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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a31b9505-688f-4a96-9007-1fa242a5b069 | uncovering-and-quantifying-social-biases-in | 2305.15377 | null | https://arxiv.org/abs/2305.15377v1 | https://arxiv.org/pdf/2305.15377v1.pdf | Uncovering and Quantifying Social Biases in Code Generation | With the popularity of automatic code generation tools, such as Copilot, the study of the potential hazards of these tools is gaining importance. In this work, we explore the social bias problem in pre-trained code generation models. We propose a new paradigm to construct code prompts and successfully uncover social biases in code generation models. To quantify the severity of social biases in generated code, we develop a dataset along with three metrics to evaluate the overall social bias and fine-grained unfairness across different demographics. Experimental results on three pre-trained code generation models (Codex, InCoder, and CodeGen) with varying sizes, reveal severe social biases. Moreover, we conduct analysis to provide useful insights for further choice of code generation models with low social bias. (This work contains examples that potentially implicate stereotypes, associations, and other harms that could be offensive to individuals in certain social groups.) | ['Tsung-Yi Ho', 'Pin-Yu Chen', 'Jian-Guang Lou', 'Daoguang Zan', 'Fengji Zhang', 'Zhe Su', 'Yan Gao', 'Xiaokang Chen', 'Yan Liu'] | 2023-05-24 | null | null | null | null | ['code-generation'] | ['computer-code'] | [-3.05576678e-02 4.28153247e-01 -2.31215537e-01 -4.02819067e-01
-6.36328906e-02 -6.08948648e-01 6.72960162e-01 5.63781023e-01
5.81295006e-02 5.07642865e-01 7.94326186e-01 -5.79691172e-01
6.29315004e-02 -6.53849781e-01 -3.64188701e-01 -2.90169865e-01
-1.64919928e-01 -2.28150144e-01 -3.67658645e-01 -4.33387309e-01
9.87394035e-01 -7.10367262e-02 -1.31250668e+00 3.45927119e-01
1.49226725e+00 2.39617705e-01 -2.36012176e-01 5.25449336e-01
2.97882259e-02 1.12082160e+00 -7.77114153e-01 -1.02908587e+00
1.68385610e-01 -2.47429162e-01 -7.17841506e-01 -3.55954587e-01
5.14240384e-01 -2.72823423e-01 1.67036176e-01 1.49290717e+00
6.06358767e-01 -2.67551839e-01 7.05685079e-01 -1.51952565e+00
-1.14436615e+00 1.17838478e+00 -6.66579247e-01 2.66596168e-01
8.74797463e-01 2.98782349e-01 1.06937492e+00 -5.63468099e-01
5.64410925e-01 1.37914586e+00 8.85599315e-01 6.47558093e-01
-1.40700138e+00 -1.29832244e+00 3.72008840e-03 -4.73648399e-01
-9.74663377e-01 -4.71207350e-01 6.29075348e-01 -1.18522406e+00
6.28281116e-01 2.53982753e-01 4.88291442e-01 1.46883941e+00
4.74056333e-01 3.22127461e-01 1.21479416e+00 -1.52119398e-01
1.11944079e-01 4.74184573e-01 3.40876460e-01 7.39580631e-01
7.91911006e-01 3.71666491e-01 -4.30869967e-01 -1.01513720e+00
1.53937861e-01 -1.37289958e-02 1.12198498e-02 -1.86746791e-01
-1.04761589e+00 1.27960765e+00 4.67800617e-01 1.60852030e-01
4.91775312e-02 1.99869365e-01 5.12963235e-01 2.05819234e-01
6.48509860e-01 9.95092869e-01 -4.00564894e-02 -3.00931662e-01
-7.72411346e-01 6.87453389e-01 7.33278990e-01 9.32691336e-01
8.09632003e-01 -7.99447298e-02 -5.25387526e-01 9.89227891e-01
3.51501435e-01 4.66733962e-01 7.28276312e-01 -9.10091639e-01
6.90337956e-01 8.71399164e-01 1.07703909e-01 -1.65002966e+00
-2.58681953e-01 -4.09911454e-01 -4.77697104e-01 9.76564139e-02
1.68119162e-01 -5.92922628e-01 2.06776359e-03 1.63927543e+00
-1.51719734e-01 -3.96726221e-01 -3.27511013e-01 4.34918821e-01
4.48162496e-01 4.41234708e-02 3.90835285e-01 2.41795748e-01
1.09088850e+00 -6.26909673e-01 -2.93144584e-01 -3.46882880e-01
9.80742097e-01 -8.55991781e-01 1.15973473e+00 -9.33739766e-02
-7.94639528e-01 -2.55861551e-01 -6.71639919e-01 2.10577905e-01
-1.31808057e-01 -5.26053905e-02 8.88728321e-01 1.28170240e+00
-9.29726660e-01 5.93032360e-01 1.89433317e-03 -2.96024233e-01
6.38295054e-01 9.16108117e-02 -3.70201282e-02 1.45687416e-01
-1.07114089e+00 6.59529090e-01 2.91787162e-02 -5.35854399e-01
-8.49670649e-01 -7.33109593e-01 -7.58622110e-01 1.91646650e-01
2.88175140e-02 -5.94798744e-01 1.04313207e+00 -1.45945668e+00
-7.67374277e-01 8.95551205e-01 2.64852136e-01 -2.38525763e-01
4.97706980e-01 -2.21239656e-01 -2.23804593e-01 -4.77092475e-01
5.89606762e-01 5.26089907e-01 8.81025970e-01 -1.34748960e+00
-2.30737984e-01 -1.20038979e-01 4.57329273e-01 -2.31912822e-01
-5.86634874e-01 5.36953986e-01 9.83862221e-01 -1.00864899e+00
-7.53097773e-01 -1.19238496e+00 -3.14796120e-01 -3.71583015e-01
-6.70797765e-01 -1.56351581e-01 -1.38363810e-02 -4.44220811e-01
1.70684505e+00 -2.27669120e+00 -2.41654664e-01 2.37626284e-01
4.59957153e-01 1.18395843e-01 -1.87891051e-01 5.28793275e-01
-3.41869384e-01 7.13216662e-01 -2.26772860e-01 -1.39208645e-01
2.07740068e-02 -3.63299131e-01 -4.61082429e-01 1.97534814e-01
1.78787172e-01 6.06593072e-01 -1.29022157e+00 -5.39698839e-01
-5.31945527e-01 -7.86467791e-02 -1.36838233e+00 2.15128452e-01
1.22962683e-01 5.63632436e-02 -4.91620690e-01 6.58773243e-01
6.46368861e-01 -1.20333128e-01 3.38428281e-03 5.02869904e-01
-3.01534742e-01 3.18578094e-01 -3.30017209e-01 1.08077037e+00
-3.72876734e-01 6.83003008e-01 -1.07505694e-01 -1.85764745e-01
9.45761263e-01 -1.02780208e-01 3.58053260e-02 -3.02767634e-01
3.46644998e-01 1.16748109e-01 3.58222157e-01 -6.96135640e-01
9.12814975e-01 -2.63188720e-01 -3.77703428e-01 8.55015635e-01
-2.85640031e-01 -6.80035772e-03 2.87195165e-02 5.23371160e-01
1.19535172e+00 -4.80784506e-01 4.67735052e-01 -8.17337453e-01
3.16411018e-01 -2.92767495e-01 3.51427048e-01 7.90838718e-01
-5.64882576e-01 3.42318654e-01 1.03925836e+00 -2.21927956e-01
-6.73040867e-01 -6.85712695e-01 2.03030985e-02 1.46805573e+00
3.78806666e-02 -7.11867034e-01 -7.91801095e-01 -1.01066256e+00
3.78894448e-01 8.18817437e-01 -7.89520860e-01 -5.65871894e-01
-1.04825146e-01 -9.22466755e-01 9.14517879e-01 1.87932551e-01
1.89632982e-01 -9.53811824e-01 -5.58788717e-01 -1.97791785e-01
-2.07985520e-01 -4.62884873e-01 -7.57581353e-01 -4.49023247e-01
-6.97781503e-01 -1.30887139e+00 -4.35355783e-01 -4.43757743e-01
1.02006245e+00 4.51479793e-01 1.23341763e+00 8.01135123e-01
-3.93750668e-02 1.62599862e-01 -5.55854023e-01 -6.22785091e-01
-8.07974398e-01 2.03257352e-01 -2.24454366e-02 -1.55964881e-01
2.73977518e-01 -4.44341719e-01 -5.45990407e-01 3.36447358e-01
-7.76956320e-01 3.83967767e-03 3.22110087e-01 4.86988574e-01
-8.99672210e-01 -4.20438439e-01 7.96344101e-01 -1.28416193e+00
1.43587291e+00 -1.22989035e+00 -2.32762158e-01 -1.44565031e-02
-9.49983478e-01 -3.35022085e-03 4.06545967e-01 -3.89782339e-01
-1.23796928e+00 -4.99787271e-01 1.64477497e-01 3.30214381e-01
7.52709210e-02 4.62970436e-01 4.49914366e-01 -2.25473970e-01
1.37539327e+00 -3.56316417e-01 -3.24468613e-02 -2.48736218e-01
1.39651313e-01 9.43310916e-01 -2.83031106e-01 -8.49856436e-01
8.90937269e-01 1.91026568e-01 -5.19205034e-01 -3.18893135e-01
-7.91078389e-01 -1.29275456e-01 -1.57605447e-02 -2.85087109e-01
6.07450902e-01 -9.26666260e-01 -6.48160577e-01 2.77797937e-01
-1.18176544e+00 -3.85361791e-01 2.25777373e-01 6.51982725e-02
-2.17210039e-01 4.69232827e-01 -4.49226350e-01 -8.36863935e-01
-1.69108808e-01 -1.10032928e+00 8.13601077e-01 1.19284846e-01
-8.96338165e-01 -9.03834701e-01 4.85697120e-01 6.69793427e-01
8.44270945e-01 3.91543031e-01 1.12862408e+00 -5.58321357e-01
-3.66316527e-01 6.86352998e-02 -5.36833763e-01 2.31878594e-01
9.96758509e-03 3.20626318e-01 -8.68442297e-01 -2.61300504e-01
-2.83284813e-01 -4.19886380e-01 6.00406110e-01 -8.92370269e-02
1.09583271e+00 -7.36856580e-01 -2.88875103e-01 2.73537755e-01
1.10904968e+00 -2.82338951e-02 4.40538049e-01 1.39277816e-01
7.55645275e-01 9.59314764e-01 5.35753846e-01 8.39871824e-01
5.33487737e-01 3.24845225e-01 4.70805615e-01 3.70369405e-01
1.71935380e-01 -4.11356360e-01 8.28899026e-01 6.27377093e-01
-8.14380422e-02 1.50666619e-02 -1.27925050e+00 5.80076516e-01
-1.56281793e+00 -1.08683586e+00 -2.66301781e-01 1.93234515e+00
8.52037311e-01 -2.17867363e-02 1.77283242e-01 -2.05504194e-01
9.05007422e-01 4.26379353e-01 -5.92234693e-02 -9.06786919e-01
2.05158547e-01 -2.28862867e-01 3.66581231e-01 4.70282972e-01
-7.39527881e-01 5.83478391e-01 7.48187351e+00 5.04729629e-01
-9.68046606e-01 -1.01902485e-01 8.42417896e-01 -9.72505063e-02
-8.14451277e-01 2.62437314e-01 -5.58286965e-01 9.75857973e-01
8.09231341e-01 -7.13603795e-01 5.77034950e-01 1.09476650e+00
9.11302492e-02 -2.94716116e-02 -1.11726546e+00 7.06023574e-01
4.02602702e-01 -8.68449152e-01 -9.48899705e-03 2.08080247e-01
1.14813662e+00 -1.39282376e-01 1.04201384e-01 5.01774192e-01
8.96218717e-01 -9.61566865e-01 8.76974285e-01 2.56978780e-01
5.36328137e-01 -4.75653738e-01 5.89578390e-01 2.92098254e-01
-4.28881288e-01 -7.25241542e-01 -3.26593131e-01 -6.99764311e-01
-3.17948639e-01 8.35385561e-01 -8.69870126e-01 7.21277669e-03
3.59079331e-01 7.56590188e-01 -1.34252667e+00 8.88673663e-01
-1.66979045e-01 5.74476838e-01 5.84780931e-01 -1.93172023e-01
-1.47110850e-01 1.23529710e-01 4.83064622e-01 1.28435349e+00
3.92571062e-01 -3.78200918e-01 -2.24694252e-01 1.31508863e+00
-3.40215862e-01 1.45519987e-01 -9.72532690e-01 -1.71184868e-01
4.85556036e-01 1.30411482e+00 -4.61351782e-01 -1.21806040e-01
-3.00647080e-01 4.30814207e-01 5.42498529e-01 1.92259490e-01
-8.87364507e-01 -3.82950485e-01 8.95336807e-01 3.46973509e-01
-5.64897358e-01 4.10587303e-02 -4.98063266e-01 -1.29842043e+00
-2.96969205e-01 -1.19384181e+00 9.68642607e-02 -5.90786219e-01
-1.42884731e+00 2.36530185e-01 6.49377005e-03 -9.82344925e-01
-4.87524830e-02 -2.44132042e-01 -1.01795328e+00 5.76686621e-01
-1.25782049e+00 -5.92286408e-01 -4.87308323e-01 1.60332412e-01
1.10154793e-01 -4.62248772e-01 4.28245515e-01 3.43846887e-01
-4.56327170e-01 7.41012335e-01 -3.85334045e-01 1.52266070e-01
1.08276320e+00 -1.15165305e+00 5.47296762e-01 5.88133097e-01
-5.33900201e-01 1.22608042e+00 6.38505280e-01 -8.69871438e-01
-6.18298590e-01 -1.05221283e+00 1.17684150e+00 -9.13956821e-01
7.59358048e-01 -3.79287958e-01 -5.01875043e-01 6.46052778e-01
2.48386845e-01 -5.47904313e-01 1.21874297e+00 3.83661807e-01
-7.15430737e-01 3.24871153e-01 -1.17692864e+00 7.35041738e-01
1.36582303e+00 -6.67673826e-01 -4.66633111e-01 2.12878779e-01
6.54514372e-01 1.53109848e-01 -5.13404310e-01 -1.07965313e-01
6.11362219e-01 -1.62788022e+00 6.39851272e-01 -5.86414814e-01
1.15324295e+00 2.13815451e-01 8.77855048e-02 -1.61936784e+00
-6.03085995e-01 -7.33558714e-01 5.25372505e-01 1.49922764e+00
4.10690010e-01 -7.62649596e-01 4.28334653e-01 1.09697998e+00
7.24852681e-02 -3.81109357e-01 -3.77203047e-01 -4.44433361e-01
2.98606992e-01 5.12631610e-02 9.09816206e-01 1.51886833e+00
5.87096691e-01 8.45134929e-02 -3.74074161e-01 -2.76833147e-01
4.89173025e-01 -5.65664619e-02 8.98520470e-01 -1.28155839e+00
-1.91227958e-01 -6.53449178e-01 -2.61388510e-01 -3.66941720e-01
6.84146345e-01 -1.15343380e+00 -2.09331900e-01 -7.82529056e-01
7.52909124e-01 -6.42212093e-01 1.39174908e-01 3.55300665e-01
-5.92660725e-01 -8.84641334e-02 3.87982875e-01 1.18506938e-01
-4.70074117e-01 4.40374970e-01 9.80310142e-01 -4.46615666e-02
1.27344117e-01 -1.17168307e-01 -1.57081890e+00 7.46252775e-01
9.61036801e-01 -8.61222446e-01 -3.01389694e-01 -4.24174428e-01
1.11487603e+00 -1.25019670e-01 4.82689679e-01 -7.95868635e-01
-2.62158573e-01 -4.13258344e-01 -2.25889891e-01 4.26546872e-01
-5.43612659e-01 -5.98758936e-01 -4.27349657e-02 6.97496355e-01
-8.81174266e-01 4.46956575e-01 -1.07041143e-01 2.71677464e-01
1.51284328e-02 -4.93993014e-01 6.09400928e-01 -6.96025044e-02
2.58781254e-01 -6.93342015e-02 -5.90486646e-01 4.56219584e-01
1.08365583e+00 1.07514694e-01 -7.71771431e-01 -4.48510915e-01
-4.92792986e-02 2.26311520e-01 8.70318055e-01 8.61825287e-01
2.11527005e-01 -1.37196934e+00 -7.33341455e-01 1.61846936e-01
4.83125478e-01 -7.62817681e-01 1.29489088e-03 5.98809659e-01
-5.03516436e-01 -2.38414798e-02 -5.15001893e-01 2.77177840e-02
-1.33600903e+00 5.16081035e-01 1.66592300e-01 -3.59663032e-02
1.99704707e-01 7.46132731e-01 3.97346020e-01 -4.64563578e-01
-2.34141961e-01 -3.96047175e-01 -1.55572295e-01 2.01364860e-01
3.79180253e-01 9.08866167e-01 -4.20758814e-01 -6.49230897e-01
-2.81969845e-01 6.41329065e-02 1.08612530e-01 2.86885023e-01
8.57890546e-01 -6.16482198e-02 -5.02471089e-01 1.43036216e-01
9.78663325e-01 6.39528036e-01 -8.21979702e-01 3.12721074e-01
2.02351406e-01 -1.16597700e+00 -5.18248260e-01 -7.53256917e-01
-8.30884635e-01 7.30235040e-01 1.48544431e-01 5.76947570e-01
4.46388274e-01 -2.71442473e-01 3.77638996e-01 -6.28592074e-02
5.02923906e-01 -8.63196254e-01 4.16956782e-01 6.13705099e-01
1.02485585e+00 -1.25178182e+00 -8.48814473e-03 -3.84751767e-01
-6.76996350e-01 6.62209511e-01 7.60476172e-01 -2.15367913e-01
6.94457531e-01 1.20383300e-01 7.44194910e-02 -2.78068632e-01
-8.25812042e-01 4.33599919e-01 -3.30024287e-02 5.48361003e-01
1.06957614e+00 4.34109211e-01 -6.49231136e-01 7.36570954e-01
-7.37470984e-01 -1.24421299e-01 1.06962132e+00 5.99209607e-01
-4.01794523e-01 -1.10776520e+00 -4.35038716e-01 7.17116714e-01
-7.02450454e-01 -3.44078392e-01 -9.72700417e-01 1.86509743e-01
4.43095356e-01 1.11422825e+00 -2.14542851e-01 -7.05727160e-01
7.38179311e-02 -2.58680917e-02 -5.92971593e-02 -9.03175533e-01
-1.46319878e+00 -8.82756531e-01 4.55819726e-01 -5.50108373e-01
-2.39967838e-01 -7.54018545e-01 -6.75109327e-01 -7.63941705e-01
-3.83356094e-01 3.41234989e-02 2.88175195e-01 3.04133236e-01
6.95699275e-01 1.66877821e-01 7.46130645e-01 -6.80400550e-01
-5.76032400e-01 -9.77758467e-01 -3.89473319e-01 9.96114135e-01
3.64381194e-01 -7.78047502e-01 -6.83240473e-01 -2.19484910e-01] | [9.01158618927002, 10.19959545135498] |
19f1bd45-c1ef-4f7d-8811-63da2b275996 | tinycd-a-not-so-deep-learning-model-for | 2207.13159 | null | https://arxiv.org/abs/2207.13159v2 | https://arxiv.org/pdf/2207.13159v2.pdf | TINYCD: A (Not So) Deep Learning Model For Change Detection | In this paper, we present a lightweight and effective change detection model, called TinyCD. This model has been designed to be faster and smaller than current state-of-the-art change detection models due to industrial needs. Despite being from 13 to 140 times smaller than the compared change detection models, and exposing at least a third of the computational complexity, our model outperforms the current state-of-the-art models by at least $1\%$ on both F1 score and IoU on the LEVIR-CD dataset, and more than $8\%$ on the WHU-CD dataset. To reach these results, TinyCD uses a Siamese U-Net architecture exploiting low-level features in a globally temporal and locally spatial way. In addition, it adopts a new strategy to mix features in the space-time domain both to merge the embeddings obtained from the Siamese backbones, and, coupled with an MLP block, it forms a novel space-semantic attention mechanism, the Mix and Attention Mask Block (MAMB). Source code, models and results are available here: https://github.com/AndreaCodegoni/Tiny_model_4_CD | ['Alessandro Ferrari', 'Gabriele Lombardi', 'Andrea Codegoni'] | 2022-07-26 | null | null | null | null | ['change-detection-for-remote-sensing-images', 'building-change-detection-for-remote-sensing'] | ['miscellaneous', 'miscellaneous'] | [-2.01649934e-01 -2.81621426e-01 -1.95937872e-01 -1.06081985e-01
-5.03791630e-01 -3.77573162e-01 7.16029823e-01 6.28536120e-02
-6.85367882e-01 3.64356726e-01 1.69180214e-01 8.63321349e-02
7.07922829e-03 -9.33488429e-01 -8.53205383e-01 -4.88967806e-01
-2.66899198e-01 1.24149114e-01 4.66688395e-01 -4.10227358e-01
1.84650183e-01 2.52208591e-01 -1.86452651e+00 -3.89034376e-02
8.55690300e-01 1.17687845e+00 1.98569730e-01 4.32051033e-01
-2.19207983e-02 5.49406588e-01 -1.89993143e-01 -2.00860679e-01
3.21157575e-01 -4.45224583e-01 -6.32346392e-01 -4.46333915e-01
7.09881723e-01 -2.79369473e-01 -5.93212962e-01 1.02583385e+00
6.91536367e-01 4.44026858e-01 2.24197626e-01 -1.08340716e+00
-9.50950503e-01 6.38158143e-01 -4.58981156e-01 6.29364371e-01
2.13619873e-01 1.27603322e-01 1.23852909e+00 -1.11384046e+00
9.70418215e-01 1.17868435e+00 7.18666196e-01 3.87035906e-01
-1.06467450e+00 -7.22212255e-01 5.15544832e-01 6.07994914e-01
-1.43782771e+00 -2.75690466e-01 8.67515385e-01 -3.63578737e-01
1.40542924e+00 1.82423621e-01 8.32304895e-01 1.10517085e+00
2.89466888e-01 9.55441654e-01 6.95431054e-01 -3.26357156e-01
2.24295482e-01 -9.37071666e-02 1.86797440e-01 7.43314803e-01
7.89362118e-02 3.66657078e-02 -5.30293524e-01 1.25706539e-01
5.48110068e-01 3.08636934e-01 -3.03485692e-01 -3.24552745e-01
-1.40856659e+00 7.51019359e-01 8.81002188e-01 8.62557590e-01
-3.95310819e-01 3.86590093e-01 4.04048651e-01 2.77126491e-01
5.60017347e-01 4.56807911e-01 -5.11542320e-01 -5.03691912e-01
-8.16953480e-01 3.67671996e-01 3.95799220e-01 7.61080265e-01
8.06282461e-01 -4.75970916e-02 -5.72519898e-02 7.68485725e-01
2.20018119e-01 4.13598329e-01 9.37059641e-01 -8.01509380e-01
4.33095604e-01 7.90993810e-01 -9.00131389e-02 -1.03523636e+00
-4.44031268e-01 -7.21694887e-01 -8.25092435e-01 2.13381331e-02
-3.46059538e-03 2.59233385e-01 -8.33154261e-01 1.94671702e+00
3.57195944e-01 2.89329022e-01 -4.19170707e-01 8.60318661e-01
5.90185165e-01 6.57954514e-01 -2.57427901e-01 1.78479165e-01
1.25992465e+00 -1.38306284e+00 -8.10351491e-01 -2.24435106e-01
6.04651868e-01 -4.96986896e-01 1.39585543e+00 1.79029882e-01
-1.13491821e+00 -8.61008167e-01 -1.35967195e+00 -2.99362481e-01
-8.13203633e-01 -2.72771828e-02 3.47600877e-01 2.75105983e-01
-1.17731655e+00 8.27297568e-01 -1.05943763e+00 -6.68989599e-01
3.12724411e-01 1.11606054e-01 -3.10605615e-01 -1.60746843e-01
-1.43121147e+00 8.35892856e-01 2.79784411e-01 -7.96489138e-03
-6.81575894e-01 -7.86132932e-01 -9.27304745e-01 7.56313428e-02
2.33620286e-01 -5.23966849e-01 1.07523668e+00 -7.37044632e-01
-1.35843897e+00 6.81120098e-01 -1.87540308e-01 -5.29377759e-01
5.09586334e-01 -4.43246216e-01 -6.47799134e-01 2.65899207e-02
-2.39358633e-03 8.32508385e-01 6.34505212e-01 -7.93699503e-01
-7.96928525e-01 -2.87381619e-01 6.32823557e-02 1.14572026e-01
-4.68902856e-01 -9.45596173e-02 -7.23892093e-01 -9.31704521e-01
5.90360202e-02 -9.19004738e-01 -5.90561330e-02 3.00091151e-02
1.62964649e-02 -2.61692971e-01 8.97840858e-01 -8.28416884e-01
1.71997285e+00 -2.32808232e+00 2.54870653e-01 -7.06282705e-02
1.90937951e-01 4.52422678e-01 -3.34316522e-01 4.49818224e-01
-8.14154521e-02 1.40822291e-01 -4.24303204e-01 -4.38858241e-01
2.44211897e-01 1.33252732e-04 6.83890730e-02 5.89794517e-01
8.73771608e-02 9.77291703e-01 -9.13367212e-01 4.16741185e-02
4.41824138e-01 4.48957235e-01 -8.45287025e-01 -1.45784050e-01
-1.07822835e-01 7.31866881e-02 -5.82963461e-04 6.89530015e-01
6.88969910e-01 -1.54367059e-01 -5.10290675e-02 -1.83025226e-02
-2.84335613e-01 4.08251107e-01 -1.30610323e+00 2.12953639e+00
-3.75747651e-01 6.67766333e-01 -1.73408359e-01 -8.19386184e-01
8.38682711e-01 -2.25080363e-02 6.88456833e-01 -1.10041070e+00
1.14056379e-01 2.85926014e-01 -5.43746725e-03 -1.93750784e-01
5.49457312e-01 2.71328241e-01 -1.92575514e-01 3.55253816e-01
1.75957069e-01 1.45877376e-01 4.85902667e-01 2.77397819e-02
1.37998307e+00 2.31739417e-01 1.96468756e-01 -3.52202892e-01
6.16314054e-01 -2.96040237e-01 8.37811410e-01 4.94192839e-01
-5.98103404e-01 5.05369663e-01 1.78487867e-01 -5.92687607e-01
-7.58509815e-01 -1.08174217e+00 -8.04104656e-02 1.14297116e+00
2.94763297e-01 -5.99439681e-01 -8.39911342e-01 -6.02635145e-01
2.48793155e-01 7.33200371e-01 -8.26612294e-01 -3.47443640e-01
-7.50421941e-01 -5.80040276e-01 3.92214507e-01 5.48513114e-01
8.88239563e-01 -1.11728561e+00 -4.88596290e-01 4.15114820e-01
-1.99342296e-01 -8.36366951e-01 -6.57807946e-01 6.38522506e-02
-7.41888165e-01 -9.87885118e-01 -5.19296587e-01 -7.37177908e-01
2.05583990e-01 2.30638340e-01 1.03160203e+00 -3.11865002e-01
-3.36771578e-01 3.77531685e-02 -5.51327944e-01 -3.15345436e-01
8.62079263e-02 2.59575248e-01 2.09370568e-01 -8.49828124e-02
5.98689854e-01 -7.03284323e-01 -8.45245957e-01 1.24564908e-01
-1.09958625e+00 -2.60887593e-01 5.61976910e-01 6.63194716e-01
7.52262712e-01 5.67561053e-02 4.65869635e-01 -4.64525878e-01
4.57861543e-01 -5.75430930e-01 -4.68984365e-01 -3.44004594e-02
-8.62577617e-01 -2.06272647e-01 5.87510288e-01 -4.36987668e-01
-7.27916420e-01 -2.75480717e-01 -3.91482890e-01 -5.35981894e-01
2.06424519e-02 3.69890600e-01 -9.25504640e-02 1.58001602e-01
6.07808650e-01 2.23375946e-01 -2.67963171e-01 -8.25751364e-01
5.23349285e-01 5.41039050e-01 3.90767246e-01 -3.71425413e-02
6.71900213e-01 5.90355992e-01 -5.87270856e-01 -6.14468575e-01
-4.44022030e-01 -5.53439975e-01 -8.60245228e-01 -1.08723536e-01
7.57207215e-01 -1.04400849e+00 -2.55278319e-01 9.69833374e-01
-9.19861078e-01 -4.43772435e-01 -4.89059478e-01 5.09175539e-01
-5.05040228e-01 6.31164983e-02 -7.96387374e-01 -1.87953502e-01
-4.94665802e-01 -8.86390448e-01 8.26513350e-01 1.61582053e-01
-2.36887515e-01 -8.74472260e-01 3.59976202e-01 1.70470789e-01
7.91077375e-01 4.26693618e-01 7.34473586e-01 -4.78279382e-01
-4.85729307e-01 -1.11890785e-01 -1.11573182e-01 4.55571234e-01
2.91515559e-01 -1.40692487e-01 -8.32679212e-01 -5.69614053e-01
-4.73296940e-02 1.93823069e-01 1.19783545e+00 4.14135188e-01
1.13544309e+00 -2.24978507e-01 -3.18475187e-01 5.97453058e-01
1.55428743e+00 2.88852423e-01 6.05117559e-01 5.29164255e-01
6.13909006e-01 1.50405020e-01 4.77612108e-01 3.92889231e-01
6.60517871e-01 7.69588530e-01 4.97700810e-01 -3.18402164e-02
-5.12667120e-01 -1.85870692e-01 6.16254747e-01 1.32831419e+00
1.51305780e-01 -1.21293850e-01 -7.68029690e-01 9.42127645e-01
-1.93486023e+00 -1.05895674e+00 9.36772022e-03 2.03980064e+00
7.43889391e-01 2.50089079e-01 7.98899159e-02 1.53390586e-01
6.19620204e-01 4.65333402e-01 -7.90778697e-01 -4.13471073e-01
-2.08457023e-01 2.01670051e-01 3.48228425e-01 3.13248694e-01
-1.16822898e+00 9.31725264e-01 6.09414434e+00 8.58635366e-01
-1.29626036e+00 5.08111417e-01 2.57016510e-01 -4.90515232e-01
-3.03866446e-01 -3.72687638e-01 -6.94273114e-01 7.58121848e-01
9.43836689e-01 1.08331898e-02 6.23977542e-01 7.92056322e-01
1.54122517e-01 -3.99360210e-02 -9.59236562e-01 9.39813852e-01
3.98732960e-01 -1.33106816e+00 -1.74784616e-01 -7.77428821e-02
8.45803499e-01 6.98818862e-01 -7.60658532e-02 6.08363390e-01
1.37973934e-01 -5.78813851e-01 9.59407270e-01 4.55745369e-01
5.69459856e-01 -6.63448036e-01 8.20627451e-01 1.83070362e-01
-1.49163210e+00 -3.03906888e-01 -1.10632025e-01 -1.57808095e-01
3.94865572e-02 6.72382534e-01 -4.02450673e-02 5.81339896e-01
1.20558631e+00 1.10686827e+00 -6.59936070e-01 9.97473776e-01
-1.16079070e-01 4.98232782e-01 -3.45634282e-01 8.69463198e-03
4.77698028e-01 -8.67967308e-02 5.26726604e-01 1.28222346e+00
5.60777724e-01 -3.78374666e-01 -9.45068747e-02 8.65338087e-01
-2.89776891e-01 7.94439986e-02 -5.12602031e-01 5.95363751e-02
5.82714856e-01 1.02869225e+00 -3.83841097e-01 -2.83352047e-01
-5.63798010e-01 1.22436047e+00 4.21775371e-01 2.69587666e-01
-1.09018445e+00 -8.72884989e-01 1.08074057e+00 6.52728230e-02
7.40508676e-01 -2.60133088e-01 -7.85470158e-02 -1.23453784e+00
3.44319791e-01 -6.90179825e-01 3.50894392e-01 -4.84959811e-01
-1.12821412e+00 5.31527460e-01 -2.53336847e-01 -1.22628069e+00
4.29042764e-02 -4.72733498e-01 -5.31947732e-01 7.68367469e-01
-1.67170238e+00 -1.07019091e+00 -2.94235468e-01 5.24540484e-01
5.99805176e-01 6.82106912e-02 6.82663620e-01 7.01432168e-01
-8.84528518e-01 6.15341485e-01 4.75243807e-01 1.27237529e-01
7.34698832e-01 -1.14153636e+00 9.51407731e-01 1.14386559e+00
9.91769731e-02 5.83173811e-01 4.61710066e-01 -4.60741490e-01
-1.36282122e+00 -1.20707381e+00 1.05902743e+00 -2.92297930e-01
8.79717469e-01 -5.02203584e-01 -9.28135395e-01 7.66779482e-01
3.01583767e-01 1.42000303e-01 4.70805645e-01 7.91360661e-02
-4.41957742e-01 -3.97151589e-01 -1.20814109e+00 5.34091711e-01
1.45290768e+00 -4.89760429e-01 -7.16731846e-01 -1.12712212e-01
1.04647160e+00 -3.16564679e-01 -1.17216086e+00 3.89386922e-01
5.36763132e-01 -9.21306729e-01 7.59316802e-01 -3.11490804e-01
1.89938113e-01 -4.17818010e-01 -4.20771331e-01 -1.44111001e+00
-8.97853613e-01 -3.63455862e-01 -5.24240553e-01 1.27835989e+00
2.71821380e-01 -8.47457647e-01 3.27483326e-01 -2.85950322e-02
-3.13677877e-01 -7.92033195e-01 -1.28014886e+00 -1.02951038e+00
2.17353299e-01 -6.49178207e-01 7.65278161e-01 1.08227336e+00
-1.17062718e-01 1.68181822e-01 -9.39857215e-02 -1.61896139e-01
2.52364337e-01 3.29849198e-02 4.23491776e-01 -1.11780798e+00
-5.25797941e-02 -9.21587825e-01 -4.45786178e-01 -1.06553996e+00
-5.31812906e-02 -8.90591741e-01 -1.59188181e-01 -1.61350584e+00
-8.00737813e-02 -3.63938123e-01 -7.92435646e-01 5.45150161e-01
-7.93951303e-02 3.02019387e-01 2.30875254e-01 2.20056742e-01
-6.33704305e-01 9.14663076e-01 1.05603242e+00 -3.11773539e-01
-2.07803562e-01 -4.27741945e-01 -6.29470944e-01 4.94624615e-01
8.10588062e-01 -2.86095619e-01 -1.58128873e-01 -6.28391981e-01
9.52219442e-02 -6.80011392e-01 7.81024992e-02 -1.41603923e+00
1.20530710e-01 4.20704968e-02 1.73628613e-01 -6.20914340e-01
3.37727457e-01 -7.51930058e-01 1.11104988e-01 7.99561441e-01
-1.31520122e-01 2.78285623e-01 3.24421346e-01 5.20288467e-01
-3.15524906e-01 1.16366521e-01 8.30243289e-01 -1.16860131e-02
-1.09015584e+00 4.13154364e-01 -3.95411164e-01 -6.77578151e-02
1.16740918e+00 -3.38046774e-02 -4.21362102e-01 -3.15060280e-02
-4.11104083e-01 2.91427135e-01 6.21413171e-01 9.48559165e-01
2.68948942e-01 -1.79950869e+00 -5.42269588e-01 3.70036870e-01
4.57798541e-01 -1.50412023e-01 6.29897833e-01 9.27814543e-01
-5.36429524e-01 5.36907554e-01 -3.03784668e-01 -4.57231164e-01
-8.12228680e-01 6.64431512e-01 3.45816523e-01 -3.87328684e-01
-6.60079718e-01 8.55963469e-01 -1.67881101e-01 -7.90022492e-01
2.49555200e-01 -7.15510011e-01 1.11477830e-01 1.85482487e-01
7.11460650e-01 7.49866784e-01 2.94792920e-01 -5.81942201e-01
-6.64768517e-01 6.38414025e-01 1.68011747e-02 7.12315291e-02
1.47878885e+00 -2.82836974e-01 -3.67183648e-02 4.71974254e-01
1.39947593e+00 -1.27702370e-01 -1.20417988e+00 -5.74096382e-01
-6.77116290e-02 -3.81207168e-01 7.61812404e-02 -9.30788755e-01
-1.07016659e+00 7.69325674e-01 9.85767961e-01 1.54616699e-01
1.24349701e+00 -9.11754370e-02 1.12387431e+00 2.70972461e-01
2.57761687e-01 -1.40096056e+00 8.28657374e-02 7.01045752e-01
1.02419448e+00 -1.18911517e+00 -2.75320798e-01 1.27809316e-01
-2.87883162e-01 5.75841665e-01 7.60895848e-01 -3.75573754e-01
9.11244929e-01 1.39632583e-01 -3.22385356e-02 -1.87460799e-02
-7.47749627e-01 -2.67616749e-01 2.09154710e-01 3.51126134e-01
2.84793526e-01 9.85252112e-02 -3.94508213e-01 5.31612575e-01
-9.78771597e-02 1.79201961e-02 1.29516721e-01 9.39421058e-01
-5.12360036e-01 -1.10520101e+00 -1.02298535e-01 3.98620397e-01
-2.86333710e-01 -4.95648896e-03 -1.29214451e-01 8.85319769e-01
5.91740489e-01 7.65485466e-01 3.58770072e-01 -6.98361039e-01
5.85980654e-01 -1.40840178e-02 1.44491404e-01 -3.35538954e-01
-6.53622627e-01 -1.88126504e-01 -1.93404347e-01 -1.00316274e+00
-3.49172235e-01 -8.66791964e-01 -1.15839720e+00 -5.86832583e-01
1.25131141e-02 -2.66203225e-01 6.09169304e-01 6.96576238e-01
7.41574287e-01 7.52079308e-01 5.73278189e-01 -9.71097827e-01
-2.45800570e-01 -1.36441839e+00 -5.38787603e-01 4.14725006e-01
3.17558348e-01 -7.87680686e-01 -3.98118854e-01 -1.80487067e-01] | [9.611506462097168, -0.9982553124427795] |
2574d070-f656-479c-ab38-c10d86897cb9 | video-frame-interpolation-with-densely | 2304.13596 | null | https://arxiv.org/abs/2304.13596v1 | https://arxiv.org/pdf/2304.13596v1.pdf | Video Frame Interpolation with Densely Queried Bilateral Correlation | Video Frame Interpolation (VFI) aims to synthesize non-existent intermediate frames between existent frames. Flow-based VFI algorithms estimate intermediate motion fields to warp the existent frames. Real-world motions' complexity and the reference frame's absence make motion estimation challenging. Many state-of-the-art approaches explicitly model the correlations between two neighboring frames for more accurate motion estimation. In common approaches, the receptive field of correlation modeling at higher resolution depends on the motion fields estimated beforehand. Such receptive field dependency makes common motion estimation approaches poor at coping with small and fast-moving objects. To better model correlations and to produce more accurate motion fields, we propose the Densely Queried Bilateral Correlation (DQBC) that gets rid of the receptive field dependency problem and thus is more friendly to small and fast-moving objects. The motion fields generated with the help of DQBC are further refined and up-sampled with context features. After the motion fields are fixed, a CNN-based SynthNet synthesizes the final interpolated frame. Experiments show that our approach enjoys higher accuracy and less inference time than the state-of-the-art. Source code is available at https://github.com/kinoud/DQBC. | ['Gangshan Wu', 'Jie Tang', 'Jie Liu', 'Chang Zhou'] | 2023-04-26 | null | null | null | null | ['video-frame-interpolation', 'motion-estimation'] | ['computer-vision', 'computer-vision'] | [-2.54993320e-01 -4.78251427e-01 -3.26741725e-01 -1.61420688e-01
-3.62352222e-01 -2.10194200e-01 5.06370902e-01 -4.65234160e-01
-3.31382900e-01 8.33261490e-01 4.54120219e-01 1.35324016e-01
2.29109943e-01 -8.16354573e-01 -7.48306274e-01 -6.51881576e-01
-1.60659283e-01 -1.57282203e-01 7.15478182e-01 -1.96426675e-01
2.45870456e-01 3.66518557e-01 -1.48917043e+00 5.92495859e-01
6.53329790e-01 8.42971861e-01 6.27324343e-01 8.98891926e-01
3.30457836e-02 1.16339707e+00 -4.39858645e-01 -5.37744910e-02
3.41910630e-01 -6.52931988e-01 -8.33867311e-01 -9.87184569e-02
5.07460177e-01 -6.62641823e-01 -5.56355417e-01 8.64441097e-01
2.97246307e-01 5.16939759e-01 3.26406986e-01 -1.12870347e+00
-6.93621218e-01 4.06152636e-01 -5.99545717e-01 7.12495685e-01
3.79865110e-01 3.90717745e-01 7.05176711e-01 -1.16968179e+00
9.57379341e-01 1.49952054e+00 4.87512201e-01 6.70894384e-01
-1.04059994e+00 -5.75678885e-01 3.19726706e-01 5.36341250e-01
-1.51297295e+00 -4.77741122e-01 7.61354089e-01 -3.65173310e-01
7.73127437e-01 2.29193106e-01 7.91471422e-01 1.00121999e+00
2.99014330e-01 6.61911130e-01 4.90019858e-01 -1.57247722e-01
9.93652642e-02 -3.63154560e-01 -4.13823962e-01 4.78543222e-01
-6.54570162e-02 2.97764510e-01 -7.47712791e-01 2.49497786e-01
1.38584745e+00 9.81751829e-03 -6.51833713e-01 -1.95866753e-03
-1.80991292e+00 5.60858846e-01 5.58240175e-01 4.96267587e-01
-5.44844687e-01 3.31053734e-01 1.62272438e-01 -3.06353886e-02
3.47786188e-01 1.08688213e-01 -2.45405301e-01 -7.09457695e-02
-1.19882262e+00 4.44628954e-01 4.09775615e-01 1.02225637e+00
1.04634440e+00 2.14673966e-01 -4.45893168e-01 6.28738284e-01
2.26072177e-01 1.70442179e-01 3.58384818e-01 -1.48957062e+00
4.80365783e-01 6.78462461e-02 3.13939661e-01 -1.22630084e+00
-1.23352006e-01 -2.15204850e-01 -1.03100002e+00 2.74013132e-01
7.24396765e-01 6.52052611e-02 -6.36715472e-01 1.61064625e+00
4.37900394e-01 8.58849764e-01 -8.17687511e-02 1.27064204e+00
7.74538875e-01 9.45374727e-01 5.32090478e-02 -1.72200546e-01
1.06570339e+00 -1.17183053e+00 -7.85033643e-01 -9.93475392e-02
3.98327738e-01 -8.61278892e-01 8.46805811e-01 1.95637688e-01
-1.27089632e+00 -1.02076912e+00 -8.09611320e-01 -3.34236890e-01
1.48222774e-01 1.65315960e-02 4.56535012e-01 -2.54105814e-02
-1.16385186e+00 7.12244213e-01 -8.94136310e-01 5.17113619e-02
4.89149868e-01 2.10464269e-01 -2.34933257e-01 -1.30387172e-01
-1.25304079e+00 7.40142047e-01 3.49932671e-01 2.35244140e-01
-8.29281509e-01 -8.28089774e-01 -8.56142879e-01 -1.10844783e-01
2.15804160e-01 -1.03836501e+00 1.08255553e+00 -1.25916314e+00
-1.47300637e+00 2.34338492e-01 -5.54620862e-01 -3.55506927e-01
7.26455867e-01 -2.04667896e-01 -2.98225582e-01 3.43600869e-01
1.62491933e-01 1.30157816e+00 1.03082895e+00 -9.55645025e-01
-9.08071637e-01 2.05216363e-01 1.45181701e-01 3.24772328e-01
1.07533380e-01 -7.74318054e-02 -6.07411325e-01 -9.14451838e-01
-1.12729579e-01 -8.14293385e-01 -2.28677824e-01 4.25187200e-01
-3.12622599e-02 -9.51468796e-02 9.71583366e-01 -6.45313203e-01
1.41866791e+00 -2.19914985e+00 2.81286389e-01 -2.39915088e-01
1.42090142e-01 3.76840591e-01 -2.34678194e-01 3.90001270e-03
4.69847396e-03 -1.21634871e-01 -3.30632529e-03 -1.60262451e-01
-6.14410400e-01 2.55431235e-01 -7.49087185e-02 4.10305202e-01
3.66360754e-01 8.30877304e-01 -1.25654578e+00 -6.31451964e-01
6.12310052e-01 9.17957246e-01 -9.00495112e-01 1.90626562e-01
-9.77225229e-02 1.15070605e+00 -3.20578724e-01 4.08211380e-01
7.64270961e-01 -3.04889947e-01 -7.43417889e-02 -4.88035232e-01
-3.49998862e-01 1.62096858e-01 -1.35555518e+00 1.99055028e+00
-4.22464341e-01 8.87911439e-01 -2.34386697e-01 -6.59322917e-01
7.73835540e-01 2.23913416e-01 5.42756796e-01 -4.05464202e-01
-1.26290079e-02 1.12730309e-01 8.82400572e-02 -5.08256078e-01
6.80789828e-01 2.53797024e-01 4.75258440e-01 -1.64853007e-01
-1.06997505e-01 3.28737907e-02 3.66909474e-01 7.22800493e-02
6.63146257e-01 6.11746609e-01 2.72976279e-01 -2.90690899e-01
8.92687023e-01 -2.11626858e-01 1.00072920e+00 4.98910397e-01
-2.67344385e-01 9.52555478e-01 2.54614055e-02 -8.49213421e-01
-1.07090914e+00 -9.67413247e-01 8.59218165e-02 7.87977993e-01
5.40643811e-01 -3.76345068e-01 -6.87657118e-01 -2.46162415e-01
-3.61331224e-01 4.90676701e-01 -4.97058094e-01 1.36814550e-01
-9.50524330e-01 -2.58183211e-01 3.46421301e-02 6.69115245e-01
7.73435473e-01 -1.06934059e+00 -8.23764741e-01 5.25733471e-01
-8.44827890e-01 -1.26115978e+00 -8.23323667e-01 -5.71116507e-01
-9.57235277e-01 -7.50436008e-01 -1.07348788e+00 -8.50093901e-01
5.88769078e-01 5.44794023e-01 1.10324466e+00 2.59597361e-01
-5.22865972e-04 -2.04787910e-01 -3.30909520e-01 1.08499691e-01
-3.99756312e-01 -1.92978054e-01 -1.97224081e-01 1.36954159e-01
9.89203528e-02 -5.58465123e-01 -1.08715165e+00 5.00109196e-01
-9.27917778e-01 5.02034009e-01 1.75303191e-01 8.58023107e-01
6.35724962e-01 -1.46054536e-01 1.95991129e-01 -3.30943942e-01
-2.84360703e-02 -3.63348097e-01 -5.69609642e-01 -7.37177720e-03
1.07023336e-01 1.22917697e-01 6.59038067e-01 -7.57501602e-01
-1.13731790e+00 -2.26932354e-02 -5.65104783e-02 -8.01629245e-01
-4.55464385e-02 4.40706909e-02 5.47173396e-02 9.61019620e-02
6.27192140e-01 1.17229577e-03 -2.67856538e-01 -1.20486416e-01
3.37406874e-01 1.86169013e-01 7.95998156e-01 -3.66604626e-01
5.76869249e-01 8.63734365e-01 -2.98560988e-02 -7.95740366e-01
-5.52079618e-01 -3.57163459e-01 -7.93053269e-01 -5.79314351e-01
1.19020212e+00 -1.09244621e+00 -5.02417684e-01 3.94366264e-01
-1.57457781e+00 -5.06009698e-01 -1.86967641e-01 7.91724622e-01
-5.82423687e-01 3.63861352e-01 -6.89816952e-01 -5.56339383e-01
-7.17699081e-02 -1.45823157e+00 1.00561941e+00 2.90454775e-01
-4.08807844e-01 -1.01388240e+00 -1.01412283e-02 3.34568731e-02
3.91551137e-01 2.11522594e-01 3.88361812e-01 3.65928531e-01
-1.20267916e+00 3.39350224e-01 -1.12509094e-01 1.80398867e-01
2.15077847e-01 3.40056628e-01 -8.72959375e-01 -7.71639794e-02
-2.33433172e-01 4.28603321e-01 8.69817734e-01 9.03976440e-01
1.10114229e+00 -3.54379177e-01 -1.36547670e-01 9.59078491e-01
1.29115856e+00 2.95216650e-01 1.02756822e+00 3.82015198e-01
9.40593421e-01 5.60685217e-01 8.21842194e-01 4.75558341e-01
4.75635558e-01 8.03721130e-01 2.70055383e-01 1.24944029e-02
-5.26052773e-01 -2.51588255e-01 3.32048297e-01 7.96264470e-01
-6.71792507e-01 -2.50994503e-01 -5.96553326e-01 6.95530295e-01
-1.94014645e+00 -1.40121830e+00 -2.22423092e-01 2.09021759e+00
7.42469728e-01 -1.13673940e-01 4.85202856e-02 1.59035046e-02
8.52582157e-01 3.32572401e-01 -3.55078667e-01 -4.39274311e-02
-1.89013526e-01 -7.73379579e-02 3.21443975e-01 8.31972182e-01
-1.01008201e+00 9.64527071e-01 5.34989929e+00 7.94105887e-01
-1.28420269e+00 1.91808790e-01 8.51647377e-01 -9.65469256e-02
-3.35196525e-01 1.65009335e-01 -8.45187247e-01 6.89035356e-01
7.09000409e-01 1.53185725e-01 5.40910542e-01 5.16053200e-01
6.13892555e-01 -3.30721170e-01 -1.02858961e+00 1.10085249e+00
-3.10419977e-01 -1.84101880e+00 2.66950540e-02 -1.46248132e-01
9.43460643e-01 -1.26882032e-01 -2.30863839e-02 -7.34342113e-02
-1.88767854e-02 -8.18845272e-01 1.09404588e+00 7.49380708e-01
7.10142255e-01 -6.04089379e-01 5.11749864e-01 2.56987274e-01
-1.62267768e+00 -7.28304172e-03 -5.62612295e-01 -1.71200261e-01
4.41195667e-01 5.76469362e-01 -4.67602581e-01 5.10186791e-01
9.45121706e-01 1.09214818e+00 -2.40030065e-01 8.77330124e-01
-1.82031482e-01 3.01956058e-01 -1.43714145e-01 2.94102460e-01
2.62756348e-01 -2.25380361e-01 6.06826186e-01 1.23010957e+00
6.39002562e-01 1.13799423e-01 3.89282517e-02 9.53413188e-01
2.61934489e-01 -4.10703756e-02 -4.84802961e-01 4.61667299e-01
5.09176672e-01 9.59123552e-01 -7.78640091e-01 -6.92335725e-01
-6.28253639e-01 1.06330395e+00 7.25930557e-02 6.74342632e-01
-1.00362110e+00 1.01813108e-01 9.21475053e-01 2.17658818e-01
4.93769497e-01 -3.87882411e-01 -2.94176582e-02 -1.48221600e+00
-6.44361600e-02 -6.26368046e-01 2.15378731e-01 -9.03565526e-01
-9.21262145e-01 7.38507092e-01 1.15108550e-01 -1.70961726e+00
-6.00445747e-01 -2.36621439e-01 -6.04521871e-01 8.17933738e-01
-1.50958276e+00 -8.74186397e-01 -5.70592880e-01 8.03488851e-01
1.00680661e+00 1.23685636e-01 4.07608032e-01 4.51880693e-01
-3.17914009e-01 2.15445787e-01 -2.50631958e-01 2.12321714e-01
7.24180639e-01 -6.33363485e-01 5.43130636e-01 1.13721085e+00
3.98669913e-02 4.25953656e-01 5.90990067e-01 -6.15714252e-01
-9.92546439e-01 -1.25201321e+00 9.20513928e-01 -3.16864371e-01
3.30204397e-01 -1.44023797e-03 -1.20088565e+00 4.43225473e-01
1.77609071e-01 5.78617573e-01 9.64796096e-02 -7.10218251e-01
-1.60020366e-01 -1.80163577e-01 -7.89128780e-01 7.96577752e-01
1.16135561e+00 -3.92617494e-01 -2.14428768e-01 -9.59791988e-02
5.14768362e-01 -4.81862009e-01 -7.50469804e-01 2.47349963e-01
6.36868358e-01 -1.30432498e+00 1.34542811e+00 -1.48291379e-01
7.43561327e-01 -7.32044637e-01 -6.85532838e-02 -1.04733157e+00
-5.00682175e-01 -8.21256161e-01 -1.74488440e-01 8.97705317e-01
2.37333079e-04 -3.00386548e-01 5.58656037e-01 4.70488966e-01
2.99422890e-02 -5.57633102e-01 -8.24507117e-01 -6.91300809e-01
-9.95665789e-02 -2.86424220e-01 6.45641565e-01 1.08661079e+00
-2.71526635e-01 1.59788579e-02 -5.86193442e-01 1.39181376e-01
4.82474834e-01 1.34610571e-03 7.35181391e-01 -7.91723371e-01
-3.18919599e-01 -3.70539457e-01 -5.25350273e-01 -1.31024075e+00
6.05719350e-02 -6.19902551e-01 2.49580052e-02 -1.35329485e+00
-1.76292092e-01 -2.06439033e-01 -7.96043128e-02 1.74618959e-01
-3.64935666e-01 4.21707749e-01 4.97606635e-01 4.58609909e-01
-3.50268722e-01 4.35890526e-01 1.85265517e+00 -4.54251766e-02
-3.93020451e-01 -1.97474256e-01 -1.17730416e-01 8.50768924e-01
8.07308197e-01 -4.05296177e-01 -5.73737502e-01 -6.84690416e-01
-1.28707677e-01 5.07562220e-01 5.01842976e-01 -1.21805835e+00
3.38477105e-01 -3.80161881e-01 6.70266628e-01 -6.16228759e-01
3.92430097e-01 -5.08307159e-01 6.29135072e-01 4.27562267e-01
-3.10734957e-01 2.85844624e-01 -9.32372082e-03 3.77251506e-01
-4.34737563e-01 -8.73372182e-02 9.32502508e-01 -3.16858768e-01
-1.04852450e+00 5.50013900e-01 -5.35133421e-01 1.91452391e-02
8.59576285e-01 -4.84578520e-01 -5.37252985e-02 -6.28068149e-01
-6.65318370e-01 -1.69845909e-01 4.10415918e-01 4.99685496e-01
8.11645389e-01 -1.35695505e+00 -8.22910428e-01 2.83146173e-01
-3.06990325e-01 3.91597331e-01 4.32877660e-01 8.95459592e-01
-8.03440928e-01 8.59255493e-02 -4.94406611e-01 -7.97209442e-01
-9.83924627e-01 5.91056228e-01 3.35188508e-01 -4.16329056e-02
-7.50922203e-01 8.96784663e-01 7.23822117e-01 3.92226338e-01
-6.06682710e-02 -7.82548487e-01 -1.54417858e-01 -1.36720300e-01
8.19075525e-01 5.28086483e-01 -3.30425173e-01 -8.96561027e-01
-2.65223414e-01 6.79369450e-01 1.58269361e-01 -2.02843755e-01
1.02595210e+00 -3.90826225e-01 8.31262320e-02 2.17639551e-01
1.27759087e+00 -1.72558784e-01 -2.06339812e+00 -8.55488926e-02
-2.53157467e-01 -1.02143729e+00 8.11924634e-04 -1.42563909e-01
-1.37566876e+00 9.23688829e-01 3.91735733e-01 -3.06440055e-01
1.28360426e+00 -3.43697697e-01 8.07183564e-01 -1.13391772e-01
4.99999464e-01 -7.97633886e-01 1.03826888e-01 4.41485018e-01
7.71619976e-01 -1.06243098e+00 -6.18210509e-02 -6.08363271e-01
-7.05330312e-01 1.31608868e+00 6.92065954e-01 -2.88482606e-01
4.82567042e-01 3.42811763e-01 9.40776691e-02 3.70893419e-01
-7.99500108e-01 -7.58871511e-02 4.26796377e-01 7.97679961e-01
4.72425908e-01 -2.56014258e-01 -1.75763384e-01 -3.99135761e-02
-3.49711552e-02 3.71057272e-01 5.08809865e-01 6.22221291e-01
-1.71236247e-01 -9.42209303e-01 -4.51796889e-01 -8.29128772e-02
-4.40187186e-01 -1.61451533e-01 4.00886923e-01 6.93135619e-01
4.05175000e-01 8.20208549e-01 4.07773435e-01 -1.98537216e-01
-1.93380099e-03 -4.39930230e-01 4.85159129e-01 -1.51141748e-01
-3.95386219e-01 3.95782053e-01 -1.17893651e-01 -9.94352818e-01
-9.83164310e-01 -5.30397236e-01 -1.35714078e+00 -4.29532826e-01
-6.11337125e-02 -1.71463966e-01 1.73200727e-01 6.31318808e-01
3.41143906e-01 6.27989769e-01 4.15077507e-01 -1.43266964e+00
3.25865187e-02 -7.40718365e-01 -2.66654402e-01 5.37036121e-01
5.81245482e-01 -6.16747856e-01 -1.50539622e-01 6.22003853e-01] | [10.723173141479492, -1.446092963218689] |
ea1f1623-0f63-4731-9196-7f9c909f008a | an-evaluation-of-deep-cnn-baselines-for-scene | 1805.06086 | null | http://arxiv.org/abs/1805.06086v1 | http://arxiv.org/pdf/1805.06086v1.pdf | An Evaluation of Deep CNN Baselines for Scene-Independent Person Re-Identification | In recent years, a variety of proposed methods based on deep convolutional
neural networks (CNNs) have improved the state of the art for large-scale
person re-identification (ReID). While a large number of optimizations and
network improvements have been proposed, there has been relatively little
evaluation of the influence of training data and baseline network architecture.
In particular, it is usually assumed either that networks are trained on
labeled data from the deployment location (scene-dependent), or else adapted
with unlabeled data, both of which complicate system deployment. In this paper,
we investigate the feasibility of achieving scene-independent person ReID by
forming a large composite dataset for training. We present an in-depth
comparison of several CNN baseline architectures for both scene-dependent and
scene-independent ReID, across a range of training dataset sizes. We show that
scene-independent ReID can produce leading-edge results, competitive with
unsupervised domain adaption techniques. Finally, we introduce a new dataset
for comparing within-camera and across-camera person ReID. | ['Michael Jamieson', 'Paul Marchwica', 'Parthipan Siva'] | 2018-05-16 | null | null | null | null | ['large-scale-person-re-identification'] | ['computer-vision'] | [-1.42574206e-01 -3.13471884e-01 1.31118655e-01 -8.07978630e-01
-3.66165459e-01 -6.47150218e-01 7.32158899e-01 1.40925512e-01
-9.88854408e-01 7.17756033e-01 4.74551141e-01 1.38495073e-01
-4.57688197e-02 -4.84394670e-01 -6.19975507e-01 -3.38068753e-01
-2.76624523e-02 6.26324177e-01 -2.16450930e-01 -1.12867929e-01
2.71834191e-02 7.63741970e-01 -1.52702105e+00 -2.38356546e-01
5.23188293e-01 3.73771876e-01 -1.97628617e-01 8.26040745e-01
1.59354761e-01 4.48606730e-01 -9.79775965e-01 -8.15900385e-01
7.00109959e-01 -1.27425209e-01 -8.77302170e-01 -7.16096833e-02
1.10968578e+00 -5.04999757e-01 -6.71990991e-01 8.59441519e-01
1.02551210e+00 3.91316324e-01 4.70945626e-01 -1.22436357e+00
-9.09356356e-01 4.06512171e-01 -4.48404282e-01 2.88257629e-01
3.47137302e-01 1.23405643e-01 3.60823542e-01 -6.18869543e-01
6.99797392e-01 1.21424556e+00 1.02371216e+00 8.05407107e-01
-1.18317497e+00 -9.82045174e-01 2.03484491e-01 1.46832988e-01
-1.71532428e+00 -6.72564685e-01 6.51947439e-01 -3.29512954e-01
9.49424684e-01 3.94929871e-02 4.60351527e-01 1.31847775e+00
-3.22550029e-01 6.91606402e-01 8.97606254e-01 -5.20919561e-01
5.11560664e-02 3.83702040e-01 3.19591999e-01 9.78414491e-02
5.69083512e-01 3.36772144e-01 -4.60637093e-01 4.73804027e-02
6.82396293e-01 -1.33801937e-01 -4.14260663e-02 -5.06758809e-01
-1.05670786e+00 4.94242191e-01 6.79871798e-01 1.22579813e-01
-2.24607378e-01 -6.00897446e-02 5.45626640e-01 1.70526817e-01
2.82045573e-01 5.80033243e-01 -3.02000225e-01 4.62930650e-02
-1.03840768e+00 5.63001156e-01 8.49030733e-01 1.37020993e+00
7.38321126e-01 6.30239770e-02 -3.26891541e-02 8.82150292e-01
-1.96348891e-01 3.44725072e-01 4.00000274e-01 -6.45121455e-01
3.16083580e-01 5.14110804e-01 3.02016109e-01 -9.45616484e-01
-6.02381945e-01 -5.62249780e-01 -1.00044060e+00 2.27264110e-02
6.13249779e-01 -3.12472075e-01 -1.01061380e+00 1.86457157e+00
2.13735670e-01 2.03588843e-01 2.67082751e-02 1.00082886e+00
9.92769539e-01 -2.40744576e-02 3.92912716e-01 5.03711164e-01
1.13425243e+00 -1.04355574e+00 -3.73078138e-01 -3.82826656e-01
5.43116987e-01 -5.88991165e-01 4.44216013e-01 1.82752833e-02
-8.98016751e-01 -1.00233996e+00 -1.00110817e+00 -2.58600712e-01
-8.00619483e-01 2.59058237e-01 3.38885367e-01 1.09778059e+00
-1.47482872e+00 2.69635946e-01 -3.86553019e-01 -1.15096664e+00
2.49227807e-01 7.46937454e-01 -8.83105397e-01 -2.72020996e-01
-9.23279226e-01 9.76082683e-01 4.62311834e-01 1.77093938e-01
-7.98529029e-01 -5.39564788e-01 -9.23749387e-01 -5.05601652e-02
-8.50267485e-02 -1.00213051e+00 1.19777787e+00 -1.06384492e+00
-1.17527592e+00 1.20019674e+00 -3.00050080e-01 -5.98553061e-01
8.09279501e-01 -3.44717056e-01 -6.72800303e-01 -2.05602184e-01
2.09174559e-01 1.13309884e+00 3.94770503e-01 -1.44368351e+00
-7.04842031e-01 -4.30929363e-01 3.13564360e-01 2.74978995e-01
-4.18968141e-01 3.84759635e-01 -6.81874931e-01 -5.02109468e-01
-5.36013305e-01 -9.81790960e-01 -2.01372758e-01 -3.05417597e-01
-4.86772120e-01 -1.56356409e-01 5.02108574e-01 -6.91522539e-01
7.02978909e-01 -2.06858778e+00 -1.38088271e-01 1.06476672e-01
1.04007021e-01 4.21097606e-01 -4.24195379e-01 3.91582906e-01
-3.83014739e-01 1.30290538e-01 9.16137397e-02 -9.72505510e-01
-7.74294734e-02 -5.84420608e-03 1.66419163e-01 7.59089887e-01
-2.27034781e-02 9.30053413e-01 -7.85430074e-01 -4.12134409e-01
3.78411770e-01 5.61194479e-01 -4.28785384e-01 3.12917322e-01
5.00765443e-01 7.48245835e-01 7.89469853e-03 4.58966464e-01
8.42439950e-01 5.47449514e-02 8.77050497e-03 -1.47172600e-01
-2.38078475e-01 -7.02934340e-02 -1.23805404e+00 1.85451663e+00
-4.72727902e-02 8.66419256e-01 1.17892794e-01 -8.25310588e-01
8.91744733e-01 1.38646439e-01 3.42288673e-01 -7.60514021e-01
1.95195854e-01 -1.72082663e-01 -1.15137190e-01 -1.42721012e-01
1.08610082e+00 2.87925392e-01 -8.57663229e-02 3.13635051e-01
1.77855670e-01 6.38236761e-01 2.85836607e-01 2.14922041e-01
8.86156380e-01 1.42541662e-01 1.01152122e-01 -4.34007078e-01
3.99357021e-01 2.91223735e-01 4.72035676e-01 1.09248698e+00
-5.56278229e-01 9.40815449e-01 -2.15657547e-01 -8.50680828e-01
-1.39973223e+00 -8.85506511e-01 -1.54210597e-01 1.05781758e+00
5.28888702e-01 -2.12131798e-01 -1.02803016e+00 -5.15550077e-01
7.65992478e-02 4.35581475e-01 -8.22692990e-01 1.04304953e-02
-8.76847208e-01 -7.93815553e-01 9.88270283e-01 8.14800322e-01
9.46380317e-01 -6.73533976e-01 -1.78493902e-01 1.36359766e-01
-7.97414854e-02 -1.40385449e+00 -6.44838214e-01 -1.97533801e-01
-4.54689473e-01 -1.19999695e+00 -1.14026141e+00 -8.95449579e-01
8.94992948e-01 6.66512966e-01 1.10272396e+00 8.84261653e-02
-3.22608382e-01 7.75456429e-01 -2.53158987e-01 -4.36626434e-01
-4.58634049e-02 3.87600869e-01 5.32079756e-01 -4.62419912e-02
8.38841259e-01 -2.77830601e-01 -8.03924620e-01 5.12706697e-01
-4.77157384e-01 -1.82157308e-01 5.46484768e-01 5.97430706e-01
8.45465809e-02 -9.72500741e-02 5.17124534e-01 -7.00280368e-01
6.42097890e-01 -2.90129960e-01 -3.99494320e-01 2.74117231e-01
-4.52460498e-01 -1.68439329e-01 5.43004870e-01 -5.05396664e-01
-1.20646977e+00 1.02774538e-01 -6.32077754e-02 -4.23578203e-01
-9.05562639e-01 7.98212439e-02 -1.23985074e-01 -3.40748012e-01
9.04281199e-01 8.83491859e-02 -3.04226667e-01 -5.77793598e-01
4.12531286e-01 7.13782012e-01 1.08419979e+00 -6.70751572e-01
9.47041094e-01 5.10657132e-01 -4.28721815e-01 -7.49325335e-01
-3.76992077e-01 -8.45012128e-01 -1.23907757e+00 -1.65343255e-01
1.00768304e+00 -1.32863533e+00 -5.57932794e-01 7.51538455e-01
-1.06806016e+00 -3.79197210e-01 -1.76653773e-01 4.60151702e-01
-2.30580252e-02 4.05524731e-01 -4.79843289e-01 -5.41901350e-01
-3.75603944e-01 -9.31584239e-01 9.10086036e-01 6.83985889e-01
-3.58705848e-01 -1.01246691e+00 1.75319925e-01 3.40426624e-01
5.35798252e-01 1.25484362e-01 2.45744720e-01 -9.98457372e-01
-4.23949748e-01 -8.47359359e-01 -5.55492759e-01 5.30132977e-03
1.05500862e-01 -3.57773870e-01 -1.30604911e+00 -8.09082627e-01
-7.77558863e-01 -5.17417304e-02 5.10554910e-01 3.42008263e-01
8.79869163e-01 2.80345380e-02 -5.75311005e-01 9.16981101e-01
1.21043587e+00 -9.10516009e-02 6.64866626e-01 8.02852869e-01
9.15751338e-01 6.69501841e-01 1.36754453e-01 1.19581707e-01
8.90994847e-01 7.68854260e-01 -2.59708613e-03 -4.20707554e-01
-3.45143646e-01 -2.82870352e-01 -5.25761284e-02 2.32917711e-01
-4.19209898e-01 -3.98303568e-01 -1.06723809e+00 9.13254499e-01
-1.68754292e+00 -8.65169585e-01 1.12509467e-01 2.38628387e+00
1.92806423e-01 -2.76607424e-01 5.99153936e-01 -3.70359838e-01
1.22900581e+00 -2.23447025e-01 -6.08259022e-01 -5.22539467e-02
-2.99561650e-01 -1.18287280e-01 9.38689768e-01 2.09389880e-01
-1.33108270e+00 9.79550242e-01 6.93337584e+00 2.20857337e-01
-9.38722134e-01 1.14070296e-01 3.43010813e-01 -1.80776045e-01
3.89605463e-01 -2.48196468e-01 -1.13222742e+00 2.65260726e-01
8.47148240e-01 -2.55029738e-01 2.39710137e-01 1.04328215e+00
-1.45934066e-02 2.54417630e-03 -1.25754404e+00 1.31306624e+00
4.03548867e-01 -8.83757234e-01 -1.48730427e-01 2.36913145e-01
8.96659553e-01 1.50670663e-01 -1.50594607e-01 3.84683490e-01
6.61436677e-01 -9.23319280e-01 7.48730719e-01 1.67068496e-01
8.09581518e-01 -8.07467699e-01 1.14946616e+00 5.00174090e-02
-1.28813517e+00 -1.04446501e-01 -4.92116183e-01 3.43147516e-02
1.62742525e-01 2.55128951e-03 -7.33433723e-01 7.74104416e-01
1.11479378e+00 6.53862715e-01 -9.94833767e-01 1.54189408e+00
6.88126590e-03 1.52410984e-01 -4.15996730e-01 3.19907486e-01
1.60540696e-02 2.08865911e-01 4.07830656e-01 1.63986146e+00
5.28098270e-02 1.88247934e-02 1.99052379e-01 5.97870350e-01
-2.52499104e-01 -1.06948353e-01 -5.39786220e-01 4.01763856e-01
6.36056483e-01 1.13817549e+00 -5.77582955e-01 -3.43812674e-01
-4.93012279e-01 1.05123734e+00 4.50940758e-01 7.77794242e-01
-5.31052530e-01 -1.94392711e-01 1.01652086e+00 6.17161058e-02
-5.28626852e-02 -4.12709355e-01 -2.85898447e-01 -1.19360495e+00
-1.08839631e-01 -5.88049531e-01 6.42427742e-01 -5.63453019e-01
-1.56311285e+00 4.99323905e-01 2.70333976e-01 -9.78806674e-01
-2.06096411e-01 -5.13385355e-01 -5.85609376e-01 1.20961988e+00
-1.63584733e+00 -1.48579335e+00 -7.37087190e-01 7.45035768e-01
3.56496781e-01 -3.10997516e-01 8.62866521e-01 6.68514490e-01
-9.07128036e-01 1.24717426e+00 6.21451549e-02 8.15240860e-01
1.23279846e+00 -1.18921137e+00 9.42197978e-01 1.22170782e+00
-2.57990718e-01 9.75712061e-01 5.64652383e-01 -6.88212395e-01
-1.20935094e+00 -1.20078230e+00 7.49298453e-01 -7.72056699e-01
-1.43331224e-02 -4.72979933e-01 -6.70871496e-01 1.02058983e+00
2.22276449e-01 -6.07670918e-02 6.82992280e-01 5.13313353e-01
-5.96929252e-01 -1.53213859e-01 -1.38226974e+00 5.54492414e-01
1.36413646e+00 -5.18241823e-01 -4.03027803e-01 -2.49362341e-03
1.66754276e-01 -4.79904681e-01 -8.26108932e-01 2.66336828e-01
6.57051623e-01 -9.12904322e-01 1.26187050e+00 -7.31583238e-01
-3.57057117e-02 -3.12218517e-01 7.48099536e-02 -1.27656305e+00
-6.79351389e-01 -1.73735097e-01 4.16075319e-01 1.45810258e+00
2.49541011e-02 -6.98404551e-01 9.78763998e-01 1.30836296e+00
-1.68451592e-01 2.02300951e-01 -7.12800324e-01 -9.79832470e-01
2.54956067e-01 -5.31864017e-02 9.77388561e-01 1.15671086e+00
-3.81756544e-01 7.55201504e-02 -5.65007865e-01 4.94711906e-01
8.15994561e-01 -3.30311537e-01 1.56460023e+00 -1.18262303e+00
1.50604680e-01 -5.13364673e-01 -8.53206277e-01 -1.05262959e+00
1.88651741e-01 -6.19132817e-01 -3.61408256e-02 -1.56941628e+00
4.62711811e-01 -5.66630900e-01 -3.17077190e-01 4.84224886e-01
-2.85770833e-01 3.63139033e-01 1.88626468e-01 4.73751724e-01
-5.58751941e-01 1.69648647e-01 6.60934031e-01 -3.29176903e-01
-1.95949703e-01 -3.33363056e-01 -7.63116479e-01 4.40704197e-01
8.83917272e-01 -2.31147364e-01 -1.26961127e-01 -8.63634527e-01
-2.19567090e-01 -6.98296428e-01 6.43751562e-01 -1.37865186e+00
7.24351227e-01 1.94964752e-01 9.68377709e-01 -4.00947630e-01
2.79905200e-01 -7.39277065e-01 3.28103811e-01 -5.86765110e-02
-2.86166042e-01 3.37933809e-01 4.86072809e-01 4.30172414e-01
-7.60960486e-03 1.26233742e-01 8.29729080e-01 -2.71946460e-01
-1.16550338e+00 3.74071658e-01 5.00541627e-02 -1.31596401e-01
8.39453995e-01 -6.10349834e-01 -3.71886760e-01 -3.61616522e-01
-4.23666209e-01 3.22433174e-01 9.28813875e-01 7.49049962e-01
2.32170925e-01 -1.07566237e+00 -8.88412476e-01 2.14517593e-01
4.10381466e-01 -6.07379191e-02 4.28752184e-01 3.37269127e-01
-4.87068981e-01 4.89600718e-01 -3.36755544e-01 -5.81463873e-01
-1.24878526e+00 6.92220151e-01 5.59952796e-01 6.91947192e-02
-6.43819273e-01 9.14481580e-01 2.86985040e-01 -9.05828655e-01
4.49876815e-01 4.76141721e-01 -1.48556530e-01 -2.67501444e-01
7.72188663e-01 4.91471916e-01 -1.94048099e-02 -1.13889825e+00
-3.20921659e-01 6.06809497e-01 -3.27429384e-01 7.50778764e-02
1.07704043e+00 -2.97209382e-01 1.37700021e-01 -2.54936516e-01
1.01675558e+00 -2.70548522e-01 -1.14659607e+00 -2.40040258e-01
-1.32018104e-01 -5.36378086e-01 -2.63007998e-01 -8.72985601e-01
-8.26218843e-01 6.19728088e-01 9.93095100e-01 -1.88777298e-01
8.50789130e-01 -1.79447576e-01 6.54739797e-01 5.81159294e-01
7.07892239e-01 -1.13676751e+00 -3.67190033e-01 2.97204912e-01
5.23914814e-01 -1.47510004e+00 3.76913324e-02 -1.26897216e-01
-3.58954340e-01 8.02622378e-01 9.20037448e-01 -1.27112463e-01
3.51861626e-01 8.53271335e-02 3.73542845e-01 2.42799502e-02
1.52950272e-01 -3.49600554e-01 1.56880096e-01 1.05655134e+00
2.32537240e-01 6.15199693e-02 2.58237779e-01 3.04250836e-01
-5.26169360e-01 1.14090815e-01 2.57524908e-01 8.59533906e-01
1.74402177e-01 -1.19777489e+00 -6.04877591e-01 1.15474053e-01
-2.73576081e-01 1.00594535e-01 -6.75828516e-01 1.14566588e+00
1.99368313e-01 1.10128856e+00 2.34706774e-01 -3.14878047e-01
5.38527191e-01 -4.71323356e-02 5.03553987e-01 -4.93127763e-01
-7.94987082e-01 -4.99966323e-01 2.54610687e-01 -2.07492694e-01
-6.02670908e-01 -6.09088182e-01 -5.45710981e-01 -8.90273690e-01
-3.63812357e-01 -3.76586951e-02 7.21332431e-01 8.56944978e-01
5.87908983e-01 1.50185168e-01 9.09886807e-02 -1.22359002e+00
6.08668923e-02 -1.06185663e+00 -4.05114084e-01 7.40830958e-01
2.22986281e-01 -5.73073506e-01 -1.21663325e-01 1.91897973e-01] | [14.64822769165039, 0.9984248280525208] |
b8ed4673-6ee6-4cde-82a4-b26adc884541 | community-detection-and-portfolio | 2112.13383 | null | https://arxiv.org/abs/2112.13383v1 | https://arxiv.org/pdf/2112.13383v1.pdf | Community detection and portfolio optimization | Community detection methods can be used to explore the structure of complex systems. The well-known modular configurations in complex financial systems indicate the existence of community structures. Here we analyze the community properties of correlation-based networks in worldwide stock markets and use community information to construct portfolios. Portfolios constructed using community detection methods perform well. Our results can be used as new portfolio optimization and risk management tools. | ['Lin Chen', 'H. Eugene Stanley', 'Gang-Jin Wang', 'Chao Wang', 'Longfeng Zhao'] | 2021-12-26 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-7.10476279e-01 -1.86332345e-01 2.41257697e-01 3.74844670e-01
4.13951159e-01 -1.19563520e+00 6.24233186e-01 4.24557418e-01
-4.25269920e-03 5.90745091e-01 1.07755519e-01 -6.46789670e-01
-5.22934556e-01 -1.31437743e+00 7.32626021e-02 -5.11149168e-01
-9.70007479e-01 6.35433435e-01 6.34027779e-01 -6.02583170e-01
5.62146902e-01 5.76742947e-01 -8.82305682e-01 3.33085924e-01
4.78533536e-01 5.40201962e-01 -2.61545151e-01 5.26574075e-01
-2.21950620e-01 6.63145661e-01 -5.95233619e-01 -5.86301029e-01
6.04502738e-01 -2.34950289e-01 -1.53502136e-01 1.30069599e-01
-4.71048176e-01 3.29996675e-01 -1.76669825e-02 1.20581317e+00
2.16430888e-01 -3.71220469e-01 5.64779401e-01 -1.06989002e+00
-1.48983747e-01 9.02382016e-01 -6.08649790e-01 4.98993844e-01
1.33003846e-01 -1.36982307e-01 1.62942719e+00 -9.96260464e-01
8.78693759e-01 1.07958794e+00 8.01477253e-01 -2.21453890e-01
-1.29424429e+00 -8.00461948e-01 -1.49726003e-01 6.89413100e-02
-1.11571503e+00 7.28037506e-02 7.43506074e-01 -1.14838791e+00
7.09130168e-01 3.90670113e-02 1.29484570e+00 3.14109474e-01
5.40063083e-01 3.90961729e-02 1.07077527e+00 -1.80401593e-01
4.67659533e-01 -2.06399634e-02 3.15742671e-01 5.37571967e-01
1.28336763e+00 2.52571285e-01 -2.53588617e-01 -7.07171738e-01
8.25557470e-01 1.54413566e-01 -9.38101783e-02 -4.68634099e-01
-1.13765383e+00 1.26922643e+00 3.83531451e-01 5.48286736e-01
-2.44379073e-01 -1.94250289e-02 3.12755644e-01 6.88869238e-01
3.63779336e-01 7.89018631e-01 8.89632031e-02 3.89760524e-01
-6.36487842e-01 2.03593239e-01 1.22067380e+00 6.31766498e-01
4.23563361e-01 -4.58057038e-02 4.56924438e-01 1.21314771e-01
5.28442860e-01 2.69615233e-01 1.04526736e-01 -9.67554212e-01
5.05671263e-01 9.46095884e-01 -1.20819524e-01 -1.21136701e+00
-3.59362274e-01 -5.79878330e-01 -1.07756376e+00 6.32261395e-01
2.60782093e-01 -5.33803646e-03 2.12112576e-01 5.67431509e-01
-2.19647940e-02 -1.59240037e-01 1.22480197e-02 3.41031373e-01
1.25099301e-01 2.54148781e-01 -5.72678804e-01 -2.03903437e-01
1.07311773e+00 -4.33325559e-01 -3.82536024e-01 4.84747291e-01
1.01382062e-01 -6.56712234e-01 -9.06744823e-02 4.18752611e-01
-8.99611294e-01 1.33520782e-01 -1.18989873e+00 1.26019382e+00
-2.04897895e-01 -6.22997165e-01 6.02115154e-01 8.78798068e-01
-1.11702251e+00 9.12317693e-01 -5.63936293e-01 1.31813526e-01
4.06582832e-01 2.42468104e-01 -1.29530683e-01 1.97438151e-01
-8.06098044e-01 6.52021348e-01 2.72487104e-01 1.63385123e-01
-5.63924193e-01 -3.68724205e-02 -3.42347771e-01 2.83099592e-01
3.62291127e-01 -5.74683011e-01 5.82072973e-01 -7.00791478e-01
-7.18373537e-01 2.30393440e-01 6.59932911e-01 -5.07579744e-01
5.85550725e-01 4.83969778e-01 -1.42297387e-01 3.73358220e-01
-1.41438851e-02 -5.57303786e-01 3.86647105e-01 -6.73366547e-01
-1.33811444e-01 1.16599470e-01 -1.89369395e-01 -3.30964714e-01
-4.12534356e-01 4.48946953e-01 6.13888681e-01 -9.89214540e-01
3.68674397e-01 -7.17565775e-01 -5.11438966e-01 1.37714494e-03
-4.68954057e-01 1.52962863e-01 9.99632105e-02 -3.46776277e-01
1.23410523e+00 -1.61618412e+00 1.39942005e-01 1.09518576e+00
8.11347365e-01 -1.29929796e-01 -6.47565871e-02 1.08409214e+00
-2.07414657e-01 6.64344013e-01 -1.61728248e-01 2.55634576e-01
-6.47214130e-02 -1.11978166e-01 -2.31868833e-01 5.63500941e-01
4.55263704e-01 6.56906903e-01 -6.83516085e-01 -3.18839163e-01
-2.19680354e-01 -4.53771353e-02 -6.35938704e-01 -1.99105948e-01
1.22350112e-01 2.66620591e-02 -4.26804453e-01 5.98995149e-01
4.87828553e-01 -4.86221403e-01 6.99956715e-01 3.62557411e-01
-2.32576922e-01 6.89466819e-02 -1.19074607e+00 2.40928113e-01
5.14226496e-01 8.08180153e-01 1.59192950e-01 -1.07504368e+00
1.02919436e+00 3.43407363e-01 5.61567545e-01 -3.81461754e-02
-1.17197759e-01 2.42743284e-01 8.46087098e-01 1.25685595e-02
9.78790689e-03 -3.21957439e-01 2.12034151e-01 8.81330490e-01
-2.57394344e-01 6.48582866e-03 6.20521426e-01 4.37203139e-01
1.53848386e+00 -8.74405205e-01 5.21399677e-01 -7.66877115e-01
5.21717131e-01 -1.31100819e-01 5.58859527e-01 4.16277826e-01
-5.88446073e-02 3.44909221e-01 1.28153539e+00 -2.84868062e-01
-1.15456915e+00 -1.05788934e+00 2.06358545e-02 2.89375633e-01
-2.50212938e-01 -4.57213432e-01 -4.57368582e-01 -1.22784443e-01
5.94389200e-01 -3.47426593e-01 -6.21153712e-01 2.94211656e-01
-4.94780242e-01 -9.61836338e-01 2.08147168e-01 4.29892153e-01
1.30819291e-01 -1.02628124e+00 -1.39139056e-01 4.66496617e-01
2.59624153e-01 -6.42419398e-01 -2.93872088e-01 -3.80506441e-02
-1.22946596e+00 -1.68166578e+00 -7.27083206e-01 -4.36538517e-01
6.87216759e-01 1.02294534e-01 1.19162333e+00 5.82397401e-01
-4.27096099e-01 1.88502535e-01 -3.15027893e-01 -2.59990066e-01
-7.24394500e-01 -2.21440613e-01 1.12749077e-01 9.07411426e-02
-3.40748996e-01 -8.20431471e-01 -5.35076618e-01 6.81952536e-01
-5.26402354e-01 -5.18187523e-01 3.08748990e-01 7.34801352e-01
2.43861936e-02 4.06332850e-01 6.80739105e-01 -6.31406665e-01
1.26616597e+00 -7.46953130e-01 -1.24760091e+00 4.58076030e-01
-9.90517318e-01 1.93636149e-01 1.41388416e-01 -3.20374340e-01
-3.40642333e-01 -3.89192641e-01 6.66759908e-01 -1.19338214e-01
7.94845343e-01 1.20624900e+00 3.17030132e-01 -4.71114039e-01
3.50069791e-01 -7.12523013e-02 3.50977600e-01 -3.92968178e-01
-8.35474730e-02 2.28555560e-01 -2.08766013e-01 -4.39910203e-01
9.37637329e-01 4.29941863e-01 4.53638494e-01 -7.32698560e-01
5.17805815e-02 -6.43673122e-01 -7.16600180e-01 -4.57596600e-01
2.78167218e-01 -7.24088728e-01 -8.34410548e-01 1.70471862e-01
-1.01118398e+00 4.75753583e-02 6.40090778e-02 1.08200058e-01
-1.81306489e-02 6.00791693e-01 -9.69642282e-01 -1.21476030e+00
-2.61477888e-01 -2.18865559e-01 1.10197492e-01 -1.75978556e-01
-7.13827759e-02 -1.54437637e+00 8.49788010e-01 4.18989807e-02
5.30243993e-01 7.11411893e-01 7.21123397e-01 -7.48202324e-01
-1.02623689e+00 -3.15222055e-01 6.39755046e-03 9.88429040e-02
-9.58792791e-02 5.22428215e-01 -2.76647061e-01 -3.53119463e-01
-5.79305962e-02 6.20319903e-01 9.16857541e-01 1.40150487e-01
1.10570461e-01 -4.75694478e-01 -1.97604522e-01 2.35411543e-02
1.44813693e+00 1.71316206e-01 4.90683049e-01 4.50678915e-01
8.72010812e-02 1.15512109e+00 1.50248662e-01 7.59985149e-01
-2.20444590e-01 2.15730399e-01 8.61447379e-02 4.27266568e-01
4.93657857e-01 1.88464537e-01 3.98580939e-01 1.40842640e+00
-5.04462421e-01 1.39362618e-01 -1.45645869e+00 3.18126351e-01
-2.06116343e+00 -1.56033301e+00 -4.60491240e-01 1.92327225e+00
4.99632865e-01 5.86231947e-01 7.26124227e-01 1.79137677e-01
1.06316805e+00 -2.59328008e-01 -9.93211344e-02 7.68517703e-02
-6.98595941e-01 3.67727101e-01 4.64041352e-01 3.20010215e-01
-7.94245839e-01 2.09702939e-01 7.72831106e+00 1.20500542e-01
-6.73596263e-01 3.75925638e-02 3.28343898e-01 -2.18861639e-01
-3.91867667e-01 1.87656179e-01 -3.16193968e-01 4.18723583e-01
7.25015879e-01 -9.88480985e-01 2.48575836e-01 5.14122963e-01
2.01578792e-02 1.70380436e-02 -6.82761669e-01 5.80634296e-01
-4.09778923e-01 -1.86419141e+00 -3.86042237e-01 7.57854581e-01
8.34164798e-01 7.22301081e-02 -2.27998421e-01 -2.81166136e-01
7.17980087e-01 -8.51231277e-01 6.42560720e-01 7.43420303e-01
8.60143229e-02 -7.95229733e-01 8.42478812e-01 1.74319044e-01
-1.73026299e+00 -4.34026062e-01 -5.39044082e-01 -4.05379683e-01
2.18738407e-01 9.33349609e-01 -7.72623122e-01 5.09652734e-01
5.18937111e-01 1.02364683e+00 -8.16358924e-01 1.67086947e+00
4.32817526e-02 5.52500486e-01 -2.42999554e-01 -4.82397050e-01
-1.95837930e-01 -8.59010160e-01 6.68965995e-01 7.04907537e-01
1.90017745e-01 -4.26487066e-02 1.69933438e-02 1.24207675e+00
2.72123247e-01 2.45431773e-02 -8.23967278e-01 -6.57056093e-01
5.10533512e-01 1.19184124e+00 -1.40071917e+00 -1.44899800e-01
-3.15743566e-01 2.49870107e-01 -1.07071482e-01 6.68451041e-02
2.42583966e-03 -4.52449411e-01 6.18441522e-01 6.57882333e-01
5.92221916e-01 -6.01394176e-01 -4.32132334e-01 -1.37712348e+00
-8.35853517e-02 -6.07352614e-01 2.96225488e-01 -4.09421802e-01
-1.77938461e+00 4.72237498e-01 -2.44286414e-02 -1.43624496e+00
-1.43149853e-01 -8.04286301e-01 -1.36390352e+00 4.27054584e-01
-9.08870280e-01 -1.46671817e-01 7.64353573e-02 3.85808825e-01
-3.27515692e-01 -9.84414518e-01 5.55626810e-01 1.37337342e-01
-5.71937263e-01 -1.99478120e-01 5.86713791e-01 5.61168075e-01
2.84382366e-02 -1.43905365e+00 8.27251792e-01 7.89039731e-01
3.41922402e-01 8.15522373e-01 4.31835115e-01 -1.12339723e+00
-9.15864944e-01 -6.36969745e-01 5.68011940e-01 -3.91409099e-01
1.57993054e+00 -5.75075924e-01 -6.69296145e-01 1.99040547e-01
4.55414295e-01 -1.79087445e-01 7.09286094e-01 3.69505025e-02
-4.90380645e-01 -1.55244237e-02 -8.65522742e-01 3.86520445e-01
7.74670124e-01 -4.04922247e-01 -5.09283960e-01 2.29358092e-01
4.83758599e-01 8.16405892e-01 -1.03138387e+00 7.01995287e-03
5.46922207e-01 -1.23046839e+00 1.01004410e+00 -3.86946619e-01
-3.00288876e-03 -1.52873248e-01 2.48143211e-01 -9.58881319e-01
-5.02204716e-01 -9.15947139e-01 -7.67456740e-03 9.26209092e-01
5.79758942e-01 -1.41981816e+00 7.84557402e-01 1.82926252e-01
8.23089719e-01 -3.25240761e-01 -9.74252045e-01 -1.05328310e+00
5.04699290e-01 4.28638130e-01 3.13008666e-01 1.08270657e+00
5.17406881e-01 3.60420078e-01 4.31766897e-01 -4.12179768e-01
1.12953460e+00 2.60372162e-01 2.96109557e-01 -2.08666849e+00
-4.48980659e-01 -1.14189076e+00 -6.17752433e-01 3.23796153e-01
-1.63692534e-01 -7.14206457e-01 -4.82089520e-01 -1.00959814e+00
3.30215394e-01 -5.98379016e-01 -3.10377568e-01 5.01237549e-02
2.77396381e-01 -1.33992312e-02 4.83408391e-01 6.50753736e-01
-4.32317138e-01 1.75711289e-01 7.81722724e-01 -1.39324889e-01
2.35489428e-01 3.44805896e-01 -5.00178397e-01 3.95672828e-01
1.00520313e+00 -6.03276074e-01 -5.45444563e-02 7.69636154e-01
1.21341074e+00 1.36188984e-01 2.55407989e-01 -8.92264724e-01
3.49504888e-01 -2.41359055e-01 -7.07526058e-02 -4.47432309e-01
-2.72456378e-01 -6.79467440e-01 7.08248258e-01 1.10285759e+00
7.27546513e-02 6.68949842e-01 -2.30201378e-01 9.01386023e-01
-5.25213003e-01 -5.74882627e-01 3.85517448e-01 -4.17966902e-01
1.57103121e-01 2.38400828e-02 -8.24716330e-01 -4.72584665e-02
1.22052729e+00 -4.83228303e-02 -5.19308627e-01 -5.35802066e-01
-8.12744141e-01 2.59885132e-01 4.86112654e-01 7.78952986e-02
6.96448028e-01 -1.18680227e+00 -1.03788960e+00 -7.94443190e-02
-7.29728267e-02 -6.26136601e-01 -6.94811583e-01 7.67160416e-01
-1.34581065e+00 2.04547152e-01 -5.13150632e-01 -4.30195332e-01
-1.37281871e+00 4.53896195e-01 6.52634323e-01 -6.43549383e-01
-2.72671372e-01 4.10710633e-01 6.81194961e-02 -1.07676564e-02
-1.60402894e-01 -9.14847851e-02 -5.42109489e-01 6.35197818e-01
4.01424557e-01 3.51168782e-01 -3.88745457e-01 -1.07638240e-01
-4.60765600e-01 4.79922026e-01 3.35021764e-01 -5.73797643e-01
1.73175418e+00 8.11035633e-02 -9.63740885e-01 4.46151555e-01
8.13368499e-01 8.85148346e-02 -5.89166641e-01 6.96807429e-02
1.09635031e+00 -2.28835285e-01 -5.46979010e-01 -1.18581988e-01
-1.30325747e+00 8.51027071e-01 9.34562832e-02 1.00612283e+00
5.73604524e-01 -4.01132815e-02 -1.18969902e-01 5.66940963e-01
6.27003908e-01 -7.98539162e-01 5.14999509e-01 4.33551639e-01
1.04359627e+00 -6.34501517e-01 4.54357386e-01 -5.11439562e-01
-4.00238991e-01 1.51748157e+00 -6.49279356e-02 -8.24631214e-01
1.52287161e+00 4.87061322e-01 -4.49944228e-01 -5.86451173e-01
-1.00141060e+00 -5.39939143e-02 3.31161946e-01 4.77559447e-01
1.59496382e-01 2.16148183e-01 -3.02367449e-01 6.03976667e-01
-2.01935962e-01 -7.04934597e-01 1.20217240e+00 9.46410358e-01
-8.22580934e-01 -1.32432199e+00 -6.03669286e-01 4.03716654e-01
-4.08930480e-01 -4.37069476e-01 -1.07674241e+00 5.57080686e-01
-4.31956619e-01 7.69757569e-01 1.86024189e-01 -3.75948638e-01
5.26036993e-02 -5.34608327e-02 2.14005679e-01 -7.80752540e-01
-1.19489360e+00 3.82786632e-01 1.41411349e-01 -2.37839013e-01
-4.23262835e-01 -1.06246674e+00 -6.55705214e-01 -7.00071394e-01
-7.46907353e-01 3.60093236e-01 2.45617956e-01 2.84981191e-01
2.56552935e-01 2.31499374e-01 8.50834787e-01 -4.27488387e-01
-5.31351149e-01 -8.50740075e-01 -1.12133729e+00 -6.99267387e-02
9.75049511e-02 -6.19700491e-01 -7.81246483e-01 -1.27072006e-01] | [6.798125267028809, 5.198350429534912] |
3317d9fa-7884-4344-af7d-964fbb6dc9f7 | instance-wise-depth-and-motion-learning-from | 1912.09351 | null | https://arxiv.org/abs/1912.09351v2 | https://arxiv.org/pdf/1912.09351v2.pdf | Instance-wise Depth and Motion Learning from Monocular Videos | We present an end-to-end joint training framework that explicitly models 6-DoF motion of multiple dynamic objects, ego-motion and depth in a monocular camera setup without supervision. Our technical contributions are three-fold. First, we propose a differentiable forward rigid projection module that plays a key role in our instance-wise depth and motion learning. Second, we design an instance-wise photometric and geometric consistency loss that effectively decomposes background and moving object regions. Lastly, we introduce a new auto-annotation scheme to produce video instance segmentation maps that will be utilized as input to our training pipeline. These proposed elements are validated in a detailed ablation study. Through extensive experiments conducted on the KITTI dataset, our framework is shown to outperform the state-of-the-art depth and motion estimation methods. Our code and dataset will be available at https://github.com/SeokjuLee/Insta-DM. | ['Seokju Lee', 'Stephen Lin', 'In So Kweon', 'Sunghoon Im'] | 2019-12-19 | null | null | null | null | ['video-instance-segmentation'] | ['computer-vision'] | [-1.87664516e-02 -1.75704956e-01 -2.52584875e-01 -4.61340934e-01
-8.58134627e-01 -7.38990963e-01 5.87646008e-01 -7.23203242e-01
-2.96700537e-01 4.62289214e-01 -1.00640748e-02 -6.40746504e-02
2.97327101e-01 -3.52130830e-01 -9.07634795e-01 -5.82457364e-01
1.78170025e-01 3.39969486e-01 4.58091617e-01 3.73014212e-01
1.85725793e-01 3.99219185e-01 -1.27610981e+00 5.99467754e-02
6.22633338e-01 8.76873374e-01 3.21568072e-01 9.52136934e-01
2.66331047e-01 1.03473043e+00 7.46556893e-02 -2.68224865e-01
6.73932433e-01 -1.70700133e-01 -9.90394771e-01 5.51087618e-01
9.84157801e-01 -1.03383696e+00 -7.36172676e-01 8.10428143e-01
3.99578780e-01 2.93448329e-01 3.93227547e-01 -9.99627650e-01
-1.35991216e-01 2.24029049e-02 -7.66073167e-01 3.42018604e-01
2.28002146e-01 3.71480495e-01 6.66639090e-01 -1.05494297e+00
9.43243206e-01 1.09593403e+00 4.62783426e-01 7.00156748e-01
-1.08613217e+00 -3.97704840e-01 5.60374320e-01 1.31399393e-01
-1.09867823e+00 -5.40421307e-01 9.64469373e-01 -5.98150432e-01
8.43036413e-01 -5.01453206e-02 6.92656517e-01 1.06167078e+00
7.94220790e-02 1.37517703e+00 7.73272634e-01 -2.21436426e-01
-9.16363206e-03 -1.53517798e-01 -8.97151977e-02 9.38741624e-01
4.53318283e-02 7.48080239e-02 -4.40032572e-01 1.93776518e-01
1.21132660e+00 -2.81391777e-02 -4.18368667e-01 -1.01703238e+00
-1.14084518e+00 5.31698465e-01 1.55962676e-01 -7.33678639e-02
-2.82730430e-01 6.09088659e-01 6.15406446e-02 -2.23670080e-01
7.84643590e-01 -1.98288724e-01 -6.08200550e-01 -3.13323021e-01
-1.02895570e+00 3.08517456e-01 5.29387534e-01 1.11252499e+00
7.74586916e-01 -1.21342525e-01 1.10213950e-01 6.47378623e-01
5.36387205e-01 4.50832993e-01 1.44986719e-01 -1.57912552e+00
5.49833477e-01 2.92825669e-01 1.84206501e-01 -6.76572323e-01
-2.38404125e-01 -2.79980212e-01 -2.51341164e-01 2.17133254e-01
3.84800285e-01 -1.23831533e-01 -9.78975773e-01 1.61037171e+00
8.69308472e-01 7.41967976e-01 -2.42018208e-01 1.13967693e+00
8.29544723e-01 3.59147668e-01 -1.00313514e-01 1.47605598e-01
1.05576813e+00 -1.58944857e+00 -4.24193770e-01 -4.98403549e-01
4.77478981e-01 -6.68521583e-01 7.34293044e-01 2.49390766e-01
-1.51250744e+00 -4.90505189e-01 -7.87635982e-01 -5.64270377e-01
1.99904755e-01 3.59732181e-01 8.00256610e-01 3.88147712e-01
-1.05538929e+00 5.03813684e-01 -1.41527891e+00 -2.06260264e-01
5.43536186e-01 2.24133909e-01 -3.27996790e-01 -2.03617692e-01
-5.55764496e-01 6.26516044e-01 2.34484375e-01 1.75855845e-01
-1.35319781e+00 -9.15526152e-01 -1.18429959e+00 -4.41879600e-01
3.83203804e-01 -1.27568102e+00 1.58677757e+00 -7.20569611e-01
-1.61305130e+00 1.29007804e+00 -4.83617604e-01 -3.04383069e-01
1.05053580e+00 -5.97521305e-01 3.42039406e-01 5.63329160e-01
1.76503092e-01 1.02665555e+00 7.12865591e-01 -1.33182311e+00
-8.47410500e-01 -3.95257592e-01 1.47855133e-01 5.38130224e-01
9.27299708e-02 -8.69427919e-02 -1.28798938e+00 -5.94629228e-01
2.38343403e-01 -1.07409275e+00 -3.18477899e-01 1.89184383e-01
-4.00384188e-01 2.10263163e-01 8.02406430e-01 -7.69813716e-01
8.67649794e-01 -1.89459133e+00 4.82034177e-01 -1.93395808e-01
2.89336115e-01 6.31453246e-02 1.14133343e-01 -9.59361047e-02
1.08557872e-01 -3.03047419e-01 -4.55099225e-01 -1.05786633e+00
-9.98506621e-02 -2.12366786e-03 -1.36958957e-01 6.85900450e-01
2.01741964e-01 1.08581555e+00 -8.92423689e-01 -3.96281213e-01
6.83202267e-01 6.21757150e-01 -5.41601717e-01 2.77713269e-01
-2.18023643e-01 8.53625238e-01 -4.00510311e-01 9.59567428e-01
8.13345075e-01 -3.14737320e-01 -8.05248544e-02 -1.84002042e-01
-9.96956378e-02 2.59397864e-01 -1.10882795e+00 2.27784324e+00
-2.77581275e-01 5.94306946e-01 3.01741928e-01 -5.33350289e-01
2.42094353e-01 6.12618849e-02 7.00853765e-01 -3.19029093e-01
1.93648949e-01 3.59632671e-02 -4.86604631e-01 -4.58154976e-01
4.82772261e-01 2.00075924e-01 2.95456171e-01 2.00398907e-01
1.03660986e-01 -2.47008875e-01 1.69379428e-01 2.92424619e-01
9.68099952e-01 1.00399780e+00 -9.36783105e-02 -1.04960212e-02
6.30467355e-01 3.11195981e-02 8.60825956e-01 3.05673063e-01
-4.77631569e-01 1.15421712e+00 2.20879719e-01 -4.20559078e-01
-1.03314495e+00 -1.04607069e+00 -1.45965278e-01 6.45064116e-01
4.68661964e-01 -1.33704275e-01 -6.59850597e-01 -6.99257910e-01
-5.38652986e-02 3.46674979e-01 -5.36119759e-01 3.14784855e-01
-6.96964502e-01 -5.51418900e-01 1.38633758e-01 8.31715226e-01
5.61199427e-01 -6.98670626e-01 -7.82323062e-01 7.02892095e-02
-3.36850315e-01 -1.67474890e+00 -6.80741608e-01 -1.09367318e-01
-1.15938115e+00 -1.08184481e+00 -9.14056420e-01 -6.68777704e-01
6.82102382e-01 6.49518609e-01 1.13394868e+00 -1.65750638e-01
-4.21082169e-01 9.55134988e-01 -1.02273509e-01 -5.97956106e-02
3.30134630e-02 7.70911500e-02 -9.75218341e-02 -1.40216976e-01
1.57356277e-01 -6.76828265e-01 -1.16724062e+00 2.82028526e-01
-7.19320476e-01 4.80461478e-01 3.68462890e-01 3.82733881e-01
9.19792712e-01 -4.45161432e-01 -2.34122351e-01 -7.95643449e-01
-3.12339455e-01 -3.99747610e-01 -9.26442146e-01 -8.02549124e-02
-1.66450322e-01 -2.33016163e-01 -7.88157135e-02 -2.19511420e-01
-1.19502783e+00 6.71803236e-01 -3.56851965e-02 -8.99892032e-01
-2.29116231e-01 -5.37552033e-03 -2.65888005e-01 -2.51475155e-01
1.74125865e-01 2.29729518e-01 -1.62987158e-01 -4.27401394e-01
4.83219415e-01 1.84349880e-01 7.83346534e-01 -6.15406752e-01
9.15708840e-01 1.13655663e+00 -3.58115546e-02 -6.83813155e-01
-7.96550333e-01 -8.85340035e-01 -1.15588820e+00 -2.40565538e-01
1.14920413e+00 -1.43560243e+00 -3.12950224e-01 7.73773432e-01
-1.15128124e+00 -9.02723491e-01 8.73176008e-02 5.11888206e-01
-8.51700068e-01 7.20938981e-01 -8.60173643e-01 -7.57468402e-01
-3.82454246e-01 -1.09807670e+00 1.42768228e+00 1.97217137e-01
3.83695103e-02 -1.23264956e+00 1.65970430e-01 8.11500430e-01
-8.00993145e-02 4.07083094e-01 1.39878824e-01 3.28600518e-02
-1.20743096e+00 8.84638056e-02 -1.68075129e-01 3.24752212e-01
1.60853378e-02 1.62682533e-01 -1.11199582e+00 -3.36970747e-01
-3.59993987e-02 -2.15046942e-01 1.11009169e+00 7.63832629e-01
1.03974807e+00 8.94632787e-02 -4.33739394e-01 1.34556305e+00
1.37319911e+00 5.33432104e-02 6.77791178e-01 5.46877265e-01
1.33260691e+00 6.20497704e-01 7.46337116e-01 2.43359566e-01
8.24684978e-01 8.34466994e-01 4.88480628e-01 -5.10439165e-02
-3.27957362e-01 -1.52487114e-01 4.59630132e-01 4.96295154e-01
-6.63160086e-02 3.55328582e-02 -7.68607080e-01 7.51665294e-01
-1.99701965e+00 -9.10921097e-01 -3.69008899e-01 2.01777577e+00
5.38267970e-01 -9.51126665e-02 3.18510711e-01 -3.03431332e-01
3.99795860e-01 1.42912462e-01 -7.31037796e-01 3.07137370e-01
-4.47159521e-02 -1.29911393e-01 6.01624787e-01 8.57458651e-01
-1.54228294e+00 1.32196081e+00 5.55906010e+00 3.57393324e-01
-9.40436006e-01 1.84836686e-01 7.25002110e-01 -4.38771486e-01
-1.74606517e-01 -2.11822422e-04 -9.05820251e-01 3.10134649e-01
5.02477765e-01 1.67004272e-01 2.88057148e-01 8.81470442e-01
4.22255516e-01 -2.88263738e-01 -1.26237273e+00 1.18458486e+00
1.05772018e-01 -1.24053299e+00 -2.92941600e-01 1.87014028e-01
1.07212067e+00 3.56408238e-01 6.13388829e-02 -8.07516873e-02
1.95265666e-01 -6.51835024e-01 9.61678565e-01 3.40016335e-01
5.48636973e-01 -5.74458539e-01 3.77842128e-01 1.60424531e-01
-1.17650330e+00 1.36378706e-01 -2.98710167e-01 -4.99011539e-02
5.48120856e-01 5.59234500e-01 -4.95007217e-01 7.62600660e-01
7.80073583e-01 1.11617076e+00 -4.25393701e-01 1.06572640e+00
-4.61165756e-01 3.81055266e-01 -2.45013818e-01 7.33981311e-01
2.75880516e-01 -3.21262449e-01 6.40006959e-01 1.18306947e+00
1.46127731e-01 3.42110246e-01 5.00928871e-02 9.14732933e-01
-9.69512984e-02 -3.15557241e-01 -4.70702797e-01 2.77282029e-01
2.62794584e-01 1.49249816e+00 -7.02684522e-01 -3.35973799e-01
-4.78395790e-01 1.44514191e+00 3.39548677e-01 4.42757457e-01
-9.88796115e-01 1.59295782e-01 9.55037773e-01 1.94878563e-01
6.31778598e-01 -4.92997259e-01 -3.12280715e-01 -1.59602296e+00
3.42373669e-01 -4.58747596e-01 2.18334258e-01 -7.66812563e-01
-1.05233335e+00 2.31817171e-01 1.21441804e-01 -1.22268820e+00
-3.54853481e-01 -6.51920676e-01 -5.76063693e-01 7.51721263e-01
-1.70095456e+00 -1.33602560e+00 -6.76939249e-01 4.48735833e-01
8.15935373e-01 2.02712715e-01 2.68663675e-01 4.81481582e-01
-7.49460995e-01 3.19941282e-01 -1.69467062e-01 2.85954148e-01
6.02650523e-01 -1.18664503e+00 6.86348796e-01 1.32194102e+00
9.52238292e-02 5.25597215e-01 4.02193785e-01 -6.22600436e-01
-1.47019851e+00 -1.26391006e+00 3.59373242e-01 -1.02316856e+00
4.62910146e-01 -2.99026966e-01 -8.02312076e-01 1.07773423e+00
1.41011886e-02 2.89197505e-01 2.97414571e-01 -4.24174815e-01
-2.13845987e-02 1.96195349e-01 -1.00594854e+00 5.35367787e-01
1.44612551e+00 -3.92807245e-01 -2.82552242e-01 2.08114102e-01
6.98839903e-01 -1.09448695e+00 -6.23964667e-01 3.82663935e-01
6.25081599e-01 -1.07984114e+00 1.18632805e+00 -2.83894718e-01
7.15400636e-01 -6.40404165e-01 -1.19909875e-01 -8.85612547e-01
-2.22634956e-01 -7.62625873e-01 -5.93828619e-01 1.07891524e+00
3.39192785e-02 -2.83492416e-01 1.13205838e+00 8.55290890e-01
-2.16911554e-01 -7.91170478e-01 -8.21427584e-01 -5.55042505e-01
1.78536661e-02 -6.40572965e-01 -8.55173022e-02 9.18569028e-01
-5.31652212e-01 1.00732245e-01 -4.43144947e-01 3.11698675e-01
9.36231017e-01 -2.82817166e-02 1.21167505e+00 -7.93240190e-01
-5.22482634e-01 -3.20270121e-01 -5.15748560e-01 -1.67469072e+00
1.23229064e-01 -6.77549362e-01 -5.73176751e-03 -1.71591747e+00
3.37572902e-01 -1.39684483e-01 3.34606133e-02 1.97555840e-01
-2.99873918e-01 4.46074158e-01 2.08450004e-01 2.68790066e-01
-7.20991552e-01 6.13896608e-01 1.29739165e+00 1.53161287e-01
-3.04313570e-01 5.98503016e-02 -3.94447833e-01 8.60483646e-01
5.45079231e-01 -3.01585644e-01 -5.55358946e-01 -7.92455077e-01
-2.79035568e-01 -4.11616452e-02 7.44857848e-01 -8.90544891e-01
9.51009244e-02 -1.53049320e-01 5.47512889e-01 -8.81864130e-01
5.95717609e-01 -6.56302392e-01 3.74867059e-02 2.20174357e-01
3.79739068e-02 -7.52324192e-03 3.35626125e-01 5.93411088e-01
1.06156766e-01 1.37765119e-02 7.37480521e-01 -1.33081332e-01
-1.03957915e+00 6.86388612e-01 -3.00068036e-02 1.06078491e-01
1.21051574e+00 -3.56706381e-01 -1.10379793e-01 -4.48288679e-01
-5.24616301e-01 5.30640841e-01 8.44475567e-01 4.43423688e-01
6.29771233e-01 -1.08981752e+00 -6.37869537e-01 -1.07523091e-01
-9.11296755e-02 5.47257602e-01 4.88043338e-01 1.03352237e+00
-9.50395703e-01 4.21475947e-01 7.79108983e-03 -9.55524087e-01
-1.40140426e+00 2.22863778e-01 5.22728860e-01 1.68433867e-03
-9.09200788e-01 1.12032366e+00 6.07190430e-01 -4.68837321e-01
4.09585208e-01 -5.22961020e-01 3.60035270e-01 -4.77754354e-01
5.85628986e-01 4.75607693e-01 -1.93150103e-01 -6.67524219e-01
-5.63594639e-01 8.98479044e-01 -9.33053568e-02 -4.77401644e-01
1.33529556e+00 -5.15788794e-01 1.51335895e-01 2.96819925e-01
1.26192665e+00 -9.63509977e-02 -2.20495367e+00 -9.66094285e-02
-3.26817304e-01 -7.65433490e-01 1.30947694e-01 -5.45191646e-01
-1.14797544e+00 9.11795974e-01 5.04968584e-01 -6.18290067e-01
1.05171025e+00 1.51769565e-02 8.26060891e-01 2.82689147e-02
3.25296372e-01 -8.93846452e-01 9.57179815e-02 5.29094040e-01
5.09177566e-01 -1.48769236e+00 7.37820044e-02 -7.21709788e-01
-6.72233224e-01 9.24276233e-01 8.74788463e-01 -1.48085967e-01
3.54355007e-01 2.15534031e-01 2.01996684e-01 -1.02051394e-02
-6.68440878e-01 -2.63717741e-01 4.90469873e-01 6.00815594e-01
4.41571712e-01 -3.69392037e-01 2.89258957e-02 1.70393080e-01
1.51422909e-02 1.13862105e-01 4.69586074e-01 1.00287032e+00
-1.22726597e-01 -1.05840790e+00 -2.22934321e-01 -2.16147840e-01
-5.01934707e-01 2.86540128e-02 -3.03012550e-01 7.64798224e-01
-7.04583973e-02 7.24199712e-01 7.87024274e-02 -2.39418313e-01
1.27256796e-01 -2.94502489e-02 8.86634529e-01 -5.84208846e-01
-2.92483233e-02 3.39628220e-01 -6.64563328e-02 -9.83757913e-01
-5.77491820e-01 -8.68586898e-01 -1.44386160e+00 -1.93212822e-01
2.06899360e-01 -4.49334919e-01 6.35710478e-01 8.28137636e-01
4.68521714e-01 4.81097728e-01 4.15536612e-01 -1.44677925e+00
-9.28025991e-02 -5.94705403e-01 -1.46825954e-01 3.86695117e-01
4.54833269e-01 -6.50633276e-01 -3.46247315e-01 3.45875323e-01] | [8.520462989807129, -1.9590915441513062] |
3ca7cb7b-08d5-471c-a24e-67e56d2dc8a3 | i3cl-intra-and-inter-instance-collaborative | 2108.01343 | null | https://arxiv.org/abs/2108.01343v3 | https://arxiv.org/pdf/2108.01343v3.pdf | I3CL:Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection | Existing methods for arbitrary-shaped text detection in natural scenes face two critical issues, i.e., 1) fracture detections at the gaps in a text instance; and 2) inaccurate detections of arbitrary-shaped text instances with diverse background context. To address these issues, we propose a novel method named Intra- and Inter-Instance Collaborative Learning (I3CL). Specifically, to address the first issue, we design an effective convolutional module with multiple receptive fields, which is able to collaboratively learn better character and gap feature representations at local and long ranges inside a text instance. To address the second issue, we devise an instance-based transformer module to exploit the dependencies between different text instances and a global context module to exploit the semantic context from the shared background, which are able to collaboratively learn more discriminative text feature representation. In this way, I3CL can effectively exploit the intra- and inter-instance dependencies together in a unified end-to-end trainable framework. Besides, to make full use of the unlabeled data, we design an effective semi-supervised learning method to leverage the pseudo labels via an ensemble strategy. Without bells and whistles, experimental results show that the proposed I3CL sets new state-of-the-art results on three challenging public benchmarks, i.e., an F-measure of 77.5% on ICDAR2019-ArT, 86.9% on Total-Text, and 86.4% on CTW-1500. Notably, our I3CL with the ResNeSt-101 backbone ranked 1st place on the ICDAR2019-ArT leaderboard. The source code will be available at https://github.com/ViTAE-Transformer/ViTAE-Transformer-Scene-Text-Detection. | ['Jian Ye', 'Bo Du', 'DaCheng Tao', 'Juhua Liu', 'Jing Zhang'] | 2021-08-03 | null | null | null | null | ['scene-text-detection'] | ['computer-vision'] | [ 1.10618673e-01 -4.17678118e-01 2.84168171e-03 -2.72600502e-01
-1.20630217e+00 -5.72633266e-01 5.27527392e-01 -1.13209315e-01
-2.68419474e-01 2.48268038e-01 8.39692950e-02 -1.19193763e-01
2.29160979e-01 -6.21996701e-01 -6.61636651e-01 -8.90976846e-01
4.55409527e-01 3.97366107e-01 5.15023768e-01 -9.72649232e-02
2.15666056e-01 -3.40814702e-02 -1.38831532e+00 7.12045670e-01
9.47974265e-01 1.07681191e+00 3.88197154e-01 4.26428229e-01
-4.75345761e-01 7.95086026e-01 -5.65193534e-01 -4.73251462e-01
1.83948189e-01 -2.32769996e-01 -2.47758001e-01 3.25191259e-01
6.45449519e-01 -2.92513192e-01 -3.53904873e-01 9.42757368e-01
5.52948415e-01 -2.88178455e-02 5.99434495e-01 -9.34502780e-01
-5.18597424e-01 6.63888097e-01 -1.13375473e+00 2.53290445e-01
-5.41584268e-02 2.70533979e-01 1.18078816e+00 -1.41378307e+00
3.53978038e-01 1.09422410e+00 7.29785740e-01 3.93236458e-01
-8.39311242e-01 -9.27378476e-01 5.55067658e-01 2.00377837e-01
-1.43666184e+00 -3.86633545e-01 7.76230574e-01 -3.87121886e-01
6.50542974e-01 2.75982171e-01 3.12391222e-01 1.27245498e+00
1.72161639e-01 1.35072100e+00 1.00205219e+00 -2.98281640e-01
-1.31313309e-01 1.33935377e-01 1.67012379e-01 7.53113985e-01
1.16062067e-01 -2.38101363e-01 -7.42235243e-01 7.63548836e-02
4.04175043e-01 2.65578836e-01 -1.69910088e-01 -9.12867859e-02
-1.13992167e+00 6.89551771e-01 3.36581022e-01 3.30633581e-01
9.35405940e-02 -8.95333737e-02 5.80558896e-01 -1.41526476e-01
6.84639990e-01 -1.03107534e-01 -4.08038616e-01 1.19880326e-01
-9.11059618e-01 -1.04929306e-01 3.48781437e-01 9.53958273e-01
5.75379431e-01 -1.26887813e-01 -4.62505877e-01 1.15935802e+00
4.79287058e-01 8.09071243e-01 4.03141409e-01 -5.87911643e-02
1.07188201e+00 7.87233710e-01 -1.86223611e-01 -7.40809202e-01
-2.08214328e-01 -7.45997071e-01 -7.75182068e-01 -1.56108022e-01
3.58770937e-01 -1.00323915e-01 -9.87223387e-01 1.34606457e+00
4.10703480e-01 1.73911005e-01 -9.13304389e-02 7.64249146e-01
1.02272534e+00 7.77835369e-01 -9.62228924e-02 1.07875481e-01
1.36338913e+00 -1.27118266e+00 -5.32694638e-01 -5.98106503e-01
6.81081891e-01 -1.14744878e+00 1.00004005e+00 3.19832742e-01
-5.90188324e-01 -5.50303757e-01 -9.43075716e-01 -1.83104929e-02
-2.26734281e-01 7.63550997e-01 9.20415744e-02 3.86383802e-01
-6.35191560e-01 3.64163928e-02 -7.02243745e-01 -2.54205436e-01
5.82296073e-01 3.70079949e-02 -7.40782395e-02 -2.96312451e-01
-7.42536008e-01 3.79986227e-01 2.71645576e-01 3.41727078e-01
-8.08659792e-01 -4.85601693e-01 -6.05910122e-01 -1.56556875e-01
6.78500414e-01 -3.79641175e-01 1.06455195e+00 -9.60485458e-01
-1.15356517e+00 7.84125626e-01 -2.36615255e-01 -1.32998109e-01
8.66242528e-01 -3.22358221e-01 -2.88476378e-01 -4.11088727e-02
3.88956249e-01 3.33852738e-01 9.47056830e-01 -1.25661838e+00
-8.71986151e-01 -5.75972915e-01 -4.89256740e-01 2.74480551e-01
-4.83069271e-01 -8.44416916e-02 -8.59987497e-01 -9.12224293e-01
1.61930233e-01 -8.02829146e-01 1.73892319e-01 3.64134461e-02
-7.03370214e-01 -4.51640189e-01 1.14585006e+00 -6.04072154e-01
1.16461980e+00 -2.27785921e+00 3.99700366e-03 -1.85459584e-01
2.90318727e-01 4.05166507e-01 -1.27375990e-01 4.35938120e-01
2.84672052e-01 8.00826587e-03 -6.45796731e-02 -7.53967106e-01
-7.89874047e-02 -4.17313389e-02 -3.63572299e-01 4.77391958e-01
2.33763665e-01 9.11588430e-01 -7.28841603e-01 -6.92423999e-01
4.23121065e-01 4.56186563e-01 -2.36641198e-01 3.97422165e-02
-2.95548141e-01 3.65638226e-01 -8.00298095e-01 8.00027728e-01
7.67567575e-01 -3.83433074e-01 -9.51121300e-02 -1.92251220e-01
-1.81595549e-01 6.91565126e-02 -1.09870887e+00 1.52524745e+00
-3.79364997e-01 7.78812945e-01 -6.56252354e-03 -8.35683107e-01
1.02520275e+00 3.92316729e-02 1.72263905e-01 -7.91004956e-01
2.32100815e-01 4.06463921e-01 -2.63690889e-01 -4.20451015e-01
3.57595414e-01 1.46803379e-01 4.39120643e-03 1.69869512e-01
-8.68239477e-02 3.56187999e-01 4.36935723e-02 3.63973767e-01
1.06721199e+00 1.43125772e-01 -1.06124334e-01 -1.35361180e-01
6.17250681e-01 -2.24115893e-01 6.45390391e-01 8.67671788e-01
-1.98172048e-01 7.94667602e-01 2.22570598e-01 -5.30354679e-01
-7.75626779e-01 -8.79245877e-01 -2.13120565e-01 1.25955081e+00
2.83959329e-01 -4.95536119e-01 -5.01472056e-01 -9.58918989e-01
-3.94321196e-02 4.75405663e-01 -6.79040551e-01 9.74039584e-02
-4.56341952e-01 -7.55323887e-01 5.66194773e-01 7.27090359e-01
7.47181058e-01 -8.45999181e-01 -1.34010807e-01 2.48966143e-01
-4.87237573e-01 -1.42576098e+00 -8.24773312e-01 2.80000240e-01
-6.26870394e-01 -8.97387207e-01 -7.09183633e-01 -7.36494958e-01
4.59358066e-01 6.85913861e-01 8.51566076e-01 1.35280803e-01
-3.94887656e-01 2.42396399e-01 -6.60869062e-01 -3.76270562e-01
-1.82568625e-01 -6.02243729e-02 -3.92519832e-01 3.92863005e-01
4.47412312e-01 -5.23182079e-02 -6.98452234e-01 7.08883703e-01
-6.80321515e-01 3.02223980e-01 6.84359491e-01 1.03774869e+00
7.61673033e-01 -5.20254485e-02 4.21086609e-01 -6.41847134e-01
-3.01725250e-02 -3.34515214e-01 -6.43525839e-01 3.55697662e-01
-1.89975590e-01 -3.13081175e-01 6.95250452e-01 -3.30726027e-01
-1.05113351e+00 2.08314255e-01 -6.70682937e-02 -5.01280725e-01
5.62231913e-02 2.85926700e-01 -3.50990891e-01 1.39004037e-01
5.73388219e-01 6.65916741e-01 -4.89550173e-01 -4.43673104e-01
8.84818435e-02 1.00898516e+00 3.91163915e-01 -6.14409089e-01
8.32312465e-01 6.34452581e-01 -4.31317210e-01 -9.13108885e-01
-1.20044446e+00 -9.82543528e-01 -4.04573649e-01 -1.69941798e-01
8.87034178e-01 -1.28453481e+00 -1.62797481e-01 1.02055073e+00
-9.91800845e-01 -4.77057010e-01 1.52662113e-01 1.49971426e-01
-1.00853942e-01 5.76369941e-01 -5.31623304e-01 -8.31848621e-01
-6.71470761e-01 -1.18719721e+00 1.65937293e+00 2.21335962e-01
3.50635320e-01 -7.04898000e-01 -2.08801761e-01 8.82316589e-01
1.57705277e-01 -9.42711718e-03 5.05137205e-01 -8.64820182e-01
-6.15238607e-01 -3.11003476e-01 -5.55451751e-01 3.12687248e-01
1.42240927e-01 1.00265346e-01 -1.11227655e+00 -4.00132030e-01
-3.87847662e-01 -3.94183844e-01 1.17644513e+00 1.16027102e-01
1.19714606e+00 4.46359962e-02 -4.74073231e-01 4.84022081e-01
1.30889273e+00 -4.86272834e-02 5.49358428e-01 3.03713858e-01
1.00965083e+00 4.06489313e-01 8.29596519e-01 5.57291567e-01
4.87730175e-01 7.45593250e-01 3.02049547e-01 -7.98010007e-02
-3.58532101e-01 -3.43226433e-01 4.32447374e-01 6.81543767e-01
3.17506075e-01 -5.54624736e-01 -1.04959249e+00 4.87615615e-01
-1.96807539e+00 -7.19304085e-01 -4.64328527e-01 2.05348921e+00
5.37491500e-01 3.22601497e-01 3.97370793e-02 1.14771472e-02
1.09674621e+00 1.74701467e-01 -6.06652021e-01 2.08485410e-01
-2.35781088e-01 -1.35603631e-02 4.36377585e-01 1.12957202e-01
-1.31263924e+00 1.13015354e+00 3.85440040e+00 1.31506526e+00
-1.13491094e+00 2.42757708e-01 7.52966762e-01 -7.97829479e-02
6.44042194e-02 -2.58168399e-01 -1.40006995e+00 6.80875838e-01
3.49636137e-01 2.52403349e-01 5.81932031e-02 7.54539073e-01
1.59250811e-01 3.00240479e-02 -7.88234115e-01 1.09272540e+00
2.76133209e-01 -1.23330033e+00 3.56684551e-02 2.25616433e-02
8.45006764e-01 4.40548182e-01 7.83569291e-02 4.15137947e-01
2.61398461e-02 -6.83793068e-01 8.56518149e-01 1.56899601e-01
7.94350386e-01 -5.89123249e-01 7.30262935e-01 4.29180384e-01
-1.57960629e+00 -6.12719692e-02 -3.23250622e-01 5.36249936e-01
-2.34985232e-01 7.49087691e-01 -5.67207396e-01 5.75429201e-01
8.66521597e-01 9.77639496e-01 -6.67606652e-01 1.25250375e+00
-3.86533886e-01 7.28223085e-01 -4.28789258e-01 -1.96845114e-01
4.28750843e-01 6.12365492e-02 5.80216169e-01 1.44722259e+00
2.23210767e-01 -2.45713815e-01 4.38457936e-01 6.61315560e-01
-2.54664183e-01 3.29803854e-01 -2.96870142e-01 1.49363726e-01
4.23924327e-01 1.44796681e+00 -9.41577137e-01 -2.61260450e-01
-5.30698836e-01 9.74623859e-01 3.28063756e-01 2.29137987e-01
-1.07149994e+00 -3.18005741e-01 2.05980569e-01 -9.58397090e-02
7.15381920e-01 1.62381008e-02 -2.36311615e-01 -1.42616796e+00
4.99153584e-01 -1.05603194e+00 3.98037016e-01 -6.49968386e-01
-1.49378443e+00 5.65773547e-01 -4.66263205e-01 -1.35671246e+00
3.74195099e-01 -6.84549212e-01 -8.49403381e-01 7.64890969e-01
-1.59728801e+00 -1.63590920e+00 -6.17485821e-01 6.91579700e-01
1.04773593e+00 -1.56689808e-01 3.62844318e-01 2.53762364e-01
-1.03754663e+00 9.53535557e-01 4.35514569e-01 5.02170920e-01
1.05655766e+00 -1.10629189e+00 5.00118256e-01 8.75774443e-01
3.83175790e-01 1.25880897e-01 2.14334518e-01 -6.31130338e-01
-1.45625126e+00 -1.36163127e+00 5.29952943e-01 -5.95974982e-01
7.61992514e-01 -7.66274929e-01 -1.03489304e+00 6.47403657e-01
-7.37693012e-02 3.83066505e-01 4.59021449e-01 1.68024719e-01
-6.97388411e-01 -2.14118212e-01 -7.32427359e-01 5.15099883e-01
9.68996823e-01 -4.63011622e-01 -3.78022194e-01 6.31854355e-01
4.15604919e-01 -5.69925010e-01 -4.23272878e-01 3.86515617e-01
4.98391628e-01 -1.00683999e+00 6.97041631e-01 -3.90872695e-02
5.45202672e-01 -4.16352779e-01 -2.73882538e-01 -8.69729936e-01
-1.09096102e-01 -4.06231046e-01 5.48088960e-02 1.45579708e+00
5.01423538e-01 -7.62813866e-01 7.92510927e-01 -7.22691268e-02
-3.57802898e-01 -9.15478587e-01 -1.10518789e+00 -7.27791667e-01
2.84358293e-01 -4.73069936e-01 2.17505977e-01 8.32463324e-01
-3.66601437e-01 3.56812924e-01 -4.80489284e-01 2.78176218e-01
6.70479357e-01 2.08372191e-01 9.68039453e-01 -1.03377545e+00
-4.06994760e-01 -3.78560156e-01 -3.40761721e-01 -1.32813859e+00
-2.98803132e-02 -7.99525976e-01 2.52480567e-01 -1.45626557e+00
5.64565778e-01 -6.04849696e-01 -3.02453488e-01 6.04372501e-01
-5.59340835e-01 2.53857523e-01 2.90411472e-01 3.37750465e-01
-1.12620652e+00 7.31916130e-01 1.11354589e+00 -3.15683752e-01
-1.91926751e-02 -9.92009044e-02 -5.76973677e-01 6.79705262e-01
7.59534776e-01 -5.61021686e-01 -9.80065614e-02 -6.47684216e-01
8.75607952e-02 -2.07151517e-01 4.65493023e-01 -9.75416780e-01
2.75383472e-01 8.30779299e-02 6.22861862e-01 -1.05358219e+00
7.30327144e-02 -6.18292511e-01 -3.90616983e-01 2.11361781e-01
-2.30464675e-02 -3.71457994e-01 3.55693668e-01 8.07197571e-01
-1.27840275e-02 -9.81804132e-02 7.90306985e-01 1.25527382e-01
-7.55277812e-01 2.86085516e-01 -1.70366764e-01 2.99795091e-01
1.01572859e+00 -1.42412469e-01 -6.97157085e-01 -8.78626481e-03
-3.05689216e-01 5.80292225e-01 3.52683514e-01 6.76759124e-01
5.02285659e-01 -9.57635939e-01 -1.06037092e+00 1.04944892e-01
6.79655969e-01 1.65833086e-01 4.97773796e-01 9.89876211e-01
-1.54315561e-01 2.34820783e-01 3.61629635e-01 -1.13716960e+00
-1.34809446e+00 2.30839640e-01 4.73140776e-01 -4.58239555e-01
-8.89786780e-01 9.65199292e-01 6.55896366e-01 -4.39824522e-01
2.50975460e-01 -3.43829133e-02 8.74560401e-02 7.38103688e-02
6.74052835e-01 6.91233873e-02 2.98107088e-01 -6.17981493e-01
-4.65807229e-01 9.04112101e-01 -5.13532817e-01 4.07244772e-01
9.64818656e-01 -1.54847682e-01 3.03843439e-01 3.42253059e-01
1.04982388e+00 2.63942033e-01 -1.40087187e+00 -5.98630905e-01
-2.36763194e-01 -4.96419668e-01 2.23815501e-01 -1.06524897e+00
-1.16151392e+00 9.11584914e-01 6.88377619e-01 -1.30913273e-01
1.04361176e+00 5.07965907e-02 8.69168639e-01 3.28015804e-01
2.34538525e-01 -1.04828310e+00 4.88301367e-01 5.71666062e-01
8.22142899e-01 -1.54264140e+00 -7.36444667e-02 -4.75269049e-01
-7.39291847e-01 1.13936317e+00 7.56974041e-01 1.48338780e-01
2.77345568e-01 3.60014200e-01 2.39700437e-01 -1.07329801e-01
-7.12148905e-01 -2.74887949e-01 3.54028106e-01 1.88003704e-01
5.15657365e-01 7.31281191e-02 8.11546221e-02 5.15248299e-01
2.57307678e-01 -2.81799465e-01 1.61216781e-01 8.32293034e-01
-4.83848661e-01 -9.78003740e-01 -4.73179966e-01 5.42747736e-01
-5.18217921e-01 -3.24796796e-01 -6.02791131e-01 7.48771906e-01
1.41318306e-01 1.09561729e+00 -1.41189307e-01 -4.74298060e-01
3.14672023e-01 1.49488449e-02 2.25508839e-01 -7.02951849e-01
-6.09938741e-01 5.67668259e-01 -5.07739671e-02 -2.76871592e-01
-4.38580252e-02 -7.22434402e-01 -1.18802416e+00 -1.10119164e-01
-7.75001645e-01 -2.06638977e-01 5.54035425e-01 9.97793078e-01
4.11766201e-01 6.09728158e-01 7.61552751e-01 -6.46783173e-01
-5.60156465e-01 -1.01492476e+00 -3.76870126e-01 1.78701952e-01
2.01561540e-01 -4.92007762e-01 -5.44033885e-01 -2.19784930e-01] | [11.996685981750488, 2.245739459991455] |
7e6f0043-9534-408c-89ac-315ccbc08148 | validating-weak-form-market-efficiency-in | 1909.05151 | null | https://arxiv.org/abs/1909.05151v1 | https://arxiv.org/pdf/1909.05151v1.pdf | Validating Weak-form Market Efficiency in United States Stock Markets with Trend Deterministic Price Data and Machine Learning | The Efficient Market Hypothesis has been a staple of economics research for decades. In particular, weak-form market efficiency -- the notion that past prices cannot predict future performance -- is strongly supported by econometric evidence. In contrast, machine learning algorithms implemented to predict stock price have been touted, to varying degrees, as successful. Moreover, some data scientists boast the ability to garner above-market returns using price data alone. This study endeavors to connect existing econometric research on weak-form efficient markets with data science innovations in algorithmic trading. First, a traditional exploration of stationarity in stock index prices over the past decade is conducted with Augmented Dickey-Fuller and Variance Ratio tests. Then, an algorithmic trading platform is implemented with the use of five machine learning algorithms. Econometric findings identify potential stationarity, hinting technical evaluation may be possible, though algorithmic trading results find little predictive power in any machine learning model, even when using trend-specific metrics. Accounting for transaction costs and risk, no system achieved above-market returns consistently. Our findings reinforce the validity of weak-form market efficiency. | ['Jeffrey Gropp', 'Samuel Showalter'] | 2019-09-11 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-6.17080569e-01 -2.06837848e-01 -7.84981191e-01 1.57201234e-02
-5.48342288e-01 -8.59385848e-01 6.98082626e-01 -1.14616016e-02
-4.28935945e-01 5.82073092e-01 2.42520258e-01 -1.12167990e+00
-3.57172340e-01 -7.78497696e-01 -2.64629662e-01 -2.38490924e-01
-1.76842898e-01 4.23413366e-01 -2.53338844e-01 -9.20618996e-02
8.48627150e-01 2.32609421e-01 -1.03962553e+00 -1.81994304e-01
4.90809292e-01 1.32379091e+00 -3.52783740e-01 3.00891548e-01
-1.61813810e-01 1.22538674e+00 -2.00590476e-01 -8.09018254e-01
9.48797226e-01 -3.22268218e-01 -1.64050370e-01 -5.39998591e-01
-2.60516703e-01 -7.77618468e-01 -1.63718194e-01 6.75213039e-01
7.60217533e-02 -3.34106296e-01 5.93536258e-01 -1.44394290e+00
-7.52975166e-01 8.74285758e-01 -5.78177392e-01 6.84856534e-01
-2.35228781e-02 4.35365945e-01 1.85318780e+00 -7.27769077e-01
2.50828594e-01 7.97116935e-01 9.29397404e-01 -4.48247910e-01
-1.21726239e+00 -1.06699562e+00 -3.25096399e-01 -2.93862820e-01
-7.79053986e-01 -1.80052087e-01 6.88520014e-01 -8.08809757e-01
1.14239907e+00 2.76885331e-01 1.24111259e+00 5.30578494e-01
8.28446329e-01 4.27140445e-01 1.42948639e+00 -2.66875476e-01
8.55284929e-03 -4.70165871e-02 1.63204758e-03 9.46456864e-02
1.13723493e+00 6.68466032e-01 -3.57820928e-01 -4.35303777e-01
1.05742610e+00 3.13310683e-01 4.25506204e-01 -2.83994246e-02
-1.26076865e+00 1.12524867e+00 -4.01753783e-02 5.66152990e-01
-6.67506635e-01 1.55737251e-03 4.82925206e-01 1.18093550e+00
7.20189452e-01 8.43547523e-01 -8.86830628e-01 -5.47223628e-01
-1.28568649e+00 5.47183096e-01 1.29121792e+00 3.04746956e-01
4.17447418e-01 4.25118655e-01 2.15829358e-01 1.47113934e-01
2.73213297e-01 7.25995481e-01 7.68314481e-01 -1.08750224e+00
2.32133776e-01 4.86413807e-01 3.06469947e-01 -1.13706779e+00
-4.63413298e-01 -6.27095878e-01 -3.34256858e-01 5.03039300e-01
7.32831359e-01 -3.34700614e-01 -3.73210013e-02 1.20776141e+00
-4.21917915e-01 -2.24734358e-02 8.11005831e-02 5.99515557e-01
-3.87455612e-01 5.33851504e-01 -1.96995988e-01 -5.10362685e-01
1.17547882e+00 -6.35268271e-01 -4.82408136e-01 -5.88794611e-02
6.44265711e-01 -9.01457667e-01 6.76496625e-01 4.84882355e-01
-1.15061629e+00 -1.41960934e-01 -6.57386124e-01 4.68186796e-01
-2.20468998e-01 -4.94761914e-01 1.13611233e+00 7.04799175e-01
-9.13780630e-01 7.73543715e-01 -7.99551845e-01 3.01619798e-01
1.53081343e-01 1.87603489e-01 2.06840992e-01 6.73771560e-01
-8.07644188e-01 1.06730735e+00 3.77876550e-01 -2.80930072e-01
-1.00275531e-01 -1.00587142e+00 -2.49230072e-01 2.11685970e-01
3.16076905e-01 -5.73152959e-01 1.40689194e+00 -1.49614227e+00
-1.07013571e+00 3.55506688e-01 3.97372305e-01 -1.02818382e+00
5.66811383e-01 -7.16722831e-02 -5.38496196e-01 -4.11132500e-02
3.43723863e-01 -6.25976622e-02 5.06896675e-01 -5.02378464e-01
-1.06304824e+00 -3.67514610e-01 -2.32642442e-01 -1.30131051e-01
-1.36471137e-01 2.61602193e-01 6.25797093e-01 -1.20072579e+00
2.75622666e-01 -6.73725784e-01 5.01691317e-03 -5.99303782e-01
3.62283409e-01 -1.06338181e-01 3.11099827e-01 -8.34087074e-01
1.65173447e+00 -1.76110971e+00 -7.48489022e-01 5.51424563e-01
1.43806681e-01 -5.41355610e-01 3.85935575e-01 6.23076499e-01
-3.81837785e-01 4.83119011e-01 1.31767601e-01 2.09230825e-01
5.59182644e-01 -1.21710457e-01 -7.41019666e-01 5.62835574e-01
1.28901377e-01 1.25950241e+00 -3.58575344e-01 -5.59899621e-02
1.34547025e-01 -4.84851331e-01 -6.63413048e-01 -3.63338381e-01
7.90730268e-02 -2.12827936e-01 -2.31887892e-01 9.66800034e-01
3.81835252e-01 -3.34886819e-01 1.26133442e-01 2.65086025e-01
-9.57235932e-01 4.33661997e-01 -9.12158549e-01 8.02439988e-01
-9.58929881e-02 8.28854024e-01 -1.40900344e-01 -9.87240970e-01
5.97008049e-01 2.39336535e-01 7.52517104e-01 -1.18552315e+00
9.31655318e-02 8.04080606e-01 6.68319643e-01 -9.95020717e-02
6.12305284e-01 -5.76713026e-01 -3.45929600e-02 9.63743329e-01
-4.41324264e-01 -1.47796441e-02 1.46487519e-01 -3.42534095e-01
9.57121432e-01 -5.73197417e-02 4.78124321e-01 -4.88209188e-01
-4.07511979e-01 4.44732934e-01 6.60662234e-01 6.88599408e-01
5.99712767e-02 1.91053703e-01 6.09207809e-01 -4.93208081e-01
-1.44507086e+00 -9.36226189e-01 -1.29261672e-01 9.50264096e-01
-4.86173958e-01 5.07408902e-02 -1.74337909e-01 -1.06334304e-02
8.08496594e-01 8.06018829e-01 -4.14150536e-01 4.20164049e-01
-1.94913417e-01 -8.08916152e-01 3.70236874e-01 4.21177894e-01
3.41799706e-01 -8.13792348e-01 -1.28194785e+00 3.70601147e-01
5.20211577e-01 -3.69842023e-01 -4.36561182e-02 2.28597760e-01
-1.18137610e+00 -1.08990908e+00 -6.73234165e-01 -1.99271604e-01
6.74921495e-04 1.21129215e-01 1.24064767e+00 9.50848460e-02
1.71463832e-01 4.13820684e-01 -2.37150744e-01 -9.30094123e-01
-2.16934443e-01 -8.52855071e-02 -4.32329066e-02 -3.25015128e-01
7.00328052e-01 -6.00047052e-01 -7.43330121e-01 4.87301685e-02
-7.73958743e-01 -3.54997575e-01 9.11155164e-01 7.49757349e-01
9.34090614e-02 1.68154255e-01 1.03084099e+00 -3.65303129e-01
7.94492543e-01 -7.50971854e-01 -9.83707428e-01 1.62521489e-02
-1.48437870e+00 -5.44083342e-02 4.36252356e-02 -2.95297295e-01
-8.73413205e-01 -8.58707726e-01 2.14969367e-01 -2.08672285e-01
4.14880276e-01 1.20201743e+00 6.91068649e-01 2.91700453e-01
2.68155664e-01 2.43245102e-02 4.67566758e-01 -3.61250550e-01
-1.46595046e-01 3.23260158e-01 2.69530535e-01 -4.10350800e-01
9.74544168e-01 5.40081799e-01 -2.07807705e-01 -4.98476714e-01
-3.84745806e-01 -4.30668294e-01 -2.03264266e-01 6.14228062e-02
5.58466434e-01 -1.08416307e+00 -6.49409294e-01 3.87191474e-01
-4.24100339e-01 -3.68213356e-01 -5.11251032e-01 1.29825044e+00
-6.09075606e-01 -2.19460472e-01 -8.51673782e-01 -1.34686637e+00
-2.77600795e-01 -7.03651428e-01 2.23447099e-01 2.91311871e-02
-7.84450829e-01 -1.18875027e+00 4.05982763e-01 3.56412023e-01
6.82287991e-01 4.49660301e-01 8.69736254e-01 -1.32484627e+00
-6.67742848e-01 -4.99283820e-01 -7.52124935e-02 1.81473210e-01
3.38124692e-01 2.18438789e-01 -5.39152026e-01 -2.39152744e-01
4.55753654e-01 -3.65935899e-02 5.79382539e-01 6.80247426e-01
1.26286864e-01 -6.33757174e-01 9.35676768e-02 2.56044000e-01
1.44586289e+00 7.27880359e-01 2.53667742e-01 9.98872638e-01
-2.10999757e-01 5.73196352e-01 5.56762338e-01 8.95090699e-01
3.37788790e-01 -5.34478948e-02 -8.60951021e-02 4.64016311e-02
7.83527493e-01 -3.64993095e-01 2.72181332e-01 6.75032079e-01
-2.02842385e-01 5.79454720e-01 -1.03890812e+00 3.00386518e-01
-1.60786080e+00 -1.58862281e+00 6.11040927e-02 1.94982278e+00
4.42364633e-01 7.95782030e-01 6.84760630e-01 1.43627778e-01
1.21757209e-01 1.85107142e-01 -6.11735702e-01 -2.93040574e-01
-2.93191850e-01 2.55067438e-01 9.62343395e-01 6.78745285e-02
-6.17340744e-01 6.12957656e-01 6.91676283e+00 2.63536185e-01
-1.20098794e+00 -1.74794406e-01 8.37053716e-01 -1.39891386e-01
-1.00489366e+00 5.02680123e-01 -2.43480176e-01 5.53928137e-01
9.29232597e-01 -7.89528191e-01 1.45320982e-01 9.41158354e-01
5.22646725e-01 -3.18058789e-01 -8.43253911e-01 8.80630314e-01
-3.90700966e-01 -1.59514153e+00 -2.83076763e-01 6.68093741e-01
7.08775878e-01 -1.27278478e-03 6.96916461e-01 3.74664932e-01
3.40763003e-01 -9.08238530e-01 1.22822356e+00 6.90253377e-01
1.83803871e-01 -9.17797387e-01 7.66950071e-01 3.69870454e-01
-8.80697310e-01 -7.00986445e-01 -2.20610186e-01 -8.27680588e-01
-2.76389830e-02 5.72319448e-01 -5.11504948e-01 2.39882633e-01
6.34760797e-01 3.82928163e-01 -2.85693496e-01 8.70282054e-01
4.87441093e-01 8.65370452e-01 -4.56651419e-01 9.47389752e-02
5.68397045e-01 -8.48924577e-01 1.25797823e-01 8.60445201e-01
7.23704815e-01 2.36379728e-01 -3.81372243e-01 9.02662694e-01
1.92321062e-01 1.32697865e-01 -1.00764251e+00 -6.89104676e-01
4.66132790e-01 7.15133369e-01 -8.92168701e-01 -8.30555782e-02
-1.30268252e+00 4.74997669e-01 -4.15038049e-01 7.56154135e-02
-4.23012465e-01 -3.50985713e-02 6.66389883e-01 4.47593421e-01
9.00968835e-02 -5.03304124e-01 -8.55989754e-01 -1.23188245e+00
4.03601490e-02 -1.07903004e+00 4.93352741e-01 -3.81091684e-01
-1.16373658e+00 -4.19055730e-01 -1.87587008e-01 -9.14762318e-01
-6.23303831e-01 -5.66059530e-01 -7.53958106e-01 8.23410094e-01
-1.42961419e+00 -5.34806609e-01 7.13234961e-01 2.67400712e-01
3.13564777e-01 -6.27091527e-01 3.58697504e-01 -1.92675143e-01
-2.28140146e-01 1.28874600e-01 7.28770792e-01 3.78646672e-01
3.90378654e-01 -1.26202905e+00 6.05044365e-01 6.02297664e-01
4.44829375e-01 7.85233557e-01 5.16842782e-01 -1.13150311e+00
-1.45432341e+00 -2.44043961e-01 9.62658226e-01 -4.94929075e-01
1.47228813e+00 3.37097853e-01 -6.44134581e-01 1.02611768e+00
5.06607592e-01 -7.27654099e-01 8.79149973e-01 4.17807512e-02
-3.63166332e-01 -4.20133859e-01 -9.37737107e-01 5.46783030e-01
3.74537259e-01 -5.34423828e-01 -1.09776461e+00 -2.98393577e-01
4.83141035e-01 3.71058851e-01 -1.11487925e+00 1.21288456e-01
1.24073160e+00 -1.30795932e+00 7.01908529e-01 -4.26844865e-01
1.12924270e-01 2.86301821e-01 -2.17744082e-01 -8.78488004e-01
-4.16971415e-01 -8.89217377e-01 3.13243657e-01 9.55486178e-01
7.18899429e-01 -1.20522606e+00 7.77619898e-01 1.01570833e+00
1.37706250e-01 -4.18588907e-01 -9.83421445e-01 -8.87774110e-01
4.47201967e-01 -6.23040855e-01 8.36604476e-01 1.31524301e+00
1.26525432e-01 -2.30392456e-01 3.76426168e-02 -3.48806947e-01
7.02925742e-01 5.92166364e-01 8.03676963e-01 -1.57760310e+00
-2.98639327e-01 -1.17428136e+00 -1.41460732e-01 -5.95517695e-01
1.96511269e-01 -7.83777535e-01 -8.38275552e-01 -1.02849913e+00
-9.48625337e-03 -2.79404312e-01 -5.20458817e-01 1.36713058e-01
4.03008789e-01 -2.45787650e-01 3.72089207e-01 5.88683307e-01
1.17534697e-01 6.10969663e-02 7.61181533e-01 1.24540851e-01
-3.67023796e-01 1.20517775e-01 -1.02151620e+00 9.26364064e-01
1.16285765e+00 -8.83439183e-02 -3.24133724e-01 -2.02325750e-02
8.71705472e-01 3.26730311e-01 5.68831086e-01 -6.50782466e-01
9.79094654e-02 -6.54421389e-01 4.83631819e-01 -7.49302983e-01
-2.73674250e-01 -9.64630485e-01 5.58494985e-01 7.76784360e-01
-7.52949193e-02 1.08002889e+00 1.33908167e-01 2.41073444e-01
-3.40810061e-01 -8.54412839e-02 3.17646265e-01 -3.63391936e-01
-4.92505550e-01 -1.33381095e-02 -4.94312912e-01 2.55957156e-01
9.89182353e-01 -4.71678197e-01 -2.77763009e-02 -4.72323805e-01
-1.34068951e-01 7.13434294e-02 6.60434425e-01 1.21317819e-01
3.72552872e-01 -1.13118529e+00 -7.24220395e-01 3.45327817e-02
-2.05199853e-01 -7.66629100e-01 -3.55216414e-01 1.04971302e+00
-7.81333804e-01 8.17554951e-01 -1.84943035e-01 -1.99440885e-02
-4.16186750e-01 5.59527755e-01 3.79240245e-01 -1.15263239e-01
-7.13583827e-01 1.91749781e-01 -6.63989633e-02 3.58756483e-01
-2.53221363e-01 -7.60356843e-01 4.41284120e-01 8.35473359e-01
4.16232109e-01 6.54448509e-01 -3.25867653e-01 -2.39249736e-01
8.08659121e-02 2.85064965e-01 2.99334049e-01 -6.97264016e-01
1.65596235e+00 -6.83693066e-02 -2.02271432e-01 9.12897706e-01
8.83352876e-01 8.14579204e-02 -1.12039328e+00 -4.62662019e-02
8.80224347e-01 -4.79751855e-01 -8.91069323e-02 -6.04727805e-01
-9.49377596e-01 3.18673074e-01 1.79015532e-01 7.68489838e-01
8.20716083e-01 -4.46523428e-01 6.47544324e-01 5.56473076e-01
3.12558979e-01 -1.59611726e+00 -1.10397100e-01 1.32192925e-01
6.43617332e-01 -1.16718304e+00 -9.15659964e-03 8.04840207e-01
-5.85465550e-01 1.05514705e+00 -1.23570390e-01 -4.04258877e-01
1.19071674e+00 3.80919933e-01 -8.61959532e-02 -1.71470299e-01
-8.29784811e-01 1.93745289e-02 -1.60352334e-01 -8.18442255e-02
5.11909187e-01 2.98451334e-01 -7.23475575e-01 9.33718383e-01
-8.66330683e-01 2.37714909e-02 4.63182300e-01 9.14402843e-01
-6.06394768e-01 -7.69329190e-01 -5.35100877e-01 1.02359354e+00
-1.23180950e+00 -3.66937935e-01 -3.72411907e-01 1.54491866e+00
-4.00917351e-01 8.79730999e-01 7.82311559e-01 -1.82908028e-01
2.26022959e-01 4.25987184e-01 -8.26957747e-02 -1.03500076e-01
-8.08311760e-01 6.51311576e-01 -2.01453581e-01 -2.56022960e-01
-3.57373387e-01 -1.17106020e+00 -9.67674732e-01 -9.59875882e-01
-2.42735013e-01 3.64800394e-01 3.99744302e-01 8.76278579e-01
-1.35475680e-01 -2.07957432e-01 8.80382419e-01 -3.11008126e-01
-1.23420358e+00 -6.98545158e-01 -1.24797606e+00 5.12581207e-02
4.14026737e-01 -3.47750962e-01 -6.23733938e-01 -2.29327649e-01] | [4.5543975830078125, 4.213082313537598] |
2e0081de-4927-479c-a9c0-6eb894963e59 | unsupervised-change-point-detection-for | 2305.11976 | null | https://arxiv.org/abs/2305.11976v1 | https://arxiv.org/pdf/2305.11976v1.pdf | Unsupervised Change Point Detection for heterogeneous sensor signals | Change point detection is a crucial aspect of analyzing time series data, as the presence of a change point indicates an abrupt and significant change in the process generating the data. While many algorithms for the problem of change point detection have been developed over time, it can be challenging to select the appropriate algorithm for a specific problem. The choice of the algorithm heavily depends on the nature of the problem and the underlying data source. In this paper, we will exclusively examine unsupervised techniques due to their flexibility in the application to various data sources without the requirement for abundant annotated training data and the re-calibration of the model. The examined methods will be introduced and evaluated based on several criteria to compare the algorithms. | ['Mario Krause'] | 2023-05-19 | null | null | null | null | ['change-point-detection'] | ['time-series'] | [ 3.59691173e-01 -6.14487350e-01 -9.29648653e-02 -2.15591207e-01
-2.31032774e-01 -9.05803859e-01 7.78789520e-01 7.22603202e-01
-2.26312160e-01 5.53357303e-01 -2.71375030e-01 -3.29836130e-01
-4.75598574e-01 -6.14019692e-01 -1.44140124e-01 -7.37352669e-01
-3.24979931e-01 3.36612225e-01 3.13624948e-01 -2.58500367e-01
5.33705354e-01 7.37009168e-01 -1.70830345e+00 -2.12536216e-01
6.72414720e-01 9.01706338e-01 2.62825727e-01 6.05291426e-01
-3.21403563e-01 4.70909506e-01 -7.25666761e-01 3.53081793e-01
2.77327687e-01 -8.08523059e-01 -3.96377295e-01 2.82761782e-01
-1.36710078e-01 3.33274007e-01 2.39699274e-01 8.81949425e-01
2.45467559e-01 1.83131605e-01 7.37678707e-01 -1.34285808e+00
2.50646621e-01 2.31907845e-01 -2.89523512e-01 7.34235883e-01
4.40762401e-01 -9.87155363e-02 8.27478707e-01 -7.79172182e-01
7.53062487e-01 5.69526911e-01 5.73807716e-01 3.16390395e-02
-1.22255719e+00 -1.41990200e-01 6.67206571e-02 3.42281282e-01
-1.24563515e+00 -3.90438586e-01 1.23394513e+00 -7.73395479e-01
6.01935089e-01 3.27423692e-01 7.49818861e-01 6.23256564e-01
2.12539405e-01 3.57651383e-01 1.14268970e+00 -6.13391280e-01
5.01248062e-01 5.41152805e-02 2.07328662e-01 1.23297051e-01
2.18784094e-01 -8.84050876e-02 -2.44380593e-01 -1.04517594e-01
5.27967513e-01 5.71214743e-02 -2.20526397e-01 -5.26363492e-01
-9.43147480e-01 6.21204555e-01 -3.22273597e-02 9.51857746e-01
-5.48279047e-01 -2.49562904e-01 5.69136918e-01 5.58506012e-01
4.43181336e-01 5.42167664e-01 -5.68577290e-01 -5.26909709e-01
-1.20717478e+00 2.79638171e-01 7.12510526e-01 4.63820189e-01
5.30371428e-01 5.20280488e-02 2.56947428e-01 5.15802503e-01
-4.01529893e-02 9.39697996e-02 6.54641211e-01 -3.84849757e-01
2.93436110e-01 8.27681422e-01 2.86584467e-01 -1.23889863e+00
-3.41474891e-01 -5.58478475e-01 -5.04027128e-01 2.22113475e-01
5.36891699e-01 -7.51250088e-02 -6.73119903e-01 1.41061354e+00
4.71755654e-01 -2.29995146e-01 -6.34094700e-02 4.06490147e-01
3.16353679e-01 8.66327524e-01 -2.87375689e-01 -7.76559651e-01
8.84893596e-01 -2.77798176e-01 -9.29185152e-01 4.37871963e-02
2.89358228e-01 -8.32130075e-01 8.57521117e-01 3.05644035e-01
-5.82156360e-01 -4.30929929e-01 -9.20615315e-01 5.26346743e-01
-6.02746367e-01 -2.28145406e-01 1.78707838e-01 1.94432929e-01
-5.74141562e-01 5.51002145e-01 -7.93005884e-01 -8.10678661e-01
-3.10027480e-01 1.87476337e-01 -1.95791740e-02 4.56806690e-01
-9.96678174e-01 9.26205337e-01 6.47809625e-01 2.97941864e-01
-4.74758565e-01 -3.46396953e-01 -3.74550700e-01 -1.22637890e-01
3.14716846e-01 -1.11019969e-01 1.10375071e+00 -1.24776340e+00
-1.06312287e+00 6.66513562e-01 -1.08920097e-01 -4.55432862e-01
9.24407721e-01 1.67195559e-01 -8.75460267e-01 -1.71638597e-02
3.67153138e-02 -2.03065470e-01 9.10948098e-01 -9.14054275e-01
-9.03721154e-01 -1.63048327e-01 -1.37488708e-01 4.56392067e-03
2.72425767e-02 1.66227892e-01 -1.36564955e-01 -8.47344458e-01
3.89055967e-01 -8.30235898e-01 -6.85183555e-02 -3.16152930e-01
1.74279697e-02 -2.34880015e-01 1.25397229e+00 -5.14431417e-01
1.57222760e+00 -2.39271784e+00 -7.90217221e-02 4.26930100e-01
-2.05825552e-01 -7.99241439e-02 3.94149959e-01 1.02280819e+00
-3.64381313e-01 -1.36793122e-01 -4.98397827e-01 2.29344189e-01
-3.22609007e-01 1.03389576e-01 -5.38668215e-01 6.57479525e-01
7.74531364e-02 3.11152130e-01 -8.44177783e-01 -3.08752388e-01
4.31305140e-01 8.03589374e-02 8.44940245e-02 1.87542543e-01
-8.44566301e-02 6.47040784e-01 -4.92226809e-01 4.61236209e-01
1.05232745e-01 -4.87634912e-02 1.17200181e-01 -1.00891106e-02
-5.79537749e-01 2.15103894e-01 -1.52724135e+00 1.02673292e+00
-2.33891055e-01 8.39453936e-01 -3.72515351e-01 -1.22318959e+00
1.18222702e+00 5.76470137e-01 1.06352401e+00 -5.00136375e-01
2.90799304e-03 5.71078062e-01 3.26263368e-01 -5.74019194e-01
5.13624549e-01 -8.50970820e-02 -2.09966395e-02 5.20658970e-01
-2.78273433e-01 -8.52583200e-02 7.60242283e-01 -3.77221018e-01
9.94211435e-01 8.34572613e-02 8.28259110e-01 -3.60299796e-01
4.53861743e-01 4.25283253e-01 6.12649977e-01 3.68702561e-01
-2.54095137e-01 2.67108232e-01 4.15499926e-01 -5.91445506e-01
-1.00502515e+00 -5.82991302e-01 -4.05154377e-01 7.13401198e-01
-7.43504893e-03 -1.49540797e-01 -3.46509486e-01 -4.38731164e-01
-4.30227779e-02 6.21843100e-01 -5.12980282e-01 1.83281928e-01
-7.09046781e-01 -8.97092283e-01 6.93584457e-02 3.49557586e-02
2.35727429e-01 -1.11598873e+00 -1.09191787e+00 6.47005498e-01
-1.55573815e-01 -7.42194712e-01 -1.50221866e-02 4.20154929e-01
-1.33199930e+00 -1.29024410e+00 -2.19487548e-01 -5.71345329e-01
7.61888921e-01 1.69298574e-01 8.75534832e-01 -2.33660907e-01
-1.24005459e-01 2.98070431e-01 -5.62887967e-01 -5.77277780e-01
-5.83444417e-01 9.07954350e-02 -1.71676036e-02 1.20141074e-01
3.58924687e-01 -5.41261256e-01 -2.95197964e-01 3.55604649e-01
-1.02692592e+00 -3.48978996e-01 2.11737528e-01 4.60545152e-01
5.66821337e-01 7.55246043e-01 6.08127058e-01 -7.42512643e-01
7.83480108e-01 -4.94273126e-01 -7.82727659e-01 2.50158310e-01
-8.46487701e-01 -3.64665464e-02 6.72988236e-01 -3.41743201e-01
-5.41923821e-01 2.58351773e-01 2.16239393e-01 -1.00370869e-01
-2.53937542e-01 8.64923060e-01 5.95817231e-02 9.96399373e-02
5.77079654e-01 2.15963244e-01 -2.64232397e-01 -5.50415277e-01
6.14676438e-02 4.27258760e-01 4.92774785e-01 -1.82355240e-01
7.36536682e-01 3.58771116e-01 3.40222530e-02 -9.91822183e-01
-2.00947717e-01 -7.02008486e-01 -9.61597025e-01 -5.90390563e-01
3.78558010e-01 -2.90114313e-01 1.31663650e-01 4.72836941e-01
-8.72317493e-01 3.46464589e-02 -5.56946576e-01 2.22020984e-01
-4.93367255e-01 2.72821784e-01 6.62169605e-02 -7.64812350e-01
-2.47774795e-01 -8.14108610e-01 3.42727989e-01 8.06381702e-02
-5.00497043e-01 -1.31604242e+00 3.95995617e-01 -2.91642219e-01
3.04163605e-01 8.20382893e-01 8.77322197e-01 -7.74981737e-01
-1.55736476e-01 -7.65814722e-01 4.14020389e-01 7.21231475e-02
8.27181578e-01 5.87621212e-01 -7.03457594e-01 -2.71175325e-01
4.34577286e-01 5.29349506e-01 2.31009826e-01 5.04980505e-01
7.98478842e-01 -1.87732831e-01 -2.72899389e-01 3.31427492e-02
1.57250845e+00 6.74073458e-01 3.11488241e-01 7.06062794e-01
2.45955408e-01 7.31195152e-01 7.67301381e-01 4.52280104e-01
-1.54106319e-01 7.74220467e-01 2.95664340e-01 -4.71559688e-02
2.18024984e-01 -1.10446326e-01 1.69229552e-01 9.50125635e-01
-1.08193858e-02 -1.30240887e-01 -1.18010545e+00 7.86628485e-01
-1.80827343e+00 -1.12119341e+00 -4.61395591e-01 2.32302666e+00
6.02566063e-01 3.72843921e-01 4.45715278e-01 8.53316188e-01
8.81779671e-01 1.89960390e-01 -4.06605512e-01 -2.38992125e-01
-1.93066910e-01 -2.56989390e-01 2.87570596e-01 2.54455745e-01
-1.15036345e+00 2.73671150e-01 6.95925474e+00 4.32016313e-01
-1.60030746e+00 -1.14922047e-01 2.16388807e-01 1.36562213e-01
5.75748086e-02 -5.52705340e-02 -3.60041291e-01 6.98700607e-01
7.98653662e-01 -5.40210843e-01 1.64728731e-01 5.34018278e-01
7.86347866e-01 -2.64485657e-01 -1.11933827e+00 8.21245730e-01
-1.65911242e-01 -7.31950462e-01 -2.56138206e-01 -1.78014293e-01
6.62813008e-01 -1.09408028e-01 -3.06524217e-01 -2.79390693e-01
-2.54260272e-01 -5.29992700e-01 8.86738658e-01 5.70809245e-01
3.19720805e-01 -5.84347427e-01 5.50859272e-01 3.15467089e-01
-1.26042306e+00 -1.67173967e-01 1.66593343e-01 -2.66975582e-01
2.13001311e-01 8.48114550e-01 -8.13851774e-01 5.60020745e-01
4.68077302e-01 6.42174780e-01 -5.67124724e-01 1.37452400e+00
-1.71076566e-01 7.55438685e-01 -6.54137671e-01 -1.17873758e-01
3.79179530e-02 -3.04427832e-01 9.27627444e-01 1.03657937e+00
4.53093350e-01 -2.83327132e-01 2.61614889e-01 5.09942830e-01
5.98611057e-01 4.56312805e-01 -7.31413782e-01 -2.09906548e-01
5.40770531e-01 1.04494226e+00 -1.06841874e+00 -1.28093824e-01
-4.58029300e-01 5.44008315e-01 -2.62321889e-01 2.97988355e-01
-4.91097420e-01 -4.74521160e-01 4.11270022e-01 4.08449739e-01
2.81249821e-01 -5.76995313e-01 -2.98290431e-01 -8.83842468e-01
3.67000073e-01 -8.00699055e-01 5.71833849e-01 -3.89571577e-01
-1.15326130e+00 5.29604375e-01 1.71210513e-01 -1.70817029e+00
-7.00000644e-01 -1.81319758e-01 -8.06104481e-01 6.31629109e-01
-1.08112073e+00 -5.61732829e-01 -3.59062225e-01 3.92053574e-01
6.81013167e-01 -2.21193694e-02 5.67084312e-01 2.52602339e-01
-4.92872387e-01 -2.22856432e-01 4.86565351e-01 -7.99718201e-02
5.14453232e-01 -1.22908866e+00 1.78446010e-01 1.44378984e+00
1.56488076e-01 5.25909841e-01 1.35055888e+00 -6.02548480e-01
-1.19288468e+00 -7.87077844e-01 9.89883184e-01 -2.83035457e-01
1.16127050e+00 -4.98767607e-02 -1.06361914e+00 3.18283468e-01
-4.54280563e-02 -2.54864872e-01 4.97867882e-01 1.11350030e-01
2.12348372e-01 -3.78464848e-01 -9.89851713e-01 3.69429052e-01
4.91001278e-01 -5.37757397e-01 -8.99854124e-01 -2.95620854e-03
-1.68675352e-02 -1.31384730e-01 -8.08624327e-01 3.17711800e-01
2.70783544e-01 -6.55426145e-01 3.32422614e-01 -3.98315102e-01
1.26400128e-01 -8.01864266e-01 2.04606634e-02 -1.20403433e+00
-3.45700681e-01 -4.05044079e-01 -9.23301131e-02 1.48185003e+00
2.50588357e-01 -5.20268381e-01 4.07696724e-01 4.14036483e-01
2.62897164e-01 -3.68176073e-01 -9.10044193e-01 -7.79342711e-01
-4.83372420e-01 -5.22188962e-01 6.32887304e-01 1.04888928e+00
1.07788809e-01 1.81866214e-01 -8.17446187e-02 1.63409054e-01
3.76816660e-01 5.68219841e-01 8.01616013e-01 -1.58985364e+00
-2.41892394e-02 -6.03277385e-01 -6.61640704e-01 -4.29427236e-01
-4.42263067e-01 -4.85140800e-01 2.37832487e-01 -1.39845109e+00
-2.73059994e-01 -4.72086459e-01 -3.66368562e-01 2.60519058e-01
-1.48775965e-01 -1.24455377e-01 -7.70327374e-02 7.28593111e-01
-3.79601754e-02 2.15120450e-01 5.27580321e-01 -6.76191822e-02
-7.14822769e-01 4.20087963e-01 -1.61396205e-01 5.40482104e-01
9.28813577e-01 -5.88109612e-01 -6.40649498e-01 -7.49007463e-02
2.86370099e-01 -1.45995900e-01 1.36569157e-01 -1.04287195e+00
1.01815693e-01 -3.21400434e-01 2.69974004e-02 -6.80174053e-01
-3.37097615e-01 -1.28428555e+00 7.89333105e-01 6.65862501e-01
-1.42307699e-01 6.78175867e-01 6.53110445e-02 5.13962269e-01
-5.15643358e-01 -6.10095561e-01 7.27168679e-01 3.11237164e-02
-1.19377649e+00 6.10360354e-02 -5.65311372e-01 -1.30060688e-01
1.26012588e+00 -4.18074310e-01 2.02627435e-01 -2.41458759e-01
-7.08714664e-01 -7.44151399e-02 6.73773587e-01 6.09897077e-01
9.96950939e-02 -1.03365219e+00 -4.89311934e-01 -7.81422888e-04
3.01736534e-01 -2.86832958e-01 -2.21925199e-01 7.27385461e-01
-6.04751587e-01 2.56742507e-01 -2.99364090e-01 -6.53817415e-01
-1.20420420e+00 6.49384558e-01 4.16621685e-01 -5.40803522e-02
-6.67509258e-01 -3.74840759e-02 -3.09162199e-01 1.49004757e-01
-5.26572727e-02 -4.88946438e-01 -4.40593421e-01 5.57137072e-01
3.00006956e-01 4.65323865e-01 3.81133169e-01 -5.88950515e-01
-3.41397405e-01 4.56991374e-01 2.91948497e-01 -9.41040292e-02
1.42987716e+00 -3.22117805e-01 -2.67632037e-01 1.25989783e+00
7.39530742e-01 -1.41414776e-01 -1.07869494e+00 -1.28892034e-01
5.77985466e-01 -4.32720214e-01 -8.05539172e-03 -3.68128061e-01
-7.02646196e-01 3.03839147e-01 8.26925039e-01 9.39438403e-01
1.33674920e+00 -3.35199982e-01 1.38639167e-01 1.35653943e-01
4.61814284e-01 -1.21934843e+00 -3.67422611e-01 3.53407711e-01
9.04243112e-01 -9.24010158e-01 1.48001641e-01 -1.52524740e-01
-2.25664586e-01 1.26013470e+00 -4.74325707e-03 -7.04278201e-02
8.56622517e-01 8.42280313e-03 1.68726832e-01 -1.87203750e-01
-5.01077354e-01 -1.67340875e-01 2.25778922e-01 3.44965219e-01
4.87229824e-01 -5.59496507e-02 -9.70642447e-01 -9.44139808e-02
-1.40467688e-01 2.11497381e-01 4.42506850e-01 1.31792736e+00
-5.12827694e-01 -1.12292123e+00 -5.17175496e-01 4.09751683e-01
-4.01845962e-01 4.26377773e-01 -4.98420119e-01 9.37338412e-01
2.25072578e-01 1.05534971e+00 1.17319368e-01 8.56876075e-02
5.63795686e-01 1.77682728e-01 1.87625736e-01 -4.11130607e-01
-5.19825578e-01 1.90930143e-01 -6.64162114e-02 -8.40975791e-02
-8.01018775e-01 -1.31527960e+00 -1.05391359e+00 -1.64450556e-01
-2.98125297e-01 4.21396613e-01 8.22831511e-01 1.04776442e+00
7.96551704e-02 4.08972919e-01 9.63973224e-01 -3.24914396e-01
-1.04575939e-01 -9.72806394e-01 -6.74412012e-01 5.68969607e-01
2.61482149e-01 -7.30133951e-01 -4.83635008e-01 4.29987282e-01] | [7.240373134613037, 3.3325536251068115] |
c177ef8a-af6b-490d-a444-4762212a2c29 | template-based-automatic-search-of-compact | 1904.02365 | null | https://arxiv.org/abs/1904.02365v2 | https://arxiv.org/pdf/1904.02365v2.pdf | Template-Based Automatic Search of Compact Semantic Segmentation Architectures | Automatic search of neural architectures for various vision and natural language tasks is becoming a prominent tool as it allows to discover high-performing structures on any dataset of interest. Nevertheless, on more difficult domains, such as dense per-pixel classification, current automatic approaches are limited in their scope - due to their strong reliance on existing image classifiers they tend to search only for a handful of additional layers with discovered architectures still containing a large number of parameters. In contrast, in this work we propose a novel solution able to find light-weight and accurate segmentation architectures starting from only few blocks of a pre-trained classification network. To this end, we progressively build up a methodology that relies on templates of sets of operations, predicts which template and how many times should be applied at each step, while also generating the connectivity structure and downsampling factors. All these decisions are being made by a recurrent neural network that is rewarded based on the score of the emitted architecture on the holdout set and trained using reinforcement learning. One discovered architecture achieves 63.2% mean IoU on CamVid and 67.8% on CityScapes having only 270K parameters. Pre-trained models and the search code are available at https://github.com/DrSleep/nas-segm-pytorch. | ['Chunhua Shen', 'Vladimir Nekrasov', 'Ian Reid'] | 2019-04-04 | null | null | null | null | ['holdout-set'] | ['computer-vision'] | [ 2.42292210e-01 6.93102553e-02 6.28582537e-02 -4.33058619e-01
-5.78055263e-01 -4.99186516e-01 5.59629083e-01 2.50489879e-02
-7.59417534e-01 5.63403726e-01 -2.47774482e-01 -3.58834565e-01
-1.59714997e-01 -6.03296161e-01 -8.10464025e-01 -7.54916191e-01
-1.04422178e-02 7.14618206e-01 4.83712554e-01 -1.79357186e-01
4.44750160e-01 6.30770564e-01 -1.77331865e+00 2.25993499e-01
5.98234236e-01 1.12427521e+00 6.64104819e-01 6.24032617e-01
5.86621687e-02 6.14595711e-01 -3.48321050e-01 -4.13318217e-01
3.94900560e-01 -3.03716660e-01 -8.99485171e-01 -9.95716304e-02
5.28773248e-01 -4.24859747e-02 -1.25618353e-01 9.26633894e-01
5.74317873e-01 2.15904012e-01 5.37482023e-01 -7.21080363e-01
-2.36843042e-02 8.22190523e-01 -3.82171810e-01 5.65577269e-01
-8.18623528e-02 3.57900351e-01 1.07420599e+00 -8.61272931e-01
7.43471444e-01 7.21746325e-01 3.27664793e-01 6.53382599e-01
-1.31971633e+00 -5.13392091e-01 7.49060586e-02 3.37645292e-01
-1.21760941e+00 -5.56356311e-01 5.94103038e-01 -3.06145877e-01
1.27154183e+00 1.85879007e-01 7.93567061e-01 1.05663526e+00
-1.56086922e-01 7.39565969e-01 1.05213118e+00 -3.60096306e-01
3.68407458e-01 2.22820193e-01 1.08408071e-01 9.08865392e-01
7.56615102e-02 -1.20549062e-02 -4.17546391e-01 2.12367445e-01
6.60517871e-01 -1.71322793e-01 -1.41633064e-01 -3.33099574e-01
-9.36080694e-01 6.56444907e-01 7.15423524e-01 6.27246857e-01
-6.81845486e-01 8.98967087e-02 2.82478333e-01 1.52729928e-01
2.29096919e-01 7.58715868e-01 -6.93126798e-01 -6.58707246e-02
-1.27358532e+00 1.50847599e-01 6.22246802e-01 5.43051541e-01
9.18527365e-01 -1.01221025e-01 -9.42684524e-03 8.59518170e-01
3.98708396e-02 1.01164579e-02 6.70282364e-01 -7.04153538e-01
2.58544892e-01 9.91759479e-01 -3.58836591e-01 -6.41835213e-01
-5.43478251e-01 -8.82526934e-01 -8.51030052e-01 4.19980794e-01
4.66231227e-01 -7.95555189e-02 -9.46370304e-01 1.33478856e+00
2.80346513e-01 1.85944781e-01 -5.72205447e-02 8.86518180e-01
5.29906631e-01 5.20067573e-01 -1.34139761e-01 -2.74803620e-02
1.33147502e+00 -1.18506372e+00 2.50231400e-02 -4.26191479e-01
5.34708560e-01 -5.64472914e-01 9.58305180e-01 6.45553291e-01
-1.29757690e+00 -6.32017553e-01 -9.13134217e-01 1.44279480e-01
-5.00586808e-01 3.71300548e-01 3.37394565e-01 3.85009438e-01
-1.43490231e+00 7.97387242e-01 -7.32286453e-01 -3.13843757e-01
6.97153747e-01 6.08220816e-01 -1.20263938e-02 1.18902966e-01
-7.04742253e-01 8.74571979e-01 5.77043116e-01 2.76147306e-01
-1.00874293e+00 -3.96650314e-01 -4.72159982e-01 2.83247381e-01
5.47771454e-01 -6.22607708e-01 1.13337374e+00 -1.23561811e+00
-1.56472838e+00 1.03603518e+00 -6.63060769e-02 -1.09761012e+00
6.79481089e-01 -2.98179360e-03 -1.61923952e-02 1.72642544e-01
-5.81643917e-02 1.02814782e+00 1.04893064e+00 -9.18550789e-01
-7.29862571e-01 -4.35637057e-01 -6.19783886e-02 1.70752451e-01
-2.99297482e-01 -7.24986475e-03 -7.17564344e-01 -3.80202919e-01
1.18955590e-01 -9.24579501e-01 -4.98573005e-01 -1.76593751e-01
-4.64254677e-01 -2.76049972e-01 4.65186685e-01 -4.79876399e-01
1.08553684e+00 -1.95730340e+00 3.51945072e-01 3.72961164e-01
-1.59802530e-02 5.08763611e-01 -1.15007669e-01 -3.35538350e-02
-2.98296381e-02 7.27258772e-02 -3.57218415e-01 -3.61942142e-01
-1.95861027e-01 -5.69713302e-02 -9.32743251e-02 2.76820570e-01
2.73876786e-01 8.22770059e-01 -5.48129201e-01 -3.85162264e-01
3.76383305e-01 4.68506485e-01 -6.45165503e-01 7.69852176e-02
-4.08313721e-01 4.25899893e-01 -3.20544064e-01 5.47666788e-01
2.72585273e-01 -5.09326994e-01 3.29924710e-02 -1.46033272e-01
-3.19652528e-01 2.61760920e-01 -1.23707736e+00 1.82488990e+00
-4.22780246e-01 5.15133142e-01 -1.74647808e-01 -1.21917903e+00
1.01534379e+00 7.64527395e-02 2.60077626e-01 -7.75928497e-01
4.37983930e-01 4.67775553e-01 1.58865064e-01 -3.69484305e-01
2.25264445e-01 4.36354548e-01 2.99554855e-01 3.20033073e-01
1.80457234e-01 2.01190159e-01 4.87395108e-01 -4.56239581e-02
1.26813924e+00 1.38345465e-01 2.12951124e-01 -2.31951028e-01
6.40380740e-01 6.55382052e-02 1.21046938e-01 7.12526262e-01
-2.53090616e-02 5.83238780e-01 4.28911358e-01 -6.24965906e-01
-1.09706092e+00 -4.69170749e-01 -2.30115891e-01 1.14401162e+00
-1.91634774e-01 -5.23005240e-02 -1.02508700e+00 -5.46409667e-01
-5.15489519e-01 6.91508830e-01 -6.56597376e-01 4.73580174e-02
-6.15980744e-01 -6.54167116e-01 5.26657879e-01 1.73464924e-01
6.46494269e-01 -1.60602999e+00 -1.24605560e+00 2.15191990e-01
2.67636210e-01 -9.10609305e-01 -1.44143820e-01 5.34820974e-01
-1.09236455e+00 -1.06294668e+00 -8.91254902e-01 -9.52924609e-01
7.57499516e-01 -1.11121662e-01 1.16236651e+00 3.06047499e-01
-6.10932112e-01 -1.76901177e-01 -2.43294597e-01 -1.83827817e-01
-2.36809075e-01 6.81613743e-01 -3.73653799e-01 1.60333589e-01
2.20178410e-01 -4.95110869e-01 -8.19382071e-01 1.42676115e-01
-6.87982440e-01 2.17113510e-01 9.35654938e-01 7.71510601e-01
7.86404252e-01 -1.46480307e-01 2.55398244e-01 -8.90053451e-01
3.69858235e-01 -2.70414203e-01 -1.01409042e+00 2.08997160e-01
-7.20678866e-01 3.67426395e-01 6.04457855e-01 -3.28392953e-01
-7.79779196e-01 5.88589549e-01 -3.45946133e-01 -4.48110491e-01
-4.77626681e-01 2.07061499e-01 1.58528328e-01 7.97300115e-02
8.73143613e-01 4.31657106e-01 -2.49188304e-01 -4.99263912e-01
2.01666832e-01 2.35490873e-01 3.85059506e-01 -2.81679779e-01
5.69256723e-01 2.41985440e-01 -2.07718611e-01 -7.69292474e-01
-5.63373923e-01 -4.69451189e-01 -7.32299030e-01 -2.50862598e-01
9.20924187e-01 -5.27429760e-01 -6.61290705e-01 4.01401728e-01
-9.96461332e-01 -6.15939379e-01 -2.47807071e-01 2.08342984e-01
-4.09897536e-01 -1.08271293e-01 -2.33105049e-01 -7.02085912e-01
-7.19890058e-01 -1.36978555e+00 8.90158892e-01 4.10120100e-01
-2.12306440e-01 -6.37221217e-01 6.68336675e-02 2.93544561e-01
3.87932271e-01 -2.11288333e-02 7.53327310e-01 -7.71367192e-01
-8.19357336e-01 -1.35099053e-01 -2.43598193e-01 3.67201090e-01
-3.38142902e-01 -5.17482683e-02 -1.00685930e+00 -6.53924569e-02
-1.44443989e-01 -3.82976294e-01 9.63110328e-01 5.91452479e-01
1.58181286e+00 -3.63306552e-01 -3.11286360e-01 7.51814246e-01
1.66312659e+00 1.49242863e-01 5.54177284e-01 6.15080655e-01
2.92800158e-01 5.62257946e-01 1.57676145e-01 3.01205784e-01
3.01281530e-02 7.16483653e-01 8.14465523e-01 5.85701969e-03
2.35121767e-03 1.96103565e-02 1.18513092e-01 2.64485121e-01
-3.01740229e-01 -2.49797106e-02 -1.00564969e+00 5.99491119e-01
-1.62237704e+00 -9.34072852e-01 9.27580297e-02 2.09820175e+00
8.14707160e-01 6.25162661e-01 1.89135134e-01 1.26782671e-01
4.93110031e-01 1.38921544e-01 -7.74338007e-01 -3.41844916e-01
-1.14144027e-01 5.45793355e-01 5.85426986e-01 5.10799348e-01
-8.74544740e-01 1.11314714e+00 4.85649872e+00 7.45042384e-01
-1.21507132e+00 -1.13091864e-01 8.89178038e-01 -3.03927153e-01
3.11503559e-02 -8.23793933e-02 -9.73681927e-01 4.38246906e-01
9.48315501e-01 4.26467836e-01 7.01519549e-01 8.54834199e-01
1.79790556e-01 -1.88031927e-01 -9.92888689e-01 9.32429254e-01
-9.80714336e-02 -1.62461424e+00 4.70043272e-02 1.27802223e-01
6.40680730e-01 5.81141531e-01 3.86520065e-02 1.00357838e-01
1.66847870e-01 -1.13888717e+00 7.00201809e-01 4.74645793e-01
5.04483342e-01 -5.98956287e-01 5.16268075e-01 5.86642325e-01
-8.28479409e-01 -1.58161506e-01 -3.19741637e-01 1.63024262e-01
-2.52316296e-01 4.47267026e-01 -1.00203621e+00 7.99419358e-02
9.11516428e-01 4.69785631e-01 -7.10105717e-01 1.20731890e+00
-2.97777891e-01 6.46394074e-01 -5.00583291e-01 -2.96518266e-01
5.55237830e-01 -1.32541388e-01 3.11468035e-01 1.18554127e+00
3.35579544e-01 -2.10528880e-01 -1.61041498e-01 7.75711656e-01
-2.14573562e-01 1.99315012e-01 -4.95842606e-01 2.88129777e-01
3.42721939e-01 1.54835761e+00 -1.16787100e+00 -3.40279549e-01
-4.82294634e-02 9.94540453e-01 6.22747302e-01 1.56493887e-01
-7.92308271e-01 -5.03862128e-02 2.71781713e-01 1.67678207e-01
6.55809581e-01 3.40003669e-02 -4.01627243e-01 -9.06441748e-01
9.39175859e-02 -9.58191037e-01 3.53254974e-01 -6.49430096e-01
-7.11777747e-01 1.12491965e+00 -3.35487008e-01 -7.75682509e-01
-4.00042653e-01 -5.71258843e-01 -5.86399257e-01 7.15443909e-01
-1.47339571e+00 -8.12800825e-01 -2.23402053e-01 5.99637210e-01
1.04203296e+00 -3.40991586e-01 8.01752210e-01 2.14258894e-01
-8.11313033e-01 4.17148471e-01 -1.19014911e-01 4.44717333e-02
2.15973988e-01 -1.21152556e+00 3.78727138e-01 7.55246520e-01
6.66157544e-01 2.71852225e-01 7.54999638e-01 -2.40230247e-01
-9.84650373e-01 -8.47391546e-01 8.59053075e-01 -3.20340991e-02
5.49732506e-01 -2.56035089e-01 -9.40411508e-01 3.97698998e-01
3.47618520e-01 -2.29068846e-02 1.61602393e-01 1.46294469e-02
-7.80997053e-02 -2.10915849e-01 -7.94952393e-01 6.67286098e-01
9.58000302e-01 -1.01726003e-01 -2.52835661e-01 2.52858847e-01
1.99179247e-01 -3.73399436e-01 -3.82021725e-01 1.99380979e-01
3.39266598e-01 -1.19759142e+00 8.29463959e-01 -4.04858559e-01
4.74833310e-01 -1.33475304e-01 9.06913355e-02 -1.02488625e+00
-2.06115797e-01 -5.47252715e-01 5.99786229e-02 8.11081529e-01
9.69029725e-01 -5.59131384e-01 1.15174735e+00 2.96744049e-01
-5.54164164e-02 -1.25929427e+00 -8.97422135e-01 -3.37865442e-01
-3.82542551e-01 -2.84894228e-01 2.37625316e-01 5.51007867e-01
-6.02323413e-01 5.26692450e-01 -1.71767816e-01 1.40304476e-01
5.07299185e-01 3.65559980e-02 5.57833672e-01 -1.16319728e+00
-5.60930550e-01 -8.89396131e-01 -1.87288523e-01 -8.45403671e-01
1.86974660e-03 -9.93698001e-01 4.91146781e-02 -1.40665495e+00
-1.05494395e-01 -5.53354621e-01 -3.09256792e-01 6.65917397e-01
1.95056021e-01 3.76711816e-01 5.35833836e-03 3.74301702e-01
-5.78473210e-01 1.80905491e-01 8.70704234e-01 -9.84826311e-02
-2.78579712e-01 2.24030286e-01 -4.46592689e-01 8.76246929e-01
1.07410562e+00 -5.18030703e-01 -3.69445622e-01 -5.14090836e-01
3.45185757e-01 -2.62235343e-01 4.90102500e-01 -1.39692962e+00
4.26756561e-01 2.79024333e-01 4.17387396e-01 -4.69673038e-01
2.82396197e-01 -8.00441444e-01 1.72026586e-02 7.44686186e-01
-4.94456917e-01 1.32881045e-01 1.86341166e-01 2.29195014e-01
-6.41741678e-02 -6.41040206e-01 8.27454627e-01 -3.72363001e-01
-9.11101997e-01 2.56373763e-01 -1.91194907e-01 -1.33916408e-01
1.06499696e+00 -3.79188657e-01 -1.06299622e-02 1.06683560e-01
-9.48781133e-01 9.10643712e-02 2.78144538e-01 2.20673382e-01
5.81171215e-01 -5.43149054e-01 -6.69927180e-01 2.29889750e-01
-2.68288136e-01 2.93010503e-01 2.19708145e-01 7.03986406e-01
-7.02933490e-01 3.41152996e-01 -3.55935812e-01 -6.29619598e-01
-1.34772849e+00 4.09619778e-01 5.77870250e-01 -5.93117833e-01
-7.28216410e-01 1.07487869e+00 -1.46788746e-01 -6.25056028e-02
4.88603622e-01 -2.44703710e-01 -4.24703985e-01 1.94682583e-01
3.18426102e-01 1.80227041e-01 3.86218935e-01 -4.85224664e-01
-2.58692741e-01 4.85147923e-01 -2.56918311e-01 3.14036272e-02
1.61854136e+00 1.84951738e-01 2.98431353e-03 1.83863506e-01
1.03429353e+00 -4.36015308e-01 -1.56016386e+00 -1.38331831e-01
2.90643156e-01 5.75866960e-02 1.08171836e-01 -8.99861634e-01
-1.33377242e+00 9.18318808e-01 8.72240663e-01 2.66919702e-01
1.32717443e+00 2.43874118e-01 4.29307282e-01 6.13663077e-01
3.28646213e-01 -1.10903096e+00 3.71840186e-02 4.03547376e-01
8.53357852e-01 -1.13785410e+00 -1.66210696e-01 -7.40886629e-02
-4.77356464e-01 1.02370715e+00 5.81665695e-01 -4.24348116e-01
5.13394415e-01 2.24705473e-01 7.23447744e-03 -4.28874254e-01
-8.57136667e-01 -4.12040949e-01 1.92975283e-01 1.64823428e-01
2.08541378e-01 -1.51347443e-01 -9.18019265e-02 8.95959735e-02
-4.15440410e-01 -7.03630596e-03 1.92922130e-01 6.54674768e-01
-5.35130084e-01 -9.86122072e-01 -3.00714350e-03 6.68010533e-01
-5.68894923e-01 -2.27331311e-01 -2.33275995e-01 3.93428713e-01
2.34753519e-01 3.88950944e-01 6.15335926e-02 -2.16532201e-01
4.00640309e-01 1.16122715e-01 2.71376312e-01 -5.55314600e-01
-8.81216288e-01 1.31445721e-01 -5.03945947e-02 -5.98530531e-01
-3.66717517e-01 -7.75122285e-01 -1.17161143e+00 1.38312444e-01
-2.14088127e-01 -5.18322140e-02 8.86406481e-01 8.85136724e-01
3.38597566e-01 7.08231986e-01 3.15567642e-01 -9.99503434e-01
-5.51345408e-01 -8.67109358e-01 -2.12857321e-01 9.84085277e-02
9.36153755e-02 -3.82733166e-01 -1.85423464e-01 4.47075628e-02] | [8.71527099609375, 2.984646797180176] |
35a53318-ab5e-4754-9145-67b50b589372 | utilizing-bert-for-aspect-based-sentiment | 1903.09588 | null | http://arxiv.org/abs/1903.09588v1 | http://arxiv.org/pdf/1903.09588v1.pdf | Utilizing BERT for Aspect-Based Sentiment Analysis via Constructing Auxiliary Sentence | Aspect-based sentiment analysis (ABSA), which aims to identify fine-grained
opinion polarity towards a specific aspect, is a challenging subtask of
sentiment analysis (SA). In this paper, we construct an auxiliary sentence from
the aspect and convert ABSA to a sentence-pair classification task, such as
question answering (QA) and natural language inference (NLI). We fine-tune the
pre-trained model from BERT and achieve new state-of-the-art results on
SentiHood and SemEval-2014 Task 4 datasets. | ['Luyao Huang', 'Xipeng Qiu', 'Chi Sun'] | 2019-03-22 | utilizing-bert-for-aspect-based-sentiment-1 | https://aclanthology.org/N19-1035 | https://aclanthology.org/N19-1035.pdf | naacl-2019-6 | ['sentence-pair-classification'] | ['natural-language-processing'] | [ 3.12369823e-01 1.27445206e-01 1.01714700e-01 -1.08474040e+00
-1.09366274e+00 -7.43414700e-01 7.34413743e-01 4.22707617e-01
-2.98336238e-01 6.05650783e-01 5.01869977e-01 -4.85719860e-01
4.30600494e-01 -9.40617204e-01 -5.82489789e-01 -2.41769224e-01
4.03502345e-01 6.26798987e-01 -1.13505475e-01 -9.43871319e-01
3.27009857e-01 -2.82341868e-01 -1.25509894e+00 9.46327627e-01
8.29025269e-01 1.51318467e+00 -6.51043534e-01 8.18752766e-01
-4.01872993e-01 1.18913674e+00 -7.50570476e-01 -1.36859107e+00
-1.19763307e-01 -3.37182492e-01 -1.23215413e+00 -8.39083567e-02
4.30099726e-01 3.80230874e-01 7.35809088e-01 1.00914025e+00
2.31427446e-01 -9.10114124e-02 6.84600949e-01 -1.23543131e+00
-6.04973495e-01 4.33148175e-01 -5.69761455e-01 1.99380323e-01
6.53444409e-01 -2.70577446e-02 1.80768991e+00 -9.00276780e-01
2.91388899e-01 1.40789390e+00 7.92916536e-01 3.02537024e-01
-8.85091424e-01 -1.75708771e-01 2.83933729e-01 1.68934330e-01
-5.90277433e-01 -2.18646169e-01 1.00675595e+00 -1.85026109e-01
1.36181664e+00 4.94758278e-01 8.67124736e-01 1.00637972e+00
6.12184644e-01 1.20515335e+00 1.45843208e+00 -2.89257705e-01
4.47051525e-01 1.14710495e-01 6.59881175e-01 3.51373106e-01
1.41842961e-01 -6.99327409e-01 -7.73494363e-01 -3.49684536e-01
-5.17149746e-01 -4.16775376e-01 2.72595972e-01 1.17042296e-01
-9.17198300e-01 1.08436596e+00 2.00015128e-01 -1.21227920e-01
-6.25983775e-01 -3.28227282e-01 7.95019925e-01 7.26274252e-01
1.14397025e+00 1.05931544e+00 -1.21561074e+00 -2.08806396e-01
-5.36559284e-01 2.99349278e-01 1.35219073e+00 6.31621659e-01
8.05833459e-01 -1.74072593e-01 -5.38132131e-01 7.05119073e-01
3.36168289e-01 8.17097604e-01 5.15541255e-01 -4.24225181e-01
5.38540006e-01 1.19617236e+00 -2.59720087e-01 -7.83220589e-01
-3.99406075e-01 -2.25885883e-01 -6.48576856e-01 -5.98380119e-02
-1.70108080e-02 -3.44740987e-01 -7.68649101e-01 1.24656117e+00
5.11234522e-01 -3.86227399e-01 3.89745682e-01 6.46955311e-01
1.03118396e+00 6.38967812e-01 1.89061880e-01 2.50955075e-02
2.06480026e+00 -1.36053729e+00 -5.08541286e-01 -9.65990007e-01
5.49844265e-01 -8.41576099e-01 1.42074120e+00 3.97432446e-01
-1.05814004e+00 -2.71358967e-01 -1.00981009e+00 -3.80194873e-01
-7.91266978e-01 -6.67029023e-02 9.50865746e-01 4.94736224e-01
-1.01545155e+00 3.89338620e-02 -4.76171404e-01 4.57039475e-02
6.44350708e-01 2.83220917e-01 -4.36660022e-01 2.24536598e-01
-1.39576900e+00 7.56517649e-01 -2.48520538e-01 2.46233925e-01
-3.35112363e-01 -7.66679943e-01 -1.40377831e+00 -6.88824803e-02
2.14231730e-01 -1.17703855e+00 1.54227698e+00 -1.27915847e+00
-1.81871271e+00 1.44748759e+00 -7.13293791e-01 -5.91543257e-01
-3.07475448e-01 -3.52963626e-01 -4.73624736e-01 -5.05976006e-02
5.31197727e-01 2.58314192e-01 9.81014013e-01 -8.66474330e-01
-6.61869645e-01 -7.78606474e-01 7.64048278e-01 2.75063932e-01
-1.92428216e-01 4.34586465e-01 -4.56042960e-02 -4.03466552e-01
-4.25308615e-01 -7.55167127e-01 -3.92766058e-01 -7.21743345e-01
-6.31170571e-01 -5.39351940e-01 3.14056337e-01 -4.14203972e-01
9.37702239e-01 -1.72875631e+00 -2.82842405e-02 5.61403558e-02
1.07353732e-01 2.03307196e-01 -3.56871635e-01 6.38014972e-02
2.27358546e-02 -9.80125219e-02 -5.13516605e-01 -8.89518917e-01
5.68815880e-02 -8.47704932e-02 -5.46829462e-01 5.57094719e-03
7.44045734e-01 1.41817594e+00 -9.24588919e-01 -4.91577953e-01
-1.56172469e-01 1.64478540e-01 -6.50582850e-01 3.02978575e-01
-5.15168130e-01 1.94071606e-01 -6.10707879e-01 1.09085047e+00
7.68107772e-01 -2.80392855e-01 -4.71052714e-02 -3.74325544e-01
2.20859155e-01 9.94188786e-01 -3.45022082e-01 1.39708698e+00
-1.04388523e+00 5.29193103e-01 5.10344245e-02 -9.30552244e-01
9.42272246e-01 -3.99224088e-03 1.45765785e-02 -9.42643166e-01
5.13549984e-01 -7.24903047e-02 -1.47508159e-01 -3.89601827e-01
8.57964098e-01 -5.66152334e-01 -7.19248056e-01 5.86732984e-01
1.10516891e-01 -7.49262750e-01 3.49364340e-01 1.88981280e-01
9.56971407e-01 -2.85280794e-01 6.59562826e-01 -5.32165349e-01
1.10549366e+00 2.86704719e-01 3.46115917e-01 3.99830878e-01
-2.66753554e-01 5.57206392e-01 9.59680796e-01 -6.23261988e-01
-5.67277491e-01 -8.51232409e-01 -2.52912566e-02 1.42413592e+00
-1.62389241e-02 -7.60371089e-01 -5.89623213e-01 -1.23281276e+00
-1.15013659e-01 9.43891823e-01 -1.04009557e+00 -5.85591346e-02
-3.99161428e-01 -1.06506813e+00 1.87051259e-02 4.99434948e-01
4.54097360e-01 -1.35832214e+00 -2.37966388e-01 -8.32964480e-02
-3.35642040e-01 -1.34474862e+00 -5.06409466e-01 1.82721153e-01
-5.86358249e-01 -9.91265118e-01 -1.07757710e-01 -8.98318768e-01
6.34361804e-01 1.87371485e-02 2.17019916e+00 -4.39803958e-01
3.87626857e-01 9.64104682e-02 -5.53214729e-01 -1.02072763e+00
-2.05635369e-01 4.02908653e-01 -2.80383170e-01 2.78757691e-01
1.13631988e+00 -3.18782985e-01 -6.06199026e-01 6.00549020e-03
-7.00220764e-01 -2.06861854e-01 4.49000567e-01 7.02463508e-01
9.27799106e-01 -3.96523714e-01 6.80667102e-01 -1.62032676e+00
1.17185414e+00 -3.29554379e-01 -2.95528054e-01 2.10693106e-01
-5.85380137e-01 -7.25794807e-02 1.01081824e+00 2.46768191e-01
-1.03783166e+00 -2.47166142e-01 -6.93151534e-01 3.43737483e-01
-1.17316851e-02 8.33675921e-01 -2.32863650e-01 2.71749794e-01
4.75266933e-01 7.10882396e-02 -1.61324412e-01 -5.84398918e-02
3.58942270e-01 8.53828430e-01 2.62985408e-01 -2.55322695e-01
3.79603803e-01 7.30254412e-01 -4.36957926e-02 -4.19996023e-01
-2.35060620e+00 -7.17583477e-01 -3.55735868e-01 7.89879486e-02
8.78028750e-01 -1.13653159e+00 -9.42762554e-01 5.36130548e-01
-1.10411978e+00 -3.64145972e-02 -5.52091360e-01 -1.99725300e-01
-4.46717024e-01 -1.10160550e-02 -5.57877243e-01 -7.01396823e-01
-1.40806544e+00 -1.03558719e+00 1.75493217e+00 2.51149267e-01
-5.97219408e-01 -1.12455833e+00 6.44091487e-01 1.05066419e+00
4.13721830e-01 -9.84016210e-02 6.26451612e-01 -8.31129730e-01
1.31648868e-01 -5.09487391e-01 5.72446957e-02 7.82960296e-01
-1.56519618e-02 -2.39672542e-01 -1.23573279e+00 3.26761603e-02
4.73054677e-01 -7.18456864e-01 8.61235499e-01 3.01818758e-01
9.19330418e-01 -3.61282438e-01 3.34468633e-01 3.87147397e-01
9.99665976e-01 -3.80323887e-01 5.20282149e-01 6.88015938e-01
5.98253489e-01 6.58518314e-01 1.12997723e+00 1.19154640e-01
1.16086888e+00 1.91347778e-01 2.92970896e-01 1.17883878e-02
3.25380057e-01 -3.87652707e-03 6.50245249e-01 1.13484156e+00
2.14485228e-01 -1.08897887e-01 -6.39741182e-01 7.31790841e-01
-1.68127036e+00 -6.32629514e-01 -3.58311564e-01 1.45379114e+00
1.09132540e+00 4.54783827e-01 -5.23481704e-02 6.17867112e-02
1.54293150e-01 7.09616959e-01 -4.81662780e-01 -1.23769915e+00
-5.95014751e-01 5.05208075e-01 -3.06876171e-02 6.44723117e-01
-1.44619381e+00 1.20034146e+00 6.13889694e+00 5.53684175e-01
-9.24731314e-01 1.67433649e-01 1.17501855e+00 5.80684915e-02
-7.43975461e-01 1.79231204e-02 -6.27646208e-01 2.67471492e-01
1.04223287e+00 -1.32919759e-01 1.37101173e-01 1.08955896e+00
-2.01338679e-01 -2.66743004e-01 -9.66296494e-01 4.92397815e-01
5.75639248e-01 -1.15453756e+00 3.67461294e-01 -5.65300941e-01
9.09897089e-01 2.66865820e-01 5.90327904e-02 7.50717402e-01
1.85910180e-01 -8.06629360e-01 5.22821605e-01 3.48990381e-01
4.73649114e-01 -7.21357644e-01 1.40671468e+00 -6.72297180e-02
-1.12117720e+00 2.30220988e-01 -2.34609440e-01 -4.40974146e-01
4.83265042e-01 1.07435191e+00 -4.27521646e-01 4.20472294e-01
8.76681685e-01 1.01997364e+00 -7.49530315e-01 2.41351351e-01
-7.59586930e-01 9.16512311e-01 1.19986415e-01 -4.90699202e-01
2.56596595e-01 -3.87334019e-01 6.49802744e-01 1.04610109e+00
-3.85817349e-01 -1.39445961e-01 -3.75712156e-01 4.65567142e-01
-5.31085908e-01 1.97146401e-01 -3.55963916e-01 -1.23297356e-01
-3.29953969e-01 1.86270976e+00 -2.73141831e-01 -5.70103049e-01
-5.55621088e-01 1.10937357e+00 4.98039901e-01 -4.08436842e-02
-3.84140104e-01 -5.34821570e-01 9.21740592e-01 -2.42971659e-01
3.30541939e-01 4.89181310e-01 -8.78050864e-01 -1.57680869e+00
3.32284421e-01 -1.21198130e+00 5.37176609e-01 -9.19073284e-01
-1.83838296e+00 9.66289401e-01 -6.61127210e-01 -1.07008660e+00
-4.22583610e-01 -9.09487009e-01 -1.18763554e+00 8.91709864e-01
-1.97736669e+00 -1.41516542e+00 -4.67712320e-02 4.00117546e-01
6.13584995e-01 -1.18449762e-01 9.08606470e-01 -1.24035887e-01
-2.53529787e-01 5.69191217e-01 -4.64148343e-01 -6.58059716e-02
7.12801635e-01 -1.65938663e+00 1.05066228e+00 5.80458641e-01
-6.35144785e-02 5.04077971e-01 7.53247738e-01 -4.10692662e-01
-1.55445552e+00 -1.13672268e+00 1.67938221e+00 -1.24061036e+00
1.24358082e+00 -6.47256076e-01 -5.42634487e-01 6.00881517e-01
4.94758606e-01 -2.23603234e-01 1.05150139e+00 7.43941844e-01
-4.86997545e-01 -3.49808127e-01 -9.44505036e-01 4.66608495e-01
4.40315783e-01 -9.46993172e-01 -1.01796162e+00 2.72580206e-01
1.00706434e+00 -2.80876428e-01 -7.76557148e-01 6.07810855e-01
4.04531956e-01 -9.86789584e-01 8.84354115e-01 -1.04506552e+00
1.11434221e+00 -1.11023366e-01 -5.74272387e-02 -1.67514944e+00
1.78128421e-01 -3.45801234e-01 -3.52904536e-02 1.11299014e+00
1.04322290e+00 -7.09663272e-01 7.80647874e-01 5.83415210e-01
-1.62248388e-01 -1.29037368e+00 -7.35529840e-01 -5.58059141e-02
9.67228934e-02 -6.56038463e-01 8.52009952e-01 6.68019295e-01
4.19924587e-01 1.45911443e+00 4.99643907e-02 -5.43212518e-02
2.37375230e-01 9.35642421e-01 7.04804003e-01 -1.06916344e+00
-3.27192932e-01 -3.70466262e-01 -1.83323681e-01 -8.50359976e-01
5.40472209e-01 -6.80616975e-01 1.71382084e-01 -1.68860722e+00
2.95787573e-01 2.20086277e-01 -3.17473888e-01 2.89039373e-01
-8.59145880e-01 5.17946482e-01 2.37373002e-02 -3.67720842e-01
-1.30006409e+00 9.23848987e-01 1.42678511e+00 -5.69580317e-01
1.27393842e-01 5.14766574e-01 -1.40393329e+00 9.13694143e-01
7.70041168e-01 -3.77983838e-01 -2.02289253e-01 -4.00492609e-01
1.25999546e+00 -2.82139271e-01 -9.43011791e-02 -2.10829571e-01
-1.79345775e-02 1.16435647e-01 -5.30340038e-02 -8.50233734e-01
5.57311714e-01 -3.75760376e-01 -1.11559224e+00 9.22109932e-02
-3.81438583e-01 3.62017244e-01 2.05154791e-01 3.34660977e-01
-8.67230415e-01 -9.47253481e-02 2.20969737e-01 -5.66084497e-02
-2.75107205e-01 1.78335905e-01 -3.14208448e-01 7.79141605e-01
5.34325004e-01 3.07032138e-01 -6.23415112e-01 -3.91615987e-01
-2.14074120e-01 5.63778281e-01 7.54781067e-02 3.63518089e-01
4.23548877e-01 -8.69221210e-01 -8.56429279e-01 5.30010946e-02
5.67338765e-01 3.29351246e-01 2.97885597e-01 8.96898448e-01
-2.39256620e-01 4.30162132e-01 2.06806213e-01 -2.60283381e-01
-1.02906621e+00 2.41238281e-01 5.79651706e-02 -7.85799265e-01
7.47568905e-02 1.08605087e+00 2.47947022e-01 -1.16454983e+00
-5.20739198e-01 -2.23042578e-01 -8.36923420e-01 2.94658333e-01
6.53300822e-01 -8.38430151e-02 4.13240433e-01 -7.54840791e-01
-7.10590541e-01 5.85734725e-01 -2.42059961e-01 8.57143104e-02
1.34594321e+00 -9.58466828e-02 -8.53376448e-01 5.98207295e-01
1.35387731e+00 2.38167256e-01 -5.05153000e-01 4.96381195e-04
-1.64943308e-01 -9.61074159e-02 5.37816575e-03 -1.04986656e+00
-7.52933681e-01 8.90046775e-01 -2.97365278e-01 5.30081809e-01
1.22558200e+00 3.55749696e-01 1.07655382e+00 6.88348591e-01
3.59736234e-02 -1.09962547e+00 -7.72391632e-02 1.16985631e+00
9.15253997e-01 -1.82550240e+00 -2.18008440e-02 -3.38686705e-01
-1.34528160e+00 6.80895507e-01 7.09970653e-01 -2.47580886e-01
7.89693534e-01 1.43193290e-01 6.17665231e-01 -7.16218591e-01
-1.09432042e+00 -2.76434630e-01 6.27558649e-01 2.22008333e-01
5.49711645e-01 3.09379250e-01 -3.61525208e-01 1.37048602e+00
-1.02036548e+00 -3.42148185e-01 3.50861341e-01 7.93601751e-01
4.85003889e-02 -8.59856188e-01 1.06349677e-01 8.22920680e-01
-7.66444564e-01 -7.01591969e-01 -8.56873989e-01 1.42176822e-01
-4.17086542e-01 1.36261380e+00 9.97524783e-02 -4.98609096e-01
5.90075970e-01 -6.71948195e-02 -2.32825223e-02 -7.19649613e-01
-1.28850925e+00 -5.69813192e-01 8.01792920e-01 -7.71373570e-01
-7.66754866e-01 -6.20311916e-01 -8.60032737e-01 -7.54042193e-02
-1.73840359e-01 3.96158516e-01 7.89941370e-01 1.38530421e+00
6.06620550e-01 5.31013131e-01 9.33047473e-01 -3.20639968e-01
-3.10766876e-01 -1.11572671e+00 -2.68350869e-01 6.10968292e-01
4.27708596e-01 8.75317901e-02 -5.60759008e-01 -2.52016068e-01] | [11.483524322509766, 6.675646781921387] |
f21d3853-a860-4b0a-8c91-f167dde9e211 | sources-of-hallucination-by-large-language | 2305.14552 | null | https://arxiv.org/abs/2305.14552v1 | https://arxiv.org/pdf/2305.14552v1.pdf | Sources of Hallucination by Large Language Models on Inference Tasks | Large Language Models (LLMs) are claimed to be capable of Natural Language Inference (NLI), necessary for applied tasks like question answering and summarization, yet this capability is under-explored. We present a series of behavioral studies on several LLM families (LLaMA, GPT-3.5, and PaLM) which probe their behavior using controlled experiments. We establish two factors which predict much of their performance, and propose that these are major sources of hallucination in generative LLM. First, the most influential factor is memorization of the training data. We show that models falsely label NLI test samples as entailing when the hypothesis is attested in the training text, regardless of the premise. We further show that named entity IDs are used as "indices" to access the memorized data. Second, we show that LLMs exploit a further corpus-based heuristic using the relative frequencies of words. We show that LLMs score significantly worse on NLI test samples which do not conform to these factors than those which do; we also discuss a tension between the two factors, and a performance trade-off. | ['Mark Steedman', 'Mark Johnson', 'Mohammad Javad Hosseini', 'Liang Cheng', 'Tianyi Li', 'Nick McKenna'] | 2023-05-23 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [ 1.99809462e-01 6.18377268e-01 -1.97402105e-01 -1.90113351e-01
-8.53055179e-01 -6.53100908e-01 9.37068641e-01 3.83119822e-01
-4.60459471e-01 8.97658408e-01 5.12763917e-01 -7.57108331e-01
-1.19795784e-01 -8.17174971e-01 -7.00213492e-01 -3.56740147e-01
6.71502799e-02 7.12898195e-01 1.91689134e-01 -1.61635086e-01
8.05745006e-01 2.57487923e-01 -1.39699042e+00 4.23597097e-01
1.11160076e+00 2.11256504e-01 1.37811556e-01 7.56967843e-01
-1.91757634e-01 1.16817224e+00 -1.17276037e+00 -6.66261375e-01
-3.49292368e-01 -7.39518523e-01 -1.19445133e+00 1.90872867e-02
4.04316574e-01 -1.69180915e-01 -1.48663953e-01 8.14260364e-01
5.49857199e-01 8.42183530e-02 9.35268223e-01 -1.13904345e+00
-6.08522117e-01 8.83603215e-01 -1.91484675e-01 4.42051858e-01
5.66248894e-01 1.91741198e-01 1.07916248e+00 -7.42168188e-01
7.76848018e-01 1.41619658e+00 6.11267626e-01 6.08039796e-01
-1.12291837e+00 -3.18330586e-01 -1.34115800e-01 1.37646481e-01
-1.35074770e+00 -8.57338428e-01 2.85624415e-01 -4.44933362e-02
1.33419192e+00 5.30296206e-01 5.30556023e-01 1.23189127e+00
5.87990224e-01 8.16946745e-01 1.12835586e+00 -5.85966527e-01
4.63004112e-01 4.79364127e-01 4.11506057e-01 6.83913529e-01
7.48624563e-01 -2.41665140e-01 -8.64881039e-01 -5.21997631e-01
3.66905659e-01 -7.48260915e-01 -2.90354490e-01 1.46946505e-01
-1.00475860e+00 1.12417877e+00 -8.62129256e-02 5.50925016e-01
-2.97937930e-01 -6.94551691e-02 -2.26876251e-02 3.39732796e-01
2.64457464e-01 9.66346860e-01 -2.57003486e-01 -2.46398062e-01
-9.87421453e-01 3.78055960e-01 1.32931423e+00 8.51047456e-01
3.49978864e-01 -7.08678737e-02 -1.96185589e-01 7.30132937e-01
3.76030326e-01 5.36581457e-01 1.04327536e+00 -7.49203503e-01
5.29657662e-01 3.80649209e-01 6.79638283e-03 -9.06671941e-01
-3.70827347e-01 -5.11040747e-01 -4.16099429e-01 -3.79779786e-01
3.65061402e-01 7.58409593e-03 -4.88691628e-01 2.01313877e+00
-2.13452861e-01 -3.52743149e-01 2.33810291e-01 1.98662996e-01
6.49571359e-01 7.03287303e-01 2.07394496e-01 -5.35417438e-01
1.16981792e+00 -6.41125679e-01 -5.73614955e-01 -6.41162455e-01
1.01749754e+00 -4.93365020e-01 1.10209846e+00 2.29427159e-01
-1.48078620e+00 -2.25299791e-01 -1.10807216e+00 6.44764081e-02
-3.14672083e-01 -2.60654151e-01 7.74240613e-01 7.97033727e-01
-1.13636506e+00 4.76085275e-01 -4.70985740e-01 -7.14340270e-01
2.42753103e-01 7.71079883e-02 -6.77008033e-02 2.05016240e-01
-1.27119589e+00 1.31918263e+00 3.88179243e-01 -6.33184314e-02
-7.91324556e-01 -1.10212013e-01 -6.84584022e-01 1.07024200e-01
7.01680556e-02 -8.07511151e-01 1.30372417e+00 -5.15212834e-01
-9.98091638e-01 1.00625050e+00 -5.91048479e-01 -6.73972249e-01
3.36464822e-01 -1.02554053e-01 -5.30928791e-01 2.62790203e-01
1.90607458e-01 5.39428294e-01 8.03500295e-01 -1.39566720e+00
-2.58639425e-01 -2.78003484e-01 -2.16085032e-01 1.83104724e-01
-2.14303628e-01 -8.55686888e-02 1.08539879e-01 -3.26853186e-01
9.42242369e-02 -6.88465476e-01 1.28554612e-01 -7.95418680e-01
-8.72206032e-01 -5.73025703e-01 3.40985239e-01 -4.52756643e-01
1.72680616e+00 -1.84371138e+00 -2.75651723e-01 3.11839402e-01
3.16451341e-01 3.85152012e-01 -1.30765721e-01 7.26607919e-01
1.12604767e-01 7.07513809e-01 -7.33956918e-02 -2.74454743e-01
1.50359884e-01 3.18231761e-01 -7.28521228e-01 2.32900172e-01
1.43854648e-01 1.26543212e+00 -6.22714460e-01 -6.42486453e-01
-2.50986904e-01 -4.03237045e-02 -5.45050621e-01 1.02039076e-01
-3.55128437e-01 -6.47878572e-02 -3.72182846e-01 3.93620044e-01
2.33567551e-01 -5.21791577e-01 2.00496957e-01 1.27502441e-01
-2.01912560e-02 7.82044709e-01 -7.30195999e-01 1.11742294e+00
-1.03053331e-01 6.59233451e-01 -3.41519266e-01 -8.08970034e-01
4.43198234e-01 2.69826800e-01 -4.40401167e-01 -6.71269238e-01
5.89352325e-02 3.45629007e-01 3.75995338e-01 -7.52087116e-01
7.06336379e-01 -2.48362929e-01 -1.24167085e-01 8.48067880e-01
3.31272297e-02 -2.86797941e-01 3.46740395e-01 7.02723145e-01
1.12605524e+00 -3.86944175e-01 6.46503448e-01 -4.72873271e-01
2.29216576e-01 9.35245678e-02 3.05121690e-01 1.52305806e+00
5.32638058e-02 1.58046037e-01 7.89633691e-01 1.18759312e-02
-9.52894032e-01 -1.05766642e+00 -3.18881452e-01 1.04554272e+00
-8.79364982e-02 -6.03330791e-01 -9.55148757e-01 -5.04472196e-01
-1.89844280e-01 1.76857150e+00 -5.46964526e-01 -5.23261368e-01
-3.84125143e-01 -9.05436933e-01 8.99106801e-01 3.19245726e-01
3.11546296e-01 -1.35229993e+00 -7.87588239e-01 1.06816933e-01
-3.61447394e-01 -7.61961877e-01 -5.07338159e-02 2.46549100e-01
-9.77351189e-01 -6.56165123e-01 -1.30173385e-01 -4.18935061e-01
7.87608445e-01 1.10979438e-01 1.38200843e+00 3.99250895e-01
4.17297147e-02 5.39647579e-01 -2.55209625e-01 -4.30824310e-01
-1.01767051e+00 3.12330902e-01 -5.77683235e-03 -6.25057995e-01
4.84766692e-01 -6.16657913e-01 -1.53252900e-01 1.55874506e-01
-1.08823133e+00 1.11618094e-01 9.67728376e-01 6.81685388e-01
2.57909335e-02 -1.32887080e-01 9.12025690e-01 -1.21929181e+00
1.17648411e+00 -6.86399817e-01 -3.79075743e-02 6.01916313e-01
-6.66978061e-01 3.66399527e-01 2.55274594e-01 -4.81050640e-01
-1.02900028e+00 -9.14015174e-01 -2.86040753e-01 4.12760705e-01
-1.88801199e-01 7.16870248e-01 -2.57840931e-01 2.06889778e-01
8.48855495e-01 5.45723736e-01 -5.75037673e-02 -2.06656024e-01
1.40617475e-01 5.81796587e-01 3.02129000e-01 -7.02157915e-01
6.23055577e-01 4.86767180e-02 -3.91770750e-01 -1.18884623e+00
-1.09076881e+00 4.54757316e-03 -6.33138046e-02 3.67953032e-02
5.88678598e-01 -5.06372988e-01 -6.56044483e-01 1.85948521e-01
-1.22284126e+00 -3.59022379e-01 -1.84501946e-01 2.98389912e-01
-6.29440665e-01 4.97059792e-01 -7.11298108e-01 -9.70797300e-01
-4.21642423e-01 -8.12353551e-01 7.64959455e-01 2.90186256e-01
-9.37027395e-01 -1.03995025e+00 3.10557913e-02 4.22848105e-01
5.56308091e-01 -2.86811769e-01 1.71626580e+00 -1.45038176e+00
-2.73065537e-01 -1.82774499e-01 9.05666053e-02 8.23318213e-02
-3.93555135e-01 -1.23680748e-01 -9.65865552e-01 -2.94389009e-01
5.20537674e-01 -5.61460137e-01 8.35687876e-01 1.95088267e-01
8.89301836e-01 -7.88984954e-01 -4.59617853e-01 9.71467942e-02
1.11111200e+00 5.17118499e-02 8.96185994e-01 1.04542896e-01
2.07549825e-01 7.01719344e-01 8.00062642e-02 2.17142582e-01
3.68797362e-01 1.88513741e-01 -5.10260649e-02 5.03978670e-01
9.92698148e-02 -7.40464509e-01 6.38816178e-01 1.13245034e+00
2.65192747e-01 -8.04013193e-01 -7.54871070e-01 5.55990517e-01
-1.54091835e+00 -1.14807796e+00 -1.21528856e-01 2.32691121e+00
1.00311887e+00 5.09218812e-01 -2.53332376e-01 5.86044528e-02
4.20692742e-01 2.15762705e-01 -5.25649428e-01 -4.83868390e-01
-4.50822294e-01 3.27802241e-01 2.26545140e-01 7.99252570e-01
-4.11027133e-01 9.25889432e-01 7.44217157e+00 9.45798099e-01
-6.31940246e-01 7.21567497e-02 7.09349930e-01 6.83453158e-02
-7.60128021e-01 1.14937037e-01 -8.61243963e-01 3.32682788e-01
1.50527918e+00 -4.23066258e-01 1.72304809e-01 4.67104882e-01
-9.54899043e-02 -6.92106009e-01 -1.32547879e+00 4.20306623e-01
4.51692075e-01 -1.16265845e+00 5.97622156e-01 3.15923154e-01
5.59096158e-01 -1.75385684e-01 7.22877160e-02 5.50047934e-01
1.29060030e-01 -1.14778662e+00 8.28307390e-01 5.29451966e-01
2.29384005e-01 -6.16271079e-01 6.07869506e-01 9.06199038e-01
-2.97389477e-01 1.01601474e-01 -5.27914584e-01 -1.10744163e-01
1.12372436e-01 4.77504283e-01 -9.83323693e-01 9.87472534e-02
-6.53682351e-02 -1.63566336e-01 -1.08134770e+00 6.79798186e-01
-3.82613748e-01 1.06568611e+00 -1.36283115e-01 -3.38615477e-01
2.61257123e-02 1.71438396e-01 6.77609384e-01 1.08240604e+00
1.96204513e-01 -5.01279384e-02 -2.43263334e-01 1.07815325e+00
-4.50054109e-02 8.78555551e-02 -6.33298516e-01 -4.97934252e-01
6.72178924e-01 8.36590171e-01 -6.25411093e-01 -6.45401180e-01
-1.22395292e-01 9.42818582e-01 4.03461307e-01 2.50742376e-01
-6.65270984e-01 -2.08539814e-01 1.68621819e-02 7.91632831e-02
-6.68058172e-02 -1.23768903e-01 -2.53917515e-01 -1.28181517e+00
-1.85807884e-01 -9.85155404e-01 2.09632769e-01 -9.91940141e-01
-1.22089159e+00 5.22459209e-01 1.85220808e-01 -4.74603415e-01
-8.08355093e-01 -4.43194598e-01 -6.32694423e-01 7.32915401e-01
-1.03694236e+00 -7.19638586e-01 1.96842745e-01 1.63053766e-01
4.98871177e-01 -4.83138487e-02 1.12478709e+00 -1.86476067e-01
-4.98372853e-01 6.27323925e-01 -1.64340064e-01 2.46668104e-02
4.99270946e-01 -1.01242733e+00 5.58874965e-01 7.44848132e-01
5.20247400e-01 1.36673009e+00 8.48214090e-01 -7.42389023e-01
-1.23915505e+00 -6.77682579e-01 1.50383317e+00 -8.77355158e-01
4.88869667e-01 -3.44355822e-01 -9.99319792e-01 8.61693263e-01
4.79902923e-01 -8.18166196e-01 1.02839530e+00 9.51185822e-02
-3.09614569e-01 5.03053367e-01 -1.06081605e+00 7.27990448e-01
9.37633455e-01 -5.14308989e-01 -1.24063945e+00 4.08243209e-01
6.92895591e-01 6.08504042e-02 -4.22580689e-01 6.85820803e-02
3.12501609e-01 -1.17248905e+00 6.76762760e-01 -9.22831297e-01
4.94480908e-01 5.32793738e-02 -1.18843012e-01 -1.32946932e+00
-1.41214117e-01 -5.52107751e-01 -1.98634684e-01 1.21253109e+00
8.67401242e-01 -9.98654068e-01 3.41164678e-01 6.70383930e-01
4.80073057e-02 -6.64495945e-01 -7.49611199e-01 -6.26273274e-01
2.30609670e-01 -4.67838615e-01 1.62598819e-01 8.60308111e-01
3.89468610e-01 8.87528181e-01 -7.25863948e-02 -2.07183585e-01
4.54475731e-01 -2.32918002e-02 5.12173533e-01 -1.19232786e+00
-4.26898867e-01 -5.08162618e-01 -9.32051316e-02 -1.08647621e+00
1.96023479e-01 -8.77230823e-01 1.31340874e-02 -1.38945496e+00
5.04208684e-01 -1.86885297e-02 2.28552207e-01 2.92608589e-01
-2.30216473e-01 -1.42079517e-01 -5.32144196e-02 3.68844271e-01
-7.37175584e-01 2.47387543e-01 8.18058193e-01 3.24073493e-01
3.88076622e-03 -6.72757402e-02 -1.18822598e+00 8.96453500e-01
9.21693087e-01 -3.52613032e-01 -5.44400871e-01 -3.87776524e-01
6.34208322e-01 1.25648409e-01 3.52357030e-01 -9.14595842e-01
3.42213303e-01 -2.96146586e-03 5.19256413e-01 -5.02217054e-01
2.10301548e-01 -1.29215971e-01 -1.23875715e-01 4.51804250e-01
-8.76254141e-01 2.23859563e-01 1.56133130e-01 3.86494309e-01
1.03101663e-01 -7.76249766e-01 5.84127605e-01 -3.61033946e-01
-3.97621781e-01 -3.06255966e-01 -8.80377710e-01 5.46155393e-01
6.07939065e-01 -6.69659069e-03 -6.24121785e-01 -7.17419624e-01
-4.80182171e-01 -8.27208161e-02 2.92357355e-01 1.40661359e-01
5.39186478e-01 -8.84318650e-01 -7.74214208e-01 1.92491829e-01
1.07354008e-01 -4.47635621e-01 -2.01519877e-02 9.39060926e-01
-2.99775690e-01 7.51860917e-01 2.60654509e-01 -1.91199690e-01
-8.14022303e-01 4.48804885e-01 1.20354883e-01 -2.67229646e-01
-2.18100131e-01 8.63961279e-01 1.15711570e-01 -1.77430719e-01
-8.10660142e-03 -2.52607763e-01 -4.90488391e-03 1.71867400e-01
6.30147636e-01 2.05578953e-01 2.29433961e-02 -4.08208787e-01
-2.12764859e-01 -1.57118425e-01 -3.78677309e-01 -3.75506371e-01
1.03785312e+00 -9.87178087e-02 -4.38038379e-01 6.80498421e-01
9.71296966e-01 4.31121022e-01 -1.83799908e-01 -1.81890428e-01
2.34180972e-01 -2.20047116e-01 -3.79277527e-01 -8.61032963e-01
-1.17434844e-01 7.25906193e-01 -2.41959572e-01 5.82202256e-01
4.86919433e-01 2.61326760e-01 5.38671613e-01 7.98454583e-01
3.45245868e-01 -1.04635286e+00 4.86503094e-02 4.39726710e-01
8.11207533e-01 -7.17003107e-01 -2.30209753e-02 -1.32927284e-01
-5.37674427e-01 7.13802576e-01 4.93762732e-01 2.23980367e-01
2.43788362e-02 1.42815366e-01 -3.36510897e-01 -3.21100771e-01
-1.21751678e+00 6.10285215e-02 -6.60439208e-02 3.41464043e-01
6.30829990e-01 1.24544371e-02 -7.13464379e-01 6.49008274e-01
-7.21709967e-01 -4.57479626e-01 6.67942524e-01 8.87500405e-01
-9.47714150e-01 -7.57679343e-01 -3.71496856e-01 8.68007481e-01
-4.39805657e-01 -5.08257091e-01 -7.64244199e-01 7.97394872e-01
-2.63938934e-01 1.26270115e+00 3.51804425e-03 -1.60109997e-01
-6.16438910e-02 7.12374091e-01 6.26552403e-01 -7.33590007e-01
-4.44712162e-01 -2.66191751e-01 4.49521959e-01 -2.76374668e-01
-8.50264356e-02 -6.40345693e-01 -1.04797471e+00 -6.00741446e-01
-5.04111230e-01 6.03234053e-01 3.43477875e-01 1.21059346e+00
3.31112713e-01 1.18295945e-01 2.41073012e-01 -2.30104089e-01
-1.01654887e+00 -1.30874598e+00 -5.09064138e-01 3.80722255e-01
-1.08527578e-01 -2.57315069e-01 -6.70844615e-01 -1.63261473e-01] | [11.663224220275879, 9.09545612335205] |
8e4defcd-3598-4e68-be43-8bc44779127d | self-paced-kernel-estimation-for-robust-blind | null | null | http://openaccess.thecvf.com/content_iccv_2017/html/Gong_Self-Paced_Kernel_Estimation_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Gong_Self-Paced_Kernel_Estimation_ICCV_2017_paper.pdf | Self-Paced Kernel Estimation for Robust Blind Image Deblurring | The challenge in blind image deblurring is to remove the effects of blur with limited prior information about the nature of the blur process. Existing methods often assume that the blur image is produced by linear convolution with additive Gaussian noise. However, including even a small number of outliers to this model in the kernel estimation process can significantly reduce the resulting image quality. Previous methods mainly rely on some simple but unreliable heuristics to identify outliers for kernel estimation. Rather than attempt to identify outliers to the model a priori, we instead propose to sequentially identify inliers, and gradually incorporate them into the estimation process. The self-paced kernel estimation scheme we propose represents a generalization of existing self-paced learning approaches, in which we gradually detect and include reliable inlier pixel sets in a blurred image for kernel estimation. Moreover, we automatically activate a subset of significant gradients w.r.t. the reliable inlier pixels, and then update the intermediate sharp image and the kernel accordingly. Experiments on both synthetic data and real-world images with various kinds of outliers demonstrate the effectiveness and robustness of the proposed method compared to the state-of-the-art methods.
