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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
8c4c721f-c893-4875-b1e1-108e43810b23 | liir-at-semeval-2020-task-12-a-cross-lingual | 2005.03695 | null | https://arxiv.org/abs/2005.03695v2 | https://arxiv.org/pdf/2005.03695v2.pdf | LIIR at SemEval-2020 Task 12: A Cross-Lingual Augmentation Approach for Multilingual Offensive Language Identification | This paper presents our system entitled `LIIR' for SemEval-2020 Task 12 on Multilingual Offensive Language Identification in Social Media (OffensEval 2). We have participated in sub-task A for English, Danish, Greek, Arabic, and Turkish languages. We adapt and fine-tune the BERT and Multilingual Bert models made available by Google AI for English and non-English languages respectively. For the English language, we use a combination of two fine-tuned BERT models. For other languages we propose a cross-lingual augmentation approach in order to enrich training data and we use Multilingual BERT to obtain sentence representations. LIIR achieved rank 14/38, 18/47, 24/86, 24/54, and 25/40 in Greek, Turkish, English, Arabic, and Danish languages, respectively. | ['Marie-Francine Moens', 'Erfan Ghadery'] | 2020-05-07 | null | https://aclanthology.org/2020.semeval-1.274 | https://aclanthology.org/2020.semeval-1.274.pdf | semeval-2020 | ['abuse-detection'] | ['natural-language-processing'] | [-4.96145993e-01 -1.81603506e-01 -1.79123253e-01 -6.62234239e-03
-1.25459397e+00 -1.09028745e+00 8.36586595e-01 3.40105653e-01
-1.27281070e+00 9.35608447e-01 4.22419041e-01 -3.00588608e-01
9.26798284e-02 -3.83326471e-01 -3.71031970e-01 1.72989499e-02
-8.81469548e-02 8.03047955e-01 3.99458259e-02 -7.51642346e-01
-1.53415084e-01 5.05238712e-01 -7.69468904e-01 6.30891383e-01
6.97464883e-01 7.89536059e-01 -3.68837893e-01 6.95723534e-01
3.64960842e-02 9.44104552e-01 -4.29027677e-01 -1.11401987e+00
4.33263510e-01 -3.82682830e-02 -8.91356707e-01 -5.23513317e-01
2.26161450e-01 -2.98353285e-01 -4.06129301e-01 1.10608375e+00
7.32240617e-01 1.35142375e-02 7.44480073e-01 -7.39027500e-01
-6.69584215e-01 1.22025871e+00 -9.55967426e-01 3.64545643e-01
4.31108087e-01 -2.43886575e-01 9.48806047e-01 -1.08698940e+00
6.88230217e-01 1.38437259e+00 7.36419141e-01 6.97685540e-01
-7.92628765e-01 -9.98083711e-01 -3.53862755e-02 1.57603458e-01
-1.71112323e+00 -3.76007885e-01 5.92937112e-01 -4.80776429e-01
1.00692499e+00 3.59446585e-01 1.19267292e-01 1.56475437e+00
-2.56112128e-01 1.13708937e+00 1.29962230e+00 -5.96847713e-01
-3.48886102e-01 5.90395033e-01 4.98785943e-01 3.66330057e-01
-5.45529276e-02 2.05816299e-01 -4.96668220e-01 -1.79277360e-01
1.54983804e-01 -6.74405277e-01 2.51984298e-01 5.64839840e-01
-9.95044529e-01 1.08375049e+00 1.02206714e-01 6.55012012e-01
-4.26575691e-01 -4.02499199e-01 1.09450781e+00 3.67693931e-01
7.93399453e-01 4.83279616e-01 -6.37748063e-01 -3.01469713e-01
-7.13219464e-01 1.76624477e-01 6.89797938e-01 6.77230835e-01
4.68142450e-01 7.06069022e-02 -1.49907142e-01 1.33175254e+00
1.24023952e-01 5.87235808e-01 9.56966817e-01 -2.27803007e-01
6.94079101e-01 4.50474173e-01 -1.47144236e-02 -4.79454875e-01
-5.13210058e-01 -4.76982951e-01 -7.20773637e-01 -2.48104960e-01
8.41373205e-02 -5.03968358e-01 -5.13986588e-01 1.63577461e+00
-1.03687290e-02 -1.14772953e-01 4.64376450e-01 5.65381229e-01
1.12761474e+00 5.35329580e-01 4.85199153e-01 -1.34410441e-01
1.30464578e+00 -8.79203677e-01 -3.55563432e-01 -1.68906748e-01
7.38600969e-01 -8.02327454e-01 1.01648629e+00 4.51984167e-01
-1.06884754e+00 -6.07721090e-01 -6.93730235e-01 7.58573553e-03
-7.69673765e-01 5.69837272e-01 2.52002776e-01 6.35713339e-01
-8.03143084e-01 2.63088524e-01 -3.80850196e-01 -1.40505075e-01
-1.02756232e-01 4.49808478e-01 -6.66818976e-01 2.15427414e-01
-1.78029168e+00 1.02909744e+00 8.56507063e-01 -1.58196256e-01
-6.70834124e-01 -2.35272273e-01 -9.17807043e-01 -4.96019423e-01
-6.34352043e-02 3.68565083e-01 9.80855405e-01 -9.54869330e-01
-1.32670045e+00 1.42349541e+00 3.79661351e-01 -7.39602447e-01
6.23325288e-01 -6.47294760e-01 -9.58713889e-01 -2.17217609e-01
3.30842227e-01 6.68504655e-01 1.49385542e-01 -6.33592367e-01
-6.16941333e-01 -4.10705179e-01 2.01611921e-01 4.63495344e-01
-7.80734599e-01 9.26192284e-01 -4.71400440e-01 -8.61152530e-01
-9.00138795e-01 -9.01854575e-01 -2.47230250e-02 -9.95022774e-01
-6.53682172e-01 -3.96386921e-01 5.35989463e-01 -1.49554634e+00
1.53224874e+00 -2.22549391e+00 1.00321911e-01 2.42224976e-01
-1.76669568e-01 5.85881829e-01 -2.05061883e-01 4.12801921e-01
-2.77042419e-01 1.28714621e-01 4.80406024e-02 -5.64594865e-01
-6.61615506e-02 -7.60010853e-02 -1.31809354e-01 2.57124424e-01
1.86096765e-02 7.96861231e-01 -8.92515481e-01 -4.29036558e-01
8.94085765e-02 4.66909051e-01 -3.87099594e-01 -6.85238317e-02
2.18166292e-01 3.49597186e-01 -3.17543298e-01 5.52410901e-01
3.62289190e-01 5.86063743e-01 -1.28663406e-02 -2.28938796e-02
-3.34740222e-01 3.68522674e-01 -8.99808109e-01 1.33229315e+00
-6.41955972e-01 4.51328754e-01 1.31187811e-01 -4.39664483e-01
8.56589079e-01 4.08385247e-01 5.30352116e-01 -6.57232821e-01
5.84329605e-01 5.76984584e-01 3.37606831e-03 -3.27035517e-01
8.10276687e-01 -1.68451160e-01 -5.04515886e-01 3.78185600e-01
3.68392803e-02 3.30874264e-01 5.55296183e-01 1.73231244e-01
7.06556022e-01 -2.59011649e-02 3.27150047e-01 -3.06728423e-01
1.03206837e+00 -2.86782831e-01 1.76410303e-01 4.24909174e-01
-3.12007040e-01 3.07161212e-01 1.21600591e-01 -1.47991166e-01
-7.31448770e-01 -8.58040810e-01 -2.69208729e-01 1.66823339e+00
-4.12903547e-01 -6.22768044e-01 -4.85677212e-01 -9.04838085e-01
-7.54568204e-02 9.23008263e-01 -5.73984802e-01 -1.21594243e-01
-6.68407798e-01 -9.16524172e-01 1.25203061e+00 4.02382016e-01
6.30690336e-01 -1.23747849e+00 2.41511345e-01 3.67102742e-01
-3.91482621e-01 -1.25440848e+00 -6.37384355e-01 1.08794242e-01
-6.22981228e-02 -9.75453496e-01 -6.51063681e-01 -6.76396072e-01
3.16152014e-02 -5.91353118e-01 9.90393221e-01 -4.25890297e-01
-1.11002535e-01 1.42951071e-01 -5.62024117e-01 -5.27476609e-01
-5.81405461e-01 4.37085867e-01 4.71418858e-01 1.72807992e-01
6.98463678e-01 1.70966133e-01 2.73993034e-02 -6.45096228e-02
-7.03628540e-01 -1.96364284e-01 1.59001321e-01 6.56585515e-01
1.77368909e-01 -2.83142030e-01 7.00010896e-01 -9.39320862e-01
9.82554734e-01 -6.40214860e-01 -3.79247576e-01 1.97040975e-01
-4.97674011e-02 -2.42348000e-01 6.07178330e-01 -6.98349059e-01
-7.21449554e-01 1.75329822e-03 -7.25580752e-01 -2.17952833e-01
1.19516909e-01 8.56729090e-01 -8.26838687e-02 1.36396900e-01
1.08531070e+00 -3.50544304e-02 -4.67107445e-01 -7.56104589e-01
3.04592788e-01 1.34633720e+00 5.43884218e-01 -6.00000203e-01
7.10240126e-01 -2.10269421e-01 -7.41163671e-01 -9.50694740e-01
-8.34071815e-01 -8.67768228e-01 -7.69866526e-01 3.64337601e-02
8.92031848e-01 -1.22671795e+00 -6.07684672e-01 7.21285701e-01
-1.07934177e+00 -3.04715127e-01 -3.49764496e-01 5.88658452e-01
1.07900230e-02 3.22561026e-01 -1.08711469e+00 -7.01786578e-01
-9.04632568e-01 -9.35512066e-01 1.07077825e+00 -1.89914986e-01
-5.01477361e-01 -1.22168899e+00 3.26691151e-01 2.35873401e-01
2.49578059e-01 1.20716229e-01 5.43261766e-01 -1.24909425e+00
6.81505024e-01 -5.36042929e-01 -3.04548621e-01 8.68460476e-01
6.44844696e-02 -2.38120586e-01 -8.20298731e-01 -4.10821199e-01
-5.44464529e-01 -7.76613832e-01 7.63357759e-01 -1.50965422e-01
7.85080552e-01 -4.68699127e-01 1.08003385e-01 3.46458226e-01
1.02412260e+00 -5.56385294e-02 2.84060687e-01 3.97374928e-01
7.60117650e-01 5.23360908e-01 5.42425036e-01 4.49740708e-01
4.62852001e-01 8.44526350e-01 -2.23579720e-01 -3.93686891e-02
8.62935409e-02 -2.18628287e-01 8.98202300e-01 8.88593614e-01
-1.17007084e-01 4.84374538e-02 -1.07933986e+00 7.12730646e-01
-1.36526954e+00 -8.11545670e-01 -3.98286693e-02 2.27164674e+00
1.32049572e+00 1.56690300e-01 6.43054187e-01 6.59550279e-02
5.36798596e-01 -1.91618174e-01 -5.79517335e-03 -7.11576164e-01
-4.71499801e-01 3.87774110e-01 8.24143767e-01 7.15841651e-01
-1.60994864e+00 1.65990174e+00 5.15696335e+00 1.31269765e+00
-1.32480943e+00 3.56387526e-01 5.94421029e-01 -2.35544308e-03
-7.91331083e-02 -4.93895382e-01 -1.24967289e+00 4.01294738e-01
1.35443068e+00 -3.80194068e-01 6.11545563e-01 7.88741112e-01
-1.51195809e-01 4.54427272e-01 -7.93352365e-01 1.01266277e+00
2.59544998e-01 -9.20687139e-01 6.80313110e-02 -8.44496116e-02
6.51336133e-01 7.09343731e-01 9.80479121e-02 1.00756359e+00
5.98163545e-01 -9.04925942e-01 7.50990868e-01 -1.11823939e-01
1.11894238e+00 -1.04177737e+00 9.56439734e-01 6.44267678e-01
-8.97098482e-01 -1.38842270e-01 -2.21321642e-01 4.25910592e-01
-5.51090613e-02 2.03743037e-02 -7.93660283e-01 5.72281063e-01
5.78840852e-01 7.63597250e-01 -9.77132678e-01 4.30831999e-01
-2.86921859e-01 6.89924896e-01 -4.94486719e-01 1.71120465e-02
5.13349056e-01 1.87684357e-01 7.63618588e-01 1.62528372e+00
5.77574922e-03 -2.84486949e-01 4.08050448e-01 5.24051078e-02
-3.99198651e-01 8.24566424e-01 -4.92816865e-01 -2.92699993e-01
8.01565349e-02 1.35117960e+00 -1.46862030e-01 -4.31004971e-01
-2.18575120e-01 1.04356825e+00 5.39988399e-01 8.39375630e-02
-8.54193449e-01 -3.13476920e-01 1.55803904e-01 -5.82206296e-04
-5.80716431e-01 -1.82096839e-01 1.10158265e-01 -1.29015732e+00
-4.08149421e-01 -1.29694331e+00 8.11430812e-01 -4.01775867e-01
-1.50186276e+00 1.27149427e+00 8.58223364e-02 -1.02640057e+00
-8.01627338e-01 -1.03824615e+00 -5.60121164e-02 1.01970315e+00
-1.27221692e+00 -1.50519848e+00 5.37661970e-01 8.65684628e-01
5.36699235e-01 -8.09864581e-01 9.64368522e-01 7.64976263e-01
-4.58465964e-01 9.40832675e-01 2.32691631e-01 7.07403719e-01
9.67522264e-01 -1.18228304e+00 2.80289620e-01 4.52787042e-01
2.45691091e-01 4.75217581e-01 3.76197934e-01 -7.48745501e-01
-7.74602890e-01 -1.35623968e+00 1.26608789e+00 -5.12215972e-01
1.17383254e+00 -4.55838531e-01 -5.96779287e-01 7.88005292e-01
2.74346322e-01 -3.82379800e-01 8.67720842e-01 4.82053429e-01
-3.82562160e-01 9.20703262e-02 -1.15715611e+00 6.39940500e-01
4.66527849e-01 -9.42539692e-01 -3.22026581e-01 7.28509009e-01
2.96072453e-01 -2.83930153e-01 -8.21684659e-01 3.42012197e-01
5.12948453e-01 -2.93806583e-01 9.35059726e-01 -8.71654928e-01
2.86905646e-01 2.52288908e-01 -2.87605584e-01 -1.39182091e+00
-1.69424042e-01 -3.71759683e-01 2.87718445e-01 1.32206130e+00
6.92572236e-01 -6.15654826e-01 5.16816497e-01 4.91592847e-02
2.70075276e-02 -4.48051929e-01 -1.06066084e+00 -7.53079057e-01
6.08497500e-01 -5.64110935e-01 9.84928980e-02 1.11509693e+00
1.32621109e-01 7.99692094e-01 -5.87904215e-01 -1.83283448e-01
3.38663876e-01 -4.74432886e-01 5.96610248e-01 -1.09122968e+00
-4.42370534e-01 -4.55425411e-01 -1.21717818e-01 -4.75177348e-01
7.82287955e-01 -1.29467189e+00 -4.58353490e-01 -9.80876267e-01
1.72994599e-01 -3.68332654e-01 -5.53533971e-01 7.44733870e-01
-1.81158736e-01 9.35780942e-01 2.15154260e-01 1.84917614e-01
-4.37206149e-01 3.95519376e-01 4.04938638e-01 -2.64212817e-01
-2.30789915e-01 -1.06680967e-01 -7.74958193e-01 8.01300406e-01
9.28636312e-01 -1.58445925e-01 1.44899830e-01 -3.99868429e-01
1.75761223e-01 -2.12562516e-01 -2.72132978e-02 -6.51174366e-01
-2.56453753e-01 2.45700821e-01 4.24842805e-01 -4.91348624e-01
4.32717115e-01 -6.91774070e-01 -3.05803031e-01 3.71018082e-01
-4.39220458e-01 2.45538205e-01 5.76543987e-01 -9.28091183e-02
-4.43974942e-01 -2.13700816e-01 8.82602274e-01 -9.48908702e-02
-5.81119716e-01 3.42465371e-01 -3.68974775e-01 2.55706251e-01
9.67917979e-01 1.06971942e-01 -1.88905567e-01 -2.41066471e-01
-1.05111682e+00 2.49894947e-01 -1.46373466e-01 8.47807884e-01
1.89734459e-01 -1.39525664e+00 -1.46853137e+00 2.13552192e-01
3.70830327e-01 -9.96606469e-01 1.39315411e-01 6.41715229e-01
-5.80394745e-01 3.95815313e-01 -1.83400750e-01 1.39918104e-01
-1.37256050e+00 3.97805005e-01 2.40967602e-01 -1.01206553e+00
1.75389066e-01 9.80862021e-01 -3.61452810e-02 -9.49674726e-01
1.34989917e-01 4.37864691e-01 -7.21011996e-01 4.03712094e-01
4.58989382e-01 3.19307595e-01 1.82062298e-01 -1.44412923e+00
-5.63763380e-01 3.12128723e-01 -5.32046437e-01 -5.09620249e-01
1.11989713e+00 1.04054995e-01 -1.58212826e-01 3.65736932e-01
1.45107818e+00 7.79477060e-01 -1.03247963e-01 -2.92881548e-01
2.79183060e-01 1.47456080e-01 -3.95874530e-02 -1.17816782e+00
-6.38763905e-01 8.44078004e-01 5.83489954e-01 -3.41421813e-02
9.14155126e-01 -9.43298265e-02 7.21894026e-01 1.90670013e-01
3.65420789e-01 -1.34344447e+00 -2.95952648e-01 1.23653603e+00
1.27634561e+00 -1.17071867e+00 -3.54736537e-01 -1.10648952e-01
-9.14588988e-01 9.05896723e-01 5.31670570e-01 -1.63704023e-01
6.24581635e-01 2.84729779e-01 1.88285589e-01 9.36428308e-02
-4.56037372e-01 -1.74649313e-01 4.81663495e-01 1.29125312e-01
8.84956837e-01 3.66356283e-01 -6.76785052e-01 9.35492456e-01
-5.28530478e-01 -2.69540757e-01 3.19100246e-02 4.49881226e-01
3.71972844e-02 -1.37491250e+00 -2.27927431e-01 2.39433840e-01
-1.24350059e+00 -6.44624054e-01 -8.95893276e-01 8.99728775e-01
1.57993212e-01 9.03201878e-01 -2.11809620e-01 -6.56729341e-01
4.34134603e-01 1.57637075e-01 1.86113030e-01 -7.23817050e-01
-1.34319270e+00 1.94268301e-01 6.73529446e-01 3.46445777e-02
5.45241870e-02 -6.58726990e-01 -9.82092977e-01 -1.60011292e-01
-1.67065829e-01 4.86099958e-01 6.94083393e-01 8.66704464e-01
-4.75578532e-02 1.54651552e-01 5.58555901e-01 -3.60376388e-01
-6.76764786e-01 -1.53695834e+00 -5.94633520e-01 4.55877483e-01
1.30689424e-03 -3.47443968e-01 -2.25255802e-01 -2.40621939e-01] | [9.742698669433594, 10.64908504486084] |
89571c1c-39bf-4e7e-a6e0-ebcf5ebcc276 | neural-pre-processing-a-learning-framework | 2303.12148 | null | https://arxiv.org/abs/2303.12148v1 | https://arxiv.org/pdf/2303.12148v1.pdf | Neural Pre-Processing: A Learning Framework for End-to-end Brain MRI Pre-processing | Head MRI pre-processing involves converting raw images to an intensity-normalized, skull-stripped brain in a standard coordinate space. In this paper, we propose an end-to-end weakly supervised learning approach, called Neural Pre-processing (NPP), for solving all three sub-tasks simultaneously via a neural network, trained on a large dataset without individual sub-task supervision. Because the overall objective is highly under-constrained, we explicitly disentangle geometric-preserving intensity mapping (skull-stripping and intensity normalization) and spatial transformation (spatial normalization). Quantitative results show that our model outperforms state-of-the-art methods which tackle only a single sub-task. Our ablation experiments demonstrate the importance of the architecture design we chose for NPP. Furthermore, NPP affords the user the flexibility to control each of these tasks at inference time. The code and model are freely-available at \url{https://github.com/Novestars/Neural-Pre-processing}. | ['Mert R. Sabuncu', 'Alan Wang', 'Xinzi He'] | 2023-03-21 | null | null | null | null | ['reconstruction', 'skull-stripping'] | ['computer-vision', 'medical'] | [ 2.46503636e-01 4.34779078e-01 -1.96783021e-02 -8.65446389e-01
-1.04886520e+00 -4.03632253e-01 5.84559977e-01 -9.73397568e-02
-8.45855355e-01 5.58996320e-01 3.59121859e-01 -1.49408847e-01
-1.37996040e-02 -3.62752885e-01 -8.07490468e-01 -6.00202084e-01
-1.52617440e-01 3.96251559e-01 -8.71758759e-02 1.85260192e-01
-2.57425122e-02 5.27569652e-01 -1.09081936e+00 1.31372735e-01
6.29056096e-01 7.27826655e-01 2.37939805e-01 5.09905159e-01
4.66703504e-01 5.69620788e-01 -1.36033352e-03 -3.84665638e-01
4.65122342e-01 -3.12734962e-01 -8.97874832e-01 -8.98024216e-02
6.08891547e-01 -5.26855528e-01 -3.31042975e-01 1.17166710e+00
7.09537208e-01 2.25293517e-01 6.08045757e-01 -1.06581044e+00
-7.51369238e-01 5.44784248e-01 -8.06764126e-01 5.32855868e-01
-1.34657934e-01 2.33281747e-01 8.75821292e-01 -1.04422486e+00
4.73956019e-01 8.52961063e-01 6.07402146e-01 4.77975786e-01
-1.45247662e+00 -5.84241927e-01 1.01813816e-01 -8.35411437e-03
-1.49643588e+00 -6.47755504e-01 7.62689412e-01 -5.57654023e-01
9.26025808e-01 1.20883197e-01 4.63932425e-01 9.20138359e-01
2.11734772e-01 6.57073081e-01 1.29585981e+00 -3.59012485e-01
2.29643866e-01 -3.83895814e-01 1.69729382e-01 7.55568564e-01
8.99287239e-02 6.05841875e-02 -3.64091605e-01 6.88162297e-02
9.87027586e-01 6.55188262e-02 -2.64910966e-01 -4.75973010e-01
-1.53101397e+00 6.36043310e-01 7.82973230e-01 2.86717415e-01
-6.08928621e-01 8.79281610e-02 2.97531068e-01 7.07386946e-03
5.81496656e-01 3.34729731e-01 -4.86521542e-01 1.27590001e-01
-1.16428220e+00 3.98266435e-01 3.07305574e-01 7.04742849e-01
5.48352838e-01 -1.01140723e-01 -3.55542213e-01 8.78696024e-01
1.91233069e-01 1.85496867e-01 4.89168048e-01 -1.00412071e+00
4.35300142e-01 1.63206026e-01 -2.97864646e-01 -4.09750372e-01
-9.96508956e-01 -6.72853887e-01 -8.72943938e-01 3.79758358e-01
6.67372942e-01 -4.10876006e-01 -1.17420256e+00 2.17587256e+00
3.97063702e-01 1.45451441e-01 -3.62581670e-01 1.26631093e+00
6.01647317e-01 2.17194125e-01 3.88885438e-01 1.37840426e-02
1.61936879e+00 -1.28693175e+00 -4.25875485e-01 -5.84814787e-01
5.91169059e-01 -5.17440259e-01 1.41459382e+00 1.62254259e-01
-1.46075916e+00 -2.80731738e-01 -9.89707291e-01 -4.75707561e-01
-2.46576726e-01 5.09433486e-02 5.78072786e-01 2.86762387e-01
-1.28627896e+00 5.15170634e-01 -1.23703063e+00 -1.94342151e-01
8.38249445e-01 6.06384695e-01 -6.36173189e-01 -8.49604011e-02
-9.55043375e-01 1.14272881e+00 2.00253263e-01 9.16677192e-02
-8.73087347e-01 -1.11884534e+00 -8.31070304e-01 -3.97804938e-02
2.88862199e-01 -9.60977614e-01 1.39536786e+00 -9.36715066e-01
-1.28060067e+00 1.26490378e+00 -8.91953632e-02 -2.17099369e-01
5.77976763e-01 -1.05129041e-01 4.49084006e-02 2.17587486e-01
2.67942369e-01 9.73323703e-01 7.88282394e-01 -9.95859563e-01
-1.41276971e-01 -7.67930746e-01 -7.22688437e-02 4.12644804e-01
-6.23507574e-02 4.48657960e-01 -6.72451854e-01 -9.54746962e-01
1.42009407e-01 -8.50964010e-01 -2.64129013e-01 2.27549493e-01
-4.27436322e-01 7.69760609e-02 2.14788005e-01 -1.03280354e+00
7.75238156e-01 -2.00214767e+00 2.90253699e-01 8.71690363e-02
4.43285048e-01 -1.93567693e-01 -3.13373595e-01 -2.82083035e-01
-6.28584504e-01 -2.02464983e-01 -7.99982846e-01 -8.03328872e-01
2.57526208e-02 -9.06483456e-02 1.83053195e-01 9.51376200e-01
1.03144191e-01 1.11352766e+00 -8.65917623e-01 -3.61416698e-01
8.20094720e-02 7.79927015e-01 -6.59012496e-01 1.21571735e-01
1.36874363e-01 7.89807975e-01 -1.51634008e-01 5.18776178e-01
6.71186447e-01 -1.11958764e-01 -1.01763502e-01 -3.97981405e-01
-9.10040364e-02 1.41244262e-01 -7.86134601e-01 2.30049801e+00
-4.46641982e-01 2.87520438e-01 4.42057878e-01 -1.03694177e+00
3.06994736e-01 1.60847977e-01 4.76163238e-01 -9.62559283e-01
3.73845279e-01 7.98043087e-02 1.20454259e-01 -3.41392845e-01
-1.13718249e-01 -4.35518533e-01 8.50790218e-02 7.82608688e-01
3.45446646e-01 -2.96398252e-02 1.18163176e-01 1.91730950e-02
1.00420356e+00 2.17378095e-01 2.87836939e-01 -5.55972934e-01
1.57423764e-01 -3.70959520e-01 5.11522472e-01 6.02601528e-01
-3.99864495e-01 1.11182785e+00 4.66982365e-01 -2.84367055e-01
-1.04291868e+00 -1.29082870e+00 -2.68547326e-01 1.44063413e+00
-3.56107771e-01 -1.71792135e-02 -1.11099720e+00 -5.75527012e-01
-1.98283210e-01 6.85138643e-01 -8.90899956e-01 -4.81127761e-03
-7.62868702e-01 -1.00430906e+00 5.75489938e-01 6.87956452e-01
1.35081530e-01 -1.02230346e+00 -6.83706164e-01 -5.66143841e-02
-8.00990313e-02 -1.14968514e+00 -7.90148199e-01 2.58757800e-01
-8.08557391e-01 -6.74535751e-01 -1.03090155e+00 -7.92835832e-01
1.03217673e+00 5.81518635e-02 1.04323566e+00 8.57399181e-02
-3.93100858e-01 1.57210872e-01 8.62225667e-02 -3.85955155e-01
2.30182514e-01 2.89544463e-01 9.57611389e-03 -8.14178139e-02
1.69909865e-01 -1.07896566e+00 -8.46487224e-01 -2.25460925e-03
-9.88663137e-01 3.71382236e-01 6.56424940e-01 4.95510876e-01
8.16342533e-01 -3.09511900e-01 4.14182454e-01 -8.93395841e-01
5.96129417e-01 -3.84280622e-01 -4.95040566e-01 1.54934764e-01
-4.28185821e-01 1.41577214e-01 3.79997194e-01 -2.77683258e-01
-9.62421536e-01 1.82446912e-01 -2.29056954e-01 -3.40692639e-01
-3.78824472e-01 5.41922987e-01 -2.16124088e-01 -4.41243313e-02
6.35133803e-01 5.78149371e-02 1.30818516e-01 -4.87273186e-01
5.36912978e-01 6.94183335e-02 1.02672994e+00 -5.73152661e-01
7.89790154e-01 6.93186581e-01 -8.58218595e-02 -3.44593227e-01
-1.03944051e+00 -7.83531442e-02 -1.28237736e+00 -1.49719343e-01
1.23862302e+00 -1.02451944e+00 -2.88745195e-01 4.90824819e-01
-1.09015572e+00 -7.79649734e-01 -2.36787990e-01 7.04474628e-01
-5.87684870e-01 1.27576783e-01 -5.55826724e-01 -1.93194076e-01
-6.49217606e-01 -1.21755743e+00 9.80880082e-01 7.55772069e-02
-3.48045290e-01 -8.78696024e-01 1.36577234e-01 5.43201387e-01
4.94938225e-01 3.63424391e-01 8.25755298e-01 -5.82871258e-01
-3.22649807e-01 -1.27614945e-01 -4.58466798e-01 1.58652931e-01
-2.90730968e-02 -4.78040099e-01 -1.08558726e+00 -4.08692300e-01
8.86041000e-02 -4.18259829e-01 8.01585078e-01 6.70170248e-01
1.42713618e+00 -1.97677687e-01 1.98853269e-01 1.19010687e+00
1.13366199e+00 -3.37365836e-01 4.56245959e-01 3.26363057e-01
8.98992240e-01 6.21744037e-01 -6.29210770e-02 1.75154835e-01
7.34549224e-01 7.14546561e-01 2.54794866e-01 -6.09588742e-01
-3.42205912e-01 1.17988549e-01 1.02200381e-01 6.84883475e-01
-1.43756226e-01 3.67777795e-01 -1.04818475e+00 4.00112331e-01
-1.67810118e+00 -7.12907910e-01 1.60411179e-01 2.09431648e+00
1.17027795e+00 1.15406707e-01 2.72466213e-01 -2.58921534e-01
4.38779384e-01 2.10269421e-01 -7.46560812e-01 -9.47295055e-02
9.14256573e-02 4.42651719e-01 6.02838635e-01 9.00046527e-01
-1.27050197e+00 9.35929656e-01 5.52869892e+00 5.21142483e-01
-1.26797414e+00 7.04561234e-01 8.61277580e-01 -6.50527418e-01
-1.15311503e-01 -2.91434854e-01 -3.58299553e-01 2.32282132e-01
8.13121676e-01 5.84081002e-03 8.63503039e-01 5.49072623e-01
3.95973712e-01 -5.39817661e-02 -1.22985971e+00 9.84541535e-01
1.78572059e-01 -1.00758183e+00 -2.39280939e-01 8.18043854e-03
4.60985094e-01 5.60958266e-01 6.58032820e-02 1.72631606e-01
8.39316696e-02 -1.14092457e+00 1.09098923e+00 5.06525755e-01
9.42282856e-01 -4.84440029e-01 3.81751686e-01 1.65178012e-02
-8.42187047e-01 1.98436394e-01 -1.03589464e-02 -5.98146990e-02
3.43113780e-01 7.02521622e-01 -3.38085651e-01 1.78200677e-01
6.56973779e-01 2.98086911e-01 -6.36936903e-01 9.07922566e-01
-4.42974299e-01 3.96562427e-01 -3.97166431e-01 7.34837294e-01
1.56748593e-01 -1.90813705e-01 2.74894536e-01 1.39048839e+00
-6.21143766e-02 4.45076048e-01 6.30672574e-02 1.15297520e+00
-2.28768647e-01 -1.57584436e-02 -2.35802725e-01 4.36017543e-01
1.76870435e-01 1.45578289e+00 -9.68964636e-01 -1.42042130e-01
-4.64713544e-01 9.74432528e-01 6.61276400e-01 5.19266129e-01
-8.95081639e-01 -3.59063327e-01 5.49014330e-01 2.15907291e-01
5.43080904e-02 -2.74898589e-01 -8.62173319e-01 -1.22924864e+00
2.23387539e-01 -6.55491829e-01 3.30502331e-01 -8.82705033e-01
-1.15789843e+00 6.76394701e-01 2.96488911e-01 -4.98845607e-01
-1.65911138e-01 -5.64061224e-01 -7.79622376e-01 1.08124638e+00
-1.66669083e+00 -1.29320586e+00 -2.23304525e-01 8.14177394e-01
3.31038207e-01 2.34137475e-01 6.51256263e-01 5.10878086e-01
-8.82160127e-01 6.69983566e-01 -3.37996930e-01 3.21820647e-01
6.86305046e-01 -1.24381018e+00 3.47543567e-01 1.19700587e+00
-1.58213437e-01 8.59612226e-01 4.01804447e-01 -4.72990125e-01
-1.14542437e+00 -1.04263854e+00 7.92798162e-01 -3.94487232e-01
6.60756588e-01 -6.53344274e-01 -1.03713119e+00 8.43070924e-01
2.68065989e-01 4.00193423e-01 7.49428213e-01 -2.99216458e-03
-3.80014330e-01 -7.49657974e-02 -1.08141220e+00 6.35210156e-01
1.15068436e+00 -6.48389280e-01 -6.85784459e-01 4.84450638e-01
5.63520312e-01 -5.70760369e-01 -8.66428494e-01 3.30093324e-01
4.21920300e-01 -7.57370114e-01 1.24409699e+00 -5.39549291e-01
4.33583081e-01 -2.28200272e-01 9.28668231e-02 -1.29903126e+00
-5.91717660e-01 -4.13230598e-01 -3.92142162e-02 9.03684080e-01
6.13414466e-01 -5.78653991e-01 5.71552157e-01 1.09533584e+00
-2.16244638e-01 -8.57473195e-01 -1.01693988e+00 -4.07111466e-01
4.57629651e-01 -4.91350383e-01 5.19281209e-01 1.15326333e+00
-9.92412586e-03 4.45952654e-01 -1.10858828e-01 2.29361221e-01
9.53006864e-01 -1.76475421e-01 2.60221332e-01 -1.00949371e+00
-2.66917050e-01 -6.62266970e-01 5.63063147e-03 -5.27048111e-01
3.49813849e-01 -1.24063134e+00 5.65513894e-02 -1.46292400e+00
4.08493042e-01 -1.83736995e-01 -4.98006046e-01 1.04081619e+00
-2.19838902e-01 5.52840590e-01 1.55932084e-01 2.29502976e-01
-3.24632108e-01 4.86876339e-01 1.29607880e+00 8.28012079e-02
-2.07187563e-01 -2.78533161e-01 -1.03834176e+00 7.87475824e-01
9.86954331e-01 -5.40039480e-01 -4.61136669e-01 -1.10459840e+00
-2.38177404e-01 -1.66067690e-01 5.47041059e-01 -8.25181007e-01
2.92221606e-01 2.41727382e-02 5.86173058e-01 -7.56376535e-02
2.32731014e-01 -6.27026081e-01 -2.49886498e-01 2.39319921e-01
-4.72773314e-01 3.36265713e-01 2.77013183e-01 -5.72293326e-02
1.21832684e-01 1.33581040e-02 1.18644309e+00 -1.52403280e-01
-3.17197949e-01 5.63242078e-01 1.37319043e-02 2.74312973e-01
8.32175314e-01 -5.96794067e-04 -1.67311490e-01 -1.55698374e-01
-9.42212641e-01 9.20083150e-02 3.65028888e-01 1.65934935e-01
4.27342057e-01 -1.08233750e+00 -8.95619392e-01 1.97443023e-01
-1.33327171e-01 2.02140614e-01 3.49413991e-01 1.30509472e+00
-5.43761671e-01 2.27798030e-01 -4.63804781e-01 -3.82130504e-01
-1.00951207e+00 3.54345381e-01 5.88921487e-01 -2.98697203e-01
-9.11487699e-01 9.49071705e-01 4.33512628e-01 -8.00011754e-01
3.90888393e-01 -3.67041767e-01 7.95669407e-02 -1.41110316e-01
6.81066990e-01 -9.23112500e-03 4.15931344e-01 -6.99826121e-01
-4.28460658e-01 3.98689151e-01 -2.68193096e-01 -4.51133788e-01
1.81612420e+00 -4.45843562e-02 -2.96078861e-01 1.84267517e-02
1.40104365e+00 -4.13644195e-01 -1.48805058e+00 -3.82520825e-01
-1.35672390e-01 -1.26903400e-01 6.85711801e-01 -1.02496183e+00
-1.44574130e+00 8.33006918e-01 6.68690145e-01 -5.37146986e-01
1.15976679e+00 1.72625668e-02 5.90369105e-01 2.82122016e-01
2.51139589e-02 -1.01188755e+00 -3.05144221e-01 4.30282891e-01
1.11462009e+00 -1.03229368e+00 1.57312915e-01 -1.78539932e-01
-6.99384034e-01 7.39633560e-01 6.85880125e-01 -1.32013246e-01
8.37978959e-01 5.06511271e-01 1.65710151e-01 -3.91696870e-01
-4.42092776e-01 -6.36115745e-02 4.55757618e-01 4.85137731e-01
5.83076775e-01 5.53035177e-03 -1.03369251e-01 6.51121259e-01
-5.21895111e-01 1.44919276e-01 2.09454969e-01 1.07856870e+00
-6.27303272e-02 -8.89116287e-01 -3.65069717e-01 5.11944771e-01
-5.36610603e-01 -3.98249954e-01 -9.37801525e-02 5.49875796e-01
5.99212274e-02 3.41713756e-01 2.43680105e-01 2.73847468e-02
4.44144547e-01 1.31294399e-01 7.90576220e-01 -6.37505889e-01
-4.82578099e-01 6.99486807e-02 -2.40179077e-01 -6.94918454e-01
-2.80929208e-01 -7.46976376e-01 -1.42891467e+00 -1.06474541e-01
2.47167334e-01 -2.58710563e-01 5.90714276e-01 1.03489244e+00
4.01209891e-01 5.17964959e-01 2.56678075e-01 -1.27851450e+00
-3.81384045e-01 -8.75499189e-01 -5.49872577e-01 4.74145859e-01
3.00455242e-01 -6.80087268e-01 -4.35853451e-02 2.49750048e-01] | [14.349220275878906, -2.2017626762390137] |
3eb92077-eef8-47de-aba4-7660dafdcbef | autodoviz-human-centered-automation-for | 2302.09688 | null | https://arxiv.org/abs/2302.09688v1 | https://arxiv.org/pdf/2302.09688v1.pdf | AutoDOViz: Human-Centered Automation for Decision Optimization | We present AutoDOViz, an interactive user interface for automated decision optimization (AutoDO) using reinforcement learning (RL). Decision optimization (DO) has classically being practiced by dedicated DO researchers where experts need to spend long periods of time fine tuning a solution through trial-and-error. AutoML pipeline search has sought to make it easier for a data scientist to find the best machine learning pipeline by leveraging automation to search and tune the solution. More recently, these advances have been applied to the domain of AutoDO, with a similar goal to find the best reinforcement learning pipeline through algorithm selection and parameter tuning. However, Decision Optimization requires significantly more complex problem specification when compared to an ML problem. AutoDOViz seeks to lower the barrier of entry for data scientists in problem specification for reinforcement learning problems, leverage the benefits of AutoDO algorithms for RL pipeline search and finally, create visualizations and policy insights in order to facilitate the typical interactive nature when communicating problem formulation and solution proposals between DO experts and domain experts. In this paper, we report our findings from semi-structured expert interviews with DO practitioners as well as business consultants, leading to design requirements for human-centered automation for DO with RL. We evaluate a system implementation with data scientists and find that they are significantly more open to engage in DO after using our proposed solution. AutoDOViz further increases trust in RL agent models and makes the automated training and evaluation process more comprehensible. As shown for other automation in ML tasks, we also conclude automation of RL for DO can benefit from user and vice-versa when the interface promotes human-in-the-loop. | ['Daniel Haehn', 'Loraine Franke', 'Elizabeth M. Daly', 'Paulito Palmes', 'Radu Marinescu', 'Inge Vejsbjerg', 'Rahul Nair', 'Werner Geyer', 'Dharmashankar Subramanian', 'Long Vu', 'Owen Cornec', 'Cole Makuch', 'Abel N. Valente', 'Shazia Afzal', 'Daniel Karl I. Weidele'] | 2023-02-19 | null | null | null | null | ['automl'] | ['methodology'] | [-3.96711826e-01 5.36910951e-01 -1.03517160e-01 -3.27610940e-01
-5.39262414e-01 -9.71806347e-01 4.08606529e-01 3.67816836e-01
-4.52752531e-01 5.71726739e-01 2.52749801e-01 -8.62439930e-01
-4.23536777e-01 -4.90043700e-01 -2.12294340e-01 -2.17829362e-01
-4.04528938e-02 7.98118472e-01 -3.72046918e-01 -2.68311024e-01
4.08034861e-01 5.15249133e-01 -1.52019131e+00 9.10839960e-02
9.60359514e-01 6.52367353e-01 2.81634718e-01 6.70396566e-01
7.64832273e-02 1.10989892e+00 -7.43227243e-01 -5.54156564e-02
5.61977863e-01 -3.05503815e-01 -9.53869879e-01 -5.08751869e-02
-5.84271140e-02 -5.51676214e-01 3.20554376e-01 5.03367841e-01
5.75032294e-01 7.47644156e-02 3.54160845e-01 -1.36398864e+00
-1.35310337e-01 6.22127354e-01 -1.17044039e-01 -3.71379033e-03
5.20941138e-01 1.20375335e+00 1.10279369e+00 -2.21615031e-01
7.24640787e-01 1.45787108e+00 3.68580014e-01 4.65439528e-01
-1.46581447e+00 -6.29745066e-01 2.54112214e-01 4.20415588e-02
-8.20725441e-01 -3.08670044e-01 5.65312266e-01 -7.94441998e-01
1.40463841e+00 6.06214516e-02 1.14560390e+00 7.31873930e-01
1.39973924e-01 8.84443879e-01 1.32762551e+00 -3.63382071e-01
8.12879562e-01 4.64679599e-01 -9.53675807e-02 8.41751993e-01
5.97831048e-02 4.64797109e-01 -3.31770241e-01 -8.01675022e-02
9.53483105e-01 -1.84840649e-01 4.32170406e-02 -5.18039286e-01
-9.32435811e-01 1.02740908e+00 1.78778589e-01 1.65499553e-01
-7.39367723e-01 1.79788917e-01 2.04262629e-01 7.87631512e-01
4.37829085e-02 1.38987947e+00 -7.01870203e-01 -6.86445355e-01
-6.20137930e-01 8.27277899e-01 1.23358870e+00 6.13046408e-01
9.51425135e-01 1.34335265e-01 -3.22303563e-01 4.04100686e-01
6.55095518e-01 2.72999942e-01 2.72825360e-01 -1.55921483e+00
2.77568847e-01 8.86443734e-01 7.43293941e-01 -7.95152366e-01
-6.28812313e-01 -5.39316773e-01 4.09339415e-03 1.19869232e+00
4.67570603e-01 -7.83929765e-01 -4.59691256e-01 1.36449862e+00
4.68835384e-01 -6.63881838e-01 2.60036021e-01 1.10278988e+00
2.51628160e-01 3.30102444e-01 2.74036527e-01 -9.56443623e-02
1.28904736e+00 -7.68185318e-01 -6.30982578e-01 -4.24250960e-01
9.52743828e-01 -4.62572783e-01 1.62016940e+00 6.95737422e-01
-9.99990225e-01 -3.32403123e-01 -1.11370671e+00 3.07016134e-01
-1.50806680e-01 -1.96924031e-01 8.44887912e-01 6.58086956e-01
-8.24866891e-01 6.89782858e-01 -9.77981448e-01 -2.87411749e-01
5.71982920e-01 4.75919127e-01 1.27667695e-01 1.56420261e-01
-8.73359621e-01 1.16188133e+00 2.81488717e-01 -2.89583057e-01
-9.74082172e-01 -9.05792236e-01 -6.31856382e-01 1.00999676e-01
7.76113153e-01 -9.28375900e-01 1.91726220e+00 -1.12661266e+00
-1.92579675e+00 3.69640857e-01 3.57538104e-01 -4.32074964e-01
6.41855359e-01 -3.11531186e-01 -4.50802036e-02 -1.88661322e-01
2.52171718e-02 6.23967052e-01 5.80313981e-01 -9.89346504e-01
-8.76995921e-01 -2.64472246e-01 4.66443151e-01 4.60446835e-01
2.49813557e-01 -9.48831066e-02 2.01328218e-01 -1.14045456e-01
-5.54090619e-01 -6.20502591e-01 -5.18769205e-01 -3.10063720e-01
1.99051738e-01 -4.76185501e-01 8.11609745e-01 -5.43994665e-01
1.50247777e+00 -1.84657824e+00 -1.13472551e-01 3.05705994e-01
5.26989698e-01 1.64894849e-01 4.16342951e-02 7.75646389e-01
1.90529212e-01 3.10142547e-01 3.18114281e-01 5.24585694e-03
5.14647603e-01 1.55820876e-01 -1.99384734e-01 6.30282462e-02
2.76335418e-01 8.05532873e-01 -1.08582008e+00 -2.22817868e-01
2.86616892e-01 -1.79628864e-01 -9.91239250e-01 6.01905227e-01
-7.52597570e-01 6.05908513e-01 -6.76070988e-01 5.01660943e-01
1.39623433e-02 -3.76582295e-01 3.08918685e-01 2.16831282e-01
-5.28316200e-01 5.18321216e-01 -1.36952007e+00 1.47846460e+00
-7.36589432e-01 4.53137130e-01 2.81238258e-01 -7.26260483e-01
9.24724817e-01 2.03964144e-01 4.88821268e-01 -7.75355399e-01
-6.65947944e-02 1.44277006e-01 3.29078376e-01 -7.56285846e-01
1.66723356e-01 1.65863231e-01 2.32724566e-02 9.91163373e-01
-2.14460984e-01 -6.42743409e-01 2.31264487e-01 -5.78401238e-02
1.33474839e+00 7.19334662e-01 5.71041405e-01 -1.55868724e-01
-9.97366756e-02 6.43181026e-01 6.44514859e-01 9.68236089e-01
-1.88736185e-01 -4.30865049e-01 6.86907053e-01 -7.40334809e-01
-9.00615692e-01 -5.98903418e-01 4.86449361e-01 1.09671938e+00
-4.07332122e-01 -5.82558513e-01 -7.90741801e-01 -8.17680776e-01
2.14418113e-01 1.14402366e+00 -3.48995060e-01 1.98736548e-01
-4.07205075e-01 1.98188238e-02 2.69379258e-01 4.53906097e-02
6.38914883e-01 -1.32578242e+00 -1.72302449e+00 6.02855206e-01
3.24981332e-01 -5.03328979e-01 -2.00732619e-01 3.75425398e-01
-7.59033620e-01 -1.03365028e+00 -1.66136518e-01 -1.95642069e-01
2.40568370e-01 -1.18850455e-01 1.20685887e+00 -7.29089379e-02
-4.53381002e-01 9.07391250e-01 -3.14213514e-01 -8.12311649e-01
-7.70115137e-01 4.20150720e-02 -2.13810131e-02 -7.43995786e-01
3.87761742e-01 -6.03970051e-01 -8.19132447e-01 5.64919293e-01
-6.21393919e-01 3.34497869e-01 5.45186639e-01 5.21027982e-01
1.49761677e-01 -9.37428325e-02 6.94325984e-01 -7.77558506e-01
1.31603694e+00 -3.91096443e-01 -1.05569935e+00 2.55463332e-01
-1.41312170e+00 3.55775267e-01 5.60745656e-01 -2.59289503e-01
-8.69592011e-01 7.99073651e-02 3.14890921e-01 -3.03474545e-01
-3.37867290e-01 6.02681816e-01 2.06985563e-01 1.96836904e-01
9.67997730e-01 -5.05243480e-01 4.92555410e-01 -2.33437121e-01
4.52990800e-01 6.03237629e-01 -1.51067063e-01 -7.53442049e-01
4.68639046e-01 -1.60937116e-01 -4.55004841e-01 -4.08950478e-01
-2.69844532e-01 -7.12283477e-02 -1.44761994e-01 -3.61889303e-01
7.51319051e-01 -5.61627328e-01 -1.45841026e+00 -1.50355220e-01
-7.65348315e-01 -1.17284715e+00 -6.82347894e-01 3.89716566e-01
-8.73182476e-01 -2.84643829e-01 3.79915200e-02 -9.94112432e-01
-3.43369365e-01 -1.33890963e+00 4.43161011e-01 5.02496660e-01
-8.35819602e-01 -1.05353904e+00 2.31952772e-01 3.75745356e-01
7.20606208e-01 1.80054590e-01 1.01351678e+00 -7.57533669e-01
-6.15533233e-01 8.94450471e-02 1.68377608e-01 -9.45157092e-03
2.32025176e-01 1.52771743e-02 -8.13688040e-01 -1.54699594e-01
-1.33078113e-01 -5.28256297e-01 -2.88901255e-02 2.01856911e-01
7.58599937e-01 -7.40861475e-01 -8.08031112e-02 2.29232997e-01
1.07149851e+00 5.77246249e-01 1.43612619e-04 7.68130898e-01
1.63004458e-01 8.71150374e-01 1.27487266e+00 7.71290362e-01
4.32237655e-01 6.10493243e-01 3.16160738e-01 -2.65841961e-01
2.88504243e-01 -4.28014427e-01 4.23737139e-01 -2.66033020e-02
-9.70313195e-05 1.41926050e-01 -1.24358737e+00 1.30779803e-01
-2.22715187e+00 -7.64511883e-01 5.31003475e-01 2.00080109e+00
8.05934191e-01 9.71089378e-02 6.40456080e-01 -2.44650483e-01
4.39185016e-02 -1.90694094e-01 -1.04603052e+00 -8.17675352e-01
5.99598169e-01 2.34498680e-01 3.94486904e-01 7.24889636e-01
-6.65272713e-01 1.10938406e+00 5.98970985e+00 4.39332426e-01
-1.02000594e+00 -1.95108742e-01 5.83490551e-01 -1.81841806e-01
-3.76594365e-01 3.78007680e-01 -6.27997696e-01 -5.86866848e-02
8.04769158e-01 -2.17154980e-01 1.02352715e+00 1.01321805e+00
1.00834000e+00 -3.96268547e-01 -1.53741610e+00 9.33606088e-01
-7.48179317e-01 -1.53487122e+00 -4.12690163e-01 2.28470996e-01
4.65291202e-01 -1.74104124e-01 -1.68548934e-02 5.55401146e-01
1.22168982e+00 -1.33259356e+00 6.28623366e-01 4.75761622e-01
3.37384373e-01 -7.42418706e-01 3.42260867e-01 6.10659063e-01
-6.38413370e-01 -6.48169398e-01 1.52208984e-01 -6.45392597e-01
-1.90523818e-01 -4.77241464e-02 -1.68354070e+00 6.65747970e-02
6.04093432e-01 3.86195958e-01 -2.98982739e-01 7.62392282e-01
-3.15721124e-01 6.48875237e-01 -2.58204490e-01 -3.79001379e-01
4.19333547e-01 -2.87116617e-01 4.16397512e-01 1.04335964e+00
-2.47861922e-01 9.39522460e-02 4.94892955e-01 1.21204662e+00
6.73213959e-01 3.89336273e-02 -6.13568246e-01 -5.47622323e-01
6.51609182e-01 1.25891900e+00 -4.67520446e-01 4.36022617e-02
5.14899530e-02 4.25756633e-01 2.16368943e-01 6.24153376e-01
-3.41990441e-01 -3.76857579e-01 9.81202185e-01 2.40938008e-01
1.91647246e-01 -4.37539607e-01 -4.42593306e-01 -4.73421574e-01
-4.13427860e-01 -1.69925928e+00 2.74506569e-01 -5.92806458e-01
-8.74586225e-01 1.19431697e-01 2.01270133e-02 -9.61439490e-01
-8.54421556e-01 -3.68363172e-01 -4.39155847e-01 9.44053233e-01
-1.21716058e+00 -8.45792532e-01 -2.05226198e-01 3.80056918e-01
6.07326984e-01 -2.77019024e-01 8.13061297e-01 -1.27986267e-01
-4.41548228e-01 2.46697396e-01 -3.62858057e-01 -2.63819724e-01
6.48869991e-01 -1.36825252e+00 1.82890534e-01 2.45426744e-01
-2.06551895e-01 8.41642737e-01 9.04413044e-01 -6.83005571e-01
-1.71307707e+00 -4.68908042e-01 1.26177087e-01 -4.08255994e-01
5.78324258e-01 -1.54120475e-01 -3.90129149e-01 4.97463435e-01
3.73043597e-01 -7.34419882e-01 6.50575399e-01 1.28522754e-01
6.48650229e-02 -2.21285164e-01 -1.23232448e+00 8.29756975e-01
6.90124333e-01 -4.23721582e-01 -5.21882415e-01 1.66165605e-01
6.97508752e-01 -2.92944402e-01 -9.78848398e-01 -1.06376596e-02
5.33794582e-01 -8.32210183e-01 5.53377569e-01 -8.28199983e-01
2.56462067e-01 -5.13163507e-01 1.60660520e-01 -1.53041279e+00
-2.75111139e-01 -1.35443652e+00 6.28038943e-02 1.00963461e+00
5.21452188e-01 -7.90278614e-01 5.62408924e-01 1.13781202e+00
-8.60433578e-02 -8.58405054e-01 -3.99178714e-01 -3.72598648e-01
-2.04495519e-01 -5.51407576e-01 7.60998785e-01 8.63621294e-01
2.13220283e-01 2.71978199e-01 1.66608855e-01 6.32820129e-02
3.14816147e-01 -1.11067612e-02 1.29476869e+00 -1.19089091e+00
-9.40620065e-01 -6.64414048e-01 1.06527552e-01 -1.10029161e+00
-3.39048922e-01 -6.61608458e-01 -3.33587569e-03 -1.69881773e+00
-4.45614874e-01 -7.54536986e-01 1.59683585e-01 8.08097541e-01
1.82282642e-01 -7.72989869e-01 4.03925270e-01 3.61518234e-01
-4.90373701e-01 1.75725505e-01 1.30907595e+00 9.50676873e-02
-1.04804111e+00 1.83226749e-01 -1.02731407e+00 7.75217474e-01
9.05679405e-01 -4.26980853e-01 -6.29160047e-01 -2.46975765e-01
5.52275062e-01 2.41322100e-01 3.44211906e-01 -8.06897581e-01
2.13674083e-01 -6.71918690e-01 2.47464135e-01 -1.11972600e-01
-8.20334703e-02 -9.36758757e-01 2.43410870e-01 7.04158425e-01
-3.84413183e-01 2.94007748e-01 3.72855872e-01 1.06743045e-01
2.12263152e-01 -2.31425807e-01 3.72757852e-01 -2.55659252e-01
-5.18775702e-01 -1.47163734e-01 -8.17255199e-01 -1.97884981e-02
1.14665270e+00 -3.64359081e-01 -3.81150872e-01 -7.64923811e-01
-7.17731416e-01 8.26698482e-01 4.53625947e-01 1.48228124e-01
3.99692178e-01 -4.78515893e-01 -3.29600066e-01 1.08902946e-01
-2.23022416e-01 1.13955349e-01 -2.72895843e-01 7.38239467e-01
-6.43446624e-01 3.61156732e-01 -3.78237158e-01 -4.40358937e-01
-1.07843828e+00 2.96961308e-01 6.59392238e-01 -5.70146739e-01
-5.76845646e-01 4.36329544e-01 -2.25243658e-01 -5.65120518e-01
4.79831487e-01 -6.14729404e-01 -1.20318189e-01 2.16734692e-01
3.20620954e-01 5.63047469e-01 -2.40418732e-01 6.32624745e-01
-1.89755380e-01 1.18176028e-01 -2.22909957e-01 -6.53016984e-01
1.48078299e+00 9.29746719e-04 3.67921114e-01 4.54438716e-01
2.67831355e-01 -6.76719472e-02 -1.78725100e+00 5.87026402e-02
3.53770703e-01 -3.12868476e-01 1.69082716e-01 -1.55918860e+00
-4.20395494e-01 7.00568914e-01 7.71097481e-01 3.62788320e-01
7.59014189e-01 -1.52387142e-01 4.93722595e-02 8.92924130e-01
4.10298973e-01 -1.52313805e+00 2.56023854e-01 2.70099849e-01
1.03800845e+00 -1.15095198e+00 1.06813021e-01 2.42626727e-01
-1.05108619e+00 1.32072413e+00 7.46810853e-01 1.40322745e-01
3.61382574e-01 2.60449916e-01 4.05421227e-01 -4.62963283e-01
-1.01282978e+00 -1.05260454e-01 -2.93276161e-01 7.15476274e-01
3.03700149e-01 1.04301006e-01 -1.33692503e-01 4.46069092e-01
-3.37830156e-01 3.27999741e-01 3.07266682e-01 1.14316392e+00
-5.79156816e-01 -1.21461618e+00 -3.64452690e-01 2.90013880e-01
-1.58706531e-01 2.50853986e-01 -4.08467233e-01 8.82517457e-01
-6.25742823e-02 1.32823777e+00 -7.64641166e-02 -1.59992486e-01
4.95566845e-01 -1.38843311e-02 2.87419766e-01 -7.08427608e-01
-1.29284930e+00 1.15535319e-01 5.19462109e-01 -8.69584441e-01
-6.10622531e-03 -6.27987802e-01 -1.40926397e+00 -1.35129511e-01
1.85883299e-01 3.73464972e-01 8.86014700e-01 8.94295037e-01
8.39910150e-01 6.43686414e-01 6.08239412e-01 -6.13181889e-01
-9.75230992e-01 -7.94858932e-01 -1.80888563e-01 -1.96054101e-01
3.20140094e-01 -6.06886387e-01 -6.29611537e-02 -3.22037518e-01] | [4.185873508453369, 1.716565489768982] |
2d7b4d4c-514b-4dbd-8df0-a59a45719d24 | review-on-the-feasibility-of-adversarial | 2303.07003 | null | https://arxiv.org/abs/2303.07003v1 | https://arxiv.org/pdf/2303.07003v1.pdf | Review on the Feasibility of Adversarial Evasion Attacks and Defenses for Network Intrusion Detection Systems | Nowadays, numerous applications incorporate machine learning (ML) algorithms due to their prominent achievements. However, many studies in the field of computer vision have shown that ML can be fooled by intentionally crafted instances, called adversarial examples. These adversarial examples take advantage of the intrinsic vulnerability of ML models. Recent research raises many concerns in the cybersecurity field. An increasing number of researchers are studying the feasibility of such attacks on security systems based on ML algorithms, such as Intrusion Detection Systems (IDS). The feasibility of such adversarial attacks would be influenced by various domain-specific constraints. This can potentially increase the difficulty of crafting adversarial examples. Despite the considerable amount of research that has been done in this area, much of it focuses on showing that it is possible to fool a model using features extracted from the raw data but does not address the practical side, i.e., the reverse transformation from theory to practice. For this reason, we propose a review browsing through various important papers to provide a comprehensive analysis. Our analysis highlights some challenges that have not been addressed in the reviewed papers. | ['Wim Mees', 'Jean-Michel Dricot', 'Thibault Debatty', 'Tayeb Kenaza', 'Benjamin Cochez', 'Islam Debicha'] | 2023-03-13 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 4.35317934e-01 3.03192616e-01 -1.25546783e-01 -1.05523564e-01
-2.65546411e-01 -9.11292255e-01 9.11486506e-01 -1.81563899e-01
-2.32862964e-01 5.35053194e-01 -6.50800109e-01 -7.27556586e-01
1.71820760e-01 -7.93982327e-01 -6.95498168e-01 -7.60442376e-01
-6.16944768e-02 -1.10488690e-01 2.41977185e-01 -2.20286772e-01
3.64964157e-01 8.22512329e-01 -1.26100945e+00 1.90820634e-01
6.98758423e-01 7.66363859e-01 -4.91806090e-01 6.54941559e-01
1.82921588e-02 6.53199136e-01 -9.50653076e-01 -8.32164347e-01
6.94120407e-01 -4.98874635e-01 -3.78951728e-01 -2.50243433e-02
4.38564211e-01 -3.54736239e-01 -4.81225789e-01 1.55355728e+00
1.47751600e-01 -2.79471934e-01 3.56216848e-01 -2.08208323e+00
-5.77732325e-01 5.17996252e-01 -4.09991771e-01 -9.71891433e-02
1.44166231e-01 4.91012573e-01 6.71093524e-01 -4.25291300e-01
3.52915168e-01 1.34090567e+00 4.48377311e-01 7.57482529e-01
-8.80026400e-01 -1.08484626e+00 9.19533819e-02 1.98806062e-01
-1.18029594e+00 -1.62811577e-01 1.10638416e+00 -3.43649060e-01
5.30003130e-01 5.03387094e-01 2.90128678e-01 1.42623293e+00
3.85057032e-01 7.45397627e-01 1.41697311e+00 -6.69850469e-01
2.87253499e-01 6.35106683e-01 -2.84909401e-02 5.59731483e-01
6.48080051e-01 6.93212926e-01 9.99212712e-02 -5.34024477e-01
5.70266604e-01 -1.12983838e-01 -1.12845942e-01 -3.47775310e-01
-7.53009796e-01 1.21015835e+00 3.33918869e-01 3.51482302e-01
5.71318790e-02 -1.07180290e-01 5.60336053e-01 5.60420275e-01
5.55711500e-02 7.25527346e-01 -4.37977046e-01 3.42656821e-01
-5.55243731e-01 1.29325584e-01 1.00398433e+00 7.06998825e-01
4.12738055e-01 4.34624761e-01 5.23623288e-01 1.90526381e-01
2.39677951e-01 6.51542485e-01 1.00510053e-01 -4.88686174e-01
3.24212760e-01 4.14851934e-01 -1.78884901e-02 -1.34690690e+00
-1.51059866e-01 -1.91955090e-01 -7.98982978e-01 6.23102307e-01
6.01048768e-01 -3.87595087e-01 -5.42186916e-01 1.60368204e+00
4.07220036e-01 3.94132912e-01 2.26177037e-01 6.78353250e-01
4.00058240e-01 5.51971316e-01 1.60601303e-01 -2.30802223e-01
9.42484438e-01 -7.51153290e-01 -5.86207747e-01 -3.72001231e-01
4.77783173e-01 -8.59704196e-01 9.11887527e-01 6.11503184e-01
-5.18212795e-01 -4.21908587e-01 -1.40863597e+00 6.68038189e-01
-6.02998316e-01 -4.75356251e-01 9.36292350e-01 1.41844308e+00
-4.53793436e-01 4.86172646e-01 -6.86834097e-01 -1.63603202e-01
4.82262611e-01 4.36664969e-01 -3.20703179e-01 5.94750382e-02
-1.57893407e+00 1.02931774e+00 3.64294559e-01 -1.18863523e-01
-9.22158241e-01 -4.64042485e-01 -7.78631806e-01 -3.31884444e-01
7.72386909e-01 -3.28674793e-01 9.13623571e-01 -1.35309875e+00
-1.30089605e+00 6.99860692e-01 3.85718614e-01 -8.89824152e-01
8.20496321e-01 -2.74164557e-01 -7.26157188e-01 7.36535862e-02
-4.48947847e-01 1.28456369e-01 1.35499477e+00 -1.39410782e+00
-5.31166852e-01 -2.61881232e-01 4.58173633e-01 -4.13754702e-01
-4.86172378e-01 3.19261849e-01 1.88121319e-01 -8.65624249e-01
-3.58554333e-01 -1.22702539e+00 -4.97897714e-01 -1.04192700e-02
-7.49225378e-01 1.47897616e-01 1.24471247e+00 -2.76294112e-01
1.13842428e+00 -2.23454070e+00 -2.39471868e-01 2.77831703e-01
7.44642224e-03 9.07317936e-01 6.47508428e-02 6.22994184e-01
-1.80633366e-01 6.40497506e-01 -2.63334751e-01 2.30214268e-01
-8.71579163e-03 1.55066758e-01 -1.04626667e+00 7.54871190e-01
3.25478047e-01 7.56973982e-01 -8.45241547e-01 -2.22853586e-01
5.95050991e-01 4.32732522e-01 -3.04018676e-01 2.42657915e-01
-2.46341079e-01 4.92572278e-01 -7.84117043e-01 5.31404495e-01
8.97057354e-01 2.38020748e-01 1.98824838e-01 1.52880168e-02
3.10961813e-01 -2.31413856e-01 -1.05536914e+00 8.21311831e-01
-2.57803649e-01 5.92915952e-01 -3.99855785e-02 -1.13788295e+00
7.82168031e-01 3.00984234e-01 1.88194409e-01 -2.38721400e-01
3.69401336e-01 1.01598896e-01 3.33091915e-01 -5.02597094e-01
1.27138585e-01 -2.90050298e-01 -2.79908448e-01 5.59391320e-01
-3.83749157e-01 -1.44343182e-01 -3.13188463e-01 6.96719214e-02
9.92778301e-01 -1.47689916e-02 6.00104809e-01 2.03679815e-01
9.75110292e-01 3.70443240e-02 4.80780214e-01 1.03074551e+00
-4.07059729e-01 7.16116652e-02 4.25010979e-01 -5.18935502e-01
-1.01448143e+00 -9.86334860e-01 -1.16048940e-01 5.08684158e-01
4.63746525e-02 -3.11333984e-01 -1.02609503e+00 -1.16102839e+00
4.32520546e-02 7.77426958e-01 -5.37580132e-01 -5.64823091e-01
-5.69866002e-01 -8.06447625e-01 1.08808565e+00 1.83395237e-01
5.85020840e-01 -1.10934877e+00 -6.61760211e-01 6.72667399e-02
3.07276994e-01 -1.42794967e+00 4.52911071e-02 -2.86481291e-01
-7.94201016e-01 -1.48206484e+00 -8.35928917e-02 -3.02515507e-01
7.99376905e-01 1.48950994e-01 7.94036806e-01 4.17919278e-01
-3.47422391e-01 3.57086718e-01 -6.01361573e-01 -9.55352187e-01
-1.13139689e+00 -2.25783855e-01 3.73584896e-01 1.44703731e-01
5.20294130e-01 -3.37892741e-01 -6.62618354e-02 3.90706807e-01
-1.23568308e+00 -3.14728051e-01 6.41016424e-01 7.69709110e-01
1.76764000e-02 5.06704271e-01 8.94224644e-01 -1.19329190e+00
5.63699543e-01 -3.63606244e-01 -8.73515248e-01 3.64307284e-01
-5.82550943e-01 -1.09547518e-01 1.18916810e+00 -8.23097289e-01
-7.02605367e-01 -7.84635171e-02 -9.28817615e-02 -5.32946289e-01
-4.63931322e-01 1.18324295e-01 -6.87677979e-01 -5.99774718e-01
7.48568118e-01 9.30879042e-02 1.22303188e-01 -6.38308823e-02
3.79712403e-01 6.26184165e-01 1.32315651e-01 -5.57360768e-01
1.47684455e+00 6.16801500e-01 3.12456906e-01 -9.27439153e-01
-8.16922367e-01 1.96028978e-01 -3.42854500e-01 -1.69819936e-01
4.42313790e-01 -1.24162816e-01 -6.11750126e-01 6.75941825e-01
-9.08116937e-01 3.03735826e-02 4.17712778e-02 3.09151500e-01
-4.07260716e-01 7.67529368e-01 -3.77370566e-01 -1.08324790e+00
-1.55240804e-01 -1.27535594e+00 2.81206101e-01 1.23312727e-01
-1.27171099e-01 -1.04449725e+00 -2.37937316e-01 3.29072535e-01
2.59016454e-01 7.10305870e-01 8.69133115e-01 -1.14702129e+00
-6.37001812e-01 -7.30350018e-01 1.42436191e-01 5.99807143e-01
1.77666888e-01 3.01028132e-01 -1.28006780e+00 -5.00401020e-01
5.38230300e-01 -1.33892924e-01 4.27198529e-01 6.69659004e-02
1.03778172e+00 -6.90045059e-01 -2.81551570e-01 4.34226960e-01
1.37059522e+00 5.58894813e-01 8.12875330e-01 5.92090130e-01
6.07659519e-01 7.62996912e-01 9.00666654e-01 9.47344005e-02
-4.02246147e-01 5.92735946e-01 8.12710106e-01 4.91077751e-02
4.16501880e-01 -1.83854863e-01 4.92727578e-01 6.76889643e-02
2.91019410e-01 -1.21223912e-01 -8.49346876e-01 3.86842564e-02
-1.58376622e+00 -1.13163781e+00 -9.02013760e-03 2.30844998e+00
5.30451179e-01 6.25618577e-01 -1.02437109e-01 5.84053755e-01
8.25118363e-01 2.67765641e-01 -5.72743893e-01 -6.55834019e-01
1.77853480e-02 1.73645362e-01 4.20188069e-01 4.25675392e-01
-1.49878550e+00 1.10381925e+00 6.03295660e+00 9.17694390e-01
-1.22104514e+00 -1.66343048e-01 5.25393188e-01 3.29388916e-01
7.26989284e-03 2.45996490e-01 -6.63769543e-01 3.61666501e-01
9.06261921e-01 -3.18474799e-01 3.74452800e-01 1.15638638e+00
4.72225770e-02 2.75809586e-01 -1.02833724e+00 5.99192560e-01
2.67537802e-01 -1.06088877e+00 1.09735265e-01 2.67985851e-01
4.47701484e-01 -5.62640131e-01 4.77442801e-01 4.02360290e-01
1.92382142e-01 -1.11292303e+00 3.41532588e-01 3.17281075e-02
3.67102832e-01 -1.22302866e+00 7.21127927e-01 6.52163863e-01
-6.02101445e-01 -1.70083702e-01 -4.68041241e-01 -1.71464115e-01
-1.19322769e-01 6.18484914e-01 -9.90401268e-01 5.61415434e-01
3.33491713e-01 3.39008301e-01 -5.71261406e-01 7.17146158e-01
-5.96760750e-01 8.04197848e-01 -2.14775324e-01 6.67278096e-03
3.21023673e-01 -8.57000947e-02 6.98448658e-01 9.26983118e-01
-1.63241073e-01 -1.58830196e-01 2.73117632e-01 7.93713093e-01
1.79866925e-01 -1.02387883e-01 -1.19864583e+00 -3.65929753e-01
5.05361855e-01 1.20224643e+00 -6.33580148e-01 -4.40363511e-02
-5.47192872e-01 5.80556870e-01 -2.11211115e-01 6.33967370e-02
-1.11193085e+00 -3.29918921e-01 6.54447317e-01 -5.88319935e-02
-3.26410048e-02 -3.32924634e-01 -3.01178932e-01 -1.04150712e+00
-1.71511054e-01 -1.43853998e+00 3.55783492e-01 -2.00185940e-01
-1.52952409e+00 5.66034794e-01 -1.11142233e-01 -1.32486141e+00
-1.93857297e-01 -6.52424276e-01 -7.04922736e-01 5.81683338e-01
-1.21717644e+00 -1.17094958e+00 2.32465565e-01 4.79552865e-01
5.05425692e-01 -4.84029889e-01 8.93728375e-01 -6.75915107e-02
-7.05198169e-01 9.09649670e-01 -1.46194890e-01 3.63824219e-01
6.61980987e-01 -7.04385579e-01 5.87120175e-01 1.19705975e+00
2.75481105e-01 8.06535602e-01 9.60998952e-01 -6.51092350e-01
-1.43290055e+00 -1.04913318e+00 4.93729442e-01 -8.16715062e-01
9.26269948e-01 -3.19314957e-01 -9.13570583e-01 7.69503236e-01
4.06317972e-03 1.04704320e-01 7.47981429e-01 -3.31386089e-01
-6.99064434e-01 1.23315930e-01 -1.62866938e+00 9.30392861e-01
4.53195393e-01 -4.30618793e-01 -6.38818800e-01 1.86840713e-01
6.92926824e-01 -4.25583161e-02 -6.42344177e-01 5.22711396e-01
4.02108222e-01 -1.00469625e+00 1.25158048e+00 -9.67351913e-01
2.52352148e-01 -2.26270229e-01 -2.42347289e-02 -1.01647639e+00
2.72332817e-01 -6.19393170e-01 -3.01280349e-01 1.18246603e+00
9.63335112e-02 -1.02697885e+00 6.26098514e-01 6.81539655e-01
3.03776711e-01 -5.91764808e-01 -8.05080712e-01 -1.10836101e+00
3.42482984e-01 -7.32615471e-01 5.38535774e-01 1.19524395e+00
-1.81147709e-01 1.06974645e-02 -7.01785862e-01 5.99628866e-01
8.64338994e-01 -9.13098529e-02 1.06507075e+00 -1.06865370e+00
-2.19172850e-01 -2.94686735e-01 -6.62667036e-01 -3.09155107e-01
3.97290826e-01 -5.42976856e-01 -3.97263795e-01 -6.58468902e-01
-1.73311532e-01 -5.36805928e-01 -3.41614068e-01 3.48544389e-01
-2.30440080e-01 1.12240277e-01 5.20206511e-01 2.19528496e-01
-7.69128576e-02 6.38735741e-02 1.12028575e+00 -2.49599218e-01
1.71025127e-01 2.32357889e-01 -7.05294847e-01 1.20250332e+00
1.11203408e+00 -6.57021463e-01 -5.73108613e-01 2.01123163e-01
1.35120958e-01 -1.03791796e-01 5.65090120e-01 -9.13463295e-01
-1.27466088e-02 -6.36604369e-01 8.21116418e-02 -1.35298997e-01
2.22965911e-01 -1.23422158e+00 1.62432380e-02 8.22208047e-01
-1.75553173e-01 -1.96035936e-01 2.17326254e-01 5.99819243e-01
-1.24872029e-01 -4.76265848e-01 9.76095974e-01 -1.92550778e-01
-7.00305283e-01 2.58626133e-01 -3.23151499e-01 -3.92792672e-02
1.57371807e+00 -1.14714511e-01 -4.66434956e-01 -4.09690678e-01
-4.40757304e-01 -2.10946009e-01 6.61046207e-01 6.40525222e-01
5.37680984e-01 -9.83661592e-01 -4.99864697e-01 4.03118849e-01
-6.97987080e-02 -4.56862926e-01 1.39655635e-01 4.96789545e-01
-4.63167608e-01 4.32612807e-01 -2.80652851e-01 -2.46011376e-01
-1.37303770e+00 1.38528562e+00 3.33112031e-01 -3.08873504e-01
-6.95956588e-01 4.47212040e-01 3.27967644e-01 -2.13192880e-01
2.39995614e-01 3.57532054e-01 -1.11276336e-01 -3.36902797e-01
7.57898569e-01 2.25488842e-01 -2.33880594e-01 -4.49194133e-01
-5.48510134e-01 3.22985083e-01 -2.31963143e-01 1.34934112e-01
9.82055008e-01 2.92239755e-01 6.89290389e-02 6.13949783e-02
8.91523123e-01 1.75746873e-01 -1.05975783e+00 -6.58259541e-02
7.80935511e-02 -7.56629050e-01 -2.73258895e-01 -6.00134671e-01
-1.10501111e+00 9.93323624e-01 4.44054991e-01 6.85723007e-01
1.03815317e+00 -5.13702273e-01 7.10956454e-01 5.89446366e-01
7.86204338e-01 -7.86567450e-01 1.44945890e-01 3.56922448e-01
7.35279262e-01 -1.30306363e+00 -1.58627890e-02 -6.25900984e-01
-6.52834177e-01 1.02945971e+00 5.71657717e-01 -5.09612739e-01
7.49655962e-01 4.98766512e-01 1.98245823e-01 1.27398714e-01
-3.43102604e-01 3.44328761e-01 3.65606435e-02 9.08436358e-01
-1.30521178e-01 -5.39622130e-03 -3.10141087e-01 5.31947911e-01
-1.79749846e-01 -1.62930027e-01 8.05278778e-01 1.05114305e+00
-3.22583914e-01 -1.49332130e+00 -8.53872955e-01 1.48263767e-01
-7.71106243e-01 7.59629235e-02 -7.10918665e-01 1.13866222e+00
1.06059611e-01 1.16460657e+00 -6.23786271e-01 -6.04452789e-01
1.26170337e-01 2.66952980e-02 4.01293784e-01 -4.50572759e-01
-7.93186784e-01 -4.59344864e-01 1.03741903e-02 -3.86122137e-01
-3.08627218e-01 -4.73574340e-01 -8.69484365e-01 -6.13087773e-01
-3.76233578e-01 5.82630970e-02 6.41846120e-01 9.60221052e-01
-1.42146021e-01 3.96730125e-01 9.54309702e-01 -5.64886808e-01
-1.00649309e+00 -5.17778277e-01 -5.25252819e-01 4.12760735e-01
2.64650524e-01 -7.51800478e-01 -8.53238404e-01 -1.51621833e-01] | [5.689700603485107, 7.661377906799316] |
8251f7f5-e901-46e5-a994-a52958f0a96a | dynamic-texture-synthesis-by-incorporating | 2104.05940 | null | https://arxiv.org/abs/2104.05940v2 | https://arxiv.org/pdf/2104.05940v2.pdf | Dynamic Texture Synthesis by Incorporating Long-range Spatial and Temporal Correlations | The main challenge of dynamic texture synthesis lies in how to maintain spatial and temporal consistency in synthesized videos. The major drawback of existing dynamic texture synthesis models comes from poor treatment of the long-range texture correlation and motion information. To address this problem, we incorporate a new loss term, called the Shifted Gram loss, to capture the structural and long-range correlation of the reference texture video. Furthermore, we introduce a frame sampling strategy to exploit long-period motion across multiple frames. With these two new techniques, the application scope of existing texture synthesis models can be extended. That is, they can synthesize not only homogeneous but also structured dynamic texture patterns. Thorough experimental results are provided to demonstrate that our proposed dynamic texture synthesis model offers state-of-the-art visual performance. | ['C. -C. Jay Kuo', 'Shiyu Mou', 'Ye Wang', 'Hong-Shuo Chen', 'Bin Wang', 'Kaitai Zhang'] | 2021-04-13 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 2.07505733e-01 -4.24221098e-01 -2.80673057e-01 4.40530293e-03
-4.81724024e-01 -2.56881446e-01 5.84880948e-01 -5.97861350e-01
2.85397798e-01 5.86887777e-01 8.18271860e-02 1.33317113e-01
-9.22942981e-02 -6.21905863e-01 -4.45172817e-01 -1.12432420e+00
-4.63051982e-02 -1.46564111e-01 6.95057869e-01 -2.60296375e-01
2.24325567e-01 4.47191298e-01 -1.45914543e+00 6.04471922e-01
5.42648137e-01 1.38065410e+00 4.32455212e-01 5.68180740e-01
8.72392803e-02 9.21779931e-01 -3.02363724e-01 -2.34696418e-01
2.74436355e-01 -6.37345016e-01 -5.03567338e-01 4.96719956e-01
3.79158944e-01 -4.31796014e-01 -6.17505252e-01 1.07275796e+00
4.27575052e-01 1.34654537e-01 2.64271319e-01 -9.37147796e-01
-5.01066625e-01 2.56680399e-01 -9.37113762e-01 2.51635820e-01
3.28385711e-01 9.67344735e-03 7.92741418e-01 -1.00284338e+00
8.22369218e-01 1.51572502e+00 5.66600978e-01 2.57491767e-01
-9.90952015e-01 -5.49796462e-01 2.18268469e-01 4.99133676e-01
-1.40781915e+00 -5.55303693e-01 9.57340837e-01 -2.05888987e-01
5.90974629e-01 4.15710241e-01 5.73210418e-01 1.07059205e+00
6.95581377e-01 6.85442507e-01 1.27023864e+00 -4.13615495e-01
-1.88458040e-01 -5.21456540e-01 -3.57986003e-01 7.39350796e-01
-1.13978676e-01 3.77365172e-01 -8.04186702e-01 5.11163287e-02
1.08520818e+00 -2.32277647e-01 -4.43499058e-01 -5.86543202e-01
-1.34689510e+00 3.95067781e-01 -2.00214684e-01 2.08129644e-01
-8.38946104e-02 1.95879847e-01 5.61570525e-01 3.82381111e-01
7.73801386e-01 -8.14172029e-02 -3.41981947e-02 -3.64216298e-01
-9.90087926e-01 3.54659438e-01 5.63938439e-01 1.02474403e+00
7.21808791e-01 3.11641008e-01 -3.81094873e-01 9.74527597e-01
3.13521653e-01 9.83482540e-01 2.71391362e-01 -9.95588064e-01
5.47891259e-01 -2.80640852e-02 -1.94118153e-02 -1.65026557e+00
1.07854173e-01 -1.59204781e-01 -1.09894967e+00 1.71474338e-01
2.38847971e-01 1.83937490e-01 -7.60194659e-01 1.44746912e+00
4.13437188e-01 3.96247268e-01 -3.00788105e-01 8.51220965e-01
4.96313930e-01 6.92675233e-01 -3.47673804e-01 -4.84128714e-01
1.06064665e+00 -1.20275414e+00 -1.21299613e+00 1.16575487e-01
-4.99401391e-02 -1.37336791e+00 7.81322956e-01 1.66498855e-01
-1.22011054e+00 -6.00999594e-01 -1.22098410e+00 4.74372469e-02
2.61602640e-01 -7.17331991e-02 4.18564677e-01 4.38715518e-01
-8.90812397e-01 4.71050352e-01 -1.07484043e+00 -2.10680291e-02
2.38881633e-02 1.55931816e-01 -1.24263570e-01 -2.35515207e-01
-9.71100748e-01 4.77035224e-01 4.66958247e-02 3.46564412e-01
-5.98335385e-01 -4.33536530e-01 -8.52600396e-01 6.76771551e-02
5.65048397e-01 -4.88194138e-01 1.10818934e+00 -9.92705107e-01
-1.89262724e+00 4.56906974e-01 -5.64023197e-01 -9.24833026e-03
7.68377364e-01 -1.54279873e-01 -6.12854958e-01 3.43910962e-01
5.52948080e-02 1.89357072e-01 1.07753932e+00 -1.25318635e+00
-5.43956935e-01 1.96197674e-01 -3.63041341e-01 2.58795589e-01
-1.19234964e-01 -4.78371345e-02 -9.48867798e-01 -1.36837137e+00
2.66699612e-01 -1.10185957e+00 -8.33059177e-02 1.49355620e-01
-3.34079534e-01 4.77415979e-01 1.40801692e+00 -5.61416805e-01
1.53578556e+00 -2.25506520e+00 2.71185309e-01 3.74055237e-01
1.19887404e-01 1.38776362e-01 -2.02352524e-01 4.15859818e-01
1.60101578e-01 -2.83935398e-01 1.00303471e-01 -2.46238515e-01
-1.87137976e-01 2.17914626e-01 -5.49735904e-01 3.67682666e-01
2.81211035e-03 6.87675655e-01 -8.09319437e-01 -4.13242429e-01
4.21476513e-01 4.70930636e-01 -6.42698944e-01 3.98425572e-02
-1.95610076e-01 6.55245662e-01 -6.27298594e-01 7.64177322e-01
1.00306559e+00 -2.93399066e-01 3.95350784e-01 -3.50015908e-01
-2.13616341e-01 -7.42308125e-02 -1.19519842e+00 1.53813577e+00
-2.78906792e-01 7.10583746e-01 -2.60726400e-02 -6.43047631e-01
8.28375220e-01 3.29945087e-01 7.74982989e-01 -9.79356706e-01
2.94063725e-02 3.21568429e-01 -1.17729262e-01 -4.71209705e-01
6.98677659e-01 -4.01945524e-02 1.15107074e-01 1.71076968e-01
-3.82371962e-01 1.34853095e-01 1.54549375e-01 -3.72832775e-01
7.66979814e-01 3.80347282e-01 1.06843948e-01 -4.12740409e-01
7.76724875e-01 -3.07854414e-01 8.38406861e-01 5.37555754e-01
-1.17472224e-02 9.13779557e-01 5.61966062e-01 -4.84036952e-01
-1.44943237e+00 -9.27980304e-01 2.93589104e-03 7.03889608e-01
5.25478184e-01 -4.35096115e-01 -5.10595441e-01 -3.47913533e-01
-2.49331981e-01 -2.07371488e-01 -5.98615766e-01 -5.06977225e-03
-1.07271588e+00 -6.83972776e-01 1.76922470e-01 2.55239427e-01
9.90696907e-01 -7.46548176e-01 -3.90222311e-01 3.34399164e-01
-5.27178228e-01 -1.28915584e+00 -8.52804244e-01 -3.58176857e-01
-7.08265126e-01 -1.02992260e+00 -9.75323021e-01 -7.32809424e-01
3.76424044e-01 5.96170604e-01 9.04473305e-01 1.27474844e-01
-6.15261830e-02 1.61866009e-01 -4.89362240e-01 2.44055331e-01
-2.67889649e-01 -1.72248870e-01 -9.95226130e-02 4.39687967e-01
-1.21749192e-01 -5.58590710e-01 -7.61666119e-01 1.01890349e+00
-9.31392193e-01 3.54560107e-01 2.58591026e-01 1.27696717e+00
1.01643503e+00 2.69642472e-01 2.45051414e-01 -6.32486403e-01
1.53094977e-01 -1.21225275e-01 -5.86042523e-01 4.68700886e-01
-2.81596392e-01 -1.10376589e-01 4.59477067e-01 -6.81094170e-01
-1.35671043e+00 -1.99480474e-01 -6.08909726e-02 -7.01540053e-01
5.08039474e-01 3.92471075e-01 -2.63239205e-01 -4.61079121e-01
2.92482134e-03 3.51955354e-01 1.04136616e-01 -4.28398669e-01
6.72918186e-02 1.35955170e-01 4.68101799e-01 -8.49567831e-01
7.90044606e-01 8.32738459e-01 2.70461112e-01 -9.44147348e-01
-6.60914123e-01 -1.45121649e-01 -5.23088336e-01 -2.50337809e-01
6.49098992e-01 -7.87938297e-01 -6.04832113e-01 9.55628812e-01
-9.16382730e-01 -6.15378261e-01 -1.56034259e-02 5.02381861e-01
-8.88476849e-01 8.35927725e-01 -6.75403237e-01 -3.97796869e-01
-6.00524992e-02 -1.48887849e+00 1.15514433e+00 -9.09587070e-02
6.26401529e-02 -1.00022757e+00 -6.78943982e-03 -4.68648411e-02
6.14894927e-01 2.67140567e-01 8.50957632e-01 4.05968219e-01
-1.05407739e+00 1.76918253e-01 -3.38194668e-01 2.35262409e-01
3.82986963e-01 1.59160510e-01 -5.76694369e-01 -4.56543952e-01
2.16467097e-01 3.63607630e-02 6.75814629e-01 4.02290672e-01
1.23012912e+00 -2.73068488e-01 -2.16835827e-01 9.80023682e-01
1.33660161e+00 4.54556823e-01 1.01071560e+00 4.99602526e-01
8.67405236e-01 4.53342646e-01 9.95089710e-01 5.29621124e-01
4.63623926e-02 1.28313565e+00 -1.24281026e-01 1.18571520e-03
-5.17812371e-01 -6.94387481e-02 2.76564211e-01 1.12764668e+00
-2.63851345e-01 -4.97821212e-01 -6.58014238e-01 1.62431866e-01
-2.15007138e+00 -1.18765450e+00 8.89289230e-02 1.99312675e+00
6.42459452e-01 -2.30621099e-02 -2.37568632e-01 2.01573715e-01
7.85594642e-01 6.71604335e-01 -2.36723855e-01 -1.86612591e-01
-5.83117783e-01 5.25525175e-02 4.25068438e-01 4.78876561e-01
-1.17779052e+00 1.00617337e+00 7.04389668e+00 1.40592897e+00
-1.44493175e+00 -1.62119702e-01 7.48996317e-01 -3.50620598e-02
-1.82820901e-01 -1.21039033e-01 -6.39106333e-01 6.29408419e-01
4.41362828e-01 -2.14280516e-01 3.48942369e-01 5.69681346e-01
4.26537335e-01 -2.22014397e-01 -6.83135331e-01 1.09133172e+00
1.31463960e-01 -1.32076442e+00 7.82133788e-02 1.71771094e-01
9.94770646e-01 -5.02289772e-01 4.68547672e-01 -1.97452173e-01
-1.26381978e-01 -8.19432974e-01 8.79785955e-01 5.96193254e-01
1.25062644e+00 -7.89818645e-01 4.58623141e-01 -2.81485379e-01
-1.75584114e+00 2.34040156e-01 -3.76352370e-01 1.52435958e-01
4.28124845e-01 5.76853216e-01 2.24923529e-02 8.31809998e-01
6.99202299e-01 1.06581688e+00 -3.48485112e-01 9.71230507e-01
3.78714092e-02 2.56223887e-01 -8.53146315e-02 3.63176942e-01
1.81696743e-01 -2.92669594e-01 7.01999009e-01 1.02903926e+00
7.57009685e-01 -8.57134536e-02 4.17735904e-01 2.72131354e-01
4.12893236e-01 -2.06549168e-02 -5.32028854e-01 2.63950586e-01
3.34152132e-01 8.30832779e-01 -7.26236880e-01 -2.90032804e-01
-5.85321546e-01 1.12319684e+00 -1.10954151e-01 5.65834045e-01
-8.58987868e-01 -1.34919852e-01 7.65685856e-01 2.20420033e-01
6.53472602e-01 -5.59588969e-01 -3.43324065e-01 -1.33748925e+00
2.37676859e-01 -1.25502038e+00 1.48052191e-02 -5.37856579e-01
-1.10946465e+00 4.58294302e-01 -3.10725868e-02 -1.56479919e+00
-2.73031682e-01 -3.82788986e-01 -1.75743625e-01 7.69002557e-01
-1.33061528e+00 -1.42001021e+00 -3.99622321e-01 6.98696494e-01
7.69505024e-01 -2.83474982e-01 4.50071871e-01 6.27735019e-01
-2.88658470e-01 6.66848063e-01 3.45665574e-01 -1.50150418e-01
8.68911564e-01 -6.06833637e-01 2.63994306e-01 8.78285050e-01
-3.04819643e-01 4.06334341e-01 6.59013450e-01 -8.10650468e-01
-1.49793112e+00 -1.03995800e+00 6.40130043e-01 -4.29993160e-02
6.82192385e-01 -3.16268444e-01 -9.45384204e-01 4.98921901e-01
1.01153888e-01 1.83652505e-01 2.36956909e-01 -3.44535351e-01
-2.63703614e-01 -1.74054936e-01 -7.49643981e-01 9.33091760e-01
1.07146180e+00 -5.52106977e-01 -3.57761011e-02 -1.17602102e-01
3.79331142e-01 -8.37029159e-01 -6.99556112e-01 7.91466117e-01
9.26899076e-01 -1.18257916e+00 1.02940893e+00 6.87921718e-02
7.28372633e-01 -4.28571075e-01 -3.46834660e-01 -8.10050070e-01
-4.72971201e-01 -9.57833529e-01 -3.01852107e-01 1.14406097e+00
-7.98380896e-02 -3.55310529e-01 7.64071822e-01 1.10353678e-01
-1.45221323e-01 -8.35486293e-01 -9.36235189e-01 -9.30388987e-01
-2.36250162e-01 -1.40339687e-01 4.37397391e-01 9.34635341e-01
-3.37111950e-01 -1.00111902e-01 -1.17190552e+00 -3.84647027e-02
5.63090503e-01 2.53287673e-01 7.33285964e-01 -5.49697518e-01
-4.95677173e-01 -2.56135076e-01 -4.49596196e-01 -1.50042415e+00
-4.57400046e-02 -2.92103022e-01 9.65423733e-02 -9.67148781e-01
2.84235597e-01 -5.43003917e-01 -1.52331397e-01 -2.37152725e-02
-1.06821455e-01 5.12272656e-01 1.42403945e-01 3.17260206e-01
-5.38937211e-01 8.59844089e-01 1.84318304e+00 1.19414233e-01
1.58758555e-02 -2.02511728e-01 -8.24972913e-02 5.63685179e-01
3.60000461e-01 -1.72602847e-01 -5.20228982e-01 -6.14366531e-01
2.69009895e-03 3.80357802e-01 3.16626966e-01 -9.79940951e-01
1.18514389e-01 -5.19933581e-01 2.49533013e-01 -7.68735766e-01
4.06300932e-01 -6.83940768e-01 4.96682376e-01 1.70561716e-01
4.90378588e-02 3.76658767e-01 2.07137302e-01 7.84128308e-01
-6.18033469e-01 4.32172477e-01 1.00931358e+00 1.61933526e-01
-7.21375763e-01 6.02816164e-01 -5.18359125e-01 -1.21434361e-01
9.30493593e-01 -3.52248311e-01 -4.59784299e-01 -3.50499690e-01
-5.02528191e-01 1.21382596e-02 7.75739312e-01 6.32857859e-01
6.05411410e-01 -1.74882150e+00 -3.50239187e-01 3.45827073e-01
-9.83775482e-02 -2.72269756e-01 7.27862239e-01 1.01421368e+00
-8.32986176e-01 3.22865516e-01 -4.08971250e-01 -7.41163552e-01
-1.45304680e+00 5.12432098e-01 2.48441771e-01 -4.04681861e-01
-1.07334387e+00 4.47463512e-01 6.84629560e-01 2.43902892e-01
8.36643949e-02 -9.47013721e-02 1.07032642e-01 -1.87526211e-01
6.50925577e-01 2.89237231e-01 -1.12596400e-01 -8.55731487e-01
-1.11434422e-01 1.03361976e+00 3.73390689e-03 -2.56043911e-01
9.54889417e-01 -7.08795667e-01 -1.19139232e-01 3.71895432e-01
1.17096066e+00 2.98939079e-01 -1.52248657e+00 -3.46649259e-01
-2.02059701e-01 -1.09982586e+00 -1.81477055e-01 -2.83684254e-01
-1.27836716e+00 5.61475754e-01 5.76519310e-01 -1.80853054e-01
1.27706718e+00 -6.13734007e-01 1.03592706e+00 1.64732590e-01
4.70597088e-01 -1.10836112e+00 3.98989201e-01 8.93493354e-01
7.71785378e-01 -9.74045098e-01 1.21071056e-01 -9.16842997e-01
-4.25982505e-01 1.22043931e+00 4.67165619e-01 -1.18947998e-01
7.69434512e-01 4.71276313e-01 1.22024633e-01 -6.96176142e-02
-7.79718459e-01 1.16677880e-01 4.09704775e-01 4.20895070e-01
5.11699378e-01 -2.39961132e-01 -4.31463480e-01 -6.62893727e-02
7.17853010e-02 -1.16112858e-01 3.47557545e-01 8.38456750e-01
-2.02584282e-01 -1.45586050e+00 -2.82072812e-01 2.97782328e-02
-7.84894884e-01 -2.32382882e-02 1.80696398e-01 6.75390422e-01
-1.72028616e-01 7.96182275e-01 -3.68671492e-02 -4.05833602e-01
1.29803464e-01 -2.87728041e-01 6.29917622e-01 -2.40181118e-01
-6.88565290e-03 8.47416997e-01 1.97554752e-01 -8.12719345e-01
-5.78997970e-01 -5.35499454e-01 -6.65329993e-01 -7.21949339e-01
-1.11068875e-01 -1.53387606e-01 2.60427803e-01 6.95325673e-01
2.15089560e-01 6.45110667e-01 8.56456935e-01 -8.91353309e-01
-1.69156581e-01 -6.09242558e-01 -8.54003608e-01 4.50627744e-01
3.74634504e-01 -9.65398729e-01 -1.62089199e-01 2.12096974e-01] | [11.0285062789917, -0.9663817882537842] |
dfec2368-b51d-45f2-a5f3-18c15aa9370f | active-learning-with-contrastive-pre-training | 2307.02744 | null | https://arxiv.org/abs/2307.02744v1 | https://arxiv.org/pdf/2307.02744v1.pdf | Active Learning with Contrastive Pre-training for Facial Expression Recognition | Deep learning has played a significant role in the success of facial expression recognition (FER), thanks to large models and vast amounts of labelled data. However, obtaining labelled data requires a tremendous amount of human effort, time, and financial resources. Even though some prior works have focused on reducing the need for large amounts of labelled data using different unsupervised methods, another promising approach called active learning is barely explored in the context of FER. This approach involves selecting and labelling the most representative samples from an unlabelled set to make the best use of a limited 'labelling budget'. In this paper, we implement and study 8 recent active learning methods on three public FER datasets, FER13, RAF-DB, and KDEF. Our findings show that existing active learning methods do not perform well in the context of FER, likely suffering from a phenomenon called 'Cold Start', which occurs when the initial set of labelled samples is not well representative of the entire dataset. To address this issue, we propose contrastive self-supervised pre-training, which first learns the underlying representations based on the entire unlabelled dataset. We then follow this with the active learning methods and observe that our 2-step approach shows up to 9.2% improvement over random sampling and up to 6.7% improvement over the best existing active learning baseline without the pre-training. We will make the code for this study public upon publication at: github.com/ShuvenduRoy/ActiveFER. | ['Ali Etemad', 'Shuvendu Roy'] | 2023-07-06 | null | null | null | null | ['facial-expression-recognition', 'active-learning', 'active-learning'] | ['computer-vision', 'methodology', 'natural-language-processing'] | [ 5.00627160e-01 4.39566284e-01 -4.34215784e-01 -7.52911150e-01
-1.01206839e+00 -1.63314328e-01 6.16105556e-01 4.37964126e-02
-6.73649728e-01 8.66963208e-01 3.62907380e-01 1.34267002e-01
1.12383477e-01 -5.54825962e-01 -5.94632268e-01 -7.61665165e-01
1.93047240e-01 5.01893878e-01 7.07122982e-02 -1.78586364e-01
8.93956348e-02 4.92416531e-01 -1.54882872e+00 3.80609810e-01
5.87554395e-01 1.02747190e+00 -2.53172845e-01 2.21279174e-01
-2.23124921e-01 9.74342704e-01 -6.64810956e-01 -4.71331596e-01
1.65584251e-01 -4.37801272e-01 -1.02737033e+00 3.73936057e-01
3.92931491e-01 -3.06812018e-01 1.50189400e-01 6.40574694e-01
8.41711581e-01 1.95424035e-01 4.60876256e-01 -1.18544507e+00
-6.06654510e-02 2.52618730e-01 -6.20599151e-01 7.38883242e-02
1.40849337e-01 8.17904025e-02 9.38428223e-01 -9.77377176e-01
7.43435502e-01 9.71720457e-01 5.87358594e-01 8.40176582e-01
-1.37248600e+00 -9.06506538e-01 -4.54664882e-03 1.28119186e-01
-1.32155144e+00 -1.20496285e+00 8.00686777e-01 -3.25727463e-01
7.32414603e-01 6.17945492e-02 6.59362257e-01 1.25084877e+00
-4.50361222e-01 1.04226041e+00 1.18646181e+00 -8.04256022e-01
4.33330357e-01 2.57488489e-01 1.50611475e-01 6.01971328e-01
-6.65201396e-02 -5.99842705e-02 -7.12859750e-01 -3.86689782e-01
3.27572465e-01 -2.07373068e-01 -1.99042991e-01 -4.42260653e-01
-4.55794305e-01 1.04104185e+00 1.72625497e-01 2.78406978e-01
-4.72793221e-01 -1.23674147e-01 3.50456446e-01 1.42982617e-01
9.92082238e-01 4.22947913e-01 -3.75939131e-01 -2.85493582e-01
-1.13141501e+00 -1.33210629e-01 6.63140714e-01 4.24919218e-01
9.64936614e-01 -1.00183347e-02 1.52304977e-01 1.10942459e+00
3.34539324e-01 9.79878306e-02 2.68116534e-01 -1.01613975e+00
1.07316174e-01 7.16633677e-01 9.13328975e-02 -7.04417825e-01
-1.46931142e-01 -2.80066788e-01 -5.48970819e-01 4.39476371e-01
4.57413435e-01 -3.45565021e-01 -9.17309761e-01 1.76363420e+00
3.60649854e-01 1.57608271e-01 -1.12806372e-01 6.64379120e-01
8.52269948e-01 4.53994036e-01 4.49775755e-01 -4.38407898e-01
6.83593512e-01 -8.59090924e-01 -5.20161390e-01 -3.33821088e-01
6.88202322e-01 -6.67601824e-01 7.60813415e-01 4.75867718e-01
-8.62002671e-01 -2.70151645e-01 -8.52587581e-01 2.14407772e-01
-1.98972538e-01 7.89223835e-02 8.42836082e-01 6.10530138e-01
-9.48822558e-01 3.37026298e-01 -6.50953472e-01 -3.65377396e-01
1.16551685e+00 5.81237674e-01 -6.52315795e-01 -2.04855204e-01
-9.24890220e-01 5.89403093e-01 2.09176332e-01 1.31404445e-01
-8.24313462e-01 -6.02442265e-01 -6.27476990e-01 -2.49765620e-01
6.06447995e-01 -1.29876614e-01 1.27706146e+00 -1.74251771e+00
-1.53071725e+00 1.17354155e+00 -2.83801794e-01 -4.35365409e-01
5.67674577e-01 -4.12928283e-01 -1.12250566e-01 6.94439113e-02
-1.96572930e-01 9.00081158e-01 7.88770378e-01 -1.24971044e+00
-2.86220759e-01 -2.73160636e-01 1.80598035e-01 1.30299613e-01
-3.13596070e-01 1.98029712e-01 -3.57072294e-01 -3.07273060e-01
-3.67378145e-01 -9.94284630e-01 -2.50197172e-01 1.33971479e-02
-6.96239434e-03 -3.59034389e-01 6.78338349e-01 -3.35870057e-01
1.02821410e+00 -2.31456280e+00 -2.17930794e-01 9.82537270e-02
2.76346743e-01 6.53265119e-01 -1.08562790e-01 1.86491206e-01
-3.16892326e-01 1.91958010e-01 -2.59168476e-01 -4.73968238e-01
-2.76932776e-01 1.50623098e-01 -5.99937439e-02 4.15364832e-01
4.24418926e-01 6.59914732e-01 -9.98825371e-01 -5.71128547e-01
4.04629996e-03 5.65177023e-01 -5.90144515e-01 2.83400416e-01
-9.78862047e-02 5.46320081e-01 -1.64577514e-01 6.54784977e-01
4.35486138e-01 -3.25268120e-01 2.65772760e-01 1.19115999e-02
1.66714653e-01 1.36940390e-01 -9.89419639e-01 1.59483492e+00
-3.32450718e-01 7.14094520e-01 -4.60060872e-02 -1.25877476e+00
1.02587712e+00 5.15604317e-01 8.27191532e-01 -8.92849803e-01
2.42264152e-01 3.17823797e-01 -3.78002003e-02 -3.28235388e-01
2.05048889e-01 -3.16694170e-01 2.46198386e-01 5.16362131e-01
2.98749566e-01 3.49542797e-01 2.51790404e-01 1.67404145e-01
1.12369490e+00 2.36916304e-01 3.72124553e-01 1.34633452e-01
2.85926700e-01 -3.29470560e-02 7.58868098e-01 5.02651870e-01
-3.70957762e-01 5.41041136e-01 5.31533897e-01 -2.71661490e-01
-7.72645772e-01 -5.18446803e-01 -1.28295437e-01 1.27060127e+00
-4.07909721e-01 -4.25757468e-01 -6.97654188e-01 -1.02371252e+00
-2.99565881e-01 4.79180098e-01 -7.55697906e-01 -1.32246524e-01
-4.45897877e-01 -1.05864978e+00 3.73682350e-01 3.03028435e-01
6.24832690e-01 -1.24532676e+00 -6.70127213e-01 6.10993654e-02
-1.52051345e-01 -7.05786049e-01 1.82206705e-01 2.34442174e-01
-7.24724770e-01 -9.72741425e-01 -7.51389861e-01 -3.12836498e-01
7.20655620e-01 -7.12281093e-02 1.25957561e+00 2.04931960e-01
-2.20478460e-01 1.39411792e-01 -5.19438207e-01 -6.34740651e-01
-2.74546236e-01 2.01729909e-01 -3.38661641e-01 2.91276872e-01
5.52524090e-01 -2.69830257e-01 -5.60898542e-01 1.75653756e-01
-6.88317299e-01 9.58928652e-03 5.64925313e-01 8.93147945e-01
5.68131030e-01 -2.35372782e-01 6.92740500e-01 -1.33285463e+00
3.40800643e-01 -6.93002403e-01 -2.31714904e-01 1.02663562e-01
-5.56769967e-01 -2.13562652e-01 1.43004417e-01 -4.75714982e-01
-1.01730824e+00 3.40063214e-01 -3.19365352e-01 -2.89187491e-01
-2.88429677e-01 3.51230651e-01 -1.93340197e-01 -7.70671368e-02
8.32239270e-01 -3.84503245e-01 2.37131879e-01 -3.48169476e-01
-8.08558334e-03 9.02550638e-01 8.20283126e-03 -3.36384445e-01
3.19188982e-01 4.75770175e-01 -2.29674146e-01 -8.05632889e-01
-1.12753582e+00 -6.01894379e-01 -6.24058068e-01 -3.99439216e-01
3.51556838e-01 -8.24532926e-01 -2.33545110e-01 4.37559038e-01
-7.97605574e-01 -6.87417328e-01 -5.76581419e-01 4.08385307e-01
-3.91229033e-01 1.41817734e-01 -1.47529289e-01 -1.04834604e+00
-3.64305705e-01 -9.23277557e-01 1.05952418e+00 2.76751727e-01
-6.96882010e-01 -8.94472063e-01 3.21833700e-01 6.82050943e-01
3.86675298e-01 4.24107403e-01 4.69060183e-01 -8.02638113e-01
-1.75723165e-01 -3.53934228e-01 6.36632070e-02 5.54062366e-01
1.32998362e-01 -5.13079166e-02 -1.39112818e+00 -2.59316623e-01
-3.40441048e-01 -8.81355047e-01 9.23714757e-01 4.17168885e-02
9.13594007e-01 -1.32204175e-01 -1.78284273e-01 2.28595510e-01
1.26402211e+00 3.21676075e-01 8.81475687e-01 2.65150070e-01
3.69835407e-01 6.86800838e-01 6.99663162e-01 3.92499566e-01
-6.92370236e-02 7.01726496e-01 2.06980467e-01 -5.74377358e-01
-5.58602773e-02 2.28521917e-02 2.22388789e-01 2.68150240e-01
-2.89170980e-01 -2.10796684e-01 -1.08285809e+00 5.17078459e-01
-1.69422340e+00 -1.07668912e+00 3.19464505e-01 2.36576486e+00
1.07340229e+00 2.20611393e-01 3.78347158e-01 5.15455365e-01
4.19236779e-01 8.17551091e-02 -4.53584194e-01 -4.03648287e-01
-1.83166470e-02 7.17752159e-01 1.93537012e-01 3.65738720e-01
-1.20914602e+00 9.43414271e-01 6.02521849e+00 8.73226404e-01
-1.46422648e+00 2.74631679e-01 9.76828754e-01 -3.00676376e-01
1.59743756e-01 9.23597440e-02 -7.50258863e-01 3.50799054e-01
1.12987626e+00 2.13319823e-01 1.20781930e-02 7.78419316e-01
2.89375573e-01 -3.75827670e-01 -8.06374192e-01 1.00940335e+00
1.43995330e-01 -1.17352092e+00 -2.73211628e-01 2.14254394e-01
5.94644308e-01 -8.94562108e-04 -2.98955649e-01 2.55625367e-01
2.75753260e-01 -1.07570136e+00 5.57937205e-01 7.04016984e-01
7.97099650e-01 -7.75352716e-01 8.85991871e-01 3.10107440e-01
-6.00107789e-01 -2.85244221e-03 -2.20193550e-01 -1.66139659e-02
5.16746938e-02 6.67998850e-01 -1.06686056e+00 2.39515722e-01
5.70870757e-01 6.27460957e-01 -6.89244211e-01 1.12485361e+00
-1.78407490e-01 1.13146961e+00 -2.49954686e-01 2.03910142e-01
5.03942072e-02 1.04868867e-01 2.07199782e-01 1.05922282e+00
-1.63991153e-01 5.98366698e-03 -1.51724285e-02 4.10413563e-01
-2.61806756e-01 3.29811364e-01 -4.50051427e-01 -4.17594880e-01
1.96071103e-01 1.31165886e+00 -6.99102640e-01 -2.80728579e-01
-3.90381277e-01 7.55423129e-01 4.45855319e-01 2.86356479e-01
-5.75845957e-01 -1.59574509e-01 1.46094680e-01 3.94629985e-01
-1.33189550e-02 1.33396581e-01 1.51836500e-02 -8.14390302e-01
-2.09735081e-01 -1.08210802e+00 4.09389585e-01 -6.67959630e-01
-1.12630057e+00 5.49887121e-01 3.81115787e-02 -1.00242889e+00
-4.78776157e-01 -2.65264511e-01 -3.34545702e-01 6.72331154e-01
-1.28086233e+00 -1.02319562e+00 -3.33139181e-01 4.02320325e-01
4.83807653e-01 -2.57180274e-01 1.03759873e+00 4.44616735e-01
-5.75522125e-01 6.48817241e-01 -1.39411777e-01 3.21605951e-01
8.83247912e-01 -9.66068268e-01 -3.71953547e-02 4.91372705e-01
3.77925932e-01 5.43742120e-01 4.41196412e-01 -3.95685196e-01
-9.32780623e-01 -7.41518557e-01 9.99495924e-01 -3.22464794e-01
3.02368313e-01 -4.70034510e-01 -9.48177278e-01 5.37110627e-01
2.18286395e-01 2.94912428e-01 1.13532341e+00 3.06428373e-01
-2.15865940e-01 -2.44456917e-01 -1.21150303e+00 2.42369175e-01
1.06993508e+00 -3.44971985e-01 -2.87409991e-01 1.39761657e-01
-3.65314074e-03 -1.47356436e-01 -5.80806851e-01 7.60162175e-01
5.76814592e-01 -1.01345873e+00 6.22501254e-01 -5.62063992e-01
3.87120098e-01 1.42578557e-01 -3.04439727e-05 -1.16306221e+00
-1.14115916e-01 -5.90381622e-01 -1.76128745e-02 1.44080389e+00
5.06283343e-01 -4.81252432e-01 1.12959325e+00 5.26622772e-01
2.11043164e-01 -1.22142792e+00 -6.88609779e-01 -2.57925302e-01
-1.15594439e-01 -3.67761254e-01 3.41985732e-01 1.03272474e+00
-3.05337548e-01 5.75316489e-01 -3.04647565e-01 -4.98852521e-01
4.45215583e-01 -2.10843280e-01 9.12440240e-01 -1.40123367e+00
-2.27731004e-01 -1.08546749e-01 -4.50043947e-01 -5.37845194e-01
3.70184273e-01 -7.16243446e-01 -1.50939925e-02 -1.40935040e+00
3.51113111e-01 -6.29217505e-01 -2.63887465e-01 9.65393782e-01
-3.55502889e-02 6.97638690e-01 4.71034050e-02 3.87416601e-01
-8.43354821e-01 4.09290284e-01 8.36392403e-01 -5.80979362e-02
-1.58213452e-01 -1.61004870e-03 -7.52283514e-01 7.37480938e-01
8.95461500e-01 -4.78325099e-01 -5.66501617e-01 -1.37506545e-01
1.81838214e-01 -2.34067872e-01 2.37035334e-01 -8.02475214e-01
-1.38212308e-01 -1.45783961e-01 4.98854160e-01 -6.08378276e-02
4.60610628e-01 -6.60672069e-01 1.89201653e-01 -3.36356685e-02
-4.27669168e-01 -5.17867327e-01 2.69841403e-01 2.84130931e-01
-2.87364215e-01 -3.18272084e-01 1.02878475e+00 -2.18730778e-01
-8.69507015e-01 2.48869359e-01 -3.61021012e-01 1.76753789e-01
8.84701192e-01 -3.80118757e-01 -1.40276387e-01 -3.64583880e-01
-8.38038385e-01 -2.59252310e-01 4.44747299e-01 2.35241145e-01
2.90661484e-01 -1.04059839e+00 -6.11364305e-01 9.11400095e-02
1.32738516e-01 7.50877708e-03 1.33768348e-02 9.30782437e-01
-2.15376422e-01 8.43372494e-02 -1.71133280e-01 -4.57440585e-01
-1.52017033e+00 -2.93470174e-02 4.03970242e-01 -2.39818603e-01
-3.84702921e-01 8.59282196e-01 -7.67336935e-02 -1.95510864e-01
2.41718963e-01 5.66573739e-01 -4.28776354e-01 5.11443257e-01
3.97983015e-01 2.36968353e-01 3.63463491e-01 -9.34153616e-01
-3.03035676e-01 2.22516671e-01 -2.40051419e-01 -1.92555159e-01
1.52554357e+00 1.65460527e-01 1.06350146e-01 5.46473384e-01
1.33255219e+00 8.74928311e-02 -1.21208346e+00 -2.42310360e-01
2.25556731e-01 -5.64322770e-01 2.08961323e-01 -7.83208013e-01
-1.25717640e+00 7.33397722e-01 8.60311806e-01 -1.09456159e-01
1.33410645e+00 8.28146935e-03 4.03730571e-01 1.38804525e-01
4.00524020e-01 -1.09726346e+00 3.06330621e-01 2.56397337e-01
6.79792881e-01 -1.33806956e+00 2.18855947e-01 -3.74877304e-01
-6.86919868e-01 7.66082227e-01 4.98795182e-01 5.21782003e-02
5.92993915e-01 2.70627826e-01 2.77610272e-01 -2.54955977e-01
-8.64956558e-01 -2.87823051e-01 9.87691954e-02 3.40243042e-01
7.89665997e-01 -1.14849344e-01 -2.99192548e-01 2.43089542e-01
5.18175177e-02 5.23575366e-01 1.18115596e-01 1.30674338e+00
-2.39701495e-01 -1.34419298e+00 -4.67412136e-02 4.63229239e-01
-6.96908891e-01 1.50780872e-01 -7.86324322e-01 7.10312486e-01
3.98754984e-01 9.41925466e-01 1.69668511e-01 -1.64229378e-01
1.61422014e-01 4.38399255e-01 4.57546979e-01 -8.53704333e-01
-4.01887208e-01 8.89758319e-02 5.01627266e-01 -5.25883317e-01
-9.72769320e-01 -6.86077297e-01 -1.07075953e+00 -2.08559379e-01
-4.66478854e-01 1.67513937e-01 4.41013008e-01 1.02405262e+00
3.77340853e-01 2.03900188e-02 5.75670719e-01 -6.26127124e-01
-1.49452180e-01 -9.52143133e-01 -3.12494427e-01 6.26636267e-01
6.63874149e-02 -8.70498538e-01 -5.54831564e-01 1.08784221e-01] | [13.554035186767578, 1.6948810815811157] |
6b5e51a0-37e7-4e89-a4d4-74abdd074fb1 | dta-physical-camouflage-attacks-using | 2203.09831 | null | https://arxiv.org/abs/2203.09831v1 | https://arxiv.org/pdf/2203.09831v1.pdf | DTA: Physical Camouflage Attacks using Differentiable Transformation Network | To perform adversarial attacks in the physical world, many studies have proposed adversarial camouflage, a method to hide a target object by applying camouflage patterns on 3D object surfaces. For obtaining optimal physical adversarial camouflage, previous studies have utilized the so-called neural renderer, as it supports differentiability. However, existing neural renderers cannot fully represent various real-world transformations due to a lack of control of scene parameters compared to the legacy photo-realistic renderers. In this paper, we propose the Differentiable Transformation Attack (DTA), a framework for generating a robust physical adversarial pattern on a target object to camouflage it against object detection models with a wide range of transformations. It utilizes our novel Differentiable Transformation Network (DTN), which learns the expected transformation of a rendered object when the texture is changed while preserving the original properties of the target object. Using our attack framework, an adversary can gain both the advantages of the legacy photo-realistic renderers including various physical-world transformations and the benefit of white-box access by offering differentiability. Our experiments show that our camouflaged 3D vehicles can successfully evade state-of-the-art object detection models in the photo-realistic environment (i.e., CARLA on Unreal Engine). Furthermore, our demonstration on a scaled Tesla Model 3 proves the applicability and transferability of our method to the real world. | ['Howon Kim', 'Se-Yoon Oh', 'Hunmin Yang', 'Thi-Thu-Huong Le', 'Youngyeo Yun', 'Harashta Tatimma Larasati', 'Hyoeun Kang', 'Yongsu Kim', 'Naufal Suryanto'] | 2022-03-18 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Suryanto_DTA_Physical_Camouflage_Attacks_Using_Differentiable_Transformation_Network_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Suryanto_DTA_Physical_Camouflage_Attacks_Using_Differentiable_Transformation_Network_CVPR_2022_paper.pdf | cvpr-2022-1 | ['real-world-adversarial-attack'] | ['adversarial'] | [ 2.05780998e-01 4.71098125e-02 2.74196208e-01 4.49718773e-01
-5.46280086e-01 -1.02963245e+00 8.12804878e-01 -8.79127502e-01
-2.27565486e-02 4.30307090e-01 -4.55247819e-01 -6.82919681e-01
3.45877409e-01 -9.06670451e-01 -1.13609970e+00 -7.99118459e-01
-1.46015733e-01 -2.58573517e-02 4.72578704e-01 -3.15683335e-01
2.72984896e-02 1.10689282e+00 -1.38426566e+00 1.89068578e-02
6.52178586e-01 1.05500054e+00 -3.40437800e-01 8.00470233e-01
3.72463644e-01 5.48416317e-01 -1.04522681e+00 -4.24764931e-01
9.29082274e-01 -1.20772146e-01 -1.12144202e-01 -2.49167129e-01
7.81768918e-01 -7.49218762e-01 -1.08697426e+00 1.19171357e+00
3.18534046e-01 -1.55000448e-01 6.33628726e-01 -1.87327611e+00
-1.16575360e+00 -1.59841225e-01 -3.65696579e-01 -1.86894745e-01
3.40159327e-01 7.41021514e-01 1.15596261e-02 -4.61849481e-01
4.50056732e-01 1.57322490e+00 5.75771213e-01 1.05744946e+00
-8.82743239e-01 -1.10498238e+00 -1.58870786e-01 -1.32503197e-01
-1.37380195e+00 -9.24196616e-02 8.66030395e-01 -3.75779063e-01
7.78718114e-01 7.16109812e-01 5.55059314e-01 1.52069318e+00
9.56248522e-01 4.80443537e-01 1.40117419e+00 -5.56639917e-02
2.24787891e-01 6.06417470e-02 -4.14471507e-01 6.69180810e-01
3.54940087e-01 9.67109084e-01 -4.53483872e-02 -5.73428810e-01
1.06683850e+00 -1.42722264e-01 -3.60066175e-01 -4.82032359e-01
-8.54968190e-01 5.54850280e-01 4.82966661e-01 -5.23076057e-01
-1.27831653e-01 5.88178813e-01 3.34576070e-01 3.97916138e-01
1.92586899e-01 5.64046025e-01 -2.63703754e-03 2.53369302e-01
-1.01649798e-01 4.59917426e-01 7.30889559e-01 9.42766786e-01
4.45408583e-01 7.36255229e-01 -2.38005921e-01 -4.37277593e-02
1.93099260e-01 1.37961757e+00 -3.63092422e-02 -9.91605401e-01
2.04831183e-01 2.23724410e-01 3.67321670e-01 -1.13117671e+00
1.25220627e-01 8.45319405e-02 -6.93233490e-01 1.39452469e+00
1.58487201e-01 -3.23039860e-01 -1.18820190e+00 1.70265281e+00
7.69309163e-01 6.37019396e-01 2.26216942e-01 8.80860388e-01
6.30169511e-01 6.38837636e-01 -6.18045256e-02 4.30736035e-01
1.12127256e+00 -5.49566627e-01 -4.53364402e-01 -7.57123455e-02
2.09904656e-01 -8.16650569e-01 1.09515011e+00 2.69426078e-01
-7.19693065e-01 -5.45031726e-01 -1.44087636e+00 2.34534398e-01
-5.92980504e-01 -4.00754690e-01 6.52710736e-01 1.24714541e+00
-8.73065412e-01 2.42797807e-01 -1.00526345e+00 3.69127877e-02
5.16488671e-01 5.35266757e-01 -5.17251492e-01 2.05122277e-01
-1.36211216e+00 1.10558856e+00 -7.97873139e-02 -1.53469011e-01
-1.67346954e+00 -8.66946757e-01 -8.56982529e-01 -1.17377564e-01
5.01312673e-01 -6.38575435e-01 9.49975729e-01 -9.49978411e-01
-1.76190615e+00 7.77009547e-01 5.56405365e-01 -4.91294444e-01
7.89533079e-01 -2.69147307e-01 -6.29039884e-01 1.76600978e-01
-4.37691778e-01 5.20138085e-01 1.28216052e+00 -1.65507174e+00
-1.51657075e-01 -4.44483347e-02 6.45365715e-01 -7.97493681e-02
-2.06429034e-01 -8.64044875e-02 -6.31867796e-02 -7.39405870e-01
-5.51396430e-01 -1.32076550e+00 -7.16146082e-03 7.25413144e-01
-6.15254760e-01 3.62024128e-01 1.82674146e+00 -5.08102536e-01
4.17939782e-01 -2.44932437e+00 -3.50895435e-01 1.55000255e-01
2.30254203e-01 8.85513186e-01 -4.63293493e-01 2.29978457e-01
1.33384224e-02 1.80125251e-01 7.09973797e-02 1.41732603e-01
2.05155671e-01 2.09441796e-01 -9.85826850e-01 9.04116273e-01
3.80022138e-01 1.03097248e+00 -7.63671696e-01 1.28742745e-02
3.87644857e-01 8.08846056e-01 -5.11333644e-01 3.29884291e-01
-3.61217052e-01 5.14390647e-01 -7.84943640e-01 5.60409665e-01
1.27381492e+00 4.51636285e-01 -3.73843312e-01 1.22980580e-01
1.78252593e-01 -2.26067230e-01 -8.12484622e-01 1.03387797e+00
-4.25968230e-01 6.81451619e-01 1.52758241e-01 -9.33353975e-02
8.21693838e-01 1.98413253e-01 7.07815215e-02 -6.29966974e-01
4.35489565e-01 -2.11040918e-02 -9.40763801e-02 -2.73672968e-01
5.27415633e-01 1.03812113e-01 -3.56686205e-01 2.17234224e-01
-6.49730682e-01 -6.12411797e-01 -6.90635383e-01 1.38356343e-01
1.42296052e+00 2.39981979e-01 -1.29047498e-01 -1.81716815e-01
3.04614753e-01 1.69919955e-03 3.75405490e-01 8.90380561e-01
-2.22884670e-01 3.35093349e-01 4.00731683e-01 -4.10413444e-01
-1.21421111e+00 -1.52031577e+00 8.46105367e-02 1.66484848e-01
7.23784149e-01 1.41002610e-01 -8.46001267e-01 -9.01832700e-01
4.61968333e-01 8.77578139e-01 -8.95989060e-01 -7.43340850e-01
-6.11509502e-01 5.96757904e-02 1.25551009e+00 2.74060279e-01
7.93088555e-01 -8.38413715e-01 -8.35649669e-01 -3.47915798e-01
4.92124081e-01 -1.04778230e+00 -4.45883453e-01 -4.16764289e-01
-1.41201794e-01 -1.06616628e+00 -2.60928720e-01 -4.60069090e-01
6.15792990e-01 6.29770637e-01 5.92238545e-01 1.90885022e-01
-2.11868078e-01 5.54150999e-01 -1.02106929e-01 -6.39772594e-01
-1.09873140e+00 -5.82769632e-01 3.46371770e-01 -4.51627076e-02
-9.88812000e-02 -4.20303255e-01 -4.98408914e-01 5.95299006e-01
-1.25963449e+00 2.32484774e-03 7.04817027e-02 5.79508841e-01
3.16940039e-01 9.45332572e-02 1.72064871e-01 -6.60025775e-01
3.44880909e-01 -3.00344735e-01 -1.16311359e+00 1.85244426e-01
-2.99896616e-02 -3.06614220e-01 9.37300622e-01 -1.15543377e+00
-1.03559613e+00 -2.92739093e-01 2.65609324e-02 -1.14582038e+00
-6.06981060e-03 -2.84808934e-01 -6.48325682e-01 -9.78119314e-01
8.59075546e-01 -7.59916604e-02 1.58234015e-02 9.45239328e-03
5.05966604e-01 4.09366369e-01 8.46249878e-01 -6.22820497e-01
1.83386886e+00 9.88041222e-01 4.18100417e-01 -5.57712555e-01
-2.57933319e-01 3.89648497e-01 7.85502717e-02 -3.03014606e-01
7.23565161e-01 -6.27436519e-01 -1.16495931e+00 1.08440137e+00
-1.24622846e+00 -5.93560576e-01 -2.78499663e-01 2.07259923e-01
-6.74559832e-01 3.55192900e-01 -4.71817583e-01 -6.49746895e-01
-8.54066163e-02 -1.24028099e+00 1.10361886e+00 1.67513743e-01
9.30709988e-02 -6.74276114e-01 -3.73729877e-02 -1.37844440e-02
5.71774840e-01 1.03458238e+00 8.69057536e-01 -2.59177327e-01
-1.06959653e+00 -5.17475009e-01 -4.34196331e-02 3.46171379e-01
2.21821859e-01 8.04890394e-02 -1.09193254e+00 -6.27593875e-01
3.46193254e-01 -2.67272800e-01 3.18982095e-01 -7.96776488e-02
1.21698511e+00 -5.96711338e-01 -4.36871082e-01 1.00563776e+00
1.28256190e+00 4.95214731e-01 1.03788340e+00 3.23567301e-01
9.65095818e-01 -7.44786635e-02 6.44966304e-01 1.03865437e-01
-3.82734746e-01 7.60829806e-01 1.07834148e+00 -2.82928377e-01
-3.19688261e-01 -4.05958951e-01 6.17929101e-01 -1.03797697e-01
3.20167333e-01 -6.94770336e-01 -5.46942711e-01 5.66061586e-02
-1.30080521e+00 -7.67193794e-01 1.29197910e-01 2.15246081e+00
3.87161404e-01 3.87999266e-01 -3.80998969e-01 -2.62655556e-01
7.26101518e-01 1.71962604e-01 -9.21656013e-01 -5.21941721e-01
-1.46720961e-01 1.94948062e-01 9.44233775e-01 4.33317274e-01
-1.06438947e+00 1.19099808e+00 6.14079857e+00 1.05425262e+00
-1.52427328e+00 1.10274717e-01 1.25125870e-01 -1.34294396e-02
-4.24261570e-01 -1.22889588e-02 -5.52428901e-01 3.97194743e-01
7.07587242e-01 -2.58997589e-01 6.28251791e-01 1.00124097e+00
-1.62175838e-02 2.62327105e-01 -1.03524137e+00 5.20208776e-01
1.83791026e-01 -1.37476563e+00 3.44855487e-01 2.75781214e-01
7.19537139e-01 -1.98924407e-01 6.88181281e-01 3.35240275e-01
7.11495757e-01 -1.02895701e+00 9.50556159e-01 1.93845123e-01
1.22320008e+00 -7.04252720e-01 2.84789085e-01 2.29732662e-01
-8.55286598e-01 1.37368202e-01 -3.08265418e-01 9.89882722e-02
1.15899496e-01 -2.01333791e-01 -7.09686816e-01 4.02807802e-01
4.17585850e-01 -1.50954396e-01 -3.23190004e-01 7.88029850e-01
-4.45382506e-01 4.50490385e-01 -4.29108709e-01 1.60665840e-01
4.12390947e-01 3.04681599e-01 1.18311739e+00 6.11145735e-01
3.69787037e-01 1.00159891e-01 1.74734682e-01 9.21874821e-01
-7.02780634e-02 -6.09011650e-01 -1.20545828e+00 1.78208604e-01
4.54436332e-01 9.07526374e-01 -4.30937737e-01 -6.56074136e-02
-1.69840246e-01 9.64934349e-01 -3.34014297e-02 5.52684486e-01
-1.54410267e+00 -3.07478786e-01 1.21845567e+00 1.60665229e-01
2.15882346e-01 -4.22624886e-01 -1.14295416e-01 -9.15993154e-01
-3.60734165e-02 -1.18528867e+00 -1.03915766e-01 -1.01645017e+00
-1.17004204e+00 6.98544681e-01 -7.24102044e-03 -1.47738147e+00
1.88775316e-01 -9.56836820e-01 -6.10331237e-01 8.89785588e-01
-1.25278366e+00 -1.54021025e+00 -2.09855646e-01 6.85532629e-01
1.06760502e-01 -3.09741676e-01 6.97205186e-01 3.29493321e-02
-1.34041816e-01 9.30682003e-01 7.04399077e-03 1.01593405e-01
5.37289321e-01 -6.82908535e-01 1.04735219e+00 1.14024317e+00
-3.09320927e-01 3.85892868e-01 8.58669579e-01 -9.92338002e-01
-1.97525144e+00 -1.25760269e+00 -2.71055311e-01 -7.12320447e-01
7.22347856e-01 -6.24443769e-01 -1.02122688e+00 7.12154210e-01
1.20287485e-01 1.76205546e-01 7.17658326e-02 -8.16157162e-01
-7.50574708e-01 6.91293254e-02 -1.63624597e+00 1.20151603e+00
9.52885807e-01 -6.99947417e-01 -2.61344552e-01 1.71042770e-01
1.18490064e+00 -9.61218238e-01 -5.10201037e-01 4.96482134e-01
6.25342727e-01 -6.95523143e-01 1.14536536e+00 -7.11794496e-01
1.39634073e-01 -7.02620208e-01 -2.36214206e-01 -1.24008012e+00
-1.38827741e-01 -1.13476527e+00 -2.22727805e-01 9.49259758e-01
-2.87447184e-01 -1.08384585e+00 6.30815327e-01 6.92732632e-01
-2.61483073e-01 -3.68361145e-01 -1.19456196e+00 -1.21064389e+00
2.25174516e-01 -3.86303931e-01 1.09673285e+00 8.91329587e-01
-6.75720811e-01 -5.32881260e-01 -7.73797750e-01 9.98041213e-01
5.84388912e-01 -2.68062681e-01 1.27168941e+00 -5.33336759e-01
-3.65293473e-01 -1.54077262e-01 -7.97695041e-01 -7.85198152e-01
4.42193210e-01 -7.24764049e-01 -1.17722657e-02 -6.73869371e-01
-2.29049683e-01 -5.93582332e-01 -2.32080724e-02 4.07859147e-01
-3.05606481e-02 4.67060715e-01 4.30022985e-01 5.12044085e-03
7.95276240e-02 6.25719905e-01 1.64182544e+00 -4.82627213e-01
1.09960914e-01 -7.43522868e-02 -4.24673676e-01 7.95856774e-01
6.22460365e-01 -6.84014678e-01 -5.23438513e-01 -6.01278365e-01
-9.47687477e-02 9.50495973e-02 1.13087809e+00 -1.03756535e+00
-8.96525681e-02 -4.82631981e-01 2.07394809e-01 -2.42548659e-01
6.99388266e-01 -1.15204680e+00 6.04085147e-01 6.84256554e-01
1.57229841e-01 -1.15756914e-01 7.73337305e-01 5.17958999e-01
3.18002790e-01 2.61923701e-01 9.86860633e-01 2.26614386e-01
-6.22406185e-01 5.81192434e-01 -4.15558249e-01 -2.40323037e-01
1.48339438e+00 -3.57085913e-01 -9.46303904e-01 -1.71643972e-01
-2.33755425e-01 -1.81674495e-01 1.14231730e+00 6.54349864e-01
6.26697123e-01 -1.57431209e+00 -5.18324077e-01 5.54971755e-01
-1.23118892e-01 -3.25113714e-01 3.56141686e-01 -6.55971393e-02
-8.26920569e-01 6.14104979e-02 -6.19103193e-01 -4.20531929e-01
-1.31670570e+00 1.12082028e+00 5.23034573e-01 9.55633149e-02
-6.65966570e-01 4.16348904e-01 8.67744029e-01 -3.75407398e-01
7.92000629e-03 -1.28386050e-01 5.08076251e-01 -1.09113908e+00
5.33141017e-01 2.85574019e-01 -1.81642920e-01 -4.91241932e-01
-2.88179755e-01 4.75821584e-01 2.16693237e-01 1.40636489e-01
6.76640511e-01 3.19304496e-01 2.15429738e-01 -2.43393973e-01
1.00634801e+00 2.61205941e-01 -1.84225261e+00 1.50626138e-01
-8.79377186e-01 -9.88154411e-01 -1.73709154e-01 -7.07105279e-01
-1.07150006e+00 6.22739017e-01 6.84488356e-01 2.94971734e-01
8.58316362e-01 -3.53788108e-01 9.57861185e-01 4.78063941e-01
6.44670725e-01 -3.40395391e-01 1.67751804e-01 3.81847441e-01
1.02234912e+00 -5.83725095e-01 -2.77016342e-01 -8.69421482e-01
-3.34630430e-01 5.90734422e-01 8.98049355e-01 -6.62317872e-01
5.62864840e-01 8.25385809e-01 3.71691406e-01 -1.36435896e-01
-3.40771645e-01 4.89507824e-01 1.12727374e-01 1.01716876e+00
-7.76907742e-01 1.36013031e-01 4.71987247e-01 4.34450731e-02
-2.12485641e-01 -3.79603356e-01 5.96958756e-01 1.00512862e+00
-6.02007806e-02 -8.76003385e-01 -9.30869341e-01 -1.36003807e-01
-3.87430310e-01 2.63453633e-01 -4.54036415e-01 1.29428041e+00
2.12003887e-01 8.36085737e-01 -1.37500569e-01 -4.64273512e-01
3.71630013e-01 -3.97130311e-01 5.99949121e-01 -1.48406416e-01
-5.74593008e-01 -3.39679927e-01 -7.03638196e-02 -7.30268598e-01
1.13170832e-01 -2.19263077e-01 -9.99898255e-01 -7.66906023e-01
-1.63301781e-01 -2.92896688e-01 5.35766542e-01 7.28214860e-01
4.43993479e-01 5.14978826e-01 8.77528846e-01 -1.03600347e+00
-8.07694912e-01 -2.79048502e-01 -4.66427714e-01 5.07333279e-01
4.57981408e-01 -9.50733721e-01 -4.59817022e-01 -9.63030383e-02] | [5.4350905418396, 7.878684043884277] |
7043733f-bd2d-44e5-bfa0-31adbde2d794 | pedestrian-behavior-maps-for-safety | 2305.04506 | null | https://arxiv.org/abs/2305.04506v1 | https://arxiv.org/pdf/2305.04506v1.pdf | Pedestrian Behavior Maps for Safety Advisories: CHAMP Framework and Real-World Data Analysis | It is critical for vehicles to prevent any collisions with pedestrians. Current methods for pedestrian collision prevention focus on integrating visual pedestrian detectors with Automatic Emergency Braking (AEB) systems which can trigger warnings and apply brakes as a pedestrian enters a vehicle's path. Unfortunately, pedestrian-detection-based systems can be hindered in certain situations such as night-time or when pedestrians are occluded. Our system addresses such issues using an online, map-based pedestrian detection aggregation system where common pedestrian locations are learned after repeated passes of locations. Using a carefully collected and annotated dataset in La Jolla, CA, we demonstrate the system's ability to learn pedestrian zones and generate advisory notices when a vehicle is approaching a pedestrian despite challenges like dark lighting or pedestrian occlusion. Using the number of correct advisories, false advisories, and missed advisories to define precision and recall performance metrics, we evaluate our system and discuss future positive effects with further data collection. We have made our code available at https://github.com/s7desai/ped-mapping, and a video demonstration of the CHAMP system at https://youtu.be/dxeCrS_Gpkw. | ['Mohan Trivedi', 'Afnan Alofi', 'Akshay Gopalkrishnan', 'Lulua Rakla', 'Samveed Desai', 'Ross Greer'] | 2023-05-08 | null | null | null | null | ['pedestrian-detection'] | ['computer-vision'] | [-4.31100667e-01 -3.23753178e-01 -1.87992439e-01 -4.38690871e-01
-8.38546872e-01 -5.84008336e-01 6.22212648e-01 5.60533106e-01
-7.47738004e-01 7.25913584e-01 2.51734287e-01 -9.45646584e-01
4.84840602e-01 -9.54081476e-01 -6.54376864e-01 -1.91312402e-01
-1.13919109e-01 1.91042423e-01 9.57635701e-01 -2.83266962e-01
5.21456711e-02 7.03014553e-01 -1.65985787e+00 2.69509047e-01
4.51144159e-01 5.25190771e-01 -1.77293643e-02 1.05678523e+00
1.59223855e-01 4.31620896e-01 -6.02708399e-01 -2.61239648e-01
4.64659780e-01 1.95954040e-01 -2.17970312e-01 -5.08359253e-01
6.95342660e-01 -8.11824560e-01 -7.51931727e-01 4.99340385e-01
6.12950146e-01 2.72301644e-01 5.66617131e-01 -1.89699113e+00
-3.40239666e-02 -2.77583927e-01 -5.23638964e-01 8.92705441e-01
6.15006745e-01 1.00991642e+00 3.78039420e-01 -8.57898355e-01
1.70167014e-01 1.22763264e+00 8.56463432e-01 5.34066200e-01
-1.09960401e+00 -9.49068666e-01 3.10169876e-01 6.89167142e-01
-1.64016604e+00 -8.09107959e-01 3.00972551e-01 -6.79648161e-01
1.04614270e+00 4.44343805e-01 8.02390993e-01 1.02567613e+00
-1.65443253e-02 8.12231421e-01 5.83702385e-01 -5.99683374e-02
2.39882678e-01 2.77601153e-01 3.09776425e-01 4.91472632e-01
6.73660040e-01 5.02853990e-01 -5.19893289e-01 -1.34397119e-01
1.03561215e-01 3.96072268e-02 -8.16048682e-02 2.20512878e-02
-8.12104881e-01 7.40699470e-01 4.62931573e-01 -4.46277648e-01
-2.10507616e-01 1.28385797e-02 2.92586088e-01 -2.62541503e-01
1.35749131e-01 -2.41371840e-01 2.09147200e-01 -1.94615662e-01
-8.45420539e-01 5.40624499e-01 5.03009915e-01 1.19516194e+00
6.71153903e-01 -1.48600921e-01 -2.78728068e-01 4.34595942e-01
3.80849272e-01 9.25129890e-01 -4.93988901e-01 -7.74979532e-01
4.88071918e-01 3.71257246e-01 6.05416298e-01 -1.15204358e+00
-4.69236702e-01 4.62226003e-01 -2.53668129e-01 6.14449084e-01
7.36946166e-01 -5.83883524e-01 -7.87061930e-01 1.19527042e+00
5.37296951e-01 2.61293173e-01 -3.64649206e-01 1.16068447e+00
9.52410579e-01 6.07713878e-01 8.05740237e-01 2.80852646e-01
1.43598604e+00 -4.52373773e-01 -5.24299800e-01 -5.30267477e-01
5.24999857e-01 -7.52742112e-01 6.94669902e-01 6.54665232e-02
-7.17764854e-01 -6.87273681e-01 -9.70825315e-01 -3.32302861e-02
-7.94722617e-01 3.14514749e-02 1.70808345e-01 8.24809968e-01
-1.15656137e+00 -5.30152433e-02 -7.92375326e-01 -8.17356288e-01
5.38191259e-01 8.20049644e-02 -1.42566994e-01 -4.54494022e-02
-1.27146173e+00 1.15896642e+00 -6.87268749e-02 8.57592300e-02
-1.03028941e+00 -6.50655448e-01 -8.59025300e-01 -2.58974582e-01
2.48794019e-01 -3.82282346e-01 1.27159548e+00 -1.46548063e-01
-4.88114148e-01 5.37084341e-01 -6.42815769e-01 -5.26828647e-01
7.08640218e-01 -4.10604060e-01 -8.27278674e-01 7.69832656e-02
4.81704742e-01 1.11921096e+00 1.57529876e-01 -1.29770827e+00
-1.39915633e+00 -1.35180512e-02 1.50593789e-02 1.20291382e-01
2.12392703e-01 1.91673532e-01 -3.34957957e-01 -2.48128287e-02
-5.64158857e-01 -7.04845905e-01 -4.41003442e-01 2.63480961e-01
-4.48754311e-01 -2.02322945e-01 1.11333680e+00 -7.45403409e-01
1.20155251e+00 -2.02228928e+00 -1.09592772e+00 3.47785443e-01
7.46069327e-02 5.58161616e-01 2.08945107e-03 3.90455365e-01
1.90505654e-01 6.74027503e-02 5.21558344e-01 -2.65447170e-01
-2.50683334e-02 5.61699346e-02 -2.65513539e-01 4.76034105e-01
1.92461297e-01 7.07468569e-01 -1.12426651e+00 -4.70984995e-01
8.97788823e-01 6.73922122e-01 -2.11383641e-01 3.51481847e-02
3.29771191e-01 1.84552848e-01 -1.55930966e-01 7.38456786e-01
8.64479899e-01 3.57413173e-01 -2.42470905e-01 -1.64246652e-03
-7.94625461e-01 4.69668299e-01 -1.33376527e+00 4.81080502e-01
9.05087516e-02 1.24632978e+00 -5.67298979e-02 -4.68754351e-01
6.53855920e-01 1.67000309e-01 1.95132330e-01 -8.27072740e-01
-1.76919237e-01 -1.32134512e-01 -4.40473616e-01 -3.31799477e-01
7.51613557e-01 5.28725266e-01 3.02858483e-02 9.18234661e-02
-7.24783003e-01 5.17358422e-01 5.00659168e-01 6.09376729e-01
1.12334979e+00 -3.12537402e-01 1.56754538e-01 -1.58388596e-02
1.62862286e-01 6.40942812e-01 4.89314318e-01 1.01190853e+00
-8.78323436e-01 4.55783993e-01 2.04181150e-01 -7.42782950e-01
-1.17625642e+00 -1.40447569e+00 -1.38696432e-01 9.50789273e-01
4.57668215e-01 -6.44706607e-01 -5.57987273e-01 -5.55824876e-01
3.39133680e-01 1.02919424e+00 -2.75534540e-01 -7.44016096e-02
-5.94051540e-01 -4.33841050e-01 5.89560926e-01 8.60038877e-01
4.87425506e-01 -5.92077971e-01 -9.59780931e-01 1.11083813e-01
-3.12419683e-01 -9.35540438e-01 -5.26773512e-01 -3.66963208e-01
9.88908038e-02 -1.35617995e+00 -5.82930863e-01 -3.64191681e-01
7.40912020e-01 1.00046682e+00 7.90590465e-01 3.04892242e-01
-5.32950342e-01 8.51262510e-01 -5.24689518e-02 -7.08083987e-01
-2.15015531e-01 -3.74013692e-01 2.62734234e-01 -3.14461768e-01
1.10448742e+00 9.20934007e-02 -1.24612820e+00 8.20780635e-01
1.22692645e-01 2.42450565e-01 -3.89054865e-02 8.57455209e-02
2.98280656e-01 -1.50670767e-01 3.39368075e-01 -1.08111516e-01
4.09033656e-01 -7.49197364e-01 -8.75631452e-01 -8.38617608e-02
-3.15764189e-01 -6.70688391e-01 4.24538516e-02 -2.46609285e-01
-9.10566986e-01 1.75247476e-01 -1.07882477e-01 5.66606149e-02
-7.97007561e-01 -3.27663600e-01 -1.06441960e-01 1.55529335e-01
9.24771786e-01 -1.03362322e-01 -3.11359257e-01 -1.22829050e-01
2.71208018e-01 8.11689973e-01 6.15181148e-01 -2.27298364e-01
6.93786144e-01 7.68142939e-01 -3.46006542e-01 -9.17213619e-01
-2.15501279e-01 -9.21225429e-01 -5.11420190e-01 -9.69737887e-01
9.25453424e-01 -1.21767271e+00 -1.23313105e+00 1.36470795e-01
-1.23981655e+00 -5.41298926e-01 1.21050589e-01 4.21185583e-01
-1.56787671e-02 2.25414768e-01 -1.91542581e-01 -1.15972638e+00
7.12942034e-02 -1.02362919e+00 8.77217829e-01 6.47780120e-01
-4.51749831e-01 -5.77169895e-01 -6.79802075e-02 2.64845639e-01
3.30933869e-01 1.39081508e-01 -2.25657020e-02 -3.17575991e-01
-7.98462987e-01 -3.83117974e-01 -4.84514683e-01 -7.58466944e-02
9.10935178e-02 9.11846757e-02 -8.91222656e-01 -2.17533767e-01
-1.02291024e+00 3.61443460e-01 7.50176311e-01 5.23941159e-01
4.05417711e-01 -3.62448692e-01 -1.05506623e+00 6.33753911e-02
9.26589489e-01 4.58092421e-01 5.83787739e-01 5.11924565e-01
5.42180717e-01 4.55624938e-01 7.68356204e-01 4.21631038e-01
9.94998455e-01 8.57154727e-01 1.62452698e-01 -4.09641743e-01
-3.79620552e-01 -4.21788365e-01 2.79367417e-01 -3.86722386e-01
-3.18965651e-02 -3.15890074e-01 -1.40078759e+00 8.57445776e-01
-1.92510414e+00 -1.28560984e+00 -6.78719163e-01 2.34529066e+00
3.91808718e-01 4.44942743e-01 7.32535124e-01 -8.87329057e-02
8.94119859e-01 -2.01171204e-01 -3.08594376e-01 -3.79868716e-01
2.63157457e-01 -5.06507158e-01 1.14625764e+00 9.65824425e-01
-1.30351806e+00 8.74261379e-01 6.31986141e+00 4.28947508e-01
-8.19537401e-01 9.91145819e-02 7.83370972e-01 -3.73610765e-01
5.53081818e-02 -1.30582437e-01 -1.44043219e+00 7.98642039e-01
1.05758941e+00 -9.21141058e-02 1.71279117e-01 8.70094478e-01
9.52636063e-01 -7.91505456e-01 -8.63115430e-01 7.28344321e-01
-1.90272555e-01 -1.19552624e+00 -3.61490399e-01 2.93225553e-02
5.61586358e-02 1.41678572e-01 -9.11336541e-02 3.31304580e-01
6.11126304e-01 -7.82429457e-01 8.50381613e-01 5.13840377e-01
7.22259581e-01 -9.00012732e-01 3.20373297e-01 1.43142879e-01
-1.62600756e+00 2.51331367e-02 -1.67360097e-01 -2.20586985e-01
7.21587420e-01 1.35362267e-01 -1.26399446e+00 -1.62839115e-01
1.01463783e+00 4.16472346e-01 -8.65286231e-01 1.59415567e+00
-4.28681016e-01 6.26051784e-01 -7.50987053e-01 -2.13153288e-01
1.03820227e-01 2.88357556e-01 7.91149199e-01 1.67867267e+00
1.42774180e-01 4.03966337e-01 3.41579705e-01 5.97491205e-01
8.14148009e-01 -3.27484637e-01 -1.02684748e+00 7.57845461e-01
7.75186777e-01 1.15516937e+00 -7.39763975e-01 -5.14521420e-01
-5.28883517e-01 4.71564978e-01 -5.22941686e-02 7.80685902e-01
-1.21937919e+00 -2.78390348e-01 1.06601846e+00 8.49699557e-01
1.48484856e-01 -4.63700861e-01 -3.81344646e-01 -5.09808660e-01
-6.56841621e-02 -2.51699597e-01 5.67324162e-01 -9.37621534e-01
-7.01216519e-01 2.07762495e-01 4.02034849e-01 -1.38994586e+00
6.02142662e-02 -5.42384863e-01 -8.89196515e-01 9.29337442e-01
-1.28656304e+00 -8.98085535e-01 -6.25782430e-01 5.72221935e-01
5.07697046e-01 1.04129896e-01 2.75488317e-01 7.60879517e-01
-5.64054310e-01 7.10442483e-01 -4.31143492e-01 2.29371786e-01
7.87781000e-01 -1.00813746e+00 8.85758936e-01 1.17397845e+00
-4.80568796e-01 3.58811527e-01 9.18821275e-01 -1.03011882e+00
-8.47155511e-01 -1.20930290e+00 9.88311648e-01 -8.72939944e-01
2.48856589e-01 -5.79546690e-01 -6.46618128e-01 8.27842355e-01
1.15776680e-01 -1.21137634e-01 6.04234636e-01 6.24793619e-02
-2.50213929e-02 -1.68258607e-01 -1.13557553e+00 1.16239917e+00
9.31272388e-01 -1.50797948e-01 -1.38925001e-01 3.97185534e-01
1.38073951e-01 -3.66803527e-01 -8.99475664e-02 1.02467686e-01
5.57271063e-01 -9.57791805e-01 1.16015124e+00 -4.00895655e-01
-4.43597466e-01 -8.50181520e-01 8.38927776e-02 -1.04840755e+00
-2.76240826e-01 -3.85784179e-01 1.57330930e-01 1.06888354e+00
4.36223716e-01 -6.95249975e-01 5.65253258e-01 1.38279116e+00
-1.46492541e-01 -2.67029107e-01 -1.06124926e+00 -5.93396723e-01
-5.21644294e-01 -8.69338095e-01 4.72915083e-01 3.71314198e-01
1.66182384e-01 -9.17715132e-02 -3.82169217e-01 7.03553259e-01
7.42251277e-01 -7.29903817e-01 1.20450485e+00 -5.85932076e-01
5.41700304e-01 -4.49364603e-01 -6.39142931e-01 -9.47292745e-01
-5.13104796e-01 -3.27771187e-01 2.16950059e-01 -1.68848681e+00
-1.78965311e-02 -5.83176017e-01 2.45398339e-02 6.97066665e-01
-1.75840259e-01 2.48576954e-01 2.40608320e-01 -7.59571791e-02
-8.64465714e-01 -5.85904159e-02 4.40152913e-01 -1.94657519e-01
-1.98594376e-01 1.75102547e-01 -4.50215071e-01 7.37913251e-01
1.08863664e+00 -3.40381891e-01 -1.42823681e-01 -2.26350829e-01
-8.98032412e-02 -3.39327008e-01 1.08839238e+00 -1.40785193e+00
6.65228009e-01 -2.84063101e-01 7.10905850e-01 -1.09973919e+00
4.33255941e-01 -6.77596927e-01 1.87707648e-01 5.76404870e-01
7.94756487e-02 3.35071713e-01 9.11528647e-01 4.66817230e-01
4.76453334e-01 2.85513788e-01 5.44338822e-01 1.43469259e-01
-1.12970710e+00 1.34845510e-01 -1.10585058e+00 -1.44574851e-01
1.44149816e+00 -5.27869880e-01 -8.16249788e-01 -5.06381691e-01
-5.07141769e-01 9.04731274e-01 5.06680071e-01 6.49952471e-01
8.44662070e-01 -1.45840275e+00 -6.04002893e-01 3.01154733e-01
8.93182382e-02 -2.88205117e-01 3.88547093e-01 7.23234773e-01
-5.49391031e-01 5.41991055e-01 -2.36694932e-01 -7.36148477e-01
-1.52182806e+00 5.00347435e-01 3.64611119e-01 6.08266950e-01
-7.95356095e-01 4.23025787e-01 1.33988082e-01 1.67453706e-01
4.06595707e-01 6.26818761e-02 -1.18634589e-01 -1.55605212e-01
1.03004122e+00 8.82057369e-01 -2.89786398e-01 -9.75242674e-01
-8.98239255e-01 -1.37213763e-04 -7.10308775e-02 -2.04408512e-01
5.48500478e-01 -4.29515868e-01 7.97697604e-01 -6.02319278e-02
5.38293362e-01 -5.19495904e-02 -1.74594772e+00 2.22309291e-01
-8.05553645e-02 -7.07111120e-01 -2.54268646e-01 -7.37603426e-01
-5.16639531e-01 7.00071096e-01 1.00236261e+00 7.47885406e-02
6.31040037e-01 1.78500786e-01 9.02386606e-01 1.05906069e-01
3.58247280e-01 -1.16234684e+00 -4.75835860e-01 1.65781930e-01
5.49182415e-01 -1.55079138e+00 -1.81098245e-02 -1.44635618e-01
-6.95048869e-01 6.39396012e-01 1.03248000e+00 2.07644105e-01
5.39111674e-01 4.72088456e-01 2.76500851e-01 -4.50947843e-02
-6.29483342e-01 -5.78050196e-01 2.28264153e-01 1.05584395e+00
-1.71317384e-01 4.80122387e-01 -1.14092693e-01 2.56754577e-01
-1.13033941e-02 -2.04414144e-01 2.92688519e-01 9.43738401e-01
-8.18350792e-01 -5.39463103e-01 -6.70237422e-01 2.47678518e-01
1.42357543e-01 -3.51456106e-02 -2.06689432e-01 8.93679440e-01
2.67224997e-01 1.46780574e+00 4.15787578e-01 -3.73092592e-01
7.91401088e-01 -1.61542684e-01 -1.23129725e-01 -2.00286061e-01
-1.20230742e-01 -2.41519302e-01 5.89110017e-01 -7.63632715e-01
7.30562210e-03 -9.86059785e-01 -1.23391080e+00 -7.95117259e-01
1.80178970e-01 -1.47682682e-01 5.21711946e-01 5.70603430e-01
5.54429114e-01 1.46612927e-01 3.20100278e-01 -1.14428484e+00
4.26959842e-01 -4.51883018e-01 1.13884084e-01 3.24494481e-01
5.80965877e-01 -8.34966063e-01 -8.38836804e-02 1.12586431e-01] | [7.754637718200684, -0.8871783018112183] |
b5920e39-0d17-48d6-ad37-641ede583759 | deleter-leveraging-bert-to-perform | 1909.03223 | null | https://arxiv.org/abs/1909.03223v1 | https://arxiv.org/pdf/1909.03223v1.pdf | Deleter: Leveraging BERT to Perform Unsupervised Successive Text Compression | Text compression has diverse applications such as Summarization, Reading Comprehension and Text Editing. However, almost all existing approaches require either hand-crafted features, syntactic labels or parallel data. Even for one that achieves this task in an unsupervised setting, its architecture necessitates a task-specific autoencoder. Moreover, these models only generate one compressed sentence for each source input, so that adapting to different style requirements (e.g. length) for the final output usually implies retraining the model from scratch. In this work, we propose a fully unsupervised model, Deleter, that is able to discover an "optimal deletion path" for an arbitrary sentence, where each intermediate sequence along the path is a coherent subsequence of the previous one. This approach relies exclusively on a pretrained bidirectional language model (BERT) to score each candidate deletion based on the average Perplexity of the resulting sentence and performs progressive greedy lookahead search to select the best deletion for each step. We apply Deleter to the task of extractive Sentence Compression, and found that our model is competitive with state-of-the-art supervised models trained on 1.02 million in-domain examples with similar compression ratio. Qualitative analysis, as well as automatic and human evaluations both verify that our model produces high-quality compression. | ['Richard Socher', 'Tong Niu', 'Caiming Xiong'] | 2019-09-07 | null | null | null | null | ['sentence-compression', 'text-compression'] | ['natural-language-processing', 'natural-language-processing'] | [ 7.10108519e-01 2.44272828e-01 -3.14316750e-01 -3.33195984e-01
-8.52169335e-01 -3.82185400e-01 4.27689523e-01 5.17397583e-01
-6.80637419e-01 6.67874038e-01 6.10341370e-01 -4.23123926e-01
1.40094534e-01 -6.82172716e-01 -8.49422991e-01 -3.96824807e-01
1.63217336e-01 7.15420425e-01 -1.28869757e-01 -3.10108364e-01
5.17110407e-01 5.12888171e-02 -1.34027994e+00 5.43018997e-01
1.02986252e+00 5.52905321e-01 7.11565793e-01 9.36620235e-01
-3.32602918e-01 6.38468146e-01 -7.97473133e-01 -6.30712748e-01
2.03869175e-02 -8.22625399e-01 -1.08026886e+00 3.73923592e-03
2.24058032e-01 -3.74964178e-01 -4.24013734e-01 1.04479539e+00
3.79043490e-01 -4.25614715e-02 7.84818709e-01 -4.04432625e-01
-6.54170930e-01 1.19785881e+00 -3.50405812e-01 1.39559761e-01
4.65895087e-01 3.40832099e-02 1.30562019e+00 -6.35481417e-01
7.92322397e-01 8.17352474e-01 4.36921090e-01 6.75613046e-01
-1.23302460e+00 -1.40634209e-01 2.15374548e-02 2.48673469e-01
-1.03976619e+00 -5.16164422e-01 7.47620881e-01 -5.50165325e-02
1.43756938e+00 4.00199205e-01 6.98351085e-01 1.16031826e+00
3.30874950e-01 9.71607149e-01 4.95799869e-01 -7.59745717e-01
3.33521783e-01 -9.74541064e-03 -1.22387193e-01 4.37035888e-01
8.65614861e-02 -5.03445446e-01 -5.30022502e-01 1.50045902e-01
1.85769927e-02 -2.23142236e-01 -3.56257558e-01 -3.76974829e-02
-1.02556634e+00 8.19202423e-01 1.20666297e-02 4.45706248e-01
-4.40509081e-01 -3.81065086e-02 7.03133285e-01 6.19893193e-01
5.71200073e-01 5.54139555e-01 -6.05803370e-01 -3.25727850e-01
-1.42456365e+00 1.73498556e-01 1.09518754e+00 1.00342131e+00
6.74475014e-01 -1.54999420e-01 -2.28746071e-01 9.69227076e-01
-8.38869512e-02 2.22954914e-01 8.34612429e-01 -8.48874748e-01
9.19180810e-01 4.80661124e-01 -1.55005410e-01 -7.85449266e-01
-1.69457987e-01 -6.15189731e-01 -1.15339851e+00 -3.23574781e-01
-3.15213739e-03 -1.00397937e-01 -6.01986289e-01 1.68771863e+00
-1.54851466e-01 -2.82055289e-01 2.68344730e-01 6.33782208e-01
4.27826881e-01 9.05596435e-01 -1.56925499e-01 -5.35849631e-01
9.29142177e-01 -1.10629857e+00 -4.74947572e-01 -3.79546940e-01
1.03988409e+00 -6.42574668e-01 1.12057948e+00 6.15944445e-01
-1.44349420e+00 -2.75320053e-01 -1.12340891e+00 -2.23636732e-01
-4.32784669e-03 3.29902977e-01 1.05819665e-01 3.88787299e-01
-9.81878400e-01 9.74609673e-01 -6.68466210e-01 -3.38798881e-01
2.31727317e-01 1.24821916e-01 -2.40475163e-01 -1.00202024e-01
-9.34073627e-01 9.04925585e-01 6.77489161e-01 -1.00474700e-01
-7.40828216e-01 -4.84549254e-01 -7.29822040e-01 5.61345756e-01
3.58849734e-01 -9.52110469e-01 1.47124851e+00 -1.25947356e+00
-1.65169322e+00 5.94416082e-01 -3.95162791e-01 -1.11331916e+00
4.54840958e-01 -3.84892464e-01 -1.32583573e-01 3.59620661e-01
-8.18751529e-02 5.61596036e-01 9.76027429e-01 -8.75530720e-01
-4.65616763e-01 -8.93955156e-02 -2.11892262e-01 3.05860668e-01
-5.65868676e-01 -3.04175839e-02 -5.54948151e-01 -8.02069187e-01
-1.65311366e-01 -6.24112904e-01 -2.87078738e-01 -6.68482423e-01
-5.40094614e-01 -1.96207985e-01 4.98439878e-01 -1.07469368e+00
1.64541769e+00 -1.96379757e+00 5.12759984e-01 -3.48469988e-02
-7.42316917e-02 3.60748649e-01 -3.27460706e-01 8.02677155e-01
1.31249458e-01 2.36869112e-01 -7.16544926e-01 -6.11464322e-01
-9.07918140e-02 4.39868793e-02 -3.73122126e-01 2.37718001e-01
1.74425930e-01 7.17157006e-01 -9.43093657e-01 -5.91749251e-01
-1.73583217e-02 1.42608091e-01 -9.26528752e-01 3.45039308e-01
-5.22342622e-01 1.59124359e-01 -4.49822731e-02 1.80835381e-01
4.12754297e-01 -9.70133170e-02 4.66890246e-01 -2.73571769e-03
7.03481138e-02 7.35423505e-01 -7.69281209e-01 1.92977977e+00
-7.09284663e-01 6.77124321e-01 -2.11567730e-01 -1.38572466e+00
9.00126815e-01 2.64522403e-01 2.81165272e-01 -8.20954144e-01
1.07787363e-01 1.87330291e-01 -3.92488949e-02 -6.78518951e-01
8.96584511e-01 7.36266822e-02 -1.51153624e-01 7.24381030e-01
2.31733844e-01 -1.96686566e-01 6.42956972e-01 5.38815439e-01
1.26336670e+00 2.93105072e-03 4.01296943e-01 1.22589245e-01
6.44701064e-01 -1.43594146e-01 4.20988441e-01 9.08045471e-01
2.04364538e-01 8.25862765e-01 6.30957425e-01 -2.31126383e-01
-1.48615754e+00 -6.08116925e-01 3.59537750e-01 1.06733704e+00
-2.11076930e-01 -8.50667000e-01 -1.14312065e+00 -6.14308238e-01
-3.21658671e-01 1.13050413e+00 -2.35300109e-01 -3.25401634e-01
-9.70561683e-01 -4.12829518e-01 5.76948583e-01 3.59764516e-01
1.42116293e-01 -1.28810096e+00 -8.31352532e-01 4.03077513e-01
-4.75947887e-01 -8.49665165e-01 -5.84013402e-01 2.76399940e-01
-1.09542727e+00 -6.84490740e-01 -4.28233325e-01 -8.45561028e-01
6.14187002e-01 -2.68118512e-02 1.20283616e+00 2.12854013e-01
-1.04962755e-02 7.71408826e-02 -6.21634305e-01 -2.77126282e-01
-8.81970346e-01 7.29160547e-01 -1.83640912e-01 -1.75660089e-01
2.30716065e-01 -6.69039011e-01 -4.56688106e-01 -3.53014112e-01
-1.10228121e+00 2.79209554e-01 9.99841928e-01 9.73017931e-01
6.23385310e-01 -6.56165630e-02 6.99153781e-01 -9.97247338e-01
1.12275445e+00 -3.96515310e-01 -1.12139173e-01 3.87861341e-01
-6.86746836e-01 4.31991458e-01 1.24535835e+00 -1.95219979e-01
-9.95490134e-01 1.75525527e-03 -3.97270977e-01 -2.13542998e-01
-2.27697790e-01 9.03363168e-01 -7.99743459e-02 7.94372678e-01
6.83879733e-01 8.21216166e-01 5.67913987e-03 -6.94842994e-01
4.86961186e-01 8.07269514e-01 7.31548071e-01 -3.84383529e-01
4.37136769e-01 6.80432022e-02 -3.90343010e-01 -9.64008331e-01
-7.76414931e-01 -3.20416063e-01 -8.06041598e-01 1.14350639e-01
5.46086907e-01 -6.74202144e-01 -3.17964226e-01 1.70285851e-01
-1.48340821e+00 -2.84414560e-01 -4.44093198e-01 3.62251103e-01
-7.33695388e-01 6.97880030e-01 -6.37067676e-01 -5.91355979e-01
-9.08145308e-01 -8.18904936e-01 9.00064230e-01 -1.04297407e-01
-5.48780978e-01 -7.43775249e-01 8.64688903e-02 2.76909769e-01
4.26033318e-01 -3.44581366e-01 1.19281030e+00 -8.88818502e-01
-4.24881160e-01 -3.96470368e-01 2.01476365e-01 7.17619658e-01
-1.93618312e-01 -4.76187132e-02 -4.93677855e-01 -3.20933700e-01
1.54129684e-01 -4.87165272e-01 1.30075145e+00 2.11811677e-01
1.58695412e+00 -7.78408766e-01 2.69195694e-03 5.69382846e-01
1.24655449e+00 -1.25551268e-01 7.00090587e-01 2.12281287e-01
4.81896698e-01 5.60340464e-01 3.52142185e-01 5.45987427e-01
2.27519006e-01 3.31507921e-01 3.65786195e-01 3.60932082e-01
5.53654023e-02 -5.35002232e-01 5.41549802e-01 1.42887259e+00
4.63751890e-02 -5.89325309e-01 -5.68732798e-01 5.85134029e-01
-1.63867140e+00 -1.13942635e+00 2.91227490e-01 2.13951874e+00
1.14781976e+00 3.95370513e-01 -2.04582334e-01 3.59803438e-01
4.14272368e-01 3.53130728e-01 -4.70094144e-01 -9.43137586e-01
-1.11828730e-01 4.20496047e-01 4.61020380e-01 4.56063211e-01
-7.91344881e-01 9.39075053e-01 5.92568874e+00 9.39636111e-01
-9.96792614e-01 -9.22429636e-02 6.47417128e-01 -3.33863080e-01
-4.86734182e-01 1.01773493e-01 -6.46859527e-01 5.24418116e-01
1.32860982e+00 -3.57155859e-01 6.18001819e-01 6.74572885e-01
2.58996546e-01 -8.46500993e-02 -1.25147986e+00 7.17644632e-01
3.63363504e-01 -1.48614645e+00 4.62799519e-01 -3.31235558e-01
5.47824800e-01 -5.25762215e-02 -2.53892899e-01 3.96740943e-01
-1.67900342e-02 -1.03485787e+00 7.28134274e-01 5.38851321e-01
6.92774594e-01 -8.45830441e-01 7.90221453e-01 8.85496855e-01
-7.09718943e-01 -2.96996683e-02 -5.57163954e-01 -7.14249089e-02
1.90100744e-01 7.22624123e-01 -1.00773656e+00 6.17847025e-01
3.59233469e-01 7.91667700e-01 -5.52122831e-01 7.70151615e-01
-3.24713498e-01 8.95405650e-01 -1.70186549e-01 -1.46608308e-01
7.12281018e-02 -2.77695060e-01 6.71503067e-01 1.82385159e+00
4.22055662e-01 -1.44036636e-01 -8.65738094e-02 5.12014210e-01
-5.51314831e-01 4.40601200e-01 -5.61002553e-01 -2.11683884e-01
4.27683294e-01 9.63653147e-01 -3.27205271e-01 -5.06187260e-01
-3.15314710e-01 1.28129661e+00 6.08058929e-01 2.09032893e-01
-4.73323017e-01 -5.76492012e-01 8.19198787e-02 2.68198196e-02
6.43320560e-01 -1.73256531e-01 -4.49875623e-01 -1.28553760e+00
3.23123276e-01 -1.09948051e+00 3.15658152e-01 -4.21753675e-01
-9.03871834e-01 5.49466968e-01 -2.05704078e-01 -1.21277511e+00
-6.24067247e-01 -2.62367018e-02 -7.53415644e-01 7.33566642e-01
-1.34959614e+00 -6.82613373e-01 1.34860262e-01 2.58767158e-01
1.03897285e+00 -2.18347415e-01 9.37504232e-01 2.58312732e-01
-5.91767311e-01 7.24727929e-01 3.20424736e-01 -7.68791186e-04
6.79165304e-01 -1.22307003e+00 4.61607516e-01 9.65304255e-01
2.67212391e-01 6.25965953e-01 9.00330067e-01 -5.61539650e-01
-1.53672409e+00 -1.05688643e+00 1.47370851e+00 1.62761528e-02
4.03029650e-01 -2.52030969e-01 -9.61123407e-01 6.79051578e-01
5.39047360e-01 -6.10848665e-01 5.94850063e-01 1.30207213e-02
-1.89196751e-01 -1.11698166e-01 -7.29111433e-01 6.93630695e-01
1.08837008e+00 -4.15388376e-01 -8.10237169e-01 2.96669811e-01
9.56248462e-01 -3.83492619e-01 -5.71762860e-01 1.26954570e-01
2.61478662e-01 -1.00364327e+00 6.74554110e-01 -6.91358030e-01
1.14721584e+00 7.30649680e-02 -9.43231955e-03 -1.41685963e+00
-6.57579824e-02 -6.01084232e-01 -5.34957767e-01 1.07231307e+00
6.30073965e-01 -9.30367857e-02 7.07267940e-01 1.41443580e-01
-4.64461833e-01 -1.02076721e+00 -8.35197270e-01 -6.49331570e-01
6.25411123e-02 -4.78178650e-01 4.95014846e-01 4.66853708e-01
1.14656858e-01 7.10432827e-01 -6.15521133e-01 -2.81715512e-01
2.71509171e-01 2.36446008e-01 8.29255998e-01 -9.12887871e-01
-6.20960057e-01 -5.39183021e-01 1.15740202e-01 -1.53078687e+00
3.51202101e-01 -1.20877433e+00 1.28803894e-01 -1.62597847e+00
3.02213699e-01 -3.42714647e-03 5.75715639e-02 1.72336176e-01
-4.69357818e-02 -2.98479617e-01 1.59179866e-01 2.41014540e-01
-7.64044166e-01 8.44435155e-01 9.04116213e-01 -3.49952132e-01
-3.86786491e-01 -8.51941470e-04 -9.04683352e-01 6.84388340e-01
9.89830434e-01 -6.05269194e-01 -2.90874630e-01 -8.16102684e-01
3.54961544e-01 2.35988453e-01 -1.39771178e-01 -9.01867092e-01
4.40890282e-01 -5.24889901e-02 2.03293532e-01 -7.75858283e-01
1.12274894e-02 -5.30422866e-01 -2.51439929e-01 5.15056491e-01
-8.73929203e-01 1.24225847e-01 -2.16949418e-01 6.09211147e-01
-3.40860903e-01 -8.54786754e-01 6.16780400e-01 -2.29796305e-01
-2.60517508e-01 1.93673968e-02 -4.07229751e-01 1.57136723e-01
5.52118540e-01 -4.59172064e-03 -1.08044088e-01 -5.50653696e-01
-4.63799149e-01 4.17201370e-02 3.80827665e-01 2.27762952e-01
6.70336246e-01 -7.44438708e-01 -1.04276860e+00 1.42100766e-01
-1.88026771e-01 2.23023161e-01 1.77057877e-01 5.57508528e-01
-6.15834653e-01 4.83605772e-01 5.86722903e-02 -3.32469046e-01
-1.13240111e+00 5.85041761e-01 -8.42884406e-02 -7.20782042e-01
-8.26598585e-01 5.41293800e-01 -3.87104928e-01 -1.80034563e-01
3.47468615e-01 -2.77433485e-01 -6.12757802e-02 4.62069781e-03
5.29448748e-01 2.72222161e-01 3.71841311e-01 -2.99948186e-01
1.38856590e-01 1.90127268e-01 -4.90241319e-01 -1.49588197e-01
1.49882674e+00 -2.39259422e-01 -3.01753789e-01 2.84289598e-01
1.33851171e+00 -1.64768193e-02 -9.13073182e-01 -2.79645503e-01
2.21743971e-01 -2.22762302e-01 -2.28231743e-01 -5.67855954e-01
-7.62698591e-01 8.34972322e-01 -7.29052275e-02 1.79569960e-01
1.29554665e+00 -9.82309878e-02 1.15467227e+00 9.13738251e-01
9.97490287e-02 -1.41453540e+00 1.52958348e-01 8.68117630e-01
8.76761377e-01 -9.51223493e-01 6.50836974e-02 -8.95727612e-03
-6.69592679e-01 1.23577607e+00 3.50892544e-01 -1.10018142e-01
2.23268375e-01 6.38915971e-02 -3.61414671e-01 2.00510398e-01
-1.15760934e+00 2.83538640e-01 7.23479455e-03 2.36287177e-01
5.70845127e-01 6.41120151e-02 -8.52832735e-01 4.95611280e-01
-7.07909882e-01 -2.94579323e-02 7.03598022e-01 8.95093501e-01
-5.99525332e-01 -1.27449155e+00 6.11804202e-02 7.40185261e-01
-4.92465079e-01 -2.40900442e-01 -3.73235106e-01 2.36285523e-01
-1.64012417e-01 7.32442439e-01 1.47427484e-01 -3.39625150e-01
2.45978087e-01 2.56128997e-01 4.81670052e-01 -7.57504225e-01
-6.55063331e-01 -2.45890599e-02 1.45320460e-01 -2.30862737e-01
-1.63905859e-01 -7.15461910e-01 -1.18007851e+00 -3.49786490e-01
-1.24976806e-01 8.33785310e-02 7.01252460e-01 8.71248543e-01
4.69540983e-01 3.84664416e-01 6.11437023e-01 -6.68791234e-01
-9.73853528e-01 -1.24624360e+00 -2.78772593e-01 6.13607526e-01
2.36857295e-01 2.94378638e-01 -2.57234514e-01 3.63943607e-01] | [12.162691116333008, 9.247700691223145] |
ad161f5d-b50d-4639-96e3-7af359a35394 | rnn-transducers-for-nested-named-entity | 2203.03543 | null | https://arxiv.org/abs/2203.03543v1 | https://arxiv.org/pdf/2203.03543v1.pdf | RNN Transducers for Nested Named Entity Recognition with constraints on alignment for long sequences | Popular solutions to Named Entity Recognition (NER) include conditional random fields, sequence-to-sequence models, or utilizing the question-answering framework. However, they are not suitable for nested and overlapping spans with large ontologies and for predicting the position of the entities. To fill this gap, we introduce a new model for NER task -- an RNN transducer (RNN-T). These models are trained using paired input and output sequences without explicitly specifying the alignment between them, similar to other seq-to-seq models. RNN-T models learn the alignment using a loss function that sums over all alignments. In NER tasks, however, the alignment between words and target labels are available from the human annotations. We propose a fixed alignment RNN-T model that utilizes the given alignment, while preserving the benefits of RNN-Ts such as modeling output dependencies. As a more general case, we also propose a constrained alignment model where users can specify a relaxation of the given input alignment and the model will learn an alignment within the given constraints. In other words, we propose a family of seq-to-seq models which can leverage alignments between input and target sequences when available. Through empirical experiments on a challenging real-world medical NER task with multiple nested ontologies, we demonstrate that our fixed alignment model outperforms the standard RNN-T model, improving F1-score from 0.70 to 0.74. | ['Laurent El Shafey', 'Mingqiu Wang', 'Izhak Shafran', 'Hagen Soltau'] | 2022-02-08 | null | null | null | null | ['nested-named-entity-recognition'] | ['natural-language-processing'] | [ 5.43124318e-01 4.08771545e-01 -2.26628155e-01 -6.39697433e-01
-1.02291751e+00 -7.04065382e-01 9.48199108e-02 2.14333192e-01
-6.96899116e-01 6.94691062e-01 3.45041156e-01 -3.39282900e-01
-2.68317331e-02 -7.11656451e-01 -6.91790581e-01 -3.10930818e-01
4.60644551e-02 5.14237881e-01 1.53286234e-01 -1.78991720e-01
-9.91082042e-02 8.86230078e-03 -1.06039000e+00 4.01534140e-01
1.15207517e+00 6.19650900e-01 2.97616333e-01 6.73234403e-01
-5.21380246e-01 9.00795519e-01 -5.28520107e-01 -6.58395946e-01
1.08947530e-01 -5.06235182e-01 -1.11213481e+00 -4.82833922e-01
2.73453236e-01 8.64451975e-02 -2.00953200e-01 9.24682736e-01
5.20668149e-01 2.06172884e-01 3.55088145e-01 -9.11579430e-01
-6.36074245e-01 8.68890822e-01 -3.53286900e-02 2.20618267e-02
4.95904595e-01 -3.09137672e-01 1.48256230e+00 -7.17432201e-01
6.78611219e-01 1.00907016e+00 8.06913733e-01 8.73555362e-01
-1.18044972e+00 -5.51333249e-01 7.62230530e-02 -4.05047387e-02
-1.33215022e+00 -3.03234726e-01 9.35781524e-02 -2.85967350e-01
1.23915637e+00 2.96841890e-01 -5.38038276e-02 1.02786732e+00
2.48663068e-01 7.77971029e-01 6.49939358e-01 -6.38073087e-01
1.30892634e-01 -3.09284866e-01 3.21274906e-01 7.75889516e-01
-5.55956364e-02 -2.72233665e-01 -4.28770810e-01 -5.02333879e-01
3.42162281e-01 1.13960110e-01 -2.15356335e-01 4.60053496e-02
-1.19652843e+00 6.07708097e-01 2.32201070e-01 4.87003326e-01
-4.12761509e-01 -3.94062288e-02 3.40653628e-01 1.43798605e-01
2.88559586e-01 5.90644836e-01 -1.09679031e+00 5.55366576e-02
-8.62678528e-01 -7.87912458e-02 1.09341586e+00 1.13411343e+00
7.48570263e-01 -4.43073064e-01 -4.67038244e-01 9.01553035e-01
1.57782316e-01 3.97635281e-01 5.34115970e-01 -7.54492879e-01
6.47657871e-01 5.87661505e-01 5.00497408e-03 -5.18290341e-01
-5.33929348e-01 -1.75015420e-01 -7.01206088e-01 -6.70195282e-01
3.41512024e-01 -3.47540081e-01 -1.15946639e+00 2.20361423e+00
3.10017556e-01 4.67887312e-01 4.15744364e-01 4.99923497e-01
8.21646094e-01 6.36180043e-01 5.28950751e-01 -1.81752611e-02
1.64595640e+00 -1.00951254e+00 -9.27241206e-01 -4.96037275e-01
1.11751604e+00 -5.92840612e-01 6.91316605e-01 -1.19045444e-01
-7.00165033e-01 -2.79661924e-01 -5.70996046e-01 -2.70744681e-01
-4.69546974e-01 1.52815059e-01 2.88850158e-01 4.90325809e-01
-9.22929525e-01 5.79840302e-01 -9.47215855e-01 -5.89322805e-01
-6.88388804e-03 3.77772421e-01 -5.65328479e-01 -2.86872298e-01
-1.49038112e+00 1.04941916e+00 6.48772836e-01 2.34768271e-01
-4.90896612e-01 -6.58878207e-01 -1.15844715e+00 2.56063908e-01
4.14692670e-01 -8.65973949e-01 1.44045460e+00 -7.61774182e-01
-1.32664144e+00 7.00517178e-01 -4.88387734e-01 -4.20666218e-01
-3.41663808e-02 -4.58843708e-01 -5.07526517e-01 -2.30042115e-01
2.16703966e-01 6.02344811e-01 1.14904985e-01 -7.08040059e-01
-5.47201872e-01 -2.34841824e-01 1.93200156e-01 -7.83338957e-03
-2.54352123e-01 2.34961003e-01 -4.44983453e-01 -4.82991606e-01
-6.16646893e-02 -1.00625980e+00 -7.63510764e-01 -4.15291131e-01
-7.28935838e-01 -4.56006467e-01 2.50652075e-01 -8.25766981e-01
1.51569140e+00 -2.07177544e+00 1.08249560e-01 1.10006273e-01
2.40742993e-02 2.24458992e-01 -5.25656044e-01 7.08101749e-01
-2.11102337e-01 3.82223040e-01 -5.92617750e-01 -3.39572430e-01
1.92441177e-02 6.84910536e-01 -1.84664339e-01 -3.43865827e-02
4.65334147e-01 9.73540962e-01 -9.11764860e-01 -5.96857250e-01
-3.75814885e-01 4.79170114e-01 -6.11711621e-01 5.52464604e-01
-3.14046770e-01 2.99331039e-01 -5.01699626e-01 4.60189492e-01
2.84839958e-01 -4.18422222e-01 7.12675512e-01 -7.32641816e-02
6.27536923e-02 6.46381736e-01 -9.56230164e-01 1.82481170e+00
-5.68058133e-01 2.25748628e-01 -1.64700583e-01 -8.69808316e-01
8.96139622e-01 8.08860302e-01 4.39851016e-01 -5.14297724e-01
-1.60609394e-01 3.60919565e-01 -9.78507623e-02 -7.09361136e-01
4.13737565e-01 -4.03535485e-01 -4.16247159e-01 1.24880612e-01
4.31089550e-01 5.47778189e-01 2.53100306e-01 9.61176232e-02
1.54073584e+00 2.56101549e-01 7.47595191e-01 7.43416250e-02
4.77418751e-01 -6.79484829e-02 1.20830929e+00 8.11078846e-01
9.26203057e-02 5.28360367e-01 4.20308620e-01 -3.25425446e-01
-8.79776418e-01 -7.68360317e-01 -3.79040539e-02 1.38424325e+00
-2.41284207e-01 -7.96519518e-01 -7.51680076e-01 -1.13293016e+00
-3.69802922e-01 7.71215856e-01 -5.44117033e-01 1.96230188e-01
-9.08273757e-01 -5.58495879e-01 9.47900832e-01 6.49457872e-01
-4.18125559e-03 -1.19782066e+00 -2.96493590e-01 4.88740027e-01
-7.56088972e-01 -1.41768014e+00 -8.43797982e-01 5.48857391e-01
-9.23016727e-01 -9.54049230e-01 -5.35308957e-01 -9.98004854e-01
7.57425964e-01 -3.98119360e-01 1.17676795e+00 -7.17995390e-02
7.48338830e-03 2.44518638e-01 -4.78710651e-01 -2.93544918e-01
-3.14462483e-01 4.76504773e-01 -1.58876538e-01 -1.19200356e-01
5.28697252e-01 -4.83109146e-01 -2.73805022e-01 3.73808354e-01
-1.18522060e+00 -1.02440335e-01 8.14907372e-01 1.01179504e+00
7.72497594e-01 -4.16229367e-01 5.88809013e-01 -1.41367435e+00
2.14793697e-01 -6.95199192e-01 -2.25945368e-01 7.41147459e-01
-3.51037592e-01 4.99334514e-01 8.20514739e-01 -5.01591384e-01
-1.03777611e+00 3.62495780e-01 -5.35934210e-01 -2.10321829e-01
-2.96044379e-01 8.12768936e-01 -4.34935778e-01 5.45845509e-01
5.44595301e-01 7.90527984e-02 -4.47197348e-01 -7.63675570e-01
4.21346873e-01 7.60471702e-01 7.01778054e-01 -5.98527908e-01
5.17171323e-01 1.48198500e-01 -3.06534171e-01 -3.92691046e-01
-1.35368443e+00 -7.21511126e-01 -7.79661596e-01 4.81384009e-01
1.18823564e+00 -7.23262429e-01 -5.28007567e-01 -8.45617205e-02
-1.57451546e+00 -2.41161779e-01 -2.00595140e-01 5.22928476e-01
-4.17938054e-01 4.12147433e-01 -8.59136999e-01 -7.16781676e-01
-5.73085010e-01 -8.14025640e-01 9.97107089e-01 2.58409202e-01
-2.88930267e-01 -1.05023110e+00 3.75348359e-01 2.63829261e-01
2.60577649e-01 1.80955473e-02 1.11477065e+00 -1.50205112e+00
-2.32927322e-01 -1.27530113e-01 1.53892443e-01 1.83371112e-01
1.96051478e-01 -5.35328507e-01 -7.31140554e-01 -2.95178443e-02
-1.99114099e-01 -7.47001693e-02 6.64430737e-01 8.68462473e-02
1.00845480e+00 -5.13461471e-01 -5.21775842e-01 3.39596778e-01
1.39267170e+00 1.36018202e-01 7.56939888e-01 2.66663969e-01
5.86628735e-01 7.41052508e-01 4.83563900e-01 5.90160713e-02
5.75882673e-01 5.36261022e-01 2.34761938e-01 -4.31960188e-02
1.84433520e-01 -6.57950282e-01 4.10589635e-01 1.09241378e+00
1.69725537e-01 -5.60686886e-01 -1.05177796e+00 6.74802661e-01
-1.92429519e+00 -7.88399398e-01 -3.02874949e-03 2.12472916e+00
9.89906251e-01 -1.27599105e-01 -2.73536593e-01 -4.64675695e-01
9.45726752e-01 -1.49606556e-01 -4.77937132e-01 -3.11504275e-01
1.10360496e-02 4.98183519e-01 4.69895840e-01 4.66000855e-01
-1.07454216e+00 9.57958877e-01 6.28395891e+00 6.66415572e-01
-7.32578874e-01 3.21147770e-01 1.95371360e-01 3.40648890e-01
-4.01277989e-01 2.95413464e-01 -1.14217532e+00 3.64118516e-01
1.36941290e+00 7.40721375e-02 1.29036203e-01 6.62148535e-01
-3.32024544e-02 3.67025286e-01 -1.33895266e+00 6.10593498e-01
8.22424516e-02 -1.01133227e+00 -1.93355698e-02 2.99121812e-02
5.06005764e-01 3.10283363e-01 -5.30273914e-01 5.11700928e-01
6.85516059e-01 -1.11443841e+00 3.50338638e-01 6.46776021e-01
7.97416449e-01 -4.41275984e-01 1.10201693e+00 6.18776917e-01
-1.20632911e+00 -1.63400285e-02 -3.11807007e-01 3.28806281e-01
4.79872435e-01 6.33178890e-01 -8.95864785e-01 9.99272704e-01
4.39706534e-01 5.63710153e-01 -1.03699930e-01 9.67303872e-01
-4.95108455e-01 8.48291039e-01 -4.13374484e-01 1.33422792e-01
1.20592073e-01 -2.08491515e-02 3.74606282e-01 1.66307640e+00
4.27045345e-01 3.13922197e-01 4.04573441e-01 4.64429051e-01
-4.09043849e-01 3.86758834e-01 -4.57055718e-01 -1.35783628e-01
5.82515478e-01 1.11571562e+00 -3.52047205e-01 -2.27158070e-01
-5.82387090e-01 1.10362279e+00 6.93577468e-01 2.77800232e-01
-6.98778450e-01 -7.25267231e-01 6.17515028e-01 -2.39491373e-01
5.81561565e-01 -4.77848500e-02 5.73334917e-02 -1.22753417e+00
7.29701445e-02 -9.60070014e-01 9.87721324e-01 -5.22040606e-01
-1.57285106e+00 8.81693840e-01 -3.27456385e-01 -9.83039260e-01
-1.91430107e-01 -5.62900901e-01 -3.31957519e-01 8.00805390e-01
-1.65200961e+00 -1.09135449e+00 2.93911606e-01 2.45182633e-01
4.07603621e-01 2.59669930e-01 1.25816047e+00 3.86641353e-01
-6.28997266e-01 6.26042068e-01 5.16931005e-02 7.98757553e-01
8.57446015e-01 -1.26485538e+00 6.13268852e-01 7.66590953e-01
3.56551796e-01 9.49258208e-01 4.29716319e-01 -6.58392847e-01
-9.85956490e-01 -1.18637693e+00 1.80914247e+00 -6.56382799e-01
4.32894617e-01 -4.11524534e-01 -1.04910457e+00 1.18583930e+00
2.37734005e-01 1.89678445e-01 1.34135711e+00 3.46807539e-01
-6.89405918e-01 1.46529138e-01 -1.00079560e+00 4.48941648e-01
1.14559436e+00 -7.07964003e-01 -9.82317984e-01 2.84999341e-01
1.00938475e+00 -5.01341045e-01 -1.22685754e+00 5.02510846e-01
4.10668641e-01 -3.95939350e-01 7.02925086e-01 -1.19143152e+00
2.86356896e-01 -3.76283705e-01 -4.11408544e-01 -1.14712811e+00
-2.09980756e-01 -6.57666028e-01 3.69521603e-02 1.32583022e+00
9.56075966e-01 -4.98765677e-01 5.04969597e-01 6.83344901e-01
-3.09732407e-01 -7.64755189e-01 -1.06557977e+00 -7.96339631e-01
6.87456727e-02 -4.62434292e-01 6.09709740e-01 1.28119469e+00
3.06557089e-01 6.21713698e-01 -4.76504982e-01 6.22838378e-01
2.17729598e-01 -8.61995369e-02 2.60409892e-01 -1.13579190e+00
-3.85550320e-01 2.25435257e-01 -2.24619105e-01 -1.42397630e+00
4.61209327e-01 -1.09281135e+00 5.61453402e-01 -1.72558403e+00
3.87642562e-01 -4.66233194e-01 -5.70630550e-01 1.08486927e+00
-4.68730122e-01 -1.56942412e-01 1.20098948e-01 1.09151699e-01
-8.00983548e-01 2.44420171e-01 7.03044176e-01 3.76872234e-02
-1.28481895e-01 -1.30256340e-01 -5.90161264e-01 5.56515336e-01
6.82857513e-01 -1.20434809e+00 1.67331025e-02 -6.18120790e-01
3.84934008e-01 3.48739684e-01 -1.56823441e-01 -6.12445176e-01
6.81737602e-01 -2.35763550e-01 -5.77413850e-02 -4.00929540e-01
4.36395071e-02 -7.67277241e-01 1.91086680e-01 2.78015792e-01
-7.18764484e-01 1.47292718e-01 -6.72173575e-02 6.80215955e-01
-2.23998338e-01 -5.22718787e-01 3.45920116e-01 -1.62905946e-01
-5.07657588e-01 3.18746448e-01 -2.89406449e-01 3.22639346e-01
6.77624583e-01 1.29336745e-01 -3.35158706e-01 -1.16916366e-01
-9.03031826e-01 5.32717943e-01 3.00087221e-02 3.71808201e-01
3.29762518e-01 -1.13701355e+00 -7.79259741e-01 -1.57956883e-01
3.15997779e-01 6.68914169e-02 1.79915667e-01 7.58982360e-01
-2.42198743e-02 5.69895864e-01 2.09919453e-01 -3.38261545e-01
-1.30574739e+00 4.92384374e-01 4.79992062e-01 -8.34916711e-01
-4.39911038e-01 6.95230067e-01 3.46109480e-01 -1.36775827e+00
1.50795043e-01 -3.12478065e-01 -2.98932433e-01 -3.29595804e-01
4.78565663e-01 4.89898995e-02 9.00638774e-02 -5.78017712e-01
-5.49666643e-01 3.19731563e-01 -1.75829858e-01 3.52800749e-02
1.39274275e+00 -6.14432879e-02 -3.37275177e-01 3.09909463e-01
1.03580320e+00 1.29574984e-01 -6.81581438e-01 -6.40590966e-01
7.75999188e-01 1.01052992e-01 -5.20420969e-01 -8.38530958e-01
-7.42024243e-01 6.58053339e-01 1.21057838e-01 -1.01664685e-01
1.05241489e+00 2.03115538e-01 1.02288795e+00 7.01432347e-01
1.68805614e-01 -7.59972632e-01 -2.21046790e-01 9.51002896e-01
1.93028674e-01 -7.97030866e-01 -5.36817193e-01 -5.58141887e-01
-5.38342237e-01 1.09454632e+00 6.60361350e-01 4.14296091e-01
4.71261263e-01 4.79819268e-01 2.57103324e-01 -1.62464231e-02
-8.93819034e-01 -4.07630354e-01 3.09914291e-01 4.94563401e-01
8.08296144e-01 -6.35997579e-02 -5.02021492e-01 9.96402562e-01
-8.67727548e-02 1.08499080e-01 1.78182378e-01 9.99651670e-01
-3.26779723e-01 -1.59891939e+00 -6.85595721e-02 3.55352789e-01
-9.52858210e-01 -4.85478848e-01 -1.76834032e-01 3.22770387e-01
1.65773794e-01 1.06776762e+00 -5.33175562e-03 -3.09345096e-01
6.17142379e-01 6.09570980e-01 1.79392882e-02 -1.03485918e+00
-8.34783554e-01 -1.58706337e-01 4.30382758e-01 -4.75246996e-01
-4.35450613e-01 -4.85541999e-01 -1.48753798e+00 2.72845984e-01
-6.74371481e-01 6.45741165e-01 5.17536819e-01 1.27975750e+00
5.93354523e-01 3.57422292e-01 3.89713705e-01 -5.61797656e-02
-6.11248434e-01 -9.66413021e-01 -2.47816950e-01 3.00799370e-01
2.29448304e-01 -6.34640306e-02 -5.46440631e-02 1.91232264e-01] | [9.691828727722168, 9.438106536865234] |
4fcf62e4-1d1f-4f86-a07f-259f2f7cc2d6 | joinedtrans-prior-guided-multi-task | 2305.11504 | null | https://arxiv.org/abs/2305.11504v1 | https://arxiv.org/pdf/2305.11504v1.pdf | JOINEDTrans: Prior Guided Multi-task Transformer for Joint Optic Disc/Cup Segmentation and Fovea Detection | Deep learning-based image segmentation and detection models have largely improved the efficiency of analyzing retinal landmarks such as optic disc (OD), optic cup (OC), and fovea. However, factors including ophthalmic disease-related lesions and low image quality issues may severely complicate automatic OD/OC segmentation and fovea detection. Most existing works treat the identification of each landmark as a single task, and take into account no prior information. To address these issues, we propose a prior guided multi-task transformer framework for joint OD/OC segmentation and fovea detection, named JOINEDTrans. JOINEDTrans effectively combines various spatial features of the fundus images, relieving the structural distortions induced by lesions and other imaging issues. It contains a segmentation branch and a detection branch. To be noted, we employ an encoder pretrained in a vessel segmentation task to effectively exploit the positional relationship among vessel, OD/OC, and fovea, successfully incorporating spatial prior into the proposed JOINEDTrans framework. There are a coarse stage and a fine stage in JOINEDTrans. In the coarse stage, OD/OC coarse segmentation and fovea heatmap localization are obtained through a joint segmentation and detection module. In the fine stage, we crop regions of interest for subsequent refinement and use predictions obtained in the coarse stage to provide additional information for better performance and faster convergence. Experimental results demonstrate that JOINEDTrans outperforms existing state-of-the-art methods on the publicly available GAMMA, REFUGE, and PALM fundus image datasets. We make our code available at https://github.com/HuaqingHe/JOINEDTrans | ['Xiaoying Tang', 'Pujin Cheng', 'Zhiyuan Cai', 'Li Lin', 'Huaqing He'] | 2023-05-19 | null | null | null | null | ['fovea-detection'] | ['medical'] | [ 5.91164641e-02 -6.65685907e-02 -1.11139275e-01 -1.90996170e-01
-6.33814692e-01 -3.86929929e-01 2.64910609e-01 -8.08766335e-02
-4.47223276e-01 4.94719326e-01 1.61239862e-01 -3.31477821e-01
-7.64749795e-02 -6.33560002e-01 -4.63620722e-01 -8.16093445e-01
2.87981004e-01 3.37625928e-02 5.87465167e-01 2.81574249e-01
4.11690235e-01 4.18537796e-01 -1.39904571e+00 1.74818009e-01
1.32302582e+00 1.10525894e+00 3.39846730e-01 4.80908692e-01
7.89963007e-02 4.27188784e-01 -1.92400128e-01 -3.67256731e-01
4.74141598e-01 -4.47561681e-01 -7.13360429e-01 3.60904634e-01
5.13870120e-01 -6.45237148e-01 -6.38913224e-03 1.19752979e+00
7.92406559e-01 -1.89485997e-01 4.22115028e-01 -7.54668236e-01
-3.70451510e-01 1.01863287e-01 -1.15775990e+00 4.94594902e-01
-2.48368025e-01 5.05466878e-01 6.32655203e-01 -7.23864377e-01
5.12319207e-01 8.25749397e-01 4.51903641e-01 3.82695943e-02
-1.02892840e+00 -6.12298489e-01 2.56724842e-02 1.53979272e-01
-1.38200200e+00 -4.98609722e-01 4.36165512e-01 -7.80635655e-01
5.02401948e-01 -7.53644556e-02 8.89465332e-01 2.34077841e-01
7.96811432e-02 7.22853601e-01 1.34244227e+00 -3.26839775e-01
-9.99127030e-02 -1.24832787e-01 1.13563873e-01 9.63601410e-01
2.29609415e-01 1.15866400e-01 8.42535794e-02 1.11712709e-01
1.10074234e+00 -2.04984754e-01 -4.68238443e-01 -1.90236896e-01
-9.05437648e-01 5.23534954e-01 6.64314628e-01 -5.03543206e-02
-5.24838567e-01 -1.89855650e-01 1.78544968e-01 -1.61275715e-01
2.17761040e-01 2.31842995e-01 -1.43265396e-01 1.73119813e-01
-9.00047958e-01 2.44715121e-02 4.02380154e-02 7.17062175e-01
7.86305785e-01 -4.74352032e-01 -4.80049223e-01 8.85181069e-01
4.33081329e-01 5.59418574e-02 4.86305237e-01 -9.89424050e-01
2.21400976e-01 7.90555358e-01 1.60407498e-01 -4.41589653e-01
-5.00268757e-01 -6.77377045e-01 -6.63133681e-01 4.36755091e-01
6.31349742e-01 -3.82312626e-01 -1.30130768e+00 1.08803964e+00
5.45018792e-01 4.60451663e-01 -4.20817852e-01 1.30148876e+00
6.56857729e-01 1.48000389e-01 1.18507119e-02 -1.89040288e-01
1.50854850e+00 -1.14650035e+00 -4.25432056e-01 -2.00730115e-01
6.70837402e-01 -1.10593259e+00 7.17822194e-01 1.97097331e-01
-1.25349116e+00 -6.55736327e-01 -7.50057518e-01 -5.72916389e-01
7.87708238e-02 9.59013641e-01 6.58590019e-01 3.15846741e-01
-1.18532503e+00 4.96676490e-02 -8.94893050e-01 -1.47690699e-01
7.25016057e-01 3.39687496e-01 -9.30859298e-02 -1.97667599e-01
-6.43580973e-01 7.81482100e-01 1.77205160e-01 4.06942517e-01
-5.13891101e-01 -6.25499427e-01 -7.57055402e-01 -2.83193529e-01
3.99573326e-01 -1.14988112e+00 1.19999814e+00 -6.06383920e-01
-1.38719022e+00 1.04716909e+00 -6.34102464e-01 -4.93601412e-01
5.22920310e-01 -1.30049646e-01 -1.79642230e-01 3.52149367e-01
1.38742894e-01 6.94854200e-01 8.17893207e-01 -7.30693519e-01
-1.14821255e+00 -4.88087833e-01 1.06751159e-01 2.39474028e-01
2.43137509e-01 2.23799631e-01 -9.05714214e-01 -4.01384294e-01
2.19316915e-01 -7.63765454e-01 -1.52078986e-01 3.19353610e-01
-7.26714313e-01 -2.50922978e-01 4.64851469e-01 -7.42691815e-01
1.13438284e+00 -2.08048320e+00 -1.49739072e-01 3.74845676e-02
5.27407348e-01 6.56850994e-01 -1.08799338e-01 -3.58056605e-01
1.07856311e-01 -6.31627906e-03 5.91476336e-02 -3.23320329e-01
-4.79761034e-01 -1.96523935e-01 2.09959939e-01 5.60238540e-01
3.55036378e-01 8.38430285e-01 -6.58228755e-01 -6.30968451e-01
3.81459892e-01 5.60005486e-01 -4.77021784e-01 -2.74968892e-02
-1.86355002e-02 5.09838521e-01 -6.21926665e-01 8.98656011e-01
7.07656980e-01 -5.41044831e-01 -1.37355402e-01 -4.70534712e-01
-5.82846165e-01 2.11816400e-01 -9.86436069e-01 1.50378823e+00
-2.07758576e-01 6.16876721e-01 6.89225346e-02 -6.37979507e-01
7.73735046e-01 1.17359236e-01 4.50844944e-01 -7.10272491e-01
5.10687530e-01 4.67864215e-01 5.51139474e-01 -5.82497478e-01
9.47913006e-02 1.62675649e-01 6.52925014e-01 1.22874856e-01
-1.15023993e-01 2.10421771e-01 4.55923170e-01 -2.61936843e-01
5.87645292e-01 3.51404011e-01 3.94604146e-01 8.94719213e-02
6.67237937e-01 -1.98560990e-02 7.84246325e-01 2.44750574e-01
-4.78446573e-01 9.94075596e-01 6.74009025e-01 -2.60965317e-01
-9.41024005e-01 -7.88503587e-01 -5.69718480e-01 3.60526145e-01
3.70168447e-01 -3.63144040e-01 -7.23825753e-01 -3.66897523e-01
-3.09293866e-02 1.17889065e-02 -5.78996539e-01 2.53258526e-01
-4.41569746e-01 -9.24370944e-01 1.81031138e-01 3.84151876e-01
9.29815590e-01 -3.98572683e-01 -5.68851590e-01 1.48353875e-01
-8.67233276e-02 -1.01876259e+00 -6.90393209e-01 -5.41372955e-01
-7.82766640e-01 -1.36104131e+00 -1.09288335e+00 -9.25474524e-01
8.07053685e-01 4.02052015e-01 5.20526052e-01 -4.46773395e-02
-7.72502720e-01 -2.79996812e-01 -6.12897892e-03 -3.06824207e-01
7.22760111e-02 -1.98937684e-01 -4.61525053e-01 4.55595106e-01
3.48794222e-01 -4.65185314e-01 -1.29463828e+00 5.40458143e-01
-4.01187003e-01 2.08504841e-01 9.95741069e-01 7.55364597e-01
8.28773141e-01 3.59616950e-02 2.42578894e-01 -5.81983924e-01
3.92403364e-01 -2.24275663e-01 -1.00656378e+00 1.79336280e-01
-5.49762130e-01 -3.69509667e-01 1.51603788e-01 -1.32011116e-01
-9.36574936e-01 2.02472374e-01 5.11030331e-02 -4.15409654e-01
-2.63421208e-01 4.02483940e-01 1.19282365e-01 -3.44117254e-01
5.25331438e-01 1.17634058e-01 4.14623260e-01 -5.18264592e-01
9.20808688e-02 8.50817084e-01 6.10260546e-01 -1.13412283e-01
3.04194540e-01 6.02536798e-01 -2.38607964e-03 -5.86938322e-01
-8.68593037e-01 -6.95670962e-01 -6.58539414e-01 -1.30347416e-01
8.53823841e-01 -9.95660186e-01 -7.19965339e-01 7.32477367e-01
-8.95752788e-01 -2.16016814e-01 -5.16668847e-03 7.35439420e-01
-2.00183913e-01 3.57992470e-01 -5.26216507e-01 -3.94551814e-01
-2.84748495e-01 -1.50115848e+00 1.08543205e+00 8.34694982e-01
2.16037586e-01 -7.66443074e-01 -2.45046586e-01 6.52543426e-01
3.72498691e-01 1.53224170e-01 8.03894818e-01 -6.52314872e-02
-1.05784047e+00 -3.44562232e-02 -6.91640019e-01 3.92360866e-01
1.52860329e-01 2.70344496e-01 -8.43829751e-01 -2.13360116e-01
-3.27854723e-01 1.95359699e-02 1.00926852e+00 1.07070017e+00
9.67543006e-01 1.62910789e-01 -3.99088502e-01 1.05786681e+00
1.32571149e+00 2.79297024e-01 7.43522286e-01 4.29541200e-01
5.12271464e-01 5.31079590e-01 7.42625415e-01 3.22921336e-01
5.84831715e-01 6.39084160e-01 3.33619207e-01 -6.07216001e-01
-7.02036381e-01 1.75141767e-01 -4.79402468e-02 2.17974707e-01
-4.10829157e-01 1.45904096e-02 -1.05088103e+00 7.52592504e-01
-1.61626077e+00 -4.04819578e-01 -2.18462497e-01 2.25970912e+00
8.83856297e-01 -1.12037778e-01 2.29937226e-01 -3.34320635e-01
8.78628850e-01 -1.70751706e-01 -7.65840888e-01 2.03355283e-01
-1.29566789e-01 1.37715846e-01 5.55416226e-01 3.98540050e-01
-1.24664772e+00 8.42489243e-01 5.18805981e+00 5.32869935e-01
-1.42018735e+00 -4.86407876e-02 7.80409932e-01 -3.36509526e-01
2.08381981e-01 6.00923114e-02 -1.04938889e+00 5.29335618e-01
1.96834177e-01 -2.89144721e-02 1.47611737e-01 2.74625957e-01
4.19593960e-01 -4.93781745e-01 -5.98143876e-01 9.21341300e-01
-1.89232126e-01 -1.39235783e+00 -2.05360144e-01 2.96905100e-01
6.14458680e-01 2.83935785e-01 2.14665219e-01 -3.06212336e-01
-8.92063528e-02 -8.77718925e-01 3.36070620e-02 7.55493701e-01
7.84396052e-01 -4.72907454e-01 8.53548646e-01 -6.69349208e-02
-9.69924331e-01 -1.76400989e-01 -2.69215673e-01 2.86682636e-01
4.89933491e-02 5.81321836e-01 -8.31591010e-01 4.67127144e-01
5.99905849e-01 9.63766575e-01 -7.84474611e-01 1.95836377e+00
-3.72314572e-01 3.55308145e-01 -1.98847562e-01 5.66502869e-01
2.00743482e-01 -5.56389511e-01 7.72382379e-01 5.40283203e-01
3.10502619e-01 3.42521854e-02 7.11837262e-02 1.06205773e+00
1.13557912e-01 2.31393889e-01 8.49859696e-03 3.11012983e-01
4.62856382e-01 1.29938722e+00 -6.22659147e-01 -4.47374508e-02
-6.56182826e-01 6.04749024e-01 5.75778596e-02 6.59509301e-01
-6.91779017e-01 -4.54631239e-01 6.25404477e-01 4.35907751e-01
2.08658069e-01 3.15322429e-02 -3.44141245e-01 -1.01994646e+00
2.64361233e-01 -5.09869516e-01 3.42704535e-01 -7.74915099e-01
-6.85963809e-01 5.16195774e-01 -3.87253821e-01 -1.40187669e+00
4.46077399e-02 -6.22307599e-01 -7.56130099e-01 1.27528024e+00
-2.07860994e+00 -1.23154914e+00 -4.31784928e-01 4.99188513e-01
2.10921243e-01 -1.83573127e-01 3.07919681e-01 3.57122153e-01
-1.14397049e+00 4.26635653e-01 1.93960108e-02 2.91492641e-01
8.85719180e-01 -1.17504275e+00 9.70958620e-02 1.15042460e+00
-4.38834101e-01 7.29354739e-01 2.46640146e-02 -6.07972026e-01
-8.26142907e-01 -1.16467083e+00 5.64875066e-01 4.05221283e-02
4.39155012e-01 1.89786285e-01 -7.28341103e-01 6.23222709e-01
-2.42815167e-03 3.60983759e-01 4.21562433e-01 -2.91581091e-04
1.04418300e-01 -2.26588681e-01 -7.92775512e-01 5.60127378e-01
6.73827291e-01 -2.41494179e-01 -2.39286944e-01 3.23065579e-01
2.91243851e-01 -7.01083899e-01 -9.21752572e-01 3.49269211e-01
4.75608826e-01 -1.16724968e+00 8.58575821e-01 -1.11067273e-01
2.73540974e-01 -5.71984887e-01 5.88572323e-01 -9.73957062e-01
-2.38695174e-01 -9.29517567e-01 1.36777788e-01 1.03216231e+00
3.53170007e-01 -9.76326168e-01 5.66646874e-01 4.87812221e-01
-4.15205091e-01 -8.88014793e-01 -7.12877989e-01 -2.18190283e-01
-8.70359764e-02 1.00111425e-01 3.99336547e-01 6.04665875e-01
-4.39016402e-01 -1.86595321e-02 -2.43401472e-02 4.57496554e-01
6.64004326e-01 5.59435487e-01 6.79154932e-01 -1.12197042e+00
-6.37208149e-02 -5.95953763e-01 -5.13381660e-01 -1.14586830e+00
-4.80654806e-01 -6.72828376e-01 -3.55204165e-01 -1.72307587e+00
1.25330403e-01 -4.55833375e-01 -3.45827162e-01 5.31098604e-01
-3.11793745e-01 3.83033007e-01 -7.72785470e-02 4.46504772e-01
-2.22504064e-01 1.95003703e-01 1.98954749e+00 2.59712756e-01
-4.44291532e-01 2.91905165e-01 -8.48648250e-01 9.20914948e-01
9.01017487e-01 -5.84463552e-02 -4.38698590e-01 -5.75479388e-01
-2.45036051e-01 1.40613511e-01 7.43455648e-01 -8.53734374e-01
6.52355909e-01 2.81638920e-01 3.85012150e-01 -5.30557811e-01
2.18278617e-01 -2.28434950e-01 -3.83324921e-01 2.56895244e-01
8.24655034e-03 -6.43056691e-01 1.77773118e-01 3.40730101e-01
-4.62044418e-01 1.00983409e-02 1.10548377e+00 4.45532873e-02
-6.33052468e-01 5.28679729e-01 -5.50570562e-02 -5.35792820e-02
1.17034435e+00 -3.37160230e-01 -6.55289292e-01 8.18566978e-02
-9.28344429e-01 6.22494400e-01 4.97848511e-01 2.09392637e-01
6.64266586e-01 -6.55136824e-01 -9.49921310e-01 5.73100865e-01
-1.70715787e-02 3.16599548e-01 4.41008955e-01 1.72976565e+00
-6.12835884e-01 4.74482626e-01 -3.30584943e-01 -7.66586542e-01
-1.24163914e+00 7.91412815e-02 8.91816497e-01 2.96650290e-01
-6.98708892e-01 8.93706501e-01 3.66762042e-01 2.03425169e-01
2.69420028e-01 -6.88629568e-01 -4.71039057e-01 1.01311743e-01
6.40129566e-01 3.57173026e-01 -5.92679866e-02 -5.90223670e-01
-8.71187449e-03 9.66287017e-01 -5.24947166e-01 4.46339518e-01
8.92773807e-01 -4.78273332e-01 -2.90085018e-01 -1.84863657e-01
9.37655807e-01 -5.47315180e-03 -1.39551997e+00 -5.88869870e-01
-3.82366419e-01 -7.30289280e-01 6.87746525e-01 -9.53411281e-01
-1.26542258e+00 1.08354533e+00 8.17502975e-01 -1.25164241e-01
1.47172642e+00 -1.71507508e-01 9.10800755e-01 -3.41244400e-01
1.33010879e-01 -7.33136177e-01 -3.87130082e-01 9.59165916e-02
6.30876422e-01 -1.31503177e+00 -3.60489413e-02 -7.84938157e-01
-6.34716332e-01 1.23257542e+00 7.58623481e-01 3.89892086e-02
5.51217079e-01 1.20423865e-02 2.31407017e-01 -8.44042525e-02
-3.62311602e-01 -8.13191533e-01 7.46118307e-01 5.35723925e-01
4.37225014e-01 -2.02378482e-01 -3.18741739e-01 2.88688987e-01
-3.79141420e-04 2.95000523e-01 5.64273059e-01 5.75855732e-01
-5.28515935e-01 -1.06194520e+00 -1.13298573e-01 6.45258069e-01
-5.71236789e-01 -2.74003595e-01 1.42629640e-02 8.18989933e-01
5.23985267e-01 8.00996006e-01 3.59843194e-01 5.85965402e-02
7.21976086e-02 -2.92056799e-01 3.26885402e-01 -6.92384064e-01
-3.12116891e-01 7.82447875e-01 5.94776645e-02 -7.30331898e-01
-3.88212383e-01 -8.29975486e-01 -1.17987156e+00 1.46465376e-01
-3.41631740e-01 -1.95860744e-01 3.78681660e-01 7.06767738e-01
6.39050722e-01 4.54075038e-01 4.29986089e-01 -5.03827691e-01
-1.95580557e-01 -9.07141685e-01 -7.47103035e-01 -5.24162538e-02
6.69714391e-01 -6.81781888e-01 -5.86649366e-02 9.26008075e-03] | [15.769103050231934, -3.965282440185547] |
5fbd7a3e-0b5d-4b3e-8ebd-f4afac308f2a | adaptive-personlization-in-federated-learning | 2207.03448 | null | https://arxiv.org/abs/2207.03448v1 | https://arxiv.org/pdf/2207.03448v1.pdf | Adaptive Personlization in Federated Learning for Highly Non-i.i.d. Data | Federated learning (FL) is a distributed learning method that offers medical institutes the prospect of collaboration in a global model while preserving the privacy of their patients. Although most medical centers conduct similar medical imaging tasks, their differences, such as specializations, number of patients, and devices, lead to distinctive data distributions. Data heterogeneity poses a challenge for FL and the personalization of the local models. In this work, we investigate an adaptive hierarchical clustering method for FL to produce intermediate semi-global models, so clients with similar data distribution have the chance of forming a more specialized model. Our method forms several clusters consisting of clients with the most similar data distributions; then, each cluster continues to train separately. Inside the cluster, we use meta-learning to improve the personalization of the participants' models. We compare the clustering approach with classical FedAvg and centralized training by evaluating our proposed methods on the HAM10k dataset for skin lesion classification with extreme heterogeneous data distribution. Our experiments demonstrate significant performance gain in heterogeneous distribution compared to standard FL methods in classification accuracy. Moreover, we show that the models converge faster if applied in clusters and outperform centralized training while using only a small subset of data. | ['Nassir Navab', 'Maximilian Frantzen', 'Richard Gaus', 'Johann Boschmann', 'Azade Farshad', 'Yousef Yeganeh'] | 2022-07-07 | null | null | null | null | ['skin-lesion-classification'] | ['medical'] | [-3.09775472e-01 8.40313360e-02 -3.33775342e-01 -5.08908212e-01
-8.83804739e-01 -4.59930032e-01 9.28154290e-02 1.79429665e-01
-3.18774581e-01 5.97519338e-01 9.64212939e-02 -2.81245083e-01
-5.12027681e-01 -5.08200288e-01 -3.72932404e-01 -1.22301447e+00
-1.96439579e-01 8.10674489e-01 7.31908455e-02 3.49279165e-01
-1.81766897e-01 4.66551363e-01 -1.14433444e+00 7.92990148e-01
9.80583668e-01 9.62589920e-01 7.21696168e-02 3.64434272e-01
-1.67277157e-01 9.15350616e-01 -6.91353142e-01 -4.64176953e-01
4.66227800e-01 -2.89146721e-01 -8.98822725e-01 1.55257195e-01
2.46093944e-01 -2.54064500e-01 -1.14574432e-01 8.19482267e-01
7.46857941e-01 -7.85232261e-02 4.97641593e-01 -1.48030138e+00
-1.47138029e-01 1.03285074e+00 -5.80615997e-01 -2.60174811e-01
-2.34026790e-01 -9.58155096e-02 5.11939168e-01 -1.97438270e-01
7.23497689e-01 7.51860619e-01 8.21425378e-01 6.22251451e-01
-1.27668417e+00 -6.77795291e-01 1.25409141e-01 2.35547945e-01
-1.39776289e+00 -1.53080359e-01 7.23704398e-01 -3.69769305e-01
-8.22914578e-03 5.51799834e-01 2.11168259e-01 1.04189444e+00
1.47279665e-01 7.20402539e-01 1.21907794e+00 -3.34461033e-01
6.46663189e-01 7.13381708e-01 8.94454867e-02 4.13600445e-01
2.16500282e-01 -3.47782701e-01 -2.66384721e-01 -8.36878061e-01
2.25237995e-01 4.48378235e-01 -3.04013878e-01 -9.20757532e-01
-9.28753197e-01 7.56165683e-01 4.47869420e-01 3.93371761e-01
-3.80375177e-01 -3.34271431e-01 5.01889884e-01 1.88587472e-01
9.52386633e-02 4.64820676e-02 -4.89020646e-01 3.96967143e-01
-1.00267577e+00 -7.60528222e-02 1.04597735e+00 8.65257919e-01
6.82954788e-01 -5.36533952e-01 -3.84154260e-01 6.01863682e-01
-9.79381576e-02 -1.12832725e-01 9.45566535e-01 -1.04478478e+00
2.28005886e-01 7.37246335e-01 -1.23053733e-02 -9.14835513e-01
-2.48997360e-01 -5.97553730e-01 -1.27568257e+00 1.77965969e-01
4.37131226e-01 -5.80905557e-01 -6.03065133e-01 1.59377158e+00
6.35921419e-01 1.64283603e-01 -9.00328532e-02 5.68066478e-01
2.86967456e-01 1.98365241e-01 2.23220941e-02 -3.16993207e-01
7.75106490e-01 -1.06140661e+00 -6.48653328e-01 5.88736534e-01
8.10721576e-01 -4.83883977e-01 5.43159008e-01 5.81615567e-01
-8.36862326e-01 -3.11937988e-01 -4.54014927e-01 5.85269451e-01
-1.26550108e-01 3.30329537e-02 6.27521694e-01 1.06526959e+00
-1.19590294e+00 6.38510108e-01 -8.60483587e-01 -3.26394200e-01
8.33855450e-01 5.98478317e-01 -3.12801480e-01 -2.01254249e-01
-5.95394254e-01 3.80570084e-01 1.33496866e-01 -2.79229403e-01
-9.15894091e-01 -1.00655091e+00 -2.18797728e-01 6.01214804e-02
2.48943821e-01 -7.32193410e-01 9.55687821e-01 -9.47086990e-01
-1.23013651e+00 6.83672249e-01 1.32384807e-01 -4.95735079e-01
1.11266196e+00 4.56712782e-01 -3.34423691e-01 5.88207655e-02
-7.67548829e-02 2.74534971e-01 8.46495390e-01 -1.43702316e+00
-8.96097541e-01 -6.53926730e-01 -4.73234743e-01 -8.44284147e-02
-6.89410388e-01 -1.88934624e-01 -2.92453855e-01 -2.56646007e-01
-8.10644925e-02 -9.20570910e-01 -7.25268483e-01 -6.34608418e-02
-4.88057017e-01 -7.77524114e-02 9.58515525e-01 -4.27708894e-01
1.22051954e+00 -2.27058792e+00 -8.21607262e-02 7.52739429e-01
6.35968506e-01 -1.58934407e-02 1.32268533e-01 3.04017276e-01
-2.94496901e-02 1.72252625e-01 8.76721069e-02 -5.93147933e-01
-1.16487525e-01 2.10947007e-01 2.23018974e-01 5.50027013e-01
-6.87448919e-01 2.51209050e-01 -5.38491070e-01 -9.74838853e-01
-4.07469347e-02 1.33870006e-01 -8.38596225e-01 4.23243701e-01
6.58833757e-02 8.05860698e-01 -5.75146079e-01 6.40534222e-01
8.81705761e-01 -4.87988472e-01 9.03597176e-01 -1.86614573e-01
8.39529559e-02 -5.63935637e-01 -1.28679240e+00 1.64816904e+00
-4.28185999e-01 -1.77568793e-01 5.83541274e-01 -1.17350197e+00
7.84338355e-01 4.42168534e-01 1.23113441e+00 -1.54782921e-01
2.79176921e-01 2.35821679e-01 1.13583151e-02 -4.55926389e-01
-4.05850260e-05 1.06660090e-01 1.25126198e-01 6.64346635e-01
5.34500405e-02 5.75711131e-01 -4.44976509e-01 2.49979943e-01
1.20583403e+00 -5.57137489e-01 -5.60793020e-02 -1.95722014e-01
1.71382412e-01 -1.82120696e-01 7.50842094e-01 1.05290091e+00
-5.48854530e-01 5.77720344e-01 4.34403807e-01 -5.36150634e-01
-7.91301548e-01 -1.02478337e+00 -2.33273908e-01 1.14016461e+00
1.15687707e-02 -3.29252064e-01 -7.32921600e-01 -1.14523780e+00
5.53279184e-02 4.28827912e-01 -7.26448536e-01 -1.01907916e-01
-3.72621417e-01 -8.31410348e-01 4.88554269e-01 1.37537882e-01
4.33204949e-01 -6.89422011e-01 -5.54446697e-01 4.52423766e-02
-2.11567611e-01 -5.85663617e-01 -4.65502501e-01 3.68680209e-01
-8.20222855e-01 -1.08345437e+00 -7.16249704e-01 -9.58739161e-01
8.58554721e-01 2.62709670e-02 8.13663483e-01 3.31489965e-02
-6.80688858e-01 4.61917609e-01 -4.06718880e-01 -2.12258518e-01
-5.82991600e-01 2.75054216e-01 -1.35120705e-01 5.96588254e-01
-2.89878082e-02 -6.78048730e-01 -9.10417736e-01 5.14242351e-01
-9.39279199e-01 -4.31329340e-01 4.65446562e-01 9.28487718e-01
4.70090032e-01 3.92386079e-01 4.71486360e-01 -1.57204902e+00
3.49208683e-01 -9.73936260e-01 -1.26474306e-01 6.32417917e-01
-8.99573028e-01 -1.72710374e-01 8.94289434e-01 -5.07076740e-01
-1.34937441e+00 4.26295817e-01 3.59129667e-01 -6.88218474e-01
-2.99011290e-01 1.77054197e-01 -3.42273921e-01 -2.40189746e-01
8.16849947e-01 8.75740312e-03 4.39089090e-01 -6.22399092e-01
3.09147924e-01 8.90852094e-01 2.59402841e-01 -7.45102644e-01
4.19796288e-01 5.91834784e-01 -2.51239866e-01 -1.74072072e-01
-2.83210725e-01 -5.49179018e-01 -5.75382829e-01 -3.54530215e-02
4.86185372e-01 -6.90886319e-01 -7.58693933e-01 5.21892667e-01
-6.30189478e-01 -1.34403750e-01 -3.74378413e-01 3.15722615e-01
-3.96205485e-01 3.78040880e-01 -4.64281976e-01 -5.34492970e-01
-4.34939802e-01 -1.06924772e+00 5.70591748e-01 1.92385048e-01
-7.37326890e-02 -1.04934323e+00 1.85327441e-01 5.24283648e-01
6.72235489e-01 4.50874180e-01 1.03057969e+00 -1.04447317e+00
-3.51795346e-01 -3.07387084e-01 3.73597175e-01 3.30026925e-01
3.85418683e-01 -2.14006111e-01 -9.12892759e-01 -7.87052035e-01
1.00653946e-01 -3.49367797e-01 5.58763564e-01 3.67502064e-01
1.83707905e+00 -5.56098223e-01 -8.32647324e-01 8.44806492e-01
1.46776915e+00 2.57407278e-01 4.32489425e-01 1.94158643e-01
5.68034410e-01 6.89216912e-01 2.77727395e-01 9.27517653e-01
2.09341526e-01 3.16393554e-01 4.75639999e-01 -2.20798925e-01
1.81732431e-01 -7.28418976e-02 -9.91717279e-02 5.35968840e-01
1.86571330e-01 -2.60493103e-02 -8.39677870e-01 5.56009769e-01
-2.04730320e+00 -8.63617063e-01 2.56938547e-01 2.45461726e+00
8.72104406e-01 -4.90171552e-01 3.37432295e-01 -2.63535142e-01
1.08719420e+00 -2.46528789e-01 -7.32036054e-01 -1.46105900e-01
3.75916846e-02 1.49724662e-01 6.33090973e-01 -1.41963204e-02
-9.30703759e-01 3.70983988e-01 5.98560762e+00 8.83174598e-01
-1.22632706e+00 4.76995230e-01 1.15258086e+00 -3.43235165e-01
-1.58187822e-01 -2.24930689e-01 -4.25670832e-01 6.77166164e-01
7.83211470e-01 -2.39944100e-01 2.31704473e-01 1.09843791e+00
-9.08810943e-02 3.09586972e-01 -1.11619508e+00 1.08651006e+00
-1.27711251e-01 -1.54325497e+00 1.68639552e-02 2.25327298e-01
1.13707221e+00 3.55744958e-02 3.01895291e-01 6.73251525e-02
6.44700825e-01 -9.14307177e-01 2.46152639e-01 4.05298173e-01
5.86546600e-01 -9.04329956e-01 7.34526157e-01 6.36240661e-01
-6.97874188e-01 -5.39882541e-01 -4.26865667e-01 7.07210243e-01
-3.50136995e-01 5.08343875e-01 -9.29699302e-01 8.30447197e-01
1.09096539e+00 2.71896739e-02 -6.47365451e-01 1.21556401e+00
6.15218282e-01 6.44310117e-01 -2.44564295e-01 2.12538138e-01
-1.17254533e-01 -1.28586888e-01 1.89129308e-01 8.67490649e-01
2.59535372e-01 -8.86644870e-02 4.26292390e-01 4.89570349e-01
-6.54115379e-02 4.71639842e-01 -3.64600658e-01 4.03847039e-01
6.68076217e-01 1.51626623e+00 -2.74369687e-01 -4.55559082e-02
-1.90268144e-01 8.50806594e-01 5.11871159e-01 1.24410532e-01
-7.13757157e-01 -1.84075370e-01 3.88090491e-01 2.89799511e-01
3.21674168e-01 3.57646406e-01 -1.48913264e-01 -1.01231754e+00
-1.02998912e-01 -1.08471549e+00 1.14131141e+00 4.65102587e-03
-1.68553817e+00 8.45575452e-01 -3.28902096e-01 -1.27345371e+00
-1.70332655e-01 -1.14888571e-01 -8.68799329e-01 6.09408855e-01
-9.67067957e-01 -1.46894789e+00 -4.37107295e-01 1.37265742e+00
-1.49528518e-01 -5.37068605e-01 9.36179519e-01 4.25979227e-01
-5.82367480e-01 1.29803061e+00 7.32947707e-01 1.43996164e-01
1.02704954e+00 -1.07515919e+00 -8.12583208e-01 3.57021183e-01
-2.18971848e-01 6.77578866e-01 2.05746233e-01 -4.75853831e-01
-1.26295817e+00 -1.46766627e+00 4.48338151e-01 -1.52881950e-01
2.38866970e-01 -1.82739899e-01 -7.04257846e-01 7.39647627e-01
2.03136906e-01 3.15559536e-01 1.26225734e+00 3.72268856e-01
-2.38346130e-01 -6.11221552e-01 -1.86437988e+00 3.12928915e-01
5.64274430e-01 1.05419643e-02 6.71924800e-02 5.82020521e-01
4.67618167e-01 -1.47987783e-01 -1.19162738e+00 7.47367218e-02
2.87942469e-01 -1.14905739e+00 5.67308247e-01 -8.05277586e-01
-1.86074190e-02 -7.83848576e-03 -2.28143945e-01 -1.34749472e+00
-5.01780033e-01 -1.01622164e+00 2.32450157e-01 1.42686260e+00
2.11303636e-01 -7.01256990e-01 1.35847485e+00 9.21421230e-01
8.75458717e-02 -7.82330275e-01 -1.05246270e+00 -6.11295700e-01
3.77126455e-01 4.03409712e-02 8.42863560e-01 1.26925445e+00
1.82434052e-01 -3.11050117e-01 -3.98980558e-01 2.58167773e-01
1.04881132e+00 4.14707422e-01 7.97396660e-01 -1.10720277e+00
-5.23959160e-01 -1.19465835e-01 -2.29453877e-01 -2.11433038e-01
7.67162070e-02 -1.03456724e+00 -3.23795199e-01 -1.02575016e+00
4.71242368e-01 -1.11194980e+00 -5.94676495e-01 7.19096541e-01
3.32266353e-02 -3.20249170e-01 1.24726281e-01 4.96914268e-01
-8.79179478e-01 1.31540909e-01 1.05981195e+00 -1.46596774e-01
-4.04944152e-01 4.70823705e-01 -7.25734413e-01 3.54410350e-01
8.75740111e-01 -4.73501354e-01 -4.16794717e-01 -2.40531072e-01
-5.62072515e-01 1.32039249e-01 1.96666166e-01 -1.01681459e+00
7.83314407e-01 -1.24177150e-02 6.04568303e-01 -4.27504659e-01
-2.35485628e-01 -1.31588686e+00 6.37598217e-01 6.41744435e-01
-3.72465640e-01 -3.50414902e-01 -3.03167373e-01 5.33470929e-01
-1.47157788e-01 2.90345345e-02 1.01318014e+00 -2.91089177e-01
-1.19017020e-01 6.62342787e-01 -1.93985209e-01 -1.15706183e-01
1.50198746e+00 -1.05599500e-01 -2.07711443e-01 -3.66647661e-01
-1.21808755e+00 4.95704442e-01 4.88674551e-01 1.22489341e-01
2.64231414e-01 -1.08106971e+00 -6.94683313e-01 1.75612926e-01
-1.63899198e-01 1.53356031e-01 7.91099966e-01 9.81868327e-01
-2.39012808e-01 3.55238700e-03 -3.23022693e-01 -7.32095242e-01
-1.26337230e+00 7.84150243e-01 5.48638523e-01 -4.13293928e-01
-4.67431039e-01 5.72596192e-01 1.41148835e-01 -8.30010593e-01
6.90759242e-01 1.62474424e-01 1.59629017e-01 1.15184970e-01
3.84224057e-01 6.52916908e-01 2.45387271e-01 -7.82279968e-02
-3.92589897e-01 -2.91310083e-02 -4.86500621e-01 3.39228421e-01
1.53418529e+00 -6.28085807e-03 -3.36421311e-01 -1.99122746e-02
1.39039850e+00 4.49947715e-02 -1.04259825e+00 -3.89055610e-01
-2.38238379e-01 -5.48352480e-01 -1.52116209e-01 -9.15430427e-01
-1.61437273e+00 3.76362413e-01 9.64424729e-01 3.53861973e-02
1.31064188e+00 1.54639751e-01 6.01546705e-01 7.81397149e-02
7.78109908e-01 -9.90066826e-01 -1.02415875e-01 -2.13933378e-01
2.90524334e-01 -9.93585825e-01 -1.69014588e-01 -2.33512014e-01
-8.20137560e-01 8.83333802e-01 7.28781998e-01 2.22012937e-01
9.14669275e-01 4.06923980e-01 2.88093418e-01 2.35453956e-02
-7.23792613e-01 5.84406137e-01 -3.49372715e-01 8.41033280e-01
-1.46479338e-01 2.64963120e-01 -2.06144191e-02 1.00314009e+00
-5.92045020e-03 -1.52220149e-02 3.02084267e-01 9.83422160e-01
-5.74698560e-02 -1.39933670e+00 -5.53145528e-01 5.57353139e-01
-5.55460274e-01 3.73632818e-01 -2.90735871e-01 5.54879546e-01
5.29349327e-01 8.73427391e-01 -3.88383791e-02 -3.99933249e-01
8.03822577e-02 6.31394163e-02 2.84046471e-01 -4.15629625e-01
-1.02382910e+00 1.31811038e-01 -4.59421456e-01 -6.65718675e-01
-4.39814329e-02 -7.70165622e-01 -1.03561747e+00 -4.63769555e-01
-1.89188734e-01 5.10137796e-01 5.81781507e-01 4.78710234e-01
6.84371650e-01 3.81154090e-01 1.51779926e+00 -3.90943497e-01
-1.08210969e+00 -4.34588969e-01 -9.84524250e-01 6.30489469e-01
1.04437821e-01 -8.84925202e-02 -4.58473086e-01 -3.43238674e-02] | [6.0947136878967285, 6.445560455322266] |
5cdf6eed-841e-40b5-b403-c9bee4766beb | review-of-medical-data-analysis-based-on | 2212.02234 | null | https://arxiv.org/abs/2212.02234v1 | https://arxiv.org/pdf/2212.02234v1.pdf | Review of medical data analysis based on spiking neural networks | Medical data mainly includes various biomedical signals and medical images, and doctors can make judgments on the physical condition of patients through medical data. However, the interpretation of medical data requires a lot of labor costs and may be misjudged, so many scholars use neural networks and deep learning to classify and study medical data, thereby improving doctors' work efficiency and accuracy, achieving early detection of diseases and early diagnosis, so it has a wide range of application prospects. However, traditional neural networks have disadvantages such as high energy consumption and high latency (slow calculation speed). This paper introduces the research on signal classification and disease diagnosis based on the third-generation neural network - pulse neural network in recent years, using medical data, such as electroencephalogram (EEG), electrocardiogram (ECG), electromyography (EMG), magnetic resonance imaging (MRI), etc., summarizes the advantages and disadvantages of pulse neural networks compared with traditional networks, and looks forward to the future development direction. | ['D. Zhao', 'L. Wang', 'X. Li'] | 2022-11-13 | null | null | null | null | ['electromyography-emg'] | ['medical'] | [ 3.38100672e-01 -3.21622074e-01 -1.58145964e-01 -3.01220119e-01
1.91057086e-01 2.07443893e-01 -1.57296017e-01 3.07569164e-03
-6.14630282e-01 9.36836183e-01 -1.60967499e-01 -2.03152806e-01
-6.12366021e-01 -7.60579705e-01 6.37279749e-02 -7.73677945e-01
-3.06930006e-01 4.18513507e-01 -9.38888788e-02 1.11478128e-01
3.86730283e-01 4.03458714e-01 -1.04446054e+00 2.90916190e-02
1.03017330e+00 1.24564636e+00 2.32436821e-01 3.42368037e-01
-4.97134067e-02 6.97832048e-01 -1.06317675e+00 1.02169529e-01
-4.07175899e-01 -5.74996829e-01 -6.68615878e-01 -1.30755022e-01
-6.94791675e-01 -2.27782115e-01 -6.06611371e-01 1.29730284e+00
9.72496390e-01 -1.90086253e-02 4.52820688e-01 -1.18307412e+00
-9.90693152e-01 3.68083656e-01 -4.01309341e-01 6.08952582e-01
1.23014450e-01 2.70001516e-02 -1.20213546e-01 -4.08491969e-01
7.33072236e-02 6.93214357e-01 1.02351820e+00 6.27927840e-01
-6.36004984e-01 -7.90284038e-01 -3.37505966e-01 7.83034265e-01
-1.24353182e+00 -5.47612384e-02 6.65362775e-01 -3.02822918e-01
7.40598857e-01 -6.30346639e-03 9.29617941e-01 1.26062226e+00
7.72594690e-01 6.34532452e-01 1.01165891e+00 -9.14760306e-02
3.78772706e-01 -1.94507927e-01 2.80064851e-01 4.08316135e-01
3.13713938e-01 2.95346856e-01 -2.71127760e-01 -2.43635755e-02
9.75126803e-01 4.34574574e-01 -8.73589218e-01 4.06069070e-01
-1.27742290e+00 5.16125977e-01 4.28109258e-01 9.70900953e-01
-9.09876168e-01 -1.22594275e-01 5.27723074e-01 3.35711032e-01
1.33465916e-01 4.06700104e-01 -5.50250173e-01 -5.56033850e-01
-8.94899786e-01 -1.37042925e-01 5.46203196e-01 2.91511297e-01
-4.11341758e-03 4.53831732e-01 7.45216617e-03 8.43347132e-01
3.70833129e-02 6.06133223e-01 1.29368627e+00 -8.20422113e-01
7.89019745e-03 3.15720618e-01 -2.47252882e-01 -1.26503158e+00
-8.64946365e-01 -4.62938905e-01 -1.62281156e+00 2.00004146e-01
-1.56809524e-01 -6.40999377e-01 -1.05718529e+00 1.27376437e+00
-2.04221457e-01 6.16283901e-02 5.72714359e-02 1.06582260e+00
1.18609548e+00 7.84934700e-01 8.78542662e-02 -7.10225284e-01
1.21124828e+00 -4.94607061e-01 -1.40247548e+00 -7.27492645e-02
1.26859620e-01 -1.87398210e-01 6.18793786e-01 9.57944214e-01
-1.11600900e+00 -7.20062435e-01 -9.37112331e-01 4.35125530e-01
-3.00375134e-01 1.28247455e-01 7.34578073e-01 3.51601928e-01
-5.80245316e-01 1.06354940e+00 -9.13168013e-01 -2.87251800e-01
5.53585231e-01 6.60642922e-01 -1.13034189e-01 2.27989987e-01
-1.64403117e+00 1.23104203e+00 6.53071046e-01 4.71632451e-01
-4.20171052e-01 -3.37508142e-01 -5.94127595e-01 -1.11123197e-01
1.57313198e-02 -6.27754033e-01 1.07047546e+00 -7.62299418e-01
-1.37878954e+00 4.27304715e-01 2.21346647e-01 -3.09147179e-01
5.19507155e-02 4.97756116e-02 -1.09788835e+00 2.99389124e-01
-2.08955258e-01 3.55751991e-01 5.69209754e-01 -4.24881220e-01
-4.91916984e-01 -6.57388151e-01 -4.35522854e-01 3.82393785e-02
-3.81817013e-01 -1.24495402e-02 1.66593954e-01 -5.95705509e-01
1.84353441e-01 -4.94995028e-01 -1.43367857e-01 -2.41773114e-01
-2.23686144e-01 -3.59903723e-01 7.49969482e-01 -9.19723868e-01
1.34587741e+00 -2.19538665e+00 2.05481753e-01 2.04186648e-01
2.44536906e-01 6.75340652e-01 2.35722542e-01 3.35201435e-02
-2.15755984e-01 -9.28355679e-02 -4.22389507e-01 6.41094208e-01
-5.28890789e-01 5.16484261e-01 3.49208117e-01 4.69383985e-01
-1.77579969e-01 9.22278404e-01 -8.06008637e-01 -4.10153538e-01
