921e+00 -4.77374673e-01 1.55908689e-01 -6.10469989e-02
-2.11360216e+00 6.55491725e-02 1.06628907e+00 -4.53793794e-01
8.51861894e-01 3.36762935e-01 4.49152499e-01 1.36609236e-02
-2.48602659e-01 3.69768739e-01 8.80529106e-01 -5.16926110e-01
2.31020302e-01 1.56780168e-01 1.00398019e-01 9.73861754e-01
4.11545247e-01 -3.21962506e-01 -9.75376144e-02 -2.57120073e-01
7.09230840e-01 -2.01012090e-01 -2.41865188e-01 -7.87578464e-01
-8.30882728e-01 8.07724178e-01 8.05895627e-01 5.88794649e-01
-3.72440042e-03 6.62927568e-01 4.39678580e-02 1.79573685e-01
1.67602882e-01 3.48788619e-01 -2.49721289e-01 -4.57197055e-02
-7.88740695e-01 -2.13213146e-01 6.90541804e-01 1.16500783e+00
1.26948619e+00 -4.26877528e-01 5.33771396e-01 8.48645926e-01
2.98064351e-01 5.17116070e-01 -2.22877845e-01 -1.25856125e+00
3.96842420e-01 9.47399437e-01 -1.24435171e-01 -1.63480055e+00
-6.08977258e-01 -1.09747507e-01 -1.06201041e+00 3.66347260e-03
5.56626737e-01 -1.19357578e-01 -3.81317973e-01 1.96267533e+00
1.38102964e-01 -1.32324904e-01 -3.73169631e-01 5.02965987e-01
-3.64351179e-03 4.78875607e-01 -1.84414551e-01 -6.65648758e-01
1.00410891e+00 -6.09659851e-01 -2.22134620e-01 1.73161641e-01
9.83472943e-01 -7.24130392e-01 1.06741726e+00 8.67695287e-02
-1.10615849e+00 -8.64506587e-02 -7.48046279e-01 5.20147622e-01
-4.00101453e-01 -2.22123161e-01 4.39043939e-01 9.79996443e-01
-1.61826801e+00 6.49804652e-01 -7.08501279e-01 -7.70791292e-01
1.77776188e-01 5.52730322e-01 -9.39994976e-02 -1.41968392e-02
-6.42669678e-01 5.21582127e-01 6.68098405e-02 -2.74268180e-01
5.80660673e-03 -4.91064250e-01 -4.56043184e-01 6.92656869e-03
2.41761848e-01 -2.33926937e-01 3.85241687e-01 -7.43782341e-01
-9.58565652e-01 8.29028547e-01 -3.03031266e-01 7.99926296e-02
4.30091679e-01 5.45539081e-01 -1.09503441e-01 5.25219202e-01
1.72662050e-01 5.70799589e-01 4.97793883e-01 -1.51944506e+00
-2.94811487e-01 -3.38907421e-01 -1.79774255e-01 2.19213322e-01
-3.67005259e-01 -8.88864845e-02 -6.12615526e-01 -2.84878731e-01
5.61686456e-01 -1.28062010e+00 -5.81827044e-01 -1.10276513e-01
-5.95295072e-01 -3.98633838e-01 7.41578758e-01 7.46605322e-02
1.52108252e+00 -2.23000455e+00 1.65907741e-01 1.21160078e+00
5.42851806e-01 -1.53755248e-01 -7.77345076e-02 8.91953170e-01
-1.82739720e-01 7.33881414e-01 -4.23361897e-01 1.42840460e-01
7.26810396e-02 3.23426537e-02 2.13873565e-01 8.02335799e-01
-1.76595822e-01 5.17853737e-01 -1.04907441e+00 -6.27648532e-01
6.66500255e-02 3.88254285e-01 -7.75487185e-01 -4.25728381e-01
3.08566570e-01 -2.14463949e-01 -3.54540855e-01 3.75859827e-01
8.05237949e-01 -5.65812886e-01 5.35801411e-01 -1.94975495e-01
-2.35930681e-01 -3.05233419e-01 -1.49912536e+00 1.31924927e+00
-1.24318868e-01 4.11305934e-01 4.59988624e-01 -1.31469214e+00
7.36154556e-01 1.73361585e-01 1.10610497e+00 -4.40197811e-02
-5.87354461e-03 3.81232172e-01 1.47659448e-03 -9.97428894e-02
1.69860184e-01 -2.37244084e-01 -8.38890821e-02 8.91519964e-01
-4.38647926e-01 -2.82260925e-01 5.91057360e-01 7.00393736e-01
1.53892899e+00 -3.57811511e-01 -1.21796876e-01 -9.98210251e-01
4.21540201e-01 -4.65496331e-02 3.87642115e-01 5.40985346e-01
-4.44819689e-01 3.06464553e-01 7.52979994e-01 7.80814290e-02
-7.69994259e-01 -1.25526655e+00 -8.13311860e-02 9.24911559e-01
4.21399206e-01 -6.11947775e-01 -1.00730598e+00 -5.06257474e-01
-9.09165889e-02 2.02847406e-01 -5.81383049e-01 -6.23062104e-02
-6.30808294e-01 -1.14889550e+00 4.49390560e-01 4.52831447e-01
2.98979431e-01 -5.58892727e-01 -4.91453946e-01 5.64083047e-02
-3.10085922e-01 -7.52959430e-01 -8.07164371e-01 2.05883950e-01
-9.80056465e-01 -1.44096339e+00 -5.96450806e-01 -1.01437461e+00
1.30673897e+00 5.26877582e-01 9.94297624e-01 2.12889731e-01
-3.99022281e-01 8.60239029e-01 -4.12227452e-01 1.91469163e-01
5.47031127e-02 1.52074695e-01 1.11213498e-01 -1.62776589e-01
3.89582992e-01 -7.57866800e-01 -9.59536433e-01 6.13683283e-01
-7.41972625e-01 -5.38144529e-01 4.65748549e-01 3.21247488e-01
7.69256949e-01 6.76054776e-01 2.88438410e-01 -7.74671197e-01
6.36161804e-01 -3.79145682e-01 -4.11874235e-01 9.30502042e-02
-7.71670401e-01 4.13249992e-02 6.56088650e-01 -2.80428410e-01
-2.78314739e-01 2.33354256e-01 -5.05806366e-03 -3.03701967e-01
2.51406014e-01 4.50318962e-01 4.27792482e-02 -3.43198180e-01
6.73151195e-01 -1.49130613e-01 -2.45970547e-01 -2.85909027e-01
7.74940252e-01 3.46324205e-01 1.00619599e-01 -7.15078354e-01
1.02353024e+00 7.92866588e-01 4.07899022e-01 -1.06817448e+00
-1.81499019e-01 -6.66094959e-01 -6.54370189e-01 -2.65328526e-01
6.15898907e-01 -4.13852543e-01 -1.10231793e+00 -9.23943445e-02
-7.28446066e-01 -3.88974905e-01 -3.58712167e-01 2.29578599e-01
-7.83192098e-01 6.65225506e-01 -5.10650277e-01 -9.36597884e-01
7.66682252e-02 -9.96358991e-01 6.51643753e-01 -1.32617936e-01
-2.32146770e-01 -1.17674255e+00 2.07782298e-01 -1.81138530e-01
2.86872447e-01 3.94872904e-01 1.26141965e+00 -9.18145776e-02
-3.11144680e-01 -2.63433129e-01 -5.96276700e-01 -9.09542516e-02
8.84812474e-02 2.72687167e-01 -3.02309722e-01 -5.81706166e-01
-3.49689960e-01 1.33687243e-01 5.07187009e-01 5.42958856e-01
1.02853251e+00 -3.86695206e-01 -7.80552983e-01 4.16319489e-01
1.81296325e+00 1.93892270e-02 4.23151553e-01 -6.30727336e-02
6.28829420e-01 5.96032381e-01 1.02544591e-01 2.56166548e-01
4.16057080e-01 4.95093584e-01 2.23592460e-01 -1.07003905e-01
4.83386219e-02 -1.08462535e-01 2.26723999e-01 1.30536723e+00
-3.08687627e-01 -9.09226388e-03 -1.18986070e+00 6.91832423e-01
-1.57585430e+00 -7.91516960e-01 -4.19935375e-01 2.24111700e+00
7.29792953e-01 1.21379949e-01 5.64543605e-01 4.36980158e-01
1.04893160e+00 -3.79309207e-02 -2.95079976e-01 -4.26737964e-01
3.70656587e-02 3.16441327e-01 8.71876299e-01 7.94255972e-01
-7.92854488e-01 7.13908136e-01 7.15577602e+00 9.60185766e-01
-8.08401704e-01 -1.01558469e-01 4.93185550e-01 -1.66841671e-01
-4.15248722e-01 7.28318617e-02 -3.63195032e-01 4.10890698e-01
7.76397705e-01 -2.37672120e-01 4.39715415e-01 8.33704710e-01
1.93782881e-01 -6.08794913e-02 -9.37465012e-01 1.19464779e+00
-2.30134904e-01 -1.00711536e+00 -3.06625515e-01 3.81762981e-01
6.23927355e-01 7.28813931e-02 3.53127159e-03 -2.41073266e-01
6.75354958e-01 -9.44974184e-01 1.42149791e-01 -1.86130136e-01
1.13521945e+00 -1.10823059e+00 1.02993034e-01 8.52834880e-02
-1.95667326e+00 8.90559107e-02 -5.29833853e-01 1.35154754e-01
1.21544249e-01 8.53454947e-01 -3.07877958e-01 3.09388965e-01
8.64398301e-01 5.12133062e-01 -3.38966399e-01 6.59572542e-01
1.42911822e-01 2.59094477e-01 -6.88970983e-01 -7.62458667e-02
3.04264396e-01 -7.86496401e-01 3.52524042e-01 1.37298858e+00
6.16670430e-01 3.82739246e-01 2.33632118e-01 6.94408417e-01
-1.88523188e-01 3.84546727e-01 -7.63194263e-01 -1.61722943e-01
6.91488743e-01 1.23176527e+00 -1.50230241e+00 -1.05372004e-01
-3.11702400e-01 8.31848502e-01 3.71930093e-01 3.97086799e-01
-6.31582499e-01 -9.95213628e-01 6.34877264e-01 4.28217620e-01
3.66155982e-01 -5.48370004e-01 -1.90244704e-01 -8.23915720e-01
-2.24347815e-01 -5.78026235e-01 3.63072246e-01 -3.28986585e-01
-1.27462828e+00 4.57715243e-01 -1.48108276e-02 -8.29188526e-01
2.83154964e-01 -6.25392318e-01 -4.49175805e-01 5.48203945e-01
-9.82783556e-01 -5.73658526e-01 1.02166243e-01 8.38972151e-01
-2.48325214e-01 5.64620793e-01 7.18299210e-01 3.60807896e-01
-5.77735424e-01 5.43297231e-01 2.87211299e-01 2.06404850e-01
4.60495293e-01 -1.41248715e+00 -1.81477964e-01 8.93017948e-01
1.12981007e-01 8.27254891e-01 8.24284077e-01 -3.23350400e-01
-1.55094957e+00 -9.19355154e-01 8.16762209e-01 -3.96877944e-01
7.80479908e-01 -1.72846526e-01 -5.33819854e-01 6.39172435e-01
2.18871050e-02 2.78290715e-02 1.04455912e+00 2.41693720e-01
-5.35688698e-01 -2.08501905e-01 -1.33913136e+00 7.20442176e-01
1.45789504e+00 -4.92558360e-01 1.84650883e-01 2.92921484e-01
2.89859176e-01 4.41806942e-01 -1.20069218e+00 2.41357461e-01
2.70061642e-01 -1.09053195e+00 1.02713811e+00 1.13851264e-01
-1.95004627e-01 -3.74057174e-01 -2.49801517e-01 -1.25447011e+00
-5.43044508e-01 -8.30979347e-01 4.27734971e-01 7.95861900e-01
5.96054018e-01 -5.93015194e-01 1.03676474e+00 4.24283892e-01
1.58400968e-01 -6.18702233e-01 -1.10887563e+00 -9.34248984e-01
6.26873374e-01 -3.89738202e-01 1.14496537e-01 1.26618993e+00
8.59613121e-01 1.93883508e-01 8.60744342e-02 5.22176065e-02
8.75808775e-01 1.39121905e-01 4.33467269e-01 -1.18850553e+00
-2.24161163e-01 -8.83624136e-01 -2.81975359e-01 -1.25279987e+00
-5.66343889e-02 -9.71765935e-01 -5.28624887e-03 -1.28057790e+00
2.13326335e-01 -1.05089951e+00 -1.82696372e-01 1.97062954e-01
3.80037129e-02 6.09089017e-01 9.51386392e-02 1.96738154e-01
-6.82568073e-01 -4.17595766e-02 9.43889081e-01 7.91777000e-02
-4.93826270e-01 -4.06777769e-01 -5.72440445e-01 8.11840951e-01
9.03531373e-01 -3.79686505e-01 -4.31310207e-01 -2.08297580e-01
7.18186975e-01 -9.63950604e-02 -2.35152245e-01 -8.16336393e-01
1.61254644e-01 -6.54034615e-02 -1.63757384e-01 -3.97830963e-01
1.59153312e-01 -9.11483824e-01 1.64682999e-01 7.88853705e-01
-9.10265744e-02 3.15844953e-01 -1.27470076e-01 6.38839781e-01
3.44742723e-02 1.13068581e-01 1.01762736e+00 -2.58707643e-01
-2.26028174e-01 3.66585284e-01 -4.44003195e-01 2.45434046e-01
1.27178597e+00 -4.10368741e-01 -2.09236845e-01 -3.82625818e-01
-7.76456237e-01 2.30624840e-01 8.10842872e-01 -2.53643453e-01
4.16190684e-01 -1.35644388e+00 -2.95634270e-01 -3.86574939e-02
1.92410238e-02 -2.78906852e-01 -7.36526847e-02 1.14419436e+00
-7.77703643e-01 2.32276127e-01 1.76172018e-01 -6.70054436e-01
-1.12788141e+00 8.77043486e-01 3.55093420e-01 -1.11780114e-01
-2.63593197e-01 9.22869742e-01 -7.69332871e-02 -2.99456447e-01
8.23117048e-02 -6.13891706e-02 3.72656435e-01 -2.53536906e-02
7.94267748e-03 8.17908227e-01 -3.56830597e-01 -8.42700481e-01
-7.47085929e-01 1.26482904e+00 3.60147595e-01 -4.07046467e-01
1.07543695e+00 -4.67489243e-01 -6.00405812e-01 3.26534659e-01
1.39016259e+00 4.02977020e-01 -8.12813103e-01 -2.55822577e-02
2.05154717e-01 -4.40593123e-01 -3.63339156e-01 -9.97004136e-02
-1.34502244e+00 8.86210203e-01 3.63650739e-01 8.49087119e-01
1.12281573e+00 3.26308072e-01 3.34056526e-01 3.31853062e-01
6.18699610e-01 -1.44195163e+00 1.84942618e-01 1.36505067e-01
3.47748458e-01 -8.13698292e-01 -4.02684249e-02 -9.24645424e-01
-1.96171865e-01 8.87835324e-01 3.31778318e-01 -3.18689615e-01
1.31338084e+00 1.28128871e-01 9.08319056e-02 -4.14588034e-01
-2.31291369e-01 -2.71519363e-01 -2.30198711e-01 5.67556858e-01
6.57066643e-01 2.20219135e-01 -7.01613903e-01 -1.04467042e-01
-1.26646653e-01 -4.97854799e-01 5.29199898e-01 6.61302805e-01
-6.02148116e-01 -1.41265905e+00 -4.38936025e-01 3.32147717e-01
-3.01118433e-01 1.10850530e-02 -5.50379694e-01 7.81425953e-01
1.72819376e-01 1.24536657e+00 1.55557245e-01 -4.01336044e-01
2.03966886e-01 6.49732351e-02 5.46576977e-01 -3.58864874e-01
-2.76299745e-01 2.80685067e-01 -3.40117097e-01 -6.61393762e-01
-7.59096622e-01 -4.89200115e-01 -1.44268572e+00 -8.48921239e-01
-3.49818975e-01 7.02404678e-01 5.56867778e-01 3.65220308e-01
4.93273944e-01 8.91935900e-02 8.67307544e-01 -4.88246650e-01
-1.19987957e-01 -5.41701972e-01 -1.07938921e+00 5.67514658e-01
-2.57690549e-01 -3.96316409e-01 -7.68328786e-01 -1.05957270e-01] | [7.082196235656738, 5.098977565765381] |