| ['Anton Van Den Hengel', 'Yanning Zhang', 'Dong Gong', 'Qinfeng Shi', 'Mingkui Tan'] | 2017-10-01 | null | null | null | iccv-2017-10 | ['blind-image-deblurring'] | ['computer-vision'] | [ 1.81880016e-02 -5.82011104e-01 2.38765091e-01 -1.57950267e-01
-6.64191127e-01 -4.89597023e-01 2.96768278e-01 -1.74194336e-01
-4.20714408e-01 6.45779073e-01 3.13248008e-01 1.95479169e-01
-2.98774481e-01 -4.52053500e-03 -6.18478417e-01 -9.08640623e-01
-3.86184896e-03 -2.70010028e-02 1.75238878e-01 3.64384472e-01
6.18028879e-01 3.63418132e-01 -1.27421677e+00 -6.09790497e-02
1.49728775e+00 8.27357590e-01 3.57457310e-01 6.46155357e-01
2.98072457e-01 1.03715169e+00 -4.94173497e-01 -4.18577231e-02
4.05075610e-01 -4.75223333e-01 -4.00410056e-01 6.49859250e-01
5.74501514e-01 -5.88906825e-01 -5.36001682e-01 1.58655620e+00
4.60905433e-01 2.62523204e-01 5.10391355e-01 -8.47920716e-01
-7.20597923e-01 1.82621822e-01 -9.61132944e-01 4.17748898e-01
2.40956903e-01 2.10700467e-01 3.17824453e-01 -1.37874568e+00
2.67591506e-01 8.24431062e-01 8.49524617e-01 5.39505258e-02
-1.34242308e+00 -3.57298732e-01 4.12926823e-02 4.55530971e-01
-1.59584486e+00 -8.99501920e-01 9.63165402e-01 -4.91722196e-01
1.92057297e-01 2.95537531e-01 3.87696952e-01 5.96458197e-01
-7.38742063e-03 7.00534284e-01 1.35916948e+00 -3.91966552e-01
2.92496711e-01 1.04482293e-01 2.06613332e-01 6.71926379e-01
4.37324941e-01 7.24477544e-02 -5.38192987e-01 -5.05903542e-01
1.06086600e+00 4.28566225e-02 -9.62873280e-01 -7.05090165e-01
-1.50355065e+00 4.16230768e-01 4.16219860e-01 1.25342026e-01
-5.92978179e-01 1.09300688e-01 1.15861423e-01 1.76251754e-01
7.23920166e-01 5.50047755e-01 -3.26997697e-01 -1.04043938e-01
-1.42598832e+00 1.15171382e-02 6.09999359e-01 6.61004961e-01
9.33552742e-01 5.66216037e-02 -2.33419627e-01 1.00770879e+00
1.28503680e-01 2.05826372e-01 5.68321764e-01 -9.80916977e-01
9.19632837e-02 2.47551322e-01 7.08082616e-01 -8.36790860e-01
1.30413532e-01 -4.28253055e-01 -8.69613171e-01 2.91468203e-01
7.02719271e-01 -2.38764688e-01 -1.14343977e+00 1.19669402e+00
2.41160706e-01 8.95337522e-01 -4.36614864e-02 1.32832944e+00
2.28178039e-01 4.49983060e-01 -6.51816547e-01 -4.75470215e-01
9.19475496e-01 -1.08694971e+00 -7.56430447e-01 -2.63965875e-01
1.58218026e-01 -1.34192848e+00 7.63889551e-01 4.59378302e-01
-1.05880928e+00 -6.23784781e-01 -8.86211634e-01 1.94190860e-01
1.72148481e-01 6.41710997e-01 4.39539433e-01 3.82907778e-01
-1.04525673e+00 6.87851191e-01 -7.32342184e-01 -1.73632354e-01
2.78321266e-01 4.85944860e-02 -2.03647912e-01 -3.22995245e-01
-7.85144925e-01 9.53320920e-01 3.56371194e-01 5.27791083e-01
-8.13540101e-01 -6.04358375e-01 -7.82837331e-01 -1.66946188e-01
4.24847037e-01 -7.05438375e-01 1.22946405e+00 -1.25483048e+00
-1.32659602e+00 4.89285827e-01 -5.91930270e-01 -3.80990028e-01
8.31299245e-01 -7.79549122e-01 -1.79527998e-01 2.58707553e-01
2.43019257e-02 2.98661202e-01 1.76840818e+00 -1.52781975e+00
-4.41865712e-01 -2.17218976e-03 -2.80149639e-01 3.90644044e-01
-2.07146585e-01 1.02036670e-01 -7.01869190e-01 -9.84579265e-01
4.57396150e-01 -8.81827116e-01 -3.39437634e-01 7.80630708e-02
-2.80633122e-01 2.12136000e-01 8.88760030e-01 -1.00927925e+00
1.13975155e+00 -2.33021235e+00 3.54092643e-02 1.27301902e-01
3.27135205e-01 1.17859311e-01 1.23601742e-01 5.67538142e-02
-1.92549482e-01 -5.75446486e-01 -2.95354873e-01 -3.00801277e-01
-3.62059921e-01 -1.78876266e-01 -4.00970072e-01 1.04837811e+00
-8.89499485e-02 4.81681079e-01 -1.09892428e+00 -3.18012208e-01
5.34387231e-01 3.72480750e-01 -2.54456580e-01 3.63131076e-01
1.51864409e-01 4.47056651e-01 -1.73701614e-01 5.89247406e-01
1.00126791e+00 -2.62702584e-01 -3.64996463e-01 -5.20262063e-01
-2.14907989e-01 -1.21812083e-01 -1.40655887e+00 1.50962055e+00
-2.30707690e-01 7.63601959e-01 2.36249804e-01 -6.16799831e-01
5.76756179e-01 2.92364419e-01 4.36442971e-01 1.86211973e-01
-9.31175798e-02 3.78770620e-01 9.34169665e-02 -5.77735007e-01
4.61672872e-01 1.43881872e-01 4.94096279e-01 2.09525555e-01
-1.70425028e-01 -9.52426121e-02 -3.73644680e-02 5.39256148e-02
9.36066210e-01 1.31213695e-01 3.49912703e-01 -1.81456521e-01
6.70863867e-01 -1.70848489e-01 5.98403752e-01 8.90358746e-01
-4.47808266e-01 1.01173520e+00 1.45891309e-01 -3.88206482e-01
-1.12757957e+00 -1.00590026e+00 -7.91991875e-02 4.75156248e-01
5.74412048e-01 -4.31580283e-02 -9.64548647e-01 -4.72321421e-01
-1.29305616e-01 4.94807571e-01 -4.63781267e-01 -6.75096661e-02
-4.08366233e-01 -9.72306132e-01 5.22497296e-02 2.38890663e-01
8.78182352e-01 -6.63658023e-01 -3.70918214e-01 7.98022747e-02
-1.81044415e-01 -1.00992286e+00 -9.66811299e-01 -2.62784362e-01
-9.36669230e-01 -1.17485523e+00 -1.07611716e+00 -8.83671045e-01
1.29628515e+00 7.14913428e-01 6.33765876e-01 4.65410948e-02
-1.78673387e-01 1.93829581e-01 -9.20591578e-02 2.05962732e-01
-3.08167368e-01 -5.44567466e-01 2.65996039e-01 5.22346735e-01
8.12305063e-02 -2.82425314e-01 -9.19725537e-01 3.28811109e-01
-5.72791636e-01 1.14920273e-01 8.44655991e-01 1.07341719e+00
2.45496169e-01 5.94633400e-01 2.22396150e-01 -5.65035343e-01
7.20483601e-01 -1.55928001e-01 -6.74725175e-01 1.45991474e-01
-5.24453759e-01 5.80663234e-02 5.73498011e-01 -7.27501929e-01
-1.29728365e+00 2.96561569e-01 5.07249951e-01 -1.02958095e+00
-1.64215967e-01 3.06944042e-01 9.56285447e-02 -4.36098486e-01
5.44493079e-01 4.98917073e-01 -2.96456087e-02 -5.02565086e-01
4.09198880e-01 5.43538153e-01 8.42899680e-01 -2.80329645e-01
1.24287724e+00 4.50614154e-01 -3.83345634e-01 -9.27515864e-01
-6.92060769e-01 -9.30386662e-01 -6.33994043e-01 -3.23530853e-01
4.29909766e-01 -1.09382045e+00 -4.23606455e-01 1.06682348e+00
-1.11293876e+00 -1.25189424e-01 -1.17387855e-02 7.32338727e-01
-4.71087277e-01 8.75601232e-01 -8.26360345e-01 -8.96560490e-01
-1.63357630e-01 -1.12534821e+00 7.35462785e-01 2.44277298e-01
-8.94278809e-02 -8.69570971e-01 -2.22890452e-01 2.70109117e-01
4.79182363e-01 -1.73642397e-01 3.97606254e-01 -3.85985412e-02
-6.80257797e-01 -3.71978670e-01 -4.62625057e-01 5.85610151e-01
6.71010315e-01 -2.80211270e-01 -1.13220239e+00 -4.89353925e-01
5.29451251e-01 1.07859328e-01 1.08247435e+00 6.31739795e-01
1.09403694e+00 -5.04221439e-01 -2.27384701e-01 7.12545812e-01
1.45929337e+00 -2.18030028e-02 6.03694916e-01 4.28664953e-01
8.19418371e-01 2.83228725e-01 7.04522431e-01 3.75148267e-01
-9.60400999e-02 5.21305323e-01 2.26188958e-01 -2.29965001e-01
-1.28192529e-01 1.74640808e-02 3.97720098e-01 6.32767200e-01
-1.57708704e-01 2.88013697e-01 -6.02626622e-01 7.50120461e-01
-1.97676110e+00 -8.79331589e-01 -1.17885113e-01 2.53215265e+00
1.10861039e+00 -1.96978692e-02 -2.32237712e-01 -5.54815121e-02
1.09389448e+00 2.19009370e-01 -6.55343771e-01 3.21997374e-01
-1.15742661e-01 -2.93659121e-01 7.38721132e-01 6.87871516e-01
-1.35491431e+00 9.43778396e-01 5.97743940e+00 7.01561272e-01
-1.07678652e+00 -7.37445503e-02 5.29773355e-01 7.16211880e-03
1.53759316e-01 3.19095135e-01 -2.47832403e-01 6.98424518e-01
3.46001953e-01 -1.39325634e-01 8.50731194e-01 8.20308030e-01
6.17942989e-01 -4.75190431e-01 -8.54576349e-01 1.23411226e+00
2.55711734e-01 -1.04527903e+00 -3.39400798e-01 -2.72343904e-01
1.01344824e+00 -1.89735845e-01 1.15751766e-01 -3.68365347e-01
1.83468729e-01 -6.56548619e-01 7.10065782e-01 1.00378168e+00
4.58394259e-01 -5.79064608e-01 7.82941520e-01 2.96280265e-01
-7.99524724e-01 -1.95561107e-02 -5.38866878e-01 -6.52018860e-02
-6.10766700e-03 1.07090843e+00 -8.30682933e-01 3.46782714e-01
7.86782324e-01 7.97692060e-01 -7.67812073e-01 1.83511126e+00
-3.70392650e-01 6.78437829e-01 -3.55042107e-02 4.89025325e-01
-3.07010617e-02 -4.62952614e-01 8.36082220e-01 1.00387621e+00
5.59743285e-01 -6.86770678e-02 1.14959247e-01 7.54713118e-01
6.24263696e-02 -1.65075779e-01 -2.52842546e-01 4.22016531e-01
5.32390535e-01 1.19830191e+00 -5.99918604e-01 -5.67930043e-01
-4.84793603e-01 1.70354533e+00 1.28303006e-01 8.04565191e-01
-7.48402357e-01 -4.13821995e-01 7.32325733e-01 -6.99953213e-02
4.21697110e-01 -3.13122302e-01 -2.76367456e-01 -1.48193407e+00
2.28627518e-01 -9.39483106e-01 8.03599507e-03 -1.12054121e+00
-1.34738970e+00 4.17531163e-01 -3.04195166e-01 -1.62841535e+00
4.74665724e-02 -2.36073881e-01 -8.29600632e-01 1.00630808e+00
-1.55689788e+00 -8.46295178e-01 -6.22875214e-01 4.78973746e-01
7.66515911e-01 1.12783864e-01 2.93702900e-01 -4.41571139e-02
-5.20649374e-01 1.22632384e-01 6.33848369e-01 6.77241012e-03
1.32346809e+00 -1.30857384e+00 2.54389316e-01 1.50795007e+00
-8.12453926e-02 9.24257934e-01 1.09201670e+00 -6.87266469e-01
-1.19233155e+00 -1.11181462e+00 4.85902399e-01 -2.18772545e-01
7.73418069e-01 -2.81417631e-02 -1.15890121e+00 5.03882408e-01
3.26360881e-01 3.01529914e-01 -7.37546338e-03 -3.20192724e-01
-1.26492023e-01 -1.68210343e-01 -9.58660483e-01 6.66305482e-01
5.38707852e-01 -3.84276211e-01 -6.70217156e-01 3.16006094e-01
3.92505735e-01 -5.87880135e-01 -4.67058659e-01 3.26645970e-01
2.42522195e-01 -9.67597187e-01 1.03096485e+00 4.72262949e-02
7.97695965e-02 -9.29783046e-01 4.15314496e-01 -1.59452593e+00
-5.54339230e-01 -9.67461050e-01 -4.22205240e-01 1.08081686e+00
-8.49779174e-02 -5.82982421e-01 6.97432399e-01 4.78950083e-01
-1.28674462e-01 -3.32114428e-01 -6.57164812e-01 -7.33487606e-01
-6.01369858e-01 -8.36242884e-02 1.94255844e-01 1.07673883e+00
-2.69252837e-01 -1.32735208e-01 -7.89906502e-01 7.45097399e-01
1.17225897e+00 1.90024376e-01 7.68771231e-01 -9.51048732e-01
-1.74663097e-01 -1.23631768e-01 -3.14315259e-01 -1.23355401e+00
-5.57641722e-02 -2.37212807e-01 5.21542311e-01 -1.38995206e+00
2.29337275e-01 -2.59544343e-01 -3.64451081e-01 6.18829876e-02
-8.18910599e-01 3.20317447e-01 -1.37559235e-01 7.61879325e-01
-4.34256554e-01 3.83230716e-01 1.11490250e+00 -4.31689918e-02
-3.16054046e-01 1.84440315e-01 -4.97320414e-01 1.05170786e+00
6.48879707e-01 -2.64160961e-01 -3.35521996e-01 -3.60978782e-01
-3.16206694e-01 -1.23382844e-01 6.98387861e-01 -1.06312490e+00
6.33227289e-01 -1.30937472e-01 8.76533449e-01 -4.94769275e-01
2.42095962e-01 -8.09797287e-01 -3.34232450e-02 2.17464522e-01
-7.08032846e-02 -2.16611668e-01 7.63291642e-02 7.27268577e-01
-2.72738278e-01 -4.05015945e-01 1.03234196e+00 -1.92182809e-01
-9.00675297e-01 1.08194068e-01 -3.42651159e-01 -2.91638345e-01
9.33009028e-01 -3.87710541e-01 -3.46486717e-01 -5.10606289e-01
-5.15901804e-01 -8.99842009e-02 9.96834517e-01 1.53344214e-01
8.68718505e-01 -1.04657888e+00 -6.09054267e-01 3.60383600e-01
-3.79897095e-02 -9.64866579e-02 2.38416031e-01 1.23970342e+00
-5.91488123e-01 5.65533061e-03 1.85581446e-01 -5.11900842e-01
-1.26061368e+00 6.84027016e-01 4.51806366e-01 2.73193181e-01
-6.92604482e-01 9.00461137e-01 1.34177327e-01 3.01032990e-01
2.21393824e-01 -3.99241894e-01 8.29780176e-02 -2.65761703e-01
5.31808555e-01 5.51658511e-01 -3.09565365e-02 -5.94809890e-01
-1.81429714e-01 5.36483705e-01 -2.82987624e-01 -2.72267051e-02
1.14794433e+00 -3.87489110e-01 -5.67901492e-01 2.90327072e-01
9.12594140e-01 4.89295483e-01 -1.67850900e+00 -4.55616057e-01
1.31888047e-01 -1.09051073e+00 3.31045270e-01 -5.37646711e-01
-9.73508179e-01 3.29313934e-01 6.99292123e-01 -9.81258675e-02
1.43288422e+00 -2.50367284e-01 6.78183436e-01 1.53643593e-01
1.66760847e-01 -1.07327235e+00 1.89056769e-01 1.65078416e-01
8.22023511e-01 -1.22070134e+00 3.12177658e-01 -4.08124447e-01
-4.81024712e-01 1.10584617e+00 5.08197904e-01 -3.69030803e-01
4.26540136e-01 6.85494067e-03 2.33455673e-01 1.85448587e-01
-3.70781362e-01 -1.66800693e-01 3.80557269e-01 3.94279480e-01
1.97341785e-01 -4.43186402e-01 -4.88199256e-02 -1.57685637e-01
1.36165738e-01 1.27028376e-01 8.02866638e-01 7.75774062e-01
-6.80623651e-01 -7.04590440e-01 -1.05190563e+00 4.48497891e-01
-3.01098585e-01 -3.86409014e-01 -8.63821581e-02 2.81137854e-01
-5.45079745e-02 8.98598731e-01 -1.86406881e-01 -5.72558232e-02
1.31205693e-01 -1.32278666e-01 4.08400387e-01 -2.78240412e-01
-8.74835625e-02 5.84116399e-01 -2.18124226e-01 -3.51246476e-01
-3.76183212e-01 -8.45496356e-01 -7.86553144e-01 -8.57312456e-02
-7.24701583e-01 2.13509157e-01 3.55882049e-01 7.22962081e-01
2.32415080e-01 2.06335828e-01 7.74549842e-01 -1.04844511e+00
-6.53658390e-01 -1.17420816e+00 -6.45524621e-01 4.35085922e-01
9.02816236e-01 -5.37257373e-01 -9.15419161e-01 5.33502221e-01] | [11.568583488464355, -2.7031612396240234] |
6cf949b1-4680-4f08-8475-efe49bd13ce1 | lcpformer-towards-effective-3d-point-cloud | 2210.12755 | null | https://arxiv.org/abs/2210.12755v2 | https://arxiv.org/pdf/2210.12755v2.pdf | LCPFormer: Towards Effective 3D Point Cloud Analysis via Local Context Propagation in Transformers | Transformer with its underlying attention mechanism and the ability to capture long-range dependencies makes it become a natural choice for unordered point cloud data. However, separated local regions from the general sampling architecture corrupt the structural information of the instances, and the inherent relationships between adjacent local regions lack exploration, while local structural information is crucial in a transformer-based 3D point cloud model. Therefore, in this paper, we propose a novel module named Local Context Propagation (LCP) to exploit the message passing between neighboring local regions and make their representations more informative and discriminative. More specifically, we use the overlap points of adjacent local regions (which statistically show to be prevalent) as intermediaries, then re-weight the features of these shared points from different local regions before passing them to the next layers. Inserting the LCP module between two transformer layers results in a significant improvement in network expressiveness. Finally, we design a flexible LCPFormer architecture equipped with the LCP module. The proposed method is applicable to different tasks and outperforms various transformer-based methods in benchmarks including 3D shape classification and dense prediction tasks such as 3D object detection and semantic segmentation. Code will be released for reproduction. | ['Jungong Han', 'Banghuai Li', 'Zhiyou Zhao', 'Zhuoxu Huang'] | 2022-10-23 | null | null | null | null | ['3d-shape-retrieval', '3d-point-cloud-classification'] | ['computer-vision', 'computer-vision'] | [ 1.12680100e-01 2.78173368e-02 -1.96030989e-01 -5.58067858e-01
-3.81156713e-01 -4.56195921e-01 5.35158038e-01 4.45580602e-01
-1.26174912e-01 3.32913220e-01 8.67187232e-02 -1.25987694e-01
-2.16762736e-01 -8.92171919e-01 -9.64058697e-01 -8.59572589e-01
-2.65652180e-01 6.47982776e-01 5.42122126e-01 -2.17870399e-02
2.46341392e-01 1.10013425e+00 -1.46810722e+00 5.89286149e-01
8.53867173e-01 1.32088006e+00 6.02557123e-01 -1.09519050e-01
-5.95735371e-01 4.87379074e-01 -3.25934678e-01 -1.67209990e-02
2.83188611e-01 1.92188114e-01 -6.91854477e-01 -9.02005062e-02
5.77995718e-01 -1.54283360e-01 -2.18366295e-01 9.03178930e-01
3.75576228e-01 2.31621876e-01 5.84842503e-01 -1.10336149e+00
-4.26670074e-01 5.41975737e-01 -7.31882930e-01 3.46139282e-01
-2.31803814e-03 1.02062546e-01 1.32639468e+00 -1.06447089e+00
5.27309179e-01 1.41580117e+00 6.27550304e-01 2.07414761e-01
-1.08360541e+00 -7.98385799e-01 8.39735806e-01 2.53753245e-01
-1.36816394e+00 -8.28212425e-02 1.21709490e+00 -3.42706710e-01
1.04608536e+00 3.10985178e-01 8.11431348e-01 1.02291310e+00
1.70524903e-02 1.07950032e+00 9.37777519e-01 1.34628907e-01
1.47466481e-01 2.44840309e-02 2.23354563e-01 6.12563550e-01
2.82779727e-02 -9.07679424e-02 -4.32280868e-01 -1.39750227e-01
8.89272988e-01 3.61156911e-01 -1.77631855e-01 -6.33113503e-01
-9.44258869e-01 7.69521415e-01 1.12474453e+00 3.28638405e-01
-5.39080024e-01 7.35437721e-02 1.31341994e-01 9.84703675e-02
6.73602283e-01 2.28166923e-01 -4.54052955e-01 2.43139073e-01
-8.92845690e-01 2.46171728e-01 4.33481276e-01 1.11602473e+00
9.62627113e-01 -2.82607883e-01 -2.64821261e-01 9.71943200e-01
3.43318135e-01 2.47550324e-01 1.00393549e-01 -6.03210688e-01
6.63512647e-01 1.01936865e+00 -2.36930370e-01 -1.27529538e+00
-3.94748002e-01 -8.27360809e-01 -9.04772043e-01 1.41343728e-01
7.07854982e-03 3.74346882e-01 -9.22825634e-01 1.52623641e+00
5.59204340e-01 4.06128526e-01 -3.90426934e-01 1.03148806e+00
8.20843399e-01 7.53114164e-01 -1.39017016e-01 2.04101756e-01
1.09571731e+00 -7.26349890e-01 -2.27544177e-02 -3.06678385e-01
2.36024424e-01 -5.18329561e-01 9.82291341e-01 1.08047120e-01
-8.94543231e-01 -5.15644968e-01 -8.60936582e-01 -2.49265030e-01
-3.39483470e-01 -3.15508842e-01 6.71815991e-01 -6.75067902e-02
-9.13196921e-01 7.48920262e-01 -9.48892057e-01 -1.14269085e-01
9.75258887e-01 5.28526366e-01 -1.92461967e-01 -2.38965765e-01
-7.38454580e-01 4.73368198e-01 2.44093284e-01 1.57218233e-01
-6.90097928e-01 -8.21044862e-01 -8.71542335e-01 8.85670334e-02
3.08575600e-01 -4.63188678e-01 7.91210830e-01 -7.47259378e-01
-1.00466955e+00 8.02883923e-01 -2.88844675e-01 -2.99626350e-01
3.85005623e-01 -1.12340435e-01 1.02645524e-01 7.12611005e-02
1.88082188e-01 8.28302383e-01 8.45875323e-01 -1.44115329e+00
-6.61040425e-01 -5.29925585e-01 1.64174274e-01 3.04832131e-01
-6.73647374e-02 -2.42954805e-01 -7.51161575e-01 -5.83353579e-01
5.27538538e-01 -8.34800184e-01 -2.39826113e-01 6.70343712e-02
-4.92755294e-01 -5.21803260e-01 1.12470138e+00 -2.68730968e-01
8.44633877e-01 -2.40117097e+00 2.05475867e-01 6.31345987e-01
4.71006542e-01 3.99079099e-02 -2.37492025e-01 1.83043763e-01
3.21228988e-02 1.80645026e-02 -4.28399652e-01 -4.41128701e-01
2.66471580e-02 5.40544629e-01 -2.91212291e-01 2.79363483e-01
5.54208040e-01 1.03262913e+00 -8.21835399e-01 -3.94296497e-01
4.50994909e-01 5.23254454e-01 -7.70909667e-01 2.00603995e-02
-3.28335404e-01 5.58213890e-01 -8.44876230e-01 8.86288941e-01
9.17745650e-01 -3.24215263e-01 -3.24157983e-01 -2.34833971e-01
-5.98306544e-02 6.66026175e-01 -8.39654148e-01 1.67682159e+00
-5.75989306e-01 4.10395026e-01 1.56038538e-01 -9.35411751e-01
1.17701924e+00 -2.01078370e-01 5.49181759e-01 -7.01758504e-01
-3.21152881e-02 1.60992756e-01 -7.09577426e-02 -2.65123278e-01
3.78185600e-01 -9.05077085e-02 1.14828587e-01 4.61370386e-02
-7.92189911e-02 -3.55833061e-02 -2.78066754e-01 6.33250698e-02
9.85827625e-01 -3.14461975e-03 -6.65213317e-02 -2.91483432e-01
4.16449428e-01 -2.03873247e-01 6.27288878e-01 5.61667085e-01
-2.03517987e-03 7.04226792e-01 6.20421708e-01 -5.09965003e-01
-8.62915277e-01 -9.90538418e-01 -2.54310727e-01 7.27012694e-01
4.48516518e-01 -3.90231490e-01 -2.14764088e-01 -9.44524646e-01
4.83804882e-01 5.48264563e-01 -7.16940701e-01 -8.16074833e-02
-6.09679282e-01 -4.94839996e-01 1.98196322e-01 7.26947248e-01
5.99859416e-01 -9.92962658e-01 -3.62854451e-01 2.46739864e-01
-1.14035651e-01 -1.06090474e+00 -2.77582169e-01 5.08528113e-01
-1.13088381e+00 -8.04906964e-01 -4.05956954e-01 -6.65806055e-01
5.99230170e-01 2.88886964e-01 1.16922081e+00 2.44759750e-02
1.00457075e-03 6.43747896e-02 -3.62938046e-01 -2.57228553e-01
1.98025644e-01 2.36323282e-01 -3.61321121e-01 2.05335304e-01
4.11315262e-01 -9.20017898e-01 -6.77851617e-01 3.64493787e-01
-6.43645346e-01 1.65879931e-02 6.77974463e-01 8.44984114e-01
1.11827159e+00 -6.64611859e-03 1.89371824e-01 -8.96577716e-01
9.25496072e-02 -5.99973440e-01 -4.67833579e-01 -1.27774164e-01
-1.30303770e-01 7.98159614e-02 4.46376413e-01 -1.48414314e-01
-7.41774499e-01 3.66587006e-02 -4.02791083e-01 -8.95283103e-01
-2.67975152e-01 3.63917649e-01 -5.31759501e-01 -1.20699875e-01
1.12406299e-01 1.96314335e-01 -2.02403128e-01 -7.90656388e-01
-8.65941048e-02 2.61941344e-01 7.03618601e-02 -6.97074950e-01
6.85846984e-01 5.28476596e-01 1.81735203e-01 -8.27440143e-01
-7.96032310e-01 -5.29058754e-01 -5.95851004e-01 2.66933478e-02
7.14432180e-01 -8.93399477e-01 -5.59192300e-01 2.46708363e-01
-1.17909050e+00 -2.78029174e-01 -5.02661765e-01 1.57345146e-01
-2.54665881e-01 1.67880952e-01 -4.63710576e-01 -5.70208848e-01
-9.85238925e-02 -1.31162572e+00 1.42872918e+00 4.42325324e-02
2.46816482e-02 -8.63008261e-01 -2.02663645e-01 7.77519047e-02
2.11309776e-01 3.45001668e-01 1.18784964e+00 -6.68268561e-01
-1.00745368e+00 -1.11716256e-01 -4.60542113e-01 3.58593345e-01
1.05960436e-01 -1.81452855e-01 -1.12912011e+00 -2.58873940e-01
4.45417091e-02 9.25203040e-02 1.14663172e+00 3.81322682e-01
1.73040020e+00 -2.33305112e-01 -6.62725687e-01 8.71255934e-01
1.25398934e+00 -5.27209453e-02 3.71117949e-01 4.30427901e-02
1.15995944e+00 6.21899068e-01 3.80087733e-01 2.80542970e-01
3.09819400e-01 6.74579382e-01 9.47605848e-01 -4.66094576e-02
3.08591016e-02 -4.27750409e-01 1.53275624e-01 8.71881485e-01
-1.82568833e-01 -2.59633183e-01 -8.26191843e-01 5.06441951e-01
-1.70545626e+00 -7.21124649e-01 -1.08004861e-01 2.10232711e+00
4.28532928e-01 3.40359181e-01 -1.60403922e-01 -1.27515242e-01
6.38528824e-01 5.09122491e-01 -6.51878715e-01 -2.40282118e-01
-2.37578526e-01 4.01137352e-01 4.43884164e-01 3.59644949e-01
-1.12722778e+00 9.27721739e-01 4.84794664e+00 1.09227872e+00
-1.15681589e+00 9.23349485e-02 6.47787392e-01 -2.61061788e-01
-6.34976566e-01 -4.45361733e-02 -7.37158060e-01 4.44325447e-01
2.65756309e-01 4.51398730e-01 7.53519684e-02 7.54515469e-01
4.30956110e-02 2.33293831e-01 -1.29312253e+00 9.43670809e-01
-1.91388309e-01 -1.22692609e+00 4.55540359e-01 2.39390507e-01
5.55741191e-01 4.59035754e-01 9.81252864e-02 8.05829391e-02
-1.04232412e-02 -9.70254123e-01 7.67329812e-01 3.95053357e-01
4.77602124e-01 -9.17833686e-01 7.27994263e-01 3.37932765e-01
-1.34427094e+00 -9.05049965e-02 -5.96373677e-01 1.23584986e-01
-1.95156112e-01 8.97956729e-01 -7.38403976e-01 7.26361990e-01
9.26699519e-01 1.21672797e+00 -4.86027837e-01 1.05091488e+00
-1.54096350e-01 4.55184937e-01 -6.44806921e-01 6.34614676e-02
6.80315256e-01 -2.08558545e-01 9.53578293e-01 1.08315325e+00
2.85425931e-01 -6.53378293e-02 2.17008397e-01 1.17345178e+00
-2.01486662e-01 1.75536834e-02 -8.39176953e-01 2.62958378e-01
3.84112895e-01 1.15722072e+00 -9.07969534e-01 -2.07502678e-01
-5.44114828e-01 8.49606216e-01 4.69258159e-01 2.44076446e-01
-7.52324104e-01 -1.00804873e-01 1.02881396e+00 3.82624298e-01
7.61528313e-01 -2.30384603e-01 -4.59319681e-01 -1.01015925e+00
1.81839362e-01 -6.04067206e-01 1.85007885e-01 -5.49284101e-01
-1.42288423e+00 7.53997743e-01 -8.82165879e-02 -1.32738268e+00
3.53934258e-01 -5.40115416e-01 -6.98762119e-01 8.79498422e-01
-1.70722234e+00 -1.13920319e+00 -4.08221811e-01 7.15932548e-01
5.28803647e-01 1.06739394e-01 4.68733281e-01 4.47440475e-01
-3.71808171e-01 5.05016267e-01 -1.13550134e-01 -4.46197996e-03
2.85206527e-01 -1.34021366e+00 3.98832470e-01 6.10377192e-01
2.47635782e-01 6.36404276e-01 1.71513528e-01 -6.68259442e-01
-1.01355386e+00 -1.35554707e+00 7.09234416e-01 -3.41317058e-01
3.17261547e-01 -7.41285264e-01 -1.10635543e+00 5.34706652e-01
-1.28073096e-01 2.07636446e-01 3.27052265e-01 2.15232149e-01
-5.39822400e-01 -3.88829142e-01 -9.84728515e-01 4.46743399e-01
1.42448831e+00 -6.62754655e-01 -4.98542219e-01 1.42751187e-01
8.04868579e-01 -3.81691396e-01 -7.06202686e-01 6.77807212e-01
1.65771008e-01 -9.39258575e-01 1.13911760e+00 -3.62217963e-01
5.21602452e-01 -3.00867498e-01 -3.56177837e-01 -1.15331471e+00
-4.86713290e-01 -3.15858334e-01 -5.14022037e-02 1.29037559e+00
3.84281904e-01 -6.92112863e-01 8.96465302e-01 3.89389247e-01
-6.02873445e-01 -1.04450178e+00 -1.17244029e+00 -5.98922074e-01
8.69582370e-02 -5.86667418e-01 8.75809371e-01 9.20422256e-01
-4.38610524e-01 2.47517809e-01 7.38855749e-02 4.57481563e-01
5.37231386e-01 6.27365768e-01 5.02742589e-01 -1.38718462e+00
-1.70130476e-01 -5.69895744e-01 -6.10483825e-01 -1.69601965e+00
2.49712430e-02 -1.37101054e+00 -7.69148842e-02 -1.33736396e+00
4.00446914e-02 -1.06836271e+00 -5.85103214e-01 6.06156290e-01
-6.38227910e-02 2.15198487e-01 2.49352768e-01 2.89610416e-01
-4.47467357e-01 8.59330893e-01 1.46267354e+00 -4.21045363e-01
-2.20358238e-01 2.42245406e-01 -5.04866242e-01 7.41723299e-01
5.89246690e-01 -5.22495151e-01 -3.18369240e-01 -6.17092013e-01
6.42909706e-02 -2.99556375e-01 7.50279605e-01 -9.83269989e-01
2.34303936e-01 6.99587092e-02 6.10197425e-01 -1.10562348e+00
5.01266897e-01 -1.22910774e+00 -5.85628524e-02 8.40281323e-02
-1.31348178e-01 -4.38475274e-02 3.51932079e-01 6.72208130e-01
-3.98367703e-01 -6.32703006e-02 6.30034864e-01 -1.81801990e-01
-6.41872287e-01 7.71799862e-01 1.55216623e-02 -1.82012971e-02
8.79002869e-01 -2.75726408e-01 7.02293739e-02 1.13918275e-01
-6.75619662e-01 3.51009786e-01 5.26984394e-01 5.06107032e-01
7.79486597e-01 -1.34469712e+00 -4.53984410e-01 5.45329273e-01
1.95309862e-01 7.95679390e-01 3.34353119e-01 8.22593391e-01
-2.82280803e-01 2.94701487e-01 -1.82916403e-01 -1.13926756e+00
-1.06554449e+00 4.94519353e-01 3.14567119e-01 -1.63028315e-01
-1.12275803e+00 1.10221052e+00 7.70691156e-01 -5.10779083e-01
2.70561665e-01 -9.97769058e-01 -1.51511610e-01 1.01013184e-01
3.78622785e-02 8.09469894e-02 1.86957806e-01 -5.81978619e-01
-6.45907342e-01 7.22738802e-01 -2.20766604e-01 4.48383778e-01
1.60952449e+00 -4.50341904e-04 -2.04144403e-01 5.30444145e-01
1.30379915e+00 7.93173760e-02 -1.51044083e+00 -4.23782915e-01
-7.10395426e-02 -5.91575682e-01 1.08862050e-01 -4.67192084e-01
-1.50653028e+00 1.03719831e+00 3.27536315e-01 1.91868693e-01
1.10179901e+00 3.88213545e-01 7.85849392e-01 1.84402511e-01
4.27382916e-01 -6.79692268e-01 -2.25199148e-01 6.81156218e-01
1.00267959e+00 -1.12817943e+00 -2.04910889e-01 -7.08634019e-01
-4.44510013e-01 7.41324842e-01 5.70012391e-01 -3.52734834e-01
8.18473101e-01 1.59631923e-01 -4.21431035e-01 -5.79743743e-01
-8.28047097e-01 -1.88386753e-01 3.33692402e-01 5.92066944e-01
4.07529250e-02 8.18933696e-02 2.34568298e-01 5.59115052e-01
-4.25158031e-02 -5.25451124e-01 -1.35189131e-01 6.76271498e-01
-2.66617239e-01 -1.03485715e+00 -1.39262930e-01 6.63670897e-01
-2.54438400e-01 -1.53984115e-01 -4.31749046e-01 7.72311449e-01
5.20722568e-01 4.47212219e-01 4.93874639e-01 -4.24916238e-01
4.79526192e-01 3.53667475e-02 3.85672927e-01 -6.45933509e-01
-7.45576739e-01 1.45578012e-01 -1.64723217e-01 -7.84281909e-01
-4.19628024e-01 -8.75420868e-01 -1.20112908e+00 -2.73133099e-01
-2.55269766e-01 -1.49877900e-02 4.22546715e-01 8.96096528e-01
6.05601370e-01 6.32989764e-01 6.85189784e-01 -1.03193021e+00
-9.22664627e-02 -7.24130094e-01 -5.28166234e-01 2.86024660e-01
4.83924538e-01 -9.52636957e-01 -2.12564379e-01 -5.46961188e-01] | [7.952996253967285, -3.480642318725586] |
a99675e0-2619-44aa-b0f6-ea621bb0f504 | superpixel-based-refinement-for-object | 2101.04574 | null | https://arxiv.org/abs/2101.04574v1 | https://arxiv.org/pdf/2101.04574v1.pdf | Superpixel-based Refinement for Object Proposal Generation | Precise segmentation of objects is an important problem in tasks like class-agnostic object proposal generation or instance segmentation. Deep learning-based systems usually generate segmentations of objects based on coarse feature maps, due to the inherent downsampling in CNNs. This leads to segmentation boundaries not adhering well to the object boundaries in the image. To tackle this problem, we introduce a new superpixel-based refinement approach on top of the state-of-the-art object proposal system AttentionMask. The refinement utilizes superpixel pooling for feature extraction and a novel superpixel classifier to determine if a high precision superpixel belongs to an object or not. Our experiments show an improvement of up to 26.0% in terms of average recall compared to original AttentionMask. Furthermore, qualitative and quantitative analyses of the segmentations reveal significant improvements in terms of boundary adherence for the proposed refinement compared to various deep learning-based state-of-the-art object proposal generation systems. | ['Simone Frintrop', 'Christian Wilms'] | 2021-01-12 | null | null | null | null | ['object-proposal-generation'] | ['computer-vision'] | [ 4.60795969e-01 6.71388566e-01 6.44568652e-02 -4.05939460e-01
-9.25593555e-01 -2.03012973e-01 5.25697708e-01 3.92939776e-01
-5.26397467e-01 7.66288757e-01 -1.56840563e-01 2.49685869e-01
6.99689835e-02 -9.95188653e-01 -9.79179323e-01 -5.28807163e-01
5.78365624e-01 8.72901738e-01 8.65945518e-01 1.38212413e-01
4.95903552e-01 4.66058880e-01 -1.70398569e+00 4.01961893e-01
8.91391218e-01 1.11808300e+00 2.76119292e-01 3.88181031e-01
-4.37660933e-01 9.10065592e-06 -8.62116277e-01 -2.41621837e-01
4.01304722e-01 -1.99494168e-01 -1.03544760e+00 3.99594456e-01
9.87182438e-01 -2.09419996e-01 3.56083930e-01 1.26738906e+00
2.74550933e-02 -5.54837845e-02 8.21001351e-01 -1.05603480e+00
-6.36982918e-01 7.53005266e-01 -6.74014032e-01 7.23576769e-02
-2.23079801e-01 8.47592428e-02 1.02399457e+00 -8.33276093e-01
6.57063603e-01 1.05634642e+00 5.86330116e-01 6.95792854e-01
-1.50571024e+00 -5.87835014e-01 1.93380862e-01 6.68602288e-02
-1.58003521e+00 -1.47459611e-01 4.52222943e-01 -5.13596892e-01
6.32205009e-01 1.36932552e-01 8.10680330e-01 5.05926251e-01
2.00437140e-02 9.60036933e-01 1.05553579e+00 -4.04069245e-01
4.87099409e-01 3.11045438e-01 2.56242216e-01 5.56161284e-01
6.76664948e-01 -2.70611346e-01 -2.13534608e-01 3.50182563e-01
1.04565632e+00 -8.73688608e-02 5.63562363e-02 -1.25560239e-01
-1.19668376e+00 6.29579604e-01 9.03031588e-01 4.86466557e-01
-7.32122302e-01 3.55163097e-01 2.46679075e-02 -4.73674864e-01
5.76562524e-01 8.31026137e-01 -4.02280450e-01 2.34904230e-01
-1.65143013e+00 4.97483492e-01 3.95118058e-01 1.06921399e+00
1.00984490e+00 -1.42591938e-01 -7.67012775e-01 4.07344520e-01
1.92105874e-01 -6.15410283e-02 3.00922155e-01 -9.42561328e-01
2.50811607e-01 1.09942186e+00 1.79584071e-01 -6.19272828e-01
-2.72436917e-01 -5.96610427e-01 -6.70352399e-01 4.50135052e-01
6.45429432e-01 1.50967404e-01 -1.53965724e+00 1.24573040e+00
5.14940381e-01 7.53310472e-02 -1.46908775e-01 9.84991550e-01
8.29093635e-01 5.50866604e-01 2.22448841e-01 1.31239742e-01
1.61293805e+00 -1.13369143e+00 -5.18588245e-01 -2.74038702e-01
2.15895638e-01 -6.96187615e-01 9.72367644e-01 2.86705554e-01
-1.05731714e+00 -7.92227149e-01 -1.02311158e+00 -3.69531848e-02
-3.62429678e-01 3.02496076e-01 4.70851362e-01 5.76441050e-01
-1.01451075e+00 5.36334515e-01 -8.25719893e-01 -3.60198677e-01
1.11320210e+00 5.84284127e-01 -1.21454902e-01 1.94164649e-01
-6.54603601e-01 7.05459416e-01 8.66018653e-01 1.39172329e-02
-6.93987191e-01 -9.79191840e-01 -5.08244336e-01 3.66762370e-01
5.19599974e-01 -6.55837774e-01 1.31506693e+00 -1.17724061e+00
-1.34124506e+00 1.02111483e+00 -7.64640514e-03 -7.63197005e-01
6.70190513e-01 -3.99125844e-01 2.78302670e-01 1.38970926e-01
3.51539224e-01 1.45148456e+00 1.06843424e+00 -1.19803309e+00
-1.17169905e+00 -1.58697709e-01 1.84481949e-01 -1.62230097e-02
-4.16696891e-02 -2.40258157e-01 -5.65614760e-01 -7.19869316e-01
3.34854394e-01 -7.11512685e-01 -5.81478953e-01 1.09197557e-01
-8.17615747e-01 -5.62049150e-01 1.03157294e+00 -2.45836616e-01
8.94731939e-01 -1.71995902e+00 -4.74066250e-02 1.54481046e-02
2.57056773e-01 4.52231348e-01 2.79912129e-02 -3.48413736e-01
2.02795729e-01 2.87628531e-01 -4.50121909e-01 -4.33587819e-01
-5.58028482e-02 -6.93694651e-02 7.19837248e-02 2.08961129e-01
6.17571294e-01 9.26567554e-01 -8.08071315e-01 -7.81714678e-01
3.81504267e-01 3.72064471e-01 -5.02820671e-01 1.85806215e-01
-6.28240764e-01 3.57621342e-01 -5.25406957e-01 6.60875916e-01
7.21424818e-01 -2.42647916e-01 -3.13497782e-01 -4.73960966e-01
-2.52876073e-01 6.93922341e-02 -1.15864861e+00 1.67706358e+00
-1.81763515e-01 5.91594756e-01 -1.43391743e-01 -7.39180267e-01
8.36219788e-01 2.80255437e-01 1.94847301e-01 -3.32432568e-01
1.88386038e-01 2.92639017e-01 5.35316579e-02 4.92834263e-02
9.15503204e-01 7.11806864e-02 8.65089074e-02 3.43620598e-01
2.73023397e-01 -3.99142087e-01 4.97841001e-01 6.12433590e-02
7.85458803e-01 2.67099589e-01 2.32523233e-01 -6.42953813e-01
4.10396278e-01 2.12446854e-01 6.17654085e-01 8.96302700e-01
-1.66272402e-01 1.11019838e+00 4.57767457e-01 -4.57732618e-01
-1.15899944e+00 -7.08050728e-01 -3.06394666e-01 7.59521961e-01
2.32799962e-01 -1.04549721e-01 -1.46549392e+00 -8.61935079e-01
-1.49575055e-01 5.07206678e-01 -8.46385956e-01 2.17280984e-01
-5.82319140e-01 -7.33233869e-01 2.77876168e-01 7.24655747e-01
8.81514251e-01 -1.41886854e+00 -1.05834103e+00 3.01918089e-01
2.42393501e-02 -1.15395188e+00 -3.34869385e-01 -6.57905312e-03
-9.94892597e-01 -9.82436061e-01 -1.05965948e+00 -6.52132630e-01
1.14793634e+00 -1.78434923e-01 9.97770727e-01 4.91898507e-02
-5.28590918e-01 -9.85178277e-02 -2.98694104e-01 -4.05805945e-01
-3.76002103e-01 5.98595679e-01 -2.92371213e-01 1.60449326e-01
2.00803906e-01 -1.94929894e-02 -8.69153142e-01 1.60721809e-01
-9.78844166e-01 3.38773757e-01 9.04006898e-01 5.03652215e-01
9.37803566e-01 1.51915075e-02 3.95075470e-01 -1.08887923e+00
2.23308414e-01 -1.95587091e-02 -9.48063731e-01 1.73415646e-01
-5.58534265e-01 2.34946176e-01 1.99346229e-01 -3.49128723e-01
-9.97419477e-01 3.57064992e-01 3.89106646e-02 -3.10976654e-01
-5.68034410e-01 1.90918855e-02 -7.29644075e-02 -2.08907668e-02
5.96717477e-01 -1.90587472e-02 -4.18495327e-01 -4.75453049e-01
4.08146292e-01 4.26597893e-01 4.79170382e-01 -3.42795551e-01
5.89252710e-01 4.62171674e-01 -6.98911548e-02 -7.69015849e-01
-1.08697629e+00 -5.28714418e-01 -9.42399323e-01 -1.74031317e-01
1.27286208e+00 -4.16356772e-01 -5.03181875e-01 5.58463752e-01
-1.46316564e+00 -5.15872240e-01 -6.85616374e-01 1.67404577e-01
-5.11216760e-01 -1.51043966e-01 -4.36378241e-01 -6.58536613e-01
-6.16889715e-01 -1.29737997e+00 1.35274518e+00 7.29151189e-01
-3.20718557e-01 -4.99538511e-01 -3.44724238e-01 4.81629729e-01
5.21329582e-01 1.65304840e-01 6.72283709e-01 -7.14349091e-01
-1.06010294e+00 -2.67976731e-01 -7.52406538e-01 2.21488819e-01
1.07089587e-01 1.49866819e-01 -9.81818676e-01 3.58015597e-02
-4.60781127e-01 3.47371735e-02 1.01206422e+00 6.82547927e-01
1.22015560e+00 -3.41315240e-01 -5.79967260e-01 2.35111549e-01
1.47943461e+00 -1.87492277e-02 6.21212840e-01 3.48687530e-01
7.91486442e-01 5.94043732e-01 7.40601122e-01 2.45813116e-01
1.16766349e-03 7.95055628e-01 5.67166924e-01 -2.48396233e-01
-5.22308052e-01 8.45926180e-02 -2.31720537e-01 -5.99794760e-02
-6.65507913e-02 -2.45484605e-01 -8.93383801e-01 1.04520643e+00
-1.84926212e+00 -6.23192012e-01 -4.77964669e-01 2.16034937e+00
8.51934314e-01 5.04877150e-01 2.18274407e-02 2.75368452e-01
1.00994706e+00 -9.61512774e-02 -4.09469843e-01 -2.67604798e-01
2.68028855e-01 4.64555442e-01 6.40360236e-01 3.58131528e-01
-1.29566848e+00 1.38112402e+00 5.69268990e+00 1.08910203e+00
-1.16198087e+00 2.28405073e-01 7.97720134e-01 2.43499994e-01
1.77377269e-01 -8.36371034e-02 -1.17791128e+00 3.38077933e-01
3.28731298e-01 1.73920229e-01 -2.50227779e-01 8.17337334e-01
-4.27977880e-03 -4.89435166e-01 -1.12414074e+00 5.90726495e-01
-5.89312725e-02 -1.61283529e+00 1.65327027e-01 9.48712081e-02
1.07870543e+00 -1.48660079e-01 -3.71689983e-02 1.11028090e-01
1.08055867e-01 -8.99818897e-01 1.09050202e+00 5.73824644e-01
5.50622225e-01 -5.69706798e-01 9.53231633e-01 7.38051459e-02
-1.05981231e+00 3.22794467e-01 -3.72798413e-01 1.82365909e-01
2.01503485e-01 8.13558102e-01 -1.29915047e+00 1.99873626e-01
7.98574805e-01 2.37373248e-01 -6.35099232e-01 1.43170822e+00
-3.98935050e-01 6.01710320e-01 -3.99679899e-01 1.46393701e-01
4.88405585e-01 4.83093113e-02 4.41356927e-01 1.26566291e+00
8.35352615e-02 -3.85880470e-03 7.40786493e-02 1.60313606e+00
-2.66776681e-01 -3.18190679e-02 5.29837124e-02 -1.41310720e-02
3.42121601e-01 1.57280087e+00 -1.49911308e+00 -8.02596211e-01
8.61058310e-02 9.86546814e-01 1.17057920e-01 1.17751546e-01
-7.13323295e-01 -4.34079468e-01 4.44504321e-01 4.80996430e-01
7.36550689e-01 5.38954288e-02 -5.71896791e-01 -5.45834422e-01
-4.00851667e-02 -2.18530893e-01 -3.53863761e-02 -6.18203700e-01
-9.65613604e-01 6.75073206e-01 -3.19438875e-02 -9.18242931e-01
1.20499335e-01 -3.62945288e-01 -6.17039382e-01 6.58416212e-01
-1.29341233e+00 -1.44117177e+00 -4.88915414e-01 -8.71859938e-02
9.16586876e-01 2.75201857e-01 5.23578346e-01 7.53893331e-02
-4.11579400e-01 4.26418841e-01 -3.74081820e-01 2.00246811e-01
3.15841049e-01 -1.38410294e+00 5.60127437e-01 9.17186677e-01
1.38330251e-01 2.66821384e-01 5.95714092e-01 -6.35632753e-01
-3.62869412e-01 -1.53799343e+00 7.21270025e-01 -2.52025813e-01
1.29689455e-01 -3.02780926e-01 -9.68971848e-01 4.50473368e-01
1.94897830e-01 2.21525848e-01 9.77719426e-02 -2.43359745e-01
1.13706002e-02 -5.83278015e-02 -1.33007407e+00 4.77400333e-01
6.40122533e-01 6.35094419e-02 -5.97894490e-01 1.85912460e-01
8.33141923e-01 -4.79639649e-01 -6.50441527e-01 5.29433012e-01
5.88813484e-01 -9.98531759e-01 7.57265568e-01 -2.82818645e-01
5.84609449e-01 -4.90264773e-01 2.00159192e-01 -8.87583077e-01
-3.65806282e-01 -2.08332628e-01 1.44685701e-01 1.41341567e+00
5.89026093e-01 -2.96357214e-01 1.35310411e+00 6.92665875e-01
-4.12909687e-01 -7.13953316e-01 -9.82262254e-01 -6.18052363e-01
-2.17279181e-01 -1.93115339e-01 6.63736820e-01 5.44236541e-01
-5.24409294e-01 3.92568074e-02 2.70204633e-01 1.72873005e-01
7.62159407e-01 2.09380969e-01 7.25745082e-01 -1.27415752e+00
4.40552235e-02 -8.04668009e-01 -6.99016571e-01 -8.07052612e-01
-1.21903121e-01 -6.19955778e-01 4.97431219e-01 -1.97193444e+00
2.25156024e-01 -6.91312492e-01 -3.03721964e-01 6.13454998e-01
-2.40479797e-01 8.52549255e-01 2.27555141e-01 5.04059196e-02
-7.47622907e-01 2.41653889e-01 1.26358604e+00 -3.39619607e-01
-4.02042121e-01 2.00797260e-01 -5.29469848e-01 8.24525356e-01
6.78435683e-01 -7.03225136e-01 6.28938004e-02 -3.49048376e-01
-1.22096919e-01 -4.74007756e-01 4.37107354e-01 -1.40172970e+00
6.72500059e-02 -1.08458966e-01 3.29255551e-01 -7.15550065e-01
3.19209933e-01 -6.35582626e-01 -1.01233795e-01 5.29422164e-01
-3.19291443e-01 -5.67125916e-01 3.84860456e-01 3.70481461e-01
-7.67606199e-02 -3.80131900e-01 9.76230562e-01 -2.84142703e-01
-7.93967247e-01 2.73062140e-01 -4.24326926e-01 2.92875078e-02
1.20940268e+00 -5.88849545e-01 -1.88568667e-01 2.09684774e-01
-6.95014477e-01 -4.12396602e-02 4.20773864e-01 2.02746183e-01
5.31331122e-01 -1.04002035e+00 -7.25621581e-01 7.84723163e-02
1.30841047e-01 7.45406926e-01 1.14856310e-01 8.59736562e-01
-7.10894704e-01 5.37499189e-01 -3.49786460e-01 -8.32864881e-01
-1.20535958e+00 2.07052186e-01 3.04931104e-01 -2.14221716e-01
-5.57084918e-01 1.25999343e+00 6.02129817e-01 -7.54128397e-02
4.73106429e-02 -9.25208330e-01 -1.33633927e-01 7.91825503e-02
2.62499303e-01 4.09715861e-01 2.36284956e-01 -5.49371541e-01
-3.68620574e-01 6.73669696e-01 -3.56143415e-01 -1.13985151e-01
1.02159309e+00 3.01218152e-01 1.61167569e-02 2.15632886e-01
6.06604040e-01 -4.44791824e-01 -1.45867455e+00 -9.76861268e-02
1.62656516e-01 -4.23111588e-01 1.45494416e-01 -7.85218060e-01
-1.15253949e+00 8.47826719e-01 5.94102204e-01 1.87799081e-01
7.81084538e-01 2.52780199e-01 7.77722478e-01 -5.11896284e-03
3.29505205e-01 -1.16741037e+00 9.91778634e-03 2.11818174e-01
8.19669902e-01 -1.31070495e+00 1.02241829e-01 -5.88265955e-01
-3.38489622e-01 9.05120075e-01 9.65826929e-01 -4.36303586e-01
4.37582672e-01 5.61438203e-02 -7.71830976e-02 -2.76139796e-01
-3.00200254e-01 -4.21914309e-01 5.77436805e-01 4.39522773e-01
3.48843634e-01 1.74103990e-01 -3.44271302e-01 4.91462320e-01
-8.76799673e-02 1.94352982e-03 4.89180267e-01 6.38186812e-01
-6.65898442e-01 -7.62925208e-01 -5.21336913e-01 5.70719481e-01
-5.22899806e-01 -8.69839732e-03 -3.04015696e-01 8.13137650e-01
5.13288319e-01 5.72796881e-01 5.36274791e-01 5.92438839e-02
2.87497789e-01 -1.38050482e-01 5.70815504e-01 -1.10692418e+00
-8.29024792e-01 5.16662598e-02 -3.69234353e-01 -5.02231240e-01
-5.73285103e-01 -4.79661435e-01 -1.58914173e+00 3.53046298e-01
-7.89264619e-01 5.60739301e-02 7.29609430e-01 1.11061788e+00
3.19294900e-01 8.60171080e-01 -4.33335900e-02 -1.39487052e+00
-6.68275952e-02 -1.12728083e+00 -4.36668903e-01 2.83159554e-01
1.32311895e-01 -6.20574892e-01 8.58051181e-02 2.32453778e-01] | [9.526917457580566, 0.47979679703712463] |
5b38dee5-2a71-4997-91ae-39e3a7681c26 | patch-mix-contrastive-learning-with-audio | 2305.14032 | null | https://arxiv.org/abs/2305.14032v2 | https://arxiv.org/pdf/2305.14032v2.pdf | Patch-Mix Contrastive Learning with Audio Spectrogram Transformer on Respiratory Sound Classification | Respiratory sound contains crucial information for the early diagnosis of fatal lung diseases. Since the COVID-19 pandemic, there has been a growing interest in contact-free medical care based on electronic stethoscopes. To this end, cutting-edge deep learning models have been developed to diagnose lung diseases; however, it is still challenging due to the scarcity of medical data. In this study, we demonstrate that the pretrained model on large-scale visual and audio datasets can be generalized to the respiratory sound classification task. In addition, we introduce a straightforward Patch-Mix augmentation, which randomly mixes patches between different samples, with Audio Spectrogram Transformer (AST). We further propose a novel and effective Patch-Mix Contrastive Learning to distinguish the mixed representations in the latent space. Our method achieves state-of-the-art performance on the ICBHI dataset, outperforming the prior leading score by an improvement of 4.08%. | ['Se-Young Yun', 'Sungnyun Kim', 'Kyongpil Tae', 'Changwan Ha', 'Byungjo Lee', 'Soyoun Son', 'Hyerim Baek', 'Won-Yang Cho', 'June-Woo Kim', 'Sangmin Bae'] | 2023-05-23 | null | null | null | null | ['audio-classification', 'sound-classification'] | ['audio', 'audio'] | [ 2.58067459e-01 -3.88912588e-01 9.03248787e-03 1.10106222e-01
-1.20353079e+00 -3.97603095e-01 3.91975850e-01 6.29556254e-02
-1.60092503e-01 3.46847326e-01 3.02887410e-01 -3.55474293e-01
3.36694233e-02 -5.42887032e-01 -4.79657412e-01 -8.29393983e-01
2.74121404e-01 2.67374843e-01 1.46042973e-01 2.86439270e-01
-5.07680595e-01 2.17561945e-01 -1.33857238e+00 5.39114118e-01
6.98349416e-01 9.92563844e-01 1.88027725e-01 9.93146956e-01
6.18513599e-02 8.18512917e-01 -5.66367805e-01 -5.14227375e-02
-1.06711403e-01 -7.19548166e-01 -5.68045557e-01 -2.06029683e-01
3.29306811e-01 -6.19480014e-01 -4.46525931e-01 5.57767570e-01
1.02501047e+00 -2.09968865e-01 6.97584450e-01 -1.04194677e+00
-3.62014502e-01 1.15463234e-01 -3.02549064e-01 3.98103356e-01
2.42331862e-01 2.06218734e-01 1.04538560e+00 -9.43188429e-01
1.32356837e-01 8.71990561e-01 8.25535059e-01 6.06571853e-01
-9.96061206e-01 -8.45339477e-01 -2.82498956e-01 3.80525678e-01
-1.27498746e+00 -2.24132523e-01 7.09526837e-01 -6.92652702e-01
8.34675074e-01 3.00394654e-01 6.55784070e-01 1.48746479e+00
-7.64935166e-02 5.91564715e-01 9.03885663e-01 -1.33584753e-01
-1.54033929e-01 7.22069293e-02 -3.62854809e-01 6.93428874e-01
2.86801577e-01 2.85197999e-02 -3.33854288e-01 -3.33870620e-01
7.36437976e-01 4.97914463e-01 -5.43201387e-01 4.35622446e-02
-1.45939088e+00 7.94084132e-01 4.96767282e-01 4.32912380e-01
-5.75274169e-01 2.40488037e-01 2.55029500e-01 -1.05105765e-01
4.30229574e-01 3.18194330e-01 -2.05875620e-01 -4.30040024e-02
-9.49493229e-01 7.52660856e-02 5.50512671e-01 2.40438774e-01
7.95827508e-02 8.75085741e-02 -5.21881402e-01 9.75346267e-01
4.82140273e-01 7.78329909e-01 4.08021659e-01 -7.79144585e-01
3.51141632e-01 1.26432627e-01 -7.55194053e-02 -7.27374554e-01
-5.43868303e-01 -8.38258445e-01 -1.24333155e+00 -3.21282715e-01
1.09962441e-01 -9.65769589e-02 -8.54574263e-01 1.53177011e+00
4.70977068e-01 8.83107424e-01 -1.79197460e-01 6.74982131e-01
1.23579836e+00 7.52740204e-01 4.76891398e-02 -3.26158226e-01
1.39617372e+00 -1.04142022e+00 -8.17094743e-01 -2.39243601e-02
3.92997295e-01 -6.49558246e-01 9.56866860e-01 4.10637647e-01
-7.52410948e-01 -6.11636341e-01 -8.47890615e-01 4.82286289e-02
7.37552941e-02 2.06907019e-01 4.69010919e-02 5.42417347e-01
-9.00253356e-01 3.64114642e-01 -1.05095375e+00 -1.54718727e-01
6.40958309e-01 2.62331385e-02 -1.40748680e-01 -2.15746477e-01
-1.22834980e+00 3.92321289e-01 -2.10238516e-01 -1.43253624e-01
-1.15033937e+00 -8.68547440e-01 -4.46732372e-01 8.79997313e-02
2.73093879e-01 -1.09026432e+00 1.32329822e+00 -1.27436548e-01
-1.24618638e+00 8.06672752e-01 -6.06857426e-02 -2.56699294e-01
5.59839904e-01 -3.58392596e-01 -4.42027897e-01 5.73971689e-01
-2.38896757e-02 4.36060011e-01 1.15991330e+00 -1.08854008e+00
-6.18719757e-01 -7.92746395e-02 -3.67879897e-01 -2.40430608e-02
-4.61037248e-01 3.88982818e-02 -3.91731352e-01 -9.74647105e-01
-2.75628328e-01 -9.51637566e-01 -1.92401838e-02 7.95296803e-02
-4.05550480e-01 -1.48374274e-01 5.91772020e-01 -8.56553137e-01
1.40720606e+00 -2.31408429e+00 -1.14674874e-01 -2.88331807e-01
7.71290720e-01 5.57884693e-01 -5.46118766e-02 3.02601039e-01
5.04403822e-02 2.02088043e-01 -3.42339814e-01 -5.31270444e-01
-1.99075565e-01 2.30780314e-03 -3.62451285e-01 4.13186908e-01
2.21296519e-01 8.83113146e-01 -8.71320605e-01 -5.52945673e-01
1.83377191e-01 1.11827838e+00 -6.68375850e-01 6.71037793e-01
5.02010360e-02 7.67297566e-01 -3.58843416e-01 6.21781290e-01
3.93674046e-01 -8.43080103e-01 -1.44548967e-01 -6.30625784e-02
4.99563180e-02 7.10945189e-01 -5.44222593e-01 1.32465005e+00
-4.42976624e-01 4.89348292e-01 1.90353449e-02 -8.27834785e-01
4.06798363e-01 1.00113750e+00 8.57238472e-01 -1.94086924e-01
2.95610446e-02 2.61358798e-01 1.23995533e-02 -7.67095983e-01
-2.65163451e-01 -4.39422280e-01 2.66615301e-01 4.45894957e-01
-2.10092708e-01 -1.44977063e-01 -4.42583531e-01 -1.97894916e-01
1.34268785e+00 -4.21285659e-01 2.82614708e-01 3.34389240e-01
3.85171831e-01 -4.20368224e-01 3.04860383e-01 6.86220348e-01
-2.50351191e-01 1.18120265e+00 1.57078803e-02 -9.86503623e-03
-6.05685711e-01 -1.13252831e+00 -2.18402207e-01 8.55630994e-01
-3.41448396e-01 -3.19729745e-01 -7.14645326e-01 -8.48997772e-01
-1.46908417e-01 3.31023008e-01 -3.82439852e-01 -2.32563242e-01
-6.52953386e-01 -7.80510783e-01 7.90945172e-01 7.54934072e-01
4.27074730e-01 -1.03947675e+00 -6.36816800e-01 3.97619992e-01
-6.05997026e-01 -1.16010928e+00 -4.28981513e-01 1.18606044e-02
-7.30273485e-01 -9.53309119e-01 -1.20836163e+00 -6.08037353e-01
1.15611531e-01 5.72634339e-01 9.27531600e-01 9.60930064e-02
-7.14647710e-01 5.41204453e-01 -3.48704100e-01 -4.78786439e-01
-4.77570325e-01 1.31428897e-01 -5.74971251e-02 3.79810445e-02
4.60273325e-02 -4.09894586e-01 -1.11782992e+00 1.39667124e-01
-7.65330076e-01 -6.49179816e-02 6.13093555e-01 8.14471006e-01
8.23610663e-01 -1.46184728e-01 7.56926000e-01 -6.74160004e-01
5.34373403e-01 -8.37248743e-01 4.94594574e-02 -3.21701616e-02
-5.47860920e-01 -3.50402117e-01 6.47179484e-01 -4.55326974e-01
-6.66192174e-01 -2.79451311e-01 -6.34042442e-01 -7.76888847e-01
-1.97489068e-01 3.81616205e-01 1.87652424e-01 2.98899949e-01
4.99097884e-01 2.39194661e-01 -1.70961782e-01 -7.84122348e-01
6.32570460e-02 1.02059221e+00 6.23812079e-01 4.80080140e-04
7.92880714e-01 5.79398990e-01 -5.93134202e-03 -8.42409372e-01
-1.07219017e+00 -7.66093493e-01 -1.94655299e-01 -3.73776853e-02
1.26613927e+00 -1.22911978e+00 -3.42684925e-01 6.87374234e-01
-1.10321152e+00 -2.44574368e-01 -2.42742389e-01 8.63639474e-01
-2.49490470e-01 4.17359233e-01 -6.23707891e-01 -6.92772806e-01
-5.31721592e-01 -1.04846764e+00 1.15640819e+00 -2.11044595e-01
-1.85393408e-01 -7.50839651e-01 6.25488639e-01 6.90309763e-01
4.81047779e-01 1.91467047e-01 8.84219766e-01 -8.66032064e-01
-4.53064591e-01 -1.58611178e-01 -2.87668049e-01 5.70235491e-01
5.50460160e-01 -1.65498942e-01 -1.38763011e+00 -4.96803552e-01
3.59048665e-01 -3.74152392e-01 1.21854794e+00 3.71245533e-01
1.34845316e+00 -3.27936739e-01 -4.80998009e-01 7.35980272e-01
8.64649773e-01 2.21948028e-01 2.58717000e-01 -2.30684891e-01
1.08033013e+00 2.95483619e-01 2.60396302e-01 5.23049772e-01
3.61598700e-01 5.67814946e-01 4.28211868e-01 -3.45464736e-01
-5.66131532e-01 -1.61877185e-01 1.61369324e-01 1.23632801e+00
4.72189113e-03 -5.57114899e-01 -1.05563819e+00 6.25224710e-01
-1.36810148e+00 -8.10200155e-01 -7.08082244e-02 2.07118464e+00
8.17109108e-01 -2.17599705e-01 2.72602171e-01 4.26226377e-01
6.49877191e-01 3.56413722e-01 -4.69080180e-01 2.98360139e-01
4.09631610e-01 4.98540431e-01 -1.24491684e-01 2.68190265e-01
-1.40555108e+00 2.86392599e-01 6.45498276e+00 7.56462157e-01
-1.42945147e+00 4.04022783e-01 5.05527675e-01 -2.03854740e-01
-3.49667907e-01 -7.60743618e-01 -5.82634628e-01 5.65958977e-01
9.37937021e-01 3.39894712e-01 3.95150542e-01 5.30528307e-01
9.86161679e-02 5.27662575e-01 -9.22396839e-01 1.29183412e+00
1.28591567e-01 -1.06366515e+00 -1.24410100e-01 9.50623751e-02
6.21536613e-01 3.36710274e-01 3.74235928e-01 1.46302953e-01
-2.86315173e-01 -1.17201078e+00 1.58816695e-01 3.50585341e-01
1.25498712e+00 -5.10425448e-01 6.52886629e-01 2.35808268e-01
-1.35755002e+00 -3.08760889e-02 4.91268747e-02 4.12706792e-01
-2.79752221e-02 6.79044604e-01 -1.34257078e+00 4.08452153e-01
7.75781751e-01 6.74804866e-01 -3.37208062e-01 1.31645787e+00
-1.95200056e-01 1.19469333e+00 -2.83349305e-01 2.53936261e-01
-1.12444356e-01 3.19797039e-01 6.17742062e-01 1.30728900e+00
6.57601476e-01 -6.27235919e-02 3.73602808e-02 7.87294328e-01
-2.55417228e-01 5.42687029e-02 -5.66791892e-01 -2.72709310e-01
4.16837335e-01 1.12576938e+00 -2.91414946e-01 -3.03749681e-01
-5.76280296e-01 8.67573380e-01 -6.43583685e-02 2.64350504e-01
-1.11785996e+00 -1.64166823e-01 5.89127719e-01 3.67392004e-01
5.28699696e-01 1.84381381e-01 6.76442981e-02 -9.90539312e-01
-2.02430531e-01 -1.05817080e+00 4.18253839e-01 -4.94001538e-01
-1.27113235e+00 6.85594678e-01 -1.96748346e-01 -1.41743290e+00
-3.86448592e-01 -3.43783915e-01 -5.55514634e-01 6.84430361e-01
-1.67716455e+00 -9.26621437e-01 -5.81228793e-01 5.23599088e-01
4.31442589e-01 -8.55818614e-02 9.51738536e-01 6.83276534e-01
-5.79972446e-01 6.60275996e-01 2.10101187e-01 1.07287362e-01
8.75134349e-01 -1.07940638e+00 4.79476064e-01 4.83983457e-01
3.13701153e-01 3.78035814e-01 3.86977583e-01 -4.63902891e-01
-9.77981269e-01 -1.53831780e+00 8.86589170e-01 -4.64820832e-01
3.61178905e-01 -1.60504684e-01 -1.30234694e+00 3.86756331e-01
6.15985580e-02 3.16948771e-01 9.02989507e-01 -4.43968564e-01
-5.18064916e-01 -1.98966578e-01 -1.02968407e+00 2.40940273e-01
8.14613044e-01 -9.56166863e-01 -5.78749716e-01 3.32172066e-01
9.51991796e-01 -1.34818926e-01 -7.56033540e-01 8.04534614e-01
5.12525439e-01 -7.76962757e-01 1.37328851e+00 -4.15320426e-01
3.81000340e-01 -2.07760796e-01 -2.72017680e-02 -1.13685584e+00
-5.14607906e-01 -5.28272212e-01 -4.45213765e-01 9.85304892e-01
9.52058583e-02 -5.66810191e-01 4.87237751e-01 -1.16888314e-01
-1.02803394e-01 -9.29446101e-01 -8.77099454e-01 -5.34854889e-01
-5.16052842e-02 -4.99916792e-01 2.42213085e-01 8.52017224e-01
-3.66150528e-01 3.80111486e-01 -5.60536265e-01 1.20429017e-01
3.79780889e-01 2.01414172e-02 3.86077255e-01 -1.40913463e+00
-8.28291357e-01 -3.95927161e-01 1.19640358e-01 -1.08090067e+00
-1.97988540e-01 -8.87045324e-01 1.90858454e-01 -1.89566791e+00
4.42748308e-01 -2.06682310e-01 -7.94490635e-01 3.78826290e-01
-6.73803866e-01 5.62197864e-01 1.12190932e-01 4.01918024e-01
-4.83023047e-01 4.60643381e-01 1.40280259e+00 -2.31056854e-01
-7.95943215e-02 2.04044580e-01 -5.88370264e-01 6.78932250e-01
8.64552438e-01 -7.31827497e-01 -4.61026698e-01 -1.91626027e-01
1.15026660e-01 2.02408135e-01 6.59695029e-01 -1.14279473e+00
-1.42837182e-01 1.83585256e-01 1.95156798e-01 -6.90737486e-01
5.61831236e-01 -6.63081527e-01 2.18145758e-01 8.22340429e-01
-3.60798419e-01 -2.74102986e-01 1.27460122e-01 8.99801612e-01
-2.46554464e-01 1.23362854e-01 8.88292432e-01 8.05610046e-02
2.68583953e-01 5.22830665e-01 -5.13897300e-01 4.83780950e-01
6.59794450e-01 3.29428136e-01 -4.49499726e-01 -3.82970244e-01
-4.21829075e-01 -1.09622113e-01 -7.99581595e-03 2.04281151e-01
7.12336540e-01 -1.33432126e+00 -9.02252674e-01 3.06708425e-01
1.02024361e-01 1.84242710e-01 4.42574471e-01 1.23117340e+00
-4.64974195e-01 5.61888993e-01 2.79415309e-01 -8.74084771e-01
-1.40141773e+00 6.27664208e-01 3.85599345e-01 -3.97388965e-01
-7.60525525e-01 9.73605037e-01 7.39517033e-01 -1.52172610e-01
4.59023446e-01 -5.72452426e-01 -3.76244515e-01 9.83131155e-02
6.08817697e-01 5.59227884e-01 8.10009912e-02 -4.04801607e-01
-5.22793889e-01 6.01916909e-01 9.33890045e-02 1.17912553e-02
1.23266411e+00 1.46344215e-01 1.93729416e-01 5.11639357e-01
1.43465960e+00 2.03673944e-01 -9.75487351e-01 -2.14999333e-01
-4.87149388e-01 -3.00470233e-01 5.83730228e-02 -7.88257241e-01
-1.06442404e+00 1.43105149e+00 8.24105442e-01 4.01225239e-01
1.12547600e+00 2.34766930e-01 1.25872397e+00 3.71794134e-01
-2.73446530e-01 -4.84909534e-01 5.91130018e-01 1.65675953e-01
9.30894434e-01 -1.16745400e+00 -3.20572466e-01 -2.87231356e-01
-4.79492068e-01 5.58792233e-01 1.19894415e-01 1.31103948e-01
7.34901130e-01 2.08336234e-01 2.57870466e-01 -1.19882941e-01
-8.35247874e-01 -2.07545936e-01 5.94878852e-01 5.65475583e-01
6.09356165e-01 1.00152440e-01 2.62189209e-02 8.42264712e-01
2.04924315e-01 3.90549675e-02 -5.40361591e-02 5.04685938e-01
-4.87465233e-01 -8.90838385e-01 -4.77369219e-01 6.15968943e-01
-1.08225584e+00 -2.66647905e-01 -4.48343039e-01 4.93976563e-01
1.60184816e-01 1.03789818e+00 -6.95107803e-02 -5.46616316e-01
1.54013023e-01 2.17683688e-01 3.13116848e-01 -7.25520730e-01
-5.58170676e-01 5.80171645e-01 -5.09840474e-02 -3.29179376e-01
-3.87550563e-01 -5.12760103e-01 -1.16117299e+00 2.37275690e-01
-1.62544176e-01 -1.36144146e-01 2.59763211e-01 6.79324031e-01
4.47106957e-01 1.07026541e+00 8.54375720e-01 -5.41155517e-01
-5.91593623e-01 -9.24625576e-01 -2.91601688e-01 2.42827132e-01
9.90827620e-01 -5.38712680e-01 -7.57822812e-01 -8.59052874e-03] | [14.568245887756348, 3.900416374206543] |
9ac04e26-3636-4c0d-8d92-494cd5fd536d | auto-encoder-based-co-training-multi-view | 2201.02978 | null | https://arxiv.org/abs/2201.02978v1 | https://arxiv.org/pdf/2201.02978v1.pdf | Auto-Encoder based Co-Training Multi-View Representation Learning | Multi-view learning is a learning problem that utilizes the various representations of an object to mine valuable knowledge and improve the performance of learning algorithm, and one of the significant directions of multi-view learning is sub-space learning. As we known, auto-encoder is a method of deep learning, which can learn the latent feature of raw data by reconstructing the input, and based on this, we propose a novel algorithm called Auto-encoder based Co-training Multi-View Learning (ACMVL), which utilizes both complementarity and consistency and finds a joint latent feature representation of multiple views. The algorithm has two stages, the first is to train auto-encoder of each view, and the second stage is to train a supervised network. Interestingly, the two stages share the weights partly and assist each other by co-training process. According to the experimental result, we can learn a well performed latent feature representation, and auto-encoder of each view has more powerful reconstruction ability than traditional auto-encoder. | ['Xin Zuo', 'Hao-jie Xie', 'Yuan-Fang Wang', 'Jian-wei Liu', 'Run-kun Lu'] | 2022-01-09 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-2.81242073e-01 -1.04548559e-01 -3.38751376e-01 -4.51342434e-01
-5.53609550e-01 -2.21765116e-01 5.27786255e-01 -5.39006233e-01
1.17355727e-01 4.17366028e-01 6.98413968e-01 3.17029715e-01
-6.94093779e-02 -7.70537853e-01 -6.80610120e-01 -6.87291205e-01
2.33020127e-01 2.56153554e-01 1.77961186e-01 1.65443625e-02
7.36972764e-02 1.56044915e-01 -1.53842950e+00 8.34143639e-01
3.79900157e-01 9.24561679e-01 3.97316515e-01 4.35122639e-01
-2.77111083e-01 1.16648149e+00 -1.40965162e-02 -3.23447794e-01
-1.28383865e-03 -4.83424693e-01 -6.22879386e-01 3.46530259e-01
-9.09601003e-02 -6.38588846e-01 -5.21249831e-01 8.46212387e-01
3.81958574e-01 -1.83010682e-01 5.93364120e-01 -1.07616723e+00
-8.91909063e-01 4.99166042e-01 -6.63843274e-01 1.15167327e-01
1.40166819e-01 -2.44681403e-01 9.93982077e-01 -1.28339255e+00
4.65927869e-01 1.41959703e+00 6.22595847e-01 3.44059527e-01
-6.37615263e-01 -5.33685625e-01 1.36327192e-01 5.10299921e-01
-1.10313976e+00 -3.31977993e-01 1.11247492e+00 -4.52213377e-01
6.35421693e-01 -2.97426283e-01 8.63384664e-01 1.09944093e+00
3.94949734e-01 1.35096276e+00 1.22727716e+00 -2.98018128e-01
-1.60708725e-01 3.99155617e-01 -2.12463304e-01 1.01055753e+00
-9.34106708e-02 1.82161748e-01 -4.80287075e-01 -1.04819890e-02
9.07933235e-01 7.83911347e-01 -2.54221857e-01 -7.77254164e-01
-1.23085034e+00 1.03426731e+00 3.95334184e-01 3.38490844e-01
-2.98630625e-01 -8.23799148e-02 6.39171958e-01 2.80484766e-01
4.07759190e-01 -1.48467690e-01 -4.94396538e-01 6.17547780e-02
-7.44546056e-01 -2.87084639e-01 4.99845982e-01 8.16736817e-01
8.38814914e-01 1.39204472e-01 2.44966283e-01 8.12982857e-01
5.97543716e-01 3.36222023e-01 9.98408258e-01 -5.78495920e-01
4.66131300e-01 9.64467168e-01 -3.59053135e-01 -7.26766169e-01
-1.19107649e-01 -5.43524623e-01 -1.12468064e+00 1.70084551e-01
-2.75135070e-01 -2.54566163e-01 -6.05073214e-01 1.36297834e+00
1.84024364e-01 3.34663481e-01 4.11630750e-01 7.14228272e-01
1.01012695e+00 8.17669690e-01 -4.20759678e-01 -5.68889022e-01
1.20591486e+00 -1.35781586e+00 -8.16390395e-01 -7.87471533e-02
2.67450511e-01 -6.54232204e-01 5.72997570e-01 5.25124550e-01
-9.89216924e-01 -1.01277804e+00 -1.29547691e+00 5.04710935e-02
-3.31900120e-01 3.93134952e-01 8.43629718e-01 -1.69247314e-01
-5.30311942e-01 4.92566466e-01 -7.05280900e-01 4.63495627e-02
4.27286118e-01 1.51416868e-01 -7.92993069e-01 -3.81703794e-01
-1.05800509e+00 8.30704808e-01 6.24881983e-01 -1.76894490e-03
-1.12899375e+00 -4.04143691e-01 -1.05416739e+00 1.40999034e-01
3.16205293e-01 -7.12056756e-01 8.91523898e-01 -1.14380336e+00
-1.38951147e+00 5.26950359e-01 -2.37554312e-01 -1.17041329e-02
2.05389574e-01 -4.84618247e-01 -7.04503298e-01 7.57302642e-02
2.55285412e-01 2.03568205e-01 1.15897071e+00 -1.47611189e+00
-5.81643105e-01 -6.43764913e-01 -1.31705523e-01 3.39479208e-01
-3.92361075e-01 -2.69766361e-01 -5.43817699e-01 -2.61351734e-01
3.72720182e-01 -5.34065247e-01 -4.47905324e-02 -3.16345125e-01
-1.10182529e-02 -3.14315557e-01 1.21436322e+00 -5.92869043e-01
1.06714940e+00 -2.27083468e+00 5.02697647e-01 -3.34812291e-02
3.87346864e-01 2.45031923e-01 -2.84817792e-03 6.42678142e-01
-4.67720449e-01 -3.17765385e-01 1.06364891e-01 -1.82787910e-01
-4.84872192e-01 3.92406762e-01 -3.03015053e-01 3.82531464e-01
-1.67394802e-02 9.48923230e-01 -9.78359580e-01 -7.11592555e-01
3.39494914e-01 4.16037530e-01 -4.15742666e-01 7.58435249e-01
1.08053923e-01 2.99088657e-01 -5.43112576e-01 4.12018538e-01
5.37388563e-01 -6.78697586e-01 2.33060330e-01 -5.71895421e-01
-1.29233431e-02 -2.08038896e-01 -1.48908412e+00 2.01441312e+00
-6.94335461e-01 1.93870395e-01 -1.79068506e-01 -1.11679745e+00
9.03481483e-01 5.51140606e-01 6.27527654e-01 -4.88502204e-01
6.72796518e-02 1.27362922e-01 -3.01532269e-01 -1.00292158e+00
-4.39365506e-02 -4.37977314e-01 3.44749987e-01 7.23583579e-01
7.91403413e-01 3.69306445e-01 -3.56828481e-01 9.46252644e-02
6.23546720e-01 4.20790464e-01 7.29203641e-01 2.39573449e-01
8.89429986e-01 -5.29557645e-01 6.69241905e-01 1.06204383e-01
2.88600534e-01 4.48567986e-01 2.67703116e-01 -8.15886855e-01
-1.06942427e+00 -9.75025117e-01 1.13259852e-01 8.98015678e-01
1.17826648e-01 -4.67028707e-01 -2.16242597e-01 -1.11627376e+00
3.17453258e-02 3.50135803e-01 -7.63576686e-01 -3.32297385e-01
-3.75617862e-01 -3.50922376e-01 -5.26240319e-02 9.04507935e-01
7.91572332e-01 -9.90697563e-01 -3.30163926e-01 2.44720802e-01
-1.78229641e-02 -9.69208956e-01 -2.59473771e-01 3.17555159e-01
-1.04256713e+00 -1.27737939e+00 -6.46338344e-01 -8.52006078e-01
7.28852749e-01 6.88159287e-01 1.03997612e+00 1.16167948e-01
5.06715551e-02 4.52608287e-01 -4.83206183e-01 -3.06168020e-01
-2.50280678e-01 -1.85274854e-01 1.97857231e-01 3.38033706e-01
3.68335575e-01 -1.07198536e+00 -3.84550542e-01 1.21744439e-01
-7.92745531e-01 2.97663510e-01 1.04709923e+00 1.19556081e+00
7.58942842e-01 5.42366356e-02 6.09404027e-01 -1.09043121e+00
2.34172136e-01 -5.69362462e-01 -4.37890232e-01 5.89762509e-01
-6.16340876e-01 3.14675242e-01 9.44741428e-01 -7.73144662e-02
-1.14640689e+00 2.88019419e-01 -1.08330555e-01 -1.15201843e+00
-1.40940145e-01 5.90236604e-01 -5.45785069e-01 1.25790864e-01
1.44059837e-01 9.02162611e-01 1.94771796e-01 -6.57771409e-01
6.11514926e-01 6.24581456e-01 3.56899738e-01 -1.58541780e-02
7.39301145e-01 4.72014904e-01 -2.70534176e-02 -3.60452026e-01
-1.28481174e+00 -5.72139561e-01 -1.01770639e+00 -2.22249284e-01
9.69776928e-01 -1.34757769e+00 -6.22661531e-01 2.65468270e-01
-1.01398349e+00 4.75426704e-01 -1.55015096e-01 9.05799508e-01
-7.56942749e-01 4.57811445e-01 -2.85541773e-01 -6.48611784e-01
-1.67570964e-01 -1.03191018e+00 9.75040376e-01 2.42907047e-01
4.75888759e-01 -1.24715126e+00 2.52668619e-01 2.64103234e-01
-2.69662272e-02 3.06772497e-02 7.75804698e-01 -8.06647718e-01
-4.26917791e-01 -8.33653361e-02 -1.25764593e-01 6.30291939e-01
3.28993946e-01 -3.52151632e-01 -1.15111184e+00 -4.94387716e-01
6.26056314e-01 -7.14271069e-01 1.00698400e+00 1.90403894e-01
1.34336352e+00 -4.02899653e-01 -3.35026592e-01 7.71707058e-01
1.83930898e+00 9.24053118e-02 4.76892978e-01 1.20109096e-01
8.39446485e-01 3.07827324e-01 4.57093984e-01 4.57682669e-01
5.46018004e-01 2.87306994e-01 5.66904187e-01 -9.05920789e-02
6.85409904e-02 -6.72974408e-01 3.81637126e-01 1.62191319e+00
-2.56550252e-01 2.54427820e-01 -3.47343743e-01 4.40355957e-01
-1.99667096e+00 -1.42556894e+00 1.65813684e-01 1.91370440e+00
4.29892957e-01 1.63824439e-01 2.53822524e-02 -1.30838361e-02
4.72119540e-01 4.29097235e-01 -5.76878965e-01 -1.20313816e-01
1.60520956e-01 -2.50280619e-01 -2.31631044e-02 2.59884536e-01
-1.13853359e+00 6.24426067e-01 6.19278240e+00 5.85928202e-01
-1.06533289e+00 1.89598009e-01 2.75530338e-01 8.12094286e-02
-4.75626588e-01 -2.88766120e-02 -7.05026627e-01 3.30167860e-01
5.20200193e-01 9.25757885e-02 3.87061983e-01 1.25268328e+00
-2.44982049e-01 2.47127965e-01 -1.31504238e+00 1.33547938e+00
5.99129319e-01 -1.52506888e+00 2.84509182e-01 -4.26563397e-02
7.19374537e-01 -8.09447095e-03 -1.44423679e-01 5.13487160e-01
2.38889039e-01 -8.35457504e-01 2.54886299e-01 8.91749144e-01
6.51702225e-01 -1.00756574e+00 9.93086398e-01 8.50117445e-01
-1.58987844e+00 -3.39891940e-01 -6.36044443e-01 1.30752474e-01
1.10487022e-01 4.58878785e-01 -3.40105176e-01 1.18709266e+00
5.30616164e-01 1.48436356e+00 -5.87070525e-01 6.86365485e-01
-2.51941085e-01 3.73680502e-01 2.95235395e-01 3.46999675e-01
2.39561141e-01 -2.50301063e-01 2.66740829e-01 8.32634926e-01
1.27137929e-01 2.11625427e-01 4.32570279e-01 6.85209692e-01
8.30215141e-02 -1.30274519e-01 -1.01250160e+00 9.03075784e-02
1.93212911e-01 1.43147171e+00 -2.02722758e-01 -5.17751694e-01
-1.04171789e+00 1.20360887e+00 6.76627517e-01 1.12517081e-01
-7.23357797e-01 -1.40138209e-01 2.44394485e-02 -2.69490868e-01
6.74865663e-01 -3.85776237e-02 1.83049422e-02 -1.51366282e+00
1.64869260e-02 -8.72466862e-01 4.44398701e-01 -9.46330011e-01
-1.44413197e+00 4.83299911e-01 -1.01960152e-01 -1.82653964e+00
-4.58903700e-01 -5.57985961e-01 -7.12433398e-01 8.83679807e-01
-1.59574401e+00 -1.61152613e+00 -3.39914143e-01 8.76192272e-01
9.17056859e-01 -9.38665986e-01 9.08477783e-01 1.93211406e-01
-2.53291100e-01 4.59325552e-01 3.52118701e-01 1.90641060e-01
4.90321934e-01 -1.20749533e+00 -1.18906304e-01 5.55445373e-01
4.37762260e-01 5.21810532e-01 -9.92988050e-03 -4.40440655e-01
-1.72099745e+00 -1.02064836e+00 6.18739307e-01 -3.88148665e-01
3.59885275e-01 -9.35718417e-02 -8.55494738e-01 9.93168831e-01
3.72530639e-01 3.78384292e-01 1.14528465e+00 2.08688408e-01
-3.81313741e-01 -2.47259542e-01 -6.64735019e-01 -3.94696519e-02
7.51442909e-01 -7.82871902e-01 -1.00554621e+00 2.40933865e-01
7.77763069e-01 -3.12544674e-01 -1.07285666e+00 2.62811959e-01
6.87065661e-01 -1.40385449e+00 1.01659703e+00 -8.95035982e-01
8.97880852e-01 -1.98169217e-01 -2.93823808e-01 -1.53853369e+00
-8.52860689e-01 1.04255825e-01 -5.89869142e-01 1.32866704e+00
5.65697998e-02 -3.04906756e-01 6.04531109e-01 -2.61197031e-01
-3.30073863e-01 -1.32014251e+00 -5.06455302e-01 -5.60503185e-01
-3.18537623e-01 4.02143076e-02 6.15568876e-01 1.07364273e+00
-1.22275397e-01 7.93516636e-01 -8.32227826e-01 1.89947426e-01
5.85656643e-01 6.30626798e-01 6.96448445e-01 -1.24636924e+00
-6.26144886e-01 3.88587192e-02 -2.93787360e-01 -1.10275316e+00
2.24667072e-01 -9.76765275e-01 -3.10895383e-01 -1.70930803e+00
7.06102669e-01 1.09064385e-01 -4.15841669e-01 3.64598036e-01
-3.00476015e-01 -2.94523835e-01 -1.85526032e-02 5.25666416e-01
-8.84436786e-01 9.35627043e-01 1.69164371e+00 1.01736382e-01
7.05284774e-02 1.91165805e-01 -7.39594400e-01 1.07379317e+00
2.72568077e-01 -2.68612891e-01 -6.75869048e-01 -4.59689617e-01
1.35975793e-01 4.83446717e-01 2.73566455e-01 -9.33636606e-01
2.60147631e-01 -6.56173900e-02 1.09800303e+00 -8.72669160e-01
3.77444208e-01 -1.26960802e+00 -4.74343076e-02 4.96196628e-01
-1.95670560e-01 2.50336707e-01 -2.87531018e-01 8.56061816e-01
-6.65100098e-01 -2.09673554e-01 6.09337687e-01 -6.87898815e-01
-1.05802035e+00 4.95880663e-01 -1.39296649e-03 1.65282916e-02
9.88465667e-01 -1.85868174e-01 8.38949755e-02 -2.81420708e-01
-7.64243722e-01 3.59128505e-01 3.48961237e-03 5.71905851e-01
1.01828706e+00 -1.97118652e+00 -5.50837815e-01 5.73345542e-01
1.79044485e-01 -2.18790956e-03 3.71840328e-01 5.41210890e-01
-9.58235040e-02 1.97705254e-01 -4.60663706e-01 -6.57016575e-01
-1.01118779e+00 9.39132392e-01 3.14429849e-01 -4.71433729e-01
-8.12554896e-01 8.38893294e-01 2.56359965e-01 -3.89121860e-01
3.24047431e-02 4.20547336e-01 -7.23397911e-01 2.55276561e-01
6.84810817e-01 2.29348585e-01 -2.55203515e-01 -6.00668669e-01
-1.66603580e-01 7.12223172e-01 -3.49085897e-01 1.37945712e-01
1.75295210e+00 -5.78482747e-02 -2.38586873e-01 9.55296576e-01
1.68706250e+00 -2.26533085e-01 -1.19941866e+00 -5.72404742e-01
-4.68331963e-01 -4.77906853e-01 4.51892167e-02 -4.84848768e-01
-1.31287503e+00 1.21637976e+00 5.93042374e-01 6.61160098e-03
1.10412276e+00 -7.32306689e-02 7.65269339e-01 4.69721079e-01
2.87255615e-01 -9.85741317e-01 6.57020926e-01 6.50525212e-01
9.26854968e-01 -1.44003141e+00 2.73235768e-01 -5.15932478e-02
-8.47913861e-01 1.37451077e+00 1.05713129e+00 -2.62394309e-01
1.11810768e+00 -5.58294244e-02 -1.91223070e-01 -4.85934407e-01
-9.30107951e-01 4.52587497e-04 3.49506706e-01 3.84065926e-01
2.83386707e-01 -2.51027256e-01 1.43076805e-02 9.82249141e-01
7.21866116e-02 1.30691484e-01 6.92540333e-02 7.63456166e-01
-4.89506900e-01 -1.02652693e+00 -3.34379449e-02 5.65671146e-01
-1.87568605e-01 1.31609231e-01 -1.03509806e-01 6.28688216e-01
5.15876830e-01 4.30499434e-01 -6.55414686e-02 -8.66396606e-01
2.07443804e-01 1.95304021e-01 4.11936104e-01 -7.37866819e-01
-3.53175163e-01 1.92394376e-01 -3.31690490e-01 -6.30196095e-01
-6.59180880e-01 -5.26756465e-01 -9.83352184e-01 -5.91725530e-03
-4.46127623e-01 3.36550057e-01 2.87838578e-01 1.32567465e+00
1.94133490e-01 7.22200274e-01 1.27406240e+00 -6.03020310e-01
-6.49441361e-01 -8.86438370e-01 -7.94469118e-01 3.53058130e-01
3.36980224e-01 -8.74357820e-01 -1.98094130e-01 1.11299336e-01] | [8.458002090454102, 4.557337284088135] |
54491e36-a913-4dbf-b978-039cbdf8ed5f | suggestion-miner-at-semeval-2019-task-9 | null | null | https://aclanthology.org/S19-2218 | https://aclanthology.org/S19-2218.pdf | Suggestion Miner at SemEval-2019 Task 9: Suggestion Detection in Online Forum using Word Graph | This paper describes the suggestion miner system that participates in SemEval 2019 Task 9 - SubTask A - Suggestion Mining from Online Reviews and Forums. The system participated in the subtasks A. This paper discusses the results of our system in the development, evaluation and post evaluation. Each class in the dataset is represented as directed unweighted graphs. Then, the comparison is carried out with each class graph which results in a vector. This vector is used as features by a machine learning algorithm. The model is evaluated on hold on strategy. The organizers randomly split (8500 instances) training set (provided to the participant in training their system) and testing set (833 instances). The test set is reserved to evaluate the performance of participants systems. During the evaluation, our system ranked 31 in the Coda Lab result of the subtask A (binary class problem). The binary class system achieves evaluation value 0.34, precision 0.87, recall 0.73 and F measure 0.78. | ['Luqman Ahmed', 'Syed Jawad Hussain', 'Humera Liaquat', 'Usman Ahmed'] | 2019-06-01 | null | null | null | semeval-2019-6 | ['suggestion-mining'] | ['natural-language-processing'] | [ 3.94245470e-03 7.51374185e-01 -2.35681504e-01 -7.16673493e-01
-4.89293039e-01 -4.63803291e-01 8.71326804e-01 5.84155679e-01
-5.90519369e-01 7.84408689e-01 -1.85075011e-02 -6.25162601e-01
-2.61218220e-01 -6.52251959e-01 -2.21331879e-01 -3.23915601e-01
-1.67159408e-01 9.13669109e-01 3.28543812e-01 -3.78045678e-01
8.99603963e-01 -7.88185447e-02 -1.79797292e+00 7.12373614e-01
6.62078857e-01 8.61778975e-01 2.20374968e-02 7.68830359e-01
-5.45380488e-02 4.53225136e-01 -7.75028229e-01 -4.95094925e-01
3.84959966e-01 -2.15653881e-01 -1.02337384e+00 1.66886091e-01
3.11400712e-01 2.60763258e-01 2.87543356e-01 8.93052101e-01
1.11035243e-01 3.38955998e-01 9.46850598e-01 -1.38705373e+00
-7.78757855e-02 1.10499442e+00 -4.41305697e-01 1.02086784e-02
5.04354596e-01 -3.95105511e-01 1.29614031e+00 -1.28586519e+00
7.19005108e-01 9.99389112e-01 2.24373773e-01 3.23679537e-01
-1.07039118e+00 -5.00873923e-01 1.00791506e-01 1.21301681e-01
-1.15241468e+00 -2.01089755e-01 5.13768673e-01 -4.34763014e-01
1.28850675e+00 5.30498147e-01 6.60935402e-01 3.00355732e-01
1.68679357e-01 4.59667593e-01 1.19282663e+00 -7.13675797e-01
4.59959328e-01 1.16229212e+00 9.58940089e-01 6.98344409e-01
5.42063773e-01 -2.67247200e-01 -6.24533713e-01 -3.01345199e-01
-1.13813043e-01 -3.13887745e-01 9.38879773e-02 3.56358923e-02
-6.66673064e-01 8.66304576e-01 9.94697139e-02 3.61549944e-01
-4.82678175e-01 -5.61304450e-01 2.62326896e-01 8.88775051e-01
6.33497596e-01 7.31218934e-01 -6.56647027e-01 2.24233538e-01
-7.70698011e-01 6.90196753e-01 1.43569672e+00 8.68693352e-01
7.96349108e-01 -4.64526564e-01 -8.26177299e-02 9.43799078e-01
5.33951342e-01 -9.88386199e-02 7.82014489e-01 -5.55504382e-01
5.09784400e-01 1.08158088e+00 1.52866200e-01 -9.70667422e-01
-5.64605117e-01 -3.18634808e-01 -3.25049996e-01 2.56934047e-01
1.03514276e-01 -3.95760506e-01 -5.88937163e-01 1.00152349e+00
2.80821770e-01 -6.98913813e-01 -4.92149219e-02 6.31314039e-01
1.23418832e+00 4.74469990e-01 1.00290244e-02 -5.42031348e-01
1.28789353e+00 -1.25549257e+00 -5.64603388e-01 2.26005808e-01
9.15748835e-01 -1.00619030e+00 8.60936999e-01 8.67812872e-01
-1.10335052e+00 -8.34645212e-01 -1.32268047e+00 4.90721762e-01
-7.39003360e-01 2.53512442e-01 4.83542472e-01 7.37175226e-01
-1.24692976e+00 5.04428983e-01 -6.67417124e-02 -4.91611719e-01
-2.34492481e-01 7.90453494e-01 -4.75082248e-01 1.93543002e-01
-1.02751815e+00 7.89910555e-01 2.72004366e-01 -3.95834804e-01
-4.07250434e-01 -2.72176057e-01 -6.19274437e-01 -7.51217157e-02
1.49862424e-01 -2.63154715e-01 1.34502316e+00 -9.55056429e-01
-1.22777438e+00 1.07595932e+00 9.08683613e-02 -5.51857531e-01
3.71570975e-01 7.79850334e-02 -5.32503307e-01 -4.02096152e-01
3.88769656e-01 4.86836076e-01 4.52374965e-01 -1.10444224e+00
-1.16640818e+00 -3.94065857e-01 1.51781276e-01 4.39473003e-01
-4.02388453e-01 2.26957142e-01 -2.69030184e-01 -1.76323563e-01
3.88203785e-02 -8.41484904e-01 -3.01529825e-01 -1.04845548e+00
-6.66346371e-01 -8.59732807e-01 3.47498208e-01 -2.01792985e-01
1.50933814e+00 -1.70664251e+00 -3.92295867e-01 7.49142587e-01
3.15412104e-01 6.54545426e-02 1.83076151e-02 6.34488046e-01
-2.98955530e-01 5.02007723e-01 2.81913191e-01 -2.75627702e-01
-1.03668720e-01 -3.26039493e-01 5.53909689e-04 1.94551736e-01
-2.18892828e-01 2.98929721e-01 -5.39396226e-01 -4.32855427e-01
-5.11148795e-02 -3.31366599e-01 -1.78583086e-01 3.38018239e-01
-9.34121981e-02 -4.78039145e-01 -4.93402600e-01 3.82234484e-01
4.56123829e-01 -2.11492673e-01 3.99052411e-01 -4.68045147e-03
-3.13026071e-01 4.26969588e-01 -1.15751314e+00 1.12385154e+00
-2.71222256e-02 4.31016117e-01 -3.04815233e-01 -1.06215894e+00
1.43223274e+00 1.85898229e-01 4.01206702e-01 -6.39445305e-01
1.22575767e-01 1.75295368e-01 2.40984306e-01 -4.20223892e-01
8.48747313e-01 3.73116463e-01 5.49147315e-02 9.17378306e-01
1.83599591e-02 -1.68315470e-01 6.60027623e-01 6.63047314e-01
1.18943453e+00 -2.84361035e-01 3.80504489e-01 -5.97130001e-01
7.47566104e-01 4.10399795e-01 -1.00826301e-01 8.96225274e-01
1.87903065e-02 3.75384331e-01 5.72860301e-01 -6.42394006e-01
-7.25937426e-01 -5.20469308e-01 -1.18783265e-02 1.23526204e+00
-1.57814205e-01 -1.03432357e+00 -6.92840159e-01 -1.08946812e+00
2.97889151e-02 7.01169789e-01 -9.75276768e-01 1.76818550e-01
-8.83795545e-02 -6.19859099e-01 -7.72984847e-02 -8.33648816e-02
1.95951506e-01 -1.13845456e+00 -3.87203425e-01 1.54212728e-01
2.55606681e-01 -5.66668510e-01 -5.71080968e-02 4.31742400e-01
-7.42304623e-01 -1.42749727e+00 -2.30957657e-01 -9.66166675e-01
6.84015274e-01 2.60920197e-01 1.26225924e+00 3.91138196e-01
-2.47247964e-01 3.96897614e-01 -5.93474686e-01 -9.12238061e-01
-2.45851338e-01 3.74459893e-01 8.45042691e-02 -2.19834149e-01
7.64480650e-01 -1.86957791e-01 -5.00586450e-01 5.55889130e-01
-3.92465502e-01 -9.11210477e-02 4.05384749e-01 7.31028378e-01
4.96556312e-01 2.87901070e-02 7.79427350e-01 -1.55501831e+00
1.29953933e+00 -6.22653663e-01 -6.51850462e-01 3.61286223e-01
-1.46987295e+00 -2.62383699e-01 2.66150177e-01 9.67237502e-02
-7.99717546e-01 -8.28599278e-03 3.98981720e-01 5.30472815e-01
-6.24334961e-02 8.12342167e-01 2.03506619e-01 1.24496564e-01
1.14547408e+00 -4.27190870e-01 3.71088348e-02 -3.37840140e-01
1.14692479e-01 9.27045584e-01 4.11781035e-02 -2.73522347e-01
4.82215405e-01 -3.22022229e-01 -2.66542077e-01 -7.89136648e-01
-6.14617884e-01 -8.09692800e-01 -3.64984572e-01 -4.71812367e-01
3.04567963e-01 -5.23670673e-01 -5.18606484e-01 -5.31292781e-02
-9.03894186e-01 -1.45151779e-01 -3.09875131e-01 3.74131978e-01
-1.10250212e-01 -1.63153023e-01 -3.72188509e-01 -1.11665213e+00
-7.67070949e-01 -7.08611727e-01 4.12504673e-01 3.35677028e-01
-6.35295749e-01 -7.46816099e-01 3.20695102e-01 4.98974413e-01
1.82672992e-01 -1.77707911e-01 8.00794363e-01 -1.66993165e+00
1.48213074e-01 -7.31221199e-01 -1.11926422e-02 3.05084974e-01
-3.81548479e-02 1.81931913e-01 -9.95404959e-01 1.74328964e-02
-1.64953902e-01 -5.58855534e-01 7.19200373e-01 1.34020254e-01
1.07444251e+00 -1.47780284e-01 -3.83760840e-01 -4.54394788e-01
1.04217827e+00 3.28893602e-01 1.88642606e-01 5.31002045e-01
1.80102825e-01 1.26711142e+00 1.19616508e+00 4.06699866e-01
3.67271096e-01 5.17539084e-01 1.87856108e-01 2.06376523e-01
2.46007323e-01 -7.92742334e-03 3.31135869e-01 8.64586890e-01
-1.68409094e-01 -4.02454346e-01 -9.14683342e-01 4.72233146e-01
-1.99403119e+00 -5.40388823e-01 -5.63043356e-01 2.22238851e+00
6.30633414e-01 6.20531142e-01 5.06386399e-01 5.09631991e-01
6.99213266e-01 -2.40749568e-01 7.48399831e-03 -9.10085797e-01
1.51909664e-01 2.90679246e-01 -2.62432527e-02 7.32645512e-01
-9.01057541e-01 6.99167132e-01 5.96796465e+00 5.10786474e-01
-6.36890292e-01 -2.32870281e-01 9.37770545e-01 1.91216365e-01
-4.33287531e-01 -6.15074337e-02 -9.67513382e-01 1.80145249e-01
1.34903133e+00 -6.14222348e-01 1.67394392e-02 8.02404881e-01
1.08410306e-02 -4.89760458e-01 -1.12487173e+00 5.48922360e-01
2.69257993e-01 -1.29207480e+00 5.86045496e-02 1.82461068e-01
7.24729955e-01 2.00451929e-02 1.10453106e-02 7.09387124e-01
2.73547560e-01 -1.09780228e+00 3.89342636e-01 3.38777333e-01
4.51454222e-01 -9.99707997e-01 1.20530641e+00 6.32498920e-01
-7.80004263e-01 7.04334527e-02 -5.45834422e-01 -3.83151561e-01
-3.91743124e-01 6.97229087e-01 -1.61759114e+00 5.49035549e-01
4.99204248e-01 4.85344887e-01 -7.92200089e-01 1.07250404e+00
-1.35418341e-01 7.82649994e-01 -1.41917408e-01 -7.45351017e-01
1.06516987e-01 -3.65104228e-01 4.07924920e-01 1.33662713e+00
-1.68721795e-01 2.03064099e-01 5.68880975e-01 1.64652616e-01
-1.54557928e-01 8.31244707e-01 -5.94252050e-01 1.39076859e-01
3.46238732e-01 1.62059999e+00 -8.63485515e-01 -5.96563458e-01
-1.17388472e-01 2.71696001e-01 2.36666918e-01 2.88568258e-01
-1.41358644e-01 -6.30312741e-01 -2.32999906e-01 3.76161873e-01
-2.33153731e-01 4.63089794e-01 -5.58721721e-01 -4.67203617e-01
-4.37531322e-02 -9.29870486e-01 5.93071342e-01 -4.70512450e-01
-1.17986572e+00 9.09750283e-01 1.37153879e-01 -1.04184723e+00
-4.58485395e-01 -6.74846113e-01 -8.03456604e-01 1.00088370e+00
-8.13971758e-01 -7.13451862e-01 -2.71318287e-01 3.61011446e-01
7.04434574e-01 -8.15662742e-01 9.80921030e-01 -1.54343201e-02
-3.90558958e-01 5.81154287e-01 -4.50653911e-01 -3.46239060e-01
8.74703109e-01 -1.68621683e+00 1.48240820e-01 2.89383858e-01
2.60598749e-01 8.23772609e-01 7.34946012e-01 -7.53526628e-01
-8.54017675e-01 -8.61778080e-01 1.54931104e+00 -3.98372084e-01
6.24444067e-01 -9.55495611e-02 -6.66628718e-01 2.28239492e-01
4.35157955e-01 -3.58395636e-01 1.06873584e+00 4.38817918e-01
-3.47202048e-02 -1.86310515e-01 -1.11404777e+00 2.07857996e-01
4.26234305e-01 -2.88294137e-01 -6.19015753e-01 6.51406348e-01
5.30284107e-01 -1.21813565e-01 -7.20931470e-01 1.69927195e-01
6.53603673e-01 -8.54376912e-01 3.02944928e-01 -8.79660666e-01
6.63430929e-01 -1.43883139e-01 2.53126640e-02 -1.05027163e+00
-9.91853699e-02 -4.86353546e-01 1.36420175e-01 1.17393970e+00
1.35603726e+00 -4.26290989e-01 8.72466207e-01 8.22420657e-01
-1.08860783e-01 -1.11434710e+00 -3.27307433e-01 -1.99818686e-01
-2.76716858e-01 -3.24533522e-01 2.47351229e-01 8.14766407e-01
4.74507868e-01 9.05411601e-01 8.17531198e-02 -6.06575131e-01
3.80508363e-01 1.75167009e-01 1.01142454e+00 -1.58157742e+00
-2.10199475e-01 -3.83689433e-01 -1.62414089e-01 -4.18163419e-01
-1.51200294e-01 -9.75741982e-01 -1.29374400e-01 -1.52879620e+00
1.02474526e-01 -4.82104868e-01 -6.17264330e-01 5.00522912e-01
-6.40009791e-02 2.84365732e-02 -1.57051563e-01 9.58387181e-02
-8.63077819e-01 -7.54912496e-02 6.79039836e-01 -1.13309152e-01
-4.05471683e-01 6.88816011e-01 -8.48512113e-01 6.22368038e-01
1.01445770e+00 -5.08653402e-01 -6.93545699e-01 3.29797387e-01
3.41685981e-01 -8.47875774e-02 -1.65701658e-01 -6.47057712e-01
3.78821075e-01 7.98054412e-02 3.13758075e-01 -8.00803781e-01
-2.01570913e-02 -6.39533579e-01 -1.79704964e-01 4.97831285e-01
-6.58113182e-01 4.50823665e-01 -1.36374816e-01 4.26872224e-01
-2.51347989e-01 -4.89647239e-01 2.38515213e-01 -1.39951110e-01
-3.89783442e-01 -8.36457592e-04 -3.32362592e-01 -3.15578073e-01
9.16491807e-01 -2.66675800e-01 -3.14227939e-01 -3.26979190e-01
-8.56038094e-01 4.18059647e-01 6.03989214e-02 3.51436108e-01
6.17166102e-01 -8.29754233e-01 -8.52132261e-01 9.45641696e-02
3.82483333e-01 -4.01800931e-01 -2.79214799e-01 7.49078572e-01
-2.07033560e-01 3.22475642e-01 -9.94724706e-02 -3.69674236e-01
-1.63572955e+00 6.34910092e-02 9.07681882e-03 -5.34915805e-01
-7.93748200e-02 1.03756249e+00 -2.90253013e-01 -4.21165943e-01
3.92505199e-01 1.24783665e-01 -1.08037102e+00 4.60746020e-01
6.78349316e-01 5.79039395e-01 5.43224156e-01 -1.36717379e-01
-2.52533138e-01 5.24512157e-02 -6.56666934e-01 -5.93506992e-01
1.36780035e+00 1.85358837e-01 -3.18942845e-01 6.56188548e-01
8.02930176e-01 2.27061734e-01 -7.38032311e-02 -7.47011378e-02
4.60362822e-01 -8.57247412e-02 -2.49776274e-01 -1.12360859e+00
-5.17152369e-01 3.37799340e-01 5.08547008e-01 1.06176412e+00
6.03180468e-01 -2.41107151e-01 -5.23106977e-02 9.14072096e-01
3.93670261e-01 -1.40675020e+00 -4.18954641e-02 4.90138292e-01
1.05880213e+00 -1.43077135e+00 4.44104552e-01 -4.17010754e-01
-8.67218256e-01 1.31907547e+00 8.34093630e-01 -2.63881594e-01
9.91558552e-01 -1.72038674e-02 -2.72459456e-05 -6.90497816e-01
-1.23258483e+00 2.92315066e-01 7.08135784e-01 4.09932017e-01
9.57162738e-01 4.19811457e-01 -1.23895371e+00 1.11691451e+00
-5.45368493e-01 -2.89169759e-01 5.92840493e-01 7.60195732e-01
-7.90206075e-01 -1.30504334e+00 1.05550289e-01 8.71121764e-01
-3.24399352e-01 9.84445028e-03 -1.10220778e+00 9.18671548e-01
6.23784624e-02 1.46084619e+00 -2.36232206e-01 -8.85622382e-01
5.74009895e-01 -7.14573413e-02 -1.29341662e-01 -1.12335694e+00
-1.45508814e+00 -7.17336908e-02 8.83043945e-01 -3.61367196e-01
-2.67175376e-01 -6.07085049e-01 -1.18008566e+00 -4.34877835e-02
-5.84877849e-01 1.11615872e+00 9.69951510e-01 6.76573515e-01
-1.28880143e-02 2.02942997e-01 1.07977247e+00 -3.08059514e-01
-6.15771651e-01 -1.34769475e+00 -7.43322194e-01 2.64853626e-01
-2.01464072e-01 -3.43286604e-01 -4.97711301e-01 -7.03116283e-02] | [10.92527961730957, 7.449750900268555] |
4b0ed255-3171-480d-878e-6f247e8be46a | cross-document-non-fiction-narrative | null | null | https://aclanthology.org/W15-4509 | https://aclanthology.org/W15-4509.pdf | Cross-Document Non-Fiction Narrative Alignment | null | ['Shakthidhar Gopavaram', 'Ben Miller', 'Ayush Shrestha', 'Jennifer Olive'] | 2015-07-01 | null | null | null | ws-2015-7 | ['graph-similarity'] | ['graphs'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.531668186187744, 3.55029296875] |
8a9f93b1-472c-45e7-8ff9-982cd33fe40d | beamformed-fingerprint-learning-for-accurate | 1804.04112 | null | http://arxiv.org/abs/1804.04112v1 | http://arxiv.org/pdf/1804.04112v1.pdf | Beamformed Fingerprint Learning for Accurate Millimeter Wave Positioning | With millimeter wave wireless communications, the resulting radiation
reflects on most visible objects, creating rich multipath environments, namely
in urban scenarios. The radiation captured by a listening device is thus shaped
by the obstacles encountered, which carry latent information regarding their
relative positions. In this paper, a system to convert the received millimeter
wave radiation into the device's position is proposed, making use of the
aforementioned hidden information. Using deep learning techniques and a
pre-established codebook of beamforming patterns transmitted by a base station,
the simulations show that average estimation errors below 10 meters are
achievable in realistic outdoors scenarios that contain mostly
non-line-of-sight positions, paving the way for new positioning systems. | ['Gabriel Falcão', 'João Gante', 'Leonel Sousa'] | 2018-04-11 | null | null | null | null | ['outdoor-positioning'] | ['miscellaneous'] | [ 4.97700647e-02 2.47065783e-01 2.79438823e-01 -1.07693270e-01
-4.41581994e-01 -5.50526917e-01 3.68872046e-01 -1.21750861e-01
-3.09072405e-01 7.33851552e-01 4.21043336e-01 -3.38379800e-01
-3.37900519e-01 -1.25773859e+00 -3.30546230e-01 -1.21368408e+00
-4.71161485e-01 1.32571563e-01 -1.02297924e-01 -9.73903015e-02
5.11374138e-02 4.52253431e-01 -9.07619178e-01 8.71861540e-03
1.76976413e-01 1.12932527e+00 3.22669655e-01 9.06873107e-01
6.39179628e-03 6.09603405e-01 -1.15430689e+00 -3.91825318e-01
6.97440729e-02 8.64655003e-02 3.28565001e-01 -2.13652894e-01
-7.06548308e-05 -3.49910706e-01 -7.09689319e-01 6.35202944e-01
8.13064754e-01 -3.53968024e-01 6.07087076e-01 -8.58871758e-01
-7.54397064e-02 4.93509382e-01 8.66886824e-02 -1.21050879e-01
4.82004911e-01 -1.93511799e-01 4.34395611e-01 -5.90712428e-01
1.65857896e-01 7.04777598e-01 1.04450881e+00 -2.23336853e-02
-3.43981713e-01 -7.36693025e-01 -3.55127037e-01 -1.41556963e-01
-1.33578074e+00 -4.14171517e-01 9.45607185e-01 -3.01418036e-01
4.37548608e-01 5.11533201e-01 6.32802606e-01 1.63573706e+00
1.07698107e+00 1.67355686e-01 6.19873941e-01 -5.31664491e-01
5.92959642e-01 1.22789443e-02 -1.84687138e-01 3.05687428e-01
7.76555896e-01 4.85959768e-01 -3.40096772e-01 -4.00756560e-02
5.58368087e-01 1.04880869e-01 -4.46748167e-01 -5.77988863e-01
-1.05123103e+00 4.73325491e-01 9.45342124e-01 8.12729537e-01
-3.57285649e-01 5.48267841e-01 -8.24448049e-01 1.60036668e-01
6.82386756e-02 5.88718474e-01 -2.06729606e-01 9.72725525e-02
-7.33780563e-01 -3.16037565e-01 8.20582032e-01 1.30301189e+00
5.38592994e-01 6.12712651e-02 -6.66422248e-02 2.44070247e-01
1.15164185e+00 1.00043964e+00 7.44427443e-02 -5.90157747e-01
7.71224797e-01 -1.25335529e-01 4.44325119e-01 -1.54035079e+00
-1.14422178e+00 -1.91306996e+00 -1.35180485e+00 1.05434537e-01
1.82694450e-01 -8.91521752e-01 -9.42683876e-01 1.59271693e+00
-7.75670558e-02 5.75962551e-02 5.23632765e-01 3.42846036e-01
7.90236533e-01 6.14500046e-01 -4.74257529e-01 -2.23090842e-01
8.14781010e-01 -5.61388671e-01 -7.26086915e-01 -6.74041927e-01
4.48471487e-01 -6.42257690e-01 -7.46241165e-03 6.80976689e-01
-7.48370171e-01 -5.13824821e-01 -1.40053093e+00 6.26753986e-01
-2.05060154e-01 -7.26275072e-02 6.74789190e-01 1.24481952e+00
-1.02412891e+00 -2.28239730e-01 -5.81068158e-01 5.55821415e-03
5.55373013e-01 2.92810619e-01 2.30603501e-01 -2.49911815e-01
-9.06299949e-01 4.52552646e-01 -3.86083901e-01 6.95964932e-01
-3.05192769e-01 -5.84396243e-01 -5.11337936e-01 7.27532804e-02
-1.73558056e-01 -7.76687860e-01 1.10317504e+00 -2.00036123e-01
-1.41845834e+00 -8.57432410e-02 -1.43941656e-01 -1.64621070e-01
4.83958691e-01 -6.36785803e-03 -9.66292143e-01 -1.32746071e-01
-1.54411830e-02 -1.24263443e-01 4.28667337e-01 -1.93941021e+00
-5.04363239e-01 -3.18526298e-01 1.05624877e-01 -1.37489021e-01
-1.10971525e-01 -8.85750830e-01 -2.73638606e-01 -3.09408933e-01
8.28729033e-01 -8.71745229e-01 -8.14548314e-01 -1.85441390e-01
-6.87195480e-01 5.36351323e-01 2.13185906e-01 -5.67260496e-02
9.72034812e-01 -2.06915760e+00 -4.26011473e-01 8.17158818e-01
3.91685277e-01 -1.61704928e-01 1.64378837e-01 8.74724090e-01
2.95008659e-01 -4.54328954e-01 2.95407861e-01 -5.26475132e-01
-4.16132621e-02 -8.89877379e-02 -3.35841954e-01 5.46883881e-01
-6.02165043e-01 6.08158827e-01 -8.74315262e-01 3.17134351e-01
1.18538126e-01 6.42383814e-01 -6.83357239e-01 9.83966663e-02
3.59700114e-01 1.01297760e+00 -8.53632689e-01 3.45839292e-01
1.14645457e+00 -2.33527645e-02 1.08895533e-01 9.82342288e-02
-3.17395091e-01 5.33650592e-02 -1.06788194e+00 1.50573242e+00
-1.01828074e+00 7.23229766e-01 -2.63578780e-02 -4.73527938e-01
1.18123031e+00 2.30122611e-01 5.23761332e-01 -1.11799848e+00
3.19634974e-01 6.62081316e-02 -1.89640328e-01 -4.29524153e-01
1.42512411e-01 4.90926839e-02 -4.84168231e-01 1.66066438e-01
-1.44425288e-01 5.94303645e-02 -5.45980871e-01 -7.54816383e-02
1.63866460e+00 -5.98408699e-01 2.56941408e-01 2.13682696e-01
3.16521257e-01 -4.83904511e-01 2.92782098e-01 1.21805704e+00
4.51105028e-01 3.91062677e-01 -2.44620144e-01 -1.99516073e-01
-3.02782625e-01 -1.63627851e+00 -2.43498102e-01 3.25449139e-01
5.76878130e-01 -1.35091141e-01 -1.70580894e-01 2.28008572e-02
-2.34593794e-01 6.79880917e-01 -7.32112080e-02 -2.20117480e-01
-4.36248302e-01 -8.21239531e-01 4.15562123e-01 1.18713960e-01
6.53694510e-01 -5.89328945e-01 -7.20270097e-01 6.58879220e-01
-2.59963483e-01 -1.16737437e+00 5.02955019e-01 5.05031049e-01
-5.13446093e-01 -6.19410157e-01 -6.83316767e-01 -7.51059473e-01
7.99432456e-01 6.04749084e-01 1.04046273e+00 -4.49103527e-02
-7.76569918e-02 5.47649026e-01 -4.15208459e-01 -6.02464736e-01
-1.62428934e-02 -1.69964224e-01 -1.00733638e-01 -2.54674759e-02
-2.08939910e-02 -9.53690469e-01 -8.40774715e-01 2.98638672e-01
-7.80799091e-02 6.12044241e-03 1.26181567e+00 2.11011305e-01
-1.56587780e-01 6.52191699e-01 2.76617169e-01 -2.75693357e-01
-6.97511584e-02 -6.09297335e-01 -5.76065540e-01 -9.87931415e-02
-7.56498799e-02 -2.26545230e-01 4.59576994e-01 5.21315932e-02
-1.17093074e+00 3.03528346e-02 -4.79766726e-01 8.28377128e-01
-2.17803210e-01 4.51070815e-01 -8.01795602e-01 -5.32984436e-01
5.49039364e-01 2.75887311e-01 -1.04134691e+00 -2.21620411e-01
2.39217177e-01 7.95183480e-01 5.82128465e-01 3.17525625e-01
1.43052697e+00 6.55051827e-01 3.22978526e-01 -1.23446703e+00
-5.46792090e-01 -5.33337116e-01 -5.06454885e-01 -2.77880013e-01
4.24156755e-01 -1.00759113e+00 -6.26071632e-01 3.76532435e-01
-1.20812547e+00 -1.48335725e-01 3.45020354e-01 9.62676644e-01
-3.12613219e-01 -4.72750328e-02 1.58991327e-03 -9.71817911e-01
2.06255570e-01 -6.60732687e-01 8.21086049e-01 9.11684036e-02
-2.08690673e-01 -9.99493897e-01 4.25860256e-01 2.02249244e-01
9.30714369e-01 2.57652048e-02 7.32186079e-01 1.04606993e-01
-1.23716867e+00 -5.03795683e-01 2.10236147e-01 -4.19523001e-01
2.11279526e-01 -1.12253416e+00 -1.09458578e+00 -4.29935247e-01
2.03024745e-01 5.37400782e-01 4.79509443e-01 8.41013789e-01
7.77662992e-01 -3.21378618e-01 -8.34110498e-01 1.12763345e+00
1.62092650e+00 4.93278295e-01 1.03692496e+00 2.31617838e-01
4.45637107e-01 4.25817519e-02 7.95207247e-02 6.27716303e-01
2.64278620e-01 6.19397461e-01 8.89566958e-01 -1.98821470e-01
-2.06138864e-01 7.53367171e-02 -6.69003651e-02 6.30998492e-01
1.09995700e-01 -1.32013059e+00 -5.25458276e-01 -2.75526810e-02
-1.18453741e+00 -8.75412345e-01 -3.87920409e-01 2.01595664e+00
5.43980114e-02 3.37481558e-01 -8.18726242e-01 1.13029137e-01
8.75372253e-03 4.07433182e-01 -1.50200084e-01 1.90489292e-01
1.44331832e-03 2.34041467e-01 1.01788044e+00 9.57412481e-01
-9.80796695e-01 1.19081371e-01 6.11987734e+00 -4.21705171e-02
-1.14336169e+00 -1.90485269e-01 -1.14154428e-01 2.35495850e-01
-6.58086896e-01 -5.75793505e-01 -6.94854736e-01 4.36501145e-01
7.92251229e-01 2.47853458e-01 -1.90401554e-01 4.38402593e-01
2.78105736e-01 -2.73238629e-01 -7.12207615e-01 1.10707021e+00
5.84032163e-02 -1.15772653e+00 -4.49081928e-01 4.76118892e-01
6.45225406e-01 2.36262698e-02 6.87212825e-01 1.13420680e-01
3.18878025e-01 -7.27789640e-01 6.16995990e-01 8.60422432e-01
2.11519122e-01 -6.49316192e-01 8.64993751e-01 4.18629616e-01
-9.15721595e-01 -4.92736042e-01 -3.51607054e-01 -2.23805934e-01
1.80557311e-01 1.43749452e+00 -1.01726282e+00 7.32799828e-01
4.02092904e-01 2.07329541e-01 -4.29523855e-01 1.59000325e+00
-5.62074661e-01 6.81914866e-01 -2.66912431e-01 -3.45096141e-01
2.15583861e-01 1.24933600e-01 5.72830498e-01 1.18132555e+00
1.19671118e+00 -1.11354679e-01 -1.60423130e-01 4.18826193e-01
-7.95420632e-02 -3.61778677e-01 -6.88674927e-01 7.63061523e-01
4.65205312e-01 9.75421190e-01 -6.68261766e-01 2.43466720e-01
-3.43193859e-01 7.82584012e-01 -7.07504630e-01 8.40565860e-01
-6.03833079e-01 -4.91892874e-01 4.95212048e-01 1.81725636e-01
5.49834967e-01 -8.71031106e-01 -3.85238528e-01 -5.84679782e-01
1.18349351e-01 -4.82094288e-03 -4.98216480e-01 -7.94709921e-01
-9.22820330e-01 4.42467958e-01 -6.38901889e-01 -1.69295824e+00
-1.58956677e-01 -5.63971937e-01 -6.35889828e-01 5.29830396e-01
-1.56680524e+00 -9.38929677e-01 -7.23569453e-01 4.68025863e-01
-3.50102149e-02 -7.17120618e-02 1.07020056e+00 5.23153722e-01
1.63255587e-01 5.55489779e-01 8.99434745e-01 2.97096163e-01
3.22495282e-01 -9.29565966e-01 1.82399660e-01 7.76962161e-01
4.54298586e-01 5.93552768e-01 9.06168759e-01 -3.63714367e-01
-1.51637042e+00 -1.22052121e+00 8.30229282e-01 -4.27640080e-01
2.00614899e-01 -7.63253570e-01 1.54015884e-01 5.80956936e-01
1.49187982e-01 -1.24295525e-01 1.23373902e+00 2.38447666e-01
5.54293115e-03 -2.52999842e-01 -1.03855824e+00 6.24459505e-01
1.16216993e+00 2.03723758e-01 -3.88961881e-01 4.51684266e-01
1.94833398e-01 -3.53133321e-01 -7.02814609e-02 2.60086387e-01
7.33687818e-01 -9.66591895e-01 1.46797311e+00 3.96229655e-01
-9.10237506e-02 -5.40474197e-04 -5.55070102e-01 -1.57940304e+00
-8.56598794e-01 -5.35018504e-01 3.13371532e-02 7.95078635e-01
4.83003050e-01 -6.85334921e-01 1.18915844e+00 -4.70551878e-01
-2.41313413e-01 -3.08044553e-01 -1.23620152e+00 -6.53797626e-01
-2.69040436e-01 -9.11617994e-01 5.34068763e-01 2.34880552e-01
-3.73081177e-01 4.24538136e-01 -2.91789562e-01 1.45750356e+00
1.04065490e+00 -1.05822928e-01 8.83369565e-01 -1.53463125e+00
-5.03890276e-01 -5.47779314e-02 -8.24578464e-01 -1.83148170e+00
-4.22879338e-01 -5.77395380e-01 5.56932509e-01 -1.97907686e+00
-6.11504614e-01 -9.40460563e-01 -2.87382275e-01 -1.45170718e-01
6.72536790e-01 6.27283096e-01 -2.47945651e-01 -2.44860500e-01
-5.86761594e-01 5.80523491e-01 9.42660213e-01 -4.31025892e-01
-1.65606022e-01 1.23693013e+00 -8.67489934e-01 9.34659302e-01
7.66233027e-01 -5.02912939e-01 -4.21552807e-01 -7.09429741e-01
6.46444440e-01 7.70143569e-02 4.33596849e-01 -2.01966310e+00
5.70780873e-01 1.44207597e-01 8.58301699e-01 -9.33344185e-01
7.83230662e-01 -1.45174456e+00 3.44987452e-01 7.82576501e-01
5.28131798e-02 -7.31994987e-01 -2.36284554e-01 1.03306353e+00
2.79463232e-01 -8.65939539e-03 3.59876990e-01 2.52727389e-01
-4.27265823e-01 1.26115441e-01 -1.08255637e+00 -6.15182281e-01
9.26731288e-01 -2.70546854e-01 -1.18484758e-01 -1.06675255e+00
-5.18459141e-01 -1.57519221e-01 -2.71090150e-01 1.85698822e-01
5.10936201e-01 -1.25122643e+00 -2.92894185e-01 3.66731614e-01
1.63244143e-01 -2.48627871e-01 2.49437913e-01 3.26440990e-01
-2.51787931e-01 1.02682638e+00 -8.74125659e-02 -6.02019727e-01
-1.05014217e+00 3.73868681e-02 3.31577241e-01 -3.77036333e-02
-6.39065683e-01 1.00864339e+00 1.56015322e-01 -3.80504042e-01
4.37591612e-01 -6.33738339e-01 -3.23675722e-01 -2.07808048e-01
4.83970314e-01 -9.92175043e-02 2.97686398e-01 -4.91684407e-01
-4.66091663e-01 1.00187397e+00 5.86248577e-01 -3.21692437e-01
1.34470689e+00 -4.47379261e-01 4.53600407e-01 1.46838784e-01
1.18201566e+00 1.01255512e+00 -9.19833779e-01 -3.41739386e-01
-1.70691028e-01 -5.65626204e-01 1.64162874e-01 -1.04578626e+00
-7.84665227e-01 6.57930493e-01 9.26816106e-01 2.05584243e-01
8.93976867e-01 8.20751488e-02 6.31036520e-01 1.00390565e+00
1.45752871e+00 -2.92886704e-01 1.29831001e-01 4.83133674e-01
4.16548342e-01 -6.69910789e-01 -1.50223136e-01 -4.59012657e-01
3.67574334e-01 1.14975131e+00 -9.33184400e-02 1.18937910e-01
1.12515223e+00 7.62530029e-01 5.36274135e-01 -1.19198039e-01
-6.84037209e-02 1.11456722e-01 -2.31080651e-01 1.19961727e+00
3.99734192e-02 -3.65038291e-02 1.61824793e-01 7.56968081e-01
-8.96377563e-01 -3.90361041e-01 5.54373145e-01 1.03156781e+00
-9.39773977e-01 -1.02336383e+00 -7.00808764e-01 1.48947924e-01
5.46438619e-03 1.40767381e-01 9.01022628e-02 4.86159682e-01
4.17822629e-01 1.39834464e+00 6.15057610e-02 -4.07361746e-01
4.42923069e-01 -4.87577587e-01 5.54219007e-01 -4.39019054e-01
2.99997807e-01 -5.03423028e-02 -7.31205270e-02 -4.55365181e-01
-1.26038015e-01 -1.64052740e-01 -1.05952394e+00 2.69096762e-01
-1.13898374e-01 2.80313462e-01 8.67901862e-01 1.00771248e+00
1.97317079e-01 1.16070580e+00 1.17428875e+00 -9.08952236e-01
-1.26262918e-01 -6.13190532e-01 -5.84131122e-01 -6.23040020e-01
6.27900660e-01 -7.04986453e-01 -4.97256070e-01 -5.71854591e-01] | [6.302882671356201, 1.113948941230774] |
9cc9588f-510e-4f30-a259-64631dc3f63a | disambiguating-confusion-sets-as-an-aid-for | null | null | https://aclanthology.org/2020.readi-1.1 | https://aclanthology.org/2020.readi-1.1.pdf | Disambiguating Confusion Sets as an Aid for Dyslexic Spelling | Spell checkers and other proofreading software are crucial tools for people with dyslexia and other reading disabilities. Most spell checkers automatically detect spelling mistakes by looking up individual words and seeing if they exist in the vocabulary. However, one of the biggest challenges of automatic spelling correction is how to deal with real-word errors, i.e. spelling mistakes which lead to a real but unintended word, such as when then is written in place of than. These errors account for 20{\%} of all spelling mistakes made by people with dyslexia. As both words exist in the vocabulary, a simple dictionary lookup will not detect the mistake. The only way to disambiguate which word was actually intended is to look at the context in which the word appears. This problem is particularly apparent in languages with rich morphology where there is often minimal orthographic difference between grammatical items. In this paper, we present our novel confusion set corpus for Icelandic and discuss how it could be used for context-sensitive spelling correction. We have collected word pairs from seven different categories, chosen for their homophonous properties, along with sentence examples and frequency information from said pairs. We present a small-scale machine learning experiment using a decision tree binary classification which results range from 73{\%} to 86{\%} average accuracy with 10-fold cross validation. While not intended as a finalized result, the method shows potential and will be improved in future research. | ['Anton Karl Ingason', "Steinunn Rut Fri{\\dh}riksd{\\'o}ttir"] | 2020-05-01 | null | null | null | lrec-2020-5 | ['spelling-correction'] | ['natural-language-processing'] | [ 4.44182068e-01 -3.17659616e-01 2.53326952e-01 -1.21148974e-01
-6.20568514e-01 -7.06151664e-01 2.96703964e-01 7.91192830e-01
-9.35652137e-01 8.30190897e-01 3.65081459e-01 -7.62420237e-01
-1.49000436e-01 -6.49637818e-01 -5.02915680e-01 -4.87563640e-01
6.00783467e-01 5.84400058e-01 2.96015114e-01 -5.33201158e-01
6.68645561e-01 2.15061963e-01 -1.72099316e+00 4.07567471e-01
1.28072929e+00 2.32212320e-01 8.59876812e-01 6.49713457e-01
-3.03504318e-01 5.42487502e-01 -1.17236578e+00 -4.88111824e-01
4.21683257e-03 -4.17520255e-01 -9.17799950e-01 -1.07565828e-01
8.02378535e-01 -9.52092856e-02 -4.70389016e-02 1.43442357e+00
4.32987124e-01 6.61467314e-02 6.33910358e-01 -3.96133423e-01
-6.20579123e-01 5.09437263e-01 -8.44537094e-02 5.86881220e-01
7.42001891e-01 -3.29100788e-02 8.83787513e-01 -7.20943034e-01
2.76607901e-01 9.41878021e-01 5.83946407e-01 5.87017417e-01
-8.41233015e-01 -6.18914604e-01 -1.25949726e-01 5.67289114e-01
-1.43589723e+00 -2.88897097e-01 1.38411179e-01 -5.89517951e-01
1.52124059e+00 4.64145273e-01 9.40865397e-01 6.75208271e-01
1.89413905e-01 3.56748730e-01 1.34982038e+00 -1.46259975e+00
1.18825294e-03 2.21355539e-02 5.64751923e-01 5.10117233e-01
7.15977490e-01 -2.93114185e-01 -4.78595316e-01 3.13795328e-01
1.98043019e-01 -2.30033696e-01 -4.98305351e-01 4.46140498e-01
-9.35228646e-01 4.71870542e-01 -2.95251310e-01 7.13137686e-01
-5.92362368e-03 -2.56694049e-01 3.22673351e-01 5.32944858e-01
-7.88223073e-02 6.33380949e-01 -4.92382437e-01 -6.11709476e-01
-1.06384587e+00 3.70170385e-01 4.92295206e-01 9.24371064e-01
2.61615902e-01 -2.49613568e-01 -1.26239255e-01 1.24197137e+00
9.42358300e-02 6.88478649e-01 1.07315624e+00 -1.83276936e-01
6.16396189e-01 7.80732036e-01 1.87302411e-01 -9.36979115e-01
-3.44318956e-01 -2.48229340e-01 -2.24965528e-01 2.44413763e-01
1.13154221e+00 2.94190377e-01 -9.23763335e-01 1.51332784e+00
-2.75569946e-01 -5.10664344e-01 -1.83980111e-02 5.81304073e-01
7.23598361e-01 3.70102376e-01 6.23306334e-02 -4.16954815e-01
1.74617505e+00 -5.72500825e-01 -7.37943351e-01 -5.04681349e-01
1.01705575e+00 -9.72529471e-01 1.44860268e+00 7.27404714e-01
-1.08962274e+00 -1.79761559e-01 -1.05258489e+00 -7.72166550e-02
-4.29124057e-01 2.28044197e-01 9.47545469e-02 1.00116181e+00
-7.48542905e-01 5.24316549e-01 -4.28722650e-01 -5.26201665e-01
2.79973930e-04 1.01629853e-01 -4.03327227e-01 -9.15148333e-02
-1.28778934e+00 1.29647577e+00 4.48872268e-01 -3.71245056e-01
8.75063092e-02 -4.91253227e-01 -7.58037984e-01 -9.74219590e-02
2.26480030e-02 -1.47624034e-02 1.30054998e+00 -1.01085186e+00
-8.95411849e-01 1.37757170e+00 -5.47966301e-01 -3.52508843e-01
3.62721741e-01 -2.75910765e-01 -9.16041553e-01 -8.34973454e-02
3.59558582e-01 -1.95218727e-01 6.44315124e-01 -4.31381136e-01
-9.49788690e-01 -5.35359621e-01 -3.73117536e-01 2.12584913e-01
-2.55332381e-01 5.74101508e-01 -1.20493788e-02 -1.00464571e+00
3.43075752e-01 -7.81350493e-01 4.48596537e-01 -3.94995391e-01
-8.48503113e-02 -4.88286406e-01 7.28207678e-02 -1.21320057e+00
1.67451310e+00 -1.93085515e+00 -3.78200471e-01 1.20612033e-01
-3.31135951e-02 9.23253477e-01 2.68204689e-01 5.31065106e-01
-2.93487400e-01 5.60823791e-02 -2.13470802e-01 6.88676387e-02
-1.15457244e-01 1.16095811e-01 -2.64908433e-01 2.96729773e-01
-1.76954806e-01 5.42936862e-01 -1.13810503e+00 -8.84065777e-02
1.38586178e-01 -3.54436599e-02 -2.88627267e-01 -1.94877654e-01
2.19715368e-02 -5.14311902e-02 5.55713832e-01 3.14491481e-01
4.40900296e-01 2.42492780e-01 1.44369513e-01 3.64613593e-01
-2.92246222e-01 1.05876994e+00 -1.40937483e+00 1.05323720e+00
-5.00231862e-01 8.95921052e-01 -5.36851466e-01 -7.28695095e-01
7.32917488e-01 9.91933495e-02 -5.09683371e-01 -1.00182962e+00
-3.33555490e-02 7.62289226e-01 4.83758211e-01 -7.63111889e-01
6.24840379e-01 -3.31814796e-01 9.97173935e-02 2.54685521e-01
-2.83774018e-01 -5.18278889e-02 5.96152723e-01 5.32558002e-02
1.19168925e+00 -3.89634311e-01 8.70511532e-01 -3.75709563e-01
7.45781958e-01 9.76245999e-02 5.10664761e-01 9.30279195e-01
-1.11337647e-01 8.35716546e-01 2.95212626e-01 -2.25670278e-01
-9.21166897e-01 -8.22952986e-01 -3.18587512e-01 8.05031121e-01
-1.84050158e-01 -7.98952699e-01 -7.28644848e-01 -2.87760735e-01
-2.50225395e-01 1.34870493e+00 -1.69901118e-01 -1.47783279e-01
-5.65666854e-01 -2.41516903e-01 5.37052572e-01 4.52638924e-01
3.08556825e-01 -1.07867944e+00 -5.86529136e-01 2.88932174e-01
-1.70116663e-01 -8.24376166e-01 -4.59983110e-01 9.87028629e-02
-4.41460609e-01 -1.15026557e+00 -6.07100546e-01 -1.14021313e+00
5.44929504e-01 4.02845472e-01 8.69868457e-01 5.80570698e-01
-1.70082673e-01 7.55853578e-02 -7.69057751e-01 -7.69681573e-01
-8.58105421e-01 -3.45502466e-01 3.79593790e-01 -6.10033214e-01
1.06075263e+00 -3.22848499e-01 -2.25033328e-01 -3.61687653e-02
-6.05034232e-01 -4.09059338e-02 2.99341768e-01 7.83341408e-01
4.01576370e-01 3.07411700e-01 7.28614777e-02 -8.34992647e-01
7.21378267e-01 -5.48769049e-02 -4.92157727e-01 2.92743206e-01
-6.84981883e-01 -6.62809191e-03 7.27660239e-01 -3.32622319e-01
-6.05955303e-01 -2.55493402e-01 -5.15812814e-01 2.78188229e-01
-5.60228229e-01 2.49240354e-01 -2.62615949e-01 1.30875394e-01
9.10599589e-01 7.15121925e-01 1.03270477e-02 -5.67800045e-01
-1.58358723e-01 1.12010992e+00 6.42301023e-01 -1.90482676e-01
4.56438571e-01 -1.20558895e-01 -4.99655783e-01 -1.19633245e+00
-6.22326612e-01 -7.90380597e-01 -3.82476240e-01 -4.78479117e-02
5.60917199e-01 -5.84799886e-01 -5.27272701e-01 9.99371290e-01
-1.20529437e+00 -1.20632365e-01 -2.02559963e-01 5.23866057e-01
-3.66109341e-01 6.89133763e-01 -1.27816051e-01 -6.81469619e-01
-2.53399730e-01 -1.07494748e+00 5.45754015e-01 2.94185221e-01
-8.33327293e-01 -7.68718362e-01 8.30237120e-02 4.63941038e-01
1.11489594e-01 -5.51837742e-01 1.36581004e+00 -1.00480413e+00
2.79418677e-01 -3.07371765e-01 -8.11207946e-03 6.52906597e-01
3.06592613e-01 -2.43257731e-01 -5.05528390e-01 -2.73506880e-01
-2.69988049e-02 1.61634728e-01 5.82359552e-01 8.13760236e-03
7.66056061e-01 -3.66953433e-01 -3.36897634e-02 1.51911139e-01
1.25872409e+00 4.53382790e-01 9.10979807e-01 5.44185698e-01
6.27577722e-01 3.95715922e-01 4.13988262e-01 2.13891283e-01
3.26484233e-01 7.49784946e-01 -9.85121951e-02 6.31336987e-01
-6.30904853e-01 -3.44534785e-01 4.54459816e-01 8.42810333e-01
6.20120056e-02 -4.05456185e-01 -1.32684064e+00 8.14653575e-01
-1.19729936e+00 -1.07677424e+00 -7.76177883e-01 2.67215872e+00
1.02605641e+00 1.91921175e-01 7.16219917e-02 1.00575411e+00
9.01739895e-01 -3.06114405e-01 1.17441893e-01 -6.68217003e-01
-3.12086880e-01 6.51227832e-01 4.37168121e-01 7.04083800e-01
-6.40425384e-01 1.28693807e+00 5.70373154e+00 1.03307652e+00
-1.10842788e+00 2.46549379e-02 -1.41564295e-01 -7.76751339e-02
-2.88727194e-01 -3.85244414e-02 -1.03847516e+00 8.94667327e-01
8.16103458e-01 6.17036037e-02 4.56908882e-01 2.85764903e-01
2.88332403e-01 -6.51050627e-01 -8.12051535e-01 1.28691018e+00
3.81753474e-01 -1.02829874e+00 -2.53313407e-02 -3.48698080e-01
4.01530474e-01 -1.68866843e-01 -4.55605894e-01 4.05453667e-02
3.74572724e-02 -1.03362453e+00 1.17690253e+00 3.71719271e-01
9.13829148e-01 -5.86528659e-01 4.72107232e-01 4.26887691e-01
-5.72981775e-01 -6.27224846e-03 -3.89430851e-01 -8.52446437e-01
-1.70605868e-01 5.84094286e-01 -7.47819245e-01 -1.23759694e-01
6.43721700e-01 4.09151733e-01 -7.82311320e-01 1.50609827e+00
-9.34437454e-01 9.45338488e-01 -2.78485566e-01 -6.17757440e-01
-2.01771036e-01 1.01864701e-02 8.91410410e-01 1.35905468e+00
5.30417979e-01 2.31217504e-01 -6.55248940e-01 3.31603318e-01
4.35333341e-01 6.56586528e-01 -3.75362575e-01 -7.88759589e-02
7.37055004e-01 6.63614750e-01 -7.21088767e-01 -1.76969379e-01
-5.22590637e-01 1.25735056e+00 5.28140128e-01 -3.72348800e-02
-2.13741049e-01 -7.51939416e-01 6.50869191e-01 3.32579821e-01
1.64332211e-01 -2.09354654e-01 -6.01339757e-01 -1.02091324e+00
3.90117586e-01 -1.10342908e+00 3.85208368e-01 -7.21704066e-01
-1.09133995e+00 1.76978573e-01 -2.45920360e-01 -1.12731659e+00
-8.53997748e-03 -1.09224474e+00 -5.82725704e-01 1.22548449e+00
-1.00185645e+00 -4.27390724e-01 -6.68706298e-02 2.80334681e-01
7.01699793e-01 -3.64268363e-01 8.84782314e-01 1.99472845e-01
-3.31060410e-01 9.24445927e-01 1.93985745e-01 1.18083589e-01
7.28461206e-01 -1.42471385e+00 3.50456059e-01 1.28848517e+00
4.73507017e-01 7.94701159e-01 1.00649893e+00 -8.34624946e-01
-6.29608870e-01 -7.52933025e-01 1.89242899e+00 -5.03538668e-01
6.15260303e-01 -2.01262712e-01 -1.03599024e+00 2.97477663e-01
4.29367647e-02 -5.02702415e-01 6.82376981e-01 -1.14480682e-01
-2.20219165e-01 1.10595569e-01 -1.09073710e+00 8.19986820e-01
1.13460851e+00 -5.00731945e-01 -1.17619669e+00 6.65633440e-01
3.25013280e-01 -3.80916685e-01 -2.03560486e-01 -1.61134899e-01
3.88155520e-01 -9.07762945e-01 4.01399016e-01 -6.91805840e-01
3.97649109e-01 -3.67611200e-01 -1.00601263e-01 -1.65576255e+00
-4.07644451e-01 -5.93510449e-01 4.24681842e-01 1.17075622e+00
4.71879363e-01 -6.43188834e-01 3.72768134e-01 6.93409324e-01
-3.01515728e-01 -2.33044103e-01 -9.24906135e-01 -1.01590669e+00
5.01468003e-01 -9.26113129e-01 4.32390809e-01 6.77686274e-01
5.92217028e-01 7.29965838e-03 5.94951287e-02 5.46646565e-02
1.00983098e-01 -4.49871689e-01 2.19146088e-01 -1.02677560e+00
-2.95292139e-01 -6.70045376e-01 -7.17837393e-01 -7.33882904e-01
1.12502752e-02 -1.08838046e+00 1.10018075e-01 -1.49809706e+00
-5.20465262e-02 -3.77572000e-01 2.30233803e-01 5.68438649e-01
-3.58412951e-01 1.49145320e-01 2.62140691e-01 8.21907073e-02
-1.04328558e-01 6.58732047e-03 8.15317929e-01 -1.72184542e-01
-1.53374508e-01 1.96182251e-01 -7.65411139e-01 8.84783804e-01
9.57638681e-01 -6.20693028e-01 -1.39588788e-01 -6.89177334e-01
6.67641222e-01 -2.80137628e-01 2.04920411e-01 -1.18779659e+00
2.49331266e-01 -2.11183727e-01 3.09671704e-02 -2.51774132e-01
-2.52590805e-01 -6.18566751e-01 -1.73469469e-01 6.35451853e-01
-1.63836002e-01 4.18196946e-01 2.70593911e-01 7.33377561e-02
1.87715098e-01 -1.15796089e+00 8.76836836e-01 -2.50502497e-01
-8.78366768e-01 -5.13984323e-01 -9.96606767e-01 4.36353087e-01
9.44289625e-01 -4.47786003e-01 -2.40077317e-01 -1.80623442e-01
-5.11961102e-01 -3.36900890e-01 6.27254009e-01 4.44911271e-01
5.22650719e-01 -9.22159553e-01 -7.03463554e-01 4.85378683e-01
4.03063834e-01 -4.31667477e-01 -8.71142671e-02 5.87602615e-01
-1.04685140e+00 5.46413064e-01 -1.41314268e-01 -6.87981863e-03
-1.70427465e+00 3.36363554e-01 4.02725130e-01 2.28335023e-01
-6.88862979e-01 1.10958660e+00 -5.64971089e-01 -6.53608292e-02
2.76901722e-01 -4.29521471e-01 -4.49991643e-01 9.67608169e-02
1.20800030e+00 4.94966656e-01 7.68881917e-01 -8.63544405e-01
-5.95113337e-01 4.84236628e-01 -5.15875638e-01 -1.64323047e-01
8.12690973e-01 -3.20951715e-02 -3.82394373e-01 5.74285865e-01
6.35589540e-01 6.10038221e-01 -3.26376557e-01 -1.40173003e-01
-9.91489664e-02 -7.14379609e-01 -5.05896434e-02 -1.15881884e+00
-4.26684409e-01 7.91203439e-01 5.95478654e-01 1.30760491e-01
8.99777830e-01 -1.49180770e-01 7.86939919e-01 4.85521913e-01
4.76159871e-01 -1.45676041e+00 -5.93995094e-01 9.57669079e-01
7.69529104e-01 -9.95611370e-01 -1.78555265e-01 -3.23787928e-01
-2.81203091e-01 1.16678774e+00 4.89374399e-01 -4.66852747e-02
1.41596392e-01 -1.67094227e-02 1.54729551e-02 6.25263825e-02
-1.69693843e-01 -4.70981598e-01 1.87177896e-01 7.15485036e-01
6.24416769e-01 2.54963756e-01 -1.40160763e+00 7.09438682e-01
-9.12731528e-01 -3.17352146e-01 7.74721444e-01 7.96511710e-01
-6.56625509e-01 -1.34908283e+00 -6.18275642e-01 8.57439518e-01
-4.76640403e-01 -6.78353548e-01 -3.76554340e-01 4.95315343e-01
3.31472576e-01 1.12871635e+00 2.06866890e-01 -2.57867724e-01
4.04565513e-01 3.20573181e-01 8.48958492e-01 -9.00000691e-01
-6.58142745e-01 -6.42889798e-01 1.19840235e-01 -2.52725091e-02
3.35575163e-01 -1.22126842e+00 -1.49120843e+00 -1.92932531e-01
-3.16923231e-01 -3.31959911e-02 4.77014095e-01 1.49589479e+00
-1.88298121e-01 1.57776266e-01 -2.23853111e-01 2.46746801e-02
-4.79429007e-01 -1.04482973e+00 -5.97694159e-01 6.64322436e-01
1.39316306e-01 -5.43791771e-01 -3.79437625e-01 -8.85853693e-02] | [10.970078468322754, 10.627943992614746] |
6474fa98-05d4-43c8-8a76-02f44c38610c | innovation-pursuit-a-new-approach-to-subspace | 1512.00907 | null | http://arxiv.org/abs/1512.00907v5 | http://arxiv.org/pdf/1512.00907v5.pdf | Innovation Pursuit: A New Approach to Subspace Clustering | In subspace clustering, a group of data points belonging to a union of
subspaces are assigned membership to their respective subspaces. This paper
presents a new approach dubbed Innovation Pursuit (iPursuit) to the problem of
subspace clustering using a new geometrical idea whereby subspaces are
identified based on their relative novelties. We present two frameworks in
which the idea of innovation pursuit is used to distinguish the subspaces.