4.14578974e-01 5.10811329e-01 3.10591191e-01 1.06983840e-01
5.10160804e-01 6.23659194e-01 -6.86267853e-01 6.09598935e-01
4.77903962e-01 -3.77462387e-01 -1.68258861e-01 -3.70202959e-01
1.99100837e-01 8.20482299e-02 -9.92716968e-01 1.62288511e+00
-1.72492936e-01 6.76131189e-01 3.76856066e-02 -1.67365813e+00
9.45253313e-01 9.01423573e-01 7.80263662e-01 -9.87574160e-01
7.04634309e-01 2.95849144e-01 3.34693015e-01 -1.45604122e+00
-3.94038022e-01 -3.04171562e-01 3.03517014e-01 3.74353826e-01
-5.65753095e-02 2.65410125e-01 -1.01659829e-02 -4.03383374e-01
8.82673562e-01 -2.12504238e-01 2.22093180e-01 -9.80366319e-02
3.58066618e-01 3.42972018e-03 7.22338974e-01 1.83612734e-01
-2.69274950e-01 2.58743197e-01 -3.75710502e-02 -6.28508687e-01
-5.52381277e-01 -8.02740872e-01 -4.32937622e-01 3.18256587e-01
2.24243626e-01 3.30899149e-01 -7.56965518e-01 -2.07642764e-01
-4.39866111e-02 4.81204271e-01 -4.80991185e-01 -7.31975555e-01
-5.19610703e-01 -1.21986449e+00 7.31585681e-01 8.57253790e-01
8.18879426e-01 -1.67418504e+00 -9.55404997e-01 5.29798150e-01
-1.29738823e-01 -7.25221574e-01 2.19955985e-02 2.70863384e-01
-1.24842036e+00 -1.19305491e+00 -1.07759893e+00 -1.06200182e+00
6.97637320e-01 -3.03214610e-01 7.26515114e-01 1.40921161e-01
-6.24101460e-01 8.80750567e-02 -1.95549071e-01 -5.26025593e-01
-1.32930085e-01 -3.00205529e-01 5.39040804e-01 -4.17693406e-01
7.00109184e-01 -9.99707401e-01 -5.99443972e-01 -7.73016363e-02
-7.51457691e-01 -1.12977393e-01 1.01217639e+00 9.22496557e-01
2.39950612e-01 7.31611669e-01 1.02757943e+00 -6.53983474e-01
1.20452142e+00 -3.99375498e-01 -5.45910373e-02 1.66254312e-01
-1.02380729e+00 -3.55536908e-01 7.27018356e-01 -5.42843223e-01
-8.89976978e-01 -5.93777061e-01 -3.59309316e-01 -3.15458506e-01
-3.70173275e-01 8.86893034e-01 1.42980427e-01 -5.08077368e-02
6.80128753e-01 4.65883046e-01 2.26803735e-01 -6.52274370e-01
-3.90868723e-01 1.14974749e+00 1.12454844e+00 -7.17709810e-02
-2.91889869e-02 -2.60038853e-01 -9.17840675e-02 -6.19740665e-01
-2.93471158e-01 8.96309614e-02 -4.24727648e-01 -2.61861771e-01
9.68627155e-01 -4.78567511e-01 -8.98009896e-01 6.70481443e-01
-1.31396234e+00 1.66415840e-01 1.34146184e-01 1.25202847e+00
-2.06784397e-01 2.87005603e-01 -1.18271244e+00 -6.84582293e-01
-8.33924353e-01 -8.52011561e-01 2.22113699e-01 5.48894048e-01
-4.31295097e-01 -1.05883133e+00 -2.45679051e-01 9.66123194e-02
5.91019392e-01 4.21663612e-01 1.10388601e+00 -3.61281455e-01
3.48724723e-01 -3.28468978e-01 -1.38205756e-02 5.90410709e-01
3.88916910e-01 -1.77133277e-01 -6.09657228e-01 -6.54324424e-03
5.31433463e-01 -2.30609328e-01 3.98318678e-01 9.18916523e-01
1.73560929e+00 -1.09847732e-01 -4.49721366e-01 4.43370134e-01
1.14099014e+00 1.32673550e+00 8.19850266e-01 -3.53111625e-02
4.15929258e-01 3.79342169e-01 2.66306847e-01 1.96582645e-01
-4.70256433e-02 -4.61902581e-02 1.35256320e-01 -2.54595011e-01
2.51560122e-01 3.28735054e-01 -7.25085065e-02 1.42250979e+00
-6.58107519e-01 -6.06781542e-02 -7.87838459e-01 4.97990102e-01
-1.63800263e+00 -8.86038125e-01 -3.00340265e-01 1.97361541e+00
9.84578013e-01 1.28993452e-01 -2.62789079e-03 7.80122697e-01
9.08512473e-01 -4.85944211e-01 -8.27704549e-01 -2.65772402e-01
1.78712443e-01 4.94790316e-01 2.20654368e-01 -1.65818080e-01
-7.86119044e-01 5.29342704e-02 7.20268488e+00 6.37579978e-01
-1.60782027e+00 1.90519616e-01 4.66731548e-01 5.54493368e-02
3.23244870e-01 -8.69984090e-01 1.24783802e-03 9.87908840e-01
1.09987807e+00 -2.58985251e-01 4.95656252e-01 6.02573693e-01
1.14278823e-01 -1.43931836e-01 -9.24956322e-01 1.70932615e+00
-1.03287302e-01 -1.15276289e+00 -4.42032099e-01 -1.89792693e-01
2.01382145e-01 -1.41792178e-01 -1.11847080e-01 2.10821748e-01
-4.19211388e-01 -1.27459776e+00 -8.28167126e-02 7.13390350e-01
8.96618783e-01 -8.32990170e-01 1.16958773e+00 5.22646368e-01
-6.20642066e-01 -1.52737781e-01 -3.91960263e-01 -4.12689030e-01
2.66360581e-01 8.72653067e-01 -1.61709979e-01 4.98458087e-01
8.73554766e-01 9.01190102e-01 7.22599104e-02 1.04813385e+00
-2.67484337e-01 5.50843179e-01 -6.01712838e-02 -2.64164269e-01
1.23884223e-01 -2.10556462e-01 2.60011792e-01 8.45271230e-01
4.41460907e-01 5.62738955e-01 -6.02453537e-02 9.06767070e-01
1.73542067e-01 -2.60860801e-01 -2.56728441e-01 -3.66721094e-01
4.01823014e-01 1.01459503e+00 -6.46400809e-01 -5.11472762e-01
-8.59722495e-02 9.82179463e-01 -4.08114672e-01 2.76250273e-01
-9.44130838e-01 -1.05174160e+00 -4.29536738e-02 -2.51410514e-01
-5.28771639e-01 1.88193887e-01 -5.22921145e-01 -8.69213521e-01
7.36871138e-02 -7.84692466e-01 2.53477275e-01 -1.06758022e+00
-1.26908779e+00 5.62977970e-01 -1.64231006e-02 -1.13740480e+00
-2.00679138e-01 -7.12681472e-01 -8.32167864e-01 8.88740182e-01
-1.02314019e+00 -1.04104906e-01 -3.89404923e-01 6.84519172e-01
3.30975205e-01 -1.57287911e-01 1.03777373e+00 8.35581183e-01
-7.16274440e-01 2.31667459e-01 8.36752132e-02 3.63578439e-01
3.90370011e-01 -1.03021550e+00 -2.58577410e-02 2.46070415e-01
-4.56996471e-01 5.35342455e-01 3.42277259e-01 -4.20591831e-01
-1.04663312e+00 -7.76628673e-01 7.99994171e-01 2.92982280e-01
1.83525980e-01 4.51482505e-01 -1.03696001e+00 2.65880674e-01
3.27944517e-01 -1.43137187e-01 8.21201384e-01 -2.46386975e-01
7.78798997e-01 -1.91080391e-01 -1.44368875e+00 2.56343126e-01
8.60281289e-01 -1.96232408e-01 -1.11675620e+00 3.39973986e-01
2.33058289e-01 -3.33628535e-01 -1.26355720e+00 7.13472724e-01
6.59742713e-01 -7.04313934e-01 7.45510519e-01 -5.55173993e-01
2.76817828e-01 -1.42369762e-01 5.71522295e-01 -1.65871942e+00
-5.88511884e-01 -1.67684838e-01 -2.12340713e-01 6.89305782e-01
7.38361180e-02 -7.29576707e-01 4.70233887e-01 5.76911092e-01
-2.83652753e-01 -1.11367893e+00 -7.39403605e-01 -6.80655599e-01
-3.55813265e-01 -2.40570158e-01 4.81763005e-01 1.19376898e+00
4.89595205e-01 5.74270546e-01 -5.83448172e-01 -2.16601044e-01
4.18012887e-01 -8.87105688e-02 -2.39332944e-01 -1.59613299e+00
-1.17172979e-01 -5.47573209e-01 -6.31577849e-01 -6.96202338e-01
-3.87108743e-01 -4.80697155e-01 -1.14593431e-02 -2.10141015e+00
-1.24366008e-01 -1.76496223e-01 -6.90543771e-01 6.81790590e-01
-5.87731898e-02 2.13629976e-01 -3.33210409e-01 2.09690005e-01
-2.05782861e-01 4.03826326e-01 1.56396735e+00 -1.75654307e-01
-2.04170749e-01 9.75775253e-03 -5.20718873e-01 9.32824016e-01
8.87877822e-01 -3.77255738e-01 -3.36092889e-01 -6.60519958e-01
-1.24783553e-01 4.72663969e-01 5.40499240e-02 -1.42731333e+00
4.25695360e-01 -1.43365443e-01 1.05405688e+00 -4.83014613e-01
1.76175505e-01 -8.98902774e-01 2.82752246e-01 1.21871650e+00
-1.67662352e-01 8.54123011e-02 -3.53115471e-03 2.40050420e-01
-4.43412423e-01 -2.94832557e-01 6.35908961e-01 -2.83145338e-01
-8.79316509e-01 5.02338827e-01 -7.71559656e-01 -1.72402874e-01
1.11100471e+00 -4.55381215e-01 -3.05976868e-02 -1.63605213e-01
-1.09332621e+00 2.08308011e-01 -3.01630586e-01 3.08773428e-01
9.69082296e-01 -1.59818292e+00 -3.18779916e-01 4.03800845e-01
-3.18459481e-01 3.10922265e-02 4.69281167e-01 1.25083339e+00
-6.63500249e-01 2.54024982e-01 -7.43278921e-01 -5.07611632e-01
-1.01195943e+00 4.34376687e-01 4.45867866e-01 2.76879638e-01
-8.56865883e-01 7.48931587e-01 -5.19602299e-01 -4.74269874e-02
5.24112940e-01 -3.81692529e-01 -5.95246732e-01 -1.59482807e-01
7.20867932e-01 6.94911301e-01 6.82027712e-02 -2.57987469e-01
-5.74687123e-01 6.62803769e-01 3.50283384e-01 3.19284469e-01
1.35300291e+00 2.65438676e-01 -3.17019731e-01 5.22280157e-01
1.07377553e+00 -8.79012525e-01 -4.18385625e-01 4.57934849e-02
-1.95411369e-01 -7.71136805e-02 4.58188206e-01 -1.22734487e+00
-1.42680860e+00 1.24306774e+00 9.00586903e-01 3.35623801e-01
1.47873044e+00 -5.12949169e-01 1.37669766e+00 5.99169374e-01
5.97076774e-01 -1.46934831e+00 -1.39347866e-01 3.63127179e-02
5.88551760e-01 -9.86830950e-01 -7.52205402e-02 1.55298084e-01
-5.85323572e-01 1.58922970e+00 4.41707373e-01 -1.93697631e-01
9.13039207e-01 3.05840343e-01 1.68654725e-01 -2.94304401e-01
-2.94403076e-01 3.58856082e-01 8.51795673e-02 7.12282121e-01
3.32069546e-01 4.68596965e-02 -8.12546790e-01 1.01051402e+00
-3.70218069e-03 8.60567510e-01 1.02576457e-01 1.22823942e+00
-6.61002517e-01 -7.63663650e-01 -5.15374124e-01 9.47006047e-01
-9.19132471e-01 4.43130061e-02 2.63880137e-02 4.48536009e-01
5.44101357e-01 9.68339503e-01 1.54897094e-01 -6.71429038e-01
2.55023152e-01 2.54378226e-02 5.18760979e-01 -1.77451968e-01
-3.16346616e-01 -1.17088340e-01 -1.32588461e-01 -2.19124377e-01
-7.89586723e-01 4.79891226e-02 -1.72490251e+00 -2.98999935e-01
-2.30078146e-01 4.36734587e-01 6.93873703e-01 1.14115870e+00
3.91327679e-01 1.16248393e+00 3.50297064e-01 -5.92717707e-01
-4.40495521e-01 -1.43723512e+00 -1.00284684e+00 3.45012248e-01
1.31046891e-01 -7.31455147e-01 -1.27163380e-01 8.45569074e-02] | [13.75610065460205, 3.3306148052215576] |
7a97e31d-a81b-4f37-b7da-db2d10bdd131 | temporally-coherent-video-harmonization-using | 1809.01372 | null | http://arxiv.org/abs/1809.01372v1 | http://arxiv.org/pdf/1809.01372v1.pdf | Temporally Coherent Video Harmonization Using Adversarial Networks | Compositing is one of the most important editing operations for images and
videos. The process of improving the realism of composite results is often
called harmonization. Previous approaches for harmonization mainly focus on
images. In this work, we take one step further to attack the problem of video
harmonization. Specifically, we train a convolutional neural network in an
adversarial way, exploiting a pixel-wise disharmony discriminator to achieve
more realistic harmonized results and introducing a temporal loss to increase
temporal consistency between consecutive harmonized frames. Thanks to the
pixel-wise disharmony discriminator, we are also able to relieve the need of
input foreground masks. Since existing video datasets which have ground-truth
foreground masks and optical flows are not sufficiently large, we propose a
simple yet efficient method to build up a synthetic dataset supporting
supervised training of the proposed adversarial network. Experiments show that
training on our synthetic dataset generalizes well to the real-world composite
dataset. Also, our method successfully incorporates temporal consistency during
training and achieves more harmonious results than previous methods. | ['Hao-Zhi Huang', 'Shi-Min Hu', 'Senzhe Xu', 'Junxiong Cai', 'Wei Liu'] | 2018-09-05 | null | null | null | null | ['video-harmonization'] | ['computer-vision'] | [ 2.88094848e-01 -1.28298223e-01 1.73399478e-01 -1.65424585e-01
-4.56827015e-01 -5.56599140e-01 6.13974571e-01 -2.49220148e-01
-4.30614680e-01 7.17384994e-01 -5.92121482e-03 -1.07969731e-01
2.00102687e-01 -8.86044025e-01 -1.09550071e+00 -5.64838052e-01
6.92372993e-02 -5.77056482e-02 3.83351862e-01 -3.97587091e-01
4.67770770e-02 4.81279194e-01 -1.31912959e+00 8.60397741e-02
1.04013407e+00 7.68683195e-01 -8.63789320e-02 7.28801906e-01
2.41969079e-01 8.25430274e-01 -7.22244620e-01 -6.07996821e-01
7.04307735e-01 -7.22804546e-01 -8.31243336e-01 2.84110129e-01
5.59012771e-01 -3.47566187e-01 -4.15618122e-01 1.12539911e+00
4.87603605e-01 3.28543365e-01 2.74465412e-01 -1.46641076e+00
-7.09987342e-01 5.97106636e-01 -6.47772729e-01 -2.82902569e-02
1.26305476e-01 3.93376231e-01 7.92634130e-01 -4.99737233e-01
7.49482691e-01 1.19751561e+00 6.85574114e-01 4.24909562e-01
-1.23145449e+00 -7.64012516e-01 4.07819264e-02 2.29884595e-01
-1.23203850e+00 -3.22372973e-01 1.10143375e+00 -3.01891983e-01
2.70801753e-01 3.50672722e-01 8.14500809e-01 1.05951571e+00
-9.12134945e-02 6.94845140e-01 1.05800986e+00 -3.12815219e-01
8.32205042e-02 1.01990090e-03 -5.77585399e-01 4.63302612e-01
2.52385549e-02 1.68816671e-02 -2.12417573e-01 2.31440261e-01
9.64572072e-01 -8.27853978e-02 -6.02177441e-01 -3.91623616e-01
-1.35995531e+00 5.86212814e-01 6.71123385e-01 3.43212008e-01
-2.83273190e-01 3.25911611e-01 4.26930815e-01 3.09896767e-01
3.92046571e-01 8.27156723e-01 -3.02028097e-03 1.03332326e-01
-1.21208823e+00 2.63032556e-01 5.23188055e-01 6.78681195e-01
7.17343688e-01 3.53856623e-01 -3.37258250e-01 6.84334278e-01
-2.33030006e-01 2.91002870e-01 2.86023855e-01 -1.38387096e+00
3.84609848e-01 2.71073997e-01 3.00992161e-01 -1.30881822e+00
-1.23106368e-01 -5.02365768e-01 -9.70665574e-01 4.99464452e-01
4.10756707e-01 -2.18255654e-01 -6.38721168e-01 2.05172276e+00
3.29196244e-01 6.19141638e-01 7.91099072e-02 1.03019369e+00
3.53826761e-01 7.79170632e-01 -9.98382494e-02 -2.02972859e-01
8.84131789e-01 -1.27285194e+00 -7.11268306e-01 2.03224108e-01
1.22015327e-01 -1.02749157e+00 1.06849861e+00 3.07979524e-01
-1.40453172e+00 -7.74353743e-01 -1.10061264e+00 -1.15361810e-01
-1.45476624e-01 -9.52980071e-02 4.25215185e-01 3.86230588e-01
-1.04677308e+00 7.85286367e-01 -7.13976383e-01 -7.76502565e-02
3.49250406e-01 1.88015714e-01 -4.62139100e-01 2.02746004e-01
-1.25303197e+00 8.31736267e-01 5.13231277e-01 2.33377233e-01
-8.89468253e-01 -8.58985424e-01 -8.93441558e-01 -8.97537544e-02
4.14145827e-01 -7.28927493e-01 1.02302706e+00 -1.50297320e+00
-1.71644759e+00 7.23345280e-01 3.39288980e-01 -5.91555357e-01
1.00233305e+00 -1.77101523e-01 -2.59047508e-01 1.70864761e-01
-3.40874344e-02 8.38318884e-01 1.03277206e+00 -1.56702924e+00
-4.61854964e-01 1.77410394e-01 3.96164149e-01 3.02493386e-02
-5.12068331e-01 -1.92411676e-01 -7.03073084e-01 -1.13927543e+00
-5.17515063e-01 -9.70432937e-01 -2.43282035e-01 1.38687119e-01
-5.37100196e-01 3.28022629e-01 9.70154464e-01 -8.64476740e-01
1.06907570e+00 -2.01471424e+00 3.18007201e-01 2.84779817e-02
2.02319860e-01 6.54398501e-01 -4.07479435e-01 1.66812211e-01
-1.68376833e-01 4.50514853e-02 -4.42258388e-01 -4.90381956e-01
-1.76568165e-01 2.28147700e-01 -2.82067031e-01 4.15957242e-01
3.27405840e-01 9.36013937e-01 -9.66300189e-01 -4.97913480e-01
3.84204119e-01 7.17710376e-01 -7.80272245e-01 2.77893573e-01
-2.96508342e-01 9.13360178e-01 -1.28717571e-01 2.18986511e-01
8.48425448e-01 2.71293838e-02 9.87901837e-02 -4.39514488e-01
-1.28933653e-01 -2.03706041e-01 -1.21794534e+00 1.72142541e+00
-6.61633730e-01 7.46571243e-01 2.37751994e-02 -1.09281027e+00
7.95797050e-01 1.03677101e-01 6.60067081e-01 -7.14173198e-01
3.18247050e-01 1.18999302e-01 -8.45945925e-02 -2.94180393e-01
6.99523807e-01 -6.04177676e-02 4.78291474e-02 3.49158406e-01
-1.99830055e-01 -3.14315051e-01 2.00293019e-01 1.76939275e-02
7.91582644e-01 2.50149161e-01 5.41574955e-02 3.52904238e-02
6.59512639e-01 -2.12220132e-01 7.23134458e-01 5.15420675e-01
-1.55447736e-01 8.66938353e-01 4.91694719e-01 -4.74074125e-01
-1.41591680e+00 -9.42425966e-01 2.43048891e-01 5.57898879e-01
4.56021041e-01 -1.63697734e-01 -9.22279596e-01 -4.66079563e-01
-2.47312978e-01 4.38509345e-01 -6.74598277e-01 -2.90019363e-01
-9.43497539e-01 -3.65767717e-01 7.08787858e-01 4.97585446e-01
1.11384058e+00 -9.06166732e-01 -5.73813498e-01 3.02651286e-01
-3.93376410e-01 -1.42493796e+00 -7.49502540e-01 -2.77476341e-01
-5.49828053e-01 -1.01274359e+00 -9.29204404e-01 -7.02194512e-01
6.16401613e-01 2.39904270e-01 1.06013548e+00 1.86538607e-01
-2.67876625e-01 -1.73768215e-02 -2.52802968e-01 -2.03352705e-01
-5.51719487e-01 -1.53796732e-01 -2.23598644e-01 3.50756586e-01
-4.50001091e-01 -8.34411919e-01 -8.41078520e-01 3.85962188e-01
-1.31289959e+00 4.43250835e-01 3.75350475e-01 7.64446557e-01
5.03374577e-01 2.70697445e-01 2.16699764e-01 -8.44725847e-01
4.20797259e-01 2.62287748e-03 -7.36318111e-01 2.68061250e-01
-3.09449792e-01 8.04903731e-02 1.06439722e+00 -6.80868864e-01
-1.06151581e+00 2.36507598e-02 -1.13141380e-01 -7.81379461e-01
1.69014364e-01 4.39100228e-02 -2.03340501e-01 -3.20166647e-01
5.20197093e-01 -8.83076899e-03 -1.83454230e-02 -9.42660943e-02
5.84193885e-01 1.05050370e-01 1.08163297e+00 -5.55926681e-01
1.13779402e+00 7.42949724e-01 8.95349756e-02 -3.22349012e-01
-5.81186593e-01 -2.78406721e-02 -5.24732172e-01 -3.29217643e-01
9.71726954e-01 -7.49677718e-01 -7.32058227e-01 6.76797569e-01
-1.18976414e+00 -5.86248219e-01 -3.68041098e-01 3.95036429e-01
-6.74950778e-01 5.33524215e-01 -5.54881513e-01 -3.64365131e-01
-3.53073180e-01 -1.18449354e+00 8.91081452e-01 2.70300061e-01
1.31019140e-02 -9.53991115e-01 1.44049123e-01 2.75357336e-01
5.51372468e-01 8.89285922e-01 5.30346632e-01 -1.05980381e-01
-5.96363723e-01 4.22740309e-03 -1.65498152e-01 5.66116631e-01
2.37076491e-01 4.33254629e-01 -5.80657184e-01 -1.89360067e-01
-6.94801211e-02 -1.32790536e-01 9.28033412e-01 2.04303324e-01
1.24559629e+00 -3.68878603e-01 1.34928018e-01 8.64102364e-01
1.37248063e+00 1.93365246e-01 9.31333363e-01 5.26074708e-01
8.43523502e-01 4.23813015e-01 6.55417740e-01 4.98227745e-01
3.16034526e-01 9.55058634e-01 5.48602045e-01 -6.24905705e-01
-4.44127679e-01 -1.94550335e-01 2.77885795e-01 5.65058589e-01
-3.04263622e-01 -3.82643372e-01 -5.80073833e-01 5.85591137e-01
-1.89155257e+00 -1.16556048e+00 1.28889605e-01 2.12312770e+00
1.05221140e+00 -2.68798467e-04 1.86555207e-01 1.99990362e-01
9.57815826e-01 3.43932003e-01 -3.21011066e-01 -2.38466114e-01
-2.19771251e-01 3.02938789e-01 4.11426157e-01 5.82799971e-01
-1.23367357e+00 1.06939924e+00 5.44130707e+00 6.59972668e-01
-1.52613604e+00 1.21738039e-01 6.75807774e-01 -7.74401650e-02
-3.72331291e-01 -1.45834833e-02 -9.67561230e-02 5.31269848e-01
3.75016689e-01 -9.89533961e-02 6.29184783e-01 5.63494503e-01
4.96786624e-01 4.80175614e-02 -9.15807605e-01 9.08799708e-01
6.28616288e-02 -1.53427267e+00 1.25398338e-01 -3.41249883e-01
1.19670010e+00 -5.53239584e-01 3.17426503e-01 -5.93289454e-03
3.87134880e-01 -9.57587540e-01 9.38250601e-01 5.32765210e-01
7.44077444e-01 -8.87331128e-01 6.35501325e-01 -8.40998627e-03
-1.11390841e+00 1.83087483e-01 -1.74644932e-01 1.18360437e-01
4.27242726e-01 4.45343047e-01 -4.65532482e-01 6.49521887e-01
5.91569483e-01 7.40153849e-01 -5.10474563e-01 1.01308203e+00
-3.96810293e-01 2.78363436e-01 -1.36228681e-01 5.56957126e-01
2.06713215e-01 -2.51019746e-01 4.28964555e-01 1.07629192e+00
3.14921141e-01 -1.21420793e-01 1.86776463e-02 8.85182977e-01
-3.83118898e-01 -3.65660116e-02 -5.61136365e-01 1.06232120e-02
2.07676917e-01 1.29903328e+00 -7.02509224e-01 -2.64262766e-01
-3.19300026e-01 1.36059725e+00 1.28050104e-01 3.65055650e-01
-1.34747660e+00 -3.05169761e-01 6.41580939e-01 -9.45275873e-02
4.24951047e-01 -2.90933013e-01 -1.74400479e-01 -1.26212823e+00
2.31140498e-02 -1.16062713e+00 1.31074503e-01 -8.42150986e-01
-1.00029790e+00 7.40298569e-01 -2.34255537e-01 -1.35951543e+00
-1.07132137e-01 -2.18391985e-01 -9.34818029e-01 6.12288773e-01
-1.69625437e+00 -1.34961879e+00 -7.32574224e-01 9.92631495e-01
4.03619617e-01 7.55246803e-02 4.34494436e-01 5.17143428e-01
-5.77405691e-01 6.68357432e-01 -1.13625020e-01 2.64869452e-01
1.05663443e+00 -1.03989327e+00 3.17746729e-01 1.34395015e+00
9.60067362e-02 4.88652408e-01 9.85502064e-01 -4.02810901e-01
-9.97364759e-01 -1.30503166e+00 3.50318700e-01 7.36514851e-02
5.35641551e-01 -1.27163172e-01 -8.40126157e-01 5.98073304e-01
5.35694599e-01 1.40803784e-01 2.34105453e-01 -6.15878344e-01
-3.80180776e-01 -3.74095798e-01 -1.07270062e+00 8.25515151e-01
9.97420192e-01 -4.34172720e-01 -3.39810461e-01 1.75321698e-01
8.22312772e-01 -5.73463917e-01 -8.55062306e-01 6.08174026e-01
4.07826751e-01 -1.15530062e+00 1.07742965e+00 -4.12789941e-01
7.75308371e-01 -6.12530112e-01 1.25755683e-01 -1.35689545e+00
8.86914413e-03 -9.43034112e-01 2.20795214e-01 1.30358028e+00
8.81343037e-02 -3.68614525e-01 5.60696125e-01 4.49381381e-01
-9.29309875e-02 -4.17715877e-01 -5.82053244e-01 -7.91265488e-01
7.99013451e-02 -3.00829262e-01 6.92785859e-01 1.15600753e+00
-5.44177353e-01 -1.38838321e-01 -9.08525586e-01 2.40942329e-01
6.21834219e-01 2.98831373e-01 1.19202435e+00 -7.06986666e-01
-4.94635791e-01 -4.52417314e-01 -3.83537829e-01 -7.69769311e-01
2.37547413e-01 -5.13186157e-01 4.71213125e-02 -1.17614484e+00
1.66888207e-01 -3.45669985e-01 -9.08411071e-02 4.57643747e-01
-3.18241894e-01 7.21760452e-01 4.61380213e-01 6.59973100e-02
-5.54968238e-01 6.08082831e-01 1.68270850e+00 -2.39775360e-01
-2.24474266e-01 -2.13706642e-01 -4.62279916e-01 5.50874174e-01
8.67043495e-01 -2.07448035e-01 -5.51386178e-01 -5.83508909e-01
7.50950677e-03 -1.59913162e-03 5.40733814e-01 -1.05027544e+00
1.83863625e-01 -2.79461354e-01 2.62868285e-01 -1.68977693e-01
2.26822600e-01 -7.53849924e-01 6.25564933e-01 4.69585210e-01
-3.51354897e-01 1.07612565e-01 1.96337491e-01 3.85877162e-01
-5.27015150e-01 3.74218412e-02 1.28513241e+00 4.95330477e-03
-6.78199232e-01 4.67498720e-01 1.07800186e-01 1.43653944e-01
1.32344162e+00 -5.92609942e-02 -2.94742346e-01 -6.17594838e-01
-5.44811130e-01 9.15814862e-02 8.71444941e-01 5.04110992e-01
4.37104285e-01 -1.41831136e+00 -5.91900110e-01 1.17727071e-01
-2.46299371e-01 -1.56082138e-01 2.59832144e-01 8.31677377e-01
-9.54962015e-01 -1.04920469e-01 -7.19452262e-01 -3.09975415e-01
-1.19697213e+00 7.90137291e-01 5.13588369e-01 -2.66691953e-01
-5.97737432e-01 6.53621793e-01 2.04988286e-01 -1.13466091e-01
1.48458824e-01 -2.29170829e-01 1.06548639e-02 -1.75768822e-01
3.77280235e-01 2.18383789e-01 -1.73817649e-01 -6.63274705e-01
-1.09466024e-01 7.12708652e-01 1.68731362e-01 -1.58446521e-01
1.37691462e+00 -1.58673808e-01 -1.91393003e-01 -6.49299324e-02
1.30124259e+00 1.77951559e-01 -1.55307448e+00 -1.10068209e-01
-4.26888555e-01 -6.24397755e-01 -1.73097461e-01 -3.14927906e-01
-1.60383689e+00 7.09646583e-01 3.89538974e-01 1.44286066e-01
1.44071710e+00 -4.12525088e-01 1.13035989e+00 -1.49269486e-02
1.07235670e-01 -9.75549638e-01 5.44609308e-01 1.65956348e-01
9.33107316e-01 -1.17858648e+00 -1.99961826e-01 -4.41922545e-01
-6.87201142e-01 1.11868763e+00 6.30276203e-01 -3.21178108e-01
2.75878847e-01 2.97708720e-01 1.24724649e-01 2.07893297e-01
-3.10928136e-01 -1.49230093e-01 1.72578767e-01 4.81879681e-01
1.68119863e-01 -2.22050622e-01 -4.87145364e-01 1.73308820e-01
-2.48220652e-01 -6.11591972e-02 5.62387168e-01 5.34723938e-01
1.12240082e-02 -1.29705870e+00 -4.40137714e-01 -3.50761414e-01
-5.80908358e-01 1.10404469e-01 -1.87661141e-01 9.22724247e-01
3.04482907e-01 8.43146026e-01 1.27086854e-02 -3.54834616e-01
2.82571018e-01 -4.56904858e-01 5.15313089e-01 -5.13955057e-02
-6.14178777e-01 -1.28838513e-02 -1.59209743e-01 -7.76127875e-01
-8.17359626e-01 -1.69614032e-01 -1.22557187e+00 -6.61243975e-01
4.11413349e-02 -8.34949091e-02 3.96583974e-01 8.22653472e-01
1.40144765e-01 7.45714486e-01 8.96657050e-01 -1.06201005e+00
-1.81107029e-01 -4.68922645e-01 -3.17231655e-01 9.45609927e-01
3.48515868e-01 -4.91470039e-01 -1.89843118e-01 4.61648524e-01] | [11.256453514099121, -1.0648107528686523] |
a09b9a31-9af5-4a35-a58f-7d34d61a4fec | r-3-reverse-retrieve-and-rank-for-sarcasm | 2004.13248 | null | https://arxiv.org/abs/2004.13248v4 | https://arxiv.org/pdf/2004.13248v4.pdf | $R^3$: Reverse, Retrieve, and Rank for Sarcasm Generation with Commonsense Knowledge | We propose an unsupervised approach for sarcasm generation based on a non-sarcastic input sentence. Our method employs a retrieve-and-edit framework to instantiate two major characteristics of sarcasm: reversal of valence and semantic incongruity with the context which could include shared commonsense or world knowledge between the speaker and the listener. While prior works on sarcasm generation predominantly focus on context incongruity, we show that combining valence reversal and semantic incongruity based on the commonsense knowledge generates sarcasm of higher quality. Human evaluation shows that our system generates sarcasm better than human annotators 34% of the time, and better than a reinforced hybrid baseline 90% of the time. | ['Nanyun Peng', 'Debanjan Ghosh', 'Smaranda Muresan', 'Tuhin Chakrabarty'] | 2020-04-28 | null | null | null | null | ['scene-text-detection'] | ['computer-vision'] | [ 1.99807674e-01 7.95526147e-01 1.16468869e-01 -5.75130463e-01
-5.73338509e-01 -5.60210645e-01 7.47272015e-01 2.56768644e-01
-1.94630057e-01 7.34637260e-01 1.01847160e+00 2.11694971e-01
5.33894539e-01 -5.74366927e-01 -5.18848449e-02 -3.21280599e-01
9.25161421e-01 6.32457554e-01 -2.29943693e-02 -9.60761905e-01
4.15328085e-01 -3.48773777e-01 -1.11361623e+00 7.43306339e-01
6.06167436e-01 2.96093285e-01 -1.58405989e-01 5.98025918e-01
-1.52144670e-01 1.41340923e+00 -1.15441513e+00 -8.68323863e-01
-1.54622709e-02 -1.24378705e+00 -1.22034979e+00 -1.60495773e-01
2.47266248e-01 1.87454522e-01 3.90485197e-01 9.91673768e-01
7.54476011e-01 3.19348663e-01 5.84393144e-01 -9.51854110e-01
-7.45705068e-01 1.36807358e+00 -1.15318440e-01 -1.29908741e-01
5.97191513e-01 3.35568795e-03 1.14126515e+00 -7.68821239e-01
9.02743995e-01 1.09772861e+00 9.04850721e-01 1.12824106e+00
-1.23283184e+00 -3.64419609e-01 -4.84070152e-01 1.72890931e-01
-8.05701554e-01 -4.52674747e-01 1.18409061e+00 -4.17388648e-01
1.06422973e+00 4.34470505e-01 9.40293908e-01 1.31889224e+00
-1.48209780e-01 4.93908912e-01 1.21668983e+00 -7.79826641e-01
3.98897707e-01 3.68380219e-01 2.64705479e-01 6.22706339e-02
-1.24692969e-01 -2.18389988e-01 -7.97917008e-01 -3.23231310e-01
2.51970649e-01 -6.53250754e-01 2.14875909e-03 3.17855589e-02
-1.08233488e+00 1.23013687e+00 1.12328708e-01 4.25511181e-01
-1.85913697e-01 8.01561326e-02 9.02712166e-01 1.97170839e-01
3.76781851e-01 1.14685082e+00 -1.23450212e-01 -5.49935520e-01
-1.02984536e+00 3.81530702e-01 1.03059769e+00 5.57783127e-01
2.16115370e-01 1.83524579e-01 -4.33729261e-01 1.24591756e+00
1.04132198e-01 5.54124475e-01 7.64330685e-01 -1.17693472e+00
8.26366395e-02 5.00437081e-01 2.65072882e-01 -1.11440551e+00
-3.64589423e-01 -1.40858442e-01 -4.56813991e-01 1.02889806e-01
1.42655373e-01 2.81683961e-03 -3.51424575e-01 2.18636417e+00
2.36210838e-01 -3.54096949e-01 4.44217741e-01 1.12474358e+00
1.29927480e+00 3.98009807e-01 2.31326997e-01 -6.53071880e-01
1.26341641e+00 -1.17818201e+00 -9.65685666e-01 -3.50071400e-01
8.12711537e-01 -9.71334159e-01 1.37396049e+00 3.37163955e-01
-1.50218892e+00 -5.76576531e-01 -1.16387546e+00 -2.59519011e-01
1.58916205e-01 6.27956837e-02 4.12627310e-01 5.38163006e-01
-7.88647056e-01 5.38571417e-01 -2.47344319e-02 -3.81424695e-01
-2.18191072e-01 -2.43382916e-01 2.00159196e-02 6.00005805e-01
-1.50752234e+00 1.26351213e+00 2.84683526e-01 -2.85748094e-01
-4.57797915e-01 -6.21772826e-01 -9.40675795e-01 -3.41928512e-01
3.50692905e-02 -1.04579115e+00 1.58694148e+00 -1.72117066e+00
-1.92337608e+00 1.37151015e+00 4.10216264e-02 -6.03487611e-01
3.15212071e-01 -3.73344600e-01 -2.28332147e-01 1.64583787e-01
2.94571042e-01 6.82570934e-01 5.06845236e-01 -1.00458229e+00
2.48073801e-01 7.54228607e-02 9.50151011e-02 6.22128189e-01
-1.84386313e-01 3.54941517e-01 4.99506414e-01 -9.50856745e-01
-1.85060695e-01 -1.01405048e+00 -8.14132169e-02 -4.87281591e-01
-5.78288198e-01 -4.06580031e-01 4.74838644e-01 -5.64905047e-01
1.28898191e+00 -2.01588702e+00 3.07180464e-01 -2.50643343e-01
8.15501660e-02 3.74045111e-02 -2.57508643e-02 8.59207451e-01
-4.67121691e-01 -9.55964923e-02 -2.50951737e-01 -3.82837862e-01
-4.61916119e-04 1.39721977e-02 -6.78110301e-01 -1.91104606e-01
-2.67750233e-01 9.89276171e-01 -1.11296928e+00 -6.97113276e-01
3.70924012e-04 3.11625659e-01 -6.58198833e-01 2.96894133e-01
-2.89899230e-01 3.26899856e-01 1.70299336e-01 -2.77809147e-02
1.91617906e-01 -2.49884069e-01 5.09461403e-01 -2.59843737e-01
2.65505761e-01 9.42027450e-01 -5.57936430e-01 1.92867470e+00
-4.61379558e-01 2.69791961e-01 -4.16423231e-01 -6.90116465e-01
1.22282159e+00 5.24599493e-01 2.63480982e-03 -3.90876591e-01
1.51480943e-01 1.21989183e-01 2.43066531e-03 -3.40099096e-01
7.12988138e-01 -1.05034912e+00 -6.32166743e-01 1.00245082e+00
-1.36014253e-01 -1.05753553e+00 2.12526381e-01 9.70170379e-01
8.71859848e-01 2.09118053e-01 7.52169967e-01 -3.37986559e-01
5.25399506e-01 3.79484475e-01 4.31011707e-01 4.90158260e-01
-2.29261279e-01 6.01131618e-01 4.82434273e-01 -3.46813858e-01
-1.23208177e+00 -1.15325856e+00 3.85713190e-01 1.08547997e+00
7.72076622e-02 -8.57311428e-01 -9.52807009e-01 -5.03427386e-01
-7.11736798e-01 1.47749627e+00 -7.47460067e-01 -3.63721490e-01
-5.14236093e-01 -4.28034395e-01 7.01014876e-01 4.68353003e-01
3.26470166e-01 -1.68355715e+00 -1.10582507e+00 2.95367122e-01
-7.03178465e-01 -9.23026145e-01 -4.89980847e-01 -1.39440402e-01
-6.99147344e-01 -8.77217293e-01 -1.07671926e-02 -7.25909472e-01
3.57494593e-01 -6.00651875e-02 1.58029926e+00 1.90421402e-01
-3.43679607e-01 2.31978536e-01 -5.84665358e-01 -3.46836209e-01
-1.15515208e+00 -5.68719625e-01 -2.78037578e-01 -5.91198087e-01
4.88602966e-01 -7.06829548e-01 -3.36256266e-01 -1.16130568e-01
-6.64612174e-01 8.90599012e-01 -1.96502909e-01 9.91468668e-01
2.16431484e-01 -5.11375487e-01 9.21978712e-01 -1.05472040e+00
1.01943815e+00 -3.17979008e-01 4.39447492e-01 -6.70092702e-02
-5.76638937e-01 -1.61279574e-01 6.77922010e-01 -3.86276841e-01
-1.23313487e+00 6.64428016e-03 -7.77992830e-02 6.90715164e-02
5.64151444e-02 1.75866276e-01 1.44225210e-01 8.90229046e-01
1.26376879e+00 -7.43980110e-02 7.25042671e-02 -6.18245378e-02
8.94574881e-01 5.44283748e-01 7.71384299e-01 -5.98611653e-01
2.77878404e-01 5.28321505e-01 -4.02713716e-01 -3.39576900e-01
-1.41673970e+00 -2.52478808e-01 -2.93397993e-01 -3.96887183e-01
9.72896874e-01 -8.49386990e-01 -2.17735797e-01 2.39635408e-01
-1.52155733e+00 -3.07056189e-01 -1.10931480e+00 1.49751410e-01
-9.13218796e-01 5.42478561e-01 -8.79569650e-01 -5.90180993e-01
-8.68433475e-01 -5.87566733e-01 7.17073023e-01 2.15831384e-01
-1.44327319e+00 -9.03588653e-01 6.29165232e-01 8.50919008e-01
3.86786520e-01 2.84763992e-01 8.75427067e-01 -8.76774311e-01
5.71729362e-01 7.11263865e-02 1.34079278e-01 4.56287146e-01
4.02752645e-02 -1.81544542e-01 -6.30293906e-01 2.66020060e-01
3.32476825e-01 -1.04706049e+00 5.07409036e-01 -1.95790157e-02
2.74951339e-01 -5.43488860e-01 2.67844141e-01 -2.47781664e-01
1.04010677e+00 -1.79745883e-01 7.70691097e-01 1.38035595e-01
3.04872811e-01 9.03887987e-01 5.25717616e-01 7.51202106e-01
5.44618666e-01 8.22168767e-01 2.41645455e-01 9.74728838e-02
-5.44147551e-01 -5.60595632e-01 6.72426879e-01 1.12887180e+00
2.03534320e-01 1.67100713e-01 -4.98657554e-01 7.95766234e-01
-1.84921598e+00 -1.57873356e+00 -6.21257961e-01 1.81793368e+00
1.68035352e+00 -4.83641624e-02 2.58561671e-01 3.24719429e-01
6.79329693e-01 2.44066536e-01 -1.74124744e-02 -1.24444199e+00
-5.68995416e-01 4.50038344e-01 -5.08874118e-01 7.62126386e-01
-5.56413949e-01 1.26764047e+00 6.75558376e+00 6.05834007e-01
-6.94424808e-01 3.88878226e-01 3.26466650e-01 -4.39339459e-01
-7.76294887e-01 3.04083019e-01 -2.05430388e-01 2.68805176e-01
7.01371908e-01 -2.38271460e-01 2.06333682e-01 8.85553241e-01
3.90278012e-01 -1.42879575e-01 -9.44007337e-01 9.15206790e-01
8.42473984e-01 -1.18672955e+00 -9.55383331e-02 -7.10623384e-01
9.75819886e-01 -4.69979554e-01 -3.98193359e-01 2.63505578e-01
7.15808034e-01 -1.00967073e+00 8.97136509e-01 4.21154886e-01
3.58575583e-01 -5.40077448e-01 6.95316315e-01 3.12194258e-01
-6.06020391e-01 2.23857015e-01 -5.62266149e-02 -5.20364046e-01
7.16896832e-01 8.00571799e-01 -8.81563723e-01 3.23741511e-02
4.91234034e-01 5.82436919e-01 -5.41532516e-01 2.30795994e-01
-1.13212502e+00 7.93447793e-01 -5.47271036e-02 -2.65677124e-01
-2.08069056e-01 -1.06612304e-02 7.75714040e-01 1.40577412e+00
-1.27791271e-01 2.41623893e-01 9.98523831e-02 1.08216643e+00
1.35340974e-01 7.07219064e-01 -4.05702084e-01 3.89245935e-02
3.59437853e-01 1.35477579e+00 -4.47586656e-01 -8.78075838e-01
1.80335701e-01 1.25859654e+00 3.78527373e-01 -9.34948847e-02
-9.14373398e-01 -1.24543674e-01 -1.89768188e-02 -1.22528836e-01
-5.26723377e-02 2.66280770e-01 -7.73694217e-01 -9.20636237e-01
-2.23048806e-01 -1.04151976e+00 4.92262691e-01 -1.50228894e+00
-1.46658325e+00 7.22568870e-01 -8.37658048e-02 -7.76890159e-01
-5.58085382e-01 1.15856226e-03 -1.00732398e+00 5.76764882e-01
-5.73071361e-01 -1.32322192e+00 -2.41531119e-01 3.15661669e-01
7.73574948e-01 6.38759136e-02 1.23756301e+00 -5.14796734e-01
1.53996378e-01 2.12203860e-01 -9.19757545e-01 -1.02504194e-01
1.04345369e+00 -1.35559499e+00 5.51566370e-02 5.66667795e-01
2.58389324e-01 6.91123307e-01 1.46055603e+00 -8.29312861e-01
-3.04280281e-01 -5.37743151e-01 1.66032910e+00 -6.16044462e-01
6.52563035e-01 -9.94463116e-02 -7.06848621e-01 2.27847949e-01
9.90975201e-01 -5.93625486e-01 1.15907454e+00 2.42166266e-01
-9.82529104e-01 5.10297954e-01 -1.04539132e+00 8.38069558e-01
9.96996462e-01 -6.56023979e-01 -1.58272564e+00 5.17584622e-01
8.14025760e-01 -1.69655457e-01 -4.77730960e-01 3.99752408e-01
3.32462490e-01 -1.17765450e+00 5.39704502e-01 -7.87884772e-01
1.18702102e+00 -2.98328310e-01 -1.77188456e-01 -1.49730444e+00
-1.03223011e-01 -6.81210279e-01 1.64274603e-01 1.18204224e+00
4.14509058e-01 9.06534716e-02 4.15204197e-01 5.21875858e-01
-2.73588836e-01 -3.95513475e-01 -7.19331384e-01 -4.52215284e-01
2.31889561e-01 -3.52991968e-01 1.55700266e-01 1.09744799e+00
1.02371705e+00 1.29116178e+00 -6.57833040e-01 -7.98957825e-01
2.18884289e-01 5.13698876e-01 8.34709704e-01 -8.55399966e-01
-8.16569984e-01 -4.88854796e-01 1.55680850e-02 -6.63505137e-01
3.40590745e-01 -1.14094996e+00 2.95200169e-01 -1.53525496e+00
7.35004902e-01 -1.31948367e-02 -5.15388511e-03 7.00157583e-01
-1.38664886e-01 6.17949307e-01 3.23848873e-01 2.52132982e-01
-8.18157136e-01 5.78829765e-01 8.84480178e-01 -9.35741421e-03
-4.19348031e-01 -5.69956362e-01 -1.00730121e+00 9.96201873e-01
9.23969448e-01 -5.11007249e-01 -5.34450293e-01 -2.11121086e-02
9.99716043e-01 5.93855083e-02 5.02326608e-01 -6.22148931e-01
-7.71867186e-02 -2.94495255e-01 -3.34161937e-01 -3.16080332e-01
5.11798918e-01 6.65586218e-02 4.04442549e-01 3.39555174e-01
-9.29550409e-01 5.07968962e-02 -3.81243117e-02 5.34826629e-02
-1.92299247e-01 -5.66390455e-01 9.54536021e-01 -2.99016774e-01
-2.22218037e-01 -1.02291262e+00 -2.83194005e-01 6.43882930e-01
7.93317795e-01 -1.29366294e-01 -5.26814640e-01 -8.18442285e-01
-8.59945595e-01 -4.39094722e-01 6.57939076e-01 5.18423378e-01
5.30686617e-01 -1.47152591e+00 -1.06683421e+00 -2.90786028e-01
2.97372907e-01 -6.54940963e-01 2.52761126e-01 8.91511679e-01
-2.90072888e-01 -9.13668945e-02 -1.47408143e-01 -1.57992736e-01
-1.48476052e+00 4.35769856e-01 2.63501942e-01 -4.07690644e-01
-6.23024523e-01 1.02489817e+00 7.10480288e-02 -3.43850881e-01
-2.86874741e-01 3.55085015e-01 -4.03090268e-01 7.67394230e-02
5.70300639e-01 -7.54470425e-03 -3.14423203e-01 -9.02669132e-01
-2.14001372e-01 2.98514098e-01 1.89148530e-01 -6.84675992e-01
9.14927363e-01 -1.51448324e-01 -4.16984051e-01 9.21597362e-01
4.72960323e-01 5.79279900e-01 -4.98046756e-01 1.03413925e-01
-1.69815198e-01 -7.72940889e-02 -4.67002422e-01 -1.40619135e+00
-3.89479935e-01 6.09121323e-01 -1.23354107e-01 2.14405060e-01
8.69785666e-01 2.30663016e-01 9.10940647e-01 1.03517257e-01
1.88973740e-01 -1.68435490e+00 8.26726496e-01 5.52472532e-01
1.33942008e+00 -8.84789407e-01 1.00266352e-01 -4.62180346e-01
-1.58030176e+00 1.00591052e+00 7.22183108e-01 -2.30127916e-01
2.52556890e-01 1.05169602e-01 5.73616147e-01 -3.30369979e-01
-1.10395265e+00 2.92036116e-01 1.00419959e-02 4.65330034e-01
9.24012780e-01 2.44008213e-01 -1.25873911e+00 1.22885752e+00
-8.90055716e-01 -2.20912442e-01 9.34823275e-01 5.27858853e-01
-4.13013130e-01 -1.15849388e+00 -6.03587814e-02 -1.19714394e-01
-4.36794192e-01 -5.13036430e-01 -1.17462933e+00 1.70614913e-01
1.92600191e-01 1.26884139e+00 -2.06855863e-01 -2.30115280e-01
1.73192143e-01 2.76068240e-01 4.58257109e-01 -9.97097373e-01
-1.27806020e+00 -1.19224384e-01 7.98020303e-01 -3.53424191e-01
-8.28515410e-01 -5.40368021e-01 -1.68354440e+00 -7.73553401e-02
-2.83704132e-01 5.45888960e-01 5.34566164e-01 1.10604405e+00
1.66904002e-01 7.10161850e-02 5.16001701e-01 -2.54732370e-01
-7.71535099e-01 -1.27906501e+00 -1.28250733e-01 1.25065529e+00
-4.73107666e-01 -2.65167087e-01 -6.63341880e-01 4.71907288e-01] | [13.073317527770996, 7.729450702667236] |
7b8bd8f3-275a-4c1e-b53b-ed31e587052c | heartbeat-classification-fusing-temporal-and | null | null | https://www.researchgate.net/publication/327263145_Heartbeat_classification_fusing_temporal_and_morphological_information_of_ECGs_via_ensemble_of_classifiers | https://www.researchgate.net/publication/327263145_Heartbeat_classification_fusing_temporal_and_morphological_information_of_ECGs_via_ensemble_of_classifiers | Heartbeat classification fusing temporal and morphological information of ECGs via ensemble of classifiers | A method for the automatic classification of electrocardiograms (ECG) based on the combination of multiple Support Vector Machines (SVMs) is presented in this work. The method relies on the time intervals between consequent beats and their morphology for the ECG characterisation. Different descriptors based on wavelets, local binary patterns (LBP), higher order statistics (HOS) and several amplitude values were employed. Instead of concatenating all these features to feed a single SVM model, we propose to train specific SVM models for each type of feature. In order to obtain the final prediction, the decisions of the different models are combined with the product, sum, and majority rules. The designed methodology approaches are tested on the public MIT-BIH arrhythmia database, classifying four kinds of abnormal and normal beats. Our approach based on an ensemble of SVMs offered a satisfactory performance, improving the results when compared to a single SVM model using the same features. Additionally, our approach also showed better results in comparison with previous machine learning approaches of the state-of-the-art. | ['V.Mondéjar-Guerr', 'J.Rouco', 'J.Novo', 'M.Ortega', 'M.G.Penedo'] | 2019-01-01 | null | null | null | biomedical-signal-processing-and-control-2019 | ['heartbeat-classification', 'electrocardiography-ecg'] | ['medical', 'methodology'] | [ 4.87591237e-01 -1.36183083e-01 1.01573437e-01 -3.39754075e-01
-4.04653639e-01 -3.41542512e-01 5.04917562e-01 8.54870915e-01
-4.09506649e-01 9.23283875e-01 -4.12308276e-01 -2.43320167e-01
-4.83702302e-01 -6.29021406e-01 4.04584641e-03 -8.77526522e-01
-1.29393876e-01 4.71382946e-01 5.31200826e-01 -6.15694150e-02
4.78203624e-01 7.69855261e-01 -1.77268195e+00 6.94164217e-01
8.36805999e-01 1.27010667e+00 -1.96598694e-02 6.04004800e-01
1.23995446e-01 3.55922520e-01 -6.87795937e-01 2.24522293e-01
-2.18602759e-03 -6.82392657e-01 -4.60498214e-01 -9.83685926e-02
-4.39957380e-02 2.52253115e-01 3.15415591e-01 6.20007396e-01
5.61727405e-01 -1.61964670e-01 9.48886633e-01 -9.62283254e-01
1.77486986e-01 3.94891500e-01 -1.02821946e-01 4.42733407e-01
6.02381766e-01 -2.78709203e-01 6.34873629e-01 -7.06785381e-01
4.73224342e-01 4.44358468e-01 1.02336776e+00 -5.99540286e-02
-1.33757019e+00 -2.85199732e-01 -4.23916847e-01 6.58729315e-01
-1.36052275e+00 -1.93311378e-01 7.54334688e-01 -7.23400176e-01
7.89884031e-01 5.58034360e-01 9.46448922e-01 6.69677794e-01
3.64215374e-01 2.16600895e-01 1.57161593e+00 -7.04834342e-01
3.43517840e-01 6.01278484e-01 5.71688354e-01 6.10549152e-01
2.59513855e-01 -8.79948810e-02 -2.13047937e-01 -5.93745053e-01
3.99727881e-01 -7.68098906e-02 -9.51196551e-02 -2.94493556e-01
-1.09299099e+00 8.02619696e-01 -1.34269983e-01 9.73258674e-01
-7.02089190e-01 -4.59473312e-01 6.33173347e-01 1.09272532e-01
2.26109460e-01 3.04226905e-01 -5.46414852e-01 -8.63732770e-02
-1.21561134e+00 2.21868709e-01 9.28871155e-01 -1.45852352e-02
6.25688612e-01 -3.62709574e-02 -2.55751699e-01 5.26547253e-01
1.50596708e-01 2.57850140e-01 6.36662662e-01 -2.79920638e-01
1.27951935e-01 8.50430906e-01 -5.67336567e-02 -1.15386987e+00
-6.71764851e-01 -5.54247081e-01 -8.42749655e-01 4.18309897e-01
5.56703448e-01 -1.08035766e-01 -6.01582468e-01 1.08991408e+00
2.04765171e-01 1.71247140e-01 2.08379000e-01 5.09046257e-01
7.28600144e-01 4.48623955e-01 -9.76296216e-02 -4.63136911e-01
1.06727612e+00 -4.49790299e-01 -8.15950811e-01 4.54598904e-01
5.60826063e-01 -7.42069483e-01 4.54456300e-01 9.65838015e-01
-8.25696468e-01 -8.41349602e-01 -1.28134620e+00 7.75506616e-01
-4.93710518e-01 5.33694685e-01 1.53502345e-01 7.98257887e-01
-8.68343651e-01 1.18306959e+00 -6.32677615e-01 -4.49271798e-01
7.09297806e-02 4.43651915e-01 -3.24595809e-01 6.20891690e-01
-9.93771315e-01 1.22589028e+00 5.43802142e-01 1.19984433e-01
5.19046523e-02 -1.88562006e-01 -7.02147782e-01 2.31718179e-02
-3.49789441e-01 -3.04468721e-01 5.07766485e-01 -1.06222641e+00
-1.45290422e+00 6.61274076e-01 -6.83783293e-02 -6.76288307e-01
6.19522691e-01 2.32006580e-01 -4.69107032e-01 4.35754687e-01
-2.35580042e-01 -6.30153865e-02 7.35332549e-01 -1.04998970e+00
-5.36446929e-01 -3.00416589e-01 -5.16341507e-01 -1.14997394e-01
-2.93673098e-01 9.15681869e-02 3.82333070e-01 -4.23865765e-01
2.90191501e-01 -7.50662804e-01 -1.36171311e-01 -5.88920653e-01
-2.82945901e-01 -1.58207923e-01 6.82085395e-01 -9.61789906e-01
1.55992639e+00 -2.34158444e+00 5.27252316e-01 7.23722637e-01
-1.20470628e-01 5.18539488e-01 5.30666232e-01 5.24030983e-01
-1.40434816e-01 -3.19788188e-01 -3.49376380e-01 -8.76514241e-03
-3.55574459e-01 3.07906598e-01 -1.23390285e-02 5.03703833e-01
8.98003057e-02 2.47180969e-01 -4.88507807e-01 -7.86595106e-01
5.99157989e-01 3.66610646e-01 -5.45673296e-02 -5.87559715e-02
3.64148021e-01 3.95029545e-01 -3.24935853e-01 2.74782300e-01
4.79320973e-01 8.29087868e-02 4.49438781e-01 -3.86296004e-01
-1.03262708e-01 -1.59069732e-01 -1.47892773e+00 1.22101033e+00
-1.46666080e-01 1.63773164e-01 -5.49801052e-01 -1.49615502e+00
1.40233243e+00 7.18851745e-01 6.15368068e-01 -1.54360339e-01
1.59676746e-01 6.36157811e-01 2.74058789e-01 -5.92057884e-01
-2.06881061e-01 -1.29707088e-03 9.68040824e-02 -1.06720179e-02
3.01651269e-01 9.79750529e-02 3.48862588e-01 -4.78427619e-01
8.66897464e-01 2.19862327e-01 8.62077594e-01 -3.43550801e-01
1.05692136e+00 -1.81166366e-01 2.31288061e-01 4.87754345e-01
-1.53393269e-01 5.64651251e-01 6.30007803e-01 -8.06264997e-01
-6.75423205e-01 -7.73910940e-01 -5.61088622e-01 2.69135326e-01
-1.62272349e-01 -2.34106988e-01 -7.86812425e-01 -6.64225459e-01
5.50872833e-02 4.56027865e-01 -4.86412674e-01 1.44581452e-01
-5.55608332e-01 -9.79855955e-01 5.94261706e-01 1.43976942e-01
1.00431934e-01 -1.24763894e+00 -1.16796076e+00 4.23807085e-01
3.10368333e-02 -8.89065385e-01 4.49828088e-01 4.66588289e-01
-1.21373165e+00 -9.98673439e-01 -6.58221245e-01 -4.44888383e-01
2.50802994e-01 -5.48839986e-01 6.63827956e-01 5.89810498e-02
-5.03596783e-01 -2.30382923e-02 -5.96896291e-01 -5.18119514e-01
-6.97183311e-01 1.08280651e-01 4.29971963e-02 4.65387553e-01
5.57515733e-02 -8.31378758e-01 -2.40841240e-01 1.42264143e-01
-5.09913206e-01 -3.01491261e-01 7.47179508e-01 8.44886005e-01
4.51375246e-01 -1.84927076e-01 8.54871750e-01 -7.40444422e-01
4.96204406e-01 -3.53851765e-01 -3.13431829e-01 2.98280746e-01
-7.73435593e-01 -8.14563781e-02 8.68112266e-01 -3.45231086e-01
-5.15250623e-01 1.70079529e-01 -2.68204957e-01 -8.12922046e-02
-5.25850058e-01 3.95336539e-01 2.62124062e-01 -4.58530098e-01
8.86888146e-01 6.22213662e-01 1.17052443e-01 -4.81276870e-01
-3.52848589e-01 7.85646498e-01 1.87806696e-01 -1.93535432e-01
4.05946016e-01 3.04186307e-02 4.06023085e-01 -9.17670846e-01
-2.82021552e-01 -5.55643797e-01 -9.76543605e-01 -3.75973016e-01
8.87227058e-01 -2.84601063e-01 -4.76940602e-01 5.66823542e-01
-1.17265832e+00 3.57262820e-01 -2.83152699e-01 6.72514975e-01
-5.48069298e-01 6.82475209e-01 -3.62305790e-01 -1.19058883e+00
-4.27262783e-01 -1.16393661e+00 6.57858014e-01 1.21937007e-01
-5.54786384e-01 -8.52177143e-01 2.02943400e-01 8.93956870e-02
3.98776233e-01 7.08477497e-01 9.35623109e-01 -1.23333991e+00
8.82328600e-02 -5.73446393e-01 2.15232521e-01 5.89903235e-01
1.84221074e-01 1.20492734e-01 -7.83037722e-01 1.71286950e-03
1.53834000e-01 1.03194788e-01 7.04499483e-01 2.59912342e-01
9.42570984e-01 8.44932720e-03 -2.86015213e-01 3.17188323e-01
1.59653711e+00 5.94241440e-01 5.00254691e-01 4.11205769e-01
2.08767895e-02 2.74780720e-01 5.79648256e-01 6.95296824e-01
-2.68730614e-02 9.56563473e-01 2.46001765e-01 -2.70319939e-01
3.06157291e-01 4.30576056e-01 1.21329233e-01 4.96136397e-01
-6.70357406e-01 2.89679021e-01 -8.69211912e-01 2.77447909e-01
-1.81166470e+00 -1.03539526e+00 -4.68365997e-01 2.41288733e+00
4.80654389e-01 4.68296379e-01 4.81021762e-01 1.12057865e+00
6.43026173e-01 -1.45497695e-02 1.63527682e-01 -8.23719501e-01
-2.23841488e-01 7.78615475e-01 2.67973632e-01 2.85135478e-01
-1.29065645e+00 1.06565036e-01 6.09336185e+00 7.65356004e-01
-1.39109850e+00 1.21783614e-01 2.87083834e-01 3.50982636e-01
2.59705305e-01 -1.72967419e-01 -6.72551274e-01 6.30389392e-01
9.85875964e-01 1.48545504e-01 1.62495464e-01 6.52767301e-01
1.16547339e-01 -3.19813430e-01 -6.67010427e-01 9.67391551e-01
2.98501074e-01 -1.18391550e+00 -1.59551620e-01 -1.77229568e-01
3.57397974e-01 -3.04420203e-01 -3.49081486e-01 -1.27500251e-01
-9.37158763e-01 -7.92956352e-01 5.65774441e-01 9.36896205e-01
3.81996095e-01 -4.98134166e-01 1.23797083e+00 3.82818818e-01
-9.49406266e-01 -3.70244414e-01 1.36523515e-01 -1.72862276e-01
1.37518987e-01 6.74638271e-01 -7.73909092e-01 1.08683980e+00
4.33415890e-01 5.89305341e-01 -6.06808901e-01 1.30989933e+00
9.19629782e-02 6.08550131e-01 -4.30995226e-01 -2.88210988e-01
-5.85990995e-02 -2.59465694e-01 6.47764027e-01 1.38707161e+00
3.85890514e-01 -1.50757447e-01 2.06096351e-01 4.23132896e-01
1.03308034e+00 7.68406868e-01 -3.95534694e-01 2.41806060e-01
1.06687747e-01 1.33074880e+00 -1.01953614e+00 -6.13108695e-01
-2.24166885e-01 8.19784880e-01 -1.41357005e-01 -7.58094564e-02
-7.04900563e-01 -7.43353784e-01 -8.38757083e-02 2.21120343e-01
3.63446563e-01 5.90941347e-02 -5.87105870e-01 -8.88624370e-01
2.60582656e-01 -7.54032314e-01 4.70273197e-01 -3.20389658e-01
-9.91537571e-01 9.35911655e-01 1.41725719e-01 -1.56537950e+00
-3.71161699e-01 -5.83888233e-01 -6.18153870e-01 1.19395316e+00
-9.48412418e-01 -8.39815676e-01 -9.93074179e-02 3.48285496e-01
1.56962544e-01 -3.63026172e-01 1.34939528e+00 3.01532537e-01
-2.61386275e-01 2.43900910e-01 -7.31326491e-02 -1.73244804e-01
4.99409497e-01 -1.32998741e+00 -3.83459032e-01 6.03399813e-01
1.77845448e-01 1.32648692e-01 9.02073860e-01 -4.39452976e-01
-6.16193473e-01 -2.92807221e-01 1.33460748e+00 -1.44769654e-01
3.83156687e-01 1.76450312e-01 -7.60804594e-01 5.63577749e-02
8.62498358e-02 5.71907833e-02 8.64907682e-01 -1.96976103e-02
1.71954781e-01 -3.35558861e-01 -1.29534936e+00 -1.95459932e-01
1.67167559e-01 -1.11566722e-01 -9.38885570e-01 9.11459401e-02
-2.88390696e-01 -2.21382156e-01 -1.15178764e+00 6.53000772e-01
9.24737573e-01 -1.45095682e+00 8.93956602e-01 -3.39509428e-01
1.72931701e-01 -3.85197282e-01 -3.71339992e-02 -1.11509347e+00
-1.81396008e-01 -1.92634091e-01 -1.12770624e-01 7.82298148e-01
4.81677234e-01 -8.90263379e-01 3.99265021e-01 -1.60635129e-01
6.60573542e-02 -1.06592238e+00 -1.05639493e+00 -5.83954394e-01
-5.80182552e-01 7.65639991e-02 1.48908347e-01 8.01254392e-01
2.98534989e-01 8.59372392e-02 -2.68001080e-01 2.38419808e-02
5.99023938e-01 2.69853950e-01 2.95317799e-01 -1.68909872e+00
-7.68960118e-01 -4.19001997e-01 -1.21893072e+00 2.22844362e-01
-1.55916318e-01 -9.29979682e-01 -4.45698112e-01 -1.34972095e+00
-1.47949114e-01 -5.39828300e-01 -7.22958565e-01 5.40783465e-01
-3.19838263e-02 4.45558757e-01 2.29576036e-01 -7.86925852e-02
-5.18425973e-03 -1.45020634e-01 6.27539635e-01 2.64399856e-01
-5.29963851e-01 5.54988801e-01 -1.32686853e-01 7.60478675e-01
7.16532946e-01 -4.49720621e-01 -1.89395398e-01 4.12031084e-01
-2.69990504e-01 4.39459592e-01 3.64552647e-01 -1.70766485e+00
-4.78896461e-02 2.21866593e-01 7.05923975e-01 -6.60599470e-01
4.18314546e-01 -7.85809994e-01 5.94312906e-01 1.05014253e+00
-1.39668822e-01 1.20237589e-01 8.05099756e-02 2.09324270e-01
-4.41491395e-01 -4.67202842e-01 8.33124638e-01 2.59938426e-02
-3.44187915e-01 -3.04275662e-01 -4.64657217e-01 -6.47675574e-01
1.30121136e+00 -6.88417852e-01 2.86770016e-01 9.72808227e-02
-1.46533978e+00 -5.43714643e-01 3.25542130e-02 -8.95351246e-02
4.74270403e-01 -9.57199693e-01 -7.28419721e-01 2.80354500e-01
2.02221066e-01 -8.01592648e-01 3.64703000e-01 1.34781945e+00
-7.09341943e-01 4.11250949e-01 -8.45404625e-01 -7.21837521e-01
-1.81605875e+00 3.87312621e-01 4.31303442e-01 -4.69493330e-01
-4.20217961e-01 2.83765733e-01 -7.48684704e-01 6.83750957e-02
5.88313956e-03 -4.24078107e-01 -9.61359501e-01 5.46762764e-01
2.28337944e-01 5.17213762e-01 4.71049368e-01 -6.90300703e-01
-5.48208416e-01 8.18948269e-01 5.66909790e-01 -1.58323094e-01
1.27207971e+00 4.17376190e-01 -2.26187423e-01 7.90159345e-01
9.34065104e-01 2.24539172e-02 -3.11126202e-01 1.74013451e-01
1.36058062e-01 -1.80750057e-01 -2.73443103e-01 -8.63637567e-01
-5.01788557e-01 8.76100361e-01 8.99351299e-01 6.59906507e-01
1.29903448e+00 -5.35785317e-01 2.57822901e-01 2.27441013e-01
5.02952039e-01 -7.72131801e-01 -5.61750174e-01 1.41859306e-02
5.35821021e-01 -6.21478438e-01 3.65355685e-02 -5.08863628e-01
-7.10857332e-01 1.76735163e+00 -1.33684635e-01 -3.92393142e-01
9.25606310e-01 1.33424312e-01 -2.12826319e-02 2.52880573e-01
-3.87811095e-01 -2.46711656e-01 4.00139868e-01 4.40568179e-01
4.41813648e-01 1.75443381e-01 -1.32312953e+00 7.98075497e-01
1.09227590e-01 4.82153505e-01 3.42213690e-01 8.63904119e-01
-5.09520054e-01 -1.47199833e+00 -4.49997813e-01 6.99353874e-01
-5.57760477e-01 1.94516227e-01 -6.55371919e-02 6.73796713e-01
6.36975884e-01 9.02722180e-01 -3.39233458e-01 -5.43262720e-01
3.91719490e-01 5.87065041e-01 6.87577546e-01 -4.33385998e-01
-1.03435993e+00 -9.16271657e-02 2.60852218e-01 -2.03810871e-01
-4.74495530e-01 -1.02809882e+00 -9.34812129e-01 4.32079285e-01
-1.94699436e-01 4.07344878e-01 9.28662837e-01 1.17747450e+00
7.10369125e-02 4.02045339e-01 8.02265406e-01 -8.43791127e-01
-7.55618274e-01 -1.01547086e+00 -8.10581267e-01 4.85146880e-01
1.87449768e-01 -7.37434268e-01 -5.28514266e-01 1.07085384e-01] | [14.182726860046387, 3.2487668991088867] |
8779bb3a-e536-49f7-ab79-cea1ed4b6887 | multi-slice-low-rank-tensor-decomposition | 2102.12056 | null | https://arxiv.org/abs/2102.12056v3 | https://arxiv.org/pdf/2102.12056v3.pdf | Multi-Slice Low-Rank Tensor Decomposition Based Multi-Atlas Segmentation: Application to Automatic Pathological Liver CT Segmentation | Liver segmentation from abdominal CT images is an essential step for liver cancer computer-aided diagnosis and surgical planning. However, both the accuracy and robustness of existing liver segmentation methods cannot meet the requirements of clinical applications. In particular, for the common clinical cases where the liver tissue contains major pathology, current segmentation methods show poor performance. In this paper, we propose a novel low-rank tensor decomposition (LRTD) based multi-atlas segmentation (MAS) framework that achieves accurate and robust pathological liver segmentation of CT images. Firstly, we propose a multi-slice LRTD scheme to recover the underlying low-rank structure embedded in 3D medical images. It performs the LRTD on small image segments consisting of multiple consecutive image slices. Then, we present an LRTD-based atlas construction method to generate tumor-free liver atlases that mitigates the performance degradation of liver segmentation due to the presence of tumors. Finally, we introduce an LRTD-based MAS algorithm to derive patient-specific liver atlases for each test image, and to achieve accurate pairwise image registration and label propagation. Extensive experiments on three public databases of pathological liver cases validate the effectiveness of the proposed method. Both qualitative and quantitative results demonstrate that, in the presence of major pathology, the proposed method is more accurate and robust than state-of-the-art methods. | ['Heng-Da Cheng', 'Haotian Wang', 'Xiancheng Zhou', 'Min Xian', 'Changfa Shi'] | 2021-02-24 | null | null | null | null | ['liver-segmentation'] | ['medical'] | [-1.46672741e-01 -1.82272136e-01 -4.96543348e-02 -1.92506790e-01
-9.87067819e-01 -5.49341440e-01 2.02117637e-01 1.78780064e-01
-2.29501259e-02 2.60528505e-01 4.87228572e-01 -2.35705495e-01
-3.00682873e-01 -4.31180686e-01 -1.91694990e-01 -1.08849669e+00
-2.05468878e-01 7.14885831e-01 2.53862530e-01 1.50740325e-01
-1.82393476e-01 7.22366512e-01 -5.47151268e-01 3.05455439e-02
9.24517035e-01 6.76886857e-01 1.32763445e-01 4.70097661e-01
1.37618288e-01 4.63392049e-01 -1.55920595e-01 6.47906587e-03
5.54638624e-01 -5.58087468e-01 -9.93282616e-01 7.55250216e-01
1.63514867e-01 -2.24640682e-01 -2.04972401e-01 1.14656711e+00
5.42274117e-01 -1.74127549e-01 6.43814087e-01 -7.39131510e-01
-1.43490314e-01 5.65864086e-01 -7.59224355e-01 2.08646610e-01
5.55031709e-02 1.61495924e-01 4.94116813e-01 -9.33294296e-01
4.15343940e-01 6.06322944e-01 6.54520690e-01 1.89677820e-01
-1.28921056e+00 -3.46140325e-01 -2.57340133e-01 -2.33197421e-01
-1.57951224e+00 -5.35722170e-03 7.78369486e-01 -6.65001333e-01
1.15962870e-01 4.38032061e-01 8.93223464e-01 1.74212232e-01
4.08595979e-01 6.70264721e-01 1.43227077e+00 -2.05338940e-01
-8.15791041e-02 -2.70687848e-01 9.08669457e-02 1.18353617e+00
4.82076734e-01 -4.35795560e-02 5.44729941e-02 -4.64449197e-01
9.04561222e-01 5.51621057e-02 -5.39625585e-01 -7.69036114e-01
-2.03691483e+00 5.86464167e-01 6.77837789e-01 7.81085134e-01
-7.31865942e-01 -2.17847988e-01 3.55968088e-01 -2.49524951e-01
3.18222553e-01 1.78569287e-01 1.19764410e-01 6.50125444e-01
-1.27176750e+00 -2.82530367e-01 6.81434274e-01 7.06666768e-01
3.67873818e-01 -2.16219239e-02 -3.15660894e-01 6.06966078e-01
5.44426858e-01 2.80552566e-01 6.66681767e-01 -5.09869814e-01
-2.25867435e-01 6.54166460e-01 -3.11630815e-01 -9.95714664e-01
-5.82689106e-01 -7.27362275e-01 -1.48395371e+00 -3.09810787e-01
4.46679622e-01 2.47895762e-01 -8.88370395e-01 1.10234082e+00
7.56191671e-01 4.68916923e-01 1.54441614e-02 1.29752994e+00
9.11029756e-01 1.67234734e-01 3.91895026e-02 -5.22731066e-01
1.62503827e+00 -8.92699540e-01 -5.27547598e-01 2.18213245e-01
7.19916344e-01 -9.84438539e-01 6.19971216e-01 1.18152246e-01
-8.69704425e-01 -8.03631768e-02 -7.47920394e-01 3.56929034e-01
4.40801471e-01 5.71161985e-01 5.17842174e-01 5.35470784e-01
-9.51333463e-01 8.49761665e-02 -1.21354115e+00 -2.10062280e-01
2.07110673e-01 5.89419901e-01 -5.53558171e-01 -3.81167948e-01
-6.43230617e-01 8.35508823e-01 2.53871530e-01 3.49700928e-01
-1.03373611e+00 -7.80119300e-01 -8.60044956e-01 -3.73779356e-01
2.68348575e-01 -7.91788816e-01 9.90514278e-01 -5.34560144e-01
-1.29986489e+00 9.17689741e-01 -1.55471265e-01 -1.85219690e-01
5.10208130e-01 4.92060423e-01 -1.06992433e-02 4.39715207e-01
1.67654753e-01 3.56586784e-01 5.70632041e-01 -1.49892783e+00
-3.75671275e-02 -3.46759647e-01 -4.35213625e-01 2.70606726e-01
2.71212552e-02 -2.13767234e-02 -5.55648208e-01 -7.82178879e-01
7.96104550e-01 -1.28986096e+00 -8.45181048e-01 -2.33375095e-02
-5.62876821e-01 1.52432933e-01 6.12429857e-01 -9.85705018e-01
9.40325081e-01 -2.15969443e+00 3.12045336e-01 5.42782128e-01
4.89042133e-01 -1.19596599e-02 9.06665809e-03 -2.26322547e-01
-1.10713346e-02 -4.65221815e-02 -5.60058236e-01 -2.53368348e-01
-4.28216249e-01 3.45110714e-01 3.88735026e-01 1.02934504e+00
-3.43540430e-01 9.07905579e-01 -9.75410581e-01 -9.55598950e-01
3.52191925e-01 5.24244249e-01 -4.19354856e-01 1.41987532e-01
3.21597010e-01 1.25146580e+00 -4.91881579e-01 8.68954718e-01
6.20317817e-01 -4.79375571e-01 3.27789605e-01 -7.79125035e-01
-1.54800072e-01 -2.02652454e-01 -9.60924506e-01 1.82153273e+00
-2.73240864e-01 -3.69629376e-02 3.34384859e-01 -6.95252836e-01
5.81677020e-01 7.06342697e-01 1.27299786e+00 -1.47860289e-01
2.46107295e-01 5.23285627e-01 1.94310084e-01 -2.98080564e-01
3.75729203e-02 -3.55649859e-01 1.36279374e-01 5.06268084e-01
-1.91838592e-01 -4.15717989e-01 2.22901821e-01 2.09013462e-01
8.29226911e-01 -3.20654511e-01 3.76254350e-01 -7.07086861e-01
9.26330924e-01 2.08055079e-01 7.13882446e-01 1.36152148e-01
-4.40287322e-01 7.50898421e-01 1.61480352e-01 -4.66091394e-01
-7.74122775e-01 -8.80387843e-01 -2.90339679e-01 2.76872724e-01
1.93490356e-01 -1.28931016e-01 -7.62537360e-01 -9.24804032e-01
-2.03916639e-01 4.75962497e-02 -3.24231565e-01 2.58717984e-01
-7.23491311e-01 -1.41567600e+00 4.00610566e-01 4.79713753e-02
5.57753205e-01 -3.22581708e-01 -3.64166141e-01 2.29841009e-01
-6.75292253e-01 -1.20577490e+00 -8.75522614e-01 -3.47600013e-01
-1.18195200e+00 -1.28194451e+00 -1.16756666e+00 -1.06246078e+00
1.45792043e+00 5.10225534e-01 9.29975152e-01 5.22964656e-01
-5.14204919e-01 3.14590245e-01 -2.05342591e-01 3.59503955e-01
-6.47309363e-01 -1.55616254e-01 8.42840150e-02 2.46327013e-01
-3.58288199e-01 -2.95888573e-01 -8.53771865e-01 5.67038298e-01
-9.96205688e-01 2.21766606e-01 9.88091886e-01 9.80215728e-01
9.87551570e-01 1.97272688e-01 1.80774238e-02 -7.36598790e-01
3.17772597e-01 -4.10064787e-01 -6.38553917e-01 3.54461372e-01
-5.26787043e-01 -7.35390112e-02 2.14254722e-01 -2.83742547e-01
-7.14200318e-01 5.75818300e-01 -1.73882972e-02 -4.47082281e-01
-1.48774870e-02 7.98027754e-01 3.16217959e-01 -7.73436964e-01
3.87102336e-01 4.12623227e-01 2.48829395e-01 -3.40460539e-01
2.19655409e-02 1.81121349e-01 4.45667356e-01 -4.84484941e-01
7.67159641e-01 6.44049346e-01 4.34355766e-01 -6.73243642e-01
-5.18343925e-01 -8.36028457e-01 -9.56829488e-01 -2.84619689e-01
8.43318284e-01 -8.08468342e-01 -4.19141352e-01 4.02102649e-01
-7.38710821e-01 -2.02172413e-01 -9.85436812e-02 9.34202194e-01
-3.25771660e-01 8.72590184e-01 -9.30936217e-01 -1.37010381e-01
-5.65967679e-01 -1.89510250e+00 9.32572901e-01 -1.04940198e-01
1.78220972e-01 -1.10901833e+00 1.08240182e-02 4.68587935e-01
4.81525123e-01 5.01339018e-01 1.02286041e+00 -4.80829954e-01
-8.82480085e-01 -1.15037426e-01 -1.53445005e-01 1.89540952e-01
3.88238072e-01 -2.52972066e-01 -1.89059615e-01 -5.56058824e-01
1.48314238e-01 2.71643847e-01 4.63874936e-01 6.86464608e-01
7.70314097e-01 -4.05231774e-01 -2.74985582e-01 6.99752986e-01
1.47298586e+00 -2.39078432e-01 1.97642490e-01 -2.91319359e-02
8.45568478e-01 4.52470511e-01 4.90771383e-01 2.69833118e-01
6.35891259e-01 5.72881520e-01 3.79417092e-01 -5.97274899e-01
-4.53691095e-01 2.58854717e-01 3.03072780e-02 1.50642204e+00
-1.09713987e-01 2.89847255e-01 -1.35033143e+00 8.36004913e-01
-1.57849526e+00 -2.93471992e-01 -5.64659715e-01 2.25533295e+00
8.19568694e-01 -4.37294960e-01 5.19899912e-02 9.65508539e-03
4.66370672e-01 -1.84341639e-01 -4.80710268e-02 4.90171790e-01
1.23182528e-01 -7.86400363e-02 6.65612996e-01 5.24642885e-01
-1.10541844e+00 5.38905203e-01 6.09586239e+00 4.93141174e-01
-1.23247457e+00 4.27721441e-01 6.03308737e-01 4.43906128e-01
-1.74984574e-01 6.57204306e-03 -2.16739878e-01 8.05793181e-02
3.14851940e-01 -2.39454523e-01 1.51996046e-01 4.60207105e-01
3.28772038e-01 -3.23866844e-01 -7.56003857e-01 7.61723876e-01
1.19122528e-01 -1.18072641e+00 8.83903354e-02 2.41223469e-01
9.35031593e-01 4.27148603e-02 -1.15773641e-01 -2.98133165e-01
2.64352322e-01 -8.13745499e-01 2.42435858e-01 4.39045280e-01
8.14311624e-01 -3.68034124e-01 1.04895806e+00 2.26820469e-01
-1.32164121e+00 3.86999339e-01 6.74754083e-02 6.73312902e-01
1.38725445e-01 1.02426076e+00 -1.30514181e+00 9.74655390e-01
3.13907951e-01 6.63171649e-01 -7.24216521e-01 1.42696536e+00
-1.93326890e-01 5.39163291e-01 -2.39574462e-01 4.86026317e-01
2.26895273e-01 -4.14916128e-01 8.19024026e-01 1.16029739e+00
3.82877886e-01 4.50492412e-01 8.41491342e-01 5.32968998e-01
6.79028034e-02 4.05407459e-01 -4.08974886e-01 3.12991142e-01
-4.08969410e-02 1.73541474e+00 -1.33586645e+00 -2.96168536e-01
-3.84268433e-01 8.61197412e-01 -3.11380059e-01 2.07124174e-01
-5.18242776e-01 4.93998230e-01 6.44215792e-02 1.58769518e-01
-2.88695216e-01 -5.11252880e-01 -2.77076691e-01 -1.37544703e+00
-2.40980580e-01 -9.23187852e-01 5.45164049e-01 -2.00111523e-01
-1.14525616e+00 7.53784835e-01 -9.64974519e-03 -1.41414928e+00
-1.53932469e-02 -2.15658516e-01 -3.84029478e-01 6.23283088e-01
-1.47747576e+00 -1.54646933e+00 -5.24231315e-01 6.61062419e-01
3.17974359e-01 1.70984671e-01 8.53386581e-01 4.90792006e-01
-3.85915518e-01 1.77274674e-01 -1.15291052e-01 5.22130907e-01
4.52570438e-01 -1.16696763e+00 -3.35489810e-01 1.09473181e+00
-3.26591283e-02 5.61288834e-01 4.47906196e-01 -7.08839715e-01
-1.52113974e+00 -1.22420120e+00 5.83385408e-01 -5.93993887e-02
4.21908408e-01 2.28241086e-01 -7.58369088e-01 7.27999926e-01
6.71721324e-02 6.15587592e-01 9.50203300e-01 -5.03320873e-01
8.68636072e-02 2.00690934e-03 -1.46351898e+00 4.45648372e-01
3.42150182e-01 -1.60938323e-01 -3.44165295e-01 6.37009621e-01
1.27583921e-01 -7.88162887e-01 -1.46423137e+00 7.67874897e-01
3.26293617e-01 -6.05735242e-01 1.04162717e+00 8.04490373e-02
-1.80526957e-01 -7.29861438e-01 -7.32433004e-03 -1.33833396e+00
-4.28570062e-01 -4.59236234e-01 2.98737466e-01 6.17761314e-01
7.39820078e-02 -3.97821814e-01 6.35259628e-01 4.98777032e-01
-3.92555982e-01 -6.70944870e-01 -1.11638021e+00 -5.56715786e-01
3.26085724e-02 1.62282154e-01 2.88335711e-01 1.20701134e+00
-1.12642616e-01 -1.98905349e-01 4.70489450e-03 6.06169581e-01
1.13307774e+00 3.32288325e-01 4.50436890e-01 -1.02096534e+00
-1.83736663e-02 -3.35637450e-01 -4.19580549e-01 -5.70118785e-01
-1.00402739e-02 -1.17201352e+00 1.85555935e-01 -1.72542262e+00
5.61729550e-01 -8.43258500e-01 -3.36974978e-01 4.50111151e-01
-1.84767708e-01 6.69530392e-01 1.13801077e-01 6.06352866e-01
-4.37968969e-01 2.46028781e-01 1.71277261e+00 -2.61907905e-01
-2.99679842e-02 1.18487086e-02 -2.81388909e-01 6.88948631e-01
5.90527773e-01 -5.60205281e-01 -5.27879819e-02 -2.55146861e-01
-3.83293688e-01 3.66209000e-01 5.74904501e-01 -8.70691419e-01
4.10389066e-01 -6.55394718e-02 2.17251420e-01 -4.49569792e-01
-8.50253850e-02 -9.99441028e-01 5.19167602e-01 1.14372432e+00
3.52061354e-02 2.55115330e-02 5.10772830e-03 2.14371845e-01
-3.74309033e-01 -4.73062657e-02 1.07127857e+00 -4.77373004e-01
-2.11652100e-01 6.03086233e-01 -3.98292601e-01 -2.43204087e-01
1.17740500e+00 1.82694525e-01 2.33064994e-01 -8.91350117e-03
-8.57142627e-01 2.85357744e-01 3.73614639e-01 -1.57704592e-01
7.81493485e-01 -1.32181680e+00 -1.11889970e+00 2.31637239e-01
-7.54906386e-02 3.59871298e-01 1.95455864e-01 1.90803957e+00
-8.72875452e-01 4.37720925e-01 5.60191423e-02 -1.00838506e+00
-1.45869303e+00 3.90075386e-01 6.30739331e-01 -7.31448293e-01
-7.69358158e-01 5.14939189e-01 2.80120939e-01 -5.38501143e-01
-1.62619606e-01 -5.84877074e-01 -6.28384622e-03 -2.78823674e-01
1.44671470e-01 7.32224882e-02 3.36193830e-01 -1.17030609e+00
-4.14370537e-01 7.44441926e-01 7.44730234e-02 5.12236133e-02
1.08406174e+00 -2.23307192e-01 -6.16449833e-01 -5.81789110e-03
9.23135161e-01 1.60603881e-01 -8.21678042e-01 -4.09761906e-01
6.81163743e-02 -5.08899808e-01 5.07301986e-01 -7.31188297e-01
-1.53016865e+00 5.17984867e-01 6.03703797e-01 -1.11184739e-01
1.21085489e+00 -9.41123813e-02 8.91257107e-01 -2.98933357e-01
6.16644263e-01 -2.24499404e-01 -3.17923985e-02 2.04144895e-01
7.79876709e-01 -1.10524988e+00 3.12613010e-01 -8.33441436e-01
-6.22070074e-01 1.09391057e+00 1.48946956e-01 9.73689780e-02
7.32835472e-01 1.92369059e-01 2.68619478e-01 -4.08221245e-01
-1.12936117e-01 -1.90617427e-01 6.47847652e-01 1.37602553e-01
5.11683762e-01 3.40207607e-01 -4.71733749e-01 1.32901892e-01
1.86579525e-01 -1.71601959e-02 3.88934672e-01 8.75846744e-01
-1.46941885e-01 -1.04770851e+00 -6.09586716e-01 2.68587649e-01
-6.99116766e-01 -8.73658061e-02 -9.67839509e-02 5.73601663e-01
-6.58867359e-02 5.78976810e-01 -4.59840924e-01 -5.47977760e-02
9.26438645e-02 -2.22392708e-01 5.19566894e-01 -5.64774215e-01
-8.11401606e-01 5.43808103e-01 -4.16545302e-01 -3.34338099e-01
-6.58057570e-01 -7.11471677e-01 -1.42852783e+00 -9.58434716e-02
-4.50570315e-01 3.20271671e-01 6.89735770e-01 7.75969326e-01
1.26693889e-01 5.37984371e-01 8.96293163e-01 -7.45604455e-01
-3.48090857e-01 -6.96767747e-01 -4.28548962e-01 3.69549662e-01
4.44498897e-01 -4.92993474e-01 -2.15456560e-01 1.56136647e-01] | [14.400650978088379, -2.6542952060699463] |
95594b7f-d09a-4777-8da1-1d92fab57209 | one-shot-learning-for-autonomous-aerial | 2206.01411 | null | https://arxiv.org/abs/2206.01411v1 | https://arxiv.org/pdf/2206.01411v1.pdf | One-shot Learning for Autonomous Aerial Manipulation | This paper is concerned with learning transferable contact models for aerial manipulation tasks. We investigate a contact-based approach for enabling unmanned aerial vehicles with cable-suspended passive grippers to compute the attach points on novel payloads for aerial transportation. This is the first time that the problem of autonomously generating contact points for such tasks has been investigated. Our approach builds on the underpinning idea that we can learn a probability density of contacts over objects' surfaces from a single demonstration. We enhance this formulation for encoding aerial transportation tasks while maintaining the one-shot learning paradigm without handcrafting task-dependent features or employing ad-hoc heuristics; the only prior is extrapolated directly from a single demonstration. Our models only rely on the geometrical properties of the payloads computed from a point cloud, and they are robust to partial views. The effectiveness of our approach is evaluated in simulation, in which one or three quadropters are requested to transport previously unseen payloads along a desired trajectory. The contact points and the quadroptors configurations are computed on-the-fly for each test by our apporach and compared with a baseline method, a modified grasp learning algorithm from the literature. Empirical experiments show that the contacts generated by our approach yield a better controllability of the payload for a transportation task. We conclude this paper with a discussion on the strengths and limitations of the presented idea, and our suggested future research directions. | ['Eliseo Ferrante', 'Claudio Zito'] | 2022-06-03 | null | null | null | null | ['one-shot-learning'] | ['methodology'] | [ 3.04421484e-01 2.51958877e-01 1.02124631e-01 1.08611383e-01
-5.16212225e-01 -1.09902275e+00 5.20599604e-01 -7.18322471e-02
-2.81453758e-01 1.01448512e+00 -5.25978029e-01 -1.62733734e-01
-7.75422633e-01 -6.95000172e-01 -1.21611738e+00 -8.10695648e-01
-6.19823515e-01 6.40113294e-01 3.12140733e-01 -4.71590698e-01
1.80382758e-01 7.59026170e-01 -1.87460864e+00 -6.63031861e-02
1.02558303e+00 1.08730352e+00 5.85685492e-01 9.76010323e-01
6.13410175e-01 6.52583912e-02 -5.11273265e-01 2.90141441e-02
9.18704987e-01 5.61598465e-02 -6.17544711e-01 4.72169995e-01
3.01816404e-01 -5.93140900e-01 -4.76098210e-02 6.21760607e-01
2.14508459e-01 3.97900760e-01 9.22790766e-01 -1.45817387e+00
9.74434018e-02 1.27955511e-01 6.65014014e-02 -3.89363885e-01
4.89833802e-01 5.41374743e-01 7.32617438e-01 -5.25223076e-01
5.91109872e-01 8.13486993e-01 5.58646262e-01 4.21363384e-01
-8.61802399e-01 -1.23749524e-01 3.02650630e-01 -2.07880557e-01
-1.13591862e+00 -4.69678417e-02 4.77425784e-01 -6.55544043e-01
8.42389286e-01 1.31236941e-01 9.92332637e-01 1.14404500e+00
6.76636457e-01 6.15145504e-01 8.38100612e-01 -3.02815735e-01
5.61821759e-01 -4.61544283e-02 -5.95483422e-01 1.09552419e+00
5.78583300e-01 3.86909366e-01 -1.77367076e-01 -2.59356409e-01
8.68153393e-01 -4.94789928e-02 -5.02477765e-01 -1.16971958e+00
-1.33610785e+00 5.33213556e-01 4.77474302e-01 -7.80054107e-02
-5.92006207e-01 8.85735825e-02 2.11580712e-02 4.99438941e-01
2.38078251e-01 6.78413808e-01 -7.90442646e-01 -8.34422633e-02
-7.21695185e-01 8.02395701e-01 1.25197983e+00 1.77150524e+00
4.36261982e-01 2.31090654e-02 1.00319572e-01 -6.20842166e-02
1.59680247e-01 7.37495184e-01 -1.81188852e-01 -1.07582653e+00
6.20444000e-01 4.45646822e-01 9.24346268e-01 -7.84699857e-01
-2.08788335e-01 -9.22164321e-02 -1.57043457e-01 7.00247705e-01
1.96145892e-01 -8.06871116e-01 -9.77872193e-01 1.30713904e+00
3.32811534e-01 -1.83968812e-01 3.00178826e-01 9.78581667e-01
5.34297004e-02 7.26475358e-01 -3.84854078e-01 -8.63536522e-02
7.40844727e-01 -8.09082627e-01 -2.43660450e-01 1.70497492e-01
4.99018729e-01 -3.95829648e-01 8.21984589e-01 7.15173185e-01
-1.00779748e+00 -4.82113928e-01 -1.10838795e+00 4.90725338e-01
-3.50505173e-01 4.39856172e-01 5.68891585e-01 1.62749395e-01
-8.78219783e-01 1.07986045e+00 -1.17626274e+00 -6.07667923e-01
1.15755893e-01 6.21033967e-01 -1.00999102e-01 9.80513096e-02
-7.08777606e-01 1.00824785e+00 3.05784106e-01 1.88531324e-01
-1.55099881e+00 -4.78493541e-01 -7.39910126e-01 -1.82597160e-01
8.22868288e-01 -9.67416108e-01 1.28715765e+00 -6.30107939e-01
-1.81733155e+00 1.74497083e-01 3.84077579e-01 -3.40449214e-01
6.70311749e-01 -5.03416181e-01 4.07423794e-01 3.97845507e-01
-1.05811514e-01 8.62444699e-01 1.14581621e+00 -1.62998295e+00
-8.17665040e-01 -6.06381483e-02 7.16318429e-01 4.96791720e-01
-1.46992402e-02 -7.46858954e-01 8.67579505e-02 -4.53387022e-01
-7.16487542e-02 -1.46482778e+00 -2.90262461e-01 5.98421276e-01
-3.09844345e-01 -7.85460472e-02 9.86595750e-01 -1.48424879e-01
2.64520705e-01 -1.58927608e+00 1.08532941e+00 1.12079777e-01
-3.98213089e-01 1.44233942e-01 -7.18880892e-02 1.06080616e+00
4.49183643e-01 -1.35851771e-01 -5.51451504e-01 -1.36623336e-02
2.54016202e-02 3.49075973e-01 -6.71387255e-01 4.21414256e-01
3.24772596e-01 5.07776082e-01 -1.11465907e+00 -1.96633399e-01
2.40969792e-01 2.77028531e-01 -5.04258692e-01 5.13024211e-01
-7.71162033e-01 4.92116868e-01 -8.78942728e-01 9.91148174e-01
3.79931003e-01 4.68594313e-01 1.96633846e-01 -1.75240681e-01
-5.63539386e-01 -1.90292388e-01 -1.01328993e+00 1.94894731e+00
-5.04888475e-01 1.26012743e-01 4.63191748e-01 -7.99261987e-01
1.06415176e+00 2.95037061e-01 5.60502589e-01 5.39002895e-01
3.09668124e-01 3.45978945e-01 -2.25716203e-01 -7.61401176e-01
6.24877810e-01 4.25521731e-02 -1.99512169e-01 -7.96336234e-02
2.80453056e-01 -1.06433594e+00 4.79061343e-02 -1.85726970e-01
9.89499152e-01 9.78475273e-01 3.61550674e-02 -4.86903489e-01
1.62121668e-01 5.56427598e-01 2.46471539e-01 4.16459233e-01
3.15775722e-01 2.77635813e-01 2.30407268e-01 -3.53284508e-01
-9.95664716e-01 -1.04154015e+00 8.13468993e-02 4.73944604e-01
4.32830185e-01 -1.46433070e-01 -9.21907842e-01 -6.99257135e-01
3.56501341e-01 6.46290362e-01 -4.53060448e-01 1.15611978e-01
-3.22050929e-01 -1.74475402e-01 1.43499836e-01 4.96354371e-01
2.04244569e-01 -8.54747772e-01 -1.54232514e+00 7.65923783e-02
7.04789087e-02 -9.20919657e-01 1.29130885e-01 2.88992673e-01
-1.17177498e+00 -1.36222482e+00 -9.56481278e-01 -7.09425628e-01
1.05020893e+00 2.81703383e-01 4.84346122e-01 -7.34147206e-02
-4.12360251e-01 1.12863338e+00 -6.52959645e-01 -5.57219207e-01
-7.32958093e-02 1.40592501e-01 4.53055590e-01 -3.24781358e-01
-3.51294428e-01 -5.06910563e-01 -2.70642221e-01 4.36305225e-01
-7.34754682e-01 -5.42862713e-02 7.46190667e-01 6.46429539e-01
4.93634045e-01 3.40665519e-01 6.76732659e-02 -2.18412951e-02
5.63911915e-01 -3.49394143e-01 -1.01881945e+00 1.81527004e-01
-3.12427543e-02 -1.04577690e-01 6.49630845e-01 -5.25353611e-01
-7.14066207e-01 5.70157230e-01 6.21966362e-01 -8.07978392e-01
-3.25228989e-01 2.91426629e-01 -1.31185487e-01 -5.20423532e-01
3.22782546e-01 4.68036672e-03 2.08933517e-01 -1.52138829e-01
1.78630754e-01 4.40594971e-01 1.26047432e-01 -1.25618875e+00
1.10564578e+00 3.56727600e-01 5.01488745e-01 -1.04318047e+00
-4.19990361e-01 -7.53206387e-02 -1.18167770e+00 -4.78625208e-01
7.23611355e-01 -6.48944080e-01 -9.30046558e-01 3.16460043e-01
-1.28713870e+00 -6.56232834e-01 -4.96254504e-01 6.31017447e-01
-1.26261759e+00 2.69634604e-01 -4.90079075e-02 -1.05697525e+00
-5.04896007e-02 -1.28315580e+00 1.37945628e+00 7.47320279e-02
1.05750583e-01 -8.57164264e-01 -1.06972056e-02 -1.67961821e-01
1.75109655e-01 8.29959333e-01 5.76026976e-01 -2.25621045e-01
-1.05060089e+00 -2.55908817e-01 4.03944165e-01 1.17210828e-01
1.50529668e-01 1.67482078e-01 -5.49061596e-01 -8.74320567e-01
-1.11878678e-01 -5.51481068e-01 4.86896545e-01 2.86399513e-01
7.70362794e-01 -4.14539874e-01 -7.77413189e-01 3.92208934e-01
1.48852384e+00 2.62264639e-01 1.19219877e-01 1.87622696e-01
3.83299649e-01 9.33507085e-01 1.30008042e+00 6.91352129e-01
3.62530500e-02 6.64514065e-01 1.14574039e+00 4.25757796e-01
5.51210701e-01 -2.80944675e-01 4.86209780e-01 4.12597179e-01
-8.01010072e-01 -6.69826388e-01 -7.38745749e-01 7.67537177e-01
-1.77445710e+00 -4.29521620e-01 1.32074669e-01 2.35465646e+00
1.87996522e-01 -6.60539791e-02 -7.48161506e-03 -1.22324176e-01
4.58565295e-01 -1.82138994e-01 -6.54836118e-01 -9.83047560e-02
5.85722864e-01 1.81099419e-02 8.35476339e-01 5.69465995e-01
-1.14599657e+00 8.83695245e-01 5.60873556e+00 1.57986581e-01
-8.10642719e-01 -5.53438008e-01 -4.69831020e-01 -7.60575682e-02
1.07996561e-01 3.28431785e-01 -8.02183390e-01 1.14295304e-01
5.34891367e-01 -3.03488374e-02 5.59085786e-01 9.23920929e-01
-3.67085189e-02 -2.89344609e-01 -1.27913523e+00 3.08321178e-01
1.33493066e-01 -8.61047626e-01 2.61237711e-01 2.07899451e-01
5.33346236e-01 -2.41034761e-01 8.08000639e-02 1.18461370e-01
2.76639134e-01 -4.42829698e-01 8.86370659e-01 9.49723065e-01
6.99783564e-01 -5.91119230e-01 6.17267072e-01 5.98966241e-01
-1.17975819e+00 -3.90533984e-01 -5.26155114e-01 -1.73105836e-01
3.46296787e-01 -9.65026096e-02 -9.95015383e-01 1.04862607e+00
5.03471076e-01 5.81157207e-01 3.28409337e-02 1.19893360e+00
-1.62724972e-01 1.42596468e-01 -5.49447119e-01 -4.94849294e-01
5.71814597e-01 -2.37999275e-01 1.14909923e+00 6.02811694e-01
8.61956835e-01 2.70491391e-01 5.97762942e-01 8.28690946e-01
4.66887355e-01 -2.96192408e-01 -1.04063618e+00 -6.54608831e-02
1.97643846e-01 1.17691803e+00 -6.52548671e-01 -5.94366528e-02
5.10997251e-02 8.08211088e-01 -1.58797935e-01 3.16589087e-01
-9.16930497e-01 -7.15162456e-01 3.81625682e-01 2.74349779e-01
7.57233858e-01 -7.80048251e-01 4.79626954e-01 -9.17740643e-01
1.51947677e-01 -5.55071294e-01 -3.22818875e-01 -9.56766188e-01
-1.03078473e+00 6.90433025e-01 7.82079816e-01 -1.91239381e+00
-2.98146546e-01 -1.01492214e+00 -4.67703551e-01 5.53218603e-01
-1.41141450e+00 -1.33890784e+00 -5.74269354e-01 2.52334982e-01
8.29683423e-01 -3.21020931e-01 8.66339266e-01 -5.76321006e-01
1.09581485e-01 -1.12175785e-01 -2.57186294e-01 -4.85054433e-01
3.34878147e-01 -1.22527862e+00 2.28945300e-01 5.07806003e-01
-4.22837794e-01 4.46287602e-01 8.75398993e-01 -1.10964131e+00
-1.92493403e+00 -1.02011943e+00 -1.37796607e-02 -5.08715868e-01
6.57488823e-01 -3.22282642e-01 -4.84634459e-01 7.63324738e-01
1.00936256e-01 -1.99785203e-01 2.90551428e-02 -3.45867753e-01
5.95626414e-01 4.65042777e-02 -1.32263672e+00 4.33843613e-01
1.06605327e+00 1.63471714e-01 -8.11620653e-01 6.42688215e-01
8.00405741e-01 -7.68596113e-01 -9.54604983e-01 7.65077174e-01
5.83892584e-01 -4.83380407e-01 7.11769104e-01 -7.48169720e-01
1.84617549e-01 -4.08302337e-01 -2.51106292e-01 -1.59599638e+00
6.73202658e-03 -1.09886539e+00 7.86876678e-03 7.07868338e-01
2.21950367e-01 -4.65303957e-01 6.90629959e-01 3.16044748e-01
-6.29898608e-01 -1.00025356e+00 -8.01752806e-01 -1.17334974e+00
4.18673232e-02 1.91367224e-01 4.49729174e-01 3.82336766e-01
-6.22392520e-02 -1.91209972e-01 -3.08509409e-01 8.31210554e-01
6.98409259e-01 7.30327889e-02 1.01814473e+00 -1.39541686e+00
-3.20082456e-01 3.57638389e-01 -2.88179338e-01 -1.08074164e+00
3.11745793e-01 -5.86194158e-01 5.58583856e-01 -1.60010898e+00
-4.94500637e-01 -7.34598815e-01 3.98522466e-01 3.87749523e-01
5.24037421e-01 -6.92242801e-01 3.25131655e-01 4.09522094e-02
-2.34052509e-01 7.91580081e-01 1.51669335e+00 -7.69516407e-03
-3.73145640e-01 4.57880825e-01 5.22959940e-02 8.18298459e-01
8.54947209e-01 -3.26815188e-01 -7.59728014e-01 -5.94835103e-01
-2.53155697e-02 4.43716526e-01 6.87233627e-01 -1.42689097e+00
4.13372606e-01 -2.26168409e-01 1.47750527e-01 -4.79774743e-01
8.37420821e-01 -1.39803541e+00 -5.29733254e-03 6.53506696e-01
-1.45797223e-01 2.54519761e-01 4.75135118e-01 1.07079637e+00
3.12151074e-01 -4.95550454e-01 2.01084569e-01 -3.45315725e-01
-3.94096494e-01 3.30709934e-01 -6.46549761e-01 -3.94959211e-01
1.58882487e+00 -2.16825604e-01 -2.62179133e-02 -1.89890429e-01
-7.75311112e-01 1.99421078e-01 8.01471412e-01 2.70870030e-01
7.95388997e-01 -6.85085356e-01 -3.94885451e-01 2.40502790e-01
-1.83378413e-01 3.92109960e-01 -1.18367739e-01 5.60429633e-01
-5.26713669e-01 4.61388856e-01 -5.25680363e-01 -7.66708493e-01
-1.02829254e+00 6.68121636e-01 2.71850646e-01 3.05601180e-01
-6.47319615e-01 5.61754644e-01 -2.59092867e-01 -5.08790970e-01
2.68140465e-01 -8.96299660e-01 2.01856598e-01 -3.22018236e-01
-1.98939398e-01 4.87165540e-01 -1.06603637e-01 -1.85081080e-01
-2.53469031e-02 1.06459713e+00 1.85254410e-01 -1.79656923e-01
1.40174639e+00 2.70289660e-01 2.45505914e-01 3.66897255e-01
3.73212636e-01 -1.18269913e-01 -1.91428506e+00 4.63839293e-01
-4.54386681e-01 -5.08479059e-01 -1.28023073e-01 -7.42189884e-01
-5.59497237e-01 7.01636016e-01 2.48141795e-01 1.09824345e-01
7.30574191e-01 -2.89737493e-01 6.37198687e-01 1.08005524e+00
1.31572187e+00 -8.75530958e-01 7.78356418e-02 3.72079015e-01
1.28106821e+00 -7.63893604e-01 1.26779243e-01 -5.63389838e-01
-6.20635867e-01 1.34376884e+00 5.69218338e-01 -5.88824093e-01
3.87832761e-01 4.55522180e-01 -4.53191221e-01 -1.29209399e-01
-8.33522558e-01 -1.62663311e-01 1.78415284e-01 9.57555473e-01
-3.04921448e-01 -4.70355526e-02 1.71139631e-02 7.35581368e-02
-1.66723505e-01 9.41775516e-02 6.32212222e-01 1.68679476e+00
-8.21876705e-01 -9.50118363e-01 -2.50869393e-01 1.54637933e-01
1.65804133e-01 3.65845412e-01 -5.89970410e-01 1.19384873e+00
1.40875429e-01 7.26704776e-01 -6.17710985e-02 -4.55953419e-01
7.20643163e-01 -1.14064239e-01 1.03344560e+00 -7.67643869e-01
-2.60774344e-01 -3.45268190e-01 5.14330752e-02 -4.43484962e-01
-4.90049988e-01 -7.76696086e-01 -1.08208930e+00 2.02179447e-01
-4.52719778e-01 3.41215223e-01 8.69322419e-01 7.66777217e-01
4.63094115e-01 4.36051875e-01 5.00246048e-01 -1.66883159e+00
-1.00368321e+00 -7.26993680e-01 -4.27712649e-01 -3.36211145e-01
3.80704433e-01 -1.44923651e+00 -5.26009262e-01 1.65655985e-01] | [4.979844093322754, 0.5540348887443542] |
bfd08953-b624-44ae-b716-6a8d6d65d7b4 | large-scale-category-structure-aware-image | null | null | http://papers.nips.cc/paper/4347-large-scale-category-structure-aware-image-categorization | http://papers.nips.cc/paper/4347-large-scale-category-structure-aware-image-categorization.pdf | Large-Scale Category Structure Aware Image Categorization | Most previous research on image categorization has focused on medium-scale data sets, while large-scale image categorization with millions of images from thousands of categories remains a challenge. With the emergence of structured large-scale dataset such as the ImageNet, rich information about the conceptual relationships between images, such as a tree hierarchy among various image categories, become available. As human cognition of complex visual world benefits from underlying semantic relationships between object classes, we believe a machine learning system can and should leverage such information as well for better performance. In this paper, we employ such semantic relatedness among image categories for large-scale image categorization. Specifically, a category hierarchy is utilized to properly define loss function and select common set of features for related categories. An efficient optimization method based on proximal approximation and accelerated parallel gradient method is introduced. Experimental results on a subset of ImageNet containing 1.2 million images from 1000 categories demonstrate the effectiveness and promise of our proposed approach. | ['Bin Zhao', 'Fei Li', 'Eric P. Xing'] | 2011-12-01 | null | null | null | neurips-2011-12 | ['image-categorization'] | ['computer-vision'] | [ 1.52553514e-01 -2.86768466e-01 -2.03241587e-01 -7.97578812e-01
-3.67523998e-01 -6.68059528e-01 4.89351243e-01 5.69953740e-01
-4.76810485e-01 1.66129336e-01 2.78363854e-01 -1.34937137e-01
-3.12339425e-01 -7.89113402e-01 -4.74179357e-01 -3.60803068e-01
-1.25764519e-01 9.47149396e-02 3.60091269e-01 -4.42436226e-02
7.10146606e-01 9.00141671e-02 -1.92194617e+00 3.17442954e-01
7.48451531e-01 1.49626589e+00 5.25613070e-01 2.99776465e-01
-4.15642679e-01 8.65946054e-01 -3.49442065e-01 -4.04045135e-01
3.98253262e-01 -1.90652251e-01 -1.04857635e+00 3.81973982e-01
8.21948946e-01 -1.53766006e-01 -7.99554214e-02 1.43931353e+00
2.02072740e-01 1.18680589e-01 6.33084416e-01 -1.50921023e+00
-7.39122152e-01 5.10289431e-01 -7.38074362e-01 3.05640727e-01
-8.61416459e-02 -1.94598347e-01 1.10968328e+00 -7.34998345e-01
5.13764501e-01 1.53498733e+00 6.11133397e-01 1.69948906e-01
-1.02510965e+00 -7.19312429e-01 3.88700455e-01 4.50140953e-01
-1.54985416e+00 5.32993712e-02 6.94586992e-01 -7.19028652e-01
5.50052226e-01 1.02247536e-01 5.16070604e-01 3.91953319e-01
1.68287635e-01 4.40475225e-01 1.20783782e+00 -2.71017730e-01
1.63450703e-01 3.83287847e-01 7.65193284e-01 7.48680353e-01
2.32951760e-01 -4.23820496e-01 -1.86430171e-01 -1.98933735e-01
4.70872700e-01 5.06914258e-01 1.74857810e-01 -6.12225175e-01
-1.29114497e+00 9.84174550e-01 1.15013456e+00 2.31447756e-01
-3.49596232e-01 1.47365242e-01 6.08683109e-01 3.26721698e-01
2.66100198e-01 2.47348592e-01 -2.14736655e-01 4.18113470e-01
-5.07964134e-01 4.97551076e-02 4.63456005e-01 9.93077993e-01
1.15649569e+00 -4.84436691e-01 6.87371790e-02 1.26305771e+00
2.64432311e-01 2.35504419e-01 8.28686833e-01 -1.22059286e+00
3.48856062e-01 1.08294129e+00 -3.14721555e-01 -1.41474271e+00
-3.07866246e-01 -4.23215449e-01 -1.23377740e+00 -2.67773960e-02
1.88768372e-01 4.43009079e-01 -9.44522738e-01 1.60975409e+00
5.01638889e-01 1.43876880e-01 2.64971461e-02 9.80673194e-01
8.45469415e-01 3.42433095e-01 3.39775473e-01 1.90788746e-01
1.61044967e+00 -9.00498748e-01 -1.96206138e-01 -3.49996030e-01
3.38298917e-01 -6.27384841e-01 1.04975915e+00 -1.12035975e-01
-5.28323889e-01 -8.44824612e-01 -7.76522279e-01 -1.59427702e-01
-6.12185597e-01 -2.69644052e-01 1.01196194e+00 3.05644393e-01
-1.02571976e+00 3.21865767e-01 -2.96336204e-01 -8.17825258e-01
8.06909323e-01 3.36998701e-01 -3.57073724e-01 -3.42891663e-01
-9.60903466e-01 4.79237199e-01 5.69674909e-01 -2.81859398e-01
-6.69977486e-01 -7.53148139e-01 -7.46911287e-01 5.38834929e-02
1.34824291e-01 -7.56612420e-01 9.85414445e-01 -1.35447419e+00
-8.40368330e-01 1.22282529e+00 8.48093163e-03 -4.86054182e-01
1.04174726e-01 -2.14274265e-02 -1.47504583e-01 4.47819591e-01
5.82665443e-01 1.18456089e+00 9.31018889e-01 -1.34386694e+00
-1.24874306e+00 -5.77724338e-01 2.57420003e-01 2.90555358e-01
-8.17686200e-01 2.25575030e-01 -3.73785973e-01 -4.88939345e-01
3.50351006e-01 -9.56155539e-01 -4.02553171e-01 2.33763784e-01
-3.11733276e-01 -6.31991923e-01 7.10391343e-01 -1.98592499e-01
9.76353049e-01 -2.04189205e+00 4.85768454e-04 1.63389415e-01
5.90528607e-01 -2.05423057e-01 -2.28956118e-01 1.73519924e-01
3.29382196e-02 2.01589018e-01 -1.67298570e-01 9.87298638e-02
-1.48835048e-01 1.30540594e-01 -3.90623957e-01 2.75259525e-01
-8.82652104e-02 8.49236727e-01 -8.38783801e-01 -6.99513674e-01
1.72120139e-01 1.08499214e-01 -5.40824234e-01 2.37054199e-01
2.70974338e-01 3.03432699e-02 -7.61922002e-01 5.89247763e-01
6.12233639e-01 -7.04939604e-01 -4.48197722e-02 -4.68379736e-01
5.81856072e-02 -1.79043636e-01 -7.23229170e-01 1.62906754e+00
-4.93671954e-01 4.66963440e-01 -1.45413697e-01 -1.37028551e+00
7.81798780e-01 -1.57306343e-01 5.28399944e-01 -5.89681149e-01
1.69044152e-01 2.29461472e-02 1.68088943e-01 -3.09093386e-01
2.29266539e-01 7.57583603e-02 -3.05604964e-01 4.18516576e-01
-3.59679833e-02 -2.52629548e-01 3.00166577e-01 5.79891622e-01
7.73109436e-01 -4.35871392e-01 4.50643271e-01 -7.24867344e-01
5.91706753e-01 2.94605911e-01 3.75873536e-01 8.99126172e-01
-3.46768439e-01 4.09261405e-01 1.29074395e-01 -7.17431188e-01
-9.75043714e-01 -8.54178190e-01 -3.42599928e-01 1.22827983e+00
7.47363150e-01 -3.62284780e-01 -6.88118637e-01 -6.28980696e-01
3.50250095e-01 3.47015895e-02 -6.70737088e-01 -1.36955112e-01
-3.07960689e-01 -7.24804699e-01 3.50788459e-02 4.31522161e-01
8.74832630e-01 -9.86974061e-01 -5.15180647e-01 3.27455886e-02
-1.10140346e-01 -1.14258707e+00 -4.31252837e-01 -7.34399930e-02
-9.14846003e-01 -1.34950769e+00 -3.42139989e-01 -1.18070483e+00
8.20711792e-01 1.13002074e+00 1.30810714e+00 2.76609957e-01
-7.29842246e-01 5.73140562e-01 -4.08250630e-01 -2.85977960e-01
-1.93944067e-01 -5.91387078e-02 7.26562887e-02 3.02152097e-01
4.63028997e-01 -6.13666594e-01 -9.55488205e-01 4.79630768e-01
-7.38748312e-01 1.34137571e-01 7.23128378e-01 8.16745639e-01
6.01110458e-01 1.89771995e-01 6.73307836e-01 -8.55366170e-01
4.29367989e-01 -8.40226531e-01 -4.71642613e-01 3.22214186e-01
-6.97939754e-01 -1.24944486e-01 6.70640826e-01 -4.64716852e-01
-8.99816871e-01 -1.19174413e-01 3.80466074e-01 -2.83004582e-01
-1.51602939e-01 4.88873839e-01 -3.30427103e-02 -2.35689223e-01
4.64099914e-01 2.83378810e-01 -1.45425931e-01 -3.46098900e-01
7.32780695e-01 9.52285588e-01 5.26157200e-01 -4.96862710e-01
7.91367888e-01 6.56846404e-01 -2.21453533e-01 -8.00996304e-01
-1.18193042e+00 -1.02521718e+00 -8.53672683e-01 -4.97015640e-02
8.94700289e-01 -1.04683387e+00 -6.27802610e-01 4.20777768e-01
-8.85341048e-01 8.18904489e-02 3.13797742e-02 2.86512196e-01
-3.34570140e-01 4.51180160e-01 -5.64542353e-01 -3.32865007e-02
-4.41713929e-01 -9.35632348e-01 9.91438210e-01 3.99475485e-01
1.59526691e-02 -8.75027537e-01 -3.14130515e-01 6.60730004e-01
4.43217158e-01 -1.43338397e-01 1.11755383e+00 -5.00540137e-01
-7.99904287e-01 -2.54252762e-01 -9.04524148e-01 3.67144018e-01
4.52852756e-01 -4.75901395e-01 -6.50595844e-01 -4.05417919e-01
-1.20526768e-01 -7.86344051e-01 1.02318501e+00 2.70979136e-01
1.72463799e+00 -2.34079689e-01 -5.21148741e-01 6.24097645e-01
1.62669933e+00 2.77905469e-03 2.00133249e-01 4.25791651e-01
1.15083182e+00 7.71700263e-01 5.99985123e-01 3.34644079e-01
6.80588961e-01 3.79290462e-01 2.39850983e-01 -2.19935551e-02
-2.15012699e-01 -3.41803990e-02 -4.46957469e-01 1.05924273e+00
1.63912088e-01 1.08211137e-01 -8.78428876e-01 4.88832325e-01
-1.73205531e+00 -8.41310501e-01 1.59714729e-01 2.01423025e+00
7.70786107e-01 1.20205611e-01 -6.65126964e-02 -1.45762637e-01
1.10231447e+00 4.14555967e-02 -8.81430686e-01 1.09309621e-01
5.74366935e-02 -4.62465614e-01 3.63486767e-01 4.71111238e-02
-1.41118550e+00 9.53380406e-01 6.27728891e+00 9.56858754e-01
-9.58414733e-01 1.58889204e-01 8.54297042e-01 5.09666324e-01
4.21306305e-02 3.27104032e-02 -7.88559616e-01 4.00636941e-01
5.61434150e-01 -5.47715545e-01 1.82937950e-01 1.16603148e+00
-2.18159810e-01 1.67067181e-02 -8.67186606e-01 1.31051481e+00
1.20228395e-01 -1.21425200e+00 4.08914298e-01 7.45732784e-02
7.80600429e-01 1.30204245e-01 4.36608680e-02 1.46901608e-01
6.20758235e-01 -7.09700584e-01 7.75953293e-01 -4.75712121e-02
6.40626729e-01 -5.50161719e-01 5.13736546e-01 1.85595870e-01
-1.69726205e+00 -5.70190370e-01 -8.31499994e-01 -5.36607653e-02
-2.72537321e-01 4.36157793e-01 -4.24584568e-01 1.31496131e-01
9.92113292e-01 1.24680960e+00 -1.07749653e+00 1.07585192e+00
1.59074947e-01 4.21044528e-01 -1.37992620e-01 2.45804954e-02
3.25138092e-01 -2.51035720e-01 5.71755087e-03 1.05748546e+00
1.00834444e-02 3.59123558e-01 6.38967395e-01 4.31625605e-01
-2.49029338e-01 1.85540766e-01 -5.91743231e-01 7.70260245e-02
6.28003359e-01 1.73351359e+00 -1.26551068e+00 -5.74006677e-01
-6.90499485e-01 8.54621947e-01 6.34180248e-01 1.90488830e-01
-5.95799387e-01 -2.89238572e-01 8.13709795e-01 -3.91005240e-02
1.28487810e-01 -1.00072809e-01 -1.13736726e-01 -1.11964500e+00
-1.67161405e-01 -7.47381270e-01 7.48575628e-01 -6.99409544e-01
-1.76635194e+00 7.70999014e-01 8.92256498e-02 -1.26421881e+00
1.37758389e-01 -5.16318381e-01 -2.94781983e-01 4.35038954e-01
-1.51420045e+00 -1.45021975e+00 -6.99981034e-01 7.99814820e-01
8.77661288e-01 -2.82671034e-01 6.24818265e-01 1.75991878e-01
-3.09734553e-01 3.23898196e-01 3.97523582e-01 3.33510876e-01
5.45081198e-01 -1.22499502e+00 2.24171132e-01 4.81586576e-01
1.09056175e-01 7.19987154e-01 2.89730549e-01 -3.52017909e-01
-1.22304273e+00 -1.29018486e+00 3.81727934e-01 -3.31235111e-01
1.01628292e+00 -4.39374894e-01 -8.76740932e-01 3.02730411e-01
1.03268452e-01 3.03128064e-01 6.35114253e-01 3.78579758e-02
-9.32402432e-01 -5.16502440e-01 -8.89995396e-01 4.44306374e-01
1.36581635e+00 -6.63914800e-01 -6.70565307e-01 5.78868210e-01
7.66243517e-01 2.81377256e-01 -9.30693269e-01 1.84103757e-01
4.63233709e-01 -7.83581972e-01 1.46922314e+00 -6.83322072e-01
4.56641197e-01 -1.60546333e-01 -5.06796718e-01 -1.19911873e+00
-5.43166876e-01 -3.61992009e-02 4.76674438e-01 1.19882858e+00
-1.71381652e-01 -6.67350948e-01 4.88499850e-01 6.04586065e-01
6.35138005e-02 -4.04208124e-01 -5.96369147e-01 -7.39072442e-01
1.18749671e-01 -3.96334194e-02 3.74293357e-01 9.47578669e-01
-2.95196414e-01 5.78551173e-01 -4.20937389e-02 1.03093520e-01
1.14599299e+00 8.46569002e-01 5.63181043e-01 -1.58639610e+00
6.31019771e-02 -6.25515997e-01 -9.73495185e-01 -7.94214845e-01
4.08807039e-01 -1.11000192e+00 -3.04384027e-02 -1.47773242e+00
9.40296352e-01 -8.48104954e-01 -6.70446932e-01 4.39350694e-01
-4.52109158e-01 6.63416564e-01 3.48639965e-01 6.66861057e-01
-1.12017417e+00 3.42111319e-01 9.85467553e-01 -4.91780102e-01
3.37061822e-01 -2.85706013e-01 -9.50083017e-01 8.72883737e-01
6.66129708e-01 -5.64573109e-01 -6.94295168e-01 -2.94720888e-01
-1.52471691e-01 -3.82331520e-01 3.86181325e-01 -1.04999900e+00
3.27467740e-01 -2.23641887e-01 2.50370324e-01 -3.01065862e-01
4.39277366e-02 -7.63256371e-01 -7.89168477e-02 5.56056321e-01
-5.63795686e-01 -5.31230122e-02 -1.31653666e-01 8.52904201e-01
-5.89858890e-01 -3.94596718e-02 9.89082098e-01 -3.30151200e-01
-1.38698554e+00 5.53757727e-01 -3.96082178e-02 2.03637153e-01
1.06950891e+00 -1.62054986e-01 -3.85614365e-01 -1.04066044e-01
-3.19544315e-01 4.19914693e-01 4.23326790e-01 8.43660831e-01
6.98971868e-01 -1.37622654e+00 -6.28939033e-01 -8.95657390e-02
4.65934873e-01 -3.30336064e-01 1.44055381e-01 2.11563274e-01
-3.37073743e-01 3.43340248e-01 -3.93966585e-01 -8.87739778e-01
-1.55056441e+00 9.80973184e-01 1.21040627e-01 1.33786812e-01
-5.84133327e-01 9.63224232e-01 9.88346934e-01 -3.37540209e-01
3.05302963e-02 -8.21765810e-02 -4.80971336e-01 1.94350019e-01
7.61998653e-01 -1.30463401e-02 -2.59002417e-01 -8.26964021e-01
-4.18131053e-01 1.14564288e+00 -2.55051941e-01 4.33412999e-01
1.17423308e+00 -6.77343607e-01 -5.01586020e-01 3.66601795e-01
1.63225985e+00 -6.21942401e-01 -1.13683426e+00 -4.88324791e-01
2.60872215e-01 -7.30694473e-01 1.06776536e-01 -5.32720447e-01
-1.08459556e+00 8.93120468e-01 9.78203535e-01 3.21408778e-01
1.24047005e+00 3.07818770e-01 6.53520942e-01 6.69870913e-01
7.88826227e-01 -7.71095037e-01 3.94757718e-01 2.84398079e-01
7.09693491e-01 -1.83854401e+00 1.17136855e-02 -6.53637230e-01
-6.42958164e-01 8.97053957e-01 6.51800036e-01 -1.87637269e-01
9.07320857e-01 -3.40805829e-01 2.36182660e-01 -2.79243350e-01
-5.23257911e-01 -3.62755477e-01 3.35163414e-01 5.04983544e-01
2.09531978e-01 1.32355159e-02 -2.18971908e-01 3.60970318e-01
2.30375323e-02 -1.93832114e-01 2.10078597e-01 8.52771401e-01
-7.49529064e-01 -9.08542931e-01 -1.83036163e-01 7.39651799e-01
-3.18620950e-01 -1.99661985e-01 -9.44732875e-02 5.07225096e-01
1.15543805e-01 9.65465665e-01 3.29365373e-01 -3.19790035e-01
2.87667718e-02 -3.96816015e-01 2.39072859e-01 -7.00715899e-01
-4.20394123e-01 -3.53907645e-01 -3.94693077e-01 -6.75750434e-01
-5.92202127e-01 -3.42571169e-01 -1.21318924e+00 -1.94934085e-01
-1.38439327e-01 1.45158619e-01 6.64611161e-01 7.21008778e-01
4.74095643e-01 4.95252497e-02 9.70217109e-01 -7.58300006e-01
-5.07743716e-01 -7.49883890e-01 -6.50957286e-01 1.08319640e+00
4.47772816e-02 -7.45712698e-01 -3.54056478e-01 4.48354930e-01] | [9.756978034973145, 2.045459270477295] |
c4a077a3-6f59-4327-b362-d98591a3e6a3 | preserving-locality-in-vision-transformers | 2304.06971 | null | https://arxiv.org/abs/2304.06971v1 | https://arxiv.org/pdf/2304.06971v1.pdf | Preserving Locality in Vision Transformers for Class Incremental Learning | Learning new classes without forgetting is crucial for real-world applications for a classification model. Vision Transformers (ViT) recently achieve remarkable performance in Class Incremental Learning (CIL). Previous works mainly focus on block design and model expansion for ViTs. However, in this paper, we find that when the ViT is incrementally trained, the attention layers gradually lose concentration on local features. We call this interesting phenomenon as \emph{Locality Degradation} in ViTs for CIL. Since the low-level local information is crucial to the transferability of the representation, it is beneficial to preserve the locality in attention layers. In this paper, we encourage the model to preserve more local information as the training procedure goes on and devise a Locality-Preserved Attention (LPA) layer to emphasize the importance of local features. Specifically, we incorporate the local information directly into the vanilla attention and control the initial gradients of the vanilla attention by weighting it with a small initial value. Extensive experiments show that the representations facilitated by LPA capture more low-level general information which is easier to transfer to follow-up tasks. The improved model gets consistently better performance on CIFAR100 and ImageNet100. | ['De-Chuan Zhan', 'Han-Jia Ye', 'Da-Wei Zhou', 'Bowen Zheng'] | 2023-04-14 | null | null | null | null | ['class-incremental-learning'] | ['computer-vision'] | [ 8.27205628e-02 6.89724535e-02 -2.69803017e-01 -4.26208466e-01
-3.37177545e-01 -2.37868503e-01 5.86497605e-01 1.80670783e-01
-4.54064161e-01 6.72507226e-01 3.85564357e-01 -2.18275517e-01
1.19714707e-01 -7.72220910e-01 -1.05868173e+00 -7.71455884e-01
-1.78345609e-02 -8.85918215e-02 6.88497186e-01 -9.77369174e-02
1.91438913e-01 5.98814785e-01 -1.40604937e+00 7.32720554e-01
8.18477631e-01 7.77536094e-01 5.39974749e-01 4.76360589e-01
-4.64180633e-02 1.16602087e+00 -5.26581466e-01 -2.00774759e-01
3.34133714e-01 -4.68206823e-01 -9.54123080e-01 -4.45536748e-02
7.75375366e-01 -3.89374465e-01 -4.15104657e-01 9.45807040e-01
3.67981285e-01 3.44805270e-01 5.95545352e-01 -9.78502870e-01
-9.12991166e-01 6.11571670e-01 -8.93566966e-01 4.96444851e-01
-1.85729340e-01 3.21553499e-01 9.97316480e-01 -1.18944216e+00
5.77545702e-01 1.16471279e+00 7.10690558e-01 4.94188428e-01
-1.11746442e+00 -4.68937963e-01 7.09552407e-01 7.25473702e-01
-1.24199021e+00 -3.78766328e-01 9.43873107e-01 -1.44512177e-01
8.17559063e-01 2.35648490e-02 7.33022988e-01 8.50388646e-01
2.50370681e-01 1.16316545e+00 7.72648454e-01 -5.86926877e-01
2.38145649e-01 2.62050420e-01 2.74864644e-01 5.78062177e-01
1.77413419e-01 -8.72387066e-02 -5.58567941e-01 4.73192543e-01
8.25464368e-01 3.29313099e-01 -3.45734775e-01 -5.10537803e-01
-9.95009601e-01 7.07484543e-01 1.16646087e+00 4.71616954e-01
-5.17270744e-01 2.49235392e-01 3.06327492e-01 3.10897887e-01
4.67420995e-01 3.24054033e-01 -3.81739497e-01 2.02218711e-01
-8.46539199e-01 -2.76958644e-01 6.12131022e-02 8.15331757e-01
9.78594959e-01 1.03504390e-01 -5.06031215e-01 1.17271984e+00
9.15991813e-02 3.15340981e-02 5.93229234e-01 -6.90234601e-01
4.51513231e-01 6.70290649e-01 -3.28809381e-01 -7.06684887e-01
-1.64564520e-01 -9.34594989e-01 -1.05429220e+00 2.74702609e-01
2.86295742e-01 1.03418723e-01 -1.18191195e+00 1.71784365e+00
1.31889880e-01 3.12540591e-01 -4.58524227e-02 6.28110766e-01
5.70250213e-01 8.51136386e-01 2.27805272e-01 -7.18639046e-03
9.63663399e-01 -1.41385925e+00 -4.44200009e-01 -6.49579465e-01
5.92373967e-01 -5.02596617e-01 1.30430877e+00 1.21529244e-01
-1.00181282e+00 -1.06243706e+00 -1.10133290e+00 -3.18842053e-01
-3.39950681e-01 5.06716184e-02 3.19023520e-01 7.52035901e-02
-1.11494088e+00 6.75528884e-01 -7.24862516e-01 -3.21699828e-01
9.02049601e-01 2.72512406e-01 -2.64133751e-01 -5.50016642e-01
-1.06164062e+00 9.05870855e-01 3.09399188e-01 1.18657686e-01
-8.42664838e-01 -8.13226163e-01 -8.13250661e-01 3.02058697e-01
6.93680942e-02 -5.22402883e-01 1.06040061e+00 -1.21930814e+00
-1.28647530e+00 4.29456413e-01 -3.87187839e-01 -5.16912460e-01
2.75225639e-01 -1.28473341e-01 1.01235121e-01 1.79339349e-01
6.87780380e-02 1.06777620e+00 1.16636229e+00 -1.21504617e+00
-7.77123153e-01 -1.85825124e-01 2.28510410e-01 2.72171974e-01
-8.06534231e-01 -4.46855456e-01 -4.98459488e-01 -7.63985217e-01
6.15951829e-02 -7.41332412e-01 -9.01977569e-02 2.16846958e-01
6.66431934e-02 -4.18171197e-01 8.87923062e-01 -5.76806843e-01
1.20519948e+00 -2.38488317e+00 7.17123374e-02 -1.20152280e-01
1.27094045e-01 6.45504415e-01 -4.34523463e-01 1.57998383e-01
-2.59003431e-01 -1.49625048e-01 -1.40609920e-01 -3.04359436e-01
-3.25853795e-01 3.04309100e-01 -4.08878297e-01 1.76178753e-01
3.97728592e-01 1.21438634e+00 -8.98149848e-01 -2.14413658e-01
3.66226196e-01 5.13664126e-01 -8.72622132e-01 -8.91227275e-02
-2.63868384e-02 2.21048668e-01 -1.82752699e-01 3.22608739e-01
8.13810110e-01 -3.31426799e-01 -2.21670762e-01 -3.88944298e-01
-1.49272308e-01 1.86437324e-01 -6.99173987e-01 1.54038763e+00
-5.53994000e-01 8.78944218e-01 -2.13095382e-01 -1.15344489e+00
7.47219682e-01 -4.46807779e-02 5.43920919e-02 -9.12999988e-01
-1.16528533e-01 -9.38821733e-02 1.35596320e-01 -1.82557717e-01
3.63534361e-01 -8.35894197e-02 4.75957543e-01 1.21311821e-01
4.92042266e-02 1.95653468e-01 -6.97218180e-02 1.94521919e-01
8.47705722e-01 -7.78788701e-02 2.58735478e-01 -2.69372553e-01
5.82838178e-01 -2.09457606e-01 4.73914921e-01 8.61322820e-01
-3.73226255e-01 7.19743371e-01 2.55075693e-01 -5.26721179e-01
-9.55347180e-01 -8.82576346e-01 -4.88200709e-02 1.43852973e+00
6.24278560e-02 -2.67618954e-01 -5.45901716e-01 -9.13123071e-01
-6.85781166e-02 7.04398096e-01 -8.54098380e-01 -8.40844989e-01
-7.23167658e-01 -6.10275924e-01 1.23757452e-01 9.42385852e-01
9.54225719e-01 -1.18086278e+00 -3.68808568e-01 3.55051577e-01
-2.61620861e-02 -7.80315399e-01 -6.68481469e-01 3.57645959e-01
-1.04717207e+00 -7.22940326e-01 -1.08823335e+00 -1.19807196e+00
8.53618085e-01 6.31584525e-01 7.31774628e-01 1.49718344e-01
-8.02485868e-02 1.08127199e-01 -3.66740376e-01 -4.59896237e-01
8.57581198e-02 3.81801158e-01 -3.86522919e-01 1.84756815e-01
2.26800278e-01 -3.98500770e-01 -7.73508608e-01 1.75853893e-01
-8.59866738e-01 1.83602020e-01 8.19518864e-01 1.17558789e+00
4.60763365e-01 5.14972284e-02 5.31138480e-01 -7.21553922e-01
3.38178843e-01 -1.88408002e-01 -1.58295706e-01 2.62183756e-01
-3.92671138e-01 2.51027972e-01 8.10065866e-01 -7.06775010e-01
-1.16855085e+00 -5.75045310e-02 -2.47481287e-01 -5.16950905e-01
1.50929660e-01 3.46342862e-01 -9.80227813e-02 -2.27936402e-01
5.48045218e-01 3.56946319e-01 -3.30466837e-01 -4.75400537e-01
4.58509833e-01 3.54252368e-01 3.70980412e-01 -3.89727652e-01
6.06202841e-01 5.04903674e-01 -3.88276935e-01 -8.68037224e-01
-1.10728264e+00 -4.65484887e-01 -8.14645827e-01 -1.00432374e-01
5.57641268e-01 -7.98024595e-01 -2.79784143e-01 5.88829339e-01
-9.94394958e-01 -7.32160866e-01 -8.50571394e-01 3.90373409e-01
-2.22933009e-01 1.48710087e-01 -5.77580869e-01 -3.62331659e-01
-1.16012238e-01 -1.05203748e+00 7.59293318e-01 3.39203686e-01
1.00240394e-01 -1.02770126e+00 1.15354545e-02 -4.60746922e-02
6.45491540e-01 -3.74940455e-01 9.63155389e-01 -1.51837334e-01
-5.77561378e-01 1.62889212e-02 -5.40831983e-01 6.11875415e-01
1.85471386e-01 -2.92296737e-01 -1.24403036e+00 -5.14130712e-01
-3.84571180e-02 -2.75502414e-01 1.57938802e+00 5.30583799e-01
1.42013288e+00 -4.28293675e-01 -3.33456010e-01 7.96656787e-01
1.30368447e+00 2.35052139e-01 8.46136749e-01 4.31577176e-01
9.17300522e-01 3.94058198e-01 3.22248638e-01 8.63373727e-02
3.94010007e-01 4.74446893e-01 3.07647645e-01 -3.56359303e-01
-7.66418159e-01 -4.06966597e-01 3.68680179e-01 8.09016645e-01
-5.84958214e-03 7.63560608e-02 -6.87348187e-01 9.11616981e-01
-1.80401719e+00 -9.15884793e-01 2.72447646e-01 2.19149232e+00
1.00542259e+00 3.42932373e-01 -3.72185886e-01 6.05053976e-02
6.43274844e-01 3.19718003e-01 -6.89279437e-01 -3.41343552e-01
-6.98582828e-02 2.62825072e-01 3.48631471e-01 6.94070876e-01
-9.55440521e-01 1.23219597e+00 5.82982969e+00 9.31192279e-01
-1.32635939e+00 6.53990582e-02 9.62086260e-01 -8.87427032e-02
-1.70035347e-01 -2.52341200e-02 -8.86029363e-01 3.03432733e-01
6.05093598e-01 1.22491689e-02 1.52915984e-01 9.18739557e-01
6.67181462e-02 -5.59453107e-02 -1.07714641e+00 7.76929617e-01
8.63465220e-02 -1.30520117e+00 5.08846879e-01 -2.46186778e-02
1.05776167e+00 1.51589483e-01 3.40358377e-01 6.56407952e-01
6.56975955e-02 -7.67632425e-01 6.59836888e-01 5.40363729e-01
6.72904611e-01 -7.09066451e-01 7.20265090e-01 2.39860252e-01
-1.22025323e+00 -2.57020324e-01 -8.01309347e-01 -1.61969319e-01
-1.78521767e-01 4.39016283e-01 -7.91192770e-01 9.69391540e-02
6.94469273e-01 1.01803184e+00 -9.37018514e-01 1.28608048e+00
-4.60422873e-01 7.21589267e-01 -1.28087267e-01 1.46051720e-01
5.00759840e-01 1.31781086e-01 1.69027045e-01 1.07817984e+00
6.33764938e-02 3.60527597e-02 -1.46690547e-01 5.01122057e-01
-4.02888894e-01 -6.46801889e-02 -5.56312323e-01 4.09693182e-01
1.03897005e-01 8.63609672e-01 -6.90649629e-01 -5.28129160e-01
-4.90842134e-01 1.14837337e+00 6.72232330e-01 5.75589836e-01
-6.13216758e-01 -3.63380373e-01 4.50251311e-01 1.80869430e-01
7.74336934e-01 -4.14694361e-02 -4.12458867e-01 -1.10760999e+00
2.07835957e-02 -5.94883263e-01 3.83536071e-01 -8.15610170e-01
-1.08455098e+00 6.59283042e-01 -1.22733220e-01 -1.00488532e+00
-5.60491951e-03 -4.97883528e-01 -8.95871222e-01 7.13254631e-01
-1.87579143e+00 -1.20113707e+00 -2.60161042e-01 7.49368668e-01
8.62817764e-01 7.23335566e-03 3.13614786e-01 3.13741565e-01
-5.87214053e-01 8.95459712e-01 2.26243839e-01 3.77059095e-02
8.11594725e-01 -9.91043627e-01 3.59487385e-01 1.00921428e+00
2.95188397e-01 7.98268437e-01 3.07820439e-01 -3.72079492e-01
-7.40349889e-01 -1.34454155e+00 9.32429731e-01 -1.28443927e-01
4.19425368e-01 -3.68354231e-01 -1.37653005e+00 8.53531837e-01
3.24199289e-01 3.39650780e-01 1.12638690e-01 4.42587622e-02
-5.43267548e-01 -5.11812449e-01 -8.27112556e-01 6.25796914e-01
9.49311316e-01 -6.49453402e-01 -6.77127600e-01 2.67596215e-01
7.41971612e-01 9.41548347e-02 -2.54273981e-01 3.10564339e-01
4.33304816e-01 -9.02584314e-01 1.02212560e+00 -4.99027699e-01
1.56121671e-01 -2.99895942e-01 1.84414998e-01 -1.49660575e+00
-8.54237974e-01 -1.79196417e-01 -5.44076525e-02 1.09144318e+00
3.85150850e-01 -4.65958774e-01 8.45006585e-01 3.03135484e-01
-2.50528663e-01 -6.46738827e-01 -7.43932188e-01 -6.70725644e-01
3.43874544e-01 -2.68866003e-01 3.91353488e-01 8.84924591e-01
-6.79361373e-02 5.87905228e-01 -3.11839163e-01 7.59894177e-02
4.55206245e-01 -3.87481973e-02 4.51731950e-01 -1.02940798e+00
-1.58844396e-01 -5.24313509e-01 -3.34335953e-01 -1.52130377e+00
-1.62609860e-01 -8.12599361e-01 1.53557267e-02 -1.65470457e+00
3.30549836e-01 -3.83206278e-01 -7.09845364e-01 8.76666903e-01
-4.89622891e-01 4.54890639e-01 4.05647337e-01 2.93776155e-01
-7.16051221e-01 8.44511330e-01 1.48469031e+00 -4.88378227e-01
-2.81716853e-01 4.79245111e-02 -7.34845996e-01 6.55776858e-01
7.92591214e-01 -3.66358429e-01 -6.73649549e-01 -7.59805679e-01
-2.28314325e-01 -6.72615945e-01 3.74402314e-01 -1.15518701e+00
5.31231523e-01 -3.29608172e-02 6.30160093e-01 -5.85241199e-01
2.08443925e-01 -6.85951054e-01 -4.97660518e-01 6.08177960e-01
-5.04168808e-01 -2.55783170e-01 3.87113154e-01 4.50143278e-01
-3.41529012e-01 -3.39478880e-01 1.13939798e+00 -1.03489682e-01
-1.03173196e+00 4.55086648e-01 -2.41069660e-01 -7.07052425e-02
8.75986457e-01 -2.51068175e-01 -2.83019960e-01 -3.27274501e-01
-7.63526380e-01 1.97398245e-01 3.20640504e-01 4.81231511e-01
8.05885553e-01 -1.30046821e+00 -7.40923166e-01 5.25554955e-01
2.51008362e-01 1.37470499e-01 3.58477712e-01 6.20057821e-01
-3.70829940e-01 4.15783495e-01 -4.16890919e-01 -5.94730258e-01
-1.02249670e+00 7.32085824e-01 3.39349002e-01 -1.95304841e-01
-7.75718927e-01 1.28227508e+00 7.98984349e-01 4.57175560e-02
4.67521846e-01 -6.30559385e-01 -3.24946344e-01 1.28429964e-01
8.36997271e-01 8.56295079e-02 7.52474144e-02 -4.06405658e-01
-2.95967013e-01 5.98801672e-01 -6.92305386e-01 1.58276394e-01
1.58395183e+00 -2.24849358e-01 7.88555443e-02 4.36496079e-01
1.38446367e+00 -1.81801066e-01 -1.87173510e+00 -5.48056543e-01
-1.88132063e-01 -3.73925388e-01 2.79446214e-01 -6.17308557e-01
-1.34393203e+00 1.23742604e+00 6.62688017e-01 -1.66121483e-01
1.18087208e+00 6.83080480e-02 5.80604315e-01 3.34008962e-01
1.47622854e-01 -9.25176084e-01 4.23973411e-01 7.27638245e-01
1.04120684e+00 -1.13647461e+00 -8.18500221e-02 -1.40078634e-01
-6.03604138e-01 9.12943423e-01 8.30800116e-01 -2.32240185e-01
7.00188339e-01 -5.01176976e-02 -1.72800481e-01 7.72240311e-02
-7.75004506e-01 -2.57149547e-01 3.29778790e-01 7.55929351e-01
2.06425190e-01 -3.82639617e-01 8.66439492e-02 2.25148827e-01
1.05652712e-01 -1.14212610e-01 3.91573220e-01 8.53537261e-01
-7.10866690e-01 -8.46653402e-01 -1.04939528e-01 3.20879042e-01
-4.53923754e-02 -4.47883457e-01 -2.45567903e-01 6.31207764e-01
3.24707538e-01 5.86217880e-01 9.14980397e-02 -1.70779526e-01
4.28196758e-01 6.07627705e-02 4.45218265e-01 -6.69608414e-01
-3.68674666e-01 -7.03966618e-02 -4.60648596e-01 -3.15724641e-01
-2.43897215e-01 -5.20324469e-01 -1.02189100e+00 -1.63318947e-01
-3.89745623e-01 1.57895945e-02 3.06355536e-01 8.48945439e-01
2.81720191e-01 9.17459726e-01 6.13652527e-01 -7.22941816e-01
-3.08663875e-01 -1.00761020e+00 -3.56115907e-01 3.38425696e-01
6.04263902e-01 -5.65305829e-01 -2.64877439e-01 1.64098412e-01] | [9.601052284240723, 1.7705763578414917] |
f5fb1371-9bd4-4249-8cd5-c0c38e58f280 | semi-blind-source-separation-via-sparse | 1212.0451 | null | http://arxiv.org/abs/1212.0451v2 | http://arxiv.org/pdf/1212.0451v2.pdf | Semi-blind Source Separation via Sparse Representations and Online Dictionary Learning | This work examines a semi-blind single-channel source separation problem. Our
specific aim is to separate one source whose local structure is approximately
known, from another a priori unspecified background source, given only a single
linear combination of the two sources. We propose a separation technique based
on local sparse approximations along the lines of recent efforts in sparse
representations and dictionary learning. A key feature of our procedure is the
online learning of dictionaries (using only the data itself) to sparsely model
the background source, which facilitates its separation from the
partially-known source. Our approach is applicable to source separation
problems in various application domains; here, we demonstrate the performance
of our proposed approach via simulation on a stylized audio source separation
task. | ['Sirisha Rambhatla', 'Jarvis D. Haupt'] | 2012-12-03 | null | null | null | null | ['audio-source-separation'] | ['audio'] | [ 4.36029583e-01 -3.20326507e-01 -1.09315254e-01 5.43693863e-02
-1.32245302e+00 -6.37046218e-01 3.78437042e-01 1.12375900e-01
4.53301892e-02 6.29148543e-01 4.38335329e-01 5.19334599e-02
-1.96151823e-01 -2.02701092e-01 -4.61909533e-01 -9.96800005e-01
-9.30113271e-02 2.90896177e-01 -2.32906967e-01 -1.15979195e-01
5.76786622e-02 5.19911826e-01 -1.46299529e+00 -4.46026437e-02
7.36399293e-01 9.11707878e-01 2.57284164e-01 9.46551144e-01
-1.20398559e-01 7.31980979e-01 -7.54650056e-01 1.30489975e-01
4.39936876e-01 -7.68798232e-01 -4.35064614e-01 5.21931171e-01
4.48953778e-01 -2.45934911e-02 -4.44867164e-01 1.21556044e+00
6.56174481e-01 3.26947898e-01 6.02896333e-01 -9.62250710e-01
-3.84006083e-01 7.09208310e-01 -7.05559552e-01 6.65075064e-01
3.17546546e-01 -4.98621643e-01 7.78992951e-01 -8.44338119e-01
-2.73941215e-02 1.04122102e+00 5.71058571e-01 3.24384779e-01
-1.18651950e+00 -6.61421597e-01 1.36113763e-01 -4.61721607e-02
-1.70567214e+00 -1.26595473e+00 1.26674283e+00 -3.09636742e-01
4.56979811e-01 5.60557246e-02 2.22775742e-01 8.42613697e-01
-2.84136921e-01 7.49869585e-01 1.27381551e+00 -5.85826874e-01
4.56875771e-01 3.07909518e-01 2.19809607e-01 5.23723602e-01
2.94359386e-01 6.89234287e-02 -9.18401957e-01 -6.10818982e-01
7.80323625e-01 -1.26261577e-01 -6.69367969e-01 -2.46296749e-01
-8.51526082e-01 6.43701375e-01 -9.39972475e-02 5.79360902e-01
-5.01469612e-01 -8.52145627e-02 -1.28430650e-01 2.14797407e-01
3.91412824e-01 3.20111625e-02 -2.11270615e-01 -1.55599162e-01
-1.41976047e+00 -7.11636320e-02 1.07489121e+00 9.52890337e-01
7.28700161e-01 8.52687001e-01 2.17838824e-01 1.00776112e+00
4.06091064e-01 9.63513434e-01 6.40915036e-01 -6.44250512e-01
4.03806984e-01 -1.54571250e-01 4.32400048e-01 -1.10693979e+00
1.43259376e-01 -8.74025106e-01 -8.68093729e-01 -9.90970507e-02
5.21124721e-01 -4.60112393e-01 -7.28119016e-01 1.63479829e+00
3.47513348e-01 1.03983533e+00 3.44294995e-01 9.63811517e-01
6.33851111e-01 9.14167345e-01 -3.64999056e-01 -6.32767677e-01
9.48771477e-01 -7.70567715e-01 -8.56825769e-01 -3.64362150e-01
-2.65877098e-01 -1.02458477e+00 3.48028272e-01 7.00955331e-01
-1.09294415e+00 -5.76044917e-01 -1.02416801e+00 3.64637673e-01
9.63548869e-02 4.64328140e-01 4.71149236e-01 9.00193930e-01
-7.67196536e-01 2.09752470e-01 -5.66505969e-01 -1.13097243e-01
1.93166018e-01 2.27500871e-01 -2.39683613e-01 -3.71481758e-03
-8.79441738e-01 3.37311268e-01 -2.19371933e-02 2.37251326e-01
-1.34142900e+00 -5.38381398e-01 -9.03759539e-01 2.48286530e-01
2.89159656e-01 -3.33519280e-01 1.00289190e+00 -1.36909330e+00
-1.63400722e+00 3.75985682e-01 -9.11022305e-01 -4.73300308e-01
-1.30791038e-01 -3.21188033e-01 -7.62339950e-01 3.10550958e-01
1.54499963e-01 -3.26852500e-01 1.79123032e+00 -1.69969904e+00
-5.83221853e-01 -2.41192505e-01 -2.41164178e-01 3.21736932e-01
-3.26131284e-01 9.17274505e-02 -3.40799689e-01 -1.07797980e+00
2.27503240e-01 -6.89567864e-01 -1.50882706e-01 -2.97564626e-01
-2.13977292e-01 4.71193165e-01 6.36202812e-01 -9.96251702e-01
1.09703684e+00 -2.50685000e+00 4.02931869e-01 5.20450830e-01
1.27814114e-01 1.84326455e-01 -1.79171711e-01 3.13163698e-01
-2.55612075e-01 -5.52531242e-01 -5.81048191e-01 -5.56289673e-01
-3.36754322e-01 -7.39783272e-02 -7.44820714e-01 8.05886269e-01
-2.02529579e-01 1.05189018e-01 -8.76580000e-01 -3.40095997e-01
-1.13391325e-01 6.38550401e-01 -2.15779722e-01 4.38457996e-01
3.75396997e-01 4.88302320e-01 -2.29182422e-01 8.12665224e-01
8.20886254e-01 4.97236103e-02 8.04853067e-02 -8.96597430e-02
2.80667003e-02 2.19689519e-03 -1.98257709e+00 1.75526762e+00
-7.16096103e-01 6.11238956e-01 9.29885447e-01 -1.23906064e+00
8.89378548e-01 7.38510132e-01 6.18413329e-01 -1.40801683e-01
2.00524867e-01 2.05159515e-01 -9.72247869e-02 -1.20187923e-01
1.83848590e-01 -4.81192291e-01 3.02558839e-01 5.15127838e-01
4.99236554e-01 9.49157923e-02 4.03308794e-02 3.20405930e-01
8.23646069e-01 -3.52860838e-01 6.54776454e-01 -2.23014116e-01
7.16819882e-01 -4.00095791e-01 4.77296174e-01 8.80018890e-01
-1.99506909e-01 6.86852992e-01 -1.08111702e-01 2.22089171e-01
-4.94432807e-01 -1.09753847e+00 1.29212672e-02 7.52146125e-01
5.82175069e-02 -3.16869050e-01 -4.48058724e-01 -4.61072832e-01
-2.81215366e-02 4.07310098e-01 -3.75875384e-01 1.06787898e-01
-5.11081040e-01 -7.19868302e-01 5.04539549e-01 2.44389966e-01
3.82449031e-02 -4.59064484e-01 -2.47617409e-01 4.06402767e-01
-2.65971482e-01 -9.74045813e-01 -6.27071202e-01 3.52572560e-01
-8.08154225e-01 -9.67298150e-01 -1.03678095e+00 -7.78229117e-01
7.05278635e-01 8.48397255e-01 6.55277133e-01 -2.46941179e-01
-1.06129095e-01 8.78437579e-01 -1.83061332e-01 -4.86475348e-01
-3.62839311e-01 -5.32715321e-01 4.48106915e-01 9.33906853e-01
7.21253604e-02 -8.00242543e-01 -8.79462510e-02 9.03215036e-02
-7.15449452e-01 -5.91791391e-01 4.56973523e-01 7.37645030e-01
6.11492336e-01 8.59066248e-01 6.85729325e-01 -6.70015514e-01
7.24629104e-01 -8.67589772e-01 -4.77976501e-01 8.56310651e-02
-1.70676082e-01 -1.89146921e-01 7.63637722e-01 -7.24421561e-01
-1.44481552e+00 2.93240160e-01 1.19222321e-01 -7.15199113e-01
-4.03874010e-01 3.75818580e-01 -4.27661031e-01 -4.00615394e-01
5.15786171e-01 7.21950531e-01 -2.94241697e-01 -8.75130534e-01
3.57969791e-01 7.74108648e-01 6.81220591e-01 -6.92046702e-01
1.16661632e+00 6.88711643e-01 -1.61158353e-01 -1.35150456e+00
-6.85972214e-01 -9.98834670e-01 -6.95033848e-01 5.33199981e-02
7.47060999e-02 -1.32236898e+00 -1.76001698e-01 4.26112682e-01
-9.23819065e-01 8.25454965e-02 -4.65449035e-01 5.87888658e-01
-3.48502189e-01 6.35567665e-01 -2.57662386e-01 -1.37451589e+00
-5.55543117e-02 -6.62222981e-01 1.16072333e+00 2.36602575e-01
-9.82492045e-02 -1.22719419e+00 3.88929546e-01 2.00479865e-01
2.87204832e-01 -1.52368650e-01 4.90102679e-01 -7.79798031e-01
-2.29491100e-01 -2.93639749e-01 2.19132707e-01 5.85950792e-01
7.92673230e-01 -5.63061476e-01 -1.16681838e+00 -6.11900389e-01
5.89155674e-01 8.36485550e-02 7.30460465e-01 4.42535698e-01
5.36238313e-01 -4.08041269e-01 -2.46583417e-01 5.68692565e-01
1.42320824e+00 2.20202446e-01 2.34351121e-02 -3.38375717e-01
8.63279223e-01 3.03338736e-01 2.63305008e-01 8.53732824e-01
-3.93117853e-02 5.39147019e-01 -6.46541491e-02 -1.28334999e-01
-3.61303180e-01 -1.39342695e-01 4.84876275e-01 1.05794895e+00
5.46196923e-02 -1.02134697e-01 -5.53638160e-01 9.73033071e-01
-1.48373353e+00 -1.04042935e+00 1.62281230e-01 2.20051742e+00
7.46351421e-01 -3.17443639e-01 2.43077919e-01 6.35374665e-01
7.31965363e-01 2.63685942e-01 -4.69735473e-01 1.10703431e-01
-3.13991815e-01 5.67982733e-01 2.96459556e-01 8.28298926e-01
-1.22090483e+00 5.36857426e-01 6.70688438e+00 1.00300741e+00
-1.07239592e+00 1.82326406e-01 1.06669880e-01 -1.10717751e-01
-1.05659023e-01 2.02953760e-02 -6.26968384e-01 4.46020395e-01
9.76898611e-01 -3.75279278e-01 7.12118387e-01 6.11379385e-01
1.84577778e-01 -1.70687392e-01 -9.89807546e-01 1.35852075e+00
6.54527485e-01 -1.00507891e+00 1.08823851e-01 -2.30254695e-01
6.85745835e-01 -4.85966027e-01 1.51730870e-04 -1.46568909e-01
2.31426090e-01 -7.75799513e-01 7.47974396e-01 3.99954706e-01
4.70960140e-01 -7.39765882e-01 3.35255206e-01 4.89363194e-01
-1.38051891e+00 -1.96572796e-01 -1.75385728e-01 -4.40398194e-02
1.19215913e-01 8.24953854e-01 -6.06060207e-01 6.29822433e-01
4.72342402e-01 7.84115851e-01 -1.44561484e-01 1.17409754e+00
-1.44679457e-01 1.20321047e+00 -4.20879483e-01 4.86327291e-01
-1.28550276e-01 -2.38378271e-01 1.10013998e+00 1.42258739e+00
4.20720041e-01 3.24238688e-01 2.52164841e-01 5.22823036e-01
1.31554574e-01 2.95190275e-01 -5.63378572e-01 -3.56782638e-02
4.49207723e-01 1.09714830e+00 -7.07533002e-01 -3.57389450e-01
-4.85972732e-01 9.29251611e-01 -2.94573121e-02 7.12927878e-01
-4.67511982e-01 -4.20580417e-01 6.38608992e-01 -7.32315406e-02
5.89534819e-01 -3.88073742e-01 -1.03511170e-01 -1.53866684e+00
-2.59414971e-01 -1.30961275e+00 2.76646852e-01 -4.48283225e-01
-1.31402075e+00 5.16805530e-01 -1.01630710e-01 -1.30229867e+00
-9.69896838e-02 -2.11437121e-01 -6.31087005e-01 1.18733633e+00
-1.57058072e+00 -8.61976504e-01 2.40413751e-02 1.16294694e+00
6.60346448e-01 -6.76369309e-01 8.62383127e-01 3.93163949e-01
-4.77734238e-01 4.02443260e-01 5.50638795e-01 6.09531477e-02
6.15006924e-01 -1.14211607e+00 -2.01977253e-01 1.29629016e+00
8.67200196e-01 8.53289604e-01 9.77810442e-01 -4.80225950e-01
-1.47290409e+00 -7.69093812e-01 5.13947248e-01 -1.47503942e-01
7.49138296e-01 -3.33822608e-01 -8.54167223e-01 4.91042167e-01
1.80576563e-01 6.53613359e-02 1.18291354e+00 6.13863915e-02
-4.03165162e-01 -2.34077916e-01 -1.03738201e+00 6.79672956e-02
4.63386744e-01 -7.02905834e-01 -7.92231381e-01 1.63705572e-01
7.52269998e-02 -2.89239377e-01 -4.65576112e-01 -2.09100142e-01
3.16008270e-01 -6.33277297e-01 1.17867088e+00 -4.11854237e-01
-1.28753141e-01 -5.08728027e-01 -4.62349236e-01 -1.48394012e+00
-3.75912309e-01 -9.89558637e-01 -5.82176089e-01 1.51108885e+00
-2.02807859e-02 -4.30298269e-01 6.39630318e-01 9.98134390e-02
3.14814538e-01 -6.97704777e-02 -1.02609932e+00 -8.12927961e-01
-2.94164807e-01 -3.70939553e-01 3.40918094e-01 9.20462251e-01
-5.15851974e-02 3.09052169e-01 -9.22288239e-01 8.03412437e-01
1.17323160e+00 4.44447905e-01 6.60768688e-01 -1.12295711e+00
-9.60391283e-01 8.45170245e-02 -9.17600319e-02 -1.25585222e+00
4.43087488e-01 -6.85735762e-01 2.05602899e-01 -1.06670880e+00
-1.32959792e-02 -3.58543634e-01 -5.32359123e-01 1.44430369e-01
-1.50527537e-01 3.39878686e-02 1.96021006e-01 3.00399303e-01
-2.07174331e-01 5.18179178e-01 6.24661863e-01 -2.79532254e-01
-3.38031769e-01 4.59745139e-01 -1.13746309e+00 8.45055580e-01
6.07267976e-01 -7.40468800e-01 -7.26427555e-01 -4.49002028e-01
-4.94195879e-01 3.43364656e-01 2.75045305e-01 -1.19972086e+00
3.63594323e-01 -1.14497572e-01 2.43049309e-01 -1.05091959e-01
5.62226295e-01 -1.08744335e+00 4.96707633e-02 -6.37691244e-02
-1.37606218e-01 -5.84774911e-01 3.51426423e-01 1.04548490e+00
-5.13017535e-01 -4.56031173e-01 7.19167113e-01 -9.87239182e-03
-5.19606590e-01 1.23120978e-01 -4.19397026e-01 7.37056583e-02
5.69149733e-01 -1.09649878e-02 3.43594313e-01 -7.51776218e-01
-1.01557028e+00 -3.67163122e-01 -8.91910866e-02 6.28553182e-02
8.04502130e-01 -1.12994099e+00 -7.52440333e-01 4.27784741e-01
-2.24694490e-01 -4.44468945e-01 2.26157263e-01 4.33594465e-01
1.14294365e-01 3.85786355e-01 -3.39877643e-02 -1.94735959e-01
-1.32238948e+00 6.40970588e-01 3.37246031e-01 1.29312545e-01
-4.74988282e-01 9.38825428e-01 2.79580206e-01 2.07113236e-01
5.13067424e-01 2.30798088e-02 -2.74361283e-01 4.46897000e-02
8.68765116e-01 7.02726841e-01 -1.06415354e-01 -1.20595515e+00
-4.05332148e-01 6.92750812e-01 4.00338560e-01 -4.35926914e-01
1.19297194e+00 -3.95592809e-01 -1.53490171e-01 7.74511456e-01
1.12525904e+00 9.61134136e-01 -1.01827049e+00 -7.15682685e-01
-3.01172674e-01 -7.41931021e-01 3.16075534e-01 -5.84002495e-01
-1.06707811e+00 8.57694030e-01 6.61028266e-01 1.10556528e-01
1.43243825e+00 -1.07608959e-01 5.49421489e-01 3.05317700e-01
3.58703345e-01 -6.52844012e-01 -1.93410277e-01 1.67001829e-01
8.49138677e-01 -1.13890409e+00 -1.58201773e-02 -4.11523044e-01
-4.56228405e-01 1.00555873e+00 4.56773005e-02 -2.94433564e-01
1.12433338e+00 5.12833655e-01 1.35975882e-01 1.49933800e-01
-2.93866813e-01 -3.87497872e-01 5.26106060e-01 8.95742059e-01
2.29829431e-01 -7.35241175e-02 4.56600189e-01 9.79949653e-01
4.54661176e-02 -2.04133883e-01 5.52429318e-01 9.04936075e-01
-5.42471886e-01 -9.26451802e-01 -1.06926811e+00 8.43128562e-02
-6.86372399e-01 -3.14913362e-01 -3.12254965e-01 1.32594228e-01
8.41080472e-02 1.42013001e+00 -3.58053952e-01 1.09530039e-01
1.90331727e-01 8.35173391e-03 3.84203821e-01 -7.58904338e-01
-1.83931142e-01 8.00749898e-01 -1.86599523e-01 -1.81673005e-01
-7.63675511e-01 -9.20849562e-01 -9.00611699e-01 1.54380664e-01
-2.48946890e-01 6.46717370e-01 4.89622593e-01 9.26421404e-01
1.01067245e-01 3.77928525e-01 7.75844753e-01 -9.92659092e-01
-4.00654107e-01 -6.07533276e-01 -1.16345501e+00 1.45855442e-01
9.95406926e-01 -6.27537608e-01 -6.27499461e-01 5.50554454e-01] | [15.229649543762207, 5.73895788192749] |
ab369f46-86af-44ab-81ec-c72b7acd3dc0 | graphrel-modeling-text-as-relational-graphs | null | null | https://aclanthology.org/P19-1136 | https://aclanthology.org/P19-1136.pdf | GraphRel: Modeling Text as Relational Graphs for Joint Entity and Relation Extraction | In this paper, we present GraphRel, an end-to-end relation extraction model which uses graph convolutional networks (GCNs) to jointly learn named entities and relations. In contrast to previous baselines, we consider the interaction between named entities and relations via a 2nd-phase relation-weighted GCN to better extract relations. Linear and dependency structures are both used to extract both sequential and regional features of the text, and a complete word graph is further utilized to extract implicit features among all word pairs of the text. With the graph-based approach, the prediction for overlapping relations is substantially improved over previous sequential approaches. We evaluate GraphRel on two public datasets: NYT and WebNLG. Results show that GraphRel maintains high precision while increasing recall substantially. Also, GraphRel outperforms previous work by 3.2{\%} and 5.8{\%} (F1 score), achieving a new state-of-the-art for relation extraction. | ['Wei-Yun Ma', 'Peng-Hsuan Li', 'Tsu-Jui Fu'] | 2019-07-01 | null | null | null | acl-2019-7 | ['joint-entity-and-relation-extraction'] | ['natural-language-processing'] | [-6.51533976e-02 9.17534888e-01 -5.71747899e-01 -4.39279079e-01
-6.41824186e-01 -5.61273932e-01 7.08902478e-01 7.54096270e-01
-4.41172987e-01 8.02955806e-01 3.03671837e-01 -3.46725464e-01
-1.75042808e-01 -1.15380204e+00 -4.55631882e-01 1.66572090e-02
-4.63100195e-01 7.20549285e-01 3.38059276e-01 -3.24056029e-01
-2.04061925e-01 5.15945196e-01 -7.48693585e-01 3.34314555e-01
8.52344453e-01 8.27465177e-01 -2.43074581e-01 5.21066487e-01
-4.08617228e-01 1.14202595e+00 -3.94121677e-01 -8.78770351e-01
2.27713101e-02 -2.30043724e-01 -1.40635955e+00 -2.77344376e-01
2.99499661e-01 -1.12165377e-01 -7.99444735e-01 7.89704740e-01
1.66440576e-01 1.60761178e-01 4.91627187e-01 -9.95709538e-01
-6.35398924e-01 1.36542988e+00 -6.98122561e-01 1.44850180e-01
4.05442327e-01 -4.83140558e-01 1.90797400e+00 -1.00184059e+00
9.24945831e-01 1.13571739e+00 5.63123405e-01 1.30812004e-02
-1.19036651e+00 -6.43958092e-01 3.48218381e-01 3.55663449e-02
-1.50941288e+00 -2.38538578e-01 4.02457923e-01 -1.41033322e-01
1.92774415e+00 9.47132558e-02 5.22590995e-01 5.02987504e-01
3.24063189e-02 7.54791021e-01 5.47158122e-01 -4.86921847e-01
-3.59004378e-01 -3.99241149e-01 5.60202956e-01 7.98512578e-01
5.16903758e-01 -2.41124898e-01 -5.70376992e-01 -5.95058985e-02
4.98865396e-01 -4.35999542e-01 -6.20113388e-02 5.61077893e-02
-9.83187139e-01 5.83052933e-01 7.63960540e-01 4.57352221e-01
-4.70905632e-01 1.17253087e-01 1.76023468e-01 1.30314931e-01
8.85151386e-01 6.54428482e-01 -8.78291547e-01 3.38537130e-03
-6.39417768e-01 3.13527167e-01 1.21633983e+00 1.23297930e+00
8.83150578e-01 -4.73350614e-01 -3.82379144e-01 8.06379139e-01
2.75133610e-01 2.23352075e-01 -1.72741786e-01 -4.33137357e-01
9.43806410e-01 1.25472689e+00 -2.88352340e-01 -1.18983471e+00
-8.18571031e-01 -5.97813129e-01 -6.32402360e-01 -4.53544110e-01
2.26139337e-01 -2.48346880e-01 -1.00575721e+00 1.41885018e+00
4.65699673e-01 1.57557622e-01 1.56784713e-01 4.75351751e-01
1.62748849e+00 4.31522280e-01 2.38832116e-01 -9.91030931e-02
1.44541550e+00 -1.12096143e+00 -1.00246596e+00 -5.49416900e-01
1.14247060e+00 -5.65593541e-01 4.29191828e-01 -9.24785659e-02
-8.61525774e-01 -2.39630058e-01 -9.10495341e-01 -4.48616803e-01
-6.15145326e-01 1.17212087e-02 9.55611944e-01 1.83175355e-01
-9.12836254e-01 7.90980339e-01 -9.91602659e-01 -4.17595416e-01
6.08269095e-01 5.66600621e-01 -6.66307867e-01 -2.22937241e-02
-1.41256642e+00 1.07138002e+00 8.60717475e-01 1.37362823e-01
-1.52523935e-01 -7.70618737e-01 -1.25223780e+00 2.72234917e-01
7.93980837e-01 -5.53559601e-01 9.94264007e-01 -1.14396773e-01
-9.04333889e-01 9.35707152e-01 -4.28348988e-01 -7.12233365e-01
-5.95251210e-02 -6.21359348e-01 -7.24706113e-01 1.09846212e-01
2.79185653e-01 4.31554556e-01 -9.72872227e-02 -9.53934550e-01
-7.01268554e-01 -3.33514839e-01 1.63627207e-01 1.46802709e-01
-1.59421816e-01 3.37753206e-01 -9.73458290e-01 -5.43984473e-01
2.77069211e-01 -7.96918392e-01 -2.48392537e-01 -6.59221530e-01
-1.09432542e+00 -7.22148180e-01 5.43409646e-01 -9.40649152e-01
1.60677826e+00 -1.61667156e+00 -4.76482231e-03 4.21399534e-01
8.91110241e-01 2.56334454e-01 -2.73586065e-01 7.37668574e-01
-3.64283025e-01 3.77907306e-01 -1.68485090e-01 -3.53654653e-01
-1.14772283e-01 2.47605056e-01 7.03361928e-02 1.21779963e-01
7.03652263e-01 1.58525074e+00 -1.01214242e+00 -6.44837141e-01
-2.15083405e-01 3.76870781e-01 -2.85359263e-01 9.62776765e-02
-1.22485250e-01 3.41744795e-02 -5.02319574e-01 4.70718145e-01
5.06342471e-01 -5.63143849e-01 7.92587519e-01 -2.66744673e-01
2.77107269e-01 1.07700431e+00 -9.10914421e-01 1.62874866e+00
-3.06153655e-01 5.36779583e-01 -3.24574143e-01 -8.14586997e-01
1.12259591e+00 3.12932163e-01 5.90121686e-01 -5.63874424e-01
8.18058550e-02 9.76623818e-02 1.25246823e-01 -3.47476542e-01
5.85094750e-01 4.50668216e-01 -1.80119351e-01 3.53073567e-01
4.00857747e-01 -4.84199971e-02 5.58391929e-01 8.76127362e-01
1.60526049e+00 2.35377386e-01 6.87332213e-01 -1.63284764e-01
1.76362887e-01 -4.09175456e-02 7.49766648e-01 5.37071943e-01
2.66657203e-01 2.89674371e-01 1.01157320e+00 -4.05383557e-01
-5.65473497e-01 -8.43193769e-01 2.66494989e-01 8.12455773e-01
5.32548912e-02 -1.14218068e+00 -3.95963162e-01 -1.26158535e+00
1.28156945e-01 6.60139143e-01 -6.01853192e-01 6.52140705e-03
-9.13740516e-01 -6.40337110e-01 5.25057554e-01 7.09981799e-01
4.40555334e-01 -1.10786963e+00 1.82016492e-01 4.34512109e-01
-3.27840596e-01 -1.83488739e+00 -2.10719138e-01 2.78455526e-01
-5.65858722e-01 -1.15037942e+00 6.94260001e-02 -8.34702551e-01
4.89215881e-01 -7.11557120e-02 1.76729226e+00 3.13340187e-01
2.75483411e-02 -2.98386455e-01 -5.61802745e-01 -1.17924280e-01
-4.96922852e-03 8.17685366e-01 -3.58212024e-01 -4.70028818e-01
7.00873077e-01 -6.19871497e-01 -3.12549353e-01 -1.48815140e-01
-4.76256013e-01 6.50198162e-02 6.33257568e-01 6.83502436e-01
4.52484846e-01 7.36040249e-02 4.62505549e-01 -1.55594397e+00
5.43147266e-01 -3.53235722e-01 -3.66084009e-01 4.04378146e-01
-8.51053357e-01 1.49737492e-01 1.40811712e-01 3.68420209e-04
-8.84652436e-01 -1.43813595e-01 -8.02520886e-02 2.07497731e-01
-6.01011440e-02 1.01634121e+00 -3.01910818e-01 3.40305686e-01
4.09921378e-01 -5.09464800e-01 -5.16169429e-01 -3.70780438e-01
7.27189541e-01 3.87843579e-01 4.56378579e-01 -4.32361454e-01
8.78895342e-01 1.58893868e-01 4.24368680e-02 -4.60027844e-01
-1.35339236e+00 -5.28324008e-01 -1.09081054e+00 2.97373474e-01
9.15906966e-01 -9.57379222e-01 -6.71405733e-01 5.22494279e-02
-1.46976674e+00 -4.06913191e-01 -2.21162289e-01 4.13965642e-01
1.60025269e-01 2.54009455e-01 -1.08635199e+00 -6.61864996e-01
-6.07737780e-01 -6.69801354e-01 1.10215163e+00 1.86094657e-01
-5.86053371e-01 -1.04742324e+00 4.20024944e-03 4.34977174e-01
-4.58820127e-02 4.36644971e-01 9.43114400e-01 -1.34123743e+00
-4.50689495e-01 -2.73578733e-01 -8.21410835e-01 -8.42322931e-02
4.15231794e-01 -9.51257721e-02 -7.68789649e-01 3.51706445e-02
-8.96995962e-01 -2.08003670e-01 1.02792585e+00 -1.06713055e-02
6.03569508e-01 -3.60403985e-01 -8.87216747e-01 3.35855365e-01
1.09637010e+00 3.08123287e-02 5.36053002e-01 1.31924912e-01
1.19036651e+00 7.41207480e-01 6.04758382e-01 1.22201048e-01
9.10381854e-01 5.62956512e-01 2.48439834e-01 -4.60219711e-01
-3.27927798e-01 -3.57296199e-01 -1.27367124e-01 7.13366091e-01
-2.22147197e-01 -4.55570847e-01 -1.31135869e+00 7.09402025e-01
-1.94313443e+00 -6.10212922e-01 -6.04467452e-01 1.61487937e+00
9.91168022e-01 4.42208529e-01 -1.46374581e-02 -1.31222710e-01
6.56101465e-01 2.94538826e-01 -1.45225003e-01 -1.71837300e-01
-2.32278839e-01 7.60853767e-01 5.27504325e-01 5.72233558e-01
-1.36410987e+00 1.65320575e+00 5.83911753e+00 6.76755965e-01
-5.00958323e-01 -1.43778026e-01 4.83549327e-01 1.91107571e-01
-2.64115930e-01 4.62216318e-01 -9.45298254e-01 -2.13264048e-01
9.37810540e-01 -2.87682656e-02 2.68167667e-02 4.23015118e-01
-1.78377971e-01 -2.55408115e-03 -9.93669927e-01 5.35212398e-01
-2.58584768e-01 -1.31804621e+00 -2.02985674e-01 1.97854042e-01
7.38247275e-01 2.82373935e-01 -5.13413191e-01 4.29632515e-01
1.01779401e+00 -1.23847997e+00 1.13391116e-01 2.85976976e-01
8.31013262e-01 -8.82407665e-01 1.08603823e+00 -1.49028897e-01
-1.82043409e+00 3.86597991e-01 8.92939279e-04 -9.50031653e-02
3.38603288e-01 1.02440619e+00 -1.13532352e+00 1.31517875e+00
6.08041584e-01 9.97136176e-01 -6.52263343e-01 4.83758092e-01
-8.30263853e-01 6.62766218e-01 -3.53227705e-01 -3.33374329e-02
1.84470311e-01 2.42418349e-02 3.85698259e-01 1.46377516e+00
-1.75150692e-01 4.08099532e-01 3.83230746e-01 7.45581686e-01
-5.82899749e-01 3.14833254e-01 -5.24300992e-01 -4.16459709e-01
6.73116863e-01 1.57225955e+00 -8.59510422e-01 -4.04989809e-01
-5.73797047e-01 8.71687591e-01 9.80156958e-01 3.51876110e-01
-6.20571971e-01 -7.87151575e-01 5.35168648e-01 -1.94126725e-01
5.29852986e-01 -3.38727474e-01 -3.30101699e-01 -1.10085940e+00
2.73924880e-02 -4.69070941e-01 6.51999712e-01 -3.76173168e-01
-1.24655187e+00 1.00880516e+00 8.16952437e-02 -6.15003109e-01
-2.87592798e-01 -4.76414025e-01 -3.86177480e-01 9.57783401e-01
-1.40705144e+00 -1.59039366e+00 -1.27106518e-01 1.58058777e-01
-1.35046588e-02 -6.94659259e-03 9.65082526e-01 1.56837419e-01
-7.19956517e-01 7.49092579e-01 -6.39344335e-01 1.00087297e+00
5.30121863e-01 -1.41882217e+00 1.23393369e+00 9.19848979e-01
7.53594697e-01 7.36586988e-01 2.69954741e-01 -1.19838202e+00
-9.77043867e-01 -1.26959205e+00 1.77420914e+00 -5.24887443e-01
1.07065749e+00 -5.05787551e-01 -9.37592328e-01 1.10158539e+00
3.66121143e-01 3.56401443e-01 6.61501348e-01 8.82509708e-01
-8.52004766e-01 3.87248620e-02 -8.18881631e-01 5.79210460e-01
1.58157980e+00 -5.16434729e-01 -5.39675057e-01 1.88451484e-01
1.18715596e+00 -5.74892640e-01 -1.43091822e+00 8.43597591e-01
4.27596241e-01 -5.28736532e-01 7.73011506e-01 -8.20607424e-01
4.37401384e-01 -3.59723752e-04 7.09414184e-02 -1.25677741e+00
-3.88804078e-01 -6.42600596e-01 -6.37854695e-01 1.55047274e+00
1.18692160e+00 -7.39118099e-01 8.65635693e-01 4.91881788e-01
8.83043110e-02 -1.02585614e+00 -5.38172901e-01 -5.94299316e-01
-2.01210622e-02 -3.95923853e-01 7.02151358e-01 1.21124983e+00
3.93367171e-01 1.02061379e+00 6.46011904e-02 3.08642715e-01
2.99151748e-01 3.48706037e-01 6.27546191e-01 -1.41732466e+00
-2.00127393e-01 -3.02835792e-01 -3.97179335e-01 -9.45528567e-01
6.31881416e-01 -1.15520954e+00 -1.59077004e-01 -2.12981725e+00
2.53623813e-01 -5.20370245e-01 -3.40613693e-01 1.02633321e+00
-5.40131986e-01 1.24846973e-01 5.51982373e-02 -1.97242871e-02
-6.17562354e-01 3.96813512e-01 1.03904057e+00 -1.66635290e-01
-3.21440846e-01 -2.54722893e-01 -8.49919438e-01 3.43627930e-01
7.22354531e-01 -5.22416592e-01 -2.89387733e-01 -4.05281484e-01
4.30740803e-01 -1.32333308e-01 -5.41560873e-02 -4.18149024e-01
2.42123321e-01 -1.57356001e-02 2.14241609e-01 -8.28629851e-01
1.90561801e-01 -4.30827528e-01 -2.34066367e-01 3.57130915e-02
-3.76508564e-01 -5.16010523e-02 1.65801018e-01 3.87894303e-01
-3.30989748e-01 2.52022088e-01 1.41293615e-01 4.06927355e-02
-4.61055189e-01 5.05278766e-01 1.94613457e-01 2.83617795e-01
7.51946211e-01 4.06339139e-01 -6.17133439e-01 -3.65559936e-01
-7.79067814e-01 4.64280695e-01 -8.84768814e-02 4.87675190e-01
5.27536690e-01 -1.27093172e+00 -1.00317931e+00 -1.81446970e-01
2.57582664e-01 5.29180229e-01 -1.94174647e-01 7.78842628e-01
-3.26018751e-01 4.22346652e-01 5.56349576e-01 -3.74722108e-02
-1.38822734e+00 3.87303352e-01 2.01068476e-01 -1.11779428e+00
-7.50349045e-01 1.09415388e+00 -6.31425250e-03 -6.55036807e-01
-1.88365951e-01 -4.27218169e-01 -4.83147472e-01 8.83602127e-02
2.60826349e-01 4.44897264e-02 3.51707876e-01 -6.96300745e-01
-5.88230550e-01 1.87285691e-01 -4.51661736e-01 -8.23154524e-02
1.50524294e+00 4.42122445e-02 -7.37155497e-01 1.74379602e-01
1.19691169e+00 2.73824483e-01 -5.92766821e-01 -7.13816047e-01
7.85585344e-01 1.37325528e-03 -1.17818654e-01 -8.45220566e-01
-1.22298193e+00 3.58968079e-01 -3.97126079e-01 4.54515934e-01
7.63346136e-01 5.93105674e-01 9.49616075e-01 3.84790808e-01
1.60780326e-01 -7.12219298e-01 -3.12046677e-01 9.39798772e-01
6.89112961e-01 -1.17096627e+00 3.89781386e-01 -1.25071061e+00
-5.24101853e-01 9.09221470e-01 7.26531804e-01 -1.28052711e-01
8.23543131e-01 6.43031240e-01 -9.49810520e-02 -5.76413572e-01
-7.88581252e-01 -7.76668668e-01 8.50940824e-01 4.32761252e-01
8.80810976e-01 2.03451976e-01 -3.38470846e-01 7.33974814e-01
-3.57277185e-01 -4.79187936e-01 -4.93088923e-02 8.94695401e-01
3.90345725e-04 -1.44352412e+00 3.72616768e-01 6.36991501e-01
-6.16304815e-01 -6.44878089e-01 -8.77230227e-01 9.80413616e-01
-5.36799394e-02 1.27990651e+00 6.01159669e-02 -6.81455016e-01
4.56214398e-01 -3.47950589e-03 3.03183138e-01 -1.02202868e+00
-7.98113942e-01 -2.43484750e-02 9.79490638e-01 -7.06350684e-01
-3.64898533e-01 -3.77610952e-01 -1.74403262e+00 -7.25003183e-02
-6.51660621e-01 2.40049005e-01 1.56000853e-01 1.16254258e+00
5.77544510e-01 8.27290893e-01 2.61582792e-01 -1.51518568e-01
2.92237639e-01 -1.31617904e+00 -4.99861896e-01 3.27865899e-01
-4.07167897e-02 -4.95115280e-01 9.60501283e-02 -3.12703490e-01] | [9.251664161682129, 8.613909721374512] |
baf14636-2f46-4ade-8967-5bc00ec4d597 | what-else-can-fool-deep-learning-addressing-1 | 1912.06960 | null | https://arxiv.org/abs/1912.06960v1 | https://arxiv.org/pdf/1912.06960v1.pdf | What Else Can Fool Deep Learning? Addressing Color Constancy Errors on Deep Neural Network Performance | There is active research targeting local image manipulations that can fool deep neural networks (DNNs) into producing incorrect results. This paper examines a type of global image manipulation that can produce similar adverse effects. Specifically, we explore how strong color casts caused by incorrectly applied computational color constancy - referred to as white balance (WB) in photography - negatively impact the performance of DNNs targeting image segmentation and classification. In addition, we discuss how existing image augmentation methods used to improve the robustness of DNNs are not well suited for modeling WB errors. To address this problem, a novel augmentation method is proposed that can emulate accurate color constancy degradation. We also explore pre-processing training and testing images with a recent WB correction algorithm to reduce the effects of incorrectly white-balanced images. We examine both augmentation and pre-processing strategies on different datasets and demonstrate notable improvements on the CIFAR-10, CIFAR-100, and ADE20K datasets. | ['Michael S. Brown', 'Mahmoud Afifi'] | 2019-12-15 | what-else-can-fool-deep-learning-addressing | http://openaccess.thecvf.com/content_ICCV_2019/html/Afifi_What_Else_Can_Fool_Deep_Learning_Addressing_Color_Constancy_Errors_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Afifi_What_Else_Can_Fool_Deep_Learning_Addressing_Color_Constancy_Errors_ICCV_2019_paper.pdf | iccv-2019-10 | ['color-constancy'] | ['computer-vision'] | [ 5.44156015e-01 -1.53452203e-01 3.01730782e-01 -6.93755150e-01
-4.41739112e-01 -4.21337485e-01 3.94921333e-01 -2.37612367e-01
-7.18843341e-01 5.26407242e-01 -4.42049131e-02 -6.16476297e-01
2.99395978e-01 -5.01878500e-01 -1.04973853e+00 -4.93640751e-01
3.24408203e-01 -3.04906011e-01 2.10993901e-01 -7.07975850e-02
4.06384677e-01 6.44295990e-01 -1.20827985e+00 3.90712172e-01
9.96490777e-01 7.98281133e-01 -4.76960875e-02 7.67398834e-01
-4.82268706e-02 9.44817960e-01 -1.17349088e+00 -4.04665172e-01
4.94101822e-01 -5.36567569e-01 -4.58219707e-01 1.46655872e-01
1.40830100e+00 -7.31463492e-01 -4.51072976e-02 1.33507955e+00
4.93258029e-01 1.42642185e-01 4.05563831e-01 -1.35757792e+00
-1.00624859e+00 3.52797359e-01 -7.63691306e-01 6.58582807e-01
-4.92109478e-01 3.64556104e-01 4.17475730e-01 -6.68130040e-01
3.78249109e-01 1.27242434e+00 7.60254920e-01 6.98842704e-01
-1.36883259e+00 -9.32295322e-01 3.72218937e-01 1.27768427e-01
-1.00362551e+00 -5.64516127e-01 6.11844540e-01 -2.05546066e-01
7.58475542e-01 2.34568238e-01 5.32908678e-01 9.96149361e-01
9.81936753e-02 6.97951078e-01 1.41607463e+00 -4.73827839e-01
1.25807002e-01 -4.73587327e-02 2.72579968e-01 5.20411015e-01
5.98384738e-01 1.17421605e-01 -3.66533011e-01 2.21385464e-01
9.04722571e-01 -3.97550225e-01 -3.23518902e-01 5.75499423e-02
-7.58497596e-01 4.77145672e-01 6.83316886e-01 2.89936457e-02
-5.48692122e-02 4.52561677e-01 2.30088294e-01 1.38635799e-01
6.45227373e-01 8.13200593e-01 -5.59998155e-01 8.23852718e-02
-8.65445495e-01 1.34778410e-01 2.41061062e-01 7.05041587e-01
6.74996495e-01 7.11287141e-01 -2.79632598e-01 1.02845597e+00
1.92416713e-01 3.76458436e-01 2.97756046e-01 -9.80323970e-01
5.36131263e-01 3.92657161e-01 -4.52543050e-02 -1.00877106e+00
-4.35605496e-01 -5.48348963e-01 -3.46446335e-01 8.38345706e-01
7.46851504e-01 -2.68702149e-01 -1.55674350e+00 1.82450581e+00
5.05687520e-02 2.57136106e-01 -3.30099687e-02 1.08292544e+00
6.84779227e-01 3.99983168e-01 5.31211317e-01 3.10200721e-01
1.05053294e+00 -1.24633777e+00 -7.84237027e-01 -7.06213236e-01
5.00935376e-01 -7.47712374e-01 1.37635779e+00 3.78046125e-01
-1.08243716e+00 -4.89808291e-01 -1.36284912e+00 -2.89863914e-01
-5.40253758e-01 2.38526091e-01 5.39066911e-01 1.14946806e+00
-1.11441028e+00 5.84505141e-01 -7.83792496e-01 -9.37838703e-02
8.50423634e-01 2.19776765e-01 -9.58398059e-02 -1.71840101e-01
-8.26913595e-01 1.07707131e+00 1.13317326e-01 1.91475108e-01
-7.33797729e-01 -1.12764490e+00 -6.72874272e-01 -1.17594324e-01
1.80448681e-01 -1.26138777e-01 1.18688977e+00 -1.80986333e+00
-1.14304626e+00 8.37736130e-01 1.79224968e-01 -5.23470342e-01
4.66590673e-01 -4.53457206e-01 -4.20275122e-01 1.99797571e-01
-1.48461297e-01 1.09299338e+00 9.67553377e-01 -1.62671089e+00
-3.06971610e-01 -3.96727026e-01 -1.91202864e-01 3.19628507e-01
-5.56125045e-01 1.51884750e-01 -5.70294201e-01 -1.09028673e+00
-1.22280857e-02 -8.58619332e-01 -1.44781977e-01 2.32501253e-01
-4.47453201e-01 4.30387050e-01 9.02487576e-01 -8.16562414e-01
7.73077548e-01 -2.15327668e+00 -7.13904917e-01 2.03468427e-01
6.31551594e-02 7.69492447e-01 -6.90474629e-01 -3.62949789e-01
-3.37600052e-01 5.45780659e-01 -2.50887126e-01 -4.23266202e-01
-2.69664198e-01 2.32143655e-01 -3.20233315e-01 3.76539648e-01
4.64731157e-01 9.70261335e-01 -4.03505147e-01 -1.43681196e-02
2.48034865e-01 5.85919440e-01 -5.80565333e-01 -2.81935692e-01
-1.65294155e-01 1.45322844e-01 2.09523618e-01 5.26617169e-01
1.16278267e+00 1.82999715e-01 -5.51838502e-02 -2.68452048e-01
-1.90260471e-03 1.58824235e-01 -6.71640098e-01 1.19503951e+00
-1.31895646e-01 1.13689768e+00 3.82372551e-02 -7.14922309e-01
6.69050038e-01 -2.59117842e-01 3.47990952e-02 -1.14288104e+00
3.20707202e-01 -1.09746397e-01 3.96484286e-01 -1.34382024e-01
5.70263982e-01 -2.29357127e-02 6.10936761e-01 2.89343566e-01
-2.02477321e-01 -2.53156107e-02 4.89043668e-02 3.02490499e-02
8.49376798e-01 1.07094035e-01 -4.30159003e-01 -3.56324613e-01
-1.80151448e-01 1.51139721e-01 5.11653006e-01 9.98026073e-01
-5.24633825e-01 9.98659194e-01 4.56559092e-01 -1.77195504e-01
-1.10161936e+00 -8.03563833e-01 9.13268179e-02 1.18788612e+00
2.02256262e-01 -6.41828552e-02 -1.10533655e+00 -3.94534618e-01
-2.75539011e-02 8.60947132e-01 -6.30398691e-01 -5.82032382e-01
-3.50755572e-01 -1.18912697e+00 9.81082797e-01 8.98744166e-01
9.77429092e-01 -9.13416207e-01 -7.21131086e-01 -3.69101763e-02
4.33170311e-02 -1.36611986e+00 -4.91710871e-01 1.94747537e-01
-9.84397352e-01 -1.02537429e+00 -7.50152349e-01 -7.07627535e-01
1.03042018e+00 5.37026346e-01 9.48922515e-01 3.12406510e-01
-5.76524138e-01 2.06533417e-01 -3.23476732e-01 -7.03309357e-01
-3.87269974e-01 -3.21365774e-01 -2.94051379e-01 -1.70959800e-01
2.30750993e-01 -8.80470648e-02 -6.16505861e-01 3.04587096e-01
-1.16133487e+00 2.04705417e-01 4.84745294e-01 5.02123177e-01
2.17880920e-01 6.45784736e-02 3.56743962e-01 -1.15802717e+00
7.47292519e-01 1.21549509e-01 -6.15248442e-01 1.52753532e-01
-7.39877105e-01 4.26384956e-02 4.24441069e-01 -7.03059733e-01
-1.32334042e+00 -1.76361665e-01 -8.94100890e-02 -5.24735153e-01
-1.74851447e-01 2.07939342e-01 -6.77885190e-02 -6.42559409e-01
9.06660020e-01 -2.99704969e-01 -8.43109414e-02 -4.76675093e-01
2.24233702e-01 2.10799351e-01 6.84693575e-01 -3.28003109e-01
7.97324479e-01 5.10695040e-01 -2.62491375e-01 -7.22973228e-01
-7.07381666e-01 -2.32245121e-02 -2.77392149e-01 -1.96190268e-01
8.80248189e-01 -8.75449598e-01 -1.89387843e-01 1.12334144e+00
-8.85956228e-01 -9.91542876e-01 3.91103365e-02 2.47635081e-01
3.83808464e-02 7.40498528e-02 -6.21082664e-01 -5.82382262e-01
-2.14885607e-01 -1.09269297e+00 5.55614710e-01 5.32970905e-01
3.25324163e-02 -8.54950726e-01 -4.72933322e-01 4.07137394e-01
7.42053092e-01 3.35016191e-01 9.74609554e-01 -2.96841651e-01
-4.23838109e-01 1.13844492e-01 -6.74993634e-01 7.08360851e-01
6.79322518e-03 3.25191468e-01 -1.32792532e+00 -7.52981901e-02
-2.59477496e-01 -2.08078250e-01 1.10008597e+00 6.14916027e-01
1.36837482e+00 -2.01014891e-01 5.72713800e-02 6.84891522e-01
1.44051123e+00 3.62461269e-01 9.29084063e-01 6.50739193e-01
8.51507485e-01 4.09723461e-01 4.01793808e-01 -9.52146575e-03
5.16842194e-02 4.62499499e-01 4.90446657e-01 -8.56735289e-01
-7.02034533e-01 7.41987536e-03 1.82740062e-01 -6.25252053e-02
1.26551852e-01 -2.91367054e-01 -8.57420802e-01 5.16419232e-01
-1.33049631e+00 -6.25426769e-01 -3.76845688e-01 1.92791343e+00
8.99223149e-01 2.26109385e-01 -1.30579442e-01 6.52467459e-02
7.40449488e-01 2.63638422e-02 -6.21049047e-01 -5.00215054e-01
-3.85117829e-01 3.55698973e-01 1.03064322e+00 4.83639717e-01
-1.10507917e+00 1.23730171e+00 6.72466707e+00 3.55578601e-01
-1.34534383e+00 4.50970642e-02 1.24225485e+00 -9.65847299e-02
-2.43055984e-01 -3.27923805e-01 -4.81778502e-01 3.24340284e-01
6.59195602e-01 5.69032669e-01 5.75593472e-01 7.02176630e-01
3.43468726e-01 -4.07345176e-01 -6.74278140e-01 9.20242727e-01
3.24191958e-01 -1.22327673e+00 3.06784213e-02 -7.57521614e-02
9.23591435e-01 1.09798551e-01 6.57076776e-01 7.79224485e-02
3.72035712e-01 -9.76677299e-01 9.26815867e-01 2.17076167e-01
7.65438735e-01 -7.74242759e-01 4.13353503e-01 -2.30182812e-01
-5.36955357e-01 -9.32842717e-02 -4.19890732e-01 1.09953292e-01
-2.15913117e-01 5.00737846e-01 -7.33622611e-01 -1.48717493e-01
8.88058782e-01 3.14402193e-01 -1.21988821e+00 1.26892149e+00
-3.51989746e-01 7.79733539e-01 -2.45980009e-01 3.22760105e-01
2.22056299e-01 -1.05040282e-01 1.55852556e-01 1.06560445e+00
-7.13782609e-02 1.36514222e-02 -4.53546703e-01 1.02703118e+00
-3.10497761e-01 -3.05886298e-01 -3.28520149e-01 -5.03618009e-02
3.22768331e-01 1.11204660e+00 -1.03005672e+00 -3.14670801e-01
-4.51713443e-01 1.18114316e+00 1.75394505e-01 9.20336127e-01
-9.87216413e-01 -2.14417174e-01 8.27046692e-01 2.69124396e-02
1.06641874e-01 -3.18105876e-01 -8.97980452e-01 -9.05947030e-01
-1.24475844e-01 -1.01358557e+00 3.61939594e-02 -1.20042157e+00
-1.03291285e+00 4.65692610e-01 -1.20877229e-01 -6.32595718e-01
4.77580100e-01 -8.53549957e-01 -7.46032894e-01 7.55343795e-01
-1.82647693e+00 -9.68830585e-01 -5.66632092e-01 4.76369917e-01
3.92665207e-01 5.86379617e-02 2.67067701e-01 4.89580661e-01
-7.08260715e-01 9.47701812e-01 3.32878754e-02 2.83144563e-01
9.53729391e-01 -1.24476552e+00 7.81229019e-01 1.40365255e+00
3.93253826e-02 6.34054542e-01 6.60263896e-01 -6.70269787e-01
-9.71864820e-01 -1.36578655e+00 1.14578933e-01 -2.42872462e-01
2.47607544e-01 -3.45393986e-01 -9.59875882e-01 6.71471059e-01
2.90559947e-01 9.18235481e-02 3.55193228e-01 -7.82163143e-02
-6.23837054e-01 -2.29406565e-01 -1.15784359e+00 8.60879302e-01
8.50045860e-01 -3.64266902e-01 -1.98710158e-01 9.09598917e-02
5.38849473e-01 -4.76707369e-01 -1.93564698e-01 3.08899522e-01
3.02682102e-01 -9.33711767e-01 9.94353712e-01 -6.63941681e-01
3.60127896e-01 -1.99710965e-01 3.35429944e-02 -1.42667615e+00
-1.28362313e-01 -3.97427768e-01 3.82039696e-01 1.33258259e+00
3.89956981e-01 -5.53492248e-01 7.44044244e-01 1.11702418e+00
-2.83614039e-01 -2.16359958e-01 -4.95368868e-01 -7.13674366e-01
2.28888825e-01 -4.71839428e-01 3.31309050e-01 9.16804850e-01
-7.52822161e-01 -1.46064773e-01 -1.73001692e-01 2.27632686e-01
4.52065378e-01 -4.66765195e-01 7.13340163e-01 -6.63088381e-01
1.21762127e-01 -6.12236559e-01 -2.26392195e-01 -6.18483961e-01
-5.65931015e-02 -4.46857303e-01 1.96984351e-01 -1.29818988e+00
-5.13790809e-02 -4.56428498e-01 -2.82133400e-01 8.53861630e-01
-5.11222839e-01 8.61008763e-01 5.53612590e-01 -5.70523962e-02
-2.41120338e-01 2.13750944e-01 1.30653846e+00 -7.57872090e-02
-7.41115063e-02 -3.62341791e-01 -1.02602017e+00 6.21011972e-01
1.11480665e+00 -3.73634547e-01 -5.71667731e-01 -1.04250860e+00
6.04532696e-02 -7.88098276e-01 7.28407085e-01 -1.29728556e+00
-4.24310230e-02 -5.30302823e-02 8.19476068e-01 -2.91422248e-01
1.85281605e-01 -8.59350443e-01 -3.41030151e-01 3.56792629e-01
-5.76823056e-01 1.96794048e-01 8.75839651e-01 4.10525501e-01
6.57711849e-02 -2.38053128e-01 1.28222775e+00 -5.14791626e-03
-8.57882440e-01 2.21103504e-02 -4.96140152e-01 1.57915786e-01
8.58550847e-01 -2.51849711e-01 -9.36108887e-01 -3.74484986e-01
-2.70020574e-01 5.25105745e-02 5.51794291e-01 5.32771528e-01
6.08870506e-01 -1.13689852e+00 -3.53824973e-01 2.68081695e-01
-1.32200450e-01 -2.58037835e-01 1.41773358e-01 6.10529482e-01
-8.57903481e-01 -6.76022619e-02 -5.53216457e-01 -2.41596729e-01
-1.39431119e+00 2.27030352e-01 6.89387858e-01 3.79603505e-01
-3.87978196e-01 1.22023034e+00 2.88803548e-01 -1.64851472e-01
3.45194429e-01 -7.18900859e-01 2.11482663e-02 -1.57652646e-01
5.38967013e-01 2.00052112e-01 4.47864354e-01 -3.10873985e-01
-2.29757532e-01 2.23805517e-01 -3.46141100e-01 -3.42863262e-01
1.15670216e+00 -1.37118801e-01 9.26775113e-02 1.71880815e-02
1.14203310e+00 -2.33084336e-01 -1.75591588e+00 7.04221278e-02
-1.17694639e-01 -6.36557937e-01 4.99047041e-01 -1.33492208e+00
-1.52022350e+00 9.92316782e-01 9.94428456e-01 -9.28262547e-02
1.31987178e+00 -5.65438271e-01 6.19999528e-01 1.43657848e-01
-3.44505340e-01 -1.36521983e+00 3.00641775e-01 4.63093460e-01
6.38341367e-01 -1.25265121e+00 -4.13600244e-02 -1.99199200e-01
-9.54949796e-01 9.12087440e-01 1.19097209e+00 -4.98950481e-02
2.56543428e-01 4.27387834e-01 6.38253450e-01 -2.88658477e-02
-3.22340071e-01 -6.46652952e-02 3.06110203e-01 7.15423644e-01
2.38155156e-01 -2.47907817e-01 7.88589865e-02 2.58599758e-01
2.16464396e-03 -1.24328338e-01 7.34361529e-01 9.57704186e-01
-3.12940031e-01 -6.92587376e-01 -6.66359782e-01 5.76708496e-01
-6.51400208e-01 -4.92090672e-01 -6.75977826e-01 9.55910265e-01
1.77832037e-01 9.60000873e-01 3.24461639e-01 -2.28704557e-01
1.56444341e-01 -2.98150741e-02 5.87523103e-01 -4.18460935e-01
-6.17600918e-01 8.06474462e-02 2.49158428e-03 -5.35278618e-01
-4.38253015e-01 -3.84068012e-01 -1.33291185e+00 -2.93322980e-01
-4.26905125e-01 -5.35399675e-01 9.96795118e-01 9.28666949e-01
2.02182710e-01 8.92974079e-01 1.28198504e-01 -6.55973554e-01
-2.97981501e-01 -9.24061477e-01 -4.88886595e-01 6.59534931e-01
5.36982775e-01 -5.30080140e-01 -3.06274742e-01 3.60143185e-01] | [11.237035751342773, -0.6990960240364075] |
c76a4767-44ba-48b4-a434-8580b8de7570 | bop-benchmark-for-6d-object-pose-estimation | 1808.08319 | null | http://arxiv.org/abs/1808.08319v1 | http://arxiv.org/pdf/1808.08319v1.pdf | BOP: Benchmark for 6D Object Pose Estimation | We propose a benchmark for 6D pose estimation of a rigid object from a single
RGB-D input image. The training data consists of a texture-mapped 3D object
model or images of the object in known 6D poses. The benchmark comprises of: i)
eight datasets in a unified format that cover different practical scenarios,
including two new datasets focusing on varying lighting conditions, ii) an
evaluation methodology with a pose-error function that deals with pose
ambiguities, iii) a comprehensive evaluation of 15 diverse recent methods that
captures the status quo of the field, and iv) an online evaluation system that
is open for continuous submission of new results. The evaluation shows that
methods based on point-pair features currently perform best, outperforming
template matching methods, learning-based methods and methods based on 3D local
features. The project website is available at bop.felk.cvut.cz. | ['Tae-Kyun Kim', 'Xenophon Zabulis', 'Stephan Ihrke', 'Jiri Matas', 'Federico Tombari', 'Fabian Manhardt', 'Dirk Kraft', 'Tomas Hodan', 'Frank Michel', 'Caner Sahin', 'Bertram Drost', 'Anders Glent Buch', 'Wadim Kehl', 'Eric Brachmann', 'Carsten Rother', 'Joel Vidal'] | 2018-08-24 | bop-benchmark-for-6d-object-pose-estimation-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Tomas_Hodan_PESTO_6D_Object_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Tomas_Hodan_PESTO_6D_Object_ECCV_2018_paper.pdf | eccv-2018-9 | ['6d-pose-estimation-using-rgbd'] | ['computer-vision'] | [ 2.54695296e-01 -2.95979287e-02 -6.12533167e-02 -2.58583248e-01
-9.14264977e-01 -7.09064662e-01 7.47943819e-01 -1.80681869e-01
-3.35296720e-01 3.07685256e-01 -8.55473280e-02 1.42542988e-01
-4.24349666e-01 -3.56603414e-01 -8.07101548e-01 -5.82851827e-01
-1.78423766e-02 1.26235211e+00 4.70596820e-01 -1.43679023e-01
4.35242414e-01 1.12031603e+00 -1.66353500e+00 -7.29228258e-02
-2.22044196e-02 1.32028234e+00 2.98170090e-01 2.41351187e-01
9.12496522e-02 -1.79759432e-02 -4.69642729e-01 -3.30186158e-01
8.38078380e-01 -9.79913622e-02 -8.44058156e-01 2.72937864e-01
7.97058940e-01 -1.15353361e-01 -3.22483450e-01 7.37118900e-01
7.55673826e-01 -1.95435844e-02 6.06469750e-01 -9.89582241e-01
5.44092944e-03 -1.67545870e-01 -2.03679755e-01 -1.75527588e-01
1.10369694e+00 1.35114361e-02 7.09836602e-01 -1.11776030e+00
1.17897880e+00 1.24208128e+00 6.48441732e-01 4.05020952e-01
-1.20417154e+00 -4.06913795e-02 -2.08242282e-01 1.64232492e-01
-1.35472465e+00 -3.36381435e-01 8.20524871e-01 -5.19111693e-01
9.56127048e-01 5.71427703e-01 6.70305848e-01 9.45933580e-01
1.59119397e-01 3.58305812e-01 1.32890582e+00 -5.98588705e-01
8.84214118e-02 1.39704153e-01 -9.55555066e-02 5.10396600e-01
2.38763735e-01 3.29069555e-01 -6.16239965e-01 -1.08631507e-01
1.03477347e+00 -9.36421081e-02 -8.96472186e-02 -1.22436309e+00
-1.39394534e+00 3.29458296e-01 4.23823029e-01 1.43383756e-01
-2.50102818e-01 -1.69136837e-01 1.65676206e-01 2.98375636e-01
3.70234221e-01 3.19087833e-01 -5.76465786e-01 -1.58991292e-01
-5.78750670e-01 5.18748820e-01 9.45357680e-01 1.17165375e+00
9.26395357e-01 -3.48689497e-01 1.31467283e-01 6.07685626e-01
3.34977865e-01 7.12992072e-01 1.92712799e-01 -9.48351920e-01
2.96973854e-01 6.25598371e-01 4.36597735e-01 -9.53170538e-01
-6.63638353e-01 1.63510721e-02 -1.89836279e-01 4.49778318e-01
4.25905734e-01 4.19910789e-01 -7.92574048e-01 1.14157557e+00
6.19353771e-01 -2.66705245e-01 -2.53658354e-01 1.15161717e+00
1.03537893e+00 9.94321927e-02 -9.15316045e-01 -1.88109010e-01
1.10115492e+00 -7.06983745e-01 -4.63110089e-01 -7.96914250e-02
1.48707535e-02 -1.14328873e+00 6.77867115e-01 6.27562463e-01
-1.21306658e+00 -6.10585451e-01 -9.32243407e-01 -7.82782361e-02
-3.82417202e-01 -4.61728349e-02 3.39924097e-01 6.98850036e-01
-1.09626722e+00 6.12860560e-01 -7.68314898e-01 -6.52856767e-01
-9.68132317e-02 7.48847902e-01 -6.61473453e-01 -1.16043687e-02
-5.75667024e-01 1.39752400e+00 3.31177145e-01 1.43593505e-01
-5.54261267e-01 -5.17283916e-01 -8.69211495e-01 -7.72183120e-01
5.48195064e-01 -6.44410670e-01 1.23931384e+00 -5.94250441e-01
-1.90813947e+00 1.56882799e+00 5.23385108e-02 -1.07734554e-01
9.95309770e-01 -2.63773173e-01 -5.17220423e-03 2.58808851e-01
-2.67832369e-01 4.27080840e-01 7.86443472e-01 -1.53944767e+00
-2.94178408e-02 -8.44747901e-01 -1.71188079e-02 3.47584069e-01
3.25891405e-01 7.77969956e-02 -7.64633417e-01 -3.90187234e-01
6.70444846e-01 -1.24870014e+00 -1.88543752e-01 1.83978528e-01
-3.01768005e-01 -9.77719650e-02 7.80062318e-01 -5.37727833e-01
5.23024321e-01 -1.83883119e+00 4.81019497e-01 2.73752391e-01
-8.37929845e-02 3.52664813e-02 -1.10085137e-01 5.35274208e-01
-2.00866342e-01 -3.20246041e-01 3.90453488e-02 -3.86416495e-01
3.39424700e-01 1.94811016e-01 -3.10441572e-03 9.21792448e-01
-8.36295262e-02 6.54270411e-01 -5.72983146e-01 -4.15940762e-01
7.29303598e-01 4.46899235e-01 -1.12773821e-01 1.48994043e-01
-1.12280965e-01 4.78620052e-01 -3.46621066e-01 9.35673714e-01
9.80955839e-01 1.35882020e-01 -5.04990621e-03 -4.55289066e-01
-2.01339051e-01 2.80746073e-01 -1.74397135e+00 2.11875153e+00
-2.42551625e-01 8.29132646e-02 -4.66674492e-02 -8.47683430e-01
1.36199188e+00 3.13577682e-01 8.26999366e-01 -5.02099454e-01
2.36010879e-01 5.33042312e-01 -3.88550639e-01 -4.28219497e-01
2.33411551e-01 5.97958788e-02 -1.35617722e-02 2.07616359e-01
2.41473228e-01 -7.23298311e-01 -1.13683082e-01 -4.80602711e-01
9.28771973e-01 7.02025533e-01 3.54471147e-01 -2.69941956e-01
5.29773593e-01 -6.26820102e-02 9.04947296e-02 5.41192234e-01
-8.22484568e-02 9.69358861e-01 1.02099381e-01 -7.48688936e-01
-1.07373786e+00 -1.27884650e+00 -6.37499213e-01 2.84682751e-01
5.67897141e-01 -3.23853552e-01 -4.45235372e-01 -3.35258007e-01
4.41186219e-01 -1.65587924e-02 -6.32185042e-01 2.77888149e-01
-6.60857975e-01 -3.72896969e-01 -1.87194914e-01 4.05499011e-01
1.92991391e-01 -8.07445645e-01 -8.87198865e-01 -1.52822644e-01
9.53483656e-02 -1.22761500e+00 -5.39315790e-02 2.49223471e-01
-1.23766220e+00 -1.07979012e+00 -6.87314749e-01 -5.82378387e-01
6.81316674e-01 9.90365446e-02 1.11427844e+00 -3.07469338e-01
-4.72679436e-01 8.89169991e-01 -4.39481556e-01 -3.98325115e-01
2.68427636e-02 -1.10055603e-01 3.69087517e-01 -3.14131618e-01
3.11203927e-01 -4.27350372e-01 -4.43235874e-01 9.21177268e-01
-5.60562372e-01 -2.30377197e-01 4.92154717e-01 6.35070801e-01
9.28547621e-01 -6.74428165e-01 -2.34736919e-01 -3.00078154e-01
-1.56385779e-01 8.39368179e-02 -9.88491952e-01 1.18642837e-01
-3.80967438e-01 -1.14868715e-01 -1.82093799e-01 -3.02861661e-01
-7.73382306e-01 6.52467966e-01 -5.38966283e-02 -6.73055947e-01
-3.43726605e-01 -3.42267528e-02 -3.04758281e-01 -5.59622407e-01
7.97340035e-01 5.71994632e-02 2.00898632e-01 -9.76965725e-01
1.34144098e-01 3.84387672e-01 6.89573050e-01 -8.48452866e-01
1.00210524e+00 6.66098356e-01 3.53025496e-01 -8.00692677e-01
-5.00032127e-01 -7.23115027e-01 -1.21279252e+00 -4.12396878e-01
5.06025970e-01 -5.70882440e-01 -6.43124104e-01 4.61630225e-01
-1.29551983e+00 -7.12757558e-02 -3.94468516e-01 7.49371409e-01
-1.09723973e+00 2.61829317e-01 -2.79111713e-01 -6.46254838e-01
1.28617818e-02 -1.43268311e+00 1.54806006e+00 -4.15651873e-02
-1.01951838e-01 -8.26755941e-01 4.38624352e-01 5.16832471e-01
-7.39960745e-03 6.47472322e-01 2.39982396e-01 -5.09147167e-01
-8.11761737e-01 -6.69653416e-01 1.35820821e-01 1.38106287e-01
7.72010833e-02 2.68708263e-02 -1.06577969e+00 -3.77631634e-01
9.61726457e-02 -2.47691855e-01 3.66411924e-01 1.85894415e-01
8.75795960e-01 1.11332022e-01 -3.49878162e-01 6.45568788e-01
1.75653696e+00 8.02685544e-02 6.63863897e-01 6.60932541e-01
4.20044869e-01 7.39971220e-01 7.45139301e-01 3.81763756e-01
-4.41495813e-02 1.39486587e+00 9.65336442e-01 1.83816105e-01
-1.66263044e-01 8.55372846e-02 1.57184973e-01 7.16117561e-01
-7.48815477e-01 1.45600244e-01 -1.11124611e+00 1.80174440e-01
-1.61348522e+00 -4.78210956e-01 -2.66291708e-01 2.65491867e+00
4.10013169e-01 1.28093109e-01 3.67792606e-01 2.70778179e-01
6.37101233e-01 -1.63047407e-02 -4.82855111e-01 -1.68429762e-01
7.55818412e-02 3.95980746e-01 4.25131291e-01 4.23840076e-01
-1.09549177e+00 6.32486582e-01 6.71071386e+00 4.69947606e-01
-1.02466369e+00 2.06547715e-02 -8.92116949e-02 2.39124279e-02
1.08200029e-01 -1.61396295e-01 -1.00865185e+00 6.96450546e-02
6.31704807e-01 3.30868140e-02 1.22482762e-01 8.82548332e-01
-2.11587951e-01 -1.72799319e-01 -1.23897052e+00 1.17113209e+00
5.10040402e-01 -1.00426316e+00 -1.33583605e-01 2.20512539e-01
6.40803277e-01 3.15824449e-01 1.27141714e-01 -3.68233234e-01
-3.92545611e-01 -7.63440430e-01 9.95176971e-01 7.46940076e-01
9.42185700e-01 -4.04299766e-01 8.01452577e-01 1.74868345e-01
-9.32075858e-01 3.10372055e-01 -4.34011906e-01 2.63093598e-03
-9.40836370e-02 3.70204002e-01 -7.70586789e-01 1.08575583e+00
9.45403397e-01 6.52401745e-01 -5.92363596e-01 1.43949151e+00
-1.20720372e-01 -1.35552287e-01 -4.75397408e-01 7.18692839e-02
-7.48190135e-02 -3.41895461e-01 8.71787131e-01 1.07387412e+00
3.91168654e-01 -3.11175257e-01 1.08909151e-02 5.43211162e-01
3.96559089e-01 1.81455165e-01 -7.14858234e-01 6.02621794e-01
8.65306333e-02 1.21292818e+00 -8.12611580e-01 1.68733403e-01
-2.07214355e-01 8.83705080e-01 -1.22315243e-01 -4.60930653e-02
-5.04895926e-01 -1.54032037e-01 4.25254822e-01 3.07076782e-01
5.06798148e-01 -4.06780213e-01 -1.03995018e-01 -9.80593562e-01
3.71185124e-01 -6.16706252e-01 2.73946404e-01 -1.10120583e+00
-1.06473434e+00 6.71179056e-01 4.68138486e-01 -1.62444103e+00
-3.20903510e-01 -1.32339227e+00 -3.48559953e-02 8.67787004e-01
-1.20724654e+00 -1.05325830e+00 -6.71639740e-01 4.24278647e-01
3.29692394e-01 -3.66429798e-02 1.01392615e+00 1.42800197e-01
-9.33099240e-02 2.41629347e-01 1.15533568e-01 -2.81540364e-01
9.81287360e-01 -1.29059267e+00 1.68204248e-01 1.97241366e-01
1.41439453e-01 3.91127050e-01 7.63433933e-01 -3.05974066e-01
-1.86061275e+00 -4.51517671e-01 8.49105656e-01 -1.17598319e+00
3.59089762e-01 -5.54820061e-01 -6.10999107e-01 6.65404320e-01
-9.34897885e-02 3.10852975e-01 2.73078829e-01 -2.45678559e-01
-2.31231734e-01 -3.70524764e-01 -1.32784581e+00 -9.61183608e-02
1.28336310e+00 -3.72865230e-01 -8.56200993e-01 5.70739031e-01
3.55012894e-01 -1.33812571e+00 -1.17641342e+00 5.58621407e-01
8.31237316e-01 -1.06278169e+00 1.30581462e+00 -2.00867161e-01
6.03350326e-02 -3.79688233e-01 -3.74212593e-01 -8.86629701e-01
-5.88649362e-02 -7.28655398e-01 -1.09020106e-01 6.63543940e-01
-8.82108733e-02 -6.68498993e-01 7.55676925e-01 3.19662958e-01
-2.25919545e-01 -6.77486420e-01 -1.17812407e+00 -1.03736448e+00
-2.51872450e-01 -3.13000947e-01 4.50851530e-01 4.77374464e-01
-3.24348032e-01 -8.81999880e-02 -3.66845541e-03 1.24399707e-01
7.38907278e-01 4.02427554e-01 1.13663757e+00 -1.42610574e+00
1.16478316e-02 -3.93391967e-01 -1.15771937e+00 -9.36530352e-01
-3.85880433e-02 -7.53023922e-01 -1.03167258e-01 -1.33868778e+00
1.20553106e-01 -2.47579485e-01 -3.88953313e-02 3.84197891e-01
4.83466238e-01 5.50036192e-01 2.71712802e-02 2.39176497e-01
-4.94239271e-01 1.48578450e-01 1.37789714e+00 1.18914276e-01
1.23409398e-01 2.83387542e-01 1.00353241e-01 7.60513365e-01
3.94752324e-01 -3.53633434e-01 7.76215717e-02 -2.36047924e-01
2.49892712e-01 2.31767986e-02 7.46729255e-01 -1.04823709e+00
1.00696169e-01 -4.30948026e-02 4.93959695e-01 -1.16708827e+00
7.26787865e-01 -1.19705188e+00 4.69324321e-01 5.43918610e-01
2.87874401e-01 2.11664796e-01 2.04512253e-01 3.63535553e-01
-1.44097492e-01 -3.72953147e-01 7.52623677e-01 -3.05421293e-01
-6.81951702e-01 3.65262806e-01 3.31109434e-01 -2.53321797e-01
1.04870236e+00 -8.64623845e-01 2.42310925e-03 -7.10733235e-02
-9.37626719e-01 -2.78977990e-01 9.27143395e-01 3.59784812e-01
5.25988042e-01 -1.49161148e+00 -6.11053467e-01 1.74246818e-01
3.35952371e-01 1.40532613e-01 -7.01625878e-03 9.66233909e-01
-7.82454550e-01 5.86342335e-01 -4.75535840e-01 -1.10033798e+00
-1.35695338e+00 5.04302800e-01 5.74497581e-01 4.10957970e-02
-3.74568850e-01 7.03835726e-01 -1.50765345e-01 -9.02303874e-01
3.60467702e-01 -2.95314252e-01 1.29819438e-01 -5.27282991e-02
1.91127673e-01 5.38853526e-01 7.39795983e-01 -1.00166106e+00
-6.61155224e-01 1.28600788e+00 4.80623394e-01 -1.83650032e-01
1.48119211e+00 3.56461294e-02 -2.38196030e-01 6.83532774e-01
1.20017099e+00 -1.75050244e-01 -1.27724040e+00 -5.07529080e-01
8.21520388e-02 -8.05942118e-01 -1.75036132e-01 -8.24926436e-01
-7.36243844e-01 6.73723280e-01 9.21431720e-01 -8.49186182e-02
9.95855272e-01 3.42854768e-01 2.05619827e-01 4.44270641e-01
7.63072550e-01 -8.62330258e-01 2.91636348e-01 5.39184093e-01
1.50798476e+00 -1.14298892e+00 4.71134752e-01 -4.00333434e-01
-1.18683323e-01 1.45414925e+00 3.01456153e-01 -2.04095855e-01
6.71463668e-01 2.22575992e-01 -3.12546082e-02 -2.07323298e-01
-1.25575453e-01 -2.49226019e-01 7.87965715e-01 7.81566679e-01
3.70365605e-02 -2.16338471e-01 -3.06442171e-01 1.77626222e-01
-3.15061718e-01 -2.28368118e-01 6.40164018e-02 1.08519971e+00
-1.23392828e-01 -1.38871074e+00 -8.65309656e-01 9.62570086e-02
-1.02629639e-01 6.53832376e-01 -6.95179522e-01 1.05543351e+00
2.01357082e-01 4.30748194e-01 2.55177617e-01 -4.04357165e-01
8.27897966e-01 3.62024009e-02 1.41334271e+00 -5.50867617e-01
-4.96660531e-01 2.65369475e-01 -1.53778372e-02 -9.58637834e-01
-9.14432108e-01 -1.09887588e+00 -6.09030128e-01 -4.59804535e-02
-4.06569600e-01 -1.84145510e-01 1.11193311e+00 5.82486510e-01
1.21325791e-01 -1.51098922e-01 6.83183610e-01 -1.60860419e+00
-7.28814542e-01 -8.01080585e-01 -6.38108969e-01 3.18137378e-01
2.94761628e-01 -1.12039196e+00 -3.73427927e-01 1.00388139e-01] | [7.571387767791748, -2.618903160095215] |
c38a5065-791a-499d-b5eb-49cd7cbf7e38 | snippet-policy-network-for-multi-class-varied | 2107.13361 | null | https://arxiv.org/abs/2107.13361v1 | https://arxiv.org/pdf/2107.13361v1.pdf | Snippet Policy Network for Multi-class Varied-length ECG Early Classification | Arrhythmia detection from ECG is an important research subject in the prevention and diagnosis of cardiovascular diseases. The prevailing studies formulate arrhythmia detection from ECG as a time series classification problem. Meanwhile, early detection of arrhythmia presents a real-world demand for early prevention and diagnosis. In this paper, we address a problem of cardiovascular disease early classification, which is a varied-length and long-length time series early classification problem as well. For solving this problem, we propose a deep reinforcement learning-based framework, namely Snippet Policy Network (SPN), consisting of four modules, snippet generator, backbone network, controlling agent, and discriminator. Comparing to the existing approaches, the proposed framework features flexible input length, solves the dual-optimization solution of the earliness and accuracy goals. Experimental results demonstrate that SPN achieves an excellent performance of over 80\% in terms of accuracy. Compared to the state-of-the-art methods, at least 7% improvement on different metrics, including the precision, recall, F1-score, and harmonic mean, is delivered by the proposed SPN. To the best of our knowledge, this is the first work focusing on solving the cardiovascular early classification problem based on varied-length ECG data. Based on these excellent features from SPN, it offers a good exemplification for addressing all kinds of varied-length time series early classification problems. | ['Vincent S. Tseng', 'Gary G. Yen', 'Yu Huang'] | 2021-07-28 | null | null | null | null | ['arrhythmia-detection'] | ['medical'] | [ 1.47277296e-01 -4.29181904e-01 -2.11253539e-01 -9.91592482e-02
-6.23298347e-01 -2.72714287e-01 -2.81694438e-03 3.47897321e-01
-4.56230968e-01 7.49804080e-01 -4.52396274e-01 -3.15056652e-01
-6.62197828e-01 -5.24353802e-01 -1.31820574e-01 -8.46947134e-01
-6.33678198e-01 3.51421535e-01 -2.34476291e-02 5.09614460e-02
2.30920926e-01 4.38546658e-01 -1.22943056e+00 2.09414456e-02
9.66130197e-01 1.33246875e+00 -1.48991287e-01 6.28372610e-01
3.59545827e-01 6.99508786e-01 -8.63328815e-01 -7.36860111e-02
1.43549651e-01 -6.82213962e-01 -5.11179984e-01 -2.42219597e-01
-4.74206775e-01 -2.51004130e-01 1.06698431e-01 7.10383594e-01
1.31439078e+00 -2.72971123e-01 4.19284374e-01 -1.40603125e+00
-1.60116464e-01 5.50219655e-01 -5.66609383e-01 6.35179877e-01
2.16257200e-01 8.32524374e-02 7.30992317e-01 -6.05206192e-01
2.84987271e-01 5.73008478e-01 9.30903971e-01 4.92759168e-01
-9.24240112e-01 -6.95659995e-01 3.42985557e-04 2.85242200e-01
-1.46374142e+00 -4.56601614e-03 1.04146338e+00 -4.93356436e-01
5.47819078e-01 3.21909785e-01 8.26097250e-01 1.07848489e+00
3.04123759e-01 6.69046819e-01 1.03406215e+00 -3.63608181e-01
2.86817610e-01 -1.38770655e-01 3.61901104e-01 2.82968789e-01
7.08320811e-02 4.71430212e-01 -1.81325704e-01 -3.98023754e-01
7.36151755e-01 2.63325214e-01 -3.49622935e-01 2.20296428e-01
-1.08522737e+00 5.48611879e-01 2.14326233e-02 4.52580541e-01
-5.65805137e-01 -1.42849818e-01 6.01239979e-01 5.67308962e-01
4.70076650e-01 4.78457630e-01 -5.81382394e-01 -4.91663039e-01
-8.61939907e-01 4.15462017e-01 8.27349663e-01 6.27768397e-01
3.73226218e-02 3.24817061e-01 -4.12887961e-01 6.22706532e-01
1.72475219e-01 3.79365832e-01 6.07889354e-01 -6.66912556e-01
1.11985452e-01 5.75304627e-01 8.16400200e-02 -1.02700281e+00
-7.96413720e-01 -1.10198176e+00 -1.16881359e+00 3.69632319e-02
2.47578382e-01 -5.32900810e-01 -3.32866371e-01 1.67798042e+00
3.69355917e-01 5.51989019e-01 1.93375975e-01 8.41717660e-01
6.53240204e-01 6.50022805e-01 1.45244703e-01 -9.81356680e-01
1.41176534e+00 -5.55040956e-01 -7.99728513e-01 1.33075088e-01
3.81678343e-01 -6.15179241e-01 4.93682086e-01 6.34449422e-01
-8.86112869e-01 -6.39563262e-01 -9.96720672e-01 6.08342946e-01
7.58081824e-02 3.70394558e-01 3.72206926e-01 4.89490122e-01
-7.76435733e-01 6.71938181e-01 -7.63534248e-01 -2.54513830e-01
3.26514632e-01 3.55137408e-01 2.36823186e-01 3.67142856e-01
-1.44121134e+00 5.51814735e-01 3.23500395e-01 3.77405554e-01
-7.50440001e-01 -8.87438118e-01 -3.48724902e-01 -6.20030463e-02
4.62889522e-01 -7.90998161e-01 1.23247421e+00 -8.92572701e-01
-1.49136090e+00 7.03574896e-01 1.73127741e-01 -6.79404557e-01
5.68379760e-01 -2.01465175e-01 -7.39175916e-01 3.07566095e-02
-1.48069575e-01 -2.71403342e-01 7.43724465e-01 -6.39453113e-01
-7.77150631e-01 -3.75386298e-01 -1.11629060e-02 1.18384100e-01
-2.64554441e-01 1.13399521e-01 1.12577230e-01 -9.45323527e-01
-2.72553623e-01 -7.32132792e-01 -3.33090097e-01 -1.36277616e-01
-1.84155703e-01 -4.49322879e-01 7.17838585e-01 -6.94560289e-01
1.91444135e+00 -2.37863207e+00 5.38753085e-02 2.41908759e-01
4.54730421e-01 5.46166360e-01 2.23441079e-01 4.13157970e-01
-2.27212518e-01 4.38455082e-02 -2.30308428e-01 2.08079070e-01
-2.29775131e-01 7.47579038e-02 -2.36409485e-01 4.04693723e-01
1.95945933e-01 7.79343188e-01 -9.30021346e-01 -3.71314704e-01
-5.32950871e-02 3.24579775e-01 -2.03310564e-01 4.17332232e-01
5.96338212e-02 5.99829495e-01 -5.75365365e-01 5.07568598e-01
3.82614940e-01 -2.46955484e-01 1.00977726e-01 -3.65093909e-02
-9.05437842e-02 -3.60143274e-01 -1.20518410e+00 1.67743742e+00
-9.14975181e-02 1.35641145e-02 -2.57311791e-01 -1.45955217e+00
1.23488128e+00 8.04249287e-01 1.05222154e+00 -6.16779208e-01
1.95299134e-01 4.75823432e-01 2.58721977e-01 -8.50530386e-01
-2.45468035e-01 -5.16496263e-02 -4.92000282e-02 3.89001042e-01
-2.23873451e-01 5.29978275e-01 3.16244304e-01 -2.86534995e-01
1.20371401e+00 -2.57274155e-02 6.20595098e-01 -3.35794806e-01
7.15019464e-01 -3.94779176e-01 1.05830550e+00 8.09943914e-01
-3.30163747e-01 3.87707472e-01 6.55644119e-01 -8.31875682e-01
-6.42574012e-01 -9.18457270e-01 -2.11765662e-01 4.76413578e-01
-4.30840962e-02 -4.94343787e-01 -6.85360253e-01 -5.71498990e-01
-1.41379222e-01 1.51245296e-01 -3.15571398e-01 -3.12407196e-01
-7.88170159e-01 -1.20142269e+00 7.98811913e-01 5.16433835e-01
4.90087390e-01 -1.21772826e+00 -1.05104971e+00 6.73899949e-01
-1.54988781e-01 -9.60604429e-01 -1.59583837e-01 9.18200389e-02
-1.02896929e+00 -1.15030730e+00 -1.00941777e+00 -8.02710712e-01
1.56883419e-01 -4.26697314e-01 1.21153700e+00 2.07511410e-01
-6.61415219e-01 1.89545497e-01 -4.26276386e-01 -7.27243304e-01
-3.66661340e-01 2.12732077e-01 -4.84120697e-02 4.31262404e-01
1.22580037e-01 -7.95335233e-01 -9.75577116e-01 2.27818564e-01
-5.94836235e-01 -3.47028941e-01 5.92499733e-01 9.55958366e-01
7.46915579e-01 1.65222526e-01 1.36433387e+00 -8.23759377e-01
8.93035591e-01 -5.83897173e-01 -5.19344985e-01 2.98186064e-01
-1.11838567e+00 -3.10032219e-01 9.01635945e-01 -4.32256967e-01
-5.03712475e-01 -1.03718080e-01 -3.41674864e-01 -2.92558432e-01
-1.61462918e-01 5.38066566e-01 1.71242267e-01 2.19253138e-01
6.09277427e-01 4.20569062e-01 2.06613019e-01 -3.41826349e-01
-3.46734971e-01 7.55228102e-01 1.96121424e-01 -5.11249781e-01
2.82417506e-01 3.23412344e-02 4.35172953e-02 -6.22253716e-01
-5.91069221e-01 -4.07704949e-01 -1.96001917e-01 -2.61419296e-01
8.35099578e-01 -7.17031956e-01 -1.02659738e+00 7.10946381e-01
-1.10678089e+00 -2.99364366e-02 -4.79516178e-01 4.35215771e-01
-5.26252747e-01 2.01493457e-01 -6.21943474e-01 -1.06317163e+00
-1.00411534e+00 -9.04097795e-01 7.66780615e-01 1.42410070e-01
-2.73452908e-01 -8.70717824e-01 1.88619554e-01 -3.61038059e-01
4.78013933e-01 7.92601109e-01 8.95753741e-01 -7.48119414e-01
-1.98094174e-01 -1.39304370e-01 2.39246204e-01 3.99391651e-01
5.90311736e-02 -1.01729244e-01 -7.40646183e-01 -4.09549266e-01
4.39917535e-01 4.01106998e-02 4.93763745e-01 5.56384742e-01
1.32805920e+00 -1.98214605e-01 -2.61449963e-01 6.54273748e-01
1.38039446e+00 8.76181722e-01 4.96235877e-01 1.14771761e-01
3.04461002e-01 1.49946153e-01 6.81382239e-01 8.33752215e-01
2.23949566e-01 4.52599913e-01 2.84043193e-01 -3.33817631e-01
2.24537224e-01 3.81485701e-01 1.37605682e-01 9.79753196e-01
-3.33293200e-01 -9.44618359e-02 -1.04675400e+00 2.59228617e-01
-2.11003995e+00 -1.11926532e+00 -4.24785446e-03 2.14138484e+00
7.87702382e-01 2.76227415e-01 4.09568518e-01 7.27482319e-01
7.37880051e-01 -6.82150871e-02 -6.59446418e-01 -3.34122896e-01
1.26647100e-01 3.53827149e-01 -2.66298085e-01 -8.67161006e-02
-1.14356124e+00 1.09292999e-01 5.73370790e+00 8.13068509e-01
-1.41125357e+00 -8.48813565e-04 8.14952433e-01 1.82867974e-01
3.49229634e-01 -2.65729815e-01 -6.11648500e-01 7.99588859e-01
1.00287843e+00 -2.99989223e-01 3.26606065e-01 6.37638092e-01
4.58170533e-01 3.89399350e-01 -1.11350894e+00 1.38767314e+00
-3.12733464e-02 -9.16057587e-01 -2.93478668e-01 -2.79421061e-01
3.19828361e-01 -3.23368132e-01 -5.13151139e-02 4.18465078e-01
-6.03952110e-01 -6.62941277e-01 5.69908917e-01 8.05171549e-01
8.22041512e-01 -9.60373461e-01 8.10567021e-01 5.65059483e-01
-1.31970406e+00 -6.04623437e-01 8.17210078e-02 -1.74665302e-01
1.45092845e-01 1.04465294e+00 -4.63599712e-01 8.54166508e-01
5.95591605e-01 1.15560436e+00 -2.71496356e-01 1.28071856e+00
5.36839738e-02 8.04503620e-01 -1.52939782e-01 -9.96711105e-02
-1.45520955e-01 -8.74629244e-02 6.37608826e-01 1.04409647e+00
5.30115008e-01 1.95039287e-01 7.07054257e-01 7.59499550e-01
2.02579856e-01 1.61557913e-01 -5.52964807e-01 1.90335020e-01
4.73289877e-01 1.18545723e+00 -5.65584719e-01 -3.88754457e-01
-1.06023863e-01 5.70699394e-01 -1.46409914e-01 3.70368898e-01
-1.05372906e+00 -7.16186762e-01 1.71502516e-01 6.00884520e-02
-1.56018361e-01 1.58493742e-01 -3.04807335e-01 -1.01394582e+00
3.46893787e-01 -1.20536101e+00 7.14303374e-01 -2.88785398e-01
-1.24603772e+00 8.56117129e-01 -1.19514003e-01 -1.69917619e+00
-3.46878052e-01 -3.63964409e-01 -6.79421246e-01 7.33285546e-01
-1.55116582e+00 -6.99683964e-01 -2.68726826e-01 5.82616448e-01
5.48791409e-01 -2.80521810e-01 1.16587985e+00 7.38297939e-01
-8.04884136e-01 5.20744383e-01 -8.48971084e-02 2.35843524e-01
5.37548780e-01 -1.24427819e+00 -3.07570323e-02 7.30788469e-01
-1.68919668e-01 4.12586749e-01 3.62214059e-01 -3.22378546e-01
-1.05459690e+00 -1.08508849e+00 6.81415558e-01 3.76008190e-02
2.72628933e-01 -2.46893205e-02 -6.02365196e-01 6.73989430e-02
1.31363019e-01 2.10835889e-01 6.81799054e-01 -1.89591542e-01
3.85343462e-01 -7.71434247e-01 -8.96298110e-01 2.86466330e-01
8.90279830e-01 -1.85552046e-01 -2.97666043e-01 2.68150747e-01
3.62256080e-01 -4.91289914e-01 -1.21739173e+00 7.92104483e-01
6.45876169e-01 -7.89848626e-01 9.47359145e-01 -3.51274610e-01
2.88286686e-01 -1.47401631e-01 4.28914040e-01 -1.14322972e+00
-4.18566644e-01 -1.02053428e+00 -5.16003549e-01 1.19668269e+00
3.79901707e-01 -5.95231116e-01 4.63088006e-01 -2.20691692e-02
-1.73715413e-01 -1.42017269e+00 -8.62957239e-01 -7.96485603e-01
-2.74197310e-01 -2.51215369e-01 5.50753236e-01 9.14203346e-01
-2.03344703e-01 4.74611342e-01 -5.16475439e-01 4.72279489e-02
7.12788582e-01 4.16660696e-01 1.64271191e-01 -1.45403922e+00
-5.18797457e-01 -4.14410442e-01 -4.55185622e-01 -5.43712378e-01
-2.87123501e-01 -8.89178813e-01 -2.06576005e-01 -1.17451084e+00
-8.79545957e-02 -6.14881635e-01 -9.63841677e-01 4.37816471e-01
-3.24987531e-01 -7.24083334e-02 3.70768793e-02 1.58327043e-01
-4.53538686e-01 2.17481226e-01 1.03111279e+00 -8.98998827e-02
-4.68971372e-01 5.61586082e-01 -5.64592481e-01 6.82601690e-01
1.17428732e+00 -6.80454373e-01 -5.89286685e-01 -2.80571908e-01
1.26731873e-01 6.07745171e-01 3.46853971e-01 -1.28746986e+00
1.57987282e-01 -1.00520123e-02 2.92774677e-01 -4.11333859e-01
-4.80844900e-02 -8.99832785e-01 1.54670283e-01 1.00206280e+00
-2.83195555e-01 5.96638083e-01 -4.05318365e-02 5.20472169e-01
-2.67521709e-01 -5.13639115e-02 6.19677603e-01 -9.88335628e-03
-5.81525385e-01 5.42872190e-01 -3.39907497e-01 4.43330824e-01
1.27862966e+00 -1.24595463e-02 -8.49591643e-02 1.44794405e-01
-9.53856349e-01 2.17806488e-01 -4.77529675e-01 3.01413864e-01
7.72624493e-01 -1.25550902e+00 -1.04155183e+00 4.15556103e-01
6.64088726e-02 -1.37602881e-01 2.52131701e-01 1.28420448e+00