260e2854-edf8-4714-9b16-f28cbb4ef013 | variational-policy-gradient-method-for | 2007.02151 | null | https://arxiv.org/abs/2007.02151v1 | https://arxiv.org/pdf/2007.02151v1.pdf | Variational Policy Gradient Method for Reinforcement Learning with General Utilities | In recent years, reinforcement learning (RL) systems with general goals beyond a cumulative sum of rewards have gained traction, such as in constrained problems, exploration, and acting upon prior experiences. In this paper, we consider policy optimization in Markov Decision Problems, where the objective is a general concave utility function of the state-action occupancy measure, which subsumes several of the aforementioned examples as special cases. Such generality invalidates the Bellman equation. As this means that dynamic programming no longer works, we focus on direct policy search. Analogously to the Policy Gradient Theorem \cite{sutton2000policy} available for RL with cumulative rewards, we derive a new Variational Policy Gradient Theorem for RL with general utilities, which establishes that the parametrized policy gradient may be obtained as the solution of a stochastic saddle point problem involving the Fenchel dual of the utility function. We develop a variational Monte Carlo gradient estimation algorithm to compute the policy gradient based on sample paths. We prove that the variational policy gradient scheme converges globally to the optimal policy for the general objective, though the optimization problem is nonconvex. We also establish its rate of convergence of the order $O(1/t)$ by exploiting the hidden convexity of the problem, and proves that it converges exponentially when the problem admits hidden strong convexity. Our analysis applies to the standard RL problem with cumulative rewards as a special case, in which case our result improves the available convergence rate. | ['Mengdi Wang', 'Csaba Szepesvari', 'Amrit Singh Bedi', 'Junyu Zhang', 'Alec Koppel'] | 2020-07-04 | null | http://proceedings.neurips.cc/paper/2020/hash/30ee748d38e21392de740e2f9dc686b6-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/30ee748d38e21392de740e2f9dc686b6-Paper.pdf | neurips-2020-12 | ['variational-monte-carlo'] | ['miscellaneous'] | [-1.05659150e-01 4.55941170e-01 -6.08703494e-01 3.10052961e-01
-1.03944659e+00 -7.58472919e-01 2.83954948e-01 2.88144294e-02
-9.74985063e-01 1.39887595e+00 6.64212480e-02 -5.91986120e-01
-4.38898683e-01 -5.17184198e-01 -7.78177083e-01 -1.15899026e+00
-1.40401438e-01 4.27178741e-01 -2.47423500e-01 -1.93692312e-01
2.78385460e-01 3.51771325e-01 -1.15595448e+00 -6.74483597e-01
1.15688598e+00 1.05112851e+00 2.92932570e-01 6.60182416e-01
7.79372826e-02 6.59157455e-01 -5.59335649e-01 1.59048051e-01
4.37802047e-01 -4.68766689e-01 -8.13863933e-01 1.70219392e-01
-3.08762461e-01 -5.49612343e-01 -1.77068070e-01 1.32739806e+00
5.93852401e-01 6.23140633e-01 4.97867078e-01 -1.34036696e+00
-2.83290565e-01 4.01565433e-01 -6.00792229e-01 1.16570264e-01
5.92647381e-02 1.49739578e-01 1.13135910e+00 -3.89880627e-01
4.61031169e-01 1.13886821e+00 1.09780930e-01 6.77167118e-01
-1.35007191e+00 -1.70180351e-01 4.86792117e-01 -5.05920388e-02
-9.13475096e-01 -1.05126850e-01 3.43853891e-01 -2.56901920e-01
7.74962246e-01 2.44492944e-02 7.12122023e-01 9.63076591e-01
1.25286162e-01 1.13316250e+00 1.38763642e+00 -3.09776813e-01
7.74817705e-01 7.32331490e-03 -1.89780384e-01 8.23437393e-01
2.44583070e-01 4.31653440e-01 -8.20592567e-02 -3.03903252e-01
8.62679660e-01 -1.12648373e-02 -2.82629609e-01 -5.99247098e-01
-9.87654507e-01 1.12371707e+00 3.50622460e-02 -1.00277796e-01
-6.02666378e-01 4.77996945e-01 1.07930273e-01 4.26064581e-01
3.97835761e-01 4.48097140e-01 -3.27185929e-01 -4.16130722e-01
-7.21967757e-01 6.53512657e-01 1.03330564e+00 8.45451832e-01
5.28557599e-01 5.09340405e-01 -3.71250600e-01 3.68815273e-01
3.18625838e-01 9.95761752e-01 -9.96636320e-03 -1.53611803e+00
5.57797015e-01 -1.93908811e-02 9.84718025e-01 -2.06204757e-01
-2.90985972e-01 -5.42287469e-01 -4.72626805e-01 5.33393562e-01
8.65605056e-01 -7.98145950e-01 -5.15400589e-01 2.11879420e+00
2.83196211e-01 -1.89454392e-01 1.49079040e-01 8.53725433e-01
-4.59730148e-01 7.66394615e-01 -1.78154826e-01 -9.56486523e-01
7.08891749e-01 -4.94464040e-01 -8.07651401e-01 3.07920836e-02
4.05833215e-01 -3.66290897e-01 1.07607257e+00 4.85235125e-01
-1.25142801e+00 1.33831888e-01 -9.14371848e-01 5.21923184e-01
3.76216061e-02 -1.20040417e-01 3.60777050e-01 5.54270864e-01
-1.09647596e+00 8.53613555e-01 -8.20827186e-01 -1.40686274e-01
2.01319516e-01 3.15085918e-01 4.36550826e-01 1.78890988e-01
-1.02780628e+00 1.06735039e+00 3.00825417e-01 1.31325983e-02
-1.38123727e+00 -4.93788183e-01 -5.51109672e-01 5.00385351e-02
1.09156597e+00 -4.60655928e-01 1.65006328e+00 -7.70579457e-01
-2.15675616e+00 2.39086881e-01 5.78078581e-03 -3.94692987e-01
1.04119456e+00 -1.83731750e-01 2.02867270e-01 9.25508291e-02
2.04246551e-01 5.81562631e-02 8.63090098e-01 -1.16224468e+00
-6.82224929e-01 -3.96385461e-01 5.90602219e-01 5.45049608e-01
1.21949669e-02 -2.72610217e-01 9.70270783e-02 -1.81719601e-01
-4.73853767e-01 -1.02915370e+00 -6.89631045e-01 -3.07173997e-01
-2.37510756e-01 -2.36286893e-01 4.57539827e-01 -3.01297098e-01
1.06270528e+00 -1.63703358e+00 3.67764682e-01 3.86470139e-01
-1.64783046e-01 -1.33767888e-01 2.10530668e-01 4.91969138e-01
4.12513793e-01 1.05760567e-01 -4.09717202e-01 -1.63243577e-01
5.16648471e-01 5.01843989e-01 -5.93583465e-01 8.97498250e-01
-2.53352076e-01 8.07249546e-01 -1.25375509e+00 -3.44132811e-01
1.24626197e-01 -1.49537176e-01 -7.96456456e-01 4.40293476e-02
-6.18763149e-01 5.73781312e-01 -1.01782012e+00 2.41881028e-01
2.92827457e-01 -1.91741839e-01 4.10313457e-01 5.48705220e-01
-4.31260318e-01 -2.01385230e-01 -1.36285353e+00 1.55343676e+00
-6.70010030e-01 1.50177926e-01 6.29653931e-01 -1.36207485e+00
3.25830340e-01 2.88215578e-01 9.95680094e-01 -2.29076102e-01
2.50787973e-01 1.99588656e-01 -3.58143151e-01 -4.58285928e-01
5.29108346e-01 -4.40462232e-01 -2.27701858e-01 5.61416805e-01
-1.12469040e-01 -1.29699245e-01 1.34966388e-01 1.39035061e-01
8.77684295e-01 3.99998516e-01 3.90028298e-01 -5.48695505e-01
1.91341177e-01 6.38229493e-03 4.37257290e-01 9.78042424e-01
-3.68334681e-01 -2.13120490e-01 1.01150656e+00 2.87906408e-01
-8.94609392e-01 -1.15152860e+00 -1.35245118e-02 1.00156724e+00
1.48218349e-01 2.31180154e-02 -6.35757148e-01 -6.80377364e-01
2.38273039e-01 8.36245000e-01 -6.27228379e-01 1.15497336e-01
-3.84389639e-01 -9.15946722e-01 3.92675251e-02 1.45312533e-01
5.45081615e-01 -8.30144644e-01 -8.64139855e-01 4.15127963e-01
-5.03145494e-02 -7.90837348e-01 -4.88133967e-01 1.95423409e-01
-8.80943358e-01 -9.87947047e-01 -1.13668096e+00 -1.40019536e-01
5.06287515e-01 -2.13261142e-01 6.32235706e-01 -6.05669200e-01
2.36077651e-01 9.76477444e-01 7.97445551e-02 -4.27305698e-01
-2.36389637e-01 -1.30707905e-01 3.15842748e-01 1.65151060e-02
-3.73593330e-01 -3.97968471e-01 -6.97640955e-01 2.14910433e-01
-6.15465164e-01 -3.97093087e-01 2.19942346e-01 8.87616932e-01
7.69320786e-01 -3.83759700e-02 7.87472427e-01 -5.38905025e-01
1.00441301e+00 -5.34216821e-01 -1.33440912e+00 1.74859419e-01
-8.69398832e-01 5.58167338e-01 7.47310281e-01 -3.51008683e-01
-1.11314213e+00 -3.94949168e-02 2.59963512e-01 -4.64053124e-01
4.66258407e-01 4.80901212e-01 3.55025753e-02 7.58676007e-02
4.01343793e-01 2.57393450e-01 2.00718924e-01 -3.58021289e-01
4.92424399e-01 3.09480041e-01 1.70674533e-01 -1.12348628e+00
3.53667200e-01 5.81427991e-01 3.03463966e-01 -6.92742288e-01
-8.98017883e-01 -1.68386579e-01 1.17748305e-02 -4.49006975e-01
7.29801714e-01 -4.58408207e-01 -1.63530576e+00 -1.09536171e-01
-7.86972106e-01 -8.92974734e-01 -9.11230206e-01 6.49868548e-01
-1.40668094e+00 3.25078338e-01 -4.44161564e-01 -1.69145834e+00
1.52152181e-01 -1.06865132e+00 5.60456038e-01 1.94802746e-01
4.53449011e-01 -1.09689534e+00 2.36744583e-01 -8.79225805e-02
3.71116668e-01 4.78345186e-01 5.54695129e-01 5.45480028e-02
-5.60132384e-01 1.85387671e-01 2.67601371e-01 3.12733084e-01
-2.71782309e-01 -3.48588139e-01 -4.84079659e-01 -5.70187509e-01
2.84084558e-01 -3.04816365e-01 7.35414505e-01 8.71751368e-01
1.00633907e+00 -6.42924309e-01 -1.69802263e-01 3.84222418e-01
1.66827583e+00 4.25665110e-01 9.29480344e-02 2.04706684e-01
2.88758993e-01 4.50189829e-01 8.35811853e-01 1.05856407e+00
1.87473759e-01 4.24268872e-01 7.12729096e-01 3.33794892e-01
7.68594921e-01 -3.44298780e-01 8.06119561e-01 2.39699513e-01
-4.06766683e-01 -1.02888934e-01 -5.00402272e-01 4.31410253e-01
-2.31462550e+00 -1.10318971e+00 4.36455637e-01 2.70202041e+00
8.48448634e-01 -7.87031204e-02 4.41717446e-01 -2.43602738e-01
5.66531539e-01 6.97139725e-02 -1.22837126e+00 -6.25266552e-01
6.18735328e-02 2.86288410e-01 1.25398624e+00 8.67188096e-01
-7.50163198e-01 7.90980935e-01 6.78346634e+00 8.88071597e-01
-8.08619618e-01 3.86843950e-01 4.55170006e-01 -4.57605600e-01
-2.26681322e-01 -3.82267907e-02 -7.09428191e-01 5.10460496e-01
8.00074160e-01 -6.28685832e-01 1.01300430e+00 9.37300622e-01
8.43992889e-01 -4.60834682e-01 -7.04957426e-01 6.99879587e-01
-6.19902730e-01 -9.48957384e-01 -6.77646637e-01 7.04460204e-01
9.65033889e-01 -1.53061692e-02 3.03128928e-01 5.02978265e-01
1.03019369e+00 -8.53142738e-01 6.84440613e-01 4.38103378e-01
7.51198411e-01 -1.09948421e+00 2.81071573e-01 5.42298853e-01
-9.28026378e-01 -5.40376663e-01 -3.44149321e-01 -1.83283493e-01
4.35245484e-01 3.54227096e-01 -4.45443183e-01 4.60149586e-01
9.89013389e-02 3.85497302e-01 5.10101318e-01 9.00619268e-01
-3.24218541e-01 4.37830687e-01 -5.70528328e-01 -4.01302159e-01
6.79006636e-01 -6.83319747e-01 7.63235390e-01 6.53087795e-01
3.65571231e-01 1.83335423e-01 5.65966964e-01 9.35424924e-01
8.47241282e-02 1.46690518e-01 -4.72027391e-01 -7.05425218e-02
1.29337266e-01 9.90237653e-01 -6.10005558e-01 -2.00938523e-01
-1.56278983e-01 8.03764760e-01 3.98768455e-01 8.15968275e-01
-8.36470783e-01 -6.19614348e-02 8.31235230e-01 -3.08323622e-01
3.87221277e-01 -4.15329516e-01 9.12640691e-02 -1.14237463e+00
7.19677582e-02 -4.34044540e-01 4.26012367e-01 -1.25660598e-01
-8.84448409e-01 6.09974097e-03 3.35834175e-01 -9.45022821e-01
-6.80323064e-01 -4.90043670e-01 -3.87718320e-01 8.30841243e-01
-1.46609998e+00 -2.88847178e-01 6.18996918e-01 7.99586296e-01
2.49081925e-01 6.10385649e-02 5.10531545e-01 -1.72852367e-01
-7.45314598e-01 2.31228918e-01 7.85692453e-01 -4.67119902e-01
1.68307528e-01 -1.70051622e+00 -5.58520496e-01 7.87834704e-01
-4.87893879e-01 2.37929955e-01 1.02307153e+00 -4.77955371e-01
-1.60485971e+00 -7.66392708e-01 9.85596031e-02 -5.37285954e-02
9.88471150e-01 -1.66111309e-02 -2.54509360e-01 6.52994394e-01
8.70739594e-02 -1.26784131e-01 6.93857148e-02 -1.07097238e-01
3.69568855e-01 1.02266282e-01 -1.17816567e+00 7.13031173e-01
9.32541072e-01 -2.81190395e-01 -3.83456424e-02 4.61062491e-01
4.69357699e-01 -4.86675411e-01 -8.30984950e-01 -6.52137920e-02
3.23719829e-01 -5.24571598e-01 6.53174341e-01 -8.74239266e-01
3.65801007e-02 -1.30759493e-01 -2.43024692e-01 -1.54818952e+00
1.39587432e-01 -1.46885121e+00 -5.09213269e-01 5.54442108e-01
3.41687888e-01 -8.51599276e-01 8.23002398e-01 5.49006402e-01
-6.88949078e-02 -1.02463853e+00 -1.25339496e+00 -1.15473545e+00
4.74617869e-01 -3.38904977e-01 1.19056791e-01 6.55115664e-01
3.87386411e-01 -5.03485799e-02 -4.90748912e-01 -8.52539856e-03
9.98373091e-01 1.99311376e-01 2.96329081e-01 -5.87171972e-01
-7.69155979e-01 -6.45319104e-01 4.44442570e-01 -1.52783072e+00
4.14457291e-01 -7.96176493e-01 2.86005229e-01 -1.50964081e+00
2.24249978e-02 -5.14265895e-01 -3.56625855e-01 2.22605392e-01
-8.08425695e-02 -6.54514968e-01 4.90739137e-01 -6.05452880e-02
-6.27363384e-01 1.00663114e+00 1.72892773e+00 6.82515150e-04
-5.06664217e-01 5.98932803e-01 -6.05613887e-01 6.41954482e-01
9.91123319e-01 -3.89315695e-01 -6.88658714e-01 7.10626924e-03
2.84174800e-01 6.79251909e-01 1.96177125e-01 -4.93154556e-01
1.86396483e-02 -8.07614446e-01 -2.58870631e-01 -2.30499327e-01
4.01939631e-01 -5.56093276e-01 -2.68748671e-01 7.25152791e-01
-5.00918746e-01 -1.09127788e-02 -1.77072957e-01 9.85925674e-01
3.14537257e-01 -4.85466093e-01 8.47635806e-01 -2.61547655e-01
-3.42203289e-01 5.98031402e-01 -7.56889522e-01 5.51408112e-01
9.39665794e-01 1.28560066e-01 -2.68114693e-02 -8.73941541e-01
-1.14337277e+00 6.56761348e-01 2.32953325e-01 -3.14658195e-01
4.37725544e-01 -1.05856121e+00 -4.58025455e-01 -5.38379848e-01
-5.16475081e-01 -2.39227697e-01 9.77279842e-02 1.09595633e+00