Underlying the first framework is an iterative method that finds the subspaces
consecutively by solving a series of simple linear optimization problems, each
searching for a direction of innovation in the span of the data potentially
orthogonal to all subspaces except for the one to be identified in one step of
the algorithm. A detailed mathematical analysis is provided establishing
sufficient conditions for iPursuit to correctly cluster the data. The proposed
approach can provably yield exact clustering even when the subspaces have
significant intersections. It is shown that the complexity of the iterative
approach scales only linearly in the number of data points and subspaces, and
quadratically in the dimension of the subspaces. The second framework
integrates iPursuit with spectral clustering to yield a new variant of
spectral-clustering-based algorithms. The numerical simulations with both real
and synthetic data demonstrate that iPursuit can often outperform the
state-of-the-art subspace clustering algorithms, more so for subspaces with
significant intersections, and that it significantly improves the
state-of-the-art result for subspace-segmentation-based face clustering. | ['Mostafa Rahmani', 'George Atia'] | 2015-12-02 | null | null | null | null | ['face-clustering'] | ['computer-vision'] | [ 5.53188752e-03 -1.00080207e-01 -2.15147976e-02 1.96794271e-01
-6.45579815e-01 -7.90819824e-01 3.78485680e-01 -1.42918602e-01
-6.87029660e-02 3.45761180e-01 4.04209793e-02 -1.94802850e-01
-6.27408326e-01 -3.12389225e-01 -3.60022664e-01 -1.24026418e+00
-1.81206107e-01 5.60044408e-01 8.00579414e-02 9.63840708e-02
2.18592256e-01 8.32867026e-01 -1.46485078e+00 -1.25297606e-01
9.44139659e-01 7.68779039e-01 -1.92763731e-01 4.45467144e-01
-2.76509047e-01 -1.66781664e-01 -1.37134656e-01 6.27801120e-02
6.50364637e-01 -5.44309914e-01 -6.43463254e-01 6.78916037e-01
2.60201007e-01 2.82749414e-01 -1.99443489e-01 1.18054581e+00
2.59029031e-01 2.59698540e-01 9.48747933e-01 -1.38872278e+00
-2.57400066e-01 5.39808124e-02 -8.42877507e-01 -1.44851893e-01
3.83603185e-01 -2.55424619e-01 6.48004055e-01 -1.14899099e+00
6.12232804e-01 1.14536667e+00 7.53828645e-01 3.75473410e-01
-1.66054177e+00 -4.24861193e-01 -3.62199247e-02 1.15065359e-01
-1.80172551e+00 -4.45496619e-01 1.19527423e+00 -6.07563496e-01
3.09529662e-01 5.67324877e-01 6.41727269e-01 3.77170116e-01
-3.35601062e-01 5.26468933e-01 1.04638040e+00 -5.75937927e-01
5.59064925e-01 2.34046161e-01 4.67335463e-01 4.85261440e-01
1.36121482e-01 -7.49502284e-03 -1.79628298e-01 -5.45943558e-01
7.95516253e-01 4.86588478e-02 -2.81525224e-01 -1.13756216e+00
-1.06179154e+00 8.41590047e-01 1.85051769e-01 6.00293636e-01
-3.92168581e-01 -2.97784507e-01 5.68754645e-03 3.26394709e-03
2.49722198e-01 3.01085472e-01 -2.04297416e-02 2.86897600e-01
-1.34162545e+00 1.36501104e-01 8.50359738e-01 8.62702310e-01
7.06704676e-01 -1.61453318e-02 2.54044414e-01 6.47049665e-01
4.01223212e-01 3.85919839e-01 2.34731793e-01 -1.13402998e+00
7.97589719e-02 7.01466441e-01 5.44998293e-05 -1.13845408e+00
-2.64788359e-01 -5.03591478e-01 -8.26984525e-01 1.56570688e-01
7.27443337e-01 -4.07383293e-02 -6.12496912e-01 1.51476681e+00
5.57689130e-01 4.57555145e-01 1.12785205e-01 9.11623120e-01
3.45506102e-01 6.48557127e-01 -3.39363694e-01 -9.67233717e-01
1.01761436e+00 -5.53806663e-01 -5.29584765e-01 4.67704862e-01
2.13836670e-01 -8.02635372e-01 5.41041017e-01 5.76828003e-01
-9.81457889e-01 -5.49516380e-01 -8.16688359e-01 6.61656201e-01
-1.24308519e-01 1.49960980e-01 3.40687484e-01 8.30187500e-01
-1.22038281e+00 6.31644845e-01 -5.58409035e-01 -6.62968278e-01
1.85085520e-01 5.22302926e-01 -3.18400234e-01 2.27986649e-01
-5.35033584e-01 2.55628198e-01 4.28250194e-01 9.98290032e-02
-4.06470984e-01 -4.31352735e-01 -4.70951200e-01 -1.05667613e-01
2.90055215e-01 -4.84287441e-01 5.71839333e-01 -1.14606106e+00
-1.09453094e+00 7.44670033e-01 -5.65528333e-01 -1.66092277e-01
4.82077926e-01 1.03504680e-01 -5.58790743e-01 4.48064625e-01
2.51023620e-01 4.26985264e-01 1.23547113e+00 -1.71229780e+00
-3.83412004e-01 -6.51315391e-01 -6.66575491e-01 1.99239656e-01
-5.52215755e-01 -1.80189088e-02 -8.20956051e-01 -4.96906877e-01
8.11579049e-01 -1.07877707e+00 -5.56975603e-01 -1.80183604e-01
-5.48313975e-01 -4.06578422e-01 1.24280107e+00 -4.49906677e-01
1.38955534e+00 -2.46025538e+00 6.18673742e-01 7.91978002e-01
2.90138274e-01 3.67336050e-02 1.15151279e-01 5.13765693e-01
-5.46471477e-01 -9.36161429e-02 -5.47053456e-01 -1.69271126e-01
-2.44229674e-01 3.60498615e-02 -3.67843270e-01 9.26246762e-01
-2.65553713e-01 3.56257647e-01 -8.70384395e-01 -6.35170281e-01
2.35892609e-01 3.45676422e-01 -2.06679896e-01 -6.34703040e-02
2.21075371e-01 6.35749876e-01 -3.48420084e-01 5.23465335e-01
9.87427473e-01 -5.45653403e-02 2.35734731e-01 -2.90631980e-01
-2.64522135e-01 -6.63271248e-01 -1.90506542e+00 1.43528843e+00
3.62081289e-01 4.02165264e-01 3.54300648e-01 -1.16132438e+00
8.66629004e-01 5.86985469e-01 1.20794523e+00 2.46539712e-01
-2.92032547e-02 4.83348608e-01 -3.28040034e-01 -2.20327184e-01
7.57372230e-02 -3.05139512e-01 6.37930408e-02 3.17939639e-01
8.86152610e-02 1.52635291e-01 3.74026954e-01 1.85351864e-01
6.55417025e-01 -1.86755627e-01 2.40593567e-01 -6.63963020e-01
9.83869255e-01 1.10034451e-01 6.85811341e-01 3.52357894e-01
-4.43410784e-01 4.56589699e-01 3.06318372e-01 -1.35498792e-01
-1.02412438e+00 -1.50062490e+00 -4.41726327e-01 4.34613973e-01
3.50555986e-01 -2.70317793e-01 -1.14992321e+00 -5.41492343e-01
7.87931457e-02 5.10510743e-01 -3.83056581e-01 2.26476580e-01
-3.37645173e-01 -6.80163443e-01 2.44682923e-01 1.84368014e-01
3.32874835e-01 -6.33991420e-01 -2.19185546e-01 1.10872187e-01
6.91173673e-02 -7.92541385e-01 -4.94490355e-01 -9.63672847e-02
-1.26755989e+00 -1.24096072e+00 -9.53178406e-01 -9.34872091e-01
1.20409942e+00 4.98658240e-01 5.51126063e-01 -3.06980401e-01
-5.29551357e-02 7.47412622e-01 -1.43040806e-01 6.94964752e-02
-3.24006736e-01 -3.72881085e-01 4.94470835e-01 5.55307746e-01
5.84784746e-01 -8.15133095e-01 -4.06051069e-01 5.34559071e-01
-8.32167208e-01 -3.07631105e-01 2.93112099e-01 6.15605056e-01
7.81852722e-01 6.80284679e-01 2.68597633e-01 -3.91813099e-01
5.67811012e-01 -6.03030086e-01 -5.78050792e-01 1.36304975e-01
-5.93323469e-01 6.94861077e-03 7.55931914e-01 -3.52398336e-01
-7.01189339e-01 6.49715066e-01 4.71405089e-01 -9.21278596e-01
-4.51578856e-01 1.64235234e-01 -2.43890271e-01 -2.11692795e-01
5.73310792e-01 5.06908536e-01 1.90608963e-01 -7.43213892e-01
5.80638170e-01 5.78151464e-01 7.95398355e-01 -4.76065129e-01
1.21580827e+00 8.26862872e-01 3.19240510e-01 -1.31085098e+00
-1.29515812e-01 -1.21426368e+00 -1.27074087e+00 -4.45379972e-01
8.02535474e-01 -5.57459354e-01 -6.11687362e-01 2.44095653e-01
-9.45930183e-01 5.42064786e-01 -4.26429957e-01 4.39208329e-01
-6.86063647e-01 8.36676300e-01 -2.30118811e-01 -1.04938769e+00
-6.27006441e-02 -9.58881676e-01 7.84408391e-01 1.44799188e-01
-2.41443381e-01 -9.67505932e-01 2.06823915e-01 2.05678567e-01
-2.15011522e-01 4.01831657e-01 7.86385000e-01 -6.82119370e-01
-4.01581258e-01 -2.78147429e-01 1.06185719e-01 1.26223087e-01
1.68727279e-01 1.31659761e-01 -6.77874327e-01 -6.41758502e-01
3.79206568e-01 4.53065991e-01 6.40047729e-01 5.97263157e-01
7.70313442e-01 -2.45910406e-01 -7.84677327e-01 6.21433675e-01
1.61982143e+00 4.01440889e-01 3.39024603e-01 3.32804732e-02
8.13321412e-01 8.54027271e-01 3.36304933e-01 2.49841303e-01
-3.13959032e-01 7.91908383e-01 1.09388856e-02 -1.68253168e-01
1.64224759e-01 4.06997278e-02 2.85512418e-01 8.21510017e-01
-9.41312462e-02 3.26078892e-01 -8.46052349e-01 7.25219488e-01
-2.07373428e+00 -1.12222171e+00 -4.45446730e-01 2.50308800e+00
3.17323118e-01 -3.14376235e-01 6.97389960e-01 5.69338977e-01
1.00784206e+00 -1.57959089e-01 -5.47523499e-01 -2.64779240e-01
-1.56149030e-01 -1.47174269e-01 4.07295078e-01 3.55925053e-01
-1.33396363e+00 7.80328333e-01 6.63394070e+00 9.82387424e-01
-7.76065707e-01 -8.12121555e-02 3.52921546e-01 7.08638802e-02
-1.56343207e-01 1.72041446e-01 -5.53510904e-01 3.48295689e-01
6.55354679e-01 -3.33235204e-01 5.53614736e-01 7.39595175e-01
3.63884330e-01 -1.96919832e-02 -9.92073834e-01 1.11277461e+00
1.03465356e-01 -1.05744660e+00 -5.35282306e-02 3.08812201e-01
9.96488035e-01 -4.53606278e-01 1.85141534e-01 -3.70941222e-01
-1.11206025e-01 -7.67538548e-01 4.10937786e-01 4.14028198e-01
6.20052934e-01 -9.59324121e-01 2.80660123e-01 4.14844722e-01
-1.29467821e+00 -1.98994443e-01 -2.07336098e-01 2.25040749e-01
9.55522507e-02 6.28192008e-01 -8.33038986e-01 6.35365367e-01
5.91845810e-01 6.75883949e-01 -4.47870761e-01 1.25738657e+00
1.93053037e-01 5.46446681e-01 -5.26958227e-01 3.02308947e-01
2.87071794e-01 -9.84897316e-01 1.21645832e+00 1.00056207e+00
4.92664307e-01 2.23506376e-01 3.22055936e-01 9.74432886e-01
3.81143272e-01 3.13106298e-01 -7.59115279e-01 2.51297653e-02
5.48204124e-01 1.16885877e+00 -1.14905989e+00 -2.63780892e-01
-2.68881291e-01 1.06821275e+00 -2.24586725e-01 7.69195080e-01
-3.27473909e-01 -1.06290273e-01 4.66922253e-01 1.99925810e-01
4.54186261e-01 -4.74099398e-01 -5.88873565e-01 -7.75273681e-01
2.38941703e-02 -6.65660262e-01 4.45171177e-01 -2.97514528e-01
-1.26804423e+00 5.53125799e-01 1.65537745e-01 -1.59844434e+00
-3.17853451e-01 -4.64749753e-01 -6.85992420e-01 8.77350628e-01
-6.38594806e-01 -9.20842707e-01 5.92318885e-02 9.69551563e-01
3.61102372e-01 -4.45497453e-01 8.27496886e-01 6.52225409e-03
-6.66180670e-01 2.63809085e-01 8.51434290e-01 -4.43379879e-02
3.31183076e-01 -1.30350995e+00 -2.17553318e-01 1.08713508e+00
3.14356357e-01 8.30291092e-01 9.68219817e-01 -5.83549023e-01
-1.36327529e+00 -8.10588479e-01 5.71685493e-01 -2.51514763e-01
5.80918550e-01 -1.14102364e-01 -8.73165429e-01 2.86768675e-01
3.85528319e-02 -2.94995576e-01 9.60669100e-01 9.07349288e-02
-1.57609656e-01 -9.76930484e-02 -1.19572592e+00 6.46378815e-01
8.00152957e-01 -4.51690346e-01 -6.58765733e-01 4.02639478e-01
2.89074630e-01 3.13153565e-01 -7.87367642e-01 2.99388647e-01
3.22890490e-01 -1.10061359e+00 1.10495734e+00 -3.54902267e-01
-1.87021434e-01 -8.27434242e-01 -1.49308115e-01 -1.04229558e+00
-5.46165884e-01 -9.28053796e-01 -1.84098214e-01 1.11785746e+00
4.22200970e-02 -5.76976895e-01 1.14868426e+00 3.01057130e-01
1.47776157e-01 -6.30207121e-01 -1.24132073e+00 -1.13459384e+00
4.08423953e-02 -2.55518377e-01 2.92326719e-01 1.13864362e+00
3.09529275e-01 2.64320612e-01 3.32578011e-02 2.54438758e-01
1.23267245e+00 3.45768809e-01 6.31138742e-01 -1.57545161e+00
-1.69202313e-01 -6.26566350e-01 -5.83458662e-01 -7.98978090e-01
1.84064120e-01 -8.09764326e-01 -1.80213079e-01 -1.14918029e+00
2.28354082e-01 -4.00089681e-01 -2.74850011e-01 -4.39122841e-02
-4.17789184e-02 6.01875074e-02 2.73638695e-01 8.23199987e-01
-4.34960395e-01 3.88770074e-01 8.07547450e-01 5.59772141e-02
-7.46682346e-01 1.52574196e-01 -5.03222823e-01 8.90982270e-01
4.64811385e-01 -1.49066404e-01 -4.17261928e-01 3.30061108e-01
-5.26009619e-01 1.44547015e-01 3.19394201e-01 -1.33257139e+00
4.07364160e-01 5.55384681e-02 4.14262593e-01 -8.96111786e-01
3.29930931e-01 -1.26391029e+00 5.73903382e-01 3.78093749e-01
7.49270320e-02 -2.40865678e-01 1.44367769e-01 8.89558315e-01
-1.98434606e-01 -1.61520720e-01 8.03432107e-01 2.02109650e-01
-7.59076893e-01 1.27334833e-01 -2.95305640e-01 -5.45866430e-01
1.53905511e+00 -8.59539568e-01 3.56455147e-01 -2.87418693e-01
-1.09430349e+00 3.01331319e-02 6.77473307e-01 5.45972548e-02
6.91704512e-01 -1.55999517e+00 -6.00376546e-01 4.67250168e-01
-6.54042959e-02 -3.53685886e-01 1.23709358e-01 1.05614841e+00
-3.50567698e-01 5.98737478e-01 -3.28043774e-02 -1.07332730e+00
-1.53150094e+00 1.10021889e+00 2.57074654e-01 5.61889969e-02
-4.35380667e-01 6.78304851e-01 2.22252786e-01 -2.97628701e-01
1.63451076e-01 2.96734691e-01 -3.58645886e-01 1.46280766e-01
2.12314785e-01 7.59603024e-01 -2.96611786e-01 -1.19646859e+00
-3.47646773e-01 1.04750645e+00 3.22937131e-01 -2.82496870e-01
1.08475161e+00 -1.11513823e-01 -5.19833207e-01 4.73116338e-01
1.39903438e+00 1.45430520e-01 -1.19639301e+00 -2.03405485e-01
1.54356614e-01 -4.23250884e-01 -8.63792840e-03 -3.28507274e-01
-9.36788023e-01 5.77177763e-01 8.75185132e-01 3.20091873e-01
1.54418325e+00 1.19937934e-01 5.31463444e-01 -4.01522592e-02
5.00250578e-01 -1.12517715e+00 -1.81813166e-01 -2.94830110e-02
7.89671183e-01 -7.67779708e-01 -8.80065709e-02 -9.43727076e-01
-3.05337548e-01 1.28783298e+00 1.05799355e-01 -2.45304391e-01
8.15619409e-01 -2.11186424e-01 -2.27603152e-01 -2.30234504e-01
-3.85556407e-02 -9.21429619e-02 6.93595290e-01 5.44983625e-01
1.75512388e-01 2.03645959e-01 -6.17778599e-01 4.30864125e-01
7.32043982e-02 -2.51130700e-01 5.96677810e-02 5.30473053e-01
-5.99970937e-01 -1.06028152e+00 -1.29518127e+00 1.56789288e-01
-4.46713753e-02 1.72594935e-01 -6.56461060e-01 5.34548342e-01
1.93700969e-01 1.11597872e+00 9.24331471e-02 -3.63902032e-01
1.71324968e-01 1.54706225e-01 3.41545939e-01 -2.59884268e-01
-7.24924430e-02 8.02533388e-01 -3.63965303e-01 -5.08704960e-01
-4.77609336e-01 -1.05655873e+00 -1.37660027e+00 -1.23933509e-01
-1.53862834e-01 7.07953572e-01 4.56385583e-01 7.59512007e-01
2.51892298e-01 1.82296280e-02 1.00743401e+00 -7.63985991e-01
-4.34881240e-01 -4.57538873e-01 -1.03005373e+00 4.42479789e-01
6.11968823e-02 -5.19446075e-01 -5.73088765e-01 2.62282372e-01] | [7.7387800216674805, 4.4289164543151855] |
0ae10564-2a8c-4724-8b48-b8f4a059507b | certification-of-semantic-perturbations-via | 2002.12463 | null | https://arxiv.org/abs/2002.12463v4 | https://arxiv.org/pdf/2002.12463v4.pdf | Certified Defense to Image Transformations via Randomized Smoothing | We extend randomized smoothing to cover parameterized transformations (e.g., rotations, translations) and certify robustness in the parameter space (e.g., rotation angle). This is particularly challenging as interpolation and rounding effects mean that image transformations do not compose, in turn preventing direct certification of the perturbed image (unlike certification with $\ell^p$ norms). We address this challenge by introducing three different kinds of defenses, each with a different guarantee (heuristic, distributional and individual) stemming from the method used to bound the interpolation error. Importantly, we show how individual certificates can be obtained via either statistical error bounds or efficient online inverse computation of the image transformation. We provide an implementation of all methods at https://github.com/eth-sri/transformation-smoothing. | ['Martin Vechev', 'Marc Fischer', 'Maximilian Baader'] | 2020-02-27 | null | http://proceedings.neurips.cc/paper/2020/hash/5fb37d5bbdbbae16dea2f3104d7f9439-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/5fb37d5bbdbbae16dea2f3104d7f9439-Paper.pdf | neurips-2020-12 | ['provable-adversarial-defense'] | ['adversarial'] | [ 1.25956908e-01 8.55424255e-02 4.76350263e-02 2.29609162e-02
-1.42620444e+00 -1.36663413e+00 3.50742847e-01 2.95827910e-02
-3.88057321e-01 7.00449586e-01 -1.13390356e-01 -5.85853755e-01
3.70202512e-02 -5.89452982e-01 -1.27871680e+00 -8.46635044e-01
-2.65099585e-01 -1.48534484e-03 7.61213154e-02 -3.00872959e-02
3.91481638e-01 4.39013749e-01 -1.00764143e+00 -1.60675883e-01
8.09780419e-01 8.55209887e-01 -5.05043149e-01 9.70970035e-01
5.43630898e-01 2.38958821e-01 -5.32833099e-01 -6.75189018e-01
8.38336647e-01 -2.55968928e-01 -7.58516848e-01 2.38555949e-02
6.67505085e-01 -6.23357236e-01 -2.35488430e-01 1.39650905e+00
4.56163168e-01 1.62553698e-01 5.14785826e-01 -1.54899204e+00
-6.05961978e-01 4.60491091e-01 -6.31488442e-01 -2.86193162e-01
4.67891842e-01 4.07327920e-01 7.08593667e-01 -6.07780099e-01
6.46345258e-01 9.30236101e-01 7.51328230e-01 2.94981182e-01
-1.48865485e+00 -6.59233749e-01 -3.23896855e-02 -1.60576925e-01
-1.74948418e+00 -4.48182970e-01 2.36697316e-01 -3.46024960e-01
1.95212156e-01 5.65758407e-01 1.08061351e-01 9.51471210e-01
-4.95367758e-02 2.49351501e-01 1.34849930e+00 -2.59558648e-01
1.77498102e-01 2.10570171e-01 -1.20550722e-01 5.72516978e-01
4.02162790e-01 2.09084809e-01 -9.14961696e-02 -7.83294201e-01
7.96650231e-01 -4.57585961e-01 -6.00392759e-01 -5.34102023e-01
-1.23695481e+00 5.57470739e-01 8.93270820e-02 -3.93035144e-01
4.36621979e-02 7.54290581e-01 3.44523787e-01 4.38959211e-01
1.60230473e-01 5.88695370e-02 -5.93102396e-01 -4.05553691e-02
-8.13213229e-01 4.70240325e-01 7.09488511e-01 1.07769513e+00
8.25016320e-01 -2.33547855e-03 2.61838082e-02 2.87585199e-01
8.76128823e-02 7.69870162e-01 -2.06805021e-01 -1.45145857e+00
3.61904800e-01 -3.27564508e-01 4.82113779e-01 -1.13354588e+00
-1.14433859e-02 4.49319668e-02 -6.18092120e-01 4.89917248e-01
9.61583197e-01 -3.85769010e-01 -6.49263918e-01 2.09385371e+00
6.00012004e-01 5.84183037e-01 -2.21014157e-01 6.93518579e-01
9.91656259e-03 4.57987249e-01 -1.63367808e-01 3.69859375e-02
1.41406226e+00 -4.28451061e-01 -1.78145975e-01 1.56862721e-01
6.38736069e-01 -8.70039284e-01 1.08406758e+00 5.56963086e-01
-1.17185426e+00 -6.47260100e-02 -1.01654756e+00 -2.14824140e-01
-2.98019469e-01 -8.65142792e-02 2.65905499e-01 9.23861265e-01
-1.13895249e+00 5.78756213e-01 -8.63563001e-01 6.63142055e-02
3.71356905e-01 5.67935586e-01 -5.79617679e-01 1.53670430e-01
-1.17800713e+00 5.37017465e-01 -1.23060636e-01 -2.53072053e-01
-8.73111069e-01 -9.22506213e-01 -7.77567387e-01 -1.29969150e-01
2.89914072e-01 -4.84633237e-01 1.07447076e+00 -6.66279256e-01
-1.40698016e+00 9.29940104e-01 -6.48559257e-02 -6.36715293e-01
1.05338717e+00 -6.50232509e-02 -2.42750291e-02 1.06381081e-01
-3.47774662e-02 4.93950456e-01 1.02521813e+00 -1.36594689e+00
-5.02020776e-01 -1.86847582e-01 1.82350725e-01 -1.04929179e-01
-1.58995628e-01 5.20749725e-02 -4.40294951e-01 -4.95321959e-01
-1.68463469e-01 -1.31268680e+00 -5.00965118e-01 2.88357109e-01
-6.99462891e-01 4.91353303e-01 4.97215956e-01 -9.61362600e-01
9.53041196e-01 -2.52562737e+00 -2.75887191e-01 6.71471477e-01
7.01306239e-02 -5.64550087e-02 -7.91498125e-02 3.49124730e-01
1.15590747e-02 5.50362527e-01 -4.36602294e-01 -2.77621239e-01
2.95398355e-01 -6.21118732e-02 -7.93796241e-01 1.18531501e+00
-1.72670126e-01 4.89575177e-01 -6.69358075e-01 -4.08477783e-01
5.48343547e-02 7.61950612e-01 -7.07041979e-01 -4.05156583e-01
-1.88561365e-01 6.91851854e-01 -2.80603707e-01 4.31266516e-01
1.18872523e+00 1.62907876e-02 2.09349006e-01 -3.20920423e-02
-6.98084608e-02 1.25158533e-01 -1.62992620e+00 1.46206069e+00
-4.27770555e-01 2.70242751e-01 3.87924165e-01 -5.15133440e-01
5.16270876e-01 3.60008150e-01 4.71564025e-01 9.12143104e-03
1.72496125e-01 2.75130332e-01 -4.06503618e-01 4.10702042e-02
5.17312288e-01 2.66790479e-01 -2.64821380e-01 5.74685812e-01
-3.74961972e-01 -3.89424235e-01 -1.31590545e-01 1.82010606e-01
8.94358575e-01 2.08406076e-01 2.40748242e-01 -3.64775389e-01
4.28919822e-01 -5.42039871e-02 6.55613720e-01 6.80772305e-01
-3.75265658e-01 7.68190801e-01 7.99245059e-01 -1.30730018e-01
-1.20800722e+00 -1.13981235e+00 -1.32345453e-01 7.02037156e-01
2.61841416e-01 -3.70380133e-01 -9.69203472e-01 -5.85175157e-01
3.22525144e-01 4.91008848e-01 -5.82920432e-01 -4.21875715e-02
-4.35330361e-01 -3.44536752e-01 1.01364541e+00 2.11360306e-01
3.00973088e-01 -2.13216662e-01 -1.19714245e-01 -1.29902810e-01
-7.07096681e-02 -1.20032334e+00 -1.08444548e+00 -9.41649526e-02
-6.19590044e-01 -1.04341221e+00 -3.36544633e-01 -5.00564754e-01
1.15171242e+00 2.31016070e-01 7.43321180e-01 3.40808511e-01
-5.37344604e-04 6.07393146e-01 8.15244764e-02 -2.38333628e-01
-2.90964872e-01 -2.34715044e-01 -1.47047769e-02 9.05606821e-02
-3.24115604e-01 -7.21780479e-01 -8.22514296e-01 3.50769311e-01
-9.22006726e-01 -4.55425352e-01 5.55155333e-03 6.69866145e-01
9.23531890e-01 7.96600059e-02 8.50098487e-03 -8.89213920e-01
6.33616686e-01 -3.41021061e-01 -1.34493756e+00 2.45430082e-01
-3.90908778e-01 -4.22547609e-02 7.48445570e-01 -5.91336906e-01
-4.54280019e-01 3.39269698e-01 -3.96759063e-03 -3.88965964e-01
-6.85982108e-02 7.34613389e-02 -1.83689430e-01 -5.71174383e-01
7.61752069e-01 2.35793203e-01 -3.15247141e-02 -7.45250359e-02
7.51533508e-01 3.79710525e-01 8.77982080e-01 -1.16969323e+00
1.30828273e+00 8.46655607e-01 2.55085677e-01 -3.04720700e-01
-3.43779147e-01 -1.84073877e-02 -1.58700854e-01 -3.90633307e-02
5.21559656e-01 -8.43917310e-01 -1.08645511e+00 3.97378236e-01
-1.07794356e+00 -5.88244379e-01 -3.26805294e-01 2.76764929e-01
-7.38433838e-01 8.15428972e-01 -5.64949751e-01 -7.37330914e-01
-1.41038835e-01 -1.43101537e+00 9.85372663e-01 -2.84562167e-03
-9.48720425e-02 -7.88126409e-01 -5.63810878e-02 1.31273374e-01
2.97133714e-01 5.97361505e-01 3.99915189e-01 -4.48548287e-01
-8.71609688e-01 -5.09038925e-01 -1.51293933e-01 4.38757271e-01
-1.18189611e-01 4.16088849e-01 -8.47857118e-01 -4.25003707e-01
-1.16356105e-01 -1.33796558e-01 4.59083408e-01 4.24157917e-01
1.40232718e+00 -7.87865639e-01 8.09905156e-02 9.98377085e-01
1.34793651e+00 -3.25595170e-01 9.11565244e-01 4.17832077e-01
3.63372207e-01 2.11202666e-01 6.41348660e-01 7.12631762e-01
3.12375098e-01 6.28400385e-01 3.16669255e-01 1.61578760e-01
1.11562051e-01 -2.05706015e-01 6.77569866e-01 6.07199073e-02
-1.15969747e-01 1.38927162e-01 -6.27217650e-01 3.57696801e-01
-1.57469440e+00 -8.54632139e-01 -1.67525038e-01 2.87708831e+00
1.15983987e+00 -1.71273679e-01 1.17426038e-01 5.20395413e-02
7.90603340e-01 -1.80860795e-02 -3.05198610e-01 -5.00738561e-01
-1.33331582e-01 2.87428975e-01 1.32150888e+00 1.07144725e+00
-9.05346036e-01 8.52206230e-01 6.18310833e+00 1.01341951e+00
-1.12937224e+00 1.19392708e-01 7.74401367e-01 5.05942963e-02
-5.06941736e-01 4.38482344e-01 -5.21562219e-01 5.14111042e-01
7.18734682e-01 -3.16433311e-01 8.66946340e-01 8.77779543e-01
1.97985679e-01 6.77016154e-02 -6.00818157e-01 6.04324698e-01
-3.71703684e-01 -1.22296238e+00 -2.19281033e-01 4.63317305e-01
8.46506655e-01 -1.77210838e-01 5.63618183e-01 -1.80823058e-01
8.24638009e-01 -8.51186633e-01 8.40961158e-01 8.97357613e-02
1.33337390e+00 -8.56127560e-01 8.93341303e-02 -3.51611637e-02
-1.04015541e+00 3.12963337e-01 -2.90685177e-01 3.92158419e-01
1.44662680e-02 6.62703753e-01 -5.58005095e-01 6.35154903e-01
4.36106652e-01 -5.46258353e-02 -2.23844409e-01 8.84224892e-01
-5.88038862e-01 6.69383645e-01 -7.21341431e-01 5.51458895e-01
-9.99641195e-02 -4.06392485e-01 6.93186104e-01 1.01548541e+00
6.42954528e-01 -8.57679546e-02 2.27391362e-01 6.94089651e-01
-2.64419734e-01 6.20885231e-02 -4.00903106e-01 2.52248645e-01
7.96995759e-01 1.13046372e+00 -5.57680070e-01 -1.41852930e-01
-2.55624890e-01 8.79359007e-01 -1.09854870e-01 4.90783364e-01
-1.33052087e+00 -3.92281502e-01 9.19137537e-01 7.94588178e-02
3.63069266e-01 -6.28660500e-01 -5.66640675e-01 -1.13024271e+00
1.38916194e-01 -1.18573904e+00 3.85938466e-01 -4.67682540e-01
-8.85830581e-01 2.15359896e-01 -7.70158991e-02 -1.25487900e+00
-1.25942439e-01 -2.99658298e-01 -5.86462200e-01 9.01672840e-01
-1.37062144e+00 -1.04321682e+00 -5.79796024e-02 9.07096207e-01
-3.65747213e-01 4.71055567e-01 6.22014761e-01 2.86065698e-01
-2.20983505e-01 1.11849833e+00 1.29253253e-01 1.16703339e-01
1.22252381e+00 -1.23365271e+00 5.03777504e-01 1.18217671e+00
6.12435006e-02 8.64440739e-01 9.44209099e-01 -3.66390914e-01
-1.59894240e+00 -1.00790024e+00 5.83107889e-01 -4.88947332e-01
9.41443682e-01 -3.70605648e-01 -6.09349787e-01 9.36294734e-01
-1.33771077e-01 4.54918474e-01 4.88103062e-01 -1.67597771e-01
-8.45843732e-01 -7.70889297e-02 -1.64954948e+00 9.89136040e-01
7.92120039e-01 -6.04815722e-01 4.58279490e-01 4.62006837e-01
7.99585044e-01 -8.18273008e-01 -1.02975404e+00 2.08809033e-01
4.91213650e-01 -8.13239515e-01 1.13555837e+00 -3.25831890e-01
-3.75701152e-02 -8.29663157e-01 -3.87552738e-01 -9.46199119e-01
7.85926357e-02 -1.32121658e+00 -1.07683271e-01 1.30426729e+00
5.50207734e-01 -1.06322289e+00 5.32785475e-01 7.81851232e-01
1.56463310e-01 -5.31483889e-01 -1.01853800e+00 -1.12392831e+00
4.08885479e-01 -6.25604928e-01 9.49633718e-01 9.42912042e-01
-5.77512793e-02 -6.82787180e-01 -5.30998588e-01 8.10892820e-01
8.43346179e-01 -1.50224566e-01 1.17557585e+00 -4.85730618e-01
-4.02146280e-01 -4.02290940e-01 -5.04467189e-01 -8.36073160e-01
-2.89194169e-03 -6.44706845e-01 -2.85499915e-02 -8.47036123e-01
-1.71881929e-01 -6.03489280e-01 -9.12786126e-02 4.54023302e-01
-9.97812450e-02 3.04440230e-01 1.98619336e-01 9.58846360e-02
-3.52041900e-01 -5.32925949e-02 1.04582298e+00 8.32467228e-02
-1.87304597e-02 2.11385116e-02 -9.29889023e-01 5.56133032e-01
1.07905483e+00 -4.63583797e-01 -2.47133911e-01 -2.90706396e-01
3.83381575e-01 1.13482222e-01 7.65562057e-01 -9.34688270e-01
2.29796112e-01 -3.18762749e-01 -3.80186439e-02 7.42193535e-02
1.53442144e-01 -7.02028513e-01 4.02001441e-01 4.14450914e-01
-4.00981486e-01 2.25707144e-02 1.23714194e-01 4.80337203e-01
6.26801699e-02 1.30273309e-02 1.09538698e+00 3.03414643e-01
-3.58368568e-02 6.40586019e-01 -7.70273432e-03 1.70386821e-01
9.91186202e-01 -1.15713120e-01 -6.41691923e-01 -6.58377945e-01
-6.16698682e-01 4.04658839e-02 1.04751754e+00 -2.69065112e-01
2.43923232e-01 -1.22501540e+00 -8.07624340e-01 -1.44053087e-01
-1.26887187e-01 1.83587410e-02 2.72920251e-01 1.22351050e+00
-6.69532180e-01 -1.33897334e-01 2.05902427e-01 -1.54450461e-01
-1.25502157e+00 6.24562621e-01 2.99863726e-01 -2.29155138e-01
-5.00613630e-01 6.99543059e-01 -1.45832412e-02 -3.53936672e-01
2.93172956e-01 -3.24446768e-01 7.81936467e-01 -4.45677161e-01
5.14819264e-01 3.45375925e-01 -1.00916326e-01 -3.91975671e-01
-3.42955947e-01 6.66683018e-01 2.18250036e-01 -5.70409238e-01
8.99460673e-01 -2.74541765e-01 -1.88144639e-01 -3.14153671e-01
1.19798517e+00 8.58109236e-01 -1.51432443e+00 4.24142890e-02
-3.04020286e-01 -7.31086016e-01 -2.17427224e-01 -4.63954002e-01
-1.15580118e+00 4.56043988e-01 3.49176347e-01 1.41905338e-01
1.03366828e+00 -4.00869519e-01 9.46363151e-01 1.47517502e-01
4.76183206e-01 -1.17515874e+00 -6.72700644e-01 2.76782990e-01
8.07071030e-01 -9.36170459e-01 2.85747617e-01 -6.17046177e-01
-5.25686622e-01 8.53538156e-01 1.02292048e-02 -5.17908633e-01
7.63633609e-01 6.29910886e-01 1.77210838e-01 4.02867198e-01
-4.74213243e-01 5.38637452e-02 3.27705331e-02 7.07265973e-01
1.83000326e-01 1.42780617e-01 -3.42461944e-01 4.18216646e-01
-3.94065499e-01 -1.16488539e-01 8.01093876e-01 8.29881966e-01
-1.37190521e-01 -1.42918003e+00 -8.48793387e-01 9.04441439e-03
-7.64288485e-01 -1.54268220e-01 -1.62899062e-01 7.11805105e-01
6.69984668e-02 8.85167658e-01 -4.00043666e-01 -3.70534509e-01
6.44000107e-03 -2.43887082e-01 4.70547885e-01 -2.02883631e-02
-4.57457483e-01 -7.23952949e-02 3.95270884e-02 -6.85561001e-01
2.66809929e-02 -8.78983676e-01 -1.40074027e+00 -9.72268105e-01
-1.76417217e-01 1.51263416e-01 7.57012129e-01 3.64633888e-01
5.31480551e-01 -1.96306139e-01 1.15500426e+00 -5.35741091e-01
-9.51978743e-01 -1.92843512e-01 -6.07337236e-01 4.20509428e-01
3.59876275e-01 -1.12206183e-01 -8.51907492e-01 3.33597720e-01] | [5.871531963348389, 7.341946125030518] |
6cc8fa1b-e61c-45be-9572-d459881eaa77 | learning-deep-structured-multi-scale-features | 1801.00524 | null | http://arxiv.org/abs/1801.00524v1 | http://arxiv.org/pdf/1801.00524v1.pdf | Learning Deep Structured Multi-Scale Features using Attention-Gated CRFs for Contour Prediction | Recent works have shown that exploiting multi-scale representations deeply
learned via convolutional neural networks (CNN) is of tremendous importance for
accurate contour detection. This paper presents a novel approach for predicting
contours which advances the state of the art in two fundamental aspects, i.e.
multi-scale feature generation and fusion. Different from previous works
directly consider- ing multi-scale feature maps obtained from the inner layers
of a primary CNN architecture, we introduce a hierarchical deep model which
produces more rich and complementary representations. Furthermore, to refine
and robustly fuse the representations learned at different scales, the novel
Attention-Gated Conditional Random Fields (AG-CRFs) are proposed. The
experiments ran on two publicly available datasets (BSDS500 and NYUDv2)
demonstrate the effectiveness of the latent AG-CRF model and of the overall
hierarchical framework. | ['Xavier Alameda-Pineda', 'Elisa Ricci', 'Wanli Ouyang', 'Xiaogang Wang', 'Nicu Sebe', 'Dan Xu'] | 2018-01-01 | learning-deep-structured-multi-scale-features-1 | http://papers.nips.cc/paper/6985-learning-deep-structured-multi-scale-features-using-attention-gated-crfs-for-contour-prediction | http://papers.nips.cc/paper/6985-learning-deep-structured-multi-scale-features-using-attention-gated-crfs-for-contour-prediction.pdf | neurips-2017-12 | ['contour-detection'] | ['computer-vision'] | [ 8.90269727e-02 1.44013569e-01 -1.19998582e-01 -4.98493105e-01
-9.82551277e-01 -4.25603509e-01 6.25316501e-01 3.03474426e-01
-2.89386690e-01 6.23525798e-01 3.36670339e-01 5.27893715e-02
1.27008870e-01 -1.02155232e+00 -6.50012791e-01 -6.12259150e-01
-9.68988165e-02 1.34710476e-01 4.82182592e-01 -2.55009383e-01
2.51335323e-01 7.94589639e-01 -1.60734427e+00 2.18595549e-01
9.88230050e-01 1.46196949e+00 1.69442430e-01 6.49087191e-01
-2.27979004e-01 5.69865644e-01 -4.55979288e-01 -4.66884077e-01
2.10084431e-02 -7.07811043e-02 -7.26548314e-01 9.78966430e-02
7.85300910e-01 -1.68686748e-01 1.45704318e-02 1.10908401e+00
1.20287642e-01 7.71831200e-02 6.63802266e-01 -7.56289423e-01
-9.27717924e-01 2.83866584e-01 -6.82185829e-01 7.12485835e-02
-9.14388448e-02 -6.32128417e-01 9.74369049e-01 -8.26390207e-01
6.56814754e-01 1.25843537e+00 8.07390034e-01 2.63863206e-01
-1.14953625e+00 -4.82900709e-01 3.77675384e-01 -1.23633496e-01
-1.29799128e+00 4.59255390e-02 7.35402584e-01 -5.29420972e-01
7.44919002e-01 -9.76596251e-02 5.50916672e-01 7.19671249e-01
5.93598604e-01 9.39981461e-01 1.25162530e+00 -4.68676150e-01
-8.10008682e-03 -2.54198611e-01 2.26309866e-01 8.60546410e-01
3.91680092e-01 2.36057103e-01 -5.37944674e-01 2.60901283e-02
1.23206043e+00 -3.62004712e-02 -7.19998404e-03 -3.69434416e-01
-9.29719627e-01 1.09886515e+00 8.68083596e-01 4.70542371e-01
-5.42107403e-01 1.36665389e-01 6.64559230e-02 -1.33243352e-01
8.38348389e-01 1.86486214e-01 -6.14482999e-01 2.96447217e-01
-1.22712767e+00 1.52833983e-01 4.16285753e-01 8.69657397e-01
1.09599125e+00 6.04407117e-03 -4.13231671e-01 5.93865514e-01
6.35817945e-01 2.45897934e-01 4.25010502e-01 -6.60398006e-01
9.72800925e-02 6.98180497e-01 -5.87715693e-02 -9.48307872e-01
-6.11309290e-01 -5.44140339e-01 -1.00440335e+00 5.77215552e-01
3.94369304e-01 -2.34695077e-01 -1.62375259e+00 1.58423507e+00
1.56368166e-01 4.04286623e-01 2.10905433e-01 6.78583503e-01
9.27244067e-01 4.34116542e-01 4.52038437e-01 2.42064655e-01
1.28801477e+00 -1.17391753e+00 -7.65793741e-01 -2.63683766e-01
1.63131207e-01 -9.78631854e-01 3.32181633e-01 2.76356846e-01
-1.01859689e+00 -1.03334296e+00 -1.19880652e+00 -2.65567601e-01
-7.59972394e-01 3.88632834e-01 1.13170338e+00 4.62729007e-01
-1.44251430e+00 7.80822873e-01 -9.07665968e-01 -8.25315341e-02
7.40145266e-01 2.07189903e-01 -3.41528356e-01 9.20075253e-02
-1.04541731e+00 7.20775306e-01 4.24809963e-01 4.22460645e-01
-5.24507642e-01 -2.43563294e-01 -1.23564911e+00 1.32409319e-01
3.25661153e-02 -3.30997199e-01 9.48422551e-01 -8.23585808e-01
-1.39440644e+00 7.06950247e-01 -2.91564524e-01 -4.90127742e-01
2.79748380e-01 -6.63913190e-01 -7.45936483e-02 2.05279350e-01
9.78961661e-02 9.54220891e-01 7.86055982e-01 -1.23463964e+00
-8.08323205e-01 -2.75317967e-01 -2.39765778e-01 -5.62960804e-02
-1.04294129e-01 5.14523014e-02 -4.53941017e-01 -1.04928517e+00
3.92771393e-01 -6.52890384e-01 -6.13615930e-01 -1.07790800e-02
-5.00175953e-01 -4.30591315e-01 6.89959586e-01 -6.52079523e-01
7.48639703e-01 -1.89922655e+00 1.89532533e-01 2.02034265e-01
1.46315679e-01 2.82684654e-01 -1.12842262e-01 1.08922735e-01
3.69857289e-02 2.10745558e-01 -4.16011542e-01 -7.33018160e-01
-2.40971297e-01 1.81799486e-01 -2.40467846e-01 1.88320905e-01
6.70898676e-01 1.29910874e+00 -5.91483414e-01 -6.60134017e-01
3.60172600e-01 7.54165411e-01 -1.21289954e-01 3.67980987e-01
-1.53063729e-01 3.46029520e-01 -2.97644943e-01 8.04696381e-01
1.04687369e+00 -3.66703570e-01 7.22605782e-03 -2.57694691e-01
-4.17751908e-01 -2.28254050e-01 -9.63658273e-01 1.87700546e+00
-2.61290312e-01 5.11067152e-01 4.53864224e-02 -6.53353631e-01
1.34881079e+00 2.92825967e-01 3.17711025e-01 -7.43072152e-01
1.57291368e-01 -2.70913471e-03 -6.08598828e-01 9.78147164e-02
8.02466094e-01 -6.63289726e-02 -2.19126046e-01 -1.10279061e-01
7.35152185e-01 -7.56641179e-02 7.46710896e-02 1.21211119e-01
5.92381060e-01 6.46629274e-01 3.67875248e-01 -4.26712096e-01
4.43803340e-01 -1.02734715e-01 7.68915832e-01 6.85529709e-01
-2.48646468e-01 1.09520364e+00 2.64151335e-01 -4.59779024e-01
-7.29778409e-01 -1.13273585e+00 -3.07740182e-01 1.17435181e+00
3.24798554e-01 -2.49148503e-01 -5.23152947e-01 -7.72459507e-01
-2.55366601e-03 3.05267990e-01 -1.05953121e+00 2.54552037e-01
-6.29526019e-01 -9.73925471e-01 3.90594035e-01 1.15588665e+00
5.55869877e-01 -1.40526688e+00 -6.63521349e-01 3.24049145e-01
7.22003505e-02 -1.24178183e+00 -1.56633720e-01 4.44679618e-01
-9.82239306e-01 -8.52178216e-01 -9.30000484e-01 -9.70317125e-01
4.88344431e-01 4.16562669e-02 1.19030213e+00 7.71106854e-02
-4.40360844e-01 2.97774076e-02 -6.86704457e-01 -4.75588650e-01
-3.42671946e-02 1.50821522e-01 -7.18168139e-01 1.93624228e-01
1.16883390e-01 -3.04848790e-01 -3.91173095e-01 -1.53307751e-01
-8.18917572e-01 2.46312186e-01 1.09728456e+00 6.47078395e-01
7.92897165e-01 -2.27363929e-01 6.91397369e-01 -9.00965810e-01
3.31130505e-01 -3.63929212e-01 -7.09665477e-01 4.26497877e-01
-6.28803253e-01 2.81854212e-01 2.22233221e-01 -8.86785164e-02
-1.45037901e+00 4.39482689e-01 -5.69219291e-01 -2.42915079e-01
-8.25834811e-01 4.33408141e-01 1.31953403e-01 -1.59283638e-01
1.59404919e-01 1.40068233e-01 -3.17536891e-01 -7.37065554e-01
6.24046504e-01 1.83421522e-01 8.90720069e-01 -4.29497868e-01
7.25947261e-01 7.29140580e-01 6.21863417e-02 -5.17034054e-01
-1.04749441e+00 -5.83143294e-01 -1.28446233e+00 -3.86357456e-02
1.41361618e+00 -9.34954226e-01 -2.07133979e-01 8.28092575e-01
-1.27290797e+00 -2.63461649e-01 -2.67978311e-01 2.49829993e-01
-5.22412956e-01 3.19433570e-01 -8.88362169e-01 -8.51214290e-01
-4.54004377e-01 -9.18338299e-01 1.49422824e+00 9.29542840e-01
3.61337155e-01 -1.20979261e+00 3.25867504e-01 -7.36177340e-02
6.31253183e-01 6.94243729e-01 5.58002532e-01 -4.39682722e-01
-5.09675145e-01 -1.61819056e-01 -6.31406844e-01 2.31564522e-01
1.26105040e-01 7.94207677e-02 -1.27244663e+00 -7.65363351e-02
-5.79334080e-01 -5.46346307e-01 1.52109814e+00 6.02829397e-01
8.73195827e-01 2.54495561e-01 -3.14078569e-01 7.51231074e-01
1.72200119e+00 -2.55884491e-02 8.07022452e-01 2.39180431e-01
7.11030483e-01 4.06539470e-01 6.45011365e-01 1.23490639e-01
5.02627611e-01 4.47746098e-01 5.05147517e-01 -8.18217695e-01
-4.16566133e-01 -1.21845946e-01 1.84806418e-02 5.49228430e-01
-3.44592899e-01 -3.66585068e-02 -5.38245678e-01 7.08420515e-01
-1.94917715e+00 -5.91072619e-01 -6.61080033e-02 1.60517645e+00
6.03168368e-01 1.96971685e-01 -2.37560451e-01 -1.86514169e-01
6.62478030e-01 3.32670122e-01 -2.86643803e-01 -3.99041116e-01
-4.48269695e-01 7.76398957e-01 4.21843469e-01 4.70991284e-01
-1.78342760e+00 1.28353608e+00 6.78852606e+00 7.35083640e-01
-1.17438340e+00 1.10378489e-02 6.83125734e-01 8.64740908e-01
-5.24628572e-02 -4.03037742e-02 -1.05811310e+00 -6.53746724e-02
7.07904696e-01 2.85867453e-01 -3.53688061e-01 9.53109562e-01
-4.88996506e-01 -1.84521884e-01 -4.20164555e-01 4.61173773e-01
1.54532596e-01 -1.57089543e+00 -5.66951893e-02 8.74923356e-03
7.83086479e-01 3.97351116e-01 -3.97866108e-02 2.87064493e-01
7.50427604e-01 -1.00534904e+00 9.18918252e-01 8.85965824e-01
7.75252223e-01 -6.76258981e-01 1.06592464e+00 8.62518325e-02
-1.80347586e+00 2.37473890e-01 -5.56459725e-01 1.17481381e-01
4.30420429e-01 5.65681279e-01 -5.14174581e-01 9.98850346e-01
8.06949973e-01 8.75629067e-01 -1.00590229e+00 1.10513580e+00
-3.77150238e-01 5.19608259e-01 -2.06428707e-01 3.78057003e-01
3.54840487e-01 -8.76785070e-03 -1.14377677e-01 1.60200715e+00
1.79408029e-01 -1.24310084e-01 2.99321949e-01 8.96345496e-01
4.52894010e-02 1.54354647e-01 -2.77705520e-01 2.45105296e-01
-1.18448630e-01 1.85333276e+00 -1.25211549e+00 -3.51993144e-01
-6.65302873e-01 7.92083204e-01 7.06043482e-01 2.48700887e-01
-6.24028087e-01 -2.89898366e-01 4.18987632e-01 -4.03614908e-01
7.54353285e-01 -4.58453357e-01 -3.62164527e-01 -1.10419631e+00
-4.25207764e-01 2.17175223e-02 4.82480049e-01 -5.90264618e-01
-1.56406605e+00 1.08343065e+00 -1.58871576e-01 -8.49282324e-01
-7.15830997e-02 -6.85102999e-01 -7.77584791e-01 9.64859903e-01
-2.21569419e+00 -1.80804801e+00 -5.14546692e-01 5.22176743e-01
5.98993301e-01 2.05428481e-01 1.11365068e+00 2.47414876e-02
-2.93063700e-01 4.05949265e-01 -2.34281808e-01 5.20756125e-01
4.70982790e-01 -1.58986986e+00 8.08923066e-01 9.43003178e-01
4.36161309e-01 4.02671099e-01 7.41979256e-02 -9.11581337e-01
-5.82834423e-01 -1.35200214e+00 8.26014817e-01 -1.62138388e-01
2.51750320e-01 -4.30219650e-01 -1.01385176e+00 6.21735871e-01
4.33789432e-01 3.90057057e-01 4.80554730e-01 5.45399860e-02
-2.91907728e-01 3.06844264e-01 -1.01974356e+00 1.13653287e-03
8.02889764e-01 -2.62336135e-01 -6.11633956e-01 -1.32536054e-01
6.95468187e-01 -3.68043989e-01 -9.76716161e-01 8.48091781e-01
5.60705006e-01 -1.11898172e+00 1.00823176e+00 -4.21019226e-01
1.94062397e-01 -6.19043186e-02 -1.93795979e-01 -1.15406120e+00
-5.59890747e-01 -1.60645232e-01 -1.15107194e-01 1.25199103e+00
3.57217938e-01 -4.28172439e-01 9.84586418e-01 6.99829832e-02
-3.11464936e-01 -9.16313887e-01 -8.95818532e-01 -3.94189209e-01
4.91727680e-01 -2.84780294e-01 5.59918702e-01 5.79423010e-01
-4.53972757e-01 1.60615802e-01 -2.90253699e-01 3.25385571e-01
7.16847837e-01 5.90456724e-01 3.73789847e-01 -1.84760499e+00
1.89367935e-01 -2.64139205e-01 -5.25405824e-01 -1.19029450e+00
1.00990117e-01 -7.60212243e-01 4.30253386e-01 -1.88142300e+00
1.19341791e-01 -5.35261035e-01 -6.15947485e-01 5.55767536e-01
-4.32780325e-01 6.27159536e-01 3.29757273e-01 -1.00527182e-02
-7.09974229e-01 6.12438619e-01 1.28956568e+00 2.17623755e-01
1.09199109e-02 5.72759546e-02 -7.54896820e-01 9.73669350e-01
7.49737084e-01 -3.49742532e-01 1.75580502e-01 -2.14087486e-01
-7.52297491e-02 -3.58858844e-03 4.01479125e-01 -1.16509795e+00
3.98703843e-01 -4.25778627e-02 9.11812186e-01 -1.05371892e+00
4.52902794e-01 -5.04125953e-01 -5.40879846e-01 1.65583670e-01
-8.99602100e-02 -3.30509134e-02 4.45227087e-01 7.75635481e-01
-5.88886976e-01 1.72109697e-02 8.99501920e-01 -1.06655799e-01
-1.03003693e+00 2.98618376e-01 -2.73561805e-01 -2.44134903e-01
9.12009656e-01 5.08254990e-02 -2.89313048e-01 -2.10316226e-01
-1.02464390e+00 -5.64294718e-02 1.93002254e-01 6.66329622e-01
8.07295501e-01 -1.24589217e+00 -7.34778762e-01 4.60304260e-01
-6.63077161e-02 2.09096774e-01 1.66069984e-01 4.59421694e-01
-3.48817080e-01 4.42345381e-01 -4.91141468e-01 -6.07995987e-01
-8.95489931e-01 2.79598832e-01 2.09343955e-01 -7.03785121e-01
-6.61851227e-01 1.18300104e+00 3.83429021e-01 -2.76292235e-01
-1.86496563e-02 -6.00440979e-01 -8.03354561e-01 1.90446839e-01
3.30168575e-01 -4.69276272e-02 7.83889890e-02 -9.98628497e-01
-2.23436192e-01 9.68030393e-01 -2.10194021e-01 4.74581541e-03
1.54445052e+00 1.78610993e-04 3.57095525e-02 1.42585590e-01
9.38642681e-01 -6.97565526e-02 -1.76510727e+00 -1.72970623e-01
3.10622752e-01 -1.83269769e-01 2.33122051e-01 -6.87405348e-01
-1.39877141e+00 9.80247855e-01 7.43339717e-01 1.04737960e-01
1.15896356e+00 1.86346799e-01 6.95413649e-01 -2.23627482e-02
3.21786314e-01 -8.73340607e-01 1.71171322e-01 5.12939870e-01
8.48100781e-01 -1.34354270e+00 -7.15204850e-02 -6.82740510e-01
-7.50004888e-01 1.49876380e+00 6.48994982e-01 -5.95259547e-01
1.00401831e+00 5.94710231e-01 3.29316497e-01 -3.67175013e-01
-3.50166708e-01 -7.22663701e-01 5.69066644e-01 5.73037922e-01
5.34496844e-01 6.16454966e-02 -6.24877959e-02 7.43327260e-01
1.12079449e-01 -6.92867935e-02 3.96955907e-01 1.07600462e+00
-6.51340842e-01 -1.20314431e+00 -3.06997925e-01 3.41366708e-01
-5.58264434e-01 -6.69738129e-02 -2.68781722e-01 6.54584527e-01
4.01452661e-01 9.37704742e-01 1.64031982e-01 -2.10346922e-01
1.06896199e-01 1.19824305e-01 2.69212484e-01 -7.12474704e-01
-5.31489074e-01 3.57645690e-01 -3.43577385e-01 -6.18704498e-01
-8.98526490e-01 -5.54744840e-01 -1.34692287e+00 5.16635001e-01
-6.31848991e-01 3.36110778e-02 7.12899387e-01 1.00286067e+00
2.73881644e-01 7.17389762e-01 2.37122402e-01 -1.37179399e+00
-6.63675219e-02 -1.10417521e+00 -6.70013309e-01 3.52083027e-01
4.42745924e-01 -9.04805958e-01 -1.45025086e-02 2.59237617e-01] | [9.60216999053955, 0.16193534433841705] |
3c2fc354-cf60-4ea9-9fde-bc0d2b3c9c21 | reward-shaping-with-subgoals-for-social | 2104.06410 | null | https://arxiv.org/abs/2104.06410v1 | https://arxiv.org/pdf/2104.06410v1.pdf | Reward Shaping with Subgoals for Social Navigation | Social navigation has been gaining attentions with the growth in machine intelligence. Since reinforcement learning can select an action in the prediction phase at a low computational cost, it has been formulated in a social navigation tasks. However, reinforcement learning takes an enormous number of iterations until acquiring a behavior policy in the learning phase. This negatively affects the learning of robot behaviors in the real world. In particular, social navigation includes humans who are unpredictable moving obstacles in an environment. We proposed a reward shaping method with subgoals to accelerate learning. The main part is an aggregation method that use subgoals to shape a reinforcement learning algorithm. We performed a learning experiment with a social navigation task in which a robot avoided collisions and then reached its goal. The experimental results show that our method improved the learning efficiency from a base algorithm in the task. | ['Seiji Yamada', 'Takato Okudo'] | 2021-04-13 | null | null | null | null | ['social-navigation'] | ['robots'] | [ 6.11788742e-02 2.96462387e-01 1.33503079e-01 -2.48657480e-01
-6.03040047e-02 -4.34558541e-02 4.20582771e-01 1.57137826e-01
-1.05032170e+00 1.16420114e+00 -2.03593925e-01 -1.13472424e-01
-1.96737796e-01 -1.06278634e+00 -5.70916355e-01 -8.45852196e-01
-3.80453736e-01 5.21078646e-01 7.92289436e-01 -7.35721707e-01
6.19948566e-01 3.69115062e-02 -1.67413151e+00 -1.77621022e-01
1.29034650e+00 5.72599471e-01 9.53611076e-01 4.19209480e-01
-1.30975127e-01 1.11808634e+00 -3.41229737e-01 2.26004824e-01
2.76537210e-01 -7.24105179e-01 -7.87026465e-01 -1.25371605e-01
-6.54417336e-01 -2.12074280e-01 -7.76151791e-02 1.21874821e+00
4.86849695e-01 6.22412801e-01 6.92185640e-01 -1.50822437e+00
-5.60797453e-02 7.73590446e-01 -5.74997604e-01 -1.49617344e-01
3.55563343e-01 1.59584656e-01 4.79419380e-01 -1.77403629e-01
5.29857755e-01 1.49461401e+00 3.65323871e-01 6.60166621e-01
-6.09157085e-01 -7.40020454e-01 2.62542814e-01 6.84669793e-01
-7.94783831e-01 3.61835301e-01 4.97404993e-01 -1.26183510e-01
9.14100170e-01 -4.50933248e-01 1.14748716e+00 6.97051287e-01
7.32468367e-01 8.37752819e-01 1.32010937e+00 -2.52777010e-01
6.45544589e-01 -2.36702293e-01 -1.70039967e-01 7.36272216e-01
8.56780112e-02 5.79328775e-01 -1.67932913e-01 3.75819318e-02
6.92613065e-01 -4.23988998e-02 2.50943035e-01 -5.40222466e-01
-8.77766371e-01 9.83284295e-01 7.77513981e-01 1.63141027e-01
-5.41286349e-01 1.96840376e-01 3.55049640e-01 6.35991871e-01
-2.47036159e-01 4.43847388e-01 -5.41374922e-01 -4.48165059e-01
1.16369613e-01 4.74410564e-01 8.67968798e-01 6.46340787e-01
1.01045156e+00 1.60958365e-01 3.23587388e-01 9.51908052e-01
3.16913992e-01 7.65662670e-01 7.19781160e-01 -9.94960189e-01
2.90955007e-01 7.67874181e-01 3.33242983e-01 -9.58589971e-01
-8.59933138e-01 -1.98925301e-01 -5.98635197e-01 1.06321490e+00
4.19053316e-01 -6.00740731e-01 -7.37136304e-01 1.59907389e+00
5.54276407e-01 -1.29539911e-02 4.26569372e-01 8.50548089e-01
2.72025287e-01 8.23734879e-01 2.25174278e-01 -5.35087824e-01
9.27281857e-01 -1.09615827e+00 -6.76335335e-01 -2.18516842e-01
8.41827750e-01 -3.97161603e-01 1.01226819e+00 7.49291480e-01
-5.82471371e-01 -6.07960880e-01 -1.10012281e+00 6.39213681e-01
-3.23244035e-01 -6.20742083e-01 5.41944742e-01 4.25359696e-01
-8.42612565e-01 9.22590852e-01 -7.90680110e-01 -5.26421666e-01
1.38967440e-01 6.46945477e-01 -1.99964911e-01 2.36787409e-01
-1.32913136e+00 1.14067614e+00 7.51950681e-01 -1.78152099e-01
-1.10762763e+00 3.26531678e-01 -6.57508135e-01 -7.38974735e-02
8.17165673e-01 -2.38721967e-01 1.50856721e+00 -1.28993356e+00
-1.88192046e+00 1.52451143e-01 3.16615313e-01 -4.95742977e-01
5.08999348e-01 -2.20528871e-01 -2.39824548e-01 -2.47876272e-01
1.66483641e-01 7.64467359e-01 7.90344119e-01 -1.17587781e+00
-1.38867807e+00 -2.79728711e-01 2.77668595e-01 8.94944131e-01
-1.14155628e-01 -4.20937270e-01 1.44578934e-01 -1.37083381e-01
7.08988234e-02 -1.27615333e+00 -9.69306290e-01 -6.07533634e-01
2.86775112e-01 -5.34717500e-01 7.63800681e-01 -1.21783763e-02
7.93478012e-01 -1.88749182e+00 2.95483857e-01 2.57861435e-01
-1.51869461e-01 7.85656348e-02 -1.56102508e-01 5.17401755e-01
2.68340141e-01 -3.34938318e-01 -2.11843967e-01 3.26929063e-01
-1.84813425e-01 4.01564449e-01 -1.58806786e-01 2.51828823e-02
-4.50323820e-01 4.85543400e-01 -1.48800743e+00 -5.36323667e-01
6.03046753e-02 -5.27031682e-02 -7.82426775e-01 5.03878236e-01
-2.02934191e-01 5.21735609e-01 -7.12514460e-01 2.79134035e-01
3.91243488e-01 2.00160056e-01 1.27767205e-01 5.77815294e-01
-1.92296222e-01 -2.04376519e-01 -1.34149885e+00 1.44900680e+00
-3.56953889e-01 2.21737921e-01 -1.01459458e-01 -1.04471254e+00
1.18320823e+00 -1.21714026e-01 7.24564016e-01 -1.00242591e+00
3.05335253e-01 4.03517485e-01 4.93203491e-01 -6.35181308e-01
5.70867956e-01 -5.97963966e-02 -7.92000219e-02 5.55360496e-01
-1.70171902e-01 -4.63979483e-01 3.47517788e-01 -8.74322206e-02
1.08252597e+00 5.45408309e-01 6.26258671e-01 -3.53632420e-01
7.29186714e-01 4.18143481e-01 6.41155839e-01 9.94716942e-01
-2.78510988e-01 -1.98197514e-01 3.99483562e-01 -5.32783985e-01
-7.02780545e-01 -8.53639066e-01 3.41676265e-01 1.17785287e+00
6.26995981e-01 -1.46585047e-01 -8.57618332e-01 -9.56357658e-01
-2.18817234e-01 6.41028643e-01 -4.33382779e-01 -4.35233235e-01
-7.95342326e-01 -7.44424999e-01 -1.59685612e-01 6.42302409e-02
7.79689908e-01 -1.99079418e+00 -1.15952742e+00 4.44014490e-01
1.40032023e-01 -5.13985634e-01 -8.11434537e-02 5.00617146e-01
-8.14797640e-01 -1.38893390e+00 -3.88463914e-01 -1.15404737e+00
8.30029786e-01 2.10240304e-01 5.18875480e-01 2.85437167e-01
5.10117561e-02 2.91712344e-01 -7.86103487e-01 -7.31576741e-01
-4.84641880e-01 1.41614452e-01 3.05753857e-01 -5.20429611e-01
4.14759308e-01 -5.51894367e-01 -5.27710557e-01 5.91083050e-01
-6.53155327e-01 2.21711680e-01 6.53965175e-01 9.02989864e-01
1.88880682e-01 5.02957284e-01 1.03905129e+00 -6.43577099e-01
1.04509783e+00 -5.23281574e-01 -9.55070376e-01 -1.10702813e-01
-8.70747924e-01 3.05884033e-01 9.39289689e-01 -4.49963659e-01
-1.08330321e+00 2.77380407e-01 1.99474394e-02 6.18039787e-01
-9.57726613e-02 5.00593185e-01 1.14999108e-01 -5.60226887e-02
8.03483546e-01 3.43306839e-01 3.76587063e-01 8.64974931e-02
2.80626863e-02 6.86775386e-01 -1.36917323e-01 -5.80978036e-01
5.96413851e-01 -9.62346271e-02 2.20642760e-01 -6.51380420e-01
-1.90384537e-01 1.26494151e-02 -2.92179197e-01 -6.78071141e-01
5.99961519e-01 -5.36807358e-01 -1.27486742e+00 7.86656380e-01
-7.85910249e-01 -8.02865803e-01 -2.26050362e-01 6.21786296e-01
-1.10085678e+00 2.47428477e-01 -2.89496779e-01 -8.67889047e-01
2.39602802e-03 -1.29437160e+00 1.91688359e-01 8.44089448e-01
1.32889420e-01 -6.56586230e-01 2.28910193e-01 -2.36066684e-01
3.43153089e-01 -1.24066882e-01 6.10929489e-01 -4.40118760e-01
-5.86411715e-01 3.65375251e-01 7.32101277e-02 -5.61726056e-02
1.30588457e-01 -4.85249519e-01 -2.40634799e-01 -2.18928188e-01
-4.91297208e-02 -7.11249709e-01 4.46980953e-01 2.78892606e-01
8.58808517e-01 -3.59734744e-02 -3.36100370e-01 9.94585231e-02
1.33812654e+00 1.38238549e+00 6.68555319e-01 9.10220206e-01
2.54047543e-01 6.93335474e-01 1.47219920e+00 4.63157982e-01
1.84452429e-01 4.69602108e-01 8.25608790e-01 2.44398370e-01
3.79861444e-01 -3.56247485e-01 5.13112605e-01 6.05361581e-01
-3.40478152e-01 -1.99729577e-02 -9.46151376e-01 2.82438427e-01
-2.45219731e+00 -9.27103221e-01 1.21280208e-01 2.06868958e+00
8.22091043e-01 3.29609662e-01 2.39876091e-01 1.06682673e-01
4.97461975e-01 -2.15695351e-01 -8.17984164e-01 -4.15952623e-01
3.14783663e-01 -2.51893044e-01 4.90962744e-01 9.27309453e-01
-8.49434733e-01 1.41633868e+00 6.02518892e+00 7.35060275e-01
-9.65084970e-01 -2.73326188e-01 2.95351654e-01 3.71632963e-01
1.16600908e-01 7.36754462e-02 -9.40655053e-01 3.37568343e-01
6.04099154e-01 -3.40626478e-01 7.61189818e-01 1.23317242e+00
2.76278883e-01 -6.98859274e-01 -5.61429024e-01 8.52602661e-01
-3.77205312e-01 -6.09196424e-01 -2.20898110e-02 -1.65171519e-01
6.70598030e-01 -9.13681835e-02 -1.35342136e-01 7.45026648e-01
9.35206056e-01 -9.36950862e-01 4.50766683e-01 4.42487687e-01
4.66226973e-02 -1.26020682e+00 7.51950800e-01 9.49833214e-01
-1.04599166e+00 -6.15294993e-01 -5.79592943e-01 -6.60874605e-01
1.97458103e-01 -4.28383164e-02 -1.07401741e+00 3.06099623e-01
6.19836867e-01 5.93864441e-01 -3.57510656e-01 1.20768356e+00
-4.25675273e-01 2.68729955e-01 -2.05291376e-01 -1.03997338e+00
6.56971097e-01 -8.07756424e-01 4.15608704e-01 4.94737923e-01
4.31661814e-01 2.54195929e-01 5.82296610e-01 9.29131880e-02
6.53540492e-01 5.47052741e-01 -8.17958415e-01 2.54798025e-01
2.10604146e-01 9.96890545e-01 -1.09849358e+00 -2.34113317e-02
-9.04414877e-02 6.93486154e-01 6.00172639e-01 2.24528074e-01
-9.37203646e-01 -6.72399044e-01 1.45897761e-01 4.45800386e-02
-3.83631103e-02 -2.73052394e-01 -2.51097735e-02 -5.16961098e-01
-3.74535173e-01 -7.37396717e-01 1.14252687e-01 -6.76252604e-01
-7.83274651e-01 7.47380614e-01 -5.12504801e-02 -1.34647357e+00
-5.29435933e-01 -6.10336185e-01 -3.65354955e-01 3.30545127e-01
-1.14125669e+00 -3.36419761e-01 -4.52723771e-01 6.62834883e-01
6.72307968e-01 -7.44703114e-01 7.99754500e-01 -4.07944098e-02
-2.25770190e-01 5.40132858e-02 7.29428157e-02 -2.11708158e-01
5.35652637e-01 -1.36013377e+00 -1.97835162e-01 2.96228617e-01
-4.45807666e-01 2.42401347e-01 8.99914503e-01 -9.92064774e-01
-9.95370626e-01 -5.90992451e-01 3.82614374e-01 3.23749810e-01
5.70758700e-01 6.61223531e-02 -4.88911927e-01 3.82328689e-01
3.57637942e-01 -5.24912715e-01 2.30895996e-01 1.28815815e-01
5.89708149e-01 -2.47757971e-01 -1.09600043e+00 1.07949388e+00
1.16944110e+00 3.36498410e-01 -6.87808514e-01 1.35165289e-01
6.28950298e-01 -4.85166609e-01 -1.83250949e-01 2.85377502e-01
4.98200029e-01 -9.49652612e-01 5.30173004e-01 -4.66309518e-01
3.91462445e-01 -5.14739990e-01 2.75886089e-01 -1.80127978e+00
-4.22231913e-01 -6.00325227e-01 3.08631718e-01 4.46532786e-01
3.58171940e-01 -8.13999951e-01 1.00775874e+00 2.95154303e-01
3.14939730e-02 -8.53222489e-01 -7.64448762e-01 -8.22850406e-01
-8.08270201e-02 2.03570165e-02 4.10024494e-01 4.62568402e-01
4.22815442e-01 5.82664013e-01 -5.50206602e-01 -2.98448861e-01
5.19562006e-01 -2.40230232e-01 1.01394188e+00 -1.26852942e+00
-7.44457096e-02 -1.98805526e-01 -2.44865030e-01 -1.13770008e+00
1.60504282e-01 -6.25176609e-01 3.89681518e-01 -1.65026712e+00
-1.47829756e-01 -7.52018213e-01 -4.09052849e-01 3.47187489e-01
-2.61103958e-01 -5.35609066e-01 2.15620160e-01 -5.47396541e-02
-8.31223547e-01 8.31645012e-01 1.68295026e+00 -4.69046198e-02
-7.25183249e-01 2.71204442e-01 -4.12898511e-01 9.06167984e-01
1.19353974e+00 -7.69408345e-01 -9.59516883e-01 9.81598124e-02
4.82536763e-01 4.07793254e-01 -3.22663873e-01 -1.47480702e+00
3.92930269e-01 -6.52702689e-01 6.81202188e-02 -4.55752999e-01
1.61034510e-01 -1.30198407e+00 -3.10358573e-02 1.39875150e+00
-2.90026158e-01 1.12099312e-01 -4.38386947e-02 7.19201148e-01
-9.03074518e-02 -6.79536581e-01 7.87193954e-01 -2.89750189e-01
-1.09531224e+00 1.27507389e-01 -8.18936110e-01 -1.18643820e-01
1.54204655e+00 -2.91639000e-01 -3.79082235e-03 -5.50712347e-01
-7.61489034e-01 7.43517935e-01 3.02515388e-01 5.11860609e-01
6.70589566e-01 -1.18276393e+00 -3.49809974e-01 1.86248049e-02
-2.04677492e-01 -7.16174766e-02 1.31007507e-01 3.83417010e-01
-7.25104630e-01 -1.34313419e-01 -9.63193476e-01 -3.89500260e-01
-1.22951829e+00 5.50899625e-01 4.28914219e-01 -4.53423381e-01
-4.07838851e-01 3.63393366e-01 -4.49207332e-03 -6.88410640e-01
4.72906083e-01 -8.99432525e-02 -1.04332197e+00 -7.12828189e-02
3.77711922e-01 4.41757351e-01 -6.06304646e-01 -2.79321223e-01
-3.47964838e-02 4.87474978e-01 -2.98612230e-02 -3.84283930e-01
1.30435252e+00 -1.00956008e-01 -2.17469577e-02 4.45147216e-01
5.82639694e-01 -3.85452837e-01 -1.46343052e+00 2.53075641e-03
1.47206336e-01 -3.33127052e-01 -3.27832282e-01 -7.54018307e-01
-7.32304335e-01 3.15942913e-01 8.64164948e-01 3.75964344e-01
9.21852767e-01 -5.97801328e-01 7.67768502e-01 1.20438004e+00
1.15763187e+00 -1.71659136e+00 4.12407100e-01 1.13907325e+00
7.83985734e-01 -1.43936491e+00 -1.45084336e-01 -1.54241860e-01
-9.56556022e-01 1.12731326e+00 1.19430852e+00 -4.62724894e-01
7.68490076e-01 1.07198037e-01 1.64819807e-01 4.44701277e-02
-5.57408690e-01 -2.58088589e-01 -4.96657044e-01 8.46771240e-01
-2.43165232e-02 3.38150896e-02 -9.47901070e-01 4.87862200e-01
-3.71633977e-01 1.43796712e-01 7.50791013e-01 1.17516267e+00
-1.32524991e+00 -1.43749905e+00 -3.28432024e-01 3.47746551e-01
8.91720504e-02 3.80748034e-01 6.71037957e-02 7.00508475e-01
3.00173551e-01 1.14812720e+00 -2.40727752e-01 -6.20782256e-01
4.49900389e-01 -3.03264797e-01 3.68136615e-01 -5.56423962e-01
-4.84798014e-01 -2.38752142e-01 -1.22176401e-01 -7.91168571e-01
-4.00880009e-01 -6.61780357e-01 -2.17614532e+00 -7.18097249e-03
4.35183290e-03 8.15591455e-01 4.84039962e-01 8.84064436e-01
-2.12225132e-02 5.33677042e-01 7.62733996e-01 -6.53245986e-01
-6.96959972e-01 -8.92991960e-01 -6.49706006e-01 2.42272347e-01
5.10069542e-02 -1.02825391e+00 -1.86274007e-01 -2.56061643e-01] | [4.022695064544678, 1.727403998374939] |
e85e3169-4526-478a-9264-a7f4dfb42623 | glyphdraw-learning-to-draw-chinese-characters | 2303.17870 | null | https://arxiv.org/abs/2303.17870v2 | https://arxiv.org/pdf/2303.17870v2.pdf | GlyphDraw: Seamlessly Rendering Text with Intricate Spatial Structures in Text-to-Image Generation | Recent breakthroughs in the field of language-guided image generation have yielded impressive achievements, enabling the creation of high-quality and diverse images based on user instructions.Although the synthesis performance is fascinating, one significant limitation of current image generation models is their insufficient ability to generate text coherently within images, particularly for complex glyph structures like Chinese characters. To address this problem, we introduce GlyphDraw, a general learning framework aiming to endow image generation models with the capacity to generate images coherently embedded with text for any specific language.We first sophisticatedly design the image-text dataset's construction strategy, then build our model specifically on a diffusion-based image generator and carefully modify the network structure to allow the model to learn drawing language characters with the help of glyph and position information.Furthermore, we maintain the model's open-domain image synthesis capability by preventing catastrophic forgetting by using parameter-efficient fine-tuning techniques.Extensive qualitative and quantitative experiments demonstrate that our method not only produces accurate language characters as in prompts, but also seamlessly blends the generated text into the background.Please refer to our \href{https://1073521013.github.io/glyph-draw.github.io/}{project page}. \end{abstract} | ['Xiaodong Lin', 'Haonan Lu', 'Di Niu', 'Ruichen Wang', 'Chen Chen', 'Mingjun Zhao', 'Jian Ma'] | 2023-03-31 | null | null | null | null | ['optical-character-recognition'] | ['computer-vision'] | [ 4.22108680e-01 1.78995192e-01 -1.99053641e-02 -1.44910336e-01
-6.18482590e-01 -5.82988679e-01 7.32354224e-01 -3.05825144e-01
-7.53135756e-02 7.21462548e-01 3.59544069e-01 -4.03212339e-01
2.48643219e-01 -9.49447393e-01 -7.77900338e-01 -5.25038183e-01
3.91242176e-01 1.38954848e-01 -3.37480903e-02 -2.72584498e-01
3.64465714e-01 2.78131664e-01 -1.50621819e+00 4.83858645e-01
1.22071970e+00 4.32774246e-01 6.85854435e-01 8.13900054e-01
-3.69014412e-01 9.96912956e-01 -7.93235779e-01 -4.95058149e-01
1.49412304e-01 -8.24627995e-01 -6.67551637e-01 3.22230667e-01
4.84865904e-01 -5.51441014e-01 -3.98619413e-01 9.34452236e-01
6.35062873e-01 -1.03882663e-01 5.42390645e-01 -1.18419111e+00
-1.35327899e+00 7.51753092e-01 -5.28222024e-01 -3.05366945e-02
3.84014785e-01 6.64652824e-01 7.78226495e-01 -1.00237107e+00
9.25968349e-01 1.03637469e+00 2.29093611e-01 8.87269378e-01
-1.28182054e+00 -7.98167706e-01 3.70618671e-01 -1.54477909e-01
-1.24745512e+00 -4.54357624e-01 7.31724381e-01 -4.38668132e-01
6.57241821e-01 3.74229640e-01 7.58929968e-01 1.47549033e+00
1.28068686e-01 9.37150776e-01 1.09809959e+00 -5.26092589e-01
-5.28256446e-02 2.42556095e-01 -4.22374278e-01 7.72453547e-01
8.39713067e-02 9.02615190e-02 -6.18417561e-01 2.28071481e-01
1.25686789e+00 -2.15524450e-01 -3.33620489e-01 -1.35005102e-01
-1.54140985e+00 8.11869800e-01 3.76773596e-01 2.45129138e-01
-2.83154547e-01 2.78741509e-01 1.22486881e-03 1.68931931e-01
4.77999330e-01 6.12494290e-01 -6.44923225e-02 3.45456749e-02
-1.04924762e+00 5.00878513e-01 5.34498572e-01 1.28935504e+00
8.22627008e-01 3.52320254e-01 -5.91769636e-01 7.98602223e-01
-3.15104276e-02 6.43481851e-01 4.46716249e-01 -1.07196844e+00
4.14614350e-01 3.81571710e-01 7.78450891e-02 -9.49791849e-01
-1.23769343e-02 -3.61186981e-01 -1.04436755e+00 2.86468565e-01
3.05751801e-01 -3.48533988e-01 -9.43659723e-01 1.78897774e+00
-6.09831437e-02 -1.72710985e-01 -1.81266770e-01 6.87088788e-01
1.02858043e+00 9.71362352e-01 8.59704539e-02 7.21054822e-02
1.27764070e+00 -1.18384016e+00 -6.61724806e-01 -2.50710160e-01
3.97132933e-01 -8.64087343e-01 1.70476365e+00 2.55860955e-01
-1.28270733e+00 -7.29353964e-01 -8.63453090e-01 -8.54938701e-02
-3.07484806e-01 2.26168111e-01 4.98587310e-01 4.71075743e-01
-1.35744166e+00 3.57517391e-01 -3.02848905e-01 -2.19032407e-01
5.65946341e-01 -4.28981632e-02 -8.86915401e-02 -3.39545198e-02
-1.11087704e+00 5.41592479e-01 3.82948220e-01 -3.12332064e-01
-8.70802343e-01 -8.44995618e-01 -7.57018209e-01 -1.74359888e-01
2.53960699e-01 -1.13187385e+00 1.23228836e+00 -1.33570158e+00
-1.57903624e+00 8.19359779e-01 -6.41208664e-02 -1.46647558e-01
8.02838266e-01 -1.21022381e-01 -1.59645543e-01 2.50483513e-01
1.96361259e-01 1.47771370e+00 1.01425099e+00 -1.61068785e+00
-5.35398960e-01 2.54751086e-01 -8.47345293e-02 3.26578230e-01
-4.91468757e-01 -1.32890448e-01 -6.66712880e-01 -1.15187573e+00
-3.96497190e-01 -6.92878723e-01 -3.22221160e-01 1.27285600e-01
-6.24833226e-01 8.05663690e-02 5.85474968e-01 -6.58682168e-01
1.37556398e+00 -2.09748197e+00 -9.07119513e-02 -3.81465219e-02
2.29067311e-01 1.15136430e-01 -5.66073418e-01 6.66182756e-01
-1.45487189e-01 3.81000817e-01 -2.16578782e-01 -3.73146355e-01
-7.26847276e-02 -8.05850849e-02 -5.13104856e-01 -1.38053492e-01
2.61677861e-01 1.31534791e+00 -9.54980552e-01 -4.40896958e-01
2.73607522e-01 4.89161581e-01 -6.90993786e-01 2.39390075e-01
-5.10949194e-01 5.76559722e-01 -2.74864286e-01 4.86452311e-01
5.07759452e-01 -5.57208300e-01 -5.12341820e-02 -4.99086455e-02
-1.45304158e-01 -8.36124122e-02 -9.08643782e-01 1.82061327e+00
-5.25495708e-01 6.24875069e-01 -2.32293218e-01 -3.92247617e-01
9.54384089e-01 1.05459183e-01 2.38721877e-01 -9.37683105e-01
-1.75377965e-01 1.77026931e-02 -3.36611748e-01 -3.86778980e-01
7.14519858e-01 1.24426611e-01 2.81518063e-04 7.04153359e-01
-2.24132821e-01 -4.82067049e-01 4.24266368e-01 5.66187143e-01
6.42308831e-01 3.51053506e-01 -1.07424341e-01 -1.56550959e-01
3.57373744e-01 2.15057656e-03 7.28000253e-02 9.89214063e-01
3.02276433e-01 9.67422962e-01 4.50118452e-01 -3.22681308e-01
-1.42741704e+00 -1.06325960e+00 1.75225183e-01 1.10533345e+00
4.80637811e-02 -4.33817416e-01 -1.02875698e+00 -4.17205572e-01
-3.89608860e-01 9.98080254e-01 -5.79394817e-01 -1.15773395e-01
-4.39912945e-01 -6.99923277e-01 5.46245277e-01 2.60438025e-01
7.60782659e-01 -1.56346130e+00 -4.64484245e-01 4.97174896e-02
-5.50211072e-01 -8.68753374e-01 -8.09039474e-01 -2.78004438e-01
-6.89949870e-01 -5.91044605e-01 -1.23803997e+00 -1.02768195e+00
1.12215161e+00 3.71135592e-01 1.29545069e+00 4.26077664e-01
-4.23669904e-01 3.88882995e-01 -3.24755609e-01 -1.72737062e-01
-6.93056047e-01 2.42974982e-01 -4.15899366e-01 -7.28522018e-02
-1.88478917e-01 -4.90815639e-01 -8.80186796e-01 9.48536545e-02
-1.33541763e+00 9.11976576e-01 5.74320853e-01 7.82863557e-01
4.53007549e-01 1.97668429e-02 5.06029010e-01 -9.61271346e-01
9.70245957e-01 -1.41995549e-01 -5.73446333e-01 2.62704730e-01
-5.69618165e-01 2.68862769e-02 5.68293333e-01 -5.09305596e-01
-1.12591481e+00 -9.17334948e-03 -1.66364312e-01 -1.57359734e-01
-2.05933869e-01 2.54105896e-01 -2.33144965e-02 2.53432274e-01
6.93268836e-01 7.54995286e-01 6.81815520e-02 -1.31152794e-01
8.22995067e-01 4.13565338e-01 6.14071548e-01 -7.35480607e-01
8.69529128e-01 3.84869426e-01 -6.12827957e-01 -6.84660077e-01
-5.30592740e-01 3.15634310e-01 -4.15796161e-01 -3.20182621e-01
8.51406515e-01 -1.10006893e+00 -4.68153834e-01 7.54804611e-01
-1.18761551e+00 -9.87430155e-01 -4.93851572e-01 -1.15014791e-01
-6.63920224e-01 3.15840244e-01 -7.35024273e-01 -4.39045012e-01
-4.19500679e-01 -1.12579143e+00 9.13609624e-01 3.16292584e-01
-3.43882203e-01 -9.10189927e-01 -1.53628424e-01 2.23630682e-01
5.43331444e-01 2.45478496e-01 1.01493466e+00 -5.47313988e-02
-9.30242538e-01 8.61896425e-02 -3.99011463e-01 2.78177589e-01
3.07608426e-01 2.74614602e-01 -6.73129916e-01 -1.93328500e-01
-3.99640679e-01 -2.40952253e-01 7.77923763e-01 2.99444795e-01
1.40433478e+00 -5.45493603e-01 -1.35012135e-01 7.31126487e-01
1.34338808e+00 6.75941408e-02 8.98941040e-01 4.24353361e-01
7.50823379e-01 6.28686726e-01 1.96103632e-01 5.92854023e-01
4.40037400e-01 4.00833696e-01 1.99571222e-01 -4.65030700e-01
-6.11207247e-01 -8.74296248e-01 3.04997444e-01 6.12950385e-01
1.92405224e-01 -5.85363686e-01 -7.12683320e-01 4.99046266e-01
-1.70873153e+00 -9.88140523e-01 -8.56928006e-02 1.94683313e+00
1.17939138e+00 6.00532554e-02 -1.24786571e-02 -2.82746673e-01
6.73136353e-01 3.25154483e-01 -3.99435043e-01 -2.16980889e-01
-3.48449171e-01 8.84159952e-02 4.31633294e-01 6.39614820e-01
-6.66639507e-01 1.29831576e+00 6.54090166e+00 9.66382623e-01
-1.20519149e+00 -1.31115034e-01 1.07889748e+00 -1.76662609e-01
-8.85437608e-01 -1.07659400e-01 -7.23139942e-01 6.70449018e-01
5.13824463e-01 -2.66927600e-01 6.66204453e-01 4.99686897e-01
5.13066888e-01 -1.59840003e-01 -6.79913402e-01 9.28608894e-01
1.46650136e-01 -1.78055215e+00 6.91749096e-01 -1.51487768e-01
9.83374059e-01 -4.53513980e-01 5.10317147e-01 2.66625378e-02
6.25009894e-01 -1.07538497e+00 1.01261365e+00 6.07838452e-01
1.19480145e+00 -5.98276556e-01 7.94067327e-03 2.75802672e-01
-7.98528910e-01 1.01609468e-01 -1.91337258e-01 5.12839071e-02
2.93416023e-01 6.57034576e-01 -6.76187098e-01 1.75499871e-01
5.35780370e-01 6.59486771e-01 -8.83942485e-01 7.13333011e-01
-5.45132637e-01 3.35066319e-01 4.97979447e-02 3.67750064e-03
8.79106522e-02 -2.15642080e-01 2.73740977e-01 1.25368857e+00
5.33749819e-01 2.78971996e-02 -5.35680503e-02 1.36044490e+00
-1.24736704e-01 1.90179840e-01 -8.46166134e-01 -1.91952601e-01
4.50846046e-01 1.23621380e+00 -8.56635213e-01 -5.25482595e-01
-1.97329625e-01 1.29218352e+00 1.99267447e-01 6.64670706e-01
-9.37905967e-01 -4.78520870e-01 3.14374089e-01 3.03147346e-01
2.08860427e-01 -3.39308411e-01 -5.55098891e-01 -1.11316490e+00
-1.05281942e-01 -1.23273480e+00 -6.89946264e-02 -1.33067822e+00
-1.07230926e+00 8.64962041e-01 -2.98724417e-02 -1.02025437e+00
-1.39351502e-01 -2.93384224e-01 -5.96265793e-01 8.88365984e-01
-1.25052798e+00 -1.31398714e+00 -5.12187302e-01 7.51387417e-01
8.02509844e-01 -2.73253411e-01 5.89838624e-01 1.69632927e-01
-5.14042199e-01 6.71427786e-01 -1.25254914e-01 9.14844424e-02
8.16363394e-01 -1.05008149e+00 7.45936692e-01 8.70942056e-01
9.76080447e-02 5.76547801e-01 6.01083517e-01 -9.56833541e-01
-1.07596719e+00 -9.73429203e-01 8.57244551e-01 -4.66052771e-01
5.24739265e-01 -6.44514084e-01 -6.97558403e-01 4.83280361e-01
7.11303949e-01 -3.82315934e-01 4.00077462e-01 -4.41978067e-01
-3.37478697e-01 1.07772328e-01 -7.58327007e-01 1.27551734e+00
1.15984559e+00 -3.26579899e-01 -4.98146154e-02 4.71322209e-01
9.46937561e-01 -5.03877759e-01 -4.65754300e-01 7.40067959e-02
3.09206843e-01 -1.11368024e+00 9.83018994e-01 -1.70779541e-01
9.74894404e-01 -3.26009333e-01 2.95019299e-01 -1.20175719e+00
-5.30894339e-01 -1.01896787e+00 2.19467580e-01 1.35400164e+00
5.85229218e-01 -4.92752999e-01 7.84380913e-01 4.94479269e-01
3.24533097e-02 -4.99839336e-01 -2.31353149e-01 -4.43563104e-01
1.68711498e-01 -2.36918047e-01 7.19272196e-01 6.68576360e-01
-2.38026455e-01 2.60693252e-01 -6.10757351e-01 -2.39561245e-01
3.75927269e-01 7.03412071e-02 8.92405808e-01 -5.52522898e-01
-3.25391293e-01 -7.80728877e-01 4.32362616e-01 -1.28641653e+00
-1.45081952e-01 -7.52811909e-01 4.59274873e-02 -1.69393289e+00
2.86121428e-01 -5.20095766e-01 2.02658042e-01 4.58916008e-01
-3.48809838e-01 5.19062161e-01 5.69205701e-01 2.89987177e-01
-4.05368239e-01 4.94053751e-01 1.87032461e+00 -3.27468403e-02
-1.41778141e-01 -1.60357296e-01 -1.14782190e+00 2.72501707e-01
8.97823274e-01 -2.41440251e-01 -6.86167121e-01 -6.74332201e-01
5.32696128e-01 2.40721623e-03 5.51457286e-01 -8.13933492e-01
6.24803640e-02 -3.23276907e-01 5.60056567e-01 -4.63438988e-01
5.76229505e-02 -3.35094988e-01 3.26111913e-01 4.44593698e-01
-6.10621870e-01 3.67644995e-01 2.51782417e-01 2.67623514e-01
-2.89409850e-02 -1.63205817e-01 6.94138169e-01 -5.10582030e-01
-6.47286952e-01 3.32112283e-01 -6.91325903e-01 6.70535862e-02
9.10820782e-01 -3.73742104e-01 -4.03296709e-01 -7.41298616e-01
-5.54553807e-01 9.94670242e-02 8.18276048e-01 6.79356277e-01
6.70066833e-01 -1.40178239e+00 -8.53574395e-01 2.71625906e-01
8.99283737e-02 1.07924215e-01 4.09550786e-01 4.05440718e-01
-7.72021115e-01 2.29818910e-01 -2.86718339e-01 -3.62070709e-01
-8.62872601e-01 6.24358535e-01 1.83945760e-01 -3.45258355e-01
-6.46610677e-01 7.44816184e-01 5.82991898e-01 -3.20908666e-01
1.22734964e-01 -1.10606275e-01 1.31955221e-01 -8.61034691e-02
6.80286467e-01 -3.74184363e-02 -3.72532547e-01 -2.07984701e-01
1.97604820e-01 3.91021520e-01 -1.78691655e-01 -4.12007362e-01
1.16726315e+00 -4.11939651e-01 -3.95232476e-02 5.60793206e-02
8.18996608e-01 -2.21798103e-02 -1.55559456e+00 -8.00796598e-02
-4.44362342e-01 -4.73109901e-01 -2.85856396e-01 -1.01524901e+00
-1.15983903e+00 7.72168517e-01 3.46450686e-01 6.06940687e-02
1.18376887e+00 -1.49225041e-01 8.12763870e-01 -2.23848261e-02
9.91545096e-02 -1.04315913e+00 8.37365389e-01 5.30418813e-01
1.06331360e+00 -1.05112195e+00 -6.93818405e-02 -2.92685717e-01
-9.81023073e-01 1.04115677e+00 7.18069971e-01 -5.99423386e-02
2.81145900e-01 2.90829390e-01 3.32557052e-01 -7.80732706e-02
-6.74987137e-01 3.52748185e-02 1.02801181e-01 8.58441353e-01
5.60276687e-01 -1.27160862e-01 -2.04380244e-01 1.49473876e-01
-4.47584987e-01 1.41550094e-01 6.80392623e-01 7.29542255e-01
-2.66708195e-01 -1.27370584e+00 -2.32802227e-01 2.27203056e-01
-1.44299150e-01 -5.21302938e-01 -3.51559132e-01 6.00500047e-01
8.57873634e-02 7.24262059e-01 8.49702209e-02 -1.31694272e-01
8.34525451e-02 -1.37372747e-01 4.27147001e-01 -5.76723695e-01
-5.57833910e-01 7.39058480e-02 -1.62472382e-01 -2.99293339e-01
-1.50301442e-01 -4.01692241e-01 -9.42395031e-01 -5.34223855e-01
1.62222281e-01 -2.12172449e-01 4.80599701e-01 4.98857290e-01
5.51095188e-01 6.62218809e-01 6.47086143e-01 -8.21428537e-01
-9.91299525e-02 -6.81945264e-01 -2.73440301e-01 4.91495728e-01
1.08790286e-01 -4.66367863e-02 -1.00360870e-01 4.13379282e-01] | [11.394576072692871, -0.25990059971809387] |
7ce1d85f-870b-4a6a-a047-a90c90a4adc8 | keyphrase-generation-with-cross-document | 2004.09800 | null | https://arxiv.org/abs/2004.09800v2 | https://arxiv.org/pdf/2004.09800v2.pdf | Keyphrase Generation with Cross-Document Attention | Keyphrase generation aims to produce a set of phrases summarizing the essentials of a given document. Conventional methods normally apply an encoder-decoder architecture to generate the output keyphrases for an input document, where they are designed to focus on each current document so they inevitably omit crucial corpus-level information carried by other similar documents, i.e., the cross-document dependency and latent topics. In this paper, we propose CDKGen, a Transformer-based keyphrase generator, which expands the Transformer to global attention with cross-document attention networks to incorporate available documents as references so as to generate better keyphrases with the guidance of topic information. On top of the proposed Transformer + cross-document attention architecture, we also adopt a copy mechanism to enhance our model via selecting appropriate words from documents to deal with out-of-vocabulary words in keyphrases. Experiment results on five benchmark datasets illustrate the validity and effectiveness of our model, which achieves the state-of-the-art performance on all datasets. Further analyses confirm that the proposed model is able to generate keyphrases consistent with references while keeping sufficient diversity. The code of CDKGen is available at https://github.com/SVAIGBA/CDKGen. | ['Yan Song', 'Shizhe Diao', 'Tong Zhang'] | 2020-04-21 | null | null | null | null | ['keyphrase-generation'] | ['natural-language-processing'] | [-9.70185734e-03 7.87281338e-03 -2.95131236e-01 1.02432452e-01
-1.06630695e+00 -5.67924500e-01 9.89282966e-01 1.15414545e-01
-1.39230996e-01 8.12406659e-01 8.72675419e-01 -2.42491841e-01
4.57666814e-03 -7.92679429e-01 -7.22643971e-01 -6.15816653e-01
4.91497934e-01 3.33779454e-01 1.69644549e-01 -4.42460716e-01
6.30015790e-01 3.85037661e-02 -1.16871357e+00 4.39383775e-01
1.00677347e+00 5.41256368e-01 5.89271545e-01 5.71053386e-01
-6.31635070e-01 6.92193747e-01 -8.84657502e-01 -5.22925615e-01
-2.99259182e-02 -5.86799026e-01 -5.65539479e-01 -5.87646924e-02
4.00039822e-01 -6.06965840e-01 -6.76921844e-01 8.56871784e-01
5.19181073e-01 -7.57566318e-02 5.55375516e-01 -1.22827089e+00
-1.28118289e+00 1.19255495e+00 -4.85534877e-01 3.61955106e-01
1.57590136e-01 -1.93108469e-02 1.38294744e+00 -1.14592826e+00
4.39027995e-01 1.25359845e+00 1.09617338e-01 4.48227465e-01
-8.63542438e-01 -9.70082581e-01 4.76562083e-01 1.66468635e-01
-1.42731845e+00 -3.08000326e-01 9.38637555e-01 -1.08882725e-01
9.19719100e-01 1.51141346e-01 6.63256645e-01 1.20690358e+00
5.52319944e-01 1.11059701e+00 5.61660349e-01 -2.89091140e-01
-1.92205817e-01 1.35765448e-01 2.68406540e-01 4.28497821e-01
3.05332243e-01 -2.47364268e-01 -5.93936980e-01 -9.91548970e-02
6.72091186e-01 1.42869696e-01 -4.94111568e-01 8.77683535e-02
-1.47033536e+00 8.89242828e-01 2.74721444e-01 4.01183426e-01
-6.70804918e-01 1.61146224e-01 4.25242931e-01 1.80083211e-04
5.20024419e-01 5.29130161e-01 -3.31619114e-01 3.97526994e-02
-9.21689451e-01 5.93405426e-01 5.99480391e-01 1.37436903e+00
5.59038341e-01 -3.68586741e-02 -9.00858581e-01 6.68696880e-01
1.50200769e-01 5.53268075e-01 7.99035788e-01 -3.28473151e-01
8.48212361e-01 7.11097777e-01 1.56826705e-01 -1.07159030e+00
8.24744403e-02 -6.57607853e-01 -9.38018739e-01 -7.18765378e-01
-4.16272968e-01 -1.49942413e-01 -1.13231504e+00 1.49332070e+00
4.94883880e-02 2.12997198e-02 7.57983625e-02 5.27870178e-01
1.06853759e+00 1.39498460e+00 -9.51448455e-02 -1.89254373e-01
1.38758755e+00 -1.28501785e+00 -1.02044797e+00 -2.44253427e-01
1.70793846e-01 -7.97210872e-01 1.18890822e+00 1.28984779e-01
-9.98137832e-01 -6.76322460e-01 -9.13140595e-01 -3.42227578e-01
-4.42909926e-01 4.05009180e-01 2.54335374e-01 1.44120120e-02
-9.52325225e-01 8.88766721e-02 -4.65917528e-01 -8.54742974e-02
1.88964948e-01 5.78743964e-02 3.75727713e-02 1.47816449e-01
-1.59171522e+00 5.24282396e-01 8.03261459e-01 1.69975515e-02
-9.87245739e-01 -9.37866211e-01 -6.67242944e-01 2.99481064e-01
6.15053117e-01 -7.95437634e-01 1.34828997e+00 -3.59051675e-01
-1.22360325e+00 2.80195057e-01 -3.59408915e-01 -3.13183397e-01
1.22217350e-01 -4.56890613e-01 -4.28607762e-01 1.99912176e-01
4.24813986e-01 8.99840057e-01 9.33179736e-01 -1.16858351e+00
-8.76287103e-01 9.19810459e-02 1.18629433e-01 4.70873803e-01
-5.77882588e-01 -2.37236712e-02 -1.12811697e+00 -1.23475814e+00
-2.34676138e-01 -6.53744996e-01 6.86438149e-03 -6.17694139e-01
-8.95187080e-01 -3.73851448e-01 8.69355857e-01 -9.06014919e-01
1.85267007e+00 -1.81564105e+00 2.09497914e-01 -2.18122005e-02
2.11851314e-01 1.27111688e-01 -2.23666728e-01 9.65879738e-01
1.71031337e-02 3.98306817e-01 -1.52719155e-01 -1.31935790e-01
-2.34642979e-02 -9.76343751e-02 -1.04724073e+00 -2.12074846e-01
2.80729204e-01 1.17471635e+00 -7.97757566e-01 -4.63124126e-01
-6.37311786e-02 4.55539793e-01 -3.12233716e-01 2.64665097e-01
-5.70750296e-01 1.47067264e-01 -8.66890848e-01 4.26346302e-01
2.91169316e-01 -3.55153412e-01 -3.06081295e-01 -3.13258380e-01
-1.29993767e-01 6.79613531e-01 -9.56609964e-01 1.52790308e+00
-4.80604321e-01 4.18995231e-01 -2.82635093e-01 -4.82631415e-01
9.51360106e-01 4.72537816e-01 2.14940012e-01 -6.04027927e-01
7.42180347e-02 7.81020224e-02 -2.72044301e-01 -1.74623728e-02
1.05958664e+00 2.62811154e-01 -2.35890642e-01 7.40824282e-01
1.17894925e-01 -2.22105205e-01 5.38983822e-01 8.42402577e-01
7.34081388e-01 -1.73330054e-01 2.51981616e-01 -3.14936310e-01
6.61034942e-01 -1.38567403e-01 1.55678317e-01 8.31217170e-01
5.96377492e-01 4.97508973e-01 5.28060853e-01 -9.48702246e-02
-1.15011513e+00 -6.00956917e-01 2.47825190e-01 9.46108401e-01
2.34045401e-01 -8.86594176e-01 -6.62568450e-01 -6.49485826e-01
-1.73775092e-01 9.54195440e-01 -6.45182431e-01 -2.91437894e-01
-5.60355365e-01 -5.85996926e-01 3.80487084e-01 4.05932516e-01
4.27359343e-01 -1.27457368e+00 -2.18223393e-01 4.68656063e-01
-4.44241762e-01 -8.59020412e-01 -1.08792973e+00 -5.36774546e-02
-3.84711266e-01 -7.80216038e-01 -1.26554644e+00 -7.84004748e-01
7.14906216e-01 5.18074453e-01 1.06225073e+00 8.63485560e-02
1.73162147e-01 3.20501268e-01 -6.15523398e-01 -4.20633078e-01
-3.59383702e-01 5.27092338e-01 -2.16423690e-01 -2.59680761e-04
2.75194973e-01 -2.86266834e-01 -5.24522722e-01 -9.87697765e-02
-1.13655972e+00 4.54854816e-01 8.53293359e-01 9.94951367e-01
5.78520536e-01 3.49284858e-01 5.20484865e-01 -7.24321544e-01
1.39121425e+00 -6.67457581e-01 -4.89388406e-01 3.47135007e-01
-7.90976822e-01 2.69522309e-01 8.01639616e-01 -4.68049914e-01
-1.07698667e+00 -6.54330015e-01 3.85232754e-02 -3.24867964e-01
3.45294476e-01 7.04512775e-01 -2.20631793e-01 6.82123244e-01
9.14078131e-02 9.39953685e-01 -5.49450278e-01 -5.20487964e-01
5.12894928e-01 7.43199229e-01 3.99250954e-01 -6.48638725e-01
8.63716960e-01 9.28799137e-02 -5.01897693e-01 -5.02837956e-01
-8.92858386e-01 -4.15403396e-01 -2.77420223e-01 -1.19555891e-02
5.82862854e-01 -1.13361418e+00 -1.57023400e-01 5.10560036e-01
-1.37072802e+00 -2.23175939e-02 -1.87892959e-01 1.40864581e-01
-4.47518975e-02 2.94492930e-01 -5.64152181e-01 -5.42231679e-01
-1.06065726e+00 -1.10624373e+00 1.30510104e+00 4.14055765e-01
-1.17217377e-01 -7.38886058e-01 -5.88534698e-02 -1.38957417e-02
2.61685759e-01 -2.80583948e-01 1.04933000e+00 -8.71971488e-01
-7.03781188e-01 -1.46079436e-01 -2.15047464e-01 2.45691553e-01
4.36658084e-01 1.05389901e-01 -4.98145044e-01 -2.47916028e-01
-2.64242858e-01 -8.78881291e-02 1.11045194e+00 1.31972685e-01
1.24599695e+00 -7.86200881e-01 -4.13491189e-01 2.86158711e-01
1.25298250e+00 5.54988444e-01 5.10959268e-01 3.01458716e-01
8.97996008e-01 4.53456998e-01 5.25364339e-01 3.76573652e-01
6.58679247e-01 5.16847312e-01 4.55533825e-02 5.84166721e-02
-1.35479450e-01 -8.01206052e-01 3.74820411e-01 1.21529841e+00
4.34219658e-01 -8.37252915e-01 -7.96034276e-01 8.41198862e-01
-1.66270614e+00 -9.60839272e-01 1.20259158e-03 1.75862563e+00
1.13815546e+00 3.56919438e-01 -2.59800762e-01 -5.63385189e-02
7.71830797e-01 4.23245370e-01 -4.23544586e-01 -1.16501696e-01
-6.28804490e-02 1.77042440e-01 3.41437668e-01 5.02554953e-01
-8.03007066e-01 1.23692310e+00 5.13167524e+00 1.02434456e+00
-1.07240009e+00 -1.84795365e-01 6.11005425e-01 7.36593083e-02
-8.16425562e-01 -9.74359363e-02 -1.32132876e+00 8.14306974e-01
7.70508528e-01 -8.16750228e-01 2.28809610e-01 6.56922758e-01
1.96769476e-01 1.63329080e-01 -7.45205283e-01 6.95439041e-01
2.66010612e-01 -1.51651239e+00 8.16000760e-01 -1.11432940e-01
8.28412235e-01 -2.86075681e-01 1.12681948e-01 4.29603100e-01
3.67009729e-01 -6.90586746e-01 8.55276525e-01 4.93574113e-01
5.86282611e-01 -8.49713981e-01 5.99074125e-01 2.61431217e-01
-1.25560904e+00 4.51326780e-02 -3.00020039e-01 3.29041868e-01
1.83903843e-01 5.81080377e-01 -9.57663000e-01 6.49272025e-01
5.65190911e-01 7.00622201e-01 -5.41196704e-01 7.00263739e-01
-5.63029468e-01 5.33226192e-01 -5.14426157e-02 -3.50291759e-01
6.35975778e-01 -1.08845197e-01 5.25730729e-01 1.37290502e+00
6.56073868e-01 1.29754841e-01 3.91766131e-02 9.89615142e-01
-2.68153220e-01 4.11051363e-01 -3.98437977e-01 -4.75161433e-01
8.33208084e-01 1.25453150e+00 -6.71076536e-01 -7.10647225e-01
-2.88389802e-01 1.05782616e+00 -3.73142608e-03 4.94484514e-01
-6.95928931e-01 -7.41477966e-01 4.21697348e-01 -7.01051205e-02
6.46730840e-01 -8.14362913e-02 4.63628694e-02 -1.14333320e+00
2.96151161e-01 -1.09474218e+00 1.90830857e-01 -1.03866100e+00
-1.09828281e+00 6.87152207e-01 2.76359707e-01 -1.15241194e+00
-2.49219880e-01 -1.85933262e-01 -7.99894691e-01 1.11858380e+00
-1.41825521e+00 -1.29037583e+00 -1.64921790e-01 4.99043494e-01
1.02915752e+00 -2.34239370e-01 5.19305706e-01 -8.96424614e-03
-4.30788070e-01 4.25268561e-01 1.83496609e-01 2.25810930e-01
6.66636944e-01 -1.19134212e+00 9.71368074e-01 1.03046548e+00
2.24614918e-01 1.06858134e+00 5.38551629e-01 -9.74107265e-01
-1.40519273e+00 -1.17643666e+00 1.10641265e+00 -2.32846573e-01
8.25517297e-01 -5.18020928e-01 -9.46036518e-01 6.70158684e-01
6.89332306e-01 -6.03976548e-01 4.38592970e-01 -3.14244837e-01
-2.23241761e-01 -2.06834257e-01 -3.62280250e-01 1.04806948e+00
6.62109017e-01 -5.32801688e-01 -8.84104848e-01 2.28090897e-01
1.41959751e+00 -4.94586349e-01 -4.81237024e-01 2.01118588e-01
3.68811131e-01 -4.81035769e-01 7.13980675e-01 -3.07212442e-01
8.20988655e-01 -3.44616950e-01 1.73601672e-01 -1.54983556e+00
-4.68058765e-01 -8.14442158e-01 -4.76586848e-01 1.68664038e+00
4.91763353e-01 -4.03488517e-01 3.88847232e-01 1.54637247e-01
-3.06848466e-01 -7.88220048e-01 -5.51381052e-01 -4.37806219e-01
8.07276170e-04 -1.99782804e-01 9.20863807e-01 7.43144810e-01
-1.96068987e-01 8.80271018e-01 -6.13329768e-01 2.77499910e-02
2.55826622e-01 2.40050733e-01 7.81821907e-01 -7.49118149e-01
-7.28468895e-02 -5.79946935e-01 3.38090360e-01 -1.39667261e+00
1.31218329e-01 -8.63161504e-01 -1.78341325e-02 -1.93995059e+00
3.21951032e-01 -1.26896873e-01 -3.34132642e-01 5.05034566e-01
-7.08304107e-01 -1.22723930e-01 2.08373204e-01 2.66445279e-01
-3.74943882e-01 9.46842253e-01 1.57091367e+00 -5.26334286e-01
-2.65482426e-01 -1.01873815e-01 -1.44234550e+00 1.88660547e-01
7.78311551e-01 -4.77243900e-01 -8.14275384e-01 -5.20442903e-01
2.07671851e-01 7.84642342e-03 1.44692674e-01 -6.61724448e-01
3.75183165e-01 -1.77848995e-01 3.28942120e-01 -9.77366328e-01
1.21256873e-01 -5.81959784e-01 -4.31216322e-02 3.70985657e-01
-5.99064112e-01 4.71802622e-01 3.71798277e-01 4.35021520e-01
-3.19214612e-01 -1.80372864e-01 6.14856854e-02 -2.58701503e-01
-6.78517759e-01 5.29908717e-01 -3.61629337e-01 1.96046934e-01
7.55944014e-01 1.96319968e-01 -5.12675405e-01 -6.00561857e-01
1.07432283e-01 4.53736275e-01 8.53788033e-02 8.77919853e-01
7.40521252e-01 -1.37532651e+00 -1.02084875e+00 1.71674058e-01
3.20949525e-01 1.17532626e-01 2.03125432e-01 2.94247150e-01
-2.86069900e-01 1.08331215e+00 1.04575410e-01 -6.29101396e-02
-9.86649930e-01 6.80427492e-01 -2.11306028e-02 -5.33673048e-01
-6.81786656e-01 8.83340478e-01 5.73326111e-01 1.27512654e-02
1.53503597e-01 -6.90643013e-01 -3.61860156e-01 2.59745866e-01
8.47847283e-01 -9.54097956e-02 -4.24082344e-03 -3.98051023e-01
1.36292968e-02 3.39131564e-01 -7.88190126e-01 -2.31427252e-01
1.04860723e+00 -1.98392108e-01 -2.60402977e-01 2.73454875e-01
9.96676028e-01 3.35381001e-01 -1.04804552e+00 -3.67898524e-01
-1.88628748e-01 -1.18029177e-01 2.46684372e-01 -8.35017443e-01
-9.91917193e-01 6.60723567e-01 -1.23403937e-01 9.72988084e-02
1.24555635e+00 7.91108534e-02 1.06036425e+00 3.17428887e-01
1.20960949e-02 -9.25562441e-01 2.08654031e-01 5.65521121e-01
1.32718956e+00 -8.36490452e-01 1.00953542e-01 -2.84403507e-02
-7.96208739e-01 9.54732358e-01 7.36215413e-01 9.72520411e-02
4.46594357e-01 9.33086723e-02 -9.41448659e-02 -1.57127514e-01
-1.11481059e+00 1.98793616e-02 5.20279944e-01 1.70323804e-01
3.28272700e-01 -2.33854666e-01 -4.91043866e-01 8.69661033e-01
-4.29430068e-01 -2.39644170e-01 6.27609074e-01 7.82869875e-01
-4.10916716e-01 -1.10171902e+00 -2.20745862e-01 5.46052754e-01
-5.94435751e-01 -7.55278230e-01 -5.77514052e-01 6.45386100e-01
-1.42197475e-01 6.67885602e-01 4.22243886e-02 -3.65950108e-01
1.52611420e-01 1.30678341e-01 -8.88006613e-02 -7.76100457e-01
-6.09848559e-01 3.88612419e-01 -1.64951280e-01 -1.00916564e-01
-5.09174690e-02 -3.15882653e-01 -1.03575611e+00 -2.48379037e-01
-2.71153033e-01 5.85344195e-01 3.51062387e-01 6.64729416e-01
6.25203192e-01 9.64348197e-01 6.12625778e-01 -5.51991105e-01
-3.36647719e-01 -1.32440960e+00 -3.26320946e-01 4.10723425e-02
4.39715803e-01 -2.11024374e-01 -4.23787422e-02 2.54204571e-01] | [12.335434913635254, 8.948945999145508] |
40b03131-31f5-4eca-a82f-f7dad842a0fa | small-noisy-and-perspective-face-detection | 2010.16164 | null | https://arxiv.org/abs/2010.16164v1 | https://arxiv.org/pdf/2010.16164v1.pdf | Small Noisy and Perspective Face Detection using Deformable Symmetric Gabor Wavelet Network | Face detection and tracking in low resolution image is not a trivial task due to the limitation in the appearance features for face characterization. Moreover, facial expression gives additional distortion on this small and noisy face. In this paper, we propose deformable symmetric Gabor wavelet network face model for face detection in low resolution image. Our model optimizes the rotation, translation, dilation, perspective and partial deformation amount of the face model with symmetry constraints. Symmetry constraints help our model to be more robust to noise and distortion. Experimental results on our low resolution face image dataset and videos show promising face detection and tracking results under various challenging conditions. | ['Seungkyu Lee', 'Sherzod Salokhiddinov'] | 2020-10-30 | null | null | null | null | ['face-model'] | ['computer-vision'] | [-8.43908861e-02 -2.96212763e-01 -1.32975042e-01 -4.12820429e-01
-7.09481835e-02 -4.99649853e-01 2.44639724e-01 -9.29569602e-01
-3.36658955e-01 5.91366768e-01 -1.82388350e-02 4.80786264e-01
-1.03548005e-01 -7.03920901e-01 -2.70802885e-01 -8.34730864e-01
-1.78045943e-01 6.34871200e-02 1.63008600e-01 -3.38920988e-02
-2.07663681e-02 1.25175273e+00 -1.51717508e+00 2.11354643e-01
-2.89577022e-02 9.31267083e-01 8.92603919e-02 4.73492980e-01
4.44251657e-01 2.44037449e-01 -2.83220261e-01 -8.05486590e-02
5.09120286e-01 4.15295102e-02 -4.63067770e-01 3.87775630e-01
5.40303528e-01 -6.92382991e-01 -3.77597302e-01 1.24387670e+00
8.41635227e-01 2.14013889e-01 5.55114031e-01 -1.01121318e+00
-8.83570790e-01 -6.20298125e-02 -9.67439473e-01 3.52358282e-01
3.91759455e-01 -2.46190116e-01 2.01522216e-01 -1.21537149e+00
8.22532237e-01 1.67271829e+00 7.97937274e-01 9.85058129e-01
-8.78727973e-01 -7.98291683e-01 -2.86881737e-02 -1.37869520e-02
-1.78547573e+00 -7.41398871e-01 7.66960382e-01 -1.29607320e-01
6.79774642e-01 2.29659274e-01 2.60044664e-01 8.89916122e-01
-2.13010311e-01 1.21509597e-01 1.00843155e+00 -3.82488579e-01
-3.01715285e-01 1.87291190e-01 -2.51486361e-01 1.12076437e+00
4.22592759e-01 1.04649581e-01 -1.68384209e-01 -7.97808841e-02
1.44377589e+00 -3.32927555e-02 -2.54075438e-01 -5.65582290e-02
-4.35037315e-01 5.79968154e-01 5.16771972e-02 3.03035080e-01
-2.14434594e-01 -1.70340449e-01 4.03678156e-02 1.64851040e-01
4.19628382e-01 -2.81359255e-01 -3.59155834e-01 2.55709112e-01
-6.07664943e-01 7.14935409e-03 3.57899189e-01 7.47670293e-01
2.89849073e-01 4.22226548e-01 -8.18721801e-02 9.69785035e-01
5.47764242e-01 6.59158707e-01 3.75550389e-01 -9.90909874e-01
-5.29931067e-03 4.46418494e-01 5.64767234e-02 -1.32540834e+00
-4.76115525e-01 -1.57934144e-01 -8.73682976e-01 5.16806245e-01
3.30391049e-01 -1.50031447e-01 -9.21684325e-01 1.40448630e+00
6.28580451e-01 1.19894855e-01 -1.41380891e-01 1.09523439e+00
1.11315966e+00 3.42923760e-01 -7.63583034e-02 -6.01172030e-01
1.51637948e+00 -4.85265076e-01 -9.49624777e-01 1.01144366e-01
-5.08515984e-02 -1.16400707e+00 6.50506914e-01 2.88499594e-01
-1.07797658e+00 -5.45555055e-01 -7.33915865e-01 1.18586700e-02
-1.70765016e-02 6.68023825e-01 2.42242649e-01 1.03134751e+00
-8.64443600e-01 4.00918275e-01 -7.22539485e-01 -2.98865914e-01
5.52148938e-01 7.70578325e-01 -7.79464245e-01 6.06560744e-02
-7.04387963e-01 7.80296862e-01 5.46072908e-02 4.43285465e-01
-4.95403737e-01 -1.99004248e-01 -7.54444301e-01 -1.94772691e-01
2.14962989e-01 -2.09939629e-01 6.47302210e-01 -5.99015653e-01
-1.60536230e+00 1.09014845e+00 -3.11837941e-01 5.21532781e-02
5.04664898e-01 1.06573515e-01 -4.38196778e-01 1.39455706e-01
-3.73997688e-01 4.09585744e-01 1.26118243e+00 -1.13984621e+00
-2.84817547e-01 -8.47336054e-01 -2.93569326e-01 7.81646296e-02
-3.81652892e-01 7.61795282e-01 -4.13811415e-01 -6.45968795e-01
2.90714771e-01 -7.70156682e-01 3.11552733e-01 3.11766446e-01
6.44023940e-02 -2.60208070e-01 1.65771210e+00 -8.36026728e-01
8.78247976e-01 -2.16974807e+00 -2.07408413e-01 2.36661926e-01
-1.90080911e-01 5.03615916e-01 -1.06264539e-01 -3.89398783e-01
-6.62923977e-02 2.70510375e-01 4.67370301e-01 -3.64578632e-03
-4.38930184e-01 1.62892878e-01 1.26372473e-02 9.18220162e-01
3.08575749e-01 5.41127801e-01 -1.67319402e-01 -7.22862542e-01
1.07017368e-01 1.31345439e+00 -3.42228025e-01 5.92458025e-02
4.54302460e-01 4.08454031e-01 -7.27289915e-01 1.19126165e+00
1.18582928e+00 3.55464786e-01 -2.18728766e-01 -5.98005414e-01
-9.53622907e-02 -5.43759644e-01 -1.61487520e+00 8.93089712e-01
-1.10677913e-01 4.38466400e-01 6.47029281e-01 -8.24195981e-01
1.15628517e+00 5.71279883e-01 3.42872858e-01 -4.82247800e-01
5.66015720e-01 -1.40021056e-01 -1.00488968e-01 -8.72268319e-01
4.81223129e-02 -1.24903902e-01 7.81085253e-01 2.63563305e-01
-4.94164750e-02 3.27255964e-01 5.64193204e-02 -5.75973630e-01
3.30487370e-01 2.51151413e-01 7.52225667e-02 -1.27323464e-01
8.39347601e-01 -9.26517308e-01 7.69905269e-01 1.81944057e-01
-4.16219532e-01 5.49311101e-01 1.46668479e-01 -6.86254144e-01
-1.01596761e+00 -7.63763070e-01 -6.18407011e-01 7.95177937e-01
-1.26389980e-01 2.04792380e-01 -8.95792007e-01 -4.48330045e-01
-1.38444915e-01 -3.44413489e-01 -6.18898928e-01 1.53992876e-01
-1.09086466e+00 -8.52554440e-01 6.44507170e-01 5.81879616e-01
6.95389390e-01 -1.05958664e+00 -4.02833849e-01 -3.18780541e-01
1.42399460e-01 -1.34736788e+00 -4.92299289e-01 -5.77936769e-01
-8.11501682e-01 -1.22255147e+00 -7.04763412e-01 -1.18143380e+00
1.11594057e+00 2.17417911e-01 4.10548240e-01 3.48395824e-01
-8.91177118e-01 2.01399758e-01 -3.77741233e-02 -3.31218243e-01
-5.91556132e-02 -6.00607693e-01 5.45948982e-01 1.19278088e-01
1.63773954e-01 -2.34334975e-01 -6.53291643e-01 6.86352313e-01
-5.43025553e-01 -6.80586278e-01 2.40331575e-01 8.39369714e-01
6.23761117e-01 6.00974500e-01 1.92981973e-01 -1.82249963e-01
4.72252369e-01 1.73326164e-01 -8.14644277e-01 3.11962396e-01
-2.39467129e-01 -2.37071559e-01 3.85750115e-01 -7.70140231e-01
-1.30131006e+00 3.49367738e-01 -5.68614677e-02 -5.63722730e-01
-2.07172126e-01 -2.19316378e-01 -1.92247257e-01 -9.41912889e-01
5.21076679e-01 1.29578277e-01 3.18339646e-01 -5.96903324e-01
-9.52165201e-02 7.28487670e-01 3.84730905e-01 -4.88224924e-01
1.01318371e+00 8.48057747e-01 4.95967388e-01 -1.33472717e+00
-3.95821214e-01 -1.70269713e-01 -8.98698747e-01 -4.08094555e-01
6.55570984e-01 -8.02740693e-01 -1.34449267e+00 2.99228936e-01
-1.03955638e+00 4.38204110e-01 2.18937740e-01 5.62277615e-01
-1.34987369e-01 4.99673694e-01 -5.42169154e-01 -1.27133298e+00
-5.99302709e-01 -1.06628442e+00 8.91660988e-01 4.47620541e-01
2.90546596e-01 -8.20564806e-01 -4.63715613e-01 1.27670184e-01
6.10030711e-01 5.48917830e-01 3.55183601e-01 5.14601432e-02
-4.27449942e-01 -4.57423508e-01 -4.04899240e-01 5.52855909e-01
3.75153244e-01 4.72041190e-01 -9.31626081e-01 -6.31620586e-01
3.63255739e-01 -2.02701569e-01 5.74739873e-01 6.21848941e-01
1.14444041e+00 -4.31034535e-01 -3.31925362e-01 9.06324565e-01
1.29685640e+00 3.13709348e-01 4.02305067e-01 6.19475208e-02
4.69754159e-01 7.52878606e-01 6.33630514e-01 3.40729505e-01
-3.18208903e-01 7.39995718e-01 2.61696815e-01 -6.77988753e-02
-4.23613101e-01 2.73189604e-01 2.74353772e-01 2.02452973e-01
-8.98496568e-01 2.55017310e-01 -4.60135221e-01 1.27746493e-01
-1.25854456e+00 -1.27475274e+00 1.30138323e-02 2.08302689e+00
5.51895499e-01 -4.27867442e-01 2.12364495e-01 1.34159224e-02
1.00002015e+00 -2.32707590e-01 -1.85274854e-01 -7.80729577e-02
-2.79704511e-01 2.71043092e-01 4.44918603e-01 4.96543050e-01
-1.16314232e+00 8.00545633e-01 6.88654327e+00 7.70755053e-01
-1.35779667e+00 1.29771724e-01 5.22055805e-01 -2.00661600e-01
3.64274532e-01 -6.04513764e-01 -1.39164281e+00 2.47867689e-01
2.33791247e-01 -5.51641807e-02 4.76412296e-01 8.27181041e-01
2.94437677e-01 1.60077602e-01 -6.12983525e-01 1.15367234e+00
2.69858360e-01 -9.32092786e-01 -8.67032039e-04 -1.09320357e-02
3.79019767e-01 -4.01631564e-01 4.63716418e-01 -2.92093188e-01
-2.96052009e-01 -1.33776748e+00 2.74122179e-01 3.52444142e-01
9.56986308e-01 -8.78239930e-01 5.02396941e-01 1.15405984e-01
-1.35918808e+00 -7.37401098e-02 -8.24090481e-01 1.71452716e-01
-3.38741466e-02 -2.50080019e-01 -7.70762324e-01 -9.85500664e-02
8.37702930e-01 2.71523863e-01 -2.87098527e-01 6.55556560e-01
9.64618027e-02 1.07414804e-01 -5.39977551e-01 1.44362912e-01
-3.09110910e-01 -2.49994516e-01 5.75830638e-01 1.14136553e+00
4.30273235e-01 5.13565898e-01 1.61172301e-01 6.56892061e-01
-1.14474215e-01 1.20140933e-01 -6.80424929e-01 2.26753175e-01
3.34979981e-01 1.68728256e+00 -8.53241444e-01 3.62654150e-01
-3.22071880e-01 5.63479125e-01 -4.29749861e-03 3.68813366e-01
-6.86876237e-01 -1.67191103e-01 6.46929204e-01 4.87395227e-01
3.27642292e-01 -2.38239571e-01 1.15330383e-01 -1.03461862e+00
1.65567651e-01 -5.42544127e-01 3.56576324e-01 -3.44871998e-01
-7.95868039e-01 8.80446553e-01 -3.85789946e-02 -9.30161893e-01
1.06102556e-01 -9.33239460e-01 -4.53663826e-01 8.77753675e-01
-1.56942117e+00 -1.50259697e+00 -3.12360555e-01 9.85576332e-01
5.59098363e-01 -4.57947940e-01 8.77937615e-01 4.91657227e-01
-7.82919109e-01 8.65506947e-01 -1.91965789e-01 4.63608682e-01
6.30978048e-01 -4.93154913e-01 -6.14666119e-02 7.11724222e-01
-2.66066045e-01 9.03419554e-01 7.05293596e-01 -5.88804781e-01
-1.58547246e+00 -1.01178491e+00 3.63851130e-01 -3.15640360e-01
1.96036458e-01 -2.97680110e-01 -8.60032082e-01 6.89486504e-01
-1.53122380e-01 6.57366514e-01 3.64983410e-01 -2.61686146e-01
-1.99731767e-01 -1.64747387e-01 -1.84076357e+00 3.24131012e-01
9.42209601e-01 -2.85720944e-01 -1.52646974e-01 4.49542612e-01
1.95753668e-03 -5.36761701e-01 -1.04736137e+00 6.18133068e-01
1.10059452e+00 -4.75345850e-01 1.42416990e+00 -5.01472473e-01
-3.15654159e-01 -3.38709861e-01 -5.85976196e-03 -5.42500377e-01
-5.50343573e-01 -5.89633763e-01 2.08122488e-02 1.31085122e+00
-5.51891439e-02 -5.18746436e-01 1.01723123e+00 5.21968246e-01
5.57021022e-01 -6.75471663e-01 -1.17125964e+00 -8.82563710e-01
-2.43614018e-01 2.12381721e-01 3.09471548e-01 8.40008914e-01
-4.67778563e-01 -1.82610631e-01 -5.73538005e-01 5.34067869e-01
9.55062866e-01 -1.72630966e-01 4.70192820e-01 -1.37747467e+00
1.93308413e-01 -1.45825416e-01 -5.04294217e-01 -2.65681952e-01
2.65941560e-01 -4.25076753e-01 -5.13330042e-01 -1.00672543e+00
2.90531993e-01 -1.62383437e-01 8.62158686e-02 5.09600282e-01
3.88340265e-01 8.79538357e-01 -8.43819380e-02 4.99170572e-02
-9.74427015e-02 2.45749444e-01 1.43407977e+00 1.08006604e-01
-1.09894767e-01 6.65611997e-02 -2.10265651e-01 1.18189180e+00
7.04225600e-01 -3.84711683e-01 -9.03695896e-02 -3.26663584e-01
-8.82724449e-02 5.24561889e-02 2.42823139e-01 -5.61791241e-01
3.19496036e-01 -3.89177799e-01 1.12855685e+00 -3.72290969e-01
7.44747341e-01 -9.23066556e-01 1.83814347e-01 6.60515487e-01
1.46901816e-01 1.92946773e-02 2.21301287e-01 1.61559671e-01
-1.06889471e-01 -3.89299780e-01 1.34876359e+00 -2.69640684e-01
-3.89315546e-01 6.11093760e-01 1.30491070e-02 -5.22603750e-01
1.18596220e+00 -5.98979354e-01 -3.24442953e-01 -1.33485779e-01
-9.40143645e-01 -1.33938044e-01 2.80870795e-01 5.69980085e-01
9.45442796e-01 -1.43842721e+00 -9.21892464e-01 8.62552881e-01
-3.83738905e-01 -3.07029426e-01 1.78435162e-01 7.59511352e-01
-5.65474629e-01 3.61160576e-01 -6.05071962e-01 -2.87462533e-01
-2.43427563e+00 4.00904775e-01 5.23097813e-01 4.48350161e-01
-4.04768020e-01 8.56375098e-01 -1.27095208e-01 -1.43712029e-01
3.13440412e-01 7.17783645e-02 -6.63214624e-01 -9.86187384e-02
8.68191779e-01 6.82731688e-01 -1.89506426e-01 -1.46299076e+00
-5.43764591e-01 1.42662036e+00 -3.04918736e-01 3.27601612e-01
1.35306501e+00 -1.27225429e-01 -3.92808974e-01 -4.64152932e-01
1.23747087e+00 6.84944838e-02 -9.83382940e-01 -1.09879062e-01
-4.37971950e-01 -1.05924249e+00 2.11677715e-01 -3.59968007e-01
-1.28854024e+00 6.55576885e-01 1.01339650e+00 9.11575034e-02
1.09700191e+00 -2.99766362e-01 3.73280376e-01 3.99809808e-01
2.23847196e-01 -1.19222605e+00 5.07588126e-02 3.34143847e-01
1.19181693e+00 -1.28428197e+00 2.62373000e-01 -8.50322127e-01
-1.02764860e-01 1.50302935e+00 7.55106986e-01 -2.10761771e-01
8.95409763e-01 6.39432967e-01 1.50435716e-02 -8.65873545e-02
-3.13603550e-01 -1.96925271e-02 4.18929696e-01 8.58813584e-01
6.72469378e-01 -3.75120848e-01 -3.58337373e-01 2.02947944e-01
2.99660955e-02 1.97078045e-02 2.18520448e-01 6.82807148e-01
-7.17415214e-01 -9.59797621e-01 -7.87654161e-01 -3.52784656e-02
-1.06847358e+00 5.44770479e-01 -1.63224474e-01 8.26599419e-01
2.10229740e-01 8.10364664e-01 7.09595010e-02 4.81393710e-02
4.64987904e-01 -1.43220887e-01 9.79870856e-01 -3.12608153e-01
-1.54995173e-01 5.80504179e-01 -2.63630301e-01 -3.26335341e-01
-5.58054209e-01 -5.72060049e-01 -1.31622756e+00 -3.33562642e-01
-5.06666481e-01 -2.20177144e-01 9.37524021e-01 5.76420903e-01
1.93922207e-01 1.15963228e-01 7.48459578e-01 -8.07331145e-01
-7.24792600e-01 -1.05405712e+00 -8.83584738e-01 4.20871586e-01
4.17550236e-01 -8.90168130e-01 -2.44074196e-01 1.58114284e-01] | [13.235306739807129, 0.6572482585906982] |
95e98473-cd57-41dc-82ce-d74357d432b3 | pessimistic-bootstrapping-for-uncertainty-1 | 2202.11566 | null | https://arxiv.org/abs/2202.11566v1 | https://arxiv.org/pdf/2202.11566v1.pdf | Pessimistic Bootstrapping for Uncertainty-Driven Offline Reinforcement Learning | Offline Reinforcement Learning (RL) aims to learn policies from previously collected datasets without exploring the environment. Directly applying off-policy algorithms to offline RL usually fails due to the extrapolation error caused by the out-of-distribution (OOD) actions. Previous methods tackle such problem by penalizing the Q-values of OOD actions or constraining the trained policy to be close to the behavior policy. Nevertheless, such methods typically prevent the generalization of value functions beyond the offline data and also lack precise characterization of OOD data. In this paper, we propose Pessimistic Bootstrapping for offline RL (PBRL), a purely uncertainty-driven offline algorithm without explicit policy constraints. Specifically, PBRL conducts uncertainty quantification via the disagreement of bootstrapped Q-functions, and performs pessimistic updates by penalizing the value function based on the estimated uncertainty. To tackle the extrapolating error, we further propose a novel OOD sampling method. We show that such OOD sampling and pessimistic bootstrapping yields provable uncertainty quantifier in linear MDPs, thus providing the theoretical underpinning for PBRL. Extensive experiments on D4RL benchmark show that PBRL has better performance compared to the state-of-the-art algorithms. | ['Zhaoran Wang', 'Peng Liu', 'Animesh Garg', 'Zhihong Deng', 'Zhuoran Yang', 'Lingxiao Wang', 'Chenjia Bai'] | 2022-02-23 | pessimistic-bootstrapping-for-uncertainty | https://openreview.net/forum?id=Y4cs1Z3HnqL | https://openreview.net/pdf?id=Y4cs1Z3HnqL | iclr-2022-4 | ['d4rl'] | ['robots'] | [-2.17143446e-01 4.09405261e-01 -6.33012295e-01 -2.15826273e-01
-9.52772439e-01 -6.96305990e-01 5.39184034e-01 3.22775692e-01
-5.42539835e-01 1.36630940e+00 -2.33903170e-01 -5.87198436e-01
-2.79823840e-01 -7.82213271e-01 -1.10970426e+00 -7.47580826e-01
-2.84057826e-01 6.46038294e-01 1.65984854e-01 1.62566811e-01
1.29675597e-01 1.63052037e-01 -1.29057550e+00 -1.48240179e-01
1.12723625e+00 1.44993043e+00 1.61406741e-01 3.02315831e-01
-2.72569768e-02 7.79496610e-01 -8.30209970e-01 -2.12394129e-02
4.89881247e-01 -3.14593881e-01 -3.14272970e-01 -1.58308208e-01
-1.56923994e-01 -1.00713217e+00 6.96930885e-02 1.34724605e+00
3.38647604e-01 3.88539582e-01 3.61903161e-01 -1.54640615e+00
-3.04935873e-01 7.88479030e-01 -5.32968283e-01 -1.71550319e-01
1.22799210e-01 4.54729885e-01 7.23262429e-01 -1.34607524e-01
3.50105077e-01 1.44686031e+00 1.85438752e-01 6.34798229e-01
-1.27381802e+00 -6.40064001e-01 5.39411962e-01 3.20608839e-02
-1.01464427e+00 9.36894957e-03 4.90243137e-01 -1.69214606e-01
5.29394627e-01 -1.04845591e-01 7.44119585e-01 1.26833546e+00
2.25154400e-01 1.00582814e+00 1.63693345e+00 -2.33142659e-01
1.05021739e+00 1.82361081e-01 -4.14327800e-01 2.97821641e-01
4.29715335e-01 7.95729578e-01 -2.15006575e-01 -3.41562778e-01
7.01415181e-01 -8.44528005e-02 -2.40683258e-02 -4.48077410e-01
-6.32479668e-01 7.64013290e-01 1.24418974e-01 -4.32860404e-01
-5.45983911e-01 3.47143084e-01 4.99185145e-01 3.38052243e-01
3.32816422e-01 4.60657090e-01 -5.28169751e-01 -4.86289918e-01
-7.60984361e-01 5.77439487e-01 6.29693687e-01 9.71951604e-01
5.07019460e-01 1.01374581e-01 -7.09263802e-01 3.35495263e-01
1.73046008e-01 5.65458179e-01 3.05179864e-01 -1.16897035e+00
5.87448061e-01 2.36946464e-01 9.76817667e-01 -4.20679569e-01
-3.34541732e-03 -2.43011460e-01 -2.34708905e-01 5.20111561e-01
7.03140557e-01 -5.03006101e-01 -7.69491315e-01 1.87516260e+00
5.01739919e-01 -1.25450576e-02 1.00577928e-01 9.46462333e-01
-1.93245605e-01 6.09821856e-01 3.83357108e-02 -7.31750369e-01
7.48228252e-01 -6.01058364e-01 -8.13167214e-01 6.25512674e-02
3.05140346e-01 9.51456353e-02 1.28329206e+00 8.05675566e-01
-8.96818578e-01 -2.56825835e-01 -9.38122988e-01 6.36377513e-01
1.71672165e-01 -2.93261819e-02 2.42395371e-01 6.03061557e-01
-6.24155939e-01 9.29761648e-01 -9.71675336e-01 2.68813282e-01
5.49927056e-01 1.76449999e-01 2.30580747e-01 1.90273598e-01
-1.13235462e+00 8.28644872e-01 6.87260211e-01 -2.33117845e-02
-1.60755682e+00 -6.62281632e-01 -5.14820635e-01 -8.70455205e-02
1.26075995e+00 -6.54636323e-02 1.75132549e+00 -9.78709817e-01
-2.15808702e+00 4.60900627e-02 3.76101702e-01 -1.08358097e+00
1.05546880e+00 -3.55269015e-01 -5.16428836e-02 1.61496084e-02
-1.23944648e-01 4.59268689e-01 1.25773096e+00 -1.29198897e+00
-6.73579514e-01 -1.98319688e-01 3.42866331e-01 1.63205400e-01
-3.46804783e-02 -5.23583293e-01 1.20089706e-02 -3.24447811e-01
-4.86225784e-01 -7.91082621e-01 -3.95764053e-01 -2.06579626e-01
-4.04055476e-01 -3.67312372e-01 5.87809682e-01 -3.91260266e-01
1.33728349e+00 -1.91368151e+00 -3.28817487e-01 1.85459331e-01
-2.47350782e-01 8.13200176e-02 8.48024860e-02 4.14128900e-01
5.04950464e-01 -4.98000942e-02 -2.47152001e-01 2.00097403e-03
5.19530058e-01 5.07207572e-01 -8.77599120e-01 4.33829010e-01
-8.75884295e-02 6.39079332e-01 -1.24553418e+00 -3.07032615e-01
1.31334156e-01 -2.89692700e-01 -4.67705220e-01 4.43928689e-01
-1.07550406e+00 6.39845014e-01 -6.86426044e-01 4.63585466e-01
6.72858119e-01 2.06574306e-01 4.79447067e-01 2.56705672e-01
-2.09839225e-01 1.00592896e-01 -1.05036294e+00 1.25761747e+00
-5.39007127e-01 -3.60683464e-02 4.93138470e-02 -9.65048134e-01
8.68427515e-01 4.14518528e-02 3.34801406e-01 -5.64247370e-01
2.76845127e-01 1.45818517e-01 -1.61518469e-01 -2.69988894e-01
2.23284960e-01 -2.88313478e-01 -2.52019405e-01 4.40946877e-01
-1.50275350e-01 -4.76975530e-01 2.82943565e-02 -2.48471022e-01
9.24398124e-01 6.77240849e-01 5.67427099e-01 -1.97891802e-01
2.28176266e-01 -1.55495435e-01 7.88070023e-01 9.91417885e-01
-6.24694824e-01 -2.52400219e-01 1.01290941e+00 -6.80780187e-02
-8.70789528e-01 -1.11007822e+00 -7.70397112e-02 7.45097578e-01
1.59308180e-01 -4.12438996e-02 -8.11136782e-01 -1.27931225e+00
5.20412445e-01 1.38960803e+00 -6.38968110e-01 -1.20771617e-01
-7.79785961e-02 -4.05061066e-01 2.24852681e-01 4.99488533e-01
4.53986108e-01 -7.46754169e-01 -9.83647466e-01 3.78199190e-01
2.47039124e-01 -9.80870366e-01 -1.60537690e-01 2.98342586e-01
-7.89305449e-01 -8.84214342e-01 -5.77517331e-01 2.06946701e-01
6.28020048e-01 -4.45678115e-01 6.97326303e-01 -6.17287695e-01
3.12955707e-01 4.35677767e-01 -1.46940008e-01 -6.66652977e-01
-4.86071229e-01 -3.49703729e-01 3.76115680e-01 -1.90774903e-01
1.49798751e-01 -4.14099544e-01 -6.19823933e-01 3.18341881e-01
-6.50340140e-01 -3.11689198e-01 4.54097003e-01 9.12967384e-01
9.97374773e-01 3.29940319e-01 8.08198214e-01 -6.38247311e-01
8.99934947e-01 -4.76933867e-01 -1.52477920e+00 4.07493204e-01
-9.25910413e-01 5.68841815e-01 1.09750867e+00 -7.38179386e-01
-1.19033062e+00 -2.34762058e-01 2.87546366e-01 -7.93501735e-01
-1.55405989e-02 7.19244853e-02 -1.31978348e-01 3.30506712e-01
4.91542011e-01 1.46081597e-01 2.10246205e-01 -3.57275426e-01
3.37935567e-01 6.46518707e-01 2.73492873e-01 -1.50317383e+00
4.87230271e-01 4.00790334e-01 1.71622321e-01 -1.75405979e-01
-1.13191760e+00 1.56313241e-01 1.37895867e-01 -4.50038016e-01
3.73314470e-01 -6.40446365e-01 -1.28537667e+00 2.33592894e-02
-6.24052107e-01 -8.38925421e-01 -8.37046862e-01 5.61156213e-01
-1.19502032e+00 2.89818943e-01 -7.33689517e-02 -1.63961458e+00
-3.35850343e-02 -1.09293222e+00 7.58457839e-01 2.70747155e-01
2.45023206e-01 -6.24763012e-01 2.26527214e-01 -8.17074478e-02
1.44347966e-01 5.08133650e-01 7.02455699e-01 -4.91823047e-01
-4.07250404e-01 9.32482556e-02 9.11355838e-02 5.17191648e-01
-1.59707054e-01 -1.45633206e-01 -7.83492267e-01 -4.97512162e-01
1.88313827e-01 -8.69728863e-01 5.43122649e-01 5.16898155e-01
1.64591873e+00 -7.94681609e-01 -7.76607767e-02 9.87791121e-02
1.47246826e+00 5.80660760e-01 3.14507842e-01 5.03769040e-01
-1.94617927e-01 4.61064041e-01 1.36699331e+00 1.27201223e+00
1.46847874e-01 5.65850377e-01 8.50010574e-01 8.10001373e-01
6.62313759e-01 -8.65711808e-01 6.57646477e-01 -1.15741849e-01
1.47177234e-01 -1.80836231e-01 -5.16662717e-01 3.42124730e-01
-2.06720138e+00 -7.87959397e-01 5.89827120e-01 2.82242179e+00
1.28710890e+00 4.88973171e-01 4.57532227e-01 -2.48521492e-01
5.85091650e-01 -8.37019086e-02 -1.33655715e+00 -6.28668785e-01
3.92943352e-01 4.08580787e-02 9.31336045e-01 4.53940064e-01
-7.66179562e-01 7.64531851e-01 5.45813179e+00 1.26723862e+00
-8.59110117e-01 8.74975473e-02 6.54807866e-01 -2.88376570e-01
-2.85036415e-01 1.42916560e-01 -9.90323007e-01 8.38991582e-01
1.01804936e+00 -2.78983742e-01 8.74435604e-01 1.20247042e+00
4.87188369e-01 -6.32208109e-01 -1.16505647e+00 6.08742058e-01
-6.82692707e-01 -8.77792597e-01 -4.31705743e-01 2.34862790e-02
9.17288363e-01 -1.66478962e-01 2.42818929e-02 7.97492623e-01
8.50071609e-01 -8.11192989e-01 1.02307081e+00 7.12363064e-01
7.94134915e-01 -1.04531813e+00 5.51889837e-01 8.19373488e-01
-5.98339915e-01 -6.17735505e-01 -4.69888031e-01 -1.14111453e-01
-4.17094976e-02 6.54112875e-01 -8.90719593e-01 4.01393980e-01
4.22579318e-01 1.76181614e-01 1.24073714e-01 6.48907125e-01
-5.58526516e-01 6.86385095e-01 -5.30951679e-01 -4.15369987e-01
4.31938261e-01 -4.47571218e-01 4.84355569e-01 5.18515766e-01
2.56146014e-01 -1.58817321e-01 4.25658375e-01 1.10655057e+00
1.51622638e-01 -2.24633694e-01 -3.37919325e-01 -2.67733514e-01
7.16339886e-01 7.38737822e-01 -3.91646773e-01 -2.69871294e-01
2.06880093e-01 4.54130203e-01 4.09894824e-01 3.33672673e-01
-1.02223337e+00 -1.62013695e-01 5.29501259e-01 -2.42332891e-01
3.40438187e-01 -1.36412203e-01 -3.79612483e-02 -7.50979245e-01
4.93862182e-02 -8.87345493e-01 3.38413715e-01 -3.54873419e-01
-1.29485714e+00 3.16206031e-02 4.46178347e-01 -1.26047695e+00
-5.85023582e-01 -5.09994149e-01 -3.74421030e-01 5.00490367e-01
-1.40441382e+00 -3.03059965e-01 3.06618065e-01 2.23402068e-01
4.65305656e-01 1.82867631e-01 3.88613403e-01 -3.97143245e-01
-5.49100697e-01 5.05105138e-01 6.08704388e-01 -5.23556292e-01
5.54769337e-01 -1.38678432e+00 -1.52408674e-01 4.02333111e-01
-6.07796848e-01 1.11284189e-01 9.84470606e-01 -7.27949679e-01
-1.35550046e+00 -1.01613450e+00 -2.15835512e-01 -1.23196818e-01
8.69353831e-01 -1.74442753e-01 -6.52323306e-01 4.10423070e-01
-1.79757193e-01 2.16732904e-01 -1.29687758e-02 -2.22297132e-01
-7.73623362e-02 -3.32677126e-01 -1.56396902e+00 5.70384741e-01
7.52442420e-01 -5.13453074e-02 -4.48481977e-01 1.90724328e-01
9.56817389e-01 -4.99991238e-01 -8.27966869e-01 4.18665290e-01
5.17051101e-01 -9.01844263e-01 5.48026562e-01 -8.01313818e-01
1.97335169e-01 -2.48689279e-01 -2.00997561e-01 -1.49945533e+00
4.31053549e-01 -9.86529231e-01 -7.07417667e-01 1.04588556e+00
-7.09315017e-03 -8.28308344e-01 5.86278379e-01 5.07109344e-01
8.36564600e-02 -1.08019996e+00 -1.19704998e+00 -1.49062634e+00
2.31451914e-01 -5.87114930e-01 8.36827219e-01 4.34135944e-01
1.78210214e-01 -5.06133974e-01 -3.84128243e-01 1.44377410e-01
9.82765734e-01 2.43608415e-01 7.49274373e-01 -5.59147298e-01
-7.82305479e-01 -2.98875898e-01 1.36723518e-01 -1.05562890e+00
3.73643547e-01 -1.95440561e-01 4.82848078e-01 -1.11225939e+00
-1.28650844e-01 -5.98304689e-01 -3.26013118e-01 4.14312094e-01
1.14141628e-01 -7.20000982e-01 1.77405849e-01 -1.18715599e-01
-8.25644255e-01 1.12173522e+00 1.48423922e+00 3.29720937e-02
-3.91338408e-01 3.09959441e-01 -2.72033334e-01 6.39495254e-01
9.32238281e-01 -4.99778837e-01 -8.08976114e-01 2.28090644e-01
2.70080656e-01 5.52468061e-01 2.49088883e-01 -7.37211168e-01
-2.36848965e-01 -9.64071572e-01 7.51561150e-02 -7.35904038e-01
-6.31569978e-03 -7.41268814e-01 -1.15710132e-01 6.72887325e-01
-5.55548966e-01 -4.70017284e-01 2.05465794e-01 1.08734667e+00
1.74896911e-01 -3.49937111e-01 7.33444691e-01 -1.70814797e-01
-4.46218848e-01 3.04838210e-01 -3.05793464e-01 4.40047979e-01
1.20805204e+00 2.61824578e-01 -2.61447191e-01 -3.95198017e-01
-5.03151715e-01 7.27740884e-01 3.07524979e-01 -3.11858729e-02
3.95780712e-01 -1.06621230e+00 -2.25898489e-01 -6.76921709e-03
-7.49549717e-02 1.02991842e-01 1.12503313e-01 7.21688628e-01
1.03403740e-01 3.30294281e-01 -5.56353629e-02 -3.92316848e-01
-3.52254868e-01 8.42119873e-01 4.70157295e-01 -5.56538224e-01
-4.27609503e-01 4.07179326e-01 -1.00072592e-01 -4.20092404e-01
5.95282555e-01 -6.75586045e-01 1.92811891e-01 -4.54429276e-02
3.07922304e-01 6.19660139e-01 -1.55641004e-01 3.70873660e-01
-9.59942266e-02 -1.15613960e-01 1.07915789e-01 -5.20303369e-01
9.56307590e-01 -1.09746546e-01 4.14829522e-01 6.01452112e-01
5.55626631e-01 -2.44159892e-01 -2.28328323e+00 -6.41057566e-02
8.53977203e-02 -5.67004502e-01 1.46620572e-01 -1.19742250e+00
-4.99977022e-01 5.94283044e-01 4.85491782e-01 2.97987193e-01
1.02088690e+00 -3.74286950e-01 5.61557472e-01 4.82958168e-01
9.30410624e-01 -1.71798742e+00 -1.69255227e-01 2.47950852e-01
8.01772475e-01 -1.17801714e+00 1.28635034e-01 4.21614081e-01
-9.64531302e-01 9.57850337e-01 7.93830812e-01 -2.68596083e-01
4.21131641e-01 2.55029976e-01 -4.31914777e-01 4.63203877e-01
-1.05842984e+00 -8.57584700e-02 -1.27469555e-01 5.39112926e-01
-4.09068078e-01 4.68131572e-01 -5.17262459e-01 8.44736636e-01
-6.62331004e-03 4.30374861e-01 3.84342313e-01 1.07573080e+00
-6.00034475e-01 -9.77387905e-01 -6.04254842e-01 3.44581813e-01
-3.07291836e-01 3.57567519e-01 2.04274505e-02 8.55669856e-01
-2.12064907e-02 8.34166229e-01 2.10612118e-02 1.08033447e-02
1.02709189e-01 -7.50322416e-02 7.32578814e-01 -2.46424764e-01
-1.31109416e-01 5.52961454e-02 6.94040060e-02 -8.56734753e-01
8.62999037e-02 -5.88419914e-01 -1.47021866e+00 -5.74247204e-02
-2.40589038e-01 3.15784097e-01 7.34236538e-01 1.07560778e+00
2.87361741e-01 2.41405040e-01 1.00304496e+00 -6.36640191e-01
-1.91972041e+00 -8.33078980e-01 -7.03010917e-01 -1.35589838e-01
4.07014072e-01 -1.13399327e+00 -5.43025792e-01 -7.61575043e-01] | [4.219857215881348, 2.4509871006011963] |
1f3186cb-7908-44a8-b5b3-f6bdcba7ac4f | a-brief-review-of-hypernetworks-in-deep | 2306.06955 | null | https://arxiv.org/abs/2306.06955v1 | https://arxiv.org/pdf/2306.06955v1.pdf | A Brief Review of Hypernetworks in Deep Learning | Hypernetworks, or hypernets in short, are neural networks that generate weights for another neural network, known as the target network. They have emerged as a powerful deep learning technique that allows for greater flexibility, adaptability, faster training, information sharing, and model compression etc. Hypernets have shown promising results in a variety of deep learning problems, including continual learning, causal inference, transfer learning, weight pruning, uncertainty quantification, zero-shot learning, natural language processing, and reinforcement learning etc. Despite their success across different problem settings, currently, there is no review available to inform the researchers about the developments and help in utilizing hypernets. To fill this gap, we review the progress in hypernets. We present an illustrative example to train deep neural networks using hypernets and propose to categorize hypernets on five criteria that affect the design of hypernets as inputs, outputs, variability of inputs and outputs, and architecture of hypernets. We also review applications of hypernets across different deep learning problem settings. Finally, we discuss the challenges and future directions that remain under-explored in the field of hypernets. We believe that hypernetworks have the potential to revolutionize the field of deep learning. They offer a new way to design and train neural networks, and they have the potential to improve the performance of deep learning models on a variety of tasks. Through this review, we aim to inspire further advancements in deep learning through hypernetworks. | ['David A. Clifton', 'Soheila Molaei', 'Ping Lu', 'Jiandong Zhou', 'Vinod Kumar Chauhan'] | 2023-06-12 | null | null | null | null | ['causal-inference', 'model-compression', 'causal-inference'] | ['knowledge-base', 'methodology', 'miscellaneous'] | [-4.93654460e-02 4.61028904e-01 -2.27225140e-01 -3.91411424e-01
2.46406108e-01 -1.98152333e-01 5.06650269e-01 -2.20233724e-01
-4.76197332e-01 9.94590342e-01 1.01098932e-01 -1.99830696e-01
-6.81946278e-01 -1.05754173e+00 -6.70908511e-01 -8.74285281e-01
-2.65616208e-01 4.01101917e-01 3.20230514e-01 -6.52157441e-02
1.86750665e-01 4.56399471e-01 -1.56891537e+00 3.88277143e-01
3.03554654e-01 1.15057826e+00 3.79873365e-02 3.92455965e-01
-2.20654920e-01 9.67955172e-01 -6.93485260e-01 -2.99484521e-01
6.47127107e-02 -1.04443952e-01 -7.48191297e-01 -5.49150765e-01
1.05855107e-01 -6.18884452e-02 -7.29525089e-01 9.01550293e-01
7.18093276e-01 1.91564664e-01 7.11739361e-01 -1.48872066e+00
-6.49883330e-01 1.26438808e+00 -1.45264551e-01 4.19851482e-01
-3.07719022e-01 2.16578603e-01 8.53101134e-01 -6.73780739e-01
3.63375634e-01 1.41537249e+00 8.96849334e-01 9.46486950e-01
-8.07209909e-01 -9.95957494e-01 1.37403190e-01 5.50892949e-01
-9.54034925e-01 -3.71835560e-01 4.40835625e-01 -2.01639339e-01
1.31589258e+00 -1.49173573e-01 7.30826497e-01 1.46676874e+00
4.79545772e-01 1.09485984e+00 6.87762320e-01 -3.83672953e-01
4.17183816e-01 -5.64998053e-02 1.75064474e-01 6.60661161e-01
2.63246447e-01 3.81014287e-01 -6.17676437e-01 9.08750575e-03
7.19345450e-01 3.19970399e-02 7.57961944e-02 -1.27149150e-01
-9.99920368e-01 1.03584838e+00 5.26085496e-01 3.72329980e-01
-3.45870256e-01 7.04023778e-01 5.63683867e-01 4.33309734e-01
1.29334033e-01 5.60457766e-01 -6.43135130e-01 3.38778342e-03
-5.63696504e-01 3.07961583e-01 8.50243032e-01 8.10679197e-01
5.14601409e-01 7.53597260e-01 -1.65760472e-01 9.54467475e-01
2.43267700e-01 1.81573480e-01 1.01302218e+00 -7.56512523e-01
3.86719555e-01 5.86501718e-01 -7.45013714e-01 -6.21117830e-01
-6.10695124e-01 -3.91502261e-01 -1.20728981e+00 2.07209617e-01
-1.03007890e-01 -7.70956874e-01 -1.26014650e+00 1.63216782e+00
-9.06192437e-02 4.19495016e-01 2.46657491e-01 4.32632297e-01
1.10340643e+00 7.26414800e-01 1.61918953e-01 9.50242579e-02
9.48270142e-01 -7.83839107e-01 -6.81493878e-01 -2.30986670e-01
3.68335843e-01 -3.21572274e-01 5.35037875e-01 5.60678780e-01
-1.12580073e+00 -4.18803006e-01 -1.16688979e+00 3.25317442e-01
-8.98930132e-01 -3.90933186e-01 9.26997721e-01 4.93794024e-01
-9.78764713e-01 8.48006248e-01 -8.80244493e-01 -1.09946348e-01
7.34449625e-01 8.75249803e-01 -6.31805435e-02 1.23968117e-01
-1.79358137e+00 9.58446145e-01 1.10630012e+00 1.83477793e-02
-1.16010809e+00 -7.81905949e-01 -7.76401103e-01 2.60768622e-01
2.10256055e-01 -6.08587325e-01 1.22837985e+00 -6.39864981e-01
-1.41678178e+00 3.39381248e-01 5.65917432e-01 -9.33202505e-01
1.22418806e-01 -1.80729032e-01 -4.86560315e-01 -4.02216822e-01
-4.35814232e-01 9.78060126e-01 6.54223740e-01 -7.79586673e-01
-7.86094189e-01 -1.21211812e-01 -1.04018494e-01 8.98929164e-02
-5.91734409e-01 -3.21879566e-01 -2.62239724e-01 -6.15142822e-01
-2.22203583e-01 -8.59983981e-01 -2.89360523e-01 -2.62500167e-01
-4.38938260e-01 -7.14994967e-01 1.05130374e+00 3.46098959e-01
1.12228405e+00 -2.10017180e+00 4.32935245e-02 3.16376507e-01
3.77699018e-01 8.03356528e-01 -1.38067991e-01 4.97565210e-01
-9.50144827e-02 2.68819779e-01 -1.19844496e-01 2.02582613e-01
-3.41023020e-02 6.19158268e-01 -2.27106959e-01 -5.50176539e-02
3.02721649e-01 1.18581772e+00 -9.24758434e-01 -3.36583436e-01
4.32984740e-01 6.20628834e-01 -2.88225114e-01 -7.86022376e-03
-3.77688408e-01 -3.09254318e-01 -5.03301382e-01 3.77395988e-01
2.99478859e-01 -9.57242548e-02 4.68337461e-02 2.01738458e-02
1.01022519e-01 1.25057414e-01 -1.29374886e+00 1.22270286e+00
-3.10930014e-01 1.00797355e+00 -3.94752204e-01 -1.24943697e+00
1.05230427e+00 6.14812016e-01 3.54210973e-01 -6.30270004e-01
2.91850865e-01 9.27764401e-02 3.75429571e-01 -5.39009571e-01
1.08758129e-01 -1.31517991e-01 1.90857157e-01 4.40047741e-01
6.87514663e-01 8.94292817e-02 1.87730551e-01 -9.28387865e-02
1.05493426e+00 -3.24651569e-01 2.66910195e-01 -1.49335563e-01
2.33003423e-01 -2.15158060e-01 6.09709442e-01 8.16945612e-01
-1.10604642e-02 3.78401458e-01 6.31605983e-01 -8.25709879e-01
-9.72658515e-01 -9.55311596e-01 -3.11824411e-01 1.09051120e+00
-3.29231203e-01 -9.81958583e-02 -5.75889289e-01 -5.17887235e-01
1.79036677e-01 7.01937079e-01 -7.69035339e-01 -5.41671574e-01
-4.70582724e-01 -1.23197722e+00 1.01993465e+00 7.29472339e-01
8.51067185e-01 -1.63641179e+00 -5.89981496e-01 3.09821934e-01
3.51922989e-01 -8.57277632e-01 2.56527334e-01 7.86684871e-01
-1.29946113e+00 -9.81180787e-01 -4.73878920e-01 -9.11692679e-01
3.32780689e-01 -2.19362646e-01 1.06614661e+00 6.08113445e-02
-2.91257620e-01 3.41217183e-02 -1.32351443e-01 -9.00714397e-01
-1.88224733e-01 5.15704334e-01 4.05317545e-01 -5.27038395e-01
6.56145513e-01 -7.79825449e-01 -4.73207593e-01 2.74196029e-01
-1.16781652e+00 -1.60002545e-01 7.26974547e-01 1.06604743e+00
2.37134352e-01 3.27127516e-01 1.03560233e+00 -1.35407960e+00
1.00713575e+00 -8.82928371e-01 -5.36814034e-01 3.07280809e-01
-9.81546700e-01 5.03180683e-01 5.76516688e-01 -5.71766913e-01
-9.71132934e-01 -2.93464094e-01 -2.34758928e-01 -4.68353570e-01
-3.76779996e-02 6.51414990e-01 1.69818431e-01 -8.54759961e-02
8.98443818e-01 -1.63844973e-01 -9.80936363e-02 -2.88986564e-01
3.90126407e-01 4.97028321e-01 2.30057225e-01 -4.12089050e-01
3.36396962e-01 -1.25703305e-01 9.92596745e-02 -5.39088011e-01
-6.86104655e-01 -1.42105237e-01 -4.34281677e-01 -8.42228010e-02
6.53236806e-01 -6.03586793e-01 -7.65138686e-01 3.93504113e-01
-9.58418727e-01 -5.52242398e-01 -3.89920086e-01 4.97373819e-01
-3.92451346e-01 -4.61718649e-01 -6.14753664e-01 -5.78645885e-01
-5.19067466e-01 -1.13733280e+00 2.35601425e-01 5.61601520e-01
-9.86244604e-02 -1.50427425e+00 5.53925261e-02 -2.09710225e-01
6.04851961e-01 1.52198151e-01 1.12109339e+00 -9.53683197e-01
-4.20869768e-01 -1.07651316e-01 -1.19919881e-01 5.48898339e-01
-3.17240953e-02 1.34257823e-01 -1.18813384e+00 8.07753578e-02
-3.06919605e-01 -8.18677247e-01 1.16068673e+00 5.50652683e-01
1.47378397e+00 -4.01970983e-01 -5.10992289e-01 7.20251024e-01
1.42092538e+00 3.69233549e-01 6.50545597e-01 4.03725296e-01
4.86571550e-01 3.69353473e-01 -2.53950004e-02 4.66928512e-01
-6.06335923e-02 1.35492921e-01 8.51862788e-01 2.10972056e-01
-2.78444648e-01 -1.59669966e-01 8.17834735e-02 9.03163612e-01
-1.63223684e-01 -4.14380819e-01 -1.01244998e+00 4.42764431e-01
-1.97036386e+00 -1.05514956e+00 2.47169778e-01 1.76014626e+00
8.71198237e-01 3.52817148e-01 -2.43415296e-01 2.54606456e-01
4.66539741e-01 1.58132598e-01 -9.44119334e-01 -6.25508606e-01
-1.50650665e-01 3.96207124e-01 6.09387994e-01 2.05282688e-01
-9.18743491e-01 1.19413674e+00 7.41973829e+00 8.49769115e-01
-1.28646207e+00 -7.84169585e-02 7.25163221e-01 -1.90054223e-01
-1.79864064e-01 -5.48089385e-01 -1.20818877e+00 2.37529561e-01
1.16678941e+00 -1.64146960e-01 3.63118500e-01 1.14035070e+00
-2.98885144e-02 2.52359092e-01 -1.11201358e+00 8.51985693e-01
-2.32777372e-01 -1.75557959e+00 1.90424830e-01 -1.09350532e-01
1.10139716e+00 5.20770252e-01 2.89169103e-01 6.81746662e-01
7.83064723e-01 -1.35448515e+00 1.77313834e-01 4.48370785e-01
6.08593881e-01 -9.01010811e-01 1.05368829e+00 2.65735507e-01
-7.47711062e-01 -3.18155378e-01 -7.71177649e-01 -1.50144652e-01
-1.78700984e-01 5.45122981e-01 -1.01414502e+00 1.34829640e-01
6.92283273e-01 7.23438561e-01 -3.11089098e-01 1.32995892e+00
-1.74320549e-01 7.19675124e-01 -2.00617537e-01 -5.24941325e-01
5.66102564e-01 2.96651930e-01 1.76738128e-01 1.22245288e+00
9.91876945e-02 1.91299230e-01 -2.08783150e-01 8.69757056e-01
-3.85602236e-01 -3.57779980e-01 -8.25062215e-01 -4.91747260e-02
6.19525373e-01 1.02151918e+00 -5.11235833e-01 -2.34027520e-01
-1.40124202e-01 2.17243314e-01 3.66164088e-01 4.20284063e-01
-5.81678510e-01 -6.15237594e-01 7.29568481e-01 -2.79611617e-01
-9.23082605e-03 -4.05613706e-02 -5.56846261e-01 -8.23583901e-01
-5.18630266e-01 -8.51489365e-01 5.32971740e-01 -6.19847894e-01
-1.14202094e+00 7.90209353e-01 3.94911677e-01 -7.76497722e-01
-6.72744095e-01 -1.14285910e+00 -7.76011348e-01 5.55759013e-01
-1.33626497e+00 -4.40920651e-01 -1.13172732e-01 6.79835200e-01
7.01355457e-01 -8.36493969e-01 8.69475365e-01 1.64836198e-01
-6.50103033e-01 6.03354156e-01 2.19660014e-01 2.93485850e-01
3.20396304e-01 -1.18129802e+00 4.17954773e-01 3.79664481e-01
1.94924414e-01 5.44812858e-01 4.89717841e-01 -4.17366236e-01
-1.21846652e+00 -1.18421268e+00 5.50896883e-01 -4.92746048e-02
5.60538173e-01 -1.22750558e-01 -8.06249857e-01 7.33932018e-01
4.06211168e-01 -1.90123692e-02 6.25662684e-01 3.27707678e-01
-1.19873099e-01 -3.23162675e-01 -1.08591354e+00 5.53708971e-01
9.25720811e-01 -9.55098644e-02 -6.93188787e-01 1.96301550e-01
7.19699740e-01 -3.31717104e-01 -7.29712963e-01 5.69203734e-01
5.96471727e-01 -9.05550480e-01 8.86073768e-01 -9.43886459e-01
5.15000761e-01 4.00964975e-01 1.15881853e-01 -1.60729492e+00
-4.37639147e-01 -4.11458224e-01 -4.75265503e-01 9.57131982e-01
5.84863424e-01 -6.68250918e-01 1.14574432e+00 6.36944532e-01
-2.66791552e-01 -1.17434120e+00 -7.23857999e-01 -6.65768385e-01
2.62989640e-01 -6.72376037e-01 6.56717360e-01 8.05928767e-01
-4.77176020e-03 3.51054430e-01 -2.65106797e-01 -2.81744897e-01
5.63264370e-01 -6.58723176e-01 1.35209605e-01 -1.55194926e+00
1.37109086e-01 -7.25351632e-01 -5.80244303e-01 -6.79010808e-01
2.73953348e-01 -8.05360615e-01 -1.96581155e-01 -1.56294477e+00
-7.70973712e-02 -6.78845704e-01 -6.74376488e-01 8.64968359e-01
2.38014966e-01 1.39686137e-01 1.19184412e-01 -2.10276879e-02
-3.13370168e-01 4.77386773e-01 1.15826297e+00 -2.28720352e-01
-2.53448665e-01 4.03340846e-01 -5.56980252e-01 8.29464734e-01
1.24612379e+00 -6.45514786e-01 -9.56615508e-01 -7.02446759e-01
4.88339841e-01 -3.33705693e-01 -1.50914770e-02 -1.22640789e+00
4.90637511e-01 -2.82218635e-01 6.78483248e-01 -3.24574500e-01
2.55151421e-01 -7.21811950e-01 -1.02788799e-01 5.90258956e-01
-6.44546270e-01 2.60559738e-01 4.82101589e-01 4.36047167e-01
-2.75768071e-01 -6.99019432e-01 8.16879749e-01 -5.33763587e-01
-9.32153523e-01 6.42139375e-01 -5.53973198e-01 2.63400078e-01
9.83427703e-01 -6.78806528e-02 -3.19636166e-01 -8.35440382e-02
-6.69636130e-01 4.39102024e-01 -5.12028396e-01 8.52391779e-01
9.45921898e-01 -1.40216708e+00 -3.87602091e-01 2.30774030e-01
-2.01993942e-01 2.65288711e-01 -9.06879008e-02 2.42798522e-01
-3.41674536e-01 7.14843869e-01 -5.59379578e-01 -4.14530128e-01
-9.18798745e-01 2.31955841e-01 5.68245709e-01 -1.46369576e-01
-5.68902671e-01 9.73169029e-01 3.63005288e-02 -4.78691697e-01
9.45324719e-01 -3.30454618e-01 -4.66759026e-01 -5.56813329e-02
6.39813364e-01 4.91155386e-01 1.70759842e-01 2.28672400e-01
-1.64983049e-01 1.69419125e-01 -2.38513902e-01 -1.02682404e-01
1.58884025e+00 3.97276133e-01 8.18071812e-02 7.15772808e-01
1.03281701e+00 -1.03778553e+00 -1.19274640e+00 -3.07995230e-01
1.39683679e-01 1.60251379e-01 2.57526696e-01 -9.00486946e-01
-1.51636803e+00 1.19968331e+00 7.09220886e-01 3.06079626e-01
8.87365520e-01 -2.04142094e-01 6.14853978e-01 7.85474896e-01
7.88012519e-02 -1.22903001e+00 3.09492111e-01 1.01969433e+00
7.14819670e-01 -1.21963227e+00 -1.65566176e-01 3.86499822e-01
-4.91298258e-01 1.56697631e+00 9.14602101e-01 -2.23194659e-01
1.09360433e+00 4.64301676e-01 -1.57165319e-01 -2.75957674e-01
-1.37540352e+00 5.29733734e-06 2.00406373e-01 8.06450784e-01
5.93947172e-01 -8.15452337e-02 2.82234512e-02 3.42415124e-01
-3.46299887e-01 2.22670063e-01 3.31846535e-01 5.88511646e-01
-7.45188594e-01 -1.09265852e+00 1.74565390e-02 7.79102027e-01
-4.01697755e-01 -3.50504935e-01 -1.88739508e-01 8.26025128e-01
3.66772443e-01 6.08466506e-01 8.43162760e-02 -5.66575885e-01
1.99060485e-01 2.02029258e-01 2.76844263e-01 -6.76210761e-01
-7.10759163e-01 -6.26746416e-01 -7.70051591e-03 -1.96787044e-01
-1.20611727e-01 -1.96204722e-01 -1.31359148e+00 -5.50171494e-01
-1.67616904e-01 1.21241167e-01 8.84166121e-01 1.04984879e+00
2.00850531e-01 8.55452836e-01 2.72633493e-01 -6.46468341e-01
-6.31897748e-01 -1.10247529e+00 -3.48152220e-01 -1.71218842e-01
1.63678616e-01 -7.81131804e-01 -1.96906015e-01 -2.40387991e-01] | [8.586774826049805, 3.2307162284851074] |
32bb771e-36bc-4f1a-922a-dcd6dc5439c1 | learning-functional-distributional-semantics-1 | 2204.10624 | null | https://arxiv.org/abs/2204.10624v1 | https://arxiv.org/pdf/2204.10624v1.pdf | Learning Functional Distributional Semantics with Visual Data | Functional Distributional Semantics is a recently proposed framework for learning distributional semantics that provides linguistic interpretability. It models the meaning of a word as a binary classifier rather than a numerical vector. In this work, we propose a method to train a Functional Distributional Semantics model with grounded visual data. We train it on the Visual Genome dataset, which is closer to the kind of data encountered in human language acquisition than a large text corpus. On four external evaluation datasets, our model outperforms previous work on learning semantics from Visual Genome. | ['Guy Emerson', 'Yinhong Liu'] | 2022-04-22 | null | https://aclanthology.org/2022.acl-long.275 | https://aclanthology.org/2022.acl-long.275.pdf | acl-2022-5 | ['language-acquisition'] | ['natural-language-processing'] | [ 3.62019956e-01 3.13676566e-01 -4.18540627e-01 -8.15457523e-01
-1.38819188e-01 -9.21167850e-01 7.16116369e-01 8.16851735e-01
-5.37840366e-01 5.20435750e-01 8.15681100e-01 -6.12185001e-01
1.01934165e-01 -6.56295776e-01 -7.54890978e-01 -2.54749745e-01
1.20457329e-01 5.95980763e-01 1.12949554e-02 -2.41946205e-01
4.68395531e-01 -5.60897626e-02 -1.62955272e+00 8.28971326e-01
4.83964503e-01 7.05099285e-01 5.01321316e-01 4.49782819e-01
-6.92866504e-01 9.72185135e-01 -6.67923033e-01 -1.59025058e-01
-3.46879572e-01 -4.43641245e-01 -9.18197393e-01 -3.39194387e-01
3.87190223e-01 9.53829885e-02 3.57547194e-01 1.09316123e+00
4.13778424e-01 -6.66634440e-02 1.11029589e+00 -1.20337462e+00
-1.59313333e+00 9.64671135e-01 -2.64192164e-01 2.23902822e-01
7.76306570e-01 2.01093584e-01 1.53241229e+00 -7.62778044e-01
1.29394233e+00 1.63383806e+00 5.24169862e-01 7.36712098e-01
-1.41655338e+00 1.63452446e-01 1.33991003e-01 1.17594272e-01
-1.15994155e+00 -1.94266606e-02 5.16278148e-01 -9.64548349e-01
1.37795770e+00 8.34255144e-02 7.44751692e-01 1.15386891e+00
-9.84915197e-02 6.79508567e-01 8.48872364e-01 -8.05488348e-01
4.47625548e-01 -3.47657502e-01 4.87682015e-01 7.39525378e-01
6.86771795e-02 -1.31569952e-01 -4.22214776e-01 -4.32516128e-01
2.45121554e-01 -2.70436615e-01 -1.44564971e-01 -7.36141145e-01
-1.26733291e+00 1.22169125e+00 6.30982041e-01 3.94654989e-01
-1.80759713e-01 5.94393611e-01 7.85756767e-01 3.98775280e-01
6.94461465e-01 5.03159523e-01 -4.34089065e-01 -1.19395360e-01
-4.53290075e-01 1.18638836e-01 6.56693757e-01 8.32053781e-01
5.62040150e-01 -8.42927303e-03 -3.00456136e-01 8.46713305e-01
7.20244169e-01 3.23187768e-01 8.09675515e-01 -4.29800898e-01
-1.43813640e-01 7.32334197e-01 -8.13090652e-02 -6.55275166e-01
-4.29846913e-01 5.07551908e-01 -1.42084181e-01 1.22844264e-01
4.06077445e-01 3.44334096e-01 -1.11088300e+00 1.98813891e+00
3.22885603e-01 1.19239666e-01 2.57814467e-01 9.35227633e-01
1.22052419e+00 5.13375938e-01 5.40427566e-01 1.47843614e-01
1.37297630e+00 -2.80059189e-01 -6.39074087e-01 -9.75674391e-02
1.07665968e+00 -3.21279287e-01 1.88838565e+00 3.18696171e-01
-4.47026730e-01 -1.30631000e-01 -1.13204670e+00 -6.54598176e-01
-8.72689128e-01 -5.28925471e-02 8.67943347e-01 5.93143880e-01
-1.16660714e+00 2.84109950e-01 -4.14729595e-01 -7.98419118e-01
4.25074190e-01 -2.55429596e-01 -3.84655714e-01 7.14745000e-02
-1.19445741e+00 8.47694039e-01 9.63881433e-01 -5.50380230e-01
-9.61393356e-01 -8.95712972e-01 -1.43054068e+00 -1.62403628e-01
1.42154157e-01 -6.80304527e-01 1.11815655e+00 -9.69782114e-01
-1.04353344e+00 1.66028845e+00 -6.71968311e-02 -4.64710802e-01
-2.01412380e-01 -1.38321742e-01 -2.53422111e-01 -7.70971701e-02
2.60619998e-01 7.74728298e-01 4.78963017e-01 -1.09533906e+00
-1.56125709e-01 -5.05885363e-01 2.01117113e-01 -2.21587345e-01
-2.36166835e-01 2.97509646e-03 1.02902003e-01 -8.50706816e-01
-6.47958994e-01 -5.97392082e-01 -6.14980906e-02 2.10525811e-01
-5.12335479e-01 -7.74107575e-01 5.76910436e-01 -3.01875561e-01
1.12602174e+00 -2.25933790e+00 3.19614828e-01 3.89840081e-02
3.75136375e-01 5.78972399e-02 -1.23272069e-01 5.46677649e-01
-3.46360356e-01 4.76073951e-01 -4.76501524e-01 2.03415096e-01
5.35402060e-01 6.63892210e-01 -5.57167649e-01 5.75780213e-01
2.29084551e-01 1.09879017e+00 -1.27341449e+00 -4.27555859e-01
1.21657789e-01 3.32929820e-01 -6.82164490e-01 1.86875775e-01
-1.01327801e+00 -2.07940146e-01 -4.87338573e-01 5.92559755e-01
2.85250813e-01 -4.22730058e-01 5.88968933e-01 1.33701220e-01
-5.65911271e-03 3.19337130e-01 -5.88917792e-01 2.25757837e+00
-1.76638931e-01 7.28786886e-01 -4.76710856e-01 -1.15552199e+00
7.75960445e-01 1.00818142e-01 -1.44277960e-01 -6.21287882e-01
4.78271842e-02 -1.91138327e-01 -1.60736680e-01 -7.54994690e-01
2.98698306e-01 -6.74078941e-01 -6.55272722e-01 6.14602923e-01
3.48403156e-01 -2.50459433e-01 2.28537530e-01 2.96255678e-01
7.69944251e-01 4.50171202e-01 9.16947603e-01 -8.07288766e-01
1.95454940e-01 3.79200608e-01 2.61922926e-01 4.11210537e-01
2.12072451e-02 3.45443845e-01 9.27861929e-01 -7.28895307e-01
-9.57637310e-01 -1.33370280e+00 -1.86189041e-01 1.71946359e+00
1.23554841e-01 -9.77161348e-01 -6.96570456e-01 -6.79035127e-01
3.15378696e-01 1.27746034e+00 -1.12149894e+00 -2.04826534e-01
-1.83624744e-01 -2.93548942e-01 1.06430590e+00 5.83620548e-01
-6.04923785e-01 -1.41982245e+00 -8.79224002e-01 -2.44506687e-01
1.11625709e-01 -5.76024473e-01 -3.88706535e-01 4.34780598e-01
-2.67332137e-01 -1.13023937e+00 -1.49088979e-01 -8.90022218e-01
3.91520858e-01 -2.93170899e-01 1.75329351e+00 1.80595577e-01
-6.74622178e-01 2.51914859e-01 -4.02293384e-01 -8.76590014e-01
-6.08741939e-01 -3.28429520e-01 -4.25716974e-02 -5.51769197e-01
8.75851333e-01 -5.09610362e-02 -5.73330112e-02 -6.15370989e-01
-1.36721385e+00 1.43355176e-01 -1.05534635e-01 1.07190359e+00
7.86513090e-01 -6.77986681e-01 6.24278665e-01 -1.35089242e+00
9.13113654e-01 -8.23272109e-01 -3.63408148e-01 4.62017775e-01
-3.94067049e-01 6.78213716e-01 6.54534400e-01 -2.74124771e-01
-9.11055267e-01 -5.76977693e-02 -2.35849291e-01 -1.55702993e-01
-4.57838237e-01 1.00133765e+00 -2.28990540e-01 6.21840239e-01
9.92801845e-01 3.36059742e-02 7.83988312e-02 -4.88668114e-01
1.27371228e+00 6.07552469e-01 3.93585712e-01 -7.68952131e-01
8.78839567e-02 4.19496566e-01 -1.16670862e-01 -1.09865832e+00
-9.25459921e-01 -4.59759086e-01 -8.19005072e-01 1.95670933e-01
1.14625478e+00 -6.51828706e-01 -4.84299958e-01 -1.03878334e-01
-9.87183809e-01 -3.77151519e-01 -4.95261282e-01 7.23458529e-02
-8.90558481e-01 4.69969176e-02 -3.69380355e-01 -4.28024173e-01
-9.45867524e-02 -6.73457980e-01 1.49770737e+00 -4.24738765e-01
-7.85629869e-01 -1.52244878e+00 4.05256003e-01 -2.55859941e-01
6.90055490e-02 6.67826235e-01 1.56999993e+00 -8.74780834e-01
2.67446637e-01 5.17056942e-01 -1.24852449e-01 -1.29734740e-01
2.51255870e-01 8.99626464e-02 -8.87265623e-01 -1.16677865e-01
-2.96916366e-01 -1.12720156e+00 1.15005362e+00 2.86663473e-01
1.18899715e+00 -1.86903611e-01 -4.04648483e-01 6.17653966e-01
1.57661617e+00 -9.90318805e-02 4.69164073e-01 3.33789408e-01
7.78094411e-01 7.22578049e-01 6.82120025e-01 4.76537943e-01
4.35003102e-01 2.47103259e-01 5.48478484e-01 -1.37299180e-01
-9.28217843e-02 -6.93341017e-01 2.47092977e-01 2.43268937e-01
5.08885264e-01 -5.39745986e-01 -1.21261299e+00 7.28230298e-01
-1.80630815e+00 -9.46354866e-01 1.18435010e-01 1.64074445e+00
1.01351178e+00 -3.98588449e-01 1.96127981e-01 -3.05133134e-01
5.33041954e-01 1.67954162e-01 -6.63097024e-01 -7.70989001e-01
-2.19750375e-01 1.80575177e-01 2.73209065e-02 2.76526898e-01
-9.78520930e-01 1.18431258e+00 7.22430849e+00 6.38113976e-01
-1.10268235e+00 1.29266724e-01 3.82363945e-01 1.47363031e-02
-1.08484638e+00 -1.34546697e-01 -2.20741406e-01 2.93281674e-01
8.63432884e-01 -3.49387079e-01 3.76207568e-02 1.04899096e+00
-2.29153454e-01 2.26605371e-01 -1.52095282e+00 1.04906070e+00
3.87527466e-01 -1.54760659e+00 5.17270207e-01 -1.47526532e-01
-5.75956665e-02 2.21407786e-01 -2.29362287e-02 2.64471266e-02
8.36372614e-01 -1.67503905e+00 1.14188576e+00 4.96840984e-01
1.25638449e+00 -6.63708866e-01 4.20725077e-01 2.28725180e-01
-9.47884738e-01 1.13986485e-01 -6.35822833e-01 -2.57116407e-01
5.56270033e-02 3.26941490e-01 -9.10473883e-01 9.69352648e-02
6.58536792e-01 1.15752268e+00 -5.99094093e-01 2.75631994e-01
-5.47629595e-01 6.39450073e-01 1.05298363e-01 -4.44181532e-01
2.93756306e-01 1.50875524e-01 3.24917912e-01 1.75502396e+00
1.10627264e-01 -1.22032426e-01 3.66362274e-01 1.15052927e+00
3.91951427e-02 4.26827043e-01 -1.45858049e+00 -6.89455807e-01
1.47616357e-01 9.66061354e-01 -6.50489807e-01 -5.27799368e-01
-4.61542964e-01 9.22238231e-01 9.83670771e-01 9.69663188e-02
-6.28299296e-01 -1.65204495e-01 9.98580933e-01 -2.08933838e-02
3.12152147e-01 6.43196478e-02 -1.23172723e-01 -1.19548666e+00
-3.94267172e-01 -7.58148909e-01 8.97828460e-01 -1.07403076e+00
-1.58952105e+00 2.69158661e-01 9.19019952e-02 -8.06817412e-01
-4.74324167e-01 -1.32142460e+00 -1.99609816e-01 6.96382344e-01
-1.00751710e+00 -1.39606202e+00 2.91871876e-01 4.46389496e-01
2.38364622e-01 -2.05158010e-01 1.28814435e+00 -4.19040054e-01
1.58431411e-01 3.07824641e-01 -1.68575928e-01 -1.38903176e-02
6.55435443e-01 -1.84459674e+00 6.91977203e-01 4.41018552e-01
5.59642315e-01 1.05478847e+00 8.80594075e-01 -6.03792071e-01
-1.50486779e+00 -1.05542982e+00 1.00885332e+00 -7.69665182e-01
8.55954051e-01 -7.34787285e-01 -8.35265756e-01 9.11303818e-01
5.34910977e-01 3.70173417e-02 1.51755738e+00 3.28382254e-01
-8.28246653e-01 8.24342012e-01 -1.22656262e+00 3.78131449e-01
1.38208747e+00 -9.24240947e-01 -1.37265134e+00 1.29612714e-01
1.16415107e+00 -4.49415855e-02 -7.43922412e-01 -3.69351916e-02
4.19181108e-01 -6.15351379e-01 9.41830099e-01 -1.19250655e+00
4.81392473e-01 -6.36963248e-01 -5.91179669e-01 -1.83819890e+00
-3.44714701e-01 -2.12416083e-01 1.66059777e-01 9.74593163e-01
4.07516479e-01 -2.60758340e-01 2.42098942e-01 -4.74159084e-02
-2.50625044e-01 -2.76666671e-01 -6.69454634e-01 -5.82045376e-01
5.44251084e-01 -6.89414859e-01 7.99788892e-01 1.42515802e+00
4.29923058e-01 7.62408257e-01 1.07281413e-02 -3.70619982e-01
4.56238270e-01 3.09756875e-01 3.50908369e-01 -1.35440540e+00
-2.58996248e-01 -3.71040642e-01 -8.37018728e-01 -6.68886304e-01
8.14502180e-01 -1.74693227e+00 1.10136345e-01 -1.66503274e+00
4.16706681e-01 6.46402314e-02 -3.93767208e-01 8.40007603e-01
-1.07911743e-01 1.68714389e-01 -8.93695876e-02 -3.68502647e-01
-7.39249647e-01 4.39761549e-01 8.08224499e-01 -3.49651217e-01
3.13145965e-01 -1.25748587e+00 -1.02996624e+00 9.07330871e-01
6.66282535e-01 -4.80601907e-01 -8.79455626e-01 -6.91921055e-01
7.61200547e-01 -2.82504737e-01 3.69043231e-01 2.77534649e-02
-2.13838413e-01 -5.13474762e-01 3.80636245e-01 -4.05069262e-01
-1.76321507e-01 -4.32015270e-01 -3.47753167e-02 4.54228610e-01
-7.83055127e-01 2.22649619e-01 2.03740984e-01 6.01166487e-01
-1.48770645e-01 -1.49806172e-01 4.41767126e-01 -2.14384407e-01
-1.47789812e+00 -1.91956341e-01 -4.46150780e-01 5.85060179e-01
9.46936607e-01 -7.83610418e-02 -5.91297090e-01 -1.15831286e-01
-8.10546756e-01 1.79350838e-01 1.01686370e+00 7.90707707e-01
8.81263196e-01 -1.47071826e+00 -6.09461129e-01 2.28558883e-01
9.97170150e-01 -5.85290432e-01 -3.67336750e-01 2.37450097e-03
-5.19031167e-01 3.19977999e-01 -2.98764616e-01 -6.19727135e-01
-1.25518167e+00 1.11276650e+00 -1.12088621e-01 3.68946701e-01
-6.38366103e-01 8.12846661e-01 6.96549177e-01 -8.79357636e-01
-1.48393929e-01 -7.43521690e-01 -6.51342869e-01 4.26118225e-02
8.35880458e-01 -1.70654997e-01 -4.26418573e-01 -9.27360654e-01
-6.85157061e-01 3.12140375e-01 4.15464431e-01 7.51433102e-03
1.42031682e+00 4.27318811e-02 -6.78650141e-01 1.28200114e+00
1.30576766e+00 -1.71520263e-01 -7.31570363e-01 -1.26285208e-02
7.38935351e-01 -3.14191490e-01 -3.07493836e-01 -8.00639749e-01
-3.55627865e-01 8.48246276e-01 4.85752076e-01 2.26503536e-01
8.12404811e-01 7.48860359e-01 2.81459391e-01 4.75648552e-01
2.15633869e-01 -9.99523878e-01 2.50936389e-01 8.42804670e-01
9.45984304e-01 -1.18787944e+00 -3.79796505e-01 -2.04367578e-01
-9.11246002e-01 1.05582619e+00 3.75916481e-01 -3.67879346e-02
6.42116010e-01 5.38741708e-01 2.92297363e-01 -8.00754130e-01
-1.11969626e+00 -6.28585458e-01 4.58823442e-01 1.01844537e+00
1.08811152e+00 4.58377421e-01 -4.05101657e-01 7.82468855e-01
-4.04968113e-01 1.85909584e-01 3.85037631e-01 8.96868050e-01
-5.21966159e-01 -1.03148758e+00 2.31431097e-01 4.16333169e-01
-4.99191701e-01 -5.72630465e-01 -9.43835378e-01 5.28638959e-01
2.47123048e-01 8.74146998e-01 3.43796223e-01 -6.56057596e-02
1.62605032e-01 3.90430123e-01 4.66975003e-01 -1.21438706e+00
-1.93020955e-01 -2.29779214e-01 3.70084018e-01 -7.30056763e-01
-6.08283818e-01 -5.16301870e-01 -2.01146078e+00 -1.79193411e-02
2.01483831e-01 8.34505484e-02 2.69413948e-01 7.75512993e-01
3.05048704e-01 4.24087822e-01 -2.44532689e-01 -2.34151825e-01
-3.50756258e-01 -8.41286838e-01 -7.04297066e-01 1.19671190e+00
2.27844536e-01 -6.99589729e-01 -1.64681271e-01 5.16858637e-01] | [10.571187019348145, 2.2932794094085693] |
bb4a6f60-12bc-4b25-b5b5-f5e1713f75c4 | fast-and-high-quality-singing-voice-synthesis | 1910.11690 | null | https://arxiv.org/abs/1910.11690v2 | https://arxiv.org/pdf/1910.11690v2.pdf | Fast and High-Quality Singing Voice Synthesis System based on Convolutional Neural Networks | The present paper describes singing voice synthesis based on convolutional neural networks (CNNs). Singing voice synthesis systems based on deep neural networks (DNNs) are currently being proposed and are improving the naturalness of synthesized singing voices. As singing voices represent a rich form of expression, a powerful technique to model them accurately is required. In the proposed technique, long-term dependencies of singing voices are modeled by CNNs. An acoustic feature sequence is generated for each segment that consists of long-term frames, and a natural trajectory is obtained without the parameter generation algorithm. Furthermore, a computational complexity reduction technique, which drives the DNNs in different time units depending on type of musical score features, is proposed. Experimental results show that the proposed method can synthesize natural sounding singing voices much faster than the conventional method. | ['Keiichi Tokuda', 'Yoshihiko Nankaku', 'Keiichiro Oura', 'Kei Hashimoto', 'Kazuhiro Nakamura', 'Shinji Takaki'] | 2019-10-24 | null | null | null | null | ['singing-voice-synthesis'] | ['speech'] | [-1.45313561e-01 -3.52310181e-01 8.05506632e-02 -2.10178327e-02
-2.78065890e-01 -4.66843992e-01 1.04359657e-01 -8.43508542e-01
-9.59764943e-02 4.43614542e-01 2.15489626e-01 6.70827627e-02
1.73078135e-01 -7.30295658e-01 -5.50923169e-01 -7.26727486e-01
2.03973223e-02 -9.60941166e-02 2.17795577e-02 -3.85988057e-01
-1.46058217e-01 6.15054846e-01 -1.83092642e+00 2.41314322e-01
5.63333571e-01 9.47452188e-01 3.14792246e-01 1.30717969e+00
-1.60642967e-01 9.41662669e-01 -1.07653248e+00 1.39920413e-01
1.07707478e-01 -1.06275594e+00 -3.49939704e-01 -2.66776290e-02