-4.35248017e-01 -5.30865416e-02 -2.06106260e-01 -8.30087602e-01
-1.08271170e+00 4.57047969e-01 4.17149007e-01 -6.62738919e-01
-7.81767488e-01 4.79461879e-01 -2.20995769e-01 8.53060409e-02
4.09437031e-01 -4.35551047e-01 -4.48698044e-01 1.10383794e-01
5.51417053e-01 5.94074786e-01 1.13379866e-01 7.39112422e-02
-4.87903655e-01 7.65384972e-01 4.32211347e-02 2.17607260e-01
1.30317271e+00 2.37979367e-01 -1.22867122e-01 5.57657778e-01
8.77622426e-01 -3.68643284e-01 -7.29017437e-01 -1.53021077e-02
1.03136413e-01 1.36888325e-01 -2.06776857e-01 -1.08640099e+00
-1.17065513e+00 8.99847150e-01 1.19242489e+00 3.77681881e-01
1.58395505e+00 -5.65418601e-01 8.58987629e-01 2.52334356e-01
3.85102242e-01 -9.24831510e-01 1.88174527e-02 2.20562473e-01
8.65714908e-01 -8.39929402e-01 -1.19305380e-01 -1.78113487e-02
-5.20755291e-01 1.27752507e+00 5.13300240e-01 -3.09716731e-01
9.34815109e-01 3.75759035e-01 3.60013954e-02 1.90292560e-02
-7.25210130e-01 1.47245139e-01 1.02710642e-01 4.00492728e-01
4.18540597e-01 -3.03524043e-02 -9.23810601e-01 1.12066412e+00
2.55354226e-01 5.34785032e-01 8.37936625e-02 1.00254083e+00
-1.82703063e-01 -1.13630652e+00 -1.03208087e-01 6.39310658e-01
-8.73077750e-01 6.62516579e-02 1.30553305e-01 5.40559053e-01
4.56600219e-01 7.66423345e-01 -2.42440954e-01 -3.68692040e-01
5.37752509e-01 8.68726447e-02 2.10271731e-01 -3.83465618e-01
-9.40451026e-01 1.81052044e-01 -2.13137001e-01 -5.06117344e-01
-4.46145117e-01 -3.89816433e-01 -1.20773053e+00 2.17170328e-01
-2.85605133e-01 3.15288156e-01 3.98211330e-01 7.80495644e-01
6.16615117e-01 1.02940297e+00 9.38249528e-01 -3.51161510e-01
-1.06100333e+00 -7.94114292e-01 -8.09085011e-01 3.69561613e-01
5.07215500e-01 -3.71546894e-01 -1.69553846e-01 3.04694306e-02] | [14.288296699523926, 3.2729508876800537] |
66bd27e8-c80c-41cc-8197-5a331463f8d2 | style-over-substance-evaluation-biases-for | 2307.03025 | null | https://arxiv.org/abs/2307.03025v1 | https://arxiv.org/pdf/2307.03025v1.pdf | Style Over Substance: Evaluation Biases for Large Language Models | As large language models (LLMs) continue to advance, accurately and comprehensively evaluating their performance becomes increasingly challenging. Conventionally, human evaluations are considered the gold standard in natural language generation. Recent advancements incorporate state-of-the-art LLMs as proxies for human judges in evaluation processes. Nonetheless, the extent to which humans and LLMs are capable evaluators remains uncertain. This study aims to investigate the behavior of both crowd-sourced human and LLM-based judges when comparing outputs from different models. To accomplish this, we curate a dataset comprising intentionally flawed machine-generated answers. Our findings indicate that despite the potentially greater danger posed by factual errors, answers with factual errors were still rated more favorably compared to answers that were too short or contained grammatical errors. This highlights a concerning bias in the evaluation process. To address this issue, we propose to independently evaluate machine-generated text across multiple dimensions, rather than merging all the evaluation aspects into a single score. We instantiate this idea with the Elo rating system, resulting in the Multi-Elo Rating System. Empirical results from our study reveal that this proposed approach significantly enhances the quality of LLM-based evaluations, particularly in terms of factual accuracy. However, notable improvement is not observed in crowd-sourced-based evaluations, suggesting the need for further investigation and refinement. | ['Alham Fikri Aji', 'Minghao Wu'] | 2023-07-06 | null | null | null | null | ['text-generation'] | ['natural-language-processing'] | [-8.78643766e-02 3.48418504e-01 1.86676607e-01 -3.60270143e-01
-1.21126366e+00 -8.06828976e-01 7.74389625e-01 5.93775451e-01
-7.21493781e-01 9.01187003e-01 4.01585281e-01 -3.59896004e-01
1.32054046e-01 -8.25839579e-01 -4.32192534e-01 -2.16891631e-01
6.54093266e-01 5.98942697e-01 -1.26247808e-01 -3.26654881e-01
6.30219638e-01 1.06018528e-01 -1.55615795e+00 4.97875065e-01
1.43725324e+00 4.73553002e-01 -1.19312905e-01 8.21635306e-01
-2.01763034e-01 8.76610875e-01 -1.45558560e+00 -1.10158801e+00
8.89749900e-02 -3.99080843e-01 -6.59927666e-01 8.91143680e-02
6.64506495e-01 -3.54713857e-01 3.72683108e-01 8.53005171e-01
8.99356425e-01 1.01693518e-01 6.52756870e-01 -1.12382853e+00
-9.76632893e-01 6.19681001e-01 -4.63602692e-02 1.51565121e-02
8.73769939e-01 5.75455308e-01 1.04690504e+00 -1.07244158e+00
6.66147172e-01 1.42142737e+00 5.77720404e-01 4.14903969e-01
-1.15971029e+00 -4.18865442e-01 -5.38450219e-02 -2.44536936e-01
-1.16837108e+00 -4.45766121e-01 3.90946120e-01 -6.93652987e-01
8.46437633e-01 1.48649752e-01 2.63679028e-01 1.14836168e+00
4.22222316e-02 5.97192526e-01 1.45874345e+00 -6.28058791e-01
3.37173194e-01 6.77124441e-01 2.82325465e-02 2.81593412e-01
6.94535375e-01 -1.64142922e-01 -6.97340250e-01 -3.66038263e-01
1.79722458e-01 -6.05763376e-01 -1.13261476e-01 2.44985998e-01
-1.19894481e+00 9.37801898e-01 8.52881372e-02 3.15765947e-01
-5.82920372e-01 -3.41239005e-01 2.34930933e-01 3.05745274e-01
8.02048385e-01 1.03497422e+00 -9.35721546e-02 -5.04017413e-01
-1.09321499e+00 7.02387750e-01 9.20281470e-01 5.15237510e-01
4.35286850e-01 -4.32377681e-02 -7.95896769e-01 9.07454967e-01
2.04376236e-01 6.73212409e-01 5.48003137e-01 -1.04040360e+00
8.80388081e-01 1.02733719e+00 6.95513964e-01 -1.21643412e+00
-8.91683251e-02 -5.68987489e-01 -3.97112489e-01 4.20225769e-01
5.60708284e-01 -2.58285344e-01 -4.43562627e-01 1.56522262e+00
1.71103776e-01 -6.39551282e-01 -3.79675999e-03 1.04346752e+00
7.19088197e-01 3.99781018e-01 3.46038967e-01 4.57060970e-02
1.21788704e+00 -9.01979625e-01 -5.99611938e-01 -2.07069010e-01
8.51721466e-01 -9.47433531e-01 1.57420945e+00 3.91397387e-01
-1.25938773e+00 -8.18839908e-01 -8.96933019e-01 1.64565876e-01
-3.11205089e-01 3.03849280e-01 2.22525492e-01 8.84104431e-01
-1.22203994e+00 4.19018626e-01 -1.07546307e-01 -3.14333647e-01
1.28691182e-01 -6.46440238e-02 -2.85108596e-01 -6.02805950e-02
-1.28529799e+00 1.14850628e+00 -4.37395610e-02 1.03293704e-02
-6.51487529e-01 -4.12742347e-01 -6.17663622e-01 -2.23605558e-01
1.31676421e-01 -6.50988817e-01 1.62554979e+00 -8.48165095e-01
-1.09764409e+00 1.01102483e+00 -4.03842032e-01 -3.20901364e-01
9.16995585e-01 -2.69530773e-01 -3.73046756e-01 -6.84451163e-02
5.76193035e-01 4.83229607e-01 4.22245294e-01 -1.47311163e+00
-6.22195482e-01 -1.79933816e-01 4.03556794e-01 3.10854048e-01
-3.07897061e-01 2.47588977e-01 2.30487987e-01 -5.07156968e-01
-3.27486813e-01 -7.29375839e-01 -1.44807786e-01 -4.14985180e-01
-3.61053556e-01 -6.20501578e-01 -1.73656456e-02 -5.27396858e-01
1.65212142e+00 -1.63464761e+00 -2.38249719e-01 -3.46529931e-02
4.43567485e-01 6.72265649e-01 -4.11799669e-01 6.24190867e-01
4.36572254e-01 6.62078321e-01 -4.86572571e-02 -7.22511828e-01
2.20201299e-01 -2.67566115e-01 -1.89762712e-01 2.13343594e-02
2.95288146e-01 1.01608694e+00 -1.11863172e+00 -6.45565569e-01
-4.51631360e-02 2.02596799e-01 -2.90950388e-01 4.07067984e-01
-1.30595893e-01 3.91902983e-01 -3.43628705e-01 6.91849411e-01
4.56680268e-01 -2.09657773e-01 -3.26123118e-01 2.96737462e-01
-2.55704671e-01 4.47325349e-01 -8.83343995e-01 1.24351096e+00
-5.94547868e-01 7.67374337e-01 -2.07251787e-01 -3.39115739e-01
1.14505577e+00 4.59765464e-01 -2.52866566e-01 -8.92578065e-01
-3.46968286e-02 5.93786001e-01 5.74700870e-02 -6.75941825e-01
9.84436452e-01 -1.29927352e-01 -1.10581934e-01 7.47862220e-01
-2.83207387e-01 -2.61606872e-01 6.57643437e-01 3.35814774e-01
1.01366675e+00 -1.44077927e-01 1.36328235e-01 7.13003725e-02
5.45335233e-01 3.04164052e-01 2.44605333e-01 1.09283924e+00
-4.44633573e-01 5.74668288e-01 4.47771490e-01 -8.70476514e-02
-9.86963809e-01 -9.23831224e-01 9.17841047e-02 1.10275233e+00
-2.93112457e-01 -5.08846819e-01 -1.05519605e+00 -7.01174796e-01
-1.05532788e-01 1.05423164e+00 -5.01522779e-01 6.45892248e-02
-3.26409101e-01 -5.84960878e-01 8.65222037e-01 3.20908606e-01
2.21287712e-01 -1.27833521e+00 -8.63109708e-01 3.32983643e-01
-6.02319181e-01 -9.06595945e-01 -2.00213566e-01 -5.87752163e-01
-7.23840177e-01 -9.11658347e-01 -9.85776722e-01 -1.98794752e-01
7.11655915e-01 2.65178531e-01 1.55379295e+00 3.30981195e-01
1.17476560e-01 3.24315935e-01 -6.58977687e-01 -6.03335202e-01
-8.91493142e-01 9.96936560e-02 -5.74182086e-02 -1.43654838e-01
5.95579684e-01 -6.69242069e-02 -6.51385546e-01 3.41740012e-01
-9.38861966e-01 -1.64068207e-01 6.10803246e-01 4.46004361e-01
3.24725322e-02 -3.01094294e-01 8.89832973e-01 -9.08731043e-01
1.64648592e+00 -4.35185015e-01 -1.62777543e-01 4.18002069e-01
-8.02883506e-01 2.46051271e-02 6.09234333e-01 -3.24838847e-01
-1.15905416e+00 -8.01962435e-01 7.55504589e-04 2.31794477e-01
-3.44849169e-01 7.41773605e-01 2.50939757e-01 1.53361261e-01
1.10374737e+00 -2.22624719e-01 -2.48452470e-01 -1.84124529e-01
1.48909062e-01 1.00405872e+00 2.86830515e-01 -7.72071779e-01
7.41076589e-01 6.35326803e-02 -5.34276187e-01 -5.26239574e-01
-1.02454579e+00 -2.38663718e-01 -2.73759305e-01 -6.39515817e-01
6.63399279e-01 -8.67779255e-01 -5.48558474e-01 3.75417382e-01
-1.59505022e+00 -2.66411990e-01 -1.27843261e-01 3.88813347e-01
-7.43842572e-02 2.44902298e-01 -5.18130720e-01 -1.20794284e+00
-4.26064610e-01 -1.24530232e+00 1.18981159e+00 2.68350810e-01
-9.40538108e-01 -9.20684934e-01 4.18668926e-01 1.01026773e+00
6.67013645e-01 3.76594633e-01 8.06048393e-01 -7.27719307e-01
-2.61079758e-01 -6.47552490e-01 -1.12549096e-01 4.26317960e-01
-2.41781905e-01 2.24547461e-01 -9.28186834e-01 -1.09975170e-02
-4.54162285e-02 -6.54382586e-01 3.46103698e-01 -8.37755483e-03
3.88897061e-01 -2.30598718e-01 1.74869090e-01 -2.34042451e-01
1.19004321e+00 -3.98815498e-02 4.96005386e-01 5.30995071e-01
3.66023213e-01 1.15498245e+00 8.60847771e-01 5.45924187e-01
6.15253747e-01 4.42013085e-01 1.12346560e-01 8.04319978e-02
-6.86278194e-02 -5.26338220e-01 4.05448169e-01 7.96966910e-01
-2.16949224e-01 -7.67706573e-01 -1.10242999e+00 5.98028541e-01
-1.66827035e+00 -7.78772771e-01 -4.16773051e-01 2.30502605e+00
6.06908441e-01 1.36525854e-01 -8.44023526e-02 2.42477164e-01
7.88837254e-01 1.35423273e-01 -2.45638981e-01 -6.65777504e-01
-3.88839066e-01 8.77786800e-02 -4.90213856e-02 4.76057589e-01
-4.84515041e-01 6.63083732e-01 6.24355459e+00 6.67176366e-01
-8.24078739e-01 1.86854854e-01 7.40706325e-01 -1.63716912e-01
-7.33293653e-01 -5.79310320e-02 -7.54184365e-01 5.53288400e-01
1.05207825e+00 -3.08010101e-01 1.48063079e-01 6.39847875e-01
6.62916064e-01 -4.02893603e-01 -8.99982750e-01 6.80449128e-01
3.28485161e-01 -8.24555695e-01 2.06420377e-01 5.72851449e-02
9.18749392e-01 -3.46030861e-01 -6.57471493e-02 5.90374351e-01
3.38674515e-01 -1.24154961e+00 1.13857675e+00 5.24655819e-01
5.03171861e-01 -4.89858896e-01 1.18827772e+00 5.77881515e-01
-6.86233044e-01 8.76812637e-02 -3.15276504e-01 -5.24404764e-01
4.50370640e-01 7.55538464e-01 -9.01507437e-01 2.69145966e-01
2.39580914e-01 4.65265550e-02 -1.11141980e+00 8.49800706e-01
-6.06219113e-01 6.33451104e-01 2.21540600e-01 -3.25933903e-01
2.45882750e-01 -1.90443229e-02 4.16373312e-01 1.13211250e+00
4.40212011e-01 -2.23813638e-01 -1.23293661e-01 8.91400695e-01
-2.18802944e-01 3.00950140e-01 -6.46538556e-01 -1.78412676e-01
7.59375751e-01 1.27056444e+00 -5.33692956e-01 -5.55143058e-01
-2.47248739e-01 7.79027700e-01 6.54668152e-01 4.65373993e-01
-7.03595340e-01 -2.37077042e-01 1.93404511e-01 2.90395260e-01
-3.81462872e-01 -4.35620733e-02 -6.30633414e-01 -9.44787979e-01
4.64581609e-01 -1.31590593e+00 -9.33125429e-03 -1.02920294e+00
-1.25144148e+00 8.76509249e-01 -1.89601168e-01 -1.12256944e+00
-3.54861915e-01 -4.43180442e-01 -6.47847772e-01 1.08314335e+00
-1.32091928e+00 -7.13234782e-01 -5.63729227e-01 7.73666874e-02
6.08029902e-01 -7.71382451e-02 6.05120182e-01 1.42522186e-01
-3.39327872e-01 7.08631337e-01 -2.59314358e-01 -3.14149559e-02
1.13786101e+00 -1.07455409e+00 3.60083997e-01 8.08276772e-01
-4.36302982e-02 8.54761004e-01 8.57863009e-01 -7.69240499e-01
-5.47919512e-01 -8.14443052e-01 1.68605816e+00 -9.10821676e-01
4.46934223e-01 -6.08560182e-02 -7.71254599e-01 -8.91524949e-04
2.28767425e-01 -5.94206333e-01 7.84960449e-01 2.34852452e-02
-2.07507506e-01 2.70146966e-01 -1.16674435e+00 6.45442903e-01
8.22802722e-01 -8.32199156e-01 -7.39661694e-01 2.17707798e-01
5.29775143e-01 -1.00724138e-01 -7.65612543e-01 2.84303755e-01
4.69613194e-01 -1.30082262e+00 5.61186671e-01 -5.19887209e-01
9.42327082e-01 -2.40047306e-01 -2.04599909e-02 -1.51938295e+00
4.83132675e-02 -2.89291203e-01 1.32971987e-01 1.60807788e+00
8.02656889e-01 -6.02748692e-01 5.68081498e-01 1.16176426e+00
-4.91145961e-02 -6.79155707e-01 -5.88393331e-01 -5.96356928e-01
1.37299612e-01 -4.47661817e-01 5.79115093e-01 7.47543514e-01
-5.71154729e-02 3.20619613e-01 -2.05447942e-01 -1.52610973e-01
3.91323537e-01 -1.61545292e-01 9.82493281e-01 -1.30819929e+00
-1.42500162e-01 -5.14981925e-01 -8.99527594e-02 -5.65935493e-01
1.78952307e-01 -6.49996519e-01 1.70642138e-01 -1.85564673e+00
1.82340190e-01 -2.82978714e-01 6.44853413e-02 -7.06407949e-02
-7.05097675e-01 2.71282732e-01 4.34196770e-01 3.60496521e-01
-7.17271984e-01 3.73140991e-01 1.34184003e+00 9.24469605e-02
-8.50913152e-02 -4.77341525e-02 -9.50339973e-01 5.96405983e-01
9.16030526e-01 -5.17863452e-01 -2.82993078e-01 -7.61829317e-01
6.82412624e-01 -4.39943932e-02 3.74777555e-01 -1.10722351e+00
2.61076123e-01 -1.03232767e-02 1.39810443e-01 -1.93008944e-01
4.36889641e-02 -2.47205377e-01 -9.52061415e-02 1.90573290e-01
-7.37708926e-01 5.05409598e-01 -1.36161208e-01 2.14588150e-01
-5.73642910e-01 -5.90626419e-01 2.40792587e-01 -2.83230931e-01
-1.36900559e-01 -2.17561290e-01 -5.56809604e-01 3.32363963e-01
7.89408743e-01 -3.94333392e-01 -6.64152980e-01 -6.01305664e-01
-1.18210182e-01 8.38068947e-02 6.84943855e-01 3.76876056e-01
3.77721578e-01 -1.15694475e+00 -1.18575728e+00 -3.11440796e-01
4.78941113e-01 -9.95381176e-02 1.76450133e-01 6.77613497e-01
-4.93725389e-01 7.47658372e-01 1.93639547e-01 -2.41402671e-01
-1.00184035e+00 8.09508637e-02 2.73251534e-02 -7.04363167e-01
2.21080184e-01 7.14247048e-01 -1.57096595e-01 -6.63236797e-01
1.59394130e-01 -1.55418888e-01 -2.83836901e-01 3.64572853e-01
5.49766362e-01 8.71586084e-01 1.43650949e-01 -7.58823812e-01
3.15049551e-02 2.63416946e-01 2.14770466e-01 -5.61986625e-01
7.33761907e-01 -2.11708784e-01 -7.88664147e-02 5.85711539e-01
7.50589669e-01 4.23520058e-01 -7.96565294e-01 3.46837938e-02
1.66372716e-01 -4.88729209e-01 -3.10779303e-01 -1.04714406e+00
-4.30087775e-01 1.03667498e+00 1.23941965e-01 3.16103518e-01
6.07183337e-01 -2.00575873e-01 5.76713264e-01 3.32596660e-01
5.75827479e-01 -1.28532481e+00 2.93855399e-01 4.38434243e-01
9.84633267e-01 -1.59532320e+00 -3.13466102e-01 3.35017592e-02
-9.38748658e-01 7.92962134e-01 8.77373338e-01 1.92219764e-01
-1.87218204e-01 -2.40487441e-01 5.44158816e-01 -1.09966151e-01
-8.63690615e-01 -1.59305185e-02 3.22701186e-01 4.78566259e-01
1.11024916e+00 2.72487640e-01 -7.53051341e-01 6.36037707e-01
-4.88577545e-01 5.81338108e-02 7.07210362e-01 7.62000203e-01
-4.06642228e-01 -1.06082129e+00 -7.43565857e-01 4.55451220e-01
-4.75634962e-01 -2.39878908e-01 -7.65128374e-01 5.43258309e-01
5.41722253e-02 1.73802769e+00 -2.78365672e-01 -3.30150425e-01
5.89480519e-01 1.96552619e-01 -1.38414772e-02 -7.89570153e-01
-1.11292887e+00 -5.52693725e-01 4.72857088e-01 -3.47262025e-01
-5.10594785e-01 -3.98285180e-01 -7.11718500e-01 -4.15097147e-01
-3.39246929e-01 4.93336082e-01 7.22607017e-01 9.14598823e-01
4.21836674e-01 2.37892807e-01 4.26787674e-01 -6.28436387e-01
-9.30690944e-01 -1.35229301e+00 -1.34010479e-01 8.07079136e-01
1.92000329e-01 -4.64141637e-01 -6.93514228e-01 -2.57042468e-01] | [11.772787094116211, 8.853104591369629] |
fd94cc55-c384-4dde-9c8d-160a25f2c01d | freepoint-unsupervised-point-cloud-instance | 2305.06973 | null | https://arxiv.org/abs/2305.06973v1 | https://arxiv.org/pdf/2305.06973v1.pdf | FreePoint: Unsupervised Point Cloud Instance Segmentation | Instance segmentation of point clouds is a crucial task in 3D field with numerous applications that involve localizing and segmenting objects in a scene. However, achieving satisfactory results requires a large number of manual annotations, which is a time-consuming and expensive process. To alleviate dependency on annotations, we propose a method, called FreePoint, for underexplored unsupervised class-agnostic instance segmentation on point clouds. In detail, we represent the point features by combining coordinates, colors, normals, and self-supervised deep features. Based on the point features, we perform a multicut algorithm to segment point clouds into coarse instance masks as pseudo labels, which are used to train a point cloud instance segmentation model. To alleviate the inaccuracy of coarse masks during training, we propose a weakly-supervised training strategy and corresponding loss. Our work can also serve as an unsupervised pre-training pretext for supervised semantic instance segmentation with limited annotations. For class-agnostic instance segmentation on point clouds, FreePoint largely fills the gap with its fully-supervised counterpart based on the state-of-the-art instance segmentation model Mask3D and even surpasses some previous fully-supervised methods. When serving as a pretext task and fine-tuning on S3DIS, FreePoint outperforms training from scratch by 5.8% AP with only 10% mask annotations. | ['Gui-Song Xia', 'Dengxin Dai', 'Li Jiang', 'Jian Ding', 'Zhikai Zhang'] | 2023-05-11 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 2.32977316e-01 3.16497922e-01 -1.78641200e-01 -6.21652305e-01
-8.90809774e-01 -8.41274083e-01 5.36084175e-01 2.25485861e-01
-2.59808242e-01 2.82283932e-01 -6.58036470e-01 -2.64501184e-01
1.22056909e-01 -8.03716898e-01 -1.00386369e+00 -4.65650737e-01
1.20892137e-01 1.03117073e+00 6.09194040e-01 -3.84468073e-03
3.64739478e-01 8.29989016e-01 -1.60982180e+00 1.20919608e-01
1.11142731e+00 1.06963241e+00 3.39336395e-01 1.22859523e-01
-8.77071321e-01 -8.32046792e-02 -4.87047493e-01 -2.58467525e-01
6.07378423e-01 4.59016897e-02 -9.17873740e-01 5.82507312e-01
4.98314202e-01 -6.12720214e-02 3.66293848e-01 1.06389558e+00
4.92917337e-02 -1.95498280e-02 7.41188884e-01 -1.29071248e+00
-1.84040651e-01 2.18632087e-01 -8.49562347e-01 -3.30650866e-01
-1.26934156e-01 2.93109298e-01 8.76037061e-01 -9.60038841e-01
2.42211699e-01 1.07243061e+00 8.71418774e-01 4.12492573e-01
-1.13550758e+00 -7.05404580e-01 4.01913345e-01 -3.63012552e-01
-1.54217041e+00 -2.86261272e-02 8.94685507e-01 -6.25662923e-01
7.94730067e-01 3.28705043e-01 6.93380833e-01 4.37038302e-01
-5.31605422e-01 8.88462484e-01 1.00693071e+00 -1.35548800e-01
3.30990493e-01 2.06088364e-01 2.37134457e-01 7.00360119e-01
2.36791931e-02 -1.18382901e-01 1.21261045e-01 -7.18050152e-02
1.02025139e+00 2.97052324e-01 9.59270597e-02 -6.94949508e-01
-1.16757333e+00 7.36004293e-01 7.62577832e-01 -1.98961888e-02
-1.22037962e-01 8.85983482e-02 1.63423657e-01 -2.58537997e-02
7.85826862e-01 4.86487925e-01 -8.62638831e-01 9.68237221e-02
-1.24812162e+00 4.97198887e-02 5.17985344e-01 1.19406557e+00
1.44277310e+00 -3.86908174e-01 2.34711170e-02 8.30035388e-01
2.65028298e-01 4.83633488e-01 1.03585847e-01 -9.27088916e-01
4.16864663e-01 1.26349747e+00 8.93755779e-02 -6.12715960e-01
-3.84449869e-01 -6.18138969e-01 -5.81964910e-01 4.83746678e-01
3.29171330e-01 1.33962452e-01 -1.43024600e+00 9.61081564e-01
7.30361581e-01 4.68343556e-01 -3.16950470e-01 9.92526531e-01
7.02799737e-01 5.82979143e-01 8.76744092e-02 2.00625688e-01
1.19997585e+00 -1.03825259e+00 3.27407499e-03 -2.91998804e-01
8.00977647e-01 -6.28892243e-01 1.36984324e+00 4.17230695e-01
-8.50845993e-01 -5.97893417e-01 -8.82798970e-01 -7.86113366e-02
-4.68193591e-01 1.61550492e-01 8.66875827e-01 5.55689573e-01
-7.55078793e-01 7.07550347e-01 -1.07079363e+00 -2.17676148e-01
9.40522015e-01 5.48694730e-01 -2.54604876e-01 7.16584325e-02
-4.99996454e-01 3.78838062e-01 5.34076691e-01 -6.80001006e-02
-6.23366177e-01 -1.05166245e+00 -7.52072871e-01 -1.41940907e-01
5.83341718e-01 -5.82140803e-01 1.02588892e+00 -8.93857837e-01
-1.43457842e+00 1.28969085e+00 2.17264052e-02 -2.91607529e-01
5.20199955e-01 -3.37317556e-01 1.21535704e-01 -2.09108870e-02
3.15579325e-01 9.71055150e-01 8.18265498e-01 -1.72773445e+00
-7.44704604e-01 -4.99084681e-01 1.84755191e-01 2.02051014e-01
1.24033011e-01 -4.37036842e-01 -9.99970794e-01 -3.99072438e-01
6.92291677e-01 -9.91836131e-01 -5.16933262e-01 1.23691849e-01
-7.21943855e-01 -3.81089628e-01 1.05437517e+00 -5.45455478e-02
6.70266569e-01 -2.30010605e+00 -1.65390313e-01 4.65315610e-01
1.45273581e-01 2.21078381e-01 7.53137469e-02 -5.81678972e-02
-2.09607128e-02 2.66950697e-01 -9.96371090e-01 -6.23452723e-01
1.87235802e-01 4.24973428e-01 -2.02687010e-01 4.72915441e-01
5.46622753e-01 8.08382750e-01 -9.47848797e-01 -6.94952846e-01
5.96083999e-01 3.14674914e-01 -6.32831216e-01 1.49188042e-01
-7.48011351e-01 7.25793242e-01 -6.10341489e-01 8.94792378e-01
9.93387103e-01 -4.69263464e-01 -4.60706055e-01 -1.95574120e-01
-1.83766782e-01 -5.91752976e-02 -1.16816139e+00 2.15303111e+00
-3.84992033e-01 1.57946691e-01 -5.41030429e-02 -1.13853848e+00
1.13491058e+00 -8.04903358e-02 6.92897260e-01 -1.71483785e-01
-1.55460760e-02 3.48238081e-01 -4.85980779e-01 -2.31585130e-01
2.29134962e-01 -9.97397229e-02 -9.95436236e-02 2.11324573e-01
-5.70654236e-02 -8.15534532e-01 -8.21783543e-02 8.01894441e-02
8.00374031e-01 5.06418705e-01 -1.17492497e-01 -9.69259143e-02
4.79557872e-01 4.69340205e-01 4.12427604e-01 5.19275010e-01
9.81800482e-02 1.14607048e+00 3.41588974e-01 -3.39771032e-01
-8.95335615e-01 -9.54972148e-01 -5.52277148e-01 7.72836268e-01
5.18133163e-01 -2.41239265e-01 -8.49338174e-01 -1.03876388e+00
2.28810564e-01 4.31139588e-01 -3.85977566e-01 2.09285885e-01
-3.00181389e-01 -6.87473893e-01 3.32175612e-01 5.24949431e-01
5.94849348e-01 -8.64516735e-01 -2.64987141e-01 3.00992094e-03
2.39731193e-01 -1.24598098e+00 -2.23082215e-01 4.09414172e-01
-1.16561806e+00 -1.07816708e+00 -7.57650077e-01 -7.04172969e-01
1.10899651e+00 3.16343814e-01 1.30092454e+00 3.75380367e-01
-6.51156753e-02 2.59752572e-01 -3.74098241e-01 -3.64338309e-01
4.31365743e-02 3.41577828e-01 -3.55456352e-01 -4.93493155e-02
5.22122800e-01 -6.59419596e-01 -7.21151412e-01 4.37530309e-01
-8.81191134e-01 1.47372767e-01 7.01916814e-01 4.21562105e-01
1.20519936e+00 -3.23697180e-02 8.95382091e-02 -1.39230597e+00
-1.08061552e-01 -4.83764708e-01 -8.56288075e-01 -7.67537057e-02
-2.83212662e-01 -7.89436549e-02 2.86325693e-01 -1.21996514e-01
-8.29264224e-01 6.02905393e-01 -2.90958583e-01 -8.22798073e-01
-5.99115849e-01 2.02025726e-01 -3.15301389e-01 -2.85380960e-01
5.67663372e-01 -9.58007202e-02 -2.15778157e-01 -8.05948138e-01
5.97022474e-01 5.59559107e-01 4.97328222e-01 -7.80354559e-01
1.22302806e+00 6.88618243e-01 2.12051030e-02 -7.09562421e-01
-1.13121700e+00 -8.90442908e-01 -1.17623985e+00 1.39899224e-01
1.00234020e+00 -9.48573530e-01 -3.44808131e-01 1.70985430e-01
-1.17727804e+00 -5.54546058e-01 -4.03318286e-01 1.09335504e-01
-5.10460973e-01 3.44121933e-01 -1.44456640e-01 -6.49271369e-01
-1.63920000e-02 -1.07372141e+00 1.61568034e+00 6.02872297e-03
6.55863509e-02 -8.70917976e-01 -1.72762528e-01 5.11986494e-01
-1.39522925e-01 4.68848765e-01 7.29661524e-01 -7.73587525e-01
-1.03579879e+00 -3.09211195e-01 -4.99292880e-01 5.01849771e-01
1.74128607e-01 6.02447987e-02 -1.01055670e+00 2.58015580e-02
-1.66382447e-01 -1.67576283e-01 6.96354508e-01 2.38239020e-01
1.76437759e+00 9.58786160e-02 -5.96252024e-01 1.04127634e+00
1.52277637e+00 -1.79359049e-01 3.11917454e-01 2.47251287e-01
1.17563415e+00 4.91569370e-01 8.66540790e-01 4.12696242e-01
3.61547083e-01 4.33897018e-01 7.28431046e-01 -5.14663398e-01
1.46020040e-01 -2.34124199e-01 -3.16302150e-01 4.83538270e-01
-7.77969137e-02 1.01539291e-01 -1.14912665e+00 6.18202150e-01
-1.75337052e+00 -3.72820646e-01 -4.00948316e-01 2.16535783e+00
8.20666254e-01 4.57464069e-01 1.76142544e-01 2.38608569e-01
6.67256236e-01 -1.42462105e-01 -7.15105951e-01 1.02014519e-01
1.91237226e-01 4.00360972e-01 6.58407509e-01 4.15160328e-01
-1.39836252e+00 1.34430468e+00 4.97321272e+00 9.67950106e-01
-1.10167861e+00 5.73960580e-02 7.11422741e-01 1.44632801e-01
-1.50143012e-01 1.63271904e-01 -7.15189993e-01 5.50262213e-01
2.85757333e-01 4.89372224e-01 2.28278991e-03 1.21188879e+00
8.15230235e-02 5.48128225e-02 -1.21928537e+00 1.06093681e+00
-2.00325683e-01 -1.28934515e+00 4.02352102e-02 1.05151413e-02
8.78593087e-01 3.11649680e-01 -8.42567235e-02 3.32388967e-01
2.60643244e-01 -9.25321162e-01 5.54250240e-01 2.17805102e-01
9.17447627e-01 -6.77525222e-01 5.33186018e-01 5.25159299e-01
-1.19964516e+00 2.83468336e-01 -4.68706161e-01 1.65022183e-02
4.44143005e-02 8.95732641e-01 -6.73443019e-01 4.97732937e-01
7.01988995e-01 7.70200431e-01 -4.84644115e-01 1.19179738e+00
-2.37064824e-01 6.16810501e-01 -7.16912091e-01 3.89928758e-01
5.74796379e-01 -4.30401951e-01 3.27544183e-01 1.17411005e+00
9.03168172e-02 3.30569178e-01 6.75550282e-01 1.09947205e+00
-7.52213225e-02 -4.06794176e-02 -4.25065041e-01 2.22144365e-01
4.62774128e-01 1.35649502e+00 -1.22381747e+00 -4.50322866e-01
-2.28158176e-01 9.90245938e-01 2.14104950e-01 2.51112342e-01
-8.07030499e-01 -4.25117642e-01 5.55257320e-01 3.80499721e-01
3.29913288e-01 -4.09290016e-01 -8.16448092e-01 -9.09888566e-01
4.10952829e-02 -3.15174937e-01 -9.40296054e-03 -6.40511513e-01
-1.28794897e+00 5.08708715e-01 1.47700071e-01 -1.63294029e+00
2.01668352e-01 -4.97636318e-01 -6.87708735e-01 6.85977876e-01
-1.84841812e+00 -1.33661485e+00 -6.36848569e-01 5.33770502e-01
6.91270113e-01 1.84746057e-01 3.85219365e-01 3.15393746e-01
-3.92532140e-01 2.89336950e-01 -1.39091104e-01 1.43756568e-01
3.76198739e-01 -1.60085821e+00 6.17434442e-01 4.38966334e-01
1.55239895e-01 4.62887228e-01 4.15291369e-01 -5.62819719e-01
-1.05184913e+00 -1.45501149e+00 3.31909239e-01 -7.05316484e-01
4.40866441e-01 -4.55722690e-01 -1.13927579e+00 6.04880273e-01
-4.82575119e-01 3.73177826e-01 4.62361455e-01 6.82884604e-02
1.68607738e-02 1.11238211e-02 -1.25639534e+00 2.89875418e-01
1.40370750e+00 -2.64801264e-01 -5.96362352e-01 7.80550122e-01
9.75322366e-01 -7.29709566e-01 -7.52963662e-01 5.39478242e-01
-1.52363122e-01 -9.14216459e-01 1.15876079e+00 -2.03609779e-01
3.03177595e-01 -6.31575704e-01 1.22015618e-01 -1.06363440e+00
-5.25370911e-02 -5.19217730e-01 3.01434904e-01 1.26824665e+00
4.12778139e-01 -4.67812061e-01 1.29078615e+00 5.85264206e-01
-6.50673687e-01 -7.72058725e-01 -7.80414462e-01 -7.42109716e-01
3.79705057e-02 -8.16015303e-01 9.17413771e-01 1.12697852e+00
-5.72599411e-01 1.11879937e-01 4.03966010e-01 5.53886831e-01
8.47420990e-01 3.26619267e-01 1.17029381e+00 -1.56734550e+00
6.87116832e-02 -4.28247452e-01 -4.37176019e-01 -1.41954863e+00
2.15499178e-01 -1.05096948e+00 1.24860324e-01 -1.57694244e+00
-2.36676529e-01 -1.28119302e+00 -2.52254605e-02 7.22028852e-01
-6.50225133e-02 5.99256158e-01 4.38045040e-02 4.65874076e-01
-7.30944276e-01 4.54127073e-01 1.28540409e+00 -2.34190285e-01
-4.48149174e-01 3.07948232e-01 -3.64331663e-01 9.87879276e-01
6.98598385e-01 -5.05347788e-01 -4.32394952e-01 -5.19375801e-01
-3.72606330e-02 -2.58961350e-01 4.99253094e-01 -1.14034331e+00
1.08168691e-01 -2.39406824e-01 2.72815764e-01 -1.15976763e+00
5.46122909e-01 -1.17898965e+00 3.03462744e-02 2.79626120e-02
7.58925453e-02 -4.52721566e-01 1.65983036e-01 3.98326874e-01
-1.84669510e-01 -3.28329921e-01 6.70509815e-01 -4.10880178e-01
-7.92727292e-01 7.25770473e-01 3.58979046e-01 -4.03188877e-02
1.09903467e+00 -7.28008687e-01 1.26325727e-01 3.23481500e-01
-7.87316561e-01 5.26094139e-01 8.24717462e-01 1.22482158e-01
4.90117848e-01 -1.00241029e+00 -3.89299214e-01 3.73068899e-01
2.10529014e-01 1.19253743e+00 1.38995528e-01 4.82824326e-01
-7.79830217e-01 2.01134011e-01 1.22466736e-01 -1.33845592e+00
-8.49056840e-01 4.21180069e-01 2.18505606e-01 1.11714631e-01
-8.52397084e-01 9.25006568e-01 5.17596483e-01 -7.56448746e-01
2.40780801e-01 -6.13097370e-01 -1.53267067e-02 -2.32285216e-01
-9.55417529e-02 6.96463585e-02 2.50493407e-01 -4.83813584e-01
-3.97852242e-01 1.06153226e+00 7.52819702e-02 2.50317574e-01
1.32158720e+00 7.86822811e-02 -4.66284864e-02 4.90547270e-01
1.13386917e+00 -1.95857376e-01 -1.64812112e+00 -1.77426070e-01
1.78742230e-01 -6.33885503e-01 1.35262445e-01 -6.85032547e-01
-1.25310409e+00 9.80759740e-01 4.30142134e-01 2.13763505e-01
8.10492873e-01 3.49555641e-01 8.28710616e-01 3.30050409e-01
4.50729638e-01 -7.91648865e-01 -1.42335445e-01 4.44159180e-01
6.72464371e-01 -1.56594872e+00 3.25191543e-02 -9.02115345e-01
-4.70106542e-01 9.01047230e-01 7.61393487e-01 -4.23990309e-01
7.68801272e-01 1.77227408e-02 2.16220483e-01 -4.56458032e-01
-7.73029700e-02 -4.06715780e-01 5.17455041e-01 6.63394570e-01
1.93189472e-01 -1.14752222e-02 9.71570238e-02 5.42413414e-01
-3.90148371e-01 -1.22442529e-01 -1.26155958e-01 8.04357350e-01
-5.63402176e-01 -9.95535195e-01 -5.69366038e-01 4.90025759e-01
-1.60909649e-02 1.11797117e-01 -3.35793614e-01 9.21551049e-01
4.21474606e-01 5.29400051e-01 4.44020122e-01 -2.56177753e-01
5.04582167e-01 -1.02003008e-01 1.94465026e-01 -1.06686080e+00
-5.47281206e-01 1.67058595e-02 -3.62006992e-01 -4.90371138e-01
-4.94641006e-01 -3.03899974e-01 -1.68580365e+00 -1.06301354e-02
-4.78322208e-01 1.29314154e-01 8.14803839e-01 1.08753729e+00
3.67739886e-01 3.80743295e-01 7.61783183e-01 -1.42442441e+00
-1.57470986e-01 -6.74702525e-01 -3.95705134e-01 6.14677727e-01
2.47083262e-01 -7.67218888e-01 -4.33196247e-01 -1.81220484e-03] | [8.028698921203613, -3.1820857524871826] |
5857fbf8-04f5-4f45-bcc8-1a4a7fa2c34f | lmd-a-learnable-mask-network-to-detect | 2211.00825 | null | https://arxiv.org/abs/2211.00825v2 | https://arxiv.org/pdf/2211.00825v2.pdf | LMD: A Learnable Mask Network to Detect Adversarial Examples for Speaker Verification | Although the security of automatic speaker verification (ASV) is seriously threatened by recently emerged adversarial attacks, there have been some countermeasures to alleviate the threat. However, many defense approaches not only require the prior knowledge of the attackers but also possess weak interpretability. To address this issue, in this paper, we propose an attacker-independent and interpretable method, named learnable mask detector (LMD), to separate adversarial examples from the genuine ones. It utilizes score variation as an indicator to detect adversarial examples, where the score variation is the absolute discrepancy between the ASV scores of an original audio recording and its transformed audio synthesized from its masked complex spectrogram. A core component of the score variation detector is to generate the masked spectrogram by a neural network. The neural network needs only genuine examples for training, which makes it an attacker-independent approach. Its interpretability lies that the neural network is trained to minimize the score variation of the targeted ASV, and maximize the number of the masked spectrogram bins of the genuine training examples. Its foundation is based on the observation that, masking out the vast majority of the spectrogram bins with little speaker information will inevitably introduce a large score variation to the adversarial example, and a small score variation to the genuine example. Experimental results with 12 attackers and two representative ASV systems show that our proposed method outperforms five state-of-the-art baselines. The extensive experimental results can also be a benchmark for the detection-based ASV defenses. | ['Kunde Yang', 'Wei-Qiang Zhang', 'Xiao-Lei Zhang', 'Jie Wang', 'Xing Chen'] | 2022-11-02 | null | null | null | null | ['speaker-verification'] | ['speech'] | [ 2.90743977e-01 1.39345601e-01 2.00989962e-01 -3.14060807e-01
-1.04904270e+00 -9.29390609e-01 3.62920523e-01 -3.90141040e-01
3.70900445e-02 4.78194445e-01 2.14325890e-01 -4.61613029e-01
1.98815674e-01 -3.80365849e-01 -4.60456491e-01 -9.36569929e-01
-1.42546207e-01 -1.80176690e-01 1.14653662e-01 -4.07407671e-01
-2.24176776e-02 3.34955841e-01 -1.10430598e+00 3.46689701e-01
7.00936794e-01 1.09686732e+00 -1.51351646e-01 6.36269450e-01
5.13144545e-02 4.70174640e-01 -1.37061894e+00 -3.65318924e-01
4.43063974e-01 -5.68175316e-01 -3.91570151e-01 -2.72529185e-01
4.43557084e-01 -3.11520696e-01 -3.84652436e-01 1.41582072e+00
7.59002984e-01 -1.91812038e-01 5.07960021e-01 -1.44767129e+00
-6.94530010e-01 9.32070553e-01 -3.83722633e-01 2.70759642e-01
3.60885233e-01 2.42946088e-01 6.56148314e-01 -8.71190548e-01
-9.26304460e-02 1.51781166e+00 5.73005497e-01 9.55487788e-01
-9.47528362e-01 -1.15292013e+00 2.80078620e-01 2.59334326e-01
-1.56276977e+00 -7.32711673e-01 1.12803483e+00 -1.69064030e-01
3.73052746e-01 8.07894289e-01 9.95183215e-02 1.20978105e+00
-6.76523522e-02 5.60149372e-01 9.30523992e-01 -3.66861969e-01
2.33847514e-01 4.11300898e-01 1.57806620e-01 4.18933153e-01
-4.99444269e-02 4.31149095e-01 -3.87776256e-01 -5.62049747e-01
2.40035713e-01 -2.96374917e-01 -8.05601776e-01 2.50300378e-01
-7.03245640e-01 8.92785847e-01 3.94490212e-01 1.27490520e-01
-1.26514107e-01 1.43338442e-02 3.38259161e-01 3.66146803e-01
1.80831790e-01 1.94649488e-01 -2.96653450e-01 1.64743036e-01
-6.47780836e-01 7.06895143e-02 6.70212865e-01 3.76554608e-01
4.96453434e-01 7.79148161e-01 -9.05041993e-02 5.25447726e-01
5.59419811e-01 7.21455276e-01 5.94072282e-01 -5.06566346e-01
7.15207219e-01 4.01453704e-01 -3.77788395e-02 -1.12130773e+00
2.18479738e-01 -3.65820259e-01 -7.44286001e-01 6.48670197e-01
3.63571405e-01 -4.30300802e-01 -7.68293083e-01 2.04204655e+00
3.94531101e-01 5.67136586e-01 2.51658082e-01 9.61633980e-01
6.72144711e-01 7.46983230e-01 -2.22259924e-01 -4.01303917e-01
1.22253716e+00 -5.40187180e-01 -9.04512167e-01 -3.77745450e-01
1.87444568e-01 -9.40092981e-01 1.09359550e+00 4.11727607e-01
-7.65656888e-01 -6.73014343e-01 -1.62384462e+00 6.26346350e-01
-1.76784739e-01 -1.05094664e-01 8.11778530e-02 1.49567080e+00
-7.27778137e-01 1.78987548e-01 -3.82436961e-01 3.40078026e-01
1.07432641e-01 3.76102507e-01 -1.66280702e-01 4.53923255e-01
-1.64655995e+00 6.17584407e-01 3.68951827e-01 3.27415973e-01
-1.16099393e+00 -4.84804332e-01 -7.73855686e-01 1.94222823e-01
1.90690473e-01 6.19592406e-02 1.20897806e+00 -1.18891609e+00
-1.60971773e+00 4.48612928e-01 -7.32782930e-02 -7.01226354e-01
5.05417585e-01 -1.07589830e-02 -9.59508598e-01 5.49697019e-02
-3.25902700e-01 -2.38689613e-02 1.51042914e+00 -1.36701477e+00
-3.17917407e-01 -2.17843160e-01 -1.11204252e-01 -6.41269162e-02
-5.63038826e-01 3.29773933e-01 1.62182510e-01 -8.67945790e-01
8.35081488e-02 -8.34767342e-01 7.66469240e-02 -3.08190465e-01
-6.73790097e-01 1.43360972e-01 1.31719255e+00 -8.70530009e-01
1.40075827e+00 -2.72107458e+00 -2.12718070e-01 3.80668968e-01
-2.58824304e-02 6.84124351e-01 -1.03093691e-01 1.77032828e-01
-4.35561895e-01 2.25503176e-01 -4.09785658e-01 -2.27646708e-01
1.47574738e-01 -1.15741350e-01 -1.01142442e+00 5.26660323e-01
2.46422574e-01 2.90681392e-01 -6.51156425e-01 -1.67324185e-01
-4.92612878e-03 5.45949221e-01 -2.62615472e-01 4.14099455e-01
1.43891767e-01 4.02592391e-01 -2.79220581e-01 3.60973597e-01
1.07300723e+00 4.49420184e-01 6.16627187e-02 -7.49844909e-02
1.76326573e-01 3.94903034e-01 -1.58882856e+00 8.52436066e-01
-1.59202471e-01 5.92589200e-01 5.89841068e-01 -9.32038546e-01
1.17319810e+00 7.36100793e-01 -1.11385114e-01 -1.65245026e-01
1.57459587e-01 2.24477112e-01 3.33569586e-01 -2.89797157e-01
-4.05046232e-02 -7.34762028e-02 -2.02125430e-01 3.39106858e-01
-2.77999759e-01 3.75748128e-02 -5.89892685e-01 1.90256804e-01
8.68361712e-01 -5.50490916e-01 3.56117725e-01 -4.80931923e-02
1.25905824e+00 -6.79357648e-01 8.62788320e-01 7.48514533e-01
-5.44484019e-01 4.48739409e-01 3.74011487e-01 -2.02150509e-01
-5.01315415e-01 -1.33610797e+00 -1.06158137e-01 7.33404160e-01
1.88648507e-01 -1.76376969e-01 -1.13997555e+00 -9.42390323e-01
-1.20920286e-01 7.14476883e-01 -4.54343587e-01 -5.20463884e-01
-7.75803626e-01 -4.27922726e-01 1.13958251e+00 1.96865588e-01
5.70916355e-01 -1.01044762e+00 1.14941977e-01 1.07676409e-01
-2.67144948e-01 -1.05934572e+00 -8.46216619e-01 -1.80978999e-02
-3.18450153e-01 -9.03960884e-01 -1.30585074e-01 -7.84442425e-01
5.04302621e-01 3.02508056e-01 5.32916784e-01 1.91776589e-01
-1.85131263e-02 -4.78701070e-02 -2.33588204e-01 -7.97626257e-01
-1.00913930e+00 -3.75078589e-01 5.83421111e-01 6.40829742e-01
2.77343661e-01 -5.39168835e-01 -4.05364364e-01 5.99516809e-01
-8.20578396e-01 -5.49549460e-01 3.36222798e-01 7.65406311e-01
2.59618282e-01 1.84379041e-01 9.72283423e-01 -4.15769309e-01
7.89701521e-01 -3.15971166e-01 -5.28665423e-01 9.72106308e-02
-2.90931106e-01 -4.72409278e-02 9.81938541e-01 -9.20462489e-01
-9.77268636e-01 -4.86000217e-02 -3.25300783e-01 -4.01537240e-01
-1.72326759e-01 2.42872491e-01 -8.99105310e-01 -2.29152396e-01
9.53881145e-01 4.90953714e-01 -1.72613394e-02 -3.13346535e-01
2.31037766e-01 1.12317038e+00 6.92757845e-01 -2.64571935e-01
1.48474455e+00 2.81128109e-01 -4.09312636e-01 -8.37472677e-01
-4.69302773e-01 -1.94860518e-01 -1.23448767e-01 -1.60514265e-01
5.86081684e-01 -6.28142357e-01 -7.96015680e-01 5.80199301e-01
-1.37927401e+00 2.13974401e-01 7.16718063e-02 4.31048453e-01
-7.64992237e-02 6.90459549e-01 -4.71063375e-01 -1.34428561e+00
-6.57638252e-01 -1.34261680e+00 7.59144783e-01 1.89727262e-01
-2.40962744e-01 -6.24761403e-01 -1.07162736e-01 3.73309910e-01
3.03080708e-01 3.09965253e-01 7.33348668e-01 -1.11181438e+00
-4.29349840e-01 -6.15817845e-01 2.66414702e-01 8.65766704e-01
4.48647201e-01 -4.86035347e-02 -1.53085792e+00 -3.96857440e-01
6.63775027e-01 -1.11323677e-01 4.66189474e-01 1.21136807e-01
1.06139910e+00 -8.83964062e-01 1.45793902e-02 6.18472099e-01
1.00788260e+00 5.66534162e-01 5.89300275e-01 4.12720665e-02
5.69243193e-01 4.33059782e-01 3.82022768e-01 1.66086644e-01
-3.15120071e-01 8.27705979e-01 6.77622795e-01 6.81645125e-02
1.81877762e-02 -2.56711245e-01 1.03612614e+00 8.70563865e-01
2.93083221e-01 -2.26592600e-01 -4.81079191e-01 2.07324252e-01
-1.36328614e+00 -1.28920412e+00 -2.97518671e-02 2.49607611e+00
8.53811502e-01 3.98714036e-01 1.53494542e-02 8.19684446e-01
1.11862016e+00 4.52815861e-01 -5.40416062e-01 -6.10347092e-01
-1.19300783e-01 2.14111969e-01 8.67769048e-02 8.26118767e-01
-1.14149070e+00 6.64754748e-01 5.83693218e+00 1.09255481e+00
-1.30510366e+00 8.02890286e-02 4.09816056e-01 1.06878765e-01
1.52617414e-02 -2.13443995e-01 -8.11258197e-01 7.30056107e-01
1.01287127e+00 -2.31427506e-01 5.49071431e-01 1.02121985e+00
4.08663541e-01 6.57451570e-01 -1.00350881e+00 9.10274148e-01
1.12741821e-01 -8.03607106e-01 5.57607375e-02 2.10891534e-02
2.45369449e-01 -4.73904610e-01 4.78783369e-01 3.21855605e-01
1.21196114e-01 -1.11910844e+00 8.33732426e-01 -1.31541893e-01
6.61771536e-01 -8.68666172e-01 7.69567609e-01 4.59052533e-01
-1.40889037e+00 -2.57061958e-01 -3.50816607e-01 2.12553367e-02
-1.22372862e-02 3.34172368e-01 -1.11949325e+00 3.81733567e-01
4.43475574e-01 -1.04283698e-01 -2.36230746e-01 6.18167281e-01
-5.61009407e-01 1.10077107e+00 -1.45817310e-01 4.02725488e-02
1.74944978e-02 3.34624872e-02 1.00710654e+00 1.12086201e+00
1.69689462e-01 -3.76615897e-02 1.60670280e-01 7.33686745e-01
-6.75130039e-02 5.22007234e-02 -5.07097960e-01 2.19978586e-01
9.96797144e-01 1.10042024e+00 -6.18234389e-02 -1.30469263e-01
-4.20962460e-02 8.52237463e-01 -4.91057783e-02 4.20293123e-01
-1.13305688e+00 -6.42241955e-01 9.23510909e-01 -1.74078852e-01
1.49017125e-01 -2.91618239e-02 -3.93759608e-01 -8.40665460e-01
3.28520715e-01 -1.42034185e+00 3.51926535e-01 -3.47467899e-01
-1.26393199e+00 9.11919951e-01 -3.60593230e-01 -1.61527300e+00
-2.88330972e-01 -3.28715414e-01 -1.25490534e+00 1.20402241e+00
-1.26024711e+00 -9.26693618e-01 -1.75272495e-01 7.43377268e-01
5.34177125e-01 -5.58586776e-01 1.01631653e+00 1.93585247e-01
-6.34847105e-01 1.26297140e+00 -1.26751751e-01 5.64124525e-01
6.55101776e-01 -1.00277805e+00 6.45575583e-01 1.21510375e+00
2.59358853e-01 7.16743171e-01 8.17209363e-01 -3.89832377e-01
-1.05115688e+00 -1.03854227e+00 4.90970939e-01 -4.62613344e-01
5.46470642e-01 -6.49111331e-01 -1.19614291e+00 4.89118367e-01
1.31447732e-01 -9.76341665e-02 7.55352736e-01 -3.80038559e-01
-6.35986626e-01 -3.18876982e-01 -1.35587728e+00 5.26579499e-01
6.26974881e-01 -7.57346392e-01 -9.08072174e-01 1.51004761e-01
9.56953645e-01 -2.73977250e-01 -2.87267148e-01 3.49427521e-01
3.06378603e-01 -9.65484321e-01 1.21616137e+00 -4.71676290e-01
-1.48260251e-01 -6.80726707e-01 -1.95711106e-01 -1.30202520e+00
-1.37202710e-01 -1.04487193e+00 -3.48201066e-01 1.63157463e+00
4.94314194e-01 -9.50632632e-01 7.06789255e-01 2.92970210e-01
-2.11974978e-01 -3.55122328e-01 -1.22921991e+00 -1.08468223e+00
-1.11181498e-01 -5.10453284e-01 9.55772638e-01 9.76849914e-01
-8.59601200e-02 9.31222811e-02 -6.56374156e-01 1.04188538e+00
7.82034516e-01 -3.37028980e-01 7.86781013e-01 -8.82160068e-01
-4.62127775e-01 -2.77327836e-01 -6.62677288e-01 -7.35834956e-01
1.88534752e-01 -6.34888709e-01 9.44494009e-02 -6.38729334e-01
-4.02211189e-01 -9.51824486e-02 -4.10606414e-01 2.92819828e-01
-6.67880535e-01 1.35897249e-01 2.21610412e-01 7.14808255e-02
9.40117240e-02 4.46830690e-01 6.56785667e-01 -5.06513774e-01
-3.25943381e-01 4.12076980e-01 -7.65432715e-01 9.02929783e-01
9.22177196e-01 -7.60923028e-01 -5.48106551e-01 -1.31155297e-01
-5.22905588e-01 1.23498827e-01 2.65152574e-01 -1.11706114e+00
3.45865078e-02 -1.53865531e-01 2.38371417e-01 -3.36732447e-01
4.41086709e-01 -9.26053226e-01 3.88030149e-02 7.11292386e-01
-3.08050931e-01 -1.89077303e-01 1.62418276e-01 5.64270318e-01
-3.51298183e-01 -3.97620469e-01 1.05010808e+00 1.73854068e-01
-3.12222779e-01 1.66020393e-01 -3.19152534e-01 8.44908282e-02
8.88811707e-01 -2.22567856e-01 -3.05422157e-01 -5.64833701e-01
-4.35433358e-01 -2.35874012e-01 2.46877572e-03 4.54266459e-01
8.10780525e-01 -1.47239280e+00 -9.41651762e-01 5.41796744e-01
-1.91611066e-01 -3.23015124e-01 3.37091982e-01 1.28986716e-01
-1.87277809e-01 1.67330712e-01 1.89208642e-01 -3.72148722e-01
-1.71453834e+00 6.98278725e-01 5.52974820e-01 -5.99735826e-02
-1.13648847e-01 9.52331960e-01 5.36082983e-01 -3.11152399e-01
5.81524670e-01 1.15203240e-03 -2.82709539e-01 -3.51000994e-01
1.12981296e+00 1.32162169e-01 1.05847502e-02 -7.97235012e-01
-5.45815825e-01 3.67982268e-01 -1.84545200e-02 -2.55913734e-01
8.43977153e-01 4.13057655e-02 1.15062900e-01 7.57732466e-02
1.30349278e+00 6.38584793e-01 -1.00493336e+00 -1.42882019e-01
-2.49720350e-01 -6.01704597e-01 -1.75136954e-01 -6.64590001e-01
-9.99174356e-01 9.33151484e-01 9.66903627e-01 7.45838523e-01
1.20919597e+00 -4.37499076e-01 9.19172049e-01 9.84130353e-02
1.90706149e-01 -7.15247989e-01 2.42262259e-01 3.10509562e-01
1.01828790e+00 -1.00724840e+00 -3.41899782e-01 -5.52113235e-01
-7.04496264e-01 9.63553786e-01 7.00458407e-01 4.30371612e-03
5.20470023e-01 2.43903100e-01 5.88169336e-01 2.97838360e-01
-2.55152524e-01 4.76127833e-01 3.37332755e-01 8.91406000e-01
-1.84013471e-02 1.03554018e-02 -1.46493325e-02 1.02088118e+00
-5.58809757e-01 -7.50984907e-01 4.42462116e-01 5.61154187e-01
-5.19836009e-01 -1.10162890e+00 -1.06391799e+00 -1.36749744e-01
-7.42416501e-01 -2.08589181e-01 -7.27940023e-01 3.52023423e-01
6.94483072e-02 1.43649971e+00 -4.17875767e-01 -8.71760368e-01
3.13229114e-01 2.17799142e-01 -2.18983456e-01 -4.13434774e-01
-7.00316966e-01 -1.87057525e-01 -9.17915329e-02 -1.80471644e-01
8.81859586e-02 -4.25864100e-01 -1.34185243e+00 -3.10514331e-01
-5.07283092e-01 4.15898442e-01 6.63132548e-01 8.80412698e-01
1.29589051e-01 4.32526529e-01 1.43192804e+00 -6.26108885e-01
-1.18968940e+00 -9.93985951e-01 -4.49904710e-01 4.65500027e-01
7.65852511e-01 -3.26870650e-01 -1.06557620e+00 -6.65447256e-03] | [14.012434959411621, 5.833187580108643] |
a137f7bc-7f8e-4d9a-ae77-5a49bedb7f0a | when-sam-meets-medical-images-an | 2304.08506 | null | https://arxiv.org/abs/2304.08506v5 | https://arxiv.org/pdf/2304.08506v5.pdf | When SAM Meets Medical Images: An Investigation of Segment Anything Model (SAM) on Multi-phase Liver Tumor Segmentation | Learning to segmentation without large-scale samples is an inherent capability of human. Recently, Segment Anything Model (SAM) performs the significant zero-shot image segmentation, attracting considerable attention from the computer vision community. Here, we investigate the capability of SAM for medical image analysis, especially for multi-phase liver tumor segmentation (MPLiTS), in terms of prompts, data resolution, phases. Experimental results demonstrate that there might be a large gap between SAM and expected performance. Fortunately, the qualitative results show that SAM is a powerful annotation tool for the community of interactive medical image segmentation. | ['Xinde Li', 'Shenghong Ju', 'Tianyi Xia', 'Chuanfei Hu'] | 2023-04-17 | null | null | null | null | ['tumor-segmentation'] | ['computer-vision'] | [ 2.12086871e-01 5.66574037e-01 -5.80650508e-01 -2.50718087e-01
-1.24073291e+00 -3.90018731e-01 7.81858116e-02 3.27997983e-01
-3.83629590e-01 3.09466898e-01 4.88373488e-02 -3.17086220e-01
1.01781890e-01 -3.63794982e-01 -7.36220032e-02 -8.78723979e-01
2.28118613e-01 8.63856733e-01 6.52647257e-01 1.53551057e-01
1.33297086e-01 4.44184065e-01 -8.58111382e-01 2.75689989e-01
8.72086346e-01 5.57202399e-01 2.10267752e-01 6.34081781e-01
-3.27899009e-01 8.74132454e-01 -5.56525946e-01 -5.82930027e-03
1.90886538e-02 -4.11583275e-01 -1.24555910e+00 5.56520224e-01
-5.19204438e-02 -9.48521644e-02 1.86576933e-01 1.20588994e+00
6.51184976e-01 -1.23602301e-01 6.24244809e-01 -1.10633862e+00
1.96973383e-02 9.06538248e-01 -7.11287141e-01 3.56510013e-01
1.62759572e-01 5.14333904e-01 6.57644987e-01 -6.78708911e-01
1.04619420e+00 7.91907072e-01 6.29719734e-01 5.68534911e-01
-1.10101271e+00 -3.08027953e-01 -2.76715070e-01 -9.32588521e-03
-1.41103339e+00 -3.40977252e-01 5.09856582e-01 -7.39943683e-01
5.13975918e-01 5.92035294e-01 8.65290999e-01 7.77133703e-01
1.28834262e-01 1.18225932e+00 9.97071028e-01 -5.26951730e-01
3.93291175e-01 2.89184898e-01 6.86088562e-01 9.25311506e-01
8.64687115e-02 -3.01564813e-01 -1.69442669e-01 -6.59566894e-02
7.59628594e-01 -1.88565582e-01 -3.62906665e-01 -4.01531130e-01
-1.37657535e+00 7.34825850e-01 3.38301271e-01 4.42086607e-01
-2.34642357e-01 -1.22007824e-01 7.03987181e-01 -7.84075633e-02
6.20225668e-01 5.68057656e-01 -1.28552660e-01 -1.42611176e-01
-1.29303753e+00 -4.14391845e-01 8.61847758e-01 9.86539423e-01
2.12942272e-01 -9.76270661e-02 -5.38423419e-01 5.25081694e-01
1.97065659e-02 1.40533773e-02 6.99697137e-01 -8.86133671e-01
-3.30492198e-01 7.53992915e-01 -2.00822145e-01 -4.02510047e-01
-6.83407307e-01 -1.59502715e-01 -8.62624288e-01 1.00996450e-01
4.89833802e-01 -8.79367813e-02 -1.26437569e+00 1.10504687e+00
4.97489750e-01 2.73874044e-01 -1.38069633e-02 9.57675576e-01
1.27650487e+00 1.52618349e-01 3.42396051e-01 -7.33974576e-01
1.47178257e+00 -1.32802653e+00 -9.76438046e-01 -6.74789175e-02
9.36814308e-01 -7.97001004e-01 1.10688794e+00 1.28919452e-01
-1.14545894e+00 -3.26439172e-01 -6.96008801e-01 1.84917510e-01
-1.39115557e-01 -5.48015498e-02 7.07525790e-01 6.88821375e-01
-8.69670093e-01 3.70643973e-01 -1.07082272e+00 -6.89154088e-01
9.24662888e-01 2.45768383e-01 -4.59898300e-02 1.63440164e-02
-7.41341114e-01 6.55457854e-01 5.46704829e-01 -4.29925829e-01
-8.66026342e-01 -8.32988560e-01 -6.57675564e-01 -2.18589529e-01
6.87303185e-01 -5.65402806e-01 1.45472014e+00 -1.01428342e+00
-1.14811003e+00 1.45913517e+00 2.40703393e-02 -4.78885591e-01
8.48908246e-01 4.54286128e-01 2.18662564e-02 5.49401939e-01
1.72973022e-01 1.01818526e+00 6.24676347e-01 -1.03279436e+00
-6.12413287e-01 -6.89374935e-03 -2.97138184e-01 9.34525654e-02
1.74730364e-02 2.12749094e-01 -6.54363275e-01 -4.77186948e-01
1.24951638e-01 -8.53350401e-01 -8.63049567e-01 1.63504899e-01
-6.30505562e-01 -3.84446114e-01 6.55981839e-01 -5.77322006e-01
1.33428383e+00 -2.31773925e+00 -2.18562379e-01 9.05260518e-02
4.20985192e-01 2.96207070e-01 2.12354913e-01 -3.40570398e-02
4.36270423e-02 4.80831146e-01 -5.01319587e-01 -2.07450204e-02
-3.33284438e-01 3.89570385e-01 2.40834951e-01 6.32710695e-01
-1.48608744e-01 1.40747225e+00 -1.07876718e+00 -1.33947659e+00
4.16971564e-01 -4.06653807e-02 -3.15816581e-01 2.41893992e-01
-2.70930648e-01 8.13491166e-01 -4.79086041e-01 1.08603168e+00
3.66695762e-01 -7.38539338e-01 3.16588283e-02 -1.90493420e-01
-1.35728598e-01 -5.53195059e-01 -8.26969206e-01 1.93018711e+00
4.00737464e-01 6.25687957e-01 1.84414610e-01 -8.05098057e-01
4.19874161e-01 6.75409794e-01 1.12742722e+00 -3.70710492e-01
4.30819482e-01 -2.38399822e-02 1.94310933e-01 -9.26607609e-01
2.13513613e-01 -1.01654150e-01 -1.28892390e-02 4.74241823e-01
-2.10408848e-02 -4.09268260e-01 5.66624105e-01 5.04701316e-01
8.90399575e-01 -3.99176538e-01 7.46363819e-01 -6.22779369e-01
2.45988086e-01 7.64326155e-01 5.98580718e-01 8.49901080e-01
-8.74336183e-01 8.09099376e-01 6.21289670e-01 -3.03174585e-01
-8.95429850e-01 -7.20847964e-01 -2.03230143e-01 7.74828672e-01
3.67934227e-01 -2.96000332e-01 -1.46143889e+00 -9.38344359e-01
-6.28591359e-01 4.92766529e-01 -5.73486507e-01 2.62318581e-01
-3.88663858e-01 -8.60395014e-01 4.79929000e-01 4.41136241e-01
2.18797565e-01 -1.23721123e+00 -1.23212743e+00 2.53530413e-01
-3.09129655e-01 -1.15799260e+00 -5.55229545e-01 1.24695756e-01
-9.34517920e-01 -1.52681541e+00 -9.26771343e-01 -1.14125037e+00
9.80087161e-01 3.42385054e-01 1.13741362e+00 3.05106342e-01
-1.02737164e+00 7.22263098e-01 -4.31534916e-01 -5.03003657e-01
-6.72661781e-01 2.17335746e-01 -5.50114751e-01 -4.92682457e-01
2.83449471e-01 -1.95748359e-02 -5.69902718e-01 5.00991106e-01
-8.62461805e-01 7.34998822e-01 6.11239254e-01 7.97789574e-01
9.59518075e-01 -6.60473853e-02 4.17999983e-01 -1.47865105e+00
5.85018098e-01 -2.61888951e-01 -4.08848226e-01 4.45500523e-01
-5.95685542e-01 -5.63698351e-01 -7.96892419e-02 -4.50826347e-01
-9.82381105e-01 3.06773812e-01 -8.64738226e-02 -3.53850573e-01
-5.07827342e-01 5.51566124e-01 3.33146691e-01 -1.08218469e-01
9.28062141e-01 9.17605087e-02 2.85421550e-01 -1.09042734e-01
3.85022581e-01 5.54810166e-01 3.93929422e-01 -5.40167950e-02
1.48588464e-01 6.59248888e-01 -2.15801179e-01 -1.04857588e+00
-1.02771962e+00 -1.00322187e+00 -7.76833951e-01 -5.02754569e-01
1.12590778e+00 -6.67769432e-01 -3.20531309e-01 2.37681776e-01
-9.22796071e-01 -5.04863501e-01 -8.64894807e-01 1.99231192e-01
-5.01723051e-01 2.65554696e-01 -1.00466990e+00 -4.69859570e-01
-8.03562760e-01 -1.59341919e+00 9.02573466e-01 6.11567020e-01
-3.77074808e-01 -1.15181136e+00 -2.45435506e-01 5.33364236e-01
2.14104325e-01 5.42764723e-01 7.87393868e-01 -1.02287281e+00
-5.29232264e-01 -2.13050365e-01 -3.05839360e-01 -1.13902740e-01
-4.29394022e-02 3.89233157e-02 -8.34803104e-01 -3.56545836e-01
8.03882629e-02 -1.74170613e-01 5.16473889e-01 1.00564778e+00
1.22940755e+00 7.70710036e-02 -6.98386669e-01 4.60502684e-01
1.27921104e+00 2.75618255e-01 4.71750081e-01 -1.08580798e-01
6.51803732e-01 3.96906942e-01 9.90266025e-01 3.18929344e-01
-4.18857858e-02 1.07513063e-01 2.42079541e-01 -7.03441203e-01
-3.07298720e-01 2.19308838e-01 -2.18564078e-01 1.10051847e+00
1.67020634e-01 1.27995580e-01 -1.39018524e+00 6.25098407e-01
-1.97335231e+00 -5.80063403e-01 -3.78415316e-01 1.71130657e+00
9.12574470e-01 3.20604920e-01 1.52634550e-02 -9.49778184e-02
7.64763176e-01 -8.34696293e-02 -5.26372969e-01 4.94187847e-02
3.17281216e-01 -1.05704442e-02 5.55956542e-01 2.58671641e-01
-1.15134358e+00 1.19405293e+00 7.61326838e+00 1.36429381e+00
-9.52156007e-01 4.53606158e-01 9.59709764e-01 2.62577981e-01
1.94866210e-01 -4.35953215e-02 -4.20952559e-01 3.48843426e-01
7.28375971e-01 -5.85274756e-01 -5.31524681e-02 1.00175738e+00
3.58460188e-01 -4.76917237e-01 -1.16515124e+00 9.22051430e-01
1.10709585e-01 -1.64099872e+00 -7.29778260e-02 -6.04025982e-02
8.31359267e-01 -3.48328166e-02 -3.21885943e-01 2.12592080e-01
2.11741537e-01 -9.97051954e-01 3.09397608e-01 4.29267228e-01
8.32425356e-01 -3.69168878e-01 8.56853068e-01 6.52543604e-01
-1.02296042e+00 3.99491012e-01 -2.26753965e-01 5.53239048e-01
5.15042305e-01 5.43120325e-01 -1.38183725e+00 3.69391918e-01
2.68257588e-01 5.59760988e-01 -7.81324804e-01 1.53106916e+00
-9.20171514e-02 7.68196106e-01 1.47139251e-01 2.96920598e-01
2.14347616e-01 -7.95882270e-02 5.82235694e-01 1.55344033e+00
-1.86024413e-01 4.38637048e-01 9.24952686e-01 5.65221071e-01
4.32624295e-02 2.68827230e-01 -2.98085064e-01 -3.42258155e-01
1.65871441e-01 1.59199178e+00 -1.64827096e+00 -8.70559037e-01
-1.52996123e-01 6.61324024e-01 -2.80331910e-01 9.95871350e-02
-9.04110730e-01 6.35379329e-02 -2.07910333e-02 1.02972530e-01
-2.33279675e-01 1.36156484e-01 -5.91425121e-01 -8.25812638e-01
-6.82104111e-01 -7.72085249e-01 6.53195083e-01 -6.79078043e-01
-1.04697907e+00 3.78332317e-01 -1.59559011e-01 -1.33334970e+00
4.94279265e-02 -2.42657557e-01 -6.15033984e-01 1.63968146e-01
-1.20740640e+00 -1.43182325e+00 -5.79331398e-01 5.47392309e-01
1.34405875e+00 1.51204228e-01 8.91692400e-01 1.16371639e-01
-5.58670878e-01 3.18069994e-01 -3.49147350e-01 3.84948552e-01
5.40157020e-01 -1.32812119e+00 -4.48227823e-02 7.06157923e-01
4.13972065e-02 1.05410434e-01 8.11126590e-01 -5.95129728e-01
-1.11292803e+00 -1.00691950e+00 2.65106350e-01 -2.17619419e-01
4.88668233e-01 2.58646876e-01 -9.87953067e-01 6.08261108e-01
3.96580458e-01 2.04502046e-01 1.15323830e+00 -3.56789201e-01
4.02783573e-01 4.98210967e-01 -1.42558956e+00 5.13951123e-01
6.11454427e-01 -3.31404135e-02 -4.93579745e-01 7.23456383e-01
7.37179935e-01 -8.67923498e-01 -1.01974261e+00 2.91685760e-01
8.88443962e-02 -7.91248381e-01 7.64148653e-01 -3.18227291e-01
2.75017351e-01 -1.89236421e-02 4.44573462e-01 -9.68080640e-01
-1.67815045e-01 -4.45534855e-01 1.23396546e-01 9.60424542e-01
3.24995697e-01 -1.98866621e-01 1.01654065e+00 9.63128865e-01
-3.73309284e-01 -7.67708123e-01 -8.35600674e-01 -3.70943159e-01
-9.17396247e-02 -2.92568982e-01 -7.48514012e-02 1.20565689e+00
2.04012856e-01 2.56230623e-01 2.25793328e-02 -2.15298831e-01
7.39082813e-01 1.15249783e-01 5.67884505e-01 -1.28308904e+00
-6.87515363e-02 -6.38620436e-01 -1.91786438e-01 -5.87099195e-01
-1.80622965e-01 -1.05725181e+00 1.74052939e-01 -1.91118979e+00
6.43404484e-01 -3.40436071e-01 -2.33084932e-01 5.26616096e-01
-2.21225217e-01 2.65773594e-01 2.67970841e-02 4.33313400e-01
-1.15084088e+00 -2.20244765e-01 1.69646931e+00 -3.26590598e-01
-1.69915080e-01 1.09305277e-01 -4.58322644e-01 1.08917928e+00
6.70301259e-01 -5.68134367e-01 -2.65441895e-01 9.84591395e-02
-3.38477969e-01 3.52976561e-01 -7.57304281e-02 -7.96837986e-01
6.86125159e-01 -5.04241467e-01 6.65866286e-02 -6.34678185e-01
-1.45916194e-01 -6.65132821e-01 2.03878358e-01 9.08580661e-01
-2.97904044e-01 -5.24847329e-01 -3.93508300e-02 3.63089263e-01
-3.45975786e-01 -5.61366141e-01 1.28963184e+00 -8.63470554e-01
-1.03714836e+00 4.01481897e-01 -4.92761284e-01 2.55712569e-01
1.56966341e+00 -5.12426913e-01 -1.05018958e-01 2.61089861e-01
-1.15104640e+00 5.27738571e-01 5.15560031e-01 -1.16002806e-01
4.77514982e-01 -7.45975256e-01 -6.05028152e-01 5.42720482e-02
1.07571982e-01 4.05210763e-01 3.73005897e-01 1.52038193e+00
-8.32256794e-01 4.44663204e-02 6.01129942e-02 -1.08034575e+00
-1.59189224e+00 6.39774978e-01 3.29962343e-01 -4.97689396e-01
-1.20933902e+00 9.08118069e-01 1.18161231e-01 7.94862509e-02
4.01584089e-01 -2.68674493e-01 -3.00623924e-01 4.38891441e-01
4.71386373e-01 3.43173981e-01 -2.42763937e-01 -3.89620751e-01
-2.32826859e-01 1.65595576e-01 -2.56271720e-01 3.57281081e-02
1.04623711e+00 -1.62415057e-01 -1.91727996e-01 5.36981463e-01
6.25976801e-01 -4.17689294e-01 -1.12175882e+00 -1.62247479e-01
3.77113640e-01 -2.74288327e-01 1.55053884e-01 -8.10546517e-01
-1.19099712e+00 7.64716804e-01 8.15186441e-01 3.49678904e-01
8.55309010e-01 3.21274161e-01 6.56759977e-01 -1.54375449e-01
3.84494364e-01 -1.35795546e+00 1.59208894e-01 1.59325823e-01
3.88883442e-01 -1.58361506e+00 7.75678605e-02 -6.86730385e-01
-9.64054406e-01 1.02248609e+00 4.88075912e-01 1.07068136e-01
5.80036819e-01 5.11810303e-01 3.23555946e-01 -4.84556437e-01
-4.11717564e-01 -3.15712035e-01 2.06345320e-01 4.87243116e-01
3.72797102e-01 3.81576419e-01 -4.53492731e-01 4.21391666e-01
2.68801510e-01 1.68739647e-01 6.69578493e-01 1.09497678e+00
-8.80469203e-01 -6.75564408e-01 -2.74421930e-01 5.28966784e-01
-4.73319411e-01 3.59594747e-02 -1.50458410e-01 7.98590004e-01
5.21799177e-02 7.05439508e-01 -1.74330577e-01 3.63901109e-01
1.14753820e-01 4.08724789e-03 2.93782145e-01 -8.47257018e-01
-7.60371089e-01 5.51909387e-01 -2.72926599e-01 -5.56227982e-01
-5.30939579e-01 -3.96065742e-01 -1.56679451e+00 2.28203893e-01
-4.94934648e-01 1.82868764e-01 5.04381418e-01 8.79426360e-01
-1.73048258e-01 9.36036825e-01 2.66527295e-01 -5.90908527e-01
-9.01133120e-02 -8.60280156e-01 -5.64477682e-01 4.75255668e-01
-5.00420891e-02 -2.05367252e-01 -2.18554586e-01 4.58887428e-01] | [14.69363021850586, -2.2842347621917725] |
6824218a-eabc-4103-9538-1f5de6db6a12 | on-evaluating-multilingual-compositional | 2306.11420 | null | https://arxiv.org/abs/2306.11420v1 | https://arxiv.org/pdf/2306.11420v1.pdf | On Evaluating Multilingual Compositional Generalization with Translated Datasets | Compositional generalization allows efficient learning and human-like inductive biases. Since most research investigating compositional generalization in NLP is done on English, important questions remain underexplored. Do the necessary compositional generalization abilities differ across languages? Can models compositionally generalize cross-lingually? As a first step to answering these questions, recent work used neural machine translation to translate datasets for evaluating compositional generalization in semantic parsing. However, we show that this entails critical semantic distortion. To address this limitation, we craft a faithful rule-based translation of the MCWQ dataset from English to Chinese and Japanese. Even with the resulting robust benchmark, which we call MCWQ-R, we show that the distribution of compositions still suffers due to linguistic divergences, and that multilingual models still struggle with cross-lingual compositional generalization. Our dataset and methodology will be useful resources for the study of cross-lingual compositional generalization in other tasks. | ['Daniel Hershcovich', 'Zi Wang'] | 2023-06-20 | null | null | null | null | ['machine-translation', 'semantic-parsing'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.53532046e-01 -1.96540169e-02 -4.13799286e-01 -6.65241778e-01
-8.63630891e-01 -1.03387475e+00 5.80075383e-01 2.77322866e-02
-4.83472049e-01 7.83505797e-01 4.31464165e-01 -8.64996910e-01
1.14171311e-01 -7.29705632e-01 -9.90786016e-01 -4.50002253e-01
2.51367927e-01 8.02576661e-01 4.11517881e-02 -3.90213370e-01
1.22643054e-01 2.45806590e-01 -1.10232258e+00 7.33779490e-01
1.08718562e+00 3.78090441e-01 2.36121565e-01 3.09550583e-01
-3.38728935e-01 2.97812670e-01 -1.43710688e-01 -7.56636083e-01
3.97857428e-01 -8.39757562e-01 -1.16969895e+00 -3.86390239e-01
8.39429975e-01 -9.10050143e-03 1.69733629e-01 1.16494560e+00
3.62139434e-01 1.02076279e-02 6.24808252e-01 -9.19957936e-01
-1.20910299e+00 1.29467893e+00 -3.91973034e-02 2.38322586e-01
2.13080302e-01 8.80282670e-02 1.48639059e+00 -7.67533004e-01
9.13277864e-01 1.55878484e+00 9.28779900e-01 9.30765867e-01
-1.56495118e+00 -6.98539138e-01 4.53654915e-01 2.55686883e-02
-1.06984353e+00 -2.64652550e-01 6.03964865e-01 -2.07390323e-01
9.55594659e-01 9.99173298e-02 3.16389501e-01 1.38290274e+00
4.45373496e-03 7.22068787e-01 1.65657616e+00 -4.78538424e-01
-4.36502043e-03 7.25037903e-02 9.73000452e-02 6.59867048e-01
3.35576922e-01 2.21526414e-01 -6.16350412e-01 6.16904907e-03
3.86590511e-01 -4.14357126e-01 -2.85784453e-01 -1.53543159e-01
-1.24455607e+00 9.52969790e-01 5.48540831e-01 4.28773999e-01
4.55219373e-02 6.03538677e-02 6.22542381e-01 6.68884099e-01
3.73431236e-01 8.59706163e-01 -8.91669333e-01 1.66349530e-01
-5.75397253e-01 4.50519979e-01 6.91328466e-01 1.09477055e+00
7.78371751e-01 2.94225453e-03 5.37198782e-01 9.51632857e-01
-2.14037463e-01 5.72384238e-01 7.48498619e-01 -1.01344466e+00
6.84796572e-01 3.18390161e-01 -4.15012151e-01 -5.18345535e-01
-3.84684384e-01 -3.67220014e-01 -2.97383577e-01 -1.55747041e-01
6.02685869e-01 -1.69438079e-01 -4.60421383e-01 2.40110707e+00
-6.92911725e-03 -6.52069271e-01 2.56809443e-01 8.74284625e-01
3.39729279e-01 5.96527040e-01 5.74536920e-01 1.56723037e-01
1.31726849e+00 -7.41727114e-01 -1.79422200e-01 -5.07588744e-01
9.30497110e-01 -7.38280773e-01 1.73747087e+00 1.03721851e-02
-1.20498562e+00 -4.42517102e-01 -9.77568686e-01 -4.04498816e-01
-5.72890103e-01 -2.76887715e-01 8.46516311e-01 7.44434178e-01
-1.01798069e+00 6.37775362e-01 -7.27436244e-01 -8.95359516e-01
2.95167714e-01 7.35615045e-02 -2.79680669e-01 -3.57445210e-01
-1.27027726e+00 1.29638970e+00 5.00447273e-01 -3.30153197e-01
-7.00681806e-01 -9.58152771e-01 -7.47555852e-01 -1.53847739e-01
1.45726353e-01 -8.57497036e-01 1.45220268e+00 -1.41305625e+00
-1.15952349e+00 1.10670567e+00 -1.59838811e-01 -3.58059824e-01
2.85317153e-01 -4.84262630e-02 -2.13349566e-01 -1.66665152e-01
2.83693343e-01 8.71665955e-01 2.75397241e-01 -1.24103677e+00
-6.83089256e-01 -7.41189539e-01 2.48960741e-02 2.90141732e-01
-1.31099477e-01 2.17233419e-01 2.75464475e-01 -8.34651649e-01
2.37557679e-01 -1.22064042e+00 1.14485569e-01 -3.23764056e-01
1.52892664e-01 -3.35259914e-01 2.63082772e-01 -7.68059909e-01
7.20619678e-01 -1.96354711e+00 3.15995604e-01 -2.35903725e-01
-4.69067305e-01 -1.45194858e-01 -3.28878045e-01 5.19699931e-01
-1.13749787e-01 4.68974531e-01 -6.59648418e-01 -2.45413601e-01
3.32218975e-01 5.53890646e-01 -6.61929131e-01 2.74084419e-01
3.36817771e-01 1.28044701e+00 -8.18560481e-01 -2.72608012e-01
-3.95983100e-01 4.74023372e-02 -8.98632348e-01 -2.75630981e-01
-5.20430624e-01 3.49968374e-01 -4.78658266e-02 5.51468015e-01
3.58996242e-01 3.25011730e-01 5.70970893e-01 7.03301951e-02
-1.53805818e-02 1.00096655e+00 -4.50397015e-01 2.01191545e+00
-7.19989181e-01 4.86507893e-01 -7.09725078e-03 -1.20032966e+00
5.10946393e-01 3.71738188e-02 -3.03389907e-01 -1.01069880e+00
-1.17112778e-01 9.01994884e-01 3.96094173e-01 -2.82744050e-01
4.72863615e-01 -1.00808537e+00 -3.48882377e-01 7.68344879e-01
1.38245314e-01 -4.13932592e-01 1.53841853e-01 -1.95900455e-01
6.87012553e-01 5.45335770e-01 -3.15988548e-02 -8.00412536e-01
3.12652826e-01 3.76191646e-01 7.42760181e-01 4.54309106e-01
8.00417513e-02 4.27096456e-01 4.07149792e-01 -5.59591830e-01
-1.26308072e+00 -1.40962029e+00 -2.94145226e-01 1.63579166e+00
-6.18364289e-02 -4.25916240e-02 -8.46941769e-01 -8.52789044e-01
-4.61864620e-02 1.19164395e+00 -4.97729570e-01 -1.80125460e-01
-1.13965869e+00 -9.87424493e-01 7.30569720e-01 8.09365630e-01
1.47765219e-01 -1.14033449e+00 -2.23172337e-01 8.32267627e-02
-2.58261949e-01 -1.19435513e+00 -5.21500349e-01 3.50389838e-01
-1.26816201e+00 -8.24286759e-01 -4.75191981e-01 -1.13691723e+00
5.55625558e-01 5.33227855e-03 1.50539839e+00 -8.07828009e-02
1.26155794e-01 1.89360350e-01 -2.67962992e-01 -4.41876769e-01
-8.71283591e-01 5.57505846e-01 1.27889565e-03 -6.88479304e-01
6.23902738e-01 -7.28168249e-01 -1.96062922e-01 3.42778862e-01
-9.09547389e-01 -2.37895340e-01 5.62014639e-01 7.76438296e-01
3.49521309e-01 -2.44978473e-01 9.50260401e-01 -1.20500374e+00
8.42486501e-01 -4.48236018e-01 -5.82103789e-01 1.44753754e-01
-5.28603673e-01 4.10171032e-01 9.50836480e-01 -4.95142847e-01
-1.25082207e+00 -2.24996060e-01 -1.15109354e-01 4.11262661e-01
-4.70986366e-02 4.73400921e-01 -2.47267872e-01 2.25149453e-01
8.87017429e-01 1.88317284e-01 -1.12773426e-01 -5.30584574e-01
8.01534235e-01 4.29915965e-01 5.36901593e-01 -1.30459797e+00
6.61357701e-01 3.01052332e-01 -1.37448087e-01 -8.08512747e-01
-8.66151989e-01 7.48669058e-02 -5.98119199e-01 6.18303955e-01
1.03422427e+00 -9.40841019e-01 -2.77608544e-01 -2.00768299e-02
-1.12486565e+00 -7.49661446e-01 -3.18229258e-01 3.26957703e-01
-6.54141963e-01 2.00717762e-01 -8.92383337e-01 -1.36109740e-01
-2.35761151e-01 -1.11731219e+00 9.20363665e-01 -4.20434266e-01
-8.16530108e-01 -1.28742588e+00 1.58895150e-01 4.20559496e-01
4.86796290e-01 -2.62760818e-01 1.92782760e+00 -6.73565030e-01
-6.69655859e-01 1.91711530e-01 -1.81277618e-01 5.47203779e-01
3.85629162e-02 -3.87704551e-01 -6.76320255e-01 -2.04846606e-01
3.91764231e-02 -5.84894061e-01 9.53675628e-01 1.02623202e-01
7.77474105e-01 -2.15259954e-01 -1.70962121e-02 7.47335553e-01
1.34894550e+00 5.63841537e-02 1.03790097e-01 2.96443015e-01
4.81870413e-01 9.78233814e-01 2.31776293e-02 -6.52734935e-01
7.68070281e-01 4.35003370e-01 -8.89570192e-02 1.14499681e-01
-4.20698315e-01 -5.59462786e-01 5.50237715e-01 1.07689703e+00
5.66615313e-02 5.57211898e-02 -9.26674187e-01 7.72670448e-01
-1.46476197e+00 -7.20854223e-01 1.42837286e-01 2.11578202e+00
1.07571781e+00 5.14496937e-02 9.13171992e-02 -1.56315640e-01
3.81579459e-01 -6.57202750e-02 -2.45817631e-01 -8.58048856e-01
-4.36003864e-01 5.27792811e-01 3.81580651e-01 7.88451672e-01
-6.96337104e-01 1.38368869e+00 6.54571152e+00 3.61577392e-01
-1.22717345e+00 2.74662763e-01 3.56409043e-01 6.41522631e-02
-7.71549225e-01 1.03631027e-01 -7.34021425e-01 3.19478422e-01
9.94062364e-01 8.70048553e-02 8.11172664e-01 5.98574460e-01
-3.10073107e-01 2.04689518e-01 -1.60715091e+00 4.09271926e-01
1.69647306e-01 -1.01321363e+00 3.15387398e-01 -2.52962172e-01
1.00648534e+00 3.60521227e-01 1.95371509e-01 5.37691534e-01
5.83638847e-01 -9.99138534e-01 8.31830144e-01 -3.14358652e-01
6.59501255e-01 -6.58984125e-01 5.58413088e-01 2.65062362e-01
-6.99041784e-01 -6.64973184e-02 -5.89584947e-01 -1.87644303e-01
9.67268124e-02 4.30056341e-02 -6.47770226e-01 3.21481794e-01
4.16535586e-01 2.91109115e-01 -6.58337057e-01 1.95403472e-01
-5.40394783e-01 6.28174484e-01 -1.76777378e-01 -1.41371220e-01
4.16119516e-01 -2.69014299e-01 2.22742692e-01 1.25429523e+00
3.77961814e-01 -1.40298754e-01 -3.92567851e-02 1.02457356e+00
-3.93311560e-01 3.22533756e-01 -7.43703485e-01 -1.90881357e-01
3.26181918e-01 6.49165452e-01 -6.00914538e-01 -2.67414838e-01
-5.61268330e-01 1.09957349e+00 7.27219939e-01 5.18507183e-01
-5.25159121e-01 1.20263502e-01 7.98126876e-01 1.66955173e-01
-2.43209582e-02 -5.10778248e-01 -5.84541619e-01 -1.18674564e+00
1.40002817e-01 -1.22094285e+00 3.63024294e-01 -6.28303707e-01
-1.60392821e+00 5.00751197e-01 8.04175511e-02 -4.03177559e-01
-3.00870448e-01 -9.53696966e-01 -2.32190475e-01 1.01300824e+00
-1.34999681e+00 -1.36927068e+00 2.93388903e-01 2.50128269e-01
7.85859466e-01 -1.05440356e-01 9.42338347e-01 2.87362546e-01
-2.95342356e-01 6.05717123e-01 -1.59300372e-01 4.81526293e-02
8.82170081e-01 -1.33660901e+00 8.11687291e-01 7.56514192e-01
3.98788363e-01 1.01105034e+00 5.64896524e-01 -5.13754725e-01
-1.24465144e+00 -1.12024915e+00 1.30633950e+00 -9.01816487e-01
7.81674802e-01 -5.17450035e-01 -1.06908023e+00 1.24537086e+00
3.52025628e-01 -2.82228619e-01 6.41877651e-01 5.17204523e-01
-8.04629803e-01 3.99499349e-02 -7.12880552e-01 9.48290169e-01
1.57614958e+00 -9.30176854e-01 -1.13642061e+00 3.01256895e-01
1.06508493e+00 8.91741272e-03 -6.10880315e-01 3.66598815e-01
5.32193184e-01 -8.04312587e-01 7.74072051e-01 -1.06521392e+00
6.66222095e-01 -8.23575854e-02 -5.68789065e-01 -1.65613055e+00
-2.35289395e-01 -2.55950242e-01 8.25752556e-01 1.18918872e+00
9.95819688e-01 -8.72870445e-01 5.60821950e-01 3.56073081e-01
-4.00449306e-01 -5.35473704e-01 -7.44678199e-01 -1.05067027e+00
1.19409502e+00 -5.53495169e-01 6.82484567e-01 1.19102514e+00
5.49054667e-02 6.99469686e-01 2.77563840e-01 -8.35484341e-02
4.80958730e-01 3.59865636e-01 4.72296774e-01 -9.25849438e-01
-2.94711202e-01 -7.39183605e-01 -2.91835777e-02 -9.31471884e-01
8.91830444e-01 -1.62366486e+00 9.57588777e-02 -1.09708059e+00
2.62548447e-01 -4.39781338e-01 3.06459088e-02 2.68060207e-01
-2.96479464e-01 2.55329043e-01 1.98159724e-01 5.99107333e-02
-1.61285937e-01 3.76497567e-01 1.06342399e+00 -1.03922479e-01
4.12358567e-02 -3.50764424e-01 -9.77104127e-01 8.11944842e-01
9.85172272e-01 -5.48148513e-01 -4.86042917e-01 -1.10164642e+00
5.48125505e-01 -4.25962895e-01 7.51287416e-02 -4.81614292e-01
-9.97708291e-02 -1.56325072e-01 2.73262542e-02 -1.47479862e-01
-7.66898170e-02 -7.39452302e-01 -1.00454442e-01 4.29910451e-01
-5.76581895e-01 5.47707021e-01 2.85205781e-01 1.14969052e-01
-1.65108353e-01 -2.50261545e-01 7.78382957e-01 -4.60919172e-01
-5.55123031e-01 -1.30506726e-02 -3.55621755e-01 7.55518138e-01
5.08569121e-01 9.43872482e-02 -5.21276414e-01 -3.23166177e-02
-4.17376697e-01 4.40826528e-02 1.02843475e+00 5.97762346e-01
-4.40040715e-02 -1.34843397e+00 -8.70911300e-01 1.54208556e-01
2.52958059e-01 -2.68625915e-01 -1.00249939e-01 5.70012510e-01
-6.23584569e-01 5.51313460e-01 -1.92142755e-01 -3.65150124e-01
-7.33333111e-01 8.22659254e-01 3.40810329e-01 1.29921138e-01
-2.83437163e-01 9.11910355e-01 4.95663464e-01 -1.24384964e+00
-3.14725488e-01 -6.51240408e-01 5.54333031e-01 -1.42735258e-01
-3.57754692e-03 2.35303357e-01 3.60225863e-03 -6.98388994e-01
-2.73665339e-01 7.95765519e-01 1.69319868e-01 -3.16863000e-01
1.24857521e+00 -2.10198283e-01 -3.32408875e-01 5.74909985e-01
1.34611106e+00 2.84757257e-01 -6.94824040e-01 -3.22075248e-01
4.27823871e-01 7.19403382e-03 -6.29856944e-01 -9.23448741e-01
-5.81675887e-01 1.07074094e+00 4.00005616e-02 -2.87259202e-02
8.97522151e-01 2.09057659e-01 9.54422653e-01 4.36961353e-01
3.83454949e-01 -1.06564081e+00 -3.57271314e-01 8.09095085e-01
9.31535780e-01 -1.22243130e+00 -1.52634770e-01 -2.31375813e-01
-6.90562367e-01 6.98111653e-01 4.55298722e-01 -3.43755670e-02
6.51105493e-02 2.04918385e-02 2.67894745e-01 -2.29924191e-02
-6.88035548e-01 -7.03084981e-03 6.47194162e-02 6.26988292e-01
8.41045439e-01 3.62501919e-01 -6.08665049e-01 4.25210953e-01
-9.22870815e-01 -4.13427055e-01 1.65680170e-01 5.23419738e-01
-1.54354721e-01 -1.40691519e+00 -2.02915013e-01 -6.72609955e-02
-5.63748479e-01 -4.59438324e-01 -7.38445878e-01 9.61971223e-01
2.56383568e-01 5.25835991e-01 -1.28465053e-02 5.23251891e-02
3.30448210e-01 7.66963542e-01 9.51187432e-01 -8.30867529e-01
-5.66815019e-01 -3.89888555e-01 3.43305558e-01 -4.62147206e-01
-2.58357465e-01 -7.77516007e-01 -1.22789443e+00 -2.47078225e-01
1.63820386e-01 5.06395437e-02 7.39378750e-01 9.05442894e-01
2.19047554e-02 3.14164966e-01 5.10704406e-02 -4.64048237e-01
-8.18142593e-01 -9.28738356e-01 -2.71000296e-01 7.79732704e-01
1.35804400e-01 -1.09496593e-01 -4.95675147e-01 1.18156970e-01] | [10.83095932006836, 9.408235549926758] |
3ee1cdb2-c699-4035-bf3d-c7933b429eb8 | da2-deep-attention-adapter-for-memory | 2012.01362 | null | https://arxiv.org/abs/2012.01362v3 | https://arxiv.org/pdf/2012.01362v3.pdf | $DA^3$:Dynamic Additive Attention Adaption for Memory-EfficientOn-Device Multi-Domain Learning | Nowadays, one practical limitation of deep neural network (DNN) is its high degree of specialization to a single task or domain (e.g., one visual domain). It motivates researchers to develop algorithms that can adapt DNN model to multiple domains sequentially, while still performing well on the past domains, which is known as multi-domain learning. Almost all conventional methods only focus on improving accuracy with minimal parameter update, while ignoring high computing and memory cost during training, which makes it difficult to deploy multi-domain learning into more and more widely used resource-limited edge devices, like mobile phones, IoT, embedded systems, etc. We observe that large memory used for activation storage is the bottleneck that largely limits the training time and cost on edge devices. To reduce training memory usage, while keeping the domain adaption accuracy performance, we propose Dynamic Additive Attention Adaption ($DA^3$), a novel memory-efficient on-device multi-domain learning method. $DA^3$ learns a novel additive attention adaptor module, while freezing the weights of the pre-trained backbone model for each domain. Differentiating from prior works, such module not only mitigates activation memory buffering for reducing memory usage during training but also serves as a dynamic gating mechanism to reduce the computation cost for fast inference. We validate $DA^3$ on multiple datasets against state-of-the-art methods, which shows great improvement in both accuracy and training time. Moreover, we deployed $DA^3$ into the popular NIVDIA Jetson Nano edge GPU, where the measured experimental results show our proposed $DA^3$ reduces the on-device training memory consumption by 19-37x, and training time by 2x, in comparison to the baseline methods (e.g., standard fine-tuning, Parallel and Series Res. adaptor, and Piggyback). | ['Deliang Fan', 'Adnan Siraj Rakin', 'Li Yang'] | 2020-12-02 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [-6.26830989e-03 -4.48577791e-01 -2.65816808e-01 -2.30077296e-01
-3.72028530e-01 -5.71295977e-01 8.42646360e-02 -2.50972867e-01
-6.99743271e-01 8.25612187e-01 -1.83383986e-01 -3.68726939e-01
-4.83324863e-02 -9.74651396e-01 -9.30131495e-01 -7.83004105e-01
5.88908136e-01 3.29031467e-01 5.65696836e-01 -6.87149391e-02
-2.28956461e-01 4.05868262e-01 -1.43673503e+00 1.47395879e-01
8.38065982e-01 1.30133212e+00 5.83590806e-01 2.29119539e-01
-2.57236153e-01 3.52690190e-01 -5.51324785e-01 -4.35132831e-01
1.50374234e-01 -7.78799951e-02 -4.00345266e-01 -3.31204325e-01
2.79990375e-01 -6.29488647e-01 -4.07711774e-01 1.25293541e+00
8.49973798e-01 7.80580789e-02 3.66728216e-01 -1.01241457e+00
-5.62437236e-01 2.59140670e-01 -5.71083665e-01 4.89640087e-01
-4.25147235e-01 1.65970370e-01 5.27368784e-01 -7.91143179e-01
4.74986374e-01 9.16394889e-01 7.29442120e-01 7.54564703e-01
-9.36918616e-01 -1.16143811e+00 4.10634220e-01 1.79403037e-01
-1.29649425e+00 -3.91519397e-01 8.50983739e-01 -2.84959108e-01
1.16579318e+00 -3.36104110e-02 5.70758998e-01 1.24688303e+00
1.44579902e-01 6.56949401e-01 8.39019358e-01 -7.69024789e-02
3.10412675e-01 4.60929349e-02 2.67059237e-01 6.09337330e-01
4.53529865e-01 -2.05486447e-01 -3.29435289e-01 3.07389759e-02
1.16657853e+00 2.96060920e-01 -1.07186936e-01 1.10731304e-01
-8.84622216e-01 5.10369599e-01 4.20294136e-01 2.09996372e-01
-3.81982177e-01 1.57765999e-01 9.97633874e-01 1.73746675e-01
4.99086738e-01 -2.15345845e-01 -9.21166062e-01 -4.05076891e-01
-6.69502139e-01 -1.16048306e-01 2.66281933e-01 1.08490634e+00
8.95435274e-01 3.57366562e-01 2.69878600e-02 9.85910237e-01
5.94866611e-02 5.35470665e-01 8.45177829e-01 -5.20730376e-01
6.27268791e-01 8.24056149e-01 -2.71948367e-01 -5.90933025e-01
-6.39469981e-01 -5.33061981e-01 -1.36626327e+00 1.90958977e-02
1.36045173e-01 -6.37444437e-01 -1.00962210e+00 1.88053870e+00
3.68043065e-01 4.61872190e-01 -1.91334002e-02 8.79912019e-01
9.81129587e-01 6.20689929e-01 2.12137222e-01 -2.14688465e-01
1.40528274e+00 -1.18281543e+00 -4.15134370e-01 -3.85077804e-01
6.63646460e-01 -5.00259697e-01 1.38615513e+00 4.37667817e-01
-8.75785589e-01 -1.00753701e+00 -1.11336923e+00 -2.75408268e-01
-3.01697493e-01 3.24280024e-01 8.92106891e-01 6.97200477e-01
-9.16620672e-01 5.71349382e-01 -1.00252855e+00 -1.78704292e-01
6.85572803e-01 7.51689196e-01 -1.31280437e-01 -1.48017943e-01
-1.10179043e+00 3.67393315e-01 4.21500385e-01 -2.96405852e-02
-7.37684250e-01 -9.87315238e-01 -4.35220152e-01 5.32574430e-02
8.73339474e-02 -8.61158967e-01 1.00360692e+00 -1.22434676e+00
-1.87126088e+00 7.41470218e-01 -1.71030477e-01 -3.64393324e-01
2.67062992e-01 -4.64888871e-01 -7.89858043e-01 -1.62576199e-01
-1.38048947e-01 6.68877721e-01 8.91504228e-01 -7.46422946e-01
-5.63219070e-01 -4.83005285e-01 4.61646803e-02 2.21457452e-01
-9.20770943e-01 -1.04052186e-01 -8.14376175e-01 -7.63681531e-01
-6.65643513e-02 -9.26745117e-01 -6.64248392e-02 1.70274638e-02
2.00056285e-02 -3.08241159e-01 1.13219142e+00 -5.27211964e-01
1.37484121e+00 -2.28971887e+00 -3.89451645e-02 -6.18237667e-02
3.24710906e-01 6.66695297e-01 1.74054489e-01 -4.11549181e-01
-1.79621335e-02 -4.87793088e-02 -3.36481035e-02 -3.75080079e-01
-1.87993169e-01 4.40659493e-01 -8.50501359e-02 1.91989705e-01
-3.28630731e-02 8.90286803e-01 -5.95266104e-01 -3.14263791e-01
1.12757757e-01 5.68778872e-01 -8.30523431e-01 6.68461248e-02
-2.29804367e-01 3.17852706e-01 -4.53525037e-01 6.06328368e-01
8.99263620e-01 -5.32590330e-01 2.17139333e-01 -5.29899359e-01
1.75895244e-02 1.65169373e-01 -1.13654661e+00 1.91469359e+00
-7.75711656e-01 4.12890017e-01 1.61211684e-01 -1.17557538e+00
9.44432914e-01 2.75942206e-01 3.90323609e-01 -1.07467949e+00
2.92562783e-01 3.45046163e-01 -1.21587448e-01 -2.67191947e-01
2.38282129e-01 -6.49556005e-03 -1.20054498e-01 7.46357217e-02
1.84077427e-01 5.40565670e-01 -2.70505905e-01 -2.79427588e-01
9.73877430e-01 1.12026483e-01 7.09588006e-02 -4.41944212e-01
4.22940046e-01 -1.63242951e-01 9.93049026e-01 3.41523260e-01
-2.97536999e-01 2.63209403e-01 1.80960983e-01 -5.86077929e-01
-1.00365102e+00 -9.02432740e-01 -1.17209926e-01 1.45488870e+00
2.87204623e-01 -1.14379078e-01 -8.77740562e-01 -7.11911917e-01
-1.55057341e-01 2.15048999e-01 -2.44280010e-01 -3.52954954e-01
-8.62708688e-01 -1.13377857e+00 6.65774703e-01 9.52095389e-01
1.11492753e+00 -9.12454724e-01 -5.94005227e-01 4.27496195e-01
1.43936768e-01 -1.15671813e+00 -4.48881924e-01 3.68449301e-01
-1.38121092e+00 -4.56453443e-01 -8.62499237e-01 -1.00689161e+00
5.74972332e-01 1.11722067e-01 1.10081267e+00 -8.06636140e-02
5.74511513e-02 4.89281230e-02 -9.82157290e-02 -3.10008228e-01
1.77176401e-01 4.40514922e-01 4.28836703e-01 -1.05404340e-01
5.06221831e-01 -9.63552535e-01 -9.26585019e-01 3.45719159e-01
-7.95609176e-01 1.05486684e-01 7.55049467e-01 9.77932990e-01
9.06089604e-01 -1.87195633e-02 8.69168460e-01 -1.03877199e+00
3.92718703e-01 -4.44506884e-01 -7.34954536e-01 6.39633685e-02
-6.75284088e-01 -7.42397681e-02 1.19963467e+00 -8.24032366e-01
-1.09699297e+00 6.05561361e-02 -5.69430053e-01 -7.89849818e-01
1.01100123e-02 2.31128842e-01 -6.12524927e-01 -2.21691206e-01
6.00079238e-01 3.20614815e-01 -3.54849070e-01 -6.64099693e-01
5.97788431e-02 6.12783015e-01 5.07126391e-01 -6.01014197e-01
3.20487201e-01 3.69319558e-01 -1.33499920e-01 -7.56656945e-01
-4.92176473e-01 -2.56633461e-01 -3.96485835e-01 1.40445560e-01
7.54090011e-01 -1.37596142e+00 -8.55556786e-01 6.66162670e-01
-1.15146172e+00 -6.35582745e-01 1.85284130e-02 3.49161208e-01
-9.91462246e-02 9.64315310e-02 -5.56827366e-01 -3.08030009e-01
-8.80470932e-01 -1.20949602e+00 9.30323482e-01 5.57667911e-01
1.53484046e-01 -1.16441464e+00 -2.20589370e-01 1.68900102e-01
6.85335159e-01 -1.40564060e-02 9.82333958e-01 -3.04763466e-01
-4.98734593e-01 2.64601022e-01 -6.39297903e-01 5.44530630e-01
-2.21059634e-03 -5.54034173e-01 -1.06829870e+00 -4.63040888e-01
6.64545447e-02 -1.79361463e-01 8.80141199e-01 5.60971081e-01
1.62131143e+00 -1.87499985e-01 -6.76542759e-01 1.11382592e+00
1.62916744e+00 5.01934469e-01 6.52153015e-01 2.29352444e-01
9.88282919e-01 -3.39680940e-01 2.79886842e-01 4.78628099e-01
3.15635204e-01 7.37193465e-01 2.88955808e-01 -2.49992028e-01
-1.43727049e-01 -7.50261098e-02 4.08900172e-01 1.18153381e+00
-1.74705669e-01 -1.54798254e-01 -7.26148486e-01 4.84384179e-01
-1.70909715e+00 -4.34819520e-01 2.44240716e-01 2.06115317e+00
8.21400881e-01 5.22246122e-01 1.65909514e-01 -1.75805405e-01
6.48965418e-01 1.69644225e-02 -1.23348498e+00 -2.57714748e-01
-1.36964336e-01 6.89165533e-01 7.16216981e-01 1.10700473e-01
-1.07581723e+00 9.05278265e-01 4.99342251e+00 1.13470006e+00
-1.57134175e+00 5.84697068e-01 7.23939121e-01 -2.82919824e-01
-1.01959288e-01 -4.40307885e-01 -1.20559263e+00 8.14664245e-01
1.01062000e+00 -1.03646472e-01 3.56561750e-01 1.38655448e+00
-6.00855313e-02 3.04560483e-01 -9.00658429e-01 1.49911189e+00
-2.19850019e-01 -1.48582697e+00 3.29703689e-02 -4.05758917e-02
7.78925180e-01 3.63786995e-01 4.20114808e-02 5.96445978e-01
-4.72248346e-03 -6.32124901e-01 4.88810003e-01 1.29609302e-01
1.29815459e+00 -7.96046138e-01 7.25866616e-01 3.05679172e-01
-1.42257583e+00 -9.65224504e-02 -5.35020232e-01 -1.33138731e-01
2.72764284e-02 8.86909425e-01 -3.25796872e-01 3.38244647e-01
1.01323915e+00 7.01170266e-01 -3.10292900e-01 5.73832512e-01
2.19927296e-01 7.15335071e-01 -4.33120072e-01 -1.35392085e-01
2.85426322e-02 -1.50382116e-01 2.37594396e-01 1.20573580e+00
5.10197639e-01 2.30934128e-01 2.05592513e-02 5.26240528e-01
-5.58979213e-01 -1.91753626e-01 -3.52463335e-01 1.51443377e-01
5.50468564e-01 1.20699275e+00 -6.95918024e-01 -4.92686182e-01
-5.00790179e-01 1.44333673e+00 3.42994511e-01 1.95792228e-01
-1.23252249e+00 -4.60599780e-01 1.04229212e+00 7.77043998e-02
2.81336516e-01 -4.00199294e-01 -6.18442476e-01 -1.17061341e+00
2.77766943e-01 -6.10945344e-01 2.36328736e-01 -4.20916647e-01
-1.11655247e+00 7.72029817e-01 -4.50507790e-01 -1.18869221e+00
2.56122530e-01 -8.40919971e-01 -3.43725413e-01 8.15456569e-01
-1.48403132e+00 -1.01057851e+00 -5.73041379e-01 7.46046066e-01
6.42693162e-01 -2.70687252e-01 8.12374473e-01 1.03859246e+00
-9.98117924e-01 1.13695872e+00 3.25551808e-01 9.82854664e-02
6.27272964e-01 -7.69545197e-01 5.91422856e-01 6.52687311e-01
-2.05822065e-01 6.36153042e-01 7.40636140e-02 -4.79450732e-01
-1.51160181e+00 -1.41732752e+00 2.32722938e-01 -7.66308457e-02
3.80709440e-01 -5.96084297e-01 -1.09505177e+00 6.53527319e-01
-4.84366305e-02 3.78949076e-01 5.86917043e-01 -7.05718771e-02
-1.23087376e-01 -6.83407426e-01 -1.09553981e+00 5.34190297e-01
1.65219080e+00 -4.63758081e-01 -1.39019161e-01 2.98030436e-01
1.02858448e+00 -6.70564651e-01 -7.82503545e-01 3.94831836e-01
4.88920897e-01 -8.14750731e-01 1.00862074e+00 -4.07104850e-01
2.30097532e-01 -1.91887021e-01 -1.90223694e-01 -1.10805142e+00
-4.64712858e-01 -4.74375218e-01 -3.75192851e-01 1.33605027e+00
2.86028236e-01 -8.83015394e-01 1.13438523e+00 4.50305551e-01
-4.76576269e-01 -9.84299481e-01 -1.20498896e+00 -7.97185957e-01
6.59000650e-02 -4.86636519e-01 5.90422571e-01 8.92260253e-01
-6.14586294e-01 5.23929238e-01 -2.39673108e-01 2.22137496e-01
3.08741212e-01 -7.71357268e-02 5.41664720e-01 -1.25879657e+00
-4.42986965e-01 -3.40503693e-01 -2.86101580e-01 -1.64177549e+00
2.98970360e-02 -6.94519043e-01 -3.89960766e-01 -1.24862361e+00
-5.15645221e-02 -7.77655363e-01 -4.52374130e-01 5.78601182e-01
-4.25387248e-02 2.64827371e-01 -8.16221312e-02 1.05956323e-01
-5.82868457e-01 5.64930499e-01 1.17202449e+00 3.41928415e-02
-4.43249345e-01 -4.44092900e-02 -7.68201530e-01 8.30066264e-01
6.63108110e-01 -3.53988111e-01 -5.51039457e-01 -1.04844892e+00
2.29559541e-01 -2.81481326e-01 3.12201142e-01 -1.45587301e+00
3.61039788e-01 3.22704427e-02 5.79360425e-01 -2.19690129e-01
3.13925803e-01 -9.79983509e-01 1.27040505e-01 3.56604248e-01
4.82032150e-01 3.09487849e-01 8.28321338e-01 4.40991193e-01
-8.94395411e-02 4.36202176e-02 9.34991062e-01 2.60389727e-02
-1.31649208e+00 5.51765919e-01 1.09995052e-01 8.63758773e-02
9.59472239e-01 -2.77357370e-01 -3.46346945e-01 2.82636106e-01
-5.54301620e-01 2.73161847e-02 2.26120964e-01 3.12829733e-01
5.16541958e-01 -1.50541890e+00 -1.47962719e-01 4.70758528e-01
-2.14013934e-01 4.60482925e-01 7.40458250e-01 5.65781295e-01
-5.02688527e-01 2.66896278e-01 -3.49279195e-01 -6.70598388e-01
-9.77205038e-01 4.48674858e-01 3.82489085e-01 -3.70523065e-01
-5.49077511e-01 1.13003278e+00 3.52557272e-01 -2.38628983e-01
4.66905907e-02 -5.21050751e-01 1.40954539e-01 -1.17857635e-01
4.22479004e-01 4.57908839e-01 3.34661692e-01 -1.43647924e-01
-5.68964839e-01 8.23501289e-01 -2.70000845e-01 3.03121924e-01
1.29891455e+00 2.25001406e-02 1.94460943e-01 2.03398719e-01
1.35956669e+00 -2.95056194e-01 -1.47923827e+00 -2.68374920e-01
-4.07029778e-01 -2.37716995e-02 2.55297989e-01 -7.29007423e-01
-1.33862162e+00 9.67268825e-01 1.04666865e+00 -2.82051265e-01
1.71303892e+00 -2.19178751e-01 1.36069345e+00 2.26616815e-01
5.34122288e-01 -1.28650057e+00 9.91524160e-02 6.44203782e-01
3.97711068e-01 -1.25717461e+00 -9.03452784e-02 -1.93846494e-01
-5.24235427e-01 9.19146478e-01 1.19565380e+00 -6.95122257e-02
8.89225721e-01 5.11917353e-01 -1.96626633e-01 1.21144712e-01
-4.40377861e-01 2.28602305e-01 5.53406496e-03 5.16793966e-01
2.55540341e-01 6.25179559e-02 -5.01239188e-02 1.26561666e+00
9.72234309e-02 3.99811596e-01 -8.71169791e-02 6.61905050e-01
-2.18461797e-01 -1.03656924e+00 4.19486538e-02 6.22068405e-01
-4.18622345e-01 -1.56830579e-01 3.03401291e-01 6.59527361e-01
7.43560255e-01 3.56311977e-01 3.88573021e-01 -6.43639386e-01
5.14816046e-01 9.27002057e-02 2.03612521e-01 -4.38661486e-01
-6.35926485e-01 -7.83168226e-02 -2.44273543e-01 -3.54813516e-01
-2.38704473e-01 -2.82059163e-01 -1.44144285e+00 -4.75776613e-01
-1.62501797e-01 -4.02077019e-01 4.96006697e-01 9.29267526e-01
8.96314561e-01 9.27712142e-01 2.93585777e-01 -7.85312235e-01
-2.17383265e-01 -7.91911721e-01 -3.05972189e-01 2.31220663e-01
1.38616011e-01 -8.14971685e-01 5.02896942e-02 -5.10796309e-02] | [8.77720832824707, 2.909540891647339] |
6c522662-a8a6-4ae4-9da7-123818bb089a | pymaf-x-towards-well-aligned-full-body-model | 2207.06400 | null | https://arxiv.org/abs/2207.06400v3 | https://arxiv.org/pdf/2207.06400v3.pdf | PyMAF-X: Towards Well-aligned Full-body Model Regression from Monocular Images | We present PyMAF-X, a regression-based approach to recovering parametric full-body models from monocular images. This task is very challenging since minor parametric deviation may lead to noticeable misalignment between the estimated mesh and the input image. Moreover, when integrating part-specific estimations into the full-body model, existing solutions tend to either degrade the alignment or produce unnatural wrist poses. To address these issues, we propose a Pyramidal Mesh Alignment Feedback (PyMAF) loop in our regression network for well-aligned human mesh recovery and extend it as PyMAF-X for the recovery of expressive full-body models. The core idea of PyMAF is to leverage a feature pyramid and rectify the predicted parameters explicitly based on the mesh-image alignment status. Specifically, given the currently predicted parameters, mesh-aligned evidence will be extracted from finer-resolution features accordingly and fed back for parameter rectification. To enhance the alignment perception, an auxiliary dense supervision is employed to provide mesh-image correspondence guidance while spatial alignment attention is introduced to enable the awareness of the global contexts for our network. When extending PyMAF for full-body mesh recovery, an adaptive integration strategy is proposed in PyMAF-X to produce natural wrist poses while maintaining the well-aligned performance of the part-specific estimations. The efficacy of our approach is validated on several benchmark datasets for body, hand, face, and full-body mesh recovery, where PyMAF and PyMAF-X effectively improve the mesh-image alignment and achieve new state-of-the-art results. The project page with code and video results can be found at https://www.liuyebin.com/pymaf-x. | ['Yebin Liu', 'Zhenan Sun', 'Liang An', 'Mengcheng Li', 'Yuxiang Zhang', 'Yating Tian', 'Hongwen Zhang'] | 2022-07-13 | null | null | null | null | ['markerless-motion-capture', '3d-human-pose-and-shape-estimation', 'human-mesh-recovery'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 2.99698204e-01 2.63138592e-01 -3.81971657e-01 -3.51854146e-01
-7.07866013e-01 -6.07700013e-02 1.72372848e-01 -4.65267777e-01
1.11152627e-01 4.57547605e-01 4.35111493e-01 4.46904123e-01
-1.12512819e-01 -5.53624988e-01 -9.96950507e-01 -4.38674748e-01
2.93237627e-01 6.02328598e-01 4.71957326e-02 -3.15914541e-01
-5.93722574e-02 4.49111462e-01 -1.51683879e+00 1.44482777e-01
8.05539012e-01 8.96353960e-01 2.87541449e-01 3.73802900e-01
1.85929343e-01 1.35187447e-01 -1.78366676e-01 -3.47370803e-01
4.80734974e-01 -1.55370191e-01 -6.94844723e-01 3.70995939e-01
1.03671575e+00 -5.99562943e-01 -3.64373744e-01 8.25621665e-01
6.16027594e-01 -1.12102538e-01 5.29882252e-01 -1.11485553e+00
-1.98942289e-01 2.63144195e-01 -1.01786196e+00 -3.74234527e-01
6.33996487e-01 2.61073679e-01 6.79483593e-01 -1.13995516e+00
1.00507581e+00 1.48846734e+00 8.65938127e-01 6.36402726e-01
-1.36714995e+00 -9.74650562e-01 2.91064560e-01 -1.20678961e-01
-1.42722058e+00 -6.91218913e-01 9.85126019e-01 -3.60349476e-01
5.96784711e-01 3.00645292e-01 9.15162683e-01 9.80929911e-01
2.46374205e-01 6.20802641e-01 6.52318656e-01 -2.68264592e-01
-2.95286745e-01 -3.68423074e-01 -3.39290142e-01 9.95735526e-01
7.91980028e-02 1.90398514e-01 -8.18316936e-01 -1.12049885e-01
1.39480329e+00 4.67900597e-02 -4.61125046e-01 -7.32783556e-01
-1.21432984e+00 2.81807572e-01 5.82503855e-01 -1.15716822e-01
-6.43489420e-01 2.52190828e-01 2.29768947e-01 -6.95238709e-02
5.20845175e-01 2.15170667e-01 -3.61993015e-01 1.18801214e-01
-1.01892364e+00 6.18082345e-01 3.69308561e-01 9.10432041e-01
8.11174452e-01 -5.75275766e-03 -1.71468377e-01 1.00256753e+00
3.21948558e-01 6.03667140e-01 7.72104189e-02 -1.34352219e+00
6.94452107e-01 7.77650297e-01 4.89822440e-02 -1.16806829e+00
-5.64902008e-01 -4.73369628e-01 -9.62531745e-01 2.41022184e-01
4.04969752e-01 2.44345561e-01 -1.02507102e+00 1.69758141e+00
8.82660687e-01 1.01557501e-01 -5.24913192e-01 1.26408255e+00
8.73827279e-01 2.32071444e-01 -1.09954290e-01 3.36120017e-02
1.36580050e+00 -9.77347970e-01 -5.42209446e-01 -4.64294702e-01
2.58511007e-01 -7.12547064e-01 1.21508050e+00 3.03314030e-01
-1.19274414e+00 -6.76771402e-01 -8.12197089e-01 -1.78023741e-01
2.80844539e-01 4.45359439e-01 2.01856613e-01 4.07431126e-02
-7.95229852e-01 7.45033920e-01 -9.72386360e-01 -1.84124738e-01
4.14917648e-01 6.00560784e-01 -7.59969592e-01 -2.91403979e-01
-7.86163509e-01 6.54536247e-01 9.31270123e-02 4.51882571e-01
-5.25151551e-01 -8.16929281e-01 -1.19184518e+00 -2.80279517e-01
5.94276309e-01 -1.20132542e+00 1.00104570e+00 -8.97177994e-01
-1.51988757e+00 8.91988218e-01 -1.82115123e-01 5.50381541e-02
8.28242183e-01 -5.51501691e-01 -1.96950678e-02 2.33238205e-01
1.89323008e-01 1.06682658e+00 1.20893240e+00 -1.34439635e+00
-2.29691446e-01 -4.76067960e-01 -2.32317254e-01 3.83432597e-01
4.45237756e-02 -3.66455913e-01 -8.30577254e-01 -1.06988251e+00
3.68050605e-01 -1.02621913e+00 -1.59580588e-01 7.22117841e-01
-5.28382540e-01 8.48439336e-03 6.73841059e-01 -1.14128697e+00
1.11639988e+00 -2.05687714e+00 6.78995907e-01 3.35947096e-01
2.45559782e-01 -8.11773315e-02 -2.33362377e-01 3.11756562e-02
-1.83986276e-01 -3.42578292e-01 -3.97303015e-01 -6.68265402e-01
-1.56540334e-01 4.02060747e-01 9.72726196e-03 7.20121443e-01
2.17077538e-01 1.07981050e+00 -6.49656892e-01 -5.90693414e-01
3.92121047e-01 6.47820473e-01 -8.04445505e-01 1.91694856e-01
-2.62687892e-01 9.37902212e-01 -2.86909878e-01 9.92926359e-01
8.30327570e-01 -2.67744005e-01 1.12800993e-01 -8.59827876e-01
1.61756858e-01 -1.27641723e-01 -1.38610923e+00 2.07868743e+00
-4.30635154e-01 3.73079851e-02 6.01713717e-01 -5.24730444e-01
6.29986346e-01 1.63587227e-01 6.96510792e-01 -6.77585542e-01
1.59099594e-01 2.03848749e-01 -2.50341773e-01 -2.63045192e-01
3.39795530e-01 -7.37004122e-03 1.52387396e-01 1.56884179e-01
-1.10289149e-01 -2.25442037e-01 -2.37274453e-01 -1.11885190e-01
5.39179862e-01 8.77022505e-01 2.45331511e-01 -5.21689430e-02
5.46525180e-01 -2.85594136e-01 7.86334932e-01 1.22853599e-01
7.12603629e-02 9.36184704e-01 9.34797525e-03 -4.40021574e-01
-9.54527617e-01 -1.01188934e+00 -8.67264122e-02 8.53741348e-01
2.31954783e-01 -6.84846580e-01 -7.28045583e-01 -5.75820684e-01
1.68682396e-01 1.01964884e-01 -6.76145434e-01 -9.39959139e-02
-9.66727138e-01 -2.39636108e-01 1.85378641e-01 6.95910215e-01
4.99485850e-01 -1.06945908e+00 -5.23471773e-01 1.73918977e-01
-4.52368587e-01 -1.01871955e+00 -8.54563236e-01 -2.96411514e-01
-8.95902038e-01 -1.13368690e+00 -9.22385871e-01 -5.66207230e-01
9.06574547e-01 -7.12216347e-02 1.04665887e+00 2.90075600e-01
-4.86923009e-01 2.02979326e-01 4.93884943e-02 2.01938003e-01
-3.36042017e-01 7.57105798e-02 1.25631139e-01 -7.11796656e-02
-4.83540118e-01 -8.48127246e-01 -9.99858975e-01 6.74460113e-01
-6.88275695e-01 4.42444116e-01 6.08080804e-01 8.28082144e-01
9.17378247e-01 -3.76693517e-01 2.50745595e-01 -5.69641769e-01
1.28902923e-02 -2.00661078e-01 -3.63996774e-01 -2.63573304e-02
-4.06890482e-01 -7.47856870e-02 3.60875010e-01 -6.72621548e-01
-9.06767249e-01 4.68407363e-01 -3.89343947e-01 -9.74558771e-01
1.00302614e-01 1.67695284e-01 -4.61760610e-01 -1.04667030e-01
4.30723369e-01 1.30891586e-02 2.33424351e-01 -7.20234036e-01
3.07704747e-01 3.05572420e-01 9.50764477e-01 -7.58701503e-01
8.19278181e-01 5.20256460e-01 6.42236322e-02 -5.80698490e-01
-8.04226756e-01 -3.32387775e-01 -8.22507858e-01 -4.54071641e-01
7.47541189e-01 -1.03907120e+00 -4.53723341e-01 6.58046305e-01
-9.60916460e-01 -4.58961695e-01 -2.22437605e-01 3.24330062e-01
-6.80194199e-01 5.23038447e-01 -5.58363616e-01 -3.88910204e-01
-6.14047587e-01 -1.19600821e+00 1.70362198e+00 -2.10008342e-02
-4.84363317e-01 -6.39537275e-01 -1.18876919e-01 6.08281195e-01
6.02488890e-02 4.00798559e-01 8.19542348e-01 2.27314383e-01
-4.55986172e-01 -3.90024222e-02 -7.63179064e-02 2.09424421e-01
3.37693781e-01 -1.03254877e-01 -7.65675604e-01 -4.22115624e-01
-4.58699286e-01 -3.23071659e-01 5.21964371e-01 5.18034756e-01
1.16241992e+00 -4.10076439e-01 -3.73317361e-01 8.98826778e-01
9.79780674e-01 -6.62240326e-01 5.24242222e-01 1.85991898e-01
1.13219881e+00 7.37591267e-01 9.72433150e-01 5.36363840e-01
4.55729485e-01 1.20958972e+00 5.86977065e-01 -2.02857211e-01
-5.34431636e-01 -6.56148016e-01 2.95431346e-01 6.54689014e-01
-1.89259410e-01 3.71702611e-01 -6.50063217e-01 2.25823820e-01
-1.74666464e+00 -5.34403503e-01 8.10301453e-02 2.19617772e+00
1.04878008e+00 -6.05676062e-02 3.63945633e-01 -4.94364835e-03
6.31576836e-01 1.46031097e-01 -7.89767265e-01 2.12751403e-01
1.11525990e-01 2.06167325e-01 2.42744505e-01 6.05587900e-01
-9.90430892e-01 9.90859151e-01 5.26808357e+00 9.05083895e-01
-1.27340877e+00 2.05728561e-02 3.17965686e-01 -2.37318471e-01
-1.80529848e-01 -1.24617815e-01 -6.59865856e-01 2.78729945e-01
1.97806552e-01 3.75025660e-01 2.95088589e-01 6.79067314e-01
2.30295256e-01 -9.44496989e-02 -1.02803993e+00 1.26070893e+00
1.99356978e-03 -1.10733318e+00 5.37384339e-02 2.15213478e-01
5.31519353e-01 -2.40425199e-01 -1.32523939e-01 8.32557976e-02
-1.98138043e-01 -1.03161561e+00 8.95115197e-01 7.33109057e-01
1.22131753e+00 -6.28533840e-01 3.04023117e-01 4.24425483e-01
-1.38638830e+00 1.48593575e-01 -2.30821416e-01 4.45838980e-02
4.15019631e-01 5.45037746e-01 -5.49412668e-01 8.36323202e-01
8.03419054e-01 8.79658222e-01 -5.60264885e-01 7.66028941e-01
-1.35629922e-01 1.85298339e-01 -5.07092834e-01 8.75286996e-01
-3.75745118e-01 -1.77262142e-01 7.22558260e-01 8.25382590e-01
8.76680613e-02 -8.16762745e-02 3.85053962e-01 9.07057703e-01
-5.61351813e-02 2.41093829e-01 -2.42672876e-01 5.00528336e-01
3.04674059e-01 1.26378584e+00 -4.86416996e-01 -2.50861108e-01
-4.55278233e-02 1.14073980e+00 3.82003605e-01 1.21309370e-01
-7.83291519e-01 2.93242007e-01 6.67604268e-01 8.02948952e-01
2.66169280e-01 -6.33984618e-03 -1.70690760e-01 -1.14936829e+00
4.05164838e-01 -1.07615554e+00 2.37361208e-01 -9.45286334e-01
-1.04296291e+00 3.91921312e-01 2.69052148e-01 -1.28746986e+00
-3.56839925e-01 -2.24089637e-01 -2.52909094e-01 7.30823994e-01
-1.22135508e+00 -1.52740073e+00 -7.12502301e-01 6.32788837e-01
6.60212159e-01 2.84385055e-01 6.49378598e-01 3.65280718e-01
-5.29785335e-01 7.28556693e-01 -6.21572971e-01 6.56472892e-02
9.52898443e-01 -8.26510251e-01 3.48684549e-01 4.90275592e-01
-8.71257931e-02 5.12825966e-01 6.49059236e-01 -1.16255784e+00
-1.40531445e+00 -1.26012683e+00 2.24777699e-01 -2.31364667e-01
5.88899851e-03 -3.48579377e-01 -9.65407670e-01 6.36414647e-01
-3.30078125e-01 3.41976672e-01 1.48252428e-01 -1.65491834e-01
-4.48794812e-01 -1.79071531e-01 -1.15360343e+00 6.36566520e-01
1.47923064e+00 -2.27636829e-01 -4.67808723e-01 1.98659346e-01
3.86929750e-01 -1.03540754e+00 -1.13644159e+00 7.99185812e-01
1.03093636e+00 -7.84658730e-01 1.28290856e+00 -2.10198537e-01
5.03812194e-01 -4.44011718e-01 -8.73127654e-02 -1.04803169e+00
-2.20748842e-01 -6.01033807e-01 -4.08314854e-01 1.07035601e+00
-7.37864748e-02 -3.49759430e-01 9.89704192e-01 5.32215834e-01
-2.30454821e-02 -1.10297740e+00 -1.14558697e+00 -4.87923831e-01
-5.58243468e-02 -2.51843274e-01 4.59227562e-01 7.40645647e-01
-4.12376285e-01 5.07483669e-02 -6.50078118e-01 2.19985008e-01
6.87413573e-01 2.40256205e-01 1.22104776e+00 -1.04606175e+00
-4.37568426e-01 -1.88962623e-01 -3.00132424e-01 -1.33200324e+00
2.28951216e-01 -7.16674387e-01 8.23705643e-02 -1.40314591e+00
7.69039690e-02 -4.20788974e-01 2.45615333e-01 7.74238527e-01
-2.46237755e-01 7.08230555e-01 3.16544712e-01 4.48942304e-01
-1.26431733e-01 6.46722496e-01 1.65434909e+00 -4.73044580e-03
-2.42557973e-01 8.80709067e-02 -3.71562988e-01 8.82954121e-01
4.24971402e-01 -3.21673661e-01 -2.74526060e-01 -2.51529098e-01
8.90311301e-02 2.58683503e-01 7.44830906e-01 -8.42423856e-01
1.35899596e-02 -1.06290452e-01 7.18633771e-01 -8.04927289e-01
7.19712019e-01 -8.02468836e-01 6.45829439e-01 4.13640678e-01
5.39157577e-02 1.11336820e-01 1.02863304e-01 5.00873804e-01
8.95528402e-03 2.20524415e-01 7.75122404e-01 -1.12141669e-01
-3.39768946e-01 5.87624490e-01 3.16215724e-01 -2.71325074e-02
7.27783918e-01 -4.56988305e-01 1.46354496e-01 -4.36240882e-01
-9.24857974e-01 3.37717354e-01 9.91462886e-01 5.32863259e-01
9.75339174e-01 -1.40183258e+00 -6.90532267e-01 5.44055402e-01
6.73535019e-02 6.37145102e-01 4.88965422e-01 1.21748376e+00
-4.81026053e-01 -1.28858000e-01 -1.87979847e-01 -9.76459205e-01
-1.57060981e+00 2.51339048e-01 6.31313920e-01 -2.55877376e-01
-1.06296146e+00 5.12167633e-01 5.37768900e-01 -6.97643340e-01
3.00975263e-01 -4.16526288e-01 1.85082272e-01 -2.88672090e-01
2.67126858e-01 3.89432281e-01 2.78951041e-02 -9.87518370e-01
-4.09095794e-01 1.35378206e+00 1.62337944e-02 6.74257651e-02
1.28286827e+00 -2.34974131e-01 -8.93826932e-02 -5.64129315e-02
9.90572572e-01 4.37336296e-01 -1.64502716e+00 -2.13504478e-01
-5.08978605e-01 -5.91777682e-01 -1.70242533e-01 -8.31116915e-01
-1.41772008e+00 6.37182236e-01 5.06470919e-01 -9.33860660e-01
1.16688573e+00 1.87387317e-01 9.00945306e-01 -1.35919988e-01
3.99931550e-01 -7.71307468e-01 1.20423280e-01 4.97671664e-02
1.49433613e+00 -9.16320384e-01 4.91671234e-01 -9.83537793e-01
-4.31721389e-01 9.90423024e-01 9.19642627e-01 -1.89973235e-01
6.00137591e-01 1.79604843e-01 4.99846563e-02 -4.11885589e-01
-3.31038892e-01 1.83027983e-01 6.29878640e-01 5.48746109e-01
9.93429795e-02 -1.03939310e-01 2.04497371e-02 4.79633987e-01
-4.88800049e-01 1.22849150e-02 -1.02093697e-01 8.41008782e-01
-1.31874466e-02 -1.13852549e+00 -7.64088154e-01 3.32955897e-01
-2.31562838e-01 1.89378157e-01 -2.49640957e-01 7.63189852e-01
2.35618114e-01 4.67658669e-01 -6.87135309e-02 -2.83457309e-01
6.09643161e-01 -1.30496591e-01 8.07237566e-01 -6.41864240e-01
-5.20649433e-01 5.47904968e-01 1.28424913e-01 -1.05924571e+00
-4.25832301e-01 -5.60302675e-01 -1.24916506e+00 -1.72828734e-01
-2.48810425e-01 -4.61635500e-01 3.96188140e-01 7.46579766e-01
5.82657814e-01 4.66308773e-01 2.07021102e-01 -1.48947382e+00
-4.29122657e-01 -9.20817852e-01 -3.10370386e-01 5.22955239e-01
3.62704515e-01 -1.12426424e+00 -6.31992742e-02 1.06878169e-02] | [7.0965375900268555, -1.1946470737457275] |
a3df19cf-7c1c-412f-a9af-afdfe862f056 | seeing-what-you-said-talking-face-generation | 2303.17480 | null | https://arxiv.org/abs/2303.17480v1 | https://arxiv.org/pdf/2303.17480v1.pdf | Seeing What You Said: Talking Face Generation Guided by a Lip Reading Expert | Talking face generation, also known as speech-to-lip generation, reconstructs facial motions concerning lips given coherent speech input. The previous studies revealed the importance of lip-speech synchronization and visual quality. Despite much progress, they hardly focus on the content of lip movements i.e., the visual intelligibility of the spoken words, which is an important aspect of generation quality. To address the problem, we propose using a lip-reading expert to improve the intelligibility of the generated lip regions by penalizing the incorrect generation results. Moreover, to compensate for data scarcity, we train the lip-reading expert in an audio-visual self-supervised manner. With a lip-reading expert, we propose a novel contrastive learning to enhance lip-speech synchronization, and a transformer to encode audio synchronically with video, while considering global temporal dependency of audio. For evaluation, we propose a new strategy with two different lip-reading experts to measure intelligibility of the generated videos. Rigorous experiments show that our proposal is superior to other State-of-the-art (SOTA) methods, such as Wav2Lip, in reading intelligibility i.e., over 38% Word Error Rate (WER) on LRS2 dataset and 27.8% accuracy on LRW dataset. We also achieve the SOTA performance in lip-speech synchronization and comparable performances in visual quality. | ['Haizhou Li', 'Robby T. Tan', 'Malu Zhang', 'Xinyuan Qian', 'Jiadong Wang'] | 2023-03-29 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Seeing_What_You_Said_Talking_Face_Generation_Guided_by_a_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Seeing_What_You_Said_Talking_Face_Generation_Guided_by_a_CVPR_2023_paper.pdf | cvpr-2023-1 | ['talking-face-generation', 'face-generation'] | ['computer-vision', 'computer-vision'] | [ 5.10458350e-02 1.14619665e-01 -3.10960382e-01 -4.04604971e-02
-1.18904936e+00 -6.86943009e-02 3.73628318e-01 -6.53258979e-01
-1.54844061e-01 6.40663087e-01 5.19716620e-01 1.41838044e-01
1.26169622e-01 -2.01670200e-01 -6.39556646e-01 -8.81134927e-01
4.91237670e-01 -1.01022452e-01 1.19266346e-01 2.26243529e-02
2.67914474e-01 3.12902927e-01 -2.06979728e+00 3.09315294e-01
9.59875345e-01 1.18099618e+00 4.46397066e-01 7.34629154e-01
-9.11863223e-02 6.32239997e-01 -6.27360880e-01 -3.73393387e-01
-8.98126364e-02 -6.15656018e-01 -4.57862407e-01 2.70557433e-01
5.27526796e-01 -3.93075109e-01 -1.12968780e-01 1.05861604e+00
1.10080492e+00 -6.19213283e-02 6.79711580e-01 -1.42186415e+00
-5.59025526e-01 3.54538590e-01 -5.33483863e-01 -1.38272434e-01
6.18844509e-01 5.39212823e-01 6.95340097e-01 -1.14107621e+00
6.35780692e-01 1.44219136e+00 4.69200075e-01 9.54401195e-01
-1.03729427e+00 -6.86587036e-01 -2.80502826e-01 5.60337663e-01
-1.47255528e+00 -1.42516720e+00 9.66971755e-01 -2.14214489e-01
5.09405315e-01 2.50986308e-01 3.53708953e-01 1.14480889e+00
-1.22322679e-01 8.43213797e-01 9.44165587e-01 -5.05509496e-01
1.81186956e-03 1.61256060e-01 -4.24709767e-01 3.70993644e-01
-3.40294063e-01 2.09010318e-01 -9.73065615e-01 4.03437793e-01
4.00860161e-01 -6.91856563e-01 -7.36501813e-01 -6.89166486e-02
-1.21246326e+00 4.81534064e-01 -1.35178342e-01 2.28234962e-01
-1.75038844e-01 4.87105101e-02 3.59195858e-01 1.50842398e-01
5.83822012e-01 -3.47702429e-02 -1.77817345e-01 -2.45509207e-01
-1.10296547e+00 -6.94279745e-02 3.71034056e-01 8.08537066e-01
2.98949659e-01 3.97151858e-01 -4.93867755e-01 1.03222942e+00
6.39121056e-01 9.53656256e-01 7.45123804e-01 -9.73800421e-01
5.57338893e-01 -4.90425341e-02 1.30321443e-01 -7.59822965e-01
-2.26105139e-01 -3.40995826e-02 -8.99738193e-01 5.00389814e-01
2.91615516e-01 -2.86652207e-01 -6.29469335e-01 2.05550289e+00
3.42518330e-01 3.41494948e-01 3.16203296e-01 1.04604805e+00
9.68213320e-01 8.06817055e-01 -1.84597876e-02 -8.81826103e-01
1.17402315e+00 -1.02918243e+00 -1.31639266e+00 2.34274730e-01
2.49933880e-02 -1.29937530e+00 1.18947864e+00 4.28790808e-01
-1.64802241e+00 -1.05521631e+00 -7.06206083e-01 6.40039295e-02
1.43746123e-01 7.22424686e-01 1.44760273e-02 7.21661270e-01
-1.40550077e+00 2.92317271e-01 -4.30153370e-01 -1.87211618e-01
1.05799183e-01 1.51142344e-01 -4.00409162e-01 3.04369658e-01
-1.15802109e+00 7.72649705e-01 1.00598082e-01 -1.39920503e-01
-8.19855332e-01 -6.58535421e-01 -9.17377651e-01 -3.31498422e-02
2.29494974e-01 -4.42923576e-01 1.15406120e+00 -9.86230314e-01
-2.27554822e+00 7.36474514e-01 -6.09955072e-01 -1.24570765e-01
7.43994653e-01 -6.81407601e-02 -6.44937277e-01 3.99503887e-01
-1.09176159e-01 1.02329206e+00 1.40689278e+00 -1.39682698e+00
-4.65005368e-01 1.13284998e-02 -5.39251745e-01 4.61010605e-01
-2.95382053e-01 9.92826074e-02 -5.66805422e-01 -6.39435768e-01
-8.44721049e-02 -6.84962690e-01 5.93664885e-01 2.84057766e-01
-3.32706422e-01 -5.38614333e-01 1.04950142e+00 -9.47424948e-01
1.13750422e+00 -2.39743710e+00 6.20034523e-02 -3.35505962e-01
-6.83900043e-02 6.74152255e-01 -3.90535116e-01 1.72589608e-02
-1.25050768e-01 1.01028152e-01 2.35994011e-02 -8.02146256e-01
-7.76266539e-03 -2.85353273e-01 -2.54058987e-01 5.52439213e-01
1.93128139e-01 7.65125334e-01 -7.34851182e-01 -8.81927788e-01
4.26233292e-01 7.68204808e-01 -4.24310505e-01 5.90998471e-01
-7.68843964e-02 3.49114925e-01 3.88049819e-02 6.46622360e-01
9.06320512e-01 2.58537680e-01 -1.23922199e-01 -4.32255954e-01
-1.43016249e-01 9.78601724e-02 -1.20238256e+00 1.58971298e+00
-7.82324374e-01 8.53141248e-01 3.77293885e-01 -6.39315963e-01
8.15822840e-01 8.46435905e-01 4.14598078e-01 -6.24354899e-01
-6.53453916e-03 3.37141514e-01 -1.87763900e-01 -9.70121264e-01
1.99878767e-01 -2.02666018e-02 5.79680324e-01 2.49313891e-01
2.98608541e-01 -3.75765949e-01 -5.07444777e-02 -2.27915198e-01
3.96058530e-01 1.37633219e-01 1.09998710e-01 -2.70936303e-02
9.44205523e-01 -8.65474105e-01 1.27485931e-01 8.70491266e-02
-3.58005226e-01 8.90035629e-01 3.66876304e-01 3.29450190e-01
-9.69403625e-01 -1.39299214e+00 -9.61421877e-02 9.05525029e-01
2.12607771e-01 -2.34129459e-01 -1.08514082e+00 -3.22994530e-01
-4.37071502e-01 5.90710998e-01 -1.17604576e-01 -9.23472568e-02
-3.90124321e-01 -1.35056242e-01 7.19417870e-01 1.85983047e-01
6.70413911e-01 -1.39565372e+00 -2.60904312e-01 -2.35856045e-02
-8.25217605e-01 -1.27493131e+00 -8.14207911e-01 -6.07524395e-01
-1.67085603e-01 -8.20441306e-01 -1.33709311e+00 -9.21300590e-01
3.83268923e-01 9.90590304e-02 7.60779202e-01 -1.64974913e-01
-1.46420598e-01 3.42062920e-01 -5.01801446e-02 -1.59600005e-01
-8.81473541e-01 -1.55426145e-01 4.06813771e-01 3.55341882e-01
-2.33967453e-01 -4.20768678e-01 -6.20178699e-01 5.60540855e-01
-6.77708626e-01 1.94112808e-01 5.62433481e-01 8.60030293e-01
4.34125453e-01 -5.80732189e-02 9.54407275e-01 2.50894815e-01
8.90437901e-01 -1.03221163e-01 -4.65841085e-01 3.77903819e-01
-4.13950294e-01 -6.72085136e-02 5.52032173e-01 -8.90773118e-01
-1.29011595e+00 -8.68015885e-02 -6.05165303e-01 -7.52264917e-01
-2.79288828e-01 -1.16182804e-01 -5.87838709e-01 3.01810801e-02
3.76072735e-01 4.96130109e-01 3.13978791e-01 -3.21149439e-01
2.71065146e-01 1.17497456e+00 7.35421240e-01 -1.34034634e-01
5.21768034e-01 1.16637640e-01 -6.81508258e-02 -1.24838305e+00
-2.14323893e-01 -4.71259922e-01 -7.64131621e-02 -5.40610433e-01
9.96278286e-01 -8.61255050e-01 -1.24004281e+00 9.10264254e-01
-1.45045030e+00 -2.79765099e-01 -2.39259303e-01 7.02452183e-01
-1.03425956e+00 6.11217678e-01 -3.23938727e-01 -1.17123628e+00
-4.08949822e-01 -1.46788013e+00 1.31943238e+00 1.16440997e-01
2.78352909e-02 -5.41272223e-01 -2.27298141e-02 5.30648589e-01
5.21535695e-01 -1.95435137e-01 4.73250002e-01 -7.57424459e-02
-1.83865562e-01 2.39474744e-01 -2.63637394e-01 6.48359478e-01
3.55556577e-01 6.70819506e-02 -1.35880017e+00 -6.54486939e-02
-5.26879840e-02 -4.75316375e-01 6.90938830e-01 6.60901785e-01
1.05024314e+00 -5.38709402e-01 -6.01604953e-02 3.93458962e-01
9.82748926e-01 1.16091140e-01 9.01948512e-01 -5.93949080e-01
2.79453516e-01 8.77540231e-01 5.92767060e-01 3.88349503e-01
9.27788690e-02 1.16272879e+00 4.08677012e-01 -2.26899832e-01
-9.97232020e-01 -4.42853391e-01 7.71039605e-01 8.09761882e-01
-4.87738773e-02 -5.87064624e-01 -5.21480560e-01 5.47787905e-01
-1.60601103e+00 -1.10945809e+00 1.88175216e-01 2.15147829e+00
1.10592520e+00 -1.69784218e-01 8.73091817e-02 4.42907393e-01
1.00877988e+00 2.90484071e-01 -4.14658308e-01 -1.48927853e-01
-2.61909932e-01 1.32434722e-02 1.64316539e-02 9.83161449e-01
-7.54087329e-01 1.11117959e+00 5.76606083e+00 1.51003587e+00
-1.50299025e+00 1.66263580e-01 6.22765183e-01 -3.66213918e-02
-1.88343376e-01 -7.18019962e-01 -7.30882764e-01 6.53354943e-01
8.00976455e-01 -9.35112610e-02 6.04601920e-01 5.89402735e-01
8.21750820e-01 -9.06694755e-02 -8.52144480e-01 1.42884088e+00
5.34895778e-01 -1.09297216e+00 -3.17950244e-03 -1.98949590e-01
5.39889395e-01 -3.96402448e-01 4.04376179e-01 -1.20712034e-01
-5.04422426e-01 -1.11583638e+00 8.64292443e-01 7.18600154e-01
1.43593645e+00 -6.79724514e-01 5.16061842e-01 2.69744694e-01
-1.34377897e+00 2.06298903e-01 -1.32762849e-01 5.27201712e-01
2.97268867e-01 2.30767652e-01 -8.69531631e-01 3.67279798e-01
5.17026186e-01 3.39705497e-01 -2.52053052e-01 8.47570539e-01
-4.95765150e-01 4.69196171e-01 -1.77819893e-01 7.74721205e-02
-2.70484000e-01 3.42398137e-01 8.11874866e-01 1.03801560e+00
5.08673072e-01 -1.10370398e-01 -3.36867750e-01 7.54885137e-01
-1.14239417e-01 3.17132443e-01 -5.38428962e-01 1.35846436e-01
2.90917128e-01 9.98281062e-01 -1.26102522e-01 -1.96525037e-01
1.57670565e-02 9.35933888e-01 -3.06396574e-01 4.53282207e-01
-9.06749368e-01 -3.28018248e-01 7.21816421e-01 2.60966808e-01
2.75283195e-02 9.84492451e-02 1.25719458e-01 -9.63216603e-01
2.42146835e-01 -9.44000006e-01 -3.51995736e-01 -1.32627463e+00
-8.18635762e-01 7.09978342e-01 -1.94574147e-01 -1.33685577e+00
-6.54057503e-01 -3.78287882e-01 -4.97618228e-01 8.58049333e-01
-1.73187864e+00 -1.12294316e+00 -3.70828986e-01 1.05499315e+00
1.03238392e+00 -3.19500923e-01 5.25391817e-01 3.14968556e-01
-1.67360082e-01 1.01589394e+00 -1.42647132e-01 -1.81438476e-01
1.14782119e+00 -7.18196809e-01 6.69092312e-03 7.44092882e-01
-2.15665512e-02 9.11147445e-02 7.50333130e-01 -3.28224689e-01
-1.16895521e+00 -8.53583395e-01 1.15142989e+00 2.18008712e-01
2.60907680e-01 -2.25825682e-01 -5.67268848e-01 -5.66710830e-02
4.74193275e-01 1.06911808e-02 3.05873305e-01 -5.97272515e-01
-1.53300203e-02 -2.79638588e-01 -1.30654001e+00 7.16515779e-01
9.95384693e-01 -6.13973141e-01 -3.97999942e-01 1.90969750e-01
8.14066887e-01 -2.64176130e-01 -4.73619282e-01 4.61188555e-01
6.21837735e-01 -1.05863607e+00 9.81287539e-01 -3.23770009e-02
3.73143733e-01 -3.08244795e-01 -6.85316846e-02 -1.32808566e+00
3.74114275e-01 -1.08583248e+00 -6.44139126e-02 1.54672110e+00
2.65064389e-01 -5.01913190e-01 4.26905960e-01 -6.93762675e-02
-5.22377267e-02 -4.93409187e-01 -1.22663331e+00 -7.68185735e-01
-1.75158903e-01 -4.21007186e-01 4.30276662e-01 5.27117491e-01
1.71182632e-01 2.34582722e-01 -8.44897747e-01 1.38616294e-01
5.59475303e-01 -2.82559335e-01 8.03557515e-01 -9.13133979e-01
-5.43753766e-02 -4.86539096e-01 -1.93745166e-01 -1.15224791e+00
4.78806615e-01 -4.03378725e-01 3.32308233e-01 -1.36545873e+00
-1.99165180e-01 1.26983121e-01 1.41542152e-01 2.86457628e-01
-8.17040354e-02 2.99179018e-01 3.43843162e-01 1.59545630e-01
-4.42602634e-01 7.83521116e-01 1.48172462e+00 -1.75382242e-01
-3.18597376e-01 3.03232670e-01 -8.55377987e-02 7.24944651e-01
7.13270664e-01 -2.05007456e-02 -5.51497102e-01 -2.39137143e-01
-2.97610581e-01 6.82139814e-01 3.52242768e-01 -1.16237831e+00
3.46817464e-01 -5.38373962e-02 7.78698996e-02 -6.99005187e-01
7.25309789e-01 -6.08435571e-01 5.75352693e-03 3.54770422e-01
-5.96047044e-01 -3.18052351e-01 1.59224793e-01 2.09004328e-01
-4.28687781e-01 -9.68474001e-02 1.23076761e+00 3.14238250e-01
-3.75129849e-01 1.98073015e-01 -3.77891451e-01 9.55087095e-02
9.09707904e-01 -1.74218759e-01 -1.53415084e-01 -8.96546006e-01
-6.87843621e-01 -1.80181190e-01 2.22693905e-01 3.72714758e-01
9.00331199e-01 -1.27450633e+00 -9.80179727e-01 3.95966709e-01
-8.63528848e-02 -4.13887233e-01 4.79881436e-01 8.48426342e-01
-1.26658067e-01 3.64821702e-01 -2.45632038e-01 -7.30854094e-01
-1.55808508e+00 4.83137637e-01 3.83468151e-01 1.25049986e-02
-3.40902746e-01 7.00426340e-01 2.11646765e-01 2.06183031e-01
6.82746112e-01 -1.58480540e-01 -2.99291313e-01 1.74174234e-01
5.46942413e-01 5.30853927e-01 -5.75269349e-02 -7.92598665e-01
-2.77828097e-01 9.76181686e-01 5.57015896e-01 -4.61358100e-01
7.13668525e-01 -5.46708643e-01 3.40514600e-01 2.27788448e-01
1.33179235e+00 5.52802563e-01 -1.29072976e+00 1.90461561e-01
-5.29679358e-01 -3.34935874e-01 6.13229349e-02 -9.14813519e-01
-1.13450575e+00 1.20130408e+00 8.82747173e-01 1.05104130e-02
1.45196664e+00 -7.97941722e-03 7.38691509e-01 -6.63842037e-02
-6.83387101e-04 -1.05264127e+00 5.00522673e-01 1.19660966e-01
1.26531351e+00 -1.45170510e+00 -4.84881580e-01 -5.02438188e-01
-8.18257451e-01 1.13999319e+00 3.95327687e-01 4.77554530e-01
5.23836195e-01 2.99042284e-01 2.34536350e-01 5.42405367e-01
-6.14413619e-01 -4.26853806e-01 4.51557636e-01 9.02195394e-01
2.93209434e-01 -1.72551304e-01 -2.23788530e-01 3.37461978e-01
-3.38694781e-01 1.58613324e-01 1.01972990e-01 1.83965471e-02
-3.56034249e-01 -7.13549674e-01 -5.03933311e-01 -3.75129879e-01
-4.82878298e-01 -2.32486054e-01 -1.49102926e-01 5.71805596e-01
3.18255633e-01 1.43414867e+00 -5.90903759e-02 -2.95918405e-01
1.45558566e-01 1.74010187e-01 2.80703515e-01 -2.58873496e-02
-1.55926839e-01 4.56631064e-01 -7.98665360e-02 -6.39510810e-01
-4.83492851e-01 -3.45184028e-01 -8.97416949e-01 -2.91263223e-01
-4.34075981e-01 1.66370571e-02 8.10127318e-01 7.17749655e-01
2.83713609e-01 4.25516307e-01 9.77038026e-01 -9.58506048e-01
-6.29443645e-01 -1.17739141e+00 -4.16212678e-01 3.58102202e-01
6.89030945e-01 -5.73217094e-01 -8.39317381e-01 2.44826376e-01] | [14.36989974975586, 5.000560283660889] |
abe6fc8f-e8dc-4d2d-8d35-0ebc4ae3457d | enhanced-vehicle-re-identification-for-its-a | 2208.06579 | null | https://arxiv.org/abs/2208.06579v1 | https://arxiv.org/pdf/2208.06579v1.pdf | Enhanced Vehicle Re-identification for ITS: A Feature Fusion approach using Deep Learning | In recent years, the development of robust Intelligent transportation systems (ITS) is tackled across the globe to provide better traffic efficiency by reducing frequent traffic problems. As an application of ITS, vehicle re-identification has gained ample interest in the domain of computer vision and robotics. Convolutional neural network (CNN) based methods are developed to perform vehicle re-identification to address key challenges such as occlusion, illumination change, scale, etc. The advancement of transformers in computer vision has opened an opportunity to explore the re-identification process further to enhance performance. In this paper, a framework is developed to perform the re-identification of vehicles across CCTV cameras. To perform re-identification, the proposed framework fuses the vehicle representation learned using a CNN and a transformer model. The framework is tested on a dataset that contains 81 unique vehicle identities observed across 20 CCTV cameras. From the experiments, the fused vehicle re-identification framework yields an mAP of 61.73% which is significantly better when compared with the standalone CNN or transformer model. | ['Radhika M. Pai', 'Ujjwal Verma', 'Manohara Pai M. M', 'Ashutosh Holla B'] | 2022-08-13 | null | null | null | null | ['vehicle-re-identification'] | ['computer-vision'] | [ 4.25039195e-02 -4.95008409e-01 -1.49558008e-01 -4.63559508e-01
-4.62169141e-01 -6.01972580e-01 8.85628760e-01 -2.29749262e-01
-4.29995805e-01 4.47566628e-01 -1.98533908e-01 -2.94853240e-01
3.40519249e-02 -6.40363574e-01 -6.79432154e-01 -6.92297876e-01
5.07847190e-01 9.18964073e-02 1.50641054e-01 -1.13753267e-02
3.36731613e-01 9.09706235e-01 -2.09920359e+00 -5.72005361e-02
8.60309482e-01 1.20809436e+00 1.02205940e-01 3.08158815e-01
4.61199842e-02 3.38094115e-01 -2.00766504e-01 -6.08489335e-01
4.94569570e-01 2.05530822e-01 -5.63141286e-01 7.92683735e-02
7.54888117e-01 -3.80904794e-01 -4.99886215e-01 1.25276434e+00
1.47887453e-01 4.09701690e-02 5.57459831e-01 -1.57782423e+00
-5.31612754e-01 1.56574428e-01 -4.26339746e-01 2.49484912e-01
-1.10061392e-01 4.39648703e-02 4.88274574e-01 -8.38002503e-01
3.64495218e-01 1.13765883e+00 7.66978920e-01 2.22462684e-01
-8.32178235e-01 -9.56955492e-01 -2.14941874e-01 9.12364662e-01
-1.56369388e+00 -3.61438096e-01 6.41551375e-01 -4.82580394e-01
7.59666741e-01 3.85193266e-02 4.63144481e-01 8.31939816e-01
1.09888740e-01 6.06428623e-01 8.58321428e-01 -2.29726791e-01
-1.18842751e-01 3.47765267e-01 1.72286764e-01 4.65587020e-01
5.28925776e-01 2.03272447e-01 -6.35740831e-02 2.95491487e-01
3.48729551e-01 3.17687213e-01 2.51347244e-01 -3.32374632e-01
-9.25844908e-01 7.67226160e-01 4.32381809e-01 1.54815674e-01
-3.02678287e-01 5.63678928e-02 6.89366996e-01 8.12892914e-02
2.50439763e-01 1.30303815e-01 -1.60451382e-01 2.11740136e-02
-6.83099151e-01 2.71394551e-01 2.19792739e-01 9.77533221e-01
9.96077597e-01 3.36379498e-01 9.40034688e-02 9.08043325e-01
3.18074524e-01 9.52253342e-01 3.63803923e-01 -8.49661171e-01
5.00759542e-01 8.94968450e-01 -1.73137453e-03 -1.26491868e+00
-2.04518899e-01 -3.02749187e-01 -9.25293386e-01 2.08496645e-01
1.24844238e-01 9.82097760e-02 -9.95662689e-01 1.43777394e+00
3.17505330e-01 5.14711440e-01 1.80916235e-01 6.68111384e-01
8.95290792e-01 4.96209949e-01 1.24299794e-01 2.87002444e-01
1.39127147e+00 -8.98157358e-01 -6.77124023e-01 1.40435949e-01
4.85900939e-01 -7.85775304e-01 2.48516232e-01 9.81214568e-02
-3.34267914e-01 -1.00111818e+00 -1.17343879e+00 1.87493652e-01
-8.31759691e-01 3.97278428e-01 1.09830618e-01 8.23609710e-01
-9.83038664e-01 7.99444392e-02 -4.94322747e-01 -8.53889227e-01
4.58408058e-01 5.92810273e-01 -6.61662102e-01 -1.84716716e-01
-1.12498069e+00 1.06919110e+00 4.24928784e-01 3.44934404e-01
-9.70068395e-01 -5.00973523e-01 -8.14312577e-01 -6.38916716e-02
1.68394357e-01 -3.74552280e-01 9.41100776e-01 -1.08238447e+00
-1.26614296e+00 7.80057192e-01 -1.90652698e-01 -6.62202954e-01
5.18573105e-01 -5.32139987e-02 -7.80856431e-01 -1.62467450e-01
2.69597501e-01 4.57475007e-01 9.17365372e-01 -1.07132483e+00
-1.11028314e+00 -4.28834170e-01 -1.40076697e-01 -5.04114293e-02
-3.09444010e-01 1.92982778e-01 -3.25974882e-01 -2.21316889e-01
-1.35398299e-01 -1.31904233e+00 2.24699274e-01 -3.41318280e-01
-2.23167151e-01 -3.98560226e-01 1.52284241e+00 -5.87376237e-01
7.11873651e-01 -1.91877604e+00 -2.85171360e-01 2.35190362e-01
1.85770854e-01 9.39392328e-01 -1.82352945e-01 1.99771553e-01
-3.11028957e-01 -5.04311100e-02 8.47339630e-02 -3.78434688e-01
-1.44381210e-01 1.54073238e-01 -1.05644979e-01 5.90683758e-01
1.68807477e-01 8.25624347e-01 -4.83177781e-01 -2.23139390e-01
8.11883330e-01 5.72601438e-01 -3.09352092e-02 1.86770737e-01
3.56733322e-01 3.89382094e-01 -3.58450294e-01 8.18085968e-01
1.21179628e+00 1.08231694e-01 -4.61079143e-02 -6.12080693e-01
-5.11466563e-01 -4.29531157e-01 -9.91611063e-01 9.99079943e-01
-1.27978355e-01 1.14528489e+00 -1.40429109e-01 -1.22799921e+00
1.00663531e+00 2.69180000e-01 5.31822145e-01 -9.01291192e-01
5.34697950e-01 2.46638805e-01 -8.89724791e-02 -7.74744749e-01
7.62410998e-01 4.28267211e-01 2.07617387e-01 1.96776763e-02
-8.81837755e-02 3.71936291e-01 1.33176092e-02 -2.28742927e-01
7.43037581e-01 -8.78881216e-02 -9.08972695e-03 -4.27299105e-02
1.18990457e+00 8.62614065e-02 5.07915080e-01 3.77204269e-01
-4.51921523e-01 3.33689988e-01 -3.28275174e-01 -9.05866742e-01
-1.36810160e+00 -7.76633859e-01 -2.59776384e-01 4.28123772e-01
3.86197686e-01 1.20764084e-01 -7.84623504e-01 -4.20393914e-01
1.67883620e-01 3.09213430e-01 -3.86496097e-01 -2.49050453e-01
-6.99459612e-01 -5.77635407e-01 8.66436124e-01 5.32055914e-01
1.24858689e+00 -6.67066872e-01 -3.89878780e-01 2.37914417e-02
-2.36641303e-01 -1.69478285e+00 -2.01609775e-01 -4.11409676e-01
-4.78085667e-01 -1.23270440e+00 -6.06235564e-01 -9.49714363e-01
5.66041112e-01 8.11642945e-01 3.63774240e-01 8.84593353e-02
-1.83349162e-01 3.19047272e-01 -4.07026708e-01 -5.01851857e-01
-4.07546550e-01 2.36254603e-01 4.25107151e-01 5.89315951e-01
6.95281863e-01 -1.42789856e-01 -5.22889197e-01 5.93043685e-01
-6.30696654e-01 -1.65606841e-01 3.99437249e-01 6.65786326e-01
2.32782334e-01 2.34819785e-01 5.59365392e-01 -2.52898723e-01
2.90361524e-01 -5.47062874e-01 -1.14588153e+00 3.65258217e-01
-7.48607755e-01 -2.36408636e-01 6.44243658e-01 -2.83828914e-01
-1.16104376e+00 1.13858752e-01 9.83211212e-03 -5.83846509e-01
-4.34519440e-01 2.97979951e-01 -1.87745467e-01 -6.05868280e-01
6.84795007e-02 2.84822106e-01 1.89898029e-01 -4.00994956e-01
1.80205375e-01 1.10816872e+00 7.59918749e-01 -1.45286173e-01
1.06754923e+00 3.81298184e-01 8.38209018e-02 -8.88271451e-01
-1.82084471e-01 -5.14418006e-01 -6.28679216e-01 -6.03859186e-01
1.02776420e+00 -1.05968261e+00 -1.10179734e+00 1.01202333e+00
-1.38780212e+00 3.79195899e-01 3.48740399e-01 7.89483011e-01
-9.45612118e-02 4.37976390e-01 -8.64164382e-02 -7.09465146e-01
-2.05409348e-01 -1.61370897e+00 9.22837615e-01 6.20999396e-01
3.22603405e-01 -7.58719802e-01 -4.01445962e-02 6.52883410e-01
5.25933325e-01 3.54953557e-02 6.33695483e-01 -6.64599180e-01
-9.52321589e-01 -4.48967725e-01 -7.02382445e-01 5.86231351e-01
3.09474468e-01 2.54223138e-01 -1.12561440e+00 -2.85072416e-01
-3.16247314e-01 2.59769410e-01 6.18104637e-01 1.76474839e-01
8.87709081e-01 1.36036128e-02 -4.38131988e-01 7.69982994e-01
1.50460958e+00 5.73193610e-01 7.72168577e-01 6.88992143e-01
9.40440416e-01 5.54269910e-01 5.33520043e-01 2.65787333e-01
7.89601922e-01 8.17906678e-01 7.61308253e-01 -1.56191200e-01
-5.90852909e-02 -4.24342155e-02 -1.00937602e-03 5.27543664e-01
-2.93467849e-01 -1.68652222e-01 -1.04373634e+00 6.56842887e-01
-1.89409196e+00 -1.07050645e+00 -2.60637224e-01 2.10960340e+00
-1.33115903e-01 -3.22991550e-01 -3.00300252e-02 5.84151819e-02
1.21108329e+00 -2.09978610e-01 -5.21672428e-01 -4.43375528e-01
-1.05828300e-01 -2.14484140e-01 1.11302066e+00 2.59739727e-01
-1.33632016e+00 1.01736128e+00 5.96776056e+00 7.24225581e-01
-1.46477211e+00 -1.00133531e-02 4.17745739e-01 5.71779191e-01
2.01037511e-01 -1.86611578e-01 -9.11764264e-01 5.18628955e-01
9.07725930e-01 -2.34905034e-01 3.50979209e-01 8.03785324e-01
3.21586460e-01 7.12697208e-02 -7.43059754e-01 1.34457827e+00
3.21150333e-01 -1.36241269e+00 2.29853630e-01 1.88592359e-01
8.35103333e-01 2.62115121e-01 3.30466926e-01 9.85962227e-02
1.25794545e-01 -9.42512214e-01 5.91375113e-01 5.03965557e-01
7.17150152e-01 -8.39342117e-01 1.33432901e+00 2.90389638e-02
-1.58355713e+00 -3.74314606e-01 -2.43000835e-01 1.83679819e-01
8.93318206e-02 -1.34932678e-02 -1.04761088e+00 8.17657828e-01
9.80303824e-01 9.49161232e-01 -6.70553505e-01 1.20796704e+00
3.89659941e-01 2.96466351e-01 -1.27729625e-01 1.42613247e-01
1.84051931e-01 -3.10964614e-01 5.47487199e-01 8.87686491e-01
5.41111410e-01 -2.71251172e-01 3.47666368e-02 6.34008825e-01
-1.06686607e-01 -6.08299412e-02 -8.89462054e-01 6.76604360e-02
5.51251769e-01 1.39445102e+00 -4.58262742e-01 -3.06752682e-01
-5.20464897e-01 4.70638901e-01 -1.24843768e-03 4.71751958e-01
-1.09842503e+00 -2.69762188e-01 1.06629121e+00 -4.80485596e-02
5.03210306e-01 -2.49678820e-01 -4.56120223e-02 -8.69819760e-01
-1.45228922e-01 -6.63859665e-01 1.56966485e-02 -7.12437212e-01
-1.05153227e+00 7.86879182e-01 1.43665969e-01 -1.60282576e+00
-9.45602506e-02 -8.13632965e-01 -4.16245908e-01 7.48678923e-01
-1.83625972e+00 -1.67318964e+00 -6.77540481e-01 5.44137836e-01
5.66611707e-01 -8.03669035e-01 3.45908672e-01 7.73813248e-01
-8.74938250e-01 7.94523239e-01 2.88253278e-01 2.16473356e-01
5.60781419e-01 -5.40926993e-01 2.34956443e-01 1.13531876e+00
-4.07576859e-01 4.88456219e-01 5.11470497e-01 -4.99044776e-01
-1.48902512e+00 -1.48534286e+00 7.97478020e-01 -2.92318970e-01
4.54555511e-01 -2.58385539e-02 -6.70074582e-01 6.88451111e-01
3.35665852e-01 -7.13010877e-02 3.50890100e-01 -4.27514881e-01
-5.39650321e-01 -8.13124537e-01 -1.23201740e+00 2.89092034e-01
5.55143356e-01 -7.31924474e-01 -2.65269428e-01 -1.87791318e-01
3.54473442e-01 -1.42897099e-01 -7.24754751e-01 6.51028574e-01
7.88085043e-01 -7.78677046e-01 9.63105321e-01 -2.64428377e-01
-1.73871607e-01 -6.59339190e-01 -3.79229993e-01 -9.55046237e-01
-1.81506664e-01 -2.26201564e-01 4.71392930e-01 1.37900054e+00
6.43589571e-02 -9.73870397e-01 6.91620290e-01 7.13079214e-01
-1.63066164e-01 -1.16673350e-01 -1.19146776e+00 -6.77695930e-01
-2.77158886e-01 -4.69536573e-01 1.00234509e+00 7.18695879e-01
-5.52346289e-01 1.39809951e-01 -6.00870252e-01 3.55433404e-01
6.41319990e-01 -2.77833849e-01 1.02380645e+00 -1.50118816e+00
7.55500674e-01 -3.57753754e-01 -1.03729892e+00 -7.31163323e-01
4.20931846e-01 -7.07701325e-01 -6.71564266e-02 -1.09629667e+00
3.09219062e-01 -3.55022222e-01 -3.77011299e-01 2.88965195e-01
2.30489790e-01 4.88251865e-01 2.64718741e-01 1.93155587e-01
-5.31327784e-01 6.80929601e-01 9.11680818e-01 -4.74428475e-01
2.69113898e-01 -1.10051036e-02 -3.82539243e-01 5.57178557e-01
9.16334748e-01 -2.73583829e-01 -3.24766159e-01 -6.01045072e-01
-1.09164521e-01 -3.03211242e-01 7.16043472e-01 -1.24804628e+00
5.63432992e-01 1.05659083e-01 1.97069973e-01 -1.03809226e+00
2.64401525e-01 -1.37453663e+00 5.87430716e-01 2.41715118e-01
6.29457906e-02 5.20350575e-01 3.57177079e-01 4.99625057e-01
-4.29809868e-01 6.84794188e-02 7.38445163e-01 2.26839036e-01
-1.07164276e+00 4.50951755e-01 -6.63895369e-01 -7.10419118e-01
1.35153282e+00 -7.23759711e-01 -3.93241853e-01 -9.70427468e-02
-9.40828323e-02 2.47162998e-01 3.65142047e-01 8.99943292e-01
6.63775802e-01 -1.60663605e+00 -6.15673900e-01 4.09306377e-01
4.26855445e-01 -4.40448642e-01 4.54251796e-01 7.18997061e-01
-5.66643894e-01 8.55777085e-01 -5.51134348e-01 -8.57346296e-01
-1.74828660e+00 4.67649162e-01 4.75984633e-01 1.89524800e-01
-2.25520343e-01 1.81216821e-01 1.86189841e-02 -5.24940073e-01
9.17455256e-02 -3.91704477e-02 -6.25302970e-01 1.87027105e-03
5.41375160e-01 7.87832260e-01 1.82283372e-01 -1.44878387e+00
-4.84676719e-01 1.04586685e+00 -1.38377964e-01 2.73534000e-01
1.19884372e+00 -5.53288400e-01 -1.09716460e-01 -8.36727917e-02
1.57415855e+00 -5.02356410e-01 -9.70634043e-01 -3.56751651e-01
4.45380295e-03 -5.52173853e-01 1.21671567e-02 -4.61413115e-01
-1.29880595e+00 7.17275321e-01 1.17447734e+00 2.98023634e-02
9.32798743e-01 -4.03754532e-01 8.15621436e-01 5.35974741e-01
4.52533364e-01 -1.14568090e+00 -2.03390449e-01 7.09305823e-01
5.29395282e-01 -1.66270804e+00 -3.49226952e-01 -2.56099224e-01
-3.83510530e-01 1.27849340e+00 6.83952332e-01 2.43342556e-02
6.34105086e-01 -1.54441759e-01 5.63129336e-02 2.28661094e-02
-1.45373583e-01 -3.41711700e-01 3.22095811e-01 7.47515917e-01
-1.43177196e-01 -7.27189099e-03 -1.03785116e-02 6.13074303e-02
1.02784961e-01 8.60883892e-02 3.33001584e-01 5.12374222e-01
-3.80800545e-01 -1.02192199e+00 -5.68268180e-01 2.33401194e-01
-2.41095573e-01 2.95344323e-01 -1.03857011e-01 7.60976493e-01
4.10643578e-01 1.35759735e+00 1.60276815e-01 -9.11975265e-01
2.43925497e-01 -2.55646169e-01 2.08237134e-02 5.61857894e-02
-3.82262379e-01 -4.21596706e-01 5.77944256e-02 -4.02182639e-01
-8.94883454e-01 -6.42099738e-01 -8.60296369e-01 -7.04361558e-01
-4.39841419e-01 -3.86720821e-02 9.91104841e-01 1.12489557e+00
4.23391223e-01 3.83531064e-01 9.40373838e-01 -8.40027571e-01
-2.29791537e-01 -6.74817443e-01 -3.11782390e-01 3.97864372e-01
4.73883122e-01 -9.93306100e-01 -2.25057855e-01 9.29164439e-02] | [8.096098899841309, -0.9949317574501038] |
7511162b-c8b9-4413-b759-39f2be05dce0 | learning-universe-model-for-partial-matching | 2210.10374 | null | https://arxiv.org/abs/2210.10374v1 | https://arxiv.org/pdf/2210.10374v1.pdf | Learning Universe Model for Partial Matching Networks over Multiple Graphs | We consider the general setting for partial matching of two or multiple graphs, in the sense that not necessarily all the nodes in one graph can find their correspondences in another graph and vice versa. We take a universe matching perspective to this ubiquitous problem, whereby each node is either matched into an anchor in a virtual universe graph or regarded as an outlier. Such a universe matching scheme enjoys a few important merits, which have not been adopted in existing learning-based graph matching (GM) literature. First, the subtle logic for inlier matching and outlier detection can be clearly modeled, which is otherwise less convenient to handle in the pairwise matching scheme. Second, it enables end-to-end learning especially for universe level affinity metric learning for inliers matching, and loss design for gathering outliers together. Third, the resulting matching model can easily handle new arriving graphs under online matching, or even the graphs coming from different categories of the training set. To our best knowledge, this is the first deep learning network that can cope with two-graph matching, multiple-graph matching, online matching, and mixture graph matching simultaneously. Extensive experimental results show the state-of-the-art performance of our method in these settings. | ['Junchi Yan', 'Tianzhe Wang', 'Jiaxin Lu', 'Zetian Jiang'] | 2022-10-19 | null | null | null | null | ['graph-matching'] | ['graphs'] | [-8.11081287e-03 2.60693967e-01 2.81997807e-02 -1.11680463e-01
-8.50911498e-01 -4.01681036e-01 2.30286852e-01 5.39713264e-01
9.98364687e-02 2.11441427e-01 -1.99289933e-01 -3.97998318e-02
-2.94669956e-01 -7.96171963e-01 -8.82103205e-01 -6.40887618e-01
-3.20211589e-01 8.23237062e-01 2.67303914e-01 -1.10187314e-01
-1.98230341e-01 7.34408855e-01 -1.20260751e+00 -2.85219043e-01
8.82263422e-01 1.00377023e+00 -5.90936542e-02 3.96513462e-01
1.84011504e-01 4.39320326e-01 -3.05151790e-01 -8.80214870e-01
7.09688425e-01 -9.38102156e-02 -3.47308576e-01 4.18176919e-01
1.34785569e+00 -2.13563278e-01 -9.74982858e-01 1.35361552e+00
5.50169289e-01 2.60671079e-01 5.17756820e-01 -1.70152271e+00
-7.84259796e-01 6.48235083e-01 -9.12065208e-01 -6.46465123e-02
6.55256093e-01 -9.86177400e-02 1.29828882e+00 -9.40073669e-01
4.45150942e-01 1.28332746e+00 8.78653467e-01 1.24510273e-01
-1.10854769e+00 -6.64353609e-01 3.93481106e-01 5.46043776e-02
-1.51482213e+00 -2.66557276e-01 1.07407975e+00 -3.82767409e-01
2.92553753e-01 2.19277710e-01 7.10887790e-01 8.59354198e-01
2.21506157e-03 7.08169937e-01 3.78191531e-01 -1.58127680e-01
-3.57099040e-03 -4.00466323e-01 7.52400458e-02 8.74447167e-01
6.47067606e-01 2.55786180e-01 -4.17918026e-01 -3.97783905e-01
6.89846575e-01 3.19579393e-01 -4.91415679e-01 -1.04354310e+00
-1.23109269e+00 5.80716372e-01 8.71832192e-01 1.74108684e-01
-8.04947466e-02 1.53128818e-01 2.90746182e-01 9.16433334e-01
5.00559211e-01 6.75483197e-02 -1.49030805e-01 5.42650580e-01
-7.71043301e-01 1.79129899e-01 8.74745429e-01 1.51806688e+00
1.01404989e+00 -5.50595857e-02 1.21063471e-01 5.02845109e-01
4.05713737e-01 2.46026054e-01 8.35839212e-02 -4.86069262e-01
7.40686715e-01 6.95412338e-01 -4.13993001e-02 -1.60930502e+00
-3.99545878e-01 -4.51820880e-01 -1.24510264e+00 -1.90556105e-02
6.60951972e-01 2.50189543e-01 -8.62285972e-01 1.70132935e+00
5.20606101e-01 8.43505204e-01 -3.81510407e-01 8.31928968e-01
1.01368177e+00 -8.08323920e-02 -5.68047404e-01 -1.68463588e-01
8.93727779e-01 -9.29930747e-01 -4.15946305e-01 -2.79500365e-01
5.80351472e-01 -7.20006049e-01 6.82114184e-01 1.32500574e-01
-9.09583092e-01 -5.88035822e-01 -1.14261115e+00 -3.83143201e-02
-3.57112736e-01 -5.38022459e-01 6.49831593e-01 4.48297620e-01
-1.22268295e+00 8.55163217e-01 -5.85879087e-01 -5.78308284e-01
2.98189491e-01 4.01509076e-01 -6.05073631e-01 -4.62319493e-01
-1.00421274e+00 4.46929902e-01 2.52783328e-01 1.60006493e-01
-4.27785665e-01 -6.63895488e-01 -1.22123182e+00 5.08701801e-02
5.64851582e-01 -9.45241332e-01 7.31800973e-01 -8.08218360e-01
-7.14072049e-01 1.16587865e+00 -1.47789106e-04 -1.41384229e-01
1.02225292e+00 1.73610121e-01 -5.43155491e-01 -2.31312379e-01
2.48181775e-01 7.81018659e-02 9.99339640e-01 -1.14548469e+00
-3.27079296e-01 -8.59741092e-01 -2.21052486e-03 6.58246204e-02
-1.38439924e-01 -2.63842553e-01 -7.21396267e-01 -7.86557794e-01
7.99181759e-01 -9.16873991e-01 -1.72448590e-01 3.97040665e-01
-5.33213198e-01 -2.00005651e-01 7.27392972e-01 -4.46083099e-01
1.03622842e+00 -2.28721690e+00 2.95400202e-01 7.63579607e-01
7.25061893e-01 -1.21133134e-01 -1.15267418e-01 6.73947930e-01
-3.57584387e-01 -1.41556352e-01 -4.07541543e-02 -6.59258246e-01
2.22462192e-01 8.38528052e-02 -8.66558179e-02 1.12201548e+00
-2.80629039e-01 9.43169355e-01 -1.24737155e+00 -4.36795473e-01
1.79956317e-01 1.15543202e-01 -2.93692678e-01 2.64364064e-01
2.49736682e-01 4.79728431e-01 -1.67382598e-01 1.10071731e+00
1.12632143e+00 -3.90585870e-01 1.38991386e-01 -3.63756031e-01
5.29757142e-01 -3.29514474e-01 -1.83785856e+00 1.92688596e+00
-1.90051347e-02 2.85560727e-01 1.55227304e-01 -1.01392865e+00
9.72642660e-01 2.19474450e-01 6.23760283e-01 -3.67915720e-01
-3.09385806e-02 5.51988900e-01 -5.36271883e-03 -1.59505144e-01
3.87124091e-01 1.66916803e-01 -1.21904962e-01 1.48594484e-01
2.47041389e-01 1.31893396e-01 1.91571444e-01 3.15956324e-01
1.15134513e+00 -2.29228437e-01 3.92215371e-01 -1.59408733e-01
3.88782710e-01 -5.80805779e-01 7.88894773e-01 9.95281637e-01
-3.69804800e-01 9.25298631e-01 2.06533954e-01 -4.89215732e-01
-9.00571465e-01 -1.27357101e+00 5.44404089e-02 7.15696633e-01
6.16976380e-01 -2.48134673e-01 -1.65904984e-01 -6.85146511e-01
5.69604456e-01 -2.72314996e-01 -3.87456954e-01 -3.00307751e-01
-5.77959120e-01 -2.31136724e-01 2.08656654e-01 3.01677138e-01
1.30341440e-01 -5.33329010e-01 3.68905991e-01 2.46448547e-01
-9.29355547e-02 -1.11983955e+00 -1.03289509e+00 -3.36826712e-01
-8.85367870e-01 -1.59371448e+00 -6.37840867e-01 -1.04466820e+00
9.09578323e-01 6.98069215e-01 1.28909576e+00 6.49451733e-01
-1.33337960e-01 4.76457119e-01 -2.28694737e-01 -9.87229571e-02
-2.90871441e-01 2.62687989e-02 2.87654787e-01 4.16020513e-01
2.32828423e-01 -1.04604387e+00 -5.54656208e-01 4.86823380e-01
-7.73629248e-01 -3.60179126e-01 3.31238598e-01 7.10916638e-01
6.22198284e-01 -9.67473723e-03 3.33585441e-01 -8.74572277e-01
3.12943578e-01 -6.18287027e-01 -6.40041888e-01 3.04094225e-01
-3.98747534e-01 -3.75390083e-01 7.01051831e-01 -4.88913804e-01
-1.84350282e-01 -5.58421314e-02 1.53236672e-01 -9.26151097e-01
-6.65356964e-03 3.64741147e-01 -5.31296194e-01 -7.26189017e-01
5.07443428e-01 -1.12743199e-01 -2.16162652e-02 -4.85651881e-01
4.10132766e-01 2.34339744e-01 7.43852139e-01 -5.85830033e-01
1.45303714e+00 6.24517798e-01 3.57873976e-01 -4.97905672e-01
-5.79242468e-01 -9.67921734e-01 -9.30736899e-01 -2.61840194e-01
3.13013643e-01 -1.06097496e+00 -5.12032032e-01 6.87308729e-01
-1.00211334e+00 2.09770799e-01 -1.27315253e-01 2.03933135e-01
-5.14959335e-01 1.02603638e+00 -5.27749360e-01 -5.35286784e-01
-1.68411076e-01 -1.06930578e+00 1.23548329e+00 1.43080577e-01
5.16819693e-02 -1.26086473e+00 -1.17736414e-01 1.16174892e-01
6.46559075e-02 7.39445806e-01 5.00537455e-01 -9.25682783e-01
-9.06040132e-01 -6.89960063e-01 -2.56837666e-01 -1.88114241e-01
2.96325177e-01 -9.30576846e-02 -7.25214839e-01 -9.96853828e-01
-2.98319250e-01 -1.79789186e-01 7.24726558e-01 6.57019019e-03
8.14833164e-01 -1.51259214e-01 -6.45723343e-01 9.55515921e-01
1.50345016e+00 -2.99573481e-01 3.23815674e-01 1.23063147e-01
1.23126793e+00 3.81372839e-01 5.67605913e-01 2.81440198e-01
4.71511304e-01 7.92174339e-01 8.06676745e-01 -2.17432529e-01
1.62138569e-03 -4.82508957e-01 3.19346339e-02 9.32559431e-01
3.65612656e-01 -2.67943978e-01 -6.80305898e-01 6.23139441e-01
-2.29851937e+00 -7.04283416e-01 -2.43417755e-01 2.53589535e+00
4.65348549e-02 -7.62568563e-02 1.89613685e-01 -5.55462874e-02
1.10663724e+00 3.49220634e-01 -6.79824531e-01 3.65030974e-01
-9.99493524e-02 -2.47688383e-01 4.04326022e-01 4.71704483e-01
-1.15224779e+00 6.06661081e-01 4.97787476e+00 6.26268566e-01
-5.37251115e-01 -7.02105686e-02 3.54416013e-01 2.54758447e-01
-3.77385139e-01 2.46325016e-01 -3.60241801e-01 4.86643285e-01
1.29863396e-01 -2.82319814e-01 5.26910901e-01 7.30833471e-01
-3.01320136e-01 3.39729786e-01 -1.59037805e+00 1.27665353e+00
1.80617154e-01 -1.00656629e+00 8.23540092e-02 1.37151092e-01
7.91330516e-01 8.24618489e-02 -1.27181262e-01 1.84971943e-01
2.21568927e-01 -6.23594165e-01 4.48605865e-01 6.14075661e-01
6.20017707e-01 -6.82120144e-01 6.44073963e-01 2.87040561e-01
-1.70669138e+00 1.53563663e-01 -6.56631529e-01 2.82459438e-01
9.07917917e-02 7.67390251e-01 -4.01557535e-01 1.35480678e+00
6.76284373e-01 9.20218766e-01 -5.39584160e-01 1.51300454e+00
2.88221478e-01 -1.22926980e-01 -5.21581650e-01 4.37298745e-01
4.89435671e-03 -6.48974717e-01 8.95938039e-01 6.99090004e-01
5.67470968e-01 -3.59941006e-01 7.58706868e-01 7.31480479e-01
-4.11829263e-01 1.36377588e-01 -1.10145354e+00 3.91644567e-01
6.79063559e-01 1.16525590e+00 -5.41992188e-01 -1.58094242e-01
-7.47171283e-01 1.12712193e+00 7.82028377e-01 4.09221560e-01
-7.80469716e-01 -3.05826247e-01 7.70720184e-01 1.22129112e-01
1.38636082e-01 -4.29079868e-03 2.07489207e-01 -1.72395372e+00
5.51214039e-01 -9.07041907e-01 8.31458807e-01 -5.16862392e-01
-1.86745429e+00 3.71580333e-01 -2.15942115e-01 -1.58535862e+00
7.89520368e-02 -4.84913230e-01 -8.30657899e-01 5.61780035e-01
-1.34740877e+00 -1.37944102e+00 -6.98607743e-01 7.96649575e-01
-6.55896365e-05 -1.60789028e-01 4.33281720e-01 7.75503099e-01
-4.65728521e-01 9.79908526e-01 1.24933802e-01 3.70314330e-01
8.85073960e-01 -1.33930194e+00 6.28233254e-01 1.32833350e+00
4.27250624e-01 4.93810892e-01 6.05581045e-01 -6.33214235e-01
-1.63103282e+00 -1.24786448e+00 7.98489153e-01 -3.88312757e-01
9.57943499e-01 -5.06881893e-01 -1.10045445e+00 9.68463063e-01
-3.74856740e-01 4.96179491e-01 3.98584068e-01 1.74123600e-01
-5.31987011e-01 -3.20877701e-01 -1.21301556e+00 7.17665076e-01
1.56371534e+00 -5.49928665e-01 -2.19842300e-01 7.14230597e-01
8.34201515e-01 -9.15745854e-01 -1.11797416e+00 5.00532806e-01
2.97134012e-01 -1.16081655e+00 1.15875542e+00 -5.57934940e-01
-3.30466956e-01 -4.48400557e-01 -3.28304023e-01 -1.35021877e+00
-4.41608816e-01 -1.10016859e+00 -5.61531901e-01 1.36281538e+00
5.23790605e-02 -8.32998753e-01 8.51098716e-01 5.40387332e-01
-2.16649994e-01 -5.96396983e-01 -1.17324936e+00 -1.05691755e+00
-1.67976290e-01 -3.21919054e-01 1.09791315e+00 1.22904587e+00
-1.79452375e-01 -4.87063155e-02 -4.46657002e-01 7.47681677e-01
1.12487054e+00 3.89068723e-01 1.14184046e+00 -1.55163443e+00
-4.51181293e-01 -4.78888422e-01 -9.92027640e-01 -1.15938663e+00
2.01650739e-01 -1.21536624e+00 -2.43052319e-01 -1.37791741e+00
5.31007163e-02 -4.72633600e-01 -2.73823559e-01 -1.54905319e-02
-3.87360305e-01 1.08861089e-01 2.46704191e-01 2.88919330e-01
-6.07378542e-01 5.15248716e-01 1.13013172e+00 -3.33745390e-01
1.16184406e-01 3.01781744e-01 -6.57782197e-01 5.11144221e-01
3.35822463e-01 -4.25906122e-01 -2.81965137e-01 -2.89338797e-01
2.96185076e-01 3.13971430e-01 4.65756387e-01 -1.06763446e+00
5.64433634e-01 1.44948006e-01 3.40274163e-02 -4.52221841e-01
-1.11512961e-02 -1.31269610e+00 4.07486230e-01 1.95339844e-01
2.53894597e-01 3.26032519e-01 -1.59858733e-01 1.19018078e+00
-2.54290223e-01 3.11103091e-02 6.95361257e-01 -1.66721269e-01
-7.47602940e-01 1.31151140e+00 4.94665563e-01 2.66163349e-01
9.67661917e-01 -5.78966796e-01 -1.37095660e-01 -5.73875546e-01
-6.85448289e-01 5.44139743e-01 9.43060338e-01 5.66903710e-01
6.85123861e-01 -1.84712279e+00 -7.96131909e-01 3.05713952e-01
6.32395387e-01 3.82839262e-01 2.53932983e-01 1.10681534e+00
-3.29448491e-01 -4.16382551e-01 2.87282944e-01 -8.24275076e-01
-1.11227918e+00 1.00572097e+00 5.45557857e-01 -3.03115457e-01
-8.91897321e-01 8.30325782e-01 2.96502799e-01 -8.31229627e-01
5.96952975e-01 -1.47481516e-01 3.99085224e-01 -8.20223838e-02
2.39136219e-01 4.30225134e-01 2.42827848e-01 -7.54549026e-01
-3.64907295e-01 7.24645734e-01 -1.03606783e-01 6.14794374e-01
8.47467661e-01 -1.78852826e-01 -3.34130049e-01 4.67596710e-01
1.42473054e+00 6.59558475e-02 -9.65901315e-01 -7.48315990e-01
-5.51161589e-03 -8.61657560e-01 -5.34475267e-01 7.02617317e-02
-1.41356695e+00 4.51949567e-01 2.52530217e-01 3.90956134e-01
8.55135083e-01 9.53004807e-02 8.76433372e-01 2.72058785e-01
7.52641320e-01 -6.95061862e-01 2.16138344e-02 1.91506192e-01
8.30437422e-01 -1.57631922e+00 1.01394832e-01 -5.17626464e-01
2.74124574e-02 1.08226311e+00 8.09343398e-01 -5.21696568e-01
8.32786441e-01 -1.02088697e-01 -1.03939906e-01 -3.85134518e-01
-1.95618734e-01 -1.83146358e-01 5.99975169e-01 6.83863521e-01
1.63255528e-01 -4.73973574e-03 2.79534221e-01 8.86172615e-03
8.11489392e-03 -5.94389319e-01 3.79598886e-01 4.96172726e-01
-2.08290562e-01 -9.71375644e-01 -3.29143375e-01 5.12430251e-01
-1.49526745e-01 6.41375706e-02 -5.92909932e-01 1.10584044e+00
-8.65382925e-02 8.64528179e-01 3.16973552e-02 -3.05657774e-01
6.50655091e-01 -3.88472319e-01 5.06514549e-01 -4.20853108e-01
-3.48396301e-01 6.41184971e-02 -2.17246816e-01 -7.83752203e-01
-3.07917833e-01 -6.24990702e-01 -8.59436274e-01 -7.01690316e-01
-6.12387776e-01 -8.40085745e-02 -1.34409994e-01 7.43693054e-01
4.03248578e-01 2.92178184e-01 7.57137477e-01 -1.08508432e+00
-5.31046391e-01 -8.08672309e-01 -9.16643918e-01 9.62017775e-01
6.95799291e-01 -7.03839779e-01 -7.73211122e-01 -7.34628499e-01] | [7.190425872802734, 6.382202625274658] |
45e960c6-c5f7-47d3-a571-8dd47d368c23 | a-generalized-parametric-3d-shape | 1803.01780 | null | http://arxiv.org/abs/1803.01780v1 | http://arxiv.org/pdf/1803.01780v1.pdf | A generalized parametric 3D shape representation for articulated pose estimation | We present a novel parametric 3D shape representation, Generalized sum of
Gaussians (G-SoG), which is particularly suitable for pose estimation of
articulated objects. Compared with the original sum-of-Gaussians (SoG), G-SoG
can handle both isotropic and anisotropic Gaussians, leading to a more flexible
and adaptable shape representation yet with much fewer anisotropic Gaussians
involved. An articulated shape template can be developed by embedding G-SoG in
a tree-structured skeleton model to represent an articulated object. We further
derive a differentiable similarity function between G-SoG (the template) and
SoG (observed data) that can be optimized analytically for efficient pose
estimation. The experimental results on a standard human pose estimation
dataset show the effectiveness and advantages of G-SoG over the original SoG as
well as the promise compared with the recent algorithms that use more
complicated shape models. | ['Meng Ding', 'Guoliang Fan'] | 2018-03-05 | null | null | null | null | ['3d-shape-representation'] | ['computer-vision'] | [-1.95992634e-01 2.94687837e-01 3.30330968e-01 -8.70858431e-02
-5.58764040e-01 -5.11097074e-01 2.77754843e-01 -3.02949339e-01
-2.34789729e-01 4.58655059e-01 9.37116593e-02 1.49807796e-01
-4.61241990e-01 -5.29889345e-01 -4.15854067e-01 -8.37934196e-01
-8.78184885e-02 1.16596460e+00 4.89418030e-01 1.51669487e-01
-6.92314133e-02 1.02655149e+00 -1.19263554e+00 -5.72281063e-01
9.33933794e-01 9.21438694e-01 2.63636678e-01 4.12866652e-01
1.84919909e-02 -4.53611314e-01 -4.66328949e-01 -5.97577393e-01
1.85406893e-01 1.55280828e-01 -4.86748844e-01 4.73357379e-01
3.06560755e-01 7.30731189e-02 -1.97234020e-01 7.04329133e-01
7.44402230e-01 3.25239271e-01 1.20564735e+00 -1.04681504e+00
-3.62544805e-01 -8.74869302e-02 -5.01936913e-01 -4.49822873e-01
2.41773814e-01 -2.16082305e-01 6.51992798e-01 -1.09119284e+00
6.22557640e-01 1.69782889e+00 1.03848410e+00 4.66385335e-01
-1.19702840e+00 -1.67587891e-01 1.01934336e-01 -3.23488384e-01
-1.56235111e+00 5.72658256e-02 8.97279918e-01 -3.96268398e-01
4.53766555e-01 2.39084885e-01 8.46208513e-01 8.48019660e-01
2.61738002e-01 9.91375446e-01 7.89776146e-01 -1.32635191e-01
1.23709060e-01 -4.44716483e-01 3.91976014e-02 1.08609855e+00
5.82118213e-01 -2.68305749e-01 -2.26623669e-01 -6.60390258e-01
1.32637525e+00 -2.64680386e-01 -2.48922899e-01 -1.19254410e+00
-1.27578318e+00 6.18941545e-01 1.39975265e-01 -6.86015338e-02
-5.89846730e-01 2.57483125e-01 1.30143985e-01 -3.70473593e-01
3.71728092e-01 2.31432930e-01 -2.88962215e-01 -1.89777941e-01
-7.77651012e-01 6.32972062e-01 8.57505500e-01 1.07364488e+00
3.86965364e-01 4.37930912e-01 -3.14459294e-01 9.15757179e-01
7.86625743e-01 9.41028714e-01 1.39553025e-01 -1.11511219e+00
2.98396647e-01 5.19682467e-01 2.73665786e-01 -9.61371183e-01
-6.65986478e-01 -7.22059548e-01 -8.01021397e-01 2.07814723e-01
6.75872743e-01 2.39782855e-02 -8.01867843e-01 1.50359964e+00
7.60618985e-01 -3.17956172e-02 -1.93562642e-01 7.75424480e-01
7.91084409e-01 4.13574964e-01 -3.77277136e-01 1.45859674e-01
1.47668803e+00 -7.97832906e-01 -6.01454198e-01 -3.59420627e-02
1.36454672e-01 -6.40489221e-01 9.29197729e-01 5.29463649e-01
-1.13483119e+00 -4.38740402e-01 -8.90552461e-01 1.83028609e-01
-8.84625036e-03 2.69357830e-01 7.05888689e-01 8.47871423e-01
-8.21069419e-01 5.51233172e-01 -1.03833878e+00 -2.11697623e-01
2.98158646e-01 4.59661424e-01 -5.29732287e-01 3.52803200e-01
-6.04661584e-01 8.41353953e-01 3.12577963e-01 2.18622223e-01
-6.51774406e-01 -3.18576604e-01 -9.61134911e-01 -2.32365489e-01
4.20244187e-01 -9.91315484e-01 1.07753396e+00 6.27685338e-02
-1.97983778e+00 7.65400052e-01 -2.42126688e-01 1.07501984e-01
7.60838985e-01 -3.37172359e-01 6.17111437e-02 2.51891971e-01
-1.83483750e-01 3.10518056e-01 1.17289126e+00 -1.37527609e+00
1.03715993e-01 -6.88321352e-01 -3.33855242e-01 2.47234672e-01
-3.34756039e-02 -9.00767595e-02 -7.40132511e-01 -1.07705224e+00
8.18786561e-01 -1.29196930e+00 -4.07010764e-01 3.17942113e-01
-4.53031957e-01 -3.48734230e-01 1.02818501e+00 -6.57399595e-01
1.11069369e+00 -1.76750171e+00 6.34107530e-01 5.24679303e-01
3.95190157e-02 2.78328001e-01 -1.00975484e-01 3.46427232e-01
3.14212918e-01 -2.07300350e-01 -3.82262141e-01 -5.55455208e-01
1.47792250e-01 5.94464540e-01 1.80418894e-01 6.73381984e-01
2.81329229e-02 1.01311243e+00 -7.45867848e-01 -5.35602510e-01
-6.80008456e-02 7.52133071e-01 -4.60731626e-01 1.12096287e-01
7.40402415e-02 4.14270729e-01 -8.50170195e-01 8.37264419e-01
6.68942869e-01 8.39492455e-02 -2.33695701e-01 -4.20795858e-01
1.28766760e-01 -2.67148316e-01 -1.59776437e+00 1.40945101e+00
-5.04868627e-02 -3.86319458e-02 4.99549747e-01 -7.10796237e-01
1.40615654e+00 4.01668102e-01 6.01008296e-01 3.58374059e-01
1.64110243e-01 3.57360899e-01 -3.02225500e-01 -2.40104154e-01
2.80374140e-01 -7.17354268e-02 -1.50002509e-01 2.53754765e-01
1.20418802e-01 -6.83153152e-01 -8.04271773e-02 -2.11514756e-01
4.39027637e-01 6.48903847e-01 3.75930101e-01 -4.49441105e-01
6.61270142e-01 -5.71543992e-01 6.05470240e-01 4.04343873e-01
-7.20207840e-02 8.52551639e-01 -8.93608704e-02 -4.07665044e-01
-8.82584572e-01 -1.45711744e+00 -1.49604946e-01 5.51407278e-01
1.68783501e-01 -2.75825292e-01 -8.50534678e-01 -4.37529266e-01
3.75157207e-01 3.30616534e-01 -2.47117400e-01 3.36302929e-02
-6.50143862e-01 -5.72403550e-01 6.48579478e-01 7.88377583e-01
4.72203791e-01 -7.08159208e-01 -6.68349683e-01 2.89839238e-01
-1.21465407e-01 -1.09998119e+00 -8.04532588e-01 -7.23482743e-02
-1.21644056e+00 -9.16279733e-01 -1.07899690e+00 -6.02458835e-01
8.11183453e-01 2.57701762e-02 6.95810080e-01 -3.53644848e-01
-1.45152330e-01 8.73645484e-01 -1.03675209e-01 -5.46733499e-01
-1.65582836e-01 -2.78393239e-01 2.91118950e-01 2.25255340e-01
-3.11399877e-01 -6.37222230e-01 -4.10982937e-01 7.19763994e-01
-7.55413651e-01 -1.74599171e-01 6.33797586e-01 7.63579190e-01
8.63587797e-01 -1.01406589e-01 2.72968113e-01 -4.92275208e-01
6.22147501e-01 6.95670024e-02 -4.66278821e-01 9.02156308e-02
-3.03525865e-01 2.40178674e-01 1.66043594e-01 -6.39117718e-01
-1.14789999e+00 2.60811508e-01 -5.70487417e-02 -5.77058733e-01
1.30396839e-02 4.27644581e-01 -5.72921515e-01 -4.29947108e-01
8.46305713e-02 3.44532311e-01 4.94425476e-01 -9.27778661e-01
4.51962113e-01 4.99992669e-01 8.86401534e-01 -1.20383847e+00
9.71406162e-01 6.71738386e-01 5.55224419e-01 -1.23634732e+00
-4.16724235e-01 -5.75984657e-01 -1.03349733e+00 -2.12794065e-01
7.51534760e-01 -4.97507066e-01 -7.61765718e-01 7.14026034e-01
-1.22654450e+00 5.84026277e-02 -2.88039565e-01 6.31287098e-01
-1.14450264e+00 9.93440092e-01 -4.73044604e-01 -1.17536330e+00
-4.75629270e-01 -1.12940395e+00 1.53728795e+00 3.60045992e-02
-4.02426004e-01 -1.12526000e+00 -4.90783863e-02 4.41302240e-01
1.77270100e-01 6.08588159e-01 7.37781823e-01 -4.59687680e-01
-2.72619665e-01 -5.37267208e-01 2.24364191e-01 2.04465404e-01
1.80943105e-02 1.36686221e-01 -4.85792398e-01 -4.93089944e-01
1.60025388e-01 8.06595385e-02 3.50380063e-01 5.45219660e-01
9.96571124e-01 -1.97284371e-01 -4.74619567e-01 7.48286486e-01
9.64809895e-01 1.93132673e-05 5.65368056e-01 2.62428194e-01
6.30296826e-01 3.63465309e-01 7.42704034e-01 5.82432628e-01
2.61135250e-01 1.07598567e+00 5.51414430e-01 9.26225930e-02
2.09111664e-02 -2.42768809e-01 3.14345509e-01 8.40829372e-01
-9.04171407e-01 7.14704394e-02 -7.18658268e-01 3.60609084e-01
-1.86821949e+00 -6.11113250e-01 -3.97071868e-01 2.36678648e+00
4.60773438e-01 -1.23946317e-01 4.32995260e-01 1.63343698e-01
6.82022154e-01 -3.76572981e-02 -3.31633031e-01 -1.63796529e-01
-1.32938817e-01 2.83826649e-01 3.00662220e-01 3.98032308e-01
-1.17487240e+00 6.43070102e-01 7.20473576e+00 1.07895386e+00
-5.33405304e-01 -1.43498749e-01 -2.28593364e-01 3.87988180e-01
-1.65625110e-01 -1.92488879e-01 -1.07467532e+00 2.44296923e-01
3.48106623e-01 -2.42295191e-01 -6.14386424e-02 9.97239351e-01
1.92031026e-01 2.23824173e-01 -7.02641606e-01 1.14013720e+00
-5.49126007e-02 -6.65691078e-01 3.16273034e-01 2.96070457e-01
5.61036885e-01 -5.29133976e-01 -1.29854642e-02 -5.54830208e-02
-7.33937919e-02 -7.32290924e-01 9.96116757e-01 7.54485548e-01
4.69372809e-01 -6.54987872e-01 6.01165235e-01 4.81665611e-01
-1.61627924e+00 3.03948134e-01 -6.18426561e-01 3.57235879e-01
5.59665620e-01 4.43089426e-01 -6.43884778e-01 9.33904827e-01
3.44864547e-01 2.32828110e-01 -3.29826713e-01 1.45677519e+00
-1.19191304e-01 3.24246407e-01 -7.53938675e-01 -8.22006837e-02
2.09052846e-01 -7.44600832e-01 1.29014647e+00 1.16725862e+00
6.02721810e-01 -6.31546006e-02 3.05148274e-01 8.63026798e-01
4.81094897e-01 2.44321987e-01 -3.88108432e-01 -2.91410070e-02
3.36824805e-01 1.27382863e+00 -8.02436173e-01 -1.56355470e-01
-1.11689739e-01 1.00943780e+00 6.69558644e-02 3.97450507e-01
-7.49201775e-01 -1.90531492e-01 6.02730334e-01 3.09781879e-01
5.43114305e-01 -7.88632393e-01 -2.01472603e-02 -8.64201665e-01
-2.67135142e-03 -6.92470193e-01 3.22678179e-01 -7.37710297e-01
-1.20380795e+00 5.43135285e-01 4.91226733e-01 -1.34845376e+00
-1.71835303e-01 -1.07313991e+00 -4.04511660e-01 6.50093138e-01
-7.68057227e-01 -1.52289867e+00 -2.08505690e-01 5.99137068e-01
3.35090339e-01 -1.45810395e-01 1.00634813e+00 -2.52424330e-01
-7.96582624e-02 4.28146064e-01 9.57436934e-02 -9.27862003e-02
2.13086635e-01 -1.37168097e+00 3.27668965e-01 3.72214973e-01
1.13179386e-02 8.01392138e-01 7.21738577e-01 -8.39778364e-01
-1.57459056e+00 -8.06502700e-01 2.98498482e-01 -5.68914533e-01
4.16322380e-01 -1.22869499e-01 -9.14817929e-01 5.38977861e-01
-4.64908957e-01 -7.18265548e-02 6.20792866e-01 -1.81805655e-01
-2.29929715e-01 3.06605905e-01 -1.23719358e+00 5.60626686e-01
1.27610600e+00 -8.47184509e-02 -9.05461967e-01 2.08445713e-02
3.08294296e-01 -5.55112481e-01 -1.32491243e+00 7.45321751e-01
9.34399426e-01 -6.22017860e-01 1.37218261e+00 -4.70785528e-01
-4.92388338e-01 -5.42167068e-01 -1.13394938e-01 -1.33737135e+00
-6.02974653e-01 -9.26190197e-01 -5.73073626e-01 8.25357318e-01
-9.50141028e-02 -6.51134968e-01 1.00626361e+00 4.79259819e-01
-3.93304825e-01 -1.08241296e+00 -1.16229081e+00 -1.45032334e+00
-1.88050848e-02 -3.03309917e-01 4.60826010e-01 3.45763266e-01
-4.45016950e-01 2.87067313e-02 -3.36239904e-01 2.59662449e-01
1.19616210e+00 9.34908316e-02 9.93120968e-01 -1.85881126e+00
-3.17281455e-01 -4.09803987e-01 -7.70243108e-01 -1.43118393e+00
7.35544264e-02 -7.53918350e-01 -2.83352006e-02 -1.67943358e+00
-6.51455000e-02 -2.31691629e-01 3.18698436e-01 2.84058481e-01
-3.02034467e-01 1.99131817e-01 2.21119188e-02 5.80852404e-02
-2.57458895e-01 1.01803768e+00 1.77307236e+00 8.24351683e-02
-1.35799780e-01 5.39168477e-01 -1.90800190e-01 1.15296733e+00
3.43631625e-01 -1.08892374e-01 -2.63279229e-01 2.80189048e-02
-2.36906216e-01 -1.01061650e-01 2.67166704e-01 -9.22072768e-01
4.14128937e-02 1.27091138e-02 2.41378590e-01 -1.06948805e+00
8.48862052e-01 -1.00798535e+00 7.02823222e-01 5.10604978e-01
4.02135670e-01 1.54852578e-02 5.13926856e-02 7.64042974e-01
-6.26263022e-02 -4.42086667e-01 7.47723281e-01 2.94917878e-02
-3.05676788e-01 5.36213458e-01 -2.80800551e-01 -1.89751655e-01
9.38070655e-01 -8.69959533e-01 2.30482876e-01 -4.51573700e-01
-1.19716394e+00 -2.90227514e-02 5.25713682e-01 1.69229656e-01
8.63206804e-01 -1.65392363e+00 -6.43397987e-01 1.64950952e-01
-3.06790392e-03 3.33616138e-01 -8.92377049e-02 9.67273295e-01
-5.83117783e-01 4.39047605e-01 -1.20557053e-02 -7.78243244e-01
-1.34883034e+00 2.76441038e-01 2.57841408e-01 -1.64522856e-01
-7.46745408e-01 6.93816423e-01 2.44354486e-01 -7.94691682e-01
1.32481515e-01 -4.53124791e-01 -1.15574328e-02 -1.96816236e-01
1.75780579e-01 8.15367937e-01 -1.49061695e-01 -1.14245522e+00
-3.10602069e-01 1.23127282e+00 5.96383214e-01 -6.97939331e-03
1.48698735e+00 -2.90810093e-02 -1.61374941e-01 2.97015220e-01
7.59853184e-01 3.28083217e-01 -1.49400198e+00 -1.54031500e-01
-3.75017375e-02 -4.48618770e-01 -3.56237352e-01 -1.03537455e-01
-8.94511700e-01 6.26947224e-01 2.49465391e-01 -4.31578979e-02
5.86297691e-01 1.62225336e-01 5.23658037e-01 3.83770764e-01
8.14738870e-01 -8.66350949e-01 2.82469422e-01 5.20123184e-01
1.45093858e+00 -3.80692095e-01 7.58789256e-02 -1.01662540e+00
-5.23028493e-01 1.38994110e+00 1.84200868e-01 -2.04694256e-01
7.72300124e-01 2.94606924e-01 -9.46474001e-02 -2.91870773e-01
-1.54743344e-02 -2.69715965e-01 9.49697971e-01 1.11128891e+00
1.95458904e-01 6.79421872e-02 -4.25131053e-01 8.78610075e-01
-2.11237624e-01 -2.41812289e-01 8.35639387e-02 8.57427359e-01
-6.18789494e-01 -1.19338667e+00 -8.82879436e-01 2.30309755e-01
-2.28436977e-01 6.25056088e-01 -3.65985841e-01 8.32331717e-01
-2.53937304e-01 4.67723429e-01 -2.99911052e-01 -5.89818973e-03
6.35853648e-01 2.74131507e-01 9.43942845e-01 -5.68184793e-01
-5.17752394e-02 4.97234315e-01 6.49979711e-02 -7.25647330e-01
-4.99246359e-01 -6.71171069e-01 -1.33025074e+00 1.75224289e-01
-5.91697633e-01 2.36613959e-01 6.10080481e-01 7.05075085e-01
2.81396687e-01 1.62941247e-01 7.02172220e-02 -1.46265173e+00
-9.10717964e-01 -9.11949098e-01 -9.94354427e-01 4.52245831e-01
-1.95092857e-01 -1.37753749e+00 -2.24729091e-01 -1.05924331e-01] | [7.0356550216674805, -1.2235417366027832] |
cfb0b0ac-b356-45c5-89df-0bd7fa70fecb | you-only-need-one-model-for-open-domain | 2112.07381 | null | https://arxiv.org/abs/2112.07381v2 | https://arxiv.org/pdf/2112.07381v2.pdf | You Only Need One Model for Open-domain Question Answering | Recent approaches to Open-domain Question Answering refer to an external knowledge base using a retriever model, optionally rerank passages with a separate reranker model and generate an answer using another reader model. Despite performing related tasks, the models have separate parameters and are weakly-coupled during training. We propose casting the retriever and the reranker as internal passage-wise attention mechanisms applied sequentially within the transformer architecture and feeding computed representations to the reader, with the hidden representations progressively refined at each stage. This allows us to use a single question answering model trained end-to-end, which is a more efficient use of model capacity and also leads to better gradient flow. We present a pre-training method to effectively train this architecture and evaluate our model on the Natural Questions and TriviaQA open datasets. For a fixed parameter budget, our model outperforms the previous state-of-the-art model by 1.0 and 0.7 exact match scores. | ['Kyoung-Gu Woo', 'Christopher D. Manning', 'Ashwin Paranjape', 'Jongwon Lee', 'Akhil Kedia', 'Haejun Lee'] | 2021-12-14 | null | null | null | null | ['hard-attention', 'triviaqa'] | ['methodology', 'miscellaneous'] | [-9.40997992e-03 5.34664690e-01 8.05641264e-02 -1.90219462e-01
-1.47753072e+00 -8.67274761e-01 6.66024685e-01 2.72430360e-01
-6.32728338e-01 6.85772479e-01 5.35925686e-01 -4.94821936e-01
-1.43407226e-01 -9.17381704e-01 -1.03209555e+00 -2.70904712e-02
2.55046487e-01 1.12845004e+00 8.47606540e-01 -6.37510777e-01
1.46308109e-01 -1.25022367e-01 -1.35585034e+00 6.38387203e-01
1.01147306e+00 7.75933206e-01 2.17218295e-01 1.10261357e+00
-3.50926548e-01 1.38068533e+00 -5.87947965e-01 -8.72732937e-01
2.01964006e-02 -4.83894706e-01 -1.61583376e+00 -3.93098533e-01
6.02638841e-01 -6.36095166e-01 -5.56482553e-01 4.26716179e-01
4.16802853e-01 4.42384541e-01 3.91885877e-01 -5.27900159e-01
-1.36115110e+00 6.74216628e-01 2.65364740e-02 3.71897072e-01
6.07309639e-01 5.81892394e-02 1.60476041e+00 -7.92727947e-01
6.21670663e-01 9.72761869e-01 3.87562752e-01 6.76740468e-01
-1.09684491e+00 -7.14189857e-02 2.02376306e-01 4.34536815e-01
-7.92844951e-01 -4.87273514e-01 4.15107518e-01 7.83153847e-02
1.28061080e+00 2.45088741e-01 3.08801264e-01 7.81983674e-01
-1.73173919e-01 1.00766706e+00 6.67736053e-01 -3.56958270e-01
1.04074106e-01 -1.37123913e-02 5.52555263e-01 8.03094804e-01
-1.50072351e-01 -2.74963021e-01 -3.41903895e-01 -2.76230901e-01
3.61941755e-01 -4.47152741e-02 -4.55476254e-01 -2.61132956e-01
-8.95732284e-01 9.70761716e-01 8.34026814e-01 1.22763515e-01
-3.96006942e-01 2.42620766e-01 1.75302714e-01 8.10238779e-01
4.33019698e-01 8.94408703e-01 -6.81020677e-01 2.83888299e-02
-9.26248550e-01 5.63548625e-01 1.13073266e+00 7.39224374e-01
6.86545193e-01 -6.44206166e-01 -7.06146717e-01 1.00603592e+00
3.58071506e-01 2.39526480e-01 4.35744226e-01 -1.06323659e+00
6.81631565e-01 6.00786507e-01 2.90060461e-01 -4.58489090e-01
-1.12351499e-01 -6.81867063e-01 -4.68106493e-02 -4.41381782e-01
5.95431507e-01 -4.53287624e-02 -9.86488998e-01 1.52679753e+00
2.25376546e-01 1.40862316e-01 3.19646984e-01 8.57407808e-01
1.01596451e+00 1.04266953e+00 1.82054281e-01 3.54669571e-01
1.61057568e+00 -1.58378744e+00 -3.77327621e-01 -2.51647800e-01
7.45869160e-01 -7.39910185e-01 1.23954630e+00 4.28727083e-02
-1.58490551e+00 -3.86116117e-01 -8.60221982e-01 -8.85437489e-01
-2.83320755e-01 -7.31372461e-02 9.80882123e-02 -6.90563396e-03
-1.43507993e+00 5.92722833e-01 -6.61492050e-01 -3.62415463e-01
2.81267852e-01 1.02558233e-01 3.69530497e-03 -3.59687448e-01
-1.35242307e+00 1.19813025e+00 5.47205508e-02 -1.60487056e-01
-1.05374730e+00 -9.48210001e-01 -5.97965360e-01 4.12743270e-01
2.35540733e-01 -1.22211289e+00 1.87797463e+00 -5.57692766e-01
-1.81947958e+00 9.30142999e-01 -3.24871540e-01 -6.90349936e-01
2.39184633e-01 -5.91035604e-01 -1.48316622e-01 4.87239242e-01
1.28491879e-01 6.23563886e-01 6.16359055e-01 -1.13575137e+00
-5.04838109e-01 -2.41954811e-02 7.97837257e-01 3.38300407e-01
-5.38133867e-02 -4.09721509e-02 -6.96798027e-01 -2.52920806e-01
-5.40215969e-02 -6.12553477e-01 -2.19303817e-01 -2.18650401e-01
2.75975969e-02 -5.49703658e-01 2.47221068e-01 -9.48432028e-01
1.24471128e+00 -1.66649532e+00 2.27902815e-01 -1.13205113e-01
3.33243728e-01 2.97953635e-01 -6.82075918e-01 6.49563313e-01
3.20798308e-01 -9.18776691e-02 -3.75101268e-01 -2.66262114e-01
2.42152438e-02 1.89617649e-01 -7.08163738e-01 -2.16601659e-02
4.88397151e-01 1.40032244e+00 -1.15777171e+00 -2.67099231e-01
-3.93334299e-01 1.83561489e-01 -9.64659095e-01 6.56794131e-01
-8.30911160e-01 6.42019436e-02 -6.00801468e-01 1.41623318e-01
2.48914510e-01 -7.81179428e-01 -1.05599083e-01 1.14575677e-01
3.75390857e-01 1.10985041e+00 -5.99830329e-01 1.91099584e+00
-8.50503623e-01 4.04943079e-01 2.84607671e-02 -6.49304807e-01
7.94912696e-01 3.63437235e-01 -1.44942686e-01 -1.08391392e+00
-1.22643173e-01 3.81637186e-01 -1.42638564e-01 -5.41603923e-01
9.25669670e-01 6.51531816e-02 9.17444974e-02 8.46436322e-01
5.38903713e-01 -6.55423105e-02 3.33935708e-01 6.96781278e-01
1.39630413e+00 2.23966554e-01 -2.83515394e-01 -1.35763258e-01
6.17819428e-01 1.27989545e-01 -1.59639314e-01 9.55340743e-01
1.99416906e-01 6.95194364e-01 2.91355819e-01 -2.83478677e-01
-9.09366727e-01 -1.19683874e+00 1.83970779e-01 1.68474209e+00
8.90207216e-02 -2.66616493e-01 -6.46396518e-01 -8.86274457e-01
-7.44031593e-02 9.04643834e-01 -5.18141270e-01 -3.02760750e-01
-8.92827928e-01 -1.13556936e-01 5.53974032e-01 5.07110178e-01
3.70037377e-01 -1.14415646e+00 -5.00551164e-01 4.32447374e-01
-5.89610398e-01 -8.57354879e-01 -3.32150787e-01 1.54732734e-01
-9.79919732e-01 -9.95155692e-01 -1.16080797e+00 -9.93776441e-01
5.37751436e-01 8.50130320e-02 2.02375698e+00 5.24322093e-01
2.45981231e-01 4.78695571e-01 -6.20652080e-01 6.14750795e-02
-4.24320668e-01 8.13209593e-01 -7.88058996e-01 -1.41365558e-01
2.72617519e-01 -4.01760787e-01 -8.84644330e-01 1.63563862e-01
-9.30843234e-01 -1.56054899e-01 5.01773596e-01 9.08470869e-01
4.12098706e-01 -9.43253338e-01 7.52874911e-01 -9.34339345e-01
1.02666450e+00 -7.96483457e-01 -6.50907338e-01 4.95773643e-01
-2.23624974e-01 5.79850197e-01 7.62412071e-01 -1.61787376e-01
-1.21700418e+00 -5.58564544e-01 -5.12260199e-01 -1.44789159e-01
1.32733598e-01 5.32335579e-01 1.91651091e-01 3.74463558e-01
8.86296034e-01 1.76027372e-01 -2.05627874e-01 -6.94837809e-01
9.56171691e-01 5.21364510e-01 5.82955718e-01 -6.32855058e-01
8.57113540e-01 1.91614941e-01 -6.66638672e-01 -3.15628886e-01
-1.35481405e+00 -6.14737391e-01 -2.10696116e-01 2.00760677e-01
7.22216189e-01 -1.00936604e+00 -6.23335123e-01 8.98573641e-03
-1.35562682e+00 -5.46339035e-01 -6.08028948e-01 9.83783826e-02
-4.11207467e-01 1.02922447e-01 -1.04031098e+00 -3.83370042e-01
-6.27866924e-01 -6.38433516e-01 1.05986083e+00 3.09602499e-01
-2.36379296e-01 -9.72035348e-01 4.91033167e-01 8.28712046e-01
8.62139881e-01 -4.94475037e-01 1.07849729e+00 -1.10238886e+00
-9.38210189e-01 -6.83258697e-02 -2.70488441e-01 1.43875003e-01
-4.24182355e-01 -5.76807797e-01 -1.01021039e+00 -1.34227082e-01
-6.92880973e-02 -8.48594606e-01 1.25789893e+00 -1.91511571e-01
8.93947661e-01 -5.49197137e-01 -6.16992749e-02 2.68778682e-01
1.18155360e+00 -2.77742893e-01 7.33147383e-01 4.30317730e-01
3.57849121e-01 5.84031224e-01 1.45764649e-01 -1.10668808e-01
9.22724009e-01 2.07105398e-01 4.03884143e-01 -9.24691150e-04
-3.37838173e-01 -5.76086998e-01 2.49601573e-01 1.03014886e+00
2.91723698e-01 -4.71631765e-01 -7.80090272e-01 1.08094323e+00
-1.77144527e+00 -9.47095215e-01 5.00158556e-02 1.98558760e+00
1.06267214e+00 -7.84934014e-02 -1.39595959e-02 -3.33881736e-01
2.24080756e-01 3.17754000e-01 -5.66778481e-01 -5.17947733e-01
1.47834122e-01 7.27696657e-01 9.27672759e-02 8.62814069e-01
-5.93259394e-01 1.18435919e+00 6.51858473e+00 4.42867458e-01
-6.12974524e-01 1.97661117e-01 5.02993286e-01 -2.42291883e-01
-8.02064657e-01 2.41806895e-01 -5.87740481e-01 1.58644170e-01
1.20692623e+00 6.31308928e-03 5.65544844e-01 6.21250153e-01
-4.95810181e-01 8.35407078e-02 -1.27909720e+00 4.03093159e-01
1.72189191e-01 -1.46232617e+00 4.70665872e-01 -3.43536615e-01
6.16754413e-01 4.63741779e-01 -1.88551053e-01 8.03875268e-01
6.86163902e-01 -9.12112594e-01 5.75362802e-01 6.03204966e-01
2.85466820e-01 -5.09015918e-01 6.23982251e-01 3.31673771e-01
-7.67047346e-01 -1.76407024e-01 -5.59034228e-01 -1.37442708e-01
3.63482445e-01 1.09168828e-01 -6.27176285e-01 5.55739284e-01
5.85171998e-01 2.32438639e-01 -5.88768303e-01 1.04294968e+00
-5.33820570e-01 8.66955876e-01 -2.53056645e-01 -3.23386103e-01
4.96341914e-01 9.46909115e-02 3.01932245e-01 9.03363645e-01
2.51144767e-02 2.91451395e-01 -1.53639942e-01 8.84399116e-01
-5.88377416e-01 3.62791978e-02 -4.50954437e-02 1.15847848e-01
3.87053430e-01 1.06883383e+00 -3.96448486e-02 -6.14425004e-01
-6.23969436e-01 1.19684660e+00 9.61787462e-01 6.58974349e-01
-6.71521127e-01 -6.54254317e-01 3.59837353e-01 2.55917490e-01
6.97680056e-01 9.96486563e-03 1.68906897e-01 -1.40996718e+00
1.85376644e-01 -1.01321495e+00 7.58481860e-01 -9.19428766e-01
-1.39376974e+00 8.31328392e-01 -3.63622189e-01 -7.00120986e-01
-5.43849826e-01 -4.60517675e-01 -6.50737643e-01 1.11133718e+00
-2.17415142e+00 -1.02037978e+00 1.46624370e-04 5.88638425e-01
5.47162235e-01 1.61897212e-01 9.23252046e-01 2.42956683e-01
-7.87520856e-02 5.99132717e-01 1.25875443e-01 3.06393981e-01
5.60958326e-01 -1.31133854e+00 7.55333066e-01 7.62954712e-01
3.36661249e-01 8.64591956e-01 5.07704258e-01 -2.06134468e-01
-1.32204509e+00 -9.08620834e-01 1.42289221e+00 -9.57590878e-01
8.57111514e-01 -2.17049092e-01 -1.31706023e+00 7.39392400e-01
7.46123195e-01 -1.78049728e-01 7.49717176e-01 4.93552387e-01
-6.56358778e-01 -4.95740995e-02 -7.65937507e-01 6.04716539e-01
9.16066885e-01 -8.13342929e-01 -1.46577013e+00 2.83845246e-01
1.11407375e+00 -4.25197631e-01 -8.53294790e-01 4.02397886e-02
3.00114006e-01 -6.74422920e-01 9.21118021e-01 -1.01704633e+00
6.59878314e-01 -2.42640018e-01 1.91385671e-02 -1.17632604e+00
-4.17255491e-01 -6.41202569e-01 -6.19753182e-01 1.04537046e+00
9.93457496e-01 -5.99427402e-01 5.90147614e-01 7.24833786e-01
-2.52748579e-01 -9.57479775e-01 -8.40191185e-01 -4.30601746e-01
4.66389537e-01 2.28984188e-02 7.06606030e-01 4.59821105e-01
8.90710875e-02 9.70432222e-01 5.24594216e-05 7.33400881e-02
1.88376486e-01 3.78517270e-01 6.76938236e-01 -1.06678379e+00
-6.14524603e-01 -3.62726033e-01 3.66985738e-01 -1.97785711e+00
9.07337591e-02 -1.16188180e+00 1.50880814e-01 -2.04245257e+00
1.08938932e-01 -3.29367101e-01 -4.23639864e-01 3.45030129e-01
-6.37949407e-01 6.33080304e-02 1.84462771e-01 2.56177217e-01
-1.22841227e+00 7.89241016e-01 1.23625970e+00 -1.49675250e-01
-9.62508693e-02 -2.09631190e-01 -1.09392500e+00 2.57965565e-01
6.91282928e-01 -4.92900908e-01 -5.57206333e-01 -1.31951058e+00
5.39812922e-01 1.68209061e-01 3.72610390e-01 -7.71340847e-01
4.71026599e-01 4.89221424e-01 1.91649765e-01 -2.86059678e-01
3.40483904e-01 -4.25698727e-01 -7.09427536e-01 2.64783561e-01
-8.59171093e-01 1.75541431e-01 6.62654266e-02 4.61313933e-01
-2.73759067e-01 -4.36747193e-01 5.17756402e-01 -2.76854783e-01
-3.55523378e-01 2.11877242e-01 -2.24806100e-01 7.40647018e-01
3.32777053e-01 1.93049446e-01 -6.44557476e-01 -7.14172363e-01
-5.31706631e-01 7.24037349e-01 2.65486419e-01 6.06515884e-01
4.05439138e-01 -9.32018399e-01 -9.81977105e-01 -8.01926702e-02
1.99696004e-01 6.79525584e-02 2.27469087e-01 3.52646738e-01
-5.59206188e-01 6.45324945e-01 2.21487179e-01 -2.57137656e-01
-6.11934841e-01 4.42146838e-01 5.25995135e-01 -9.60347652e-01
-3.95017952e-01 1.08937788e+00 -3.50557715e-02 -8.91427696e-01
9.80730727e-02 -3.74015898e-01 -4.30091977e-01 -3.50496545e-02
6.89658165e-01 3.30563486e-01 2.18410790e-01 -2.90659904e-01
-1.67774633e-01 4.45960343e-01 -4.80153918e-01 -3.13861549e-01
1.11952317e+00 -1.67034477e-01 -2.84474692e-03 5.11228032e-02
1.28412867e+00 -1.50900558e-01 -9.71801996e-01 -6.12866938e-01
6.53189942e-02 -2.65406482e-02 -2.30991498e-01 -1.16874337e+00
-6.50057077e-01 8.45278680e-01 6.63467497e-02 2.73332447e-01
9.04800951e-01 4.99627680e-01 1.21895218e+00 8.51788640e-01
8.36212486e-02 -7.69643128e-01 3.61889243e-01 1.07130134e+00
1.02924275e+00 -9.72925663e-01 -4.41098571e-01 1.34491652e-01
-4.90469873e-01 6.76582515e-01 6.00832999e-01 -3.79189670e-01
2.07455561e-01 -3.44363391e-01 3.13832104e-01 -2.52705723e-01
-1.21936893e+00 -4.53774989e-01 4.11465049e-01 2.58135647e-01
6.48755431e-01 -4.28320318e-01 -1.33303016e-01 6.41940534e-01
-3.18621904e-01 8.62239674e-02 2.47487575e-01 8.69659424e-01
-6.13977134e-01 -1.10309672e+00 -5.26976399e-02 5.14485002e-01
-7.15758979e-01 -4.91421342e-01 -2.63697416e-01 3.61669302e-01
-5.70123017e-01 1.24903607e+00 2.72183120e-01 8.37116167e-02
5.67889035e-01 3.74132127e-01 3.50219458e-01 -6.71545327e-01
-1.06402004e+00 -6.17203653e-01 4.51556444e-01 -6.40100360e-01
5.61926328e-02 -1.15221299e-01 -1.24812365e+00 8.08768440e-03
-4.68598068e-01 7.38618374e-01 2.19749942e-01 8.84465754e-01
9.39854383e-01 5.71019232e-01 3.97375643e-01 -1.77164793e-01
-8.57758880e-01 -9.64172363e-01 1.13493785e-01 5.17949045e-01
5.24362683e-01 -3.16067971e-02 -2.03498647e-01 -5.41672930e-02] | [11.271342277526855, 7.959482669830322] |
9f77aa55-4a9b-4d60-a92d-bb8cb64d47a1 | autoassign-differentiable-label-assignment | 2007.03496 | null | https://arxiv.org/abs/2007.03496v3 | https://arxiv.org/pdf/2007.03496v3.pdf | AutoAssign: Differentiable Label Assignment for Dense Object Detection | Determining positive/negative samples for object detection is known as label assignment. Here we present an anchor-free detector named AutoAssign. It requires little human knowledge and achieves appearance-aware through a fully differentiable weighting mechanism. During training, to both satisfy the prior distribution of data and adapt to category characteristics, we present Center Weighting to adjust the category-specific prior distributions. To adapt to object appearances, Confidence Weighting is proposed to adjust the specific assign strategy of each instance. The two weighting modules are then combined to generate positive and negative weights to adjust each location's confidence. Extensive experiments on the MS COCO show that our method steadily surpasses other best sampling strategies by large margins with various backbones. Moreover, our best model achieves 52.1% AP, outperforming all existing one-stage detectors. Besides, experiments on other datasets, e.g., PASCAL VOC, Objects365, and WiderFace, demonstrate the broad applicability of AutoAssign. | ['Jian-Feng Wang', 'Benjin Zhu', 'Songtao Liu', 'Zeming Li', 'Jian Sun', 'Fuhang Zong', 'Zhengkai Jiang'] | 2020-07-07 | null | null | null | null | ['dense-object-detection'] | ['computer-vision'] | [ 4.44471352e-02 -1.95353568e-01 -4.23189789e-01 -6.18903399e-01
-6.57632947e-01 -5.96351445e-01 4.10384983e-01 -1.02476224e-01
-5.64877689e-01 4.08098578e-01 -3.34431738e-01 5.47807217e-02
4.08176184e-01 -4.37180489e-01 -6.19146645e-01 -6.24089003e-01
1.29504070e-01 3.64743173e-01 9.26373661e-01 4.39759418e-02
2.88934499e-01 5.74108839e-01 -1.58404863e+00 3.68922591e-01
7.78426945e-01 1.33849180e+00 3.89681906e-01 4.80345607e-01
-6.47981837e-02 4.03185099e-01 -5.97260416e-01 -5.14034927e-01
2.54662722e-01 7.80157372e-02 -3.91915202e-01 4.56302047e-01
7.74876356e-01 -2.65194267e-01 -7.04292357e-02 1.12588120e+00
3.87463331e-01 8.39478243e-03 9.60599422e-01 -1.32619667e+00
-8.46131504e-01 3.01978230e-01 -1.14054906e+00 3.16405535e-01
-3.46253030e-02 1.49067774e-01 9.71223712e-01 -1.49141669e+00
3.90998572e-01 1.37202179e+00 5.94273686e-01 7.56117463e-01
-1.14086664e+00 -8.54161322e-01 6.63005888e-01 3.50718975e-01
-1.59032118e+00 -4.20536280e-01 6.85410619e-01 -3.58069420e-01
4.58980709e-01 1.30371362e-01 3.89833421e-01 9.01087701e-01
-1.28064156e-01 1.23473191e+00 9.42728937e-01 -4.30434942e-01
2.39884779e-01 4.92741227e-01 9.14172009e-02 7.60482430e-01
3.33644718e-01 -1.01028435e-01 -3.44916314e-01 -6.11918941e-02
6.38243318e-01 5.83117269e-02 -1.20047159e-01 -8.20623040e-01
-1.02693129e+00 6.25002444e-01 7.43196785e-01 -1.72402024e-01
-1.23982824e-01 1.16016328e-01 2.35127017e-01 -7.78748617e-02
3.63686562e-01 2.59240836e-01 -5.29766321e-01 5.27981579e-01
-5.58379889e-01 -7.12995902e-02 3.81064177e-01 1.35338378e+00
7.60945320e-01 -5.72783872e-02 -6.06987059e-01 1.24484468e+00
5.58991730e-01 7.11866200e-01 3.72582674e-01 -8.44950438e-01
3.38050723e-01 7.41871715e-01 1.70034587e-01 -8.25384498e-01
-2.43951425e-01 -6.53056920e-01 -6.02011859e-01 3.60771418e-01
4.58553672e-01 -2.34858058e-02 -1.35824656e+00 1.54042768e+00
5.89861333e-01 2.01787531e-01 -2.85006195e-01 9.77885008e-01
8.01862180e-01 3.04879189e-01 4.34007883e-01 1.54269695e-01
1.55342138e+00 -1.11230648e+00 -4.21388477e-01 -5.61390162e-01
2.43336916e-01 -9.15911019e-01 9.39144909e-01 3.58220458e-01
-7.76467800e-01 -7.15761423e-01 -1.16394937e+00 3.33196551e-01
-3.01731557e-01 6.98581576e-01 7.21248627e-01 6.92258596e-01
-6.47288740e-01 2.40149990e-01 -7.06399977e-01 -3.00706267e-01
6.92172110e-01 2.50504583e-01 -2.23458067e-01 -1.78498790e-01
-7.38666296e-01 6.54526532e-01 4.62838858e-01 3.73739786e-02