1.38813913e-01 4.23719823e-01 5.73397242e-02 -2.86604881e-01
-5.43485641e-01 5.73558450e-01 6.60918117e-01 -6.57516956e-01
-2.76635200e-01 4.43607837e-01 -4.07983288e-02 -4.45872433e-02
2.33374089e-01 -4.62197751e-01 7.89419413e-02 1.15846373e-01
1.90888733e-01 6.55063927e-01 -5.64465106e-01 -9.81061161e-02
4.49515171e-02 2.04613522e-01 3.27691942e-01 -9.44431186e-01
1.14630926e+00 -3.71921718e-01 2.64871210e-01 4.30671513e-01
9.83636439e-01 -1.34655163e-01 -1.87639463e+00 -3.21737558e-01
-2.86189858e-02 -2.17082232e-01 1.09280579e-01 -6.17175996e-01
-1.03875196e+00 5.90444088e-01 4.65922087e-01 3.59540284e-01
8.14611137e-01 -1.33284464e-01 3.58971089e-01 5.58926761e-01
7.41429567e-01 -1.60788381e+00 1.44213084e-02 7.22693920e-01
7.81383336e-01 -1.06011915e+00 -1.15252351e-02 2.76976019e-01
-7.54601002e-01 9.19144273e-01 2.26289466e-01 -4.27380830e-01
6.54524624e-01 7.90839940e-02 -5.24541020e-01 2.27785513e-01
-6.54242218e-01 -6.85884356e-01 -3.32842350e-01 3.72062504e-01
-3.64362076e-02 4.86635625e-01 -6.60647094e-01 2.89721966e-01
-3.78266573e-02 8.71854797e-02 6.94958389e-01 1.10182750e+00
-6.65784001e-01 -1.09273076e+00 -2.10808665e-01 3.72929096e-01
-6.64851606e-01 6.25583604e-02 2.25223020e-01 7.19105959e-01
-3.68564099e-01 9.67473149e-01 -1.85181111e-01 2.97409058e-01
2.31755167e-01 -7.82244951e-02 9.04478192e-01 -3.52538198e-01
-1.50279686e-01 2.32360497e-01 -1.01382375e-01 -8.61587703e-01
-3.92312557e-01 -7.94487298e-01 -1.27685118e+00 -3.83504242e-01
-1.74060866e-01 4.19101745e-01 6.15005851e-01 1.03806388e+00
-1.86220556e-02 3.52886498e-01 8.11653733e-01 -7.49535441e-01
-1.41332078e+00 -7.47882962e-01 -9.03274596e-01 -2.15508088e-01
5.51652670e-01 -8.67262244e-01 -4.18850780e-01 -5.72973669e-01] | [4.241131782531738, 2.5760788917541504] |
93eb11c1-d429-4c23-9047-4bf2dbda8e87 | auxiliary-tasks-and-exploration-enable | 2104.04112 | null | https://arxiv.org/abs/2104.04112v2 | https://arxiv.org/pdf/2104.04112v2.pdf | Auxiliary Tasks and Exploration Enable ObjectNav | ObjectGoal Navigation (ObjectNav) is an embodied task wherein agents are to navigate to an object instance in an unseen environment. Prior works have shown that end-to-end ObjectNav agents that use vanilla visual and recurrent modules, e.g. a CNN+RNN, perform poorly due to overfitting and sample inefficiency. This has motivated current state-of-the-art methods to mix analytic and learned components and operate on explicit spatial maps of the environment. We instead re-enable a generic learned agent by adding auxiliary learning tasks and an exploration reward. Our agents achieve 24.5% success and 8.1% SPL, a 37% and 8% relative improvement over prior state-of-the-art, respectively, on the Habitat ObjectNav Challenge. From our analysis, we propose that agents will act to simplify their visual inputs so as to smooth their RNN dynamics, and that auxiliary tasks reduce overfitting by minimizing effective RNN dimensionality; i.e. a performant ObjectNav agent that must maintain coherent plans over long horizons does so by learning smooth, low-dimensional recurrent dynamics. Site: https://joel99.github.io/objectnav/ | ['Erik Wijmans', 'Abhishek Das', 'Dhruv Batra', 'Joel Ye'] | 2021-04-08 | null | null | null | null | ['auxiliary-learning'] | ['methodology'] | [-3.94111753e-01 2.35067576e-01 1.95141420e-01 -1.39937580e-01
-5.03562987e-01 -6.71371043e-01 6.94936275e-01 -2.33674049e-01
-7.88251579e-01 5.35994947e-01 3.06126595e-01 -2.10259303e-01
-6.10692911e-02 -5.27417660e-01 -9.29199934e-01 -5.35379112e-01
-5.40059805e-01 5.83255768e-01 -4.34470996e-02 -4.40431297e-01
1.45457715e-01 2.47350782e-01 -1.48369765e+00 -1.50851607e-01
8.24319959e-01 6.41256869e-01 6.54600799e-01 1.01935112e+00
2.50836253e-01 1.14724672e+00 -2.36508086e-01 2.29983523e-01
4.92708623e-01 -2.07945585e-01 -7.06260204e-01 -2.59700626e-01
2.74868190e-01 -5.62074542e-01 -6.45007551e-01 8.68724704e-01
4.25244153e-01 6.63551927e-01 5.64074695e-01 -1.20955098e+00
-9.63540196e-01 5.49427330e-01 -3.15653831e-01 7.71842673e-02
-4.19795848e-02 9.89906073e-01 8.86797071e-01 -7.62196660e-01
8.26628089e-01 1.37242246e+00 6.38674080e-01 8.60409141e-01
-1.45088995e+00 -2.69858927e-01 8.37014794e-01 -1.98841491e-03
-1.07962894e+00 -5.94957173e-01 3.79763693e-01 -4.47362751e-01
1.53342581e+00 -6.49086982e-02 8.34314764e-01 1.39149749e+00
2.76328593e-01 7.96701789e-01 6.92856312e-01 1.90015510e-01
6.23396456e-01 -2.88697779e-01 -1.07196175e-01 8.62827837e-01
1.22019477e-01 5.49586773e-01 -4.21263188e-01 1.33973747e-01
9.47111368e-01 1.20058477e-01 -2.08391622e-01 -8.16113949e-01
-1.14252377e+00 8.70114088e-01 1.18391597e+00 -9.35779214e-02
-6.03784263e-01 8.78069520e-01 3.15955818e-01 1.91615701e-01
-5.25787361e-02 8.36656451e-01 -3.94170284e-01 -3.82980466e-01
-3.42947155e-01 8.68835568e-01 7.23190367e-01 1.14906275e+00
5.74093223e-01 5.04303277e-01 1.21471904e-01 5.32997549e-01
6.15812123e-01 5.64692080e-01 4.60429877e-01 -1.37274945e+00
2.86888778e-01 5.03599644e-01 3.45675200e-01 -5.20494878e-01
-7.26451218e-01 -6.60831034e-01 -5.22298574e-01 9.30101633e-01
2.33595863e-01 -2.99191058e-01 -1.59124577e+00 2.04569340e+00
9.43069756e-02 -1.09216407e-01 4.34528381e-01 1.28224707e+00
7.30130374e-01 7.79216766e-01 1.95851058e-01 4.69220251e-01
9.29455698e-01 -1.49709189e+00 -2.86843121e-01 -8.49160373e-01
6.14612162e-01 -7.89302140e-02 1.38752115e+00 2.25967288e-01
-1.20870149e+00 -3.12817007e-01 -9.68477011e-01 -1.93413749e-01
-3.49399984e-01 -5.69952764e-02 6.98761225e-01 -1.38279378e-01
-1.42674887e+00 5.42932034e-01 -1.36295843e+00 -5.12000799e-01
5.33019960e-01 3.25659305e-01 -3.47736448e-01 2.25571603e-01
-5.25005281e-01 1.10836899e+00 2.79998839e-01 3.09380352e-01
-1.85890520e+00 -7.03062594e-01 -1.03169811e+00 1.23292483e-01
4.85952795e-01 -9.56723154e-01 1.77750623e+00 -7.84144461e-01
-1.61877632e+00 2.39140585e-01 -5.76395057e-02 -8.25605392e-01
5.35790563e-01 -5.97126544e-01 2.97928929e-01 -2.33195409e-01
6.28506839e-02 1.04767394e+00 4.58116770e-01 -1.45657778e+00
-5.62129021e-01 -3.57799232e-01 1.12307981e-01 6.83146715e-01
1.24621436e-01 -6.13647699e-01 -4.43063170e-01 -2.89923549e-01
1.72749296e-01 -1.12115800e+00 -8.35201979e-01 2.46974140e-01
-2.14361131e-01 -1.59740254e-01 5.93729556e-01 -5.95100582e-01
5.44938207e-01 -1.80816460e+00 7.14531541e-01 -4.05527540e-02
5.20616770e-01 -1.13173887e-01 -5.28728187e-01 3.50795984e-01
3.72404486e-01 1.12394631e-01 -2.36811161e-01 -6.72581494e-01
1.83825582e-01 1.65510133e-01 -3.82895678e-01 4.38156635e-01
-1.22025488e-02 1.29853106e+00 -1.05633962e+00 1.60886288e-01
1.75081447e-01 5.56579053e-01 -7.86033750e-01 1.92638472e-01
-7.73146331e-01 4.25799698e-01 -3.93585652e-01 6.35946214e-01
2.08118483e-01 -3.23722363e-01 7.98051804e-02 4.19572294e-01
-4.13505524e-01 4.74376798e-01 -7.88360894e-01 2.16343737e+00
-4.57258701e-01 8.41484904e-01 3.46702605e-01 -5.08592963e-01
7.48020768e-01 -1.60814524e-01 6.99793249e-02 -1.07695889e+00
-2.28755386e-03 1.89336732e-01 1.68542400e-01 -3.71368349e-01
7.04436362e-01 2.89447606e-01 1.40182063e-01 7.45324194e-02
-2.09621396e-02 -2.47551072e-02 3.84280761e-03 2.48910174e-01
1.24586248e+00 6.41581118e-01 6.66343197e-02 -2.86284149e-01
-2.75472552e-01 3.94906908e-01 3.58076900e-01 1.03437269e+00
-2.72673339e-01 3.60365659e-01 3.29794943e-01 -5.83456755e-01
-1.37956786e+00 -1.22140479e+00 4.28688318e-01 1.33110809e+00
2.97861993e-01 -2.96194583e-01 -3.51319671e-01 -2.44707540e-01
9.62503403e-02 8.06047916e-01 -9.19225991e-01 -9.10255462e-02
-9.02463973e-01 -3.94521654e-01 2.77401745e-01 7.19731152e-01
4.04639363e-01 -1.35941088e+00 -1.20687187e+00 4.14796472e-01
2.40290463e-01 -5.68954468e-01 -2.70946354e-01 5.36969066e-01
-8.51418853e-01 -7.40390062e-01 -7.61643171e-01 -7.16477871e-01
5.43510020e-01 3.51919264e-01 1.07240367e+00 -4.51452509e-02
-1.86351165e-01 3.16841841e-01 -1.95024937e-01 -1.62843585e-01
-3.69997993e-02 4.42605078e-01 2.61935920e-01 -7.79374182e-01
-4.24291827e-02 -8.38741601e-01 -9.29703295e-01 8.51002187e-02
-4.92049873e-01 2.85446793e-01 6.51306629e-01 8.86579812e-01
4.31517601e-01 -6.77820265e-01 3.79807085e-01 -1.90776333e-01
6.67559028e-01 -6.08489454e-01 -1.12951744e+00 -1.00807406e-01
-7.35392094e-01 4.85034764e-01 5.50140083e-01 -6.02462232e-01
-8.73713017e-01 7.97741488e-02 2.03260854e-02 -6.72473788e-01
2.51683332e-02 4.65465397e-01 3.28713030e-01 1.70605388e-04
1.03881931e+00 3.98703188e-01 2.20877245e-01 -3.56292844e-01
5.70735693e-01 2.39207298e-02 6.11678123e-01 -4.30701405e-01
7.88187981e-01 5.11435509e-01 -2.42023125e-01 -4.96657401e-01
-3.46369505e-01 -1.64689198e-01 -7.40144253e-02 -1.15406424e-01
8.93228889e-01 -1.11128008e+00 -1.40489185e+00 2.19694883e-01
-1.02998865e+00 -1.45837355e+00 -3.73623848e-01 4.08385545e-01
-9.64170098e-01 -2.90452391e-01 -4.37536508e-01 -9.94284868e-01
-4.87381697e-01 -1.13301122e+00 8.48733246e-01 5.97673476e-01
-1.27808541e-01 -6.75156951e-01 4.48728055e-01 -2.34952271e-01
8.26711237e-01 7.19767883e-02 4.42870051e-01 -3.35815310e-01
-1.09930336e+00 3.89455765e-01 -1.49409920e-01 -2.62435168e-01
-2.38126576e-01 -3.33204776e-01 -8.27181280e-01 -5.27551591e-01
-3.71593833e-01 -6.08518958e-01 1.23847103e+00 5.28246403e-01
7.73563027e-01 -5.41165590e-01 -4.77561682e-01 9.80725884e-01
1.40425634e+00 3.67137730e-01 4.32498306e-01 8.20713937e-01
6.64781213e-01 4.67200875e-01 2.67941624e-01 3.36435229e-01
7.71058619e-01 6.69399142e-01 1.12044013e+00 1.26821741e-01
-7.45107280e-03 -5.73359907e-01 5.37858129e-01 1.28489122e-01
-1.33639976e-01 -3.34708393e-01 -1.14724350e+00 8.18786442e-01
-2.37936330e+00 -9.12223041e-01 5.04333675e-01 1.82601893e+00
4.79907751e-01 2.38905966e-01 2.53489822e-01 -8.80717695e-01
5.25636710e-02 3.20564091e-01 -1.27654004e+00 -1.83995992e-01
1.31026823e-02 -4.29823697e-01 7.04173923e-01 7.68718123e-01
-9.19626713e-01 1.38526404e+00 5.73931265e+00 3.04499626e-01
-1.01394200e+00 -6.45637587e-02 4.03997809e-01 -6.25253379e-01
-3.29599798e-01 -2.86804494e-02 -7.38166034e-01 1.33173466e-02
8.89761448e-01 -8.89698267e-02 1.12814355e+00 1.39761853e+00
3.96803707e-01 -6.02862723e-02 -1.13250101e+00 8.49294662e-01
-1.32197857e-01 -1.50636423e+00 -2.53695428e-01 2.56077111e-01
5.08103907e-01 8.62958312e-01 4.66151595e-01 7.37694383e-01
1.16025794e+00 -1.53480113e+00 1.07741380e+00 7.31646121e-01
3.26895088e-01 -3.91170919e-01 3.86128366e-01 4.56081092e-01
-1.20369804e+00 -3.23453575e-01 -2.93241233e-01 -2.88636982e-01
3.55923563e-01 -4.54561621e-01 -9.46805239e-01 -3.67994197e-02
1.01568079e+00 6.81826174e-01 -3.99262398e-01 1.19591856e+00
-8.31941441e-02 1.24672122e-01 -5.19023120e-01 -4.38655287e-01
7.97090232e-01 -1.63439065e-01 8.27845633e-01 8.05769503e-01
1.94600433e-01 8.14527795e-02 3.24978948e-01 1.25268114e+00
-5.15055098e-02 -4.17230695e-01 -9.44503844e-01 -8.06704685e-02
2.85362720e-01 1.08242762e+00 -4.65013504e-01 -1.38804778e-01
-5.61730377e-03 8.43017817e-01 9.42120731e-01 8.47157359e-01
-6.74779117e-01 -1.60970122e-01 1.17045093e+00 -1.38245344e-01
4.70138609e-01 -6.31021917e-01 -3.63692254e-01 -9.81125951e-01
6.10179603e-02 -7.58913875e-01 -5.70655130e-02 -9.33981597e-01
-9.11603153e-01 8.22267234e-01 -3.65515828e-01 -9.48199689e-01
-5.63095391e-01 -5.53876162e-01 -5.52542150e-01 7.58107662e-01
-1.40502512e+00 -1.09837461e+00 -5.22306979e-01 2.52875239e-01
9.52508092e-01 -2.29010388e-01 7.12389350e-01 -2.95467615e-01
-3.69107604e-01 3.08472306e-01 3.34935226e-02 -5.31541072e-02
1.61034822e-01 -1.30259550e+00 9.15722489e-01 7.47463405e-01
-9.80495736e-02 7.93251693e-01 9.61031377e-01 -7.90624201e-01
-1.77815604e+00 -1.10471630e+00 1.96489319e-01 -6.16032004e-01
5.78081608e-01 -6.12557948e-01 -8.50032449e-01 1.13545501e+00
2.94203788e-01 -2.50805505e-02 -1.61523059e-01 2.07642287e-01