4.47204083e-01 -4.84956473e-01 -4.53249335e-01 9.15766418e-01
1.00200081e+00 4.12968099e-01 3.57770741e-01 -8.32839251e-01
-5.28525114e-01 9.22181368e-01 2.75705546e-01 -3.33749056e-02
-1.45764761e-02 5.39016724e-01 1.18259132e+00 -1.04055727e+00
3.51072550e-01 1.15986812e+00 8.03740203e-01 1.13101387e+00
-6.41523600e-01 -1.01002288e+00 -5.67617953e-01 2.53406674e-01
-1.19603658e+00 -6.95400178e-01 1.21918488e+00 -1.97274446e-01
7.94416964e-01 4.89386439e-01 1.04158354e+00 8.28884065e-01
-1.63693443e-01 9.64645922e-01 3.83383304e-01 -4.77771640e-01
-2.56204829e-02 -3.82276237e-01 -4.82942432e-01 4.43968087e-01
-6.48722649e-01 4.31303948e-01 -7.72307515e-01 1.32954776e-01
1.13907790e+00 -4.16345119e-01 -3.09387296e-01 5.14760792e-01
-1.15303528e+00 7.57940710e-01 4.31785405e-01 7.32631028e-01
-4.94592428e-01 4.92871255e-01 7.18029499e-01 2.82360405e-01
1.68364584e-01 7.10790217e-01 -1.00897618e-01 -4.33670640e-01
-1.36613905e+00 6.34763896e-01 8.11988354e-01 5.89562953e-01
2.62888963e-03 1.10580194e+00 -3.12641472e-01 1.18665254e+00
-9.03751478e-02 3.00495118e-01 8.23914826e-01 -1.19681346e+00
1.36112303e-01 -5.52519551e-03 -7.32573420e-02 -9.79970455e-01
-3.28343600e-01 -6.49767160e-01 -9.71321046e-01 7.13892356e-02
2.16359407e-01 -5.28883994e-01 -5.80462575e-01 1.76384735e+00
5.70751615e-02 5.16859353e-01 6.49547204e-02 1.10300779e+00
1.15161633e+00 1.27073753e+00 -4.48473424e-01 -4.83925998e-01
8.75612855e-01 -1.35748839e+00 -1.47317183e+00 2.99099207e-01
-8.04665536e-02 -8.29606295e-01 1.21905613e+00 5.04198074e-01
-1.41360664e+00 -1.25071478e+00 -1.03448534e+00 -1.35217816e-01
7.61710629e-02 3.75454336e-01 3.04983675e-01 3.63957971e-01
-1.10602963e+00 1.12923813e+00 -5.19955337e-01 2.52795845e-01
1.41395424e-02 4.13620085e-01 7.69972876e-02 7.59314716e-01
-1.38264179e+00 3.30534637e-01 3.10373962e-01 5.60907423e-01
-1.16374719e+00 -7.12101996e-01 -5.18530428e-01 2.83229232e-01
-1.19989768e-01 -6.32010221e-01 1.83589375e+00 -1.10580087e+00
-2.32333183e+00 1.69328883e-01 -1.36941969e-01 -4.42380995e-01
2.01278895e-01 -2.47397795e-01 -7.15011120e-01 1.08299688e-01
-4.02485788e-01 5.16947925e-01 1.29230118e+00 -8.24496388e-01
-5.03554225e-01 2.24957749e-01 -3.25390309e-01 1.78557143e-01
-6.38796926e-01 3.17522377e-01 -3.61492448e-02 -9.01291192e-01
1.35782489e-03 -8.19483101e-01 -9.35298651e-02 -1.54037595e-01
-5.05580187e-01 -4.29244906e-01 9.75823045e-01 -6.50473356e-01
1.71997523e+00 -2.10428596e+00 2.67195970e-01 -2.49090180e-01
6.31013364e-02 8.52583289e-01 -2.29966626e-01 4.43873763e-01
-9.14769396e-02 -1.10938914e-01 -2.67620116e-01 -7.82517046e-02
-1.95728317e-02 1.12723291e-01 -5.53616405e-01 -5.87456627e-03
1.87235802e-01 9.09022570e-01 -7.06513882e-01 -2.43560344e-01
2.36187339e-01 5.66033065e-01 -5.90900242e-01 8.82335961e-01
-3.35137993e-01 6.55751526e-01 -8.24330933e-03 2.51104593e-01
3.17885667e-01 4.12564695e-01 -5.86351119e-02 -1.82277009e-01
-4.73075807e-01 5.70452213e-01 -9.78725433e-01 1.56534159e+00
-9.30492699e-01 7.74535298e-01 2.66863018e-01 -7.50998080e-01
1.38059497e+00 9.65234518e-01 1.80714756e-01 -1.55227318e-01
2.68990099e-01 5.20638108e-01 4.94501054e-01 -7.89142847e-01
5.07240176e-01 -7.02325761e-01 2.12872699e-01 2.72664189e-01
3.09008777e-01 -7.18039572e-01 2.25437973e-02 -7.00621068e-01
7.27932513e-01 9.70008895e-02 -6.35250732e-02 8.54247361e-02
6.65672779e-01 -3.55258107e-01 8.36679399e-01 3.17864180e-01
-1.81935336e-02 7.84177959e-01 1.26622900e-01 -7.41029203e-01
-1.31685865e+00 -6.85345411e-01 2.95495931e-02 9.79744554e-01
-4.82896745e-01 -2.58263767e-01 -1.02197826e+00 1.55976877e-01
-3.38135570e-01 6.45941198e-01 -1.57275885e-01 -1.25467241e-01
-1.11979139e+00 -8.91228318e-02 1.08543921e+00 6.45870209e-01
4.53522086e-01 -2.03098249e+00 -3.88914108e-01 8.73539329e-01
-1.50652513e-01 -8.60613227e-01 -9.18094814e-01 -7.01528862e-02
-9.54786658e-01 -5.32127142e-01 -1.10076213e+00 -1.13212347e+00
-1.34487152e-01 -1.27649188e-01 7.73261070e-01 1.35360748e-01
-1.13694482e-01 -2.71224439e-01 -1.96040809e-01 -4.15479302e-01
-7.91445017e-01 1.17322572e-01 5.84240735e-01 1.72037497e-01
4.38534655e-02 -1.03668106e+00 -5.18735230e-01 -9.94640812e-02
-8.94937098e-01 -1.64269265e-02 2.84146249e-01 1.05795300e+00
5.37502587e-01 1.13667876e-01 1.20495892e+00 -3.57071072e-01
1.15046191e+00 -1.28168821e-01 -4.22247976e-01 -2.12564126e-01
-3.05885058e-02 -3.11543047e-01 1.39135695e+00 -8.19034755e-01
-8.94167900e-01 2.60864235e-02 -7.00341105e-01 -1.02909613e+00
-2.19008908e-01 2.51851857e-01 1.79051328e-02 2.97587484e-01
4.97849584e-01 4.40582693e-01 1.92797124e-01 -7.00400949e-01
3.50888878e-01 1.10920250e+00 9.30481672e-01 -3.02253991e-01
8.19988370e-01 -8.68392512e-02 -1.17811330e-01 -1.34114742e+00
-6.23890817e-01 -1.54316738e-01 -5.94581485e-01 -5.11353910e-01
8.19923162e-01 -6.09008610e-01 -1.05750322e+00 8.88898969e-01
-1.55608916e+00 -2.78564990e-01 -5.41143000e-01 7.82331645e-01
-8.11191022e-01 8.50148872e-02 -1.03949118e+00 -1.08184874e+00
-7.56775618e-01 -1.02942729e+00 5.30837238e-01 4.87791628e-01
-2.39968523e-01 -7.57738173e-01 2.41635472e-01 1.87479258e-01
6.53049648e-01 1.45976156e-01 7.02839196e-01 -3.17037672e-01
-2.30325103e-01 -1.23642907e-01 4.07912165e-01 1.04041028e+00
2.69888252e-01 3.52237374e-01 -1.33512485e+00 1.72386896e-02
1.96312115e-01 -2.72360504e-01 5.73412120e-01 7.23901331e-01
1.45823610e+00 -5.87246656e-01 5.89320958e-01 7.01319337e-01
7.07726538e-01 5.86648881e-01 5.50571144e-01 -3.26742470e-01
8.84864032e-01 5.91846526e-01 3.22941035e-01 3.56087714e-01
-1.08782619e-01 5.59394956e-01 3.33138138e-01 -1.77147731e-01
-5.72615743e-01 -4.84444529e-01 5.57386816e-01 1.97232473e+00
-4.14798796e-01 -1.28132686e-01 -2.39938229e-01 6.84848130e-01
-1.43487275e+00 -1.26714098e+00 -1.54327482e-01 1.90645945e+00
9.69397366e-01 -2.95209557e-01 3.30825627e-01 6.41333640e-01
9.10714388e-01 5.15511394e-01 -8.05423558e-01 -9.45728004e-01
-2.07227282e-02 7.54176378e-01 -2.72703826e-01 3.97994310e-01
-8.51735473e-01 1.17950344e+00 6.64298248e+00 9.94299173e-01
-1.60700262e+00 -2.10928872e-01 1.13130026e-01 -3.59334916e-01
-1.40646517e-01 -5.12987435e-01 -5.15135467e-01 2.71069735e-01
1.15056980e+00 -1.58531725e-01 8.62400115e-01 8.04965973e-01
6.96563005e-01 8.28292370e-01 -7.49929726e-01 1.01800668e+00
-2.66771968e-02 -1.53183270e+00 1.19956158e-01 -4.25986469e-01
7.60928988e-01 -5.34247398e-01 2.62806535e-01 2.64797866e-01
-2.71465808e-01 -1.22225666e+00 8.34111333e-01 6.16414607e-01
1.14877617e+00 -1.20469809e+00 5.10837018e-01 4.71350461e-01
-1.44154561e+00 -2.64129341e-01 -3.40395182e-01 -3.42161566e-01
3.29204977e-01 3.36937934e-01 -1.04913712e+00 1.90047652e-01
5.15411913e-01 6.57178164e-01 2.05552414e-01 9.11173820e-01
-3.43484223e-01 1.34905088e+00 -1.03227876e-01 -4.28430766e-01
2.50829518e-01 -9.49878097e-02 8.04107487e-01 1.15713573e+00
7.24705637e-01 6.37406483e-02 -3.34120840e-01 1.22312427e+00
-2.51545489e-01 3.00069183e-01 -5.65630555e-01 -5.83318949e-01
4.23777878e-01 1.29122496e+00 -3.26956697e-02 -2.30935976e-01
-7.26681426e-02 7.01465845e-01 -3.22718590e-01 2.27548808e-01
-5.74497104e-01 -8.19944143e-01 7.21798956e-01 1.48625270e-01
2.75575489e-01 -4.15564358e-01 -4.92352471e-02 -7.71138847e-01
-1.29327819e-01 -1.02850771e+00 -2.30618760e-01 -8.98022056e-01
-1.06985331e+00 9.81923640e-01 -7.33017743e-01 -1.58659196e+00
-7.20830500e-01 -4.06982243e-01 -1.16238511e+00 9.52555656e-01
-1.38284087e+00 -9.17221963e-01 -5.02816252e-02 5.08357346e-01
9.91734862e-01 -3.91769141e-01 9.97428596e-01 4.02071357e-01
-4.40195799e-01 5.22521794e-01 -1.06454417e-01 2.23577052e-01
3.85572761e-01 -1.09722519e+00 5.56340337e-01 5.30233979e-01
3.31636876e-01 4.23508912e-01 6.98481560e-01 -8.05466548e-02
-1.16991568e+00 -9.87668037e-01 1.06128716e+00 3.83415222e-01
5.17274082e-01 -1.93771675e-01 -9.87618268e-01 5.93114533e-02
5.02661765e-01 7.68281100e-03 6.71237946e-01 -2.74560571e-01
6.64969757e-02 -2.62905151e-01 -7.01715171e-01 5.20606518e-01
7.95470357e-01 -7.53726244e-01 -5.74203551e-01 6.99877292e-02
9.15348947e-01 -4.30468500e-01 -6.83911979e-01 4.23252612e-01
7.40925968e-01 -1.05888748e+00 6.89884305e-01 -6.90327823e-01
6.01138473e-01 -3.98374915e-01 1.79981172e-01 -1.50740623e+00
-2.17416078e-01 -1.11009586e+00 -3.33296090e-01 1.15290964e+00
2.46197060e-01 1.26080234e-02 4.29162294e-01 -1.46585181e-01
-5.18972158e-01 -4.92748648e-01 -8.43739092e-01 -9.22445834e-01
1.95874617e-01 -4.53642368e-01 9.80895340e-01 8.57687712e-01
-2.23129854e-01 4.69413072e-01 -8.82510483e-01 -3.99061367e-02
2.38233194e-01 2.72286028e-01 7.23479569e-01 -1.16702950e+00
-4.05345857e-01 -4.20671552e-01 -1.07458845e-01 -1.19285417e+00
3.70500743e-01 -7.27529287e-01 4.04480606e-01 -1.17374980e+00
-4.86616492e-01 2.15471219e-02 -2.29891270e-01 2.06554711e-01
-2.67111696e-02 2.41330683e-01 3.76131862e-01 -4.43274863e-02
1.20578408e-01 9.59809184e-01 1.95375192e+00 2.80538984e-02
-5.23544192e-01 5.04009008e-01 1.00019060e-01 9.06461656e-01
1.03187525e+00 -1.93620101e-01 -2.99573153e-01 -2.56673157e-01
-3.01825523e-01 8.22238207e-01 1.53801873e-01 -1.41066134e+00
4.02669698e-01 5.13081849e-02 6.41467273e-02 -8.54866803e-01
8.12214136e-01 -4.36299443e-01 5.42649142e-02 4.50436831e-01
-6.84590518e-01 -2.62502670e-01 7.09055588e-02 -2.94693913e-02
-7.63996720e-01 -3.61535698e-01 9.70194280e-01 -9.96550396e-02
-2.28752047e-01 4.45193589e-01 -4.74786192e-01 -2.79839188e-01
3.69197994e-01 7.39133358e-02 4.49424505e-01 -8.58539224e-01
-9.44880664e-01 -4.93161887e-01 -4.46643323e-01 5.73308885e-01
8.08429062e-01 -1.85948515e+00 -7.77010858e-01 2.70825177e-01
-4.25413460e-01 3.40706967e-02 5.72849274e-01 2.62712300e-01
-6.46536291e-01 4.30875301e-01 -2.24823862e-01 -2.91392207e-01
-1.21464872e+00 2.68163949e-01 6.28412187e-01 1.05647817e-01
-6.22959852e-01 1.00334013e+00 6.84780674e-03 -7.02189803e-01
2.87465066e-01 -3.66520703e-01 -5.56133151e-01 -1.10836783e-02
6.43195868e-01 5.11627734e-01 -2.49291688e-01 -7.54742384e-01
2.30870754e-01 5.54902852e-01 4.68613207e-01 -1.57603949e-01
1.38100588e+00 2.40990564e-01 -1.53311074e-01 7.81389594e-01
1.07135749e+00 1.49246052e-01 -1.15846026e+00 3.81934978e-02
-2.86761910e-01 -1.66867569e-01 2.69581825e-01 -4.32953656e-01
-1.37215948e+00 1.45647466e+00 3.54461133e-01 3.66724938e-01
1.22357166e+00 -4.91081387e-01 1.58060002e+00 2.52960980e-01
-8.08592364e-02 -1.09923053e+00 1.93059385e-01 7.12235689e-01
1.32054615e+00 -7.08016574e-01 -7.93942869e-01 1.58020034e-01
-5.84862649e-01 1.73435211e+00 6.26424551e-01 -5.82167268e-01
5.01654744e-01 3.47369015e-01 4.42344576e-01 1.51383400e-01
-7.73366332e-01 -6.66891560e-02 2.12028593e-01 3.73708278e-01
6.59789145e-01 1.85373098e-01 -3.93562943e-01 1.06866848e+00
-8.11891913e-01 1.87527373e-01 4.26462293e-01 5.66880777e-02
-5.50432742e-01 -1.04490685e+00 -4.58853006e-01 8.11259449e-02
-7.21915960e-01 -2.91732490e-01 -3.50134313e-01 3.02974343e-01
2.81020045e-01 1.02567363e+00 1.04584865e-01 -8.81993473e-01
4.68171924e-01 2.10356385e-01 2.77723908e-01 -5.16133547e-01
-9.60564613e-01 4.02762294e-01 -4.09412049e-02 -2.07876131e-01
-2.31540293e-01 -1.96983606e-01 -1.47130191e+00 -9.94785577e-02
-3.89070094e-01 3.65428090e-01 7.47657895e-01 6.11733258e-01
1.24140985e-01 9.95267272e-01 1.33155668e+00 -1.08276749e+00
-6.53374851e-01 -1.22634649e+00 -9.44087446e-01 -1.46157388e-02
6.77597880e-01 -1.52102441e-01 -5.95059097e-01 1.90573692e-01] | [15.543153762817383, 6.185049057006836] |
21079fbf-459a-4574-9dda-938112f57cff | a-lightweight-instrument-agnostic-model-for | 2203.09893 | null | https://arxiv.org/abs/2203.09893v2 | https://arxiv.org/pdf/2203.09893v2.pdf | A Lightweight Instrument-Agnostic Model for Polyphonic Note Transcription and Multipitch Estimation | Automatic Music Transcription (AMT) has been recognized as a key enabling technology with a wide range of applications. Given the task's complexity, best results have typically been reported for systems focusing on specific settings, e.g. instrument-specific systems tend to yield improved results over instrument-agnostic methods. Similarly, higher accuracy can be obtained when only estimating frame-wise $f_0$ values and neglecting the harder note event detection. Despite their high accuracy, such specialized systems often cannot be deployed in the real-world. Storage and network constraints prohibit the use of multiple specialized models, while memory and run-time constraints limit their complexity. In this paper, we propose a lightweight neural network for musical instrument transcription, which supports polyphonic outputs and generalizes to a wide variety of instruments (including vocals). Our model is trained to jointly predict frame-wise onsets, multipitch and note activations, and we experimentally show that this multi-output structure improves the resulting frame-level note accuracy. Despite its simplicity, benchmark results show our system's note estimation to be substantially better than a comparable baseline, and its frame-level accuracy to be only marginally below those of specialized state-of-the-art AMT systems. With this work we hope to encourage the community to further investigate low-resource, instrument-agnostic AMT systems. | ['Sebastian Ewert', 'Gabriel Meseguer-Brocal', 'David Rubinstein', 'Juan José Bosch', 'Rachel M. Bittner'] | 2022-03-18 | null | null | null | null | ['music-transcription'] | ['music'] | [ 4.04372692e-01 -3.95671666e-01 -1.81067392e-01 -6.19333461e-02
-1.25368416e+00 -8.33597481e-01 2.18221724e-01 -1.84613734e-01
-2.42070019e-01 2.99548507e-01 1.25811696e-01 1.09970868e-01
8.71959925e-02 -2.17416272e-01 -3.79142642e-01 -5.54403007e-01
7.42959278e-03 -3.28079164e-02 -1.18921988e-03 -5.42973690e-02
2.20627144e-01 3.17727149e-01 -1.75080526e+00 4.13016260e-01
4.79218632e-01 1.15566993e+00 2.62928754e-01 1.00680852e+00
8.24705139e-02 5.66295683e-01 -9.39063847e-01 -3.21651548e-01
2.81280011e-01 -6.53469563e-01 -4.59260017e-01 -2.00789541e-01
8.14146221e-01 -9.65818241e-02 -1.03625469e-01 7.53893971e-01
1.06511557e+00 2.93381691e-01 3.40295374e-01 -9.28171456e-01
-3.31029624e-01 8.01465571e-01 -3.27152401e-01 1.21046387e-01
2.57689655e-01 2.48304307e-01 1.27066135e+00 -8.72730851e-01
3.10731113e-01 9.64664340e-01 9.90831614e-01 4.03618664e-01
-1.23927879e+00 -7.91885734e-01 1.29683271e-01 8.69516879e-02
-1.41210759e+00 -7.64466226e-01 9.01669145e-01 -1.05237231e-01
1.18884075e+00 4.52671140e-01 7.05218732e-01 1.13207555e+00
-8.57185200e-02 9.80613589e-01 7.14055240e-01 -4.08446938e-01
1.50684282e-01 -4.26719487e-01 -3.46606970e-01 3.54929984e-01
-2.52567917e-01 -1.74568012e-01 -1.05160117e+00 5.08750342e-02
9.31123137e-01 -2.35155985e-01 -2.99989939e-01 1.59967795e-01
-1.26543772e+00 4.32482600e-01 1.99066073e-01 4.40032393e-01
-3.98798078e-01 4.87032920e-01 8.36138904e-01 2.95322865e-01
3.44199896e-01 7.96129167e-01 -4.75560218e-01 -8.32479298e-01
-1.59578228e+00 4.32781160e-01 7.17660189e-01 8.61469567e-01
7.29138330e-02 5.53248584e-01 -4.74177331e-01 1.22222364e+00
-3.13084811e-01 2.19482183e-01 6.38765395e-01 -1.24391317e+00
3.16088378e-01 1.30243257e-01 -1.76797993e-02 -7.33338773e-01
-3.97062480e-01 -8.93922567e-01 -7.94341147e-01 5.49802482e-02
4.47900355e-01 -3.92586850e-02 -6.17798626e-01 1.98678052e+00
-9.38436687e-02 3.29999804e-01 -2.82908767e-01 1.08607006e+00
5.19712687e-01 6.75594985e-01 -1.89934447e-01 -4.19902474e-01
1.38828111e+00 -9.23366845e-01 -8.02259982e-01 -1.83452681e-01
2.40592957e-01 -1.25918639e+00 1.36998248e+00 8.96594048e-01
-1.45347166e+00 -8.58281374e-01 -9.35175657e-01 -1.55869946e-01
2.38553256e-01 5.53751290e-01 8.65314901e-01 6.22578681e-01
-7.81853080e-01 8.52781653e-01 -7.77270555e-01 -2.07598835e-01
2.11799324e-01 2.90550679e-01 1.31509185e-01 4.62237954e-01
-8.81141305e-01 4.28451478e-01 1.61796555e-01 1.87853903e-01
-7.48290002e-01 -7.93215156e-01 -4.43488866e-01 2.36734480e-01
4.42424595e-01 -5.69089711e-01 1.83985889e+00 -8.40498924e-01
-1.91985297e+00 6.40517473e-01 -2.88698912e-01 -5.54873884e-01
2.40924582e-01 -4.90451545e-01 -5.46416700e-01 1.54193878e-01
-1.38650328e-01 6.13755584e-01 1.08653688e+00 -7.40926743e-01
-6.95934415e-01 6.24806955e-02 -3.83695364e-02 2.74238795e-01
-5.72450280e-01 1.18770942e-01 -6.98484898e-01 -1.26472068e+00
5.88700622e-02 -1.24675071e+00 8.27406347e-02 -1.86795354e-01
-4.14929539e-01 -1.63982019e-01 5.56327581e-01 -5.73403955e-01
1.77853751e+00 -2.36581540e+00 1.08381413e-01 -2.91197568e-01
-3.11513305e-01 2.60472655e-01 -1.28141671e-01 3.60877156e-01
-1.23637244e-02 -1.27774447e-01 -2.31815383e-01 -4.72584665e-01
2.25778580e-01 -3.02214324e-02 -4.64264125e-01 1.07283875e-01
4.55143973e-02 6.91216171e-01 -6.99842870e-01 -3.06959748e-01
1.61644489e-01 4.28176999e-01 -8.07979107e-01 1.21835880e-01
-3.08456153e-01 4.01829660e-01 1.42697282e-02 8.59755933e-01
1.38629720e-01 3.07712071e-02 1.98007256e-01 -4.07554775e-01
-1.80699214e-01 7.22235918e-01 -1.23743379e+00 2.15044665e+00
-7.95008481e-01 8.58767629e-01 1.41757697e-01 -7.21187234e-01
8.06504905e-01 7.24911809e-01 4.11958486e-01 -3.55233788e-01
4.03904654e-02 3.73879999e-01 3.99481982e-01 -4.04941946e-01
7.31080413e-01 -1.17323667e-01 -2.16531441e-01 4.62965816e-01
1.96789816e-01 -4.45039988e-01 2.59068310e-01 -1.54115155e-01
1.08207619e+00 2.85862148e-01 1.65869132e-01 5.29891700e-02
9.96190980e-02 -3.78457397e-01 4.79368359e-01 6.98135793e-01
-2.34464660e-01 9.59118724e-01 1.70174390e-01 -1.53754696e-01
-9.23388422e-01 -1.02773261e+00 -5.17830066e-03 1.55112207e+00
-2.75717199e-01 -7.49323666e-01 -7.39117146e-01 -2.12347522e-01
-1.98322535e-01 5.42428434e-01 -1.75616845e-01 -1.41335100e-01
-8.89432669e-01 -4.41971898e-01 1.06433022e+00 6.38476133e-01
4.47670996e-01 -1.19849420e+00 -6.36047125e-01 5.46838462e-01
-4.21111733e-01 -9.79766488e-01 -8.93593550e-01 2.19049901e-01
-9.64191616e-01 -6.07960045e-01 -8.71593952e-01 -6.77176297e-01
-1.24348961e-01 4.51295003e-02 1.14468980e+00 -2.56803513e-01
-2.65739053e-01 1.97404489e-01 -2.74079740e-01 -5.69050729e-01
-1.02494918e-01 4.58696127e-01 3.98381263e-01 6.58276007e-02
4.09301035e-02 -1.05642807e+00 -8.70032728e-01 2.21718088e-01
-7.79462457e-01 -1.24009356e-01 6.37662053e-01 7.78853536e-01
5.73606312e-01 -8.30729753e-02 9.24073339e-01 -5.47442138e-01
8.04698884e-01 9.91240069e-02 -1.35034129e-01 -1.17754549e-01
-4.84659493e-01 -2.38776565e-01 7.07978189e-01 -1.01349843e+00
-8.38026702e-01 4.78986092e-02 -2.66914576e-01 -6.33196115e-01
-3.72503176e-02 4.01290298e-01 -1.79677308e-02 1.86394960e-01
7.10038483e-01 1.73425898e-01 -4.51497138e-01 -7.10815012e-01
2.22127885e-01 7.32026041e-01 1.04552090e+00 -7.97216475e-01
5.35105944e-01 4.77617644e-02 -8.38384181e-02 -8.07625651e-01
-1.01559019e+00 -4.82813358e-01 -2.98233300e-01 -3.94375980e-01
3.97152036e-01 -1.00125766e+00 -9.08251166e-01 3.65884602e-01
-1.00116694e+00 -2.77675390e-01 -3.67345273e-01 6.41442239e-01
-7.61813581e-01 1.28961593e-01 -8.79137456e-01 -1.11452031e+00
-5.19143820e-01 -9.43986773e-01 1.23324442e+00 1.02322131e-01
-8.80071700e-01 -5.02560496e-01 -1.01975678e-02 2.58988202e-01
4.70616043e-01 1.22910835e-01 6.39442801e-01 -2.16546685e-01
-4.11492884e-01 -1.78786322e-01 1.65773734e-01 3.99223506e-01
1.64849505e-01 8.51635933e-02 -1.47590077e+00 -1.96525350e-01
-1.27392426e-01 -3.02067578e-01 8.11444819e-01 6.37472391e-01
1.43725729e+00 -2.20511153e-01 4.23866250e-02 6.40797675e-01
9.38339591e-01 8.51118788e-02 4.09602821e-01 6.96287677e-02
5.84777296e-01 2.21793815e-01 6.93161190e-01 6.10405564e-01
-1.15488090e-01 1.16968143e+00 2.39704311e-01 -3.22702639e-02
-5.19648135e-01 -2.96939582e-01 5.04975200e-01 1.25244093e+00
-5.35671413e-01 -2.58059859e-01 -4.43650424e-01 6.54671311e-01
-1.76504326e+00 -1.16105342e+00 7.40672946e-02 2.40918493e+00
1.12926185e+00 2.61077166e-01 5.95582247e-01 5.47917604e-01
6.27103090e-01 2.39412308e-01 -7.51139343e-01 -6.39458120e-01
-1.70107126e-01 6.92187428e-01 7.79004246e-02 -6.15869947e-02
-9.56076443e-01 8.01941931e-01 6.64311266e+00 1.17317688e+00
-1.38033307e+00 6.92420453e-02 1.69108555e-01 -9.29041803e-01
1.80235773e-01 -3.20009470e-01 -5.35023332e-01 3.36397260e-01
1.16842806e+00 -7.47509897e-02 7.23048508e-01 6.94754004e-01
3.41483414e-01 2.28370205e-01 -1.19356477e+00 1.42011499e+00
-5.15092583e-03 -1.19391942e+00 -2.30109617e-01 -8.31270665e-02
6.16545737e-01 -2.59063810e-01 3.13487887e-01 5.74748516e-01
-3.48879308e-01 -9.27302122e-01 1.07724988e+00 3.81064266e-01
1.14230967e+00 -9.11259830e-01 2.46135205e-01 1.03667475e-01
-1.58771777e+00 -1.79279417e-01 -6.92591742e-02 -4.17478621e-01
3.02527100e-01 6.11318290e-01 -7.62368441e-01 3.65943223e-01
8.45742404e-01 6.79855943e-01 -1.91117883e-01 1.08903873e+00
-2.13077188e-01 1.09260464e+00 -4.03153896e-01 2.76488960e-02
-2.29654051e-02 1.49181098e-01 6.60926521e-01 1.59231770e+00
7.20011950e-01 -2.03309860e-03 -1.15170153e-02 5.46432674e-01
-3.99531037e-01 1.91418454e-01 -3.40389967e-01 -2.76370466e-01
6.70603931e-01 1.28675938e+00 -6.34606183e-01 -2.59390175e-01
-1.64232045e-01 1.19123054e+00 1.97487265e-01 1.70365453e-01
-9.07529891e-01 -4.47639972e-01 9.06102657e-01 -1.46555915e-01
4.65646505e-01 -3.24418068e-01 -4.92480576e-01 -1.00158691e+00
1.11630335e-01 -1.05909860e+00 3.07137012e-01 -8.34667921e-01
-1.12947798e+00 5.90526760e-01 -4.32233453e-01 -1.57689404e+00
-6.18619323e-01 -4.83222455e-01 -5.70193648e-01 6.23833120e-01
-9.35174882e-01 -8.93862724e-01 1.68685451e-01 3.74725610e-01
9.53326941e-01 -5.66939730e-03 1.18380105e+00 4.86067027e-01
-6.23632371e-01 9.57979381e-01 -1.31535903e-01 2.90754694e-03
1.06079888e+00 -1.22029424e+00 3.55073512e-01 7.78507769e-01
8.01330924e-01 7.27120340e-01 7.30166197e-01 -3.08314592e-01
-1.45631874e+00 -9.44513023e-01 7.65120506e-01 -3.97933245e-01
4.14950281e-01 -5.37824273e-01 -7.16673613e-01 4.71760035e-01
5.45149632e-02 -2.57491678e-01 7.56492019e-01 6.79118931e-01
-4.91202295e-01 -2.39719182e-01 -6.34993076e-01 8.68098080e-01
1.23048866e+00 -9.44685757e-01 -4.07065839e-01 6.81244880e-02
5.63848078e-01 -5.78314006e-01 -1.03324795e+00 4.20735866e-01
1.10158801e+00 -1.03593540e+00 1.06343961e+00 -1.88587308e-01
2.87536681e-01 -4.02340978e-01 -1.37807786e-01 -1.19421363e+00
-3.67881954e-01 -9.65545297e-01 -4.84182447e-01 1.45668674e+00
3.41681302e-01 -9.98893008e-02 5.94150722e-01 1.78699985e-01
-4.59536850e-01 -5.86297214e-01 -9.68669057e-01 -9.79590058e-01
-3.66688997e-01 -9.45752978e-01 4.33618903e-01 8.53426039e-01
1.04428813e-01 5.43871045e-01 -7.57507324e-01 9.22920331e-02
2.64577448e-01 5.09806514e-01 7.37415612e-01 -1.14317477e+00
-7.20655084e-01 -7.55224586e-01 -1.42251492e-01 -1.04015243e+00
-6.93222806e-02 -7.67740488e-01 1.48733899e-01 -9.80138540e-01
-5.20434715e-02 -2.35582605e-01 -4.80293572e-01 5.03453672e-01
-6.81076050e-02 8.68370116e-01 4.21564579e-01 2.15885401e-01
-4.61621255e-01 2.98362166e-01 1.15371823e+00 -7.77813867e-02
-4.97901499e-01 3.05496693e-01 -5.70297182e-01 8.12439203e-01
8.35018098e-01 -3.17126393e-01 -2.76652038e-01 -3.03319603e-01
1.26871958e-01 1.13801867e-01 4.55181897e-02 -1.37262034e+00
1.76657051e-01 1.15326092e-01 3.42806131e-01 -4.48532730e-01
8.36119890e-01 -4.47777569e-01 3.24647367e-01 1.65338755e-01
-5.34219325e-01 -4.46777679e-02 4.52232182e-01 2.75446981e-01
-3.59825075e-01 -1.87630743e-01 5.78097641e-01 -1.03685714e-01
-3.85408252e-01 2.14615986e-01 -2.90711164e-01 -2.60902122e-02
2.63851166e-01 -2.92581141e-01 1.31569013e-01 -5.24322093e-01
-5.82128406e-01 -5.22416115e-01 2.39017501e-01 3.98892969e-01
2.60085881e-01 -1.36621213e+00 -6.61985219e-01 -8.22735019e-03
1.68710709e-01 -2.45976686e-01 4.04548228e-01 8.25658441e-01
-1.04160771e-01 3.62243980e-01 -8.52390379e-03 -6.69125140e-01
-1.39191937e+00 3.31281990e-01 2.08866864e-01 -1.70584470e-01
-6.70527101e-01 9.73896086e-01 5.23406267e-02 -1.39587745e-01
5.25106132e-01 -6.83566988e-01 1.27580926e-01 7.91888461e-02
5.07167876e-01 4.35121745e-01 2.59229600e-01 -3.20472836e-01
-9.35891941e-02 5.91680169e-01 1.35035202e-01 -3.85109752e-01
1.23365295e+00 1.04040034e-01 3.17840219e-01 9.93991077e-01
7.69710660e-01 3.67747873e-01 -1.20917642e+00 5.46255037e-02
-2.37143517e-01 -4.06159192e-01 1.62137628e-01 -9.09189463e-01
-9.28229570e-01 9.75176096e-01 4.95997399e-01 3.27513576e-01
1.60040927e+00 -3.37216020e-01 1.08430994e+00 3.69611502e-01
3.12459856e-01 -1.28838837e+00 1.15789451e-01 4.67353582e-01
8.95222783e-01 -8.54543507e-01 -2.14337170e-01 -1.70609012e-01
-4.82104897e-01 1.17207170e+00 3.00262779e-01 -4.59095538e-02
1.10089906e-01 4.55586523e-01 1.55549929e-01 1.56259239e-01
-8.51730943e-01 -2.01434672e-01 3.99212718e-01 3.28708649e-01
9.97401178e-01 1.27942011e-01 -3.04377526e-02 8.34374964e-01
-7.23366499e-01 1.11962527e-01 1.68208212e-01 5.65982938e-01
-2.44578481e-01 -1.16763103e+00 -4.55783904e-01 4.44894284e-01
-9.09947217e-01 -2.48653948e-01 -1.83769807e-01 5.33177435e-01
1.12643495e-01 1.09556735e+00 -8.61963630e-02 -3.49022716e-01
5.13586283e-01 3.78878534e-01 8.40486169e-01 -5.16584694e-01
-1.10062885e+00 6.07980907e-01 1.83097601e-01 -5.76939225e-01
-4.36683714e-01 -7.09084690e-01 -1.11823106e+00 -1.25503451e-01
-1.77690297e-01 -1.15587190e-01 6.70929551e-01 7.08375871e-01
4.53538150e-01 9.51029241e-01 3.74918729e-01 -1.18267894e+00
-3.06680053e-01 -1.26750970e+00 -6.78372204e-01 3.32174063e-01
2.24191695e-01 -4.23523575e-01 -2.15353847e-01 2.42246673e-01] | [15.78536605834961, 5.376358985900879] |
2aeb2598-4b9c-4ec3-878c-c0e0ef95ea77 | real-time-human-centric-segmentation-for | 2108.07199 | null | https://arxiv.org/abs/2108.07199v1 | https://arxiv.org/pdf/2108.07199v1.pdf | Real-time Human-Centric Segmentation for Complex Video Scenes | Most existing video tasks related to "human" focus on the segmentation of salient humans, ignoring the unspecified others in the video. Few studies have focused on segmenting and tracking all humans in a complex video, including pedestrians and humans of other states (e.g., seated, riding, or occluded). In this paper, we propose a novel framework, abbreviated as HVISNet, that segments and tracks all presented people in given videos based on a one-stage detector. To better evaluate complex scenes, we offer a new benchmark called HVIS (Human Video Instance Segmentation), which comprises 1447 human instance masks in 805 high-resolution videos in diverse scenes. Extensive experiments show that our proposed HVISNet outperforms the state-of-the-art methods in terms of accuracy at a real-time inference speed (30 FPS), especially on complex video scenes. We also notice that using the center of the bounding box to distinguish different individuals severely deteriorates the segmentation accuracy, especially in heavily occluded conditions. This common phenomenon is referred to as the ambiguous positive samples problem. To alleviate this problem, we propose a mechanism named Inner Center Sampling to improve the accuracy of instance segmentation. Such a plug-and-play inner center sampling mechanism can be incorporated in any instance segmentation models based on a one-stage detector to improve the performance. In particular, it gains 4.1 mAP improvement on the state-of-the-art method in the case of occluded humans. Code and data are available at https://github.com/IIGROUP/HVISNet. | ['Yujiu Yang', 'Haoqian Wang', 'Xinyuan Zhao', 'Weihao Xia', 'Chenyu Tian', 'Ran Yu'] | 2021-08-16 | null | null | null | null | ['video-instance-segmentation'] | ['computer-vision'] | [-2.29643732e-02 -1.49918497e-01 -1.76765040e-01 -3.00883174e-01
-3.09037745e-01 -2.93286115e-01 3.57756108e-01 -8.95743594e-02
-7.09274411e-01 6.94294035e-01 -1.56452522e-01 3.98505367e-02
4.08201993e-01 -5.99409461e-01 -8.85500073e-01 -6.05483234e-01
1.11427672e-01 4.71585006e-01 8.53422463e-01 3.43973823e-02
-3.21868109e-03 1.65847123e-01 -1.68086267e+00 1.90079421e-01
8.18229914e-01 7.64375150e-01 2.09376812e-01 3.96695673e-01
1.34574816e-01 5.04623950e-01 -8.39375436e-01 -6.11897409e-01
3.86311293e-01 -3.07821065e-01 -6.00363433e-01 4.10793006e-01
5.66641271e-01 -6.29769862e-01 -3.81709754e-01 1.10892761e+00
4.63520199e-01 2.08711743e-01 2.99539447e-01 -1.38073266e+00
-3.43891010e-02 3.52728993e-01 -9.77331519e-01 5.71345985e-01
6.37428880e-01 5.48817515e-01 5.68996191e-01 -6.77840412e-01
5.94946802e-01 1.53231597e+00 6.41360641e-01 5.96093059e-01
-9.88905430e-01 -8.46887052e-01 5.11664748e-01 5.81852913e-01
-1.65328097e+00 -3.76938194e-01 5.45880556e-01 -5.65457225e-01
4.12124395e-01 4.03973699e-01 9.21404064e-01 1.11403680e+00
-1.68090016e-01 1.26743591e+00 9.19017255e-01 5.27811386e-02
1.83268338e-01 1.35081038e-01 3.22845608e-01 7.55557477e-01
5.81154466e-01 -4.40767817e-02 -1.80727974e-01 2.60833930e-02
7.49086976e-01 1.25456333e-01 -2.09596902e-01 -3.10328662e-01
-1.21351588e+00 6.25908077e-01 4.38786119e-01 -3.62771861e-02
-2.99521476e-01 -1.67465404e-01 4.88550782e-01 -2.67137110e-01
4.07784373e-01 -4.78954017e-02 -2.65892714e-01 1.26300678e-01
-1.12433267e+00 5.70124567e-01 5.74368596e-01 9.58628714e-01
6.13193214e-01 -1.08025581e-01 -5.70233881e-01 6.95288241e-01
5.41182347e-02 5.57229996e-01 7.11666569e-02 -9.40607965e-01
5.29242873e-01 7.62636125e-01 3.60441804e-01 -8.78673971e-01
-5.36317706e-01 -5.13987839e-01 -8.36249590e-01 1.57521549e-03
6.29074395e-01 -5.00614166e-01 -1.03846860e+00 1.52290511e+00
9.13840413e-01 5.14205694e-01 -2.80417591e-01 1.46379685e+00
9.65796530e-01 7.21538186e-01 3.47572505e-01 -1.63377285e-01
1.78942037e+00 -1.16935599e+00 -6.10280991e-01 -4.36524183e-01
2.24964857e-01 -4.48983371e-01 7.43991494e-01 3.11123252e-01
-8.89470935e-01 -8.70488226e-01 -8.54606867e-01 1.83105811e-01
-3.85238349e-01 3.06524545e-01 2.90679872e-01 6.38646007e-01
-6.11544311e-01 2.11976409e-01 -7.13231862e-01 -5.73608696e-01
4.85299200e-01 2.18759060e-01 -3.06116462e-01 1.46713983e-02
-1.19969213e+00 5.04277706e-01 4.55562860e-01 9.61069688e-02
-9.38212514e-01 -5.89544535e-01 -8.91045272e-01 -9.98679101e-02
1.06855524e+00 -7.95122683e-01 1.11211360e+00 -9.99109685e-01
-1.01174867e+00 8.54281843e-01 -2.52885401e-01 -5.47746062e-01
1.04465818e+00 -5.21364987e-01 -3.24223787e-01 3.03976625e-01
2.54906625e-01 8.86985660e-01 9.19159949e-01 -1.25441837e+00
-8.82634640e-01 -3.98369551e-01 2.24139556e-01 2.14620918e-01
7.17096627e-02 3.65953177e-01 -1.06605887e+00 -5.58817744e-01
-3.05675179e-01 -1.05686760e+00 -2.66677201e-01 -7.51024671e-03
-6.19614482e-01 -3.27171922e-01 8.93918931e-01 -9.56974506e-01
1.33597493e+00 -2.03268099e+00 -5.57515025e-02 -6.06918968e-02
1.53373346e-01 7.13893950e-01 1.87359869e-01 -8.20791423e-02
2.49699473e-01 -2.17622258e-02 -1.89057514e-01 -3.50780398e-01
-1.01870112e-01 1.12211384e-01 2.66648978e-01 5.73214591e-01
1.58685893e-01 6.79601669e-01 -7.77582705e-01 -9.37167943e-01
3.96409720e-01 4.34035003e-01 -6.11200809e-01 2.36622423e-01
-5.20627536e-02 6.29035294e-01 -3.72915089e-01 7.94527769e-01
8.71137619e-01 -2.01467514e-01 -1.13239037e-02 -1.61125794e-01
-1.29396543e-01 -2.80182630e-01 -1.46089852e+00 1.23917234e+00
3.04639339e-01 3.87337416e-01 1.85265690e-01 -8.87950838e-01
5.28860927e-01 2.56123543e-01 3.17583799e-01 -5.06086826e-01
1.61735520e-01 -2.21452162e-01 2.77141351e-02 -6.28349125e-01
4.98031557e-01 4.53759849e-01 -4.24582995e-02 -2.68682718e-01
-1.63799360e-01 6.21632874e-01 7.38528192e-01 1.28877968e-01
7.61804581e-01 1.74955577e-01 3.15076262e-01 -2.32013807e-01
7.51203120e-01 -4.86161076e-02 1.00487697e+00 9.19108808e-01
-7.68266797e-01 6.01098120e-01 5.45335889e-01 -5.14049351e-01
-8.75622272e-01 -8.96370530e-01 3.30803767e-02 1.14325213e+00
5.67392349e-01 -3.97204489e-01 -1.24479890e+00 -6.50979340e-01
-4.94659245e-02 3.11314017e-01 -4.33843493e-01 2.00577572e-01
-7.49530911e-01 -7.56830931e-01 4.29567754e-01 4.97689098e-01
9.96222556e-01 -1.08676422e+00 -9.49303985e-01 -3.99834998e-02
-6.18875682e-01 -1.43291676e+00 -4.58314747e-01 -4.76370484e-01
-4.37635422e-01 -1.27200866e+00 -1.10823107e+00 -6.19609356e-01
6.32799983e-01 3.67784828e-01 9.10231590e-01 1.47405297e-01
-4.95844424e-01 1.79986626e-01 -3.39669049e-01 -2.43818209e-01
1.16816737e-01 3.42598185e-02 1.26147429e-02 3.11533689e-01
4.07947123e-01 9.95697379e-02 -9.30185854e-01 7.80459881e-01
-5.62448263e-01 2.33473092e-01 3.88743162e-01 5.33459902e-01
5.73633194e-01 2.24052832e-01 2.64032125e-01 -9.22444582e-01
9.05923024e-02 -4.56173807e-01 -6.27307832e-01 2.41442248e-01
6.63432777e-02 -3.89169455e-01 4.30982858e-01 -5.75318575e-01
-1.05236697e+00 2.69020051e-01 -8.32768157e-02 -4.56097394e-01
-5.24209976e-01 -3.47313434e-02 -3.55113178e-01 1.50511429e-01
3.54564279e-01 6.84140325e-02 -2.42721230e-01 -4.33265954e-01
2.91998056e-03 5.93764305e-01 6.22288108e-01 -4.98107255e-01
6.40223742e-01 4.83591855e-01 -2.98753738e-01 -9.21868682e-01
-7.83056259e-01 -7.76536763e-01 -5.62111676e-01 -6.17069602e-01
1.14323330e+00 -1.16697478e+00 -7.95387506e-01 6.24340117e-01
-1.04457116e+00 -3.33742321e-01 1.98471650e-01 3.51967514e-01
-3.04891646e-01 4.41712916e-01 -6.25816762e-01 -9.98019278e-01
-2.37113714e-01 -1.22455299e+00 1.10397995e+00 6.30851269e-01
-5.60364407e-03 -4.10827726e-01 -4.54170823e-01 6.86605275e-01
4.57427055e-02 3.11634541e-01 1.48414969e-01 -6.03275836e-01
-7.03845382e-01 -1.96563646e-01 -3.33693027e-01 2.14784816e-01
-4.13074285e-01 8.74543712e-02 -7.44065404e-01 -3.58070731e-01
-3.18316579e-01 1.19893178e-01 8.95403087e-01 5.25414646e-01
1.17587805e+00 -3.27920139e-01 -5.65856457e-01 3.70800793e-01
1.06355929e+00 3.08187515e-01 6.49525821e-01 1.46326408e-01
7.72279799e-01 6.42141521e-01 9.91994858e-01 6.64634526e-01
3.74398410e-01 8.83428276e-01 2.02198550e-01 -3.33459228e-01
-3.87346186e-02 -1.19185880e-01 3.26270610e-01 9.81993228e-02
-3.36205333e-01 -2.69741565e-01 -8.51539850e-01 4.83736455e-01
-2.09893918e+00 -1.21409130e+00 -1.62521496e-01 2.21948671e+00
4.82173681e-01 3.75070274e-01 8.02211761e-01 1.19739778e-01
1.37263656e+00 2.00400069e-01 -5.45123994e-01 2.12620080e-01
1.94930285e-01 -2.68466562e-01 5.41634381e-01 2.80845612e-01
-1.56110406e+00 9.97193694e-01 5.16149378e+00 9.16301787e-01
-7.45466948e-01 1.90307692e-01 7.87070274e-01 -3.84629481e-02
6.16204381e-01 -2.29371935e-01 -1.28210032e+00 8.91089678e-01
5.30396044e-01 7.40623325e-02 2.62498945e-01 9.10763443e-01
3.89988273e-01 -3.56026679e-01 -7.74312317e-01 1.18239558e+00
2.14449033e-01 -8.32111239e-01 -7.01243877e-02 -1.54715121e-01
4.59266782e-01 -3.27793926e-01 -1.94068298e-01 4.91748780e-01
-2.12302580e-01 -4.78157640e-01 8.27439249e-01 4.40809280e-01
3.73244584e-01 -7.65082240e-01 8.74420643e-01 4.47845936e-01
-1.41231549e+00 -9.69979353e-03 -3.29046935e-01 -6.95771649e-02
3.72029543e-01 3.28609794e-01 -4.26511914e-01 3.94610763e-01
1.02195120e+00 5.26410758e-01 -8.39832544e-01 1.48849380e+00
-1.75779402e-01 4.75643992e-01 -4.06063914e-01 1.34886324e-01
1.74949571e-01 -7.31673390e-02 6.14130914e-01 1.40237272e+00
5.12147248e-02 2.98249751e-01 7.62133956e-01 6.54463530e-01
2.42177606e-01 -4.34451215e-02 -2.71514893e-01 4.21902359e-01
3.54426712e-01 1.27099705e+00 -9.65570033e-01 -8.81499946e-01
-5.00794232e-01 9.70663726e-01 -1.08231947e-01 5.37444234e-01
-1.43714607e+00 -2.48719990e-01 5.49772143e-01 6.15143955e-01
5.35130858e-01 -3.10120191e-02 1.66448936e-01 -1.22169697e+00
1.42641932e-01 -8.54753673e-01 6.16140962e-01 -5.81242442e-01
-9.13422167e-01 4.53093588e-01 4.12731588e-01 -1.11286604e+00
-6.35779053e-02 -4.46121544e-01 -3.78533959e-01 3.17382246e-01
-1.19047880e+00 -9.82696652e-01 -6.81079745e-01 8.65711808e-01
7.86153436e-01 1.30782112e-01 -6.98377788e-02 7.72544384e-01
-1.14882934e+00 5.96182287e-01 -5.23300171e-01 5.00318706e-01
6.53443336e-01 -8.72007251e-01 2.93322086e-01 1.14524126e+00
-2.75872946e-01 4.11126465e-01 8.65438938e-01 -7.72862434e-01
-9.59741831e-01 -1.32609069e+00 5.43444991e-01 -1.56753495e-01
2.17104852e-01 -4.59872246e-01 -8.20779860e-01 7.30592787e-01
6.82116523e-02 2.10236862e-01 3.44521582e-01 -1.65193096e-01
3.13377976e-02 -8.45222622e-02 -1.23493767e+00 7.09019184e-01
1.24330890e+00 6.16608607e-03 -4.25499588e-01 3.79902393e-01
5.65060019e-01 -6.06624603e-01 -5.37121952e-01 5.95492303e-01
4.59008962e-01 -1.17017305e+00 1.21856582e+00 -3.74371976e-01
1.57908753e-01 -6.48911297e-01 1.11144133e-01 -7.58427143e-01
-3.68689567e-01 -4.55081344e-01 -2.36318320e-01 1.16096199e+00
1.16990227e-02 -5.39033592e-01 8.25387955e-01 6.15735471e-01
4.81395535e-02 -5.43554187e-01 -8.12840819e-01 -8.56805623e-01
-4.73124087e-01 -4.86931354e-02 4.53791171e-01 5.97963810e-01
-2.81767517e-01 2.16686487e-01 -6.00977719e-01 3.45414758e-01
8.52027476e-01 -1.00569502e-01 1.10928559e+00 -1.11008966e+00
-2.03107774e-01 -2.46929675e-01 -6.47253871e-01 -1.21277475e+00
1.18620330e-02 -2.53232032e-01 7.71592557e-02 -1.39686716e+00
5.22666335e-01 -1.57421470e-01 -8.48095268e-02 3.90675008e-01
-5.68660080e-01 3.49842966e-01 5.33724844e-01 1.45682171e-01
-1.23190284e+00 2.20201239e-01 1.19136536e+00 -1.60564169e-01
-8.64290074e-02 1.77372217e-01 -3.45311940e-01 9.29154396e-01
8.33063066e-01 -3.15055281e-01 -7.74029549e-03 -1.66639045e-01
-4.46200788e-01 8.13181847e-02 8.15927029e-01 -1.61233795e+00
3.14114362e-01 -2.32969344e-01 5.73932886e-01 -7.98036933e-01
4.61603373e-01 -7.60024607e-01 3.88192534e-01 7.16999233e-01
4.35833894e-02 -1.71062015e-02 9.84317511e-02 5.08472323e-01
-1.79404795e-01 -1.04576871e-01 9.38043237e-01 -3.09429526e-01
-1.14169097e+00 3.78190398e-01 -2.89266706e-01 1.96179748e-01
1.45693719e+00 -2.74905503e-01 -4.56458032e-01 -9.93505344e-02
-7.19467461e-01 4.65073615e-01 4.20832723e-01 4.81720984e-01
3.85871917e-01 -1.04182565e+00 -7.38610268e-01 2.33403414e-01
-9.08587966e-03 -1.00779146e-01 6.11568511e-01 9.22911465e-01
-5.12018144e-01 3.48560333e-01 -2.58315325e-01 -8.40546489e-01
-1.51535523e+00 7.96327353e-01 3.09535325e-01 -1.60298437e-01
-6.02909923e-01 5.44453263e-01 5.70446432e-01 -6.19156733e-02
4.20621544e-01 -2.22561076e-01 -3.07788700e-01 1.65517062e-01
7.27750242e-01 6.76403821e-01 -3.74995083e-01 -9.59067047e-01
-6.29334867e-01 4.95926917e-01 -5.46379201e-02 1.97161376e-01
7.17029512e-01 -1.15882568e-01 2.49654621e-01 1.50168136e-01
7.45497584e-01 -3.05943310e-01 -1.58531332e+00 -1.58874318e-01
-3.00157398e-01 -6.40909910e-01 -3.79921973e-01 -5.54478526e-01
-1.29907751e+00 6.48815632e-01 7.24778593e-01 8.40713605e-02
8.26246321e-01 -7.30401501e-02 9.65678453e-01 5.84499463e-02
4.61903244e-01 -1.17624617e+00 -1.31304398e-01 2.37498671e-01
5.10141253e-01 -1.41704106e+00 8.87869149e-02 -6.89257145e-01
-8.63972068e-01 5.31262338e-01 1.08148956e+00 -1.58253878e-01
3.88239622e-01 -3.27579938e-02 -1.25384554e-01 1.29386773e-02
-3.29150736e-01 -5.23081541e-01 2.52242804e-01 4.33329076e-01
8.80348012e-02 2.89838523e-01 -4.59936023e-01 5.46918750e-01
9.36770514e-02 2.20826603e-02 2.35755786e-01 7.37304866e-01
-5.56771934e-01 -5.39065123e-01 -7.53120482e-01 4.05709684e-01
-5.20895243e-01 3.23713958e-01 -1.71069607e-01 9.96253729e-01
5.60960770e-01 1.12438345e+00 1.28500938e-01 -2.40515426e-01
4.92993474e-01 -1.12254001e-01 2.71650463e-01 -3.16305518e-01
-5.98677933e-01 1.02451310e-01 1.33023620e-01 -7.46824324e-01
-6.45087183e-01 -7.62210488e-01 -1.10335982e+00 -5.23239434e-01
-2.45427594e-01 1.10104876e-02 2.06834584e-01 7.87457585e-01
2.29402214e-01 5.09874284e-01 1.11025728e-01 -1.15511012e+00
-7.28652179e-02 -7.74501443e-01 -4.56512451e-01 6.93713188e-01
1.81417525e-01 -1.00882018e+00 -2.14000881e-01 1.44529536e-01] | [8.251976013183594, -0.42179787158966064] |
c5f3e1c2-3cb8-47e9-835a-bb56da9785f0 | instant-one-shot-word-learning-for-context | 2107.02268 | null | https://arxiv.org/abs/2107.02268v1 | https://arxiv.org/pdf/2107.02268v1.pdf | Instant One-Shot Word-Learning for Context-Specific Neural Sequence-to-Sequence Speech Recognition | Neural sequence-to-sequence systems deliver state-of-the-art performance for automatic speech recognition (ASR). When using appropriate modeling units, e.g., byte-pair encoded characters, these systems are in principal open vocabulary systems. In practice, however, they often fail to recognize words not seen during training, e.g., named entities, numbers or technical terms. To alleviate this problem we supplement an end-to-end ASR system with a word/phrase memory and a mechanism to access this memory to recognize the words and phrases correctly. After the training of the ASR system, and when it has already been deployed, a relevant word can be added or subtracted instantly without the need for further training. In this paper we demonstrate that through this mechanism our system is able to recognize more than 85% of newly added words that it previously failed to recognize compared to a strong baseline. | ['Alexander Waibel', 'Sebastian Stüker', 'Juan Hussain', 'Christian Huber'] | 2021-07-05 | null | null | null | null | ['sequence-to-sequence-speech-recognition'] | ['speech'] | [ 4.63405907e-01 -5.01501001e-02 3.15209366e-02 -2.66085267e-01
-1.06123948e+00 -7.85618424e-01 5.84753096e-01 4.22573328e-01
-8.27360034e-01 6.50140226e-01 1.68005899e-01 -9.01116133e-01
6.12878382e-01 -5.62804103e-01 -7.05405653e-01 -3.49940568e-01
2.28914723e-01 5.47668636e-01 2.33413652e-01 -4.90141124e-01
2.08988741e-01 5.74154615e-01 -1.32005584e+00 3.94902527e-01
4.85447675e-01 5.11600375e-01 3.42578083e-01 9.87563252e-01
-4.90967572e-01 6.40833497e-01 -9.35015023e-01 -2.82112598e-01
9.52349752e-02 9.94456094e-03 -8.16295087e-01 -5.85056841e-02
4.08250779e-01 -4.58221346e-01 -6.03374720e-01 8.35874856e-01
6.59943223e-01 4.20881867e-01 3.98118705e-01 -3.48137051e-01
-6.24598145e-01 8.46136272e-01 -2.11676862e-02 4.57798809e-01
5.49750686e-01 1.91334546e-01 9.76797998e-01 -1.41565597e+00
4.54358011e-01 1.04087985e+00 2.32171491e-01 9.23918724e-01
-1.08548892e+00 -4.33341831e-01 1.84430838e-01 -7.66605511e-02
-1.54006875e+00 -1.21523237e+00 1.20328687e-01 -3.36003006e-02
1.79765904e+00 4.20537859e-01 1.27174258e-01 1.04785609e+00
-2.96597123e-01 7.85716593e-01 3.77671212e-01 -7.52642453e-01
3.10592562e-01 -2.31682304e-02 4.81172055e-01 5.01591742e-01
1.07641771e-01 -1.34067446e-01 -5.60520351e-01 -1.46269649e-02
4.19890553e-01 -3.37919772e-01 -2.28599146e-01 2.74295837e-01
-1.25855720e+00 5.71305692e-01 -4.83027063e-02 4.52601343e-01
-4.91282165e-01 1.14381783e-01 5.14011800e-01 3.31776828e-01
2.52259225e-01 4.78066415e-01 -6.20370209e-01 -5.66832900e-01
-1.06907415e+00 -3.13501745e-01 7.39140093e-01 9.64028001e-01
4.64710325e-01 4.23267126e-01 -1.12003289e-01 1.10700846e+00
5.90034202e-02 7.90395021e-01 8.71280015e-01 -2.30301306e-01
5.36655724e-01 3.13262165e-01 -7.85578117e-02 -2.52128512e-01
3.51702329e-03 -3.31019968e-01 -5.60520172e-01 -2.48279721e-01
1.55326381e-01 -1.29391506e-01 -1.49796498e+00 1.41187310e+00
-2.05199625e-02 2.43444681e-01 4.84658539e-01 5.30797362e-01
7.80325890e-01 1.00372326e+00 -1.84071846e-02 -2.38599852e-01
1.43052888e+00 -8.85868251e-01 -6.82740271e-01 -8.58220458e-01
8.23393285e-01 -9.22402203e-01 8.42882037e-01 1.26509503e-01
-1.36241567e+00 -5.27615130e-01 -1.02908671e+00 -6.26837239e-02
-6.35741472e-01 3.98762077e-02 6.69726729e-02 6.75198495e-01
-1.24625504e+00 2.86304504e-01 -7.98849165e-01 -3.54201615e-01
-1.50083616e-01 3.47128600e-01 -5.12325287e-01 3.18468059e-03
-1.30416024e+00 9.21741605e-01 5.08610129e-01 -1.35690317e-01
-6.71843410e-01 -5.83126843e-01 -8.81245673e-01 2.31198132e-01
2.82367319e-01 -3.01443547e-01 1.79942667e+00 -8.02157640e-01
-1.44107819e+00 7.84859419e-01 -6.00353479e-01 -7.61771441e-01
-6.42702878e-02 -2.30924845e-01 -7.25550711e-01 1.34687796e-01
-3.03413719e-01 4.31675822e-01 8.39068234e-01 -6.92386329e-01
-6.74300313e-01 -1.57479092e-01 -1.60717979e-01 1.12349361e-01
-3.98137927e-01 5.41205823e-01 -5.18666983e-01 -8.60206425e-01
-4.32430506e-02 -9.38023686e-01 -4.62164842e-02 -6.89837277e-01
-1.49161711e-01 -1.28591701e-01 5.07725775e-01 -9.58500028e-01
1.56956363e+00 -2.37451196e+00 -6.45807311e-02 3.99778813e-01
-1.28564820e-01 1.18191981e+00 -3.50458801e-01 6.36001885e-01
-2.42252454e-01 3.31037223e-01 -2.11597994e-01 -3.90017390e-01
-9.68754515e-02 3.19181830e-01 -5.62402248e-01 7.84596354e-02
3.80430907e-01 8.88145745e-01 -7.55325258e-01 -1.70706175e-02
2.54713207e-01 4.85922009e-01 -1.64539680e-01 3.62651557e-01
-2.54484385e-01 -1.67905882e-01 -3.73259149e-02 2.99017459e-01
2.42777213e-01 -1.87851578e-01 1.09103739e-01 3.76384377e-01
4.19863388e-02 1.20337260e+00 -1.03217566e+00 1.27725720e+00
-7.35551655e-01 6.71082556e-01 -9.06416029e-03 -8.04838538e-01
9.91331875e-01 7.89541125e-01 -1.27671465e-01 -6.62549615e-01
-1.08313570e-02 5.96546173e-01 1.20651722e-01 2.97277234e-02
7.98640788e-01 5.39148152e-02 1.12902083e-01 4.34917808e-01
4.10055518e-02 6.26588836e-02 2.13335171e-01 2.89013594e-01
1.39082408e+00 -6.18141234e-01 5.52367985e-01 2.33565584e-01
8.92503083e-01 -1.59067288e-01 4.36133966e-02 8.18055332e-01
2.02317551e-01 6.32714033e-01 -2.31646568e-01 -2.36466393e-01
-1.38389683e+00 -8.34471345e-01 1.04518399e-01 1.34256983e+00
-4.30051416e-01 -5.83595753e-01 -6.81984067e-01 -5.31768203e-01
-5.67980185e-02 1.11653113e+00 2.98924837e-02 -4.08499151e-01
-9.81097400e-01 -2.49058683e-03 9.08599019e-01 6.79328084e-01
-9.42124128e-02 -1.08501434e+00 -7.36292675e-02 6.30193055e-01
-3.28035317e-02 -1.18848121e+00 -7.50982106e-01 4.26342845e-01
-5.77489138e-01 -3.68451416e-01 -8.40477049e-01 -8.85466635e-01
5.87994099e-01 4.73894507e-01 1.11801112e+00 4.32695270e-01
1.09694719e-01 1.84374973e-01 -5.20642877e-01 7.66957877e-03
-1.01141536e+00 3.70814919e-01 1.52200580e-01 -5.00109978e-02
6.55422509e-01 -1.96736053e-01 -1.81594640e-01 3.04840177e-01
-9.50267792e-01 -1.29823312e-01 6.79843605e-01 8.67992878e-01
3.16911697e-01 -5.10363519e-01 6.57840550e-01 -5.78116059e-01
7.35243201e-01 -1.90247983e-01 -3.62837046e-01 4.90681440e-01
-4.80540961e-01 3.66846979e-01 6.90180302e-01 -6.34545863e-01
-6.55543745e-01 4.54609729e-02 -7.34784126e-01 -2.15546042e-01
-3.17132711e-01 7.53755152e-01 1.16525944e-02 1.12748869e-01
6.46645486e-01 7.00446069e-01 -1.61431342e-01 -5.47050834e-01
5.53870022e-01 1.43262088e+00 5.96422791e-01 -3.33813339e-01
6.53910875e-01 -2.35561997e-01 -7.97614932e-01 -1.13263786e+00
-4.10001457e-01 -9.42539155e-01 -4.07753855e-01 1.88288927e-01
5.30864060e-01 -9.79780376e-01 -3.17034274e-01 3.79743606e-01
-1.37354195e+00 -2.55402625e-01 -2.79770821e-01 4.06102955e-01
-1.85452342e-01 3.98836732e-01 -6.63201451e-01 -1.01959085e+00
-6.02378130e-01 -9.45211172e-01 1.05803823e+00 -5.97760873e-03
-4.12494838e-01 -8.20116520e-01 -1.43063545e-01 2.01656029e-01
6.63073123e-01 -8.78263533e-01 8.37461233e-01 -1.31706619e+00
-2.93293506e-01 -4.71754432e-01 1.31350741e-01 6.68666482e-01
1.13881849e-01 7.27966055e-02 -8.46003294e-01 -4.12021816e-01
-3.75815988e-01 -1.62611455e-01 8.71544898e-01 -1.80872858e-01
7.98103511e-01 -4.72367197e-01 -1.29916593e-01 2.44042009e-01
7.61197746e-01 6.15600049e-01 7.74797499e-01 1.23841152e-01
3.80015314e-01 2.01375216e-01 8.75488445e-02 1.24671578e-01
9.22295731e-03 6.99704885e-01 -1.28140002e-01 1.42496422e-01
-1.90287381e-01 -3.42924565e-01 6.95821583e-01 1.19803905e+00
4.02085006e-01 -5.91211140e-01 -1.11152709e+00 6.66757226e-01
-1.41826808e+00 -7.48115420e-01 1.65580690e-01 2.47238636e+00
1.11904895e+00 4.40626979e-01 -1.74117759e-01 8.83487985e-02
8.09695482e-01 2.20301300e-01 -2.85064906e-01 -8.03075194e-01
-1.99197635e-01 5.80577075e-01 7.22157717e-01 7.96575129e-01
-7.44881868e-01 1.38524079e+00 6.67577457e+00 1.11134243e+00
-1.30037379e+00 -1.17557146e-01 3.50560725e-01 1.00804284e-01
-4.83389169e-01 -1.53012216e-01 -1.09001279e+00 2.35301197e-01
1.64497089e+00 -2.24625736e-01 7.30978191e-01 7.52314091e-01
-1.00336939e-01 3.13734442e-01 -1.11080408e+00 1.06099916e+00
2.47392341e-01 -1.15912616e+00 1.96030259e-01 -2.21773833e-01
2.05234677e-01 1.52880862e-01 -6.00802377e-02 5.28219819e-01
2.88548976e-01 -1.23204756e+00 6.43176317e-01 4.15264033e-02
9.90415692e-01 -5.85387230e-01 5.99361300e-01 4.49622691e-01
-9.74586010e-01 1.67749316e-01 -3.07296008e-01 -7.11996853e-03
3.52766514e-01 2.98620492e-01 -1.32046103e+00 1.80078387e-01
-1.95393324e-01 1.81240588e-01 -5.33376098e-01 7.92064786e-01
-4.34783936e-01 1.10880494e+00 -4.55839753e-01 -4.35100526e-01
3.95351768e-01 3.56662780e-01 5.53864837e-01 1.43909502e+00
3.56283009e-01 2.50924230e-01 1.57761190e-03 3.90114076e-02
-2.85104185e-01 2.72745460e-01 -5.34925580e-01 -5.65397441e-01
6.65102601e-01 8.58515024e-01 -3.69776070e-01 -8.44019413e-01
-4.87920552e-01 1.19127584e+00 5.54724276e-01 3.62146676e-01
-3.95779014e-01 -4.70882475e-01 8.47879767e-01 6.33133482e-03
6.55967236e-01 -7.24310398e-01 -5.99420816e-02 -1.23643410e+00
9.64979976e-02 -1.31100190e+00 1.23346791e-01 -8.48113716e-01
-1.04658794e+00 8.61534476e-01 -5.08892179e-01 -6.19658470e-01
-6.73816144e-01 -7.96738505e-01 -3.21109861e-01 1.01925600e+00
-1.18433273e+00 -5.63897610e-01 4.33762014e-01 3.60810131e-01
8.78222108e-01 -4.75805670e-01 1.08873987e+00 4.38259423e-01
-4.34189081e-01 8.74569416e-01 2.25043431e-01 5.52142620e-01
6.47083938e-01 -1.14517009e+00 1.32944119e+00 1.25248742e+00
7.32784331e-01 1.07127035e+00 5.15621424e-01 -7.87908316e-01
-1.48254490e+00 -7.84485996e-01 1.27992511e+00 -3.02531689e-01
7.44978607e-01 -6.48937583e-01 -1.30395722e+00 7.10988164e-01
1.01267189e-01 -7.73711596e-03 6.67738497e-01 1.10069506e-01
-6.66467726e-01 7.22341463e-02 -6.08650386e-01 7.30376363e-01
9.05419648e-01 -9.68503237e-01 -9.67920244e-01 2.77979314e-01
1.21935761e+00 -5.41492581e-01 -4.42057759e-01 1.33160189e-01
4.01726335e-01 -1.82649851e-01 9.20045078e-01 -1.03475428e+00
-6.56027719e-02 -2.26264566e-01 -3.85440677e-01 -1.41041672e+00
-1.95345074e-01 -8.57593238e-01 -4.24217850e-01 1.17435622e+00
9.03901637e-01 -6.93430126e-01 3.74599040e-01 6.18509531e-01
-3.25664312e-01 -2.86222428e-01 -1.09527755e+00 -9.37372565e-01
-7.91291445e-02 -7.12477148e-01 6.97506189e-01 5.66997647e-01
5.14871813e-02 7.04081297e-01 -2.36639440e-01 2.60481238e-01
-7.38866478e-02 -4.62846369e-01 5.98977625e-01 -6.85315967e-01
-1.65251300e-01 -4.35824305e-01 -5.37316084e-01 -1.54945827e+00
3.03391576e-01 -8.37282419e-01 3.79698783e-01 -1.33391845e+00
-3.25356007e-01 -4.33870614e-01 -3.42720836e-01 5.78048348e-01
-3.94133300e-01 3.52624916e-02 1.43473208e-01 -2.44545098e-02
-4.56645697e-01 2.63909519e-01 4.65385228e-01 -1.31707817e-01
-6.10936098e-02 -1.60009302e-02 -4.78128821e-01 3.18185389e-01
9.24123108e-01 -4.76807654e-01 -4.84853163e-02 -6.01613104e-01
1.71513064e-03 2.01516479e-01 -2.93524936e-02 -8.83172691e-01
4.56503391e-01 1.92626826e-02 -2.66035516e-02 -4.80233669e-01
3.88562530e-01 -6.12020254e-01 -9.15933326e-02 2.62485921e-01
-6.16255105e-01 2.35346138e-01 2.48042837e-01 2.53779292e-01
-2.13174522e-01 -5.76075971e-01 5.61087489e-01 5.31516187e-02
-8.70338321e-01 4.04422320e-02 -9.31141257e-01 1.89492375e-01
6.02358341e-01 -1.24584585e-01 -4.24214005e-01 -5.23892641e-01
-5.46133757e-01 8.33432656e-03 3.72771800e-01 6.30680561e-01
7.50192225e-01 -9.63360310e-01 -7.81798005e-01 4.61838186e-01
3.23888898e-01 -2.62080371e-01 -7.74234906e-02 3.04820597e-01
-4.48103875e-01 7.99724042e-01 3.61585885e-01 -1.52896896e-01
-1.55044127e+00 7.32073963e-01 3.24676156e-01 -1.34563074e-01
-5.09390950e-01 8.72521877e-01 -2.36789390e-01 -3.21853042e-01
4.44146603e-01 -3.45387965e-01 5.84710389e-02 -2.16993123e-01
1.06158102e+00 2.23870259e-02 7.30512798e-01 -7.59750605e-01
-5.44570506e-01 3.99319008e-02 -4.10665005e-01 -5.15689373e-01
9.83583152e-01 -1.39238581e-01 1.71125099e-01 4.06590343e-01
1.08001721e+00 3.78217846e-01 -4.66588706e-01 -4.83068973e-01
2.03429475e-01 -2.65514910e-01 -6.53609633e-02 -8.02336812e-01
-6.35477185e-01 1.03679872e+00 2.68054605e-01 3.50971639e-01
7.31919467e-01 -1.18567996e-01 1.27122283e+00 8.60430539e-01
2.98121572e-01 -1.03549027e+00 -1.88899457e-01 1.17303193e+00
5.84674120e-01 -1.12314475e+00 -6.04525149e-01 -1.75445989e-01
-4.24163699e-01 1.14499199e+00 2.13715523e-01 1.74635693e-01
4.22028601e-01 4.57370490e-01 2.58244276e-01 3.01313579e-01
-1.07260430e+00 -4.13371772e-01 5.10374129e-01 4.51544195e-01
4.56743360e-01 1.42829999e-01 -2.78065234e-01 2.39141449e-01
-3.08572084e-01 -3.96104693e-01 5.44026732e-01 9.26628053e-01
-8.21226537e-01 -1.19882774e+00 -2.68061668e-01 3.30748588e-01
-6.83063686e-01 -7.95842707e-01 -3.47894758e-01 1.71610132e-01
-5.29400289e-01 1.12182367e+00 2.94479579e-01 -3.26461673e-01
3.87629658e-01 6.45412087e-01 6.88818470e-02 -1.05537868e+00
-6.90678239e-01 4.08426067e-03 4.38403845e-01 -2.55585283e-01
2.24509954e-01 -4.41281736e-01 -1.36730671e+00 -2.85704941e-01
-5.06857216e-01 2.30488673e-01 8.95429969e-01 8.51820648e-01
4.18399900e-01 3.90194565e-01 4.26447213e-01 -1.55354366e-01
-8.61936212e-01 -9.93393898e-01 -2.00290576e-01 2.02636600e-01
5.80302536e-01 -3.98185402e-02 -2.79398024e-01 4.46150377e-02] | [14.284862518310547, 6.835782051086426] |
a1231e75-93c3-4548-8c81-90173cec8273 | flycap-markerless-motion-capture-using | 1610.09534 | null | http://arxiv.org/abs/1610.09534v3 | http://arxiv.org/pdf/1610.09534v3.pdf | FlyCap: Markerless Motion Capture Using Multiple Autonomous Flying Cameras | Aiming at automatic, convenient and non-instrusive motion capture, this paper
presents a new generation markerless motion capture technique, the FlyCap
system, to capture surface motions of moving characters using multiple
autonomous flying cameras (autonomous unmanned aerial vehicles(UAV) each
integrated with an RGBD video camera). During data capture, three cooperative
flying cameras automatically track and follow the moving target who performs
large scale motions in a wide space. We propose a novel non-rigid surface
registration method to track and fuse the depth of the three flying cameras for
surface motion tracking of the moving target, and simultaneously calculate the
pose of each flying camera. We leverage the using of visual-odometry
information provided by the UAV platform, and formulate the surface tracking
problem in a non-linear objective function that can be linearized and
effectively minimized through a Gaussian-Newton method. Quantitative and
qualitative experimental results demonstrate the competent and plausible
surface and motion reconstruction results | ['Wei Cheng', 'Lu Fang', 'Kaiwen Guo', 'Lan Xu', 'Guyue Zhou', 'Yebin Liu', 'Qionghai Dai'] | 2016-10-29 | null | null | null | null | ['markerless-motion-capture'] | ['computer-vision'] | [ 2.11733043e-01 -3.53582621e-01 5.34989275e-02 1.77530020e-01
-4.08240825e-01 -1.09527314e+00 4.28806484e-01 -6.92546844e-01
-5.88818789e-01 3.03024441e-01 -4.01161760e-01 3.48073065e-01
8.93019512e-02 -2.60313064e-01 -7.17043161e-01 -4.63739574e-01
1.46842688e-01 3.89879286e-01 7.42925525e-01 -4.35917191e-02
1.22705594e-01 7.47230887e-01 -1.47025859e+00 -4.28287685e-01
5.48489511e-01 8.23224425e-01 5.42136312e-01 1.21076310e+00
8.02914381e-01 5.53227007e-01 -2.61643857e-01 -3.77106726e-01
5.22598028e-01 -5.39934300e-02 -3.15479875e-01 5.68674266e-01
8.80124748e-01 -6.18962646e-01 -4.31432664e-01 1.11112475e+00
1.00543708e-01 4.21487302e-01 2.38074809e-01 -1.25286210e+00
-1.12105682e-01 -4.18006212e-01 -6.33391738e-01 -1.05884187e-01
1.04754865e+00 8.56889188e-02 4.82911706e-01 -8.03490162e-01
1.14413798e+00 9.49613512e-01 8.30317378e-01 6.09482706e-01
-7.42707372e-01 -1.12999596e-01 1.16364941e-01 -1.72598615e-01
-1.66555119e+00 -2.68388599e-01 8.31683815e-01 -6.27565324e-01
4.33226794e-01 3.80497158e-01 1.19541681e+00 7.84930944e-01
4.13352758e-01 5.04107237e-01 4.52341020e-01 -2.64689535e-01
7.44214579e-02 -2.92885244e-01 -3.20031792e-01 1.10326028e+00
3.72756332e-01 1.32701099e-01 -6.02718174e-01 -1.09364532e-01
1.39112568e+00 4.04433310e-01 -6.03362978e-01 -9.87359345e-01
-1.40706432e+00 4.99121934e-01 1.38653174e-01 -1.58417284e-01
-4.19890344e-01 3.33761930e-01 -4.69399124e-01 -1.33670598e-01
2.20724300e-01 8.41227025e-02 -1.76468074e-01 -5.21899819e-01
-1.05982625e+00 3.10632020e-01 5.59309661e-01 1.37215555e+00
7.31718421e-01 2.87684709e-01 5.31244218e-01 -2.70990103e-01
7.50266492e-01 1.13317275e+00 2.35463724e-01 -1.42134726e+00
2.54281074e-01 8.51347089e-01 6.43730044e-01 -1.21238816e+00
-1.90865293e-01 3.48303318e-01 -2.60491967e-01 4.16537970e-01
2.48086900e-01 -4.12504077e-01 -5.10382891e-01 1.13873911e+00
8.18780839e-01 4.84575570e-01 -7.21114576e-02 1.39445531e+00
2.54850507e-01 3.97832096e-01 -4.56497043e-01 -2.33024135e-01
1.01427794e+00 -7.64185965e-01 -7.38100827e-01 -4.01644200e-01
3.83398086e-01 -6.81228817e-01 7.84577549e-01 4.23659205e-01
-1.20762730e+00 -2.98900902e-01 -1.15393984e+00 -1.17093489e-01
1.98760867e-01 4.17738557e-01 4.44199413e-01 2.58098453e-01
-9.23366427e-01 3.93536568e-01 -1.43975163e+00 -2.89477885e-01
-9.63282064e-02 4.88760650e-01 -6.90093994e-01 1.09409774e-03
-6.02275372e-01 8.24434161e-01 -8.75676572e-02 3.54222864e-01
-1.18126249e+00 -3.70447099e-01 -9.13451910e-01 -6.69284165e-01
4.81070548e-01 -9.26199019e-01 1.00348687e+00 -9.89689231e-01
-1.90140820e+00 9.75120842e-01 -1.42771617e-01 -6.37438847e-03
6.82232857e-01 -9.40016866e-01 9.50327236e-03 4.71255332e-01
3.87872495e-02 4.28125143e-01 7.85231650e-01 -1.27459109e+00
-5.87394834e-01 -6.60734355e-01 -4.59122472e-02 6.90456271e-01
9.07025242e-04 4.63057645e-02 -9.27743077e-01 -2.25582436e-01
2.79918462e-01 -1.42862916e+00 -2.63858944e-01 4.93734807e-01
-3.01193476e-01 5.54884970e-01 1.14662325e+00 -5.22030830e-01
9.30935919e-01 -1.89310038e+00 8.37903500e-01 -8.14008638e-02
1.20380342e-01 2.28039280e-01 3.29133421e-01 -1.03423648e-01
6.85196996e-01 -6.47972941e-01 -2.57654458e-01 -4.80288446e-01
-4.26347047e-01 6.90004751e-02 -1.78448394e-01 9.73613560e-01
-3.45007926e-01 8.36040974e-01 -9.21747565e-01 -2.90511787e-01
3.69491994e-01 9.36472654e-01 -3.35979491e-01 4.52220410e-01
4.29740809e-02 4.92976308e-01 -6.87832355e-01 1.00378656e+00
5.09478867e-01 9.81619135e-02 -9.46771726e-02 -7.62527585e-02
-4.26571906e-01 -5.07483006e-01 -1.56793439e+00 2.21821690e+00
5.75101823e-02 6.51074648e-01 5.86617112e-01 8.90218467e-02
7.59130001e-01 1.12991475e-01 5.94838381e-01 1.86147153e-01
3.31375211e-01 -1.57386865e-04 -7.79538572e-01 -3.48810613e-01
8.83563995e-01 7.49247149e-02 -7.98769444e-02 -3.86097133e-02
-1.25799105e-01 -4.04595107e-01 -3.31154048e-01 -1.30102232e-01
8.93524826e-01 7.80774593e-01 1.56899929e-01 4.78355438e-02
4.82375741e-01 5.68087816e-01 5.68428576e-01 9.70846601e-03
-1.67981759e-01 8.92631471e-01 -4.50200289e-01 -4.75414187e-01
-8.24240208e-01 -1.06134927e+00 2.75331587e-01 5.38698196e-01
8.26065302e-01 -2.91517377e-01 -1.07983541e+00 -7.45370463e-02
-1.50347147e-02 -2.79614292e-02 -4.26984131e-01 -2.50280648e-02
-7.39914000e-01 -2.19106480e-01 3.28132808e-01 3.98774058e-01
3.96908551e-01 -3.94308031e-01 -1.57816648e+00 1.95410233e-02
-3.78600396e-02 -1.37458587e+00 -7.63464451e-01 -3.76007557e-01
-8.57616305e-01 -1.29602623e+00 -7.82501161e-01 -5.09893477e-01
6.11824811e-01 7.86710382e-01 5.09787798e-01 -2.38575131e-01
-2.62592167e-01 1.03643095e+00 -1.66125983e-01 -7.50201568e-02
1.18438534e-01 -5.22338033e-01 5.64179540e-01 2.04790518e-01
-7.47542232e-02 -2.30004996e-01 -6.50635242e-01 5.12703538e-01
-5.53633332e-01 -5.11150174e-02 1.41529351e-01 5.60762249e-02
1.01504910e+00 -2.92359918e-01 -8.37848842e-01 -1.41439557e-01
3.31417262e-03 -8.75892267e-02 -1.10265410e+00 1.48162618e-01
-1.89812947e-02 -6.73572123e-01 1.85609907e-01 -7.50067890e-01
-7.11304665e-01 1.00672793e+00 5.36603034e-01 -1.10597873e+00
2.92723272e-02 -2.03291662e-02 -3.91063094e-01 -7.23266423e-01
3.76532644e-01 3.54355305e-01 9.40321535e-02 -9.85425636e-02
5.80106378e-01 3.02391142e-01 1.05821955e+00 -8.49651396e-02
1.16530490e+00 1.12352395e+00 1.78414166e-01 -1.08651197e+00
-5.19783676e-01 -4.61330444e-01 -1.46628308e+00 -8.08871329e-01
1.37846076e+00 -1.18699718e+00 -1.10065532e+00 7.78651595e-01
-1.21291661e+00 -2.88269073e-01 -1.89018697e-01 8.04090917e-01
-8.36423516e-01 5.39801955e-01 -3.36176366e-01 -9.75177944e-01
-3.34060073e-01 -1.10558271e+00 1.63806272e+00 4.31702316e-01
-1.87635690e-01 -1.25001442e+00 5.98856151e-01 3.18711191e-01
-2.23008081e-01 8.55445743e-01 -4.78785723e-01 4.70956653e-01
-1.08813286e+00 -5.66287875e-01 5.03450990e-01 -1.81385800e-01
-3.60260941e-02 5.10273933e-01 -6.43433928e-01 -4.76558387e-01
1.91637412e-01 1.47944123e-01 1.62014097e-01 5.42052329e-01
-1.21630520e-01 -9.67725441e-02 -5.28330803e-01 1.35471261e+00
1.56330228e+00 1.54805675e-01 3.03192824e-01 6.29149318e-01
1.32925975e+00 5.32656193e-01 1.05546463e+00 4.93576825e-01
6.33564115e-01 1.17598701e+00 7.06786633e-01 1.46074638e-01
3.05014819e-01 -2.63418853e-01 1.03251624e+00 8.04535747e-01
-5.30914485e-01 4.79075834e-02 -9.04602110e-01 2.93444395e-01
-1.76789749e+00 -7.60357678e-01 -6.29459858e-01 2.40034890e+00
2.55379170e-01 -3.64718348e-01 1.22519121e-01 -2.18482763e-01
6.58239067e-01 -1.57424822e-01 -5.58563054e-01 2.14811027e-01
-1.70003638e-01 -4.51659024e-01 7.09227026e-01 8.34722281e-01
-1.14299619e+00 1.33441043e+00 6.24404335e+00 -2.57119108e-02
-1.19654262e+00 -1.86420195e-02 -5.04581094e-01 -2.32810199e-01
1.11750416e-01 2.09802225e-01 -9.84011054e-01 3.53434592e-01
7.90452898e-01 -1.50033414e-01 2.93371975e-01 9.54812706e-01
9.13718119e-02 -2.53251582e-01 -8.30573738e-01 1.01305175e+00
3.45650911e-01 -1.16413569e+00 -1.10834159e-01 2.39474401e-01
8.13582957e-01 7.75507689e-02 9.60326567e-02 -6.50689244e-01
3.05975407e-01 -3.67350399e-01 1.10856712e+00 1.04850566e+00
7.34654367e-01 -4.82327044e-01 2.25558281e-01 6.20815277e-01
-1.52762961e+00 7.62720359e-03 -4.51604247e-01 -2.42726132e-01
5.97704530e-01 2.12318497e-03 -3.16845924e-01 4.93621528e-01
5.76756001e-01 8.12640131e-01 -4.34783131e-01 8.93086016e-01
-1.42382622e-01 -1.15362465e-01 -3.49339604e-01 2.34341592e-01
7.90048987e-02 -6.94666147e-01 1.05505276e+00 5.55340469e-01
6.41325712e-01 3.25955927e-01 3.13028336e-01 7.70896792e-01
4.20642614e-01 -3.14245611e-01 -7.88393259e-01 2.61762202e-01
2.81262994e-01 1.48185754e+00 -7.57332027e-01 1.15334883e-01
-2.27647483e-01 1.53378773e+00 3.86194289e-02 6.94913194e-02
-8.65290523e-01 -1.77931003e-02 8.22598040e-01 3.11945349e-01
2.15555772e-01 -9.72558677e-01 2.26906925e-01 -1.60129964e+00
2.22674116e-01 -2.71895438e-01 -7.08662644e-02 -1.15154552e+00
-4.15831298e-01 6.60525501e-01 -1.14675291e-01 -1.80217493e+00
-1.62802249e-01 -5.32684386e-01 -6.31823063e-01 5.08315563e-01
-1.11073601e+00 -1.48916972e+00 -7.10449159e-01 1.08328342e+00
3.59215230e-01 -7.36640543e-02 7.34213769e-01 -1.27183318e-01
-5.87228477e-01 7.42024407e-02 4.34023142e-02 -4.31547537e-02
4.26204085e-01 -1.09497476e+00 2.53002405e-01 1.39608359e+00
3.61010544e-02 6.52028382e-01 5.19652903e-01 -9.73966718e-01
-2.34333491e+00 -1.05756927e+00 2.94046998e-01 -1.24218249e+00
5.96865237e-01 -4.70700026e-01 -5.83236635e-01 1.04491663e+00
-4.34527427e-01 1.57789350e-01 2.77089477e-01 -8.82608354e-01
4.33316141e-01 1.35223195e-01 -1.01964426e+00 4.41711426e-01
9.07649100e-01 -4.13722694e-01 -5.85153818e-01 1.21356539e-01
7.01281369e-01 -1.07440901e+00 -8.15775335e-01 -4.56163175e-02
6.77740216e-01 -5.32730043e-01 1.09200108e+00 -1.37289539e-01
-6.92426488e-02 -8.00760210e-01 -4.30186957e-01 -7.84992397e-01
-9.59256813e-02 -1.30556047e+00 -4.14339125e-01 9.40325379e-01
-3.67204994e-01 -1.55869126e-02 1.26002312e+00 9.37678397e-01
-1.76736563e-01 -3.57385397e-01 -1.16760039e+00 -6.08813584e-01
-5.30847669e-01 -1.85083896e-01 -5.29655665e-02 7.59025395e-01
-1.18998371e-01 2.51127612e-02 -7.06233084e-01 7.35393822e-01
9.22262073e-01 -7.00820587e-04 1.09337294e+00 -1.30543411e+00
7.28307292e-02 9.39066336e-02 -9.20333028e-01 -1.41061032e+00
9.67269316e-02 -3.65818292e-01 1.50752366e-01 -1.15584624e+00
-2.36115351e-01 2.83776313e-01 7.31225967e-01 -1.18001826e-01
-1.98023379e-01 1.87930912e-01 2.60398716e-01 6.82101786e-01
-8.25701714e-01 6.00898623e-01 1.17297423e+00 4.08720225e-01
-2.98267633e-01 1.19974367e-01 -1.22850947e-02 1.21491754e+00
-1.25686139e-01 -3.40894192e-01 -2.59304374e-01 -7.25815296e-01
1.90562800e-01 4.75011855e-01 7.05676019e-01 -1.33684778e+00
7.85978615e-01 -2.75793135e-01 3.97558123e-01 -5.28549194e-01
8.77524972e-01 -1.29988837e+00 6.04253411e-01 5.95215559e-01
2.36641809e-01 6.82273507e-01 2.79286224e-02 7.41455913e-01
-7.11484477e-02 5.51050305e-02 5.52846968e-01 -5.63870966e-02
-9.02940810e-01 5.69430172e-01 -4.87058878e-01 -6.32603690e-02
1.62172234e+00 -7.61531591e-01 -3.11361309e-02 -2.91899502e-01
-6.38450086e-01 2.86839642e-02 1.30208397e+00 3.14107180e-01
1.13769519e+00 -1.37409365e+00 -3.43212426e-01 3.76900524e-01
-2.16841444e-01 3.80898774e-01 2.99670428e-01 9.12198901e-01
-1.17479491e+00 2.40880370e-01 -1.73438370e-01 -1.03414023e+00
-1.38250601e+00 3.12138677e-01 5.51091790e-01 3.85408461e-01
-7.60647833e-01 9.21089232e-01 -1.26222238e-01 -4.06037837e-01
-1.51040509e-01 -2.81835049e-01 -3.90351266e-02 -3.21122080e-01
5.57188332e-01 7.45745480e-01 -2.99205095e-01 -1.48302078e+00
-5.50582409e-01 1.56652689e+00 4.44848329e-01 -3.51221085e-01
1.16359115e+00 -6.05314016e-01 1.80583969e-01 4.67460275e-01
8.98963392e-01 1.61762729e-01 -2.01493526e+00 8.58954340e-02
-2.70450890e-01 -7.43074059e-01 3.22588086e-02 -1.56890396e-02
-8.34675491e-01 7.07546651e-01 5.90453207e-01 -3.19440871e-01
8.76541853e-01 -3.57191890e-01 7.38976300e-01 3.57268631e-01
8.12794328e-01 -8.39082420e-01 -1.37627527e-01 5.07814288e-01
5.47892094e-01 -8.82753611e-01 1.69689596e-01 -5.13323843e-01
-8.75841737e-01 1.23278201e+00 6.13005996e-01 -5.29521763e-01
2.99962252e-01 4.63089436e-01 2.42516741e-01 -2.24704906e-01
-4.67762828e-01 1.83867916e-01 4.20692801e-01 8.55450332e-01
-1.90048158e-01 -5.54894134e-02 4.93676424e-01 3.80408913e-01
-2.29020536e-01 -3.64984013e-02 8.71236086e-01 1.28541899e+00
-5.27709901e-01 -6.41569555e-01 -6.09318256e-01 -4.96090442e-01
-1.36788070e-01 2.81447738e-01 -7.95416951e-01 8.97585034e-01
-6.34090528e-02 7.43005097e-01 1.72581419e-01 -5.91664851e-01
7.51852751e-01 -3.44044030e-01 7.42724240e-01 -5.12433767e-01
-4.21774805e-01 2.77877659e-01 -4.50819820e-01 -1.05683196e+00
-6.38403833e-01 -1.00329542e+00 -1.33040285e+00 -1.57597587e-01
-4.75019604e-01 -1.80568367e-01 6.39302313e-01 6.78332448e-01
2.00162679e-01 4.58071493e-02 5.81649601e-01 -1.52454686e+00
-2.14110792e-01 -3.48871887e-01 -7.21090138e-01 2.40168646e-01
4.76270735e-01 -8.77413750e-01 -3.50934774e-01 4.50801164e-01] | [7.297806262969971, -1.374818205833435] |
308d49a0-87b0-4e2f-9e12-a0260ed5c2d9 | llmscore-unveiling-the-power-of-large | 2305.11116 | null | https://arxiv.org/abs/2305.11116v1 | https://arxiv.org/pdf/2305.11116v1.pdf | LLMScore: Unveiling the Power of Large Language Models in Text-to-Image Synthesis Evaluation | Existing automatic evaluation on text-to-image synthesis can only provide an image-text matching score, without considering the object-level compositionality, which results in poor correlation with human judgments. In this work, we propose LLMScore, a new framework that offers evaluation scores with multi-granularity compositionality. LLMScore leverages the large language models (LLMs) to evaluate text-to-image models. Initially, it transforms the image into image-level and object-level visual descriptions. Then an evaluation instruction is fed into the LLMs to measure the alignment between the synthesized image and the text, ultimately generating a score accompanied by a rationale. Our substantial analysis reveals the highest correlation of LLMScore with human judgments on a wide range of datasets (Attribute Binding Contrast, Concept Conjunction, MSCOCO, DrawBench, PaintSkills). Notably, our LLMScore achieves Kendall's tau correlation with human evaluations that is 58.8% and 31.2% higher than the commonly-used text-image matching metrics CLIP and BLIP, respectively. | ['William Yang Wang', 'Xin Eric Wang', 'Xiujun Li', 'Xianjun Yang', 'Yujie Lu'] | 2023-05-18 | null | null | null | null | ['text-matching'] | ['natural-language-processing'] | [ 5.00533283e-01 -1.78632453e-01 -1.82898238e-01 -3.37236017e-01
-9.67689395e-01 -6.31024837e-01 9.07966912e-01 2.37081528e-01
-1.51640236e-01 2.02510700e-01 1.69260949e-01 -1.85657144e-01
1.14648826e-01 -5.98379076e-01 -6.92972898e-01 -3.01687270e-01
4.72498238e-01 2.43122831e-01 4.26613718e-01 -1.67558957e-02
4.83540773e-01 2.33821571e-01 -1.92772412e+00 1.01977825e+00
9.26090419e-01 1.35878348e+00 3.55489105e-01 6.95108891e-01
-3.78659129e-01 1.07259226e+00 -5.57159424e-01 -8.70293379e-01
1.70793951e-01 -4.77765650e-01 -6.42296016e-01 1.65030569e-01
1.22827375e+00 -3.38528872e-01 1.05516762e-01 1.02543604e+00
3.93358827e-01 -2.63651460e-01 8.43952656e-01 -1.57518232e+00
-6.76072299e-01 4.96310979e-01 -7.72428870e-01 -1.85228735e-01
6.93909645e-01 4.05670434e-01 1.26198852e+00 -1.08764958e+00
8.57170641e-01 1.43221939e+00 5.77766955e-01 -2.37862580e-02
-1.24645221e+00 -6.93959534e-01 -8.69155526e-02 9.65274796e-02
-1.30918324e+00 -3.15440446e-01 2.66466886e-01 -6.45186424e-01
1.05663204e+00 5.10168195e-01 5.09892762e-01 8.17221642e-01
2.75929064e-01 7.08226442e-01 1.46683347e+00 -5.72128773e-01
3.07706697e-03 3.42309177e-01 -2.91796446e-01 6.40256345e-01
7.18216971e-02 5.95595352e-02 -6.86539292e-01 1.33125216e-01
7.60246158e-01 -1.53611168e-01 1.33671433e-01 -2.27334440e-01
-1.39225948e+00 3.65136236e-01 2.54431814e-01 3.23000938e-01
-1.99313834e-01 1.60564452e-01 4.08919930e-01 3.16840470e-01
2.19253197e-01 4.44687992e-01 -5.66114672e-02 -6.31885091e-03
-1.13649487e+00 2.80177504e-01 4.60720360e-01 1.13076258e+00
7.67722666e-01 -2.97327265e-02 -7.29198754e-01 8.89467418e-01
9.76524781e-03 8.41030538e-01 3.32170784e-01 -1.04875565e+00
4.79500473e-01 8.09846640e-01 2.91389897e-02 -1.18885911e+00
-2.22757265e-01 -4.89774734e-01 -6.20567203e-01 4.23669338e-01
1.62919238e-01 5.07323086e-01 -5.26305139e-01 1.51653600e+00
2.18614321e-02 -9.13740695e-02 -1.24669604e-01 1.01355970e+00
1.05095327e+00 4.13483977e-01 3.22380513e-01 7.20681399e-02
1.66444814e+00 -1.08607197e+00 -5.14311373e-01 -2.34585673e-01
4.66166943e-01 -1.19237006e+00 1.79199624e+00 3.15639555e-01
-1.38879561e+00 -9.10838902e-01 -1.16090477e+00 -8.22588652e-02
-3.12274784e-01 3.70367497e-01 2.10712343e-01 5.06258726e-01
-1.19992125e+00 3.02018046e-01 -4.83256504e-02 -4.83625710e-01
2.09156901e-01 -1.06119560e-02 -3.25174630e-01 -8.46580863e-02
-8.19670260e-01 9.10108268e-01 3.38319212e-01 -5.42040229e-01
-8.18636537e-01 -6.93540871e-01 -6.61495745e-01 -7.01876432e-02
3.30475271e-01 -8.13452125e-01 1.12436664e+00 -1.10148287e+00
-1.18156230e+00 1.36079323e+00 2.96702813e-02 -2.88168848e-01
7.35142767e-01 1.46249048e-02 -5.25907934e-01 3.08111429e-01
4.15529668e-01 1.07724011e+00 9.54909563e-01 -1.43911469e+00
-9.11838055e-01 -1.03591859e-01 1.49886459e-01 3.18700224e-01
-3.09882224e-01 1.81387052e-01 -8.74698997e-01 -7.19586790e-01
-8.43554735e-02 -6.55316114e-01 2.95750797e-01 1.65966764e-01
-3.61543179e-01 -1.14229945e-02 4.63762850e-01 -4.88622695e-01
1.35554731e+00 -2.15583825e+00 -1.98016129e-02 1.43988803e-01
2.53362626e-01 1.38389468e-01 -4.71231222e-01 6.28249347e-01
-2.45600822e-03 4.65087295e-02 -9.21219066e-02 -4.08604562e-01
2.59486049e-01 -2.49685675e-01 -1.96398661e-01 1.13833912e-01
2.07682073e-01 1.22268522e+00 -8.34453881e-01 -9.74120736e-01
4.00977880e-01 8.57664421e-02 -5.35691202e-01 2.66838998e-01
-3.50782007e-01 -1.40640363e-01 -6.22399268e-04 7.54989326e-01
6.55182838e-01 -4.37783271e-01 -5.77991654e-04 -5.89296997e-01
-1.72025919e-01 -2.74719477e-01 -9.29555297e-01 1.71739316e+00
-4.30074990e-01 8.10249209e-01 -2.54425079e-01 -3.87720615e-01
8.74967098e-01 9.68242958e-02 3.23357671e-01 -1.39885366e+00
-5.10733165e-02 2.56794959e-01 -2.28084117e-01 -5.88644385e-01
6.88501835e-01 -4.41396795e-02 -1.42176107e-01 4.84225750e-01
-1.27697214e-01 -5.06340444e-01 3.24421763e-01 4.09039319e-01
8.16419423e-01 3.01999450e-01 2.95632064e-01 -3.15227509e-01
7.36394405e-01 2.01722570e-02 -7.40810186e-02 7.97524750e-01
-1.39083475e-01 9.50696647e-01 5.98553598e-01 -2.10252866e-01
-1.28636277e+00 -1.24256444e+00 -1.66590232e-02 1.22709739e+00
3.40863049e-01 -5.91609061e-01 -1.08646142e+00 -3.56907934e-01
1.18975807e-03 6.46523237e-01 -5.74387610e-01 6.97773099e-02
-4.34044469e-03 -2.42371336e-01 7.24955142e-01 4.26824957e-01
6.03611171e-01 -1.05353022e+00 -7.27756321e-01 -1.99558452e-01
-2.95590281e-01 -1.56564689e+00 -6.47862077e-01 -3.94093782e-01
-5.74447334e-01 -9.47968364e-01 -5.08872986e-01 -7.83361375e-01
5.80583096e-01 2.07846865e-01 1.50277686e+00 2.35610574e-01
-4.54345554e-01 2.48211086e-01 -2.98714072e-01 -7.10186884e-02
-4.69061613e-01 -2.31721222e-01 -2.83608586e-01 2.31111869e-02
1.11972012e-01 -1.69866040e-01 -8.13224256e-01 5.81628621e-01
-1.31456053e+00 7.06800997e-01 9.37202156e-01 6.13609016e-01
7.12640643e-01 -1.53219968e-01 2.88610220e-01 -6.07885897e-01
6.29270196e-01 2.82023903e-02 -5.43500304e-01 6.07952714e-01
-7.94878483e-01 -1.02661617e-01 4.06197071e-01 -4.45477575e-01
-8.26991677e-01 -1.88696757e-01 2.04526871e-01 -3.48360300e-01
-3.47909480e-02 3.63981336e-01 3.37203778e-03 -4.18885238e-02
6.95423961e-01 2.38133863e-01 -1.23416848e-01 9.00251139e-03
4.94767010e-01 6.85319960e-01 8.28174770e-01 -7.10575998e-01
6.96153462e-01 4.17810857e-01 -1.60798430e-01 -5.20684183e-01
-7.45532393e-01 -3.77790213e-01 -5.50586760e-01 -5.53672254e-01
1.00496411e+00 -8.92481327e-01 -7.50607133e-01 4.05162185e-01
-1.06255019e+00 -3.32952052e-01 -1.22916318e-01 2.32124433e-01
-7.14845181e-01 3.50318968e-01 -4.66806710e-01 -5.52457452e-01
-4.66172516e-01 -1.24943364e+00 1.50000036e+00 -4.85623181e-02
-4.12594527e-01 -5.73493779e-01 -3.90017062e-01 6.71468794e-01
3.93985659e-01 3.26624990e-01 9.51228201e-01 -1.23773962e-01
-5.12016177e-01 -1.85915858e-01 -9.01908875e-01 2.77433455e-01
-2.01830298e-01 1.35208160e-01 -9.30045664e-01 -3.80210429e-02
-3.44781190e-01 -4.13613826e-01 4.62394774e-01 2.70596147e-01
1.04436755e+00 -8.72403905e-02 -3.70958410e-02 4.39718425e-01
1.51317275e+00 7.58489519e-02 1.00534034e+00 4.97648269e-01
6.40904844e-01 7.25708544e-01 8.11794758e-01 4.37923044e-01
4.73227113e-01 9.07570004e-01 2.86849946e-01 -4.01252776e-01
-6.21720970e-01 -3.90225083e-01 4.63353038e-01 7.38735378e-01
1.74472392e-01 -2.78280973e-01 -9.72366452e-01 2.03858897e-01
-1.75768864e+00 -8.77018571e-01 -3.22467417e-01 2.06785178e+00
7.21363783e-01 1.74537271e-01 6.04330152e-02 -4.99584638e-02
6.08115554e-01 -3.32488753e-02 -4.00128871e-01 -4.16485876e-01
-4.34325486e-01 2.31806487e-02 3.72649729e-01 4.01861221e-01
-8.45265448e-01 1.08328176e+00 6.55297279e+00 1.28171909e+00
-1.07399583e+00 1.17649816e-01 7.86464393e-01 -5.49729615e-02
-4.72716272e-01 2.12051626e-02 -4.20541763e-01 4.11691487e-01
7.41737425e-01 -8.20038766e-02 3.22261870e-01 5.48759818e-01
1.29321873e-01 -2.51634538e-01 -1.20127285e+00 1.31958377e+00
3.49439383e-01 -1.22432530e+00 4.29103374e-01 -1.18174464e-01
1.06270027e+00 -3.11909616e-01 3.04748684e-01 8.43345448e-02
-5.45053184e-02 -1.19551444e+00 1.33374548e+00 6.24925375e-01
1.24718738e+00 -4.91555899e-01 5.57125747e-01 2.33395845e-02
-1.30257344e+00 1.87597841e-01 -8.93970281e-02 9.89862755e-02
-2.22548880e-02 3.35742533e-01 -7.29377747e-01 4.65620100e-01
6.25615954e-01 5.21960318e-01 -1.05719507e+00 6.90103889e-01
-7.84314945e-02 7.51664564e-02 2.14180171e-01 5.69279306e-02
1.92947850e-01 -8.57321844e-02 1.95797101e-01 1.48625433e+00
2.48768508e-01 -3.08980763e-01 2.82902271e-01 1.12240314e+00
-8.65137111e-03 5.27512491e-01 -3.61125767e-01 -1.84799716e-01
5.31161070e-01 1.26922762e+00 -8.19041789e-01 -6.69042408e-01
-5.50539792e-01 1.22147167e+00 3.00792586e-02 2.44340718e-01
-1.04870915e+00 -3.31081003e-01 2.82208353e-01 3.96026552e-01
9.16094985e-03 1.02108248e-01 -6.02261364e-01 -8.89077187e-01
1.06431939e-01 -1.19713938e+00 2.60661483e-01 -1.52947831e+00
-1.16490161e+00 6.85650170e-01 2.16847777e-01 -1.51996946e+00
-1.09352410e-01 -5.41383147e-01 -4.05931443e-01 7.68376529e-01
-1.26242077e+00 -1.54368258e+00 -7.86000490e-01 5.39274752e-01
5.01158297e-01 -1.79374933e-01 6.34807527e-01 3.78494829e-01
-3.62218827e-01 8.10347736e-01 -3.13506454e-01 -7.36326948e-02
9.57729399e-01 -1.11236179e+00 4.46024388e-01 7.67473996e-01
7.65587613e-02 4.34449315e-01 6.89633846e-01 -4.86243606e-01
-1.36483407e+00 -8.52985263e-01 8.23342562e-01 -4.95360732e-01
6.70371056e-01 -2.15715647e-01 -6.30349755e-01 1.44589767e-01
3.90874416e-01 -2.11459622e-01 3.75314295e-01 -3.70534658e-01
-8.06569338e-01 -2.84075230e-01 -1.05538607e+00 9.28816080e-01
1.11459529e+00 -8.98460269e-01 -3.26745272e-01 1.21348955e-01
6.90654159e-01 -3.81893367e-01 -1.09118474e+00 3.75145227e-01
1.03956914e+00 -1.29935110e+00 1.14082491e+00 -1.34233370e-01
9.62556660e-01 -4.85332429e-01 -5.40649712e-01 -7.14797676e-01
-2.68244624e-01 -2.98199803e-01 3.15053374e-01 1.20850050e+00
4.73569155e-01 -3.29794914e-01 4.00529653e-01 4.71370280e-01
-1.35836229e-01 -5.72585583e-01 -4.33356136e-01 -8.04357648e-01
-1.82157069e-01 -7.10543692e-01 6.94976270e-01 7.90328383e-01
-8.25275630e-02 2.21319154e-01 -3.08094114e-01 -2.54984796e-01
6.81232512e-01 3.10403109e-01 1.03683543e+00 -9.16779995e-01
-2.18637645e-01 -9.36806858e-01 -6.08803332e-01 -7.08105326e-01
-1.00756422e-01 -1.03347278e+00 -5.84009252e-02 -1.64520609e+00
6.75075889e-01 -1.36669412e-01 -1.95997909e-01 3.63713861e-01
-8.64528120e-02 7.41942823e-01 4.06764209e-01 3.57217193e-01
-9.19774890e-01 9.84453112e-02 1.42885828e+00 -3.14422011e-01
1.87629342e-01 -7.15947926e-01 -6.69869244e-01 5.55282593e-01
5.61062515e-01 -2.02297941e-01 -3.40757877e-01 -4.65352029e-01
3.29835594e-01 -2.92223040e-02 5.15478432e-01 -1.24887681e+00
1.34471565e-01 -2.88530827e-01 3.99792731e-01 -6.18375897e-01
1.14521809e-01 -6.78094685e-01 4.55017269e-01 5.29200375e-01
-5.71958601e-01 2.59359211e-01 2.04416916e-01 1.71798497e-01
-3.57368410e-01 -5.01372851e-02 7.72942424e-01 6.18675128e-02
-1.01114833e+00 1.01935603e-01 1.85261201e-02 6.24555312e-02
9.45037365e-01 -5.38702965e-01 -5.56908250e-01 -3.22810918e-01
-2.47897401e-01 -3.57749686e-02 9.09334362e-01 6.99326277e-01
7.38005340e-01 -1.58371639e+00 -7.83247948e-01 2.25224063e-01
8.36650550e-01 -5.89138448e-01 2.78265506e-01 9.52268004e-01
-6.44651771e-01 4.51443017e-01 -3.91625553e-01 -9.33903694e-01
-1.43068767e+00 4.18101996e-01 1.02106981e-01 -1.13796599e-01
-4.75358993e-01 4.55293864e-01 5.94429493e-01 -7.98367783e-02
1.53338507e-01 -5.46876371e-01 1.45020351e-01 9.35949758e-02
4.89568532e-01 2.69507915e-01 -1.36374114e-02 -8.77519906e-01
-3.00758570e-01 9.86695528e-01 1.28519893e-01 -4.34864730e-01
7.78946340e-01 -1.88083366e-01 -2.11178422e-01 2.41112083e-01
1.21718311e+00 -2.48239338e-02 -1.06020927e+00 -1.75769284e-01
4.67450656e-02 -5.85972846e-01 -1.14270858e-01 -1.12829566e+00
-7.97773063e-01 7.38789618e-01 7.73153126e-01 3.81941274e-02
1.23005617e+00 -6.35990233e-04 6.88606262e-01 -1.23814717e-02
3.71740758e-01 -1.20351171e+00 7.28053510e-01 2.80334294e-01
1.03672040e+00 -1.29000115e+00 7.73910508e-02 -3.37338388e-01
-1.05160737e+00 8.83800209e-01 7.25024998e-01 1.61849618e-01
5.11372313e-02 3.60507905e-01 8.90123919e-02 -2.62794465e-01
-7.14510262e-01 -2.72770345e-01 8.17112029e-01 4.64531928e-01
6.16922855e-01 2.17804030e-01 -2.27858052e-01 3.47410202e-01
-3.17755401e-01 -7.37904459e-02 5.40047772e-02 5.76766670e-01
-5.40817976e-01 -9.49198484e-01 -3.96403462e-01 4.01697338e-01
-3.09819996e-01 -2.78128296e-01 -4.92207885e-01 5.95136344e-01
1.62045941e-01 1.00915337e+00 1.94982111e-01 -6.62035227e-01
5.49129605e-01 -1.46425948e-01 6.07354522e-01 -4.12224025e-01
-8.46425414e-01 1.15495957e-01 4.02812921e-02 -8.45380425e-01
-4.43855196e-01 -3.28217983e-01 -1.21122909e+00 -3.97671640e-01
-5.82309291e-02 -3.27915609e-01 7.41779625e-01 6.96950197e-01
3.81190032e-01 4.23981100e-01 2.85978585e-01 -6.36957109e-01
-5.50354607e-02 -8.41852963e-01 -3.57499897e-01 1.06057477e+00
-1.54625271e-02 -4.50648904e-01 -1.82078108e-01 2.49603763e-01] | [11.132686614990234, 0.9965345859527588] |
9b849a7e-50de-4ec1-9981-6b92ab3334a2 | prior-knowledge-and-memory-enriched | null | null | https://aclanthology.org/2022.findings-acl.297 | https://aclanthology.org/2022.findings-acl.297.pdf | Prior Knowledge and Memory Enriched Transformer for Sign Language Translation | This paper attacks the challenging problem of sign language translation (SLT), which involves not only visual and textual understanding but also additional prior knowledge learning (i.e. performing style, syntax). However, the majority of existing methods with vanilla encoder-decoder structures fail to sufficiently explore all of them. Based on this concern, we propose a novel method called Prior knowledge and memory Enriched Transformer (PET) for SLT, which incorporates the auxiliary information into vanilla transformer. Concretely, we develop gated interactive multi-head attention which associates the multimodal representation and global signing style with adaptive gated functions. One Part-of-Speech (POS) sequence generator relies on the associated information to predict the global syntactic structure, which is thereafter leveraged to guide the sentence generation. Besides, considering that the visual-textual context information, and additional auxiliary knowledge of a word may appear in more than one video, we design a multi-stream memory structure to obtain higher-quality translations, which stores the detailed correspondence between a word and its various relevant information, leading to a more comprehensive understanding for each word. We conduct extensive empirical studies on RWTH-PHOENIX-Weather-2014 dataset with both signer-dependent and signer-independent conditions. The quantitative and qualitative experimental results comprehensively reveal the effectiveness of PET. | ['Xingshan Zeng', 'Meng Zhang', 'Zhou Zhao', 'Tao Jin'] | null | null | null | null | findings-acl-2022-5 | ['sign-language-translation'] | ['computer-vision'] | [ 4.25092131e-01 -4.51116979e-01 -3.49974990e-01 -3.60634506e-01
-9.12780404e-01 -5.53798914e-01 5.49204409e-01 -6.73126996e-01
-3.75619024e-01 4.97239143e-01 9.45315719e-01 -2.54163414e-01
4.00864244e-01 -4.02113199e-01 -7.39038348e-01 -6.22883558e-01
2.97513574e-01 -2.21400559e-02 -2.09637910e-01 -2.78548837e-01
1.40115097e-01 -1.26318827e-01 -1.22064805e+00 7.04826891e-01
1.01358104e+00 7.60455191e-01 5.22290170e-01 6.50359690e-01
-1.93841681e-01 8.24680805e-01 -5.43610215e-01 -6.16663694e-01
1.06182076e-01 -8.25968087e-01 -4.55393463e-01 2.00586155e-01
4.01016533e-01 -6.53889418e-01 -4.57098395e-01 8.11893284e-01
7.91729867e-01 -2.91105378e-02 5.20002842e-01 -1.17842364e+00
-9.71021354e-01 5.39385617e-01 -5.37714601e-01 -3.88555378e-02
3.54760230e-01 8.35958540e-01 1.20700479e+00 -9.90134001e-01
8.42297614e-01 1.23462474e+00 1.56705126e-01 6.62210882e-01
-5.95350325e-01 -7.78354108e-01 4.67559129e-01 5.68708539e-01
-1.13139975e+00 -2.98638672e-01 8.36736917e-01 -1.89603344e-01
6.21442020e-01 1.25643328e-01 9.62645888e-01 1.57320046e+00
-7.56214708e-02 1.30281270e+00 1.17851388e+00 -3.30839723e-01
-2.11752191e-01 -2.81142211e-03 -2.75439352e-01 8.51606727e-01
-4.44418006e-02 2.03477487e-01 -1.06570983e+00 3.43228847e-01
6.63810372e-01 -2.78533340e-01 -4.91785169e-01 4.09425870e-02
-1.51300967e+00 5.91062963e-01 1.32674202e-01 -6.90331832e-02
-3.25073510e-01 1.76969185e-01 4.22435194e-01 2.16029599e-01
2.73281173e-03 -5.55107519e-02 -2.85984099e-01 -3.92841697e-01
-7.81471252e-01 -1.35998493e-02 5.16691983e-01 1.16893184e+00
5.64937353e-01 1.94329903e-01 -6.77142143e-01 5.05484998e-01
6.79302216e-01 1.01836312e+00 6.05396152e-01 -5.38853347e-01
1.03924572e+00 5.77249885e-01 -2.35127732e-01 -7.09253728e-01
7.14581832e-02 -4.68929380e-01 -7.16965914e-01 -3.10234036e-02
1.50749072e-01 -3.48160475e-01 -1.18221402e+00 1.91324866e+00
1.23803243e-01 3.34312558e-01 8.55344683e-02 1.25130153e+00
1.00918472e+00 8.13990712e-01 2.20377952e-01 -1.49769500e-01
1.50768280e+00 -1.01489866e+00 -7.49213696e-01 -3.95888895e-01
5.08376956e-01 -8.67910087e-01 1.45659411e+00 1.16185367e-01
-7.60579407e-01 -4.55214173e-01 -9.84148324e-01 -2.86689728e-01
9.59258601e-02 5.08474529e-01 5.48840821e-01 3.78220052e-01
-6.30220056e-01 4.00781743e-02 -6.17549479e-01 -2.46535718e-01
3.50188017e-01 -1.76421553e-02 -1.70566708e-01 -3.16386491e-01
-1.28979909e+00 7.12362468e-01 4.47429828e-02 4.20905441e-01
-8.65321517e-01 -3.82867336e-01 -9.27399576e-01 -3.18526626e-01
2.74261832e-01 -9.31759357e-01 1.14309657e+00 -1.15820408e+00
-1.76610208e+00 5.54004252e-01 -4.72245067e-01 -6.03055060e-02
6.52052701e-01 -2.63979614e-01 -2.33428389e-01 1.52790397e-01
-1.12020299e-01 7.38887131e-01 8.59575033e-01 -1.18213761e+00
-5.99205077e-01 -1.62247673e-01 4.28725891e-02 7.10800469e-01
-3.26902121e-01 2.48222947e-01 -7.82965481e-01 -9.61676121e-01
-1.86717615e-01 -7.03027725e-01 1.68250110e-02 -1.16925202e-02
-4.69544202e-01 6.80334195e-02 5.08459866e-01 -1.26940203e+00
1.30841362e+00 -2.12325859e+00 5.45826733e-01 1.58872068e-01
-6.43524453e-02 1.61337212e-01 -5.46175480e-01 3.14631820e-01
2.95136839e-01 -7.11015314e-02 -2.72541702e-01 -3.60624403e-01
1.29073665e-01 4.04062480e-01 -5.02242506e-01 3.40566516e-01
1.44101948e-01 1.25100625e+00 -8.57016206e-01 -7.28236854e-01
1.32113164e-02 6.33192360e-01 -5.83922625e-01 2.85029024e-01
-3.49535078e-01 7.69899428e-01 -5.97744226e-01 8.15371394e-01
4.00281608e-01 -1.57379553e-01 1.34455815e-01 -4.04495865e-01
-2.81859078e-02 2.39835069e-01 -9.36475337e-01 1.94226253e+00
-5.26112139e-01 7.53991961e-01 -1.49885654e-01 -6.07194901e-01
7.45612681e-01 2.83578902e-01 8.92174244e-02 -8.75550330e-01
2.09206343e-01 1.63013414e-01 -1.41157180e-01 -9.15977359e-01
3.16342026e-01 -1.50755376e-01 -1.33243680e-01 4.31651175e-01
6.23249114e-02 3.50511014e-01 1.52014881e-01 2.59864777e-01
8.34549069e-01 5.75752676e-01 -1.62779912e-02 4.15315449e-01
3.74752581e-01 -6.33639246e-02 7.23388016e-01 3.50477576e-01
-1.40274808e-01 6.26698017e-01 3.43235254e-01 1.55985922e-01
-8.60847473e-01 -9.60706055e-01 4.54921007e-01 1.21315300e+00
4.29525912e-01 -4.77681518e-01 -6.33415997e-01 -6.38181090e-01
-4.78139818e-01 6.81073129e-01 -4.73441899e-01 -1.50602013e-01
-8.23125124e-01 -5.27888715e-01 6.87778533e-01 7.84366548e-01
7.41093218e-01 -1.22087514e+00 -4.74905759e-01 -8.78277719e-02
-7.27819443e-01 -1.15420842e+00 -1.03912318e+00 -4.04498875e-01
-5.72473228e-01 -8.41643512e-01 -9.78306830e-01 -9.17140543e-01
6.90843284e-01 1.42825261e-01 7.66961336e-01 8.51676390e-02
-1.09655932e-01 3.99091393e-01 -6.24882162e-01 -4.56550978e-02
-2.56529778e-01 -1.34662554e-01 -3.11688095e-01 4.46395487e-01
1.97977468e-01 -3.75875533e-01 -6.66144669e-01 2.04734638e-01
-7.27045298e-01 7.24495471e-01 1.21789098e+00 1.03239667e+00
4.64695156e-01 -6.40737712e-01 2.46765509e-01 -3.83701801e-01
5.03267884e-01 -3.06365520e-01 -3.03091973e-01 5.34634233e-01
-1.56907663e-01 2.85926044e-01 4.28837776e-01 -6.35796607e-01
-1.35108674e+00 1.24367490e-01 -1.24273181e-01 -3.61204445e-01
1.07832318e-02 7.16262817e-01 -7.23326564e-01 1.66620970e-01
1.52415723e-01 9.08273876e-01 -7.42615387e-02 -4.19639856e-01
6.92889035e-01 7.56385267e-01 6.96577787e-01 -5.34278035e-01
9.48939562e-01 3.78403395e-01 -2.60228068e-01 -5.93683004e-01
-6.43920898e-01 -1.47623077e-01 -5.60076773e-01 -4.10745054e-01
9.43123937e-01 -9.92618978e-01 -5.89776218e-01 6.29407287e-01
-1.30190897e+00 -4.41112459e-01 -3.36080119e-02 6.27414525e-01
-5.51615417e-01 6.83743954e-01 -5.30689597e-01 -6.41261876e-01
-2.13337526e-01 -1.31803906e+00 1.15662754e+00 2.05677390e-01
-8.58227164e-02 -6.57093883e-01 -2.88614817e-02 7.04193175e-01
1.70211241e-01 -5.71141094e-02 9.29770648e-01 -2.25459740e-01
-1.02917743e+00 -7.95451179e-03 -4.22566056e-01 3.83977741e-01
-2.00882137e-01 -1.02042586e-01 -6.90468013e-01 -7.36131594e-02
-3.80707979e-01 -2.16677174e-01 9.93288398e-01 1.13815434e-01
7.16636419e-01 -5.66861153e-01 1.41056720e-02 8.10776949e-01
1.12351143e+00 6.74994336e-03 7.16912031e-01 1.37069896e-01
9.54589963e-01 4.83145893e-01 7.58098066e-01 4.48334068e-01
9.30368245e-01 8.03733587e-01 1.95183709e-01 -9.01932344e-02
-6.37749732e-01 -7.47800589e-01 1.00171220e+00 1.06029570e+00
-4.27894324e-01 -4.21189815e-01 -5.88794172e-01 5.96334994e-01
-1.76390553e+00 -1.00739145e+00 1.06395610e-01 1.93476391e+00
1.04166853e+00 -3.75252545e-01 -4.22781073e-02 -1.94178998e-01
7.51960337e-01 2.85779446e-01 -4.61360276e-01 4.25795130e-02
-4.37872708e-01 -4.64371331e-02 4.13662463e-01 3.55607688e-01
-6.91202939e-01 1.14878893e+00 5.24409580e+00 9.06941712e-01
-1.37372100e+00 1.17353469e-01 2.05560774e-01 -2.68264771e-01
-7.41025388e-01 5.48845492e-02 -7.50155330e-01 5.80084801e-01
4.97201651e-01 -9.96716227e-03 6.14993334e-01 4.70815152e-01
4.84410316e-01 1.04946978e-01 -7.22864866e-01 1.00412273e+00
4.49936450e-01 -1.16880178e+00 4.89992976e-01 1.26349807e-01
6.02091253e-01 2.79784948e-02 1.66905895e-01 4.01082516e-01
7.93840513e-02 -8.48618507e-01 9.98598695e-01 6.65749550e-01
1.14162302e+00 -5.43103993e-01 5.88728309e-01 2.47841939e-01
-1.43347645e+00 -3.48483287e-02 5.55343628e-02 1.90012231e-01
6.41012073e-01 4.90878597e-02 -4.98603970e-01 5.64173758e-01
2.00105727e-01 7.91999340e-01 -4.38144654e-01 9.81162548e-01
-9.24653232e-01 9.06882882e-01 -1.85675636e-01 -1.98219270e-01
1.99872673e-01 4.12441380e-02 6.72457755e-01 1.33164847e+00
4.32086915e-01 1.53053463e-01 1.87405854e-01 6.85661793e-01
-9.10020843e-02 4.30700064e-01 -3.16328973e-01 -1.19553737e-01
3.39132607e-01 9.26465869e-01 -2.44968489e-01 -3.66486162e-01
-6.71809256e-01 1.35810053e+00 6.54265136e-02 7.38050282e-01
-1.03746891e+00 -1.85571194e-01 6.72582269e-01 -4.80938703e-02
4.57627773e-01 -4.00415510e-01 -3.65584463e-01 -1.54857492e+00
3.43901187e-01 -9.98236239e-01 3.21075737e-01 -1.01341343e+00
-1.08283579e+00 4.12585646e-01 -1.48317397e-01 -1.46425080e+00
-1.73787355e-01 -6.05722666e-01 -7.86371589e-01 7.80096412e-01
-1.81839883e+00 -1.87092030e+00 -4.01376367e-01 7.79314697e-01
7.08771527e-01 -2.76399195e-01 5.18123925e-01 3.22728962e-01
-6.09479606e-01 8.97205710e-01 -2.94714421e-01 5.07417142e-01
8.85755002e-01 -8.37550342e-01 3.51419359e-01 1.17623055e+00
3.74107838e-01 3.95301282e-01 3.84764761e-01 -9.59913373e-01
-1.73377109e+00 -1.15737617e+00 1.04427385e+00 -2.71762133e-01
5.79669595e-01 -2.46588439e-01 -4.88016635e-01 6.28048182e-01
3.48145843e-01 -3.29422683e-01 6.59432411e-01 -2.90246755e-01
-5.38716376e-01 9.79274064e-02 -4.88027006e-01 9.15667415e-01
1.39900100e+00 -7.29406536e-01 -6.22808099e-01 6.85821027e-02
7.14186311e-01 -5.49111128e-01 -3.48647326e-01 3.74977469e-01
7.67268777e-01 -6.58132553e-01 6.92607403e-01 -6.70795858e-01
7.63318717e-01 -4.80361313e-01 -3.30651373e-01 -1.20054626e+00
2.00756956e-02 -6.98676884e-01 -2.02961624e-01 1.33010876e+00
4.53079790e-01 -1.86129704e-01 5.32017291e-01 3.58895600e-01
-2.78517127e-01 -7.76278138e-01 -7.86788642e-01 -6.19859219e-01
-2.68747568e-01 -7.40522087e-01 4.78838980e-01 6.34313703e-01
2.69745588e-02 4.78683829e-01 -9.95270669e-01 2.00632930e-01
5.56763172e-01 3.81129473e-01 9.41108942e-01 -4.63421762e-01
-5.32948434e-01 -3.54776204e-01 -4.02500063e-01 -1.63193214e+00
2.25758806e-01 -9.68200088e-01 2.20198333e-01 -1.68793809e+00
4.83966827e-01 1.16955459e-01 -1.66158646e-01 5.75952888e-01
-3.83510232e-01 2.43300989e-01 3.96947056e-01 1.89544648e-01
-6.77654862e-01 8.99263322e-01 1.68403912e+00 -1.64927393e-01
2.89562661e-02 -1.61437184e-01 -6.78784311e-01 5.39776564e-01
4.64446515e-01 -3.56870890e-02 -3.72203350e-01 -8.34687710e-01
8.59731436e-02 1.50234148e-01 6.78720415e-01 -6.38769865e-01
2.94804841e-01 -3.81212264e-01 2.42795736e-01 -6.56841934e-01
2.96694249e-01 -4.98372406e-01 -3.99388760e-01 2.54960537e-01
-4.10607606e-01 -7.04180226e-02 -5.45668304e-02 7.26901710e-01
-3.03122312e-01 1.37705371e-01 4.08229589e-01 -4.11597602e-02
-1.13756955e+00 5.18372357e-01 -2.60890841e-01 2.12359846e-01
6.67699337e-01 -2.29105636e-01 -2.98583150e-01 -6.66395068e-01
-4.45766181e-01 3.76451075e-01 2.49470770e-01 6.40209258e-01
8.00273240e-01 -1.54125500e+00 -1.06429899e+00 3.47928584e-01
3.44294846e-01 -3.06710631e-01 5.50854623e-01 8.88106585e-01
-2.48133108e-01 1.12380341e-01 -2.63137311e-01 -4.79417652e-01
-1.41626155e+00 1.43122062e-01 1.11426823e-02 -1.48360645e-02
-6.76526129e-01 9.61155713e-01 3.55332792e-01 -9.19917226e-02
2.67992586e-01 -2.95551896e-01 -8.77769291e-02 5.05520254e-02
6.05016887e-01 -1.04321605e-02 -5.37376225e-01 -8.68583024e-01
-1.66932955e-01 7.11706638e-01 1.18148744e-01 -5.83225787e-01
1.06941462e+00 -3.43281567e-01 1.60153896e-01 1.70850351e-01
9.58081603e-01 1.28218561e-01 -1.41950250e+00 -4.16105181e-01
-3.87585342e-01 -5.91361403e-01 -3.86345312e-02 -1.09042728e+00
-1.11911917e+00 1.18242323e+00 2.81365126e-01 -9.21939611e-01
1.34804165e+00 -4.60180230e-02 1.23298144e+00 2.03779697e-01
2.86214322e-01 -1.05639899e+00 1.69440791e-01 5.47516644e-01
9.90986109e-01 -1.14099717e+00 -3.27672362e-01 -2.51606643e-01
-1.18703759e+00 9.78283584e-01 6.29932225e-01 3.29364687e-01
5.60203716e-02 1.85989633e-01 4.11498070e-01 1.48922101e-01
-7.49509454e-01 -4.46632355e-01 6.05097115e-01 6.56234026e-01
2.31552124e-01 1.27241798e-02 -1.90813646e-01 9.52777386e-01
-2.01818123e-01 1.18016273e-01 1.79651752e-01 6.61838055e-01
-2.52711564e-01 -1.16334379e+00 -1.70088604e-01 2.37333719e-02
5.93827702e-02 -4.82107669e-01 -4.83134538e-01 5.22466183e-01
2.05351219e-01 6.51694953e-01 -3.14322382e-01 -6.98342681e-01
2.68964708e-01 5.86650930e-02 5.31427801e-01 -3.93067092e-01
-3.50594163e-01 2.78221548e-01 2.27063239e-01 -5.13259172e-01
-3.21003377e-01 -7.36090243e-01 -1.30472541e+00 -2.13498294e-01
-9.42197815e-02 -3.25361907e-01 5.67281008e-01 1.31625700e+00
3.28235179e-01 4.68886256e-01 4.16117013e-01 -7.11317897e-01
-4.69197482e-01 -9.41980064e-01 -2.55299479e-01 3.79675537e-01
2.30988011e-01 -5.49641669e-01 -8.68368149e-02 3.04968089e-01] | [9.210939407348633, -6.512601375579834] |
03712ae6-903c-4f9c-9ae4-d49e082e8dce | audio-visual-speech-recognition-using-deep | 1611.02879 | null | http://arxiv.org/abs/1611.02879v1 | http://arxiv.org/pdf/1611.02879v1.pdf | Audio Visual Speech Recognition using Deep Recurrent Neural Networks | In this work, we propose a training algorithm for an audio-visual automatic
speech recognition (AV-ASR) system using deep recurrent neural network
(RNN).First, we train a deep RNN acoustic model with a Connectionist Temporal
Classification (CTC) objective function. The frame labels obtained from the
acoustic model are then used to perform a non-linear dimensionality reduction
of the visual features using a deep bottleneck network. Audio and visual
features are fused and used to train a fusion RNN. The use of bottleneck
features for visual modality helps the model to converge properly during
training. Our system is evaluated on GRID corpus. Our results show that
presence of visual modality gives significant improvement in character error
rate (CER) at various levels of noise even when the model is trained without
noisy data. We also provide a comparison of two fusion methods: feature fusion
and decision fusion. | ['Abhinav Thanda', 'Shankar M Venkatesan'] | 2016-11-09 | null | null | null | null | ['audio-visual-speech-recognition'] | ['speech'] | [ 3.65423471e-01 -2.55188107e-01 3.58234048e-01 -3.72442335e-01
-1.10840499e+00 -3.23293388e-01 7.52309799e-01 -1.49763748e-01
-6.60858393e-01 4.09391224e-01 4.61484224e-01 -3.89347434e-01
3.66831034e-01 -1.28186718e-01 -4.91805464e-01 -8.53387654e-01
3.66173297e-01 9.00856629e-02 -1.12927154e-01 -1.45100296e-01
1.13637254e-01 5.82511663e-01 -1.73866665e+00 6.74320817e-01
1.92196369e-01 1.05350995e+00 4.75400925e-01 1.53217936e+00
1.56628201e-03 1.05646694e+00 -8.44351172e-01 1.84929505e-01
4.98620756e-02 -3.76387805e-01 -8.58322859e-01 1.09955877e-01
5.64002283e-02 -2.32726231e-01 -5.88235617e-01 7.50999808e-01
8.17879617e-01 7.73470700e-01 6.80048048e-01 -9.19193864e-01
-5.79951525e-01 4.77575660e-01 -2.38844305e-01 3.45902234e-01
2.84551740e-01 1.35692298e-01 1.00932872e+00 -1.27602339e+00
3.63431841e-01 1.43412161e+00 4.52289402e-01 8.44696939e-01
-1.32513583e+00 -3.36905003e-01 -1.52494842e-02 5.43440223e-01
-1.41330612e+00 -9.64959085e-01 6.85474038e-01 -2.80439496e-01
1.78582561e+00 3.29425782e-01 5.22522509e-01 1.23396456e+00
-2.05068123e-02 8.66187453e-01 7.03128099e-01 -7.63691306e-01
2.31354579e-01 -2.36844402e-02 1.33597583e-01 4.02155191e-01
-7.20529795e-01 2.83713967e-01 -7.42724180e-01 9.44780409e-02
4.66794878e-01 -1.01508565e-01 -7.49593824e-02 3.13553333e-01
-1.02069283e+00 9.02458012e-01 1.80184782e-01 5.31582415e-01
-6.32308781e-01 3.62251937e-01 6.98330879e-01 5.48201501e-01
3.71456146e-01 -1.82776943e-01 -3.00613761e-01 -2.80946374e-01
-1.10595274e+00 -3.94601732e-01 3.15407068e-01 2.76356578e-01
4.88323897e-01 7.50361204e-01 -3.27266455e-01 1.44619274e+00
8.17886174e-01 5.47462285e-01 9.39241827e-01 -1.03897142e+00
3.39203805e-01 1.27154604e-01 -2.99037904e-01 -6.64837718e-01
-2.47025311e-01 6.67973282e-03 -8.21688235e-01 4.80129600e-01
7.95694068e-02 -7.13624060e-02 -1.48801816e+00 1.53313375e+00
-6.15770072e-02 4.07937735e-01 7.30496764e-01 9.30449545e-01
1.03318226e+00 1.18148184e+00 5.06020412e-02 -4.22540158e-01
1.20978177e+00 -9.54040349e-01 -1.06171048e+00 1.49657264e-01
6.90984368e-01 -9.15767848e-01 7.12493420e-01 4.56345052e-01
-1.07550001e+00 -8.71329546e-01 -1.12926447e+00 -1.81942642e-01
-4.22044575e-01 4.81332272e-01 -8.38142037e-02 5.89769006e-01
-1.56501126e+00 3.51423621e-01 -9.83117044e-01 -3.40309292e-01
2.31349841e-02 6.19433105e-01 -3.61472517e-01 2.60802776e-01
-1.05627489e+00 9.35018539e-01 4.57912445e-01 4.19227958e-01
-1.39673591e+00 -1.67934239e-01 -9.10536706e-01 -3.89649309e-02
8.46427679e-03 -3.39501143e-01 1.47211111e+00 -1.21875751e+00
-2.07779503e+00 5.14180720e-01 -4.37370390e-01 -8.22339296e-01
1.01872738e-02 -1.03159577e-01 -5.18468857e-01 1.86119333e-01
-5.25929272e-01 9.89667177e-01 1.07817459e+00 -1.14360797e+00
-7.52925336e-01 -3.18327956e-02 -4.50533867e-01 4.17250156e-01
-9.31036025e-02 2.15041488e-01 -2.17448935e-01 -7.25103617e-01
-1.40234670e-02 -9.10955966e-01 2.38164775e-02 -6.53266013e-01
-2.57666707e-01 -5.55965900e-01 1.22467625e+00 -1.04673278e+00
1.21041763e+00 -2.38324142e+00 3.00633341e-01 2.54368037e-01
2.11966336e-02 5.98428190e-01 -3.24073285e-01 3.53477865e-01
-3.79554182e-01 1.59087945e-02 7.17251599e-02 -8.76226783e-01
-2.92037427e-01 4.51822639e-01 -4.25305277e-01 3.35457206e-01
1.06764458e-01 7.23495960e-01 -3.40663224e-01 -3.70947480e-01
7.58834481e-01 1.01579511e+00 -2.28910014e-01 3.92335862e-01
-2.41928827e-02 2.54326999e-01 1.24621868e-01 3.29235286e-01
1.25512332e-01 6.70096651e-02 -1.75185472e-01 -1.73590437e-01
-5.53550161e-02 5.17929435e-01 -1.05968690e+00 1.63030851e+00
-5.54834008e-01 1.14133835e+00 -3.08219474e-02 -9.82883632e-01
1.01771390e+00 9.15003717e-01 2.24446401e-01 -7.18985021e-01
3.13390762e-01 2.77790427e-02 6.68955094e-04 -1.89390615e-01
5.96221805e-01 7.75060058e-02 3.58742028e-01 4.27980930e-01
4.54006881e-01 1.89126715e-01 -2.25389481e-01 1.83479562e-01
1.03573120e+00 -1.58632889e-01 -7.08884448e-02 3.14164370e-01
8.63816440e-01 -3.25636178e-01 3.08317721e-01 6.29164159e-01
-1.98949948e-01 7.35272586e-01 1.76757425e-02 -3.94131571e-01
-1.33656776e+00 -9.88808990e-01 2.60712475e-01 1.28921127e+00
-5.47956228e-01 -5.02914846e-01 -6.04778469e-01 -4.36177343e-01
-4.11253631e-01 8.45876694e-01 -5.01162767e-01 -2.06310570e-01
-5.29972255e-01 -4.26754862e-01 7.55207241e-01 5.61535776e-01
1.69268847e-01 -1.41555250e+00 -3.37074429e-01 2.91758537e-01
-2.69682854e-01 -1.08887982e+00 -2.65937716e-01 4.33730721e-01
-5.81302285e-01 -3.48071843e-01 -7.28265584e-01 -8.57916892e-01
3.06848705e-01 1.53649986e-01 5.23819685e-01 4.49487381e-02
-1.27119794e-01 6.30932033e-01 -4.75272179e-01 -2.12294921e-01
-9.76616800e-01 -2.65512198e-01 2.21259400e-01 2.32435837e-01
3.71260911e-01 -2.69273847e-01 -4.00982112e-01 9.95818153e-03
-6.85294449e-01 1.20822407e-01 2.36597553e-01 1.00954521e+00
5.60709238e-01 -2.71864116e-01 5.40887058e-01 5.90586103e-02
7.73417652e-01 2.13331413e-02 -5.70447981e-01 3.54850471e-01
-2.00036526e-01 3.66608232e-01 3.32067996e-01 -6.66594625e-01
-1.14659047e+00 4.65403646e-01 -2.93924063e-01 -9.52314377e-01
-2.53987819e-01 4.02665406e-01 2.83558816e-01 2.37057641e-01
5.12988269e-01 4.01019692e-01 5.69248060e-03 -5.39247870e-01
4.86552715e-01 1.20891845e+00 6.33372068e-01 -3.18086557e-02
1.33529589e-01 1.41832307e-01 -3.89464676e-01 -1.18549550e+00
-3.15052569e-01 -5.12749970e-01 -5.34603059e-01 -4.29615319e-01
1.21609890e+00 -9.31212008e-01 -8.80903959e-01 4.83756661e-01
-1.32074344e+00 -2.80035049e-01 -2.51851439e-01 8.51720095e-01
-5.16056716e-01 3.18895817e-01 -7.28391111e-01 -1.44909596e+00
-5.78134358e-01 -1.28867102e+00 9.89286602e-01 4.14756946e-02
-1.80328302e-02 -9.26852942e-01 2.86257058e-01 4.32857335e-01
2.70198911e-01 -2.61803508e-01 4.62114602e-01 -8.65687907e-01
-1.16835982e-01 5.27330041e-02 -8.32022056e-02 8.87689829e-01
3.60759050e-02 3.76179874e-01 -1.56171036e+00 -2.86808312e-01
3.84197496e-02 -4.34513777e-01 1.19051933e+00 6.44988000e-01
8.90083909e-01 -1.49193808e-01 7.54450262e-03 2.87939787e-01
1.10047913e+00 5.82739890e-01 6.17759943e-01 -8.48867968e-02
7.97238290e-01 2.96184897e-01 1.79725513e-01 3.83344471e-01
1.35701314e-01 6.94425523e-01 1.55054808e-01 7.42251240e-03
-4.68029380e-01 1.16317913e-01 8.13706160e-01 1.46853435e+00
1.15055302e-02 -4.41704929e-01 -9.95890617e-01 5.61757445e-01
-1.94653201e+00 -9.93897021e-01 2.73101062e-01 2.11629558e+00
6.51610017e-01 -9.06736329e-02 1.38205692e-01 4.16518807e-01
9.01742399e-01 -2.54840739e-02 -1.94844022e-01 -9.18028772e-01
-2.50359088e-01 1.14103384e-01 2.25061670e-01 8.17230761e-01
-1.03973532e+00 1.12087142e+00 6.65935040e+00 9.95130479e-01
-1.36483300e+00 2.04857349e-01 6.47547662e-01 -1.71242520e-01
6.59020841e-02 -2.39387602e-01 -6.85933828e-01 1.56189620e-01
1.73785329e+00 4.06072527e-01 7.42256343e-01 3.18897128e-01
4.65394378e-01 -5.19772731e-02 -9.99698818e-01 1.55075645e+00
1.58308595e-01 -1.28549266e+00 5.55023402e-02 -4.17878404e-02
3.27042192e-01 4.46170330e-01 2.00299516e-01 2.94789910e-01
6.02745771e-01 -1.35212994e+00 5.77206790e-01 6.39157355e-01
6.27655208e-01 -9.73754704e-01 6.41513765e-01 2.48195529e-02
-1.30800772e+00 -1.28752142e-01 -2.68828690e-01 2.19771758e-01
2.44418755e-01 1.75162069e-02 -1.14377546e+00 2.96682626e-01
4.98758316e-01 5.30218840e-01 -3.50102782e-01 6.07283235e-01
1.44328535e-01 8.50273669e-01 -4.70146149e-01 8.15423205e-02
3.67027730e-01 2.06421122e-01 5.31440675e-01 1.43931675e+00
2.60115921e-01 2.21575890e-02 -1.68848075e-02 3.30600560e-01
-6.40180856e-02 8.89272988e-02 -6.17746949e-01 -3.13402355e-01
4.11424994e-01 1.16990948e+00 -6.11427546e-01 -5.70043564e-01
-3.47368509e-01 1.00095129e+00 3.14324826e-01 4.73169833e-01
-6.57946706e-01 -1.55009970e-01 4.73135859e-01 -6.97008669e-01
4.91642594e-01 -3.09125245e-01 -3.65876779e-02 -9.75981295e-01
-3.19109708e-01 -8.54997277e-01 3.05375338e-01 -1.13977528e+00
-8.54325533e-01 8.96487653e-01 -2.88628638e-01 -9.40481365e-01
-7.43020654e-01 -5.02152026e-01 -3.76222968e-01 9.03039753e-01
-1.04735374e+00 -1.03566170e+00 2.82755643e-01 7.80851185e-01
9.46164131e-01 -7.62082338e-01 9.71244276e-01 2.62586623e-01
-5.62512755e-01 5.74177742e-01 1.40462622e-01 3.03896159e-01
4.16423023e-01 -1.13681757e+00 3.00556213e-01 9.33869362e-01
6.10447884e-01 1.51925325e-01 5.98178267e-01 -5.62560499e-01
-1.19288874e+00 -9.27440107e-01 8.22957993e-01 -2.84197032e-01
3.91808629e-01 -3.51614386e-01 -9.35288429e-01 5.34175694e-01
6.48832917e-01 1.05652191e-01 6.40147388e-01 -2.89587110e-01
-4.35015231e-01 -1.28964325e-02 -8.13402593e-01 4.52351213e-01
3.45176339e-01 -1.16088367e+00 -7.52151668e-01 1.29181042e-01
1.04799378e+00 -8.40926468e-02 -6.34618759e-01 1.60469159e-01
4.59772468e-01 -4.85416859e-01 8.83076668e-01 -5.33211946e-01
-2.66629189e-01 -3.57170939e-01 -7.00368941e-01 -1.34390509e+00
-2.03389470e-02 -8.15957308e-01 -1.09933995e-01 1.20281065e+00
6.04128242e-01 -2.61905879e-01 3.12282056e-01 2.48269811e-01
-1.93122640e-01 -1.82655528e-01 -1.31700563e+00 -4.06492352e-01
-4.33563352e-01 -9.28209484e-01 -5.99617101e-02 6.59570694e-01
-1.63498268e-01 7.41730094e-01 -7.44939268e-01 1.16855972e-01
1.78232178e-01 -6.56476617e-01 2.73769408e-01 -9.62767661e-01
-1.59863845e-01 -2.97024041e-01 -6.06415570e-01 -7.22104490e-01
2.73514718e-01 -7.30534315e-01 1.80497885e-01 -1.48125732e+00
-1.07800849e-01 1.89188257e-01 -6.01478040e-01 7.76915610e-01
3.99991423e-01 2.08289802e-01 3.90104383e-01 1.41367242e-01
-6.76365852e-01 6.44761026e-01 6.09752297e-01 -1.08177617e-01
-5.42214930e-01 -1.43078506e-01 9.35823992e-02 4.47493106e-01
7.87835896e-01 -5.15090585e-01 -2.69049078e-01 -3.46432000e-01
-1.08565599e-01 3.22829455e-01 2.93154985e-01 -1.09733593e+00
4.71774876e-01 3.57081383e-01 5.61575294e-01 -1.03308058e+00
9.37484980e-01 -7.64295340e-01 9.04986337e-02 3.94229531e-01
-6.87029243e-01 3.05255234e-01 2.53268123e-01 4.56471443e-01
-3.54419261e-01 -7.40562826e-02 8.79534125e-01 2.90046006e-01
-6.74627244e-01 -3.78533751e-01 -1.00819027e+00 -5.46432674e-01
6.91626847e-01 -2.63338178e-01 -1.83183774e-02 -6.34990990e-01
-1.32972133e+00 -5.42062744e-02 -1.84930190e-01 7.10306644e-01
1.04147828e+00 -1.49620473e+00 -8.53341520e-01 3.78848255e-01
-2.03262225e-01 -5.32346129e-01 3.34974080e-01 7.20717669e-01
-2.59650111e-01 5.60136378e-01 1.96608547e-02 -9.15701926e-01
-1.89407849e+00 5.19412458e-01 5.53539217e-01 1.95831861e-02
-5.46142757e-01 7.38602698e-01 7.94333871e-03 -2.28973567e-01
7.86670506e-01 -4.35213178e-01 -6.16174757e-01 1.51091680e-01
7.20261276e-01 3.17060560e-01 2.28795543e-01 -1.22607028e+00
-4.39562291e-01 3.47364664e-01 -1.78989902e-01 -9.48634207e-01
1.31316936e+00 -4.13892388e-01 3.13062072e-01 8.49613369e-01
1.50382507e+00 -4.93438065e-01 -1.11769176e+00 -3.64106923e-01
-2.18929857e-01 1.20162629e-01 6.32495642e-01 -7.71546602e-01
-1.03621519e+00 1.20285916e+00 1.27091599e+00 2.49187797e-01
1.11501086e+00 -1.77050918e-01 5.33793271e-01 6.28709555e-01
-4.77927804e-01 -1.37743819e+00 1.92023665e-01 8.80054474e-01
9.10668373e-01 -1.08446193e+00 -5.20712078e-01 5.10531068e-01
-9.63955104e-01 1.30079067e+00 1.60047591e-01 -1.26468182e-01
7.34974921e-01 3.65747809e-01 4.14789170e-01 5.45849912e-02
-1.26839936e+00 -4.03543383e-01 5.00312090e-01 5.43540180e-01
2.93851465e-01 -7.18300268e-02 2.63903320e-01 9.80539247e-02
-4.96052280e-02 -2.55544692e-01 3.48103315e-01 7.53938496e-01
-5.93789637e-01 -8.13335955e-01 -5.38115799e-01 9.69692543e-02
-6.14833891e-01 -2.48324826e-01 -3.17381501e-01 1.86610952e-01
-3.73344392e-01 1.37408364e+00 2.18377247e-01 -7.19891489e-01
-2.63807606e-02 4.51320648e-01 2.09377915e-01 -4.01937217e-01
-8.24361444e-01 7.68143415e-01 5.82106151e-02 -3.47074240e-01
-4.86580402e-01 -5.74727774e-01 -1.36860132e+00 1.40503431e-02
-5.27209640e-01 2.22824037e-01 1.05127716e+00 1.09290767e+00
3.48285019e-01 8.80388975e-01 6.83748662e-01 -8.79129410e-01
-3.55608076e-01 -1.18405104e+00 -2.16644958e-01 6.02769256e-02
8.26141179e-01 -3.91537547e-01 -4.43241239e-01 4.03947651e-01] | [14.358404159545898, 5.176866054534912] |
9c9a1ee2-1fc0-43c2-9b6f-1279ea39b529 | methods-for-the-frugal-labeler-multi-class | null | null | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0263656 | https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0263656&type=printable | Methods for the frugal labeler: Multi-class semantic segmentation on heterogeneous labels | Deep learning increasingly accelerates biomedical research, deploying neural networks for multiple tasks, such as image classification, object detection, and semantic segmentation. However, neural networks are commonly trained supervised on large-scale, labeled datasets. These prerequisites raise issues in biomedical image recognition, as datasets are generally small-scale, challenging to obtain, expensive to label, and frequently heterogeneously labeled. Furthermore, heterogeneous labels are a challenge for supervised methods. If not all classes are labeled for an individual sample, supervised deep learning approaches can only learn on a subset of the dataset with common labels for each individual sample; consequently, biomedical image recognition engineers need to be frugal concerning their label and ground truth requirements. This paper discusses the effects of frugal labeling and proposes to train neural networks for multi-class semantic segmentation on heterogeneously labeled data based on a novel objective function. The objective function combines a class asymmetric loss with the Dice loss. The approach is demonstrated for training on the sparse ground truth of a heterogeneous labeled dataset, training within a transfer learning setting, and the use-case of merging multiple heterogeneously labeled datasets. For this purpose, a biomedical small-scale, multi-class semantic segmentation dataset is utilized. The heartSeg dataset is based on the medaka fish’s position as a cardiac model system. Automating image recognition and semantic segmentation enables high-throughput experiments and is essential for biomedical research. Our approach and analysis show competitive results in supervised training regimes and encourage frugal labeling within biomedical image recognition. | ['Markus Reischl', 'Christian Pylatiuk', 'Luca Rettenberger', 'Mark Schutera'] | 2022-02-08 | null | null | null | plos-one-2022-2 | ['2d-semantic-segmentation'] | ['computer-vision'] | [ 7.28612661e-01 1.72828272e-01 -2.75024980e-01 -5.63124478e-01
-8.83221567e-01 -5.78882277e-01 -1.14626102e-02 3.81031781e-01
-7.82282591e-01 7.58119047e-01 -4.33543444e-01 -2.91512161e-02
5.77842109e-02 -6.84434772e-01 -6.70309424e-01 -9.52413023e-01
2.25376651e-01 9.92747188e-01 -5.67409918e-02 3.92425925e-01
-7.78823867e-02 6.91877007e-01 -1.40632737e+00 3.70502234e-01
7.23539352e-01 1.09569657e+00 2.98562855e-01 4.72818822e-01
-1.29040942e-01 4.30116057e-01 -5.82074225e-01 -2.16595426e-01
3.48737389e-01 -6.12722874e-01 -9.09612894e-01 2.45337918e-01
6.10608518e-01 4.67346422e-02 2.53030539e-01 1.10236907e+00
6.84763789e-01 -2.22552106e-01 9.04009342e-01 -1.17026722e+00
-1.61715612e-01 4.18410957e-01 -3.58854771e-01 -6.90810531e-02
-3.57108682e-01 -2.69316286e-02 8.39153826e-01 -6.37611151e-01
6.71514928e-01 7.44522572e-01 8.99179697e-01 7.43965089e-01
-1.26105297e+00 -6.66856050e-01 -2.18074471e-01 -1.50740162e-01
-1.42740703e+00 -2.59104043e-01 5.92541158e-01 -7.72876203e-01
2.84164339e-01 2.03746885e-01 6.75296664e-01 7.00336039e-01
-3.31967585e-02 7.90706277e-01 1.14717782e+00 -9.80435908e-02
4.37650740e-01 7.80625865e-02 1.03481039e-01 6.40935659e-01
3.05080026e-01 -1.77506447e-01 -2.85054535e-01 -8.26582685e-03
6.36622190e-01 1.80623204e-01 -6.50194436e-02 -4.95339781e-01
-1.33361578e+00 8.10116172e-01 3.52845281e-01 4.99902129e-01
-2.75875509e-01 -4.13369723e-02 5.12089133e-01 1.82932854e-01
5.51295757e-01 5.45139968e-01 -6.85347915e-01 1.56330913e-01
-1.42031467e+00 -9.33309793e-02 7.47973144e-01 7.83399224e-01
9.06973422e-01 -1.29129887e-01 -1.59067616e-01 8.86245310e-01
1.80175677e-01 4.88929003e-01 6.39311671e-01 -9.83950138e-01
4.95207161e-02 8.29361022e-01 -3.49641740e-01 -9.11630452e-01
-8.14171851e-01 -6.59057856e-01 -1.18727875e+00 1.57500252e-01
6.76387608e-01 -8.66163075e-02 -1.21065760e+00 1.69418156e+00
5.38402677e-01 1.25485048e-01 2.13212535e-01 8.93708467e-01
9.51018870e-01 3.52065593e-01 2.14291647e-01 -8.12870115e-02
1.34876108e+00 -7.07463861e-01 -4.20418382e-01 -9.53209475e-02
1.10487747e+00 -4.86006200e-01 8.30695629e-01 2.26904854e-01
-8.32012773e-01 -3.61488998e-01 -8.60593915e-01 1.72485015e-03
-5.20707667e-01 3.74832809e-01 3.66644830e-01 7.82375276e-01
-8.16221774e-01 5.84222853e-01 -7.71079242e-01 -3.04860145e-01
1.16918552e+00 7.16487586e-01 -4.47015315e-01 -2.65784919e-01
-7.62733102e-01 6.79091632e-01 5.68513989e-01 1.75057009e-01
-1.01044941e+00 -9.69690859e-01 -7.94667244e-01 3.83026525e-02
6.81817997e-03 -6.63160145e-01 7.70054460e-01 -1.17851412e+00
-1.21132505e+00 1.46257269e+00 3.55435550e-01 -3.06027472e-01
6.32726610e-01 4.17156488e-01 1.15400506e-02 3.77301067e-01
3.78900975e-01 1.04901659e+00 6.52008057e-01 -1.20427346e+00
-4.48005021e-01 -5.86095989e-01 -4.20538813e-01 -1.22725936e-02
-2.18122840e-01 -2.37053022e-01 9.36778553e-04 -5.78068972e-01
2.36425936e-01 -1.07841551e+00 -2.92484820e-01 1.21214762e-01
-3.94431382e-01 1.08548805e-01 9.54649389e-01 -5.58229327e-01
5.04899681e-01 -2.27973151e+00 6.71625361e-02 2.36182272e-01
2.79779971e-01 3.18260342e-01 -1.82345271e-01 -1.72119111e-01
1.87858799e-03 1.20217554e-01 -8.10303926e-01 -2.09544018e-01
-1.89093873e-01 1.73674345e-01 3.59992594e-01 7.73705125e-01
1.50120646e-01 9.24495518e-01 -8.21979821e-01 -7.15157628e-01
8.06099102e-02 3.73944700e-01 -5.47779083e-01 3.04469824e-01
-6.52842149e-02 8.59396219e-01 -2.69339263e-01 9.05386150e-01
5.15275657e-01 -5.44152141e-01 1.21087343e-01 -5.66870570e-01
2.33643278e-01 -5.42800725e-01 -8.99294674e-01 1.90593481e+00
-4.52397555e-01 3.61366719e-01 2.97598571e-01 -1.63164949e+00
8.24994385e-01 4.20171797e-01 9.52994525e-01 -4.85039055e-01
4.31671500e-01 6.11397862e-01 -2.76280846e-02 -4.32804734e-01
-3.35921384e-02 -6.41743720e-01 -1.31918743e-01 4.10146326e-01
3.45194787e-01 -4.07999784e-01 3.02111953e-01 -4.00861464e-02
8.98434281e-01 -1.57863587e-01 -1.87599026e-02 -4.35396194e-01
3.25891614e-01 2.53339827e-01 6.47215486e-01 4.40838248e-01
-3.40027720e-01 9.72688973e-01 4.59175557e-01 -5.87715387e-01
-9.79242086e-01 -6.92691207e-01 -4.60597813e-01 9.50323343e-01
1.25691697e-01 2.86618292e-01 -8.98836851e-01 -1.01979566e+00
-9.58408192e-02 1.44821540e-01 -6.35941088e-01 -1.32409990e-01
-3.33331883e-01 -1.01433563e+00 7.67007649e-01 2.42930487e-01
3.99861872e-01 -1.10312748e+00 -8.00452650e-01 1.32551700e-01
-1.86626043e-03 -1.18479228e+00 -3.18296850e-01 4.45268720e-01
-8.41657043e-01 -1.32153356e+00 -1.25644672e+00 -1.15101469e+00
1.04432273e+00 -1.58214599e-01 1.08426189e+00 1.59587473e-01
-7.09454954e-01 2.94191450e-01 -1.36460841e-01 -2.77530909e-01
-5.05214989e-01 2.93620646e-01 -1.92444876e-01 2.11949617e-01
9.36773494e-02 -3.13071758e-01 -7.38627493e-01 4.67996627e-01
-1.16528702e+00 6.13837540e-02 6.98768854e-01 1.28957212e+00
8.17121148e-01 -1.38338387e-01 9.68111217e-01 -1.29992831e+00
1.95609052e-02 -4.56079036e-01 -5.02841353e-01 3.44610453e-01
-3.53792518e-01 -2.51052171e-01 4.79614556e-01 -4.14198399e-01
-6.94890141e-01 5.12104928e-01 -2.00448781e-01 -3.39225769e-01
-3.81352246e-01 6.55131996e-01 -1.55157983e-01 -2.64426887e-01
6.66135430e-01 -1.09767936e-01 3.13070178e-01 -1.10672943e-01
1.67187393e-01 7.61351824e-01 3.61894310e-01 -5.48155546e-01
2.29205564e-01 6.38851881e-01 2.08184257e-01 -8.55941355e-01
-9.21627343e-01 -5.43356955e-01 -8.69987726e-01 -2.08238944e-01
1.12734616e+00 -8.49159598e-01 -4.59674209e-01 6.42871380e-01
-8.49323928e-01 -6.11310780e-01 -5.29558361e-01 5.45005798e-01
-5.11916637e-01 2.60662735e-01 -6.08999491e-01 -2.01596290e-01
-4.34841394e-01 -1.47194707e+00 1.34323716e+00 2.36321375e-01
-2.18511745e-01 -1.06212938e+00 -1.86096579e-02 7.09151208e-01
3.92125994e-01 3.60821038e-01 9.58139837e-01 -9.86747265e-01
-2.68650740e-01 -4.72792268e-01 -4.47824478e-01 4.89550322e-01
3.12961996e-01 -2.52570003e-01 -1.04535937e+00 -4.25321192e-01
-1.90749004e-01 -8.55326712e-01 7.95235038e-01 5.42317033e-01
1.29031098e+00 4.57636826e-02 -3.37379694e-01 6.65952921e-01
1.17020810e+00 1.69554323e-01 2.25813001e-01 -1.74130090e-02
9.68585074e-01 8.60639572e-01 3.84495020e-01 1.86128676e-01
1.27590835e-01 4.09240633e-01 3.30518991e-01 -6.16333187e-01
-1.55046821e-01 2.86665469e-01 -2.88267583e-01 7.70537078e-01
3.38084847e-01 -2.84013804e-02 -1.04048765e+00 5.54450691e-01
-1.51453257e+00 -5.31664729e-01 1.29155681e-01 2.06779218e+00
1.04591644e+00 -3.51940632e-01 -4.10494581e-02 8.73538777e-02
8.94879103e-01 -3.70082736e-01 -7.77868748e-01 1.09639339e-01
-2.13709906e-01 2.00410664e-01 5.91438353e-01 1.43198296e-01
-1.34058964e+00 7.94920385e-01 5.69825363e+00 8.96663904e-01
-1.42365551e+00 2.60631919e-01 1.23988116e+00 -2.54020025e-03
-3.77452821e-02 -4.40767318e-01 -6.91649258e-01 4.57324147e-01
7.92407155e-01 1.88332602e-01 -2.83788480e-02 6.32668793e-01
-7.08738640e-02 -1.64459050e-02 -1.12968898e+00 1.08149958e+00
8.44766796e-02 -1.34235013e+00 1.47092436e-02 -4.62681316e-02
9.69138920e-01 1.66097626e-01 1.32465139e-01 9.32417363e-02
7.61357993e-02 -1.20443428e+00 5.10659635e-01 9.89749357e-02
1.00743008e+00 -5.44865131e-01 1.00863802e+00 1.92144483e-01
-7.75063813e-01 1.85311437e-01 -2.73474038e-01 5.01505256e-01
1.38385683e-01 8.75006199e-01 -8.11030388e-01 3.49877596e-01
5.37153780e-01 8.09707940e-01 -4.48970705e-01 1.11527193e+00
3.24326664e-01 4.70962614e-01 -3.77395988e-01 1.77365512e-01
2.41042599e-01 -3.41658443e-01 1.75496433e-02 1.00723684e+00
5.00633180e-01 -1.37377024e-01 4.06573266e-01 7.15292037e-01
-3.56664628e-01 3.10575753e-01 -4.47985083e-01 -3.11590761e-01
8.62976760e-02 1.44154191e+00 -1.43736923e+00 -4.03541952e-01
-1.95150241e-01 7.93939233e-01 1.08113505e-01 2.08683208e-01
-5.97162426e-01 -1.60551026e-01 2.10071966e-01 -1.37778535e-01
2.84002963e-02 2.74410427e-01 -6.35123372e-01 -1.01570201e+00
-3.56818169e-01 -8.52663219e-01 5.00027776e-01 -3.07252675e-01
-1.36600101e+00 5.24758995e-01 -1.66988626e-01 -1.23802698e+00
1.66667879e-01 -6.27281547e-01 -9.36897248e-02 6.01587355e-01
-1.48553240e+00 -1.34863007e+00 -3.93829346e-01 4.18312818e-01
3.46839666e-01 -2.55038500e-01 8.16039026e-01 7.78273165e-01
-5.75163722e-01 3.77102435e-01 1.00666106e-01 2.79414207e-01
6.19809210e-01 -1.04654586e+00 -3.48164588e-01 3.46887469e-01
1.58488661e-01 3.01827461e-01 3.66442919e-01 -5.20640016e-01
-9.34905827e-01 -1.34506857e+00 4.97932583e-01 4.27023731e-02
3.66755128e-01 -1.72435135e-01 -8.97335351e-01 4.76691842e-01
-2.76366681e-01 5.52593470e-01 1.19410980e+00 -2.27849677e-01
-7.70005882e-02 -1.40967906e-01 -1.54764009e+00 2.22314820e-01
5.59110463e-01 -5.19661963e-01 -7.96620622e-02 7.72423565e-01
2.99717039e-01 -4.70905215e-01 -1.22192860e+00 6.66476607e-01
5.80955684e-01 -6.23800576e-01 8.58450711e-01 -4.60391015e-01
3.14092338e-01 -3.40788305e-01 -5.03322482e-02 -1.26714778e+00
1.99862063e-01 3.25204656e-02 5.66068649e-01 9.20855939e-01
5.70015490e-01 -4.83784586e-01 1.25281417e+00 6.04120910e-01
-3.57918531e-01 -6.50914192e-01 -1.13625002e+00 -6.40833020e-01
2.81132162e-01 4.05535288e-02 2.92556107e-01 1.22949278e+00
-3.05138767e-01 1.91302240e-01 -1.17186256e-01 -1.15210511e-01
7.80521810e-01 2.68856406e-01 4.30547029e-01 -1.49136555e+00
1.04016503e-02 -4.33900774e-01 -6.61021173e-01 -5.21034300e-01
4.16309565e-01 -1.24009418e+00 2.41561010e-01 -1.35859585e+00
2.87190944e-01 -8.96472812e-01 -2.45936528e-01 6.29028440e-01
1.06817469e-01 8.57787907e-01 -4.69522923e-02 2.52422750e-01
-5.68952501e-01 2.88623601e-01 1.27831054e+00 -7.37431467e-01
1.74137508e-03 3.58893499e-02 -3.24796259e-01 5.78912795e-01
9.11475599e-01 -7.84308612e-01 -3.01102221e-01 -2.81581968e-01
2.16015410e-02 -5.84870614e-02 3.47879112e-01 -1.12130547e+00
2.32961044e-01 1.47070363e-01 3.14923555e-01 -3.65443707e-01
2.03326881e-01 -1.09729099e+00 4.19559568e-01 7.25408077e-01
-5.10375559e-01 -4.65560228e-01 6.98137134e-02 3.31708074e-01
-4.30501997e-01 -3.39171112e-01 1.29562283e+00 -3.79931569e-01
-4.49586749e-01 5.94817400e-01 -3.17755550e-01 3.47423822e-01
1.27566862e+00 -2.52079546e-01 -1.51347145e-01 1.57076970e-01
-9.52998340e-01 7.79626705e-03 5.08379936e-01 -1.04339160e-01
4.19224262e-01 -1.08077061e+00 -6.84634924e-01 2.13816226e-01
2.15459809e-01 5.15950322e-01 4.63279873e-01 1.03486443e+00
-8.55850101e-01 1.38537243e-01 -3.71191114e-01 -1.05152905e+00
-1.23819411e+00 3.50257427e-01 4.26118433e-01 -2.28307813e-01
-2.78901547e-01 8.99425209e-01 3.65459710e-01 -8.00595522e-01
-1.61260851e-02 -3.36425632e-01 -2.46428341e-01 4.55874473e-01
1.22543208e-01 1.62162006e-01 3.16835731e-01 -8.06368828e-01
-1.00020580e-01 6.91789925e-01 1.08279325e-01 3.70131969e-01
1.19271886e+00 5.21811955e-02 -2.97042340e-01 4.79850233e-01
1.60289669e+00 -5.73104978e-01 -1.12544072e+00 -1.60454094e-01
4.88811731e-02 -1.31465614e-01 3.56894881e-02 -7.36611187e-01
-1.62694240e+00 1.04526412e+00 8.01146150e-01 7.65880570e-02
8.96833658e-01 3.91626470e-02 6.92264497e-01 4.04161394e-01
3.66142511e-01 -1.07745659e+00 3.62272002e-02 1.94045454e-01
4.07751113e-01 -1.54029036e+00 -5.36927134e-02 -3.76945287e-01
-5.13445377e-01 9.29531813e-01 4.25129026e-01 2.59942353e-01
6.95607364e-01 3.44258130e-01 4.91112739e-01 -3.66434902e-01
-4.78590876e-02 -6.91936612e-02 1.01147026e-01 5.07161021e-01
3.81837398e-01 1.20645612e-01 -8.65457207e-02 4.37462211e-01
1.50357634e-01 1.51634768e-01 2.02591762e-01 8.82230997e-01
-2.36221060e-01 -9.76610363e-01 -1.21532165e-01 8.01672935e-01
-7.97637522e-01 -1.56061044e-02 -1.28594622e-01 4.20141071e-01
2.63541371e-01 5.77173948e-01 4.04385589e-02 -9.06346142e-02
5.03043551e-03 -3.81206460e-02 2.49126777e-01 -7.99342096e-01
-7.07710922e-01 -8.78250301e-02 -2.37755969e-01 -2.16267928e-01
-7.47844338e-01 -5.59279919e-01 -1.38150120e+00 1.89679340e-01
-2.98222095e-01 2.33097181e-01 7.95591891e-01 1.07575464e+00
3.08748305e-01 4.86216813e-01 5.35807014e-01 -9.52446222e-01
-1.41371936e-01 -6.89649999e-01 -8.29513848e-01 6.42016768e-01
1.37861088e-01 -6.24806166e-01 -3.95221531e-01 4.20904607e-01] | [14.70792007446289, -2.3366856575012207] |
67653c00-677f-44f5-8b32-aad26062f138 | hdpv-slam-hybrid-depth-augmented-panoramic | 2301.11823 | null | https://arxiv.org/abs/2301.11823v3 | https://arxiv.org/pdf/2301.11823v3.pdf | HDPV-SLAM: Hybrid Depth-augmented Panoramic Visual SLAM for Mobile Mapping System with Tilted LiDAR and Panoramic Visual Camera | This paper proposes a novel visual simultaneous localization and mapping (SLAM) system called Hybrid Depth-augmented Panoramic Visual SLAM (HDPV-SLAM), that employs a panoramic camera and a tilted multi-beam LiDAR scanner to generate accurate and metrically-scaled trajectories. RGB-D SLAM was the design basis for HDPV-SLAM, which added depth information to visual features. It aims to solve the two major issues hindering the performance of similar SLAM systems. The first obstacle is the sparseness of LiDAR depth, which makes it difficult to correlate it with the extracted visual features of the RGB image. A deep learning-based depth estimation module for iteratively densifying sparse LiDAR depth was suggested to address this issue. The second issue pertains to the difficulties in depth association caused by a lack of horizontal overlap between the panoramic camera and the tilted LiDAR sensor. To surmount this difficulty, we present a hybrid depth association module that optimally combines depth information estimated by two independent procedures, feature-based triangulation and depth estimation. During a phase of feature tracking, this hybrid depth association module aims to maximize the use of more accurate depth information between the triangulated depth with visual features tracked and the deep learning-based corrected depth. We evaluated the efficacy of HDPV-SLAM using the 18.95 km-long York University and Teledyne Optech (YUTO) MMS dataset. The experimental results demonstrate that the two proposed modules contribute substantially to the performance of HDPV-SLAM, which surpasses that of the state-of-the-art (SOTA) SLAM systems. | ['Yujia Zhang', 'Gunho Sohn', 'Mohammad Moein Sheikholeslami', 'Zahra Arjmandi', 'Amin Alizadeh Naeini', 'Mostafa Ahmadi'] | 2023-01-27 | null | null | null | null | ['simultaneous-localization-and-mapping'] | ['computer-vision'] | [-8.19171667e-02 -3.44812900e-01 -3.13614495e-02 -3.33819002e-01
-7.16334164e-01 -1.98499739e-01 5.66451907e-01 -1.85700595e-01
-7.12112904e-01 8.30322623e-01 -1.85597286e-01 -1.71413541e-01
-1.77743912e-01 -8.65365744e-01 -5.24224043e-01 -4.77295756e-01
6.60067275e-02 7.87660718e-01 2.58925825e-01 -1.74952537e-01
5.28014839e-01 8.86561036e-01 -1.78249514e+00 -2.96776563e-01
9.59837198e-01 1.01959002e+00 6.96246743e-01 4.27844316e-01
-3.48446608e-01 2.53398806e-01 -1.02341361e-01 1.30185649e-01
5.29090106e-01 -1.90541238e-01 -2.67581761e-01 1.50875319e-02
8.60291243e-01 -5.17480075e-01 -2.83117980e-01 7.36244559e-01
4.07238007e-01 -3.29000503e-02 6.08106434e-01 -1.57057977e+00
5.02170958e-02 -3.62408996e-01 -8.98004770e-01 -1.78089216e-01
4.62795556e-01 -8.28190893e-02 6.83017552e-01 -1.19522047e+00
7.19042778e-01 1.04887021e+00 7.95253098e-01 9.37775299e-02
-9.23719287e-01 -7.18211055e-01 -3.56004775e-01 1.22380234e-01
-1.82269800e+00 -2.31428728e-01 7.14479804e-01 -6.07455194e-01
1.03002071e+00 -7.62759298e-02 1.05778956e+00 5.37186146e-01
6.73093319e-01 7.41085261e-02 1.14658999e+00 -5.13197482e-01
3.04958880e-01 8.46115872e-02 -4.62395966e-01 1.11843896e+00
5.02192974e-01 4.47214693e-01 -9.52430487e-01 -7.76707754e-02
1.05022824e+00 1.31872222e-01 -3.21069896e-01 -1.13489676e+00
-1.00647438e+00 1.04417479e+00 7.68763304e-01 6.32347092e-02
-2.62589514e-01 1.93109632e-01 -6.17857166e-02 -1.55523531e-02
1.83282807e-01 2.62105674e-01 -2.59968758e-01 -1.53908774e-01
-1.21960819e+00 -6.41379282e-02 3.39870453e-01 1.05735910e+00
1.51925254e+00 2.02301294e-01 7.63300419e-01 2.94631481e-01
7.06054807e-01 9.55874979e-01 3.48012894e-01 -1.04551435e+00
5.61556578e-01 7.28218615e-01 -1.21872919e-02 -9.23255324e-01
-5.23565471e-01 -4.16394435e-02 -5.46196520e-01 6.34809911e-01
1.92317203e-01 3.42124179e-02 -9.24653172e-01 1.21608615e+00
4.38760877e-01 -1.10049598e-01 1.04138412e-01 8.94048691e-01
5.80022633e-01 3.43011975e-01 -4.87345695e-01 -1.05982721e-01
8.12012613e-01 -5.73937893e-01 -5.53406954e-01 -5.95923781e-01
6.49313509e-01 -8.01728368e-01 7.84099698e-01 1.04434229e-01
-4.58834916e-01 -4.52404141e-01 -1.31683922e+00 -1.46368980e-01
-2.76972175e-01 -6.61808997e-02 6.77492619e-01 2.82786012e-01
-1.06634450e+00 2.44639263e-01 -8.51529062e-01 -7.01055884e-01
6.48916885e-02 4.18041497e-01 -7.63405740e-01 -2.34532058e-01
-8.38082910e-01 1.24585056e+00 2.35270619e-01 7.58942440e-02
-6.44087136e-01 -5.39026976e-01 -1.24862051e+00 -4.51811790e-01
-5.57515882e-02 -7.34250546e-01 8.56853366e-01 -5.83965600e-01
-1.33556223e+00 8.83652985e-01 -4.75314051e-01 -3.44813436e-01
6.29557669e-01 -3.46809149e-01 2.59490367e-02 2.84750402e-01
4.96523172e-01 9.47644114e-01 5.69402456e-01 -1.51308835e+00
-9.33326304e-01 -6.95366442e-01 -5.75737894e-01 5.84854603e-01
3.18832636e-01 -7.81414509e-01 -3.62072527e-01 1.76383168e-01
8.98493826e-01 -9.10282612e-01 -1.75236553e-01 3.12305331e-01
4.90094535e-02 4.10469890e-01 1.14076984e+00 -1.66282818e-01
5.95488727e-01 -1.92755103e+00 2.16760617e-02 2.06845507e-01
6.00533895e-02 -1.31482765e-01 1.88143075e-01 5.29078782e-01
4.94463980e-01 -4.26099032e-01 -1.76388443e-01 -7.03214645e-01
-4.36776191e-01 5.36368608e-01 -2.35194668e-01 8.74287605e-01
-3.08864534e-01 7.01614261e-01 -8.59513223e-01 -6.96269631e-01
8.61486614e-01 5.68366051e-01 -2.48898998e-01 1.22447066e-01
3.80659997e-02 6.86778307e-01 -1.35259047e-01 9.76270378e-01
1.03690076e+00 3.74290496e-01 -4.04984243e-02 -8.97191390e-02
-7.49622703e-01 1.04996286e-01 -1.21000397e+00 2.16622925e+00
-5.57739615e-01 7.63262749e-01 1.75175503e-01 -2.20784321e-01
1.22336471e+00 -1.24042884e-01 5.33206284e-01 -1.02263308e+00
5.56893721e-02 6.61430240e-01 -4.83069927e-01 -2.41514638e-01
9.88455057e-01 -4.12086368e-01 4.53951117e-03 -1.06055439e-01
9.33014378e-02 -9.00463164e-01 -5.13910115e-01 8.45225677e-02
6.44406497e-01 4.50038731e-01 6.79895580e-01 -1.66841358e-01
4.27395135e-01 4.32971388e-01 7.00517893e-01 4.76141006e-01
-2.34320894e-01 6.73572063e-01 -1.68186873e-01 -3.11535746e-01
-1.05928040e+00 -1.20362079e+00 -3.32727611e-01 1.45191655e-01
6.60257638e-01 -2.45129779e-01 -4.76358905e-02 -3.81706089e-01
4.33422893e-01 2.95016795e-01 -3.23984444e-01 1.87729850e-01
-3.39643419e-01 -1.97184861e-01 2.21408620e-01 3.36595982e-01
7.32483804e-01 -4.91346180e-01 -1.10412872e+00 1.34599194e-01
-1.79399610e-01 -1.01758051e+00 1.71211168e-01 4.24983978e-01
-1.01875925e+00 -1.08539164e+00 -2.77128249e-01 -4.91552562e-01
3.55215818e-01 6.76219344e-01 5.35767138e-01 -2.94055760e-01
-2.13140190e-01 4.44714546e-01 -1.08682044e-01 -1.17555171e-01
1.86391026e-02 -1.07873008e-01 3.42798799e-01 -3.71196032e-01
3.13137531e-01 -6.62210464e-01 -4.47435260e-01 3.53650868e-01
-3.59724402e-01 1.69285327e-01 6.56723320e-01 5.60666382e-01
8.64833474e-01 -3.62566233e-01 2.51446627e-02 -2.81486630e-01
-1.62598088e-01 -2.60673374e-01 -1.13657367e+00 -3.33647043e-01
-7.94930637e-01 -2.08709031e-01 7.96164870e-02 5.05701564e-02
-7.22942889e-01 6.52539253e-01 -1.21185929e-01 -5.55368483e-01
2.36873403e-02 3.63000005e-01 -1.15965001e-01 -7.53665686e-01
5.24502039e-01 4.06555086e-01 1.05797090e-01 -1.52822599e-01
2.83209234e-01 7.49404132e-01 5.56120932e-01 -1.96898773e-01
9.94982004e-01 1.01182783e+00 6.46398008e-01 -1.21201026e+00
-5.36518812e-01 -7.56888628e-01 -1.10426831e+00 -2.84363508e-01
7.73925304e-01 -1.27919352e+00 -4.22310621e-01 4.69603658e-01
-1.10625732e+00 -1.72521725e-01 -1.43326983e-01 8.64977181e-01
-7.53799200e-01 6.65920138e-01 -1.78306311e-01 -9.61817861e-01
-1.09262869e-01 -1.06959152e+00 1.26122844e+00 2.00768769e-01
-1.25780359e-01 -8.80685568e-01 4.86630946e-01 2.80779481e-01
1.82163432e-01 4.80413765e-01 3.60076964e-01 2.39475533e-01
-1.02988303e+00 -1.10875167e-01 -1.56983986e-01 -4.76151286e-03
1.39357567e-01 -8.32384229e-02 -1.01545477e+00 -4.49536145e-01
-2.91475914e-02 -1.87036708e-01 6.84581637e-01 2.61533380e-01
2.06603095e-01 3.06891769e-01 -5.25332510e-01 1.13259351e+00
1.95658016e+00 2.06608430e-01 5.75246751e-01 8.12937975e-01
9.39274490e-01 2.90409654e-01 1.19913352e+00 5.16708195e-01
7.13427126e-01 8.63294542e-01 9.85875547e-01 -1.08367682e-01
7.07928464e-02 -6.16996408e-01 1.79194674e-01 7.21051455e-01
1.61863416e-01 2.15079054e-01 -1.11279511e+00 6.38826430e-01
-1.79407287e+00 -4.50667560e-01 -4.42808926e-01 2.55797982e+00
2.38748550e-01 -1.31083820e-02 -4.87656564e-01 9.70448256e-02
2.61987746e-01 1.89004868e-01 -3.30457181e-01 -3.30934107e-01
-1.23709172e-01 1.33822262e-01 8.32409680e-01 1.10983944e+00
-8.15213978e-01 1.21006024e+00 5.13907146e+00 4.37515795e-01
-1.28865981e+00 8.27436000e-02 -6.30915284e-01 6.72878698e-02
-2.78861374e-01 4.26499367e-01 -1.12792039e+00 2.96808720e-01
7.14443088e-01 4.01826389e-02 1.33397907e-01 1.05393136e+00
2.01429278e-01 -9.23134625e-01 -8.83337379e-01 1.24199033e+00
2.03055263e-01 -1.19005740e+00 -8.73421039e-03 5.31540811e-01
7.42789030e-01 4.35921341e-01 -1.66749522e-01 3.46612372e-02
-7.97026083e-02 -6.98587239e-01 8.10303271e-01 2.89875805e-01
1.14981687e+00 -7.28216171e-01 7.46455967e-01 5.76214790e-01
-1.59718859e+00 -2.51630042e-02 -4.48876500e-01 -4.98794705e-01
2.82994241e-01 6.36740446e-01 -1.25858831e+00 8.63080680e-01
6.19302750e-01 6.19058728e-01 -4.73697126e-01 1.21459091e+00
-4.21028614e-01 -3.24478060e-01 -6.26519263e-01 4.05863643e-01
3.54956120e-01 -2.38038436e-01 4.60304826e-01 7.48826385e-01
7.37716734e-01 -3.89284968e-01 2.14254811e-01 6.94138944e-01
2.97538042e-01 -8.29140469e-02 -1.23742759e+00 6.04022324e-01
7.57276952e-01 1.16936004e+00 -4.45088297e-01 -1.01813346e-01
-3.04598510e-01 9.43778813e-01 2.94825733e-01 3.71291786e-02
-6.11416876e-01 -1.32071838e-01 7.57045925e-01 4.16007787e-01
8.62089321e-02 -8.48693490e-01 -5.10255575e-01 -1.08824003e+00
-1.90168515e-01 -1.46638036e-01 4.94538210e-02 -1.11831892e+00
-6.88736022e-01 3.70383382e-01 -1.81556821e-01 -1.64028180e+00
-2.90358871e-01 -3.46910000e-01 -7.94187933e-02 1.14685225e+00
-1.91398001e+00 -1.31695306e+00 -9.36468959e-01 5.91161728e-01
3.00796926e-01 1.24403268e-01 6.62033021e-01 1.54601976e-01
9.63758081e-02 -2.68274955e-02 1.20848916e-01 -3.80994678e-01
8.06353509e-01 -1.08463824e+00 1.17871538e-01 9.76593375e-01
1.41010985e-01 2.97546446e-01 5.26826859e-01 -1.02130544e+00
-1.51780081e+00 -9.97736335e-01 1.01757097e+00 -4.80300188e-01
3.16482216e-01 -3.98806989e-01 -6.21145964e-01 7.97419310e-01
-2.00494707e-01 2.54693422e-02 2.14483604e-01 -3.33952725e-01
2.14881320e-02 -3.39647263e-01 -1.26515973e+00 2.47609243e-01
8.01066756e-01 -7.91389585e-01 -6.75043643e-01 -2.34309491e-02
4.12397146e-01 -8.66212845e-01 -4.22169060e-01 7.18323648e-01
7.22141087e-01 -1.32816923e+00 7.29986966e-01 4.68476564e-01
1.07416764e-01 -6.83323562e-01 -6.93547904e-01 -1.09363520e+00
-1.64829083e-02 -1.52449578e-01 -5.60351945e-02 9.42146122e-01
-1.63893968e-01 -8.14877808e-01 9.49736357e-01 1.08854324e-01
-2.22396880e-01 -5.36441505e-01 -1.53344011e+00 -7.88170636e-01
-2.68682629e-01 -3.76356155e-01 2.62439340e-01 8.42378914e-01
-3.01648647e-01 2.85098493e-01 -2.64556289e-01 6.19750381e-01
9.44411159e-01 2.02215627e-01 1.17742860e+00 -1.44955432e+00
2.03521132e-01 1.13202743e-01 -7.55329847e-01 -1.06628430e+00
7.63625978e-03 -6.67570531e-01 1.96932703e-01 -1.77683234e+00
-1.87147841e-01 -5.47723830e-01 3.25724453e-01 1.30565241e-01
3.44360113e-01 2.66196042e-01 1.31687909e-01 4.91135091e-01
-1.92535400e-01 8.71071160e-01 9.93619204e-01 3.52154702e-01
-2.10433915e-01 -2.77163476e-01 4.20465246e-02 7.89848566e-01
4.15915132e-01 -5.61550379e-01 -3.07965666e-01 -5.93794465e-01
8.73885676e-02 3.56910169e-01 4.51533616e-01 -1.36922443e+00
5.34288347e-01 -1.88466638e-01 4.48049188e-01 -1.30624914e+00
9.31881189e-01 -1.20323324e+00 2.54922926e-01 7.03946471e-01
6.17315948e-01 2.33257830e-01 1.19752198e-01 4.62952673e-01
-2.77999312e-01 -2.54290737e-02 9.95961845e-01 -2.22689301e-01
-1.23951781e+00 2.93083787e-01 -2.74047226e-01 -3.99033099e-01
1.12828422e+00 -7.42899001e-01 -2.75156260e-01 -3.38671863e-01
-1.00999109e-01 1.88352942e-01 1.13293481e+00 2.12748975e-01
1.22747648e+00 -1.32349634e+00 -1.84824020e-01 6.87075675e-01
3.08041602e-01 4.52670604e-01 1.81758970e-01 9.06949162e-01
-1.04900610e+00 5.74889123e-01 -4.33735192e-01 -1.17208195e+00
-1.15185344e+00 2.98767477e-01 4.17999536e-01 1.41853437e-01
-7.67570138e-01 6.16488039e-01 -1.60341002e-02 -6.23524070e-01
5.11086434e-02 -3.10312688e-01 1.31215706e-01 1.78561658e-02
1.40129879e-01 4.20715511e-01 1.72894955e-01 -9.71855342e-01
-7.31066823e-01 1.31063747e+00 4.24174488e-01 -4.25930113e-01
1.01219344e+00 -6.45009577e-01 -3.02341413e-02 6.45111918e-01
1.07319713e+00 2.92085320e-01 -1.33270133e+00 -1.30074307e-01
-1.79005504e-01 -9.08688307e-01 2.80250102e-01 -3.31108660e-01
-6.88798487e-01 1.10078299e+00 8.11494946e-01 -5.43129742e-01
7.62046814e-01 -2.91937053e-01 6.00709856e-01 2.51534432e-01
1.08526206e+00 -7.79999256e-01 -4.49955203e-02 7.88379908e-01
6.07899785e-01 -1.43744326e+00 3.38021427e-01 -2.88649589e-01
-5.21839917e-01 1.20381856e+00 7.82491863e-01 7.25341663e-02
2.44929433e-01 3.22303087e-01 3.78843576e-01 -2.63672709e-01
-8.35175216e-02 -3.48354727e-01 -1.22688234e-01 7.01094627e-01
-6.89848810e-02 -1.62489414e-01 -3.97706665e-02 -3.44644129e-01
-2.71978587e-01 5.32932170e-02 5.06149709e-01 1.24788749e+00
-8.78949821e-01 -8.43999863e-01 -6.05819285e-01 -1.00143522e-01
4.27783668e-01 -1.34419566e-02 -2.02151030e-01 1.27083135e+00
4.43601966e-01 6.21273875e-01 2.87395805e-01 -5.86914599e-01
1.59837723e-01 -1.12009250e-01 6.26425147e-01 -4.77968782e-01
5.47424480e-02 -2.87102144e-02 -1.58560932e-01 -8.86602342e-01
-3.08585405e-01 -5.53776145e-01 -1.43802929e+00 -2.79701561e-01
-1.88214317e-01 1.11345410e-01 1.28873312e+00 8.48541558e-01
2.52308637e-01 -5.12002632e-02 5.83739579e-01 -1.25765502e+00
-1.06952563e-01 -6.99530959e-01 -8.58423948e-01 -2.07574919e-01
6.58721805e-01 -1.05936110e+00 -5.15055120e-01 -5.45685709e-01] | [7.519136428833008, -2.247694969177246] |
bd00fd62-2c7d-4b8b-b730-40a0b20c0fe1 | towards-goal-feasibility-and-diversity | 2206.07170 | null | https://arxiv.org/abs/2206.07170v1 | https://arxiv.org/pdf/2206.07170v1.pdf | Towards Goal, Feasibility, and Diversity-Oriented Deep Generative Models in Design | Deep Generative Machine Learning Models (DGMs) have been growing in popularity across the design community thanks to their ability to learn and mimic complex data distributions. DGMs are conventionally trained to minimize statistical divergence between the distribution over generated data and distribution over the dataset on which they are trained. While sufficient for the task of generating "realistic" fake data, this objective is typically insufficient for design synthesis tasks. Instead, design problems typically call for adherence to design requirements, such as performance targets and constraints. Advancing DGMs in engineering design requires new training objectives which promote engineering design objectives. In this paper, we present the first Deep Generative Model that simultaneously optimizes for performance, feasibility, diversity, and target achievement. We benchmark performance of the proposed method against several Deep Generative Models over eight evaluation metrics that focus on feasibility, diversity, and satisfaction of design performance targets. Methods are tested on a challenging multi-objective bicycle frame design problem with skewed, multimodal data of different datatypes. The proposed framework was found to outperform all Deep Generative Models in six of eight metrics. | ['Faez Ahmed', 'Lyle Regenwetter'] | 2022-06-14 | null | null | null | null | ['design-synthesis'] | ['adversarial'] | [ 7.14957789e-02 1.41841725e-01 -1.70187056e-01 -5.33384025e-01
-8.08726370e-01 -4.55077916e-01 5.87622583e-01 -1.81963608e-01
3.79999250e-01 8.31473470e-01 2.14678437e-01 -2.05711387e-02
-3.64973545e-01 -9.20434952e-01 -6.46300316e-01 -6.31131709e-01
4.05441433e-01 8.44900191e-01 -5.71259201e-01 -2.90670961e-01
3.23459089e-01 3.35394174e-01 -1.60840440e+00 3.02883804e-01