-8.32025886e-01 -6.21032953e-01 -7.16193199e-01 -1.88921198e-01
4.58532989e-01 -5.03513396e-01 1.36715543e+00 -9.91499066e-01
-1.29131281e+00 1.06780052e+00 -1.59021363e-01 -2.70088911e-01
6.37407184e-01 -5.42308390e-02 -6.98609650e-01 1.26576751e-01
1.93572640e-01 1.08778441e+00 1.15085316e+00 -1.39857864e+00
-1.25963759e+00 -1.04570858e-01 -8.80873017e-03 1.46900490e-01
-5.55502653e-01 -7.73027074e-03 -1.14599097e+00 -6.35582805e-01
2.10802361e-01 -8.45974922e-01 -3.16003263e-01 5.68326533e-01
-3.62657785e-01 -4.31519181e-01 7.92197168e-01 -1.03111289e-01
1.08985734e+00 -2.43826652e+00 -3.26896101e-01 3.32937419e-01
2.35320970e-01 2.84697562e-01 -3.17936331e-01 -1.26008689e-01
1.32815372e-02 -2.70911366e-01 -1.10046767e-01 -3.24107379e-01
5.77437393e-02 1.03200138e-01 7.77511671e-02 6.14086568e-01
6.77536368e-01 6.90454364e-01 -8.50657880e-01 -8.49812269e-01
2.75121421e-01 2.96611607e-01 -7.01982975e-01 1.03389874e-01
-5.51518761e-02 -2.07429454e-01 -4.86746639e-01 1.07390392e+00
9.53279495e-01 -5.01913607e-01 3.63348238e-02 -4.57057178e-01
2.01965511e-01 -3.42847586e-01 -1.52144468e+00 1.25652921e+00
-1.89728111e-01 3.85710835e-01 2.32064500e-01 -6.80360436e-01
1.00288415e+00 -1.45031154e-01 1.28460452e-01 -5.45497835e-01
8.01797435e-02 1.08425111e-01 1.01154763e-02 -2.01829091e-01
5.16876638e-01 3.78654122e-01 1.16781399e-01 -1.94661729e-02
5.34953438e-02 1.93315014e-01 3.93975079e-01 3.35871637e-01
7.72448838e-01 9.64591280e-02 3.34561229e-01 -2.87489623e-01
6.31294668e-01 -9.49965641e-02 7.57498980e-01 8.49163830e-01
-6.97612464e-01 9.15385664e-01 7.65703991e-02 -2.51492262e-01
-7.89342523e-01 -1.22908986e+00 -5.53713381e-01 1.43676960e+00
5.09471178e-01 2.81918701e-03 -6.45832181e-01 -1.13467264e+00
3.35107088e-01 4.31679934e-01 -7.92772591e-01 -2.10062683e-01
-3.46374750e-01 -8.83428514e-01 2.63682026e-02 9.57796633e-01
3.56017917e-01 -9.15708244e-01 -2.27771088e-01 1.09667271e-01
1.24880988e-02 -1.08268785e+00 -8.90302479e-01 1.43300876e-01
-4.85691428e-01 -1.22533417e+00 -8.99105966e-01 -9.68820095e-01
9.97017920e-01 5.20476282e-01 1.09809756e+00 1.18031595e-02
-4.86894757e-01 4.26238596e-01 -3.70055884e-01 -4.51652616e-01
-1.08015701e-01 8.76255259e-02 1.46187618e-01 3.93303037e-01
5.54800689e-01 -1.18846692e-01 -1.02137470e+00 8.27015936e-01
-6.26998127e-01 -3.08431298e-01 9.38606858e-01 7.65946805e-01
7.63764977e-01 -2.74502635e-01 5.67671657e-01 -8.45106244e-01
2.22011358e-01 -3.59343916e-01 -6.33627355e-01 4.60069776e-01
-7.25671351e-01 -1.09967843e-01 1.58495367e-01 -9.31347549e-01
-1.16264248e+00 5.34325719e-01 1.22045718e-01 -5.45519292e-01
-2.00151145e-01 -2.60913342e-01 -2.80842662e-01 -2.61590362e-01
8.59244823e-01 -2.59891106e-03 -2.10989833e-01 -3.72122467e-01
4.91369933e-01 6.79228961e-01 9.21245873e-01 -4.91282046e-01
9.49418545e-01 5.04909337e-01 -4.16082174e-01 -3.77726793e-01
-9.32846785e-01 -8.69858861e-01 -6.46068811e-01 -3.03249955e-01
6.80452287e-01 -1.10925305e+00 -6.03513479e-01 3.91517520e-01
-9.28225935e-01 -1.50762033e-02 -2.09980100e-01 3.64602387e-01
-1.91896215e-01 2.96217352e-01 -5.63868940e-01 -8.07426751e-01
-2.72046924e-01 -9.28173602e-01 1.17583513e+00 5.34066260e-01
4.78727603e-03 -6.98448002e-01 -3.91207606e-01 7.69126713e-02
2.63629109e-01 -6.75458908e-02 2.41661191e-01 -5.25696158e-01
-4.61843073e-01 -4.99244183e-01 -5.98348737e-01 3.62213224e-01
1.16867170e-01 3.35807294e-01 -1.13603508e+00 -3.72079939e-01
-6.31620228e-01 -3.60569030e-01 1.15398920e+00 2.87463248e-01
1.46521068e+00 1.03742152e-01 -6.03251934e-01 5.75109601e-01
1.18953538e+00 8.85062292e-02 4.99618769e-01 3.79223078e-01
6.16223454e-01 3.22683990e-01 1.04368365e+00 5.46344936e-01
2.64296204e-01 6.34685278e-01 6.59778535e-01 -3.33975255e-01
-3.46764117e-01 -2.13750020e-01 1.60238877e-01 2.92933345e-01
1.49719724e-02 -7.66915530e-02 -4.39990759e-01 4.59333479e-01
-1.80946422e+00 -9.02059734e-01 -2.18021348e-01 2.01622462e+00
9.39916432e-01 3.90268862e-01 2.26356566e-01 -1.21553734e-01
1.00224078e+00 -1.13814540e-01 -8.30994189e-01 1.67107075e-01
-2.12386206e-01 -1.16505049e-01 7.79164374e-01 2.19703153e-01
-1.49897707e+00 9.78488624e-01 6.94306421e+00 1.20441639e+00
-8.21293533e-01 1.11815007e-02 7.02231228e-01 6.19358523e-03
3.26887444e-02 -3.57060581e-01 -1.25766075e+00 4.46745515e-01
1.04227804e-01 3.59306671e-02 7.87951201e-02 1.39440274e+00
-2.79047847e-01 5.02634160e-02 -9.53679442e-01 8.79420459e-01
7.70749599e-02 -1.00863636e+00 -1.46352530e-01 -3.41227859e-01
7.82346487e-01 -6.23207586e-03 2.00390711e-01 4.78337824e-01
5.47888219e-01 -5.27999341e-01 9.40460384e-01 3.43637675e-01
9.18047547e-01 -6.17256761e-01 6.56728327e-01 8.41550976e-02
-1.42010999e+00 -9.46814939e-02 -6.21242046e-01 3.76373589e-01
3.52225862e-02 4.83719170e-01 -8.95468414e-01 1.08032592e-01
8.63870263e-01 6.02665186e-01 -9.61676419e-01 1.38008988e+00
-4.30948138e-01 5.67370355e-01 -2.80341655e-01 -1.77636910e-02
1.06202938e-01 2.21755028e-01 2.53145009e-01 1.49345911e+00
4.09859121e-02 -4.87041138e-02 4.38700318e-01 7.20267773e-01
-1.12181101e-02 3.60826179e-02 1.06584445e-01 3.85207891e-01
6.86994851e-01 1.75187337e+00 -7.48379350e-01 -4.36786950e-01
-3.74932826e-01 1.18052042e+00 3.51406962e-01 4.26372230e-01
-1.04212916e+00 -3.61281425e-01 5.76306820e-01 2.95177754e-02
6.35998428e-01 8.46013874e-02 -8.30582008e-02 -9.66943622e-01
5.36808223e-02 -5.88670135e-01 6.62400126e-01 -6.50270402e-01
-1.66888273e+00 5.18794656e-01 -3.81276309e-02 -1.53452170e+00
6.55723959e-02 -1.01124918e+00 -6.69007361e-01 6.79707587e-01
-1.63836634e+00 -1.18228829e+00 -4.68104690e-01 5.84813654e-01
6.29871845e-01 -2.31761709e-01 5.16035974e-01 5.11196256e-01
-7.99242020e-01 1.00179136e+00 -9.11869481e-02 1.97179452e-01
1.12052786e+00 -1.37715590e+00 2.94078767e-01 7.89153874e-01
-3.56032029e-02 5.14452577e-01 5.26056707e-01 -3.73567700e-01
-1.02481484e+00 -1.32939756e+00 4.67871308e-01 -4.74380076e-01
6.24063134e-01 -3.55042756e-01 -9.79885995e-01 4.04423773e-01
5.40322158e-03 6.18072212e-01 4.91157919e-01 1.07278690e-01
-7.12628245e-01 -4.45322931e-01 -1.22643518e+00 5.54104209e-01
1.20300066e+00 -9.12259966e-02 -3.25291991e-01 4.36915725e-01
5.33568144e-01 -5.07777095e-01 -5.91308296e-01 6.04770839e-01
5.94650745e-01 -6.67429090e-01 1.19006050e+00 -4.02400732e-01
6.99777231e-02 -7.80654550e-01 -1.76605627e-01 -1.09955835e+00
-9.32293832e-01 -1.32033274e-01 -2.62639046e-01 1.46920729e+00
6.06037855e-01 -5.91984570e-01 8.99519801e-01 5.67831039e-01
-1.84694603e-01 -6.49032354e-01 -6.96965516e-01 -8.72961402e-01
-2.93340206e-01 -3.52344185e-01 5.58449268e-01 7.63200283e-01
-4.27496672e-01 2.47437194e-01 -2.38232002e-01 4.55569118e-01
8.95901680e-01 3.37803848e-02 6.96414530e-01 -1.07931352e+00
-4.10800993e-01 -5.85546792e-01 -4.10476059e-01 -1.21606946e+00
-2.15644278e-02 -6.97936296e-01 2.22646564e-01 -1.14554858e+00
4.70993042e-01 -8.05499196e-01 -7.90862143e-01 6.09869719e-01
-5.75390100e-01 8.36331189e-01 7.61750415e-02 2.11327285e-01
-1.00855577e+00 4.25530732e-01 1.21762741e+00 -5.22319913e-01
5.20259514e-02 1.66114464e-01 -7.78667986e-01 8.80993545e-01
7.25471556e-01 -4.65990692e-01 -2.36774087e-01 -1.94679469e-01
-2.81155735e-01 -6.67646706e-01 2.55858839e-01 -1.04387391e+00
2.43086353e-01 -4.12525028e-01 8.64150941e-01 -7.79645920e-01
2.15554312e-01 -8.84213984e-01 -2.67118514e-01 3.20704788e-01
-2.08105877e-01 -8.63809437e-02 1.51734814e-01 8.45508516e-01
-7.82141536e-02 -1.60997555e-01 1.15244353e+00 1.90723345e-01
-1.37288737e+00 6.24669135e-01 -3.62958461e-02 9.52556655e-02
1.24922240e+00 -3.88814807e-01 -2.54896671e-01 1.02673359e-01
-5.58127403e-01 7.18587518e-01 3.16886425e-01 5.74771047e-01
5.44317603e-01 -1.64893401e+00 -7.10403204e-01 1.83986381e-01
7.85481811e-01 -1.79938152e-01 2.09600702e-01 6.52699351e-01
-1.00963503e-01 -7.66975582e-02 -4.77807596e-04 -9.30388331e-01
-1.28425658e+00 6.61611736e-01 4.07354534e-01 1.67994872e-01
-4.38429475e-01 1.16603637e+00 6.01561427e-01 -2.70457208e-01
5.61889827e-01 -1.16749212e-01 -3.82087529e-01 -3.29609811e-02
8.30417931e-01 1.62397012e-01 -1.22414604e-01 -4.66579348e-01
-6.93814695e-01 5.31466603e-01 -4.65388030e-01 2.41180837e-01
8.86224508e-01 -8.27446282e-02 3.91968310e-01 8.06481540e-02
8.86823058e-01 2.29399968e-02 -1.80660641e+00 -4.52810585e-01
-2.21465658e-02 -5.69249094e-01 -2.63726771e-01 -8.99582624e-01
-1.40592444e+00 5.92622519e-01 9.19166148e-01 -9.55719501e-02
1.03148568e+00 2.01450601e-01 2.69648433e-01 2.65443414e-01
1.83653891e-01 -1.31090677e+00 3.65266889e-01 2.02135772e-01
8.03094685e-01 -1.58244932e+00 -2.47187670e-02 -7.30108380e-01
-7.81136751e-01 9.46653306e-01 1.22549713e+00 -5.78185990e-02
5.22801638e-01 3.29149395e-01 2.15840057e-01 1.82313666e-01
-5.67442596e-01 -4.79670435e-01 7.16827214e-01 8.04522514e-01
2.76305079e-01 2.03751683e-01 -2.26584002e-01 4.49764162e-01
3.40281397e-01 -2.78589636e-01 1.27832945e-02 5.89799464e-01
-7.89566517e-01 -9.99687135e-01 -5.64503670e-01 4.42944020e-01
-2.79161513e-01 1.62796944e-01 -3.32974255e-01 8.57006967e-01
2.58193254e-01 9.36515450e-01 1.20702535e-01 -2.56752849e-01
5.52075028e-01 -2.33510315e-01 3.31445843e-01 -5.93167484e-01
-2.93626368e-01 1.02682650e-01 1.33649642e-02 -4.78326857e-01
-2.98271835e-01 -6.22594416e-01 -1.31004608e+00 -3.89450863e-02
-7.70247877e-01 -1.64677471e-01 4.32068676e-01 4.88414407e-01
2.32581466e-01 5.70765495e-01 6.92550957e-01 -8.52487683e-01
-8.26489925e-01 -9.39018309e-01 -5.92261553e-01 6.14287496e-01
7.29273781e-02 -8.53553116e-01 -3.15084130e-01 2.06695292e-02] | [9.234554290771484, 1.3202623128890991] |
bfb466ee-5b16-41d1-a057-fef28448e8d4 | semi-offline-reinforcement-learning-for | 2306.09712 | null | https://arxiv.org/abs/2306.09712v1 | https://arxiv.org/pdf/2306.09712v1.pdf | Semi-Offline Reinforcement Learning for Optimized Text Generation | In reinforcement learning (RL), there are two major settings for interacting with the environment: online and offline. Online methods explore the environment at significant time cost, and offline methods efficiently obtain reward signals by sacrificing exploration capability. We propose semi-offline RL, a novel paradigm that smoothly transits from offline to online settings, balances exploration capability and training cost, and provides a theoretical foundation for comparing different RL settings. Based on the semi-offline formulation, we present the RL setting that is optimal in terms of optimization cost, asymptotic error, and overfitting error bound. Extensive experiments show that our semi-offline approach is efficient and yields comparable or often better performance compared with state-of-the-art methods. | ['Rui Yan', 'Yi Liu', 'Jie Cao', 'Li Dong', 'Victor Ye Dong', 'Yiqiao Jin', 'Xiting Wang', 'Changyu Chen'] | 2023-06-16 | null | null | null | null | ['text-generation', 'offline-rl'] | ['natural-language-processing', 'playing-games'] | [-2.94054955e-01 6.26658872e-02 -6.25081539e-01 2.52080038e-02
-8.84434760e-01 -8.37037385e-01 2.90847272e-01 1.23743623e-01
-8.54143739e-01 8.98964286e-01 -2.36688778e-01 -4.66229945e-01
-2.30608627e-01 -4.82201934e-01 -7.36710548e-01 -4.14461017e-01
-6.49317324e-01 3.25254411e-01 -9.97650176e-02 -1.72233865e-01
2.42481574e-01 4.29577142e-01 -1.50111902e+00 -4.91350383e-01
8.23848546e-01 1.08833349e+00 4.21602696e-01 6.40873551e-01
1.22063883e-01 8.98422539e-01 -2.87482888e-01 -7.53497779e-02
6.96899831e-01 -5.21073401e-01 -7.86499441e-01 -2.37600282e-01
-3.06174517e-01 -8.02709460e-01 -6.33257866e-01 9.48491395e-01
5.67498147e-01 4.98875439e-01 -3.50120058e-03 -1.26334190e+00
-5.55059969e-01 8.39487255e-01 -3.88486177e-01 3.16431940e-01
5.08605421e-01 3.90228093e-01 1.03366303e+00 -5.46257079e-01
4.32813138e-01 1.13200092e+00 3.28188866e-01 7.28442907e-01
-1.46441686e+00 -5.03975213e-01 6.15654230e-01 2.32357025e-01
-9.09460425e-01 -4.64391023e-01 6.81138754e-01 -2.04613190e-02
9.41474974e-01 -1.11382119e-02 8.68457317e-01 1.23827720e+00
-5.76264411e-02 1.33753955e+00 1.58251059e+00 -3.25595021e-01
8.41903031e-01 7.27628097e-02 -3.26478481e-01 5.88785708e-01
1.56788472e-02 8.86202991e-01 -7.53119111e-01 -1.26275241e-01
1.11890054e+00 -1.68499231e-01 -9.16216448e-02 -6.31609440e-01
-1.03809583e+00 7.31874883e-01 3.11651617e-01 -7.92590603e-02
-3.18577290e-01 5.17125607e-01 2.88129717e-01 7.68795788e-01
-3.62623446e-02 7.92060554e-01 -3.56067836e-01 -7.51385987e-01
-5.97078979e-01 3.05419087e-01 8.04795146e-01 1.16376615e+00
7.50164032e-01 2.20980674e-01 -1.79947734e-01 5.81369877e-01
6.00493997e-02 4.03778434e-01 6.46916330e-01 -1.29615581e+00
5.07860303e-01 2.71587610e-01 8.30288649e-01 -5.80419421e-01
-3.17265391e-01 -6.54408038e-01 -5.21223247e-02 3.03710192e-01
2.01507196e-01 -3.99183333e-01 -2.76478618e-01 2.00633121e+00
4.34513688e-01 -5.99317104e-02 1.99315444e-01 9.31223392e-01
2.20002115e-01 2.32116982e-01 -1.08637102e-01 -4.76057112e-01
6.90432966e-01 -1.39233613e+00 -8.98357093e-01 -6.26000345e-01
4.53358620e-01 1.48334112e-02 1.49688315e+00 4.75934356e-01
-1.31872320e+00 -1.86637372e-01 -1.21777618e+00 2.76330948e-01
-1.82277024e-01 1.08433753e-01 9.32489097e-01 5.80482244e-01
-1.02453446e+00 1.12925982e+00 -1.07339585e+00 -5.27730398e-02
1.84520647e-01 3.15669656e-01 -6.23824820e-02 2.56679207e-01
-9.40249324e-01 9.01899457e-01 3.71930122e-01 2.09731251e-01
-1.50385988e+00 -2.98672915e-01 -7.63106048e-01 8.54469463e-03
9.68132079e-01 -2.36957684e-01 1.86649930e+00 -7.49088883e-01
-2.25356936e+00 3.42671841e-01 4.28745374e-02 -6.51058555e-01
9.32187200e-01 -5.04835486e-01 -1.45950526e-01 2.97669381e-01
-1.48820490e-01 4.73968387e-01 6.88975275e-01 -1.12359762e+00
-5.03614187e-01 -2.04041630e-01 4.53394085e-01 5.92063248e-01
-4.33233947e-01 -2.22991288e-01 -3.09379071e-01 -2.95031428e-01
-6.20127432e-02 -9.15842712e-01 -5.67805290e-01 -7.65127763e-02
2.42004190e-02 -1.20104589e-01 3.67176384e-01 -3.19772482e-01
1.17370570e+00 -1.99930441e+00 -2.26149522e-03 1.74848303e-01
1.68198928e-01 -1.63836807e-01 -2.95457840e-01 7.16211140e-01
3.20794940e-01 -7.14200810e-02 1.51941180e-01 -3.90559852e-01
3.80268961e-01 2.75390357e-01 -5.13313651e-01 4.24575061e-01
-1.34417236e-01 1.09144366e+00 -1.54729378e+00 -3.47398281e-01
4.21040356e-02 -2.48573825e-01 -6.23594165e-01 6.25765204e-01
-1.82524279e-01 6.32839262e-01 -5.58530033e-01 6.97074533e-01
1.84980780e-01 -2.40001857e-01 5.82839966e-01 6.07483566e-01
-1.26504391e-01 4.90430832e-01 -1.34547544e+00 1.71736324e+00
-7.39699423e-01 3.57387155e-01 2.09773481e-01 -7.92047620e-01
6.77123308e-01 1.94548517e-01 3.33961487e-01 -1.00308073e+00
1.44588500e-01 2.94350237e-01 -2.59661019e-01 -4.63608474e-01
5.97061872e-01 2.50619829e-01 -1.94118023e-02 7.23241925e-01
1.12368114e-01 1.71987116e-01 2.21703187e-01 -1.04530424e-01
1.25823665e+00 7.06337571e-01 3.88643831e-01 -5.45692369e-02
-9.35025960e-02 -2.92656153e-01 5.46850562e-01 1.40636730e+00
-7.04942346e-01 -2.52574950e-01 5.47330320e-01 -2.83063054e-01
-5.82433164e-01 -1.04254770e+00 2.89720625e-01 1.48542476e+00
4.69087809e-01 -2.68265843e-01 -6.01212442e-01 -7.37089455e-01
2.00132832e-01 6.65434897e-01 -6.79445922e-01 -1.47310212e-01
-5.18738747e-01 -1.22382805e-01 3.97882223e-01 7.87079811e-01
5.03831089e-01 -1.10367155e+00 -1.30933881e+00 4.05223429e-01
-7.01989466e-03 -9.98430133e-01 -4.87265527e-01 6.60071671e-01
-1.16640794e+00 -7.79266655e-01 -4.49252218e-01 -3.61438185e-01
6.02411509e-01 3.74637991e-01 9.56680834e-01 2.39730794e-02
-9.31731611e-02 7.26059735e-01 -3.75987977e-01 -1.59296870e-01
-9.90783423e-02 -2.01052219e-01 4.15265590e-01 -2.96823949e-01
-1.30636021e-01 -7.16849566e-01 -9.25285578e-01 2.32565418e-01
-5.65473914e-01 -2.28668734e-01 6.62898540e-01 9.44488227e-01
5.79983115e-01 -1.04231246e-01 8.16134810e-01 -5.83530366e-01
7.18435287e-01 -4.33971107e-01 -1.03124011e+00 1.90882012e-01
-1.13992584e+00 4.74150300e-01 9.15462196e-01 -6.49443090e-01
-9.26248491e-01 7.95564279e-02 3.12826246e-01 -7.93984830e-01
1.02308027e-01 4.59972948e-01 9.53004658e-02 -3.52570474e-01
6.70928597e-01 3.82850885e-01 9.35646966e-02 -5.67639768e-01
4.28687900e-01 2.61455625e-01 4.06425655e-01 -9.69719172e-01
7.40431070e-01 3.24254692e-01 -2.10782617e-01 -2.59423673e-01
-9.14740384e-01 -2.31123537e-01 -1.74788415e-01 -3.08066934e-01
2.96338145e-02 -7.50328898e-01 -1.18591881e+00 2.32322924e-02
-4.34324950e-01 -1.01158679e+00 -6.68875217e-01 5.03740788e-01
-1.12690258e+00 2.41936997e-01 -5.56516349e-01 -1.44222975e+00
-3.71837802e-02 -1.14924550e+00 6.19789958e-01 4.14368242e-01
1.72197014e-01 -9.17028785e-01 2.89238729e-02 1.89869888e-02
5.15777171e-01 7.59066641e-02 4.30210024e-01 -4.36190307e-01
-6.56840384e-01 2.94476077e-02 1.93183169e-01 1.59659460e-02
-1.54347733e-01 -5.87274790e-01 -8.80862117e-01 -7.07087755e-01
-8.60340521e-02 -9.67798710e-01 5.53905845e-01 6.14172518e-02
1.46122420e+00 -6.93798125e-01 -3.34398523e-02 5.21744967e-01
1.45961297e+00 2.71946341e-01 3.13938707e-01 7.30339408e-01
8.46087486e-02 4.10559446e-01 1.12478018e+00 1.07187736e+00
4.63777959e-01 4.06848818e-01 5.93577206e-01 1.59855619e-01
5.10564327e-01 -9.73584294e-01 7.29914367e-01 4.48724896e-01
1.64049178e-01 1.74447998e-01 -4.43803549e-01 5.66370070e-01
-2.26040459e+00 -9.12888110e-01 9.70894635e-01 2.61368251e+00
1.18688834e+00 1.89581886e-01 5.46093643e-01 -1.28894955e-01
2.50327796e-01 1.69676589e-03 -1.34096324e+00 -4.51023430e-01
1.91652954e-01 1.83623269e-01 6.97695494e-01 3.98941189e-01
-8.50867450e-01 1.01423955e+00 8.21938610e+00 6.82250917e-01
-8.74869108e-01 1.98582515e-01 4.57056940e-01 -5.11868596e-01
-9.81866643e-02 1.14559703e-01 -5.99724472e-01 1.73013806e-01
9.92495298e-01 -3.92472982e-01 1.43620098e+00 1.36694467e+00
2.07235932e-01 -2.29766667e-01 -1.66839004e+00 9.88512754e-01
-3.65427911e-01 -9.76263762e-01 -6.02594614e-01 4.78543788e-02
6.05649054e-01 3.71125643e-03 2.18100682e-01 8.06392372e-01
7.59802639e-01 -9.07687068e-01 1.05096626e+00 2.04976171e-01
7.23838568e-01 -8.62519264e-01 2.70440042e-01 6.70928657e-01
-1.08574688e+00 -6.62261307e-01 -1.19864039e-01 -2.71522850e-01
-2.32976243e-01 4.62704487e-02 -5.17921209e-01 2.80361474e-01
7.56928444e-01 3.61281723e-01 -3.20965320e-01 8.84846985e-01
-6.06348932e-01 4.21271741e-01 -2.36328393e-01 -5.15520036e-01
4.65906918e-01 -3.28591764e-01 5.30371845e-01 7.77829945e-01
-3.20932604e-02 1.11449994e-02 6.14354789e-01 1.11035824e+00
-2.37040222e-02 -1.99092273e-02 -6.78923309e-01 -4.22312021e-01
9.72856820e-01 1.10003614e+00 -5.14645934e-01 -2.35688165e-02
3.50876674e-02 8.64792764e-01 8.95519495e-01 4.19711828e-01
-9.16072428e-01 -4.92723137e-01 3.69809777e-01 -4.03890222e-01
3.32690746e-01 -4.54390854e-01 1.07564842e-02 -1.06617582e+00
1.39982179e-01 -9.77955759e-01 2.01349765e-01 -2.65827596e-01
-9.21893120e-01 3.03599536e-01 1.45083098e-02 -1.21578979e+00
-4.83630419e-01 -3.96234453e-01 -3.62874776e-01 2.21703023e-01
-1.83496344e+00 -5.56708038e-01 -4.11858149e-02 3.73073816e-01
4.19206679e-01 8.25251788e-02 7.92951047e-01 -8.47712383e-02
-7.12978601e-01 1.03883433e+00 4.68778163e-01 -3.37301195e-01
3.26028049e-01 -1.54509866e+00 1.78254381e-01 4.93661284e-01
-8.11598971e-02 7.15260208e-01 6.23081207e-01 -2.90810525e-01
-2.11396527e+00 -7.23992705e-01 1.25748321e-01 -1.99351653e-01
7.99528480e-01 -4.26472634e-01 -2.81687260e-01 7.68465638e-01
-4.67560478e-02 -2.34129727e-02 6.93494260e-01 4.00125802e-01
-2.00544447e-01 -7.46524185e-02 -1.17381358e+00 1.04502010e+00
1.19987214e+00 -5.66112995e-01 -2.89087176e-01 3.47995222e-01
8.56010199e-01 -8.54138374e-01 -7.12502062e-01 -3.09864171e-02
7.05577374e-01 -9.52222705e-01 8.04475784e-01 -6.10060513e-01
1.44034505e-01 1.07894264e-01 3.97876725e-02 -1.38116252e+00
-6.70083836e-02 -1.50852919e+00 -6.67860389e-01 5.60013473e-01
3.24472666e-01 -9.36149836e-01 6.26583338e-01 5.00071406e-01
8.49390998e-02 -1.18639863e+00 -9.37779725e-01 -1.44789624e+00
-5.13154529e-02 -3.38124782e-01 6.17747724e-01 5.64135253e-01
5.32470047e-01 -2.41463497e-01 -4.35323894e-01 3.31363194e-02
8.16668570e-01 5.00642896e-01 7.44564950e-01 -6.35263503e-01
-9.41274703e-01 -3.99015516e-01 2.54615217e-01 -1.61179233e+00
3.44622016e-01 -5.42185009e-01 2.51343131e-01 -1.16796958e+00
-1.01012560e-02 -5.77034593e-01 -6.96000159e-01 6.47619367e-01
-7.87222013e-03 -3.48423064e-01 1.42171904e-01 2.19541982e-01
-1.10381496e+00 9.95472968e-01 1.21810722e+00 3.17733973e-01
-6.32525206e-01 -1.29487365e-01 -7.22383082e-01 5.73437631e-01
8.78435969e-01 -3.05231482e-01 -5.55785894e-01 -3.12514782e-01
3.22634459e-01 4.87635106e-01 1.34303540e-01 -8.56094718e-01
1.68182060e-01 -6.77257478e-01 1.26089737e-01 -3.50094855e-01
3.49997580e-01 -6.58712804e-01 -4.75194991e-01 5.19066870e-01
-8.80609155e-01 1.29359856e-01 1.61764577e-01 9.21800494e-01
2.30938613e-01 -3.40993166e-01 6.26122355e-01 -2.48372987e-01
-5.00290453e-01 2.18403488e-01 -3.87849122e-01 3.16032946e-01
9.57149327e-01 -2.98818737e-01 -2.21807331e-01 -5.95139921e-01
-7.55189121e-01 8.43128920e-01 3.01984578e-01 3.32690150e-01
6.30246997e-01 -1.28822768e+00 -3.43770310e-02 7.81695768e-02
-2.81858444e-02 -2.98196107e-01 -3.97015810e-02 8.97860825e-01
-1.97934538e-01 3.28163356e-01 -1.68732405e-01 -2.00385258e-01
-5.90885639e-01 6.60923839e-01 3.07337314e-01 -5.72901130e-01
-6.48260176e-01 6.92436576e-01 -1.87899172e-01 -5.33145249e-01
9.11473989e-01 -2.12913185e-01 3.73129435e-02 -3.94151181e-01
4.79797214e-01 6.50747061e-01 -2.47402743e-01 3.01711708e-01
-1.03974238e-01 -6.50974438e-02 -4.18807985e-03 -6.06278121e-01
1.23376656e+00 -1.90918803e-01 3.55277181e-01 7.74896681e-01
6.74909055e-01 -1.92583427e-01 -1.80724847e+00 -3.56936187e-01
1.67553186e-01 -8.23669076e-01 3.04993212e-01 -9.31992650e-01
-7.42428184e-01 4.96644109e-01 7.38761723e-01 1.64310306e-01
1.04267061e+00 -3.42886090e-01 6.45502567e-01 1.05188763e+00
9.79099691e-01 -1.93228567e+00 4.67251480e-01 5.78409314e-01
8.14617276e-01 -1.33861995e+00 -5.91795221e-02 2.18331829e-01
-7.99010992e-01 8.93171191e-01 8.74448419e-01 -3.72959226e-01
3.64459068e-01 2.04685524e-01 -3.98590505e-01 1.38067782e-01
-1.12192476e+00 -2.57679105e-01 -4.10655469e-01 4.48277444e-01
-1.05790950e-01 2.03204989e-01 -1.15696296e-01 7.54203260e-01
-1.70742765e-01 6.49842918e-02 3.99468511e-01 1.46926606e+00
-5.19283533e-01 -1.05264533e+00 -9.30015370e-02 1.80648789e-01
-2.68807918e-01 2.01326415e-01 -3.26665670e-01 6.65797830e-01
-5.73085546e-01 1.12209511e+00 -2.52983093e-01 -3.87899935e-01
1.82408825e-01 -1.99774116e-01 8.82132232e-01 -3.54304671e-01
-6.85092568e-01 1.38208777e-01 7.79020265e-02 -1.27306199e+00
-1.39487693e-02 -5.67478240e-01 -1.32578707e+00 -1.06425330e-01
-5.16567767e-01 2.24758536e-01 6.52090967e-01 9.03780162e-01
5.51229656e-01 3.14296424e-01 1.35982704e+00 -9.69297111e-01
-1.54860175e+00 -5.99288404e-01 -6.54243648e-01 -1.65715441e-01
5.80483615e-01 -9.66647029e-01 -5.99411011e-01 -6.13007903e-01] | [4.0844221115112305, 2.1860949993133545] |
fedd411a-5cf0-4d3b-8ad6-94f6d813f308 | catchbackdoor-backdoor-testing-by-critical | 2112.13064 | null | https://arxiv.org/abs/2112.13064v2 | https://arxiv.org/pdf/2112.13064v2.pdf | CatchBackdoor: Backdoor Testing by Critical Trojan Neural Path Identification via Differential Fuzzing | The success of deep neural networks (DNNs) in real-world applications has benefited from abundant pre-trained models. However, the backdoored pre-trained models can pose a significant trojan threat to the deployment of downstream DNNs. Existing DNN testing methods are mainly designed to find incorrect corner case behaviors in adversarial settings but fail to discover the backdoors crafted by strong trojan attacks. Observing the trojan network behaviors shows that they are not just reflected by a single compromised neuron as proposed by previous work but attributed to the critical neural paths in the activation intensity and frequency of multiple neurons. This work formulates the DNN backdoor testing and proposes the CatchBackdoor framework. Via differential fuzzing of critical neurons from a small number of benign examples, we identify the trojan paths and particularly the critical ones, and generate backdoor testing examples by simulating the critical neurons in the identified paths. Extensive experiments demonstrate the superiority of CatchBackdoor, with higher detection performance than existing methods. CatchBackdoor works better on detecting backdoors by stealthy blending and adaptive attacks, which existing methods fail to detect. Moreover, our experiments show that CatchBackdoor may reveal the potential backdoors of models in Model Zoo. | ['Zhaoyan Ming', 'Yue Yu', 'Ting Wang', 'Chong Fu', 'Yao Cheng', 'Jinyin Chen', 'Ruoxi Chen', 'Haibo Jin'] | 2021-12-24 | null | null | null | null | ['dnn-testing'] | ['adversarial'] | [ 2.27153718e-01 -1.50634870e-01 -1.62724108e-01 1.15833774e-01
-2.95748353e-01 -1.19503701e+00 5.46826422e-01 -3.18315506e-01
-8.44079554e-02 6.69160485e-01 -5.42420864e-01 -9.20111179e-01
-2.27883365e-02 -7.15599418e-01 -1.36861110e+00 -8.01655591e-01
-1.90539181e-01 -1.64491162e-01 5.86486816e-01 -3.87716740e-01
1.71419531e-01 5.52434564e-01 -1.20626295e+00 4.18648690e-01
5.03824234e-01 6.54678166e-01 -1.95975661e-01 7.24018514e-01
1.58028752e-01 6.53889000e-01 -1.24393499e+00 -7.03252614e-01
6.26683712e-01 -3.92721027e-01 -1.33740798e-01 -7.72584558e-01
6.80200696e-01 -6.81490362e-01 -5.08196712e-01 1.59565687e+00
1.98060334e-01 -4.99466032e-01 5.61117865e-02 -1.75425196e+00
-2.99609452e-01 1.12749386e+00 -4.76258934e-01 5.40045261e-01
-1.43328652e-01 5.41797221e-01 7.41559327e-01 -3.95668656e-01
4.11313772e-01 1.31485009e+00 7.80540049e-01 9.45867836e-01
-9.79453802e-01 -1.52447701e+00 1.10668615e-01 -9.14539322e-02
-1.20400822e+00 -3.74191850e-01 7.26031840e-01 -2.07219437e-01
7.55644560e-01 3.64479005e-01 3.93931240e-01 1.94725084e+00
5.74085534e-01 4.89825219e-01 5.69015622e-01 -1.45959705e-01
1.65121332e-01 -1.24676933e-03 2.62773097e-01 7.05283463e-01
7.33914018e-01 8.29036236e-01 -2.14091137e-01 -5.11599481e-01
7.39718437e-01 2.22515389e-01 -2.88842887e-01 2.94615984e-01
-8.06600511e-01 8.27864587e-01 4.52953130e-01 1.40404209e-01
7.01883435e-02 6.33299232e-01 5.00016868e-01 3.51173192e-01
-2.78909087e-01 6.41992927e-01 -6.87468886e-01 7.00371116e-02
-6.31999791e-01 2.68738985e-01 5.96958280e-01 7.28745997e-01
5.60216308e-01 4.54774976e-01 1.57680027e-02 4.28598784e-02
2.48095393e-01 4.01724845e-01 4.07543570e-01 -4.09742951e-01
6.66854501e-01 6.72858119e-01 -5.59397221e-01 -8.80144179e-01
-1.83039382e-02 -7.27245152e-01 -5.91646850e-01 3.22562754e-01
4.53039706e-01 -5.73466241e-01 -1.05006003e+00 1.90053380e+00
2.08153054e-01 8.29257250e-01 8.62278193e-02 4.88913387e-01
2.76788145e-01 4.74317253e-01 -2.14978307e-01 3.65884274e-01
1.12883592e+00 -7.34859586e-01 -3.81005198e-01 -3.46146584e-01
8.15027118e-01 -2.40291461e-01 8.43437672e-01 3.95463169e-01
-4.45078880e-01 -1.36959776e-01 -1.43464959e+00 7.24070609e-01
-4.80801791e-01 -2.24809453e-01 5.17094016e-01 1.42310452e+00
-8.00049543e-01 6.97615206e-01 -7.57587969e-01 4.39154774e-01
6.65059984e-01 4.72291410e-01 -1.52221471e-01 2.55783787e-03
-1.39884412e+00 4.19250846e-01 5.28423131e-01 2.07164630e-01
-1.86637092e+00 -9.13118958e-01 -5.84190726e-01 -1.10903673e-01
4.79219198e-01 -1.89447463e-01 9.58587468e-01 -9.32029188e-01
-9.46363747e-01 3.39907378e-01 3.96351486e-01 -8.24506521e-01
7.79178441e-01 -9.85641330e-02 -6.95385695e-01 1.68212950e-01
-1.55229464e-01 2.22320452e-01 1.04139531e+00 -1.26304114e+00
-4.00785208e-01 -2.13299826e-01 2.49269202e-01 -7.11885333e-01
-6.52133703e-01 1.31131217e-01 4.32155952e-02 -7.80747712e-01
-3.37556928e-01 -8.40340734e-01 -2.16259822e-01 -3.08133602e-01
-1.12837803e+00 3.11271667e-01 1.26274157e+00 -1.02790251e-01
1.15507042e+00 -2.23327112e+00 -7.71285295e-01 5.03688097e-01
5.25002301e-01 8.36704135e-01 -2.94543087e-01 1.72807544e-01
-1.99375197e-01 6.33407414e-01 1.75858885e-02 1.16355494e-01
-1.70676783e-01 2.26911813e-01 -1.04134548e+00 5.28453529e-01
2.19806954e-01 9.03999984e-01 -7.87071347e-01 6.89268336e-02
-2.55295724e-01 3.26327205e-01 -4.53510672e-01 -9.01984051e-02
-3.39397401e-01 9.93785262e-03 -4.65919197e-01 9.14112985e-01
7.74136662e-01 1.12521142e-01 1.12606073e-02 1.88885048e-01
2.82040656e-01 1.91073626e-01 -7.37413764e-01 5.39053500e-01
2.23885655e-01 7.75941968e-01 -2.42928371e-01 -6.64464235e-01
8.87584865e-01 6.39262944e-02 -2.18737587e-01 -4.09981430e-01
4.12226081e-01 3.53227347e-01 4.82769758e-01 -1.81098536e-01
7.26148263e-02 2.55170148e-02 -9.10570472e-02 3.89917314e-01
4.59520966e-02 6.85813487e-01 -1.39227062e-01 2.14167446e-01
1.72075272e+00 -5.41891754e-01 -2.88807213e-01 -2.42119245e-02
1.52004942e-01 -6.03746511e-02 7.63542712e-01 1.27064562e+00
-3.45107764e-01 1.58610240e-01 1.01790726e+00 -3.57138604e-01
-8.26758504e-01 -1.10482085e+00 -1.08577073e-01 6.98525846e-01
1.56772703e-01 -3.55891854e-01 -1.07896316e+00 -1.15151012e+00
-2.71018911e-02 6.33515000e-01 -7.56327212e-01 -9.45528090e-01
-4.64859724e-01 -5.44350445e-01 1.87407053e+00 5.50432920e-01
6.04477167e-01 -8.80258858e-01 -7.42208660e-01 -5.26951216e-02
4.29251939e-01 -1.32340705e+00 -3.51178169e-01 4.61314470e-01
-9.84290957e-01 -1.42338955e+00 -3.12186688e-01 -5.47183037e-01
7.31676042e-01 1.44957855e-01 6.38847053e-01 6.03431106e-01
-3.27217579e-01 -4.42877650e-01 -2.16952160e-01 -5.59451163e-01
-8.01960766e-01 -2.27283716e-01 3.78587186e-01 -1.68441966e-01
2.55723715e-01 -6.38349950e-01 -1.87523842e-01 5.50138593e-01
-1.08669674e+00 -5.79116881e-01 7.70772755e-01 6.29102290e-01
3.44651908e-01 5.60850024e-01 4.32701916e-01 -1.07491171e+00
6.59858584e-01 -6.56714439e-01 -9.23762083e-01 2.53993660e-01
-4.87576723e-01 1.80741727e-01 1.14065659e+00 -1.12291408e+00
-3.43957782e-01 -2.58472562e-01 -2.27693349e-01 -1.17910397e+00
-3.23193640e-01 1.82836369e-01 -5.05321801e-01 -4.63513553e-01
9.69322562e-01 1.06241949e-01 -1.72110409e-01 -2.98048556e-01
-1.43874988e-01 -1.67854056e-01 5.73188066e-01 -5.04381537e-01
1.30133569e+00 3.62891108e-01 1.19882762e-01 -5.77867687e-01
-2.46938944e-01 3.54563653e-01 1.22423090e-01 -2.17870697e-01
2.77859271e-01 -4.82360363e-01 -7.85785973e-01 6.54344857e-01
-1.21387839e+00 -3.97881538e-01 1.90549344e-01 3.77668925e-02
3.83340955e-01 3.62366199e-01 -5.66694319e-01 -8.68920922e-01
-2.02540457e-01 -1.48967063e+00 5.13451040e-01 1.71509832e-01
9.22175273e-02 -8.41100216e-01 -2.88314298e-02 -5.25225773e-02
8.27829316e-02 6.25073254e-01 1.25349236e+00 -1.34303951e+00
-6.59973979e-01 -6.82073355e-01 2.12536842e-01 4.13802683e-01
-3.30765456e-01 5.48803449e-01 -1.26147223e+00 -1.29888922e-01
-1.05540551e-01 -7.27635771e-02 8.27499926e-01 2.13033736e-01
1.14819229e+00 -6.80228770e-01 -6.44906461e-01 9.26671445e-01
1.40467942e+00 5.06686628e-01 7.40979910e-01 4.04999942e-01
8.15526962e-01 2.55574912e-01 -9.93241370e-03 1.43546239e-01
-3.58355641e-01 2.69101053e-01 1.39000154e+00 5.89398332e-02
4.00964141e-01 -6.91546798e-01 9.47971463e-01 -1.63585022e-01
4.33851242e-01 -4.51965213e-01 -1.00437713e+00 2.82333016e-01
-1.41505706e+00 -8.35433662e-01 -2.78758168e-01 2.10039210e+00
5.62031686e-01 8.49872231e-01 2.57611368e-02 5.30531704e-01
9.57666516e-01 -1.60832837e-01 -7.56151497e-01 -5.00038803e-01
-3.07380289e-01 3.47440809e-01 8.15043569e-01 1.98724210e-01
-8.42974782e-01 8.78375471e-01 6.19046688e+00 9.01431084e-01
-1.04168379e+00 2.48546302e-02 6.46849930e-01 -2.18255013e-01
-5.54706156e-01 1.32071704e-01 -1.30899179e+00 5.66880882e-01
1.07822478e+00 3.01075667e-01 2.51558751e-01 9.87056553e-01
-1.37625724e-01 3.29489768e-01 -1.22975492e+00 5.62016845e-01
-8.43132883e-02 -1.53621173e+00 1.84079871e-01 5.50873876e-01
6.58887923e-01 7.72608742e-02 3.50762218e-01 3.58440757e-01
4.24370557e-01 -1.12601650e+00 1.05236554e+00 4.27106805e-02
4.93345916e-01 -1.11116004e+00 8.34323049e-01 5.57258070e-01
-7.14495122e-01 -4.33611542e-01 -3.61651182e-01 1.67894155e-01
-3.41434270e-01 6.04003906e-01 -1.00968373e+00 2.43638996e-02
6.68874621e-01 4.51155268e-02 -7.49230146e-01 8.53587568e-01
-5.12821138e-01 1.03616250e+00 -4.53233719e-01 -3.35816681e-01
4.61534649e-01 2.86715835e-01 8.14371288e-01 8.56092393e-01
2.59600192e-01 -6.65698767e-01 -4.99323159e-01 1.36451459e+00
-4.39639747e-01 -6.58100307e-01 -1.03160667e+00 -3.83050829e-01
8.11492801e-01 1.07684243e+00 -7.88654029e-01 3.68925408e-02
1.86881304e-01 5.47696948e-01 7.35134725e-03 3.24860692e-01
-1.23019481e+00 -5.61494231e-01 1.13539600e+00 8.74141902e-02
3.02104801e-01 1.01984870e-02 -2.53809780e-01 -8.12285066e-01
1.97151795e-01 -1.14364481e+00 2.29328051e-01 -3.02559763e-01
-1.03915966e+00 7.63612390e-01 -1.18143819e-01 -1.13614213e+00
-1.25656247e-01 -9.27292049e-01 -1.24206090e+00 5.72939277e-01
-1.14775836e+00 -8.56833875e-01 1.22101769e-01 9.13038671e-01
7.46801198e-02 -5.77154756e-01 5.08027792e-01 1.82318270e-01
-1.18235087e+00 1.40593135e+00 -7.10204914e-02 6.62511349e-01
2.30323941e-01 -7.32746422e-01 8.09621394e-01 1.64912426e+00
2.26306543e-01 1.01091218e+00 6.99104190e-01 -9.25718904e-01
-1.46718609e+00 -1.26341927e+00 2.73480207e-01 -4.62703824e-01
9.32612360e-01 -7.94619083e-01 -7.99894929e-01 7.45372951e-01
-2.50239164e-01 1.67126343e-01 4.83330220e-01 -4.02365088e-01
-8.14244747e-01 -3.63527238e-02 -1.29045153e+00 9.20491815e-01
8.24108303e-01 -3.82721901e-01 -1.45734131e-01 8.22345167e-02
9.88062739e-01 -2.49185517e-01 -1.54156042e-02 2.70148158e-01
4.59947824e-01 -9.37094986e-01 9.93548810e-01 -9.38563943e-01
6.40975416e-01 -1.28741130e-01 -2.34376825e-02 -8.97698522e-01
3.94368559e-01 -9.58975077e-01 -5.15142083e-01 1.33158946e+00
6.53204501e-01 -9.83662546e-01 9.80602443e-01 1.30768880e-01
-1.29515424e-01 -6.03806973e-01 -1.11890602e+00 -1.19531965e+00
1.93744257e-01 -7.76902020e-01 8.67725909e-01 9.98226523e-01
-4.92358208e-01 -6.28631562e-02 -2.81724334e-01 6.89897299e-01
8.49824488e-01 -6.55663550e-01 6.72807038e-01 -1.01943290e+00
-4.19387221e-01 -6.32962763e-01 -5.45975804e-01 -4.94550467e-01
2.88112253e-01 -7.42804825e-01 -1.45774633e-02 -3.38979751e-01
-2.00450957e-01 -4.26449478e-01 -3.95497590e-01 8.19856942e-01
-1.13735020e-01 3.82672042e-01 -1.31154619e-02 6.30477145e-02
-2.78326292e-02 -1.23751694e-02 8.05940866e-01 -2.15142280e-01
-4.65566367e-02 2.33752012e-01 -8.48852754e-01 7.63952076e-01
8.82151186e-01 -1.00227654e+00 -5.57339489e-01 -3.65557641e-01
5.68227112e-01 -2.98920095e-01 9.72689867e-01 -1.20660937e+00
8.20557252e-02 1.69950705e-02 2.41008967e-01 -3.92522395e-01
-5.17377444e-02 -8.63336563e-01 -3.40747274e-02 8.90610397e-01
-1.27635047e-01 3.88423324e-01 5.90916574e-01 7.69170046e-01
2.40648091e-01 -5.45424044e-01 6.64776504e-01 -6.26568682e-04
-5.78248143e-01 3.83774549e-01 -5.04405379e-01 8.10468011e-03
1.21347761e+00 -5.10476172e-01 -7.03948855e-01 8.87244269e-02
-8.35627541e-02 8.15666616e-02 2.16968670e-01 3.22158337e-01
8.91422629e-01 -1.05348504e+00 -2.92149097e-01 6.63281024e-01
-6.43940642e-02 -2.38323539e-01 1.02746077e-01 4.86359537e-01
-5.96546471e-01 1.57829195e-01 -2.47196838e-01 -3.72420669e-01
-1.28708208e+00 6.17770076e-01 6.77226484e-01 -1.99076071e-01
-5.95004857e-01 1.15383041e+00 9.06776041e-02 -3.02529126e-01
3.64845067e-01 -3.30087572e-01 1.71176478e-01 -3.59148294e-01
6.76917911e-01 2.17044473e-01 1.66639447e-01 -1.64879948e-01
-6.54961467e-01 -1.51792644e-02 -2.78873593e-01 7.11780787e-02
1.02525198e+00 6.85199082e-01 1.16029643e-01 -3.80829684e-02
1.24287009e+00 8.28036442e-02 -1.21279573e+00 1.75052658e-01
-9.58185717e-02 -3.16966683e-01 -4.99523990e-02 -8.01532805e-01
-1.39268899e+00 9.72257137e-01 4.72764760e-01 1.27009496e-01
9.07383323e-01 -1.67801902e-01 1.02902544e+00 5.89695930e-01
2.57304698e-01 -4.44277316e-01 2.68835157e-01 5.35004973e-01
2.56108165e-01 -5.67139030e-01 -6.24276221e-01 -5.51361702e-02
-2.39974886e-01 1.13319552e+00 9.72811580e-01 -3.03411514e-01
5.67896783e-01 9.02042508e-01 1.05441928e-01 -1.48504212e-01
-8.92862737e-01 4.93365258e-01 -3.00830662e-01 6.36742592e-01
-4.45040077e-01 -1.27714127e-01 4.42621768e-01 8.99622977e-01
-3.44614238e-01 -4.91262019e-01 6.02690458e-01 9.03488278e-01
-2.60247797e-01 -9.67478752e-01 -6.27566814e-01 4.09545898e-01
-6.96005523e-01 -3.01430494e-01 -8.24619770e-01 8.61795127e-01
2.95094281e-01 8.28355670e-01 -3.22704911e-02 -1.35053742e+00
2.04886168e-01 -3.11993938e-02 1.25687003e-01 -3.35396975e-01
-1.09634137e+00 -2.05681652e-01 -2.02969536e-01 -4.90519255e-01
6.34375930e-01 -8.91520083e-02 -1.08670163e+00 -5.58522284e-01
-6.45060837e-01 -1.20773897e-01 6.11668825e-01 9.90501463e-01
1.81416169e-01 7.51293004e-01 7.86867619e-01 -4.02857542e-01
-9.03864086e-01 -6.32515728e-01 -5.52210569e-01 3.55128087e-02
7.13469505e-01 -6.70739412e-01 -8.05603743e-01 -5.08663416e-01] | [5.709142208099365, 7.698341369628906] |
5e9f1cb8-47a0-4bb2-abf1-4ca4d7d9825b | automated-graph-generation-at-sentence-level | null | null | https://aclanthology.org/2020.coling-main.240 | https://aclanthology.org/2020.coling-main.240.pdf | Automated Graph Generation at Sentence Level for Reading Comprehension Based on Conceptual Graphs | This paper proposes a novel miscellaneous-context-based method to convert a sentence into a knowledge embedding in the form of a directed graph. We adopt the idea of conceptual graphs to frame for the miscellaneous textual information into conceptual compactness. We first empirically observe that this graph representation method can (1) accommodate the slot-filling challenges in typical question answering and (2) access to the sentence-level graph structure in order to explicitly capture the neighbouring connections of reference concept nodes. Secondly, we propose a task-agnostic semantics-measured module, which cooperates with the graph representation method, in order to (3) project an edge of a sentence-level graph to the space of semantic relevance with respect to the corresponding concept nodes. As a result of experiments on the QA-type relation extraction, the combination of the graph representation and the semantics-measured module achieves the high accuracy of answer prediction and offers human-comprehensible graphical interpretation for every well-formed sample. To our knowledge, our approach is the first towards the interpretable process of learning vocabulary representations with the experimental evidence. | ['Chun-Shien Lu', 'Wan-Hsuan Lin'] | 2020-12-01 | null | null | null | coling-2020-8 | ['miscellaneous'] | ['miscellaneous'] | [ 3.19360226e-01 9.57990110e-01 -9.57575887e-02 -3.87437463e-01
-4.36165154e-01 -4.71153170e-01 8.40190232e-01 8.23964596e-01
-1.77026585e-01 4.55487013e-01 4.09945577e-01 -8.97711635e-01
-6.51884973e-01 -1.21297419e+00 -4.50442553e-01 -2.39063144e-01
-1.83997266e-02 7.00586855e-01 3.81222665e-01 -5.76007783e-01
3.65529150e-01 2.70573288e-01 -1.49284589e+00 4.35251683e-01
1.01570237e+00 1.08449042e+00 2.43710101e-01 3.65345269e-01
-7.51415014e-01 9.42140460e-01 -3.91649485e-01 -6.61153078e-01
-2.93725699e-01 -4.76901740e-01 -1.37460697e+00 4.05590087e-02
2.39661232e-01 3.65266114e-01 -9.88199040e-02 9.14963186e-01
-1.18323145e-02 1.20910332e-01 7.39260912e-01 -9.90908265e-01
-6.93169713e-01 7.89572239e-01 -1.65607125e-01 1.23152830e-01
7.61529326e-01 -3.73906553e-01 1.55695999e+00 -7.06970096e-01
7.97732234e-01 1.52239442e+00 2.57966429e-01 2.92875826e-01
-1.12714398e+00 2.46079341e-01 2.77535915e-01 5.16525269e-01
-1.02129602e+00 5.18107824e-02 1.02401650e+00 -3.79779935e-01
1.00271475e+00 5.56263208e-01 6.05808258e-01 7.06070423e-01
3.13695073e-02 5.74831843e-01 8.72929037e-01 -8.47387910e-01
3.84758502e-01 2.10435361e-01 7.78917670e-01 8.39597106e-01
2.86130905e-01 -3.77012283e-01 -3.79726112e-01 -1.60620645e-01
3.99666935e-01 -1.39611423e-01 -4.75615978e-01 -6.98455930e-01
-8.60488057e-01 1.08724797e+00 6.33957982e-01 6.24425650e-01
-2.76937455e-01 3.49144302e-02 5.03434539e-01 3.34028214e-01
4.33591574e-01 6.95222676e-01 -5.00449598e-01 2.20023379e-01
-5.71519256e-01 1.27399027e-01 1.07637179e+00 8.99734974e-01
8.32803488e-01 -3.68467808e-01 -3.22856545e-01 5.39121807e-01
7.06757069e-01 -8.14656094e-02 4.65916574e-01 -5.85467398e-01
6.23947084e-01 1.19425356e+00 -3.24527659e-02 -1.33977199e+00
-4.25912350e-01 -5.60760200e-01 -4.50774491e-01 -1.28094763e-01
3.87178302e-01 3.53516847e-01 -5.03613174e-01 1.63165116e+00
4.53798443e-01 -2.16543302e-01 2.71486193e-01 8.07712734e-01
1.06014180e+00 4.90019232e-01 2.86508888e-01 -1.48248777e-01
1.95365059e+00 -8.37010384e-01 -8.69818330e-01 -2.13184983e-01
9.47579265e-01 -3.14164430e-01 1.30247581e+00 -6.58673048e-02
-8.56919706e-01 -3.91204894e-01 -1.18302238e+00 -3.29283148e-01
-5.57167768e-01 -8.49584043e-02 6.37802482e-01 5.04759848e-01
-1.01447976e+00 5.56417525e-01 -3.90411437e-01 -5.64141750e-01
2.04377994e-01 -1.34216705e-02 -4.48987186e-01 -7.34844580e-02
-1.52920079e+00 1.21860921e+00 7.75944531e-01 1.15173697e-01
-1.39308676e-01 -4.90096599e-01 -1.12653208e+00 3.60724062e-01
7.80776680e-01 -1.11902702e+00 8.00461829e-01 -8.43331397e-01
-1.11772871e+00 8.93721640e-01 -2.96932489e-01 -5.81120729e-01
1.78791672e-01 2.82504819e-02 -2.72223085e-01 4.29935247e-01
-6.72553573e-03 3.49950552e-01 7.11533189e-01 -1.33541298e+00
-2.68695712e-01 -5.74621320e-01 6.29195094e-01 2.66804367e-01
-4.73418295e-01 -1.97896332e-01 -2.48671055e-01 -5.31891525e-01
3.52878779e-01 -3.76367837e-01 -8.26853216e-02 -9.90452543e-02
-3.86654615e-01 -6.78638816e-01 6.17997110e-01 -6.96856558e-01
1.52523100e+00 -1.73971593e+00 5.02255857e-01 2.76842654e-01
6.56810522e-01 3.58664207e-02 8.67726281e-02 8.75762284e-01
-2.78861880e-01 1.60068363e-01 -2.85099149e-01 -1.77912474e-01
2.17690602e-01 4.29042786e-01 -3.93132985e-01 8.84496495e-02
3.20963353e-01 1.23153651e+00 -1.15730584e+00 -6.03296936e-01
1.36289448e-01 9.22912881e-02 -5.56418717e-01 1.70650750e-01
-4.80715424e-01 -7.20166937e-02 -6.62550986e-01 2.48615950e-01
3.41958493e-01 -3.22350413e-01 6.81926906e-01 -4.05685693e-01
3.45553100e-01 6.87125862e-01 -9.93517280e-01 1.69611335e+00
-4.91751283e-01 4.81416643e-01 -2.13393226e-01 -1.13200176e+00
1.19302547e+00 1.80824786e-01 -1.18538387e-01 -6.90529406e-01
-1.24108814e-01 1.43439278e-01 -3.87835130e-02 -5.47982991e-01
7.64861405e-01 -2.65803188e-01 -1.08094625e-01 3.49350035e-01
2.92114198e-01 -1.81008279e-01 1.58454075e-01 6.72484279e-01
9.76947367e-01 1.48105875e-01 6.34805024e-01 -5.66120684e-01
9.01409864e-01 -2.17929572e-01 -5.35116754e-02 5.19795358e-01
3.49731654e-01 1.57279417e-01 1.03093135e+00 -4.69429731e-01
-9.27348733e-01 -8.74396265e-01 -6.70460844e-03 8.39607120e-01
6.29994497e-02 -9.64105964e-01 -7.93317199e-01 -8.51538181e-01
-7.06465319e-02 1.15988147e+00 -7.57240176e-01 -2.42228687e-01
-3.74462098e-01 -3.88215855e-02 2.08702594e-01 5.21078825e-01
4.36253585e-02 -1.03099906e+00 -5.42792618e-01 1.22669794e-01
-2.73462445e-01 -9.07169580e-01 -3.98982763e-02 2.80746549e-01
-9.04083788e-01 -1.18096757e+00 -5.17817885e-02 -7.55917668e-01
7.26792634e-01 -4.42176126e-02 1.28567779e+00 5.97493827e-01
2.75286794e-01 5.11106849e-01 -7.28365064e-01 -1.37818396e-01
-5.08008540e-01 -6.45492524e-02 -4.41945910e-01 3.89255062e-02
4.45997894e-01 -5.98346949e-01 -3.48473072e-01 -2.44093463e-02
-1.08547306e+00 1.76957548e-01 1.05671257e-01 8.99000168e-01
4.06169742e-01 -1.01545893e-01 6.11230552e-01 -1.03583229e+00
1.04685307e+00 -4.84044492e-01 -3.59970778e-01 6.85995281e-01
-7.49855220e-01 6.98851705e-01 5.75075984e-01 5.04829846e-02
-9.25553739e-01 -2.43906096e-01 -1.14939071e-01 3.28085832e-02
5.24068400e-02 1.00411499e+00 -3.65040869e-01 3.22899193e-01
7.87641287e-01 1.24218278e-01 3.07922661e-02 -4.93908525e-01
9.34030056e-01 4.79884058e-01 3.34921032e-01 -7.94575870e-01
7.76647747e-01 1.71623856e-01 3.61162841e-01 -6.46929026e-01
-9.17248070e-01 -6.53945267e-01 -8.90237927e-01 2.97739729e-02
8.78314912e-01 -4.44539934e-01 -6.64794564e-01 -2.72193313e-01
-1.36746156e+00 9.81431976e-02 -5.61291337e-01 1.92543603e-02
-5.83553016e-01 8.96774650e-01 -3.96140844e-01 -8.67177963e-01
-3.65331650e-01 -7.80276954e-01 1.12500012e+00 1.00530528e-01
-4.51809376e-01 -1.41940427e+00 -1.04914695e-01 5.58930814e-01
8.06109160e-02 1.46878362e-01 1.50128174e+00 -9.52113271e-01
-4.35543716e-01 -3.38296480e-02 -4.20279443e-01 1.02560088e-01
-5.44702336e-02 -3.75638962e-01 -8.89328599e-01 1.59386322e-02
1.58746868e-01 -1.75121039e-01 7.77556419e-01 -1.36444524e-01
1.07178807e+00 -4.75133926e-01 -1.42366603e-01 -9.09156539e-03
1.39185619e+00 -1.61410004e-01 8.94093513e-01 3.00456703e-01
5.76890111e-01 8.19124341e-01 5.88017404e-01 1.80547893e-01
6.43399358e-01 8.93778324e-01 5.42528391e-01 1.10858835e-01
-8.71134177e-02 -6.84650898e-01 -9.68792811e-02 9.04964566e-01
-5.13129756e-02 -3.04875433e-01 -8.49106610e-01 3.85496616e-01
-2.01299000e+00 -9.52257097e-01 -4.81618404e-01 1.94621158e+00
7.38096476e-01 2.07179934e-01 -1.73535615e-01 3.14273089e-01
4.91791546e-01 2.50964791e-01 1.62111312e-01 -6.27648175e-01
2.25835547e-01 2.76207566e-01 -1.54873286e-03 8.55349004e-01
-7.55580604e-01 8.61056507e-01 5.12778521e+00 9.96598184e-01
-3.89901072e-01 2.04521231e-03 2.22567230e-01 7.75237441e-01
-1.03235722e+00 3.76231194e-01 -3.93847048e-01 1.27114937e-01
8.50377798e-01 -2.34054461e-01 2.01320633e-01 7.11269081e-01
-1.33946732e-01 -6.89612031e-02 -1.28737593e+00 6.80407465e-01
1.44520178e-01 -1.54583907e+00 5.49650848e-01 -8.55929479e-02
2.39497181e-02 -7.82312751e-01 -4.68947768e-01 2.75643736e-01
-1.84994102e-01 -1.06658983e+00 8.98946166e-01 8.16801190e-01
5.41010976e-01 -5.74705482e-01 8.80397141e-01 4.93968248e-01
-1.36600506e+00 -1.67577699e-01 -4.65487897e-01 -3.02854449e-01
2.60928124e-01 4.74635422e-01 -1.01728094e+00 1.32389247e+00
2.29723707e-01 4.88251060e-01 -7.35625923e-01 6.23754382e-01
-6.74709558e-01 5.36911905e-01 1.20962240e-01 -4.32160884e-01
2.53117949e-01 -3.39604765e-01 5.60073674e-01 1.16618943e+00
1.63101748e-01 2.67757922e-01 -1.77461788e-01 1.07062793e+00
-8.41012411e-03 3.48857105e-01 -6.30730450e-01 -1.90881118e-01
4.13359255e-01 1.18495464e+00 -7.99924135e-01 -4.59126890e-01
-3.17555994e-01 7.96789587e-01 7.67642796e-01 2.35836625e-01
-4.31110978e-01 -4.60031480e-01 2.72624731e-01 1.07847750e-01
2.56763369e-01 -2.64495701e-01 -2.93736160e-01 -1.17081773e+00
3.13017011e-01 -6.18105412e-01 5.27657449e-01 -1.08837914e+00
-1.20084894e+00 7.01307058e-01 -1.16621966e-04 -9.22676146e-01
-3.39306951e-01 -7.59761930e-01 -5.84576666e-01 1.06715226e+00
-1.48345923e+00 -1.06357765e+00 -1.82546347e-01 3.28418344e-01
4.08341408e-01 5.04455455e-02 1.20024502e+00 -7.32861534e-02
-2.09074065e-01 2.72713929e-01 -4.05643672e-01 3.14285010e-02
5.48941344e-02 -1.59793246e+00 2.74685353e-01 6.61207199e-01
4.88233089e-01 9.30136859e-01 8.58300865e-01 -4.34796870e-01
-1.31025434e+00 -6.84079766e-01 1.49963439e+00 -8.02435637e-01
9.54826176e-01 -3.63885045e-01 -1.23941219e+00 5.41617572e-01
2.11996228e-01 -2.46574640e-01 5.91785789e-01 4.49735284e-01
-5.30634403e-01 1.00914083e-01 -8.03807139e-01 5.19671321e-01
1.06633294e+00 -8.48928094e-01 -1.64555132e+00 2.85567254e-01
1.14808452e+00 -2.56104261e-01 -8.54160011e-01 1.55673012e-01
5.66492304e-02 -7.22307742e-01 9.99597371e-01 -9.56236422e-01
5.64484239e-01 -3.52407902e-01 -1.58108130e-01 -1.26931763e+00
-2.76432335e-01 -3.67518842e-01 -3.47018868e-01 1.14933980e+00
6.09306097e-01 -5.86194932e-01 4.61352944e-01 4.47900623e-01
-1.79848403e-01 -1.03624833e+00 -1.05671084e+00 -4.50085789e-01
-9.21412408e-02 -4.61384118e-01 6.02218509e-01 8.89937341e-01
5.87367475e-01 7.76347578e-01 8.17411616e-02 -1.16465636e-01
3.66852701e-01 3.42931092e-01 3.32362294e-01 -1.58483040e+00
-4.31523353e-01 -3.63599509e-01 -6.95920169e-01 -1.16829431e+00
2.04565004e-01 -1.21548986e+00 -3.64266902e-01 -2.06623244e+00
4.69264984e-02 -2.98673898e-01 -1.57350823e-01 2.28246540e-01
-1.66280061e-01 -3.72477174e-01 2.95784920e-01 -7.52373785e-02
-6.77046180e-01 5.33367395e-01 1.32547569e+00 -1.67676628e-01
1.43464014e-01 -2.11652398e-01 -1.07793891e+00 5.97482502e-01
5.65043926e-01 -3.32650453e-01 -8.74335170e-01 -1.61872730e-01
7.31369495e-01 2.52557009e-01 5.61638892e-01 -3.13666910e-01
1.61362693e-01 -1.01423979e-01 -7.55377710e-02 -2.90647119e-01
1.11348994e-01 -9.45713401e-01 -6.33844286e-02 3.25665593e-01
-4.79543895e-01 -8.97934511e-02 -6.16073124e-02 7.07314074e-01
-3.73975664e-01 -6.20107055e-01 2.39248604e-01 -5.07926829e-02
-7.85735965e-01 -2.19130576e-01 -8.03766772e-02 2.58283854e-01
6.07773006e-01 -2.10706457e-01 -4.69399869e-01 -3.84474397e-01
-7.50103593e-01 2.55557448e-01 2.05895126e-01 3.63931984e-01
7.23475575e-01 -1.26785362e+00 -3.68744224e-01 -8.35352689e-02
3.82585049e-01 -1.32985055e-01 9.87143591e-02 5.87842882e-01
-4.33174312e-01 6.14691913e-01 -7.40856230e-02 -3.62597823e-01
-1.06070852e+00 8.53973567e-01 2.12410584e-01 -5.81106663e-01
-7.47138381e-01 6.09615028e-01 6.08140491e-02 -3.61730218e-01
2.57443730e-03 -6.13210380e-01 -7.98181057e-01 3.28119040e-01
3.73574078e-01 1.01477191e-01 2.99938768e-01 -5.34804106e-01
-3.48664552e-01 4.27255154e-01 2.04764634e-01 -4.31743711e-02
1.14620221e+00 -1.99394807e-01 -4.73330259e-01 3.68060946e-01
1.05747950e+00 -2.47646589e-02 -6.75118387e-01 -3.47945333e-01
6.33573413e-01 -2.89571494e-01 -1.45928442e-01 -6.65157139e-01
-4.38419163e-01 1.00760877e+00 -9.85234529e-02 7.20691621e-01
8.67978632e-01 5.58678925e-01 3.55285496e-01 5.72972238e-01
4.01407212e-01 -8.73099148e-01 1.87597170e-01 4.80580747e-01
1.09992111e+00 -7.82481968e-01 5.87837324e-02 -7.89334655e-01
-5.76032579e-01 1.49249375e+00 2.67041266e-01 2.73903221e-01
3.72319579e-01 -2.66024292e-01 -2.22676814e-01 -7.31562614e-01
-8.04188430e-01 -3.28308493e-01 7.12660670e-01 6.18085027e-01
5.03235579e-01 2.30948612e-01 -7.97852635e-01 7.98522115e-01
-2.65608639e-01 -3.04815680e-01 4.03172851e-01 6.65856600e-01
-6.93577230e-01 -1.25094879e+00 -1.11284435e-01 2.99160093e-01
1.51926354e-01 -2.46152833e-01 -4.56100136e-01 7.05021262e-01
-1.29504442e-01 1.02480423e+00 -8.91984999e-02 -1.88172445e-01
5.43164194e-01 4.56296891e-01 4.88605320e-01 -8.03000331e-01
-4.22251701e-01 -4.76550281e-01 5.07126272e-01 -4.28537279e-01
-3.60052109e-01 -2.95604885e-01 -1.46111405e+00 4.02970947e-02
-3.67728710e-01 4.95369166e-01 4.50963616e-01 1.32318783e+00
2.99202710e-01 5.93836427e-01 1.94750860e-01 -2.70064563e-01
-5.98503649e-01 -8.71183932e-01 -5.96491814e-01 7.52158940e-01
-2.34486461e-02 -6.65343404e-01 -3.26562911e-01 -2.66493291e-01] | [10.494146347045898, 8.013036727905273] |
312f6bfb-b3c9-44d2-bf03-a84edd647966 | cern-confidence-energy-recurrent-network-for | 1704.03058 | null | http://arxiv.org/abs/1704.03058v1 | http://arxiv.org/pdf/1704.03058v1.pdf | CERN: Confidence-Energy Recurrent Network for Group Activity Recognition | This work is about recognizing human activities occurring in videos at
distinct semantic levels, including individual actions, interactions, and group
activities. The recognition is realized using a two-level hierarchy of Long
Short-Term Memory (LSTM) networks, forming a feed-forward deep architecture,
which can be trained end-to-end. In comparison with existing architectures of
LSTMs, we make two key contributions giving the name to our approach as
Confidence-Energy Recurrent Network -- CERN. First, instead of using the common
softmax layer for prediction, we specify a novel energy layer (EL) for
estimating the energy of our predictions. Second, rather than finding the
common minimum-energy class assignment, which may be numerically unstable under
uncertainty, we specify that the EL additionally computes the p-values of the
solutions, and in this way estimates the most confident energy minimum. The
evaluation on the Collective Activity and Volleyball datasets demonstrates: (i)
advantages of our two contributions relative to the common softmax and
energy-minimization formulations and (ii) a superior performance relative to
the state-of-the-art approaches. | ['Song-Chun Zhu', 'Sinisa Todorovic', 'Tianmin Shu'] | 2017-04-10 | cern-confidence-energy-recurrent-network-for-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Shu_CERN_Confidence-Energy_Recurrent_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Shu_CERN_Confidence-Energy_Recurrent_CVPR_2017_paper.pdf | cvpr-2017-7 | ['group-activity-recognition'] | ['computer-vision'] | [ 4.36141521e-01 2.97047943e-01 -2.57694095e-01 -4.02998924e-01
-7.73169518e-01 -6.50423095e-02 6.32113159e-01 -1.93400651e-01
-5.04850149e-01 7.11628377e-01 5.08067608e-01 7.43003860e-02
-3.46949309e-01 -4.34111416e-01 -1.11422658e+00 -7.91510761e-01
-4.32354689e-01 1.97002530e-01 3.35423499e-02 2.79640228e-01
2.20540151e-01 1.96335673e-01 -1.76384687e+00 4.39039886e-01
5.53920865e-01 1.45917523e+00 -1.13106363e-01 6.76622510e-01
1.19237199e-01 1.40437341e+00 -5.04367888e-01 -3.50031614e-01
-6.00911155e-02 -5.92274070e-01 -1.10954678e+00 9.84199718e-02
1.78857982e-01 -1.29625499e-01 -2.65641481e-01 7.87447155e-01
1.89349547e-01 6.96405828e-01 5.39899826e-01 -1.04591346e+00
-4.42449033e-01 4.40784812e-01 -1.61355689e-01 5.53205013e-02
6.85863733e-01 -1.08321004e-01 9.57166255e-01 -8.26561093e-01
5.77264845e-01 9.06711400e-01 9.30407107e-01 5.58462739e-01
-8.45673025e-01 -2.74908125e-01 4.03984934e-01 4.76774186e-01
-1.42449391e+00 -5.77341139e-01 4.98864949e-01 -3.86135489e-01
1.55468369e+00 3.61061729e-02 7.39708424e-01 1.66665876e+00
2.58506507e-01 9.39002514e-01 6.44177258e-01 -5.17371178e-01
3.44146371e-01 -1.74756631e-01 -1.01529598e-01 8.83878827e-01
-2.38558009e-01 1.00965440e-01 -1.02993643e+00 -8.61593410e-02
4.31228399e-01 -3.93828936e-02 -1.16105698e-01 -2.84646988e-01
-1.39089203e+00 3.71092319e-01 2.13477150e-01 4.18930501e-01
-5.63041091e-01 7.36489594e-01 3.91686112e-01 -1.37974560e-01
6.10505104e-01 2.22567514e-01 -4.69596654e-01 -7.35911489e-01
-1.18486261e+00 6.27902597e-02 8.29755545e-01 4.34541672e-01
3.38262469e-01 -9.13311690e-02 -2.87390083e-01 6.90119028e-01
3.77127975e-01 -3.29987109e-01 5.92343390e-01 -1.00074327e+00
3.22763920e-01 3.59984726e-01 1.40345603e-01 -6.25403643e-01
-3.57084900e-01 -5.30558705e-01 -6.42595947e-01 2.26465836e-01
4.22495067e-01 -3.66839230e-01 -8.78326535e-01 1.95215702e+00
3.19300815e-02 6.69233859e-01 8.08904972e-03 7.64451087e-01
4.23427910e-01 6.18346632e-01 3.30909222e-01 -2.55626649e-01
1.13620925e+00 -1.11394918e+00 -7.05817282e-01 -3.09401721e-01
4.03587669e-01 -1.40495315e-01 5.94590664e-01 4.11626965e-01
-1.30101812e+00 -5.81919491e-01 -1.12790573e+00 -9.02672336e-02
-5.00778615e-01 1.90886408e-01 6.74134791e-01 3.39500368e-01
-1.15086627e+00 1.03504431e+00 -1.33309865e+00 -3.85435790e-01
2.17719689e-01 3.73473644e-01 -1.65041849e-01 6.55234993e-01
-1.24684763e+00 1.21879637e+00 5.05391479e-01 3.26614320e-01
-6.73376799e-01 -3.91019553e-01 -9.76438940e-01 3.17933530e-01
4.19762999e-01 -5.52440941e-01 1.34089291e+00 -1.32070458e+00
-1.81978822e+00 7.54660189e-01 -3.48445475e-01 -7.13122487e-01
5.37083447e-01 -2.80672640e-01 -3.23945373e-01 -2.70061046e-02
-3.03058803e-01 7.62769639e-01 5.25197148e-01 -7.22303629e-01
-5.43313086e-01 1.06794657e-02 -1.02956392e-01 4.76266086e-01
-3.49771589e-01 2.90047407e-01 -4.44117308e-01 -6.91702664e-01
-1.34106353e-03 -9.33771074e-01 -2.44643122e-01 -1.93013772e-01
-2.20590800e-01 -5.17178059e-01 4.46373254e-01 -8.03260982e-01
1.25530767e+00 -2.03838801e+00 4.18194830e-01 2.24304900e-01
-1.90600827e-01 -1.95828974e-01 2.25527957e-01 2.65684456e-01
-9.60215777e-02 1.03040755e-01 -1.29407600e-01 -6.28562093e-01
3.55435550e-01 9.00568217e-02 -7.80533850e-02 4.16457176e-01
2.77232111e-01 9.81237352e-01 -5.86559951e-01 -5.68878174e-01
2.87879646e-01 7.54441023e-01 -2.95575112e-01 6.67490736e-02
-2.90409118e-01 1.52921751e-01 -2.40801021e-01 5.03954351e-01
3.87908407e-02 -4.31020379e-01 4.47217047e-01 -1.15869857e-01
-9.24683958e-02 5.09949625e-01 -1.28915942e+00 1.99821353e+00
-2.35257760e-01 6.38339341e-01 -2.47058854e-01 -1.28411233e+00
4.02737737e-01 6.18451893e-01 8.34312797e-01 -4.64143783e-01
1.37826607e-01 -1.44575872e-02 -5.42159796e-01 -4.96778250e-01
4.30567324e-01 7.21026585e-02 -3.23568463e-01 3.09549272e-01
6.09625340e-01 7.19151437e-01 1.86346933e-01 -1.27377734e-01
1.06147456e+00 9.35724080e-01 1.92868218e-01 -6.19239397e-02
3.76506448e-01 -6.34727418e-01 6.64239287e-01 6.56305552e-01
-1.95113242e-01 4.63527799e-01 5.02836406e-01 -4.53073770e-01
-4.77874190e-01 -8.59999359e-01 8.95463228e-02 1.37774956e+00
1.85717698e-02 -4.69871074e-01 -8.85829031e-01 -5.75124085e-01
-4.39514786e-01 5.38600147e-01 -8.82125854e-01 -2.77528763e-01
-6.72480643e-01 -7.25631773e-01 6.11847222e-01 7.96437681e-01
5.47782421e-01 -1.23212731e+00 -1.00325501e+00 2.30628029e-01
-4.55378920e-01 -1.22126615e+00 -4.02804732e-01 6.17225468e-01
-6.89235926e-01 -9.11989212e-01 -4.62430865e-01 -5.83966911e-01
2.44108036e-01 -5.97925782e-01 1.09514475e+00 -1.79168865e-01
1.46754347e-02 4.12714422e-01 -1.21065415e-01 -1.22604519e-01
-3.06550246e-02 9.53476205e-02 7.22525120e-02 1.52213603e-01
6.50469005e-01 -6.08034492e-01 -5.60264230e-01 9.37174708e-02
-3.84617180e-01 1.51724536e-02 6.07878566e-01 6.04536295e-01
6.64681733e-01 -2.96680946e-02 3.39201272e-01 -2.30449572e-01
4.41118330e-01 -4.72731203e-01 -2.85391599e-01 6.71235740e-01
-5.59164643e-01 3.02749008e-01 2.03806415e-01 -4.52520400e-01
-1.26482391e+00 3.57720464e-01 -1.63338229e-01 -4.54716533e-01
-2.18854398e-01 5.58933139e-01 -4.58929576e-02 -2.16315184e-02
2.86502957e-01 2.53094286e-01 -3.76090258e-01 -5.12074590e-01
1.01677030e-01 4.16014761e-01 7.14780450e-01 -5.76266527e-01
6.50515333e-02 9.05601755e-02 -6.18502498e-02 -7.00211823e-01
-8.94728184e-01 -1.76843032e-01 -6.02945805e-01 -6.29076183e-01
1.46609068e+00 -8.55354369e-01 -1.22583926e+00 5.64030170e-01
-1.02175820e+00 -4.63623911e-01 -3.82442713e-01 7.65686989e-01
-7.48387456e-01 3.88419867e-01 -8.20892990e-01 -1.05180693e+00
-2.64360905e-01 -8.40608060e-01 9.74585176e-01 3.09144855e-01
-3.87230963e-01 -1.07096899e+00 1.25183672e-01 5.29454350e-01
2.75765419e-01 6.59140348e-01 3.91904920e-01 -5.57514369e-01
-3.73277754e-01 -1.56376660e-01 2.82236755e-01 4.25441682e-01
-3.82793576e-01 1.34629071e-01 -9.97986257e-01 -6.59835711e-02
-1.80770263e-01 -5.71488559e-01 9.60100293e-01 7.17516899e-01
1.52689147e+00 -1.30496949e-01 -3.65295112e-01 6.82651758e-01
9.94850814e-01 1.87609673e-01 6.32207036e-01 3.31209421e-01
4.16751534e-01 3.40566665e-01 3.07777882e-01 4.42798972e-01
3.96458954e-01 8.92622650e-01 2.91612595e-01 -1.15705188e-02
7.29514062e-02 -4.47454572e-01 7.50676274e-01 4.54433501e-01
-5.30306518e-01 -2.93011159e-01 -6.70735300e-01 4.11998987e-01
-2.20429111e+00 -1.41634905e+00 2.89891422e-01 2.20941520e+00
5.88583469e-01 2.81249046e-01 1.78515360e-01 -9.64256749e-03
5.01785934e-01 4.01540279e-01 -6.03425026e-01 -5.34517944e-01
4.19850424e-02 2.99621314e-01 3.51372808e-01 4.44346219e-01
-1.37562048e+00 6.46666646e-01 7.12670660e+00 8.21760714e-01
-8.53412092e-01 2.47488379e-01 6.02133453e-01 -6.77070916e-01
2.65450150e-01 -1.42829940e-01 -6.20274901e-01 6.61795855e-01
1.32650137e+00 2.19782650e-01 5.02195239e-01 7.45238125e-01
1.76947609e-01 -2.33344451e-01 -1.32787037e+00 9.60496247e-01
1.69915661e-01 -1.31542432e+00 -3.60659868e-01 2.29782194e-01
4.92742538e-01 2.38162413e-01 -3.23596030e-01 5.54147720e-01
3.55107993e-01 -1.11792123e+00 1.25502038e+00 9.91714776e-01
5.13794243e-01 -8.44758451e-01 5.72566986e-01 4.33159322e-01
-1.31212711e+00 -1.88642085e-01 -5.76149859e-02 -1.94474578e-01
4.10917282e-01 4.15215373e-01 -4.15669493e-02 5.04419088e-01
9.76323962e-01 8.15772891e-01 -3.18264246e-01 7.04601943e-01
-2.74129570e-01 4.81359184e-01 -2.59360522e-01 1.19765112e-02
3.32999200e-01 -1.41214490e-01 3.08148295e-01 1.40262961e+00
2.90849060e-01 3.50836627e-02 2.67216831e-01 7.03870177e-01
-4.05712426e-01 -4.73052561e-01 -1.00563385e-01 -1.62367627e-01
1.40060797e-01 9.32168007e-01 -6.15414798e-01 -3.10888529e-01
-3.39988917e-01 1.19939697e+00 4.56589311e-01 3.63438249e-01
-1.11768377e+00 -3.40524286e-01 6.15995169e-01 -2.85135895e-01
3.11961204e-01 -4.40596640e-02 -3.28359194e-02 -1.29985845e+00
4.01949763e-01 -5.53029358e-01 6.37538433e-01 -7.72800088e-01
-9.22059476e-01 3.59486341e-01 1.35059819e-01 -1.05169153e+00
-6.39605999e-01 -4.90800947e-01 -6.21087134e-01 5.81850708e-01
-1.17227829e+00 -1.04705822e+00 -1.31077901e-01 4.32215750e-01
4.57452238e-01 1.31359205e-01 9.02876675e-01 3.62363636e-01
-6.65601552e-01 5.49504459e-01 -7.33541325e-02 4.38446179e-02
3.48118126e-01 -1.17047286e+00 2.54391849e-01 9.36993480e-01
1.86871529e-01 3.56301099e-01 6.27414763e-01 -3.04609120e-01
-9.90477562e-01 -7.87494659e-01 1.19451046e+00 -4.55197275e-01
4.65694249e-01 -4.63370740e-01 -6.58848107e-01 1.05653214e+00
2.83613801e-01 -6.33368315e-03 7.74524331e-01 2.62816161e-01
6.52093589e-02 1.96775854e-01 -9.17803586e-01 3.22743148e-01
1.35761523e+00 -4.84986901e-01 -6.29911304e-01 3.69004786e-01
3.38256687e-01 -3.58815700e-01 -9.70309973e-01 3.73417914e-01
9.43875372e-01 -1.05262792e+00 9.19014692e-01 -6.87048197e-01
3.01158965e-01 -6.31371588e-02 -1.43107429e-01 -9.54005837e-01
-2.42104277e-01 -6.02765083e-01 -7.41190851e-01 9.23617065e-01
4.53747928e-01 -2.59293318e-01 9.88948345e-01 8.45207989e-01
-1.34837106e-01 -1.04184520e+00 -1.17184830e+00 -6.86592400e-01
-3.60965759e-01 -7.01520085e-01 2.83490568e-01 8.09575558e-01
2.77769268e-01 1.56021221e-02 -8.46048474e-01 -1.21526234e-01
5.63837886e-01 -1.90666065e-01 2.06227854e-01 -1.15255296e+00
-6.16988003e-01 -5.32991052e-01 -4.35564727e-01 -1.11380064e+00
4.47144955e-01 -5.59952617e-01 2.13662282e-01 -1.66475141e+00
2.32222378e-01 2.57323146e-01 -8.20728838e-01 6.58746660e-01
3.02743256e-01 1.48461103e-01 -1.12224974e-01 -1.11830980e-01
-1.11975515e+00 6.47681534e-01 4.16700631e-01 1.05688825e-01
-2.13563681e-01 7.05015212e-02 -4.07131582e-01 1.07173312e+00
4.95810598e-01 -4.45115954e-01 -1.13638319e-01 -4.26718861e-01
3.85821491e-01 1.74154595e-01 5.95696807e-01 -1.31841087e+00
4.05118853e-01 -1.91556722e-01 6.73992515e-01 -4.26890045e-01
6.28001213e-01 -5.56933403e-01 3.61747414e-01 4.72623289e-01
-7.01198101e-01 -3.50720525e-01 -1.17641658e-01 6.35567009e-01
-2.49434620e-01 -1.11474723e-01 6.09674811e-01 -2.15859935e-01
-7.39659131e-01 4.28096503e-02 -7.35709310e-01 -3.75333220e-01
1.07071173e+00 -6.07513726e-01 2.57204939e-02 -4.62518096e-01
-1.05151045e+00 1.91521972e-01 2.11937860e-01 6.44704700e-01
3.68919790e-01 -1.29644608e+00 -3.43581736e-01 1.81150530e-02
-1.28752172e-01 -4.19586390e-01 3.38881493e-01 1.04934835e+00
-9.55691636e-02 3.48499060e-01 -5.47955632e-02 -3.40249270e-01
-1.17114878e+00 1.45136923e-01 9.17157292e-01 -4.98358101e-01
-4.31898743e-01 9.85527933e-01 -3.12856406e-01 -2.62733549e-01
7.17808366e-01 -4.14465547e-01 -1.43740505e-01 1.20745879e-02
3.38006437e-01 5.25212705e-01 1.77933216e-01 -7.07372248e-01
-7.67471135e-01 4.84241277e-01 4.10477698e-01 -2.19368920e-01
1.50090563e+00 8.49498212e-02 6.59519807e-02 4.56381112e-01
1.01320159e+00 -3.84744376e-01 -1.47709298e+00 -1.96482688e-02
3.73289466e-01 -2.08960101e-02 2.37609699e-01 -9.39422607e-01
-8.80435646e-01 6.54022932e-01 7.62918532e-01 -3.00871730e-02
1.04454398e+00 -2.14938782e-02 6.56674564e-01 6.20736599e-01
1.24866016e-01 -1.64321387e+00 4.68960553e-02 3.97357494e-01
5.42203665e-01 -1.01560545e+00 -8.45528767e-02 2.89445579e-01
-5.74462771e-01 1.09518242e+00 5.14170825e-01 3.06427032e-02
5.84356368e-01 2.59769410e-01 -4.34730381e-01 -3.19451362e-01
-1.08598065e+00 -6.17077798e-02 5.38179994e-01 1.16489768e-01
6.84833825e-01 -1.68679193e-01 -2.10100427e-01 6.26154780e-01
1.02797434e-01 4.34414506e-01 2.21967518e-01 1.02798676e+00
-3.81185919e-01 -7.77596474e-01 -1.37225622e-02 3.57439727e-01
-8.68870795e-01 2.83597022e-01 -3.81608337e-01 4.50755060e-01
3.93349409e-01 9.76393759e-01 1.52481228e-01 -5.21677315e-01
3.35621595e-01 6.02454782e-01 4.46530819e-01 -2.48231798e-01
-6.78251386e-01 -1.75554380e-01 3.15846056e-01 -1.07949650e+00
-8.30388665e-01 -6.05682075e-01 -1.20703721e+00 -1.57691449e-01
-2.63782531e-01 2.22521037e-01 7.64046848e-01 1.30256259e+00
4.45341885e-01 6.60985589e-01 1.51414976e-01 -1.33112264e+00
-4.68933374e-01 -1.04017758e+00 -3.30234736e-01 3.06931615e-01
1.78706214e-01 -7.71187246e-01 -3.26187849e-01 1.85637459e-01] | [8.282123565673828, 0.5472071170806885] |
fe0a905f-c649-4635-9d30-3ab985a2d5c7 | icartoonface-a-benchmark-of-cartoon-person | 1907.13394 | null | https://arxiv.org/abs/1907.13394v3 | https://arxiv.org/pdf/1907.13394v3.pdf | Cartoon Face Recognition: A Benchmark Dataset | Recent years have witnessed increasing attention in cartoon media, powered by the strong demands of industrial applications. As the first step to understand this media, cartoon face recognition is a crucial but less-explored task with few datasets proposed. In this work, we first present a new challenging benchmark dataset, consisting of 389,678 images of 5,013 cartoon characters annotated with identity, bounding box, pose, and other auxiliary attributes. The dataset, named iCartoonFace, is currently the largest-scale, high-quality, richannotated, and spanning multiple occurrences in the field of image recognition, including near-duplications, occlusions, and appearance changes. In addition, we provide two types of annotations for cartoon media, i.e., face recognition, and face detection, with the help of a semi-automatic labeling algorithm. To further investigate this challenging dataset, we propose a multi-task domain adaptation approach that jointly utilizes the human and cartoon domain knowledge with three discriminative regularizations. We hence perform a benchmark analysis of the proposed dataset and verify the superiority of the proposed approach in the cartoon face recognition task. We believe this public availability will attract more research attention in broad practical application scenarios. | ['Junhui Liu', 'Yifan Zhao', 'Yi Zheng', 'Xiangju Lu', 'Mengyuan Ren', 'Jia Li', 'He Yan'] | 2019-07-31 | null | null | null | null | ['person-recognition'] | ['computer-vision'] | [ 3.44411641e-01 -4.71682608e-01 -1.07599616e-01 -3.68724257e-01
-6.01739764e-01 -7.59289265e-01 5.77159464e-01 -4.73509789e-01
9.82513800e-02 7.43209779e-01 -1.79900214e-01 7.39657730e-02
2.69705236e-01 -3.76039088e-01 -7.71773517e-01 -9.51591313e-01
1.06916584e-01 4.19397324e-01 -1.10269785e-01 9.37863588e-02
2.93413997e-01 6.20725930e-01 -1.65999413e+00 4.65098560e-01
8.23060453e-01 1.44848776e+00 -7.33430907e-02 3.66396122e-02
2.34498352e-01 6.83902144e-01 -3.75908047e-01 -1.07808077e+00
3.50308537e-01 -2.22332865e-01 -3.79307836e-01 5.68763494e-01
9.65453506e-01 -2.85570234e-01 -1.20888717e-01 9.59276676e-01
4.58852530e-01 2.97852904e-02 7.01135457e-01 -1.60468459e+00
-6.80168986e-01 1.79312691e-01 -1.07869768e+00 -2.31208637e-01
3.96533906e-01 1.54778510e-01 7.99384594e-01 -1.25800085e+00
9.32794750e-01 1.44275188e+00 6.00756824e-01 3.28540444e-01
-9.11349118e-01 -1.23362231e+00 4.61297959e-01 4.49885190e-01
-1.36658347e+00 -6.69063866e-01 9.37313139e-01 -5.57892203e-01
1.57631963e-01 -8.36943462e-03 3.56551826e-01 1.47250164e+00
-3.31850164e-02 7.78031290e-01 1.31229234e+00 -2.35223144e-01
-2.51262798e-04 5.11374295e-01 -4.60195184e-01 9.42635596e-01
2.80531526e-01 -1.61230937e-01 -6.52903557e-01 -1.53735608e-01
5.78975618e-01 7.54612759e-02 -1.11455090e-01 -7.31928468e-01
-7.48599172e-01 6.03561699e-01 -5.77759519e-02 -5.44045540e-03
-1.42671600e-01 -3.72987539e-01 5.07082522e-01 1.56096801e-01
5.43521881e-01 -1.56954139e-01 -4.15475905e-01 -3.78210209e-02
-6.47843480e-01 2.92053539e-02 7.47612238e-01 1.37596667e+00
4.49511737e-01 2.27627158e-01 1.97140053e-01 1.13618922e+00
2.98576981e-01 7.60072947e-01 7.85004124e-02 -4.48061854e-01
4.95009959e-01 6.14512205e-01 -7.53401890e-02 -1.36550570e+00
-2.26603672e-01 -1.42347351e-01 -8.35451066e-01 -3.36756147e-02
4.93151993e-01 -1.57381177e-01 -5.96491516e-01 1.38471353e+00
4.43706095e-01 3.62521261e-01 -9.71144736e-02 9.03114855e-01
8.04618776e-01 4.13685352e-01 8.28920826e-02 -2.70772874e-01
1.57759809e+00 -1.04477036e+00 -4.69524771e-01 1.43234413e-02
8.34698156e-02 -1.20771670e+00 8.31642568e-01 6.61838233e-01
-4.87222761e-01 -5.08263648e-01 -9.65709746e-01 2.33772472e-01
-2.89158046e-01 6.98910773e-01 5.95003128e-01 9.50663388e-01
-4.32945043e-01 2.41466925e-01 -2.26876512e-01 -4.51768726e-01
6.73354805e-01 7.87061900e-02 -5.96435964e-01 -6.49880528e-01
-7.09339738e-01 5.88888466e-01 -4.56727110e-02 9.37248990e-02
-8.43039632e-01 -4.96260345e-01 -7.19750881e-01 -4.70507860e-01
7.80734837e-01 -2.20376551e-02 7.68248558e-01 -1.16469371e+00
-1.42283547e+00 9.26755607e-01 -6.54252321e-02 -7.98131749e-02
6.65719450e-01 -1.57730415e-01 -8.81955087e-01 1.81841761e-01
1.47309959e-01 4.32509273e-01 1.40910888e+00 -1.34730124e+00
-7.13692307e-01 -5.23009539e-01 -1.17049336e-01 -2.44475175e-02
-5.31800389e-01 1.91207126e-01 -7.73160398e-01 -8.26692760e-01
-2.23238632e-01 -1.22890270e+00 4.33968008e-01 2.41960317e-01
-5.28742194e-01 -1.72151968e-01 1.27280283e+00 -8.25198948e-01
8.38299751e-01 -2.34748936e+00 -4.83356928e-03 6.78109229e-02
-2.16772139e-01 1.45084545e-01 -2.85886496e-01 2.97637999e-01
-6.42047897e-02 -1.43564135e-01 -3.09593797e-01 -2.85247892e-01
1.15740910e-01 -2.41899058e-01 -3.78015012e-01 6.27847016e-01
4.53591406e-01 3.60553682e-01 -4.89680618e-01 -6.43772185e-01
9.06675979e-02 4.88565058e-01 -3.70768815e-01 -6.55974820e-03
-1.41170487e-01 7.06258953e-01 -4.14076716e-01 1.44863868e+00
1.15933943e+00 -5.60266823e-02 4.89480913e-01 -5.85809529e-01
-6.76744506e-02 -6.10601068e-01 -1.25624347e+00 1.39783478e+00
-3.91327769e-01 6.67950749e-01 1.82802439e-01 -7.15524137e-01
1.04912031e+00 2.48953566e-01 8.41728270e-01 -7.11505532e-01
2.81830102e-01 2.43430525e-01 -1.56701535e-01 -5.29711723e-01
3.12255770e-01 -2.63558961e-02 -1.61231145e-01 2.93558240e-01
1.51194027e-02 7.65877664e-02 3.93538505e-01 -1.93988495e-02
6.46094143e-01 4.40954328e-01 2.86119491e-01 -1.10909000e-01
7.28183270e-01 -8.83736759e-02 7.19079018e-01 1.69323444e-01
-3.44801992e-01 5.10826945e-01 5.38188040e-01 -6.05308235e-01
-9.58057165e-01 -5.98686039e-01 -3.87124181e-01 1.25249135e+00
1.73173383e-01 -8.75188485e-02 -6.38922215e-01 -9.34147298e-01
3.29623342e-01 1.07912228e-01 -4.98843372e-01 1.73058197e-01
-6.44631565e-01 -9.57066834e-01 4.46104854e-01 3.04041147e-01
8.28487933e-01 -1.11525106e+00 -1.50614455e-01 -1.78367674e-01
-1.36816636e-01 -1.71302247e+00 -6.60177886e-01 -4.13993925e-01
-4.33265179e-01 -1.59943867e+00 -8.39524686e-01 -1.12621462e+00
6.52125239e-01 1.21918693e-01 9.08786297e-01 -1.35650992e-01
-4.50782984e-01 5.15950859e-01 -3.75361979e-01 -2.69147307e-01
4.43951823e-02 -1.41302526e-01 1.58550635e-01 7.52827346e-01
2.22271279e-01 -3.14529300e-01 -3.82824630e-01 6.25382900e-01
-5.31119168e-01 -8.26454014e-02 8.06973755e-01 7.29160130e-01
4.04260874e-01 -2.75161099e-02 5.82485616e-01 -1.26881361e+00
2.54579544e-01 -4.88151699e-01 -8.76497805e-01 4.40043241e-01
-2.36140132e-01 -4.87033188e-01 8.33535433e-01 -5.14652252e-01
-1.26142550e+00 4.56675231e-01 1.91771716e-01 -4.50702757e-01
-3.25518012e-01 7.66257197e-02 -5.94279230e-01 -5.36210358e-01
2.61385918e-01 2.58382559e-01 -2.60248721e-01 -4.47417289e-01
3.08312953e-01 6.41197085e-01 4.57726926e-01 -8.73059034e-01
9.55557406e-01 5.31222343e-01 5.89473359e-02 -1.09110022e+00
-5.00055432e-01 -2.89392531e-01 -8.32175732e-01 -5.23647785e-01
5.68085670e-01 -1.12724686e+00 -8.92883837e-01 8.79865885e-01
-1.02046883e+00 1.21609867e-01 4.04660970e-01 2.29225159e-02
-3.04062366e-01 5.44542372e-01 -3.10144603e-01 -8.42888355e-01
-2.32093148e-02 -1.04592395e+00 1.14105475e+00 8.86632428e-02
3.86744350e-01 -5.79088449e-01 -4.04176623e-01 4.79476780e-01
-1.90874636e-02 5.66590250e-01 8.73917282e-01 -5.78046203e-01
-8.33460808e-01 -2.43462652e-01 -6.52941108e-01 2.71547079e-01
1.74043790e-01 9.67377424e-02 -9.93006647e-01 -4.90963370e-01
-2.81215161e-01 -5.31792760e-01 6.71474636e-01 -1.62774220e-01
1.23811412e+00 -1.98896676e-01 -3.56635451e-01 5.61980486e-01
1.21605706e+00 5.17894030e-01 3.42159867e-01 1.68033361e-01
9.18600440e-01 6.90474808e-01 9.12836313e-01 9.08317924e-01
1.92732900e-01 7.70944953e-01 3.45007002e-01 9.42478552e-02
-7.99967125e-02 -1.90769359e-01 3.79298836e-01 6.90044403e-01
-1.83980718e-01 -2.57437348e-01 -6.66709363e-01 2.48339072e-01
-1.51108325e+00 -7.48121679e-01 1.59397170e-01 1.87184644e+00
4.75563675e-01 -1.00281313e-01 2.50766993e-01 -5.17181121e-02
1.00059271e+00 1.20304048e-01 -9.23180640e-01 1.43246308e-01
-1.96114019e-01 -4.89877723e-02 2.82052100e-01 -1.82240963e-01
-1.50889409e+00 9.55013454e-01 5.16103268e+00 7.74676204e-01
-1.17499924e+00 2.01715175e-02 7.15268016e-01 2.07721829e-01
3.59671563e-01 -3.96684289e-01 -8.85763109e-01 6.79724813e-01
3.26006562e-01 1.66945413e-01 5.03863931e-01 1.05641770e+00
-2.17425510e-01 1.77483484e-01 -1.05759287e+00 1.23087239e+00
7.33658433e-01 -9.37445879e-01 1.83489919e-01 -3.54041830e-02
8.84123325e-01 -6.00834548e-01 4.81600523e-01 2.99486190e-01
-2.75748875e-02 -7.01983035e-01 7.23403275e-01 2.45564491e-01
1.20526075e+00 -7.99361169e-01 5.93788683e-01 -1.34901434e-01
-1.47625196e+00 -3.69030088e-01 -1.46220237e-01 5.30197203e-01
-2.89937388e-02 3.86883080e-01 -6.40990138e-01 6.26616716e-01
6.33864701e-01 1.07511127e+00 -4.96057421e-01 1.01095760e+00
-5.26358224e-02 2.84710288e-01 -4.08932613e-03 7.49681890e-02
-1.15819164e-01 -2.85326481e-01 3.48303229e-01 1.04590011e+00
2.85917133e-01 1.32469917e-02 4.21126842e-01 2.98813999e-01
-4.24034029e-01 4.89328951e-01 -6.36530042e-01 -1.98537543e-01
6.05219960e-01 1.39397967e+00 -7.60409653e-01 -1.56195670e-01
-7.90965676e-01 9.21545148e-01 1.67720094e-02 2.49535620e-01
-1.14790535e+00 -2.05581561e-01 6.96990132e-01 -8.68222415e-02
4.75839972e-01 -4.99675795e-02 7.41908476e-02 -1.21137595e+00
1.98714018e-01 -1.31992733e+00 4.18745697e-01 -4.84448701e-01
-1.57228971e+00 8.67585301e-01 -1.79642096e-01 -1.49266160e+00
4.86114919e-02 -1.11913502e+00 -3.16262484e-01 3.37058872e-01
-1.50957060e+00 -1.66364694e+00 -5.28041542e-01 7.58165240e-01
7.51393974e-01 -8.12682271e-01 5.82058728e-01 8.12526703e-01
-9.83518302e-01 7.82307565e-01 1.71912640e-01 2.59157956e-01
1.22645283e+00 -6.12282693e-01 9.75547582e-02 5.95813870e-01
-4.82069813e-02 2.92405456e-01 4.33180571e-01 -6.64954424e-01
-1.66033232e+00 -1.29629588e+00 3.74604136e-01 -2.77033418e-01
6.15976512e-01 -6.55630827e-01 -5.78051031e-01 6.07841611e-01
2.82744803e-02 1.65171802e-01 6.64424598e-01 -2.31682882e-02
-5.09807944e-01 -4.57588255e-01 -1.21190989e+00 3.91583592e-01
1.11486208e+00 -4.17900294e-01 9.66845155e-02 5.33762991e-01
1.73343405e-01 -3.27920794e-01 -7.94775844e-01 4.99659151e-01
8.17696393e-01 -7.42978752e-01 8.74558628e-01 -3.28784764e-01
3.88230592e-01 -9.44495574e-02 -3.32948893e-01 -8.01322877e-01
2.98677161e-02 -4.16311860e-01 7.37179965e-02 1.51339614e+00
1.38718188e-01 -5.31295359e-01 1.05254793e+00 3.12202662e-01
1.59835704e-02 -5.10621071e-01 -8.71186614e-01 -6.99185669e-01
-1.54710293e-01 -7.31984153e-02 6.51055932e-01 1.02988052e+00
-2.97927290e-01 9.47596878e-02 -1.01312697e+00 7.92233720e-02
7.68168032e-01 5.83483696e-01 9.75134730e-01 -1.38473260e+00
2.11051524e-01 -8.40291753e-02 -3.55211169e-01 -8.05576146e-01
4.29611802e-01 -6.86690331e-01 -5.27329072e-02 -7.02244997e-01
3.21643263e-01 -4.70586121e-01 1.15638841e-02 4.60662305e-01
3.17978114e-01 6.91870570e-01 3.71306390e-01 1.00491896e-01
-7.85271466e-01 6.02844119e-01 1.28016150e+00 -3.26404393e-01
3.08972478e-01 -2.34659575e-02 -5.31092227e-01 7.89465308e-01
6.69896126e-01 -2.83149898e-01 -1.98897690e-01 -1.47186100e-01
-3.74284029e-01 -1.03115171e-01 2.79043943e-01 -9.72889960e-01
1.47596598e-01 -2.52601564e-01 6.80277944e-01 -6.29782259e-01
6.58665657e-01 -1.19151008e+00 1.16490070e-02 2.44303092e-01
1.75117314e-01 1.27880991e-01 3.52594078e-01 6.26028657e-01
-1.95675239e-01 4.02093008e-02 9.23332691e-01 -6.65596798e-02
-1.16661596e+00 6.33809388e-01 -1.20711416e-01 5.33036776e-02
1.40215015e+00 -9.64548215e-02 -3.60027701e-01 -3.82180512e-02
-5.26226401e-01 -2.62157526e-02 3.67195219e-01 7.43970156e-01
5.92313111e-01 -1.44794214e+00 -7.15614617e-01 4.25667286e-01
3.16812277e-01 -5.43296516e-01 4.62218821e-01 7.15315580e-01
-2.77640730e-01 4.63238686e-01 -7.73334146e-01 -3.86476666e-01
-1.59512508e+00 6.42555833e-01 -7.95552284e-02 1.11017115e-01
-4.74168628e-01 5.92525244e-01 2.19515964e-01 -3.33198577e-01
3.36764932e-01 2.54269630e-01 -3.58793646e-01 4.63749141e-01
3.21606070e-01 4.72605288e-01 -1.21539272e-02 -1.06642032e+00
-6.06565177e-01 8.80649269e-01 -1.96096256e-01 3.48412126e-01
1.14427924e+00 2.48512626e-02 -8.40312988e-02 4.79502827e-02
1.04450750e+00 1.04139976e-01 -1.35656810e+00 -1.69881418e-01
-1.14247091e-02 -8.61369729e-01 -5.79984069e-01 -8.58616352e-01
-1.42887259e+00 7.80711949e-01 7.17160106e-01 -2.31904954e-01
1.21520150e+00 -2.97893792e-01 7.35140979e-01 2.79768020e-01
6.71369553e-01 -1.16088784e+00 3.83331776e-01 2.79324532e-01
1.16103637e+00 -1.49188280e+00 2.32106328e-01 -9.73742604e-01
-8.66056025e-01 1.07698095e+00 1.12324035e+00 1.77490324e-01
5.35658956e-01 1.18277669e-01 4.32347208e-02 1.58384383e-01
-5.49022675e-01 6.88238069e-02 2.32230410e-01 7.20089257e-01
3.04014295e-01 -4.39809971e-02 9.44550633e-02 7.05788553e-01
-1.53748207e-02 -1.76983699e-01 1.27936989e-01 7.49954283e-01
-5.71768060e-02 -1.06584656e+00 -5.10370433e-01 2.94440299e-01
-3.03036720e-01 9.86466929e-02 -5.26814759e-01 9.37082350e-01
3.80074739e-01 8.78266454e-01 -2.38146439e-01 -3.71267319e-01
3.91894102e-01 2.54669809e-03 4.91395980e-01 -2.15195194e-01
-1.35409534e-01 -2.33550090e-03 7.65623674e-02 -3.08189720e-01
-4.03599054e-01 -1.08405089e+00 -6.17084086e-01 -4.75476921e-01
-7.36250803e-02 -1.22015052e-01 6.54487908e-01 5.70153832e-01
3.03883940e-01 2.90275425e-01 8.69962215e-01 -9.02502537e-01
-2.11540192e-01 -8.56277525e-01 -7.57600129e-01 5.87232351e-01
8.17918181e-02 -1.07977343e+00 -7.77577087e-02 3.13484520e-01] | [13.282976150512695, 0.6846533417701721] |
ea6da077-d339-47ce-b7eb-e49458886522 | on-the-complexity-of-representation-learning | 2212.09429 | null | https://arxiv.org/abs/2212.09429v1 | https://arxiv.org/pdf/2212.09429v1.pdf | On the Complexity of Representation Learning in Contextual Linear Bandits | In contextual linear bandits, the reward function is assumed to be a linear combination of an unknown reward vector and a given embedding of context-arm pairs. In practice, the embedding is often learned at the same time as the reward vector, thus leading to an online representation learning problem. Existing approaches to representation learning in contextual bandits are either very generic (e.g., model-selection techniques or algorithms for learning with arbitrary function classes) or specialized to particular structures (e.g., nested features or representations with certain spectral properties). As a result, the understanding of the cost of representation learning in contextual linear bandit is still limited. In this paper, we take a systematic approach to the problem and provide a comprehensive study through an instance-dependent perspective. We show that representation learning is fundamentally more complex than linear bandits (i.e., learning with a given representation). In particular, learning with a given set of representations is never simpler than learning with the worst realizable representation in the set, while we show cases where it can be arbitrarily harder. We complement this result with an extensive discussion of how it relates to existing literature and we illustrate positive instances where representation learning is as complex as learning with a fixed representation and where sub-logarithmic regret is achievable. | ['Alessandro Lazaric', 'Matteo Pirotta', 'Andrea Tirinzoni'] | 2022-12-19 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 5.32978714e-01 2.90314078e-01 -1.11923218e+00 -1.49678066e-01
-1.22893488e+00 -1.00225902e+00 3.65085661e-01 1.41347885e-01
3.01714018e-02 1.01738155e+00 1.32720426e-01 -6.58834875e-01
-7.20052421e-01 -9.44204926e-01 -1.17699099e+00 -8.91416550e-01
-2.54798383e-01 5.88018179e-01 -4.31635916e-01 -8.70811939e-02
1.12810358e-01 5.01131952e-01 -1.29053390e+00 3.10189337e-01
6.24285042e-01 1.23428392e+00 3.51305306e-02 6.99256420e-01
-2.33108729e-01 5.55243254e-01 -3.72696817e-01 -4.04113472e-01
6.55885637e-01 -5.86323678e-01 -8.75077009e-01 3.29591364e-01
4.46583390e-01 -1.43502042e-01 -4.77268517e-01 8.95697594e-01
1.11730203e-01 2.40752086e-01 6.40530288e-01 -1.06300032e+00
-7.57875443e-01 7.82862842e-01 -5.00621259e-01 1.45838931e-01
2.22738653e-01 -2.63333917e-01 1.64886034e+00 -2.46392652e-01
2.34787047e-01 1.16930830e+00 3.82181078e-01 3.68970186e-01
-1.81695795e+00 -5.42158842e-01 6.77487493e-01 1.71359897e-01
-8.38982105e-01 -3.44345510e-01 6.54388726e-01 -4.30767566e-01
5.76063275e-01 7.15266287e-01 8.22586536e-01 9.53432977e-01
-4.29287910e-01 1.15754485e+00 1.13585043e+00 -7.40915596e-01
4.37397450e-01 1.97824359e-01 4.71195102e-01 4.56364125e-01
5.55518150e-01 5.29731750e-01 -4.14329439e-01 -4.60750937e-01
8.98864627e-01 3.23870540e-01 -2.60016769e-01 -1.13481963e+00
-8.25465977e-01 1.33458269e+00 5.18425822e-01 1.45106390e-01
-1.94237903e-01 6.33864701e-01 3.03331167e-01 8.24006200e-01
3.90828878e-01 5.78480542e-01 -4.31662887e-01 5.28940745e-02
-7.93735683e-01 4.13012505e-01 8.46124113e-01 1.11138320e+00
1.04112411e+00 3.83565910e-02 -4.11042482e-01 7.35746086e-01
-2.45449711e-02 2.91332811e-01 1.10564448e-01 -7.28708923e-01
7.11923897e-01 1.39086589e-01 3.95189136e-01 -5.82921207e-01
-1.94044501e-01 -6.90445006e-01 -4.14951771e-01 2.18203105e-02
7.07291484e-01 -1.92754626e-01 -6.39802217e-01 1.90184593e+00
1.04029082e-01 8.58519450e-02 -1.75847970e-02 7.43759453e-01
6.10463172e-02 6.18655503e-01 -3.34040433e-01 -5.48124492e-01
9.87820327e-01 -1.06738544e+00 -5.65556705e-01 -5.05281210e-01
6.76166356e-01 -3.48681122e-01 6.88478470e-01 2.64933825e-01
-9.53511000e-01 9.83880386e-02 -9.92133498e-01 2.11412445e-01
-2.39959896e-01 -1.66592285e-01 1.18480551e+00 1.04897082e+00
-7.25291669e-01 5.94045579e-01 -5.22664845e-01 -2.64610678e-01
4.79280055e-01 4.55809563e-01 -1.34719461e-01 -3.66124898e-01
-1.01426411e+00 6.35185003e-01 4.40146267e-01 -1.77895948e-02
-8.09870183e-01 -7.26345837e-01 -7.26745486e-01 1.46130100e-01
7.44458437e-01 -6.66348040e-01 1.51852632e+00 -1.55797982e+00
-1.41589785e+00 3.63938510e-01 -7.76445717e-02 -8.52928638e-01
6.36433184e-01 -1.00513607e-01 1.37042189e-02 -2.94169486e-02
-1.99286357e-01 -9.79533568e-02 9.86330926e-01 -1.12407160e+00
-6.04016960e-01 -3.29615325e-01 8.77878785e-01 8.40945318e-02
-2.57930279e-01 -2.26798922e-01 9.80390534e-02 -6.63969278e-01
3.82486172e-02 -1.07004273e+00 -4.29386258e-01 2.69209202e-02
-3.38914007e-01 1.95090733e-02 3.79008740e-01 -2.62437314e-01
1.32701468e+00 -2.08683729e+00 2.50249356e-01 5.08804798e-01
-2.35017046e-01 -1.96976066e-01 -2.70579278e-01 7.29919314e-01
-3.61014694e-01 1.55309200e-01 -2.29484200e-01 -1.02451265e-01
2.58227408e-01 4.21507359e-01 -5.74480832e-01 7.74063289e-01
-1.10836953e-01 9.04690921e-01 -1.03459311e+00 2.37816013e-02
9.58043784e-02 -8.09069276e-02 -9.40232038e-01 -1.08909346e-01
-4.66059655e-01 6.37709722e-02 -7.72077501e-01 6.86712265e-01
3.47169876e-01 -3.26854825e-01 5.12735188e-01 2.98101693e-01
4.00621407e-02 3.66286725e-01 -1.32349980e+00 1.43901801e+00
-7.29019761e-01 6.11064851e-01 -3.90464813e-02 -1.72373331e+00
5.22755682e-01 3.05623561e-01 4.51839328e-01 -3.75063390e-01
-2.71624237e-01 3.31258118e-01 -4.04217690e-02 -6.73597977e-02
2.85143465e-01 -4.31854546e-01 -7.60436282e-02 6.55371070e-01
-5.17067909e-02 1.45385236e-01 9.70738102e-03 1.56405903e-02
1.02715433e+00 -5.18613160e-02 5.81769288e-01 -9.51401591e-02
-5.70537187e-02 -1.58145279e-01 3.95482957e-01 1.30643976e+00
1.78865954e-01 4.42752868e-01 8.30689430e-01 -3.30272555e-01
-9.02774751e-01 -9.01866198e-01 -2.47600675e-01 1.35705185e+00
-1.48712426e-01 -3.39279532e-01 -2.56193012e-01 -6.84276283e-01
4.70454454e-01 5.78802884e-01 -8.83694172e-01 -1.98921204e-01
-4.16386843e-01 -6.69481814e-01 -8.80054757e-02 5.20955861e-01
-4.77143452e-02 -6.62200928e-01 -4.03195381e-01 2.75791109e-01
3.25422138e-01 -5.41161895e-01 -4.13740069e-01 6.98923409e-01
-1.12356710e+00 -1.16536009e+00 -6.47361159e-01 -4.15555924e-01
5.79294503e-01 5.35734177e-01 1.01517081e+00 -1.96125612e-01
-5.84652014e-02 6.35029018e-01 -3.83559853e-01 -3.35641533e-01
1.75444752e-01 1.15891047e-01 -3.48830581e-01 2.00716019e-01
1.29333918e-03 -5.65083206e-01 -6.04819834e-01 -1.25371860e-02
-9.19007182e-01 -1.74295545e-01 6.11438572e-01 1.34153712e+00
5.28514087e-01 -2.89044380e-01 7.62125671e-01 -1.25607550e+00
5.67562938e-01 -8.44417751e-01 -7.82366931e-01 5.39213061e-01
-4.69597548e-01 1.75431415e-01 4.68088895e-01 -6.74275875e-01
-5.39638817e-01 2.29598269e-01 5.43064475e-01 -4.59482163e-01
4.32522058e-01 7.50456452e-01 7.53319487e-02 -1.03167929e-01
6.81609511e-01 2.08478257e-01 -1.19000591e-01 -4.86631930e-01
6.20376825e-01 3.69050056e-01 -1.99421957e-01 -1.03948224e+00
6.33316755e-01 1.87585294e-01 1.02485642e-01 -5.57620704e-01
-1.19034100e+00 -2.61255115e-01 -1.34026960e-01 5.57688326e-02
1.48997009e-02 -6.46930099e-01 -5.15192568e-01 -4.15445596e-01
-7.65806258e-01 -5.18993616e-01 -8.53514373e-01 6.44992113e-01
-1.19496131e+00 -3.98613922e-02 -2.61289805e-01 -1.48356056e+00
1.45231828e-01 -9.64280128e-01 7.36351728e-01 -9.73491073e-02
4.11158055e-02 -9.25185204e-01 1.15368161e-02 2.79051960e-01
3.14183056e-01 2.92143494e-01 1.21462691e+00 -6.47385001e-01
-1.00272548e+00 -2.20872283e-01 -1.36707976e-01 1.44114390e-01
1.71891227e-01 -3.30587715e-01 -8.40260506e-01 -6.07146859e-01
-2.99755543e-01 -4.67399180e-01 1.10076368e+00 7.23104656e-01
1.40420949e+00 -7.27668881e-01 -3.25051934e-01 5.70040643e-01
1.57848954e+00 3.67031038e-01 3.10713321e-01 5.54310620e-01
1.54284298e-01 5.75922489e-01 6.14736199e-01 4.64796245e-01
-1.99051961e-01 6.95588410e-01 5.84773481e-01 3.20555240e-01
3.36910903e-01 -3.98884475e-01 1.13329008e-01 -4.93949354e-02
-3.02177761e-02 -1.17683206e-02 -4.90058541e-01 6.17000520e-01
-2.34841037e+00 -1.06146121e+00 5.32213628e-01 2.59932089e+00
7.99669623e-01 -1.68429818e-02 5.50648212e-01 1.84082165e-01
8.05482268e-01 2.62450635e-01 -9.18079674e-01 -5.86347401e-01
-2.02654209e-02 2.18414545e-01 8.86546791e-01 4.89142150e-01
-1.03730774e+00 5.54258525e-01 6.97288656e+00 9.43574905e-01
-9.25282359e-01 1.79860190e-01 7.45623410e-01 -6.26672924e-01
-6.11366630e-01 1.61831781e-01 -4.90343362e-01 2.83987880e-01
7.45551944e-01 -3.44701797e-01 1.03468251e+00 1.05073988e+00
-2.13034496e-01 1.68749243e-01 -1.53963709e+00 1.13179016e+00
-3.40936005e-01 -1.59287465e+00 -7.35629722e-03 4.25175428e-01
1.01705706e+00 -2.47800276e-01 3.93882304e-01 4.85685974e-01
5.59011817e-01 -1.22090495e+00 7.40104854e-01 4.24521893e-01
8.68824244e-01 -1.10175121e+00 2.96349376e-01 3.73694569e-01
-8.61526608e-01 -7.83180714e-01 -4.34647709e-01 -1.71187088e-01
-2.42477149e-01 3.52409840e-01 -3.68384570e-01 6.75501525e-01
1.86715201e-01 8.27016294e-01 7.03372285e-02 1.43437052e+00
1.54370656e-02 7.06092834e-01 -4.07909572e-01 -7.06402138e-02
5.34395874e-01 -4.35973078e-01 4.37603414e-01 1.09582186e+00
4.26603884e-01 -6.29467294e-02 3.15431684e-01 5.45168757e-01
-2.05978304e-01 2.80269682e-01 -8.31367970e-01 -2.15833902e-01
4.33223456e-01 8.13222528e-01 -2.39178285e-01 -1.68379277e-01
-5.54738879e-01 4.74018753e-01 7.47078061e-01 6.38528168e-01
-4.82023805e-01 1.85584411e-01 9.59283054e-01 1.46233127e-01
5.62340736e-01 7.47767910e-02 -2.24254116e-01 -9.99691308e-01
7.81501606e-02 -7.73834825e-01 7.22281277e-01 -2.14633912e-01
-1.31921077e+00 -3.88674624e-02 1.17754705e-01 -1.24652445e+00
-4.45601642e-01 -5.51725388e-01 -3.00428778e-01 7.06490159e-01
-1.77493405e+00 -8.73408258e-01 5.55369854e-01 5.68562984e-01
4.05968130e-01 -2.26613075e-01 8.88917208e-01 2.30912920e-02
-4.46339309e-01 7.34612703e-01 8.10915351e-01 -2.51025289e-01
2.66340762e-01 -1.36315334e+00 -2.96990305e-01 3.95531684e-01
4.17391032e-01 7.19598413e-01 6.30365133e-01 -9.45078358e-02
-1.91594863e+00 -1.07732344e+00 2.91780770e-01 -1.04757823e-01
1.06989646e+00 -4.72599804e-01 -4.82596099e-01 1.18732953e+00
-2.03437164e-01 3.84013355e-01 8.85752916e-01 5.70182979e-01
-5.60456991e-01 -3.47911358e-01 -1.11776793e+00 5.40560007e-01
1.24685979e+00 -6.69200361e-01 -2.43764967e-01 6.36110425e-01
8.00452232e-01 -2.86326557e-01 -5.99376023e-01 6.88383803e-02
8.58211994e-01 -6.92079008e-01 1.08315957e+00 -1.27757931e+00
1.49100438e-01 1.18830353e-01 -6.08763576e-01 -1.35076511e+00
-4.27504838e-01 -9.25782144e-01 -7.22483218e-01 8.37608457e-01
5.59811831e-01 -9.71042693e-01 8.94655108e-01 4.55088228e-01
1.11020282e-01 -1.04027283e+00 -1.29607141e+00 -1.07669115e+00
3.59065711e-01 -4.61822659e-01 6.60954952e-01 8.50371122e-01
8.83338451e-02 6.93788677e-02 -6.29884422e-01 1.76333729e-02
4.78675902e-01 8.60101759e-01 5.90657234e-01 -1.02682257e+00
-8.15333009e-01 -5.33572555e-01 -3.15978646e-01 -1.56525266e+00
1.86920106e-01 -1.12863231e+00 -3.53927135e-01 -1.25302279e+00
3.40360045e-01 -8.65951657e-01 -6.94797993e-01 2.66614139e-01
2.85098404e-01 -2.35767394e-01 3.40996385e-01 3.79896052e-02
-4.15891677e-01 2.53738821e-01 1.35591507e+00 -3.92654538e-01
-2.49459729e-01 5.78631580e-01 -1.20557523e+00 1.68359086e-01
7.54842281e-01 -4.24925923e-01 -6.62550807e-01 -1.50024444e-01
5.10909796e-01 5.55335283e-01 3.88170034e-01 -2.97577292e-01
-6.70090038e-03 -6.36489332e-01 9.55447555e-02 -3.18633080e-01
4.92775559e-01 -1.04152858e+00 1.47243142e-02 3.47382635e-01
-7.72284746e-01 -4.73258555e-01 -7.36496001e-02 1.26472032e+00
3.60204689e-02 -6.21748447e-01 5.07337868e-01 -6.06468655e-02
-8.90912786e-02 4.59351629e-01 -7.32930526e-02 2.01291051e-02
1.03262258e+00 -3.11737418e-01 -2.46858105e-01 -6.95058167e-01
-8.98154259e-01 1.16392866e-01 1.93573490e-01 4.30282729e-04
2.90452540e-01 -1.49553347e+00 -5.25739670e-01 -5.47159985e-02
3.20787162e-01 -4.98211272e-02 -3.45226973e-02 5.86447895e-01
1.01374194e-01 6.73024952e-01 1.97030917e-01 -2.44733304e-01
-9.29736912e-01 8.41810584e-01 3.97082061e-01 -4.43545222e-01
-3.77722561e-01 6.31849766e-01 3.49371195e-01 -2.04884067e-01
3.56980294e-01 -3.95549506e-01 1.02269679e-01 2.91744530e-01
5.25464237e-01 2.39027411e-01 -1.52978808e-01 -1.28780678e-01
1.41473293e-01 3.38840097e-01 -1.64520741e-01 2.44235843e-02
1.46284533e+00 9.80358347e-02 1.14811286e-01 6.87807262e-01
1.21873915e+00 -9.33401883e-02 -1.44832385e+00 -6.94807529e-01
1.11315675e-01 -8.86056304e-01 1.72119904e-02 -5.66947162e-01
-1.17089212e+00 5.65399885e-01 4.38168973e-01 6.91852331e-01
8.63130331e-01 2.36550868e-01 2.43835151e-01 7.45402515e-01
6.27638459e-01 -1.05497110e+00 -6.61076009e-02 3.48018408e-01
1.09308612e+00 -1.04556859e+00 9.87184942e-02 -2.01286465e-01
-1.42761931e-01 1.21562099e+00 -1.27813024e-02 -4.08945620e-01
6.74365520e-01 -1.40917361e-01 -7.14781940e-01 1.81881666e-01
-8.41391802e-01 -4.03148472e-01 2.70699143e-01 5.63926756e-01
3.19151372e-01 5.30273795e-01 -1.95552930e-01 5.25142729e-01
-2.88328677e-01 -2.30598852e-01 3.27072740e-01 9.66709912e-01
-3.87432456e-01 -1.26602387e+00 -4.63419765e-01 9.79634881e-01
-4.36618745e-01 -4.11532782e-02 -2.13768408e-01 6.69723153e-01
-3.98529530e-01 9.31446433e-01 8.29970986e-02 1.30115375e-01
2.33799383e-01 6.00513406e-02 8.22469711e-01 -9.26332772e-01
-2.56440192e-01 4.18962911e-02 2.40063787e-01 -3.96701366e-01
-4.27014619e-01 -9.42893386e-01 -5.70544124e-01 -2.45737508e-01
-5.98818719e-01 2.45450616e-01 5.06404877e-01 7.87275255e-01
-6.43951893e-02 4.05483752e-01 8.94842148e-01 -7.09021211e-01
-1.01441836e+00 -6.28705323e-01 -9.87372041e-01 -5.34106493e-02
9.66501117e-01 -7.84067929e-01 -3.34838957e-01 -5.26900291e-01] | [4.5696702003479, 3.3141748905181885] |
7606bb63-7821-4ca3-a730-0381e7cb88e0 | tad-a-large-scale-benchmark-for-traffic | 2209.12386 | null | https://arxiv.org/abs/2209.12386v1 | https://arxiv.org/pdf/2209.12386v1.pdf | TAD: A Large-Scale Benchmark for Traffic Accidents Detection from Video Surveillance | Automatic traffic accidents detection has appealed to the machine vision community due to its implications on the development of autonomous intelligent transportation systems (ITS) and importance to traffic safety. Most previous studies on efficient analysis and prediction of traffic accidents, however, have used small-scale datasets with limited coverage, which limits their effect and applicability. Existing datasets in traffic accidents are either small-scale, not from surveillance cameras, not open-sourced, or not built for freeway scenes. Since accidents happened in freeways tend to cause serious damage and are too fast to catch the spot. An open-sourced datasets targeting on freeway traffic accidents collected from surveillance cameras is in great need and of practical importance. In order to help the vision community address these shortcomings, we endeavor to collect video data of real traffic accidents that covered abundant scenes. After integration and annotation by various dimensions, a large-scale traffic accidents dataset named TAD is proposed in this work. Various experiments on image classification, object detection, and video classification tasks, using public mainstream vision algorithms or frameworks are conducted in this work to demonstrate performance of different methods. The proposed dataset together with the experimental results are presented as a new benchmark to improve computer vision research, especially in ITS. | ['Shiguo Lian', 'Yibing Nan', 'Chuwen Huang', 'Yajun Xu'] | 2022-09-26 | null | null | null | null | ['video-classification'] | ['computer-vision'] | [ 9.71919596e-02 -4.72346395e-01 -2.99324870e-01 -2.24273518e-01
-5.23680091e-01 -7.46991113e-02 4.67729390e-01 -3.11786503e-01
-6.02019727e-01 6.89938366e-01 1.26509055e-01 -2.99617141e-01
-6.56764209e-02 -8.03922057e-01 -3.65927905e-01 -7.92882144e-01
2.57718921e-01 1.71480909e-01 8.25339675e-01 -3.74954641e-01
3.47629607e-01 5.61411142e-01 -1.87551856e+00 2.67139584e-01
7.04249978e-01 1.01129198e+00 3.10283870e-01 4.03437823e-01
-1.60160735e-01 8.26208413e-01 -4.04077977e-01 -6.00420475e-01
6.11233652e-01 -7.26731047e-02 -3.95800233e-01 1.90758511e-01
4.53302145e-01 -4.48064774e-01 -8.35434854e-01 9.80890095e-01
3.84367794e-01 1.79095969e-01 6.00835979e-01 -1.64089882e+00
-4.85232562e-01 -2.87081391e-01 -6.31238997e-01 6.71893120e-01
-4.17047441e-02 5.98141909e-01 5.25041580e-01 -5.84883869e-01
4.33344275e-01 1.30240929e+00 6.90838754e-01 4.65020508e-01
-4.87355053e-01 -7.89948463e-01 1.75090674e-02 7.71769285e-01
-1.15157366e+00 -5.76318324e-01 8.02405715e-01 -6.65355623e-01
9.69063401e-01 3.48240793e-01 6.43143475e-01 1.29612029e+00
1.68845266e-01 6.98878288e-01 8.83069575e-01 -2.35677660e-01
-4.92550097e-02 1.77411526e-01 6.17831588e-01 7.15435803e-01
5.37362933e-01 1.56502455e-01 -1.81489233e-02 1.58986121e-01
4.72930729e-01 3.46968681e-01 6.23480640e-02 -1.92564934e-01
-1.27456093e+00 8.00364196e-01 3.73161584e-01 1.74388722e-01
-5.71319103e-01 -3.08656007e-01 8.99665773e-01 2.64542550e-01
4.03666735e-01 -2.28377700e-01 -2.89725691e-01 -3.93154114e-01
-2.49366105e-01 3.56533200e-01 1.21284425e-01 1.08945239e+00
6.04358017e-01 -5.11241667e-02 -2.66994357e-01 1.09701943e+00
1.47572309e-02 7.87020266e-01 3.20718467e-01 -8.33662689e-01
7.90854812e-01 9.19679821e-01 -1.32861763e-01 -1.39322841e+00
-5.30631781e-01 1.32832482e-01 -9.61630940e-01 2.35861048e-01
5.72459221e-01 -1.86275318e-01 -8.91390443e-01 1.18153346e+00
3.78274560e-01 2.23107308e-01 4.30563428e-02 1.18277240e+00
1.02571797e+00 6.06365502e-01 2.68138558e-01 -3.25485140e-01
1.50995374e+00 -1.03206289e+00 -9.23291445e-01 -1.42178923e-01
7.31356025e-01 -8.50593090e-01 9.85591412e-01 2.14474127e-01
-7.07460940e-01 -8.69246304e-01 -4.68598157e-01 2.08009295e-02
-7.10105002e-01 2.29131296e-01 4.03858840e-01 7.26507127e-01
-4.79649007e-01 -4.39847231e-01 -4.17208493e-01 -9.12907481e-01
8.70875239e-01 -4.57846373e-02 -4.46328968e-01 -2.83476710e-01
-1.08316684e+00 1.17409265e+00 2.92081565e-01 1.69248253e-01
-8.31445813e-01 -3.04188281e-01 -6.66641176e-01 -4.57336992e-01
7.98175812e-01 -5.58419704e-01 1.04097152e+00 -8.22584391e-01
-7.15932310e-01 9.10229921e-01 -6.98465481e-02 -6.56400383e-01
6.83755815e-01 -3.51780728e-02 -8.18812490e-01 2.58687995e-02
3.37322444e-01 4.19663161e-01 6.56836271e-01 -9.09451425e-01
-1.25629687e+00 -4.79027182e-01 1.93518251e-01 -6.20117933e-02
-3.77127767e-01 2.99252748e-01 -6.66420460e-01 -5.05058765e-01
-3.89988899e-01 -1.09108353e+00 -1.87021062e-01 -1.15404636e-01
-3.40212852e-01 -4.18111205e-01 1.29118586e+00 -4.89960074e-01
1.27577364e+00 -2.00223637e+00 -4.46271479e-01 -2.02950597e-01
7.58310780e-02 8.88462543e-01 -1.99683513e-02 3.09914201e-01
1.58985093e-01 -2.37456873e-01 -9.30673331e-02 1.67849123e-01
-2.34071419e-01 3.50538015e-01 -2.26162866e-01 2.66989768e-01
1.27913311e-01 6.43932223e-01 -6.98261321e-01 -7.88833499e-01
8.15256953e-01 1.46538660e-01 -2.45121136e-01 1.07179679e-01
2.13517874e-01 2.82095611e-01 -8.32810342e-01 7.03397214e-01
6.95039213e-01 4.56544191e-01 -6.18842423e-01 -5.27853787e-01
-5.27156472e-01 -2.62985647e-01 -7.83402503e-01 1.09963226e+00
-1.35466814e-01 9.23003912e-01 -1.76026478e-01 -1.22771430e+00
8.46000850e-01 2.25729391e-01 5.76809406e-01 -1.03489375e+00
2.67384738e-01 2.51097735e-02 3.11883874e-02 -1.55243218e+00
3.78842860e-01 2.08787084e-01 1.99437095e-03 -2.48813644e-01
-3.26497823e-01 3.20079088e-01 6.47441745e-01 4.46485206e-02
9.93652403e-01 -1.48812994e-01 -9.99911502e-03 2.00342447e-01
8.30792129e-01 6.88034773e-01 6.04636788e-01 5.69395840e-01
-7.45357096e-01 4.62535739e-01 2.86059648e-01 -8.95666063e-01
-1.16426027e+00 -7.66219914e-01 -3.15007716e-01 8.08824420e-01
1.58765927e-01 -1.20314464e-01 -8.68011236e-01 -7.50970364e-01
-2.93536156e-01 4.94782686e-01 -3.74167591e-01 -2.15472817e-01
-5.52857161e-01 -8.96668375e-01 6.07512951e-01 3.61564845e-01
1.12131929e+00 -1.20719624e+00 -5.10650396e-01 6.37723580e-02
-2.58890331e-01 -1.54436696e+00 -2.71089554e-01 -7.59184361e-01
-3.37568849e-01 -1.67732978e+00 -8.27188969e-01 -9.59152639e-01
5.91299057e-01 1.06568360e+00 6.72181785e-01 -4.65963110e-02
-5.79922080e-01 3.60517740e-01 -2.41578251e-01 -8.99763823e-01
-1.27391368e-01 -2.23860919e-01 1.64424017e-01 4.07466829e-01
1.06719160e+00 -3.60081382e-02 -5.35186350e-01 6.78662658e-01
-6.97956145e-01 4.90057170e-02 6.99876845e-01 4.84330356e-01
1.95658132e-01 2.96704680e-01 7.32127011e-01 -6.13808274e-01
5.59133768e-01 -6.23525202e-01 -6.51539803e-01 7.29728714e-02
-1.46349207e-01 -7.08253741e-01 4.06418979e-01 3.20126244e-04
-1.23664343e+00 -2.87317932e-01 -1.58090189e-01 -5.92568517e-01
-8.18362832e-01 1.70646727e-01 -1.02505021e-01 2.57407129e-01
4.16087449e-01 1.25676662e-01 1.38997793e-01 -2.19487846e-01
3.91243026e-02 1.09467781e+00 5.33543944e-01 -1.40478626e-01
5.58930755e-01 4.15679187e-01 1.33490682e-01 -1.24383020e+00
-7.63107777e-01 -7.45198011e-01 -6.22303426e-01 -7.45095968e-01
1.32325006e+00 -9.02165174e-01 -6.26165152e-01 7.73090839e-01
-1.09801531e+00 3.26927364e-01 1.55900300e-01 8.53316009e-01
-2.68989056e-01 4.87174422e-01 -2.09666878e-01 -9.56217587e-01
-2.42725555e-02 -1.18568897e+00 7.67688453e-01 1.46638796e-01
3.77614826e-01 -5.86962104e-01 -1.11168705e-01 1.09720325e+00
3.69126469e-01 8.80053863e-02 6.19993627e-01 -3.53539526e-01
-7.13031173e-01 -4.69480336e-01 -6.22449696e-01 5.22608280e-01
2.61693507e-01 3.02267671e-01 -8.71703386e-01 2.65698105e-01
-2.05732971e-01 -2.26399630e-01 9.44208622e-01 4.89317507e-01
1.19557476e+00 -1.91927910e-01 -5.14143884e-01 1.07446853e-02
1.03862011e+00 5.80089986e-01 8.12603295e-01 6.46276653e-01
7.16125488e-01 1.00670910e+00 1.04150105e+00 2.84518003e-01
6.05444849e-01 6.77819669e-01 2.99468011e-01 -1.76805064e-01
-2.63292968e-01 2.08177134e-01 3.70697320e-01 8.73116076e-01
-5.22007823e-01 -2.65777826e-01 -1.12935460e+00 7.43977308e-01
-2.12493420e+00 -1.44567454e+00 -9.07314181e-01 1.96952558e+00
7.64789954e-02 3.79265904e-01 6.00718141e-01 2.24681213e-01
9.41473544e-01 1.84101705e-02 -2.16799021e-01 -2.59179056e-01
-7.17588738e-02 -6.23455644e-01 4.83713359e-01 -1.17043227e-01
-1.31757665e+00 7.67691135e-01 5.40768051e+00 1.13643837e+00
-8.88648510e-01 2.19747469e-01 6.37917578e-01 1.02338627e-01
4.73237306e-01 -4.22708571e-01 -8.67476642e-01 7.63627112e-01
9.54442620e-01 1.22543007e-01 -3.07760537e-02 8.36551785e-01
8.17498326e-01 -1.25454634e-01 -4.23668116e-01 1.22541702e+00
1.29438182e-02 -1.10809934e+00 1.51717559e-01 -9.46304128e-02
5.19058943e-01 2.75549769e-01 -5.86705431e-02 4.67017055e-01
-1.47698894e-01 -4.77089524e-01 3.11151117e-01 6.88347995e-01
2.66269088e-01 -8.74159455e-01 1.05207062e+00 5.32035053e-01
-1.13730192e+00 -2.93086886e-01 -7.16042638e-01 -1.37835339e-01
3.02764118e-01 2.57905453e-01 -5.69556057e-01 3.62323731e-01
1.02544010e+00 9.49519873e-01 -7.21975446e-01 1.38508761e+00
3.77116501e-01 7.08340108e-01 -1.01024965e-02 -1.16630904e-01
5.84423125e-01 -4.61710960e-01 5.96652865e-01 1.17640710e+00
2.92901218e-01 1.72686338e-01 3.35728526e-01 3.74076009e-01
3.33815724e-01 6.58502430e-02 -1.38013434e+00 2.93854326e-01
1.65628597e-01 1.36045575e+00 -4.34599280e-01 -4.49153185e-01
-1.22255039e+00 2.55294561e-01 -1.39993206e-01 2.69301087e-01
-1.29621220e+00 -3.87706786e-01 8.68876755e-01 2.69772202e-01
1.58492118e-01 -2.71337628e-01 -7.51957744e-02 -9.46526885e-01
5.38551919e-02 -5.61353862e-01 5.24625003e-01 -8.29287648e-01
-1.39661229e+00 6.01763725e-01 3.47474575e-01 -1.60970604e+00
3.59331101e-01 -8.27675343e-01 -6.73403442e-01 3.55644017e-01
-1.54853904e+00 -1.32815909e+00 -5.57239532e-01 9.54438686e-01
1.15011752e+00 -8.07881415e-01 1.51137322e-01 8.10589671e-01
-1.30949914e+00 2.12775782e-01 -3.61349545e-02 2.55230546e-01
4.77127552e-01 -4.61884677e-01 1.38689160e-01 7.56996274e-01
-4.62293625e-01 1.04969501e-01 4.72528487e-01 -5.56574583e-01
-1.47484469e+00 -1.44907117e+00 6.82516754e-01 -6.87847555e-01
6.74929440e-01 4.25351970e-03 -6.77266181e-01 5.53991854e-01
3.35158199e-01 2.40897574e-02 3.06320667e-01 -2.59492129e-01
1.56976238e-01 -4.96808410e-01 -1.17354798e+00 7.29141653e-01
1.12248886e+00 -1.60963133e-01 -8.05670381e-01 5.44115543e-01
4.50011045e-01 1.63813889e-01 -3.11851889e-01 3.32124382e-01
3.87060642e-01 -1.07458496e+00 1.06273389e+00 -7.90988684e-01
1.37458250e-01 -4.55933422e-01 -1.17302887e-01 -7.81637669e-01
-2.89965332e-01 -2.84495372e-02 3.23173851e-01 1.44450092e+00
7.80852288e-02 -7.52185106e-01 5.29671192e-01 7.79283166e-01
-5.56631207e-01 -4.08557326e-01 -1.01391935e+00 -7.36577928e-01
-2.87255526e-01 -9.00582910e-01 4.33697730e-01 5.98736525e-01
-5.38590908e-01 4.12936717e-01 -6.32983387e-01 1.94781750e-01
6.98058188e-01 -2.04654470e-01 1.21653008e+00 -1.13395751e+00
7.06158161e-01 -5.12438238e-01 -9.58554745e-01 -7.11504936e-01
-1.23169664e-02 -3.31002653e-01 -1.65755913e-01 -1.31647754e+00
2.63580620e-01 -3.21195215e-01 -2.16775358e-01 3.04721445e-01
-7.85741061e-02 4.72956479e-01 -1.47875808e-02 3.27895790e-01
-7.19194174e-01 3.89331490e-01 1.31129670e+00 -4.63632911e-01
2.80383617e-01 8.89294297e-02 -2.65869647e-01 1.02017784e+00
8.13371062e-01 -2.15253517e-01 -5.17601192e-01 -3.79382014e-01
-3.95419806e-01 -1.38215765e-01 7.10592866e-01 -1.30037022e+00
2.48557091e-01 -4.76433992e-01 5.06969355e-02 -9.17769194e-01
4.52882379e-01 -1.22245502e+00 -5.59504293e-02 3.72583777e-01
-1.54441431e-01 9.20085236e-02 9.70678777e-02 6.38312519e-01
-3.72042269e-01 -1.26062781e-01 5.99719167e-01 -1.88746437e-01
-1.37075365e+00 6.27241492e-01 -7.99151540e-01 -9.18817148e-02
1.74517858e+00 -5.00623286e-01 -5.33821225e-01 -1.76964700e-02
-2.06425548e-01 2.89768338e-01 5.65153882e-02 9.97929335e-01
6.41963243e-01 -1.42926741e+00 -7.71435380e-01 3.66678506e-01
5.21183252e-01 -2.03136921e-01 5.36216140e-01 1.04772604e+00
-6.11023843e-01 8.89081120e-01 -6.71420217e-01 -6.74715698e-01
-1.49880826e+00 1.00590456e+00 -2.74782497e-02 3.44254464e-01
-7.13071287e-01 2.89635897e-01 4.02910203e-01 -2.68288046e-01
1.93039358e-01 -3.76259923e-01 -6.61909640e-01 8.28130916e-02
5.94499350e-01 8.13783467e-01 8.99610389e-03 -1.28922880e+00
-2.50791878e-01 8.25802028e-01 5.41661941e-02 5.35817862e-01
1.10916007e+00 -5.19741535e-01 1.01078905e-01 3.40738147e-01
1.09477043e+00 -4.09010112e-01 -1.03231013e+00 -1.81286559e-01
-1.72040612e-01 -9.02437627e-01 1.28180295e-01 -4.22785193e-01
-1.42852604e+00 9.50463533e-01 9.43527520e-01 4.81227010e-01
1.07372022e+00 -1.81750670e-01 1.31365299e+00 7.32646167e-01
5.41221142e-01 -1.18449318e+00 -1.38793781e-01 3.81849915e-01
8.16945016e-01 -1.76875162e+00 -3.89229625e-01 -3.84717524e-01
-1.06287086e+00 1.18775702e+00 1.00709295e+00 7.81534910e-02
4.98212308e-01 -2.33128697e-01 3.12060863e-01 -2.69008666e-01
-6.71147823e-01 -5.84448755e-01 2.61247694e-01 8.06400895e-01
-1.67751517e-02 -1.88140258e-01 -3.96123916e-01 3.43339324e-01
3.31009865e-01 2.00174019e-01 2.59264439e-01 9.50241506e-01
-8.23924363e-01 -7.92994082e-01 -6.07065737e-01 6.96936488e-01
-3.48457068e-01 3.76186699e-01 -1.35541439e-01 8.51881385e-01
6.62218750e-01 1.34988904e+00 -1.81643926e-02 -2.67366022e-01
7.43530631e-01 -2.31457114e-01 -3.50662433e-02 -1.53352180e-02
-1.40511110e-01 -3.24312985e-01 3.94738495e-01 -3.75403851e-01
-6.62230492e-01 -7.12680578e-01 -8.74665260e-01 -4.88506019e-01
-2.53561914e-01 -3.64863798e-02 4.33381855e-01 8.33670616e-01
2.10911006e-01 3.85612518e-01 7.43519843e-01 -7.20670342e-01
1.00742832e-01 -8.95954370e-01 -3.50791752e-01 6.43830836e-01
3.62485766e-01 -9.71920788e-01 2.48310789e-02 1.14058703e-01] | [7.7908782958984375, -0.49544262886047363] |
1ee9a0f3-c7df-41be-80d4-9d09635acab4 | multivariate-multi-frequency-and-multimodal | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chen_Multivariate_Multi-Frequency_and_Multimodal_Rethinking_Graph_Neural_Networks_for_Emotion_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chen_Multivariate_Multi-Frequency_and_Multimodal_Rethinking_Graph_Neural_Networks_for_Emotion_CVPR_2023_paper.pdf | Multivariate, Multi-Frequency and Multimodal: Rethinking Graph Neural Networks for Emotion Recognition in Conversation | Complex relationships of high arity across modality and context dimensions is a critical challenge in the Emotion Recognition in Conversation (ERC) task. Yet, previous works tend to encode multimodal and contextual relationships in a loosely-coupled manner, which may harm relationship modelling. Recently, Graph Neural Networks (GNN) which show advantages in capturing data relations, offer a new solution for ERC. However, existing GNN-based ERC models fail to address some general limits of GNNs, including assuming pairwise formulation and erasing high-frequency signals, which may be trivial for many applications but crucial for the ERC task. In this paper, we propose a GNN-based model that explores multivariate relationships and captures the varying importance of emotion discrepancy and commonality by valuing multi-frequency signals. We empower GNNs to better capture the inherent relationships among utterances and deliver more sufficient multimodal and contextual modelling. Experimental results show that our proposed method outperforms previous state-of-the-art works on two popular multimodal ERC datasets. | ['Heng Tao Shen', 'Shuyuan Zhu', 'Jie Shao', 'Feiyu Chen'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 1.62401229e-01 -7.34133869e-02 7.95318633e-02 -4.81076568e-01
-3.15757573e-01 -2.50334680e-01 6.30391002e-01 2.67817795e-01
-1.88948959e-01 4.91733223e-01 5.80571890e-01 6.46673888e-03
-6.76893830e-01 -4.97046471e-01 -2.30722949e-01 -6.45211399e-01
-4.73074138e-01 2.76736498e-01 -3.85494411e-01 -7.24176288e-01
-1.13071702e-01 2.97915906e-01 -1.55483818e+00 4.53164786e-01
8.74156773e-01 1.04623127e+00 -1.73116550e-01 6.37915611e-01
-4.38621968e-01 8.55204165e-01 -4.86713290e-01 -5.37441730e-01
-3.64493221e-01 -7.00596869e-01 -4.44198251e-01 -1.14736781e-01
-7.64087737e-02 3.05403024e-01 -3.20732296e-01 1.00552845e+00
7.13632226e-01 3.29134107e-01 7.14423776e-01 -1.48251605e+00
-5.07198691e-01 9.91290033e-01 -5.15696049e-01 -5.65962829e-02
6.32105708e-01 -3.93729568e-01 1.09250832e+00 -7.63429761e-01
3.49512756e-01 1.54686570e+00 8.26696455e-01 5.14398158e-01
-1.03800583e+00 -5.59595585e-01 4.93293256e-01 3.93567413e-01
-1.23210680e+00 -3.13577622e-01 1.31955135e+00 4.64162566e-02
9.00807083e-01 6.37113392e-01 6.24894500e-01 1.46729052e+00
-6.73503950e-02 7.93862522e-01 9.96255100e-01 -3.55428278e-01
-6.04649894e-02 -3.77559625e-02 2.32752562e-01 1.72667861e-01
-3.90819371e-01 -4.64485824e-01 -9.06779289e-01 -2.74359941e-01
3.87913465e-01 1.46302059e-01 -5.80031514e-01 -2.40660474e-01
-1.18442142e+00 7.21826196e-01 3.25557530e-01 6.26747370e-01
-5.85484505e-01 -8.28758068e-03 5.52537858e-01 3.53719682e-01
4.59058136e-01 2.78576255e-01 -2.64805853e-01 -5.50017953e-01
-2.90201068e-01 1.13601051e-01 9.53599751e-01 8.74537468e-01
2.82516211e-01 9.23194066e-02 -6.02910183e-02 1.32705212e+00
3.52081239e-01 2.65159786e-01 3.77786994e-01 -6.19690955e-01
6.93522990e-01 8.07667911e-01 -3.27540070e-01 -1.87205040e+00
-8.32706273e-01 -3.64657849e-01 -1.40538275e+00 -7.45574534e-01
1.30872587e-02 -3.06035042e-01 -3.77020687e-01 1.90402365e+00
1.91861942e-01 1.87065601e-01 2.51922160e-01 1.04013407e+00
1.37928116e+00 5.86636364e-01 1.80809572e-01 -3.72771055e-01
1.23883080e+00 -6.16519690e-01 -1.43473840e+00 -3.18448804e-02
4.63313282e-01 -6.60073936e-01 6.16746306e-01 3.91860604e-01
-9.16060090e-01 -3.58338982e-01 -6.60811365e-01 1.53842092e-01
-5.94549954e-01 -2.38964245e-01 1.02509880e+00 5.36199570e-01
-1.06560004e+00 2.59580195e-01 -2.62164176e-01 -4.49164689e-01
-2.21777428e-02 3.01497102e-01 -5.20463586e-01 -1.36343136e-01
-1.76856875e+00 8.36173952e-01 4.21337247e-01 9.56752717e-01
-2.22959340e-01 -2.61693001e-01 -1.02728081e+00 9.90406871e-02
6.72470093e-01 -3.17148417e-01 5.75735092e-01 -6.66259706e-01
-1.39524472e+00 1.69445127e-01 -1.47791393e-02 -1.88337296e-01
6.99170753e-02 -1.96886063e-02 -1.11160648e+00 4.95654345e-02
-5.74960947e-01 4.71193343e-01 4.53621209e-01 -1.60113394e+00
-1.84407830e-01 -3.25719148e-01 4.13175523e-02 6.73082948e-01
-6.87736928e-01 2.73461696e-02 -6.65339708e-01 -5.13133049e-01
3.26735437e-01 -7.45905578e-01 -3.84872742e-02 -6.08758152e-01
-4.80760515e-01 -4.44907516e-01 8.10990214e-01 -5.24187863e-01
1.61643493e+00 -2.21947336e+00 6.48387432e-01 4.81788576e-01
3.00916314e-01 1.06994517e-01 -4.37901586e-01 9.35171902e-01
-8.94303918e-02 8.37578550e-02 -1.03170291e-01 -7.15799272e-01
4.40055639e-01 5.04492223e-01 4.33094054e-02 1.96416944e-01
1.61886916e-01 1.02347171e+00 -7.02300906e-01 -5.92778325e-01
1.19540036e-01 9.50260341e-01 -3.27944994e-01 1.70600861e-01
4.76743095e-02 3.57971936e-01 -1.34796649e-01 6.90439165e-01
7.10688531e-01 2.35192515e-02 7.30721772e-01 -5.00796616e-01
3.37992191e-01 -2.14715198e-01 -1.29786360e+00 1.66724443e+00
-2.69259363e-01 5.94411016e-01 2.33203828e-01 -1.36384559e+00
1.12968826e+00 4.97107029e-01 5.92967927e-01 -7.89998531e-01
4.04905349e-01 -7.70509019e-02 1.62589118e-01 -9.03847277e-01
7.77525544e-01 -1.40607759e-01 -4.14130062e-01 1.06782414e-01
1.86488569e-01 9.81381908e-02 -4.08530002e-03 3.16589206e-01
6.08030319e-01 -1.36627480e-01 1.63509458e-01 5.33827730e-02
6.59503520e-01 -6.14666522e-01 6.34116232e-01 5.53763926e-01
-2.04975531e-01 4.75013018e-01 8.95531595e-01 -1.71670616e-01
-4.04852599e-01 -5.18070698e-01 7.74345323e-02 1.07290542e+00
2.30320290e-01 -6.72023714e-01 -4.35391545e-01 -4.37624454e-01
-1.99620664e-01 5.35736263e-01 -6.45079255e-01 -1.00593328e-01
-3.72567415e-01 -8.07890654e-01 6.91046119e-01 3.57329130e-01
3.00980866e-01 -1.05411923e+00 3.88765708e-02 3.47442359e-01
-8.92518640e-01 -1.38413811e+00 -1.25390097e-01 2.34993652e-01
-4.79083687e-01 -9.05015528e-01 -5.14485717e-01 -6.27369702e-01
2.51077712e-01 1.91702217e-01 1.17939568e+00 5.30712157e-02
6.54040650e-02 7.22627342e-01 -7.07773149e-01 -2.99928993e-01
-7.40558580e-02 2.39626858e-02 -8.84079747e-03 5.01316786e-01
5.38764715e-01 -7.56879032e-01 -3.69268030e-01 2.22674504e-01
-1.00945818e+00 -8.22782964e-02 2.78752655e-01 8.52823138e-01
2.78733939e-01 3.98109764e-01 7.87771881e-01 -7.43384480e-01
1.27500927e+00 -8.32618773e-01 1.67448074e-01 4.48014259e-01
-2.15521768e-01 -2.77764857e-01 5.81266999e-01 -6.75361991e-01
-1.09125924e+00 -4.78181273e-01 -2.43536960e-02 -4.02259201e-01
-2.38625422e-01 1.01571822e+00 -3.16166490e-01 6.84930012e-02
1.46157920e-01 1.83315963e-01 -4.97982241e-02 -2.78275251e-01
4.65325058e-01 6.78528726e-01 3.93212557e-01 -8.03804576e-01
2.48404637e-01 8.54838863e-02 2.19524682e-01 -8.25257480e-01
-4.09869134e-01 -5.01404643e-01 -2.81806618e-01 -5.87188840e-01
8.56058300e-01 -8.43765676e-01 -1.14877486e+00 3.19244891e-01
-1.12158430e+00 9.44186151e-02 2.60169476e-01 5.29719234e-01
-1.95354849e-01 5.12401700e-01 -5.42996705e-01 -1.31700063e+00
-9.96735618e-02 -9.61662352e-01 9.98529255e-01 3.04103166e-01
-3.24750990e-01 -1.29843628e+00 -1.55219406e-01 4.44139957e-01
6.04440510e-01 5.88366151e-01 1.01705706e+00 -5.70285380e-01
-1.77867201e-04 -8.11140165e-02 -2.18521789e-01 1.35755569e-01
5.45508647e-03 -2.34803438e-01 -8.33190322e-01 7.39218369e-02
-1.12761356e-01 -3.57070923e-01 6.43068016e-01 1.11384004e-01
1.07000291e+00 -1.23044588e-01 -8.06212351e-02 2.59684861e-01
1.18543017e+00 2.89871186e-01 7.19676614e-01 -2.51039956e-02
9.03642535e-01 1.13818574e+00 3.43126327e-01 5.32804906e-01
9.18230057e-01 5.98379731e-01 7.47559190e-01 -2.11342528e-01
3.61513317e-01 7.28566051e-02 2.51946803e-02 1.47170246e+00
-3.97600561e-01 -7.44064748e-01 -8.55987549e-01 5.02597570e-01
-2.32058620e+00 -8.96951258e-01 -4.88432050e-01 1.48414302e+00
7.49119580e-01 -3.02595258e-01 -1.17362745e-01 2.12399423e-01
6.18474722e-01 2.96485931e-01 -6.38834462e-02 -5.91135144e-01
-6.15473509e-01 -1.68737948e-01 -1.81767195e-01 2.49793410e-01
-9.26392794e-01 5.86269140e-01 5.70400667e+00 7.25795269e-01
-7.48699546e-01 -1.33460447e-01 3.65392506e-01 6.97818643e-04
-6.85291171e-01 -2.89536923e-01 -3.82398427e-01 3.05922300e-01
7.74681449e-01 3.59429777e-01 6.68188870e-01 3.08487713e-01
5.64580709e-02 4.57908399e-02 -9.21307385e-01 1.48760426e+00
4.55799282e-01 -7.27298498e-01 -2.28764620e-02 -5.12796417e-02
6.39912665e-01 -4.90722686e-01 8.03538784e-02 6.63312614e-01
5.72188534e-02 -1.15167379e+00 2.20563814e-01 1.03861356e+00
2.07231537e-01 -1.05604029e+00 1.17460346e+00 2.53173560e-01
-1.16274655e+00 -2.31427290e-02 -2.04250962e-01 -2.93318778e-02
2.57890165e-01 4.78383183e-01 -2.85895735e-01 1.17654610e+00
7.94151902e-01 7.31824100e-01 -5.92344761e-01 4.76177186e-01
1.96354493e-01 3.55829865e-01 -2.08496839e-01 -3.47213954e-01
2.99997807e-01 -3.51093471e-01 4.87429798e-01 1.41252387e+00
3.13006639e-01 3.47325593e-01 5.83156571e-02 5.23537636e-01
-4.19308543e-02 3.19293082e-01 -6.11965060e-01 -2.97622502e-01
4.19917047e-01 1.31248009e+00 -4.63627815e-01 -3.15479515e-03
-4.64926779e-01 7.86249697e-01 4.25591260e-01 6.46349192e-01
-8.89053762e-01 -2.93056160e-01 4.74052250e-01 -6.58614993e-01
1.49590984e-01 -2.81470448e-01 7.50531256e-02 -1.16615546e+00
5.06373383e-02 -1.16282582e+00 5.73961437e-01 -8.07413697e-01
-1.68504417e+00 7.42030442e-01 5.23798680e-03 -1.22221911e+00
-3.16685766e-01 -3.94546211e-01 -4.10527289e-01 7.55069196e-01
-1.47944236e+00 -1.39559889e+00 -4.11586732e-01 9.14210200e-01
1.59898832e-01 -1.52032226e-01 1.08496404e+00 4.00456667e-01
-5.49192965e-01 6.23941720e-01 -2.02744201e-01 -9.73619148e-02
6.84644103e-01 -1.13921940e+00 -4.46862310e-01 5.22449613e-01
4.17594276e-02 9.10638452e-01 7.13120461e-01 -4.09899235e-01
-1.77985954e+00 -6.47905350e-01 1.14261281e+00 -1.00943081e-01
7.53826976e-01 -5.94201446e-01 -1.04369521e+00 3.08719367e-01
5.27303219e-01 -1.66072175e-01 1.13354993e+00 6.82306588e-01
-4.10029173e-01 -4.03943025e-02 -8.62193048e-01 7.25210249e-01
1.12599528e+00 -8.17942500e-01 -4.27259982e-01 -8.46126527e-02
6.31821573e-01 -3.29202741e-01 -1.23430836e+00 6.05565608e-01
6.05648577e-01 -1.06962800e+00 9.10678446e-01 -5.11605084e-01
4.07193422e-01 -1.70891017e-01 -6.11647964e-01 -1.44142878e+00
-2.91649140e-02 -6.19035661e-01 -3.58002514e-01 1.70026100e+00
2.89867580e-01 -5.44811189e-01 2.34060526e-01 6.44163132e-01
9.95181799e-02 -7.41732836e-01 -8.17696512e-01 -3.34878296e-01
-3.65847826e-01 -8.63739312e-01 7.77580500e-01 1.38730526e+00
4.49731737e-01 4.48720694e-01 -9.06098962e-01 1.80708319e-01
3.97756398e-01 -1.28218919e-01 5.70870876e-01 -1.24577200e+00
-1.96389541e-01 -5.93813717e-01 -5.16257584e-01 -6.48894429e-01
3.02847654e-01 -6.19986355e-01 -4.68188673e-02 -1.54701591e+00
6.50110766e-02 -2.90195495e-01 -5.73790073e-01 1.97276250e-01
-3.04653764e-01 1.96414366e-01 2.94695377e-01 -9.69612375e-02
-9.24300432e-01 8.88203442e-01 1.11139655e+00 -1.93591088e-01
-2.01743096e-01 -4.55196410e-01 -8.11032593e-01 4.45083827e-01
6.47889018e-01 -3.83749157e-02 -6.20135725e-01 -3.24885994e-01
6.24049664e-01 3.45832616e-01 3.87031764e-01 -6.33190393e-01
4.35588211e-01 -7.86555558e-02 2.64540076e-01 -7.69614458e-01
6.50165915e-01 -1.22762644e+00 4.39191878e-01 -3.83566588e-01
-4.05067176e-01 7.28855431e-02 2.25759283e-01 7.61453629e-01
-6.97693527e-01 1.11535214e-01 1.12603558e-02 1.30396888e-01
-7.39089727e-01 8.90689343e-02 -2.50857353e-01 -1.52591497e-01
4.78748530e-01 3.23044397e-02 -2.36011103e-01 -9.71900702e-01
-9.10099387e-01 4.85800803e-01 -2.50579238e-01 8.07948053e-01
7.23702431e-01 -1.72142148e+00 -4.48482513e-01 1.67720933e-02
3.93921822e-01 -2.00781420e-01 9.06248093e-01 1.02351093e+00
2.57973701e-01 3.31485480e-01 5.61143877e-03 -6.46065176e-01
-1.39390337e+00 3.63281101e-01 2.46846765e-01 -3.12214822e-01
-2.43822560e-01 7.36116290e-01 -6.78980574e-02 -8.02268922e-01
5.31138122e-01 -2.38900837e-02 -9.28626657e-01 4.59528506e-01
1.54932901e-01 1.12025820e-01 -1.40520230e-01 -1.01360834e+00
-4.79240209e-01 5.67263782e-01 1.81583479e-01 1.31556932e-02
1.39182615e+00 -4.29476440e-01 -4.36299026e-01 7.43450344e-01
1.24173868e+00 -3.14814329e-01 -7.43051350e-01 -5.08176029e-01
-1.08477987e-01 -8.30178857e-02 -6.64181113e-02 -7.93713570e-01
-9.51025784e-01 7.67187774e-01 3.80653024e-01 5.79481125e-01
1.42483354e+00 -5.35471924e-02 5.14501750e-01 3.30871403e-01
1.64104998e-01 -1.27002990e+00 1.26633495e-01 6.50683343e-01
1.09838879e+00 -1.27007878e+00 -2.10033596e-01 -4.68260288e-01
-9.67795134e-01 1.29217052e+00 5.24552405e-01 4.61001396e-01
6.68431103e-01 1.11043915e-01 1.20300867e-01 -5.18578172e-01
-6.89342976e-01 -2.77033687e-01 4.99978662e-01 6.87994957e-01
5.47221780e-01 2.34281793e-01 -4.32261944e-01 8.65457416e-01
-5.71382940e-02 -4.63973850e-01 1.24302298e-01 7.55748630e-01
2.56365776e-01 -1.07570815e+00 -3.19299757e-01 1.61987633e-01
-3.07595104e-01 -1.50602192e-01 -5.49700558e-01 8.88405561e-01
2.72041738e-01 1.34279740e+00 -1.63396075e-01 -6.90488935e-01
4.68773752e-01 1.72498688e-01 1.99979872e-01 3.63132879e-02
-6.09916449e-01 3.13138574e-01 4.89303261e-01 -4.70957190e-01
-1.10711861e+00 -4.32828188e-01 -1.02019215e+00 -2.38244027e-01
-3.15637201e-01 1.65673509e-01 7.83053398e-01 9.41879988e-01
3.72856796e-01 8.44752252e-01 5.61274230e-01 -7.15554416e-01
1.36135323e-02 -1.23740685e+00 -6.99438095e-01 7.62306929e-01
2.21551657e-01 -7.49817312e-01 -3.29654872e-01 -2.37979069e-01] | [13.14797306060791, 5.271522045135498] |
53acfa5d-985c-42a5-b67c-4280a9733b3f | autotrigger-named-entity-recognition-with | 2109.04726 | null | https://arxiv.org/abs/2109.04726v3 | https://arxiv.org/pdf/2109.04726v3.pdf | AutoTriggER: Label-Efficient and Robust Named Entity Recognition with Auxiliary Trigger Extraction | Deep neural models for named entity recognition (NER) have shown impressive results in overcoming label scarcity and generalizing to unseen entities by leveraging distant supervision and auxiliary information such as explanations. However, the costs of acquiring such additional information are generally prohibitive. In this paper, we present a novel two-stage framework (AutoTriggER) to improve NER performance by automatically generating and leveraging ``entity triggers'' which are human-readable cues in the text that help guide the model to make better decisions. Our framework leverages post-hoc explanation to generate rationales and strengthens a model's prior knowledge using an embedding interpolation technique. This approach allows models to exploit triggers to infer entity boundaries and types instead of solely memorizing the entity words themselves. Through experiments on three well-studied NER datasets, AutoTriggER shows strong label-efficiency, is capable of generalizing to unseen entities, and outperforms the RoBERTa-CRF baseline by nearly 0.5 F1 points on average. | ['Jay Pujara', 'Fred Morstatter', 'Xiang Ren', 'James Allan', 'Elizabeth Boschee', 'Bill Yuchen Lin', 'Sheikh Muhammad Sarwar', 'Ravi Kiran Selvam', 'Dong-Ho Lee'] | 2021-09-10 | null | null | null | null | ['low-resource-named-entity-recognition'] | ['natural-language-processing'] | [ 6.82501346e-02 6.52441084e-01 -1.61543131e-01 -7.65181601e-01
-9.90319848e-01 -8.47052395e-01 6.35084033e-01 6.42304242e-01
-8.38557661e-01 9.83245254e-01 4.82112288e-01 -3.85389209e-01
3.05512697e-01 -7.32332885e-01 -8.62139702e-01 -2.05290869e-01
1.22946806e-01 5.28491795e-01 -2.92948812e-01 1.93295404e-02
4.02560970e-03 1.75310999e-01 -1.18021989e+00 1.78820252e-01
1.13784862e+00 4.33454335e-01 1.61350489e-01 4.57521975e-01
-2.62007028e-01 6.53271019e-01 -5.82309246e-01 -7.85723686e-01
-7.91666061e-02 -1.48868516e-01 -9.99454260e-01 -2.13293433e-01
5.13320863e-01 -3.00330698e-01 -1.45645708e-01 6.33014977e-01
4.15887058e-01 4.19731379e-01 5.96935213e-01 -7.32830882e-01
-1.32189047e+00 1.11661208e+00 -1.89356714e-01 2.15145230e-01
2.10796203e-02 1.97597570e-03 1.35320270e+00 -1.10322106e+00
6.89205825e-01 8.01567316e-01 8.15395594e-01 1.05061102e+00
-1.30718195e+00 -5.83243728e-01 2.89446622e-01 -8.39726701e-02
-1.12644231e+00 -5.71294606e-01 3.72027069e-01 -2.58611917e-01
1.19276154e+00 1.50331661e-01 2.69251708e-02 1.24647212e+00
-3.15089881e-01 7.86855817e-01 9.51220930e-01 -5.25630713e-01
2.46762648e-01 3.79446000e-01 4.05795604e-01 8.18268895e-01
4.12521958e-01 4.39586155e-02 -4.41697806e-01 -8.77355486e-02
5.24992764e-01 -1.15876123e-01 -2.53814608e-01 2.50030935e-01
-1.04763436e+00 7.90825188e-01 9.20836210e-01 2.50154793e-01
-7.00634658e-01 1.03114665e-01 8.02823231e-02 -2.92978883e-01
5.15534043e-01 1.00191617e+00 -8.61078143e-01 1.65667847e-01
-8.96373212e-01 -4.04441496e-03 7.97051430e-01 8.87508929e-01
9.08499122e-01 2.95064002e-02 -3.77488971e-01 8.53129208e-01
3.59453231e-01 3.55090052e-01 4.34758365e-01 -7.34900594e-01
5.33708513e-01 5.61976552e-01 3.43403906e-01 -5.45152009e-01
-4.88384664e-01 -7.18786061e-01 -5.54290175e-01 -2.98034281e-01
2.73301065e-01 -5.96100688e-01 -1.31592155e+00 2.08017302e+00
4.13300306e-01 3.10350746e-01 3.60221595e-01 6.14214063e-01
1.09147513e+00 6.56102479e-01 8.34931493e-01 3.21265727e-01
1.44858038e+00 -9.88440394e-01 -5.41211188e-01 -5.83296895e-01
8.96234095e-01 -3.77183944e-01 9.88607943e-01 -2.18026072e-01
-7.13780105e-01 -4.24460351e-01 -8.07157099e-01 -5.41255891e-01
-6.44984305e-01 3.08199108e-01 7.77543604e-01 4.08656538e-01
-1.13484323e+00 7.54967213e-01 -8.28206360e-01 -2.74797559e-01
5.15897334e-01 4.50583667e-01 -6.12776458e-01 -9.36253294e-02
-1.34996092e+00 1.03468537e+00 7.82366037e-01 2.14259461e-01
-6.37369573e-01 -9.62024391e-01 -1.21883655e+00 3.74249279e-01
2.06518471e-01 -8.84626508e-01 1.22885764e+00 -4.92459089e-01
-1.14906681e+00 7.08197176e-01 -4.62796867e-01 -5.54757655e-01
1.58166721e-01 -4.96286422e-01 -5.11966646e-01 -1.80123627e-01
3.22982937e-01 1.15804040e+00 2.62883574e-01 -1.26102901e+00
-4.86762434e-01 -7.39822015e-02 2.29965374e-01 7.37960637e-02
-4.29580390e-01 -2.08862767e-01 -1.98309228e-01 -6.12746179e-01
-2.53573477e-01 -9.13734257e-01 -5.98980129e-01 -5.47503591e-01
-9.10443544e-01 -5.26115894e-01 2.22308248e-01 -8.89280260e-01
1.11613214e+00 -1.87378204e+00 -1.75394773e-01 2.59047132e-02
3.48129869e-01 3.20017606e-01 -2.26450384e-01 2.39872620e-01
-6.72468264e-03 6.14153326e-01 -3.19221795e-01 -5.44716120e-01
2.21485212e-01 3.62253666e-01 -4.88425642e-01 -1.68265775e-01
6.46718442e-01 1.15845132e+00 -1.11048949e+00 -3.48098665e-01
2.33567562e-02 7.49793649e-01 -6.72080398e-01 2.52585590e-01
-3.25171828e-01 4.42966133e-01 -4.29084241e-01 4.09736633e-01
2.92893887e-01 -6.74929440e-01 2.29060590e-01 -1.13101140e-01
-3.35397734e-03 8.57023120e-01 -7.92973220e-01 1.59911454e+00
-7.96104670e-01 5.85691929e-01 -3.46105456e-01 -4.79015738e-01
8.76844347e-01 4.52286839e-01 -8.28820616e-02 -2.25985363e-01
-1.40327811e-02 1.82956174e-01 -4.16801721e-01 -2.66436368e-01
8.11642349e-01 -1.61945537e-01 -1.92680627e-01 4.30187464e-01
2.50039935e-01 4.95475680e-01 8.17715600e-02 3.67142588e-01
1.27672172e+00 1.18974581e-01 2.62674570e-01 -9.12835896e-02
1.95937023e-01 1.04552008e-01 7.59145319e-01 8.17743182e-01
6.61032125e-02 3.06140780e-01 -7.46028721e-02 -3.92946959e-01
-9.88577664e-01 -1.06242502e+00 -3.83236557e-02 1.08074510e+00
-1.41722217e-01 -5.16564965e-01 -6.81199372e-01 -1.31128395e+00
8.74517411e-02 1.40196323e+00 -8.22794199e-01 4.22621891e-02
-5.43202281e-01 -7.18054235e-01 6.01677001e-01 1.10333979e+00
4.76082742e-01 -1.18597448e+00 -1.18443780e-01 4.40956801e-01
-3.53711665e-01 -1.24224234e+00 -5.17133415e-01 5.21456122e-01
-8.65938365e-01 -5.76322854e-01 -5.48041940e-01 -7.60009110e-01
1.18200731e+00 -2.45202214e-01 1.38200283e+00 1.97165400e-01
-2.51744427e-02 4.14395556e-02 -1.69579521e-01 -2.94179797e-01
-2.46570677e-01 4.75597918e-01 -1.70991108e-01 -3.31421584e-01
6.02510512e-01 -3.26830208e-01 -4.54903245e-01 5.82621396e-02
-8.36436570e-01 -1.88793745e-02 8.57248187e-01 9.73291039e-01
5.21686077e-01 -2.19587684e-01 7.76665211e-01 -1.51837671e+00
3.60649437e-01 -7.38317013e-01 -2.69118994e-01 3.41452718e-01
-8.83845627e-01 6.91014707e-01 7.58792937e-01 -2.33640224e-01
-1.45136189e+00 1.88927516e-01 -2.59946138e-01 5.60837761e-02
-4.62106287e-01 5.79596877e-01 -1.70913458e-01 4.55510378e-01
7.60610044e-01 -1.39442608e-01 -8.78488719e-01 -7.71388948e-01
7.18549013e-01 5.54877102e-01 8.95215273e-01 -7.91670978e-01
1.00032854e+00 2.41737410e-01 -4.69537318e-01 -5.32542050e-01
-1.61022818e+00 -4.03907090e-01 -6.24582231e-01 3.19776148e-01
9.19305086e-01 -1.07712007e+00 -3.63534808e-01 -1.25775427e-01
-1.36925042e+00 -3.12199622e-01 -2.86169529e-01 4.83392149e-01
1.30179048e-01 6.09009229e-02 -8.73701513e-01 -4.70601588e-01
-4.45269346e-01 -5.54786682e-01 9.59283948e-01 6.90348327e-01
-4.58271027e-01 -1.31529117e+00 9.11079645e-02 5.55100739e-01
3.94547969e-01 2.71830469e-01 9.32783842e-01 -1.29679513e+00
-5.64085245e-01 -2.36491188e-01 -2.64955431e-01 7.82877132e-02
3.04914061e-02 -1.74869806e-01 -1.20974672e+00 5.20252436e-02
-5.37827313e-01 -2.32821539e-01 1.00670111e+00 -2.65597440e-02
9.71993685e-01 -6.72787666e-01 -5.37264884e-01 5.13147831e-01
1.36129999e+00 -3.50344032e-01 3.71277452e-01 4.56021667e-01
8.11719418e-01 6.31859899e-01 2.32366562e-01 1.77121669e-01
9.04369652e-01 3.28153849e-01 2.19603226e-01 -3.53425056e-01
-7.54350573e-02 -8.31090271e-01 1.10524699e-01 6.68703556e-01
-3.41310389e-02 -5.78251839e-01 -8.42724502e-01 9.01524663e-01
-1.56256449e+00 -8.23755026e-01 -6.68843091e-02 1.94620001e+00
1.18339491e+00 1.04207778e-02 -4.18578982e-01 -4.81582046e-01
9.65920091e-01 -6.59352541e-02 -7.07487345e-01 -2.69391030e-01
-9.45651084e-02 3.67314816e-01 6.63629353e-01 6.24774575e-01
-1.16591263e+00 1.29876745e+00 5.83733225e+00 3.38888735e-01
-9.16831613e-01 8.57311115e-02 5.40052176e-01 3.61385584e-01
-7.74133563e-01 1.81438357e-01 -1.23521376e+00 2.55262643e-01
1.16797364e+00 -4.67017107e-03 2.75223106e-01 8.15051973e-01
-1.19009078e-01 3.87214899e-01 -1.21332622e+00 3.83750319e-01
-1.79056048e-01 -1.37774944e+00 1.51102424e-01 5.83654381e-02
8.83838654e-01 2.94798762e-01 -6.19110577e-02 5.99858582e-01
1.06939971e+00 -1.08593416e+00 4.46674109e-01 3.55498135e-01
7.58797288e-01 -4.99747992e-01 8.79424572e-01 2.57564813e-01
-1.01565552e+00 1.02283776e-01 -2.89834738e-01 1.82919338e-01
3.38456482e-01 6.79722309e-01 -1.47507107e+00 4.86135215e-01
3.77059102e-01 5.75780094e-01 -5.87464154e-01 8.36488783e-01
-1.09131801e+00 8.05658519e-01 -4.58418310e-01 6.18408807e-03
3.03592026e-01 4.45923448e-01 1.88514516e-01 1.43390489e+00
2.05663800e-01 3.15652281e-01 -2.71887444e-02 1.14456344e+00
-7.73678899e-01 9.30733979e-02 -3.68638754e-01 -3.77204895e-01
9.38441992e-01 1.48222482e+00 -6.41205370e-01 -5.18988073e-01
-1.75948501e-01 1.08790457e+00 8.15666199e-01 4.02704716e-01
-8.61867130e-01 -5.94301045e-01 3.26773107e-01 -2.65146941e-01
4.97615725e-01 -1.37880608e-01 -4.93651807e-01 -1.40770257e+00
-9.85748246e-02 -3.00710499e-01 6.33155048e-01 -7.70607948e-01
-1.36252785e+00 8.74011457e-01 -3.90525281e-01 -5.07745206e-01
-3.59776974e-01 -4.84597385e-01 -6.86742425e-01 9.15935040e-01
-1.89972198e+00 -1.26142883e+00 -2.08217427e-01 2.67942607e-01
4.43687677e-01 3.04015696e-01 1.06593740e+00 2.41237864e-01
-7.45101988e-01 8.03977013e-01 -7.43914470e-02 5.43022692e-01
7.97466934e-01 -1.77362406e+00 7.91788578e-01 9.69396830e-01
5.48134327e-01 1.14000130e+00 7.33631313e-01 -7.83787549e-01
-7.26595104e-01 -1.29418659e+00 1.67550993e+00 -9.08299148e-01
4.54123855e-01 -3.59761715e-01 -1.04567945e+00 1.05367243e+00
2.48575881e-01 2.87114736e-02 1.08790362e+00 6.96964920e-01
-7.18097627e-01 2.91307390e-01 -1.15013993e+00 4.89389837e-01
1.00250077e+00 -6.45263195e-01 -9.90182102e-01 2.58660585e-01
8.73108149e-01 -3.38116437e-01 -9.54292774e-01 2.46009111e-01
1.77735925e-01 -3.28241616e-01 8.90520215e-01 -1.00872838e+00
4.91652846e-01 -3.34803522e-01 7.90660530e-02 -1.49747944e+00
-3.45682472e-01 -5.64405382e-01 -1.30367756e-01 1.72165251e+00
1.11849701e+00 -5.72142661e-01 7.04855323e-01 1.23921680e+00
-3.30272526e-01 -5.00271618e-01 -3.94201458e-01 -6.35076880e-01
-4.50952090e-02 -4.09818962e-02 8.01318288e-01 1.32249784e+00
-2.70217080e-02 6.54423118e-01 -1.58308491e-01 6.17888153e-01
5.68141341e-01 -1.02118097e-01 4.81869042e-01 -1.30142844e+00
-2.30628088e-01 9.17696953e-02 7.00315386e-02 -1.11571109e+00
5.88702619e-01 -1.24728048e+00 3.80763382e-01 -1.78489637e+00
1.52749732e-01 -8.54425371e-01 -5.90898216e-01 1.22101796e+00
-7.38231122e-01 1.62377268e-01 1.44040128e-02 7.71273896e-02
-7.38876939e-01 3.96789938e-01 6.55355692e-01 1.04577310e-01
-9.29875299e-02 -3.70011568e-01 -1.08836913e+00 6.93189859e-01
6.81908429e-01 -7.71185338e-01 -6.36389852e-02 -9.48272645e-01
2.10048035e-01 -2.58017629e-01 3.79591018e-01 -6.27823472e-01
3.17765564e-01 1.81810752e-01 6.12533748e-01 -3.27723235e-01
6.85932264e-02 -5.21413982e-01 3.73462178e-02 3.33322994e-02
-8.30534935e-01 9.14729461e-02 2.81185240e-01 8.27809751e-01
1.06930209e-03 -3.28407854e-01 3.96079719e-01 -9.41772684e-02
-8.14310133e-01 2.37095967e-01 1.76403895e-02 4.34064388e-01
5.85494220e-01 1.65315762e-01 -6.51769757e-01 -1.26797035e-01
-8.14210951e-01 3.97178203e-01 5.05392849e-01 3.69584352e-01
2.18246266e-01 -1.16402411e+00 -7.39914238e-01 -5.35054058e-02
7.34667331e-02 1.26363216e-02 2.78584063e-01 2.70808786e-01
-6.99034333e-02 4.95125353e-01 1.29381180e-01 -2.50565648e-01
-8.52266729e-01 4.09440547e-01 6.71539530e-02 -5.25817037e-01
-5.58176756e-01 1.11729634e+00 1.53303266e-01 -7.68986285e-01
-2.94673685e-02 -2.26971865e-01 -2.49220863e-01 -1.90428972e-01
4.30073798e-01 1.35135219e-01 5.04195914e-02 -4.22907323e-01
-2.97637284e-01 5.59217334e-02 -4.60980952e-01 -1.36834100e-01
1.64884472e+00 -8.22133645e-02 2.57260144e-01 4.15566489e-02
9.47056651e-01 4.25225496e-01 -1.31914103e+00 -4.28007007e-01
6.56336784e-01 -7.06865489e-02 3.96663547e-02 -1.39747345e+00
-7.92265773e-01 8.56020629e-01 -6.59935325e-02 6.77300245e-02
6.85569584e-01 1.51077762e-01 1.19809949e+00 6.35617733e-01
1.12404190e-01 -8.42923403e-01 -3.11751693e-01 5.61386347e-01
3.68639916e-01 -1.30124700e+00 -3.87483746e-01 -2.41142437e-01
-6.88989580e-01 1.03455770e+00 7.01714933e-01 1.18882947e-01
2.67801404e-01 9.92506221e-02 2.64216512e-01 -1.77651972e-01
-7.82776773e-01 -2.50175148e-01 4.79114145e-01 4.57615644e-01
6.97745085e-01 1.34705812e-01 4.48634773e-02 9.36236501e-01
-2.84719735e-01 -2.16363177e-01 4.34990406e-01 6.11063540e-01
-4.07591552e-01 -1.20983338e+00 5.03457934e-02 3.48275661e-01
-7.09634721e-01 -6.38721406e-01 -4.07581627e-01 6.44937992e-01
1.28701432e-02 9.61619437e-01 -1.39309645e-01 -1.76930159e-01
1.97551876e-01 3.81252170e-01 -2.13340130e-02 -1.06689954e+00
-7.15808332e-01 -4.98159707e-01 5.42528629e-01 -2.65888840e-01
-2.25748584e-01 -4.71585810e-01 -1.68404877e+00 -5.77159524e-02
-6.35615230e-01 5.91081202e-01 8.03251207e-01 1.18929005e+00
9.14520264e-01 4.15487170e-01 3.42937142e-01 -3.34983081e-01
-5.43332875e-01 -7.79903948e-01 -1.00572959e-01 4.66674685e-01
2.12315947e-01 -4.43933517e-01 -3.88636827e-01 1.51328757e-01] | [9.70616626739502, 9.460282325744629] |
002cb121-b557-43ea-bdeb-054f93bc320a | fuzzy-jets | 1509.02216 | null | http://arxiv.org/abs/1509.02216v1 | http://arxiv.org/pdf/1509.02216v1.pdf | Fuzzy Jets | Collimated streams of particles produced in high energy physics experiments
are organized using clustering algorithms to form jets. To construct jets, the
experimental collaborations based at the Large Hadron Collider (LHC) primarily
use agglomerative hierarchical clustering schemes known as sequential
recombination. We propose a new class of algorithms for clustering jets that
use infrared and collinear safe mixture models. These new algorithms, known as
fuzzy jets, are clustered using maximum likelihood techniques and can
dynamically determine various properties of jets like their size. We show that
the fuzzy jet size adds additional information to conventional jet tagging
variables. Furthermore, we study the impact of pileup and show that with some
slight modifications to the algorithm, fuzzy jets can be stable up to high
pileup interaction multiplicities. | ['Lester Mackey', 'Benjamin Nachman', 'Conrad Stansbury', 'Ariel Schwartzman'] | 2015-09-07 | null | null | null | null | ['jet-tagging'] | ['graphs'] | [-6.97574675e-01 -3.04272324e-01 -9.69031602e-02 -2.23982900e-01
-3.17161709e-01 -8.74296486e-01 8.93440485e-01 5.17152965e-01
-4.49456006e-01 6.16865635e-01 7.88066164e-02 -3.05842996e-01
-2.24241704e-01 -8.72451782e-01 -1.75126642e-01 -9.46435571e-01
-7.81475753e-02 1.59047735e+00 1.06809378e+00 2.23772570e-01
1.82242185e-01 1.03344953e+00 -1.01912272e+00 5.35499513e-01
4.23460007e-01 5.26016653e-01 1.33796930e-01 1.16996801e+00
-1.79136768e-01 5.22969961e-01 -4.34849709e-01 -6.22536898e-01
5.16305268e-01 -8.30881774e-01 -2.87272513e-01 -4.03990626e-01
-7.09842622e-01 4.17785048e-01 9.03740432e-03 9.79508579e-01
5.24990320e-01 3.69179249e-01 8.48454475e-01 -1.41718864e+00
5.79738319e-02 8.19714248e-01 -8.91290978e-02 5.01137853e-01
-4.87546414e-01 -2.45788828e-01 6.35955691e-01 -8.68569195e-01
1.06202626e+00 1.47412491e+00 2.96015590e-01 8.93250108e-02
-1.04887235e+00 -7.69862294e-01 -6.84577823e-01 -1.12391844e-01
-8.26872647e-01 -3.48810434e-01 5.90543091e-01 -6.23660862e-01
1.01937234e+00 4.79257435e-01 5.64673543e-01 4.73850071e-01
6.68114364e-01 1.85339287e-01 1.11188173e+00 -7.27948546e-01
5.18857479e-01 4.58139092e-01 -5.01227789e-02 4.86695439e-01
6.61285400e-01 4.81502205e-01 -2.68821359e-01 -5.41022897e-01
6.97357297e-01 -3.57365698e-01 4.49101120e-01 -9.73189473e-01
-1.20527887e+00 1.06833172e+00 5.36823347e-02 6.12753451e-01
2.87072420e-01 -1.18396096e-02 5.14310598e-01 -2.88932234e-01
4.27913129e-01 8.14053595e-01 -4.73736197e-01 4.87098172e-02
-5.72538376e-01 3.32292199e-01 8.19610775e-01 9.30360317e-01
1.93774223e-01 -3.52854997e-01 -1.08042255e-01 5.39991558e-01
3.81203294e-01 3.86276156e-01 1.66397378e-01 -8.01622927e-01
2.67928720e-01 -2.78481022e-02 3.81820649e-01 -4.48477328e-01
-6.85393155e-01 -3.16013426e-01 -5.98953485e-01 4.00370419e-01
3.23647112e-01 -5.41252673e-01 -7.62491345e-01 8.44799519e-01
6.86944902e-01 -3.34812135e-01 1.71826303e-01 6.50767922e-01
5.45686066e-01 6.12714350e-01 2.98553675e-01 -6.82274044e-01
1.00576556e+00 -1.00591719e+00 -4.19608563e-01 4.68582362e-01
1.77458107e-01 -1.24383926e+00 -2.45047718e-01 5.12618005e-01
-1.37123549e+00 -7.00801611e-01 -6.19280875e-01 3.23344707e-01
6.74155504e-02 -1.57803640e-01 1.35360539e+00 9.99354184e-01
-7.34927475e-01 8.27422917e-01 -1.29496658e+00 -4.03023332e-01
-3.59388173e-01 3.39165002e-01 1.99247301e-01 5.81889987e-01
-4.13333982e-01 4.68444020e-01 8.08346093e-01 -3.85383606e-01
-1.50016740e-01 -4.09024328e-01 -1.16998687e-01 3.47344540e-02
-2.69692149e-02 -8.32591832e-01 1.44659507e+00 -1.28836393e-01
-1.38217711e+00 2.59712994e-01 -3.13131660e-01 -4.08035815e-01
2.33538136e-01 1.20233588e-01 -6.42018795e-01 4.80497539e-01
1.79473460e-02 3.62252593e-01 2.52247006e-01 -1.51477015e+00
-8.44184518e-01 6.92373365e-02 -7.45745420e-01 -2.53320821e-02
5.36637664e-01 1.25681973e+00 -2.62107641e-01 -4.56228197e-01
6.39699042e-01 -1.09199357e+00 -8.42728436e-01 -1.12637722e+00
-5.14420271e-01 -2.84868479e-01 3.20158631e-01 -4.05993760e-02
4.17326719e-01 -1.76028883e+00 2.91789919e-01 3.42709422e-01
3.89868915e-01 -1.77538440e-01 5.03815532e-01 6.85201466e-01
-2.10401323e-02 3.85288358e-01 3.58078122e-01 -6.39387742e-02
1.45646513e-01 -4.13650414e-03 1.58334244e-02 8.76634195e-02
-2.16080591e-01 6.94010317e-01 -1.02495909e+00 -4.41734254e-01
3.22330058e-01 -5.78662306e-02 -6.62746549e-01 1.24263950e-01
-9.59708393e-02 7.43537247e-01 -4.37259048e-01 1.61021963e-01
9.44406271e-01 4.59055513e-01 3.36812317e-01 -1.38348907e-01
-9.11699295e-01 3.32485996e-02 -8.76155853e-01 8.40062141e-01
2.53254116e-01 7.48127848e-02 4.14086103e-01 -5.26348948e-01
8.83075237e-01 2.04065964e-01 8.23754787e-01 -3.66252601e-01
2.07579792e-01 9.33403969e-02 2.91660160e-01 -7.76022747e-02
9.76580679e-01 -9.30143535e-01 -2.68701404e-01 1.47647679e-01
2.81481862e-01 -3.11321944e-01 4.56618518e-01 6.45282745e-01
1.13892210e+00 -4.16856229e-01 -8.49446841e-03 -2.04022571e-01
-3.86938542e-01 3.72589052e-01 8.04099560e-01 1.04144871e+00
1.69925302e-01 3.91238123e-01 5.49397469e-01 -4.84175272e-02
-1.50971997e+00 -1.52189910e+00 -4.80696827e-01 7.56248057e-01
3.10390204e-01 -9.06354785e-01 -5.52294403e-02 -5.85808992e-01
-1.83306113e-01 7.78702915e-01 2.40520015e-01 -9.61763225e-03
-4.55635220e-01 -1.60480428e+00 1.08699799e-01 5.03538966e-01
-4.30566855e-02 -1.10052288e+00 -1.19611174e-01 2.08756402e-01
4.44761932e-01 -1.20874572e+00 2.92477190e-01 7.03345835e-01
-7.41767049e-01 -9.26762640e-01 1.85390949e-01 -4.19269562e-01
4.35282230e-01 1.14341736e-01 1.05041003e+00 -1.98799327e-01
-3.39330852e-01 1.63772672e-01 -9.30830777e-01 -6.80546820e-01
-1.12426782e+00 -4.84351248e-01 4.00283247e-01 -5.06071091e-01
2.00582564e-01 -2.73496836e-01 -6.39853850e-02 3.03677946e-01
-4.38982785e-01 -1.44151570e-02 5.93275547e-01 3.72441590e-01
4.84334886e-01 1.09743655e+00 1.02419250e-01 -7.44488537e-01
1.56898916e-01 -5.30215204e-01 -9.36460376e-01 -2.41693944e-01
-1.32130831e-01 1.75088525e-01 3.92737776e-01 -1.94032162e-01
-1.31457484e+00 3.44273821e-02 -2.01514270e-02 -2.02752545e-01
-4.37186897e-01 -2.81853795e-01 -3.10409576e-01 -3.73420417e-01
1.85174182e-01 -5.68603814e-01 -7.98087120e-01 -7.97915041e-01
4.14901614e-01 6.74815625e-02 5.79116523e-01 -5.79282939e-01
1.17713988e+00 3.28396708e-01 3.11478555e-01 -6.01683736e-01
-4.83491868e-01 -9.17949677e-01 -1.19126332e+00 -1.29297644e-01
1.35001802e+00 -4.94138658e-01 -6.25049949e-01 2.25909962e-03
-1.13661385e+00 5.95405586e-02 -3.80464554e-01 1.19205809e+00
-2.37111419e-01 4.05427635e-01 -8.20183575e-01 -1.10076368e+00
1.82939798e-01 -7.26616442e-01 7.71913111e-01 2.80967832e-01
2.41072088e-01 -7.58422613e-01 5.27149200e-01 1.25028238e-01
4.57582995e-02 1.87698826e-01 1.31366813e+00 -1.06176448e+00
-7.44102538e-01 -2.54331324e-02 -7.75262117e-02 -1.68554872e-01
-3.50041687e-01 2.88005382e-01 6.09485544e-02 -2.93101043e-01
2.33475521e-01 2.81136006e-01 1.05462146e+00 3.86597961e-01
6.30718887e-01 2.01657459e-01 -1.06954193e+00 2.84691364e-01
1.29088664e+00 7.34537065e-01 1.09668933e-01 -6.58453405e-02
4.43937093e-01 3.57767701e-01 6.30445182e-01 3.83248895e-01
-1.82076812e-01 7.91140556e-01 -1.67062044e-01 2.64700260e-02
-2.48481557e-02 1.46995127e-01 1.78152099e-01 1.44986880e+00
-3.59930605e-01 -3.86273474e-01 -6.65647686e-01 1.10159591e-01
-1.67607653e+00 -1.28675628e+00 -1.01539505e+00 2.21887803e+00
4.58364159e-01 5.80728471e-01 1.42821923e-01 -1.54374868e-01
8.66235495e-01 -4.66130137e-01 2.70063043e-01 -7.45172560e-01
2.25139007e-01 7.64983833e-01 7.28952229e-01 6.77114308e-01
-8.94060969e-01 9.11202133e-01 7.14037848e+00 7.02593744e-01
-3.62633586e-01 6.63829446e-01 -1.20153032e-01 -4.60592598e-01
-2.57197291e-01 4.46022272e-01 -1.05547500e+00 5.96324980e-01
8.25173557e-01 -1.31709442e-01 -6.22822605e-02 4.54828471e-01
2.23335680e-02 -6.24440193e-01 -7.55004048e-01 5.87562978e-01
-3.07061732e-01 -1.17315328e+00 -4.72181827e-01 8.51130188e-02
8.68515015e-01 2.39395723e-01 -8.48565280e-01 7.36077726e-01
1.13107252e+00 -3.35275978e-01 6.37023211e-01 5.61304986e-01
9.37911496e-02 -1.09039736e+00 6.39915943e-01 2.55567223e-01
-1.21945274e+00 4.50068340e-02 -8.06074500e-01 1.87727794e-01
7.35375524e-01 1.33580518e+00 -1.29874957e+00 8.61944258e-01
6.41807318e-01 -2.71475226e-01 -8.06897461e-01 1.72607791e+00
9.84469354e-02 8.72460544e-01 -6.36085272e-01 -1.37532443e-01
-1.83510691e-01 -5.98368466e-01 8.92429352e-01 1.28448856e+00
6.26534283e-01 3.62725526e-01 6.09624565e-01 4.17440057e-01
2.93150425e-01 1.77797247e-02 -2.76055992e-01 -2.39218429e-01
3.43413889e-01 1.53442383e+00 -1.89104116e+00 -4.76384044e-01
3.97267118e-02 6.04549706e-01 -1.08149521e-01 -5.90734044e-03
-5.76414347e-01 -9.38678607e-02 9.34989005e-02 -5.57669904e-03
5.63036561e-01 -9.67641354e-01 -1.29462555e-02 -9.63924587e-01
-5.12140870e-01 6.68106005e-02 2.42074445e-01 -7.02499211e-01
-1.18665707e+00 3.64481598e-01 4.15437698e-01 -6.38853192e-01
-2.35102296e-01 -5.56764126e-01 -9.69143689e-01 3.11202496e-01
-4.95428383e-01 -7.15385258e-01 2.69149661e-01 -3.43641703e-04
3.41342151e-01 -4.21248943e-01 4.60466594e-01 1.78608656e-01
-2.87112623e-01 -2.86039531e-01 6.48377597e-01 -1.97883576e-01
7.57083416e-01 -1.60912895e+00 1.93631843e-01 6.65958464e-01
2.88072050e-01 7.68269598e-02 1.04688811e+00 -1.38204896e+00
-1.17303753e+00 -1.16654170e+00 9.44045007e-01 -6.01900995e-01
8.18053186e-01 -9.94144261e-01 -2.05801278e-01 5.09113967e-01
4.75681216e-01 9.76345167e-02 7.02485561e-01 2.59961247e-01
3.63197803e-01 2.71881104e-01 -9.91454482e-01 -6.66356087e-02
9.35307205e-01 5.46204597e-02 -5.18961728e-01 8.98601830e-01
6.06248021e-01 -1.06292419e-01 -7.59607494e-01 9.15005624e-01
-3.72345820e-02 -1.25237072e+00 8.54175389e-01 -8.34169507e-01
-4.02860910e-01 -3.04661930e-01 1.88060597e-01 -1.40436447e+00
-9.89591360e-01 -6.16688609e-01 4.14159596e-01 1.36301959e+00
7.91660026e-02 -2.38066316e-01 7.19986558e-01 8.95733759e-03
-2.54620165e-01 2.86738854e-02 -9.23121989e-01 -8.90021145e-01
2.74528295e-01 -3.08264464e-01 5.70109300e-02 8.43113720e-01
-2.24213339e-02 4.88307387e-01 -8.47454593e-02 6.37883961e-01
6.13929987e-01 3.05215001e-01 3.58883977e-01 -1.73409915e+00
-7.74551511e-01 -2.40195751e-01 -3.90359640e-01 -1.68050021e-01
-2.60780752e-01 -1.31710923e+00 1.17000088e-01 -1.28586650e+00
3.00954223e-01 -9.35572624e-01 -7.07837567e-02 -1.83039248e-01
3.34129125e-01 1.89030498e-01 4.04026151e-01 1.24378186e-02
-1.14400697e+00 2.58810103e-01 6.05480313e-01 3.65536422e-01
-1.81796253e-01 2.66895473e-01 1.30378038e-01 9.21538591e-01
9.35140133e-01 -8.06022167e-01 2.20170841e-01 1.26990095e-01
4.50217694e-01 5.33255041e-01 2.22395390e-01 -1.47571158e+00
3.14163029e-01 -1.15057461e-01 9.98008192e-01 -9.84748900e-01
3.06519121e-01 -3.13835889e-01 6.29607677e-01 1.37615681e-01
-5.93674257e-02 1.80492863e-01 7.84012526e-02 3.72305065e-01
-4.84779067e-02 -9.94565070e-01 7.94044316e-01 -4.93861407e-01
1.12976000e-01 1.96811389e-02 -1.09768581e+00 -2.87128776e-01
1.42740214e+00 7.81286538e-01 -1.52386108e-03 2.05701888e-01
-1.10741985e+00 4.98087667e-02 3.13496679e-01 3.74208868e-01
-1.10416763e-01 -1.49265444e+00 -3.45449269e-01 -2.04027355e-01
-5.00696123e-01 -3.72691363e-01 -3.28947663e-01 8.07246745e-01
-3.91341954e-01 7.22418606e-01 -8.00877586e-02 -4.76044118e-01
-1.44423771e+00 1.08595634e+00 -2.09527109e-02 -4.16275620e-01
-5.31539619e-01 8.93045187e-01 -3.61724682e-02 -3.30509990e-01
-4.58719432e-01 -1.83918029e-01 4.28802595e-02 1.09233551e-01
2.22457886e-01 3.57080966e-01 2.74970800e-01 -4.41637665e-01
-1.03414059e-01 1.21993996e-01 2.07583427e-01 -3.03681582e-01
9.93109703e-01 2.53651857e-01 -5.60938299e-01 3.92566770e-01
6.73616648e-01 1.22435486e+00 -5.27237773e-01 2.79869020e-01
2.98571765e-01 -2.87042975e-01 -1.14853270e-01 -7.18902469e-01
-6.90990925e-01 5.88906646e-01 3.32190394e-01 8.49757910e-01
3.59868497e-01 8.11999381e-01 5.73330164e-01 3.97309005e-01
7.11409390e-01 -8.29186320e-01 -2.67878145e-01 3.32630605e-01
4.20272917e-01 -6.86100066e-01 -7.88587853e-02 -7.91073203e-01
-4.48879987e-01 1.25300729e+00 4.57723260e-01 1.04443774e-01
5.06766617e-01 8.46593976e-01 -5.06559551e-01 -2.01011509e-01
-1.02808094e+00 -3.24392974e-01 1.20051242e-01 3.22251529e-01
1.89306378e-01 4.13087904e-01 -7.69551277e-01 7.21457779e-01
-5.68077266e-01 -6.96784914e-01 6.76241338e-01 8.36113036e-01
-9.85161006e-01 -1.73121607e+00 -8.70767951e-01 6.38068855e-01
-3.15358490e-01 8.23909044e-02 -5.46136439e-01 5.68460703e-01
6.28447354e-01 1.18992627e+00 1.36119574e-01 -1.70890406e-01
-9.95295271e-02 4.05225843e-01 8.36990297e-01 -8.55897009e-01
-7.01331735e-01 5.92167616e-01 4.22621012e-01 -6.95414096e-02
-4.91521925e-01 -9.64585066e-01 -1.53408790e+00 1.81673840e-01
-5.08552313e-01 1.40628135e+00 9.25257206e-01 1.02068532e+00
-1.44958064e-01 5.29349506e-01 6.40188158e-01 -1.33069682e+00
1.25726089e-01 -5.69570243e-01 -6.90594494e-01 -9.80273411e-02
-5.84505320e-01 -7.53462017e-01 -3.35842967e-01 -3.82351190e-01] | [15.70753002166748, 2.9174845218658447] |
5af461e3-d13f-4e12-a19c-1d39971a9f79 | a-question-type-driven-and-copy-loss-enhanced | 2005.11665 | null | https://arxiv.org/abs/2005.11665v1 | https://arxiv.org/pdf/2005.11665v1.pdf | A Question Type Driven and Copy Loss Enhanced Frameworkfor Answer-Agnostic Neural Question Generation | The answer-agnostic question generation is a significant and challenging task, which aims to automatically generate questions for a given sentence but without an answer. In this paper, we propose two new strategies to deal with this task: question type prediction and copy loss mechanism. The question type module is to predict the types of questions that should be asked, which allows our model to generate multiple types of questions for the same source sentence. The new copy loss enhances the original copy mechanism to make sure that every important word in the source sentence has been copied when generating questions. Our integrated model outperforms the state-of-the-art approach in answer-agnostic question generation, achieving a BLEU-4 score of 13.9 on SQuAD. Human evaluation further validates the high quality of our generated questions. We will make our code public available for further research. | ['Nan Jiang', 'Yunfang Wu', 'Xiuyu Wu'] | 2020-05-24 | a-question-type-driven-and-copy-loss-enhanced-1 | https://aclanthology.org/2020.ngt-1.8 | https://aclanthology.org/2020.ngt-1.8.pdf | ws-2020-7 | ['type-prediction'] | ['computer-code'] | [ 0.22092012 0.5264859 0.24585132 -0.30950218 -1.1985079 -0.84336394
0.52272916 0.14950958 -0.2276954 1.0711993 0.5363187 -0.40046278
0.10883699 -1.0760618 -0.69595456 0.00930414 0.74285287 0.5203562
0.65015966 -0.6993538 0.59438586 -0.06628245 -1.5141984 0.8666496
1.2501278 0.66487914 0.5776088 1.0684443 -0.78493303 1.083798
-1.161868 -0.8433418 -0.06310654 -1.0297056 -1.3626152 -0.18378717
0.59365475 -0.26705545 0.02858637 0.73570067 0.56253344 0.07246608
0.64947164 -1.0512612 -0.89548707 0.6882501 0.04416494 0.37241098
0.88896495 0.14164183 1.2354759 -0.68165207 0.5806354 1.09868
0.18657482 0.98242354 -0.7506339 -0.42739534 -0.14388546 0.38889536
-0.78635347 -0.30251798 0.62614995 -0.36178035 0.83687454 0.7441655
0.26780018 0.8276092 0.38971528 0.7375719 1.1323302 -0.5077848
0.19594927 0.30997038 0.25034627 0.5941547 0.07072055 -0.45300373
-0.37727678 -0.09484292 0.2704035 -0.52282864 -0.22864404 0.4323678
-1.0502874 0.997761 0.20081383 0.31433278 -0.26935953 -0.04079327
0.20453939 0.6834757 0.15837649 1.0450028 -0.65544796 -0.26094857
-0.80684334 0.8168815 1.147517 1.179473 0.66210824 -0.3599361
-1.1793776 0.7369552 0.12243121 0.33310017 0.745292 -0.97837174
0.7402164 1.0532932 0.17896754 -0.80215496 -0.01296449 -0.4381475
-0.5415425 -0.15779705 0.5260935 -0.5182923 -0.54707307 1.5847521
0.33587107 -0.3709474 0.09185302 0.55498165 1.3043398 0.9432529
-0.34184843 0.04439846 1.7756143 -1.4026445 -0.8486321 -0.30284598
0.47840136 -1.1303573 1.3986728 0.08358929 -1.3444693 -0.7606746
-0.7032931 -0.26905048 -0.09779882 0.16629496 0.07857385 0.78061855
-0.8988778 0.30357048 0.00879809 -0.2776579 -0.05837694 -0.11199191
-0.01697747 0.03646294 -1.3611974 0.8163759 0.17451298 -0.5012688
-0.5708524 -0.8079229 -0.63804495 0.28456008 0.3724164 -1.0951747
1.8362286 -0.7611308 -1.4621406 0.6466905 -0.57644373 -0.33906627
0.36792016 -0.0692792 -0.19158013 0.3057237 0.5244903 0.7249603
0.7742488 -0.9778511 -0.6448783 0.02489985 0.4397102 0.10840914
-0.20799886 0.1255217 -0.11026973 -0.8088136 -0.19544366 -0.64244306
-0.04821746 -0.14599916 -0.46239644 -0.57510823 0.28215 -1.0155835
1.463529 -1.3808979 -0.20387655 -0.32416296 -0.1614189 0.23527406
-0.4281118 0.79827017 0.10024597 0.32785264 -0.3301836 0.11667199
0.1610086 -0.1169861 -0.47241917 -0.6408851 0.49028146 1.1473315
-0.84138376 -0.5093209 -0.48806754 -0.17376901 -0.7475578 0.8397005
-0.838728 0.16628334 -0.3422731 0.1841605 0.6757646 -0.25068033
-0.2698697 0.23540474 0.20388265 0.9115488 -1.061312 1.4754244
-0.60338175 0.3054603 -0.3228536 -0.25752044 1.09073 0.41653353
-0.45232734 -0.79490495 -0.17755283 0.1506333 0.14338776 -0.97603565
0.79396755 -0.20124857 -0.16537108 0.7472119 0.16053125 -0.49703187
0.8466769 0.49190262 1.2328432 0.05379347 0.0660407 -0.15331922
1.0760107 -0.00797205 0.20350064 0.89260036 0.05647514 0.93565255
0.85225093 0.03437737 -0.9201018 -1.007869 0.31514195 1.0250348
-0.33798718 -0.5036033 -1.0684528 -1.1256781 -0.30064753 1.3268487
-0.6003494 -0.08890872 -0.7081372 -0.19505379 0.49520716 0.25826898
0.36231545 -1.3420784 -0.31424594 0.31955203 -1.0178624 -0.7578303
-0.85622287 -0.5134213 -0.7080609 -1.1222284 -0.78644943 -0.92195004
0.6363525 0.07634683 1.3688247 0.32272005 0.08414737 0.2066836
-0.81680864 -0.44562438 -0.91340417 0.40407747 -0.78569466 -0.06707554
0.19532616 -0.1892612 -0.6724578 0.16463752 -1.0431067 0.07641403
0.46450347 0.68812853 0.18983746 -0.4093902 1.1172739 -0.8894431
1.3573372 -0.64095974 -0.29615322 0.508984 -0.3130061 0.4757
0.89330524 -0.08446141 -1.1897361 -0.3692555 -0.8376652 0.5534402
-0.24533845 0.2559535 -0.09627372 0.31968698 0.79468375 0.5203659
0.05035855 -0.501457 0.3013708 0.6603331 0.15742493 -0.4085859
0.8395347 -0.13050276 -0.31862894 -0.5154201 -1.1244593 -0.4454675
-0.14529714 -0.2466503 0.844913 -0.54280645 -0.5405976 0.23337248
-1.7057672 -0.02138934 -0.4733293 -0.10658383 -0.5295236 0.45909044
-0.5152775 -0.74214154 -0.71201956 -0.9165226 0.83576983 0.42486766
-0.4857997 -0.70899135 0.34358075 1.096091 0.5058244 -0.102341
1.0981996 -0.8260248 -0.6534387 -0.08684821 -0.06972096 0.524563
0.16201735 -0.33241448 -0.6785175 0.0842149 0.32232273 -0.32841685
0.7922444 -0.30040172 1.0344267 -0.7744547 0.341988 -0.18177363
1.1864021 -0.07390061 0.9357399 0.27314085 0.2004763 1.0182307
0.60499614 0.16438088 0.7689671 0.54157525 0.3777472 0.40660307
-0.5177291 -0.5076981 0.35170743 1.1732298 0.5934643 -0.5657519
-0.60109496 0.8312394 -1.4912459 -1.2484959 -0.8448572 1.9543068
1.2596264 -0.03391333 -0.10124129 0.18276602 0.638527 0.01521661
-0.09747571 -0.7470613 0.05928103 0.57604855 -0.2778209 0.66524464
-0.43569222 0.7386475 5.9321933 0.8239465 -0.43564385 0.17835894
0.43850893 0.12799862 -0.9215294 0.13110557 -1.0147616 0.8467505
1.052934 -0.4559313 -0.01302843 0.5000309 0.20302872 -0.25150007
-0.88426685 0.34980455 0.49082094 -1.4236411 0.5270009 -0.48233578
0.7339409 -0.69650036 -0.24027054 0.5171771 0.16860977 -0.8032532
0.6137322 0.8336464 0.29805836 -0.82113606 0.83812726 0.7986731
-0.875142 -0.04494911 -0.51828045 -0.25064006 0.37491587 0.58233416
-1.1091243 0.70374596 0.437708 -0.23034407 -1.0180628 1.2156163
-0.77200884 0.7819647 0.17943719 -0.6000368 0.17628108 0.08464836
0.41170463 0.9838027 0.62677187 0.08975362 -0.32692093 1.0919532
-0.34567755 0.4731638 -0.18363774 0.2043594 0.3799651 1.3350576
-0.1271478 -0.35938808 -0.11409812 1.2500976 0.3940513 -0.06841808
-0.637761 -0.94156706 0.18846953 0.43919885 0.32282674 0.1478754
-0.30716482 -1.1018279 0.4533807 -1.1592822 0.33703777 -1.1236432
-1.3997306 0.74431473 -0.34052825 -1.065626 -0.5508638 -0.2796124
-1.1691191 1.1051552 -1.749711 -0.9138671 -0.34123704 0.2460294
0.8757446 -0.08938676 0.6124908 0.17731683 -0.1728711 0.91574395
-0.19633093 0.12295305 0.883682 -1.42943 0.76142275 0.99887246
-0.14545768 0.58115065 0.7139368 -0.6523687 -1.0954713 -1.259872
1.9499516 -0.65594983 0.41902813 -0.2974924 -0.98326135 0.1961987
0.7758278 -0.91509646 0.82015073 -0.40325618 -0.24043536 0.0070809
-1.2969714 0.6086709 0.480756 -0.23308206 -1.1640604 0.6182221
1.2644914 -0.08501591 -0.5689601 0.05884732 0.05405669 -1.1412755
0.5403496 -0.56323504 1.0497847 -0.37405255 0.23748228 -1.2773968
-0.3335452 -0.6492394 -0.15569213 1.6454765 0.8679537 -0.5226881
0.77241015 0.44188315 -0.16345629 -0.73744833 -0.69200796 -0.75148135
0.40003094 0.01693573 0.9580198 0.47760555 -0.10768154 0.83182186
-0.15189847 -0.23900256 0.23151319 0.20918353 0.79000187 -0.83925366
-0.512724 -0.2957574 0.2442387 -1.3754145 -0.04532517 -1.0316982
0.06421395 -2.0096893 0.1987352 -0.0589899 0.4483136 0.06052376
-0.8363396 0.14487362 0.4437972 -0.28024203 -0.41879776 0.7705139
1.5898974 -0.0471682 0.18481387 0.39433455 -1.242497 0.29397961
1.2800018 -0.65666133 -0.37069172 -0.5980977 0.5799936 0.2086939
0.10694861 -1.0031037 0.2205764 -0.16344619 0.15573467 -0.74973303
0.02421632 -0.36179546 -0.4015014 0.6622676 -0.7205241 0.4572262
-0.03073251 0.21442263 -0.28282377 -1.1089265 0.6551779 -0.5155051
-0.12461504 -0.02363534 -0.55938596 0.7389736 0.890977 0.02493119
-0.43397772 -0.81716704 -0.20054626 0.43105003 0.11998182 0.6669551
0.7419373 -1.3386137 -1.3317978 -0.10306196 0.21214215 -0.28014615
0.29959306 0.18629465 -0.7482495 0.5235235 -0.03525137 0.09541896
-1.2426878 0.35395217 0.19416262 -0.69261533 0.05671531 0.9481242
-0.02980176 -0.98118615 -0.08012626 -0.16005059 -0.6308018 0.14648151
0.95156896 0.5684675 0.26929647 -0.17779018 0.04283478 0.28626522
-0.21597295 -0.26539665 0.8555058 -0.10229071 -0.44424137 0.16502735
1.0850374 0.18655355 -0.60362506 0.04084357 -0.05928437 -0.32639137
-0.6646392 -1.2596891 -0.52885085 0.9099802 0.13703384 0.431569
0.9364855 0.06527245 1.3397415 0.38897264 0.04949853 -1.0160077
0.5815885 0.97286415 1.3971951 -0.9679359 -0.40783048 -0.46175686
-0.65644276 0.944378 1.0266466 0.01747504 0.0199969 -0.18000999
0.04680771 0.12334885 -1.1829579 -0.14117509 0.4003826 0.48647583
0.7150309 -0.29282418 -0.92614263 0.7407019 -0.8108339 -0.19429862
1.0021375 0.86765397 -0.80364984 -1.5735282 -0.49606094 0.70296514
-0.54820603 -0.38625014 -0.82507217 0.08117177 -0.07517626 1.6746392
-0.143512 -0.16958462 0.67122424 0.39172024 0.40625882 -0.9190827
-1.3166416 -0.6720689 0.37737826 -0.14672697 0.18395312 -0.46253446
-1.1474401 -0.2715398 -0.35808516 0.5769628 0.46929234 0.74659044
0.6235818 0.55127335 0.7580748 0.29769713 -1.0145471 -1.1370273
-0.00836323 0.45082605 0.278207 0.0954383 -0.28315842 0.09323046] | [11.53770923614502, 8.206329345703125] |
eae62db8-221c-4791-a413-98e13678f335 | dhrl-a-graph-based-approach-for-long-horizon | 2210.05150 | null | https://arxiv.org/abs/2210.05150v3 | https://arxiv.org/pdf/2210.05150v3.pdf | DHRL: A Graph-Based Approach for Long-Horizon and Sparse Hierarchical Reinforcement Learning | Hierarchical Reinforcement Learning (HRL) has made notable progress in complex control tasks by leveraging temporal abstraction. However, previous HRL algorithms often suffer from serious data inefficiency as environments get large. The extended components, $i.e.$, goal space and length of episodes, impose a burden on either one or both high-level and low-level policies since both levels share the total horizon of the episode. In this paper, we present a method of Decoupling Horizons Using a Graph in Hierarchical Reinforcement Learning (DHRL) which can alleviate this problem by decoupling the horizons of high-level and low-level policies and bridging the gap between the length of both horizons using a graph. DHRL provides a freely stretchable high-level action interval, which facilitates longer temporal abstraction and faster training in complex tasks. Our method outperforms state-of-the-art HRL algorithms in typical HRL environments. Moreover, DHRL achieves long and complex locomotion and manipulation tasks. | ['H. Jin Kim', 'Inkyu Jang', 'Jigang Kim', 'Seungjae Lee'] | 2022-10-11 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [-2.79358536e-01 1.66239142e-01 -3.82282615e-01 1.57501131e-01
-6.22773826e-01 -4.96794194e-01 4.24868137e-01 3.00530493e-01
-6.11058712e-01 1.19237542e+00 1.37697071e-01 -1.28326371e-01
-2.88829893e-01 -9.38261807e-01 -7.46790469e-01 -6.85875833e-01
-5.82522452e-01 2.04884484e-01 6.91933811e-01 -4.02995884e-01
3.06254566e-01 3.24758589e-01 -1.63438821e+00 5.62401786e-02
7.71265864e-01 7.17660487e-01 4.14539784e-01 5.33652484e-01
2.33915687e-01 1.08384109e+00 -5.23758888e-01 4.04640436e-01
3.99649948e-01 -5.81137121e-01 -7.07526982e-01 9.96302515e-02
6.05707355e-02 -5.44552982e-01 -5.32947004e-01 5.81539690e-01
5.27176619e-01 6.53817415e-01 1.74138561e-01 -1.58512473e+00
-2.13941589e-01 6.72726989e-01 -5.62456071e-01 -1.51826486e-01
2.50857174e-01 5.72906673e-01 8.71762037e-01 -2.02134281e-01
7.11855352e-01 1.39308155e+00 6.73835337e-01 6.29513323e-01
-1.33806133e+00 -4.39101011e-01 6.32939398e-01 9.84983519e-02
-8.24399292e-01 -9.99674723e-02 5.65745115e-01 -4.64792818e-01
1.38960600e+00 -3.21231991e-01 1.04835534e+00 1.03292716e+00
5.23426175e-01 7.71309376e-01 1.33505499e+00 -9.99585539e-02
7.31290877e-01 -5.23953259e-01 -2.88644433e-01 9.07833636e-01
1.69341981e-01 6.08508050e-01 -6.40000641e-01 -6.74323291e-02
1.36281252e+00 -7.17906281e-02 -2.39931210e-03 -7.94906676e-01
-1.11215818e+00 8.90824199e-01 5.59191108e-01 -5.65535836e-02
-4.25118744e-01 5.77349842e-01 6.57560468e-01 4.46484476e-01
-2.20412865e-01 8.07883143e-01 -3.54033560e-01 -4.27845359e-01
-5.58296680e-01 7.03744352e-01 7.31769860e-01 1.12808168e+00
7.99137890e-01 3.02017748e-01 -3.21125805e-01 5.98226845e-01
-1.57322526e-01 8.76968801e-02 3.61868083e-01 -1.45671940e+00
3.78143221e-01 5.77209473e-01 4.22519714e-01 -5.16355813e-01
-7.78412700e-01 -3.49820763e-01 -5.82896352e-01 6.79931164e-01
4.37838376e-01 -1.98232770e-01 -8.79859507e-01 2.08376336e+00
5.39119720e-01 -9.80011523e-02 1.16090558e-01 6.16433322e-01
7.67230913e-02 7.10197091e-01 2.57135600e-01 -2.34025061e-01
9.86562669e-01 -1.36071193e+00 -5.31430542e-01 -3.47684056e-01
7.24998534e-01 2.21154895e-02 1.33245909e+00 3.85175049e-01
-1.48262358e+00 -6.34024382e-01 -1.11366987e+00 -1.35170266e-01
-3.92493337e-01 -2.69437790e-01 8.32956851e-01 4.78533022e-02
-1.02863359e+00 1.03646135e+00 -1.12212658e+00 -1.50476158e-01
2.07903720e-02 4.52363014e-01 -1.50912777e-01 1.33714527e-02
-1.25690317e+00 9.66505706e-01 6.04250431e-01 -1.95496961e-01
-1.36918366e+00 -5.21999955e-01 -1.15975356e+00 1.36348531e-01
8.02631438e-01 -5.91804385e-01 1.58158326e+00 -1.84308141e-01
-1.71424615e+00 2.13735536e-01 4.48159456e-01 -5.95541656e-01
5.51905155e-01 -4.09240574e-01 1.62684679e-01 5.08198440e-02
1.97873965e-01 8.61389935e-01 6.95448697e-01 -1.08959091e+00
-9.35161173e-01 -3.08937490e-01 4.90747273e-01 5.35399616e-01
-4.54361625e-02 -7.37525284e-01 -3.41724932e-01 -4.24998403e-01
-2.73187339e-01 -9.73331988e-01 -6.65151477e-01 -6.86983243e-02
2.03654736e-01 -5.43244481e-01 6.59151196e-01 -3.82113993e-01
1.48155653e+00 -2.00882959e+00 7.51733601e-01 -2.47437254e-01
2.41248578e-01 1.38843060e-01 -2.62741446e-01 9.92673635e-01
3.64425123e-01 -1.22922145e-01 -1.20701797e-01 -1.60492137e-01
3.10030013e-01 6.38875723e-01 -1.56749159e-01 2.61221349e-01
-5.95204830e-02 6.77139819e-01 -1.51783717e+00 -4.25738543e-01
1.17343150e-01 -3.92634347e-02 -9.28416908e-01 4.18314666e-01
-5.63228786e-01 5.61891377e-01 -4.15597826e-01 3.47633779e-01
7.10255001e-03 -7.01373741e-02 2.82446116e-01 2.62816489e-01
-5.55319846e-01 3.88688028e-01 -1.16649628e+00 2.03413534e+00
-5.30255377e-01 1.74430579e-01 3.94917168e-02 -7.07959175e-01
6.60942256e-01 1.81660473e-01 6.45908177e-01 -1.02713394e+00
-2.16388956e-01 1.20306663e-01 -8.74354541e-02 -3.64658415e-01
4.83146548e-01 -1.78787615e-02 -4.68039185e-01 1.61322206e-01
6.85985200e-03 -5.69289565e-01 7.69407332e-01 1.47143835e-02
1.26691520e+00 8.93363655e-01 5.52899778e-01 -2.31639802e-01
1.37181997e-01 7.53227100e-02 9.97207403e-01 1.09039044e+00
-4.34239507e-01 -1.49765223e-01 8.95079613e-01 -4.22914416e-01
-1.27945566e+00 -1.16532981e+00 3.19251895e-01 1.24156046e+00
9.01478603e-02 -6.53473377e-01 -7.10604310e-01 -5.17322898e-01
3.68732274e-01 4.74720925e-01 -7.52756953e-01 -3.03282559e-01
-8.58842254e-01 -1.97470441e-01 5.15738785e-01 8.18625152e-01
7.58582592e-01 -1.45758557e+00 -1.18514764e+00 6.34307802e-01
6.43469617e-02 -1.09085226e+00 -3.00758779e-01 4.85652566e-01
-9.34423149e-01 -8.63825560e-01 -3.54841441e-01 -6.42204881e-01
2.83623248e-01 1.11213125e-01 1.24567461e+00 5.21808863e-02
-3.95998955e-01 1.71287030e-01 -3.66212457e-01 -2.10496873e-01
-2.47545131e-02 1.03671901e-01 1.72736779e-01 -8.76353383e-01
-3.37254137e-01 -8.39718282e-01 -6.61473453e-01 1.75464913e-01
-7.42729545e-01 2.26522610e-01 4.90687340e-01 1.10980630e+00
4.48852420e-01 2.21654475e-01 7.52309561e-01 -4.58857119e-01
7.48268187e-01 -1.86487734e-01 -1.03791094e+00 -2.82232463e-02
-4.90708470e-01 3.61421704e-01 8.83820653e-01 -3.47826064e-01
-8.03510070e-01 -7.47607574e-02 4.29592729e-02 -2.24674135e-01
6.23170882e-02 3.48143220e-01 1.91550627e-01 1.96375966e-01
5.96850455e-01 8.18109065e-02 -6.89578280e-02 -1.62670583e-01
3.61567914e-01 1.26804173e-01 2.15360925e-01 -9.38166916e-01
5.69931567e-01 8.15504938e-02 2.66796738e-01 -7.75757194e-01
-7.58671939e-01 -2.93744177e-01 -4.69784975e-01 -2.46959865e-01
7.71169603e-01 -8.52130234e-01 -1.15521586e+00 5.12483537e-01
-6.18697584e-01 -1.23918653e+00 -6.70781970e-01 3.27310890e-01
-1.27801967e+00 1.39122447e-02 -9.76833999e-01 -8.53542447e-01
1.82106197e-01 -1.35650539e+00 1.02972770e+00 2.27471247e-01
-6.08854666e-02 -7.33750224e-01 2.79435009e-01 -1.27127200e-01
2.86319166e-01 5.09444892e-01 1.07634425e+00 1.47101834e-01
-6.04192615e-01 3.56419742e-01 3.50459032e-02 2.87558194e-02
-6.26858249e-02 -3.56889278e-01 -3.16925228e-01 -8.29102635e-01
-2.59435087e-01 -9.91989970e-01 6.98457539e-01 2.81013012e-01
1.11891782e+00 -3.11318040e-01 -1.51865438e-01 4.28026766e-01
1.52458847e+00 4.87063766e-01 5.28099179e-01 6.46665096e-01
4.76757616e-01 4.72982258e-01 9.66757357e-01 8.49652290e-01
3.45958173e-01 5.55987000e-01 5.29477537e-01 -1.22977994e-01
-1.13057448e-02 -4.80746776e-01 5.59862614e-01 5.98639250e-01
-1.61961973e-01 -5.86588718e-02 -7.73122549e-01 5.78833044e-01
-2.35178304e+00 -1.05998611e+00 4.74798054e-01 2.21625161e+00
9.48125362e-01 4.18919176e-01 5.14034748e-01 9.93705615e-02
1.37722746e-01 2.96433896e-01 -1.08366823e+00 -5.49902558e-01
4.39136684e-01 -4.23594825e-02 5.56516409e-01 5.79139054e-01
-8.80681276e-01 1.18661249e+00 6.46176100e+00 7.25634515e-01
-7.28985071e-01 -2.99183130e-01 -8.11024457e-02 -2.27030262e-01
1.26305416e-01 -2.08760574e-02 -9.07868683e-01 2.16887563e-01
7.14877844e-01 -1.34061664e-01 9.46684301e-01 1.04344046e+00
2.77090281e-01 -3.18793327e-01 -1.28230739e+00 5.47575116e-01
-6.08296633e-01 -1.17431223e+00 -5.02643585e-02 1.65885419e-01
6.87994480e-01 -6.66427240e-02 -7.27046505e-02 1.03804982e+00
9.55719173e-01 -9.21647966e-01 6.18349314e-01 1.22577541e-01
6.83866560e-01 -1.02575409e+00 1.46582156e-01 6.29969060e-01
-1.52261400e+00 -6.04841769e-01 -2.50497162e-01 -5.21270037e-01
-2.57529714e-03 -1.57636374e-01 -3.81929785e-01 5.89103758e-01
7.31117964e-01 6.79429173e-01 -2.43396938e-01 8.19095612e-01
-3.00922483e-01 7.83089176e-02 -1.36237532e-01 6.54404908e-02
8.65413070e-01 -2.16091782e-01 2.75449425e-01 8.75670612e-01
-1.82761401e-02 1.17357671e-01 8.92596245e-01 4.72613752e-01
2.07595393e-01 -4.04789716e-01 -8.56579602e-01 -2.88528949e-01
5.06956697e-01 8.66244256e-01 -5.03986180e-01 -2.75706828e-01
-4.11150932e-01 6.61193013e-01 1.02652431e+00 4.13893640e-01
-1.07662868e+00 -4.39876646e-01 8.16205502e-01 -6.84887096e-02
2.70615399e-01 -8.39872241e-01 -2.15817951e-02 -1.02366006e+00
-1.69690818e-01 -1.00991106e+00 4.06545430e-01 -5.45693576e-01
-9.21030760e-01 3.57731998e-01 3.68976109e-02 -1.17458665e+00
-4.01212364e-01 -3.66184145e-01 -2.53112428e-03 4.10928905e-01
-1.67059445e+00 -7.13188469e-01 -2.74242342e-01 7.34074712e-01
9.07776654e-01 2.51869380e-01 8.37539315e-01 1.34530431e-02
-5.97490311e-01 2.96749175e-01 1.49742991e-01 -2.25463480e-01
4.99456465e-01 -1.58852577e+00 2.99148947e-01 5.54809213e-01
-6.84194207e-01 3.64396989e-01 6.06592000e-01 -7.28776872e-01
-1.31856954e+00 -8.98795009e-01 2.67478794e-01 -8.69031325e-02
7.42728531e-01 -4.24028188e-01 -6.72328711e-01 7.95425653e-01
1.66352570e-01 -4.28604424e-01 2.80251354e-01 1.86566621e-01
-1.37536690e-01 2.50391793e-02 -8.82281065e-01 1.11402655e+00
1.35237336e+00 -3.50826979e-01 -6.15686238e-01 1.80805415e-01
1.01653790e+00 -5.97760677e-01 -1.05485213e+00 5.06612659e-01
6.65986478e-01 -1.10433435e+00 9.79687631e-01 -7.13657975e-01
6.08342409e-01 -1.96625218e-01 9.93747041e-02 -1.51069510e+00
-6.22915924e-01 -7.69848526e-01 -3.66775364e-01 5.55298567e-01
2.85316706e-02 -4.69500154e-01 5.99495828e-01 3.42534095e-01
-3.28657150e-01 -1.08724010e+00 -6.53896689e-01 -1.25433552e+00
1.73321798e-01 9.28964764e-02 3.69657487e-01 6.42732263e-01
4.57874268e-01 3.40397060e-01 -6.20194435e-01 5.57800103e-03
8.69674385e-01 2.82172590e-01 7.75995374e-01 -9.15445089e-01
-5.73294520e-01 -4.61185396e-01 1.01341955e-01 -1.14463091e+00
-5.25891259e-02 -3.73592079e-01 3.43393475e-01 -1.72065759e+00
-1.26339972e-01 -5.17592788e-01 -2.65753567e-01 6.83415115e-01
-3.60254422e-02 -4.35373992e-01 4.26109672e-01 9.68133062e-02
-8.10766160e-01 8.63164425e-01 1.78680074e+00 3.39206219e-01
-7.24218428e-01 -4.48926806e-01 -8.84314179e-02 7.51760364e-01
1.11415613e+00 -3.69664937e-01 -7.91842222e-01 -4.49497402e-01
5.22299819e-02 7.27095842e-01 3.70027609e-02 -1.26890278e+00
2.74706274e-01 -6.23505771e-01 1.66684017e-01 -3.82408410e-01
4.30209309e-01 -6.47828639e-01 -1.42044038e-01 8.97856414e-01
-6.59027576e-01 4.42319751e-01 2.79476970e-01 7.10185587e-01
-5.15894443e-02 2.02025935e-01 9.19028580e-01 -6.08785629e-01
-7.36934781e-01 3.63349348e-01 -5.79160333e-01 2.10478708e-01
1.20634043e+00 -1.41934380e-01 -3.73700440e-01 -9.39808190e-02
-9.25735414e-01 8.61684680e-01 5.76194763e-01 4.62503314e-01
3.66190344e-01 -1.16710496e+00 -2.00082079e-01 1.80756208e-02
-5.54311052e-02 2.35039219e-01 2.84703046e-01 6.66318238e-01
-3.93271506e-01 4.09801513e-01 -8.86224031e-01 -2.14920416e-01
-7.97900140e-01 8.30659032e-01 1.84154108e-01 -1.00098956e+00
-8.65936816e-01 4.83460486e-01 2.00306520e-01 -2.28318378e-01
6.35621905e-01 -3.17214847e-01 1.92038920e-02 7.28825331e-02
2.61579305e-01 5.98038971e-01 -4.11686271e-01 1.52395904e-01
-7.53497407e-02 3.69359314e-01 -7.86577985e-02 -3.30159456e-01
1.31946206e+00 2.14457437e-02 8.07124227e-02 6.06274784e-01
6.18779302e-01 -5.90540290e-01 -2.10471749e+00 1.97684430e-02
1.33787678e-03 -2.65071511e-01 -7.63442814e-02 -6.89031839e-01
-4.76085603e-01 8.61183584e-01 1.18840329e-01 2.74920404e-01
1.04551363e+00 -3.91992182e-01 9.11929965e-01 6.00560904e-01
9.33496058e-01 -1.55018234e+00 5.35674334e-01 7.86737204e-01
9.81055558e-01 -9.06568050e-01 4.47438331e-03 -2.82831341e-02
-5.94537556e-01 9.32349503e-01 1.17765939e+00 -4.30397600e-01
2.05502622e-02 2.62348384e-01 -4.12776798e-01 -1.14692107e-01
-1.12325847e+00 -5.61536252e-01 -4.54482615e-01 6.41913772e-01
6.54888377e-02 7.24301860e-02 -5.33078790e-01 1.24430209e-01
-2.39954814e-02 8.18777606e-02 5.54120600e-01 1.74839568e+00
-8.41880918e-01 -1.16664600e+00 -3.52071077e-02 4.82029989e-02
-7.51743689e-02 1.40616834e-01 1.53552100e-01 1.30127478e+00
-2.78650708e-02 6.76550686e-01 -4.92422730e-02 -1.44791827e-01
5.85568964e-01 -1.19099535e-01 8.67696226e-01 -6.00689590e-01
-6.26606524e-01 6.53844029e-02 1.35400400e-01 -1.23084950e+00
1.16422456e-02 -4.77202803e-01 -1.57738829e+00 -3.04100364e-01
1.43830016e-01 3.21331948e-01 1.76092237e-01 5.95410585e-01
3.35153729e-01 8.27447116e-01 5.16999841e-01 -1.22042632e+00
-1.03510714e+00 -4.93182421e-01 -8.01208675e-01 2.72250950e-01
6.29370391e-01 -1.05945516e+00 -1.32167086e-01 -2.28634089e-01] | [4.164243221282959, 1.5676871538162231] |
5b1ecf6b-3d82-4e20-95a5-436ae5f14e7f | robust-neural-malware-detection-models-for | 1806.10741 | null | http://arxiv.org/abs/1806.10741v1 | http://arxiv.org/pdf/1806.10741v1.pdf | Robust Neural Malware Detection Models for Emulation Sequence Learning | Malicious software, or malware, presents a continuously evolving challenge in
computer security. These embedded snippets of code in the form of malicious
files or hidden within legitimate files cause a major risk to systems with
their ability to run malicious command sequences. Malware authors even use
polymorphism to reorder these commands and create several malicious variations.
However, if executed in a secure environment, one can perform early malware
detection on emulated command sequences.
The models presented in this paper leverage this sequential data derived via
emulation in order to perform Neural Malware Detection. These models target the
core of the malicious operation by learning the presence and pattern of
co-occurrence of malicious event actions from within these sequences. Our
models can capture entire event sequences and be trained directly using the
known target labels. These end-to-end learning models are powered by two
commonly used structures - Long Short-Term Memory (LSTM) Networks and
Convolutional Neural Networks (CNNs). Previously proposed sequential malware
classification models process no more than 200 events. Attackers can evade
detection by delaying any malicious activity beyond the beginning of the file.