-4.55544293e-01 1.69186771e-01 -8.58431339e-01 1.35883641e+00
1.45161760e+00 -2.72499472e-01 -4.45215017e-01 1.14582367e-01
1.03171921e+00 -6.55785799e-01 -2.62780309e-01 1.68919474e-01
7.01700509e-01 -7.49924898e-01 9.93869781e-01 -8.72467458e-01
2.15496600e-01 -5.53003192e-01 -2.38807276e-01 -1.57659245e+00
-5.41139662e-01 -8.54635358e-01 -3.76136243e-01 3.33518326e-01
6.07249677e-01 -6.17264926e-01 9.08609509e-01 5.78676283e-01
-3.91694427e-01 -7.64280140e-01 -7.92466164e-01 -8.70349467e-01
3.09978463e-02 -1.96170926e-01 3.27902108e-01 2.70562887e-01
-7.14243874e-02 4.20876592e-01 -4.65707928e-01 3.28102916e-01
5.73381424e-01 -8.28805715e-02 9.48278248e-01 -8.77987266e-01
-2.63194591e-01 -8.38338971e-01 -7.54492357e-02 -1.48013711e+00
6.95762336e-02 -6.81575298e-01 5.37600696e-01 -1.82883012e+00
5.66978715e-02 -5.83561957e-01 2.17962544e-02 6.39857233e-01
2.09269688e-01 -2.37323627e-01 5.82701325e-01 3.65878046e-01
-9.31931734e-01 1.17829239e+00 1.14587533e+00 -2.08966076e-01
-8.00233066e-01 -3.28162104e-01 -7.47239470e-01 6.30646408e-01
9.10124540e-01 -2.86516637e-01 -5.01606941e-01 -1.01159894e+00
3.00205946e-01 -7.84166455e-02 8.06727171e-01 -9.96616244e-01
5.06705761e-01 -3.09008271e-01 4.54128385e-01 -4.09624159e-01
7.79711306e-01 -5.60321748e-01 -3.60132195e-02 8.51969361e-01
-5.56469619e-01 4.53881323e-01 4.56728011e-01 7.62736082e-01
2.85753638e-01 3.35024036e-02 5.68240404e-01 -4.85886931e-01
-9.73522961e-01 3.82504702e-01 -5.61262727e-01 2.62266193e-02
8.91556740e-01 -1.17332786e-01 -7.26038158e-01 -5.57866395e-01
-7.80051172e-01 8.21343780e-01 6.25752807e-01 5.27257860e-01
8.63719344e-01 -1.05311763e+00 -5.64213037e-01 -4.21318039e-02
-1.27258841e-02 3.35832089e-01 2.84516066e-01 5.22767007e-01
-6.51607156e-01 4.10676539e-01 -3.54478657e-01 -5.40331781e-01
-7.03321934e-01 4.75004107e-01 8.63451362e-01 -6.49894625e-02
-1.08456230e+00 1.00140166e+00 4.51550096e-01 -7.59535491e-01
5.74514270e-01 -4.66253936e-01 -6.99212849e-02 -3.69668573e-01
4.11202073e-01 2.80650824e-01 -4.36925739e-01 -1.13878518e-01
-3.07148546e-01 3.71592224e-01 -2.49990702e-01 -5.88275492e-01
1.57652700e+00 -1.56024266e-02 3.33199352e-01 3.41739833e-01
8.33462954e-01 -5.37804127e-01 -2.09731483e+00 -6.38144389e-02
-1.56832188e-01 -1.42379358e-01 1.26403764e-01 -8.70860338e-01
-7.19367743e-01 6.19044483e-01 7.17778027e-01 9.48785543e-02
6.26210332e-01 1.49663001e-01 3.55279446e-01 1.03898215e+00
3.74013931e-01 -8.54827881e-01 2.83082604e-01 1.01539588e+00
1.33365428e+00 -1.39264882e+00 -2.99259335e-01 3.77707899e-01
-8.05114806e-01 8.20379615e-01 1.04062819e+00 -5.26159406e-01
1.79395735e-01 1.83602884e-01 7.11519597e-03 -3.89062673e-01
-1.24580097e+00 -4.47796911e-01 -1.75158475e-02 7.31157005e-01
-1.11447237e-01 -7.67994486e-03 4.50622499e-01 3.48463655e-01
-4.01133478e-01 -1.58765718e-01 4.01307046e-01 1.03006876e+00
-7.53672600e-01 -2.97071487e-01 7.46719837e-02 2.95943409e-01
2.05150470e-01 -1.65022597e-01 -2.36522444e-02 8.72341871e-01
-4.76355433e-01 5.75554729e-01 1.77516982e-01 -2.57244408e-01
3.58096659e-01 -2.56243527e-01 1.42740011e-01 -4.68689471e-01
-5.70444226e-01 -9.00071487e-02 2.41432320e-02 -1.05348527e+00
1.81872413e-01 -6.78185344e-01 -1.56635761e+00 -2.68674970e-01
1.99761435e-01 -1.52102664e-01 8.18122566e-01 5.28008878e-01
6.85153484e-01 7.22934127e-01 -2.40500141e-02 -1.41094434e+00
-8.48117113e-01 -9.32905912e-01 -1.51866843e-04 -1.13254152e-01
9.35222149e-01 -7.48221695e-01 -3.20971161e-01 -2.46544555e-01] | [4.500492095947266, 0.6341466903686523] |
36b3707d-510c-4b5c-bc74-a1e339575f47 | generate-then-select-open-ended-visual | 2305.18842 | null | https://arxiv.org/abs/2305.18842v1 | https://arxiv.org/pdf/2305.18842v1.pdf | Generate then Select: Open-ended Visual Question Answering Guided by World Knowledge | The open-ended Visual Question Answering (VQA) task requires AI models to jointly reason over visual and natural language inputs using world knowledge. Recently, pre-trained Language Models (PLM) such as GPT-3 have been applied to the task and shown to be powerful world knowledge sources. However, these methods suffer from low knowledge coverage caused by PLM bias -- the tendency to generate certain tokens over other tokens regardless of prompt changes, and high dependency on the PLM quality -- only models using GPT-3 can achieve the best result. To address the aforementioned challenges, we propose RASO: a new VQA pipeline that deploys a generate-then-select strategy guided by world knowledge for the first time. Rather than following the de facto standard to train a multi-modal model that directly generates the VQA answer, RASO first adopts PLM to generate all the possible answers, and then trains a lightweight answer selection model for the correct answer. As proved in our analysis, RASO expands the knowledge coverage from in-domain training data by a large margin. We provide extensive experimentation and show the effectiveness of our pipeline by advancing the state-of-the-art by 4.1% on OK-VQA, without additional computation cost. Code and models are released at http://cogcomp.org/page/publication_view/1010 | ['Bing Xiang', 'Dan Roth', 'Patrick Ng', 'Vittorio Castelli', 'Zhiguo Wang', 'William Yang Wang', 'Alexander Hanbo Li', 'Yuhao Zhang', 'Henghui Zhu', 'Pramuditha Perera', 'Gukyeong Kwon', 'Sheng Zhang', 'Xingyu Fu'] | 2023-05-30 | null | null | null | null | ['answer-selection'] | ['natural-language-processing'] | [-4.07001749e-02 3.55643362e-01 -1.28945261e-01 -3.11823070e-01
-1.25834596e+00 -1.00916660e+00 7.79111862e-01 1.90211877e-01
-3.28321368e-01 5.11388540e-01 3.31007510e-01 -4.65889722e-01
1.86417177e-01 -8.74502718e-01 -9.45790172e-01 -6.90530390e-02
3.91696841e-01 9.34189796e-01 5.60586870e-01 -3.38090122e-01
1.31734952e-01 4.71886918e-02 -1.58554959e+00 8.93568695e-01
9.50680912e-01 8.14022541e-01 1.22734837e-01 7.22806454e-01
-6.51665926e-01 1.37010932e+00 -6.40570998e-01 -7.89116740e-01
1.34757593e-01 -5.16709447e-01 -1.23100471e+00 -3.71857762e-01
6.69554353e-01 -3.46266538e-01 -1.73573270e-01 7.15170383e-01
5.55598974e-01 5.94479665e-02 5.30957222e-01 -1.34766436e+00
-1.23724687e+00 6.65207624e-01 -1.73678905e-01 -1.11661881e-01
6.60493970e-01 7.30182648e-01 1.33504188e+00 -1.02319002e+00
1.17434597e+00 1.40022957e+00 3.24853927e-01 8.09314609e-01
-1.02390385e+00 -3.90751302e-01 1.79939091e-01 5.85483015e-01
-1.15100873e+00 -4.65107381e-01 6.73639238e-01 -3.15439343e-01
1.37396300e+00 2.26914376e-01 4.66960728e-01 1.11369228e+00
-2.03129798e-01 1.29919040e+00 1.06286085e+00 -5.54728866e-01
2.47737154e-01 5.31316996e-02 1.47177860e-01 8.84518027e-01
-7.50578046e-02 -2.19782367e-01 -7.51623034e-01 -1.15731638e-02
4.44901586e-01 -7.20271111e-01 -2.39466488e-01 -4.85218138e-01
-1.29011977e+00 8.23362410e-01 6.08075976e-01 -3.96948569e-02
-3.06537002e-01 2.89581239e-01 4.06551659e-01 3.56465667e-01
1.60543710e-01 7.77543008e-01 -5.63340247e-01 -1.60457283e-01
-6.95263505e-01 4.95286167e-01 7.13591516e-01 8.92078161e-01
8.04924309e-01 -7.75616542e-02 -5.80595493e-01 6.41290605e-01
3.20933729e-01 5.32884657e-01 1.84326366e-01 -1.04528594e+00
7.18435168e-01 1.04946232e+00 1.81918740e-01 -6.93267941e-01
-2.79657513e-01 -1.59813389e-01 -2.67506540e-01 3.34547520e-01
7.66267478e-01 4.99815494e-02 -1.32919228e+00 1.70897675e+00
4.39334840e-01 -3.37129802e-01 3.13996285e-01 9.87210095e-01
1.34637165e+00 8.62282991e-01 3.94033104e-01 2.55592525e-01
1.43035984e+00 -1.15401006e+00 -5.99587798e-01 -5.09810984e-01
6.42662942e-01 -5.32270491e-01 1.67811179e+00 3.10170531e-01
-1.15704656e+00 -4.61783916e-01 -8.60455751e-01 -5.15820682e-01
-4.72335607e-01 5.70458621e-02 5.25776923e-01 2.64297277e-01
-1.22938395e+00 -4.48879227e-02 -5.70715725e-01 -4.13475513e-01
6.39003575e-01 -4.16463017e-02 -3.42536986e-01 -4.01567370e-01
-1.23228896e+00 1.12548482e+00 3.52847695e-01 -7.90057182e-02
-1.09713769e+00 -8.24063301e-01 -1.03957462e+00 -1.33599520e-01
6.93701446e-01 -1.07620370e+00 1.39163697e+00 -1.05913448e+00
-1.50321972e+00 8.97349238e-01 -4.32692677e-01 -4.88858432e-01
5.40856779e-01 -2.44970307e-01 -2.71844685e-01 5.16694307e-01
2.21533775e-01 1.14468586e+00 6.92995191e-01 -1.27509308e+00
-4.41579372e-01 -1.51287660e-01 6.95037961e-01 2.67394096e-01
2.01768160e-01 -4.78025861e-02 -7.28289902e-01 -1.94221854e-01
-1.64645851e-01 -6.23250246e-01 -8.03083852e-02 4.26426996e-03
-1.27982795e-01 -4.88597453e-01 4.17580575e-01 -7.45939314e-01
9.16811228e-01 -1.92249441e+00 5.13706058e-02 -5.19951545e-02
1.48447871e-01 2.98869818e-01 -4.86325413e-01 3.74608994e-01
3.91442567e-01 4.12774719e-02 -1.21516451e-01 -2.31835380e-01
2.79922217e-01 1.84243664e-01 -7.01440692e-01 -3.29998955e-02
6.94808722e-01 1.59896588e+00 -1.12125719e+00 -6.59705520e-01
1.71001926e-01 2.95317471e-01 -5.45767665e-01 3.02614838e-01
-1.04386139e+00 3.13326776e-01 -2.44567633e-01 6.47703588e-01
4.66833174e-01 -5.60437500e-01 9.61224139e-02 -1.26394719e-01
9.08202901e-02 4.72867012e-01 -9.24200177e-01 1.94561076e+00
-5.02931833e-01 6.29864335e-01 -3.32846582e-01 -4.02855456e-01
7.47732043e-01 2.97830403e-01 -1.92255333e-01 -1.11886144e+00
-2.95816958e-02 2.07709908e-01 -1.33263126e-01 -7.79017448e-01
5.04841447e-01 1.60269327e-02 -1.50079593e-01 2.79975504e-01
4.50481236e-01 -3.68372172e-01 3.57779711e-01 6.90770984e-01
1.10262048e+00 6.97998822e-01 2.02342480e-01 1.47627071e-01
3.39596242e-01 7.26536512e-01 3.11269075e-01 8.20642650e-01
-2.17189908e-01 6.83081388e-01 7.19183743e-01 -2.86589712e-01
-8.89637649e-01 -1.14898515e+00 3.42275858e-01 1.15713632e+00
1.40009433e-01 -4.38167214e-01 -5.33944905e-01 -9.03069794e-01
-1.41503006e-01 1.15491760e+00 -6.67269647e-01 5.76024577e-02
-4.24675196e-01 -1.73044130e-01 8.31164777e-01 5.97817898e-01
4.39611077e-01 -1.56340158e+00 -9.56584752e-01 7.19701052e-02
-4.87492561e-01 -1.21622729e+00 -3.92544381e-02 -2.05901280e-01
-4.08972710e-01 -1.04384983e+00 -5.72296381e-01 -6.34303749e-01
4.68874395e-01 -1.33072570e-01 1.70810962e+00 6.64505363e-02
-3.18341963e-02 6.37037158e-01 -5.74830115e-01 -4.67024952e-01
-4.62073416e-01 1.62842907e-02 -5.69600940e-01 -1.57121852e-01
3.35219592e-01 -1.24679052e-03 -6.18142724e-01 -5.01777381e-02
-9.13836300e-01 4.13252175e-01 5.18928468e-01 6.09216034e-01
6.69437051e-01 -4.37095940e-01 7.30207086e-01 -8.33690763e-01
4.91081029e-01 -3.49782139e-01 -5.99605501e-01 7.68485665e-01
-3.54724258e-01 2.76844084e-01 5.04745424e-01 -3.20460528e-01
-1.30359709e+00 6.07943982e-02 -2.09372669e-01 -3.42226058e-01
-1.77084252e-01 6.33020103e-01 -3.20731312e-01 3.47699225e-01
9.77194726e-01 2.08445013e-01 -3.43311191e-01 -1.71709999e-01
1.06862211e+00 1.99789435e-01 7.72459090e-01 -6.71265781e-01
6.08003557e-01 5.81255317e-01 -3.83965850e-01 -4.67058539e-01
-9.02176380e-01 -2.67310530e-01 -2.65165597e-01 -2.77190179e-01
1.10107017e+00 -9.65034008e-01 -7.28508890e-01 2.44530424e-01
-1.33326733e+00 -7.38081634e-01 -3.82211149e-01 4.06992473e-02
-4.51548964e-01 1.70067072e-01 -4.29740667e-01 -8.17336142e-01
-4.99574602e-01 -9.86465693e-01 9.01479661e-01 2.71473765e-01
-3.72498274e-01 -7.89510190e-01 1.09034464e-01 7.66105711e-01
3.84413093e-01 1.83875009e-01 1.06187665e+00 -6.76243782e-01
-8.43827784e-01 1.46740213e-01 -3.87728631e-01 -1.46682281e-02
-3.24549198e-01 5.01735173e-02 -1.10098112e+00 9.13639516e-02
-5.07665992e-01 -9.03839946e-01 9.74279284e-01 5.64247854e-02
7.30401635e-01 -1.53820381e-01 7.00492188e-02 3.09125960e-01
1.39996624e+00 -8.11063796e-02 6.55688643e-01 5.24636626e-01
7.87143111e-01 6.45025134e-01 6.41545177e-01 4.99127507e-02
1.02155650e+00 4.17384684e-01 7.47385025e-01 6.23408929e-02
-6.18727624e-01 -6.30707622e-01 4.54248309e-01 3.33495110e-01
2.41242215e-01 -4.46237713e-01 -1.23801851e+00 1.11268079e+00