6.18385732e-01 1.04883409e+00 -6.44952729e-02 5.70585966e-01
3.69258940e-01 6.95513666e-01 -9.44490790e-01 -3.17558765e-01
9.59363859e-03 -5.89175105e-01 -4.32789683e-01 1.83925778e-01
1.86043441e-01 -8.27613398e-02 -1.54018641e-01 7.09752738e-01
1.07988405e+00 5.25183938e-02 9.88012731e-01 -1.51299453e+00
-1.03553903e+00 4.24059719e-01 -3.43865871e-01 -4.01294231e-01
1.34802252e-01 3.50070834e-01 1.25469100e+00 -8.36722732e-01
3.01486254e-01 1.02395678e+00 4.62603956e-01 5.20062506e-01
-1.51308286e+00 -4.47436899e-01 -2.24132895e-01 -2.07761884e-01
-1.25303185e+00 -4.07195389e-01 1.02244365e+00 -7.33358383e-01
8.25311959e-01 1.42223686e-01 4.97032940e-01 1.38449776e+00
5.75934231e-01 5.04288912e-01 6.56072676e-01 -1.96368456e-01
7.22031057e-01 7.34203905e-02 -3.80194873e-01 2.66495079e-01
1.34213313e-01 3.43422294e-01 -5.33256471e-01 -5.02392687e-02
4.93829787e-01 -4.27458286e-01 6.82315305e-02 -5.24354100e-01
-7.18598604e-01 1.06965172e+00 1.69344962e-01 6.46792278e-02
-3.69912118e-01 6.58091843e-01 2.51376361e-01 8.81651193e-02
3.50542009e-01 8.06763351e-01 -2.64489651e-01 -3.79591823e-01
-9.73611295e-01 6.06406748e-01 6.15054131e-01 1.19374239e+00
5.63685477e-01 7.26374030e-01 -3.64659011e-01 9.21734989e-01
5.60869992e-01 4.60963339e-01 4.02249992e-01 -6.93793952e-01
2.92410970e-01 4.78897274e-01 2.26659343e-01 -1.34427261e+00
-3.25713456e-01 -8.93625975e-01 -7.42528617e-01 3.52197230e-01
-4.85669300e-02 -3.62609267e-01 -1.01940918e+00 1.84757829e+00
1.11640528e-01 -5.54425299e-01 -1.64129987e-01 8.87402594e-01
9.21095788e-01 8.54818046e-01 5.65077411e-03 1.13203816e-01
6.86271071e-01 -4.57773626e-01 -4.96562421e-01 -1.86898842e-01
5.89690804e-01 -1.00584507e+00 1.10060704e+00 4.52339888e-01
-9.92465317e-01 -8.14211667e-01 -1.08487487e+00 3.17592204e-01
-2.30924245e-02 4.32222396e-01 5.20438671e-01 1.27286232e+00
-8.12965870e-01 3.75307113e-01 -4.18628335e-01 1.43274933e-01
4.46243882e-01 4.64051217e-01 2.56710827e-01 2.12825388e-01
-8.24580371e-01 6.56140924e-01 1.80961266e-01 1.82806641e-01
-1.49890769e+00 -1.01090574e+00 -6.51981950e-01 1.56340018e-01
9.20158103e-02 -7.93783545e-01 1.18968344e+00 -8.05436194e-01
-1.67734218e+00 4.84554887e-01 5.38711071e-01 -5.77750914e-02
3.88204426e-01 -2.43314266e-01 -4.81273413e-01 -8.28099906e-01
-1.08576089e-01 5.79216480e-01 8.38667989e-01 -1.42309809e+00
-3.01228106e-01 -4.05199192e-02 -6.98757470e-02 -2.36650467e-01
-1.14010625e-01 -5.54683805e-01 2.41097257e-01 -6.41792953e-01
-2.96540529e-01 -1.01064169e+00 -1.57800347e-01 -6.21345460e-01
-7.62760222e-01 -8.66612196e-02 8.74885619e-01 -4.87568647e-01
1.48952639e+00 -1.69225407e+00 2.96276242e-01 4.13216710e-01
-6.60948316e-03 5.51099777e-02 1.58315748e-02 6.65921509e-01
-1.69142753e-01 5.18471934e-02 -3.34310369e-03 -1.57412931e-01
5.46073198e-01 2.31531128e-01 -1.81360170e-01 4.92722631e-01
4.28425670e-01 9.79355633e-01 -5.78516781e-01 3.40937003e-02
2.67835736e-01 3.28946650e-01 -1.06348193e+00 1.36727214e-01
-6.55768275e-01 3.98857623e-01 -3.33078414e-01 5.76777041e-01
3.91008466e-01 -2.52164930e-01 2.78102726e-01 -3.79155338e-01
1.28968358e-01 -7.95466602e-02 -1.10994148e+00 1.48633766e+00
-7.66016066e-01 6.19683743e-01 -4.90518153e-01 -1.10253465e+00
1.44503689e+00 1.76226720e-01 7.01205313e-01 -7.66811788e-01
4.24930573e-01 1.24638960e-01 2.35467643e-01 -2.89375901e-01
5.82182884e-01 -1.56826019e-01 -6.22969747e-01 5.28965473e-01
7.00239092e-02 -5.11607349e-01 -7.83831105e-02 -3.13967824e-01
8.84713650e-01 2.81764477e-01 -2.01933533e-01 -4.21279609e-01
2.22926080e-01 -2.03564450e-01 5.34477770e-01 3.51771593e-01
4.10373300e-01 8.12493801e-01 5.96618056e-01 -5.00194848e-01
-1.60394919e+00 -1.13607979e+00 2.35267207e-01 8.97575319e-01
-2.76322305e-01 -3.29690188e-01 -4.76728678e-01 -4.24142897e-01
2.22039372e-02 1.18059719e+00 -7.29956925e-01 -6.58237219e-01
-4.26573277e-01 -8.84143591e-01 4.03058439e-01 3.85682642e-01
9.98397022e-02 -7.68194377e-01 -7.19667792e-01 2.43337363e-01
1.41727123e-02 -5.80924571e-01 -4.37090755e-01 -1.21259063e-01
-3.62004995e-01 -7.70613074e-01 -6.61692560e-01 -7.25281954e-01
6.15048826e-01 -6.17694318e-01 1.48116004e+00 -2.48048618e-01
-6.18185639e-01 1.95951760e-01 -6.88168555e-02 -4.45026517e-01
-7.98658073e-01 3.64372760e-01 -1.29356667e-01 -5.59418313e-02
-6.01911638e-03 -6.86186016e-01 -7.08063006e-01 6.78065658e-01
-8.33181441e-01 1.66988477e-01 4.79066163e-01 9.23225343e-01
3.87510687e-01 1.85584307e-01 9.91672039e-01 -3.93494070e-01
8.70089948e-01 -5.44052064e-01 -5.64269364e-01 2.98116952e-01
-7.08917856e-01 3.80783230e-01 4.95508641e-01 -4.10143942e-01
-9.84569371e-01 -2.01143846e-01 -1.81168571e-01 -2.35443354e-01
2.36207843e-02 3.68853062e-01 -4.90942776e-01 2.52520293e-01
7.39352942e-01 -6.27432317e-02 -1.75569370e-01 -2.92469442e-01
4.45016354e-01 6.11066818e-01 2.56086588e-01 -1.12012136e+00
5.93260884e-01 -1.28160179e-01 2.37233952e-01 -5.51696301e-01
-4.05141860e-01 2.40587935e-01 3.35099152e-03 -5.79752505e-01
5.06214976e-01 -5.17280936e-01 -8.01227868e-01 3.25914234e-01
-9.53569591e-01 8.92573316e-03 -4.35748488e-01 2.85333663e-01
-1.01571774e+00 -2.40569636e-01 1.14547282e-01 -8.11686575e-01
-3.38680297e-01 -1.38006508e+00 1.21548975e+00 7.72174969e-02
-6.81866527e-01 -9.38997149e-01 3.20296645e-01 7.22440034e-02
6.67658269e-01 8.98914397e-01 1.23107445e+00 -1.77513689e-01
-3.12711477e-01 -3.52586687e-01 1.85740337e-01 5.70779622e-01
5.14333904e-01 3.51830512e-01 -7.64784992e-01 -2.35212758e-01
-2.21008867e-01 -4.38497275e-01 1.08818859e-01 6.68640912e-01
1.07874501e+00 -3.26986432e-01 1.64453369e-02 3.58893305e-01
1.47127271e+00 4.90080088e-01 8.12377512e-01 -3.71318907e-02
5.75578570e-01 6.29697084e-01 4.06868905e-01 1.00645506e+00
-6.79479912e-02 8.37973475e-01 5.24481058e-01 2.54808385e-02
-1.21232204e-01 -3.20209712e-01 1.02779798e-01 3.52483958e-01
9.41007361e-02 -6.83459103e-01 -1.02567327e+00 6.03681564e-01
-1.81558669e+00 -8.39475393e-01 3.76579687e-02 1.98279107e+00
5.74118376e-01 2.39006504e-01 3.24444979e-01 1.14279933e-01
6.53575301e-01 -8.23072791e-02 -6.78144276e-01 -7.99277067e-01
5.11715561e-03 4.59318727e-01 2.58081287e-01 1.38247535e-01
-6.17376745e-01 3.33745211e-01 6.19581747e+00 9.23280120e-01
-1.03161323e+00 -1.60693139e-01 1.13662672e+00 -3.25829953e-01
-8.25320899e-01 -1.53538764e-01 -6.36390746e-01 5.69597840e-01
8.82966518e-01 -3.61380458e-01 8.86734203e-02 1.14244413e+00
4.18299228e-01 1.15329497e-01 -1.50929642e+00 1.06742322e+00
-1.08106330e-01 -1.53020895e+00 -8.09030235e-02 3.24824691e-01
1.32170820e+00 -5.97492993e-01 8.00856829e-01 3.28576624e-01
4.30288613e-01 -1.63095319e+00 1.00306296e+00 5.14850974e-01
7.12748528e-01 -1.22639275e+00 5.71629226e-01 3.22754011e-02
-7.29854286e-01 5.05109970e-03 -2.05045287e-02 1.40333384e-01
3.27348024e-01 7.08362162e-01 -1.03748477e+00 4.29245770e-01
2.33445466e-01 3.48603189e-01 -3.03179085e-01 7.90944636e-01
1.98743999e-01 4.67823237e-01 -1.96552694e-01 -3.48455042e-01
3.04724991e-01 7.82551393e-02 3.21955949e-01 8.43040705e-01
7.92890191e-01 -5.81281006e-01 -4.18734290e-02 1.50565493e+00
9.78314206e-02 -1.90215483e-01 -4.75802094e-01 -4.66092259e-01
2.59292066e-01 8.15492153e-01 -2.20838249e-01 2.35248670e-01
1.24542199e-01 3.54202479e-01 -3.67084444e-01 3.26273829e-01
-1.48053145e+00 -1.52516127e-01 9.24942791e-01 3.85294050e-01
2.95995206e-01 -2.20454752e-01 -7.30168879e-01 -3.46911907e-01
-1.40106201e-01 -1.01935196e+00 -6.29311502e-02 -7.38311112e-01
-1.14541030e+00 4.53456163e-01 1.30614564e-01 -1.33119690e+00
-6.37926400e-01 -5.22714257e-01 -3.23205024e-01 9.27738667e-01
-6.98966622e-01 -1.32478464e+00 -2.14340568e-01 4.99871969e-02
6.43869162e-01 -5.94321847e-01 5.46728194e-01 6.12478673e-01
-5.09040058e-01 6.54306769e-01 -1.01126498e-02 -3.08737606e-01
3.74071389e-01 -9.31709588e-01 4.74957675e-01 3.67208332e-01
-1.39561787e-01 2.39441991e-01 1.21326935e+00 -5.60513794e-01
-1.35423124e+00 -1.00981104e+00 3.01419884e-01 -3.08147073e-01
2.33806491e-01 -2.69941717e-01 -2.44305372e-01 1.84305817e-01
4.71793339e-02 -5.19619465e-01 8.07118416e-01 -1.35949120e-01
-6.33746907e-02 -1.01148978e-01 -1.13025630e+00 2.76599020e-01
7.52743006e-01 -1.96123496e-02 8.46375450e-02 2.41627231e-01
3.12253147e-01 -4.80896384e-01 -1.12521231e+00 4.94192272e-01
7.42566466e-01 -8.22940409e-01 9.43598211e-01 -8.99163246e-01
9.74640608e-01 -2.75271386e-01 -5.98490357e-01 -1.57931399e+00
-3.54702175e-01 -7.39622176e-01 -5.50594404e-02 1.37081420e+00
6.20322585e-01 -2.85132192e-02 1.19660234e+00 8.53663623e-01
-3.81341189e-01 -9.99314904e-01 -6.90675676e-01 -6.97804511e-01
-2.92118508e-02 -5.05319476e-01 1.07887471e+00 4.45126891e-01
-5.81548393e-01 4.59913731e-01 -7.98457086e-01 -1.05728887e-01
5.35726249e-01 1.40308768e-01 1.01657677e+00 -7.93586969e-01
-5.78955531e-01 -6.97372139e-01 -6.54044688e-01 -7.40532577e-01
-1.53727397e-01 -5.48870623e-01 1.25089556e-01 -1.39864433e+00
-2.06162661e-01 -4.69700903e-01 1.74989045e-01 2.81504467e-02
4.22096282e-01 -9.12825689e-02 -1.21934377e-01 -3.72840852e-01
-6.11053668e-02 1.12728345e+00 1.22682977e+00 -3.04860413e-01
-2.51191586e-01 2.29595006e-01 -7.61396766e-01 2.71695465e-01
9.23625886e-01 -2.17417940e-01 -8.72465312e-01 -3.77653718e-01
7.01620162e-01 -1.09144710e-01 3.10020864e-01 -1.02735925e+00
-3.12551469e-01 -3.82494211e-01 4.85049486e-01 -5.66723228e-01
2.67340153e-01 -6.58497334e-01 9.65827584e-01 3.49390119e-01
-3.32942784e-01 2.41116568e-01 1.49064347e-01 3.31733704e-01
-1.87753476e-02 1.14513949e-01 8.45578730e-01 2.82492161e-01
-5.73665023e-01 8.33036005e-02 -1.87659383e-01 5.66439852e-02
1.06728327e+00 -1.86250597e-01 -2.93151468e-01 -5.99064708e-01
-7.46254921e-01 -2.16197073e-02 2.98037112e-01 8.09784770e-01
7.65485942e-01 -1.83957434e+00 -8.34125757e-01 1.18316486e-01
2.17751205e-01 -8.53640735e-02 2.59500653e-01 3.45966667e-01
-5.98341167e-01 4.54158902e-01 -2.65591741e-01 -7.72639930e-01
-7.30939627e-01 3.27097356e-01 4.55552936e-01 -1.67048737e-01
-5.78042418e-02 7.85962522e-01 3.30425292e-01 -4.16944414e-01
-4.54891063e-02 -4.49503452e-01 4.00254875e-01 -1.45609185e-01
-5.10380678e-02 5.60301900e-01 3.34552437e-01 -4.97338206e-01
-2.33108729e-01 3.89822990e-01 3.31763864e-01 -5.12856469e-02
1.37784851e+00 1.69382975e-01 3.90734166e-01 1.12397164e-01
1.34782720e+00 -5.24695694e-01 -1.07062316e+00 2.30865747e-01
-1.62047908e-01 -4.54028696e-01 2.69337475e-01 -8.42141986e-01
-1.24920082e+00 6.11711502e-01 8.36709380e-01 1.33532241e-01
1.05348182e+00 -1.52330413e-01 8.47710252e-01 -1.38791025e-01
2.24403933e-01 -1.35709727e+00 6.20673537e-01 1.56099468e-01
1.17924738e+00 -8.96215260e-01 -1.70400783e-01 9.26185325e-02
-6.29416823e-01 8.99318397e-01 6.63411677e-01 -1.33697271e-01
7.59790480e-01 3.04275393e-01 -4.32293177e-01 -4.08580452e-01
-7.16446221e-01 2.55052477e-01 5.56761146e-01 6.55930579e-01
6.79337978e-01 1.35761186e-01 1.39303878e-02 4.78576452e-01
-4.63026136e-01 -1.00277662e-01 1.04572296e-01 8.41574430e-01
-3.80152702e-01 -1.19774568e+00 -3.76053005e-01 4.43738788e-01
-4.79855910e-02 3.66346091e-01 -2.57029891e-01 7.44622350e-01
2.83599198e-01 9.11714792e-01 -6.96311519e-02 -6.32710278e-01
5.38794875e-01 -3.19097608e-01 5.09281993e-01 -5.68731487e-01
-3.68049204e-01 -4.59335707e-02 3.15493584e-01 -3.34252477e-01
-1.26947373e-01 -4.35147285e-01 -7.08576441e-01 -5.51960051e-01
-2.74243236e-01 1.02445921e-02 8.47742498e-01 6.90410137e-01
3.77484679e-01 1.02775300e+00 8.94511044e-01 -5.76992273e-01
-5.70352495e-01 -7.72178352e-01 -3.05384547e-01 2.88632095e-01
3.64375487e-02 -9.04131174e-01 1.48050517e-01 8.82969946e-02] | [5.813911437988281, 3.305405855178833] |
f99ab37a-708b-47d2-9dae-57d4d4a57f8d | identity-aware-multi-sentence-video | 2008.09791 | null | https://arxiv.org/abs/2008.09791v1 | https://arxiv.org/pdf/2008.09791v1.pdf | Identity-Aware Multi-Sentence Video Description | Standard video and movie description tasks abstract away from person identities, thus failing to link identities across sentences. We propose a multi-sentence Identity-Aware Video Description task, which overcomes this limitation and requires to re-identify persons locally within a set of consecutive clips. We introduce an auxiliary task of Fill-in the Identity, that aims to predict persons' IDs consistently within a set of clips, when the video descriptions are given. Our proposed approach to this task leverages a Transformer architecture allowing for coherent joint prediction of multiple IDs. One of the key components is a gender-aware textual representation as well an additional gender prediction objective in the main model. This auxiliary task allows us to propose a two-stage approach to Identity-Aware Video Description. We first generate multi-sentence video descriptions, and then apply our Fill-in the Identity model to establish links between the predicted person entities. To be able to tackle both tasks, we augment the Large Scale Movie Description Challenge (LSMDC) benchmark with new annotations suited for our problem statement. Experiments show that our proposed Fill-in the Identity model is superior to several baselines and recent works, and allows us to generate descriptions with locally re-identified people. | ['Jae Sung Park', 'Anna Rohrbach', 'Trevor Darrell'] | 2020-08-22 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3739_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123660358.pdf | eccv-2020-8 | ['video-description', 'gender-prediction'] | ['computer-vision', 'computer-vision'] | [ 1.42182216e-01 -1.43627360e-01 -8.69259238e-02 -6.99114561e-01
-1.09132588e+00 -7.48967767e-01 9.72356081e-01 1.94102779e-01
-3.50535452e-01 6.52310491e-01 5.78267753e-01 5.27161598e-01
1.92849800e-01 -4.13201600e-01 -6.67192578e-01 -3.38411450e-01
1.21360011e-01 7.73457587e-01 1.00715578e-01 -5.84910028e-02
8.30205828e-02 1.08035445e-01 -1.66181242e+00 6.73368514e-01
4.17348027e-01 9.21805441e-01 -1.85500886e-02 6.07891023e-01
1.22543618e-01 1.08362889e+00 -4.32004571e-01 -1.11422825e+00
8.16532969e-02 -5.04788220e-01 -1.02099574e+00 3.12089831e-01
9.35599029e-01 -4.18951273e-01 -2.93641627e-01 8.20360959e-01
5.17285168e-01 3.35400432e-01 8.31014276e-01 -1.42795348e+00
-6.29603028e-01 8.07112515e-01 -3.61776561e-01 -9.97785758e-03
1.00750947e+00 -2.53035635e-01 1.24869490e+00 -9.66815948e-01
1.09597409e+00 1.23354030e+00 1.04250121e+00 9.30647492e-01
-1.13127506e+00 -5.03002644e-01 2.05954850e-01 4.35221612e-01
-1.86404359e+00 -7.55411208e-01 6.31339192e-01 -6.06625676e-01
6.62592649e-01 3.27235043e-01 4.77829039e-01 1.63149095e+00
-4.81878638e-01 7.38207221e-01 5.07463336e-01 -1.65747136e-01
-6.48788884e-02 4.24728453e-01 -1.25542849e-01 5.38623750e-01
-2.92540761e-03 -5.43134451e-01 -7.98178196e-01 -9.41852108e-02
5.01449883e-01 -1.20231688e-01 -1.41795710e-01 -3.15345943e-01
-1.28525519e+00 6.30778968e-01 -2.89810281e-02 2.94109583e-01
-2.21543759e-01 1.26698121e-01 7.91744649e-01 -1.18483216e-01
6.26252830e-01 3.99800360e-01 -2.80632265e-02 -2.96941340e-01
-1.24258769e+00 8.06220651e-01 8.17467451e-01 1.20446455e+00
6.36597514e-01 -5.05465329e-01 -6.65051341e-01 9.58783090e-01
-7.12690651e-02 2.50742376e-01 2.85615653e-01 -1.09982336e+00
5.04557967e-01 3.82338345e-01 2.79934317e-01 -1.15146673e+00
-2.03408062e-01 -1.94756493e-01 -7.24632144e-01 -6.48833752e-01
3.13707620e-01 7.16417357e-02 -3.17679733e-01 2.00778151e+00
2.59664923e-01 5.62238753e-01 1.40372682e-02 8.30140710e-01
1.16910553e+00 4.67993677e-01 1.44949734e-01 -8.26064870e-02
1.55808973e+00 -9.49559093e-01 -5.34965754e-01 -7.42212832e-02
4.36821193e-01 -3.70864302e-01 5.27422249e-01 -1.20652840e-01
-1.32301879e+00 -7.05606818e-01 -7.21735477e-01 -4.05371100e-01
-2.89472222e-01 5.03803730e-01 2.73138463e-01 4.37688202e-01
-1.19484890e+00 5.31654596e-01 -4.50506300e-01 -6.79997683e-01
2.69456804e-01 3.50911617e-01 -7.88018227e-01 -5.93272597e-02
-1.19118047e+00 7.61142313e-01 3.70784670e-01 -1.70433670e-01
-5.61847210e-01 -7.46265411e-01 -1.32798111e+00 2.02984530e-02
3.00439090e-01 -1.08308566e+00 1.19425416e+00 -1.15903664e+00
-9.67697978e-01 1.40846980e+00 -6.21881843e-01 -4.73637670e-01
6.29486561e-01 -7.91913550e-03 -2.99618453e-01 4.69902068e-01
6.50761366e-01 8.98532987e-01 7.03280628e-01 -1.43586457e+00
-9.56444681e-01 -2.98278600e-01 2.37789840e-01 2.13700905e-01
-5.60762465e-01 3.57835293e-01 -1.06066096e+00 -6.70744598e-01
-4.07153845e-01 -9.57325697e-01 1.29721314e-01 -3.46252829e-01
-4.40823525e-01 -3.43151987e-01 3.26883197e-01 -8.45363438e-01
1.40394938e+00 -2.05872822e+00 4.91423935e-01 -4.37875576e-02
2.59111583e-01 -1.60141334e-01 -1.43488213e-01 5.98192632e-01
3.59076671e-02 5.74937165e-02 2.80538667e-02 -1.46054947e+00
3.10517490e-01 -9.08360556e-02 -1.19983718e-01 3.52112681e-01
2.21228525e-01 6.51888549e-01 -8.96200418e-01 -8.06902170e-01
-2.04211310e-01 7.51918435e-01 -6.57400787e-01 3.57732683e-01
9.10533243e-04 5.08971035e-01 -8.00683349e-02 6.37759864e-01
5.16338944e-01 -1.41802400e-01 2.59095669e-01 -4.53297675e-01
-1.77274838e-01 2.10070927e-02 -1.12480104e+00 1.84170604e+00
-3.06604117e-01 4.68768686e-01 -1.37798682e-01 -9.61731195e-01
7.15980411e-01 6.10632479e-01 7.40885437e-01 -3.12129676e-01
-4.93458994e-02 2.23810419e-01 -6.78350627e-01 -6.67368531e-01
7.55235910e-01 -1.22846112e-01 -7.14674652e-01 2.43468180e-01
5.10307014e-01 3.93813610e-01 7.25738287e-01 4.98888493e-01
8.99509430e-01 3.70831043e-01 2.45673075e-01 1.57331809e-01
9.99879181e-01 -2.21123561e-01 7.09526956e-01 7.23055840e-01
-1.93207979e-01 1.07601428e+00 6.56126380e-01 -5.46175778e-01
-1.24323440e+00 -8.37396979e-01 1.98839754e-01 1.21396065e+00
1.55793563e-01 -8.58075261e-01 -7.89321661e-01 -8.11586201e-01
-2.76288092e-02 3.57582033e-01 -8.76100481e-01 5.01283146e-02
-6.34020329e-01 -3.77552450e-01 6.04414403e-01 5.34835517e-01
4.33735400e-01 -6.47971749e-01 -6.91468120e-02 5.94093092e-02
-7.50449002e-01 -1.69442213e+00 -8.24380100e-01 -6.14189744e-01
2.95628887e-02 -8.32934022e-01 -8.87789309e-01 -9.47378218e-01
6.59134209e-01 -1.64802060e-01 1.29552269e+00 -2.20635720e-02
-1.42444730e-01 7.14853108e-01 -5.01903713e-01 -7.51114916e-03
-3.46868664e-01 4.17726815e-01 3.03988278e-01 6.78482950e-01
4.95796889e-01 -5.73679924e-01 -4.82240885e-01 9.41336378e-02
-5.31984389e-01 2.22150579e-01 2.68846512e-01 6.91639245e-01
4.49138671e-01 -3.48413944e-01 7.11674094e-01 -8.68537784e-01
1.40108481e-01 -6.00106776e-01 -6.46300986e-02 4.22312915e-01
3.16884229e-03 -1.84267640e-01 7.49015450e-01 -2.52263933e-01
-9.74884152e-01 4.75715071e-01 -2.30928570e-01 -4.99040455e-01
-1.35852724e-01 2.43823245e-01 -3.64262819e-01 2.02563643e-01
1.67783707e-01 3.75401706e-01 -1.61497533e-01 -3.89791697e-01
1.45136341e-01 6.77019835e-01 1.01322186e+00 -5.44237494e-01
9.55089271e-01 5.09004176e-01 -1.85748354e-01 -5.50944328e-01
-1.06539857e+00 -7.83987582e-01 -1.06154943e+00 -4.72964346e-01
1.09862125e+00 -1.37912154e+00 -9.27954197e-01 3.38822603e-01
-1.31241941e+00 1.97688013e-01 -8.05653408e-02 2.01667026e-01
-7.12438226e-01 3.77919227e-01 -5.47928870e-01 -7.69140601e-01
-2.15404451e-01 -7.86549926e-01 1.52554595e+00 2.01682955e-01
-5.58446407e-01 -1.10298312e+00 8.63561183e-02 6.93882704e-01
1.64162993e-01 3.60636115e-01 4.08703893e-01 -1.06147075e+00
-3.32046896e-01 -4.15781736e-01 -2.22623572e-01 -9.62380040e-03
-8.96760747e-02 -1.51460245e-01 -8.75198424e-01 -2.46759489e-01
-4.32448983e-01 -4.32493329e-01 9.15973485e-01 6.93539828e-02
9.14052069e-01 -4.59344953e-01 -3.87295574e-01 6.72749698e-01
1.29130232e+00 -3.37286502e-01 4.16674256e-01 3.52722734e-01
9.46401775e-01 7.76496768e-01 5.79817235e-01 7.69049823e-01
1.18042696e+00 1.12645829e+00 4.21055593e-02 8.80221352e-02
-6.04040688e-03 -4.79987562e-01 3.31245303e-01 5.01548946e-01
-3.31122279e-01 -1.58177078e-01 -7.17310309e-01 8.03645670e-01
-2.14090562e+00 -1.53928185e+00 1.45455807e-01 2.07235885e+00
7.66702592e-01 -2.80929327e-01 7.40904093e-01 -1.16376907e-01
9.10454214e-01 1.38775870e-01 -2.30913386e-01 -2.66432520e-02
-1.17593341e-01 -2.66230226e-01 1.55537009e-01 3.13977033e-01
-1.54902768e+00 7.90501416e-01 5.59242344e+00 6.65078044e-01
-7.09526062e-01 1.41975254e-01 5.37924826e-01 -2.82545596e-01
-1.90371409e-01 -1.83484301e-01 -1.17505431e+00 4.77512807e-01
8.11121285e-01 -3.81562531e-01 1.75988615e-01 6.83636129e-01
1.71281412e-01 1.75018281e-01 -1.62643147e+00 1.33187413e+00
8.75037253e-01 -1.35858250e+00 3.31383526e-01 -2.42729694e-01
6.18538141e-01 -7.27678776e-01 7.29048857e-03 4.76337850e-01
-3.01550329e-01 -8.89693439e-01 1.20327902e+00 5.98355532e-01
9.64994729e-01 -7.27622509e-01 8.32302272e-01 2.72948742e-02
-1.60846210e+00 -1.26981631e-01 -1.36115432e-01 4.26154509e-02
4.83418196e-01 3.67174521e-02 -7.14563966e-01 7.18550861e-01
6.21079087e-01 1.19071317e+00 -8.04108083e-01 7.87811160e-01
2.05194443e-01 -3.76954265e-02 1.40412048e-01 1.76955327e-01
-3.17224562e-02 3.56091112e-02 4.88788515e-01 1.50924468e+00
4.91226643e-01 -1.19986068e-02 2.55477369e-01 8.19042504e-01
-3.39837551e-01 2.40032047e-01 -4.03188229e-01 6.03245348e-02
6.41134381e-01 1.25688601e+00 -5.31889617e-01 -5.23969889e-01
-5.78213036e-01 1.50529075e+00 4.43555623e-01 8.02464932e-02
-1.04470015e+00 -7.93688595e-02 6.90432191e-01 2.11516872e-01
3.59507889e-01 5.87114580e-02 1.96726725e-01 -1.43258488e+00
1.30981028e-01 -7.02242732e-01 5.80855429e-01 -6.36267781e-01
-1.38500512e+00 6.22051299e-01 1.35077879e-01 -1.30525637e+00
-6.51962817e-01 -1.56087726e-01 -3.97549301e-01 5.88784635e-01
-1.39848471e+00 -2.03100443e+00 -2.74896979e-01 6.14760339e-01
7.65847385e-01 -2.82471150e-01 8.80917907e-01 7.40125537e-01
-8.45498502e-01 9.73394811e-01 -1.65252283e-01 4.77137387e-01
1.15619993e+00 -1.30926239e+00 3.73800546e-01 8.30844879e-01
1.38458937e-01 5.84071577e-01 9.14358854e-01 -5.68661273e-01
-1.12692118e+00 -1.25826120e+00 1.57530725e+00 -7.82773077e-01
4.75544453e-01 -7.40143895e-01 -5.08629978e-01 9.12970901e-01
-3.76439877e-02 -5.39897941e-02 8.89752805e-01 5.18382750e-02
-5.21205664e-01 -1.06199257e-01 -9.21688616e-01 3.94458205e-01
1.12947059e+00 -7.23600268e-01 -3.81907910e-01 3.62948090e-01
4.23035145e-01 -3.89684081e-01 -9.49981570e-01 8.50192159e-02
8.21016490e-01 -1.01055944e+00 1.20124781e+00 -4.88890529e-01
9.19369817e-01 -3.19567621e-02 -1.36233270e-01 -8.33611846e-01
-3.24317306e-01 -6.02012753e-01 -2.57376600e-02 2.16154861e+00
3.56242508e-01 3.19155008e-02 6.60766125e-01 8.12334180e-01
-4.63312864e-03 -3.11048359e-01 -7.09019899e-01 -7.33386397e-01
-3.11965644e-01 -1.24501370e-01 5.24952888e-01 8.75951171e-01
1.43050879e-01 5.46781600e-01 -9.91400838e-01 2.32568100e-01
7.07608044e-01 6.01427481e-02 7.71583259e-01 -1.16419721e+00
-2.28525728e-01 -2.19744131e-01 -6.02106810e-01 -8.48394454e-01
6.97320580e-01 -8.60361040e-01 -2.35107895e-02 -1.54292810e+00
9.42017496e-01 -6.21317662e-02 1.12981968e-01 3.55760366e-01
-3.67680132e-01 7.50614882e-01 3.69526684e-01 3.74563366e-01
-1.40239429e+00 4.13021475e-01 4.04749066e-01 -4.67953794e-02
6.41642585e-02 4.79590222e-02 -7.05837667e-01 5.87793410e-01
2.17037112e-01 -3.43675405e-01 -1.32855311e-01 -2.86931753e-01
2.86858261e-01 6.75937384e-02 5.16652346e-01 -1.09880340e+00
3.68457288e-01 1.74674481e-01 4.03118432e-01 -4.38325822e-01
6.67488098e-01 -5.42189777e-01 3.57918859e-01 -4.27791523e-03
-5.51265538e-01 1.19221658e-01 -2.06759691e-01 3.93120259e-01
-3.84533495e-01 -1.83719337e-01 4.36709881e-01 -2.69736379e-01
-9.26828861e-01 3.81603479e-01 -1.52901113e-01 1.26301005e-01
1.06295884e+00 -3.33394736e-01 -2.02697143e-02 -6.84656084e-01
-9.74389613e-01 4.31852698e-01 7.44954944e-01 5.91858387e-01
3.28835070e-01 -1.67783964e+00 -1.06744933e+00 -1.93102285e-01
4.34987754e-01 -4.82532740e-01 5.39508104e-01 6.98579848e-01
-2.89751887e-01 3.31266254e-01 -1.59549803e-01 -4.83408839e-01
-1.64312124e+00 5.51383078e-01 1.31208390e-01 -2.72281498e-01
-3.61668199e-01 1.03796601e+00 9.36140269e-02 -2.86796927e-01
3.29236448e-01 3.12815130e-01 -7.04633236e-01 5.34415245e-01
8.16820085e-01 2.13995546e-01 -2.61408150e-01 -1.42999196e+00
-5.82903504e-01 7.32330263e-01 -1.05244838e-01 -2.32325256e-01
1.44014001e+00 -6.19327605e-01 -7.95482248e-02 5.73514104e-01
1.44512975e+00 -2.27278490e-02 -1.12989807e+00 -2.06204429e-01
2.53070961e-03 -4.05842662e-01 -6.08841360e-01 -4.63801026e-01
-7.85704076e-01 4.04864281e-01 7.90123269e-02 2.20091715e-02
9.87569749e-01 2.55692005e-01 1.00697434e+00 -4.79905009e-02
2.32987225e-01 -1.06182039e+00 1.35114849e-01 4.01699066e-01
7.71112561e-01 -1.35791028e+00 -4.56564538e-02 -5.56183338e-01
-9.01129782e-01 1.02330935e+00 6.49309456e-01 2.21153125e-01
4.11019623e-02 -5.67404442e-02 -2.08295509e-01 -3.72648309e-03
-7.66648650e-01 -4.16592389e-01 4.38534349e-01 6.77930534e-01
4.02058274e-01 -3.17319185e-01 -1.97893664e-01 1.17435908e+00
-1.77051604e-01 -1.06891662e-01 4.92279828e-01 4.84806508e-01
-2.19590310e-02 -1.22177351e+00 -2.87452757e-01 6.48868009e-02
-6.83107972e-01 6.60724267e-02 -5.87028503e-01 4.90654647e-01
4.77325797e-01 7.75868893e-01 3.52477372e-01 -5.17676950e-01
3.98991823e-01 2.02895716e-01 4.46206540e-01 -6.33040369e-01
-7.39690661e-01 -1.55985296e-01 4.29168612e-01 -4.91488039e-01
-7.82104492e-01 -1.18157482e+00 -7.83743322e-01 -3.08290899e-01
1.57834411e-01 1.85725808e-01 3.57129872e-01 9.66225922e-01
3.47259641e-01 1.74104884e-01 6.90134823e-01 -1.10853851e+00
-1.46296501e-01 -6.98332310e-01 -6.27761364e-01 1.06518829e+00
1.75598875e-01 -5.30848324e-01 -2.97754437e-01 6.58057094e-01] | [14.505143165588379, 0.9507742524147034] |
d3685ea8-780b-4fca-a727-8500c1e35c4c | rethinking-and-designing-a-high-performing | 2011.14936 | null | https://arxiv.org/abs/2011.14936v2 | https://arxiv.org/pdf/2011.14936v2.pdf | Rethinking and Designing a High-performing Automatic License Plate Recognition Approach | In this paper, we propose a real-time and accurate automatic license plate recognition (ALPR) approach. Our study illustrates the outstanding design of ALPR with four insights: (1) the resampling-based cascaded framework is beneficial to both speed and accuracy; (2) the highly efficient license plate recognition should abundant additional character segmentation and recurrent neural network (RNN), but adopt a plain convolutional neural network (CNN); (3) in the case of CNN, taking advantage of vertex information on license plates improves the recognition performance; and (4) the weight-sharing character classifier addresses the lack of training images in small-scale datasets. Based on these insights, we propose a novel ALPR approach, termed VSNet. Specifically, VSNet includes two CNNs, i.e., VertexNet for license plate detection and SCR-Net for license plate recognition, integrated in a resampling-based cascaded manner. In VertexNet, we propose an efficient integration block to extract the spatial features of license plates. With vertex supervisory information, we propose a vertex-estimation branch in VertexNet such that license plates can be rectified as the input images of SCR-Net. In SCR-Net, we introduce a horizontal encoding technique for left-to-right feature extraction and propose a weight-sharing classifier for character recognition. Experimental results show that the proposed VSNet outperforms state-of-the-art methods by more than 50% relative improvement on error rate, achieving > 99% recognition accuracy on CCPD and AOLP datasets with 149 FPS inference speed. Moreover, our method illustrates an outstanding generalization capability when evaluated on the unseen PKUData and CLPD datasets. | ['Lap-Pui Chau', 'Yunhao Zhou', 'Zhen-Peng Bian', 'Yi Wang'] | 2020-11-30 | null | null | null | null | ['license-plate-recognition', 'license-plate-detection'] | ['computer-vision', 'computer-vision'] | [ 1.54294744e-01 -6.09792113e-01 -8.53484794e-02 -2.61337936e-01
-7.56385028e-01 -5.78561246e-01 1.26543447e-01 -5.08474886e-01
-3.44473839e-01 3.74699980e-01 -4.07507181e-01 -4.09803033e-01
1.15162194e-01 -1.01188457e+00 -9.65662539e-01 -5.62826991e-01
5.27650833e-01 1.34617344e-01 6.17266655e-01 -3.33475888e-01
5.58813810e-01 9.98326361e-01 -1.41031957e+00 2.67477304e-01
8.54566097e-01 1.29613352e+00 2.12865785e-01 6.08493745e-01
-1.72377229e-01 7.69950509e-01 -6.10526323e-01 -6.88191175e-01
5.11020660e-01 1.03705272e-01 -1.76509172e-01 3.42314482e-01
3.30345452e-01 -4.78378832e-01 -7.14933991e-01 8.99251759e-01
4.51771677e-01 -2.93624494e-03 4.34963435e-01 -9.32162702e-01
-7.86576867e-01 2.08952174e-01 -6.45498276e-01 3.89092453e-02
-3.10773663e-02 2.72600502e-01 8.06516409e-01 -1.23713481e+00
4.12742406e-01 7.30149806e-01 1.01716328e+00 3.86164457e-01
-6.76134884e-01 -6.61447346e-01 1.47120684e-01 1.27937451e-01
-1.82200098e+00 -2.70377725e-01 7.73460925e-01 -1.92015395e-01
1.06913090e+00 3.30538839e-01 5.84288836e-01 7.68144965e-01
-7.17731044e-02 1.16488063e+00 6.98894203e-01 -2.36889228e-01
2.81996708e-02 1.69027727e-02 1.14817806e-01 8.07984710e-01
3.25461894e-01 -2.75401711e-01 -1.45714387e-01 4.93897378e-01
1.35655916e+00 4.48617309e-01 -6.24964647e-02 1.04917601e-01
-8.83618951e-01 3.96895885e-01 4.11258489e-01 1.81770474e-01
-2.09828317e-01 9.31597352e-02 4.29988950e-01 -6.13605343e-02
3.78288388e-01 2.10778818e-01 -1.01860590e-01 -1.84606388e-01
-9.95379984e-01 -2.79214680e-02 4.39353108e-01 1.15355539e+00
7.17879951e-01 5.53299725e-01 -2.14811161e-01 1.23456252e+00
2.71577835e-01 7.87383497e-01 5.08770585e-01 -3.24314266e-01
8.93567860e-01 9.48968709e-01 -1.18108489e-01 -1.10432076e+00
-2.39447743e-01 -6.24758124e-01 -8.65886807e-01 1.98161051e-01
2.25388259e-01 -9.09797102e-03 -1.05563080e+00 8.62402022e-01
-3.41925889e-01 4.03206527e-01 1.26495570e-01 8.76782537e-01
7.45985091e-01 9.14947689e-01 -3.97188336e-01 3.67807269e-01
1.42025149e+00 -1.17695904e+00 -4.90177780e-01 -3.02142829e-01
4.89785939e-01 -6.02739334e-01 8.39147389e-01 2.31785849e-01
-8.70693266e-01 -7.13469207e-01 -1.32771564e+00 -7.23967329e-02
-4.70787287e-01 1.07959569e+00 3.03674042e-01 7.63306201e-01
-9.18746531e-01 2.80493587e-01 -5.69300473e-01 4.70291004e-02
5.97880483e-01 6.00656569e-01 -2.74850100e-01 -2.08389148e-01
-8.91691089e-01 6.02304995e-01 1.07551098e-01 6.53603196e-01
-4.34296817e-01 -3.57636243e-01 -7.66533673e-01 2.84144700e-01
3.56889546e-01 -9.09510255e-02 8.61927867e-01 -1.01555324e+00
-1.83992887e+00 6.20019197e-01 -5.92069142e-02 -3.40183705e-01
6.87616885e-01 -2.93941889e-02 -7.73688018e-01 2.35112220e-01
-1.59033582e-01 4.69181597e-01 8.47741306e-01 -8.78029883e-01
-8.04893196e-01 -2.95899630e-01 -1.97523817e-01 2.10068643e-01
-3.20883602e-01 -7.86023960e-02 -1.10132754e+00 -7.83521712e-01
2.74786621e-01 -8.96285594e-01 -1.10988738e-02 -1.12014301e-01
-5.55508733e-01 -1.10324189e-01 9.87529814e-01 -6.63113058e-01
1.08302486e+00 -2.44238019e+00 -6.19512737e-01 4.11637366e-01
2.41765454e-01 9.48767304e-01 -7.33446479e-02 1.71700507e-01
2.94784456e-02 8.20733234e-02 -8.14985186e-02 -3.16339821e-01
-8.06299374e-02 -1.20227560e-01 -5.94207466e-01 5.36150694e-01
5.89099824e-01 1.08930266e+00 -3.00269336e-01 -2.34789908e-01
5.39358437e-01 4.49716687e-01 -2.65712976e-01 -2.74953060e-02
3.09126824e-01 -2.74281025e-01 -4.41221714e-01 1.04985964e+00
1.06280160e+00 -2.51034349e-01 -1.18516460e-01 -2.05449432e-01
-4.26363319e-01 -1.13002732e-01 -1.18808174e+00 9.98439848e-01
-1.98479593e-01 9.58418489e-01 -8.00267905e-02 -8.54314089e-01
1.63201714e+00 9.81047601e-02 -1.35733038e-01 -8.31853509e-01
8.87062103e-02 5.77183425e-01 -3.83855432e-01 -2.23869830e-01
8.42570126e-01 2.89464891e-01 -4.84512933e-02 -9.43666026e-02
-1.17109761e-01 2.85825402e-01 2.20457595e-02 -4.59390357e-02
7.97852814e-01 -1.00458443e-01 -1.28173739e-01 9.48805287e-02
7.99711525e-01 -7.57324919e-02 6.01707160e-01 7.84657121e-01
-1.87851131e-01 7.80293167e-01 4.02083874e-01 -6.06685877e-01
-1.03758764e+00 -7.84903646e-01 -1.83565319e-01 6.68077230e-01
3.62078577e-01 7.60583952e-02 -6.28330231e-01 -3.73997003e-01
-4.80728596e-02 2.65998185e-01 -2.69595653e-01 1.20705694e-01
-9.21359479e-01 -6.53895199e-01 1.11363590e+00 1.09262455e+00
1.20851827e+00 -6.68217361e-01 -2.16409341e-02 2.02322617e-01
2.27062300e-01 -1.40519083e+00 -5.17389297e-01 -1.56390548e-01
-9.49626088e-01 -7.76612103e-01 -9.64301646e-01 -1.15570998e+00
7.16585517e-01 5.11511028e-01 4.70937759e-01 7.10080564e-02
-7.43753538e-02 -4.51203883e-02 -2.21397027e-01 -1.07645094e-01
-3.36940028e-03 1.25047505e-01 -1.20367415e-01 5.05512714e-01
4.30626422e-01 -1.35808095e-01 -5.69156110e-01 6.81691229e-01
-7.35680044e-01 8.34410414e-02 1.03882861e+00 5.68542600e-01
5.91874182e-01 -2.35683739e-01 6.31195426e-01 -6.55698538e-01
5.72786033e-01 1.04041742e-02 -1.09002292e+00 4.08335686e-01
-4.46151078e-01 -4.56543952e-01 9.11090076e-01 -2.11322308e-01
-9.11638916e-01 3.25359374e-01 -5.82183838e-01 -7.27210283e-01
-3.97146828e-02 1.52248591e-01 -1.78285047e-01 -4.50649917e-01
1.93131119e-01 8.86377335e-01 2.36869920e-02 -2.91817814e-01
-1.69334710e-02 1.11416674e+00 7.54358947e-01 -2.22409710e-01
8.13964903e-01 4.10827279e-01 -1.59884602e-01 -9.70677674e-01
-4.25022319e-02 -6.33047879e-01 -5.84376395e-01 -5.03546178e-01
7.17639625e-01 -1.12647128e+00 -1.12462974e+00 1.11492920e+00
-1.05788743e+00 4.09972556e-02 3.27642150e-02 3.52386862e-01
-2.05342263e-01 3.97667706e-01 -9.89659905e-01 -9.53370214e-01
-3.47857982e-01 -1.41921771e+00 1.07950616e+00 4.90950078e-01
5.23616195e-01 -6.55534208e-01 -4.39830452e-01 4.74917144e-01
5.08531034e-01 -1.28902733e-01 3.74262333e-01 -1.02808237e+00
-1.09025216e+00 -8.19848657e-01 -8.07652771e-01 5.49666524e-01
-1.36596024e-01 2.60964036e-01 -1.04228544e+00 1.08779967e-02
-4.13010746e-01 -1.18790492e-01 1.00503731e+00 3.33967388e-01
1.08353174e+00 -8.26218128e-02 -2.64135480e-01 7.77969778e-01
1.59058833e+00 4.24353540e-01 1.10294366e+00 4.12349433e-01
9.55973566e-01 6.98844269e-02 3.54174793e-01 3.30127358e-01
2.77014047e-01 6.18413150e-01 3.24285924e-01 -2.51924843e-01
-6.86021661e-03 -3.76478374e-01 4.38721865e-01 9.56281543e-01
-3.38232398e-01 -3.58068228e-01 -8.89747560e-01 3.15270662e-01
-1.66416323e+00 -8.68499339e-01 -2.57799059e-01 2.05007362e+00
2.49106407e-01 2.08268523e-01 7.42059480e-03 1.04557559e-01
1.23549414e+00 2.07016259e-01 -5.55652201e-01 -3.09238434e-01
-4.15489346e-01 -2.35376179e-01 9.30366874e-01 1.72810450e-01
-1.17790508e+00 1.10532391e+00 5.70658875e+00 1.18767548e+00
-1.43327475e+00 -3.55419546e-01 8.47138345e-01 3.22988540e-01
1.00336373e-01 -5.33227146e-01 -1.26809359e+00 5.12784958e-01
6.28480673e-01 3.08803022e-01 2.83921391e-01 9.27755117e-01
-1.41766295e-01 4.40391511e-01 -7.04919040e-01 1.14022410e+00
3.01014215e-01 -1.76225114e+00 1.53682202e-01 -3.01467236e-02
3.63927186e-01 1.24266975e-01 2.74793029e-01 4.85035837e-01
-3.88357826e-02 -8.73723090e-01 7.89127111e-01 4.04367864e-01
1.03876662e+00 -1.01808465e+00 9.17158902e-01 2.71503538e-01
-1.44563401e+00 -2.08709776e-01 -5.85939348e-01 1.96736708e-01
-5.45468889e-02 3.30346435e-01 -7.95495033e-01 5.17292142e-01
2.71576494e-01 9.42669511e-01 -6.38907492e-01 1.10046697e+00
-1.76000208e-01 6.28228664e-01 -9.76283476e-02 -3.22979718e-01
6.48258150e-01 -4.22195822e-01 4.01778072e-01 1.45979452e+00
3.83124977e-01 9.82485637e-02 -2.63244599e-01 9.48281884e-01
-4.38193083e-01 -8.44968259e-02 -3.81392688e-01 2.55540032e-02
5.59038103e-01 1.28388572e+00 -9.79923189e-01 -2.71001875e-01
-5.87559223e-01 1.04847085e+00 1.32235453e-01 4.13815409e-01
-1.02913654e+00 -8.77970099e-01 4.00494009e-01 -1.83957238e-02
9.19255733e-01 -7.92893618e-02 -3.26334774e-01 -1.25120330e+00
2.93956995e-01 -5.98859668e-01 1.86893772e-02 -7.47718990e-01
-1.08220005e+00 6.65244579e-01 -7.39890397e-01 -1.72705793e+00
1.73703626e-01 -1.16087520e+00 -6.10840380e-01 8.08240175e-01
-1.76803100e+00 -1.16441453e+00 -1.36211246e-01 5.81689119e-01
7.55129099e-01 -5.57093382e-01 3.74766529e-01 5.19174457e-01
-1.15740204e+00 8.12670290e-01 5.03735363e-01 1.08125055e+00
3.17068428e-01 -7.84526348e-01 8.00967753e-01 1.04649878e+00
-8.05448070e-02 4.27536756e-01 -2.01792195e-01 -6.19340062e-01
-1.73971510e+00 -1.59517872e+00 4.96908486e-01 -5.40323853e-02
4.07377154e-01 -6.27924085e-01 -9.79156733e-01 8.16942573e-01
-3.71936917e-01 2.91886598e-01 5.18257141e-01 -3.55477840e-01
-4.96944427e-01 -4.43714410e-01 -1.08092618e+00 6.54459000e-01
7.05353737e-01 -5.20506144e-01 -3.03625703e-01 1.02128677e-01
5.68738759e-01 -5.86015046e-01 -5.39635599e-01 1.38694972e-01
7.10325539e-01 -7.05183327e-01 1.07651687e+00 2.18817890e-02
5.48400521e-01 -5.41017234e-01 -1.90875456e-01 -7.92454660e-01
-3.59544545e-01 -1.24189727e-01 2.64749289e-01 1.26322567e+00
6.47465229e-01 -9.89325404e-01 7.78865516e-01 6.15887940e-01
-4.70626712e-01 -8.93646479e-01 -1.02598405e+00 -9.22558725e-01
-2.28958264e-01 -5.52001715e-01 6.26624763e-01 6.79861069e-01
-4.25952047e-01 -3.22273709e-02 -4.63409245e-01 4.37486023e-01
2.54646420e-01 1.52868658e-01 4.83847469e-01 -9.17847991e-01
-1.27540454e-01 -5.22043765e-01 -8.78650546e-01 -1.49353886e+00
-7.45155215e-02 -6.09634936e-01 -1.08174440e-02 -1.41129935e+00
-8.59967694e-02 -3.13559622e-01 -2.92472959e-01 3.87709647e-01
9.29091349e-02 6.70839787e-01 4.50108618e-01 5.64867139e-01
-8.29076409e-01 2.77735174e-01 1.15059114e+00 -3.33034188e-01
-1.73999935e-01 1.86255828e-01 -4.48033094e-01 6.32031918e-01
5.61829746e-01 -9.65361074e-02 5.17420880e-02 -6.00808263e-01
1.35160476e-01 8.25949162e-02 3.26665670e-01 -1.11858857e+00
7.05641389e-01 3.91903400e-01 8.04638326e-01 -8.04999411e-01
4.52350587e-01 -6.75903678e-01 -2.75961995e-01 3.97972047e-01
3.02115735e-02 7.45669082e-02 4.92757052e-01 6.03809357e-01
-3.87189537e-01 -2.26160139e-01 5.88045180e-01 1.40346810e-01
-1.13940942e+00 2.50599235e-01 -6.82984114e-01 -3.32113117e-01
1.01399338e+00 -8.76152039e-01 -5.31634450e-01 -8.57510418e-02
-1.16768330e-01 -5.39774001e-02 3.52458447e-01 3.67631614e-01
9.77203786e-01 -1.27866161e+00 -7.05515265e-01 5.05551457e-01
1.24087594e-01 -8.30157548e-02 4.41653162e-01 5.59239864e-01
-1.08497906e+00 6.94474936e-01 -1.75670609e-01 -6.59445941e-01
-1.03201163e+00 1.14374466e-01 3.82834911e-01 -1.57273710e-01
-7.11404681e-01 8.68709385e-01 -6.25225157e-02 -5.16359806e-01
-2.56583095e-02 -2.68132389e-01 -2.93425679e-01 -8.00177827e-02
6.26065373e-01 5.12258768e-01 4.52118278e-01 -6.77266538e-01
-4.62897599e-01 8.58248770e-01 -2.97797799e-01 1.93717808e-01
1.43114901e+00 2.30644464e-01 8.41492042e-02 1.20737359e-01
1.07580054e+00 1.11214764e-01 -1.53687203e+00 -9.66890976e-02
-2.77282894e-01 -6.03733003e-01 -1.62850078e-02 -6.60771787e-01
-1.39347923e+00 8.70936513e-01 4.62349325e-01 -1.59861580e-01
1.08645189e+00 -4.66590673e-01 8.45434964e-01 6.31021440e-01
2.61567086e-01 -1.27778947e+00 -2.08951369e-01 6.28612638e-01
6.39384747e-01 -9.52655554e-01 -1.85166374e-01 -4.64027613e-01
-7.87347615e-01 1.47707653e+00 6.38799012e-01 -4.70360041e-01
1.21575579e-01 4.13938105e-01 -1.47811591e-03 6.92547439e-03
-3.85311306e-01 1.21575423e-01 2.22837448e-01 2.70550519e-01
6.97659329e-02 8.42668489e-02 2.68668175e-01 8.33194673e-01
2.27378801e-01 -2.31390446e-02 5.85229516e-01 8.25221658e-01
-4.25993323e-01 -6.13040507e-01 -4.58983451e-01 4.81218606e-01
-1.61289945e-01 -2.16698781e-01 -3.12019765e-01 9.96658802e-01
-1.53459325e-01 6.13499820e-01 4.70567256e-01 -6.38070166e-01
5.28286934e-01 -5.60455993e-02 -3.27772349e-02 -2.18846887e-01
-6.12572432e-01 -1.99305192e-02 -1.53946713e-01 -3.10091287e-01
6.35435358e-02 -4.84912127e-01 -1.11751878e+00 -4.01138604e-01
-5.88551342e-01 -2.10267454e-01 8.06740999e-01 9.33065593e-01
6.81843817e-01 6.68900609e-01 7.84221411e-01 -7.78688550e-01
-4.81539220e-01 -6.67759538e-01 -8.27382267e-01 -3.64920907e-02
1.39508560e-01 -2.49144569e-01 -1.90513298e-01 -1.43825158e-01] | [9.849594116210938, -4.92580509185791] |
261151b1-7d76-4499-8b16-035b4000f9c3 | images-speak-in-images-a-generalist-painter | 2212.02499 | null | https://arxiv.org/abs/2212.02499v2 | https://arxiv.org/pdf/2212.02499v2.pdf | Images Speak in Images: A Generalist Painter for In-Context Visual Learning | In-context learning, as a new paradigm in NLP, allows the model to rapidly adapt to various tasks with only a handful of prompts and examples. But in computer vision, the difficulties for in-context learning lie in that tasks vary significantly in the output representations, thus it is unclear how to define the general-purpose task prompts that the vision model can understand and transfer to out-of-domain tasks. In this work, we present Painter, a generalist model which addresses these obstacles with an "image"-centric solution, that is, to redefine the output of core vision tasks as images, and specify task prompts as also images. With this idea, our training process is extremely simple, which performs standard masked image modeling on the stitch of input and output image pairs. This makes the model capable of performing tasks conditioned on visible image patches. Thus, during inference, we can adopt a pair of input and output images from the same task as the input condition, to indicate which task to perform. Without bells and whistles, our generalist Painter can achieve competitive performance compared to well-established task-specific models, on seven representative vision tasks ranging from high-level visual understanding to low-level image processing. In addition, Painter significantly outperforms recent generalist models on several challenging tasks. | ['Tiejun Huang', 'Chunhua Shen', 'Yue Cao', 'Wen Wang', 'Xinlong Wang'] | 2022-12-05 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Images_Speak_in_Images_A_Generalist_Painter_for_In-Context_Visual_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Images_Speak_in_Images_A_Generalist_Painter_for_In-Context_Visual_CVPR_2023_paper.pdf | cvpr-2023-1 | ['keypoint-detection', 'personalized-segmentation'] | ['computer-vision', 'computer-vision'] | [ 6.39018655e-01 -7.35483319e-02 4.33182865e-02 -4.21201676e-01
-4.25124943e-01 -6.42992258e-01 1.01311553e+00 -2.08245099e-01
-5.37632048e-01 4.63550597e-01 -1.90870315e-01 -4.06726092e-01
3.77893671e-02 -5.00934422e-01 -9.57520962e-01 -7.90034294e-01
5.16003907e-01 3.14268202e-01 1.95257545e-01 -5.98274171e-02
4.70289260e-01 3.26217711e-01 -1.81139159e+00 8.34556401e-01
5.55066645e-01 8.18965614e-01 1.01817143e+00 6.62622988e-01
-2.17742383e-01 8.15838277e-01 -7.76303589e-01 -1.81066290e-01
1.26866579e-01 -3.13530177e-01 -8.65868390e-01 2.85204947e-01
9.20389831e-01 -2.56070346e-01 1.34743154e-02 9.76971388e-01
2.40082949e-01 1.04609698e-01 7.87746370e-01 -1.08784294e+00
-1.19676936e+00 3.10121447e-01 -5.33592224e-01 2.44951084e-01
3.86139035e-01 4.80997950e-01 1.03496051e+00 -1.26093721e+00
6.74029708e-01 1.20028448e+00 2.08953336e-01 7.69255221e-01
-1.66433024e+00 -3.48374128e-01 6.24618530e-01 3.39065790e-01
-1.13853633e+00 -2.66843170e-01 7.55736470e-01 -6.25432014e-01
1.03334105e+00 2.93053836e-01 7.32339144e-01 1.41007674e+00
1.90304354e-01 9.53884184e-01 1.62086725e+00 -5.28675437e-01
2.30889171e-01 2.48682514e-01 -3.56550962e-02 4.62232560e-01
3.55391181e-03 4.83086079e-01 -5.80643833e-01 2.42759421e-01
8.57745647e-01 6.52011111e-02 -5.14769614e-01 -2.84202099e-01
-1.34012508e+00 6.18766546e-01 7.46808052e-01 2.99835235e-01
-3.78260761e-01 1.28024444e-01 1.38107850e-03 3.27267081e-01
3.00170839e-01 5.84439397e-01 -5.20756960e-01 4.62065220e-01
-8.67882550e-01 1.46146595e-01 5.79174519e-01 8.72831166e-01
1.04392052e+00 1.18500497e-02 -3.85356933e-01 7.81345963e-01
9.71834660e-02 4.05080497e-01 3.29324484e-01 -1.03919148e+00
3.11957657e-01 3.23621660e-01 4.22643051e-02 -7.22797215e-01
-2.79436260e-01 -4.76665556e-01 -6.17452800e-01 6.93785131e-01
6.09721959e-01 7.77211487e-02 -1.32250464e+00 1.84578395e+00
3.64535078e-02 2.21804142e-01 7.56690279e-02 1.10683298e+00
7.12369859e-01 8.70918512e-01 3.48025352e-01 -1.51625395e-01
1.65165269e+00 -1.08100891e+00 -4.90857929e-01 -9.94054258e-01
2.13276878e-01 -7.69466460e-01 1.71414781e+00 6.54778659e-01
-8.10745895e-01 -1.00832570e+00 -9.75136340e-01 -3.26384872e-01
-6.85880661e-01 1.87020347e-01 6.01414740e-01 1.92702502e-01
-1.20450997e+00 3.11818361e-01 -2.89397210e-01 -4.27966863e-01
3.35137367e-01 1.31219804e-01 -3.59011561e-01 -3.18996787e-01
-7.85858154e-01 1.31904125e+00 5.44456601e-01 8.25361162e-02
-1.16929543e+00 -7.28377879e-01 -8.69873047e-01 7.03926235e-02
5.49303710e-01 -1.00876975e+00 1.42995179e+00 -1.47219801e+00
-1.19111037e+00 1.30586052e+00 -2.81281382e-01 -2.62855321e-01
1.48063943e-01 -3.22611243e-01 -1.89212024e-01 8.22257921e-02
6.98340759e-02 1.14435446e+00 1.34205461e+00 -1.84813190e+00
-5.94263136e-01 -2.00128272e-01 2.20642343e-01 3.98948252e-01
1.40780568e-01 -1.43913135e-01 -5.91665506e-01 -5.94640553e-01
-8.92544538e-02 -7.70601571e-01 -1.08198635e-01 2.45162681e-01
-2.67029762e-01 -1.61536589e-01 8.95171702e-01 -2.42768690e-01
5.35569787e-01 -2.36507916e+00 1.63633257e-01 -1.77867442e-01
7.77205601e-02 1.87875912e-01 -3.75073940e-01 2.65542299e-01
-1.53668389e-01 -6.70447424e-02 -4.08701599e-01 -3.94255638e-01
-9.25299749e-02 6.44272506e-01 -5.54081559e-01 1.13281459e-01
5.19208968e-01 1.11446965e+00 -1.02519715e+00 -4.01680052e-01
4.34022427e-01 3.87816846e-01 -2.77263373e-01 4.74873096e-01
-5.76054335e-01 6.98540926e-01 -1.04209915e-01 4.38911557e-01
4.30313706e-01 -4.31881070e-01 2.00275052e-03 -1.52543545e-01
-8.40079337e-02 1.89125910e-01 -7.84690738e-01 1.96464586e+00
-7.04502702e-01 8.88908029e-01 2.33696476e-01 -1.02512109e+00
6.23221815e-01 2.31492937e-01 -1.60510108e-01 -8.51225138e-01
-5.90698421e-02 -2.24645883e-02 1.39534831e-01 -7.21150696e-01
2.73734212e-01 -1.65561393e-01 8.25392231e-02 4.88560230e-01
2.30282545e-01 -4.71580386e-01 2.60497361e-01 6.82585835e-02
6.30139470e-01 4.18362886e-01 3.80429476e-01 -1.65069044e-01
4.52619314e-01 7.03023449e-02 1.71258956e-01 1.05729318e+00
-5.65457866e-02 8.62951279e-01 3.91281515e-01 -5.68335235e-01
-7.95378506e-01 -1.22156584e+00 -6.32963330e-02 1.61938524e+00
2.30002720e-02 -3.69764008e-02 -7.20143497e-01 -6.34602785e-01
-6.04262911e-02 8.88075590e-01 -8.80220950e-01 4.92423251e-02
-4.44339961e-01 -4.28209543e-01 -2.00198051e-02 4.47581857e-01
3.78654420e-01 -1.67661667e+00 -7.33580410e-01 -5.97490855e-02
6.54567480e-02 -1.16163635e+00 -3.65526229e-01 4.38423872e-01
-7.09427118e-01 -1.08799481e+00 -7.49553680e-01 -1.15626395e+00
9.76249814e-01 5.12778044e-01 1.40680408e+00 1.72864497e-01
-4.26921099e-01 5.04805267e-01 -1.59630522e-01 -4.20192003e-01
-2.29774550e-01 -3.62062484e-01 -4.04012531e-01 8.23037401e-02
3.32090944e-01 -4.63504881e-01 -6.62351191e-01 9.85253230e-02
-1.09085786e+00 3.58355880e-01 1.02622235e+00 1.00586426e+00
6.66258812e-01 -4.93656874e-01 3.82566839e-01 -1.25348711e+00
6.61186159e-01 -3.33292753e-01 -5.09002030e-01 4.29563820e-01
-6.16125703e-01 4.81543690e-02 6.89827979e-01 -6.06875420e-01
-1.14664614e+00 3.43071282e-01 1.54947955e-02 -4.25539672e-01
-6.05681479e-01 4.72537220e-01 -2.42419243e-02 7.54374489e-02
8.84612143e-01 5.28103352e-01 -1.10167116e-01 -3.22351068e-01
8.73428702e-01 3.81953448e-01 8.67136002e-01 -7.59097934e-01
6.79143369e-01 3.63614947e-01 -2.46604726e-01 -7.20930457e-01
-1.04694140e+00 -2.93551117e-01 -6.24923944e-01 -1.79013282e-01
9.42299128e-01 -8.58645260e-01 -5.87770939e-01 3.21095854e-01
-1.35168803e+00 -6.17511272e-01 -3.26675415e-01 2.37850934e-01
-6.91738665e-01 1.19477794e-01 -3.86176109e-01 -7.33762860e-01
5.94768524e-02 -1.21825004e+00 1.11495447e+00 3.11073810e-01
-1.28606290e-01 -9.56303895e-01 -2.89213657e-01 2.01872274e-01
3.54698569e-01 -8.25661570e-02 1.10376048e+00 -3.20970237e-01
-7.98393428e-01 1.93924606e-01 -5.58844388e-01 4.99694556e-01
4.97671403e-02 -2.02705979e-01 -1.47635460e+00 -1.66547060e-01
3.37531596e-01 -5.05909026e-01 1.11569750e+00 3.16856176e-01
1.53587413e+00 -2.59162366e-01 -3.12802613e-01 7.16671169e-01
1.48136032e+00 1.39188662e-01 5.68476260e-01 2.31564537e-01
5.23157001e-01 7.96731949e-01 4.15328443e-01 -3.20622623e-02
2.55027175e-01 5.40305972e-01 6.52838767e-01 -2.94587553e-01
-3.61211628e-01 -3.09806257e-01 3.45266998e-01 2.07320228e-01
-5.64892925e-02 -1.10087909e-01 -7.34482706e-01 5.74996173e-01
-1.88823712e+00 -9.79997694e-01 1.15797095e-01 1.97848952e+00
9.52196121e-01 1.12924911e-01 -2.81393528e-01 -2.53679097e-01
5.01225829e-01 3.63698423e-01 -6.55031979e-01 -3.95616502e-01
-1.45587638e-01 3.55494499e-01 -7.96866324e-03 5.71937799e-01
-1.02549767e+00 1.10976207e+00 6.86752605e+00 6.35709882e-01
-1.27702105e+00 1.52903959e-01 5.70330739e-01 7.22895470e-03
-2.38833740e-01 1.34660378e-01 -5.13418317e-01 3.88193309e-01
5.19203842e-01 2.42094129e-01 6.79515123e-01 8.98226202e-01
1.10195890e-01 -4.01176989e-01 -1.66628432e+00 1.08720267e+00
3.36995512e-01 -1.32420266e+00 2.68556505e-01 -2.56228745e-01
5.48059046e-01 -7.63674080e-02 4.92302924e-01 3.53952318e-01
2.08065778e-01 -1.33381534e+00 6.49751782e-01 4.56404686e-01
8.78400266e-01 -4.73404750e-02 2.42997929e-01 5.99575877e-01
-7.29772389e-01 -2.57430702e-01 -3.88030618e-01 -3.06191236e-01
2.56234482e-02 3.46888483e-01 -9.16761875e-01 1.94118395e-01
6.57334983e-01 5.29669821e-01 -6.57930851e-01 8.52608085e-01
-7.52448618e-01 4.03454095e-01 -8.31045657e-02 1.30713284e-01
2.44421646e-01 -1.07380778e-01 1.27839372e-01 1.43214345e+00
1.82592168e-01 4.17857617e-02 2.60927558e-01 1.23853528e+00
1.54394105e-01 -7.24616647e-02 -9.08898950e-01 3.67158324e-01
4.23404187e-01 1.31030929e+00 -4.28309768e-01 -3.69445026e-01
-6.09423876e-01 1.10769498e+00 5.56841314e-01 9.04234171e-01
-4.50361490e-01 -8.57040584e-02 3.91081393e-01 1.02520764e-01
3.49390566e-01 -2.69097924e-01 -4.71777141e-01 -1.05962098e+00
1.44592058e-02 -8.70561600e-01 1.84089914e-01 -1.47869551e+00
-1.31678152e+00 5.44785917e-01 7.22195134e-02 -9.92411077e-01
-3.31426263e-01 -1.18944597e+00 -8.15139413e-01 1.00988305e+00
-1.81510997e+00 -1.38071394e+00 -4.13726807e-01 8.48015130e-01
7.70967662e-01 4.96926159e-02 7.70406544e-01 -5.13840057e-02
-2.17792124e-01 2.47754663e-01 -3.14428478e-01 1.08929470e-01
7.68815637e-01 -1.43702900e+00 2.71210283e-01 8.71275544e-01
7.65847504e-01 7.21146822e-01 7.52066791e-01 -2.05412880e-01
-1.23261297e+00 -8.50015938e-01 7.51568973e-01 -6.37695670e-01
4.84161168e-01 -5.78106999e-01 -1.05585861e+00 8.80633652e-01
6.55582488e-01 2.19929606e-01 3.82113636e-01 1.81735799e-01
-6.46562755e-01 -1.54619351e-01 -7.45622098e-01 7.38890290e-01
1.01687777e+00 -8.36627483e-01 -1.07922328e+00 5.25359750e-01
5.68027675e-01 -3.70684594e-01 -2.61050433e-01 1.61272138e-01
2.92391419e-01 -1.01958692e+00 1.17295480e+00 -5.20397365e-01
4.38134015e-01 -5.05612493e-01 -1.06780343e-01 -1.34058523e+00
-4.22801822e-01 -4.04084027e-01 -2.02102065e-02 8.70608032e-01
5.76011539e-01 -6.07011914e-01 4.40782011e-01 3.46914500e-01
-2.16166601e-01 -6.55387461e-01 -6.88158810e-01 -5.94569981e-01
4.57996465e-02 -4.99349624e-01 1.95237666e-01 6.17655873e-01
-2.79079169e-01 7.48082876e-01 -2.55518138e-01 1.25750467e-01
4.28034246e-01 3.98772031e-01 7.29163766e-01 -1.22885835e+00
-5.65528691e-01 -4.73886907e-01 1.16216689e-01 -1.39706016e+00
2.66155541e-01 -8.91655505e-01 3.13420624e-01 -1.52842367e+00
2.57059067e-01 -3.25203061e-01 -4.36689794e-01 7.38959610e-01
-4.28899884e-01 1.84553847e-01 5.51760852e-01 3.80237520e-01
-3.42140317e-01 1.64414614e-01 1.47805011e+00 -3.10471743e-01
8.47316440e-03 -8.75630155e-02 -9.53438997e-01 7.51072764e-01
6.39067292e-01 -3.50059032e-01 -6.48398221e-01 -7.81624198e-01
2.35005200e-01 -2.09973380e-01 8.09726655e-01 -8.31522524e-01
2.20586702e-01 -3.79295349e-01 6.02102280e-01 -3.23625147e-01
6.37047946e-01 -8.65665734e-01 -2.54010651e-02 3.22203964e-01
-4.28971887e-01 -5.34118935e-02 1.44743323e-01 5.87637186e-01
-2.35491931e-01 -5.22636890e-01 6.58225000e-01 -5.27439594e-01
-1.23105752e+00 7.74005130e-02 -3.51179779e-01 -7.80765563e-02
9.06753778e-01 -3.56697023e-01 -6.76120877e-01 -2.01470912e-01
-9.75250661e-01 -1.70521736e-02 5.88736176e-01 4.68311697e-01
8.14363301e-01 -9.36453223e-01 -5.55965066e-01 2.72424012e-01
3.31893027e-01 2.50920504e-01 2.00614735e-01 6.15136087e-01
-1.33998305e-01 3.08425844e-01 -3.11100006e-01 -9.02765274e-01
-9.94551480e-01 1.01206863e+00 7.79550821e-02 -5.31630665e-02
-6.22591317e-01 9.08555865e-01 1.01580060e+00 -2.03978226e-01
1.77209377e-01 -4.45389926e-01 -2.11937279e-01 -5.71986549e-02
6.39911234e-01 -5.56349337e-01 -1.69944897e-01 -2.77966142e-01
-6.12606928e-02 7.06692219e-01 -2.01713055e-01 -2.48225689e-01
9.28694129e-01 -9.06110257e-02 -7.27297291e-02 6.37007236e-01
9.07472014e-01 -3.65909487e-01 -1.78274775e+00 -3.45769525e-01
-4.09022421e-02 -3.96982610e-01 -1.50981620e-01 -1.23927510e+00
-8.69199753e-01 1.24244833e+00 3.96677315e-01 2.08887011e-01
1.31895578e+00 2.38478944e-01 1.91157103e-01 5.22119522e-01
2.08476439e-01 -1.01726758e+00 3.98038864e-01 6.26459360e-01
1.22704065e+00 -1.47142804e+00 -1.31943956e-01 -2.70344347e-01
-9.08550620e-01 9.53101993e-01 8.21328342e-01 -1.92556188e-01
3.82084340e-01 1.84692323e-01 4.06682283e-01 -2.71854013e-01
-1.09771931e+00 -5.43712139e-01 3.73091877e-01 1.11404026e+00
3.46724182e-01 8.56657140e-03 -5.28744236e-02 3.25787812e-01
1.67398690e-03 -1.08425237e-01 3.17716718e-01 8.26917887e-01
-7.08752513e-01 -1.01525259e+00 -4.77438331e-01 2.19655544e-01
-1.79463610e-01 -3.51206273e-01 -5.39674819e-01 1.00571370e+00
3.59216511e-01 7.92333841e-01 1.61226809e-01 -9.98497456e-02
1.49655014e-01 3.95091891e-01 6.29245579e-01 -1.03271055e+00
-6.19700134e-01 -6.35285452e-02 -1.81031272e-01 -4.26992238e-01
-5.65132380e-01 -3.13872695e-01 -1.08995962e+00 2.79968649e-01
-3.53088509e-03 -1.90836802e-01 6.69120431e-01 1.13413644e+00
2.98410803e-01 4.14707929e-01 2.70783842e-01 -1.15329051e+00
-3.72929960e-01 -9.49405193e-01 -3.48514408e-01 7.55784988e-01
5.07927597e-01 -6.46720529e-01 -2.26853043e-01 4.03635651e-01] | [10.229183197021484, 1.6889493465423584] |
731f55c6-5e9b-46cb-9104-1ec3dd60e0a5 | justdeep-at-nlp4if-2019-task-1-propaganda | null | null | https://aclanthology.org/D19-5016 | https://aclanthology.org/D19-5016.pdf | JUSTDeep at NLP4IF 2019 Task 1: Propaganda Detection using Ensemble Deep Learning Models | The internet and the high use of social media have enabled the modern-day journalism to publish, share and spread news that is difficult to distinguish if it is true or fake. Defining {``}fake news{''} is not well established yet, however, it can be categorized under several labels: false, biased, or framed to mislead the readers that are characterized as propaganda. Digital content production technologies with logical fallacies and emotional language can be used as propaganda techniques to gain more readers or mislead the audience. Recently, several researchers have proposed deep learning (DL) models to address this issue. This research paper provides an ensemble deep learning model using BiLSTM, XGBoost, and BERT to detect propaganda. The proposed model has been applied on the dataset provided by the challenge NLP4IF 2019, Task 1 Sentence Level Classification (SLC) and it shows a significant performance over the baseline model. | ['Hani Al-Omari', 'Samira Shaikh', 'Ola AlTiti', 'Malak Abdullah'] | 2019-11-01 | null | null | null | ws-2019-11 | ['logical-fallacies', 'propaganda-detection'] | ['miscellaneous', 'natural-language-processing'] | [-9.94720832e-02 2.69228876e-01 -4.22496438e-01 -1.06974654e-01
-5.32154620e-01 -4.05043542e-01 1.05499208e+00 4.70526695e-01
-2.55606800e-01 1.00934863e+00 7.11588442e-01 -5.71211994e-01
5.60632348e-01 -8.02424848e-01 -9.23170686e-01 -4.31261778e-01
5.25603235e-01 2.30654955e-01 -8.72398093e-02 -5.73286414e-01
7.64086246e-01 -1.01883434e-01 -9.94300187e-01 8.38140368e-01
8.59831333e-01 8.46846282e-01 -5.02198160e-01 2.19176993e-01
-5.59767425e-01 1.57474554e+00 -1.34957802e+00 -7.39072204e-01
-2.51187235e-01 -5.87294579e-01 -9.65413749e-01 -3.65987629e-01
3.42843175e-01 -4.90308791e-01 -4.49824519e-02 1.37285435e+00
4.57555830e-01 -4.32148010e-01 5.54357588e-01 -1.08386719e+00
-1.19885778e+00 1.10645628e+00 -7.72613943e-01 4.71187621e-01
4.71670628e-01 -2.64564246e-01 6.89516068e-01 -5.78207374e-01
7.54575431e-01 1.72432542e+00 7.55932748e-01 4.25004780e-01
-9.01987135e-01 -7.85780787e-01 -1.73916332e-02 3.88856173e-01
-7.32297301e-01 -1.21622004e-01 1.14027524e+00 -6.69802666e-01
6.60622537e-01 1.90426931e-01 8.48800182e-01 2.37652469e+00
7.52286792e-01 9.78787422e-01 1.80045533e+00 -3.94028664e-01
3.14298570e-02 3.92006159e-01 3.13369483e-01 6.82886362e-01
3.19760531e-01 -5.08911982e-02 -6.81452870e-01 -2.14865103e-01
2.20857817e-03 -2.87695616e-01 -2.55312294e-01 6.29968286e-01
-1.07567143e+00 1.46873045e+00 7.48992622e-01 5.48071980e-01
-3.97992879e-01 2.53697872e-01 8.28605115e-01 6.34110749e-01
1.38800871e+00 7.12363958e-01 -1.83540415e-02 -2.36772120e-01
-8.58929813e-01 5.08692503e-01 9.39854562e-01 2.20781460e-01
-5.74529395e-02 1.06392823e-01 -2.31045276e-01 7.35201359e-01
4.18154478e-01 5.08838773e-01 7.65017033e-01 -2.39681169e-01
4.89337057e-01 3.39192420e-01 3.15299183e-01 -1.95242405e+00
-3.96183550e-01 -6.56589925e-01 -6.59529567e-01 6.85108677e-02
1.17776908e-01 -2.44241267e-01 -7.29367733e-01 1.20279801e+00
1.04520030e-01 1.22676669e-02 -2.11285681e-01 9.87039328e-01
1.21715438e+00 1.21972525e+00 1.15411073e-01 -2.21688733e-01
1.25074708e+00 -1.12062967e+00 -1.39592218e+00 -2.76819050e-01
7.61176527e-01 -1.01059794e+00 7.95678139e-01 6.00627244e-01
-7.53639460e-01 -2.35828012e-01 -1.32835031e+00 -1.24436297e-01
-8.23818803e-01 -2.02197731e-01 2.97667563e-01 6.91904604e-01
-5.53459466e-01 6.28035605e-01 -4.12687570e-01 -5.53454906e-02
8.77575696e-01 -5.67137122e-01 6.04685359e-02 2.75385767e-01
-1.81888294e+00 1.32807398e+00 4.78414237e-01 3.75627689e-02
-7.82859147e-01 -3.77958626e-01 -3.46286863e-01 -4.02627677e-01
9.49354693e-02 -2.16074646e-01 1.14195299e+00 -1.63696790e+00
-1.48042881e+00 1.13583040e+00 5.10903060e-01 -8.97932887e-01
1.09545910e+00 -7.45980203e-01 -8.00498903e-01 -1.40355349e-01
3.37944418e-01 9.20834467e-02 1.25719285e+00 -1.14834881e+00
-3.08609396e-01 -1.19106039e-01 -2.78917141e-02 -7.85496011e-02
-1.36706024e-01 5.13879418e-01 8.67574394e-01 -8.89598429e-01
4.40177247e-02 -5.22421122e-01 6.29400611e-01 -4.03962523e-01
-1.05591035e+00 -3.86826426e-01 1.05257916e+00 -1.17229402e+00
1.16183972e+00 -1.88471329e+00 -2.29926273e-01 -3.42881978e-01
5.04889667e-01 4.46358681e-01 3.82896155e-01 6.13221765e-01
7.36655593e-02 7.26401091e-01 3.44268829e-01 1.16184860e-01
-8.88537541e-02 -2.20728174e-01 -7.64060259e-01 7.31758356e-01
5.56998737e-02 7.59721100e-01 -1.22084010e+00 -3.57751906e-01
-1.94256067e-01 2.48801455e-01 -5.50745837e-02 -1.33937106e-01
-5.00760198e-01 3.42354149e-01 -6.50480986e-01 4.75370139e-01
7.05521286e-01 -3.55450511e-01 -3.44801024e-02 -2.43282199e-01
-3.59432250e-01 8.27337861e-01 -2.09016308e-01 8.40598285e-01
-7.26889372e-02 1.13373387e+00 -2.96790242e-01 -1.12271452e+00
8.97357047e-01 1.72127530e-01 -4.22995612e-02 -7.65014946e-01
7.35505223e-01 3.96059692e-01 -1.45533979e-01 -7.65252411e-01
3.99147004e-01 -3.32065135e-01 -3.52002800e-01 4.23709661e-01
-3.46018374e-01 1.25843272e-01 -4.13728416e-01 1.46974832e-01
9.55853283e-01 -3.77581790e-02 2.89436817e-01 -1.97957933e-01
2.12817326e-01 3.07487398e-01 1.90151140e-01 8.59264255e-01
-3.53303462e-01 8.34032446e-02 6.87573314e-01 -7.55223751e-01
-8.22243631e-01 -9.11005855e-01 -4.96296138e-02 1.04489279e+00
2.45692343e-01 4.98547256e-02 -4.76480186e-01 -9.96599019e-01
5.10923453e-02 1.09067595e+00 -7.55621374e-01 -2.65069276e-01
-3.55135769e-01 -7.61152148e-01 7.84410596e-01 -1.97086379e-01
1.07764840e+00 -1.11798084e+00 -5.42926610e-01 4.78568703e-01
-5.69447339e-01 -8.74685705e-01 2.54243314e-01 -1.96378902e-01
-2.38039017e-01 -9.18427706e-01 -4.63899761e-01 -6.31807446e-01
1.18701383e-01 -4.29488719e-02 9.58509624e-01 7.04178587e-02
3.00958842e-01 -4.88138139e-01 -9.57645059e-01 -8.15754592e-01
-1.01886559e+00 -1.64014339e-01 -1.37733787e-01 3.44467945e-02
3.55270535e-01 3.19976881e-02 -1.94347501e-01 -3.06945711e-01
-6.88418031e-01 5.06119072e-01 3.83538604e-01 1.02785039e+00
-1.68550909e-01 -1.42402351e-01 1.06142628e+00 -1.33645046e+00
1.13912070e+00 -9.73368227e-01 -3.44832502e-02 -1.41941860e-01
-5.13632059e-01 -2.50539064e-01 6.35414422e-01 -4.65000749e-01
-1.00159764e+00 -1.12638354e+00 -2.90791631e-01 1.55284896e-01
5.93667664e-02 9.69211221e-01 3.71178299e-01 1.07463308e-01
9.14619625e-01 -4.62725125e-02 -4.84802462e-02 -4.96654153e-01
3.27196896e-01 1.19915771e+00 2.04795543e-02 -3.16171558e-03
5.24575531e-01 4.96196717e-01 -5.30281842e-01 -5.92820227e-01
-1.66157675e+00 4.98706326e-02 1.88911930e-01 -5.14466405e-01
7.52765000e-01 -7.89968371e-01 -3.77699792e-01 7.84440994e-01
-1.80325508e+00 2.01502204e-01 4.42602247e-01 2.26926699e-01
2.03856677e-02 2.94595808e-01 -9.39494193e-01 -7.01937735e-01
-5.91086507e-01 -6.41835690e-01 4.15802151e-01 -7.61361048e-02
-3.76488566e-01 -8.74326646e-01 -3.85673866e-02 7.47838914e-01
5.56549013e-01 9.65745330e-01 9.64671016e-01 -1.15827513e+00
9.41574648e-02 -4.60079551e-01 -1.98576897e-01 5.91488719e-01
-6.07189583e-03 5.00065722e-02 -7.68812299e-01 7.60476738e-02
2.59941697e-01 -7.59043574e-01 8.96325648e-01 1.65055260e-01
8.42688560e-01 -1.21781194e+00 -2.76825488e-01 -8.48440081e-02
1.12348938e+00 7.03780502e-02 6.60187721e-01 8.27595949e-01
5.28133452e-01 4.26702797e-01 4.91758049e-01 4.50518429e-01
2.23791227e-01 3.29617798e-01 4.96821851e-01 1.71392709e-01
6.14515468e-02 -5.61821520e-01 5.70859134e-01 8.39535594e-01
2.76863724e-01 -6.94626927e-01 -8.80352139e-01 2.11970031e-01
-1.72850573e+00 -1.51470757e+00 -6.06799424e-01 1.38084722e+00
8.62312496e-01 5.31731308e-01 -1.64288253e-01 3.19548398e-01
8.67498815e-01 8.11774015e-01 3.88705693e-02 -1.07049549e+00
-3.86808574e-01 -1.07777216e-01 3.09811532e-01 3.39701116e-01
-1.41591263e+00 1.05261767e+00 5.73032093e+00 8.43348086e-01
-1.63484740e+00 6.68659270e-01 9.74133670e-01 1.11999452e-01
-2.72322804e-01 -3.65610182e-01 -4.56038952e-01 1.09207225e+00
8.47633779e-01 -1.46213189e-01 7.04152137e-02 8.68759811e-01
4.31207836e-01 -6.88751601e-03 -4.36721087e-01 7.06667066e-01
4.84870195e-01 -1.64263344e+00 8.69879574e-02 -1.91993669e-01
9.43893433e-01 3.42657536e-01 2.22716793e-01 5.56980431e-01
2.02736959e-01 -9.60702777e-01 1.24223173e+00 4.00188774e-01
1.00034796e-01 -4.01404470e-01 1.10509646e+00 6.46793306e-01
3.08900118e-01 6.16442747e-02 -1.15987502e-01 -4.71928656e-01
2.85886765e-01 7.91109681e-01 -8.22199881e-01 1.73655793e-01
5.79122543e-01 6.74244761e-01 -4.98396069e-01 6.23458147e-01
-5.39706826e-01 1.24057400e+00 5.76011091e-02 -8.79451394e-01
6.26318455e-01 2.23194748e-01 8.68664861e-01 1.13884842e+00
1.73882499e-01 -4.41849709e-01 5.48152253e-02 1.02602792e+00
-4.24383193e-01 2.73804754e-01 -6.20823383e-01 -6.44497812e-01
2.78264046e-01 9.00494158e-01 -5.22381663e-01 -4.75993156e-01
-2.64427960e-01 9.80080545e-01 3.19102705e-01 1.02012061e-01
-1.29593623e+00 -9.18081105e-02 -1.01960078e-02 3.65061641e-01
-2.73864537e-01 2.10717350e-01 -2.93212116e-01 -1.15782690e+00
-2.49741465e-01 -9.36093152e-01 5.22174500e-02 -7.69125581e-01
-1.71317947e+00 5.35548031e-01 -2.78392017e-01 -8.25687230e-01
1.52738363e-01 -6.24366701e-01 -4.91625637e-01 5.46805024e-01
-1.33927476e+00 -1.16350126e+00 -1.30249262e-01 -1.84239149e-01
4.37169135e-01 -1.35752723e-01 3.98704678e-01 1.18154228e-01
-2.61331230e-01 1.85234264e-01 2.47451842e-01 3.97515863e-01
7.05399334e-01 -9.02980685e-01 2.18826272e-02 5.33993423e-01
-1.53186187e-01 1.62400499e-01 1.29689062e+00 -9.17446434e-01
-8.61664116e-01 -1.00626945e+00 1.28192496e+00 -4.95935142e-01
1.03289747e+00 -2.74185985e-01 -8.07582021e-01 4.82484281e-01
5.32912970e-01 -3.91340435e-01 5.67477584e-01 -5.01650572e-02
-6.41920924e-01 1.41278312e-01 -1.39780879e+00 2.63294190e-01
5.34990072e-01 -2.01514527e-01 -1.00097477e+00 9.33242500e-01
7.54621685e-01 -4.09237564e-01 -2.40277365e-01 -3.14918607e-02
4.89101142e-01 -9.38214242e-01 4.98210013e-01 -9.31717396e-01
1.42990887e+00 1.82537869e-01 2.43821651e-01 -1.51663208e+00
-3.76304179e-01 -4.44234282e-01 -1.34553686e-01 1.11875284e+00
4.15311426e-01 -7.81509221e-01 4.04888749e-01 -1.53398812e-01
-1.38008967e-01 -5.64501286e-01 -1.05235195e+00 -5.55703163e-01
3.63127172e-01 -3.38920921e-01 1.77100822e-01 1.58377266e+00
1.07949167e-01 5.85182726e-01 -7.73422718e-01 -1.22486554e-01
3.34203035e-01 4.07468304e-02 3.14208329e-01 -1.15891826e+00
-4.38561775e-02 -6.74384177e-01 -4.05730247e-01 -8.81774962e-01
2.09804773e-01 -8.61385167e-01 -1.62137628e-01 -1.57632184e+00
-2.55187918e-02 3.71115543e-02 -1.08188041e-01 2.04757452e-01
2.17469245e-01 1.69765726e-01 -6.19441941e-02 1.95174053e-01
-3.92794222e-01 4.31397438e-01 1.26062405e+00 -5.51285446e-01
2.12720335e-01 -1.91456378e-01 -8.83664012e-01 8.82644176e-01
8.47208858e-01 -6.21049285e-01 3.31778735e-01 -4.34243381e-01
1.01134348e+00 -1.33848265e-01 7.35446095e-01 -6.30833626e-01
-3.25688481e-01 -7.84427375e-02 2.76825011e-01 -6.59072161e-01
4.53875549e-02 -3.81677151e-01 -9.40978825e-02 8.11726391e-01
-5.76936364e-01 -1.55467182e-01 -1.34456441e-01 7.54148960e-01
-2.85969555e-01 -1.75605059e-01 9.73786056e-01 -1.70229882e-01
-3.20277929e-01 -3.83187830e-01 -7.60624051e-01 2.03138411e-01
8.51261258e-01 2.05220804e-01 -1.13218832e+00 -7.04088569e-01
-4.17716891e-01 -2.87894875e-01 -8.67280662e-02 7.33816385e-01
4.79334563e-01 -1.35765064e+00 -1.14892972e+00 -4.14554328e-01
4.32475582e-02 -7.35088229e-01 8.54092464e-02 9.11429107e-01
-1.08732188e+00 2.67881244e-01 -1.67620435e-01 -5.26893660e-02
-8.80047500e-01 4.71908092e-01 1.41495004e-01 -2.04352766e-01
-4.53588396e-01 9.27279234e-01 -4.94446516e-01 3.85399535e-02
-2.21789762e-01 -1.14274859e-01 -4.32705164e-01 3.54788303e-01
5.58126450e-01 5.38821518e-01 -1.39861912e-01 -8.86079311e-01
-2.51610875e-01 -3.97466928e-01 -3.65795255e-01 4.13813554e-02
1.16630864e+00 7.46846125e-02 -4.44431514e-01 7.94409037e-01
1.37313843e+00 -9.85703990e-03 -2.87029356e-01 -9.08042565e-02
8.02741274e-02 -5.37286162e-01 4.50888008e-01 -1.57719028e+00
-5.86748958e-01 7.89876401e-01 3.94692719e-01 1.04685080e+00
2.91275501e-01 -4.73554097e-02 9.86514390e-01 1.46502897e-01
2.42536470e-01 -1.43796360e+00 3.92494112e-01 6.84621215e-01
1.42427278e+00 -1.37167859e+00 -1.13658361e-01 -1.80909544e-01
-5.18320858e-01 1.14424193e+00 4.49442118e-01 -4.18733865e-01
5.67858100e-01 -1.54526234e-01 2.36766502e-01 -4.02427435e-01
-5.86225152e-01 5.07906914e-01 1.17599033e-01 6.93721399e-02
8.15573931e-01 3.08927506e-01 -1.16354382e+00 7.12782562e-01
-3.24444890e-01 -1.35416403e-01 6.82359815e-01 7.40578115e-01
-6.62087262e-01 -4.01350498e-01 -4.31976527e-01 7.56521165e-01
-1.14502668e+00 -1.05302848e-01 -9.49609280e-01 6.18087530e-01
4.10331696e-01 1.19164908e+00 -8.86444598e-02 -4.98300701e-01
-2.35627457e-01 5.71915396e-02 1.42289370e-01 -3.27735186e-01
-8.64155769e-01 -1.97244719e-01 6.95468009e-01 -2.14998111e-01
-5.19268215e-01 -1.92992613e-01 -7.98011541e-01 -6.82918906e-01
-5.59022129e-01 8.99202600e-02 8.94589245e-01 1.24806821e+00
1.69584081e-01 3.41989934e-01 6.31849527e-01 -2.81159490e-01
-7.48681188e-01 -1.42817938e+00 -2.32760578e-01 6.37154877e-01
4.15382177e-01 -7.08989859e-01 -7.35825777e-01 -1.97368845e-01] | [8.247309684753418, 10.343926429748535] |
7508ae93-0ae9-476f-a318-e8f3037e505a | poet-a-self-learning-framework-for-profinet | 2305.03175 | null | https://arxiv.org/abs/2305.03175v1 | https://arxiv.org/pdf/2305.03175v1.pdf | POET: A Self-learning Framework for PROFINET Industrial Operations Behaviour | Since 2010, multiple cyber incidents on industrial infrastructure, such as Stuxnet and CrashOverride, have exposed the vulnerability of Industrial Control Systems (ICS) to cyber threats. The industrial systems are commissioned for longer duration amounting to decades, often resulting in non-compliance to technological advancements in industrial cybersecurity mechanisms. The unavailability of network infrastructure information makes designing the security policies or configuring the cybersecurity countermeasures such as Network Intrusion Detection Systems (NIDS) challenging. An empirical solution is to self-learn the network infrastructure information of an industrial system from its monitored network traffic to make the network transparent for downstream analyses tasks such as anomaly detection. In this work, a Python-based industrial communication paradigm-aware framework, named PROFINET Operations Enumeration and Tracking (POET), that enumerates different industrial operations executed in a deterministic order of a PROFINET-based industrial system is reported. The operation-driving industrial network protocol frames are dissected for enumeration of the operations. For the requirements of capturing the transitions between industrial operations triggered by the communication events, the Finite State Machines (FSM) are modelled to enumerate the PROFINET operations of the device, connection and system. POET extracts the network information from network traffic to instantiate appropriate FSM models (Device, Connection or System) and track the industrial operations. It successfully detects and reports the anomalies triggered by a network attack in a miniaturized PROFINET-based industrial system, executed through valid network protocol exchanges and resulting in invalid PROFINET operation transition for the device. | ['Jürgen Beyerer', 'Christian Haas', 'Markus Karch', 'Ankush Meshram'] | 2023-04-29 | null | null | null | null | ['network-intrusion-detection', 'self-learning'] | ['miscellaneous', 'natural-language-processing'] | [ 6.34982407e-01 2.04890408e-02 3.09419185e-02 2.07418859e-01
2.30998993e-01 -1.05386031e+00 6.22336268e-01 2.33458742e-01
2.89964497e-01 2.79559076e-01 -9.33528066e-01 -1.13876081e+00
-5.71301401e-01 -9.81606960e-01 -3.93476903e-01 -3.60226899e-01
-4.06069875e-01 4.21446830e-01 5.87117910e-01 3.29866409e-01
2.81956792e-01 9.66679454e-01 -1.62123454e+00 1.24726944e-01
5.85982949e-02 1.11705601e+00 -1.09045610e-01 9.70895708e-01
-1.67498723e-01 4.27604824e-01 -1.32975531e+00 3.14612031e-01
3.37270617e-01 -1.99265346e-01 -7.58575857e-01 -3.37791815e-02
-2.70901322e-01 -2.73211807e-01 -2.21523955e-01 1.22941768e+00
-2.97816396e-01 -2.40299582e-01 1.21249937e-01 -2.15359426e+00
4.73175138e-01 6.06249392e-01 3.55533175e-02 1.07774623e-02
3.43092293e-01 8.71902466e-01 4.01860923e-01 1.64184988e-01
6.94700241e-01 1.21196520e+00 7.15376362e-02 1.62156057e-02
-1.18392670e+00 -1.01536596e+00 5.96332066e-02 1.44699007e-01
-9.16856110e-01 7.90993199e-02 7.67051995e-01 -4.40798014e-01
1.19031501e+00 6.16928041e-01 3.15253943e-01 1.58627284e+00
7.61584997e-01 -3.14755589e-01 7.92869747e-01 -4.37017173e-01
5.68137825e-01 6.17643073e-02 6.14905715e-01 1.73265025e-01
7.95757949e-01 6.42319679e-01 2.15425640e-01 -3.92686814e-01
9.70876932e-01 4.11770105e-01 1.59908950e-01 1.47153944e-01
-1.29641843e+00 7.09695593e-02 -2.90695101e-01 6.19838655e-01
-5.69567502e-01 -9.93437022e-02 1.18499839e+00 5.96884072e-01
-6.86865509e-01 8.50118697e-01 -9.49893236e-01 -5.27241886e-01
-5.58940507e-02 -3.39096069e-01 1.41604912e+00 1.20875764e+00
6.12367153e-01 3.59805495e-01 4.30364102e-01 -3.80679846e-01
2.21740022e-01 3.42906654e-01 -1.39337808e-01 -8.75135422e-01
3.28559354e-02 9.75744903e-01 -1.52837232e-01 -7.97214985e-01
-3.54864568e-01 -4.78472076e-02 -7.23967016e-01 6.50982499e-01
2.77401537e-01 -2.06118286e-01 -5.86681724e-01 1.13939977e+00
2.82628119e-01 5.99647999e-01 1.09125949e-01 2.32206807e-01
-2.00236380e-01 6.62768662e-01 2.08981186e-01 -5.04619241e-01
1.41490102e+00 -4.45978232e-02 -9.35617566e-01 3.76071066e-01
5.21467507e-01 -8.02061319e-01 5.04143238e-01 7.98043370e-01
-4.26010996e-01 -5.32680571e-01 -1.39728117e+00 1.39775038e+00
-7.47610450e-01 -4.61736232e-01 3.32010865e-01 7.61129618e-01
-2.07021177e-01 7.53552794e-01 -9.66262817e-01 -2.50636280e-01
-1.13790452e-01 4.46013182e-01 -2.35341087e-01 3.19574565e-01
-1.18154657e+00 7.27924943e-01 9.81859028e-01 5.66532873e-02
-1.53672147e+00 -5.53688884e-01 -7.40941465e-01 1.80532653e-02
7.64754117e-01 -7.06468448e-02 1.03771436e+00 -4.77312684e-01
-1.47336531e+00 2.81946603e-02 7.39714801e-01 -6.05195224e-01
2.00413540e-01 2.85042882e-01 -1.52451098e+00 1.14897422e-01
-4.76031035e-01 -4.85234320e-01 8.16634476e-01 -1.11488390e+00
-7.62962461e-01 -2.36533925e-01 3.58876854e-01 -1.05997181e+00
3.71199027e-02 2.85385489e-01 5.04037857e-01 -3.86786982e-02
-3.65267962e-01 -9.21555817e-01 -1.22731894e-01 -8.03348184e-01
-1.17275131e+00 1.54250309e-01 1.54796553e+00 -1.92442685e-01
1.19377506e+00 -2.10889339e+00 -6.45155966e-01 9.49203789e-01
-2.32936233e-01 5.93968391e-01 -5.10074431e-03 9.44887757e-01
-6.04689181e-01 1.80714086e-01 -4.16903831e-02 4.47609127e-01
2.28281051e-01 4.10660744e-01 -5.99993348e-01 5.38561285e-01
-3.27645205e-02 6.65702000e-02 -7.36591220e-01 -7.47826770e-02
9.84303057e-01 1.78350676e-02 1.36292309e-01 3.35603744e-01
-4.72630858e-01 7.31547415e-01 -6.32653177e-01 7.68899083e-01
6.21450186e-01 1.76642761e-01 6.16464376e-01 -2.70673126e-01
-5.13883352e-01 2.78885961e-01 -1.44415915e+00 9.84769404e-01
-7.34127164e-01 3.94698754e-02 3.64593595e-01 -6.37023807e-01
8.58866334e-01 9.01011109e-01 6.67987943e-01 -2.18650460e-01
3.51482570e-01 2.59720534e-01 1.13799095e-01 -3.53139311e-01
-3.02933425e-01 2.41682127e-01 -2.39775866e-01 5.78184903e-01
1.30685270e-01 -1.72353480e-02 4.02223349e-01 -5.70394956e-02
1.91775239e+00 -5.47902063e-02 5.31308055e-01 -2.43557319e-01
1.00526106e+00 9.71366689e-02 4.84462678e-01 4.23139840e-01
7.14520887e-02 -7.72900641e-01 7.27315724e-01 -6.10742509e-01
-7.89621234e-01 -1.28481853e+00 5.85329458e-02 9.56718065e-03
6.75883889e-02 -6.77325368e-01 -4.33884919e-01 -1.18413365e+00
-3.16164821e-01 1.31251705e+00 1.05999961e-01 -6.31803513e-01
-8.58627796e-01 -6.00909293e-02 5.77851415e-01 1.21894881e-01
2.74239033e-01 -1.35749292e+00 -8.62212956e-01 5.54087937e-01
5.16756177e-01 -1.60017955e+00 1.23541690e-02 5.40200412e-01
-8.32913578e-01 -1.88637412e+00 9.79017437e-01 -3.88561010e-01
6.53238058e-01 -4.36333865e-01 7.70254374e-01 1.04893394e-01
-8.40112984e-01 4.53284144e-01 -1.84109062e-02 -5.97768962e-01
-1.31203485e+00 -2.70536840e-01 3.27722549e-01 -5.05637228e-02
2.91940868e-01 -8.71757150e-01 -6.68845028e-02 6.90989137e-01
-1.14654839e+00 -7.69966245e-01 5.27611136e-01 2.40105331e-01
2.42069438e-01 1.15805030e+00 5.38976133e-01 -8.54578555e-01
4.63401318e-01 -4.82283711e-01 -1.37244296e+00 3.08115520e-02
-9.57710743e-01 -1.95499197e-01 1.40963125e+00 -7.00633585e-01
-5.95404387e-01 7.11058378e-02 1.34320155e-01 -5.38373172e-01
-1.30962515e+00 1.70295626e-01 -6.11723304e-01 3.45328480e-01
1.80884704e-01 6.38507381e-02 2.64278781e-02 -2.61542410e-01
-3.63387913e-01 4.08311039e-01 5.16856015e-01 -5.70859969e-01
1.12383544e+00 2.28664130e-01 4.08133596e-01 -8.48325789e-01
-1.18163042e-02 -2.40120322e-01 -2.40699768e-01 -5.61861038e-01
6.62429154e-01 8.08987990e-02 -1.69089127e+00 2.93553978e-01
-1.52534699e+00 1.54357171e-04 -5.02449155e-01 4.79538530e-01
-3.52467477e-01 3.72959197e-01 -7.36594975e-01 -1.05435109e+00
-6.30193278e-02 -1.25375843e+00 5.85898995e-01 -2.08629683e-01
-6.25941873e-01 -6.89684153e-01 1.81084797e-01 -4.31023985e-01
2.48312548e-01 6.23575211e-01 1.28282094e+00 -1.22384322e+00
-7.78854489e-01 -6.73269510e-01 2.23179251e-01 7.65017152e-01
6.84410810e-01 5.30256152e-01 -7.41833448e-01 -4.76574272e-01
2.20343530e-01 5.53414643e-01 -7.45082676e-01 -1.99742109e-01
1.07877707e+00 -5.00088334e-01 -4.81494367e-01 2.05792397e-01
1.42715430e+00 1.09697163e+00 6.02321744e-01 2.95102626e-01
3.33357066e-01 6.27905548e-01 7.04339445e-01 4.53514278e-01
-6.75747037e-01 5.14958560e-01 1.19064581e+00 3.27178985e-01
4.30108666e-01 -8.10592398e-02 7.75000930e-01 2.95904100e-01
1.42878711e-01 4.43332680e-02 -5.76540768e-01 -1.58619434e-01
-1.11303091e+00 -1.06930959e+00 -3.69885594e-01 2.33169103e+00
1.41379699e-01 7.88318813e-01 -9.23851579e-02 6.45993054e-01
9.00686741e-01 -2.58677542e-01 -5.95175743e-01 -6.82145119e-01
5.17703831e-01 2.94730127e-01 6.46207035e-01 1.63393214e-01
-6.08286023e-01 3.85127246e-01 4.74350929e+00 4.71450567e-01
-7.96742201e-01 -1.41016617e-01 -2.37976357e-01 5.13117731e-01
5.14399350e-01 4.70446825e-01 -7.35720932e-01 6.42816246e-01
1.89474988e+00 -8.84199589e-02 4.62098718e-01 7.92272568e-01
6.03453517e-01 3.06238055e-01 -1.38891983e+00 5.92093945e-01
-6.83535993e-01 -9.44169641e-01 -8.16712305e-02 4.74511147e-01
1.73424467e-01 -4.24120277e-01 -5.12204587e-01 1.76953554e-01
5.09908915e-01 -5.35430789e-01 5.46113729e-01 1.88943058e-01
4.38456416e-01 -9.99886453e-01 7.68381834e-01 1.75039276e-01
-1.33285832e+00 -4.38979566e-01 4.74670589e-01 9.54437107e-02
6.97250247e-01 6.68748200e-01 -1.14384246e+00 1.01940644e+00
4.93027687e-01 2.04262748e-01 1.34493550e-02 5.33175290e-01
-2.24330425e-01 8.65007401e-01 -4.16253686e-01 2.75581062e-01
3.96057144e-02 -2.57983565e-01 1.14024508e+00 6.27034605e-01
1.63956746e-01 -5.08828878e-01 1.92703679e-01 1.03103995e+00
5.72877824e-01 -7.27066159e-01 -9.53234315e-01 -5.56279540e-01
8.31710935e-01 1.43357468e+00 -8.78906369e-01 -5.37865199e-02
-1.48500904e-01 4.50081050e-01 -9.17929828e-01 3.63860637e-01
-9.29262161e-01 -7.01844513e-01 9.41624522e-01 1.49266109e-01
4.91390042e-02 -5.98343849e-01 2.72111028e-01 -2.22941101e-01
-6.33133426e-02 -1.13379097e+00 4.47298944e-01 -3.60963881e-01
-1.38249385e+00 8.19889605e-01 1.76151320e-01 -1.49818051e+00
-4.30895835e-01 -8.40195060e-01 -8.20698321e-01 3.90556157e-01
-8.00970316e-01 -9.44503188e-01 -5.45046404e-02 9.42999899e-01
2.95848906e-01 -4.94419783e-02 1.26213634e+00 2.28873864e-01
-9.35684204e-01 -1.06435955e-01 -7.61079133e-01 -9.11287814e-02
2.63524741e-01 -9.50535357e-01 3.83558333e-01 1.07862318e+00
-1.69377878e-01 7.57250845e-01 9.95803058e-01 -1.06102622e+00
-1.72563303e+00 -1.02740574e+00 2.88510084e-01 -1.88534632e-01
1.23336530e+00 -4.37530130e-01 -4.79913533e-01 1.17344439e+00
2.66873747e-01 -1.79083779e-01 3.52662206e-01 -4.98948306e-01
-7.19623417e-02 -2.32987881e-01 -1.24986804e+00 5.07520974e-01
6.67918563e-01 -3.80928278e-01 -3.53035897e-01 3.80878061e-01
9.34178531e-01 7.68400282e-02 -1.22166133e+00 2.53518850e-01
6.07011421e-03 -8.68024409e-01 9.30718362e-01 -7.48780251e-01
-2.95204043e-01 -6.52637482e-01 7.38522857e-02 -8.82573783e-01
1.63942918e-01 -1.30847502e+00 -4.70567137e-01 1.36333227e+00
8.41422528e-02 -1.37112653e+00 2.56261408e-01 1.68289483e-01
-3.58038366e-01 -3.01734731e-02 -1.15342224e+00 -1.30232942e+00
-9.42958772e-01 -6.29118800e-01 1.13229966e+00 1.05337894e+00
3.47065896e-01 1.18745454e-01 2.52734989e-01 9.61694658e-01
5.71387649e-01 -1.08306147e-01 6.85556054e-01 -1.61931968e+00
-1.67190492e-01 -3.69532853e-01 -7.02098310e-01 2.68103834e-02
2.09411040e-01 -3.88366461e-01 -1.98194116e-01 -9.08976793e-01
-7.86443949e-01 -3.29935879e-01 -5.14598310e-01 5.28397024e-01
1.19443440e+00 -4.96184677e-01 1.69829186e-02 -1.04346976e-01
-1.87101871e-01 -2.75217921e-01 6.29729927e-01 -3.28411430e-01
-1.32416755e-01 4.91620183e-01 4.07584190e-01 6.46006048e-01
9.17839408e-01 -5.54930270e-01 -5.36581278e-01 6.89412117e-01
1.45528302e-01 4.52660769e-01 7.42831886e-01 -1.38586700e+00
2.09279865e-01 -3.40996325e-01 -2.51414329e-01 -7.97742844e-01
2.38354784e-02 -2.02632594e+00 9.07334507e-01 1.00320256e+00
9.36536416e-02 6.12996399e-01 2.90850252e-01 6.07219756e-01
-1.68564886e-01 -1.63827464e-01 4.15920317e-01 1.71121523e-01
-5.35734951e-01 3.42842400e-01 -9.06921983e-01 -7.64328063e-01
1.60298193e+00 -2.12134033e-01 -3.55593711e-01 2.16352537e-01
-8.08128834e-01 -1.38641790e-01 2.29223087e-01 4.68542933e-01
4.31356698e-01 -7.92937636e-01 2.05634683e-01 9.14772332e-01
8.51239543e-03 -2.51475453e-01 2.52793044e-01 7.92859018e-01
-2.07079589e-01 4.37381268e-01 -3.96808714e-01 -3.83612096e-01
-1.19960558e+00 1.07447278e+00 3.17044139e-01 -5.98918319e-01
-6.51164591e-01 -2.81375021e-01 -4.07296568e-01 -4.03243005e-01
7.78204277e-02 -7.24399984e-01 -7.10239634e-02 -4.01732057e-01
5.63716769e-01 1.04501963e+00 3.99802476e-01 -6.91522062e-02
-5.35677969e-01 7.40520954e-02 1.07136399e-01 1.89891398e-01
1.03098929e+00 1.37769789e-01 -4.51434761e-01 4.22905356e-01
9.15189981e-01 -2.84108847e-01 -9.49092209e-01 4.94125664e-01
3.77782047e-01 -9.95154381e-02 -1.17658690e-01 -1.00476503e+00
-1.11883593e+00 3.67390007e-01 4.95150030e-01 1.11060309e+00
9.81998563e-01 -3.00640643e-01 8.74280930e-01 2.13744581e-01
1.02671468e+00 -7.23893940e-01 -1.81508794e-01 5.29523432e-01
2.03177109e-01 -3.29999924e-01 -3.56033623e-01 -6.68193281e-01
-4.23667468e-02 1.51867151e+00 6.17960274e-01 7.33154477e-04
9.04316485e-01 1.14997077e+00 -3.29662204e-01 -5.44780195e-01
-5.31427860e-01 6.75100744e-01 -5.03618121e-01 8.76186609e-01
-6.37491226e-01 1.39490768e-01 2.07840502e-01 5.17089725e-01
7.35867694e-02 3.81821245e-02 6.56387687e-01 1.15836442e+00
-8.45633969e-02 -1.47077966e+00 -5.34417748e-01 2.45365918e-01
-3.47006053e-01 4.58686054e-01 -2.43401587e-01 1.11428571e+00
2.67390877e-01 1.23905694e+00 2.16467187e-01 -7.21629083e-01
7.84014404e-01 4.37049679e-02 6.78906962e-02 -5.19222379e-01
-8.51423800e-01 -2.74700731e-01 4.80486810e-01 -1.03845179e+00
2.24699676e-01 -5.48773766e-01 -1.48324656e+00 -3.69047910e-01
-2.77591079e-01 -1.36773735e-02 8.46308768e-01 1.01436687e+00
4.44220394e-01 1.21519697e+00 8.97916794e-01 -4.96005952e-01
-7.75079489e-01 -7.58757949e-01 -9.69006896e-01 1.55941859e-01
2.49562711e-01 -7.18099058e-01 -9.47604597e-01 -6.27055019e-02] | [5.34089469909668, 7.137630462646484] |
7d7368e2-1b8f-4722-b47e-af48530f26ba | 6d-dynamic-camera-relocalization-from-single | null | null | http://openaccess.thecvf.com/content_cvpr_2016/html/Feng_6D_Dynamic_Camera_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Feng_6D_Dynamic_Camera_CVPR_2016_paper.pdf | 6D Dynamic Camera Relocalization From Single Reference Image | Dynamic relocalization of 6D camera pose from single reference image is a costly and challenging task that requires delicate hand-eye calibration and precision positioning platform to do 3D mechanical rotation and translation. In this paper, we show that high-quality camera relocalization can be achieved in a much less expensive way. Based on inexpensive platform with unreliable absolute repositioning accuracy (ARA), we propose a hand-eye calibration free strategy to actively relocate camera into the same 6D pose that produces the input reference image, by sequentially correcting 3D relative rotation and translation. We theoretically prove that, by this strategy, both rotational and translational relative pose can be effectively reduced to zero, with bounded unknown hand-eye pose displacement. To conquer 3D rotation and translation ambiguity, this theoretical strategy is further revised to a practical relocalization algorithm with faster convergence rate and more reliability by jointly adjusting 3D relative rotation and translation. Extensive experiments validate the effectiveness and superior accuracy of the proposed approach on laboratory tests and challenging real-world applications. | ['Qian Zhang', 'Fei-Peng Tian', 'Jizhou Sun', 'Wei Feng'] | 2016-06-01 | null | null | null | cvpr-2016-6 | ['camera-relocalization'] | ['computer-vision'] | [ 1.56140849e-01 -3.75515455e-03 -2.39107460e-01 3.77205983e-02
-5.05104780e-01 -9.59456086e-01 2.50806063e-01 -5.90928137e-01
-5.44159055e-01 5.84131718e-01 -3.09158683e-01 -2.78219432e-01
-2.18131185e-01 1.26733676e-01 -9.37618911e-01 -7.95707107e-01
7.07943439e-01 2.49829277e-01 1.76018223e-01 2.35038847e-02
5.86797059e-01 7.59988308e-01 -1.26419437e+00 -9.32242334e-01
6.92836523e-01 5.21772444e-01 2.79734284e-01 7.95558453e-01
8.27058554e-01 3.57646644e-01 -5.02959609e-01 -7.93909058e-02
5.20171285e-01 -2.09400624e-01 -5.04826367e-01 6.18189692e-01
5.06291211e-01 -6.03458107e-01 2.78354734e-01 1.10101926e+00
5.03658950e-01 3.35098221e-03 2.71312654e-01 -8.57235193e-01
-4.41610277e-01 -2.77883280e-02 -1.13596690e+00 -3.10218781e-02
8.65683496e-01 -1.08753778e-01 4.56116796e-01 -8.94255340e-01
7.41002560e-01 6.87898755e-01 6.41312659e-01 3.78158927e-01
-1.34893632e+00 -6.44742012e-01 1.00717902e-01 -2.43717909e-01
-1.91633511e+00 -3.17049414e-01 7.77730405e-01 -3.26194435e-01
5.37594080e-01 2.38654584e-01 4.95735675e-01 8.38111222e-01
2.92135864e-01 5.54150641e-02 8.44401240e-01 -6.23735964e-01
-1.89883307e-01 3.52799177e-01 -1.64997414e-01 5.52332819e-01
8.45046997e-01 -1.26313537e-01 -2.29533017e-01 2.67611235e-01
1.16293871e+00 2.34188065e-01 -7.41956174e-01 -1.03277636e+00
-1.54893148e+00 2.23559275e-01 2.34970436e-01 1.84614081e-02
-1.61816448e-01 4.40718606e-02 -4.12377447e-01 1.19173266e-01
1.93102866e-01 5.30073106e-01 -3.78860772e-01 -1.63211912e-01
-4.97693449e-01 -1.31516740e-01 4.04773384e-01 1.54524481e+00
7.71740198e-01 -1.10470511e-01 3.79948139e-01 3.95463079e-01
2.59680480e-01 1.03878319e+00 4.87046897e-01 -9.79998291e-01
4.81897324e-01 3.89924526e-01 7.58206844e-01 -1.00854921e+00
-4.17746395e-01 -3.51020187e-01 -4.92703587e-01 2.45606855e-01
3.81503373e-01 -1.49704680e-01 -4.56516117e-01 1.42112386e+00
7.68963099e-01 5.76462299e-02 6.75955042e-03 1.34079325e+00
1.24757215e-01 2.53996432e-01 -6.45370364e-01 -5.90948224e-01
1.17350161e+00 -7.29404151e-01 -6.54933631e-01 -1.72107548e-01
5.99464476e-01 -1.27217102e+00 9.55889523e-01 5.91167748e-01
-8.71179283e-01 -5.03827989e-01 -1.37499177e+00 -8.58541429e-02
3.73893023e-01 6.17849648e-01 2.01160401e-01 7.79440463e-01
-9.88798440e-01 2.42675152e-02 -8.10139179e-01 -5.41587591e-01
-4.44012105e-01 6.93759441e-01 -6.60193801e-01 1.37832165e-01
-5.77642918e-01 9.61851716e-01 2.18176350e-01 4.23200965e-01
-3.53942722e-01 -4.77661788e-01 -5.97416759e-01 -4.34034675e-01
8.07520151e-01 -7.20998645e-01 1.11401117e+00 -4.56169814e-01
-2.17821026e+00 1.00693357e+00 -2.87741184e-01 -6.48397952e-02
4.65691775e-01 -6.70234978e-01 3.59633602e-02 1.94106892e-01
-2.43374035e-02 2.88873136e-01 1.17238271e+00 -1.32856345e+00
-3.86375219e-01 -6.16748869e-01 -2.45967451e-02 7.59797990e-01
-1.88105404e-01 -9.90920663e-02 -6.61298633e-01 -2.49974743e-01
8.39223742e-01 -1.44303346e+00 2.96189748e-02 7.11221099e-02
-3.20891678e-01 2.53271908e-01 7.71402657e-01 -1.26074255e-01
8.66600275e-01 -2.07094955e+00 3.78325164e-01 7.11398274e-02
2.62514412e-01 2.12733939e-01 2.81776488e-01 -3.70875327e-03
-2.38359988e-01 -1.81054741e-01 3.91694635e-01 -4.59713191e-01
-5.35217464e-01 -3.93086635e-02 -3.43526810e-01 1.13329709e+00
-2.21523300e-01 6.04279995e-01 -8.14599514e-01 -2.43307084e-01
4.44196820e-01 5.60655832e-01 -3.73614758e-01 1.51166022e-01
4.48690981e-01 7.46973097e-01 -2.94737935e-01 6.79137886e-01
8.74300241e-01 -7.51362443e-02 2.34536171e-01 -3.42120200e-01
-2.45528892e-01 -6.47374466e-02 -1.43915153e+00 1.69952047e+00
-4.85697985e-01 5.63468575e-01 2.04268456e-01 -5.30540287e-01
1.08356154e+00 1.91692904e-01 3.64700705e-01 -4.51113909e-01
5.70692241e-01 1.70423776e-01 -4.67270613e-01 -2.75191903e-01
8.35315585e-01 -2.13382337e-02 9.17881206e-02 5.21168530e-01
-2.19124526e-01 -5.26829839e-01 -4.19695556e-01 -2.17155203e-01
2.57011741e-01 3.80723566e-01 5.36222637e-01 -2.38401473e-01
8.01975012e-01 -2.88253099e-01 3.66913050e-01 3.77312392e-01
-1.65095523e-01 7.39328504e-01 1.80682212e-01 -2.11276472e-01
-1.13959110e+00 -8.62506270e-01 8.59871656e-02 4.77662653e-01
9.15101707e-01 -4.78454009e-02 -8.43341708e-01 -2.12845296e-01
-2.16883689e-01 1.40332028e-01 -3.18586379e-01 -1.02980137e-01
-7.35657394e-01 -4.74445522e-01 6.32152930e-02 2.24403232e-01
2.38809034e-01 4.24883850e-02 -8.96304786e-01 -2.19420806e-01
-1.56570598e-01 -1.19609094e+00 -7.12858379e-01 -2.23616764e-01
-9.24103796e-01 -1.11526144e+00 -8.52413118e-01 -5.91908574e-01
1.23890316e+00 1.18966711e+00 5.63797832e-01 -9.97463167e-02
3.25926393e-02 5.27828157e-01 -3.31390500e-01 -5.75809516e-02
3.23426649e-02 -4.01482061e-02 5.64047635e-01 2.20006317e-01
-4.17854562e-02 -3.56785178e-01 -7.82684088e-01 1.07673514e+00
-5.65665603e-01 5.53702079e-02 5.98886788e-01 5.65249264e-01
7.43965149e-01 -2.68407986e-02 -8.93933550e-02 -3.37811589e-01
2.63641119e-01 2.53722429e-01 -1.29222095e+00 1.98364928e-02
-7.92064905e-01 -4.37172130e-02 3.51016968e-01 -6.80368185e-01
-1.00382769e+00 3.34186405e-01 3.87593865e-01 -6.55529082e-01
7.31103271e-02 -1.17599182e-01 -1.13474615e-01 -4.76447195e-01
7.28049755e-01 1.06085025e-01 2.18608841e-01 -3.25995594e-01
5.74298918e-01 7.63667643e-01 8.87722492e-01 -4.87206161e-01
1.13235581e+00 8.42308044e-01 3.17079514e-01 -7.95262456e-01
-5.37357509e-01 -5.08166492e-01 -1.10865510e+00 -4.36384767e-01
7.20034301e-01 -1.10779107e+00 -1.10706723e+00 4.05782819e-01
-1.06403470e+00 2.19364136e-01 2.58017510e-01 8.98326933e-01
-3.70023966e-01 7.35626698e-01 1.12209849e-01 -6.62088335e-01
-1.93702698e-01 -1.44659793e+00 1.49228275e+00 2.48746872e-01
-2.09451139e-01 -6.30399585e-01 -1.37311555e-02 3.39108318e-01
4.20580581e-02 2.00277433e-01 -1.42800078e-01 3.22554916e-01
-8.39980364e-01 -5.09951711e-01 -4.28381190e-02 7.65171126e-02
2.34297320e-01 2.45293856e-01 -5.67732930e-01 -6.43120944e-01
4.34739172e-01 5.76699786e-02 8.05912912e-02 2.58908749e-01
3.21824461e-01 -9.24887322e-03 -3.43102217e-01 9.55592990e-01
1.42800355e+00 2.56339312e-01 4.20824230e-01 5.40561140e-01
8.55643451e-01 2.47594729e-01 1.04734385e+00 3.30011845e-01
2.06571579e-01 9.65044975e-01 5.67307115e-01 1.59729719e-01
8.40903372e-02 -2.12420017e-01 3.87218684e-01 6.96178198e-01
-4.35775608e-01 -1.61973983e-01 -4.95715588e-01 1.54839516e-01
-1.63179934e+00 -4.26619202e-01 -2.47154340e-01 2.70333982e+00
6.15831316e-01 2.99502886e-03 -1.97459489e-01 2.16308683e-01
9.19963598e-01 -1.56431779e-01 -4.61402535e-01 3.67259746e-03
1.50579095e-01 -2.86723763e-01 1.05812752e+00 8.33518267e-01
-7.45250285e-01 9.61284697e-01 6.02909088e+00 2.87258863e-01
-1.62238669e+00 -2.61362828e-03 -2.74724606e-02 -2.18810856e-01
-4.42995653e-02 1.84709802e-01 -1.19995105e+00 1.98912978e-01
4.54298228e-01 1.07167155e-01 3.85924667e-01 8.43096435e-01
1.02735043e-01 -4.36330676e-01 -1.00566232e+00 1.42406535e+00
4.53741103e-01 -8.73543799e-01 -2.93612480e-01 1.10590674e-01
6.69246912e-01 -4.54269469e-01 2.25579858e-01 -4.14325655e-01
-1.71674758e-01 -2.96666324e-01 8.79163802e-01 4.17823136e-01
1.08937335e+00 -6.63095713e-01 5.67957699e-01 4.63916987e-01
-9.56943989e-01 -7.79558113e-03 -4.74496722e-01 -1.37343824e-01
2.19706163e-01 3.15735936e-01 -9.47223604e-01 7.04894900e-01
3.82970184e-01 6.01825058e-01 -4.84565079e-01 4.97206986e-01
-5.66163242e-01 -5.65201864e-02 -4.33196515e-01 4.40007634e-02
-2.58636147e-01 -6.60749197e-01 7.26695299e-01 3.18820804e-01
5.31888187e-01 1.99818075e-01 -2.78027147e-01 4.45718825e-01
1.91953376e-01 -2.23817363e-01 -6.74332976e-01 5.94700933e-01
5.00484228e-01 1.19385159e+00 -7.62062013e-01 2.85643190e-02
-1.25319555e-01 1.16163659e+00 -4.29772250e-02 4.64530289e-01
-1.19425881e+00 -4.49647546e-01 5.07687092e-01 2.09551007e-01
2.68172145e-01 -6.40132666e-01 -2.71599367e-02 -1.44694579e+00
3.96584272e-01 -3.70309025e-01 -1.26860082e-01 -1.31770098e+00
-1.95974529e-01 5.95122695e-01 8.47401917e-02 -1.87742674e+00
-3.73135269e-01 -5.72804332e-01 -5.50011769e-02 6.89544380e-01
-1.31361926e+00 -1.03700864e+00 -5.18145025e-01 6.23953462e-01
2.64642209e-01 1.94518581e-01 4.76116121e-01 1.00503273e-01
-6.04565918e-01 9.12677824e-01 8.74097720e-02 -3.04951370e-01
1.25478208e+00 -8.79067481e-01 -1.35851607e-01 1.12868726e+00
-1.01293184e-01 8.70525301e-01 7.84739614e-01 -2.34282330e-01
-1.89901090e+00 -7.57555842e-01 6.22626841e-01 -9.86344516e-01
6.90553606e-01 -4.96613771e-01 -3.47669810e-01 1.01600707e+00
2.27989051e-02 -5.19531071e-02 1.11195318e-01 -3.12532932e-01
-1.56861052e-01 -5.49844503e-01 -1.02033746e+00 7.21033037e-01
9.76023257e-01 -4.83047009e-01 -5.70532382e-01 1.70025080e-01
8.29803586e-01 -9.96864676e-01 -6.39657676e-01 1.08145393e-01
6.86822414e-01 -8.19024146e-01 1.14746690e+00 2.72399813e-01
-3.03195745e-01 -9.58061695e-01 1.68547332e-02 -8.72975469e-01
-1.00652814e-01 -1.41406131e+00 2.26077422e-01 1.02470970e+00
3.36145237e-02 -8.74528706e-01 6.20615959e-01 2.73449659e-01
5.13335988e-02 -3.01417947e-01 -9.59244907e-01 -8.75147998e-01
-4.49968517e-01 -1.50652096e-01 3.68960321e-01 6.72188103e-01
7.68652707e-02 6.00326538e-01 -7.21898139e-01 8.13522279e-01
6.80418134e-01 1.26071781e-01 1.39724803e+00 -1.04042685e+00
-2.03014072e-02 7.71863982e-02 -6.85570359e-01 -1.67493594e+00
-9.71766040e-02 -1.85212299e-01 -2.62496825e-02 -7.18826532e-01
-4.00328413e-02 -4.76463623e-02 1.64844140e-01 -5.67152426e-02
-1.51033290e-02 4.98773456e-01 -6.77387491e-02 5.44408917e-01
-4.99517202e-01 2.04822719e-01 1.32051694e+00 3.14356327e-01
-4.61557567e-01 4.47455980e-02 -6.90962434e-01 6.45597517e-01
4.66862381e-01 -4.40669000e-01 -7.89836228e-01 -6.87165380e-01
6.48725450e-01 3.25436860e-01 2.77963579e-01 -9.58103240e-01
2.72321194e-01 -8.47073346e-02 2.25845441e-01 -7.30386674e-01
4.13107634e-01 -1.19153249e+00 3.36880744e-01 2.88439035e-01
1.82508722e-01 1.66156262e-01 -1.05789304e-01 5.84835351e-01
3.66611741e-02 -1.15606397e-01 8.81332874e-01 2.97926873e-01
-2.38345385e-01 9.48048532e-02 -1.57200828e-01 -6.37443960e-01
1.46035695e+00 -7.22929299e-01 -2.94410110e-01 -4.55809832e-01
-6.65255487e-01 -1.64563701e-01 1.03894329e+00 3.97833258e-01
4.67387974e-01 -1.19352877e+00 -1.33900836e-01 5.90175629e-01
1.58155233e-01 2.29327813e-01 1.13458477e-01 1.42655194e+00
-9.49546039e-01 7.52481699e-01 1.49672776e-01 -1.22622073e+00
-1.57524240e+00 8.29009295e-01 2.95134872e-01 2.79014558e-01
-5.36640763e-01 5.45361221e-01 -2.11845674e-02 -2.63087332e-01
-3.16203646e-02 -6.05628908e-01 2.96144243e-02 -2.55994916e-01
4.06811327e-01 2.57703573e-01 8.89641047e-02 -8.37874770e-01
-3.66778314e-01 1.53917646e+00 -7.04541430e-03 -1.51415825e-01
8.93283367e-01 -9.24168706e-01 -2.13660393e-02 -2.97173299e-02
1.10773814e+00 3.66160780e-01 -1.36274290e+00 3.53216305e-02
-4.32513267e-01 -9.27928984e-01 -1.34445265e-01 -8.12310800e-02
-8.53523970e-01 5.67813158e-01 6.12053633e-01 -2.25595504e-01
1.02684760e+00 -1.14370003e-01 2.94936091e-01 7.10165322e-01
6.30527437e-01 -9.80718791e-01 4.05875072e-02 1.99767113e-01
7.61252284e-01 -1.04113841e+00 4.17923629e-01 -5.64057469e-01
-3.70178282e-01 1.05773115e+00 5.36260664e-01 -1.31262094e-01
4.09844726e-01 4.36519310e-02 1.01085149e-01 1.78845316e-01
-2.56705582e-01 2.73344785e-01 5.96674606e-02 5.39133132e-01
1.22172013e-01 -1.82961434e-01 -2.08732933e-01 5.52579947e-02
-7.35455528e-02 -1.84988886e-01 1.02220368e+00 9.61079895e-01
-4.08955842e-01 -9.32878017e-01 -9.65814114e-01 -6.33815110e-01
-1.84294671e-01 3.40865910e-01 -2.38986105e-01 1.07998633e+00
-2.23582029e-01 8.92357588e-01 2.09099740e-01 -3.66434634e-01
4.44688559e-01 -3.88657302e-01 7.98934102e-01 -2.76404619e-01
2.24123180e-01 6.08291626e-01 -3.93362164e-01 -6.88632011e-01
-6.80012226e-01 -4.85947490e-01 -1.08850253e+00 -2.76764542e-01
-7.89453804e-01 -1.26007423e-02 7.29044020e-01 7.14425802e-01
4.82693583e-01 7.80332759e-02 9.77068603e-01 -1.06269491e+00
-6.01887047e-01 -7.68749177e-01 -5.65803409e-01 7.09030107e-02
7.17327356e-01 -8.29043388e-01 -7.82128632e-01 2.52673715e-01] | [7.903773307800293, -2.2010228633880615] |
c3bfbfc6-13e7-4be6-aa1f-b30cca690c88 | towards-interactive-language-modeling-1 | null | null | https://openreview.net/forum?id=vD5JzgHTt9Q | https://openreview.net/pdf?id=vD5JzgHTt9Q | Towards Interactive Language Modeling | Interaction between caregivers and children plays a critical role in human language acquisition and development. Given this observation, it is remarkable that explicit interaction plays little to no role in artificial language modeling---which also targets the acquisition of human language, yet by artificial models. Moreover, an interactive approach to language modeling has the potential to make language models substantially more versatile and to considerably impact downstream applications. Motivated by these considerations, we pioneer the space of interactive language modeling. First we present a road map in which we detail the steps that need to be taken towards interactive language modeling. We then lead by example and take the first steps on this road map, showing the initial feasibility of our approach. As such, this work aims to be the start of a larger research agenda on interactive language modeling. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['language-acquisition'] | ['natural-language-processing'] | [ 5.88560477e-02 1.06842160e+00 -5.12971468e-02 -5.47899127e-01
-1.65986598e-01 -3.70005012e-01 7.29357421e-01 4.48737949e-01
-3.09498012e-01 4.73276883e-01 3.37323368e-01 -7.13921726e-01
7.45044053e-02 -7.77677417e-01 -4.74055022e-01 2.51574162e-02
-7.89984912e-02 5.31147718e-01 1.14966318e-01 -2.95871556e-01
-7.78951049e-02 5.19539893e-01 -1.63512683e+00 3.35882664e-01
8.15864921e-01 -2.88902581e-01 3.41861516e-01 4.78127629e-01
-6.05052590e-01 1.22366929e+00 -2.75262594e-01 -4.99799132e-01
-5.09558797e-01 -5.61731398e-01 -1.16382909e+00 -2.99677160e-02
-2.74370074e-01 -5.50759971e-01 2.46903077e-02 6.14673972e-01
2.52771348e-01 4.56048176e-02 5.41593432e-01 -1.03680694e+00
-5.27987659e-01 1.10981989e+00 1.05455965e-01 -2.77354568e-01
8.37162733e-01 -9.84067023e-02 5.97675323e-01 -8.52350593e-01
7.66219616e-01 1.35984409e+00 4.10014749e-01 8.90033066e-01
-1.28968346e+00 -3.74338388e-01 6.23212636e-01 -2.49747992e-01
-1.31410360e+00 -5.08446097e-01 6.23885274e-01 -7.28621364e-01
1.33351648e+00 1.30071625e-01 1.24721503e+00 6.75083458e-01
-2.94008881e-01 1.04310679e+00 8.94687474e-01 -1.13659298e+00
-1.44820690e-01 5.19181907e-01 4.41432774e-01 6.32040620e-01
1.55686587e-01 7.19949976e-02 -5.98365426e-01 1.03785008e-01
8.23011875e-01 -4.59794044e-01 1.34540955e-03 -2.99073517e-01
-9.29742992e-01 5.79117417e-01 -1.47492915e-01 9.64465737e-01
-3.22824046e-02 2.48710051e-01 2.02676252e-01 4.09228623e-01
4.22408819e-01 2.74642497e-01 -3.59594971e-01 -3.82034123e-01
-7.59651721e-01 3.07446122e-01 8.73269975e-01 1.00071454e+00
4.17866647e-01 3.10274977e-02 2.75131881e-01 8.67913604e-01
7.52667964e-01 1.17768146e-01 -1.34508148e-01 -8.81006181e-01
1.27179604e-02 7.48095810e-01 1.07601359e-01 -4.64540690e-01
-4.54795986e-01 3.65185877e-03 -1.11864716e-01 3.65325600e-01
6.36102557e-01 -1.19962178e-01 -5.17453194e-01 1.74679089e+00
3.01081330e-01 -2.65624672e-01 -2.61894148e-02 2.22784296e-01
9.59243894e-01 4.13547575e-01 6.60307527e-01 -4.03071553e-01
1.14729083e+00 -5.71083665e-01 -7.68007934e-01 1.19132265e-01
1.26039433e+00 -6.58088446e-01 1.14090395e+00 3.62767339e-01
-1.66537952e+00 -2.85139441e-01 -6.86894536e-01 -2.42134333e-01
-5.12180209e-01 -1.15599841e-01 1.12224782e+00 9.28536832e-01
-1.32123911e+00 2.22700924e-01 -1.19133818e+00 -6.47881150e-01
-2.94682849e-02 5.28020382e-01 -2.97100991e-01 5.40870167e-02
-1.07529986e+00 1.16142190e+00 4.58536059e-01 -7.75754526e-02
-4.35717553e-01 -6.73390925e-01 -9.38643038e-01 -2.32270777e-01
1.26203652e-02 -6.25357389e-01 1.65645421e+00 -1.22452044e+00
-1.45774603e+00 1.28366482e+00 -2.83476979e-01 -1.17025569e-01
6.77974820e-01 -3.70318621e-01 -2.20137715e-01 -1.73092291e-01
-2.11253569e-01 7.40813434e-01 -2.53126621e-01 -1.45511496e+00
-6.43677115e-01 -1.02902465e-01 5.37680209e-01 1.38896912e-01
-1.26313195e-01 9.73240376e-01 -5.72401226e-01 -6.39845014e-01
-1.26312733e-01 -5.75353682e-01 -3.05428177e-01 1.28426822e-02
1.66637257e-01 -4.90089476e-01 -2.78841108e-02 -6.99157596e-01
1.65906382e+00 -1.85563743e+00 2.35093161e-02 -5.06442152e-02
4.36480880e-01 2.68992335e-01 2.17707798e-01 1.08023691e+00
-1.48490697e-01 2.48117849e-01 -1.48377255e-01 -3.39493275e-01
1.11415237e-01 5.09162173e-02 7.22971652e-03 2.89143855e-03
1.21940635e-02 1.07039106e+00 -8.84580195e-01 -5.09896219e-01
4.10298288e-01 8.62820804e-01 -5.28593719e-01 2.16560602e-01
-2.87669688e-01 5.83988607e-01 -5.03562570e-01 3.74596357e-01
2.11027488e-01 2.16401532e-01 2.79268563e-01 6.89511538e-01
-7.68980205e-01 6.67584300e-01 -8.44877541e-01 1.34686589e+00
-5.97519398e-01 7.31621206e-01 2.08851099e-01 -8.24769020e-01
6.37023270e-01 7.29733527e-01 2.15364233e-01 -3.93575341e-01
5.65721281e-02 3.18190694e-01 3.73869985e-01 -5.73039532e-01
1.50652438e-01 -2.30709419e-01 1.97196066e-01 8.06932926e-01
3.65813659e-03 -3.74967128e-01 8.42510909e-02 1.73251957e-01
5.05264461e-01 4.60021734e-01 6.32506967e-01 -3.69482309e-01
6.89242780e-01 1.76500976e-02 7.01704398e-02 5.54025650e-01
2.59571373e-01 1.59097850e-01 4.80927199e-01 -4.94825065e-01
-9.22427416e-01 -8.01151812e-01 -5.80405304e-03 1.23547840e+00
-5.48393786e-01 -6.00423396e-01 -1.21531165e+00 -4.12300587e-01
-4.73540783e-01 1.36804426e+00 -4.71905589e-01 2.53495604e-01
-8.02553236e-01 -3.60499173e-01 4.85269904e-01 4.95744616e-01
-1.64268255e-01 -1.43118656e+00 -7.28561699e-01 3.37801278e-01
2.78440744e-01 -7.69711375e-01 3.40995103e-01 -1.78503338e-02
-7.56498516e-01 -7.22488284e-01 -7.25900948e-01 -1.20667267e+00
7.84082115e-01 -9.31458026e-02 1.29769647e+00 7.07295477e-01
5.27010448e-02 8.31465125e-01 -5.32798409e-01 -7.92012513e-01
-8.87655318e-01 -1.53717041e-01 -1.22046463e-01 -6.89860284e-01
5.00798643e-01 -8.30838203e-01 -4.03656997e-02 -1.90093875e-01
-8.61093760e-01 8.33654225e-01 -1.57038141e-02 3.38575870e-01
-1.35365929e-02 -1.64909109e-01 3.50614637e-01 -1.16159689e+00
7.22229898e-01 -2.05677390e-01 -4.83110249e-01 5.13975620e-01
-4.73460257e-01 -1.79074824e-01 3.27821136e-01 -2.80402839e-01
-1.04684389e+00 1.63022146e-01 -4.80383545e-01 5.27244031e-01
-5.65589011e-01 5.17680526e-01 -2.85626203e-01 4.62278239e-02
2.67586321e-01 2.08192170e-01 -1.29683673e-01 -6.07531190e-01
4.52165604e-01 5.79176724e-01 1.13590255e-01 -7.59721160e-01
6.03592753e-01 -8.51748046e-03 -2.99306095e-01 -1.19134033e+00
-5.03800884e-02 4.52070646e-02 -1.02539349e+00 -3.23861361e-01
7.82696545e-01 -8.95517766e-01 -4.95034933e-01 5.28753400e-01
-1.33628416e+00 -8.68654430e-01 -1.41148388e-01 5.56453168e-01
-5.59679806e-01 2.14432478e-01 -5.26138902e-01 -1.35634792e+00
1.17602296e-01 -8.75554323e-01 5.91705739e-01 7.41233826e-02
-8.33713949e-01 -1.42225540e+00 -1.50582045e-01 2.70755112e-01
6.61920682e-02 8.78768191e-02 1.35551536e+00 -7.83334672e-01
-3.89132142e-01 1.01923048e-01 2.54008710e-01 -7.38287112e-03
-6.04204871e-02 3.38992506e-01 -8.67135227e-01 1.26209155e-01
-6.78300261e-02 -2.11345240e-01 -4.12162170e-02 1.28184766e-01
5.34260869e-01 4.47603278e-02 -3.77367646e-01 8.34890828e-02
1.10350335e+00 6.07879162e-01 3.72451931e-01 2.13435233e-01
6.24498248e-01 1.31136930e+00 5.28084338e-01 -5.50581962e-02
9.27779496e-01 7.57778704e-01 -3.43396634e-01 -3.21857542e-01
-1.62124664e-01 -5.08747220e-01 2.08728433e-01 1.05306411e+00
-2.28749976e-01 -3.01849609e-03 -1.57045043e+00 7.23491907e-01
-1.67450333e+00 -6.27067327e-01 -4.08936113e-01 2.07509136e+00
8.38526607e-01 2.64876395e-01 3.20497483e-01 3.02978694e-01
3.74386758e-01 -3.81078571e-01 1.37652725e-01 -8.25511515e-01
2.45819643e-01 3.01906168e-01 -3.93323749e-01 1.12217402e+00
-4.59517956e-01 1.30277276e+00 7.41731453e+00 2.74632692e-01
-8.18684876e-01 4.15650792e-02 5.42970717e-01 2.33572230e-01
-7.91944802e-01 7.99611807e-02 -5.75566530e-01 1.85671940e-01
9.10837889e-01 -3.93873572e-01 6.62882745e-01 4.66036439e-01
7.44323075e-01 -3.58925283e-01 -1.46416163e+00 4.45761204e-01
-2.93344140e-01 -8.83282661e-01 8.75486135e-02 -1.25060990e-01
1.38524607e-01 -2.95839280e-01 -4.82680231e-01 2.51215369e-01
4.83407944e-01 -1.17994165e+00 8.58779073e-01 3.88698518e-01
7.45750070e-01 -6.55692101e-01 2.32524097e-01 8.35539639e-01
-1.11714566e+00 3.03492218e-01 1.87160030e-01 -7.50472128e-01
3.21676344e-01 2.99570747e-02 -8.09795141e-01 1.56748161e-01
3.00483644e-01 7.81821460e-02 -2.75625974e-01 7.63842762e-01
-5.03055871e-01 7.40947008e-01 -2.49460995e-01 -2.32250378e-01
-2.94156652e-02 -1.16442926e-01 2.21066728e-01 1.47051203e+00
2.82048043e-02 6.36332631e-01 -5.41001819e-02 7.30406165e-01
5.50777972e-01 5.48767567e-01 -9.25449371e-01 -2.66741335e-01
4.30041611e-01 6.14788055e-01 -9.17294264e-01 -3.23161155e-01
-9.52234983e-01 6.43917561e-01 3.53922635e-01 3.47603798e-01
-4.86895859e-01 -1.17854498e-01 5.94086647e-01 5.78590214e-01
-2.60231704e-01 -5.03670156e-01 -4.93437171e-01 -1.02350366e+00
-3.78004014e-01 -9.72338200e-01 -1.04489438e-01 -8.34340215e-01
-4.23889548e-01 3.67825866e-01 5.03427088e-01 -4.63836998e-01
-6.80014968e-01 -3.72570932e-01 -4.50362474e-01 1.01339793e+00
-6.34725988e-01 -1.49943292e+00 2.10551694e-01 7.34835118e-02
7.14541554e-01 1.24813922e-01 1.17466617e+00 1.18584529e-01
-2.72949159e-01 4.40085083e-01 -3.56174380e-01 4.08029445e-02
-1.58749014e-01 -1.09455478e+00 8.70030165e-01 6.99527562e-01
2.79700011e-01 1.07453251e+00 8.74451458e-01 -7.85845280e-01
-1.06611717e+00 -4.80735600e-01 1.66286564e+00 -6.01954162e-01
7.40341425e-01 -6.36457264e-01 -9.56471980e-01 1.01025808e+00
4.37575251e-01 -6.21662915e-01 8.88596654e-01 1.04391769e-01
3.10499161e-01 5.35070956e-01 -9.83649015e-01 8.68928432e-01
1.39513838e+00 -6.39482617e-01 -8.09401810e-01 1.45106748e-01
8.71318698e-01 -2.95135111e-01 -7.97104716e-01 3.21691930e-01
8.27156961e-01 -1.04839087e+00 7.80905366e-01 -6.03577495e-01
3.20434719e-01 -8.84891488e-03 1.70658693e-01 -7.99960732e-01
-1.44310772e-01 -8.50191116e-01 1.69308111e-01 1.72549307e+00
7.57941663e-01 -5.80933392e-01 6.17766559e-01 1.31323397e+00
-2.38369014e-02 -9.62965488e-01 -4.51676339e-01 -2.39547536e-01
5.84767461e-01 -1.13290799e+00 6.36842787e-01 8.91763151e-01
4.25865084e-01 5.12268394e-02 -9.79386568e-02 -1.06967904e-01
2.34334603e-01 -4.43973482e-01 7.75789738e-01 -1.38855982e+00
-2.92710543e-01 -4.72266614e-01 -3.74509096e-02 -9.47842598e-01
1.55505463e-01 -6.92464113e-01 -3.25604707e-01 -1.80841923e+00
-5.14521934e-02 -3.94694835e-01 8.74328464e-02 2.96150237e-01
1.42256007e-01 -1.50289744e-01 3.07902575e-01 -1.69064462e-01
-1.96899727e-01 6.68652132e-02 1.07072854e+00 4.51586992e-01
-5.36923289e-01 1.18051879e-01 -8.64925146e-01 1.23037755e+00
8.65164876e-01 -2.21053123e-01 -7.22334743e-01 -7.45944619e-01
4.52440858e-01 2.38673568e-01 -1.43826026e-02 -6.83129251e-01
1.90439429e-02 -4.35989946e-01 5.75886965e-02 -3.90006334e-01
1.91690177e-01 -8.25148106e-01 3.88844460e-01 5.61267793e-01
-4.94826555e-01 1.42467832e-02 3.33303928e-01 -3.14612359e-01
6.65926486e-02 -4.80486035e-01 4.05323535e-01 -2.83222169e-01
-4.39987868e-01 -3.52583885e-01 -1.19364119e+00 -1.78382725e-01
1.05086172e+00 -5.30092716e-01 4.34856236e-01 -4.38987106e-01
-1.05586803e+00 2.82062888e-01 8.10631931e-01 3.77834141e-01
2.54713297e-01 -8.93154144e-01 -4.66619134e-01 2.42469579e-01
-8.42580944e-02 1.10239191e-02 -2.06685796e-01 5.00752628e-01
-8.09401989e-01 8.02060604e-01 1.51962668e-01 -1.21288143e-01
-1.70998871e+00 3.89510185e-01 3.18082213e-01 4.67514396e-02
-6.01141870e-01 9.35721040e-01 4.85156178e-01 -6.01386011e-01
3.43229473e-01 -4.85273629e-01 -6.91234589e-01 -1.07368730e-01
6.73780024e-01 3.43617857e-01 -3.75476658e-01 -8.87409151e-01
-3.23951721e-01 5.07206321e-01 1.65497288e-01 -5.08247972e-01
1.37875116e+00 -3.55130911e-01 -5.15499115e-01 1.04552555e+00
4.67525542e-01 2.88948387e-01 -6.68379366e-01 8.38272497e-02
3.99712861e-01 -1.82557739e-02 -3.40800911e-01 -8.99261951e-01
-3.30851704e-01 9.65404093e-01 2.43551597e-01 4.87474591e-01
9.99059796e-01 2.96488911e-01 2.29994908e-01 1.13399334e-01
5.47068596e-01 -8.92585814e-01 -6.87307656e-01 3.76597822e-01
8.21411788e-01 -7.71306098e-01 -2.13162601e-02 -9.00641084e-01
-3.68546367e-01 1.15088677e+00 6.46004498e-01 4.11209673e-01
6.40511096e-01 6.07954681e-01 2.30151340e-01 -2.47017920e-01
-1.02249169e+00 -1.48906037e-01 -9.91829205e-03 8.54984701e-01
1.34256399e+00 3.15847814e-01 -8.86310101e-01 6.80859089e-01
-3.67784500e-01 4.68179762e-01 3.02930206e-01 1.05328691e+00
-5.52310526e-01 -1.82211614e+00 -3.32011431e-01 -4.02056612e-02
-5.96854448e-01 -2.05324948e-01 -7.79066980e-01 1.13644814e+00
4.16112155e-01 1.04262245e+00 -2.81974584e-01 -1.16021939e-01
2.88165867e-01 4.15717930e-01 7.07063317e-01 -9.24744606e-01
-9.80118096e-01 -8.93017426e-02 5.23224711e-01 -2.60100007e-01
-3.90207410e-01 -8.06707323e-01 -1.45491076e+00 -3.07865590e-01
-1.28030404e-01 2.64373869e-01 5.08024096e-01 1.05874491e+00
-3.09360802e-01 2.48919979e-01 2.96826400e-02 -5.76342702e-01
2.84879684e-01 -7.20558345e-01 -2.44106561e-01 -3.79182667e-01
1.03304915e-01 -3.58425945e-01 -3.41941193e-02 2.49345452e-01] | [10.390406608581543, 8.772350311279297] |
5e49aca9-f879-42c5-a354-d9db789439de | single-image-dehazing-via-combining-the-prior | 2111.05701 | null | https://arxiv.org/abs/2111.05701v2 | https://arxiv.org/pdf/2111.05701v2.pdf | Single image dehazing via combining the prior knowledge and CNNs | Aiming at the existing single image haze removal algorithms, which are based on prior knowledge and assumptions, subject to many limitations in practical applications, and could suffer from noise and halo amplification. An end-to-end system is proposed in this paper to reduce defects by combining the prior knowledge and deep learning method. The haze image is decomposed into the base layer and detail layers through a weighted guided image filter (WGIF) firstly, and the airlight is estimated from the base layer. Then, the base layer image is passed to the efficient deep convolutional network for estimating the transmission map. To restore object close to the camera completely without amplifying noise in sky or heavily hazy scene, an adaptive strategy is proposed based on the value of the transmission map. If the transmission map of a pixel is small, the base layer of the haze image is used to recover a haze-free image via atmospheric scattering model, finally. Otherwise, the haze image is used. Experiments show that the proposed method achieves superior performance over existing methods. | ['Wangming Xu', 'Shiqian Wu', 'Chaobing Zheng', 'Yuwen Li'] | 2021-11-10 | null | null | null | null | ['image-dehazing', 'single-image-haze-removal'] | ['computer-vision', 'computer-vision'] | [ 3.04795653e-01 -2.34898061e-01 7.03582287e-01 -2.72609830e-01
-2.61965871e-01 1.53102249e-01 -1.02271236e-01 -4.97601300e-01
-2.11131454e-01 4.85909700e-01 1.74987718e-01 -7.75609724e-03
-3.57696302e-02 -1.02014744e+00 -4.93730187e-01 -1.39082587e+00
2.60147959e-01 -2.96532899e-01 5.22928536e-01 -3.29996198e-01
1.21486321e-01 1.01585634e-01 -1.57608283e+00 3.05862278e-01
1.17048216e+00 1.09746742e+00 6.12071037e-01 4.73854810e-01
-8.56828466e-02 1.08097076e+00 -6.24663532e-01 1.42081931e-01
3.86903703e-01 -3.99138361e-01 -2.37574447e-02 4.64247674e-01
4.43354398e-01 -8.94916832e-01 -6.90453410e-01 1.48882461e+00
5.49468756e-01 2.62467563e-01 2.88928539e-01 -8.04083943e-01
-6.31516576e-01 1.07446045e-01 -5.58000386e-01 1.47924781e-01
-4.78595644e-01 3.51117581e-01 2.18296468e-01 -1.02317989e+00
-1.34134628e-02 1.04302549e+00 5.04800975e-01 2.30660319e-01
-5.51434278e-01 -7.38713562e-01 2.53430456e-01 2.07028776e-01
-1.39486647e+00 -4.78604645e-01 8.84837806e-01 -7.70359486e-02
3.97188604e-01 -1.58391297e-02 6.43927932e-01 6.69536218e-02
2.44289309e-01 6.78141057e-01 9.79755402e-01 -3.01335543e-01
-8.68122801e-02 1.79895721e-02 7.87500292e-02 7.82582283e-01
5.89823425e-01 8.36794823e-02 -2.77363271e-01 3.91536146e-01
5.17317653e-01 4.39411074e-01 -6.89032793e-01 2.04056010e-01
-8.05152178e-01 4.74570394e-01 8.19269955e-01 -3.28758662e-03
-6.04948103e-01 -5.30272862e-03 -2.72348136e-01 2.70587146e-01
6.58130646e-01 -9.26228538e-02 -1.48970604e-01 6.63405895e-01
-1.08558881e+00 2.41998032e-01 2.92019516e-01 7.02487767e-01
1.19835114e+00 2.26752475e-01 -1.06767185e-01 9.53510463e-01
8.47656965e-01 8.75097632e-01 -7.45616248e-03 -1.04737628e+00
4.19372410e-01 5.44485450e-01 3.63271803e-01 -9.65244412e-01
-1.09887794e-01 -6.88023984e-01 -1.20135093e+00 7.75571764e-01
-1.19459376e-01 -3.56902778e-01 -1.28698671e+00 1.08040583e+00
3.51571500e-01 4.11067069e-01 2.28433430e-01 1.38796008e+00
1.02272689e+00 1.28089738e+00 -4.49380517e-01 -4.04481649e-01
9.18203533e-01 -1.17339385e+00 -9.76073325e-01 -5.05297422e-01
1.88634261e-01 -1.01976800e+00 5.45433104e-01 6.31208062e-01
-1.19420993e+00 -5.53331733e-01 -1.28238380e+00 -1.84463724e-01
-8.74890685e-02 2.25926101e-01 2.41679654e-01 4.50787008e-01
-1.04654527e+00 2.41601273e-01 -6.31845474e-01 2.17931792e-01
3.14653188e-01 1.62704572e-01 2.05177650e-01 -7.90678501e-01
-1.19465292e+00 7.02135861e-01 4.48386610e-01 8.01366448e-01
-1.14442360e+00 -4.34295207e-01 -7.59861648e-01 9.56836641e-02
1.87199384e-01 -8.20143878e-01 7.96763182e-01 -9.97539818e-01
-1.51691878e+00 2.38966271e-01 -1.82128828e-02 -1.89648926e-01
1.26508191e-01 -4.15286034e-01 -4.68739629e-01 1.67639911e-01
-1.67901561e-01 2.16532961e-01 1.14224076e+00 -1.49144983e+00
-8.36754978e-01 -2.78914154e-01 1.87313966e-02 5.92045605e-01
-1.36711299e-01 -2.24852100e-01 -7.96042323e-01 -3.34067464e-01
4.13566113e-01 -6.17709339e-01 -2.19518870e-01 1.47581533e-01
-3.33189845e-01 3.32939148e-01 1.05899048e+00 -9.28837061e-01
1.08751976e+00 -2.41242409e+00 -7.58678839e-02 6.25764206e-02
3.47848117e-01 6.06559277e-01 -3.01772863e-01 5.96858189e-02
1.34885520e-01 -3.32231373e-01 -5.37693501e-01 -1.30991429e-01
-4.40834016e-01 1.23231774e-02 -1.33503273e-01 7.44413614e-01
2.44884454e-02 4.05002236e-01 -6.15271091e-01 -4.46651161e-01
5.48835039e-01 7.29687810e-01 -3.10292065e-01 6.19493604e-01
-8.00667331e-02 3.31468523e-01 -4.91016477e-01 4.58736539e-01
1.52794254e+00 1.43184736e-01 -4.72851962e-01 -2.13683501e-01
-5.04293621e-01 2.79669743e-02 -1.19295967e+00 1.11419332e+00
-6.09387755e-01 6.60899162e-01 6.74623847e-01 -6.61859393e-01
1.05034781e+00 3.24949883e-02 -2.44385097e-02 -6.50236249e-01
1.18999198e-01 1.30184099e-01 -3.89171876e-02 -8.74360144e-01
1.91249937e-01 -2.18253314e-01 6.52035415e-01 5.47355972e-03
-4.05148089e-01 -3.44372720e-01 -3.02141607e-01 1.33037239e-01
8.80206883e-01 -8.40619877e-02 -3.30797255e-01 -5.34533449e-02
7.88883448e-01 -1.35560066e-01 6.13135815e-01 6.05386972e-01
4.72170254e-03 8.71434093e-01 -2.72542119e-01 -6.38779700e-01
-8.92561972e-01 -1.09818542e+00 5.28968088e-02 6.49997234e-01
7.07621634e-01 2.78992555e-03 -6.78636670e-01 2.17049085e-02
-4.81280744e-01 6.38111711e-01 -2.48064488e-01 -5.35073042e-01
-4.57325965e-01 -1.05858099e+00 -4.85711284e-02 -1.76858768e-01
1.44154549e+00 -9.00617838e-01 -1.81948408e-01 2.12782085e-01
-4.30175126e-01 -8.80805194e-01 -2.24785462e-01 -2.89100200e-01
-8.78993869e-01 -8.82468998e-01 -7.36581802e-01 -7.10638106e-01
6.88953698e-01 1.15132475e+00 7.69622743e-01 7.08880663e-01
-7.64876455e-02 -1.27615467e-01 -2.58251131e-01 -6.58980250e-01
-1.74484283e-01 -5.80460668e-01 -2.61454016e-01 4.34874177e-01
2.17972308e-01 -4.56058294e-01 -1.14039874e+00 1.42440051e-01
-1.12181735e+00 2.40685135e-01 9.49114084e-01 7.62350142e-01
3.73198450e-01 8.85834098e-01 -5.75859137e-02 -6.15767658e-01
1.20041586e-01 -4.95776683e-01 -1.03119862e+00 -1.10291176e-01
-5.78335226e-01 -3.05015892e-01 3.35572153e-01 1.44609392e-01
-1.71315825e+00 -3.95785905e-02 -1.30982280e-01 -4.82625663e-01
-8.70913118e-02 5.23482144e-01 -4.64144558e-01 -2.61247933e-01
2.67179847e-01 5.85892379e-01 5.89699037e-02 -3.95448655e-01
5.08675054e-02 8.56674135e-01 6.54901028e-01 2.05695018e-01
1.21513212e+00 6.77960575e-01 -4.89969961e-02 -1.05238163e+00
-1.02421916e+00 -5.37132382e-01 -3.15631956e-01 -4.37431365e-01
1.04040730e+00 -1.38944960e+00 -4.30998087e-01 9.91521955e-01
-1.24110055e+00 -2.34353572e-01 3.83659542e-01 7.62086093e-01
1.45447195e-01 4.71393049e-01 -6.54454470e-01 -1.15682900e+00
-4.93775219e-01 -1.04891384e+00 9.93743181e-01 5.47533810e-01
1.08710647e+00 -6.26040101e-01 -2.68900871e-01 5.57751179e-01
6.08659089e-01 -1.43593818e-01 6.39537990e-01 4.12423700e-01
-1.39087975e+00 -9.53444541e-02 -5.92609406e-01 8.95139158e-01
2.06220165e-01 -5.59580773e-02 -1.10304618e+00 -2.89043099e-01
4.35403913e-01 8.46965536e-02 1.43175435e+00 5.68214655e-01
1.05231977e+00 -2.18385026e-01 5.85868247e-02 1.11749816e+00
1.78543103e+00 1.72681779e-01 1.13919866e+00 4.97432172e-01
8.71593952e-01 5.43227732e-01 7.61237681e-01 1.62467778e-01
1.94743350e-01 1.95861325e-01 1.02170384e+00 -4.77262646e-01
-5.24461150e-01 3.06593984e-01 5.05857706e-01 8.57245684e-01
-3.35374959e-02 -4.97985840e-01 -4.96008635e-01 5.08574486e-01
-1.64382267e+00 -7.93570518e-01 -5.44479847e-01 2.18936515e+00
4.50230181e-01 1.22503392e-01 -8.09576631e-01 -6.26121685e-02
7.25270331e-01 3.39160800e-01 -4.38906640e-01 1.17966786e-01
-2.22134486e-01 -2.12456435e-01 6.67622805e-01 6.86383128e-01
-8.57048452e-01 8.50100398e-01 5.53941441e+00 6.00117266e-01
-9.35191810e-01 -5.45175606e-03 3.26379955e-01 -1.18523449e-01
-2.50210404e-01 -4.10683900e-02 -6.17700458e-01 6.78119242e-01
5.09186685e-01 3.96287680e-01 7.73083448e-01 2.70756066e-01
5.88250339e-01 -3.91108513e-01 -1.88886315e-01 1.08086038e+00
1.48554966e-01 -9.58708763e-01 1.97023922e-03 -7.88933784e-02
8.62334847e-01 2.39840180e-01 1.38419539e-01 6.69729188e-02
2.53429532e-01 -7.42296576e-01 3.46456468e-01 8.70510459e-01
4.05724734e-01 -8.24475229e-01 1.18478370e+00 5.79410732e-01
-8.48024070e-01 -6.29905760e-02 -9.44922328e-01 -1.95463210e-01
1.05569020e-01 1.34357548e+00 -3.50967199e-01 6.28168464e-01
1.00788188e+00 5.20517707e-01 -2.74006426e-01 1.28765440e+00
-5.14943957e-01 8.14673901e-01 -3.95228446e-01 4.88550693e-01
3.56124222e-01 -6.38643801e-01 5.43300331e-01 9.78316367e-01
5.40434182e-01 5.72565198e-01 1.04178868e-01 7.82246768e-01
-6.74619973e-02 -3.20990562e-01 -5.35679221e-01 4.76706117e-01
1.32812425e-01 1.29681730e+00 -2.70409971e-01 -4.06631649e-01
-6.95020854e-01 1.11928725e+00 -1.87006399e-01 8.08400691e-01
-5.98084688e-01 -6.64174139e-01 4.48670954e-01 -2.10363325e-02
3.64095956e-01 -1.26969352e-01 9.58733726e-03 -1.01341474e+00
-1.34398108e-02 -6.25740528e-01 1.82997584e-02 -1.36386716e+00
-1.01936185e+00 3.38293225e-01 -3.29072446e-01 -1.24171865e+00
6.27116978e-01 -6.51827097e-01 -9.52643812e-01 1.31187689e+00
-2.15223956e+00 -9.98257875e-01 -1.02597833e+00 6.85183942e-01
5.33596396e-01 -1.25177216e-03 1.91832244e-01 5.25970221e-01
-6.38609946e-01 -1.00063324e-01 4.48451519e-01 -1.39862806e-01
6.19854629e-01 -9.09310222e-01 -5.33601753e-02 1.37586761e+00
-5.95847249e-01 3.39417905e-01 7.53790617e-01 -7.14071929e-01
-1.22963011e+00 -1.53677619e+00 4.55635875e-01 9.36079323e-02
5.95437363e-02 -6.90026060e-02 -1.27612007e+00 4.29738760e-01
3.36935550e-01 3.48177701e-01 1.36823684e-01 -5.17969429e-01
-1.15605053e-02 -5.18377602e-01 -9.15892184e-01 4.63811219e-01
5.82612216e-01 -2.23566785e-01 -4.54345018e-01 4.85858649e-01
1.00250936e+00 -4.60448325e-01 -4.10653293e-01 4.75820243e-01
2.64981747e-01 -1.23468924e+00 9.23783004e-01 9.19961780e-02
4.25495803e-01 -9.60501850e-01 -8.34468156e-02 -1.49574268e+00
-7.12018847e-01 -2.69863069e-01 9.19453502e-02 1.14777565e+00
2.09399253e-01 -4.62711245e-01 6.29971266e-01 1.54656827e-01
-7.01673627e-01 -3.00417513e-01 -3.61753315e-01 -2.06380472e-01
-3.58736724e-01 3.29052331e-03 6.63092852e-01 6.40928030e-01
-9.38900352e-01 3.67369443e-01 -7.08685577e-01 1.19361627e+00
1.09654582e+00 1.68359995e-01 6.88822985e-01 -1.06775081e+00
7.57067725e-02 1.06388099e-01 -1.25878587e-01 -1.19455349e+00
-1.78935274e-01 -4.80697840e-01 6.99933827e-01 -1.99850261e+00
1.39040366e-01 -3.99134718e-02 -5.32749057e-01 -3.05424500e-02
-6.39795184e-01 3.94079536e-01 6.73780963e-03 4.62389857e-01
-5.63387990e-01 7.87357152e-01 1.54899883e+00 -4.86629874e-01
-1.00712933e-01 1.45590737e-01 -4.16141093e-01 8.05547476e-01
7.57511139e-01 -4.19962019e-01 -4.77426291e-01 -1.00396776e+00
3.87185439e-02 4.86220866e-02 4.20924217e-01 -1.17406285e+00
5.73111713e-01 -5.34075014e-02 8.10793042e-01 -8.82299185e-01
3.99051309e-01 -8.78244042e-01 -1.08012170e-01 3.22301775e-01
1.02785230e-01 -6.75347865e-01 2.16923393e-02 6.72242343e-01
-4.12397563e-01 -3.34617436e-01 1.00094962e+00 -2.43687123e-01
-7.04810262e-01 5.79063952e-01 -4.42760915e-01 -5.02071679e-01
5.22079885e-01 -2.12391675e-01 -4.58866358e-01 -5.17549813e-01
-2.64523864e-01 4.42280889e-01 3.56790453e-01 4.78959158e-02
1.14702463e+00 -9.88914192e-01 -1.06370115e+00 3.31432223e-01
9.01077390e-02 4.60652500e-01 7.28388190e-01 7.80107915e-01
-8.94030988e-01 -1.19687051e-01 5.71491383e-02 -4.21482980e-01
-1.20965767e+00 5.96946001e-01 6.99466705e-01 2.11390197e-01
-8.09058666e-01 8.31020713e-01 6.41011178e-01 -2.61685371e-01
6.91212639e-02 -6.58826306e-02 -1.56086758e-01 -6.20357990e-01
8.36435914e-01 4.47308034e-01 1.68577403e-01 -5.60462832e-01
3.20046172e-02 7.27131367e-01 1.18430294e-02 9.38545540e-02
1.38304448e+00 -7.42907047e-01 -5.92332184e-01 3.85723971e-02
1.05439663e+00 1.69232879e-02 -1.38708496e+00 -4.14922118e-01
-7.98313379e-01 -7.29320824e-01 9.20697153e-01 -6.68432593e-01
-1.59058499e+00 1.11923778e+00 9.56574917e-01 -1.07265174e-01
1.62105334e+00 -4.30163711e-01 1.03117263e+00 6.03958070e-01
3.29407421e-03 -6.94911480e-01 -4.88486663e-02 3.97205830e-01
5.60950279e-01 -1.29134297e+00 3.03591728e-01 -3.72895867e-01
-2.31626183e-01 1.10293841e+00 7.75918663e-01 -1.73091188e-01
7.67030597e-01 1.15824625e-01 3.89115095e-01 -4.45003152e-01
-4.58402574e-01 -1.79134548e-01 -7.64832720e-02 5.83652139e-01
-3.93125676e-02 -4.07080710e-01 1.78415492e-01 1.47606656e-01
2.35343248e-01 8.41930658e-02 6.66541278e-01 4.92694229e-01
-1.28893757e+00 -3.61166686e-01 -7.86700368e-01 3.26615900e-01
-3.21927607e-01 -4.33882475e-01 1.15460753e-01 3.32404345e-01
5.89060009e-01 1.39209330e+00 7.47835040e-02 -2.68733859e-01
1.24390528e-01 -5.05797088e-01 1.93185568e-01 -5.21969020e-01
-8.62230584e-02 3.08730930e-01 -1.69176623e-01 -3.54143828e-01
-4.47991818e-01 -6.62662834e-02 -1.20332265e+00 -1.92116365e-01
-5.86713672e-01 1.01642534e-01 4.72823650e-01 7.52538860e-01
3.51119712e-02 6.77916884e-01 8.85117650e-01 -8.54719400e-01
-4.00316939e-02 -1.03654468e+00 -1.05927444e+00 1.42117977e-01
9.28669989e-01 -4.39997405e-01 -7.07131743e-01 6.41804859e-02] | [10.885265350341797, -3.1652872562408447] |
907c7e1c-418d-4185-90b9-2b85335a4d6a | multimodality-multi-lead-ecg-arrhythmia | 2210.06297 | null | https://arxiv.org/abs/2210.06297v1 | https://arxiv.org/pdf/2210.06297v1.pdf | Multimodality Multi-Lead ECG Arrhythmia Classification using Self-Supervised Learning | Electrocardiogram (ECG) signal is one of the most effective sources of information mainly employed for the diagnosis and prediction of cardiovascular diseases (CVDs) connected with the abnormalities in heart rhythm. Clearly, single modality ECG (i.e. time series) cannot convey its complete characteristics, thus, exploiting both time and time-frequency modalities in the form of time-series data and spectrogram is needed. Leveraging the cutting-edge self-supervised learning (SSL) technique on unlabeled data, we propose SSL-based multimodality ECG classification. Our proposed network follows SSL learning paradigm and consists of two modules corresponding to pre-stream task, and down-stream task, respectively. In the SSL-pre-stream task, we utilize self-knowledge distillation (KD) techniques with no labeled data, on various transformations and in both time and frequency domains. In the down-stream task, which is trained on labeled data, we propose a gate fusion mechanism to fuse information from multimodality.To evaluate the effectiveness of our approach, ten-fold cross validation on the 12-lead PhysioNet 2020 dataset has been conducted. | ['Ngan Le', 'Morten Olgaard Jensen', 'Jingxian Wu', 'Donald Adjeroh', 'Patel Brijesh', 'Duc Le', 'Thinh Phan'] | 2022-09-30 | null | null | null | null | ['self-knowledge-distillation', 'ecg-classification'] | ['computer-vision', 'medical'] | [ 4.25625771e-01 -1.25653833e-01 -3.88378315e-02 -4.91740972e-01
-8.54462445e-01 -4.29112613e-01 3.40690196e-01 5.40995836e-01
-2.51622021e-01 7.97231734e-01 1.62015989e-01 -3.35425317e-01
-5.06741345e-01 -5.81274867e-01 -1.55820489e-01 -7.19888985e-01
-2.13022381e-01 1.12688541e-03 -3.85841012e-01 -6.69749230e-02
-4.25207131e-02 1.14219964e-01 -1.14750648e+00 4.05745685e-01
9.57445800e-01 1.09113789e+00 -2.27313444e-01 6.93707705e-01
-1.15746267e-01 7.17189550e-01 -3.72118324e-01 -1.91081062e-01
-4.67772931e-02 -8.55756879e-01 -4.98541713e-01 -1.20611228e-02
-1.73649117e-01 3.01741153e-01 -1.83577523e-01 9.92427945e-01
9.23126757e-01 -2.67447680e-01 6.30497515e-01 -1.14556396e+00
-5.26460819e-02 6.63054764e-01 -5.72811484e-01 2.57112473e-01
3.30791593e-01 -1.21633284e-01 5.94884455e-01 -8.51017356e-01
3.23894083e-01 8.75543594e-01 7.22478926e-01 2.41715237e-01
-1.13065720e+00 -5.34564137e-01 -2.39088625e-01 2.62987256e-01
-1.42748916e+00 -3.09663862e-01 1.37556577e+00 -3.63658339e-01
2.39413217e-01 3.25962871e-01 7.67962933e-01 1.04780066e+00
5.00724792e-01 4.27411586e-01 1.31716430e+00 -4.92699355e-01
2.83394814e-01 1.33896351e-01 1.98109150e-01 5.05830169e-01
8.69450197e-02 1.38077410e-02 -7.29128063e-01 -3.04551452e-01
5.50800264e-01 1.51659936e-01 -1.89534992e-01 -1.71485215e-01
-1.55278802e+00 3.68025362e-01 2.13423632e-02 4.29711848e-01
-4.37574685e-01 -3.21482569e-01 7.40835845e-01 7.20487654e-01
3.42796743e-01 1.27589434e-01 -4.95034695e-01 -1.73816785e-01
-8.84742796e-01 -1.65690720e-01 7.57288873e-01 5.66193700e-01
5.83062887e-01 -3.86425965e-02 -1.46480247e-01 6.26363575e-01
4.06177908e-01 6.18842065e-01 8.41588378e-01 -6.29181445e-01
6.06365085e-01 8.46879482e-01 -2.25982442e-01 -9.66153920e-01
-6.65418983e-01 -8.44217479e-01 -1.51178062e+00 -9.02124643e-02
1.73192635e-01 -6.72190666e-01 -7.30422556e-01 1.89557326e+00
3.74242038e-01 5.08537769e-01 4.08273190e-01 8.31882060e-01
1.11056244e+00 4.47544932e-01 2.50849515e-01 -7.91061282e-01
1.48620892e+00 -2.24238381e-01 -8.88342083e-01 2.58320630e-01
7.19181836e-01 -5.55937827e-01 4.67684984e-01 5.66603065e-01
-7.19964266e-01 -8.28242064e-01 -1.22960877e+00 2.74637163e-01
-3.72302741e-01 4.01396394e-01 2.08632335e-01 7.52020836e-01
-5.66401243e-01 6.15602851e-01 -6.29575372e-01 -2.28486478e-01
3.01731080e-01 1.38071269e-01 -7.47264862e-01 2.57916376e-02
-1.53060949e+00 7.59952307e-01 5.04704118e-01 2.49166146e-01
-5.51269650e-01 -5.35478055e-01 -7.69926012e-01 -1.67053223e-01
2.25512728e-01 -6.96267664e-01 5.27612984e-01 -7.71777570e-01
-1.14185953e+00 6.40991449e-01 4.64611761e-02 -3.40927362e-01
3.15770149e-01 -3.96709144e-02 -9.97845650e-01 2.78557748e-01
-9.42692235e-02 1.69097736e-01 1.09643054e+00 -1.07580113e+00
-5.27556241e-01 -7.64103830e-01 -4.58407253e-01 3.10924739e-01
-5.27999461e-01 -3.20168614e-01 2.69062892e-02 -7.20123053e-01
3.69061589e-01 -6.97726488e-01 -7.09767407e-03 -3.53077620e-01
-5.09405434e-01 -2.13963948e-02 8.75398755e-01 -7.49661684e-01
1.61156905e+00 -2.40728211e+00 3.23507220e-01 4.79936987e-01
5.16243875e-01 2.82163978e-01 2.92083118e-02 5.73634088e-01
-5.24967134e-01 -2.66954809e-01 -5.08163393e-01 -2.69107819e-01
-1.48824111e-01 1.54667735e-01 -1.06060363e-01 3.69478703e-01
2.53864735e-01 8.47987950e-01 -8.21380198e-01 -7.86840081e-01
2.97417462e-01 4.32011276e-01 4.11141366e-02 1.09224088e-01
3.06423873e-01 1.09223902e+00 -4.49081361e-01 4.92802620e-01
4.47093576e-01 -2.52178490e-01 3.72967541e-01 -8.59539449e-01
-4.03502807e-02 -3.41342747e-01 -1.14847195e+00 2.23000216e+00
-2.18906298e-01 1.35747939e-01 -3.75740618e-01 -1.19357574e+00
9.68929708e-01 7.81347394e-01 8.42074275e-01 -5.64808846e-01
2.66912073e-01 1.80536628e-01 9.81214643e-02 -9.43079293e-01
-2.83667296e-01 -2.95164049e-01 -1.38002649e-01 3.88965994e-01
3.26041490e-01 3.62184048e-01 1.16136394e-01 9.90053564e-02
1.07873523e+00 1.18339024e-01 5.37264466e-01 -6.28805831e-02
8.42149973e-01 -2.92153627e-01 4.92947429e-01 4.24736440e-01
-1.29738003e-01 5.62065542e-01 5.32899320e-01 -3.43460172e-01
-7.14956105e-01 -8.43904197e-01 -2.29710385e-01 4.72512513e-01
8.43263045e-02 -5.79859674e-01 -3.48518312e-01 -6.31297231e-01
-1.66547671e-01 1.39579266e-01 -6.44979358e-01 -4.14741486e-01
-2.32862890e-01 -1.06042254e+00 8.06509972e-01 4.96021003e-01
5.00791967e-01 -7.95226097e-01 -8.01923096e-01 3.02181095e-01
-2.19516292e-01 -8.77477169e-01 -6.95298016e-02 3.35044891e-01
-1.04773605e+00 -1.21910048e+00 -9.28201675e-01 -5.49117088e-01
4.61550862e-01 -2.37978101e-01 7.76204228e-01 -2.50939339e-01
-4.51737970e-01 5.39456129e-01 -4.86684442e-01 -4.25475329e-01
-2.72298098e-01 -2.97950823e-02 -1.05669245e-01 7.92867482e-01
3.11761051e-01 -9.59516346e-01 -7.30361104e-01 -8.09396431e-02
-8.61451745e-01 4.14495580e-02 8.25754046e-01 8.57441664e-01
4.55194682e-01 8.99490044e-02 1.17652380e+00 -8.92371416e-01
7.61925220e-01 -6.32745624e-01 5.78111224e-02 4.11833048e-01
-6.71254218e-01 -1.94810688e-01 4.80066955e-01 -3.75126392e-01
-9.44938123e-01 2.44095013e-01 3.60683538e-02 -4.52613860e-01
-3.31803232e-01 1.04454660e+00 -2.70068675e-01 1.11246660e-01
7.11292446e-01 4.49111342e-01 1.70977399e-01 -4.58811700e-01
2.89045602e-01 9.64515209e-01 7.40276277e-01 -5.03484488e-01
6.50853395e-01 3.95928055e-01 3.27817142e-01 -5.69082499e-01
-6.51085198e-01 -5.21414280e-01 -7.13974416e-01 -4.10742044e-01
9.78672147e-01 -1.00609291e+00 -5.92637897e-01 4.95449811e-01
-7.93908119e-01 4.88168478e-01 -3.71718049e-01 9.02175069e-01
-3.28657717e-01 5.88784158e-01 -3.76048863e-01 -9.54171300e-01
-5.89035988e-01 -5.56045473e-01 9.24115598e-01 1.74597129e-01
-1.38732925e-01 -9.98771071e-01 2.60288537e-01 1.35620028e-01
2.43903384e-01 6.40073180e-01 1.13063014e+00 -8.58667612e-01
8.52296501e-02 -3.36767554e-01 -1.26710787e-01 5.43586969e-01
4.54686433e-01 -5.67774415e-01 -9.93041456e-01 -1.93463296e-01
3.90398413e-01 -2.40060836e-01 5.93883395e-01 1.01163432e-01
7.86390066e-01 1.99228078e-01 -3.35257016e-02 4.16488260e-01
1.35594404e+00 3.42119277e-01 5.26434481e-01 -3.56822073e-01
8.67712080e-01 5.07410169e-01 4.47588742e-01 6.76936328e-01
6.07988954e-01 1.29797578e-01 1.92374513e-02 -4.34424043e-01
7.00658709e-02 3.92538292e-04 1.48728549e-01 1.24401426e+00
-2.68850863e-01 1.10821933e-01 -9.27220285e-01 2.68281341e-01
-2.01245737e+00 -8.35115492e-01 -5.98679706e-02 2.34037280e+00
8.23824227e-01 6.70748800e-02 7.85563439e-02 8.38041425e-01
6.42102182e-01 -1.31934196e-01 -6.90779805e-01 2.03999370e-01
-2.19158545e-01 2.18138486e-01 4.54206504e-02 2.99269836e-02
-1.19986653e+00 -3.46041210e-02 4.98569107e+00 7.48141885e-01
-1.35373712e+00 2.00579390e-01 5.04162312e-01 3.44296694e-01
-1.55367926e-01 -1.21799156e-01 -3.55795138e-02 5.02532303e-01
1.00838721e+00 -1.55079007e-01 1.41340703e-01 1.70090839e-01
2.68132806e-01 2.32051108e-02 -1.16009212e+00 1.46975613e+00
2.68138349e-01 -9.34214056e-01 -1.65645376e-01 -3.16638172e-01
3.06230903e-01 -2.66883254e-01 -1.61249220e-01 2.98713982e-01
-6.54795349e-01 -8.36480975e-01 3.66481841e-01 8.87606561e-01
9.27415729e-01 -6.22191966e-01 1.06124556e+00 3.71474504e-01
-1.39161885e+00 -1.12795502e-01 2.88239151e-01 3.23932409e-01
1.85491756e-01 8.05701613e-01 -2.23991230e-01 1.51994932e+00
3.94159734e-01 1.22683406e+00 -5.88511407e-01 9.84197617e-01
2.16102958e-01 6.34701371e-01 -1.11220188e-01 5.09096265e-01
-3.09627920e-01 -2.01877713e-01 5.01281917e-01 9.97610331e-01
3.88884574e-01 2.99367100e-01 2.42244378e-01 4.49407160e-01
2.67735302e-01 3.83845419e-01 -4.59430218e-01 -1.11346640e-01
2.82007605e-01 1.36679196e+00 -5.99015415e-01 -5.28087676e-01
-4.13613081e-01 6.20207667e-01 -3.21737796e-01 3.35187316e-01
-7.83833504e-01 -6.33040845e-01 -3.58344227e-01 -1.88241318e-01
-1.59223199e-01 -2.98141455e-03 -3.61704081e-01 -1.27111924e+00
7.66736642e-02 -9.84459162e-01 5.88334858e-01 -6.15501583e-01
-1.64634681e+00 7.01603115e-01 5.43089882e-02 -1.74713230e+00
-7.41157383e-02 -1.66484565e-01 -4.71434742e-01 9.80733037e-01
-1.49513280e+00 -1.39044631e+00 -4.11779851e-01 9.91279721e-01
8.73997062e-02 -3.71546268e-01 1.07391381e+00 9.04747844e-01
-5.48314333e-01 3.63134682e-01 -2.89941102e-01 1.77709371e-01
8.41361523e-01 -1.17568874e+00 -3.87952924e-01 4.73081201e-01
4.15102541e-02 6.31380916e-01 2.52904415e-01 -6.85761571e-01
-1.49978614e+00 -1.10392833e+00 7.65044391e-01 -7.10942969e-02
2.91856557e-01 8.51760507e-02 -7.97358632e-01 1.13608085e-01
4.43518646e-02 1.36549026e-01 1.07481790e+00 -4.99396957e-02
-1.54476866e-01 -4.78293598e-01 -9.30434048e-01 1.36936694e-01
5.78977704e-01 -7.71644056e-01 -7.63674557e-01 7.97324702e-02
2.28743494e-01 -2.82683194e-01 -1.40357316e+00 7.76293695e-01
6.01173878e-01 -6.39770687e-01 8.24613273e-01 -5.58909416e-01
2.71140516e-01 -6.12061977e-01 -1.54732093e-01 -1.14884984e+00
-3.58122811e-02 -6.84465647e-01 -1.84164912e-01 1.21636772e+00
4.16535437e-01 -6.52029037e-01 3.84371936e-01 6.99722394e-02
-8.43437240e-02 -7.03831077e-01 -1.04379952e+00 -3.79457861e-01
-4.39204931e-01 -4.09873039e-01 4.45712239e-01 1.16908884e+00
4.31609690e-01 6.55395269e-01 -6.19810581e-01 1.97929069e-02
8.78840625e-01 1.54170126e-01 3.41924936e-01 -1.38638115e+00
-3.52760196e-01 1.09337181e-01 -4.98662472e-01 -2.41430163e-01
-2.76334822e-01 -9.59392548e-01 -3.84127319e-01 -1.25380766e+00
2.17908263e-01 -2.80648828e-01 -9.76725459e-01 4.94615078e-01
-1.77497342e-01 4.07466590e-01 1.56308040e-01 1.87717065e-01
-5.99514663e-01 5.03484607e-01 1.13763785e+00 2.92603951e-03
-2.79473037e-01 -4.52811234e-02 -5.47682822e-01 6.20468915e-01
7.28055477e-01 -4.16262954e-01 -8.26869011e-01 5.01171798e-02
2.76051372e-01 7.08950698e-01 4.23189461e-01 -1.14846385e+00
3.22002828e-01 3.38842273e-01 5.37720382e-01 -6.63821876e-01
3.36749315e-01 -9.11960304e-01 4.05562639e-01 4.60402340e-01
-4.46111649e-01 2.27766752e-01 2.41377391e-02 7.18484879e-01
-5.39268076e-01 1.64163649e-01 3.46970260e-01 -3.38587612e-02
-2.73533642e-01 2.11266175e-01 -1.79083630e-01 -3.96730825e-02
9.95807707e-01 -4.25535031e-02 -1.20884486e-01 -2.43649170e-01
-1.29013646e+00 1.58083647e-01 -3.03470850e-01 3.41908097e-01
7.22112238e-01 -1.37264931e+00 -8.52290213e-01 4.00064439e-01
3.42204422e-01 -3.26671034e-01 6.45676911e-01 1.37577069e+00
-1.56573653e-01 1.10594034e-01 -3.60298783e-01 -7.16844440e-01
-1.07325983e+00 3.97819430e-01 1.53200805e-01 -2.34208494e-01
-5.78089237e-01 2.12247595e-01 -2.63867497e-01 -1.96648523e-01
2.19663367e-01 -1.04564480e-01 -6.56181097e-01 4.57212508e-01
3.88472080e-01 2.58027762e-01 1.83887526e-01 -4.53743190e-01
-3.81482899e-01 7.23605812e-01 3.22998762e-01 -3.81294489e-01
1.05715370e+00 -2.08370388e-01 -2.34878734e-01 9.23937201e-01
1.00240564e+00 -2.34198600e-01 -4.03453022e-01 -2.46902525e-01
1.72804482e-02 1.32140324e-01 -1.63811129e-02 -9.39725697e-01
-9.24349010e-01 1.02561855e+00 9.94594872e-01 2.36379996e-01
1.52749693e+00 -4.66771007e-01 7.14739501e-01 1.55493289e-01
2.86319107e-01 -8.71443748e-01 -9.46353376e-02 -1.33977234e-01
5.51854193e-01 -1.17137384e+00 1.40884882e-02 -2.89636821e-01
-7.51942098e-01 1.15006900e+00 1.07947990e-01 1.73270658e-01
1.13280213e+00 1.05939418e-01 3.45273435e-01 -2.78979152e-01
-7.32581556e-01 -5.76048680e-02 4.54383790e-01 3.80486816e-01
4.10820872e-01 5.44706881e-02 -6.21291041e-01 8.44043493e-01
2.79606402e-01 3.59544188e-01 2.57250778e-02 1.12100649e+00
-2.33204048e-02 -1.27606094e+00 -4.29252386e-01 4.66418713e-01
-4.64657784e-01 8.07042271e-02 -2.93505043e-02 2.34405622e-01
5.12931466e-01 1.16438925e+00 -4.88897502e-01 -7.75704443e-01
2.56193012e-01 3.81972224e-01 4.48106438e-01 -3.58961076e-01
-6.88481569e-01 3.08336288e-01 -4.01502289e-02 -2.66097277e-01
-8.78316998e-01 -5.03474474e-01 -1.09583795e+00 3.36456507e-01
-2.15516269e-01 1.94752216e-01 6.11056983e-01 1.15168738e+00
4.51018959e-01 8.73734891e-01 8.39149594e-01 -3.95632118e-01
-5.12825847e-01 -1.00931072e+00 -6.76775992e-01 5.81229568e-01
3.94522339e-01 -3.03065419e-01 -1.22129038e-01 4.19004977e-01] | [14.234299659729004, 3.2613325119018555] |
0944894a-6286-422b-8daa-7e7fad576700 | end-to-end-neural-sentence-ordering-using | 1611.04953 | null | http://arxiv.org/abs/1611.04953v2 | http://arxiv.org/pdf/1611.04953v2.pdf | End-to-End Neural Sentence Ordering Using Pointer Network | Sentence ordering is one of important tasks in NLP. Previous works mainly
focused on improving its performance by using pair-wise strategy. However, it
is nontrivial for pair-wise models to incorporate the contextual sentence
information. In addition, error prorogation could be introduced by using the
pipeline strategy in pair-wise models. In this paper, we propose an end-to-end
neural approach to address the sentence ordering problem, which uses the
pointer network (Ptr-Net) to alleviate the error propagation problem and
utilize the whole contextual information. Experimental results show the
effectiveness of the proposed model. Source codes and dataset of this paper are
available. | ['Xinchi Chen', 'Xuanjing Huang', 'Xipeng Qiu', 'Jingjing Gong'] | 2016-11-15 | null | null | null | null | ['sentence-ordering'] | ['natural-language-processing'] | [ 2.01454565e-01 5.52786700e-02 2.09955111e-01 -8.44785810e-01
-6.81600511e-01 -4.14752185e-01 1.62315313e-02 2.54540414e-01
-6.11743033e-01 6.34472847e-01 5.21127462e-01 -4.11205173e-01
-1.07860319e-01 -6.28354847e-01 -5.74056387e-01 -1.83812425e-01
1.43161923e-01 2.17497632e-01 3.36237252e-01 -3.26083511e-01
7.25150883e-01 1.48253053e-01 -9.72414076e-01 6.19711936e-01
1.16370726e+00 7.21658111e-01 5.06830692e-01 6.72326028e-01
-3.90358269e-01 8.09497297e-01 -6.36370242e-01 -3.76555443e-01
4.42010574e-02 -4.52164739e-01 -9.51760113e-01 -3.26129138e-01
1.17188312e-01 -3.70333076e-01 -2.43837178e-01 1.14832556e+00
7.86328018e-01 4.06483144e-01 2.07157999e-01 -7.68307388e-01
-7.01523542e-01 1.02555490e+00 -6.67173088e-01 4.43180919e-01
5.08081675e-01 1.44072389e-03 1.09976959e+00 -9.62167919e-01
3.03080231e-01 1.23501337e+00 6.81292355e-01 2.75132447e-01