We present specialized models that can handle extremely long sequences while
successfully performing malware detection in an efficient way. We present an
implementation of the Convoluted Partitioning of Long Sequences approach in
order to tackle this vulnerability and operate on long sequences. We present
our results on a large dataset consisting of 634,249 file sequences, with
extremely long file sequences. | ['Karthik Selvaraj', 'Mady Marinescu', 'Jack W. Stokes', 'Rakshit Agrawal'] | 2018-06-28 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [ 3.67642522e-01 -6.47641718e-01 -2.03660622e-01 -2.15401743e-02
-2.86536038e-01 -1.07211924e+00 7.49948382e-01 1.77154601e-01
-5.65970957e-01 1.05482429e-01 -3.67723942e-01 -9.75272417e-01
2.39872709e-01 -7.77235925e-01 -8.17499101e-01 -5.55315912e-01
-6.80274904e-01 2.40616530e-01 4.41312104e-01 3.33873071e-02
5.40390253e-01 8.57543647e-01 -1.36475039e+00 8.87243807e-01
2.33725861e-01 7.89036810e-01 1.91761985e-01 1.32271254e+00
-2.24579334e-01 1.16144013e+00 -1.15481305e+00 -4.43493724e-02
1.85470030e-01 1.33986771e-01 -6.35293901e-01 -4.50614691e-01
5.02952859e-02 -6.82758510e-01 -5.07625937e-01 9.84563708e-01
1.16691917e-01 -3.05904150e-01 5.94109237e-01 -1.27779925e+00
-1.15882121e-01 9.83570755e-01 -1.76674798e-01 6.57275975e-01
4.17554349e-01 5.79892039e-01 6.34203970e-01 -5.06553710e-01
5.43853462e-01 9.18068111e-01 8.45365822e-01 5.33307135e-01
-9.53931868e-01 -6.65919840e-01 -2.69495934e-01 3.19081217e-01
-9.93071437e-01 -1.91769168e-01 5.41391969e-01 -7.55349338e-01
1.65579867e+00 2.19756335e-01 1.43432006e-01 1.84045112e+00
8.11244011e-01 3.98876876e-01 5.79342902e-01 -1.13159090e-01
5.81604801e-02 -1.40416861e-01 5.56229115e-01 5.94327807e-01
1.13193683e-01 6.28409266e-01 -9.71299538e-04 -6.86228931e-01
4.92599234e-02 8.81897509e-01 -1.82896122e-01 3.53357583e-01
-1.11229920e+00 1.05366218e+00 1.37277842e-01 4.56563920e-01
-2.01011226e-01 3.61494660e-01 1.20611930e+00 6.51536286e-01
4.84361872e-02 4.51699555e-01 -7.12065995e-01 -4.54534203e-01
-1.12067413e+00 -1.26419172e-01 1.16356742e+00 3.90331745e-01
4.63566452e-01 4.22524273e-01 -1.79474633e-02 1.56037122e-01
2.00998217e-01 4.72331524e-01 7.66775787e-01 -1.60159498e-01
2.68176079e-01 2.65718907e-01 -4.83656615e-01 -9.44191456e-01
-2.85897017e-01 -2.96368092e-01 -5.59216976e-01 3.27150285e-01
3.73084247e-01 -9.18064564e-02 -8.68100226e-01 1.32444465e+00
-1.47105679e-01 5.27161837e-01 -1.17717192e-01 -1.01238146e-01
3.11333388e-01 1.05189598e+00 -1.39867561e-02 -1.88896552e-01
1.32365417e+00 -6.20949984e-01 -4.72639769e-01 4.74904897e-03
7.91276097e-01 -5.34672499e-01 7.49589026e-01 4.79752153e-01
-4.18905437e-01 -4.02203113e-01 -1.13478088e+00 6.18103743e-01
-9.55583930e-01 -3.18386734e-01 3.50024775e-02 9.27230239e-01
-9.38277602e-01 8.52907360e-01 -1.06074262e+00 -1.04994662e-02
3.45125079e-01 5.37729144e-01 -8.37414786e-02 4.65707332e-01
-1.13031137e+00 6.98795557e-01 6.83915615e-01 -2.09745422e-01
-1.85467386e+00 -4.92325157e-01 -8.42507660e-01 1.85396433e-01
5.08388698e-01 2.21481487e-01 1.26036167e+00 -6.84488237e-01
-1.10804379e+00 7.35925317e-01 2.67452057e-02 -9.84212160e-01
2.14490563e-01 -1.15024053e-01 -5.13448358e-01 1.20365344e-01
-3.91034782e-01 -5.14179543e-02 1.56588507e+00 -1.02339959e+00
-4.04075056e-01 -9.35600474e-02 -4.11370695e-02 -8.53212655e-01
-6.28872395e-01 5.41154444e-01 2.98228592e-01 -6.60401285e-01
-8.68211687e-01 -8.42490137e-01 -3.82726081e-03 -8.09430480e-01
-6.80375755e-01 3.47547159e-02 1.64598584e+00 -8.22531402e-01
1.38721955e+00 -2.19666529e+00 -2.54943043e-01 3.15609187e-01
4.09448892e-01 8.00623655e-01 -1.01073682e-01 6.44079506e-01
-3.04675579e-01 3.76142442e-01 -3.35687250e-01 -1.47663251e-01
-1.39436156e-01 1.59001857e-01 -1.09311640e+00 4.93210107e-01
1.33128077e-01 9.09310341e-01 -6.52307093e-01 -9.12839249e-02
1.38267785e-01 4.21246827e-01 -2.93967277e-01 4.96321887e-01
-4.93699700e-01 1.71931148e-01 -6.24388121e-02 7.32751667e-01
3.50098699e-01 -1.56667486e-01 3.20798308e-02 1.92907184e-01
8.69221240e-02 4.52283502e-01 -4.95421201e-01 8.95257771e-01
-6.84692621e-01 8.62852097e-01 2.68459320e-02 -9.87792730e-01
5.54297686e-01 4.35506850e-01 3.49173129e-01 2.34937090e-02
5.32595992e-01 2.48392388e-01 3.55207920e-01 -5.74988246e-01
3.46923500e-01 2.30505913e-01 -2.78545767e-01 9.15001035e-01
3.16486810e-03 3.59017402e-01 8.24169144e-02 1.83071196e-01
1.77903712e+00 -4.09359664e-01 4.66958225e-01 1.79882243e-01
7.35874891e-01 -7.39392266e-02 6.09949566e-02 1.03467858e+00
-2.29529724e-01 -7.90811181e-02 8.55944693e-01 -5.81603169e-01
-1.05434954e+00 -1.15046418e+00 1.26105044e-02 1.41021872e+00
-5.92477322e-01 -6.36473358e-01 -7.27292299e-01 -1.23581624e+00
-8.32449272e-02 7.52228379e-01 -7.38568664e-01 -4.50801134e-01
-1.03384876e+00 -5.25593460e-01 1.34913731e+00 4.24382657e-01
4.57173027e-02 -1.46533310e+00 -9.77709055e-01 2.86624938e-01
3.98786098e-01 -1.04326594e+00 -4.60755974e-01 9.95401859e-01
-6.26544654e-01 -1.29483736e+00 1.25703946e-01 -6.34719491e-01
2.58048326e-01 -2.72296846e-01 9.59219217e-01 2.41278991e-01
-5.34047186e-01 3.24370027e-01 -5.07742882e-01 -3.61773342e-01
-1.29863715e+00 2.32592180e-01 4.34636213e-02 -2.85056472e-01
5.47537446e-01 -4.91316706e-01 2.27771848e-01 -2.82636727e-03
-1.30004394e+00 -8.04596543e-01 4.30050164e-01 8.37223470e-01
-5.65499254e-02 2.13192418e-01 3.27798754e-01 -1.06125939e+00
6.01171076e-01 -9.64251161e-01 -7.40478516e-01 7.41721014e-04
-3.55424613e-01 1.24295063e-01 1.55297637e+00 -9.90989804e-01
-4.21968281e-01 -6.29445612e-02 -5.17875671e-01 -9.87416387e-01
-4.29143876e-01 2.74714649e-01 3.46712284e-02 -1.42058402e-01
7.78806567e-01 7.29584038e-01 2.53501445e-01 -1.14706837e-01
5.01452349e-02 8.61655712e-01 4.04695451e-01 -3.45670760e-01
1.10772264e+00 4.52126473e-01 -1.65634021e-01 -7.31373489e-01
-2.01097682e-01 -3.55720252e-01 -5.13769329e-01 -5.21533936e-02
7.10461259e-01 -2.39665017e-01 -1.02219784e+00 8.45445335e-01
-1.31299317e+00 -4.04645652e-01 1.38444803e-03 -2.06988201e-01
-1.95796877e-01 6.22582436e-01 -1.12930620e+00 -6.76221550e-01
-4.85830486e-01 -1.53980529e+00 8.87572825e-01 -3.23108315e-01
-1.96561053e-01 -1.23609424e+00 4.78804298e-02 -4.46880609e-01
5.72259545e-01 3.33160102e-01 1.10990024e+00 -1.71920264e+00
-3.52097183e-01 -6.30066514e-01 2.63789862e-01 5.26512086e-01
1.73881222e-02 4.67192709e-01 -1.08801293e+00 -5.55246294e-01
4.54293221e-01 -1.04345255e-01 1.00339413e+00 4.97725382e-02
1.43574893e+00 -6.21433079e-01 -5.61913133e-01 8.63410652e-01
1.36488688e+00 7.84285665e-01 4.59416062e-01 3.16168666e-01
8.57498169e-01 3.79640698e-01 1.35596856e-01 5.03317416e-01
-4.34213191e-01 3.00801724e-01 8.91307056e-01 6.59736156e-01
3.98117870e-01 4.58297580e-02 1.12150133e+00 5.61439514e-01
5.68346202e-01 -2.21515119e-01 -1.23146582e+00 2.15394780e-01
-1.06588125e+00 -1.44740939e+00 -1.19018160e-01 2.17093134e+00
6.02104664e-01 5.44248462e-01 6.20935559e-02 3.70715916e-01
6.88428223e-01 4.95294690e-01 -3.84700984e-01 -9.61044371e-01
2.33676732e-01 5.74751794e-01 7.20924079e-01 3.77862096e-01
-1.35590911e+00 7.62228251e-01 6.43183756e+00 1.19028449e+00
-1.58463693e+00 3.48396420e-01 3.73614311e-01 -1.80978537e-01
1.39393046e-01 -1.85535282e-01 -1.07583773e+00 9.06897724e-01
2.05234790e+00 1.30801663e-01 5.64534426e-01 1.11434674e+00
4.15110774e-02 2.32399464e-01 -1.20267785e+00 6.74129844e-01
1.59648642e-01 -1.45399022e+00 2.36042961e-02 2.65749723e-01
3.00570607e-01 2.27970064e-01 4.35969770e-01 6.47308826e-01
2.69913524e-01 -1.39087605e+00 5.45371830e-01 5.11436053e-02
9.02707636e-01 -8.63370717e-01 7.10762978e-01 5.44292331e-01
-1.29047179e+00 -7.00738311e-01 -5.17417826e-02 1.34614661e-01
-5.38346097e-02 4.20170039e-01 -1.40079153e+00 6.95789382e-02
4.74630594e-01 7.89417267e-01 -7.16099381e-01 3.42510700e-01
1.89599749e-02 9.98214483e-01 -1.84575021e-01 -2.11721182e-01
5.34515500e-01 4.50563639e-01 7.20919013e-01 1.53465569e+00
1.84379235e-01 -6.71693087e-01 2.23984972e-01 7.36006975e-01
9.96285751e-02 -3.59396160e-01 -1.13963938e+00 -5.25622308e-01
4.37752455e-01 1.26552260e+00 -8.33097219e-01 -5.05879462e-01
-3.31764638e-01 7.75835633e-01 1.12495460e-01 1.49655089e-01
-1.15189028e+00 -6.21390820e-01 7.69148767e-01 1.06065385e-01
5.08942485e-01 -3.40528727e-01 -1.81708425e-01 -1.02537370e+00
-3.15151870e-01 -1.29369402e+00 4.91024971e-01 1.02591492e-01
-1.25344062e+00 8.46983016e-01 -9.99531895e-02 -1.17433739e+00
-8.87289166e-01 -1.04777956e+00 -1.07804310e+00 5.20342112e-01
-8.65869462e-01 -9.06482100e-01 7.86851943e-02 6.19760692e-01
6.16718650e-01 -4.70472276e-01 6.89610422e-01 2.04054415e-01
-4.70851898e-01 6.04190290e-01 3.20310205e-01 5.99573135e-01
4.35249358e-01 -1.08210874e+00 6.65374041e-01 1.10561657e+00
8.80852193e-02 9.24089670e-01 6.01461351e-01 -1.07986498e+00
-1.42345321e+00 -1.36458540e+00 3.41186762e-01 -6.93905234e-01
1.00512099e+00 -6.63502812e-01 -1.05378723e+00 1.09128809e+00
2.23707840e-01 -2.30028480e-02 7.26310074e-01 -5.11615872e-01
-8.94346237e-01 3.29967827e-01 -1.18692601e+00 2.32098907e-01
6.04330778e-01 -9.43387568e-01 -5.23196638e-01 1.95140377e-01
7.88530290e-01 -1.19535848e-01 -5.48909664e-01 2.92886078e-01
2.27423429e-01 -1.09404361e+00 1.07367873e+00 -7.18180120e-01
4.71050024e-01 -1.12780742e-01 -5.53195961e-02 -9.45516825e-01
5.34614846e-02 -6.80918097e-01 -8.70932937e-01 6.51672602e-01
2.12205365e-01 -8.47569704e-01 4.64711279e-01 -4.65665817e-01
-1.02703013e-01 -7.09768474e-01 -7.97394037e-01 -9.95782673e-01
1.40829593e-01 -7.72647023e-01 4.50737566e-01 8.78584623e-01
-1.87111139e-01 -8.09828714e-02 -4.22931612e-01 3.22805971e-01
5.83234310e-01 -1.52944669e-01 4.55389589e-01 -9.73696172e-01
-6.81092978e-01 -6.75877154e-01 -4.35371220e-01 -6.00006104e-01
6.35211766e-01 -1.02029383e+00 3.01398039e-02 -4.54335630e-01
1.14250369e-01 1.28223509e-01 -1.24341868e-01 4.66596335e-01
3.72624435e-02 9.02357027e-02 3.84193398e-02 3.93634737e-01
-3.13016862e-01 -9.78874564e-02 1.36688545e-01 -2.25828648e-01
8.38583428e-03 2.60343943e-02 2.35004991e-01 7.82027125e-01
8.74520719e-01 -7.48423994e-01 -1.51302859e-01 1.33511677e-01
9.12584588e-02 -3.40214483e-02 4.80724722e-01 -1.08500195e+00
1.49020627e-01 1.30904675e-01 3.44595164e-01 -9.16188598e-01
1.48159280e-01 -8.86275053e-01 2.32613599e-03 1.25808048e+00
-2.99243331e-01 4.76348817e-01 1.60346135e-01 6.50649011e-01
-3.31518687e-02 -6.21148527e-01 8.58102798e-01 -1.89201280e-01
-6.91643536e-01 3.21782947e-01 -1.21197891e+00 -1.47766098e-01
1.60736227e+00 -1.11940995e-01 -3.66134763e-01 -1.90959692e-01
-7.06005037e-01 -3.44246060e-01 5.79573989e-01 4.41955119e-01
7.43099093e-01 -7.13582277e-01 -2.81719655e-01 4.89794135e-01
-5.19857444e-02 -6.57860041e-01 1.24723306e-02 6.68651044e-01
-7.57881641e-01 5.49911201e-01 -2.39683494e-01 -6.71639442e-01
-1.61745620e+00 1.20943105e+00 3.28984201e-01 -7.11213589e-01
-3.45387697e-01 6.13124013e-01 -3.85101169e-01 -5.08536458e-01
2.81679839e-01 -1.57891273e-01 -3.23691189e-01 4.26725037e-02
8.74399126e-01 2.25980818e-01 -2.16472102e-03 -5.23791015e-01
-4.72531438e-01 -6.39327615e-02 -5.30433655e-01 2.69496620e-01
1.14867246e+00 5.07230580e-01 -4.99415845e-01 5.54661632e-01
1.71268761e+00 1.79982126e-01 -8.48178983e-01 4.93027717e-02
2.90977567e-01 -2.63103932e-01 -4.75478321e-01 -4.36664462e-01
-9.35812533e-01 1.06305552e+00 4.81126577e-01 6.55408502e-01
9.95279253e-01 -1.29582435e-01 1.18200898e+00 6.47865713e-01
3.70300502e-01 -3.11407089e-01 4.40941364e-01 1.20064521e+00
8.17822441e-02 -8.31768394e-01 -6.16729021e-01 2.20398959e-02
-4.95748632e-02 1.57439339e+00 5.40206134e-01 -4.62335169e-01
7.31632411e-01 1.14394438e+00 -2.37170950e-01 -2.09689572e-01
-9.29821849e-01 3.26468170e-01 -2.02325732e-01 7.71521091e-01
1.91074282e-01 -6.25150129e-02 1.23316646e-01 2.92639136e-01
-5.99411950e-02 -6.75644636e-01 8.79813552e-01 1.24238217e+00
-5.13093531e-01 -1.04329550e+00 -7.89280057e-01 9.11526561e-01
-9.77032185e-01 -5.17475605e-01 -2.69203067e-01 3.76212895e-01
4.56126370e-02 8.58383536e-01 1.77543983e-01 -1.02433550e+00
-3.54520261e-01 3.98681492e-01 -5.10202907e-02 -7.61158466e-01
-1.36057675e+00 -3.64295989e-01 -2.13309601e-01 -6.11983478e-01
3.96114796e-01 -5.52770495e-01 -1.30442166e+00 -6.69604123e-01
-7.09765479e-02 -1.07856080e-01 6.34305477e-01 9.30662811e-01
2.72755831e-01 8.83927822e-01 8.61176670e-01 -9.19654727e-01
-1.09120941e+00 -1.03767598e+00 -2.37577707e-01 4.15669948e-01
6.77726328e-01 -4.05606121e-01 -7.74814487e-01 1.63672894e-01] | [14.394762992858887, 9.659435272216797] |
08005743-93fc-4ec4-a05f-2a3a41ff1c53 | tf-gnn-graph-neural-networks-in-tensorflow | 2207.03522 | null | https://arxiv.org/abs/2207.03522v1 | https://arxiv.org/pdf/2207.03522v1.pdf | TF-GNN: Graph Neural Networks in TensorFlow | TensorFlow GNN (TF-GNN) is a scalable library for Graph Neural Networks in TensorFlow. It is designed from the bottom up to support the kinds of rich heterogeneous graph data that occurs in today's information ecosystems. Many production models at Google use TF-GNN and it has been recently released as an open source project. In this paper, we describe the TF-GNN data model, its Keras modeling API, and relevant capabilities such as graph sampling, distributed training, and accelerator support. | ['Bryan Perozzi', 'David Wong', 'Lisa Wang', 'Kevin Villela', 'Anton Tsitsulin', 'Jennifer She', 'Mihir Paradkar', 'John Palowitch', 'Vahab Mirrokni', 'Brandon Mayer', 'André Linhares', 'Silvio Lattanzi', 'Filipe Miguel Gonçalves de Almeida', 'Jonathan Halcrow', 'Neslihan Bulut', 'Peter Battaglia', 'Sami Abu-El-Haija', 'Sibon Li', 'Alvaro Sanchez-Gonzalez', 'Jan Pfeifer', 'Dustin Zelle', 'Martin Blais', 'Arno Eigenwillig', 'Oleksandr Ferludin'] | 2022-07-07 | null | null | null | null | ['graph-sampling'] | ['graphs'] | [-7.63839126e-01 -1.85604066e-01 -5.64844251e-01 -3.32018971e-01
3.88440222e-01 -3.19172978e-01 6.14629269e-01 1.02133594e-01
-2.01171748e-02 3.53687167e-01 2.40773141e-01 -1.05063331e+00
-6.87121879e-03 -1.05078530e+00 -4.21743095e-01 -2.35261843e-01
-6.07856333e-01 4.71040398e-01 2.62202710e-01 -4.26041871e-01
-2.13856339e-01 8.57802510e-01 -1.18631446e+00 2.84543991e-01
1.49569824e-01 1.03369093e+00 -1.46826550e-01 1.17492497e+00
-2.34707341e-01 1.57704294e+00 -5.11549950e-01 -5.74451089e-01
4.34877902e-01 1.22488968e-01 -6.95465386e-01 -6.97577000e-01
7.62028396e-01 -3.98856193e-01 -9.95903313e-01 1.04276323e+00
4.59479719e-01 2.91873395e-01 -2.10892648e-01 -1.70644951e+00
-8.97488952e-01 8.67722213e-01 1.00202471e-01 6.50993347e-01
-5.09083942e-02 3.85024041e-01 9.30369854e-01 -6.43985391e-01
9.66026187e-01 1.16760385e+00 1.10393584e+00 4.06973571e-01
-8.56035471e-01 -4.40454602e-01 -1.24475442e-01 8.82813111e-02
-1.04535198e+00 -3.86185050e-01 6.94161177e-01 -3.86484832e-01
1.58221304e+00 1.88543424e-01 1.04788804e+00 1.42681384e+00
6.18146777e-01 5.61238229e-01 5.48488438e-01 -1.32944480e-01
3.32007200e-01 -2.96586186e-01 5.73840439e-01 9.78694856e-01
4.41946149e-01 3.77116613e-02 -6.65179253e-01 -5.67812920e-01
9.35656428e-01 7.01055005e-02 1.34498239e-01 -2.70930171e-01
-1.04208708e+00 8.03347886e-01 1.15780115e+00 1.76097870e-01
-5.69483459e-01 9.06322718e-01 1.05153370e+00 3.88686746e-01
8.66678655e-01 1.68884799e-01 -4.16518867e-01 -2.42397562e-01
-6.17030144e-01 1.31997332e-01 1.23890412e+00 1.34122026e+00
5.91451943e-01 6.01975203e-01 8.47099498e-02 5.32002151e-01
4.32678223e-01 -1.72758556e-03 5.06054044e-01 -7.41950154e-01
4.19351190e-01 7.65375018e-01 -8.06817770e-01 -1.03828895e+00
-5.03994823e-01 -7.08480000e-01 -1.06035328e+00 -1.47574335e-01
-7.17754650e-04 -3.49288911e-01 -1.14994717e+00 9.49769378e-01
3.30094278e-01 3.83140057e-01 -1.35667354e-01 6.83721840e-01
1.60493064e+00 2.77548015e-01 4.95575368e-02 6.86138928e-01
8.39735150e-01 -1.75285256e+00 -5.08181810e-01 -3.52285236e-01
1.03722858e+00 -3.78902972e-01 8.41233015e-01 7.72583112e-02
-9.11345005e-01 -3.00291777e-01 -9.12441909e-01 -4.99947011e-01
-9.50758874e-01 -3.46207201e-01 1.59888327e+00 7.70630717e-01
-1.96987057e+00 9.67696846e-01 -1.14989233e+00 -7.83865750e-01
3.54716629e-01 2.39704624e-01 -5.06982923e-01 -3.62067431e-01
-1.13365054e+00 5.71126997e-01 5.46436012e-01 1.40265197e-01
-1.01355314e+00 -6.20484114e-01 -6.99939787e-01 1.58078372e-01
1.84508592e-01 -9.64988708e-01 1.24911928e+00 -5.72687089e-01
-1.20441949e+00 4.05883819e-01 4.74555552e-01 -6.67798877e-01
1.33144483e-01 3.61488640e-01 -8.72787356e-01 -1.89260632e-01
-3.62718165e-01 2.48730302e-01 3.34425360e-01 -2.42199153e-01
1.04655758e-01 -3.61247391e-01 1.72559768e-01 -9.54939276e-02
-3.25411558e-01 1.34188741e-01 -5.63066959e-01 -3.36913496e-01
-2.28628993e-01 -8.34545612e-01 -5.33283114e-01 -1.55208513e-01
-7.74330139e-01 -2.42592841e-01 9.02977705e-01 -5.07409990e-01
9.93248105e-01 -2.13787031e+00 -2.80996442e-01 4.26356405e-01
1.13905430e+00 3.30706060e-01 -2.91524976e-01 7.20014334e-01
-1.07605703e-01 2.30113968e-01 6.14264071e-01 -4.30343568e-01
5.64438626e-02 3.04171294e-01 1.74446002e-01 3.50884706e-01
-3.55663300e-01 1.08553016e+00 -9.12593126e-01 -1.77587792e-01
-3.15578878e-02 6.94010735e-01 -6.14819825e-01 2.95146797e-02
-4.53915924e-01 -1.53677925e-01 -2.68512517e-01 9.02193487e-01
5.33947110e-01 -9.46697235e-01 4.51028615e-01 -1.02185085e-01
1.81836337e-01 4.54865128e-01 -6.81198716e-01 1.65626013e+00
-1.09098114e-01 8.02871585e-01 3.05009723e-01 -4.30572033e-01
9.73377287e-01 4.09199148e-02 2.91542262e-01 -4.86502051e-01
2.48187691e-01 7.63464719e-02 5.90173900e-02 -2.02034310e-01
1.12519693e+00 4.10025030e-01 3.47511917e-01 3.09219420e-01
7.38188684e-01 3.51562798e-01 4.62162316e-01 8.22605848e-01
1.67720854e+00 -3.07085216e-01 -3.10522644e-03 -4.31355685e-01
-1.54606491e-01 3.33783850e-02 1.46895081e-01 9.02779937e-01
-2.16843590e-01 -7.05344975e-02 7.51627862e-01 -1.09276950e+00
-1.16773582e+00 -8.77326548e-01 4.23543394e-01 1.21365273e+00
-5.07168412e-01 -1.37244403e+00 -5.62023580e-01 -7.77134717e-01
1.28767267e-01 4.55005229e-01 -4.42445904e-01 -1.57916080e-02
-1.18647389e-01 -5.28855145e-01 6.31109178e-01 3.72710258e-01
5.06242335e-01 -1.28085458e+00 2.06381813e-01 1.31660961e-02
6.02866352e-01 -1.11438143e+00 -3.92941535e-01 1.64912209e-01
-1.05969405e+00 -1.02767479e+00 -2.21142069e-01 -5.33261955e-01
2.99012333e-01 4.31456119e-01 1.77153826e+00 5.12222767e-01
-2.04393238e-01 2.57838488e-01 -2.12970823e-01 -2.51194209e-01
-3.18716943e-01 6.79419458e-01 -1.39491946e-01 -3.90360504e-01
2.91988581e-01 -8.49019825e-01 -4.18756515e-01 -2.68047839e-01
-5.76725960e-01 1.28377721e-01 8.71610940e-02 4.04755741e-01
2.80165702e-01 1.57536156e-02 -4.42146361e-02 -1.28082395e+00
1.10004199e+00 -1.02446461e+00 -7.10502923e-01 7.94850290e-02
-8.78505349e-01 -1.26934499e-01 8.33262920e-01 -1.11073077e-01
-2.59203255e-01 -4.77091163e-01 -1.00519903e-01 -8.02830219e-01
2.39220724e-01 1.09787202e+00 2.72954375e-01 -5.26327074e-01
9.35112774e-01 -2.48040438e-01 2.43575290e-01 -5.82053065e-01
3.71557623e-01 4.35725451e-01 1.44360095e-01 -4.49664325e-01
1.90757319e-01 1.55582055e-01 2.12921113e-01 -7.22168982e-01
-4.71981108e-01 -3.26760620e-01 -6.12189472e-02 -3.91434640e-01
4.75832999e-01 -9.16309476e-01 -8.50268364e-01 6.17885709e-01
-9.72797453e-01 -8.72369289e-01 -2.12197572e-01 4.63559479e-01
1.22157224e-01 -7.52148554e-02 -1.11323762e+00 -2.46651962e-01
-1.12643898e+00 -1.01471138e+00 5.58904290e-01 2.36776412e-01
3.69357347e-01 -1.55451918e+00 2.18890339e-01 -9.39248279e-02
1.30634522e+00 2.49782056e-01 6.63163543e-01 -1.03977704e+00
-8.37357879e-01 -3.64836127e-01 -6.54032350e-01 5.02767980e-01
-3.75816762e-01 4.69597042e-01 -5.12055278e-01 -6.13397837e-01
-4.78662461e-01 -2.33367637e-01 5.03119528e-01 3.66972834e-01
1.25615764e+00 -4.97180611e-01 -3.27400088e-01 1.23966515e+00
1.71781766e+00 -2.75971204e-01 3.84231508e-01 2.29257405e-01
1.35907555e+00 -2.10210192e-03 -5.22220135e-01 1.94024935e-01
6.17862344e-01 1.07853159e-01 8.23798537e-01 -2.66741186e-01
-3.04190874e-01 -2.95434535e-01 1.87295273e-01 1.39839458e+00
-2.05432817e-01 -3.63052547e-01 -1.31840599e+00 1.05277203e-01
-1.89680493e+00 -8.36128235e-01 -7.13609517e-01 1.67556083e+00
1.90390572e-01 2.31269583e-01 1.87590539e-01 -6.68446958e-01
6.41190588e-01 5.76515794e-01 -6.22881114e-01 -7.66488254e-01
2.49233823e-02 3.11839998e-01 7.86239326e-01 2.70788044e-01
-5.50479233e-01 1.01250672e+00 7.35538197e+00 4.31995630e-01
-1.29072690e+00 4.49587435e-01 4.72342700e-01 -2.22948864e-01
-2.85793185e-01 2.13701770e-01 -7.88646758e-01 3.60265613e-01
1.84081542e+00 -6.39956772e-01 1.09545422e+00 1.49955821e+00
-3.10769349e-01 3.90574902e-01 -7.96498299e-01 7.38811851e-01
-2.36382801e-02 -2.06752157e+00 4.35245261e-02 3.80254894e-01
3.49407583e-01 1.54655576e+00 -3.14658433e-01 7.14736223e-01
1.13273597e+00 -9.74639475e-01 3.89070481e-01 4.50973511e-01
8.48888159e-01 -7.31535733e-01 7.59586573e-01 9.61821824e-02
-1.14952290e+00 1.73761651e-01 -5.20087242e-01 -3.03640157e-01
-1.96631681e-02 1.07807648e+00 -1.04805243e+00 5.42410374e-01
8.97530437e-01 1.01020813e+00 -1.06256711e+00 1.12699747e+00
1.80969872e-02 6.59683108e-01 -3.10801029e-01 -4.07758415e-01
2.49346748e-01 -2.76988357e-01 3.22744727e-01 1.10277390e+00
2.78907269e-01 -5.33576787e-01 2.73842096e-01 7.07317948e-01
-6.59095883e-01 7.25421533e-02 -1.06702375e+00 -3.96605283e-01
5.64372659e-01 1.63757265e+00 -7.36399591e-01 -4.58499402e-01
-3.62637669e-01 6.28399968e-01 8.85803878e-01 2.20875591e-01
-7.36927748e-01 -4.79945928e-01 6.49702847e-01 2.07773540e-02
-8.44404101e-02 -5.85162044e-01 1.82305679e-01 -1.31018043e+00
-2.37995312e-01 -7.98373699e-01 3.91424537e-01 -1.11868310e+00
-1.39522302e+00 1.21079409e+00 -4.69523370e-01 -3.77016068e-01
1.12310378e-02 -9.84216452e-01 -7.17606187e-01 9.67469335e-01
-1.03951252e+00 -1.32972229e+00 -7.13528156e-01 9.11441207e-01
-1.81568727e-01 -4.96944994e-01 9.82694864e-01 4.90672857e-01
-8.95224690e-01 3.27222794e-01 1.59768909e-01 3.24142307e-01
1.37053385e-01 -1.28072429e+00 1.53136110e+00 8.54456604e-01
5.48865721e-02 9.53945816e-01 3.40289652e-01 -8.89888525e-01
-1.87955773e+00 -1.41622138e+00 8.02386582e-01 -1.54836237e-01
1.22865105e+00 -8.22858512e-01 -6.47386670e-01 1.38981545e+00
2.74857908e-01 9.05362487e-01 5.21487117e-01 7.11199224e-01
-6.53366864e-01 -2.24822745e-01 -9.75990951e-01 6.02192342e-01
1.19232893e+00 -8.15433860e-01 5.02720237e-01 1.07674277e+00
1.00220156e+00 -7.43994236e-01 -1.22597861e+00 -7.92283043e-02
2.45110422e-01 -1.07805979e+00 6.28068864e-01 -7.52338767e-01
1.42217800e-01 9.00819153e-02 -1.54520914e-01 -1.32830477e+00
-5.80165803e-01 -9.90358889e-01 -7.33542979e-01 7.91694164e-01
4.01697397e-01 -9.13562298e-01 1.21781552e+00 4.07947749e-01
-3.10731024e-01 -6.80786610e-01 -7.32380986e-01 -8.18347871e-01
-3.37921709e-01 -5.52555144e-01 1.10208011e+00 1.11353290e+00
-1.32473260e-01 5.96094668e-01 -1.56737208e-01 -1.66209236e-01
6.28684878e-01 -5.44977367e-01 1.02749348e+00 -1.35490394e+00
-4.86301929e-01 -3.52440715e-01 -6.37249947e-01 -5.01252532e-01
-1.22259080e-03 -1.48396921e+00 -7.45122671e-01 -1.81305695e+00
-2.00009868e-01 -5.99257350e-01 -2.92592436e-01 6.98475659e-01
5.10556340e-01 1.11485496e-01 2.80111372e-01 2.76379734e-01
-6.10798120e-01 1.49630517e-01 1.01378953e+00 -1.71714500e-01
9.20860171e-02 -2.85639822e-01 -6.14097178e-01 3.22662771e-01
9.22170520e-01 -3.42927307e-01 -4.71419483e-01 -6.27513230e-01
5.74200392e-01 -6.01868890e-02 3.94636273e-01 -1.24842846e+00
5.27271330e-01 1.19012237e-01 2.02234894e-01 -3.56036603e-01
3.04382801e-01 -5.22296727e-01 5.60900688e-01 2.04612896e-01
1.37113661e-01 9.67148423e-01 8.37298334e-02 2.56617785e-01
1.20534390e-01 2.73035020e-01 2.95469284e-01 -2.92545527e-01
-5.87892056e-01 9.05752003e-01 6.35498911e-02 -9.19306427e-02
7.29857028e-01 1.68975338e-01 -1.27114379e+00 -4.45495039e-01
-5.81110835e-01 5.09332716e-02 5.11582196e-01 3.94145668e-01
4.33665305e-01 -1.39772999e+00 -4.35778707e-01 5.27583778e-01
1.18670687e-02 -1.73779249e-01 2.95284212e-01 1.02150488e+00
-1.26382279e+00 4.15977150e-01 -2.82963097e-01 -2.31643498e-01
-9.02155876e-01 8.31799030e-01 6.36590779e-01 -5.64989924e-01
-1.00491977e+00 8.55152905e-01 -8.18522334e-01 -8.77820909e-01
2.81076312e-01 -3.69592994e-01 2.07992271e-01 -5.26646852e-01
4.48926508e-01 4.99887139e-01 7.19476283e-01 -1.54763892e-01
-2.71017194e-01 -5.66259325e-01 -2.18045846e-01 4.04066056e-01
1.42061114e+00 5.58176458e-01 -7.70050883e-01 5.03515303e-01
1.21797597e+00 -4.27199453e-01 -6.20262921e-01 5.06769270e-02
-2.15422526e-01 -2.82709505e-02 5.30650377e-01 -7.03776479e-01
-1.65773475e+00 6.22543633e-01 2.09780231e-01 7.81282365e-01
5.98415434e-01 -7.67674521e-02 8.52449358e-01 4.29239124e-01
5.16078115e-01 -7.79117107e-01 -2.87452221e-01 1.06941247e+00
5.84253550e-01 -5.23723125e-01 2.46168688e-01 -1.76907778e-01
-1.02009915e-01 1.13277650e+00 7.69993663e-01 -3.52534235e-01
1.07564282e+00 3.67564112e-01 -5.72704859e-02 -8.38535249e-01
-1.22570813e+00 2.37906352e-01 -6.63351417e-02 6.31531179e-01
5.33208191e-01 3.03232044e-01 1.43736854e-01 2.38165893e-02
-4.99284118e-01 2.17454031e-01 7.42942095e-01 7.91001618e-01
2.65044689e-01 -8.74464571e-01 1.82406664e-01 1.08853948e+00
-3.36442232e-01 -6.57001555e-01 -3.28695416e-01 8.34727645e-01
-3.40473056e-01 5.21753609e-01 1.28106400e-01 -8.63569081e-01
1.03179947e-01 -3.41041088e-01 2.50456780e-01 -6.17030323e-01
-1.23802066e+00 -6.22716665e-01 6.92415655e-01 -1.09056950e+00
3.33095074e-01 -1.75181888e-02 -8.93917322e-01 -1.34956098e+00
-1.82123721e-01 1.60077885e-01 1.40793967e+00 2.50267148e-01
1.01233244e+00 8.81310821e-01 1.55813426e-01 -1.06784654e+00
-1.31884083e-01 -1.07031941e+00 -9.25581217e-01 -3.77501436e-02
6.26142099e-02 -9.07493457e-02 -6.06168330e-01 -7.77933180e-01] | [6.97572135925293, 5.796433448791504] |
3b45a4ab-f47e-47c3-8b72-ab171535ad07 | deep-generative-models-in-engineering-design | 2110.10863 | null | https://arxiv.org/abs/2110.10863v4 | https://arxiv.org/pdf/2110.10863v4.pdf | Deep Generative Models in Engineering Design: A Review | Automated design synthesis has the potential to revolutionize the modern engineering design process and improve access to highly optimized and customized products across countless industries. Successfully adapting generative Machine Learning to design engineering may enable such automated design synthesis and is a research subject of great importance. We present a review and analysis of Deep Generative Machine Learning models in engineering design. Deep Generative Models (DGMs) typically leverage deep networks to learn from an input dataset and synthesize new designs. Recently, DGMs such as feedforward Neural Networks (NNs), Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and certain Deep Reinforcement Learning (DRL) frameworks have shown promising results in design applications like structural optimization, materials design, and shape synthesis. The prevalence of DGMs in engineering design has skyrocketed since 2016. Anticipating continued growth, we conduct a review of recent advances to benefit researchers interested in DGMs for design. We structure our review as an exposition of the algorithms, datasets, representation methods, and applications commonly used in the current literature. In particular, we discuss key works that have introduced new techniques and methods in DGMs, successfully applied DGMs to a design-related domain, or directly supported the development of DGMs through datasets or auxiliary methods. We further identify key challenges and limitations currently seen in DGMs across design fields, such as design creativity, handling constraints and objectives, and modeling both form and functional performance simultaneously. In our discussion, we identify possible solution pathways as key areas on which to target future work. | ['Faez Ahmed', 'Amin Heyrani Nobari', 'Lyle Regenwetter'] | 2021-10-21 | null | null | null | null | ['design-synthesis'] | ['adversarial'] | [ 1.45057857e-01 1.90663144e-01 -1.78865522e-01 -2.86130141e-02
-1.69908464e-01 -5.94473422e-01 4.50065494e-01 -5.00871778e-01
5.15317082e-01 8.12862098e-01 2.53100246e-01 -3.67844403e-01
-4.76419926e-01 -1.22581995e+00 -6.96138799e-01 -7.34699428e-01
3.04025441e-01 4.53681976e-01 -7.49343634e-01 -3.94676536e-01
1.78662777e-01 8.42272401e-01 -1.44321597e+00 2.87067086e-01
7.37190425e-01 9.50281441e-01 -1.12079298e-02 4.12361205e-01
5.67377210e-02 6.69483304e-01 -6.68835402e-01 -3.35829467e-01
1.15251400e-01 -7.23614514e-01 -5.71181536e-01 7.27867335e-02
-9.30457041e-02 -1.70314550e-01 -4.10814941e-01 4.55904812e-01
1.10226905e+00 2.27520958e-01 9.86341357e-01 -1.17323136e+00
-1.51338458e+00 6.70299172e-01 -2.64901295e-02 -3.89375299e-01
1.90731362e-01 6.61920011e-01 8.58765841e-01 -1.24256277e+00
5.46763599e-01 1.05395460e+00 6.48099184e-01 9.39433217e-01
-1.34905553e+00 -6.49258673e-01 -1.15913220e-01 8.35323930e-02
-1.19785750e+00 -3.36537927e-01 1.38359988e+00 -5.78036726e-01
1.18255699e+00 8.90115947e-02 9.58835781e-01 1.34399712e+00
5.54390609e-01 8.79999220e-01 7.67927885e-01 -3.50195587e-01
6.43186092e-01 -2.47460708e-01 -5.55535197e-01 4.68217909e-01
1.21228501e-01 6.24399543e-01 -3.38318348e-01 6.22285157e-03
1.17937875e+00 -1.83865100e-01 9.90573615e-02 -5.26503444e-01
-8.31499219e-01 1.29726756e+00 5.53382099e-01 2.64734924e-01
-5.32224417e-01 6.25246108e-01 1.22009523e-01 2.11166695e-01
2.53173888e-01 1.14237237e+00 -3.52257133e-01 -3.00655980e-02
-8.44078720e-01 6.97924078e-01 5.52797079e-01 1.03394389e+00
6.49337053e-01 1.10740876e+00 -9.72822756e-02 1.03172529e+00
3.69861722e-01 3.62568021e-01 3.28658432e-01 -9.34184492e-01
9.39830730e-04 5.38816273e-01 -2.21021995e-01 -9.33680713e-01
-3.21618736e-01 -6.71701729e-01 -9.22691405e-01 4.31191206e-01
-3.96365345e-01 -4.13678676e-01 -9.86011744e-01 1.58808100e+00
1.91365153e-01 -4.38084751e-01 -8.74497145e-02 6.93088055e-01
1.08232999e+00 7.33878374e-01 5.51948249e-02 1.28898220e-02
6.99710727e-01 -7.26612687e-01 -6.28924429e-01 -1.46083012e-01
1.97172523e-01 -9.27651286e-01 9.80775714e-01 4.33381706e-01
-1.25590670e+00 -7.78249264e-01 -1.15172696e+00 4.79681864e-02
-5.17015874e-01 2.69210517e-01 9.25091147e-01 9.39892054e-01
-9.14381921e-01 8.59079361e-01 -7.27462351e-01 -1.83408288e-03
7.80059278e-01 6.15524411e-01 3.92684072e-01 1.03662364e-01
-1.14412296e+00 7.20212579e-01 1.90781325e-01 3.11897814e-01
-1.28556645e+00 -1.30154002e+00 -7.56856263e-01 -1.32948220e-01
3.71164113e-01 -1.27155221e+00 1.12522602e+00 -6.78980410e-01
-2.15908194e+00 2.66574055e-01 3.79700601e-01 -3.74105237e-02
1.25178888e-01 -1.35915592e-01 -5.66982567e-01 -5.43352008e-01
-2.50946283e-01 6.84091032e-01 9.51262534e-01 -1.24293625e+00
3.18359211e-02 -2.43555624e-02 -2.18672425e-01 -3.86033624e-01
1.18193217e-03 -4.44707900e-01 5.00528336e-01 -1.11587334e+00
-2.71748662e-01 -8.70633721e-01 -4.49432164e-01 -1.40002444e-01
-5.14105737e-01 -2.03804433e-01 9.87053216e-01 -4.50464219e-01
1.46893251e+00 -1.67788136e+00 7.35342324e-01 1.77894980e-01
1.79715857e-01 3.60308439e-01 5.01193898e-03 8.65622342e-01
-2.33984202e-01 2.03505903e-01 -3.07540596e-01 1.34345293e-01
3.33027273e-01 3.54809076e-01 -2.73176879e-01 8.71923342e-02
4.54625219e-01 1.71913195e+00 -6.72286749e-01 2.26064950e-01
5.88761687e-01 5.21818578e-01 -1.03006661e+00 1.73777878e-01
-7.20006824e-01 5.52531719e-01 -6.25797749e-01 9.24925208e-01
2.39463881e-01 -1.13265216e-01 3.45527008e-02 -4.65192258e-01
-1.37919486e-01 4.25921269e-02 -1.03966820e+00 1.49168742e+00
-8.11645210e-01 4.66245353e-01 -2.61681437e-01 -1.05469763e+00
1.22275925e+00 2.99989015e-01 6.49255753e-01 -5.93926907e-01
2.15724558e-01 2.84413010e-01 1.00183904e-01 -3.78736109e-01
2.75792271e-01 -2.06397355e-01 -2.28201911e-01 3.87344003e-01
2.28437886e-01 -7.32377231e-01 -6.22638315e-02 -4.62011606e-01
7.77372003e-01 4.17797625e-01 1.61287412e-01 -1.44507706e-01
4.29689944e-01 -1.10360943e-01 3.75476658e-01 2.69549608e-01
3.60065579e-01 4.40929294e-01 2.09681734e-01 -7.24592388e-01
-1.44320786e+00 -1.25250793e+00 1.10676244e-01 6.90337956e-01
-5.82039416e-01 -3.95646453e-01 -4.43826467e-01 -2.09753394e-01
1.66801527e-01 9.16235626e-01 -6.77048385e-01 -6.71002984e-01
-7.71381736e-01 -7.80117810e-01 3.52843255e-01 8.38084340e-01
2.39201680e-01 -1.34730661e+00 -3.03297907e-01 5.27473092e-01
4.47969884e-01 -5.78697741e-01 -1.35262191e-01 7.32484162e-02
-8.25946987e-01 -7.51373410e-01 -8.96728098e-01 -9.73763347e-01
3.98472667e-01 -5.03138363e-01 1.26021683e+00 -3.59453708e-01
-7.09302783e-01 4.02896076e-01 -7.96257481e-02 -4.35686916e-01
-8.23656678e-01 2.51201063e-01 5.65248355e-02 -3.86705548e-01
-5.68465889e-02 -9.77870166e-01 -7.93394208e-01 3.41738820e-01
-8.78680766e-01 1.15204200e-01 6.20995104e-01 9.59834933e-01
6.45164192e-01 6.55423552e-02 1.19832385e+00 -5.45553505e-01
7.87122846e-01 -4.87664521e-01 -4.71211612e-01 1.93694681e-01
-9.40935850e-01 2.76539296e-01 7.34833002e-01 -4.23836529e-01
-1.03564024e+00 -1.37510508e-01 -6.03571594e-01 -4.21920031e-01
9.32882831e-04 5.72077394e-01 -5.05011857e-01 -1.88711822e-01
5.83114028e-01 -1.79846107e-03 1.35415122e-01 -4.67771769e-01
7.37088919e-01 4.16494489e-01 1.21547468e-01 -9.27990794e-01
8.03376257e-01 -1.14609160e-01 3.17108989e-01 -7.37439454e-01
-1.91460818e-01 5.79037309e-01 -2.20784962e-01 -3.83836597e-01
6.29491746e-01 -4.29174602e-01 -7.12075591e-01 3.94476026e-01
-8.07122946e-01 -3.96875232e-01 -8.88297617e-01 1.36169985e-01
-1.00883758e+00 -1.86925188e-01 -5.09629130e-01 -6.04573905e-01
-5.45602024e-01 -1.50096142e+00 8.85179996e-01 3.53319138e-01
-6.93269551e-01 -1.19651532e+00 1.00554436e-01 1.15640484e-01
9.17898297e-01 7.82754302e-01 1.52980578e+00 4.90136482e-02
-5.78363478e-01 5.91187812e-02 5.20423830e-01 5.14157593e-01
4.95764345e-01 3.12281042e-01 -7.64038205e-01 2.94243917e-02
-2.18770981e-01 -1.78892300e-01 4.72119898e-01 8.61966610e-01
1.29016364e+00 -2.88119346e-01 -1.80931360e-01 6.03895009e-01
1.34442413e+00 6.67210817e-01 7.83120155e-01 -3.97606455e-02
9.22734618e-01 3.79482836e-01 3.24695371e-02 6.54155076e-01
-2.26666689e-01 5.63338220e-01 3.18323553e-01 -1.16265446e-01
-4.90215451e-01 -4.65037733e-01 2.27251500e-01 7.07821727e-01
-2.98445195e-01 -3.33844841e-01 -6.54627204e-01 3.34746867e-01
-1.56197131e+00 -8.18025231e-01 2.15721697e-01 1.67834115e+00
6.98850095e-01 -1.44730419e-01 2.00543150e-01 1.63317427e-01
5.86458802e-01 5.01567647e-02 -1.02427542e+00 -8.05274427e-01
-1.75326705e-01 1.19095767e+00 2.09828079e-01 -3.42516750e-02
-6.59827530e-01 8.62130225e-01 7.19890642e+00 8.76589954e-01
-1.13089955e+00 -2.31769949e-01 7.33805835e-01 -1.91894904e-01
-8.32009256e-01 -1.21289834e-01 -5.09081542e-01 3.47866327e-01
8.09586704e-01 -2.42482096e-01 6.80924714e-01 1.18120754e+00
2.57209420e-01 5.71121752e-01 -1.21912014e+00 8.42602968e-01
-2.46773079e-01 -2.01578641e+00 3.10336918e-01 3.51426065e-01
1.38666749e+00 -5.63098490e-01 7.03604579e-01 3.33538055e-01
3.88617933e-01 -1.36935866e+00 6.49361551e-01 5.32701850e-01
7.95036793e-01 -1.12731814e+00 2.82142699e-01 -3.75835210e-01
-1.18195844e+00 -1.95706010e-01 -9.72748250e-02 -1.78782325e-02
3.86474311e-01 5.40213406e-01 -5.86003542e-01 5.47806621e-01
2.40276232e-01 9.24347878e-01 -2.20642924e-01 6.89210236e-01
-2.28869468e-01 5.34875989e-01 -1.22805703e-02 -2.93303818e-01
1.32849157e-01 -2.61926144e-01 4.67336267e-01 4.96361047e-01
5.54081976e-01 -2.27547213e-01 -1.70889646e-01 1.85168064e+00
-1.69825573e-02 -2.54279613e-01 -4.85408872e-01 -7.34571815e-01
3.59302580e-01 8.47113013e-01 -4.12268430e-01 8.93411264e-02
-1.10445738e-01 4.33675051e-01 -3.48386616e-01 5.88914394e-01
-1.13969278e+00 -2.95904607e-01 8.96954894e-01 3.20044219e-01
3.56373429e-01 -3.97624999e-01 -6.16896868e-01 -5.64282060e-01
-3.46794397e-01 -1.07013535e+00 -2.16909312e-02 -8.06553006e-01
-1.31304491e+00 2.63548195e-01 -8.67429562e-03 -1.12503099e+00
-6.22578979e-01 -8.06771696e-01 -5.09837091e-01 7.04162240e-01
-9.99315441e-01 -1.30960512e+00 3.31589789e-03 1.75889656e-01
8.38653147e-01 -6.35476530e-01 6.81974113e-01 3.82881939e-01
-7.48657107e-01 6.11857176e-01 1.35758519e-01 -1.36914000e-01
2.49505922e-01 -8.58536124e-01 7.42266119e-01 3.61891538e-01
-1.48697302e-01 5.91234684e-01 7.08631217e-01 -7.21589267e-01
-1.85962427e+00 -1.11828351e+00 6.67809471e-02 -2.17723295e-01
4.48488683e-01 -1.50949150e-01 -5.61244965e-01 4.62766141e-01
2.18363091e-01 -4.84294534e-01 9.19886827e-01 -2.29008928e-01
2.39385277e-01 3.96132022e-02 -1.14068413e+00 6.05618715e-01
1.22735894e+00 -3.29248250e-01 -1.89408898e-01 7.50994086e-02
7.09252059e-01 -4.37310070e-01 -1.34069514e+00 6.65795088e-01
7.48324454e-01 -6.11194432e-01 1.31265306e+00 -8.07849288e-01
1.02950847e+00 -1.38375476e-01 -1.48031399e-01 -1.46651626e+00
-7.37128556e-01 -9.09109712e-01 -6.34285033e-01 1.22840774e+00
3.42888176e-01 -4.04635787e-01 9.36422348e-01 7.90163517e-01
-7.74880826e-01 -1.21724749e+00 -5.27655542e-01 -7.29825795e-01
4.81031597e-01 -2.97218859e-01 1.11492741e+00 5.06111860e-01
-5.49814105e-01 2.13190332e-01 -4.14840847e-01 -4.09908980e-01
3.34513426e-01 2.82282144e-01 6.63439989e-01 -9.93613601e-01
-4.04785246e-01 -9.44302797e-01 -2.82279909e-01 -8.05267096e-01
6.03092350e-02 -8.91398430e-01 -3.58483225e-01 -1.69819057e+00
-4.67462957e-01 -3.46428543e-01 4.13779430e-02 2.50540763e-01
3.61815602e-01 3.12766954e-02 -1.53152183e-01 -2.81020641e-01
4.09224361e-01 1.16015708e+00 1.68748641e+00 -3.68335307e-01
-5.10022759e-01 1.70389965e-01 -9.82876897e-01 3.02181125e-01
8.37406158e-01 2.49021978e-04 -5.31952083e-01 -1.66939571e-01
6.41562104e-01 -1.92944601e-01 2.87441581e-01 -8.87217224e-01
-2.08108902e-01 -4.13295448e-01 8.39458585e-01 -4.75193441e-01
1.22502320e-01 -6.29387438e-01 9.52040076e-01 5.70659280e-01
-2.34246969e-01 1.81739464e-01 3.23749810e-01 2.74098396e-01
1.27458246e-02 -2.87428573e-02 7.62683630e-01 -1.20841831e-01
-7.35974193e-01 2.85081446e-01 -6.61054969e-01 -2.48404965e-01
9.99766827e-01 -3.57285589e-01 -2.12326143e-02 -2.27250591e-01
-8.87613595e-01 -3.59740667e-02 2.66715407e-01 6.85477972e-01
8.32125247e-01 -1.88195324e+00 -4.37793702e-01 3.97568971e-01
-1.71378985e-01 -7.91298226e-02 4.30291951e-01 1.09351993e-01
-5.96723914e-01 4.31093276e-01 -4.80109602e-01 -3.01266551e-01
-5.24406552e-01 7.65327811e-01 4.01829720e-01 1.59997381e-02
-2.75122344e-01 6.75763190e-01 9.67605412e-02 -2.67872244e-01
-1.83424681e-01 -4.71620560e-01 2.43510604e-01 -1.20506950e-01
-1.57674998e-02 8.13966811e-01 1.25459954e-01 -2.49180287e-01
-2.18686461e-01 7.50157118e-01 2.39128768e-01 2.26645350e-01
1.72836137e+00 2.73033917e-01 1.56743601e-01 8.52001458e-02
1.17661834e+00 -4.65567976e-01 -1.21201742e+00 2.01336876e-01
-4.33328211e-01 5.38960434e-02 2.65148789e-01 -8.47917855e-01
-1.58128524e+00 7.11131275e-01 6.42411649e-01 -5.08870780e-02
1.19403696e+00 1.06238201e-01 1.03979123e+00 -2.75295116e-02
2.42604926e-01 -1.25963497e+00 5.77522814e-01 3.07981193e-01
1.39802158e+00 -6.83471501e-01 1.20783657e-01 -5.65814897e-02
-3.29957813e-01 1.34078228e+00 4.85416621e-01 -3.60673517e-01
9.39790487e-01 4.20801967e-01 -5.69995821e-01 -3.41291279e-01
-4.66732025e-01 3.55471000e-02 5.91840625e-01 8.07728708e-01
5.82271576e-01 -3.35369408e-02 -3.80080491e-02 8.06307912e-01
-4.61411834e-01 8.56939629e-02 7.02114627e-02 1.17915785e+00
-1.97457850e-01 -1.65152502e+00 -1.63851410e-01 4.26922411e-01
4.74655069e-02 1.38839958e-02 -5.30441642e-01 7.00738311e-01
2.52120316e-01 7.51838386e-01 -2.34055102e-01 -5.71973085e-01
5.14268875e-01 -5.59881516e-02 8.02025616e-01 -5.34636855e-01
-6.82972908e-01 5.14917150e-02 -1.32945105e-01 -3.36092114e-01
-2.73483664e-01 -3.53129834e-01 -8.52205217e-01 -5.87988436e-01
-2.78518826e-01 -2.07962826e-01 6.76192999e-01 7.17192888e-01
7.31940866e-01 1.17946947e+00 5.93777299e-01 -9.86595154e-01
-1.14874840e-01 -5.41939974e-01 -2.89999068e-01 -3.00420195e-01
-5.16047217e-02 -9.67985272e-01 6.58468828e-02 1.22206517e-01] | [5.80819034576416, 3.299329996109009] |
e0d9aa05-714d-46e1-a62f-ebb0403533d3 | metacomp-learning-to-adapt-for-online-depth | 2207.10623 | null | https://arxiv.org/abs/2207.10623v1 | https://arxiv.org/pdf/2207.10623v1.pdf | MetaComp: Learning to Adapt for Online Depth Completion | Relying on deep supervised or self-supervised learning, previous methods for depth completion from paired single image and sparse depth data have achieved impressive performance in recent years. However, facing a new environment where the test data occurs online and differs from the training data in the RGB image content and depth sparsity, the trained model might suffer severe performance drop. To encourage the trained model to work well in such conditions, we expect it to be capable of adapting to the new environment continuously and effectively. To achieve this, we propose MetaComp. It utilizes the meta-learning technique to simulate adaptation policies during the training phase, and then adapts the model to new environments in a self-supervised manner in testing. Considering that the input is multi-modal data, it would be challenging to adapt a model to variations in two modalities simultaneously, due to significant differences in structure and form of the two modal data. Therefore, we further propose to disentangle the adaptation procedure in the basic meta-learning training into two steps, the first one focusing on the depth sparsity while the second attending to the image content. During testing, we take the same strategy to adapt the model online to new multi-modal data. Experimental results and comprehensive ablations show that our MetaComp is capable of adapting to the depth completion in a new environment effectively and robust to changes in different modalities. | ['Liping Xie', 'Mingming Gong', 'Wei Ji', 'Shanshan Zhao', 'Yang Chen'] | 2022-07-21 | null | null | null | null | ['depth-completion'] | ['computer-vision'] | [ 5.04533410e-01 2.72677224e-02 -6.42189905e-02 -3.72203410e-01
-4.36476886e-01 -3.99375111e-01 2.91089058e-01 -1.34250760e-01
-4.45832878e-01 5.69253862e-01 8.76169205e-02 4.66061048e-02
1.05280630e-01 -8.87840450e-01 -8.31502795e-01 -7.42128372e-01
1.82958096e-01 3.81134808e-01 4.11869526e-01 -1.23102508e-01
2.50690550e-01 4.30504769e-01 -2.04409790e+00 4.96661991e-01
9.28192437e-01 9.40390050e-01 5.70307255e-01 8.04921448e-01
-1.49254441e-01 7.38330007e-01 -3.58898818e-01 6.44612387e-02
5.18164217e-01 -3.54755878e-01 -8.00485551e-01 4.66029644e-01
5.79978287e-01 -5.06816328e-01 -4.54983026e-01 9.17486310e-01
6.24069691e-01 2.27091342e-01 2.11408913e-01 -9.31647301e-01
-2.21621647e-01 1.67833835e-01 -3.93987089e-01 9.08314949e-04
5.05376220e-01 3.13822895e-01 3.14835370e-01 -9.11220431e-01
6.40153706e-01 8.76237035e-01 4.64567214e-01 9.92760718e-01
-1.11387110e+00 -4.34113055e-01 4.28758413e-01 2.56382465e-01
-1.05637395e+00 -5.63884914e-01 9.15323496e-01 -1.99670032e-01
7.93616056e-01 -5.46572246e-02 9.70387161e-01 1.17487276e+00
1.90069359e-02 8.41794968e-01 1.26808369e+00 -5.26722550e-01
5.89605212e-01 1.38145089e-01 -4.92027581e-01 5.45331895e-01
-1.62375331e-01 3.31097603e-01 -9.71009433e-01 2.23708600e-01
7.62643754e-01 -4.13152911e-02 -3.70254785e-01 -6.84192598e-01
-1.18049622e+00 3.84753168e-01 4.37310487e-01 2.78875709e-01
-3.63015056e-01 -1.41111031e-01 2.15649709e-01 4.13178951e-01
2.37811714e-01 3.38860214e-01 -6.79571509e-01 -3.09874594e-01
-8.55568171e-01 -2.64467914e-02 5.69468617e-01 7.38251209e-01
1.02995014e+00 1.98786519e-02 2.18684092e-01 8.43604147e-01
2.68902272e-01 4.49714690e-01 8.86479557e-01 -9.06411350e-01
6.83513105e-01 6.13540351e-01 -2.37475231e-01 -5.95873892e-01
-4.20324117e-01 -3.81295055e-01 -7.64146745e-01 4.83840883e-01
3.91539752e-01 -1.35618433e-01 -1.35260022e+00 1.84457564e+00
6.15285575e-01 2.10051820e-01 1.04740404e-01 7.85902202e-01
6.62743449e-01 3.37515414e-01 -2.28117302e-01 -2.29321286e-01
6.39281332e-01 -8.34774256e-01 -5.54629385e-01 -6.76101744e-01
5.88347018e-01 -5.22393882e-01 1.25858271e+00 4.86405492e-01
-1.28674567e+00 -8.55494797e-01 -1.20160949e+00 1.68000191e-01
-2.68132448e-01 -3.50735694e-01 5.28853655e-01 3.91787499e-01
-1.02982569e+00 6.08853638e-01 -9.62932289e-01 -4.54495281e-01
3.42862725e-01 4.62104946e-01 -5.63833654e-01 -5.17547905e-01
-1.02019513e+00 6.82056129e-01 5.51342547e-01 2.78847098e-01
-9.56512451e-01 -5.20887613e-01 -9.48907137e-01 -2.79135615e-01
2.41134644e-01 -8.84284556e-01 1.23055196e+00 -1.37250972e+00
-1.93085909e+00 7.18286037e-01 -2.05231756e-01 8.07878524e-02
5.00287592e-01 -1.86436903e-02 -2.38432407e-01 2.27468848e-01
-4.22046967e-02 7.68159509e-01 1.07160771e+00 -1.50128019e+00
-7.74748266e-01 -6.96459353e-01 1.46988377e-01 5.61999261e-01
-4.84541595e-01 -8.12149763e-01 -7.36075640e-01 -4.27706152e-01
7.90781975e-01 -9.53676105e-01 -1.57915682e-01 -2.30266037e-03
-4.67537381e-02 5.23685157e-01 7.53458261e-01 -9.44401920e-02
9.59594250e-01 -2.19498825e+00 4.10871476e-01 3.94985974e-01
2.83617619e-02 5.10126650e-02 -2.04439148e-01 2.49914214e-01
-9.68443379e-02 -3.65608245e-01 -2.35198170e-01 -6.54077470e-01
-4.79391724e-01 5.45591116e-01 2.79052891e-02 2.80739188e-01
-3.98999043e-02 6.38837397e-01 -1.09344971e+00 -2.81634361e-01
7.51102269e-02 1.83867499e-01 -8.08128417e-01 6.14586294e-01
-1.68559134e-01 1.05077767e+00 -2.74749011e-01 7.97067761e-01
6.15507364e-01 -1.30331546e-01 2.00257227e-01 -3.59303713e-01
9.21281204e-02 5.28707989e-02 -1.24515152e+00 2.24192286e+00
-7.12829888e-01 3.07311743e-01 -1.32164499e-02 -1.03244901e+00
6.00845516e-01 1.86413720e-01 4.80260432e-01 -1.18097878e+00
-2.22857539e-02 2.39861131e-01 1.61641285e-01 -8.07515979e-01
3.64560515e-01 -2.95657814e-01 2.72740573e-01 4.03236866e-01
1.26205817e-01 -3.57997984e-01 1.92289508e-03 3.33118215e-02
1.11567128e+00 3.10623914e-01 -4.50825281e-02 4.49983507e-01
3.58357310e-01 -1.80710807e-01 6.49235547e-01 8.02308321e-01
-3.03600103e-01 7.30871022e-01 -4.20808271e-02 -4.15985286e-01
-8.55993986e-01 -1.04228604e+00 -6.43090084e-02 9.03666675e-01
2.44404480e-01 -1.61211193e-01 -4.57324594e-01 -7.65594304e-01
-1.41614124e-01 2.66088843e-01 -7.29946911e-01 -5.30714095e-01
-4.68979806e-01 -6.11324906e-01 9.87212509e-02 5.63172460e-01
1.03202212e+00 -1.13506365e+00 -8.18653047e-01 2.69265890e-01
-1.49972498e-01 -1.29725039e+00 -1.79692507e-01 4.51568186e-01
-1.27504802e+00 -9.31648135e-01 -6.19425178e-01 -7.70419121e-01
9.72206235e-01 2.52643138e-01 8.99471045e-01 3.47832479e-02
4.11580317e-02 8.07838559e-01 -4.93968219e-01 -1.22847244e-01
-4.00438756e-01 8.58241394e-02 6.64862767e-02 2.44278520e-01
-1.15341507e-01 -9.11368608e-01 -8.63679230e-01 2.52554208e-01
-1.29715323e+00 2.94681579e-01 6.84387505e-01 8.86565566e-01
7.00997531e-01 2.34824985e-01 4.28650290e-01 -9.37151492e-01
6.72730878e-02 -3.80482972e-01 -1.59026027e-01 7.27740377e-02
-7.06592500e-01 -1.08645763e-02 6.67136014e-01 -6.94362760e-01
-1.10424674e+00 3.64456505e-01 -3.07439148e-01 -3.75715315e-01
-2.13370681e-01 7.47857153e-01 -4.10572916e-01 -4.15547192e-01
7.14312911e-01 3.85865003e-01 -1.72923040e-02 -2.32858628e-01
6.48660734e-02 3.79680842e-01 5.03589094e-01 -6.71583056e-01
1.02676940e+00 6.69302106e-01 -1.09498367e-01 -4.49765801e-01
-9.09797728e-01 -2.61710495e-01 -8.05240989e-01 -4.86740112e-01
6.03019357e-01 -8.30683053e-01 -3.78754944e-01 1.00553453e+00
-6.49258435e-01 -9.44287360e-01 -5.70856214e-01 4.85215068e-01
-6.03649735e-01 3.52679133e-01 -3.50759685e-01 -4.43222463e-01
1.47417977e-01 -1.03838313e+00 9.45466578e-01 2.93141305e-01
2.04909623e-01 -1.28136897e+00 3.34649354e-01 3.37717831e-01
4.94828910e-01 1.73832804e-01 7.58300245e-01 -8.07912946e-02
-5.63486814e-01 -3.45147163e-01 2.03516290e-01 4.40198481e-01
3.83401752e-01 -2.91836619e-01 -1.09039903e+00 -6.10234022e-01
2.59282649e-01 -6.92802727e-01 7.26279378e-01 2.09564697e-02
1.31711340e+00 3.16785760e-02 -2.57890373e-02 9.88025606e-01
1.43024826e+00 4.10240553e-02 8.39873731e-01 8.15542996e-01
7.60898471e-01 6.42609119e-01 6.69787288e-01 4.45810229e-01
5.25941312e-01 5.69841027e-01 7.57754982e-01 -3.90240774e-02
-3.77940945e-02 -4.45437312e-01 4.88927424e-01 9.13830578e-01
1.01686366e-01 8.31241682e-02 -9.25507247e-01 4.66463596e-01
-1.85545766e+00 -8.27351272e-01 5.50274909e-01 2.41206455e+00
8.91290009e-01 4.04189736e-01 -9.54243317e-02 1.81748584e-01
2.83061564e-01 5.31053916e-02 -1.04322684e+00 -1.18804723e-01
-2.14429319e-01 1.57480866e-01 2.51605451e-01 3.09324414e-01
-7.70106494e-01 7.79592693e-01 6.36135578e+00 2.88767934e-01
-1.47487140e+00 -3.04371361e-02 4.39026147e-01 -3.04399252e-01
-3.85858119e-01 1.42279834e-01 -5.23855627e-01 3.89966100e-01
7.11419046e-01 2.73016840e-01 6.42411709e-01 5.94677031e-01
1.78784445e-01 -5.22860348e-01 -1.29913938e+00 1.06399417e+00
2.51484632e-01 -1.02366638e+00 7.75956511e-02 -2.00601667e-01
1.10021186e+00 4.17785160e-02 2.71889001e-01 3.63993734e-01
-1.76187158e-01 -7.12477446e-01 6.34344637e-01 4.96318638e-01
6.88930273e-01 -3.30907345e-01 5.49066126e-01 7.42156446e-01
-9.51040447e-01 -3.48436296e-01 -1.60532549e-01 -1.35028243e-01
-4.44711074e-02 5.88778555e-01 -6.18862152e-01 3.42827708e-01
8.71221960e-01 8.10371876e-01 -6.79273307e-01 1.07724285e+00
-2.38778621e-01 3.09345752e-01 -1.85014278e-01 5.62417924e-01
-1.17788859e-01 1.09699950e-01 3.40787590e-01 6.52916491e-01
2.89672583e-01 -1.97306834e-02 2.04594404e-01 3.57751429e-01
8.20083246e-02 -8.86624008e-02 -6.50517881e-01 2.64636964e-01
3.58822644e-01 9.18450296e-01 -3.89633805e-01 -2.47998118e-01
-4.72370267e-01 1.31451917e+00 5.05434155e-01 6.65594816e-01
-5.21600723e-01 1.00506924e-01 2.54183173e-01 1.93714276e-01
3.40815991e-01 -2.65650272e-01 -3.59127164e-01 -1.40406382e+00
2.15736672e-01 -1.06909513e+00 3.71299297e-01 -9.56515729e-01
-1.18887591e+00 6.12842083e-01 -4.53475453e-02 -1.49582779e+00
-2.81590253e-01 -3.74952942e-01 -5.11424184e-01 6.37541413e-01
-1.77575934e+00 -1.12295878e+00 -7.61024892e-01 9.95052338e-01
5.38979530e-01 -1.53253913e-01 7.73703694e-01 2.43251652e-01
-6.32378697e-01 7.45101631e-01 2.06533130e-02 -3.28007936e-01
8.18744063e-01 -1.12731087e+00 -1.63997471e-01 8.24302554e-01
7.01079071e-02 4.13206726e-01 5.39289296e-01 -5.58387518e-01
-1.51633930e+00 -9.06139553e-01 2.06640497e-01 -1.27130419e-01
4.15841252e-01 -2.32758969e-01 -1.13507795e+00 5.69120765e-01
-1.90794945e-01 2.98227578e-01 6.45848811e-01 3.77329551e-02
-2.46168271e-01 -3.32614899e-01 -1.21321619e+00 6.10894263e-01
1.17079008e+00 -6.81438506e-01 -3.53862792e-01 2.81507939e-01
3.93192083e-01 -9.93566751e-01 -7.78491974e-01 7.28168309e-01
5.88279486e-01 -1.27588081e+00 7.22361207e-01 -3.44638258e-01
5.19285023e-01 -2.10424721e-01 -3.30426782e-01 -1.43751919e+00
-1.51072428e-01 -2.54997551e-01 -4.65453893e-01 9.03656125e-01
2.43165419e-01 -7.01225340e-01 1.18087566e+00 6.13383293e-01
-2.76472211e-01 -8.66433561e-01 -1.02095461e+00 -5.77302694e-01
8.13501794e-03 -5.88689089e-01 4.60075885e-01 8.28637958e-01
-2.80102808e-02 8.81654248e-02 -2.18406767e-01 3.89960229e-01
3.68314177e-01 1.30476266e-01 1.00085664e+00 -9.65471685e-01
-7.20235348e-01 3.47361830e-03 -4.04937536e-01 -1.45787728e+00
1.66926697e-01 -5.71438849e-01 2.72546291e-01 -1.39232647e+00
9.58905369e-02 -6.54601574e-01 -2.85253286e-01 4.93036360e-01
-3.53253305e-01 2.82976538e-01 -3.54408063e-02 4.73109275e-01
-5.33315182e-01 7.80877411e-01 1.50197470e+00 -1.70946211e-01
-4.92775619e-01 1.33059785e-01 -5.19851148e-01 7.01722860e-01
6.78482890e-01 -2.89640784e-01 -7.85305321e-01 -6.88879967e-01
4.48203057e-01 1.45291639e-02 1.43951088e-01 -1.23615408e+00
4.72490162e-01 -1.44110009e-01 5.73284328e-01 -2.76460618e-01
3.91567230e-01 -1.00344968e+00 -8.88998061e-02 3.30534488e-01
-8.27842057e-02 -2.60677245e-02 4.44335669e-01 6.08209550e-01
-4.56832558e-01 -3.04664820e-01 7.46891022e-01 -3.17928016e-01
-1.09860694e+00 6.17630720e-01 -1.85669139e-01 -1.88354962e-02
8.86192739e-01 -8.13559115e-01 -1.28644601e-01 -4.47684079e-01
-1.12662041e+00 2.14056686e-01 8.34294617e-01 2.73521662e-01
9.39006686e-01 -1.36010849e+00 -2.74550915e-01 6.48030639e-01
3.41824025e-01 4.96662498e-01 6.11216426e-01 7.77474046e-01
-3.20865095e-01 -3.17193955e-01 -3.51605833e-01 -8.67372990e-01
-8.03543806e-01 4.22970921e-01 5.76527297e-01 -1.96345165e-01
-4.91022766e-01 7.60630727e-01 1.69204935e-01 -4.50918376e-01
4.42752331e-01 -8.01146701e-02 -5.26998974e-02 -1.99769497e-01
4.05643523e-01 -1.35298684e-01 2.85685211e-01 -3.08294713e-01
-7.06236586e-02 8.18527043e-01 -2.11380884e-01 -2.08436489e-01
1.27202797e+00 -4.51989472e-01 1.63048491e-01 8.72758746e-01
1.06672585e+00 -2.96715694e-03 -1.70440125e+00 -4.70471442e-01
-4.06280071e-01 -7.09370136e-01 2.54590567e-02 -8.04104209e-01
-1.26651871e+00 7.62833536e-01 8.96982491e-01 -7.73461685e-02
1.61218596e+00 -2.27118805e-01 7.93466747e-01 4.28324640e-01
4.97280896e-01 -1.29744995e+00 7.38388538e-01 5.61998725e-01
6.76998377e-01 -1.55702472e+00 -7.89826140e-02 -1.29451290e-01
-4.96147931e-01 1.16743934e+00 1.08235228e+00 1.78716525e-01
6.60590053e-01 9.33051407e-02 2.84598172e-01 -8.90042782e-02
-6.47624791e-01 -1.55735567e-01 1.90720588e-01 8.59517455e-01
3.74922231e-02 -5.04249692e-01 5.21566033e-01 -8.91665146e-02
1.92558039e-02 2.16315109e-02 4.62950706e-01 1.25288880e+00
-3.18102449e-01 -1.26038253e+00 -3.27776462e-01 2.76563704e-01
1.71585619e-01 1.06909022e-01 -1.38674289e-01 7.00159848e-01
3.82373929e-01 7.18329608e-01 7.74779636e-03 -5.80161095e-01
6.61286116e-01 -4.34910618e-02 8.33016634e-01 -9.81968939e-01
-3.27073902e-01 -7.23039359e-02 -3.26279134e-01 -7.96428502e-01
-7.71669269e-01 -6.70590997e-01 -1.26594317e+00 -7.82506764e-02
-1.34010628e-01 -1.88917354e-01 6.56670690e-01 1.20275223e+00
2.11226448e-01 6.20970190e-01 9.91933763e-01 -1.17502201e+00
-4.08556610e-01 -7.31565237e-01 -3.87851655e-01 5.66341281e-01
4.59275335e-01 -6.69035971e-01 -5.29469371e-01 -3.80819254e-02] | [8.6929931640625, -2.337602376937866] |
6c83d004-463e-4263-9a11-52134e126f90 | san-bert-extractive-summarization-for | 2304.01894 | null | https://arxiv.org/abs/2304.01894v1 | https://arxiv.org/pdf/2304.01894v1.pdf | San-BERT: Extractive Summarization for Sanskrit Documents using BERT and it's variants | In this work, we develop language models for the Sanskrit language, namely Bidirectional Encoder Representations from Transformers (BERT) and its variants: A Lite BERT (ALBERT), and Robustly Optimized BERT (RoBERTa) using Devanagari Sanskrit text corpus. Then we extracted the features for the given text from these models. We applied the dimensional reduction and clustering techniques on the features to generate an extractive summary for a given Sanskrit document. Along with the extractive text summarization techniques, we have also created and released a Sanskrit Devanagari text corpus publicly. | ['Mahabala Rao M G', 'Jammi Kunal', 'Sampath Lonka', 'Kartik Bhatnagar'] | 2023-04-04 | null | null | null | null | ['extractive-summarization', 'extractive-document-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.67713219e-01 1.94899529e-01 3.29418406e-02 -3.73333037e-01
-9.69262958e-01 -8.32585871e-01 9.80467379e-01 1.55090794e-01
-4.13700610e-01 1.10952401e+00 1.34478545e+00 -1.69044554e-01
-4.34708476e-01 -5.06219566e-01 -3.06051463e-01 -6.28109217e-01
5.66971526e-02 9.18356299e-01 -2.68932343e-01 -2.66412973e-01
7.06505716e-01 2.96112359e-01 -5.89865446e-01 3.55282336e-01
9.74051297e-01 3.49846750e-01 2.35719889e-01 1.54324996e+00
-2.22692624e-01 1.42763889e+00 -9.64578092e-01 -6.30957425e-01
-1.27838627e-01 -7.53418207e-01 -1.26314247e+00 -4.57873344e-02
3.50152612e-01 -6.01741433e-01 -1.03982985e+00 5.47720253e-01
3.54877174e-01 3.19344908e-01 1.18365479e+00 -7.08293617e-01
-1.10769773e+00 1.28207719e+00 -4.15016234e-01 4.74357784e-01
1.11477464e-01 -5.71402371e-01 1.40084720e+00 -6.03549540e-01
8.70410383e-01 1.03362036e+00 3.49350899e-01 5.57709455e-01
-6.45090342e-01 -2.91585177e-01 -2.92139024e-01 3.12841415e-01
-1.01089239e+00 -6.11788452e-01 9.00330782e-01 -1.78296372e-01
1.34267676e+00 5.81679940e-01 7.66031206e-01 1.07663214e+00
4.31910247e-01 1.55852759e+00 6.32360518e-01 -2.39933848e-01
-6.14790851e-03 -2.55759031e-01 7.28350818e-01 5.65760016e-01
3.01001258e-02 -4.34870422e-01 -4.91643190e-01 -4.59832579e-01
3.71775895e-01 -1.63271755e-01 -1.40918732e-01 3.50865632e-01
-1.10287726e+00 1.17849135e+00 -2.54308656e-02 3.60547334e-01
-6.08193278e-01 9.95569676e-02 7.90515900e-01 4.27738577e-01
6.54264808e-01 6.28615022e-01 -4.72888350e-01 -4.64041293e-01
-1.35221350e+00 2.04346165e-01 8.84543240e-01 9.25518334e-01
3.33642662e-01 1.56979561e-01 -1.80806190e-01 1.03545976e+00
1.97817311e-01 6.45022988e-01 7.06557393e-01 -6.68777406e-01
7.70454526e-01 3.58416289e-01 -2.04343021e-01 -8.52245688e-01
-2.88334996e-01 -9.19287577e-02 -1.16196513e+00 -1.06481314e+00
-4.54920232e-01 -3.15801740e-01 -9.66763318e-01 8.99005651e-01
-2.55379707e-01 1.49248883e-01 8.09793115e-01 4.10280555e-01
1.27627337e+00 1.37345755e+00 -5.17408907e-01 -6.35885298e-01
9.36376810e-01 -1.30559444e+00 -9.60731864e-01 -6.22657547e-03
6.21000588e-01 -8.14486980e-01 5.57130635e-01 5.29934704e-01
-1.01116502e+00 -1.92205030e-02 -8.86313379e-01 -4.57327247e-01
5.86596541e-02 4.42563504e-01 4.33806628e-01 2.74574965e-01
-7.96751142e-01 4.41172510e-01 -8.58735979e-01 -5.91721177e-01
2.76037574e-01 1.93760917e-01 -3.38205129e-01 1.27478287e-01
-1.02537537e+00 8.01772594e-01 9.37472701e-01 1.89411074e-01
-9.89614666e-01 -1.77538455e-01 -8.51121902e-01 -3.17026041e-02
2.24982724e-01 -4.95273530e-01 1.24906361e+00 -7.83939719e-01
-1.82654393e+00 4.07018393e-01 -6.25594407e-02 -7.17727482e-01
5.75288199e-02 -6.19516909e-01 -3.03912252e-01 4.33690459e-01
-2.91173607e-01 1.28015086e-01 8.07401001e-01 -9.87834752e-01
-4.27256793e-01 -3.10343385e-01 -2.76183575e-01 5.51140070e-01
-3.06593537e-01 5.73162735e-01 -3.01747978e-01 -1.05407262e+00
-2.47166008e-02 -7.66758084e-01 -1.14288665e-01 -1.31149983e+00
-9.32602823e-01 -2.46212035e-01 9.34633553e-01 -1.63427114e+00
1.73823464e+00 -1.83374298e+00 5.39862812e-01 -3.06155160e-02
1.35530323e-01 4.89733636e-01 -5.04042625e-01 1.05394709e+00
3.91674459e-01 2.15853095e-01 -3.78366262e-01 -1.16827190e-01
-2.97065794e-01 6.01190746e-01 -7.57542312e-01 2.93332666e-01
4.57985252e-02 1.10248768e+00 -9.05692816e-01 -6.29203200e-01
6.23098761e-02 2.55164027e-01 -4.95521039e-01 4.39157546e-01
-4.97681908e-02 2.55474269e-01 -6.46314561e-01 1.97689921e-01
2.02107787e-01 2.35994130e-01 1.27812654e-01 1.21779591e-01
1.52504593e-01 7.66474187e-01 -5.07734597e-01 1.59518480e+00
-2.75531918e-01 7.15650022e-01 -1.70132607e-01 -8.87857974e-01
1.36285520e+00 3.48957896e-01 4.54148501e-01 -1.30192593e-01
2.14332372e-01 1.55502766e-01 -1.51894450e-01 -7.16641247e-01
1.49351931e+00 6.31712228e-02 -3.76823664e-01 9.83623803e-01
4.07978058e-01 -6.05287611e-01 4.25731927e-01 8.69944394e-01
1.23183560e+00 -1.05628461e-01 7.18256056e-01 -7.10440502e-02
5.89640975e-01 5.93675487e-03 1.91157833e-01 7.27383256e-01
1.44335881e-01 8.31358433e-01 4.93398845e-01 -4.90501821e-01
-1.34948850e+00 -8.94413531e-01 2.29319274e-01 1.00762618e+00
-4.12988782e-01 -1.01788163e+00 -4.90915060e-01 -1.11065948e+00
-4.10543382e-01 1.07160950e+00 -4.69596624e-01 -1.90736175e-01
-9.64603603e-01 -8.13503504e-01 9.55346406e-01 3.79934907e-01
6.67065740e-01 -1.06158698e+00 -1.37687936e-01 2.40912527e-01
-5.38345933e-01 -6.81416571e-01 -7.43591487e-01 2.02929992e-02
-7.47395158e-01 -8.52069080e-01 -6.36118829e-01 -1.01335526e+00
3.25146884e-01 7.95480702e-03 9.45952535e-01 -2.36643016e-01
3.95644486e-01 3.44040871e-01 -7.77769566e-01 -2.29972422e-01
-8.95614862e-01 4.58404928e-01 -1.56593636e-01 -1.88442364e-01
2.59889811e-02 -2.76201129e-01 -8.00094455e-02 -3.82385194e-01
-1.04632640e+00 3.55939120e-01 5.41364729e-01 8.94987881e-01
2.29200274e-01 -7.35445023e-02 6.93069577e-01 -9.88153994e-01
1.18267572e+00 -3.18076581e-01 8.73816013e-02 4.30078536e-01
-1.99318051e-01 3.37681621e-01 7.14036942e-01 -2.40834698e-01
-1.28655756e+00 -1.73060447e-01 -4.37676579e-01 1.48197934e-01
2.65906692e-01 1.01841748e+00 1.42921194e-01 6.31355524e-01
4.73884165e-01 8.14969599e-01 -3.11255246e-01 -5.86395860e-01
7.77334988e-01 1.64498734e+00 4.62844610e-01 -3.63486260e-01
4.33564574e-01 8.34009945e-02 -6.94427967e-01 -1.34157848e+00
-8.14496160e-01 -4.02153939e-01 -8.02965939e-01 1.06995769e-01
7.30263412e-01 -7.49513030e-01 1.02711849e-01 3.91628534e-01
-1.46214187e+00 -2.10688449e-02 -4.10018295e-01 4.29425627e-01
-5.69926858e-01 6.57800615e-01 -1.11232674e+00 -9.96634007e-01
-1.14520741e+00 -4.99486953e-01 1.07901967e+00 -2.67871115e-02
-3.98352414e-01 -9.59908783e-01 6.83317125e-01 4.62795019e-01
9.40157846e-02 -8.68597776e-02 1.06498325e+00 -1.32709706e+00
9.11643438e-05 -7.88111389e-02 -2.68513151e-02 7.83165991e-01
1.59329966e-01 3.01249474e-01 -4.68170494e-01 -3.75904232e-01
1.87453941e-01 -3.98215652e-01 1.13569283e+00 3.59473884e-01
5.20039082e-01 -1.23730981e+00 1.53594762e-01 3.92040521e-01
1.00554550e+00 3.81207556e-01 5.54288328e-01 7.50747621e-02
8.67382109e-01 2.16213092e-01 2.35036820e-01 3.93741190e-01
6.47731602e-01 1.34350255e-01 -2.70495802e-01 5.50947666e-01
2.14538258e-02 -5.23204803e-01 7.73997784e-01 2.27477694e+00
-1.91098139e-01 -5.32416224e-01 -8.36842835e-01 5.84096670e-01
-1.97097576e+00 -1.03416789e+00 -5.55444062e-02 1.44957983e+00
9.28403020e-01 -4.81694609e-01 -7.52725080e-02 5.30072004e-02
3.86892825e-01 6.80009484e-01 -2.59680957e-01 -9.73339677e-01
-3.10804874e-01 3.69164236e-02 3.21873933e-01 6.41058981e-01
-1.06389248e+00 1.16155970e+00 6.65889931e+00 6.93861187e-01
-8.67914021e-01 -9.11403969e-02 3.85848016e-01 -3.65178078e-01
-5.45295954e-01 -1.79426335e-02 -6.06799424e-01 1.93024457e-01
1.47075820e+00 -7.09077835e-01 6.10041559e-01 3.86286199e-01
2.00437665e-01 1.75072491e-01 -8.17940295e-01 9.12271440e-01
8.07138145e-01 -1.56407881e+00 5.30372024e-01 -6.26101643e-02
8.53795111e-01 4.46301639e-01 -4.69104946e-01 6.30366683e-01
6.87973797e-01 -8.73840570e-01 6.01862609e-01 4.50996399e-01
5.20767510e-01 -1.00793517e+00 9.02509093e-01 6.94561362e-01
-7.40365624e-01 -1.94322407e-01 -5.88819325e-01 4.11076322e-02
1.99852571e-01 3.12878758e-01 -1.10580468e+00 9.53775346e-01
7.26249591e-02 1.25295019e+00 -6.56642437e-01 5.38781285e-01
-4.28260624e-01 1.16029012e+00 -7.40301237e-02 -4.26202267e-01
4.97031301e-01 -4.27267373e-01 8.90774190e-01 1.34483218e+00
3.23551923e-01 1.82572931e-01 2.14947965e-02 1.28553882e-01
-3.70316178e-01 4.10241902e-01 -5.12621641e-01 -6.39841974e-01
3.91070575e-01 1.15179408e+00 -3.55288953e-01 -7.06876397e-01
-2.36912537e-02 1.38494015e+00 3.64302188e-01 4.24725622e-01
-5.29584408e-01 -3.57854038e-01 -6.24690292e-05 -4.48887497e-01
2.02582240e-01 -6.70569479e-01 -1.87266409e-01 -1.64517832e+00
-3.24860632e-01 -1.09893548e+00 6.69671237e-01 -8.87780488e-01
-1.16028965e+00 9.19218242e-01 2.16197986e-02 -7.50722051e-01
-4.49965119e-01 -1.41045600e-01 -4.68168497e-01 4.70102072e-01
-1.08999717e+00 -1.38213325e+00 6.14962935e-01 3.90040606e-01
9.94284987e-01 -6.22135699e-01 9.06687856e-01 -2.13978663e-02
-5.74157476e-01 -4.33305316e-02 7.90408075e-01 4.19653624e-01
3.55262786e-01 -1.31012952e+00 5.63122213e-01 9.94921565e-01
7.40012884e-01 7.77005374e-01 6.03615582e-01 -9.14242506e-01
-1.47716808e+00 -1.26033485e+00 1.32754374e+00 -4.38395292e-01
9.16755199e-01 -2.89212137e-01 -7.96825409e-01 1.28138411e+00
8.77268314e-01 -8.57623816e-01 8.06785703e-01 -2.91816611e-02
1.00045814e-03 2.84538507e-01 -7.05716193e-01 5.97668946e-01
7.29908407e-01 -4.53990698e-01 -1.48045135e+00 5.03278077e-01
8.24667275e-01 -1.92638651e-01 -6.86405420e-01 -1.64076146e-02
3.78841430e-01 -4.14456695e-01 3.75804394e-01 -9.95510697e-01
8.28663468e-01 4.36751321e-02 -1.37394562e-01 -1.76482594e+00
-4.64805663e-01 -8.61003399e-01 -4.73796964e-01 1.55826962e+00
3.35965514e-01 -2.16056034e-01 5.39841950e-01 -7.53303170e-02
-2.25645527e-01 -5.34115374e-01 -8.54057670e-01 -5.17622471e-01
2.17862710e-01 -3.42871457e-01 3.62408102e-01 7.50803709e-01
9.69671160e-02 1.14861429e+00 -1.02466881e+00 -1.74738139e-01
6.39188662e-02 1.32246509e-01 7.01497674e-01 -9.85124171e-01
-1.40395433e-01 -1.73376173e-01 -1.97228894e-01 -1.29146826e+00
3.58889073e-01 -1.39326155e+00 -4.01676958e-03 -2.02364182e+00
6.58642232e-01 1.44157186e-01 -3.47532071e-02 2.35931680e-01
-7.98950121e-02 5.71340024e-02 3.08464736e-01 5.49295723e-01
-7.90035129e-01 8.75827491e-01 1.02876115e+00 -3.11463624e-01
-4.02558088e-01 -1.52291209e-01 -9.33833361e-01 5.66067576e-01
8.91413927e-01 -4.76805240e-01 -3.36008817e-01 -7.01277435e-01
4.09316748e-01 2.02749774e-01 -3.23249072e-01 -3.83668572e-01
5.17329834e-02 -9.98973250e-02 7.13369250e-02 -1.05169368e+00
1.51482999e-01 -3.42589825e-01 -6.41993955e-02 2.36388981e-01
-7.94023931e-01 2.69568294e-01 -2.29500756e-01 4.04886842e-01
-3.29632550e-01 -4.86623377e-01 4.34250683e-01 6.25744238e-02
-3.89065653e-01 1.88985378e-01 -7.21553087e-01 -1.99064035e-02
6.80655420e-01 1.21451870e-01 -4.67964292e-01 -6.81332350e-01
-4.80091512e-01 2.44045705e-01 1.08918466e-01 4.58941996e-01
9.03244734e-01 -1.18178630e+00 -1.34744799e+00 3.20457518e-01
-3.85241657e-01 -1.02682009e-01 3.54480520e-02 6.19782627e-01
-8.00202012e-01 4.48209345e-01 -1.20802410e-01 -7.29958564e-02
-1.46193349e+00 3.01237702e-01 -1.59925133e-01 -6.85002089e-01
-7.90224075e-01 4.95828062e-01 -3.84729356e-01 -7.62670934e-01
-1.39677152e-01 -2.66369164e-01 -6.13241374e-01 1.55323461e-01
7.02407002e-01 5.66095293e-01 -2.57937051e-02 -1.00346982e+00
-2.03386858e-01 3.85535806e-02 -4.47816223e-01 -3.16374183e-01
1.69147110e+00 -3.72795314e-01 -5.42476416e-01 5.29975176e-01
1.16299057e+00 4.96329397e-01 -5.58528662e-01 -1.50663450e-01
3.01520735e-01 -4.03791368e-02 2.01121718e-02 -8.30442369e-01
-6.91299558e-01 5.50529122e-01 -3.58472526e-01 3.11795652e-01
1.01666820e+00 4.24109958e-02 1.17140555e+00 8.96649659e-01
-2.55691588e-01 -1.27483344e+00 -1.15105264e-01 1.35746348e+00
1.20807505e+00 -5.48931956e-01 1.12261735e-01 1.31275535e-01
-1.27322185e+00 1.30581772e+00 -1.03527986e-01 -4.10389036e-01
4.31168526e-01 3.54599118e-01 1.19882531e-01 -9.47999209e-02
-1.12295258e+00 2.24162772e-01 2.21986935e-01 3.98094028e-01
5.29197574e-01 1.48863882e-01 -7.73103237e-01 8.34152639e-01
-8.86126101e-01 -2.38613129e-01 1.22176695e+00 7.28531718e-01
-5.00542939e-01 -7.94858873e-01 -1.09524995e-01 1.00595415e+00
-4.93779212e-01 -3.90154451e-01 -1.03496146e+00 4.20585960e-01
-5.88153481e-01 1.02541614e+00 2.43058559e-02 -5.42231798e-01
6.56803325e-02 2.14191496e-01 3.41041327e-01 -8.61606956e-01
-8.21986020e-01 3.90874684e-01 5.92628360e-01 1.90562978e-01
-4.08153832e-01 -7.59545088e-01 -1.31112790e+00 -1.53055251e-01
-3.30230951e-01 7.36475170e-01 5.60409904e-01 9.80201900e-01
2.53372014e-01 3.50116462e-01 7.24138021e-01 -6.43324316e-01
-4.12783891e-01 -1.42593968e+00 -7.16161430e-01 1.53840721e-01
3.66681755e-01 2.95264632e-01 -1.07668219e-02 2.05395833e-01] | [12.481362342834473, 9.522987365722656] |
9902f73f-4723-4c32-8797-c5290ab5694a | semi-supervised-few-shot-atomic-action | 2011.08410 | null | https://arxiv.org/abs/2011.08410v1 | https://arxiv.org/pdf/2011.08410v1.pdf | Semi-Supervised Few-Shot Atomic Action Recognition | Despite excellent progress has been made, the performance on action recognition still heavily relies on specific datasets, which are difficult to extend new action classes due to labor-intensive labeling. Moreover, the high diversity in Spatio-temporal appearance requires robust and representative action feature aggregation and attention. To address the above issues, we focus on atomic actions and propose a novel model for semi-supervised few-shot atomic action recognition. Our model features unsupervised and contrastive video embedding, loose action alignment, multi-head feature comparison, and attention-based aggregation, together of which enables action recognition with only a few training examples through extracting more representative features and allowing flexibility in spatial and temporal alignment and variations in the action. Experiments show that our model can attain high accuracy on representative atomic action datasets outperforming their respective state-of-the-art classification accuracy in full supervision setting. | ['Chi-Keung Tang', 'Yu-Wing Tai', 'Sizhe Song', 'Xiaoyuan Ni'] | 2020-11-17 | null | null | null | null | ['atomic-action-recognition'] | ['computer-vision'] | [ 5.89139998e-01 -1.48166418e-01 -6.86419010e-01 -3.21094960e-01
-9.30771232e-01 -1.21311285e-01 7.13367939e-01 -3.18669766e-01
-3.80751938e-01 6.89446747e-01 5.17305255e-01 4.58529174e-01
-1.68091372e-01 -1.41028762e-01 -4.70161259e-01 -7.88547099e-01
-8.65364373e-02 2.83518314e-01 4.12513375e-01 6.90201968e-02
-2.30375081e-02 4.87686515e-01 -1.82381868e+00 4.20234501e-01
7.11476684e-01 1.06731653e+00 -1.26987901e-02 4.24838275e-01
9.99773517e-02 9.47362661e-01 -2.65266031e-01 -1.34479016e-01
3.50631922e-01 -7.02622354e-01 -7.55881488e-01 8.12705219e-01
6.84043527e-01 -7.31946707e-01 -5.04509211e-01 7.02700317e-01
3.99178445e-01 3.44987869e-01 6.24650955e-01 -1.45025134e+00
-8.27663839e-01 2.41212398e-01 -4.48104590e-01 2.33754218e-01
6.75881088e-01 4.54495370e-01 1.21222973e+00 -8.83505464e-01
7.03956068e-01 1.01101542e+00 6.24405324e-01 7.55157113e-01
-1.19202650e+00 -3.78781974e-01 4.81363773e-01 6.06087446e-01
-1.18301940e+00 -5.75299919e-01 6.94195330e-01 -3.59649301e-01
1.25801921e+00 -8.54374096e-03 8.49100351e-01 1.39309406e+00
-8.26224536e-02 1.34086227e+00 9.46680188e-01 -2.89729923e-01
2.26804867e-01 -3.01320255e-01 -1.79325708e-03 7.62720704e-01
-5.48594110e-02 -1.43739298e-01 -7.30440974e-01 2.20576435e-01
6.95598185e-01 5.46980739e-01 -2.55925387e-01 -7.29152441e-01
-1.30579162e+00 5.40441751e-01 2.44723871e-01 3.54188532e-01
-5.81834853e-01 1.25841469e-01 5.25384545e-01 1.43127620e-01
2.15175986e-01 1.61860019e-01 -4.61268187e-01 -6.43195331e-01
-8.28072488e-01 1.04658236e-03 2.10205793e-01 9.24177289e-01
7.20854759e-01 7.30502829e-02 -3.68893772e-01 6.63537681e-01
-2.03312024e-01 5.11155069e-01 7.28463829e-01 -1.15967751e+00
5.28719604e-01 9.17498648e-01 3.12960707e-03 -7.60515273e-01
-3.62317830e-01 -1.85285751e-02 -7.65300751e-01 1.26452640e-01
3.98574024e-01 3.75585288e-01 -9.61807191e-01 1.56254947e+00
4.00621891e-01 3.06441784e-01 -4.46881540e-03 7.47021317e-01
4.32156861e-01 1.67914480e-01 1.26912743e-01 -2.94018805e-01
1.13157880e+00 -1.47622859e+00 -7.23987818e-01 -1.85957491e-01
1.02021527e+00 -2.19372809e-01 1.28320312e+00 1.85962304e-01
-7.86958218e-01 -6.85805202e-01 -9.05829787e-01 -1.15447380e-01
-2.62647092e-01 4.01620507e-01 8.86419415e-01 2.75687635e-01
-5.94066083e-01 7.03660488e-01 -1.16393137e+00 -5.06253004e-01
8.53690386e-01 3.28064442e-01 -1.01167417e+00 -3.17579985e-01
-7.64281452e-01 6.62064075e-01 2.97441870e-01 -6.84947819e-02
-8.37436259e-01 -4.83208776e-01 -1.10610163e+00 -2.10705906e-01
6.87000990e-01 -4.30098325e-01 1.03020334e+00 -1.08567166e+00
-1.64159763e+00 7.19419062e-01 -5.08365035e-03 -4.73090112e-01
4.76207376e-01 -4.07743722e-01 -4.89040732e-01 4.66981977e-01
2.73055345e-01 6.59039021e-01 9.15288508e-01 -6.25714123e-01
-7.62474954e-01 -5.79781175e-01 1.89027831e-01 2.41108134e-01
-6.22693539e-01 1.15698993e-01 -4.65320885e-01 -5.17583013e-01
3.92664298e-02 -9.28004503e-01 -2.90883839e-01 1.43570915e-01
-9.03895572e-02 -2.84118086e-01 9.91910994e-01 -4.33087915e-01
1.24160612e+00 -2.35654616e+00 4.33207929e-01 -4.59421366e-01
-1.07998274e-01 3.88712764e-01 -3.36134404e-01 3.34522277e-01
-3.44069116e-02 -3.27883869e-01 -2.15730131e-01 -4.79215652e-01
1.05702452e-01 4.62203801e-01 -1.87227607e-01 5.79208910e-01
3.76076192e-01 1.09979653e+00 -1.13825715e+00 -6.48757100e-01
4.98068780e-01 3.85123223e-01 -4.41170275e-01 2.01220319e-01
-1.53653577e-01 5.92653811e-01 -7.49144971e-01 1.06069267e+00
1.01045705e-01 -4.04265732e-01 7.06597641e-02 -2.34877646e-01
2.47468755e-01 9.14029405e-02 -1.15484834e+00 2.01479769e+00
-2.28210628e-01 3.28677863e-01 -4.30636019e-01 -1.16494930e+00
5.26771069e-01 1.91139325e-01 8.12622070e-01 -6.40072644e-01
-6.07923903e-02 3.79427038e-02 -1.50977001e-01 -6.62371278e-01
1.85258627e-01 2.31493577e-01 -9.66770500e-02 5.92148185e-01
2.31362477e-01 2.51710594e-01 4.33129013e-01 1.98139429e-01
1.34307706e+00 6.43908143e-01 5.82594812e-01 2.56972611e-01
5.19759297e-01 -2.23957598e-01 7.49508798e-01 4.09097373e-01
-6.14823103e-01 5.42819202e-01 2.39260212e-01 -6.84003890e-01
-6.76835597e-01 -7.26977587e-01 4.96196263e-02 1.25614119e+00
9.40403491e-02 -5.94909072e-01 -5.37475705e-01 -1.30046129e+00
-2.55237017e-02 3.45807880e-01 -8.85685027e-01 -1.72489181e-01
-6.78121865e-01 -4.00797069e-01 3.21941763e-01 1.12103426e+00
6.41713679e-01 -1.22550428e+00 -8.11046064e-01 1.55682251e-01
-1.07240364e-01 -1.45190966e+00 -4.92627531e-01 1.15753664e-02
-9.88296092e-01 -1.17445219e+00 -6.98160827e-01 -3.47521663e-01
6.93021715e-01 4.74076331e-01 7.54464269e-01 -1.57231912e-01
-5.32863259e-01 8.36184859e-01 -7.30630100e-01 1.17964549e-02
-2.59860456e-02 -1.95846885e-01 1.98551595e-01 3.61261934e-01
5.14975429e-01 -7.43171751e-01 -6.68564439e-01 3.66716266e-01
-8.81935179e-01 -2.25441325e-02 8.00134718e-01 9.56670821e-01
7.35916913e-01 -3.64737570e-01 3.80770922e-01 -5.03432155e-01
-1.10799238e-01 -1.28059641e-01 -2.55240705e-02 4.32957798e-01
-3.75938594e-01 -9.48544443e-02 6.65430188e-01 -4.93733108e-01
-9.11108136e-01 4.71219122e-01 3.17175120e-01 -6.70583665e-01
-4.69545484e-01 1.02895893e-01 -2.20953569e-01 -8.86508897e-02
5.26933670e-01 5.34851611e-01 5.75175844e-02 -4.69812423e-01
5.13803720e-01 5.17754376e-01 4.53152597e-01 -2.67046362e-01
4.50469762e-01 7.43543684e-01 -2.49053594e-02 -7.89975703e-01
-1.09936309e+00 -5.01281977e-01 -1.37661135e+00 -2.30873838e-01
1.01041460e+00 -9.35407579e-01 -1.82805166e-01 5.74185908e-01
-6.35905623e-01 -4.58816618e-01 -6.16032541e-01 6.39393628e-01
-9.17509496e-01 8.50789905e-01 -5.79274356e-01 -7.60474443e-01
-1.63069982e-02 -1.10353243e+00 1.52269137e+00 -4.22851332e-02
-2.72775263e-01 -6.53058767e-01 7.07666352e-02 6.69760823e-01
6.48722202e-02 3.65233272e-01 4.74964768e-01 -6.27371430e-01
-8.27345312e-01 -5.21516383e-01 -7.04667941e-02 5.77796876e-01
6.02804959e-01 -2.65803728e-02 -7.30531335e-01 -2.94142067e-01
-4.39183980e-01 -9.27617550e-01 1.03908503e+00 7.36063272e-02
1.20620263e+00 -2.67491974e-02 -3.49018574e-01 4.80288029e-01
8.91430140e-01 -5.68273328e-02 6.43357575e-01 2.88166642e-01
9.00805831e-01 2.72393614e-01 1.08854783e+00 6.14504099e-01
2.89302677e-01 1.00845599e+00 3.32233697e-01 1.86014995e-01
-2.90763408e-01 -1.22509621e-01 7.63237178e-01 5.43563545e-01
-5.64545572e-01 1.57529250e-01 -4.24036235e-01 5.91858745e-01
-2.18678308e+00 -1.28164375e+00 3.11366469e-01 2.08438540e+00
6.50569677e-01 -4.54120897e-02 5.42519152e-01 2.06118122e-01
3.11163664e-01 5.28781533e-01 -6.96245193e-01 5.09201437e-02
-4.04022187e-02 -1.07260775e-02 1.53530747e-01 -3.15015912e-02
-1.38724434e+00 1.01478076e+00 6.38095427e+00 9.17844355e-01
-8.61853838e-01 9.64394733e-02 2.79945463e-01 -2.51865029e-01
1.92414701e-01 -2.25597799e-01 -8.48515630e-01 4.60389793e-01
5.82670808e-01 2.70256311e-01 1.53406918e-01 9.92798030e-01
-9.13455412e-02 -2.91169360e-02 -1.37401843e+00 1.07537067e+00
4.26052213e-01 -1.19727695e+00 2.54846752e-01 2.30055917e-02
7.95410156e-01 -1.00475401e-01 -2.26389617e-01 4.60591912e-01
1.82320312e-01 -7.43506670e-01 3.74820977e-01 4.81438696e-01
8.66354644e-01 -3.65034133e-01 3.75685930e-01 2.36891106e-01
-1.26137960e+00 -4.84028488e-01 -2.30539232e-01 -2.18596086e-01
2.86043435e-01 4.24808301e-02 -1.01199359e-01 5.74860215e-01
6.40018702e-01 1.36289465e+00 -6.15897179e-01 6.60508692e-01
-2.27082014e-01 3.49756688e-01 -1.88408390e-01 1.62580401e-01
4.76026058e-01 -1.18677713e-01 2.62501955e-01 1.04463768e+00
2.01124460e-01 2.84357905e-01 6.42130494e-01 1.11092143e-01
2.30534241e-01 1.33932948e-01 -6.40085161e-01 -3.48071605e-01
8.32644403e-02 1.13006258e+00 -5.60737848e-01 -4.63620514e-01
-8.07728767e-01 1.34683514e+00 5.35830021e-01 3.00511271e-01
-9.60687339e-01 2.90513206e-02 7.74143457e-01 -1.24837779e-01
6.41632795e-01 -2.73847550e-01 2.44808570e-01 -1.43977809e+00
3.65095824e-01 -1.02573395e+00 6.85219884e-01 -4.63346213e-01
-1.03362131e+00 4.17741686e-01 2.91327760e-02 -1.65170717e+00
-3.72901350e-01 -6.80993378e-01 -3.90680701e-01 7.22474279e-03
-1.24396873e+00 -1.52521229e+00 -3.78336132e-01 8.22136700e-01
8.65811348e-01 -4.25753742e-01 1.04914284e+00 3.19479704e-01
-8.98144543e-01 7.77622640e-01 -2.27514678e-03 1.49134532e-01
6.55054390e-01 -9.71724153e-01 8.96434709e-02 7.79376745e-01
5.65266967e-01 1.85584500e-01 2.50440896e-01 -3.80750567e-01
-1.46404672e+00 -1.01077962e+00 6.62281275e-01 -6.65752172e-01
8.44341874e-01 -1.95351154e-01 -7.08281040e-01 9.38536823e-01
-8.61458853e-02 4.64392960e-01 8.27429652e-01 2.02603675e-02
-5.84635079e-01 -1.81311786e-01 -7.78650582e-01 6.91503823e-01
1.61723399e+00 -6.51696742e-01 -5.97534359e-01 5.51877737e-01
4.16176558e-01 -6.82082549e-02 -7.76882231e-01 6.14417434e-01
6.79129064e-01 -1.18694174e+00 9.54050124e-01 -1.10391736e+00
4.62397814e-01 -3.30557197e-01 -3.93824905e-01 -8.74875367e-01
-5.10837317e-01 -4.72064316e-01 -4.52266604e-01 1.01964486e+00
6.12813383e-02 -3.88423204e-01 7.48126209e-01 7.31428027e-01
-1.84047922e-01 -1.01978922e+00 -1.02087545e+00 -1.13286650e+00
-4.94586021e-01 -4.54480201e-01 3.23039979e-01 9.18334067e-01
1.99927002e-01 1.73858017e-01 -7.34274387e-01 -1.95014179e-01
5.16272366e-01 4.88372982e-01 1.12204659e+00 -9.92236674e-01
-5.23239136e-01 -3.34436983e-01 -1.05619204e+00 -1.19285297e+00
4.28968310e-01 -6.08342707e-01 -1.32461965e-01 -1.37877202e+00
4.81359959e-01 2.68678600e-03 -5.92870951e-01 8.64607513e-01
-1.68759778e-01 5.99369287e-01 1.57305360e-01 2.66857445e-01
-1.38769281e+00 1.01508868e+00 1.21237874e+00 -2.37618595e-01
-5.18279746e-02 -1.81774423e-01 -3.26325268e-01 8.25768232e-01
4.82685745e-01 -1.11635186e-01 -4.58264023e-01 -2.56813347e-01
-4.54959840e-01 -1.82444751e-01 4.34203207e-01 -1.26484883e+00
-4.23415266e-02 -3.96383584e-01 3.45546126e-01 -3.37713242e-01
6.13951564e-01 -8.64729941e-01 -6.13401420e-02 2.59960443e-01
-4.29396152e-01 -2.75218040e-01 -1.50396749e-01 9.13436472e-01
-3.58762503e-01 1.71371117e-01 5.36593497e-01 -9.90217328e-02
-1.25833452e+00 7.02515543e-01 -1.06324427e-01 2.10574288e-02
1.38972390e+00 -6.14313364e-01 -9.94540006e-02 -3.08599442e-01
-9.59294260e-01 1.34087250e-01 5.88912606e-01 4.87128019e-01
4.81704980e-01 -1.70798135e+00 -4.54677016e-01 2.73493081e-01
5.89674354e-01 -3.76388341e-01 6.30372524e-01 1.30858850e+00
2.37804148e-02 3.01145971e-01 -4.63887870e-01 -6.80429757e-01
-1.47685015e+00 6.11566782e-01 3.09983082e-02 -3.17412853e-01
-1.05690825e+00 6.65901482e-01 2.16686159e-01 -4.19828184e-02
3.63714188e-01 -2.88104087e-01 -1.13476619e-01 1.13783695e-01
7.94596195e-01 5.08388162e-01 -3.33011150e-01 -9.90658760e-01
-4.91816550e-01 7.53619611e-01 -1.28761873e-01 3.12979579e-01
1.28408062e+00 -1.39209822e-01 2.97846079e-01 5.66918790e-01
1.14618456e+00 -3.96723628e-01 -1.78516161e+00 -4.67502505e-01
-7.11736977e-02 -8.44214499e-01 -2.59490132e-01 -4.95228827e-01
-9.34157193e-01 9.06996071e-01 4.17604148e-01 -3.46042626e-02
1.15736127e+00 4.54579443e-02 1.03268802e+00 6.69307113e-01
5.96372545e-01 -1.38267148e+00 5.20520449e-01 3.54792714e-01
7.99276471e-01 -1.41371274e+00 2.95945227e-01 -2.97512531e-01
-9.62829649e-01 9.61635411e-01 6.76153362e-01 -1.17776535e-01
4.41653013e-01 -1.18924871e-01 6.13969192e-02 -1.62032664e-01
-6.37687325e-01 -5.77119231e-01 3.83788258e-01 6.41704977e-01
1.75314009e-01 -4.62025553e-02 -6.20239601e-02 4.17437077e-01
4.90408033e-01 1.45506263e-01 1.10587232e-01 1.00217283e+00
-3.43232483e-01 -1.09641635e+00 2.80648917e-01 4.44585681e-01
-3.54354203e-01 2.48064950e-01 -4.05935198e-01 9.68566179e-01
9.00360290e-03 5.91494560e-01 -1.62381709e-01 -4.45907712e-01
3.83485675e-01 3.26654315e-01 7.49767244e-01 -6.18951857e-01
4.15443210e-03 -1.28937215e-01 1.83465585e-01 -1.26957619e+00
-1.00059772e+00 -9.70326602e-01 -1.24713933e+00 2.85495847e-01
-2.44856462e-01 -3.52684319e-01 1.95784923e-02 1.32191861e+00
7.25295663e-01 2.61031717e-01 5.79707742e-01 -1.19114387e+00
-7.52941549e-01 -9.54266369e-01 -8.23263109e-01 9.36775446e-01
2.27212459e-01 -1.10544443e+00 -3.81508857e-01 1.85908139e-01] | [8.315871238708496, 0.6965299844741821] |
877dd124-de10-4113-baef-bad4ca038dc1 | learning-graph-models-for-template-free-1 | null | null | https://arxiv.org/abs/2006.07038 | https://arxiv.org/pdf/2006.07038.pdf | Learning Graph Models for Template-Free Retrosynthesis | Retrosynthesis prediction is a fundamental problem in organic synthesis, where the task is to identify precursor molecules that can be used to synthesize a target molecule. A key consideration in building neural models for this task is aligning model design with strategies adopted by chemists. Building on this viewpoint, this paper introduces a graph-based approach that capitalizes on the idea that the graph topology of precursor molecules is largely unaltered during a chemical reaction. The model first predicts the set of graph edits transforming the target into incomplete molecules called synthons. Next, the model learns to expand synthons into complete molecules by attaching relevant leaving groups. This decomposition simplifies the architecture, making its predictions more interpretable, and also amenable to manual correction. Our model achieves a top-1 accuracy of 53.7%, outperforming previous template-free and semi-template-based methods | ['Regina Barzilay', 'Andreas Krause', 'Connor W. Coley', 'Charlotte Bunne', 'Vignesh Ram Somnath'] | 2021-06-04 | null | null | null | arxiv-2021-6 | ['retrosynthesis'] | ['medical'] | [ 7.86230922e-01 6.41957283e-01 -4.98002976e-01 -2.09251702e-01
-2.88923364e-02 -9.91334558e-01 7.70056844e-01 6.07207000e-01
3.04574780e-02 8.99201989e-01 2.15291008e-01 -6.64770603e-01
2.04672322e-01 -9.11345303e-01 -9.75593805e-01 -6.49506450e-01
2.99484823e-02 4.40226257e-01 5.74637875e-02 -3.61684889e-01
3.51752877e-01 1.01163793e+00 -1.02873731e+00 3.54099393e-01
8.98757696e-01 5.82192659e-01 3.32522452e-01 4.28502202e-01
-1.62130803e-01 8.81404877e-01 -3.90139431e-01 -6.18969023e-01
3.24525446e-01 -5.42376637e-01 -8.87022376e-01 -3.32741499e-01
3.25345010e-01 1.36906505e-01 -4.14372027e-01 8.52295697e-01
2.68962413e-01 2.34735593e-01 1.08697295e+00 -8.49081337e-01
-6.54200852e-01 1.07824433e+00 2.08949909e-01 -2.12821916e-01
3.10631067e-01 -2.61677653e-02 1.32003701e+00 -1.10801578e+00
1.09409916e+00 1.13414514e+00 3.08033526e-01 5.43093681e-01
-1.44317091e+00 -5.47366560e-01 4.31120604e-01 6.96285143e-02
-1.20686841e+00 -6.23667836e-01 6.06586993e-01 -6.34494722e-01
1.47885251e+00 2.72390187e-01 7.81561971e-01 8.97067189e-01
7.32252836e-01 2.53033251e-01 4.33533937e-01 -3.01527888e-01
4.13908035e-01 -1.66577473e-01 -2.79266268e-01 8.06610286e-01
3.39126259e-01 1.77503735e-01 -5.42783976e-01 1.25183195e-01
4.49347049e-01 -2.36494876e-02 -1.01026952e-01 -5.30059576e-01
-1.23332131e+00 8.26211572e-01 7.44253099e-01 2.59919703e-01
-4.09180552e-01 1.50515154e-01 2.03566849e-01 1.41831756e-01
1.93320990e-01 1.34092629e+00 -5.14624655e-01 4.41069812e-01
-8.60986054e-01 1.45715222e-01 9.71056283e-01 1.26074624e+00
8.18164408e-01 2.25028425e-01 1.06311515e-01 2.57365316e-01
3.14403653e-01 -2.29024544e-01 1.03352718e-01 -5.66609263e-01
3.48505020e-01 7.64505625e-01 -1.14680372e-01 -8.12269568e-01
-5.07343113e-01 -3.99086118e-01 -6.69324100e-01 -3.49069797e-02
2.12410033e-01 -4.92864884e-02 -1.14600706e+00 1.68012178e+00
3.36789906e-01 -3.81289065e-01 1.05227135e-01 3.15843225e-01
9.13498163e-01 1.03694689e+00 2.65700758e-01 -5.61503768e-01
7.59221375e-01 -1.21030819e+00 -5.24775982e-01 4.49982882e-02
8.18912685e-01 -6.85460806e-01 3.63179266e-01 5.47966957e-01
-1.00584292e+00 -5.17849267e-01 -1.37547505e+00 -3.12951714e-01
-8.49430203e-01 -1.52813330e-01 8.24637771e-01 6.16879523e-01
-8.59411299e-01 1.29124987e+00 -5.87692499e-01 -2.07650110e-01
2.47814283e-01 8.26139450e-01 -4.30422753e-01 1.48900807e-01
-9.72555101e-01 1.18541670e+00 1.11813164e+00 1.08760983e-01
-1.29350054e+00 -9.54280138e-01 -7.54032314e-01 8.04559141e-02
5.44753730e-01 -8.72538984e-01 1.15395927e+00 -9.19306576e-01
-1.74256086e+00 4.22601938e-01 -3.01632226e-01 -4.20064777e-01
2.04682142e-01 4.16023999e-01 -3.61554235e-01 -2.19146356e-01
-2.05227926e-01 8.85810733e-01 7.80669987e-01 -9.85151887e-01
-4.03207302e-01 -5.92511520e-02 1.09300196e-01 1.59845829e-01
1.41550541e-01 -1.50832430e-01 -1.65996850e-01 -6.76325500e-01
1.05449885e-01 -9.89163756e-01 -6.28413081e-01 -2.85424381e-01
-8.42826962e-01 -2.74731219e-01 1.10717490e-01 -3.47121805e-01
1.17975795e+00 -1.42252135e+00 8.04949105e-01 4.43994671e-01
5.24358630e-01 1.55603558e-01 -2.45872796e-01 1.15166092e+00
-7.07167685e-01 3.71692449e-01 -7.01915324e-02 1.51293382e-01
-1.36908025e-01 -2.41455689e-01 -4.49852556e-01 1.38198718e-01
2.81417698e-01 9.62077737e-01 -1.01506388e+00 4.37179692e-02
1.30456388e-01 2.13271052e-01 -6.96844161e-01 8.26967061e-02
-8.36165786e-01 5.43085217e-01 -4.26164210e-01 6.81574941e-01
2.76806533e-01 -1.25684589e-01 8.31582010e-01 -4.16309714e-01
-3.71624500e-01 7.48399496e-01 -8.00828874e-01 1.77396119e+00
-1.44993812e-01 2.86622882e-01 -4.86086071e-01 -8.33981574e-01
1.14180315e+00 2.60766149e-01 3.97418261e-01 -3.03917974e-01
1.49359286e-01 2.94218361e-01 3.50015670e-01 -2.54607890e-02
5.64120829e-01 -2.14006320e-01 2.90020913e-01 2.49202535e-01
2.48204187e-01 -3.85661542e-01 7.12890208e-01 3.50051761e-01
8.42306614e-01 3.81978750e-01 9.12761927e-01 -3.32265735e-01
6.24938905e-01 -9.60922262e-05 6.04366422e-01 4.51487601e-01
3.91960114e-01 2.38408178e-01 5.15558541e-01 -8.28393579e-01
-1.34347415e+00 -8.51589680e-01 3.32401216e-01 1.08395159e+00
-1.56805664e-01 -9.63806510e-01 -7.92215824e-01 -6.36706471e-01
-1.48564801e-01 8.53969038e-01 -6.58163667e-01 -5.32157660e-01
-4.36646760e-01 -3.90133828e-01 2.70522326e-01 2.64500111e-01
-2.95234919e-01 -9.75762784e-01 1.67164028e-01 7.31175601e-01
1.91941544e-01 -5.77663422e-01 -5.68152010e-01 6.64768338e-01
-9.55952406e-01 -1.18341076e+00 -3.00930709e-01 -7.78760493e-01
9.09928739e-01 1.62462294e-01 9.34312761e-01 4.74954024e-02
-2.34168440e-01 -3.96741629e-01 -8.51451010e-02 -6.01185620e-01
-1.05190647e+00 4.76401448e-01 1.09347217e-01 -1.97556064e-01
-5.82930632e-02 -7.01641738e-01 -3.97556514e-01 -2.34956872e-02
-8.09580266e-01 4.45011497e-01 6.04815543e-01 4.45303082e-01
7.37107992e-01 -1.47758424e-01 5.54847002e-01 -9.67546582e-01
4.12107468e-01 -3.50970298e-01 -6.43835247e-01 6.06869042e-01
-9.41327274e-01 4.33073461e-01 1.12646592e+00 -2.41451502e-01
-8.71001124e-01 6.84735358e-01 -2.49832664e-02 6.59753308e-02
2.31390353e-02 6.72118425e-01 -6.06789827e-01 -2.66370386e-01
7.31444538e-01 1.74714133e-01 -2.52576441e-01 -3.00479829e-01
7.19605684e-01 2.05766514e-01 1.40799835e-01 -6.16768479e-01
6.39930248e-01 -6.96543902e-02 5.64008594e-01 -9.14265752e-01
-5.79567015e-01 -2.43929643e-02 -1.08218491e+00 7.49385729e-02
7.01706707e-01 -7.88498700e-01 -8.12308848e-01 -5.83729669e-02
-1.29150462e+00 -2.77711600e-01 -1.36482358e-01 1.64351583e-01
-6.21620774e-01 5.47822535e-01 -3.07979643e-01 -2.26755828e-01
-3.47734690e-01 -1.35372102e+00 5.09238243e-01 -1.87244900e-02
-4.82596338e-01 -9.92992818e-01 1.08108543e-01 1.55888513e-01
-1.33411959e-01 3.08489084e-01 1.45947099e+00 -1.12334979e+00
-8.87179673e-01 -4.52810572e-03 1.92063078e-01 -2.92194206e-02
4.10358876e-01 3.29979718e-01 -7.43828416e-01 -2.47490585e-01
-5.64992309e-01 -6.05204776e-02 8.42327714e-01 2.04712152e-01
1.03034139e+00 -4.27278370e-01 -6.25644743e-01 6.01034820e-01
1.04974091e+00 7.26896644e-01 4.93302524e-01 1.67320698e-01
9.13595855e-01 6.93703294e-01 2.31718600e-01 6.09879084e-02
7.55375028e-02 5.22390127e-01 6.24670208e-01 1.08539529e-01
-3.36258207e-03 -7.97526479e-01 3.15717340e-01 5.69773972e-01
-3.11170548e-01 -7.59226680e-01 -7.60812640e-01 3.85095142e-02
-1.67356074e+00 -7.94613719e-01 -1.67145401e-01 2.12220716e+00
9.56648886e-01 8.65925029e-02 1.10329501e-01 4.04676907e-02
5.71387947e-01 1.45992190e-01 -7.11705446e-01 -6.50352895e-01
1.31787911e-01 3.78401309e-01 6.55963838e-01 7.17452943e-01
-9.30243790e-01 1.48869860e+00 7.43719864e+00 6.98099434e-01
-1.18959951e+00 -5.82047582e-01 4.58247930e-01 1.29079714e-01
-4.68314826e-01 3.67560238e-01 -8.79470408e-01 2.14502886e-01
1.00293720e+00 -4.62067872e-01 7.27026165e-01 7.19742358e-01
3.58082980e-01 4.41124976e-01 -1.89634621e+00 4.15797949e-01
-1.50932744e-01 -2.11266112e+00 7.07178116e-01 9.85241905e-02
8.28452826e-01 -2.54935533e-01 -2.37619191e-01 9.23068542e-03
4.35989022e-01 -1.33913624e+00 9.66150522e-01 3.75556767e-01
7.21315384e-01 -7.03825831e-01 -5.98766888e-03 1.30947441e-01
-1.21374428e+00 7.58449435e-02 -4.41123098e-01 -3.76435854e-02
3.24191973e-02 2.34217227e-01 -1.51795101e+00 7.59302258e-01
-2.29781970e-01 1.03043091e+00 -4.05396432e-01 7.73647726e-01
-6.74370468e-01 1.98626906e-01 2.94371881e-03 -2.01543495e-01
2.10356623e-01 -5.02765238e-01 3.35171342e-01 1.17333865e+00
2.63377815e-01 3.92053416e-03 2.84562737e-01 1.07012689e+00
-5.02372324e-01 2.84325331e-01 -9.60552812e-01 -5.76414883e-01
5.73159456e-01 1.13955224e+00 -9.34256256e-01 -3.01670194e-01
-1.33703843e-01 7.75539875e-01 3.08541507e-01 4.41749603e-01
-6.44705892e-01 -5.02843857e-01 4.26505923e-01 -4.13035415e-03
4.03179765e-01 -3.03163111e-01 -4.83175293e-02 -8.39130938e-01
-4.38412994e-01 -9.92348433e-01 1.16902582e-01 -6.11702204e-01
-5.95011890e-01 4.45495218e-01 -3.23119253e-01 -8.90258789e-01
-1.32850930e-01 -8.77062619e-01 -7.15838730e-01 6.76854014e-01
-1.29927194e+00 -1.17212391e+00 1.97567791e-01 4.12111059e-02
6.38535321e-01 -2.17538476e-01 8.91935170e-01 -6.81343600e-02
-7.74256945e-01 3.77482206e-01 1.32306337e-01 -4.44135606e-01
5.84843993e-01 -1.22842717e+00 8.57330859e-01 8.34850371e-01
2.76101470e-01 1.17639792e+00 8.48925054e-01 -9.03500795e-01
-1.54484689e+00 -1.38306999e+00 1.15759885e+00 -4.00354087e-01
6.34921432e-01 -5.60861766e-01 -5.37724018e-01 6.99050128e-01
1.00273587e-01 -5.13589680e-01 1.04574275e+00 -9.25619304e-02
-6.05805337e-01 3.63920592e-02 -7.06191242e-01 9.01107967e-01
1.28820205e+00 -4.97785807e-01 -3.58828813e-01 7.27342486e-01
9.72991049e-01 -3.30811858e-01 -1.10359585e+00 5.04957810e-02
5.07737577e-01 -6.30170763e-01 1.09777713e+00 -1.21702003e+00
5.13630033e-01 -4.02649492e-01 3.32166441e-02 -1.43333840e+00
-5.71770608e-01 -1.16442049e+00 -1.42565876e-01 5.92588246e-01
1.18886042e+00 -3.33931655e-01 8.30425501e-01 6.60319269e-01
-5.98882377e-01 -5.72798550e-01 -4.95708346e-01 -8.06910038e-01
1.58066899e-01 1.41596466e-01 7.06108749e-01 9.88367915e-01
2.54864603e-01 8.13613474e-01 -3.48718435e-01 -1.27963468e-01
9.58758742e-02 2.25322656e-02 7.30634093e-01 -1.42205095e+00
-3.20373960e-02 -5.52180946e-01 -6.99434280e-02 -1.05997527e+00
2.96569377e-01 -1.62400043e+00 -1.21933073e-01 -1.49574637e+00
-2.27096919e-02 -1.44592762e-01 -2.13725433e-01 6.04485929e-01
2.47085527e-01 -1.66796133e-01 2.52887905e-01 1.22748435e-01
-3.66002381e-01 6.79188609e-01 1.25247347e+00 -2.83916414e-01
-4.41878140e-01 -2.02155024e-01 -9.07363355e-01 5.66041708e-01
7.47482121e-01 -6.90684080e-01 -4.75224525e-01 -9.50385928e-02
7.40882277e-01 -1.20612562e-01 -3.45420361e-01 -5.09907007e-01
2.89014548e-01 -5.80153584e-01 2.82810897e-01 -5.29924572e-01
2.48479962e-01 -7.41023481e-01 6.68156147e-01 8.69492352e-01
-5.72738886e-01 -7.84196705e-02 -2.98578963e-02 6.29407167e-01
1.96155608e-01 -2.93598771e-01 5.79386711e-01 -3.94074023e-01
-4.52011675e-01 6.37639523e-01 -6.92382216e-01 -5.28406739e-01
1.06446242e+00 -3.80993664e-01 -2.31618345e-01 -9.54533368e-02
-9.20241714e-01 -1.24985069e-01 5.66606224e-01 4.53094691e-01
4.01310176e-01 -1.07016182e+00 -1.87130123e-01 1.85443033e-02
1.51548639e-01 2.08189827e-03 -3.06854457e-01 4.62850094e-01
-9.02188182e-01 8.96140039e-01 -1.76979244e-01 -4.17014882e-02
-1.17658329e+00 9.22741592e-01 4.44435507e-01 -4.30140235e-02
-1.54387295e-01 8.84802520e-01 3.93549889e-01 -5.43886542e-01
2.46260405e-01 -6.35920227e-01 -1.39336616e-01 1.26818612e-01
2.98183560e-01 2.93287218e-01 3.48244876e-01 -4.30996299e-01
-2.25741968e-01 3.36448312e-01 -5.61463237e-01 4.51021254e-01
1.46330810e+00 3.96127760e-01 -2.43964761e-01 3.01058628e-02
9.56869245e-01 -1.13110393e-01 -1.02401698e+00 -1.85289402e-02
2.00986192e-01 1.70489192e-01 -1.38914749e-01 -8.03260863e-01
-5.88902175e-01 8.01029384e-01 -1.49833232e-01 -6.91501424e-02
5.20970702e-01 -2.97766298e-01 3.82358670e-01 1.23813772e+00
3.58756371e-02 -1.04883218e+00 3.45863774e-02 5.35589159e-01
1.05664861e+00 -8.61513734e-01 2.96302319e-01 -7.50362694e-01
-2.42971629e-01 1.53700507e+00 3.06416333e-01 2.08338380e-01
2.51236707e-01 -3.23008150e-01 -5.49236953e-01 -2.15060845e-01
-8.54806781e-01 2.06726253e-01 4.10709709e-01 5.85598469e-01
6.81057990e-01 -5.27439453e-02 -2.73978889e-01 3.42934221e-01
-3.30925852e-01 -2.77495980e-01 5.98603010e-01 8.74583900e-01
-6.05770469e-01 -1.60688198e+00 2.09078893e-01 8.63081664e-02
-2.47944683e-01 -4.27861363e-01 -1.23440731e+00 6.35407865e-01
1.74172893e-01 8.04803967e-01 -4.40464318e-01 -3.90575171e-01
3.33950490e-01 2.53479391e-01 7.18240261e-01 -1.05163872e+00
-6.64662063e-01 1.60506107e-02 3.88463736e-01 -4.91631806e-01
-1.90884620e-01 -3.57849568e-01 -1.24519467e+00 -2.79195458e-01
-3.85496438e-01 3.82820398e-01 6.57491446e-01 8.21517706e-01
5.28666258e-01 6.10821486e-01 7.58300424e-01 -8.19090426e-01
-2.70441055e-01 -6.61200464e-01 -2.38764718e-01 -1.04157172e-01
1.27470523e-01 -4.03938264e-01 5.15191350e-03 3.34801763e-01] | [4.523787498474121, 6.084957122802734] |
14b7ac01-9bbe-477b-8979-61fdc319f9d3 | through-the-fairness-lens-experimental | 2307.02726 | null | https://arxiv.org/abs/2307.02726v1 | https://arxiv.org/pdf/2307.02726v1.pdf | Through the Fairness Lens: Experimental Analysis and Evaluation of Entity Matching | Entity matching (EM) is a challenging problem studied by different communities for over half a century. Algorithmic fairness has also become a timely topic to address machine bias and its societal impacts. Despite extensive research on these two topics, little attention has been paid to the fairness of entity matching. Towards addressing this gap, we perform an extensive experimental evaluation of a variety of EM techniques in this paper. We generated two social datasets from publicly available datasets for the purpose of auditing EM through the lens of fairness. Our findings underscore potential unfairness under two common conditions in real-world societies: (i) when some demographic groups are overrepresented, and (ii) when names are more similar in some groups compared to others. Among our many findings, it is noteworthy to mention that while various fairness definitions are valuable for different settings, due to EM's class imbalance nature, measures such as positive predictive value parity and true positive rate parity are, in general, more capable of revealing EM unfairness. | ['Divesh Srivastava', 'Abolfazl Asudeh', 'Fatemeh Nargesian', 'Nikola Danevski', 'Nima Shahbazi'] | 2023-07-06 | null | null | null | null | ['fairness', 'fairness'] | ['computer-vision', 'miscellaneous'] | [-7.89577365e-02 2.90733904e-01 -4.26933467e-01 -4.82583404e-01
-1.26805127e-01 -4.04647976e-01 5.79850972e-01 7.13955641e-01
-7.74144769e-01 9.63676512e-01 3.78448278e-01 -2.62071550e-01
-8.76763538e-02 -1.07232010e+00 -3.06630284e-01 -2.60374069e-01
-9.50419754e-02 4.21073496e-01 -3.21597070e-01 -1.40116870e-01
6.40226483e-01 2.59656519e-01 -1.42935133e+00 -3.40052471e-02
1.37829220e+00 4.75369126e-01 -8.25994849e-01 1.11895517e-01
-9.04681906e-02 8.27261269e-01 -6.02364719e-01 -1.61602187e+00
4.94186848e-01 -2.72927135e-01 -8.72883856e-01 -5.50062835e-01
7.16032445e-01 -2.20257312e-01 -5.12799799e-01 1.62266576e+00
7.16413975e-01 -2.78556496e-02 6.85973763e-01 -1.68973839e+00
-7.05850005e-01 1.01236022e+00 -8.40979993e-01 2.97599465e-01
3.56975496e-01 1.89023316e-01 1.49673867e+00 -3.73803347e-01
6.27929628e-01 1.31958985e+00 9.00747597e-01 3.15777391e-01
-1.31409192e+00 -1.25703657e+00 -4.59650084e-02 2.64691085e-01
-1.50816321e+00 -5.33633113e-01 4.34688359e-01 -4.32341486e-01
3.81498247e-01 6.11651242e-01 3.87529999e-01 8.11780691e-01
1.62370265e-01 5.13439596e-01 1.40541863e+00 -2.05738381e-01
2.04110384e-01 3.90986681e-01 1.35848641e-01 1.58515915e-01
1.09405279e+00 3.03987384e-01 -5.29791057e-01 -9.62036431e-01
1.22817144e-01 2.59301867e-02 -1.34004056e-01 -2.93674111e-01
-9.42263544e-01 1.09717679e+00 3.21807712e-01 1.86624140e-01
-2.78328866e-01 -6.14153892e-02 4.27366167e-01 5.36028206e-01
6.58041596e-01 7.76599407e-01 -2.05781177e-01 -1.76795885e-01
-1.05820847e+00 6.22510314e-01 1.06867504e+00 6.10550821e-01
7.00036764e-01 -4.32942212e-01 -1.74161732e-01 6.31061971e-01
1.38822913e-01 5.54952562e-01 2.34236971e-01 -1.04323018e+00
6.44124448e-01 7.37667859e-01 2.79557854e-01 -1.62632442e+00
-2.02348158e-01 -3.62482131e-01 -1.13536513e+00 8.00898969e-02
6.90209806e-01 -8.68470818e-02 -4.66280803e-03 2.04710078e+00
3.06848377e-01 9.28388909e-03 -3.19256008e-01 9.76099133e-01
6.04601562e-01 -1.40419498e-01 6.51009917e-01 -1.11611426e-01
1.54591393e+00 -3.34647685e-01 -6.35884523e-01 -1.65435195e-01
6.45517409e-01 -7.80720949e-01 5.47210813e-01 -1.82382122e-01
-1.01598859e+00 8.90805051e-02 -6.31551445e-01 1.95236370e-01
-5.41174889e-01 -5.74991763e-01 8.55035245e-01 1.09171200e+00
-7.37330198e-01 6.95500374e-01 -1.28450036e-01 -5.46539187e-01
8.14627767e-01 1.76993504e-01 -4.58788574e-01 6.67270645e-02
-1.62091720e+00 1.15170658e+00 -1.30137533e-01 -4.41109478e-01
-1.11119694e-03 -8.75917256e-01 -6.22895122e-01 2.18086094e-01
2.37140059e-01 -8.52368176e-01 8.53093565e-01 -1.30779850e+00
-6.30306721e-01 1.56333041e+00 9.42872018e-02 -5.35876811e-01
1.16881382e+00 1.47534302e-02 -6.94733679e-01 -2.76888490e-01
3.05275559e-01 2.65211850e-01 3.84906888e-01 -9.92489874e-01
-5.59458613e-01 -6.91242635e-01 8.00733417e-02 2.40891576e-01
-4.40083861e-01 4.89921391e-01 4.85805571e-01 -9.28692341e-01
-2.61369377e-01 -8.46976399e-01 -3.47233295e-01 1.84955463e-01
-4.46242183e-01 -1.24028705e-01 1.74576432e-01 -4.68168914e-01
1.66776514e+00 -1.75847316e+00 -3.87050271e-01 4.06063616e-01
6.21442795e-01 3.20886552e-01 1.42458364e-01 3.87819827e-01
-1.64012536e-01 6.59755051e-01 -3.93681109e-01 -1.42257929e-01
3.36740643e-01 -3.72536778e-01 -1.48485795e-01 9.62830722e-01
-5.69881201e-02 8.01168382e-01 -9.87670839e-01 -6.62151039e-01
-9.47360173e-02 -8.37824196e-02 -5.91092885e-01 6.15820810e-02
4.19956475e-01 2.29714271e-02 -2.41476774e-01 7.19218493e-01
1.04821098e+00 -1.79453000e-01 4.76848394e-01 5.78478388e-02
-1.18826404e-01 3.25758338e-01 -1.27439058e+00 8.06399465e-01
2.14229941e-01 7.09887266e-01 1.29475489e-01 -9.73639727e-01
6.75967336e-01 1.98703900e-01 3.95919442e-01 -8.88000488e-01
3.50703657e-01 2.43371099e-01 1.32040843e-01 -2.75481939e-01
1.07696712e+00 -4.32230353e-01 -2.43960306e-01 7.96053112e-01
-5.12047172e-01 2.96079725e-01 5.11354133e-02 3.62155348e-01
9.60856199e-01 -5.73396921e-01 7.85317421e-01 -5.54071426e-01
3.37316602e-01 -5.49260378e-02 9.83680844e-01 9.08093512e-01
-9.98727560e-01 4.30176079e-01 4.54720289e-01 -4.60592210e-01
-1.15092885e+00 -9.36877131e-01 -3.71709526e-01 9.05492842e-01
4.47823942e-01 -1.65539682e-01 -5.11581302e-01 -6.32957339e-01
8.24157476e-01 6.42336786e-01 -7.84259200e-01 -2.51330018e-01
-2.28044987e-01 -1.18286431e+00 9.49009717e-01 6.47121221e-02
4.88191634e-01 -8.06561589e-01 -5.61106443e-01 -9.40830335e-02
-4.40091997e-01 -8.43624651e-01 -5.04162133e-01 -5.22931457e-01
-6.92653000e-01 -1.30072045e+00 -7.30913460e-01 -1.95739999e-01
7.36348808e-01 3.41613710e-01 1.79783475e+00 5.07704556e-01
-2.32230112e-01 2.63161182e-01 -1.99614361e-01 -6.21933937e-01
-3.88065159e-01 6.83441237e-02 2.68113971e-01 8.15875977e-02
9.93534267e-01 -5.23726642e-01 -7.61136293e-01 4.55699116e-01
-8.03349137e-01 -2.87501037e-01 3.70881230e-01 3.17126840e-01
-2.21090108e-01 -3.12394112e-01 7.94651926e-01 -1.44327569e+00
7.89946377e-01 -1.00286365e+00 -3.26679647e-01 2.87857920e-01
-1.08165348e+00 -3.20822567e-01 2.95519799e-01 -2.61221766e-01
-9.41316009e-01 -9.05799091e-01 4.23372298e-01 2.51077443e-01
3.87993939e-02 1.63674846e-01 1.50075899e-02 -8.45664814e-02
7.92821825e-01 -4.65032727e-01 -1.24546131e-02 -2.15481654e-01
1.51517898e-01 9.57142293e-01 2.19522968e-01 -6.75041974e-01
9.67694104e-01 5.28992057e-01 -2.52763540e-01 -3.40566576e-01
-4.91022170e-01 -2.65037328e-01 -2.21112579e-01 -1.81588694e-01
5.01047552e-01 -9.23971474e-01 -1.10100591e+00 4.21533942e-01
-7.61839867e-01 1.15074195e-01 -7.16960728e-02 3.15390676e-01
1.06886495e-02 6.79327071e-01 -5.98267555e-01 -1.02513444e+00
-5.19386351e-01 -7.46804118e-01 4.08464193e-01 4.49228793e-01
-7.56797373e-01 -1.01986492e+00 2.24351346e-01 5.07430494e-01
6.00957036e-01 2.99414665e-01 7.70733476e-01 -8.18111241e-01
-2.16269344e-01 -2.66761810e-01 -5.81642210e-01 -2.55007714e-01
4.90653934e-03 9.67735276e-02 -9.91763949e-01 -3.95486981e-01
-5.60495079e-01 -1.02540776e-01 6.14331722e-01 1.00398555e-01
9.46902394e-01 -4.74311203e-01 -3.98944288e-01 2.96104789e-01
1.36035526e+00 -2.83678055e-01 6.18262231e-01 6.56692088e-01
2.07276359e-01 9.44502771e-01 6.98561668e-01 7.55805910e-01
8.98323774e-01 5.96191466e-01 2.74995536e-01 -1.55832961e-01
2.83002615e-01 -2.87676901e-01 1.00138858e-01 3.62209529e-01
-1.08229496e-01 -9.56282094e-02 -1.02211857e+00 5.89407325e-01
-1.72024596e+00 -1.31158566e+00 -1.47775456e-01 2.49729061e+00
8.68689716e-01 -1.74494281e-01 3.94814283e-01 -3.24763544e-02
1.27528846e+00 1.83538169e-01 -3.63808155e-01 -5.12323022e-01
-6.06375813e-01 -2.45379344e-01 7.82408059e-01 1.23488672e-01
-8.84719729e-01 5.01466274e-01 6.54735851e+00 5.22027671e-01
-7.09809661e-01 -2.09732503e-01 1.05099499e+00 5.07876985e-02
-7.07384527e-01 2.17610434e-01 -2.29530111e-01 9.05636847e-01
6.63242102e-01 -9.38811719e-01 1.49276733e-01 4.90144759e-01
7.08631240e-03 -1.81503907e-01 -9.22334909e-01 9.38600779e-01
-1.06006674e-01 -1.00413787e+00 -7.06237629e-02 2.49110878e-01
8.92986536e-01 -1.99321136e-01 1.89701080e-01 2.87145585e-01
8.14097703e-01 -1.21036422e+00 6.77751243e-01 3.58196020e-01
7.36877680e-01 -8.03666890e-01 9.30619717e-01 1.36709034e-01
-6.67353392e-01 -8.74258205e-02 -4.88798440e-01 -3.82487386e-01
-3.21849138e-02 1.11376929e+00 -1.48226142e-01 5.83388567e-01
8.25637162e-01 2.88149446e-01 -5.52614748e-01 1.26894951e+00
2.22815767e-01 4.36318159e-01 -1.25353914e-02 1.19778953e-01
-1.66282266e-01 -3.03246200e-01 6.07487321e-01 1.28945363e+00
1.01859115e-01 1.30266875e-01 -3.20010126e-01 9.65497434e-01
-6.70144081e-01 3.14792007e-01 -7.52804697e-01 -1.40822187e-01
1.03635228e+00 1.33191597e+00 -5.05216658e-01 -2.87304729e-01
-5.40818691e-01 6.42488778e-01 4.24169332e-01 9.95655209e-02
-7.42493570e-01 -3.30186635e-01 1.16752136e+00 8.26331526e-02
-5.54473877e-01 2.84970701e-01 -6.33884013e-01 -1.47343004e+00
-2.59920150e-01 -1.08364189e+00 8.46862912e-01 -2.52696514e-01
-1.84335077e+00 -7.41443411e-03 -3.69711459e-01 -1.11485422e+00
2.55486697e-01 -8.93942788e-02 -6.52861536e-01 9.22770858e-01
-1.61878359e+00 -5.91255724e-01 -3.62916380e-01 2.06261680e-01
-3.54646921e-01 -1.68585598e-01 7.40999699e-01 8.11224222e-01
-5.65243006e-01 1.18763959e+00 -5.84323145e-02 2.13472456e-01
1.32755423e+00 -1.08435261e+00 4.78701532e-01 7.10288763e-01
-2.43418872e-01 7.60398984e-01 9.82667923e-01 -7.56689012e-01
-7.30731070e-01 -8.59736502e-01 1.39321423e+00 -7.24545538e-01
4.34442610e-01 5.73657220e-03 -8.12860847e-01 6.33780181e-01
-5.01009682e-03 -1.60358146e-01 1.06984770e+00 3.82484913e-01
-6.93034530e-01 3.15752178e-02 -1.78079033e+00 8.21895719e-01
1.32139266e+00 -6.29085660e-01 -6.19277894e-01 -1.55495912e-01
1.98382601e-01 -1.11227632e-01 -9.82613385e-01 4.10776287e-01
7.56559491e-01 -1.32352924e+00 7.83466518e-01 -9.14844990e-01
5.56593597e-01 -4.71676812e-02 -7.22400174e-02 -1.24518740e+00
-5.30220151e-01 -6.11045241e-01 2.40403563e-01 1.57243621e+00
3.10743332e-01 -1.07601833e+00 8.62124562e-01 1.31915879e+00
7.25608110e-01 -3.03689986e-01 -7.91267633e-01 -5.19405901e-01
4.87494260e-01 -2.22055186e-02 1.03376448e+00 1.87791514e+00
2.51879960e-01 -1.86823651e-01 -4.57254559e-01 1.15047591e-02
1.02299416e+00 3.63957822e-01 8.96162808e-01 -1.60807586e+00
2.14474455e-01 -8.16105962e-01 -5.92290521e-01 -1.99554548e-01
3.66246819e-01 -9.89457369e-01 -5.96883774e-01 -9.24299955e-01
9.82951105e-01 -8.05210173e-01 -2.49548018e-01 8.73501673e-02
-6.47712350e-01 2.31490567e-01 3.31411362e-01 2.66284466e-01
-5.08189976e-01 2.82147795e-01 8.54194582e-01 -1.43527180e-01
3.31292808e-01 7.44330958e-02 -1.38176775e+00 5.79879820e-01
7.62976229e-01 -7.67865300e-01 2.17951775e-01 -2.03444391e-01
6.81581497e-01 -1.98507354e-01 3.82655293e-01 -5.07827997e-01
2.98786044e-01 -5.67611933e-01 1.57200605e-01 9.69792455e-02
-3.37693334e-01 -8.37277651e-01 4.26353037e-01 5.98663926e-01
-4.85387355e-01 3.37400705e-01 -9.76835638e-02 4.33637440e-01
-7.36081004e-02 6.80259336e-03 8.77692103e-01 -1.93072319e-01
-3.55605036e-01 3.94245028e-01 -1.53292149e-01 6.40842319e-01
9.43306684e-01 -2.76419431e-01 -7.19896853e-01 -4.71225411e-01
3.02955043e-02 4.24932569e-01 1.04706466e+00 4.23744172e-01
-5.45149371e-02 -1.42944503e+00 -1.09612870e+00 -1.91833019e-01
4.27258015e-01 -7.72958696e-01 1.70707464e-01 9.70494330e-01
-2.59870827e-01 6.20045029e-02 -4.09235895e-01 2.70029921e-02
-1.24757659e+00 3.35090160e-01 4.58394915e-01 -1.14111319e-01
-2.21512392e-01 5.27605474e-01 1.93316236e-01 -6.42498016e-01
1.08236879e-01 4.96858239e-01 -7.51570240e-03 3.57660323e-01
5.03767431e-01 1.06298578e+00 -1.91795245e-01 -9.47997928e-01
-5.18971384e-01 1.38878420e-01 1.13351375e-01 2.05206335e-01
1.08594716e+00 -3.39600414e-01 -4.16801870e-01 7.10864589e-02
7.88888931e-01 4.52423692e-01 -4.29261148e-01 -3.15790504e-01
3.55424315e-01 -1.04693937e+00 -2.66180784e-01 -7.01832175e-01
-1.03642583e+00 4.67167705e-01 2.97922820e-01 4.36272085e-01
7.35241115e-01 -5.60344577e-01 5.74045479e-01 -1.58166170e-01
5.46813428e-01 -1.37179196e+00 -5.07825255e-01 6.24318533e-02
2.84322828e-01 -1.45104241e+00 4.18497384e-01 -2.99422294e-01
-6.23091578e-01 5.15697598e-01 6.81067169e-01 1.16455875e-01
4.66724813e-01 1.13030290e-02 4.14592959e-02 -3.92296314e-02
-4.89900827e-01 -7.74172917e-02 6.88497648e-02 3.36655885e-01
7.81608462e-01 5.68472207e-01 -1.01073337e+00 6.29038513e-01
-3.66483837e-01 -1.41091764e-01 7.24871397e-01 5.65931678e-01
-1.27028033e-01 -9.78238642e-01 -3.92585605e-01 8.96211624e-01
-1.09779036e+00 -8.68922248e-02 -7.31052995e-01 6.08265579e-01
-6.21845238e-02 1.02330625e+00 1.89098045e-01 -1.97370425e-01
2.60949194e-01 -1.66649312e-01 8.54472294e-02 -2.32679635e-01
-9.94246006e-01 -1.13038111e+00 3.06814551e-01 -6.23995960e-01
-4.95787174e-01 -8.01800370e-01 -8.34686697e-01 -1.43979728e+00
-4.32956904e-01 2.72886932e-01 2.11767018e-01 6.22523367e-01
4.63127106e-01 -2.65844285e-01 6.45041645e-01 -2.45661855e-01
-7.60162771e-01 -5.97463071e-01 -8.85115147e-01 1.13018453e+00
2.00594272e-02 -7.48087943e-01 -7.09594369e-01 -5.96865296e-01] | [8.877676010131836, 5.348504543304443] |
70e0de97-8daf-491a-a726-13ccf90ef453 | towards-end-to-end-car-license-plate-location | 2008.10916 | null | https://arxiv.org/abs/2008.10916v2 | https://arxiv.org/pdf/2008.10916v2.pdf | Towards End-to-end Car License Plate Location and Recognition in Unconstrained Scenarios | Benefiting from the rapid development of convolutional neural networks, the performance of car license plate detection and recognition has been largely improved. Nonetheless, most existing methods solve detection and recognition problems separately, and focus on specific scenarios, which hinders the deployment for real-world applications. To overcome these challenges, we present an efficient and accurate framework to solve the license plate detection and recognition tasks simultaneously. It is a lightweight and unified deep neural network, that can be optimized end-to-end and work in real-time. Specifically, for unconstrained scenarios, an anchor-free method is adopted to efficiently detect the bounding box and four corners of a license plate, which are used to extract and rectify the target region features. Then, a novel convolutional neural network branch is designed to further extract features of characters without segmentation. Finally, the recognition task is treated as sequence labeling problems, which are solved by Connectionist Temporal Classification (CTC) directly. Several public datasets including images collected from different scenarios under various conditions are chosen for evaluation. Experimental results indicate that the proposed method significantly outperforms the previous state-of-the-art methods in both speed and precision. | ['Shuxin Qin', 'Sijiang Liu'] | 2020-08-25 | null | null | null | null | ['license-plate-detection'] | ['computer-vision'] | [ 2.12218806e-01 -9.67684388e-01 -1.45224437e-01 -3.47634614e-01
-8.27338398e-01 -6.88295782e-01 3.25615585e-01 -6.76634312e-01
-5.29903591e-01 3.38346362e-01 -5.38587451e-01 -3.08219492e-01
4.58802998e-01 -5.52084863e-01 -5.47705650e-01 -1.03254044e+00
4.66421753e-01 1.22636735e-01 6.95514679e-01 3.21974903e-02
5.97197890e-01 7.10740566e-01 -1.38515174e+00 1.38523683e-01
7.74321377e-01 1.21033120e+00 2.93116719e-01 3.17934036e-01
-1.62969217e-01 6.23916805e-01 -4.18319672e-01 -4.17135030e-01
4.47647899e-01 8.29724129e-03 -2.67280370e-01 6.08505666e-01
2.37214267e-01 -5.78646779e-01 -7.81373382e-01 1.18149555e+00
4.36836988e-01 1.35926917e-01 5.53192914e-01 -1.06354547e+00
-5.12559950e-01 -2.02583209e-01 -6.93788469e-01 2.50185788e-01
-2.91814744e-01 3.12091708e-01 5.13691425e-01 -1.09538019e+00
3.29761118e-01 7.09099591e-01 8.17304075e-01 4.73511994e-01
-4.86371994e-01 -8.86122882e-01 1.69983268e-01 3.60518217e-01
-1.64883828e+00 -5.37072182e-01 7.91157246e-01 -3.50653708e-01
6.04008019e-01 1.15116484e-01 4.14213777e-01 8.03490341e-01
-1.39768586e-01 1.17621112e+00 7.34164000e-01 -1.90943643e-01
-1.32660661e-02 6.44200742e-02 2.54234225e-01 7.08667576e-01
1.07086994e-01 -2.60625575e-02 1.00556955e-01 4.68362778e-01
8.92118037e-01 5.93621552e-01 -1.25058174e-01 -1.56197220e-01
-9.36831415e-01 5.63292086e-01 2.99603164e-01 1.59566611e-01
-2.55658329e-01 -1.25807360e-01 5.28766632e-01 -2.86223412e-01
2.82078505e-01 -2.31480658e-01 -2.63879150e-01 -1.82813242e-01
-9.56041336e-01 1.46604508e-01 4.49162662e-01 1.13041675e+00
6.82640254e-01 2.22114861e-01 -7.20396191e-02 1.05308974e+00
3.03471088e-01 5.12338638e-01 4.12752211e-01 -2.98363030e-01
6.94779158e-01 7.11256266e-01 1.81306705e-01 -1.17111480e+00
-3.74576241e-01 -3.54102850e-01 -8.52013111e-01 1.10064693e-01
4.16161865e-01 -2.35754415e-01 -1.13717353e+00 8.60225201e-01
1.34690627e-01 6.38847530e-01 8.94556567e-02 1.24472380e+00
7.17771530e-01 1.03147113e+00 -1.19637132e-01 -1.68314427e-02
1.57849526e+00 -1.41618979e+00 -6.51061833e-01 -5.78016520e-01
5.23999870e-01 -8.80261183e-01 6.58680201e-01 2.80623466e-01
-6.66352272e-01 -6.55141473e-01 -1.27989364e+00 6.88701682e-03
-4.92696226e-01 9.32342291e-01 4.71823961e-01 6.86652482e-01
-4.46309745e-01 1.79449245e-02 -6.75240517e-01 -4.66992669e-02
6.83333337e-01 3.77024055e-01 -2.30717182e-01 -4.46577460e-01
-9.43233728e-01 7.44381070e-01 4.77259934e-01 9.24199462e-01
-7.14224517e-01 -1.51728570e-01 -6.88049674e-01 -7.77705833e-02
4.59998131e-01 7.26489425e-02 1.18609393e+00 -1.03069282e+00
-1.60325575e+00 7.25790679e-01 -5.11231646e-02 -9.93122607e-02
5.45959532e-01 -4.65224795e-02 -7.16214240e-01 -2.88048908e-02
-7.56058991e-02 2.48208985e-01 7.26061285e-01 -1.07250237e+00
-1.07697213e+00 -2.92121857e-01 -2.72313148e-01 2.24851161e-01
-3.18698317e-01 3.62532735e-01 -1.28999960e+00 -5.20313144e-01
1.00739948e-01 -1.01722419e+00 -1.39925450e-01 -1.69018418e-01
-3.98031682e-01 -1.48334712e-01 1.44218731e+00 -9.98891830e-01
9.07672167e-01 -2.45508122e+00 -4.51062232e-01 5.22738919e-02
-4.56611961e-02 9.17879462e-01 4.56934050e-02 8.17954689e-02
1.67067096e-01 -8.75058845e-02 -3.42833966e-01 -4.12766904e-01
-1.01641215e-01 1.26047686e-01 -2.50577629e-01 7.30717778e-01
3.83918405e-01 7.28773892e-01 -3.66906911e-01 -5.08251131e-01
5.19756496e-01 2.74871737e-01 -9.13511869e-03 2.80735284e-01
9.56936628e-02 1.53426602e-01 -5.79282999e-01 9.90640223e-01
1.34572113e+00 -1.18371891e-02 -1.40479714e-01 -2.69219968e-02
-4.34340626e-01 -1.71347782e-01 -1.22467208e+00 1.17333329e+00
-3.42945516e-01 8.99990976e-01 5.77366650e-02 -1.21444321e+00
1.24322248e+00 1.68516576e-01 1.43873960e-01 -7.97905028e-01
2.59168267e-01 5.28851509e-01 -2.50710845e-01 -8.21869075e-01
4.65413749e-01 1.16182685e-01 -1.52051449e-01 -1.22493356e-02
-3.88170272e-01 2.92476445e-01 1.47447661e-01 -3.62168342e-01
7.71119595e-01 1.25049874e-01 -1.14486150e-01 1.35082290e-01
9.56710458e-01 1.60525993e-01 9.85175967e-01 3.41239750e-01
-4.92634803e-01 7.67298818e-01 1.27520785e-01 -6.90363944e-01
-1.03872907e+00 -8.32118154e-01 6.42008334e-02 8.48695338e-01
6.80073380e-01 3.00139368e-01 -7.57681549e-01 -7.35512853e-01
-2.86657214e-01 2.74275422e-01 -1.83682725e-01 2.14212090e-02
-1.02402127e+00 -7.62647808e-01 7.00145245e-01 7.80412138e-01
1.20714211e+00 -9.17160869e-01 -2.83179402e-01 2.43334532e-01
-1.18388586e-01 -1.51523209e+00 -6.76112056e-01 -2.45907679e-01
-6.13668442e-01 -9.69042480e-01 -9.61290061e-01 -1.41030264e+00
6.58674777e-01 7.36711085e-01 3.88605893e-01 1.83159322e-01
-2.46945694e-01 -3.98387372e-01 -2.27224901e-01 -1.80664271e-01
1.35578914e-02 1.21247299e-01 -1.98575199e-01 5.75624526e-01
6.12987697e-01 -1.66479811e-01 -6.29843593e-01 7.03229904e-01
-9.12720442e-01 3.47949378e-02 1.07853138e+00 6.61236644e-01
3.35510641e-01 1.45459026e-01 4.60272789e-01 -4.44496870e-01
3.86780262e-01 -2.56586313e-01 -1.01246130e+00 3.30697089e-01
-2.39373237e-01 -4.05959487e-01 7.40448713e-01 -3.80469799e-01
-1.17961562e+00 5.61624885e-01 -3.03337663e-01 -5.07563829e-01
-4.67655718e-01 3.91908914e-01 -3.82286131e-01 -2.56514907e-01
-5.00277393e-02 9.56463277e-01 -3.18826251e-02 -5.26210308e-01
1.50374873e-02 1.24526072e+00 8.12275290e-01 -1.52875364e-01
8.39203656e-01 3.56676102e-01 -3.77832055e-01 -9.70593870e-01
-5.30716836e-01 -7.67825305e-01 -7.20131457e-01 -4.89549935e-01
8.80856216e-01 -9.10538077e-01 -8.29026759e-01 1.14264941e+00
-1.37624097e+00 1.95524588e-01 5.79330921e-01 5.21276355e-01
-1.85157254e-01 5.39763987e-01 -6.72190726e-01 -8.78967822e-01
-2.59347528e-01 -1.44228351e+00 1.11410058e+00 6.60415411e-01
6.77355528e-01 -6.55196011e-01 -3.13832700e-01 5.70535004e-01
3.56059015e-01 3.93550918e-02 3.98567557e-01 -8.65987182e-01
-1.14789438e+00 -7.61209309e-01 -7.46744931e-01 5.80932200e-01
-5.77998683e-02 2.27610931e-01 -8.86636257e-01 -6.10080585e-02
-2.47198388e-01 -1.73498839e-01 8.37933004e-01 1.96842968e-01
1.06362915e+00 -2.46945098e-01 -5.66724896e-01 7.46139348e-01
1.46171927e+00 6.47345483e-01 9.39426363e-01 5.67390919e-01
7.47845113e-01 2.74452537e-01 6.40801370e-01 2.96558738e-01
2.83148229e-01 6.93050086e-01 3.77637535e-01 -2.19622433e-01
1.89074889e-01 -1.13892872e-02 4.25286025e-01 8.85451734e-01
3.38050304e-03 -2.26336628e-01 -1.06016076e+00 5.41695595e-01
-1.77478111e+00 -8.86117160e-01 -1.97371662e-01 1.83382607e+00
4.24372584e-01 1.98039502e-01 9.72572267e-02 3.62333804e-02
1.24621332e+00 1.67467937e-01 -5.64764917e-01 -6.65172786e-02
-7.49619491e-03 -3.90159041e-01 7.17680812e-01 -1.70644913e-02
-1.44686615e+00 1.19688940e+00 5.79760361e+00 1.18402421e+00
-1.56333029e+00 -1.44002840e-01 5.73859632e-01 4.43164915e-01
4.46393758e-01 -1.81477293e-01 -1.07259262e+00 6.95074975e-01
4.96873647e-01 1.64105520e-01 2.32247204e-01 9.47017670e-01
1.36123285e-01 3.17382693e-01 -7.08344281e-01 1.15611255e+00
9.04103145e-02 -1.49734426e+00 -2.76943773e-01 -4.22647074e-02
6.22430205e-01 5.74741587e-02 1.93111405e-01 5.74957788e-01
-2.31136441e-01 -8.24753821e-01 6.78406179e-01 4.59991187e-01
7.38628268e-01 -1.05236304e+00 1.11351550e+00 6.09412372e-01
-1.38113701e+00 -2.08461851e-01 -4.49362367e-01 7.34853521e-02
6.01387136e-02 2.36185938e-01 -7.68292367e-01 5.20204484e-01
3.07271123e-01 7.96421885e-01 -3.75794917e-01 1.37288642e+00
-8.24139193e-02 6.45413995e-01 -1.35368809e-01 -2.74601251e-01
7.80451715e-01 -3.99372309e-01 1.68295756e-01 1.52043772e+00
3.77538085e-01 3.66138816e-02 3.70411932e-01 7.36335099e-01
-1.36487007e-01 9.03575942e-02 -3.55067939e-01 2.39529926e-02
5.09029210e-01 1.43923748e+00 -8.91664684e-01 -1.75761268e-01
-7.23747015e-01 1.05546522e+00 4.72524613e-02 4.09149140e-01
-1.25659347e+00 -8.45451236e-01 4.92218316e-01 -1.86819151e-01
7.68076956e-01 -3.72889072e-01 -1.97978333e-01 -1.13609111e+00
3.40181679e-01 -6.65094972e-01 1.72878176e-01 -5.73269129e-01
-1.05202544e+00 4.63696957e-01 -5.38850784e-01 -1.67623878e+00
1.71634555e-02 -1.05995965e+00 -9.72392380e-01 7.59917617e-01
-1.67021263e+00 -1.18204451e+00 -4.30033505e-01 5.37763000e-01
8.58410299e-01 -4.02043790e-01 2.59394258e-01 6.70645833e-01
-1.37069058e+00 7.29423881e-01 6.14133298e-01 9.17488396e-01
3.79676908e-01 -5.14344156e-01 4.43492413e-01 1.34096217e+00
-2.94763148e-01 3.26619387e-01 1.22348949e-01 -4.39485103e-01
-1.40169084e+00 -1.46871531e+00 5.08573174e-01 1.25666186e-01
4.19786513e-01 -4.52522635e-01 -8.47474813e-01 5.58218300e-01
-5.43813854e-02 4.27961528e-01 3.29777628e-01 -6.35741353e-01
-2.00543761e-01 -3.20654541e-01 -9.51033413e-01 5.55930793e-01
6.23499870e-01 -2.28828803e-01 -3.32452387e-01 3.40315372e-01
4.35862809e-01 -5.16335487e-01 -3.89707536e-01 2.30963409e-01
5.04511595e-01 -6.44189596e-01 8.21022749e-01 -3.02816808e-01
3.42547238e-01 -6.68319046e-01 -7.02738613e-02 -8.19686770e-01
-1.53405398e-01 -3.33231688e-01 1.65384173e-01 1.36182714e+00
2.73873329e-01 -6.97578609e-01 9.54392791e-01 5.79723120e-01
-4.28710848e-01 -6.22014403e-01 -1.06554890e+00 -8.48232388e-01
-1.79976851e-01 -2.43024841e-01 4.71163005e-01 7.06945240e-01
-5.38458347e-01 1.11105286e-01 -5.61229646e-01 4.93814379e-01
4.08530056e-01 3.05947244e-01 6.60213411e-01 -8.51921022e-01
2.95375019e-01 -4.92372125e-01 -7.55677402e-01 -1.55612516e+00
1.69131190e-01 -4.96878952e-01 5.44836164e-01 -1.33070409e+00
1.38324305e-01 -3.63381058e-01 -1.43036991e-01 2.55735368e-01
-1.23651609e-01 4.29671109e-01 2.26357400e-01 4.46143270e-01
-7.00016141e-01 4.30926591e-01 1.07335436e+00 -4.12685364e-01
1.27670653e-02 3.09064627e-01 -3.11982810e-01 8.23726535e-01
8.14878285e-01 -3.11851680e-01 1.37530058e-03 -5.70070207e-01
-4.28584903e-01 4.59852591e-02 2.77249545e-01 -1.13534403e+00
6.28044069e-01 -2.04892740e-01 5.59435308e-01 -1.04234755e+00
3.54975879e-01 -8.65910411e-01 -3.27377558e-01 3.43448907e-01
8.20434093e-02 -8.46853778e-02 3.21654767e-01 5.80580175e-01
-4.55804735e-01 -4.91203368e-01 8.13164115e-01 9.43610296e-02
-1.41810703e+00 5.82114220e-01 -6.55205548e-01 -1.25206158e-01
1.41759968e+00 -5.66431463e-01 -2.86203414e-01 5.43977087e-03
-1.28067657e-01 1.69758722e-01 1.92006826e-01 4.70153391e-01
8.81954849e-01 -1.42916512e+00 -7.94108927e-01 2.76075184e-01
2.20270932e-01 6.08831234e-02 5.69301724e-01 6.87232435e-01
-1.01127768e+00 6.14180326e-01 -2.37386957e-01 -7.03059793e-01
-1.19608986e+00 5.66900611e-01 5.84510326e-01 -6.85533090e-03
-6.43568575e-01 6.47724271e-01 -6.49443939e-02 -4.77173567e-01
2.51611650e-01 -6.68229610e-02 -3.08135897e-01 -1.52775913e-01
7.25281537e-01 3.29148322e-01 1.91072300e-01 -8.73461962e-01
-5.87719142e-01 7.65897572e-01 -3.35213333e-01 1.74393058e-01
1.27223396e+00 -5.30898981e-02 8.38911384e-02 -5.99668510e-02
1.47898996e+00 -2.59038895e-01 -1.56680131e+00 -2.47337684e-01
-1.01970494e-01 -5.70531666e-01 4.61766571e-02 -5.20873308e-01
-1.48392534e+00 9.63833690e-01 5.91987252e-01 -1.18119523e-01
1.08655524e+00 -6.44141018e-01 1.14124012e+00 5.46967149e-01
9.26142633e-02 -1.32176864e+00 -1.40465796e-01 5.71124077e-01
5.21122098e-01 -1.17788339e+00 -2.20369473e-01 -4.32661027e-01
-7.04455733e-01 1.51353478e+00 8.56028855e-01 -2.24364713e-01
3.42629701e-01 2.96170384e-01 2.22608581e-01 2.36861452e-01
-1.81808636e-01 -2.06247661e-02 1.24543741e-01 3.98135930e-01
4.52344790e-02 -1.72310486e-01 -4.17631045e-02 7.76985943e-01
4.89628792e-01 5.88465780e-02 4.30849224e-01 1.01616633e+00
-5.43253064e-01 -6.76862121e-01 -5.18571973e-01 1.79578945e-01
-3.08967531e-01 8.45036209e-02 -1.56773433e-01 9.95514572e-01
3.18734378e-01 8.76036942e-01 1.64514482e-01 -5.49251497e-01
2.49899030e-01 -2.08320748e-02 -1.01195619e-01 -1.70240715e-01
-1.61377564e-01 2.52002738e-02 -4.98543531e-02 -8.39946643e-02
-2.28417307e-01 -6.64362013e-01 -1.22965324e+00 -3.38282853e-01
-5.82805157e-01 -1.04349568e-01 7.88353205e-01 1.07195354e+00
2.49506101e-01 4.52511996e-01 9.46204901e-01 -9.19545233e-01
-7.04202652e-01 -7.83536732e-01 -5.09544313e-01 1.51567042e-01
2.24057049e-01 -5.98600447e-01 -1.01280093e-01 2.68712789e-01] | [9.78979206085205, -4.985507488250732] |
69118286-3f1f-4e7a-8f68-aa58d1a17ff7 | texture-generation-with-neural-cellular | 2105.07299 | null | https://arxiv.org/abs/2105.07299v1 | https://arxiv.org/pdf/2105.07299v1.pdf | Texture Generation with Neural Cellular Automata | Neural Cellular Automata (NCA) have shown a remarkable ability to learn the required rules to "grow" images, classify morphologies, segment images, as well as to do general computation such as path-finding. We believe the inductive prior they introduce lends itself to the generation of textures. Textures in the natural world are often generated by variants of locally interacting reaction-diffusion systems. Human-made textures are likewise often generated in a local manner (textile weaving, for instance) or using rules with local dependencies (regular grids or geometric patterns). We demonstrate learning a texture generator from a single template image, with the generation method being embarrassingly parallel, exhibiting quick convergence and high fidelity of output, and requiring only some minimal assumptions around the underlying state manifold. Furthermore, we investigate properties of the learned models that are both useful and interesting, such as non-stationary dynamics and an inherent robustness to damage. Finally, we make qualitative claims that the behaviour exhibited by the NCA model is a learned, distributed, local algorithm to generate a texture, setting our method apart from existing work on texture generation. We discuss the advantages of such a paradigm. | ['Ettore Randazzo', 'Eyvind Niklasson', 'Alexander Mordvintsev'] | 2021-05-15 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 4.44623113e-01 3.94413948e-01 3.59442919e-01 1.76234975e-01
-2.74312198e-01 -8.44409645e-01 1.21473682e+00 -2.17773877e-02
4.28084321e-02 1.06257689e+00 1.45802004e-02 -1.67017654e-01
-2.33059719e-01 -1.28273261e+00 -1.02873373e+00 -1.44826615e+00
-2.78102100e-01 4.68953848e-01 3.22904319e-01 -4.18703854e-01
3.55069667e-01 8.54487419e-01 -1.73097968e+00 1.62492797e-01
4.60810214e-01 8.31208587e-01 4.03381959e-02 9.80420768e-01
3.24347913e-01 1.02234340e+00 -1.43364117e-01 -2.24950969e-01
1.82904795e-01 -8.21131647e-01 -9.74924445e-01 3.48131239e-01
-3.11995368e-03 2.06323668e-01 -1.47029459e-01 6.86862409e-01
2.70716906e-01 1.90487757e-01 1.14873827e+00 -8.09571981e-01
-7.93322027e-01 2.29184046e-01 -1.38216376e-01 -1.66054517e-01
2.18996197e-01 4.06096607e-01 4.74356592e-01 -6.51875675e-01
1.11633933e+00 9.90323305e-01 6.77534759e-01 8.96962702e-01
-1.72833014e+00 2.37662420e-02 -3.29736620e-01 -4.34476972e-01
-1.27461398e+00 -6.97899461e-01 5.29972374e-01 -5.63859344e-01
5.72051167e-01 4.58413333e-01 8.06922376e-01 1.32151663e+00
7.46613979e-01 3.60509127e-01 1.64819527e+00 -6.94534540e-01
6.02889776e-01 -8.03743675e-02 -5.80337286e-01 9.92041230e-01
1.34792015e-01 2.89921612e-01 -4.10687387e-01 -1.85457692e-01
1.56069791e+00 -3.61479014e-01 -2.57384256e-02 -4.47641164e-01
-1.40189254e+00 4.74005222e-01 2.09782109e-01 2.85007864e-01
-3.70293438e-01 5.56352854e-01 6.29149228e-02 3.52635682e-01
3.93220365e-01 6.15123928e-01 -2.88924132e-03 -6.11979589e-02
-4.01049823e-01 3.51797193e-01 9.54904199e-01 8.21365893e-01
8.75666678e-01 6.29668832e-02 2.03112975e-01 5.18569529e-01
-9.35629308e-02 4.68089700e-01 2.13286817e-01 -1.18703055e+00
-4.27885711e-01 1.70769468e-01 1.52502134e-01 -1.08607590e+00
-1.87420234e-01 -1.50301352e-01 -1.15638602e+00 6.61898136e-01
6.61303937e-01 -2.42248893e-01 -7.48139858e-01 1.85887110e+00
2.52381176e-01 -5.24989516e-03 4.24346924e-02 4.86152261e-01
-1.11454174e-01 7.58805454e-01 -1.63866580e-01 -2.93426812e-01
9.23881352e-01 -5.27574301e-01 -2.28919327e-01 4.63426560e-01
6.05644643e-01 -6.18428171e-01 1.04058635e+00 3.40029329e-01
-1.28015006e+00 -2.16265127e-01 -6.90463364e-01 2.20603481e-01
-4.15345192e-01 -3.69788706e-01 7.63352752e-01 2.71025300e-01
-1.33228123e+00 1.06401277e+00 -9.67849076e-01 -7.93076336e-01
2.81948119e-01 1.89917594e-01 -3.84795815e-01 3.89647990e-01
-6.27856135e-01 7.64612615e-01 -5.55216335e-02 -5.83035648e-02
-1.03905356e+00 -3.56489927e-01 -3.70135367e-01 -3.13921601e-01
2.93286331e-03 -1.05814803e+00 9.61552024e-01 -1.13789392e+00
-1.99159718e+00 1.01950920e+00 -1.01188041e-01 -3.74212354e-01
6.16487920e-01 5.18586099e-01 -2.48040874e-02 1.66467130e-01
-6.75656497e-02 8.50211501e-01 7.98696995e-01 -1.56192791e+00
-1.42771825e-01 4.23034690e-02 -1.03316531e-01 -9.19922814e-03
3.53060588e-02 -3.63648176e-01 3.20946395e-01 -8.66834044e-01
5.99739738e-02 -1.12755227e+00 -5.54609358e-01 1.62705466e-01
-3.62950653e-01 -9.53179747e-02 7.12772131e-01 -1.78306520e-01
6.78077698e-01 -1.91223240e+00 2.68071115e-01 5.54142237e-01
1.08033516e-01 -4.56291974e-01 -8.14259648e-02 9.40661371e-01
1.90470710e-01 3.50502014e-01 -4.63602483e-01 7.76512101e-02
-5.70015796e-02 4.20352906e-01 -3.73291343e-01 4.97632504e-01
3.16830158e-01 9.67045248e-01 -8.88392448e-01 -3.21580917e-01
-3.04238528e-01 4.59100425e-01 -5.12042344e-01 -7.73406774e-02
-3.78059596e-01 8.71471584e-01 -5.10631680e-01 4.26406533e-01
2.50214189e-02 -2.91905940e-01 6.42887354e-02 2.91079015e-01
-3.08587819e-01 -7.29985535e-02 -8.95190239e-01 1.48429716e+00
-5.00882149e-01 6.90384030e-01 2.24033501e-02 -8.38748813e-01
9.92456138e-01 2.77183890e-01 3.80787849e-01 -5.71680307e-01
1.02915436e-01 4.00472671e-01 8.97083338e-03 -4.48801488e-01
1.84267417e-01 -2.99782187e-01 2.04328652e-02 1.00538599e+00
-1.46752253e-01 -5.17954409e-01 -8.54706671e-03 5.87678403e-02
1.35020709e+00 2.48841256e-01 1.56730358e-02 -9.75638032e-01
8.99600238e-02 1.25338748e-01 9.00042057e-02 9.44861650e-01
3.49040687e-01 8.54813814e-01 5.48548520e-01 -7.16487765e-01
-1.75495470e+00 -1.17508745e+00 -1.20237984e-01 5.25492072e-01
1.97583795e-01 -1.72994286e-02 -1.13707447e+00 -1.85980558e-01
-3.87260079e-01 1.98033452e-01 -1.05136514e+00 -9.13558230e-02
-6.13329768e-01 -6.27392411e-01 5.95515788e-01 2.68625617e-01
3.99717271e-01 -1.55403924e+00 -7.15483427e-01 4.67677385e-01
2.87193239e-01 -6.40794873e-01 -1.62589446e-01 3.23202670e-01
-9.77405667e-01 -7.81938195e-01 -7.02696562e-01 -1.03986037e+00
1.08214056e+00 -1.40618935e-01 1.02787483e+00 3.26105356e-01
-3.82876426e-01 5.32424212e-01 -7.26025254e-02 -7.41807818e-02
-7.93036044e-01 2.82223374e-02 2.82382071e-01 3.28030646e-01
-5.12444198e-01 -1.03070617e+00 -6.23246789e-01 3.38825971e-01
-1.17845786e+00 4.35868382e-01 5.79166651e-01 8.01892996e-01
7.98039496e-01 2.65679687e-01 5.71276128e-01 -8.35549891e-01
7.74684668e-01 -3.21303636e-01 -2.79334247e-01 8.64611566e-02
-3.38019431e-01 2.63285726e-01 6.51554227e-01 -5.85615873e-01
-1.03638768e+00 1.39708400e-01 1.19733572e-01 1.33490294e-01
-5.81064105e-01 2.09525749e-01 1.78945661e-01 -4.29928780e-01
9.84443426e-01 5.02308846e-01 4.26601768e-01 -1.25932187e-01
4.15196598e-01 -5.16883982e-03 6.40592933e-01 -1.33651209e+00
8.53475988e-01 9.12584186e-01 3.39448035e-01 -1.06380963e+00
-1.66621372e-01 3.98124754e-01 -6.54293418e-01 -3.47912222e-01
7.06094682e-01 -2.37883776e-01 -7.84306586e-01 8.52717459e-01
-1.04473591e+00 -1.18221188e+00 -9.36117649e-01 -1.39873967e-01
-1.29304087e+00 -1.63059875e-01 -9.52884495e-01 -8.08174312e-01
2.43287422e-02 -7.24014938e-01 8.39638352e-01 1.31844476e-01
-4.79311496e-01 -1.34624910e+00 3.80365670e-01 -4.14237767e-01
7.75339484e-01 9.01025712e-01 1.18926346e+00 1.19372115e-01
-7.82188118e-01 9.16421488e-02 1.65205762e-01 -1.60511315e-01
2.26921797e-01 4.08143461e-01 -7.48891771e-01 -1.33595541e-01
-8.33106637e-02 -2.71493167e-01 5.74421406e-01 3.47720504e-01
9.93045270e-01 -6.96769655e-01 -4.52831835e-01 3.21009874e-01
1.33913970e+00 2.21040651e-01 9.09142256e-01 2.45849907e-01
4.74641979e-01 9.19519782e-01 -3.52580994e-01 1.04097605e-01
2.03762539e-02 4.32490706e-01 6.10698201e-02 -2.09687114e-01
-2.04758674e-01 -3.08384031e-01 1.90348297e-01 1.09064376e+00
-6.70104742e-01 -1.82929575e-01 -8.58126104e-01 4.67256278e-01
-1.82736492e+00 -1.08971155e+00 -1.00524396e-01 2.05926299e+00
9.54705775e-01 1.73162043e-01 1.43338099e-01 2.06983695e-03
5.89644969e-01 -2.06224263e-01 -3.48348588e-01 -6.09061062e-01
-6.05258405e-01 4.47569788e-01 4.21906799e-01 4.94383752e-01
-8.29411149e-01 1.00187230e+00 7.57435846e+00 8.54349375e-01
-1.16798401e+00 -1.23156846e-01 1.00350583e+00 2.46517867e-01
-5.45388222e-01 2.84048408e-01 -1.49975866e-01 2.89000511e-01
7.97600091e-01 6.51591495e-02 7.61968493e-01 3.02546144e-01
3.21595877e-01 -3.09031785e-01 -9.84344363e-01 5.34304500e-01
-3.90715063e-01 -1.79101229e+00 2.63217717e-01 4.58139539e-01
1.24982953e+00 -4.26263779e-01 4.14500795e-02 -4.64453369e-01
7.41839051e-01 -1.22545111e+00 9.43716466e-01 1.12008357e+00
8.23627114e-01 -5.55994153e-01 1.56382367e-01 3.52244794e-01
-9.06865656e-01 2.57754833e-01 -2.21988574e-01 -3.53960007e-01
1.16417013e-01 5.65084696e-01 -2.31454372e-01 1.43430727e-02
4.37016100e-01 4.65730280e-01 -4.76797849e-01 8.04678738e-01
8.52181539e-02 7.00013101e-01 -5.15244365e-01 -3.07348013e-01
2.59502500e-01 -4.56671983e-01 5.47089934e-01 1.12857866e+00
5.02123773e-01 3.75626266e-01 -2.21073285e-01 1.06993222e+00
2.07142562e-01 -4.25002053e-02 -1.24182594e+00 4.05099317e-02
9.70072895e-02 1.11321008e+00 -1.39084303e+00 -2.03477353e-01
1.29327893e-01 8.60752046e-01 1.46105900e-01 5.05029023e-01
-4.49239582e-01 -3.00385982e-01 4.26719934e-01 4.96120155e-01
1.71438366e-01 -5.37996948e-01 -4.45283800e-01 -9.99378979e-01
-1.30426466e-01 -7.81501949e-01 -4.72488493e-01 -1.02471960e+00
-1.30855310e+00 6.16757393e-01 -3.31471145e-01 -1.00318718e+00
-1.68447226e-01 -5.06103098e-01 -7.94902682e-01 5.01221120e-01
-6.74237788e-01 -1.21298528e+00 -4.15078662e-02 6.32116854e-01
1.51667044e-01 -1.36284586e-02 9.78537679e-01 -2.84316063e-01
-2.69165516e-01 1.21070139e-01 1.22049816e-01 -3.09623647e-02
2.28297338e-01 -1.20664382e+00 4.00930554e-01 6.72019243e-01
1.97277859e-01 6.48307085e-01 7.64486849e-01 -5.08329034e-01
-1.41824257e+00 -7.67755628e-01 6.54586673e-01 -5.75409174e-01
7.65765667e-01 -5.51507175e-01 -8.40136528e-01 4.00014311e-01
3.71814609e-01 3.11218332e-02 5.85531741e-02 -2.88238853e-01
-1.27012089e-01 9.13434997e-02 -9.50875580e-01 9.78369832e-01
1.40987611e+00 -6.08301282e-01 4.84694541e-02 3.21020067e-01
2.53749818e-01 -2.24293932e-01 -7.44710147e-01 1.01416245e-01
8.11285734e-01 -1.02356970e+00 5.48390865e-01 -5.80265701e-01
7.12540567e-01 -2.42849857e-01 2.30230112e-02 -1.39355135e+00
-5.87655246e-01 -1.18840325e+00 2.41977230e-01 1.21673775e+00
6.33984149e-01 -8.11252415e-01 7.55692363e-01 4.01021838e-01
3.14722136e-02 -8.75938237e-01 -6.65564716e-01 -8.72024119e-01
4.71074998e-01 2.14897413e-02 4.85350639e-01 1.04405475e+00
-7.13297725e-02 9.86259431e-03 -8.58919844e-02 -1.48202002e-01
5.10462642e-01 -7.73678645e-02 7.12628663e-01 -1.23701108e+00
-3.17938656e-01 -7.71814585e-01 -3.57661009e-01 -7.23254263e-01
-9.62922946e-02 -7.25269198e-01 2.11101353e-01 -1.18293798e+00
3.81918363e-02 -9.82189775e-01 2.95733567e-02 2.81618148e-01
4.72012192e-01 4.40412849e-01 -2.92250514e-01 5.09732842e-01
-2.62044996e-01 2.98988551e-01 1.56880832e+00 1.22322515e-01
-8.59900415e-02 -3.01742196e-01 -5.78761697e-01 6.81671798e-01
7.89231777e-01 -4.24780130e-01 -3.38909119e-01 -2.24949375e-01
7.45480776e-01 1.67822987e-02 6.92786276e-01 -1.07501125e+00
4.07176673e-01 -4.43520308e-01 2.71948755e-01 3.49638551e-01
-1.86500587e-02 -3.38933498e-01 6.41197979e-01 3.35237682e-01
-5.14156520e-01 1.59915432e-01 -4.06673327e-02 5.92677951e-01
6.55228198e-02 7.76504874e-02 8.52743030e-01 -4.25678611e-01
-2.57584602e-01 1.62742645e-01 -1.02109015e+00 -7.00357705e-02
1.02254522e+00 -4.55373466e-01 -3.35652143e-01 -5.10326803e-01
-8.76436949e-01 -4.75624025e-01 1.00271451e+00 -9.81826261e-02
2.22780794e-01 -1.47236311e+00 -5.14136791e-01 2.87838668e-01
-2.55426496e-01 1.14588767e-01 -5.44788651e-02 7.24415004e-01
-9.15733337e-01 -4.92133200e-02 -5.11325777e-01 -5.96413314e-01
-3.67546350e-01 3.88357013e-01 4.89449590e-01 6.04956262e-02
-7.94274926e-01 6.17008686e-01 3.73332083e-01 -3.05315316e-01
-4.23150033e-01 -1.35314927e-01 3.78574282e-01 -2.96338260e-01
1.32466152e-01 1.23438507e-01 -1.06196642e-01 -3.03498656e-01
1.16302662e-01 8.57653797e-01 3.96815121e-01 -5.00583231e-01
1.35335720e+00 -1.21100850e-01 -4.75875586e-01 5.47914684e-01
7.22890258e-01 1.21314071e-01 -1.61278820e+00 9.21484455e-02
-1.99520424e-01 1.16519786e-01 -4.76871341e-01 -4.89513814e-01
-7.68127620e-01 8.05320263e-01 1.03351526e-01 9.30417180e-01
8.04374933e-01 8.92873928e-02 5.87659657e-01 3.25110286e-01
8.49684477e-01 -1.03458118e+00 2.10607022e-01 5.29454589e-01
9.98516321e-01 -6.60356820e-01 -3.87003899e-01 -1.88201874e-01
-2.73627669e-01 1.29162884e+00 2.29903102e-01 -7.21394062e-01
7.05931365e-01 7.69886672e-01 1.28563168e-02 -2.18084544e-01
-1.07677472e+00 6.36393651e-02 -1.53170198e-01 7.88783908e-01
2.12701038e-01 1.44261748e-01 -5.59157282e-02 -1.80433646e-01
-3.83653939e-01 -2.02267766e-01 7.90630460e-01 1.02291989e+00
-3.54736447e-01 -9.82745111e-01 -1.87326878e-01 2.46559560e-01
-1.88510567e-01 1.96418583e-01 -5.80110073e-01 7.87647367e-01
5.85303865e-02 4.81894821e-01 1.55533418e-01 -2.93076545e-01
-6.30503893e-02 6.75908625e-02 8.40386689e-01 -3.50943208e-01
-2.15995669e-01 1.12009197e-01 -5.08267917e-02 -4.21667516e-01
-5.10191560e-01 -8.62787306e-01 -9.92850959e-01 -6.49016500e-01
-1.44703286e-02 4.95707393e-02 3.39037418e-01 9.69493091e-01
5.05622745e-01 2.30774507e-01 5.56482553e-01 -1.15422356e+00
1.49116218e-01 -5.26606321e-01 -8.08761060e-01 4.20402110e-01
2.41548195e-01 -5.06716728e-01 -3.14753026e-01 7.63625443e-01] | [11.454760551452637, -0.3632754683494568] |
eb1d6c9d-18c5-4907-942b-929572e77a38 | data-augmentation-via-dependency-tree-1 | 1903.09460 | null | http://arxiv.org/abs/1903.09460v1 | http://arxiv.org/pdf/1903.09460v1.pdf | Data Augmentation via Dependency Tree Morphing for Low-Resource Languages | Neural NLP systems achieve high scores in the presence of sizable training
dataset. Lack of such datasets leads to poor system performances in the case
low-resource languages. We present two simple text augmentation techniques
using dependency trees, inspired from image processing. We crop sentences by
removing dependency links, and we rotate sentences by moving the tree fragments
around the root. We apply these techniques to augment the training sets of
low-resource languages in Universal Dependencies project. We implement a
character-level sequence tagging model and evaluate the augmented datasets on
part-of-speech tagging task. We show that crop and rotate provides improvements
over the models trained with non-augmented data for majority of the languages,
especially for languages with rich case marking systems. | ['Mark Steedman', 'Gözde Gül Şahin'] | 2019-03-22 | data-augmentation-via-dependency-tree | https://aclanthology.org/D18-1545 | https://aclanthology.org/D18-1545.pdf | emnlp-2018-10 | ['text-augmentation'] | ['natural-language-processing'] | [ 4.58295703e-01 1.38489127e-01 -2.29429051e-01 -3.80466342e-01
-7.28566647e-01 -7.49746621e-01 8.02548885e-01 -5.87292835e-02
-1.05608809e+00 1.03295887e+00 5.75735509e-01 -4.44833875e-01
4.74899143e-01 -4.03135628e-01 -5.52475274e-01 -4.46737081e-01
1.57502502e-01 5.92589200e-01 2.10636452e-01 -4.85454023e-01
-9.70621109e-02 4.20646876e-01 -1.13805282e+00 7.43556678e-01
8.07792425e-01 -1.72010511e-02 4.02759045e-01 6.68817043e-01
-5.89443624e-01 7.09890366e-01 -6.03684843e-01 -6.79398715e-01
2.17807412e-01 -2.72585928e-01 -8.27416718e-01 1.54505208e-01
5.58265686e-01 1.10312290e-02 -3.11533868e-01 9.34831977e-01
3.03503007e-01 -1.52552754e-01 4.70637977e-01 -1.05441344e+00
-9.32394445e-01 1.04207492e+00 -5.82115591e-01 4.03252840e-01
1.49430469e-01 -1.35249242e-01 9.15752709e-01 -9.49548423e-01
9.48337495e-01 1.22685039e+00 6.33860052e-01 8.27207088e-01
-1.09312010e+00 -4.80965436e-01 2.35245273e-01 -1.99457750e-01
-1.17677593e+00 -6.04547858e-01 5.05708158e-01 -2.86176145e-01
1.70619333e+00 -5.35179749e-02 3.74851972e-01 1.00092494e+00
-1.16200931e-01 8.26184094e-01 1.09939992e+00 -1.12496233e+00
-2.89358824e-01 1.40745535e-01 3.69590670e-01 7.50642300e-01
2.25774825e-01 -1.77972913e-01 -5.59702039e-01 7.19026551e-02
5.44396043e-01 -2.83898413e-01 1.77826323e-02 -1.20339869e-02
-1.49112582e+00 7.30278611e-01 1.39303580e-01 8.15391600e-01
-2.49964118e-01 2.71066055e-02 6.68896616e-01 3.32627684e-01
6.32201314e-01 4.64374661e-01 -9.68991458e-01 1.56406045e-01
-7.57746041e-01 -1.02022979e-02 6.70403600e-01 1.24448562e+00
4.91390407e-01 2.69449025e-01 -4.40475404e-01 1.14356756e+00
7.10447654e-02 5.15711546e-01 7.51855016e-01 -5.40259421e-01
1.00075626e+00 4.95574117e-01 -1.35318816e-01 1.89704262e-02
-3.62918198e-01 -2.74020851e-01 -7.81449735e-01 -7.24563375e-02
3.95083308e-01 -4.09259707e-01 -1.41435397e+00 1.89772582e+00
6.22425601e-02 -4.77618158e-01 5.80949843e-01 1.85349211e-01
6.14524662e-01 7.19866514e-01 5.45221627e-01 -2.19149321e-01
1.60688937e+00 -1.31340098e+00 -1.02915859e+00 -5.82814872e-01
1.36711001e+00 -9.10272300e-01 1.31024778e+00 1.10781536e-01
-8.80548358e-01 -5.63503265e-01 -9.59836185e-01 -1.96075991e-01
-7.31696308e-01 5.72934389e-01 9.01125312e-01 8.78533185e-01
-1.26778889e+00 2.84973174e-01 -7.74582922e-01 -5.55517435e-01
3.09139371e-01 4.19279128e-01 -8.88657331e-01 -2.99595386e-01
-1.20401442e+00 1.14669204e+00 7.88294137e-01 -2.06286967e-01
-4.22927499e-01 -3.22420865e-01 -1.15722597e+00 -4.11433578e-01
1.71711799e-02 -3.62514526e-01 1.25787556e+00 -1.03829765e+00
-1.04633009e+00 1.24058092e+00 -2.51698703e-01 -5.51284909e-01
2.43345588e-01 -4.66915101e-01 -4.63285059e-01 -1.91890001e-01
9.26155075e-02 1.22391725e+00 5.42431116e-01 -8.74713302e-01
-6.00209296e-01 -2.97315985e-01 -2.81612694e-01 2.20762238e-01
-6.46503747e-01 6.13077700e-01 -4.38939214e-01 -8.94635439e-01
-1.82631388e-01 -1.06175423e+00 -4.63191807e-01 -4.10597444e-01
-3.48165363e-01 -2.21240088e-01 8.89045954e-01 -1.05764246e+00
1.03535104e+00 -2.14618897e+00 -2.49069363e-01 -3.50477755e-01
-3.36948693e-01 5.80958366e-01 -6.20022595e-01 4.66542453e-01
-3.91098022e-01 5.65737665e-01 -5.03289282e-01 -5.02802730e-01
-2.86809236e-01 8.01384747e-01 -2.25168586e-01 9.99200270e-02
8.58795583e-01 8.14512372e-01 -6.48429334e-01 -7.52193511e-01
7.28922710e-02 4.95177180e-01 -5.08482635e-01 5.80202136e-03
-2.61573732e-01 3.29431742e-01 -1.27489731e-01 5.35728633e-01
4.87407029e-01 1.72847286e-01 1.94255486e-01 2.02916101e-01
-2.40233704e-01 7.72561669e-01 -8.52949619e-01 1.88052893e+00
-6.12020552e-01 7.01792538e-01 -1.94019094e-01 -6.55906677e-01
8.36733758e-01 5.83009362e-01 -1.13982402e-01 -4.71324056e-01
-1.10527545e-01 1.94509491e-01 4.54870105e-01 -5.46201766e-01
7.52013147e-01 -3.01110089e-01 -2.70986021e-01 2.87173927e-01
4.36324567e-01 -2.20717654e-01 7.03287780e-01 3.80256087e-01
1.27121198e+00 3.65148008e-01 5.38087428e-01 6.43409193e-02
5.39402962e-01 1.31890342e-01 3.83450270e-01 6.42107725e-01
-2.04157591e-01 5.56202352e-01 3.77040476e-01 -2.99394459e-01
-1.62602520e+00 -7.13997483e-01 -1.76363245e-01 1.57140374e+00
-9.79771495e-01 -2.61624306e-01 -7.55643487e-01 -9.62938368e-01
-4.39142436e-01 8.05070758e-01 -5.50840259e-01 3.16035807e-01
-9.74639177e-01 -1.07044756e+00 8.54130983e-01 8.95503640e-01
4.79406446e-01 -1.58739424e+00 9.91052911e-02 1.96026579e-01
-2.93783695e-02 -1.64779854e+00 -1.41129687e-01 6.81482494e-01
-8.45517814e-01 -5.19118130e-01 -8.20328355e-01 -1.30792904e+00
8.36796761e-01 -2.29529977e-01 1.10571921e+00 2.23397002e-01
1.47593334e-01 -1.66285515e-01 -6.23599589e-01 -4.50343847e-01
-7.68091917e-01 3.77451986e-01 3.09646949e-02 -6.28854871e-01
4.75014925e-01 -1.73840374e-01 3.54806095e-01 -1.94612250e-01
-8.16380560e-01 2.69327492e-01 9.34107780e-01 8.20901394e-01
4.52805907e-01 -3.75099033e-01 5.53614378e-01 -1.29987025e+00
2.76097447e-01 -1.48695905e-03 -3.95496875e-01 2.96658784e-01
-6.06937483e-02 4.22106326e-01 4.84575450e-01 -4.81999546e-01
-1.28457928e+00 6.35295570e-01 -5.38173854e-01 1.91318378e-01
-4.16694492e-01 3.15566748e-01 -3.27395737e-01 1.91399142e-01
6.67934835e-01 -6.02912903e-02 -5.79404473e-01 -5.77354372e-01
7.28103399e-01 8.18175852e-01 6.46434128e-01 -4.65547115e-01
7.04132080e-01 1.84750482e-01 -2.69933730e-01 -1.04385567e+00
-9.61230218e-01 -4.94872570e-01 -1.27575719e+00 3.58399153e-01
8.82233500e-01 -1.03564632e+00 3.14814955e-01 4.10504133e-01
-1.51944613e+00 -4.81914699e-01 -4.40607488e-01 6.67465210e-01
-1.75601259e-01 3.05867940e-01 -7.87746072e-01 -7.92184532e-01
-3.83410960e-01 -7.94862032e-01 9.32361007e-01 -1.02255255e-01
-1.93768144e-01 -9.10493076e-01 3.94313902e-01 2.53854841e-01
5.44523112e-02 -1.29010156e-01 1.19200659e+00 -1.25272632e+00
-1.86089445e-02 -3.00558209e-01 -1.36148766e-01 6.03282273e-01
2.65817083e-02 1.13291010e-01 -1.07718158e+00 -5.11898771e-02
-3.63413662e-01 -3.67753595e-01 9.59269643e-01 -9.21117980e-03
6.33058608e-01 -2.28866532e-01 -3.30217957e-01 2.90655851e-01
1.14357555e+00 1.01784833e-01 7.01001644e-01 3.55438948e-01
8.30885410e-01 7.28811979e-01 3.90786976e-01 -1.80042360e-03
7.70843625e-02 3.24451119e-01 -1.91209674e-01 -2.50076383e-01
-5.64476728e-01 -2.52571046e-01 5.41615844e-01 1.33016253e+00
6.49138242e-02 -6.21370375e-01 -1.20290458e+00 9.97070491e-01
-1.44603884e+00 -5.33553481e-01 -5.41577160e-01 1.86405814e+00
1.31349242e+00 3.58142912e-01 -1.69629052e-01 3.15181911e-01
8.43104601e-01 7.09206040e-04 8.53648558e-02 -6.51937962e-01
-6.06379569e-01 5.83164215e-01 6.16340101e-01 5.58958530e-01
-1.28717434e+00 1.58598340e+00 7.01395082e+00 7.07989812e-01
-6.36364758e-01 4.97167379e-01 2.72274762e-01 1.46220326e-01
-1.22484192e-01 1.96655959e-01 -1.15544903e+00 1.02477660e-03
1.23851323e+00 4.51573789e-01 -5.68624996e-02 7.01205432e-01
-2.65161276e-01 3.18753123e-02 -9.78919506e-01 4.78090882e-01
1.96458980e-01 -1.09445500e+00 2.93298274e-01 -1.98626906e-01
9.16667283e-01 5.34339845e-01 -1.61255196e-01 5.94045222e-01
8.00896645e-01 -8.10448706e-01 3.76780421e-01 5.32660633e-02
8.58432353e-01 -5.45384765e-01 8.98429215e-01 2.51905233e-01
-8.52733016e-01 3.59624296e-01 -4.05123949e-01 -2.21677035e-01
3.09026450e-01 3.33108991e-01 -1.07309854e+00 7.18793049e-02
3.52148056e-01 2.55973935e-01 -9.78491843e-01 8.29555988e-01
-7.86417305e-01 7.41011739e-01 -3.48947734e-01 -6.51275143e-02
2.17821658e-01 9.88173485e-02 3.24050248e-01 1.81597447e+00
8.93849730e-02 -8.89827609e-02 5.81962168e-02 8.43329281e-02
-4.40569460e-01 5.28319359e-01 -7.95483649e-01 -3.33429337e-01
2.68407345e-01 1.37340975e+00 -7.96186924e-01 -6.92937613e-01
-8.55184853e-01 1.04329371e+00 6.63636267e-01 1.96675733e-01
-3.05887580e-01 -4.06443268e-01 7.10646361e-02 -4.91302386e-02
2.92500407e-01 -4.59914029e-01 -2.86782116e-01 -1.06619906e+00
-1.07508227e-01 -8.31318200e-01 4.91728812e-01 -1.02550590e+00
-1.13302326e+00 9.26760077e-01 -5.14197536e-02 -7.14890480e-01
-3.79787147e-01 -8.62698019e-01 -2.26691112e-01 9.23181653e-01
-1.55339241e+00 -1.66031373e+00 1.38108626e-01 3.94934893e-01
7.15430558e-01 -4.52758580e-01 1.12273920e+00 3.04024220e-01
-6.07499719e-01 5.81627250e-01 -5.70697412e-02 7.86388159e-01
9.36265588e-01 -1.29609752e+00 9.47053254e-01 1.18498051e+00
7.40628064e-01 4.87108499e-01 4.28642839e-01 -9.01870310e-01
-8.28343928e-01 -1.25136721e+00 1.57564068e+00 -5.27924299e-01
8.37934196e-01 -6.93093419e-01 -1.17308283e+00 1.10632706e+00
6.43292665e-01 1.10151924e-01 8.24565947e-01 3.21665198e-01
-5.99864960e-01 4.26566780e-01 -1.06015801e+00 7.29752719e-01
1.06369650e+00 -6.15011811e-01 -1.20219207e+00 6.66003466e-01
9.17204976e-01 -1.06123194e-01 -5.43533742e-01 2.79407382e-01
1.59031615e-01 -1.96052343e-01 5.81928909e-01 -1.02505934e+00
3.82718414e-01 -1.81054175e-01 -2.14094833e-01 -1.23399675e+00
-1.51951581e-01 -5.28990269e-01 4.13562775e-01 1.67927623e+00
1.02454436e+00 -4.19080555e-01 6.84845507e-01 2.86527097e-01
-3.50534827e-01 3.87779139e-02 -8.84630382e-01 -7.94454873e-01
3.34427088e-01 -5.35238862e-01 2.07027480e-01 1.13163030e+00
-5.71686812e-02 7.42187679e-01 -1.91871494e-01 -1.09230407e-01
1.84900105e-01 -6.03120387e-01 5.14262497e-01 -1.02580464e+00
-1.44962803e-01 -1.24619817e-02 -2.43800595e-01 -4.68404293e-01
5.25023460e-01 -1.09448457e+00 1.83601066e-01 -1.58466494e+00
3.13522846e-01 -5.24127841e-01 -5.81608191e-02 1.18094897e+00
-1.88154548e-01 6.21805310e-01 1.71376184e-01 1.37164369e-01
-4.49952215e-01 3.11396092e-01 9.00881410e-01 1.04653828e-01
-2.69664228e-01 -5.13239861e-01 -3.83847475e-01 1.15948451e+00
9.20889318e-01 -8.27120960e-01 8.32398087e-02 -7.12367475e-01
3.75850588e-01 -3.26227069e-01 -2.11545601e-01 -1.09693336e+00
-1.34702235e-01 2.65962388e-02 4.74101067e-01 -7.41129160e-01
1.60654381e-01 -5.43975532e-01 -4.19747114e-01 5.38717449e-01
-4.78674799e-01 3.46687019e-01 5.28814316e-01 4.32769842e-02
-1.47450611e-01 -7.10171402e-01 7.31967509e-01 -2.91812181e-01
-6.98898613e-01 2.14460958e-02 -6.53407276e-01 2.21220523e-01
5.70690334e-01 5.45799620e-02 -4.45023060e-01 1.51007205e-01
-7.92350948e-01 -4.65985723e-02 3.69517803e-01 5.14049888e-01
1.00049771e-01 -1.19529569e+00 -9.54109132e-01 1.99349657e-01
3.51617306e-01 -2.62095690e-01 -2.19353735e-01 4.59050119e-01
-6.32781148e-01 6.52326584e-01 -3.69121552e-01 -1.22147515e-01
-1.53086782e+00 7.38046825e-01 -1.37484921e-02 -6.29167438e-01
-3.00548941e-01 8.10875833e-01 1.38155341e-01 -7.30015874e-01
-3.31273824e-02 -3.84598553e-01 -5.58886886e-01 5.67285568e-02
5.14423847e-01 -1.78184554e-01 3.29101026e-01 -8.10601592e-01
-4.42709684e-01 2.82965153e-01 -6.43549263e-01 -6.70666754e-01
1.38349473e+00 1.07612573e-01 -1.47384599e-01 2.98304915e-01
9.06989694e-01 3.16705316e-01 -7.71320343e-01 -3.45694989e-01
2.63513565e-01 7.75088742e-02 -7.32890517e-02 -8.78087878e-01
-6.26290977e-01 1.06288850e+00 4.97912645e-01 -1.88901946e-01
8.49203289e-01 8.19506496e-02 4.13980126e-01 7.77326286e-01
2.87016761e-02 -1.09578371e+00 -8.09775889e-02 1.18687105e+00
7.84585059e-01 -1.12089753e+00 2.01472994e-02 -5.43864906e-01
-8.43310118e-01 6.63175821e-01 6.78633869e-01 -6.03608489e-02
3.20778757e-01 7.75157630e-01 1.53215185e-01 1.48374155e-01
-8.39785039e-01 -6.33290529e-01 1.48682058e-01 9.91367698e-01
1.05717969e+00 -6.14173897e-02 -5.98033071e-01 5.09984672e-01
-5.21720886e-01 -3.81252974e-01 6.80629730e-01 1.13225079e+00
-3.82103294e-01 -1.57360756e+00 -3.32700819e-01 3.45799714e-01
-9.13477063e-01 -8.34603548e-01 -6.37204945e-01 1.25416970e+00
3.17523450e-01 6.73812330e-01 2.70876586e-01 1.72239505e-02
2.52976716e-01 7.59817719e-01 5.51330745e-01 -1.07550967e+00
-8.15404356e-01 1.82203844e-01 6.71424687e-01 1.71860576e-01
-5.19394636e-01 -8.20701241e-01 -1.32969844e+00 2.30570003e-01
-3.89316410e-01 5.03592230e-02 8.39940190e-01 1.14740586e+00
-1.91267654e-02 5.12846887e-01 -4.32877913e-02 -4.91642296e-01
-9.55000147e-02 -1.56714308e+00 -2.67876297e-01 3.38861674e-01
1.53603271e-01 -2.14721084e-01 -5.72329853e-03 3.98937941e-01] | [10.367616653442383, 9.933441162109375] |
09d6a6ef-80dd-4c62-bff3-3edc175f1656 | parallel-multi-dimensional-lstm-with | 1506.07452 | null | http://arxiv.org/abs/1506.07452v1 | http://arxiv.org/pdf/1506.07452v1.pdf | Parallel Multi-Dimensional LSTM, With Application to Fast Biomedical Volumetric Image Segmentation | Convolutional Neural Networks (CNNs) can be shifted across 2D images or 3D
videos to segment them. They have a fixed input size and typically perceive
only small local contexts of the pixels to be classified as foreground or
background. In contrast, Multi-Dimensional Recurrent NNs (MD-RNNs) can perceive
the entire spatio-temporal context of each pixel in a few sweeps through all
pixels, especially when the RNN is a Long Short-Term Memory (LSTM). Despite
these theoretical advantages, however, unlike CNNs, previous MD-LSTM variants
were hard to parallelize on GPUs. Here we re-arrange the traditional cuboid
order of computations in MD-LSTM in pyramidal fashion. The resulting
PyraMiD-LSTM is easy to parallelize, especially for 3D data such as stacks of
brain slice images. PyraMiD-LSTM achieved best known pixel-wise brain image
segmentation results on MRBrainS13 (and competitive results on EM-ISBI12). | ['Wonmin Byeon', 'Marijn F. Stollenga', 'Marcus Liwicki', 'Juergen Schmidhuber'] | 2015-06-24 | parallel-multi-dimensional-lstm-with-1 | http://papers.nips.cc/paper/5642-parallel-multi-dimensional-lstm-with-application-to-fast-biomedical-volumetric-image-segmentation | http://papers.nips.cc/paper/5642-parallel-multi-dimensional-lstm-with-application-to-fast-biomedical-volumetric-image-segmentation.pdf | neurips-2015-12 | ['brain-image-segmentation'] | ['medical'] | [ 5.84039629e-01 1.78728044e-01 5.05419895e-02 -3.54685426e-01
-6.76721632e-01 -3.46449167e-01 3.09309423e-01 -1.15040570e-01
-9.40589666e-01 4.67187166e-01 5.57537824e-02 -5.15832543e-01
2.24646062e-01 -6.60643756e-01 -9.57907259e-01 -9.02010918e-01
-2.77735054e-01 3.45196813e-01 6.12142146e-01 1.58088818e-01
3.92746963e-02 6.77381456e-01 -1.12937379e+00 7.83926547e-01
1.60318241e-01 9.67632949e-01 7.02596903e-01 8.25489879e-01
-1.92835674e-01 8.53217781e-01 -4.00782257e-01 2.11058006e-01
4.99234954e-03 -2.72018969e-01 -1.06464016e+00 2.34372634e-02
8.11050892e-01 -3.48683298e-01 -6.11009359e-01 9.97031927e-01
6.49238586e-01 1.99798241e-01 2.57746339e-01 -5.75372756e-01
-6.47258520e-01 6.46586835e-01 -6.27518654e-01 8.88721943e-01
-1.63791165e-01 -3.06572020e-02 5.51408291e-01 -9.36405420e-01
8.11795533e-01 1.14957607e+00 6.58917964e-01 6.92903578e-01
-1.29297340e+00 -2.09613055e-01 5.74784994e-01 2.07119286e-01
-7.96571314e-01 4.01734263e-02 2.99557239e-01 -2.56386667e-01
1.49256527e+00 7.88054392e-02 9.98922288e-01 1.11453891e+00
5.87317467e-01 1.11024690e+00 1.22088432e+00 -1.77547082e-01
1.63344145e-01 -6.51110351e-01 4.03895736e-01 6.07791543e-01
-1.84574425e-01 -2.95177519e-01 -3.92964989e-01 8.47403556e-02
1.32248938e+00 3.68069381e-01 -1.79308310e-01 -1.63302109e-01
-1.75181687e+00 6.10659003e-01 9.51752901e-01 5.54711580e-01
-5.39133608e-01 4.09919769e-01 7.22042024e-01 2.63414651e-01
4.62760776e-01 2.88655353e-03 -4.87955064e-01 2.09060445e-01
-1.30602360e+00 -3.21894661e-02 2.07293600e-01 5.66091895e-01
4.92615789e-01 4.50949110e-02 -2.81861693e-01 7.71346867e-01
-8.34603533e-02 2.00200886e-01 8.13413382e-01 -7.78600276e-01
2.80607879e-01 3.53574574e-01 -4.12198097e-01 -6.80464625e-01
-9.09995735e-01 -3.91772985e-01 -1.25760758e+00 2.31406599e-01
3.48101288e-01 -2.76648611e-01 -1.62263417e+00 1.37199044e+00
1.89931229e-01 3.58123064e-01 -2.71914542e-01 1.09044373e+00
9.92296457e-01 6.48163319e-01 -5.60838953e-02 -9.54757929e-02
1.53159499e+00 -1.10516834e+00 -5.18163621e-01 -5.20138025e-01
7.10325956e-01 -4.08307701e-01 8.39074433e-01 2.61476696e-01
-1.29406047e+00 -4.91126627e-01 -8.36791515e-01 -3.92921269e-01
-4.86904263e-01 -1.35170579e-01 6.10054493e-01 3.15202534e-01
-1.79568207e+00 6.04210436e-01 -1.30591035e+00 -1.21089540e-01
7.29439676e-01 5.98653436e-01 -2.83032387e-01 1.25064254e-01
-1.12228107e+00 8.06140959e-01 3.57448161e-01 6.55220866e-01
-1.01850045e+00 -8.13121259e-01 -6.32136405e-01 -2.43724480e-01
7.14727566e-02 -7.55752802e-01 1.30625248e+00 -9.50985610e-01
-1.32105207e+00 1.31511259e+00 -4.90399837e-01 -8.87400150e-01
4.37967777e-01 -8.16370472e-02 -3.81334051e-02 4.57820147e-01
-8.88427645e-02 1.27897167e+00 9.53432202e-01 -7.47331560e-01
-5.23766100e-01 -6.32475019e-01 1.69981383e-02 2.32980847e-01
2.78106302e-01 2.05296874e-01 -3.99010301e-01 -7.55914152e-01
7.28314877e-01 -9.73525405e-01 -9.05867994e-01 6.33897334e-02
-5.61899781e-01 -4.08676453e-02 9.47473466e-01 -4.05308813e-01
6.54063642e-01 -1.86536062e+00 3.13633353e-01 -2.03743577e-01
5.09487331e-01 4.07306910e-01 1.88782606e-02 -3.10866445e-01
-4.80267853e-01 -7.39128664e-02 -3.54886115e-01 -3.15768659e-01
-3.28801930e-01 6.10257566e-01 -4.26341474e-01 6.83766365e-01
1.12421781e-01 1.19428408e+00 -7.63024807e-01 -4.74671602e-01
3.02494615e-01 7.67609715e-01 -3.74091506e-01 -2.67537206e-01
-2.23349258e-01 4.46035117e-01 -2.39223257e-01 4.06078041e-01
6.97467864e-01 -3.60732049e-01 4.71991189e-02 -1.84087560e-01
-2.94850171e-01 2.70951182e-01 -8.64741206e-01 2.00910497e+00
-3.41095358e-01 9.71439302e-01 1.40668690e-01 -1.43057096e+00
4.08376426e-01 4.74813938e-01 4.18006212e-01 -8.86180818e-01
3.05753779e-02 1.26459897e-01 3.05432975e-02 -3.02694142e-01
2.63582498e-01 6.99090492e-03 1.34442717e-01 6.12989783e-01
1.18267119e-01 1.23688437e-01 1.57454029e-01 -4.31743748e-02
9.17434514e-01 4.04449292e-02 -1.77712902e-01 -2.79886752e-01
2.17699125e-01 -9.18463171e-02 4.40012127e-01 9.98813987e-01
-3.05427879e-01 1.12191784e+00 6.76731944e-01 -1.04468656e+00
-8.33710313e-01 -1.12006605e+00 -2.50347376e-01 1.22907817e+00
-2.03313902e-01 2.21108735e-01 -9.34036851e-01 -4.34613317e-01
-5.21490216e-01 3.28335345e-01 -8.76254201e-01 3.38535607e-01
-1.13687575e+00 -1.02958250e+00 6.24302864e-01 5.48572063e-01
5.21181107e-01 -1.46037745e+00 -1.38919842e+00 5.77681065e-01
1.05267623e-02 -1.14315104e+00 -3.71792823e-01 7.29146004e-01
-1.28760302e+00 -8.05168390e-01 -1.30045807e+00 -1.19779515e+00
8.93025994e-01 2.58696467e-01 1.04942143e+00 -1.50792211e-01
-4.83268142e-01 -5.38673028e-02 8.32232088e-02 -5.40494733e-03
1.54724211e-01 8.45377520e-02 -2.46118605e-01 -4.15002853e-01
1.71464548e-01 -7.52553999e-01 -9.89128411e-01 2.15781718e-01
-9.17037666e-01 3.19727093e-01 6.72978699e-01 9.13497329e-01
1.19410455e+00 -3.78622532e-01 1.88118353e-01 -1.04034328e+00
3.17161858e-01 -1.02238476e-01 -3.14346313e-01 3.52818727e-01
1.58596784e-01 -8.86927992e-02 5.26452899e-01 -6.20533645e-01
-7.17250645e-01 1.22477010e-01 -3.29590380e-01 -5.43001115e-01
-3.43255609e-01 2.60386229e-01 5.70065498e-01 -5.67992702e-02
3.07113796e-01 4.76883262e-01 -2.29840085e-01 -2.90213227e-01
2.12179229e-01 5.91255352e-02 6.39635801e-01 -3.02034050e-01
-1.11386701e-01 7.08985150e-01 -9.82157513e-02 -9.40234780e-01
-7.89299011e-01 -1.25621811e-01 -1.18010819e+00 -2.17892766e-01
1.34064698e+00 -6.54191613e-01 -7.26703703e-01 7.27457106e-01
-1.32954979e+00 -9.31751907e-01 -2.00071245e-01 4.37548369e-01
-6.15338981e-01 -1.97786931e-02 -1.24015117e+00 -2.56844372e-01
-4.91715521e-01 -1.48570538e+00 1.07258153e+00 1.86589152e-01
-4.73019108e-02 -1.05496454e+00 -1.74987599e-01 -4.82149534e-02
3.71224076e-01 4.17318016e-01 1.13664782e+00 -2.94401705e-01
-4.43459600e-01 1.18372872e-01 -4.82093573e-01 5.63517474e-02
-2.81646308e-02 -2.09397256e-01 -9.01417673e-01 -1.27186522e-01
1.87266365e-01 -2.24322051e-01 1.41636443e+00 1.34010887e+00
1.71491468e+00 -1.00251779e-01 -3.65996867e-01 6.77321792e-01
1.14283001e+00 2.94356588e-02 6.71157062e-01 2.16144785e-01
7.06693888e-01 3.55091691e-01 -7.95450713e-03 -1.22761630e-01
1.01961531e-01 4.00161058e-01 3.14329714e-01 -7.82428384e-01
-3.94926101e-01 2.41199479e-01 2.57752478e-01 9.36057806e-01
-1.62796065e-01 1.98791280e-01 -1.08358860e+00 7.01499343e-01
-1.67203569e+00 -8.61915886e-01 -2.21478686e-01 1.76992893e+00
8.19315612e-01 2.94364661e-01 1.28255775e-02 -2.04373822e-01
8.16710234e-01 4.80250686e-01 -9.13401008e-01 -6.38991714e-01
-3.61671925e-01 2.65945762e-01 7.04815984e-01 3.17755759e-01
-1.23011684e+00 1.03951240e+00 7.11566401e+00 7.49866843e-01
-1.49648690e+00 3.73678684e-01 1.15649402e+00 -5.37434697e-01
-3.04989666e-02 -4.85634387e-01 -7.25019515e-01 1.49625286e-01
9.33555245e-01 4.73242939e-01 2.57405251e-01 4.85329121e-01
2.27073714e-01 -3.61829191e-01 -1.04274058e+00 1.07883608e+00
-1.57751530e-01 -1.74141181e+00 2.65550632e-02 -8.99242684e-02
7.27958679e-01 9.21175063e-01 3.61085266e-01 1.15343727e-01
2.13180184e-01 -1.32202423e+00 6.95957661e-01 3.03676844e-01
7.08735228e-01 -6.00344479e-01 4.63139147e-01 2.55081445e-01
-9.19572890e-01 3.08523893e-01 -7.70039737e-01 7.32699484e-02
2.94863135e-01 6.09445512e-01 -5.80225706e-01 -3.14782895e-02
1.09397197e+00 6.36947930e-01 -4.15096670e-01 7.49733329e-01
3.11415959e-02 4.43301260e-01 -3.52365464e-01 1.43271759e-01
9.92588997e-01 -1.67512909e-01 3.42477083e-01 1.47362053e+00
4.91702184e-02 1.95342138e-01 1.70870684e-02 8.37094188e-01
-5.72666563e-02 -3.87273014e-01 -4.06537890e-01 1.44604191e-01
-2.86205381e-01 1.25162804e+00 -1.49602389e+00 -4.80522454e-01
-4.47766423e-01 1.13799977e+00 4.19203222e-01 6.81953967e-01
-7.51485467e-01 -1.93486825e-01 5.57498395e-01 -8.19725692e-02
6.95895493e-01 -4.25052196e-01 -3.91828954e-01 -8.50763381e-01
-1.44200817e-01 -5.91384351e-01 3.84861082e-01 -7.62476683e-01
-8.63612890e-01 1.02100086e+00 -2.68360049e-01 -8.32693338e-01
8.94541740e-02 -9.71912146e-01 -4.52688962e-01 6.31128669e-01
-1.57196796e+00 -1.03051221e+00 2.72981465e-01 7.34147847e-01
7.07122982e-01 3.11206907e-01 8.18190455e-01 1.31796956e-01
-7.02777684e-01 4.20384966e-02 -9.04422905e-03 3.60428780e-01
3.04601341e-01 -1.20150471e+00 7.90309846e-01 7.69405186e-01
3.11416000e-01 7.25908995e-01 4.38896090e-01 -4.92242038e-01
-1.17292440e+00 -1.12934446e+00 6.68009162e-01 4.02428359e-02
6.91026270e-01 -5.33616006e-01 -9.88452673e-01 1.04116642e+00
4.72154349e-01 6.46278203e-01 3.75171334e-01 -6.17610887e-02
-1.24693967e-01 3.61261606e-01 -1.01570606e+00 6.64969206e-01
1.11966491e+00 -6.38319671e-01 -7.32357383e-01 6.78675711e-01
7.84329891e-01 -1.03797138e+00 -8.30118954e-01 2.29214981e-01
4.43709284e-01 -1.15614486e+00 1.20089233e+00 -6.02142453e-01
3.15289885e-01 -1.61227286e-01 5.36143743e-02 -1.14413691e+00
-1.31357998e-01 -3.46780807e-01 1.34105116e-01 1.40364319e-01
4.35750991e-01 -6.54829443e-01 8.76185894e-01 4.47771907e-01
-4.95665520e-01 -1.14419448e+00 -1.26583755e+00 -3.68408203e-01
2.96987057e-01 -6.23439312e-01 1.19181745e-01 6.41287506e-01
-4.46998566e-01 -1.10110879e-01 8.31671357e-02 1.68587491e-01
4.67077971e-01 3.81056905e-01 -9.85493362e-02 -6.79481208e-01
2.75937766e-02 -6.92159295e-01 -2.39921302e-01 -1.43305051e+00
1.65145919e-01 -1.08054900e+00 -3.58313359e-02 -1.70767283e+00
1.81101501e-01 -1.60496175e-01 -5.98006308e-01 8.08870792e-01
1.72026768e-01 5.04494250e-01 -1.03946574e-01 -8.25830996e-02
-6.46786690e-01 7.48452023e-02 1.67329645e+00 -2.44461745e-01
-2.45723054e-01 9.23881028e-03 -8.43583047e-03 8.80761981e-01
6.22880995e-01 -5.52430511e-01 -2.19623432e-01 -9.10526812e-01
9.67303887e-02 1.84591338e-01 7.34090626e-01 -7.75542080e-01
6.80189669e-01 1.90998420e-01 7.62054920e-01 -1.11000431e+00
2.28958771e-01 -2.81328529e-01 -1.49338007e-01 6.46728992e-01
-4.63182211e-01 1.79449022e-01 4.23804313e-01 2.11131558e-01
-2.95148522e-01 4.72756512e-02 7.68438816e-01 -6.74132705e-01
-8.64095509e-01 6.13425136e-01 -9.47433770e-01 3.09876911e-02
6.73692584e-01 -5.08358061e-01 -1.75069228e-01 1.34278983e-01
-1.13721621e+00 -2.98046526e-02 5.31381229e-03 1.99380279e-01
8.88045192e-01 -9.68694031e-01 -4.19011891e-01 2.58427858e-01
-7.05569088e-01 5.07763863e-01 7.69974887e-01 1.23914146e+00
-7.53269732e-01 7.45932579e-01 -4.47499365e-01 -1.11999750e+00
-1.06045055e+00 2.69890964e-01 7.86500394e-01 -2.38654003e-01
-1.36818302e+00 1.34389329e+00 5.53500831e-01 -4.93789673e-01
2.53174692e-01 -1.03923905e+00 -1.43585548e-01 1.90570116e-01
7.58707702e-01 -9.27784964e-02 4.10957754e-01 -5.28725445e-01
-4.93363172e-01 6.38905406e-01 -2.73777813e-01 -5.26535399e-02
1.62926590e+00 8.73247220e-04 -4.83472109e-01 6.77067637e-01
1.33486903e+00 -7.71959484e-01 -1.58350670e+00 -3.32308382e-01
-1.18906586e-03 4.76628616e-02 5.50919712e-01 -3.55307758e-01
-1.54746246e+00 1.24449825e+00 6.48163617e-01 7.54761249e-02
1.03974748e+00 -5.48872864e-03 1.22584724e+00 4.77845252e-01
2.83411056e-01 -1.05898952e+00 -1.17468588e-01 7.68761635e-01
7.02618182e-01 -9.26589787e-01 -2.21152022e-01 1.45119861e-01
-5.99423289e-01 1.47932732e+00 3.40908349e-01 -4.59982693e-01
6.93875253e-01 4.30031270e-01 2.98210457e-02 -4.33396250e-01
-7.32161582e-01 -1.20775640e-01 2.49953359e-01 3.95873576e-01
5.26226699e-01 5.89810349e-02 4.24376726e-01 1.44911692e-01
-1.19571080e-02 -1.13464564e-01 4.87093955e-01 7.75889516e-01
-2.95248717e-01 -6.74700022e-01 -1.60707906e-02 5.51815629e-01
-5.93148053e-01 -2.19127312e-01 1.26667827e-01 5.74333549e-01
2.97493320e-02 3.51568103e-01 5.49883723e-01 1.52384549e-01
2.59502307e-02 -1.24678329e-01 6.61719799e-01 -5.47429085e-01
-7.24435568e-01 2.68078089e-01 -2.93985784e-01 -8.41808498e-01
-6.06985569e-01 -7.20350444e-01 -1.66458404e+00 -8.66381451e-02
2.55228937e-01 -4.24876183e-01 5.48695862e-01 1.07454157e+00
1.31782234e-01 8.85974884e-01 5.46306111e-02 -1.26588809e+00
1.63939342e-01 -6.36210501e-01 -5.56157053e-01 -9.99393389e-02
7.03282773e-01 -2.87373662e-01 6.53124899e-02 3.14952917e-02] | [14.44774341583252, -2.5688815116882324] |
7d098c14-4850-42ba-9459-37665524016d | agile3d-attention-guided-interactive-multi | 2306.00977 | null | https://arxiv.org/abs/2306.00977v1 | https://arxiv.org/pdf/2306.00977v1.pdf | AGILE3D: Attention Guided Interactive Multi-object 3D Segmentation | During interactive segmentation, a model and a user work together to delineate objects of interest in a 3D point cloud. In an iterative process, the model assigns each data point to an object (or the background), while the user corrects errors in the resulting segmentation and feeds them back into the model. From a machine learning perspective the goal is to design the model and the feedback mechanism in a way that minimizes the required user input. The current best practice segments objects one at a time, and asks the user to provide positive clicks to indicate regions wrongly assigned to the background and negative clicks to indicate regions wrongly assigned to the object (foreground). Sequentially visiting objects is wasteful, since it disregards synergies between objects: a positive click for a given object can, by definition, serve as a negative click for nearby objects, moreover a direct competition between adjacent objects can speed up the identification of their common boundary. We introduce AGILE3D, an efficient, attention-based model that (1) supports simultaneous segmentation of multiple 3D objects, (2) yields more accurate segmentation masks with fewer user clicks, and (3) offers faster inference. We encode the point cloud into a latent feature representation, and view user clicks as queries and employ cross-attention to represent contextual relations between different click locations as well as between clicks and the 3D point cloud features. Every time new clicks are added, we only need to run a lightweight decoder that produces updated segmentation masks. In experiments with four different point cloud datasets, AGILE3D sets a new state of the art, moreover, we also verify its practicality in real-world setups with a real user study. | ['Theodora Kontogianni', 'Konrad Schindler', 'Bastian Leibe', 'Francis Engelmann', 'Jonas Schult', 'Sabarinath Mahadevan', 'Yuanwen Yue'] | 2023-06-01 | null | null | null | null | ['interactive-segmentation'] | ['computer-vision'] | [ 1.67203307e-01 -7.06588924e-02 1.71562843e-02 -2.65510291e-01
-6.79224789e-01 -7.48722136e-01 1.12881690e-01 4.22855824e-01
-4.80602711e-01 1.02578782e-01 -3.50880504e-01 -4.22385186e-01
1.91114232e-01 -7.52337158e-01 -8.41662765e-01 -5.63447356e-01
9.54355896e-02 1.01280284e+00 9.34872806e-01 2.67434776e-01
4.14228231e-01 8.60303998e-01 -1.51958215e+00 1.44167721e-01
8.35593641e-01 1.00535011e+00 7.35364974e-01 7.12508380e-01
-4.39445227e-01 -1.31561741e-01 -7.02947319e-01 -2.06980124e-01
4.38953578e-01 1.45311773e-01 -6.75139785e-01 4.43764716e-01
5.87125897e-01 -2.98770458e-01 1.72828197e-01 8.85601819e-01
3.12016517e-01 8.20581242e-02 5.49560130e-01 -1.11324847e+00
-4.89958465e-01 1.38950035e-01 -1.01440585e+00 1.37284681e-01
1.77375898e-01 3.47269863e-01 1.13531470e+00 -1.00192070e+00
5.50119877e-01 1.04274487e+00 3.89170587e-01 3.86259735e-01
-1.44277596e+00 -5.22557378e-01 7.31730938e-01 -1.55701816e-01
-1.27669752e+00 4.50312421e-02 6.66854024e-01 -4.69587713e-01
8.14727426e-01 6.18564606e-01 9.50016081e-01 5.12714446e-01
-1.34374887e-01 1.03629148e+00 5.43066323e-01 -1.76266521e-01
3.36298078e-01 1.71595767e-01 2.64254093e-01 6.31685913e-01
-1.26418889e-01 -1.33466661e-01 -4.70777154e-01 -1.45073980e-01
9.09748852e-01 1.99794501e-01 -2.69171387e-01 -7.19728410e-01
-1.10015619e+00 6.09136343e-01 6.37385249e-01 1.22721292e-01
-5.08447409e-01 1.82023481e-01 -1.63696289e-01 6.57124026e-03
4.02547449e-01 2.45933890e-01 -5.47826529e-01 -1.14596086e-02
-9.61238325e-01 4.23473060e-01 5.78381896e-01 1.10134041e+00
8.94926190e-01 -7.08569229e-01 -2.44522676e-01 6.23560369e-01
4.14592266e-01 4.98156846e-01 -1.37079388e-01 -7.73487687e-01
5.24713397e-01 8.94177377e-01 3.59952748e-01 -1.01165271e+00
-3.04996014e-01 -4.56614166e-01 -3.27592403e-01 5.07025123e-01
3.11371207e-01 1.96081564e-01 -1.22979701e+00 1.52015364e+00
7.35447228e-01 1.95046186e-01 -6.44723237e-01 1.21304166e+00
4.46813583e-01 5.77690125e-01 -2.25611031e-02 5.44792376e-02
1.39096308e+00 -7.71012247e-01 -2.44329467e-01 -3.66822541e-01
4.35341597e-01 -9.23047602e-01 1.34095073e+00 3.80243599e-01
-1.21435535e+00 -6.71381950e-01 -6.81917250e-01 -1.68305904e-01
-2.53008276e-01 9.03173015e-02 4.60847735e-01 2.33299881e-01
-7.75079072e-01 5.02620876e-01 -1.06663585e+00 -1.22558065e-01
4.63437140e-01 4.89498973e-01 6.52885586e-02 2.86051244e-01
-6.65723741e-01 6.52676940e-01 7.24529400e-02 1.96289018e-01
-7.30156839e-01 -7.18191564e-01 -5.34688830e-01 1.77694827e-01
8.57722282e-01 -6.56656384e-01 1.37708294e+00 -7.48463273e-01
-1.12828970e+00 8.78889203e-01 -4.68303770e-01 -1.99037865e-01
6.50960803e-01 -5.18171728e-01 1.24477684e-01 9.24493074e-02
4.03567284e-01 1.00495601e+00 7.97520995e-01 -1.57275105e+00
-1.07976818e+00 -5.99155009e-01 1.57644734e-01 3.35148931e-01
2.10739955e-01 -1.85474589e-01 -1.44398522e+00 -3.83813232e-01
7.50624359e-01 -1.05382907e+00 -3.21205795e-01 3.18199694e-01
-6.97074473e-01 -3.05977285e-01 1.00483656e+00 -2.33823642e-01
1.21808791e+00 -2.27993059e+00 1.61138475e-01 5.46589851e-01
3.76748770e-01 1.10332854e-01 9.91363749e-02 1.25438198e-02
5.31896316e-02 2.23707065e-01 -3.61550182e-01 -5.64805865e-01
1.17937950e-02 2.70244390e-01 -1.87625512e-01 2.85699308e-01
2.35925496e-01 8.31821859e-01 -8.10757041e-01 -6.30773306e-01
3.98638427e-01 1.77345559e-01 -8.70092869e-01 1.83395013e-01
-5.46545625e-01 2.20465943e-01 -6.38154387e-01 5.93574286e-01
8.49382877e-01 -5.31656682e-01 -2.78126925e-01 -1.59756154e-01
-3.68044555e-01 1.95818573e-01 -1.49956834e+00 1.57598591e+00
-3.71066928e-01 3.87966514e-01 1.39783025e-01 -5.06040335e-01
7.25823224e-01 -8.72679614e-03 2.73996890e-01 -4.76094276e-01
-1.13483883e-01 1.42836064e-01 -2.68566161e-01 -5.16600072e-01
3.15411270e-01 3.38109136e-01 1.81456983e-01 5.00258744e-01
-2.85840362e-01 -1.25634208e-01 1.44381216e-02 3.11564207e-01
9.50030506e-01 1.66403279e-01 -1.86106503e-01 2.27956727e-01
2.90450662e-01 -1.42926753e-01 3.90066445e-01 1.02815509e+00
-1.98533642e-03 7.85356283e-01 6.36888206e-01 -2.21145436e-01
-7.53261626e-01 -1.18599474e+00 1.15725277e-02 1.12250054e+00
7.32230484e-01 -1.78590015e-01 -6.00724459e-01 -9.97069180e-01
2.65555054e-01 7.85696089e-01 -6.87463105e-01 1.44956186e-01
-5.55925965e-01 -1.89846307e-01 -1.94427624e-01 4.71808583e-01
2.78957784e-01 -9.99521494e-01 -1.10144150e+00 1.84549078e-01
-4.42390665e-02 -7.23708212e-01 -8.85643363e-01 3.93120259e-01
-9.14154589e-01 -1.18545771e+00 -6.86909556e-01 -7.44974375e-01
1.02419519e+00 5.01371145e-01 1.12867272e+00 3.13069373e-01
-2.37257466e-01 2.32893050e-01 -2.59206206e-01 -1.77328780e-01
-6.67126700e-02 1.42197549e-01 -4.25441295e-01 1.13260269e-01
3.58529925e-01 -3.01968306e-01 -6.99279308e-01 7.53243029e-01
-7.58294642e-01 2.72253782e-01 5.68787634e-01 4.83617902e-01
9.72097695e-01 -1.44898832e-01 1.34489741e-02 -1.03494895e+00
2.26204932e-01 -2.66877115e-01 -1.00932133e+00 1.04047500e-01
-1.73315197e-01 -1.10463295e-02 4.90917899e-02 -3.31871033e-01
-7.24142313e-01 3.05350482e-01 -1.19786859e-01 -7.89315879e-01
-2.79776514e-01 2.34757319e-01 -2.36010373e-01 1.72615781e-01
5.07717073e-01 1.41203599e-02 -3.89560342e-01 -8.08668554e-01
4.22294438e-01 5.53250074e-01 4.34330076e-01 -3.40655357e-01
7.59225965e-01 4.63793874e-01 -2.66312808e-01 -5.96861005e-01
-6.12836361e-01 -6.99379504e-01 -9.23224688e-01 -2.61458546e-01
1.01322889e+00 -5.04547894e-01 -9.74227190e-01 1.32756174e-01
-1.40418315e+00 -2.20435157e-01 -1.07709005e-01 1.35351375e-01
-2.38226220e-01 1.10111721e-01 -8.13633427e-02 -8.21288705e-01
1.37795836e-01 -1.25961292e+00 1.39015067e+00 2.26956502e-01
-2.60969132e-01 -5.52105904e-01 -3.14526707e-01 2.12007478e-01
-7.85419121e-02 2.33176053e-02 9.82586861e-01 -6.68167531e-01
-1.11486149e+00 -3.97906184e-01 -3.72631431e-01 -4.89207432e-02
2.13372275e-01 1.98121801e-01 -7.82725751e-01 2.31979229e-02
-2.32698217e-01 2.81926751e-01 6.59289181e-01 4.59329218e-01
1.52616560e+00 -2.50943881e-02 -7.44281590e-01 4.16021615e-01
1.13436294e+00 3.49024147e-01 2.66025275e-01 1.63281914e-02
7.12626636e-01 3.99424881e-01 6.33207083e-01 2.36037940e-01
1.98509902e-01 8.86631608e-01 7.40988374e-01 -4.12360787e-01
9.67363641e-02 -2.50549376e-01 -2.84909070e-01 1.82560399e-01
5.03931157e-02 -2.91265786e-01 -7.88746476e-01 4.99345601e-01
-1.89015305e+00 -5.85547090e-01 -3.10855478e-01 2.48701572e+00
6.75600290e-01 6.33544862e-01 1.35114431e-01 -1.00434527e-01
7.08648801e-01 -1.05473898e-01 -8.73144448e-01 1.17510132e-01
2.69473344e-01 1.15841664e-01 5.14760911e-01 6.56750143e-01
-1.02000988e+00 1.08162272e+00 5.40026236e+00 6.41064405e-01
-1.21651602e+00 -5.83333187e-02 6.78035915e-01 -4.15218234e-01
-4.61004227e-01 7.53990784e-02 -8.72761011e-01 5.18798590e-01
8.64012539e-02 2.92893946e-01 2.86477745e-01 9.33583081e-01
4.60108280e-01 -4.25922394e-01 -1.39712703e+00 9.81610119e-01
-2.37504691e-01 -1.18367326e+00 1.44383302e-02 6.13310374e-02
2.87978977e-01 9.76741388e-02 5.11272065e-02 -2.37147976e-02
9.72747803e-02 -6.23871505e-01 1.04084480e+00 4.06073719e-01
2.82092512e-01 -6.96812570e-01 3.77744883e-01 4.77260053e-01
-9.88323808e-01 1.54388368e-01 -7.72610605e-02 2.41201356e-01
3.07870507e-01 4.79211837e-01 -1.11188579e+00 1.49615735e-01
8.65260124e-01 2.08695516e-01 -4.83320981e-01 1.34717536e+00
-3.19227368e-01 4.22773749e-01 -7.01384544e-01 -1.92093849e-01
4.02603954e-01 -1.14788212e-01 8.24191451e-01 1.04313505e+00
7.38708824e-02 5.89228943e-02 3.54866743e-01 1.36443865e+00
1.81090608e-01 -7.69609883e-02 -1.55317694e-01 2.83429563e-01
6.03702545e-01 1.03021753e+00 -1.23429036e+00 -3.52691829e-01
-2.65942097e-01 1.17269492e+00 2.30252460e-01 5.92228711e-01
-1.00670516e+00 -3.61729831e-01 6.12122893e-01 4.51991022e-01
7.40953028e-01 -3.27095091e-01 -5.98797619e-01 -7.76488841e-01
1.41101778e-01 -3.66473496e-01 2.24579796e-01 -9.55541193e-01
-9.24472213e-01 3.51118118e-01 -1.92217469e-01 -1.15494454e+00
2.10667238e-01 -3.18224877e-01 -4.98347491e-01 1.11390913e+00
-1.16593182e+00 -6.42989933e-01 -3.32590669e-01 4.40179616e-01
7.58780897e-01 6.14342511e-01 3.31992805e-01 4.28818017e-01
-3.61620039e-01 2.86865562e-01 -4.68820453e-01 4.20267247e-02
3.68924290e-01 -1.45478785e+00 6.37352467e-01 6.12642825e-01
4.88370001e-01 6.20120645e-01 6.72635794e-01 -8.66325319e-01
-9.40912366e-01 -8.27868283e-01 8.68734956e-01 -4.88765150e-01
1.02482088e-01 -5.77119172e-01 -1.23197889e+00 5.73622108e-01
-3.06665719e-01 -1.76758140e-01 4.71163958e-01 2.33577058e-01
1.13014996e-01 5.78131452e-02 -9.45786119e-01 8.86012614e-01
9.84405160e-01 -2.34239995e-01 -3.68076742e-01 3.01429719e-01
8.78686428e-01 -7.63347566e-01 -1.63135633e-01 1.38856649e-01
3.58333886e-01 -1.11781883e+00 8.85357320e-01 -4.98542130e-01
9.26442817e-02 -8.18949521e-01 4.01690863e-02 -1.08026016e+00
-3.69572341e-01 -5.05041242e-01 -1.12835728e-01 8.62398028e-01
6.70951426e-01 -3.30561697e-01 1.02753413e+00 9.32721496e-01
-2.96226114e-01 -1.04595339e+00 -9.60783064e-01 -3.68469268e-01
-5.36293566e-01 -7.17201829e-01 7.14138329e-01 5.20285547e-01
-4.76223618e-01 1.91311732e-01 1.72103997e-02 7.60504544e-01
4.24201399e-01 4.49879408e-01 9.34645355e-01 -1.21641052e+00
-2.52778530e-01 -5.71392298e-01 -1.18182935e-01 -1.88580728e+00
-5.31616151e-01 -6.78059638e-01 1.70588031e-01 -1.52382147e+00
-1.62558839e-01 -9.54309702e-01 -6.85545653e-02 6.25928760e-01
-3.25492591e-01 2.11953700e-01 3.64350885e-01 1.65393487e-01
-6.08636081e-01 1.02396198e-01 1.28768551e+00 -1.61581952e-02
-7.02549934e-01 6.57076895e-01 -5.46413839e-01 8.98631811e-01
2.92205662e-01 -5.16460001e-01 -2.54852265e-01 -6.53820753e-01
2.70896137e-01 1.41748622e-01 5.99947751e-01 -7.16564715e-01
4.05913770e-01 -1.07394621e-01 5.89227617e-01 -1.22223020e+00
3.92617434e-01 -1.19918466e+00 -1.12404395e-02 2.97655374e-01
-2.48863310e-01 -1.48643553e-01 1.86842874e-01 5.25186300e-01
1.72535956e-01 -4.14413005e-01 6.56036615e-01 -2.26392522e-01
-4.81417358e-01 5.89227736e-01 -1.21724337e-01 -1.99025780e-01
1.07086325e+00 -6.65672719e-01 3.76349390e-01 -4.13739495e-02
-1.20251477e+00 6.89613283e-01 5.42366087e-01 4.75622356e-01
5.71960211e-01 -1.02907073e+00 -2.04109862e-01 5.29143929e-01
-4.05186825e-02 6.26292646e-01 2.33636096e-01 6.84537292e-01
-3.79466981e-01 2.15317309e-01 4.49468821e-01 -1.17627704e+00
-1.48441362e+00 4.77454036e-01 3.02995861e-01 1.71703711e-01
-7.02825487e-01 1.21223247e+00 4.91676241e-01 -1.46219805e-01
5.44748068e-01 -7.15218723e-01 1.25446599e-02 1.11772738e-01
2.77542651e-01 1.60964340e-01 1.05485916e-01 -2.60885209e-01
-4.05606478e-01 7.35129714e-01 -2.82751292e-01 -1.75473377e-01
1.07800412e+00 -2.01488674e-01 1.66436732e-01 6.02175832e-01
9.45435643e-01 7.55416304e-02 -1.65610206e+00 -2.51527816e-01
-6.05367050e-02 -1.04618311e+00 -1.35037908e-02 -9.68489885e-01
-1.03793144e+00 7.80303061e-01 5.29785097e-01 5.94844997e-01
9.28422093e-01 4.69379514e-01 6.50052369e-01 9.31466147e-02
2.10707217e-01 -1.07850301e+00 1.64690152e-01 3.08686495e-01
7.70231485e-01 -1.00302076e+00 -2.20452413e-01 -6.50519848e-01
-7.19551682e-01 8.93675327e-01 8.41933250e-01 5.75798936e-02
5.77033699e-01 1.49835810e-01 3.30521539e-02 -3.66443932e-01
-6.95774078e-01 -1.89097300e-01 2.07780436e-01 3.25498164e-01
9.38428380e-03 -8.97120908e-02 1.13361314e-01 4.13452119e-01
-1.37712419e-01 -1.56523898e-01 -5.60969152e-02 7.71284938e-01
-7.57488191e-01 -8.76034856e-01 -4.51516032e-01 4.41721678e-01
-7.47133493e-02 6.44114092e-02 -3.22401404e-01 7.37785518e-01
3.82150769e-01 5.69166481e-01 6.60656512e-01 -5.06658182e-02
6.09359741e-01 -1.10026538e-01 1.20977260e-01 -8.14697027e-01
-5.50900102e-01 5.35608292e-01 -3.51639628e-01 -5.82078993e-01
-6.59543439e-04 -7.01954424e-01 -1.58677232e+00 8.94599408e-02
-6.21415675e-01 1.54518321e-01 7.31948912e-01 8.09544027e-01
6.51559412e-01 5.92352927e-01 6.48678064e-01 -9.96754825e-01
-2.85962410e-02 -5.87820649e-01 -2.63178587e-01 3.33886892e-01
3.73374760e-01 -7.75573671e-01 -3.34902287e-01 1.81040075e-02] | [9.347145080566406, -0.2928410768508911] |
7d58862f-bfe2-4770-90d8-09bc097bb4b2 | knowledge-hypergraph-embedding-meets | 2102.09557 | null | https://arxiv.org/abs/2102.09557v1 | https://arxiv.org/pdf/2102.09557v1.pdf | Knowledge Hypergraph Embedding Meets Relational Algebra | Embedding-based methods for reasoning in knowledge hypergraphs learn a representation for each entity and relation. Current methods do not capture the procedural rules underlying the relations in the graph. We propose a simple embedding-based model called ReAlE that performs link prediction in knowledge hypergraphs (generalized knowledge graphs) and can represent high-level abstractions in terms of relational algebra operations. We show theoretically that ReAlE is fully expressive and provide proofs and empirical evidence that it can represent a large subset of the primitive relational algebra operations, namely renaming, projection, set union, selection, and set difference. We also verify experimentally that ReAlE outperforms state-of-the-art models in knowledge hypergraph completion, and in representing each of these primitive relational algebra operations. For the latter experiment, we generate a synthetic knowledge hypergraph, for which we design an algorithm based on the Erdos-R'enyi model for generating random graphs. | ['David Poole', 'David Vazquez', 'Perouz Taslakian', 'Bahare Fatemi'] | 2021-02-18 | null | null | null | null | ['hypergraph-embedding'] | ['graphs'] | [-1.48726046e-01 1.00061333e+00 -4.69407856e-01 2.56120833e-03
-3.96907032e-02 -7.19201028e-01 8.67545605e-01 5.42149365e-01
6.08605109e-02 6.72314167e-01 3.84770542e-01 -5.50061524e-01