-1.87189519e+00 -8.46114337e-01 -3.35115761e-01 1.87122321e+00
1.01124537e+00 7.77927712e-02 3.06870677e-02 -2.23579735e-01
1.25312060e-01 2.49258697e-01 -5.83492279e-01 -4.48419154e-01
-3.62367660e-01 4.11788315e-01 1.06612608e-01 7.14677751e-01
-7.71692276e-01 1.40855205e+00 6.15002775e+00 8.43412042e-01
-8.99128258e-01 1.99926794e-01 1.94528997e-01 -1.81391075e-01
-9.02313769e-01 3.22124779e-01 -5.31383097e-01 -1.55956866e-02
7.42314637e-01 1.79792166e-01 4.04032916e-01 6.35202527e-01
-1.85297489e-01 -3.32471758e-01 -9.07428861e-01 8.14768493e-01
2.29374692e-01 -1.58535159e+00 5.08640051e-01 -5.72255373e-01
6.25341833e-01 1.13592260e-01 1.02426466e-02 6.77645862e-01
6.69155121e-01 -1.15815890e+00 1.16476953e+00 4.85900819e-01
8.23360622e-01 -5.05204201e-01 5.13694584e-01 2.50714123e-01
-1.07396126e+00 2.01545730e-02 -2.39446923e-01 -1.61860690e-01
3.37366492e-01 4.72892910e-01 -1.05152380e+00 7.67087460e-01
6.79448068e-01 2.92578608e-01 -9.19279277e-01 7.80103683e-01
-9.21254992e-01 7.07373679e-01 -6.40400723e-02 -1.04096241e-01
2.90618360e-01 2.79285640e-01 4.02065128e-01 9.94080365e-01
-2.56039258e-02 8.93214643e-02 -2.12883856e-02 9.09637034e-01
-1.32442012e-01 1.74931869e-01 -3.85263145e-01 -8.06536451e-02
3.59998703e-01 1.12925637e+00 -5.21405041e-01 -5.75189888e-01
-4.52246606e-01 1.01650739e+00 7.17716813e-01 5.25496364e-01
-8.82015765e-01 -2.80868798e-01 2.23204166e-01 -6.02612784e-03
3.46205860e-01 -1.25838473e-01 -2.99778849e-01 -1.21332133e+00
2.07489744e-01 -1.19908869e+00 7.20154226e-01 -1.38493800e+00
-1.22837532e+00 7.74957180e-01 -2.92904936e-02 -7.87458360e-01
-3.70092839e-01 -6.76412404e-01 -3.06153476e-01 7.86507845e-01
-1.80496454e+00 -1.72643518e+00 -2.40263596e-01 6.88278496e-01
7.08901942e-01 5.00597395e-02 7.43749022e-01 -8.66200030e-02
-7.87105411e-02 4.00364071e-01 -6.42867923e-01 1.38194710e-01
7.97850907e-01 -1.43927848e+00 7.30093539e-01 9.82780397e-01
4.48199004e-01 4.44505930e-01 6.88766360e-01 -7.43831456e-01
-1.38892651e+00 -9.55787182e-01 1.25300550e+00 -1.21492386e+00
7.39509940e-01 -3.18712026e-01 -1.05016613e+00 7.76209295e-01
5.33363402e-01 8.09299853e-03 4.31815863e-01 1.62456464e-02
-9.68742847e-01 2.10215881e-01 -9.87431526e-01 7.47821748e-01
1.05267453e+00 -7.34452188e-01 -9.90092158e-01 8.12620446e-02
9.78236318e-01 -6.37257040e-01 -5.05755723e-01 3.56133640e-01
4.55057681e-01 -8.41493309e-01 7.98354983e-01 -7.13969409e-01
6.56526387e-01 -7.69027352e-01 -1.59079507e-01 -1.11971045e+00
-1.43362790e-01 -4.42745537e-01 -3.16043019e-01 1.31202209e+00
7.28399754e-01 -4.83732015e-01 4.71597970e-01 6.58827841e-01
-6.55524582e-02 -6.03631437e-01 -9.20625746e-01 -4.96198624e-01
1.28990794e-02 -7.90112674e-01 5.90891182e-01 7.59212077e-01
-8.69608372e-02 6.33616745e-01 -9.36466232e-02 2.47586161e-01
2.93427050e-01 1.60363212e-01 9.76094663e-01 -8.42093289e-01
-3.90989214e-01 -2.92072952e-01 -3.83994915e-02 -1.02195132e+00
2.04960927e-01 -9.14871335e-01 1.07041754e-01 -2.29109526e+00
1.20048471e-01 -1.56610206e-01 -5.89602590e-02 9.81361985e-01
-4.50362772e-01 3.31264079e-01 4.75780934e-01 9.06457603e-02
-1.04195487e+00 4.25946712e-01 1.34513342e+00 -2.74830788e-01
-2.19870120e-01 -5.97443402e-01 -7.84158289e-01 5.46888888e-01
6.63918018e-01 -1.54258415e-01 -6.96109176e-01 -1.04396594e+00
9.26102102e-01 -1.09646738e-01 6.49144709e-01 -7.49437809e-01
1.10114984e-01 -1.56428173e-01 2.47376159e-01 -5.42230785e-01
3.62250268e-01 -5.42574286e-01 -1.09720066e-01 2.15401426e-01
-3.94432366e-01 1.04168706e-01 4.25368071e-01 2.84921944e-01
-1.46701977e-01 -7.48011097e-02 3.53359252e-01 -2.25332767e-01
-1.28777730e+00 1.24141984e-02 -2.38476887e-01 3.17234337e-01
8.18090260e-01 -1.04508549e-02 -1.04674244e+00 -5.68677187e-01
-3.41580868e-01 6.30967498e-01 4.66465741e-01 5.74006021e-01
6.30688369e-01 -1.15886784e+00 -9.47477221e-01 -2.21850425e-01
6.50785625e-01 1.42265290e-01 4.33045357e-01 5.11781394e-01
-6.23688340e-01 3.96931976e-01 -4.82021160e-02 -4.96300340e-01
-9.92006004e-01 7.01778352e-01 2.47071475e-01 -4.17119622e-01
-3.75475794e-01 9.32084322e-01 9.69463587e-02 -6.93721056e-01
-6.48510689e-03 -2.95643836e-01 -2.87595481e-01 2.94247083e-02
4.81370956e-01 4.24669236e-02 5.83356023e-02 -5.01745045e-01
-6.22058868e-01 3.58192712e-01 2.07375407e-01 -5.00828326e-01
1.04181552e+00 4.91705686e-02 -4.11560945e-02 3.91339064e-01
6.86511159e-01 8.44690055e-02 -1.22747767e+00 -2.78819084e-01
2.37649474e-02 -3.30061793e-01 -3.43199819e-01 -1.41018152e+00
-7.01649308e-01 9.23890233e-01 2.82711804e-01 -9.51292813e-02
1.04496336e+00 4.33129549e-01 5.88184953e-01 3.31047207e-01
3.51114213e-01 -9.01560307e-01 2.95341611e-01 6.64022863e-01
1.08173490e+00 -1.31952679e+00 -3.00647318e-01 -1.14118606e-01
-1.10052276e+00 7.72640347e-01 1.00897455e+00 1.26358226e-01
-8.79697427e-02 -1.72845095e-01 6.50534809e-01 -3.12865824e-01
-1.07273936e+00 -5.93935728e-01 4.60003138e-01 8.20943832e-01
2.15429574e-01 -4.75715064e-02 5.33453822e-02 4.74747181e-01
-1.76551655e-01 7.47312307e-02 2.53467619e-01 7.94588327e-01
-2.60508984e-01 -9.68751967e-01 -2.83478349e-01 9.50819254e-02
-1.55681014e-01 -2.83673167e-01 -6.39213860e-01 7.88696706e-01
9.17006284e-02 1.23353779e+00 -1.74401365e-02 -1.56029090e-01
5.29183447e-01 3.95158470e-01 5.83399832e-01 -5.22412002e-01
-6.74095154e-01 -4.63080108e-01 3.26704502e-01 -5.88800967e-01
-3.81170720e-01 -3.16879749e-01 -1.53231990e+00 -2.27212738e-02
2.01816559e-02 -1.31608680e-01 3.77906978e-01 1.04613280e+00
6.41129494e-01 5.34340262e-01 -1.40655592e-01 -3.29706907e-01
-2.89582312e-01 -7.41456151e-01 2.64175445e-01 6.23728454e-01
2.81654298e-01 -3.62030476e-01 -8.69523063e-02 6.29687533e-02] | [10.881878852844238, 1.8471035957336426] |
8eb6951a-d317-43c7-a6a5-366c57a42bca | language-model-as-an-annotator-exploring | 2105.12544 | null | https://arxiv.org/abs/2105.12544v2 | https://arxiv.org/pdf/2105.12544v2.pdf | Language Model as an Annotator: Exploring DialoGPT for Dialogue Summarization | Current dialogue summarization systems usually encode the text with a number of general semantic features (e.g., keywords and topics) to gain more powerful dialogue modeling capabilities. However, these features are obtained via open-domain toolkits that are dialog-agnostic or heavily relied on human annotations. In this paper, we show how DialoGPT, a pre-trained model for conversational response generation, can be developed as an unsupervised dialogue annotator, which takes advantage of dialogue background knowledge encoded in DialoGPT. We apply DialoGPT to label three types of features on two dialogue summarization datasets, SAMSum and AMI, and employ pre-trained and non pre-trained models as our summarizes. Experimental results show that our proposed method can obtain remarkable improvements on both datasets and achieves new state-of-the-art performance on the SAMSum dataset. | ['Ting Liu', 'Bing Qin', 'Libo Qin', 'Xiaocheng Feng', 'Xiachong Feng'] | 2021-05-26 | null | https://aclanthology.org/2021.acl-long.117 | https://aclanthology.org/2021.acl-long.117.pdf | acl-2021-5 | ['conversational-response-generation'] | ['natural-language-processing'] | [ 2.90764570e-01 7.80554235e-01 -1.52576491e-01 -5.64485908e-01
-8.83603454e-01 -7.37147272e-01 1.20569348e+00 1.47720933e-01
-2.12862819e-01 1.08880734e+00 8.65419269e-01 -7.09218234e-02
4.51247007e-01 -6.71885848e-01 -3.07838563e-02 -2.08390921e-01
2.60590047e-01 8.41042697e-01 3.30137223e-01 -9.33201969e-01
3.11468393e-01 -2.62272984e-01 -8.97525191e-01 7.41166174e-01
1.29171073e+00 7.75468707e-01 1.24897979e-01 1.03623426e+00
-6.96204543e-01 1.24878800e+00 -1.01476371e+00 -6.73810303e-01
-1.62944242e-01 -8.80380869e-01 -1.49346125e+00 1.60075173e-01
-5.80568761e-02 -5.83886623e-01 -4.60640848e-01 6.33007109e-01
4.61984932e-01 2.26957589e-01 8.73638570e-01 -9.09312785e-01
-4.08895463e-01 1.07238150e+00 -8.21906552e-02 -1.51182935e-01
9.03740287e-01 1.32402092e-01 1.22699320e+00 -5.85031688e-01
5.98718584e-01 1.41523123e+00 4.20891076e-01 8.96534503e-01
-1.05435061e+00 -1.71947822e-01 -1.12588875e-01 8.23116377e-02
-5.67879677e-01 -7.20815718e-01 9.04878497e-01 -2.76687622e-01
1.12173605e+00 6.31689906e-01 2.62436986e-01 1.39211881e+00
-1.44356400e-01 1.18505847e+00 1.02753472e+00 -4.48098242e-01
1.06426552e-01 4.19162273e-01 4.37881172e-01 6.06560767e-01
-5.73097587e-01 -6.85586572e-01 -6.74956679e-01 -5.53761780e-01
3.32110792e-01 -5.32888234e-01 -4.64546919e-01 1.67460628e-02
-1.06413782e+00 1.24799466e+00 1.31841138e-01 1.80190668e-01
-1.34503946e-01 -5.68872273e-01 8.28280330e-01 4.72824275e-01
6.05135739e-01 8.72817457e-01 -6.32453501e-01 -4.73731697e-01
-5.43137312e-01 3.02924812e-01 1.73207557e+00 1.10978889e+00
7.15701878e-01 -2.29667664e-01 -6.30340457e-01 1.46873140e+00
-4.15231325e-02 2.03426808e-01 8.07301700e-01 -1.07261479e+00
6.12022042e-01 9.58347499e-01 2.26533040e-01 -6.81038678e-01
-4.87409562e-01 3.35676730e-01 -8.29731405e-01 -5.50062120e-01
3.21766019e-01 -6.37119174e-01 -3.54747295e-01 1.45318842e+00
1.48439541e-01 -3.12651783e-01 7.39584565e-01 5.86311758e-01
1.54084587e+00 8.53044629e-01 -1.24694891e-01 -3.93879384e-01
1.31686485e+00 -1.57918990e+00 -8.60325456e-01 -2.38231122e-01
6.96433783e-01 -6.87986255e-01 1.14398956e+00 8.81037638e-02
-1.06895852e+00 -3.20971310e-01 -8.15740108e-01 -2.20485106e-01
-2.57243901e-01 6.21894896e-02 6.24821842e-01 4.69663799e-01
-8.59050035e-01 5.27348340e-01 -5.46826363e-01 -6.67614579e-01
-1.08165741e-01 -5.72832115e-03 -1.98930681e-01 3.80672187e-01
-1.40907574e+00 1.01981056e+00 4.54945862e-01 -3.25213283e-01
-7.67769754e-01 -2.41302624e-01 -9.85190749e-01 1.80210888e-01
6.13817871e-01 -7.80726790e-01 1.98457766e+00 -7.40577281e-01
-2.49764228e+00 7.23084271e-01 -4.70115580e-02 -6.31347239e-01
5.03026366e-01 -3.08845460e-01 -2.02715136e-02 4.10490423e-01
4.15169708e-02 5.19840062e-01 5.70856750e-01 -1.04949164e+00
-5.25669038e-01 -5.90147302e-02 5.23916125e-01 6.35463834e-01
-5.88649929e-01 1.15714453e-01 -2.60166258e-01 -3.48308772e-01
-4.99948502e-01 -8.65162253e-01 -3.43960732e-01 -7.10955858e-01
-9.34271872e-01 -6.21267557e-01 7.29940355e-01 -8.16221178e-01
1.14815593e+00 -1.52241588e+00 3.33130449e-01 -5.33567727e-01
2.15728626e-01 4.72125709e-01 -1.35030858e-02 1.05793321e+00
6.42114878e-01 -5.49513474e-02 -3.87238264e-01 -3.81157219e-01
1.94138467e-01 1.64810896e-01 -3.85428935e-01 -2.82460064e-01
2.26408765e-01 7.48498261e-01 -9.58550394e-01 -5.98465383e-01
2.57788509e-01 -2.24252746e-01 -3.93620521e-01 8.71783078e-01
-7.11311042e-01 7.35350728e-01 -7.84870148e-01 8.61618668e-02
2.39193365e-01 -3.23416740e-01 4.82201338e-01 1.03299253e-01
1.83396507e-02 7.98196495e-01 -3.87601227e-01 2.15095377e+00
-6.40894175e-01 3.93174410e-01 1.58145323e-01 -9.43338633e-01
1.05266154e+00 5.93854666e-01 2.01031744e-01 -1.36650264e-01
3.58200341e-01 -1.14956059e-01 -1.17293388e-01 -4.33810472e-01
9.74110067e-01 2.55503267e-01 -7.00837493e-01 8.38364184e-01
5.38018644e-01 -2.72039771e-01 2.81739593e-01 7.15079129e-01
1.27374005e+00 -3.72321010e-01 7.03024030e-01 -1.72099486e-01
8.94488335e-01 2.85936058e-01 3.03453028e-01 8.85080218e-01
-3.39641385e-02 2.60999203e-01 8.11226785e-01 -1.59129817e-02
-7.17451513e-01 -5.48207164e-01 3.25501114e-02 1.47751153e+00
-4.98265326e-02 -8.11163068e-01 -9.45303977e-01 -1.16317296e+00
-2.90861756e-01 8.36695850e-01 -1.85970634e-01 -1.08329564e-01
-5.33606172e-01 -5.98248661e-01 8.31765115e-01 3.26979458e-01
9.71793532e-01 -1.15602219e+00 -3.44377398e-01 3.64866734e-01
-6.44174516e-01 -1.24365914e+00 -3.94825369e-01 -4.26918678e-02
-5.75436234e-01 -9.54709351e-01 -6.41595960e-01 -7.17498541e-01
2.66862601e-01 6.47995174e-02 1.14676452e+00 -1.56573072e-01
2.48426259e-01 5.33696890e-01 -7.03065455e-01 -2.19905227e-01