-5.71016967e-01 -4.15131092e-01 3.81390184e-01 4.31195766e-01
-1.08476079e+00 -3.67735207e-01 1.00241625e+00 -3.97227556e-02
1.53631759e+00 2.41775766e-01 3.56107444e-01 6.55804217e-01
2.49290287e-01 1.17322612e+00 7.48935223e-01 -6.17743671e-01
-1.29819646e-01 -1.78787544e-01 6.98182940e-01 5.42263567e-01
-4.55460511e-02 -1.19000830e-01 -4.68498915e-01 1.51972592e-01
3.73371810e-01 2.17687041e-02 -2.84992367e-01 3.27120841e-01
-8.72651041e-01 4.33207214e-01 5.73662102e-01 5.06316662e-01
-4.18565035e-01 -1.46248266e-01 6.22510493e-01 2.83325374e-01
5.01757324e-01 6.10973120e-01 -7.54969835e-01 -2.73258060e-01
-1.07381320e+00 1.31925136e-01 7.64648020e-01 9.33965087e-01
4.09494191e-01 -1.80952936e-01 -5.36318600e-01 7.77930439e-01
4.80264336e-01 -2.60130107e-01 6.36996210e-01 -6.13883436e-01
1.00952482e+00 6.91744864e-01 9.50913429e-02 -1.23287451e+00
-7.20574796e-01 -4.63739067e-01 -8.41050625e-01 -5.26012182e-01
1.11999378e-01 -3.45739156e-01 -8.29586148e-01 1.53758585e+00
1.41696647e-01 2.50419378e-01 1.27350420e-01 9.57227170e-01
1.15428996e+00 8.76202226e-01 -5.54299075e-03 -1.60389885e-01
1.31519318e+00 -1.46921813e+00 -1.01666141e+00 -2.88624018e-01
7.79614866e-01 -8.67901862e-01 1.13579261e+00 1.41699240e-01
-1.31015134e+00 -5.85547268e-01 -9.30691361e-01 -4.74300385e-01
-1.35353923e-01 1.86603859e-01 3.87316018e-01 3.24475884e-01
-9.09511626e-01 7.50172496e-01 -8.62027228e-01 -3.74558240e-01
1.23326816e-01 4.28440362e-01 -1.09760806e-01 -3.19687240e-02
-1.70803487e+00 9.00556505e-01 8.03765237e-01 8.07131112e-01
-3.06793571e-01 -4.78169411e-01 -8.41175199e-01 4.21768248e-01
1.21851273e-01 -4.96961355e-01 1.46795845e+00 -5.57534516e-01
-1.49193680e+00 2.28902653e-01 -5.44222772e-01 -3.81482303e-01
3.17167222e-01 -4.99557495e-01 -3.33767444e-01 3.27363163e-02
-3.99583131e-02 4.75329787e-01 8.52051601e-02 -7.02511668e-01
-5.23159027e-01 -2.03070268e-01 3.15438122e-01 5.84315300e-01
-4.62854385e-01 3.82592708e-01 -4.96346116e-01 -5.25956988e-01
1.64646879e-01 -5.08470476e-01 -4.73897219e-01 -4.94625986e-01
-6.43207669e-01 -6.00049138e-01 4.58205849e-01 -1.15508401e+00
1.75879395e+00 -2.12943864e+00 -1.32497966e-01 -2.16644794e-01
-1.47096932e-01 3.94190490e-01 -2.83401966e-01 7.98564672e-01
4.33275998e-02 3.63098383e-01 -4.17193115e-01 -3.92908335e-01
-5.93729597e-03 -3.82465050e-02 -6.82202131e-02 4.54915464e-02
4.53261524e-01 8.61122906e-01 -9.43194211e-01 -6.54506683e-01
-5.56308813e-02 2.31766015e-01 -5.49255192e-01 3.42497319e-01
-1.25986293e-01 3.68112683e-01 -3.04432929e-01 3.71224254e-01
7.95181274e-01 -1.10470675e-01 1.77199438e-01 -2.74122000e-01
-9.60678235e-02 7.99476981e-01 -9.69392836e-01 1.82487428e+00
-3.59572083e-01 4.01522636e-01 -1.52102768e-01 -8.44008684e-01
8.82732689e-01 3.89313847e-01 1.27648935e-01 -7.06040919e-01
1.92947283e-01 4.69080471e-02 2.93836594e-01 -9.52953041e-01
9.23379123e-01 -4.29259129e-02 -5.67140430e-02 2.34724522e-01
-1.91615656e-01 3.16516668e-01 3.94769788e-01 1.87266290e-01
1.16077363e+00 2.63629586e-01 2.99068302e-01 9.83623415e-02
5.83764434e-01 4.00986411e-02 8.49691153e-01 6.19050384e-01
-4.31905895e-01 6.48903549e-01 5.00567496e-01 -3.81839097e-01
-7.77672708e-01 -5.33665657e-01 -4.94218022e-02 9.52833533e-01
6.95869932e-03 -3.73280078e-01 -6.33130491e-01 -7.66892731e-01
-4.50538844e-01 8.76063406e-01 -2.63277948e-01 -9.09041017e-02
-8.99339080e-01 -7.28473663e-01 5.57449043e-01 7.49542296e-01
7.81685412e-01 -1.19028270e+00 -3.50012511e-01 5.26038945e-01
-6.91414654e-01 -1.00519121e+00 -6.19478524e-01 8.43269154e-02
-9.43917096e-01 -7.62952268e-01 -4.82779980e-01 -1.09461653e+00
7.22630382e-01 2.29904681e-01 1.00587821e+00 1.74383894e-01
2.78736264e-01 -4.42320079e-01 -6.64979041e-01 -7.62307271e-02
1.13391474e-01 3.52264643e-01 -2.03706414e-01 -2.32976496e-01
6.22667015e-01 -3.84517223e-01 -6.45761728e-01 3.19497250e-02
-6.44711435e-01 6.07880540e-02 6.45595014e-01 8.69563580e-01
3.33984017e-01 2.49463558e-01 6.61921620e-01 -1.17945862e+00
1.20786548e+00 -4.56866711e-01 -1.79995045e-01 4.01890278e-01
-5.09278595e-01 3.29947807e-02 8.32223892e-01 -6.74667358e-02
-1.33557785e+00 5.69792539e-02 -6.30641580e-01 -5.05073108e-02
-2.43005380e-01 1.04361153e+00 -3.07406306e-01 3.19481492e-01
1.74315155e-01 2.20485538e-01 -6.02557361e-01 -7.73652911e-01
1.78063929e-01 9.11099792e-01 3.49099845e-01 -3.00936401e-01
1.42409965e-01 -2.32844666e-01 -3.22507054e-01 -4.51641500e-01
-1.08414817e+00 -5.38792968e-01 -6.63583398e-01 4.91501577e-02
6.76603138e-01 -7.11693585e-01 -6.49150372e-01 3.37439835e-01
-1.74535537e+00 1.83101505e-01 4.01434332e-01 5.40631950e-01
-4.98841926e-02 6.88747048e-01 -9.34242785e-01 -6.94453120e-01
-6.53096020e-01 -1.03812432e+00 6.81372285e-01 3.97659093e-01
-2.20393702e-01 -9.62882996e-01 8.90882860e-04 4.24314469e-01
3.11244816e-01 -1.49553657e-01 7.87642062e-01 -9.80945408e-01
-2.99181134e-01 -3.04389954e-01 -4.11733210e-01 2.31254205e-01
-1.11679167e-01 -1.27766475e-01 -7.45761871e-01 -1.49817377e-01
2.30850741e-01 -6.56722412e-02 9.62850690e-01 3.62732410e-01
1.35495508e+00 -5.19586205e-01 -1.95119932e-01 3.95356774e-01
1.48192966e+00 3.26280594e-01 6.55938447e-01 3.15855712e-01
7.57149041e-01 8.15184355e-01 7.56257355e-01 2.79014587e-01
7.27786958e-01 3.13645989e-01 2.42964432e-01 4.91236523e-02
6.24828339e-02 -3.93807024e-01 2.31510371e-01 1.27482152e+00
3.22723448e-01 -6.15896642e-01 -1.02058005e+00 6.04004323e-01
-2.09914589e+00 -8.92738760e-01 -5.44952750e-01 1.60860908e+00
9.04795647e-01 3.22301865e-01 -4.02385086e-01 5.96484430e-02
9.71444309e-01 1.50591090e-01 -3.37903380e-01 -8.36165309e-01
7.89089352e-02 -1.38616368e-01 1.97257474e-01 6.57457173e-01
-1.10126078e+00 9.91885483e-01 6.57622671e+00 5.13097405e-01
-8.37095439e-01 9.79947224e-02 4.93566781e-01 -9.99438465e-02
-2.33744636e-01 9.03032348e-02 -1.10901606e+00 7.53141165e-01
9.29611087e-01 6.17675930e-02 1.17872782e-01 5.35171211e-01
4.99090195e-01 -1.73178956e-01 -9.22875047e-01 6.10814214e-01
1.67476714e-01 -9.90483880e-01 -2.57417262e-01 -5.99963963e-01
4.54712421e-01 -4.29913700e-02 -4.08551455e-01 4.67405796e-01
1.08877286e-01 -6.38467193e-01 2.80349195e-01 5.49991608e-01
2.94982374e-01 -8.59809220e-01 1.13005817e+00 6.11959398e-01
-1.18855190e+00 -7.14324489e-02 -4.96726125e-01 -5.86648107e-01
6.10065579e-01 8.00295174e-01 -8.82163465e-01 7.74040520e-01
6.39334321e-01 7.77782023e-01 -5.59902370e-01 1.38744664e+00
-6.58710420e-01 7.79179215e-01 -1.94630042e-01 -3.89416307e-01
2.98133492e-01 8.15809146e-02 3.31420749e-01 1.44874346e+00
1.61581695e-01 1.84702173e-01 2.40734458e-01 5.41930795e-01
-2.13842139e-01 1.40837893e-01 -4.62320924e-01 -1.05776429e-01
6.50129974e-01 1.23161352e+00 -5.25412261e-01 -1.98498666e-01
-3.69276732e-01 1.11594796e+00 7.53340364e-01 2.05343634e-01
-7.93329716e-01 -9.91175413e-01 -1.99136045e-02 -3.52448672e-01
3.23805124e-01 -2.61096954e-01 -5.70868254e-01 -1.12750053e+00
3.55884552e-01 -4.57351387e-01 2.93940067e-01 -7.51385987e-01
-1.34111273e+00 8.27477455e-01 -2.41576612e-01 -1.03142917e+00
6.45461082e-02 -3.65946472e-01 -8.44184458e-01 1.02044511e+00
-1.67211759e+00 -9.51690793e-01 1.02961972e-01 1.07125819e-01
6.94890916e-01 2.21480951e-01 6.21109128e-01 5.32608867e-01
-9.32826638e-01 6.93427145e-01 9.24313441e-02 3.65166575e-01
6.34727478e-01 -1.17503440e+00 4.95492369e-01 1.28719807e+00
-2.09640965e-01 9.87624526e-01 6.37943268e-01 -7.35216379e-01
-9.80401337e-01 -9.56837356e-01 1.76114953e+00 -9.16578770e-02
4.76916730e-01 -2.34960943e-01 -1.02770579e+00 6.39398932e-01
7.83486724e-01 -2.99612880e-01 7.39155710e-01 4.58431870e-01
5.85166104e-02 9.02963616e-03 -1.03625584e+00 5.41748524e-01
9.85958993e-01 -3.08594555e-01 -1.02452219e+00 3.60604316e-01
1.24106658e+00 -6.93964720e-01 -6.48384392e-01 4.03411895e-01
2.55085051e-01 -7.71690547e-01 4.94443119e-01 -5.88713050e-01
8.55253875e-01 -3.95893693e-01 8.31794590e-02 -1.34901857e+00
-5.17135799e-01 -3.11567008e-01 -9.17242616e-02 1.59817600e+00
6.71232879e-01 -4.54881042e-01 6.17191315e-01 8.36556017e-01
-5.52400947e-01 -1.12634575e+00 -6.97202325e-01 -6.63945615e-01
-2.55055726e-02 -2.88793951e-01 6.98216796e-01 8.50232601e-01
3.41039985e-01 7.11773694e-01 -5.15701354e-01 2.30610222e-01
5.73304389e-03 3.82809006e-02 7.60263205e-02 -7.30882585e-01
-3.87376755e-01 -2.75368214e-01 1.30643129e-01 -1.39051771e+00
2.11103082e-01 -9.65400279e-01 5.94997227e-01 -2.07146788e+00
1.55813813e-01 -3.12363327e-01 -5.00970960e-01 5.43370485e-01
-6.39445066e-01 -2.46277064e-01 3.22420418e-01 -1.97130423e-02
-8.51829290e-01 6.95247531e-01 1.28210330e+00 -4.37611863e-02
-2.74328262e-01 -2.03092426e-01 -8.13006938e-01 5.77829838e-01
1.14258885e+00 -6.54335082e-01 -3.30220103e-01 -1.32545948e+00
5.19521594e-01 1.99219853e-01 -1.18962921e-01 -8.31029415e-01
7.11259842e-01 -6.36189356e-02 3.11942428e-01 -1.12228131e+00
2.18056992e-01 -7.52624631e-01 -3.80433112e-01 3.49230051e-01
-6.38225913e-01 4.84692246e-01 1.38090447e-01 5.45075119e-01
-5.26546717e-01 -7.06037223e-01 3.84062707e-01 -2.42538810e-01
-5.60406029e-01 6.64763302e-02 -1.33096159e-01 -8.44673738e-02
7.54820943e-01 -5.64051531e-02 -3.52626145e-01 2.84514087e-03
-4.66188222e-01 7.19350517e-01 -5.67330327e-03 3.87796551e-01
7.84373045e-01 -1.13175976e+00 -7.57481039e-01 -3.11997291e-02
-9.97542590e-02 1.87581390e-01 4.23076600e-01 8.52838755e-01
-6.11511230e-01 5.76841295e-01 3.37914191e-02 -1.23367660e-01
-1.08919156e+00 4.06862855e-01 2.80002981e-01 -6.27682567e-01
-3.33996445e-01 1.20331740e+00 -2.46666774e-01 -8.14916253e-01
2.92975962e-01 -2.71675169e-01 -6.52316213e-01 -5.47357984e-02
6.31340921e-01 2.66958326e-01 1.85002178e-01 -1.71967670e-01
-5.55862725e-01 2.06948951e-01 -4.88465160e-01 2.95614544e-02
1.36358345e+00 -3.64805549e-01 -5.50754905e-01 3.27752590e-01
1.12234116e+00 -2.49396697e-01 -8.68534982e-01 -1.74327463e-01
3.09906423e-01 -3.89531851e-01 -6.48487806e-02 -1.03981066e+00
-7.80831933e-01 1.20155513e+00 1.14146069e-01 6.41680881e-02
1.29493272e+00 -5.19550025e-01 1.26855981e+00 3.73431593e-01
-1.19985584e-02 -1.03681886e+00 -2.02928901e-01 1.12218666e+00
7.88137317e-01 -1.32825983e+00 -2.51642972e-01 -4.15334523e-01
-5.64372540e-01 1.14132905e+00 9.89081025e-01 -2.23722681e-01
6.14548564e-01 1.06747337e-01 -5.12242541e-02 8.01795945e-02
-8.69302034e-01 2.50877738e-02 7.91553706e-02 2.15929985e-01
9.26691532e-01 -1.95230559e-01 -1.07674301e+00 1.05025411e+00
-4.08599414e-02 1.11946434e-01 4.03445631e-01 1.06225073e+00
-5.78880966e-01 -1.23411942e+00 -4.68389764e-02 4.82556880e-01
-6.38035655e-01 -6.72619402e-01 -3.06348264e-01 1.20227411e-01
7.43148178e-02 1.24343550e+00 9.28938854e-03 -4.18601096e-01
3.92553300e-01 2.81433072e-02 2.08784282e-01 -7.39953101e-01
-1.01308310e+00 -1.89498633e-01 3.32467526e-01 -2.63628036e-01
-2.60565549e-01 -4.14461613e-01 -1.55144238e+00 -1.77667156e-01
-4.73763496e-01 4.28818986e-02 4.93165106e-01 1.06983685e+00
5.95787704e-01 8.79265010e-01 6.22649968e-01 -4.58904296e-01
-4.90274280e-01 -1.40479326e+00 -1.97243154e-01 1.73748896e-01
1.70709163e-01 -1.12941220e-01 -1.53145179e-01 -3.53390694e-01] | [10.827478408813477, 8.979944229125977] |
5a56916e-67b7-4704-bfa8-3b5e9d9588a6 | do-vision-language-pretrained-models-learn | 2203.17271 | null | https://arxiv.org/abs/2203.17271v3 | https://arxiv.org/pdf/2203.17271v3.pdf | Do Vision-Language Pretrained Models Learn Composable Primitive Concepts? | Vision-language (VL) pretrained models have achieved impressive performance on multimodal reasoning and zero-shot recognition tasks. Many of these VL models are pretrained on unlabeled image and caption pairs from the internet. In this paper, we study whether representations of primitive concepts--such as colors, shapes, or the attributes of object parts--emerge automatically within these pretrained VL models. We propose a two-step framework, Compositional Concept Mapping (CompMap), to investigate this. CompMap first asks a VL model to generate primitive concept activations with text prompts, and then learns to construct a composition model that maps the primitive concept activations (e.g. the likelihood of black tail or red wing) to composite concepts (e.g. a red-winged blackbird). We show that a composition model can be reliably learn from ground truth primitive concepts. We thus hypothesize that if primitive concepts indeed emerge in a VL pretrained model, its primitive concept activations can be used to learn a composition model similar to the one designed by experts. We propose a quantitative metric to measure the degree of similarity, and refer to the metric as the interpretability metric. We also measure the classification accuracy when using the primitive concept activations and the learned composition model to predict the composite concepts, and refer to it as the usefulness metric. Our study reveals that state-of-the-art VL pretrained models learn primitive concepts that are highly useful for fine-grained visual recognition on the CUB dataset, and compositional generalization tasks on the MIT-States dataset. However, we observe that the learned composition models have low interpretability in our qualitative analyses. Our results reveal the limitations of existing VL models, and the necessity of pretraining objectives that encourage the acquisition of primitive concepts. | ['Chen Sun', 'Ellie Pavlick', 'Usha Bhalla', 'Tian Yun'] | 2022-03-31 | null | null | null | null | ['fine-grained-visual-recognition'] | ['computer-vision'] | [ 2.60698736e-01 2.10976839e-01 -1.43303066e-01 -4.89437580e-01
-2.94569165e-01 -8.28708470e-01 9.61897552e-01 7.13156611e-02
-3.68329763e-01 4.99549299e-01 1.73550725e-01 -3.56625468e-01
5.20511940e-02 -7.68238425e-01 -1.04428577e+00 -5.04967391e-01
8.03012326e-02 7.72990823e-01 -9.16598961e-02 -2.62614489e-01
2.66880661e-01 3.39997321e-01 -1.84574962e+00 7.44372725e-01
8.68928850e-01 9.77641761e-01 2.06200540e-01 5.78140080e-01
-5.02816141e-01 9.72220659e-01 -5.43525994e-01 -4.13585424e-01
2.22939283e-01 -5.84364116e-01 -1.03008485e+00 1.44556031e-01
7.87882566e-01 -3.66059124e-01 -4.20746915e-02 9.21837926e-01
-1.24174125e-01 1.99024737e-01 1.17175567e+00 -1.51407659e+00
-1.27927935e+00 6.91502750e-01 -1.89402491e-01 -1.37706092e-02
6.42310441e-01 3.93332750e-01 1.48185718e+00 -1.17263234e+00
6.70312822e-01 1.52030480e+00 3.50599229e-01 9.59178209e-01
-1.46484959e+00 -4.94885206e-01 4.95470732e-01 3.71395081e-01
-1.42507768e+00 -1.47858575e-01 6.14100277e-01 -6.45452559e-01
7.31840432e-01 2.24250734e-01 6.90845430e-01 1.35146904e+00
-2.90677071e-01 1.04218185e+00 1.32177496e+00 -4.06657398e-01
4.00160640e-01 3.51509564e-02 2.01811209e-01 1.00299597e+00
1.25703856e-01 3.08924794e-01 -7.03570724e-01 1.92645611e-03
6.92976832e-01 -1.25699192e-01 -2.25176811e-01 -4.39423323e-01
-1.24232471e+00 8.31455708e-01 8.87205064e-01 2.87742049e-01
-1.36814028e-01 2.54290611e-01 -1.65073238e-02 2.79354185e-01
6.55264035e-02 7.35629737e-01 -3.13093036e-01 3.31253082e-01
-6.01500332e-01 4.83351871e-02 8.13678086e-01 9.95189667e-01
1.21690345e+00 1.94223430e-02 -1.85245678e-01 6.29085779e-01
4.11265194e-01 5.54023087e-01 5.16840398e-01 -8.35049748e-01
1.97187632e-01 9.27975595e-01 -2.53865153e-01 -6.26174510e-01
-1.22581057e-01 -1.26996771e-01 -6.03105485e-01 2.08286181e-01
2.25529969e-01 1.26519859e-01 -1.22006214e+00 1.95027959e+00
-4.85275723e-02 1.95681140e-01 3.89669925e-01 1.04614925e+00
1.12000811e+00 6.17148638e-01 3.54249209e-01 1.99170396e-01
1.31746304e+00 -9.28488016e-01 -1.58864066e-01 -5.15753865e-01
3.28827024e-01 -1.78355828e-01 1.54486728e+00 1.65440843e-01
-6.88730478e-01 -8.21558118e-01 -9.15194988e-01 -8.49140957e-02
-6.48758888e-01 9.53800827e-02 6.85163260e-01 4.51156706e-01
-1.15087855e+00 4.04465020e-01 -4.37610984e-01 -7.24194586e-01
4.39534277e-01 7.80614391e-02 -4.43408132e-01 -2.16202125e-01
-8.43095303e-01 9.24442291e-01 6.66887105e-01 -1.56387806e-01
-1.70462120e+00 -5.54083526e-01 -9.94328678e-01 2.33515903e-01
3.94814938e-01 -9.11644816e-01 1.08928967e+00 -1.64837086e+00
-1.16724908e+00 1.31067419e+00 3.08806989e-02 -4.06535864e-01
5.55999838e-02 6.93649128e-02 -1.58352658e-01 4.06836689e-01
4.27587926e-02 1.31106055e+00 1.07933116e+00 -1.91951728e+00
-6.38217747e-01 -1.52495459e-01 5.27323186e-01 2.51241297e-01
-1.65970236e-01 -3.26648921e-01 -1.94075257e-01 -5.49252570e-01
1.43519521e-01 -9.17132497e-01 1.64164439e-01 3.69177073e-01
-2.50040859e-01 -3.87091964e-01 5.16036808e-01 -3.24065387e-01
5.49613774e-01 -1.92476928e+00 4.30202901e-01 1.57803714e-01
3.24384421e-01 3.75860259e-02 -5.81046999e-01 2.29318559e-01
-1.51184693e-01 3.49032611e-01 -4.96853739e-01 -1.81863070e-01
1.54007047e-01 8.08210492e-01 -5.97062707e-01 1.84038922e-01
6.27458096e-01 1.15383708e+00 -1.00792432e+00 -3.00876379e-01
8.62186030e-02 1.39662996e-01 -5.08376122e-01 5.66857815e-01
-7.17374980e-01 3.12503099e-01 -2.26705462e-01 9.28198814e-01
2.45866507e-01 -3.09037328e-01 2.16650367e-01 -1.90398023e-01
1.93225712e-01 -1.88236348e-02 -6.93763435e-01 1.69625926e+00
-3.31469804e-01 7.04984009e-01 -2.55413085e-01 -1.16054416e+00
1.00031137e+00 3.11163254e-02 -6.67263791e-02 -5.78258216e-01
7.28625953e-02 -6.22589178e-02 1.22265376e-01 -5.77838123e-01
2.54394054e-01 -3.90190333e-01 -6.94151819e-02 4.61737871e-01
4.58459616e-01 -2.95904130e-01 5.55736162e-02 3.37375581e-01
7.80059457e-01 4.09158081e-01 3.67439359e-01 -3.54286611e-01
6.39115810e-01 1.69833183e-01 2.22415417e-01 8.59048009e-01
-5.33986241e-02 4.91093129e-01 3.91696781e-01 -4.95957822e-01
-8.18138659e-01 -1.46765852e+00 2.22866312e-01 1.56022096e+00
2.91271985e-01 -4.12789792e-01 -5.63980877e-01 -7.99440026e-01
7.15067014e-02 9.28925395e-01 -9.28264678e-01 -3.74061972e-01
-1.68294400e-01 -3.32046181e-01 7.00919807e-01 6.79471135e-01
4.12065744e-01 -1.29523742e+00 -7.37982392e-01 -2.86515027e-01
-2.95351028e-01 -1.06664217e+00 -3.33360843e-02 2.09898204e-01
-7.78950810e-01 -1.09708595e+00 -3.21849078e-01 -8.91633213e-01
1.07320642e+00 1.33788556e-01 1.47330582e+00 6.44715965e-01
-2.55146950e-01 1.01795614e+00 -5.96428156e-01 -2.92926461e-01
-4.25131440e-01 -4.22146887e-01 1.36173680e-01 2.81535476e-01
3.36053073e-01 -2.12056831e-01 -3.16773832e-01 2.07769603e-01
-1.11827636e+00 3.34222674e-01 6.96516693e-01 1.02474737e+00
5.36809683e-01 -3.74381542e-01 2.30157524e-01 -6.06403649e-01
5.15027463e-01 -4.68599111e-01 -3.43221188e-01 6.86884344e-01
-4.62293059e-01 5.56894660e-01 7.08195567e-01 -6.31939828e-01
-1.12958133e+00 1.04524486e-01 2.09725257e-02 -6.11794174e-01
-7.08317697e-01 6.89481795e-01 -3.11490521e-02 -2.10592777e-01
8.72886121e-01 4.05556113e-01 -2.75824308e-01 -3.21319520e-01
8.73517573e-01 3.18466663e-01 7.98207521e-01 -1.32678258e+00
9.00333226e-01 5.73266864e-01 -6.85701221e-02 -1.01812065e+00
-9.74193871e-01 -3.53700489e-01 -7.07704067e-01 -1.81120455e-01
9.38253760e-01 -8.99594963e-01 -7.15295017e-01 1.05776459e-01
-1.30372488e+00 -4.84001726e-01 -3.30597192e-01 5.12040481e-02
-6.74332798e-01 3.47249597e-01 -4.41350460e-01 -8.50991964e-01
-1.38215303e-01 -8.92954409e-01 1.07654381e+00 3.30636084e-01
-1.75372049e-01 -8.48511696e-01 -8.31156597e-02 3.53134453e-01
7.83915073e-02 1.12133868e-01 1.38944113e+00 -6.38502717e-01
-7.17277586e-01 2.90065736e-01 -3.82198483e-01 2.29264751e-01
-8.88837352e-02 6.56772256e-02 -1.11082566e+00 -2.31001139e-01
-2.05914766e-01 -8.71790707e-01 1.05593443e+00 -3.35861444e-02
1.33547401e+00 -3.96023214e-01 -1.24626040e-01 6.82582319e-01
1.42416489e+00 5.84848784e-02 5.32257259e-01 -5.16886637e-02
6.90405488e-01 8.65520000e-01 4.56432819e-01 8.38408098e-02
3.66468549e-01 3.83158326e-01 5.18244088e-01 4.26277705e-02
-2.03946903e-01 -6.40333116e-01 5.94772577e-01 5.50127566e-01
-2.94777781e-01 -2.46840999e-01 -1.17987585e+00 5.02687514e-01
-1.87172151e+00 -9.18604612e-01 2.01803282e-01 1.93544269e+00
8.52445900e-01 -9.14944708e-02 -2.87134852e-02 -3.57242018e-01
3.95777822e-01 -1.91513225e-01 -4.60612118e-01 -3.36720735e-01
-2.31149644e-01 3.44571829e-01 -5.77621795e-02 5.39635420e-01
-8.77441406e-01 1.26870787e+00 6.47278976e+00 4.81216401e-01
-9.55975473e-01 -6.11408986e-02 5.20336032e-01 3.72513980e-01
-6.07473850e-01 2.35115886e-01 -6.07569158e-01 -7.14915292e-03
6.48132563e-01 -9.49798226e-02 6.47057056e-01 9.51673985e-01
-4.19923335e-01 -5.25112636e-02 -1.68908846e+00 1.09060431e+00
6.85563564e-01 -1.28659785e+00 8.63563895e-01 -2.18507186e-01
6.63592041e-01 -2.62516916e-01 1.21911086e-01 5.96422911e-01
4.40172791e-01 -1.37369573e+00 9.03520465e-01 7.64760256e-01
8.40729177e-01 -2.43807256e-01 3.75458062e-01 3.44764531e-01
-1.33801103e+00 -2.03430727e-01 -5.21209896e-01 -1.23503104e-01
-2.85258293e-01 -2.73841918e-01 -9.84625280e-01 4.34704602e-01
5.17278552e-01 5.65766931e-01 -1.01471889e+00 7.83324182e-01
-7.47714698e-01 4.33784693e-01 -7.09940419e-02 -1.43542334e-01
3.63481075e-01 -1.68534815e-01 3.27902734e-01 1.02948034e+00
1.01738021e-01 1.93056464e-01 2.74881005e-01 1.61250758e+00
-1.18341222e-01 -2.04077348e-01 -6.75740063e-01 -3.15466344e-01
9.93602052e-02 1.19987512e+00 -5.53282142e-01 -5.40061831e-01
-4.36369061e-01 1.05532289e+00 6.76460505e-01 5.46407759e-01
-6.72169209e-01 1.78624630e-01 7.57711887e-01 -2.64973730e-01
6.11176938e-02 -2.23024681e-01 -2.25232109e-01 -1.28560519e+00
-4.37616646e-01 -9.41944122e-01 5.48451662e-01 -1.25211596e+00
-1.67881787e+00 6.79969132e-01 2.07674265e-01 -1.01062298e+00
-2.81815171e-01 -1.21527255e+00 -6.39850736e-01 5.83929479e-01
-1.37903750e+00 -1.42835307e+00 -6.39859617e-01 9.08627808e-01
6.38748229e-01 -3.51291776e-01 9.92666066e-01 -3.37191910e-01
-1.47044241e-01 3.90215755e-01 -5.71668267e-01 2.85015553e-01
4.22006279e-01 -1.45967638e+00 2.27737278e-01 8.53565693e-01
8.61043334e-01 8.39690208e-01 5.53565919e-01 -4.93911624e-01
-1.39321303e+00 -9.04071808e-01 5.41520536e-01 -7.31367171e-01
6.70788765e-01 -4.77561444e-01 -1.00060296e+00 7.29883790e-01
7.71072507e-02 -3.59725729e-02 7.87712753e-01 1.63119197e-01
-9.29735839e-01 6.39229044e-02 -8.42718780e-01 7.57492304e-01
1.25770819e+00 -9.85420287e-01 -1.37614620e+00 2.09291369e-01
7.01599777e-01 6.57580495e-02 -4.04923290e-01 3.22307467e-01
5.00763774e-01 -7.76091695e-01 1.13962543e+00 -1.20483387e+00
6.33231163e-01 -4.46856886e-01 -3.40874940e-01 -1.40786695e+00
-4.48797375e-01 8.59435946e-02 -1.75504297e-01 9.12548363e-01
4.62726057e-01 -3.22582364e-01 6.39305472e-01 5.59541166e-01
-1.50608942e-01 -3.00213903e-01 -7.03536153e-01 -8.99761200e-01
4.12416793e-02 -3.29314917e-01 3.71535689e-01 1.00218749e+00
-6.87848404e-02 6.79629862e-01 -1.73665378e-02 1.91349044e-01
6.44720852e-01 3.91729683e-01 7.39262104e-01 -1.42675674e+00
-3.67486417e-01 -4.94037151e-01 -3.28076005e-01 -1.07219803e+00
5.86955845e-01 -1.13129079e+00 3.11772466e-01 -1.45586276e+00
2.93412924e-01 -3.52881074e-01 -2.72035301e-01 8.50936115e-01
-2.08392948e-01 1.55076236e-01 4.71563935e-01 1.98959216e-01
-5.61660171e-01 7.12854981e-01 1.07365692e+00 -6.22844219e-01
9.55176726e-02 -5.28964221e-01 -6.34522021e-01 7.86626220e-01
5.76103449e-01 -1.81570321e-01 -5.40189803e-01 -5.78960896e-01
3.75787407e-01 -2.80740052e-01 9.77730036e-01 -9.86930609e-01
5.67694195e-02 -4.43964392e-01 4.58953351e-01 -1.91588968e-01
4.74800140e-01 -1.07466018e+00 -1.98083501e-02 4.87154692e-01
-4.52682823e-01 -1.08192742e-01 2.29733393e-01 6.42748713e-01
-1.97819591e-01 -4.43852425e-01 5.92223525e-01 -3.78697395e-01
-1.44489121e+00 1.76768079e-01 -3.32649201e-01 1.24613225e-01
8.74433458e-01 -2.09431529e-01 -7.10130990e-01 -4.81871247e-01
-6.46811485e-01 2.59427428e-01 5.26184559e-01 6.66233242e-01
1.08339393e+00 -1.36636591e+00 -6.82018340e-01 3.09931159e-01
7.69094467e-01 -3.26671362e-01 -1.37176320e-01 4.24017191e-01
-5.28415740e-01 3.03026199e-01 -5.94939411e-01 -7.55926013e-01
-1.10720110e+00 8.83085132e-01 4.67834055e-01 3.49948466e-01
-3.21759045e-01 1.01699340e+00 8.29948545e-01 -5.66975892e-01
1.68598458e-01 -5.79857409e-01 -2.04254657e-01 -5.93114682e-02
4.54772770e-01 -3.12795073e-01 -5.10632038e-01 -7.43350148e-01
-3.21205318e-01 6.04281485e-01 2.33831435e-01 -6.37100264e-02
1.03231251e+00 6.26222119e-02 -2.27665603e-01 5.58371484e-01
7.93908656e-01 -4.57852423e-01 -1.23781669e+00 -3.11835825e-01
1.75067246e-01 -2.88394004e-01 -3.63371938e-01 -9.59665537e-01
-7.37811029e-01 1.10482740e+00 4.81890023e-01 -1.28348842e-01
1.05433547e+00 4.32546437e-01 3.60117741e-02 8.67367268e-01
4.81124997e-01 -9.16803718e-01 6.35579467e-01 7.80233026e-01
1.14203358e+00 -1.37887335e+00 -3.83202285e-01 -2.92991310e-01
-9.94372249e-01 1.10436463e+00 1.05767024e+00 7.97199365e-03
1.48232087e-01 -1.47276074e-01 -9.68265980e-02 -3.55938315e-01
-8.43050241e-01 -7.04414368e-01 7.94467032e-01 8.97879660e-01
2.23023340e-01 3.31807375e-01 1.28139138e-01 7.10061491e-01
-1.17299892e-01 -3.24325323e-01 3.47329259e-01 6.18771076e-01
-5.67492783e-01 -7.85683572e-01 -3.74365151e-01 2.97266871e-01
3.64454895e-01 -3.28474730e-01 -1.06142211e+00 5.93536079e-01
4.13251311e-01 7.94592917e-01 2.53190160e-01 -6.03305042e-01
1.36654586e-01 4.98385549e-01 7.07822025e-01 -9.83388722e-01
-4.49639231e-01 -5.42280853e-01 9.83871147e-02 -5.32620132e-01
-5.59649587e-01 -2.07795262e-01 -1.33567405e+00 -8.78775641e-02
1.66031450e-01 -7.60758370e-02 3.35766733e-01 1.20286870e+00
-4.38148603e-02 2.22693115e-01 1.30531371e-01 -7.86906242e-01
-2.99554855e-01 -8.51103544e-01 -3.15376997e-01 9.56422269e-01
2.72333711e-01 -7.67376959e-01 -4.16139662e-01 3.29986602e-01] | [10.347854614257812, 1.9953994750976562] |
ed290824-5a95-411f-b6a1-1b8b87c9c075 | out-of-the-box-embodied-navigation-in-the | 2105.05873 | null | https://arxiv.org/abs/2105.05873v1 | https://arxiv.org/pdf/2105.05873v1.pdf | Out of the Box: Embodied Navigation in the Real World | The research field of Embodied AI has witnessed substantial progress in visual navigation and exploration thanks to powerful simulating platforms and the availability of 3D data of indoor and photorealistic environments. These two factors have opened the doors to a new generation of intelligent agents capable of achieving nearly perfect PointGoal Navigation. However, such architectures are commonly trained with millions, if not billions, of frames and tested in simulation. Together with great enthusiasm, these results yield a question: how many researchers will effectively benefit from these advances? In this work, we detail how to transfer the knowledge acquired in simulation into the real world. To that end, we describe the architectural discrepancies that damage the Sim2Real adaptation ability of models trained on the Habitat simulator and propose a novel solution tailored towards the deployment in real-world scenarios. We then deploy our models on a LoCoBot, a Low-Cost Robot equipped with a single Intel RealSense camera. Different from previous work, our testing scene is unavailable to the agent in simulation. The environment is also inaccessible to the agent beforehand, so it cannot count on scene-specific semantic priors. In this way, we reproduce a setting in which a research group (potentially from other fields) needs to employ the agent visual navigation capabilities as-a-Service. Our experiments indicate that it is possible to achieve satisfying results when deploying the obtained model in the real world. Our code and models are available at https://github.com/aimagelab/LoCoNav. | ['Rita Cucchiara', 'Lorenzo Baraldi', 'Silvia Cascianelli', 'Marcella Cornia', 'Federico Landi', 'Roberto Bigazzi'] | 2021-05-12 | null | null | null | null | ['pointgoal-navigation'] | ['robots'] | [-3.01765770e-01 2.22431913e-01 3.47766161e-01 -2.65845537e-01
-1.74537057e-03 -7.99930334e-01 8.01264346e-01 -2.42814094e-01
-8.33807588e-01 8.09890330e-01 -1.19620219e-01 -4.53700781e-01
1.35345116e-01 -8.55634093e-01 -9.11688328e-01 -5.99566996e-01
-4.64418292e-01 7.58791447e-01 4.85155284e-01 -6.65810227e-01
8.00609067e-02 6.36686683e-01 -2.01308656e+00 -2.00038001e-01
6.60989881e-01 6.75178528e-01 8.78871083e-01 8.59039009e-01
2.65351921e-01 7.45445132e-01 -3.89372289e-01 5.19294739e-02
6.22016668e-01 -9.79667976e-02 -5.93824744e-01 -2.28775427e-01
1.68318540e-01 -4.84105676e-01 -3.74483705e-01 9.20900583e-01
5.40786386e-01 2.40029395e-01 1.52794942e-01 -1.47541070e+00
-1.26369879e-01 3.25259596e-01 -6.86315224e-02 3.64338681e-02
6.26742542e-01 5.56225717e-01 4.47009146e-01 -5.92460513e-01
8.86560321e-01 1.19245553e+00 4.94630992e-01 4.56785232e-01
-7.13380992e-01 -2.69645482e-01 3.66774291e-01 3.13205421e-01
-1.27309394e+00 -5.13227642e-01 5.17351151e-01 -2.50994444e-01
1.09222817e+00 1.81490526e-01 1.08374822e+00 1.64671111e+00
2.09893480e-01 6.65016234e-01 1.04366243e+00 -3.67949754e-01
7.95077503e-01 1.45897821e-01 -4.14831698e-01 7.66317010e-01
3.24875027e-01 5.64123571e-01 -4.76790458e-01 1.32434681e-01
9.40066040e-01 -1.86970949e-01 -3.75522971e-01 -1.05988085e+00
-1.32365203e+00 6.20024264e-01 7.50411570e-01 2.86573976e-01
-4.30061996e-01 4.37103868e-01 1.57046199e-01 2.66881198e-01
-8.98070708e-02 4.35925335e-01 -4.30124134e-01 -3.65598500e-01
-3.96213382e-01 3.69744092e-01 9.68779206e-01 1.10983300e+00
6.40289128e-01 1.69616595e-01 7.75745749e-01 2.31117100e-01
5.43619812e-01 6.41434669e-01 4.62614387e-01 -1.24682021e+00
1.47174686e-01 3.68888766e-01 4.87178773e-01 -1.01609671e+00
-8.25170934e-01 -3.96929532e-01 -3.41900289e-01 1.03257143e+00
5.14303088e-01 -1.94506064e-01 -1.06050336e+00 1.72032487e+00
5.30932546e-01 4.33138870e-02 2.90881008e-01 1.28101897e+00
5.29141963e-01 5.03661871e-01 1.02006393e-02 5.22633255e-01
1.24602413e+00 -1.07945025e+00 -3.12474966e-01 -8.06400239e-01
5.94810247e-01 -4.07265961e-01 1.14907086e+00 3.82307738e-01
-7.54106164e-01 -3.92771244e-01 -1.29020226e+00 1.85649469e-01
-7.36968577e-01 -2.24885657e-01 1.06539249e+00 5.47479153e-01
-1.38215828e+00 3.47252876e-01 -1.29900384e+00 -1.04407954e+00
9.26176459e-02 2.69618422e-01 -6.28791630e-01 1.63890049e-02
-1.04218292e+00 1.47208822e+00 3.73400539e-01 2.60243416e-01
-1.34512079e+00 -1.58145547e-01 -9.70605254e-01 -2.26898044e-01
3.80263776e-01 -1.10001135e+00 1.34083104e+00 -9.93052244e-01
-1.73128271e+00 6.99398220e-01 1.92650184e-01 -5.77337682e-01
8.94853830e-01 -2.14152485e-01 -9.64195095e-03 6.47593464e-04
9.58650187e-02 8.26189756e-01 2.86842644e-01 -1.60016978e+00
-6.90897763e-01 -4.49842870e-01 6.99458301e-01 7.44003534e-01
2.35609382e-01 -4.02605951e-01 -4.68831569e-01 -2.50561833e-02
8.29545707e-02 -1.06306458e+00 -6.88983679e-01 2.02776834e-01
1.29832715e-01 3.97170335e-01 5.73433220e-01 -2.96342224e-01
1.47099152e-01 -1.97062647e+00 2.87720442e-01 -3.56956050e-02
-1.55020520e-01 -5.70784397e-02 -1.29180849e-01 6.81711555e-01
3.71489137e-01 -2.10510254e-01 -1.35877296e-01 -3.55605632e-01
2.80591279e-01 4.99715954e-01 -1.77925065e-01 6.18032634e-01
-6.19932890e-01 7.75764048e-01 -1.16863763e+00 -1.81333125e-01
6.09654427e-01 6.37995422e-01 -5.09050012e-01 2.06845164e-01
-2.38032043e-01 7.03029275e-01 -5.58433831e-01 6.28038943e-01
6.32854760e-01 9.11992639e-02 2.50175267e-01 3.56616169e-01
-4.62968439e-01 8.47154632e-02 -1.11761558e+00 2.41804862e+00
-6.03968143e-01 5.32296658e-01 4.46005166e-01 -6.92325711e-01
6.62854612e-01 -9.60660819e-03 8.17756802e-02 -9.16649878e-01
2.37618402e-01 2.78468221e-01 -1.06219999e-01 -6.12308443e-01
6.52050257e-01 -1.21022835e-02 -8.84791836e-02 8.91678482e-02
9.58770812e-02 -5.29999495e-01 1.00560077e-01 8.37124810e-02
1.28425908e+00 7.19127297e-01 2.54028857e-01 -3.05312604e-01
1.67816356e-01 5.14958322e-01 2.19909996e-01 9.25822318e-01
-2.54638225e-01 4.44001734e-01 -7.60650784e-02 -6.91627383e-01
-1.15698791e+00 -1.19668388e+00 1.45844907e-01 1.00762522e+00
7.24889755e-01 -2.27158919e-01 -7.43202984e-01 -3.53975624e-01
-1.98626280e-01 7.79653728e-01 -6.73652828e-01 1.25712529e-01
-5.78763485e-01 -4.75326896e-01 4.04659688e-01 3.55346590e-01
5.68357468e-01 -1.26444948e+00 -1.75581336e+00 1.12426251e-01
1.14366673e-01 -1.01458299e+00 4.92201865e-01 3.37867200e-01
-5.58064103e-01 -9.69660759e-01 -5.12624443e-01 -6.49608731e-01
6.08759999e-01 4.35066283e-01 1.00985134e+00 1.35715663e-01
-1.51224375e-01 7.59913266e-01 -5.91228426e-01 -4.26869810e-01
-2.51971930e-01 5.62517233e-02 1.52392894e-01 -6.88426316e-01
-4.03847806e-02 -8.54724884e-01 -7.70860374e-01 2.92963058e-01
-7.07951903e-01 3.15470338e-01 4.75382447e-01 4.79510218e-01
2.28997812e-01 -2.43232310e-01 6.38198629e-02 -2.75058895e-01
2.57050544e-01 -4.91356105e-01 -1.01324093e+00 -3.54308411e-02
-2.69486219e-01 -8.76305774e-02 6.29603028e-01 -1.92478836e-01
-9.51900065e-01 2.29037240e-01 -1.84619576e-01 -5.32967597e-02
-6.15205169e-01 4.81478393e-01 -1.47902116e-01 -3.01200747e-01
7.52294660e-01 1.85951777e-02 -4.92669828e-02 -2.04063624e-01
4.47707146e-01 4.42866445e-01 5.08570254e-01 -5.11356771e-01
7.69469440e-01 9.17590678e-01 -1.28291130e-01 -8.13850820e-01
-1.80047154e-01 -7.95940384e-02 -4.69561934e-01 -3.93077195e-01
6.85694933e-01 -9.44225550e-01 -9.17895675e-01 4.26542670e-01
-1.02710104e+00 -1.02288556e+00 -3.47395569e-01 6.25419617e-01
-9.15706813e-01 1.40998587e-01 -1.86260745e-01 -8.60931098e-01
1.13426775e-01 -1.31549227e+00 9.18290317e-01 5.82813144e-01
3.36761912e-03 -9.91747439e-01 4.65402842e-01 5.09290919e-02
5.95373631e-01 1.87048510e-01 8.41195509e-02 -2.93434441e-01
-7.53095806e-01 -2.26826351e-02 1.02742545e-01 -3.77324164e-01
-2.71099001e-01 -3.47454816e-01 -1.04983437e+00 -3.74671668e-01
2.76419390e-02 -3.02788794e-01 5.23661375e-01 1.87568009e-01
5.70312619e-01 1.36012897e-01 -4.83715415e-01 8.53138626e-01
1.42821240e+00 4.36153084e-01 6.51010633e-01 1.22001421e+00
4.60979581e-01 5.00272274e-01 7.28881121e-01 3.86874318e-01
8.51934016e-01 8.94220054e-01 1.09212255e+00 -1.41226903e-01
3.59973572e-02 -4.21028197e-01 4.05723244e-01 2.53936529e-01
-4.02740538e-01 -4.58525717e-01 -1.15513301e+00 5.64429820e-01
-2.02026200e+00 -8.83541346e-01 1.83683470e-01 2.09455633e+00
1.32897170e-02 -5.68699576e-02 -1.80438161e-01 -4.06366706e-01
2.50303388e-01 2.08445847e-01 -5.06305456e-01 -3.16017210e-01
-1.31124288e-01 -2.97269493e-01 6.51487052e-01 6.59605086e-01
-9.28310633e-01 1.13155437e+00 5.82379436e+00 1.90806389e-01
-1.11607015e+00 9.70851257e-02 2.07736567e-02 2.05012932e-02
-1.84449255e-01 3.93714577e-01 -3.94332528e-01 3.34404975e-01
9.74592984e-01 -7.65404403e-02 7.69993544e-01 1.16324019e+00
3.15461248e-01 -6.86016500e-01 -1.00424707e+00 8.16596270e-01
3.14194188e-02 -1.09422934e+00 -3.73039395e-01 2.33912647e-01
2.73629934e-01 6.64520860e-01 5.22652231e-02 2.94054002e-01
7.92563796e-01 -1.13661194e+00 1.19275653e+00 4.30641055e-01
1.73468679e-01 -4.90247309e-01 7.63982594e-01 6.65304840e-01
-9.45125639e-01 -8.14971998e-02 -5.26639104e-01 -4.95131791e-01
3.96794051e-01 -1.03249945e-01 -1.14512420e+00 6.71844959e-01
9.05142665e-01 3.42162639e-01 -5.11322200e-01 1.33229649e+00
-3.88663888e-01 -5.33302203e-02 -6.55203521e-01 -2.78024763e-01
3.95338029e-01 -2.82451481e-01 5.34081936e-01 8.38132679e-01
5.88615716e-01 2.50248308e-03 1.38031870e-01 5.57495534e-01
3.91526759e-01 -2.17617124e-01 -1.07003915e+00 4.50166672e-01
3.50251973e-01 1.20558441e+00 -9.80935276e-01 -4.12035324e-02
-2.65957534e-01 1.09153759e+00 5.21243393e-01 4.44100738e-01
-9.81888235e-01 -7.04974309e-02 6.02441788e-01 -2.86783539e-02
1.83107376e-01 -5.94343066e-01 -5.72016202e-02 -1.17443037e+00
4.51830728e-03 -7.86240220e-01 -2.84977560e-03 -1.24877834e+00
-7.03611970e-01 9.65321541e-01 4.89871092e-02 -1.18092203e+00
-3.94136459e-01 -6.84993386e-01 -2.79769361e-01 4.83760804e-01
-1.43569028e+00 -1.37189662e+00 -7.15094090e-01 3.45574528e-01
4.62691277e-01 -9.82637480e-02 1.00838745e+00 -8.96264706e-03
-4.60905284e-02 -5.99054061e-02 -5.42300679e-02 -2.38760397e-01
3.29535335e-01 -1.13318932e+00 7.22640514e-01 9.22712624e-01
1.90048113e-01 3.77605289e-01 1.26578915e+00 -4.36185271e-01
-1.65712500e+00 -4.25198883e-01 1.46725282e-01 -6.84148669e-01
6.78485632e-01 -5.86695075e-01 -4.77984279e-01 1.05206323e+00
3.76044393e-01 -3.35984007e-02 6.23339862e-02 -1.07460529e-01
5.45243435e-02 1.74259782e-01 -1.18763363e+00 1.01281500e+00
1.43156028e+00 -8.13443661e-02 -6.36781335e-01 1.80109680e-01
4.67549890e-01 -8.36836219e-01 -2.86247224e-01 3.45676690e-01
6.44680798e-01 -1.44818544e+00 9.40854371e-01 -2.04772711e-01
-1.09428704e-01 -6.68173373e-01 -3.77569854e-01 -1.60895944e+00
9.28764492e-02 -4.51457143e-01 3.24593097e-01 5.27623177e-01
3.51098478e-01 -8.93154621e-01 8.44080150e-01 5.60059726e-01
-4.11550343e-01 -2.48618335e-01 -1.17984152e+00 -7.60566115e-01
-6.03124611e-02 -5.11477053e-01 6.05921268e-01 6.01344228e-01
7.08778501e-02 4.77995574e-02 -2.27074087e-01 4.79819834e-01
5.29556990e-01 -1.52137689e-02 1.25762093e+00 -1.01977754e+00
-2.75787711e-01 -2.07977667e-01 -6.95638716e-01 -1.05217326e+00
1.64608613e-01 -4.95210022e-01 3.49731356e-01 -1.74870420e+00
-1.59465775e-01 -6.90163136e-01 2.92786330e-01 3.32976013e-01
4.63274330e-01 3.63215469e-02 3.45788717e-01 -4.87887822e-02
-7.50087380e-01 6.91453874e-01 1.09317791e+00 1.53200954e-01
-1.52695745e-01 -2.15774789e-01 -3.16198200e-01 1.07682765e+00
9.49911177e-01 -2.33218148e-01 -5.18287539e-01 -7.64936864e-01
5.70236444e-01 -8.65416899e-02 7.68829167e-01 -1.47960317e+00
5.19389153e-01 -1.82869852e-01 2.28412867e-01 -2.25922465e-01
7.57532775e-01 -1.15625691e+00 5.48101425e-01 5.54715872e-01
1.96392715e-01 2.57246196e-01 3.22408736e-01 4.35853541e-01
1.37217209e-01 -3.39161754e-01 3.98657501e-01 -6.23571336e-01
-1.33369660e+00 -8.76946524e-02 -6.34518802e-01 -3.48684639e-01
1.24303317e+00 -3.97026300e-01 -5.96847594e-01 -5.18061578e-01
-6.17720008e-01 3.05582970e-01 1.29156578e+00 4.34280455e-01
6.29780173e-01 -8.62736583e-01 -2.58041799e-01 2.65078455e-01
6.63727969e-02 1.58804715e-01 3.21829647e-01 5.76850593e-01
-1.15802956e+00 2.67999291e-01 -6.22566223e-01 -5.74412048e-01
-8.43776941e-01 6.31270945e-01 4.86944735e-01 1.82458892e-01
-8.30931604e-01 6.47538960e-01 2.92921782e-01 -8.26241791e-01
1.98587492e-01 -2.52196133e-01 -3.27544697e-02 -4.32459176e-01
4.80335325e-01 1.18378401e-01 6.52672490e-03 -7.06209362e-01
-5.30025363e-01 4.80951399e-01 4.43464547e-01 -6.16012573e-01
1.59206116e+00 -3.91823024e-01 2.30946809e-01 3.98057520e-01
5.36015809e-01 4.23245467e-02 -1.60785723e+00 5.13398707e-01
-2.31772617e-01 -4.34277982e-01 -6.48328140e-02 -8.15946877e-01
-7.21652806e-01 5.86334288e-01 8.08648646e-01 1.66990638e-01
7.84282684e-01 8.50103348e-02 1.41159520e-01 7.57383466e-01
1.38095450e+00 -9.43084836e-01 -1.83800489e-01 6.71512246e-01
9.95913804e-01 -1.20668006e+00 -1.18050501e-01 -6.21330664e-02
-6.81576431e-01 8.57035339e-01 6.91289485e-01 -3.05535346e-01
2.96321720e-01 5.64138591e-01 4.85177487e-01 -2.79837430e-01
-6.14602149e-01 -4.27316368e-01 -5.38184822e-01 1.13402176e+00
-1.44739792e-01 -1.33527834e-02 1.39716506e-01 1.69171140e-01
-6.72813237e-01 -2.09031533e-02 8.47772181e-01 1.24126112e+00
-5.92221677e-01 -8.65410745e-01 -4.52242762e-01 -4.93688792e-01
2.10514382e-01 2.03700960e-01 -2.29664236e-01 1.21378744e+00
5.69553673e-02 8.48129332e-01 -1.02528006e-01 -2.96526134e-01
4.21851903e-01 -3.81264657e-01 6.25551164e-01 -3.42264950e-01
-3.89177650e-01 -3.72754633e-01 2.28837684e-01 -9.35483098e-01
-4.13227499e-01 -4.91365314e-01 -1.48663926e+00 -3.67633373e-01
1.48325175e-01 1.26035795e-01 1.14051354e+00 7.06665635e-01
4.15066868e-01 3.41221094e-01 7.37192258e-02 -1.60695481e+00
-1.32220805e-01 -6.80723011e-01 -3.23058069e-01 -3.16946618e-02
2.94659227e-01 -9.27096844e-01 -4.41724658e-01 -2.35204414e-01] | [4.675886631011963, 0.6931507587432861] |
2f949901-6230-4b61-b3f9-2ad15c415788 | masked-video-distillation-rethinking-masked | 2212.04500 | null | https://arxiv.org/abs/2212.04500v2 | https://arxiv.org/pdf/2212.04500v2.pdf | Masked Video Distillation: Rethinking Masked Feature Modeling for Self-supervised Video Representation Learning | Benefiting from masked visual modeling, self-supervised video representation learning has achieved remarkable progress. However, existing methods focus on learning representations from scratch through reconstructing low-level features like raw pixel RGB values. In this paper, we propose masked video distillation (MVD), a simple yet effective two-stage masked feature modeling framework for video representation learning: firstly we pretrain an image (or video) model by recovering low-level features of masked patches, then we use the resulting features as targets for masked feature modeling. For the choice of teacher models, we observe that students taught by video teachers perform better on temporally-heavy video tasks, while image teachers transfer stronger spatial representations for spatially-heavy video tasks. Visualization analysis also indicates different teachers produce different learned patterns for students. Motivated by this observation, we design a spatial-temporal co-teaching method for MVD. Specifically, we distill student models from both video teachers and image teachers by masked feature modeling. Extensive experimental results demonstrate that video transformers pretrained with spatial-temporal co-teaching outperform models distilled with a single teacher on a multitude of video datasets. Our MVD with vanilla ViT achieves state-of-the-art performance compared with previous supervised or self-supervised methods on several challenging video downstream tasks. For example, with the ViT-Large model, our MVD achieves 86.4% and 76.7% Top-1 accuracy on Kinetics-400 and Something-Something-v2, outperforming VideoMAE by 1.2% and 2.4% respectively. When a larger ViT-Huge model is adopted, MVD achieves the state-of-the-art performance with 77.3% Top-1 accuracy on Something-Something-v2 and 41.1 mAP on AVA v2.2. Code will be available at \url{https://github.com/ruiwang2021/mvd}. | ['Yu-Gang Jiang', 'Lu Yuan', 'Mengchen Liu', 'Xiyang Dai', 'Yinpeng Chen', 'Zuxuan Wu', 'Dongdong Chen', 'Rui Wang'] | 2022-12-08 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Masked_Video_Distillation_Rethinking_Masked_Feature_Modeling_for_Self-Supervised_Video_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Masked_Video_Distillation_Rethinking_Masked_Feature_Modeling_for_Self-Supervised_Video_CVPR_2023_paper.pdf | cvpr-2023-1 | ['action-classification', 'self-supervised-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 1.04097696e-02 5.28951846e-02 -3.93385977e-01 -3.07143748e-01
-1.01592493e+00 -4.30872202e-01 5.56802273e-01 -1.36341542e-01
-2.66902715e-01 3.44862670e-01 1.71170816e-01 -2.49798745e-01
7.25793689e-02 -4.97117758e-01 -1.30769944e+00 -7.76072383e-01
-1.89444143e-02 -7.13432133e-02 3.66649359e-01 -1.05205335e-01
-1.30040394e-02 8.76163244e-02 -1.95557845e+00 7.52728641e-01
6.99288607e-01 8.04373622e-01 4.70356703e-01 7.33453751e-01
6.35434017e-02 1.35913455e+00 -4.40490425e-01 -5.11207767e-02
1.88173473e-01 -4.24279153e-01 -7.37619758e-01 1.75358102e-01
1.00770080e+00 -5.72076440e-01 -8.21503997e-01 7.29895651e-01
3.36079359e-01 4.47328061e-01 5.75710654e-01 -1.18212056e+00
-8.36780548e-01 7.44264841e-01 -8.37458909e-01 4.01138335e-01
3.79251182e-01 4.21122640e-01 6.68235362e-01 -1.22345948e+00
4.28191364e-01 9.68413234e-01 4.53239411e-01 6.79601133e-01
-1.34397650e+00 -7.43998945e-01 3.60359997e-01 4.79262769e-01
-1.58429229e+00 -2.32792258e-01 6.22026324e-01 -6.66773677e-01
8.29569459e-01 2.45484896e-02 7.71191537e-01 1.15265656e+00
-1.50256768e-01 1.13986754e+00 1.31393731e+00 -3.27188760e-01
-1.47823825e-01 3.66307825e-01 -1.42650679e-02 9.60331500e-01
-2.06262514e-01 6.02795482e-02 -9.22536969e-01 4.83186126e-01
8.31883371e-01 2.46484995e-01 -5.50632298e-01 -3.51654619e-01
-1.10678554e+00 5.99142253e-01 6.25569940e-01 2.24972725e-01
-1.61762461e-01 3.61208290e-01 -9.40922648e-03 5.39180696e-01
5.41342318e-01 -9.66611058e-02 -4.03234005e-01 -4.16127652e-01
-1.05888319e+00 -1.24334432e-02 1.35067821e-01 1.04559839e+00
1.02563107e+00 4.68317419e-01 -3.20671707e-01 6.12870216e-01
2.53600568e-01 4.40761715e-01 6.50006652e-01 -9.28608000e-01
3.21107090e-01 4.48042631e-01 -3.43484998e-01 -5.84034920e-01
-1.91682316e-02 -3.76301914e-01 -5.31443059e-01 3.71143430e-01
5.19428968e-01 1.73882663e-01 -1.30554247e+00 1.71919656e+00
4.21590775e-01 8.74700367e-01 1.19938388e-01 8.12610030e-01
1.43780792e+00 8.99336100e-01 3.98151800e-02 1.13281440e-02
1.25101233e+00 -1.31580377e+00 -4.47538763e-01 2.11716909e-02
7.20654011e-01 -5.48411250e-01 1.24293292e+00 5.66088259e-01
-1.26803160e+00 -1.04122019e+00 -8.40936184e-01 -2.73308247e-01
-1.68445602e-01 7.32132122e-02 3.25338274e-01 4.48039293e-01
-1.25753927e+00 6.10483706e-01 -8.66292119e-01 -1.36775762e-01
7.48768866e-01 2.89444029e-01 -5.99312603e-01 -4.45219100e-01
-6.62941337e-01 5.70199251e-01 -9.04181749e-02 -2.11702913e-01
-1.83115137e+00 -1.30292010e+00 -8.54869068e-01 8.82689357e-02
2.95214146e-01 -3.62764716e-01 1.31990385e+00 -1.00271142e+00
-1.39555824e+00 1.01730263e+00 -1.84158795e-02 -2.52664179e-01
4.71633732e-01 -3.68989885e-01 -5.25451191e-02 5.30432045e-01
-4.52159867e-02 1.05113912e+00 1.08722091e+00 -1.25662994e+00
-5.54481149e-01 -1.93264842e-01 1.20806865e-01 2.69501865e-01
-4.95887250e-01 -2.78400123e-01 -6.18412673e-01 -6.29442573e-01
-8.25426504e-02 -9.13266182e-01 -1.42077617e-02 4.99749184e-02
-1.20321698e-02 -3.91202480e-01 9.67408538e-01 -6.22871935e-01
1.04759610e+00 -2.32685828e+00 2.55192399e-01 -2.02855319e-01
4.13636535e-01 3.04913640e-01 -3.12478572e-01 1.60485074e-01
-3.26481998e-01 -5.78367487e-02 3.20768245e-02 -4.56316918e-01
-3.28605860e-01 1.33314148e-01 -2.82262474e-01 5.00938058e-01
1.97461247e-01 8.14600706e-01 -9.53250468e-01 -4.19233650e-01
4.20696944e-01 9.37026560e-01 -7.87975132e-01 5.67164004e-01
-2.58961599e-02 6.91924691e-01 -1.63338378e-01 6.67515695e-01
4.65554178e-01 -3.01723093e-01 -8.03781375e-02 -8.09933543e-02
-2.15657681e-01 2.73760110e-01 -8.62913132e-01 2.10348511e+00
-4.14705068e-01 9.74892676e-01 1.15266092e-01 -1.14460325e+00
5.26340544e-01 4.66505945e-01 5.62173247e-01 -7.23060191e-01
2.17090119e-02 -1.06393479e-01 -2.17400193e-01 -6.57871425e-01
3.21875304e-01 1.61936596e-01 2.91608095e-01 3.53249699e-01
5.63135982e-01 -1.92627102e-01 -2.50305589e-02 5.54444313e-01
1.20201457e+00 5.23047268e-01 -1.55794203e-01 -2.26812467e-01
2.55259991e-01 -5.42986430e-02 2.88152337e-01 5.83154082e-01
-2.08832964e-01 7.89603949e-01 2.88045555e-01 -4.00034934e-01
-5.67224026e-01 -1.12950885e+00 8.53490364e-03 1.47899127e+00
-4.32967320e-02 -5.33061028e-01 -7.56580770e-01 -7.83652604e-01
1.91006772e-02 5.04513204e-01 -8.51908445e-01 -2.02677935e-01
-5.26177406e-01 3.13094854e-02 2.95407176e-01 6.12724185e-01
4.08258438e-01 -8.99618983e-01 -4.55176741e-01 -1.58223435e-01
-1.24923680e-02 -1.18293643e+00 -3.03465515e-01 1.71928734e-01
-7.88370967e-01 -9.27777767e-01 -8.76894414e-01 -9.66436982e-01
8.07998955e-01 1.02272153e+00 1.05829847e+00 3.34084183e-01
-3.53974044e-01 8.94517362e-01 -3.83166879e-01 -2.90352404e-02
5.10168523e-02 -1.47471160e-01 1.65450007e-01 -1.37623167e-02
2.56063938e-01 -7.91739106e-01 -6.27759278e-01 2.17725262e-01
-1.01826000e+00 3.68294954e-01 5.21163762e-01 7.15441763e-01
6.96404934e-01 -2.82656580e-01 1.59437016e-01 -6.93450689e-01
-1.44614026e-01 -6.68001354e-01 -4.36769158e-01 1.38507664e-01
-3.20887327e-01 -1.42978042e-01 4.67030495e-01 -7.96376050e-01
-9.13163364e-01 2.02927157e-01 6.49008527e-02 -1.35317767e+00
-2.69476831e-01 2.57089764e-01 1.06668785e-01 -1.43800959e-01
6.33067012e-01 4.19989794e-01 -6.52903467e-02 -4.63186681e-01
5.02644658e-01 3.79983872e-01 5.98198473e-01 -8.02913010e-01
1.09348106e+00 3.50722134e-01 -4.02544379e-01 -9.62629080e-01
-9.72392738e-01 -3.94500911e-01 -6.35700822e-01 -5.10484278e-01
1.21295500e+00 -1.58432078e+00 -6.51606023e-01 4.22083229e-01
-6.52963400e-01 -9.47633684e-01 -3.68445128e-01 6.47997022e-01
-5.31499505e-01 -2.32857503e-02 -5.25846004e-01 -4.65483546e-01
2.03677312e-01 -1.44457209e+00 7.99824595e-01 3.64930749e-01
1.41306534e-01 -8.34224522e-01 9.42340703e-04 8.49427521e-01
3.35350752e-01 -8.23368058e-02 5.83774328e-01 -4.71966118e-01
-9.20640290e-01 3.17153662e-01 -7.81905279e-02 3.67071301e-01
3.10515165e-02 6.95741996e-02 -1.45214057e+00 -4.23049569e-01
-2.25312725e-01 -8.63705158e-01 1.01326919e+00 3.25947583e-01
1.65166855e+00 -1.55269176e-01 -5.08732870e-02 1.03953850e+00
1.27958965e+00 -1.31864354e-01 4.11693096e-01 1.65112868e-01
1.25469911e+00 6.62563324e-01 3.75529319e-01 2.77217060e-01
6.08404398e-01 7.18754530e-01 6.17885530e-01 -1.50111569e-02
-6.19035304e-01 -5.46718359e-01 8.09914351e-01 1.04167032e+00
-2.52000749e-01 9.99670848e-02 -8.11071217e-01 8.21627140e-01
-1.62588227e+00 -1.04258585e+00 -1.98612452e-01 2.05363512e+00
9.60988700e-01 -1.49890855e-01 2.01700643e-01 2.27000266e-02
3.07274759e-01 2.71176666e-01 -2.91914105e-01 -7.27687702e-02
1.97981894e-01 3.24994445e-01 3.42959404e-01 3.95058006e-01
-1.06941283e+00 1.06909466e+00 5.03732681e+00 8.25862527e-01
-1.23224378e+00 2.45179266e-01 6.83830261e-01 -4.86302167e-01
-3.28258336e-01 -1.70982376e-01 -7.33953476e-01 3.77903730e-01
9.56435502e-01 1.17101043e-01 4.64337945e-01 8.78786504e-01
1.23160891e-01 5.65593727e-02 -1.18699503e+00 1.18510926e+00
2.05961898e-01 -1.39552820e+00 -2.39811908e-03 -1.45054273e-02
1.19279766e+00 8.03171992e-02 5.19119978e-01 7.12384105e-01
3.69756192e-01 -1.22839320e+00 7.57885993e-01 2.24989042e-01
8.86500776e-01 -4.65304166e-01 1.75044447e-01 1.41987413e-01
-1.37619722e+00 -5.64962476e-02 -2.03353077e-01 -4.11872342e-02
-2.62935758e-01 1.92268774e-01 -5.95838249e-01 2.38395736e-01
1.30104065e+00 1.05969691e+00 -8.52833092e-01 8.06562603e-01
-5.22147119e-01 1.11168766e+00 -7.87258595e-02 4.11613941e-01
4.10885721e-01 -2.89866813e-02 1.49816960e-01 1.10176170e+00
3.17942619e-01 3.36339921e-01 2.88161337e-01 6.24875844e-01
-2.66327411e-01 -3.69503945e-02 -8.33664596e-01 1.59599841e-01
3.86749953e-01 1.22640407e+00 -4.15429771e-01 -4.49247032e-01
-8.08208406e-01 8.80088151e-01 5.78631699e-01 5.64488649e-01
-1.00130939e+00 4.81381156e-02 7.42465794e-01 3.20147276e-01
6.82743132e-01 -1.48430720e-01 2.81762183e-01 -1.24502659e+00
-2.04817593e-01 -1.07081604e+00 3.09416860e-01 -1.03489852e+00
-8.24023426e-01 4.51788396e-01 2.61756599e-01 -1.38141656e+00
-1.55937985e-01 -6.58907712e-01 -8.30797076e-01 3.81797969e-01
-1.60918808e+00 -1.18168175e+00 -7.00008512e-01 1.09763050e+00
8.79386902e-01 -2.16633320e-01 4.76154268e-01 2.78301746e-01
-7.95874417e-01 7.21322119e-01 8.50857608e-03 1.27962038e-01
7.46943772e-01 -1.23638129e+00 -2.91836653e-02 7.45310426e-01
6.51757479e-01 3.42515707e-01 4.06227976e-01 -1.86558738e-01
-1.71277273e+00 -1.08152807e+00 3.95747095e-01 -5.43795824e-01
6.13480330e-01 -2.85496861e-01 -1.14415991e+00 9.96478438e-01
5.82061291e-01 5.43817163e-01 6.59619629e-01 -2.90577650e-01
-6.47871792e-01 -1.00832991e-01 -9.98477340e-01 4.16330695e-01
1.01766372e+00 -7.18068898e-01 -5.01127541e-01 3.87889624e-01
9.31694329e-01 -3.98533702e-01 -9.41483498e-01 1.99446321e-01
4.08911973e-01 -1.07887101e+00 1.04944849e+00 -5.78515172e-01
6.46648347e-01 -2.38197386e-01 -2.49323979e-01 -1.33439529e+00
-2.01007947e-01 -5.87698698e-01 -3.87171924e-01 1.13237631e+00
1.81608140e-01 -9.71595943e-02 9.66273844e-01 2.26330996e-01
-3.88698310e-01 -1.00008881e+00 -6.53771281e-01 -5.87730408e-01
1.73516288e-01 -5.17105579e-01 1.91620156e-01 1.24057913e+00
-2.16163114e-01 1.23419151e-01 -3.22431862e-01 1.92979023e-01
6.56333447e-01 -2.81994920e-02 1.04975605e+00 -8.41522932e-01
-4.66971278e-01 -3.84353608e-01 -3.65260869e-01 -1.34467065e+00
2.22369462e-01 -8.71589780e-01 -1.09803915e-01 -1.38202667e+00
2.82102287e-01 -2.51570016e-01 -4.09934372e-01 7.30175614e-01
-2.57478148e-01 5.48286080e-01 3.29262257e-01 8.25929493e-02
-6.80227637e-01 6.40878916e-01 1.42495704e+00 -2.95422018e-01
-1.15091890e-01 -1.57699242e-01 -4.59242254e-01 6.46651030e-01
6.71746910e-01 -4.63432878e-01 -7.80290604e-01 -6.80107117e-01
-1.74390331e-01 8.26451853e-02 4.08516377e-01 -1.06076717e+00
1.25869632e-01 -1.75654143e-01 5.50733566e-01 -4.84064758e-01
5.09881079e-01 -6.89184546e-01 -2.31868371e-01 4.14872557e-01
-3.23933959e-01 9.12021399e-02 3.59491736e-01 4.67763275e-01
-2.86699384e-01 -1.20711505e-01 7.07075298e-01 -1.50676802e-01
-9.41904485e-01 3.27403009e-01 -5.03993094e-01 1.64203763e-01
1.21057749e+00 -2.17192680e-01 -5.83451807e-01 -6.26616657e-01
-7.93910623e-01 2.70395517e-01 3.20540041e-01 4.94306922e-01
9.93122339e-01 -1.31608796e+00 -8.06934059e-01 4.22527999e-01
8.29474181e-02 2.99141586e-01 5.52840829e-01 1.02984786e+00
-5.37297130e-01 8.95084515e-02 -2.56761193e-01 -1.06299782e+00
-1.52446902e+00 5.77366769e-01 9.47461203e-02 4.47793491e-02
-6.64197385e-01 1.39113772e+00 5.81709027e-01 -3.26966554e-01
5.63187122e-01 -4.29987103e-01 -2.14552984e-01 -2.01352425e-02
7.41541445e-01 2.61054486e-01 -1.50474802e-01 -7.68702149e-01
-1.22966006e-01 7.68936038e-01 -2.08120689e-01 9.38255414e-02
1.55103254e+00 1.53056756e-01 3.90701950e-01 4.34603572e-01
1.35451686e+00 1.21155391e-02 -1.85206401e+00 -2.88720518e-01
-4.69036877e-01 -5.71036100e-01 2.46807784e-01 -4.14024591e-01
-1.54160190e+00 1.26924336e+00 7.02552497e-01 -1.09807491e-01
1.10309827e+00 1.95028067e-01 3.37454259e-01 1.41864821e-01
1.60682037e-01 -5.56239963e-01 6.91419899e-01 2.92765498e-01
5.78504324e-01 -1.44758129e+00 1.52856305e-01 -2.11593121e-01
-7.83715844e-01 7.09676325e-01 1.16078711e+00 -1.05103329e-01
6.11220956e-01 -2.53604259e-02 1.77726433e-01 -1.88500121e-01
-1.11950254e+00 -3.22856128e-01 4.97846007e-01 5.77320755e-01
5.40784478e-01 -6.36371821e-02 5.06695807e-01 3.24869245e-01
-1.14370711e-01 -2.89809614e-01 6.42206728e-01 9.85561252e-01
-4.51720178e-01 -7.11607039e-01 -2.94213235e-01 1.01696752e-01
-3.36105078e-01 -1.02156647e-01 -3.79567854e-02 9.23066974e-01
7.19470903e-02 8.56597722e-01 1.83515221e-01 -7.12256312e-01
1.53252244e-01 6.36728182e-02 8.43128383e-01 -8.28207970e-01
-8.54787648e-01 3.32390927e-02 -3.99957746e-01 -8.16536188e-01
-5.13654947e-01 -4.56801683e-01 -1.25172985e+00 -3.09614182e-01
-9.88269132e-03 1.48423925e-01 5.46414554e-01 6.27660573e-01
1.75113052e-01 7.56787717e-01 6.53121293e-01 -1.12483776e+00
-3.08239043e-01 -9.74375129e-01 -3.14313114e-01 4.28813577e-01
5.03978014e-01 -8.12866747e-01 -4.46220338e-01 2.69907385e-01] | [9.354334831237793, 0.8955991864204407] |
5cb7cd3e-1900-4b48-925d-d98a8fd52d7b | outpainting-by-queries | 2207.05312 | null | https://arxiv.org/abs/2207.05312v1 | https://arxiv.org/pdf/2207.05312v1.pdf | Outpainting by Queries | Image outpainting, which is well studied with Convolution Neural Network (CNN) based framework, has recently drawn more attention in computer vision. However, CNNs rely on inherent inductive biases to achieve effective sample learning, which may degrade the performance ceiling. In this paper, motivated by the flexible self-attention mechanism with minimal inductive biases in transformer architecture, we reframe the generalised image outpainting problem as a patch-wise sequence-to-sequence autoregression problem, enabling query-based image outpainting. Specifically, we propose a novel hybrid vision-transformer-based encoder-decoder framework, named \textbf{Query} \textbf{O}utpainting \textbf{TR}ansformer (\textbf{QueryOTR}), for extrapolating visual context all-side around a given image. Patch-wise mode's global modeling capacity allows us to extrapolate images from the attention mechanism's query standpoint. A novel Query Expansion Module (QEM) is designed to integrate information from the predicted queries based on the encoder's output, hence accelerating the convergence of the pure transformer even with a relatively small dataset. To further enhance connectivity between each patch, the proposed Patch Smoothing Module (PSM) re-allocates and averages the overlapped regions, thus providing seamless predicted images. We experimentally show that QueryOTR could generate visually appealing results smoothly and realistically against the state-of-the-art image outpainting approaches. | ['Rui Zhang', 'Jie Sun', 'Kaizhu Huang', 'Xi Yang', 'Penglei Gao', 'Kai Yao'] | 2022-07-12 | null | null | null | null | ['image-outpainting'] | ['computer-vision'] | [ 5.92172563e-01 2.51683623e-01 -9.82315913e-02 -2.48152375e-01
-8.33273172e-01 -2.65007224e-02 4.55291688e-01 -3.66981775e-01
-2.01895088e-01 6.42364442e-01 2.14702800e-01 -1.13303833e-01
1.38783738e-01 -8.57584178e-01 -1.28210223e+00 -5.51628709e-01
5.29090166e-01 -1.02205679e-01 1.55474544e-01 -1.81851447e-01
2.04494670e-01 3.34774226e-01 -1.45633793e+00 5.26163876e-01
1.23697686e+00 1.20689046e+00 9.86152232e-01 6.33945942e-01
-2.16988787e-01 1.21801758e+00 -5.16099513e-01 -5.11351526e-01
1.62580520e-01 -5.38150370e-01 -5.65736592e-01 3.26292589e-02
7.88515091e-01 -6.96543455e-01 -7.81694651e-01 9.64380860e-01
4.95878696e-01 4.21316810e-02 3.19700718e-01 -9.24933255e-01
-1.32849395e+00 3.45729798e-01 -6.35122001e-01 6.27093315e-02
1.30867943e-01 4.70002234e-01 8.82196784e-01 -1.15781784e+00
6.46139562e-01 1.16693354e+00 6.54261827e-01 5.44386506e-01
-1.39531243e+00 -5.70696414e-01 2.10172102e-01 4.01302934e-01
-1.40948772e+00 -3.97791892e-01 1.12941706e+00 -1.00691482e-01
1.02155876e+00 5.15500426e-01 6.74390078e-01 1.22823346e+00
3.75676036e-01 1.07449007e+00 9.21508074e-01 -3.55125666e-01
7.93990716e-02 1.12536602e-01 -5.25952160e-01 4.58732069e-01
-4.76886272e-01 3.08911145e-01 -6.40368223e-01 3.31436306e-01
1.28723657e+00 3.01019490e-01 -5.01335502e-01 -3.44524556e-03
-1.13535464e+00 8.04731786e-01 9.00740802e-01 -3.33209150e-03
-6.44989967e-01 3.69767874e-01 3.28177720e-01 3.65231872e-01
6.73281789e-01 3.58132094e-01 -3.02359760e-01 2.11045906e-01
-1.29866970e+00 2.88856626e-01 1.90911800e-01 1.20016444e+00
9.40687001e-01 4.10600960e-01 -6.34911954e-01 1.07344031e+00
7.58987814e-02 3.91783774e-01 4.08186078e-01 -1.04589963e+00
4.44347799e-01 4.80782658e-01 -1.68386940e-03 -9.38399315e-01
1.79761797e-01 -6.88054442e-01 -1.13758707e+00 1.58335969e-01
-8.39909613e-02 8.86793658e-02 -9.03218508e-01 1.73611772e+00
9.35499817e-02 3.19786042e-01 -9.29127336e-02 9.27381754e-01
8.57530773e-01 8.46574664e-01 1.60781443e-01 -2.39615962e-01
1.22904503e+00 -1.42223930e+00 -5.83280444e-01 -4.24257308e-01
2.61479825e-01 -8.51921797e-01 1.36412978e+00 2.48256385e-01
-1.26608026e+00 -1.04937482e+00 -9.15641785e-01 -6.24751151e-01
-1.17148094e-01 2.42083564e-01 3.01499605e-01 -3.14644212e-03
-1.40160966e+00 5.37637651e-01 -4.15756255e-01 -1.92097262e-01
6.76229179e-01 9.45210084e-02 -3.40398937e-01 -2.62717009e-01
-1.12265968e+00 9.28096473e-01 2.63118744e-01 2.77271569e-01
-1.14533687e+00 -8.98263872e-01 -8.90918255e-01 2.22098097e-01
3.55099082e-01 -9.80204344e-01 1.17402875e+00 -1.39530027e+00
-1.50949419e+00 6.16775334e-01 -3.91140044e-01 -7.87324429e-01
5.61891794e-01 -3.74631077e-01 -1.38391092e-01 2.77906239e-01
1.63342044e-01 1.31100202e+00 1.39862537e+00 -1.33623493e+00
-3.11634362e-01 -1.07742667e-01 8.53350759e-02 2.66461521e-01
-2.42123693e-01 -1.80236667e-01 -5.96940219e-01 -1.26010501e+00
-2.17831567e-01 -5.18299937e-01 -2.71693617e-01 3.12093884e-01
-3.35797817e-01 3.67451310e-02 1.01226008e+00 -9.51824963e-01
1.24496591e+00 -2.12770247e+00 1.64969906e-01 -3.87783051e-01
2.43472755e-01 3.97587568e-01 -4.76488620e-01 5.42570531e-01
-2.56604075e-01 -2.14470282e-01 -3.44592154e-01 -7.12091923e-01
-3.57160717e-01 3.62674952e-01 -6.77726805e-01 5.02707325e-02
5.99208057e-01 1.33685243e+00 -7.19694614e-01 -5.21497071e-01
3.97233516e-01 5.63068986e-01 -8.56382847e-01 4.77584481e-01
-5.43120384e-01 4.72243398e-01 -2.70337969e-01 5.92825592e-01
8.63278449e-01 -1.41579553e-01 -3.87738615e-01 -4.92012292e-01
-7.47979134e-02 -1.20651387e-01 -5.02924442e-01 2.11954927e+00
-8.47461760e-01 7.61875153e-01 2.97215879e-02 -1.06701183e+00
1.09761727e+00 1.38706341e-01 3.67453694e-01 -1.10171402e+00
-1.82681069e-01 -1.64377727e-02 -3.46065879e-01 -6.14216268e-01
6.11003757e-01 -1.01403758e-01 4.01848584e-01 9.46658775e-02
2.70064771e-01 -6.08959384e-02 -2.54130661e-01 1.80464968e-01
9.11902428e-01 5.58646202e-01 -9.81395617e-02 1.17123155e-02
5.00202298e-01 -6.63310215e-02 2.03219250e-01 7.74632096e-01
1.14062823e-01 1.07805157e+00 1.86635002e-01 -4.15083081e-01
-1.37207520e+00 -1.06638801e+00 8.48950595e-02 1.11559248e+00
2.10529104e-01 -2.45902419e-01 -8.82641613e-01 -4.81351137e-01
-2.04675585e-01 8.36155176e-01 -7.53824592e-01 -4.67099011e-01
-5.19100845e-01 -2.33508691e-01 4.49311495e-01 6.00095868e-01
8.87627482e-01 -1.38668072e+00 -5.94125569e-01 3.91740292e-01
-3.79422814e-01 -8.45474899e-01 -9.36478317e-01 5.10501042e-02
-8.94346118e-01 -5.91206253e-01 -1.20080793e+00 -8.31126332e-01
6.73919559e-01 2.98202395e-01 1.11363304e+00 -9.68078077e-02
-2.77389646e-01 2.55344629e-01 -3.24065149e-01 -1.89297289e-01
-2.99897492e-01 -9.85606834e-02 -4.65081781e-01 2.39919901e-01
-2.64330134e-02 -8.14125538e-01 -1.08606541e+00 4.31787074e-02
-1.24735010e+00 5.42864680e-01 1.07236421e+00 1.19606185e+00
7.56033719e-01 -3.63496065e-01 7.06953704e-01 -5.76129317e-01
5.85747421e-01 -3.80203515e-01 -2.40661189e-01 2.37083301e-01
-4.92857099e-01 -1.62610970e-02 9.59940493e-01 -6.78206444e-01
-1.22913444e+00 -1.09935015e-01 -4.47162569e-01 -1.14405823e+00
-4.47658524e-02 3.58544618e-01 -1.08775258e-01 -1.92514345e-01
6.16478801e-01 1.01683116e+00 1.79646045e-01 -5.32612383e-01
5.22379339e-01 4.87077296e-01 6.93951130e-01 -3.38647515e-01
4.97207373e-01 3.57667595e-01 -2.13481352e-01 -7.06612349e-01
-7.62743115e-01 -8.33103359e-02 -2.06528783e-01 -2.60020107e-01
8.34005594e-01 -1.10575426e+00 -6.32189810e-01 3.62869233e-01
-1.54026330e+00 -4.78187174e-01 -5.29234946e-01 1.39739469e-01
-7.95835793e-01 4.15839404e-01 -7.02235162e-01 -6.21468902e-01
-5.82769156e-01 -1.29363751e+00 1.30337620e+00 2.03638464e-01
8.24879948e-03 -7.19554663e-01 -2.32825175e-01 3.04187417e-01
5.94126523e-01 -2.02921331e-02 7.43651032e-01 -1.47928327e-01
-9.24670100e-01 1.68929234e-01 -6.14900947e-01 8.46955895e-01
-2.11297005e-01 -4.62804765e-01 -1.21772623e+00 -1.92264870e-01
3.20996433e-01 -2.43957832e-01 1.06411564e+00 4.92303669e-01
1.44024634e+00 -7.29023933e-01 -1.06791183e-01 8.51632774e-01
1.66422594e+00 6.98570302e-03 1.12914407e+00 1.46476850e-01
8.19244802e-01 4.79139835e-01 5.30085325e-01 3.72311473e-01
2.94630945e-01 6.45006597e-01 7.33440936e-01 -4.29437339e-01
-4.76169169e-01 -7.23433256e-01 4.06142712e-01 5.82167506e-01
1.94009602e-01 -2.33749643e-01 -2.65305370e-01 5.34385085e-01
-1.72560894e+00 -9.14237380e-01 2.43311614e-01 2.00729656e+00
1.00313580e+00 -1.20462552e-01 -3.08304399e-01 -3.63985240e-01
6.00778699e-01 3.10613781e-01 -7.62478948e-01 -3.87559056e-01
-9.40910503e-02 4.07215863e-01 2.96694666e-01 4.70371395e-01
-7.68221140e-01 1.07853603e+00 5.49760914e+00 1.44082665e+00
-1.31637216e+00 3.07286799e-01 9.17807937e-01 8.55190400e-03
-4.49037611e-01 3.56925256e-03 -4.00355011e-01 6.23042524e-01
6.39863193e-01 5.27320206e-02 6.75889015e-01 8.70266795e-01
4.14863497e-01 -7.53555968e-02 -1.05146122e+00 1.14102209e+00
1.48996115e-01 -1.67256856e+00 5.66569209e-01 -1.13088317e-01
6.32572472e-01 2.46278686e-03 2.61911780e-01 4.61350828e-01
-2.01499328e-01 -1.11477137e+00 9.04585838e-01 8.85751963e-01
1.12273824e+00 -6.15238786e-01 4.10500884e-01 4.72160429e-01
-9.52234209e-01 -1.27047613e-01 -5.12246728e-01 4.48035486e-02
8.19661692e-02 5.66138148e-01 -5.15046299e-01 6.78465843e-01
7.43980527e-01 7.14632392e-01 -5.81599474e-01 8.40069413e-01
4.06398624e-02 3.72391254e-01 -6.53009638e-02 3.59852016e-01
2.95537829e-01 -2.27355033e-01 4.52693135e-01 1.03769720e+00
4.23809558e-01 -8.46610889e-02 -3.85881104e-02 1.43064785e+00
-2.01981306e-01 3.61372493e-02 -6.66289330e-01 2.99805611e-01
3.88448894e-01 1.09862852e+00 -2.11495265e-01 -4.70512629e-01
-4.72946107e-01 1.47875965e+00 3.56738001e-01 6.84455633e-01
-1.01977944e+00 -3.50982815e-01 5.30924499e-01 1.83339670e-01
5.65929711e-01 9.51157734e-02 -3.51543635e-01 -1.07722068e+00
2.42142364e-01 -7.89240301e-01 -2.09027156e-01 -1.33565497e+00
-1.13467312e+00 7.66871274e-01 -2.46937916e-01 -1.40505767e+00
-1.20264538e-01 -2.04216808e-01 -6.82687879e-01 1.08085108e+00
-1.54327011e+00 -1.55260706e+00 -3.59708011e-01 7.54858375e-01
1.04885936e+00 3.82237993e-02 6.08394444e-01 3.41303796e-01
-2.13665202e-01 7.58704901e-01 -7.92516395e-03 -1.18303396e-01
6.77495182e-01 -8.84440124e-01 2.57023722e-01 8.21130693e-01
2.25201137e-02 5.22880554e-01 6.68792486e-01 -6.31163001e-01
-1.48893595e+00 -1.54781842e+00 5.35435319e-01 -2.81948261e-02
3.47382784e-01 -2.37808183e-01 -9.80077565e-01 6.54594779e-01
6.26331270e-01 1.33159295e-01 1.08649477e-01 -5.50291955e-01
-3.77648562e-01 -4.13855046e-01 -1.08624673e+00 8.75240982e-01
1.02353609e+00 -7.25599527e-01 -3.04018229e-01 2.16107637e-01
1.20239258e+00 -2.47278184e-01 -8.37746322e-01 3.25126052e-01
3.83925855e-01 -1.28696930e+00 1.17088282e+00 -1.41940191e-01
9.14154053e-01 -2.58892328e-01 -1.13277256e-01 -1.15955520e+00
-3.18680674e-01 -6.87084496e-01 -1.50348186e-01 1.27357495e+00
3.93078364e-02 -4.72458184e-01 7.03468204e-01 1.94252715e-01
-4.26369548e-01 -1.06871200e+00 -9.00238216e-01 -5.23329914e-01
-6.57064840e-02 -5.09597838e-01 4.04899478e-01 5.27311206e-01
-4.58788335e-01 3.29954624e-01 -8.05223346e-01 -1.55179441e-01
4.56733465e-01 1.04848765e-01 6.59989059e-01 -5.40419459e-01
-6.70476317e-01 -4.01668370e-01 -9.91849154e-02 -1.65397894e+00
-1.12237811e-01 -7.01417863e-01 1.38123915e-01 -1.34689879e+00
1.37905210e-01 -1.69431090e-01 -8.80841389e-02 3.35136682e-01
-1.53671518e-01 5.35718143e-01 4.01043713e-01 3.05639952e-01
-3.25457782e-01 1.05313647e+00 1.77926874e+00 -3.05112153e-01
1.07486576e-01 -2.29642510e-01 -7.05842793e-01 3.08635324e-01
3.91144633e-01 -1.49105489e-01 -7.05669343e-01 -7.20606744e-01
-9.22052339e-02 4.18768972e-01 8.11812758e-01 -9.66839612e-01
2.60566741e-01 -2.84593627e-02 5.66281021e-01 -6.32828116e-01
5.59319317e-01 -7.67092228e-01 2.56061137e-01 2.65806466e-01
-4.98893410e-01 3.52126211e-02 1.89014390e-01 7.56036222e-01
-6.15603447e-01 -3.18460241e-02 7.86489248e-01 -3.24642211e-01
-7.21535027e-01 5.48778892e-01 -5.40152006e-02 -1.53052256e-01
9.21490729e-01 -3.35955828e-01 -1.38015792e-01 -4.96960372e-01
-6.72234595e-01 5.43844439e-02 4.40406263e-01 3.50437224e-01
1.04373288e+00 -1.39387238e+00 -7.05088377e-01 3.95367503e-01
2.32692901e-02 2.19567090e-01 9.05952036e-01 8.55937421e-01
-4.96820509e-01 3.20058286e-01 -2.96692491e-01 -7.02672482e-01
-7.09574699e-01 9.94255006e-01 1.75784737e-01 -2.54798174e-01
-7.25979507e-01 9.34368908e-01 8.34840834e-01 -1.19703375e-01
1.23759851e-01 -4.00199711e-01 1.21072032e-01 -2.11354136e-01
4.77881908e-01 2.85260519e-03 -1.71389371e-01 -4.63297218e-01
9.92215574e-02 4.09841508e-01 -3.64737272e-01 7.74033442e-02
1.17500997e+00 -2.94838965e-01 -1.54972523e-01 -6.38033077e-03
1.28507018e+00 -3.76630515e-01 -1.73362041e+00 -2.36543387e-01
-4.74631250e-01 -5.17378688e-01 7.27369636e-03 -7.33014524e-01
-1.16712129e+00 1.04611218e+00 5.00834703e-01 -1.95041038e-02
1.52373922e+00 -1.36422515e-01 1.12213564e+00 -1.00732528e-01
1.82495400e-01 -8.18429530e-01 3.38200361e-01 2.03007475e-01
1.50833559e+00 -1.10747898e+00 -3.06436956e-01 -2.57302463e-01
-6.69036984e-01 9.77830470e-01 7.93283045e-01 -5.01833200e-01
3.13828021e-01 5.84979318e-02 -1.78462893e-01 2.46621385e-01
-8.76163960e-01 -5.14300726e-02 2.21799627e-01 7.44192600e-01
1.30492195e-01 -2.49394044e-01 -3.62680145e-02 4.93963182e-01
4.85556833e-02 3.35788846e-01 1.21823758e-01 4.95078564e-01
-2.87113339e-01 -8.56588542e-01 -2.83134162e-01 4.57146287e-01
-1.59771129e-01 -6.78242683e-01 6.40555173e-02 5.81002235e-01
3.42917562e-01 5.48521996e-01 9.59697291e-02 -4.05786365e-01
1.39779195e-01 -1.78105667e-01 4.89585221e-01 -3.34465474e-01
-5.16999483e-01 2.68744469e-01 -3.40759575e-01 -7.81564057e-01
-1.98699251e-01 -2.30722547e-01 -7.61610925e-01 -8.24496448e-02
-1.40625775e-01 -2.91370958e-01 2.70853639e-01 8.51419449e-01
7.90314496e-01 7.17300653e-01 5.33264160e-01 -1.07661200e+00
-5.48415184e-01 -1.23422503e+00 -3.38487744e-01 2.80975699e-01
4.39457595e-01 -4.26887572e-01 5.79957478e-02 2.60627776e-01] | [11.294840812683105, -1.128976583480835] |
ddd11eda-9de6-4670-87c1-1e9c32e6f8be | transformers-meet-directed-graphs | 2302.00049 | null | https://arxiv.org/abs/2302.00049v2 | https://arxiv.org/pdf/2302.00049v2.pdf | Transformers Meet Directed Graphs | Transformers were originally proposed as a sequence-to-sequence model for text but have become vital for a wide range of modalities, including images, audio, video, and undirected graphs. However, transformers for directed graphs are a surprisingly underexplored topic, despite their applicability to ubiquitous domains, including source code and logic circuits. In this work, we propose two direction- and structure-aware positional encodings for directed graphs: (1) the eigenvectors of the Magnetic Laplacian - a direction-aware generalization of the combinatorial Laplacian; (2) directional random walk encodings. Empirically, we show that the extra directionality information is useful in various downstream tasks, including correctness testing of sorting networks and source code understanding. Together with a data-flow-centric graph construction, our model outperforms the prior state of the art on the Open Graph Benchmark Code2 relatively by 14.7%. | ['Cosmin Paduraru', 'Stephan Günnemann', 'Ali Taylan Cemgil', 'Daniel Mankowitz', 'Yujia Li', 'Simon Geisler'] | 2023-01-31 | null | null | null | null | ['graph-property-prediction'] | ['graphs'] | [ 5.45140505e-01 1.67854801e-01 -4.89616990e-01 -1.96918696e-01
-3.61618638e-01 -1.01002681e+00 3.52740705e-01 5.25145233e-01
3.21446359e-01 4.62678462e-01 3.39214861e-01 -1.06177938e+00
-3.22244108e-01 -9.36923444e-01 -9.05441284e-01 -4.34053987e-01
-5.30455351e-01 3.65615487e-01 5.42504072e-01 -3.39857072e-01
6.77522659e-01 3.79652858e-01 -1.27645004e+00 3.14109623e-01
7.39153922e-01 8.24372888e-01 -2.92859431e-02 8.90810609e-01
-2.55730063e-01 1.09845006e+00 -2.39847936e-02 -5.35205841e-01
-1.02596886e-01 -5.02598047e-01 -1.03915095e+00 -1.18671075e-01
5.32296777e-01 -2.90313060e-03 -6.31990075e-01 1.12419891e+00
4.16059613e-01 -3.68742853e-01 4.18788046e-01 -1.28015530e+00
-6.99261725e-01 1.06744850e+00 -8.30480397e-01 1.55815765e-01
5.24442971e-01 1.48192734e-01 1.55314994e+00 -4.42027658e-01
9.29613113e-01 9.26168561e-01 6.98025286e-01 2.35402226e-01
-1.26532733e+00 -5.21677792e-01 8.78893360e-02 4.53767121e-01
-1.14795029e+00 -4.66333270e-01 8.76554847e-01 -4.03058857e-01
1.22321117e+00 2.12008610e-01 7.15052843e-01 8.60816002e-01
3.68492723e-01 6.27572477e-01 7.39733160e-01 -1.53609529e-01
1.53909162e-01 -5.66849470e-01 1.73435032e-01 1.13387978e+00
6.90169275e-01 -3.05260032e-01 -9.64961171e-01 -2.31992781e-01
5.72320879e-01 -2.88326532e-01 -4.75546688e-01 -9.08248365e-01
-1.19385076e+00 5.19032061e-01 5.89708626e-01 1.18058279e-01
-2.37018405e-03 9.05637741e-01 6.67448640e-01 3.41183782e-01
2.12886557e-01 3.46380591e-01 -3.65505427e-01 -4.95159179e-01
-6.99691057e-01 8.59132130e-03 1.01867831e+00 1.35817456e+00
7.49103189e-01 5.29102273e-02 -8.41037557e-02 3.65726769e-01
5.45117974e-01 5.89476824e-01 -9.68889371e-02 -4.56560254e-01
7.35634804e-01 9.43903625e-01 -7.63688922e-01 -1.23279893e+00
-4.97825116e-01 -4.64230478e-01 -8.65666270e-01 -5.65737963e-01
4.38135296e-01 2.00955942e-01 -6.43653929e-01 1.70760214e+00
-1.54208411e-02 8.96753892e-02 -3.97948921e-01 5.78273833e-01
7.83455312e-01 3.94272834e-01 -3.61843824e-01 2.62200963e-02
1.11585343e+00 -8.23162735e-01 -5.63718021e-01 -3.89063805e-01
9.21216667e-01 -4.23128575e-01 8.68777514e-01 3.76180768e-01
-1.06126666e+00 6.82940558e-02 -1.04310989e+00 -2.44312629e-01
-1.49804011e-01 -3.76862615e-01 1.06129742e+00 8.80383074e-01
-1.40425861e+00 4.59305227e-01 -6.94603205e-01 -4.83867079e-01
4.90421206e-01 4.66570735e-01 -2.20696479e-01 -3.95485431e-01
-6.26693487e-01 5.78058213e-02 -1.04028538e-01 -1.96010202e-01
-8.34114134e-01 -7.47413456e-01 -8.28929603e-01 3.54604006e-01
3.73848766e-01 -6.27626717e-01 7.43707955e-01 -2.74790734e-01
-1.28592789e+00 8.48916233e-01 -3.14653516e-01 -3.84801745e-01
1.67704105e-01 4.56529379e-01 -2.28929237e-01 1.93082228e-01
-1.24493249e-01 1.86559871e-01 5.50188124e-01 -9.19623196e-01
-1.05223797e-01 -4.91682589e-01 1.55828610e-01 -7.97668323e-02
-6.17189646e-01 -4.33052659e-01 -7.39409029e-01 -4.71666902e-01
3.02761078e-01 -1.06498528e+00 4.97675575e-02 1.10079356e-01
-8.93418908e-01 6.98718131e-02 7.65376985e-01 -4.66772258e-01
1.73269391e+00 -2.16872954e+00 5.12746513e-01 5.96644163e-01
7.92334199e-01 -2.95490265e-01 -3.83239239e-02 6.91936433e-01
-1.60276085e-01 4.08082604e-01 -4.24553692e-01 2.99595773e-01
6.33926615e-02 -1.66773289e-01 -1.19297735e-01 5.61507642e-01
-3.96771878e-02 1.30722797e+00 -1.12284863e+00 -2.56880224e-01
-2.32268170e-01 6.48569465e-02 -9.64804053e-01 -2.64502496e-01
-4.07450974e-01 6.41469881e-02 -3.47187310e-01 6.94058359e-01
6.99489832e-01 -7.99202561e-01 8.01538706e-01 -2.18002915e-01
1.82764143e-01 5.15679300e-01 -7.03794301e-01 2.10904002e+00
-1.91641033e-01 1.10801816e+00 1.79251194e-01 -9.05041635e-01
7.19567657e-01 -1.28807083e-01 4.96244013e-01 -6.22674942e-01
1.10643255e-02 2.01083899e-01 3.51965666e-01 -4.51240629e-01
5.89346051e-01 2.88522691e-01 -2.40556568e-01 6.32825673e-01
-7.83167928e-02 -4.64002073e-01 3.47719908e-01 9.77848172e-01
2.04311609e+00 -1.06920682e-01 -1.04945377e-01 -5.17266750e-01
1.89985856e-01 -1.98643014e-01 1.84203580e-01 5.40229082e-01
5.15237125e-03 5.42352974e-01 1.28863025e+00 5.36305457e-02
-8.77806067e-01 -9.80494976e-01 1.11027665e-01 9.05039012e-01
2.77054101e-01 -1.11605513e+00 -6.53331876e-01 -6.34485722e-01
5.23690321e-02 3.34097475e-01 -3.53477836e-01 -4.88333583e-01
-3.90178680e-01 -5.85633457e-01 9.12098169e-01 5.68009138e-01
3.21728468e-01 -1.50997669e-01 -1.59344658e-01 -2.71774624e-02
-1.46765202e-01 -1.05614090e+00 -7.16326892e-01 4.21061575e-01
-9.97258663e-01 -1.33928192e+00 -2.20060244e-01 -8.90292764e-01
7.36064672e-01 2.63658166e-01 1.45819092e+00 3.63210618e-01
-1.90657631e-01 7.28619158e-01 -2.01113611e-01 6.92804828e-02
-3.44768912e-01 2.70604700e-01 -3.35313767e-01 -1.30203292e-01
1.69699511e-03 -7.86164284e-01 -4.91675079e-01 2.72904634e-01
-8.89603853e-01 1.43061489e-01 4.37586933e-01 7.10825205e-01
3.36348772e-01 -9.85801369e-02 2.05553457e-01 -1.24931693e+00
7.59999394e-01 -5.30691028e-01 -7.14695156e-01 2.41334125e-01
-5.06252229e-01 4.80839282e-01 5.94749272e-01 -1.16577275e-01
-5.19403338e-01 -8.38323534e-02 -1.07474223e-01 -1.17409199e-01
4.08742070e-01 6.75448239e-01 -2.36002028e-01 -6.20219171e-01
5.98203361e-01 1.59293815e-01 -2.15608850e-01 4.04605754e-02
4.50178772e-01 2.02119201e-01 3.83669645e-01 -8.28210890e-01
8.38321328e-01 3.01933885e-01 6.47032559e-01 -8.01702619e-01
-4.76901457e-02 -2.99689740e-01 -3.21511805e-01 -1.60016388e-01
6.46015048e-01 -5.17608047e-01 -1.04611540e+00 4.01737928e-01
-1.16376436e+00 -4.54464763e-01 -7.06932619e-02 -1.11014120e-01
-3.78454566e-01 5.27367473e-01 -8.89591098e-01 -4.62186873e-01
-6.69127628e-02 -1.37530839e+00 1.25198603e+00 -1.95731297e-01
-3.67821939e-02 -1.28594851e+00 -1.53448507e-01 1.43738955e-01
4.99101818e-01 6.95976391e-02 1.63208079e+00 -4.22796607e-01
-1.30810475e+00 8.36291537e-02 -4.19608206e-01 -1.78210825e-01
-1.68377012e-01 -6.67959265e-03 -7.02466369e-01 -2.32319281e-01
-5.99641562e-01 -4.25229877e-01 1.02237356e+00 9.63691995e-02
1.54469562e+00 -1.41707972e-01 -6.68697298e-01 8.58780921e-01
1.28943253e+00 -8.50982368e-02 8.00200284e-01 -2.60116607e-01
9.93606389e-01 9.74725857e-02 -1.91813290e-01 4.22001988e-01
5.63932121e-01 1.93602234e-01 9.28901613e-01 2.86687970e-01
-3.37617517e-01 -7.08148420e-01 1.89887658e-01 1.23881555e+00
9.09192190e-02 -6.29535019e-01 -1.12862146e+00 5.57433069e-01
-1.56972921e+00 -7.86371827e-01 -7.41810203e-01 2.05692959e+00
5.24247169e-01 3.09216022e-01 -2.69252360e-01 3.66242409e-01
6.45010948e-01 4.00943458e-01 -7.41358578e-01 -1.98491409e-01
-2.24949524e-01 3.31441075e-01 8.44313741e-01 2.29958028e-01
-5.43897152e-01 6.83448255e-01 6.60946941e+00 7.88150430e-01
-9.69856501e-01 -6.58762082e-02 5.90242147e-01 1.67329624e-01
-1.29370391e+00 3.59729201e-01 -4.67024028e-01 2.36014158e-01
7.71356404e-01 -3.12563807e-01 7.80343235e-01 4.29933190e-01
-3.50603402e-01 -1.02748439e-01 -1.44755745e+00 1.06084192e+00
5.03105111e-02 -1.71174812e+00 6.68523088e-02 2.87918985e-01
8.96968007e-01 1.02656439e-01 1.49811760e-01 2.53040846e-02
2.71627516e-01 -9.20618117e-01 6.92020476e-01 1.83570027e-01
1.25709224e+00 -4.45573002e-01 2.67463893e-01 -1.56011268e-01
-1.59960461e+00 -1.88152324e-02 -2.41644513e-02 -3.78354602e-02
4.19969596e-02 8.88765037e-01 -6.01211965e-01 5.66080987e-01
3.71188641e-01 1.15570545e+00 -7.24434495e-01 9.79153574e-01
-3.28280151e-01 1.02443099e+00 -1.68531537e-01 -3.68192434e-01
1.00218333e-01 -9.84830111e-02 6.39931560e-01 1.07665539e+00
4.22608614e-01 -1.00328684e-01 -2.86513418e-01 7.80654073e-01
-8.39476943e-01 -2.09717572e-01 -1.07090902e+00 -6.11864805e-01
4.76102859e-01 1.07968533e+00 -1.10142100e+00 1.03470609e-01
-4.11265969e-01 1.02512085e+00 3.34298730e-01 5.11411965e-01
-9.14638758e-01 -6.91474676e-01 5.48793435e-01 3.28395277e-01
4.27060902e-01 -4.69788998e-01 -5.49920738e-01 -1.15576553e+00
3.62259001e-02 -8.68514180e-01 1.88683927e-01 -6.90715611e-01
-9.32617784e-01 1.22652464e-01 -3.31249386e-01 -9.14945602e-01
5.88143431e-02 -7.74046659e-01 -4.83531922e-01 3.87925059e-01
-1.35140598e+00 -6.22732997e-01 -3.42361718e-01 4.12616163e-01
3.26066688e-02 4.46468107e-02 4.48176861e-01 5.09530306e-01
-3.09699893e-01 5.59141815e-01 1.49669066e-01 1.27323568e-01
5.24514556e-01 -1.15746284e+00 7.46212542e-01 9.28390861e-01
1.81478992e-01 7.20341265e-01 4.78353381e-01 -7.16695011e-01
-2.78944492e+00 -9.84584272e-01 5.34577608e-01 -2.14542478e-01
1.13414240e+00 -7.72286892e-01 -6.87927783e-01 8.13551843e-01
1.89674526e-01 1.70870870e-01 3.71704012e-01 2.28894055e-01
-6.23985469e-01 -1.09700277e-01 -7.76456535e-01 6.53191149e-01
1.97767758e+00 -9.71955717e-01 3.27251256e-01 4.79857981e-01
9.44737315e-01 -6.83250666e-01 -7.29353309e-01 2.28018090e-01
4.26109314e-01 -1.15023232e+00 7.92629957e-01 -3.53170753e-01
7.60181367e-01 -2.29283854e-01 -2.73324311e-01 -1.32550085e+00
-3.95976603e-01 -1.01939154e+00 -3.03693771e-01 1.07044935e+00
6.77377582e-01 -5.05472124e-01 9.54026878e-01 4.75910492e-02
-3.85706574e-01 -8.75623524e-01 -7.72928059e-01 -5.62858343e-01
-2.33373508e-01 -8.72397065e-01 5.55474460e-01 9.51897264e-01
4.32745516e-01 7.32976019e-01 -1.87847130e-02 5.63408174e-02
5.64570844e-01 1.52732402e-01 5.32274246e-01 -1.17001200e+00
-4.02596384e-01 -7.76812375e-01 -7.10536182e-01 -1.66821122e+00
2.30322972e-01 -1.53751671e+00 -1.26020968e-01 -1.73444772e+00
3.85050654e-01 -4.36296493e-01 3.89853925e-01 3.06459367e-01
1.85757577e-01 4.84786481e-02 -4.61779088e-02 -3.30713153e-01
-8.13975990e-01 5.28953612e-01 1.06519639e+00 -5.85629702e-01
2.71135598e-01 -3.45479012e-01 -8.84694397e-01 3.12545270e-01
4.32805777e-01 -3.36921126e-01 -7.34883368e-01 -7.76351750e-01
1.18343079e+00 2.09274575e-01 1.31676912e-01 -7.52901852e-01
5.12847543e-01 1.65967017e-01 -2.08509013e-01 -4.37576473e-01
-2.77150720e-02 -7.76483953e-01 1.16152942e-01 5.74016690e-01
-3.42368513e-01 3.22131813e-01 3.05691957e-01 1.02851093e+00
-6.34033233e-02 8.24968070e-02 3.23137641e-01 1.36342093e-01
-7.89194286e-01 4.67114598e-01 -2.93075174e-01 5.53726256e-01
8.57495070e-01 -1.32240623e-01 -8.71564031e-01 -7.19178677e-01
-1.22514255e-01 2.67340958e-01 6.69010520e-01 2.50983834e-01
6.61084950e-01 -1.16179311e+00 -3.22752416e-01 3.12151372e-01
2.77994215e-01 -1.23546556e-01 9.22985151e-02 1.03307748e+00
-7.03221798e-01 4.81472135e-01 2.46406853e-01 -8.13064814e-01
-9.72278357e-01 4.78701979e-01 1.30709320e-01 -3.91348332e-01
-2.79807895e-01 8.99184167e-01 2.38311902e-01 -4.28543001e-01
1.62783310e-01 -6.53621912e-01 4.43778574e-01 -3.25396419e-01
-4.74484125e-03 4.22219187e-01 2.79146582e-01 -3.46301615e-01
-6.37628317e-01 6.61630929e-01 3.64891917e-01 2.37243250e-01
1.23050129e+00 1.08535970e-02 -6.54920340e-01 4.88201141e-01
1.34735143e+00 2.30365112e-01 -6.93075359e-01 -7.25951493e-02
1.27939850e-01 -3.37397158e-01 -6.42288327e-02 -4.63994533e-01
-1.22555828e+00 1.10601532e+00 -3.73167582e-02 4.71769959e-01
1.19840443e+00 1.07644305e-01 6.35117292e-01 4.20652330e-01
7.04662621e-01 -5.22878766e-01 2.68929213e-01 8.70873094e-01
3.82126898e-01 -6.31346226e-01 3.02153546e-03 -8.41907442e-01
-2.39716858e-01 1.14284015e+00 3.91211003e-01 4.44669068e-01
6.85994804e-01 6.99783564e-01 -8.23377132e-01 -3.88034016e-01
-9.59599018e-01 -7.10177124e-02 2.65341818e-01 5.90043366e-01
7.90366888e-01 -1.11379191e-01 -6.44069985e-02 1.68264061e-01
1.74780022e-02 -1.45769536e-01 7.37080753e-01 9.69643354e-01
-2.10159048e-01 -1.02156281e+00 6.04544058e-02 9.05767381e-01
-8.72279406e-02 -6.30832672e-01 -6.31099641e-01 2.83310235e-01
-4.64413583e-01 8.70260358e-01 3.02141756e-02 -6.88441098e-01
1.63088709e-01 -1.51436582e-01 6.73943400e-01 -5.83721220e-01
-3.51738125e-01 -4.77508038e-01 4.14595604e-01 -6.79754198e-01
-1.92470148e-01 -4.21264291e-01 -1.41279054e+00 -7.89332986e-01
-2.87740827e-01 -9.86873917e-03 6.39963925e-01 5.43471277e-01
7.49527395e-01 9.10267889e-01 2.71045774e-01 -5.73739827e-01
-1.55628398e-01 -4.54103440e-01 -6.81360304e-01 2.17621878e-01
2.85218447e-01 -3.58002216e-01 -2.08510756e-01 -1.68770686e-01] | [6.94898796081543, 6.211045265197754] |
2c2ccbe9-3d58-4bbf-a11e-ada4bebe83d4 | humans-can-decipher-adversarial-images | 1809.04120 | null | http://arxiv.org/abs/1809.04120v3 | http://arxiv.org/pdf/1809.04120v3.pdf | Humans can decipher adversarial images | How similar is the human mind to the sophisticated machine-learning systems
that mirror its performance? Models of object categorization based on
convolutional neural networks (CNNs) have achieved human-level benchmarks in
assigning known labels to novel images. These advances promise to support
transformative technologies such as autonomous vehicles and machine diagnosis;
beyond this, they also serve as candidate models for the visual system itself
-- not only in their output but perhaps even in their underlying mechanisms and
principles. However, unlike human vision, CNNs can be "fooled" by adversarial
examples -- carefully crafted images that appear as nonsense patterns to humans
but are recognized as familiar objects by machines, or that appear as one
object to humans and a different object to machines. This seemingly extreme
divergence between human and machine classification challenges the promise of
these new advances, both as applied image-recognition systems and also as
models of the human mind. Surprisingly, however, little work has empirically
investigated human classification of such adversarial stimuli: Does human and
machine performance fundamentally diverge? Or could humans decipher such images
and predict the machine's preferred labels? Here, we show that human and
machine classification of adversarial stimuli are robustly related: In eight
experiments on five prominent and diverse adversarial imagesets, human subjects
reliably identified the machine's chosen label over relevant foils. This
pattern persisted for images with strong antecedent identities, and even for
images described as "totally unrecognizable to human eyes". We suggest that
human intuition may be a more reliable guide to machine (mis)classification
than has typically been imagined, and we explore the consequences of this
result for minds and machines alike. | ['Chaz Firestone', 'Zhenglong Zhou'] | 2018-09-11 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [ 3.86548072e-01 2.85535157e-01 1.13069698e-01 -5.00536561e-01
-1.57101810e-01 -1.05453205e+00 8.52003634e-01 -7.09170103e-02
-7.89028645e-01 4.36755329e-01 -1.50284529e-01 -5.50418794e-01
2.81785488e-01 -6.11781418e-01 -8.24211001e-01 -6.01544559e-01
1.78178668e-01 4.65990186e-01 1.61093056e-01 -2.42767990e-01
2.16718063e-01 6.78710639e-01 -1.50594556e+00 4.46192741e-01
4.46852773e-01 9.59442794e-01 -1.67230293e-01 5.42042315e-01
1.99123904e-01 9.36673939e-01 -7.11775184e-01 -9.54633534e-01
6.00504279e-01 -4.08168912e-01 -9.25915480e-01 2.61353523e-01
8.43279302e-01 -1.09675921e-01 -4.15033400e-01 1.37258875e+00
9.62794349e-02 -1.19030483e-01 8.81323636e-01 -1.37454796e+00
-1.40309882e+00 3.14014167e-01 -4.42897618e-01 1.82935908e-01
2.30573714e-01 8.50817144e-01 6.59737766e-01 -6.57081008e-01
4.83425796e-01 1.57268941e+00 7.37780631e-01 9.04966474e-01
-1.58133197e+00 -9.01586652e-01 6.16134331e-02 4.18723911e-01
-1.23481512e+00 -5.47082365e-01 4.30513620e-01 -8.33427072e-01
6.63946807e-01 3.28165412e-01 5.83631277e-01 1.37668562e+00
3.48096728e-01 3.92067790e-01 1.42131996e+00 -3.47188026e-01
4.98264506e-02 7.06577063e-01 2.83258571e-03 5.89448810e-01
3.98327261e-01 5.88537395e-01 -7.69297257e-02 1.55368876e-02
4.00682420e-01 -1.62162349e-01 -3.35897297e-01 -1.50298044e-01
-1.24764323e+00 7.55685270e-01 8.83565366e-01 2.88176239e-01
-3.32794696e-01 1.92071870e-01 2.29201779e-01 5.49615979e-01
7.87053034e-02 1.11728871e+00 -2.75172174e-01 5.19342661e-01
-4.73811656e-01 1.79027393e-01 6.06406987e-01 6.33580923e-01
7.25940526e-01 3.50557178e-01 5.00952788e-02 4.71280247e-01
-8.59379992e-02 5.88333845e-01 7.33446062e-01 -8.63559842e-01
-1.06372595e-01 5.34458220e-01 1.77666247e-01 -1.40378308e+00
-3.31634611e-01 -4.25902426e-01 -8.54377508e-01 7.89697230e-01
6.92376852e-01 1.33053437e-01 -9.78129983e-01 1.92939329e+00
-3.25063318e-02 -3.37504178e-01 6.00647666e-02 1.25680459e+00
5.23974299e-01 2.75449365e-01 4.63791221e-01 1.45287588e-01
1.64038157e+00 -6.39843762e-01 -1.33909315e-01 -6.94578826e-01
3.19203168e-01 -6.89434588e-01 1.26802242e+00 3.03678632e-01
-1.01521862e+00 -7.50105441e-01 -1.08962810e+00 1.72851041e-01
-6.26114726e-01 -3.56517762e-01 5.81051171e-01 8.70675862e-01
-9.79648471e-01 6.70198679e-01 -1.87927544e-01 -4.90371376e-01
8.27557504e-01 2.80175418e-01 -5.82816422e-01 -7.94036314e-03
-1.11245668e+00 1.49096572e+00 4.12011236e-01 -4.58291359e-02
-1.05673730e+00 -3.78799289e-01 -6.53271914e-01 5.88730089e-02
7.58504495e-02 -7.53658533e-01 1.31888914e+00 -2.21359849e+00
-7.39032924e-01 1.74596190e+00 -5.89074939e-02 -7.48311460e-01
3.87879848e-01 2.96903044e-01 -7.95947075e-01 1.63953677e-01
9.42012519e-02 8.54321122e-01 1.15179765e+00 -1.70085752e+00
-4.56501454e-01 -4.94860053e-01 1.61787689e-01 -1.21598065e-01
-1.75319090e-02 1.64392158e-01 4.20670152e-01 -7.05412388e-01
4.38623279e-02 -1.07419908e+00 -1.08143382e-01 4.03587401e-01
-3.81441832e-01 -4.81861532e-02 5.70124149e-01 -2.71830231e-01
4.42210495e-01 -2.27564383e+00 -5.56284606e-01 1.52420208e-01
4.73573476e-01 6.84682906e-01 -2.78265297e-01 -1.66880917e-02
-4.52674091e-01 5.13479412e-01 2.38649826e-03 2.51831323e-01
2.21163169e-01 1.20309182e-01 -5.84474981e-01 6.77790999e-01
4.92058039e-01 1.24195814e+00 -1.03624606e+00 -2.20091328e-01
7.45853931e-02 6.79746792e-02 -5.47005832e-02 1.84139818e-01
1.73524797e-01 3.38378221e-01 -8.66585597e-03 4.46141601e-01
6.02907896e-01 -2.90667772e-01 -5.32843992e-02 -2.38495231e-01
3.63863319e-01 -1.90613449e-01 -5.20185530e-01 7.00892806e-01
-5.04658520e-02 9.97073829e-01 3.08038499e-02 -1.00667369e+00
7.25847006e-01 2.02903569e-01 -2.73726523e-01 -8.95101368e-01
3.19119155e-01 1.67120174e-01 7.10757077e-01 -5.86874425e-01
2.10492253e-01 -7.25594938e-01 8.10031295e-02 6.10172868e-01
-1.13299778e-02 -3.13976258e-01 -3.76080841e-01 2.00155884e-01
1.03512025e+00 -4.40981448e-01 2.59311229e-01 -2.32968464e-01
4.31765407e-01 3.76381934e-01 4.92311597e-01 1.14703834e+00
-5.60574412e-01 5.29174626e-01 2.44222507e-01 -8.64355743e-01
-1.13855112e+00 -1.40766799e+00 -2.00911298e-01 1.13799798e+00
2.91036397e-01 3.67066592e-01 -6.26028895e-01 -7.02482641e-01
2.76564777e-01 9.67398763e-01 -9.26641881e-01 -6.75037324e-01
-1.46153450e-01 -3.67228836e-01 8.32050681e-01 3.65529746e-01
2.73948371e-01 -1.43024182e+00 -6.82084680e-01 -1.53407216e-01
3.56037498e-01 -1.06663442e+00 -2.64557213e-01 2.13014498e-01
-2.27545187e-01 -1.32889879e+00 -4.93573695e-01 -8.53964448e-01
9.92119431e-01 3.12691540e-01 1.45482183e+00 4.03214842e-01
-4.10931975e-01 5.35815239e-01 -1.51238710e-01 -9.25062001e-01
-9.46077406e-01 -6.19912386e-01 5.38923562e-01 2.00534612e-01
6.14361763e-01 -4.48506415e-01 -6.94177210e-01 6.16738677e-01
-9.93193746e-01 -1.55916959e-01 6.96262479e-01 8.74053657e-01
-1.22084040e-02 -8.49475190e-02 7.16464579e-01 -1.04301155e+00
5.57198286e-01 -4.94248748e-01 -3.10225815e-01 3.39082509e-01
-6.88149929e-01 -2.30010673e-01 9.31861699e-01 -7.85802007e-01
-7.61541426e-01 -8.69358927e-02 2.25117415e-01 -6.23728514e-01
-5.57741344e-01 -7.70510435e-02 -5.00741825e-02 -4.03879225e-01
1.10256910e+00 1.39800757e-01 -3.01885400e-02 -5.32396697e-02
3.94935191e-01 5.85009396e-01 1.01011419e+00 -5.84494412e-01
1.26863205e+00 5.98473549e-01 -2.45994315e-01 -5.79089165e-01
-9.04648960e-01 9.61864293e-02 -5.49482524e-01 -2.18446046e-01
1.01561034e+00 -7.62109458e-01 -9.40388262e-01 5.32534778e-01
-1.29326415e+00 -1.77289471e-01 -4.82661247e-01 2.39049077e-01
-3.60866815e-01 2.44605932e-02 -3.56296986e-01 -7.26717710e-01
8.90359730e-02 -1.05455589e+00 4.31084543e-01 2.94851482e-01
-6.21264517e-01 -9.25925910e-01 -3.89700383e-01 3.68502587e-01
5.29047310e-01 1.62355036e-01 1.30747163e+00 -1.18586838e+00
-3.49534690e-01 -3.96940112e-01 -5.51667571e-01 7.45604515e-01
4.35784981e-02 -1.48823336e-01 -1.39208710e+00 -2.03873396e-01
2.30862334e-01 -8.04622889e-01 8.32024038e-01 -1.13096908e-01
1.00129223e+00 -6.70380414e-01 -2.83234894e-01 4.84963924e-01
1.22843504e+00 4.66822356e-01 6.00592673e-01 3.44407290e-01
5.42631507e-01 8.55812252e-01 6.68249354e-02 -2.02183068e-01
2.59508174e-02 4.50535327e-01 4.71277386e-01 -3.22879076e-01
-8.41416419e-02 -2.69493520e-01 2.10048079e-01 2.10029081e-01
1.77867547e-01 -3.29386979e-01 -1.07435393e+00 4.87907112e-01
-1.39721382e+00 -1.20881987e+00 -1.45983882e-02 2.19075227e+00
5.92067063e-01 3.97303164e-01 -1.39838383e-01 -4.58013974e-02
9.11382616e-01 -3.33451070e-02 -8.18085968e-01 -7.83225536e-01
-4.44512069e-01 2.05555111e-01 4.46890086e-01 -3.83778405e-03
-1.16621852e+00 9.42731738e-01 7.07253981e+00 6.53406799e-01
-1.27258241e+00 6.90238550e-02 9.75533426e-01 8.27207342e-02
-2.14000121e-01 -4.47240584e-02 -1.08985245e-01 3.28927457e-01
7.42748320e-01 -4.11374420e-01 4.83173013e-01 8.44706237e-01
-8.52410272e-02 1.60232093e-02 -1.51887286e+00 9.38799739e-01
6.44349083e-02 -1.13991117e+00 2.70876676e-01 -7.93140307e-02
5.75066030e-01 -1.51967406e-01 5.74458838e-01 3.14988077e-01
5.94947994e-01 -1.65960634e+00 1.11914062e+00 3.65075439e-01
6.75766110e-01 -3.60110670e-01 5.97066224e-01 3.53782088e-01
-3.93494278e-01 -2.19464555e-01 -4.84913796e-01 -3.66369158e-01
-3.46276075e-01 3.59122045e-02 -8.50906670e-01 -3.12853009e-01
5.36407590e-01 1.40608594e-01 -1.02816939e+00 7.44961262e-01
-1.51241601e-01 5.47607422e-01 1.47940665e-01 1.65623516e-01
3.86371434e-01 4.09452498e-01 5.84572375e-01 1.05747569e+00
-2.14871570e-01 2.49161392e-01 -5.53696789e-02 1.20402515e+00
-2.06805721e-01 -2.84511566e-01 -9.03887630e-01 -2.48614967e-01
4.63692099e-01 1.17972946e+00 -7.87737489e-01 -3.87537450e-01
-2.98426807e-01 9.75122154e-01 1.98966593e-01 5.25906563e-01
-6.45242989e-01 -2.78265953e-01 7.42322147e-01 1.83392197e-01
-1.55087620e-01 2.14350358e-01 -3.53711247e-01 -9.57729399e-01
-1.23424545e-01 -1.23825014e+00 8.47893581e-02 -1.16635942e+00
-1.87420201e+00 8.53155732e-01 -4.47595745e-01 -1.09463525e+00
-1.46341115e-01 -9.55819309e-01 -7.06389010e-01 9.25088704e-01
-9.47692513e-01 -1.10893714e+00 -2.39522606e-01 6.38868332e-01
1.97475359e-01 -4.06262368e-01 9.32521224e-01 -1.93677694e-01
-1.88648164e-01 7.76795685e-01 -6.03681542e-02 5.67414403e-01
8.22484851e-01 -1.04179740e+00 6.12175345e-01 7.20229149e-01
5.04459560e-01 5.55239975e-01 7.31114984e-01 -1.66690513e-01
-1.02553141e+00 -1.03328896e+00 4.21582729e-01 -8.57043982e-01
6.62344754e-01 -2.15366378e-01 -1.22020566e+00 6.65895879e-01
2.13163555e-01 7.65999779e-02 4.45643783e-01 -1.29726887e-01
-1.05915475e+00 1.17431171e-02 -1.41656184e+00 6.72747850e-01
8.16543639e-01 -8.06851447e-01 -8.30736756e-01 3.85819525e-01
5.68046391e-01 6.40518740e-02 -3.35590869e-01 1.88190624e-01
7.92138040e-01 -1.20506728e+00 1.08821034e+00 -1.33499825e+00
4.53915983e-01 -2.79472739e-01 -1.24428384e-01 -1.43050194e+00
-7.70167410e-01 -2.20810160e-01 5.10472596e-01 9.57837284e-01
4.14861351e-01 -8.31756473e-01 4.69933599e-01 1.02027929e+00
1.63134765e-02 -4.89784539e-01 -7.88508892e-01 -1.03983331e+00
3.17375094e-01 -4.19752151e-01 3.78958106e-01 1.31259096e+00
-1.46972388e-01 2.89584726e-01 -4.91361879e-02 7.92973861e-02
5.73212504e-01 -1.49282515e-01 6.36265099e-01 -1.08348823e+00
-2.36207962e-01 -8.32528353e-01 -1.01811993e+00 -2.85841495e-01
6.04645491e-01 -1.05269599e+00 1.33142591e-01 -8.31664860e-01
2.21522018e-01 -3.86208326e-01 -4.93555754e-01 5.43446302e-01
-1.75598815e-01 7.27005482e-01 5.65515637e-01 4.11065698e-01
-3.81752789e-01 -5.92164919e-02 1.16462970e+00 -4.96880323e-01
3.90355200e-01 -7.42164031e-02 -1.18683195e+00 1.10956967e+00
6.21804178e-01 -5.75687170e-01 -1.51001245e-01 -5.66978633e-01
1.52278930e-01 -3.14441264e-01 1.11384809e+00 -1.07884920e+00
1.45024166e-01 -3.14211190e-01 8.72271240e-01 3.20294410e-01
1.84974372e-01 -9.60430324e-01 1.82901382e-01 5.77077210e-01
-5.53137958e-01 1.07959352e-01 2.87548870e-01 4.67174351e-01
-2.01060213e-02 -3.48798066e-01 1.29909301e+00 -4.82240915e-01
-9.95977104e-01 3.99199091e-02 -6.74629271e-01 2.68574953e-01
1.21428323e+00 -3.93166631e-01 -7.36626923e-01 -5.37109494e-01
-6.85814798e-01 -1.05193250e-01 8.52596223e-01 4.73232538e-01
6.02537096e-01 -1.21265292e+00 -7.17182457e-01 1.82853520e-01
2.85874188e-01 -6.10837460e-01 2.79227309e-02 4.74035829e-01
-2.59688020e-01 2.95270503e-01 -5.17123461e-01 -3.40949953e-01
-8.79333854e-01 1.12306738e+00 5.69258511e-01 4.70251530e-01
-1.89994112e-01 1.05866051e+00 9.23517704e-01 -7.97746927e-02
-3.72905321e-02 1.63017988e-01 1.92914933e-01 -1.82457402e-01
6.36742949e-01 -1.32644385e-01 -3.27555597e-01 -1.06274176e+00
-5.43930888e-01 1.75908759e-01 -2.07094222e-01 2.00904146e-01
8.98545682e-01 6.34310469e-02 -4.19127084e-02 4.84901041e-01
1.04002273e+00 -2.27636740e-01 -8.02563310e-01 -2.24764094e-01
-5.08027785e-02 -5.76481521e-01 -4.24966514e-01 -1.15748823e+00
-9.47918713e-01 1.06476653e+00 7.91697979e-01 3.51079673e-01
9.70275223e-01 6.45806119e-02 4.19752181e-01 4.91682291e-01
4.09142107e-01 -6.02938771e-01 1.79501817e-01 1.70577839e-01
9.46228027e-01 -1.60448611e+00 -1.74820691e-01 7.54333064e-02
-8.71971130e-01 8.81536186e-01 8.72772753e-01 -2.84105778e-01
2.95357943e-01 -2.40718588e-01 4.22928363e-01 -1.38164088e-01
-7.75396228e-01 -2.27046102e-01 4.69826043e-01 1.05037248e+00
2.26163790e-01 8.79690796e-02 2.41187036e-01 7.00433850e-01
-3.29148471e-01 -1.81901738e-01 5.20072937e-01 4.13686126e-01
-4.05793577e-01 -6.42453194e-01 -6.70754492e-01 5.46211183e-01
-3.92079800e-01 -1.64588556e-01 -6.67094111e-01 8.95034671e-01
5.36460280e-01 9.38150585e-01 2.96275496e-01 -7.38568962e-01
2.54309416e-01 1.97112903e-01 5.18783092e-01 -6.22028887e-01
-8.79068375e-01 -7.32157767e-01 -1.81307748e-01 -3.56516510e-01
-1.55173600e-01 -3.65777254e-01 -8.20664108e-01 -4.85599041e-01
-9.66179743e-02 -2.09555798e-03 4.32412714e-01 9.38529849e-01
1.62096754e-01 5.60103431e-02 6.57534599e-01 -8.76038671e-01
-6.70200467e-01 -6.93299532e-01 -5.41822672e-01 1.05139804e+00
4.05250758e-01 -5.03327131e-01 -5.39759338e-01 2.88217902e-01] | [10.032305717468262, 2.3425228595733643] |
8356560a-b081-41a9-8709-69203ab81b7c | self-supervised-video-centralised-transformer | 2203.13166 | null | https://arxiv.org/abs/2203.13166v4 | https://arxiv.org/pdf/2203.13166v4.pdf | Self-supervised Video-centralised Transformer for Video Face Clustering | This paper presents a novel method for face clustering in videos using a video-centralised transformer. Previous works often employed contrastive learning to learn frame-level representation and used average pooling to aggregate the features along the temporal dimension. This approach may not fully capture the complicated video dynamics. In addition, despite the recent progress in video-based contrastive learning, few have attempted to learn a self-supervised clustering-friendly face representation that benefits the video face clustering task. To overcome these limitations, our method employs a transformer to directly learn video-level representations that can better reflect the temporally-varying property of faces in videos, while we also propose a video-centralised self-supervised framework to train the transformer model. We also investigate face clustering in egocentric videos, a fast-emerging field that has not been studied yet in works related to face clustering. To this end, we present and release the first large-scale egocentric video face clustering dataset named EasyCom-Clustering. We evaluate our proposed method on both the widely used Big Bang Theory (BBT) dataset and the new EasyCom-Clustering dataset. Results show the performance of our video-centralised transformer has surpassed all previous state-of-the-art methods on both benchmarks, exhibiting a self-attentive understanding of face videos. | ['Maja Pantic', 'Stavros Petridis', 'Pingchuan Ma', 'Yiming Lin', 'Yiming Luo', 'Jie Shen', 'Mingzhi Dong', 'Yujiang Wang'] | 2022-03-24 | null | null | null | null | ['face-clustering'] | ['computer-vision'] | [-2.01530173e-01 -3.86204273e-01 1.41275860e-02 -5.21090329e-01
-4.64226097e-01 -1.89493895e-01 6.69380844e-01 -6.07191443e-01
-1.17787592e-01 2.72145808e-01 2.63042122e-01 3.72447789e-01
-3.91343683e-01 -3.59417021e-01 -8.58674228e-01 -1.02214956e+00
-5.30430317e-01 3.67060304e-01 4.06145081e-02 -4.15542312e-02
3.23406048e-02 3.18370551e-01 -2.01662230e+00 6.39671445e-01
2.32187524e-01 1.05441236e+00 2.20647559e-01 6.49362028e-01
1.58804357e-01 1.17195129e+00 -1.42895058e-01 -3.22131813e-01
2.83512801e-01 -6.95759892e-01 -7.60730684e-01 5.01210332e-01
8.27064455e-01 -4.80584204e-02 -5.94578445e-01 7.20756650e-01
3.96177828e-01 1.80733934e-01 6.53723955e-01 -1.51083934e+00
-5.33714831e-01 4.86742079e-01 -6.03367865e-01 5.22478104e-01
3.19628835e-01 2.09085015e-03 7.58721590e-01 -9.57935631e-01
9.31836307e-01 1.57179534e+00 6.36368215e-01 7.71051288e-01
-1.03380430e+00 -6.68203294e-01 3.13393354e-01 8.53154242e-01
-1.47263598e+00 -7.92734742e-01 9.48159456e-01 -4.33020473e-01
9.03779209e-01 2.69970149e-02 7.47224629e-01 1.15292335e+00
1.23267574e-02 7.32967615e-01 1.11410284e+00 -1.71385452e-01
2.53056616e-01 -1.53212145e-01 -3.96252573e-01 8.99946511e-01
-3.36138397e-01 -7.88316354e-02 -7.00084746e-01 1.92493767e-01
5.26708543e-01 5.85248359e-02 -2.02044144e-01 -6.87204361e-01
-9.10515428e-01 7.34689295e-01 4.58510935e-01 5.96715748e-01
-2.13576555e-01 3.26988637e-01 5.26784956e-01 5.83277225e-01
8.48315179e-01 -1.46244615e-01 -3.66591722e-01 -2.19246730e-01
-1.30354083e+00 3.35416272e-02 6.85208082e-01 8.92769575e-01
8.71782899e-01 1.55569494e-01 -9.01238918e-02 6.42076850e-01
2.42574841e-01 2.06929937e-01 4.76042718e-01 -1.23958445e+00
3.59844901e-02 4.01414126e-01 -5.25770783e-01 -1.14559460e+00
-2.74747521e-01 -1.57758221e-01 -8.52332950e-01 -3.05496017e-03
4.12765145e-01 2.66751915e-01 -6.15980566e-01 1.88773859e+00
2.71919906e-01 1.02613056e+00 -7.67909512e-02 7.53381073e-01
7.35742152e-01 6.03303492e-01 -7.39436299e-02 -6.31256104e-01
1.27150416e+00 -1.07459772e+00 -7.82987058e-01 3.88030440e-01
4.74874526e-01 -5.60479701e-01 7.64226854e-01 3.78213674e-01
-1.14521110e+00 -6.83325052e-01 -7.00353384e-01 2.21787989e-01
-3.99992973e-01 -4.01732251e-02 6.51036441e-01 8.84035647e-01
-1.68129468e+00 7.17767477e-01 -9.02571559e-01 -6.85201049e-01
8.96656752e-01 5.00263631e-01 -7.91056156e-01 -3.49396437e-01
-7.65842617e-01 4.33274776e-01 6.98412955e-02 4.17371839e-02
-1.25593626e+00 -8.31023693e-01 -7.62533486e-01 1.56959042e-01
5.43196738e-01 -6.63975418e-01 9.14011002e-01 -1.44225264e+00
-1.64967012e+00 9.71514344e-01 -4.01140153e-01 -3.50811779e-01
3.51925254e-01 2.01679505e-02 -3.37451100e-01 6.38975024e-01
-1.44312158e-01 8.69305432e-01 1.48251390e+00 -1.19376266e+00
-3.13112259e-01 -5.64981580e-01 -4.28501368e-02 1.46905661e-01
-7.78256774e-01 1.41585588e-01 -7.42658496e-01 -6.04467690e-01
-3.08146447e-01 -8.88831735e-01 3.09017718e-01 -7.71893263e-02
2.21846744e-01 -4.85146701e-01 1.44349241e+00 -2.21808970e-01
1.07355976e+00 -2.23405147e+00 5.13418496e-01 -2.22662866e-01
2.78385520e-01 1.98646352e-01 -2.79306948e-01 3.98845524e-01
-3.19314927e-01 -2.45512314e-02 -6.71598092e-02 -7.77381837e-01
-1.88680053e-01 2.18634114e-01 -2.17787232e-02 8.07385623e-01
1.93180367e-01 8.64722192e-01 -9.61618662e-01 -6.98378086e-01
3.66235107e-01 8.95832837e-01 -9.71635103e-01 2.39510342e-01
-8.81172419e-02 3.55983466e-01 2.08054855e-02 6.61495090e-01
7.86403239e-01 -1.71560839e-01 2.26900697e-01 -3.27460855e-01
1.30203590e-01 -4.77352381e-01 -8.38958025e-01 1.73324394e+00
-3.56718779e-01 8.39554667e-01 1.89631671e-01 -1.41376281e+00
4.10850316e-01 4.66791481e-01 1.12128174e+00 -5.81858516e-01
1.40432596e-01 -9.59588662e-02 -2.93869942e-01 -6.84517682e-01
6.48926497e-02 -1.26793519e-01 4.75686073e-01 3.30182403e-01
7.45046079e-01 1.40637979e-01 2.05180183e-01 3.22049201e-01
1.04622662e+00 4.22327965e-01 -7.88994282e-02 -4.84990537e-01
8.01198721e-01 -4.11368161e-01 4.81931448e-01 3.10488433e-01
-4.07875061e-01 6.60501719e-01 5.18277407e-01 -4.31415856e-01
-7.57392585e-01 -8.16810489e-01 -5.80373853e-02 1.33171272e+00
-1.66718423e-01 -8.34742308e-01 -1.13779581e+00 -8.11656237e-01
-2.04173744e-01 -9.83346030e-02 -8.28209639e-01 -8.62188637e-02
-5.48299789e-01 -7.00861752e-01 2.87761420e-01 2.99259007e-01
5.44425011e-01 -1.05051756e+00 -2.63346612e-01 6.13512564e-03
-2.64700443e-01 -1.41022182e+00 -5.71237445e-01 -1.64903730e-01
-7.47943103e-01 -1.37228823e+00 -7.28366196e-01 -1.06558657e+00
5.51059902e-01 5.95663190e-01 1.18511415e+00 3.91498767e-02
-4.64339793e-01 1.13193083e+00 -4.08696145e-01 2.64140889e-02
1.33864075e-01 -1.87898859e-01 3.50676596e-01 5.80009818e-01
5.79317272e-01 -6.77804828e-01 -6.07612252e-01 4.76515859e-01
-8.43799293e-01 -3.16447794e-01 1.40478581e-01 7.81634033e-01
2.25614354e-01 3.46235067e-01 5.85013807e-01 -5.48818409e-01
1.86197668e-01 -6.88853145e-01 -3.06977302e-01 7.84097463e-02
-1.85465023e-01 -1.92619666e-01 6.89561784e-01 -4.42270815e-01
-1.05138910e+00 2.23920390e-01 9.82494205e-02 -1.17576325e+00
-7.45123113e-03 1.32007867e-01 -1.71945080e-01 -2.63674557e-01
2.13447481e-01 3.06881428e-01 2.44347572e-01 -1.60550877e-01
4.25522596e-01 4.37326223e-01 4.87699449e-01 -6.08077407e-01
8.37481856e-01 9.02406931e-01 7.93025270e-02 -1.17236769e+00
-4.92450356e-01 -6.60989106e-01 -9.04042482e-01 -6.61831498e-01
1.04874194e+00 -1.19515336e+00 -1.09209538e+00 6.12907112e-01
-7.97443926e-01 -3.66020828e-01 -1.36299908e-01 3.63414466e-01
-1.07190979e+00 6.23092830e-01 -5.57674944e-01 -7.29690433e-01
8.74800310e-02 -1.06518805e+00 1.18853343e+00 2.24102233e-02
3.36280555e-01 -1.13851166e+00 5.66177927e-02 4.20800537e-01
4.77449954e-01 2.48916805e-01 3.78118485e-01 -1.46840945e-01
-6.23238802e-01 2.52559781e-01 -6.81908354e-02 3.77675623e-01
1.16751485e-01 2.38460496e-01 -1.31956375e+00 -6.35694683e-01
2.90693462e-01 -4.95936841e-01 1.12657452e+00 5.35421431e-01
1.43931079e+00 -2.57554501e-01 -3.66032600e-01 8.28707159e-01
1.21760643e+00 -1.24492139e-01 7.51472652e-01 3.82438954e-03
7.35495389e-01 9.22716737e-01 3.48209530e-01 4.14158672e-01
5.78212082e-01 9.37245786e-01 4.42431986e-01 2.34301955e-01
-2.83937961e-01 -1.26492932e-01 7.60893822e-01 1.02220118e+00
-4.41758990e-01 -2.08098963e-01 -4.96494979e-01 5.59500813e-01
-2.00329304e+00 -1.55403841e+00 2.49810040e-01 1.73363388e+00
3.58622611e-01 -3.55385721e-01 4.29192781e-01 9.00943354e-02
6.67718768e-01 2.10611224e-01 -2.86986738e-01 -3.26671153e-02
-1.42930284e-01 1.75882310e-01 -3.17148827e-02 1.61127999e-01
-1.47856605e+00 1.23393881e+00 6.15853214e+00 9.20093238e-01
-1.16985595e+00 4.24281985e-01 5.93762696e-01 -3.64660531e-01
2.30756536e-01 -2.28285462e-01 -5.41363180e-01 6.03305876e-01
1.18474305e+00 -9.28464457e-02 6.20294929e-01 7.82251537e-01
2.03358546e-01 2.20005766e-01 -1.34258664e+00 1.51335561e+00
6.11093700e-01 -1.31271875e+00 8.09677411e-03 2.88058579e-01
8.77155900e-01 -7.39669576e-02 4.07682687e-01 3.89487267e-01
-1.58950210e-01 -1.12823212e+00 4.77087170e-01 5.41342497e-01
6.66408181e-01 -7.78720975e-01 5.38453043e-01 -1.33960217e-01
-1.68922162e+00 -4.56387341e-01 -5.60784340e-01 1.97744090e-02
-1.44245997e-01 3.12987477e-01 -3.61113518e-01 6.40789211e-01
1.20527136e+00 1.42027473e+00 -8.14587355e-01 7.32050657e-01
4.39061016e-01 6.39948070e-01 -2.31182665e-01 4.82044786e-01
4.01629657e-01 -2.22504809e-01 3.45289975e-01 1.15859127e+00
3.83750588e-01 5.85464984e-02 1.29214942e-01 4.38772053e-01
-2.57836848e-01 2.35677749e-01 -8.27610552e-01 -1.72557160e-01
5.46037816e-02 1.44104159e+00 -8.74442756e-01 -2.74723321e-01
-6.70983315e-01 1.01340473e+00 3.02362919e-01 2.29556531e-01
-7.79824734e-01 1.74795821e-01 8.63565445e-01 2.16711715e-01
8.06074440e-01 -1.62064120e-01 6.33101463e-01 -1.44720495e+00
-1.19515009e-01 -8.45407426e-01 5.38795114e-01 -6.22069478e-01
-1.37371123e+00 8.06633532e-01 2.25706309e-01 -1.19403374e+00
-4.70238179e-01 -6.88022137e-01 -6.96683943e-01 2.31135022e-02
-1.37393296e+00 -1.41882277e+00 -4.65070456e-01 1.32825768e+00
8.40546370e-01 -6.79039717e-01 5.81669211e-01 5.15748441e-01
-6.35105073e-01 5.94909728e-01 1.57411367e-01 -8.87515321e-02
9.26919103e-01 -1.09605217e+00 -1.12605155e-01 5.99089503e-01
2.90555865e-01 5.54939985e-01 3.70079607e-01 -2.84309983e-01
-1.77675748e+00 -1.11818850e+00 4.64852780e-01 -6.30208850e-01
5.90758801e-01 -6.46740675e-01 -7.40224838e-01 6.80259883e-01
3.95858705e-01 5.36418438e-01 6.92786932e-01 3.01884441e-03
-5.49873888e-01 -4.51878637e-01 -1.21813440e+00 4.24494773e-01
1.46587276e+00 -7.17428207e-01 -3.71558011e-01 4.44609672e-01
3.71251881e-01 3.15671772e-01 -8.52230370e-01 2.48151124e-01
5.11705399e-01 -1.23725665e+00 1.04776824e+00 -2.84738094e-01
3.93651366e-01 -1.79836079e-01 -2.62924373e-01 -1.26025188e+00
-4.08143431e-01 -7.91303158e-01 -4.18157130e-01 1.30600154e+00
-5.19795954e-01 -2.87358433e-01 9.34838653e-01 -1.37511402e-01
-5.05729467e-02 -6.83539331e-01 -1.19350696e+00 -9.13990498e-01
1.89704858e-02 -3.62198114e-01 3.84605080e-01 1.07074142e+00
6.26789927e-02 6.34478033e-02 -6.64026201e-01 -2.25033864e-01
9.55884695e-01 -2.53936529e-01 7.13918924e-01 -1.22649431e+00
-1.50226671e-02 -5.21105945e-01 -9.88925695e-01 -7.00866342e-01
6.52994633e-01 -1.01331973e+00 -2.90880144e-01 -9.61396396e-01
5.45420051e-01 1.01660110e-01 -3.15750003e-01 1.98901117e-01
1.50965855e-01 8.69436622e-01 3.14970881e-01 9.80990827e-02
-9.99903738e-01 7.53211796e-01 1.08382821e+00 -2.49495134e-01
8.31847265e-02 -4.55512047e-01 -5.20820498e-01 5.42239249e-01
5.86076498e-01 -3.29005629e-01 -7.17639029e-01 -1.24264039e-01
-1.78961858e-01 -2.48785347e-01 3.45310092e-01 -1.31228936e+00
4.46964681e-01 1.00227758e-01 4.94183004e-01 -4.89432961e-01
6.25793457e-01 -8.87751758e-01 5.16504161e-02 2.92287856e-01
5.47006801e-02 2.05505267e-02 3.15312445e-02 8.24778914e-01
-3.78965676e-01 2.84405023e-01 8.54922056e-01 -2.92284012e-01
-8.91871691e-01 7.19858348e-01 -4.87778127e-01 -4.34851609e-02
1.41753733e+00 -3.24552774e-01 -7.30289146e-02 -5.82086384e-01
-6.94244504e-01 2.28004739e-01 4.99545395e-01 7.55971193e-01
7.44295180e-01 -1.39107978e+00 -7.00238824e-01 3.19786340e-01
4.39042114e-02 -5.45516193e-01 6.37260079e-01 1.06585538e+00
-2.50689685e-01 4.68578190e-01 -5.84338605e-01 -1.03239727e+00
-1.49835849e+00 9.99443769e-01 3.66826802e-01 9.12939832e-02
-6.94197834e-01 7.74526596e-01 4.44666475e-01 -2.71957308e-01
2.84789503e-01 1.42938495e-01 -4.32353079e-01 4.24521595e-01
6.73508286e-01 3.24572355e-01 -3.59322093e-02 -1.09952641e+00
-5.68771780e-01 1.04716194e+00 -5.13388440e-02 2.15298101e-01
1.64071953e+00 -2.78490335e-01 -3.28629404e-01 4.07491475e-01
1.62920332e+00 -5.01154065e-01 -1.42103875e+00 -7.62718692e-02
-4.34590802e-02 -6.47220314e-01 1.13478579e-01 -1.64404437e-02
-1.58637297e+00 1.01902890e+00 8.19481134e-01 8.12346563e-02
1.30908823e+00 1.77155122e-01 4.36199963e-01 3.48384857e-01
4.13448066e-01 -1.12117970e+00 8.04666221e-01 3.98825645e-01
8.33463490e-01 -1.26473737e+00 -1.15784943e-01 -3.55176747e-01
-6.52922690e-01 1.11866546e+00 7.32426345e-01 -2.08159342e-01
1.08092690e+00 5.98127320e-02 -1.27818555e-01 -4.50256944e-01
-1.02352273e+00 -3.39132935e-01 1.31475002e-01 8.16423297e-01
3.73882085e-01 -3.64499539e-01 1.12057954e-01 1.52091518e-01
2.41899341e-02 1.08290255e-01 4.47006732e-01 6.49395287e-01
-1.25411838e-01 -9.87201750e-01 -2.85086572e-01 2.11552724e-01
-4.86911237e-01 2.59304106e-01 -2.78683573e-01 7.38425195e-01
1.53062880e-01 1.10059106e+00 3.22544754e-01 -5.00807405e-01
-1.26108885e-01 2.09374622e-01 8.65090847e-01 -4.06585276e-01
-3.89991105e-01 1.92569256e-01 -5.49351156e-01 -9.72350478e-01
-1.17641258e+00 -7.36459851e-01 -7.16765881e-01 -6.16123676e-01
1.37366459e-01 1.69075593e-01 4.28233415e-01 7.24684238e-01
3.93292934e-01 4.36491758e-01 9.49368536e-01 -1.55398059e+00
-5.06554544e-03 -8.56007278e-01 -6.23944581e-01 6.36541009e-01
3.06996137e-01 -1.03115880e+00 -4.88728821e-01 3.22902292e-01] | [13.338793754577637, 1.171366572380066] |
3eb05c1b-565f-41a7-906e-56291440ef8f | generalization-and-robustness-implications-in | 2107.00637 | null | https://arxiv.org/abs/2107.00637v3 | https://arxiv.org/pdf/2107.00637v3.pdf | Generalization and Robustness Implications in Object-Centric Learning | The idea behind object-centric representation learning is that natural scenes can better be modeled as compositions of objects and their relations as opposed to distributed representations. This inductive bias can be injected into neural networks to potentially improve systematic generalization and performance of downstream tasks in scenes with multiple objects. In this paper, we train state-of-the-art unsupervised models on five common multi-object datasets and evaluate segmentation metrics and downstream object property prediction. In addition, we study generalization and robustness by investigating the settings where either a single object is out of distribution -- e.g., having an unseen color, texture, or shape -- or global properties of the scene are altered -- e.g., by occlusions, cropping, or increasing the number of objects. From our experimental study, we find object-centric representations to be useful for downstream tasks and generally robust to most distribution shifts affecting objects. However, when the distribution shift affects the input in a less structured manner, robustness in terms of segmentation and downstream task performance may vary significantly across models and distribution shifts. | ['Francesco Locatello', 'Ole Winther', 'Bernhard Schölkopf', 'Michele De Vita', 'Samuele Papa', 'Andrea Dittadi'] | 2021-07-01 | null | null | null | null | ['systematic-generalization'] | ['reasoning'] | [ 6.01607442e-01 -1.91595733e-01 -1.05030172e-01 -5.45008779e-01
-3.53144407e-01 -8.88283908e-01 6.39266551e-01 3.30965638e-01
-3.82933348e-01 4.36476380e-01 2.15758666e-01 -1.05376847e-01
-2.82736659e-01 -7.21171141e-01 -1.12122035e+00 -9.06557322e-01
1.09684743e-01 5.35156727e-01 4.36557263e-01 -7.14028999e-02
4.67239648e-01 8.71665299e-01 -1.69068503e+00 3.75254393e-01
5.30020356e-01 7.93988645e-01 1.52115032e-01 5.90461552e-01
1.55441940e-01 5.30740976e-01 -7.73734808e-01 -3.29000317e-02
5.97355545e-01 -2.77514726e-01 -8.03083181e-01 3.41063768e-01
9.77939129e-01 -1.04891382e-01 -2.94547081e-01 1.04990983e+00
2.73335963e-01 5.01678050e-01 9.52196419e-01 -1.05089259e+00
-1.18058825e+00 4.87007678e-01 -6.72605872e-01 3.50042224e-01
-2.22516760e-01 5.40546238e-01 9.13711071e-01 -4.81313407e-01
5.61987698e-01 1.53403342e+00 4.57810700e-01 4.73754317e-01
-1.91371298e+00 -3.78533989e-01 6.51002944e-01 -6.16824627e-03
-1.09621143e+00 -3.66220981e-01 6.35031104e-01 -6.46580875e-01
7.21992254e-01 2.61259764e-01 1.57594651e-01 9.76566792e-01
2.10045233e-01 7.05785930e-01 9.69136119e-01 -1.83841735e-01
2.06535786e-01 -4.29598428e-02 1.77870527e-01 1.57211021e-01
6.08646750e-01 2.06078485e-01 -3.59768808e-01 8.89562294e-02
7.35912859e-01 -6.84007183e-02 -1.21001817e-01 -4.94450003e-01
-1.21390486e+00 5.56717396e-01 8.32173109e-01 1.35876909e-01
-1.34963065e-01 4.14517730e-01 3.41951549e-01 2.09758282e-01
5.48149347e-01 1.10211539e+00 -8.21701884e-01 1.43485263e-01
-8.24904740e-01 4.50342268e-01 4.54347700e-01 1.01008415e+00
9.36417997e-01 3.01414654e-02 -4.13475692e-01 8.29829574e-01
7.20758364e-02 2.41333842e-01 4.37084049e-01 -8.84075701e-01
3.29239339e-01 5.20459831e-01 6.74286410e-02 -9.52650309e-01
-4.67928886e-01 -4.79845583e-01 -5.25087237e-01 3.33096355e-01
7.11369395e-01 -7.81988539e-03 -1.44430923e+00 1.98843896e+00
1.57274514e-01 -5.72515987e-02 -1.29347220e-01 9.99678493e-01
7.10067868e-01 4.65753108e-01 4.09536839e-01 1.44621342e-01
1.06732833e+00 -8.33154738e-01 -2.65516520e-01 -5.97768724e-01
6.36251390e-01 -6.59931898e-01 1.07279408e+00 2.26462618e-01
-9.01935041e-01 -7.16440260e-01 -8.79581273e-01 -1.73499957e-01
-6.57111943e-01 -1.60542801e-01 3.44638020e-01 4.13784355e-01
-9.13153589e-01 7.23526657e-01 -6.98281884e-01 -4.07794058e-01
7.07098544e-01 4.01164711e-01 -2.45057195e-01 -2.54220754e-01
-7.08068967e-01 9.76920724e-01 5.29447079e-01 -1.83999240e-01
-1.04017484e+00 -1.05297995e+00 -9.12730455e-01 9.12911296e-02
2.70768523e-01 -7.13957548e-01 1.04271042e+00 -1.24987519e+00
-9.87286866e-01 7.27969766e-01 1.22926114e-02 -2.21934333e-01
3.25397849e-01 -2.39887074e-01 3.57116126e-02 -9.02186558e-02
2.12277800e-01 1.03549469e+00 1.06637836e+00 -1.67749453e+00
-4.20888156e-01 -5.97095788e-01 1.22642972e-01 2.99736828e-01
-1.35436580e-01 -1.74277768e-01 -2.15350389e-01 -8.82084787e-01
2.70616919e-01 -1.13474178e+00 -3.47793758e-01 1.93924755e-01
-5.30614316e-01 -2.39979289e-02 1.14270735e+00 -3.08531612e-01
6.04083240e-01 -2.52039623e+00 2.45225206e-01 1.27319291e-01
2.03808397e-02 3.83094512e-02 -5.00284016e-01 9.52790678e-02
-2.79993653e-01 5.48576117e-01 -2.41379485e-01 -2.10878387e-01
-1.36069044e-01 2.92088866e-01 -1.34562403e-01 6.40649498e-01
5.90891540e-01 8.98815691e-01 -8.17182958e-01 -1.57434091e-01
1.09740302e-01 2.41396517e-01 -5.60953438e-01 -5.70513643e-02
-5.54792523e-01 6.63161397e-01 -3.51100683e-01 5.49389243e-01
6.05202496e-01 -1.94521934e-01 -1.12946235e-01 -9.33752880e-02
1.96405143e-01 2.49754533e-01 -1.06749642e+00 1.46012664e+00
-3.70651245e-01 8.64669442e-01 -1.44824147e-01 -9.35605466e-01
8.08828533e-01 -2.11626187e-01 2.61068791e-01 -5.72244644e-01
1.87602136e-02 -1.33214772e-01 5.56356907e-01 -3.96114409e-01
5.79431415e-01 -1.06656522e-01 4.01987024e-02 5.29004872e-01
-5.33989780e-02 -3.91641110e-01 2.58631051e-01 3.85539792e-02
1.00308180e+00 2.16481864e-01 4.04356197e-02 -4.77460951e-01
-2.39575192e-01 7.24411383e-02 4.92700040e-01 1.07722068e+00
-8.66088197e-02 8.26130211e-01 5.35811007e-01 -1.62020087e-01
-1.09014201e+00 -1.16164935e+00 -4.54486459e-01 1.69476736e+00
3.82380456e-01 1.07082054e-01 -3.35251182e-01 -4.68454987e-01
4.62107122e-01 8.46464753e-01 -9.14745092e-01 -5.45148313e-01
-3.24365497e-01 -8.39205921e-01 4.17711169e-01 9.71828103e-01
3.04897487e-01 -1.21072221e+00 -5.74317038e-01 5.33896089e-02
2.04700604e-01 -8.93687904e-01 -2.97269583e-01 4.00819302e-01
-1.00133038e+00 -8.70062411e-01 -5.21051645e-01 -7.00180590e-01
8.53991389e-01 2.96654195e-01 1.15541911e+00 2.27929819e-02
-4.37183410e-01 3.51145208e-01 -3.96488070e-01 -5.02912104e-01
-2.12917864e-01 9.37417820e-02 -1.70207158e-01 1.51563182e-01
9.18292031e-02 -5.31945705e-01 -6.12787247e-01 3.86879355e-01
-1.31295061e+00 -2.00741082e-01 6.57293856e-01 5.11822760e-01
4.95940059e-01 9.95074585e-02 5.64200163e-01 -1.23560250e+00
4.41260219e-01 -6.13516629e-01 -2.85197645e-01 1.72430232e-01
-3.19167703e-01 2.25829273e-01 4.12621349e-01 -8.34829390e-01
-1.04467356e+00 4.70496602e-02 5.41024745e-01 -5.10440946e-01
-6.52530849e-01 3.37861389e-01 -3.57265770e-01 1.21871799e-01
1.03681469e+00 9.47724655e-03 -1.76441029e-01 -4.43050772e-01
5.69456995e-01 2.34488890e-01 3.32557797e-01 -9.29363310e-01
7.20556378e-01 4.86379385e-01 7.15498552e-02 -7.58662403e-01
-8.94038677e-01 -4.08045739e-01 -8.19805562e-01 4.73759361e-02
8.24311852e-01 -7.60832131e-01 -7.06055909e-02 6.95715249e-01
-9.16452527e-01 -8.77354741e-01 -3.58311504e-01 1.71012610e-01
-5.30371070e-01 -1.34652540e-01 -4.11196053e-01 -3.65158916e-01
4.53834891e-01 -1.27904630e+00 1.03248477e+00 2.67590046e-01
-3.09022516e-01 -9.97637331e-01 -4.05068398e-01 2.32115090e-01
3.94152850e-01 3.00732762e-01 1.31007671e+00 -8.61875474e-01
-6.23026252e-01 9.64178443e-02 -4.26163524e-01 3.61036271e-01
5.59726655e-01 2.23955721e-01 -1.05522311e+00 -2.92003810e-01
-2.91299760e-01 -3.88636082e-01 1.11038423e+00 5.79846680e-01
1.64504564e+00 -2.01074883e-01 -3.33527923e-01 5.66527009e-01
1.41273594e+00 1.01542532e-01 6.12053633e-01 3.33284914e-01
8.02421093e-01 9.02159333e-01 5.10412216e-01 2.05909505e-01
-6.26343265e-02 3.58770251e-01 4.05564725e-01 -4.59838025e-02
-3.47054303e-01 1.21466871e-02 -1.33121893e-01 -1.21319845e-01
1.92219362e-01 -5.31311750e-01 -9.75807428e-01 7.84502864e-01
-1.75617826e+00 -6.85240149e-01 -3.65688801e-02 1.96695006e+00
9.59457159e-01 2.57042468e-01 2.32102554e-02 -3.34187686e-01
8.29841733e-01 1.55626804e-01 -9.34485018e-01 -2.69046098e-01
-9.95309800e-02 -1.14664426e-02 5.76079845e-01 2.07447901e-01
-1.05075514e+00 1.04933023e+00 6.46986246e+00 4.64587927e-01
-1.22673225e+00 -2.15810835e-01 9.73305523e-01 -3.44072253e-01
-3.21992129e-01 -3.48741263e-02 -6.63442075e-01 3.98513734e-01
4.21967387e-01 2.18537033e-01 2.60853708e-01 7.33416855e-01
5.56110330e-02 -3.37757558e-01 -1.54187703e+00 5.23586929e-01
-3.85765880e-02 -1.15407097e+00 2.60410488e-01 1.27968416e-01
1.02888048e+00 -3.27615291e-02 4.07660872e-01 2.22111642e-01
5.04665434e-01 -1.16019070e+00 7.81621933e-01 4.91154492e-01
6.87533855e-01 -4.70431179e-01 4.82456088e-01 1.92248687e-01
-7.53096879e-01 -6.78857341e-02 -2.94289380e-01 -6.89919293e-02
-2.97609776e-01 3.62590075e-01 -4.82021511e-01 4.69008051e-02
8.54641020e-01 6.01846814e-01 -1.03025138e+00 1.02937078e+00
-1.34344488e-01 6.15291238e-01 -3.15616876e-01 3.76575924e-02
1.58324644e-01 4.62466329e-02 4.97169882e-01 1.24640453e+00
-2.65292794e-01 -1.11648716e-01 2.84133792e-01 1.08431149e+00
-2.00994164e-01 -2.88112313e-01 -7.45139360e-01 4.43392731e-02
5.41703880e-01 9.89697874e-01 -8.46556246e-01 -2.76898295e-01
-2.47028276e-01 6.00665629e-01 3.55975509e-01 9.01918411e-01
-5.81261456e-01 -2.13931516e-01 9.23660517e-01 1.90487519e-01
6.12798691e-01 -1.07595690e-01 -7.38075972e-01 -8.30397248e-01
-9.67164040e-02 -7.85113454e-01 1.56169310e-01 -8.90936732e-01
-1.75088584e+00 1.60954446e-01 4.04755414e-01 -7.91173220e-01
9.87367705e-02 -7.62701929e-01 -5.50446153e-01 7.19837606e-01
-1.21153641e+00 -1.17929745e+00 -2.03864738e-01 4.31719840e-01
6.25881135e-01 1.07110240e-01 3.85200262e-01 -1.69821233e-01
-4.16236848e-01 4.39459443e-01 8.89219940e-02 2.28417411e-01
7.70409644e-01 -1.24028873e+00 3.06770891e-01 8.67317796e-01
2.19737157e-01 7.65792787e-01 7.92798817e-01 -6.17349982e-01
-1.01558864e+00 -1.45297325e+00 1.40800685e-01 -5.27338684e-01
6.02063715e-01 -3.55545253e-01 -1.14699256e+00 8.90619159e-01
-8.26529786e-02 3.25924486e-01 6.29468262e-01 4.78619337e-01
-7.04572141e-01 -6.79322109e-02 -1.15227771e+00 7.59745121e-01
1.16276836e+00 -4.09069031e-01 -5.29053450e-01 3.59591782e-01
7.70963013e-01 -3.02919328e-01 -7.07569838e-01 5.90835631e-01
2.55622059e-01 -7.10112453e-01 8.73641193e-01 -1.09568560e+00
5.39982259e-01 -2.75421947e-01 -4.24690872e-01 -1.60999143e+00
-7.58769214e-01 -7.96988681e-02 1.94739744e-01 1.33716595e+00
6.10864162e-01 -5.83258092e-01 5.09089649e-01 8.04012775e-01
-2.88620591e-01 -5.99823833e-01 -5.86769342e-01 -8.70436013e-01
4.18765128e-01 -3.16070557e-01 5.17027497e-01 7.26539969e-01
-6.52220428e-01 3.14527899e-01 2.15398118e-01 3.31897885e-01
2.83896446e-01 2.01437607e-01 7.34113574e-01 -1.08919966e+00
-2.58431196e-01 -6.63840652e-01 -5.96542060e-01 -1.03247750e+00
1.94514275e-01 -6.78950191e-01 4.29095685e-01 -1.45873880e+00
2.12533861e-01 -7.12587535e-01 -3.47645462e-01 5.95600128e-01
-3.05547833e-01 1.43653467e-01 4.07455385e-01 3.18837076e-01
-4.52889979e-01 3.93817097e-01 1.45977473e+00 -2.78731525e-01
-2.33378962e-01 -2.95811109e-02 -9.51788783e-01 6.70469403e-01
9.28232670e-01 -5.43925047e-01 -5.80829322e-01 -8.22475076e-01
5.11075631e-02 -5.61449289e-01 6.65658176e-01 -8.66619766e-01
-8.49191472e-02 -4.41953212e-01 7.28010893e-01 -2.64239103e-01
3.43575060e-01 -7.45715976e-01 -3.17762047e-01 1.65600404e-01
-7.01069891e-01 -1.66558743e-01 6.04453325e-01 7.49646604e-01
1.64134920e-01 -3.03275406e-01 1.02985108e+00 -2.97653884e-01
-8.59437704e-01 3.03358316e-01 -3.63019258e-01 3.58639926e-01
9.94957209e-01 -4.21963692e-01 -5.86872816e-01 -7.13654160e-02
-6.98707819e-01 -4.97583523e-02 6.34031832e-01 6.33535028e-01
4.37642515e-01 -1.06416297e+00 -7.61415541e-01 1.82740688e-01
3.10430318e-01 4.18241680e-01 -2.41757203e-02 3.79726678e-01
-3.13271433e-01 -5.91485985e-02 -2.83071429e-01 -8.78340840e-01
-1.04124010e+00 4.93775308e-01 3.43946427e-01 2.68084019e-01
-3.20113957e-01 1.08141875e+00 7.53845334e-01 -4.64108199e-01
9.79388878e-02 -5.55154860e-01 1.51166484e-01 9.27859247e-02
1.25817984e-01 4.67985272e-02 4.28088494e-02 -4.16282326e-01
-2.91341782e-01 4.06415254e-01 -2.44070113e-01 1.98067240e-02
1.37261450e+00 1.95517913e-02 -3.95332389e-02 7.62507498e-01
1.08423030e+00 -3.88591617e-01 -1.62950683e+00 -2.15761513e-01
-1.27521068e-01 -6.84196591e-01 -6.53856853e-03 -8.40951502e-01
-1.09972298e+00 7.55799055e-01 4.81100619e-01 2.49445081e-01
9.52357173e-01 4.33748275e-01 2.09498063e-01 3.46294016e-01
1.33993372e-01 -9.98744309e-01 3.71258736e-01 4.09422934e-01
9.59357738e-01 -1.36821032e+00 1.53987661e-01 -2.42226899e-01
-5.92535079e-01 8.50581944e-01 1.01592696e+00 -1.77553609e-01
6.91134930e-01 1.48755088e-01 -8.45227838e-02 -2.02347174e-01
-6.91955447e-01 -2.00368300e-01 4.63045418e-01 7.61318266e-01
3.10374856e-01 1.10105887e-01 4.30520892e-01 1.27583118e-02
-1.00906096e-01 -5.02499461e-01 4.12624747e-01 1.00989747e+00
-3.76028180e-01 -6.22489393e-01 -4.22917694e-01 8.77084017e-01
-2.00286165e-01 -1.67212605e-01 -5.86907625e-01 9.67386484e-01
3.97836417e-01 9.47662234e-01 5.62833190e-01 -1.54079258e-01
9.48037431e-02 -5.62440567e-02 6.11144483e-01 -1.29139924e+00
-4.04775739e-01 -7.38256127e-02 -1.01848595e-01 -3.66917759e-01
-4.72112566e-01 -9.20922935e-01 -1.09935760e+00 -9.15076733e-02
-4.16061461e-01 -5.78707814e-01 5.64571023e-01 1.11741507e+00
4.24943656e-01 7.66192973e-01 2.71525890e-01 -9.22726750e-01
-4.61779594e-01 -1.16152561e+00 -6.67894363e-01 8.90682578e-01
4.67658997e-01 -7.74575651e-01 -4.55022037e-01 4.12353665e-01] | [9.625937461853027, 1.8888870477676392] |
3b5aeab2-cd78-49f7-953a-e60ad1d0730d | synthetic-pseudo-anomalies-for-unsupervised | 2303.05112 | null | https://arxiv.org/abs/2303.05112v1 | https://arxiv.org/pdf/2303.05112v1.pdf | Synthetic Pseudo Anomalies for Unsupervised Video Anomaly Detection: A Simple yet Efficient Framework based on Masked Autoencoder | Due to the limited availability of anomalous samples for training, video anomaly detection is commonly viewed as a one-class classification problem. Many prevalent methods investigate the reconstruction difference produced by AutoEncoders (AEs) under the assumption that the AEs would reconstruct the normal data well while reconstructing anomalies poorly. However, even with only normal data training, the AEs often reconstruct anomalies well, which depletes their anomaly detection performance. To alleviate this issue, we propose a simple yet efficient framework for video anomaly detection. The pseudo anomaly samples are introduced, which are synthesized from only normal data by embedding random mask tokens without extra data processing. We also propose a normalcy consistency training strategy that encourages the AEs to better learn the regular knowledge from normal and corresponding pseudo anomaly data. This way, the AEs learn more distinct reconstruction boundaries between normal and abnormal data, resulting in superior anomaly discrimination capability. Experimental results demonstrate the effectiveness of the proposed method. | ['Zhiqiang Wu', 'Lvdong Chen', 'Chenxing Gao', 'Caidan Zhao', 'Xiangyu Huang'] | 2023-03-09 | null | null | null | null | ['video-anomaly-detection', 'one-class-classification'] | ['computer-vision', 'miscellaneous'] | [ 3.97107005e-01 -2.52761722e-01 -9.20276567e-02 -8.65371674e-02
-2.22411558e-01 -3.16766441e-01 4.24397975e-01 -3.39974910e-02
-1.71699479e-01 3.87586474e-01 8.88418853e-02 -9.96108502e-02
2.43340731e-01 -7.27642059e-01 -8.47682238e-01 -9.48285878e-01
-1.81613445e-01 -2.41588160e-01 -1.56723894e-02 -2.25471389e-02
2.04363644e-01 3.91734689e-01 -1.83758557e+00 4.08908695e-01
1.16568029e+00 1.39769506e+00 -2.72716045e-01 5.16513586e-01
-2.89522231e-01 8.23251903e-01 -8.09376240e-01 -3.25177282e-01
5.82082033e-01 -5.50970435e-01 -1.46648183e-01 4.13749337e-01
5.28810024e-01 -7.73803115e-01 -9.03485954e-01 1.43413615e+00
1.73139825e-01 9.28217918e-02 4.65000212e-01 -1.60900486e+00
-5.69887340e-01 1.31334081e-01 -5.37820756e-01 7.54428208e-01
4.75094616e-01 3.22872251e-01 9.08293366e-01 -7.07693577e-01
1.84502870e-01 8.91244948e-01 5.36442220e-01 7.62079298e-01
-9.30428267e-01 -8.80676627e-01 3.44004333e-01 5.97164035e-01
-1.24046254e+00 -5.14581442e-01 1.05533218e+00 -1.09002046e-01
5.83578587e-01 4.97084588e-01 7.55481482e-01 1.52318704e+00
3.21109682e-01 1.01182055e+00 8.38070035e-01 -2.68700831e-02
2.22093344e-01 7.12851435e-02 -9.58440006e-02 6.25994027e-01
5.95457077e-01 1.48282558e-01 -4.67637599e-01 -4.35227513e-01
4.92287219e-01 6.37537360e-01 -7.52688050e-01 -2.12557524e-01
-8.81043613e-01 5.46650052e-01 2.91813370e-02 2.02292711e-01
-7.12902069e-01 -2.95311213e-01 6.81169569e-01 8.32066238e-01
1.81636170e-01 1.19231284e-01 -1.50857285e-01 -1.03891343e-01
-6.13153815e-01 -6.32805750e-02 4.89954680e-01 7.55140781e-01
5.16480088e-01 9.33619380e-01 2.39682328e-02 5.30031383e-01
2.94243485e-01 3.92756969e-01 9.74687159e-01 -5.07885933e-01
4.15307701e-01 7.19286799e-01 -2.99519688e-01 -1.28908587e+00
2.87138104e-01 -5.18680632e-01 -1.15170372e+00 2.43705362e-01
2.74213105e-01 5.38975485e-02 -8.72492433e-01 1.60753441e+00
1.30330637e-01 9.17537272e-01 5.37056088e-01 6.99755549e-01
3.80763233e-01 8.14150572e-01 -1.08705118e-01 -4.26671386e-01
1.02003086e+00 -5.09716809e-01 -1.01233888e+00 -1.36297047e-01
6.18495226e-01 -5.24009526e-01 9.95345652e-01 7.23456979e-01
-7.75853515e-01 -5.15355527e-01 -1.38314784e+00 5.82865715e-01
-1.66220084e-01 -1.75067678e-01 5.60917020e-01 7.27222323e-01
-6.92138016e-01 5.02210200e-01 -8.00629497e-01 -1.02710910e-01
5.45063078e-01 1.97585568e-01 -6.01396978e-01 -1.51414290e-01
-1.08701384e+00 2.73781240e-01 4.95126128e-01 1.21988460e-01
-8.80581737e-01 -5.01691937e-01 -9.54723954e-01 -1.41170556e-02
2.95493096e-01 -1.68276653e-02 8.59386623e-01 -1.61495924e+00
-1.10520518e+00 4.89051044e-01 -4.11655679e-02 -7.58451879e-01
4.31396782e-01 -1.69115081e-01 -1.23647332e+00 2.44928494e-01
-2.14185208e-01 4.30532210e-02 1.43098378e+00 -1.20645416e+00
-7.74098158e-01 -3.51216495e-01 -3.19007069e-01 7.11923316e-02
-7.92671263e-01 -3.22618037e-01 -2.20837727e-01 -1.04471862e+00
5.17867684e-01 -5.64692914e-01 -8.89457688e-02 2.35077478e-02
-4.11183029e-01 2.17120156e-01 1.37744796e+00 -8.62858295e-01
1.43282902e+00 -2.56477308e+00 -2.38646641e-01 7.78677225e-01
3.20465803e-01 2.64840901e-01 -1.76637188e-01 8.77301209e-03
-5.39145708e-01 -1.86582953e-01 -3.79173636e-01 1.17047839e-01
-4.22048032e-01 5.97746551e-01 -8.26032221e-01 5.82013786e-01
4.89724159e-01 4.00377393e-01 -6.96216166e-01 -1.63521215e-01
3.20772752e-02 2.31087551e-01 -5.44366181e-01 5.43312967e-01
1.41129866e-01 5.62165558e-01 -5.67467451e-01 9.01117086e-01
8.47611904e-01 1.49479117e-02 -2.40564030e-02 -9.67368856e-02
3.49781930e-01 -1.52727306e-01 -1.42562854e+00 1.23016095e+00
9.78151243e-03 4.95908409e-01 -2.20596850e-01 -1.28114235e+00
8.52875233e-01 5.32150805e-01 6.41225457e-01 -8.14370990e-01
7.94078968e-03 2.65197337e-01 2.89095521e-01 -5.99602342e-01
2.65755624e-01 2.27507800e-01 3.20542693e-01 4.77493465e-01
4.00028750e-03 6.06455982e-01 -2.07990184e-01 8.18358064e-02
1.30563986e+00 -2.21426532e-01 2.69769073e-01 2.89490670e-01
8.44094276e-01 -4.08410847e-01 1.04749918e+00 6.63680434e-01
-3.00030768e-01 4.04226184e-01 6.11713409e-01 -6.68145120e-01
-9.62126791e-01 -1.17090654e+00 -1.04094468e-01 5.68788111e-01
1.90329105e-01 -4.39852327e-01 -6.74112201e-01 -1.12946272e+00
-7.63633475e-02 5.85740626e-01 -3.83359224e-01 -7.58225143e-01
-5.57795525e-01 -8.56623948e-01 7.04929352e-01 4.24047887e-01
7.22299457e-01 -9.25954938e-01 -3.43796015e-01 -1.11138649e-01
-1.04768105e-01 -1.14139175e+00 -3.40752631e-01 -1.36715755e-01
-1.16375995e+00 -1.17436123e+00 -3.17541480e-01 -6.09778643e-01
1.08830035e+00 3.17306817e-01 7.11263955e-01 5.25831699e-01
-3.53074074e-02 4.55664635e-01 -4.73748237e-01 -1.94411322e-01
-7.25919664e-01 -6.97694659e-01 6.06475592e-01 6.62919760e-01
6.98160589e-01 -7.46118665e-01 -4.81271774e-01 1.66473299e-01
-1.38761842e+00 -4.25584823e-01 8.45559955e-01 9.67722297e-01
7.26679146e-01 5.15208781e-01 4.09629941e-01 -7.70276725e-01
4.52762306e-01 -7.58816302e-01 -3.66339773e-01 -3.47434841e-02
-6.85234010e-01 6.37854710e-02 1.11900628e+00 -7.20717490e-01
-9.23033416e-01 -1.84136391e-01 -3.24913770e-01 -8.51915240e-01
-4.45826262e-01 7.50188157e-02 -4.56349343e-01 -2.27118492e-01
4.77980733e-01 9.12289679e-01 2.54757941e-01 -1.90693170e-01
-3.54717940e-01 6.03235364e-01 7.02071667e-01 -4.36076194e-01
1.01451695e+00 4.92756516e-01 -1.95976079e-01 -9.59409654e-01
-3.45741361e-01 -2.58103520e-01 -7.67105892e-02 -1.91141382e-01
5.16113877e-01 -8.64930511e-01 -2.76513577e-01 6.24722719e-01
-7.81985581e-01 2.63183028e-01 -5.08658588e-01 5.44150412e-01
-3.08796197e-01 1.08039606e+00 -5.10524333e-01 -8.58376086e-01
-1.23138644e-01 -1.04840744e+00 4.27175462e-01 1.04982242e-01
6.89631552e-02 -6.78395629e-01 -2.49354422e-01 -1.50234848e-01
1.11295305e-01 3.32120717e-01 9.37606394e-01 -1.27106237e+00
-5.93997777e-01 -6.73205853e-01 6.30149916e-02 8.75500679e-01
3.99177253e-01 2.75271654e-04 -1.12152123e+00 -4.92241263e-01
5.27418613e-01 8.26634392e-02 7.47254074e-01 -8.64580050e-02
1.86328197e+00 -7.03880191e-01 -1.05974495e-01 9.71173167e-01
1.17128146e+00 4.57723618e-01 1.00441563e+00 3.64515245e-01
7.25473762e-01 2.90010512e-01 4.25148040e-01 6.47288620e-01
-6.53341040e-02 1.68930799e-01 7.58849621e-01 1.68283597e-01
3.12803745e-01 -2.39419550e-01 8.16719830e-01 1.00685275e+00
-6.57969192e-02 -3.40940773e-01 -5.37883103e-01 2.30517656e-01
-1.48673511e+00 -1.23007286e+00 1.24064438e-01 2.43012309e+00
5.72588384e-01 3.63971263e-01 -1.40445277e-01 8.50036263e-01
5.87674916e-01 4.16329473e-01 -7.47634292e-01 -2.21394181e-01
-4.66267645e-01 1.08557336e-01 1.96322069e-01 -1.84091061e-01
-1.14956617e+00 3.96165103e-01 5.92579460e+00 8.95722866e-01
-1.28266060e+00 -2.55507261e-01 6.69577539e-01 1.70390308e-01
-3.03332359e-01 -1.96717277e-01 -2.61670679e-01 9.73263800e-01
8.28184247e-01 1.07204817e-01 3.94383788e-01 8.10298920e-01
-6.70849234e-02 1.98330939e-01 -1.10757220e+00 1.15091932e+00
2.69182205e-01 -1.00438261e+00 6.77710235e-01 8.21637362e-02
5.63936532e-01 -5.05092382e-01 2.63621598e-01 3.13140512e-01
-4.40707386e-01 -7.47245431e-01 3.96631807e-01 6.19164646e-01
6.13828063e-01 -7.76556730e-01 8.61995578e-01 2.03153893e-01
-9.56450760e-01 -5.94064474e-01 -4.80647266e-01 1.00736693e-01
-3.38409781e-01 5.44302106e-01 -5.06964922e-01 5.79212070e-01
8.13158095e-01 9.50850189e-01 -5.75093091e-01 8.67477596e-01
9.04713720e-02 8.35702598e-01 -1.37091458e-01 4.65170115e-01
6.07350841e-02 -4.60430235e-01 1.01267648e+00 8.16965461e-01
6.51140511e-01 9.10902694e-02 1.47495255e-01 4.46970314e-01
-5.55913858e-02 6.88670576e-02 -9.58058119e-01 -1.76093400e-01
3.10765326e-01 8.59990358e-01 -3.85345876e-01 -2.38419905e-01
-7.46343255e-01 1.30822706e+00 -1.96582481e-01 5.23893893e-01
-7.80969918e-01 -1.90840319e-01 9.85185385e-01 -1.03592344e-01
1.03898436e-01 2.03041196e-01 -9.80900694e-03 -1.49972153e+00
4.44493055e-01 -1.45640719e+00 8.03812444e-01 -3.35621655e-01
-1.50691366e+00 6.70037091e-01 -4.13554102e-01 -1.98268664e+00
-2.95799494e-01 -6.31686807e-01 -9.36608911e-01 3.55143130e-01
-1.42171764e+00 -6.48416221e-01 -3.97340268e-01 9.38169241e-01
5.50563335e-01 -8.08224976e-01 7.71480501e-01 3.27104062e-01
-8.40358257e-01 1.05676222e+00 1.04565628e-01 3.48553926e-01
5.30787587e-01 -1.07154059e+00 1.48150951e-01 1.41094017e+00
2.27038279e-01 3.75189215e-01 6.97620034e-01 -7.20461190e-01
-1.37482786e+00 -1.10379362e+00 2.52904221e-02 3.43436114e-02
6.30113184e-01 -3.22894044e-02 -1.53720307e+00 6.96038663e-01
3.55980732e-02 3.79660040e-01 9.12068486e-01 -4.30150658e-01
-6.08145893e-01 -3.58120948e-01 -1.22664762e+00 7.55649865e-01
8.55105042e-01 -5.99654496e-01 -6.26014829e-01 -5.56646788e-04
4.46372151e-01 -3.41743350e-01 -5.22629738e-01 5.08879304e-01
2.85820872e-01 -1.10784686e+00 8.53163779e-01 -7.81584382e-01
3.20225745e-01 -4.33334440e-01 -4.12684262e-01 -1.18987226e+00
6.09940104e-02 -4.93457705e-01 -1.02487004e+00 1.02296138e+00
3.87591310e-02 -9.83612418e-01 7.54541099e-01 1.55554116e-01
-2.85239607e-01 -6.80517972e-01 -9.42654729e-01 -8.36525559e-01
-5.54677069e-01 -5.78817129e-01 7.24521101e-01 1.28573203e+00
-3.65097463e-01 -4.80020404e-01 -5.26408851e-01 6.44805431e-01
5.78836381e-01 -2.61329740e-01 5.47094226e-01 -1.20724928e+00
-2.59459704e-01 -2.74346709e-01 -1.02111268e+00 -8.30794394e-01
3.23421091e-01 -7.32939541e-01 -1.36302054e-01 -5.15922248e-01
-7.28551745e-02 -2.32136026e-01 -7.76718795e-01 1.85444996e-01
-3.17435116e-01 3.15781683e-01 -3.81339461e-01 2.27135986e-01
-3.57630432e-01 6.79038405e-01 8.12644839e-01 -9.75309610e-02
-4.22307104e-02 1.34146690e-01 -5.81740201e-01 8.40793371e-01
7.89479494e-01 -3.51394027e-01 -5.60964823e-01 -2.21794039e-01
-4.69963439e-02 -2.63386220e-01 2.94441938e-01 -1.45299220e+00
4.16703857e-02 -2.17950828e-02 8.52488160e-01 -3.97351533e-01
1.34159267e-01 -1.22406650e+00 -9.41993855e-03 5.04376888e-01
-1.07878342e-01 4.23536390e-01 2.15610713e-01 1.17661691e+00
-5.17178953e-01 -1.56116337e-01 5.17229974e-01 1.61799863e-01
-1.04614615e+00 7.22463667e-01 -4.38569695e-01 -5.93345650e-02
1.18785119e+00 -7.13665783e-01 -2.33291090e-02 -5.42630255e-01
-4.86507267e-01 5.21237962e-02 5.00721812e-01 4.48552310e-01
1.11955893e+00 -1.67015517e+00 -5.14519215e-01 1.05513465e+00
3.41028154e-01 -2.28257582e-01 6.38083220e-01 8.08585942e-01
-4.47988302e-01 -1.13621049e-01 -5.08083224e-01 -7.27733850e-01
-1.11721575e+00 7.58510530e-01 2.53320754e-01 6.75079525e-02
-9.29711580e-01 3.29148203e-01 1.70283973e-01 6.93238676e-02
3.04952383e-01 1.62143037e-02 -3.29097629e-01 -1.86709136e-01
8.59284461e-01 3.19917321e-01 -1.22633274e-03 -7.16817319e-01
-3.71956713e-02 2.20790207e-01 -2.65796721e-01 3.61888319e-01
9.78975058e-01 -6.72052130e-02 -1.57568622e-02 3.32647800e-01
1.02941644e+00 2.66039431e-01 -1.27380514e+00 -2.41874814e-01
5.41043468e-02 -9.04300928e-01 -3.93536389e-02 -5.71570285e-02
-1.36025524e+00 5.76310039e-01 9.15522456e-01 4.47376341e-01
1.65026426e+00 -5.56069732e-01 1.03293884e+00 4.47154224e-01
-1.57514244e-01 -9.46557701e-01 3.55668157e-01 1.98955446e-01
5.23151219e-01 -1.24689710e+00 -2.27550060e-01 -1.14673823e-01
-7.89587557e-01 1.18072665e+00 9.93732095e-01 -3.77267033e-01
4.16668534e-01 7.88331702e-02 1.17856404e-02 4.60097753e-02
-6.78928494e-01 1.44149616e-01 3.88486117e-01 7.63397813e-01
-8.65020007e-02 -3.78678769e-01 6.49525747e-02 7.00357616e-01
8.89706686e-02 -6.36433959e-01 7.46473491e-01 9.13746595e-01
-2.29290128e-01 -8.65148842e-01 -5.80013514e-01 9.05650854e-01
-7.15811074e-01 9.70377102e-02 -1.16237126e-01 6.00908697e-01
1.34466505e-02 6.88003719e-01 5.16300738e-01 -7.00684488e-01
2.39506289e-01 3.32406878e-01 -1.69195551e-02 -2.14283168e-01
-1.15923248e-01 -1.80624053e-01 -4.67847258e-01 -7.81516910e-01
-7.25073740e-02 -6.35717094e-01 -1.18977320e+00 -3.30746710e-01
-1.35558650e-01 2.27016464e-01 1.64634988e-01 9.10586119e-01
3.15588951e-01 6.09816790e-01 1.18358862e+00 -2.55323529e-01
-7.65263140e-01 -5.05025685e-01 -6.94444656e-01 7.19731748e-01
8.04131448e-01 -5.11562526e-01 -7.20567226e-01 -6.74010739e-02] | [7.666881561279297, 2.0637166500091553] |