-8.03875327e-01 -1.77983761e+00 -8.99617672e-01 -3.07645708e-01
-7.63925672e-01 1.32078850e+00 4.08906519e-01 -4.70874339e-01
-3.34493607e-01 7.16892540e-01 -1.47136009e+00 2.09472656e-01
4.96931791e-01 2.69755036e-01 -5.57150781e-01 7.51692712e-01
-3.13713908e-01 1.16621614e+00 -4.58038688e-01 -1.22765326e+00
1.55830428e-01 -6.26536682e-02 -1.25585473e+00 -1.94193140e-01
3.75035614e-01 -1.98442444e-01 -1.35219038e+00 9.32045460e-01
1.23748347e-01 2.28705108e-01 8.54831815e-01 -1.65553057e+00
-1.08135355e+00 1.47957551e+00 -1.20119065e-01 3.63832316e-03
7.73975492e-01 -1.85136855e-01 1.60789108e+00 -6.01830900e-01
1.08479869e+00 1.39554620e+00 7.41524339e-01 3.78311783e-01
-1.56503737e+00 -2.37814173e-01 -3.16202700e-01 6.28148377e-01
-1.70823884e+00 -7.60549903e-02 3.45593333e-01 -3.65991175e-01
1.12592673e+00 3.19913119e-01 8.76564920e-01 7.86170900e-01
-3.29196975e-02 5.48623383e-01 6.45352840e-01 -5.57543993e-01
1.23808481e-01 2.07148522e-01 5.89893818e-01 1.22457457e+00
1.09363627e+00 3.54429156e-01 -3.93143535e-01 -5.83358288e-01
8.25708628e-01 -1.50589481e-01 -2.63473064e-01 -1.08046401e+00
-1.02998543e+00 8.90634656e-01 6.65508330e-01 -1.15620963e-01
-2.68600762e-01 5.62538922e-01 4.62750345e-01 4.92614180e-01
-2.04737961e-01 3.97324681e-01 -3.33997548e-01 3.95292997e-01
-2.43328959e-01 3.73246729e-01 1.56739604e+00 1.46974134e+00
7.97467291e-01 -1.18758783e-01 -2.18048841e-01 2.40732044e-01
2.52249479e-01 3.17647755e-01 -1.96586445e-01 -7.64438331e-01
3.31495076e-01 9.01977897e-01 -1.27120882e-01 -9.86642182e-01
-2.16091529e-01 -1.07364781e-01 -5.61908901e-01 -2.33428739e-02
2.00319886e-01 3.05185527e-01 -9.34850931e-01 1.42193460e+00
2.68046290e-01 4.19982493e-01 5.68878233e-01 2.67302126e-01
1.16386986e+00 3.75972420e-01 5.79309575e-02 9.97312590e-02
1.50343895e+00 -6.28081918e-01 -5.21419644e-01 2.85231650e-01
8.99330974e-01 1.08300433e-01 6.22129321e-01 8.07763040e-02
-1.14788687e+00 -1.41993552e-01 -9.58196163e-01 -1.09249629e-01
-8.75373423e-01 -5.51857829e-01 1.30975962e+00 8.94458950e-01
-1.14927030e+00 4.87017900e-01 -6.54567003e-01 -2.53183931e-01
4.40175712e-01 3.61005068e-01 -8.92279029e-01 -3.57373476e-01
-1.59266126e+00 1.03264499e+00 1.11255538e+00 -2.49078378e-01
-1.15953553e+00 -7.35437214e-01 -1.30824327e+00 5.88940680e-01
7.44432509e-01 -1.14315844e+00 8.47862303e-01 9.06501263e-02
-9.74920630e-01 9.05941844e-01 2.87378073e-01 -7.78756440e-01
-1.97386537e-02 2.27214068e-01 -6.19566500e-01 2.06403568e-01
-4.05009687e-01 1.86938524e-01 4.22433436e-01 -1.58274746e+00
-3.11097682e-01 -3.98765326e-01 8.44082415e-01 4.87254672e-02
-2.71695465e-01 -3.53897125e-01 -5.08347690e-01 -1.86725780e-01
-7.69914985e-02 -9.32842970e-01 -2.07146823e-01 -3.76690149e-01
-8.17576528e-01 -2.07734138e-01 3.61090064e-01 -2.45157450e-01
1.36125350e+00 -1.82244778e+00 3.76235455e-01 9.10803616e-01
8.75277758e-01 9.32153463e-02 -3.26211527e-02 9.55432355e-01
-2.56707400e-01 4.07014489e-01 8.14353079e-02 2.40966380e-01
4.37316120e-01 7.90435314e-01 -3.39479089e-01 1.90464005e-01
-2.01000065e-01 1.26466012e+00 -1.10579753e+00 -6.54427707e-01
1.11622863e-01 3.60602885e-01 -6.95917130e-01 -2.91826744e-02
-4.75981921e-01 -6.49717391e-01 -3.48104566e-01 4.03385192e-01
4.41204935e-01 -4.55225140e-01 9.52193379e-01 -2.62638658e-01
8.13944638e-01 1.40933976e-01 -1.56751144e+00 1.30572355e+00
-2.49927536e-01 3.95176560e-01 -4.09003526e-01 -5.99275887e-01
6.87085271e-01 3.60863239e-01 1.53949499e-01 -1.11229599e-01
-5.52441366e-02 -3.28940123e-01 -1.43678738e-02 -3.33980381e-01
7.80563295e-01 2.07237191e-02 -9.79912952e-02 5.68486631e-01
3.82867843e-01 -2.26951167e-01 4.18564677e-01 1.09671140e+00
1.70294929e+00 -2.19025850e-01 4.85045284e-01 -3.76557335e-02
3.00637633e-01 1.15674928e-01 1.25345081e-01 9.96022463e-01
4.29845005e-01 -1.41290044e-02 9.53316152e-01 -5.40684700e-01
-1.02518666e+00 -1.54131472e+00 1.24272451e-01 9.15150523e-01
1.84061691e-01 -1.23986840e+00 -3.58868003e-01 -7.92241931e-01
5.58546782e-01 1.02635360e+00 -8.49069417e-01 -4.76304293e-01
-1.65549949e-01 -4.68393177e-01 8.91896904e-01 5.95797300e-01
-2.22017374e-02 -9.67883706e-01 8.52869451e-02 5.71000092e-02
2.34605417e-01 -1.14122427e+00 7.16530606e-02 8.49456713e-02
-6.71900868e-01 -1.78002393e+00 2.09280163e-01 -7.25709379e-01
7.01790690e-01 -4.85885609e-03 1.50980389e+00 2.77089238e-01
-2.70822495e-01 8.43765199e-01 -3.20693851e-01 -2.88815051e-01
-6.96540594e-01 1.40851110e-01 -3.56658325e-02 -5.17315328e-01
4.28407103e-01 -8.08957636e-01 -3.79390754e-02 5.18590063e-02
-1.17641938e+00 -3.17079067e-01 4.29583997e-01 7.94037580e-01
5.67397475e-01 7.73494661e-01 8.54260623e-02 -1.68166625e+00
8.82771254e-01 -5.55421650e-01 -6.82829738e-01 7.39582658e-01
-7.59231925e-01 5.72101533e-01 5.16976297e-01 -1.22314341e-01
-8.88023615e-01 -1.82916790e-01 4.77420509e-01 -5.51247656e-01
1.43505707e-01 9.71170545e-01 -2.59279072e-01 -1.57878727e-01
9.99773085e-01 1.21841259e-01 -2.52676040e-01 -1.04399867e-01
1.12666094e+00 3.74513529e-02 6.74476683e-01 -9.83223796e-01
1.32545900e+00 4.64594275e-01 5.31815648e-01 -5.73744416e-01
-6.85205162e-01 -1.65833488e-01 -5.73831439e-01 3.05707753e-01
4.69711125e-01 -8.01815391e-01 -1.26848495e+00 -2.63452798e-01
-1.00540566e+00 -2.58306235e-01 -8.21887434e-01 3.06553960e-01
-6.97560966e-01 4.32559133e-01 -7.38953948e-01 -4.37645286e-01
-1.22044504e-01 -6.52185738e-01 5.48168540e-01 -6.30510375e-02
-6.36740848e-02 -1.31917846e+00 3.24617952e-01 1.47655040e-01
2.59827524e-02 1.76757559e-01 1.68709326e+00 -1.09125173e+00
-1.12586749e+00 -2.65119284e-01 -2.48761192e-01 -1.42352030e-01
-3.41931015e-01 1.97427750e-01 -4.35641736e-01 -8.54733810e-02
-8.89927924e-01 -3.05703700e-01 6.72390103e-01 -2.01591268e-01
1.04604590e+00 -4.41624641e-01 -5.69897175e-01 6.80890560e-01
1.75757957e+00 -2.90782958e-01 1.24698925e+00 1.52685642e-01
8.89861703e-01 1.91537902e-01 5.52476093e-04 2.16859221e-01
7.90205181e-01 5.00123560e-01 3.50791603e-01 4.53553677e-01
-1.90033302e-01 -7.36986279e-01 1.72693711e-02 5.84687829e-01
-4.38244283e-01 -2.61956155e-01 -1.07626700e+00 6.81431711e-01
-1.84459448e+00 -1.14774477e+00 -2.11524963e-01 2.14244580e+00
9.50756133e-01 1.98381156e-01 -4.85570543e-02 4.11007181e-03
6.03725910e-01 -2.09308439e-03 -2.50746612e-03 -4.84393537e-01
-1.66628331e-01 7.09715486e-01 8.21346819e-01 7.41953850e-01
-6.29905701e-01 1.19700158e+00 7.09446001e+00 4.83688444e-01
9.07374322e-02 -1.14412859e-01 -3.48464400e-01 4.98595327e-01
-9.28899705e-01 4.05924767e-01 -6.69387460e-01 -2.43631572e-01
9.98413801e-01 -8.43359888e-01 9.59577441e-01 9.08882797e-01
-1.01540017e+00 3.39530110e-01 -1.55421436e+00 7.02611804e-01
-1.15954429e-02 -1.73762941e+00 6.01104677e-01 1.45874336e-01
7.97859907e-01 -3.54728788e-01 -3.97518069e-01 7.92974651e-01
1.35653150e+00 -1.27710891e+00 1.44286826e-01 6.50202751e-01
7.22498596e-01 -8.50612283e-01 6.98058426e-01 -1.48326859e-01
-1.31511569e+00 -1.56878844e-01 -4.61845040e-01 4.57629949e-01
-1.10211395e-01 2.36223400e-01 -1.09873915e+00 1.23883259e+00
1.07573919e-01 2.71398783e-01 -5.71500897e-01 9.20198977e-01
-7.79750347e-01 4.95222688e-01 -2.24642396e-01 2.28464350e-01
-2.48163387e-01 -1.62664324e-01 4.79099482e-01 1.03215361e+00
-6.35225922e-02 3.70309919e-01 3.49760689e-02 8.53716791e-01
-5.69531620e-01 -3.33445221e-01 -1.05404174e+00 -4.01787311e-01
9.46601629e-01 9.34887290e-01 -3.75670314e-01 -5.43624818e-01
-4.15830314e-01 5.76519728e-01 5.38948596e-01 5.14080822e-01
-7.39508748e-01 -6.56743228e-01 5.89813590e-01 9.87201780e-02
5.70627689e-01 -1.36821374e-01 1.33462146e-01 -1.19631732e+00
-2.03051329e-01 -7.98717022e-01 9.67235267e-01 -8.65110517e-01
-1.28453135e+00 2.90980309e-01 4.08071935e-01 -3.97780210e-01
-1.72891736e-01 -7.38745809e-01 -4.08977479e-01 6.54738009e-01
-1.13545358e+00 -1.20473158e+00 -3.15934420e-01 9.41620111e-01
-5.41129410e-01 -3.44766319e-01 1.37049270e+00 1.77418962e-02
-2.89753705e-01 6.30190015e-01 -8.49030539e-02 3.69090915e-01
1.48008205e-02 -1.72455955e+00 5.24398446e-01 4.69774991e-01
6.23875201e-01 8.78995776e-01 7.79202998e-01 -6.93716049e-01
-2.00990582e+00 -1.07174122e+00 8.11220229e-01 -8.12157393e-01
1.01883423e+00 -1.91368684e-01 -9.46941257e-01 1.63930357e+00
-2.25050017e-01 2.57332891e-01 5.91038465e-01 7.72530019e-01
-9.35426116e-01 -1.86599895e-01 -1.00739062e+00 7.95010984e-01
1.42394269e+00 -8.63109291e-01 -1.11816835e+00 6.04424894e-01
9.49972272e-01 -3.64834905e-01 -1.51634562e+00 1.52943805e-01
3.91528666e-01 -6.27786815e-01 1.42578828e+00 -1.22970450e+00
3.61352980e-01 -3.04706603e-01 -1.42939106e-01 -1.29627323e+00
-5.91916800e-01 -6.45906329e-01 -1.01548672e+00 8.55581641e-01
4.81177658e-01 -6.86392188e-01 1.21827316e+00 5.40602803e-01
2.01881036e-01 -3.84197652e-01 -6.05465114e-01 -1.01487708e+00
-2.22311616e-01 -2.50613958e-01 1.09270906e+00 1.16988671e+00
4.92114842e-01 5.98429203e-01 -9.86236800e-03 6.14755929e-01
9.73178923e-01 3.44547182e-01 1.13605309e+00 -1.57194483e+00
-3.90867829e-01 -2.51420856e-01 -1.13815093e+00 -4.18335140e-01
5.67644417e-01 -1.41007137e+00 -6.91555858e-01 -2.13584709e+00
3.52658510e-01 -4.92348284e-01 -2.22823068e-01 6.64085805e-01
2.74706502e-02 -4.43754017e-01 -1.70984715e-01 -1.61041975e-01
-5.80507517e-01 5.91736436e-01 1.13460898e+00 -3.39198202e-01
-1.64476931e-02 -2.59394795e-01 -7.45350838e-01 3.43209952e-01
3.43069434e-01 -4.48746592e-01 -8.50061834e-01 -9.61517915e-02
7.76433170e-01 4.30876911e-01 5.23980021e-01 -6.10729098e-01
5.82823515e-01 -1.64714321e-01 -1.09124452e-01 -1.70804188e-01
3.02966833e-01 -9.08989549e-01 6.85030460e-01 3.74462485e-01
-3.02332431e-01 -3.12628336e-02 -5.75619983e-04 9.43738937e-01
-1.25128940e-01 -2.39351362e-01 1.41301990e-01 -9.28917751e-02
-9.51696098e-01 5.31255007e-01 1.88023210e-01 2.55027086e-01
1.15080810e+00 -1.36176616e-01 -8.85144711e-01 -4.69468057e-01
-1.17063928e+00 2.67984390e-01 5.08837283e-01 6.16831817e-02
7.00135529e-01 -1.39788723e+00 -6.56073868e-01 2.82592676e-03
5.31835020e-01 -3.10905725e-02 7.84606040e-02 3.26805204e-01
-8.15706134e-01 2.87482858e-01 -1.89799160e-01 1.80798844e-01
-1.14587593e+00 1.22623944e+00 2.78541386e-01 -5.25279343e-01
-8.00388813e-01 5.86687505e-01 6.41765222e-02 -6.77779615e-01
3.03343326e-01 -8.32950845e-02 -1.38091892e-01 -2.02629194e-01
3.13265443e-01 5.67171276e-01 -1.05651520e-01 -1.78723902e-01
-2.41428748e-01 -6.35938570e-02 -2.08967209e-01 1.53635964e-01
1.23729753e+00 3.84725302e-01 -4.26341176e-01 -3.12361307e-02
7.81558871e-01 1.33813471e-01 -3.01086992e-01 -6.66358650e-01
5.36892237e-03 -5.24779439e-01 -2.02467933e-01 -5.20540893e-01
-8.75476241e-01 3.35156560e-01 -1.56966329e-01 5.55005789e-01
5.53689837e-01 4.06397998e-01 3.22058260e-01 1.22922659e+00
7.05556810e-01 -7.24969327e-01 -2.49435544e-01 5.85526586e-01
6.77945554e-01 -5.37406206e-01 5.01728594e-01 -9.53490734e-01
-5.71467280e-01 9.76106882e-01 3.82161140e-01 -7.56081641e-02
8.28017116e-01 1.28627643e-01 -5.95668316e-01 -9.24234211e-01
-1.07403767e+00 -4.37370181e-01 1.66878954e-01 1.02561438e+00
-1.82433873e-01 5.82942188e-01 1.37885049e-01 3.22077364e-01
-3.75323623e-01 -9.35387537e-02 8.21151733e-01 7.42474437e-01
-1.54071361e-01 -1.17237067e+00 -5.93395606e-02 5.96248686e-01
-3.81952561e-02 -3.10648799e-01 -5.02727449e-01 1.29136181e+00
-2.41785228e-01 4.14438307e-01 -2.43497699e-01 -5.46861768e-01
4.83547986e-01 1.20346069e-01 9.45508003e-01 -8.73873711e-01
-2.24414781e-01 -1.04433143e+00 6.29818022e-01 -3.85620207e-01
1.36807233e-01 -1.78359821e-01 -1.32212615e+00 -8.98155391e-01
-3.29598337e-01 4.02295500e-01 2.29968026e-01 3.89865667e-01
3.40525299e-01 5.83810866e-01 1.93422645e-01 -9.94877219e-02
-7.29789019e-01 -3.82275254e-01 -1.22186756e+00 7.21829832e-01
-2.21476674e-01 -7.63208210e-01 -2.63930351e-01 -2.71675996e-02] | [8.742385864257812, 7.675741195678711] |
b164b1a9-4ca2-49a7-80b8-894c30db3560 | optimal-and-efficient-binary-questioning-for | 2307.01578 | null | https://arxiv.org/abs/2307.01578v1 | https://arxiv.org/pdf/2307.01578v1.pdf | Optimal and Efficient Binary Questioning for Human-in-the-Loop Annotation | Even though data annotation is extremely important for interpretability, research and development of artificial intelligence solutions, most research efforts such as active learning or few-shot learning focus on the sample efficiency problem. This paper studies the neglected complementary problem of getting annotated data given a predictor. For the simple binary classification setting, we present the spectrum ranging from optimal general solutions to practical efficient methods. The problem is framed as the full annotation of a binary classification dataset with the minimal number of yes/no questions when a predictor is available. For the case of general binary questions the solution is found in coding theory, where the optimal questioning strategy is given by the Huffman encoding of the possible labelings. However, this approach is computationally intractable even for small dataset sizes. We propose an alternative practical solution based on several heuristics and lookahead minimization of proxy cost functions. The proposed solution is analysed, compared with optimal solutions and evaluated on several synthetic and real-world datasets. On these datasets, the method allows a significant improvement ($23-86\%$) in annotation efficiency. | ['Gabriele Facciolo', 'Josselin Kherroubi', 'Jean-Michel Morel', 'Franco Marchesoni-Acland'] | 2023-07-04 | null | null | null | null | ['active-learning', 'few-shot-learning', 'active-learning'] | ['methodology', 'methodology', 'natural-language-processing'] | [ 5.96069396e-01 6.88398838e-01 -4.41309363e-01 -5.21333218e-01
-9.07822192e-01 -5.21903574e-01 1.46472692e-01 6.77223086e-01
-6.43958926e-01 9.30591881e-01 -8.58384296e-02 -3.96206528e-01
-4.48795468e-01 -6.96654499e-01 -3.08202595e-01 -5.26412427e-01
1.63717270e-01 7.73808956e-01 1.89546749e-01 1.92656498e-02
3.05577666e-01 -8.10682774e-04 -1.77628374e+00 5.93354777e-02
9.51774538e-01 1.21809316e+00 2.33522117e-01 6.72021329e-01
-6.36351168e-01 8.59038651e-01 -5.59309423e-01 -6.36488438e-01
4.71224427e-01 -5.50340533e-01 -1.31312120e+00 3.34391057e-01
2.08690673e-01 2.96430681e-02 7.17196018e-02 9.98565257e-01
4.95492965e-01 4.74159658e-01 3.81655097e-01 -1.25906324e+00
-4.11692619e-01 4.52257156e-01 -8.07840824e-02 2.92779654e-01
3.16152424e-01 -5.97159863e-02 1.47698748e+00 -6.73686683e-01
6.96260750e-01 6.20728374e-01 5.08958638e-01 5.16462088e-01
-1.20650649e+00 -4.94616963e-02 -1.95603088e-01 6.60255849e-01
-1.29837263e+00 -4.77104396e-01 4.67236847e-01 -3.74332458e-01
9.19251025e-01 5.01664340e-01 3.23359907e-01 4.00731653e-01
-3.44569206e-01 6.49064541e-01 1.05259013e+00 -8.52139235e-01
8.47098172e-01 4.66600686e-01 7.70048141e-01 8.21577847e-01
2.81839609e-01 -5.53857744e-01 -2.99708009e-01 -2.50070423e-01
5.27512245e-02 -1.79185160e-02 -2.86033034e-01 -3.60750735e-01
-7.16256261e-01 9.82722104e-01 1.11645944e-01 3.02140951e-01
-2.67709374e-01 -3.19876134e-01 3.24555457e-01 5.55213451e-01
4.96815801e-01 7.31302142e-01 -6.15817785e-01 -1.88392073e-01
-9.87028420e-01 1.07831443e-02 1.24679136e+00 1.11782622e+00
9.35892344e-01 -2.57138491e-01 -2.25591063e-01 8.62687826e-01
7.50333862e-03 -2.12250069e-01 4.02983516e-01 -1.09431100e+00
6.22280657e-01 8.63423109e-01 3.84503931e-01 -8.55001807e-01
-4.49030280e-01 -1.95874393e-01 -5.33316731e-01 -2.58957058e-01
6.78546786e-01 -3.06429416e-01 -4.60309148e-01 1.33497179e+00
5.41078866e-01 -5.06645106e-02 6.72610477e-02 7.85338640e-01
6.85429633e-01 6.67185903e-01 -7.95728061e-03 -8.97485793e-01
1.32550955e+00 -1.11486113e+00 -9.80593741e-01 -3.49510550e-01
8.24457407e-01 -4.68134552e-01 1.10950160e+00 4.40388709e-01
-8.82525384e-01 -3.75318497e-01 -8.98848236e-01 -3.26701999e-01
-3.16262007e-01 1.35339409e-01 7.02228248e-01 8.13001513e-01
-8.97979438e-01 3.98179442e-01 -4.58293110e-01 -6.50472879e-01
3.44470710e-01 4.46299016e-01 -1.86517164e-01 -1.84977785e-01
-1.00824559e+00 7.96298325e-01 6.69123471e-01 -1.75202861e-01
-5.07637918e-01 -3.54042411e-01 -6.69039428e-01 3.60589713e-01
1.09130907e+00 -1.86445624e-01 1.53306127e+00 -1.09580076e+00
-1.35582566e+00 7.26104200e-01 -2.60110825e-01 -7.14179516e-01
2.30996951e-01 -2.03342400e-02 7.32670203e-02 1.62467852e-01
-1.52404979e-01 5.39164960e-01 4.46621001e-01 -8.13359737e-01
-6.02780223e-01 -2.00351790e-01 4.95391816e-01 2.05424309e-01
-7.48849213e-01 -1.11505017e-02 -2.83576190e-01 -2.88115174e-01
1.22724771e-02 -8.76635015e-01 -5.02681553e-01 2.84866005e-01
-2.08173931e-01 -3.66471678e-01 5.83980143e-01 -6.80256009e-01
1.36002457e+00 -1.95454490e+00 3.30887847e-02 -1.37922764e-01
-4.13307995e-02 2.49111652e-01 1.29912525e-01 4.36626315e-01
-1.09412428e-02 6.11569248e-02 -4.50205624e-01 -2.00713500e-01
-1.19483650e-01 6.71959341e-01 -2.65949041e-01 3.90419930e-01
1.70931146e-02 7.89220750e-01 -7.68954813e-01 -7.49450922e-01
-2.45660916e-02 -1.86175600e-01 -6.43884599e-01 4.47432369e-01
-5.51803648e-01 8.05570036e-02 -1.97985426e-01 5.66055655e-01
2.88741857e-01 -5.79311848e-01 3.53470594e-01 1.26617119e-01
-4.10935320e-02 7.50470012e-02 -1.33109188e+00 1.57674658e+00
-4.51302409e-01 6.21303558e-01 -2.32297624e-03 -1.59194231e+00
1.09601367e+00 4.89924312e-01 3.90397251e-01 -3.33407581e-01
2.46765345e-01 9.52191204e-02 -2.12143268e-02 -7.42296994e-01
5.77580690e-01 -5.67899272e-02 -1.81424156e-01 3.82549316e-01
3.75844836e-01 1.76978245e-01 5.93358517e-01 1.00086100e-01
1.29069269e+00 -9.92602482e-02 9.21739519e-01 -2.87186265e-01
4.81094986e-01 2.94892788e-01 5.44559479e-01 7.99284041e-01
-1.77349731e-01 2.75832742e-01 7.50489414e-01 -4.95681643e-01
-1.09380734e+00 -2.19852418e-01 6.52134717e-02 1.11310923e+00
1.04434885e-01 -5.47658324e-01 -9.78830397e-01 -7.45431244e-01
-4.51564252e-01 1.01740825e+00 -6.14875734e-01 1.69252053e-01
-2.76720285e-01 -7.05028236e-01 1.84576258e-01 4.49583493e-02
4.79491442e-01 -8.91499996e-01 -1.22782958e+00 2.46289223e-01
-2.95857370e-01 -9.31217432e-01 -2.93821059e-02 6.73820019e-01
-9.31567013e-01 -1.13201535e+00 -3.90564740e-01 -6.21132314e-01
6.06226861e-01 6.01697154e-02 9.08032954e-01 2.94495672e-01
-4.91076708e-01 3.70752603e-01 -6.53726518e-01 -3.38522553e-01
-1.27667353e-01 3.18317831e-01 -3.06021750e-01 1.12129159e-01
4.80372995e-01 -1.34684771e-01 -3.41083020e-01 3.25606465e-01
-8.80318105e-01 -1.34569004e-01 3.12243015e-01 1.02811313e+00
7.10534811e-01 1.28184184e-01 5.12762010e-01 -1.22001565e+00
5.32438159e-01 -7.55748212e-01 -6.16616964e-01 5.69587171e-01
-7.33772397e-01 3.16413462e-01 8.72576058e-01 -3.01092476e-01
-1.04945147e+00 4.17706549e-01 -8.77800882e-02 -5.93116544e-02
-2.69475996e-01 4.43492264e-01 -2.56768078e-01 -5.04041500e-02
8.70996892e-01 1.20294005e-01 -2.45422184e-01 -5.30286908e-01
2.42928326e-01 9.32747602e-01 2.74708241e-01 -3.03734928e-01
3.00648153e-01 2.75717407e-01 3.32916789e-02 -8.95597875e-01
-1.27531683e+00 -7.41897285e-01 -5.93671322e-01 -1.80676103e-01
6.69316590e-01 -1.79770693e-01 -6.17624164e-01 -2.78741717e-01
-1.16480649e+00 3.76071525e-03 -7.79327512e-01 4.75109555e-02
-7.25836694e-01 4.82632875e-01 -1.62510127e-01 -1.13798320e+00
-3.86705726e-01 -1.07714236e+00 6.10536218e-01 2.08970100e-01
-4.78770494e-01 -7.53983676e-01 1.76822379e-01 4.97217864e-01
1.48111507e-01 2.76546925e-02 1.16361332e+00 -1.29692864e+00
-4.93601531e-01 -2.90782481e-01 7.12085664e-02 5.08203693e-02
-1.26759499e-01 -5.30010939e-01 -9.06573534e-01 1.89543478e-02
3.10126901e-01 -5.14086366e-01 5.37176669e-01 -8.54953602e-02
1.25275087e+00 -7.74190545e-01 -1.11095376e-01 1.98522463e-01
1.66144359e+00 3.92954230e-01 4.98333156e-01 2.02263191e-01
1.71953455e-01 8.65849972e-01 1.02714753e+00 6.40986323e-01
1.26247779e-01 7.52845228e-01 5.33699095e-01 3.91857386e-01
2.21732214e-01 1.55219734e-02 -1.72410473e-01 8.72560918e-01
6.41122535e-02 -5.39099813e-01 -1.18780088e+00 6.17349327e-01
-1.98797274e+00 -7.94291019e-01 -1.21245302e-01 2.27191257e+00
7.72089303e-01 9.12720487e-02 -1.05271831e-01 7.25766718e-01
6.97742581e-01 -1.31274581e-01 -5.43264866e-01 -6.02280080e-01
3.47389951e-02 2.95920759e-01 3.37515086e-01 4.39522207e-01
-1.10705602e+00 4.85531926e-01 5.96939278e+00 9.48344767e-01
-6.29782379e-01 5.95668912e-01 7.84443378e-01 -1.59179375e-01
5.99593334e-02 2.94962257e-01 -8.40374410e-01 3.80236775e-01
1.13933611e+00 -3.79379988e-01 2.98869967e-01 1.16611540e+00
-2.22246587e-01 -4.95882869e-01 -1.05716181e+00 9.67699170e-01
2.49946982e-01 -1.21982217e+00 -3.59329909e-01 -1.63496003e-01
6.04852736e-01 -5.09385586e-01 -4.08212513e-01 3.74437183e-01
-5.15852757e-02 -6.96933329e-01 5.62223434e-01 5.01397789e-01
6.82893693e-01 -7.09825575e-01 8.80385697e-01 9.55762386e-01
-8.64283502e-01 -6.35829806e-01 -6.44280910e-01 -4.01821107e-01
2.68067569e-02 2.66700566e-01 -1.04425323e+00 5.02920985e-01
4.00105536e-01 2.16806725e-01 -5.45934021e-01 1.31568170e+00
9.07081924e-03 8.11041534e-01 -2.83737034e-01 -5.21127820e-01
1.21032625e-01 -3.94325554e-02 1.69999152e-01 1.17999482e+00
3.39183867e-01 6.94085717e-01 1.59343749e-01 2.47943863e-01
-1.88771132e-02 5.64483345e-01 -4.31208223e-01 -7.34383911e-02
4.30246085e-01 1.33386230e+00 -1.18872726e+00 -4.22285765e-01
-5.01979530e-01 8.15258265e-01 6.09835148e-01 1.37616452e-02
-6.07381463e-01 -4.90413934e-01 9.22045708e-02 6.60562441e-02
3.87992233e-01 1.17105946e-01 -5.13574123e-01 -9.03835535e-01
-4.96051162e-02 -7.56831825e-01 8.33769083e-01 -5.69520473e-01
-8.49181592e-01 4.60374475e-01 5.41940778e-02 -1.04116523e+00
-2.61365354e-01 -4.55746144e-01 -2.18367100e-01 3.91286433e-01
-1.10807240e+00 -4.18349653e-01 -4.10798758e-01 2.70257711e-01
9.68054712e-01 -5.84798157e-02 1.00828803e+00 3.60926867e-01
-5.48331559e-01 5.59515715e-01 1.25876680e-01 -2.78365821e-01
3.08736563e-01 -1.18170035e+00 -1.80611521e-01 7.39119232e-01
3.33568990e-01 2.41373941e-01 8.87485802e-01 -3.94245297e-01
-1.43066347e+00 -9.33090150e-01 1.23669791e+00 -9.09967721e-02
4.62764949e-01 -3.43562812e-01 -1.08392286e+00 4.67670232e-01
2.89119005e-01 1.99445337e-01 9.55279410e-01 -1.87022075e-01
2.24297438e-02 -3.05720959e-02 -1.35562205e+00 2.36540407e-01
8.90964210e-01 -2.13684216e-01 -6.71369076e-01 5.76600790e-01
7.03175068e-01 -1.32218659e-01 -7.39076912e-01 7.83131272e-02
4.12108712e-02 -7.19545841e-01 4.24525082e-01 -7.09707022e-01
3.55784237e-01 -6.77350238e-02 -2.21788689e-01 -7.94525683e-01
-7.89156780e-02 -6.43703043e-01 -3.45139503e-01 1.08198607e+00
4.72176880e-01 -2.65403986e-01 9.33452547e-01 9.80574369e-01
-1.19378231e-02 -1.08646333e+00 -1.25562525e+00 -4.91653323e-01
-5.69664121e-01 -3.69826168e-01 2.44975582e-01 9.53723192e-01
2.41884097e-01 4.91840720e-01 -4.43734914e-01 -2.96106309e-01
4.81739074e-01 2.50501454e-01 5.05092800e-01 -1.39369094e+00
-5.39719045e-01 1.12151749e-01 -3.97812486e-01 -9.76171851e-01
6.57538548e-02 -8.24200034e-01 2.10193098e-01 -1.42652905e+00
1.84203848e-01 -4.28670049e-01 8.09888914e-02 6.98836446e-01
-1.02589317e-02 -5.63655645e-02 2.31794849e-01 7.17324987e-02
-9.70378816e-01 1.59769908e-01 6.21204376e-01 -1.42770469e-01
1.39876241e-02 1.54784974e-02 -4.87397581e-01 7.60556102e-01
7.84294665e-01 -6.35968745e-01 -6.10628068e-01 -9.82131287e-02
5.85267186e-01 5.51850379e-01 -6.81268722e-02 -9.00365055e-01
6.18553698e-01 -1.76607132e-01 -2.86815435e-01 -3.61829907e-01
3.71632963e-01 -1.11455333e+00 1.66481942e-01 4.34917510e-01
-8.48014116e-01 -1.09558739e-01 -2.39061639e-01 6.93752944e-01
-1.98755473e-01 -1.24779451e+00 7.51277626e-01 -3.01752269e-01
-8.66392553e-01 1.75103948e-01 -3.17663014e-01 3.55318516e-01
1.46055329e+00 -3.11408430e-01 -1.79286778e-01 -3.83547753e-01
-8.01277816e-01 7.47061297e-02 2.46703416e-01 3.73318642e-02
3.84506464e-01 -9.20069456e-01 -3.61964077e-01 -2.26976454e-01
1.46357283e-01 -9.47601199e-02 9.65119749e-02 6.75492108e-01
-5.03874481e-01 7.21895576e-01 -3.14148259e-03 -2.88302958e-01
-1.40943062e+00 8.38708341e-01 3.56018334e-03 -2.62374938e-01
-2.20028877e-01 8.21726918e-01 -3.57532948e-01 -1.77702233e-01
4.55051064e-01 -6.08693017e-03 -5.89164972e-01 4.81198937e-01
4.58502263e-01 7.55795360e-01 2.21977264e-01 -3.56091529e-01
-1.29027948e-01 1.53170511e-01 1.39847681e-01 2.76588239e-02
1.20365310e+00 -2.26319104e-01 -1.49017945e-01 6.50253832e-01
1.00641239e+00 -4.04186785e-01 -8.34046423e-01 -4.64340925e-01
6.70595229e-01 -5.52019596e-01 -1.78801641e-01 -5.20225644e-01
-6.47453725e-01 9.10908282e-01 6.31684899e-01 6.97441816e-01
1.23244679e+00 -9.64149311e-02 5.06218374e-01 9.53014076e-01
5.32040715e-01 -1.14776778e+00 3.31381857e-02 4.88457739e-01
3.94845039e-01 -1.25543272e+00 -4.60794307e-02 -4.73828852e-01
-6.38707638e-01 1.16095626e+00 7.17582822e-01 1.66414738e-01
4.01996255e-01 1.89117998e-01 -3.43810260e-01 -1.14894770e-02
-1.11975920e+00 -2.04263896e-01 9.93573200e-03 2.20232144e-01
3.54631007e-01 -1.73220411e-01 -8.89552593e-01 7.40976989e-01
-7.06223175e-02 1.57436118e-01 6.26936018e-01 1.05783427e+00
-7.93477654e-01 -9.81014550e-01 -1.57315671e-01 8.33836257e-01
-5.53030849e-01 -8.04593936e-02 -3.10164124e-01 4.99641687e-01
3.98249865e-01 1.01110864e+00 2.27560014e-01 -3.95532437e-02
1.64370075e-01 5.05954087e-01 1.81745261e-01 -1.06327140e+00
-6.22199357e-01 -3.46822858e-01 3.86429906e-01 -2.56088823e-01
-5.86411536e-01 -5.88693678e-01 -1.02913022e+00 1.66091949e-01
-6.58323050e-01 6.36984229e-01 5.49165070e-01 1.11576247e+00
1.34773538e-01 9.49031189e-02 4.39188689e-01 -3.30144852e-01
-7.39892662e-01 -7.87988126e-01 -4.05729711e-01 1.74938843e-01
-2.86603849e-02 -4.92312163e-01 -3.18890125e-01 1.65478453e-01] | [9.643092155456543, 4.531122207641602] |
3efaad36-d0e9-43eb-9b9e-5ebea80d7a54 | swift-and-sure-hardness-aware-contrastive | 2201.00565 | null | https://arxiv.org/abs/2201.00565v2 | https://arxiv.org/pdf/2201.00565v2.pdf | Swift and Sure: Hardness-aware Contrastive Learning for Low-dimensional Knowledge Graph Embeddings | Knowledge graph embedding (KGE) has shown great potential in automatic knowledge graph (KG) completion and knowledge-driven tasks. However, recent KGE models suffer from high training cost and large storage space, thus limiting their practicality in real-world applications. To address this challenge, based on the latest findings in the field of Contrastive Learning, we propose a novel KGE training framework called Hardness-aware Low-dimensional Embedding (HaLE). Instead of the traditional Negative Sampling, we design a new loss function based on query sampling that can balance two important training targets, Alignment and Uniformity. Furthermore, we analyze the hardness-aware ability of recent low-dimensional hyperbolic models and propose a lightweight hardness-aware activation mechanism. The experimental results show that in the limited training time, HaLE can effectively improve the performance and training speed of KGE models on five commonly-used datasets. After training just a few minutes, the HaLE-trained models are competitive compared to the state-of-the-art models in both low- and high-dimensional conditions. | ['Quan Z. Sheng', 'Yu Liu', 'Kai Wang'] | 2022-01-03 | null | null | null | null | ['knowledge-graph-embeddings', 'knowledge-graph-embeddings'] | ['graphs', 'methodology'] | [-1.93769664e-01 5.30171357e-02 -5.09381235e-01 -1.67619195e-02
-6.57695770e-01 -3.35683405e-01 2.20402688e-01 2.89017558e-01
-4.95036066e-01 4.67687458e-01 2.86193956e-02 -3.41347009e-01
-4.84302521e-01 -1.13568187e+00 -7.71001935e-01 -6.49153233e-01
-1.97093546e-01 5.29632807e-01 3.82887602e-01 -1.27903074e-01
5.02367280e-02 3.62423927e-01 -1.43609715e+00 -1.90501913e-01
1.04922938e+00 9.14192498e-01 8.29005092e-02 4.14382786e-01
-1.28378406e-01 4.52295840e-01 -1.95100293e-01 -6.13070369e-01
2.11563230e-01 -3.28082405e-02 -7.24077702e-01 -3.35102290e-01
4.34005797e-01 -2.17229277e-01 -8.24502587e-01 9.28015172e-01
9.09868717e-01 6.18891530e-02 5.55767357e-01 -1.37876177e+00
-9.78304386e-01 5.53632021e-01 -3.68123233e-01 -3.08487732e-02
7.58131361e-03 -1.96863785e-01 1.22795415e+00 -1.17684722e+00
5.51159739e-01 9.31884348e-01 7.82872260e-01 4.37136531e-01
-1.15870810e+00 -5.76846659e-01 1.75829589e-01 7.79039025e-01
-1.76226640e+00 -4.30718698e-02 1.04491186e+00 -1.45232797e-01
9.89049673e-01 -2.06902381e-02 7.78668463e-01 6.97378814e-01
-1.50304332e-01 9.82709169e-01 8.12231600e-01 -4.69729453e-01
2.65890568e-01 2.43992820e-01 -2.68680300e-03 1.04046214e+00
6.39082015e-01 -2.43164878e-02 -5.73433876e-01 -2.84009308e-01
6.04918301e-01 -1.59425706e-01 -4.06215519e-01 -9.68178093e-01
-9.97817576e-01 9.89942491e-01 5.72497129e-01 -3.95395309e-02
-1.65795416e-01 2.59634137e-01 4.59544063e-01 2.58224756e-01
4.42798853e-01 4.48217601e-01 -4.23865259e-01 -6.62140995e-02
-5.37572980e-01 2.49721915e-01 9.58782613e-01 9.75421071e-01
8.59204590e-01 -2.14658201e-01 -2.39428505e-01 7.57143259e-01
7.74521232e-02 3.26048255e-01 8.60478953e-02 -3.41610461e-01
6.32938921e-01 9.41083789e-01 -1.63661644e-01 -1.18607259e+00
-3.72511238e-01 -5.62382042e-01 -9.41872299e-01 -1.68891698e-01
8.90179574e-02 2.68862754e-01 -7.30202854e-01 1.59337986e+00
6.44335270e-01 2.79843688e-01 7.09135607e-02 7.91107953e-01
7.30756283e-01 5.47084272e-01 1.67040024e-02 -2.24514216e-01
1.16210997e+00 -1.10465956e+00 -7.10382938e-01 -2.75053810e-02
1.09260643e+00 -4.05392468e-01 1.44103158e+00 3.48431408e-01
-9.74100590e-01 -1.81780070e-01 -1.24939334e+00 -2.26730660e-01
-7.68886209e-01 8.10453072e-02 1.14501524e+00 7.08719134e-01
-9.90099728e-01 3.84738684e-01 -6.51046336e-01 -1.72212601e-01
5.18677771e-01 3.48359942e-01 -2.86952853e-01 -5.67548811e-01
-1.45362067e+00 6.57105625e-01 5.90048492e-01 7.39687532e-02
-5.22406340e-01 -8.56160641e-01 -8.77167165e-01 3.09976071e-01
8.18450391e-01 -8.19223881e-01 6.64756417e-01 1.26554951e-01
-1.44380665e+00 5.76146662e-01 2.31954575e-01 -3.49322557e-01
4.61019397e-01 -3.68642539e-01 -2.78941572e-01 2.19044268e-01
-3.07981730e-01 4.06743437e-01 7.23790348e-01 -1.03290987e+00
-1.74743935e-01 -4.48268950e-01 9.07824039e-02 3.63789886e-01
-1.17093527e+00 -6.15525544e-01 -1.01439464e+00 -5.16840637e-01
6.87674880e-02 -8.84243011e-01 -7.78298825e-02 1.19300680e-02
-1.75373435e-01 -6.42120659e-01 8.37864459e-01 -4.16035265e-01
1.71507347e+00 -2.02679729e+00 2.57756501e-01 2.69299060e-01
4.94735122e-01 5.67903757e-01 -1.18746027e-01 5.86159229e-01
3.18471074e-01 9.68009084e-02 -8.25406909e-02 -2.10545540e-01
3.74238282e-01 3.83730620e-01 -1.31535307e-01 2.81913579e-01
4.52936664e-02 1.25839221e+00 -1.09790206e+00 -7.33625174e-01
1.27208754e-01 5.18680573e-01 -7.46293366e-01 1.72703534e-01
-2.79166013e-01 -3.05406213e-01 -4.69521403e-01 5.65527797e-01
7.69923866e-01 -5.87907910e-01 3.12709659e-01 -3.16027820e-01
4.39663142e-01 1.24406688e-01 -1.25152361e+00 1.73135054e+00
-4.77120727e-01 2.36094192e-01 -2.88037151e-01 -1.13174140e+00
7.84561694e-01 -9.78589617e-03 2.26117805e-01 -8.48404348e-01
-1.06364183e-01 2.56252468e-01 -3.66786331e-01 -3.86043966e-01
6.63578272e-01 2.58102447e-01 2.26993039e-02 3.98504436e-01
-4.72536311e-02 7.47379055e-03 1.44705594e-01 5.84800065e-01
1.20755374e+00 -1.24935992e-01 -2.17889715e-02 -4.40668501e-02
3.09449315e-01 -3.39902341e-01 4.98567790e-01 6.33642137e-01
-1.21886961e-01 2.22252682e-01 5.55651784e-01 -3.33738685e-01
-9.90867198e-01 -9.39002573e-01 5.42462543e-02 1.13166559e+00
3.62598777e-01 -9.07345295e-01 -5.43099523e-01 -9.59896505e-01
3.14318478e-01 4.81086224e-01 -6.06177270e-01 -6.26642585e-01
-4.24752355e-01 -9.06676233e-01 5.09342790e-01 7.08518445e-01
5.60564458e-01 -7.06393123e-01 2.19628098e-03 1.76032364e-01
1.12768136e-01 -1.31355953e+00 -5.79887331e-01 -9.69396532e-02
-7.58472502e-01 -1.03121388e+00 -5.71427464e-01 -9.09000158e-01
6.60890341e-01 4.81425107e-01 1.11250103e+00 8.45298469e-02
-3.75585794e-01 5.78285277e-01 -5.34774303e-01 -2.19366670e-01
9.57028866e-02 4.47070628e-01 3.24304285e-03 -2.59476125e-01
4.69189286e-01 -6.96222246e-01 -7.66305506e-01 3.15976828e-01
-9.41715658e-01 7.67948031e-02 9.78594840e-01 1.01685798e+00
7.64603317e-01 3.66509169e-01 8.34290743e-01 -8.58369172e-01
7.80921042e-01 -3.29094350e-01 -7.02332735e-01 5.90720475e-01
-1.49321640e+00 3.47782701e-01 6.45555913e-01 -5.40087581e-01
-6.01795197e-01 -2.19610766e-01 1.02040403e-01 -7.66263604e-01
6.27833605e-01 8.60167682e-01 -2.07050890e-01 -4.37323779e-01
5.35435259e-01 4.38165158e-01 -1.73943058e-01 -4.56705332e-01
6.17966950e-01 3.91789466e-01 4.40570980e-01 -7.47691572e-01
9.24900055e-01 1.54073611e-01 1.69442713e-01 -6.41087413e-01
-8.68597090e-01 -5.86064339e-01 -3.09397519e-01 1.34293139e-01
4.23054963e-01 -8.78961325e-01 -8.22926581e-01 3.43775213e-01
-7.92948961e-01 -2.81519771e-01 -3.04417670e-01 5.99668860e-01
-4.14306641e-01 6.45423889e-01 -4.96224135e-01 -6.86229825e-01
-6.10740364e-01 -7.33283758e-01 8.83025289e-01 9.18500312e-03
4.47153866e-01 -1.13667059e+00 9.28119943e-02 3.94488961e-01
4.21519130e-01 4.79795225e-02 1.36961961e+00 -4.63687092e-01
-9.14306343e-01 -3.66950065e-01 -4.91670609e-01 3.99871320e-01
-1.91424012e-01 -4.20088083e-01 -6.86955929e-01 -4.55175340e-01
-4.42993969e-01 -7.23460734e-01 1.10474086e+00 -5.67989871e-02
1.42658448e+00 -3.19691867e-01 -3.61326277e-01 8.14907908e-01
1.66692030e+00 -5.04703939e-01 6.48916483e-01 3.17654371e-01
9.73929882e-01 1.40101895e-01 6.17927432e-01 4.02098805e-01
7.44354546e-01 6.81175113e-01 4.81732965e-01 7.48992059e-03
-6.49398863e-02 -6.67421103e-01 6.99866191e-02 1.27830815e+00
7.15105832e-02 -1.65738031e-01 -7.23161042e-01 8.66335154e-01
-2.10725427e+00 -6.06017053e-01 1.97140440e-01 2.41212583e+00
1.07947397e+00 2.41188139e-01 2.49802470e-02 1.92712814e-01
4.44015503e-01 2.55242884e-01 -7.26878405e-01 -2.26360813e-01
-1.42752483e-01 1.19541399e-01 6.74365580e-01 2.45182112e-01
-9.66743529e-01 1.10989821e+00 5.87902164e+00 1.23450184e+00
-8.75256240e-01 3.20216231e-02 1.15378566e-01 8.57345238e-02
-5.01640260e-01 -5.59938923e-02 -9.73958492e-01 2.36466020e-01
7.58347571e-01 -4.19322282e-01 5.16805530e-01 1.19379377e+00
-3.91878635e-01 2.47327834e-01 -1.05475521e+00 1.18084288e+00
1.49169222e-01 -1.47161186e+00 1.79644480e-01 2.28368700e-01
7.78076172e-01 -1.67716295e-01 3.22411843e-02 7.45946407e-01
3.24425218e-03 -9.66167033e-01 8.83111581e-02 2.61363178e-01
1.00889099e+00 -9.11504328e-01 7.63736606e-01 2.65149087e-01
-1.30350161e+00 7.46069327e-02 -6.64540768e-01 1.74385205e-01
-8.26994777e-02 6.86863482e-01 -9.89048421e-01 8.42338443e-01
5.40752292e-01 5.61215758e-01 -7.15689182e-01 9.42955852e-01
-3.41233581e-01 7.09841847e-01 -4.96299624e-01 -2.14822486e-01
9.30373967e-02 -9.52182710e-02 2.46104911e-01 1.08304000e+00
7.50684589e-02 5.22729196e-02 1.98310509e-01 7.35374212e-01
-4.82131392e-01 2.41318733e-01 -5.13707995e-01 -5.25875390e-01
7.09067464e-01 1.27188885e+00 -2.64605165e-01 -2.09469512e-01
-4.98742372e-01 1.07575238e+00 8.92798007e-01 3.32761496e-01
-8.70953619e-01 -7.17534006e-01 2.94405162e-01 8.98401216e-02
7.31733084e-01 -3.88937742e-01 -4.60616201e-02 -1.12015045e+00
5.23518622e-01 -4.52240318e-01 6.22900486e-01 -3.87184858e-01
-1.44370353e+00 2.57938415e-01 -6.18111566e-02 -1.00382745e+00
4.72398475e-02 -6.81272566e-01 -3.90580863e-01 5.63236773e-01
-2.00768876e+00 -1.27372837e+00 -3.45358610e-01 6.48101747e-01
-1.13697320e-01 -6.10208996e-02 1.02360618e+00 5.94093263e-01
-5.82951903e-01 1.24352598e+00 3.99446607e-01 5.15515469e-02
6.50246203e-01 -1.36994922e+00 4.41914558e-01 4.04903769e-01
1.12671427e-01 5.84816456e-01 2.93107331e-01 -4.60036159e-01
-2.04248524e+00 -1.06757045e+00 8.09285641e-01 -2.54007012e-01
6.78317666e-01 -5.90761721e-01 -1.18974245e+00 3.65549535e-01
-4.47475582e-01 3.73256087e-01 7.68503964e-01 4.83538657e-01
-7.34519184e-01 -3.09342235e-01 -8.93035114e-01 5.40155411e-01
1.34827709e+00 -6.16061866e-01 -3.80540848e-01 6.80092335e-01
1.06722879e+00 -2.82409638e-01 -1.18538606e+00 6.38510525e-01
5.06171346e-01 -6.14281356e-01 1.07984340e+00 -6.41651809e-01
1.27146766e-01 -1.78093091e-01 8.18715394e-02 -1.08367705e+00
-4.06877697e-01 -6.07081831e-01 -9.99686062e-01 9.85354543e-01
2.20732078e-01 -7.66601980e-01 1.26702094e+00 4.75026071e-01
-7.09749535e-02 -1.37313354e+00 -8.80047202e-01 -1.21151173e+00
-7.78074116e-02 -3.42582673e-01 4.28729683e-01 1.07584667e+00
8.52326006e-02 5.63658237e-01 -4.75490957e-01 1.82256237e-01
7.10010469e-01 2.56653130e-01 9.96492863e-01 -1.19659877e+00
-3.71085703e-01 -2.41693243e-01 -6.55084908e-01 -1.18888152e+00
5.64740226e-02 -1.08491910e+00 -4.76350754e-01 -1.82127810e+00
2.83710986e-01 -6.39296591e-01 -3.96791816e-01 5.78292072e-01
-4.70830411e-01 -2.42352728e-02 -2.13504694e-02 7.78384581e-02
-9.60993350e-01 1.20100498e+00 1.31596768e+00 -9.12294388e-02
-2.62775570e-01 -3.98964643e-01 -6.50764585e-01 2.21680254e-01
6.07128978e-01 -3.60912889e-01 -7.68431604e-01 -3.19252700e-01
6.81366086e-01 -4.15435016e-01 2.01981232e-01 -8.26549232e-01
5.50436139e-01 -5.53607680e-02 6.70833215e-02 -6.34596586e-01
3.23109210e-01 -7.29949534e-01 -2.20357016e-01 3.03670347e-01
1.00791328e-01 -6.39297813e-02 1.27672568e-01 1.04134178e+00
-1.34077147e-01 3.55762169e-02 3.40043783e-01 3.76252532e-01
-9.22196567e-01 7.37202048e-01 4.19695646e-01 2.08876923e-01
9.97488499e-01 -2.75522590e-01 -5.39782465e-01 -3.01180631e-01
-2.72311807e-01 5.94134152e-01 3.64600420e-01 2.22186863e-01
8.47410977e-01 -1.57040405e+00 -4.43500251e-01 1.83731541e-01
6.29648268e-01 2.75263578e-01 4.71574277e-01 8.37872624e-01
-3.83133143e-01 4.12716687e-01 2.94578016e-01 -3.46935540e-01
-1.01428795e+00 9.82209086e-01 -1.68496162e-01 -6.81452870e-01
-6.78446472e-01 7.94591486e-01 -4.31014411e-03 -5.77641368e-01
6.03377819e-01 -3.87144871e-02 9.69404727e-02 -2.29834273e-01
3.76914412e-01 5.19627452e-01 2.36864790e-01 5.53642288e-02
-2.51403272e-01 5.95560133e-01 -3.89183372e-01 3.69591027e-01
1.32901740e+00 2.37883255e-01 1.41563013e-01 1.24890395e-01
1.36939597e+00 -1.38039008e-01 -8.96131575e-01 -6.45628214e-01
-9.58962515e-02 -5.00257194e-01 2.54773140e-01 -5.72504044e-01
-8.90405595e-01 8.88047397e-01 5.55078387e-01 1.73990473e-01
1.14676809e+00 1.36608016e-02 1.11494279e+00 8.18327188e-01
3.96799982e-01 -1.41735554e+00 2.63975739e-01 4.48048621e-01
7.06306458e-01 -1.18068933e+00 3.28543633e-01 -7.21105933e-01
-3.85893673e-01 8.58239651e-01 6.19586706e-01 1.43749282e-01
7.62734592e-01 -2.36852527e-01 -4.33574110e-01 -4.42157537e-01
-8.04867923e-01 -2.12353423e-01 4.52663481e-01 5.35383880e-01
-8.41312036e-02 2.34526792e-03 -5.39863944e-01 6.01535380e-01
-1.73839182e-02 5.46879433e-02 6.74333945e-02 8.75703752e-01
-3.04663062e-01 -1.13842940e+00 2.21044675e-01 5.20427287e-01
-2.82730311e-02 -2.31913492e-01 -3.64165753e-01 1.00523651e+00
-3.22010279e-01 4.36667413e-01 -3.66737306e-01 -6.55196726e-01
3.90141606e-01 1.78986907e-01 6.77103400e-01 -4.28107262e-01
-6.23670258e-02 -3.51122528e-01 4.59255986e-02 -4.09509599e-01
-6.28967136e-02 9.81801897e-02 -1.12585056e+00 -4.73439187e-01
-7.77598500e-01 2.91180164e-01 6.11250043e-01 5.09382546e-01
8.45074236e-01 2.99180984e-01 5.39581537e-01 -4.60509092e-01
-9.89791989e-01 -7.16178596e-01 -8.45206380e-01 5.21881402e-01
-1.68349102e-01 -8.31216574e-01 -4.32831436e-01 -3.54400486e-01] | [8.7445068359375, 7.838283061981201] |
7eddf43a-1973-4c8c-b03f-01cc1be0d570 | em-k-indexing-for-approximate-query-matching | 2111.04070 | null | https://arxiv.org/abs/2111.04070v1 | https://arxiv.org/pdf/2111.04070v1.pdf | Em-K Indexing for Approximate Query Matching in Large-scale ER | Accurate and efficient entity resolution (ER) is a significant challenge in many data mining and analysis projects requiring integrating and processing massive data collections. It is becoming increasingly important in real-world applications to develop ER solutions that produce prompt responses for entity queries on large-scale databases. Some of these applications demand entity query matching against large-scale reference databases within a short time. We define this as the query matching problem in ER in this work. Indexing or blocking techniques reduce the search space and execution time in the ER process. However, approximate indexing techniques that scale to very large-scale datasets remain open to research. In this paper, we investigate the query matching problem in ER to propose an indexing method suitable for approximate and efficient query matching. We first use spatial mappings to embed records in a multidimensional Euclidean space that preserves the domain-specific similarity. Among the various mapping techniques, we choose multidimensional scaling. Then using a Kd-tree and the nearest neighbour search, the method returns a block of records that includes potential matches for a query. Our method can process queries against a large-scale dataset using only a fraction of the data $L$ (given the dataset size is $N$), with a $O(L^2)$ complexity where $L \ll N$. The experiments conducted on several datasets showed the effectiveness of the proposed method. | ['Gary Glonek', 'Matthew Roughan', 'Samudra Herath'] | 2021-11-07 | null | null | null | null | ['entity-resolution'] | ['natural-language-processing'] | [-2.83483509e-02 -2.87228048e-01 -3.62874925e-01 -4.78737324e-01
-8.06115270e-01 -4.91830081e-01 1.26461685e-01 7.22067535e-01
-7.85759270e-01 7.81156063e-01 -5.01843281e-02 -1.03164628e-01
-7.57864952e-01 -1.59624410e+00 -5.31685174e-01 -7.81347603e-02
-2.50244915e-01 8.51419628e-01 7.83945560e-01 -2.38282621e-01
5.87334752e-01 1.03363013e+00 -1.91670263e+00 2.43375793e-01
9.48771358e-01 1.19636357e+00 2.74988085e-01 3.85386348e-01
-7.37294614e-01 3.94548565e-01 -6.95199668e-01 -4.36898112e-01
5.53377211e-01 9.64861959e-02 -9.91184115e-01 -6.48846269e-01
1.90231308e-01 -1.69239208e-01 -2.54076123e-01 1.01539385e+00
7.30168641e-01 2.35087261e-01 1.61149368e-01 -1.34222233e+00
-4.44895625e-01 3.69244039e-01 -7.01025009e-01 4.10798073e-01
5.99663675e-01 -7.52682030e-01 6.14400923e-01 -1.06819141e+00
9.72963631e-01 1.19014502e+00 6.74820065e-01 -7.54851624e-02
-1.02457976e+00 -9.53107834e-01 -3.16418946e-01 3.57203007e-01
-2.14903665e+00 -3.80575806e-01 4.62582588e-01 1.59924090e-01
1.12533677e+00 6.95317268e-01 1.45470724e-01 -2.02764049e-01
4.33629826e-02 1.97484896e-01 6.07310891e-01 -3.84111851e-01
3.81364942e-01 2.98976094e-01 8.24346021e-02 1.13566503e-01
7.14607716e-01 -1.71131611e-01 -5.35185575e-01 -6.75159991e-01
6.28228843e-01 2.74814516e-01 7.30219185e-02 -3.49628419e-01
-9.88918126e-01 8.28869581e-01 4.35945064e-01 5.73218048e-01
-6.86781943e-01 -3.04364525e-02 4.82721806e-01 4.76737350e-01
1.08878188e-01 2.73368835e-01 -4.43291426e-01 -2.37780586e-02
-1.01726365e+00 6.11696184e-01 9.94574130e-01 1.30423319e+00
8.94997239e-01 -6.41107023e-01 1.72516674e-01 8.30107749e-01
-1.77846089e-01 5.28019726e-01 3.33614349e-01 -8.75734091e-01
6.35273874e-01 1.07436812e+00 4.24771547e-01 -1.57060134e+00
-4.54319894e-01 1.32391423e-01 -9.54469979e-01 -1.17539704e-01
6.56144321e-02 4.59867865e-01 -1.85597211e-01 1.53193116e+00
8.74143779e-01 7.81771392e-02 2.00880274e-01 4.70530033e-01
5.40868580e-01 6.77817702e-01 1.23611145e-01 -6.43442750e-01
1.54299200e+00 -2.55062133e-01 -9.25092340e-01 3.23141247e-01
9.08144772e-01 -9.86098468e-01 7.31285572e-01 1.42614236e-02
-1.24870241e+00 -4.34881121e-01 -8.10021162e-01 -1.95285693e-01
-9.24380422e-01 -1.71539918e-01 5.52628815e-01 3.53651464e-01
-9.25823092e-01 2.48654470e-01 -5.12422442e-01 -6.43008828e-01
-6.07388355e-02 7.51609325e-01 -5.29052198e-01 -3.19084972e-01
-1.30344832e+00 7.92004526e-01 6.09365404e-01 -1.52385816e-01
3.24966222e-01 -6.46586597e-01 -5.68046272e-01 5.03949635e-02
3.71869206e-01 -4.22779918e-01 8.17208409e-01 -2.52439594e-03
-4.23175842e-01 7.74055004e-01 -4.66885298e-01 -3.56834561e-01
2.69570500e-01 6.26393477e-04 -7.98143506e-01 1.43033937e-01
4.26182210e-01 2.90112793e-01 1.29726129e-02 -8.12023699e-01
-1.14752662e+00 -7.30787814e-01 -1.78009287e-01 2.25104883e-01
-4.56333935e-01 4.42511946e-01 -5.81557810e-01 -5.16420364e-01
5.45408905e-01 -6.56971931e-01 -2.43378684e-01 1.68807372e-01
2.11441129e-01 -4.86541748e-01 8.23109746e-01 -5.34718871e-01
1.94348407e+00 -2.02332616e+00 -2.22021833e-01 6.47083640e-01
7.31682703e-02 1.46886811e-01 1.68718487e-01 8.53359163e-01
1.62278205e-01 1.85972929e-01 -1.28204331e-01 2.31257811e-01
3.67921852e-02 2.07267895e-01 -4.89709795e-01 3.06704044e-01
-3.68743539e-01 5.64604700e-01 -5.31388283e-01 -9.36079681e-01
-1.32888019e-01 3.09535772e-01 -4.29806292e-01 1.77671507e-01
4.23836112e-02 -5.10348678e-01 -7.10306764e-01 8.10139358e-01
1.09427643e+00 -1.75357759e-01 1.73309147e-01 -3.48971575e-01
-4.06069219e-01 -1.27945065e-01 -2.02281189e+00 1.51090646e+00
-2.44749025e-01 -4.27800082e-02 1.11826681e-01 -9.51269865e-01
1.29523742e+00 2.92238414e-01 8.05952728e-01 -1.07992983e+00
-3.73116732e-01 9.85037923e-01 -5.37663817e-01 -4.04046208e-01
1.00495017e+00 1.23088263e-01 -4.18484718e-01 5.45527637e-01
-7.72303879e-01 2.35299602e-01 4.43726748e-01 4.29073162e-02
1.23837960e+00 -5.29225290e-01 5.92987955e-01 -1.45109728e-01
7.47416794e-01 4.26770121e-01 7.05720663e-01 5.96876919e-01
3.12364474e-02 -5.42696044e-02 -3.32745491e-03 -7.62241423e-01
-1.00539052e+00 -7.34368086e-01 -4.12743807e-01 1.06365252e+00
7.04596639e-01 -4.11583960e-01 -4.08083558e-01 -1.93247691e-01
4.76165444e-01 4.48905408e-01 -2.13000506e-01 1.59137156e-02
-1.03686130e+00 -6.22354686e-01 5.60900986e-01 5.13563633e-01
6.81852281e-01 -1.15017879e+00 -6.76815987e-01 5.26822805e-01
-1.12552576e-01 -8.49520206e-01 -2.33599007e-01 -9.80248302e-02
-8.92455816e-01 -1.06164348e+00 -3.64391804e-01 -9.95184898e-01
5.45782268e-01 3.86910379e-01 9.65247393e-01 2.28428364e-01
-5.34956217e-01 -5.71642332e-02 -3.36396098e-01 -2.80704588e-01
3.18902209e-02 2.75863141e-01 3.47100645e-01 -4.17367488e-01
1.07454503e+00 -5.18877685e-01 -6.61358297e-01 7.77795255e-01
-1.10145426e+00 -4.81454432e-01 5.90562165e-01 3.51543725e-01
1.15478671e+00 4.90004838e-01 8.74339521e-01 -8.55152905e-01
8.77015769e-01 -7.24031150e-01 -8.14277351e-01 5.58689654e-01
-1.08971918e+00 9.02905166e-02 5.86393058e-01 -3.83981913e-01
-6.81708097e-01 -8.03145915e-02 -2.58248858e-02 -1.77008197e-01
8.69628638e-02 5.51367819e-01 -1.69923931e-01 -2.89405677e-02
6.49430573e-01 2.48783141e-01 -2.83435106e-01 -7.31644928e-01
2.76720405e-01 1.01133502e+00 5.91472387e-01 -7.84013689e-01
7.25371540e-01 5.76499462e-01 2.48457059e-01 -3.75333190e-01
-6.38804361e-02 -7.49185503e-01 -5.74427485e-01 5.01422167e-01
5.26444435e-01 -7.26213872e-01 -8.79911602e-01 -9.97943878e-02
-1.09183145e+00 5.52097976e-01 -4.32952821e-01 3.53518903e-01
-4.21820194e-01 5.07536978e-02 -2.57646501e-01 -6.73357308e-01
-5.85143209e-01 -7.59459674e-01 8.54435802e-01 2.26150677e-01
-2.30995491e-01 -3.93673003e-01 3.76579225e-01 6.64312625e-03
7.18695223e-01 2.50406027e-01 9.13268983e-01 -1.04440856e+00
-8.08975399e-01 -5.09966195e-01 -6.26141250e-01 -7.11729050e-01
1.46688610e-01 -4.39016640e-01 -2.42007896e-01 -2.81670958e-01
-1.81837603e-01 6.80748373e-02 1.24834180e-02 -1.20117158e-01
1.16167808e+00 -3.17198277e-01 -8.10456455e-01 6.26292288e-01
1.69041705e+00 6.14995360e-01 6.35986447e-01 7.12043583e-01
1.95734307e-01 6.56063557e-01 1.42053199e+00 7.25225031e-01
5.23878694e-01 8.73057663e-01 -2.46176198e-02 -8.47058669e-02
3.55263203e-01 -2.56813318e-01 -4.54357564e-01 5.76285005e-01
3.13940883e-01 -1.61272168e-01 -9.42246914e-01 7.49232948e-01
-1.90115786e+00 -1.10665715e+00 -9.31421518e-02 2.41427374e+00
9.32096839e-01 -1.52809620e-01 -2.72054989e-02 4.03047204e-01
8.88943613e-01 -1.53155923e-01 -7.08227813e-01 -5.12650251e-01
-7.98980817e-02 5.25991738e-01 7.32583165e-01 1.89064115e-01
-6.87576711e-01 6.65629327e-01 5.80082846e+00 8.02082896e-01
-1.02074599e+00 -2.01507062e-01 1.94286540e-01 -9.87171978e-02
-2.93358505e-01 -7.66007155e-02 -1.21665561e+00 4.17146087e-01
1.10786796e+00 -7.63114095e-01 4.06125933e-01 9.89582717e-01
9.52564850e-02 -4.95601557e-02 -9.85012531e-01 1.18452740e+00
-2.79669732e-01 -1.44253254e+00 1.63704023e-01 3.17676604e-01
2.95387596e-01 -2.91117162e-01 -3.86140317e-01 1.20352544e-01
5.02179265e-02 -8.48626554e-01 2.39269510e-01 4.88723308e-01
1.11708009e+00 -1.10119629e+00 8.04898679e-01 4.24743384e-01
-1.99217498e+00 -1.49666026e-01 -7.56163299e-01 2.82363683e-01
2.05609933e-01 3.74628246e-01 -5.12690187e-01 5.68596303e-01
1.16503906e+00 -4.64495942e-02 -3.76662582e-01 1.02055907e+00
8.89542520e-01 -4.04796958e-01 -9.07836139e-01 8.20394605e-02
-3.87084395e-01 -2.04528198e-01 1.13304570e-01 1.00755274e+00
8.73212099e-01 6.04832768e-01 7.65231177e-02 4.01193917e-01
-3.80075395e-01 6.73908472e-01 -7.38052785e-01 5.54269888e-02
1.45016479e+00 8.50673616e-01 -5.96793830e-01 -3.57341021e-01
-4.18303102e-01 6.93731189e-01 3.30963641e-01 -2.29143035e-02
-6.05764151e-01 -1.13329542e+00 5.98296106e-01 4.98697996e-01
3.05862546e-01 -9.26674232e-02 -2.37182334e-01 -6.25145674e-01
4.90899533e-01 -8.48490298e-01 9.42171216e-01 -3.68462622e-01
-9.95926559e-01 7.48682976e-01 2.67760277e-01 -1.17233944e+00
-3.60787898e-01 -4.96530347e-03 -1.63102895e-01 1.02300811e+00
-1.44828057e+00 -7.71692753e-01 -4.48755413e-01 8.93950045e-01
1.04301259e-01 8.01612511e-02 9.65131223e-01 8.72860789e-01
-2.78061479e-01 6.97255254e-01 2.75768697e-01 -1.68148547e-01
8.37786138e-01 -8.58190238e-01 3.49576831e-01 5.55893362e-01
-2.58704871e-01 1.12899852e+00 5.85129917e-01 -7.96196938e-01
-1.54795194e+00 -8.57573152e-01 1.60082829e+00 -2.71996167e-02
2.81707436e-01 -1.13796301e-01 -1.22846103e+00 3.53027284e-01
-4.04191971e-01 4.77489233e-01 7.59857893e-01 -2.39014253e-01
-2.18463957e-01 -6.52734637e-01 -1.95061183e+00 2.36562967e-01
1.25796509e+00 -4.58039194e-01 -6.10655427e-01 7.73111880e-02
8.23211312e-01 -2.57381022e-01 -1.57149243e+00 5.94062746e-01
5.70299566e-01 -6.86998904e-01 1.24254394e+00 -3.24112207e-01
-4.86586303e-01 -7.53884435e-01 -6.29022777e-01 -2.97626585e-01
-5.31200916e-02 -4.10470963e-01 -2.42770284e-01 1.29336119e+00
9.10002813e-02 -7.41726160e-01 7.21514165e-01 1.25603390e+00
5.61761141e-01 -7.21285403e-01 -9.98334110e-01 -7.14021981e-01
-2.50986665e-01 -4.94093029e-03 1.35975051e+00 1.05580282e+00
-6.74445406e-02 -1.92570284e-01 1.28567025e-01 4.02121663e-01
6.40091062e-01 6.51109278e-01 9.78648484e-01 -1.65177429e+00
2.57302165e-01 8.21383111e-03 -3.94727051e-01 -8.51073742e-01
-3.43568802e-01 -5.83455741e-01 -5.31505406e-01 -1.49847877e+00
-1.90715920e-02 -1.14594901e+00 -3.06459427e-01 2.00907111e-01
2.09261417e-01 -4.53375885e-03 -1.41904429e-01 6.71263397e-01
-5.57086587e-01 1.03501372e-01 4.13103670e-01 2.26855919e-01
-4.58769411e-01 -9.68972221e-02 -5.69977939e-01 1.94797188e-01
4.92777348e-01 -8.27889323e-01 -4.49106306e-01 -3.48474741e-01
4.91398931e-01 3.67186189e-01 -2.88490772e-01 -8.87730718e-01
8.34737957e-01 -4.92508262e-01 2.21359476e-01 -1.03402627e+00
1.26068324e-01 -1.22559845e+00 8.50781202e-01 3.57353270e-01
-1.97613955e-01 7.33677030e-01 7.48456866e-02 4.81265992e-01
-5.55739701e-01 -1.24307476e-01 6.03769898e-01 3.02469032e-03
-8.62119436e-01 5.48405766e-01 4.46005225e-01 -1.72076840e-02
1.42078912e+00 -4.26696241e-01 -2.86414564e-01 2.07034677e-01
-3.68547708e-01 3.99921596e-01 8.02271843e-01 2.22062528e-01
6.49429739e-01 -1.55973351e+00 -5.01620889e-01 1.70259893e-01
2.83893824e-01 3.48023146e-01 2.78888270e-02 3.09376121e-01
-8.71199250e-01 6.37317836e-01 -1.89867005e-01 -3.49241704e-01
-1.42439795e+00 8.62084150e-01 -3.16165313e-02 -4.31585729e-01
-4.38914210e-01 4.52091813e-01 -3.47180456e-01 -4.47775126e-01
2.29808033e-01 -5.73043525e-02 -8.38824436e-02 1.45521209e-01
8.68479371e-01 7.40965664e-01 1.15356646e-01 -6.35226369e-01
-6.37748301e-01 7.41963565e-01 -9.05031785e-02 -5.21280803e-02
1.41123176e+00 -3.57294500e-01 -5.75482190e-01 5.31049035e-02
1.29245639e+00 2.86432683e-01 -2.52872974e-01 -6.62451744e-01
6.25076950e-01 -8.84220779e-01 -3.35865617e-01 -2.49455750e-01
-6.77081048e-01 2.13027343e-01 7.81628132e-01 3.45580995e-01
1.42026854e+00 -7.00772703e-02 1.18848395e+00 6.12404287e-01
9.27946568e-01 -1.27095234e+00 -6.36674225e-01 -5.60258515e-02
7.79212534e-01 -9.38479006e-01 2.57248491e-01 -3.94586444e-01
-5.48099466e-02 9.83375728e-01 6.65136755e-01 1.43514639e-02
7.97483087e-01 4.11897242e-01 -2.25248680e-01 -3.02275777e-01
-6.66523397e-01 8.20512176e-02 -1.61269560e-01 2.60556906e-01
1.54914856e-01 -1.44446656e-01 -1.08547020e+00 6.12038076e-01
-2.29097292e-01 3.82752478e-01 6.49892837e-02 1.48280764e+00
-7.82488585e-01 -1.52369928e+00 -5.55985272e-01 4.30999368e-01
-5.04896462e-01 1.94081217e-01 -9.11078379e-02 7.56112397e-01
1.79480895e-01 7.54104137e-01 3.82319510e-01 -1.60417065e-01
8.47205460e-01 1.36218190e-01 -1.63088232e-01 -1.37147665e-01
-4.73198116e-01 -3.82366627e-01 -1.23622961e-01 -7.66247690e-01
-2.28141874e-01 -4.19856429e-01 -1.85049284e+00 -8.12923312e-01
-3.08183610e-01 6.49553895e-01 9.25142825e-01 2.95427293e-01
8.58869672e-01 -2.69386560e-01 6.81744218e-01 -3.94223407e-02
-5.76403737e-01 -5.45086503e-01 -7.15407312e-01 4.32923496e-01
-6.48893043e-02 -5.48282385e-01 -4.06364724e-02 -1.93462417e-01] | [8.235560417175293, 6.7888994216918945] |
28ed3f7d-17b2-4e60-94d8-78ce99818813 | varnet-exploring-variations-for-unsupervised | null | null | https://ieeexplore.ieee.org/document/8594264 | https://ieeexplore.ieee.org/document/8594264 | VarNet: Exploring Variations for Unsupervised Video Prediction | Unsupervised video prediction is a very challenging task due to the complexity and diversity in natural scenes. Prior works directly predicting pixels or optical flows either have the blurring problem or require additional assumptions. We highlight that the crux for video frame prediction lies in precisely capturing the inter-frame variations which encompass the movement of objects and the evolution of the surrounding environment. We then present an unsupervised video prediction framework - Variation Network (VarNet) to directly predict the variations between adjacent frames which are then fused with current frame to generate the future frame. In addition, we propose an adaptively re-weighting mechanism for loss function to offer each pixel a fair weight according to the amplitude of its variation. Extensive experiments for both short-term and long-term video prediction are implemented on two advanced datasets - KTH and KITTI with two evaluating metrics - PSNR and SSIM. For the KTH dataset, the VarNet outperforms the state-of-the-art works up to 11.9% on PSNR and 9.5% on SSIM. As for the KITTI dataset, the performance boosts are up to 55.1% on PSNR and 15.9% on SSIM. Moreover, we verify that the generalization ability of our model excels other state-of-the-art methods by testing on the unseen CalTech Pedestrian dataset after being trained on the KITTI dataset. Source code and video are available at https://github.com/jinbeibei/VarNet. | ['Jing Ye', 'Shice Liu', 'Qiankun Tang', 'Yiming Zeng', 'Yu Hu', 'Beibei Jin'] | 2018-10-01 | null | null | null | ieee-rsj-international-conference-on-4 | ['video-prediction'] | ['computer-vision'] | [ 6.80522546e-02 -4.49415743e-01 -1.06802091e-01 -3.29994112e-01
-1.54475257e-01 -1.95666067e-02 2.80802667e-01 -5.03064275e-01
-3.04710746e-01 8.87289762e-01 1.77917615e-01 9.83442664e-02
5.61689883e-02 -6.16615593e-01 -8.00168812e-01 -9.37691033e-01
-2.53879458e-01 -3.85599583e-01 5.77681303e-01 -1.73179120e-01
8.11445117e-02 1.73923090e-01 -1.58340418e+00 1.99807376e-01
7.46991873e-01 1.32075274e+00 3.29235137e-01 7.80023277e-01
1.70966879e-01 1.25057197e+00 -3.26388389e-01 -4.40756738e-01
4.68000144e-01 -4.30966288e-01 -3.46172184e-01 2.66917527e-01
4.96998101e-01 -6.43005550e-01 -8.68452132e-01 9.49568510e-01
4.41472173e-01 3.28413755e-01 3.35740477e-01 -1.23747492e+00
-5.60432851e-01 3.74848396e-01 -5.65362334e-01 5.85298538e-01
2.36119583e-01 4.93288010e-01 6.15743995e-01 -9.18506444e-01
5.94594955e-01 1.04591835e+00 5.90946257e-01 4.88315433e-01
-8.93258870e-01 -7.00333118e-01 2.70165622e-01 7.95473039e-01
-1.26079845e+00 -5.67448318e-01 7.67868042e-01 -5.35364509e-01
6.24045014e-01 1.16504796e-01 6.66749299e-01 1.20110774e+00
4.89560753e-01 8.22008133e-01 7.44485080e-01 -7.47882426e-02
6.90678060e-02 -1.04860410e-01 -9.49742645e-02 5.60644925e-01
-1.56292960e-01 5.10199189e-01 -5.35901368e-01 5.20731211e-01
7.26213574e-01 4.87703718e-02 -5.86266994e-01 -1.62960127e-01
-1.19298315e+00 3.84625196e-01 4.47165668e-01 -6.46848157e-02
-4.17586923e-01 3.72856557e-01 3.66015971e-01 2.23842338e-01
5.52940905e-01 -1.50663182e-01 -4.59176242e-01 -3.56360018e-01
-9.66505349e-01 9.45704803e-02 4.69614506e-01 9.33020830e-01
7.07074940e-01 3.99017930e-01 -3.37866187e-01 8.24002683e-01
3.70525330e-01 3.82311970e-01 4.40844208e-01 -1.10486448e+00
4.50792700e-01 1.28043905e-01 1.21538088e-01 -1.27365398e+00
-4.16170582e-02 -6.30482972e-01 -9.64303195e-01 3.09737027e-01
4.58939612e-01 -3.19317341e-01 -8.36048663e-01 1.54015827e+00
2.09091216e-01 8.40117335e-01 6.24007219e-03 1.08255434e+00
8.47659469e-01 9.56085205e-01 4.84390333e-02 -4.48328048e-01
8.79647195e-01 -1.34230614e+00 -7.58553088e-01 -1.80739209e-01
1.20413326e-01 -7.51225770e-01 6.89414084e-01 3.34897339e-01
-1.04283941e+00 -1.08512151e+00 -9.74380076e-01 1.42711386e-01
1.25377521e-01 9.72026065e-02 3.30804735e-01 3.93666029e-01
-1.07302892e+00 6.52011395e-01 -7.94237554e-01 -2.10957289e-01
5.11685491e-01 9.97169539e-02 -1.36187166e-01 -2.47258306e-01
-1.12250710e+00 5.96476436e-01 3.10116112e-01 1.81228995e-01
-1.11850131e+00 -1.02206731e+00 -7.50787377e-01 3.80084477e-02
4.06063288e-01 -5.48277795e-01 9.54547644e-01 -1.36229491e+00
-1.61201298e+00 4.26078796e-01 -2.36986086e-01 -7.88024426e-01
7.35058665e-01 -3.59481543e-01 -5.24881005e-01 2.73494095e-01
-8.62086713e-02 8.25646460e-01 8.33620429e-01 -1.05919909e+00
-8.44699919e-01 6.69123232e-02 9.33850780e-02 1.45868510e-01
-3.46806765e-01 -2.74571925e-02 -8.02387774e-01 -9.25716579e-01
-1.74701095e-01 -9.11467910e-01 -1.50579214e-01 3.84897023e-01
-1.37881905e-01 1.56586528e-01 1.03964055e+00 -8.93624485e-01
1.33751833e+00 -2.28487253e+00 -6.81058615e-02 -3.61883432e-01
6.74195364e-02 5.07717550e-01 -1.06503524e-01 -5.73968813e-02
-3.02465912e-02 -2.33570382e-01 -9.56005827e-02 -2.08370253e-01
-2.36464188e-01 1.79596081e-01 -1.99398652e-01 3.38687867e-01
1.44705459e-01 7.00824738e-01 -9.14071441e-01 -3.43537956e-01
5.46766162e-01 7.79450893e-01 -4.74615991e-01 1.87693626e-01
-8.09568092e-02 5.47394574e-01 -1.41006336e-01 5.29576659e-01
7.91576087e-01 -2.02874988e-01 -1.43129095e-01 -3.72328937e-01
-1.19352840e-01 -2.16966495e-01 -1.06410742e+00 1.47389579e+00
-1.74834713e-01 1.07315874e+00 -3.56175452e-01 -9.01338518e-01
7.99356461e-01 1.86476052e-01 5.61190486e-01 -7.27199614e-01
1.54186144e-01 -5.67983575e-02 5.05748130e-02 -6.68297052e-01
4.96768147e-01 3.15798283e-01 3.78956646e-01 -9.55973491e-02
2.64371745e-02 4.98593837e-01 4.15345013e-01 1.11618496e-01
8.46230984e-01 3.47443998e-01 -4.71975021e-02 -2.60328203e-01
7.94551313e-01 -4.44064319e-01 1.04972410e+00 4.98781592e-01
-6.12920284e-01 7.68474281e-01 2.79410183e-01 -6.24711692e-01
-9.75007534e-01 -1.08111739e+00 -1.33396927e-02 8.10322940e-01
5.04251838e-01 -2.98543900e-01 -6.00227416e-01 -4.29271847e-01
-3.00808758e-01 5.93875945e-01 -4.96823430e-01 -9.29830447e-02
-5.03476858e-01 -6.38970554e-01 1.67741582e-01 6.09322131e-01
9.93392527e-01 -8.81016016e-01 -5.55579305e-01 2.23408148e-01
-3.25673163e-01 -1.47996092e+00 -5.19691586e-01 -5.01494646e-01
-6.67955339e-01 -9.58991468e-01 -7.21745729e-01 -7.32893050e-01
3.91035527e-01 3.20160240e-01 9.73113000e-01 6.33234307e-02
-2.65315861e-01 3.13924223e-01 -3.36482555e-01 -1.42489284e-01
-1.67691275e-01 -3.77291292e-01 -3.60989608e-02 2.94494927e-01
9.85588878e-02 -6.50942385e-01 -1.04656661e+00 5.58363318e-01
-7.01190770e-01 3.38227004e-01 2.19202474e-01 7.51489639e-01
6.84820831e-01 2.94991404e-01 1.93823427e-01 -3.72170806e-01
-8.40892121e-02 -4.90971744e-01 -6.48152053e-01 1.30960599e-01
-3.91176075e-01 -2.67412096e-01 5.98693728e-01 -5.07881105e-01
-1.34236026e+00 1.98511183e-02 4.92884703e-02 -6.07732356e-01
-1.44897863e-01 8.58417377e-02 -1.66316912e-01 -1.95651501e-03
3.09751302e-01 3.22580189e-01 -7.92420190e-03 -1.58528283e-01
7.33138015e-03 3.40247810e-01 6.72382832e-01 -1.81279659e-01
7.52134562e-01 5.49189150e-01 -3.76081988e-02 -7.79288709e-01
-7.63205171e-01 -1.95575938e-01 -4.21728313e-01 -7.41584957e-01
9.14358497e-01 -1.19766414e+00 -5.90358973e-01 8.42894316e-01
-8.99352193e-01 -5.99677920e-01 -1.18642002e-01 5.96438110e-01
-4.16542917e-01 5.79368472e-01 -8.47385406e-01 -6.57214165e-01
-3.35366696e-01 -1.35107744e+00 5.37723660e-01 4.09103632e-01
3.24343920e-01 -1.07298458e+00 -3.25142294e-01 3.81481141e-01
5.40207744e-01 2.97375441e-01 2.86327153e-01 1.19333282e-01
-8.89658928e-01 1.32239476e-01 -2.87086487e-01 6.90781236e-01
8.35320279e-02 3.65693420e-01 -8.18167508e-01 -3.01027179e-01
1.49913393e-02 1.27281100e-01 9.68298018e-01 6.66862309e-01
1.36747432e+00 -1.18049294e-01 -1.42533481e-02 9.09255445e-01
1.52859640e+00 4.54775691e-01 1.06452096e+00 3.72491509e-01
8.43561888e-01 3.46056461e-01 7.54667342e-01 5.63605249e-01
3.81115049e-01 8.43837202e-01 5.33286929e-01 1.46897376e-01
-4.75023150e-01 -1.03422038e-01 6.16970718e-01 7.08696723e-01
-2.75462598e-01 -4.92422879e-01 -7.21401691e-01 4.50163215e-01
-2.02148223e+00 -1.32315028e+00 -2.30293199e-01 2.02197623e+00
4.24584687e-01 1.98598102e-01 -6.79595843e-02 1.41025826e-01
7.18297541e-01 2.98841268e-01 -5.10166943e-01 1.01093531e-01
-2.82065183e-01 -1.50432780e-01 5.15689731e-01 5.16068816e-01
-1.28633428e+00 9.75277901e-01 5.44605446e+00 8.61020923e-01
-1.30338931e+00 4.06292081e-02 1.18080783e+00 -2.36855313e-01
2.70210415e-01 -1.37928590e-01 -7.33397245e-01 1.05621195e+00
9.02383626e-01 -1.14488728e-01 3.30550551e-01 6.58234835e-01
6.36897385e-01 -2.76127219e-01 -7.98975468e-01 1.17507899e+00
1.95130289e-01 -1.33930993e+00 -7.10928440e-02 -1.50419056e-01
9.37900782e-01 9.85800996e-02 1.20870478e-01 1.77566499e-01
-7.43863136e-02 -8.40432882e-01 8.16962540e-01 8.60629678e-01
5.33402026e-01 -6.68281376e-01 8.30565572e-01 7.86011219e-02
-1.29855037e+00 -1.88599765e-01 -5.16132951e-01 -2.30622351e-01
4.65811014e-01 6.50198579e-01 -1.37391955e-01 4.66753691e-01
9.52553391e-01 1.28461349e+00 -5.79231083e-01 1.31047642e+00
-3.69223595e-01 7.92286694e-01 -1.15513995e-01 3.30414474e-01
1.99873880e-01 -2.40545422e-01 6.17073596e-01 1.18603289e+00
5.17812490e-01 1.43531710e-01 1.59921676e-01 4.79295343e-01
3.64922686e-04 -1.26341254e-01 -9.27627087e-02 4.98879075e-01
2.84154207e-01 1.02687526e+00 -4.20060694e-01 -4.21905100e-01
-4.07846630e-01 1.02190900e+00 -1.35011315e-01 6.41577303e-01
-1.20486891e+00 6.40712818e-03 9.62881088e-01 1.32197933e-03
7.92570829e-01 -6.64694756e-02 -4.61776070e-02 -1.28354132e+00
2.85374969e-01 -5.83348393e-01 1.87376425e-01 -1.00772405e+00
-1.18064475e+00 6.65875435e-01 -1.30133286e-01 -1.54658628e+00
-1.44425958e-01 -6.45679712e-01 -5.72976768e-01 4.75352079e-01
-1.71109295e+00 -7.35267520e-01 -7.66146898e-01 5.81278920e-01
8.83170366e-01 -3.37556750e-01 2.27099538e-01 6.14836514e-01
-9.36644673e-01 6.03931367e-01 2.76488513e-01 2.51770288e-01
5.68055391e-01 -7.24734068e-01 1.85184225e-01 1.32602990e+00
-1.32505909e-01 -4.72597294e-02 7.23597109e-01 -3.66115957e-01
-1.20231628e+00 -1.38523877e+00 6.31036401e-01 -1.43573508e-01
6.14907801e-01 -8.77704993e-02 -8.28339458e-01 5.57015836e-01
2.19857946e-01 6.05996013e-01 3.11259657e-01 -5.96780598e-01
-1.50283456e-01 -5.12605786e-01 -9.94993389e-01 5.90757847e-01
1.22564781e+00 -1.02266289e-01 5.49629033e-02 1.12407319e-01
7.37191677e-01 -4.85653639e-01 -8.61040831e-01 3.90866131e-01
6.40607595e-01 -1.44496918e+00 1.11700678e+00 -1.44075304e-01
9.28650320e-01 -5.39104283e-01 -3.37256730e-01 -1.04331219e+00
-4.70323592e-01 -6.06497228e-01 -4.19735104e-01 1.21575761e+00
2.68796477e-02 -4.08483624e-01 7.29528189e-01 6.44174576e-01
-3.25412117e-02 -8.94337893e-01 -7.14804590e-01 -6.84163868e-01
-2.70141602e-01 -7.37664700e-01 3.78323823e-01 5.99129319e-01
-4.58484203e-01 1.07413456e-01 -8.52430940e-01 1.95553750e-01
8.20091844e-01 -2.07366735e-01 6.94948196e-01 -6.66083694e-01
-3.56978863e-01 -2.65533566e-01 -9.47483301e-01 -1.28645074e+00
7.25005865e-02 -3.77950788e-01 5.25902659e-02 -1.21182251e+00
1.92376003e-01 -1.22688852e-01 -5.46078026e-01 7.29458407e-02
-3.30772460e-01 4.59281206e-01 5.05902410e-01 1.44151047e-01
-7.46326447e-01 7.35993147e-01 1.16436195e+00 -1.19240046e-01
-1.13606684e-01 -2.28867158e-02 -3.43057543e-01 7.41231620e-01
9.29505825e-01 -7.93327019e-02 -6.30921721e-01 -7.41505146e-01
-3.07202160e-01 1.72342896e-01 6.90190256e-01 -1.49994552e+00
2.24804133e-01 -1.64343238e-01 6.56643033e-01 -5.54165125e-01
3.49364311e-01 -7.81406701e-01 4.08853024e-01 3.74942333e-01
-2.81323045e-01 -6.84868842e-02 2.13484854e-01 6.75253093e-01
-3.30901295e-01 1.02440186e-01 8.95368457e-01 1.50606722e-01
-1.32908797e+00 7.22172558e-01 -3.95597190e-01 -7.71124428e-03
1.19634819e+00 -3.88738483e-01 -1.55267224e-01 -5.25466263e-01
-6.09054208e-01 1.71775818e-01 3.96519542e-01 5.33951998e-01
8.74474585e-01 -1.36427557e+00 -8.08522642e-01 1.57772094e-01
-1.58305034e-01 -2.77466476e-01 6.55221581e-01 9.02916849e-01
-6.74030662e-01 1.30610332e-01 -4.30693418e-01 -7.21852541e-01
-1.35966933e+00 4.72343594e-01 4.16026354e-01 -1.11413978e-01
-6.56838000e-01 9.48946536e-01 4.88262624e-01 2.69046336e-01
3.87162477e-01 -2.18295038e-01 -1.27646104e-01 -2.14682445e-01
8.51831853e-01 5.60300887e-01 -4.01446760e-01 -8.54590237e-01
-2.77519584e-01 4.85966384e-01 8.24914593e-03 3.41720670e-01
1.45051265e+00 -4.54327911e-01 1.58157200e-01 1.63693905e-01
1.27830744e+00 -3.87340099e-01 -2.09544110e+00 -3.13249707e-01
-3.64819944e-01 -7.91138887e-01 1.79566726e-01 -5.96225560e-01
-1.62942469e+00 6.91914916e-01 1.03698766e+00 -2.69646764e-01
1.43636131e+00 -4.46438491e-01 1.00712383e+00 4.38108929e-02
3.10903668e-01 -9.12364006e-01 1.98561102e-01 5.01602709e-01
7.01311588e-01 -1.34511697e+00 1.20707184e-01 -4.88233328e-01
-6.41464412e-01 1.05434382e+00 7.80878186e-01 -3.28297764e-01
8.13798368e-01 1.76223665e-01 1.53175488e-01 2.42196098e-01
-8.18054974e-01 1.34698013e-02 2.61585504e-01 5.53159297e-01
3.72310787e-01 -1.60268143e-01 -1.11990280e-01 1.27939403e-01
-6.26191869e-02 2.78558195e-01 5.36187768e-01 5.67752481e-01
-1.99065596e-01 -6.77330136e-01 -1.96826681e-01 2.54693687e-01
-4.89561260e-01 -8.59688222e-02 3.00635457e-01 5.81782103e-01
2.91090995e-01 1.11204481e+00 1.51821643e-01 -5.37092924e-01
1.30011603e-01 -4.42489177e-01 2.90746182e-01 4.35056835e-02
-2.07407817e-01 5.29042855e-02 -1.11702934e-01 -7.87828445e-01
-8.17013979e-01 -6.57295823e-01 -9.43856478e-01 -5.45074761e-01
8.52348283e-02 -2.67742515e-01 2.39945218e-01 8.04110646e-01
3.17516744e-01 7.89546549e-01 6.56134248e-01 -1.01385319e+00
-5.64562716e-02 -7.64298081e-01 -2.96213835e-01 6.00709319e-01
3.15052301e-01 -6.55939400e-01 -3.85078818e-01 5.45517445e-01] | [10.470507621765137, -1.2027567625045776] |
6c8c2b6b-5592-4b72-af30-b932bc821999 | multi-scale-evolutionary-neural-architecture | 2304.10749 | null | https://arxiv.org/abs/2304.10749v2 | https://arxiv.org/pdf/2304.10749v2.pdf | Emergence of Brain-inspired Small-world Spiking Neural Network through Neuroevolution | Human brain is the product of evolution during hundreds over millions of years and can engage in multiple advanced cognitive functions with low energy consumption. Brain-inspired artificial intelligence serves as a computational continuation of this natural evolutionary process, is imperative to take inspiration from the evolutionary mechanisms of brain structure and function. Studies suggest that the human brain's high efficiency and low energy consumption may be closely related to its small-world topology and critical dynamics. However, existing efforts on the performance-oriented structural evolution of spiking neural networks (SNNs) are time-consuming and ignore the core structural properties of the brain. In this paper, we propose a multi-objective Evolutionary Liquid State Machine (ELSM) with the combination of small-world coefficient and criticality as evolution goals and simultaneously integrate the topological properties of spiking neural networks from static and dynamic perspectives to guide the emergence of brain-inspired efficient structures. | ['Yiting Dong', 'Yi Zeng', 'Bing Han', 'Feifei Zhao', 'Wenxuan Pan'] | 2023-04-21 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 5.60149811e-02 -1.83364347e-01 4.41847205e-01 -7.84712136e-02
6.71081603e-01 -3.93328547e-01 6.05590343e-01 -2.75731772e-01
-3.58465016e-01 1.01475251e+00 -1.71247050e-01 -3.34478989e-02
-3.36749524e-01 -6.63029969e-01 -4.83242393e-01 -1.07040417e+00
-2.91434199e-01 2.15600818e-01 4.87517267e-01 -7.47700274e-01
5.40245116e-01 5.77714801e-01 -1.62360132e+00 -5.12213051e-01
1.37259078e+00 7.26428270e-01 3.03255230e-01 5.51608086e-01
-5.93764335e-02 2.91953951e-01 -2.42754087e-01 -2.75352508e-01
2.91903138e-01 -1.06976044e+00 -4.58517224e-01 -3.83341849e-01
-4.68002498e-01 6.45712197e-01 -3.63419354e-01 9.53895211e-01
6.37847960e-01 2.65270248e-02 7.93114305e-01 -9.70466197e-01
-8.11365008e-01 4.90690678e-01 -2.42872044e-01 6.03306890e-01
-5.67654908e-01 6.02000296e-01 7.54742205e-01 -6.16560519e-01
6.33937120e-01 6.72035038e-01 4.52688158e-01 9.01537120e-01
-1.09489310e+00 -7.09506631e-01 -3.38763386e-01 1.84550077e-01
-9.86730337e-01 -4.99617428e-01 8.83206844e-01 -3.04564565e-01
1.49470830e+00 1.59187280e-02 1.39776814e+00 7.69946754e-01
1.04133904e+00 1.98639706e-02 1.17906654e+00 -2.95566380e-01
6.22665524e-01 -4.61370140e-01 -6.24207072e-02 9.62321699e-01
7.40098000e-01 4.19830978e-01 -9.11317050e-01 -6.24242052e-02
6.11378551e-01 -4.06616628e-01 -2.85736769e-01 -3.04686278e-01
-1.01529002e+00 4.21079487e-01 5.09065628e-01 5.54787457e-01
-3.26215178e-01 6.38929963e-01 5.48291355e-02 -1.14541883e-02
-6.17550351e-02 1.10295665e+00 -3.76636028e-01 -3.55990708e-01
-7.54077375e-01 2.27903835e-02 7.12642908e-01 3.25061560e-01
4.21255231e-01 3.22679549e-01 2.88345873e-01 5.43411970e-01
2.56011844e-01 5.13328075e-01 4.58624244e-01 -1.07149088e+00
-5.24060488e-01 1.01182973e+00 -4.35954362e-01 -5.88623762e-01
-4.25063729e-01 -6.58946276e-01 -9.07723069e-01 5.62666118e-01
1.69274658e-01 -5.31155989e-02 -7.86543787e-01 1.91830277e+00
2.00164407e-01 -2.11876705e-01 -8.13643187e-02 6.26257658e-01
4.30316061e-01 5.23084819e-01 1.18003584e-01 -2.48123690e-01
9.50930834e-01 -6.32291973e-01 -4.10403907e-01 -2.60896713e-01
1.40757143e-01 -1.90299004e-01 8.49494696e-01 -7.98304826e-02
-1.31873310e+00 -3.20469625e-02 -1.36832356e+00 3.14136893e-01
-4.45687711e-01 -5.24214327e-01 6.43699586e-01 9.12933767e-01
-1.36573911e+00 7.39143610e-01 -1.11822319e+00 -5.68082035e-01
5.18660784e-01 5.17468572e-01 5.05489483e-02 5.48558712e-01
-8.93652558e-01 1.24433184e+00 5.34051895e-01 -2.56402522e-01
-4.54089731e-01 -7.35017657e-01 6.31069578e-03 1.41254634e-01
1.08627547e-02 -1.28828430e+00 7.74789691e-01 -6.28530085e-01
-1.97767472e+00 7.09175110e-01 -1.61792010e-01 -2.56230444e-01
7.41570443e-02 6.17948651e-01 4.19162214e-02 9.44040716e-03
-4.10522491e-01 7.60085821e-01 3.07825357e-01 -8.89020622e-01
-1.58623219e-01 -4.43972677e-01 -7.04564691e-01 -9.32433009e-02
-7.15744615e-01 -1.18902035e-01 3.71128052e-01 -4.11909074e-01
3.21150959e-01 -9.45762813e-01 -3.92179042e-01 -1.54843247e-02
1.75978079e-01 -1.59327909e-01 7.15472698e-01 -1.87502399e-01
9.82830584e-01 -1.68741465e+00 6.77231073e-01 1.89056829e-01
4.72672760e-01 1.08620889e-01 7.07413033e-02 2.55663872e-01
2.91793227e-01 2.00935170e-01 -5.14078915e-01 4.60389763e-01
-3.10366005e-01 -3.98556050e-03 1.52754307e-01 3.62339735e-01
1.50977865e-01 1.50515747e+00 -8.80172312e-01 -1.97430521e-01
-1.24487735e-01 4.22912419e-01 -5.51324904e-01 -1.69137672e-01
-1.44917309e-01 5.88281095e-01 -2.68773168e-01 5.89874923e-01
2.89883792e-01 -3.68417889e-01 2.03926861e-01 8.28320235e-02
-4.87541020e-01 -4.94313575e-02 -1.30782157e-01 1.33699811e+00
1.36314956e-02 7.02837825e-01 4.60833982e-02 -9.84573007e-01
1.00956559e+00 -1.30079448e-01 7.83692241e-01 -1.19971418e+00
4.35993224e-01 7.37544358e-01 9.30678904e-01 -3.40237916e-01
2.41044909e-01 -2.82202035e-01 1.84938818e-01 7.22043097e-01
3.26101452e-01 -4.01951194e-01 2.50719547e-01 -7.36006200e-02
1.27319920e+00 1.44751649e-02 1.91581205e-01 -1.11799622e+00
2.05114722e-01 -1.71545953e-01 5.29406548e-01 4.57141668e-01
-5.46013355e-01 2.60914788e-02 2.15540126e-01 -4.39997047e-01
-1.61185348e+00 -1.12776494e+00 -1.21509947e-01 7.29936004e-01
2.92140812e-01 -4.76793982e-02 -9.32887912e-01 3.83991748e-01
-4.64007676e-01 4.16067839e-01 -6.10451519e-01 -8.44903648e-01
-8.50695670e-01 -1.27148604e+00 6.50174916e-01 1.45605817e-01
7.51888812e-01 -1.22884166e+00 -1.49125624e+00 5.12067735e-01
1.74719453e-01 -7.10488856e-01 -2.66299486e-01 4.54195201e-01
-9.28871334e-01 -8.14506471e-01 -5.07834792e-01 -5.28281927e-01
5.36897361e-01 -1.43730670e-01 1.13306570e+00 3.98390442e-01
-8.93946528e-01 -1.61922470e-01 6.68094084e-02 -4.17885602e-01
-3.93945754e-01 2.53688604e-01 2.91183889e-01 -4.87125188e-01
6.12473376e-02 -1.39170873e+00 -9.77503717e-01 1.40319228e-01
-5.58625758e-01 2.35406175e-01 4.04484987e-01 7.65172899e-01
4.26987380e-01 3.39769930e-01 6.84594810e-01 3.35954241e-02
7.27814734e-01 -3.68653804e-01 -4.85273719e-01 5.33949137e-01
-8.61643076e-01 4.78093952e-01 6.31381750e-01 -4.49742943e-01
-8.67262602e-01 -1.69799000e-01 7.37609342e-02 4.00458515e-01
3.85184377e-01 2.62344539e-01 -2.17502173e-02 -4.87906456e-01
7.19905734e-01 7.50893652e-01 1.41034871e-01 1.64029986e-01
-5.38057163e-02 8.36867094e-02 3.57755691e-01 -9.11548972e-01
4.48351711e-01 4.36224103e-01 5.97107410e-01 -1.14155102e+00
-1.43396899e-01 5.37850261e-01 -4.50616717e-01 -6.32987380e-01
8.49547982e-01 -5.03890067e-02 -9.57482934e-01 8.36806655e-01
-9.53133047e-01 -3.45102489e-01 -3.91861647e-01 -3.69353034e-02
-5.13097584e-01 -1.32538840e-01 -4.15249377e-01 -1.01878619e+00
-7.39202082e-01 -8.54099989e-01 3.56609792e-01 8.10930848e-01
-2.26500869e-01 -7.08311558e-01 2.74548948e-01 -1.31528363e-01
9.85868275e-01 5.63751817e-01 1.33060896e+00 6.32198900e-02
-8.22573185e-01 4.29519147e-01 1.52544796e-01 -3.50625992e-01
4.66486849e-02 4.37172353e-01 -5.20222366e-01 -2.04016268e-01
2.19600141e-01 -5.46101332e-02 9.77428138e-01 6.56308115e-01
6.40769541e-01 -6.06696196e-02 -2.40250573e-01 7.01020300e-01
1.53788650e+00 7.66468346e-01 4.69105333e-01 3.48004937e-01
2.17004761e-01 8.13927293e-01 -4.54069495e-01 2.34213084e-01
1.16461165e-01 1.98231667e-01 4.07769978e-01 3.12237889e-01
3.02800518e-02 1.30287465e-02 3.60973924e-01 1.37841368e+00
-6.20050192e-01 -2.78479159e-01 -1.23348951e+00 3.10085565e-01
-1.63376451e+00 -1.11094713e+00 -3.28465924e-02 1.90413129e+00
6.67578161e-01 2.23268270e-01 1.64340660e-01 -1.36237249e-01
6.32992983e-01 -3.16134483e-01 -1.36567128e+00 -4.31638300e-01
-7.59501934e-01 5.54082453e-01 3.32443655e-01 -1.05470270e-01
-7.81026930e-02 8.97803068e-01 7.30986118e+00 5.15909195e-01
-1.20711815e+00 2.95110524e-01 5.35569847e-01 -5.71188569e-01
-4.16261494e-01 -2.29505263e-03 -4.37396735e-01 7.67919898e-01
1.32625365e+00 -6.27456486e-01 8.33070576e-01 1.26166195e-01
2.07406506e-02 -7.56310374e-02 -6.36851966e-01 7.06631720e-01
-5.24789512e-01 -1.77204549e+00 -1.92192227e-01 5.51445067e-01
8.88004005e-01 3.95233840e-01 1.87436163e-01 -3.45130920e-01
2.19691798e-01 -1.15070128e+00 9.39550996e-01 7.51944542e-01
6.38208270e-01 -9.00007546e-01 1.54410839e-01 3.80012929e-01
-1.23998296e+00 -4.37526733e-01 -2.94212699e-01 -7.57273361e-02
4.18886244e-01 5.52871346e-01 -4.99936342e-02 -2.35777199e-01
9.50726628e-01 3.75562787e-01 -4.75269437e-01 1.11309588e+00
2.53333271e-01 2.81028956e-01 -4.51171607e-01 -7.72833645e-01
6.02743067e-02 -5.78763068e-01 8.35774601e-01 6.99723125e-01
5.66699505e-01 4.08898592e-01 -6.58776700e-01 1.75746655e+00
-9.94837284e-02 -2.88696170e-01 -6.22757792e-01 -5.55603623e-01
5.18728852e-01 1.06472731e+00 -1.47259378e+00 7.09021091e-02
2.75071532e-01 5.66400945e-01 2.44645029e-01 4.47932631e-02
-8.58477294e-01 -4.91443455e-01 7.98014879e-01 1.37417585e-01
8.19788054e-02 -7.23381996e-01 -7.98466563e-01 -8.70968521e-01
-4.15079743e-02 -4.04115230e-01 -3.51223022e-01 -4.75423008e-01
-8.32445920e-01 7.46787786e-01 -6.86155021e-01 -3.75750095e-01
2.16483131e-01 -4.49733078e-01 -9.18588042e-01 5.20751059e-01
-1.19106412e+00 -6.52459025e-01 -1.49826601e-01 2.10229278e-01
2.96669364e-01 -3.45023125e-01 5.86906075e-01 -2.23942578e-01
-7.52483428e-01 3.74786109e-01 4.34046656e-01 -5.45442045e-01
-1.86658263e-01 -8.74925435e-01 6.63452983e-01 8.09605479e-01
-3.41003388e-01 5.19592404e-01 8.47423494e-01 -6.73301220e-01
-1.51553774e+00 -4.48071390e-01 6.06724322e-01 -2.75266081e-01
6.53175592e-01 -4.71926451e-01 -8.21411312e-01 -1.80132240e-01
3.51084918e-01 -4.40302342e-01 6.26822472e-01 -3.71234387e-01
1.60151497e-02 2.66815990e-01 -1.24253535e+00 1.02130878e+00
1.77495551e+00 -3.37653071e-01 -3.71913016e-01 -9.59689021e-02
5.80407381e-01 3.60436350e-01 -7.71619380e-01 4.98220712e-01
8.29283476e-01 -1.07862747e+00 8.17880452e-01 -4.56134617e-01
5.23841262e-01 -1.91957235e-01 1.94990858e-01 -1.25300121e+00
-6.67949677e-01 -8.00968468e-01 -2.18129411e-01 7.94636250e-01
4.06444073e-01 -9.92058992e-01 7.47223437e-01 6.74508572e-01
-2.48719737e-01 -9.91647243e-01 -1.20455301e+00 -8.91079485e-01
4.29215670e-01 3.24716151e-01 5.62694132e-01 5.14974415e-01
7.86515139e-03 1.96017250e-01 2.76991457e-01 -4.19635355e-01
7.92883396e-01 -7.48138595e-03 -8.67665410e-02 -1.44964492e+00
-1.06940640e-03 -1.29251122e+00 -2.63764411e-01 -2.03362972e-01
2.05706134e-01 -7.91137159e-01 3.60234529e-02 -1.10480559e+00
4.50744808e-01 -3.70587200e-01 -3.75747383e-01 4.26422060e-02
2.01168746e-01 -3.24552841e-02 1.06562883e-01 4.20932472e-01
-3.49521726e-01 9.64307189e-01 1.28553176e+00 3.24153244e-01
-2.33570695e-01 -5.90164542e-01 -7.95808077e-01 4.69418377e-01
1.09237683e+00 -5.80240309e-01 -3.52818877e-01 -2.48952508e-01
7.25910723e-01 -3.00036818e-01 4.14169356e-02 -1.27716804e+00
5.68210840e-01 -4.01543528e-01 1.25723764e-01 4.98391502e-02
2.57475466e-01 -4.07560706e-01 5.53098381e-01 1.23819947e+00
-1.55439392e-01 1.89107150e-01 -9.88759995e-02 4.21324819e-01
3.37062925e-01 -3.12707573e-02 1.30467343e+00 -3.04770440e-01
-4.24275935e-01 2.54968196e-01 -8.27218413e-01 1.61792934e-01
1.12085915e+00 -8.98220301e-01 -6.21771753e-01 3.06356579e-01
-2.91275412e-01 -8.40264931e-02 8.00134718e-01 1.97821677e-01
5.92774808e-01 -9.88469303e-01 -5.23568392e-01 1.33428529e-01
-3.25896889e-01 -4.37236220e-01 1.33972913e-01 7.29978919e-01
-7.98331261e-01 4.40071464e-01 -1.09754205e+00 -4.40713048e-01
-7.89221764e-01 1.83833435e-01 7.97291338e-01 -6.28920868e-02
-4.00634080e-01 1.06055832e+00 -8.99017006e-02 -2.24887550e-01
-4.37263191e-01 1.17904320e-01 7.45498762e-02 -3.98265421e-01
2.12464318e-01 5.71416676e-01 -1.47496030e-01 -5.41995466e-01
-4.28364009e-01 7.30950534e-01 4.78741169e-01 -1.70534819e-01
1.90918243e+00 -7.30044842e-02 -7.85931945e-01 1.92715824e-01
7.07817018e-01 -5.50819039e-01 -1.18104517e+00 2.64381379e-01
3.15310329e-01 6.21735752e-02 1.61920525e-02 -7.26611793e-01
-1.25243950e+00 7.22144663e-01 5.31197071e-01 2.95777749e-02
1.19419277e+00 1.49956234e-02 1.10230184e+00 5.04904270e-01
7.81398058e-01 -1.28938532e+00 5.37303448e-01 6.81633711e-01
6.55372858e-01 -6.77848041e-01 -3.50141793e-01 4.30447794e-02
4.24494632e-02 1.16046023e+00 9.24221933e-01 -2.40108624e-01
7.53133595e-01 5.78326166e-01 -7.20435262e-01 -4.41944897e-01
-1.11919868e+00 5.36120869e-02 -3.00043188e-02 5.23930848e-01
4.81776118e-01 8.03213567e-02 -6.60248935e-01 3.19267541e-01
-4.94374484e-01 5.50010391e-02 4.20775682e-01 1.12200499e+00
-1.12973714e+00 -1.04095566e+00 5.11779264e-02 4.80733246e-01
-1.88225314e-01 -1.03893459e-01 -6.86075032e-01 2.05910847e-01
2.26302758e-01 4.15632099e-01 -2.86551062e-02 -4.96184647e-01
-8.50609541e-02 1.49489626e-01 9.05388713e-01 -6.68515712e-02
-6.78670287e-01 -4.82036889e-01 -4.49991286e-01 -2.12229103e-01
-2.86197186e-01 -6.25909030e-01 -1.59390700e+00 -7.45982468e-01
-2.69609988e-01 -2.44004950e-02 9.34329331e-01 7.56737947e-01
7.56341815e-01 5.64175963e-01 4.15885538e-01 -7.91574538e-01
-7.26671293e-02 -3.84859800e-01 -5.15331984e-01 -1.16301589e-01
-6.63922131e-02 -6.80459261e-01 -2.08677202e-01 -2.25557879e-01] | [8.038875579833984, 2.78082013130188] |
1ee5a48c-9465-49b7-81cb-0335fd911b40 | influence-of-various-text-embeddings-on | 2305.03144 | null | https://arxiv.org/abs/2305.03144v1 | https://arxiv.org/pdf/2305.03144v1.pdf | Influence of various text embeddings on clustering performance in NLP | With the advent of e-commerce platforms, reviews are crucial for customers to assess the credibility of a product. The star ratings do not always match the review text written by the customer. For example, a three star rating (out of five) may be incongruous with the review text, which may be more suitable for a five star review. A clustering approach can be used to relabel the correct star ratings by grouping the text reviews into individual groups. In this work, we explore the task of choosing different text embeddings to represent these reviews and also explore the impact the embedding choice has on the performance of various classes of clustering algorithms. We use contextual (BERT) and non-contextual (Word2Vec) text embeddings to represent the text and measure their impact of three classes on clustering algorithms - partitioning based (KMeans), single linkage agglomerative hierarchical, and density based (DBSCAN and HDBSCAN), each with various experimental settings. We use the silhouette score, adjusted rand index score, and cluster purity score metrics to evaluate the performance of the algorithms and discuss the impact of different embeddings on the clustering performance. Our results indicate that the type of embedding chosen drastically affects the performance of the algorithm, the performance varies greatly across different types of clustering algorithms, no embedding type is better than the other, and DBSCAN outperforms KMeans and single linkage agglomerative clustering but also labels more data points as outliers. We provide a thorough comparison of the performances of different algorithms and provide numerous ideas to foster further research in the domain of text clustering. | ['Rohan Saha'] | 2023-05-04 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [-5.70499599e-01 -2.23089412e-01 -1.59625351e-01 -5.51216662e-01
-4.49773401e-01 -7.48369455e-01 6.31681740e-01 8.91858101e-01
-4.90339696e-01 -1.09701499e-01 5.20423234e-01 -4.19708073e-01
-5.62601447e-01 -6.74156427e-01 -7.87914917e-02 -7.63247132e-01
2.57939756e-01 7.35700428e-01 1.87780544e-01 -6.09966666e-02
5.33160210e-01 3.44251841e-01 -1.50870740e+00 2.49936685e-01
8.96827281e-01 8.30286860e-01 -9.35915783e-02 5.02596140e-01
-4.05580729e-01 2.75460988e-01 -7.13419259e-01 -7.41164327e-01
2.19900817e-01 -3.18185270e-01 -4.21436697e-01 2.94307649e-01
-8.03470239e-02 2.60315001e-01 1.47836223e-01 1.05986130e+00
3.73564988e-01 1.52813911e-01 1.07313383e+00 -1.21376157e+00
-8.79280746e-01 9.46854532e-01 -7.88221776e-01 -9.30765644e-02
1.61516607e-01 -3.82961333e-01 1.31856179e+00 -1.02892721e+00
4.90178645e-01 1.22495139e+00 6.74683392e-01 -9.93205328e-03
-1.20087206e+00 -3.02933127e-01 1.66257322e-01 1.66646197e-01
-1.61503947e+00 -1.80754840e-01 4.79317278e-01 -5.80251038e-01
4.69802082e-01 2.63460308e-01 5.36696196e-01 4.89769876e-01
6.94588348e-02 1.66381747e-01 1.08989298e+00 -4.56449986e-01
7.04553962e-01 7.65042961e-01 5.20212650e-01 1.27754778e-01
8.13964248e-01 -4.41666096e-01 -1.57544702e-01 -4.62400407e-01
1.00336060e-01 1.69098809e-01 -7.69319832e-02 -6.88205838e-01
-9.51609910e-01 1.39458513e+00 3.74456167e-01 4.65584636e-01
-2.13390216e-01 -2.05756202e-01 4.06641811e-01 2.09610686e-01
4.09190267e-01 6.55643523e-01 -3.07331920e-01 1.02931917e-01
-1.00630915e+00 4.46910635e-02 8.09185028e-01 5.15838265e-01
6.48546934e-01 -2.11347759e-01 1.04904817e-02 1.15722775e+00
5.03565252e-01 5.71183302e-02 6.91223264e-01 -7.10731506e-01
-2.91379984e-03 6.91272020e-01 2.07769975e-01 -1.42654467e+00
-4.00387168e-01 -7.87514895e-02 -4.61038291e-01 1.73214361e-01
5.14043927e-01 -1.92344368e-01 -7.25502610e-01 8.96441817e-01
1.84132352e-01 -2.42630154e-01 -5.10840714e-02 9.96847332e-01
6.67353332e-01 7.03857303e-01 -1.28993392e-01 -2.65494794e-01
1.22024560e+00 -7.75601387e-01 -6.93341315e-01 7.43158534e-02
8.86693120e-01 -1.01458311e+00 9.68574762e-01 5.55929065e-01
-5.94382703e-01 -5.01195788e-01 -1.18257403e+00 3.22053850e-01
-6.61103666e-01 -7.27667138e-02 3.75344157e-01 8.50725412e-01
-9.55759168e-01 6.01771116e-01 -5.20394027e-01 -4.83347327e-01
-7.63585046e-02 2.55978316e-01 -2.34975621e-01 -1.83495149e-01
-9.91915524e-01 9.67994869e-01 2.15002880e-01 -2.59318650e-01
-6.34283051e-02 -1.75883785e-01 -7.54220545e-01 1.26517028e-01
-4.18095253e-02 -1.57347023e-01 5.61943054e-01 -1.12858880e+00
-8.89509022e-01 6.64408803e-01 2.23961566e-02 -2.73607314e-01
1.34867460e-01 1.65138319e-01 -7.55768418e-01 7.06981122e-02
1.24798216e-01 5.99423110e-01 5.57454765e-01 -1.51185894e+00
-5.30441642e-01 -5.63209772e-01 -2.51426339e-01 2.69305497e-01
-4.85652685e-01 2.45455921e-01 -4.74933714e-01 -5.06689966e-01
1.85953826e-01 -1.00524592e+00 -3.99036467e-01 -4.71441120e-01
-1.67271197e-01 -5.33668697e-01 5.47094762e-01 -2.94291645e-01
1.54381502e+00 -2.39963198e+00 -1.16810217e-01 5.30323684e-01
1.46059781e-01 8.55101123e-02 6.93781599e-02 4.49866205e-01
-9.56142172e-02 5.88671148e-01 -1.59953088e-02 -1.84956044e-01
1.56944007e-01 1.45317301e-01 6.27484843e-02 6.00231230e-01
-1.12186782e-01 5.04763722e-01 -8.37412000e-01 -5.47596753e-01
1.31029323e-01 5.90831518e-01 -3.65351349e-01 -2.42897078e-01
1.37289122e-01 -1.85072705e-01 -1.91469550e-01 2.78541535e-01
4.93746847e-01 -2.22949654e-01 4.56581086e-01 -1.29858404e-01
-9.46718454e-02 3.09593886e-01 -1.46995175e+00 9.39168274e-01
-1.37038708e-01 6.36252284e-01 -5.68638071e-02 -8.23317111e-01
1.20598733e+00 1.67395726e-01 5.32480240e-01 -1.38943478e-01
2.41084322e-01 -1.26463454e-02 2.69448936e-01 -9.63040814e-02
8.79696608e-01 -1.33191466e-01 -1.61274999e-01 8.17104042e-01
-3.58632982e-01 4.66886088e-02 2.43774638e-01 4.89209980e-01
8.67105842e-01 -6.37020111e-01 1.19297139e-01 -4.36703116e-01
1.19281106e-01 2.33611345e-01 4.61135536e-01 3.20028752e-01
-2.51441151e-01 8.62765729e-01 8.24819982e-01 -2.06402674e-01
-9.70740557e-01 -1.07009625e+00 -2.96022385e-01 9.56049025e-01
3.45609128e-01 -7.98550189e-01 -5.23886502e-01 -8.12161148e-01
2.28436589e-01 9.87587154e-01 -7.05388188e-01 -1.24855861e-01
7.27109909e-02 -1.01104915e+00 3.75025272e-02 4.35660958e-01
-3.90925974e-01 -5.81925631e-01 -1.48043320e-01 -1.72018826e-01
4.20521870e-02 -8.16210270e-01 -5.01990259e-01 3.24903727e-01
-6.88364863e-01 -1.06755149e+00 -2.65597284e-01 -8.10605586e-01
8.70027661e-01 4.90169495e-01 9.52413082e-01 8.54585543e-02
2.43190974e-01 4.08665985e-01 -1.02040219e+00 -2.09661826e-01
-3.65542084e-01 -1.23255976e-01 3.28587949e-01 1.77976847e-01
9.99029517e-01 -8.14554989e-02 -4.99500155e-01 5.54311275e-01
-1.03869081e+00 -8.59865785e-01 2.79476851e-01 4.45919186e-01
4.77884948e-01 7.48793364e-01 4.60805625e-01 -1.06889534e+00
1.14693511e+00 -6.68807328e-01 -1.11077875e-01 1.18160695e-01
-1.16215026e+00 3.21095362e-02 6.25127673e-01 -4.03486103e-01
-4.44877028e-01 -8.92947689e-02 1.84680536e-01 -3.95625979e-01
1.04033619e-01 6.51992798e-01 7.62859657e-02 4.46272999e-01
7.04848349e-01 -4.37238902e-01 -9.57434895e-06 -3.01063597e-01
5.18078804e-01 1.12249780e+00 -1.54965743e-01 -7.98942745e-02
5.59127092e-01 4.60435659e-01 -7.50918150e-01 -7.76874542e-01
-3.79985631e-01 -8.48776817e-01 -7.47986734e-01 -8.87911245e-02
1.07296777e+00 -6.15825415e-01 -1.22043744e-01 -2.05495596e-01
-7.57245123e-01 1.87121645e-01 -2.06689805e-01 6.25695229e-01
1.43595114e-01 5.70532978e-01 -3.95273715e-01 -5.70984483e-01
5.04705310e-02 -1.39402068e+00 4.86407220e-01 5.53864092e-02
-7.91330457e-01 -1.14004838e+00 9.24153775e-02 4.37897027e-01
4.62250821e-02 8.04338306e-02 1.08278668e+00 -1.20529687e+00
1.24193296e-01 -3.09359223e-01 -6.92166314e-02 3.64305854e-01
1.62276566e-01 5.75928867e-01 -6.63716674e-01 -1.96100846e-01
-6.43209219e-02 1.50516540e-01 8.54932129e-01 6.03854954e-01
5.00307024e-01 -5.14216907e-02 -3.51245582e-01 4.77465540e-02
1.46549308e+00 2.80600190e-01 6.47915304e-01 3.80958468e-01
7.05281377e-01 9.30866718e-01 7.57192433e-01 4.09652382e-01
4.11368400e-01 4.29209799e-01 2.85382330e-01 1.03821762e-01
4.28882301e-01 2.21445393e-02 6.67037666e-01 1.24183249e+00
5.59609711e-01 -2.72956163e-01 -8.59679043e-01 7.10622489e-01
-1.62137580e+00 -7.87544489e-01 -4.36698496e-01 2.33612585e+00
3.37889105e-01 2.07431972e-01 4.76387382e-01 5.30039668e-01
9.78532016e-01 1.03851907e-01 -1.65478215e-01 -1.13705516e+00
6.93792328e-02 -7.46498629e-02 6.31686091e-01 5.36786556e-01
-7.96751022e-01 6.49397910e-01 6.05485439e+00 5.97461641e-01
-8.86026740e-01 2.00233539e-03 8.00541997e-01 -1.14646599e-01
-6.57098234e-01 5.28092170e-03 -7.12213337e-01 6.32389188e-01
1.09379172e+00 -9.08204541e-02 4.47621346e-02 8.37758899e-01
3.64916921e-01 -3.11960787e-01 -8.78716230e-01 7.70164371e-01
2.82800198e-01 -1.00476289e+00 -7.45003372e-02 1.67457029e-01
8.41682315e-01 9.57050081e-03 9.94415805e-02 8.97544716e-03
8.94359708e-01 -1.07400036e+00 3.07814986e-01 4.62126397e-02
4.26241070e-01 -1.10425746e+00 9.08586442e-01 -2.03107465e-02
-9.34403598e-01 -1.91340044e-01 -5.50511479e-01 1.30983308e-01
9.35703292e-02 8.61011982e-01 -6.41249120e-01 3.24409425e-01
9.80241358e-01 5.82534194e-01 -8.50058854e-01 8.15903544e-01
8.29804689e-02 4.78267282e-01 -2.17212379e-01 -4.10877407e-01
3.14984053e-01 -6.21709287e-01 2.29570985e-01 1.12966955e+00
3.25053066e-01 -1.33197144e-01 4.06838804e-02 5.42280972e-01
1.33021727e-01 3.96004468e-01 -3.83065701e-01 -2.01693162e-01
8.01610887e-01 1.34489274e+00 -1.27382052e+00 -2.15644717e-01
-5.52318513e-01 6.61050081e-01 5.65459905e-03 2.56241947e-01
-6.86450720e-01 -5.67941487e-01 8.06819737e-01 4.91007566e-01
6.20608747e-01 -2.32183963e-01 -6.83822393e-01 -7.83241332e-01
-1.93580851e-01 -7.84642577e-01 5.41678369e-01 -7.68109262e-01
-1.41009581e+00 5.62647343e-01 -2.42822692e-01 -1.22367406e+00
1.06758311e-01 -5.40688694e-01 -7.28834569e-01 4.25867230e-01
-8.91308010e-01 -2.32058719e-01 -1.93459599e-03 1.62255988e-01
2.26920992e-01 -7.93112144e-02 4.65244085e-01 1.31321281e-01
-4.36044306e-01 4.84625489e-01 7.07448244e-01 2.52133578e-01
1.12561524e+00 -1.42855012e+00 8.70129745e-03 5.00787258e-01
3.70389014e-01 7.63805568e-01 7.83753395e-01 -5.72163701e-01
-8.63165915e-01 -9.91528988e-01 1.01624572e+00 -6.25817239e-01
8.55502844e-01 -1.25454977e-01 -8.21174681e-01 3.07668000e-01
3.77011597e-01 -4.12653685e-01 1.13234663e+00 5.22879422e-01
-4.58365023e-01 -2.22392380e-01 -1.04902446e+00 6.51166856e-01
2.10755169e-01 -2.00091094e-01 -5.24290800e-01 1.28428578e-01
7.77242362e-01 4.69045162e-01 -1.21380353e+00 1.05178714e-01
4.84958500e-01 -1.21656883e+00 6.76823854e-01 -2.49473095e-01
4.05115932e-01 -5.95434070e-01 -1.77158952e-01 -1.60550773e+00
-7.30008602e-01 1.30668879e-01 4.80883390e-01 1.39272749e+00
7.41119266e-01 -4.92786288e-01 7.48555124e-01 7.73283243e-01
1.56520754e-01 -8.85482073e-01 -4.82793957e-01 -5.00751317e-01
2.96625525e-01 -3.90679091e-01 5.53486943e-01 1.25011110e+00
4.43742901e-01 6.61757290e-01 7.86692351e-02 1.03405058e-01
1.79493442e-01 1.09087877e-01 6.49063826e-01 -1.44043767e+00
8.25035013e-03 -6.59739077e-01 -5.24234354e-01 -5.81096470e-01
-6.25251010e-02 -9.63945568e-01 -2.08161250e-01 -1.74295568e+00
-1.49200737e-01 -6.25442147e-01 -4.63178277e-01 -6.19911999e-02
-2.19146147e-01 1.33013815e-01 2.88101733e-01 3.12720001e-01
-6.81385338e-01 1.46516517e-01 5.64268410e-01 -6.11126050e-02
-4.51011002e-01 -8.49590749e-02 -1.13028312e+00 6.20946765e-01
8.08879495e-01 -6.03029966e-01 -4.24445361e-01 -2.62449563e-01
5.51677287e-01 -3.83242816e-01 -2.74317026e-01 -6.88829362e-01
3.41118217e-01 1.43753380e-01 4.45000887e-01 -4.70742345e-01
3.01859695e-02 -1.02141047e+00 9.91269872e-02 9.13978890e-02
-4.18414712e-01 6.54902458e-01 -2.27428094e-01 6.97740316e-01
-2.41243884e-01 -5.65708280e-01 7.84810483e-01 6.17533438e-02
-2.51055211e-01 -1.80586755e-01 -6.14814103e-01 -6.43690452e-02
1.02485001e+00 -5.61269522e-01 3.10937725e-02 -4.37735200e-01
-7.90662825e-01 3.39565158e-01 1.01518703e+00 6.00699842e-01
4.93100911e-01 -1.31725323e+00 -5.98683774e-01 -6.66330531e-02
2.57047176e-01 -3.29349637e-01 -3.20054799e-01 7.74621427e-01
-5.22064090e-01 1.66736841e-01 2.93351144e-01 -5.49971282e-01
-1.36323178e+00 7.87786365e-01 -3.49886715e-02 -1.31742239e-01
-1.81752965e-01 5.06490707e-01 -3.20783630e-02 -5.03238797e-01
1.60293475e-01 -1.44886717e-01 -6.68207765e-01 7.18784034e-01
3.60802323e-01 7.00164914e-01 1.17181197e-01 -8.92823994e-01
-3.68396759e-01 5.86397171e-01 -2.19430536e-01 -2.42772236e-01
1.16890216e+00 -2.48233691e-01 -1.97108015e-01 7.80967355e-01
1.15439773e+00 1.25150323e-01 -6.60243690e-01 3.11097614e-02
3.13616484e-01 -4.28562194e-01 1.41766682e-01 -4.72138137e-01
-1.03982973e+00 6.07867718e-01 4.69057918e-01 8.68740439e-01
7.08016336e-01 5.20518012e-02 5.30397058e-01 -1.45075008e-01
-1.29607871e-01 -1.58864105e+00 -1.33345434e-02 1.62543222e-01
2.59012312e-01 -1.23978150e+00 3.49450022e-01 -2.00690255e-01
-1.34406185e+00 9.29102004e-01 3.01875591e-01 -2.43582085e-01
1.01732719e+00 -8.44434369e-03 4.46416765e-01 -3.69239718e-01
-6.30781293e-01 -2.31778219e-01 1.90387517e-01 4.64099377e-01
5.92289567e-01 2.35494107e-01 -5.89886487e-01 5.06792307e-01
-2.42088795e-01 -7.11324394e-01 7.77716815e-01 4.09532905e-01
-5.36759257e-01 -1.10159552e+00 -5.92326760e-01 8.90194118e-01
-5.35146296e-01 -1.52635120e-03 -9.40592408e-01 6.49325132e-01
2.21436799e-01 1.52631414e+00 3.91859144e-01 -6.79089904e-01
1.38132706e-01 1.03792086e-01 -1.46096572e-01 -7.65329659e-01
-7.33833313e-01 3.06344926e-01 7.22269267e-02 9.81547683e-02
-3.88385355e-01 -8.75603437e-01 -1.31884730e+00 -6.02284670e-01
-8.06666136e-01 7.35153615e-01 7.82302737e-01 6.03762865e-01
5.10991633e-01 3.02694775e-02 7.89247036e-01 -3.85327607e-01
-2.42184848e-01 -7.92633832e-01 -8.56704473e-01 6.39469683e-01
-1.58341363e-01 -5.67567766e-01 -8.44257712e-01 -1.26864463e-01] | [10.374053955078125, 7.254085540771484] |
3d62d4e3-0de1-48c4-99df-66ad10d56118 | hylda-end-to-end-hybrid-learning-domain | 2201.05585 | null | https://arxiv.org/abs/2201.05585v2 | https://arxiv.org/pdf/2201.05585v2.pdf | Domain Adaptation in LiDAR Semantic Segmentation via Alternating Skip Connections and Hybrid Learning | In this paper we address the challenging problem of domain adaptation in LiDAR semantic segmentation. We consider the setting where we have a fully-labeled data set from source domain and a target domain with a few labeled and many unlabeled examples. We propose a domain adaption framework that mitigates the issue of domain shift and produces appealing performance on the target domain. To this end, we develop a GAN-based image-to-image translation engine that has generators with alternating connections, and couple it with a state-of-the-art LiDAR semantic segmentation network. Our framework is hybrid in nature in the sense that our model learning is composed of self-supervision, semi-supervision and unsupervised learning. Extensive experiments on benchmark LiDAR semantic segmentation data sets demonstrate that our method achieves superior performance in comparison to strong baselines and prior arts. | ['Liu Bingbing', 'Yang Liu', 'Shubhra Aich', 'Yannis Y. He', 'Mrigank Rochan', 'Eduardo R. Corral-Soto'] | 2022-01-14 | null | null | null | null | ['lidar-semantic-segmentation'] | ['computer-vision'] | [ 6.62244022e-01 5.06934404e-01 -4.77231205e-01 -7.96490848e-01
-1.13661301e+00 -6.66628003e-01 5.98806679e-01 -5.08953035e-01
-4.22843516e-01 8.07864368e-01 -1.96785048e-01 -1.96603760e-01
4.38519299e-01 -8.68555784e-01 -1.01006794e+00 -3.68813485e-01
6.50711000e-01 1.14647377e+00 3.74670357e-01 -4.00082693e-02
-1.74723789e-01 1.98556632e-01 -1.10452604e+00 -8.31232220e-02
1.28778005e+00 9.02122080e-01 2.15944156e-01 1.98068395e-01
-3.85272264e-01 4.58717406e-01 -2.56969661e-01 -5.81074715e-01
5.65370977e-01 -5.05682945e-01 -9.62923944e-01 5.47195911e-01
9.30793166e-01 -2.81601727e-01 -8.85512903e-02 1.07989728e+00
3.67312104e-01 -9.52470116e-03 8.82270873e-01 -1.44228363e+00
-7.82361031e-01 4.72329527e-01 -8.67248893e-01 -5.00966050e-02
-2.63702512e-01 2.14229152e-01 8.68303359e-01 -9.71478343e-01
8.21794748e-01 1.40354502e+00 5.92134058e-01 9.26251709e-01
-1.68967450e+00 -9.18630540e-01 3.10523719e-01 -1.26908362e-01
-1.22704852e+00 -3.10576290e-01 8.60217094e-01 -5.88465214e-01
5.41544199e-01 -6.39260471e-01 3.86865228e-01 1.36847651e+00
-4.78392869e-01 7.95506120e-01 1.06037033e+00 -2.68357605e-01
4.23723429e-01 1.14520885e-01 -2.46362742e-02 3.90862912e-01
1.67094126e-01 2.38335699e-01 -4.02841479e-01 1.63431093e-01
8.26967180e-01 -3.57988477e-01 2.93588072e-01 -7.86285937e-01
-7.95107782e-01 9.39788699e-01 6.04777217e-01 -1.51894033e-01
-6.64378852e-02 1.38946295e-01 1.58703864e-01 1.89060077e-01
8.21972668e-01 1.09764114e-01 -5.08875132e-01 4.00237560e-01
-1.03294313e+00 2.90226251e-01 5.88371396e-01 1.47735620e+00
1.13797200e+00 2.43223622e-01 1.10258229e-01 8.81611586e-01
1.28922343e-01 8.40790153e-01 1.43655062e-01 -1.24937081e+00
6.04105711e-01 4.30112541e-01 -7.88907483e-02 -1.17515162e-01
7.93995485e-02 -2.16226757e-01 -6.23361945e-01 1.93068326e-01
3.26016635e-01 -1.73758700e-01 -1.35324693e+00 2.02284169e+00
4.68338758e-01 5.58063745e-01 1.85272053e-01 7.23501682e-01
6.43968463e-01 3.32436174e-01 4.80222911e-01 3.06205928e-01
1.01425612e+00 -1.17579055e+00 -3.40470046e-01 -9.13128495e-01
1.29788399e-01 -3.90409857e-01 1.24778259e+00 -3.70528065e-02
-1.05698037e+00 -9.33922768e-01 -9.49838340e-01 -3.06793898e-01
-3.64770591e-01 -6.19419711e-03 2.74699599e-01 5.12020767e-01
-1.06027126e+00 3.75393808e-01 -7.13320374e-01 -6.18270695e-01
8.73662055e-01 2.81956553e-01 -1.83910087e-01 -2.15708107e-01
-7.58326471e-01 6.98753357e-01 5.91459990e-01 -4.54006970e-01
-1.02968180e+00 -9.02577162e-01 -1.04557991e+00 -3.07743520e-01
5.58494687e-01 -1.12509465e+00 1.72931170e+00 -1.23122048e+00
-1.42243898e+00 1.38532507e+00 -2.18451917e-01 -6.05769515e-01
6.58515811e-01 -2.29941070e-01 -1.88650101e-01 2.01728493e-01
8.16852570e-01 1.49811053e+00 1.05215359e+00 -1.49055731e+00
-8.34780991e-01 -4.88550365e-01 -2.97208697e-01 2.44765103e-01
5.15437573e-02 -4.52699333e-01 -4.31944311e-01 -6.15814567e-01
-2.53754407e-02 -1.04854321e+00 -3.29615682e-01 8.87052622e-03
-3.66596311e-01 -2.04788998e-01 9.82096732e-01 -2.39957631e-01
3.02632093e-01 -2.23287320e+00 -5.61946519e-02 -5.57536783e-04
2.82263067e-02 1.15787670e-01 -3.60428512e-01 7.55613074e-02
-1.47477821e-01 1.33692399e-01 -8.42913270e-01 -5.84120214e-01
-5.30093908e-02 5.83327353e-01 -6.16740704e-01 1.74572483e-01
7.95470655e-01 1.25260806e+00 -9.70886230e-01 -8.51390660e-01
7.53103569e-02 1.68029487e-01 -6.04595304e-01 4.43851203e-01
-7.03930080e-01 1.03991592e+00 -6.25172675e-01 6.75676644e-01
1.00607741e+00 -3.13584208e-01 4.96564731e-02 1.35769369e-02
1.73281461e-01 3.90264899e-01 -7.72529185e-01 2.36617422e+00
-4.26864207e-01 3.60064447e-01 8.85745510e-02 -1.03494823e+00
1.01318252e+00 -2.36527752e-02 4.04497176e-01 -7.65074849e-01
-4.91563156e-02 2.51951009e-01 -3.21452618e-01 -1.04720943e-01
3.20302904e-01 -4.16927129e-01 -2.34822452e-01 4.47142571e-01
5.35294175e-01 -8.03844392e-01 1.65397331e-01 2.56147444e-01
7.73109734e-01 6.17798626e-01 -6.98321592e-03 -2.21330628e-01
2.46274307e-01 3.17887425e-01 7.34220922e-01 5.72637856e-01
-2.11078510e-01 7.76144326e-01 2.09053621e-01 6.11640811e-02
-1.20855069e+00 -1.54532290e+00 -1.10814296e-01 1.07640827e+00
2.66588330e-01 8.56976956e-02 -1.05105019e+00 -1.13788974e+00
1.86020136e-01 8.12785387e-01 -4.44553912e-01 -1.53902527e-02
-5.97751975e-01 -1.46853477e-01 5.37790060e-01 8.80033195e-01
1.05577493e+00 -1.00103557e+00 -2.58052677e-01 3.03782895e-02
-2.10504562e-01 -1.73218536e+00 -5.33284664e-01 2.32398778e-01
-1.09186351e+00 -9.13906217e-01 -5.68383932e-01 -1.12310493e+00
6.44501865e-01 2.84434497e-01 1.53395677e+00 -5.12809634e-01
3.82947102e-02 3.58900905e-01 -6.62629083e-02 -5.55603564e-01
-4.00857002e-01 4.08814907e-01 -3.25261235e-01 -1.98075920e-01
5.53100646e-01 -7.81872094e-01 -4.76677865e-01 2.67547429e-01
-1.05053210e+00 8.94112326e-03 5.88455021e-01 7.76663423e-01
8.67363214e-01 -3.86725843e-01 7.32594788e-01 -1.36362600e+00
2.17441678e-01 -4.89971697e-01 -7.53794909e-01 -1.36330808e-02
-6.94036245e-01 6.83238208e-02 4.77035880e-01 -1.94872603e-01
-1.35355580e+00 6.01083100e-01 1.58813279e-02 -5.36940336e-01
-5.37856281e-01 1.01376045e-02 -5.88809848e-01 -7.71875307e-02
6.75112724e-01 -7.03864405e-03 -9.92466807e-02 -5.90373755e-01
9.49523389e-01 5.84822774e-01 1.13375556e+00 -9.55734968e-01
1.28628933e+00 7.18805790e-01 -1.72773883e-01 -5.56777596e-01
-1.02975452e+00 -4.13419425e-01 -1.14754581e+00 2.00991452e-01
1.00489581e+00 -1.46932209e+00 1.04611531e-01 4.04122084e-01
-1.15538645e+00 -7.69719839e-01 -8.15711379e-01 -9.80047509e-02
-8.66759956e-01 2.22591683e-01 -2.98299015e-01 -3.93292248e-01
-5.94604947e-02 -1.01728499e+00 1.54778337e+00 2.66039819e-01
4.57591051e-03 -9.32732761e-01 1.39550880e-01 5.19822717e-01
1.31638616e-01 2.45690092e-01 7.30454862e-01 -6.67156398e-01
-7.75341690e-01 4.30748045e-01 -5.66189885e-01 6.73572779e-01
8.18595514e-02 -4.10639316e-01 -1.15897846e+00 -1.06503248e-01
-1.30610436e-01 -7.45222628e-01 1.09622216e+00 2.74522126e-01
1.18197656e+00 8.33261386e-02 -4.34391469e-01 8.27184319e-01
1.40764499e+00 -8.71068053e-03 4.87260103e-01 1.51618049e-01
7.97112823e-01 7.53775477e-01 8.39194357e-01 -8.89975652e-02
7.43372858e-01 5.02997041e-01 3.59119773e-01 -5.28192341e-01
-4.33269709e-01 -7.08989203e-01 2.67055687e-02 2.37964466e-01
4.53745574e-01 -2.39848286e-01 -9.73594725e-01 9.22422588e-01
-1.89526606e+00 -5.53276479e-01 1.12283520e-01 1.83668339e+00
9.49706852e-01 1.53312489e-01 3.90500247e-01 -4.26788896e-01
7.62702584e-01 8.52695853e-02 -1.01335835e+00 -2.22122893e-01
-1.05791748e-01 7.55566239e-01 7.65904307e-01 4.15754884e-01
-1.33464122e+00 1.64452422e+00 6.45022106e+00 7.10615873e-01
-8.54420424e-01 3.77968103e-01 6.40197277e-01 2.49376282e-01
-3.33356589e-01 4.45924401e-02 -7.29606032e-01 3.52814049e-01
7.80538499e-01 -7.03186467e-02 2.22748429e-01 1.09168518e+00
-9.86658931e-02 9.57313459e-03 -1.32461107e+00 7.40866899e-01
-1.15376249e-01 -1.02580953e+00 8.34660083e-02 1.13185972e-01
1.03495836e+00 3.58895630e-01 2.20607504e-01 2.13705167e-01
1.02057898e+00 -9.07238781e-01 6.76374555e-01 -5.73662631e-02
1.24110794e+00 -5.50152779e-01 2.26066455e-01 2.75207996e-01
-1.09760273e+00 1.56601876e-01 -3.09201151e-01 2.02633858e-01
2.21716583e-01 4.66382205e-01 -9.08593297e-01 5.38343787e-01
6.82986200e-01 1.06033421e+00 -4.80721772e-01 6.21624768e-01
-6.86447322e-01 7.33934760e-01 -3.13964337e-01 7.32043803e-01
3.51696104e-01 -5.68121731e-01 1.84163898e-01 1.10895109e+00
9.26021561e-02 4.37430963e-02 4.16952103e-01 1.43533516e+00
-4.80193436e-01 -1.16160125e-01 -1.01785910e+00 1.71322435e-01
6.51746273e-01 1.08996952e+00 -6.60509050e-01 -5.05438983e-01
-4.94929880e-01 1.19617796e+00 3.42870474e-01 5.75315356e-01
-7.56636620e-01 1.89531613e-02 6.97730422e-01 1.82963267e-01
6.21022880e-01 -6.95179477e-02 -6.83795869e-01 -1.11225533e+00
-5.26001677e-02 -6.03006601e-01 4.16114986e-01 -8.39374244e-01
-1.74240673e+00 2.64291763e-01 1.42704353e-01 -1.09266853e+00
-3.73764426e-01 -4.28897202e-01 -4.87415940e-01 8.09540749e-01
-2.05155683e+00 -1.71315515e+00 -4.03672129e-01 7.29896843e-01
7.93925703e-01 -1.13391310e-01 4.49209780e-01 3.04324150e-01
-3.25838804e-01 5.47924459e-01 -2.06067398e-01 3.74389380e-01
9.82773840e-01 -1.52293289e+00 1.05579042e+00 9.36647534e-01
2.35732898e-01 2.09973976e-01 2.88959473e-01 -8.57191324e-01
-7.10938632e-01 -1.65061927e+00 6.79900825e-01 -6.90482557e-01
6.24775767e-01 -6.16933346e-01 -8.08600843e-01 1.15387571e+00
3.37209016e-01 8.81455019e-02 5.18902302e-01 -2.31593117e-01
-5.88562191e-01 -1.29292399e-01 -1.35845709e+00 3.78351748e-01
1.50611186e+00 -5.47603667e-01 -7.55736649e-01 2.84125626e-01
9.40703869e-01 -5.64981341e-01 -5.12577593e-01 6.09080076e-01
8.93162638e-02 -7.78205335e-01 1.16222167e+00 -5.93332350e-01
6.29818857e-01 -2.36235470e-01 -1.28340244e-01 -1.34501851e+00
-7.35579208e-02 -2.88545966e-01 3.79669040e-01 1.57714176e+00
3.40974599e-01 -5.85839748e-01 1.09120321e+00 4.51180577e-01
-2.96894521e-01 -1.57784104e-01 -9.10174489e-01 -9.71346021e-01
6.71891987e-01 -2.89290965e-01 6.44135654e-01 8.21450174e-01
-7.03957140e-01 7.86124468e-01 -3.79747921e-03 6.62925616e-02
1.00019884e+00 3.38182151e-01 1.08189619e+00 -1.36646020e+00
-1.19271062e-01 -1.63170785e-01 -5.41698113e-02 -1.33859265e+00
7.53539443e-01 -1.01059282e+00 2.93075591e-01 -1.36379540e+00
2.07930610e-01 -4.28566754e-01 -2.77993195e-02 6.22506142e-01
7.93640222e-03 4.80617464e-01 5.87568991e-02 1.56869709e-01
-5.24458289e-01 6.64407372e-01 1.17635715e+00 -3.60016584e-01
-8.58360901e-02 -1.48153380e-01 -9.63741064e-01 6.92604601e-01
7.84381449e-01 -6.72856629e-01 -7.12346375e-01 -6.80139780e-01
-2.70455897e-01 -3.82452935e-01 5.58368504e-01 -7.72742867e-01
-1.08892871e-02 -2.32367471e-01 1.94709271e-01 -6.53945625e-01
2.22347185e-01 -8.91883075e-01 -2.26577014e-01 -4.57278267e-02
-3.44388753e-01 -2.37535909e-01 1.68235645e-01 7.08197594e-01
-2.81694323e-01 1.00833684e-01 1.08022070e+00 -2.34472707e-01
-1.10459304e+00 5.41270494e-01 2.07837611e-01 7.46116221e-01
9.51564968e-01 -2.78793961e-01 -3.63858461e-01 -2.27605075e-01
-6.10360324e-01 4.91853565e-01 9.47813392e-01 3.79916042e-01
3.32546473e-01 -1.42067337e+00 -6.84864283e-01 1.70092613e-01
4.21324015e-01 7.79241502e-01 -1.80920810e-01 3.55566025e-01
-7.93732405e-02 1.25069574e-01 -2.30403721e-01 -9.43807423e-01
-6.91739440e-01 3.43996584e-01 1.99042857e-01 -1.64779246e-01
-4.96608019e-01 8.51114154e-01 6.22279286e-01 -9.25982118e-01
1.99856143e-02 -1.72860622e-01 3.50067288e-01 -1.20715544e-01
-1.20911039e-01 3.54532450e-02 -7.46083483e-02 -5.98121881e-01
-3.01167965e-01 8.30784440e-01 9.41154137e-02 -3.60995203e-01
1.18556464e+00 -2.10234717e-01 1.50083184e-01 3.75380367e-01
1.02825809e+00 -3.50257933e-01 -1.71462083e+00 -5.78787565e-01
8.18395242e-02 -4.22625452e-01 -2.79760450e-01 -9.64014411e-01
-1.20904601e+00 1.09605491e+00 4.89529878e-01 -2.95463055e-01
1.21911526e+00 4.20956701e-01 1.01544452e+00 2.27928251e-01
2.53697157e-01 -1.31841588e+00 3.31448436e-01 5.44350624e-01
3.75719517e-01 -1.52931058e+00 -3.63625884e-01 -7.39127159e-01
-9.54172194e-01 6.07047081e-01 1.05560398e+00 -5.18223941e-01
4.62085873e-01 2.91733146e-01 3.08757633e-01 -4.32837158e-02
-4.01878953e-01 -7.53675759e-01 1.23248994e-01 1.15616560e+00
1.18363753e-01 -2.62529869e-02 4.23255861e-02 4.50282782e-01
-3.55642229e-01 2.86306262e-01 1.54297113e-01 7.74194658e-01
-3.08713317e-01 -1.49381447e+00 -4.97565605e-02 4.28881310e-02
-6.80953488e-02 -1.01520538e-01 -8.51061106e-01 1.03166497e+00
3.26878816e-01 9.08991337e-01 3.10462862e-01 -1.60614446e-01
3.67307454e-01 3.09437007e-01 5.00264466e-01 -1.07053578e+00
-1.98745221e-01 2.25113511e-01 -1.04660124e-01 -5.32207131e-01
-6.30222142e-01 -6.76640809e-01 -1.33585322e+00 7.61538222e-02
-9.86800343e-03 -2.82527227e-02 4.97492880e-01 9.45577085e-01
5.00704288e-01 3.19060266e-01 6.22545779e-01 -7.80694366e-01
-5.07699013e-01 -7.77443707e-01 -6.16786659e-01 6.11060917e-01
2.68798143e-01 -7.31933296e-01 5.76323606e-02 4.51374769e-01] | [9.772359848022461, 1.2888256311416626] |
3d69f946-8c0d-471b-a894-bdfd7a909edb | code-translation-with-compiler | 2207.03578 | null | https://arxiv.org/abs/2207.03578v5 | https://arxiv.org/pdf/2207.03578v5.pdf | Code Translation with Compiler Representations | In this paper, we leverage low-level compiler intermediate representations (IR) to improve code translation. Traditional transpilers rely on syntactic information and handcrafted rules, which limits their applicability and produces unnatural-looking code. Applying neural machine translation (NMT) approaches to code has successfully broadened the set of programs on which one can get a natural-looking translation. However, they treat the code as sequences of text tokens, and still do not differentiate well enough between similar pieces of code which have different semantics in different languages. The consequence is low quality translation, reducing the practicality of NMT, and stressing the need for an approach significantly increasing its accuracy. Here we propose to augment code translation with IRs, specifically LLVM IR, with results on the C++, Java, Rust, and Go languages. Our method improves upon the state of the art for unsupervised code translation, increasing the number of correct translations by 11% on average, and up to 79% for the Java -> Rust pair with greedy decoding. We extend previous test sets for code translation, by adding hundreds of Go and Rust functions. Additionally, we train models with high performance on the problem of IR decompilation, generating programming source code from IR, and study using IRs as intermediary pivot for translation. | ['Francois Charton', 'Hugh Leather', 'Gabriel Synnaeve', 'Patrick Labatut', 'Baptiste Roziere', 'Marc Szafraniec'] | 2022-06-30 | null | null | null | null | ['code-translation'] | ['computer-code'] | [ 4.03880537e-01 1.91245005e-02 -6.28587604e-01 -1.43532544e-01
-1.23805606e+00 -8.86771917e-01 4.05696630e-01 5.94125688e-02
-3.17701721e-03 4.16177750e-01 2.86054343e-01 -1.11784685e+00
4.88324732e-01 -9.13933337e-01 -1.11865127e+00 -6.85534105e-02
2.11475641e-01 1.65081084e-01 -9.21857059e-02 -5.16548634e-01
3.13303500e-01 -2.30653748e-01 -1.26697922e+00 8.01373363e-01
1.14548445e+00 2.01317146e-01 6.26229718e-02 7.16107905e-01
-5.32244146e-01 9.17876959e-01 -6.36510491e-01 -5.54475725e-01
2.66929150e-01 -7.76280463e-01 -1.07766557e+00 -5.82710683e-01
4.25476313e-01 -4.63375449e-02 6.82253614e-02 9.98978198e-01
8.95433873e-02 -3.46717656e-01 5.11975944e-01 -9.97249782e-01
-1.07362366e+00 1.28507113e+00 -6.53630376e-01 -1.84463590e-01
5.19996524e-01 1.06445506e-01 1.15572679e+00 -7.57442117e-01
7.56786466e-01 1.12463617e+00 8.57189715e-01 8.34149837e-01
-1.50560296e+00 -4.28503603e-01 -1.69708058e-01 -2.91017681e-01
-1.30424976e+00 -5.15960336e-01 3.68982375e-01 -5.73468089e-01
1.63334262e+00 3.28578562e-01 2.57923007e-01 1.02379656e+00
3.81395370e-01 6.85393572e-01 9.15903986e-01 -7.78442442e-01
1.00686483e-03 3.76368128e-02 4.79764938e-02 8.15897644e-01
4.92697172e-02 1.88008562e-01 -7.58002400e-02 -2.76777923e-01
3.04397196e-01 -2.23722398e-01 -1.80386871e-01 -1.64775774e-01
-1.55623257e+00 8.29117954e-01 4.99180645e-01 4.46165562e-01
2.29972348e-01 3.84282887e-01 8.12732875e-01 7.84082711e-01
2.25510225e-01 6.63468957e-01 -6.87574744e-01 -3.14744443e-01
-9.89682674e-01 9.50900465e-02 9.34795737e-01 1.20136833e+00
9.45159853e-01 2.45098665e-01 -9.75268707e-02 7.73608387e-01
1.22011460e-01 6.14174962e-01 8.42521846e-01 -7.46720612e-01
8.90986323e-01 9.21945930e-01 -3.75221074e-01 -7.90200353e-01
1.44530414e-02 -3.85534018e-01 -3.35496068e-01 1.05554633e-01
3.12522978e-01 -2.16520671e-02 -7.61205494e-01 1.80109894e+00
-2.14323893e-01 -4.87514436e-01 4.24390346e-01 5.87921619e-01
4.63862777e-01 7.73373246e-01 -2.81003803e-01 5.14921993e-02
9.69867706e-01 -1.12334836e+00 -5.03750816e-02 -5.51195979e-01
1.44909394e+00 -9.79000270e-01 1.42498946e+00 1.39282420e-01
-9.47583318e-01 -4.55713421e-01 -9.65895414e-01 -1.19599506e-01
-2.34933361e-01 3.10974270e-01 8.54251325e-01 5.82423508e-01
-1.34495008e+00 7.11938918e-01 -7.94692099e-01 -2.86591440e-01
1.03033803e-01 3.62986773e-01 -2.89742857e-01 -1.62348732e-01
-7.92760015e-01 9.28279698e-01 4.22578484e-01 -1.25151470e-01
-6.61298871e-01 -7.24702239e-01 -1.07141316e+00 -4.37265597e-02
1.53864712e-01 -6.16527498e-01 1.42721844e+00 -1.53240919e+00
-1.58207953e+00 8.42420399e-01 -2.66430527e-01 -3.89473110e-01
2.59814531e-01 -8.95975828e-02 -3.05451125e-01 -4.59021032e-01
3.24826807e-01 7.01244712e-01 6.16565287e-01 -9.88732278e-01
-2.88746417e-01 2.50159130e-02 2.51557529e-01 -2.35165104e-01
-7.18372911e-02 4.13327694e-01 -1.31636769e-01 -5.07877290e-01
-2.48769298e-01 -1.22112572e+00 -1.09301507e-01 -5.05233526e-01
-2.92089015e-01 4.28044051e-02 3.76050681e-01 -9.14885640e-01
1.08418298e+00 -2.23961997e+00 5.67492306e-01 5.26683442e-02
-3.47492918e-02 4.57710214e-02 -5.41356206e-01 4.60094661e-01
-2.75273055e-01 5.58690190e-01 -5.60786068e-01 1.17614686e-01
8.12960789e-02 3.43497902e-01 -6.09410465e-01 9.88641083e-02
4.41819549e-01 1.24903119e+00 -9.31298852e-01 -3.27962965e-01
-2.59069830e-01 1.39400661e-01 -1.10246062e+00 1.14437535e-01
-5.47552109e-01 2.72803277e-01 -2.26981327e-01 6.36631548e-01
1.41582042e-01 -1.52513638e-01 3.04911137e-01 2.90063590e-01
-2.75011122e-01 8.63344252e-01 -5.22913218e-01 2.02279782e+00
-1.21168315e+00 7.31713951e-01 -2.94600517e-01 -8.55393112e-01
1.07281244e+00 1.22491218e-01 -3.82312224e-03 -7.20875621e-01
-1.05099238e-01 7.34145284e-01 3.27237159e-01 -3.08775753e-01
5.27902544e-01 1.64779206e-03 -5.51270247e-01 7.09141970e-01
-6.81473985e-02 -3.25443625e-01 2.51154512e-01 7.19573200e-02
1.27149308e+00 7.75651217e-01 2.23642781e-01 -3.28027725e-01
3.47069532e-01 4.45047051e-01 3.32336277e-01 5.41943014e-01
5.38279891e-01 5.49422801e-01 7.43682504e-01 -4.45393890e-01
-1.18415129e+00 -7.64095068e-01 1.79501504e-01 1.23500192e+00
-2.62903869e-01 -6.94915175e-01 -1.04000401e+00 -9.24639702e-01
-2.23352745e-01 1.03845370e+00 -5.20154178e-01 -4.69736367e-01
-1.11150372e+00 -7.72383988e-01 9.08053458e-01 5.07826090e-01
3.98990475e-02 -8.90252531e-01 -5.82623661e-01 2.07994178e-01
-4.93664473e-01 -6.90948963e-01 -7.27922976e-01 3.98509383e-01
-1.03828871e+00 -7.91500151e-01 -2.90612340e-01 -8.16321671e-01
7.92895734e-01 -1.86270922e-02 1.49293041e+00 4.60422814e-01
-1.00418933e-01 -2.27211073e-01 -3.97981912e-01 7.74686933e-02
-1.38168073e+00 5.18586755e-01 -2.96539515e-01 -5.76723933e-01
2.03217924e-01 -5.29017985e-01 1.34440571e-01 2.57897466e-01
-9.28476870e-01 2.41202086e-01 8.15685213e-01 1.04722834e+00
3.20657007e-02 -5.17344892e-01 2.37086385e-01 -9.43838298e-01
3.58765692e-01 -4.21829939e-01 -6.51766539e-01 2.55479097e-01
-6.59310758e-01 6.98441565e-01 1.18728364e+00 -5.48513353e-01
-8.87926280e-01 -9.92486253e-02 -2.02423498e-01 -8.13052505e-02
1.14250302e-01 7.36593843e-01 1.75393626e-01 -1.52646810e-01
1.34534597e+00 2.13508382e-01 -7.74364993e-02 -3.36887449e-01
5.15416622e-01 5.99188089e-01 4.12971556e-01 -1.14838099e+00
9.74502981e-01 -1.60322756e-01 -4.13777381e-01 -1.79438263e-01
-4.33493435e-01 1.14179865e-01 -5.38185477e-01 4.27389890e-01
4.26007301e-01 -7.73818016e-01 -6.90620020e-02 5.48996544e-03
-1.42164278e+00 -6.46804571e-01 -1.77401587e-01 2.00994223e-01
-5.70863545e-01 3.84896785e-01 -8.99287522e-01 -1.29966021e-01
-4.02505636e-01 -1.59342313e+00 1.23017824e+00 -3.37529272e-01
-6.58873320e-01 -7.94576883e-01 3.17624032e-01 1.97715729e-01
6.85998559e-01 5.60557023e-02 1.61625457e+00 -3.23750615e-01
-6.89140975e-01 5.09728231e-02 -2.30024040e-01 3.03870142e-01
6.10218123e-02 1.17164657e-01 -5.78977168e-01 -2.68365622e-01
-3.62161785e-01 -2.92208940e-01 6.17963552e-01 -2.51662552e-01
5.67676067e-01 -5.83574653e-01 -2.18469352e-01 7.54769325e-01
1.57630169e+00 1.05442461e-02 6.75554216e-01 5.11636317e-01
7.88461745e-01 3.94844323e-01 2.13020608e-01 -1.55203566e-01
3.03509444e-01 7.52511203e-01 3.88143748e-01 -1.21428199e-01
-2.33830899e-01 -4.33799267e-01 1.15905738e+00 1.32011938e+00
9.34823453e-02 1.87440336e-01 -1.24853122e+00 7.55073845e-01
-1.67276406e+00 -4.84402090e-01 -2.48907030e-01 2.16473818e+00
1.25821626e+00 -7.45962188e-02 -1.02159709e-01 -1.04647532e-01
4.20893401e-01 -3.08817834e-01 -1.75595209e-01 -9.74854410e-01
3.34045023e-01 5.34673154e-01 6.04561448e-01 5.98521769e-01
-5.46185017e-01 1.20187867e+00 6.23726892e+00 5.83069861e-01
-1.32685852e+00 1.65458113e-01 3.13406527e-01 3.26595306e-01
-7.77584136e-01 5.53415298e-01 -6.92763567e-01 2.77272552e-01
1.39223611e+00 -2.66503662e-01 9.88391459e-01 1.03052580e+00
-2.08177552e-01 2.32902005e-01 -1.46943939e+00 6.25287235e-01
1.57861039e-01 -1.16678166e+00 2.75230538e-02 -1.65309235e-01
8.33404422e-01 4.03493643e-01 -2.06694350e-01 8.31395507e-01
5.92919409e-01 -1.09011221e+00 9.01142180e-01 8.73853732e-03
9.36844587e-01 -6.21979833e-01 7.37104237e-01 2.18957990e-01
-1.00117266e+00 3.54525484e-02 -3.23264539e-01 -2.21258953e-01
-4.44811374e-01 3.39309305e-01 -1.01870418e+00 6.52983904e-01
4.17764068e-01 6.95484519e-01 -9.06217039e-01 4.56887186e-01
-4.46527630e-01 6.03412688e-01 -1.55544236e-01 -9.44500715e-02
1.80365130e-01 -1.82587393e-02 2.67095327e-01 1.47416461e+00
5.41904390e-01 -5.49032450e-01 2.57546037e-01 1.26709127e+00
-2.73547858e-01 3.06279749e-01 -9.34921443e-01 -2.30607599e-01
4.61989790e-02 1.01242030e+00 -6.82329774e-01 -3.80789101e-01
-5.58908403e-01 9.87657428e-01 4.07815039e-01 1.85177267e-01
-9.50572848e-01 -4.98328149e-01 5.69168270e-01 -1.20486088e-01
8.46327543e-02 -2.58803427e-01 -3.23592037e-01 -1.46852207e+00
3.10559362e-01 -1.51772976e+00 -2.55614575e-02 -5.86737454e-01
-7.05357492e-01 8.49511087e-01 -7.27947876e-02 -1.25069809e+00
-7.58785844e-01 -4.95957613e-01 -5.11200070e-01 8.65487516e-01
-1.41995025e+00 -1.06584775e+00 2.95664340e-01 2.66665705e-02
6.02443993e-01 -1.21996887e-01 9.51118290e-01 3.52329254e-01
-3.41081679e-01 8.01502883e-01 1.18226677e-01 3.81285787e-01
6.13606989e-01 -1.10012829e+00 1.03207743e+00 1.13815165e+00
1.75258711e-01 1.17498076e+00 4.90595281e-01 -4.75383282e-01
-1.93760383e+00 -1.21651483e+00 1.04415429e+00 -7.16287255e-01
7.94246078e-01 -4.58143085e-01 -8.40217590e-01 8.31849694e-01
1.72204614e-01 -1.65023074e-01 3.84045541e-01 1.41743869e-01
-8.67406785e-01 2.05141619e-01 -7.36454427e-01 8.79068255e-01
9.90297556e-01 -7.54483998e-01 -6.65666044e-01 2.50051677e-01
1.09756386e+00 -5.32535732e-01 -9.46464777e-01 1.39003009e-01
4.18960571e-01 -6.07667744e-01 7.95062244e-01 -5.93028426e-01
1.12257302e+00 -4.28484410e-01 -3.44332963e-01 -1.41268516e+00
5.97573863e-03 -5.78771532e-01 1.86874017e-01 1.25811863e+00
9.41536069e-01 -5.51051319e-01 4.97492075e-01 2.82845289e-01
-3.33705753e-01 -3.82918686e-01 -5.42354584e-01 -8.96767378e-01
5.60634136e-01 -5.20157754e-01 7.28268683e-01 1.10662043e+00
4.19639111e-01 6.01298690e-01 -1.52098432e-01 -2.74836451e-01
1.92435697e-01 5.57096064e-01 9.59681809e-01 -7.24251688e-01
-9.18993771e-01 -6.35059178e-01 -9.76051912e-02 -9.01879787e-01
5.81258237e-01 -1.69434285e+00 3.43632907e-01 -1.13649738e+00
2.54803330e-01 -4.36032355e-01 2.13637903e-01 1.08706284e+00
2.72130277e-02 2.76012480e-01 1.97329968e-01 3.74935865e-01
-1.72118321e-01 1.53884232e-01 6.61144614e-01 -5.24164438e-01
-1.90710098e-01 -3.54092300e-01 -9.06998158e-01 4.16618615e-01
8.46251547e-01 -8.11820507e-01 -1.63739070e-01 -9.91927803e-01
7.39346623e-01 1.63210496e-01 1.03199363e-01 -7.31933355e-01
-3.38103563e-01 -1.24570727e-01 -1.40243158e-01 1.59203693e-01
-4.35964972e-01 -5.61640382e-01 2.26133734e-01 5.72965443e-01
-6.95953190e-01 4.76354986e-01 3.98724020e-01 -3.25646698e-02
-1.65884778e-01 -5.74227154e-01 6.58827484e-01 -4.37081397e-01
-5.00877202e-01 -2.63373405e-01 -2.82238960e-01 2.52382725e-01
5.15469432e-01 -2.48176921e-02 -4.36040580e-01 9.97364596e-02
-4.26875353e-02 -2.76676416e-01 1.06836963e+00 7.87984610e-01
3.13698709e-01 -1.10611105e+00 -7.55648375e-01 4.14744526e-01
3.90648127e-01 -2.98304439e-01 -6.12803459e-01 8.69380713e-01
-6.35795593e-01 3.15407842e-01 -1.83872715e-01 -6.55091703e-01
-1.10700083e+00 7.17925429e-01 2.23650321e-01 -3.59571755e-01
-5.10470510e-01 3.09364438e-01 -2.28355993e-02 -1.11633801e+00
-4.92276371e-01 -7.98073053e-01 3.32930714e-01 -5.99471033e-01
2.94081599e-01 -9.84582901e-02 4.31844443e-01 -5.21265030e-01
-3.51783454e-01 7.16401815e-01 -1.29968300e-01 4.43543382e-02
1.16224265e+00 4.60050851e-01 -7.87244380e-01 3.08039308e-01
1.57187259e+00 3.33759278e-01 -5.22479475e-01 -6.01815768e-02
3.70869428e-01 -3.80332500e-01 -3.09456348e-01 -7.97052026e-01
-8.91606867e-01 8.87523532e-01 1.49142355e-01 -1.53920799e-01
8.65231633e-01 -5.81867695e-02 8.57562125e-01 6.80217147e-01
7.54814565e-01 -5.63471675e-01 -2.12068230e-01 8.48691225e-01
4.65249389e-01 -1.02775598e+00 -4.16901231e-01 -1.27058759e-01
-2.51917690e-01 1.37745309e+00 4.77216274e-01 -1.13199107e-01
-2.40609869e-01 6.73305511e-01 9.56014469e-02 3.19254160e-01
-8.63070190e-01 1.82619527e-01 8.50290284e-02 4.56218779e-01
9.27156568e-01 1.20037138e-01 -3.59692782e-01 1.87707454e-01
-3.51200342e-01 5.61177470e-02 8.46941173e-01 1.14808762e+00
-6.86416179e-02 -1.69986629e+00 -4.80057389e-01 3.76362503e-01
-6.33783698e-01 -7.43272722e-01 -4.09763694e-01 7.71444976e-01
-1.74918160e-01 7.37585366e-01 -2.78792083e-01 -5.41854262e-01
1.50098681e-01 1.94969743e-01 5.47307074e-01 -9.59000349e-01
-1.01850033e+00 -2.53795564e-01 2.68917799e-01 -5.32766402e-01
3.72620858e-02 -4.84212577e-01 -1.27199364e+00 -3.51879567e-01
-2.47771755e-01 2.58464903e-01 7.43397772e-01 9.02026057e-01
5.68367720e-01 4.98978734e-01 5.08499026e-01 -6.68359518e-01
-5.88079572e-01 -6.47654414e-01 4.72474515e-01 2.93529093e-01
2.91608155e-01 -6.78114891e-02 -2.48994663e-01 3.78143072e-01] | [7.71728515625, 7.849681377410889] |
693e1164-dc23-49ec-ae7d-574088aa0473 | lifelong-person-re-identification-via-1 | 2211.16201 | null | https://arxiv.org/abs/2211.16201v1 | https://arxiv.org/pdf/2211.16201v1.pdf | Lifelong Person Re-Identification via Knowledge Refreshing and Consolidation | Lifelong person re-identification (LReID) is in significant demand for real-world development as a large amount of ReID data is captured from diverse locations over time and cannot be accessed at once inherently. However, a key challenge for LReID is how to incrementally preserve old knowledge and gradually add new capabilities to the system. Unlike most existing LReID methods, which mainly focus on dealing with catastrophic forgetting, our focus is on a more challenging problem, which is, not only trying to reduce the forgetting on old tasks but also aiming to improve the model performance on both new and old tasks during the lifelong learning process. Inspired by the biological process of human cognition where the somatosensory neocortex and the hippocampus work together in memory consolidation, we formulated a model called Knowledge Refreshing and Consolidation (KRC) that achieves both positive forward and backward transfer. More specifically, a knowledge refreshing scheme is incorporated with the knowledge rehearsal mechanism to enable bi-directional knowledge transfer by introducing a dynamic memory model and an adaptive working model. Moreover, a knowledge consolidation scheme operating on the dual space further improves model stability over the long term. Extensive evaluations show KRC's superiority over the state-of-the-art LReID methods on challenging pedestrian benchmarks. | ['Jingya Wang', 'Shenghua Gao', 'Zimo Liu', 'Ye Shi', 'Chunlin Yu'] | 2022-11-29 | null | null | null | null | ['person-re-identification'] | ['computer-vision'] | [-2.94826776e-01 -3.43069285e-01 -1.28736511e-01 -9.12214816e-02
1.22997113e-01 -2.02450678e-01 5.06149530e-01 3.09451103e-01
-9.13060546e-01 9.66892838e-01 1.77098230e-01 -1.02621481e-01
-3.69691521e-01 -9.88539279e-01 -6.78696215e-01 -6.66410983e-01
3.01624206e-03 2.98361033e-01 5.60211897e-01 -2.15784937e-01
1.65852070e-01 4.51437116e-01 -1.73240578e+00 -1.32437661e-01
1.17785478e+00 6.69249535e-01 4.69260186e-01 2.26852164e-01
-6.54503778e-02 5.67481458e-01 -2.48759493e-01 -6.84256494e-01
-9.11547393e-02 -1.58763126e-01 -8.81541610e-01 -3.11731815e-01
2.05707073e-01 -2.24763513e-01 -5.59913754e-01 8.71662796e-01
8.12245965e-01 5.44459403e-01 3.16775411e-01 -1.22853065e+00
-9.58438694e-01 3.20582598e-01 -5.48007131e-01 6.46785021e-01
3.34724784e-01 1.67266503e-01 3.16333562e-01 -1.15741384e+00
4.08304513e-01 1.07037175e+00 9.25395191e-01 6.48128688e-01
-1.04586732e+00 -8.85672987e-01 5.59992433e-01 8.61646056e-01
-1.55690551e+00 -2.86180228e-01 4.71490592e-01 -3.15825880e-01
9.04727697e-01 -9.10702497e-02 1.11455739e+00 9.42635000e-01
3.78433943e-01 7.62647688e-01 1.05790412e+00 -2.83301324e-01
1.61136776e-01 1.57123029e-01 6.28731072e-01 4.65668201e-01
6.66653693e-01 1.84764832e-01 -1.15514278e+00 2.14552641e-01
5.78381300e-01 1.37270883e-01 -2.91706741e-01 -2.97416568e-01
-1.09570384e+00 2.34519526e-01 4.65967298e-01 3.21158975e-01
-6.37288690e-01 -2.83937782e-01 2.80770868e-01 4.53681588e-01
3.01186770e-01 1.20657794e-01 -3.24527532e-01 -1.04123168e-01
-8.02223146e-01 3.44736338e-01 4.37322825e-01 6.58447146e-01
8.22412372e-01 -1.56324059e-01 -4.48075920e-01 7.98653185e-01
-7.02248514e-02 5.45754135e-01 8.49316418e-01 -4.78956550e-01
2.41912603e-01 7.20217049e-01 3.05781476e-02 -9.95476782e-01
-5.67583978e-01 -8.90613914e-01 -1.03038919e+00 -3.52407917e-02
3.16520065e-01 2.62343764e-01 -6.88927829e-01 2.14317083e+00
3.49450886e-01 5.79609931e-01 7.66495764e-02 7.61185646e-01
7.04941273e-01 3.15747738e-01 5.62568605e-01 -5.37600756e-01
1.37161088e+00 -9.12739098e-01 -7.97683537e-01 -3.16528618e-01
3.83150250e-01 -3.86166245e-01 9.70581234e-01 2.68874198e-01
-1.15779865e+00 -9.23351645e-01 -1.02876568e+00 -2.41602561e-03
-6.46750093e-01 -2.85259038e-01 6.23644292e-01 5.38820803e-01
-1.10498977e+00 3.42570961e-01 -4.75455046e-01 -4.55739498e-01
3.97554308e-01 2.80210108e-01 -5.82081199e-01 -3.18050534e-01
-1.61747479e+00 1.24254668e+00 6.50178373e-01 2.74118334e-01
-6.74593866e-01 -9.50232804e-01 -5.90932488e-01 6.05328828e-02
2.71805048e-01 -1.02079499e+00 8.02543581e-01 -5.98913193e-01
-1.21144414e+00 7.52010167e-01 -1.99900940e-01 -5.33264279e-01
4.33254421e-01 -5.87060511e-01 -6.48033738e-01 -2.37760529e-01
3.17713469e-02 6.77641809e-01 6.78782225e-01 -1.16282904e+00
-7.83931971e-01 -6.63034678e-01 -1.11282177e-01 5.55365562e-01
-8.19644630e-01 -4.71452981e-01 -6.65269017e-01 -8.58184457e-01
-4.02234830e-02 -8.68822396e-01 1.42648786e-01 -1.34877384e-01
3.22744131e-01 -2.15302289e-01 7.23387778e-01 -1.01715875e+00
1.65292227e+00 -1.97793651e+00 4.08775985e-01 5.32002784e-02
1.61971018e-01 7.40874946e-01 -1.69743463e-01 1.72450751e-01
-6.61102906e-02 -4.25432980e-01 -3.76529992e-02 -4.01141852e-01
-2.00371787e-01 1.59352571e-01 -2.25665152e-01 1.43243402e-01
-2.90300846e-01 1.21684444e+00 -1.12013423e+00 -3.43149185e-01
-3.46176662e-02 4.78338659e-01 -3.17396432e-01 -2.36203475e-03
2.63018370e-01 3.42641562e-01 1.93312705e-01 3.87361050e-01
9.35167968e-01 -1.94656596e-01 1.14235384e-02 -2.97973976e-02
-2.12101340e-01 -3.07178587e-01 -1.06835747e+00 1.75527859e+00
-2.93606430e-01 1.15695730e-01 -2.89428294e-01 -6.75317585e-01
7.93287039e-01 -1.42653994e-02 1.94125324e-01 -1.25124288e+00
-1.27985239e-01 3.99682634e-02 -1.71988130e-01 -1.57812268e-01
7.92155981e-01 -1.36396483e-01 5.63004762e-02 4.74688470e-01
1.82827204e-01 6.18567586e-01 -2.25851666e-02 2.48736873e-01
8.54188979e-01 1.94127150e-02 2.64307231e-01 -3.17212999e-01
8.46811533e-01 -2.92469859e-01 9.17604983e-01 7.07684457e-01
-5.48746109e-01 5.01405187e-02 -2.10390657e-01 -4.82975453e-01
-6.27722859e-01 -1.32278442e+00 8.13767090e-02 1.25090969e+00
5.95692754e-01 -2.66453356e-01 -6.41651392e-01 -5.14991701e-01
1.84674650e-01 7.69473910e-01 -6.32114232e-01 -9.32795763e-01
-5.10604143e-01 -8.58061254e-01 4.97272760e-01 5.89843810e-01
1.26919305e+00 -1.08725500e+00 -5.93950391e-01 2.40259975e-01
-4.03295338e-01 -7.08031416e-01 -7.11336792e-01 -1.57422125e-01
-8.13787103e-01 -9.26692247e-01 -1.02214599e+00 -8.71638417e-01
5.02090156e-01 7.46993780e-01 9.44690466e-01 1.56824812e-01
-1.18812002e-01 7.38216400e-01 -3.03334296e-01 -3.36237609e-01
4.67343442e-02 3.14071596e-01 5.82151353e-01 8.17953944e-02
5.25757909e-01 -7.05308080e-01 -8.04521978e-01 3.43764991e-01
-7.33946919e-01 3.35702926e-01 6.62625909e-01 9.98988867e-01
5.35195291e-01 4.34044659e-01 1.06562793e+00 -4.70605046e-01
6.79566503e-01 -6.05735958e-01 -1.80214673e-01 6.31841779e-01
-1.14810944e+00 -4.18769978e-02 1.96421981e-01 -7.10016727e-01
-1.50887501e+00 -3.07193667e-01 -8.40451643e-02 -2.68896550e-01
1.62893221e-01 3.84325475e-01 -2.08303005e-01 -9.58514214e-02
3.29926193e-01 7.73655415e-01 -1.86397340e-02 -5.17856956e-01
3.67738783e-01 3.74041378e-01 8.88234615e-01 -5.14525294e-01
6.56553268e-01 3.08613420e-01 -2.36459076e-01 -6.91952288e-01
-7.64104187e-01 -3.36743355e-01 -9.02621806e-01 -5.65705836e-01
6.51109338e-01 -1.02605462e+00 -9.37200189e-01 1.25934947e+00
-1.00390208e+00 -3.12763631e-01 -5.05707920e-01 4.53385264e-01
-3.81965786e-01 2.78997540e-01 -4.97610301e-01 -5.26125610e-01
-4.52330381e-01 -4.59515274e-01 3.95747215e-01 6.30776048e-01
5.07836454e-02 -9.24946964e-01 2.29782373e-01 2.32475027e-01
6.05335355e-01 -3.63604039e-01 1.02592480e+00 -2.51124710e-01
-4.02912319e-01 1.28061056e-01 -3.50410044e-01 2.03967571e-01
-4.98761050e-02 -9.77781236e-01 -7.27239549e-01 -5.66627979e-01
-1.38756290e-01 -1.92897603e-01 1.11077034e+00 -1.79515947e-02
8.42266381e-01 -6.07314818e-02 -3.98399472e-01 2.88083196e-01
1.04752505e+00 1.08585171e-01 8.46925080e-01 5.30220449e-01
4.87609565e-01 6.07061267e-01 7.09129751e-01 3.00787240e-01
1.01513970e+00 7.10905731e-01 2.74462160e-02 2.65469551e-01
-5.47198355e-01 -6.71248138e-01 1.07939698e-01 1.03622401e+00
-2.13688225e-01 1.69114500e-01 -9.15288746e-01 7.38000929e-01
-2.27363586e+00 -1.13600659e+00 1.05074272e-01 2.56763196e+00
1.07873392e+00 1.33506596e-01 2.67515238e-02 2.18888402e-01
7.45926380e-01 -1.53756380e-01 -8.48335207e-01 2.36231610e-01
-3.23206842e-01 -1.17333755e-02 2.45496675e-01 3.95281821e-01
-7.65926838e-01 1.03634918e+00 6.07383585e+00 8.49979222e-01
-9.35193717e-01 4.81427252e-01 3.58988523e-01 -5.32977283e-02
1.78351961e-02 -2.27593571e-01 -1.07956588e+00 5.94150484e-01
8.33224297e-01 -4.56433594e-01 4.69394267e-01 5.18859267e-01
-1.74487486e-01 -4.27352041e-01 -6.74046874e-01 1.25686693e+00
1.42251834e-01 -1.00123906e+00 1.94114923e-01 -1.73349977e-01
5.42646706e-01 -4.61487591e-01 1.57936856e-01 6.58604026e-01
-1.30896755e-02 -6.88180685e-01 8.14257801e-01 1.32186568e+00
5.07558703e-01 -8.40735257e-01 6.26462519e-01 6.93960845e-01
-1.32388270e+00 -3.65236521e-01 -2.53732532e-01 -1.51914924e-01
1.33413911e-01 4.89097655e-01 -1.90380454e-01 5.66996574e-01
9.55301046e-01 6.98544621e-01 -1.04774046e+00 1.27808034e+00
-1.91828564e-01 2.20660582e-01 1.57664910e-01 2.78258681e-01
-3.55613828e-01 1.28605798e-01 3.59510392e-01 1.11259615e+00
3.78772706e-01 2.82039493e-01 -1.08306341e-01 3.64353329e-01
1.94924604e-02 -1.37914876e-02 -4.85971123e-01 4.89926875e-01
7.92534709e-01 8.77239048e-01 -5.36059797e-01 -2.23566562e-01
-2.00837433e-01 1.28826010e+00 6.40485168e-01 5.38917899e-01
-6.87192678e-01 -1.96650639e-01 5.48393011e-01 3.37344289e-01
2.27127030e-01 -5.04645884e-01 -1.78039953e-01 -1.14987195e+00
-1.79779921e-02 -4.64855850e-01 6.69715583e-01 -7.10951209e-01
-1.39052522e+00 3.10306191e-01 -1.50277568e-02 -8.70348275e-01
2.76378185e-01 -1.76075146e-01 -3.84675205e-01 9.12362158e-01
-1.77249110e+00 -1.37983716e+00 -6.87555611e-01 9.92805302e-01
4.06560600e-01 -2.36824378e-01 7.80794442e-01 6.87933564e-01
-5.49749732e-01 8.82992327e-01 -1.44455820e-01 -2.71078438e-01
8.28630149e-01 -7.71284461e-01 1.15523338e-01 9.58052754e-01
-2.96199650e-01 9.42591131e-01 4.65386420e-01 -9.70768452e-01
-1.40728974e+00 -1.30464137e+00 1.08587718e+00 -2.79300034e-01
2.79134274e-01 -2.26124093e-01 -1.38454258e+00 4.77264315e-01
4.60363068e-02 -1.61767006e-01 5.86241841e-01 2.97143966e-01
-4.19728547e-01 -4.11405563e-01 -9.10060763e-01 5.52623868e-01
1.55736303e+00 -4.94359016e-01 -8.51168513e-01 -2.68644542e-01
6.94079459e-01 -1.25598684e-01 -6.72456026e-01 4.28255022e-01
7.30594158e-01 -8.45060050e-01 1.17955792e+00 -5.51491678e-01
-3.74103278e-01 -5.17531931e-01 6.12772591e-02 -1.29566300e+00
-8.19523096e-01 -2.17393801e-01 -6.28293693e-01 1.47004533e+00
-1.72559634e-01 -8.27173054e-01 4.98569041e-01 5.28681159e-01
1.03448853e-01 -5.89281499e-01 -1.12366092e+00 -1.10228395e+00
-7.15640932e-03 -2.01251686e-01 6.91489398e-01 8.31309021e-01
-1.23311572e-01 3.76970619e-01 -6.31580710e-01 2.01007456e-01
7.48449981e-01 -1.46065429e-01 5.70266843e-01 -1.44480622e+00
-7.84733857e-05 -3.68088961e-01 -3.12892526e-01 -9.14928734e-01
2.32701287e-01 -9.19811845e-01 -2.43434489e-01 -1.38161039e+00
5.91541231e-01 -4.95283395e-01 -8.39395881e-01 5.47640979e-01
-4.76161569e-01 1.07837342e-01 2.67711937e-01 3.34689230e-01
-1.02151394e+00 1.00147581e+00 1.09118545e+00 -5.65834902e-03
-3.89218003e-01 -1.00676186e-01 -9.16269660e-01 5.19784510e-01
5.98856568e-01 -2.26028159e-01 -6.12533331e-01 -3.45971286e-01
3.57329041e-01 -1.13300234e-01 6.77007794e-01 -1.36008263e+00
8.36375892e-01 -4.49057706e-02 4.43111479e-01 -6.29623950e-01
2.77758360e-01 -7.11336493e-01 4.16062295e-01 6.48568630e-01
7.16046542e-02 2.10533559e-01 2.82364756e-01 8.83791327e-01
-1.46203101e-01 8.64400715e-03 8.32632244e-01 3.22582163e-02
-1.23031664e+00 5.24945199e-01 -3.41594338e-01 -1.15916297e-01
1.39919686e+00 -2.92811900e-01 -6.04444623e-01 4.32123197e-03
-8.95389199e-01 5.25357127e-01 3.85332644e-01 9.11432207e-01
9.01021242e-01 -1.67399669e+00 -4.36080158e-01 2.80744851e-01
1.87976658e-01 -4.28363413e-01 8.57055664e-01 8.29507172e-01
5.91850355e-02 4.54606861e-01 -6.29663527e-01 -1.69362873e-01
-1.22581315e+00 7.81347334e-01 2.89800912e-01 -4.57496494e-01
-4.81752634e-01 7.22358704e-01 1.29210278e-02 -2.58965284e-01
4.60716695e-01 4.56599951e-01 -5.67664206e-01 3.71302664e-01
9.45884764e-01 7.32091188e-01 9.36615393e-02 -6.04676425e-01
-4.43810791e-01 5.09059072e-01 -2.67574728e-01 1.17121063e-01
1.18146944e+00 -6.23968303e-01 -1.48002505e-01 5.97049057e-01
4.36656803e-01 -4.73521739e-01 -1.08507347e+00 -7.08654284e-01
1.10317236e-02 -5.41983366e-01 3.50307673e-02 -9.45554972e-01
-8.01101923e-01 8.07173908e-01 9.88303959e-01 -4.61156100e-01
1.36613941e+00 -2.46013150e-01 1.03311944e+00 4.55047965e-01
8.93369019e-01 -1.38153672e+00 4.52955574e-01 7.44929910e-01
7.65178502e-01 -9.91106689e-01 -8.42712894e-02 -1.10064512e-02
-5.48941433e-01 6.59519911e-01 8.68654430e-01 3.24348241e-01
9.15769875e-01 -1.65110841e-01 -4.82936472e-01 5.81582040e-02
-6.38548851e-01 -2.45254666e-01 1.55560628e-01 6.71878338e-01
-6.55823275e-02 -9.56297964e-02 -4.87898916e-01 1.17936981e+00
1.41869903e-01 4.79805291e-01 4.99200076e-02 9.29997444e-01
-6.19032860e-01 -1.08912361e+00 -2.94899464e-01 3.61866415e-01
1.61033362e-01 6.04881234e-02 2.27308229e-01 5.10927498e-01
6.25226259e-01 7.04390943e-01 -2.12089330e-01 -5.63923180e-01
5.21943390e-01 3.73892188e-01 5.30964792e-01 -3.11667621e-01
-4.38822597e-01 -6.40029430e-01 -3.50653470e-01 -3.54162455e-01
-3.13993126e-01 -8.11667562e-01 -1.15655959e+00 -6.61151528e-01
-1.18900463e-01 1.58174202e-01 3.02919567e-01 8.29762459e-01
6.11000001e-01 5.66573143e-01 2.03645214e-01 -4.98644531e-01
-1.87243462e-01 -7.73144305e-01 -6.55322373e-01 4.28000778e-01
2.09679944e-03 -1.10450351e+00 2.24087641e-01 3.64188142e-02] | [9.815508842468262, 3.186896562576294] |
25f879b7-7394-4a5e-aed6-4376cfa6357e | enhancing-semantic-code-search-with | 2204.03293 | null | https://arxiv.org/abs/2204.03293v3 | https://arxiv.org/pdf/2204.03293v3.pdf | CoCoSoDa: Effective Contrastive Learning for Code Search | Code search aims to retrieve semantically relevant code snippets for a given natural language query. Recently, many approaches employing contrastive learning have shown promising results on code representation learning and greatly improved the performance of code search. However, there is still a lot of room for improvement in using contrastive learning for code search. In this paper, we propose CoCoSoDa to effectively utilize contrastive learning for code search via two key factors in contrastive learning: data augmentation and negative samples. Specifically, soft data augmentation is to dynamically masking or replacing some tokens with their types for input sequences to generate positive samples. Momentum mechanism is used to generate large and consistent representations of negative samples in a mini-batch through maintaining a queue and a momentum encoder. In addition, multimodal contrastive learning is used to pull together representations of code-query pairs and push apart the unpaired code snippets and queries. We conduct extensive experiments to evaluate the effectiveness of our approach on a large-scale dataset with six programming languages. Experimental results show that: (1) CoCoSoDa outperforms 14 baselines and especially exceeds CodeBERT, GraphCodeBERT, and UniXcoder by 13.3%, 10.5%, and 5.9% on average MRR scores, respectively. (2) The ablation studies show the effectiveness of each component of our approach. (3) We adapt our techniques to several different pre-trained models such as RoBERTa, CodeBERT, and GraphCodeBERT and observe a significant boost in their performance in code search. (4) Our model performs robustly under different hyper-parameters. Furthermore, we perform qualitative and quantitative analyses to explore reasons behind the good performance of our model. | ['Wenchao Gu', 'Hongbin Sun', 'Dongmei Zhang', 'Shi Han', 'Hongyu Zhang', 'Lun Du', 'Yanlin Wang', 'Ensheng Shi'] | 2022-04-07 | null | null | null | null | ['code-search', 'code-search'] | ['computer-code', 'computer-vision'] | [-4.32808883e-02 -4.02915120e-01 -5.69052994e-01 -2.59845942e-01
-1.20459461e+00 -6.16653323e-01 4.82948184e-01 1.64166242e-01
-3.81128401e-01 2.12370232e-01 2.57567316e-01 -5.08406162e-01
1.64751664e-01 -4.73741829e-01 -7.43667305e-01 -3.70157897e-01
-1.32319346e-01 2.78813481e-01 3.13910425e-01 -1.86637849e-01
4.63227481e-01 1.49308309e-01 -1.56589794e+00 5.46142995e-01
1.13666725e+00 7.60517538e-01 6.15498543e-01 6.97836697e-01
-3.97547156e-01 1.12769639e+00 -4.50762928e-01 -3.49455565e-01
1.47078991e-01 -3.70886505e-01 -9.26091075e-01 -2.78278381e-01
3.20409119e-01 -1.94682688e-01 -2.64668852e-01 1.20044327e+00
3.92698675e-01 -7.69515708e-02 5.30094147e-01 -1.09842265e+00
-9.38100874e-01 7.52371788e-01 -8.75523567e-01 3.36155385e-01
4.66120750e-01 2.40303308e-01 1.24749637e+00 -1.20004475e+00
3.52516830e-01 9.89930212e-01 5.51825404e-01 6.92977190e-01
-1.01167798e+00 -8.27185392e-01 1.01139270e-01 1.38448223e-01
-1.38811481e+00 -4.45154905e-01 6.47175252e-01 -5.75779378e-01
1.31967950e+00 4.00161475e-01 3.54623914e-01 9.49848950e-01
-1.26227871e-01 1.22031701e+00 6.37386382e-01 -6.12571359e-01
2.05800831e-02 3.64483416e-01 2.21785963e-01 1.14598846e+00
-1.13136098e-01 1.27138451e-01 -3.35607052e-01 -5.23973644e-01
4.62872416e-01 1.40250683e-01 -2.33673811e-01 -4.54293042e-01
-8.78285944e-01 1.05773389e+00 6.19172335e-01 2.07976848e-01
-5.53100221e-02 3.37840766e-01 6.17876053e-01 2.37417221e-01
2.32935369e-01 7.37246394e-01 -5.24022996e-01 -3.81660968e-01
-9.57947314e-01 2.39344761e-01 6.03023171e-01 1.26693881e+00
7.49604225e-01 7.27688372e-02 -3.97091419e-01 1.18506396e+00
3.83133620e-01 4.19041514e-01 8.53579819e-01 -6.58860087e-01
7.99084544e-01 7.96813965e-01 -3.08934867e-01 -7.70589411e-01
-2.62651183e-02 -5.08481860e-01 -4.73057091e-01 -3.29630226e-02
-6.63006157e-02 1.27624422e-01 -1.08575356e+00 1.70188189e+00
-3.71578038e-01 -9.78442207e-02 4.11005616e-02 6.68712735e-01
9.11897898e-01 5.90937614e-01 1.22510634e-01 3.02959710e-01
1.17402256e+00 -1.45843673e+00 -8.87086689e-02 -5.42330086e-01
9.17769730e-01 -8.94151807e-01 1.70158660e+00 -5.88001451e-03
-1.18481135e+00 -4.52578783e-01 -8.74186873e-01 -1.53305277e-01
-1.73583493e-01 4.60850477e-01 7.23178864e-01 5.01836717e-01
-1.03811669e+00 1.09738596e-01 -8.69244516e-01 -2.21817866e-01
4.18485403e-01 3.82940471e-01 1.23200849e-01 -1.55364871e-01
-8.34580481e-01 4.42866713e-01 3.10623318e-01 -3.64039272e-01
-1.01346028e+00 -5.64459682e-01 -9.24228966e-01 5.22846282e-01
3.03946763e-01 -4.13830668e-01 1.22359920e+00 -9.90122139e-01
-1.00394678e+00 1.04151094e+00 -3.28397989e-01 -4.22630996e-01
2.52076864e-01 -2.00449124e-01 -3.12023431e-01 3.76139767e-02
2.82051653e-01 8.90013814e-01 4.35465336e-01 -1.15295839e+00
-5.00901461e-01 7.86152035e-02 2.14843884e-01 5.58269359e-02
-4.90990430e-01 3.61941516e-01 -1.06155050e+00 -8.57932687e-01
3.84010598e-02 -9.49665308e-01 -6.64119422e-02 -1.49919689e-01
-2.39758730e-01 -3.00227851e-01 4.54903513e-01 -5.27815104e-01
1.55752897e+00 -2.44814944e+00 4.82197218e-02 1.11220829e-01
1.29667550e-01 3.04763824e-01 -4.36540365e-01 1.96222991e-01
-6.72370791e-02 3.69169533e-01 -4.37778562e-01 -1.57480925e-01
-1.02743313e-01 -1.07726626e-01 -5.65486789e-01 3.32698822e-02
3.93447280e-01 1.06694245e+00 -9.46093619e-01 -4.62397426e-01
-2.54852295e-01 1.54796198e-01 -1.18780804e+00 1.89985275e-01
-3.65189672e-01 -1.72623739e-01 -4.97611940e-01 9.82727766e-01
3.37754995e-01 -6.66514814e-01 -1.32047027e-01 2.23868266e-01
-1.37813704e-03 4.16702271e-01 -6.00358546e-01 1.73336411e+00
-6.29805028e-01 6.65429056e-01 -1.48629919e-01 -7.80033231e-01
8.09830189e-01 -5.63087463e-02 9.05722007e-02 -1.00172031e+00
-2.05862045e-01 4.04922634e-01 -1.80977479e-01 -7.57716298e-01
6.86651886e-01 3.37594926e-01 -2.80472845e-01 4.21243519e-01
-4.09098640e-02 4.43214104e-02 2.68507987e-01 3.89711887e-01
1.33715689e+00 1.25630528e-01 9.95375775e-03 -3.29856724e-01
6.79306924e-01 3.37819532e-02 3.45627427e-01 9.34776485e-01
-9.26543921e-02 6.52482927e-01 5.07649601e-01 -2.64483482e-01
-8.22002172e-01 -8.65901828e-01 1.57891840e-01 1.49194193e+00
1.43924743e-01 -6.13524020e-01 -4.94897217e-01 -9.60582674e-01
-1.23944134e-01 5.63295484e-01 -4.98897076e-01 -3.86112958e-01
-7.44267046e-01 -8.43128741e-01 7.96371341e-01 7.09549308e-01
5.13166189e-01 -1.10367227e+00 -5.16085804e-01 -1.38587877e-01
-2.73085326e-01 -6.74042106e-01 -8.09186399e-01 4.72519904e-01
-7.10557938e-01 -1.08030438e+00 -6.46160066e-01 -1.23617208e+00
8.73652756e-01 3.83440405e-01 1.36478329e+00 5.85815370e-01
-2.10186958e-01 2.04814568e-01 -4.94160414e-01 1.50623294e-02
-4.74606514e-01 3.93922597e-01 -5.03821492e-01 -6.77736640e-01
2.71062165e-01 -2.01055244e-01 -5.40105283e-01 2.61706501e-01
-8.57017696e-01 -1.60908043e-01 7.75175989e-01 1.01745367e+00
3.06435853e-01 -3.20295215e-01 4.05896485e-01 -8.51673901e-01
8.56514931e-01 -8.55642617e-01 -7.46066570e-01 3.88487875e-01
-8.54455829e-01 4.21734035e-01 6.56714559e-01 -4.59432751e-01
-8.55252266e-01 1.09945886e-01 -2.45785281e-01 -4.69025016e-01
3.37998867e-01 7.92099953e-01 1.85983971e-01 -3.98494862e-02
1.00337362e+00 4.51523244e-01 -1.54175296e-01 -4.63755131e-01
1.39864996e-01 7.01941192e-01 4.88805473e-01 -8.34254980e-01
6.88948870e-01 1.63511634e-01 -7.27941275e-01 -2.56608725e-01
-4.63590741e-01 -5.42994320e-01 -4.19602059e-02 3.13142896e-01
5.76556325e-01 -1.17778134e+00 -4.19071943e-01 2.70855933e-01
-8.08740854e-01 -3.97640556e-01 7.52890781e-02 3.25832069e-01
-2.97436893e-01 3.87759387e-01 -7.90981174e-01 -5.67181647e-01
-4.35718000e-01 -1.58364284e+00 1.03945768e+00 1.63662165e-01
-1.60141960e-01 -7.79973626e-01 4.17214148e-02 2.74326444e-01
5.57827950e-01 -4.27826792e-01 1.23018074e+00 -7.94930220e-01
-7.94754326e-01 -1.42600328e-01 -3.60961765e-01 1.91774428e-01
-2.31973808e-02 1.36638850e-01 -9.43570375e-01 -4.52374816e-01
-2.43169188e-01 -7.41865516e-01 1.11314964e+00 8.12209677e-03
1.55930543e+00 -3.49609375e-01 -4.58998203e-01 6.34824693e-01
1.49306917e+00 4.48879033e-01 5.00019908e-01 3.77881587e-01
6.20845199e-01 2.12696731e-01 4.83235031e-01 2.65521020e-01
3.95166039e-01 6.85568094e-01 4.85145628e-01 6.65006191e-02
-1.50637850e-01 -3.41082186e-01 5.54056287e-01 9.37161148e-01
4.85455781e-01 -1.58583313e-01 -1.24008691e+00 7.92135179e-01
-1.73615634e+00 -8.05536091e-01 1.89308599e-01 2.13453817e+00
1.12485158e+00 2.07899541e-01 3.97700891e-02 -3.75415981e-01
5.74975431e-01 -2.87133697e-02 -5.20084083e-01 -2.53980726e-01
-2.37353034e-02 2.82836944e-01 3.49392354e-01 3.80567938e-01
-1.06536341e+00 9.98067737e-01 6.32820511e+00 9.10934865e-01
-1.18348229e+00 -7.96022564e-02 6.73298657e-01 -1.57695204e-01
-5.29063046e-01 8.66842493e-02 -6.37662828e-01 6.55385077e-01
6.29457533e-01 -1.85714662e-01 6.71577156e-01 1.24005938e+00
-3.44349444e-01 9.01884772e-03 -1.19569862e+00 1.04262233e+00
2.15189695e-01 -1.37149560e+00 -9.86585617e-02 -1.86329842e-01
9.14412796e-01 4.55322593e-01 2.97649235e-01 9.75823343e-01
5.36852956e-01 -8.73227119e-01 7.84371376e-01 1.18838109e-01
5.63953757e-01 -6.00309610e-01 5.52922785e-01 2.18472794e-01
-1.13956964e+00 -4.10281509e-01 -2.78747320e-01 1.47288918e-01
-3.90316874e-01 2.84868836e-01 -7.80794144e-01 1.22523800e-01
6.96869254e-01 5.35095632e-01 -1.03589547e+00 1.33933520e+00
-2.19828635e-01 6.92032397e-01 -3.90110016e-02 -2.21873268e-01
4.41883743e-01 2.59544939e-01 1.87030062e-01 1.46217239e+00
3.10773641e-01 -4.48738068e-01 2.16563076e-01 1.41350996e+00
-4.66499180e-01 4.16105837e-02 -3.25524688e-01 -1.52096003e-01
6.38581991e-01 1.00924838e+00 -6.02163970e-01 -4.47950512e-01
-7.04803228e-01 9.11888480e-01 5.04557610e-01 4.56401378e-01
-9.57593918e-01 -5.23975670e-01 3.27279836e-01 -2.26312563e-01
4.29498315e-01 1.37262374e-01 -1.61970705e-01 -1.21592081e+00
3.98987174e-01 -1.13313150e+00 5.25401115e-01 -6.64859831e-01
-9.92230296e-01 6.92243218e-01 -3.49949040e-02 -1.23009586e+00
-3.92690450e-01 -2.66445756e-01 -5.52014947e-01 9.10874188e-01
-1.50133336e+00 -7.43304551e-01 -2.45427176e-01 3.05137038e-01
7.62064219e-01 -5.41956902e-01 6.88950956e-01 5.36342740e-01
-5.49112380e-01 1.00561202e+00 1.06136151e-01 3.85645807e-01
5.86473405e-01 -1.21762633e+00 6.25404477e-01 9.35318649e-01
2.34819815e-01 1.00178373e+00 2.71511942e-01 -6.30001843e-01
-1.41293073e+00 -1.03040659e+00 6.50720298e-01 -4.66760129e-01
6.11333847e-01 -4.24014896e-01 -1.07755899e+00 5.96666515e-01
1.23110414e-01 1.53372675e-01 4.76780504e-01 1.11395001e-01
-7.75701284e-01 1.17305361e-01 -9.29478824e-01 5.64123929e-01
7.59714425e-01 -8.51113141e-01 -4.77868795e-01 2.57085145e-01
8.35470498e-01 -5.21595180e-01 -3.52135569e-01 3.78043652e-01
4.00992483e-01 -9.34402585e-01 8.74927342e-01 -4.39200312e-01
7.72699893e-01 -1.60280809e-01 -2.24303484e-01 -9.92446125e-01
-2.68916160e-01 -3.22220176e-01 -5.28060421e-02 1.19252980e+00
6.51803732e-01 -2.51749873e-01 8.14912319e-01 5.57228982e-01
-3.35286468e-01 -9.77384329e-01 -4.44253862e-01 -7.90729225e-01
2.08940342e-01 -3.12138587e-01 5.00425339e-01 9.73715961e-01
2.25875184e-01 2.96613216e-01 -1.21310346e-01 -7.92968199e-02
8.49849805e-02 2.88519531e-01 4.88458633e-01 -7.19812334e-01
-5.90048313e-01 -7.65707314e-01 1.48728974e-02 -1.35001385e+00
4.19123322e-01 -1.38174748e+00 1.23576880e-01 -1.16646934e+00
8.54219258e-01 -6.17373586e-01 -3.63113940e-01 6.97541595e-01
-5.33451796e-01 -3.06014232e-02 5.41639477e-02 4.82565284e-01
-7.21426845e-01 5.39901257e-01 6.52681828e-01 -4.11512256e-01
-2.64100075e-01 2.52830936e-03 -9.02712107e-01 4.91840035e-01
7.31000543e-01 -4.27495420e-01 -6.38231277e-01 -7.83313572e-01
4.90935832e-01 -1.60121955e-02 2.65261739e-01 -7.04645872e-01
2.70129889e-01 1.36229411e-01 6.99901804e-02 -3.35383832e-01
-1.43731758e-01 -6.13705575e-01 -3.62107545e-01 6.23142481e-01
-6.75119519e-01 4.40123469e-01 5.25803506e-01 3.70019883e-01
-3.28038633e-01 -6.54194951e-01 6.56708479e-01 -3.29026610e-01
-8.33555758e-01 9.51442216e-03 -2.80321985e-01 5.26321292e-01
5.54809690e-01 -6.21053670e-03 -3.84409785e-01 -1.70242667e-01
-1.68989882e-01 4.58677530e-01 5.76816142e-01 7.50652015e-01
6.89124763e-01 -1.37398696e+00 -4.29874957e-01 2.81165123e-01
5.52857816e-01 -1.61338389e-01 -1.68731570e-01 4.88845766e-01
-5.16929269e-01 4.99017715e-01 1.32630378e-01 -5.37268341e-01
-1.30347896e+00 5.80954194e-01 2.38373697e-01 -2.37078890e-01
-2.21193999e-01 1.18669653e+00 2.49117136e-01 -5.83675265e-01
5.28314173e-01 -5.40868163e-01 -1.11289518e-02 -2.38779783e-01
3.93862456e-01 1.53239459e-01 4.33495715e-02 -3.06618690e-01
-5.34581602e-01 3.49007875e-01 -4.88272399e-01 2.08435372e-01
1.25120533e+00 1.73082754e-01 -1.83825210e-01 7.91906565e-02
1.53166914e+00 3.41279417e-01 -1.04465199e+00 -1.73498094e-01
2.59939611e-01 -4.25849050e-01 -1.67361438e-01 -1.01481593e+00
-1.33284283e+00 7.07436025e-01 6.72506630e-01 -8.13227706e-03
1.14637828e+00 2.99007356e-01 5.30502617e-01 4.08545434e-01
1.57529026e-01 -7.96539187e-01 4.30568784e-01 5.68359315e-01
7.12820530e-01 -1.36168563e+00 -1.74982712e-01 -1.75166979e-01
-4.95786816e-01 8.92319202e-01 8.99187446e-01 6.95066452e-02
2.76977360e-01 3.80123407e-01 -5.25457002e-02 -3.39691252e-01
-9.61141467e-01 -5.00041246e-02 3.00381839e-01 1.24218501e-01
7.65768170e-01 -2.39837646e-01 -6.54165214e-03 4.75899875e-01
-1.73108913e-02 -2.33640134e-01 3.78240943e-01 1.09693599e+00
-3.86296481e-01 -1.15202880e+00 -7.13163838e-02 5.25204301e-01
-5.39436400e-01 -6.57211602e-01 -4.29137915e-01 6.56130552e-01
-3.22376460e-01 7.88164616e-01 -7.34212324e-02 -4.88426447e-01
1.87085688e-01 2.45363880e-02 2.19827116e-01 -7.93731987e-01
-8.45525384e-01 1.50270406e-02 -3.31956804e-01 -4.91635948e-01
-2.79633421e-02 -5.14287174e-01 -1.43180597e+00 1.49307176e-01
-5.13240993e-01 3.07863057e-01 4.89007175e-01 5.24034739e-01
6.28690243e-01 5.43555856e-01 5.45848668e-01 -1.93239987e-01
-5.99771798e-01 -7.62204349e-01 7.40176886e-02 3.75286907e-01
4.12019968e-01 -4.32014585e-01 -5.38796365e-01 -9.67114121e-02] | [7.553072929382324, 8.002159118652344] |
207ddc76-4e4e-441c-a9d3-02eb84354d42 | novel-visual-category-discovery-with-dual | 2107.03358 | null | https://arxiv.org/abs/2107.03358v2 | https://arxiv.org/pdf/2107.03358v2.pdf | Novel Visual Category Discovery with Dual Ranking Statistics and Mutual Knowledge Distillation | In this paper, we tackle the problem of novel visual category discovery, i.e., grouping unlabelled images from new classes into different semantic partitions by leveraging a labelled dataset that contains images from other different but relevant categories. This is a more realistic and challenging setting than conventional semi-supervised learning. We propose a two-branch learning framework for this problem, with one branch focusing on local part-level information and the other branch focusing on overall characteristics. To transfer knowledge from the labelled data to the unlabelled, we propose using dual ranking statistics on both branches to generate pseudo labels for training on the unlabelled data. We further introduce a mutual knowledge distillation method to allow information exchange and encourage agreement between the two branches for discovering new categories, allowing our model to enjoy the benefits of global and local features. We comprehensively evaluate our method on public benchmarks for generic object classification, as well as the more challenging datasets for fine-grained visual recognition, achieving state-of-the-art performance. | ['Kai Han', 'Bingchen Zhao'] | 2021-07-07 | null | http://proceedings.neurips.cc/paper/2021/hash/c203d8a151612acf12457e4d67635a95-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/c203d8a151612acf12457e4d67635a95-Paper.pdf | neurips-2021-12 | ['fine-grained-visual-recognition'] | ['computer-vision'] | [ 5.36114156e-01 3.64620984e-01 -4.65814024e-01 -6.66332126e-01
-8.79326582e-01 -8.03139448e-01 7.84908116e-01 2.86771536e-01
-2.02149108e-01 6.62675142e-01 1.67736217e-01 -8.53129402e-02
-1.53997079e-01 -5.95607877e-01 -6.42632365e-01 -9.78457570e-01
1.13821916e-01 7.81954408e-01 3.18802327e-01 4.01693791e-01
1.50880188e-01 2.97255546e-01 -1.96565926e+00 7.02234983e-01
4.91162717e-01 1.34668589e+00 9.83848702e-03 2.87026465e-01
-9.33749303e-02 1.01018465e+00 -2.53938317e-01 -4.23286319e-01
2.40977287e-01 -2.28692576e-01 -1.41143906e+00 6.22516334e-01
5.76279283e-01 4.33251485e-02 1.45498112e-01 9.69239712e-01
2.51585543e-01 9.27414149e-02 1.03504801e+00 -1.28992617e+00
-4.27627295e-01 7.70725250e-01 -6.64322555e-01 -9.19707343e-02
-2.36489922e-02 4.28920984e-02 1.16675770e+00 -6.91200078e-01
6.60555363e-01 1.22424638e+00 5.66397369e-01 6.18983805e-01
-1.45448506e+00 -6.68693185e-01 5.84764242e-01 4.35772717e-01
-1.49086165e+00 -4.19021815e-01 6.39725149e-01 -7.73945034e-01
3.59392315e-01 3.34567994e-01 1.92516148e-01 9.30577695e-01
-4.72564846e-01 1.00494266e+00 1.54318261e+00 -3.77738386e-01
2.05003202e-01 4.73088026e-01 4.30509657e-01 3.52241784e-01
2.31210083e-01 7.58853927e-02 -3.19371998e-01 -9.59879905e-02
1.68003321e-01 9.36795473e-02 -2.58785039e-01 -1.11474860e+00
-1.32269764e+00 8.06967378e-01 7.99180388e-01 2.25829750e-01
-2.75706083e-01 -9.62482765e-02 3.60102326e-01 8.26594383e-02
5.82882702e-01 4.88957584e-01 -7.50988901e-01 3.78764421e-01
-8.82653713e-01 -1.81400388e-01 6.53138041e-01 7.12373614e-01
1.04554284e+00 -7.02636719e-01 -5.07365763e-01 1.02711928e+00
6.25138283e-01 2.01788962e-01 4.86392885e-01 -7.83405125e-01
2.31880859e-01 7.12326765e-01 -1.06637500e-01 -5.78615665e-01
-2.60595232e-01 -9.22549665e-01 -9.12629664e-01 4.07809876e-02
4.48280066e-01 3.35533589e-01 -1.40670168e+00 1.62794220e+00
4.74150807e-01 2.95552373e-01 2.54926961e-02 7.58859634e-01
1.26954019e+00 3.85550737e-01 1.47355199e-01 -1.02380909e-01
1.32273674e+00 -1.30910373e+00 -8.77355263e-02 -2.00267658e-01
6.02484763e-01 -4.66731519e-01 5.32995164e-01 4.19528276e-01
-4.76935565e-01 -6.92596555e-01 -6.83778942e-01 2.03891724e-01
-4.33674335e-01 1.97877645e-01 5.90948641e-01 3.41069281e-01
-1.00691235e+00 4.57484692e-01 -3.65116924e-01 -3.42082500e-01
9.40738440e-01 2.51810670e-01 -5.32975554e-01 -4.07938629e-01
-7.41801143e-01 5.09525418e-01 8.05107892e-01 -1.04059637e-01
-1.10539567e+00 -5.89178860e-01 -7.07810700e-01 -7.06167053e-03
3.64069968e-01 -5.89895368e-01 1.03809595e+00 -1.07742965e+00
-1.05650318e+00 1.43021059e+00 1.06685720e-01 -3.30950171e-01
4.06622350e-01 2.79373705e-01 -5.48344478e-02 -2.60069445e-02
5.27797401e-01 1.11305404e+00 8.58870268e-01 -1.84670222e+00
-9.49730039e-01 -4.31163758e-01 2.56700125e-02 2.52041221e-01
-2.16362223e-01 -4.52290893e-01 -5.43387055e-01 -5.34944236e-01
3.32387447e-01 -9.87133145e-01 -2.63142735e-01 -1.78708032e-01
-6.22683942e-01 -6.28294051e-01 5.50994217e-01 -2.64206558e-01
7.88981676e-01 -2.07947516e+00 2.12517336e-01 3.76038790e-01
5.64123094e-01 2.36026868e-01 -1.87573031e-01 -1.08756021e-01
-2.13050023e-01 2.05281097e-02 -1.79274961e-01 -3.47337812e-01
-5.69049306e-02 2.04606593e-01 -4.09718722e-01 1.38719916e-01
2.30178803e-01 1.04557693e+00 -1.04340839e+00 -4.03803170e-01
1.83290280e-02 1.11248553e-01 -4.32788759e-01 2.15207696e-01
-6.56230841e-03 8.69794130e-01 -2.95123130e-01 6.97760880e-01
7.63925076e-01 -7.17706501e-01 3.61699075e-01 -4.10372406e-01
3.11634392e-01 1.55818462e-01 -1.12085915e+00 1.38644993e+00
-3.56982350e-01 2.69992888e-01 -3.73461306e-01 -1.32154763e+00
8.86674106e-01 -1.61758959e-02 2.42048576e-01 -4.71650869e-01
9.21427086e-02 1.26229897e-01 -2.57557660e-01 -1.77966014e-01
1.11783452e-01 -2.46300787e-01 -1.96955681e-01 4.57503676e-01
4.34398979e-01 -3.09057906e-02 1.47924826e-01 1.51555091e-01
7.70674825e-01 1.70891788e-02 4.39414233e-01 -2.11023882e-01
3.78487408e-01 -8.22820812e-02 4.45702523e-01 1.12432182e+00
-2.00253263e-01 5.55963576e-01 3.18488002e-01 -3.98499876e-01
-6.35369301e-01 -1.17449057e+00 -3.17600220e-01 1.21845782e+00
3.61436576e-01 -1.33938015e-01 -3.95938605e-01 -1.48315907e+00
2.54395723e-01 3.76690626e-01 -8.69955957e-01 -2.72093326e-01
7.13006184e-02 -6.40384555e-01 2.69801438e-01 4.92817879e-01
4.94967282e-01 -9.37257290e-01 -2.59435307e-02 -1.17733724e-01
-5.05848706e-01 -1.01618445e+00 -9.72098038e-02 6.32732809e-01
-7.57442415e-01 -1.08819616e+00 -7.91868925e-01 -1.00284219e+00
9.36834335e-01 4.70216006e-01 1.40872455e+00 -5.58141507e-02
-2.38266811e-01 3.19580436e-01 -5.35372257e-01 -8.98757577e-02
-2.84546524e-01 8.00689310e-02 -2.46136323e-01 5.25434315e-01
3.81178916e-01 -2.19074160e-01 -4.71663833e-01 6.63759112e-01
-7.71898389e-01 2.24836886e-01 7.86802411e-01 1.06359017e+00
8.22837293e-01 2.03794375e-01 5.89292705e-01 -1.04639947e+00
-1.36470884e-01 -7.12491810e-01 -2.05656916e-01 3.92951548e-01
-5.05921543e-01 1.93972141e-01 2.64973938e-01 -5.42954803e-01
-9.99630272e-01 2.57329315e-01 2.03176439e-01 -2.42398664e-01
-5.85203290e-01 3.89127672e-01 -3.07039410e-01 -1.41393915e-01
5.84564328e-01 2.10302979e-01 -2.39924550e-01 -6.63396478e-01
7.70433068e-01 9.85741198e-01 6.15380347e-01 -5.47189236e-01
6.92070842e-01 6.55273318e-01 -2.80653059e-01 -3.88430119e-01
-1.55926454e+00 -1.18729973e+00 -1.03634071e+00 -1.38656929e-01
6.88346386e-01 -1.11316681e+00 -3.27836424e-01 5.01177490e-01
-8.04712951e-01 -3.22405100e-01 -6.25242054e-01 3.48330021e-01
-5.57806790e-01 4.10191208e-01 -1.75192937e-01 -3.55689436e-01
3.63460146e-02 -1.06442988e+00 1.43849862e+00 1.74959898e-01
8.43382925e-02 -9.58438337e-01 1.00047164e-01 7.87992299e-01
4.85992432e-02 1.31127834e-01 6.85816467e-01 -9.88135934e-01
-5.51590800e-01 -1.05584502e-01 -7.44633496e-01 5.86784184e-01
2.47020185e-01 -5.74094892e-01 -1.22041404e+00 -4.45038050e-01
-3.43375057e-01 -9.12065506e-01 1.58120859e+00 1.21537611e-01
1.20408404e+00 -2.28155002e-01 -7.18984544e-01 4.88702536e-01
1.14331126e+00 -1.65323466e-01 3.61284435e-01 2.77808726e-01
9.54646111e-01 7.77375460e-01 7.36511946e-01 2.57964253e-01
6.93947256e-01 7.69215941e-01 4.53818500e-01 -2.01057509e-01
-4.04182792e-01 -5.00921793e-02 -8.81490931e-02 6.51111901e-01
2.86583696e-02 -1.43807873e-01 -9.20058966e-01 6.33920908e-01
-1.94308543e+00 -6.97706997e-01 1.11382574e-01 2.22179914e+00
1.08137465e+00 -4.66964068e-03 8.09675977e-02 1.03035904e-01
9.76408422e-01 -3.00914943e-02 -5.74790537e-01 1.04459010e-01
2.82568820e-02 1.34685263e-01 4.65671062e-01 7.74394125e-02
-1.74597192e+00 8.93504381e-01 5.55129433e+00 1.16719437e+00
-9.62121189e-01 6.91465363e-02 9.53556240e-01 2.95482546e-01
-1.15776934e-01 6.14994057e-02 -9.09294188e-01 3.21773052e-01
5.41784585e-01 2.23135605e-01 4.59774584e-02 8.69218528e-01
-5.50178945e-01 -7.62619451e-02 -1.18262851e+00 9.58026052e-01
2.38004550e-01 -1.18341434e+00 4.83513772e-01 3.86669077e-02
1.10530078e+00 9.57686529e-02 -1.55096818e-02 3.76246274e-01
5.92787087e-01 -9.46078122e-01 7.33245909e-01 5.62687695e-01
8.86440754e-01 -4.91885126e-01 9.18081462e-01 3.18596303e-01
-1.26779163e+00 -1.74685434e-01 -1.59951523e-01 2.31011808e-01
-2.89690405e-01 7.77102292e-01 -8.87816727e-01 8.51093709e-01
9.18900371e-01 1.19221807e+00 -1.00820005e+00 1.31943870e+00
-2.08300263e-01 6.99057519e-01 -5.38927764e-02 3.64427805e-01
2.53112376e-01 1.96715832e-01 1.09397843e-01 1.08471453e+00
-1.21136382e-01 -3.03707302e-01 5.25472879e-01 4.75216240e-01
-3.12049687e-01 -4.91075590e-02 -3.54918838e-01 1.56687841e-01
1.52499616e-01 1.40254557e+00 -1.12243521e+00 -6.56855643e-01
-2.08847582e-01 8.70155573e-01 5.76723516e-01 2.20564142e-01
-6.13590419e-01 -2.34242141e-01 3.48165751e-01 -1.71405032e-01
5.95320344e-01 3.20189685e-01 -2.15540305e-02 -1.32731283e+00
1.06322234e-02 -7.52264202e-01 7.76117861e-01 -5.17928541e-01
-1.82061887e+00 6.30019844e-01 3.11246831e-02 -1.37736106e+00
-8.74237642e-02 -6.67275786e-01 -4.53677848e-02 5.24526179e-01
-1.87120736e+00 -1.54575431e+00 -5.11582077e-01 4.61858839e-01
3.71253192e-01 -3.71357873e-02 7.13398695e-01 2.85462022e-01
-1.19246319e-01 6.23363197e-01 2.36085847e-01 1.66691586e-01
9.15752411e-01 -1.32671559e+00 5.91935106e-02 5.00891507e-01
3.75872731e-01 1.43727958e-01 2.26854041e-01 -4.25468266e-01
-5.06259084e-01 -1.59045792e+00 7.77241170e-01 -7.54226923e-01
5.40177703e-01 -4.48827595e-01 -8.72781396e-01 5.55929720e-01
-3.24955016e-01 3.19732964e-01 8.63731802e-01 1.79773286e-01
-9.51072156e-01 -3.91945504e-02 -1.03749275e+00 4.28198203e-02
1.17294431e+00 -4.98341769e-01 -7.22681880e-01 4.22785848e-01
5.84368348e-01 9.80334636e-03 -7.00264096e-01 8.60632002e-01
4.92451131e-01 -7.50757515e-01 1.08246326e+00 -7.30431557e-01
4.20168459e-01 -5.26537716e-01 -2.96053737e-01 -1.42638421e+00
-6.27744198e-01 -4.72263396e-02 1.18681215e-01 1.46747506e+00
3.43604743e-01 -6.46344006e-01 7.29114890e-01 -1.42803207e-01
9.34405997e-02 -5.92154980e-01 -7.22820163e-01 -9.93333459e-01
1.02463514e-01 -2.50140041e-01 3.29693407e-01 1.08247626e+00
-3.16007882e-01 5.54796040e-01 -2.86224395e-01 5.95781840e-02
9.06934857e-01 7.03480482e-01 7.17116654e-01 -1.69371426e+00
-3.41837406e-01 -4.82195139e-01 -7.82367527e-01 -1.18653166e+00
2.85492241e-01 -1.52463555e+00 4.42392707e-01 -1.58013570e+00
1.10917008e+00 -8.11116278e-01 -6.21771753e-01 7.93763995e-01
-4.10523355e-01 1.00188267e+00 1.67808384e-01 5.18470883e-01
-1.36225224e+00 4.61925149e-01 1.08714247e+00 -6.02931499e-01
5.45581430e-02 7.75789991e-02 -1.08998561e+00 6.83808863e-01
4.55629140e-01 -4.92539078e-01 -3.84329528e-01 -5.52366860e-02
-2.29583889e-01 -5.70197821e-01 6.70298278e-01 -8.97531569e-01
6.65960535e-02 -3.73527827e-03 3.38464320e-01 -5.24529934e-01
-2.24232767e-02 -6.55575335e-01 4.38700728e-02 1.98468685e-01
-6.30157590e-01 -9.44544435e-01 -7.34690130e-02 8.51849377e-01
-3.13545644e-01 4.61960286e-02 9.66095567e-01 -3.11969798e-02
-1.00503671e+00 4.55736935e-01 8.26183483e-02 1.79136366e-01
1.06492901e+00 -1.84480429e-01 -4.25436527e-01 -1.29080698e-01
-1.15625703e+00 2.91223377e-01 4.54169452e-01 6.23022616e-01
3.46102685e-01 -1.45305443e+00 -8.01086843e-01 1.58840314e-01
8.75032723e-01 5.06646633e-02 3.57594848e-01 7.43901670e-01
1.37722790e-01 4.78391320e-01 -1.44276753e-01 -1.13166070e+00
-1.43397772e+00 7.32399106e-01 1.39006466e-01 -6.79727554e-01
-2.23315939e-01 1.05252397e+00 9.07055676e-01 -1.01534081e+00
2.56187022e-01 9.46161374e-02 -6.13599598e-01 3.58063608e-01
3.93877536e-01 7.17855841e-02 1.47914276e-01 -8.96717846e-01
-5.64430475e-01 7.23211646e-01 -2.99599171e-01 4.18236941e-01
1.08419740e+00 -3.95101815e-01 -3.17780197e-01 5.84940732e-01
1.36964083e+00 -2.93290526e-01 -1.22809958e+00 -7.46286929e-01
3.12240392e-01 -4.22845244e-01 -8.77960585e-03 -1.21414483e+00
-9.74541843e-01 6.26899004e-01 7.48603046e-01 2.29137257e-01
1.07934952e+00 8.34492922e-01 2.81569183e-01 2.77896464e-01
5.86872935e-01 -8.17055404e-01 2.62457937e-01 5.80503047e-01
6.00789547e-01 -1.54152977e+00 -1.18657283e-01 -6.09500468e-01
-6.32442296e-01 9.33801711e-01 3.98389637e-01 3.00021619e-01
8.07793438e-01 -3.58761400e-01 1.77929014e-01 -1.63642555e-01
-7.31405258e-01 -7.61331499e-01 8.45085204e-01 8.01385820e-01
1.27834648e-01 3.93639237e-01 1.08651303e-01 5.63019156e-01
1.44540727e-01 -8.54424238e-02 1.07350044e-01 7.44354486e-01
-4.37274158e-01 -1.14089477e+00 -2.09177569e-01 8.83222163e-01
-2.32869998e-01 -8.95336643e-02 -5.46384633e-01 4.14799273e-01
5.69332778e-01 1.04468215e+00 2.08813548e-01 -4.37468082e-01
-4.38346937e-02 -3.33287716e-02 4.39466029e-01 -1.08044553e+00
-1.81189939e-01 -1.33799523e-01 7.74772614e-02 -5.06997764e-01
-7.59006202e-01 -5.55391252e-01 -7.62410879e-01 3.12560380e-01
-5.13914883e-01 3.20200264e-01 4.28100735e-01 1.04829752e+00
4.86219764e-01 5.36889672e-01 1.00574875e+00 -1.04149318e+00
-4.05151516e-01 -8.01049054e-01 -6.61452115e-01 6.94562495e-01
3.04740340e-01 -1.03639853e+00 -6.27713084e-01 1.91354483e-01] | [9.689432144165039, 2.883833646774292] |
85c4c2d6-27cc-4f46-aff2-9a0023ae7a23 | unsupervised-sentence-compression-using | 1809.02669 | null | http://arxiv.org/abs/1809.02669v1 | http://arxiv.org/pdf/1809.02669v1.pdf | Unsupervised Sentence Compression using Denoising Auto-Encoders | In sentence compression, the task of shortening sentences while retaining the
original meaning, models tend to be trained on large corpora containing pairs
of verbose and compressed sentences. To remove the need for paired corpora, we
emulate a summarization task and add noise to extend sentences and train a
denoising auto-encoder to recover the original, constructing an end-to-end
training regime without the need for any examples of compressed sentences. We
conduct a human evaluation of our model on a standard text summarization
dataset and show that it performs comparably to a supervised baseline based on
grammatical correctness and retention of meaning. Despite being exposed to no
target data, our unsupervised models learn to generate imperfect but reasonably
readable sentence summaries. Although we underperform supervised models based
on ROUGE scores, our models are competitive with a supervised baseline based on
human evaluation for grammatical correctness and retention of meaning. | ['Thibault Févry', 'Jason Phang'] | 2018-09-07 | unsupervised-sentence-compression-using-1 | https://aclanthology.org/K18-1040 | https://aclanthology.org/K18-1040.pdf | conll-2018-10 | ['unsupervised-abstractive-sentence-compression'] | ['natural-language-processing'] | [ 8.04146826e-01 9.22759116e-01 2.51955409e-02 -6.93545043e-01
-1.40009916e+00 -6.12448215e-01 5.81850827e-01 5.19748032e-01
-4.38011140e-01 8.51280451e-01 1.01763582e+00 -1.82306319e-01
3.02474290e-01 -5.50932527e-01 -9.09021854e-01 -6.51445463e-02
2.21342668e-01 4.52603161e-01 -2.72985488e-01 -3.07496607e-01
5.02315998e-01 -2.64761835e-01 -1.32190692e+00 5.34059346e-01
1.26840329e+00 3.43258679e-01 2.86094666e-01 1.07961762e+00
-9.39690974e-03 8.70912731e-01 -1.01048422e+00 -7.33456135e-01
-1.98681038e-02 -8.73983502e-01 -1.03815317e+00 2.68607229e-01
9.27238226e-01 -4.25486088e-01 -4.18195635e-01 9.47590172e-01
5.45060694e-01 6.12649545e-02 6.60306215e-01 -4.77662712e-01
-7.73207247e-01 9.51667845e-01 2.34445697e-03 4.98424843e-02
8.05626571e-01 7.54692545e-03 1.38502192e+00 -8.30172122e-01
9.04626489e-01 1.27333713e+00 7.60592043e-01 4.59425300e-01
-1.41210806e+00 3.45354453e-02 -2.93362346e-02 -3.22930604e-01
-6.30355537e-01 -9.81221735e-01 4.08880889e-01 1.88914031e-01
1.62136757e+00 4.48352307e-01 4.27559346e-01 1.14373660e+00
4.17437404e-01 1.09304440e+00 2.75146872e-01 -5.55948198e-01
2.29843482e-01 -1.68888196e-01 2.42262602e-01 3.04600924e-01
3.36194426e-01 -3.38201374e-01 -5.30822814e-01 -9.68262032e-02
-1.30313665e-01 -3.49240959e-01 -4.88424301e-01 1.03077292e-01
-1.04875731e+00 7.28965104e-01 2.21388415e-01 1.85135543e-01
-4.75968093e-01 9.19047222e-02 7.56207466e-01 5.81940234e-01
7.90249050e-01 8.00180674e-01 -2.79887557e-01 -3.30973297e-01
-1.56534767e+00 5.37255585e-01 1.11736476e+00 1.33480883e+00
5.14236391e-01 8.93456936e-02 -3.72987658e-01 8.55859220e-01
-1.81792334e-01 4.93862391e-01 6.98255122e-01 -1.18431699e+00
8.04588199e-01 3.25100809e-01 -9.53076407e-02 -7.52700090e-01
-1.48814976e-01 -4.98001188e-01 -9.26647067e-01 -4.53406364e-01
-1.47725657e-01 -2.57347375e-01 -6.76670611e-01 1.71651733e+00
-4.22732890e-01 -2.11520210e-01 5.15965044e-01 3.96171540e-01
9.07322466e-01 8.55464399e-01 -9.66275707e-02 -6.30733728e-01
7.87243128e-01 -1.06300581e+00 -7.55387127e-01 -5.53795040e-01
1.01112247e+00 -7.16169178e-01 1.26055598e+00 2.50784308e-01
-1.67470777e+00 -4.37632501e-01 -1.18118334e+00 -5.39196849e-01
2.12429315e-01 -1.15774376e-02 3.02222520e-01 2.90491581e-01
-1.27540374e+00 1.12304080e+00 -7.49596715e-01 -5.42029381e-01
3.99281889e-01 3.21217254e-02 -4.92424607e-01 -8.44506100e-02
-8.91750515e-01 9.09351289e-01 6.20984614e-01 -2.26486474e-01
-5.06010473e-01 -6.49159312e-01 -1.26707804e+00 3.02938133e-01
-6.39574864e-06 -1.13671637e+00 1.69800031e+00 -1.04352605e+00
-1.30372858e+00 6.90396488e-01 -3.81449193e-01 -9.30723608e-01
3.75441641e-01 -4.33319628e-01 -4.18223515e-02 4.39481944e-01
3.38146418e-01 7.66315758e-01 6.80577755e-01 -9.34367299e-01
-4.17104244e-01 -1.46010648e-02 -1.42673925e-01 4.30931836e-01
-3.76236916e-01 -1.58533022e-01 -7.94298574e-02 -9.29430604e-01
-1.80879891e-01 -4.72835094e-01 -1.80010587e-01 -6.74255192e-01
-6.81422710e-01 -1.14983924e-01 5.88481426e-01 -1.03717053e+00
1.37188733e+00 -1.84925854e+00 3.23869556e-01 -3.17616642e-01
7.73509070e-02 5.78505322e-02 -6.62782669e-01 7.90088058e-01
-1.50828538e-02 2.91336179e-01 -8.38442922e-01 -8.79460871e-01
-9.49315429e-02 1.45270020e-01 -5.40328979e-01 7.74156973e-02
4.82429475e-01 9.89535511e-01 -1.11791134e+00 -4.88501638e-01
-1.90775231e-01 -3.09314616e-02 -8.76662016e-01 4.25634056e-01
-3.03778827e-01 2.41540857e-02 8.15189257e-03 2.56676942e-01
3.24334532e-01 7.72801181e-03 1.22842059e-01 1.36646748e-01
2.81652838e-01 8.91169965e-01 -5.19452333e-01 2.07582140e+00
-5.26815176e-01 7.02048659e-01 -6.23776689e-02 -1.09697044e+00
8.61323237e-01 3.17325294e-01 6.61740676e-02 -5.46853840e-01
-5.38942812e-04 1.90332308e-01 -2.10046649e-01 -5.11281490e-01
1.12276101e+00 -3.15207273e-01 -4.53930616e-01 8.17395747e-01
4.61720020e-01 -8.25035751e-01 6.26239598e-01 8.22875977e-01
1.46443009e+00 3.37609798e-02 1.76409826e-01 -1.78786278e-01
3.06232274e-01 5.92823140e-02 2.36156076e-01 9.45960522e-01
2.58309096e-01 9.88537788e-01 6.01411402e-01 -5.24078682e-02
-1.55938792e+00 -9.27218974e-01 1.09185837e-01 9.25488114e-01
-2.72137046e-01 -8.53873789e-01 -1.08665204e+00 -6.49493694e-01
-2.45968774e-01 1.53189766e+00 -1.13567963e-01 -4.27898109e-01
-7.07080305e-01 -4.74467903e-01 6.92859828e-01 5.48864663e-01
1.53539717e-01 -1.29554915e+00 -4.68668312e-01 3.53978693e-01
-6.11333489e-01 -1.01722932e+00 -6.67278230e-01 7.70778954e-02
-1.11705458e+00 -7.55125761e-01 -4.11204427e-01 -9.59985018e-01
5.67500055e-01 3.51039357e-02 1.69107509e+00 1.12952717e-01
1.16591275e-01 3.24308544e-01 -5.01992702e-01 -5.83637118e-01
-1.10740221e+00 5.16121984e-01 -8.27322602e-02 -6.60116911e-01
1.52691767e-01 -7.05611885e-01 -3.23908895e-01 -4.22107458e-01
-1.14396799e+00 2.35598326e-01 6.84560359e-01 9.91901875e-01
4.48418051e-01 -3.29287320e-01 1.06383121e+00 -1.12598062e+00
1.18483603e+00 -2.41346538e-01 1.48576334e-01 3.52177680e-01
-4.70836341e-01 2.27168053e-01 8.13294172e-01 2.36425921e-02
-1.01429558e+00 -8.40550661e-02 -3.50874841e-01 1.09969050e-01
-1.25245452e-01 6.88869894e-01 -6.72064349e-02 7.95790672e-01
7.52146840e-01 4.15396631e-01 6.15056716e-02 -4.31231350e-01
6.86481655e-01 9.67206001e-01 1.06270576e+00 -3.45833361e-01
4.87644345e-01 -5.02230413e-02 -4.77620512e-01 -7.84231663e-01
-1.20716226e+00 -2.20541552e-01 -5.91495872e-01 2.85453707e-01
5.39083660e-01 -9.95675206e-01 2.26590663e-01 -1.65164657e-02
-1.43913734e+00 -8.97332206e-02 -8.42108607e-01 2.88002610e-01
-1.01205325e+00 9.68270600e-01 -7.67939568e-01 -4.16028976e-01
-1.09796405e+00 -5.49878836e-01 1.29631090e+00 -1.04035228e-01
-7.89331377e-01 -9.11239207e-01 1.25134990e-01 2.59232521e-01
4.07063961e-01 1.64357759e-02 1.03575838e+00 -9.15503800e-01
-4.60034683e-02 -5.84918141e-01 8.00154880e-02 7.62275040e-01
1.08873554e-01 -2.82174349e-01 -7.95686483e-01 -4.06101763e-01
2.71432549e-01 -7.84307063e-01 1.33484185e+00 3.77230048e-01
8.92366529e-01 -7.54862607e-01 1.31029069e-01 3.86586696e-01
1.05026245e+00 -4.07974064e-01 7.45981991e-01 2.75668409e-02
3.43921155e-01 7.38233805e-01 4.40568626e-01 2.51813084e-01
3.35175663e-01 -8.89793038e-02 1.32874042e-01 1.18908472e-01
-2.67477423e-01 -6.80951476e-01 5.77112913e-01 1.21967721e+00
3.14516902e-01 -7.31136143e-01 -4.71234530e-01 7.41637766e-01
-1.75732803e+00 -1.27260411e+00 -2.44112639e-03 2.14324975e+00
1.16547024e+00 3.83356541e-01 -8.71049333e-03 4.07806076e-02
5.84704936e-01 3.75943244e-01 -3.99434507e-01 -8.20127070e-01
-3.98627490e-01 3.04489285e-01 1.72698051e-01 7.08294392e-01
-8.73757064e-01 1.04390264e+00 7.40936947e+00 7.16055751e-01
-6.76037252e-01 -1.82698771e-01 7.02784657e-01 -3.82557690e-01
-7.17073560e-01 1.92805409e-01 -4.87163424e-01 2.63446927e-01
1.31632364e+00 -5.43272138e-01 2.56191701e-01 5.77613890e-01
3.83724570e-01 -1.75643176e-01 -1.47267199e+00 6.24157846e-01
4.53409255e-01 -1.32221210e+00 3.69381815e-01 -3.40549976e-01
7.47695923e-01 3.77499702e-04 -4.45830941e-01 5.01815796e-01
2.41480261e-01 -1.24550366e+00 6.97603643e-01 6.25388443e-01
7.82774389e-01 -6.05572164e-01 1.00828028e+00 5.81830144e-01
-6.02923751e-01 1.28942326e-01 -6.86928809e-01 -3.23915958e-01
5.77029288e-01 6.62760079e-01 -8.09549391e-01 7.15570748e-01
9.62199643e-02 1.12698638e+00 -7.63492405e-01 7.53953576e-01
-3.62295717e-01 7.45272219e-01 -2.60018170e-01 2.16067135e-02
3.27711962e-02 3.46137658e-02 7.25465596e-01 1.52965438e+00
3.95347804e-01 -1.05988704e-01 1.13865227e-01 6.28521979e-01
-4.25082177e-01 2.02405706e-01 -8.62390339e-01 -3.50709766e-01
3.45775366e-01 8.50909173e-01 -1.81882501e-01 -8.33759964e-01
-2.87365407e-01 1.25115049e+00 4.54898298e-01 1.56716511e-01
-4.38960969e-01 -7.41481245e-01 -7.66392201e-02 -5.51468730e-02
3.02971691e-01 -2.05687322e-02 -7.25122154e-01 -1.44086099e+00
4.71485674e-01 -9.54691887e-01 1.59809113e-01 -7.56700277e-01
-1.12350678e+00 7.30060160e-01 -3.39533202e-02 -1.12599552e+00
-7.54974961e-01 1.08835995e-01 -1.01585186e+00 6.54453635e-01
-1.25520039e+00 -8.82689059e-01 8.42886269e-02 -1.07171655e-01
1.04972959e+00 -8.97501186e-02 9.27392066e-01 -1.90914869e-01
-3.60047251e-01 7.11143732e-01 1.31529629e-01 -1.40407577e-01
9.09782529e-01 -1.41989732e+00 7.63891101e-01 1.04343712e+00
6.76491037e-02 5.50919116e-01 1.10431218e+00 -8.06773067e-01
-1.05880034e+00 -1.13352203e+00 1.46708322e+00 -4.67600346e-01
4.10593688e-01 -3.15804541e-01 -9.29962337e-01 9.10305202e-01
7.37839341e-01 -6.40997827e-01 6.15202963e-01 1.39578030e-01
-1.47704899e-01 2.20681652e-01 -9.11992610e-01 4.05558079e-01
1.21392214e+00 -4.07578945e-01 -1.37929881e+00 6.16848648e-01
1.00585902e+00 -2.17159599e-01 -6.80945158e-01 3.75707924e-01
2.78341830e-01 -8.86407316e-01 5.44428647e-01 -7.18024969e-01
1.19896662e+00 1.86215490e-01 -8.29022378e-02 -1.63822460e+00
-8.94724950e-02 -5.77169120e-01 -1.19357221e-02 1.36091983e+00
5.78748465e-01 -1.12449862e-01 7.03175366e-01 5.21910965e-01
-6.62527978e-01 -5.59873760e-01 -6.69045150e-01 -6.84609473e-01
3.52332205e-01 -2.22298846e-01 3.30083728e-01 4.45059568e-01
4.67065871e-01 1.04503572e+00 -2.55680054e-01 -3.60941738e-01
4.82571840e-01 1.87930211e-01 7.99879313e-01 -9.87400532e-01
-3.81050855e-01 -4.46439832e-01 -6.83671981e-02 -1.32097459e+00
5.32956481e-01 -1.14628780e+00 3.75390440e-01 -1.91027308e+00
5.06642103e-01 3.70388091e-01 3.26402158e-01 3.26597303e-01
-3.62009197e-01 9.15131532e-03 1.65141627e-01 1.84107468e-01
-9.80186582e-01 9.88706946e-01 9.29342151e-01 -3.86135042e-01
-2.77066439e-01 -1.30614400e-01 -1.09962773e+00 5.81932306e-01
6.65781558e-01 -4.80066776e-01 -5.49590766e-01 -7.16391444e-01
1.23143442e-01 2.23708779e-01 -2.84730475e-02 -8.85715187e-01
1.59594446e-01 2.73618639e-01 1.84340641e-01 -7.12374985e-01
-4.60144058e-02 -1.21240951e-01 -3.10099185e-01 3.72826070e-01
-8.74125898e-01 8.17651823e-02 1.52818179e-02 5.67405462e-01
-4.04149175e-01 -6.66564345e-01 5.03615856e-01 -1.59149632e-01
-4.44546677e-02 -2.55562454e-01 -5.26999712e-01 3.67341399e-01
4.47851002e-01 -2.62995571e-01 -2.90546298e-01 -8.62689376e-01
-6.83870316e-01 3.68821800e-01 7.01008558e-01 2.88183093e-01
6.41058266e-01 -8.97402883e-01 -1.16905689e+00 -1.07472213e-02
9.78498999e-03 3.68197888e-01 8.45240653e-02 3.98638427e-01
-4.92108077e-01 4.26346272e-01 1.43939137e-01 -3.50631416e-01
-1.13384974e+00 3.92446339e-01 3.80177908e-02 -4.97905612e-01
-8.49760890e-01 4.73182887e-01 -5.57437949e-02 -5.18896699e-01
2.19029889e-01 -4.44191515e-01 -3.33003793e-03 -1.10089481e-01
7.16537118e-01 2.25818351e-01 3.35374594e-01 -2.66619384e-01
1.86880380e-01 1.54229805e-01 -3.30281198e-01 -8.82093981e-02
1.59136128e+00 -3.52024764e-01 -2.19720632e-01 1.48852170e-01
1.30251324e+00 7.79274106e-02 -9.36866641e-01 -5.54424189e-02
2.46536031e-01 -1.14461288e-01 -2.65756518e-01 -7.66697466e-01
-3.88030916e-01 6.13543332e-01 -1.74970850e-01 1.71132565e-01
1.17911077e+00 -7.70891830e-02 1.20548129e+00 9.63814199e-01
-5.45796901e-02 -1.16264260e+00 9.28902104e-02 8.65953803e-01
1.19099927e+00 -1.18803620e+00 1.16184995e-01 -3.54658514e-01
-6.28654718e-01 1.19468236e+00 2.33461931e-01 -3.80807936e-01
1.24485470e-01 1.93140581e-01 -2.06981152e-01 3.98930572e-02
-1.06652391e+00 1.47512883e-01 -1.34913279e-02 5.93642414e-01
7.14333355e-01 -1.23072006e-01 -6.78853035e-01 6.21162713e-01
-1.07375276e+00 -3.86897922e-02 9.24071729e-01 8.94964516e-01
-7.37793326e-01 -1.08655989e+00 5.82360402e-02 8.15922320e-01
-4.89818633e-01 -4.12746400e-01 -5.80817759e-01 2.76568681e-01
-6.37452364e-01 1.24659097e+00 2.97933578e-01 -2.44884163e-01
4.12343562e-01 2.58124739e-01 4.01034534e-01 -1.08113956e+00
-6.53216720e-01 9.11910608e-02 7.58194029e-01 -2.36957893e-01
-2.20600843e-01 -7.06023216e-01 -1.16363966e+00 -3.97439927e-01
-1.61043361e-01 2.54992425e-01 2.65003055e-01 1.01814258e+00
5.83056927e-01 4.17319804e-01 6.07635379e-01 -9.32134807e-01
-1.08209300e+00 -1.38961971e+00 -3.08555514e-01 7.79991746e-01
3.45744371e-01 4.05956119e-01 -3.16499889e-01 3.10022324e-01] | [12.351906776428223, 9.388152122497559] |
1c031807-f9b9-45c7-8d9c-32d4607f7577 | improving-distantly-supervised-relation-1 | 1804.06987 | null | http://arxiv.org/abs/1804.06987v1 | http://arxiv.org/pdf/1804.06987v1.pdf | Improving Distantly Supervised Relation Extraction using Word and Entity Based Attention | Relation extraction is the problem of classifying the relationship between
two entities in a given sentence. Distant Supervision (DS) is a popular
technique for developing relation extractors starting with limited supervision.