-1.10384738e+00 5.99184990e-01 -3.65197510e-01 -2.24802554e-01
1.95705771e-01 -4.50551093e-01 -3.34170669e-01 2.76834518e-01
-4.98658568e-01 4.41223770e-01 3.39692324e-01 1.22981405e+00
-4.06669587e-01 -5.42634606e-01 1.20365548e+00 -1.34259474e+00
1.35249805e+00 -3.37048411e-01 1.78684667e-02 3.71520340e-01
-2.83369511e-01 2.04165906e-01 8.62441301e-01 -1.73634395e-01
-1.74382198e+00 -2.58809060e-01 -2.50008643e-01 3.23146373e-01
-3.46817970e-01 7.08398044e-01 -3.43197852e-01 1.97206050e-01
7.57280231e-01 2.95316339e-01 9.29178949e-03 -6.42630577e-01
8.68705213e-01 1.34331143e+00 5.71857274e-01 -7.26663589e-01
2.41870016e-01 -3.83787975e-02 -6.73332810e-01 -9.53376591e-01
-1.09580493e+00 -6.02962077e-01 -6.00882292e-01 -3.22593413e-02
7.25821912e-01 -8.31442356e-01 -4.77395624e-01 4.89013642e-01
-1.34250760e+00 -5.10563493e-01 -1.42484024e-01 -1.80545095e-02
-7.75305986e-01 6.39573514e-01 -1.06873369e+00 -7.53229737e-01
-8.07703316e-01 -7.24625707e-01 7.73611546e-01 4.68274951e-01
-6.48863733e-01 -1.03452337e+00 3.11561406e-01 7.48287261e-01
3.73121977e-01 -2.50199437e-03 9.95541513e-01 -1.64814544e+00
-1.26635879e-01 -3.92896533e-02 -1.06920652e-01 4.50618714e-01
2.99600214e-01 -3.26520562e-01 -1.05539966e+00 -2.35088281e-02
1.03700183e-01 -9.18470681e-01 7.97739089e-01 -1.46227792e-01
7.43302703e-01 -9.20764983e-01 -2.26499379e-01 -1.09986305e-01
5.86889029e-01 1.27688646e-01 4.01262015e-01 4.75678071e-02
4.13928360e-01 8.21033537e-01 6.68895543e-01 7.51765966e-01
8.05203438e-01 6.49844885e-01 -1.08682275e-01 9.53998044e-02
1.24508925e-01 -3.11738759e-01 3.74483854e-01 1.14421892e+00
1.74473256e-01 -4.45405304e-01 -7.23937750e-01 5.29983461e-01
-2.34596562e+00 -7.55948782e-01 7.94270784e-02 1.63986564e+00
1.62816072e+00 -7.10614622e-02 2.64889359e-01 -4.17119622e-01
5.28801203e-01 5.56418300e-01 -4.36214268e-01 -7.51973510e-01
-8.45278502e-02 2.08684523e-02 -6.00737073e-02 5.07868946e-01
-1.11355293e+00 1.36061382e+00 6.03894186e+00 6.55934930e-01
-7.18965292e-01 3.59838009e-02 4.41431969e-01 2.94143885e-01
8.13163444e-03 -2.91652121e-02 -6.45448804e-01 3.14583838e-01
1.04778099e+00 -7.33721554e-01 2.05449253e-01 9.61555839e-01
-1.18623264e-01 1.65088531e-02 -1.37072647e+00 5.99492431e-01
2.85286933e-01 -1.29081357e+00 1.11653663e-01 -3.09596270e-01
7.27215648e-01 -2.85240561e-01 -4.91495788e-01 9.11092699e-01
8.97130489e-01 -8.00662994e-01 -4.58451360e-02 2.94594109e-01
4.09910172e-01 -5.50329685e-01 8.33623767e-01 7.02524841e-01
-5.87927580e-01 1.35228708e-01 -4.08496320e-01 -2.08991602e-01
2.74077207e-01 4.00044322e-02 -1.53973925e+00 8.10757041e-01
1.43054202e-01 6.37865007e-01 -3.24212730e-01 5.30258179e-01
-4.13162589e-01 6.58449233e-01 1.34551078e-02 -3.26786339e-01
2.52199978e-01 -8.87954757e-02 7.22599506e-01 1.36588359e+00
-2.41455302e-01 3.61481965e-01 7.19562531e-01 5.83084702e-01
-3.99615437e-01 3.37332487e-01 -5.68090498e-01 -2.33139187e-01
4.39298511e-01 1.59297061e+00 -1.56826913e-01 -7.14643061e-01
-5.20467103e-01 1.10790861e+00 3.74896914e-01 1.11213706e-01
-4.95747328e-01 -5.52387774e-01 3.29622358e-01 -4.21416610e-01
3.26989219e-02 7.04632550e-02 -4.05199528e-02 -1.43606246e+00
-3.17344040e-01 -1.13031578e+00 4.79040325e-01 -4.62463945e-01
-1.51819170e+00 7.79585719e-01 1.33288369e-01 -7.80604184e-01
-1.03068054e+00 -2.34828055e-01 -1.04293036e+00 6.94795668e-01
-1.11021256e+00 -1.24051929e+00 -2.40818992e-01 6.00465178e-01
1.20144534e+00 -4.52510267e-01 1.33979726e+00 -4.14917111e-01
-5.58743477e-01 4.78605390e-01 1.85774624e-01 4.95263487e-01
1.09205127e+00 -1.63317335e+00 3.74285936e-01 1.57408923e-01
-1.37369394e-01 4.31327522e-01 8.14586639e-01 -4.96480107e-01
-1.04182351e+00 -9.09702837e-01 8.31006885e-01 -4.73040551e-01
8.16702902e-01 -3.82440686e-01 -8.95195067e-01 7.74674237e-01
1.08941269e+00 -9.70874548e-01 1.05570841e+00 4.76643860e-01
-3.40956077e-02 4.34757143e-01 -1.08223367e+00 6.34177029e-01
7.24202156e-01 -4.82571095e-01 -1.32952154e+00 6.58658087e-01
8.93797755e-01 -5.98616779e-01 -1.03027344e+00 2.82225072e-01
2.00558066e-01 -8.12781215e-01 6.62610412e-01 -9.84948814e-01
6.14998400e-01 3.27771336e-01 1.29367277e-01 -1.73335946e+00
2.19915569e-01 -1.08202171e+00 -2.70291120e-01 1.68568492e+00
5.02395570e-01 -4.85748589e-01 5.12529790e-01 7.59977758e-01
-4.42326933e-01 -5.00733018e-01 -5.79873264e-01 -4.07304198e-01
1.50802389e-01 3.21553260e-01 5.02678514e-01 1.10593402e+00
1.05567098e+00 1.42845964e+00 -5.54957926e-01 -4.70374972e-01
1.08845711e-01 4.06554878e-01 1.34951937e+00 -1.35246682e+00
-1.62676573e-01 -3.25450778e-01 1.17554933e-01 -1.64088941e+00
7.00595140e-01 -7.16309309e-01 2.09374234e-01 -1.63389623e+00
3.42120349e-01 -1.06736071e-01 2.24908516e-01 4.10911202e-01
-4.00008768e-01 -4.90997046e-01 2.76186913e-02 1.35120407e-01
-1.01290870e+00 1.09008205e+00 1.16951132e+00 -2.86440372e-01
-5.52374363e-01 2.50034243e-01 -8.93141627e-01 1.00201404e+00
8.45760167e-01 -1.55863509e-01 -4.92511213e-01 -1.36312082e-01
-4.16162401e-01 7.51193762e-01 -8.37944373e-02 -5.55040598e-01
3.87997061e-01 -1.82866633e-01 -7.93075264e-02 -5.54505110e-01
5.77457309e-01 -1.00255720e-01 -6.69884682e-01 -1.29318774e-01
-9.89910662e-01 -4.64019895e-01 -3.90399732e-02 5.88884056e-01
-3.52813423e-01 -4.64646190e-01 3.80473197e-01 -4.15254593e-01
-4.35042232e-01 -1.31770357e-01 -5.90665042e-01 3.90248537e-01
6.24724448e-01 3.52291048e-01 -8.37584198e-01 -1.05050826e+00
-6.77970707e-01 6.44383907e-01 3.66712958e-01 4.28700596e-01
2.73036420e-01 -1.04136872e+00 -8.35775256e-01 -3.10443610e-01
1.37628376e-01 4.43866625e-02 2.10517794e-01 5.24016917e-01
-1.16947308e-01 7.31430411e-01 -2.30232865e-01 -4.38802272e-01
-1.34366202e+00 1.25715062e-01 2.89805550e-02 -8.51088762e-01
-5.99921525e-01 6.01329684e-01 2.25000247e-01 -8.84758353e-01
2.28070617e-01 3.61218080e-02 -6.51504934e-01 7.68283829e-02
6.22348785e-01 2.27769658e-01 -1.80814341e-01 -4.87621933e-01
7.42152035e-02 -2.11884454e-01 -5.08707285e-01 -2.83196658e-01
1.07642496e+00 -3.56147885e-01 -1.37481540e-01 4.94427979e-01
8.01554322e-01 6.39583245e-02 -1.02654958e+00 -6.91833913e-01
2.68235356e-01 -1.26853928e-01 -3.79924476e-01 -1.08793724e+00
-4.72296506e-01 7.34710455e-01 -3.02426547e-01 7.06166565e-01
7.65090764e-01 1.54043838e-01 1.08200514e+00 1.05823636e+00
2.54948407e-01 -1.18186390e+00 3.88467312e-01 1.02147567e+00
1.14953566e+00 -1.32554638e+00 -1.88998446e-01 -5.02997220e-01
-1.36234748e+00 1.17761922e+00 9.72343624e-01 4.35577929e-02
2.16353357e-01 -1.44668579e-01 2.13644400e-01 -4.67258878e-02
-1.12884486e+00 -1.73647463e-01 1.44527331e-01 4.22661275e-01
5.70931256e-01 4.22747731e-02 -4.97528881e-01 1.15017688e+00
-4.70506787e-01 -3.77964497e-01 7.96171486e-01 9.11697030e-01
-6.83359563e-01 -1.23041654e+00 1.69343352e-01 5.11712611e-01
-2.56081104e-01 -8.65609050e-02 -1.31192148e+00 5.77464759e-01
-8.32340002e-01 1.42229927e+00 -2.64369011e-01 -4.39562380e-01
3.41833234e-01 4.41855907e-01 2.63066590e-01 -1.08029282e+00
-9.67819571e-01 -1.14260219e-01 1.04416084e+00 -1.83953375e-01
-4.51519817e-01 -3.07657570e-01 -1.10094774e+00 -2.78624505e-01
-5.00878453e-01 6.91398203e-01 2.12918267e-01 9.89741921e-01
3.32446307e-01 4.37559158e-01 8.79693866e-01 -6.38387918e-01
-9.93311167e-01 -1.78111231e+00 -2.07595423e-01 4.04369235e-01
-1.30038885e-02 -2.94226408e-01 -1.98883638e-01 -2.28339154e-03] | [12.70919418334961, 8.147547721862793] |
f3db9a4c-7b10-4b39-add2-d6f8c2086389 | ecg-atk-gan-robustness-against-adversarial | 2110.09983 | null | https://arxiv.org/abs/2110.09983v3 | https://arxiv.org/pdf/2110.09983v3.pdf | ECG-ATK-GAN: Robustness against Adversarial Attacks on ECGs using Conditional Generative Adversarial Networks | Automating arrhythmia detection from ECG requires a robust and trusted system that retains high accuracy under electrical disturbances. Many machine learning approaches have reached human-level performance in classifying arrhythmia from ECGs. However, these architectures are vulnerable to adversarial attacks, which can misclassify ECG signals by decreasing the model's accuracy. Adversarial attacks are small crafted perturbations injected in the original data which manifest the out-of-distribution shifts in signal to misclassify the correct class. Thus, security concerns arise for false hospitalization and insurance fraud abusing these perturbations. To mitigate this problem, we introduce the first novel Conditional Generative Adversarial Network (GAN), robust against adversarial attacked ECG signals and retaining high accuracy. Our architecture integrates a new class-weighted objective function for adversarial perturbation identification and new blocks for discerning and combining out-of-distribution shifts in signals in the learning process for accurately classifying various arrhythmia types. Furthermore, we benchmark our architecture on six different white and black-box attacks and compare them with other recently proposed arrhythmia classification models on two publicly available ECG arrhythmia datasets. The experiment confirms that our model is more robust against such adversarial attacks for classifying arrhythmia with high accuracy. | ['Xingjun Ma', 'Alireza Tavakkoli', 'Sharif Amit Kamran', 'Khondker Fariha Hossain'] | 2021-10-17 | null | null | null | null | ['arrhythmia-detection'] | ['medical'] | [ 5.32056630e-01 9.34992135e-02 3.34154725e-01 -2.96211869e-01
-1.20806956e+00 -1.12301123e+00 -4.29764502e-02 1.28671052e-02
4.01928872e-02 9.42383111e-01 -2.18319163e-01 -3.91507715e-01
1.04891257e-02 -6.21964514e-01 -8.09582949e-01 -7.27893710e-01
-5.58296919e-01 2.05968723e-01 -4.15504724e-02 -6.28750920e-02
-2.00944886e-01 5.34748614e-01 -6.79659426e-01 6.14117920e-01
7.57883072e-01 8.88439476e-01 -1.00831318e+00 1.07838273e+00
6.66274309e-01 8.11921537e-01 -1.44442010e+00 -4.53468740e-01
4.38015759e-01 -7.81493545e-01 -6.05155230e-01 -4.99390930e-01
2.09513098e-01 -1.61870793e-01 -2.97664821e-01 1.12709081e+00
1.04219663e+00 -6.81769192e-01 7.30175138e-01 -1.32126880e+00
-4.32638288e-01 8.28816712e-01 -1.62884042e-01 3.96390766e-01
4.64267045e-01 4.73447233e-01 4.11355793e-01 -3.45076621e-01
2.99293637e-01 8.10345352e-01 1.10874259e+00 8.19955826e-01
-1.28398168e+00 -9.91506159e-01 -2.46339038e-01 -2.11673036e-01
-1.46418512e+00 -1.11504063e-01 1.15778959e+00 -4.25235927e-01
3.95195782e-01 9.35826600e-01 5.78475118e-01 1.74566483e+00
6.99332654e-01 3.99031818e-01 1.09601104e+00 -8.34571347e-02
2.33065888e-01 -4.42164019e-02 -1.64645642e-01 2.57408142e-01
1.63464263e-01 4.10182208e-01 -3.85391742e-01 -9.38678920e-01
4.52864766e-01 5.82132395e-03 -7.00469375e-01 5.67272566e-02
-1.20518136e+00 5.41534245e-01 2.95696139e-01 2.43199557e-01
-3.42259973e-01 1.78997606e-01 6.27626002e-01 6.10106647e-01
2.31463432e-01 7.99644649e-01 -5.67498624e-01 -9.82642993e-02
-7.56140411e-01 2.33476251e-01 7.39387393e-01 5.88255227e-01
-5.73872179e-02 6.58475161e-01 -6.16688013e-01 5.08977830e-01
-2.10550558e-02 7.70375848e-01 6.31250918e-01 -4.03432399e-01
2.52515793e-01 3.04081351e-01 -1.18928879e-01 -1.12003541e+00
-4.13522452e-01 -9.75298703e-01 -1.23517096e+00 1.86274469e-01
3.38310868e-01 -5.59889734e-01 -9.61658716e-01 1.80775917e+00
2.35912334e-02 4.73594099e-01 3.88944894e-01 6.67109489e-01
5.85468829e-01 2.50951052e-01 4.89716511e-03 -1.11063480e-01
1.14299655e+00 -2.19294980e-01 -7.66477227e-01 -6.61596879e-02
2.39106998e-01 -6.27520978e-01 7.77267396e-01 6.34985805e-01
-1.06376302e+00 -5.74677408e-01 -1.39281213e+00 6.91628158e-01
-9.62476358e-02 -3.73926789e-01 2.80447066e-01 1.43724644e+00
-6.96022391e-01 7.40836978e-01 -6.76846445e-01 3.24460536e-01
7.43911862e-01 4.18283194e-01 -1.46365270e-01 3.95040929e-01
-1.73048866e+00 6.39216244e-01 3.48320007e-02 1.59611180e-01
-1.20674336e+00 -1.04897285e+00 -5.91098964e-01 -1.48907185e-01