f48711fd-c122-4a80-9a0a-f453be3616c7 | valhalla-visual-hallucination-for-machine | 2206.00100 | null | https://arxiv.org/abs/2206.00100v1 | https://arxiv.org/pdf/2206.00100v1.pdf | VALHALLA: Visual Hallucination for Machine Translation | Designing better machine translation systems by considering auxiliary inputs such as images has attracted much attention in recent years. While existing methods show promising performance over the conventional text-only translation systems, they typically require paired text and image as input during inference, which limits their applicability to real-world scenarios. In this paper, we introduce a visual hallucination framework, called VALHALLA, which requires only source sentences at inference time and instead uses hallucinated visual representations for multimodal machine translation. In particular, given a source sentence an autoregressive hallucination transformer is used to predict a discrete visual representation from the input text, and the combined text and hallucinated representations are utilized to obtain the target translation. We train the hallucination transformer jointly with the translation transformer using standard backpropagation with cross-entropy losses while being guided by an additional loss that encourages consistency between predictions using either ground-truth or hallucinated visual representations. Extensive experiments on three standard translation datasets with a diverse set of language pairs demonstrate the effectiveness of our approach over both text-only baselines and state-of-the-art methods. Project page: http://www.svcl.ucsd.edu/projects/valhalla. | ['Nuno Vasconcelos', 'David Cox', 'Rogerio Feris', 'Chen', 'Chun-Fu', 'Yoon Kim', 'Rameswar Panda', 'Yi Li'] | 2022-05-31 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Li_VALHALLA_Visual_Hallucination_for_Machine_Translation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Li_VALHALLA_Visual_Hallucination_for_Machine_Translation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 3.68616283e-01 1.41136616e-01 -2.64325857e-01 -3.47175837e-01
-1.07297242e+00 -3.84235620e-01 9.93811190e-01 -4.32535827e-01
-9.03225467e-02 7.33027160e-01 2.62909591e-01 -4.41102594e-01
7.70525455e-01 -4.96575892e-01 -1.03548717e+00 -5.67404926e-01
6.64487362e-01 5.92115700e-01 -3.34956408e-01 -9.93198380e-02
-2.13384703e-02 1.64202258e-01 -9.59382594e-01 7.02941716e-01
9.70825016e-01 9.43267643e-01 4.74741399e-01 4.57888484e-01
7.41806068e-03 7.69568741e-01 -4.86118525e-01 -7.49526680e-01
3.07594508e-01 -8.21560562e-01 -6.32657111e-01 4.60831404e-01
6.00223124e-01 -3.85871351e-01 -5.88331163e-01 1.07884669e+00
6.86611950e-01 1.68525130e-02 6.80010140e-01 -1.35861647e+00
-1.45768547e+00 3.36375058e-01 -7.33579457e-01 -1.44503966e-01
3.67192686e-01 4.98649359e-01 8.80869925e-01 -1.65111423e+00
8.61615658e-01 1.55600667e+00 2.51428634e-01 5.67604601e-01
-1.57980824e+00 -6.77476048e-01 1.24302574e-01 2.58318186e-01
-1.23534513e+00 -7.79673219e-01 7.28556037e-01 -3.40423197e-01
1.11176288e+00 3.99337560e-02 5.52023411e-01 1.57089293e+00
3.61130893e-01 1.15834093e+00 1.27827120e+00 -5.20258784e-01
-7.96999931e-02 6.32093549e-01 -4.92079616e-01 8.30731809e-01
-2.35845730e-01 -2.24732384e-02 -5.87007999e-01 5.97392358e-02
7.70824254e-01 1.15570217e-01 -5.16963899e-01 -6.51538670e-01
-1.59037364e+00 8.82184803e-01 6.27279699e-01 -2.98325289e-02
-4.74291265e-01 4.36284877e-02 1.48038328e-01 4.36315864e-01
6.92111075e-01 3.11619192e-01 1.56257048e-01 2.27151766e-01
-1.10270846e+00 4.16184068e-02 4.00066108e-01 1.12758958e+00
6.83501542e-01 1.93924025e-01 -4.25938815e-01 9.74999905e-01
3.76253694e-01 8.41344178e-01 4.73115027e-01 -7.89743662e-01
1.01001251e+00 4.33570445e-01 8.06907341e-02 -8.47310245e-01
1.84228078e-01 -4.83550996e-01 -9.19973135e-01 2.33571127e-01
1.50009632e-01 -1.03639722e-01 -1.17100370e+00 1.73574972e+00
2.71382509e-03 -1.11787848e-01 3.96434665e-01 1.22616792e+00
7.62144506e-01 1.07717955e+00 -2.67438859e-01 -1.96611315e-01
9.61775422e-01 -1.49323213e+00 -7.84418285e-01 -5.54249287e-01
1.79077178e-01 -1.04793000e+00 1.33815062e+00 6.69072270e-02
-1.48053539e+00 -5.30268610e-01 -1.00651073e+00 -5.38057089e-01
-2.25089401e-01 4.58401948e-01 1.29345238e-01 -1.06822504e-02
-1.19717586e+00 1.59960225e-01 -7.77380466e-01 -4.31675196e-01
3.56736332e-01 1.71735927e-01 -4.04625356e-01 -2.68951237e-01
-1.15226829e+00 1.30253768e+00 -6.68870732e-02 2.97181070e-01
-1.17086411e+00 -2.47845694e-01 -9.87753749e-01 5.42935133e-02
1.12409495e-01 -1.02338946e+00 1.34449279e+00 -1.42381525e+00
-1.54588175e+00 9.09774125e-01 -4.40992534e-01 -2.59759128e-01
8.72159064e-01 -9.53861102e-02 -2.37549096e-01 2.69831985e-01
1.79638281e-01 1.14176798e+00 1.14482617e+00 -1.52340889e+00
-9.82468054e-02 -2.91713923e-01 -2.26151142e-02 6.44096494e-01
-2.03722149e-01 -1.21339455e-01 -6.76057935e-01 -7.25816131e-01
-8.02958310e-02 -9.45629239e-01 4.94779237e-02 3.73824686e-01
-6.05206192e-01 1.55944303e-01 9.56662655e-01 -8.45991194e-01
6.41335845e-01 -2.02210665e+00 6.71722233e-01 -2.17019960e-01
1.49550691e-01 -6.38554469e-02 -4.95894223e-01 5.01376450e-01
-4.09444235e-02 -2.62756050e-01 -2.94942200e-01 -8.33087981e-01
1.24672323e-01 -5.69226802e-04 -7.52518356e-01 5.64004600e-01
3.19076031e-01 1.28001440e+00 -7.55286753e-01 -6.63443387e-01
3.88923317e-01 9.59119081e-01 -2.30427295e-01 4.46464628e-01
-3.16001743e-01 4.59597111e-01 -1.37333915e-01 7.37986565e-01
5.43962240e-01 -4.96893525e-01 1.67166479e-02 -2.12899342e-01
1.31734252e-01 1.67948738e-01 -3.36988181e-01 1.85006189e+00
-6.90720797e-01 1.03551066e+00 -1.65029056e-02 -8.88232589e-01
9.35444772e-01 3.80116910e-01 -6.37632906e-02 -9.43876505e-01
2.28030011e-02 1.83344170e-01 -3.31787050e-01 -3.30099493e-01
5.16408503e-01 -3.47140431e-01 1.64650410e-01 4.15031374e-01
1.82698265e-01 -2.37720415e-01 -2.09764928e-01 2.40361542e-01
4.20430481e-01 4.68311071e-01 9.81020927e-02 3.10964525e-01
2.06537232e-01 2.64752746e-01 2.55540669e-01 3.92182946e-01
-8.73122439e-02 9.87128794e-01 3.30771297e-01 -1.05593719e-01
-1.29305840e+00 -1.27438271e+00 1.38740107e-01 7.99000919e-01
3.04796249e-01 8.75180066e-02 -6.03250206e-01 -7.12891996e-01
-2.43498012e-01 9.46588218e-01 -6.05386078e-01 -2.12653071e-01
-3.96826178e-01 -4.13571566e-01 2.90982425e-01 5.19493937e-01
6.20079339e-01 -1.09359944e+00 -2.44653791e-01 -1.04072459e-01
-6.93331957e-01 -1.18693304e+00 -7.64627397e-01 -1.44369453e-01
-7.24224031e-01 -4.83417839e-01 -1.39852083e+00 -1.07067966e+00
9.40297067e-01 4.90334749e-01 1.14719725e+00 -3.01666468e-01
5.36506698e-02 1.94298089e-01 1.17428945e-02 -1.31752342e-01
-4.17891890e-01 -2.19094202e-01 -2.27851152e-01 1.56832054e-01
2.04353243e-01 -2.39070088e-01 -6.25031829e-01 2.21474141e-01
-6.39741480e-01 7.66352594e-01 9.09365714e-01 1.22778594e+00
5.87035894e-01 -8.03392708e-01 2.60499179e-01 -4.55799669e-01
7.23750651e-01 -4.63220775e-01 -4.50319260e-01 5.46106935e-01
-4.23800707e-01 -1.38576075e-01 7.46050179e-01 -6.08551621e-01
-1.08985615e+00 1.00729605e-02 3.36408317e-01 -1.07907152e+00
2.49024644e-01 4.00704920e-01 -1.75352454e-01 1.59735754e-01
4.75827992e-01 6.32174134e-01 1.58940107e-01 -1.91629142e-01
4.54111934e-01 8.47900510e-01 5.05900145e-01 -2.88902193e-01
7.22325802e-01 3.23767751e-01 -3.85076582e-01 -6.52210057e-01
-5.79087615e-01 -3.36291292e-03 -4.28055882e-01 -1.78435326e-01
9.30025101e-01 -1.15154243e+00 -3.11665475e-01 2.18892872e-01
-1.40651786e+00 -3.86390269e-01 2.94901244e-02 6.50577664e-01
-7.65377045e-01 2.84018368e-01 -5.29364288e-01 -6.08909845e-01
-4.18494701e-01 -1.54422593e+00 1.37562454e+00 -1.05187237e-01
-3.49567533e-02 -9.61472929e-01 -6.92138523e-02 6.66673481e-01
3.54129195e-01 1.53346464e-01 9.08297718e-01 -2.42590040e-01
-7.94327736e-01 1.06375441e-02 -5.04598022e-01 3.43126595e-01
-7.48499483e-02 -3.15879166e-01 -9.28701341e-01 -5.01228034e-01
-1.23910688e-01 -6.90543294e-01 9.70815122e-01 1.93334043e-01
8.41922939e-01 -4.41091597e-01 -2.52870232e-01 6.29132152e-01
1.27664316e+00 2.53279470e-02 5.92643142e-01 1.82318792e-01
7.50388503e-01 4.41047847e-01 4.97975677e-01 1.07279822e-01
5.30519545e-01 8.80663514e-01 4.09992844e-01 -6.15339041e-01
-4.24582958e-01 -5.86973071e-01 7.95260966e-01 8.23702157e-01
1.14132665e-01 -5.53059042e-01 -8.35542142e-01 6.55947626e-01
-1.97223842e+00 -8.48464370e-01 3.60153139e-01 2.09127378e+00
8.98564577e-01 -2.72413969e-01 -2.07712874e-01 -5.83346307e-01
8.31679404e-01 3.26761067e-01 -8.22944045e-01 -5.26005864e-01
-2.93742955e-01 -1.55967146e-01 1.10006005e-01 6.59722447e-01
-7.79280305e-01 1.12607658e+00 5.74133253e+00 4.96108055e-01
-1.41723847e+00 2.67968804e-01 7.32781947e-01 -2.42439255e-01
-4.34718609e-01 -3.25541422e-02 -3.33594859e-01 2.27468967e-01
6.39346540e-01 -2.44608134e-01 6.82807326e-01 5.87549508e-01
3.09019148e-01 2.55075425e-01 -1.17325795e+00 1.10907435e+00
5.08718550e-01 -1.32795012e+00 4.52033699e-01 3.61774191e-02
9.40072477e-01 9.25948024e-02 5.63637912e-01 3.65323395e-01
1.89616129e-01 -1.21961808e+00 8.71944308e-01 5.38838744e-01
1.12984002e+00 -4.42133635e-01 5.60320437e-01 2.39510447e-01
-7.97764361e-01 2.48225898e-01 -4.47133988e-01 1.25593156e-01
1.96327969e-01 1.60486713e-01 -9.80485976e-01 3.80685866e-01
2.89035976e-01 8.26740444e-01 -4.57742006e-01 5.87260485e-01
-3.46546650e-01 3.94085824e-01 1.23631269e-01 5.76982535e-02
4.41831589e-01 -1.73084185e-01 4.42976505e-01 1.16874146e+00
4.60984677e-01 -3.46122772e-01 2.82770634e-01 1.17833257e+00
-4.15026397e-01 1.79361969e-01 -7.82981575e-01 -2.50907928e-01
3.42889220e-01 9.45002913e-01 -2.57477790e-01 -5.30761361e-01
-7.27592230e-01 1.55353284e+00 5.16314626e-01 9.98589277e-01
-9.81826544e-01 -2.89376259e-01 4.82895017e-01 7.56253861e-03
2.28275537e-01 -1.25599056e-01 -3.39693308e-01 -1.64524162e+00
1.76572397e-01 -1.04602051e+00 -8.78571346e-02 -1.40186989e+00
-1.17405045e+00 8.11996877e-01 -2.02609509e-01 -1.55444813e+00
-3.81935090e-01 -5.11715829e-01 -5.40654361e-01 1.28354049e+00
-1.42940903e+00 -1.60471618e+00 -1.84949100e-01 7.16718197e-01
8.99507523e-01 -2.94987679e-01 8.16282868e-01 -4.17349562e-02
-4.08590257e-01 7.56195009e-01 8.40446874e-02 1.63868055e-01
1.03748763e+00 -1.07797194e+00 4.51607347e-01 8.09628546e-01
1.69976234e-01 4.31331664e-01 7.45408952e-01 -5.65351784e-01
-1.73775375e+00 -1.24948514e+00 1.09790599e+00 -2.46347263e-01
5.12971580e-01 -6.11151457e-01 -7.66500175e-01 9.89135504e-01
1.02417827e+00 3.57345911e-03 3.50203484e-01 -4.05269146e-01
-4.51259047e-01 9.95365083e-02 -9.93612051e-01 1.01842821e+00
6.98793352e-01 -8.24684858e-01 -6.43100977e-01 5.99913239e-01
6.80382609e-01 -4.78840679e-01 -5.38215280e-01 1.13219239e-01
4.92591292e-01 -6.68214023e-01 9.39795256e-01 -5.17191648e-01
1.11778164e+00 -3.62199694e-01 -2.76399106e-01 -1.58877480e+00
-2.09091440e-01 -5.01724958e-01 -1.01573408e-01 8.27662289e-01
8.70249331e-01 -5.32858372e-01 4.29197133e-01 3.02171707e-01
-8.31598639e-02 -8.12807679e-01 -7.30613351e-01 -6.31139576e-01
2.58880168e-01 -7.48709515e-02 2.55131692e-01 1.04714715e+00
2.00195864e-01 7.63438702e-01 -8.62577677e-01 6.44858330e-02
6.71323895e-01 4.90427196e-01 8.77970457e-01 -2.90030330e-01
-5.20489693e-01 -3.81093860e-01 -2.43848413e-01 -1.34822762e+00
3.40882480e-01 -1.14108241e+00 1.98673651e-01 -1.91421735e+00
5.83622098e-01 3.11692655e-01 -6.87237531e-02 6.92355454e-01
-9.63873044e-02 6.20515227e-01 4.11649525e-01 4.01297599e-01
-4.33702558e-01 1.02413332e+00 1.66109192e+00 -5.79300761e-01
7.81004354e-02 -2.49081522e-01 -5.94300210e-01 4.30396557e-01
7.73953259e-01 -6.06071837e-02 -5.49883187e-01 -1.01717949e+00
-4.89005931e-02 5.03873527e-01 5.97346663e-01 -5.05698383e-01
1.37810484e-01 -1.58385664e-01 8.09803367e-01 -5.54450393e-01
8.20658982e-01 -6.46976709e-01 -2.80732140e-02 2.62418240e-01
-6.32748067e-01 3.65410030e-01 5.83755672e-02 4.44498897e-01
-4.65673506e-01 2.43221804e-01 7.97653079e-01 -1.20838463e-01
-3.00133288e-01 2.44773969e-01 -1.76774621e-01 -1.05542503e-01
8.40078771e-01 -2.69287705e-01 -5.24573982e-01 -7.98179209e-01
-7.79306889e-01 2.81366765e-01 7.93511152e-01 7.06543207e-01
1.10878789e+00 -1.58839262e+00 -9.93297100e-01 1.20222099e-01
3.29657406e-01 -4.60746616e-01 1.17299622e-02 1.02956259e+00
-3.18174690e-01 3.92897666e-01 -2.46889859e-01 -6.93877399e-01
-1.29477012e+00 7.53582656e-01 3.12866658e-01 -5.38167823e-03
-7.06380367e-01 5.86540461e-01 6.68838382e-01 -4.40089494e-01
2.60619462e-01 -5.06418422e-02 2.15860710e-01 -3.64635199e-01
3.87318641e-01 -4.74099629e-02 -2.81156123e-01 -8.72154415e-01
-2.30623171e-01 4.17467266e-01 -1.84404150e-01 -6.47716522e-01
1.04736733e+00 -4.22605962e-01 8.89255777e-02 5.75078785e-01
1.37453496e+00 -3.08380753e-01 -1.36930919e+00 -4.91396695e-01
-5.47672391e-01 -6.59707248e-01 4.26131189e-02 -1.15459645e+00
-1.14767921e+00 1.48286676e+00 5.43142259e-01 -3.97955924e-01
1.21368146e+00 1.75948758e-02 7.37843752e-01 3.73069823e-01
1.25437155e-01 -7.46161759e-01 2.25699395e-01 3.76980573e-01
1.42883301e+00 -1.53034914e+00 -2.21305832e-01 -1.39173552e-01
-1.16931510e+00 8.92124593e-01 6.84635639e-01 2.05006618e-02
-4.19712700e-02 -5.88543415e-02 3.79866868e-01 1.89692736e-01
-1.07307005e+00 -1.39620434e-02 4.95924145e-01 4.72485989e-01
6.03524506e-01 1.34503722e-01 -2.87177507e-02 1.13636330e-01
-9.33205634e-02 -5.94197623e-02 2.68749774e-01 6.57335639e-01
-2.85272896e-01 -8.52012455e-01 -4.73305821e-01 -1.06920768e-02
-1.43545792e-01 -3.50053906e-01 -8.49260628e-01 5.88692904e-01
-1.91543996e-01 8.81299436e-01 -2.50850455e-03 -3.20886701e-01
1.51106417e-01 9.65623483e-02 6.25427485e-01 -5.10422826e-01
-3.62057656e-01 2.48221189e-01 -1.47333324e-01 -4.62988228e-01
-1.37447566e-01 -5.31312287e-01 -9.80141521e-01 -2.35436440e-01
-1.19168237e-01 1.73523501e-02 6.50453925e-01 7.00844526e-01
4.55839068e-01 2.11362824e-01 6.71148539e-01 -1.04772949e+00
-5.00716686e-01 -1.07589400e+00 2.23844964e-02 3.61566722e-01
5.39385438e-01 -4.75334704e-01 -1.82898238e-01 2.76671916e-01] | [11.43542194366455, 1.4557660818099976] |
1ba5d297-e4ea-4dda-bdfb-2a23dcde91e0 | improving-robustness-of-jet-tagging | 2203.13890 | null | https://arxiv.org/abs/2203.13890v2 | https://arxiv.org/pdf/2203.13890v2.pdf | Improving Robustness of Jet Tagging Algorithms with Adversarial Training | Deep learning is a standard tool in the field of high-energy physics, facilitating considerable sensitivity enhancements for numerous analysis strategies. In particular, in identification of physics objects, such as jet flavor tagging, complex neural network architectures play a major role. However, these methods are reliant on accurate simulations. Mismodeling can lead to non-negligible differences in performance in data that need to be measured and calibrated against. We investigate the classifier response to input data with injected mismodelings and probe the vulnerability of flavor tagging algorithms via application of adversarial attacks. Subsequently, we present an adversarial training strategy that mitigates the impact of such simulated attacks and improves the classifier robustness. We examine the relationship between performance and vulnerability and show that this method constitutes a promising approach to reduce the vulnerability to poor modeling. | ['Alexander Schmidt', 'Andrzej Novak', 'Spandan Mondal', 'Xavier Coubez', 'Annika Stein'] | 2022-03-25 | null | null | null | null | ['jet-tagging'] | ['graphs'] | [-5.42712733e-02 -3.59695405e-01 -7.52529502e-02 -4.40099478e-01
-8.28612328e-01 -1.10559726e+00 7.79384434e-01 4.54796076e-01
-3.52691978e-01 6.19627774e-01 -2.11307049e-01 -5.84936798e-01
-3.44342403e-02 -9.24417734e-01 -9.49804485e-01 -8.71225595e-01
1.61192231e-02 5.32206953e-01 3.74975294e-01 -2.53880441e-01
2.69234091e-01 8.62350583e-01 -8.30456853e-01 1.72296375e-01
4.11750942e-01 9.36179221e-01 -6.47501767e-01 6.17848456e-01
-4.10799906e-02 6.58574045e-01 -7.41581142e-01 -5.39943218e-01
4.67653692e-01 -3.56976718e-01 -5.36360681e-01 -8.24178159e-01
2.06849635e-01 6.73286393e-02 -3.77610505e-01 1.62073982e+00
5.35363317e-01 1.41163588e-01 5.91922283e-01 -1.28520274e+00
4.99235243e-02 8.64918470e-01 -2.60288864e-01 5.40239453e-01
-2.19550803e-01 3.93779635e-01 4.78393942e-01 -2.68964022e-01
4.62467790e-01 1.21231818e+00 6.70489967e-01 3.23620379e-01
-1.39234912e+00 -1.15598202e+00 -4.93487775e-01 -1.84045807e-01
-8.88418913e-01 -3.70384961e-01 8.73961568e-01 -5.30162513e-01
5.93814552e-01 1.55781686e-01 2.90252984e-01 1.35948265e+00
6.45968258e-01 -3.94218490e-02 1.22109115e+00 -6.15437806e-01
5.35769343e-01 5.36658406e-01 3.35066646e-01 3.92761052e-01
2.77564198e-01 9.34699953e-01 -3.59726280e-01 -5.64239502e-01
4.63292569e-01 -6.16115987e-01 2.62308627e-01 -5.08169711e-01
-8.00775170e-01 1.17058229e+00 5.45086205e-01 3.13678831e-01
4.94330898e-02 2.43330628e-01 7.97615647e-01 3.15551817e-01
4.64910328e-01 1.19060028e+00 -6.06524467e-01 8.78590494e-02
-7.57201016e-01 4.73382026e-01 5.65742552e-01 4.25747782e-01
7.82624632e-02 3.80739450e-01 -2.03097060e-01 3.94717872e-01
-2.38068588e-02 6.32591605e-01 6.43773228e-02 -7.22531319e-01
1.19051300e-01 1.25647694e-01 4.41441610e-02 -6.86351717e-01
-6.09189212e-01 -7.94770777e-01 -6.77700222e-01 6.84337497e-01
6.13117099e-01 -2.48221874e-01 -8.71312559e-01 1.72224832e+00
4.23964590e-01 -1.17237814e-01 8.90105888e-02 6.29096568e-01
5.69208860e-01 2.72420466e-01 4.95702714e-01 6.69672042e-02
1.16709018e+00 -5.23809373e-01 -2.19487965e-01 -6.87958002e-02
3.02563310e-01 -7.19141424e-01 5.14403284e-01 2.36606389e-01
-8.53048980e-01 -5.05262911e-01 -1.25307071e+00 3.79551381e-01
-5.23011684e-01 -5.55670142e-01 1.11910427e+00 1.19214129e+00
-4.37643021e-01 1.11953497e+00 -8.95975649e-01 -3.36552076e-02
-4.02311869e-02 6.21632576e-01 -2.81468064e-01 4.15397972e-01
-1.44261765e+00 7.97773957e-01 6.75948143e-01 -2.22923845e-01
-8.31323087e-01 -1.06272101e+00 -5.13666868e-01 7.13131577e-02
8.28439966e-02 -2.99018264e-01 1.48700202e+00 -6.16691947e-01
-1.23713911e+00 6.16193295e-01 3.87324452e-01 -7.61463225e-01
6.51913285e-01 1.15843661e-01 -8.35528851e-01 -1.70261413e-01
-1.76410273e-01 3.06353688e-01 8.74201417e-01 -1.02347934e+00
-7.74285719e-02 -7.41721969e-03 3.10563952e-01 -6.12406909e-01
-2.23468348e-01 6.92666650e-01 1.18039539e-02 -6.13830268e-01
-5.02765141e-02 -1.18238342e+00 -3.59667808e-01 -5.53000987e-01
-6.00359499e-01 2.82629222e-01 3.67544264e-01 -4.11290616e-01
4.82054681e-01 -2.17034864e+00 -2.50570238e-01 5.00559151e-01
6.76902533e-02 3.72987539e-01 1.85769275e-01 2.25532860e-01
-2.27448404e-01 3.49558115e-01 -1.11580662e-01 1.59062847e-01
7.69549981e-02 -2.84926027e-01 -4.02431130e-01 3.65519255e-01
1.22692950e-01 7.46535838e-01 -7.08780348e-01 7.27552548e-02
2.39293665e-01 2.44811296e-01 -7.10751832e-01 2.90535390e-01
-4.95488554e-01 1.04880595e+00 -3.94246280e-01 5.38803697e-01
7.59161294e-01 1.37875408e-01 1.06471293e-01 -3.71412814e-01
-2.84188129e-02 3.69390011e-01 -7.36858785e-01 1.03676379e+00
-3.11252952e-01 2.81571418e-01 2.01677948e-01 -7.98590541e-01
7.36219585e-01 -4.90630604e-02 3.20166677e-01 -8.07704926e-01
7.37492144e-01 2.15607926e-01 5.93103528e-01 7.96614960e-02
5.59605777e-01 -6.27235889e-01 -5.61257899e-01 4.08851862e-01
1.35811225e-01 -1.52050778e-01 -6.25781249e-03 2.13439822e-01
1.14042771e+00 -4.05222267e-01 6.51324168e-02 -5.57747245e-01
2.06429541e-01 8.08165446e-02 6.71514213e-01 1.12503946e+00
-2.90338695e-01 1.27087891e-01 6.12495661e-01 -4.41583008e-01
-1.24314737e+00 -9.76270974e-01 -4.82879639e-01 8.09559882e-01
-3.05126999e-02 -2.80556291e-01 -6.43529117e-01 -8.81834984e-01
2.86318600e-01 9.39330220e-01 -4.64231610e-01 -9.71501231e-01
-5.14923394e-01 -1.29082382e+00 8.25248182e-01 7.69327581e-01
2.93575466e-01 -1.02159846e+00 -1.03099607e-01 5.92243439e-03
4.88621742e-01 -1.19075441e+00 3.14157605e-02 7.56787896e-01
-6.62543237e-01 -1.01227105e+00 9.82914418e-02 -9.09550041e-02
3.12655240e-01 -3.60143006e-01 1.38633895e+00 1.57345146e-01
-1.80033311e-01 -8.78401548e-02 -2.69105047e-01 -5.95118642e-01
-1.32530463e+00 8.24311897e-02 5.37207186e-01 -4.29289073e-01
2.80888110e-01 -5.55105627e-01 -1.81709155e-01 8.34394544e-02
-7.34341800e-01 -4.45553988e-01 3.00858825e-01 7.73844182e-01
1.87076315e-01 4.53408390e-01 3.77226174e-01 -1.32261431e+00
4.45637107e-01 -5.74721098e-01 -1.28554368e+00 -8.61903802e-02
-5.84523857e-01 4.41091746e-01 9.13628817e-01 -4.51381505e-01
-9.04689431e-01 -2.92707793e-02 -4.74888682e-01 -5.60312450e-01
-1.90154314e-01 1.55518994e-01 -3.09475213e-01 -8.58078659e-01
8.21676791e-01 -5.44976830e-01 -4.47569013e-01 -6.31913006e-01
2.52604336e-01 1.72882691e-01 6.90206707e-01 -9.09174562e-01
1.30633974e+00 -3.67422588e-02 4.30168569e-01 -4.49898392e-01
-7.86629498e-01 2.96532419e-02 -5.02164543e-01 -5.32087013e-02
6.71707153e-01 -5.91460288e-01 -6.01990998e-01 4.76616621e-01
-8.70756567e-01 -9.36936364e-02 2.63674837e-02 5.51132977e-01
-1.04223147e-01 1.09067313e-01 -6.35000110e-01 -6.43799782e-01
-2.16447040e-01 -1.34312189e+00 4.72227097e-01 2.93350726e-01
-8.58857781e-02 -8.57516289e-01 1.56489789e-01 7.52198324e-02
5.20403743e-01 3.98395896e-01 1.51273537e+00 -1.12591851e+00
-4.83781010e-01 -3.97254288e-01 -7.14131817e-02 3.12696815e-01
-3.41170967e-01 1.62731558e-02 -1.28583324e+00 -5.05135953e-01
1.78609148e-01 -4.77522463e-01 7.53543437e-01 1.58482715e-01
1.17165744e+00 2.17925429e-01 -2.12845042e-01 7.78340101e-01
1.24447215e+00 3.37839305e-01 2.31591195e-01 3.02821130e-01
5.40183604e-01 2.78819293e-01 3.21744323e-01 -4.66832519e-02
-5.54907501e-01 8.13570678e-01 3.96039575e-01 -1.98572218e-01
9.98238474e-02 -2.50560075e-01 1.11309953e-01 4.77319151e-01
2.31380776e-01 -2.16973588e-01 -7.47381866e-01 -4.02175263e-02
-1.19581711e+00 -6.99459136e-01 -1.93717241e-01 2.24515843e+00
6.20096326e-01 8.97797465e-01 1.24042071e-01 1.47353873e-01
6.58106148e-01 -8.13398659e-02 -7.36561298e-01 -6.23117685e-01
1.15377769e-01 6.73651278e-01 8.85270774e-01 3.24532956e-01
-1.22868717e+00 9.19378638e-01 7.17449141e+00 6.24393582e-01
-1.20417941e+00 2.90830463e-01 4.80757564e-01 -1.51884794e-01
-1.87103853e-01 1.98854525e-02 -6.63774610e-01 6.16884291e-01
1.24077189e+00 -4.85883392e-02 4.34610128e-01 8.90592575e-01
-1.68502823e-01 7.72848576e-02 -1.06256878e+00 3.90928745e-01
-2.69414425e-01 -1.37483060e+00 -3.70640904e-01 -3.53008658e-02
6.32486939e-01 2.29130566e-01 1.51463842e-03 8.24653983e-01
6.44161642e-01 -8.35363626e-01 7.76474595e-01 1.25731647e-01
6.64176702e-01 -1.10678947e+00 8.50571692e-01 1.32977545e-01
-7.17699111e-01 1.03510723e-01 -4.64748174e-01 1.85367599e-01
9.00726095e-02 9.32288110e-01 -6.91901803e-01 4.71089274e-01
5.05469382e-01 -2.98815697e-01 -6.51461601e-01 9.92817283e-01
-9.17542651e-02 9.02651012e-01 -3.12233865e-01 1.37900501e-01
-8.40364322e-02 -6.93146046e-03 6.62832975e-01 1.07490325e+00
-4.37909766e-04 -1.98050290e-01 2.60556936e-01 9.50376689e-01
-3.81102771e-01 -2.72251874e-01 -6.45400822e-01 -4.81889904e-01
4.78661448e-01 1.22390115e+00 -1.21006489e+00 -8.97322819e-02
-1.14572823e-01 4.56016123e-01 2.25956649e-01 -6.21445198e-03
-9.43546951e-01 -1.23982012e-01 8.93643320e-01 -1.37160137e-01
2.83075788e-04 -2.23502442e-01 -2.78507888e-01 -9.28025305e-01
-3.16019058e-01 -1.04253578e+00 3.06372374e-01 -2.15297863e-01
-1.28626299e+00 5.33511162e-01 6.96745291e-02 -8.93476963e-01
-3.42724472e-01 -7.96906829e-01 -8.60222876e-01 7.72574127e-01
-1.13335824e+00 -8.41977715e-01 2.44210824e-01 1.12560458e-01
6.07306324e-03 -5.40543377e-01 8.69116366e-01 3.51586521e-01
-4.66571689e-01 8.59141111e-01 3.45348865e-01 1.21408701e-01
6.33019507e-01 -1.22739053e+00 9.08432841e-01 9.52075243e-01
2.33556807e-01 6.76305711e-01 1.20263052e+00 -8.51309001e-01
-1.31633854e+00 -1.16586542e+00 3.83168578e-01 -5.67650080e-01
8.70535731e-01 -9.09795642e-01 -9.05505002e-01 3.53425294e-01
-1.07733691e-02 2.52940953e-01 5.45618415e-01 3.49801570e-01
-6.05486929e-01 1.05403677e-01 -1.45949256e+00 2.19739377e-01
7.37344861e-01 -8.28857064e-01 -2.44186550e-01 4.98991579e-01
7.54780591e-01 -4.92526889e-01 -7.95256257e-01 7.69592106e-01
3.91832262e-01 -1.05468345e+00 1.05089271e+00 -1.12213063e+00
6.12030216e-02 -6.34931400e-03 6.01433218e-03 -1.43268526e+00
-4.65205789e-01 -4.15562600e-01 1.34116739e-01 1.39416862e+00
4.66212600e-01 -4.88191634e-01 8.32253993e-01 6.79970503e-01
2.24412948e-01 2.79602804e-03 -6.23121262e-01 -9.78668034e-01
6.75148070e-01 -5.69784403e-01 7.55618691e-01 9.60410178e-01
-4.04057771e-01 4.66017053e-02 -3.31269413e-01 6.50540233e-01
5.99529147e-01 3.53644826e-02 5.19818425e-01 -1.26794624e+00
-6.99515998e-01 -2.23709151e-01 -4.35249895e-01 1.71953410e-01
4.76276189e-01 -1.24460196e+00 1.37592172e-02 -3.17543566e-01
1.46010116e-01 -7.02809036e-01 -7.92118549e-01 1.55708373e-01
-1.67977177e-02 5.90036154e-01 3.30267906e-01 -2.67290622e-01
-2.63872236e-01 2.81804025e-01 4.36938614e-01 -2.87919372e-01
2.38597989e-01 1.55681014e-01 -4.50612843e-01 8.39690208e-01
9.43689942e-01 -1.07810712e+00 -8.82737190e-02 3.44400816e-02
4.39608455e-01 1.53531553e-02 4.99924421e-01 -1.29442823e+00
-8.15015435e-02 -7.87958428e-02 7.40064740e-01 -2.79617637e-01
2.20920160e-01 -5.79427123e-01 2.51107544e-01 4.90709871e-01
-4.26858932e-01 1.00348905e-01 6.38104081e-01 3.46856773e-01
9.39570516e-02 -6.20938301e-01 1.08593404e+00 -1.88980415e-01
-2.23725736e-01 1.42634004e-01 -3.22987229e-01 1.49509504e-01
7.32827485e-01 8.16992521e-01 -1.40920550e-01 3.54880169e-02
-5.36425173e-01 -9.86389518e-02 5.27059436e-01 4.20883566e-01
-2.97323614e-01 -1.28304148e+00 -2.74889022e-01 4.05219316e-01
-4.05620486e-02 -7.79645979e-01 4.11389470e-02 3.37040484e-01
-2.23655701e-01 4.75159019e-01 -2.83393562e-01 -2.67345965e-01
-1.20494473e+00 8.60277176e-01 5.06293178e-01 -2.96115816e-01
-3.57566059e-01 8.73433948e-01 6.24318756e-02 -6.54568195e-01
1.27229288e-01 -2.04092681e-01 5.97833320e-02 -3.79527301e-01
4.07921404e-01 1.00686029e-01 6.63659155e-01 -2.83899754e-01
-3.10856164e-01 6.72001839e-02 -2.89578795e-01 1.14835031e-01
9.15797710e-01 6.04510128e-01 3.03975120e-02 2.14880764e-01
9.78065610e-01 4.66222882e-01 -7.90420413e-01 1.21343732e-01
1.66384771e-01 -3.61296803e-01 3.81984204e-01 -1.13533020e+00
-1.14224017e+00 8.31836224e-01 8.74949574e-01 4.37015831e-01
6.33029044e-01 -1.73363499e-02 7.03200459e-01 3.58655453e-01
4.84651893e-01 -6.04537666e-01 -3.17088634e-01 4.86326665e-01
5.00686049e-01 -1.32920694e+00 -1.46925449e-01 -3.35274726e-01
-1.60162225e-01 9.08059657e-01 5.66162467e-01 -6.39751088e-03
5.72615445e-01 7.58904517e-01 1.93687469e-01 -2.70008534e-01
-4.56051379e-01 3.39830339e-01 2.00512782e-01 3.10408324e-01
2.38618508e-01 2.55696923e-01 -8.28744620e-02 8.24853361e-01
-5.44628382e-01 -3.56049091e-01 3.83909106e-01 7.33746767e-01
-6.64034858e-02 -1.52224529e+00 -6.22491181e-01 5.00175834e-01
-6.63131595e-01 -1.16914645e-01 -4.28604126e-01 7.00387299e-01
1.82924509e-01 8.12214553e-01 -2.62668818e-01 -5.70971966e-01
1.24026515e-01 4.68662053e-01 5.32223284e-01 -4.77416933e-01
-1.07160926e+00 -2.71356404e-01 2.31955767e-01 -3.93121690e-01
5.19419946e-02 -5.67859232e-01 -1.06020319e+00 -4.56768781e-01
-4.10835594e-01 4.51313704e-01 8.92272353e-01 1.04083943e+00
2.61634868e-02 8.56099486e-01 6.22988820e-01 -7.71719754e-01
-1.01354885e+00 -7.83648551e-01 -6.57190144e-01 3.70675027e-01
2.51963258e-01 -1.05846882e+00 -5.38093925e-01 -6.79416418e-01] | [15.684991836547852, 2.925086498260498] |
d1795b0d-70b8-4417-9c97-56cf68eea750 | soccer-line-mark-segmentation-with-stochastic | 2108.06432 | null | https://arxiv.org/abs/2108.06432v2 | https://arxiv.org/pdf/2108.06432v2.pdf | Soccer line mark segmentation and classification with stochastic watershed transform | Augmented reality applications are beginning to change the way sports are broadcast, providing richer experiences and valuable insights to fans. The first step of augmented reality systems is camera calibration, possibly based on detecting the line markings of the playing field. Most existing proposals for line detection rely on edge detection and Hough transform, but radial distortion and extraneous edges cause inaccurate or spurious detections of line markings. We propose a novel strategy to automatically and accurately segment and classify line markings. First, line points are segmented thanks to a stochastic watershed transform that is robust to radial distortions, since it makes no assumptions about line straightness, and is unaffected by the presence of players or the ball. The line points are then linked to primitive structures (straight lines and ellipses) thanks to a very efficient procedure that makes no assumptions about the number of primitives that appear in each image. The strategy has been tested on a new and public database composed by 60 annotated images from matches in five stadiums. The results obtained have proven that the proposed strategy is more robust and accurate than existing approaches, achieving successful line mark detection even in challenging conditions. | ['Narciso García', 'Carlos Cuevas', 'Daniel Berjón'] | 2021-08-14 | null | null | null | null | ['line-detection'] | ['computer-vision'] | [ 1.55571178e-01 -1.72514662e-01 2.11760670e-01 4.30284888e-02
-3.92305940e-01 -7.23577261e-01 5.85507691e-01 4.90483314e-01
-5.60882568e-01 5.62902153e-01 -2.83663094e-01 -9.48933661e-02
2.37955227e-02 -8.16259563e-01 -7.25535691e-01 -2.74401098e-01
-6.95132688e-02 6.33029699e-01 9.28944767e-01 -5.62927783e-01
5.23370385e-01 8.73403072e-01 -1.49880314e+00 -7.40332380e-02
5.94081819e-01 5.43914318e-01 -2.22627997e-01 9.04894233e-01
-7.23002106e-02 4.85696673e-01 -7.37016499e-01 -7.01344311e-01
3.39853108e-01 -3.89302671e-01 -4.67748344e-01 4.38382745e-01
4.15097624e-01 -2.29144827e-01 -3.30673814e-01 9.91332293e-01
2.88784027e-01 8.59216750e-02 5.30933559e-01 -1.08073735e+00
4.93529402e-02 2.64870852e-01 -9.23010707e-01 3.39662552e-01
8.42458069e-01 -7.67007396e-02 5.89934826e-01 -7.54988432e-01
8.30706477e-01 7.75004029e-01 1.15685320e+00 -4.64353301e-02
-9.98314142e-01 -2.69634634e-01 -2.34787375e-01 6.44865856e-02
-1.62724757e+00 -2.37703860e-01 8.16699088e-01 -4.06099945e-01
3.46695423e-01 5.74765682e-01 9.57047582e-01 3.64370078e-01
2.08462346e-02 7.08917975e-01 9.44774806e-01 -9.27637339e-01
1.12455428e-01 3.02990168e-01 1.93229154e-01 7.31586039e-01
3.94903213e-01 -3.67437303e-02 -3.24668884e-01 -5.37066385e-02
1.14456439e+00 -1.33981466e-01 -4.45122272e-01 -7.22493231e-01
-1.16248858e+00 5.37674904e-01 1.81948304e-01 4.80763495e-01
-4.43316072e-01 -1.18708469e-01 3.48678589e-01 -1.39390796e-01
2.09437892e-01 1.98194474e-01 1.44119635e-01 -3.63839865e-01
-1.30546832e+00 2.83380300e-01 7.69749403e-01 8.70146573e-01
5.64672947e-01 -1.58981428e-01 2.41734415e-01 7.14510798e-01
7.69886014e-04 1.43385246e-01 3.79649520e-01 -6.21782601e-01
4.11762029e-01 5.83753228e-01 2.87037760e-01 -1.50007582e+00
-6.54881239e-01 -4.11933780e-01 -2.73221761e-01 4.88639086e-01
9.41409707e-01 -1.44981891e-01 -5.60235083e-01 8.04373741e-01
5.56033194e-01 2.28476658e-01 -2.99962163e-01 8.99171650e-01
6.45441890e-01 4.27084059e-01 -4.54377949e-01 7.55936950e-02
1.48520219e+00 -5.13552189e-01 -6.61914110e-01 -2.74347961e-02
5.23045540e-01 -1.11669791e+00 8.81472588e-01 6.34980977e-01
-1.15647805e+00 -4.88565981e-01 -1.20881569e+00 8.01518783e-02
-2.07945570e-01 3.87589395e-01 4.28045094e-01 9.70611155e-01
-5.73485434e-01 5.18672228e-01 -6.74676061e-01 -3.72798175e-01
-2.54595075e-02 3.08069795e-01 -4.96315509e-01 1.49880961e-01
-6.99336410e-01 6.94865763e-01 4.45516199e-01 5.36085963e-01
-9.69270617e-03 -3.49488497e-01 -8.23198795e-01 4.76046838e-03
5.57451189e-01 -3.89371574e-01 1.07797992e+00 -9.60262716e-01
-1.66678500e+00 1.01742053e+00 1.80903181e-01 -5.33216417e-01
1.23224032e+00 -4.36575413e-01 -3.64602745e-01 3.53414178e-01
-1.06292218e-01 1.81362703e-01 4.63098615e-01 -1.37406874e+00
-8.76479566e-01 -3.55759174e-01 8.71766731e-02 3.81843209e-01
1.14537254e-01 2.28699297e-01 -8.05750549e-01 -5.58679819e-01
7.53674030e-01 -9.01108027e-01 -9.90880132e-02 1.27459228e-01
-6.18801177e-01 1.99976653e-01 7.00289130e-01 -9.01501417e-01
1.06486154e+00 -1.95928216e+00 -4.09330636e-01 6.47872269e-01
-1.13460300e-02 2.85042137e-01 5.21138787e-01 4.18194592e-01
2.04558671e-02 -3.41197848e-01 -2.62985557e-01 -8.32875222e-02
-1.94202796e-01 -7.74321705e-02 -3.90344039e-02 8.91380847e-01
-3.33334327e-01 3.32538128e-01 -6.92827463e-01 -6.81672037e-01
5.93433559e-01 5.26696682e-01 -1.13235109e-01 -2.09510610e-01
2.70719707e-01 1.89641416e-01 -1.24975502e-01 4.72813815e-01
7.13401675e-01 5.03916502e-01 -6.14388958e-02 -5.16107976e-02
-4.46489125e-01 -1.77019715e-01 -1.99009395e+00 1.52762377e+00
-1.70991063e-01 8.57193172e-01 -2.24576309e-01 -6.69677496e-01
1.33702266e+00 2.98050404e-01 6.37533367e-01 -4.77373928e-01
1.89465746e-01 2.86609948e-01 -2.41703168e-01 -2.77813882e-01
9.62706387e-01 1.59705788e-01 4.74569760e-02 1.72438309e-01
-2.80382723e-01 -3.12919497e-01 3.93825471e-01 3.67383882e-02
6.32343292e-01 4.59630787e-01 3.15436184e-01 7.07457885e-02
6.04978800e-01 2.24178553e-01 4.26116377e-01 6.59830213e-01
-3.91345937e-03 9.47157621e-01 4.17628229e-01 -4.53079551e-01
-1.11929536e+00 -7.77892709e-01 -9.23316404e-02 5.23305178e-01
6.27820075e-01 -2.67507195e-01 -9.86795366e-01 -3.20759743e-01
-3.60362232e-01 4.63651955e-01 -3.54124278e-01 5.41083775e-02
-7.22503066e-01 -4.74189430e-01 5.01454651e-01 3.79289448e-01
5.19127905e-01 -7.67198682e-01 -9.26945448e-01 2.42257580e-01
-5.69658168e-02 -1.19843984e+00 -3.49480748e-01 -2.80654639e-01
-9.94746089e-01 -1.29537725e+00 -1.03933346e+00 -6.05136395e-01
7.07690716e-01 3.93144339e-01 9.57721651e-01 9.71585810e-02
-3.68178546e-01 4.26885724e-01 -3.95958215e-01 -2.96027660e-01
-2.71045268e-01 -2.23213956e-01 -2.84406841e-01 2.34473273e-01
1.37219548e-01 -2.10256632e-02 -4.43985850e-01 6.45899832e-01
-6.28630519e-01 -1.26062110e-01 2.40044087e-01 4.98484701e-01
5.64462781e-01 2.17431694e-01 -3.06116134e-01 -1.04594839e+00
4.50327516e-01 1.23612665e-01 -7.82055259e-01 7.23468512e-02
-4.72155996e-02 -4.01790023e-01 2.02550724e-01 -1.12973116e-01
-9.84679282e-01 5.18419325e-01 -1.68580368e-01 -7.59589523e-02
-5.68183184e-01 6.53117970e-02 -6.08508959e-02 -4.18967336e-01
8.72780502e-01 1.49090499e-01 -2.65193760e-01 -3.26136798e-01
4.23077941e-01 4.13891345e-01 1.10802698e+00 -2.48765945e-01
8.72820556e-01 7.69119918e-01 -4.67862152e-02 -1.36188948e+00
-1.38808817e-01 -8.79637003e-01 -1.12492216e+00 -6.88244820e-01
6.43110037e-01 -4.80839163e-01 -7.71961391e-01 5.00038445e-01
-1.06571448e+00 1.78186253e-01 -5.07275701e-01 8.32241833e-01
-5.73604226e-01 8.71596634e-01 -5.78655601e-01 -9.20084357e-01
-5.87968230e-02 -8.91996145e-01 8.64494741e-01 5.85543573e-01
-1.18672974e-01 -8.26214433e-01 1.85495883e-01 3.16526085e-01
-2.11787984e-01 5.46273291e-01 1.74307197e-01 -4.17737871e-01
-2.57944494e-01 -9.73414719e-01 1.55290142e-01 -2.77350154e-02
-1.41045555e-01 6.23204708e-01 -7.87673712e-01 2.00075611e-01
-2.35585794e-01 4.12945330e-01 3.58181775e-01 4.90406781e-01
6.13069654e-01 1.54330611e-01 -3.21732432e-01 5.96357226e-01
1.25773001e+00 2.05330148e-01 9.35294390e-01 7.83109188e-01
5.59768498e-01 5.01418054e-01 7.91146338e-01 4.27262157e-01
2.89418608e-01 1.02996421e+00 2.54237801e-01 -5.27618766e-01
-1.49233192e-01 -3.01669687e-01 1.42875269e-01 3.62215936e-01
-5.86930037e-01 1.22210309e-01 -9.52125192e-01 3.69426548e-01
-1.58921897e+00 -9.73421514e-01 -9.94198561e-01 2.60439539e+00
2.81191587e-01 4.98152316e-01 6.37422383e-01 9.11992311e-01
1.04429805e+00 -3.38820100e-01 -1.51615287e-03 -5.70373118e-01
-2.00293764e-01 2.67681535e-02 8.30404043e-01 4.34686214e-01
-1.11506939e+00 8.77719104e-01 6.33229876e+00 5.76810062e-01
-1.08424401e+00 -2.49415010e-01 1.41322121e-01 4.47232634e-01
1.43309027e-01 1.09326862e-01 -5.56323111e-01 2.62038589e-01
1.11930154e-01 8.12024325e-02 -2.00533450e-01 7.71046638e-01
2.05140740e-01 -6.95748329e-01 -5.15742302e-01 9.41315353e-01
1.89365089e-01 -1.15052259e+00 -5.89368761e-01 -1.46393195e-01
5.55175424e-01 -4.73227382e-01 -2.20692828e-01 -2.65718192e-01
3.37811597e-02 -3.97613645e-01 1.06890130e+00 7.84345984e-01
4.19552922e-01 -8.53975773e-01 7.84423828e-01 3.23215485e-01
-1.15285861e+00 1.71187431e-01 -3.18369448e-01 -1.40269756e-01
1.99383155e-01 4.03955519e-01 -1.24117696e+00 7.12927163e-01
3.72493684e-01 1.83359206e-01 -6.29969001e-01 1.88058996e+00
-2.47346938e-01 5.01781285e-01 -6.71142578e-01 1.05923556e-01
3.78829823e-03 -5.54141641e-01 7.01849699e-01 1.33949757e+00
3.18386644e-01 -1.07225165e-01 -3.12394556e-02 5.05744338e-01
4.66593653e-01 6.07992053e-01 -4.80212152e-01 5.18064857e-01
2.89303303e-01 1.17920065e+00 -1.68776917e+00 -1.31514713e-01
-3.31582427e-01 1.18157005e+00 -3.24630380e-01 1.81157932e-01
-8.42529058e-01 -5.11632740e-01 -6.99039251e-02 4.61409122e-01
2.19829872e-01 -5.91684818e-01 -4.33886021e-01 -9.46728289e-01
1.31826416e-01 -6.66010380e-01 5.00710666e-01 -7.48853326e-01
-4.41975206e-01 6.43214941e-01 -1.19602449e-01 -1.44769371e+00
-2.40523696e-01 -3.95946831e-01 -6.53241336e-01 5.18554688e-01
-1.05696893e+00 -1.12032413e+00 -4.81829792e-01 5.48517466e-01
5.11518121e-01 1.06650338e-01 5.09096503e-01 4.16110128e-01
-4.59700406e-01 5.93125165e-01 1.65243044e-01 3.44264686e-01
4.56547678e-01 -1.26612592e+00 2.95614153e-01 1.18737245e+00
7.33234882e-01 2.22678468e-01 1.09305167e+00 -5.46944380e-01
-9.28093731e-01 -2.26816818e-01 6.78114772e-01 -2.65375197e-01
3.50878596e-01 -3.29353213e-01 -7.95069396e-01 5.29011071e-01
-3.17937016e-01 5.44544756e-02 3.96862417e-01 -4.28879149e-02
1.39139414e-01 -2.73298472e-02 -9.81134653e-01 3.05268526e-01
6.01956427e-01 -2.56344974e-01 -6.09332502e-01 2.86990911e-01
-2.46396929e-01 -9.16545749e-01 -5.62705934e-01 1.16996989e-01
6.70286953e-01 -1.34907484e+00 1.05398881e+00 -6.10268563e-02
-8.89061913e-02 -6.05479717e-01 4.11486059e-01 -8.91536891e-01
1.91243753e-01 -6.33520305e-01 3.98673534e-01 1.19874048e+00
2.08086416e-01 -2.36017093e-01 1.10070932e+00 5.62379658e-01
-2.26186603e-01 -1.43055007e-01 -8.51924539e-01 -5.93262494e-01
-5.37365139e-01 -4.82279181e-01 3.04633796e-01 9.27074254e-01
-3.30841653e-02 -2.30125681e-01 -3.27921748e-01 2.64605314e-01
5.31083703e-01 -5.95914386e-02 1.26871526e+00 -1.29992270e+00
-3.74456614e-01 -5.72474360e-01 -1.05886197e+00 -9.03489232e-01
-5.08052647e-01 -2.60581851e-01 -2.34744921e-02 -1.40385044e+00
-4.04646099e-01 -5.96236467e-01 4.08113182e-01 5.91105036e-02
-6.68981597e-02 7.78915882e-01 2.78693944e-01 2.16527224e-01
-2.89458871e-01 7.95421749e-02 9.92546856e-01 3.28185409e-01
-4.65085983e-01 6.24189973e-01 2.64632348e-02 1.35689461e+00
3.86927783e-01 -3.17350090e-01 5.94455972e-02 -6.75016716e-02
1.04397282e-01 9.40041393e-02 3.44643474e-01 -1.32656455e+00
5.26149869e-01 2.68886566e-01 4.05922532e-01 -7.03967690e-01
4.54433084e-01 -9.72090483e-01 3.50485444e-01 4.84616697e-01
7.48560354e-02 2.60654658e-01 1.66798070e-01 3.39874387e-01
-1.69027552e-01 -7.51602829e-01 5.62134445e-01 -5.55461273e-02
-8.93021047e-01 -2.66629398e-01 -3.23923081e-01 -4.84047323e-01
1.48453283e+00 -1.03681266e+00 3.13716829e-01 -4.92105961e-01
-1.00590312e+00 -2.05787838e-01 6.56764090e-01 1.21767193e-01
5.28589189e-01 -9.44591045e-01 -6.37031674e-01 2.08679974e-01
-7.21093118e-02 -4.63436060e-02 2.40331277e-01 8.61554861e-01
-1.51825047e+00 1.26602799e-01 -3.38364899e-01 -7.63438284e-01
-1.62905085e+00 3.28003824e-01 5.38052678e-01 1.34534329e-01
-1.09174967e+00 6.06328130e-01 -3.84549886e-01 -2.54808143e-02
1.48123652e-01 -3.90429169e-01 -5.52297592e-01 2.05126658e-01
4.09718573e-01 8.34869504e-01 2.55626798e-01 -9.96584713e-01
-1.48701683e-01 9.71633136e-01 1.55818507e-01 -3.05756897e-01
1.07786787e+00 -8.07918236e-02 3.47834468e-01 5.26835442e-01
4.40942198e-01 5.52517474e-01 -1.17376149e+00 -1.39969841e-01
1.45184159e-01 -8.10816050e-01 -7.29436427e-02 -3.70155066e-01
-9.44116294e-01 8.20671618e-01 6.76136613e-01 3.95520806e-01
8.79195392e-01 -4.06025320e-01 5.22699654e-01 -1.81431055e-01
4.14848119e-01 -1.12569237e+00 -3.42408538e-01 6.13168180e-02
5.68828762e-01 -6.85806990e-01 1.64499611e-01 -8.32323551e-01
-5.96553206e-01 1.59125876e+00 2.73236930e-01 -3.29589069e-01
2.67915547e-01 3.12415957e-01 1.62134141e-01 -1.79447621e-01
3.61878514e-01 -2.78839171e-01 2.65679806e-01 6.13215089e-01
3.49052042e-01 -1.11178830e-02 -7.41184771e-01 6.29662126e-02
-4.19837147e-01 1.30687384e-02 9.57517684e-01 1.03112316e+00
-4.51161236e-01 -8.18976879e-01 -1.06357288e+00 -6.26741722e-02
-4.27855581e-01 4.23341513e-01 -4.07239825e-01 1.20859897e+00
1.56288847e-01 6.83726847e-01 1.92403153e-01 -5.92984557e-02
8.02437782e-01 -2.77389288e-01 5.68547964e-01 -2.71916509e-01
-6.23521149e-01 3.44456047e-01 4.22594175e-02 -3.36630702e-01
-2.43924513e-01 -9.85982955e-01 -1.40885627e+00 -2.37868860e-01
-8.68236721e-01 3.40786546e-01 1.00072443e+00 7.68372893e-01
-1.97229505e-01 2.27051497e-01 4.52350944e-01 -8.48689139e-01
-9.90639031e-02 -5.79075217e-01 -8.06733429e-01 4.65756595e-01
-1.23567648e-01 -6.34871185e-01 7.28641376e-02 2.12194815e-01] | [8.14599895477295, -1.5667246580123901] |
5f3f6ebf-1df1-4335-b1a8-4efad34319bf | subtask-dominated-transfer-learning-for-long | 2112.00527 | null | https://arxiv.org/abs/2112.00527v1 | https://arxiv.org/pdf/2112.00527v1.pdf | Subtask-dominated Transfer Learning for Long-tail Person Search | Person search unifies person detection and person re-identification (Re-ID) to locate query persons from the panoramic gallery images. One major challenge comes from the imbalanced long-tail person identity distributions, which prevents the one-step person search model from learning discriminative person features for the final re-identification. However, it is under-explored how to solve the heavy imbalanced identity distributions for the one-step person search. Techniques designed for the long-tail classification task, for example, image-level re-sampling strategies, are hard to be effectively applied to the one-step person search which jointly solves person detection and Re-ID subtasks with a detection-based multi-task framework. To tackle this problem, we propose a Subtask-dominated Transfer Learning (STL) method. The STL method solves the long-tail problem in the pretraining stage of the dominated Re-ID subtask and improves the one-step person search by transfer learning of the pretrained model. We further design a Multi-level RoI Fusion Pooling layer to enhance the discrimination ability of person features for the one-step person search. Extensive experiments on CUHK-SYSU and PRW datasets demonstrate the superiority and effectiveness of the proposed method. | ['Shibao Zheng', 'Qin Zhou', 'Hua Yang', 'Chuang Liu'] | 2021-12-01 | null | null | null | null | ['person-search'] | ['computer-vision'] | [-1.20011546e-01 -5.23390889e-01 -5.36131151e-02 -3.77399027e-01
-9.74061131e-01 -3.39825720e-01 5.51258266e-01 -4.09705400e-01
-9.11168337e-01 5.20021319e-01 2.04046398e-01 2.16230527e-01
-1.54712006e-01 -5.83397985e-01 -4.70781267e-01 -6.28302336e-01
5.85972428e-01 7.37018943e-01 1.19770728e-01 8.80861878e-02
3.55713293e-02 3.63708198e-01 -1.70047164e+00 2.33854979e-01
9.22887623e-01 9.65199709e-01 1.45689473e-01 6.58209920e-01
2.40552172e-01 4.09532219e-01 -7.82765388e-01 -6.39281571e-01
4.28716272e-01 -5.41755795e-01 -7.80011058e-01 7.58536011e-02
8.05376828e-01 -5.88882208e-01 -4.89793271e-01 1.09304202e+00
1.09747577e+00 2.76872694e-01 6.42284393e-01 -1.37522316e+00
-6.99199677e-01 -7.56184608e-02 -8.86321306e-01 5.32100618e-01
5.19669592e-01 1.69517681e-01 6.39683366e-01 -1.13320494e+00
8.18160474e-02 1.52677989e+00 8.77366364e-01 7.83149660e-01
-9.19450402e-01 -1.04979825e+00 -3.50369141e-02 5.70019901e-01
-1.81651473e+00 -3.39768767e-01 4.34497595e-01 -5.64172745e-01
5.08807421e-01 2.75123119e-01 4.69657421e-01 1.01608562e+00
-3.43249947e-01 1.04100645e+00 9.55788970e-01 -3.17189872e-01
-3.68595541e-01 2.37678185e-01 3.21550429e-01 6.29792213e-01
4.66676831e-01 2.46581048e-01 -4.91745651e-01 -1.24981031e-01
9.06149268e-01 3.05773616e-01 -2.37441048e-01 -9.10572410e-02
-1.06623995e+00 6.73739314e-01 5.70936382e-01 2.38648281e-01
-2.98402458e-01 -7.34551474e-02 3.07837576e-01 1.41562194e-01
2.82680809e-01 2.08000988e-01 9.54493731e-02 3.49853992e-01
-1.14660096e+00 6.28363252e-01 4.11039054e-01 8.38911712e-01
6.78161144e-01 -4.64505136e-01 -9.63378191e-01 1.07109284e+00
2.17677653e-02 6.08867049e-01 7.34987378e-01 -3.22145313e-01
5.87032080e-01 6.41007364e-01 4.24449712e-01 -7.19339967e-01
-3.68583202e-01 -8.30812931e-01 -1.04832673e+00 -8.56493562e-02
7.84511805e-01 -8.28603804e-02 -1.03077173e+00 1.64344525e+00
4.27068144e-01 1.28476083e-01 -2.37170011e-01 1.41765463e+00
1.00222766e+00 3.34107518e-01 2.78594196e-01 1.26220345e-01
1.85348225e+00 -1.37525177e+00 -3.93658340e-01 -3.87206852e-01
2.32158020e-01 -5.23673236e-01 8.23948741e-01 -1.69521600e-01
-8.70600045e-01 -1.13133276e+00 -8.54190350e-01 -1.31225616e-01
-2.80584604e-01 5.93200982e-01 2.08295390e-01 8.82611752e-01
-7.55807757e-01 1.21078037e-01 -1.48582220e-01 -3.83946002e-01
6.92481935e-01 4.47985709e-01 -4.49903190e-01 -3.70061487e-01
-1.27390194e+00 7.48653173e-01 3.18680882e-01 2.30667502e-01
-7.12082982e-01 -7.09332287e-01 -6.44499600e-01 1.83242664e-01
4.31086212e-01 -1.03477502e+00 1.03831744e+00 -7.07783222e-01
-1.07288194e+00 1.32551908e+00 -3.61598849e-01 -2.88399637e-01
8.95152688e-01 -2.51773894e-01 -3.86084199e-01 4.48467284e-02
6.17660522e-01 4.39440072e-01 9.70167756e-01 -1.17198086e+00
-1.09089255e+00 -8.14400733e-01 -2.51565903e-01 6.00840092e-01
-2.80666649e-01 2.87937045e-01 -8.92608166e-01 -8.01888824e-01
-1.58951610e-01 -8.20730567e-01 -3.60533632e-02 -7.23933876e-02
-3.36704910e-01 -6.32653773e-01 3.92994434e-01 -9.61453736e-01
1.00837922e+00 -1.91945648e+00 -2.85510756e-02 5.59347048e-02
1.10290334e-01 4.04076040e-01 -1.64899752e-01 -1.55076563e-01
-4.85945307e-02 -3.07957381e-01 3.05580020e-01 -7.94128060e-01
-1.93183497e-02 -4.31836784e-01 6.30567688e-03 5.15498877e-01
-2.00273365e-01 1.14509678e+00 -6.64393485e-01 -6.38730347e-01
9.39789116e-02 1.78628057e-01 -2.08504930e-01 5.41916430e-01
6.19336128e-01 5.11793137e-01 -2.71033078e-01 8.45468879e-01
7.73418427e-01 -1.49747297e-01 -5.26609778e-01 -3.99479806e-01
1.67080816e-02 -2.71391332e-01 -1.14640856e+00 1.36804044e+00
-1.18540786e-01 6.46466985e-02 -1.25871133e-02 -8.26067209e-01
7.28589892e-01 8.02366585e-02 2.90971518e-01 -9.79052722e-01
-6.63570911e-02 2.25496560e-01 -2.16728702e-01 -4.74398464e-01
5.67723155e-01 -1.51882589e-01 -3.88941914e-01 4.06312883e-01
1.52676344e-01 6.59271479e-01 2.56002434e-02 -9.18400139e-02
5.61479747e-01 -2.85772793e-02 2.47526258e-01 -1.67644083e-01
1.01132035e+00 -1.65408731e-01 6.28281295e-01 1.21453285e+00
-7.13065267e-01 8.47718835e-01 -2.18603671e-01 -6.03573322e-01
-9.41733956e-01 -8.37007046e-01 -2.73694191e-02 1.58817434e+00
5.89392900e-01 1.00674227e-01 -8.90591681e-01 -8.05444479e-01
1.75811082e-01 1.74870014e-01 -6.59123540e-01 -2.80551255e-01
-5.41445374e-01 -1.19116795e+00 8.70131731e-01 4.84777093e-01
1.02826238e+00 -9.72516418e-01 -4.18079734e-01 -4.46389467e-02
-5.88444173e-01 -1.09220040e+00 -1.06151569e+00 -3.17447603e-01
-1.14778720e-01 -1.06161249e+00 -1.77725327e+00 -1.11812103e+00
7.64502048e-01 4.47117269e-01 7.32200205e-01 6.03454039e-02
-4.54293489e-01 4.07530636e-01 -8.72660354e-02 -2.72401810e-01
1.87194332e-01 1.04003340e-01 3.29420775e-01 4.05394167e-01
8.57542157e-01 2.21923683e-02 -1.01205552e+00 8.39876652e-01
-3.27399254e-01 1.54154710e-02 6.59988046e-01 1.08545017e+00
5.77967525e-01 1.86577037e-01 6.57212317e-01 -2.39885241e-01
6.96769476e-01 -1.12863891e-01 -2.28441417e-01 7.34742045e-01
-5.72253346e-01 -3.22925568e-01 2.63572961e-01 -6.88863695e-01
-1.14551187e+00 5.11833318e-02 -8.21571276e-02 -3.57160538e-01
-2.49029309e-01 -1.45446852e-01 -4.18522477e-01 -2.59109408e-01
5.30952930e-01 6.37942314e-01 -1.93305045e-01 -5.39705336e-01
-3.92137542e-02 9.27358747e-01 1.14919496e+00 -3.79733652e-01
9.99962568e-01 4.41937119e-01 -4.65657264e-01 -4.79814053e-01
-1.01663804e+00 -1.01044703e+00 -6.26194715e-01 -3.09529006e-01
9.90197480e-01 -1.43695343e+00 -8.19409490e-01 1.03884745e+00
-9.59888756e-01 -8.61420184e-02 -1.53029710e-01 3.67164910e-01
-3.26480955e-01 5.43966770e-01 -4.22847927e-01 -9.39688444e-01
-8.07273746e-01 -9.69293356e-01 1.20241427e+00 8.39994907e-01
4.02624533e-02 -3.77047032e-01 1.62405577e-02 8.48572612e-01
2.53675163e-01 -2.33795702e-01 3.13170761e-01 -7.50451684e-01
-4.61135030e-01 -7.38148868e-01 -6.84179187e-01 4.33384962e-02
-1.89779475e-02 -1.18284416e+00 -1.02443814e+00 -6.43271923e-01
-1.17558450e-01 -4.76016700e-01 1.21915245e+00 2.84156173e-01
1.20114732e+00 -5.38327917e-03 -5.07411242e-01 7.32739925e-01
1.11867976e+00 -8.74137431e-02 5.46109796e-01 4.74152088e-01
9.45603013e-01 5.92731118e-01 7.05842555e-01 2.39251405e-01
8.20262313e-01 1.03455234e+00 -1.81909338e-01 -3.69194895e-01
-5.13896406e-01 -4.61268932e-01 9.00375322e-02 -1.42095596e-01
-2.88985431e-01 -3.44518870e-02 -7.48512864e-01 6.31578088e-01
-1.98503017e+00 -1.21729684e+00 1.91147283e-01 2.38216901e+00
6.27849042e-01 -2.71667689e-01 5.59033871e-01 6.82396293e-02
1.16490889e+00 -3.70960794e-02 -5.26420414e-01 4.70366508e-01
-2.21498042e-01 -2.79979974e-01 4.92593288e-01 2.91389734e-01
-1.48258078e+00 8.28100860e-01 5.31124401e+00 1.23746037e+00
-6.02549791e-01 3.31048191e-01 6.33657575e-01 -2.90422179e-02
3.39267671e-01 -4.63787913e-01 -1.51038861e+00 7.74572551e-01
2.80760467e-01 -1.40559733e-01 3.17831725e-01 7.75931716e-01
-1.54885426e-01 -1.92964271e-01 -1.13874292e+00 1.82448173e+00
5.78297615e-01 -8.38400245e-01 4.64234315e-03 -1.40179582e-02
6.24923348e-01 -4.51125592e-01 8.39654207e-02 7.09809303e-01
-1.09698020e-01 -1.07392597e+00 6.47765875e-01 6.45207942e-01
1.00621545e+00 -7.62882769e-01 9.16132212e-01 4.94159877e-01
-1.43294144e+00 -4.89287198e-01 -4.77111995e-01 1.87249631e-01
3.28068376e-01 2.07669511e-01 -4.96970743e-01 6.39459014e-01
1.17295706e+00 1.81246787e-01 -9.56938505e-01 1.43326426e+00
1.06161438e-01 -9.48859081e-02 -2.96636581e-01 1.94470555e-01
-1.27337247e-01 5.26256561e-02 5.18656552e-01 1.25279021e+00
2.92969346e-01 3.18651348e-02 2.93495059e-01 8.94808412e-01
-7.70039111e-02 -1.76385760e-01 3.92419547e-02 4.27719146e-01
2.20128357e-01 1.10356259e+00 -4.01899159e-01 -3.93382996e-01
-2.04868227e-01 1.45289850e+00 3.48057568e-01 5.69326520e-01
-7.25067616e-01 -3.44422013e-01 4.68227923e-01 1.91112876e-01
2.79577643e-01 2.20973209e-01 -1.69502437e-01 -1.25687063e+00
1.35341108e-01 -8.37767005e-01 9.66320515e-01 -4.92650539e-01
-1.63295376e+00 5.61885953e-01 -2.73117796e-03 -1.05803728e+00
-1.81830689e-01 -2.76645094e-01 -5.86545467e-01 1.37170208e+00
-1.59147716e+00 -1.64188194e+00 -8.15118909e-01 9.82766092e-01
6.06497228e-01 -5.73268771e-01 6.26801968e-01 7.15147853e-01
-9.58463848e-01 1.29933393e+00 -3.04387093e-01 4.05465037e-01
9.98689234e-01 -1.00411499e+00 2.49499947e-01 1.06564999e+00
-2.54710048e-01 4.66243625e-01 3.58717352e-01 -7.83772111e-01
-9.44871902e-01 -1.11695075e+00 1.01098847e+00 -4.83864248e-01
-4.28283066e-02 -4.08018887e-01 -7.35002756e-01 3.40430319e-01
-4.20049131e-01 1.58172622e-02 6.79070354e-01 1.16868056e-01
-2.80449033e-01 -1.94810942e-01 -1.19640446e+00 4.06768501e-01
1.16967535e+00 -6.98297977e-01 -6.04597569e-01 2.23131806e-01
9.15338099e-02 -2.95670986e-01 -5.17525375e-01 4.38135982e-01
6.93798423e-01 -7.89321780e-01 1.61668682e+00 -2.66868353e-01
-8.23281631e-02 -3.84850442e-01 1.78018898e-01 -8.88401866e-01
-7.02673793e-01 -2.96098292e-01 -1.48327410e-01 1.41119266e+00
-1.80256456e-01 -4.85395014e-01 8.41109872e-01 7.17510343e-01
4.83259946e-01 -3.78874183e-01 -1.09370983e+00 -8.19090009e-01
-1.32283092e-01 1.35359541e-01 7.07027316e-01 4.78221476e-01
-4.94483620e-01 3.28710139e-01 -7.94362664e-01 3.12464178e-01
1.06209075e+00 3.58058274e-01 1.13289154e+00 -1.18127680e+00
-3.21769685e-01 -4.23868418e-01 -2.69410998e-01 -1.32262325e+00
-9.89370048e-02 -7.27534831e-01 1.57563329e-01 -1.48458433e+00
8.67738128e-01 -4.92540717e-01 -3.58190149e-01 2.76222348e-01
-8.47530603e-01 4.86605078e-01 2.37024039e-01 6.50334656e-01
-8.48202825e-01 6.32474363e-01 1.07917321e+00 -2.02542096e-01
-1.49855465e-01 3.76500398e-01 -8.79367113e-01 4.23934907e-01
4.38285202e-01 -2.96025813e-01 -8.47023651e-02 -4.08766031e-01
-3.56144577e-01 -2.10190326e-01 1.01044202e+00 -1.23435962e+00
7.86355197e-01 2.51361489e-01 1.06858146e+00 -9.05948520e-01
4.06510144e-01 -4.38117772e-01 -8.43966305e-02 4.93010134e-01
-1.25206798e-01 -3.29813153e-01 -5.97953089e-02 6.20312572e-01
-1.72992170e-01 -1.97411925e-01 8.53198349e-01 -2.78224289e-01
-8.37902904e-01 6.81692123e-01 1.12946793e-01 -2.35845689e-02
9.46549118e-01 -5.43061376e-01 -1.59675032e-01 -3.34556907e-01
-4.48719561e-01 7.36396372e-01 2.87522793e-01 5.30786574e-01
4.65686053e-01 -1.45949721e+00 -9.35290456e-01 3.19335699e-01
2.33352631e-01 -1.19420197e-02 7.26370215e-01 6.99483216e-01
-3.47024724e-02 4.10058022e-01 -2.98762977e-01 -4.51513052e-01
-1.47069418e+00 5.71461320e-01 7.77917862e-01 -5.31395137e-01
-4.76226926e-01 1.15769315e+00 6.73479199e-01 -3.91222179e-01
4.02796984e-01 6.53222978e-01 -4.28575933e-01 1.54839367e-01
9.87023354e-01 5.07615805e-01 -2.62213081e-01 -9.87263024e-01
-4.97459263e-01 7.51349211e-01 -3.02883655e-01 -7.27939792e-03
8.90852451e-01 -4.34005439e-01 1.68578714e-01 -1.27446413e-01
9.42897797e-01 -4.48917955e-01 -1.12687743e+00 -4.48840529e-01
-1.34406030e-01 -6.08580410e-01 -2.26549447e-01 -7.96767771e-01
-6.79403186e-01 6.82019114e-01 1.07157075e+00 -2.72586286e-01
9.62406993e-01 4.52424660e-02 1.06586111e+00 1.72596663e-01
4.08667833e-01 -1.27439582e+00 2.70626724e-01 2.29754791e-01
9.47352827e-01 -1.47105527e+00 -1.67962629e-03 -3.64676088e-01
-5.48155904e-01 6.68914914e-01 9.91442859e-01 -1.80048235e-02
2.67258584e-01 -2.57163554e-01 -9.33009386e-02 7.10975528e-02
2.23288268e-01 -5.86126685e-01 6.73190594e-01 7.51248121e-01
-3.45540464e-01 5.37005216e-02 -6.87465966e-02 1.25468445e+00
-1.53333127e-01 1.08097807e-01 -3.06422353e-01 3.96065265e-01
-2.27135420e-01 -8.86088490e-01 -9.01088417e-01 3.89740646e-01
-3.94746780e-01 2.09171891e-01 -3.07027936e-01 5.30602813e-01
6.17362618e-01 8.39659095e-01 1.05246350e-01 -4.11505640e-01
4.31077063e-01 8.39710906e-02 4.90639120e-01 -2.81788558e-01
-6.80548668e-01 -1.22849114e-01 -1.71903774e-01 -3.46925169e-01
-3.08001101e-01 -5.00431478e-01 -5.59112430e-01 -2.01781422e-01
-2.97487617e-01 9.31436010e-03 1.52185470e-01 9.51393902e-01
1.50543869e-01 2.32374370e-01 4.11674738e-01 -9.62000966e-01
-7.55029857e-01 -1.00009489e+00 -4.60451365e-01 7.86142468e-01
3.30709696e-01 -7.73551524e-01 -1.59985572e-01 -2.76207268e-01] | [14.833560943603516, 0.7984473705291748] |
b3c9735f-89b7-43da-9e30-17fcc077e255 | model-aware-gesture-to-gesture-translation | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Hu_Model-Aware_Gesture-to-Gesture_Translation_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Hu_Model-Aware_Gesture-to-Gesture_Translation_CVPR_2021_paper.pdf | Model-Aware Gesture-to-Gesture Translation | Hand gesture-to-gesture translation is a significant and interesting problem, which serves as a key role in many applications, such as sign language production. This task involves fine-grained structure understanding of the mapping between the source and target gestures. Current works follow a data-driven paradigm based on sparse 2D joint representation. However, given the insufficient representation capability of 2D joints, this paradigm easily leads to blurry generation results with incorrect structure. In this paper, we propose a novel model-aware gesture-to-gesture translation framework, which introduces hand prior with hand meshes as the intermediate representation. To take full advantage of the structured hand model, we first build a dense topology map aligning the image plane with the encoded embedding of the visible hand mesh. Then, a transformation flow is calculated based on the correspondence of the source and target topology map. During the generation stage, we inject the topology information into generation streams by modulating the activations in a spatially-adaptive manner. Further, we incorporate the source local characteristic to enhance the translated gesture image according to the transformation flow. Extensive experiments on two benchmark datasets have demonstrated that our method achieves new state-of-the-art performance. | ['Houqiang Li', 'Weichao Zhao', 'Wengang Zhou', 'Weilun Wang', 'Hezhen Hu'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['gesture-to-gesture-translation', 'sign-language-production'] | ['computer-vision', 'natural-language-processing'] | [ 3.94777685e-01 -3.28179747e-01 -1.31590515e-01 -3.31242740e-01
-4.46836054e-01 -3.77755105e-01 7.10769415e-01 -6.89236760e-01
-7.43457302e-02 3.54981095e-01 6.22713447e-01 1.90490812e-01
-2.85767228e-03 -7.46669412e-01 -6.57717645e-01 -8.21409166e-01
4.95176703e-01 5.66680312e-01 2.18053684e-01 -1.43809512e-01
4.04139727e-01 4.05646801e-01 -1.22271860e+00 3.87291968e-01
7.18394101e-01 7.09103763e-01 2.64503151e-01 4.08337206e-01
-5.07438898e-01 3.71568024e-01 -3.09219092e-01 -1.68344244e-01
4.57938254e-01 -7.25845456e-01 -6.29785120e-01 2.20205873e-01
4.05332685e-01 -5.31920552e-01 -5.68642080e-01 9.69022512e-01
5.85624635e-01 2.37811998e-01 5.84054172e-01 -9.81980145e-01
-5.85566342e-01 4.54560965e-01 -8.30112755e-01 -2.82690406e-01
4.17578071e-01 4.73378450e-01 7.73845077e-01 -1.06853127e+00
1.08202672e+00 1.49092996e+00 1.52550280e-01 6.22841299e-01
-1.03760195e+00 -8.41712952e-01 4.51466382e-01 2.28601798e-01
-1.38902879e+00 -2.25244492e-01 1.11028421e+00 -4.88041848e-01
5.36220729e-01 2.64140189e-01 1.02189338e+00 1.08420742e+00
-1.21606246e-01 7.82861471e-01 1.15352786e+00 -4.34530139e-01
-1.15784165e-02 -5.29046476e-01 -3.21764112e-01 7.62433410e-01
-2.06556067e-01 1.68226868e-01 -9.09958780e-01 9.46303084e-02
1.49017179e+00 1.17754489e-01 -4.50421780e-01 -4.41167772e-01
-1.57758403e+00 4.92534012e-01 7.21699178e-01 4.31979120e-01
-5.77979863e-01 2.67949432e-01 -7.97212124e-02 2.96390038e-02
2.14342460e-01 1.35554656e-01 5.05755143e-03 -2.88312107e-01
-8.65933836e-01 2.73849756e-01 4.81207907e-01 8.27375650e-01
5.60114980e-01 -8.78271237e-02 -4.50787097e-01 5.38983285e-01
6.36052668e-01 5.56193411e-01 3.94801050e-01 -8.29812825e-01
9.02853549e-01 7.95149386e-01 -4.05542627e-02 -1.00236356e+00
-1.43077239e-01 -2.32533708e-01 -1.01140666e+00 3.86426657e-01
6.99794531e-01 2.46792704e-01 -1.16984797e+00 1.57083488e+00
5.56322038e-01 4.32488710e-01 -4.10559297e-01 1.38939285e+00
4.89223331e-01 5.66423297e-01 -8.19965675e-02 -7.86027685e-03
1.27013826e+00 -1.00520372e+00 -7.33718693e-01 1.40927956e-01
1.20303966e-02 -9.33143079e-01 1.23067176e+00 1.45376444e-01
-1.08941603e+00 -5.37830889e-01 -7.40126908e-01 -2.43874162e-01
1.46536738e-01 9.99299958e-02 3.59090120e-01 1.60109103e-01
-6.08169019e-01 4.46110338e-01 -1.18927252e+00 -2.41853297e-01
3.92087340e-01 1.00268140e-01 -2.87488252e-01 -1.63936317e-01
-7.21941948e-01 6.64346039e-01 1.03721619e-01 5.63327432e-01
-3.90973777e-01 -5.13369024e-01 -7.18000412e-01 -2.14721993e-01
2.37006664e-01 -9.68661249e-01 1.01749825e+00 -6.68273807e-01
-2.00885057e+00 5.03168106e-01 -4.28965986e-01 3.19612145e-01
9.95314002e-01 -1.46732360e-01 1.57111168e-01 1.24498464e-01
-9.07402709e-02 7.00538993e-01 1.02154136e+00 -1.32346737e+00
-5.20923197e-01 -6.68069482e-01 -1.03132263e-01 6.42133832e-01
-1.26205593e-01 -1.11057930e-01 -7.13480294e-01 -9.48614657e-01
7.03193724e-01 -9.01251018e-01 -1.39806554e-01 4.18554187e-01
-4.61613148e-01 -1.01985313e-01 8.78948748e-01 -9.30177689e-01
1.27554381e+00 -2.07103372e+00 9.03641284e-01 5.73328972e-01
1.27977654e-01 1.95378941e-02 -2.17031151e-01 2.48235464e-01
3.36059123e-01 -1.88922375e-01 -3.68191719e-01 -2.95141250e-01
-9.68421716e-03 1.28222153e-01 -4.28043693e-01 2.98755199e-01
1.68205321e-01 1.16242909e+00 -9.47324157e-01 -5.79547644e-01
4.58709151e-01 7.73832321e-01 -4.74072486e-01 4.84894633e-01
-3.47714990e-01 1.09450698e+00 -8.67550910e-01 6.84601605e-01
5.08600295e-01 -6.93584681e-02 6.70362338e-02 -5.64773440e-01
-2.92425931e-01 1.20433062e-01 -1.24210405e+00 2.34203434e+00
-3.72160912e-01 3.74573499e-01 3.92149128e-02 -6.50467217e-01
8.71889532e-01 1.87784687e-01 4.07318383e-01 -5.24981320e-01
1.88886479e-01 3.34911913e-01 4.60780114e-02 -4.92439359e-01
1.30465686e-01 -6.54135123e-02 3.17072272e-01 6.24728382e-01
-4.21387941e-01 -3.71361613e-01 -2.07800165e-01 -1.99587330e-01
6.08181357e-01 7.32564151e-01 -7.65756145e-02 1.85402423e-01
3.84904295e-01 -6.24461398e-02 2.01051369e-01 5.11668585e-02
1.98497638e-01 9.40872967e-01 2.12102622e-01 -4.84807968e-01
-8.96999240e-01 -9.62993979e-01 2.25334987e-01 6.26557946e-01
3.92586052e-01 -1.15412809e-01 -8.36119056e-01 -4.15117085e-01
-2.29876086e-01 2.32790321e-01 -6.00184023e-01 2.43167114e-02
-1.20183718e+00 -4.05795932e-01 2.95541257e-01 7.02441812e-01
7.69911170e-01 -1.22390819e+00 -7.86345661e-01 2.27120593e-01
-4.08635974e-01 -9.05446768e-01 -9.29776013e-01 -5.76685309e-01
-9.17766809e-01 -8.93163979e-01 -1.21940470e+00 -9.14121211e-01
8.40068996e-01 1.30458191e-01 5.37324667e-01 2.82195657e-01
-2.65627831e-01 -3.63249052e-03 -2.62786329e-01 2.40243658e-01
-2.46144429e-01 5.53191043e-02 -2.29591027e-01 3.94862682e-01
-1.00336410e-02 -7.96495318e-01 -7.99724042e-01 3.94612759e-01
-8.90827894e-01 5.61428905e-01 7.78883219e-01 8.52760673e-01
7.33878136e-01 -4.14625019e-01 -3.49248908e-02 -1.74957335e-01
5.75397849e-01 7.36868531e-02 -3.43005538e-01 3.97446454e-01
-1.08602867e-01 4.05074120e-01 2.26573810e-01 -6.88632846e-01
-1.20843840e+00 4.28950310e-01 -6.31711930e-02 -7.10091293e-01
6.94650486e-02 3.24314505e-01 -4.68041420e-01 4.82063442e-02
2.85928190e-01 3.56941909e-01 1.22101910e-01 -7.32004821e-01
6.58864558e-01 5.91780007e-01 4.66400981e-01 -6.36208355e-01
1.00120783e+00 5.97623289e-01 1.19730562e-01 -6.44313633e-01
-2.24903956e-01 -1.79131553e-01 -1.11182940e+00 -2.59745210e-01
9.62980151e-01 -3.57110679e-01 -4.99755591e-01 8.09483826e-01
-1.44789255e+00 -5.68464816e-01 -2.23184645e-01 6.30216300e-01
-5.66347897e-01 3.83019269e-01 -4.66203183e-01 -4.82421190e-01
-3.02594632e-01 -1.36008275e+00 1.39014673e+00 9.36448425e-02
-2.25023121e-01 -5.91414034e-01 7.68063366e-02 1.58223540e-01
2.24650308e-01 4.84264702e-01 9.43607330e-01 1.49647340e-01
-1.05418670e+00 1.13110237e-01 -3.29231381e-01 -1.05782546e-01
4.72566336e-01 2.74929237e-02 -6.06039941e-01 -5.16612194e-02
-3.82698089e-01 -6.60576001e-02 6.55619681e-01 2.92501926e-01
9.07454312e-01 -1.74459681e-01 -2.57743031e-01 7.73755312e-01
1.01935196e+00 -9.90224332e-02 5.57280600e-01 1.53892376e-02
1.21401012e+00 8.16777945e-01 5.45274198e-01 3.37444395e-01
3.42987239e-01 1.18266487e+00 1.81946591e-01 6.71380237e-02
-7.62733102e-01 -8.28334570e-01 2.15900034e-01 8.95407438e-01
-7.16624856e-01 2.00726643e-01 -6.66103244e-01 1.44232124e-01
-1.96499312e+00 -7.78998554e-01 -7.44890943e-02 2.13780761e+00
9.57676291e-01 -1.49676010e-01 -2.10464913e-02 7.83981476e-03
6.49804831e-01 2.86279947e-01 -6.06534183e-01 2.00054809e-01
1.86136886e-01 3.82810801e-01 -4.00849171e-02 7.27494955e-01
-5.23293138e-01 1.17984593e+00 4.98922920e+00 7.81060934e-01
-1.46585536e+00 -7.07780197e-02 2.08926597e-03 -8.75431672e-02
-3.70431125e-01 -3.09488382e-02 -4.32090849e-01 4.28968012e-01
-2.23294608e-02 6.18601814e-02 6.28892660e-01 2.68599987e-01
4.98768628e-01 2.90259331e-01 -1.13973236e+00 1.09775591e+00
-5.43535836e-02 -1.32229769e+00 4.36150610e-01 3.00842494e-01
7.13592649e-01 -3.75782490e-01 -1.28047973e-01 -1.72801346e-01
-1.23793855e-01 -9.60296094e-01 9.93162751e-01 7.22231448e-01
1.18666923e+00 -3.79743189e-01 2.32039958e-01 3.28790933e-01
-1.70108747e+00 3.42461646e-01 -2.12930515e-02 -1.45030871e-01
5.05900323e-01 1.68790430e-01 -5.40934741e-01 4.72468734e-01
4.73177463e-01 6.44224942e-01 -1.55795678e-01 7.99393058e-01
-6.27946019e-01 3.32722843e-01 -4.46216077e-01 2.95056030e-02
5.21069616e-02 -3.75784427e-01 5.09546101e-01 8.88886154e-01
3.89417380e-01 4.06097949e-01 2.29482681e-01 1.05379295e+00
-1.24428077e-02 6.97245821e-02 -3.78079742e-01 1.49074346e-01
4.60708350e-01 9.34229255e-01 -9.36630011e-01 -3.32613260e-01
-6.47568852e-02 1.44669521e+00 1.30636990e-01 6.04681671e-01
-4.84806836e-01 -1.61859572e-01 5.91756701e-01 9.56399739e-02
1.07563287e-01 -6.02390289e-01 -5.45472324e-01 -1.24822080e+00
4.74330783e-01 -7.05129743e-01 -1.29947975e-01 -5.37946582e-01
-9.17487741e-01 4.74188894e-01 -1.35800848e-02 -1.42300868e+00
-3.94877732e-01 -3.62710804e-01 -6.57236099e-01 1.16316533e+00
-1.31681836e+00 -1.50949502e+00 -7.39417672e-01 7.70532787e-01
7.79376566e-01 2.56706476e-01 7.64131725e-01 1.46851316e-01
-2.30867103e-01 4.30277377e-01 -3.37162197e-01 3.89697850e-01
5.97715914e-01 -8.41114163e-01 6.53676927e-01 7.01896966e-01
4.03659254e-01 5.54472089e-01 2.26062819e-01 -8.73267412e-01
-1.49856496e+00 -9.00089264e-01 6.44966006e-01 -5.47830045e-01
3.50678056e-01 -1.66969568e-01 -9.03602958e-01 4.61244285e-01
-7.26741031e-02 2.43988875e-02 7.25649074e-02 -4.29950595e-01
-3.83334756e-01 1.46159261e-01 -8.40194046e-01 7.45511889e-01
1.62767529e+00 -4.28853720e-01 -6.74737096e-01 1.18948556e-01
5.00660717e-01 -8.79785657e-01 -6.36234462e-01 1.94046184e-01
1.16376996e+00 -6.26399219e-01 8.82176042e-01 -5.30467391e-01
6.12490356e-01 -5.34459829e-01 -8.81330967e-02 -1.37742078e+00
-2.55715966e-01 -6.62892520e-01 -2.69264579e-01 1.01546228e+00
-1.83825456e-02 -2.89662570e-01 8.74751627e-01 6.35963798e-01
1.93087701e-02 -8.62889051e-01 -1.02928662e+00 -5.39577663e-01
-9.48180333e-02 -1.69128418e-01 8.55932772e-01 7.18963683e-01
-1.83209926e-01 2.69963983e-02 -2.53963202e-01 1.05305754e-01
6.64082468e-01 4.65011775e-01 9.24069822e-01 -9.01607275e-01
-2.72964448e-01 -7.69996524e-01 -3.85499179e-01 -1.60148191e+00
-4.27293852e-02 -7.71448076e-01 1.58473805e-01 -1.78339732e+00
-4.60737525e-03 -5.21743536e-01 4.84335013e-02 3.75934482e-01
-1.83930084e-01 4.41982627e-01 4.31400180e-01 5.69838822e-01
1.39919281e-01 7.43188620e-01 1.94492102e+00 -3.09823025e-02
-5.12711763e-01 -7.75785670e-02 -1.58696011e-01 7.10974753e-01
5.19236863e-01 -5.08059449e-02 -3.30051214e-01 -9.08740819e-01
-2.54962891e-01 1.73525475e-02 5.26161730e-01 -6.04665995e-01
2.41442263e-01 -5.30527234e-01 2.65066445e-01 -4.75136250e-01
5.25450349e-01 -8.44031930e-01 2.73249179e-01 5.87083399e-01
-3.64538372e-01 -2.84619331e-02 -3.71172130e-01 4.77232814e-01
-2.03167439e-01 2.29466602e-01 5.31941295e-01 -8.25835988e-02
-5.62615395e-01 5.51436722e-01 4.34851140e-01 -1.79328352e-01
8.09459031e-01 -4.54031497e-01 1.37101799e-01 -2.50078291e-01
-5.91071486e-01 1.63880512e-01 5.95382333e-01 6.37749851e-01
1.02446330e+00 -1.55816782e+00 -8.70985031e-01 5.89464962e-01
-1.52004417e-02 5.22512734e-01 1.72646269e-01 8.60162497e-01
-5.16434848e-01 1.53556243e-01 -3.80601585e-01 -8.52809668e-01
-1.16874206e+00 -1.00817636e-01 2.08381206e-01 -6.94630966e-02
-9.98769879e-01 8.38736296e-01 3.51356030e-01 -2.33347580e-01
2.88700312e-01 -6.59346581e-01 1.32485360e-01 -2.00506166e-01
6.38008118e-01 4.52197701e-01 -9.20572951e-02 -8.96762788e-01
-2.14817613e-01 1.37017500e+00 3.08496207e-01 -6.90602303e-01
1.11067557e+00 1.09388925e-01 -2.55899858e-02 1.38969138e-01
1.03743732e+00 5.41812740e-02 -1.69246686e+00 -3.10892045e-01
-2.80529618e-01 -9.83354032e-01 -2.15133473e-01 -6.64076447e-01
-1.20504546e+00 1.28291464e+00 4.11932796e-01 -5.47558725e-01
9.79892313e-01 3.12109981e-02 1.01400733e+00 8.27658921e-02
5.72690189e-01 -7.13654935e-01 1.96839377e-01 2.90401191e-01
1.38530576e+00 -8.04665625e-01 -1.51366323e-01 -6.37166798e-01
-4.16517168e-01 1.18026769e+00 6.22700036e-01 -9.28337649e-02
5.11735082e-01 2.02384479e-02 2.22813085e-01 -1.51286989e-01
-1.41360149e-01 -4.42265347e-02 4.81418163e-01 6.53195918e-01
2.63089508e-01 2.14143351e-01 -3.90767664e-01 2.25128427e-01
-3.12013119e-01 1.86110362e-01 -1.45661592e-01 9.76902425e-01
-1.55711904e-01 -1.44245327e+00 -5.35961330e-01 9.55120772e-02
2.87230641e-01 3.32267577e-04 -5.72303057e-01 5.19716203e-01
1.07871138e-01 4.45132196e-01 -6.85304403e-05 -5.28749347e-01
6.06115639e-01 6.59900978e-02 9.92169201e-01 -5.91434717e-01
-2.97185272e-01 3.74683082e-01 -3.89120996e-01 -7.61230230e-01
-5.96609533e-01 -6.76357567e-01 -1.60653603e+00 -3.18954326e-02
3.68093350e-03 -3.08130711e-01 7.55729020e-01 1.00268233e+00
3.54185224e-01 3.53619158e-01 4.28251952e-01 -1.43593240e+00
-6.08782947e-01 -1.04411852e+00 -2.81465262e-01 6.33624136e-01
2.11377397e-01 -9.02881861e-01 3.46432813e-02 2.18498707e-01] | [11.234539031982422, -0.897448718547821] |
7835c19f-c869-49cc-b83f-3db63a343f3a | shallow-attention-network-for-polyp | 2108.00882 | null | https://arxiv.org/abs/2108.00882v1 | https://arxiv.org/pdf/2108.00882v1.pdf | Shallow Attention Network for Polyp Segmentation | Accurate polyp segmentation is of great importance for colorectal cancer diagnosis. However, even with a powerful deep neural network, there still exists three big challenges that impede the development of polyp segmentation. (i) Samples collected under different conditions show inconsistent colors, causing the feature distribution gap and overfitting issue; (ii) Due to repeated feature downsampling, small polyps are easily degraded; (iii) Foreground and background pixels are imbalanced, leading to a biased training. To address the above issues, we propose the Shallow Attention Network (SANet) for polyp segmentation. Specifically, to eliminate the effects of color, we design the color exchange operation to decouple the image contents and colors, and force the model to focus more on the target shape and structure. Furthermore, to enhance the segmentation quality of small polyps, we propose the shallow attention module to filter out the background noise of shallow features. Thanks to the high resolution of shallow features, small polyps can be preserved correctly. In addition, to ease the severe pixel imbalance for small polyps, we propose a probability correction strategy (PCS) during the inference phase. Note that even though PCS is not involved in the training phase, it can still work well on a biased model and consistently improve the segmentation performance. Quantitative and qualitative experimental results on five challenging benchmarks confirm that our proposed SANet outperforms previous state-of-the-art methods by a large margin and achieves a speed about 72FPS. | ['Shuguang Cui', 'S. Kevin Zhou', 'Zhen Li', 'Ruimao Zhang', 'Yiwen Hu', 'Jun Wei'] | 2021-08-02 | null | null | null | null | ['video-polyp-segmentation'] | ['computer-vision'] | [ 2.02357382e-01 -2.10383870e-02 -1.66526318e-01 9.78119522e-02
-4.05509442e-01 -2.25107953e-01 -1.57109529e-01 2.30488807e-01
-3.08774322e-01 4.21102643e-01 -4.34817597e-02 -3.12561214e-01
2.25844920e-01 -8.25544536e-01 -5.82910061e-01 -1.07124329e+00
3.98368955e-01 -1.55951634e-01 6.54910684e-01 1.01049036e-01
1.02378145e-01 1.86946228e-01 -1.14930081e+00 2.06035644e-01
1.42044139e+00 8.46795857e-01 3.11199665e-01 4.79145259e-01
-2.83394635e-01 4.83373404e-01 -6.09960198e-01 -2.64998585e-01
2.34694570e-01 -3.97246361e-01 -3.86914521e-01 2.48828501e-01
1.66152894e-01 -5.09270668e-01 -3.13831031e-01 1.56030869e+00
5.06000638e-01 -1.92749888e-01 3.22779149e-01 -7.22013712e-01
-5.19293368e-01 4.12355900e-01 -9.21479404e-01 2.72878349e-01
-3.90771627e-02 4.11789298e-01 5.72204292e-01 -6.51913524e-01
7.53560811e-02 1.02681088e+00 7.09020019e-01 4.39240515e-01
-9.59768534e-01 -5.34198523e-01 4.31013852e-01 -1.48968488e-01
-1.26029658e+00 2.34682653e-02 6.55787826e-01 -1.69445693e-01
1.68529764e-01 4.28169012e-01 1.03192353e+00 6.69388175e-01
2.95270473e-01 1.00531852e+00 7.91116238e-01 -1.59904540e-01
-2.68561952e-02 1.71115488e-01 -7.73885101e-02 9.55983698e-01
6.76650822e-01 -7.62364343e-02 1.81793526e-01 1.11660913e-01
9.30587053e-01 4.07147467e-01 -5.29735923e-01 -1.22463107e-01
-1.34095085e+00 5.14936328e-01 8.51593137e-01 4.08263922e-01
-9.99500081e-02 3.33545581e-02 4.03482497e-01 -2.90579408e-01
1.29496962e-01 3.74636531e-01 -2.17628658e-01 2.02963352e-01
-7.64675617e-01 -1.84957191e-01 5.26829362e-01 6.61899447e-01
6.01184011e-01 -7.99540058e-02 -3.91448766e-01 6.84646726e-01
3.83507133e-01 4.54889357e-01 7.54961193e-01 -5.48703611e-01
4.24801499e-01 1.03140938e+00 1.27490312e-01 -1.33601987e+00
-4.18312460e-01 -8.31923306e-01 -1.11086130e+00 -6.46316074e-03
6.33975506e-01 -9.78306159e-02 -1.12889636e+00 1.32613719e+00
3.21669728e-01 2.59020150e-01 -1.37357444e-01 1.07274711e+00
1.02049255e+00 6.00484908e-01 1.30764008e-01 -1.46897972e-01
1.50850344e+00 -1.14134812e+00 -5.97588420e-01 -5.50572932e-01
3.81205052e-01 -7.77694941e-01 1.19153321e+00 3.34720939e-01
-1.07447004e+00 -5.83992302e-01 -1.23433137e+00 -5.78489937e-02
-7.17370063e-02 5.91545641e-01 6.92471862e-01 7.64444947e-01
-5.52881479e-01 2.12696910e-01 -9.70269144e-01 -1.15368776e-01
6.73284531e-01 2.73293793e-01 1.60613999e-01 -3.55729729e-01
-8.68056893e-01 2.78703898e-01 2.89117038e-01 5.15043080e-01
-4.48544979e-01 -7.37577260e-01 -7.65443027e-01 1.96948260e-01
2.86602527e-01 -6.15882695e-01 9.42011297e-01 -1.14392686e+00
-1.48376989e+00 5.30678391e-01 -3.47610079e-02 -6.52122647e-02
7.16041327e-01 -5.78078963e-02 -1.76857948e-01 2.04101384e-01
-5.55404015e-02 6.43900096e-01 6.60003901e-01 -1.29156828e+00
-8.49345505e-01 -2.11892068e-01 4.08486351e-02 3.90892535e-01
-5.01482725e-01 -3.78228962e-01 -1.02749670e+00 -7.24731147e-01
4.13094938e-01 -8.85019183e-01 -6.11645937e-01 3.39940965e-01
-5.43869436e-01 3.16611886e-01 5.91976106e-01 -7.62667418e-01
1.34961867e+00 -2.60610223e+00 -1.49503976e-01 4.99542952e-02
2.82527447e-01 5.11703670e-01 6.68262616e-02 -5.24272144e-01
2.14666143e-01 2.23694488e-01 -4.25147891e-01 3.26412208e-02
-3.71506065e-01 2.83435881e-01 1.78407341e-01 4.96919721e-01
1.90456852e-01 8.48958373e-01 -1.12418664e+00 -8.34689438e-01
3.40922832e-01 3.81043732e-01 -5.83019316e-01 -7.93054979e-03
-1.16439268e-01 3.39347869e-01 -5.64864516e-01 7.79369652e-01
1.14304245e+00 -4.18481469e-01 -1.26592647e-02 -4.55170155e-01
-1.46702245e-01 -7.01053217e-02 -1.21488893e+00 1.41220689e+00
-3.42920423e-01 2.62948841e-01 1.32613733e-01 -8.01838160e-01
6.45241499e-01 9.38755497e-02 3.76645625e-01 -6.38422906e-01
3.91757220e-01 4.30089772e-01 3.36892903e-01 -6.61811233e-01
2.35148281e-01 9.74599943e-02 1.65384874e-01 -3.00219711e-02
-5.56737006e-01 9.09046456e-02 2.59190083e-01 1.17652072e-02
7.27738678e-01 -2.15593249e-01 1.72777489e-01 -2.13973030e-01
6.89983010e-01 -1.22781061e-02 1.13181138e+00 4.38768178e-01
-4.89064991e-01 8.42195809e-01 5.86399078e-01 -4.16668326e-01
-5.84832430e-01 -8.17378461e-01 -6.42142370e-02 5.45867920e-01
7.99908340e-01 7.97511935e-02 -6.55035138e-01 -7.89638400e-01
-3.05206656e-01 1.14306122e-01 -6.37624860e-01 -2.56484181e-01
-6.92422330e-01 -1.43396139e+00 3.03959548e-01 6.75081551e-01
8.63467276e-01 -8.39917541e-01 -6.98397517e-01 3.31934363e-01
-3.55826765e-01 -9.11776960e-01 -7.62049675e-01 -3.98849696e-02
-9.04399812e-01 -1.17665982e+00 -1.16645408e+00 -1.11142731e+00
1.32531691e+00 5.59659898e-01 8.51756632e-01 6.53782248e-01
-4.83557314e-01 -4.16001678e-01 -3.30498785e-01 -3.50159556e-01
-3.02463770e-01 -1.08662128e-01 -6.01494193e-01 -9.68662556e-03
1.45814106e-01 -2.22780645e-01 -1.20064580e+00 3.18273515e-01
-9.72906947e-01 2.75518924e-01 9.78714883e-01 1.08565688e+00
7.09469497e-01 4.55496192e-01 1.26483634e-01 -8.23919594e-01
1.22853182e-01 -1.34952694e-01 -7.51538575e-01 1.64356142e-01
-2.65133649e-01 -3.28272402e-01 7.93768942e-01 -6.19399965e-01
-1.15538490e+00 2.23003194e-01 -1.49613112e-01 -1.06363796e-01
6.02491312e-02 3.21603477e-01 -2.65987128e-01 -2.52385944e-01
3.87254983e-01 3.18889260e-01 3.78126316e-02 -2.35468984e-01
-1.64943054e-01 4.91853178e-01 3.42228651e-01 -5.82192093e-03
6.02257013e-01 6.57333732e-01 -1.40473038e-01 -6.07350707e-01
-8.05675387e-01 -3.30908686e-01 1.46435376e-03 2.83723120e-02
9.17541742e-01 -9.22086418e-01 -5.72007358e-01 7.53813744e-01
-8.61354947e-01 -2.72739708e-01 -3.89628969e-02 5.55401981e-01
2.39457130e-01 6.13420904e-01 -9.61730957e-01 -4.71022248e-01
-4.81159985e-01 -1.47071218e+00 7.47385800e-01 9.99690175e-01
2.88050920e-01 -8.13197792e-01 -4.18201923e-01 1.85756832e-01
3.61621469e-01 2.10385844e-01 9.94435012e-01 -1.52029067e-01
-8.26697290e-01 -1.52892888e-01 -7.28635252e-01 2.33703405e-01
3.84758562e-01 3.83694172e-01 -7.51311779e-01 -3.19922328e-01
-4.73336205e-02 1.62947178e-01 1.03826678e+00 6.94191098e-01
1.39434326e+00 -1.81463435e-01 -6.89376354e-01 8.97988498e-01
1.45170033e+00 2.28987545e-01 6.60865784e-01 2.45976537e-01
7.44200170e-01 3.17615032e-01 6.83688283e-01 2.15559244e-01
3.06261688e-01 1.74631044e-01 6.05236888e-01 -7.66692460e-01
-4.26755339e-01 -2.00916693e-01 -7.95170106e-03 6.84360743e-01
2.28212908e-01 -1.75932810e-01 -5.95544279e-01 7.37744629e-01
-1.64921176e+00 -3.62329304e-01 -2.74060220e-01 2.30392647e+00
8.14332008e-01 2.83768326e-01 -9.22263041e-02 1.17892034e-01
7.23633409e-01 -4.05922681e-02 -6.66638672e-01 1.47956640e-01
-8.13999474e-02 -1.63915083e-01 5.73933125e-01 2.29268566e-01
-1.14978659e+00 4.96392906e-01 5.46290922e+00 8.91658783e-01
-1.35796475e+00 -1.78669527e-01 1.12219751e+00 1.95695758e-01
-2.92842507e-01 -2.42104873e-01 -7.43221164e-01 8.76664877e-01
-1.56072259e-01 1.59177676e-01 7.10963458e-02 7.00079560e-01
-1.88663043e-02 -3.47436100e-01 -5.54648042e-01 8.41799498e-01
-4.48658355e-02 -1.02689159e+00 -8.10580924e-02 1.16288736e-02
8.54117393e-01 -1.84302583e-01 1.14613540e-01 1.15932927e-01
3.38131413e-02 -7.95623541e-01 3.87377739e-01 2.20333695e-01
5.68552077e-01 -7.66708732e-01 1.00659049e+00 2.08434552e-01
-1.17639494e+00 -2.34421596e-01 -6.39419377e-01 4.61742163e-01
-1.43441439e-01 1.08886421e+00 -5.92049420e-01 4.61458772e-01
7.00517416e-01 6.64067805e-01 -7.68438876e-01 1.64236593e+00
-3.30465168e-01 4.13224757e-01 -3.81795436e-01 -1.34370685e-01
1.75784633e-01 -1.16787419e-01 4.76867884e-01 1.20546651e+00
4.00947213e-01 -4.19239923e-02 2.02856943e-01 9.48246837e-01
5.69027252e-02 7.56359994e-02 1.74008742e-01 1.37875840e-01
2.71863520e-01 1.40625417e+00 -1.14225984e+00 -3.14416081e-01
-5.52591741e-01 9.65403080e-01 -1.31444007e-01 4.32639629e-01
-9.36385095e-01 -4.29289341e-01 1.69608593e-01 -6.68537691e-02
3.85780245e-01 2.01322168e-01 -4.40189511e-01 -1.19476724e+00
2.46424556e-01 -8.72589111e-01 3.67160559e-01 -3.09796453e-01
-1.03250766e+00 4.76113617e-01 -6.89514399e-01 -1.46412945e+00
6.62050605e-01 -5.15182495e-01 -8.10428858e-01 6.28249109e-01
-1.86074233e+00 -9.26222026e-01 -7.54342377e-01 2.57444233e-01
5.53137124e-01 3.94676805e-01 4.54257220e-01 5.96824586e-01
-9.63326454e-01 7.01051056e-01 1.21938000e-02 2.69881845e-01
6.72005117e-01 -1.28095829e+00 1.39557630e-01 9.78686273e-01
-3.22340548e-01 4.94415700e-01 4.21847761e-01 -6.28126800e-01
-1.10200047e+00 -1.15320468e+00 3.18523675e-01 1.65739790e-01
1.81197256e-01 -6.69726878e-02 -9.97632325e-01 1.72141865e-01
-1.91792786e-01 3.53251249e-01 5.01171827e-01 -4.91848677e-01
1.63091287e-01 -2.36623421e-01 -1.16012096e+00 7.77049780e-01
5.20309627e-01 1.77784115e-01 -1.44136116e-01 3.55662763e-01
6.95354998e-01 -9.11069989e-01 -5.28988481e-01 6.34556413e-01
4.40234423e-01 -1.10525858e+00 9.43912625e-01 2.17352495e-01
4.53477323e-01 -6.38507485e-01 4.31867391e-01 -1.13473368e+00
-4.33575809e-01 -3.29420835e-01 2.16690391e-01 1.00458837e+00
4.21075881e-01 -7.50117540e-01 9.60474908e-01 4.96172369e-01
-2.52483010e-01 -1.03433299e+00 -6.67542219e-01 -3.78210753e-01
-4.00079899e-02 2.29419529e-01 6.13415599e-01 5.97767174e-01
-1.54203609e-01 -2.23597020e-01 8.85583926e-03 4.71080542e-01
4.39576745e-01 3.21249783e-01 4.16480720e-01 -9.38724101e-01
-3.40477347e-01 -5.94334245e-01 -1.31371543e-01 -1.32278967e+00
-5.57522714e-01 -3.28859180e-01 2.74729311e-01 -1.63348448e+00
3.73600096e-01 -7.67192781e-01 -3.38364094e-01 3.14256936e-01
-8.80198002e-01 4.10320401e-01 -1.07585192e-01 3.15904580e-02
-5.09123266e-01 2.88407952e-01 1.80764925e+00 -3.13214511e-01
-2.38732517e-01 2.45683163e-01 -1.01500404e+00 9.29628789e-01
7.00056851e-01 -3.40669900e-01 -1.61841393e-01 -5.36713243e-01
2.50440449e-01 -9.07339305e-02 3.58233511e-01 -1.07950962e+00
1.66643903e-01 -5.91545440e-02 7.93208480e-01 -5.73770285e-01
-1.05441660e-01 -8.47462058e-01 -2.51541346e-01 1.08296013e+00
1.75568968e-01 -4.84024048e-01 3.33217800e-01 5.99094570e-01
-2.75196642e-01 -3.01014721e-01 1.07437968e+00 -3.14345419e-01
-3.57009470e-01 3.87152225e-01 -3.36834222e-01 -5.29306456e-02
9.75318789e-01 -3.84136617e-01 -3.40298921e-01 9.29069966e-02
-3.12730700e-01 5.62899053e-01 6.75193965e-01 9.77585763e-02
4.63698864e-01 -1.05210710e+00 -5.08063316e-01 5.02372921e-01
4.23821434e-03 6.37610137e-01 6.91518307e-01 1.23376548e+00
-9.47530270e-01 9.73604321e-02 1.39009550e-01 -6.39480293e-01
-1.02380979e+00 4.48005229e-01 5.72062850e-01 -3.83943230e-01
-7.62036622e-01 1.06067002e+00 5.77091694e-01 -1.12576716e-01
2.28218004e-01 -7.73295999e-01 -1.94297999e-01 -2.59289831e-01
6.97354436e-01 1.96009696e-01 -2.40601450e-02 -1.99133918e-01
-3.17206264e-01 5.81417739e-01 -3.30305248e-01 5.77798784e-01
9.89844918e-01 -1.77995935e-01 2.47328617e-02 -9.42901522e-02
8.09215009e-01 3.85899127e-01 -1.52339756e+00 3.82399820e-02
-6.65892005e-01 -6.64828479e-01 2.48215586e-01 -7.84821212e-01
-1.56103337e+00 9.70756114e-01 7.64189184e-01 2.53458768e-01
1.22151518e+00 -4.69712526e-01 1.21987200e+00 -2.25625530e-01
-2.15590239e-01 -8.57265353e-01 9.82267857e-02 5.13770580e-02
3.11308920e-01 -1.31104422e+00 1.92172036e-01 -8.31722379e-01
-5.60159564e-01 1.08715713e+00 7.79939830e-01 -2.06979513e-01
2.71024823e-01 3.70725602e-01 2.08439842e-01 7.77685195e-02
-5.09034209e-02 -3.88745479e-02 1.91426262e-01 3.79329443e-01
2.95518875e-01 -9.68433470e-02 -3.18202734e-01 6.99363470e-01
2.66398400e-01 2.05850396e-02 5.08739412e-01 6.96471512e-01
-4.50915694e-01 -8.24004471e-01 -3.89701039e-01 5.16574204e-01
-7.30548024e-01 -1.79788843e-01 3.03759351e-02 6.79599881e-01
3.94495338e-01 8.27300668e-01 -3.50497290e-02 8.29991028e-02
2.01458946e-01 -7.55129814e-01 2.48741090e-01 -4.80023444e-01
-5.20328939e-01 5.48875153e-01 -2.70997107e-01 -3.99535328e-01
-1.92883074e-01 -3.89947534e-01 -1.37098944e+00 8.06940719e-02
-6.71772480e-01 -2.75619468e-03 2.73125976e-01 6.94967330e-01
1.31120577e-01 1.06225538e+00 5.38281262e-01 -4.60673660e-01
-3.03637892e-01 -6.43612564e-01 -4.73552227e-01 2.95532435e-01
4.19396639e-01 -3.39943796e-01 -6.00416899e-01 -1.11776516e-01] | [14.640573501586914, -2.74558687210083] |
9ce41e91-b21c-4b51-82f7-c35460281857 | tensorizing-flows-a-tool-for-variational | 2305.02460 | null | https://arxiv.org/abs/2305.02460v1 | https://arxiv.org/pdf/2305.02460v1.pdf | Tensorizing flows: a tool for variational inference | Fueled by the expressive power of deep neural networks, normalizing flows have achieved spectacular success in generative modeling, or learning to draw new samples from a distribution given a finite dataset of training samples. Normalizing flows have also been applied successfully to variational inference, wherein one attempts to learn a sampler based on an expression for the log-likelihood or energy function of the distribution, rather than on data. In variational inference, the unimodality of the reference Gaussian distribution used within the normalizing flow can cause difficulties in learning multimodal distributions. We introduce an extension of normalizing flows in which the Gaussian reference is replaced with a reference distribution that is constructed via a tensor network, specifically a matrix product state or tensor train. We show that by combining flows with tensor networks on difficult variational inference tasks, we can improve on the results obtained by using either tool without the other. | ['Hongli Zhao', 'Michael Lindsey', 'Yuehaw Khoo'] | 2023-05-03 | null | null | null | null | ['tensor-networks'] | ['methodology'] | [ 9.90449339e-02 1.88938141e-01 -7.79008865e-02 -3.33469272e-01
-5.51403821e-01 -7.71922171e-01 1.08864164e+00 -3.53563040e-01
-2.25701377e-01 9.70923781e-01 4.01837319e-01 -3.57468575e-01
-4.06837277e-02 -9.86268878e-01 -7.51777589e-01 -9.51616585e-01
1.84825450e-01 8.80152166e-01 -2.24423677e-01 4.69296686e-02
5.40974513e-02 5.32338440e-01 -1.23063743e+00 -1.04863457e-01
6.22779489e-01 6.91396177e-01 -1.64780304e-01 8.89547467e-01
-4.08857554e-01 8.14391136e-01 -5.74168622e-01 -6.07647896e-01
1.21406332e-01 -6.44457459e-01 -8.95449162e-01 5.61470203e-02
5.36465287e-01 -5.35513937e-01 -2.88695931e-01 1.16394746e+00
8.44259337e-02 4.22434062e-01 1.16392183e+00 -1.37725830e+00
-6.98917627e-01 7.55942166e-01 -9.79593173e-02 2.76157800e-02
-1.99403185e-02 -4.38209623e-03 1.19060397e+00 -6.41286314e-01
7.68698752e-01 1.41563046e+00 5.22469223e-01 6.09206617e-01
-2.07171750e+00 -2.30531290e-01 -1.54037893e-01 -1.75032407e-01
-1.14304805e+00 -3.87428820e-01 7.54656434e-01 -6.86719537e-01
6.56392515e-01 6.64766803e-02 6.40619516e-01 1.39763451e+00
-2.90477499e-02 7.63803184e-01 6.71308100e-01 -1.54169694e-01
4.34529901e-01 1.26464948e-01 -2.26093337e-01 5.43324411e-01
2.69787526e-03 4.15288284e-02 -2.60557830e-01 -3.00222188e-01
9.51048017e-01 -1.62063967e-02 -1.71060920e-01 -7.24875987e-01
-1.01990521e+00 1.32536554e+00 3.30452234e-01 2.28143737e-01
-2.09839061e-01 6.10055685e-01 4.64516848e-01 1.29747540e-02
5.72842240e-01 3.46145689e-01 -8.61187205e-02 -3.69123280e-01
-1.24875617e+00 5.99607527e-01 1.13051903e+00 6.42593086e-01
9.07976210e-01 3.75008523e-01 -3.23998809e-01 5.43118894e-01
5.09762228e-01 6.19042873e-01 8.36717784e-02 -1.42437196e+00
1.76029444e-01 1.34591922e-01 1.66462630e-01 -5.36268115e-01
2.12012008e-02 -2.59193629e-01 -6.98879123e-01 9.92579386e-02
1.04674554e+00 -4.30392861e-01 -9.45824921e-01 2.18856072e+00
2.74146050e-01 -1.42132863e-01 3.73022966e-02 5.73489904e-01
1.06860690e-01 7.25263357e-01 5.09432293e-02 8.13843384e-02
8.33252549e-01 -4.32541251e-01 -6.71632707e-01 2.78990626e-01
4.04299885e-01 -5.91619432e-01 8.52960110e-01 5.04161417e-01
-1.09267139e+00 -2.41610065e-01 -7.96441734e-01 -3.42972010e-01
-3.99388343e-01 -1.49566337e-01 5.71121573e-01 7.56277740e-01
-1.10743964e+00 9.19080794e-01 -1.19224846e+00 -2.52344042e-01
5.68418801e-01 1.53431267e-01 -4.18764837e-02 9.48590115e-02
-8.89385223e-01 8.65817487e-01 2.53060579e-01 5.72990328e-02
-1.18115795e+00 -9.47573304e-01 -7.70023346e-01 1.70657292e-01
1.22319475e-01 -9.62694466e-01 1.28224254e+00 -8.83883297e-01
-1.67579257e+00 3.18751752e-01 -7.80301243e-02 -3.17329675e-01
7.60661840e-01 1.85252652e-02 9.17164534e-02 7.98930451e-02
-2.02034526e-02 7.59800017e-01 1.07913196e+00 -9.87282932e-01
-1.85751721e-01 -2.65491694e-01 2.39646003e-01 -2.11854085e-01
-6.28284514e-02 -5.33473790e-01 1.10513180e-01 -2.18820035e-01
-2.78379619e-01 -7.31874526e-01 -2.58862395e-02 4.48517762e-02
-7.94090092e-01 -4.85667646e-01 8.18095624e-01 -4.52656239e-01
9.06466901e-01 -1.90101123e+00 5.37053823e-01 4.45806980e-01
2.68347353e-01 4.34494801e-02 -1.37516228e-03 4.93840784e-01
7.66240358e-02 3.11298162e-01 -4.61091697e-01 -6.09631538e-01
5.29035151e-01 6.63566887e-01 -5.70384681e-01 4.55406636e-01
4.14291352e-01 8.05252075e-01 -1.04835773e+00 -3.50205988e-01
1.80566743e-01 9.69772875e-01 -8.45488250e-01 1.39258474e-01
-5.11372566e-01 4.83659953e-01 -1.87687784e-01 1.77124552e-02
5.79120040e-01 -2.97376990e-01 1.70247182e-01 -2.01311201e-01
4.94909994e-02 4.14847702e-01 -1.14082599e+00 1.75088966e+00
-5.01744270e-01 8.11801732e-01 9.90156550e-04 -1.09023571e+00
6.25966966e-01 2.72224724e-01 3.49580020e-01 6.45165890e-02
2.50463039e-01 3.44001874e-02 4.79100598e-03 -3.59856963e-01
5.86753428e-01 -7.64330506e-01 2.38622352e-01 6.16319835e-01
5.43345511e-01 -4.70116466e-01 6.37882411e-01 4.63745564e-01
6.44774139e-01 6.22440755e-01 -2.40782931e-01 -3.33935797e-01
2.73719519e-01 -1.55560270e-01 1.72426239e-01 7.57948220e-01
2.92218685e-01 6.94438159e-01 1.17022300e+00 -1.39027268e-01
-1.60121453e+00 -1.72833097e+00 -5.46866596e-01 8.31294119e-01
-6.71517849e-01 -4.65265661e-01 -7.93101192e-01 -5.25901794e-01
1.59873322e-01 1.09172416e+00 -8.26612055e-01 -1.31568328e-01
-3.68988693e-01 -8.11881602e-01 7.35581040e-01 7.18522251e-01
1.87577888e-01 -5.64386606e-01 -2.37876981e-01 2.11680919e-01
-5.95296919e-02 -8.47656429e-01 -5.27378261e-01 1.88830927e-01
-9.24758494e-01 -7.36764610e-01 -7.89306700e-01 -2.10782856e-01
6.46928608e-01 -5.34845471e-01 1.16317761e+00 -3.74188632e-01
-1.93617232e-02 5.86565256e-01 2.65696675e-01 -1.07065700e-01
-6.84603274e-01 3.13047528e-01 -9.86306369e-02 9.67190564e-02
1.18074536e-01 -8.69308293e-01 -1.89709932e-01 -1.74689487e-01
-1.31836414e+00 -1.09373100e-01 5.82207255e-02 9.27169025e-01
1.17580526e-01 -3.22894543e-01 3.65332842e-01 -9.67870474e-01
7.27630675e-01 -6.47459865e-01 -6.88041747e-01 -5.31312935e-02
-4.21619862e-01 6.69795811e-01 6.15996003e-01 -5.76449931e-01
-1.07201803e+00 -1.56543285e-01 -1.53801605e-01 -6.89251006e-01
6.00694828e-02 5.80987632e-01 4.02749032e-02 4.35119748e-01
5.88385165e-01 9.96421129e-02 4.13005203e-01 -4.49260592e-01
9.90136862e-01 2.40364298e-01 7.17535257e-01 -9.74835038e-01
8.06779921e-01 5.46589136e-01 3.85748893e-01 -8.21491659e-01
-8.13790441e-01 1.79648697e-01 -6.71587765e-01 -2.02991068e-02
1.08179522e+00 -4.75476950e-01 -9.84170318e-01 2.23893017e-01
-1.25507212e+00 -2.95842737e-01 -7.81235635e-01 5.22579908e-01
-6.07994020e-01 1.71255454e-01 -5.70662320e-01 -1.04095232e+00
1.31011799e-01 -1.06622839e+00 9.43890035e-01 2.25813001e-01
-3.17756683e-01 -1.58395565e+00 3.88671160e-01 -5.98359331e-02
7.53286302e-01 1.98263168e-01 1.15076113e+00 -7.05779076e-01
-6.61833823e-01 -1.34219632e-01 -8.83984342e-02 7.04547107e-01
4.49159294e-02 5.81253827e-01 -1.01138437e+00 1.38634220e-02
-1.80505991e-01 -2.68390715e-01 8.40700805e-01 5.38272619e-01
8.49250972e-01 -3.27874601e-01 -9.69748795e-02 7.10997462e-01
1.29456162e+00 -1.70129716e-01 5.75405777e-01 -2.62050629e-01
8.66752565e-01 3.71958703e-01 -5.98559320e-01 3.96046966e-01
2.46041849e-01 3.15227300e-01 2.78865367e-01 3.64005148e-01
1.46435350e-01 -5.24095356e-01 3.08024526e-01 4.69816089e-01
-2.03521281e-01 -9.42443237e-02 -6.71443224e-01 5.04346728e-01
-1.58014381e+00 -1.18411827e+00 1.82468876e-01 2.17409658e+00
9.90543425e-01 1.32138193e-01 3.25792104e-01 -1.78511947e-01
3.48533750e-01 1.74258038e-01 -7.16976523e-01 -3.86506140e-01
-8.86820182e-02 4.30723757e-01 2.32146412e-01 9.80410933e-01
-8.78074229e-01 5.11060596e-01 7.31449699e+00 7.31582701e-01
-1.22310865e+00 2.08595879e-02 2.86086261e-01 -2.17723906e-01
-8.39232504e-01 1.66248500e-01 -6.59193993e-01 4.56945330e-01
9.78616595e-01 -2.27564171e-01 7.72183180e-01 5.54679871e-01
-9.17306170e-02 -1.56457171e-01 -1.56141853e+00 6.43551230e-01
-1.57645330e-01 -1.43842280e+00 2.77853072e-01 3.70598286e-01
6.46769047e-01 -1.02314472e-01 3.93365115e-01 3.37578446e-01
7.28470445e-01 -1.23193800e+00 6.02121770e-01 8.35567415e-01
4.93405908e-01 -7.19605625e-01 4.47972149e-01 3.15319717e-01
-7.16118217e-01 3.22862625e-01 -3.45905691e-01 -2.32720245e-02
2.63916194e-01 5.39985299e-01 -9.66515124e-01 2.57048130e-01
1.05532214e-01 6.56718969e-01 -1.11641698e-01 6.84485793e-01
-2.63694942e-01 8.38153899e-01 -6.00350082e-01 -1.54152676e-01
2.52953708e-01 -9.07333612e-01 8.06293905e-01 1.02824283e+00
1.91545755e-01 -5.06190836e-01 -1.75381646e-01 1.86928999e+00
-8.65692720e-02 -2.28402853e-01 -6.30388319e-01 -5.42263806e-01
8.42057616e-02 1.33826709e+00 -4.52777028e-01 -4.75197822e-01
-1.34967312e-01 5.23677886e-01 3.41845065e-01 7.66957819e-01
-8.02945614e-01 -3.33539665e-01 8.61731946e-01 -1.31647483e-01
5.80051363e-01 -3.99003059e-01 -1.83766291e-01 -1.32522821e+00
-2.51550674e-01 -3.31605554e-01 6.59011258e-03 -8.37165117e-01
-1.49388194e+00 3.87437522e-01 4.21247184e-01 -5.85202098e-01
-9.09605861e-01 -6.49502039e-01 -6.48979902e-01 1.23745990e+00
-9.83454049e-01 -8.74121726e-01 3.05605650e-01 4.73735541e-01
-9.86820459e-02 1.04521923e-02 7.46377289e-01 2.33821824e-01
-5.73933959e-01 1.90752581e-01 3.18033665e-01 2.40516290e-01
2.57667929e-01 -1.75830841e+00 6.97727948e-02 6.40470088e-01
2.76279449e-01 8.86771262e-01 1.07277179e+00 -4.16107297e-01
-1.36094451e+00 -6.52053416e-01 7.40104258e-01 -7.17495561e-01
9.40635085e-01 -5.95035911e-01 -9.15062487e-01 1.11590004e+00
3.22580159e-01 -3.00805783e-03 4.80182827e-01 2.09107161e-01
-4.41003829e-01 -7.00780749e-02 -1.09585273e+00 4.82061863e-01
6.48537934e-01 -7.54137099e-01 -3.20957065e-01 1.84459314e-01
4.93563354e-01 -3.77408952e-01 -1.03351641e+00 -2.10238650e-01
6.56478226e-01 -8.29365551e-01 8.32445085e-01 -1.14077032e+00
6.36216164e-01 -2.27216706e-01 -3.92176956e-01 -1.59919012e+00
-9.94904861e-02 -6.66418374e-01 -1.79314494e-01 1.35958254e+00
4.32224900e-01 -6.01951599e-01 7.70482719e-01 8.53269696e-01
1.14800617e-01 -5.08983672e-01 -9.19153273e-01 -5.62554181e-01
6.73782170e-01 -4.38168019e-01 6.87886655e-01 5.60366035e-01
-3.23318660e-01 5.59887648e-01 -2.92507201e-01 -1.87523022e-01
8.09921980e-01 -1.68702543e-01 8.29295933e-01 -1.29610109e+00
-4.54791933e-01 -6.70105338e-01 -1.58743620e-01 -1.22170794e+00
3.59253228e-01 -1.17521787e+00 -7.02706873e-02 -1.31430626e+00
1.66479647e-01 -8.95348564e-02 2.26981744e-01 2.43748650e-01
1.74693510e-01 8.86674002e-02 2.44208515e-01 -4.63675857e-02
-8.02884102e-02 7.82911897e-01 1.41005266e+00 -1.97288781e-01
-1.36876144e-02 -4.53464538e-02 -5.32586992e-01 7.34892547e-01
4.29754645e-01 -4.18778896e-01 -5.78598678e-01 -2.15481818e-01
5.60470521e-01 1.23831913e-01 5.20131171e-01 -5.58277130e-01
1.07393600e-01 -1.35009373e-02 5.53046405e-01 -3.70635003e-01
5.13855934e-01 -5.22975147e-01 4.06406790e-01 7.93295205e-02
-4.24639583e-01 -2.58759409e-01 -7.15118572e-02 4.12390649e-01
1.26381582e-02 -4.28406388e-01 6.48097456e-01 -6.36449130e-03
1.43520199e-02 1.32653967e-01 -5.06342709e-01 4.10943866e-01
3.26997459e-01 5.00486046e-02 -2.97778904e-01 -5.50102174e-01
-1.06743562e+00 -1.20223127e-02 3.36360812e-01 -5.66256978e-02
2.94312000e-01 -1.48741519e+00 -6.54492140e-01 2.89753646e-01
-3.55392069e-01 3.39330174e-02 9.63878855e-02 1.00046897e+00
-3.95326316e-01 1.63986653e-01 -1.76751032e-01 -8.02195013e-01
-4.48337346e-01 3.30957204e-01 7.34488189e-01 -1.51622191e-01
-3.40362847e-01 5.99696040e-01 4.48420458e-02 -5.30316114e-01
2.95104310e-02 -4.96596664e-01 2.94092923e-01 2.80516714e-01
2.29727328e-01 5.08746564e-01 -3.19345683e-01 -4.78889823e-01
-5.50442077e-02 2.15327442e-01 1.02269374e-01 -6.51851058e-01
1.25962317e+00 1.34914555e-03 -2.06662819e-01 9.30796623e-01
1.51006973e+00 -1.34902596e-01 -1.67230690e+00 -1.12084918e-01
-4.50188845e-01 -2.60417104e-01 -3.51353064e-02 -5.30063987e-01
-1.22876894e+00 1.12944293e+00 1.81552649e-01 6.40629530e-01
4.33698684e-01 7.17189088e-02 4.81120527e-01 2.32323840e-01
-1.00047663e-01 -7.95011997e-01 -3.77421058e-03 5.36988974e-01
6.69406354e-01 -1.05442369e+00 -9.30194184e-02 2.31560484e-01
-4.07526761e-01 1.39635515e+00 5.39538264e-02 -3.56064379e-01
9.06367838e-01 2.15907916e-01 -1.71564728e-01 -6.72710314e-02
-5.78488886e-01 -1.19044131e-03 4.19253498e-01 6.61282003e-01
5.12928367e-01 1.22236848e-01 1.49195865e-01 4.78562973e-02
-5.06149411e-01 -5.35917319e-02 6.35198534e-01 4.77515459e-01
-8.46993104e-02 -1.18389750e+00 -2.52897859e-01 5.17971635e-01
-5.23476064e-01 -2.03233942e-01 -4.18282226e-02 7.93132246e-01
-5.20471521e-02 6.10332847e-01 4.92626727e-01 1.25885836e-03
-1.24538511e-01 5.94573200e-01 8.23333144e-01 -2.39715666e-01
-3.58729511e-02 1.51583433e-01 -7.30893910e-02 -2.78851509e-01
-4.67172235e-01 -9.97807860e-01 -1.09036672e+00 -6.33681774e-01
5.68988174e-03 3.87389690e-01 7.83162773e-01 1.11744905e+00
-1.08304068e-01 4.34852153e-01 3.11811477e-01 -1.03776360e+00
-9.16906774e-01 -1.23928797e+00 -8.85185421e-01 3.34897876e-01
6.52082920e-01 -7.04822600e-01 -6.57014370e-01 2.27320388e-01] | [6.966035842895508, 3.9199397563934326] |
2d508261-8c1b-4970-af66-281b1f46d960 | stable-learning-via-sparse-variable | 2212.00992 | null | https://arxiv.org/abs/2212.00992v1 | https://arxiv.org/pdf/2212.00992v1.pdf | Stable Learning via Sparse Variable Independence | The problem of covariate-shift generalization has attracted intensive research attention. Previous stable learning algorithms employ sample reweighting schemes to decorrelate the covariates when there is no explicit domain information about training data. However, with finite samples, it is difficult to achieve the desirable weights that ensure perfect independence to get rid of the unstable variables. Besides, decorrelating within stable variables may bring about high variance of learned models because of the over-reduced effective sample size. A tremendous sample size is required for these algorithms to work. In this paper, with theoretical justification, we propose SVI (Sparse Variable Independence) for the covariate-shift generalization problem. We introduce sparsity constraint to compensate for the imperfectness of sample reweighting under the finite-sample setting in previous methods. Furthermore, we organically combine independence-based sample reweighting and sparsity-based variable selection in an iterative way to avoid decorrelating within stable variables, increasing the effective sample size to alleviate variance inflation. Experiments on both synthetic and real-world datasets demonstrate the improvement of covariate-shift generalization performance brought by SVI. | ['Xingxuan Zhang', 'Renzhe Xu', 'Yong Lin', 'Zheyan Shen', 'Yue He', 'Peng Cui', 'Han Yu'] | 2022-12-02 | null | null | null | null | ['variable-selection'] | ['methodology'] | [ 4.64451611e-01 -2.01121822e-01 -4.39171970e-01 -6.83001339e-01
-3.17881912e-01 -2.13830575e-01 7.16686323e-02 -4.48656641e-02
-4.06969994e-01 1.06733191e+00 2.24570587e-01 -2.75234073e-01
-3.53415549e-01 -6.80530369e-01 -5.47029495e-01 -9.87484336e-01
4.19359803e-02 5.57156950e-02 -1.81911826e-01 1.16302386e-01
2.05910623e-01 9.04206485e-02 -1.47628939e+00 -2.73198396e-01
1.42719972e+00 7.43646264e-01 4.19623315e-01 -8.91756564e-02
-2.22879320e-01 1.38498425e-01 -5.47581732e-01 1.24167232e-02
5.70432365e-01 -5.03631473e-01 -2.66467452e-01 3.62738252e-01
4.37365890e-01 -3.47672738e-02 -1.14689074e-01 1.21645415e+00
3.86913925e-01 3.62361997e-01 6.09616637e-01 -1.48184741e+00
-8.47974837e-01 8.25859010e-01 -1.15559685e+00 2.88209349e-01
-1.96103260e-01 -1.69897199e-01 9.55557168e-01 -9.14779782e-01
2.95774162e-01 1.21734285e+00 6.50981784e-01 4.77357447e-01
-1.49566019e+00 -1.17100024e+00 4.60856825e-01 7.97213241e-02
-1.51020646e+00 -2.99274325e-01 1.02707720e+00 -4.87202972e-01
2.83050478e-01 3.65182877e-01 3.74622375e-01 1.00276327e+00
-1.73594281e-02 7.35881567e-01 8.90846372e-01 -2.88963497e-01
4.00754750e-01 4.70926166e-01 4.18264091e-01 1.90125242e-01
8.01493108e-01 1.27345800e-01 -6.54493809e-01 -3.22455645e-01
1.04427481e+00 3.66071671e-01 -4.49696958e-01 -8.94100428e-01
-1.00789869e+00 1.04639661e+00 4.62285995e-01 1.20326273e-01
-3.05454731e-01 -3.51713568e-01 4.25412655e-01 5.57507575e-01
5.33596873e-01 4.97448534e-01 -6.64235234e-01 2.74127603e-01
-1.02782309e+00 1.40695333e-01 1.51507780e-01 1.24848378e+00
7.68913627e-01 4.75143850e-01 -2.42418081e-01 1.02662945e+00
1.13866806e-01 5.39166510e-01 8.38293076e-01 -6.64510071e-01
7.64703274e-01 6.51883245e-01 9.48913544e-02 -1.08808744e+00
-3.47702593e-01 -9.88088429e-01 -1.42726266e+00 1.51548132e-01
6.32220626e-01 -4.09290045e-01 -7.21240759e-01 2.02161574e+00
3.93263876e-01 3.55899245e-01 1.14807738e-02 9.20376301e-01
3.02503616e-01 2.96906888e-01 1.35537535e-01 -6.29955471e-01
1.03405523e+00 -6.82191849e-01 -6.57279432e-01 -3.79560173e-01
6.09149277e-01 -3.72746676e-01 1.23740768e+00 3.72059435e-01
-6.76683009e-01 -6.84204042e-01 -1.20376956e+00 2.95279592e-01
1.60882637e-01 3.90892774e-02 8.10602725e-01 6.97990119e-01
-3.92793268e-01 5.20100355e-01 -6.95630908e-01 1.70167983e-01
5.79340339e-01 4.67314094e-01 -4.88017082e-01 -9.45250392e-02
-1.15747166e+00 2.49889284e-01 3.15530777e-01 -1.27277654e-02
-1.94125772e-01 -1.10672331e+00 -9.99748290e-01 2.79744327e-01
4.55252767e-01 -7.70238400e-01 7.34885335e-01 -1.16950095e+00
-9.14769650e-01 2.48287603e-01 -4.45015877e-01 -4.67864245e-01
5.17657876e-01 -1.82682052e-01 -2.27698758e-01 -4.48593199e-01
1.20351583e-01 2.48432919e-01 1.30054867e+00 -1.15045369e+00
-5.95248282e-01 -5.74083447e-01 -4.58466917e-01 1.36827260e-01
-8.24371219e-01 -3.15114975e-01 -7.30655715e-02 -1.02917147e+00
5.07857680e-01 -8.05065334e-01 -6.20102346e-01 -6.05961420e-02
-1.13113008e-01 -5.50550632e-02 7.26974010e-01 -3.70701998e-01
1.66584575e+00 -2.49080634e+00 1.73691899e-01 3.62170428e-01
3.77364963e-01 4.88456786e-02 -2.09952667e-01 -1.23675577e-01
-5.45212626e-01 -1.77431390e-01 -4.97975171e-01 -3.00798416e-01
-3.73130172e-01 2.67995834e-01 -4.08866495e-01 5.83608627e-01
4.62481603e-02 3.82688016e-01 -9.03856516e-01 -3.99004817e-01
5.42882718e-02 1.65484771e-01 -8.29735637e-01 -1.34879565e-02
2.46830106e-01 5.65424979e-01 -6.14671648e-01 4.72917587e-01
1.03904104e+00 -3.06013077e-01 1.36251092e-01 -7.13919178e-02
9.08223838e-02 -1.30436063e-01 -1.70154822e+00 1.45023620e+00
-3.77948344e-01 2.74269581e-01 1.05771653e-01 -1.38117337e+00
1.16318035e+00 7.62797594e-02 5.53761244e-01 -4.74275559e-01
-1.03567960e-02 2.44071782e-01 7.26384297e-02 -2.08007187e-01
3.87239218e-01 -4.31866944e-01 -5.82257099e-02 1.72486544e-01
-2.87993908e-01 1.78551137e-01 -1.99492443e-02 2.18249839e-02
5.99048853e-01 -2.11472303e-01 5.49147427e-01 -5.54800928e-01
7.10821688e-01 -3.77791196e-01 1.29380643e+00 7.39938855e-01
-1.70549378e-01 6.80133104e-01 2.84574062e-01 -3.04545879e-01
-9.60324526e-01 -7.71972120e-01 -4.79997277e-01 1.03174710e+00
8.90771821e-02 -2.34416276e-01 -2.95986325e-01 -6.17009342e-01
4.47846770e-01 6.43493593e-01 -7.55031645e-01 -3.75147641e-01
-3.90965104e-01 -9.94181216e-01 -1.69257700e-01 7.35457361e-01
3.26754361e-01 -6.41960502e-01 -2.78536022e-01 2.88119197e-01
1.40037462e-01 -3.00168872e-01 -7.06614912e-01 4.36487824e-01
-1.04180360e+00 -7.62151659e-01 -9.78325605e-01 -7.61452198e-01
1.04192865e+00 6.92414939e-01 7.32540905e-01 5.71314804e-03
-6.10278882e-02 -2.87838191e-01 -2.35977650e-01 -4.59206551e-01
7.49861673e-02 1.52217299e-01 3.34026694e-01 5.46453930e-02
3.69104952e-01 -6.99721158e-01 -4.74889785e-01 4.64527190e-01
-8.39433372e-01 -1.78626701e-01 4.27408844e-01 1.26494765e+00
3.90825838e-01 2.96307862e-01 9.02471185e-01 -1.00606680e+00
4.25681412e-01 -7.07716465e-01 -6.06523633e-01 1.58270493e-01
-7.62112141e-01 3.36250275e-01 6.53537571e-01 -9.88325357e-01
-1.04825890e+00 5.35481796e-02 3.77385885e-01 -5.30282915e-01
3.11293095e-01 6.44470632e-01 -3.55092317e-01 1.01368092e-01
7.68305421e-01 1.29543066e-01 1.58883661e-01 -6.04115069e-01
6.94058016e-02 7.11171627e-01 1.28698692e-01 -5.18615484e-01
1.03670049e+00 3.33823234e-01 7.99639598e-02 -5.54757774e-01
-9.88112152e-01 -3.97717357e-01 -5.09402931e-01 3.75156671e-01
1.83065027e-01 -1.12952673e+00 -2.83665746e-01 3.11075538e-01
-4.25836682e-01 -1.20090604e-01 -4.14006978e-01 7.45521009e-01
-2.36798674e-01 4.94621933e-01 5.59919253e-02 -7.65752017e-01
-3.59816134e-01 -9.20634270e-01 5.58504462e-01 2.58868515e-01
-4.03494328e-01 -9.19180989e-01 -1.30427599e-01 1.09824948e-01
4.85508710e-01 -2.56102975e-03 8.20083976e-01 -6.39205873e-01
-1.47904545e-01 -2.11368594e-02 -1.86617419e-01 3.63273382e-01
5.43463409e-01 -2.90521622e-01 -5.66960096e-01 -6.64919734e-01
2.92713195e-01 -2.68028583e-02 8.63713145e-01 6.50080085e-01
1.38864255e+00 -2.73516536e-01 -4.29812402e-01 6.93457782e-01
1.21304500e+00 1.99458271e-01 2.05770776e-01 2.95515597e-01
6.76040709e-01 4.95874941e-01 8.33005667e-01 8.59967649e-01
-5.05685573e-03 4.45380926e-01 -3.66014987e-02 -1.78683266e-01
1.32742062e-01 -2.41630271e-01 -9.00055021e-02 6.08264506e-01
3.22535396e-01 8.70424360e-02 -5.85759342e-01 6.63913846e-01
-1.85166359e+00 -8.23916137e-01 -2.54921824e-01 2.58266735e+00
1.14136124e+00 1.17177457e-01 8.72920603e-02 3.28629583e-01
8.36025953e-01 2.59499177e-02 -8.90208125e-01 1.18997008e-01
-2.17809647e-01 -1.52492225e-01 7.02086449e-01 4.47849244e-01
-1.02218688e+00 5.63439369e-01 5.72611809e+00 9.81639206e-01
-1.15568554e+00 -9.66963470e-02 6.16343498e-01 -1.93125695e-01
-6.19953394e-01 2.73600593e-02 -1.01956689e+00 6.74487293e-01
4.71738458e-01 -5.54714262e-01 2.46351972e-01 1.08336437e+00
1.80505186e-01 -1.54302688e-02 -9.18189347e-01 1.13800085e+00
-4.93885688e-02 -8.84219050e-01 -1.52551442e-01 -6.49824589e-02
8.34185243e-01 -3.52233797e-01 8.54560286e-02 4.40531015e-01
1.77296087e-01 -7.30177402e-01 4.45250899e-01 2.57821172e-01
6.23275220e-01 -7.00897634e-01 5.32849729e-01 4.18835670e-01
-1.04751074e+00 -3.99444997e-01 -8.12464535e-01 -2.37420008e-01
2.82547586e-02 1.02686143e+00 -6.49278462e-01 4.18264329e-01
5.29112995e-01 8.75865698e-01 -5.18253744e-01 1.12688804e+00
-2.06498280e-02 7.47728050e-01 -3.79247665e-01 2.22707689e-01
-1.21079385e-01 -4.91258115e-01 4.99851316e-01 6.54294968e-01
5.13531208e-01 -6.77010641e-02 2.55378425e-01 5.47930956e-01
1.61996663e-01 1.85512453e-01 -4.17168498e-01 3.39951217e-01
7.72041559e-01 8.32305610e-01 -3.63497198e-01 -4.01936054e-01
-5.41155636e-01 5.86144090e-01 2.53902078e-01 5.07776856e-01
-7.04066098e-01 -3.85852486e-01 7.53230929e-01 1.78125173e-01
4.13667768e-01 -2.49657154e-01 -8.88734818e-01 -1.38819432e+00
6.77860901e-02 -8.69928002e-01 5.07109880e-01 -2.45923519e-01
-1.53794587e+00 2.25429446e-01 1.41864149e-02 -1.38039136e+00
5.46298921e-02 -1.89057644e-02 -4.67528939e-01 1.11068738e+00
-1.34026814e+00 -5.14868557e-01 -1.58953667e-01 6.42828822e-01
6.59361601e-01 -2.31255144e-01 4.62292492e-01 3.70473683e-01
-8.14257920e-01 1.06186688e+00 4.63229865e-01 -2.69681275e-01
1.08004749e+00 -1.08077574e+00 4.38177809e-02 7.30309010e-01
-1.92206264e-01 1.21291447e+00 9.14101601e-01 -8.49033058e-01
-9.89244282e-01 -1.09477031e+00 6.80567563e-01 -5.31087704e-02
6.28390908e-01 -4.11219388e-01 -1.46223831e+00 6.42662227e-01
-3.04553211e-01 3.86389755e-02 9.72679615e-01 5.00391245e-01
-4.40717161e-01 -4.32053715e-01 -1.25469887e+00 6.18028760e-01
1.07292533e+00 6.28936887e-02 -6.08001232e-01 1.29847392e-01
6.78942382e-01 -5.64341396e-02 -8.49841654e-01 5.81826091e-01
4.76758301e-01 -5.32062292e-01 9.95618880e-01 -6.69098377e-01
1.04494274e-01 -4.35329497e-01 -5.44652455e-02 -1.36671114e+00
-7.63281822e-01 -4.34773654e-01 4.75923605e-02 1.35318720e+00
4.37097907e-01 -8.21108520e-01 9.08388913e-01 1.01851845e+00
7.37099946e-02 -4.48760927e-01 -9.95137513e-01 -9.89510357e-01
1.42595083e-01 -1.26975819e-01 8.32361281e-01 1.17309976e+00
5.21251075e-02 5.05191386e-01 -7.53714740e-01 8.04443732e-02
8.27219188e-01 3.20848495e-01 8.66878092e-01 -1.47325504e+00
-4.29214329e-01 -1.60622910e-01 -2.06465885e-01 -1.07944691e+00
8.88449177e-02 -6.59078360e-01 -1.77691609e-01 -9.41884458e-01
3.97809535e-01 -8.14282477e-01 -5.51291168e-01 6.28973663e-01
-7.76804566e-01 -3.12249921e-02 6.42536283e-02 3.76987040e-01
-1.32790223e-01 8.37631941e-01 1.39483583e+00 -2.78785545e-02
-5.51094651e-01 3.39507222e-01 -1.24928057e+00 4.54320282e-01
8.26042235e-01 -5.49496949e-01 -8.39662015e-01 -3.06807637e-01
-1.15182949e-02 7.58662671e-02 -6.86653852e-02 -8.33127260e-01
-3.16620688e-03 -5.53405404e-01 6.41964853e-01 -4.52057928e-01
-6.63041919e-02 -9.98137236e-01 2.98403531e-01 4.17812645e-01
-3.57364267e-01 -2.17233837e-01 4.53086942e-02 7.69691825e-01
-1.15321070e-01 -2.96334893e-01 8.91474783e-01 2.25662053e-01
-2.73703694e-01 3.08303624e-01 -1.06091410e-01 -3.22913155e-02
9.73172247e-01 -2.87752271e-01 2.51948625e-01 -1.43210560e-01
-4.94553328e-01 4.93163466e-01 4.25983250e-01 5.22138357e-01
3.30719143e-01 -1.38805830e+00 -6.82672560e-01 7.77128994e-01
6.04901984e-02 1.51554585e-01 5.51379800e-01 6.58464015e-01
3.19538802e-01 2.02098534e-01 -1.35712549e-01 -4.17570144e-01
-1.33553100e+00 1.05731702e+00 -2.37392560e-01 -5.53110316e-02
-5.19295275e-01 9.51996267e-01 4.78093892e-01 -2.13720709e-01
2.41538346e-01 -3.84435356e-01 -6.60152212e-02 1.93270877e-01
6.24708056e-01 2.27500498e-01 -1.73837543e-01 -1.21197671e-01
-2.10318297e-01 3.58045131e-01 -3.68047237e-01 1.88084170e-01
1.44398820e+00 -3.83161664e-01 2.19294220e-01 3.68633181e-01
1.12401998e+00 1.47373900e-01 -1.51576066e+00 -7.15009928e-01
-1.17709026e-01 -9.17659163e-01 3.05592585e-02 -2.35046625e-01
-1.11976159e+00 6.61425531e-01 3.55395496e-01 6.02980852e-02
1.18673563e+00 -4.28342938e-01 6.75050974e-01 2.10817188e-01
4.12186742e-01 -9.91176903e-01 -2.30129153e-01 1.60090834e-01
7.91238904e-01 -1.42316377e+00 2.82653153e-01 -4.64588881e-01
-6.98470294e-01 6.87163711e-01 7.52849162e-01 -2.02568680e-01
7.21502244e-01 1.81322917e-02 -1.33642361e-01 2.15351641e-01
-4.21253741e-01 2.47545660e-01 2.62956798e-01 4.85487074e-01
3.28313917e-01 9.67383236e-02 -6.23330534e-01 9.60864782e-01
-2.16514125e-01 -6.97442368e-02 3.33712548e-01 8.65954578e-01
-5.16535759e-01 -1.25047016e+00 -6.25247478e-01 6.51826560e-01
-1.84101805e-01 -1.63997039e-01 9.17577296e-02 6.19661331e-01
-7.90121555e-02 9.04523492e-01 2.83456296e-02 -1.41733050e-01
2.50516266e-01 -2.88118683e-02 1.70857087e-01 -7.21427977e-01
-1.43601522e-01 2.09129423e-01 -4.15888727e-01 -1.10091887e-01
-1.93149269e-01 -8.76279891e-01 -9.49563682e-01 -9.66314971e-02
-6.30100429e-01 2.80410379e-01 2.76412547e-01 7.41886199e-01
4.63744015e-01 4.09751832e-01 9.37580824e-01 -5.72292030e-01
-1.10350716e+00 -1.06015849e+00 -9.77778673e-01 5.56589305e-01
4.15220261e-01 -9.58018243e-01 -6.06479466e-01 -9.04064253e-02] | [10.299111366271973, 3.274691581726074] |
283c421f-3425-424f-8bfd-cb6dc7f4e3b7 | coarsenconf-equivariant-coarsening-with | 2306.14852 | null | https://arxiv.org/abs/2306.14852v1 | https://arxiv.org/pdf/2306.14852v1.pdf | CoarsenConf: Equivariant Coarsening with Aggregated Attention for Molecular Conformer Generation | Molecular conformer generation (MCG) is an important task in cheminformatics and drug discovery. The ability to efficiently generate low-energy 3D structures can avoid expensive quantum mechanical simulations, leading to accelerated screenings and enhanced structural exploration. Several generative models have been developed for MCG, but many struggle to consistently produce high-quality conformers. To address these issues, we introduce CoarsenConf, which coarse-grains molecular graphs based on torsional angles and integrates them into an SE(3)-equivariant hierarchical variational autoencoder. Through equivariant coarse-graining, we aggregate the fine-grained atomic coordinates of subgraphs connected via rotatable bonds, creating a variable-length coarse-grained latent representation. Our model uses a novel aggregated attention mechanism to restore fine-grained coordinates from the coarse-grained latent representation, enabling efficient autoregressive generation of large molecules. Furthermore, our work expands current conformer generation benchmarks and introduces new metrics to better evaluate the quality and viability of generated conformers. We demonstrate that CoarsenConf generates more accurate conformer ensembles compared to prior generative models and traditional cheminformatics methods. | ['Aditi S. Krishnapriyan', 'Danny Reidenbach'] | 2023-06-26 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 1.00529216e-01 -6.73438609e-02 1.72654375e-01 -3.13267499e-01
-1.05694008e+00 -7.65340447e-01 6.91484451e-01 4.24017668e-01
-3.67858559e-02 1.39055848e+00 4.43645597e-01 -4.29802716e-01
1.01511247e-01 -1.22563517e+00 -1.11534834e+00 -9.74385619e-01
6.87450496e-03 7.05005884e-01 -3.03040594e-01 -3.96336764e-01
2.09970102e-01 5.40560007e-01 -1.23295772e+00 4.21767265e-01
1.38211560e+00 2.02503145e-01 -1.32082580e-02 4.65037763e-01
2.63799965e-01 2.62639761e-01 -5.56895316e-01 -3.38645220e-01
1.46485603e-04 -6.81437135e-01 -6.84783697e-01 -5.50592482e-01
6.23601317e-01 -2.06230789e-01 -5.34197465e-02 9.43818569e-01
6.99786484e-01 5.59058070e-01 1.09401011e+00 -7.44642377e-01
-1.08921480e+00 6.00163817e-01 -1.38511449e-01 -1.83120057e-01
5.51955342e-01 4.92368639e-01 1.21450222e+00 -7.89940417e-01
1.04709160e+00 1.37177753e+00 5.62551320e-01 4.78180051e-01
-1.96572506e+00 -7.40224361e-01 9.34951082e-02 6.74812496e-02
-1.43771565e+00 -7.19959289e-02 5.68265080e-01 -5.85507870e-01
1.51918924e+00 1.61823899e-01 9.18228090e-01 1.45875752e+00
8.47041965e-01 1.34330437e-01 9.47378993e-01 -9.30672511e-03
8.17053318e-01 -4.32094306e-01 4.53776829e-02 6.38108671e-01
5.77426136e-01 1.51194677e-01 -3.89993042e-01 -6.64221704e-01
8.29176307e-01 1.42673552e-01 -2.35936359e-01 -2.20670417e-01
-1.04206312e+00 1.31758749e+00 7.44840682e-01 -1.34184463e-02
-8.65687907e-01 7.05851540e-02 -2.16824338e-01 -2.46533439e-01
2.27537572e-01 1.17321825e+00 -1.42600313e-01 1.09483607e-01
-1.00544047e+00 7.53774285e-01 8.56261015e-01 7.16159701e-01
1.05609429e+00 2.87336081e-01 -2.89170533e-01 1.31378427e-01
3.14689755e-01 1.87233627e-01 2.81498015e-01 -7.30996013e-01
-9.68093649e-02 6.39349401e-01 1.63315739e-02 -5.39365649e-01
-2.28504241e-01 -2.47447997e-01 -9.02679086e-01 8.99984762e-02
-2.10054174e-01 -8.33235681e-02 -1.17324805e+00 1.75972831e+00
5.57244420e-01 -2.13631600e-01 3.19710188e-02 6.22356415e-01
9.77097988e-01 1.09338069e+00 6.50217295e-01 -2.11706623e-01
9.21711743e-01 -4.70668465e-01 -2.99188495e-01 4.79629129e-01
3.83334249e-01 -5.05994856e-01 7.11202860e-01 2.63675123e-01
-1.05957818e+00 -4.24839318e-01 -1.16039515e+00 -3.65940213e-01
-5.00411093e-01 -3.76811385e-01 9.56200242e-01 5.90959609e-01
-9.74268675e-01 1.12733698e+00 -8.93968642e-01 1.98727936e-01
3.48981321e-01 5.90610623e-01 -4.89855409e-01 2.30997667e-01
-1.24699163e+00 6.62315905e-01 6.51333451e-01 -2.85191894e-01
-1.13188934e+00 -1.00539124e+00 -9.98993516e-01 -2.57224403e-02
-3.86975259e-02 -1.36912715e+00 8.10241580e-01 -2.58603781e-01
-1.57383740e+00 9.48988497e-02 -3.38752389e-01 -4.63262439e-01
-1.75866261e-02 1.87177002e-01 4.80863079e-02 -1.65056050e-01
8.13563615e-02 1.01151705e+00 6.06204152e-01 -9.92057145e-01
2.18366966e-01 -3.03085059e-01 -1.09307155e-01 2.51142055e-01
1.57165691e-01 -6.13057911e-01 1.52490526e-01 -6.20619595e-01
-9.39637274e-02 -9.29106474e-01 -7.99680114e-01 -8.79064083e-01
-7.62339771e-01 -2.16174796e-01 3.32205594e-01 -4.34723765e-01
1.10801482e+00 -1.34742534e+00 7.88912416e-01 4.96890634e-01
5.59747756e-01 -7.07338527e-02 -2.20415756e-01 6.64588511e-01
-3.98538500e-01 5.02047300e-01 -1.82033271e-01 7.10694492e-02
6.90269917e-02 -3.36956680e-02 -6.35878965e-02 4.83751260e-02
2.35446468e-01 1.19894648e+00 -9.16846156e-01 -3.95493656e-02
1.55196324e-01 9.61099148e-01 -1.36656296e+00 1.43929586e-01
-6.80412054e-01 6.52895212e-01 -7.19876051e-01 5.33697248e-01
3.26412946e-01 -5.85203111e-01 2.69356459e-01 -5.17968953e-01
-1.78381532e-01 4.05026019e-01 -7.31328607e-01 1.47124457e+00
3.68322432e-02 -1.18986815e-01 -7.97625303e-01 -3.16718012e-01
8.40674579e-01 1.35330826e-01 5.49652636e-01 -5.57586551e-01
-9.15290962e-04 1.55495675e-02 6.99259788e-02 2.10499004e-01
7.17771709e-01 -3.27518582e-01 -9.92584974e-02 4.64568406e-01
2.65342325e-01 -4.96651620e-01 3.55523288e-01 3.89373809e-01
7.95447707e-01 5.11811137e-01 5.54542542e-01 -1.77945450e-01
1.42204702e-01 8.92965402e-03 3.56141388e-01 6.44451022e-01
3.72042626e-01 4.46250141e-01 3.45677942e-01 -7.25257874e-01
-1.37747478e+00 -1.09743786e+00 7.84962322e-05 9.71459627e-01
-3.04448128e-01 -9.28029239e-01 -1.15695691e+00 -3.17160219e-01
6.02551438e-02 7.33093381e-01 -8.49004924e-01 -5.18843412e-01
-2.37457871e-01 -1.37722385e+00 1.78853884e-01 2.96183169e-01
-1.32801145e-01 -1.06397486e+00 -2.21506786e-02 6.35655344e-01
3.05933151e-02 -4.56014037e-01 -5.47150552e-01 2.64025122e-01
-7.52274632e-01 -1.03054667e+00 -6.84589505e-01 -2.00849175e-01
6.92131042e-01 -2.05928683e-01 1.24955583e+00 -4.62765336e-01
-6.67588830e-01 -1.85497388e-01 -1.67432353e-01 -3.52225870e-01
-7.46299684e-01 1.82952687e-01 2.51186550e-01 -2.79789895e-01
3.29194188e-01 -7.26832151e-01 -8.17338228e-01 -2.11544663e-01
-8.07732821e-01 9.85430703e-02 4.77708369e-01 8.88919950e-01
1.39683199e+00 -4.13688779e-01 3.81305486e-01 -9.47143912e-01
9.76754963e-01 -3.96004766e-01 -6.87137663e-01 1.57015282e-03
-7.26392865e-01 8.07718456e-01 6.63623393e-01 -2.08694190e-01
-8.47037852e-01 -3.07428744e-03 -5.15862286e-01 -2.13648662e-01
-1.74835488e-01 4.98507529e-01 -1.61229223e-01 -2.22032249e-01
8.60787511e-01 3.40749711e-01 -3.01213890e-01 -2.82947987e-01
7.12225556e-01 7.93013126e-02 1.91320285e-01 -9.54746008e-01
4.48733300e-01 5.51032089e-02 2.92604119e-01 -9.09016669e-01
-4.35966074e-01 1.09616823e-01 -5.73621333e-01 3.63415688e-01
1.28767681e+00 -8.55639517e-01 -1.14458334e+00 3.50496024e-02
-1.08537126e+00 -3.56548190e-01 -3.32985640e-01 4.28032607e-01
-6.14783704e-01 4.68760282e-01 -6.66456819e-01 -3.77268672e-01
-7.23071277e-01 -1.77589381e+00 1.35814071e+00 1.57600209e-01
-6.72842383e-01 -7.91707218e-01 6.82329595e-01 1.83049798e-01
1.49465144e-01 9.24821198e-01 1.34720409e+00 -5.59024215e-01
-9.04860795e-01 3.70238684e-02 3.34116638e-01 -1.53293461e-01
3.46228741e-02 1.94835991e-01 -5.74511826e-01 -3.89656067e-01
-6.09944165e-01 -3.16196829e-01 1.00462770e+00 7.39690065e-01
1.04905772e+00 -5.91144443e-01 -3.17208260e-01 8.24632227e-01
1.23171759e+00 4.60216463e-01 4.91412789e-01 -4.22520563e-02
1.04529536e+00 8.06989223e-02 -8.70853141e-02 5.56733906e-01
3.92147362e-01 4.91419166e-01 3.28345001e-01 -3.29895988e-02
2.50391126e-01 -6.22658670e-01 3.53752255e-01 5.08922458e-01
-6.56791806e-01 -4.28277314e-01 -6.42125070e-01 -1.55981660e-01
-1.42546713e+00 -1.24057937e+00 -1.20538406e-01 2.14209914e+00
1.28451765e+00 -3.06631804e-01 1.91494286e-01 -4.28170115e-01
4.15443957e-01 1.02293111e-01 -8.72083068e-01 -3.91174972e-01
-4.44803983e-02 8.44071269e-01 1.97658017e-01 9.35906887e-01
-8.83709729e-01 1.18866396e+00 7.12513447e+00 6.38344586e-01
-9.93203759e-01 -1.75593019e-01 8.95233333e-01 -1.75683677e-01
-8.35484684e-01 6.45516440e-03 -1.16817951e+00 4.33508158e-01
1.14656329e+00 -2.72192657e-01 3.73167038e-01 8.91572535e-01
2.84435004e-01 4.22531515e-01 -1.33508849e+00 6.90302014e-01
-9.19264033e-02 -2.04074788e+00 8.74349833e-01 4.51254427e-01
1.16389215e+00 -9.61652920e-02 8.82843807e-02 1.60614654e-01
9.13726270e-01 -1.46943736e+00 4.59169567e-01 5.96758664e-01
8.03068042e-01 -1.08515906e+00 2.98737288e-01 -7.82068893e-02
-1.02300131e+00 5.07891238e-01 -4.29338247e-01 2.75445670e-01
9.14010853e-02 4.18450952e-01 -9.93269801e-01 6.07889771e-01
3.03983688e-01 6.11277163e-01 -4.62184638e-01 5.75552762e-01
-6.00498468e-02 2.96353608e-01 -3.04433763e-01 -3.70272219e-01
3.66759986e-01 -7.79675186e-01 4.04795945e-01 9.92361546e-01
4.04867500e-01 2.79326469e-01 2.98650652e-01 1.58502507e+00
-3.44258010e-01 6.21822663e-02 -4.76133645e-01 -4.10440445e-01
5.30556262e-01 9.71992612e-01 -2.60952652e-01 -4.01726276e-01
1.25678688e-01 9.06308055e-01 1.62507802e-01 6.69324875e-01
-8.09555829e-01 -3.10191453e-01 7.26513803e-01 3.66696976e-02
1.96850836e-01 -2.81312346e-01 3.15239340e-01 -1.11584866e+00
-5.82192838e-01 -1.26660180e+00 1.93456307e-01 -6.75972342e-01
-1.03813469e+00 8.44823241e-01 -9.67695937e-02 -4.36113238e-01
-4.68241334e-01 -4.86999184e-01 -5.09797156e-01 1.10959840e+00
-9.47382033e-01 -9.64637339e-01 -1.64917424e-01 4.02645171e-01
2.53185004e-01 -1.02066629e-01 1.21089089e+00 -1.58243507e-01
-7.17424452e-01 5.02883136e-01 2.80677259e-01 -5.53725302e-01
6.96179688e-01 -1.53705359e+00 8.35019588e-01 5.04036367e-01
9.83837247e-02 1.32236457e+00 8.58524203e-01 -1.06074047e+00
-1.55262256e+00 -1.44738269e+00 3.86211723e-01 -7.09200799e-01
2.40818709e-01 -2.98623145e-01 -9.45298254e-01 5.47541738e-01
8.11966211e-02 -4.37052310e-01 1.07675660e+00 4.68415730e-02
-5.47820807e-01 4.78181720e-01 -9.27219629e-01 7.20294356e-01
8.58324647e-01 -5.02377629e-01 -2.90609568e-01 3.97704154e-01
8.98825347e-01 -3.25773120e-01 -1.43998706e+00 2.51163900e-01
4.78188485e-01 -9.79494631e-01 1.30438280e+00 -1.07470477e+00
4.53975409e-01 -2.73579001e-01 -7.26449043e-02 -1.41797006e+00
-7.97985315e-01 -1.05697012e+00 -2.68805742e-01 5.08534729e-01
5.98543644e-01 -5.15632093e-01 9.59539950e-01 5.91188192e-01
-2.38179237e-01 -8.56689572e-01 -5.26449680e-01 -4.33990687e-01
4.87088621e-01 2.47101605e-01 1.04933667e+00 6.76052094e-01
-2.52607912e-01 6.06271386e-01 -1.72383204e-01 -8.43946338e-02
6.83591247e-01 2.95971781e-01 6.18462026e-01 -1.25184655e+00
-5.79399228e-01 -5.35932481e-01 -1.86883256e-01 -7.79444814e-01
4.59941775e-02 -1.10192335e+00 -2.80536532e-01 -1.53025770e+00
4.36480880e-01 8.27822387e-02 -1.86434351e-02 5.00700891e-01
-4.07272130e-01 3.31129819e-01 -1.79130450e-01 -9.80866142e-03
-5.69710672e-01 9.92225289e-01 1.31089854e+00 -2.27493703e-01
-6.14778101e-01 -5.70965350e-01 -8.62780750e-01 3.26339841e-01
5.82073987e-01 -2.84436911e-01 -2.36659542e-01 2.67641008e-01
5.93661487e-01 -7.86119476e-02 2.83235669e-01 -7.86655545e-01
-1.44001842e-01 -3.41221184e-01 8.56354237e-01 -7.24471986e-01
3.97912800e-01 5.62573038e-02 6.89972579e-01 4.27378535e-01
-2.19909787e-01 1.29312277e-02 8.40617791e-02 7.59745359e-01
1.21484632e-02 2.44938910e-01 8.39536488e-01 -4.15880084e-01
-1.64013103e-01 9.38974202e-01 -5.45955181e-01 -2.26658821e-01
8.78075838e-01 -2.76709646e-01 -1.00549012e-01 -1.45263448e-01
-9.37028468e-01 -1.44257382e-01 7.69017577e-01 2.23832000e-02
6.05774581e-01 -1.30099666e+00 -5.94844282e-01 3.08131903e-01
-6.69068843e-02 7.87951723e-02 4.53741610e-01 2.92211413e-01
-6.88995123e-01 5.80454469e-01 -2.52103686e-01 -3.77381563e-01
-9.98480976e-01 4.87583280e-01 4.03666884e-01 -3.02983016e-01
-1.56513482e-01 8.86667132e-01 4.11756724e-01 -3.44712555e-01
-3.08361232e-01 -3.99833620e-01 9.43371132e-02 8.46823156e-02
3.88002396e-01 2.11426452e-01 2.07659513e-01 -4.81500268e-01
-1.52451143e-01 5.02033353e-01 -4.27260131e-01 2.40099519e-01
1.58924437e+00 5.67793071e-01 -1.28688112e-01 -1.78625971e-01
1.05993140e+00 -6.40092939e-02 -1.38213348e+00 4.96709049e-01
-5.70474863e-01 2.31040046e-01 9.58289281e-02 -6.13398433e-01
-2.98329741e-01 7.85780013e-01 2.51270503e-01 -1.36269420e-01
3.92786980e-01 -1.19049691e-01 7.80532598e-01 8.72187495e-01
3.77206534e-01 -7.48917818e-01 1.26204312e-01 4.92025822e-01
8.60521257e-01 -1.04216886e+00 2.36851335e-01 -6.09270036e-02
-4.43740577e-01 9.81632590e-01 3.14175785e-01 -5.90277873e-02
1.55147925e-01 -6.86624041e-03 -5.30884266e-01 -4.19662118e-01
-6.44669235e-01 6.90855682e-02 5.03221989e-01 6.28304958e-01
8.56959760e-01 2.47370750e-01 2.69170720e-02 6.58427119e-01
-5.34716785e-01 -3.66955847e-01 3.25926393e-01 5.81154943e-01
-4.62132633e-01 -1.51762366e+00 -2.15993747e-01 3.27469081e-01
-2.57899851e-01 -6.14961803e-01 -8.10904026e-01 5.08241951e-01
-1.11683505e-02 5.61728656e-01 -2.40861326e-01 -3.30571353e-01
4.26348485e-02 1.09024867e-01 7.36708522e-01 -8.34195495e-01
-5.57001054e-01 1.99674964e-01 -2.08092019e-01 -7.93476343e-01
8.11668858e-02 -4.66230929e-01 -1.12262177e+00 -5.06584406e-01
-2.15045005e-01 6.60600841e-01 3.85392427e-01 5.27519941e-01
1.02103221e+00 7.31080174e-01 3.16282541e-01 -1.30183637e+00
-4.24118966e-01 -6.50807261e-01 -1.12014636e-01 4.23742235e-01
2.74618179e-01 -3.86780649e-01 1.83071718e-02 1.81186393e-01] | [4.968823432922363, 5.687071800231934] |
f9af235d-4eea-4616-8a2f-d6f45b4cb2a5 | using-interventions-to-improve-out-of | 2210.10636 | null | https://arxiv.org/abs/2210.10636v2 | https://arxiv.org/pdf/2210.10636v2.pdf | Using Interventions to Improve Out-of-Distribution Generalization of Text-Matching Recommendation Systems | Given a user's input text, text-matching recommender systems output relevant items by comparing the input text to available items' description, such as product-to-product recommendation on e-commerce platforms. As users' interests and item inventory are expected to change, it is important for a text-matching system to generalize to data shifts, a task known as out-of-distribution (OOD) generalization. However, we find that the popular approach of fine-tuning a large, base language model on paired item relevance data (e.g., user clicks) can be counter-productive for OOD generalization. For a product recommendation task, fine-tuning obtains worse accuracy than the base model when recommending items in a new category or for a future time period. To explain this generalization failure, we consider an intervention-based importance metric, which shows that a fine-tuned model captures spurious correlations and fails to learn the causal features that determine the relevance between any two text inputs. Moreover, standard methods for causal regularization do not apply in this setting, because unlike in images, there exist no universally spurious features in a text-matching task (the same token may be spurious or causal depending on the text it is being matched to). For OOD generalization on text inputs, therefore, we highlight a different goal: avoiding high importance scores for certain features. We do so using an intervention-based regularizer that constraints the causal effect of any token on the model's relevance score to be similar to the base model. Results on Amazon product and 3 question recommendation datasets show that our proposed regularizer improves generalization for both in-distribution and OOD evaluation, especially in difficult scenarios when the base model is not accurate. | ['Amit Sharma', 'Emre Kiciman', 'Yashoteja Prabhu', 'Parikshit Bansal'] | 2022-10-07 | null | null | null | null | ['product-recommendation'] | ['miscellaneous'] | [ 3.09030384e-01 -2.70619220e-03 -4.44447875e-01 -5.72212994e-01
-3.67268920e-01 -6.90007091e-01 4.99054611e-01 2.92674452e-01
-2.96893507e-01 4.53113735e-01 4.13464785e-01 -4.22436386e-01
-5.26287377e-01 -9.17986572e-01 -9.52798605e-01 -3.50339025e-01
8.65904614e-02 4.03634667e-01 1.90855965e-01 -2.67925888e-01
3.06767255e-01 1.69484243e-01 -1.69124365e+00 6.69239342e-01
8.36542070e-01 9.90383089e-01 3.10329795e-01 4.20271546e-01
-2.65716076e-01 2.44824037e-01 -4.98606026e-01 -7.40399301e-01
6.14222050e-01 -3.80084455e-01 -6.00136936e-01 -3.31257768e-02
1.01452196e+00 -2.05502152e-01 -1.32566348e-01 1.01639557e+00
1.16212226e-01 4.88354087e-01 1.04998469e+00 -1.16869295e+00
-1.28226185e+00 7.96837389e-01 -2.83252537e-01 2.44666904e-01
4.19364423e-01 -1.49800226e-01 1.63440585e+00 -1.04462552e+00
5.66668630e-01 1.41949582e+00 8.14828157e-01 3.98842096e-01
-1.55382967e+00 -5.75059414e-01 8.53896856e-01 -1.42968699e-01
-1.06953061e+00 9.78224650e-02 3.34066600e-01 -6.66111708e-01
6.53494120e-01 4.87296194e-01 2.52731651e-01 1.36608493e+00
4.43355918e-01 6.49937510e-01 9.14980292e-01 -3.05825919e-01
2.43244648e-01 6.09488785e-01 3.76125395e-01 1.12924829e-01
5.21131039e-01 2.73232669e-01 -7.03405678e-01 -3.52066517e-01
4.71435815e-01 3.44953626e-01 -1.52157187e-01 -3.29285800e-01
-7.94656932e-01 1.05889785e+00 3.72142047e-01 3.32438320e-01
-5.07656753e-01 -2.76972055e-02 7.53420517e-02 7.43254304e-01
5.62868416e-01 8.42844248e-01 -6.92444623e-01 2.78902262e-01
-6.79626465e-01 5.41600287e-01 7.65967965e-01 9.35985684e-01
7.18981922e-01 -3.06027949e-01 -5.07030487e-01 8.44628453e-01
2.81390309e-01 3.56642485e-01 5.96721768e-01 -5.67983210e-01
1.82513267e-01 4.85370189e-01 2.57456213e-01 -1.13407266e+00
-4.14999276e-01 -7.13894725e-01 -5.05749702e-01 2.92173997e-02
5.98976314e-01 -1.14570344e-02 -7.05280542e-01 1.92241645e+00
1.61096588e-01 -2.32616067e-03 -3.02103549e-01 1.04239500e+00
5.76224685e-01 3.44967902e-01 2.70302445e-01 -2.79064685e-01
1.30042434e+00 -5.50943613e-01 -5.88362098e-01 -2.24556074e-01
7.26874471e-01 -8.41191769e-01 1.73774755e+00 5.45973003e-01
-6.67847574e-01 -7.11879194e-01 -8.50157619e-01 4.75387536e-02
-6.08792603e-01 -2.51092285e-01 7.30913460e-01 6.15270555e-01
-5.52437961e-01 9.88521516e-01 -1.01891436e-01 -6.34150028e-01
1.25655293e-01 3.42616737e-01 3.41379456e-02 -1.47345200e-01
-1.44046485e+00 7.37253785e-01 5.08440435e-02 -2.67462164e-01
-3.72429252e-01 -1.21595550e+00 -3.26880813e-01 1.69222206e-01
4.71003324e-01 -9.74611461e-01 1.13659549e+00 -1.19207036e+00
-1.06394172e+00 4.64345336e-01 -9.64261778e-03 -3.19267958e-01
5.27088225e-01 -4.69713360e-01 -5.53136706e-01 -5.43101072e-01
2.20746189e-01 3.20471525e-01 1.25537169e+00 -1.15307140e+00
-9.48818266e-01 -5.68333507e-01 3.17666471e-01 1.78880885e-01
-4.83636498e-01 -4.11926895e-01 -3.92037779e-01 -8.63258719e-01
-1.85137503e-02 -9.44060504e-01 -1.53663546e-01 -7.46248364e-02
-2.22719207e-01 -4.82921273e-01 5.36950290e-01 -1.84706330e-01
1.55032754e+00 -2.08290744e+00 -1.68313444e-01 3.55535179e-01
2.86112241e-02 -1.88039035e-01 -3.12235892e-01 2.99021989e-01
-1.98957250e-01 2.99462050e-01 2.65366793e-01 -1.63400009e-01
2.55264580e-01 2.12226778e-01 -5.97610891e-01 1.55221030e-01
-6.31979555e-02 7.87599564e-01 -8.72290492e-01 5.37085310e-02
-7.25250393e-02 8.41719061e-02 -9.55441356e-01 -1.77902076e-02
-4.73252177e-01 4.81562465e-02 -4.94087428e-01 1.86911121e-01
4.71700579e-01 -4.11006421e-01 3.77805382e-02 -3.01190317e-01
1.64208516e-01 3.89467508e-01 -1.38500535e+00 1.25292456e+00
-7.65814900e-01 2.85878450e-01 -4.26573962e-01 -7.60933757e-01
7.59132266e-01 -8.01989809e-02 3.65599751e-01 -7.70541191e-01
-2.74436116e-01 1.73930764e-01 1.48820905e-02 -4.71159488e-01
8.34195971e-01 -3.27745944e-01 -5.99455573e-02 5.84454179e-01
-1.52090341e-02 2.01991260e-01 7.59686604e-02 2.79934824e-01
9.75722432e-01 4.30779532e-02 1.22340776e-01 -4.67263997e-01
2.43090630e-01 -7.56493583e-02 3.95263374e-01 1.34612525e+00
3.71386498e-01 5.98357558e-01 4.50529575e-01 -2.99311936e-01
-9.06020641e-01 -9.62234080e-01 -4.46806431e-01 1.76797307e+00
1.46637961e-01 -4.49179590e-01 -2.74884850e-01 -1.22946656e+00
6.33018613e-01 1.14834058e+00 -9.23601389e-01 -3.64981145e-01
-6.60050139e-02 -7.10011065e-01 -5.58424853e-02 3.25154692e-01
-2.97878683e-01 -7.24158525e-01 6.35368004e-02 3.23170424e-01
7.30555207e-02 -6.56187654e-01 -9.89982724e-01 2.54131824e-01
-8.26486826e-01 -9.04055357e-01 -5.94236255e-01 -3.10343266e-01
7.05161452e-01 3.90773296e-01 1.34327364e+00 -1.09654423e-02
-2.32178904e-02 5.26378036e-01 -3.91811997e-01 -3.36735338e-01
-3.98271888e-01 1.40068486e-01 4.17138010e-01 2.24699929e-01
6.81837678e-01 -4.02889520e-01 -7.35237300e-01 7.09529400e-01
-1.06643546e+00 -4.18160439e-01 4.94497865e-01 1.07595253e+00
4.43073779e-01 3.10134180e-02 9.06624079e-01 -1.44252801e+00
1.03942394e+00 -7.25596845e-01 -4.62644994e-01 2.59552866e-01
-1.28973472e+00 2.44481698e-01 8.94530892e-01 -1.00844145e+00
-9.87464547e-01 -2.97000319e-01 8.28916579e-02 -3.80739152e-01
-2.60350332e-02 6.41047120e-01 1.08235650e-01 5.10710418e-01
1.18289208e+00 -2.49110088e-01 -4.48360026e-01 -7.37559855e-01
4.87747401e-01 6.03621602e-01 3.11105669e-01 -5.56054652e-01
6.61970198e-01 1.76677406e-01 -1.38916835e-01 -4.60566998e-01
-1.29925168e+00 -7.01913059e-01 -3.71041298e-01 1.20616697e-01
4.23609704e-01 -6.14712715e-01 -8.04144144e-01 -2.67305344e-01
-7.24299848e-01 -1.40005574e-01 -6.48503125e-01 7.44331002e-01
-2.94837832e-01 1.63998783e-01 -1.79281875e-01 -5.90126216e-01
-2.20784023e-02 -7.92228460e-01 9.14034188e-01 -2.64932029e-02
-5.50134242e-01 -1.24019420e+00 -4.56545502e-02 6.38082996e-02
4.44665343e-01 -4.37365770e-01 1.26518500e+00 -1.04137480e+00
-2.81471521e-01 -1.89995140e-01 -1.04442172e-01 3.50084990e-01
3.91084254e-01 -1.68076172e-01 -7.65531898e-01 -2.62245327e-01
-9.84930769e-02 5.17327972e-02 1.00752640e+00 4.89740282e-01
9.51346159e-01 -5.11445820e-01 -2.28544608e-01 1.17207676e-01
1.19294977e+00 -7.03367218e-02 2.68983126e-01 7.35944286e-02
4.41464186e-01 7.36091435e-01 9.80059087e-01 4.87641156e-01
4.97251153e-02 1.01198208e+00 2.70510703e-01 4.81778160e-02
-8.21311846e-02 -7.22735405e-01 4.56055224e-01 1.68461308e-01
1.88748002e-01 -3.41861576e-01 -1.95831686e-01 2.62284100e-01
-1.93574834e+00 -9.16326046e-01 -3.24110061e-01 2.82152700e+00
7.58633256e-01 2.44970500e-01 1.64296359e-01 -4.49326128e-01
5.57996273e-01 -2.98322231e-01 -8.88448894e-01 -3.63236636e-01
-5.79916090e-02 -8.02025422e-02 5.14181912e-01 5.76271415e-01
-7.72304237e-01 7.06663907e-01 6.09412909e+00 9.40022051e-01
-9.90159333e-01 1.23489708e-01 4.56215590e-01 -2.65222579e-01
-9.21451092e-01 -1.47189155e-01 -1.12510192e+00 5.31430840e-01
6.89374745e-01 -2.95837164e-01 3.68552446e-01 8.60492349e-01
4.50068265e-01 5.53545216e-03 -1.67023563e+00 7.97281325e-01
-8.84741638e-03 -1.22206342e+00 3.67955416e-01 1.82982773e-01
9.47783828e-01 -2.59872139e-01 4.02110279e-01 5.94919622e-01
5.12455642e-01 -8.72282267e-01 6.51162207e-01 5.67641973e-01
5.20100832e-01 -4.42535788e-01 5.87947607e-01 3.65988672e-01
-7.08606243e-01 -1.72778547e-01 -7.36980975e-01 3.88596021e-02
-9.17615667e-02 7.80610979e-01 -7.91865408e-01 2.60602534e-01
7.52884746e-01 5.79197943e-01 -5.21843791e-01 7.24691391e-01
-4.45873365e-02 5.30480087e-01 -2.68640816e-01 -4.84543527e-03
1.21206157e-01 -2.07454458e-01 5.35107672e-01 1.09713960e+00
5.28686821e-01 -3.60637046e-02 4.16597910e-02 1.06971765e+00
-2.54518032e-01 3.54411632e-01 -6.99002504e-01 1.50729448e-01
2.10561633e-01 1.07404912e+00 -3.01982641e-01 -2.75432706e-01
-6.69445395e-01 9.21226144e-01 9.21397954e-02 5.64278245e-01
-6.20575547e-01 3.01611573e-02 7.78364122e-01 4.04342085e-01
4.81808037e-01 3.86368155e-01 -3.77009392e-01 -1.13643014e+00
-4.93140109e-02 -9.30057883e-01 6.28392100e-01 -6.88668251e-01
-1.90427887e+00 2.44695663e-01 -6.56493306e-02 -1.30872190e+00
-2.32741997e-01 -5.64411938e-01 -3.08175683e-01 9.75930274e-01
-1.20973635e+00 -7.62072980e-01 -3.01145464e-02 5.96141756e-01
7.09142685e-01 2.49444414e-02 5.82341492e-01 3.63362849e-01
-1.53173536e-01 8.66834641e-01 2.09547222e-01 -5.43905973e-01
1.15720642e+00 -1.36055088e+00 1.58043712e-01 5.79025269e-01
4.04323399e-01 1.03155673e+00 1.08059251e+00 -8.31646323e-01
-1.22865188e+00 -1.17625380e+00 9.28170562e-01 -6.73974156e-01
7.07856536e-01 -3.30179125e-01 -1.03954506e+00 7.91512370e-01
-2.56803930e-01 -1.14617892e-01 6.98153973e-01 8.21074545e-01
-6.28034770e-01 -2.21343532e-01 -1.01189148e+00 7.29435086e-01
1.23919117e+00 -4.25838619e-01 -7.09625959e-01 5.66442907e-01
7.77119040e-01 2.35912222e-02 -9.55155253e-01 1.73848823e-01
7.75273442e-01 -8.23354840e-01 8.88490140e-01 -1.32564187e+00
3.67253244e-01 -1.36347398e-01 -3.48387897e-01 -1.54412997e+00
-7.18034983e-01 -6.34452462e-01 -7.05314428e-02 1.09324515e+00
7.68937707e-01 -5.60980976e-01 7.70073056e-01 8.49168599e-01
-1.78132076e-02 -5.67784011e-01 -4.15404648e-01 -9.93090630e-01
2.13627949e-01 -7.60346591e-01 6.31224513e-01 9.24149036e-01
5.03960215e-02 7.07593262e-01 -4.29996252e-01 3.56987774e-01
4.13371623e-01 3.25563490e-01 7.13641107e-01 -1.48648345e+00
-6.28520906e-01 -5.54744959e-01 8.15913826e-02 -1.49058104e+00
1.09982276e-02 -9.61592793e-01 -1.80149198e-01 -1.19469881e+00
1.23694114e-01 -7.34639049e-01 -4.49487954e-01 2.31469139e-01
-3.52817386e-01 1.46482602e-01 1.65342823e-01 2.71984339e-01
-3.86655450e-01 1.50021642e-01 1.33226895e+00 -2.35074386e-01
-4.71837968e-01 7.23597169e-01 -1.10390723e+00 5.69123089e-01
3.97363096e-01 -7.38731146e-01 -5.28852999e-01 2.82168575e-02
8.51622283e-01 -1.79890797e-01 2.05374181e-01 -2.67908007e-01
1.30418122e-01 -3.56977493e-01 3.09413254e-01 -3.99191171e-01
-2.01474521e-02 -1.04000151e+00 2.05611050e-01 2.45433584e-01
-9.52098250e-01 -1.50177017e-01 -6.68905973e-02 8.48711491e-01
1.13497257e-01 -5.37405908e-01 4.16349560e-01 1.32414252e-02
-5.30588031e-01 3.79001737e-01 -2.06423789e-01 1.78917110e-01
5.17293930e-01 -2.89367974e-01 -2.30105445e-01 -5.00074744e-01
-9.37328637e-01 4.53389585e-02 2.20724285e-01 9.86079872e-01
3.25347066e-01 -1.36727929e+00 -6.85875952e-01 2.59288579e-01
5.05594492e-01 -5.16051173e-01 2.65121967e-01 8.10255766e-01
5.49492359e-01 4.94026870e-01 3.07424814e-01 -4.79800761e-01
-1.06437612e+00 8.04508567e-01 2.78462470e-01 -3.89336437e-01
-3.54379356e-01 7.19794750e-01 9.34404671e-01 -3.46439064e-01
3.02100897e-01 -6.66801870e-01 -1.88392296e-01 3.59203011e-01
5.17933607e-01 2.07946733e-01 3.69755775e-01 -1.60989031e-01
3.19109075e-02 4.10626739e-01 -4.69968855e-01 1.14282474e-01
1.09205151e+00 -4.75297302e-01 4.19201881e-01 6.35964215e-01
1.07338560e+00 3.11496586e-01 -1.20590007e+00 -5.60848534e-01
-1.85133528e-03 -7.25013971e-01 8.37542191e-02 -1.02092087e+00
-8.49608243e-01 5.65515280e-01 5.91476500e-01 6.87909484e-01
7.45947123e-01 1.62841767e-01 3.35832536e-01 5.12761533e-01
1.05655156e-01 -1.16179037e+00 8.03877190e-02 4.17423636e-01
1.03848302e+00 -1.27520800e+00 2.47172415e-02 -2.50434726e-01
-6.63505554e-01 9.76888299e-01 5.18303812e-01 1.14717990e-01
7.87110090e-01 -1.19057149e-01 -3.92281748e-02 -1.35876805e-01
-1.00497615e+00 -1.12021692e-01 7.55708516e-01 5.18367887e-01
4.93903577e-01 -5.27358241e-03 -3.75236750e-01 8.86458635e-01
-2.26805717e-01 -3.21312584e-02 3.79009485e-01 3.00672382e-01
-3.20325017e-01 -1.01934648e+00 -2.62337327e-01 1.06896985e+00
-5.28072238e-01 -2.92230159e-01 -1.73760399e-01 7.27666974e-01
9.13254172e-02 8.72821331e-01 1.22380055e-01 -4.06069726e-01
7.37341762e-01 2.14974936e-02 2.58027017e-01 -9.68382061e-01
-7.47534990e-01 -7.87571538e-03 1.46693597e-02 -5.74501097e-01
-1.17315955e-01 -7.60131478e-01 -9.29846823e-01 -3.02353054e-01
-6.32359385e-01 1.23242214e-01 4.70795304e-01 8.53102982e-01
4.35507357e-01 4.04139727e-01 7.85834372e-01 -3.61635298e-01
-1.02599132e+00 -9.18827772e-01 -9.19457972e-01 1.05608904e+00
1.86349124e-01 -7.93953538e-01 -6.96735740e-01 1.88500583e-02] | [9.856765747070312, 5.588678359985352] |
25751f0f-1869-4390-a709-840b7b146bbc | convolutional-neural-networks-for-sentence | 1408.5882 | null | http://arxiv.org/abs/1408.5882v2 | http://arxiv.org/pdf/1408.5882v2.pdf | Convolutional Neural Networks for Sentence Classification | We report on a series of experiments with convolutional neural networks (CNN)
trained on top of pre-trained word vectors for sentence-level classification
tasks. We show that a simple CNN with little hyperparameter tuning and static
vectors achieves excellent results on multiple benchmarks. Learning
task-specific vectors through fine-tuning offers further gains in performance.