We note that most of the sentences in the distant supervision relation
extraction setting are very long and may benefit from word attention for better
sentence representation. Our contributions in this paper are threefold.
Firstly, we propose two novel word attention models for distantly- supervised
relation extraction: (1) a Bi-directional Gated Recurrent Unit (Bi-GRU) based
word attention model (BGWA), (2) an entity-centric attention model (EA), and
(3) a combination model which combines multiple complementary models using
weighted voting method for improved relation extraction. Secondly, we introduce
GDS, a new distant supervision dataset for relation extraction. GDS removes
test data noise present in all previous distant- supervision benchmark
datasets, making credible automatic evaluation possible. Thirdly, through
extensive experiments on multiple real-world datasets, we demonstrate the
effectiveness of the proposed methods. | ['Siddhesh Khandelwal', 'Sharmistha Jat', 'Partha Talukdar'] | 2018-04-19 | null | null | null | null | ['relationship-extraction-distant-supervised'] | ['natural-language-processing'] | [ 3.51822257e-01 7.74027169e-01 -4.64033157e-01 -4.44350839e-01
-8.93525958e-01 -1.85258210e-01 7.48891771e-01 4.12787527e-01
-3.27684432e-01 9.12867665e-01 7.40734160e-01 -6.06557310e-01
-3.76025438e-02 -8.20876956e-01 -4.62543219e-01 -4.23616678e-01
-3.62152606e-02 4.16675597e-01 2.45178536e-01 -5.89486718e-01
-6.89933002e-02 3.94582182e-01 -9.35288370e-01 1.27826944e-01
8.21746826e-01 7.77661324e-01 -4.73409779e-02 6.62990451e-01
8.32193121e-02 1.22724700e+00 -6.81107581e-01 -6.82092965e-01
4.78385612e-02 -4.26352620e-01 -1.67446756e+00 -2.68978644e-02
-1.01238914e-01 -2.53252327e-01 -3.22232515e-01 7.52511024e-01
5.10712326e-01 3.79963368e-01 5.82338750e-01 -9.98320282e-01
-9.13538456e-01 1.12335193e+00 -5.95190763e-01 4.03983206e-01
4.13130701e-01 -3.65327686e-01 1.56458473e+00 -9.59523499e-01
7.48190582e-01 1.13854635e+00 3.34180951e-01 2.86339074e-01
-1.04067612e+00 -4.14621830e-01 3.98841262e-01 3.28922600e-01
-1.42495406e+00 -4.60965514e-01 8.54574323e-01 -1.92832053e-01
1.89437842e+00 2.65707254e-01 2.73915559e-01 1.10714900e+00
5.15909865e-02 9.29145038e-01 5.66461146e-01 -7.68304288e-01
-2.83210188e-01 -4.14939746e-02 7.94730246e-01 3.15071583e-01
3.03803444e-01 -2.88385630e-01 -5.81387341e-01 -1.04096189e-01
3.12457085e-01 -2.79640496e-01 -5.87065279e-01 3.25130783e-02
-1.03915215e+00 8.91442001e-01 6.34021640e-01 5.98644853e-01
-3.86318326e-01 -2.49122858e-01 5.09198189e-01 3.23849499e-01
9.82188225e-01 5.81856430e-01 -7.89089978e-01 7.55720809e-02
-4.39349294e-01 -1.10693738e-01 7.76990175e-01 1.27000737e+00
5.06627202e-01 -3.06525707e-01 -5.70358515e-01 8.55989993e-01
3.26216429e-01 8.98093060e-02 5.11548221e-01 -1.34789526e-01
1.00490022e+00 7.75294006e-01 -9.36001390e-02 -9.31140482e-01
-4.91607368e-01 -5.33232212e-01 -9.84458148e-01 -4.48527664e-01
-1.68018728e-01 -3.04120958e-01 -6.23276114e-01 1.61055863e+00
2.82604367e-01 -8.73131969e-04 4.50306773e-01 6.43149316e-01
1.48715389e+00 6.24102116e-01 1.49142489e-01 -4.26612854e-01
1.43579316e+00 -1.53132153e+00 -1.23580205e+00 -3.68081301e-01
9.21688020e-01 -4.62723821e-01 6.08046055e-01 -7.45306611e-02
-9.52109933e-01 -5.41236639e-01 -1.32776403e+00 -5.61868787e-01
-6.36988103e-01 9.44909975e-02 8.06550860e-01 1.78752631e-01
-8.19142818e-01 5.63846827e-01 -7.83009946e-01 -3.76113087e-01
4.15711254e-01 5.37796855e-01 -4.84336853e-01 3.26039493e-01
-1.66617727e+00 1.12701118e+00 6.71977758e-01 3.22973341e-01
-1.53938875e-01 -2.85575569e-01 -1.44764459e+00 2.32873604e-01
6.49977028e-01 -6.87531233e-01 1.08771753e+00 -3.72279793e-01
-1.24768639e+00 8.82813156e-01 -5.50567448e-01 -6.01100862e-01
-3.06704104e-01 -6.48018181e-01 -4.17167366e-01 -1.31401509e-01
2.22618446e-01 2.92629153e-01 4.27766711e-01 -1.10288322e+00
-4.57889825e-01 -2.75824398e-01 1.35801863e-02 3.15964639e-01
-2.98934340e-01 3.10535908e-01 -4.26381141e-01 -7.19306409e-01
-1.01365581e-01 -5.61416030e-01 -3.23002696e-01 -9.31829154e-01
-1.07670617e+00 -8.29407036e-01 6.46258175e-01 -8.88941765e-01
1.59289765e+00 -1.87929833e+00 4.57885593e-01 -7.85194039e-02
3.94750863e-01 6.79253578e-01 -1.50887370e-01 6.28521562e-01
-6.03829801e-01 2.93179780e-01 -8.70313793e-02 -6.08876109e-01
-2.94614702e-01 1.75559819e-01 -3.50604981e-01 -1.03202742e-02
8.83027434e-01 1.42605793e+00 -1.12901509e+00 -6.06938362e-01
-1.44484922e-01 4.61440742e-01 1.04434930e-01 4.99455690e-01
1.14047244e-01 1.33525699e-01 -4.88608152e-01 5.55599630e-01
3.16428989e-01 -5.43238401e-01 3.23664397e-01 -2.21522227e-01
3.63419235e-01 8.37935388e-01 -8.60318422e-01 1.35628736e+00
-4.08377260e-01 4.65476036e-01 -4.18906868e-01 -8.70048881e-01
1.13955700e+00 6.18267953e-01 7.12874755e-02 -3.61735761e-01
4.63742763e-01 -6.60228282e-02 2.07802802e-02 -3.69161129e-01
7.91938186e-01 1.19964890e-02 -9.57433414e-03 4.05902654e-01
5.62285781e-01 1.76101953e-01 4.18423593e-01 4.93972778e-01
1.25955462e+00 9.09517035e-02 1.05696952e+00 -6.50751218e-02
6.97976172e-01 -4.64483380e-01 7.12099075e-01 4.17774826e-01
-1.03966430e-01 6.52616799e-01 7.39343822e-01 -4.60465342e-01
-3.68545085e-01 -5.31720400e-01 -9.05905217e-02 9.31060016e-01
4.19884324e-02 -8.86959136e-01 -4.42588359e-01 -1.15223205e+00
-2.90622920e-01 5.97635090e-01 -7.78347313e-01 -2.38353670e-01
-5.29013038e-01 -8.58338594e-01 3.12040478e-01 9.76967454e-01
3.66532028e-01 -1.44669163e+00 -4.62406538e-02 2.25189656e-01
-4.74624991e-01 -1.35420179e+00 -4.67491597e-01 8.37689519e-01
-6.48423016e-01 -1.03439534e+00 -3.76159936e-01 -9.98860478e-01
6.40216351e-01 4.20794755e-01 1.42667174e+00 2.15878710e-01
2.95461804e-01 -1.49696261e-01 -7.81298816e-01 -3.96860600e-01
4.73596941e-04 6.96555197e-01 -1.75336167e-01 -1.00757152e-01
8.72277021e-01 -3.29534262e-01 -1.03156306e-02 -1.50773600e-01
-5.44067144e-01 1.15877002e-01 6.76056802e-01 1.21961045e+00
4.80349779e-01 -1.02077737e-01 7.63940871e-01 -1.47922921e+00
8.50065529e-01 -5.89624166e-01 -1.27394516e-02 4.41925675e-01
-6.25972211e-01 1.51553914e-01 3.11986506e-01 -1.35785863e-01
-1.09221494e+00 -1.66254431e-01 -2.76902258e-01 1.02402493e-01
-6.49807826e-02 7.62285233e-01 -5.24538517e-01 4.39581305e-01
5.66380143e-01 -2.04059362e-01 -4.26022887e-01 -3.49557757e-01
4.22805190e-01 8.52197707e-01 4.39109594e-01 -3.05569470e-01
7.55664051e-01 -4.47400063e-02 -2.21939981e-01 -7.74402082e-01
-1.53015590e+00 -6.72949374e-01 -1.12752187e+00 4.19676572e-01
1.08425510e+00 -9.21681404e-01 -4.35047418e-01 2.89326996e-01
-1.48929572e+00 -3.48756641e-01 -4.17370200e-01 3.75985473e-01
3.84886600e-02 2.33227625e-01 -1.08335268e+00 -8.56109500e-01
-8.81007433e-01 -8.70285332e-01 1.23769796e+00 3.14614475e-01
-6.29312277e-01 -1.25747144e+00 1.30820453e-01 7.20520556e-01
5.59911272e-03 2.84286708e-01 7.48898268e-01 -1.15303278e+00
-1.78744838e-01 -2.58691669e-01 -3.94205302e-01 2.32985064e-01
4.87499684e-01 -1.32563889e-01 -1.03147745e+00 -1.28219217e-01
-3.10922921e-01 -5.18461168e-01 1.02146804e+00 3.44576798e-02
6.81147575e-01 -1.56277537e-01 -5.42732239e-01 3.13764215e-01
9.85548735e-01 1.73734739e-01 6.40705287e-01 3.92624438e-01
1.12728679e+00 5.76975942e-01 8.08648884e-01 -7.20250160e-02
5.13207018e-01 7.43273616e-01 7.90894255e-02 -5.26612103e-01
-1.56623498e-01 -2.15202153e-01 1.64535329e-01 1.29338086e+00
-3.27693611e-01 -5.33362448e-01 -8.22669268e-01 8.74206424e-01
-2.08355093e+00 -6.80418849e-01 -5.03967822e-01 1.74942076e+00
1.16975319e+00 4.36790407e-01 -1.09091848e-02 3.40703994e-01
5.44842184e-01 2.76908547e-01 -1.39753714e-01 -6.32201254e-01
-1.87784195e-01 5.60895860e-01 2.66973138e-01 7.04749286e-01
-1.41837084e+00 1.24542153e+00 5.30761242e+00 7.82230794e-01
-4.21175212e-01 4.99804355e-02 6.74055398e-01 2.98978597e-01
-2.47327566e-01 7.96062499e-02 -1.18634999e+00 -6.21575229e-02
9.43976998e-01 -5.67675866e-02 -2.56780863e-01 6.37445509e-01
-2.45391294e-01 1.58039108e-01 -1.17443419e+00 5.30070901e-01
2.43233159e-01 -1.02242172e+00 3.95128578e-02 -2.07705161e-04
6.26919866e-01 -3.55672874e-02 -5.32691717e-01 5.44169843e-01
5.17642200e-01 -1.04585028e+00 1.31214395e-01 2.60473818e-01
6.32563472e-01 -9.97893214e-01 1.63956392e+00 4.98256609e-02
-1.38265264e+00 4.25699711e-01 -2.00431421e-01 -2.08555236e-01
4.18075830e-01 7.10165977e-01 -8.86361420e-01 1.28918910e+00
4.86180305e-01 1.14448977e+00 -8.43605578e-01 2.81197637e-01
-8.66669655e-01 7.71741450e-01 6.17095716e-02 -9.36378837e-02
1.91954717e-01 -1.87649027e-01 3.93607736e-01 1.37657154e+00
-3.49303216e-01 2.81565785e-01 -1.46209925e-01 5.56068718e-01
-2.19778001e-01 1.87581271e-01 -7.55660713e-01 -5.60109988e-02
2.65316010e-01 1.39315438e+00 -7.06839442e-01 -2.97620416e-01
-7.16837883e-01 9.27003860e-01 8.51512492e-01 3.83799285e-01
-4.35620964e-01 -8.45143318e-01 4.44018811e-01 -5.01953423e-01
5.55373847e-01 3.40779312e-02 -2.03006312e-01 -1.16959238e+00
1.92030847e-01 -6.93098009e-01 5.67873180e-01 -6.52472079e-01
-1.34320533e+00 1.17096162e+00 -7.54104182e-02 -9.65017557e-01
-2.32571483e-01 -2.90223956e-01 -5.22580445e-01 1.08574080e+00
-1.68610477e+00 -1.39243543e+00 1.22589161e-04 2.23656178e-01
5.16192615e-01 -9.75397527e-02 1.07201028e+00 2.19457224e-01
-1.01289308e+00 7.50520110e-01 -5.88078439e-01 7.72521973e-01
5.97948253e-01 -1.59329069e+00 9.32948530e-01 8.99739027e-01
5.80218315e-01 8.99141192e-01 3.24758321e-01 -7.84606040e-01
-8.27555180e-01 -1.22097504e+00 1.82080543e+00 -5.50773323e-01
8.47865224e-01 -4.90428299e-01 -1.16961181e+00 1.14125168e+00
5.88702023e-01 4.98540327e-02 9.98346806e-01 7.70272613e-01
-1.96313620e-01 6.26825988e-02 -7.30022669e-01 5.39883256e-01
1.12502038e+00 -5.25431991e-01 -9.30980980e-01 4.19627964e-01
1.36453724e+00 -4.04856294e-01 -1.13955224e+00 5.99599540e-01
-1.66611187e-02 -4.08391804e-01 8.04935992e-01 -9.07650054e-01
7.06264257e-01 -1.74993500e-01 1.55472055e-01 -1.28614831e+00
-4.89992917e-01 -7.09365427e-01 -7.88015962e-01 1.92030203e+00
8.46927583e-01 -5.03384352e-01 4.33848441e-01 3.82162690e-01
-2.58910172e-02 -1.22172678e+00 -5.58252454e-01 -6.97154582e-01
-6.18561953e-02 -1.84620813e-01 5.78328848e-01 1.12959659e+00
4.01278794e-01 1.54871380e+00 -4.85793471e-01 7.95668811e-02
4.87129726e-02 2.50382572e-01 5.99019587e-01 -1.08879566e+00
-3.65341485e-01 -1.54163435e-01 -5.09330511e-01 -1.29945016e+00
3.92627090e-01 -1.04690433e+00 6.57976046e-02 -1.71383882e+00
4.50857699e-01 1.20110735e-01 -3.21809620e-01 7.82072783e-01
-8.13808322e-01 -7.31509626e-02 -1.53786868e-01 -6.98080361e-02
-8.43762636e-01 9.08887744e-01 1.08841896e+00 -1.01305544e-01
-2.20433116e-01 7.08097294e-02 -1.10692024e+00 4.74342287e-01
6.76960766e-01 -4.43787068e-01 -4.56645936e-01 -2.66480267e-01
2.44124755e-01 -1.12766087e-01 -1.83190197e-01 -2.62672275e-01
1.20359488e-01 1.64291307e-01 7.21792057e-02 -7.17032492e-01
1.11331306e-02 -5.56976140e-01 -5.98487496e-01 7.72698373e-02
-3.19359541e-01 -2.44922206e-01 -2.56204512e-02 4.95549798e-01
-5.15600741e-01 -1.00476280e-01 3.00748944e-01 2.69800603e-01
-4.49952543e-01 1.31659135e-01 -1.35705784e-01 2.77383506e-01
8.05880606e-01 2.89594233e-01 -4.67780560e-01 -4.07265157e-01
-6.40285671e-01 5.66334605e-01 -2.08115369e-01 6.56352878e-01
5.52324772e-01 -1.55835545e+00 -8.94547760e-01 -2.62686587e-03
3.85535538e-01 5.08531749e-01 -3.38256806e-01 6.46882296e-01
-9.14823785e-02 6.54238164e-01 5.43828011e-01 2.93258368e-03
-1.56275141e+00 6.82052970e-01 -2.98961047e-02 -9.90873814e-01
-6.77818239e-01 1.21852565e+00 -5.04568778e-02 -4.85136062e-01
1.25913858e-01 -5.47196865e-01 -7.62098491e-01 1.04689807e-01
5.92419326e-01 -3.62191834e-02 4.24696654e-01 -8.52697492e-01
-4.85905975e-01 2.69631058e-01 -5.49871802e-01 1.47160858e-01
1.54792428e+00 -6.62248507e-02 -2.33857945e-01 6.15451932e-01
1.10803342e+00 -1.00890502e-01 -7.38761842e-01 -6.68395758e-01
5.14474213e-01 -2.18581874e-02 7.14934571e-03 -5.13500094e-01
-1.05558002e+00 5.37695706e-01 -2.48583600e-01 5.17967165e-01
1.08294332e+00 3.64339530e-01 1.06655800e+00 3.97762567e-01
1.11736171e-01 -8.18313956e-01 -2.64459997e-01 8.13737869e-01
1.05611789e+00 -1.47118747e+00 2.69907057e-01 -8.44382584e-01
-7.36771047e-01 8.55649292e-01 6.53276503e-01 1.64569035e-01
7.03588128e-01 4.10851628e-01 -1.55059551e-03 -4.70756322e-01
-9.41402853e-01 -6.56930745e-01 6.40563905e-01 6.01575375e-01
9.01854873e-01 -1.46973372e-01 -4.67694968e-01 1.15052080e+00
-2.34515995e-01 -4.43946570e-01 3.06058615e-01 1.10525835e+00
6.09153509e-02 -1.33729911e+00 2.36924618e-01 3.43782187e-01
-4.12853688e-01 -4.90346789e-01 -9.26656544e-01 6.12457395e-01
-1.03990421e-01 1.39230013e+00 -1.55179128e-01 -5.18557072e-01
4.13282931e-01 1.59086749e-01 1.02649868e-01 -1.12879610e+00
-7.73675382e-01 -1.28993541e-01 5.62796175e-01 -3.66626799e-01
-5.20646036e-01 -5.41354954e-01 -1.14678466e+00 8.36068094e-02
-9.23852384e-01 4.15926248e-01 -1.57632064e-02 1.34427702e+00
4.08920556e-01 1.03964293e+00 5.39177656e-01 -5.46868145e-01
-9.76107791e-02 -1.47490990e+00 -5.31517267e-01 4.53112274e-01
3.47610474e-01 -6.60522223e-01 -2.34058455e-01 7.72562101e-02] | [9.285841941833496, 8.637656211853027] |
27a0c9a7-6f16-4e27-b14b-64e76ec14334 | crossway-diffusion-improving-diffusion-based | 2307.01849 | null | https://arxiv.org/abs/2307.01849v1 | https://arxiv.org/pdf/2307.01849v1.pdf | Crossway Diffusion: Improving Diffusion-based Visuomotor Policy via Self-supervised Learning | Sequence modeling approaches have shown promising results in robot imitation learning. Recently, diffusion models have been adopted for behavioral cloning, benefiting from their exceptional capabilities in modeling complex data distribution. In this work, we propose Crossway Diffusion, a method to enhance diffusion-based visuomotor policy learning by using an extra self-supervised learning (SSL) objective. The standard diffusion-based policy generates action sequences from random noise conditioned on visual observations and other low-dimensional states. We further extend this by introducing a new decoder that reconstructs raw image pixels (and other state information) from the intermediate representations of the reverse diffusion process, and train the model jointly using the SSL loss. Our experiments demonstrate the effectiveness of Crossway Diffusion in various simulated and real-world robot tasks, confirming its advantages over the standard diffusion-based policy. We demonstrate that such self-supervised reconstruction enables better representation for policy learning, especially when the demonstrations have different proficiencies. | ['Michael S. Ryoo', 'Jinghuan Shang', 'Varun Belagali', 'Xiang Li'] | 2023-07-04 | null | null | null | null | ['self-supervised-learning', 'imitation-learning'] | ['computer-vision', 'methodology'] | [ 7.69108161e-02 9.99958441e-02 -4.63214576e-01 7.95698259e-03
-2.65049934e-01 -2.29067385e-01 1.01995826e+00 -4.01253253e-01
-9.14558351e-01 1.08725834e+00 3.13941479e-01 -1.52258307e-01
-4.56696637e-02 -3.82453322e-01 -7.80875385e-01 -1.05090809e+00
-2.38861859e-01 4.76907164e-01 2.54354179e-01 -8.39200988e-02
3.56765598e-01 4.45086896e-01 -1.42998636e+00 -5.41804321e-02
9.88482296e-01 5.69051802e-01 1.03092372e+00 7.73454249e-01
-9.23069790e-02 1.44367313e+00 -5.73852003e-01 2.40590215e-01
3.65980357e-01 -7.58511722e-01 -4.18669462e-01 8.99820179e-02
-2.58088320e-01 -6.01707757e-01 -9.42863762e-01 1.24194276e+00
5.70548594e-01 2.75340915e-01 1.07870495e+00 -1.20354199e+00
-7.93037951e-01 5.56056082e-01 -4.84920919e-01 1.68574393e-01
2.94822603e-01 8.00736666e-01 4.16051030e-01 -2.17324838e-01
8.91256511e-01 1.63179481e+00 1.04012258e-01 9.63380814e-01
-1.15478420e+00 -6.95663631e-01 2.46419370e-01 5.22050977e-01
-7.71634459e-01 -6.18278310e-02 7.03017175e-01 -4.73680973e-01
1.02312326e+00 -5.06778359e-01 8.77527714e-01 1.93937981e+00
5.64227402e-01 1.38660693e+00 1.66630471e+00 -3.68851095e-01
5.93960226e-01 8.40571709e-03 -4.70583022e-01 8.97168577e-01
-9.29679349e-02 7.13267803e-01 -3.50127846e-01 9.26743671e-02
1.24799871e+00 2.48310226e-03 -2.88414836e-01 -9.64188039e-01
-1.31439245e+00 6.31962895e-01 2.52569437e-01 2.01260656e-01
-7.12146044e-01 4.42504287e-01 2.30424553e-01 7.40322828e-01
7.76330417e-05 3.09117347e-01 -2.70059854e-01 -6.01510346e-01
-4.04605955e-01 8.45355093e-02 7.89166689e-01 1.05838799e+00
5.53021491e-01 3.92766684e-01 -2.90137112e-01 7.91943192e-01
4.10674900e-01 7.31707931e-01 1.10450399e+00 -1.39741206e+00
2.59752542e-01 1.07126504e-01 2.31461793e-01 -3.12161177e-01
-2.19731301e-01 -9.76484716e-02 -8.07994246e-01 9.95819569e-01
3.13311458e-01 -3.41342777e-01 -1.15765297e+00 1.93594182e+00
9.73015837e-03 3.18730295e-01 4.36190993e-01 7.57852197e-01
-8.19969177e-02 7.60371625e-01 2.47408494e-01 -3.41806531e-01
4.34584945e-01 -1.19476795e+00 -8.10346603e-01 -1.57254320e-02
6.42054796e-01 -2.53038824e-01 1.16839516e+00 6.14772975e-01
-9.36287701e-01 -5.80168903e-01 -9.95759904e-01 3.28408808e-01
-1.26282930e-01 9.46777984e-02 3.48846763e-01 2.33474717e-01
-1.06207061e+00 8.77343535e-01 -1.04843831e+00 -4.34312284e-01
5.66401720e-01 2.88493782e-01 -2.13150337e-01 1.05789835e-02
-9.37052906e-01 1.28592002e+00 5.29964745e-01 -3.32700580e-01
-1.90245986e+00 -1.52803391e-01 -6.74510419e-01 -1.20032512e-01
5.49372315e-01 -5.99810302e-01 1.50041962e+00 -9.75375473e-01
-2.31021142e+00 3.15496445e-01 1.50557026e-01 -7.57287562e-01
7.63022542e-01 -1.77969813e-01 -4.03493829e-02 3.36942561e-02
-3.90128978e-02 8.37095261e-01 1.21674478e+00 -1.21076858e+00
-5.68726599e-01 -1.91316396e-01 -1.13698818e-01 4.98539865e-01
-3.35665971e-01 -5.49078047e-01 -2.01765686e-01 -6.80565417e-01
-4.16006416e-01 -1.05655146e+00 -5.56878388e-01 1.46782026e-01
-4.97722961e-02 -2.97104239e-01 8.75817716e-01 -3.90071332e-01
7.24649847e-01 -2.04780030e+00 7.29568005e-01 -2.50030681e-02
5.24325483e-02 4.03342277e-01 -4.38341320e-01 5.15617073e-01
3.79993230e-01 -4.49233621e-01 -3.95279795e-01 -4.16786402e-01
1.70286492e-01 7.47092605e-01 -1.93256631e-01 4.36607033e-01
-1.00135162e-01 9.18802977e-01 -1.25355101e+00 -3.62344980e-01
3.69999558e-01 2.58035362e-01 -3.14345837e-01 3.43090862e-01
-5.14136374e-01 9.45752501e-01 -5.49934447e-01 1.47970155e-01
1.86122954e-01 4.35404517e-02 1.75930202e-01 4.78321582e-01
-3.92670371e-02 -1.55406892e-01 -9.43146050e-01 2.08334398e+00
-5.04719496e-01 6.78393126e-01 1.14115067e-01 -1.10823846e+00
8.81436944e-01 1.48525104e-01 5.01911223e-01 -9.35827672e-01
1.20767720e-01 1.59095079e-01 2.77820647e-01 -8.20015013e-01
2.61940330e-01 -1.09685296e-02 1.53294578e-01 5.49872637e-01
4.44352627e-01 -4.00455058e-01 1.64798528e-01 3.05305690e-01
1.16605008e+00 8.11131775e-01 3.09977084e-01 -2.22475920e-02
4.14492279e-01 -7.41601968e-03 3.07606667e-01 1.06017733e+00
-5.72604537e-01 9.36708227e-02 5.67620933e-01 1.71743348e-01
-1.22485161e+00 -1.08962607e+00 5.16165614e-01 6.24474823e-01
2.06863329e-01 5.56971245e-02 -7.36562252e-01 -9.57759261e-01
1.58315480e-01 8.65179718e-01 -6.87214255e-01 -4.61100608e-01
-5.47608912e-01 -3.10436845e-01 4.48242426e-01 4.91074234e-01
5.71626008e-01 -1.59462249e+00 -6.66879475e-01 4.13573980e-01
3.21710944e-01 -7.92610943e-01 -3.29645306e-01 2.76359767e-01
-9.02969301e-01 -8.49934578e-01 -1.27919614e+00 -8.70750129e-01
6.34434462e-01 8.12463984e-02 3.41728568e-01 -3.95663321e-01
-1.15715727e-01 8.40270817e-01 -2.69526303e-01 -2.45562896e-01
-9.71876085e-01 -2.04305634e-01 4.33664262e-01 -2.23885849e-01
3.88659872e-02 -5.21264136e-01 -4.94985908e-01 9.28509161e-02
-8.38199735e-01 -1.77295674e-02 9.77683604e-01 1.05726552e+00
3.42454731e-01 -1.52947173e-01 7.16399491e-01 -4.06291753e-01
1.13304543e+00 -4.93843913e-01 -7.85051465e-01 2.74865739e-02
-7.64013469e-01 5.61869860e-01 6.50150895e-01 -1.00111520e+00
-1.40154994e+00 1.23705640e-01 -7.94478208e-02 -7.64474034e-01
-2.68645287e-01 1.57610416e-01 3.06088924e-01 9.49339569e-03
5.31909585e-01 7.52151728e-01 6.49571002e-01 -4.00927454e-01
5.62152267e-01 7.25010216e-01 2.91508168e-01 -5.63299060e-01
4.53701824e-01 4.56615657e-01 -1.53073460e-01 -7.91666389e-01
-1.63165420e-01 -2.75278538e-01 -5.64235806e-01 -3.85888368e-01
8.32746446e-01 -7.25973964e-01 -9.03227627e-01 9.10642505e-01
-9.28480923e-01 -1.12037790e+00 -6.96661770e-01 1.04676986e+00
-1.40782046e+00 6.69407010e-01 -7.59811819e-01 -1.12026739e+00
3.71222526e-01 -1.46221828e+00 5.73607743e-01 2.32857302e-01
1.38010949e-01 -1.01117492e+00 3.59747618e-01 -3.02286804e-01
3.63759220e-01 -2.52690852e-01 7.56329656e-01 -3.69173765e-01
-6.63349152e-01 3.71036142e-01 1.17703304e-01 6.91957831e-01
1.02253377e-01 -4.81576234e-01 -5.69801509e-01 -5.04861653e-01
2.47532725e-01 -7.08962500e-01 9.65167522e-01 5.12274444e-01
9.76055086e-01 -4.05158959e-02 -3.53489935e-01 2.98033595e-01
1.18737245e+00 6.09087586e-01 5.47469199e-01 6.20671570e-01
4.74958152e-01 5.32761455e-01 6.75794899e-01 5.17580092e-01
1.60534590e-01 3.10609400e-01 5.64321458e-01 1.10365152e-01
-3.88262570e-01 -6.36444271e-01 9.38855886e-01 7.40071833e-01
-1.59217827e-02 -1.29864290e-01 -5.26150107e-01 4.45901096e-01
-2.05168247e+00 -1.05930638e+00 3.14949065e-01 2.07759595e+00
8.78342450e-01 1.30836412e-01 1.75134227e-01 -3.01162630e-01
3.99199188e-01 1.99570432e-01 -1.17147219e+00 -1.92156419e-01
-2.12047920e-01 -1.27654776e-01 6.71072066e-01 5.55127442e-01
-6.84702575e-01 1.25587976e+00 6.46725368e+00 9.44019198e-01
-1.05422306e+00 9.06312987e-02 1.85289338e-01 -1.21170003e-02
8.49274546e-03 -2.92915791e-01 -5.83036721e-01 5.35755038e-01
6.78931653e-01 -1.03376791e-01 7.16332793e-01 8.11283946e-01
4.77026165e-01 -4.16138083e-01 -8.53108764e-01 9.60860908e-01
-4.35572006e-02 -1.04545665e+00 -8.37234631e-02 3.06379139e-01
1.00684714e+00 3.76039296e-01 3.09578210e-01 6.25266850e-01
1.07390571e+00 -6.71987593e-01 6.41356945e-01 7.37401545e-01
4.52296138e-01 -3.29298615e-01 2.38585636e-01 9.98343587e-01
-5.19532621e-01 -6.05810881e-01 -4.99895871e-01 -6.59651607e-02
2.74603963e-01 -1.14324458e-01 -8.14736307e-01 1.39273763e-01
3.96868140e-01 9.98146832e-01 -5.75048365e-02 1.00616133e+00
-5.49218953e-01 5.76222181e-01 -1.17368162e-01 -4.58536506e-01
6.62842691e-01 -5.53204656e-01 7.37148941e-01 9.27913547e-01
4.04874265e-01 -2.32969254e-01 1.49437636e-01 8.39662254e-01
1.32434443e-01 -7.67819956e-02 -1.02890062e+00 -1.73618376e-01
1.32980689e-01 6.72632575e-01 -3.95334303e-01 -6.27681613e-01
-3.10765207e-01 1.42502558e+00 6.14097714e-01 7.00433731e-01
-7.84887314e-01 -9.53737646e-03 6.37694180e-01 -5.28239429e-01
5.76027036e-01 -7.35023141e-01 2.37703413e-01 -1.07896459e+00
-2.43805647e-01 -9.67110097e-01 -1.43325701e-01 -7.65543401e-01
-1.08077753e+00 2.73418933e-01 9.97156128e-02 -1.36796939e+00
-5.59728682e-01 -6.99131489e-01 -1.19467407e-01 6.44097567e-01
-1.73159015e+00 -6.23623073e-01 1.79305837e-01 7.63905883e-01
9.32674170e-01 -5.16360998e-01 4.47477221e-01 -1.31089061e-01
-3.78076106e-01 1.05904676e-01 7.39198804e-01 -2.53216535e-01
6.39883399e-01 -1.46655703e+00 1.14196949e-01 4.33478981e-01
2.34536044e-02 2.47086912e-01 5.88173568e-01 -8.35651755e-01
-1.38602865e+00 -6.99516416e-01 2.45843604e-02 -8.09254646e-02
8.89531732e-01 -1.08181216e-01 -6.60846055e-01 6.78342462e-01
4.81700718e-01 -5.09436667e-01 -4.12582681e-02 -6.62443936e-01
1.20000020e-01 1.63777411e-01 -1.07291853e+00 1.00470996e+00
1.30039120e+00 -3.78140658e-01 -7.09647954e-01 2.23981783e-01
5.51707745e-01 -2.08569229e-01 -5.98084867e-01 1.76045690e-02
4.45795327e-01 -7.35325813e-01 7.29456127e-01 -7.09085882e-01
2.74899215e-01 -6.63880631e-02 1.12749033e-01 -1.94087934e+00
-2.96114415e-01 -8.66939545e-01 -4.02356237e-01 6.18110716e-01
2.01150015e-01 -7.48841703e-01 7.71183491e-01 -6.92905933e-02
-5.94234951e-02 -5.42495549e-01 -8.45453918e-01 -1.28209007e+00
2.16995373e-01 -2.30558440e-01 1.08306743e-01 5.92087209e-01
1.85905263e-01 1.31472260e-01 -7.63196230e-01 -2.34487876e-01
7.50971317e-01 -4.12316918e-01 8.39873254e-01 -8.02880406e-01
-5.37345469e-01 -6.58903301e-01 -2.26131141e-01 -1.89120686e+00
5.24680674e-01 -6.64026499e-01 4.82928485e-01 -1.57213402e+00
-3.16635333e-02 -3.66501778e-01 -3.47162604e-01 1.56468600e-01
2.72594579e-03 -4.43281591e-01 3.72799248e-01 4.95262057e-01
-5.64019561e-01 1.24382424e+00 2.00293255e+00 -2.21340895e-01
-2.97212303e-01 9.09327809e-03 -6.63920268e-02 7.32664406e-01
9.09089029e-01 -5.40878117e-01 -8.67337108e-01 -3.88289541e-01
-4.44522470e-01 2.41930410e-01 -2.64462344e-02 -1.14086688e+00
3.64792854e-01 -3.81186068e-01 1.55374527e-01 -1.72453493e-01
5.37585735e-01 -8.35223198e-01 -3.82958323e-01 1.09850121e+00
-6.97883070e-01 -7.72236139e-02 -5.50631061e-02 1.21484613e+00
6.09875731e-02 -3.88847232e-01 7.46530414e-01 -3.76021773e-01
-1.11317956e+00 2.10111454e-01 -1.02386498e+00 -2.11740300e-01
1.44282115e+00 -1.78779304e-01 -1.12007208e-01 -6.59341693e-01
-1.01066542e+00 2.31571868e-01 5.88604331e-01 4.97678220e-01
6.95511580e-01 -1.09809887e+00 -4.48812366e-01 2.99451530e-01
-1.71184376e-01 -5.01682997e-01 2.57919692e-02 8.40360045e-01
-2.11546525e-01 3.05331856e-01 -6.67382658e-01 -5.62114954e-01
-8.67167890e-01 8.16571653e-01 2.87918776e-01 -3.02588284e-01
-8.20817411e-01 5.55496871e-01 1.97454110e-01 -4.14121836e-01
5.29586732e-01 -4.26658839e-01 -2.96695322e-01 -3.93188238e-01
2.18805343e-01 4.94836628e-01 -7.32648909e-01 -4.04457539e-01
3.10961753e-01 2.53065407e-01 -3.27616073e-02 -7.22903132e-01
1.10584986e+00 -2.52445728e-01 3.60539347e-01 6.25243664e-01
9.28123772e-01 -4.45434362e-01 -2.20438957e+00 -3.27891171e-01
-2.57463753e-02 -2.93961585e-01 -5.68863861e-02 -7.80111670e-01
-7.03194439e-01 9.34383273e-01 7.29745209e-01 3.11264861e-02
7.35094547e-01 -1.31479576e-01 6.02399290e-01 6.53330743e-01
8.10138106e-01 -1.37375021e+00 5.82550526e-01 6.68487728e-01
8.67275238e-01 -1.20888090e+00 -3.77211094e-01 2.74276227e-01
-1.04202890e+00 8.07461381e-01 6.83196306e-01 -4.02434111e-01
4.27542686e-01 3.15021783e-01 1.01052269e-01 1.45241737e-01
-9.09550667e-01 -4.70611960e-01 -3.70833397e-01 1.23409534e+00
-2.02916965e-01 -7.19076395e-02 -3.30081016e-01 -1.86025038e-01
2.98173219e-01 2.79864639e-01 6.33298635e-01 1.24339402e+00
-7.08276987e-01 -1.27582741e+00 -1.32973194e-01 3.04073960e-01
1.03086114e-01 2.77003497e-01 -1.82537183e-01 7.49705732e-01
-4.04920876e-01 6.49343789e-01 -2.03832909e-01 -2.55406111e-01
1.52149275e-01 -9.17183906e-02 9.17874992e-01 -4.13755953e-01
-1.40965417e-01 4.47435491e-02 -2.40722954e-01 -6.89278901e-01
-3.95029008e-01 -7.97796845e-01 -1.40196002e+00 1.70504868e-01
5.10337502e-02 -1.73400901e-03 8.29326391e-01 9.32261705e-01
3.05017143e-01 6.13168716e-01 6.29346609e-01 -9.30258811e-01
-1.30429959e+00 -1.18265545e+00 -7.17557788e-01 3.92360717e-01
5.49860835e-01 -1.05295146e+00 -3.05452824e-01 -4.57582138e-02] | [4.342376708984375, 1.2459006309509277] |
e873a185-f0c3-4e5d-bc86-0c856130e502 | learning-3d-human-shape-and-pose-from-dense | 1912.13344 | null | https://arxiv.org/abs/1912.13344v2 | https://arxiv.org/pdf/1912.13344v2.pdf | Learning 3D Human Shape and Pose from Dense Body Parts | Reconstructing 3D human shape and pose from monocular images is challenging despite the promising results achieved by the most recent learning-based methods. The commonly occurred misalignment comes from the facts that the mapping from images to the model space is highly non-linear and the rotation-based pose representation of body models is prone to result in the drift of joint positions. In this work, we investigate learning 3D human shape and pose from dense correspondences of body parts and propose a Decompose-and-aggregate Network (DaNet) to address these issues. DaNet adopts the dense correspondence maps, which densely build a bridge between 2D pixels and 3D vertices, as intermediate representations to facilitate the learning of 2D-to-3D mapping. The prediction modules of DaNet are decomposed into one global stream and multiple local streams to enable global and fine-grained perceptions for the shape and pose predictions, respectively. Messages from local streams are further aggregated to enhance the robust prediction of the rotation-based poses, where a position-aided rotation feature refinement strategy is proposed to exploit spatial relationships between body joints. Moreover, a Part-based Dropout (PartDrop) strategy is introduced to drop out dense information from intermediate representations during training, encouraging the network to focus on more complementary body parts as well as neighboring position features. The efficacy of the proposed method is validated on both indoor and real-world datasets including Human3.6M, UP3D, COCO, and 3DPW, showing that our method could significantly improve the reconstruction performance in comparison with previous state-of-the-art methods. Our code is publicly available at https://hongwenzhang.github.io/dense2mesh . | ['Zhenan Sun', 'Wanli Ouyang', 'Jie Cao', 'Hongwen Zhang', 'Guo Lu'] | 2019-12-31 | null | null | null | null | ['3d-human-pose-and-shape-estimation', '3d-human-reconstruction', 'human-mesh-recovery'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-1.67286009e-01 1.06432125e-01 -8.44117552e-02 -3.18305701e-01
-4.49837089e-01 -1.90699231e-02 3.87832612e-01 -2.26580739e-01
-1.83665082e-01 5.20336151e-01 3.55742097e-01 5.77520311e-01
-4.61542718e-02 -7.50915527e-01 -9.97726679e-01 -4.87343937e-01
-7.39270821e-02 7.78083205e-01 2.84004807e-01 -2.71571577e-01
-3.34139198e-01 4.87842351e-01 -1.55664241e+00 5.81742488e-02
7.36489773e-01 1.01957345e+00 1.15104854e-01 2.14083150e-01
1.27941832e-01 3.83328378e-01 -2.05976978e-01 -2.88185447e-01
4.60943311e-01 -3.08438182e-01 -4.47410107e-01 3.76424015e-01
4.08407211e-01 -3.43665093e-01 -5.03731906e-01 7.54794717e-01
7.75133967e-01 6.25235289e-02 3.39219719e-01 -1.10455132e+00
-1.95430771e-01 -2.48226430e-02 -7.05718040e-01 -3.74182642e-01
6.67122722e-01 1.60337642e-01 6.25837564e-01 -1.14771569e+00
8.61293197e-01 1.45806861e+00 8.60272527e-01 4.63477015e-01
-9.63487387e-01 -7.66282618e-01 3.47268343e-01 2.01165885e-01
-1.57710505e+00 -3.64888012e-01 1.01875901e+00 -3.97043705e-01
6.19906843e-01 1.27494633e-01 1.18550825e+00 1.20961821e+00
2.89412320e-01 8.19516122e-01 8.54625046e-01 -1.44885108e-01
-1.27195597e-01 -3.37315768e-01 -3.27227056e-01 1.04948473e+00
1.65332258e-01 2.72230744e-01 -7.93719590e-01 -6.75256997e-02
1.30357838e+00 3.14236313e-01 -3.56875390e-01 -8.58711362e-01
-1.19673061e+00 4.75265682e-01 8.69110346e-01 -8.65087137e-02
-5.62663794e-01 1.48381755e-01 8.32944270e-03 -9.46405903e-02
5.91014802e-01 -3.19067053e-02 -5.20946324e-01 4.01532277e-02
-7.23124087e-01 4.05344844e-01 5.15714407e-01 9.87290740e-01
7.38909304e-01 -8.46906751e-02 -1.44718178e-02 8.53056550e-01
6.50029778e-01 6.43517196e-01 2.34021053e-01 -9.37970042e-01
7.52795517e-01 7.92138696e-01 6.04476929e-02 -1.25724483e+00
-7.64400125e-01 -6.55263484e-01 -1.09302974e+00 7.74212554e-02
3.52839381e-01 -4.64981981e-03 -1.08252120e+00 1.52961171e+00
8.21503043e-01 2.05539152e-01 -5.28849602e-01 1.46317148e+00
7.55602539e-01 3.64270568e-01 -2.13698089e-01 1.91586971e-01
1.19385421e+00 -9.89385068e-01 -4.01823282e-01 -3.62134546e-01
2.33943880e-01 -5.71549594e-01 6.24941945e-01 2.16977626e-01
-1.19874620e+00 -9.60406005e-01 -8.56036127e-01 -1.45848542e-01
6.73372298e-02 1.69609532e-01 4.13987279e-01 2.20314667e-01
-6.28771961e-01 6.14048362e-01 -1.25204134e+00 -3.11973274e-01
3.93904328e-01 4.57499772e-01 -6.08834088e-01 -3.61193746e-01
-1.02683628e+00 8.03542316e-01 3.63460109e-02 5.68968296e-01
-6.26377463e-01 -6.64630592e-01 -1.08126748e+00 -3.42439383e-01
3.82382601e-01 -1.18064833e+00 8.05125833e-01 -5.97380102e-01
-1.56169415e+00 8.10614467e-01 -1.03568405e-01 -2.17912868e-01
8.28305125e-01 -6.01244211e-01 1.93032715e-02 1.37514561e-01
-1.57845784e-02 8.64740193e-01 8.83839190e-01 -1.26975489e+00
-4.01660860e-01 -7.57124960e-01 -1.19101770e-01 5.67675829e-01
5.27258068e-02 -4.54797655e-01 -1.01813293e+00 -8.12802970e-01
7.90890396e-01 -1.06139064e+00 -2.92208493e-01 4.52816904e-01
-4.72797543e-01 8.52719322e-02 3.35410804e-01 -9.12378490e-01
8.29240561e-01 -1.83497560e+00 7.40919530e-01 3.45490515e-01
1.98523328e-01 -1.23978324e-01 8.41244217e-03 2.34732389e-01
1.46118268e-01 -4.19449866e-01 -4.64831404e-02 -7.98775196e-01
-7.99073651e-02 4.05438811e-01 2.53249198e-01 8.30137908e-01
6.10242300e-02 9.63644981e-01 -7.19164491e-01 -5.22946835e-01
4.27038610e-01 9.64873195e-01 -6.49306595e-01 2.90250242e-01
-9.33392942e-02 9.92672980e-01 -5.64600706e-01 8.85686457e-01
8.04175913e-01 -3.35811526e-01 -1.99955050e-03 -5.05782902e-01
1.02188632e-01 1.83801323e-01 -1.36178982e+00 2.28613448e+00
-3.14036459e-01 -1.98582605e-01 1.51496023e-01 -7.21477985e-01
9.66198862e-01 3.30645114e-01 8.05623233e-01 -6.83711648e-01
2.30915397e-01 6.95919544e-02 -2.25752905e-01 -1.57647684e-01
9.54210609e-02 6.08032309e-02 1.81653835e-02 1.16233714e-02
6.69654310e-02 -1.15158632e-01 -3.06189299e-01 -2.86657304e-01
6.88522875e-01 7.89633453e-01 1.23966835e-01 8.58589336e-02
4.62959260e-01 -1.52915761e-01 8.90761971e-01 2.97539502e-01
4.81123850e-02 1.05477881e+00 -2.74891052e-02 -5.91577411e-01
-9.04113591e-01 -1.21575093e+00 -1.95992775e-02 6.66634083e-01
4.60472673e-01 -4.66113150e-01 -5.47236800e-01 -5.34370542e-01
2.38336220e-01 -5.13874218e-02 -5.36907196e-01 -1.89548552e-01
-7.94926822e-01 -5.53468764e-01 1.19555920e-01 6.97897851e-01
6.37253046e-01 -8.00869823e-01 -6.21057928e-01 2.00340122e-01
-3.94368440e-01 -1.02461100e+00 -4.30466503e-01 -5.68413036e-03
-1.08630407e+00 -1.07851315e+00 -1.01611459e+00 -5.28227329e-01
9.09272492e-01 -3.49433571e-02 8.64308596e-01 1.53068826e-01
-2.26228520e-01 2.61642843e-01 -3.66325855e-01 -3.16111818e-02
6.31626025e-02 1.33471325e-01 2.37299576e-01 5.14820404e-02
-3.64814550e-02 -7.70428598e-01 -8.90144646e-01 6.03068650e-01
-3.49966228e-01 4.78284359e-01 6.81737483e-01 8.05898786e-01
9.01265502e-01 -2.90435404e-01 7.33574405e-02 -4.70187724e-01
-1.80849031e-01 -3.84872794e-01 -3.72214168e-01 -7.49579668e-02
-2.74005204e-01 -1.12990230e-01 2.78738141e-01 -3.57607812e-01
-9.15867031e-01 6.01383865e-01 -3.36866647e-01 -7.43516445e-01
-1.70016199e-01 3.99482340e-01 -3.27990860e-01 -2.10073087e-02
4.09421086e-01 1.73489958e-01 2.15840638e-01 -8.70446682e-01
1.43442407e-01 1.18395768e-01 5.43121278e-01 -6.24466300e-01
1.07327092e+00 5.83794832e-01 8.72060508e-02 -6.64691091e-01
-7.63342619e-01 -4.48333830e-01 -9.89217520e-01 -4.37700480e-01
7.86080956e-01 -1.31678760e+00 -5.01572192e-01 7.02840805e-01
-1.17087722e+00 -2.21743628e-01 -2.04717278e-01 6.82029545e-01
-6.19007766e-01 3.54147702e-01 -7.14850545e-01 -5.58740437e-01
-3.50815117e-01 -1.02702212e+00 1.35641325e+00 9.18082446e-02
-2.78309852e-01 -6.56360805e-01 4.59606349e-02 4.92934167e-01
-8.31133202e-02 4.79612052e-01 4.66910481e-01 -5.48174530e-02
-6.78884804e-01 -2.77808160e-01 7.46170357e-02 1.54167295e-01
1.16296493e-01 -4.76973772e-01 -7.35623240e-01 -4.86785173e-01
-2.07219690e-01 -3.47356051e-01 5.24068296e-01 4.70449030e-01
8.85762095e-01 -1.44221848e-02 -4.35564041e-01 9.04209793e-01
9.87074852e-01 -4.26879317e-01 4.95913297e-01 3.18625122e-01
1.11563194e+00 6.61661506e-01 7.75837362e-01 7.66853631e-01
7.21716106e-01 1.02785826e+00 5.78567147e-01 -2.47119069e-01
-4.01621461e-01 -6.89366758e-01 2.33569488e-01 1.03664589e+00
-4.44095224e-01 2.55171657e-01 -7.94807851e-01 3.29676181e-01
-1.93436396e+00 -4.80866969e-01 -1.64574888e-02 2.35592675e+00
7.94714510e-01 1.55853376e-01 1.92224011e-01 7.84159005e-02
4.78982568e-01 1.39861554e-01 -5.47637820e-01 3.68209839e-01
1.15251355e-01 1.83949247e-01 2.30916262e-01 3.76876086e-01
-1.06829166e+00 7.43036866e-01 4.74077415e+00 5.02725124e-01
-9.53281641e-01 1.36424318e-01 3.70833248e-01 -2.63767779e-01
-8.86942074e-03 -2.48802930e-01 -8.52684081e-01 3.42864007e-01
3.00604254e-01 4.17937577e-01 7.22812265e-02 6.57312572e-01
1.87367722e-01 -9.12358053e-03 -1.04341257e+00 1.07037532e+00
5.15110381e-02 -9.82289135e-01 -6.97209910e-02 1.09929182e-01
7.24028111e-01 1.58205424e-02 -1.40947819e-01 8.52833465e-02
-5.07738143e-02 -8.75442326e-01 9.95588899e-01 8.78824770e-01
6.18257046e-01 -7.81272233e-01 6.53536499e-01 6.28621995e-01
-1.51487410e+00 1.02038875e-01 -4.23023105e-01 -1.42409787e-01
2.98119456e-01 6.46218300e-01 -4.55497444e-01 9.44024205e-01
9.79344368e-01 8.89398456e-01 -5.01986265e-01 1.02245438e+00
-3.13844442e-01 1.63993895e-01 -6.62604570e-01 4.21819419e-01
-2.03699857e-01 -1.66748479e-01 4.59758431e-01 6.08528018e-01
3.33627433e-01 2.91514955e-02 5.36184788e-01 8.72194588e-01
7.90509656e-02 -6.11797497e-02 -2.62853533e-01 5.98195851e-01
3.14524978e-01 1.12671864e+00 -5.03526032e-01 -1.15667135e-01
-3.52189839e-01 1.16145432e+00 2.52388984e-01 2.11327702e-01
-8.85117531e-01 2.49250606e-01 6.64927900e-01 5.69018662e-01
3.65998298e-01 -4.78966951e-01 -2.55914211e-01 -1.20360088e+00
3.61112088e-01 -8.12572300e-01 2.63393819e-01 -6.19604588e-01
-1.18339288e+00 4.71445918e-01 1.44339457e-01 -1.49946618e+00
-2.73860335e-01 -2.72018224e-01 -1.83710545e-01 8.46602976e-01
-1.28362548e+00 -1.44403088e+00 -5.68668008e-01 7.53675699e-01
4.30048585e-01 2.27669299e-01 5.66029370e-01 4.21923906e-01
-3.82352531e-01 5.84240139e-01 -3.68138760e-01 2.77250618e-01
6.58432543e-01 -8.21727395e-01 4.17752832e-01 5.77547610e-01
2.11051330e-01 4.11430657e-01 5.02252579e-01 -9.52483416e-01
-1.61896074e+00 -1.07631040e+00 7.30431199e-01 -4.67514127e-01
-1.76523216e-02 -4.74647552e-01 -6.29542768e-01 5.91663182e-01
-4.84773129e-01 4.82980818e-01 1.71360776e-01 3.22180614e-02
-1.42293066e-01 -1.63894922e-01 -9.43538487e-01 4.55311298e-01
1.55932212e+00 -2.21418440e-01 -5.43090403e-01 2.18280941e-01
3.79869938e-01 -9.88490283e-01 -9.80555892e-01 6.28196478e-01
8.46594512e-01 -9.39537823e-01 1.34919763e+00 -1.69648364e-01
3.55351090e-01 -4.21840250e-01 -1.06424399e-01 -1.19763160e+00
-3.24021637e-01 -3.34681928e-01 -4.79283243e-01 1.01052177e+00
-5.33776581e-02 -3.93253833e-01 1.19993234e+00 4.02030110e-01
-1.12089828e-01 -9.69259262e-01 -1.14991784e+00 -6.97748244e-01
-1.55994594e-01 -3.16904724e-01 5.10034919e-01 5.31024098e-01
-3.33059579e-01 9.59540680e-02 -6.48593843e-01 3.99947464e-01
7.85675645e-01 2.51899838e-01 1.09319723e+00 -1.24108565e+00
-4.50418025e-01 6.61113486e-02 -6.68582797e-01 -1.66877937e+00
-2.82110013e-02 -7.50113428e-01 1.45012319e-01 -1.44604945e+00
-7.17912465e-02 -4.73282844e-01 -7.15485066e-02 5.23750901e-01
-8.88331085e-02 5.22101581e-01 3.36941361e-01 2.28433728e-01
-3.90985757e-01 9.58147585e-01 1.66517556e+00 -1.68332420e-02
-1.27068922e-01 3.18377286e-01 -1.10907562e-01 8.71147811e-01
5.11565864e-01 -4.34330314e-01 -3.23865324e-01 -5.29274464e-01
7.12136850e-02 2.22310677e-01 7.82245517e-01 -1.35420585e+00
2.92025477e-01 9.70404968e-02 9.77645457e-01 -9.61148322e-01
8.45154762e-01 -9.56773520e-01 5.35098493e-01 5.55651009e-01
6.43598810e-02 -2.17748359e-02 -1.34754136e-01 6.04174197e-01
-1.51581898e-01 3.70664924e-01 6.12198532e-01 -3.05135131e-01
-4.91176009e-01 7.95871615e-01 2.55978227e-01 -2.74140798e-02
7.08820522e-01 -4.96323287e-01 2.18352735e-01 -4.23334628e-01
-9.50138688e-01 3.21856499e-01 6.50533378e-01 6.06813371e-01
8.50694418e-01 -1.55850768e+00 -6.75629020e-01 4.93067652e-01
6.63971677e-02 6.16378307e-01 4.43161279e-01 1.17085767e+00
-4.73752290e-01 1.81272298e-01 -3.14693660e-01 -1.00476146e+00
-1.09327471e+00 2.15005025e-01 4.22803938e-01 -1.79570556e-01
-1.07380712e+00 8.90087485e-01 2.40252718e-01 -7.43569970e-01
3.57821882e-01 -3.62548113e-01 9.66469944e-02 -2.70227760e-01
1.40244737e-01 4.85510409e-01 1.16801094e-02 -9.89521205e-01
-5.13537288e-01 1.04725158e+00 2.33890951e-01 1.05346769e-01
1.36673403e+00 -3.18986207e-01 9.56525952e-02 3.26747358e-01
1.17977858e+00 -1.39286965e-01 -1.64905846e+00 -3.47127646e-01
-4.17568594e-01 -6.45239353e-01 -1.95158809e-01 -6.50295734e-01
-1.28272772e+00 8.33668113e-01 6.42418504e-01 -6.39404237e-01
9.29993927e-01 8.00125003e-02 9.86533999e-01 2.17955094e-03
8.05770338e-01 -1.03256619e+00 4.34179269e-02 4.29415971e-01
1.19069958e+00 -1.05628967e+00 3.75725359e-01 -7.15790153e-01
-3.31751704e-01 9.56875503e-01 8.39039028e-01 -2.32053503e-01
8.08982790e-01 5.97830378e-02 1.05709024e-02 -2.45854989e-01
-3.56523693e-01 -2.20649719e-01 6.71788931e-01 7.03051627e-01
3.10779959e-01 -6.95848018e-02 -1.16969332e-01 6.82448924e-01
-2.17958525e-01 5.47325388e-02 -2.30599016e-01 9.54577327e-01
-1.50601313e-01 -1.12114108e+00 -6.65033340e-01 1.55073449e-01
-7.45507851e-02 3.14341724e-01 -1.80270970e-01 8.68568659e-01
4.51723129e-01 5.66875994e-01 5.74967600e-02 -5.81218362e-01
6.29340887e-01 -4.52704206e-02 7.04278409e-01 -6.06004536e-01
-4.43505824e-01 3.75235319e-01 5.48414188e-03 -1.09043396e+00
-5.38994133e-01 -6.83268905e-01 -1.39230359e+00 -1.70242369e-01
-1.25206113e-01 -2.71448553e-01 3.80094707e-01 9.23404217e-01
5.46404243e-01 4.66831982e-01 4.09839600e-01 -1.52652419e+00
-4.93227929e-01 -8.49554002e-01 -4.42559093e-01 4.99161661e-01
2.25109920e-01 -1.14140427e+00 1.66564621e-02 -2.05202773e-02] | [7.056962490081787, -1.0247230529785156] |
22ba9740-e191-44e1-8dac-3d9e76385279 | dsmnet-deep-high-precision-3d-surface | 2304.04200 | null | https://arxiv.org/abs/2304.04200v1 | https://arxiv.org/pdf/2304.04200v1.pdf | DSMNet: Deep High-precision 3D Surface Modeling from Sparse Point Cloud Frames | Existing point cloud modeling datasets primarily express the modeling precision by pose or trajectory precision rather than the point cloud modeling effect itself. Under this demand, we first independently construct a set of LiDAR system with an optical stage, and then we build a HPMB dataset based on the constructed LiDAR system, a High-Precision, Multi-Beam, real-world dataset. Second, we propose an modeling evaluation method based on HPMB for object-level modeling to overcome this limitation. In addition, the existing point cloud modeling methods tend to generate continuous skeletons of the global environment, hence lacking attention to the shape of complex objects. To tackle this challenge, we propose a novel learning-based joint framework, DSMNet, for high-precision 3D surface modeling from sparse point cloud frames. DSMNet comprises density-aware Point Cloud Registration (PCR) and geometry-aware Point Cloud Sampling (PCS) to effectively learn the implicit structure feature of sparse point clouds. Extensive experiments demonstrate that DSMNet outperforms the state-of-the-art methods in PCS and PCR on Multi-View Partial Point Cloud (MVP) database. Furthermore, the experiments on the open source KITTI and our proposed HPMB datasets show that DSMNet can be generalized as a post-processing of Simultaneous Localization And Mapping (SLAM), thereby improving modeling precision in environments with sparse point clouds. | ['Weiquan Liu', 'Cheng Wang', 'Yu Zang', 'Xiuhong Lin', 'Zhiyong Wang', 'Changjie Qiu'] | 2023-04-09 | null | null | null | null | ['simultaneous-localization-and-mapping', 'point-cloud-registration'] | ['computer-vision', 'computer-vision'] | [-2.75911212e-01 -4.32171226e-01 -7.29517564e-02 -3.86388302e-01
-7.28970945e-01 -1.65075794e-01 5.64417720e-01 -1.21321678e-01
-1.36912495e-01 5.11688232e-01 -3.67811888e-01 5.25350459e-02
-9.09616873e-02 -1.22396004e+00 -9.39754844e-01 -4.61663991e-01
9.40773115e-02 1.29028690e+00 7.44823396e-01 -6.98200017e-02
2.48449132e-01 1.01156855e+00 -1.72994018e+00 -1.40893355e-01
1.11425877e+00 9.56687629e-01 7.18537748e-01 1.81383878e-01
-4.87074733e-01 2.05890954e-01 -1.33549228e-01 5.25797233e-02
3.90633672e-01 5.81815302e-01 -2.78999329e-01 2.77908128e-02
6.85291052e-01 -3.08116317e-01 -1.15619943e-01 9.33683336e-01
4.92466122e-01 2.03978773e-02 6.40930593e-01 -1.36101019e+00
-1.11000896e-01 -9.38021764e-02 -8.50181997e-01 -4.18617457e-01
8.96736085e-02 7.78161362e-02 4.87018526e-01 -1.29555535e+00
5.14613748e-01 1.40470731e+00 1.00659811e+00 -2.77582817e-02
-1.20668530e+00 -1.04585600e+00 1.08444534e-01 9.22963768e-02
-2.00810838e+00 -2.22364411e-01 9.42499280e-01 -5.90712309e-01
7.40001380e-01 5.80238514e-02 8.53257596e-01 6.16636395e-01
1.36866286e-01 2.86537826e-01 8.61932874e-01 1.10280868e-02
2.55011410e-01 -1.29895046e-01 -2.05361962e-01 4.86408830e-01
4.24823970e-01 2.52609938e-01 -5.34380496e-01 -5.67965150e-01
1.16294503e+00 3.15301836e-01 -1.23909809e-01 -9.62511122e-01
-1.24415970e+00 6.93182290e-01 5.70262074e-01 -2.75432110e-01
-4.56487864e-01 3.36650789e-01 -3.24549526e-03 -3.20618957e-01
7.03719616e-01 -1.77729204e-01 -5.57192862e-01 -1.38034103e-02
-1.20717013e+00 4.68921304e-01 4.97769475e-01 1.48164213e+00
1.50244009e+00 5.50652342e-03 5.00093460e-01 7.27411985e-01
7.96147168e-01 1.16080821e+00 -6.99518397e-02 -1.18527639e+00
4.84963268e-01 7.78536618e-01 8.02532360e-02 -1.12480867e+00
-2.80567437e-01 -3.91807199e-01 -8.24503124e-01 1.38354495e-01
-3.66495699e-01 3.24887007e-01 -7.04203844e-01 1.21542442e+00
7.69568145e-01 8.65053415e-01 -1.86165974e-01 6.49159968e-01
6.47080660e-01 6.77470982e-01 -2.24976763e-01 -1.19826347e-01
1.06658077e+00 -6.55865669e-01 -3.45471829e-01 -1.62864774e-01
4.33741152e-01 -7.38603652e-01 8.01005542e-01 3.26901048e-01
-7.68061817e-01 -7.18169510e-01 -8.94239008e-01 3.68377054e-03
7.48457480e-03 1.20593440e-02 5.17592192e-01 2.04757154e-01
-8.71424675e-01 4.79608446e-01 -1.09906316e+00 -1.29802123e-01
4.31648672e-01 3.98341149e-01 -2.56561458e-01 -4.30492312e-01
-6.06147528e-01 7.61156440e-01 2.39803240e-01 2.59321779e-02
-7.68400431e-01 -1.33525360e+00 -7.88736224e-01 -2.51640230e-01
1.95471495e-01 -1.13183129e+00 8.39749873e-01 -1.68563068e-01
-1.24083412e+00 6.81013823e-01 -4.09342051e-01 -1.87926844e-01
3.74299169e-01 -2.84683704e-01 -2.03386307e-01 8.39693323e-02
4.25259084e-01 6.85373724e-01 6.85278714e-01 -1.87293625e+00
-6.00889564e-01 -6.79433048e-01 -3.49759847e-01 1.51075855e-01
3.27697784e-01 -3.76906842e-01 -5.47499478e-01 1.52942669e-02
7.13863254e-01 -9.86933887e-01 -3.59277755e-01 2.70354778e-01
-1.84904054e-01 -3.70485708e-02 1.32088101e+00 -2.36086085e-01
6.02309227e-01 -2.17526746e+00 -6.48107007e-02 3.03873330e-01
1.15608320e-01 -1.86611619e-02 -1.83347408e-02 4.49480891e-01
4.01245356e-01 -1.14338748e-01 -3.35370958e-01 -7.75375724e-01
-1.41579688e-01 6.35189712e-01 -4.51524436e-01 6.72719419e-01
-8.96189362e-02 5.16907513e-01 -8.10767591e-01 -8.54327202e-01
6.85399890e-01 7.18172967e-01 -5.60673475e-01 -1.78816300e-02
-3.01387310e-01 6.71291530e-01 -6.28767610e-01 8.85542452e-01
1.56470740e+00 -1.40153661e-01 -3.14808726e-01 -4.09256369e-01
-4.65920240e-01 -2.97645815e-02 -1.42328823e+00 2.23750234e+00
-5.31739414e-01 6.25131056e-02 3.01088095e-01 -4.00578767e-01
1.19118965e+00 1.04842251e-02 7.68312991e-01 -2.61514664e-01
-1.83196813e-01 4.46848750e-01 -4.47799712e-01 7.32957572e-03
6.22967839e-01 -2.17871785e-01 4.22261178e-01 -1.73860595e-01
-2.08656698e-01 -9.02957618e-01 -5.79515517e-01 1.70504451e-01
6.15410864e-01 4.96751040e-01 5.80662265e-02 -2.23737717e-01
5.55568039e-01 4.00172353e-01 8.15751612e-01 3.54523033e-01
2.13586435e-01 9.49222803e-01 -2.89866269e-01 -2.48032764e-01
-9.92853642e-01 -1.19164455e+00 -5.65554500e-01 1.71769693e-01
6.21639252e-01 -4.91371125e-01 -1.84863463e-01 -1.00828081e-01
5.48142374e-01 5.21502614e-01 7.82974884e-02 6.16836399e-02
-6.63482249e-01 -4.49020058e-01 1.13400154e-01 3.75196189e-01
6.31417871e-01 -4.64755595e-01 -4.50773627e-01 2.09891826e-01
-7.01208860e-02 -1.33065343e+00 -3.59957144e-02 -2.90348977e-01
-1.27589953e+00 -1.14914262e+00 -1.49386853e-01 -4.74608481e-01
5.11535883e-01 7.33439267e-01 1.15009308e+00 1.86052456e-01
1.32644236e-01 5.59180200e-01 -1.43674716e-01 -5.47934055e-01
3.38643305e-02 -2.88210902e-02 4.38918829e-01 -1.80589899e-01
3.11942101e-01 -1.14911985e+00 -5.01827836e-01 4.62693274e-01
-5.32001734e-01 2.94447571e-01 5.23002326e-01 3.17913860e-01
1.39682949e+00 1.06390798e-02 2.99878493e-02 -4.42439765e-01
-1.31079480e-01 -5.95000565e-01 -8.95561814e-01 -1.51598319e-01
-3.81837398e-01 -4.86339301e-01 1.55979082e-01 -2.55728751e-01
-7.46887147e-01 5.09468853e-01 -1.60219789e-01 -1.18706095e+00
-7.84482583e-02 3.82405937e-01 -3.36402506e-01 -5.98397315e-01
3.63024652e-01 5.58255553e-01 2.25449558e-02 -7.64390528e-01
2.73715407e-01 4.93808627e-01 5.46771824e-01 -9.74572062e-01
1.25939894e+00 1.08351314e+00 5.08143842e-01 -1.10618794e+00
-4.78569120e-01 -7.77577281e-01 -1.03424501e+00 -5.74224107e-02
6.18814588e-01 -1.43857503e+00 -5.44370770e-01 2.89557636e-01
-1.47248793e+00 -3.82727124e-02 -2.97454327e-01 6.16027057e-01
-7.56820202e-01 5.50267696e-01 -1.25806406e-01 -9.50955212e-01
-1.68954670e-01 -1.12036169e+00 1.79050279e+00 -2.09067538e-01
3.94419640e-01 -7.19857156e-01 3.92792970e-01 2.05199271e-01
6.42812923e-02 4.26124483e-01 5.94396591e-01 3.06695998e-02
-1.38985884e+00 -3.77262272e-02 -2.69044101e-01 9.90579650e-02
1.08121946e-01 2.74458796e-01 -8.52980435e-01 -3.20596814e-01
2.43700713e-01 1.24672636e-01 4.39763933e-01 4.93151218e-01
1.09226191e+00 2.53671467e-01 -8.10525835e-01 1.14188969e+00
1.75628936e+00 -1.63242117e-01 5.63249469e-01 2.63526469e-01
1.07272100e+00 4.19951200e-01 1.01730525e+00 4.75246102e-01
8.78148556e-01 1.03870082e+00 9.32006299e-01 3.14993821e-02
-2.47005466e-02 -5.30884326e-01 -8.00187979e-03 1.14119303e+00
-1.37172744e-01 4.73893344e-01 -1.19038641e+00 4.13532585e-01
-1.84930277e+00 -6.54352188e-01 -8.01695228e-01 2.26218128e+00
4.36577529e-01 -2.11088404e-01 -2.22229391e-01 -1.84140220e-01
5.54763973e-01 1.01298213e-01 -4.51236516e-01 4.90832418e-01
-6.69147670e-02 8.79254043e-02 7.14436233e-01 5.24293184e-01
-6.83571160e-01 1.11480641e+00 5.37427044e+00 8.62327397e-01
-1.19199538e+00 3.70350093e-01 -2.33614773e-01 1.06247447e-01
-4.11152720e-01 3.85509878e-01 -1.12785614e+00 6.10238016e-01
8.23893130e-01 -1.26294106e-01 7.11698458e-02 1.24158740e+00
3.62337261e-01 -5.36709540e-02 -8.35959077e-01 1.38053477e+00
-2.33548447e-01 -1.53453064e+00 2.14428693e-01 5.05851686e-01
5.93142509e-01 7.64967024e-01 -2.60873526e-01 1.73183143e-01
2.50562817e-01 -6.66667998e-01 9.28025901e-01 6.69788957e-01
8.84923816e-01 -6.23417974e-01 4.68607724e-01 9.60079432e-01
-1.74246705e+00 3.49966943e-01 -1.04008603e+00 -2.73940377e-02
4.93123204e-01 9.42673028e-01 -6.76415384e-01 1.04472625e+00
6.85960233e-01 9.26824987e-01 -4.61676866e-01 1.27686048e+00
5.07560447e-02 2.73299783e-01 -8.12479317e-01 6.07800961e-01
-2.36188457e-03 -5.53171098e-01 6.23897016e-01 7.80363739e-01
6.65393770e-01 1.66401476e-01 5.86914599e-01 1.12855351e+00
1.91709608e-01 -5.88873448e-03 -7.32656598e-01 6.66504920e-01
9.97927070e-01 1.24583220e+00 -3.63482088e-01 -2.97735274e-01
-3.41400832e-01 3.52503359e-01 1.64001092e-01 8.04603994e-02
-7.90446937e-01 2.75563836e-01 9.54207778e-01 6.13563716e-01
2.24975660e-01 -8.07945251e-01 -3.55262220e-01 -1.27397501e+00
-2.89192367e-02 -3.44850600e-01 -2.93193907e-01 -1.10888493e+00
-1.11812556e+00 3.67401004e-01 3.34012061e-01 -1.69610417e+00
1.51200980e-01 -2.98732579e-01 -4.85916764e-01 1.06274319e+00
-2.01365709e+00 -1.56915700e+00 -7.60869086e-01 6.71588182e-01
5.09337366e-01 7.51587600e-02 5.25433123e-01 4.37292576e-01
-6.85538650e-02 -4.17630792e-01 1.58215836e-01 -5.11090755e-01
4.49409962e-01 -9.02957320e-01 5.11956394e-01 4.96059984e-01
1.64211616e-01 6.43026233e-01 3.68767917e-01 -1.09572291e+00
-1.71936750e+00 -1.44834876e+00 3.71808290e-01 -6.31987989e-01
3.11110765e-01 -4.33414161e-01 -1.12157810e+00 6.68463171e-01
-5.59550047e-01 3.41759920e-01 2.14801699e-01 -2.50086896e-02
-1.07265987e-01 -2.88896561e-01 -1.24427676e+00 1.64875075e-01
1.23102367e+00 -3.97648007e-01 -5.39997160e-01 6.00124121e-01
1.00699508e+00 -8.06859732e-01 -9.40299094e-01 9.40621734e-01
3.54021817e-01 -9.10747051e-01 1.44356751e+00 1.09146602e-01
6.84961900e-02 -8.42873693e-01 -5.73755085e-01 -1.14510357e+00
-3.83382559e-01 -1.34371296e-01 -2.51018643e-01 1.24031150e+00
-3.06383967e-01 -5.97967982e-01 1.02516246e+00 2.04103321e-01
-5.94178140e-01 -5.63860118e-01 -1.23681748e+00 -9.69823718e-01
-3.58693331e-04 -7.00945497e-01 1.28178477e+00 9.43092644e-01
-8.72305274e-01 2.70371847e-02 -1.30703375e-01 8.59145641e-01
9.97632027e-01 2.31718987e-01 1.39489448e+00 -1.87588048e+00
1.39215916e-01 2.97443196e-02 -5.53843021e-01 -1.29244220e+00
2.43091017e-01 -6.48429215e-01 -6.21116124e-02 -1.65850985e+00
-1.27505019e-01 -1.15653193e+00 2.87806422e-01 -1.14806361e-01
1.87221259e-01 1.90985333e-02 1.80808350e-01 9.04861152e-01
-2.60819346e-01 9.43930149e-01 1.15004468e+00 -1.39433821e-03
5.93034504e-03 3.21467035e-02 2.69846916e-02 8.74428391e-01
3.37151408e-01 -5.95767140e-01 -4.31091905e-01 -8.26793313e-01
2.39954516e-02 8.63345638e-02 5.20784497e-01 -1.43581104e+00
4.05430734e-01 -5.01892507e-01 1.91816375e-01 -1.67420781e+00
1.02370059e+00 -1.29930186e+00 7.53497183e-01 3.30904394e-01
7.34756827e-01 6.06864393e-02 1.33413717e-01 7.15713680e-01
-1.58602700e-01 -7.20932931e-02 7.68732667e-01 -2.30102256e-01
-5.65994024e-01 1.00821292e+00 4.15426672e-01 -3.77390593e-01
9.04765546e-01 -4.69159931e-01 -1.31220803e-01 1.66576095e-02
-2.03094870e-01 4.79070306e-01 1.13135147e+00 8.71203244e-02
8.23464096e-01 -1.51950395e+00 -6.33931160e-01 3.17133367e-01
2.64669150e-01 1.17637563e+00 3.59569907e-01 7.24899590e-01
-9.22126234e-01 2.33808577e-01 1.58960566e-01 -1.47583926e+00
-9.99469459e-01 3.21796358e-01 2.32308209e-01 1.49905697e-01
-9.51295316e-01 5.50904930e-01 3.10937077e-01 -9.94165659e-01
-2.57716715e-01 -4.81871098e-01 1.63867682e-01 -3.56140465e-01
1.36742204e-01 4.03787315e-01 1.02568619e-01 -1.04430044e+00
-4.31332380e-01 1.23183608e+00 5.46604276e-01 7.59529993e-02
1.39003170e+00 -1.58650756e-01 -3.38509977e-01 5.96657693e-01
9.64663684e-01 1.55746952e-01 -1.29241586e+00 -3.37937742e-01
-1.19278178e-01 -9.98778403e-01 1.48123294e-01 -1.16990216e-01
-8.80724430e-01 9.71307456e-01 5.04132748e-01 -3.18685353e-01
5.44535875e-01 -4.61607091e-02 8.24516594e-01 2.57149428e-01
1.28045475e+00 -5.62645853e-01 -3.70016515e-01 5.27611852e-01
9.28909719e-01 -1.11566925e+00 4.95330244e-01 -9.38557506e-01
-2.21714646e-01 8.85814548e-01 6.60039127e-01 -3.59866023e-01
8.82370949e-01 2.46619821e-01 -2.54005969e-01 -4.49937671e-01
-4.20597523e-01 4.64495979e-02 4.08943668e-02 1.01906395e+00
-3.07177484e-01 6.53201714e-02 1.89716667e-01 2.79987872e-01
-4.18743342e-01 1.63104191e-01 1.62371263e-01 8.78207922e-01
-7.78714895e-01 -1.12715781e+00 -7.38981783e-01 3.60701352e-01
4.87035334e-01 3.39021091e-03 4.95605469e-02 7.79338896e-01
3.49332243e-01 3.91300291e-01 3.68312269e-01 -4.54089493e-01
4.82387960e-01 -2.39905864e-01 4.48981524e-01 -9.72250044e-01
-1.48655418e-02 4.74216975e-02 -3.79917413e-01 -8.03896248e-01
-5.05918622e-01 -7.92779803e-01 -1.33359659e+00 -4.58337545e-01
-4.91346657e-01 1.54249623e-01 1.12247562e+00 6.77860558e-01
6.70065641e-01 1.33508459e-01 5.81974030e-01 -1.56381893e+00
-3.99048895e-01 -8.74442875e-01 -8.40120554e-01 -9.08292551e-03
8.07555467e-02 -1.19012427e+00 -2.21564382e-01 -3.29359531e-01] | [8.034125328063965, -2.920042037963867] |
edfeff20-60f2-4cab-af00-268fd3fddf7c | topology-aware-uncertainty-for-image | 2306.05671 | null | https://arxiv.org/abs/2306.05671v1 | https://arxiv.org/pdf/2306.05671v1.pdf | Topology-Aware Uncertainty for Image Segmentation | Segmentation of curvilinear structures such as vasculature and road networks is challenging due to relatively weak signals and complex geometry/topology. To facilitate and accelerate large scale annotation, one has to adopt semi-automatic approaches such as proofreading by experts. In this work, we focus on uncertainty estimation for such tasks, so that highly uncertain, and thus error-prone structures can be identified for human annotators to verify. Unlike most existing works, which provide pixel-wise uncertainty maps, we stipulate it is crucial to estimate uncertainty in the units of topological structures, e.g., small pieces of connections and branches. To achieve this, we leverage tools from topological data analysis, specifically discrete Morse theory (DMT), to first capture the structures, and then reason about their uncertainties. To model the uncertainty, we (1) propose a joint prediction model that estimates the uncertainty of a structure while taking the neighboring structures into consideration (inter-structural uncertainty); (2) propose a novel Probabilistic DMT to model the inherent uncertainty within each structure (intra-structural uncertainty) by sampling its representations via a perturb-and-walk scheme. On various 2D and 3D datasets, our method produces better structure-wise uncertainty maps compared to existing works. | ['Chao Chen', 'Prateek Prasanna', 'Xiaoling Hu', 'Yikai Zhang', 'Saumya Gupta'] | 2023-06-09 | null | null | null | null | ['topological-data-analysis'] | ['graphs'] | [ 5.90363108e-02 4.61991370e-01 1.64057434e-01 -1.31135613e-01
-8.21037471e-01 -8.14126253e-01 3.80493462e-01 4.93130118e-01
3.50220241e-02 8.43201399e-01 4.71554957e-02 -3.83746445e-01
-2.48368666e-01 -8.57082903e-01 -8.98807943e-01 -6.02192879e-01
-6.03475124e-02 6.56874120e-01 5.62392414e-01 3.35499763e-01
3.72751057e-01 5.08414686e-01 -1.03139985e+00 -1.90717161e-01
1.35581911e+00 9.37430501e-01 1.33444583e-02 4.70536739e-01
-3.08692306e-01 4.08975989e-01 -2.94323355e-01 -4.23764348e-01
-2.21107323e-02 -3.07029509e-03 -8.79520714e-01 1.53271705e-02
1.94513157e-01 -2.52083540e-01 4.70875800e-02 1.18638229e+00
1.65376216e-01 1.20619955e-02 1.18828261e+00 -9.99428868e-01
-2.74209589e-01 7.99655855e-01 -9.03573036e-01 -2.67786738e-02
7.48003135e-03 -9.91151333e-02 9.68792140e-01 -6.56460822e-01
6.85706317e-01 1.21335256e+00 6.64699376e-01 1.35739237e-01
-1.39668584e+00 -3.69080961e-01 4.19215739e-01 -1.82546191e-02
-1.63140750e+00 -1.20437093e-01 1.05764842e+00 -7.05221295e-01
-2.17691343e-02 1.49692580e-01 3.27738881e-01 8.50687981e-01
1.91603839e-01 6.87467813e-01 1.09644175e+00 -8.60896930e-02
5.38944602e-01 -4.11702953e-02 -1.21499971e-01 7.83156216e-01
3.26557696e-01 -2.51527101e-01 -8.49563554e-02 -1.98069721e-01
9.70046520e-01 -2.69768625e-01 -1.72621891e-01 -4.48535889e-01
-1.22149849e+00 3.79395604e-01 5.23576200e-01 1.09039769e-01
-2.11822331e-01 1.32732853e-01 1.48718968e-01 -4.07448441e-01
5.36916018e-01 2.58528352e-01 -2.50615627e-01 -9.53330249e-02
-1.06537056e+00 9.36011821e-02 6.28461719e-01 8.49755347e-01
7.69731939e-01 -4.64833945e-01 -1.13520131e-01 6.67608798e-01
6.48037732e-01 2.21570149e-01 -2.77474552e-01 -1.06120670e+00
4.88406986e-01 5.51794291e-01 3.21173966e-01 -1.25292385e+00
-2.24003062e-01 -3.95009607e-01 -1.19205701e+00 2.77621537e-01
7.48284638e-01 -9.31610614e-02 -9.62389231e-01 1.59103572e+00
6.54575169e-01 5.08853018e-01 -3.85161191e-01 8.43816578e-01
5.86940706e-01 4.68426079e-01 -1.43267512e-01 -1.58084124e-01
1.18865764e+00 -5.46249449e-01 -3.41004759e-01 2.06285462e-01
2.81319171e-01 -6.13792181e-01 6.60115123e-01 3.87772739e-01
-9.00921941e-01 -1.56268463e-01 -9.18620050e-01 9.27980766e-02
-1.58021912e-01 1.26049131e-01 1.01020440e-01 4.56487715e-01
-5.84010661e-01 7.94449031e-01 -1.05042660e+00 2.54601091e-02
8.14364374e-01 -1.42042711e-01 -1.58149704e-01 7.22933412e-02
-9.80175853e-01 6.26722515e-01 2.61949331e-01 4.28484768e-01
-9.28975940e-01 -7.57374287e-01 -8.67138684e-01 -3.06041166e-02
6.40086055e-01 -5.03018439e-01 7.16583967e-01 -8.30131769e-02
-1.31539619e+00 5.19790709e-01 -1.93771824e-01 -2.95173913e-01
8.88484597e-01 1.90406702e-02 2.42056862e-01 2.20264688e-01
1.74542982e-02 6.39055371e-01 7.68968880e-01 -1.74732292e+00
-2.29235262e-01 -3.00455987e-01 1.13591880e-01 -1.14408582e-01
8.37233439e-02 -3.36316973e-01 -4.14692104e-01 -6.21768236e-01
6.61254406e-01 -8.62892807e-01 -5.50458908e-01 5.41570008e-01
-1.02907145e+00 -7.21288472e-02 6.15640759e-01 -7.77982414e-01
1.13318813e+00 -1.97355330e+00 1.94954798e-01 6.41868472e-01
6.56281769e-01 -3.35542969e-02 3.37994963e-01 9.44015607e-02
5.26877761e-01 4.95657116e-01 -7.77312398e-01 -3.82788837e-01
6.91820830e-02 2.65309721e-01 -1.31946579e-01 4.25212264e-01
4.30361003e-01 7.40847826e-01 -9.38672900e-01 -1.02388501e+00
3.82944614e-01 3.84271264e-01 -2.58651137e-01 -1.73029542e-01
-4.93522227e-01 7.11933017e-01 -9.17908132e-01 9.27081048e-01
1.04972076e+00 -3.60316873e-01 -3.11036017e-02 -4.28652138e-01
-2.61488497e-01 -7.24105164e-02 -1.58949852e+00 1.51065063e+00
-3.56901854e-01 2.91738391e-01 3.83180499e-01 -9.14068341e-01
9.26926255e-01 1.24839142e-01 4.87962157e-01 1.30566001e-01
1.92385733e-01 2.42671713e-01 -2.04580888e-01 -5.94095029e-02
2.24039167e-01 1.84624121e-01 -6.54099211e-02 4.29541498e-01
-3.56123120e-01 -4.10275906e-01 4.57708463e-02 4.41694230e-01
1.19040108e+00 2.00422719e-01 1.40489727e-01 -3.75224948e-01
4.10627037e-01 -2.07479268e-01 7.96949804e-01 4.46843117e-01
-3.12595218e-01 8.78471255e-01 7.84906805e-01 -2.19295949e-01
-1.03911328e+00 -1.34940696e+00 -5.35518706e-01 8.07251260e-02
4.75646943e-01 -2.88447738e-01 -8.47965181e-01 -7.65477896e-01
1.34946257e-01 4.20997947e-01 -6.68323398e-01 1.77857012e-01
-1.86060145e-01 -5.04760623e-01 4.17689741e-01 5.80456436e-01
5.40494919e-01 -3.74233097e-01 -3.81555855e-01 3.26208800e-01
-2.67988414e-01 -1.08487701e+00 -4.45050687e-01 -9.90699828e-02
-8.04325521e-01 -1.05796194e+00 -6.93331838e-01 -2.25015387e-01
9.87421393e-01 -1.37893260e-01 8.70493054e-01 -6.88642859e-02
-4.63893384e-01 1.49522275e-01 -2.58507818e-01 -3.16905171e-01
-9.72654819e-02 -3.22481394e-02 -1.27074465e-01 5.21242395e-02
-3.73593718e-01 -8.46707821e-01 -5.36898494e-01 7.80755341e-01
-7.64452815e-01 1.08155094e-01 6.48702383e-01 7.53490508e-01
1.19124866e+00 5.30218244e-01 3.02323818e-01 -7.47335017e-01
3.77324104e-01 -4.88128155e-01 -8.07256639e-01 4.09884423e-01
-5.64912677e-01 2.88579285e-01 3.17376405e-01 -2.68670738e-01
-1.08298743e+00 1.50283292e-01 6.97546676e-02 -5.05565047e-01
-1.36008367e-01 5.80002725e-01 -2.75989443e-01 -1.82544902e-01
4.39172000e-01 -7.67765641e-02 -1.47192925e-01 -4.07814711e-01
4.47872549e-01 5.61253786e-01 4.56985533e-01 -1.07963407e+00
8.77896070e-01 6.56639576e-01 4.66317087e-01 -5.88136315e-01
-7.11822033e-01 -1.66322395e-01 -9.93672967e-01 -3.84711534e-01
6.84092581e-01 -5.57314634e-01 -5.89864790e-01 2.84043998e-01
-1.20711398e+00 -1.20197736e-01 -1.59020409e-01 3.76913995e-01
-3.59998673e-01 6.00778043e-01 -5.25715411e-01 -1.03199613e+00
-4.67896871e-02 -1.15737307e+00 1.28393388e+00 2.61312783e-01
-1.55999213e-01 -9.65775490e-01 -7.51857238e-04 3.02833766e-01
2.35311687e-01 7.14103341e-01 8.57779324e-01 -1.62187502e-01
-9.99870121e-01 -1.07118897e-01 -5.61377585e-01 2.28897974e-01
2.39106286e-02 6.33396327e-01 -7.14390337e-01 2.69216120e-01
-4.12373245e-01 -6.11195862e-02 7.20741093e-01 6.68324828e-01
1.43638206e+00 1.68862939e-02 -6.01613641e-01 3.38958025e-01
1.06187260e+00 -3.84072125e-01 7.14126885e-01 -4.44992818e-02
7.16376364e-01 8.48374784e-01 6.95328295e-01 5.60011685e-01
6.24247909e-01 4.53224808e-01 7.13202715e-01 2.87317514e-01
1.35651603e-01 -4.85272259e-01 -1.81417406e-01 5.54482937e-01
-1.30892649e-01 -1.86511844e-01 -9.77606833e-01 5.64747214e-01
-1.96488249e+00 -6.01672590e-01 -3.50402504e-01 2.20946836e+00
8.13221633e-01 3.78316045e-01 -1.19068481e-01 2.73827044e-03
9.32898283e-01 1.10892601e-01 -6.71673298e-01 7.56865665e-02
4.77773547e-02 -2.77688235e-01 4.93830413e-01 5.30562639e-01
-9.63949084e-01 6.26497746e-01 5.84741592e+00 1.09073567e+00
-7.67200112e-01 -2.64172405e-01 9.96992052e-01 1.57146141e-01
-7.09128141e-01 1.79465488e-01 -6.17449403e-01 7.52019644e-01
3.80341977e-01 -7.81991407e-02 1.79685280e-01 7.08781719e-01
3.14653665e-01 -3.55560631e-01 -7.90334463e-01 6.06310725e-01
-3.99311125e-01 -1.32596850e+00 -2.12619200e-01 1.82230234e-01
8.39168370e-01 -2.50550747e-01 -2.53495783e-01 -2.65035212e-01
5.26289821e-01 -9.81913149e-01 6.70005202e-01 9.36800361e-01
6.41560435e-01 -6.18050456e-01 6.80972815e-01 4.88504291e-01
-1.41826677e+00 4.59297448e-01 -4.70128894e-01 4.06777829e-01
3.79309952e-01 1.47518396e+00 -6.64542258e-01 7.24095345e-01
6.47576630e-01 6.92759514e-01 -2.90735155e-01 1.21922290e+00
-4.81552273e-01 6.60073638e-01 -7.86756814e-01 7.13387579e-02
-8.79736990e-02 -4.30083364e-01 7.48577952e-01 7.86318958e-01
5.03820419e-01 3.85641679e-02 1.36820540e-01 1.43797278e+00
-1.00140825e-01 7.50373024e-03 -3.95606756e-01 -4.02247123e-02
8.40440571e-01 1.51479578e+00 -1.10803711e+00 -6.69001564e-02
5.18202223e-02 7.51139641e-01 3.28056812e-01 1.10368334e-01
-7.78494239e-01 -2.63321251e-01 3.29184055e-01 2.50278771e-01
2.91626632e-01 -4.34604317e-01 -7.29740441e-01 -1.03939176e+00
3.80865663e-01 -1.70484006e-01 1.56666860e-01 -7.65163779e-01
-1.38851213e+00 2.78412342e-01 1.55019611e-02 -1.29862905e+00
1.16857268e-01 -4.35247779e-01 -9.00938451e-01 7.90603101e-01
-1.30801272e+00 -1.23276258e+00 -2.91501671e-01 4.52466421e-02
1.19114190e-01 5.16140997e-01 3.78376663e-01 1.11363925e-01
-6.15226626e-01 3.66972417e-01 1.49113655e-01 -5.30015957e-03
4.73675460e-01 -1.37974823e+00 1.51948467e-01 7.98164070e-01
5.23827448e-02 6.24706030e-01 7.17776358e-01 -9.54583943e-01
-9.38493252e-01 -1.12606919e+00 5.74386895e-01 -5.90592921e-01
8.14365685e-01 -3.25162441e-01 -9.86669838e-01 3.31304163e-01
-5.04101276e-01 4.26328719e-01 2.93321878e-01 9.30381268e-02
-2.32316241e-01 1.30886689e-01 -1.50341165e+00 4.73328471e-01
1.07179201e+00 -3.48546684e-01 -2.89499879e-01 1.86027214e-02
6.29699349e-01 -5.01596451e-01 -1.14260805e+00 6.08796954e-01
7.11426497e-01 -7.17003167e-01 8.01492810e-01 -7.97135457e-02
6.21176124e-01 -8.27924192e-01 -8.07440095e-03 -1.31398487e+00
-8.35470259e-02 -5.56208193e-01 -2.13647082e-01 1.53578341e+00
4.80871201e-01 -4.30068135e-01 8.82106125e-01 7.17373013e-01
-2.22162366e-01 -9.09978271e-01 -1.31611085e+00 -7.23329365e-01
7.34949559e-02 -5.59551001e-01 7.49623597e-01 7.19342828e-01
-1.68417007e-01 -1.36358961e-01 -2.34362140e-01 6.22169197e-01
8.88003707e-01 7.54530877e-02 5.03080606e-01 -1.57855797e+00
-5.36148138e-02 -4.77060795e-01 -5.27301073e-01 -1.00159061e+00
-1.49041014e-02 -5.25398970e-01 3.13015729e-01 -1.58915019e+00
3.73828458e-03 -8.54480982e-01 -9.31200162e-02 9.78115350e-02
-1.88618183e-01 2.84088906e-02 -3.77400637e-01 2.02053726e-01
-7.73156285e-01 6.88052714e-01 1.33728874e+00 -1.41434684e-01
6.04988821e-02 1.30142063e-01 -6.62317872e-01 8.74392390e-01
5.56489646e-01 -3.29040319e-01 -3.04867417e-01 -1.71832934e-01
3.31009030e-01 3.62116061e-02 5.53396106e-01 -9.60584939e-01
2.66605049e-01 -2.30923533e-01 1.75954044e-01 -9.63282526e-01
4.05898653e-02 -8.92356932e-01 1.32929087e-01 4.65865759e-03
-1.34799153e-01 -5.31334877e-01 5.04360609e-02 1.00260520e+00
-1.68264434e-01 -3.81504864e-01 6.66611731e-01 -3.38207595e-02
-1.95491984e-01 4.64881480e-01 -2.36880034e-01 -2.82374807e-02
1.24311948e+00 9.80836991e-03 -3.30034912e-01 -2.17567697e-01
-7.68798590e-01 5.76329172e-01 6.46098495e-01 -2.54512936e-01
5.29083908e-01 -1.29432869e+00 -6.70954764e-01 -2.13110939e-01
2.35275373e-01 7.92420149e-01 4.83667970e-01 9.67413008e-01
-4.99696016e-01 1.85768772e-02 8.43227804e-02 -8.29766452e-01
-1.04076135e+00 1.70643851e-01 2.17121989e-01 -2.64841795e-01
-3.65751863e-01 9.90586519e-01 -1.43141765e-02 -4.86321062e-01
1.08582959e-01 -6.56165063e-01 -1.09968327e-01 2.58731823e-02
2.73365855e-01 6.05062544e-01 -2.08054706e-01 -3.62878710e-01
-2.67930329e-01 7.92858958e-01 -3.97518985e-02 -1.81252416e-02
1.13881028e+00 -4.13148373e-01 -1.77279964e-01 4.40752834e-01
6.44600987e-01 1.66891932e-01 -1.69832039e+00 -3.56292963e-01
-6.35262951e-02 -4.89071339e-01 2.13402867e-01 -6.38308406e-01
-8.70840490e-01 1.05657721e+00 2.53967464e-01 2.09331587e-01
6.75819933e-01 2.99914271e-01 6.57577932e-01 5.82059249e-02
5.87293446e-01 -1.14497590e+00 -2.13195011e-01 1.52887225e-01
7.58812606e-01 -1.24367929e+00 1.79335698e-01 -9.03342485e-01
-3.99490714e-01 1.12550807e+00 4.26257521e-01 7.43164420e-02
8.71301115e-01 3.22519898e-01 -3.34049702e-01 -1.99116856e-01
-5.29497385e-01 -1.25986543e-02 4.70247060e-01 5.73787034e-01
6.04311302e-02 2.28715196e-01 -6.28048256e-02 6.42469406e-01
5.15805781e-02 -1.06040157e-01 4.67361212e-01 7.98124850e-01
-5.53246140e-01 -1.04078662e+00 -4.06506956e-01 7.50242889e-01
-9.67119709e-02 4.63663414e-02 -2.88226664e-01 2.85030574e-01
1.84319898e-01 6.46273017e-01 -5.95760196e-02 -3.60699385e-01
8.31708685e-02 -3.67760152e-01 3.42546403e-01 -4.92504060e-01
2.58184761e-01 1.81179084e-02 1.06202334e-01 -4.14945006e-01
-2.04186723e-01 -9.25726235e-01 -1.36359227e+00 -1.44252673e-01
-4.50369298e-01 1.05495863e-01 8.30261230e-01 1.11034846e+00
3.88486087e-01 5.05626261e-01 5.31048000e-01 -7.29735255e-01
-3.96888793e-01 -8.47131431e-01 -7.92104959e-01 4.58613895e-02
-2.63756290e-02 -1.03672338e+00 -5.88986039e-01 -1.02265790e-01] | [9.348950386047363, -0.30207139253616333] |
ba43d44b-f255-4302-a5fa-a8c0d957b396 | use-image-clustering-to-facilitate-technology | 2112.08604 | null | https://arxiv.org/abs/2112.08604v1 | https://arxiv.org/pdf/2112.08604v1.pdf | Use Image Clustering to Facilitate Technology Assisted Review | During the past decade breakthroughs in GPU hardware and deep neural networks technologies have revolutionized the field of computer vision, making image analytical potentials accessible to a range of real-world applications. Technology Assisted Review (TAR) in electronic discovery though traditionally has dominantly dealt with textual content, is witnessing a rising need to incorporate multimedia content in the scope. We have developed innovative image analytics applications for TAR in the past years, such as image classification, image clustering, and object detection, etc. In this paper, we discuss the use of image clustering applications to facilitate TAR based on our experiences in serving clients. We describe our general workflow on leveraging image clustering in tasks and use statistics from real projects to showcase the effectiveness of using image clustering in TAR. We also summarize lessons learned and best practices on using image clustering in TAR. | ['Adam Dabrowski', 'Han Qin', 'Hilary Quatinetz', 'Fusheng Wei', 'Haozhen Zhao'] | 2021-12-16 | null | null | null | null | ['image-clustering'] | ['computer-vision'] | [-8.47155377e-02 -4.37089294e-01 9.86058861e-02 -3.92927349e-01
-6.35206938e-01 -6.20620012e-01 5.45019984e-01 3.01995903e-01
-4.87164348e-01 -1.68789640e-01 -3.10117062e-02 -5.40847301e-01
-1.19417779e-01 -4.12654102e-01 -6.44698322e-01 -4.89976704e-01
-1.05521567e-01 1.94383115e-01 1.90710589e-01 1.06601395e-01
7.58038282e-01 5.72785258e-01 -1.85206509e+00 7.31887102e-01
2.65728414e-01 7.65216827e-01 5.77285230e-01 7.35516012e-01
-3.48555326e-01 1.04015338e+00 -8.19915950e-01 -4.22129482e-01
3.64212096e-01 1.48345023e-01 -6.36075258e-01 3.97865832e-01
2.32777193e-01 -3.58432204e-01 -7.53662884e-02 7.88440883e-01
6.49721861e-01 -6.37929440e-02 3.42410713e-01 -1.28409648e+00
-5.03903270e-01 6.70473576e-01 -1.16554046e+00 6.31213784e-01
4.67106104e-02 2.33044222e-01 6.16753519e-01 -8.37496102e-01
7.02315390e-01 1.07639253e+00 6.86153710e-01 -4.77334149e-02
-8.09879839e-01 -6.68861091e-01 1.04434334e-01 4.32955682e-01
-1.07721186e+00 -4.36624587e-01 4.29225653e-01 -5.71535170e-01
1.31855845e+00 2.04916224e-02 6.26304030e-01 4.48270351e-01
7.19890743e-02 8.67922306e-01 8.61714661e-01 -8.18266988e-01
3.36301357e-01 2.55189657e-01 3.17609340e-01 7.14753807e-01
1.12122878e-01 -4.99401808e-01 -3.53552818e-01 1.96853243e-02
6.96391404e-01 -7.58340731e-02 1.59530923e-01 -1.19999476e-01
-1.31356013e+00 6.96019948e-01 2.37438381e-01 4.08783555e-01
-2.85833776e-01 5.03976703e-01 9.16377485e-01 1.30386902e-02
4.78963166e-01 3.57067049e-01 -2.48346597e-01 -4.30005491e-01
-7.18290627e-01 1.83297411e-01 5.53321183e-01 1.02939188e+00
7.14200497e-01 -1.44110456e-01 1.31568342e-01 8.31253767e-01
2.13137940e-01 2.16113180e-02 3.86826277e-01 -1.42806506e+00
1.54538542e-01 6.10609829e-01 -9.59090814e-02 -1.01828599e+00
-1.73255086e-01 -4.25515711e-01 -3.43188345e-01 3.92993480e-01
2.33116239e-01 -6.00264855e-02 -8.82966459e-01 7.95647681e-01
2.36396343e-01 -7.20257238e-02 -4.06699240e-01 7.56813943e-01
8.55505168e-01 5.23271024e-01 9.03793946e-02 1.64412841e-01
1.32865524e+00 -7.92623222e-01 -4.24120843e-01 2.30552122e-01
8.58884454e-01 -1.11594665e+00 1.10070598e+00 1.01435530e+00
-1.09268928e+00 -3.74568820e-01 -9.86961007e-01 -1.04047932e-01
-6.14414394e-01 1.80080727e-01 9.22573984e-01 9.99392569e-01
-1.27002656e+00 4.30250436e-01 -9.37143981e-01 -7.90201664e-01
6.91835046e-01 4.98197466e-01 -1.41225487e-01 -2.24795595e-01
-1.94210649e-01 4.03357238e-01 1.89250663e-01 -3.15510899e-01
-7.73160458e-01 -7.16335118e-01 -1.42517880e-01 1.16792396e-02
2.47095019e-01 -3.72547328e-01 1.34696555e+00 -1.08682275e+00
-1.02622437e+00 1.10988617e+00 1.78835779e-01 -2.05801561e-01
1.91946983e-01 -1.10580988e-01 -1.78057238e-01 2.28805229e-01
3.09453066e-02 7.62260258e-01 5.85327387e-01 -1.07700157e+00
-1.06551504e+00 -4.60239351e-01 -8.77188146e-02 2.43040726e-01
-6.17965698e-01 5.92539310e-01 -1.21296096e+00 -1.88116997e-01
-3.67896944e-01 -7.12480068e-01 -3.45066071e-01 -1.52288124e-01
-3.35519239e-02 -3.61245751e-01 1.04681873e+00 -3.33001554e-01
8.18719685e-01 -2.20981860e+00 -4.37902689e-01 3.31863463e-01
3.83937567e-01 2.36895651e-01 3.87944803e-02 4.19672012e-01
3.83804590e-02 3.03574890e-01 4.03227568e-01 -4.84188527e-01
-1.24444835e-01 2.24600583e-02 -5.76221496e-02 3.98686647e-01
-2.73520797e-01 6.79845393e-01 -7.45086491e-01 -5.08806705e-01
3.91666353e-01 4.42821234e-01 -3.46224457e-01 -2.28912041e-01
-1.24196157e-01 3.27729024e-02 -3.02517653e-01 8.17871809e-01
5.57503521e-01 -6.13822460e-01 2.38830909e-01 -2.96230949e-02
-6.20647430e-01 -1.92091420e-01 -8.23879719e-01 1.70075572e+00
-3.55551928e-01 1.14305472e+00 2.08512992e-01 -1.05711269e+00
5.81104159e-01 -2.57644113e-02 6.37730300e-01 -7.84903586e-01
1.90632224e-01 -7.88148269e-02 9.45632309e-02 -6.38327241e-01
5.74069798e-01 5.46981096e-01 4.23848599e-01 6.77500963e-01
-2.30172321e-01 1.47106007e-01 2.81050205e-01 5.62297225e-01
1.42145860e+00 1.37393981e-01 -1.51486963e-01 -2.62043238e-01
3.89348641e-02 7.97709882e-01 1.62499920e-02 8.48761439e-01
-2.11138770e-01 6.85502350e-01 2.16335937e-01 -5.40420055e-01
-1.42422545e+00 -5.79880774e-01 -3.60737070e-02 1.30224395e+00
-4.37948294e-02 -5.57399273e-01 -9.06663239e-01 -4.25349832e-01
-1.69963956e-01 2.61583507e-01 -3.50242555e-01 4.90253091e-01
-4.23336655e-01 -9.42458928e-01 3.89907122e-01 4.74644154e-01
6.64601505e-01 -1.12799120e+00 -8.84204328e-01 2.09877416e-01
2.15723768e-01 -1.28854203e+00 -1.51812166e-01 1.92247614e-01
-7.46691883e-01 -8.52394462e-01 -5.97037792e-01 -7.94071376e-01
4.25877988e-01 1.06782317e+00 1.32517111e+00 4.23913240e-01
-1.04508758e+00 8.16802800e-01 -5.24465442e-01 -8.47606182e-01
-1.89424634e-01 9.09608230e-02 -2.89201945e-01 -4.56011087e-01
7.58270442e-01 -3.41413170e-01 -7.75520682e-01 2.27005824e-01
-8.37117970e-01 8.50853100e-02 5.31399488e-01 2.06930384e-01
3.34167600e-01 1.60318807e-01 4.43140209e-01 -8.47866416e-01
6.53653443e-01 -5.60959637e-01 -7.20223486e-01 2.19582513e-01
-6.85649335e-01 -4.13218737e-01 1.10984586e-01 -2.29124412e-01
-9.69390213e-01 1.00460812e-01 8.82032737e-02 -4.86386061e-01
-3.32197636e-01 7.17253625e-01 3.43389899e-01 -2.60546535e-01
7.97577083e-01 -2.07572073e-01 2.31249765e-01 -1.91619992e-01
4.77230787e-01 9.49297130e-01 4.74819869e-01 -5.41284204e-01
5.37486710e-02 7.74617493e-01 -2.78541505e-01 -9.51257885e-01
-1.88730195e-01 -8.87706876e-01 -1.72114447e-01 -5.55813909e-01
9.18548882e-01 -1.04100311e+00 -9.33648229e-01 2.13484764e-01
-1.00793993e+00 -3.73930216e-01 1.04255997e-01 4.14772570e-01
-2.04777986e-01 5.27583718e-01 -4.95756060e-01 -6.82948589e-01
-3.24277461e-01 -1.49560058e+00 9.25266385e-01 3.05432439e-01
2.17659622e-02 -7.69778490e-01 -2.39909858e-01 6.14165246e-01
4.09534812e-01 3.24178748e-02 7.02688932e-01 -4.36462224e-01
-9.31212544e-01 -1.99702695e-01 -6.19770527e-01 1.80391148e-01
-1.56687334e-01 4.82716978e-01 -9.91910577e-01 -9.40622762e-02
-2.34424591e-01 -1.32858887e-01 6.64698899e-01 5.07379472e-01
1.62334657e+00 3.18603456e-01 -6.43500745e-01 3.28438401e-01
1.53123915e+00 4.90725338e-01 7.85090864e-01 8.71962965e-01
8.13041806e-01 7.10264623e-01 5.83997428e-01 6.19843960e-01
2.86113888e-01 4.27245736e-01 3.99818927e-01 -2.17848167e-01
3.49751674e-02 5.73672950e-01 2.53372878e-01 7.83356965e-01
-2.75003910e-01 -1.46961138e-01 -1.27802300e+00 6.92395806e-01
-2.01405764e+00 -7.41356969e-01 -5.52366555e-01 1.90780568e+00
2.23039582e-01 8.65501445e-03 2.39434376e-01 -2.37878576e-01
8.51210654e-01 -6.55088186e-01 -3.20880622e-01 -5.65530717e-01
2.44993344e-01 5.69022261e-02 6.32709384e-01 -1.97355926e-01
-9.09028292e-01 6.96949422e-01 6.49332094e+00 9.28663373e-01
-1.21122289e+00 2.78363377e-01 9.75479126e-01 -4.03066128e-02
6.23761937e-02 -9.52141434e-02 -6.31045341e-01 3.58873218e-01
7.82043755e-01 -6.36385307e-02 3.24225754e-01 1.17943645e+00
3.48841161e-01 -5.16360223e-01 -9.41846967e-01 1.18295324e+00
-6.41514882e-02 -1.67853987e+00 -1.08366549e-01 3.18384886e-01
7.44997561e-01 4.87928122e-01 3.50054234e-01 6.08125180e-02
5.06684184e-01 -9.19003785e-01 5.01445353e-01 2.79657468e-02
4.47501421e-01 -9.46366787e-01 6.95459723e-01 2.17409357e-01
-8.26604366e-01 -5.40949516e-02 -3.37949485e-01 -1.34601504e-01
-3.92254561e-01 7.69530058e-01 -1.25554144e+00 4.12424803e-01
1.44972444e+00 6.19365156e-01 -6.91586614e-01 1.42126393e+00
5.25798142e-01 6.70719266e-01 -2.70961821e-01 -9.00364295e-02
3.03617388e-01 -3.15182835e-01 1.83061555e-01 1.48190832e+00
1.93059355e-01 8.60483646e-02 3.43115404e-02 6.30901456e-01
8.57548229e-03 1.00231066e-01 -5.81067145e-01 -2.00010344e-01
5.10109246e-01 1.71446991e+00 -1.66831613e+00 -4.19195235e-01
-7.82269180e-01 7.01547146e-01 8.19981098e-03 1.23094119e-01
-6.42283201e-01 -4.69269753e-01 4.22950745e-01 8.60896930e-02
4.71334189e-01 -4.84861463e-01 -6.10280991e-01 -6.24025822e-01
-3.37892532e-01 -1.01069844e+00 9.03146937e-02 -9.72434461e-01
-1.06916153e+00 4.33563292e-01 7.56125944e-03 -9.56140220e-01
1.57808542e-01 -7.07750380e-01 -6.28203094e-01 5.35060406e-01
-8.26514900e-01 -9.58449066e-01 -6.02876365e-01 4.01921421e-01
6.56532705e-01 -3.33880991e-01 3.06965351e-01 4.86727804e-01
-6.34994805e-01 3.16383630e-01 4.16709810e-01 -9.03642923e-02
7.38784790e-01 -1.22374582e+00 8.07436824e-01 7.01781929e-01
2.46304423e-01 8.29561353e-01 5.23148835e-01 -5.53994119e-01
-1.70901668e+00 -8.18849087e-01 -4.75715511e-02 -4.93130177e-01
5.62550604e-01 -5.04411161e-01 -5.35853505e-01 5.57777047e-01
6.89468384e-01 -3.13338786e-01 5.61718047e-01 6.49478585e-02
-7.77733400e-02 9.29444507e-02 -1.00344110e+00 6.76408648e-01
7.50403345e-01 -3.45281035e-01 1.64771676e-01 5.51214814e-01
4.22652096e-01 -2.80964673e-01 -7.27714777e-01 1.66311150e-03
5.13376892e-01 -1.07553828e+00 9.99217629e-01 -3.87662463e-02
5.34628391e-01 -5.74236572e-01 -1.43087208e-02 -7.12448061e-01
-2.39733785e-01 -5.30510545e-01 4.14260805e-01 1.18428671e+00
7.75040966e-03 -9.65378433e-02 1.13582921e+00 6.88953400e-01
-3.36217552e-01 -5.20495415e-01 -2.98543036e-01 -4.20931906e-01
-1.90706864e-01 -8.29531014e-01 2.55379379e-01 9.94751692e-01
-6.11389577e-02 8.00182149e-02 2.11534187e-01 -9.74641219e-02
6.78209245e-01 -1.84268892e-01 1.02759433e+00 -1.13108122e+00
-3.56691599e-01 -5.28273880e-01 -5.85564077e-01 -6.66394114e-01
-3.98175061e-01 -7.44079888e-01 -2.72386577e-02 -1.60363519e+00
6.40991211e-01 -4.32909906e-01 -3.51454020e-02 4.12283182e-01
1.65104896e-01 7.12380290e-01 3.43795091e-01 4.63081270e-01
-9.71935809e-01 -2.96013743e-01 7.85717726e-01 -7.47359917e-02
-3.46127823e-02 -4.46848065e-01 -8.64820480e-01 6.92690551e-01
7.81106830e-01 -2.74004579e-01 -3.30909759e-01 -5.59671998e-01
4.22909230e-01 -5.67199886e-01 1.19991690e-01 -1.12395370e+00
5.65458775e-01 3.83512825e-01 6.85572565e-01 -7.24780917e-01
-3.99967916e-02 -8.00063431e-01 2.57183135e-01 1.56113490e-01
-2.20077068e-01 3.46242994e-01 3.42269897e-01 3.16276610e-01
-2.08038650e-02 -4.29222912e-01 4.30419296e-01 -5.24793804e-01
-1.20422006e+00 -1.82046574e-02 -8.35388720e-01 -4.81039584e-01
1.12830520e+00 -4.40242052e-01 -5.06876647e-01 -1.99554786e-01
-4.73810822e-01 2.10725695e-01 6.60773337e-01 2.77537018e-01
4.71859336e-01 -7.55537093e-01 -4.17004436e-01 -3.94407690e-01
1.21134028e-01 -9.23437402e-02 1.32670969e-01 8.25419486e-01
-1.00779748e+00 1.27668425e-01 -2.79495001e-01 -1.04143524e+00
-1.73721242e+00 6.59074605e-01 -6.81538088e-03 1.26862600e-01
-8.13731790e-01 6.97785676e-01 3.08777183e-01 2.42185593e-03
4.65140194e-01 -1.30695671e-01 -4.98665050e-02 -9.49227661e-02
6.97843432e-01 6.52299464e-01 5.96375346e-01 1.75881628e-02
-2.49895737e-01 4.08972472e-01 -6.63399220e-01 -3.70921493e-01
1.44707179e+00 -3.58809471e-01 -3.24464262e-01 4.22833771e-01
1.14499688e+00 -3.02343935e-01 -8.16065550e-01 2.49655321e-01
1.67874336e-01 -3.04906428e-01 5.19239366e-01 -7.94519961e-01
-1.11483943e+00 8.38330925e-01 9.85051453e-01 5.08702457e-01
1.14428043e+00 1.54578403e-01 3.21833432e-01 5.51416159e-01
5.12429118e-01 -1.29924834e+00 2.21786365e-01 2.42500037e-01
4.49487269e-01 -1.18776190e+00 4.67317581e-01 -4.70900893e-01
-7.13342011e-01 1.29444468e+00 4.53480005e-01 -1.07955195e-01
5.70613027e-01 5.25426030e-01 1.80449262e-01 -6.37175262e-01
-6.81254387e-01 -7.50731379e-02 -2.98185855e-01 7.29093671e-01
7.57811308e-01 -2.19295681e-01 -5.43914624e-02 -6.74155075e-03
2.31457785e-01 5.49860075e-02 8.86063814e-01 1.31832778e+00
-4.75003242e-01 -1.08578360e+00 -6.88341260e-01 6.46042466e-01
-9.31805372e-01 -1.18019976e-01 -4.23666269e-01 8.62776458e-01
2.67961413e-01 1.03015769e+00 3.68572086e-01 -4.34855998e-01
-1.18930429e-01 -4.73836660e-01 5.19373834e-01 -5.30661523e-01
-1.12931800e+00 3.22081923e-01 8.76247287e-02 -4.84950930e-01
-5.10851204e-01 -8.62944305e-01 -1.22284710e+00 -5.16397595e-01
-3.46502662e-01 -4.34275791e-02 1.51530898e+00 8.14971745e-01
7.21298516e-01 5.32765329e-01 2.51764357e-01 -1.18593836e+00
1.91550434e-01 -6.92602992e-01 -4.27096784e-01 6.72955289e-02
-2.20018834e-01 -4.22179908e-01 4.00175378e-02 3.87874693e-01] | [9.23216438293457, 1.9962278604507446] |
1be86529-4d20-442c-a9a1-39369b0e8903 | model-and-data-agreement-for-learning-with | 2212.01054 | null | https://arxiv.org/abs/2212.01054v2 | https://arxiv.org/pdf/2212.01054v2.pdf | Model and Data Agreement for Learning with Noisy Labels | Learning with noisy labels is a vital topic for practical deep learning as models should be robust to noisy open-world datasets in the wild. The state-of-the-art noisy label learning approach JoCoR fails when faced with a large ratio of noisy labels. Moreover, selecting small-loss samples can also cause error accumulation as once the noisy samples are mistakenly selected as small-loss samples, they are more likely to be selected again. In this paper, we try to deal with error accumulation in noisy label learning from both model and data perspectives. We introduce mean point ensemble to utilize a more robust loss function and more information from unselected samples to reduce error accumulation from the model perspective. Furthermore, as the flip images have the same semantic meaning as the original images, we select small-loss samples according to the loss values of flip images instead of the original ones to reduce error accumulation from the data perspective. Extensive experiments on CIFAR-10, CIFAR-100, and large-scale Clothing1M show that our method outperforms state-of-the-art noisy label learning methods with different levels of label noise. Our method can also be seamlessly combined with other noisy label learning methods to further improve their performance and generalize well to other tasks. The code is available in https://github.com/zyh-uaiaaaa/MDA-noisy-label-learning. | ['Dongchao Wen', 'Hongzhi Shi', 'Yunfeng Yin', 'Xingchen Cui', 'Weihong Deng', 'Yuhang Zhang'] | 2022-12-02 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [-4.75456342e-02 -1.98816419e-01 9.08373222e-02 -7.57379830e-01
-1.37002861e+00 -3.97035301e-01 1.64791316e-01 1.83351308e-01
-7.23820269e-01 9.93020833e-01 -1.33975551e-01 1.56292737e-01
8.77548605e-02 -5.39351523e-01 -7.39747047e-01 -9.68638122e-01
3.63240242e-01 3.82622302e-01 8.70384723e-02 7.14725032e-02
-1.64695978e-01 -1.82595894e-01 -1.63589060e+00 3.57963413e-01
7.33859718e-01 1.29566026e+00 1.11364082e-01 6.54596984e-02
-1.20175965e-01 8.31001043e-01 -6.71865821e-01 -5.23900986e-01
2.89091200e-01 -2.55995363e-01 -7.29287028e-01 -1.64044909e-02
7.07847416e-01 -4.87023592e-02 2.71602497e-02 1.46308029e+00
9.22818065e-01 3.26433271e-01 4.12422836e-01 -1.11232162e+00
-5.43253958e-01 6.38892293e-01 -4.52505410e-01 -1.41226858e-01
-1.64222628e-01 4.48346138e-02 9.94055390e-01 -1.10482383e+00
4.64498073e-01 1.38486946e+00 8.00975859e-01 6.93141997e-01
-1.34435117e+00 -9.92466688e-01 4.33117360e-01 3.32282335e-01
-1.35931063e+00 -2.83476561e-01 7.32551038e-01 -1.77166402e-01
1.37002200e-01 1.04207255e-01 2.21619710e-01 1.42866850e+00
-1.85247585e-01 8.35909605e-01 1.38462842e+00 -2.59994537e-01
4.07340914e-01 8.79379213e-02 4.06919777e-01 4.88928527e-01
2.41844609e-01 1.39533699e-01 -4.99695778e-01 -1.44710168e-01
4.61577848e-02 2.58942880e-02 -3.85247082e-01 -6.72525242e-02
-9.91567671e-01 8.87913108e-01 6.07322633e-01 1.81906417e-01
-5.37800305e-02 5.24026871e-01 5.89257658e-01 4.60744292e-01
9.68061209e-01 4.23423022e-01 -6.68513358e-01 3.76187563e-02
-9.11126673e-01 2.52560437e-01 4.07925069e-01 7.03450263e-01
8.93351257e-01 -1.39926508e-01 -3.49919558e-01 1.23167980e+00
3.64733994e-01 3.86596054e-01 3.78867030e-01 -1.13666916e+00
4.21257138e-01 2.67448664e-01 3.96091752e-02 -7.84472108e-01
-4.46844608e-01 -8.53088617e-01 -8.12938631e-01 1.30782142e-01
4.25282001e-01 -1.60456195e-01 -1.02785027e+00 1.95585287e+00
3.86723161e-01 3.89675677e-01 -1.37946516e-01 1.07922661e+00
8.42001617e-01 5.12567878e-01 3.39478970e-01 -1.06156133e-01
1.21378267e+00 -9.79732037e-01 -8.01932752e-01 -3.31110358e-01
1.03786826e+00 -7.75834858e-01 1.25445950e+00 5.47242880e-01
-6.27739489e-01 -5.67620337e-01 -8.44291151e-01 -1.12022109e-01
-3.06580603e-01 2.48673022e-01 1.82968602e-01 5.08396089e-01
-6.83306575e-01 7.02246726e-01 -4.38019216e-01 -6.65913522e-03
7.83719599e-01 5.06438091e-02 -9.97532085e-02 -4.79667664e-01
-1.35202312e+00 6.07914686e-01 3.49542022e-01 1.47483528e-01
-1.05233681e+00 -7.19429016e-01 -6.55400455e-01 -1.75504342e-01
6.33115232e-01 -4.43508238e-01 1.36269438e+00 -9.34133410e-01
-8.88828456e-01 7.75142968e-01 -3.25844921e-02 -2.72320777e-01
8.42446327e-01 -2.79718548e-01 -1.91213682e-01 -1.19907811e-01
2.54913539e-01 7.77832448e-01 7.63466954e-01 -1.70637596e+00
-6.51329160e-01 -3.54668856e-01 -9.76740345e-02 5.04821427e-02
-2.36654148e-01 -2.03636408e-01 -8.37712437e-02 -9.18510854e-01
3.80244374e-01 -1.07010829e+00 -1.40513793e-01 1.07900456e-01
-3.68821055e-01 -2.95421600e-01 8.31658244e-01 -1.71877146e-01
1.11237216e+00 -2.21007872e+00 -2.24368528e-01 8.15908387e-02
2.79222637e-01 3.19951862e-01 -3.57104778e-01 -8.54642391e-02
-1.35198191e-01 4.52523887e-01 -2.20028609e-01 -8.44758272e-01
-1.01864547e-01 2.10276216e-01 -4.45845574e-02 4.31694448e-01
-1.32914009e-02 6.39331877e-01 -1.20767987e+00 -3.81584048e-01
-4.77138422e-02 3.30539942e-01 -1.44135222e-01 -4.67927270e-02
-2.58191913e-01 6.17203712e-01 -3.59703362e-01 6.18127704e-01
8.75803888e-01 -1.72342524e-01 -1.89009055e-01 -5.07520974e-01
4.85057175e-01 1.90390736e-01 -1.38481271e+00 1.42849350e+00
-7.05020905e-01 3.55240732e-01 7.09121898e-02 -8.74791443e-01
7.74315536e-01 3.81732792e-01 2.80222207e-01 -5.88540435e-01
3.16378802e-01 5.20060360e-01 -3.41215581e-01 -4.69962388e-01
2.18527704e-01 -3.70150089e-01 -5.17450161e-02 4.79581416e-01
1.62121326e-01 -2.74860431e-02 2.56897688e-01 8.39287341e-02
8.33638251e-01 -1.37077138e-01 -2.35860795e-01 -1.30784526e-01
2.19281361e-01 -4.70871180e-01 9.27758873e-01 1.01656771e+00
-5.06036043e-01 9.05635655e-01 2.38792390e-01 -3.53450447e-01
-8.00425589e-01 -7.61791289e-01 -5.00892103e-01 1.26638865e+00
1.12787142e-01 -2.04876468e-01 -7.15773165e-01 -1.24971569e+00
-2.73966268e-02 6.73837125e-01 -5.13588428e-01 -2.24720925e-01
-2.72084028e-01 -9.49827492e-01 6.02826834e-01 3.27487975e-01
5.06479919e-01 -9.72193778e-01 -3.10084820e-02 3.35586965e-01
-5.47140658e-01 -1.02835727e+00 -4.13535058e-01 6.57917857e-01
-5.95589161e-01 -1.09511399e+00 -5.34745455e-01 -8.89789999e-01
7.64279962e-01 2.61388212e-01 1.15495348e+00 3.19751531e-01
6.47940487e-02 -6.03670813e-03 -7.24977374e-01 -4.17158812e-01
-4.33417439e-01 2.40838062e-02 5.59759252e-02 1.64683968e-01
2.22910255e-01 -2.29126140e-01 -3.99254888e-01 5.50889671e-01
-1.00810409e+00 -4.02148277e-01 -2.67340578e-02 1.28604066e+00
9.23368633e-01 2.90994108e-01 9.25372720e-01 -1.20843995e+00
4.23659325e-01 -6.06210589e-01 -4.33278352e-01 2.55229563e-01
-6.32243633e-01 7.39671439e-02 6.67470098e-01 -6.39395833e-01
-1.02541721e+00 -8.53157640e-02 -2.29689747e-01 -4.63997692e-01
-1.01216592e-01 3.30598682e-01 -2.63476610e-01 1.30757973e-01
8.30228627e-01 -3.64894480e-01 -3.33098978e-01 -7.66824067e-01
3.12587559e-01 7.62545288e-01 -2.39710100e-02 -6.69023693e-01
4.65840191e-01 5.10515749e-01 -1.12018503e-01 -2.87791818e-01
-1.47408390e+00 -5.07451117e-01 -9.61491391e-02 -2.58690208e-01
5.02500057e-01 -1.11977696e+00 -3.28021079e-01 9.28004920e-01
-9.99748468e-01 -2.98407853e-01 -3.77633601e-01 4.64206308e-01
-1.82224348e-01 2.19059855e-01 -7.66440392e-01 -7.04183102e-01
-2.11701185e-01 -1.48161054e+00 1.18484521e+00 1.11588418e-01
3.42467837e-02 -8.35926533e-01 -2.85097718e-01 4.69009936e-01
2.83320487e-01 -4.78181690e-02 7.20140994e-01 -8.06762815e-01
-3.20601285e-01 -2.06560969e-01 -2.31499389e-01 9.24402595e-01
-2.75912620e-02 -2.70892143e-01 -1.31218791e+00 -4.18464005e-01
1.23003654e-01 -1.01867402e+00 1.33443737e+00 3.51993412e-01
1.43019676e+00 -7.79579952e-02 -7.36881420e-02 4.46030825e-01
1.39580154e+00 -2.16073588e-01 3.18682402e-01 2.56929964e-01
9.20421243e-01 5.61196744e-01 8.03766251e-01 2.53103405e-01
3.02427739e-01 6.07640564e-01 5.13197362e-01 -1.18242381e-02
-3.92786205e-01 -1.33431554e-01 1.88973114e-01 7.50486910e-01
5.72801292e-01 -4.65954512e-01 -7.04343498e-01 4.79478478e-01
-1.83441544e+00 -6.80544794e-01 -3.42649162e-01 2.10125971e+00
1.20414782e+00 1.37423038e-01 -3.55879992e-01 3.10790300e-01
8.84775758e-01 2.26871222e-01 -5.68728924e-01 6.84372783e-02
-2.98121274e-01 -5.90256322e-03 4.95743066e-01 4.44507271e-01
-1.42660129e+00 8.76694679e-01 4.98545694e+00 1.44673777e+00
-1.01065397e+00 6.99427128e-01 1.27081466e+00 -4.92366731e-01
-3.01311553e-01 -8.29434246e-02 -9.48360503e-01 7.99837053e-01
7.84864902e-01 5.05443871e-01 2.70657271e-01 8.25868964e-01
2.77361989e-01 -3.62897485e-01 -1.01053548e+00 1.07798541e+00
-1.15860775e-01 -9.22089934e-01 -2.68414289e-01 -1.61916047e-01
1.11100900e+00 3.77318323e-01 1.26248434e-01 4.34149355e-01
3.98801595e-01 -8.78923357e-01 1.03760052e+00 4.20211643e-01
5.24065793e-01 -6.63552105e-01 1.19382441e+00 3.90705615e-01
-8.54076087e-01 -2.56734997e-01 -6.28134847e-01 1.50905684e-01
1.48274332e-01 1.60006762e+00 -2.46790811e-01 2.56834716e-01
9.55880523e-01 6.06504440e-01 -7.28131354e-01 1.04451394e+00
-4.51155603e-01 9.36686337e-01 -4.95453626e-01 1.80057749e-01
2.11265802e-01 -5.33681773e-02 3.02271307e-01 8.85464072e-01
3.06785703e-01 -2.36237049e-01 4.30751890e-01 6.28192186e-01
-6.20984375e-01 1.21460810e-01 -3.51116747e-01 5.38194954e-01
6.49585903e-01 1.15844131e+00 -5.40653169e-01 -3.72737020e-01
-2.11605296e-01 5.91686487e-01 3.75962257e-01 4.58979219e-01
-7.55928397e-01 1.93583295e-02 4.86577213e-01 -4.66019586e-02
1.07141651e-01 2.95893937e-01 -4.77458477e-01 -1.16377950e+00
1.55984715e-01 -9.49195087e-01 2.10127041e-01 -6.47984564e-01
-1.86884427e+00 6.75040245e-01 -3.50709617e-01 -1.31150019e+00
3.02625656e-01 -2.08405986e-01 -2.02637449e-01 6.51692510e-01
-1.75772977e+00 -9.87579346e-01 -2.57881999e-01 1.69633806e-01
6.88486993e-01 1.96305275e-01 5.98627448e-01 6.82550907e-01
-5.57460725e-01 8.12585771e-01 5.24341702e-01 6.21784367e-02
1.18490458e+00 -1.14520001e+00 -9.61429253e-03 6.47802830e-01
1.76494166e-01 1.29301101e-01 4.99555618e-01 -6.39704525e-01
-3.26983482e-01 -1.61079347e+00 7.70940244e-01 -3.56747955e-01
4.32559550e-01 -4.14194435e-01 -1.17191029e+00 2.76095569e-01
-2.17445567e-01 6.37735784e-01 5.47140419e-01 -2.60833281e-06
-6.01145267e-01 -4.36966658e-01 -1.38059950e+00 3.37404877e-01
1.01190746e+00 -4.77868944e-01 8.45619477e-03 5.77923417e-01
7.73740530e-01 -2.60802001e-01 -7.93086410e-01 5.59748173e-01
1.45900294e-01 -8.15790117e-01 6.97835386e-01 -2.24321932e-01
2.25937262e-01 -3.92473370e-01 -3.85619700e-01 -1.70938110e+00
-4.92685065e-02 -8.59730914e-02 1.39014766e-01 1.49414182e+00
4.85784501e-01 -5.02910674e-01 5.91256440e-01 3.52520823e-01
-6.62718043e-02 -9.23053145e-01 -1.20283341e+00 -8.57149959e-01
1.67851135e-01 -7.30919719e-01 4.18465406e-01 1.00332391e+00
-5.71904123e-01 1.72304362e-01 -5.05180836e-01 -1.79848626e-01
8.70764494e-01 -3.79566938e-01 2.34233916e-01 -1.30148530e+00
-7.42378607e-02 -1.23724736e-01 -1.51330620e-01 -7.64915526e-01
3.59913588e-01 -8.95394027e-01 3.97701412e-01 -1.23256528e+00
1.10413790e-01 -9.49135065e-01 -6.26436830e-01 7.77212977e-01
-5.42149186e-01 6.90661132e-01 4.13054079e-01 3.27547878e-01
-9.29877043e-01 8.88462722e-01 1.26127970e+00 -2.93098629e-01
2.15221241e-01 9.65839475e-02 -5.79310238e-01 7.67430127e-01
8.58570158e-01 -1.20914149e+00 -3.89666170e-01 -4.70055491e-01
3.46362919e-01 -5.52603126e-01 3.55180264e-01 -1.00527084e+00
-1.09916866e-01 1.85717404e-01 1.44964188e-01 -2.56188154e-01
2.37787589e-01 -8.89369667e-01 -3.27271223e-02 1.58292934e-01
-6.73673391e-01 -3.70901108e-01 -2.44658515e-01 7.63440371e-01
-2.25024253e-01 -7.28199601e-01 1.17509365e+00 -2.50485331e-01
-4.00579780e-01 1.57342836e-01 -1.16242640e-01 4.95106488e-01
8.18450689e-01 2.00883195e-01 -5.65706551e-01 -3.26739728e-01
-8.53453338e-01 4.58032161e-01 4.70131665e-01 3.59702945e-01
3.33111942e-01 -1.35884440e+00 -7.91130900e-01 -9.88711119e-02
3.91274154e-01 2.82967299e-01 4.18104261e-01 7.81406999e-01
-6.76132888e-02 -8.53274539e-02 5.06421149e-01 -6.07644498e-01
-1.03402197e+00 2.84897357e-01 5.02278388e-01 -4.95172650e-01
-2.74718732e-01 1.08033621e+00 7.23230317e-02 -7.17596054e-01
6.33785307e-01 -2.86266714e-01 1.12273954e-01 4.03948933e-01
4.31028813e-01 4.27244514e-01 5.57226360e-01 -5.88255048e-01
-1.57466561e-01 4.79268581e-01 -9.03746635e-02 1.35663122e-01
1.16630495e+00 -3.65455210e-01 -1.03691611e-02 6.51696086e-01
1.58914447e+00 -3.37603688e-01 -1.36791503e+00 -6.10643804e-01
4.12500687e-02 -4.38227713e-01 3.99827391e-01 -1.04419637e+00
-1.52248693e+00 7.58831382e-01 1.10213137e+00 -6.01626262e-02
1.00862420e+00 1.12725608e-01 8.56557071e-01 3.53932440e-01
4.61409986e-01 -1.38682699e+00 2.73879021e-01 4.83126998e-01
6.36570811e-01 -1.71086192e+00 -1.69567034e-01 -3.31187099e-01
-6.98375881e-01 5.55929303e-01 5.55960238e-01 1.04763187e-01
8.73246729e-01 1.28812850e-01 5.19149542e-01 -8.22692215e-02
-6.56450510e-01 -3.28034580e-01 2.62076408e-03 4.72884595e-01
2.77557045e-01 1.02718167e-01 -4.95120734e-01 7.43710279e-01
2.60388762e-01 -2.63132015e-03 5.20871878e-01 8.22586238e-01
-4.29146022e-01 -1.28518033e+00 -5.96458972e-01 6.68733001e-01
-6.14780784e-01 -2.20167935e-01 3.11686397e-02 1.98809832e-01
7.05695868e-01 1.30515015e+00 -2.20587462e-01 -2.92638928e-01
2.81218350e-01 3.47784668e-01 1.33431509e-01 -7.15571225e-01
-6.09394908e-01 1.80593520e-01 1.17998742e-01 -4.77394432e-01
-5.50807655e-01 -4.41035569e-01 -1.26977956e+00 -9.66017172e-02
-7.39046454e-01 1.59764200e-01 7.54607320e-01 9.18520093e-01
3.04974020e-01 4.95750308e-01 7.21063972e-01 -7.16936648e-01
-9.05589581e-01 -1.21143675e+00 -7.65285194e-01 8.52966130e-01
3.21932346e-01 -9.12700295e-01 -7.14022696e-01 -4.53106053e-02] | [9.411529541015625, 3.8753931522369385] |
74125783-67c1-4b8c-84e8-ea9b069a01de | understanding-image-retrieval-re-ranking-a | 2012.07620 | null | https://arxiv.org/abs/2012.07620v2 | https://arxiv.org/pdf/2012.07620v2.pdf | Understanding Image Retrieval Re-Ranking: A Graph Neural Network Perspective | The re-ranking approach leverages high-confidence retrieved samples to refine retrieval results, which have been widely adopted as a post-processing tool for image retrieval tasks. However, we notice one main flaw of re-ranking, i.e., high computational complexity, which leads to an unaffordable time cost for real-world applications. In this paper, we revisit re-ranking and demonstrate that re-ranking can be reformulated as a high-parallelism Graph Neural Network (GNN) function. In particular, we divide the conventional re-ranking process into two phases, i.e., retrieving high-quality gallery samples and updating features. We argue that the first phase equals building the k-nearest neighbor graph, while the second phase can be viewed as spreading the message within the graph. In practice, GNN only needs to concern vertices with the connected edges. Since the graph is sparse, we can efficiently update the vertex features. On the Market-1501 dataset, we accelerate the re-ranking processing from 89.2s to 9.4ms with one K40m GPU, facilitating the real-time post-processing. Similarly, we observe that our method achieves comparable or even better retrieval results on the other four image retrieval benchmarks, i.e., VeRi-776, Oxford-5k, Paris-6k and University-1652, with limited time cost. Our code is publicly available. | ['Yi Yang', 'Errui Ding', 'Xiao Tan', 'Zhedong Zheng', 'Minyue Jiang', 'Xuanmeng Zhang'] | 2020-12-14 | null | null | null | null | ['drone-view-target-localization'] | ['computer-vision'] | [ 2.68426351e-03 -5.59140325e-01 -2.54265994e-01 -4.39099595e-02
-1.19742656e+00 -5.48930049e-01 6.91304982e-01 5.18456936e-01
-4.86082852e-01 3.94166172e-01 1.68238714e-01 -4.00062054e-01
-2.81217039e-01 -1.03800547e+00 -6.22897804e-01 -6.61464155e-01
-1.98862091e-01 2.87163466e-01 2.12102890e-01 -1.80870205e-01
4.40658152e-01 5.40293992e-01 -1.63206017e+00 2.39481732e-01
6.24821186e-01 9.80064273e-01 6.80814087e-02 7.52127886e-01
1.93491966e-01 4.43860590e-01 -5.97278953e-01 -3.87137085e-01
3.48355919e-01 3.94856520e-02 -7.92154133e-01 -2.45232955e-01
4.10523653e-01 -5.80601037e-01 -7.75165260e-01 1.19158101e+00
5.52696943e-01 1.62003726e-01 6.15769625e-01 -1.23467779e+00
-7.52445400e-01 3.00130427e-01 -8.16044092e-01 -4.18849625e-02
3.26925814e-01 -3.67691338e-01 1.24318266e+00 -9.48997974e-01
6.73081398e-01 1.02903485e+00 2.45521054e-01 1.37967899e-01
-7.70509958e-01 -7.98661709e-01 -1.54113978e-01 4.15980637e-01
-1.68255019e+00 -2.40648016e-01 5.03026128e-01 -1.73984572e-01
6.32455885e-01 5.36196351e-01 7.21854925e-01 4.26286846e-01
2.06668347e-01 8.98923695e-01 8.71772289e-01 -2.67508209e-01
3.56143676e-02 -4.08813000e-01 5.23658752e-01 6.59435451e-01
5.03267169e-01 3.94662432e-02 -5.45228899e-01 -4.67472672e-01
6.18694961e-01 3.53938788e-01 -2.94654131e-01 -2.27045119e-01
-1.25440431e+00 9.04395044e-01 8.30775917e-01 3.85270976e-02
-2.13725254e-01 4.51542437e-01 6.22727394e-01 5.97825408e-01
6.18373573e-01 6.15661740e-02 -1.34295508e-01 2.42734566e-01
-8.69256914e-01 -6.86011557e-03 5.65905094e-01 8.20936322e-01
1.04487300e+00 -4.23991472e-01 -3.25424612e-01 1.00898731e+00
4.48440492e-01 7.38250911e-01 4.09069926e-01 -5.17421603e-01
2.72184432e-01 4.57446814e-01 -8.88950229e-02 -1.56588256e+00
-2.26701513e-01 -4.91408169e-01 -1.31604695e+00 -3.00743550e-01
8.66981670e-02 3.35784554e-01 -1.06475472e+00 1.30410337e+00
2.14440495e-01 4.34543014e-01 -1.00933053e-01 1.10619950e+00
1.15101755e+00 8.43494415e-01 -2.78778344e-01 -5.67418262e-02
1.42165732e+00 -1.30836904e+00 -2.08765134e-01 1.05438709e-01
6.47331953e-01 -8.93820941e-01 8.12768042e-01 3.10151845e-01
-7.34299660e-01 -4.35416847e-01 -1.11477065e+00 -1.15867563e-01
-4.25887078e-01 9.30304304e-02 9.49765980e-01 3.29065919e-01
-1.44002569e+00 7.05274582e-01 -3.73056471e-01 -2.37151816e-01
2.08920136e-01 4.64634985e-01 -5.18788040e-01 -4.36497122e-01
-1.19918942e+00 2.76436865e-01 2.98375010e-01 2.58488029e-01
-7.68605351e-01 -3.95454437e-01 -5.66346467e-01 1.51768699e-01
3.38510454e-01 -5.87304413e-01 7.87345767e-01 -4.81227398e-01
-1.25717950e+00 9.48153734e-01 -1.60274744e-01 -1.52093038e-01
2.47523397e-01 -2.74140805e-01 -4.71156895e-01 1.99985355e-01
3.24835442e-02 4.83311445e-01 8.66905510e-01 -1.18084610e+00
-5.66684961e-01 -3.61721635e-01 4.91090529e-02 2.60631561e-01
-4.70872104e-01 -1.80448115e-01 -1.33948565e+00 -5.35721898e-01
3.43005627e-01 -1.15235615e+00 -2.70422429e-01 -2.46189073e-01
-4.62750763e-01 -2.55603909e-01 3.87493163e-01 -3.53864908e-01
1.38344753e+00 -2.22923946e+00 -5.73913716e-02 8.95146370e-01
7.15628088e-01 4.05590624e-01 -5.65951288e-01 6.06983483e-01
-1.04860492e-01 7.19341114e-02 7.80844837e-02 -2.32668713e-01
-1.11839734e-01 2.71729492e-02 -4.38434035e-01 7.46748984e-01
-1.85821295e-01 1.24983025e+00 -1.12469995e+00 -3.95368248e-01
5.45745380e-02 5.84579647e-01 -4.15625304e-01 -1.13357417e-01
3.26195300e-01 -2.20221840e-02 -5.69666803e-01 5.95726550e-01
9.28850830e-01 -8.23975205e-01 8.24170783e-02 -2.60896116e-01
1.72598869e-01 5.08437306e-02 -1.21775651e+00 1.66170192e+00
-2.79827505e-01 6.33780479e-01 -1.75513431e-01 -8.94219637e-01
6.46948636e-01 -2.69972205e-01 3.66999269e-01 -1.03234577e+00
5.50935641e-02 4.39523429e-01 -5.17248750e-01 1.68216303e-01
1.15535212e+00 5.86900413e-01 -6.82212263e-02 6.34247243e-01
-3.38711500e-01 1.54253662e-01 2.19701082e-01 6.05212212e-01
1.19171679e+00 -5.16453087e-01 1.81070313e-01 -2.34072328e-01
5.08955240e-01 -3.90457027e-02 1.97439641e-02 9.83079612e-01
2.52445221e-01 6.37008429e-01 3.17671120e-01 -3.62829596e-01
-7.01562166e-01 -8.56804788e-01 1.22046545e-01 1.02518916e+00
7.62348652e-01 -9.71759319e-01 -3.76002878e-01 -5.11009216e-01
3.70639078e-02 -6.34762645e-02 -5.01519203e-01 -3.53989691e-01
-5.13584673e-01 -8.34271729e-01 3.91733527e-01 7.80251920e-02
4.80857313e-01 -9.46855962e-01 -2.25874409e-01 -6.62387609e-02
-3.31677347e-02 -6.80340171e-01 -7.57206857e-01 -1.91719100e-01
-8.18093777e-01 -1.23966670e+00 -8.59926105e-01 -8.64118993e-01
8.68332088e-01 1.11333525e+00 1.39461088e+00 9.73389089e-01
-2.61474878e-01 4.33005959e-01 -5.34126222e-01 1.53615773e-01
9.29020643e-02 1.70392796e-01 -2.97587812e-01 -1.04812682e-01
2.42343217e-01 -2.55756915e-01 -1.01231778e+00 2.75598794e-01
-1.12481594e+00 5.22612408e-03 7.32397139e-01 1.08148110e+00
9.77149904e-01 4.82172780e-02 3.18548471e-01 -8.92115831e-01
8.05992126e-01 -4.39128488e-01 -8.00070941e-01 5.30347168e-01
-8.31084669e-01 -6.37909323e-02 3.66481662e-01 -2.51651198e-01
-2.06359908e-01 -1.39093116e-01 3.77928242e-02 -3.17865193e-01
5.36406100e-01 9.13567901e-01 3.02834362e-01 -2.20402241e-01
3.52167934e-01 2.79334158e-01 -1.44653291e-01 -1.64981648e-01
4.81791079e-01 7.53012240e-01 3.43886703e-01 -4.61135387e-01
1.13082492e+00 5.33467829e-01 3.24036211e-01 -7.33033419e-01
-6.40769601e-01 -8.77520442e-01 -8.78238454e-02 -1.38475716e-01
3.09131682e-01 -1.09280825e+00 -9.69255686e-01 6.24825358e-01
-1.01327479e+00 -2.53723562e-01 3.03437710e-02 3.30673456e-01
-2.52351277e-02 8.54525506e-01 -8.99685502e-01 -2.90267915e-01
-8.56937528e-01 -1.19264317e+00 1.43566298e+00 1.35080531e-01
3.18144858e-01 -5.64315975e-01 3.61061171e-02 1.38996780e-01
4.32277679e-01 -1.55750111e-01 8.36722553e-01 -5.72195411e-01
-9.79578435e-01 -4.48535770e-01 -9.04452443e-01 -4.14882833e-03
-1.14244387e-01 -1.08529381e-01 -7.81781137e-01 -7.37296641e-01
-6.62779212e-01 -3.59696835e-01 1.17007017e+00 1.93775356e-01
1.34735191e+00 -1.46664251e-02 -3.77370596e-01 6.40370250e-01
1.57223380e+00 -1.64912388e-01 9.32766974e-01 3.64558578e-01
6.59952819e-01 2.41130814e-01 8.71501029e-01 3.49541545e-01
3.89543533e-01 5.37579894e-01 3.17143172e-01 -2.49803290e-01
-5.54792173e-02 -1.96387157e-01 1.49277985e-01 1.37696993e+00
-1.17435148e-02 -4.34294403e-01 -7.73085177e-01 4.40075576e-01
-2.05507636e+00 -5.02914608e-01 -3.53900880e-01 2.55251050e+00
6.32658124e-01 -2.42206514e-01 -2.10660547e-01 4.40762602e-02
7.72951365e-01 3.29254061e-01 -3.45798135e-01 2.05207080e-01
1.22993276e-03 3.45305473e-01 7.15932250e-01 6.78226292e-01
-9.27502215e-01 1.00171959e+00 5.60695601e+00 1.34662879e+00
-1.12876320e+00 -1.15738228e-01 7.46447146e-01 1.52977044e-03
-4.09547240e-01 -4.83302735e-02 -5.94423354e-01 3.95111322e-01
6.16238475e-01 -2.39348382e-01 6.41552866e-01 6.76481605e-01
-4.45035517e-01 -6.65090308e-02 -9.20298338e-01 1.49831963e+00
2.08554208e-01 -1.42996490e+00 3.21621597e-01 1.56525567e-01
5.90536952e-01 3.35975766e-01 2.93524209e-02 4.04551327e-01
2.38623947e-01 -1.04234552e+00 2.69284964e-01 3.79099041e-01
1.06813836e+00 -8.54221642e-01 8.12099338e-01 -6.65414482e-02
-1.40509820e+00 4.11102861e-01 -6.76857948e-01 4.01888043e-01
-6.90664873e-02 1.08736551e+00 -3.67217571e-01 8.78542006e-01
8.37308824e-01 8.54164660e-01 -7.29510367e-01 1.19554067e+00
-8.35489705e-02 3.06102514e-01 -3.97007585e-01 -1.54687241e-01
1.69865206e-01 -3.82774353e-01 2.93553442e-01 1.07164466e+00
3.51638436e-01 2.45944746e-02 1.97255149e-01 5.30282035e-02
-5.00126183e-01 2.55926996e-01 -4.26746637e-01 1.09203406e-01
3.34007382e-01 1.53422022e+00 -8.35527003e-01 -5.94077706e-01
-2.62189031e-01 1.20721030e+00 4.15803462e-01 4.70666975e-01
-8.18974853e-01 -5.54997206e-01 3.78053874e-01 -1.72153324e-01
8.69975332e-03 -1.70085534e-01 2.33324260e-01 -1.33114815e+00
1.13309078e-01 -9.21974957e-01 5.28199375e-01 -6.93731725e-01
-1.34876454e+00 7.40686536e-01 -2.35798225e-01 -1.23760009e+00
-4.76389527e-02 -5.96454501e-01 -8.98005515e-02 7.27812707e-01
-1.97390425e+00 -9.79984581e-01 -6.88884318e-01 6.94934189e-01
-4.48009185e-02 -5.76626658e-02 7.15014219e-01 7.63715684e-01
-3.68863940e-01 8.00856173e-01 3.83699179e-01 1.82596400e-01
9.30502355e-01 -9.46132123e-01 4.34670657e-01 5.17660677e-01
5.04528940e-01 8.47133398e-01 2.74325870e-02 -6.10565603e-01
-1.83141589e+00 -1.09149873e+00 8.40043902e-01 -1.67684779e-02
7.60810614e-01 -2.80262589e-01 -7.95954823e-01 1.50264591e-01
-2.65621319e-02 3.69605094e-01 4.62697268e-01 1.63354024e-01
-5.24201393e-01 -2.69182533e-01 -6.21781468e-01 7.88624883e-01
1.21121252e+00 -8.08640599e-01 5.37590943e-02 5.89692771e-01
6.38855815e-01 -5.17902553e-01 -7.90220797e-01 3.79317611e-01
6.68255806e-01 -6.42335773e-01 1.21047759e+00 -3.63717496e-01
3.36557806e-01 -4.23580974e-01 -3.42053443e-01 -1.18065274e+00
-4.77201313e-01 -6.14194036e-01 -3.95050161e-02 8.95208418e-01
2.82804519e-01 -8.22053373e-01 8.76723170e-01 1.58999488e-01
2.57908344e-01 -8.88433039e-01 -7.20353365e-01 -5.27228892e-01
-2.86503762e-01 -3.69238198e-01 6.47573590e-01 7.91869640e-01
-3.90357167e-01 3.77039731e-01 -4.58027661e-01 1.52552962e-01
5.46428025e-01 5.41452587e-01 9.79277194e-01 -1.30870342e+00
-3.44562709e-01 -3.25177938e-01 -3.84722352e-01 -1.53059077e+00
9.95034650e-02 -1.12657082e+00 -5.34543954e-02 -1.60923207e+00
5.73022664e-01 -7.93571115e-01 -7.50026584e-01 3.94079268e-01
-3.42985600e-01 6.85392201e-01 5.82554750e-02 5.96259296e-01
-1.02693343e+00 4.32541966e-01 1.23304605e+00 -3.38804394e-01
-1.05913710e-02 -1.96172372e-01 -6.74220741e-01 2.64839143e-01
6.98355973e-01 -5.53750277e-01 -3.98660600e-01 -6.20118678e-01
8.85771155e-01 5.26978634e-02 5.27483702e-01 -5.73664129e-01
6.25638604e-01 2.28770107e-01 1.54270768e-01 -9.28808153e-01
2.96851635e-01 -8.01712096e-01 1.23083912e-01 4.82476801e-01
-4.16170731e-02 3.72168452e-01 8.08384120e-02 7.58893251e-01
-4.96965826e-01 -1.20521083e-01 3.20685625e-01 1.34165317e-01
-7.29430377e-01 6.69759572e-01 -5.17250076e-02 -1.05664104e-01
6.38253093e-01 -5.50509840e-02 -7.31977165e-01 -5.86959600e-01
-1.45506963e-01 3.27488601e-01 5.01790106e-01 3.63458365e-01
8.87132227e-01 -1.51147187e+00 -8.64070237e-01 1.01562224e-01
4.13770765e-01 -4.29575704e-02 3.70545864e-01 5.98101795e-01
-7.43380904e-01 3.35229814e-01 2.76443928e-01 -7.25692689e-01
-1.42741156e+00 5.32056093e-01 -1.51199654e-01 -7.04674363e-01
-6.50077045e-01 6.74408495e-01 1.05623707e-01 -1.55730963e-01
2.54641205e-01 1.16623798e-02 -1.61292002e-01 2.09938869e-01
5.93559027e-01 3.85961771e-01 2.36607179e-01 -4.80256975e-01
-4.85364914e-01 7.89864957e-01 -3.16803813e-01 1.05317384e-01
1.40479505e+00 9.85607877e-02 -6.90915167e-01 -7.56803751e-02
1.63221431e+00 2.74438322e-01 -3.80062371e-01 -2.55697012e-01
-1.06966324e-01 -5.49470305e-01 1.61640286e-01 -3.41926038e-01
-1.49224806e+00 4.44190681e-01 5.34104288e-01 5.40663041e-02
1.27817178e+00 -8.21234882e-02 9.59702909e-01 8.82993042e-01
7.11416483e-01 -9.84424949e-01 7.97372684e-02 4.44114864e-01
8.01608026e-01 -1.22474015e+00 5.07190287e-01 -5.18430173e-01
-1.39513597e-01 8.62767994e-01 1.95657477e-01 -4.07121032e-01
8.05228829e-01 -1.96700916e-01 -1.74559742e-01 -5.38507283e-01
-5.79120517e-01 4.36790213e-02 7.23059893e-01 2.98863463e-02
1.48770019e-01 2.01335132e-01 -3.86616647e-01 1.56994596e-01
-4.12900820e-02 -1.80964574e-01 2.06631869e-01 6.66758716e-01
-9.53888223e-02 -1.15113485e+00 -1.57990843e-01 8.84742260e-01
-3.04018706e-01 -5.75743556e-01 -2.75799930e-01 7.46703923e-01
-4.40371543e-01 1.04705346e+00 -7.81543180e-02 -6.54182851e-01
1.15182951e-01 -5.74105024e-01 2.09705725e-01 -4.95231926e-01
-5.41844249e-01 7.15357661e-02 -7.16640949e-02 -8.30664635e-01
-2.80403614e-01 -9.35962126e-02 -9.69068944e-01 -6.05940342e-01
-6.76203132e-01 4.14384604e-01 5.34159720e-01 3.64918441e-01
7.25795448e-01 3.59306633e-01 6.31332695e-01 -5.98426104e-01
-3.40195656e-01 -5.74204147e-01 -5.36277056e-01 3.48125726e-01
2.24688202e-01 -3.72679144e-01 -4.59067971e-01 -6.39178395e-01] | [10.822872161865234, 0.6888476014137268] |
85015ecd-0100-41ab-b550-b5acf0635b67 | revitalizing-optimization-for-3d-human-pose | 2105.13965 | null | https://arxiv.org/abs/2105.13965v2 | https://arxiv.org/pdf/2105.13965v2.pdf | Revitalizing Optimization for 3D Human Pose and Shape Estimation: A Sparse Constrained Formulation | We propose a novel sparse constrained formulation and from it derive a real-time optimization method for 3D human pose and shape estimation. Our optimization method, SCOPE (Sparse Constrained Optimization for 3D human Pose and shapE estimation), is orders of magnitude faster (avg. 4 ms convergence) than existing optimization methods, while being mathematically equivalent to their dense unconstrained formulation under mild assumptions. We achieve this by exploiting the underlying sparsity and constraints of our formulation to efficiently compute the Gauss-Newton direction. We show that this computation scales linearly with the number of joints and measurements of a complex 3D human model, in contrast to prior work where it scales cubically due to their dense unconstrained formulation. Based on our optimization method, we present a real-time motion capture framework that estimates 3D human poses and shapes from a single image at over 30 FPS. In benchmarks against state-of-the-art methods on multiple public datasets, our framework outperforms other optimization methods and achieves competitive accuracy against regression methods. Project page with code and videos: https://sites.google.com/view/scope-human/. | ['Mustafa Mukadam', 'Todd Murphey', 'Weipeng Xu', 'Donglai Xiang', 'Kalyan Vasudev Alwala', 'Taosha Fan'] | 2021-05-28 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Fan_Revitalizing_Optimization_for_3D_Human_Pose_and_Shape_Estimation_A_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Fan_Revitalizing_Optimization_for_3D_Human_Pose_and_Shape_Estimation_A_ICCV_2021_paper.pdf | iccv-2021-1 | ['3d-human-pose-and-shape-estimation'] | ['computer-vision'] | [-2.84199025e-02 -1.74986228e-01 -5.70716187e-02 -7.42980465e-02
-7.33070791e-01 -4.64566439e-01 2.80542135e-01 -3.09090286e-01
-6.37947679e-01 4.10330355e-01 4.72935140e-01 2.38375276e-01
3.18675071e-01 -2.80848872e-02 -8.65857601e-01 -1.85417563e-01
-2.25763053e-01 1.02290165e+00 1.00579485e-01 -2.11036637e-01
5.93815856e-02 4.61485863e-01 -1.20522988e+00 -2.98754454e-01
1.75297901e-01 7.98828244e-01 -1.24667406e-01 1.06781387e+00
8.16303372e-01 4.10395056e-01 -2.36475557e-01 -2.03163356e-01
5.79767764e-01 -4.40157764e-02 -6.67796195e-01 5.01555622e-01
7.66062856e-01 -6.74734116e-01 -5.06821692e-01 5.90540588e-01
9.50921834e-01 3.39156598e-01 3.14199626e-01 -1.24632490e+00
-1.60444766e-01 -1.96786642e-01 -5.20077765e-01 7.00755939e-02
1.12995875e+00 3.30965757e-01 8.39858472e-01 -1.22494507e+00
7.62973487e-01 1.43043149e+00 1.07210350e+00 5.12913048e-01
-1.10310543e+00 -1.23079896e-01 3.76648307e-02 -1.67376712e-01
-1.45667815e+00 -6.78475678e-01 3.91237885e-01 -5.40977836e-01
1.28345406e+00 2.36636415e-01 1.10507357e+00 1.17061639e+00
2.48060152e-01 1.04063690e+00 6.15191460e-01 -3.54115725e-01
-1.35454327e-01 -4.35092598e-01 -1.00487471e-01 1.13992834e+00
3.22448701e-01 9.37357768e-02 -6.79040909e-01 -3.68933976e-01
1.15739501e+00 2.54304707e-02 -1.49140447e-01 -7.54041851e-01
-1.77460945e+00 6.04699016e-01 4.93401065e-02 -4.10267889e-01
-3.78981173e-01 5.71780026e-01 2.50591487e-01 -1.54026434e-01
3.43180597e-01 1.43083438e-01 -5.84488451e-01 -5.15058100e-01
-7.68902659e-01 9.08565402e-01 1.10648417e+00 1.17886412e+00
3.71833742e-01 2.03856714e-02 1.11282878e-01 6.18641198e-01
3.94820482e-01 8.45743179e-01 2.12220401e-01 -1.35701561e+00
3.35094750e-01 3.84166151e-01 4.56837356e-01 -1.27107751e+00
-7.55833089e-01 -3.54517549e-01 -5.55589080e-01 3.79392505e-02
6.41331553e-01 -2.48218462e-01 -5.70132315e-01 1.48211038e+00
8.31030011e-01 3.43238749e-02 -5.64427614e-01 1.39273047e+00
4.91765022e-01 3.04974228e-01 -3.69847625e-01 1.30493850e-01
1.35255826e+00 -1.21951258e+00 -4.73868817e-01 -5.54360271e-01
4.74674642e-01 -8.77175212e-01 8.52909148e-01 4.82472628e-01
-1.39631271e+00 -4.32503432e-01 -9.08030868e-01 -3.30816031e-01
2.54264325e-01 2.86188930e-01 8.09526324e-01 5.14321327e-01
-8.96595836e-01 6.76878273e-01 -1.29801083e+00 -5.83968997e-01
-8.08514131e-04 5.19779623e-01 -6.06908262e-01 1.18579186e-01
-7.10482478e-01 1.02038264e+00 -1.14419408e-01 1.94637343e-01
-8.72630239e-01 -5.52552819e-01 -1.10612643e+00 -4.36684549e-01
6.96799159e-01 -1.16437614e+00 1.41755319e+00 -9.82370898e-02
-1.61855483e+00 1.04970670e+00 -3.15580457e-01 -3.83637488e-01
7.25697041e-01 -1.06952608e+00 2.63754070e-01 3.30538958e-01
1.36925340e-01 6.56262338e-01 8.74162018e-01 -7.30183482e-01
-8.49745274e-02 -3.36680144e-01 -8.37535188e-02 3.70104849e-01
2.37466730e-02 6.56499341e-02 -1.13618517e+00 -7.51143157e-01
1.02367125e-01 -1.46875572e+00 -6.12736166e-01 3.46721143e-01
-5.49305141e-01 1.23003826e-01 3.81232828e-01 -8.84988248e-01
1.13472319e+00 -1.60500503e+00 7.20520973e-01 1.64733618e-01
4.75963801e-01 -1.17650047e-01 2.51193762e-01 2.91903198e-01
2.59720623e-01 -4.08319205e-01 -2.44681090e-01 -8.17947388e-01
3.70830029e-01 1.30150229e-01 3.33689153e-01 1.07423389e+00
-9.49457809e-02 1.01148820e+00 -7.48867989e-01 -7.24807322e-01
4.07299668e-01 6.12697661e-01 -9.73587632e-01 3.08583796e-01
1.13872208e-01 5.21391034e-01 -2.77334541e-01 1.05288470e+00
3.59720081e-01 -5.92084587e-01 3.08603585e-01 -3.73983681e-01
1.39595836e-01 7.38324458e-03 -1.54060400e+00 2.19556785e+00
-9.68853664e-03 3.72872859e-01 2.31229812e-01 -8.76446784e-01
6.80822492e-01 3.14379275e-01 9.24032331e-01 4.52757394e-03
3.75245988e-01 1.43201962e-01 -4.93692935e-01 -2.84938395e-01
4.92011935e-01 2.36480936e-01 -2.59057730e-01 3.79034072e-01
7.47737885e-02 -1.87318251e-01 1.23250663e-01 2.50253469e-01
1.10173988e+00 7.27531970e-01 4.96621132e-01 -2.37066522e-01
2.68088132e-01 4.50162180e-02 5.62849998e-01 4.16490257e-01
-4.77009594e-01 6.70180142e-01 2.92262405e-01 -7.16704369e-01
-1.24691021e+00 -1.19762480e+00 1.91176787e-01 9.55559075e-01
1.22722751e-02 -8.10264528e-01 -7.74921477e-01 -2.24564746e-01
4.69063073e-01 -2.14364886e-01 -5.03494084e-01 2.95774013e-01
-9.88699496e-01 -3.58208328e-01 5.54296136e-01 6.96864426e-01
2.84778804e-01 -4.70267773e-01 -1.01716971e+00 2.49595314e-01
-3.00000429e-01 -1.60229599e+00 -1.08201098e+00 -3.50046307e-01
-9.65463221e-01 -1.22624314e+00 -9.92842555e-01 -5.07721603e-01
6.01809680e-01 1.16888490e-02 1.32163632e+00 1.19604193e-01
-6.99363172e-01 9.81389523e-01 -7.36207440e-02 -1.36737198e-01
2.39728317e-01 -1.03406943e-01 5.48438311e-01 -3.91737729e-01
2.21114978e-01 -4.66589957e-01 -8.91654432e-01 5.56855261e-01
-3.20305973e-01 7.81356320e-02 3.88928086e-01 7.00176239e-01
8.32808554e-01 -6.45899653e-01 -2.31024235e-01 -4.63908195e-01
3.08983117e-01 -2.79053628e-01 -6.08075917e-01 -1.42763883e-01
-3.10576230e-01 3.74291204e-02 -4.00388502e-02 -5.47871828e-01
-5.79867959e-01 7.78255343e-01 -4.37733196e-02 -6.53027654e-01
7.89148659e-02 1.88082978e-01 1.72270030e-01 -3.49022597e-01
5.94344020e-01 2.50111939e-03 4.50177640e-01 -5.89049518e-01
3.13400328e-01 6.83897641e-03 8.80756378e-01 -7.68689692e-01
9.45377827e-01 7.19832897e-01 2.33328432e-01 -8.63366961e-01
-7.70002186e-01 -8.03411067e-01 -1.04072857e+00 -4.47220802e-01
7.44607508e-01 -1.19050348e+00 -1.07760394e+00 4.17910635e-01
-1.10282505e+00 -3.31758559e-01 9.31431055e-02 8.08842540e-01
-1.00748134e+00 7.51882255e-01 -8.24273705e-01 -8.03497016e-01
-4.40866530e-01 -1.00229204e+00 1.64362538e+00 -2.35320061e-01
-7.68618703e-01 -9.12675798e-01 2.36201793e-01 6.00237250e-01
1.88983664e-01 8.66589606e-01 -1.41171917e-01 -1.14455245e-01
-3.89005929e-01 -5.45040011e-01 2.02709362e-01 6.96368739e-02
-1.82125971e-01 -2.38350540e-01 -1.80393144e-01 -6.19163632e-01
-1.13124654e-01 -3.55482370e-01 4.67668056e-01 6.00449622e-01
5.88098168e-01 -3.24276507e-01 -4.76711363e-01 8.54584575e-01
1.06514728e+00 -6.64418876e-01 3.10616761e-01 3.31260264e-01
9.63525772e-01 3.01008642e-01 7.65376508e-01 1.00721192e+00
5.63831449e-01 1.09781897e+00 2.66761601e-01 3.08762081e-02
4.08907887e-03 -1.81216791e-01 5.90210557e-01 8.65739226e-01
-6.46446347e-01 2.85060078e-01 -8.97342682e-01 4.79216576e-01
-1.99061847e+00 -7.64406562e-01 -9.09642950e-02 2.23030305e+00
6.35776520e-01 -1.46880513e-02 8.24855149e-01 -9.23854038e-02
4.85190719e-01 6.44686595e-02 -6.20847821e-01 1.72435921e-02
1.10623002e-01 3.12671691e-01 6.39118850e-01 7.25227237e-01
-1.21793783e+00 9.10990655e-01 7.07773161e+00 2.81571299e-01
-4.50986534e-01 5.40943146e-02 -4.79867868e-02 -6.89854026e-01
2.95142651e-01 -1.40434220e-01 -1.10231912e+00 9.69638303e-02
6.08138204e-01 2.58213701e-03 5.00615954e-01 8.75281036e-01
2.35697404e-01 2.38835793e-02 -1.21771801e+00 1.25203884e+00
2.55826533e-01 -1.02220643e+00 -3.28861982e-01 3.03036958e-01
5.82623065e-01 -1.46911461e-02 -1.39074594e-01 -5.49432226e-02
1.97251916e-01 -1.03538013e+00 8.96772027e-01 6.76579177e-01
5.67312717e-01 -6.02812111e-01 4.12086904e-01 4.37783182e-01
-1.55229163e+00 2.67687023e-01 -3.86014044e-01 -3.13119471e-01
6.60570621e-01 6.05151534e-01 -5.21804631e-01 1.08221188e-01
8.57838273e-01 8.16574097e-01 -4.29608673e-01 9.13019359e-01
-2.29392871e-01 3.20633203e-01 -7.38648415e-01 -3.88571881e-02
-5.16837202e-02 -4.06160951e-02 9.87701833e-01 1.33520150e+00
1.54310301e-01 2.90663511e-01 6.19697988e-01 6.50691986e-01
1.45573750e-01 -2.63608946e-03 -4.13620800e-01 2.25244034e-02
2.84059584e-01 1.13844788e+00 -5.08877814e-01 -2.21573934e-01
-2.63828188e-01 1.39255500e+00 3.03875536e-01 6.50907978e-02
-9.88367736e-01 -1.99104026e-01 6.40839458e-01 1.73479557e-01
3.67114127e-01 -1.12724030e+00 -9.78055149e-02 -1.43384480e+00
3.66696358e-01 -1.14904892e+00 3.99396628e-01 -4.75893706e-01
-1.02690220e+00 2.39832193e-01 1.73315048e-01 -1.20153403e+00
-5.84240496e-01 -7.51776576e-01 -5.97020723e-02 5.69273949e-01
-8.79617274e-01 -1.13914120e+00 -4.03342813e-01 9.99084353e-01
4.61147815e-01 9.26410481e-02 7.52756596e-01 2.11413071e-01
-4.20477241e-01 6.25582993e-01 -3.76618594e-01 2.05441907e-01
7.64580309e-01 -1.22357833e+00 9.20034528e-01 8.09623539e-01
5.72834797e-02 7.76860595e-01 8.53466332e-01 -8.21293533e-01
-2.18845415e+00 -4.78877008e-01 7.99210012e-01 -1.03110409e+00
5.35692990e-01 -5.33283710e-01 -2.73692578e-01 9.24741387e-01
-2.09731191e-01 2.26109475e-01 6.95126235e-01 1.65041760e-01
-1.17746510e-01 4.81256902e-01 -9.84244645e-01 5.77084601e-01
1.44787157e+00 -2.50449121e-01 -6.77268207e-01 6.82562053e-01
6.31172657e-01 -1.15908539e+00 -1.16563439e+00 3.19376886e-01
1.11508214e+00 -8.45534921e-01 1.51599336e+00 -5.07065654e-01
1.07099727e-01 -3.73227179e-01 -3.67945403e-01 -8.13653290e-01
-4.39039171e-01 -1.07520843e+00 -8.31792414e-01 2.01241836e-01
-3.02082989e-02 -3.83175939e-01 1.04084492e+00 5.46161056e-01
3.57153043e-02 -9.93898392e-01 -7.54983127e-01 -9.53442633e-01
-3.09819192e-01 -4.79074180e-01 1.15875639e-01 4.85541195e-01
-1.14123244e-02 1.66436449e-01 -9.05799866e-01 2.72387296e-01
1.23409939e+00 -1.61635414e-01 1.40328813e+00 -9.68650818e-01
-6.95884287e-01 -2.77866889e-02 -6.72581136e-01 -1.68414879e+00
-1.55254155e-02 -4.87184197e-01 1.43740431e-03 -1.31337142e+00
9.19867381e-02 2.92219613e-02 3.14442456e-01 4.37141478e-01
-2.18835697e-01 5.96999586e-01 4.48629111e-01 3.73441219e-01
-8.35388839e-01 3.22325438e-01 1.09704459e+00 2.20917493e-01
-6.93235621e-02 -6.19043931e-02 -2.80433893e-01 9.58225369e-01
5.15454829e-01 -3.10480326e-01 1.98088199e-01 -5.49065113e-01
1.80749625e-01 9.90445837e-02 7.35667169e-01 -1.29097807e+00
3.90095681e-01 4.81832912e-03 6.40501559e-01 -7.50663400e-01
7.63943315e-01 -5.01207829e-01 2.17502832e-01 7.58300126e-01
-1.74784400e-02 3.69756073e-01 1.19226821e-01 5.55339396e-01
2.41292134e-01 2.19525337e-01 5.59864938e-01 -3.05922538e-01
-6.48444474e-01 5.44874191e-01 -1.95430011e-01 2.72803694e-01
6.99438632e-01 -3.43826234e-01 3.33545685e-01 -6.28729165e-01
-9.34409559e-01 2.74185836e-01 5.73249161e-01 3.07194084e-01
7.94551730e-01 -1.53531766e+00 -8.40154588e-01 -1.07960582e-01
-1.35201544e-01 3.08209658e-02 9.09932852e-02 1.14296937e+00
-8.78380358e-01 4.49852198e-01 -1.28963739e-01 -9.85906482e-01
-1.36010420e+00 3.20705563e-01 1.96145907e-01 -3.21520180e-01
-7.31541216e-01 7.77612150e-01 -3.29807699e-01 -6.91585243e-01
2.92508125e-01 -1.25723273e-01 4.07084435e-01 -6.19221091e-01
4.44887698e-01 8.69987905e-01 -2.20866606e-01 -9.17573750e-01
-7.06426919e-01 1.00904560e+00 3.86970937e-01 -2.36466616e-01
1.31308067e+00 -1.04216449e-01 1.23878561e-01 2.64519572e-01
1.29422700e+00 3.25964063e-01 -1.42742288e+00 -2.32606307e-01
-2.71517962e-01 -6.46435440e-01 -3.54628801e-01 -3.65974993e-01
-8.19565654e-01 4.24610943e-01 3.90772074e-01 -4.67006207e-01
7.08967030e-01 -1.19431382e-02 1.18228245e+00 6.02413177e-01
5.45292437e-01 -1.16992509e+00 3.64821911e-01 6.65635884e-01
9.75957990e-01 -1.25452995e+00 6.92078173e-01 -7.15110481e-01
-5.61228812e-01 1.11835945e+00 3.96849632e-01 -6.57837331e-01
6.17236137e-01 4.12262172e-01 -2.21812055e-01 -3.68308693e-01
-5.46663284e-01 -1.17805138e-01 6.97966576e-01 6.03566229e-01
5.34241974e-01 1.16370536e-01 -1.26936778e-01 4.79340851e-01
-4.81891572e-01 6.83124438e-02 1.14686489e-01 1.25305152e+00
-4.21398848e-01 -9.81264532e-01 -6.47009432e-01 5.28788306e-02
-4.49528545e-01 1.42311975e-01 -3.15373987e-01 9.57531333e-01
-3.90114576e-01 7.98092723e-01 -2.23500818e-01 -4.11177605e-01
6.21500134e-01 -1.16108172e-01 9.96534884e-01 -5.07140219e-01
-3.92629236e-01 2.35872880e-01 2.93181390e-01 -1.44336641e+00
-5.25580108e-01 -8.67392242e-01 -1.22536170e+00 -6.32688403e-01
-1.75864980e-01 -3.33027422e-01 5.24159491e-01 9.01678503e-01
4.30565685e-01 -7.71487951e-02 1.96815535e-01 -1.61784089e+00
-7.27823436e-01 -6.63594127e-01 -4.43877161e-01 4.60146904e-01
1.38035744e-01 -1.08183503e+00 -3.51152480e-01 1.59760758e-01] | [7.038627624511719, -0.880335807800293] |
2dc7a97a-3e03-4b35-942b-e6573be26c07 | point-cloud-quality-assessment-large-scale | 2012.11895 | null | https://arxiv.org/abs/2012.11895v4 | https://arxiv.org/pdf/2012.11895v4.pdf | Point Cloud Quality Assessment: Dataset Construction and Learning-based No-Reference Metric | Full-reference (FR) point cloud quality assessment (PCQA) has achieved impressive progress in recent years. However, in many cases, obtaining the reference point clouds is difficult, so no-reference (NR) metrics have become a research hotspot. Few researches about NR-PCQA are carried out due to the lack of a large-scale PCQA dataset. In this paper, we first build a large-scale PCQA dataset named LS-PCQA, which includes 104 reference point clouds and more than 22,000 distorted samples. In the dataset, each reference point cloud is augmented with 31 types of impairments (e.g., Gaussian noise, contrast distortion, local missing, and compression loss) at 7 distortion levels. Besides, each distorted point cloud is assigned with a pseudo quality score as its substitute of Mean Opinion Score (MOS). Inspired by the hierarchical perception system and considering the intrinsic attributes of point clouds, we propose a NR metric ResSCNN based on sparse convolutional neural network (CNN) to accurately estimate the subjective quality of point clouds. We conduct several experiments to evaluate the performance of the proposed NR metric. The results demonstrate that ResSCNN exhibits the state-of-the-art (SOTA) performance among all the existing NR-PCQA metrics and even outperforms some FR metrics. The dataset presented in this work will be made publicly accessible at http://smt.sjtu.edu.cn. The source code for the proposed ResSCNN can be found at https://github.com/lyp22/ResSCNN. | ['Le Yang', 'Yiling Xu', 'Qi Yang', 'Yipeng Liu'] | 2020-12-22 | null | null | null | null | ['blind-image-quality-assessment'] | ['computer-vision'] | [ 3.71925184e-03 -7.33442783e-01 1.88496843e-01 -2.66859293e-01
-1.03686082e+00 -2.38941029e-01 1.25794917e-01 4.70521674e-02
5.62291332e-02 4.97946948e-01 1.74639553e-01 5.38339429e-02
-2.73372173e-01 -9.12332594e-01 -6.32572114e-01 -7.19204664e-01
1.20462403e-01 3.49029191e-02 2.06230208e-01 -3.82565945e-01
4.04879212e-01 7.11142480e-01 -1.70546925e+00 1.29423395e-01
1.19218802e+00 1.49527335e+00 3.48032504e-01 4.20235813e-01
1.78285003e-01 3.36862594e-01 -9.11973357e-01 -6.13797009e-01
3.34033579e-01 -1.44532025e-01 -3.13434094e-01 -3.05364281e-02
3.85036141e-01 -4.96497601e-01 -5.07071495e-01 1.27500010e+00
8.01462829e-01 1.55598953e-01 3.31215560e-01 -1.52436638e+00
-1.06041718e+00 -1.39605291e-02 -5.43066025e-01 2.42969438e-01
2.52322465e-01 2.87368417e-01 9.02692437e-01 -1.04598927e+00
1.28470063e-01 1.17549944e+00 4.22550946e-01 1.32822961e-01
-7.69633472e-01 -9.36973095e-01 -2.63100982e-01 7.99892306e-01
-1.78346002e+00 -3.33767414e-01 9.33760166e-01 -1.23176143e-01
5.13255656e-01 4.04842794e-01 4.30037618e-01 6.93197668e-01
1.06982008e-01 6.32882118e-01 7.75210261e-01 7.37798139e-02
2.88918227e-01 -3.21758777e-01 -3.01533371e-01 2.61961251e-01
2.25432694e-01 1.47807270e-01 -2.14771315e-01 -3.09094265e-02
9.31213021e-01 1.95812985e-01 -6.37712717e-01 -2.35499024e-01
-1.27302039e+00 4.21727747e-01 9.00648773e-01 3.06715041e-01
-4.30155426e-01 4.81486060e-02 3.41832846e-01 3.35537702e-01
2.13987872e-01 1.56167433e-01 -2.64032900e-01 -2.64500409e-01
-6.52372122e-01 1.88070059e-01 1.78931624e-01 1.18796527e+00
6.83612823e-01 1.43971458e-01 -2.41278782e-01 1.06715238e+00
2.20752805e-01 9.49011326e-01 3.08737516e-01 -1.38271809e+00
7.28478193e-01 5.06317437e-01 3.46758515e-01 -1.59869182e+00
-7.16237426e-02 -6.36586368e-01 -1.24196017e+00 2.42653221e-01
-1.21927470e-01 1.99099645e-01 -4.92551446e-01 1.19350541e+00
9.17719826e-02 2.65425533e-01 2.30436083e-02 1.26949871e+00
9.81408656e-01 1.03070068e+00 -3.86335373e-01 -2.88482100e-01
9.81052458e-01 -7.22229123e-01 -7.45541811e-01 4.88894761e-01
7.84553364e-02 -9.29782450e-01 1.16629946e+00 7.86499679e-01
-1.26175964e+00 -9.26105618e-01 -1.20563388e+00 -1.30917445e-01
-6.80055469e-02 2.97635823e-01 1.62818506e-01 4.60600346e-01
-1.09761608e+00 5.32359660e-01 -6.80684268e-01 9.61264595e-03
5.30247092e-01 2.55353421e-01 -1.86222568e-01 -5.55650890e-01
-1.16112411e+00 5.25382459e-01 -5.48200011e-02 3.54269534e-01
-8.08548748e-01 -6.69893086e-01 -5.20084679e-01 1.96463495e-01
2.70896107e-01 -5.19514501e-01 1.07788980e+00 -7.85578430e-01
-1.30847681e+00 2.75792569e-01 -1.95392355e-01 9.43857655e-02
2.25390971e-01 -2.13292256e-01 -8.93831253e-01 3.63232970e-01
7.88714439e-02 4.40282643e-01 6.22766614e-01 -1.60041654e+00
-6.16793156e-01 -3.65245432e-01 1.78036600e-01 3.25249434e-01
-2.16689780e-01 5.70440479e-02 -7.08793998e-01 -6.28624856e-01
2.59020507e-01 -5.03489554e-01 9.55064967e-03 3.33152473e-01
-3.09136182e-01 -1.26918197e-01 6.48284793e-01 -6.37286782e-01
1.14662790e+00 -2.37690568e+00 -1.46579742e-02 1.42114356e-01
1.58687234e-01 3.46482605e-01 -3.53571236e-01 3.67702574e-01
-8.07639956e-02 7.27592856e-02 -3.19656372e-01 -1.35452002e-01
7.24789351e-02 1.04339838e-01 -2.01721758e-01 4.84988183e-01
1.72226161e-01 5.98679543e-01 -8.19978774e-01 -3.63573104e-01
4.11780000e-01 6.51198268e-01 -2.96319425e-01 2.07215294e-01
2.61975616e-01 4.14649129e-01 -4.29077536e-01 9.43159699e-01
1.34097672e+00 -1.67594820e-01 -6.04527712e-01 -6.06534839e-01
-1.52149066e-01 -1.62880421e-01 -1.28425574e+00 1.74360561e+00
-5.02495825e-01 5.11985898e-01 -3.11596338e-02 -6.22765720e-01
1.15115488e+00 3.37932378e-01 6.06659830e-01 -9.54116464e-01
1.77564129e-01 4.71639663e-01 -4.36017290e-02 -4.23238695e-01
6.64909005e-01 1.01940908e-01 2.25603834e-01 -1.28867850e-01
-6.93015680e-02 -2.19300300e-01 -2.12088786e-03 3.15833502e-02
1.02031875e+00 -2.73282319e-01 -5.74842189e-03 2.31767491e-01
8.51055086e-01 -1.70602679e-01 8.22122097e-01 2.33577535e-01
-4.38538641e-01 1.12419415e+00 -1.65754259e-02 -1.52433962e-01
-1.15655541e+00 -1.36683416e+00 -2.19777256e-01 4.85533595e-01
6.35665774e-01 -1.72437027e-01 -3.54966164e-01 -5.26210852e-03
-1.34418204e-01 6.07868314e-01 -1.22667722e-01 -1.54556438e-01
-4.69632030e-01 -3.24903995e-01 4.72503006e-01 4.09385949e-01
9.28855598e-01 -1.01051784e+00 -2.12894548e-02 5.15960641e-02
-5.43360472e-01 -9.64560688e-01 -4.37517107e-01 -6.92686737e-01
-8.45529258e-01 -9.64239061e-01 -9.92602289e-01 -5.90464532e-01
4.94036406e-01 7.31591105e-01 1.15597558e+00 2.84569114e-01
2.17491478e-01 1.48975968e-01 -6.47370517e-01 -2.93272704e-01
-4.25691018e-03 -2.83564150e-01 1.04062654e-01 8.57214406e-02
3.82922262e-01 -7.00115263e-01 -9.09138083e-01 7.03078508e-01
-1.08747876e+00 -3.01887929e-01 7.30408967e-01 3.81277025e-01
9.12841260e-01 4.89337087e-01 4.51818019e-01 7.49418139e-02
5.79115331e-01 -5.37999451e-01 -6.83443606e-01 -8.32178816e-02
-3.87975335e-01 -6.33764923e-01 7.69003451e-01 -5.62258065e-02
-7.86674261e-01 -4.82905358e-01 -4.94984806e-01 -9.03249383e-01
-3.44936997e-01 4.45083618e-01 -5.84014654e-01 -1.94769502e-01
3.93555135e-01 3.21805388e-01 -2.67670631e-01 -4.87226188e-01
1.58627898e-01 9.34607625e-01 7.94080019e-01 -4.85373080e-01
1.20259464e+00 5.15888572e-01 -2.06580712e-03 -6.58265173e-01
-5.80892503e-01 -5.10961771e-01 -3.38826358e-01 -3.00907552e-01
5.81281245e-01 -1.07594371e+00 -7.96904683e-01 5.95419407e-01
-1.16818380e+00 8.77117813e-02 -1.57902256e-01 5.60404301e-01
-5.03820896e-01 5.40151715e-01 -4.10159916e-01 -5.49179137e-01
-4.06769067e-01 -1.30070853e+00 1.05593944e+00 2.86814511e-01
5.15584469e-01 -3.49319279e-01 -1.08574852e-01 4.54466254e-01
5.99227190e-01 1.67113021e-01 3.52291197e-01 -8.71788338e-02
-8.93154740e-01 -1.81908876e-01 -6.10317290e-01 6.69367373e-01
2.37946123e-01 1.27702683e-01 -6.94092453e-01 -4.37732041e-01
2.51429558e-01 -5.22000045e-02 3.44959527e-01 3.56676668e-01
1.65748239e+00 -2.54724979e-01 1.74151644e-01 8.11576486e-01
1.69042194e+00 3.87387455e-01 1.11851656e+00 2.89736241e-01
7.45429397e-01 1.27905414e-01 8.98692727e-01 5.53883851e-01
4.83681947e-01 8.43291700e-01 7.87546337e-01 9.19618160e-02
-2.00909469e-02 4.77325693e-02 2.55598128e-01 1.30528057e+00
-3.72907907e-01 -4.05969083e-01 -8.97638500e-01 5.92850566e-01
-1.36793041e+00 -9.20590699e-01 -3.63827139e-01 2.06260467e+00
4.70955372e-01 -1.27569176e-02 -2.35798493e-01 6.27385855e-01
8.37933958e-01 1.53951332e-01 -7.01804578e-01 -8.74202698e-02
-2.92300463e-01 9.48736593e-02 3.10276598e-01 7.30285496e-02
-8.04883957e-01 2.70390689e-01 4.43273306e+00 1.10886824e+00
-1.05839217e+00 1.59370229e-01 4.95222777e-01 -9.13257226e-02
-1.88261062e-01 -4.15896565e-01 -2.88005710e-01 7.85846770e-01
8.58177662e-01 -1.85712636e-01 5.92523932e-01 6.75954044e-01
3.69574219e-01 2.25401744e-01 -6.58845186e-01 1.52972925e+00
1.49479941e-01 -1.18102765e+00 1.88145384e-01 -1.58513765e-04
7.64813483e-01 2.43915558e-01 3.89847755e-01 9.79667231e-02
-1.73110180e-02 -8.55889976e-01 5.25568843e-01 6.77682281e-01
1.23868322e+00 -9.98547316e-01 1.15376341e+00 8.99187848e-02
-1.35195315e+00 -2.00822711e-01 -8.74714017e-01 9.91676226e-02
1.35587618e-01 7.94455051e-01 -2.26010079e-03 1.12491393e+00
1.03211772e+00 8.83043051e-01 -5.57201445e-01 1.58647549e+00
-7.93625861e-02 4.92400408e-01 -1.00448914e-01 2.54366994e-01
3.68843302e-02 -2.08593681e-01 5.90863526e-01 5.61021686e-01
9.44833815e-01 4.40554291e-01 -2.17620194e-01 8.04218888e-01
-1.68641880e-01 1.50152758e-01 -1.90649167e-01 3.50910485e-01
7.33520746e-01 1.15805078e+00 -7.55294859e-02 -2.53504872e-01
-4.43010718e-01 8.06671500e-01 -2.34739557e-01 3.27625722e-01
-7.33386457e-01 -7.02308118e-01 9.02712584e-01 -2.19446436e-01
3.56840670e-01 -1.77489772e-01 -3.12491864e-01 -1.24865437e+00
3.59242141e-01 -8.62196624e-01 2.66211126e-02 -1.41440666e+00
-1.57962298e+00 6.17862761e-01 -2.01890096e-01 -2.04077029e+00
2.82350957e-01 -2.81414628e-01 -6.13949597e-01 1.04356718e+00
-1.66301703e+00 -7.57590711e-01 -9.67978179e-01 7.60960400e-01
4.76636916e-01 -1.00093357e-01 5.58892965e-01 7.15681553e-01
-3.68404567e-01 5.99742770e-01 4.89745140e-01 1.13990761e-01
6.21238232e-01 -9.21732426e-01 3.74453008e-01 7.56055117e-01
-3.55919957e-01 3.72156829e-01 5.96008480e-01 -3.60018402e-01
-1.31326234e+00 -1.26624930e+00 4.21562195e-01 -2.68745750e-01
5.14339685e-01 6.91410303e-02 -1.15531933e+00 5.97648434e-02
3.85048687e-02 2.91714758e-01 3.91985416e-01 -4.48356450e-01
-1.35947242e-01 -6.09503925e-01 -1.39695263e+00 3.79179209e-01
9.89868701e-01 -4.17346150e-01 -2.96429306e-01 1.29676953e-01
1.14261127e+00 -5.45552433e-01 -1.21497071e+00 5.89094639e-01
2.07922801e-01 -1.17549741e+00 1.02760506e+00 3.33297104e-01
7.29348242e-01 -8.74621928e-01 -5.86445093e-01 -1.40799212e+00
-5.14559686e-01 4.34192605e-02 -6.05174862e-02 1.25564611e+00
-1.03510387e-01 -5.42088449e-01 5.34022093e-01 -9.31195449e-03
-6.19779587e-01 -8.31033230e-01 -9.98402357e-01 -8.61771107e-01
-6.12025894e-02 -5.02326787e-01 1.27036834e+00 7.51049161e-01
-6.19328976e-01 -8.06419849e-02 -8.94140154e-02 7.02501297e-01
6.58941686e-01 7.21102282e-02 6.85844958e-01 -1.08723021e+00
-6.05048961e-04 -3.28583747e-01 -8.10969412e-01 -1.09354961e+00
-3.91484201e-01 -5.31743526e-01 -6.85449988e-02 -1.76290834e+00
-4.58784141e-02 -5.58850765e-01 -5.94905853e-01 5.26815355e-02
-1.01275966e-01 4.24633861e-01 3.31071675e-01 5.55153787e-01
-6.82557821e-01 1.00347650e+00 1.57858932e+00 -2.85175145e-01
4.83764224e-02 4.95443046e-02 -6.81512952e-01 3.74302328e-01
1.06783438e+00 -1.07407629e-01 -3.67104441e-01 -8.26083362e-01
1.70701936e-01 2.10947722e-01 4.01425481e-01 -1.63239372e+00
2.50321507e-01 -1.55871600e-01 4.60663229e-01 -1.07314324e+00
5.73756456e-01 -1.12068903e+00 3.40927392e-01 1.24888524e-01
3.06976996e-02 3.35981131e-01 4.98582125e-02 4.22078252e-01
-6.46365941e-01 8.34409846e-04 6.02176428e-01 2.55483352e-02
-6.29882991e-01 6.59640670e-01 2.70147920e-01 -2.23865092e-01
6.79705560e-01 -2.64710218e-01 -6.38858080e-01 -4.29477602e-01
-2.64715791e-01 2.22933829e-01 7.39258230e-01 3.74882311e-01
1.20545936e+00 -1.74208808e+00 -9.01896656e-01 1.55279448e-03
4.03597176e-01 3.09857309e-01 7.30820715e-01 6.41892850e-01
-6.76814198e-01 3.32269460e-01 -4.82641041e-01 -6.53740168e-01
-9.95320797e-01 6.46598399e-01 2.44575918e-01 2.25736707e-01
-3.38190675e-01 5.59540570e-01 -2.28190999e-02 -4.15924877e-01
1.28051504e-01 -2.51312584e-01 -3.89812201e-01 -4.61811811e-01
6.82628870e-01 7.12432504e-01 3.48992139e-01 -1.09976888e+00
-1.54301465e-01 6.93248630e-01 3.60207379e-01 1.80272356e-01
1.39821148e+00 -3.63920003e-01 -2.00296938e-01 2.82113671e-01
1.37060821e+00 -1.62725486e-02 -1.06219041e+00 -2.72297978e-01
-5.92628479e-01 -1.04052818e+00 1.69205859e-01 -7.39937663e-01
-1.47126567e+00 9.47567761e-01 1.00940120e+00 5.07338382e-02
1.61731863e+00 -2.27215946e-01 1.12661409e+00 2.05643147e-01
6.29280508e-01 -6.21390939e-01 4.74889204e-02 2.57131904e-01
1.32613397e+00 -1.21852183e+00 -2.78910296e-03 -2.93607593e-01
-5.91470778e-01 8.75454664e-01 6.36161208e-01 -3.14213604e-01
5.41495740e-01 -3.21771353e-01 1.61533490e-01 -1.35684401e-01
-4.67468917e-01 4.70092371e-02 2.67816752e-01 7.11994052e-01
3.32714058e-02 1.07925840e-01 -1.07799344e-01 4.21922386e-01
-4.88971233e-01 1.14730626e-01 6.42403007e-01 5.20913243e-01
-3.17269027e-01 -7.99332678e-01 -6.87752306e-01 5.38898528e-01
-3.12611669e-01 2.16658376e-02 2.20855311e-01 2.42661670e-01
3.90154004e-01 1.32224548e+00 1.79667920e-01 -7.06155837e-01
6.41303718e-01 -8.09203148e-01 4.05092053e-02 -2.56294399e-01
-1.18377164e-01 -2.11320013e-01 -4.64644343e-01 -7.99218059e-01
-6.26978159e-01 -5.10640204e-01 -1.09875655e+00 -6.63528085e-01
-2.23660722e-01 1.03083588e-01 6.85685873e-01 3.82515907e-01
3.40922803e-01 6.00053668e-01 1.05996728e+00 -1.00195491e+00
-1.93998590e-01 -8.63980532e-01 -6.09922290e-01 4.17173982e-01
4.86303777e-01 -7.26696014e-01 -4.43407506e-01 -1.72296435e-01] | [11.8110933303833, -1.8576875925064087] |
8d0cd889-6efb-4638-972d-947442f320c1 | forecasting-crime-with-deep-learning | 1806.01486 | null | http://arxiv.org/abs/1806.01486v1 | http://arxiv.org/pdf/1806.01486v1.pdf | Forecasting Crime with Deep Learning | The objective of this work is to take advantage of deep neural networks in
order to make next day crime count predictions in a fine-grain city partition.