-2.21976891e-01 -6.38354301e-01 8.83491695e-01 -1.21744025e+00
-1.23019087e+00 8.67737114e-01 5.81772447e-01 -9.13817167e-01
8.01569760e-01 -6.68129250e-02 -1.09279048e+00 1.42485369e-02
-2.75185913e-01 -1.79365173e-01 1.44463420e+00 -1.24325955e+00
-5.97810149e-02 -4.92327631e-01 -2.48673335e-01 -4.07883257e-01
-2.95306206e-01 -6.88113347e-02 2.35518262e-01 -1.39053929e+00
-6.19789995e-02 -9.89688993e-01 -2.59483904e-01 -4.55918700e-01
-6.98055267e-01 6.31592691e-01 7.35924780e-01 -5.88320494e-01
1.52781510e+00 -2.27487659e+00 -2.13098079e-01 6.62109911e-01
3.48992944e-01 3.96451920e-01 1.82238489e-01 1.79949954e-01
-4.41104621e-01 3.84495854e-01 -6.37084901e-01 3.33879203e-01
-2.24442512e-01 1.79665178e-01 -8.09418499e-01 6.78000987e-01
2.96630323e-01 1.11964083e+00 -7.76602328e-01 6.98292349e-03
-2.27608383e-01 4.40550029e-01 -4.59980816e-01 4.65204865e-01
3.44348222e-01 8.94083679e-01 -4.93614256e-01 6.33175552e-01
6.50599182e-01 2.89634883e-01 2.99479216e-01 -2.32874408e-01
7.55617559e-01 -1.05992444e-01 -1.04590535e+00 1.30195951e+00
-3.80041562e-02 2.84424782e-01 -3.26787144e-01 -9.97001410e-01
9.03927207e-01 7.01524734e-01 5.42144895e-01 -5.99494465e-02
2.81009227e-01 1.85154155e-01 4.04273033e-01 -2.89604247e-01
-2.94619620e-01 -2.09912643e-01 -5.33592999e-01 4.67594922e-01
-3.00082061e-02 -2.85549283e-01 -6.22882724e-01 7.26209506e-02
1.49904358e+00 -1.38292566e-01 2.67892808e-01 -2.61141509e-01
4.89076555e-01 -3.74650270e-01 8.28110754e-01 1.04944956e+00
-4.15223747e-01 8.69008005e-01 5.00811815e-01 -8.53449941e-01
-5.66937625e-01 -1.36091971e+00 -2.97143489e-01 2.64511943e-01
-2.49541342e-01 -1.83524385e-01 -9.83422697e-01 -1.09723914e+00
3.08448151e-02 3.22657257e-01 -7.32419074e-01 -9.72016454e-01
-5.51009595e-01 -1.30147851e+00 1.77382922e+00 6.60938919e-01
1.85557872e-01 -9.83865619e-01 -7.66328871e-01 3.79018158e-01
-1.82664827e-01 -1.00334334e+00 -4.22131181e-01 3.16726774e-01
-7.56140053e-01 -1.24919713e+00 -5.69695354e-01 -4.30278897e-01
6.32494748e-01 -8.81564140e-01 1.35538638e+00 4.74593453e-02
-8.39061022e-01 3.29744637e-01 -3.84170204e-01 -1.16098678e+00
-1.07438672e+00 -3.16491514e-01 3.97092551e-01 3.60801250e-01
9.16648880e-02 -7.41447330e-01 -7.16717660e-01 2.94517696e-01
-9.74588156e-01 -7.57242560e-01 1.59889862e-01 9.27623391e-01
4.37130868e-01 1.96319416e-01 1.04058814e+00 -1.22288358e+00
8.28843832e-01 -4.08054978e-01 -5.25752455e-02 4.80933767e-03
-6.01154208e-01 -3.00576776e-01 9.47386682e-01 -7.20918894e-01
-3.50869715e-01 -4.76883072e-03 -3.92565668e-01 -5.52774549e-01
-9.32533592e-02 9.58906785e-02 -2.56298333e-01 -1.85044542e-01
1.16263211e+00 2.14540452e-01 -1.57553390e-01 -2.82484535e-02
-6.66585937e-02 6.17736518e-01 8.50463092e-01 -3.91402721e-01
1.26487911e+00 3.48652959e-01 -7.38951042e-02 -2.46282443e-01
-6.54207349e-01 8.55494589e-02 -3.69359344e-01 -1.60600454e-01
6.17704451e-01 -7.40711212e-01 -7.04831302e-01 8.40662003e-01
-8.03548455e-01 -1.20161148e-02 -5.91849685e-01 1.07215650e-01
-4.96023655e-01 4.17055398e-01 -5.75455785e-01 -7.50858307e-01
-8.66933167e-01 -8.52170527e-01 7.86620200e-01 -2.82086700e-01
-5.54363906e-01 -7.42252946e-01 2.05490276e-01 -5.67410998e-02
5.64468503e-01 1.23572874e+00 9.31251764e-01 -1.03830040e+00
1.70231331e-02 -7.43169844e-01 7.43446231e-01 8.04836631e-01
3.63202095e-01 -4.56336327e-03 -1.36650670e+00 -4.94321465e-01
4.84685093e-01 -1.62050664e-01 2.79090196e-01 2.92226970e-01
1.39276230e+00 -5.01228869e-01 -3.31927568e-01 1.00141299e+00
1.08777606e+00 4.71482664e-01 9.54426944e-01 -9.38093588e-02
7.18619585e-01 -3.13374624e-02 2.10577145e-01 3.44282210e-01
-2.67345279e-01 4.77527112e-01 4.60150242e-01 -5.30159652e-01
2.29020506e-01 -9.16380137e-02 3.99252445e-01 4.61889714e-01
-1.37879029e-01 -1.17746681e-01 -1.00692022e+00 2.26546153e-01
-1.50250578e+00 -1.03836632e+00 2.01734100e-02 2.29865241e+00
1.10800219e+00 4.16694522e-01 1.12484485e-01 9.10903335e-01
4.82545018e-01 -1.39398038e-01 -7.78866947e-01 -5.57362139e-01
-3.20067078e-01 7.16768503e-01 2.80496418e-01 8.28196704e-02
-1.32595134e+00 4.67595369e-01 6.82450247e+00 4.78568703e-01
-1.23602021e+00 1.67525068e-01 9.05041933e-01 1.32451400e-01
-3.27567831e-02 -5.50300837e-01 -7.67727569e-02 6.13252759e-01
1.09823823e+00 -2.95298100e-02 2.14196518e-01 6.33376002e-01
-9.23738778e-02 8.57675791e-01 -1.15829384e+00 1.12041545e+00
1.92309663e-01 -1.15810597e+00 1.94824263e-01 -3.77589077e-01
6.55135214e-01 -3.44286472e-01 1.91590175e-01 2.61472046e-01
9.17584598e-02 -1.41829276e+00 6.40489936e-01 6.47341669e-01
1.09562707e+00 -1.04264390e+00 9.63073969e-01 1.43367425e-01
-7.86656082e-01 -1.19642176e-01 7.89498314e-02 2.11823940e-01
-2.15452537e-01 6.00344598e-01 -9.50831056e-01 5.62016129e-01
7.95340836e-01 5.02345800e-01 -5.37185252e-01 4.44599837e-01
-1.59908459e-01 1.11614585e+00 -3.31489705e-02 4.92490143e-01
-3.01253051e-01 3.38598639e-01 8.41527820e-01 1.08138525e+00
8.31153244e-02 5.70460781e-02 1.74064949e-01 7.69771695e-01
-1.74365297e-01 -1.63481951e-01 -7.76976705e-01 1.73486769e-01
3.83666515e-01 9.50213671e-01 -4.14417326e-01 -2.26701736e-01
2.76468664e-01 1.06310809e+00 -3.40862095e-01 2.79401243e-01
-1.21011126e+00 -6.45146430e-01 6.15206003e-01 9.94285718e-02
-3.30514461e-01 4.46844578e-01 -5.28242052e-01 -1.04620755e+00
2.46611938e-01 -1.63214040e+00 6.22344255e-01 -4.02802497e-01
-1.49928725e+00 1.21465755e+00 -5.17324090e-01 -1.61168051e+00
-3.73822242e-01 -2.43852273e-01 -6.07403755e-01 9.46831405e-01
-9.39098895e-01 -1.19294333e+00 -9.77882072e-02 1.03436863e+00
1.68960094e-01 -3.78997117e-01 1.45221257e+00 3.79806012e-01
-2.49170557e-01 1.22419298e+00 -3.38238716e-01 6.22299135e-01
7.94590414e-01 -1.32803965e+00 6.43521428e-01 1.08366597e+00
2.26758897e-01 4.66045588e-01 5.72969198e-01 -5.83494484e-01
-1.13321912e+00 -1.45811701e+00 2.80307174e-01 -1.08094990e+00
3.68903339e-01 -5.19596636e-01 -1.00763452e+00 7.25082278e-01
-5.97571246e-02 2.97008008e-01 1.02384329e+00 -3.20859283e-01
-5.09911180e-01 -2.36766189e-01 -1.51426172e+00 4.88744080e-01
7.67277360e-01 -6.43636346e-01 -6.33191824e-01 3.66112977e-01
4.04996544e-01 -5.89512229e-01 -1.21243131e+00 9.17300820e-01
6.36516929e-01 -8.04006457e-01 1.19665086e+00 -1.06856406e+00
1.83533967e-01 -3.55639786e-01 6.23392574e-02 -1.51157248e+00
-2.10105374e-01 -1.31077039e+00 -3.63677979e-01 9.73828375e-01
5.65544009e-01 -8.73434126e-01 3.25358570e-01 1.79606661e-01
-1.82020906e-02 -8.68457556e-01 -8.68156552e-01 -6.15232587e-01
1.34543210e-01 -6.53348267e-01 9.09304440e-01 1.26226628e+00
-7.15879723e-02 -5.10138609e-02 -6.68891907e-01 6.76919281e-01
5.73174059e-01 -2.18631715e-01 5.97114027e-01 -9.74805295e-01
-3.24641049e-01 -6.74036220e-02 -8.76407504e-01 2.14271098e-01
-1.17801890e-01 -9.33512688e-01 -1.46816701e-01 -6.28965318e-01
-1.65363625e-01 -4.65374470e-01 -7.16539085e-01 5.63187897e-01
-6.31842554e-01 9.11065519e-01 -2.62073465e-02 8.23764801e-02
1.18576668e-01 -3.88359167e-02 7.92271912e-01 -3.52724701e-01
-8.70031565e-02 4.92760628e-01 -7.54793465e-01 1.00643790e+00
8.73362541e-01 -9.70200956e-01 -2.78389543e-01 1.01393208e-01
1.91717565e-01 3.80085483e-02 5.11138618e-01 -1.26192808e+00
-1.93279609e-01 4.30995941e-01 7.29959369e-01 -1.07960194e-01
-2.50817928e-02 -9.85478699e-01 4.81051117e-01 9.04536426e-01
-3.46935570e-01 1.98545173e-01 2.71190912e-01 5.28393805e-01
-2.34472424e-01 3.38484466e-01 9.98186886e-01 1.07230671e-01
2.05262944e-01 3.64460081e-01 -4.23268706e-01 4.38248068e-01
1.18392551e+00 -1.72626331e-01 -1.77306328e-02 -2.97015250e-01
-1.16288352e+00 -1.77280411e-01 5.91881499e-02 4.95144278e-01
6.87235713e-01 -1.16503143e+00 -9.62661088e-01 7.19566226e-01
1.19151630e-01 -3.29349458e-01 4.07342046e-01 6.38611436e-01
-4.97414649e-01 -1.55171096e-01 -3.20201874e-01 -6.17173374e-01
-1.38122690e+00 5.89185357e-01 8.81193399e-01 -3.51318806e-01
-6.47004724e-01 7.15797603e-01 -1.18788831e-01 -3.47785175e-01
1.81739166e-01 -3.51940691e-01 6.49611950e-02 -3.31609398e-01
4.02632147e-01 2.35678479e-01 2.79476136e-01 -2.75238305e-01
-5.91038883e-01 4.94694591e-01 2.81264037e-01 3.29213887e-01
9.14024770e-01 3.46924514e-01 2.68350597e-02 4.68464494e-01
8.04671228e-01 2.66854882e-01 -6.72856927e-01 2.17263982e-01
-3.59080762e-01 -2.25194961e-01 -3.60403746e-01 -1.33068573e+00
-1.16675210e+00 7.08626091e-01 1.07689559e+00 5.39851010e-01
1.52692080e+00 -5.79147220e-01 7.89555132e-01 -6.37547299e-02
4.30000931e-01 -5.74570775e-01 -1.64995581e-01 -6.82197735e-02
8.61964941e-01 -8.12032819e-01 -1.43105879e-01 -1.37167260e-01
-8.58924508e-01 8.15302193e-01 1.84503406e-01 -3.05850595e-01
9.29298401e-01 6.92328334e-01 7.25710869e-01 -1.28150806e-01
-2.64631003e-01 6.77941263e-01 4.27831680e-01 9.59302545e-01
1.39931336e-01 3.55401367e-01 -1.11956954e-01 1.20153153e+00
-3.23933095e-01 -2.11493403e-01 4.68148291e-01 8.52055848e-01
4.78166133e-01 -1.12092507e+00 -5.85102022e-01 5.27198851e-01
-1.35378194e+00 -1.80166945e-01 -4.94052768e-01 4.69369382e-01
2.63418615e-01 8.00767183e-01 -3.57606947e-01 -5.29334009e-01
4.82967824e-01 3.66229475e-01 3.10865521e-01 -4.51737940e-01
-1.31863022e+00 -1.02169581e-01 -1.64075091e-01 -5.34553409e-01
-8.30899104e-02 -3.53118718e-01 -9.38334644e-01 2.52943963e-01
-1.46829411e-01 1.20851107e-01 2.49574736e-01 5.42733788e-01
5.52266181e-01 1.17136192e+00 1.07757556e+00 -3.10016960e-01
-1.05505908e+00 -8.67991745e-01 -7.72869766e-01 7.84589112e-01
5.84555387e-01 -6.49638921e-02 -5.69362462e-01 3.41331482e-01] | [14.308389663696289, 3.15895414352417] |
1140664c-bf0c-4c96-a895-69206b4bacea | calibration-tests-beyond-classification-1 | 2210.13355 | null | https://arxiv.org/abs/2210.13355v1 | https://arxiv.org/pdf/2210.13355v1.pdf | Calibration tests beyond classification | Most supervised machine learning tasks are subject to irreducible prediction errors. Probabilistic predictive models address this limitation by providing probability distributions that represent a belief over plausible targets, rather than point estimates. Such models can be a valuable tool in decision-making under uncertainty, provided that the model output is meaningful and interpretable. Calibrated models guarantee that the probabilistic predictions are neither over- nor under-confident. In the machine learning literature, different measures and statistical tests have been proposed and studied for evaluating the calibration of classification models. For regression problems, however, research has been focused on a weaker condition of calibration based on predicted quantiles for real-valued targets. In this paper, we propose the first framework that unifies calibration evaluation and tests for general probabilistic predictive models. It applies to any such model, including classification and regression models of arbitrary dimension. Furthermore, the framework generalizes existing measures and provides a more intuitive reformulation of a recently proposed framework for calibration in multi-class classification. In particular, we reformulate and generalize the kernel calibration error, its estimators, and hypothesis tests using scalar-valued kernels, and evaluate the calibration of real-valued regression problems. | ['Dave Zachariah', 'Fredrik Lindsten', 'David Widmann'] | 2022-10-21 | calibration-tests-beyond-classification | https://openreview.net/forum?id=-bxf89v3Nx | https://openreview.net/pdf?id=-bxf89v3Nx | iclr-2021-1 | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [ 2.31613070e-01 2.58128822e-01 -7.04121709e-01 -8.25089037e-01
-9.76292670e-01 -5.90152323e-01 4.57702279e-01 5.19222617e-01
-1.34189636e-01 9.02358234e-01 -2.56827712e-01 -5.19521177e-01
-7.17234433e-01 -9.53080475e-01 -5.74587882e-01 -7.00471163e-01
6.17750250e-02 5.90708792e-01 1.05976984e-01 2.57620156e-01
3.51186216e-01 5.59686840e-01 -1.37311602e+00 -8.64594951e-02
1.12967455e+00 1.15904105e+00 -4.25306320e-01 5.25074422e-01
2.17063904e-01 4.75233376e-01 -5.65091670e-01 -8.35010529e-01
-5.78169152e-02 -8.61819014e-02 -4.84202355e-01 -2.51271129e-01
2.52783060e-01 9.33323577e-02 7.76068866e-02 1.15510774e+00
1.48946783e-02 7.05289654e-03 1.42338204e+00 -1.76160514e+00
-8.64584982e-01 7.98650861e-01 -3.79893392e-01 -1.54344112e-01
6.93907440e-02 -5.43553412e-01 9.98556376e-01 -8.05802047e-01
-2.48220310e-01 1.33946311e+00 8.95636022e-01 2.33748764e-01
-1.47234392e+00 -6.07412517e-01 3.23521867e-02 3.18575799e-02
-1.36372280e+00 -4.47393693e-02 4.53334898e-01 -8.53124976e-01
1.30317241e-01 2.00378552e-01 4.12621759e-02 1.06809342e+00
4.92430389e-01 3.74534041e-01 1.26936960e+00 -6.80820346e-01
5.73251665e-01 6.37720942e-01 6.16751850e-01 1.91871181e-01
1.02808630e+00 3.17857683e-01 -3.06613714e-01 -5.10361493e-01
5.91214538e-01 -5.18624224e-02 -2.73005307e-01 -6.58361375e-01
-9.51857865e-01 1.34326911e+00 -5.96186966e-02 1.67427912e-01
-2.39806995e-01 -1.48227327e-02 1.58843949e-01 -2.93131862e-02
7.76762724e-01 3.87423337e-01 -4.84002948e-01 1.70348197e-01
-9.46568608e-01 1.40663743e-01 1.13741028e+00 8.47748578e-01
6.10941827e-01 2.29773112e-02 -3.06404650e-01 6.89479947e-01
3.40244830e-01 7.41133094e-01 1.77399114e-01 -7.58991897e-01
2.29967788e-01 1.81886077e-01 4.68492627e-01 -9.26544726e-01
-2.29435161e-01 -5.12472928e-01 -8.30779612e-01 2.89906144e-01
5.77163637e-01 -6.65059686e-02 -6.24722421e-01 1.80793202e+00
1.81436077e-01 1.00947194e-01 2.23244771e-01 6.63496494e-01
2.99552590e-01 4.89428014e-01 3.50497067e-01 -2.60910630e-01
9.29506183e-01 -3.81541371e-01 -6.63962126e-01 1.11351311e-01
4.58102256e-01 -6.58729315e-01 9.42612052e-01 5.74658632e-01
-5.96476197e-01 -5.62235236e-01 -1.08822381e+00 3.54470164e-01
-4.09047633e-01 1.72747135e-01 6.60927355e-01 1.23173416e+00
-5.76931119e-01 7.68570781e-01 -6.93881989e-01 -1.49681717e-01
1.52230233e-01 9.20376629e-02 -2.22756296e-01 -9.74935517e-02
-1.11337757e+00 1.18858135e+00 5.48682749e-01 1.37960345e-01
-4.73935038e-01 -7.27847576e-01 -8.25284898e-01 9.83121321e-02
1.59840539e-01 -3.31742048e-01 1.36035025e+00 -9.76773381e-01
-1.38164544e+00 5.82046330e-01 6.38611168e-02 -5.04262447e-01
5.41306376e-01 -2.72890449e-01 -5.66519916e-01 -1.95090950e-01
1.27259567e-02 2.86713261e-02 7.82723665e-01 -1.10992455e+00
-5.11938035e-01 -3.69717419e-01 -2.54924837e-02 -3.04309100e-01
-1.39540806e-01 -3.82563360e-02 2.20507875e-01 -7.01319218e-01
1.02598093e-01 -8.90885830e-01 -2.65289456e-01 -1.54447824e-01
-5.53754330e-01 -2.67020017e-01 3.29955459e-01 -3.05468351e-01
1.17281866e+00 -2.00964808e+00 -5.14874905e-02 5.98032773e-01
-2.36628801e-01 -1.38902977e-01 2.04702049e-01 1.93862766e-01
-3.05122346e-01 2.44374916e-01 -6.44675136e-01 8.08690116e-02
2.54450411e-01 2.50182867e-01 -7.84998834e-01 7.34375000e-01
3.60681206e-01 5.62371075e-01 -5.55451155e-01 -3.59637499e-01
2.58559018e-01 4.73283947e-01 -1.60276309e-01 1.52973428e-01
-2.47526988e-02 1.09292932e-01 -4.76335645e-01 6.39713824e-01
8.16447318e-01 -1.81986108e-01 -2.01133285e-02 -9.05085355e-02
1.87533144e-02 -1.20729759e-01 -1.37536788e+00 8.79840434e-01
-3.53261024e-01 3.84850919e-01 -4.25857931e-01 -1.39652240e+00
1.22045279e+00 1.18620902e-01 2.52857029e-01 1.55196816e-01
1.33949116e-01 2.61262566e-01 -3.63613248e-01 -2.21585557e-02
3.16943258e-01 -4.79660302e-01 -4.67818946e-01 1.45095125e-01
-2.75108889e-02 -3.41663688e-01 -1.68836206e-01 -1.60069644e-01
5.59803605e-01 2.07620889e-01 7.22406924e-01 -4.66166705e-01
4.26197648e-01 2.81242058e-02 4.71561790e-01 9.93012547e-01
-3.36714042e-03 5.05315244e-01 7.00666487e-01 5.53158764e-03
-8.59516025e-01 -1.47511518e+00 -8.48597050e-01 9.42716956e-01
-5.75726619e-03 1.76961198e-02 -5.00790536e-01 -4.87359285e-01
4.60430354e-01 1.15510023e+00 -8.28651845e-01 -3.43738824e-01
1.18104011e-01 -1.09545100e+00 4.72231477e-01 7.68875718e-01
1.32754492e-03 -3.08423072e-01 -3.46885502e-01 1.24472521e-01
8.95570144e-02 -9.13719296e-01 8.85066912e-02 4.37990427e-01
-1.00792658e+00 -1.45202398e+00 -7.05475271e-01 -2.70415425e-01
4.31360424e-01 -7.32067972e-02 1.20277095e+00 -5.01467586e-01
1.43321380e-01 6.90820932e-01 -3.08592081e-01 -6.81302249e-01
-4.69142586e-01 -2.67415494e-01 3.67798120e-01 -1.62979856e-03
4.88942504e-01 -2.48479515e-01 -1.06428377e-01 6.46748900e-01
-8.45265508e-01 -4.39132482e-01 5.70299327e-01 1.02979577e+00
7.70537555e-01 4.11208495e-02 8.52583408e-01 -1.02664196e+00
6.74467325e-01 -7.12806106e-01 -1.02486980e+00 7.00875640e-01
-9.31988239e-01 3.34760070e-01 5.36668301e-01 -6.00146413e-01
-1.06573451e+00 -9.16445181e-02 2.81155020e-01 -3.65393281e-01
-1.58833250e-01 6.88807845e-01 -2.16998950e-01 -1.48378648e-02
7.81455576e-01 -1.83498904e-01 -2.98918486e-01 -3.53468210e-01
2.88699687e-01 6.01783037e-01 5.66350579e-01 -9.97980177e-01
8.23586047e-01 1.38381734e-01 4.22618121e-01 -5.11518717e-01
-1.27156544e+00 -2.90575415e-01 -8.46725702e-01 -2.73279585e-02
5.94745517e-01 -6.91563845e-01 -6.24956310e-01 9.79891419e-02
-9.38762486e-01 1.98581144e-02 -2.61457413e-01 1.04228616e+00
-8.76852393e-01 2.48583615e-01 -5.04167639e-02 -1.40518570e+00
-3.70573364e-02 -8.79821837e-01 8.18751872e-01 1.60503581e-01
-1.48188934e-01 -1.20128608e+00 1.61258101e-01 -2.12305367e-01
2.48234123e-01 3.93634409e-01 1.00244200e+00 -1.14916968e+00
-3.92840467e-02 -5.93773723e-01 -2.79984713e-01 6.25180304e-01
-1.35954350e-01 3.52659523e-01 -1.03563726e+00 -3.10384911e-02
-4.65564989e-02 -2.57705390e-01 8.98045003e-01 7.46364534e-01
1.35232460e+00 -3.20823997e-01 -3.43043089e-01 3.53699178e-01
1.34922874e+00 1.10239223e-01 3.06266218e-01 1.86822996e-01
-6.32771151e-03 8.76538515e-01 9.50739503e-01 5.18436849e-01
7.16725364e-02 4.63751614e-01 2.63967276e-01 1.84372112e-01
7.22820818e-01 -2.60684878e-01 1.77568570e-01 2.61701584e-01
-3.96948960e-03 -1.46392763e-01 -8.15281808e-01 2.24131912e-01
-1.94338465e+00 -9.07397151e-01 -3.05724651e-01 2.84821463e+00
7.81128824e-01 1.85659409e-01 1.53588861e-01 3.85636687e-01
1.19416392e+00 -4.43654358e-01 -5.59003890e-01 -2.54537821e-01
-1.41579181e-01 4.33520555e-01 7.12253749e-01 5.48740506e-01
-1.35041296e+00 4.29662913e-01 7.41494036e+00 8.33619535e-01
-7.89444387e-01 7.07183219e-03 9.22894299e-01 4.45720345e-01
-2.91679054e-01 1.61732078e-01 -9.20445621e-01 3.95310849e-01
1.24104345e+00 -3.43893379e-01 -2.25899547e-01 1.43239045e+00
-5.70064820e-02 -3.28044146e-01 -1.38400161e+00 6.69501781e-01
-2.41754338e-01 -9.09041822e-01 -1.71516731e-01 -1.30868366e-03
5.99850237e-01 -5.57106853e-01 3.79719526e-01 4.67781454e-01
4.35156941e-01 -1.27227771e+00 7.03029633e-01 9.62775290e-01
8.15295756e-01 -9.59612787e-01 1.13188422e+00 3.03358793e-01
-6.82372272e-01 3.30160335e-02 -7.86939561e-01 6.58198772e-03
-2.36467555e-01 1.16375077e+00 -7.11498857e-01 5.42339265e-01
4.09801930e-01 4.59893197e-01 -4.44351852e-01 1.26728952e+00
-2.13863373e-01 8.74656916e-01 -2.80142844e-01 -7.31802583e-02
-1.06650665e-01 -2.85168529e-01 9.27498862e-02 1.30634153e+00
6.58849239e-01 -4.56171557e-02 1.98880713e-02 9.95912373e-01
4.24292475e-01 1.70617372e-01 -4.55081254e-01 1.26997456e-01
6.67701602e-01 8.71716857e-01 -4.86325294e-01 -2.44850650e-01
-3.99856657e-01 3.98905188e-01 2.14129910e-01 4.34266329e-01
-1.08465898e+00 -4.26868975e-01 4.90781248e-01 -1.55150384e-01
4.27828915e-02 -1.84180066e-01 -6.33941591e-01 -1.14105487e+00
-1.64869756e-01 -3.35591912e-01 6.51019394e-01 -5.25750458e-01
-1.73394060e+00 3.14775437e-01 6.29881501e-01 -1.24518323e+00
-4.41536933e-01 -1.13968372e+00 -4.64308023e-01 1.01530790e+00
-1.55621803e+00 -9.01633978e-01 -8.66644979e-02 4.70117956e-01
-1.14894226e-01 -1.15432575e-01 1.00893617e+00 -2.89350301e-01
-5.24994791e-01 6.65511310e-01 4.27510291e-01 -4.71882597e-02
9.44376528e-01 -1.41932571e+00 -2.67723113e-01 7.09230304e-01
-3.74324881e-02 5.21349490e-01 1.01409864e+00 -5.17109156e-01
-9.42431867e-01 -1.15351200e+00 6.54939473e-01 -5.62526524e-01
8.85313630e-01 6.86882883e-02 -1.02244604e+00 6.20746076e-01
-5.06730616e-01 2.56778188e-02 1.06331718e+00 4.93107170e-01
-7.26054728e-01 -2.02189639e-01 -1.30671585e+00 5.68984672e-02
2.38392726e-01 -2.90580302e-01 -7.70584643e-01 3.74077618e-01
5.14790416e-01 -9.03710574e-02 -1.35636067e+00 7.23298848e-01
7.69395113e-01 -9.80653584e-01 9.64512765e-01 -7.67370164e-01
2.19500273e-01 -1.70738414e-01 -5.81377208e-01 -1.17490208e+00
-2.76110530e-01 -1.50379017e-01 -7.61000216e-02 1.08360803e+00
5.69961667e-01 -8.51343095e-01 4.61933255e-01 9.37043965e-01
1.28359385e-02 -5.57594240e-01 -1.07919395e+00 -1.09394944e+00
5.36117315e-01 -9.98973727e-01 3.98485452e-01 8.53436053e-01
-6.58475910e-04 -9.94396284e-02 -3.84600401e-01 4.85590577e-01
9.46092665e-01 2.10727707e-01 5.80068469e-01 -1.74725842e+00
-4.04841423e-01 -4.26048934e-01 -6.52084470e-01 -5.65007806e-01
3.95476162e-01 -5.90678930e-01 6.55724332e-02 -1.11087000e+00
2.47876689e-01 -7.06731915e-01 -3.98430794e-01 3.79604697e-01
-1.61676899e-01 1.28596751e-02 -1.74210414e-01 2.63055503e-01
-1.22256294e-01 4.48331654e-01 6.70902014e-01 1.20716669e-01
7.13402778e-02 7.31903613e-01 -8.53789866e-01 8.64692926e-01
9.01383340e-01 -5.17457724e-01 -3.37920606e-01 3.48074168e-01
1.83065295e-01 1.44715279e-01 6.76099837e-01 -9.30656493e-01
7.95911029e-02 -5.99775672e-01 6.31965399e-01 -5.83424032e-01
1.72061741e-01 -8.41699660e-01 1.44953847e-01 3.12021077e-01
-5.32579362e-01 -2.31588945e-01 -4.23124712e-03 9.25639212e-01
-2.51186103e-01 -6.53296232e-01 1.01880360e+00 3.69456798e-01
-2.84218699e-01 1.70192420e-01 -1.62560627e-01 -3.00496500e-02
1.11126375e+00 -2.17168704e-01 -2.73067772e-01 -4.61700439e-01
-7.91249216e-01 -1.25313535e-01 2.75636792e-01 8.04655030e-02
5.51620722e-01 -1.27587712e+00 -7.96326756e-01 -1.30512133e-01
3.60372841e-01 -3.43884706e-01 -1.61033064e-01 7.26249695e-01
-2.06024479e-02 7.05654144e-01 8.80805701e-02 -6.95225537e-01
-7.62635350e-01 7.35950530e-01 3.25353712e-01 -7.99702294e-03
2.63070408e-02 5.30304134e-01 3.25219452e-01 -3.54362041e-01
2.92001814e-01 -6.06328368e-01 -1.47071004e-01 -1.48513764e-01
4.41074759e-01 4.65085536e-01 -1.33332461e-01 -4.59022254e-01
-3.57106626e-01 5.60616434e-01 3.72808337e-01 -1.23171978e-01
1.06335843e+00 1.28600016e-01 5.97590879e-02 9.10471737e-01
7.51992047e-01 -5.24057867e-03 -1.08789515e+00 -2.90272683e-01
4.20247644e-01 -4.24709201e-01 -1.22635193e-01 -9.20216680e-01
-3.70889932e-01 9.60476458e-01 3.11130434e-01 3.09643716e-01
9.30341721e-01 8.38525444e-02 -3.01010400e-01 4.74098116e-01
3.05016398e-01 -8.44007730e-01 -2.34753489e-01 3.62985551e-01
8.25730026e-01 -1.37642395e+00 9.89723206e-02 -5.47668695e-01
-7.34928310e-01 1.50074160e+00 4.18746352e-01 -4.08230089e-02
1.02565312e+00 2.00986505e-01 -4.29559350e-01 4.69322741e-01
-4.07392114e-01 2.91327052e-02 7.71175385e-01 9.30167794e-01
6.72396541e-01 4.79557216e-01 -3.84791911e-01 1.26909554e+00
-2.90538847e-01 -9.12308246e-02 5.49811542e-01 4.85126108e-01
-7.06342816e-01 -8.78482580e-01 -7.38876164e-01 5.00688791e-01
-4.08861339e-01 -4.22259532e-02 -1.03450589e-01 8.97382081e-01
-3.26800913e-01 1.19082987e+00 1.31763555e-02 -3.18497390e-01
2.40537047e-01 2.05460936e-01 3.40502173e-01 -4.90256995e-01
2.70327926e-01 -2.59220153e-01 2.85536535e-02 -1.83805600e-01
-6.18929029e-01 -6.92271471e-01 -7.55401850e-01 -3.17157358e-01
-6.77434444e-01 5.60094476e-01 7.94596195e-01 8.28109503e-01
-9.01917145e-02 1.76364750e-01 6.73526108e-01 -5.02878726e-01
-1.20914638e+00 -1.00832701e+00 -9.16040421e-01 -7.26624057e-02
8.68876129e-02 -1.09936953e+00 -6.58926487e-01 -1.47157803e-01] | [7.876157283782959, 4.2327656745910645] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.