We make predictions using Chicago and Portland crime data, which is augmented
with additional datasets covering weather, census data, and public
transportation. The crime counts are broken into 10 bins and our model predicts
the most likely bin for a each spatial region at a daily level. We train this
data using increasingly complex neural network structures, including variations
that are suited to the spatial and temporal aspects of the crime prediction
problem. With our best model we are able to predict the correct bin for overall
crime count with 75.6% and 65.3% accuracy for Chicago and Portland,
respectively. The results show the efficacy of neural networks for the
prediction problem and the value of using external datasets in addition to
standard crime data. | ['Diego Klabjan', 'Alexander Stec'] | 2018-06-05 | null | null | null | null | ['crime-prediction'] | ['miscellaneous'] | [-2.46134654e-01 -8.52961168e-02 7.89420605e-02 -5.55632353e-01
-5.11979997e-01 -2.22352698e-01 5.73041916e-01 5.26081324e-01
-7.74988472e-01 1.06285739e+00 7.53931522e-01 -4.61893886e-01
-3.04852247e-01 -1.54370570e+00 -5.66299975e-01 -2.65253652e-02
-3.23725998e-01 6.70146585e-01 -1.00000598e-01 -4.07078177e-01
1.08897440e-01 5.86711824e-01 -1.25795722e+00 3.68675679e-01
7.30737627e-01 7.81243443e-01 -1.52587369e-01 5.06622553e-01
9.76766944e-02 6.49717987e-01 -2.32067049e-01 -3.90915364e-01
4.72407192e-01 1.89347759e-01 -9.94855523e-01 -2.88743764e-01
3.54022771e-01 -5.29468954e-01 -5.23340940e-01 5.09792566e-01
3.44334871e-01 6.20535076e-01 6.67917788e-01 -9.08352673e-01
-4.36746538e-01 7.46056914e-01 -4.04189020e-01 6.99696243e-01
3.08736473e-01 1.75676808e-01 7.70473778e-01 -4.87073123e-01
3.44714403e-01 9.23571885e-01 1.06569755e+00 -7.05208583e-03
-1.25376821e+00 -7.24967837e-01 1.22670140e-02 1.76063150e-01
-1.64601743e+00 -5.06041408e-01 3.32668453e-01 -8.04576814e-01
1.69464457e+00 1.30864337e-01 7.36568928e-01 9.08352137e-01
-7.35268444e-02 3.98052186e-01 8.12293828e-01 -8.91464651e-02
2.10330829e-01 -2.21042812e-01 4.60355908e-01 2.09969446e-01
3.24839562e-01 3.00434381e-01 -3.55543792e-02 -4.59626503e-02
5.12253582e-01 2.95405567e-01 3.13631207e-01 4.52537805e-01
-5.87229490e-01 9.91360366e-01 5.23916304e-01 3.06786329e-01
-4.68701333e-01 1.24186859e-01 4.31931198e-01 -1.99346021e-01
1.10803854e+00 3.07043403e-01 -4.97767180e-01 -4.52770233e-01
-1.29127550e+00 7.50481725e-01 5.60945451e-01 2.73060352e-01
8.54024768e-01 -6.66916370e-02 -5.89965954e-02 8.99425983e-01
-1.05934694e-01 4.27148014e-01 7.50050023e-02 -7.81269670e-01
1.09144068e+00 5.37942469e-01 1.76978841e-01 -1.33269882e+00
-1.01280236e+00 -1.39406592e-01 -1.02410579e+00 1.61367096e-02
6.62024021e-01 -4.61037308e-01 -8.87919724e-01 1.44504821e+00
4.19590995e-02 4.36546892e-01 -2.89664418e-01 5.94620883e-01
4.11604077e-01 6.15505457e-01 3.60822976e-01 2.96281427e-01
8.35629702e-01 -2.30888873e-01 -2.01901749e-01 -4.30922687e-01
8.77461910e-01 -1.01134598e-01 5.69647253e-01 4.21945639e-02
-1.16290867e+00 -6.44300461e-01 -6.51898265e-01 -5.71935698e-02
-9.21484232e-01 -2.20211715e-01 6.74040675e-01 7.58955359e-01
-1.04421520e+00 7.66920090e-01 -1.14176142e+00 -4.40201253e-01
8.47336948e-01 3.12887728e-01 -3.21736574e-01 -1.42380863e-01
-1.34866583e+00 9.75290239e-01 6.04512036e-01 -1.40431941e-01
-1.37440234e-01 -8.11422706e-01 -1.19616449e+00 4.69025373e-01
-4.73893970e-01 -2.33443052e-01 6.12457275e-01 -4.69483137e-01
-6.14440203e-01 6.93912268e-01 -1.80191234e-01 -6.60023928e-01
3.75743389e-01 1.83569059e-01 -4.78540957e-01 -3.28424722e-01
6.79534853e-01 5.52410722e-01 -2.93535680e-01 -6.06616855e-01
-1.12991452e+00 -5.60427248e-01 -4.37151119e-02 -1.53148383e-01
-4.02214885e-01 6.30735606e-02 -2.25700419e-02 -4.42477435e-01
-1.99975416e-01 -6.03951931e-01 -5.10644794e-01 -1.02057171e+00
-2.92427152e-01 -3.58609438e-01 2.60784626e-01 -1.09982109e+00
1.62579048e+00 -1.87628651e+00 -6.11083746e-01 4.73212659e-01
3.14351730e-02 2.64878929e-01 -1.85046583e-01 4.58088577e-01
-2.97054648e-01 4.11976695e-01 -1.03757702e-01 -5.23095965e-01
2.12840885e-01 1.39503703e-01 -3.43229681e-01 5.31962454e-01
3.90824080e-01 8.51141632e-01 -6.20863914e-01 -1.28021926e-01
5.18470705e-01 4.01043504e-01 -9.02375340e-01 -3.38711530e-01
2.65285790e-01 2.19813287e-01 -2.29283392e-01 3.81883383e-01
9.41631377e-01 7.29668187e-03 -2.05957323e-01 5.44534445e-01
-3.61744583e-01 7.66469777e-01 -1.13460910e+00 1.03842914e+00
-3.67749035e-01 8.38719428e-01 -3.39525551e-01 -1.05996263e+00
8.34980130e-01 6.59525618e-02 5.68230331e-01 -1.20686662e+00
-6.60133287e-02 -7.90647715e-02 -1.43965244e-01 -3.60949069e-01
8.02133858e-01 -2.32018773e-02 -3.75046253e-01 7.39041924e-01
-3.77673894e-01 2.51312077e-01 5.73329329e-01 -1.76799089e-01
1.29057121e+00 -3.58954996e-01 2.25068077e-01 -3.06185573e-01
-4.80914377e-02 1.08626790e-01 6.17829621e-01 9.44925487e-01
-2.63954163e-01 7.06077337e-01 4.77502227e-01 -1.36766946e+00
-1.29223323e+00 -9.81413245e-01 -5.68262577e-01 1.01154912e+00
-5.86882234e-01 -2.07584992e-01 -4.98315305e-01 -2.40048990e-01
7.30558559e-02 7.64577925e-01 -8.42202365e-01 1.88818038e-01
-7.30105102e-01 -1.31039143e+00 7.68882334e-01 6.30406857e-01
6.87334895e-01 -9.43419158e-01 -5.66748321e-01 4.48426425e-01
-3.85998428e-01 -1.01226103e+00 1.70857355e-01 1.67889088e-01
-4.21533883e-01 -1.01977634e+00 -2.22589388e-01 -3.51535529e-01
1.42134935e-01 -2.12127909e-01 1.24017990e+00 6.02166168e-02
-1.70022041e-01 -3.57590541e-02 -2.96669662e-01 -4.58096564e-01
6.56861067e-02 5.77586234e-01 8.04307014e-02 -2.07511961e-01
9.24615622e-01 -9.51720595e-01 -4.90298361e-01 -2.50395119e-01
-6.65212274e-01 1.03063799e-01 -2.85732269e-01 5.19916415e-01
9.34793949e-02 4.51182693e-01 4.28796083e-01 -6.76188529e-01
7.29445636e-01 -9.73515987e-01 -4.78817195e-01 -2.07903519e-01
-3.76222908e-01 -4.54494774e-01 5.93310475e-01 2.51235813e-01
-8.02266419e-01 1.31492037e-02 -4.69083458e-01 2.54231721e-01
-8.76595676e-01 6.88306332e-01 3.22340488e-01 5.69140971e-01
8.39802623e-01 -6.13670126e-02 -7.94180393e-01 -3.73576403e-01
-2.56720573e-01 4.70537901e-01 2.76181519e-01 -7.03051507e-01
4.30276543e-01 4.23198998e-01 -3.85149010e-02 -6.06578529e-01
-7.08808124e-01 -4.39438492e-01 -1.03719831e+00 1.09752871e-01
1.03289497e+00 -9.20018315e-01 -8.38162839e-01 4.53748912e-01
-1.29669368e+00 -7.53209352e-01 -1.44288421e-01 3.92574966e-01
-3.26139480e-01 -7.00088292e-02 -7.59398699e-01 -9.66119945e-01
-4.87572290e-02 -7.31270909e-01 7.95917749e-01 7.44976625e-02
-3.68177474e-01 -1.22046065e+00 4.93941128e-01 3.20594013e-01
3.39429229e-01 6.61805689e-01 8.81949604e-01 -6.98794246e-01
-1.40100077e-01 -3.13024819e-01 -5.92402577e-01 -7.01647922e-02
-7.98616838e-03 -5.22668324e-02 -1.10604143e+00 3.41839269e-02
-8.35223556e-01 -1.35982111e-01 1.23233879e+00 7.75936604e-01
1.34183776e+00 -2.92135239e-01 -2.81069010e-01 7.04526544e-01
1.45689607e+00 1.39233053e-01 7.70062327e-01 6.58457875e-01
4.42126274e-01 6.52139068e-01 1.00525118e-01 9.21249390e-01
9.09784496e-01 4.71302778e-01 3.46264839e-01 -4.01851475e-01
4.40924197e-01 -2.42866084e-01 -3.15771490e-01 -8.71203765e-02
-4.36008841e-01 -1.54461890e-01 -1.88180923e+00 1.11088598e+00
-1.98056483e+00 -1.58829999e+00 -1.58221617e-01 1.88450694e+00
3.89933914e-01 2.14140356e-01 3.07247043e-01 3.15963715e-01
4.82607603e-01 1.27361611e-01 -1.21752292e-01 -6.67934835e-01
-2.41236687e-02 7.82244980e-01 7.32811987e-01 6.58747196e-01
-1.55944705e+00 9.77216363e-01 7.10502625e+00 7.84609020e-01
-7.37973988e-01 -6.60917312e-02 1.39851785e+00 -5.47368526e-01
1.34687334e-01 -2.79374301e-01 -8.35263133e-01 6.33353591e-01
1.31341231e+00 6.77292049e-02 7.65131474e-01 4.72572356e-01
7.60334611e-01 -3.24762017e-01 -7.30404794e-01 7.27731764e-01
-4.33610499e-01 -1.69913721e+00 -3.39768469e-01 3.23335797e-01
8.57341707e-01 4.91051823e-01 6.27536103e-02 6.18637502e-01
9.21263874e-01 -1.62573421e+00 4.51703608e-01 6.23731434e-01
6.52304828e-01 -1.29146492e+00 8.80881667e-01 6.46531582e-01
-1.11093283e+00 -2.88068384e-01 -5.16265690e-01 -1.21078765e+00
2.93445945e-01 5.72749436e-01 -9.28271651e-01 -1.13677606e-02
1.15987384e+00 9.79294777e-01 -4.51418728e-01 7.85397768e-01
2.78501809e-01 6.38294458e-01 -6.00244880e-01 2.31512055e-01
6.23585820e-01 -1.29399151e-01 -2.57859319e-01 1.49896014e+00
5.58755040e-01 5.72496891e-01 2.39797026e-01 1.00368333e+00
8.41817707e-02 -2.53041804e-01 -8.47083092e-01 4.79972839e-01
5.25481403e-01 7.98718095e-01 -5.78211546e-01 -2.44548604e-01
-1.99777961e-01 3.43013555e-01 8.72142375e-01 2.73233920e-01
-8.26759756e-01 -2.72078961e-01 1.09836340e+00 4.18863416e-01
2.04836786e-01 -2.01828644e-01 -9.13053155e-01 -8.70943010e-01
-2.39583001e-01 -1.86653271e-01 4.42363650e-01 -6.41625166e-01
-1.31093359e+00 3.15581650e-01 1.67023867e-01 -5.45746446e-01
-4.55706269e-01 -5.44602871e-01 -1.03699708e+00 1.21251285e+00
-1.43120444e+00 -1.12069297e+00 1.19262911e-01 6.95149899e-01
1.80515587e-01 -1.36871248e-01 8.95171404e-01 5.05353212e-01
-7.98548996e-01 4.39235717e-01 3.51874113e-01 7.82096267e-01
1.45137802e-01 -1.03274655e+00 9.42678511e-01 9.31689739e-01
-1.33407325e-01 3.46466213e-01 3.59654427e-01 -5.85898101e-01
-1.83423489e-01 -1.53945315e+00 1.65279078e+00 -7.45673358e-01
5.96103489e-01 -3.65677178e-01 -4.88787949e-01 8.62880111e-01
-1.85085818e-01 -9.86875892e-02 9.11749244e-01 8.03180277e-01
-5.45641035e-02 -8.16875845e-02 -1.37913299e+00 3.61886948e-01
9.98389125e-01 -5.51932931e-01 -3.38014245e-01 3.31517786e-01
2.85349667e-01 -2.70181239e-01 -8.79080951e-01 1.17075384e-01
3.98856729e-01 -1.36721027e+00 1.02911508e+00 -7.57629156e-01
9.81879234e-01 4.20396566e-01 -2.36668944e-01 -1.44552433e+00
-8.32759917e-01 3.69610786e-01 5.57549298e-01 1.11841464e+00
6.36079729e-01 -6.35778904e-01 1.01173091e+00 1.53978968e+00
-6.45535216e-02 -5.44552684e-01 -1.46908450e+00 -2.80371696e-01
6.86183214e-01 -1.25187600e+00 1.30209506e+00 1.16268015e+00
3.43906790e-01 -6.79428130e-02 -2.45403394e-01 2.65860230e-01
2.39896536e-01 -1.32673994e-01 6.85365260e-01 -1.25301492e+00
8.56677815e-02 -6.18022501e-01 -5.43996453e-01 -4.85515475e-01
4.15539443e-01 -7.48116076e-01 -5.68334103e-01 -1.55956721e+00
2.43611217e-01 -8.27645123e-01 -1.91481143e-01 7.76582837e-01
7.93865770e-02 5.99859059e-01 1.78806216e-01 -3.39334637e-01
-3.66395563e-01 1.96670175e-01 5.23026109e-01 -2.88645953e-01
-3.17476481e-01 1.90397706e-02 -6.12069070e-01 7.24409580e-01
1.12027228e+00 -3.45490545e-01 1.85461603e-02 -9.13326919e-01
5.31357765e-01 1.29017845e-01 5.38330674e-01 -1.33562577e+00
1.41738325e-01 -4.85335439e-01 1.03529668e+00 -8.36395621e-01
3.18642139e-01 -7.92944968e-01 1.30209224e-02 1.57505542e-01
-2.28611872e-01 1.70642331e-01 5.96887469e-01 2.02996567e-01
-4.29409407e-02 4.20396835e-01 5.19116819e-01 -1.79957703e-01
-5.41711211e-01 6.75767481e-01 -6.31801724e-01 -7.15087876e-02
8.20388079e-01 -5.00360548e-01 -3.63310456e-01 -3.09300810e-01
-1.05704463e+00 3.49166960e-01 8.36424083e-02 2.63214916e-01
4.10997510e-01 -1.49738955e+00 -1.07391942e+00 5.47208071e-01
-1.91407934e-01 -6.56744018e-02 3.55196714e-01 4.13028330e-01
-5.12733698e-01 8.27496767e-01 -4.25439954e-01 -1.59741089e-01
-6.73551381e-01 3.30709338e-01 4.74295825e-01 -7.72286415e-01
-4.42774057e-01 5.42894781e-01 -2.04845622e-01 -7.95950115e-01
-3.43195260e-01 -3.80314916e-01 -4.61639196e-01 -8.35200474e-02
6.48513734e-01 7.32345998e-01 1.35523692e-01 -7.37509012e-01
-1.74606353e-01 -4.55319732e-02 2.92463243e-01 -7.19456524e-02
1.96934390e+00 -1.04908101e-01 -1.96996741e-02 1.80390835e-01
1.08741832e+00 -3.92651439e-01 -1.07409787e+00 9.19847935e-03
-1.32477075e-01 -5.89821279e-01 1.43650517e-01 -8.30060601e-01
-9.41951513e-01 9.32336748e-01 4.51665848e-01 5.68563104e-01
9.71410930e-01 -2.80635059e-01 8.88684154e-01 1.79537550e-01
3.12912375e-01 -1.45484877e+00 -7.69428134e-01 1.24550414e+00
2.90143400e-01 -1.37289011e+00 -1.17291503e-01 2.74534911e-01
-1.68464050e-01 9.56528306e-01 5.57865083e-01 -5.25097549e-01
1.00673270e+00 2.59477258e-01 -5.08961856e-01 -2.00312957e-01
-5.85746229e-01 -3.41036975e-01 -4.67705391e-02 7.70286620e-01
4.92955744e-01 6.22176707e-01 1.43096540e-02 8.78162086e-01
-7.82035470e-01 -6.74346015e-02 3.73313189e-01 3.49781036e-01
-4.37330782e-01 -6.48088753e-01 -5.88946283e-01 9.40932691e-01
-5.30469775e-01 -3.40247363e-01 -2.60741580e-02 8.27572763e-01
6.82702243e-01 1.21672356e+00 7.72325039e-01 -4.20367897e-01
2.17855543e-01 -3.35574858e-02 -6.88209757e-02 -5.43596566e-01
-7.94841170e-01 -5.14369726e-01 3.36468667e-01 -5.19585192e-01
7.98192471e-02 -8.81702125e-01 -9.63722587e-01 -1.31276965e+00
3.95436525e-01 -1.46096528e-01 2.69212425e-01 1.17141283e+00
2.70087868e-01 3.07847053e-01 5.80796957e-01 -1.16601813e+00
2.06413925e-01 -1.11043894e+00 -9.31061685e-01 5.80797017e-01
4.97126013e-01 -3.90129954e-01 -2.75667235e-02 -2.45444685e-01] | [6.742462635040283, 1.9540774822235107] |
f26a386b-214c-4596-a526-5fdc18efb4f3 | bias-in-automated-image-colorization-metrics | 2202.08143 | null | https://arxiv.org/abs/2202.08143v1 | https://arxiv.org/pdf/2202.08143v1.pdf | Bias in Automated Image Colorization: Metrics and Error Types | We measure the color shifts present in colorized images from the ADE20K dataset, when colorized by the automatic GAN-based DeOldify model. We introduce fine-grained local and regional bias measurements between the original and the colorized images, and observe many colorization effects. We confirm a general desaturation effect, and also provide novel observations: a shift towards the training average, a pervasive blue shift, different color shifts among image categories, and a manual categorization of colorization errors in three classes. | ['Doina Bucur', 'Floris Weers', 'Frank Stapel'] | 2022-02-16 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 1.25348419e-01 -4.10545170e-01 -8.23617801e-02 -4.92952526e-01
-7.79570818e-01 -1.03209460e+00 7.47114658e-01 -4.04082716e-01
-4.69552457e-01 7.19909251e-01 8.70552808e-02 -5.44482796e-03
2.62887627e-01 -3.68856132e-01 -7.22832799e-01 -9.94403064e-01
1.56136245e-01 1.94997817e-01 1.00408174e-01 -9.55611318e-02
2.06535801e-01 5.50498068e-01 -1.10559213e+00 4.71412003e-01
8.75973821e-01 1.04395795e+00 -2.62218326e-01 9.98301387e-01
5.00843599e-02 7.20954478e-01 -6.58893228e-01 -4.11645055e-01
6.02811933e-01 -7.06130564e-01 -6.39756680e-01 2.97439039e-01
1.29070950e+00 -5.03015757e-01 -2.86852509e-01 1.25444889e+00
3.34475309e-01 6.04593679e-02 7.06025302e-01 -1.39446390e+00
-1.11895204e+00 3.06292504e-01 -1.11428320e+00 3.77691299e-01
-1.24146687e-02 4.15327966e-01 7.90180266e-01 -8.82877648e-01
7.93419719e-01 1.14842284e+00 5.58874071e-01 7.39584029e-01
-1.57542646e+00 -7.25104570e-01 2.47572035e-01 3.76257241e-01
-1.45363474e+00 -3.37163389e-01 7.11447835e-01 -5.54666102e-01
4.48893279e-01 4.11728948e-01 7.57094920e-01 1.08730054e+00
1.71545386e-01 5.30416787e-01 2.01708913e+00 -5.16318023e-01
4.31843489e-01 1.58251882e-01 4.67480123e-02 5.42305470e-01
2.93799847e-01 2.87521064e-01 -6.70521021e-01 8.26445073e-02
5.97049236e-01 -1.37789443e-01 -4.27128643e-01 -2.98667490e-01
-1.24221885e+00 4.22169298e-01 7.82773793e-01 -8.02060962e-02
-1.80380300e-01 4.66631323e-01 9.36802104e-02 2.07526624e-01
6.60841286e-01 5.04052162e-01 -5.97562015e-01 -1.65573955e-01
-1.18286681e+00 -5.46914004e-02 2.46269837e-01 1.09332716e+00
1.16946006e+00 2.89637744e-01 -7.66977072e-02 7.62378752e-01
-1.60674334e-01 9.60499108e-01 3.12408268e-01 -1.29311299e+00
1.65390953e-01 2.98121542e-01 2.58736517e-02 -7.57795095e-01
-2.66953170e-01 -1.36331826e-01 -7.68124223e-01 7.00712979e-01
6.15289807e-01 -1.98256895e-01 -1.45278013e+00 1.69796312e+00
-5.82903950e-03 -2.02770665e-01 -2.63428122e-01 1.00609875e+00
4.61465329e-01 3.32303643e-01 2.17712149e-01 2.25951821e-01
1.08923602e+00 -1.06185424e+00 -5.29235542e-01 -3.44461620e-01
2.25361839e-01 -6.59429789e-01 1.30625367e+00 5.58941126e-01
-9.18991208e-01 -5.81763387e-01 -1.15217984e+00 2.95994598e-02
-6.91289246e-01 2.85783440e-01 5.59160054e-01 8.77198279e-01
-1.42754316e+00 6.85412884e-01 -5.16521454e-01 -5.34638524e-01
4.14222568e-01 -2.86094069e-01 -2.90212214e-01 -3.86873364e-01
-5.95547318e-01 7.07660258e-01 3.87019038e-01 2.72044033e-01
-9.84923601e-01 -9.88687396e-01 -3.14199775e-01 -4.68884766e-01
1.04987480e-01 -2.32139044e-02 8.05214643e-01 -1.58655989e+00
-1.30944085e+00 1.30532014e+00 -3.14119607e-02 -1.16627396e-03
7.07761228e-01 -1.15546748e-01 -8.59730005e-01 1.68534443e-01
-3.00123960e-01 1.02446878e+00 9.83081818e-01 -1.85923564e+00
-7.19745278e-01 -2.91204929e-01 -8.02425891e-02 1.92041039e-01
1.36967516e-02 -1.70959115e-01 -6.30286574e-01 -8.87881815e-01
-2.52832891e-03 -1.32354820e+00 5.78888990e-02 9.80715305e-02
-8.11412871e-01 4.95805949e-01 5.11020660e-01 -9.49849427e-01
8.78839016e-01 -2.12257957e+00 -5.16812541e-02 4.16312367e-01
5.54759741e-01 -2.42405280e-01 -3.93335491e-01 -9.04940814e-02
-4.27870870e-01 2.52519727e-01 -2.37022579e-01 -9.20917373e-03
-3.04678101e-02 1.43403307e-01 -1.66910976e-01 7.14338243e-01
1.71930909e-01 9.84084427e-01 -8.31385314e-01 -2.41717100e-01
3.39907944e-01 2.15716630e-01 -4.08354700e-01 7.67360926e-02
2.09365468e-02 3.02027583e-01 2.67801583e-01 8.80631924e-01
1.40593791e+00 7.74302334e-02 3.19240652e-02 -6.44453943e-01
1.07569164e-02 -3.07383001e-01 -7.62205899e-01 1.48206747e+00
-1.73180997e-01 1.07045233e+00 1.32052094e-01 -3.70920867e-01
5.36117256e-01 -2.81686902e-01 2.51216769e-01 -1.01640689e+00
-1.36792198e-01 -2.63061076e-02 -3.21625844e-02 4.06102240e-02
8.40492845e-01 -3.07873338e-01 1.37091056e-01 4.60516453e-01
1.98148757e-01 -3.58528018e-01 -3.53884138e-03 3.22021991e-01
6.87291682e-01 1.59032896e-01 -4.48025763e-01 -6.94375813e-01
-4.62154076e-02 1.20570168e-01 3.79399508e-01 7.89914429e-01
-2.98937470e-01 1.25738370e+00 5.78697622e-01 -2.77645051e-01
-1.17570281e+00 -1.21237350e+00 1.84046589e-02 1.27545702e+00
3.57699335e-01 -2.06219494e-01 -8.18204463e-01 -9.28762615e-01
1.18613467e-01 7.03562915e-01 -1.24503946e+00 -4.46038216e-01
-2.47351453e-01 -1.07985604e+00 4.06544775e-01 6.38394356e-01
5.08726180e-01 -7.77114153e-01 -1.67133272e-01 -5.08339167e-01
-9.58765894e-02 -1.08131659e+00 -6.06359243e-01 4.58142847e-01
-5.71461856e-01 -1.10301042e+00 -7.91884303e-01 -3.72088015e-01
8.49342644e-01 2.75043011e-01 1.64964581e+00 1.05777439e-02
-4.42886919e-01 5.80808699e-01 -4.96933728e-01 -2.68785357e-01
-5.68508804e-01 -2.66186833e-01 -2.32469216e-01 -1.91828370e-01
2.19842404e-01 8.73397067e-02 -1.04946423e+00 4.53027695e-01
-8.83909523e-01 1.22312799e-01 4.63491410e-01 5.53704441e-01
4.77280229e-01 -1.56701967e-01 -5.59622794e-02 -9.36183214e-01
2.38317221e-01 -1.66480184e-01 -5.82017303e-01 4.71241027e-01
-1.07926273e+00 1.63903072e-01 3.53625536e-01 -1.61897033e-01
-1.30034792e+00 -2.54124433e-01 3.00530195e-01 -3.30617934e-01
-3.79548699e-01 -1.99399158e-01 -6.32358938e-02 -4.22879785e-01
8.41645777e-01 9.10030976e-02 -3.35778683e-01 -2.78515488e-01
9.83702779e-01 3.20373654e-01 7.89702356e-01 -7.92129934e-01
9.63311493e-01 6.95987940e-01 -4.16299820e-01 -6.83208883e-01
-7.04455853e-01 -9.02864411e-02 -7.18999743e-01 -4.94983613e-01
1.08466268e+00 -9.81687665e-01 -2.30101272e-01 8.40696275e-01
-6.97212338e-01 -1.03802335e+00 -5.26997566e-01 2.95482799e-02
-1.86076790e-01 1.82018533e-01 -5.16939759e-01 -2.87721664e-01
-8.87713283e-02 -9.25969958e-01 9.11132812e-01 5.36975384e-01
2.38910783e-02 -9.99873340e-01 1.88138932e-01 2.06053015e-02
4.02175725e-01 3.77636343e-01 1.02527773e+00 1.92856774e-01
-3.39141816e-01 1.38934031e-01 -8.44155073e-01 6.06508136e-01
4.23522413e-01 6.06604099e-01 -1.15321743e+00 -5.63323975e-01
-7.95139253e-01 -3.38524371e-01 1.24981678e+00 4.42689598e-01
1.21550655e+00 1.81725517e-01 4.17339709e-03 1.00430226e+00
1.80649269e+00 1.68573052e-01 7.22684443e-01 3.73908013e-01
9.78264153e-01 2.92313606e-01 2.31346622e-01 1.15362115e-01
-5.17036207e-03 4.68558341e-01 3.16042989e-01 -8.16050887e-01
-7.14772344e-01 -1.78980649e-01 2.74667710e-01 4.06192094e-01
-2.36030668e-01 -1.29544005e-01 -6.72453225e-01 4.80903268e-01
-1.28913677e+00 -6.50987685e-01 -1.08974837e-01 2.16854811e+00
8.57566833e-01 -1.66029006e-01 1.01676725e-01 -3.35563123e-01
7.68923461e-01 2.62573898e-01 -8.21900010e-01 -3.75971556e-01
-6.06830537e-01 2.63225406e-01 1.21853924e+00 3.41649890e-01
-8.32735896e-01 1.10943866e+00 8.92956066e+00 5.87090671e-01
-1.42081892e+00 -4.31473665e-02 1.16973937e+00 -1.32022932e-01
-4.58859652e-01 -1.29533142e-01 -1.65349945e-01 4.43844855e-01
5.78416586e-01 7.14502260e-02 9.39889312e-01 5.25468409e-01
-2.09115148e-01 -3.49329114e-01 -9.97242630e-01 1.24198222e+00
1.51181713e-01 -9.73461509e-01 -1.69592705e-02 -3.29334065e-02
1.27840352e+00 4.09261853e-01 4.31202769e-01 -1.75280776e-02
8.55861962e-01 -8.81006658e-01 1.17784178e+00 5.69290280e-01
1.35763001e+00 -5.45792401e-01 1.95510104e-01 -8.04645479e-01
-9.21508074e-01 7.69171715e-02 -5.10905802e-01 4.22807366e-01
-3.33205104e-01 6.86973989e-01 -4.62541044e-01 3.83753151e-01
1.15189183e+00 8.49292338e-01 -1.45643198e+00 6.78567350e-01
-4.91491646e-01 8.54241729e-01 -2.54792813e-02 4.49589610e-01
2.99496446e-02 -6.68552339e-01 1.51281580e-02 1.34701097e+00
1.46026358e-01 -3.82863075e-01 -4.55858916e-01 1.03985023e+00
-1.69672638e-01 -4.95007455e-01 2.22766940e-02 -2.84913424e-02
2.32770666e-01 1.59401262e+00 -9.89042342e-01 -5.52572966e-01
-3.26284021e-01 1.66729164e+00 5.15086614e-02 1.02735126e+00
-1.18286121e+00 -3.18898618e-01 8.37719440e-01 -2.24139705e-01
1.98255822e-01 -1.78209022e-01 -4.09310251e-01 -1.28890324e+00
-7.83506781e-02 -8.24013174e-01 3.57580930e-01 -1.36983931e+00
-1.35736370e+00 4.28909510e-01 -7.85593837e-02 -1.01256585e+00
1.75109714e-01 -8.28017771e-01 -6.47138178e-01 6.47533894e-01
-1.50841105e+00 -9.98299658e-01 -9.35940266e-01 6.30778015e-01
-1.38221547e-01 5.33832163e-02 5.21118164e-01 3.70420992e-01
-5.62758327e-01 7.91002512e-01 5.66068411e-01 1.12094671e-01
1.34848082e+00 -1.81103957e+00 4.41610038e-01 1.06345212e+00
1.19689152e-01 1.45238042e-01 7.67219424e-01 -2.65097588e-01
-1.22734118e+00 -1.25411415e+00 -3.26428741e-01 -7.09430993e-01
4.06877637e-01 -3.80599290e-01 -6.00652993e-01 7.55828798e-01
6.08487666e-01 2.82943964e-01 3.42177540e-01 1.09742567e-01
-7.54461884e-01 -3.87180209e-01 -1.10693479e+00 4.87943351e-01
9.20117915e-01 -6.86404824e-01 8.04443210e-02 1.26147969e-02
2.80857027e-01 -2.09465131e-01 -5.34438431e-01 -4.83239293e-02
8.03951204e-01 -1.15083373e+00 8.70773256e-01 -5.38058639e-01
4.94871527e-01 -4.26625818e-01 -1.98195189e-01 -1.93234336e+00
-5.70357919e-01 -8.54411498e-02 5.34544766e-01 1.26120508e+00
5.05563378e-01 -4.58100379e-01 9.41896617e-01 8.76321912e-01
9.95033011e-02 9.64980870e-02 -7.71967173e-01 -5.98974645e-01
4.98420596e-01 -1.99739859e-01 4.56587672e-01 1.06011367e+00
-4.93192732e-01 -1.68738648e-01 -3.94127876e-01 -4.54696082e-02
7.53780127e-01 5.38597815e-02 6.81556284e-01 -7.88316429e-01
-6.51937500e-02 -5.75422227e-01 -2.45124146e-01 -6.50664926e-01
-1.25983328e-01 -7.62033701e-01 2.55008280e-01 -1.21122885e+00
5.37692964e-01 -5.48395276e-01 -8.87500644e-01 4.65729117e-01
-3.34001929e-01 9.64415789e-01 2.34846950e-01 -1.30662378e-02
-7.75154054e-01 3.35535675e-01 1.15029693e+00 -5.83392918e-01
1.78338587e-01 -8.24787438e-01 -8.17302465e-01 5.18522918e-01
7.55171359e-01 -3.59864295e-01 -1.67800441e-01 -6.84637845e-01
5.15321136e-01 -7.49340355e-01 5.38766682e-01 -1.22252059e+00
-4.23269749e-01 -3.62501502e-01 1.02348733e+00 -3.36649120e-01
4.06194851e-02 -8.72060478e-01 1.47061974e-01 3.51099879e-01
-3.12292159e-01 1.66896790e-01 3.16517085e-01 4.71523076e-01
5.49528562e-02 3.85132551e-01 1.26009417e+00 -8.99252445e-02
-1.24372053e+00 1.73279926e-01 -3.86443347e-01 3.19739670e-01
9.05878723e-01 -2.62609303e-01 -6.29240692e-01 -2.73718983e-01
-5.38234651e-01 -1.43337846e-01 1.10764885e+00 5.53747058e-01
1.72464371e-01 -1.60996461e+00 -5.73814988e-01 2.56121516e-01
5.92477739e-01 -3.90780777e-01 5.12186170e-01 6.58686221e-01
-8.48457634e-01 -1.90180153e-01 -4.69325095e-01 -5.89861512e-01
-1.05328417e+00 4.03053582e-01 6.84446216e-01 3.63634169e-01
-4.46167707e-01 1.01049602e+00 3.60575825e-01 -2.51208097e-01
-2.47247115e-01 -6.01294398e-01 4.11928356e-01 -3.65031674e-03
4.25920457e-01 5.80836892e-01 3.16432387e-01 -4.02643681e-01
-4.84089822e-01 8.49792361e-01 1.06271490e-01 -1.70519203e-01
9.43636358e-01 -4.64086920e-01 -2.39236042e-01 4.68619227e-01
1.10043168e+00 1.61377534e-01 -1.64445472e+00 -1.30365536e-01
-3.67991894e-01 -7.51102805e-01 1.78598493e-01 -1.33852172e+00
-1.67815650e+00 5.75511158e-01 1.29490888e+00 1.39515251e-01
1.20695353e+00 1.16592839e-01 2.56048352e-01 2.64772624e-02
1.83766991e-01 -1.57313681e+00 2.24670414e-02 8.09947401e-02
8.07527483e-01 -1.41454673e+00 2.39436775e-01 3.39633487e-02
-8.74040544e-01 9.57078516e-01 8.72408688e-01 -1.65405065e-01
4.30429786e-01 2.16916576e-01 6.68680727e-01 -1.13950625e-01
-5.43998361e-01 2.86251325e-02 2.20665336e-01 9.62408662e-01
3.60776573e-01 3.60152453e-01 2.67554671e-01 2.54149712e-03
-1.68840855e-01 -4.03278857e-01 7.82021701e-01 4.30242926e-01
1.27121344e-01 -6.14575505e-01 -4.52110291e-01 3.76655102e-01
2.16255151e-02 -1.77466944e-01 -8.49119246e-01 8.35258007e-01
2.54658192e-01 8.70194197e-01 5.03816724e-01 -6.94730520e-01
2.76774406e-01 8.04494172e-02 8.16618502e-01 1.08865269e-01
-1.64498448e-01 -2.11662322e-01 -1.24905147e-01 -9.13847446e-01
-5.22609472e-01 -3.51498008e-01 -8.05502117e-01 -5.26736617e-01
-2.85439491e-01 -8.65145996e-02 7.69057512e-01 4.61051226e-01
2.54634470e-01 6.90203488e-01 7.53016412e-01 -1.04734409e+00
2.21136346e-01 -9.58688676e-01 -1.10435891e+00 1.04004622e+00
5.76376975e-01 -4.36648905e-01 -7.86987424e-01 3.71041596e-01] | [11.29468059539795, -1.2415286302566528] |
9a8d28fd-df4f-4c26-bc23-6cc76ec47f17 | constructing-effective-personalized-policies | 1612.08082 | null | http://arxiv.org/abs/1612.08082v3 | http://arxiv.org/pdf/1612.08082v3.pdf | Constructing Effective Personalized Policies Using Counterfactual Inference from Biased Data Sets with Many Features | This paper proposes a novel approach for constructing effective personalized
policies when the observed data lacks counter-factual information, is biased
and possesses many features. The approach is applicable in a wide variety of
settings from healthcare to advertising to education to finance. These settings
have in common that the decision maker can observe, for each previous instance,
an array of features of the instance, the action taken in that instance, and
the reward realized -- but not the rewards of actions that were not taken: the
counterfactual information. Learning in such settings is made even more
difficult because the observed data is typically biased by the existing policy
(that generated the data) and because the array of features that might affect
the reward in a particular instance -- and hence should be taken into account
in deciding on an action in each particular instance -- is often vast. The
approach presented here estimates propensity scores for the observed data,
infers counterfactuals, identifies a (relatively small) number of features that
are (most) relevant for each possible action and instance, and prescribes a
policy to be followed. Comparison of the proposed algorithm against the
state-of-art algorithm on actual datasets demonstrates that the proposed
algorithm achieves a significant improvement in performance. | ['Mihaela van der Schaar', 'William R. Zame', 'Qiaojun Feng', 'Onur Atan'] | 2016-12-23 | null | null | null | null | ['counterfactual-inference'] | ['miscellaneous'] | [ 4.52355981e-01 3.29301476e-01 -8.85418594e-01 -4.93213177e-01
-5.84987104e-01 -3.20619285e-01 7.86814094e-01 5.08480668e-01
-8.57585847e-01 1.29055202e+00 5.23682237e-01 -4.82433349e-01
-6.20787859e-01 -8.77511203e-01 -8.18290651e-01 -9.15845096e-01
-2.52559811e-01 8.34060609e-01 -1.30321518e-01 2.30130136e-01
4.57194567e-01 3.98389250e-01 -1.67881763e+00 1.14112131e-01
7.78630733e-01 9.44552541e-01 1.36402264e-01 3.76907051e-01
2.23203599e-02 8.03572953e-01 -4.47925448e-01 -5.54365516e-01
3.33839625e-01 -2.47614309e-01 -4.84751284e-01 3.01582873e-01
4.17907476e-01 -5.89740336e-01 -1.27947599e-01 9.85680699e-01
2.60601401e-01 4.01196182e-01 9.29230988e-01 -1.07328689e+00
-3.05882812e-01 8.01869392e-01 -4.20057893e-01 2.49260515e-01
1.32560953e-01 2.38389447e-01 1.10927963e+00 -1.77600794e-03
4.98500466e-01 1.26612699e+00 -1.80821195e-01 4.62189734e-01
-1.24516809e+00 -7.22649693e-01 4.69981074e-01 1.67594135e-01
-4.75489825e-01 -2.95253813e-01 7.54219651e-01 -4.54313606e-01
5.62747478e-01 2.14664415e-01 6.85624599e-01 1.18381155e+00
3.73809725e-01 6.35131598e-01 1.48787713e+00 -3.86544108e-01
8.10268760e-01 4.52913761e-01 -9.31078494e-02 2.88620174e-01
3.88207495e-01 7.58573472e-01 -1.61878645e-01 -6.39116764e-01
4.23491448e-01 4.77995038e-01 -1.31392768e-02 -4.43657517e-01
-1.00948656e+00 9.92466629e-01 1.65524200e-01 3.09944227e-02
-1.13502908e+00 -6.67127594e-02 2.45305926e-01 2.67176956e-01
4.47794944e-02 5.80812037e-01 -7.81401455e-01 -3.96494903e-02
-7.93497086e-01 7.63159513e-01 5.98375440e-01 3.46715152e-01
8.01929295e-01 -1.81603804e-01 -5.21340907e-01 4.39520270e-01
-1.09938167e-01 4.21521008e-01 7.49615252e-01 -1.03318191e+00
7.69937158e-01 5.19823968e-01 6.83627009e-01 -6.51409388e-01
-5.44652268e-02 -2.82178849e-01 -4.73681986e-01 3.71097326e-01
6.76851869e-01 -3.89061362e-01 -7.66876101e-01 1.91847610e+00
5.06722867e-01 1.21729828e-01 8.38156343e-02 6.85115755e-01
-1.20978415e-01 4.28147584e-01 3.58127266e-01 -8.18576097e-01
1.17764640e+00 -2.47781977e-01 -5.71266234e-01 -3.67218494e-01
2.22725466e-01 -4.55563515e-01 7.87383497e-01 3.67115647e-01
-9.81894672e-01 -1.90634638e-01 -6.62789345e-01 7.25258172e-01
-8.38215426e-02 -9.05943140e-02 8.37577045e-01 4.89943653e-01
-3.59643012e-01 9.54678357e-01 -2.23134771e-01 -7.64653683e-02
6.17480338e-01 5.25724411e-01 -1.54677212e-01 -5.10790106e-03
-1.17470753e+00 7.92220414e-01 3.60490978e-01 -4.25622433e-01
-1.10740161e+00 -5.47833085e-01 -2.53228247e-01 4.98725593e-01
1.12927151e+00 -6.36608183e-01 1.42025673e+00 -1.31758988e+00
-1.16521454e+00 3.69335651e-01 2.78795208e-03 -8.36415291e-01
9.58118737e-01 7.61864632e-02 -3.70670348e-01 -3.53998095e-01
2.19683915e-01 1.57968819e-01 1.05079365e+00 -6.49830818e-01
-1.23628294e+00 -8.19182098e-01 2.86746293e-01 4.47125494e-01
-2.42811125e-02 -2.61829227e-01 3.91505420e-01 -5.10882556e-01
-3.25900435e-01 -8.83739769e-01 -6.71343088e-01 -4.73870158e-01
-3.59407365e-01 -2.18965605e-01 3.64070952e-01 -7.60591403e-02
1.05587292e+00 -1.71389127e+00 -2.03253061e-01 3.30680519e-01
-2.02179566e-01 -5.72057106e-02 2.97542989e-01 4.39047635e-01
-2.02204719e-01 -1.01721939e-02 -4.46864292e-02 3.22025180e-01
-1.11973464e-01 1.68292508e-01 -5.87746561e-01 4.64197755e-01
-3.10812503e-01 1.81417495e-01 -7.95656383e-01 -2.03901753e-01
3.59308660e-01 -3.14384431e-01 -5.41649759e-01 2.69491702e-01
-3.96430969e-01 4.06218678e-01 -1.03449059e+00 9.54748169e-02
3.37658346e-01 -1.99652612e-02 3.97070378e-01 2.67951250e-01
-2.49363288e-01 4.73633140e-01 -1.42492425e+00 8.13055158e-01
-3.61822963e-01 5.90882031e-03 -5.60772061e-01 -1.34721065e+00
5.32666564e-01 5.38414717e-01 6.36868119e-01 -5.68501592e-01
1.41068667e-01 1.05508678e-01 2.30880588e-01 -6.25873268e-01
2.05359086e-01 -4.16310459e-01 -1.07694492e-01 4.86249238e-01
-1.93932399e-01 2.52682477e-01 2.55993724e-01 1.92182083e-02
9.28154051e-01 -1.52991995e-01 1.11626709e+00 -9.31716263e-02
4.60193306e-01 4.84622531e-02 9.46316361e-01 1.16391420e+00
-1.78794801e-01 -1.58272028e-01 6.77790344e-01 -5.01206100e-01
-9.70476270e-01 -7.75484800e-01 -2.05894420e-03 9.40784812e-01
-2.30705842e-01 5.26288748e-01 -3.22331190e-01 -9.56493437e-01
4.77137893e-01 1.13193130e+00 -9.51329172e-01 -1.71660390e-02
-2.00729951e-01 -8.02496493e-01 -3.82321268e-01 1.67225361e-01
2.94981182e-01 -1.21799994e+00 -8.90404820e-01 5.04695892e-01
1.59469098e-01 -6.35592818e-01 -3.83779615e-01 -1.10256501e-01
-1.19020438e+00 -1.35902429e+00 -4.62320000e-01 -2.10952647e-02
5.63004851e-01 -1.83149800e-01 9.47474420e-01 -4.30199981e-01
-1.60602972e-01 5.63924313e-02 1.40408985e-02 -8.34160328e-01
-3.04720044e-01 -4.88574177e-01 2.50991851e-01 3.60088408e-01
4.46305752e-01 -3.90933633e-01 -8.18183184e-01 -2.22952545e-01
-6.66838884e-01 -3.13275307e-01 8.44944477e-01 9.50581551e-01
3.92201781e-01 3.71000588e-01 7.98327327e-01 -1.56161213e+00
6.24490678e-01 -6.24883056e-01 -9.29695010e-01 3.62244040e-01
-1.05175889e+00 3.38784844e-01 8.18735421e-01 -5.98102868e-01
-1.51354289e+00 -1.98516659e-02 3.36887419e-01 2.08811518e-02
-4.98370945e-01 1.69117466e-01 -4.50168192e-01 7.91401863e-01
5.97164154e-01 7.63640553e-02 -1.23275995e-01 -5.34489155e-01
3.07224333e-01 6.53474033e-01 2.85832524e-01 -7.40200996e-01
3.06159854e-01 4.08562720e-01 2.50660986e-01 -2.93540746e-01
-8.52103710e-01 -2.12459415e-01 -4.39585298e-02 7.24481931e-03
4.03826684e-01 -7.75629759e-01 -1.05878019e+00 7.00212969e-03
-3.58712375e-01 9.50550139e-02 -5.48174679e-01 8.94478738e-01
-6.73065782e-01 -1.70712918e-02 1.47626236e-01 -1.38116276e+00
1.76530540e-01 -1.21081495e+00 3.57015252e-01 4.42995876e-01
-1.52084783e-01 -9.15523648e-01 1.45635381e-02 3.58089745e-01
8.63210708e-02 2.26655081e-01 1.22472525e+00 -1.01717126e+00
-7.13424444e-01 -3.90472651e-01 2.55983740e-01 1.01040944e-01
3.11857522e-01 -8.01390037e-02 -7.23702610e-01 -2.93512970e-01
3.88831086e-02 -1.02862850e-01 4.28485632e-01 1.02107990e+00
1.16178119e+00 -9.34518516e-01 -3.73871475e-01 -2.45403022e-01
1.53099847e+00 6.55885160e-01 3.60804588e-01 3.75800997e-01
-1.04512110e-01 6.78660393e-01 1.01559186e+00 8.53710949e-01
2.25314155e-01 8.11203063e-01 5.30133605e-01 1.88266993e-01
7.21791029e-01 -2.61001468e-01 2.33830348e-01 -4.43327129e-01
-9.63733345e-02 1.43608406e-01 -1.45667404e-01 8.63068104e-01
-1.96578479e+00 -1.36721587e+00 2.56757379e-01 2.82574081e+00
8.21217775e-01 1.45373613e-01 4.75419641e-01 7.65262023e-02
7.13531494e-01 1.36364877e-01 -8.96316528e-01 -6.72316849e-01
2.94445693e-01 2.01369915e-02 8.24523389e-01 4.69086289e-01
-9.24429059e-01 3.85459870e-01 5.66507149e+00 6.76417470e-01
-9.53220904e-01 -1.73090905e-01 9.14291799e-01 -3.47720653e-01
-2.82726109e-01 2.53002375e-01 -7.79910088e-01 7.23088562e-01
1.00562608e+00 -6.01273775e-01 3.69705081e-01 1.01834416e+00
6.15707934e-01 -5.56535244e-01 -1.42006195e+00 2.99415559e-01
-5.91291070e-01 -1.29232991e+00 2.54127234e-01 5.15460551e-01
9.25750911e-01 -4.71443504e-01 2.72275478e-01 2.50027269e-01
7.67881513e-01 -5.52424014e-01 5.41810930e-01 6.10789180e-01
3.76281053e-01 -9.41697240e-01 6.03580594e-01 7.87446797e-01
-3.06195080e-01 -7.87037551e-01 -3.21183175e-01 -4.59755421e-01
-3.95948529e-01 8.26355577e-01 -1.23871291e+00 3.16708893e-01
3.60088438e-01 2.59397000e-01 1.13556525e-02 9.65490043e-01
-2.36034617e-01 7.00888932e-01 1.65210143e-01 -2.02505201e-01
4.21842009e-01 -1.63737908e-01 5.38194895e-01 5.30366719e-01
4.02676165e-01 9.28369686e-02 1.30133644e-01 5.41039050e-01
2.11986005e-02 3.00291508e-01 -8.14167857e-01 4.68770079e-02
5.57491660e-01 8.83236408e-01 -2.20087558e-01 -6.99135363e-01
-4.00048345e-01 4.18803841e-01 1.78786457e-01 3.44524086e-01
-3.69046599e-01 1.55634552e-01 8.13625455e-01 1.56215608e-01
2.08229288e-01 8.22812736e-01 -7.41215199e-02 -8.68624508e-01
-1.73903763e-01 -9.32786703e-01 9.02568758e-01 -3.33607346e-01
-1.29019320e+00 -4.27929573e-02 3.10285121e-01 -1.05065131e+00
-8.27132583e-01 -3.59411061e-01 -5.66260755e-01 1.01374507e+00
-1.42290640e+00 -2.74595648e-01 5.31940937e-01 4.26331550e-01
5.26015222e-01 -3.51996183e-01 7.28176713e-01 -1.22569747e-01
-3.42277318e-01 2.06304610e-01 4.16252345e-01 -2.24924996e-01
4.37684208e-01 -1.24417007e+00 -3.13603222e-01 5.28903246e-01
-9.96841118e-02 3.52243096e-01 9.56448674e-01 -6.79246962e-01
-8.59217107e-01 -9.45359230e-01 1.01134300e+00 -1.11099571e-01
5.20377457e-01 2.34904453e-01 -5.79118967e-01 6.78593218e-01
-1.50448978e-01 -4.09625024e-01 5.43185115e-01 1.65041536e-01
1.15848236e-01 -4.40216720e-01 -1.62533689e+00 7.41150320e-01
4.96078193e-01 -6.05438128e-02 -8.85743439e-01 3.03131849e-01
1.59068435e-01 -7.71991313e-02 -4.93732721e-01 9.05469730e-02
7.12182283e-01 -1.20428145e+00 8.15673828e-01 -1.20437431e+00
4.03336734e-01 2.13088483e-01 -3.94532196e-02 -1.59952664e+00
-3.18516582e-01 -6.19530320e-01 7.55756423e-02 9.10870373e-01
4.38710004e-01 -8.53043735e-01 8.15467596e-01 1.09644568e+00
5.60463727e-01 -8.34916651e-01 -1.17421472e+00 -4.50441539e-01
-2.38727957e-01 -1.99314598e-02 1.22018206e+00 8.76149297e-01
-1.21967442e-01 -7.31147453e-02 -5.81973970e-01 1.35306790e-01
9.05490279e-01 6.85434937e-01 6.42855167e-01 -1.30564034e+00
-4.66066450e-01 -3.18965852e-01 -3.99083681e-02 -5.58176637e-01
1.38700217e-01 -3.53009313e-01 -3.20541024e-01 -1.00214088e+00
5.04685521e-01 -4.55828071e-01 -6.21927679e-01 2.75546014e-01
-6.03161871e-01 -7.96091080e-01 9.45429206e-02 -2.14142613e-02
-8.59251525e-03 1.94940448e-01 1.27293837e+00 5.79472184e-02
-2.48885393e-01 7.21730947e-01 -1.04494214e+00 7.60416508e-01
6.57474279e-01 -7.61963367e-01 -4.73357618e-01 3.70311588e-01
-2.39581302e-01 7.51366794e-01 2.79332280e-01 -4.62382793e-01
-1.45856559e-01 -9.77716386e-01 5.07795632e-01 -1.92619830e-01
7.13784471e-02 -1.17939889e+00 2.38208488e-01 6.86173797e-01
-9.74322557e-01 -1.60319179e-01 -4.25577968e-01 9.57136393e-01
1.11158133e-01 -4.95708615e-01 5.60258627e-01 -5.97545147e-01
-4.60650742e-01 4.22025502e-01 -2.15609774e-01 3.48606147e-02
1.07113016e+00 5.54655604e-02 -6.34246022e-02 -5.59274614e-01
-7.39051461e-01 9.00930762e-02 3.63405287e-01 2.11980864e-01
2.74237841e-01 -1.11751878e+00 -6.99765623e-01 -5.32120746e-03
1.72234271e-02 -4.39858466e-01 3.70512456e-01 4.25471306e-01
5.33196628e-01 4.90140647e-01 -2.93903053e-01 2.00900123e-01
-1.09029496e+00 6.61358297e-01 1.26822963e-01 -8.55182946e-01
-3.66100609e-01 2.37200588e-01 4.40960407e-01 1.11213259e-01
1.52490258e-01 -1.29937842e-01 -4.70825702e-01 2.70742804e-01
4.63986367e-01 5.90771854e-01 -9.19763967e-02 -4.00486231e-01
1.53344376e-02 -1.11629494e-01 -3.78379792e-01 -1.72498599e-01
1.47625792e+00 -1.64838172e-02 2.18150362e-01 5.36588967e-01
7.59914517e-01 -1.48707658e-01 -1.50047064e+00 -4.64615375e-01
-1.65500324e-02 -1.06743097e+00 5.29691949e-02 -1.02980769e+00
-8.70333672e-01 4.95156139e-01 7.10219562e-01 3.14212382e-01
1.10046446e+00 -3.23619962e-01 1.48110837e-01 2.32120633e-01
4.87551242e-01 -1.37358141e+00 -3.40557605e-01 -2.76029348e-01
6.25575006e-01 -1.27197230e+00 2.48880222e-01 1.40310928e-01
-9.52429712e-01 6.02338195e-01 3.20468307e-01 -2.64808893e-01
4.37211663e-01 -3.47094804e-01 -1.67275995e-01 -1.15509391e-01
-1.05536771e+00 -1.80792764e-01 2.72157490e-01 3.21257889e-01
2.13215545e-01 6.00513399e-01 -5.97109139e-01 2.77948052e-01
-6.68680891e-02 2.36527085e-01 7.38563299e-01 7.37157106e-01
-3.50462824e-01 -1.10094321e+00 -5.66661775e-01 1.16158521e+00
-8.38138282e-01 2.37173319e-01 -7.76932910e-02 7.14103639e-01
2.76631266e-01 7.47074723e-01 8.34691748e-02 3.44345957e-01
6.12061024e-01 2.17926174e-01 3.36335957e-01 -5.30067086e-01
-3.96049827e-01 1.00349769e-01 1.53879270e-01 -6.64228439e-01
-5.54689884e-01 -1.16426289e+00 -9.09587920e-01 -3.42241600e-02
-1.06998660e-01 3.02735865e-01 5.43233633e-01 1.14590251e+00
1.80901498e-01 3.79053980e-01 1.16959178e+00 -5.40395737e-01
-1.28677976e+00 -7.62882531e-01 -9.98042941e-01 7.67894506e-01
5.49980521e-01 -9.83369231e-01 -5.09421170e-01 -3.10261399e-01] | [8.393753051757812, 5.441229820251465] |
77bc4c55-0356-4df6-92dc-ab4324434a36 | partial-feature-selection-and-alignment-for | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Fu_Partial_Feature_Selection_and_Alignment_for_Multi-Source_Domain_Adaptation_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Fu_Partial_Feature_Selection_and_Alignment_for_Multi-Source_Domain_Adaptation_CVPR_2021_paper.pdf | Partial Feature Selection and Alignment for Multi-Source Domain Adaptation | Multi-Source Domain Adaptation (MSDA), which dedicates to transfer the knowledge learned from multiple source domains to an unlabeled target domain, has drawn increasing attention in the research community. By assuming that the source and target domains share consistent key feature representations and identical label space, existing studies on MSDA typically utilize the entire union set of features from both the source and target domains to obtain the feature map and align the map for each category and domain. However, the default setting of MSDA may neglect the issue of "partialness", i.e., 1) a part of the features contained in the union set of multiple source domains may not present in the target domain; 2) the label space of the target domain may not completely overlap with the multiple source domains. In this paper, we unify the above two cases to a more generalized MSDA task as Multi-Source Partial Domain Adaptation (MSPDA). We propose a novel model termed Partial Feature Selection and Alignment (PFSA) to jointly cope with both MSDA and MSPDA tasks. Specifically, we firstly employ a feature selection vector based on the correlation among the features of multiple sources and target domains. We then design three effective feature alignment losses to jointly align the selected features by preserving the domain information of the data sample clusters in the same category and the discrimination between different classes. Extensive experiments on various benchmark datasets for both MSDA and MSPDA tasks demonstrate that our proposed PFSA approach remarkably outperforms the state-of-the-art MSDA and unimodal PDA methods. | ['Huimin Lu', 'Kai Zuo', 'Yanli Ji', 'Chao Ma', 'Zuo Cao', 'Xing Xu', 'Ming Zhang', 'Yangye Fu'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['partial-domain-adaptation'] | ['methodology'] | [ 2.52641708e-01 -3.13869864e-01 -3.95160973e-01 -4.98626500e-01
-6.87991083e-01 -6.87136650e-01 4.56420064e-01 5.51149398e-02
-1.44331977e-01 7.83590436e-01 2.93903351e-02 3.13379407e-01
-3.11226219e-01 -7.82290161e-01 -5.83969474e-01 -1.02050936e+00
3.71701658e-01 5.21586359e-01 3.09981972e-01 -5.18707512e-03
4.84498851e-02 2.47833982e-01 -1.29886281e+00 1.24317229e-01
1.14784527e+00 1.08132100e+00 3.67405862e-01 -1.82839036e-01
-3.48917425e-01 1.08558387e-01 -5.84733844e-01 -1.53486114e-02
5.00956595e-01 -5.17945766e-01 -6.28650784e-01 3.30824822e-01
2.13620752e-01 -9.09502134e-02 -2.91069150e-01 1.25953472e+00
3.12206000e-01 1.94969103e-01 9.14908230e-01 -1.73240566e+00
-8.84498835e-01 2.68056393e-01 -8.35602999e-01 1.80371433e-01
7.54662380e-02 -2.53663421e-01 9.02671278e-01 -8.46994638e-01
5.29717505e-01 1.17576885e+00 4.02126580e-01 4.09189880e-01
-1.24127114e+00 -8.96097302e-01 4.69345838e-01 1.23616397e-01
-1.44809377e+00 -2.68203259e-01 9.36719239e-01 -5.59602380e-01
3.58085424e-01 -1.35914713e-01 1.69294655e-01 9.92826879e-01
-6.75170124e-02 7.83013344e-01 1.17908382e+00 -3.81798685e-01
3.97268802e-01 5.20271778e-01 3.94414485e-01 2.76289731e-01
3.07792455e-01 -6.29501939e-02 -3.77137721e-01 -3.90698254e-01
7.39302456e-01 1.24882467e-01 -1.76211625e-01 -1.02036464e+00
-1.29922402e+00 8.88375759e-01 2.68550009e-01 2.22844675e-01
-4.00864452e-01 -7.11586237e-01 3.29698712e-01 4.66481119e-01
2.88569957e-01 1.64361581e-01 -5.43861866e-01 4.00078923e-01
-4.88819063e-01 3.01904440e-01 6.04968488e-01 1.25597453e+00
1.20024121e+00 -1.80718467e-01 -5.63966576e-03 1.07707489e+00
2.40382239e-01 7.36549914e-01 8.15425456e-01 -6.09138668e-01
7.31235564e-01 9.76593673e-01 2.24074095e-01 -9.04641032e-01
-2.55042076e-01 -4.41829830e-01 -8.13097775e-01 4.98408172e-03
5.80007374e-01 -1.57368526e-01 -7.43452191e-01 2.09014416e+00
5.84621847e-01 2.73979783e-01 4.72044230e-01 9.17228639e-01
6.81672335e-01 6.38086557e-01 -8.80467799e-03 -2.59260535e-01
1.00712538e+00 -7.85426497e-01 -3.86433572e-01 -4.58493173e-01
3.57730895e-01 -5.12896121e-01 9.48829114e-01 1.01242580e-01
-5.82235932e-01 -7.32693791e-01 -1.27819777e+00 2.53653198e-01
-3.37050557e-01 3.18095505e-01 3.65488201e-01 2.26635188e-01
-2.70184994e-01 2.73979723e-01 -4.62006480e-01 -3.46060157e-01
3.72319728e-01 3.76574457e-01 -7.87897646e-01 -4.33943778e-01
-1.25513434e+00 5.74861765e-01 5.35996556e-01 -4.03975308e-01
-6.29517972e-01 -6.92768216e-01 -8.00678492e-01 3.03333700e-02
4.63222414e-01 -5.00871599e-01 1.00792468e+00 -1.38576579e+00
-1.08326209e+00 7.31032789e-01 -2.71347553e-01 -4.91199084e-02
2.23898798e-01 -1.71283275e-01 -6.67647481e-01 -1.60884216e-01
4.71526325e-01 4.76740152e-01 9.28238630e-01 -1.18109369e+00
-9.69253480e-01 -4.93455201e-01 -2.97695905e-01 5.09209216e-01
-4.91514444e-01 -1.37708858e-01 -2.70413518e-01 -6.42950654e-01
2.20490456e-01 -8.33160281e-01 1.95412070e-01 -4.99399938e-02
-3.25132847e-01 -3.67804408e-01 9.92489874e-01 -3.22218835e-01
1.07282615e+00 -2.51421809e+00 3.53705168e-01 3.37557942e-01
-1.09242620e-02 2.90046751e-01 -3.30370486e-01 1.95610598e-01
-2.66438484e-01 -4.54516023e-01 -5.04488111e-01 1.02869868e-02
5.50587028e-02 3.75861049e-01 -5.56807756e-01 4.30757701e-01
2.42129192e-01 3.32742095e-01 -8.99749577e-01 -3.48219872e-01
1.12513214e-01 6.06815889e-02 -2.26285264e-01 3.09708089e-01
9.33421701e-02 4.45454359e-01 -7.16469884e-01 6.12970412e-01
9.91711318e-01 -2.26267561e-01 1.68218285e-01 -1.78484738e-01
1.86309233e-01 7.15802088e-02 -1.56213295e+00 1.60454226e+00
-8.81877542e-02 1.24771439e-01 -1.09965608e-01 -1.14857352e+00
1.24575317e+00 1.15620010e-01 8.02081704e-01 -6.37915850e-01
-8.60131532e-02 4.22870666e-01 7.94781595e-02 -9.65115875e-02
7.78620616e-02 -4.24679995e-01 -2.77563989e-01 4.60595220e-01
3.70666891e-01 3.18657517e-01 -2.72173081e-02 -1.35285884e-01
7.29319632e-01 2.67799525e-03 6.86252952e-01 -3.26929241e-01
7.11917043e-01 7.76469782e-02 1.06524420e+00 4.53093082e-01
-5.06747186e-01 7.12904692e-01 4.56012845e-01 -6.68419451e-02
-9.19614255e-01 -1.36199045e+00 -1.71211943e-01 1.11313033e+00
5.81515789e-01 1.28638506e-01 -5.63376904e-01 -1.20285845e+00
2.72850513e-01 5.37156403e-01 -6.14714265e-01 -4.23806041e-01
-2.91804731e-01 -7.22551823e-01 2.51557380e-01 5.89284420e-01
6.37814164e-01 -8.11720610e-01 -4.45959382e-02 1.76879540e-01
5.15559269e-03 -9.65606034e-01 -6.60942376e-01 2.84506857e-01
-6.15035653e-01 -1.22015226e+00 -8.61700118e-01 -1.05692732e+00
7.43518353e-01 4.70181584e-01 7.11275220e-01 -6.49544299e-01
3.30658704e-01 1.58799127e-01 -3.96974742e-01 -2.83150852e-01
-2.48197362e-01 8.08998346e-02 3.43158126e-01 3.57249290e-01
8.55097294e-01 -5.98642290e-01 -1.80535465e-01 6.80878878e-01
-8.84548604e-01 -1.84727356e-01 7.23067284e-01 8.74820411e-01
8.28628123e-01 2.42117465e-01 1.10695791e+00 -9.27503586e-01
5.51089585e-01 -1.09240687e+00 -3.55657935e-01 3.89217645e-01
-5.59070468e-01 5.10780737e-02 6.89125001e-01 -7.01169133e-01
-1.00698137e+00 2.87614971e-01 3.20787996e-01 -6.57222688e-01
-4.44193780e-01 4.34051126e-01 -8.69158983e-01 2.06204504e-01
5.08700550e-01 3.99931818e-01 2.47497946e-01 -6.21859014e-01
9.80737135e-02 8.05114925e-01 6.16678178e-01 -6.24854565e-01
9.51671660e-01 3.87594402e-01 -1.72635943e-01 -4.02197868e-01
-7.34212935e-01 -6.84099376e-01 -9.14250076e-01 3.67238760e-01
4.75250870e-01 -9.87634063e-01 1.40261307e-01 7.01658070e-01
-8.07254493e-01 9.29461494e-02 -4.47111279e-01 5.41926742e-01
-3.48361313e-01 5.24979472e-01 1.33382604e-01 -2.50462741e-01
-6.71991929e-02 -1.17757249e+00 8.26721966e-01 5.51643670e-01
-1.27883881e-01 -9.76514161e-01 1.59459114e-01 -1.88791603e-01
5.88959381e-02 1.58288702e-01 1.07136250e+00 -1.34388483e+00
1.98438503e-02 -1.05358586e-01 -3.04707497e-01 6.21697247e-01
8.28608811e-01 -4.62921232e-01 -7.80587196e-01 -3.90432328e-01
3.77144851e-02 -3.07907790e-01 6.46439016e-01 3.15308928e-01
6.61632180e-01 -8.64702240e-02 -4.94909704e-01 2.93791980e-01
1.19399643e+00 4.39641386e-01 2.00879902e-01 3.89811397e-01
5.76437056e-01 6.99164629e-01 1.02629042e+00 5.06159067e-01
2.51719922e-01 7.63081491e-01 1.57639399e-01 3.46860252e-02
-9.58340801e-03 -2.51986712e-01 4.69722092e-01 5.80123544e-01
6.00533664e-01 -1.09102510e-01 -7.27362394e-01 7.16776967e-01
-1.88116908e+00 -5.99126458e-01 2.77231485e-01 2.39101362e+00
7.27544844e-01 -8.40339437e-02 4.10496533e-01 -2.64021102e-02
1.26254904e+00 -8.56426135e-02 -1.06390238e+00 -6.27560392e-02
-3.11181158e-01 -5.63593805e-02 2.70038247e-01 1.16867185e-01
-1.46396017e+00 5.46156228e-01 5.24405670e+00 9.13891315e-01
-1.09862149e+00 -3.01827248e-02 1.72360346e-01 1.72390461e-01
-2.19987914e-01 -3.96110001e-04 -9.00523782e-01 6.91401958e-01
4.26440746e-01 -5.56223750e-01 2.07756534e-01 1.13766217e+00
-3.11198145e-01 1.62340835e-01 -1.32797980e+00 8.23628068e-01
-6.35731313e-03 -6.55184209e-01 1.68467715e-01 1.26495287e-01
7.70590127e-01 -1.07910082e-01 1.69102430e-01 4.97711390e-01
2.61206478e-01 -5.61895967e-01 5.48324168e-01 1.37479842e-01
9.45122540e-01 -8.34387720e-01 7.84440875e-01 3.99008214e-01
-1.31740344e+00 -2.12631613e-01 -5.29503107e-01 2.22451836e-01
-1.60327181e-01 4.75362897e-01 -5.62458396e-01 8.40196431e-01
4.65185881e-01 1.03155351e+00 -4.61792380e-01 1.00357032e+00
1.17622457e-01 3.18162978e-01 -3.01716477e-01 4.34703439e-01
1.32354423e-01 -2.39675015e-01 6.52442217e-01 7.53499448e-01
5.55468798e-01 -1.23167008e-01 4.58492398e-01 7.31574118e-01
-4.25287476e-03 2.27981567e-01 -5.93632758e-01 -4.86708526e-03
9.84647632e-01 8.80618334e-01 -2.14116037e-01 -3.04476559e-01
-7.60769188e-01 9.52292144e-01 2.52313644e-01 3.69404644e-01
-6.56986535e-01 -5.22860169e-01 9.97783720e-01 -9.86999348e-02
4.08403158e-01 1.08800672e-01 -2.23371804e-01 -1.32333171e+00
3.50217409e-02 -8.05486023e-01 7.98037767e-01 -3.33053201e-01
-2.00122976e+00 4.52444255e-01 1.65403008e-01 -1.70039070e+00
-1.19049229e-01 -4.89219785e-01 -5.07920623e-01 1.07991719e+00
-1.64199078e+00 -1.07503617e+00 -2.57296354e-01 9.94961381e-01
4.67904776e-01 -5.28523922e-01 8.55057478e-01 3.05009007e-01
-5.76211452e-01 8.09015095e-01 6.12562478e-01 1.58538982e-01
1.09701598e+00 -1.07313585e+00 -6.40600771e-02 6.75923228e-01
-1.97816521e-01 6.40554130e-01 3.20790112e-01 -5.45825839e-01
-9.76690352e-01 -1.33831394e+00 6.80177569e-01 1.39888162e-02
4.96760547e-01 -1.33682892e-01 -1.28335989e+00 8.28979611e-01
-1.40447259e-01 1.08223915e-01 7.98925996e-01 -5.81720509e-02
-5.75619459e-01 -4.11844224e-01 -1.40399277e+00 1.42060325e-01
7.27604806e-01 -2.71783292e-01 -9.47948933e-01 7.08405003e-02
4.86896455e-01 -1.89496219e-01 -8.26287627e-01 5.01671493e-01
3.16443413e-01 -6.37488425e-01 8.78766358e-01 -5.92117548e-01
1.97677895e-01 -4.90647823e-01 -4.42723125e-01 -1.56788671e+00
-5.14319003e-01 9.87725183e-02 -2.86687631e-02 1.60600603e+00
2.99076717e-02 -8.07779551e-01 4.48651433e-01 4.32417452e-01
-1.35033384e-01 -3.23106557e-01 -1.18533468e+00 -1.00888157e+00
4.11395490e-01 1.05522364e-01 8.78897667e-01 1.23933625e+00
-9.20318812e-02 4.26880181e-01 -2.42169052e-01 3.77722770e-01
5.28480232e-01 5.87742209e-01 6.03580534e-01 -1.58873463e+00
-1.79749250e-01 -2.79454678e-01 -2.78935403e-01 -9.38400686e-01
4.09526557e-01 -1.00183177e+00 9.25568771e-03 -1.18902469e+00
4.74134266e-01 -7.36870587e-01 -8.12552452e-01 7.13438630e-01
-2.68106669e-01 -2.07658887e-01 -1.26138898e-02 6.14683867e-01
-4.35610741e-01 6.82768703e-01 1.00276828e+00 1.16124935e-03
-3.98334354e-01 6.29316121e-02 -1.05729532e+00 5.74209869e-01
6.74711585e-01 -5.41999996e-01 -6.53851032e-01 -2.59392917e-01
-4.52028126e-01 -2.20019624e-01 1.85566783e-01 -1.02078569e+00
5.38842194e-02 -5.41033208e-01 4.75878865e-01 -2.75465459e-01
9.99866948e-02 -1.06436825e+00 1.20187318e-02 1.53252408e-01
-2.24724323e-01 -3.53569627e-01 1.97022855e-01 8.55678976e-01
-4.60706025e-01 -1.09827988e-01 9.90648091e-01 7.39362910e-02
-1.13331151e+00 2.30227917e-01 4.23897337e-03 6.72147572e-02
1.34776592e+00 -1.88843206e-01 -4.11084563e-01 6.47897692e-03
-6.08729184e-01 3.18592191e-01 5.47391593e-01 6.38306737e-01
5.39309740e-01 -1.61760986e+00 -6.95956409e-01 5.11604846e-01
4.11847979e-01 1.68947220e-01 3.18616569e-01 5.24770796e-01
3.05257916e-01 3.18596333e-01 -6.23213172e-01 -6.12012506e-01
-1.09056115e+00 7.12182283e-01 2.32052743e-01 -1.91035062e-01
-3.93863410e-01 6.23920977e-01 8.03241491e-01 -7.66879320e-01
3.15102073e-03 7.49188364e-02 -3.99524242e-01 1.84131697e-01
3.43930095e-01 3.45021069e-01 -1.07619353e-01 -8.05994868e-01
-5.77390432e-01 5.96390009e-01 -3.25932741e-01 2.03442976e-01
1.23707151e+00 -3.23438883e-01 1.53413907e-01 3.70398730e-01
1.24746358e+00 -6.24058135e-02 -1.39080250e+00 -9.70654190e-01
-7.27201328e-02 -4.38600868e-01 -2.25763872e-01 -8.27587664e-01
-1.00029051e+00 7.07399368e-01 6.75384998e-01 -2.24446654e-01
1.43577671e+00 1.91739500e-01 7.15217113e-01 1.21863343e-01
2.08682805e-01 -1.19321525e+00 4.39649783e-02 4.05558109e-01
7.12900102e-01 -1.26577246e+00 -1.42008454e-01 -4.25644219e-01
-1.14192843e+00 1.01317990e+00 9.87164497e-01 -2.87071019e-01
5.73895156e-01 -9.34939310e-02 -1.71953395e-01 1.40187591e-01
-4.59494293e-01 -3.25765461e-02 3.42644989e-01 8.23813140e-01
-1.85780711e-02 1.14055254e-01 -2.09417224e-01 1.30386710e+00
2.43389234e-01 -1.14974007e-02 2.07620591e-01 9.12330985e-01
-4.83658910e-01 -1.39314198e+00 -4.32359338e-01 4.11678046e-01
-3.62921096e-02 2.76387870e-01 -4.88474488e-01 8.38924110e-01
3.71118844e-01 7.30260909e-01 1.65064856e-01 -4.88094836e-01
4.17716473e-01 3.02634090e-01 2.40830272e-01 -7.48833716e-01
-1.20557778e-01 1.01166062e-01 -3.82684588e-01 -6.71273768e-02
-3.58936489e-01 -8.70749354e-01 -1.36747527e+00 6.90205162e-03
-2.97424316e-01 1.64353296e-01 2.27935299e-01 1.00852132e+00
5.86572051e-01 1.95453390e-01 8.63025188e-01 -3.19307566e-01
-8.45940828e-01 -9.55397129e-01 -1.06803143e+00 6.26283526e-01
3.67998898e-01 -1.22914386e+00 -2.76707113e-01 -7.63887027e-03] | [10.380990982055664, 3.117393970489502] |
af459993-94de-4432-9853-775917537470 | self-attentive-3d-human-pose-and-shape | 2103.14182 | null | https://arxiv.org/abs/2103.14182v2 | https://arxiv.org/pdf/2103.14182v2.pdf | Self-Attentive 3D Human Pose and Shape Estimation from Videos | We consider the task of estimating 3D human pose and shape from videos. While existing frame-based approaches have made significant progress, these methods are independently applied to each image, thereby often leading to inconsistent predictions. In this work, we present a video-based learning algorithm for 3D human pose and shape estimation. The key insights of our method are two-fold. First, to address the inconsistent temporal prediction issue, we exploit temporal information in videos and propose a self-attention module that jointly considers short-range and long-range dependencies across frames, resulting in temporally coherent estimations. Second, we model human motion with a forecasting module that allows the transition between adjacent frames to be smooth. We evaluate our method on the 3DPW, MPI-INF-3DHP, and Human3.6M datasets. Extensive experimental results show that our algorithm performs favorably against the state-of-the-art methods. | ['Ming-Hsuan Yang', 'Robinson Piramuthu', 'Marco Piccirilli', 'Yun-Chun Chen'] | 2021-03-26 | null | null | null | null | ['3d-human-pose-and-shape-estimation'] | ['computer-vision'] | [-1.61925346e-01 -1.64682552e-01 -9.67056379e-02 -3.45855743e-01
-4.97968137e-01 -6.94317520e-02 4.09424484e-01 -2.01572821e-01
-4.51714754e-01 4.97912526e-01 3.75935197e-01 3.44358146e-01
2.71043271e-01 -2.32380465e-01 -6.21535540e-01 -3.23855639e-01
-3.66767317e-01 4.57882524e-01 6.09844506e-01 -1.40857846e-01
-5.84317595e-02 3.59922528e-01 -1.45195389e+00 3.14084470e-01
4.26193923e-01 1.04260921e+00 1.43907294e-01 8.22938323e-01
3.42358112e-01 9.94823098e-01 -2.47366920e-01 -2.06929237e-01
2.70672858e-01 -4.21170652e-01 -8.23075473e-01 5.90732336e-01
3.78301829e-01 -6.76963925e-01 -7.54946053e-01 5.24581313e-01
6.92263007e-01 3.65395457e-01 4.03279990e-01 -1.07917666e+00
-9.92127880e-03 -2.08344936e-01 -7.18106210e-01 2.32496783e-01
8.01849365e-01 2.46859446e-01 6.56452954e-01 -9.21599865e-01
8.22218359e-01 1.54104602e+00 8.29742849e-01 4.52034980e-01
-1.00357842e+00 -3.34637195e-01 3.97986680e-01 5.05684972e-01
-1.29346979e+00 -5.15627623e-01 8.31515551e-01 -6.23295069e-01
1.00316203e+00 -5.43827526e-02 1.12070501e+00 1.06059813e+00
4.64639992e-01 1.23932946e+00 8.33999574e-01 -3.47223908e-01
-2.48713675e-03 -5.35906971e-01 -1.07780531e-01 8.94723833e-01
-1.67713836e-01 1.67878211e-01 -8.06617677e-01 1.24071930e-02
9.51981187e-01 1.23950802e-01 -2.61390299e-01 -5.47793567e-01
-1.35166025e+00 5.34083366e-01 2.27422297e-01 6.81402311e-02
-6.53264642e-01 3.06552619e-01 3.65209728e-01 -8.03895891e-02
8.07867467e-01 -3.25886786e-01 -3.95935535e-01 -1.98215216e-01
-1.01595068e+00 6.79832995e-01 5.83648741e-01 8.10563982e-01
3.48616809e-01 -1.58814847e-01 -2.66077816e-01 6.71379924e-01
4.97679353e-01 3.50320250e-01 3.00585419e-01 -1.17798436e+00
3.34594607e-01 2.00378373e-01 4.85091507e-01 -1.27774358e+00
-6.31588042e-01 -5.62326461e-02 -5.58452666e-01 -1.05120931e-02
5.21736503e-01 -6.80491328e-02 -8.11914563e-01 1.45880616e+00
7.06210136e-01 5.25008917e-01 -3.38864058e-01 1.26692522e+00
6.03877604e-01 6.76314652e-01 1.05957441e-01 -2.01821864e-01
1.08299267e+00 -1.40995419e+00 -7.28776276e-01 -3.13041687e-01
3.18551481e-01 -6.02407932e-01 5.57857096e-01 2.57084131e-01
-1.28624463e+00 -9.35864151e-01 -6.82944298e-01 -3.63581032e-02
2.56896943e-01 -8.63292739e-02 3.22515607e-01 8.05926248e-02
-8.38246942e-01 6.62527323e-01 -1.31903994e+00 -4.38324779e-01
1.30081952e-01 2.05427364e-01 -5.14979839e-01 -1.44834176e-01
-9.67296422e-01 7.73666441e-01 2.89802134e-01 3.11679929e-01
-6.54989183e-01 -2.52386689e-01 -9.60327983e-01 -2.41067544e-01
5.61274946e-01 -8.34797800e-01 1.44860971e+00 -8.12933445e-01
-1.54590893e+00 7.01577723e-01 -5.52651525e-01 -5.35919905e-01
8.73198509e-01 -7.81778514e-01 -7.60716349e-02 3.79346192e-01
-2.79879589e-02 9.87770021e-01 8.03950012e-01 -1.18678141e+00
-7.67917156e-01 -3.32080215e-01 -6.61838353e-02 5.25612056e-01
-1.97623391e-02 1.77493168e-03 -1.13834405e+00 -8.02826583e-01
1.06019139e-01 -1.27335382e+00 -3.93145323e-01 1.08917587e-01
-2.87177771e-01 -1.11467421e-01 6.97922349e-01 -9.10404146e-01
1.27743316e+00 -1.84198320e+00 5.74720442e-01 -2.41194274e-02
8.49631801e-02 2.20299467e-01 1.56700276e-02 1.89913377e-01
2.41695419e-01 -4.40793723e-01 -5.63933863e-04 -7.94448555e-01
-1.03747219e-01 2.94165254e-01 2.80002598e-02 6.65590465e-01
4.95583862e-02 1.04225540e+00 -8.52252960e-01 -7.48490334e-01
5.53484917e-01 8.00582230e-01 -7.86905289e-01 4.80648667e-01
-1.59539774e-01 8.58771980e-01 -4.02893692e-01 6.71097636e-01
4.13543135e-01 -3.70775610e-01 2.70702809e-01 -3.00274611e-01
-1.41389430e-01 1.77006647e-02 -1.15806210e+00 1.99085271e+00
-9.07387063e-02 4.41647202e-01 -1.21113658e-01 -8.18218946e-01
4.12426293e-01 4.40304548e-01 9.03535903e-01 -4.67797011e-01
1.65959805e-01 -3.79059948e-02 -2.99547732e-01 -6.58182383e-01
5.02317786e-01 1.65590093e-01 1.15061909e-01 1.50402859e-01
2.48689726e-02 2.68414676e-01 7.26038814e-02 -9.76880863e-02
9.69221473e-01 7.63251007e-01 4.27668393e-01 9.16974619e-02
5.39043546e-01 -1.63829163e-01 7.90352464e-01 4.49023575e-01
-6.37146592e-01 8.37833524e-01 2.34937131e-01 -9.53553379e-01
-1.01547325e+00 -9.30937231e-01 3.41077209e-01 9.19907510e-01
3.85391563e-01 -5.60081184e-01 -7.25238562e-01 -6.86902285e-01
1.89986043e-02 6.32079393e-02 -5.38154304e-01 2.47723505e-01
-9.74932253e-01 -3.80648941e-01 2.79515777e-02 7.69491494e-01
5.14793932e-01 -1.02517557e+00 -1.05905163e+00 4.86155778e-01
-6.25173330e-01 -1.39117444e+00 -8.28480542e-01 -3.87491286e-01
-8.89707208e-01 -1.01720989e+00 -1.01798248e+00 -5.98997056e-01
4.33801919e-01 4.02368546e-01 1.09084308e+00 4.70747016e-02
-3.20171773e-01 6.26171112e-01 -3.89689386e-01 -7.48086497e-02
1.06868170e-01 -1.87960044e-01 2.72002280e-01 1.00134537e-01
2.17981681e-01 -3.23368520e-01 -7.69024611e-01 5.62322497e-01
-4.95588511e-01 2.05427364e-01 3.39234859e-01 8.74914587e-01
8.61978292e-01 -2.60515988e-01 1.87525555e-01 -4.93860990e-01
5.07023260e-02 -2.81390011e-01 -3.90921623e-01 1.43220976e-01
-1.33948117e-01 -1.46103472e-01 1.05511345e-01 -5.27029872e-01
-1.14103138e+00 6.76285446e-01 -1.85302496e-01 -6.95330441e-01
-2.78572828e-01 1.83402568e-01 1.41122732e-02 -1.35198119e-03
1.96093947e-01 9.25228894e-02 6.22925460e-02 -4.11269158e-01
1.06600672e-01 2.65288621e-01 7.55259037e-01 -4.80926663e-01
5.83856702e-01 6.91598237e-01 -9.60397348e-02 -1.02041733e+00
-9.76271689e-01 -6.84983134e-01 -1.01582229e+00 -6.78982973e-01
1.16086483e+00 -1.29770684e+00 -7.28762805e-01 7.07183182e-01
-1.36961508e+00 -3.93367678e-01 5.65085486e-02 7.01025307e-01
-9.26521838e-01 6.76763117e-01 -8.77916455e-01 -1.00182223e+00
-1.34405151e-01 -1.15609002e+00 1.41630447e+00 -5.86904772e-03
-5.14721334e-01 -9.22473490e-01 2.36605808e-01 4.88632411e-01
-5.70905246e-02 4.72581178e-01 8.40037167e-02 -2.65261918e-01
-5.25917530e-01 -1.97923467e-01 -1.23769157e-02 3.83385271e-02
-8.49953219e-02 -3.59653294e-01 -7.06953406e-01 -3.83568347e-01
-8.71470869e-02 -3.13324183e-01 8.12958419e-01 7.74000108e-01
1.04511511e+00 -1.08182676e-01 -4.45931584e-01 5.48402369e-01
8.10568273e-01 2.31329929e-02 5.21030962e-01 2.58518815e-01
8.28587174e-01 7.50090063e-01 9.78846073e-01 7.31789231e-01
7.22824156e-01 1.12634420e+00 2.67781228e-01 6.32095709e-03
-8.90501067e-02 -2.79357612e-01 4.24780130e-01 7.81422853e-01
-5.28361797e-01 -2.91310161e-01 -7.45406449e-01 4.48336929e-01
-2.28423619e+00 -1.16673648e+00 5.88449277e-02 2.03236032e+00
3.19060862e-01 1.13847807e-01 5.08220732e-01 -2.49642506e-02
6.33703709e-01 3.69082600e-01 -4.14507210e-01 2.84771860e-01
1.78246155e-01 -1.51858687e-01 1.43170789e-01 5.15241742e-01
-1.40762067e+00 9.12545621e-01 6.38306713e+00 4.09598202e-01
-6.97859168e-01 2.13098526e-02 5.29802263e-01 -4.31412280e-01
2.53357500e-01 -1.78611681e-01 -7.67632127e-01 4.51168567e-01
4.91811961e-01 2.09433943e-01 1.13304436e-01 5.70526838e-01
4.63148892e-01 -1.79798141e-01 -1.18020022e+00 1.17206907e+00
2.27197170e-01 -9.79150891e-01 -3.05343986e-01 4.12935168e-02
6.30323470e-01 -6.23555295e-02 -2.73066223e-01 5.23539558e-02
-1.17493801e-01 -7.34361231e-01 1.07967925e+00 8.34600449e-01
2.77789384e-01 -8.36841345e-01 7.30062366e-01 4.36455011e-01
-1.59911311e+00 8.89701471e-02 -1.08302332e-01 -9.95681956e-02
7.10272491e-01 4.87279624e-01 -3.65000159e-01 5.03313720e-01
9.93662953e-01 1.09562027e+00 -3.31485868e-01 1.01837480e+00
-1.18890159e-01 2.56745905e-01 -2.74826974e-01 3.65094632e-01
1.11726396e-01 4.37541753e-02 6.02702737e-01 1.18699396e+00
2.74640590e-01 4.24247950e-01 7.82397270e-01 3.66684586e-01
2.04506367e-01 -1.22616358e-01 -4.86634403e-01 2.36907676e-01
9.55578610e-02 9.24314022e-01 -6.29020512e-01 -4.04187620e-01
-6.90926433e-01 1.35227740e+00 3.36668849e-01 3.14730525e-01
-1.00388384e+00 1.47411138e-01 6.62821770e-01 3.03699940e-01
4.69396412e-01 -6.48839593e-01 2.45864213e-01 -1.43714654e+00
2.01546207e-01 -7.19335079e-01 5.41173100e-01 -7.19101667e-01
-1.16456640e+00 4.34180677e-01 1.65466353e-01 -1.30705512e+00
-5.94495416e-01 -3.30275416e-01 -2.95137137e-01 4.00966287e-01
-1.35030627e+00 -1.20266867e+00 -3.32702279e-01 5.65744758e-01
9.15438831e-01 8.29785392e-02 4.14717883e-01 2.90487766e-01
-4.27689493e-01 2.43010327e-01 -2.16372773e-01 1.78916276e-01
8.39901090e-01 -8.16364169e-01 7.25812137e-01 7.77175009e-01
1.49837628e-01 2.17014864e-01 6.60355210e-01 -8.51661861e-01
-1.39026177e+00 -9.59848344e-01 1.03142846e+00 -4.70573306e-01
3.65646899e-01 -1.75676480e-01 -8.10645580e-01 7.89178014e-01
3.11308671e-02 3.76228184e-01 3.78661782e-01 -1.35863364e-01
-5.80537356e-02 1.01951845e-01 -8.65708768e-01 5.15191078e-01
1.27591324e+00 -1.85818061e-01 -5.85351467e-01 2.49031633e-01
4.65892494e-01 -7.80615330e-01 -7.52185047e-01 4.95661199e-01
9.78644848e-01 -1.19088161e+00 1.24963915e+00 -4.48252559e-01
3.22737992e-01 -2.99779803e-01 -2.36226439e-01 -1.12679112e+00
-4.85352457e-01 -5.67553163e-01 -6.66040480e-01 6.62246585e-01
-1.84591249e-01 -5.36335111e-02 9.76016939e-01 5.91496706e-01
-1.22669218e-02 -7.03311861e-01 -8.95567775e-01 -7.74687231e-01
-4.37033623e-01 -4.95424896e-01 2.74815559e-02 5.15130699e-01
-4.05970775e-02 1.91591084e-02 -1.07840872e+00 6.76673651e-02
8.48350346e-01 4.47491109e-02 1.02864230e+00 -1.13696337e+00
-5.17185211e-01 -4.43138182e-02 -5.37656426e-01 -1.55198991e+00
3.42161238e-01 -1.58302039e-01 3.99832815e-01 -1.23491621e+00
2.84076244e-01 1.87307820e-01 -7.50266984e-02 3.00294727e-01
-2.99569249e-01 3.02215219e-01 4.47112411e-01 2.28336737e-01
-9.94798005e-01 5.37557840e-01 1.08090591e+00 2.05538161e-02
-2.43363589e-01 3.14129211e-05 1.72314391e-01 1.04126227e+00
4.61629599e-01 -2.26491630e-01 -1.93319067e-01 -5.02588689e-01
-3.40812683e-01 3.78797740e-01 7.04634726e-01 -1.25963330e+00
3.21094394e-01 -1.54708937e-01 7.33533800e-01 -9.41571474e-01
6.21253252e-01 -7.11211503e-01 2.01431885e-01 6.74391091e-01
-2.13649571e-01 3.03861797e-01 4.27389257e-02 8.86084318e-01
-2.07912505e-01 3.51112574e-01 8.68545771e-01 -2.78331399e-01
-9.97084677e-01 6.16749048e-01 -3.36984426e-01 7.52253532e-02
1.00717473e+00 -9.97012779e-02 3.47986072e-01 -6.08873427e-01
-8.27983201e-01 3.93517852e-01 5.25833189e-01 5.65371931e-01
8.13646257e-01 -1.47354126e+00 -6.29221797e-01 1.31214455e-01
4.63967659e-02 3.14011574e-02 5.26680708e-01 9.78922307e-01
-4.40953702e-01 3.92681986e-01 -1.98221087e-01 -1.00949574e+00
-1.46368015e+00 5.36330283e-01 2.18676597e-01 -3.16868752e-01
-9.46190476e-01 6.83951497e-01 2.64109194e-01 -1.19294226e-01
4.31366295e-01 -1.50617287e-01 -1.38747558e-01 -1.21742249e-01
5.56167603e-01 5.34667730e-01 -3.05823296e-01 -1.19688737e+00
-4.94159073e-01 9.15327907e-01 8.51622000e-02 -1.55859470e-01
1.34365189e+00 -4.50527370e-01 2.81450808e-01 3.90433729e-01
1.10198414e+00 -2.49328732e-01 -1.87608993e+00 -2.58381367e-01
-1.76708922e-01 -7.37022042e-01 -2.27322832e-01 -3.06869835e-01
-1.02487504e+00 7.84037471e-01 5.87334633e-01 -1.37058049e-01
1.04604709e+00 -9.74279568e-02 1.16650712e+00 1.60372838e-01
5.39842188e-01 -1.26330650e+00 3.31687450e-01 7.03844368e-01
7.14294195e-01 -1.38723505e+00 2.05612689e-01 -3.53062958e-01
-6.54020369e-01 9.98987377e-01 7.12090909e-01 -2.33251117e-02
5.06551266e-01 7.14295954e-02 -1.00508459e-01 6.49149343e-03
-8.59225690e-01 -2.65787184e-01 6.70795023e-01 4.37077820e-01
6.51029766e-01 -3.16903681e-01 -2.08104134e-01 3.11193913e-01
3.38089436e-01 3.49187016e-01 1.48314312e-01 1.18937206e+00
-4.70571846e-01 -1.00331306e+00 -5.27230263e-01 -5.56141958e-02
-4.72269475e-01 3.16366673e-01 -1.93981871e-01 8.12259674e-01
3.59554328e-02 8.40835989e-01 2.79818568e-02 -5.24480402e-01
4.91144270e-01 9.57014114e-02 7.44545341e-01 -2.33597010e-01
-2.33222961e-01 5.39274395e-01 1.03143722e-01 -1.13548994e+00
-8.57335806e-01 -9.07251179e-01 -1.22361159e+00 -3.40885699e-01
2.10702233e-02 -1.34649292e-01 1.99392468e-01 1.07496500e+00
4.51947540e-01 4.20823216e-01 3.62223327e-01 -1.67541075e+00
-2.14593962e-01 -6.55186951e-01 -3.92741561e-01 5.86313725e-01
4.29239452e-01 -1.01080596e+00 -8.29359703e-03 4.25102443e-01] | [7.257760524749756, -0.51537024974823] |
e1daab01-6e1e-4379-96ee-1a90f354f31e | krm-based-dialogue-management | 1912.00669 | null | https://arxiv.org/abs/1912.00669v1 | https://arxiv.org/pdf/1912.00669v1.pdf | KRM-based Dialogue Management | A KRM-based dialogue management (DM) is proposed using to implement human-computer dialogue system in complex scenarios. KRM-based DM has a well description ability and it can ensure the logic of the dialogue process. Then a complex application scenario in the Internet of Things (IOT) industry and a dialogue system implemented based on the KRM-based DM will be introduced, where the system allows enterprise customers to customize topics and adapts corresponding topics in the interaction process with users. The experimental results show that the system can complete the interactive tasks well, and can effectively solve the problems of topic switching, information inheritance between topics, change of dominance. | ['Wei Zheng', 'Xiaoyu Chi', 'Wenwu Qu'] | 2019-12-02 | null | null | null | null | ['dialogue-management'] | ['natural-language-processing'] | [-3.34046483e-01 6.16702199e-01 3.13674241e-01 -6.23128653e-01
2.75340490e-02 -5.51975965e-01 9.16112483e-01 1.63882058e-02
5.59642762e-02 6.27164245e-01 2.43151724e-01 -2.99398661e-01
-3.80436122e-01 -9.26396906e-01 5.58124006e-01 -2.16094911e-01
2.21866682e-01 1.22885609e+00 7.51262963e-01 -9.30368662e-01
6.21109605e-01 -5.81782423e-02 -1.27521133e+00 4.01836544e-01
6.18275821e-01 6.87687933e-01 8.63561928e-01 5.49640357e-01
-1.09780014e+00 6.09025061e-01 -7.41752923e-01 -1.50231153e-01
4.62796800e-02 -1.51517883e-01 -1.54145157e+00 5.49648702e-01
-5.16228795e-01 -4.07836556e-01 1.09102346e-01 8.09203684e-01
3.89108658e-01 2.07840115e-01 6.74031258e-01 -1.73277414e+00
-6.07663915e-02 8.03408504e-01 -3.69043797e-02 -4.29143459e-01
7.93170631e-01 -4.52314056e-02 2.91899085e-01 -3.92474174e-01
5.49207151e-01 1.82142639e+00 2.26229027e-01 7.28220880e-01
-6.28743589e-01 -4.65427995e-01 1.89440846e-01 2.01207250e-01
-9.20875192e-01 2.41532609e-01 4.60324496e-01 -4.77724075e-01
8.92831564e-01 3.98417383e-01 8.16855967e-01 2.46998146e-01
6.03438735e-01 3.23324353e-01 1.09013355e+00 -3.56106222e-01
2.90993601e-01 1.11343610e+00 8.65617275e-01 1.48221120e-01
-1.81509778e-01 -9.24565732e-01 -2.55976021e-01 -4.49795246e-01
8.83248389e-01 -2.14533463e-01 1.84918538e-01 -6.78414106e-02
-1.14662111e+00 9.14866745e-01 -6.32750750e-01 5.69695592e-01
-3.03414315e-01 -7.47032225e-01 7.03332007e-01 7.03867853e-01
1.18269980e-01 4.47515875e-01 -9.73249078e-01 -4.57442939e-01
2.61236671e-02 3.80787432e-01 1.75628805e+00 1.34161413e+00
5.70126057e-01 -3.85323972e-01 -2.66645104e-01 8.41747224e-01
8.16492856e-01 4.27352525e-02 6.74866319e-01 -1.09000182e+00
6.26184940e-02 1.07357991e+00 2.63546228e-01 -9.17025387e-01
-7.38260806e-01 7.89734781e-01 -8.41880023e-01 5.84334880e-02
3.85322869e-02 -3.62792045e-01 -3.32742572e-01 1.16187584e+00
8.55006754e-01 -6.13847733e-01 4.16639447e-01 3.37253302e-01
1.23707938e+00 1.02869177e+00 3.43685508e-01 -6.09142125e-01
1.94369662e+00 -6.91444218e-01 -1.47695625e+00 4.57799077e-01
3.89596641e-01 -9.21880186e-01 1.29892707e+00 8.63622308e-01
-9.43392575e-01 -5.58532834e-01 -6.17458642e-01 1.02491654e-01
-7.10790098e-01 -2.69476563e-01 8.78627002e-01 6.96510255e-01
-1.08995187e+00 1.48335788e-02 9.23769921e-02 -9.11252677e-01
-5.85592210e-01 7.64199138e-01 -6.03206046e-02 4.78623211e-01
-1.80238891e+00 1.10657394e+00 8.55639696e-01 9.57013294e-02
-4.83306319e-01 -2.47095928e-01 -3.92834991e-01 -1.23735212e-01
3.63628626e-01 -8.87879014e-01 1.68194556e+00 -4.69606370e-01
-2.36914706e+00 7.23498225e-01 5.56678355e-01 6.82005361e-02
4.16796058e-01 -4.43844125e-02 -6.06056929e-01 5.69429621e-02
1.04929797e-01 4.99482006e-01 4.00812149e-01 -1.07842648e+00
-1.20848453e+00 -1.55779764e-01 3.25011820e-01 7.35064268e-01
-1.49111181e-01 2.64047503e-01 -4.90265816e-01 2.12628156e-01
-1.09120719e-01 -6.29177749e-01 -4.24001038e-01 -5.99354088e-01
-4.22980696e-01 -8.87772858e-01 1.32403302e+00 -2.37865493e-01
1.34185171e+00 -1.80790949e+00 -1.36631325e-01 1.95042312e-01
2.22976040e-02 2.00171560e-01 4.13290262e-01 1.00596488e+00
5.26544273e-01 3.34709615e-01 3.28073442e-01 2.51809388e-01
4.86472338e-01 2.48343736e-01 -1.09917127e-01 -5.06572664e-01
-5.54343402e-01 6.59589916e-02 -4.47234809e-01 -8.11744988e-01
6.20565355e-01 -1.52535826e-01 -8.29740167e-02 4.07686293e-01
-6.23766124e-01 5.89425266e-01 -7.33527720e-01 1.62287235e-01
6.74182653e-01 -3.03373300e-02 4.94102299e-01 -1.22795768e-01
-2.27084890e-01 1.27643153e-01 -1.57240748e+00 1.44252145e+00
-7.18796849e-01 1.60735220e-01 5.60986042e-01 -5.28532922e-01
1.07084394e+00 7.76355326e-01 6.05263829e-01 -2.05579519e-01
2.82243431e-01 -3.50426972e-01 -1.87887698e-01 -8.81642818e-01
7.76624382e-01 8.86044204e-02 -4.55508977e-01 6.96944892e-01
-1.19100800e-02 -8.23977768e-01 1.38298015e-03 3.29174936e-01
5.21323383e-01 -4.10544753e-01 7.82459378e-01 -6.97922111e-01
9.54683661e-01 3.22664201e-01 3.57529432e-01 7.08998740e-01
-8.72025192e-02 -4.56019700e-01 3.93577904e-01 -4.38769966e-01
-8.25160563e-01 -7.32161343e-01 -2.60594159e-01 1.16525960e+00
6.13147497e-01 -4.98426169e-01 -7.49766111e-01 -6.78598642e-01
-3.62366498e-01 8.30933213e-01 2.64624566e-01 1.83689237e-01
-2.42572963e-01 -4.97648835e-01 -1.08245343e-01 -5.50896823e-01
1.41774619e+00 -1.32799900e+00 -5.41757584e-01 5.62291026e-01
-2.63967693e-01 -9.35955286e-01 -2.01037489e-02 -2.17295364e-01
-7.68473685e-01 -8.66677523e-01 9.85843837e-02 -1.08693969e+00
1.30434707e-01 1.71433583e-01 9.72656727e-01 -1.18916273e-01
-4.24801521e-02 7.80113757e-01 -2.46028379e-01 -6.59507811e-01
-7.54849494e-01 2.69052953e-01 -1.51721522e-01 -4.29181367e-01
5.65418303e-01 -4.71857846e-01 -4.47937638e-01 1.02899766e+00
-9.36816394e-01 4.31346625e-01 -6.07364532e-03 4.93973821e-01
-2.12242112e-01 7.40235984e-01 8.74858022e-01 -1.27671051e+00
1.51269984e+00 -4.13423449e-01 -6.15118027e-01 4.04763639e-01
-9.40429628e-01 -2.97584474e-01 2.56253362e-01 -4.36912775e-01
-1.70799541e+00 -1.07628144e-02 -8.61228406e-02 9.85933661e-01
-5.07479310e-01 6.72209486e-02 -8.36478889e-01 3.44729036e-01
-4.69820276e-02 -1.15722924e-01 5.30796051e-02 -3.86470318e-01
4.83104795e-01 1.41143525e+00 1.26012236e-01 -6.49118841e-01
2.06300154e-01 6.60403892e-02 -2.33037740e-01 -8.73211145e-01
1.26530334e-01 -6.75641060e-01 -5.35086274e-01 -7.14116156e-01
1.03682804e+00 -6.66437507e-01 -1.52368009e+00 5.15912175e-01
-1.45236599e+00 -2.80821562e-01 -1.17781922e-01 2.91974187e-01
-7.56730020e-01 3.77168417e-01 -8.23085010e-01 -1.06229711e+00
-5.00007629e-01 -9.83273864e-01 7.80446231e-01 6.01498127e-01
-5.44061184e-01 -1.17267668e+00 -2.29899377e-01 1.86901852e-01
4.13747281e-01 -2.54243165e-01 1.69359481e+00 -9.81775522e-01
-4.29465741e-01 1.71837479e-01 1.38710052e-01 1.57052353e-02
3.32442164e-01 1.28885014e-02 -3.88597757e-01 3.77093107e-01
4.87033457e-01 9.00866836e-02 -2.12247610e-01 -1.14588588e-01
5.61779022e-01 -4.92460847e-01 -4.03685421e-01 -5.96651435e-01
1.20336819e+00 1.01644325e+00 7.85731256e-01 5.68645835e-01
-5.82216606e-02 1.15513325e+00 1.27604949e+00 8.34388018e-01
7.32196212e-01 7.71997094e-01 1.34433955e-01 -1.64955407e-02
4.72334027e-01 2.17259333e-01 6.80527985e-02 1.14274466e+00
5.07912219e-01 -2.09437102e-01 -7.13281751e-01 3.35862227e-02
-2.34500384e+00 -6.14911675e-01 -4.86012042e-01 1.62657380e+00
1.10316324e+00 2.42299989e-01 2.50039279e-01 -2.60395780e-02
9.57584262e-01 -3.49498600e-01 -2.45969608e-01 -9.76886272e-01
5.67900777e-01 -4.19335604e-01 -4.16270345e-02 7.72421002e-01
-9.09829021e-01 1.12760162e+00 7.01483870e+00 5.85959613e-01
-6.07474208e-01 3.46176028e-01 3.04578453e-01 7.11766720e-01
-1.93705946e-01 -3.93231399e-02 -1.02504122e+00 4.24636453e-01
5.61411917e-01 -7.29044437e-01 2.15816602e-01 9.73429143e-01
5.67674756e-01 -5.36254227e-01 -8.23503077e-01 9.70514536e-01
-3.88731778e-01 -8.10253322e-01 4.20447618e-01 -5.37768379e-02
1.62544340e-01 -9.02828038e-01 -1.51295319e-01 4.12905544e-01
7.59729743e-01 -4.32237238e-01 2.60050744e-02 6.39304042e-01
2.07207292e-01 -7.26802051e-01 9.35252368e-01 7.19007611e-01
-1.08822405e+00 -1.34849772e-01 -6.84429109e-01 -1.03964262e-01
4.60636228e-01 4.64968055e-01 -1.37362719e+00 6.09977186e-01
5.81113398e-01 -2.06296429e-01 -5.17905317e-02 7.27427959e-01
3.09346139e-01 -1.23166321e-02 -2.58879691e-01 -3.69802028e-01
-1.25821128e-01 -6.67453587e-01 4.80437547e-01 1.19210684e+00
-4.34128940e-02 6.36509597e-01 6.23918295e-01 3.20063710e-01
6.48348033e-01 7.79196382e-01 -6.46552920e-01 4.16838080e-01
4.89844650e-01 1.50088978e+00 -6.86181903e-01 -5.90277493e-01
-1.67097852e-01 7.57148504e-01 -7.41995752e-01 -5.29975407e-02
-5.59062064e-01 -4.99829501e-01 4.70128000e-01 9.77993831e-02
-3.86289388e-01 -3.15247953e-01 -1.08221091e-01 -4.60523605e-01
-5.39405167e-01 -9.27120805e-01 4.09867346e-01 -6.68179631e-01
-1.31304562e+00 5.28113723e-01 6.37096643e-01 -9.54345226e-01
-3.58243227e-01 -2.94273674e-01 -8.02192748e-01 5.79265594e-01
-8.13212693e-01 -1.03020179e+00 -2.87669122e-01 8.23197782e-01
1.14897633e+00 -5.08989036e-01 1.08608234e+00 7.07558766e-02
-2.63877958e-01 -3.81124049e-01 -3.03713232e-01 -6.92161322e-01
8.26543033e-01 -1.34659433e+00 -2.17357904e-01 -2.39146620e-01
-8.60107601e-01 7.33706713e-01 6.56245887e-01 -7.11713910e-01
-1.14919889e+00 -5.50884187e-01 9.46038842e-01 -4.98176068e-02
2.79674590e-01 -3.85668874e-01 -5.81838250e-01 3.36887449e-01
9.69418764e-01 -1.27824748e+00 8.57685745e-01 -1.00549003e-02
6.02446556e-01 -2.39803329e-01 -1.50650620e+00 5.31572521e-01
5.07973075e-01 -2.39599571e-01 -7.83585846e-01 8.59335721e-01
1.02370715e+00 -1.92986876e-01 -1.06330514e+00 1.36119843e-01
3.22914809e-01 -7.07776666e-01 5.86981893e-01 -5.67695618e-01
-3.02899927e-01 -4.05045748e-01 1.99229792e-02 -8.21448088e-01
-2.44152531e-01 -1.04955196e+00 5.84455073e-01 2.00015521e+00
2.53637344e-01 -8.48561227e-01 3.32285345e-01 1.49164450e+00
3.67987424e-01 -1.42563045e-01 -5.79694808e-01 -1.56355381e-01
-2.49570236e-01 1.46321654e-02 8.98310363e-01 8.39488328e-01
8.09697211e-01 9.97597277e-01 -3.34041715e-01 4.97674495e-02
-1.89466953e-01 -2.54527688e-01 9.81032610e-01 -1.72226501e+00
-2.15429142e-01 -8.85375887e-02 -1.66628242e-01 -1.19561946e+00
-1.22494057e-01 -1.46661118e-01 -8.58712867e-02 -1.97173798e+00
-1.69429600e-01 -3.53401631e-01 3.72535408e-01 7.53832906e-02
3.03690583e-01 -8.32219124e-01 3.86181593e-01 3.07383418e-01
-9.55409586e-01 2.40000084e-01 1.14507091e+00 -1.85860604e-01
-7.85349667e-01 5.19865572e-01 -5.41281581e-01 7.90815532e-01
1.05337715e+00 -1.01717159e-01 -9.21628475e-01 3.84592950e-01
2.85187513e-01 5.95206261e-01 -3.85709316e-01 -5.66873908e-01
4.72838670e-01 -4.77421969e-01 -1.50209963e-01 -8.95468891e-01
1.09900191e-01 -1.34026504e+00 3.69185716e-01 4.33166862e-01
-3.34385008e-01 2.60452658e-01 1.37173198e-02 3.10861856e-01
-7.72627965e-02 -6.49306834e-01 3.59837353e-01 -5.31034768e-01
-1.00140083e+00 -1.29401937e-01 -1.31055689e+00 -4.31851447e-01
1.49597263e+00 -1.76657245e-01 -2.98414230e-01 -9.26599801e-01
-9.20548379e-01 8.10820699e-01 2.56240387e-02 2.55220741e-01
3.56058538e-01 -9.75524902e-01 -4.84514497e-02 -1.13747761e-01
-3.13761622e-01 7.45586231e-02 1.87320441e-01 1.69837102e-01
-5.95243096e-01 4.44278240e-01 -4.47303444e-01 -4.70456064e-01
-1.57352400e+00 4.78658855e-01 1.71708405e-01 -5.75829029e-01
-4.12203908e-01 2.17971206e-01 2.57183045e-01 -9.22946513e-01
6.13609374e-01 -6.27774522e-02 -9.77568328e-01 1.47521511e-01
3.94928664e-01 5.31392932e-01 -3.25764656e-01 7.23034563e-03
-9.21043158e-02 4.64540988e-01 -8.26916695e-02 -7.35413134e-01
7.49915540e-01 -1.01627290e+00 -7.09913909e-01 8.60101521e-01
4.77178454e-01 -3.35534126e-01 -3.82454306e-01 -4.90856431e-02
2.10682109e-01 -8.75840411e-02 -4.70325857e-01 -1.00611913e+00
-3.13522905e-01 5.43118179e-01 7.65123248e-01 1.17910314e+00
9.66948092e-01 -1.12987459e-01 6.71417832e-01 9.25454140e-01
5.03080487e-01 -1.72087061e+00 1.59000367e-01 6.59644723e-01
9.32207823e-01 -8.40665340e-01 3.66100781e-02 -9.73340690e-01
-1.24347556e+00 1.59110045e+00 9.51113701e-01 5.05770683e-01
1.37703860e+00 3.80926967e-01 1.58629447e-01 -5.83120346e-01
-1.07553864e+00 1.07834913e-01 -5.13559699e-01 8.44063640e-01
1.40433669e-01 1.20679885e-02 -1.06151974e+00 7.36477017e-01
-4.11628693e-01 2.10534390e-02 8.12634826e-01 8.35299015e-01
-8.81586134e-01 -1.62548614e+00 -4.06355441e-01 2.30395749e-01
-2.52684921e-01 4.46375668e-01 -7.06178963e-01 1.05430305e+00
2.47916371e-01 1.65188134e+00 4.90861014e-02 -5.11403620e-01
5.36133409e-01 3.24674219e-01 -2.52289683e-01 -8.62646520e-01
-1.05656624e+00 -5.02334312e-02 5.56293905e-01 -2.21820131e-01
-3.67820978e-01 -3.49233806e-01 -1.62774336e+00 -2.75535077e-01
-4.74211901e-01 5.79842031e-01 7.64921784e-01 8.08797956e-01
2.53579825e-01 2.65881807e-01 9.11425829e-01 -6.90882057e-02
-5.05281329e-01 -1.18226540e+00 -8.18963289e-01 1.95670828e-01
-4.11175191e-01 -2.99987406e-01 1.22751608e-01 1.05746090e-01] | [12.979720115661621, 7.984541416168213] |
4cea371a-2ec2-4b26-b4bf-de25a2db5d25 | multires-netvlad-augmenting-place-recognition | 2202.09146 | null | https://arxiv.org/abs/2202.09146v1 | https://arxiv.org/pdf/2202.09146v1.pdf | MultiRes-NetVLAD: Augmenting Place Recognition Training with Low-Resolution Imagery | Visual Place Recognition (VPR) is a crucial component of 6-DoF localization, visual SLAM and structure-from-motion pipelines, tasked to generate an initial list of place match hypotheses by matching global place descriptors. However, commonly-used CNN-based methods either process multiple image resolutions after training or use a single resolution and limit multi-scale feature extraction to the last convolutional layer during training. In this paper, we augment NetVLAD representation learning with low-resolution image pyramid encoding which leads to richer place representations. The resultant multi-resolution feature pyramid can be conveniently aggregated through VLAD into a single compact representation, avoiding the need for concatenation or summation of multiple patches in recent multi-scale approaches. Furthermore, we show that the underlying learnt feature tensor can be combined with existing multi-scale approaches to improve their baseline performance. Evaluation on 15 viewpoint-varying and viewpoint-consistent benchmarking datasets confirm that the proposed MultiRes-NetVLAD leads to state-of-the-art Recall@N performance for global descriptor based retrieval, compared against 11 existing techniques. Source code is publicly available at https://github.com/Ahmedest61/MultiRes-NetVLAD. | ['Sourav Garg', 'Michael Milford', 'Ahmad Khaliq'] | 2022-02-18 | null | null | null | null | ['visual-place-recognition'] | ['computer-vision'] | [-2.33033985e-01 -5.72523296e-01 -2.39555880e-01 -3.16143662e-01
-1.25773847e+00 -8.17355454e-01 8.86557281e-01 4.58390951e-01
-5.62431633e-01 4.09834176e-01 4.22366202e-01 1.52058616e-01
-1.30335584e-01 -7.23709822e-01 -8.38789523e-01 -4.02207702e-01
9.42434892e-02 3.51077318e-01 5.05369902e-01 -3.10595930e-01
4.55103338e-01 9.66110110e-01 -1.71549773e+00 1.97233617e-01
3.91766369e-01 1.15426838e+00 3.39381486e-01 5.25745571e-01
1.05394237e-01 6.77202225e-01 -2.15117291e-01 -2.58941390e-03
4.75686103e-01 3.01769264e-02 -6.93032861e-01 -1.76749900e-01
1.01901031e+00 -3.79610091e-01 -6.50423467e-01 9.22598302e-01
6.16556168e-01 3.83999318e-01 3.76540214e-01 -1.02404606e+00
-6.70546114e-01 -2.90709019e-01 -4.42396343e-01 2.50486672e-01
6.11855209e-01 5.01203127e-02 1.17375469e+00 -1.10827911e+00
9.34255362e-01 1.12318313e+00 7.66452849e-01 -6.59566093e-03
-1.32991040e+00 -4.09053624e-01 -3.66715394e-04 2.98244327e-01
-1.98338234e+00 -5.51960349e-01 6.53128088e-01 -3.83032918e-01
1.37784827e+00 1.22382030e-01 4.92359996e-01 8.46080482e-01
2.77501911e-01 5.45644641e-01 9.41071570e-01 -1.08453006e-01
1.29115507e-01 -4.72677588e-01 -3.11008692e-01 9.19039428e-01
2.89774984e-01 4.39175754e-04 -8.78581166e-01 -1.79094449e-01
1.01597214e+00 2.89493352e-01 -7.57223070e-02 -8.49526525e-01
-1.50382292e+00 8.66182208e-01 1.18175495e+00 3.42410326e-01
-5.79804361e-01 5.06750703e-01 3.02941531e-01 7.48664560e-03
2.89042830e-01 4.93579268e-01 -8.87559876e-02 -4.71040374e-03
-1.06741440e+00 5.13022363e-01 3.07193816e-01 1.08831835e+00
1.28596938e+00 -1.54087305e-01 -2.54672259e-01 7.19355404e-01
3.63123149e-01 5.73495507e-01 3.85658383e-01 -1.06439185e+00
3.29863757e-01 5.83336771e-01 2.28198975e-01 -1.27137196e+00
-4.13040906e-01 -5.07422924e-01 -5.56636333e-01 1.19801186e-01
2.62902211e-02 6.54213667e-01 -1.13208532e+00 1.29696488e+00
3.32827538e-01 2.26251483e-01 -4.08945158e-02 1.19757354e+00
8.70114088e-01 5.44613719e-01 -2.21499577e-01 5.64527750e-01
1.27019942e+00 -1.16232884e+00 -2.91985840e-01 -4.06750500e-01
4.80164230e-01 -9.02343571e-01 6.46694899e-01 2.92223562e-02
-7.76199341e-01 -4.32018638e-01 -1.19171858e+00 -7.70451605e-01
-6.09761596e-01 7.99540877e-02 7.17953563e-01 1.06345825e-01
-1.42601752e+00 4.22309369e-01 -8.77306700e-01 -5.54838657e-01
3.02904069e-01 3.57329339e-01 -8.63089859e-01 -4.74032342e-01
-8.09738159e-01 9.74469483e-01 2.67172784e-01 5.61554767e-02
-9.16945279e-01 -4.54085737e-01 -1.21759140e+00 -2.27700919e-01
-9.49354284e-03 -6.70909524e-01 1.12754071e+00 -2.16649085e-01
-9.82060850e-01 8.35869908e-01 -4.82103229e-01 -5.07010460e-01
3.15719634e-01 -2.86764503e-01 8.25317390e-03 3.52845281e-01
4.80357856e-01 1.02151132e+00 4.53130156e-01 -1.21726608e+00
-6.99463248e-01 -3.24657112e-01 1.67919219e-01 3.97007734e-01
2.62628824e-01 -1.91967443e-01 -7.36420274e-01 -3.41488600e-01
6.63355470e-01 -8.82980108e-01 -3.70209932e-01 1.56807169e-01
2.27104127e-02 -2.24226043e-02 6.05687857e-01 -4.90451604e-01
6.73049629e-01 -2.14298964e+00 1.77065343e-01 1.14582315e-01
2.60915607e-01 -1.36902079e-01 -4.24094707e-01 7.39145458e-01
4.00705040e-01 -2.33754352e-01 9.26467180e-02 -6.20361149e-01
5.99415191e-02 2.69600451e-01 -1.81534737e-01 9.80018973e-01
1.86515212e-01 1.02745235e+00 -1.04279852e+00 -3.08647394e-01
7.28328586e-01 9.26749527e-01 -6.94505990e-01 -3.28258867e-03
1.15280509e-01 5.18472910e-01 -4.09483761e-01 9.87466931e-01
6.14728093e-01 -1.76983744e-01 -3.41744184e-01 -4.46485937e-01
-4.18688774e-01 5.79213560e-01 -1.05808938e+00 2.36636138e+00
-5.72602749e-01 7.94273257e-01 -1.75534904e-01 -7.19202280e-01
1.05945265e+00 -1.26308799e-01 5.10849476e-01 -9.63015795e-01
7.71844611e-02 3.98751974e-01 -4.43646938e-01 2.19757602e-01
9.84897852e-01 2.41624877e-01 -3.32407624e-01 -2.00370133e-01
3.88200700e-01 -1.52209595e-01 -2.08364874e-02 5.27804010e-02
1.26509559e+00 2.71190405e-01 4.53690350e-01 -1.40771568e-01
4.37355906e-01 2.50082165e-01 4.41432118e-01 7.64426112e-01
-2.04993948e-01 1.04888749e+00 -3.55736539e-02 -5.85103452e-01
-1.17537534e+00 -1.08360982e+00 -1.48849025e-01 8.50699961e-01
4.48337227e-01 -6.95638657e-01 -1.10823162e-01 -2.67283976e-01
2.07708225e-01 1.60050809e-01 -5.23704350e-01 1.47376865e-01
-4.73073632e-01 -1.13237105e-01 6.47031546e-01 4.58867818e-01
6.10475838e-01 -7.63517916e-01 -8.54112923e-01 2.64617354e-01
-3.05271000e-01 -1.33943415e+00 -2.38685817e-01 2.72166699e-01
-5.31974494e-01 -7.54820168e-01 -7.02975869e-01 -8.16700578e-01
4.85640585e-01 6.48285925e-01 8.86593759e-01 -3.37864421e-02
-2.82509714e-01 5.24294376e-01 -5.29582798e-01 1.12212479e-01
1.71127170e-01 2.83747375e-01 1.56332642e-01 -1.74736962e-01
2.39604831e-01 -6.64533138e-01 -9.44488645e-01 6.94410056e-02
-6.34656608e-01 -4.48319502e-02 7.42506981e-01 8.01304698e-01
1.07804799e+00 -6.69982791e-01 2.66488716e-02 -5.83251938e-02
1.56185895e-01 -3.20871979e-01 -7.39008725e-01 5.98151721e-02
-2.21985027e-01 1.84692129e-01 2.80322850e-01 -1.33707091e-01
-4.10195023e-01 3.72731060e-01 -2.28095919e-01 -7.71689415e-01
-2.22782180e-01 4.69465226e-01 1.43641010e-01 -6.55443370e-01
5.70952117e-01 4.61304367e-01 -2.89193511e-01 -4.58238870e-01
5.72235823e-01 4.30858344e-01 6.70230567e-01 -4.82770473e-01
9.35018420e-01 6.45114064e-01 2.21305087e-01 -6.58181787e-01
-6.94207847e-01 -9.18889344e-01 -9.01538372e-01 1.15960829e-01
8.90692234e-01 -1.40275681e+00 -4.40676332e-01 1.99537098e-01
-1.25229824e+00 -1.87766418e-01 -4.39490564e-02 4.71475065e-01
-6.30912960e-01 7.06808418e-02 -5.00689924e-01 -3.59627187e-01
-2.76613891e-01 -1.21561682e+00 1.82887542e+00 1.37047559e-01
-2.15240195e-02 -7.14317977e-01 2.54001677e-01 2.82263488e-01
4.96988237e-01 4.07529712e-01 1.33934855e-01 -4.86971915e-01
-1.18634105e+00 -3.98317695e-01 -3.04040253e-01 -1.35432124e-01
-2.70378701e-02 -4.61970747e-01 -8.93949866e-01 -5.84060907e-01
-5.08149922e-01 -4.21994030e-01 9.69367743e-01 5.81442975e-02
7.20848083e-01 -1.15890548e-01 -2.12702334e-01 1.02446556e+00
1.78387713e+00 -1.81057349e-01 5.90589345e-01 7.40632951e-01
9.59503889e-01 1.89571232e-01 7.36616313e-01 4.62374270e-01
7.37166524e-01 9.69656110e-01 5.88463485e-01 -7.99711123e-02
-2.35802814e-01 -4.84960854e-01 1.98450789e-01 6.08056068e-01
-1.68976653e-03 1.54717296e-01 -1.10049558e+00 8.17656875e-01
-1.90924096e+00 -8.03570032e-01 1.49356321e-01 2.18202090e+00
3.20224404e-01 -3.35676372e-01 -2.48960942e-01 -2.57921904e-01
3.55360687e-01 6.12469673e-01 -2.46904582e-01 -6.26834035e-02
-2.31550664e-01 2.03428984e-01 1.02213192e+00 7.42591321e-01
-1.21901309e+00 1.33630490e+00 5.37934208e+00 6.67608738e-01
-1.42959857e+00 3.17468584e-01 5.83342388e-02 -1.75432935e-01
-1.94531858e-01 1.81104302e-01 -9.19909179e-01 -2.13041678e-01
7.13468373e-01 1.84263401e-02 4.96791393e-01 8.52018237e-01
-1.17406011e-01 -2.79283226e-01 -1.05056465e+00 1.46921003e+00
2.39894181e-01 -1.62242901e+00 8.46175849e-02 2.54893184e-01
7.53353715e-01 8.89266014e-01 3.06768585e-02 2.00870737e-01
3.50945927e-02 -1.00688052e+00 1.01856041e+00 2.92472839e-01
7.26891756e-01 -7.45611131e-01 5.70482075e-01 9.19784382e-02
-1.76915383e+00 5.07701226e-02 -6.52926922e-01 -2.21434329e-02
6.95491284e-02 2.10643798e-01 -8.46224964e-01 9.34837639e-01
8.17524016e-01 8.85060310e-01 -9.96778905e-01 1.24702919e+00
1.29068168e-02 -2.29146853e-01 -4.75290805e-01 2.44087934e-01
6.07896328e-01 2.10722759e-01 4.96986330e-01 1.19814694e+00
4.87829804e-01 -2.28080034e-01 3.88826281e-01 6.29603148e-01
8.85909870e-02 8.57177973e-02 -9.47627485e-01 2.51044154e-01
6.72291815e-01 1.35868549e+00 -8.17422271e-01 -1.83813363e-01
-4.74256992e-01 1.24239814e+00 6.47335172e-01 1.76502660e-01
-5.88916183e-01 -2.33476296e-01 9.68116522e-01 1.47108063e-01
6.64480686e-01 -7.37374187e-01 -3.58790695e-03 -1.30401790e+00
-2.91773994e-02 -5.64631820e-01 5.08851297e-02 -8.33156526e-01
-9.22539055e-01 7.22246468e-01 -7.42743313e-02 -1.36072183e+00
-2.45242536e-01 -4.44404334e-01 -6.59484118e-02 9.22495663e-01
-1.86191487e+00 -1.60629010e+00 -5.62578022e-01 7.51027167e-01
4.89100426e-01 1.02518193e-01 9.69371259e-01 3.68772745e-01
1.24618538e-01 3.99336845e-01 1.09292939e-01 1.65094241e-01
7.32622981e-01 -9.54327106e-01 6.10540867e-01 9.34343636e-01
4.48845476e-01 8.14081192e-01 4.14047241e-01 -5.67892671e-01
-1.73874915e+00 -1.25331390e+00 8.69593322e-01 -5.72975338e-01
5.19213915e-01 -6.40175760e-01 -6.58690393e-01 6.31933689e-01
1.26318112e-01 6.69514716e-01 3.10590923e-01 -6.91712871e-02
-6.31222486e-01 -1.34433031e-01 -1.08514309e+00 5.18478632e-01
1.08313787e+00 -9.70814884e-01 -4.09672469e-01 8.66572931e-02
5.70472419e-01 -8.58374357e-01 -1.04853213e+00 4.28775877e-01
6.09604895e-01 -1.01940608e+00 1.18892825e+00 9.19938609e-02
1.25633389e-01 -8.19828570e-01 -8.13213706e-01 -9.70023274e-01
-5.71342289e-01 -2.54256934e-01 9.85008776e-02 9.01071072e-01
-3.48701216e-02 -6.27463400e-01 3.97090197e-01 6.22843541e-02
-2.16237724e-01 -6.14848733e-01 -1.19083691e+00 -7.43185580e-01
-2.90701449e-01 -2.97041059e-01 5.26786983e-01 8.15449893e-01
-4.64299530e-01 1.48072824e-01 -2.87880719e-01 5.97944796e-01
6.86610460e-01 1.74662769e-01 8.83795619e-01 -1.01725686e+00
3.79540697e-02 -2.75199890e-01 -1.13239682e+00 -1.28085518e+00
1.08396253e-02 -1.08888483e+00 2.47040227e-01 -1.97225428e+00
-5.65428510e-02 -4.16457325e-01 -5.35832345e-01 6.78604424e-01
3.72290045e-01 8.08161199e-01 3.33963156e-01 4.08886433e-01
-1.05612314e+00 6.44008219e-01 9.46865320e-01 -1.60348676e-02
-6.74435496e-02 -6.49136364e-01 -3.62957329e-01 3.52429777e-01
5.78669846e-01 -3.97290498e-01 -1.92354932e-01 -5.51801443e-01
9.32681039e-02 -8.23799968e-02 7.99842894e-01 -1.33438921e+00
5.19760370e-01 4.08274531e-02 6.36339962e-01 -8.64165246e-01
7.04865336e-01 -7.42639422e-01 2.84125626e-01 2.95412242e-01
-5.10532968e-02 4.43039298e-01 2.42840201e-01 4.96457726e-01
-4.69600886e-01 2.30021134e-01 4.61412400e-01 -1.82321101e-01
-1.20437407e+00 4.64581490e-01 -4.30468582e-02 -2.94384897e-01
6.95580423e-01 -1.90313637e-01 -3.44942391e-01 -2.32571363e-01
-3.14466000e-01 2.47599706e-02 9.81685996e-01 7.77040660e-01
8.08505118e-01 -1.56491435e+00 -4.46369380e-01 1.57607168e-01
5.40665209e-01 3.16002250e-01 2.36642092e-01 9.38236535e-01
-9.55018461e-01 6.86377943e-01 -4.40074891e-01 -9.31596994e-01
-1.04546368e+00 3.21834743e-01 3.17722946e-01 -6.29958287e-02
-8.25330019e-01 6.19248033e-01 1.42776668e-01 -4.97836888e-01
7.56255016e-02 -3.17691654e-01 7.22312629e-02 -4.73224595e-02
4.56960738e-01 4.09867838e-02 2.87365317e-01 -1.30718076e+00
-8.15822065e-01 9.71506536e-01 2.62723472e-02 -1.88637987e-01
1.47981822e+00 -2.20885172e-01 -1.89653188e-01 2.89037496e-01
1.52987111e+00 2.87407860e-02 -1.29071736e+00 -2.87377566e-01
-3.65119092e-02 -6.45031154e-01 2.22604230e-01 -2.63120025e-01
-6.72099650e-01 6.62326157e-01 6.97933912e-01 -4.68499929e-01
8.26149642e-01 2.20050469e-01 5.29250264e-01 6.55130029e-01
9.02891457e-01 -6.02962852e-01 3.59969884e-02 8.03003967e-01
1.08887601e+00 -1.31587493e+00 2.26198703e-01 -1.28273427e-01
-4.47881758e-01 8.81133139e-01 5.11496127e-01 -5.96859455e-01
3.78883302e-01 -3.83174047e-02 5.17135412e-02 -2.97358125e-01
-4.52760428e-01 -5.45506656e-01 6.16475463e-01 5.44074595e-01
3.04527700e-01 -1.24777898e-01 -3.00754923e-02 6.50137737e-02
-2.12900236e-01 -2.59085655e-01 1.25716507e-01 1.21639442e+00
-4.63028669e-01 -1.07068586e+00 -3.17960948e-01 1.26676252e-02
-2.00825423e-01 -2.60015160e-01 -1.85134366e-01 6.71251476e-01
1.06352128e-01 4.91449237e-01 1.49564371e-01 -4.90867555e-01
2.88478285e-01 -1.08264729e-01 6.87589049e-01 -5.37080824e-01
-4.22120661e-01 1.46071762e-01 -5.42223305e-02 -1.17249513e+00
-4.73150074e-01 -6.80069685e-01 -1.16553307e+00 -2.15631068e-01
2.85015833e-02 -2.17097551e-01 8.34104657e-01 7.57927120e-01
7.29812980e-01 2.82474309e-01 2.13686496e-01 -1.58203745e+00
-7.96224251e-02 -7.49499142e-01 -2.20962554e-01 1.78705707e-01
7.20676959e-01 -9.24139857e-01 -5.60427494e-02 -2.71689057e-01] | [7.670206546783447, -1.995654821395874] |
1cedc504-42f5-4844-8d78-f44f3557151a | minimum-description-length-clustering-to | 2306.14937 | null | https://arxiv.org/abs/2306.14937v2 | https://arxiv.org/pdf/2306.14937v2.pdf | Minimum Description Length Clustering to Measure Meaningful Image Complexity | Existing image complexity metrics cannot distinguish meaningful content from noise. This means that white noise images, which contain no meaningful information, are judged as highly complex. We present a new image complexity metric through hierarchical clustering of patches. We use the minimum description length principle to determine the number of clusters and designate certain points as outliers and, hence, correctly assign white noise a low score. The presented method has similarities to theoretical ideas for measuring meaningful complexity. We conduct experiments on seven different sets of images, which show that our method assigns the most accurate scores to all images considered. Additionally, comparing the different levels of the hierarchy of clusters can reveal how complexity manifests at different scales, from local detail to global structure. We then present ablation studies showing the contribution of the components of our method, and that it continues to assign reasonable scores when the inputs are modified in certain ways, including the addition of Gaussian noise and the lowering of the resolution. | ['Thomas Lukasiewicz', 'Louis Mahon'] | 2023-06-26 | null | null | null | null | ['clustering'] | ['methodology'] | [ 2.39548787e-01 -1.22850470e-01 3.21756691e-01 -1.80331022e-01
-5.78309000e-01 -7.83429086e-01 3.76214296e-01 4.61219490e-01
-3.48134220e-01 5.80372773e-02 2.38048375e-01 -1.02380499e-01
-4.85931247e-01 -5.90289116e-01 -2.09234029e-01 -7.25957692e-01
-2.20550746e-01 1.17000841e-01 6.14612997e-01 -8.80082697e-02
5.69419146e-01 6.70651317e-01 -1.89722669e+00 4.58001673e-01
6.75725102e-01 7.58219600e-01 9.68644768e-02 8.78926456e-01
-8.34654197e-02 7.23291814e-01 -1.00283754e+00 -1.36888966e-01
4.88800615e-01 -6.56794786e-01 -6.36602223e-01 5.84617198e-01
5.93974471e-01 9.78051648e-02 1.26737788e-01 1.43771040e+00
3.89141202e-01 5.56765706e-04 7.62607157e-01 -9.74865377e-01
-4.67451185e-01 3.02149951e-01 -5.64419508e-01 3.07688743e-01
4.90422994e-01 1.33632749e-01 8.60219419e-01 -8.81830394e-01
7.33581960e-01 1.34750819e+00 8.21899891e-01 1.28727168e-01
-1.67059910e+00 -1.55432776e-01 4.35812436e-02 2.86214411e-01
-1.53632569e+00 -4.66023386e-01 7.03999937e-01 -7.04903483e-01
6.38668954e-01 6.56664491e-01 4.05799776e-01 6.12256706e-01
-2.03269091e-03 1.51732758e-01 1.54649162e+00 -5.51433563e-01
5.48249483e-01 5.85120395e-02 1.08916879e-01 3.58516932e-01
6.14515722e-01 -8.35310146e-02 -1.56899716e-03 -2.52233863e-01
6.30057871e-01 -1.73499942e-01 -1.95061132e-01 -4.54972357e-01
-1.04701376e+00 6.26069009e-01 3.41157675e-01 6.59677863e-01
-2.12775677e-01 -3.13285626e-02 1.66533127e-01 5.16754329e-01
1.82593197e-01 6.61201477e-01 -1.12353712e-01 1.23400830e-01
-9.75599468e-01 -1.52540550e-01 7.11854756e-01 7.36391842e-01
7.59284377e-01 -1.60467982e-01 -3.50717418e-02 7.31634736e-01
-1.28239051e-01 1.45024315e-01 3.03433776e-01 -1.49341047e+00
-1.69013409e-04 7.04401851e-01 9.17026624e-02 -1.44794095e+00
-6.24061823e-01 -2.75903076e-01 -7.53535509e-01 8.49833071e-01
4.52433079e-01 1.84433788e-01 -9.50717509e-01 1.40275586e+00
-2.63660308e-02 -1.79642752e-01 -1.16148137e-01 7.11635411e-01
6.00368977e-01 3.79294872e-01 -1.17441520e-01 -4.26428616e-01
1.36226082e+00 -4.56575751e-01 -7.88785398e-01 1.37765542e-01
2.66473830e-01 -1.01383197e+00 1.26221788e+00 5.77421963e-01
-1.18223000e+00 -8.49452913e-01 -1.22105753e+00 1.96290925e-01
-5.50663173e-01 -9.59496573e-02 1.36895955e-01 7.82917202e-01
-1.45138502e+00 9.50479865e-01 -6.03694856e-01 -5.09938717e-01
8.28463361e-02 2.15438396e-01 -2.74806172e-01 5.85090369e-02
-6.12105727e-01 8.62320781e-01 2.94710308e-01 -3.12645547e-02
-4.64990586e-01 -9.98728126e-02 -5.53448617e-01 1.36346549e-01
1.70539021e-01 -4.23232287e-01 5.88652074e-01 -1.30597699e+00
-9.22845244e-01 7.81637490e-01 -1.08996138e-01 -1.17229693e-01
4.57610160e-01 3.37807894e-01 -3.66617650e-01 5.72663665e-01
2.61987820e-02 5.59278250e-01 1.00954700e+00 -1.89296126e+00
-4.97776181e-01 -1.98580563e-01 -1.05161451e-01 1.51908726e-01
-3.06933045e-01 1.59456864e-01 -5.34808636e-01 -7.15641677e-01
5.40421605e-01 -8.17091763e-01 -5.73718905e-01 -1.16919123e-01
-2.68987536e-01 9.06436518e-02 5.85200965e-01 -3.94343555e-01
1.59213388e+00 -2.44575334e+00 4.32621874e-02 8.24091375e-01
6.19316280e-01 -1.98923066e-01 -2.44879290e-01 4.46237445e-01
-3.24532300e-01 5.90203941e-01 -2.75080383e-01 -3.92270274e-02
-5.30825518e-02 2.77053297e-01 3.64436328e-01 3.58412325e-01
1.28254101e-01 4.53395069e-01 -6.40235007e-01 -5.81818640e-01
2.02233821e-01 2.48377904e-01 -4.83162254e-01 -2.52224505e-01
3.81748974e-01 1.08702540e-01 1.38734151e-02 4.41261142e-01
8.12334478e-01 -1.74908251e-01 9.67032686e-02 -3.51088911e-01
-1.93008631e-01 -3.18780482e-01 -1.49793613e+00 9.93493855e-01
2.07933247e-01 7.65872478e-01 8.78699496e-02 -6.71827435e-01
7.41296947e-01 8.71000662e-02 4.94938225e-01 -6.25089943e-01
3.90102640e-02 -1.77788045e-02 3.65323216e-01 -6.61125064e-01
4.29602444e-01 1.55450359e-01 -7.05236644e-02 2.75921643e-01
-1.91752985e-01 -2.29422137e-01 4.76989329e-01 1.78819999e-01
1.47904062e+00 -5.96725106e-01 3.99121076e-01 -4.61557925e-01
4.21923608e-01 -1.20038398e-01 3.83683860e-01 9.82048213e-01
-4.27318990e-01 9.98361230e-01 7.57250011e-01 -3.18735778e-01
-1.36804593e+00 -1.17474711e+00 -1.74231425e-01 8.82937729e-01
3.19351286e-01 -5.59253454e-01 -1.08197057e+00 -3.73206079e-01
-2.08288491e-01 2.46872947e-01 -7.29516387e-01 1.88756436e-02
-2.85059392e-01 -7.45642960e-01 2.37563923e-01 1.04472123e-01
1.74712732e-01 -1.06666982e+00 -7.71265388e-01 -1.29692197e-01
-3.12486202e-01 -1.09549022e+00 -1.92830071e-01 2.99895376e-01
-9.78278458e-01 -1.10077655e+00 -5.36787629e-01 -8.72928679e-01
1.09388471e+00 4.77300346e-01 1.15694261e+00 5.32647967e-01
-4.62758780e-01 4.51884419e-01 -6.50330603e-01 1.27627388e-01
-5.58311582e-01 -3.35683405e-01 1.20758405e-02 -1.02377929e-01
3.83263856e-01 -5.14899492e-01 -6.49956584e-01 5.17625451e-01
-1.05050218e+00 -3.30486238e-01 6.01455986e-01 4.35386509e-01
6.22608364e-01 9.28995848e-01 -4.64769267e-03 -7.58657396e-01
9.98606086e-01 -4.65946132e-03 -3.22276741e-01 1.02637440e-01
-5.76921046e-01 -4.17629369e-02 5.78018844e-01 -3.90645444e-01
-5.49206018e-01 1.55377820e-01 1.80422485e-01 -2.30237111e-01
-5.25268853e-01 2.55271167e-01 -9.84098986e-02 -3.42799932e-01
1.10706115e+00 -1.36008829e-01 -1.32094428e-01 -4.17482853e-01
3.75663996e-01 4.87490594e-01 4.95468229e-01 -2.66226202e-01
9.98104692e-01 7.31719136e-01 -7.33637959e-02 -9.48664546e-01
-3.34023714e-01 -6.43179297e-01 -7.78554380e-01 -3.64047199e-01
6.77492321e-01 -4.33402032e-01 -5.98364234e-01 2.06503019e-01
-7.72368550e-01 8.55350588e-03 -5.71985066e-01 2.74252713e-01
-5.62695146e-01 8.48275840e-01 -5.95318198e-01 -7.02064216e-01
2.20081404e-01 -1.18192625e+00 7.91418672e-01 8.28693956e-02
-4.17915195e-01 -6.95692360e-01 -1.60285279e-01 -2.73675285e-02
3.56426090e-01 5.40755510e-01 9.79448020e-01 -2.10261643e-01
-3.09252292e-01 -2.28201032e-01 -2.41557285e-01 3.74520332e-01
1.65328294e-01 3.97460282e-01 -8.63775551e-01 -2.89327532e-01
4.69809592e-01 1.76947802e-01 8.68884802e-01 5.21982729e-01
1.03034639e+00 -3.99123490e-01 -5.76003827e-02 3.62641424e-01
1.68868959e+00 2.81793088e-01 9.99785721e-01 5.06907463e-01
3.64667565e-01 1.00699520e+00 3.12877476e-01 2.95217723e-01
-7.09972754e-02 5.42281091e-01 3.35096836e-01 -4.84086424e-01
-2.27837831e-01 2.34842062e-01 3.00816000e-01 1.02554381e+00
-2.12544441e-01 5.14676124e-02 -8.30797493e-01 3.41366261e-01
-1.58398211e+00 -1.19210172e+00 -3.28580827e-01 2.26098323e+00
6.08801723e-01 4.48158413e-01 2.67214507e-01 5.77039897e-01
9.55604851e-01 -2.04784349e-01 -2.32322127e-01 -4.12340760e-01
-2.99414337e-01 -9.16026067e-03 4.96457756e-01 7.08121359e-01
-1.15403211e+00 6.83238268e-01 8.11598873e+00 8.84823978e-01
-7.16867983e-01 -1.47779316e-01 6.52331471e-01 1.09546609e-01
-1.94026053e-01 1.50563091e-01 -1.02978893e-01 5.61414301e-01
6.37016118e-01 -3.83793078e-02 3.82859051e-01 6.43032372e-01
4.12146747e-01 -4.06504840e-01 -8.30650806e-01 9.21714127e-01
9.59762931e-02 -7.94142067e-01 5.88219911e-02 5.93295209e-02
7.78526962e-01 -2.77012438e-01 1.38550684e-01 -3.72247607e-01
2.19692573e-01 -1.05924463e+00 7.02432454e-01 7.40716636e-01
4.95279223e-01 -6.11934364e-01 6.81052566e-01 9.41512063e-02
-1.11610365e+00 -2.84198880e-01 -6.21687174e-01 -2.10463911e-01
-2.93556064e-01 6.38445079e-01 -2.13831678e-01 6.95389509e-02
9.49356914e-01 1.80103585e-01 -1.31284010e+00 1.34130979e+00
6.01413101e-02 3.08328569e-01 -3.94986808e-01 1.20832779e-01
-4.12779339e-02 -3.43814284e-01 3.95654321e-01 1.24945748e+00
4.41112429e-01 -1.75442994e-02 1.19269572e-01 7.27380335e-01
4.56368029e-01 1.99748337e-01 -5.59801042e-01 3.18118930e-01
4.39095557e-01 1.20008147e+00 -1.52900243e+00 -4.30340052e-01
-3.82708699e-01 1.02856982e+00 -2.27954667e-02 5.50526261e-01
-3.81569147e-01 -6.62191987e-01 3.79324883e-01 1.77077845e-01
3.19781274e-01 -3.01571667e-01 -6.91510439e-01 -8.09226811e-01
6.83444813e-02 -1.04995453e+00 3.25975955e-01 -8.33968043e-01
-1.27858102e+00 6.62288666e-01 -1.34839108e-02 -1.57001412e+00
4.92169708e-02 -4.65915531e-01 -4.62667495e-01 4.13734466e-01
-9.48478460e-01 -3.33147407e-01 -6.22235537e-01 4.82734531e-01
2.21158981e-01 1.32990807e-01 7.35324204e-01 1.43073127e-01
-1.60330847e-01 4.07381177e-01 2.46291429e-01 1.45485640e-01
5.19233227e-01 -1.52712166e+00 3.90455276e-01 1.00091279e+00
1.92907661e-01 7.23827779e-01 1.02393258e+00 -3.98053467e-01
-5.82486570e-01 -5.03844440e-01 6.94175422e-01 -5.78609824e-01
6.07760966e-01 -4.31737214e-01 -9.49500203e-01 -8.32803100e-02
9.19788033e-02 -2.60399759e-01 6.51493847e-01 -3.06036174e-02
-4.04610127e-01 -2.87989248e-02 -1.22851956e+00 7.15995967e-01
9.35504019e-01 -4.59256560e-01 -4.98049021e-01 9.32889730e-02
5.07137179e-01 2.02209249e-01 -9.77048039e-01 3.42365026e-01
4.11188811e-01 -1.69943881e+00 9.82050300e-01 -2.45127473e-02
2.10423082e-01 -6.71444833e-01 -1.08451441e-01 -1.15221560e+00
-9.58998501e-01 -4.49750870e-01 3.21799368e-01 9.13507164e-01
3.53772551e-01 -3.42256546e-01 6.86337352e-01 2.62865126e-01
2.20789120e-01 -2.54308015e-01 -9.15462911e-01 -1.00052214e+00
-1.33850783e-01 -3.67358923e-01 1.90432161e-01 9.60230052e-01
1.67516142e-01 1.74526989e-01 -6.36709929e-02 4.52275425e-02
8.80455971e-01 -1.60500005e-01 5.25777876e-01 -1.44084227e+00
-5.49551882e-02 -9.27347898e-01 -1.07247138e+00 -4.80580628e-01
-6.09270155e-01 -4.05199170e-01 1.33728713e-01 -1.40823638e+00
3.26577157e-01 -1.50975481e-01 -2.82944322e-01 1.66673228e-01
-3.04821163e-01 6.12037480e-01 3.66703361e-01 5.83311319e-01
-7.10587025e-01 -1.41058967e-01 1.05608511e+00 -3.24641764e-02
-9.32514518e-02 -2.26846814e-01 -8.75439525e-01 1.03542459e+00
7.18393207e-01 -4.51776505e-01 -1.94313332e-01 -1.14389218e-01
3.84948254e-01 -4.34303612e-01 1.56484142e-01 -1.49654734e+00
3.01384449e-01 1.20011717e-01 6.34946644e-01 -2.77428836e-01
2.07223799e-02 -1.32742453e+00 2.80645341e-01 5.94448924e-01
-3.32591832e-01 1.81023777e-01 2.20567174e-02 4.34903979e-01
-2.63609260e-01 -4.48593944e-01 1.12221742e+00 -1.89026877e-01
-7.15288877e-01 -2.94724494e-01 -7.16843307e-01 -1.92958757e-01
1.03099406e+00 -7.65625894e-01 -2.16701061e-01 -5.48626781e-01
-1.21778631e+00 -1.24868736e-01 9.83977139e-01 2.44093895e-01
6.72639549e-01 -1.23330879e+00 -5.25596917e-01 3.31362724e-01
2.01049358e-01 -5.20238101e-01 1.23042434e-01 7.24701881e-01
-7.83542573e-01 -2.08213478e-01 -2.81288505e-01 -8.58524144e-01
-1.56456935e+00 8.20709705e-01 2.70121396e-01 2.14322228e-02
-5.43762267e-01 4.62186903e-01 2.36616120e-01 -4.37898980e-03
4.34064418e-01 -5.52369893e-01 -4.20511723e-01 3.44127983e-01
5.78679681e-01 4.58817959e-01 8.22055563e-02 -8.36558104e-01
-1.85755908e-01 1.17452145e+00 1.91150784e-01 -6.21286035e-02
9.96089518e-01 -4.95140553e-01 -3.24128181e-01 4.72868651e-01
1.12313080e+00 1.68690503e-01 -1.12080443e+00 9.40023214e-02
4.43607658e-01 -5.47024429e-01 -2.06308603e-01 -4.94333476e-01
-7.50162482e-01 6.99214101e-01 1.08545625e+00 8.74981344e-01
1.58726561e+00 1.13905624e-01 1.14835463e-02 3.02482903e-01
1.54621869e-01 -1.37952149e+00 3.46491784e-01 3.08776438e-01
7.40818441e-01 -1.11119115e+00 2.71939367e-01 -5.90217173e-01
-4.56376910e-01 1.13171065e+00 2.79677361e-01 -2.11650386e-01
5.72163820e-01 6.10996187e-01 4.28761572e-01 -4.31465000e-01
-5.36990464e-01 -6.83091164e-01 1.77171201e-01 7.65616536e-01
1.51627451e-01 -7.14440644e-02 -4.34976369e-01 1.58741549e-01
-2.93499321e-01 -3.56809139e-01 8.01412523e-01 8.05265367e-01
-9.69354689e-01 -7.34515309e-01 -8.76920819e-01 4.45526719e-01
-6.02843344e-01 -4.63992953e-02 -8.34397554e-01 6.79351687e-01
3.17895204e-01 1.20244157e+00 1.26066953e-01 -6.84350550e-01
5.50885975e-01 -2.29062065e-01 3.44413847e-01 -4.78011280e-01
-5.21464229e-01 4.84389663e-01 -2.66600966e-01 -7.08479285e-01
-6.34728134e-01 -5.45998991e-01 -8.97663057e-01 -3.43062133e-01
-3.54484022e-01 2.17148736e-01 4.40580130e-01 4.09151763e-01
1.86686814e-01 4.58876967e-01 7.96058595e-01 -9.36949849e-01
-1.50186718e-01 -8.69335830e-01 -5.99944651e-01 9.14379835e-01
3.30296487e-01 -3.20094764e-01 -1.00789905e+00 4.39274281e-01] | [11.460186958312988, -2.2179174423217773] |
a7975424-bfab-438e-817f-27f0528214b1 | a-versatile-multi-agent-reinforcement | 2306.07542 | null | https://arxiv.org/abs/2306.07542v1 | https://arxiv.org/pdf/2306.07542v1.pdf | A Versatile Multi-Agent Reinforcement Learning Benchmark for Inventory Management | Multi-agent reinforcement learning (MARL) models multiple agents that interact and learn within a shared environment. This paradigm is applicable to various industrial scenarios such as autonomous driving, quantitative trading, and inventory management. However, applying MARL to these real-world scenarios is impeded by many challenges such as scaling up, complex agent interactions, and non-stationary dynamics. To incentivize the research of MARL on these challenges, we develop MABIM (Multi-Agent Benchmark for Inventory Management) which is a multi-echelon, multi-commodity inventory management simulator that can generate versatile tasks with these different challenging properties. Based on MABIM, we evaluate the performance of classic operations research (OR) methods and popular MARL algorithms on these challenging tasks to highlight their weaknesses and potential. | ['Jiang Bian', 'Lei Song', 'Li Zhao', 'Chuheng Zhang', 'Wei Jiang', 'Zhihao Liu', 'Xianliang Yang'] | 2023-06-13 | null | null | null | null | ['multi-agent-reinforcement-learning'] | ['methodology'] | [-6.03913665e-01 -2.89040387e-01 -6.36973530e-02 2.64524132e-01
-2.71159112e-01 -6.82997286e-01 4.60751444e-01 1.90463260e-01
-4.09315020e-01 1.46374369e+00 -5.33930361e-01 -2.49823421e-01
-6.13347173e-01 -7.00894237e-01 -6.71226501e-01 -6.75699830e-01
-8.47208798e-01 1.21502328e+00 2.63827205e-01 -7.68169582e-01
1.09066859e-01 3.98921251e-01 -1.52372479e+00 -3.32820088e-01
9.28418040e-01 8.36313426e-01 3.45844567e-01 6.39575183e-01
3.11830640e-02 9.89498258e-01 -9.43273544e-01 -3.57913896e-02
5.40010273e-01 -1.18765928e-01 -3.81259620e-01 -6.32496644e-03
-5.37109256e-01 -5.27279496e-01 5.70221953e-02 7.93973684e-01
3.81774008e-01 3.57732862e-01 3.67258728e-01 -2.23122907e+00
-3.44367117e-01 7.85559297e-01 -8.04458976e-01 2.42400929e-01
4.76941615e-02 5.34636676e-01 6.29407465e-01 -2.09767655e-01
4.24337536e-01 1.51538944e+00 2.22643569e-01 2.71873772e-01
-8.66534591e-01 -6.26531124e-01 5.18386960e-01 2.38342136e-02
-7.63622880e-01 3.29666555e-01 2.64427960e-01 -2.36097962e-01
9.39743221e-01 -1.61566027e-02 7.88183570e-01 9.72476959e-01
8.99061501e-01 7.06011415e-01 1.40910840e+00 7.48304874e-02
6.44574761e-01 6.21427111e-02 -2.27379516e-01 4.04616535e-01
5.64493537e-01 5.01820564e-01 -2.43749768e-01 -1.37093991e-01
8.58553886e-01 6.12600881e-04 4.95129406e-01 -5.41081607e-01
-1.34190583e+00 9.91883278e-01 4.62013558e-02 -3.51507246e-01
-9.16386783e-01 2.49212310e-01 5.09966195e-01 8.58220577e-01
1.32602394e-01 8.74276042e-01 -6.55546308e-01 -1.79523900e-01
-9.12926272e-02 8.48743081e-01 1.09410036e+00 1.10346735e+00
5.72042584e-01 4.83494669e-01 -9.13490653e-02 2.81421989e-01
3.25087607e-01 6.46853566e-01 2.04549760e-01 -1.63325953e+00
2.58600652e-01 7.44260028e-02 6.91632271e-01 -4.21275288e-01
-7.93803275e-01 -3.62881005e-01 -7.11437643e-01 9.47849452e-01
2.27262110e-01 -7.13580251e-01 -3.73463929e-01 1.40796173e+00
4.13837612e-01 5.37044890e-02 3.64423305e-01 8.64567935e-01
-1.07666878e-02 7.61282563e-01 -6.94453940e-02 -4.43618774e-01
1.15617752e+00 -1.32940292e+00 -8.25667560e-01 6.10001478e-03
1.64422348e-01 -5.53778291e-01 6.70984387e-01 5.30910790e-01
-1.30620456e+00 -2.89470971e-01 -9.50783551e-01 8.25897574e-01
-6.82341218e-01 -6.15014613e-01 8.75357985e-01 2.12994888e-01
-8.28391552e-01 5.65351844e-01 -7.48039007e-01 4.43194285e-02
-1.23621918e-01 5.30132473e-01 2.74134010e-01 1.50628999e-01
-1.31228161e+00 1.17176247e+00 4.18307245e-01 -8.81401077e-02
-1.34907126e+00 -7.68901348e-01 -5.94340205e-01 6.55019060e-02
1.16082168e+00 -5.66697180e-01 1.64773810e+00 -6.50895238e-01
-1.86970067e+00 -1.59116909e-01 8.87276649e-01 -5.08925796e-01
7.81135917e-01 -4.04981934e-02 -4.20091212e-01 -1.50345668e-01
1.20813943e-01 3.59512061e-01 5.90089440e-01 -1.27722907e+00
-9.74259973e-01 4.93373834e-02 5.38836598e-01 4.69315052e-01
3.05267036e-01 -1.31824970e-01 3.55789006e-01 -5.23819387e-01
-8.34886193e-01 -1.22087228e+00 -6.41283154e-01 -6.98186636e-01
-5.24267554e-02 -1.74079448e-01 7.74132013e-01 -6.47170097e-02
4.17802870e-01 -1.95450938e+00 2.36675173e-01 3.78480069e-02
-4.28145602e-02 6.80403337e-02 -3.85405362e-01 1.06525302e+00
4.28139150e-01 2.97926869e-02 1.91153973e-01 -5.03251925e-02
6.91106677e-01 4.44418103e-01 -1.27311543e-01 1.21214256e-01
3.04815710e-01 8.73454571e-01 -1.20992243e+00 -3.17175627e-01
2.15503484e-01 -3.09312284e-01 -3.28722328e-01 3.44919920e-01
-8.39380443e-01 5.48722386e-01 -6.00787818e-01 9.03913677e-01
7.08472908e-01 -2.92448588e-02 3.27723682e-01 4.74112332e-01
-3.55235189e-01 -4.46088314e-01 -1.49591470e+00 1.07655406e+00
-5.91347456e-01 5.59132583e-02 6.92788959e-01 -8.18058372e-01
5.77455878e-01 9.68003124e-02 8.62606168e-01 -1.02536678e+00
1.39275283e-01 2.57698357e-01 2.82192618e-01 -3.87142330e-01
5.54832816e-01 -8.74508545e-02 -2.37129927e-01 8.83728623e-01
-1.49439439e-01 -4.41827446e-01 7.46246040e-01 1.70230772e-02
1.20408750e+00 4.36742380e-02 1.85294583e-01 -4.55803782e-01
3.49803790e-02 2.91993737e-01 7.70864010e-01 9.79087234e-01
-5.18948972e-01 -3.67316723e-01 6.51850522e-01 -4.54978168e-01
-8.24099541e-01 -1.32343531e+00 3.33430707e-01 1.26522553e+00
5.72439492e-01 1.91020519e-01 -2.70560980e-01 -4.68213707e-01
8.19842994e-01 6.75721765e-01 -3.71007711e-01 1.70713305e-01
-4.46715444e-01 -9.05711770e-01 -5.24210334e-02 1.66810304e-01
4.25829709e-01 -1.56774175e+00 -8.05496395e-01 8.32741916e-01
5.20261705e-01 -1.18251395e+00 -2.74143159e-01 2.19582036e-01
-2.10270643e-01 -1.28127575e+00 -4.39471066e-01 -6.63233161e-01
2.37305969e-01 2.87442058e-01 1.36269641e+00 -1.89430147e-01
-3.42428029e-01 5.95990658e-01 -1.82894409e-01 -9.87497926e-01
-7.34745860e-01 4.60713729e-02 3.25322628e-01 -2.95643866e-01
-2.04697058e-01 -3.57315302e-01 -4.36470985e-01 5.69846272e-01
-1.12181258e+00 -6.04668744e-02 8.41563880e-01 8.81071448e-01
3.31357718e-01 5.46807408e-01 1.30648553e+00 -7.14929283e-01
1.23252499e+00 -7.27275491e-01 -1.29226744e+00 2.90770233e-01
-7.10362554e-01 -5.02492562e-02 8.19591880e-01 -5.83806396e-01
-9.24711525e-01 -3.79625738e-01 4.96063948e-01 -1.45637780e-01
1.71048611e-01 5.75953484e-01 3.21773410e-01 1.01100817e-01
-1.90814063e-01 9.01347622e-02 8.15987706e-01 1.30183637e-01
2.58595906e-02 2.91079938e-01 1.08811997e-01 -8.07524025e-01
8.71555805e-01 -3.56109105e-02 3.15470457e-01 -1.95248246e-01
-2.26205796e-01 1.54176578e-01 -7.41250962e-02 -2.72196978e-01
5.32743990e-01 -8.96192729e-01 -1.67126942e+00 7.33705640e-01
-7.46253848e-01 -8.53964806e-01 -4.16504085e-01 5.68829894e-01
-1.03394639e+00 -1.38353258e-01 -8.89287949e-01 -8.31115127e-01
-1.01683401e-01 -1.56926417e+00 5.17395139e-01 3.01564276e-01
3.12992394e-01 -1.02148438e+00 4.78971392e-01 6.98768348e-02
9.38400269e-01 4.33639526e-01 9.44283664e-01 -4.89207447e-01
-8.73034835e-01 2.45424315e-01 2.26962656e-01 1.29711464e-01
1.51969761e-01 1.50191382e-01 -2.19140097e-01 -7.81688690e-01
-2.61495888e-01 -6.91041410e-01 -2.40215547e-02 1.97582960e-01
6.93566084e-01 -2.62309581e-01 -5.18403836e-02 -1.03246547e-01
1.35898674e+00 6.15026653e-01 4.90297657e-03 8.91716957e-01
1.20635606e-01 5.17031431e-01 1.30761492e+00 9.28300560e-01
7.60404944e-01 5.97107410e-01 9.62410748e-01 2.67294310e-02
7.09330738e-01 2.90541172e-01 5.83251953e-01 8.41798782e-01
5.15593775e-02 -2.67927825e-01 -7.32148767e-01 1.79807037e-01
-2.27530336e+00 -9.74954605e-01 2.70896912e-01 1.94349790e+00
6.81894660e-01 4.90650833e-01 5.52654862e-01 -4.99323130e-01
5.57554543e-01 -3.42010856e-02 -1.17028832e+00 -7.98125625e-01
-1.36752337e-01 -6.89315945e-02 6.31980121e-01 2.52080709e-01
-7.57446766e-01 8.14562678e-01 6.84303570e+00 5.12158394e-01
-8.69579017e-01 -1.75680441e-03 4.43228811e-01 -2.88263261e-01
1.24278829e-01 -4.00156349e-01 -6.23273194e-01 6.55962944e-01
1.13861120e+00 -8.01556408e-01 1.20651805e+00 8.67322147e-01
4.84761328e-01 -2.83111513e-01 -8.93721104e-01 5.06990969e-01
-5.59786141e-01 -1.31646705e+00 -1.77145764e-01 2.98015743e-01
9.32722092e-01 2.48958498e-01 9.76498127e-02 1.12256134e+00
1.22739577e+00 -9.09065902e-01 6.05336547e-01 2.65655637e-01
-9.65389386e-02 -1.14419103e+00 9.94141757e-01 3.54326099e-01
-1.08336568e+00 -4.84855592e-01 -1.62932903e-01 -2.43382931e-01
2.94191509e-01 6.01328127e-02 -6.99532509e-01 6.56146824e-01
6.16726518e-01 2.30190635e-01 -1.57113045e-01 8.99914563e-01
3.60487908e-01 -4.38069291e-02 -1.82272062e-01 -3.38008702e-01
6.59350693e-01 -6.15274727e-01 3.73043478e-01 4.89783466e-01
1.63427919e-01 -3.17856103e-01 9.67065632e-01 6.91594005e-01
1.43569738e-01 -2.17161968e-01 -6.93094134e-01 -1.58568293e-01
6.40158772e-01 1.46317708e+00 -7.01475501e-01 -2.05038309e-01
-4.37914193e-01 4.84429657e-01 7.42194653e-02 3.54824513e-01
-1.25087953e+00 -3.79986316e-01 1.22568083e+00 -2.25358278e-01
1.05086908e-01 -5.87298989e-01 3.05401623e-01 -5.96707463e-01
-1.71159714e-01 -1.30113566e+00 1.76649004e-01 -5.45983315e-01
-1.53698218e+00 3.26703489e-01 4.58744690e-02 -1.26048875e+00
-5.86211085e-01 -6.36036336e-01 -5.77712893e-01 4.27156538e-01
-2.04070783e+00 -7.51921177e-01 -9.10505205e-02 4.84737217e-01
7.23851025e-01 -6.18505836e-01 5.63652337e-01 4.47241664e-02
-8.47282290e-01 -7.61838108e-02 6.81689858e-01 -3.82331550e-01
5.34400344e-01 -1.70853162e+00 3.46423626e-01 4.48232032e-02
-8.16373348e-01 -1.49121257e-02 6.54111028e-01 -5.93249977e-01
-2.06211615e+00 -1.25285029e+00 -3.70277405e-01 -1.61606640e-01
1.27120054e+00 -2.54232824e-01 -5.60373664e-01 7.45066822e-01
7.59462774e-01 -2.28058621e-01 1.99950963e-01 -2.48100370e-01
4.63900745e-01 -4.77714151e-01 -1.18043733e+00 6.61814034e-01
5.68844795e-01 2.20600411e-01 -1.76153686e-02 5.55528641e-01
9.02976692e-01 -5.11719406e-01 -1.18162823e+00 2.11309671e-01
2.04889536e-01 -6.40333474e-01 7.26695061e-01 -7.53641665e-01
7.97130018e-02 -1.03215143e-01 5.13829701e-02 -2.17777205e+00
-3.29797596e-01 -1.19132245e+00 -1.24728516e-01 1.12800443e+00
2.85673320e-01 -1.14603925e+00 2.87588745e-01 4.23650593e-01
-1.25730366e-01 -4.69376415e-01 -8.97976577e-01 -1.38093591e+00
2.78852612e-01 4.24870968e-01 1.13639128e+00 7.97721028e-01
-1.62428543e-02 1.42914698e-01 -4.14441198e-01 2.35464096e-01
9.32334781e-01 2.64620632e-01 8.70395541e-01 -1.03722191e+00
-4.96756166e-01 -4.96098846e-01 2.23328397e-01 -1.10829137e-01
3.77970755e-01 -2.16575772e-01 9.09224078e-02 -1.27577734e+00
-9.40146372e-02 -7.75073051e-01 -5.81571579e-01 2.05265835e-01
1.15093209e-01 -3.58458549e-01 6.85294092e-01 7.06359297e-02
-1.13942015e+00 8.01383197e-01 1.47983587e+00 -3.78436446e-01
-2.02485040e-01 5.02430908e-02 -3.76282394e-01 3.83276761e-01
1.17296815e+00 -2.61264920e-01 -4.72134769e-01 -2.43602991e-01
3.05073082e-01 6.36499286e-01 4.35616709e-02 -9.18359637e-01
4.94585894e-02 -1.02460837e+00 -1.48236915e-01 -5.30067623e-01
2.24409550e-01 -9.49663341e-01 3.81732374e-01 8.24194252e-01
2.16260552e-03 1.17230535e+00 2.81917423e-01 4.74433422e-01
-2.24102020e-01 -1.27229333e-01 5.30648232e-01 -5.18937469e-01
-8.68960559e-01 4.49812979e-01 -8.36474061e-01 3.23988557e-01
1.76880813e+00 4.23528671e-01 -9.29352403e-01 -4.23167467e-01
-4.58349317e-01 1.30529094e+00 3.37572962e-01 5.69546580e-01
1.54964924e-01 -1.05400348e+00 -8.31648052e-01 -8.64032358e-02
-1.74777359e-01 -8.35823789e-02 3.87769304e-02 6.36696637e-01
-7.32018590e-01 3.93067330e-01 -9.42689300e-01 -1.72877505e-01
-6.15555644e-01 9.06814754e-01 4.47030455e-01 -7.73828626e-01
-2.86028206e-01 -1.99668124e-01 -2.23114677e-02 -6.10487103e-01
1.78769559e-01 -1.75663188e-01 1.23742029e-01 1.44984007e-01
3.01506519e-01 7.35474706e-01 -1.41060770e-01 1.54643178e-01
-1.49648547e-01 5.55498293e-03 -1.99122846e-01 -1.73982531e-01
1.53593886e+00 -7.89433345e-02 4.76829000e-02 5.30765414e-01
2.48317212e-01 -4.08097178e-01 -1.47616363e+00 -2.96743214e-02
1.19608261e-01 -6.39946014e-03 -1.82103157e-01 -9.80682552e-01
-1.09635139e+00 2.47498348e-01 4.00006473e-01 9.26666379e-01
6.85247660e-01 -5.34817159e-01 6.17814958e-01 5.05977929e-01
9.26484942e-01 -1.50254095e+00 4.06431198e-01 7.59786665e-01
1.00047565e+00 -1.36523747e+00 -1.04480639e-01 -3.93542871e-02
-1.02579308e+00 8.56064975e-01 1.02544677e+00 -3.46794605e-01
6.75562680e-01 6.49848163e-01 2.56665319e-01 -1.03110991e-01
-1.35638261e+00 -2.42450878e-01 -8.94001842e-01 6.69774354e-01
-4.15876567e-01 4.17858034e-01 -1.40163481e-01 2.32554987e-01
7.01104030e-02 3.22474614e-02 1.06932509e+00 1.43008125e+00
-1.59382358e-01 -1.01868582e+00 -4.16723728e-01 3.53502691e-01
-2.93011636e-01 4.46167380e-01 6.47612736e-02 1.36454940e+00
-2.51629651e-01 8.55396271e-01 2.74016142e-01 1.75601631e-01
3.46343935e-01 -4.50381339e-01 1.92894801e-01 -3.89611304e-01
-9.11673903e-01 -5.61994091e-02 -8.23009610e-02 -4.36836481e-01
-2.70765778e-02 -6.28194153e-01 -1.36897695e+00 -7.07890332e-01
8.79941285e-02 4.05674785e-01 8.20380330e-01 6.68461204e-01
5.59253216e-01 1.00747180e+00 1.16145968e+00 -9.55461621e-01
-1.32241285e+00 -7.75854290e-01 -1.18902183e+00 1.64996475e-01
3.74958843e-01 -1.12940371e+00 -2.25047156e-01 -7.44884312e-01] | [3.7728445529937744, 1.969111442565918] |
7bf484fc-2da5-4e47-bc08-6d09cbee0c9b | on-the-complexity-of-dark-chinese-chess | 2112.02989 | null | https://arxiv.org/abs/2112.02989v1 | https://arxiv.org/pdf/2112.02989v1.pdf | On the complexity of Dark Chinese Chess | This paper provides a complexity analysis for the game of dark Chinese chess (a.k.a. "JieQi"), a variation of Chinese chess. Dark Chinese chess combines some of the most complicated aspects of board and card games, such as long-term strategy or planning, large state space, stochastic, and imperfect-information, which make it closer to the real world decision-making problem and pose great challenges to game AI. Here we design a self-play program to calculate the game tree complexity and average information set size of the game, and propose an algorithm to calculate the number of information sets. | ['Tongwei Lu', 'Cong Wang'] | 2021-12-06 | null | null | null | null | ['card-games'] | ['playing-games'] | [-4.76973385e-01 -1.77704707e-01 2.48303235e-01 2.32907474e-01
-2.53808498e-01 -8.42788696e-01 1.75499376e-02 -1.72078967e-01
-7.60767698e-01 1.16869080e+00 -2.75301844e-01 -8.69424820e-01
-5.63504696e-01 -1.06739104e+00 -6.48508668e-02 -5.99706948e-01
-3.20593894e-01 6.50761902e-01 9.33947206e-01 -9.78605449e-01
6.11410081e-01 -4.87454273e-02 -9.27635789e-01 -2.70487219e-01
6.31439626e-01 8.65523815e-01 5.36741972e-01 9.68184829e-01
3.36416923e-02 1.08444738e+00 -7.86155045e-01 -3.53050679e-01
6.36986136e-01 -6.41912520e-01 -9.66341972e-01 -3.16012412e-01
-1.16625929e+00 -3.50915015e-01 -5.06168783e-01 1.12284839e+00
5.23670197e-01 1.75879076e-01 3.69289786e-01 -1.33750558e+00
2.29903460e-01 9.63518322e-01 -4.53787327e-01 4.19583172e-01
2.87077665e-01 3.34492058e-01 7.69268692e-01 2.38990068e-01
4.43127304e-01 7.32797205e-01 3.46304685e-01 5.66648901e-01
-4.75591868e-01 -8.39686394e-01 1.01228923e-01 3.09631467e-01
-1.49518847e+00 1.22940607e-01 2.35578477e-01 -2.78623253e-01
8.11581135e-01 4.74279761e-01 1.18961883e+00 2.59544373e-01
6.03474021e-01 6.72390342e-01 1.33586824e+00 -2.45219305e-01
8.06650221e-01 -4.94547397e-01 -1.23965859e-01 5.14728367e-01
4.85357761e-01 4.70827162e-01 -1.50794923e-01 -8.40986893e-02
9.47648287e-01 -2.99461693e-01 2.26629511e-01 -3.78431231e-02
-6.21795356e-01 8.07382584e-01 5.89367636e-02 1.68471321e-01
-1.73183054e-01 2.34241679e-01 1.28454760e-01 4.07278270e-01
-1.05401702e-01 7.72340000e-01 -3.94289404e-01 -9.30867434e-01
-6.47607982e-01 6.03591144e-01 1.04150856e+00 8.20783317e-01
4.54870909e-01 7.58062601e-02 1.59803554e-01 2.19312520e-03
1.90604597e-01 2.89342135e-01 3.37756008e-01 -1.07216179e+00
7.12389112e-01 6.02036953e-01 3.50374371e-01 -8.77402484e-01
-7.37013340e-01 -3.55122596e-01 -7.26660132e-01 5.72129607e-01
4.56255645e-01 -7.92672932e-01 -3.53382796e-01 1.35614359e+00
-1.31537709e-02 -1.11911893e-01 2.04289198e-01 7.20272720e-01
3.14922303e-01 5.97135007e-01 -5.76267898e-01 -1.80486664e-01
1.08911872e+00 -7.11329460e-01 -5.36920965e-01 -5.23609936e-01
2.76609957e-01 -2.72543013e-01 6.79652154e-01 5.71730137e-01
-1.49083614e+00 4.79127374e-03 -1.25712597e+00 5.80368340e-01
-1.70821428e-01 -4.64119375e-01 8.54036093e-01 9.84355509e-01
-1.03474641e+00 3.63928139e-01 -8.07074487e-01 7.09125176e-02
5.52584603e-02 5.11886656e-01 1.95515931e-01 1.84251830e-01
-1.35799968e+00 1.03295755e+00 9.26442683e-01 -1.09638851e-02
-7.99628735e-01 -1.70392431e-02 -6.12568617e-01 2.30308384e-01
1.18596315e+00 -4.16286826e-01 1.47812486e+00 -2.06241012e-01
-2.03979588e+00 3.22103709e-01 5.22733450e-01 -3.08867931e-01
4.71857131e-01 2.81150252e-01 1.05228864e-01 -3.57823998e-01
4.54008989e-02 -7.91269764e-02 1.67506728e-02 -5.24929821e-01
-8.72006118e-01 -2.56829679e-01 7.75148630e-01 6.92747474e-01
2.37190828e-01 2.55266130e-01 -5.47207177e-01 -2.26863831e-01
8.69364962e-02 -8.96064699e-01 -1.29003644e+00 -1.04139745e+00
-4.07558590e-01 6.50611743e-02 -2.08741844e-01 -3.41009229e-01
1.99967170e+00 -1.81946790e+00 5.39213978e-02 4.03291881e-01
3.97292197e-01 1.59593657e-01 1.97452307e-01 9.36734200e-01
3.39386642e-01 1.72919422e-01 1.40030876e-01 6.03325665e-01
9.87145752e-02 4.61084336e-01 2.54185885e-01 3.23303114e-03
-4.01740044e-01 1.03053129e+00 -1.02850878e+00 -1.64489254e-01
-1.85593858e-01 -1.09586751e+00 -7.12809503e-01 6.80571124e-02
-3.13726962e-01 2.82413274e-01 -8.20303738e-01 2.50249684e-01
4.80129212e-01 -1.72162905e-01 4.80418950e-01 8.13050628e-01
-5.23471415e-01 3.59048903e-01 -1.63340318e+00 1.22863221e+00
-6.90644756e-02 1.91825867e-01 -5.39199263e-02 -6.38932288e-01
5.46056092e-01 3.80087793e-02 2.03479350e-01 -7.59312391e-01
5.75239539e-01 8.89710858e-02 6.31886780e-01 -2.79961437e-01
8.94978106e-01 -9.45734680e-02 -7.86553681e-01 7.57843018e-01
-4.16536212e-01 -6.53594792e-01 7.03198314e-01 3.78472537e-01
1.40691674e+00 -2.42574111e-01 9.13050115e-01 -4.17643696e-01
1.57407060e-01 1.64924771e-01 9.05333757e-01 1.11215258e+00
-9.95944068e-02 2.02548906e-01 1.18043411e+00 -4.53845441e-01
-7.62031376e-01 -5.57206631e-01 7.83956826e-01 7.10728049e-01
6.97513521e-01 -1.02906966e+00 -8.07655036e-01 -2.13052630e-01
-3.88399184e-01 4.57836717e-01 -3.91908169e-01 -1.58453032e-01
-3.03839505e-01 -6.65361643e-01 4.59311754e-01 3.37581247e-01
8.48037124e-01 -7.57073104e-01 -1.07259405e+00 6.63891196e-01
-1.69946492e-01 -8.56464386e-01 -5.14134109e-01 2.92694777e-01
-3.95523250e-01 -1.45960796e+00 -2.13004947e-01 -3.65259856e-01
6.04731031e-02 1.38530880e-01 1.01089346e+00 1.96349978e-01
9.30442358e-04 -1.15684979e-02 -3.89343202e-01 -7.54894674e-01
-2.63916612e-01 1.76760629e-01 -1.45841628e-01 -8.00310910e-01
-1.70030728e-01 -3.28052431e-01 -4.19589490e-01 6.30091488e-01
-6.66876733e-01 4.92971390e-01 3.55135530e-01 8.74912322e-01
2.66562998e-01 1.09341133e+00 1.63164288e-02 -7.06927538e-01
1.12282157e+00 -3.00875753e-01 -1.28247845e+00 2.35993057e-01
-3.38422149e-01 -6.58758879e-02 6.45737886e-01 -1.18491344e-01
-8.93583119e-01 -1.34596288e-01 -8.63276199e-02 4.69896972e-01
5.81585169e-01 9.68151212e-01 -2.91457474e-01 -2.04385638e-01
3.81913751e-01 3.06009442e-01 -9.00800526e-02 -2.54078209e-02
-8.13604221e-02 6.01204097e-01 1.84594885e-01 -5.54684997e-01
6.11259222e-01 -5.81764989e-03 2.38975614e-01 -3.18235159e-01
-4.25314188e-01 -2.58620501e-01 -2.41413787e-01 -3.07593226e-01
5.56160271e-01 -3.99463832e-01 -1.48886287e+00 9.90942061e-01
-7.28720307e-01 -6.30985320e-01 -3.23804468e-01 2.32914045e-01
-7.21012294e-01 2.04548150e-01 -6.74694419e-01 -1.19430161e+00
1.97228923e-01 -1.12083030e+00 1.59331277e-01 7.06936359e-01
2.63530135e-01 -5.80410838e-01 3.14603180e-01 1.54932976e-01
3.91511381e-01 1.80244893e-01 6.52209401e-01 -3.06672662e-01
-7.17307389e-01 -1.21198192e-01 1.69831961e-01 6.60567805e-02
-1.85888886e-01 -2.45947167e-01 9.91622508e-02 -2.35017970e-01
2.12294489e-01 -1.64287370e-02 3.75072323e-02 4.53093767e-01
7.77225494e-01 -4.21691030e-01 -4.01371568e-02 2.77329743e-01
1.60437524e+00 1.18279910e+00 9.50238466e-01 6.45011604e-01
3.79855419e-03 1.78742826e-01 1.03092945e+00 1.22675085e+00
7.06431091e-01 7.17346311e-01 5.04277170e-01 3.80758524e-01
7.75132775e-01 -3.36697370e-01 2.94487357e-01 1.04787219e+00
-3.43913406e-01 -4.13169175e-01 -1.03059089e+00 1.66823015e-01
-1.89747274e+00 -9.08275962e-01 9.22815595e-03 2.11716890e+00
8.42204392e-01 5.65229297e-01 4.90026087e-01 3.81273955e-01
6.63021684e-01 -9.59912166e-02 -5.50152183e-01 -5.11770129e-01
-1.16126969e-01 3.39030862e-01 1.04819524e+00 5.17447889e-01
-7.13643610e-01 1.31308281e+00 7.34286451e+00 1.27296352e+00
-6.09083652e-01 -1.69981897e-01 5.89011014e-01 -1.33887067e-01
-5.74816093e-02 5.03554285e-01 -6.75727189e-01 7.49536395e-01
7.59260476e-01 -9.90980923e-01 8.49985898e-01 3.46486717e-01
1.30348608e-01 -8.51604700e-01 -4.72473860e-01 9.43915904e-01
-4.91378605e-01 -1.38469684e+00 -5.79052448e-01 4.23296332e-01
8.67476523e-01 -2.28423923e-01 -1.05746895e-01 3.51990640e-01
1.34279716e+00 -9.00263488e-01 8.36737454e-01 2.00472921e-01
4.80657488e-01 -9.69473422e-01 9.14746642e-01 8.49575698e-01
-1.31164813e+00 -5.50780058e-01 -4.83258367e-01 -1.05838883e+00
3.60421628e-01 3.42081673e-03 -4.40868199e-01 8.74885738e-01
6.08688712e-01 3.55807506e-02 -3.48762929e-01 1.41626954e+00
-3.42438579e-01 4.19813126e-01 -5.98940253e-01 -7.14331269e-01
6.90928519e-01 -5.99292040e-01 4.41855609e-01 1.36416063e-01
4.19989198e-01 1.03786123e+00 3.80597889e-01 4.37122494e-01
6.74129128e-01 -1.81700349e-01 -2.77230799e-01 -1.55996099e-01
6.29394054e-01 6.14864290e-01 -1.06890774e+00 -1.49375826e-01
6.79797530e-02 3.90970498e-01 -1.55064330e-01 -8.81327912e-02
-7.59426355e-01 -4.74562764e-01 5.38978398e-01 -8.25362056e-02
1.07027106e-01 -5.86695910e-01 -5.03782749e-01 -9.67521906e-01
-1.98569089e-01 -1.01184046e+00 5.02058983e-01 -7.15651095e-01
-6.58670008e-01 5.16347468e-01 6.09228015e-02 -1.17924392e+00
-6.49123311e-01 -4.76126671e-01 -9.01455939e-01 6.72877967e-01
-7.98663139e-01 -3.00132692e-01 2.01826245e-02 5.38349152e-01
3.16493541e-01 -4.96604860e-01 5.31666636e-01 -1.15787715e-01
-6.61697209e-01 2.82076895e-01 4.56216693e-01 -6.50531501e-02
-3.63551885e-01 -1.32573903e+00 8.48846912e-01 9.60223079e-01
-3.71187776e-01 1.99522421e-01 8.26643527e-01 -5.86373270e-01
-1.48501503e+00 -2.47490391e-01 2.83242583e-01 -1.05879262e-01
8.46620679e-01 -2.75590032e-01 1.56676069e-01 2.93755949e-01
4.37649786e-02 -8.33340764e-01 3.56159657e-01 6.13451004e-03
4.73720342e-01 3.12686823e-02 -9.35218275e-01 1.03938484e+00
1.07075846e+00 5.91338985e-02 -3.93981338e-01 -7.06310123e-02
3.97487193e-01 -8.53038251e-01 -1.77056685e-01 -7.05524534e-02
4.66062903e-01 -1.16480386e+00 4.85344350e-01 -6.82491779e-01
9.22494307e-02 -3.68169695e-01 7.36542493e-02 -1.57921815e+00
-5.83460629e-01 -1.28996336e+00 4.60985959e-01 6.41461551e-01
3.06320786e-01 -7.67054796e-01 1.22128844e+00 8.73991609e-01
1.16210938e-01 -7.14710891e-01 -1.19425941e+00 -1.04978216e+00
1.49210721e-01 -3.74249965e-01 9.40529883e-01 3.95007998e-01
5.15332222e-01 2.55650710e-02 -7.45858550e-01 -2.25614801e-01
2.93927670e-01 1.11366630e-01 7.11510956e-01 -1.03604460e+00
-8.24121177e-01 -5.26660264e-01 -6.05671108e-01 -1.22948623e+00
-5.08280993e-01 -2.74941027e-01 1.48357078e-01 -1.35246587e+00
5.15025780e-02 -7.09943950e-01 -2.61185884e-01 2.67131597e-01
-1.06218763e-01 -2.02557623e-01 5.52508593e-01 -2.03698292e-01
-1.06561065e+00 2.16425017e-01 1.39018595e+00 1.09414905e-01
-2.74435818e-01 5.33804238e-01 -1.07738519e+00 6.91600025e-01
9.42494810e-01 -4.57369298e-01 -5.32904863e-01 -2.66938776e-01
9.93930578e-01 8.96941841e-01 -4.67344940e-01 -1.12101901e+00
7.37178385e-01 -9.43947375e-01 -5.28334737e-01 -5.13773978e-01
2.37516150e-01 -5.42500854e-01 5.31419456e-01 1.21193266e+00
9.14165229e-02 5.73826253e-01 8.50110874e-02 3.38527173e-01
-2.21835062e-01 -5.34314752e-01 4.26783830e-01 -4.84389335e-01
-9.48711574e-01 2.54322290e-01 -1.19061327e+00 4.51537251e-01
1.41851616e+00 -4.36865509e-01 -2.24245608e-01 -8.76790404e-01
-5.94984591e-01 6.73964441e-01 3.49274099e-01 -1.09249800e-01
4.31553334e-01 -8.56798291e-01 -6.49481356e-01 -1.24809265e-01
-3.33552033e-01 -3.59086767e-02 5.16733229e-01 3.06404471e-01
-1.09240663e+00 3.69107455e-01 -6.65330231e-01 1.86325058e-01
-1.20557606e+00 1.99242458e-01 2.21449420e-01 -9.77288008e-01
-2.38144174e-01 7.27559924e-01 2.58575439e-01 -1.22782946e-01
-1.42096788e-01 -1.13661252e-01 -3.00259918e-01 -3.19159061e-01
6.41207397e-01 4.90528017e-01 -1.86275363e-01 2.44938023e-02
-2.56444603e-01 1.13277450e-01 1.78779125e-01 -4.98986900e-01
1.30511081e+00 -2.87849665e-01 -2.07484514e-01 1.77379549e-01
6.93921894e-02 -1.32013664e-01 -9.68324184e-01 4.87597398e-02
1.95074156e-01 -6.67454183e-01 1.88191831e-02 -1.00947988e+00
-1.09640777e+00 4.83315051e-01 6.14794195e-02 4.57556248e-01
1.14169919e+00 -3.73049915e-01 5.89231670e-01 6.54985905e-01
1.39861655e+00 -1.30507231e+00 -2.26910785e-01 1.21995831e+00
3.35661381e-01 -3.80278051e-01 9.34373029e-03 -2.26032242e-01
-1.05142307e+00 7.91097164e-01 6.93193913e-01 -2.22457081e-01
6.22836709e-01 7.94513226e-01 -7.56928548e-02 -4.23800647e-02
-1.05635083e+00 -4.32161540e-01 -5.63061833e-01 5.41815341e-01
-5.52308619e-01 3.01992834e-01 -8.33931625e-01 1.47513592e+00
-7.76717901e-01 2.94602454e-01 1.29326427e+00 1.32777596e+00
-4.82632786e-01 -1.31067908e+00 -3.26660424e-01 5.00483811e-01
-5.01228750e-01 -2.94967473e-01 -4.11725610e-01 7.46028244e-01
-1.36190942e-02 1.02271998e+00 -1.62019387e-01 -7.34530747e-01
3.84288281e-01 -6.29332900e-01 4.45453823e-01 -5.10218799e-01
-7.84104705e-01 -1.62203684e-01 1.43119618e-01 -5.03531158e-01
1.85782313e-01 -4.24239695e-01 -9.56592083e-01 -9.49829280e-01
-6.18303001e-01 7.67740011e-01 4.76418674e-01 1.05906355e+00
4.05364856e-02 4.46615219e-01 7.99494565e-01 -3.88223231e-01
-6.79334044e-01 -4.52940851e-01 -1.25714588e+00 -3.56670290e-01
-5.84467113e-01 -6.82438016e-01 7.79169053e-02 -8.67569447e-01] | [3.498565912246704, 1.5135672092437744] |
8b277f60-64bc-455c-b742-430b30a328a9 | drone-based-volume-estimation-in-indoor | 2211.08013 | null | https://arxiv.org/abs/2211.08013v1 | https://arxiv.org/pdf/2211.08013v1.pdf | Drone-based Volume Estimation in Indoor Environments | Volume estimation in large indoor spaces is an important challenge in robotic inspection of industrial warehouses. We propose an approach for volume estimation for autonomous systems using visual features for indoor localization and surface reconstruction from 2D-LiDAR measurements. A Gaussian Process-based model incorporates information collected from measurements given statistical prior information about the terrain, from which the volume estimate is computed. Our algorithm finds feasible trajectories which minimize the uncertainty of the volume estimate. We show results in simulation for the surface reconstruction and volume estimate of topographic data. | ['John Lygeros', 'Alisa Rupenyan', 'Stefan Stevšić', 'Dominic Liao-McPherson', 'Samuel Balula'] | 2022-11-15 | null | null | null | null | ['indoor-localization'] | ['computer-vision'] | [-4.66785580e-02 9.93049219e-02 3.54396760e-01 -6.58688426e-01
-6.27005517e-01 -5.45755267e-01 3.21710318e-01 7.63067901e-01
-5.41101158e-01 7.92493224e-01 -4.50388640e-01 -4.35782135e-01
-4.34169322e-01 -1.14188981e+00 -9.85599041e-01 -6.07998848e-01
-3.31568092e-01 1.03155887e+00 9.58379731e-02 -5.66847771e-02
5.04314423e-01 8.71758461e-01 -1.14818990e+00 -4.89405334e-01
9.38392878e-01 1.14979875e+00 1.05597460e+00 8.42419982e-01
-1.43044919e-01 -8.62854496e-02 -5.28511882e-01 3.05673420e-01
5.12390137e-01 2.88536608e-01 -4.27218169e-01 5.87292075e-01
-1.09416552e-01 -1.85208887e-01 -2.68312334e-03 1.00668347e+00
-9.59374383e-02 1.82294831e-01 1.08403993e+00 -1.10297835e+00
-1.55241385e-01 -9.62433405e-03 -6.19043171e-01 -1.90469950e-01
1.98053524e-01 4.70047444e-02 3.31238151e-01 -1.10337293e+00
7.87719369e-01 1.30671501e+00 6.20540380e-01 -2.54716247e-01
-1.12022519e+00 1.38160557e-01 1.71622690e-02 9.22954641e-03
-1.76966453e+00 1.66949585e-01 3.91943336e-01 -3.68111759e-01
5.61168194e-01 -8.91655833e-02 8.52588117e-01 1.55814156e-01
1.30634594e+00 4.51606691e-01 1.02924788e+00 -1.77003086e-01
6.78660274e-01 3.19421500e-01 -2.50243574e-01 9.19247210e-01
8.83663893e-01 2.81791780e-02 -2.12617278e-01 -1.12268798e-01
1.08102643e+00 1.51177734e-01 -3.86276096e-02 -1.15191805e+00
-8.74200642e-01 8.96448076e-01 6.02030754e-01 -4.50731933e-01
-8.08244288e-01 2.85105377e-01 1.52908012e-01 2.84762830e-01
3.60113800e-01 3.62760663e-01 -5.94461203e-01 -3.74042988e-02
-5.33383369e-01 2.55465448e-01 1.03625429e+00 1.57001793e+00
8.94731402e-01 -9.16059762e-02 1.45413667e-01 4.79126349e-02
8.18658054e-01 1.31869209e+00 -3.16434324e-01 -1.03463376e+00
5.59463441e-01 3.69270056e-01 6.62569642e-01 -5.93138039e-01
-4.04060423e-01 7.40630776e-02 -4.92935508e-01 6.10614419e-01
4.24522340e-01 -2.30709359e-01 -1.06565535e+00 6.35773420e-01
5.42094171e-01 -3.78815591e-01 1.48211733e-01 5.01433432e-01
1.65807635e-01 4.71355796e-01 -4.70858634e-01 -3.71840328e-01
9.24708843e-01 -3.69975716e-01 -8.28269005e-01 -1.59087092e-01
2.84274489e-01 -5.57184160e-01 4.25234586e-01 6.77371562e-01
-7.51964808e-01 -3.18170577e-01 -1.09639513e+00 2.38475591e-01
-1.68519288e-01 8.98991376e-02 3.35358411e-01 3.15539360e-01
-9.40068603e-01 8.47711504e-01 -1.05836535e+00 -3.62741262e-01
3.29858035e-01 3.85670960e-01 -2.92772859e-01 -2.37133160e-01
-2.55307168e-01 1.29425371e+00 5.59432991e-02 5.31898737e-01
-9.70056951e-01 -3.02613229e-01 -1.13706493e+00 -5.13079226e-01
2.97120661e-01 -3.26849490e-01 1.16875434e+00 6.29608393e-01
-1.37421203e+00 2.95238584e-01 -5.29082239e-01 -5.53822935e-01
6.17327332e-01 -5.72324395e-01 4.33720648e-01 -1.08176485e-01
2.98343927e-01 7.78751448e-02 7.57376671e-01 -1.51112556e+00
-5.27710021e-01 -6.91513062e-01 -6.03539944e-01 2.77824640e-01
6.69670403e-01 -8.93599510e-01 3.98807004e-02 4.76203799e-01
8.80075395e-01 -8.29046309e-01 -1.09361005e+00 3.72623265e-01
-6.17399871e-01 -1.16581261e-01 9.87685740e-01 -6.18606567e-01
-7.87099265e-03 -1.64351022e+00 -1.33986145e-01 5.89214206e-01
-9.08120871e-02 -7.39959300e-01 3.35596144e-01 5.88628888e-01
9.48791921e-01 -2.72129864e-01 -1.91837057e-01 -2.98143119e-01
6.37046574e-03 3.82397979e-01 -1.03054434e-01 1.06532836e+00
2.07944110e-01 6.48235738e-01 -1.05635965e+00 -6.03546917e-01
8.89694571e-01 2.32293263e-01 2.54333094e-02 2.42538005e-01
2.18970794e-02 5.71748555e-01 -7.61020601e-01 7.19441593e-01
1.09713447e+00 1.75875083e-01 1.14802182e-01 -9.11862999e-02
-5.67977726e-01 -1.27510905e-01 -1.28515947e+00 1.81426454e+00
-8.46937656e-01 4.66614813e-01 7.33276129e-01 -4.75776315e-01
1.53569090e+00 -1.25703976e-01 6.02028906e-01 -5.77639900e-02
2.68602341e-01 2.51959503e-01 -6.31137371e-01 -3.66452783e-01
9.14961874e-01 -1.23859555e-01 -3.60172242e-01 -4.28114347e-02
-1.98248908e-01 -1.47598886e+00 -8.14571455e-02 6.18348597e-03
1.20820105e+00 3.87364358e-01 5.11867702e-01 -4.98968542e-01
8.98757875e-02 3.10634583e-01 3.92943829e-01 6.72302902e-01
-1.79472178e-01 1.27712801e-01 3.34749855e-02 -3.66230190e-01
-1.19330728e+00 -1.81279862e+00 -5.84742665e-01 -1.22726813e-01
8.50031376e-01 1.73489884e-01 -3.21786612e-01 -4.95303065e-01
6.61255419e-01 6.94788635e-01 -3.68304998e-01 3.92200947e-01
-4.15044755e-01 -3.26780885e-01 -4.61900890e-01 5.31679034e-01
1.28240079e-01 -6.98687673e-01 -8.98896277e-01 3.58485252e-01
2.18251050e-01 -1.07869005e+00 1.27336681e-01 7.26139605e-01
-1.29869592e+00 -1.38335812e+00 -1.57236233e-01 -5.35597801e-01
1.03083432e+00 2.34547883e-01 8.34917307e-01 -4.00622398e-01
-6.78081989e-01 7.00492799e-01 1.24046341e-01 -1.12117934e+00
-2.79643655e-01 -5.89419603e-01 2.58931756e-01 -6.94042623e-01
2.52808452e-01 -2.77810812e-01 -3.91603112e-01 3.41833979e-01
2.83792298e-02 -5.15214324e-01 5.13779461e-01 5.13299465e-01
1.11367190e+00 2.71327227e-01 -4.54044342e-03 -4.61222082e-01
8.25141311e-01 -4.77495074e-01 -1.20639086e+00 -2.38888711e-02
-5.65055132e-01 1.46533936e-01 2.10066929e-01 6.63500354e-02
-8.62281561e-01 6.33700490e-01 2.36843705e-01 -3.72322023e-01
-3.33079576e-01 1.08823709e-01 3.82254794e-02 -2.07440034e-01
4.20062602e-01 -1.36942824e-03 2.57016391e-01 -4.13710296e-01
3.67245048e-01 5.28380096e-01 3.65603536e-01 -4.31420773e-01
1.11311638e+00 7.86664367e-01 8.26159179e-01 -1.22766376e+00
-3.57899219e-01 -7.30826080e-01 -1.44045699e+00 -2.74478883e-01
7.52959967e-01 -8.32977831e-01 -1.01073623e+00 -1.95126906e-01
-1.30872214e+00 -2.47866362e-01 -9.32469785e-01 7.44084775e-01
-9.96742308e-01 2.93590248e-01 -3.10817689e-01 -1.54621804e+00
-1.98359922e-01 -1.20134926e+00 1.42350423e+00 1.16235398e-01
-4.64183390e-02 -1.03067064e+00 -1.12103529e-01 -3.93936098e-01
1.47748724e-01 6.33789480e-01 3.25957030e-01 2.24230681e-02
-1.11673057e+00 -4.45685446e-01 -5.25747500e-02 9.18261781e-02
2.91861922e-01 -1.72464088e-01 -6.97995842e-01 -2.71483272e-01
4.51535434e-01 2.26836383e-01 3.55602533e-01 1.02480912e+00
7.49762893e-01 1.59906760e-01 -8.57943952e-01 3.66163075e-01
1.61513996e+00 2.42961735e-01 1.84534132e-01 3.52175325e-01
3.30443531e-01 7.45368123e-01 1.75201106e+00 7.03459680e-01
4.01773870e-01 4.37769115e-01 9.43173170e-01 2.57736236e-01
7.03643382e-01 -2.13498935e-01 1.06862001e-01 5.78780174e-01
3.55772600e-02 -1.31974161e-01 -8.18454027e-01 8.05248618e-01
-1.59467566e+00 -4.09553111e-01 -6.51731551e-01 2.33902454e+00
8.00252035e-02 -9.25158802e-03 -5.08009672e-01 3.41001526e-03
6.29262269e-01 -3.74932855e-01 -4.89311844e-01 -5.73363900e-01
5.44145703e-01 4.33204956e-02 1.41504169e+00 9.28788543e-01
-1.04716516e+00 6.19292974e-01 6.71320105e+00 2.42499635e-01
-2.85649002e-01 2.65532304e-02 -1.43717021e-01 1.58184022e-01
-1.58817261e-01 -2.36068070e-02 -1.11965644e+00 9.61274058e-02
6.70001149e-01 -2.40510643e-01 8.51232409e-02 1.12448084e+00
5.48081458e-01 -9.07349825e-01 -8.97418141e-01 7.78165817e-01
-2.42023215e-01 -1.04761517e+00 -5.89214683e-01 4.84353304e-01
7.34369099e-01 2.46054247e-01 -1.55556321e-01 -2.87175085e-02
8.62525940e-01 -4.76345837e-01 8.06495786e-01 7.55039752e-01
5.59048533e-01 -8.53083611e-01 9.09966707e-01 5.87096453e-01
-1.57244396e+00 -2.26439953e-01 -1.07426918e+00 7.51939043e-02
1.03833687e+00 9.89374757e-01 -1.78665400e+00 5.58056891e-01
6.14345074e-01 3.12114537e-01 -8.25192109e-02 1.41200852e+00
-5.71837485e-01 -1.41183794e-01 -6.54988825e-01 -4.79360640e-01
3.95845920e-02 -7.36566186e-01 7.53329694e-01 6.82479739e-01
6.24837399e-01 -3.22414279e-01 8.56343091e-01 1.08902514e+00
5.35065353e-01 6.57048672e-02 -1.35148752e+00 5.98700225e-01
5.52631259e-01 1.16628015e+00 -9.05717969e-01 1.19844355e-01
2.43003294e-01 8.43894899e-01 -6.70815706e-02 7.05590099e-02
-5.17061710e-01 -5.58314860e-01 4.53400552e-01 2.43020400e-01
2.69187272e-01 -1.07871795e+00 -4.56744105e-01 -3.10562938e-01
8.18762332e-02 5.47676265e-01 -3.48816663e-01 -7.72706926e-01
-1.00120473e+00 3.51494402e-01 2.93059438e-01 -1.27723002e+00
-3.85940105e-01 -9.27790463e-01 -5.40324330e-01 9.19010401e-01
-1.46543586e+00 -8.38893473e-01 -6.53245032e-01 -1.22218570e-02
9.67821658e-01 1.80441484e-01 5.66833377e-01 -8.24474573e-01
3.72654438e-01 -5.61984777e-01 5.50961792e-01 -4.74766344e-01
1.35520324e-01 -1.70829999e+00 5.72583973e-01 5.03696918e-01
-2.54133254e-01 4.20860708e-01 1.16108799e+00 -1.36251605e+00
-1.89360845e+00 -1.24951470e+00 3.33148986e-01 -8.28096092e-01
6.11004353e-01 -4.67330277e-01 -5.16751647e-01 6.08557403e-01
-1.58770114e-01 9.32198316e-02 -6.37234598e-02 -1.88775569e-01
7.56452978e-01 9.00149867e-02 -1.65024483e+00 1.08652316e-01
7.98752964e-01 -9.99735221e-02 -3.68078768e-01 5.57670295e-01
7.58976758e-01 -7.48353243e-01 -1.14238811e+00 3.72643709e-01
4.21573639e-01 -2.48071432e-01 7.74916351e-01 2.20715046e-01
-3.71557504e-01 -2.80595243e-01 -1.75316453e-01 -1.51832640e+00
1.51911765e-01 -4.81079698e-01 1.58161312e-01 6.55645728e-01
2.58892119e-01 -5.22879303e-01 1.12453902e+00 4.40944940e-01
-1.51786551e-01 -6.78576410e-01 -1.18813062e+00 -1.12563729e+00
-1.92368865e-01 -4.05854166e-01 2.09847659e-01 1.62029415e-01
-1.42071158e-01 3.29399575e-03 9.88710150e-02 7.21118391e-01
1.33339059e+00 8.20174143e-02 8.67295504e-01 -1.60993671e+00
5.70819139e-01 6.81283891e-01 -5.98261535e-01 -1.13118219e+00
8.24261606e-02 -4.60140109e-01 9.56908941e-01 -2.28503799e+00
-3.11999500e-01 -6.12711370e-01 5.03415942e-01 -2.22194269e-01
4.78104085e-01 -3.68817568e-01 3.51257771e-02 -3.76255400e-02
-3.65529776e-01 7.55432427e-01 1.41997981e+00 5.49571291e-02
-5.52217662e-01 4.55729723e-01 2.40628988e-01 8.11772227e-01
8.70128274e-01 -5.25026143e-01 -4.62466925e-01 -1.91216722e-01
-1.87058732e-01 4.05874282e-01 1.88383862e-01 -1.02824807e+00
2.12414563e-01 -4.22938973e-01 5.93806505e-01 -1.51144409e+00
8.21731150e-01 -1.32364690e+00 -3.13697696e-01 7.13421047e-01
2.45970473e-01 4.18409765e-01 1.31909683e-01 1.31058657e+00
1.94339827e-01 -6.16775095e-01 4.38457549e-01 -3.16002786e-01
-7.09292531e-01 2.71074623e-01 -6.95642054e-01 -5.78083217e-01
1.29387987e+00 -3.37743223e-01 2.59934440e-02 -3.87794793e-01
-8.19957852e-01 4.59950119e-01 6.55466199e-01 1.58587858e-01
1.38402534e+00 -9.35055435e-01 -6.05865717e-01 2.93978721e-01
-2.44381636e-01 5.93008935e-01 -2.57864356e-01 7.61326909e-01
-8.79889548e-01 5.20457566e-01 5.71606494e-02 -1.22235024e+00
-1.12459695e+00 4.67670381e-01 3.15974578e-02 -1.65175483e-01
-5.61450362e-01 7.03866959e-01 -1.28661171e-01 -7.65381455e-01
-9.40102711e-02 -8.75598669e-01 1.51627630e-01 -5.01590788e-01
1.14040479e-01 6.43135905e-01 4.23629135e-02 -2.55733907e-01
-2.02064574e-01 6.84173882e-01 4.53404337e-01 -4.13491100e-01
1.35454428e+00 -7.16159284e-01 -2.79094912e-02 8.42343986e-01
8.06489229e-01 7.32053351e-03 -1.46830201e+00 6.74747899e-02
3.12886775e-01 -7.01187253e-01 -3.10972985e-02 -1.71719834e-01
-3.17959994e-01 9.65139270e-01 6.10861659e-01 -1.25328526e-01
3.15497994e-01 -3.63842808e-02 1.79109871e-01 8.23352277e-01
1.38960958e+00 -1.04518270e+00 -1.09341796e-02 6.26800954e-01
1.02434683e+00 -1.27942789e+00 5.08891284e-01 -1.04969859e+00
-6.31853580e-01 1.18971086e+00 4.84498262e-01 -4.61402237e-01
9.46016192e-01 7.28867292e-01 -5.90235256e-02 -5.36961377e-01
-5.46438515e-01 -1.02259241e-01 -1.31960392e-01 9.80640471e-01
-1.39683068e-01 1.73500240e-01 6.18945733e-02 -3.36491585e-01
-2.02062592e-01 -3.62796128e-01 7.32259095e-01 1.49778914e+00
-1.19130862e+00 -7.04314947e-01 -8.07509065e-01 5.69074571e-01
5.02525866e-02 3.37131321e-01 2.94750482e-01 9.93569613e-01
-2.39822105e-01 1.06521690e+00 4.59321439e-01 -5.03126159e-02
5.97156167e-01 -4.20057178e-01 8.39546859e-01 -9.12543058e-01
4.45579961e-02 -1.38643980e-01 7.22422302e-02 -6.37398362e-01
3.92693907e-01 -8.67228329e-01 -1.46547115e+00 3.04552615e-01
-8.28162491e-01 4.86637622e-01 1.79096365e+00 6.67600393e-01
-1.27933517e-01 3.61189753e-01 6.98785901e-01 -1.21935070e+00
-8.50645065e-01 -8.46257925e-01 -1.15640414e+00 -2.21616060e-01
1.43808443e-02 -8.54869783e-01 -2.79370397e-01 -2.18463391e-01] | [7.198177337646484, -2.015690565109253] |
489856e6-a305-405f-8a10-76dc93b2748b | mtna-a-neural-multi-task-model-for-aspect | null | null | https://aclanthology.org/I17-2026 | https://aclanthology.org/I17-2026.pdf | MTNA: A Neural Multi-task Model for Aspect Category Classification and Aspect Term Extraction On Restaurant Reviews | Online reviews are valuable resources not only for consumers to make decisions before purchase, but also for providers to get feedbacks for their services or commodities. In Aspect Based Sentiment Analysis (ABSA), it is critical to identify aspect categories and extract aspect terms from the sentences of user-generated reviews. However, the two tasks are often treated independently, even though they are closely related. Intuitively, the learned knowledge of one task should inform the other learning task. In this paper, we propose a multi-task learning model based on neural networks to solve them together. We demonstrate the improved performance of our multi-task learning model over the models trained separately on three public dataset released by SemEval workshops. | ['Wubai Zhou', 'Wei Xue', 'Tao Li', 'Qing Wang'] | 2017-11-01 | mtna-a-neural-multi-task-model-for-aspect-1 | https://aclanthology.org/I17-2026 | https://aclanthology.org/I17-2026.pdf | ijcnlp-2017-11 | ['extract-aspect'] | ['natural-language-processing'] | [ 1.48829758e-01 6.37952536e-02 -4.08944964e-01 -6.89477324e-01
-1.02239203e+00 -5.59137940e-01 3.38933855e-01 4.09444690e-01
-3.12935978e-01 4.73145068e-01 3.14885944e-01 -4.26302522e-01
1.19856820e-01 -7.91909754e-01 -5.50042391e-01 -5.72262824e-01
4.01717037e-01 5.84385395e-01 -1.96177572e-01 -2.54439294e-01
2.41578281e-01 -1.91167101e-01 -1.03938890e+00 6.36395633e-01
7.85516202e-01 1.16773427e+00 -2.38553546e-02 2.85578281e-01
-7.20840871e-01 7.09222615e-01 -4.53803957e-01 -9.84014571e-01
-3.41453813e-02 -1.24417067e-01 -7.36083090e-01 3.44935000e-01
-7.65693933e-02 -1.12060085e-03 4.24058735e-01 1.20692277e+00
4.38592620e-02 -3.76039222e-02 7.67008901e-01 -1.16398275e+00
-8.29432368e-01 8.95201564e-01 -9.68798339e-01 -1.01814292e-01
5.92731424e-02 -1.36901259e-01 1.63785207e+00 -8.07892680e-01
1.91857189e-01 8.79744768e-01 3.82755667e-01 1.71444818e-01
-8.56970966e-01 -6.13090575e-01 6.98666215e-01 1.75202027e-01
-6.80822968e-01 -2.74594486e-01 1.03144634e+00 -3.37051600e-01
1.01010156e+00 -6.76092207e-02 7.23527372e-01 8.21168721e-01
1.75658703e-01 1.30090857e+00 8.90817881e-01 -2.21498758e-01
2.33507127e-01 6.45423293e-01 4.61641401e-01 2.51538128e-01
4.54972535e-01 -6.13622844e-01 -5.79809606e-01 -3.61415669e-02
2.50507772e-01 2.17423111e-01 1.19648017e-02 -1.03261165e-01
-8.76075089e-01 1.06004345e+00 6.49548173e-02 3.34242761e-01
-6.84972823e-01 -1.53989792e-01 4.43212360e-01 4.87541139e-01
9.60602641e-01 4.03789043e-01 -1.09707308e+00 -8.75263587e-02
-4.44242626e-01 2.02032685e-01 1.11411810e+00 9.01919067e-01
9.15882409e-01 4.38900739e-02 8.42125155e-03 8.64263833e-01
6.47316933e-01 4.58545357e-01 5.94138801e-01 -7.57041052e-02
4.23556477e-01 9.99154985e-01 -3.35066989e-02 -9.99724686e-01
-2.84663945e-01 -4.41086739e-01 -7.50246048e-01 -3.42825145e-01
1.22131176e-01 -3.51386458e-01 -8.84637296e-01 1.29163969e+00
2.27574527e-01 -2.05516964e-01 5.34373149e-02 7.04625368e-01
9.34832811e-01 7.33976901e-01 3.28914046e-01 -1.12192318e-01
1.63307476e+00 -1.33530366e+00 -1.00274110e+00 -6.00394905e-01
6.24526441e-01 -9.20173347e-01 8.47821832e-01 6.43681586e-01
-7.59321094e-01 -2.35747918e-01 -1.11703408e+00 1.78321823e-03
-6.69446945e-01 2.98811853e-01 9.21319246e-01 5.01849294e-01
-6.20796204e-01 3.12262833e-01 -5.35606503e-01 -3.87757905e-02
4.60828096e-01 1.26407072e-01 -7.07051530e-02 2.73047723e-02
-1.09877598e+00 7.46433139e-01 1.88403782e-02 1.90615267e-01
-5.37411809e-01 -3.41419488e-01 -9.20142353e-01 1.65924981e-01
5.36652029e-01 -4.59668428e-01 1.55898321e+00 -1.19582534e+00
-1.25605428e+00 7.43820488e-01 -4.75316972e-01 -3.61103266e-01
-5.16997315e-02 -5.00997484e-01 -5.06300509e-01 -3.34670365e-01
2.68382639e-01 1.10375367e-01 1.03101420e+00 -1.12313783e+00
-8.90151262e-01 -5.82637787e-01 3.74438971e-01 2.79566050e-01
-5.25045455e-01 1.85883000e-01 -5.16969085e-01 -4.44295973e-01
-1.30350858e-01 -5.65999806e-01 -4.72605526e-01 -2.23405644e-01
-4.57017124e-01 -4.86115158e-01 7.70367920e-01 -7.44827986e-01
8.98820162e-01 -1.99235368e+00 4.43060556e-03 9.29592177e-02
3.05855423e-01 2.41591066e-01 -3.53613585e-01 1.70112267e-01
-2.77337097e-02 2.14903161e-01 -1.18841685e-01 -5.05116284e-01
2.70673749e-03 1.10604048e-01 -3.78018886e-01 -1.50099872e-02
3.30724716e-01 1.01140130e+00 -7.99469531e-01 -2.66219139e-01
-2.39866868e-01 4.23582464e-01 -1.61356449e-01 2.16652513e-01
-4.13051277e-01 1.71080515e-01 -8.43877017e-01 7.52379537e-01
4.05803353e-01 -4.83683527e-01 1.82580605e-01 -4.11227107e-01
3.30446094e-01 6.51386976e-01 -8.35749686e-01 1.36082184e+00
-8.89607847e-01 4.45243090e-01 -9.22089294e-02 -1.31637371e+00
9.46834922e-01 5.14273643e-01 5.18374324e-01 -7.00004041e-01
3.19435388e-01 -6.43697241e-03 -9.91697982e-02 -4.34090406e-01
5.60967326e-01 -5.32736361e-01 -2.46233121e-01 9.97873008e-01
1.56016797e-01 -1.45778552e-01 3.23900759e-01 1.09413974e-01
5.67684710e-01 -3.23269106e-02 6.36151731e-01 6.04771487e-02
5.90529799e-01 1.17578392e-03 7.22488642e-01 2.52420396e-01
-8.83574486e-02 4.61118847e-01 7.24700511e-01 -7.10074604e-01
-7.36248016e-01 -6.66899681e-01 2.60878988e-02 9.94154572e-01
-6.16346821e-02 -5.84874451e-01 -2.98859954e-01 -1.23068345e+00
-7.97223598e-02 6.78629577e-01 -6.54508650e-01 -9.38967094e-02
-3.13323855e-01 -9.21562314e-01 -1.82067126e-01 5.75292885e-01
2.87117451e-01 -1.05094373e+00 -1.56072259e-01 2.68782377e-01
-2.71678716e-01 -1.43594086e+00 -4.54647720e-01 2.69443899e-01
-6.68023407e-01 -1.20806026e+00 -4.54235464e-01 -7.39818871e-01
6.00052357e-01 4.13065791e-01 1.43833601e+00 8.40688124e-02
1.85314342e-01 1.18816309e-01 -6.76658928e-01 -9.87136185e-01
-1.19310699e-01 4.08241481e-01 -1.20630652e-01 3.01811397e-01
1.07150149e+00 -5.52061915e-01 -3.34392399e-01 1.88663140e-01
-8.60740244e-01 1.39640331e-01 7.92916834e-01 4.87374961e-01
6.07845783e-01 2.32320756e-01 7.79791296e-01 -1.47802436e+00
9.14612234e-01 -5.02640009e-01 -4.57338899e-01 4.10248607e-01
-8.42028439e-01 -4.66233343e-02 5.97106397e-01 -1.80571839e-01
-1.09738719e+00 -1.40253708e-01 -1.19501971e-01 1.94195542e-03
-1.78962313e-02 1.03676081e+00 -3.15339893e-01 3.66987377e-01
-9.08853207e-03 1.70959741e-01 -3.48049849e-01 -4.46846336e-01
1.23481378e-01 7.58300126e-01 -2.38970533e-01 -3.29772681e-01
7.39549696e-01 2.53570259e-01 -4.55736250e-01 -6.72830284e-01
-1.50321507e+00 -7.44135380e-01 -5.53412974e-01 -3.15207206e-02
8.03164601e-01 -1.11800325e+00 -4.00156707e-01 4.13441360e-01
-1.23839343e+00 1.79421648e-01 -1.71867177e-01 2.54092515e-01
-6.81169778e-02 2.51154229e-02 -4.64812100e-01 -6.81878448e-01
-7.28374362e-01 -1.19103515e+00 1.04351759e+00 5.22090793e-01
-2.17038974e-01 -1.32307839e+00 4.75120731e-02 6.87204838e-01
5.59627533e-01 -3.56899321e-01 1.00190389e+00 -1.29025674e+00
-4.18216646e-01 -5.05131185e-01 -3.31365675e-01 5.87164879e-01
6.70713246e-01 -6.10981397e-02 -1.03371727e+00 5.46930321e-02
2.93975919e-01 -3.47891271e-01 8.96890104e-01 4.36319649e-01
9.45428789e-01 -4.39678460e-01 -5.97444102e-02 2.19043225e-01
1.24381888e+00 1.53467074e-01 3.52163941e-01 4.12240505e-01
8.23611438e-01 7.82013834e-01 7.20292151e-01 4.05757397e-01
6.98566020e-01 3.82534027e-01 6.97109625e-02 -7.67877745e-03
3.23399603e-01 -5.77305667e-02 4.31138515e-01 1.34712350e+00
5.75739816e-02 -7.66612068e-02 -6.56861722e-01 7.14744449e-01
-1.94717360e+00 -4.91152644e-01 -1.58014044e-01 1.78044295e+00
9.07462835e-01 2.84484535e-01 -5.96580561e-03 -1.82123765e-01
3.71930927e-01 4.45664734e-01 -5.88378370e-01 -6.54079735e-01
-2.89568398e-02 1.34703964e-01 1.39399320e-01 2.23454729e-01
-1.30230713e+00 8.48921239e-01 5.83685064e+00 6.97992146e-01
-9.58458245e-01 2.74329007e-01 8.99710119e-01 -1.79626122e-01
-7.80926228e-01 6.52682781e-02 -8.58840585e-01 2.60756046e-01
8.33261490e-01 -5.21855652e-01 1.11828938e-01 1.21484542e+00
7.74736404e-02 -1.39542654e-01 -1.04711270e+00 6.41330421e-01
2.56827116e-01 -9.85024154e-01 3.33619535e-01 -6.37975857e-02
8.13037574e-01 -6.35800436e-02 1.17655937e-02 5.72313786e-01
4.43226546e-01 -8.52017462e-01 3.11520875e-01 7.99112692e-02
8.51623341e-02 -8.14363420e-01 1.07978117e+00 1.65921569e-01
-1.33724070e+00 3.02750558e-01 -3.46486002e-01 -1.83916286e-01
5.07087767e-01 1.16974378e+00 -4.83833849e-01 6.41213238e-01
5.68437576e-01 1.12145138e+00 -3.28707367e-01 7.98793733e-01
-6.19161963e-01 6.71827614e-01 1.89544693e-01 -2.68076211e-01
2.74003714e-01 -6.55208409e-01 2.92624027e-01 8.69934976e-01
1.41549438e-01 -9.68996435e-02 1.21826954e-01 6.09114110e-01
-4.53756750e-01 3.68773997e-01 -4.18328971e-01 -5.17544627e-01
-2.23478451e-01 1.78842843e+00 -6.72418118e-01 -3.60812485e-01
-8.67789447e-01 6.61908984e-01 3.57249796e-01 3.71818870e-01
-6.30060852e-01 -3.62922460e-01 8.82823050e-01 -1.91111520e-01
4.41485435e-01 -7.87269995e-02 -6.27677083e-01 -1.46343398e+00
3.01045597e-01 -8.93720031e-01 2.05885842e-01 -4.60529596e-01
-1.54728973e+00 6.42340064e-01 -4.26291853e-01 -1.20059276e+00
-2.87005037e-01 -7.68614590e-01 -9.02870297e-01 8.63893926e-01
-2.08498073e+00 -1.34804153e+00 1.53730974e-01 1.81969896e-01
8.10563385e-01 -2.04781607e-01 6.84207082e-01 3.48516703e-01
-5.62989235e-01 3.70114565e-01 -2.13306293e-01 3.92629504e-01
6.42244041e-01 -1.28650320e+00 5.46442389e-01 7.02335954e-01
4.13545609e-01 5.55278957e-01 5.22386014e-01 -5.60451448e-01
-1.34031308e+00 -1.02390027e+00 1.24914932e+00 -3.58818740e-01
7.46851444e-01 -1.66621432e-01 -9.53001440e-01 8.69721413e-01
4.60105509e-01 -4.86499101e-01 1.12747133e+00 8.28043878e-01
-5.03080249e-01 -3.30684602e-01 -6.76662564e-01 2.11560085e-01
2.20175520e-01 -5.55637956e-01 -6.82079196e-01 4.33295429e-01
9.07010019e-01 -6.53981566e-02 -6.18343294e-01 3.31719398e-01
5.45892358e-01 -6.44256830e-01 5.45736670e-01 -8.28130186e-01
8.09567332e-01 -5.13691902e-02 -1.14438444e-01 -1.61572671e+00
1.12872511e-01 -2.78606564e-02 -1.04734622e-01 1.42943168e+00
1.09464586e+00 -5.88905990e-01 7.01364875e-01 8.45905244e-01
2.04051167e-01 -8.94501567e-01 -4.48270828e-01 -3.23578328e-01
-2.58074347e-02 -7.31924474e-01 7.32688487e-01 8.47835720e-01
-2.18436699e-02 9.95946348e-01 -3.18202496e-01 2.83709951e-02
3.75296146e-01 5.81611812e-01 6.06083632e-01 -1.16520119e+00
-4.23934728e-01 -5.32976687e-01 1.29259109e-01 -9.79771376e-01
9.51462612e-02 -5.54520428e-01 -3.48559744e-03 -1.77945900e+00
5.81031024e-01 -3.75726610e-01 -5.89155316e-01 6.92498624e-01
-4.95964706e-01 2.75495816e-02 1.05620818e-02 5.32316975e-02
-8.14609647e-01 7.37243176e-01 1.22580934e+00 -4.58561957e-01
-1.30936414e-01 5.17162263e-01 -1.40225220e+00 8.77073884e-01
8.50347579e-01 -5.99677980e-01 -4.30795759e-01 -6.23318076e-01
9.55152154e-01 -2.84509629e-01 -3.28683406e-01 -3.54027122e-01
9.40047055e-02 -1.01577140e-01 1.07231446e-01 -6.46350145e-01
2.09524363e-01 -7.80967712e-01 -4.78278637e-01 -7.82129839e-02
-2.90910274e-01 1.71566740e-01 4.34738882e-02 6.05276525e-01
-6.52455509e-01 -3.59506845e-01 1.75282359e-01 -2.18172550e-01
-5.31155348e-01 5.08314073e-01 -4.78316784e-01 -9.14870873e-02
7.41209626e-01 3.33968461e-01 -2.12994233e-01 -6.79527164e-01
-5.06490529e-01 5.29654086e-01 6.47167638e-02 6.74561918e-01
7.24406898e-01 -1.11214304e+00 -7.01810479e-01 9.74030942e-02
2.25315124e-01 2.46879116e-01 1.99263707e-01 7.99958885e-01
1.67940438e-01 4.94241476e-01 2.17279777e-01 -2.22743824e-01
-1.09883249e+00 4.54968423e-01 -4.75388318e-02 -8.70475650e-01
-7.07660466e-02 7.14799285e-01 2.49776363e-01 -6.90082848e-01
1.12525202e-01 -2.18609899e-01 -7.98183084e-01 5.03369212e-01
7.76787102e-01 -2.08981767e-01 3.52080971e-01 -4.05280024e-01
-1.62983000e-01 5.88496625e-01 -6.51512504e-01 3.50625217e-02
1.63968647e+00 -2.21228316e-01 -2.73535490e-01 7.01590657e-01
1.22739661e+00 -1.41647398e-01 -6.72591090e-01 -5.47459483e-01
1.62824482e-01 -1.38508722e-01 1.40148804e-01 -8.53469074e-01
-1.45006835e+00 9.44387496e-01 -1.80824846e-01 3.94275248e-01
1.23089767e+00 1.33769900e-01 1.04119384e+00 4.27597493e-01
2.90777773e-01 -1.12370801e+00 1.34085685e-01 4.92650032e-01
6.07894778e-01 -1.55070198e+00 1.82231367e-01 -3.68452817e-01
-1.14342833e+00 9.44528043e-01 6.32331371e-01 7.34024346e-02
9.62902904e-01 2.60020256e-01 4.43639815e-01 -4.25722808e-01
-1.08862138e+00 -2.77844220e-01 5.14897168e-01 4.63926673e-01
6.67495430e-01 3.41127254e-02 -2.16752276e-01 1.24109423e+00
2.89230421e-02 -2.14436695e-01 3.48533392e-01 9.68010485e-01
-2.60084838e-01 -1.42954326e+00 1.89616889e-01 7.92860091e-01
-6.58791661e-01 -3.01311105e-01 -5.04577875e-01 3.44533503e-01
-3.69554937e-01 1.08348882e+00 -1.18429154e-01 -4.29573655e-01
3.00282389e-01 1.11864142e-01 -2.20813416e-02 -7.11879015e-01
-6.39088452e-01 1.42009199e-01 5.14054537e-01 -3.64521146e-01
-7.32670307e-01 -6.60090506e-01 -8.48200142e-01 -1.23438211e-02
-2.97566801e-01 4.17749256e-01 9.53164458e-01 1.53361917e+00
2.74894029e-01 5.72524130e-01 9.35032368e-01 -3.12320828e-01
-3.59197974e-01 -1.01704466e+00 -7.00555921e-01 2.94682831e-01
1.15628473e-01 -2.70640105e-01 -2.43112564e-01 4.73540723e-02] | [11.395238876342773, 6.681830406188965] |
5950bf6c-ada0-4466-9b75-ea1a5f5b565b | position-guided-text-prompt-for-vision | 2212.09737 | null | https://arxiv.org/abs/2212.09737v2 | https://arxiv.org/pdf/2212.09737v2.pdf | Position-guided Text Prompt for Vision-Language Pre-training | Vision-Language Pre-Training (VLP) has shown promising capabilities to align image and text pairs, facilitating a broad variety of cross-modal learning tasks. However, we observe that VLP models often lack the visual grounding/localization capability which is critical for many downstream tasks such as visual reasoning. In this work, we propose a novel Position-guided Text Prompt (PTP) paradigm to enhance the visual grounding ability of cross-modal models trained with VLP. Specifically, in the VLP phase, PTP divides the image into $N\times N$ blocks, and identifies the objects in each block through the widely used object detector in VLP. It then reformulates the visual grounding task into a fill-in-the-blank problem given a PTP by encouraging the model to predict the objects in the given blocks or regress the blocks of a given object, e.g. filling `P" or ``O" in aPTP ``The block P has a O". This mechanism improves the visual grounding capability of VLP models and thus helps them better handle various downstream tasks. By introducing PTP into several state-of-the-art VLP frameworks, we observe consistently significant improvements across representative cross-modal learning model architectures and several benchmarks, e.g. zero-shot Flickr30K Retrieval (+4.8 in average recall@1) for ViLT \cite{vilt} baseline, and COCO Captioning (+5.3 in CIDEr) for SOTA BLIP \cite{blip} baseline. Moreover, PTP achieves comparable results with object-detector based methods, and much faster inference speed since PTP discards its object detector for inference while the later cannot. Our code and pre-trained weight will be released at \url{https://github.com/sail-sg/ptp}. | ['Shuicheng Yan', 'Mike Zheng Shou', 'Pan Zhou', 'Alex Jinpeng Wang'] | 2022-12-19 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Position-Guided_Text_Prompt_for_Vision-Language_Pre-Training_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Position-Guided_Text_Prompt_for_Vision-Language_Pre-Training_CVPR_2023_paper.pdf | cvpr-2023-1 | ['visual-reasoning', 'zero-shot-cross-modal-retrieval', 'visual-reasoning'] | ['computer-vision', 'miscellaneous', 'reasoning'] | [-1.11007787e-01 -1.24325827e-01 -4.83320266e-01 -1.56404167e-01
-1.04505658e+00 -8.53841722e-01 8.49725246e-01 2.77590062e-02
-3.75436753e-01 3.50523621e-01 2.57216860e-02 -4.80272591e-01
1.81098163e-01 -6.15168989e-01 -1.16261995e+00 -7.00530469e-01
4.06139016e-01 6.22720242e-01 3.39345485e-01 8.04632716e-03
1.60558343e-01 2.76980549e-01 -1.44482434e+00 8.55441153e-01
3.96329015e-01 9.66312230e-01 4.41213071e-01 6.92929924e-01
-3.18857938e-01 8.69455099e-01 -2.28972480e-01 -6.03112578e-01
2.84057885e-01 1.16114460e-01 -9.28836048e-01 -1.39056370e-01
1.14784884e+00 -4.36592072e-01 -6.50354564e-01 9.32023108e-01
3.65371346e-01 1.79027095e-01 7.95718849e-01 -1.58323264e+00
-9.39529717e-01 5.41250348e-01 -1.06626868e+00 2.82898962e-01
1.97161824e-01 4.33425814e-01 1.52409315e+00 -1.41371417e+00
8.09718311e-01 1.32439125e+00 5.25743306e-01 5.64388931e-01
-1.32846701e+00 -8.93039763e-01 4.19769049e-01 3.00416619e-01
-1.52478409e+00 -5.33713520e-01 5.38442612e-01 -4.76455420e-01
1.08789504e+00 2.82533258e-01 4.37418073e-01 1.22248912e+00
-9.07182768e-02 1.45788682e+00 9.40133393e-01 -4.16537315e-01
-1.21638119e-01 1.03542097e-01 2.54941970e-01 9.37799454e-01
1.91410948e-02 -8.37070048e-02 -8.09983552e-01 1.01945542e-01
6.11356735e-01 6.33914769e-02 -1.86877832e-01 -5.46015441e-01
-1.38181496e+00 5.77685773e-01 8.24295342e-01 1.63300470e-01
-1.74623966e-01 6.96890950e-01 2.81892031e-01 -1.28235698e-01
2.06306100e-01 1.85198620e-01 -4.33688939e-01 3.11241478e-01
-1.12337351e+00 3.36089104e-01 3.50215912e-01 1.15595841e+00
8.41933668e-01 -3.24951202e-01 -7.37682164e-01 8.45555186e-01
6.04516149e-01 7.54060566e-01 -9.22472402e-02 -1.05621529e+00
8.79983008e-01 5.36034822e-01 1.46058187e-01 -8.34266067e-01
-5.24444580e-02 -2.76682794e-01 -7.19467044e-01 1.61279812e-01
5.88357747e-01 3.51844221e-01 -1.32205784e+00 1.60326922e+00
2.96679288e-01 1.95239380e-01 1.21630028e-01 8.02582026e-01
1.27683687e+00 9.27683830e-01 3.85310024e-01 2.97342628e-01
1.78029644e+00 -1.47086334e+00 -3.96293253e-01 -7.21019328e-01
6.31622136e-01 -8.75944614e-01 1.38757360e+00 1.94984064e-01
-1.01833832e+00 -5.12714028e-01 -7.04631090e-01 -6.09684587e-01
-5.63531876e-01 3.47429901e-01 4.37283635e-01 -8.53527151e-03
-1.21811748e+00 8.10297802e-02 -5.78056753e-01 -4.62559819e-01
7.04412580e-01 5.57645485e-02 -4.09000099e-01 -4.93540972e-01
-9.58711684e-01 7.02307343e-01 4.88349497e-01 2.18283758e-02
-1.24084318e+00 -8.24874938e-01 -8.95986557e-01 5.93597516e-02
5.03847003e-01 -8.35175991e-01 1.25194359e+00 -8.05238426e-01
-7.02628970e-01 1.45354021e+00 -3.91993612e-01 -5.42831123e-01
5.79563797e-01 -5.17272234e-01 -4.42941785e-02 2.48075292e-01
3.03610146e-01 1.48097479e+00 1.07880020e+00 -1.62586200e+00
-7.77613640e-01 -2.41158426e-01 2.73677409e-01 3.83494258e-01
-1.74503997e-02 2.44719442e-02 -1.32036197e+00 -5.00805259e-01
2.10702252e-02 -9.65716004e-01 1.44053489e-01 3.64628017e-01
-6.22662127e-01 -4.64253217e-01 9.37881887e-01 -6.86545014e-01
9.90197062e-01 -2.12744689e+00 -2.75120847e-02 -1.33373305e-01
2.99128473e-01 2.68595308e-01 -3.82232517e-01 4.51083422e-01
5.29354773e-02 1.02558851e-01 1.00288264e-01 -6.52229726e-01
1.97480321e-01 3.26831937e-01 -6.07301056e-01 2.52836913e-01
1.32411718e-01 1.31453466e+00 -8.37580562e-01 -8.96074593e-01
3.95915627e-01 4.30887312e-01 -4.11607862e-01 -2.26824097e-02
-6.31142616e-01 1.63147390e-01 -2.32362986e-01 7.91938066e-01
5.01258373e-01 -7.13277578e-01 -5.70612811e-02 -4.49964166e-01
6.17737398e-02 7.52559006e-02 -9.58262026e-01 1.76939297e+00
-2.74576843e-01 9.15417492e-01 -1.89072341e-02 -7.22476006e-01
4.89439487e-01 1.06730454e-01 1.82638496e-01 -8.38057399e-01
-1.03419527e-01 -1.61819905e-01 -4.16135937e-01 -4.54931021e-01
6.17952645e-01 2.79467463e-01 4.07401696e-02 3.29345226e-01
1.06565788e-01 -1.81866791e-02 2.95771599e-01 6.61982179e-01
9.33155417e-01 2.83633053e-01 -5.87958796e-03 -1.91231363e-03
3.90572757e-01 2.84109026e-01 1.57371715e-01 1.13187480e+00
-1.90636292e-01 6.30694211e-01 4.02087122e-01 -2.43984357e-01
-1.01510608e+00 -1.22935486e+00 -9.13148820e-02 1.48285210e+00
3.89836788e-01 -4.19631213e-01 -4.05704975e-01 -9.48971331e-01
1.88267857e-01 8.39298844e-01 -6.40864372e-01 1.07161403e-01
-5.78886330e-01 -4.27877992e-01 6.56090975e-01 7.18766749e-01
5.72431266e-01 -1.31593370e+00 -2.19900668e-01 -3.13952595e-01
-5.86617172e-01 -1.37644804e+00 -4.78845775e-01 7.95844868e-02
-4.21153843e-01 -1.01186717e+00 -6.96634591e-01 -1.07091022e+00
6.71016395e-01 6.78980529e-01 1.45416737e+00 2.40365237e-01
-2.25270048e-01 5.76035917e-01 -2.36500159e-01 -3.56147259e-01
-8.64771232e-02 -1.24961257e-01 -2.66981483e-01 -1.54869795e-01
4.76315558e-01 6.64420947e-02 -9.36230361e-01 2.95640200e-01
-7.48973727e-01 4.82140720e-01 6.20232761e-01 7.76477993e-01
9.68325138e-01 -3.22908998e-01 -5.90759479e-02 -4.94440168e-01
-1.52090273e-03 -2.11783201e-01 -4.80978906e-01 5.16721904e-01
-3.66655618e-01 -2.42004972e-02 2.27883086e-01 -3.50975007e-01
-8.78369868e-01 -5.19743674e-02 1.11698203e-01 -8.93233478e-01
-5.16160391e-02 1.87461123e-01 -8.07467252e-02 1.51385829e-01
4.92912143e-01 4.31223571e-01 -3.82187605e-01 -3.58078033e-01
7.64099538e-01 4.77438539e-01 7.24166572e-01 -7.03359783e-01
8.72188568e-01 7.16180086e-01 -2.44981647e-01 -5.92899024e-01
-1.14594650e+00 -8.33777010e-01 -3.82722914e-01 -2.18605265e-01
1.03518522e+00 -1.31459785e+00 -6.99358642e-01 3.26553822e-01
-1.23636222e+00 -7.05940545e-01 -3.36492844e-02 1.49069786e-01
-4.80088800e-01 3.00479025e-01 -6.00140214e-01 -6.39828503e-01
-6.11015439e-01 -1.04594982e+00 1.52730191e+00 1.88144013e-01
1.51128322e-02 -6.71172619e-01 -1.25534654e-01 7.70768940e-01
-1.26888201e-01 -3.21410447e-01 9.41329300e-01 -5.01689613e-01
-9.11389947e-01 3.81356739e-02 -8.31262529e-01 2.27940857e-01
-3.91896129e-01 1.54876001e-02 -1.09295285e+00 -4.08984691e-01
-7.22580314e-01 -4.92568940e-01 1.18222845e+00 2.95518726e-01
1.20489442e+00 -2.12856099e-01 -7.33703136e-01 6.16401315e-01
1.55046129e+00 -1.11587614e-01 6.58923924e-01 3.79193276e-01
9.83190596e-01 5.06763935e-01 8.42266619e-01 -1.17250010e-02
4.36655402e-01 8.24101746e-01 6.64337099e-01 -2.66189009e-01
-4.86311078e-01 -5.25892675e-01 3.96545142e-01 7.51287565e-02
2.52586633e-01 -5.68383276e-01 -1.28900146e+00 8.59941125e-01
-2.11009073e+00 -8.88469458e-01 -2.72868901e-01 1.81457961e+00
8.58302832e-01 -4.68646362e-02 -2.84260035e-01 -2.91784316e-01
5.94549298e-01 4.45045531e-01 -3.95089686e-01 8.67292434e-02
-1.09877765e-01 7.53646269e-02 4.40949529e-01 4.09097046e-01
-1.23760414e+00 1.47346604e+00 5.11769867e+00 1.06627011e+00
-8.10805738e-01 1.39944091e-01 6.38259590e-01 -1.86351880e-01
-1.22552462e-01 8.04164335e-02 -1.22398877e+00 2.23044574e-01
3.15229237e-01 2.38398656e-01 2.63071030e-01 7.85067320e-01
1.75607935e-01 -2.20148176e-01 -1.36872303e+00 1.30271983e+00
2.32607365e-01 -1.54611814e+00 4.07670230e-01 -4.87017557e-02
6.45238936e-01 3.45019162e-01 2.80652136e-01 5.39815664e-01
4.58263516e-01 -1.09966135e+00 9.12660420e-01 3.71971041e-01
8.03865910e-01 -2.81178117e-01 4.10929650e-01 2.03634202e-01
-1.35413706e+00 9.79109630e-02 -2.52824783e-01 3.82528305e-01
8.41645747e-02 2.82923937e-01 -8.84971023e-01 4.23812658e-01
1.12355351e+00 7.05878496e-01 -7.38555431e-01 8.54593515e-01
-4.49822992e-01 3.95059556e-01 -1.93455353e-01 2.49352142e-01
6.30525410e-01 6.22454844e-02 4.37381089e-01 1.29614270e+00
2.17664726e-02 -8.79771858e-02 1.80087596e-01 8.68748724e-01
-3.65430474e-01 7.39850923e-02 -5.61893165e-01 2.68319286e-02
4.24887121e-01 1.29034150e+00 -7.16248274e-01 -5.68922698e-01
-4.99966919e-01 1.00216997e+00 5.25345683e-01 5.53493321e-01
-1.18999922e+00 4.89097498e-02 6.11411452e-01 1.30956143e-01
5.99481761e-01 -1.10816739e-01 -1.78290412e-01 -1.12602401e+00
1.34284183e-01 -8.40013385e-01 6.19414866e-01 -1.45434916e+00
-1.17076516e+00 2.25087076e-01 3.12221944e-02 -9.83388603e-01
1.71189398e-01 -8.44321489e-01 -3.06932092e-01 7.42572904e-01
-1.58697832e+00 -1.71924174e+00 -4.53677893e-01 7.91245997e-01
8.20501626e-01 2.82142311e-01 4.83411789e-01 1.78296611e-01
-5.52078247e-01 5.32761276e-01 3.86296846e-02 4.97295946e-01
9.87268209e-01 -1.21570599e+00 3.88654947e-01 9.06684697e-01
7.83534288e-01 5.29342234e-01 6.23422265e-01 -5.00455320e-01
-1.21610093e+00 -1.37326181e+00 9.10015285e-01 -7.84837306e-01
7.34341919e-01 -4.18467045e-01 -8.77083063e-01 1.02359140e+00
3.02317679e-01 2.60746866e-01 2.70256430e-01 3.39225978e-02
-9.60229993e-01 -1.10709094e-01 -7.44808078e-01 9.03668463e-01
1.17509222e+00 -6.58393025e-01 -7.11356282e-01 7.14426160e-01
7.54009366e-01 -5.12268424e-01 -4.10227180e-01 2.54342675e-01
6.30944192e-01 -7.72039533e-01 1.48671269e+00 -6.42839015e-01
5.55601358e-01 -5.14833033e-01 -2.98388630e-01 -6.34988666e-01
-3.11769933e-01 -2.00320572e-01 -2.81512916e-01 1.43862784e+00
4.88441974e-01 -1.96879908e-01 8.40238035e-01 4.00772214e-01
-5.26844636e-02 -6.53885305e-01 -7.25049675e-01 -5.65181375e-01
1.52555876e-03 -6.14359200e-01 5.84916361e-02 8.56069326e-01
-4.48197007e-01 3.86873394e-01 -2.32959777e-01 4.07806635e-01
6.44776106e-01 1.92980617e-01 9.64805722e-01 -7.83635914e-01
-3.41116130e-01 -3.71987969e-01 5.96485771e-02 -1.39820969e+00
2.39740208e-01 -1.09251523e+00 1.51148468e-01 -1.94606745e+00
6.41785145e-01 -5.62812626e-01 -3.71561170e-01 9.92453694e-01
-2.29664639e-01 6.33950472e-01 5.03815889e-01 4.64817345e-01
-1.05850434e+00 2.41913274e-01 1.16418552e+00 -5.49582124e-01
-4.32482846e-02 -3.81128877e-01 -4.45671111e-01 7.87828028e-01
5.76323450e-01 -4.32407618e-01 -2.32540175e-01 -8.04174483e-01
2.99765497e-01 -2.00271443e-01 1.02584636e+00 -7.36862481e-01
2.66365021e-01 -7.86143020e-02 4.66327369e-01 -1.03394771e+00
4.99123126e-01 -7.38946438e-01 1.20461872e-02 2.63347030e-01
-2.13796526e-01 1.20565463e-02 4.16812181e-01 6.88256025e-01
-6.21032529e-02 -5.91034815e-02 7.56554246e-01 -1.81550935e-01
-1.23309076e+00 3.49304676e-01 1.14674792e-01 1.77842066e-01
1.01054215e+00 -1.02677502e-01 -8.23811471e-01 -9.75439101e-02
-6.10829532e-01 6.69014752e-01 5.42943239e-01 6.13892674e-01
5.99901259e-01 -1.11182976e+00 -5.53614140e-01 -1.01731211e-01
5.77136219e-01 3.00852597e-01 4.35894459e-01 1.05240989e+00
-6.12080991e-01 6.56173348e-01 1.17148399e-01 -1.08191347e+00
-1.61699319e+00 7.23868787e-01 2.76220769e-01 -4.70188379e-01
-8.60698581e-01 1.06630147e+00 7.36383200e-01 -9.46001932e-02
4.52074230e-01 -2.18526185e-01 1.64007843e-02 7.07655251e-02
5.00105917e-01 -3.02042626e-02 -2.17336062e-02 -6.94834292e-01
-5.88397086e-01 6.37502968e-01 -3.85577232e-01 -1.87636733e-01
9.16998684e-01 -1.38449296e-01 -9.55510736e-02 2.78086305e-01
1.02094817e+00 -2.08547920e-01 -1.37014318e+00 -5.42411804e-01
-1.44989833e-01 -3.58087957e-01 4.72843125e-02 -9.82656121e-01
-9.72626507e-01 9.21693742e-01 4.11819577e-01 -2.22640708e-01
8.43288720e-01 5.67067504e-01 6.16052747e-01 4.74899650e-01
3.01933855e-01 -8.38557422e-01 4.48048234e-01 5.13841033e-01
9.16682720e-01 -1.37345541e+00 -3.13236117e-02 -2.64757633e-01
-7.59705305e-01 5.49968123e-01 7.82056451e-01 4.93636243e-02
2.00571075e-01 -6.98778033e-02 9.43514630e-02 -3.91009212e-01
-8.36659610e-01 -4.35988694e-01 5.06127596e-01 4.50664282e-01
1.39537439e-01 -1.20891623e-01 5.00193797e-02 1.43661305e-01
1.14633970e-01 -2.53369331e-01 -3.98745053e-02 7.99724340e-01
-4.19431686e-01 -8.14228535e-01 -4.49130684e-01 2.97462583e-01
-3.43549103e-01 -6.23441756e-01 -4.51825917e-01 1.10221040e+00
1.92131281e-01 7.36157000e-01 1.24034800e-01 -1.32229896e-02
1.06259651e-01 3.20547760e-01 4.76541430e-01 -5.92633307e-01
-4.74373072e-01 1.51508069e-02 3.27596143e-02 -6.62861824e-01
-4.07000095e-01 -5.26194930e-01 -1.26265574e+00 -2.21664920e-01
-2.72489190e-01 -2.60440022e-01 4.36316878e-01 8.20764184e-01
3.67678404e-01 3.34990591e-01 -9.55339000e-02 -7.51749992e-01
-6.89366534e-02 -7.23101556e-01 -2.17363715e-01 5.90719700e-01
2.41002068e-01 -6.38337135e-01 -3.11441243e-01 2.82225162e-01] | [10.521842956542969, 1.4499952793121338] |
eb314ea3-ad19-4f0c-b1eb-8315966e768c | bridging-the-visual-gap-wide-range-image | 2103.15149 | null | https://arxiv.org/abs/2103.15149v2 | https://arxiv.org/pdf/2103.15149v2.pdf | Bridging the Visual Gap: Wide-Range Image Blending | In this paper we propose a new problem scenario in image processing, wide-range image blending, which aims to smoothly merge two different input photos into a panorama by generating novel image content for the intermediate region between them. Although such problem is closely related to the topics of image inpainting, image outpainting, and image blending, none of the approaches from these topics is able to easily address it. We introduce an effective deep-learning model to realize wide-range image blending, where a novel Bidirectional Content Transfer module is proposed to perform the conditional prediction for the feature representation of the intermediate region via recurrent neural networks. In addition to ensuring the spatial and semantic consistency during the blending, we also adopt the contextual attention mechanism as well as the adversarial learning scheme in our proposed method for improving the visual quality of the resultant panorama. We experimentally demonstrate that our proposed method is not only able to produce visually appealing results for wide-range image blending, but also able to provide superior performance with respect to several baselines built upon the state-of-the-art image inpainting and outpainting approaches. | ['Wei-Chen Chiu', 'Ya-Chu Chang', 'Chia-Ni Lu'] | 2021-03-28 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Lu_Bridging_the_Visual_Gap_Wide-Range_Image_Blending_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Lu_Bridging_the_Visual_Gap_Wide-Range_Image_Blending_CVPR_2021_paper.pdf | cvpr-2021-1 | ['image-outpainting'] | ['computer-vision'] | [ 5.46664655e-01 8.74202624e-02 -3.13228779e-02 -2.82958180e-01
-9.63392973e-01 -1.97047859e-01 6.58449113e-01 -2.59013683e-01
-7.90667683e-02 5.89244962e-01 1.91207245e-01 -8.07290450e-02
-2.62706243e-02 -8.64412963e-01 -1.23317134e+00 -6.96299672e-01
3.97499561e-01 -1.45437106e-01 5.88904927e-03 -4.10295427e-01
2.28018835e-01 5.47409475e-01 -1.39918590e+00 1.87327698e-01
1.03534782e+00 1.13116848e+00 3.45334888e-01 6.88425481e-01
1.38310203e-02 9.91476715e-01 -5.71684062e-01 -5.04681647e-01
4.40068483e-01 -8.87885749e-01 -4.79437828e-01 4.26663041e-01
8.51738274e-01 -5.73461771e-01 -5.52356243e-01 1.11203122e+00
3.97194803e-01 1.79116145e-01 4.79387552e-01 -1.28604794e+00
-9.68124628e-01 3.41602713e-01 -9.68351185e-01 -2.81987488e-01
3.22195411e-01 1.75652444e-01 6.74041510e-01 -6.86525285e-01
7.06307411e-01 1.05934930e+00 5.21850765e-01 5.08979976e-01
-1.05167449e+00 -6.77234054e-01 -5.30645922e-02 8.31708163e-02
-1.03984964e+00 -3.65097195e-01 1.22745681e+00 -9.18226596e-03
4.15182799e-01 2.68887788e-01 7.94012904e-01 8.23581457e-01
4.51460928e-01 8.41726422e-01 1.29946625e+00 -5.39115846e-01
-1.09803833e-01 -2.71208808e-02 -4.70148355e-01 5.66153526e-01
-3.35731596e-01 2.65964985e-01 -2.72273153e-01 2.39684924e-01
1.16539145e+00 3.57641399e-01 -4.35057819e-01 -3.89181167e-01
-1.23041308e+00 5.83892167e-01 8.41582716e-01 1.94097593e-01
-4.72766012e-01 3.65210027e-01 1.09734058e-01 3.34062636e-01
4.87615556e-01 4.75007147e-01 1.06113024e-01 2.89983422e-01
-1.27785885e+00 4.29448932e-01 2.62570262e-01 1.04084349e+00
8.80251408e-01 1.06615089e-01 -3.22086513e-01 8.39633167e-01
-5.88227669e-03 3.23243231e-01 3.08581084e-01 -1.07492054e+00
6.54044569e-01 3.98690492e-01 2.62289703e-01 -9.77835119e-01
8.01703706e-02 -1.53514832e-01 -9.70952630e-01 5.11683881e-01
8.42311531e-02 -5.20262346e-02 -9.74797606e-01 1.65663123e+00
3.03017050e-01 2.96159923e-01 8.07757825e-02 9.17224944e-01
7.20986843e-01 1.24010038e+00 7.88321346e-02 -2.67241538e-01
1.25654364e+00 -1.52135623e+00 -9.16118622e-01 -2.94600904e-01
-7.78571218e-02 -1.06766486e+00 1.15137923e+00 1.65162772e-01
-1.54924428e+00 -7.71373391e-01 -1.22108185e+00 -6.21199489e-01
-1.56395420e-01 5.67426812e-03 4.32571977e-01 5.03509268e-02
-1.01597345e+00 4.53204513e-01 -3.83649170e-01 -1.18410721e-01
2.71371782e-01 -8.97578597e-02 -6.78279459e-01 -2.91452497e-01
-1.04756391e+00 1.01752603e+00 2.57899284e-01 1.28749609e-01
-7.35526323e-01 -6.89619243e-01 -1.04592443e+00 1.66994959e-01
1.60848007e-01 -9.17332411e-01 9.65602756e-01 -1.33438694e+00
-1.67360389e+00 8.33315909e-01 1.16297491e-01 -3.76782149e-01
6.73331320e-01 -3.16591382e-01 -2.54955322e-01 9.82345343e-02
1.19262256e-01 1.18180954e+00 1.25973201e+00 -1.41173887e+00
-5.45262158e-01 -3.71672586e-02 -1.46028444e-01 4.97760475e-01
-1.30052328e-01 -1.83305368e-02 -7.22110331e-01 -1.25202799e+00
-5.61271980e-02 -7.31107831e-01 -2.60491431e-01 4.51339424e-01
-3.20325047e-01 2.08770677e-01 1.12507081e+00 -1.06281579e+00
8.38804185e-01 -2.26027036e+00 3.62919986e-01 -1.60923943e-01
-9.81395245e-02 2.16536000e-01 -4.14630949e-01 6.00902617e-01
-1.84790283e-01 -2.49852285e-01 -5.52677035e-01 -5.57346225e-01
-2.52612025e-01 5.72884046e-02 -8.32453787e-01 3.20993304e-01
3.67412359e-01 1.11969280e+00 -7.89473772e-01 -3.24929953e-01
5.65520287e-01 6.59513474e-01 -3.17842275e-01 7.80140519e-01
-3.39063257e-01 5.09671390e-01 2.15743780e-02 5.61886787e-01
1.00683391e+00 9.44022611e-02 -2.27365628e-01 -3.61918449e-01
-2.75145769e-02 -2.10379839e-01 -8.06129575e-01 2.11517525e+00
-7.94124603e-01 5.22649825e-01 6.59985170e-02 -8.14955473e-01
1.04562676e+00 9.19993147e-02 2.42575720e-01 -7.74284124e-01
4.57158796e-02 2.07158491e-01 -3.65599811e-01 -5.15514195e-01
7.61232913e-01 -2.35729724e-01 -1.28326818e-01 3.68183166e-01
-3.00644506e-02 -6.51871324e-01 -7.12114871e-02 -1.93137489e-02
5.74120164e-01 4.34951216e-01 3.64797771e-01 1.78475574e-01
4.84562635e-01 -1.35768339e-01 3.49923909e-01 4.99706626e-01
1.09736398e-01 1.08939624e+00 3.08874309e-01 -2.54271537e-01
-1.51323414e+00 -1.11479843e+00 1.23382568e-01 6.94104493e-01
5.72261035e-01 1.98257016e-03 -8.79736364e-01 -4.93605673e-01
-2.64461070e-01 5.75902581e-01 -5.64435959e-01 -2.70988494e-01
-6.61434591e-01 -3.50130737e-01 2.72948951e-01 4.11877811e-01
1.07911503e+00 -1.27713335e+00 -4.33172137e-01 2.30197400e-01
-4.26043332e-01 -1.12400532e+00 -9.88649487e-01 -3.29462662e-02
-7.80299127e-01 -8.97066593e-01 -1.31409633e+00 -1.12401903e+00
5.71533859e-01 4.94812101e-01 8.26976359e-01 -1.85962915e-01
-2.61891872e-01 -1.20290853e-01 -3.41873080e-01 3.18703353e-02
-6.45235837e-01 -1.63854450e-01 -6.18687749e-01 2.73480892e-01
-3.07080120e-01 -8.03618193e-01 -9.34260070e-01 2.41579935e-01
-1.48828995e+00 5.64192772e-01 1.01261866e+00 1.06237698e+00
6.01280391e-01 6.74317032e-02 4.73199725e-01 -6.76838338e-01
5.50515950e-01 -3.16272885e-01 -4.25402373e-01 4.88307267e-01
-4.61193234e-01 7.54932389e-02 8.12375486e-01 -5.88771462e-01
-1.19081438e+00 2.33234167e-02 -4.31375623e-01 -5.94721138e-01
-9.72216427e-02 6.30982295e-02 -3.77484441e-01 -2.92957217e-01
2.52321243e-01 5.03835976e-01 1.52295128e-01 -2.72920698e-01
8.01919758e-01 5.83689153e-01 9.40305829e-01 -2.57419378e-01
9.06332493e-01 5.56323647e-01 -7.30895698e-02 -4.62866634e-01
-6.43680573e-01 -4.92499955e-02 -2.43173465e-01 -3.66427153e-02
1.08749652e+00 -1.08918083e+00 -2.60248154e-01 7.52758563e-01
-1.19158852e+00 -3.34744632e-01 -3.60979736e-01 1.52508751e-01
-8.87886167e-01 4.79496956e-01 -7.50848293e-01 -3.90931904e-01
-5.62882543e-01 -1.25005341e+00 1.41053247e+00 3.58326972e-01
3.00045609e-01 -7.21905768e-01 1.84188541e-02 4.98725742e-01
5.56020916e-01 5.59630871e-01 9.47099388e-01 1.92570582e-01
-7.69974887e-01 -5.63919283e-02 -4.35451716e-01 5.42963445e-01
2.20432803e-01 -1.42761245e-01 -7.89834678e-01 -3.08039546e-01
5.87601475e-02 -4.85570014e-01 1.05252075e+00 2.78405041e-01
1.11353958e+00 -4.15246993e-01 -1.23444565e-01 9.46795881e-01
1.56070828e+00 2.97417969e-01 1.20621610e+00 2.99791813e-01
8.37596059e-01 6.10610366e-01 8.29513013e-01 8.40338990e-02
1.85249314e-01 9.60542798e-01 5.34603477e-01 -6.48071408e-01
-4.24664229e-01 -7.63845503e-01 3.67747426e-01 3.47487062e-01
3.51656467e-01 -2.81923652e-01 -2.12047368e-01 4.29510623e-01
-1.93524408e+00 -9.15316582e-01 4.01732117e-01 2.02799559e+00
8.68673682e-01 -2.65066445e-01 -1.68578610e-01 -4.70161475e-02
8.42074037e-01 5.57373822e-01 -5.19603431e-01 -4.30627018e-01
-1.14087202e-01 2.00377047e-01 3.03683043e-01 5.69897175e-01
-1.06559074e+00 1.05195248e+00 5.98582888e+00 1.09217227e+00
-1.32754457e+00 9.20187980e-02 8.61554623e-01 1.35385975e-01
-3.41206372e-01 7.00593134e-03 -3.31454426e-01 4.55711931e-01
3.92747849e-01 3.19344968e-01 6.31269217e-01 6.81195974e-01
7.88345858e-02 -1.39471754e-01 -7.79124916e-01 1.10933924e+00
4.37628090e-01 -1.42880309e+00 2.90432215e-01 -2.03728244e-01
1.08764410e+00 -4.86485302e-01 3.25248688e-01 -2.21947711e-02
-1.52515143e-01 -9.77144837e-01 9.05185580e-01 5.36386132e-01
9.35135305e-01 -9.55348730e-01 4.35698241e-01 2.14105412e-01
-9.21632051e-01 -5.93277030e-02 -3.06117654e-01 1.57191992e-01
3.63405883e-01 4.76908475e-01 -2.95903713e-01 6.72848046e-01
5.43929517e-01 6.97956443e-01 -5.47476292e-01 1.06838250e+00
-4.02090520e-01 -6.33508787e-02 8.16116184e-02 5.08069456e-01
2.49590144e-01 -3.16882670e-01 3.95690113e-01 8.51576507e-01
6.12616956e-01 -2.92329103e-01 -1.97043613e-01 1.05624855e+00
-2.76383966e-01 1.19102165e-01 -8.98150802e-01 2.42124870e-01
1.64595932e-01 1.29799700e+00 -4.81162637e-01 -3.86711329e-01
-3.64872336e-01 1.51504922e+00 2.20545352e-01 3.11769754e-01
-1.00956583e+00 -6.40687108e-01 3.82812768e-01 1.66998133e-01
3.79427403e-01 3.72173339e-02 -6.15549497e-02 -1.18861294e+00
2.13151351e-01 -9.32021976e-01 8.93344283e-02 -1.33163345e+00
-1.23166776e+00 7.40884423e-01 6.05604984e-03 -1.41716528e+00
-2.96053112e-01 -1.91196725e-01 -9.26911891e-01 9.18583870e-01
-1.72652030e+00 -1.74625242e+00 -5.16061306e-01 7.41838038e-01
6.35604560e-01 -4.11636829e-02 5.00008166e-01 2.72095054e-01
-1.54462367e-01 4.81473476e-01 1.50660902e-01 -1.13047689e-01
1.02381408e+00 -9.43841815e-01 4.18839216e-01 1.16490805e+00
4.16057892e-02 2.50660330e-01 7.19139993e-01 -4.64175224e-01
-1.26727247e+00 -1.42817998e+00 6.00704968e-01 1.02366187e-01
1.61464900e-01 -3.72697234e-01 -7.45052993e-01 6.27423465e-01
7.67032862e-01 2.06095532e-01 5.28475977e-02 -6.95472538e-01
-4.47385073e-01 -4.15385157e-01 -1.27477717e+00 7.98546970e-01
7.61109591e-01 -4.67419088e-01 -5.81493080e-01 8.81007090e-02
9.05695319e-01 -5.38400829e-01 -7.39842892e-01 4.54824179e-01
3.93518925e-01 -1.14521575e+00 1.13260591e+00 -4.79834601e-02
9.67299640e-01 -4.10202086e-01 -1.53573453e-02 -1.29889548e+00
-7.20046610e-02 -8.36733162e-01 -1.25880959e-02 1.35259521e+00
-5.65207228e-02 -3.95489246e-01 4.92189199e-01 2.36591235e-01
-1.58184454e-01 -7.21640110e-01 -6.83070421e-01 -4.13163990e-01
1.86348185e-01 1.31481951e-02 7.17296720e-01 6.84299350e-01
-3.56797785e-01 1.96499541e-01 -1.10094774e+00 1.99122690e-02
6.02590263e-01 6.83932245e-01 9.78915393e-01 -3.04737210e-01
-5.36487639e-01 -3.89302403e-01 -2.20545366e-01 -1.22393060e+00
1.59703493e-01 -5.16612709e-01 2.41024971e-01 -1.58906758e+00
2.42424533e-01 -2.83662647e-01 3.52590010e-02 2.41338685e-01
-2.80314893e-01 8.06061924e-01 4.69930500e-01 2.23373190e-01
-3.03704441e-01 7.97727704e-01 1.78853488e+00 -3.07281047e-01
-1.72630921e-01 -1.57883480e-01 -8.20820451e-01 3.43193531e-01
5.22724867e-01 -2.11272538e-01 -5.05864441e-01 -4.32960540e-01
-1.03537008e-01 6.03008330e-01 6.48590207e-01 -1.04891443e+00
1.50804222e-01 -1.45917490e-01 4.76049751e-01 -4.77031201e-01
5.94723284e-01 -9.16709363e-01 3.56108755e-01 2.91942090e-01
-4.30196404e-01 1.46954313e-01 4.02977839e-02 5.43110847e-01
-5.96529722e-01 -4.72737029e-02 1.16343045e+00 -1.68393239e-01
-7.61708498e-01 3.86531591e-01 1.30719338e-02 -2.41880104e-01
1.29327953e+00 -1.20521775e-02 -3.50026011e-01 -6.79571986e-01
-3.61237258e-01 -6.79993555e-02 7.55225360e-01 5.84903300e-01
8.91958356e-01 -1.47727025e+00 -6.16610467e-01 4.19732571e-01
1.17660515e-01 -4.73367088e-02 4.90328550e-01 5.44792533e-01
-7.91566670e-01 2.86416374e-02 -6.97354019e-01 -2.77908921e-01
-1.00655854e+00 8.82185757e-01 2.60609657e-01 -3.65732163e-01
-7.05953956e-01 5.70622265e-01 4.53925312e-01 -2.68647909e-01
2.57423133e-01 -2.88505912e-01 1.63292274e-01 -3.95322919e-01
5.10499835e-01 7.94428121e-03 -2.32342854e-01 -5.26710689e-01
1.09070852e-01 9.29868162e-01 -6.35800734e-02 -1.36082619e-01
1.20134223e+00 -3.92093301e-01 -2.99287647e-01 1.12195604e-01
1.29047501e+00 -1.32122263e-02 -1.70746875e+00 -3.77635583e-02
-6.56785965e-01 -6.85887933e-01 -6.17433265e-02 -7.56766200e-01
-1.19379950e+00 9.76326466e-01 6.34589195e-01 -1.27186120e-01
1.59734404e+00 -2.11502910e-01 1.24188912e+00 -1.89789534e-01
1.76861957e-01 -6.98464453e-01 2.70357817e-01 -5.59749547e-04
1.21206415e+00 -1.34270608e+00 -6.03718683e-02 -3.83247077e-01
-7.20510542e-01 1.04819858e+00 6.64215088e-01 -5.81552804e-01
3.43909413e-01 1.45065174e-01 1.34469241e-01 2.06111535e-01
-5.42194724e-01 9.51320305e-02 2.51846403e-01 6.12286806e-01
1.89243764e-01 -3.37725401e-01 -2.77758271e-01 -2.40168143e-02
4.17681858e-02 -1.11107975e-02 4.19147432e-01 5.84644675e-01
-2.22981021e-01 -1.15588462e+00 -4.68925625e-01 -4.49408405e-02
-4.09953684e-01 -2.03866363e-01 -7.32990876e-02 8.52169514e-01
2.79357314e-01 5.80619574e-01 1.65632755e-01 -2.31680021e-01
1.55962795e-01 -2.81360537e-01 6.06415868e-01 -2.38143459e-01
-4.75362301e-01 2.56354719e-01 -2.94816673e-01 -6.25614405e-01
-4.64420378e-01 -8.98030922e-02 -7.67219365e-01 -2.22602144e-01
-5.93694746e-02 -7.42464736e-02 6.07948065e-01 6.62610352e-01
3.44584674e-01 4.80492979e-01 7.79457152e-01 -1.17479420e+00
-3.54374170e-01 -8.44191551e-01 -6.42925143e-01 5.90924323e-01
5.39738476e-01 -2.39200741e-01 -7.41632953e-02 1.38707086e-01] | [11.387192726135254, -1.0996010303497314] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.