We additionally propose a simple modification to the architecture to allow for
the use of both task-specific and static vectors. The CNN models discussed
herein improve upon the state of the art on 4 out of 7 tasks, which include
sentiment analysis and question classification. | ['Yoon Kim'] | 2014-08-25 | convolutional-neural-networks-for-sentence-1 | https://aclanthology.org/D14-1181 | https://aclanthology.org/D14-1181.pdf | emnlp-2014-10 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 1.30197704e-01 -1.49746865e-01 -1.45875692e-01 -8.22498024e-01
-7.47857451e-01 -5.87298751e-01 4.91260082e-01 2.13176280e-01
-9.77877319e-01 4.48169321e-01 4.52962458e-01 -7.27082849e-01
1.64576516e-01 -5.20012438e-01 -5.62038124e-01 -3.17244828e-01
6.75963089e-02 6.35192692e-02 2.60087878e-01 -8.40519786e-01
3.01707149e-01 2.20330998e-01 -1.29128087e+00 7.24157035e-01
2.63413072e-01 1.06783032e+00 -2.03461632e-01 1.12939060e+00
-5.04587293e-01 7.12442696e-01 -9.24598694e-01 -6.22570813e-01
5.31017184e-02 2.33740315e-01 -1.06690824e+00 -2.20755592e-01
6.85639203e-01 -1.61048293e-01 -2.21270606e-01 5.75255811e-01
4.56323951e-01 4.58737940e-01 4.48518455e-01 -1.04156709e+00
-1.02413797e+00 6.42507255e-01 -7.28339106e-02 5.54765463e-01
-2.06427858e-03 1.52187034e-01 1.59258151e+00 -9.58063662e-01
2.55519569e-01 9.87059891e-01 1.06982744e+00 6.99630976e-01
-9.60042655e-01 -4.85979170e-01 4.00413483e-01 2.31200978e-01
-7.58993685e-01 -4.34300900e-01 6.08257771e-01 -3.13561618e-01
1.96577120e+00 7.15618655e-02 3.49591345e-01 1.25436866e+00
3.30408722e-01 9.14720595e-01 4.47939157e-01 -5.00899613e-01
1.41249120e-01 1.75131723e-01 1.03923500e+00 5.12742579e-01
2.41328195e-01 -4.37580317e-01 -4.56722975e-01 -4.15100843e-01
1.52860209e-01 -1.73258603e-01 -2.70553201e-01 -5.13601676e-02
-1.01685429e+00 1.18928885e+00 3.98392379e-01 5.75312793e-01
-1.40548706e-01 3.98549169e-01 8.47492278e-01 4.79750425e-01
7.25206792e-01 9.89804804e-01 -1.35813904e+00 -3.02062392e-01
-7.26151943e-01 2.51709640e-01 1.18751633e+00 8.33116174e-01
6.51175857e-01 3.85770142e-01 -6.60340011e-01 1.04982018e+00
-8.49783793e-02 5.30816987e-02 9.09544468e-01 -6.17332697e-01
5.23293734e-01 5.99228382e-01 -1.59835413e-01 -7.26226807e-01
-6.99556828e-01 -6.59558177e-01 -6.23305082e-01 -3.22417498e-01
9.45569798e-02 -5.33125103e-01 -1.16681659e+00 1.44725001e+00
-1.49121329e-01 -1.83024615e-01 1.94279864e-01 3.80962878e-01
1.41959512e+00 8.27488899e-01 1.51494041e-01 3.00334424e-01
1.57319319e+00 -1.39565516e+00 -8.37301850e-01 -4.35715109e-01
9.85315681e-01 -6.35670424e-01 1.36309505e+00 3.28216106e-01
-9.38996792e-01 -5.26367426e-01 -1.22495425e+00 -2.99935699e-01
-9.73474681e-01 5.93631454e-02 7.76994348e-01 8.08486223e-01
-1.36072946e+00 4.43403125e-01 -4.75889534e-01 -3.83600742e-01
5.43108940e-01 4.17261928e-01 -2.66280115e-01 2.10004687e-01
-1.42049575e+00 8.89031053e-01 3.67419988e-01 6.34143800e-02
-4.52838123e-01 -9.23330188e-01 -9.91561234e-01 3.17322820e-01
2.10863780e-02 -5.05141437e-01 1.81047022e+00 -9.11068916e-01
-1.53423393e+00 6.03523672e-01 -1.60812572e-01 -4.70498949e-01
-1.84535787e-01 -3.79620969e-01 -2.49223068e-01 2.44664447e-03
-2.28812978e-01 5.51077783e-01 7.15123594e-01 -8.13984752e-01
-4.52407181e-01 2.12160528e-01 4.92506057e-01 -5.11018857e-02
-1.23971975e+00 3.47225994e-01 -5.03590107e-01 -5.70192456e-01
-5.86603820e-01 -6.96726739e-01 -3.87810200e-01 -1.78899214e-01
-3.95616621e-01 -6.24936521e-01 6.19247854e-01 -3.34832221e-01
1.10584939e+00 -1.85667610e+00 -1.60361052e-01 -3.49374232e-03
1.70712262e-01 5.39666176e-01 -5.54051161e-01 3.29454899e-01
-3.86554867e-01 4.58140969e-01 -6.94340542e-02 -6.59527659e-01
-4.19089058e-03 2.14688897e-01 -1.98208794e-01 1.23378918e-01
4.22768623e-01 1.23207402e+00 -7.78218329e-01 -1.09764948e-01
-1.30904928e-01 4.42604125e-01 -6.30873322e-01 1.73379079e-01
-2.85659432e-01 -4.10897493e-01 -3.08877528e-01 4.01122600e-01
4.38382238e-01 -5.58994114e-01 -3.81468832e-02 -2.20829472e-01
1.58567667e-01 5.49955189e-01 -6.77174330e-01 1.50969660e+00
-7.33372509e-01 1.01139772e+00 4.59806062e-02 -9.31526601e-01
6.36016011e-01 3.65799606e-01 1.39913619e-01 -3.62326920e-01
3.22287083e-01 -2.19839260e-01 -8.52659121e-02 -5.89157045e-01
8.39529514e-01 8.38245302e-02 -1.40782788e-01 5.42161822e-01
7.25388527e-01 -5.69920614e-02 3.07434827e-01 2.12256238e-01
1.17749453e+00 -3.94922495e-01 2.45574117e-01 -4.76374805e-01
5.79676867e-01 4.51269001e-02 1.51552632e-01 8.27255368e-01
-2.08463445e-01 6.86078072e-01 6.64356947e-01 -7.82072544e-01
-1.02815950e+00 -5.53098142e-01 -2.30367720e-01 1.85244644e+00
-6.88386798e-01 -7.76302278e-01 -5.91771424e-01 -8.39843094e-01
7.80539289e-02 6.86817884e-01 -1.20169401e+00 -1.08152919e-01
-5.82625687e-01 -1.17896998e+00 7.23607481e-01 1.01887238e+00
2.34831810e-01 -1.15800369e+00 -3.17402184e-01 2.76885927e-01
1.93478744e-02 -1.19456434e+00 -3.97369176e-01 6.72105372e-01
-8.20315599e-01 -9.09217417e-01 -4.92361486e-01 -9.98378634e-01
3.35091799e-01 3.30571890e-01 1.58294988e+00 3.37058812e-01
-1.08059466e-01 4.49501038e-01 -6.62683368e-01 -6.81540489e-01
-5.71657345e-02 8.86182249e-01 -2.51590133e-01 -1.61184654e-01
6.49304807e-01 -8.28170180e-02 -5.35753250e-01 -9.97939706e-02
-1.09405077e+00 -4.72851336e-01 1.64489686e-01 1.05650806e+00
7.01979771e-02 -4.68653560e-01 6.21110797e-01 -1.14965534e+00
1.26723421e+00 -5.73278010e-01 -1.96761340e-01 2.17251271e-01
-4.67896104e-01 1.00824922e-01 7.26261020e-01 -4.72839624e-01
-6.96575224e-01 -3.06576461e-01 -6.55479372e-01 -1.59701139e-01
-2.95431800e-02 5.14407158e-01 2.91960776e-01 -9.31086913e-02
9.65158939e-01 -8.87847543e-02 -2.47963011e-01 -3.89805853e-01
3.86076510e-01 7.64441788e-01 1.24495164e-01 -3.26801211e-01
4.66106206e-01 1.78577542e-01 -3.88581216e-01 -7.82835603e-01
-1.24687076e+00 -6.42920196e-01 -4.44846988e-01 1.58687979e-01
8.99769664e-01 -7.64963448e-01 -6.33291483e-01 4.13738579e-01
-1.35096455e+00 -5.11300504e-01 -1.97261840e-01 9.46074352e-02
-8.42743069e-02 5.70068806e-02 -7.86016226e-01 -3.40860724e-01
-8.72273862e-01 -1.03906584e+00 1.01191461e+00 9.78341326e-02
-4.32534426e-01 -1.49428236e+00 1.32476524e-01 1.72817931e-01
1.01936424e+00 -2.07713708e-01 1.04361665e+00 -1.31865704e+00
1.57575116e-01 -4.88390714e-01 -1.22943863e-01 8.68530810e-01
-2.97079720e-02 1.14197247e-01 -1.27794921e+00 -2.48893216e-01
-7.49709159e-02 -6.09927833e-01 1.24211466e+00 4.34261382e-01
1.55358207e+00 -8.18722472e-02 -5.66137508e-02 7.79775739e-01
1.23313284e+00 -1.84893459e-01 3.36937159e-01 7.01800823e-01
6.79111958e-01 5.41613877e-01 -9.50641371e-03 2.55598754e-01
4.36531663e-01 3.11582863e-01 3.75253707e-01 -1.06220327e-01
7.59374797e-02 4.10929352e-01 9.67038348e-02 9.65568662e-01
2.44188279e-01 -5.53575933e-01 -1.06360185e+00 8.98398161e-01
-1.71756530e+00 -6.31085277e-01 -1.54856667e-01 1.43010414e+00
8.23067009e-01 4.13596213e-01 -2.38761365e-01 7.77379200e-02
3.03535730e-01 6.17177129e-01 -4.16950375e-01 -1.04203010e+00
-1.46701068e-01 7.82421231e-01 4.82101589e-01 5.04413188e-01
-1.36577713e+00 1.11695123e+00 8.12956905e+00 5.27834117e-01
-9.83807266e-01 3.02361310e-01 4.96672362e-01 -3.92285973e-01
-4.90707397e-01 -4.39769089e-01 -1.06779599e+00 -6.74225688e-02
1.32493484e+00 -2.59595290e-02 -7.24945441e-02 9.66831267e-01
-3.44658047e-01 3.90588999e-01 -1.00675154e+00 6.80147290e-01
4.25149202e-01 -1.65030038e+00 1.38407275e-01 -4.70999300e-01
8.80170345e-01 4.33471560e-01 2.47698992e-01 8.99229407e-01
4.78603065e-01 -1.08973026e+00 2.66888320e-01 1.75640374e-01
5.40515482e-01 -7.60188460e-01 1.22678435e+00 -1.66994650e-02
-8.57184947e-01 -1.23202004e-01 -6.08652949e-01 -8.67948681e-02
-2.12106302e-01 4.09877032e-01 -8.28418136e-01 1.05488308e-01
9.56255913e-01 7.32095182e-01 -8.37132990e-01 6.18503511e-01
-1.66242898e-01 1.09546447e+00 -6.71056807e-02 -6.75559103e-01
7.43806660e-01 5.90085924e-01 1.56891167e-01 1.81805146e+00
-2.54797369e-01 -9.02927816e-02 -2.94073135e-01 3.36726815e-01
-4.29503679e-01 3.78673136e-01 -3.62625629e-01 2.35340312e-01
8.75116140e-02 1.56373107e+00 -2.79129535e-01 -4.37137991e-01
-6.45557463e-01 5.46967328e-01 7.19807446e-01 5.01702845e-01
-4.64627296e-01 -8.79992187e-01 1.13054180e+00 -4.19757336e-01
7.18394279e-01 -3.63149643e-01 -5.53441703e-01 -1.43618858e+00
-5.29093593e-02 -7.76419580e-01 4.39836651e-01 -6.05281651e-01
-1.67096829e+00 9.39058244e-01 -1.45944208e-01 -7.88284540e-01
-3.33185822e-01 -1.22001791e+00 -9.21750784e-01 8.42544317e-01
-1.92343473e+00 -8.62851262e-01 -2.63686776e-01 5.70359051e-01
8.28016937e-01 -3.57425779e-01 1.12481844e+00 -1.32695213e-01
-6.19080484e-01 1.02256036e+00 2.51384377e-01 5.66779137e-01
7.63653278e-01 -1.42409611e+00 1.04445589e+00 4.01871979e-01
-6.69455156e-02 7.52075732e-01 4.65258390e-01 -1.49894461e-01
-1.31080568e+00 -1.21719253e+00 1.24459469e+00 -7.25695729e-01
9.20855939e-01 -7.78030217e-01 -1.01311040e+00 8.93729568e-01
7.60822356e-01 -8.71179551e-02 1.17965543e+00 7.79327393e-01
-7.82714188e-01 -1.01659097e-01 -7.79605389e-01 6.21408701e-01
5.18128276e-01 -5.42158365e-01 -7.27295935e-01 6.49210334e-01
1.11591017e+00 -2.72139430e-01 -8.26019049e-01 3.43405575e-01
5.20307302e-01 -5.36136389e-01 8.99801373e-01 -1.32158446e+00
6.05739474e-01 4.55330789e-01 -1.62013277e-01 -1.56308842e+00
-4.56820935e-01 -4.06121671e-01 -1.69915900e-01 7.86886275e-01
8.20622206e-01 -6.87874079e-01 8.12374175e-01 7.25990176e-01
-3.42311025e-01 -1.00659919e+00 -7.44150221e-01 -6.20211244e-01
6.33350313e-01 -8.09297502e-01 5.98139107e-01 8.33089292e-01
-2.33183187e-02 6.35497808e-01 -4.16340195e-02 -2.41187081e-01
8.67545232e-02 -2.28878364e-01 5.31488478e-01 -1.00310624e+00
-3.53044197e-02 -8.45702887e-01 -2.48770759e-01 -9.65331495e-01
5.10407805e-01 -9.36191797e-01 2.15656251e-01 -1.65421700e+00
-1.75367631e-02 2.21295863e-01 -6.60159588e-01 7.67592669e-01
-5.00625908e-01 4.40028727e-01 1.05857946e-01 -1.71171755e-01
-7.25367904e-01 5.04914999e-01 9.97551620e-01 -2.17140019e-01
4.74489527e-03 -1.00200877e-01 -1.03750288e+00 4.73607481e-01
1.16513968e+00 -4.90048438e-01 -2.57813305e-01 -8.54093909e-01
6.39593720e-01 -6.83768392e-01 -1.05049282e-01 -6.89201593e-01
3.34844351e-01 1.58248678e-01 4.18120623e-01 -4.64703888e-01
3.90909165e-01 -3.38013679e-01 -9.50913668e-01 1.56051144e-01
-9.09546316e-01 3.09937119e-01 6.48991525e-01 3.87348235e-01
-1.93146586e-01 -5.57853043e-01 6.02723777e-01 -1.69659302e-01
-5.50498545e-01 2.25972131e-01 -6.76174462e-01 2.62436450e-01
5.43090820e-01 1.19616807e-01 -5.73812425e-01 -5.71874917e-01
-5.18383682e-01 3.32356602e-01 -2.96285540e-01 7.85288393e-01
6.58436000e-01 -1.09864318e+00 -8.44234288e-01 1.47662595e-01
4.62897658e-01 -4.10461694e-01 7.67275691e-02 2.21936107e-01
-5.67988753e-01 9.29231524e-01 9.06987041e-02 -4.44003999e-01
-1.09619236e+00 2.83862591e-01 4.57653522e-01 -5.85082114e-01
-9.85933319e-02 1.25717676e+00 -5.64770140e-02 -8.47243488e-01
3.61610025e-01 -5.98434985e-01 -7.38228738e-01 2.16350973e-01
6.62306964e-01 -3.89266349e-02 6.96232438e-01 -2.82875836e-01
-3.75756472e-01 4.46891665e-01 -5.89700997e-01 5.49745373e-02
1.40823758e+00 2.73953080e-01 -3.18435282e-02 5.19232333e-01
1.76457191e+00 -4.89187211e-01 -7.59939671e-01 -4.10297990e-01
-9.79237780e-02 2.61657313e-02 3.93393695e-01 -7.24242747e-01
-1.14600849e+00 1.09173322e+00 1.09938413e-01 3.44002277e-01
8.87416780e-01 -1.84452891e-01 1.01368880e+00 1.14851773e+00
-2.13249534e-01 -1.07188177e+00 4.09509629e-01 1.31292844e+00
7.76479125e-01 -1.32772648e+00 1.23626348e-02 2.19969936e-02
-6.84003294e-01 1.51777661e+00 7.51196384e-01 -4.41234082e-01
1.23866940e+00 3.50090802e-01 2.84992337e-01 -3.14734608e-01
-1.40303147e+00 -2.15369731e-01 5.73530793e-01 4.10246611e-01
7.58351207e-01 -1.83215246e-01 -8.70064944e-02 8.59391809e-01
-4.13172007e-01 -3.05545121e-01 5.82540929e-01 1.19341111e+00
-4.74452525e-01 -8.85612369e-01 8.04826468e-02 8.06906164e-01
-9.11195219e-01 -6.31230116e-01 -3.34466100e-01 7.93992102e-01
-2.15359420e-01 1.12636566e+00 4.00207579e-01 -4.88816291e-01
6.18198693e-01 3.94203573e-01 4.29134555e-02 -9.04113650e-01
-1.55885112e+00 -5.61594188e-01 4.21990424e-01 -5.12022913e-01
-2.62361079e-01 -4.76128280e-01 -7.18450010e-01 -1.71225145e-01
-4.31913972e-01 1.08630210e-01 1.01535630e+00 8.50896418e-01
3.01923424e-01 8.56508017e-01 5.11343181e-01 -6.54039979e-01
-8.34196687e-01 -1.28045380e+00 -2.45804995e-01 2.99517989e-01
6.98129117e-01 -1.86644375e-01 -5.52070379e-01 -9.20751840e-02] | [10.787358283996582, 7.766941070556641] |
120856c2-ef44-4a7f-9b7b-6924a56dc9d8 | semi-supervised-federated-learning-for-1 | 2305.05110 | null | https://arxiv.org/abs/2305.05110v1 | https://arxiv.org/pdf/2305.05110v1.pdf | Semi-Supervised Federated Learning for Keyword Spotting | Keyword Spotting (KWS) is a critical aspect of audio-based applications on mobile devices and virtual assistants. Recent developments in Federated Learning (FL) have significantly expanded the ability to train machine learning models by utilizing the computational and private data resources of numerous distributed devices. However, existing FL methods typically require that devices possess accurate ground-truth labels, which can be both expensive and impractical when dealing with local audio data. In this study, we first demonstrate the effectiveness of Semi-Supervised Federated Learning (SSL) and FL for KWS. We then extend our investigation to Semi-Supervised Federated Learning (SSFL) for KWS, where devices possess completely unlabeled data, while the server has access to a small amount of labeled data. We perform numerical analyses using state-of-the-art SSL, FL, and SSFL techniques to demonstrate that the performance of KWS models can be significantly improved by leveraging the abundant unlabeled heterogeneous data available on devices. | ['Tao Zhang', 'Jie Ding', 'Eric W. Tramel', 'Enmao Diao'] | 2023-05-09 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 1.21768452e-01 -2.16059387e-01 -6.30991459e-01 -3.34671855e-01
-1.36949360e+00 -6.10714912e-01 1.80903584e-01 -1.03462920e-01
-1.24810874e-01 1.00790715e+00 1.67374220e-02 -5.66415370e-01
-2.32930183e-01 -4.62872058e-01 -7.33516574e-01 -4.88205194e-01
-1.15478113e-01 3.25981945e-01 5.72496690e-02 1.77445874e-01
-4.82197225e-01 3.51485848e-01 -1.70272970e+00 3.73511285e-01
5.31440854e-01 1.46552122e+00 4.16838303e-02 7.38158286e-01
-2.58909553e-01 1.12501442e+00 -5.30471981e-01 -3.20845515e-01
2.34559268e-01 -9.62771699e-02 -9.19833064e-01 -2.26508841e-01
1.96888790e-01 -5.24645925e-01 -3.25001389e-01 8.15408587e-01
7.15640604e-01 7.76848581e-04 2.29962066e-01 -1.74447799e+00
-2.71622151e-01 7.58982420e-01 -1.15942195e-01 -5.39039671e-02
3.58867794e-01 -2.59380966e-01 1.04227662e+00 -7.06990302e-01
5.14858305e-01 7.54546106e-01 8.01088095e-01 4.29804116e-01
-8.54015708e-01 -8.15300107e-01 1.20677717e-01 2.23496243e-01
-1.40840709e+00 -7.33831227e-01 7.59399474e-01 -1.22835129e-01
7.18295813e-01 5.25572360e-01 2.80452102e-01 9.52502608e-01
-4.31194335e-01 1.31123269e+00 9.81122911e-01 -7.15330362e-01
6.24313116e-01 3.45914215e-01 2.52639148e-02 6.33318543e-01
-3.77688296e-02 -2.40000218e-01 -1.19792163e+00 -8.15168977e-01
2.29275197e-01 3.94152045e-01 -2.47910231e-01 -4.07928467e-01
-9.91647184e-01 5.83963394e-01 -1.23810753e-01 1.87258676e-01
-3.80072296e-01 -1.16851367e-01 5.97208858e-01 5.64523876e-01
7.17193007e-01 -1.30536750e-01 -9.69745576e-01 -5.85719526e-01
-1.29040146e+00 -2.67683059e-01 1.02470148e+00 1.12966669e+00
9.95430052e-01 -2.08691701e-01 2.47259960e-02 6.72460914e-01
7.59388059e-02 6.86681688e-01 6.81788146e-01 -1.15780032e+00
6.56426668e-01 6.33312047e-01 3.36623758e-01 -4.08608973e-01
1.92209538e-02 -1.42003253e-01 -6.64699852e-01 -4.93373930e-01
2.30771363e-01 -4.19037104e-01 -3.76467675e-01 1.67651534e+00
5.59291959e-01 5.54277599e-01 4.12030797e-03 6.56717122e-01
4.04161185e-01 2.75122851e-01 -2.04237103e-01 -4.45270509e-01
7.69117355e-01 -1.08628249e+00 -8.18654597e-01 2.57530183e-01
1.01877260e+00 -5.98287404e-01 1.31345510e+00 4.33632076e-01
-9.02138591e-01 -2.62266368e-01 -8.81244779e-01 1.28924534e-01
-4.27926600e-01 2.25268304e-01 9.02292252e-01 1.06626105e+00
-1.04984450e+00 3.79623562e-01 -9.86278296e-01 -1.69417918e-01
7.40302086e-01 6.98201180e-01 -4.06195551e-01 -3.46297592e-01
-1.05917430e+00 -2.05526520e-02 -2.86125064e-01 -1.95537776e-01
-9.00041282e-01 -7.32203960e-01 -4.65326458e-01 1.14078246e-01
5.57272434e-01 -5.08758783e-01 1.76113892e+00 -8.04652929e-01
-1.52924240e+00 6.26295924e-01 -3.60334009e-01 -4.13279623e-01
4.02953714e-01 -1.26064524e-01 -5.66727281e-01 2.47167572e-01
5.25186732e-02 -4.52930704e-02 9.95851755e-01 -9.94548619e-01
-8.78300369e-01 -5.78369975e-01 5.71564622e-02 -9.29626971e-02
-1.13185358e+00 -8.05503782e-03 -3.56323510e-01 -4.25144136e-01
-3.00313920e-01 -8.46123040e-01 4.86240201e-02 5.89257479e-02
-3.78690571e-01 -2.47525692e-01 1.26546216e+00 -4.37279552e-01
1.31264019e+00 -2.37411880e+00 -3.31770897e-01 2.28211746e-01
3.30892831e-01 3.18290710e-01 -7.27332383e-02 4.87171412e-01
5.88869691e-01 -6.55485643e-03 2.54304796e-01 -7.71516144e-01
1.88996628e-01 2.39231974e-01 -3.60007882e-01 3.29047620e-01
-5.64254045e-01 8.35645199e-01 -8.49777162e-01 -7.45286644e-01
-2.13534050e-02 2.72069126e-01 -4.77315158e-01 4.09547359e-01
-2.11510375e-01 2.53278404e-01 -5.56642711e-01 9.55979228e-01
3.13945472e-01 -7.63267636e-01 4.55860376e-01 -1.56224454e-02
2.31073990e-01 2.26329744e-01 -1.34409690e+00 1.83267581e+00
-1.01850951e+00 2.83994734e-01 5.04214883e-01 -8.56331825e-01
5.30521095e-01 6.97366357e-01 8.28139961e-01 -3.82490903e-01
-4.52925414e-02 4.67134625e-01 -8.74128699e-01 -3.64696443e-01
3.58993411e-01 1.64862826e-01 -6.45756572e-02 1.18428874e+00
2.66569138e-01 5.09412289e-01 -4.46610868e-01 5.37557960e-01
1.37478733e+00 -3.04108024e-01 3.97782736e-02 2.08329901e-01
3.95355642e-01 -2.96493024e-01 4.21620280e-01 9.52063024e-01
-2.35147625e-01 3.99881601e-02 -8.88727680e-02 -2.49745339e-01
-6.17120624e-01 -9.22464907e-01 1.53024018e-01 1.55229747e+00
-3.53571102e-02 -6.37712419e-01 -6.50145113e-01 -8.76734674e-01
1.63218096e-01 1.24900252e-01 -9.44937691e-02 -1.83115125e-01
3.03973537e-02 -2.00766325e-01 7.34193444e-01 5.62719703e-01
4.42978710e-01 -7.29549408e-01 -5.09366155e-01 1.29672825e-01
-3.93870860e-01 -1.28762376e+00 -4.47836578e-01 3.87299418e-01
-7.34649837e-01 -1.01699197e+00 -6.33788705e-01 -7.32372046e-01
4.65704501e-01 6.43688440e-01 8.31378877e-01 -5.39206490e-02
-2.87961215e-01 8.10062289e-01 -4.64945644e-01 -3.99484903e-01
-1.06256798e-01 4.19526130e-01 5.26403964e-01 4.00122255e-01
3.77964795e-01 -6.59419835e-01 -4.19268548e-01 4.92726326e-01
-7.43640542e-01 -3.51278573e-01 1.24236815e-01 9.73209262e-01
6.09075367e-01 3.63535285e-01 8.24687660e-01 -1.18374467e+00
4.27493572e-01 -6.30387306e-01 -3.05607527e-01 5.36408067e-01
-6.15713716e-01 -1.61452711e-01 8.40134203e-01 -6.96314216e-01
-1.07201183e+00 1.77388981e-01 1.33081526e-01 -7.61696994e-01
-5.02952337e-02 3.91007990e-01 -3.90093654e-01 -4.50326502e-01
3.74431431e-01 2.93530934e-02 -1.15027405e-01 -8.02433789e-01
4.65238184e-01 1.75835443e+00 3.98816258e-01 -7.83018351e-01
4.49380338e-01 6.99315846e-01 -4.56838995e-01 -6.62262678e-01
-8.92917812e-01 -8.04666817e-01 -2.71985054e-01 -8.82211104e-02
-9.15652514e-02 -1.11993968e+00 -8.55802476e-01 4.36741173e-01
-5.64101815e-01 -5.40842652e-01 -7.38901734e-01 4.75771099e-01
-7.21351027e-01 2.78568506e-01 -5.74336529e-01 -1.23354101e+00
-6.04155362e-01 -8.47461820e-01 1.30836868e+00 -9.95556265e-02
-1.56359360e-01 -9.16947961e-01 -7.84461275e-02 6.91738248e-01
4.72541183e-01 -2.93326885e-01 8.40748072e-01 -9.17229295e-01
-3.74946624e-01 -5.81673682e-01 1.12639174e-01 2.69597024e-01
5.04435480e-01 -3.91321927e-01 -1.35584092e+00 -5.30595481e-01
-7.68671781e-02 -9.00650561e-01 2.12497503e-01 5.61409667e-02
1.47304666e+00 -4.44619656e-01 -4.83216822e-01 5.87873042e-01
1.14243615e+00 -1.25293821e-01 -1.09426945e-01 4.75627836e-03
4.94889468e-01 3.17082345e-01 6.48067474e-01 9.14815903e-01
3.65398198e-01 5.28226554e-01 1.91360056e-01 -1.30837515e-01
3.35832201e-02 -6.29342973e-01 1.76879808e-01 1.01274717e+00
2.80121118e-01 -1.24252178e-01 -6.15428030e-01 6.72106326e-01
-2.03823733e+00 -5.74383080e-01 4.02351081e-01 2.37639403e+00
8.76597643e-01 -2.30025858e-01 3.10069680e-01 7.43026197e-01
6.90119922e-01 -3.64550017e-02 -8.92157495e-01 7.67710283e-02
-3.86490747e-02 4.08955872e-01 5.12858748e-01 7.18275085e-02
-9.17068005e-01 6.81000352e-01 6.50858355e+00 1.05644822e+00
-1.03879762e+00 6.70699298e-01 5.02050400e-01 -4.87646163e-01
-2.77516127e-01 -3.16357911e-01 -6.37614667e-01 6.83870554e-01
1.41047645e+00 -1.64658353e-01 8.43237817e-01 1.09617460e+00
9.28414911e-02 1.65872872e-01 -1.10407972e+00 1.47209990e+00
-3.36482495e-01 -1.49232090e+00 -3.51169378e-01 8.63021612e-02
8.74350071e-01 3.63989204e-01 5.89103997e-02 2.64184922e-01
4.31864470e-01 -5.76186240e-01 5.09869993e-01 2.58562565e-01
1.37263989e+00 -6.00585282e-01 6.62757695e-01 7.37002134e-01
-1.29985821e+00 -4.10972446e-01 -1.60899351e-03 -2.46000718e-02
-5.46605624e-02 6.73896134e-01 -6.38212562e-01 5.62200129e-01
9.53631163e-01 4.18779105e-01 -2.92108089e-01 6.07986987e-01
2.55939335e-01 1.03119516e+00 -6.32753015e-01 -1.85874954e-03
-1.04451299e-01 3.18858176e-01 -8.67001414e-02 7.16252029e-01
2.99143851e-01 -1.32365093e-01 3.22556794e-01 2.73207158e-01
-6.06214464e-01 3.13892603e-01 -5.63573480e-01 -2.45650649e-01
9.97542799e-01 1.18030632e+00 -3.35371017e-01 -4.52438742e-01
-6.84094310e-01 9.94932771e-01 3.15661520e-01 3.55281323e-01
-3.33913386e-01 -3.01405370e-01 6.13589108e-01 1.14294827e-01
2.62673020e-01 -9.58375409e-02 9.54490453e-02 -1.22359514e+00
2.20661640e-01 -9.84052598e-01 5.30760467e-01 -4.81269658e-01
-1.41655099e+00 4.48807061e-01 -3.50025535e-01 -1.15897131e+00
-4.47883487e-01 -1.88137770e-01 -1.62499055e-01 5.33582330e-01
-1.56527281e+00 -1.36983573e+00 -1.79633290e-01 1.42007875e+00
2.48559237e-01 -4.90245402e-01 1.23576009e+00 6.95664465e-01
-3.56165737e-01 1.01019084e+00 6.72966957e-01 -1.83271646e-01
6.38830066e-01 -7.45322883e-01 3.56863551e-02 3.75545651e-01
5.74740291e-01 5.65409958e-01 2.79507879e-02 -4.05503035e-01
-1.76197910e+00 -1.12400997e+00 8.48990560e-01 -9.35394764e-02
8.00863624e-01 -5.52736282e-01 -4.88571882e-01 6.30663395e-01
-4.60922003e-01 8.63045454e-01 1.35736024e+00 4.21499431e-01
-4.52920765e-01 -3.15890223e-01 -1.41284144e+00 1.36044174e-01
1.21269047e+00 -1.32819891e+00 1.53916344e-01 7.02036262e-01
8.71065378e-01 -5.86548410e-02 -9.93604362e-01 1.00326061e-01
6.05192006e-01 -6.38918400e-01 6.53085053e-01 -9.45514321e-01
-4.94572818e-01 2.05982663e-02 -6.53141499e-01 -8.69076014e-01
3.59475911e-01 -1.25650847e+00 -8.57743144e-01 1.31157100e+00
2.08193272e-01 -6.33425176e-01 1.21347058e+00 7.97714293e-01
3.45308185e-01 -6.54042244e-01 -1.32785857e+00 -9.85024154e-01
-5.18232346e-01 -9.07186151e-01 1.15532398e+00 9.32014346e-01
3.73562902e-01 -3.10727698e-03 -5.59845209e-01 2.27254257e-02
6.08927608e-01 2.48196259e-01 7.49340713e-01 -1.27590513e+00
-4.86571878e-01 5.36382735e-01 -3.73245239e-01 -9.71193314e-01
5.36611795e-01 -7.21368551e-01 -4.50192004e-01 -9.56914842e-01
-9.17306534e-05 -1.03940094e+00 -5.23432076e-01 9.16310012e-01
2.09035635e-01 4.70421821e-01 1.00412242e-01 6.84789479e-01
-1.25376582e+00 5.38882256e-01 5.61712682e-01 -2.05557019e-01
-2.85422742e-01 6.22827411e-01 -5.28897941e-01 4.79348868e-01
7.97755539e-01 -2.96522558e-01 -6.56499863e-01 -3.10750812e-01
1.89151555e-01 1.93671182e-01 1.74265683e-01 -7.94173837e-01
5.62904537e-01 6.85370862e-02 -5.89516759e-02 -2.12964311e-01
4.06763345e-01 -1.29178548e+00 3.57594085e-03 1.07772388e-01
-3.42076778e-01 -3.67764652e-01 -1.14385739e-01 8.34199905e-01
-1.89155787e-01 1.39893442e-01 1.46106318e-01 6.39070049e-02
-4.21024293e-01 4.25601810e-01 -2.01912999e-01 -2.98782904e-02
9.92284417e-01 5.12090847e-02 -5.16028292e-02 -7.14992344e-01
-7.34327793e-01 9.97395888e-02 3.75573725e-01 1.93808049e-01
4.15853530e-01 -1.31644893e+00 7.06536472e-02 4.58998293e-01
1.39435291e-01 -6.82537407e-02 3.07314187e-01 6.09911680e-01
1.14648946e-01 6.13357425e-01 3.38264644e-01 -4.78112340e-01
-1.31715214e+00 6.95074975e-01 -2.60515995e-02 -1.40067115e-01
-3.98263782e-01 6.97730422e-01 -6.97964013e-01 -6.40730739e-01
8.16574812e-01 1.37261385e-02 5.39083898e-01 -1.29816785e-01
5.69368958e-01 6.96996927e-01 6.69713616e-01 -3.11910927e-01
-4.21894312e-01 -5.63359298e-02 1.11938365e-01 -7.24043921e-02
1.14652491e+00 -2.86257982e-01 1.40757322e-01 6.11423671e-01
1.48051071e+00 1.48636833e-01 -1.11185098e+00 -8.48040760e-01
-2.07545519e-01 -3.73929888e-01 3.47998142e-01 -6.13249719e-01
-1.17513335e+00 6.66622281e-01 7.03467071e-01 1.05171338e-01
1.23565316e+00 8.51337761e-02 1.27290571e+00 5.28396010e-01
1.25507712e+00 -1.17817187e+00 -1.35662004e-01 -1.24482699e-01
-2.12678567e-01 -1.27528501e+00 -2.83009946e-01 -2.81554788e-01
-4.22855467e-01 6.07232273e-01 1.04371615e-01 5.80753028e-01
1.13680482e+00 6.64282322e-01 8.97333920e-02 3.34500149e-02
-8.85504305e-01 3.11641023e-02 -9.34669450e-02 6.48019135e-01
7.53188580e-02 1.63691863e-01 2.82829702e-01 1.00015819e+00
-3.11469752e-02 4.81192648e-01 -6.89632073e-02 1.49784112e+00
7.00057112e-03 -1.27254510e+00 -2.55351365e-01 7.90397823e-01
-5.85104406e-01 -2.67352298e-04 -3.51462096e-01 1.17778331e-01
-2.34074909e-02 1.35861540e+00 -2.40948915e-01 -6.22276306e-01
-3.73502001e-02 3.44269753e-01 1.82307124e-01 -5.58039606e-01
-4.30855036e-01 -8.79833251e-02 -9.99328271e-02 -6.47817016e-01
-6.58202946e-01 -3.93879116e-01 -1.03863251e+00 -3.18819165e-01
-7.13193893e-01 5.00820160e-01 8.79241884e-01 1.01823628e+00
7.61476040e-01 1.57983452e-01 1.16186404e+00 -7.07186222e-01
-9.63394105e-01 -5.74684739e-01 -1.03216970e+00 1.16567232e-01
3.80095124e-01 -4.95942652e-01 -4.56150144e-01 7.24591836e-02] | [5.89973783493042, 6.243646144866943] |
361fab87-c854-43fa-ab83-4d7e2648deab | turku-neural-parser-pipeline-an-end-to-end | null | null | https://aclanthology.org/K18-2013 | https://aclanthology.org/K18-2013.pdf | Turku Neural Parser Pipeline: An End-to-End System for the CoNLL 2018 Shared Task | In this paper we describe the TurkuNLP entry at the CoNLL 2018 Shared Task on Multilingual Parsing from Raw Text to Universal Dependencies. Compared to the last year, this year the shared task includes two new main metrics to measure the morphological tagging and lemmatization accuracies in addition to syntactic trees. Basing our motivation into these new metrics, we developed an end-to-end parsing pipeline especially focusing on developing a novel and state-of-the-art component for lemmatization. Our system reached the highest aggregate ranking on three main metrics out of 26 teams by achieving 1st place on metric involving lemmatization, and 2nd on both morphological tagging and parsing. | ['Niko Miekka', 'Tapio Salakoski', 'Akseli Leino', 'Filip Ginter', 'Jenna Kanerva'] | 2018-10-01 | null | null | null | conll-2018-10 | ['morphological-tagging'] | ['natural-language-processing'] | [-4.05187547e-01 3.54207307e-01 -1.04077473e-01 -5.58946848e-01
-1.55937839e+00 -1.11548483e+00 3.38941514e-01 4.10050273e-01
-8.12198639e-01 8.27138603e-01 5.46656191e-01 -4.90831435e-01
2.34714672e-01 -3.06903809e-01 -6.93807423e-01 -2.15501204e-01
4.10002656e-02 6.45653307e-01 1.10113300e-01 -7.90986642e-02
-1.24737971e-01 3.13984185e-01 -6.21187687e-01 4.94622976e-01
9.66913164e-01 4.22464490e-01 -2.07965914e-03 7.16798902e-01
-4.57816899e-01 6.63851023e-01 -4.87707138e-01 -1.17985213e+00
1.71412975e-01 -2.36117020e-01 -1.33670652e+00 -4.64476407e-01
8.06240261e-01 1.62080660e-01 3.16176116e-01 1.03498507e+00
4.62583423e-01 -1.73906878e-01 2.46926859e-01 -3.77014399e-01
-8.63202214e-01 1.70724380e+00 -1.25565276e-01 4.10545766e-01
5.51695883e-01 -2.20805794e-01 1.56462538e+00 -9.39719975e-01
1.02527404e+00 1.26483119e+00 8.07983160e-01 4.92004722e-01
-1.09792995e+00 -4.34475780e-01 4.74097252e-01 -1.12834319e-01
-8.31455350e-01 -3.70392531e-01 2.78977424e-01 -3.57257575e-01
1.47715187e+00 -2.18487263e-01 1.59863427e-01 1.03606474e+00
1.14569172e-01 9.65852439e-01 1.59789371e+00 -7.91370332e-01
-1.46887094e-01 8.31084326e-02 6.19124830e-01 6.56640470e-01
4.14150327e-01 -5.29426038e-01 -4.93737578e-01 2.73940533e-01
3.39587539e-01 -1.02722776e+00 1.59087151e-01 2.02619687e-01
-1.31650400e+00 7.28044510e-01 1.49423564e-02 7.64156342e-01
-1.01906374e-01 -1.50018856e-01 6.38150692e-01 3.13509166e-01
8.26966584e-01 6.59527361e-01 -1.30591142e+00 -3.49262357e-01
-8.66711318e-01 -9.16026905e-02 1.03166997e+00 7.83957779e-01
6.80860698e-01 -2.61002123e-01 -6.96823820e-02 8.95856440e-01
1.80850789e-01 5.26314497e-01 3.74937832e-01 -5.22473216e-01
1.03268206e+00 4.24601406e-01 -3.03018212e-01 -1.34339094e-01
-7.29692340e-01 -3.26721787e-01 -1.63650453e-01 -2.43554711e-01
6.63844764e-01 -6.11272991e-01 -9.93647039e-01 1.83425021e+00
2.12280631e-01 -3.56055409e-01 1.78322524e-01 4.05026913e-01
1.03207695e+00 4.26912963e-01 7.23239243e-01 -1.30416349e-01
1.55264127e+00 -1.14395308e+00 -8.94009531e-01 -4.79835242e-01
1.27626026e+00 -1.23660636e+00 1.09315288e+00 5.31197846e-01
-1.22524738e+00 -4.41653550e-01 -9.62493241e-01 -4.84993935e-01
-7.44890988e-01 2.59219766e-01 8.63331914e-01 8.04232180e-01
-1.39317739e+00 6.26428485e-01 -9.05944049e-01 -7.77320743e-01
-8.11992735e-02 1.51221871e-01 -8.78564775e-01 1.26361936e-01
-1.17149270e+00 1.17174065e+00 5.31160653e-01 -2.25527301e-01
-2.34782055e-01 -7.01569021e-01 -9.85848010e-01 -4.07887846e-01
4.93615009e-02 -3.27162117e-01 1.36264825e+00 -5.41745603e-01
-1.64807022e+00 1.47424936e+00 -6.47906289e-02 -4.53491896e-01
3.65725547e-01 -1.00247300e+00 -4.23898876e-01 -2.11650357e-01
3.79923671e-01 6.55944288e-01 2.82454025e-02 -9.79169488e-01
-9.58460152e-01 -1.48230955e-01 1.19647430e-02 1.43930301e-01
-4.19177227e-02 6.32436454e-01 -3.17303538e-01 -6.28077209e-01
2.07807526e-01 -6.88321233e-01 -1.42363608e-01 -1.21976268e+00
-4.25606012e-01 -6.35191202e-01 4.62634742e-01 -1.36061895e+00
1.20539916e+00 -1.74819601e+00 3.17256004e-01 -4.68273073e-01
-3.58732402e-01 2.90197134e-01 -1.85057521e-01 5.88600457e-01
-2.62267232e-01 6.64369285e-01 -1.59503177e-01 -1.04636919e+00
1.25375748e-01 3.75041515e-01 -1.79837033e-01 2.41855741e-01
5.54151952e-01 1.16768646e+00 -1.06026006e+00 -6.95832789e-01
1.76148117e-01 2.14491576e-01 -2.62865275e-01 1.45820305e-01
3.17642838e-02 6.25323296e-01 -1.02802210e-01 6.47666633e-01
7.75764883e-01 5.03255904e-01 4.52972084e-01 -4.01766300e-02
-5.00673532e-01 1.10566616e+00 -8.83795261e-01 2.16814709e+00
-7.64348090e-01 2.10902840e-01 1.27934203e-01 -5.42876482e-01
6.14233851e-01 4.79851156e-01 1.58935785e-01 -3.67817014e-01
2.06562102e-01 5.96066117e-01 -2.42627058e-02 -3.32749009e-01
4.47621942e-01 -6.36606338e-03 -7.57483184e-01 2.75240004e-01
9.28453743e-01 -1.93655789e-01 5.69828272e-01 2.83352107e-01
1.09565842e+00 7.12821007e-01 6.79113865e-01 -7.22722471e-01
6.06225550e-01 -2.79914849e-02 5.33991635e-01 5.08551121e-01
-2.10342675e-01 5.91188967e-01 7.25379765e-01 -3.38740826e-01
-9.20584023e-01 -1.17849028e+00 -2.16040403e-01 1.46473694e+00
-7.00120628e-01 -5.61953425e-01 -1.10380447e+00 -1.39935935e+00
-4.53707278e-01 8.74571383e-01 -6.84995890e-01 6.38353229e-01
-1.04156733e+00 -6.56965673e-01 9.56131995e-01 5.07575989e-01
2.90952057e-01 -1.31209540e+00 5.23198731e-02 3.59160811e-01
-4.49389398e-01 -1.83118224e+00 -1.50017753e-01 6.87860548e-01
-7.45455563e-01 -7.96487212e-01 -2.77598351e-01 -1.32844400e+00
2.57457178e-02 -4.30058748e-01 1.59522712e+00 -3.58027428e-01
2.42895588e-01 1.71899691e-01 -8.76330853e-01 -5.54540157e-01
-5.83457708e-01 7.45153546e-01 -1.07437961e-01 -4.72819090e-01
4.90464628e-01 -5.17759323e-01 1.61373112e-02 -4.22591478e-01
-5.03888726e-01 -2.28999317e-01 8.51374328e-01 2.95206606e-01
5.84126353e-01 -7.48114347e-01 3.68725866e-01 -1.21405137e+00
4.68050241e-01 -2.21385270e-01 -3.88094902e-01 4.00946319e-01
-1.90688401e-01 1.85565978e-01 7.89262593e-01 1.02091938e-01
-1.14184904e+00 1.28810719e-01 -7.39171982e-01 3.88045430e-01
-4.34436798e-01 6.19268000e-01 -4.67877716e-01 2.57244319e-01
1.96919993e-01 -4.17458743e-01 -7.34716833e-01 -1.03283262e+00
8.00869882e-01 3.62600416e-01 6.72748387e-01 -7.80288696e-01
6.35990143e-01 6.71267658e-02 -1.91024616e-01 -5.18778801e-01
-1.49009657e+00 -5.72415411e-01 -1.57213020e+00 3.95212293e-01
1.44014668e+00 -9.40582633e-01 -6.84640482e-02 6.69028640e-01
-1.39287686e+00 -5.40652990e-01 -3.41210842e-01 3.68448496e-01
-3.29637766e-01 5.22061050e-01 -1.20246363e+00 -2.75670111e-01
-5.88054061e-01 -9.87995744e-01 1.06086040e+00 -9.54851508e-02
-1.65014952e-01 -1.43743145e+00 7.03035176e-01 2.72578895e-01
1.10765032e-01 2.46325821e-01 9.47505414e-01 -9.56814885e-01
4.58979011e-02 4.25021984e-02 -1.73327908e-01 6.49886847e-01
2.27425229e-02 7.45813479e-04 -9.72728789e-01 -1.36057034e-01
-1.22402996e-01 -4.61003631e-01 8.68090808e-01 2.53699839e-01
2.10180655e-01 1.45486325e-01 -1.50617868e-01 6.84434533e-01
1.48512924e+00 -3.13588679e-01 3.95186305e-01 5.79379857e-01
7.36777663e-01 5.79308093e-01 6.38630271e-01 -2.30157465e-01
7.22138107e-01 4.52797621e-01 2.13440508e-01 -1.32923469e-01
-4.10593271e-01 -4.18450058e-01 7.76206672e-01 1.49504769e+00
1.86835811e-01 -2.01574221e-01 -1.15279388e+00 8.04370701e-01
-1.56704080e+00 -1.67012453e-01 -5.09141147e-01 2.07769680e+00
9.94320095e-01 3.09459597e-01 -1.38936102e-01 -2.88583666e-01
5.54683030e-01 1.77419990e-01 2.71692693e-01 -1.07539785e+00
-5.01796544e-01 9.71153200e-01 6.56624317e-01 7.91694999e-01
-1.64112306e+00 2.02368712e+00 6.46859646e+00 6.84341311e-01
-9.25079107e-01 6.81185484e-01 3.81917208e-01 4.11586791e-01
-6.57180920e-02 2.82756865e-01 -1.33557689e+00 -1.12203341e-02
1.37625170e+00 2.71002352e-01 -2.05986165e-02 6.88995779e-01
-1.79297805e-01 -2.15662152e-01 -1.01070380e+00 3.10364455e-01
-8.28445479e-02 -7.13494420e-01 6.75202087e-02 -1.33189589e-01
7.12348759e-01 7.41560519e-01 -1.95683733e-01 5.75851917e-01
9.18189764e-01 -8.92099440e-01 8.91292632e-01 -5.15336171e-02
8.06938410e-01 -8.24756622e-01 1.14345348e+00 8.32856894e-02
-1.39719284e+00 4.96202946e-01 -2.44543970e-01 -1.76522791e-01
7.47642398e-01 6.23611271e-01 -5.50532460e-01 8.28290880e-01
7.32928038e-01 7.34621286e-01 -7.86778271e-01 6.69871449e-01
-1.25817549e+00 1.23807037e+00 -2.72732049e-01 3.24910253e-01
7.14037776e-01 -2.65188813e-01 5.63078165e-01 2.20493078e+00
-7.77530968e-02 -1.59865335e-01 2.82248437e-01 4.81066518e-02
-1.37368158e-01 8.05600107e-01 -2.74874717e-01 1.49989268e-02
3.72984946e-01 1.92518044e+00 -1.02306652e+00 -3.12915385e-01
-5.35162866e-01 9.83927727e-01 1.21112514e+00 -1.08429916e-01
-8.37065399e-01 -3.72398376e-01 6.02194011e-01 -2.96590984e-01
4.36488032e-01 -6.96829975e-01 -4.76491302e-01 -1.13335562e+00
-5.18035591e-02 -6.61328912e-01 6.33077383e-01 -1.76414832e-01
-1.39246309e+00 8.88480067e-01 -7.01760352e-02 -4.89862740e-01
1.31296977e-01 -9.74372447e-01 -4.44912821e-01 8.70651364e-01
-1.65457749e+00 -1.76513076e+00 2.76024550e-01 1.87216878e-01
4.93686944e-01 1.88972890e-01 1.28609371e+00 3.50234717e-01
-6.84974253e-01 8.55010808e-01 -1.60793439e-01 5.20937443e-01
1.32727301e+00 -1.81447613e+00 1.13502967e+00 1.42983377e+00
5.64306438e-01 5.07464468e-01 4.68319476e-01 -6.48165345e-01
-8.39401722e-01 -9.90781426e-01 1.69132650e+00 -1.00521839e+00
1.27851808e+00 -7.07141817e-01 -6.44216716e-01 1.31234968e+00
8.10891628e-01 -2.45290905e-01 7.07103372e-01 6.26739144e-01
-5.30433118e-01 4.64045763e-01 -9.08669889e-01 2.79819518e-01
1.20915687e+00 -3.57809454e-01 -9.27263498e-01 5.01905203e-01
1.03789926e+00 -6.92148268e-01 -1.17622018e+00 4.10977006e-01
2.80873924e-01 -8.32859397e-01 3.52060676e-01 -7.14492202e-01
4.12798941e-01 1.43163115e-01 -1.44497037e-01 -1.49946833e+00
-5.33881903e-01 -7.25597680e-01 4.77798581e-01 1.91019857e+00
9.75170434e-01 -6.32166207e-01 3.84848088e-01 -3.62793505e-02
-6.39697969e-01 -5.35321355e-01 -1.10939562e+00 -7.64953375e-01
8.25560451e-01 -6.10556483e-01 1.27468765e-01 8.74548376e-01
1.52616113e-01 6.99828446e-01 -2.27339529e-02 1.48774475e-01
3.90555263e-01 -4.09461707e-01 4.80781943e-01 -1.03203380e+00
-2.96814106e-02 -3.63819718e-01 -2.76153088e-01 -8.63146782e-01
4.98539835e-01 -1.31824613e+00 7.80134127e-02 -1.92966855e+00
1.32845759e-01 -2.83873141e-01 -3.18084866e-01 7.78046787e-01
-4.46584165e-01 3.95936221e-01 3.93970042e-01 -8.00314993e-02
-8.42173159e-01 -1.46315634e-01 7.43399858e-01 3.89529586e-01
-6.90537766e-02 -3.16023409e-01 -6.33766472e-01 8.98277938e-01
8.45218003e-01 -6.65270030e-01 4.14584994e-01 -9.92744982e-01
5.04095495e-01 -5.35131335e-01 -3.58110189e-01 -1.12744904e+00
-2.64916480e-01 3.26141357e-01 -5.73251434e-02 -7.10444272e-01
-4.02715877e-02 -2.13668987e-01 -4.56629336e-01 2.33464643e-01
9.95590761e-02 4.18208987e-01 4.15418565e-01 -2.86697835e-01
-2.34375954e-01 -4.20594245e-01 7.16098666e-01 -3.19336563e-01
-8.09433222e-01 -2.36275280e-03 -4.46450293e-01 5.58815956e-01
7.03996718e-01 2.84290671e-01 -4.01412129e-01 8.47424343e-02
-1.01026750e+00 2.03244805e-01 2.71475047e-01 5.43177485e-01
-2.73044139e-01 -7.86429822e-01 -1.09143162e+00 -2.34114587e-01
-2.81243593e-01 -1.45611569e-01 -1.27663359e-01 8.91326070e-01
-7.59109259e-01 8.23656738e-01 -1.56573564e-01 -2.75373489e-01
-1.22024083e+00 3.37550193e-01 1.04585081e-01 -1.37894678e+00
-3.06674123e-01 1.32557988e+00 -1.13929309e-01 -9.14451241e-01
-1.90885171e-01 -6.42069995e-01 -4.00008321e-01 3.11642349e-01
2.29899272e-01 2.42666677e-01 5.75927198e-01 -7.92765200e-01
-5.39823472e-01 6.82059228e-01 -2.97832072e-01 -4.20641810e-01
1.39773011e+00 -1.27273500e-01 -6.42202079e-01 6.62943423e-01
1.01457119e+00 8.53129208e-01 -7.64290154e-01 -5.87100536e-02
6.39850616e-01 3.32358718e-01 -2.79180229e-01 -1.18640471e+00
-6.71840787e-01 7.76577175e-01 4.32843417e-02 2.71851212e-01
6.30804300e-01 3.11737776e-01 8.94089401e-01 2.82287300e-01
5.85542381e-01 -1.27227712e+00 -4.28039998e-01 1.47135925e+00
5.01124501e-01 -1.21336651e+00 1.66434422e-02 -7.95320034e-01
-6.77293003e-01 1.03123641e+00 4.05842245e-01 -3.64637375e-01
6.44581318e-01 6.56387568e-01 6.15727186e-01 -7.47880386e-03
-3.72432679e-01 -5.30973136e-01 2.58311749e-01 6.22996151e-01
1.24515259e+00 4.08047944e-01 -9.53190923e-01 8.19589317e-01
-9.23190892e-01 -5.70351481e-01 2.65453070e-01 7.16124594e-01
-3.43235493e-01 -1.78591239e+00 1.19740114e-01 -1.02004595e-01
-1.26915109e+00 -6.39989495e-01 -6.83913648e-01 1.27719879e+00
3.05768281e-01 1.02251399e+00 -1.13371022e-01 -1.49825200e-01
4.76953804e-01 4.16560650e-01 6.62303448e-01 -1.17325675e+00
-1.11909842e+00 -1.90947413e-01 5.81725240e-01 -6.02475286e-01
-4.23623890e-01 -1.05945802e+00 -1.28601575e+00 1.54821187e-01
-2.53162920e-01 3.06618631e-01 8.28609288e-01 1.19100666e+00
-3.53044868e-02 3.19418222e-01 9.51401368e-02 -9.72924113e-01
-3.78859341e-01 -1.46494675e+00 -2.99105257e-01 2.86840826e-01
-2.91241974e-01 -1.95010051e-01 -4.13041115e-01 -1.11434817e-01] | [10.443948745727539, 9.98353099822998] |
5020b002-e6f3-4571-a31c-bedaf1124316 | hybrid-coarse-fine-classification-for-head | 1901.06778 | null | https://arxiv.org/abs/1901.06778v2 | https://arxiv.org/pdf/1901.06778v2.pdf | Hybrid coarse-fine classification for head pose estimation | Head pose estimation, which computes the intrinsic Euler angles (yaw, pitch, roll) from the human, is crucial for gaze estimation, face alignment, and 3D reconstruction. Traditional approaches heavily relies on the accuracy of facial landmarks. It limits their performances, especially when the visibility of the face is not in good condition. In this paper, to do the estimation without facial landmarks, we combine the coarse and fine regression output together for a deep network. Utilizing more quantization units for the angles, a fine classifier is trained with the help of other auxiliary coarse units. Integrating regression is adopted to get the final prediction. The proposed approach is evaluated on three challenging benchmarks. It achieves the state-of-the-art on AFLW2000, BIWI and performs favorably on AFLW. The code has been released on Github. | ['Zhenghua Chen', 'Haofan Wang', 'Yi Zhou'] | 2019-01-21 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-3.99631023e-01 -3.37495878e-02 -2.30863944e-01 -6.04334295e-01
-5.48423290e-01 -1.72671169e-01 4.82297629e-01 -2.08853871e-01
-4.05962586e-01 5.75457931e-01 1.81314811e-01 -3.42558958e-02
2.89784849e-01 -1.67751774e-01 -5.65124571e-01 -7.89759576e-01
1.78086191e-01 3.32680047e-01 -6.28141239e-02 -1.55115321e-01
3.00119847e-01 5.91963828e-01 -1.60187984e+00 -2.56920218e-01
5.19630313e-01 1.25864017e+00 -2.21578568e-01 3.06220740e-01
2.64035553e-01 4.91143465e-01 -3.35773438e-01 -4.07069743e-01
1.05804041e-01 3.69592719e-02 -4.04503167e-01 -5.95824048e-02
7.51100659e-01 -5.26231110e-01 7.08530918e-02 9.91414547e-01
6.76605523e-01 -3.56236883e-02 6.36341095e-01 -1.19026136e+00
-1.86270818e-01 9.29126814e-02 -9.45193172e-01 -2.70215366e-02
6.01345718e-01 -2.37509236e-02 8.23477328e-01 -1.20029092e+00
5.28094172e-01 1.23860645e+00 6.56653821e-01 5.24687052e-01
-7.77642667e-01 -1.01792073e+00 4.13043499e-02 5.16024113e-01
-1.90137470e+00 -8.83246899e-01 7.22786605e-01 -2.38987610e-01
6.64604962e-01 -8.68205428e-02 4.94120061e-01 7.53355503e-01
3.20878513e-02 4.97006953e-01 8.95682275e-01 -3.41204137e-01
-9.39795151e-02 6.65725246e-02 -2.35532969e-02 9.92265642e-01
6.27593249e-02 -1.54583737e-01 -7.60437965e-01 1.40314817e-01
4.35889989e-01 -1.18960686e-01 -4.88562375e-01 -2.02232465e-01
-7.33634531e-01 6.31480575e-01 5.67493021e-01 3.59219201e-02
-3.94573510e-01 -8.01565573e-02 1.00450523e-01 -2.52241679e-02
6.22687638e-01 -1.64809629e-01 -2.51047045e-01 -1.44381389e-01
-1.25797629e+00 1.63085118e-01 7.27463543e-01 1.00980413e+00
8.31166923e-01 -4.52685393e-02 3.24730054e-02 6.74823940e-01
8.08479190e-01 7.13129401e-01 3.99987996e-01 -8.39929044e-01
3.53007793e-01 4.71722215e-01 1.13652661e-01 -9.74744499e-01
-7.35351384e-01 -1.83312654e-01 -7.09177852e-01 2.88594782e-01
4.32255208e-01 -3.03341627e-01 -8.73177350e-01 1.64815819e+00
7.51249909e-01 3.81565750e-01 -3.79785061e-01 1.10695624e+00
8.97304177e-01 4.92948472e-01 -1.57237500e-01 -2.65954256e-01
1.31583512e+00 -9.78187144e-01 -8.43560100e-01 -1.83656633e-01
4.23195690e-01 -1.02364993e+00 7.55398035e-01 4.13857073e-01
-1.03221154e+00 -5.54516673e-01 -1.19270599e+00 -3.51962835e-01
-3.09742838e-01 5.14303803e-01 2.14376613e-01 7.30010569e-01
-1.27813292e+00 2.91619360e-01 -9.18147206e-01 -3.16756725e-01
3.94378752e-01 6.91361070e-01 -4.74828660e-01 1.26981527e-01
-8.89328718e-01 9.90154088e-01 -2.02304438e-01 6.06542945e-01
-4.34838355e-01 -2.65724957e-01 -1.01739717e+00 -8.08913633e-02
1.52724028e-01 -3.35972250e-01 1.24548113e+00 -5.31337678e-01
-1.88575017e+00 7.96744823e-01 -6.18852198e-01 -1.22290716e-01
5.57164669e-01 -4.15746868e-01 -2.27212384e-01 -9.38804150e-02
-1.99722692e-01 9.35700536e-01 1.19046545e+00 -8.33252311e-01
-5.55068970e-01 -7.89796352e-01 -8.03859085e-02 2.43151084e-01
-2.47801840e-01 3.53483185e-02 -8.94185007e-01 -1.44625813e-01
1.11181572e-01 -1.02617085e+00 2.57470101e-01 2.41650984e-01
-3.25397313e-01 -5.36669075e-01 7.71207869e-01 -8.54442596e-01
1.31220853e+00 -2.07862210e+00 4.00407566e-03 2.93440044e-01
2.47412473e-01 2.23077521e-01 1.27946019e-01 -2.53912807e-01
-1.15357921e-03 -3.32786202e-01 2.45322198e-01 -8.61474633e-01
-2.54578260e-03 -2.90448904e-01 1.89537868e-01 9.37503576e-01
1.23271741e-01 5.10262668e-01 -2.89276600e-01 -6.84311330e-01
1.12637594e-01 8.40247273e-01 -7.44063973e-01 1.13134548e-01
9.29720402e-02 3.65965694e-01 -3.37480605e-01 6.82024300e-01
9.58048761e-01 -1.49852708e-01 8.07859153e-02 -4.22878295e-01
-2.03739166e-01 3.47713560e-01 -1.23901069e+00 1.48346639e+00
-5.31581700e-01 7.93220937e-01 1.98460668e-01 -4.35904801e-01
8.95920515e-01 3.60740364e-01 2.61923194e-01 -5.14894068e-01
5.67787230e-01 1.56568855e-01 -1.59674287e-01 -2.80711263e-01
3.52895737e-01 3.60472322e-01 2.48082235e-01 3.69998544e-01
2.00708151e-01 2.76330672e-02 8.38756859e-02 -3.79006356e-01
3.83732617e-01 4.82113034e-01 5.11825740e-01 -1.78179175e-01
8.46861005e-01 -6.37526989e-01 5.89800954e-01 -2.65938044e-01
-2.44717732e-01 5.74901402e-01 6.55595005e-01 -4.13158089e-01
-7.61961639e-01 -7.14379489e-01 -1.73746943e-01 1.18345273e+00
-5.07248752e-03 -6.03498518e-01 -1.18003619e+00 -4.70803022e-01
-1.60163864e-01 2.93515414e-01 -8.22685719e-01 1.46465778e-01
-5.95439792e-01 -4.73699212e-01 3.89615506e-01 4.86013860e-01
3.93201381e-01 -7.54608572e-01 -4.63681757e-01 -1.53798848e-01
-1.79128349e-01 -1.21450293e+00 -6.91298664e-01 -1.79475814e-01
-4.15052712e-01 -1.04692447e+00 -7.13013113e-01 -6.61944091e-01
7.69684196e-01 6.26899824e-02 6.92796767e-01 1.86487302e-01
5.55544049e-02 -2.92621434e-01 2.19368059e-02 -4.76550907e-01
3.15975308e-01 2.79330820e-01 2.99688607e-01 7.90202618e-02
6.34060502e-01 -3.31156045e-01 -6.57149851e-01 4.06029403e-01
-2.53698438e-01 -1.56035759e-02 2.46207565e-01 6.50770128e-01
4.25169796e-01 -4.75553155e-01 1.96902424e-01 -5.56214690e-01
2.33781651e-01 -1.18514210e-01 -9.79952097e-01 7.13130236e-02
-4.72883672e-01 9.26546603e-02 3.64645720e-01 -1.21214852e-01
-9.08697367e-01 1.18688926e-01 -5.05337298e-01 -3.80874902e-01
-9.38597694e-02 1.05112173e-01 -2.88307637e-01 -2.14262202e-01
3.37279737e-01 -3.72831017e-01 7.14080855e-02 -4.80563968e-01
2.18282908e-01 9.20799375e-01 3.33801061e-01 -6.21395931e-02
5.05899131e-01 3.46092582e-01 9.65886712e-02 -9.49270248e-01
-8.00785422e-01 -3.72357339e-01 -9.65132177e-01 -1.47879303e-01
8.38688612e-01 -9.79534209e-01 -1.17990482e+00 4.84975100e-01
-1.21489429e+00 -4.44738604e-02 4.44887221e-01 6.03536844e-01
-3.06045651e-01 9.41079333e-02 -3.18855494e-01 -6.33379459e-01
-5.57970703e-01 -1.50359070e+00 1.29557538e+00 5.65526843e-01
-3.16906273e-01 -7.46096313e-01 -2.60233767e-02 1.03477687e-01
1.78251028e-01 -1.93390902e-02 3.54870409e-01 -2.07309350e-01
-3.74669045e-01 -3.53738189e-01 -1.69812992e-01 1.67991489e-01
5.37359044e-02 3.64255995e-01 -1.28750896e+00 -2.31443211e-01
-9.79858637e-02 -3.60152662e-01 5.75186849e-01 3.60606492e-01
1.08156478e+00 -3.03135693e-01 -2.25902408e-01 9.70613539e-01
1.03555155e+00 -8.18231851e-02 4.04042214e-01 8.97133946e-02
7.07912624e-01 4.82082158e-01 6.49118721e-01 6.99597180e-01
7.78418481e-01 8.15704882e-01 3.19854766e-01 1.36457860e-01
-7.99287930e-02 2.90685780e-02 4.87953931e-01 7.76338637e-01
-4.40190554e-01 -4.50617261e-03 -8.46236408e-01 9.71176177e-02
-1.41327953e+00 -5.78891575e-01 2.51350790e-01 2.20250010e+00
7.25424290e-01 1.62798852e-01 5.86544387e-02 1.83592439e-01
5.46816409e-01 3.19255054e-01 -2.88721412e-01 -4.31951165e-01
2.67439067e-01 3.89775962e-01 4.15669143e-01 6.92256689e-01
-1.06074858e+00 1.10744870e+00 5.70684338e+00 6.31303489e-01
-1.70137823e+00 1.65258020e-01 6.49296641e-01 -3.32515478e-01
1.77592471e-01 -4.93332118e-01 -1.49407887e+00 3.72021317e-01
8.39552879e-01 3.46965402e-01 4.77733314e-01 7.11737692e-01
2.96104878e-01 -1.91674218e-01 -1.17423356e+00 1.19473600e+00
3.68181020e-01 -8.01414728e-01 -4.63866919e-01 3.75600941e-02
4.33032483e-01 2.07838133e-01 2.00164258e-01 1.58445105e-01
-2.44601026e-01 -1.07756507e+00 6.72435522e-01 5.77492535e-01
1.04764652e+00 -8.76656950e-01 8.83901298e-01 2.14561060e-01
-1.10228395e+00 2.15052038e-01 -2.49438509e-01 1.19819708e-01
7.01266602e-02 2.11983457e-01 -1.05753183e+00 1.60403296e-01
6.92749202e-01 6.76422536e-01 -7.23300219e-01 8.31579566e-01
-5.92496276e-01 2.94069260e-01 -6.55669928e-01 -1.37861133e-01
-8.99229944e-03 -7.01213405e-02 1.56429008e-01 9.70643699e-01
3.90407503e-01 -1.41862452e-01 -8.37301090e-02 3.91414672e-01
-3.63560379e-01 2.05146194e-01 -3.18692207e-01 3.75002533e-01
3.69248509e-01 1.52620411e+00 -3.28810453e-01 -6.69751540e-02
-4.98388201e-01 7.63671100e-01 4.90050733e-01 1.93429023e-01
-9.08293068e-01 -2.67370135e-01 7.00300336e-01 1.40208840e-01
3.39062363e-01 -2.34487116e-01 -1.89238161e-01 -9.61884677e-01
1.04335621e-01 -8.91755998e-01 7.37413317e-02 -7.71643579e-01
-6.15609527e-01 8.46549869e-01 -2.23178804e-01 -1.09879625e+00
-6.73286140e-01 -7.13918030e-01 -4.52981204e-01 9.71322656e-01
-1.62042952e+00 -1.15548742e+00 -6.04379594e-01 6.77567482e-01
3.73457998e-01 2.49598343e-02 8.47347677e-01 4.84435290e-01
-8.78970861e-01 1.03747296e+00 -1.32927716e-01 1.02898628e-01
1.10968423e+00 -9.66257930e-01 3.81656051e-01 6.19089782e-01
-8.40671286e-02 5.55677593e-01 5.79348147e-01 -2.17742249e-01
-1.17443478e+00 -8.00678790e-01 1.16582346e+00 -2.05844834e-01
3.42235088e-01 -4.05533731e-01 -6.43599570e-01 6.84496224e-01
2.66497463e-01 1.91965610e-01 5.62001884e-01 1.79438740e-01
-2.51910418e-01 -4.80974436e-01 -1.07519817e+00 4.91738826e-01
5.17048597e-01 -4.99790490e-01 -1.46124125e-01 1.77854858e-02
2.65995622e-01 -7.27014422e-01 -7.27775335e-01 1.83516815e-01
9.93288994e-01 -9.41074729e-01 6.99560881e-01 -2.07066119e-01
1.29056558e-01 -3.72588485e-01 -3.12386695e-02 -1.18208611e+00
-1.13537818e-01 -6.16317332e-01 -1.54230908e-01 1.05370188e+00
3.64520133e-01 -5.06134033e-01 8.29066157e-01 3.92799348e-01
3.20993811e-01 -9.35890079e-01 -9.94391382e-01 -6.30146414e-02
-3.25240105e-01 -2.04228804e-01 8.34546566e-01 5.48637629e-01
-1.71803281e-01 5.45692384e-01 -3.47014219e-01 2.84392923e-01
5.70241451e-01 -1.19273670e-01 9.61222172e-01 -1.25266373e+00
2.24593759e-01 -4.23765630e-01 -5.00069797e-01 -1.02398634e+00
3.08019996e-01 -3.95750701e-01 6.45160973e-02 -9.98130500e-01
-3.21501642e-01 -1.05723932e-01 -1.11316927e-01 5.55316031e-01
-1.11298010e-01 7.04847097e-01 1.75831527e-01 3.08921877e-02
-5.44900477e-01 5.14063716e-01 1.13039339e+00 6.01839833e-02
-1.53845221e-01 1.95814103e-01 -5.01352787e-01 1.17982399e+00
9.48429048e-01 -1.77714989e-01 -8.24561119e-02 -5.71633220e-01
6.82967901e-02 -8.89547244e-02 2.42411601e-03 -1.11148071e+00
4.38441187e-01 2.13126451e-01 8.13281596e-01 -8.95031154e-01
7.69321978e-01 -7.26866364e-01 -3.01342100e-01 1.13648690e-01
2.31293458e-02 2.25161463e-01 5.86451814e-02 -3.91939469e-02
-2.35475481e-01 -6.15120828e-02 9.82202649e-01 3.85817647e-01
-4.20769840e-01 5.00298858e-01 1.13995738e-01 -2.44621411e-01
7.64052212e-01 -1.15218662e-01 -8.10751468e-02 -4.60870206e-01
-6.32667661e-01 1.99334085e-01 4.71822917e-01 3.07286859e-01
5.35892904e-01 -1.28440118e+00 -5.40577471e-01 5.98488986e-01
9.11284704e-03 4.04648632e-02 -5.28161786e-02 1.22887135e+00
-7.03014731e-01 5.70662379e-01 -3.10028464e-01 -6.72578275e-01
-1.59192109e+00 2.16220900e-01 3.19565207e-01 -1.58371255e-02
-6.67399615e-02 8.56662333e-01 4.54752631e-02 -2.04697013e-01
5.83553493e-01 -4.49889392e-01 -6.48582697e-01 4.18795526e-01
7.85055041e-01 3.66071761e-01 3.24152172e-01 -1.10694158e+00
-5.39089024e-01 9.90812421e-01 -9.65490267e-02 -4.31206897e-02
1.23754585e+00 -3.91268820e-01 -2.28083700e-01 1.79133832e-01
1.50215769e+00 2.84633309e-01 -1.25669885e+00 -2.04665244e-01
-2.32287407e-01 -3.55843782e-01 2.89799422e-01 -3.12464088e-01
-1.24756396e+00 1.42815900e+00 7.92486250e-01 -3.91151905e-01
1.20893025e+00 -2.76611120e-01 5.75630128e-01 3.73887539e-01
2.77563870e-01 -9.94644701e-01 -2.51266897e-01 6.56846046e-01
8.12679291e-01 -1.35254109e+00 2.20939010e-01 -3.65117639e-01
-4.10811305e-01 1.16527867e+00 7.34276295e-01 4.30445075e-02
9.72682238e-01 3.24824244e-01 2.49175131e-01 9.39416364e-02
-5.88190436e-01 -9.01764780e-02 6.44262493e-01 2.99121022e-01
8.41005802e-01 -5.07535376e-02 -1.18149631e-01 3.55695397e-01
-5.06797552e-01 2.06091493e-01 1.33769408e-01 5.82303047e-01
-3.96378994e-01 -8.63031030e-01 -4.05966818e-01 1.42244488e-01
-6.67144179e-01 -5.01568168e-02 -9.80899408e-02 7.64474988e-01
2.71664977e-01 7.21117496e-01 2.09675819e-01 -2.52574176e-01
2.01592386e-01 1.83123909e-02 6.39966011e-01 -3.21187824e-01
-2.57609189e-01 3.69499505e-01 -2.01277677e-02 -8.58965397e-01
-3.55682731e-01 -6.34230256e-01 -1.14867699e+00 -6.39686465e-01
-3.24331373e-01 -1.37668073e-01 9.05467272e-01 8.55796576e-01
1.56697735e-01 1.60670415e-01 8.07466209e-01 -1.22984445e+00
-4.05943930e-01 -1.33557677e+00 -5.02306521e-01 -7.78453499e-02
6.95770800e-01 -9.77968097e-01 -2.31910124e-01 3.26456912e-02] | [13.647063255310059, 0.28115278482437134] |
fa1b0cfd-dea2-4ff7-bc3a-d3baa9eb5d87 | flsea-underwater-visual-inertial-and-stereo | 2302.12772 | null | https://arxiv.org/abs/2302.12772v1 | https://arxiv.org/pdf/2302.12772v1.pdf | FLSea: Underwater Visual-Inertial and Stereo-Vision Forward-Looking Datasets | Visibility underwater is challenging, and degrades as the distance between the subject and camera increases, making vision tasks in the forward-looking direction more difficult. We have collected underwater forward-looking stereo-vision and visual-inertial image sets in the Mediterranean and Red Sea. To our knowledge there are no other public datasets in the underwater environment acquired with this camera-sensor orientation published with ground-truth. These datasets are critical for the development of several underwater applications, including obstacle avoidance, visual odometry, 3D tracking, Simultaneous Localization and Mapping (SLAM) and depth estimation. The stereo datasets include synchronized stereo images in dynamic underwater environments with objects of known-size. The visual-inertial datasets contain monocular images and IMU measurements, aligned with millisecond resolution timestamps and objects of known size which were placed in the scene. Both sensor configurations allow for scale estimation, with the calibrated baseline in the stereo setup and the IMU in the visual-inertial setup. Ground truth depth maps were created offline for both dataset types using photogrammetry. The ground truth is validated with multiple known measurements placed throughout the imaged environment. There are 5 stereo and 8 visual-inertial datasets in total, each containing thousands of images, with a range of different underwater visibility and ambient light conditions, natural and man-made structures and dynamic camera motions. The forward-looking orientation of the camera makes these datasets unique and ideal for testing underwater obstacle-avoidance algorithms and for navigation close to the seafloor in dynamic environments. With our datasets, we hope to encourage the advancement of autonomous functionality for underwater vehicles in dynamic and/or shallow water environments. | ['Tali treibitz', 'Yelena Randall'] | 2023-02-24 | null | null | null | null | ['simultaneous-localization-and-mapping'] | ['computer-vision'] | [-1.16427436e-01 -2.29455590e-01 9.37568367e-01 -4.20408845e-01
-3.43971759e-01 -1.01929832e+00 2.28774175e-01 2.69710887e-02
-1.28224099e+00 6.21006012e-01 2.15182960e-01 1.09456889e-01
-1.18258387e-01 -8.55565429e-01 -8.77363026e-01 -7.31660783e-01
-3.84767592e-01 5.88147700e-01 4.11261708e-01 -4.37190533e-01
4.96073484e-01 5.26437104e-01 -1.83217168e+00 -7.42299497e-01
5.17807305e-01 6.85201526e-01 6.23513162e-01 1.19701695e+00
2.21146092e-01 -5.39477058e-02 -1.49032846e-01 -2.64124811e-01
8.26464593e-01 -9.17702019e-02 -3.13589424e-02 -6.71733618e-02
1.11267197e+00 -9.15585577e-01 -3.16686660e-01 1.12991512e+00
8.03861678e-01 2.32567653e-01 -1.10235577e-02 -5.21182656e-01
4.17109542e-02 -1.45490944e-01 -1.82542413e-01 2.25323960e-01
5.63236713e-01 2.04061866e-01 6.56939745e-01 -8.56004775e-01
7.35718071e-01 9.34947550e-01 9.30298030e-01 3.85716289e-01
-6.21347129e-01 -1.99449614e-01 -2.91298181e-01 6.05220906e-02
-1.03202772e+00 -5.66400826e-01 1.38451438e-02 -5.67988753e-01
9.99941468e-01 -8.04970786e-02 1.01210296e+00 6.16575301e-01
4.68617827e-01 -3.43535990e-01 1.03792226e+00 -2.93724298e-01
5.29993415e-01 -2.34641254e-01 -2.14022815e-01 4.22660232e-01
9.56540823e-01 2.78607816e-01 -9.01054084e-01 2.23458204e-02
6.60524368e-01 5.97980499e-01 -7.68943548e-01 -3.55397791e-01
-8.25228930e-01 6.22892559e-01 2.49067858e-01 -2.97751248e-01
-1.31205931e-01 1.12777337e-01 1.45299345e-01 5.27937233e-01
3.00210770e-02 3.23279738e-01 -7.30628133e-01 -4.22428608e-01
-4.20229107e-01 1.03244290e-01 1.05620801e+00 1.08631837e+00
1.26733434e+00 2.44931504e-01 1.26693916e+00 3.82285923e-01
7.66004086e-01 1.46431553e+00 7.26677001e-01 -1.23028195e+00
4.97933030e-01 3.95815611e-01 2.99539030e-01 -7.39567459e-01
-7.92533994e-01 2.83424646e-01 -9.46041942e-02 7.56590247e-01
2.88906157e-01 -4.54753578e-01 -8.76227617e-01 9.81084347e-01
2.62698978e-01 1.58494592e-01 9.37970281e-01 1.10901833e+00
8.81899714e-01 3.21226865e-01 -8.32887888e-01 -1.52404755e-01
1.31508553e+00 -1.19925655e-01 -5.37056863e-01 -8.42382133e-01
6.63266420e-01 -5.75652063e-01 4.79388237e-01 -5.08568846e-02
-5.23733079e-01 3.90926488e-02 -1.25619555e+00 7.12030903e-02
-4.48324889e-01 -5.89193344e-01 2.79932678e-01 5.16923964e-01
-1.44332302e+00 2.44144112e-01 -1.17542326e+00 -6.06113195e-01
-3.23741704e-01 2.28668004e-01 -8.77479255e-01 -4.55955982e-01
-1.09967339e+00 1.03651488e+00 -1.57622680e-01 3.53093058e-01
-1.13383174e+00 -4.30218756e-01 -1.52003372e+00 -5.31842172e-01
-3.07865143e-01 -3.57607096e-01 1.04385304e+00 -4.38883811e-01
-1.26478958e+00 8.40730429e-01 3.15529630e-02 -5.61455965e-01
5.43049991e-01 -5.53001225e-01 -5.32595739e-02 2.44526401e-01
2.88454503e-01 2.67144829e-01 3.17737490e-01 -1.36451602e+00
-8.52190614e-01 -8.77449095e-01 2.38662437e-01 8.60635579e-01
-6.79581091e-02 -4.58785772e-01 -3.81051719e-01 3.15767586e-01
8.04574668e-01 -1.08223701e+00 -2.82905400e-01 2.87376821e-01
2.04469278e-01 9.54871237e-01 8.71666193e-01 -4.27866846e-01
2.73480624e-01 -2.01809049e+00 -3.29756252e-02 -2.56110281e-01
-2.58824766e-01 -2.00831741e-01 1.80434316e-01 8.58940721e-01
6.37097239e-01 -2.01806799e-01 -6.02538586e-02 -7.04283237e-01
-4.31321234e-01 9.51690018e-01 -5.61372153e-02 1.13601589e+00
-9.48872268e-01 2.51754165e-01 -9.19880152e-01 -1.71567127e-01
4.03273791e-01 2.01953441e-01 -4.55881745e-01 4.40037638e-01
4.26234424e-01 5.48642755e-01 1.66747928e-01 9.28590477e-01
1.14528501e+00 7.87789166e-01 8.03023353e-02 7.67616555e-02
-1.04446125e+00 -1.36029080e-01 -1.34840596e+00 1.71484053e+00
-6.11468136e-01 1.03111184e+00 8.30052197e-01 -6.47171959e-02
9.28863704e-01 5.68355322e-02 1.09862886e-01 -9.11494911e-01
5.01117017e-03 4.36077088e-01 -2.84349442e-01 -1.02836215e+00
8.98932040e-01 -6.22243881e-01 1.83980867e-01 -2.19800934e-01
-2.32325450e-01 -5.22146046e-01 1.25102073e-01 8.99937972e-02
8.92452121e-01 -1.43730372e-01 -9.48554575e-02 -4.83502150e-01
-8.57464522e-02 2.71721408e-02 7.05451667e-01 6.75939143e-01
1.37037903e-01 9.98159230e-01 -3.32750976e-01 -8.41198802e-01
-1.05190611e+00 -1.02124691e+00 -4.35069352e-01 5.14478981e-01
1.07318437e+00 1.51179418e-01 -1.39631793e-01 3.62246066e-01
2.18338788e-01 -2.25048140e-01 -5.95313609e-01 3.17321450e-01
-5.10106742e-01 -5.26431501e-01 3.17539006e-01 3.24768007e-01
4.18837458e-01 -4.86780554e-01 -1.25631750e+00 3.89836103e-01
9.78067517e-02 -1.37174106e+00 2.37050280e-01 1.61881939e-01
-1.32658494e+00 -1.16108561e+00 -4.54014510e-01 -6.58712626e-01
6.40264571e-01 6.87660158e-01 7.07082450e-01 1.89912878e-02
-1.15345560e-01 7.00794697e-01 -6.80857956e-01 -4.91107821e-01
-5.90498447e-02 -8.26368630e-01 6.54702008e-01 -2.58076191e-01
-5.06466515e-02 -5.60612857e-01 -9.00135040e-01 4.43831742e-01
-6.51207507e-01 -4.37918812e-01 1.09513208e-01 4.71131504e-01
1.92823902e-01 -5.59139431e-01 -4.09060389e-01 -6.19949028e-02
-4.34740067e-01 -4.11674440e-01 -1.20561361e+00 -5.27659774e-01
-1.33696213e-01 -4.60728317e-01 2.56553143e-01 1.36770934e-01
-6.13100827e-01 9.77818370e-02 -4.58371732e-03 -1.90753549e-01
3.97517532e-02 6.83061957e-01 8.68135318e-02 -5.68116426e-01
7.70105302e-01 4.17339116e-01 3.56257200e-01 -5.00400484e-01
-2.91462421e-01 6.21705115e-01 5.08624494e-01 1.16228499e-01
9.39806402e-01 1.30058634e+00 1.21166371e-01 -1.35909224e+00
-4.15862381e-01 -7.57226169e-01 -6.31251395e-01 -3.85663509e-01
8.27683806e-01 -1.62494552e+00 -5.74159384e-01 9.91481364e-01
-6.71812296e-01 -4.51773465e-01 1.51505634e-01 8.97211015e-01
-5.13568753e-03 7.91983187e-01 -5.61118186e-01 -9.10228908e-01
-2.77488321e-01 -1.31481218e+00 1.14848888e+00 7.87460804e-01
4.08169150e-01 -1.27192676e+00 5.89217484e-01 2.72286832e-01
3.63587171e-01 3.24816227e-01 -6.11637652e-01 1.80147439e-01
-7.00868011e-01 -4.99639958e-01 1.20721489e-01 8.95166668e-05
2.44462281e-01 8.49373490e-02 -9.08124685e-01 -8.81165862e-01
4.45175283e-02 -3.36236626e-01 7.36806571e-01 3.14559668e-01
-5.45045137e-01 -8.51518884e-02 -6.56668693e-02 1.09233677e+00
1.92968094e+00 1.57294735e-01 7.03666389e-01 9.64308202e-01
6.39518142e-01 5.06265342e-01 7.30544269e-01 7.65935779e-01
9.29845452e-01 4.89342898e-01 1.24930942e+00 -8.18509236e-02
3.91914070e-01 2.16883868e-01 4.18493629e-01 5.17490566e-01
-5.54858804e-01 -2.94466075e-02 -9.23752487e-01 8.80200148e-01
-1.24498725e+00 -7.25661397e-01 -5.31935215e-01 2.44757748e+00
2.15698525e-01 -2.21827596e-01 -7.92612374e-01 -2.14105621e-01
3.35449219e-01 -4.68300357e-02 -4.15249795e-01 -1.08997755e-01
-3.13511521e-01 -3.63506585e-01 1.62063730e+00 1.18319631e+00
-8.12827170e-01 6.40026152e-01 5.24809313e+00 -8.49662006e-01
-1.23767626e+00 -1.73647851e-01 -6.75930917e-01 1.31136179e-01
-3.61265391e-01 4.19155657e-01 -1.42405653e+00 2.73844093e-01
8.99916172e-01 2.89283156e-01 1.66976467e-01 8.98829520e-01
5.45348585e-01 -7.26275742e-01 -5.40842712e-01 9.29668188e-01
2.42278278e-01 -1.31367099e+00 -3.88282120e-01 4.99322116e-01
9.27289605e-01 9.85315800e-01 -6.46956921e-01 -3.68533820e-01
2.64284790e-01 -1.84919611e-01 1.19469988e+00 5.57345450e-01
8.44601214e-01 -2.01336890e-01 1.30139780e+00 2.96476215e-01
-1.17282999e+00 -2.26245806e-01 -6.15158558e-01 -9.73698080e-01
5.74605942e-01 2.81053841e-01 -7.61865795e-01 1.71380267e-01
1.44792950e+00 8.19580853e-01 -1.91339016e-01 1.37941122e+00
-1.19273297e-01 9.85401962e-03 -8.89558673e-01 -4.29574400e-02
2.57461399e-01 -6.04835570e-01 7.67706752e-01 7.36791253e-01
7.70357728e-01 1.96720660e-01 -2.59217639e-02 -1.40304103e-01
1.06299467e-01 -2.27820724e-01 -9.55281854e-01 7.20885277e-01
6.11489117e-01 1.00268614e+00 -4.26852703e-01 -8.82982761e-02
-4.04667020e-01 7.09142983e-01 -3.76260638e-01 8.08486342e-02
-1.12740010e-01 -4.16872174e-01 1.32538784e+00 -7.88781606e-03
2.67229602e-02 -1.10421252e+00 -1.40230656e-01 -1.14838755e+00
-3.94249223e-02 -1.08409613e-01 -4.35398985e-03 -7.92907238e-01
-7.01357305e-01 3.71061593e-01 -1.77379712e-01 -1.83703744e+00
1.05360255e-01 -8.43652070e-01 -5.77849329e-01 6.58294559e-01
-1.84936368e+00 -5.79195499e-01 -1.23594451e+00 1.27255633e-01
4.34194982e-01 3.48076522e-01 8.24757934e-01 1.51574552e-01
1.04557157e-01 -1.22448646e-01 9.09901440e-01 4.80617322e-02
7.05546916e-01 -1.40684700e+00 2.49707550e-01 1.16671383e+00
-2.18722194e-01 5.38086414e-01 1.24949789e+00 -7.81964540e-01
-2.27080584e+00 -6.47314787e-01 4.17290300e-01 -5.29012322e-01
9.00073469e-01 -2.77550131e-01 -8.46573532e-01 7.34766960e-01
-1.48411945e-01 2.14823663e-01 5.10748208e-01 -5.18405318e-01
2.80359715e-01 -4.88524735e-01 -1.03619683e+00 2.60926455e-01
7.82296419e-01 -1.74651206e-01 -4.09532458e-01 3.03561896e-01
3.81953806e-01 -1.18825686e+00 -8.39188457e-01 5.21454751e-01
9.57603216e-01 -1.33348918e+00 7.25690424e-01 4.83005583e-01
-1.92234386e-02 -7.87732720e-01 -5.84014714e-01 -1.26011622e+00
4.78431195e-01 -2.23249242e-01 7.83537984e-01 7.50957012e-01
3.43628526e-01 -1.12509298e+00 8.33591938e-01 3.73287320e-01
-8.55704486e-01 3.68862420e-01 -1.57105088e+00 -7.89518595e-01
-2.65163273e-01 -2.56280959e-01 -8.34802464e-02 4.08163249e-01
-2.24140793e-01 -9.52921435e-02 -4.14855301e-01 1.22722030e+00
1.02931452e+00 2.35979274e-01 1.17789972e+00 -1.42941511e+00
5.43220080e-02 3.31563890e-01 -1.05020046e+00 -1.04528999e+00
-4.55574363e-01 7.08249435e-02 6.21499538e-01 -1.92384708e+00
-4.20735747e-01 -1.41668081e-01 8.53607535e-01 1.86216936e-01
3.78260821e-01 5.85213006e-01 -2.69752651e-01 4.75683719e-01
-4.39528972e-02 4.94618267e-01 8.90189350e-01 3.31624776e-01
-1.70810074e-01 -2.51811445e-01 1.07581474e-01 8.12339008e-01
3.01188618e-01 -4.16916281e-01 -1.28153991e-02 -1.20602703e+00
5.73661506e-01 2.44898364e-01 3.55919003e-02 -1.40012312e+00
7.48336673e-01 -4.13856320e-02 5.59487939e-02 -2.79894829e-01
7.94658244e-01 -1.01006949e+00 2.83407539e-01 9.03023958e-01
7.03091741e-01 1.48365438e-01 9.67696831e-02 6.65116668e-01
-3.14786911e-01 -6.24679923e-01 1.10996509e+00 -6.27455056e-01
-1.52112103e+00 1.62673324e-01 -3.54741782e-01 4.89086732e-02
8.59332144e-01 -8.74468684e-01 -6.87871993e-01 -4.85399544e-01
-4.27560210e-01 4.44031239e-01 1.39790249e+00 1.08938850e-01
1.20136869e+00 -1.88233495e-01 -7.11370647e-01 4.64123636e-01
4.67769504e-01 4.74945724e-01 5.64034581e-01 5.54471612e-01
-1.84040856e+00 -4.66129333e-01 -2.02932999e-01 -9.63453889e-01
-1.27622569e+00 -2.10891396e-01 7.68059313e-01 8.79053593e-01
-9.54586685e-01 9.72953856e-01 -1.64021850e-01 -4.67554390e-01
-1.74235806e-01 -2.39403054e-01 -1.87031344e-01 1.67043120e-01
8.06465030e-01 3.84493411e-01 6.73689395e-02 -8.96485865e-01
-4.13649112e-01 1.47966981e+00 3.79751325e-01 -2.06695363e-01
1.35225976e+00 -9.24855053e-01 -1.11589283e-01 3.96622092e-01
1.14140773e+00 4.05567557e-01 -1.76795673e+00 1.37638748e-01
-4.33692873e-01 -6.41051292e-01 2.28400137e-02 -7.46868842e-04
-7.12595463e-01 8.03413093e-01 1.00720108e+00 1.27365306e-01
6.49582386e-01 1.49980905e-02 4.50112045e-01 5.83873212e-01
1.00376618e+00 -7.42339313e-01 -3.40680122e-01 1.32633007e+00
6.24830067e-01 -1.49100220e+00 2.42862403e-01 9.71963331e-02
-4.75186586e-01 1.40524769e+00 5.21364510e-01 -6.31744787e-02
3.83843094e-01 7.08402216e-01 1.11134899e+00 -1.32199496e-01
-1.48358613e-01 -4.17017132e-01 -8.02735984e-01 6.12167776e-01
-3.91772509e-01 -2.96122134e-01 4.95236292e-02 -4.38660592e-01
-4.26024914e-01 -7.02545047e-01 1.49197876e+00 1.43829060e+00
-1.08171189e+00 -4.46333617e-01 -6.58451974e-01 -1.25284627e-01
-1.42890200e-01 -1.64044276e-01 3.01894248e-01 6.59234345e-01
-7.29343435e-03 8.74326289e-01 3.21362406e-01 -2.82683462e-01
6.41427934e-01 -6.41951025e-01 8.62162188e-02 -5.73186159e-01
1.16188362e-01 -1.85740411e-01 2.79839575e-01 -3.29684496e-01
-5.38219988e-01 -9.93041694e-01 -1.63822329e+00 1.21125929e-01
-2.01190084e-01 5.09458661e-01 1.35396349e+00 7.43173420e-01
-2.11896040e-02 -3.18936735e-01 7.53110886e-01 -1.75613308e+00
-7.44856000e-02 -1.24435353e+00 -9.04136896e-01 -4.76072282e-02
6.97830260e-01 -6.89744532e-01 -1.34964085e+00 1.86691850e-01] | [7.518667221069336, -1.8004757165908813] |
c2c776f0-039c-4985-8cfa-6a48bcf022ce | one-shot-and-partially-supervised-cell-image | 2304.07991 | null | https://arxiv.org/abs/2304.07991v1 | https://arxiv.org/pdf/2304.07991v1.pdf | One-shot and Partially-Supervised Cell Image Segmentation Using Small Visual Prompt | Semantic segmentation of microscopic cell images using deep learning is an important technique, however, it requires a large number of images and ground truth labels for training. To address the above problem, we consider an efficient learning framework with as little data as possible, and we propose two types of learning strategies: One-shot segmentation which can learn with only one training sample, and Partially-supervised segmentation which assigns annotations to only a part of images. Furthermore, we introduce novel segmentation methods using the small prompt images inspired by prompt learning in recent studies. Our proposed methods use a pre-trained model based on only cell images and teach the information of the prompt pairs to the target image to be segmented by the attention mechanism, which allows for efficient learning while reducing the burden of annotation costs. Through experiments conducted on three types of microscopic cell image datasets, we confirmed that the proposed method improved the Dice score coefficient (DSC) in comparison with the conventional methods. | ['Kazuhiro Hotta', 'Sota Kato'] | 2023-04-17 | null | null | null | null | ['one-shot-segmentation'] | ['computer-vision'] | [ 3.79853338e-01 1.14642151e-01 -8.03424791e-02 -4.66416806e-01
-7.35929966e-01 -3.49109471e-01 1.03699155e-01 4.10634607e-01
-9.90225196e-01 9.69180703e-01 -5.22441030e-01 6.36715963e-02
1.13750279e-01 -6.95199609e-01 -7.25576580e-01 -1.04069316e+00
3.79240423e-01 5.25247097e-01 7.45828807e-01 4.32813466e-01
5.54442883e-01 2.80930817e-01 -9.81993973e-01 1.15739025e-01
1.12390721e+00 1.04601824e+00 7.06300616e-01 4.42169666e-01
-4.86356109e-01 9.70261812e-01 -6.38505876e-01 -1.76440686e-01
1.05848461e-01 -4.91897255e-01 -1.15621030e+00 3.96530062e-01
1.86457783e-02 -4.52061236e-01 1.76078185e-01 1.22714508e+00
4.47147101e-01 1.51435092e-01 6.32315755e-01 -9.54674602e-01
-4.30684090e-01 3.88288945e-01 -6.72239125e-01 3.44212741e-01
-2.31010169e-01 9.87269506e-02 6.27492726e-01 -5.96882105e-01
6.26074910e-01 9.44931924e-01 5.13481796e-01 6.47170961e-01
-1.25143015e+00 -7.45677650e-01 1.99620888e-01 1.95662543e-01
-1.37251019e+00 -2.43332669e-01 7.06765771e-01 -5.67124605e-01
3.78632307e-01 -3.40539515e-01 6.72405839e-01 5.43518126e-01
-1.01749394e-02 8.56396079e-01 1.31921577e+00 -3.62763494e-01
4.55456614e-01 6.43233582e-02 4.62743193e-01 9.43195164e-01
1.63590834e-01 -2.86098003e-01 1.00454479e-01 4.06647623e-02
1.07519543e+00 1.95086166e-01 -2.78119802e-01 -1.61224902e-01
-1.22959733e+00 6.14736795e-01 3.92146677e-01 4.23304319e-01
-7.05328062e-02 -2.22257711e-02 3.61756355e-01 -1.12386502e-01
3.94649297e-01 1.83532923e-01 -6.04077995e-01 1.44088462e-01
-9.27609324e-01 -1.75977170e-01 6.20752752e-01 8.90827298e-01
1.25599253e+00 -3.99291992e-01 -1.05430484e-01 7.22092211e-01
2.73152739e-01 1.68289870e-01 5.34910679e-01 -1.10548544e+00
1.72403291e-01 8.03116143e-01 1.38266444e-01 -8.36121082e-01
-3.60886902e-01 -1.06869087e-01 -7.86155283e-01 -4.28560451e-02
6.91995621e-01 -2.24629551e-01 -1.10448956e+00 1.59681261e+00
4.83141571e-01 4.97342378e-01 -2.61705946e-02 9.03641820e-01
7.85911143e-01 6.24694228e-01 2.03017771e-01 -6.63363338e-01
9.49029803e-01 -1.24070764e+00 -8.84942114e-01 6.44555092e-02
8.42219949e-01 -6.03128195e-01 1.10037696e+00 2.42437959e-01
-6.83213174e-01 -6.54539704e-01 -8.81636679e-01 -1.61601231e-01
-4.82105732e-01 2.55267113e-01 5.15845537e-01 3.49587888e-01
-9.60554481e-01 5.87711453e-01 -9.44824398e-01 -5.10918379e-01
9.12859797e-01 6.60602927e-01 -1.80821881e-01 3.37674022e-02
-6.27118945e-01 2.86533266e-01 6.50945008e-01 -1.10638579e-02
-1.11043692e+00 -4.23527837e-01 -4.63449061e-01 9.79193598e-02
3.01347166e-01 -2.38525495e-01 1.02678311e+00 -1.07412946e+00
-1.63549817e+00 9.89703774e-01 -1.68490380e-01 -1.75308168e-01
3.02429855e-01 -5.68271503e-02 2.48724133e-01 5.72950661e-01
1.57520965e-01 1.20759976e+00 3.05970222e-01 -1.46124911e+00
-7.47604787e-01 -2.12924376e-01 7.82976821e-02 -6.91176206e-03
-3.05519223e-01 -2.33322263e-01 -5.28694451e-01 -3.13943326e-01
1.75067428e-02 -6.69161320e-01 -4.39120352e-01 7.34596476e-02
-4.73807067e-01 -2.26538703e-01 1.03430510e+00 -5.93309999e-01
7.01014280e-01 -1.98689783e+00 -1.10062212e-01 6.08163364e-02
2.69096106e-01 3.75729591e-01 -2.13529289e-01 -4.23750691e-02
2.69610286e-01 1.82621181e-01 -5.08743763e-01 -2.82255232e-01
-4.43629384e-01 3.27180654e-01 7.00937398e-03 4.48644161e-01
3.38857272e-03 7.61365235e-01 -1.07713568e+00 -1.33603287e+00
1.22993745e-01 1.50416076e-01 -4.14687067e-01 5.79166889e-01
-2.67036140e-01 8.58062446e-01 -4.93701667e-01 6.41503096e-01
7.19417691e-01 -6.11807942e-01 4.42986563e-02 -2.60477871e-01
-2.80500180e-03 -4.08977687e-01 -8.62137914e-01 1.67515457e+00
7.71714747e-02 2.20227554e-01 4.66507338e-02 -1.43821657e+00
8.68404388e-01 1.99485391e-01 6.35350287e-01 -5.45263529e-01
4.61885959e-01 4.00275826e-01 -1.27589688e-01 -8.27558875e-01
-2.36470684e-01 -1.08267225e-01 2.14777008e-01 5.94282389e-01
2.45616958e-01 1.16172679e-01 4.04205173e-01 1.43220887e-01
8.41446340e-01 9.11409929e-02 2.00652629e-01 -3.21469873e-01
8.45112145e-01 9.88866612e-02 1.15499115e+00 5.27006626e-01
-5.69930255e-01 4.32035148e-01 4.86710638e-01 -5.32517076e-01
-8.86041522e-01 -6.37496650e-01 -1.15260959e-01 8.29281628e-01
7.68856466e-01 1.31259426e-01 -1.38322866e+00 -9.47073162e-01
-4.79444563e-01 9.80823487e-02 -5.40238798e-01 1.41412616e-01
-5.02821207e-01 -7.79306054e-01 3.32812905e-01 4.65454221e-01
8.09259474e-01 -1.12666249e+00 -5.53641021e-01 3.54088359e-02
-1.85335994e-01 -1.27574122e+00 -4.83442813e-01 1.99962318e-01
-1.14389122e+00 -1.36575353e+00 -9.27264810e-01 -1.36636269e+00
1.40936148e+00 3.29844564e-01 5.27373612e-01 4.72815126e-01
-1.97003245e-01 1.81891724e-01 -3.18458706e-01 -2.18717620e-01
-4.63923141e-02 1.88927025e-01 -1.94208726e-01 1.73784807e-01
3.54120970e-01 -3.58220816e-01 -7.12729812e-01 2.46563122e-01
-9.98758197e-01 3.27097088e-01 6.89605057e-01 1.08617151e+00
1.08011830e+00 -1.18543664e-02 7.40311384e-01 -1.29219258e+00
2.07787260e-01 -1.52579069e-01 -7.62483001e-01 2.93701530e-01
-5.00547767e-01 -2.98750460e-01 8.79355788e-01 -3.97334069e-01
-1.18605661e+00 3.49846363e-01 -3.59064825e-02 -2.46494427e-01
-3.45288098e-01 1.94636688e-01 -1.05498552e-01 -4.47280884e-01
1.57173917e-01 3.44682813e-01 2.37191960e-01 -4.27235276e-01
5.18451678e-03 6.94814384e-01 4.85205770e-01 -5.24071932e-01
2.79977083e-01 7.58357286e-01 -6.32157028e-02 -6.25733733e-01
-1.00630808e+00 -5.44665039e-01 -1.09334183e+00 -1.73978791e-01
1.08968616e+00 -6.20630860e-01 -8.20527017e-01 8.74764264e-01
-1.29749119e+00 -6.74341083e-01 1.13072265e-02 4.17124659e-01
-7.30802357e-01 5.97120643e-01 -9.87393796e-01 -4.60090905e-01
-3.33516270e-01 -1.19944823e+00 9.37153637e-01 6.57636106e-01
2.33477965e-01 -1.22460163e+00 7.80445933e-02 4.05221134e-01
1.01354457e-01 7.25139827e-02 1.18505776e+00 -9.21144664e-01
-8.64278495e-01 4.07094397e-02 -5.46135366e-01 3.12460691e-01
4.04085308e-01 -1.28512144e-01 -9.49354649e-01 -4.32982057e-01
9.56600308e-02 -6.91727221e-01 8.09162855e-01 4.64709014e-01
1.64135408e+00 -1.63286924e-01 -7.61022925e-01 6.31301522e-01
1.59509146e+00 4.70440328e-01 4.90258366e-01 1.59227420e-02
8.35597873e-01 6.89766467e-01 7.94155359e-01 7.14783072e-02
2.78309643e-01 2.80390769e-01 2.54518181e-01 -5.31739295e-01
-2.65743062e-02 -1.00323014e-01 -2.02103630e-01 1.08334088e+00
6.02572821e-02 -2.02631816e-01 -6.79351389e-01 5.90789318e-01
-1.97489822e+00 -6.18049204e-01 8.04537386e-02 1.95727825e+00
1.12345183e+00 2.23193727e-02 -9.65907350e-02 7.12834969e-02
8.83119762e-01 -1.96312085e-01 -7.53605485e-01 2.06416905e-01
1.33584201e-01 -5.73569797e-02 3.39715451e-01 5.05082726e-01
-1.07179606e+00 1.21706617e+00 6.55892134e+00 1.08756554e+00
-1.24229825e+00 1.90199509e-01 1.21005511e+00 2.70733267e-01
-6.50705351e-03 1.52708292e-01 -7.32238531e-01 8.68479788e-01
4.77820873e-01 3.72326076e-02 1.68429896e-01 6.10051811e-01
2.65464246e-01 -3.43432844e-01 -1.10422790e+00 9.64333355e-01
-3.30590531e-02 -1.42956185e+00 8.71006995e-02 -2.55324505e-02
8.77213717e-01 -2.06499964e-01 -4.29796219e-01 1.09435759e-01
1.57788858e-01 -8.27355385e-01 1.17101103e-01 6.94097042e-01
5.93797565e-01 -5.82472444e-01 1.10493028e+00 5.69640160e-01
-9.67242837e-01 2.12835204e-02 -6.40376508e-01 9.29825827e-02
5.89833148e-02 7.44177938e-01 -6.85220957e-01 1.59044459e-01
5.17843187e-01 7.62968838e-01 -5.46698570e-01 1.29632545e+00
4.66918349e-02 7.70331323e-01 -1.26908496e-01 -5.88471908e-03
2.18861669e-01 -4.12672549e-01 -5.15869819e-02 1.05552542e+00
1.90291718e-01 3.56453925e-01 6.14984572e-01 8.75899732e-01
-1.94415808e-01 2.66718596e-01 -1.68865398e-01 1.41240731e-02
5.10030270e-01 1.57005250e+00 -1.51537728e+00 -5.78436494e-01
-3.18388760e-01 9.47843492e-01 6.12078190e-01 5.18041372e-01
-6.52548909e-01 -6.79061472e-01 -1.97452769e-01 5.03830798e-02
1.75932512e-01 1.27421021e-01 -2.57459581e-01 -7.49417603e-01
-4.94445324e-01 -3.15870017e-01 2.33565181e-01 -6.36415660e-01
-1.21068168e+00 1.19027264e-01 -2.49714598e-01 -9.32377338e-01
2.88956910e-01 -6.15988910e-01 -6.58730626e-01 4.99331921e-01
-1.71996582e+00 -9.99468863e-01 -5.57035267e-01 4.28327113e-01
5.78523636e-01 8.04130360e-02 6.12686574e-01 4.16813105e-01
-7.97076762e-01 3.63552272e-01 9.95919555e-02 4.00122076e-01
6.70385063e-01 -1.21140027e+00 -3.02919656e-01 5.73841691e-01
-1.24401860e-01 5.07335126e-01 1.15257226e-01 -5.43384612e-01
-7.19083905e-01 -1.07689238e+00 7.38193512e-01 1.15371257e-01
1.46228194e-01 -9.01874751e-02 -1.12033856e+00 5.94274700e-01
2.90653735e-01 3.20848912e-01 9.73701417e-01 -5.23948491e-01
3.28377694e-01 -2.66858190e-01 -1.24502873e+00 4.08747345e-01
7.45449185e-01 -2.20673010e-01 -3.17405105e-01 4.76742536e-01
8.77817869e-01 -3.46322387e-01 -7.46018410e-01 2.78454453e-01
1.64083481e-01 -7.75127113e-01 5.47244906e-01 -1.68343514e-01
2.63496995e-01 -6.16890907e-01 2.71953285e-01 -9.52587962e-01
-3.19080740e-01 -3.85414243e-01 1.95644766e-01 1.48418093e+00
1.81385890e-01 -5.04111409e-01 9.63111341e-01 3.16104293e-01
-2.91239619e-01 -1.06857085e+00 -6.43852413e-01 -5.33751607e-01
1.38018861e-01 3.87791395e-01 3.83491039e-01 9.67786133e-01
-2.52266861e-02 2.85303921e-01 -6.42693117e-02 -1.26858786e-01
8.08216333e-01 2.12516338e-01 7.39603996e-01 -1.29733610e+00
-1.04415286e-02 -1.69404849e-01 -3.57466400e-01 -1.35458112e+00
3.03735524e-01 -8.47465098e-01 4.54853773e-01 -1.53353775e+00
7.81275332e-01 -5.33618450e-01 -5.05816042e-01 5.74092090e-01
-3.88538867e-01 2.10259676e-01 -2.05796868e-01 4.95200187e-01
-1.11550820e+00 3.83470327e-01 1.66187811e+00 -2.26203069e-01
-1.42306881e-03 -3.14616263e-01 -6.12183332e-01 8.58419359e-01
6.96783781e-01 -5.45912862e-01 -5.96178949e-01 -5.21553755e-01
-3.53503764e-01 -3.89322862e-02 1.10177264e-01 -1.09261775e+00
6.76265895e-01 -3.39316338e-01 5.14346540e-01 -5.91121852e-01
3.85005139e-02 -7.84967244e-01 -2.84727156e-01 6.43129230e-01
-5.28140664e-01 -5.74881673e-01 -7.37722814e-02 7.30816782e-01
-1.74117997e-01 -5.24907887e-01 1.17930377e+00 -4.27897781e-01
-6.80792093e-01 5.04214704e-01 -2.53160238e-01 4.34972271e-02
1.34101713e+00 -3.61990839e-01 -3.52467686e-01 8.78934115e-02
-5.29810488e-01 5.42214453e-01 6.62930012e-01 -3.22079092e-01
4.89538938e-01 -1.17207694e+00 -9.50042307e-02 4.15880866e-02
-1.18426301e-01 5.38911998e-01 3.13452899e-01 9.44236338e-01
-7.32931495e-01 4.38414246e-01 -3.31477135e-01 -8.28991294e-01
-1.01845872e+00 6.49661422e-01 2.89097011e-01 -3.52406740e-01
-3.71809006e-01 9.35395479e-01 5.70071101e-01 -5.81820011e-01
1.98021784e-01 -2.66867191e-01 -3.97651941e-01 -1.45389199e-01
2.97496140e-01 2.90275574e-01 -3.63648504e-01 -5.20269871e-01
-2.26840109e-01 9.71462607e-01 -4.22041327e-01 2.79270709e-01
1.01772094e+00 -3.31741363e-01 -3.60634178e-01 4.88337368e-01
1.28058600e+00 -3.94881964e-01 -1.73442435e+00 -2.59481043e-01
1.13592267e-01 -3.04435432e-01 -4.06855866e-02 -5.82804024e-01
-1.35444760e+00 1.09040868e+00 5.96566558e-01 -1.65237188e-02
1.10042775e+00 -1.45989224e-01 1.30357146e+00 4.50514525e-01
4.67214197e-01 -1.33984435e+00 4.87519532e-01 2.70802736e-01
-4.71000522e-02 -1.53840566e+00 -8.44241381e-02 -6.16563737e-01
-3.65511805e-01 9.80526090e-01 9.65227723e-01 -8.93274918e-02
6.34604692e-01 1.49595991e-01 3.10967773e-01 -1.31823286e-01
-5.56221902e-01 -1.57739729e-01 -3.71682703e-01 6.38070107e-01
3.34776580e-01 -2.98763305e-01 -5.42542219e-01 8.24432373e-01
4.44992393e-01 4.99228835e-01 4.31215882e-01 9.15485382e-01
-8.55097234e-01 -1.02739358e+00 1.02104759e-02 5.75006783e-01
-6.34200156e-01 1.98554665e-01 -3.46576273e-01 4.47637379e-01
4.77320194e-01 8.63905489e-01 2.10427746e-01 -1.99240088e-01
-2.57881999e-01 -2.13360220e-01 3.86644810e-01 -7.68598258e-01
-1.15974464e-01 2.73043364e-01 -6.10353649e-01 -3.79019678e-01
-8.79955947e-01 -1.84451789e-01 -1.90747166e+00 1.17606975e-01
-5.23544371e-01 3.76136363e-01 7.42496103e-02 1.29815447e+00
2.86615133e-01 4.64363933e-01 6.65801048e-01 -8.51586103e-01
-1.25264972e-01 -7.81331539e-01 -7.11115360e-01 4.66580272e-01
1.25845850e-01 -5.79818785e-01 -2.54711121e-01 6.43250704e-01] | [14.651575088500977, -2.9231162071228027] |
a985f5ae-8863-4e0a-aaef-fe509440198d | automatic-relation-aware-graph-network | 2205.15678 | null | https://arxiv.org/abs/2205.15678v1 | https://arxiv.org/pdf/2205.15678v1.pdf | Automatic Relation-aware Graph Network Proliferation | Graph neural architecture search has sparked much attention as Graph Neural Networks (GNNs) have shown powerful reasoning capability in many relational tasks. However, the currently used graph search space overemphasizes learning node features and neglects mining hierarchical relational information. Moreover, due to diverse mechanisms in the message passing, the graph search space is much larger than that of CNNs. This hinders the straightforward application of classical search strategies for exploring complicated graph search space. We propose Automatic Relation-aware Graph Network Proliferation (ARGNP) for efficiently searching GNNs with a relation-guided message passing mechanism. Specifically, we first devise a novel dual relation-aware graph search space that comprises both node and relation learning operations. These operations can extract hierarchical node/relational information and provide anisotropic guidance for message passing on a graph. Second, analogous to cell proliferation, we design a network proliferation search paradigm to progressively determine the GNN architectures by iteratively performing network division and differentiation. The experiments on six datasets for four graph learning tasks demonstrate that GNNs produced by our method are superior to the current state-of-the-art hand-crafted and search-based GNNs. Codes are available at https://github.com/phython96/ARGNP. | ['Qingming Huang', 'Zheng-Jun Zha', 'Jiebo Luo', 'Xinzhe Han', 'Liang Li', 'Shaofei Cai'] | 2022-05-31 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Cai_Automatic_Relation-Aware_Graph_Network_Proliferation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Cai_Automatic_Relation-Aware_Graph_Network_Proliferation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['graph-regression'] | ['graphs'] | [ 4.80639264e-02 2.54471600e-01 -5.08740067e-01 4.99965101e-02
-1.65846407e-01 -4.59777594e-01 6.77450418e-01 3.47623646e-01
-2.67794847e-01 5.98972201e-01 -2.32368290e-01 -7.08238006e-01
-5.89769721e-01 -1.40865803e+00 -5.69191933e-01 -6.69671535e-01
-2.31817067e-01 6.60771668e-01 4.01869655e-01 -2.69353598e-01
2.74921954e-01 5.82914412e-01 -1.34186792e+00 -3.01502980e-02
7.73643017e-01 9.35943961e-01 2.39669964e-01 5.91590822e-01
-3.68215591e-01 8.38697314e-01 -3.13261777e-01 -3.23644817e-01
1.70886815e-01 -4.69041973e-01 -1.02066660e+00 -2.46222079e-01
2.06359774e-01 1.58251494e-01 -9.67374742e-01 1.01280260e+00
4.58004504e-01 1.15588024e-01 4.79719996e-01 -1.36836839e+00
-5.86670339e-01 1.04197431e+00 -4.44349289e-01 6.73507452e-01
7.84287006e-02 1.97386071e-01 1.59789324e+00 -5.49363077e-01
8.23124170e-01 9.16326284e-01 4.51348454e-01 5.59479177e-01
-1.16709387e+00 -7.16665328e-01 2.38317654e-01 1.76828653e-01
-1.52126026e+00 -1.24506310e-01 9.21785235e-01 -2.88320899e-01
1.30202770e+00 2.17541441e-01 1.30476284e+00 9.86346126e-01
1.56938881e-01 6.55771971e-01 5.36046743e-01 -1.86919585e-01
1.38110995e-01 -4.90989983e-01 1.35936365e-01 1.39251888e+00
4.20905948e-01 2.25760881e-03 -7.04512835e-01 3.02464571e-02
1.15732658e+00 1.84185833e-01 -2.99225390e-01 -2.49363929e-01
-1.07655585e+00 9.14648056e-01 9.97223258e-01 4.76144791e-01
-1.60504282e-01 5.99189162e-01 4.66860265e-01 3.44702423e-01
1.88470423e-01 6.81847572e-01 -4.73361671e-01 3.22190255e-01
-7.92461574e-01 -5.43736219e-02 1.00976884e+00 7.60649443e-01
9.66951251e-01 -7.01923668e-02 -1.38621509e-01 6.41817570e-01
5.18391669e-01 -1.43161997e-01 4.45547909e-01 -5.96201599e-01
3.60087603e-01 1.49500430e+00 -9.62170482e-01 -1.31690240e+00
-6.67951584e-01 -8.03659856e-01 -1.31316185e+00 -2.10397601e-01
4.32212830e-01 2.75291264e-01 -1.02805221e+00 1.54246879e+00
2.94392079e-01 1.32708818e-01 -1.23806328e-01 6.82542026e-01
1.40629184e+00 3.51103187e-01 -5.45803346e-02 1.15733426e-02
1.48450446e+00 -1.08290017e+00 -3.40648055e-01 -3.03764760e-01
9.72154498e-01 -2.11567935e-02 8.64740491e-01 2.80804574e-01
-9.58877802e-01 -1.18374750e-01 -1.04113114e+00 -1.57769769e-01
-6.43646598e-01 -3.67520094e-01 1.22577500e+00 1.98954195e-01
-1.47160578e+00 6.55539572e-01 -6.61217034e-01 -4.85263199e-01
8.55425239e-01 6.55027032e-01 -2.61512041e-01 9.27747190e-02
-1.23144305e+00 6.09380603e-01 6.28597379e-01 1.55738026e-01
-1.02890038e+00 -4.41138178e-01 -8.55402470e-01 3.97670805e-01
7.03965187e-01 -1.13619542e+00 9.32478011e-01 -1.01034395e-01
-1.18557954e+00 1.04389465e+00 7.98672140e-02 -4.50861096e-01
-5.34015149e-03 5.46153903e-01 -5.00786975e-02 4.25234377e-01
-1.86508641e-01 7.98414290e-01 3.30354661e-01 -8.77583623e-01
-3.90073001e-01 -3.88271332e-01 1.61119580e-01 2.34907940e-01
-4.80638534e-01 -4.26914483e-01 -8.93457294e-01 -5.75712681e-01
6.18011713e-01 -6.91376209e-01 -4.30865526e-01 8.86134654e-02
-7.06070662e-01 -4.66128200e-01 4.36345428e-01 -2.94458605e-02
1.60150087e+00 -1.62289095e+00 3.27208042e-01 6.54068291e-01
1.11989594e+00 6.39224201e-02 -2.02675313e-01 5.79212308e-01
1.10893965e-01 4.64706063e-01 -2.86114309e-02 9.41558704e-02
-1.28805369e-01 3.41839790e-01 4.13706936e-02 3.81920278e-01
-3.71935144e-02 1.41998291e+00 -1.03002381e+00 -7.84282565e-01
-2.48102814e-01 1.05913401e-01 -5.08470178e-01 -2.23428771e-01
-4.39830244e-01 1.65127605e-01 -6.12801433e-01 1.01649785e+00
9.20169652e-02 -1.11621439e+00 3.96130204e-01 -5.84627241e-02
3.15070421e-01 3.73799652e-01 -6.53451443e-01 1.63240647e+00
-3.57625857e-02 5.60073376e-01 -2.20457315e-02 -1.19929385e+00
1.06739295e+00 5.74638657e-02 4.62101549e-01 -7.51242697e-01
2.77450800e-01 2.66379893e-01 2.25203514e-01 -9.49156135e-02
2.54904591e-02 3.26220095e-01 6.94584548e-02 4.88887250e-01
3.06598395e-01 -9.76280347e-02 6.45874858e-01 7.12866008e-01
1.84021735e+00 -4.67991322e-01 2.82716990e-01 -3.39419812e-01
5.91758966e-01 -1.22544095e-02 4.26322967e-01 7.96159625e-01
-7.95480087e-02 2.39686266e-01 8.89966726e-01 -5.14614642e-01
-5.54385424e-01 -8.58949602e-01 3.49235356e-01 1.19839799e+00
3.02242845e-01 -8.90885890e-01 -4.91337478e-01 -7.92560458e-01
-5.39483353e-02 -2.22599804e-02 -7.85363436e-01 -4.36262131e-01
-6.12190068e-01 -8.05719614e-01 9.02353287e-01 3.30968022e-01
5.03122389e-01 -1.11950755e+00 -2.86634356e-01 6.01470061e-02
1.39618412e-01 -8.29169631e-01 -3.17586720e-01 4.35744286e-01
-1.14440978e+00 -1.38257325e+00 -1.74346268e-01 -1.20838785e+00
9.61753249e-01 9.58702788e-02 1.26294696e+00 9.76567030e-01
-4.70313668e-01 2.63937294e-01 -1.90394968e-01 -7.12571386e-03
-2.12818384e-01 6.52248085e-01 -3.91962707e-01 -4.36620831e-01
1.82854906e-01 -9.70939755e-01 -7.71881580e-01 1.59419477e-01
-7.97237754e-01 3.17123502e-01 8.54156733e-01 8.50202203e-01
7.85263002e-01 3.81137311e-01 4.90235507e-01 -1.06937945e+00
8.28341305e-01 -5.43676078e-01 -6.84063733e-01 3.19468617e-01
-1.06165767e+00 3.24851662e-01 4.27501857e-01 -2.82436877e-01
-3.72452736e-01 -1.09363876e-01 -1.45394076e-03 -2.04580575e-01
1.74228400e-01 9.40097809e-01 1.05966143e-01 -4.23830360e-01
7.02690303e-01 3.65558028e-01 2.16149271e-01 -1.36788189e-01
1.32379234e-01 -1.29496157e-01 4.17586088e-01 -5.06719470e-01
7.10860133e-01 3.67130339e-01 6.18082762e-01 -4.61763322e-01
-5.08363366e-01 -3.76427919e-01 -3.22084039e-01 -2.26852626e-01
6.09055758e-01 -4.99269754e-01 -1.25595713e+00 3.70690256e-01
-8.93248022e-01 -6.02013707e-01 -9.99820679e-02 3.11945267e-02
-3.45107853e-01 2.37182200e-01 -1.07619846e+00 -3.75832826e-01
-7.10999727e-01 -9.36300218e-01 7.51093388e-01 1.66379407e-01
-9.66578126e-02 -1.23483503e+00 8.18578005e-02 1.33315414e-01
3.07602286e-01 1.72528788e-01 1.35843849e+00 -9.89926457e-01
-1.10218906e+00 1.56552568e-02 -3.90494645e-01 -5.49621582e-01
-7.21779838e-02 -3.78876477e-02 -4.50230479e-01 -1.50107041e-01
-6.10508204e-01 -3.13605458e-01 1.24340868e+00 1.67389229e-01
1.25977087e+00 -5.25693834e-01 -9.04269338e-01 8.90362501e-01
1.42684305e+00 2.07667872e-02 4.46695566e-01 4.59066957e-01
8.66151690e-01 3.91510159e-01 -1.32425904e-01 8.72760266e-02
5.72007895e-01 1.77661091e-01 7.35905766e-01 -1.84623078e-01
-2.37135619e-01 -4.04389381e-01 -2.00948566e-01 8.91189873e-01
-9.32562947e-02 -5.97182274e-01 -1.35706091e+00 4.26263630e-01
-2.02382374e+00 -5.63391924e-01 -5.78658246e-02 1.48675227e+00
9.49363708e-01 2.25548834e-01 -1.28291160e-01 2.01014608e-01
6.34054244e-01 1.24331027e-01 -8.32193255e-01 5.46010137e-02
-1.21984661e-01 1.90294117e-01 3.85029316e-01 3.11329216e-01
-6.48418307e-01 1.20409060e+00 5.72769165e+00 1.03581882e+00
-7.70828962e-01 -1.43535450e-01 7.47072220e-01 1.06371790e-01
-6.55305743e-01 4.22685668e-02 -6.92587018e-01 -3.04403640e-02
5.66411018e-01 -3.16898614e-01 6.92415953e-01 5.57032108e-01
-3.37915838e-01 1.41938441e-02 -1.18393826e+00 1.14765584e+00
-3.19343418e-01 -1.87315392e+00 3.32108051e-01 3.19126457e-01
5.38148820e-01 2.59953678e-01 -2.68910527e-01 3.94105762e-01
6.52512252e-01 -1.28466392e+00 1.89770982e-01 3.77451569e-01
6.24256790e-01 -4.13773477e-01 4.08023566e-01 4.20666933e-01
-1.45294261e+00 -1.12364337e-01 -2.10044622e-01 -2.50023790e-03
-1.55866444e-01 6.53985500e-01 -9.97028351e-01 5.07214546e-01
5.67312002e-01 5.84553897e-01 -8.77573729e-01 1.07287741e+00
-3.13986242e-01 5.44370949e-01 -4.15762424e-01 -5.51248789e-01
2.95559913e-01 -1.25920326e-01 5.18065453e-01 1.02713633e+00
2.65076220e-01 1.46465123e-01 2.61978172e-02 1.13307869e+00
-4.89126891e-01 2.23204400e-02 -7.36499667e-01 -6.11499310e-01
6.20061874e-01 1.46128225e+00 -1.46069384e+00 -9.03577581e-02
-1.20019615e-01 5.10624409e-01 9.21455681e-01 4.83444810e-01
-5.12148142e-01 -3.94183159e-01 1.86313510e-01 1.16492003e-01
2.49645337e-01 -1.18175656e-01 -3.07164013e-01 -8.36155176e-01
-4.28865075e-01 -6.38605297e-01 8.85760069e-01 -3.42279792e-01
-1.22623527e+00 7.16835499e-01 -2.46418625e-01 -5.95580816e-01
7.43566975e-02 -5.48036993e-01 -6.86887860e-01 4.00125325e-01
-1.24546361e+00 -1.02240777e+00 -4.28927362e-01 6.64292574e-01
3.41807216e-01 -3.79191935e-01 6.43742204e-01 8.55621025e-02
-6.92832232e-01 5.49704492e-01 -7.17976570e-01 2.51548350e-01
-1.03963882e-01 -1.15963376e+00 5.10970354e-01 6.16494954e-01
4.09241289e-01 8.79885077e-01 2.34904394e-01 -8.31944942e-01
-1.77356064e+00 -8.85087609e-01 8.09019685e-01 -7.46658593e-02
9.19689059e-01 -4.93197709e-01 -8.45342875e-01 5.50720215e-01
-3.99310291e-02 1.70448542e-01 5.57879567e-01 2.70451903e-01
-3.77294809e-01 -2.05487579e-01 -7.72433102e-01 8.62133622e-01
1.69558501e+00 -4.39453959e-01 -5.41749448e-02 3.07363570e-01
9.69498158e-01 -4.24609363e-01 -8.08124423e-01 5.93754292e-01
3.62267315e-01 -6.97771370e-01 9.35334921e-01 -5.02289414e-01
2.37322658e-01 -2.94441253e-01 1.79079562e-01 -9.52145457e-01
-5.84198177e-01 -6.40525699e-01 -6.81331456e-01 8.49018931e-01
8.64780664e-01 -9.25860107e-01 1.23325968e+00 2.81645834e-01
-1.51771888e-01 -1.44686258e+00 -6.73243284e-01 -5.30010223e-01
-2.67560184e-01 -1.01735920e-01 5.06066859e-01 9.01512742e-01
1.40995041e-01 9.17211115e-01 2.53525108e-01 -4.78528515e-02
4.47088420e-01 1.20439261e-01 5.58669627e-01 -1.48598647e+00
-4.36403394e-01 -1.17261171e+00 -5.28330803e-01 -1.06712842e+00
2.30941519e-01 -1.62558460e+00 -2.27699429e-01 -2.02503538e+00
1.92102000e-01 -6.21464670e-01 -3.97477686e-01 8.78095508e-01
3.83286104e-02 1.42303824e-01 -1.71153054e-01 2.19409794e-01
-8.91002834e-01 4.51560795e-01 1.53221178e+00 -2.92095393e-01
-9.93574858e-02 -2.13649839e-01 -1.00280666e+00 5.92292249e-01
1.06517196e+00 -3.70646954e-01 -8.24807405e-01 -2.90176928e-01
1.08073497e+00 2.86590233e-02 3.80902827e-01 -7.89732814e-01
1.00576544e+00 -7.30490759e-02 1.81359380e-01 -6.38059974e-01
6.12958819e-02 -3.92617345e-01 2.27105230e-01 6.96219683e-01
-4.28465188e-01 -1.53361782e-02 1.51212495e-02 7.19012320e-01
-9.63401124e-02 -1.41826272e-03 2.95135617e-01 -4.59884882e-01
-5.99865317e-01 8.43666375e-01 -2.38796934e-01 1.21594563e-01
7.00253069e-01 -5.10836780e-01 -7.73832440e-01 -1.68287456e-01
-6.08548999e-01 6.33752227e-01 1.66959643e-01 2.91161053e-03
8.50690961e-01 -1.11369431e+00 -4.22742933e-01 1.69793889e-01
1.29705414e-01 4.73320931e-01 4.35867794e-02 9.94049549e-01
-7.36263275e-01 3.85243446e-01 1.05677232e-01 -4.67503786e-01
-1.06178880e+00 6.92065120e-01 2.90903807e-01 -7.54106164e-01
-6.29133642e-01 1.45117259e+00 4.48717773e-01 -5.94244778e-01
3.77316177e-01 -2.60207236e-01 -3.70615602e-01 -1.95922896e-01
-5.63363507e-02 4.20128047e-01 3.21223736e-02 -4.30382676e-02
-3.35206211e-01 2.40927011e-01 -1.60206467e-01 3.17081898e-01
1.28346491e+00 1.12381741e-01 -7.01185346e-01 1.56155229e-01
1.01328444e+00 -4.00452256e-01 -8.05225790e-01 -3.22434276e-01
2.55978405e-01 2.14154571e-02 2.54617371e-02 -6.04993343e-01
-1.45608127e+00 5.44433713e-01 -1.69301122e-01 4.72554028e-01
1.21982670e+00 5.22931397e-01 7.30395198e-01 7.79890895e-01
2.73506969e-01 -7.96807647e-01 3.58788401e-01 6.68980956e-01
7.86553860e-01 -9.38426375e-01 8.20919871e-02 -7.85437465e-01
-1.24694770e-02 1.09093940e+00 1.04129815e+00 -5.31974249e-03
7.97761023e-01 2.08748370e-01 -4.58200961e-01 -1.13085151e+00
-1.18556929e+00 -3.93197834e-01 3.29264522e-01 3.55335116e-01
1.15251310e-01 -1.82791620e-01 -1.82396919e-01 6.24791801e-01
-3.22388023e-01 -2.40451127e-01 2.73317069e-01 6.44566119e-01
-5.57239413e-01 -9.45710838e-01 2.51534581e-01 8.35669279e-01
-2.07267746e-01 -4.86581087e-01 -7.33563125e-01 7.83716440e-01
-1.85283348e-01 5.64659834e-01 -1.06120937e-01 -5.84254980e-01
-1.02245763e-01 -2.69243121e-01 5.13149500e-01 -7.16095984e-01
-4.98238266e-01 -1.06723025e-01 1.60029277e-01 -5.67717791e-01
-2.16055270e-02 -1.61979899e-01 -1.72180009e+00 -4.14219916e-01
-5.21353185e-01 1.47637397e-01 2.94133484e-01 6.65178657e-01
4.94926155e-01 7.20485151e-01 1.02160156e-01 -4.93906856e-01
-6.70526773e-02 -6.70581698e-01 -5.82308054e-01 -8.63145813e-02
1.39478534e-01 -6.52470231e-01 -3.66116703e-01 -5.26584506e-01] | [7.014630317687988, 6.157727241516113] |
c8f425b5-1bb3-4120-bb0d-112af9d9cb07 | review-learning-alleviating-catastrophic | 2210.09394 | null | https://arxiv.org/abs/2210.09394v1 | https://arxiv.org/pdf/2210.09394v1.pdf | Review Learning: Alleviating Catastrophic Forgetting with Generative Replay without Generator | When a deep learning model is sequentially trained on different datasets, it forgets the knowledge acquired from previous data, a phenomenon known as catastrophic forgetting. It deteriorates performance of the deep learning model on diverse datasets, which is critical in privacy-preserving deep learning (PPDL) applications based on transfer learning (TL). To overcome this, we propose review learning (RL), a generative-replay-based continual learning technique that does not require a separate generator. Data samples are generated from the memory stored within the synaptic weights of the deep learning model which are used to review knowledge acquired from previous datasets. The performance of RL was validated through PPDL experiments. Simulations and real-world medical multi-institutional experiments were conducted using three types of binary classification electronic health record data. In the real-world experiments, the global area under the receiver operating curve was 0.710 for RL and 0.655 for TL. Thus, RL was highly effective in retaining previously learned knowledge. | ['Kwangsoo Kim', 'Hyeong-Jin Yoon', 'Rae Woong Park', 'Dae Jung Kim', 'Hyung Joon Joo', 'Yaeji Lim', 'Dongkyeong Lim', 'Jieun Choi', 'Suhyeon Kim', 'Ye Seul Yang', 'Sunghyuk Choi', 'Jaesung Yoo'] | 2022-10-17 | null | null | null | null | ['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning'] | ['methodology', 'natural-language-processing'] | [ 3.26310508e-02 1.11736089e-01 -1.57072231e-01 -3.10140699e-01
-3.72280598e-01 -3.31894875e-01 1.81205183e-01 3.84691983e-01
-8.02188277e-01 1.55920982e+00 -7.26639852e-02 -1.14367507e-01
-7.66589120e-02 -8.45147669e-01 -1.03765643e+00 -8.58728528e-01
-9.08904672e-02 2.36178096e-02 2.27396488e-01 3.49942535e-01
4.38424684e-02 4.23789233e-01 -1.03330743e+00 4.78102237e-01
7.25985527e-01 7.85958767e-01 1.01436622e-01 3.46376389e-01
-4.27705087e-02 7.46061146e-01 -9.75852907e-01 -3.96288306e-01
5.00403307e-02 -5.08973956e-01 -5.16090453e-01 -5.70659220e-01
2.62510572e-02 -3.81923229e-01 -6.27323747e-01 9.85688925e-01
8.12960207e-01 -5.84317707e-02 4.10514295e-01 -1.16479278e+00
-9.72254455e-01 4.60075498e-01 -2.52561599e-01 4.95880365e-01
-2.19402388e-01 6.37182072e-02 2.51529992e-01 -8.81673396e-01
5.90466738e-01 8.27075779e-01 9.38608468e-01 1.00420260e+00
-1.17356682e+00 -1.14139414e+00 -1.86828241e-01 9.46695134e-02
-1.47615945e+00 -3.75376523e-01 4.93012369e-01 -2.01006263e-01
7.39006102e-01 -1.44205213e-01 8.68369401e-01 1.16148531e+00
1.37844884e+00 6.96998239e-01 1.03739727e+00 -1.15680471e-01
5.51176727e-01 5.35102785e-01 8.68319124e-02 6.31796479e-01
8.34947467e-01 1.97393402e-01 -1.02785361e+00 -5.35781085e-01
6.86472058e-01 4.93617356e-01 -4.91428256e-01 -4.10012186e-01
-7.42992878e-01 6.12779081e-01 3.63798380e-01 1.77438527e-01
-3.65194201e-01 7.56149963e-02 3.01970601e-01 7.53833115e-01
3.54907900e-01 6.07205108e-02 -6.36595905e-01 2.61881739e-01
-7.96195388e-01 -1.41961560e-01 7.02624917e-01 8.09135973e-01
6.93480372e-01 1.13349937e-01 -1.57614455e-01 4.36686993e-01
1.61569193e-01 3.76709163e-01 1.12397718e+00 -2.29011729e-01
-2.04287097e-02 5.39398253e-01 -2.19149441e-01 -7.82132447e-01
-3.71350586e-01 -9.03049588e-01 -1.06868315e+00 -3.22186463e-02
-2.24275049e-02 -3.40161681e-01 -9.93798077e-01 1.98939347e+00
-9.39861163e-02 4.36955869e-01 4.40639973e-01 3.07950020e-01
9.48995471e-01 4.26472753e-01 2.60367304e-01 -5.53236306e-01
7.82095134e-01 -4.42112237e-01 -7.58580148e-01 5.76715656e-02
5.45425892e-01 -3.30146030e-02 9.49125826e-01 3.64929885e-01
-8.03458631e-01 -5.21740019e-01 -1.32783043e+00 1.62170112e-01
-4.07605916e-01 -1.48765475e-01 3.95319760e-01 6.41482711e-01
-1.26933932e+00 5.90084374e-01 -7.56863892e-01 -2.14693338e-01
1.09070289e+00 5.78661084e-01 -4.36546475e-01 -2.14832872e-01
-1.48922825e+00 6.12638831e-01 3.76115888e-01 -2.07054064e-01
-1.22674584e+00 -9.63050544e-01 -2.90969908e-01 2.88563162e-01
-1.15312226e-01 -8.71860921e-01 8.31836045e-01 -9.52904403e-01
-1.23842752e+00 8.53664279e-01 4.98976186e-02 -9.08373237e-01
5.69984615e-01 -3.09576303e-01 -6.10728920e-01 -1.49980530e-01
-1.43841773e-01 5.28447747e-01 9.47343826e-01 -1.12568986e+00
-2.97637552e-01 -5.19096971e-01 -6.91417754e-01 -5.23027666e-02
-5.45045853e-01 -7.55633652e-01 -4.54214700e-02 -6.54792249e-01
-1.95511162e-01 -9.72695351e-01 8.85467753e-02 1.10767305e-01
-2.10879389e-02 1.43830240e-01 7.57253528e-01 -6.16267979e-01
1.34215224e+00 -2.48024631e+00 -3.40592653e-01 1.43048227e-01
2.89900661e-01 8.07270646e-01 4.27458659e-02 4.28793490e-01
-1.63541045e-02 5.43611720e-02 -3.44288051e-01 -1.28781050e-01
-7.29735792e-01 2.52868265e-01 -5.15429854e-01 4.62105811e-01
-1.61597922e-01 1.18958127e+00 -8.49370897e-01 -1.17169976e-01
-2.93666154e-01 4.89469200e-01 -2.80681908e-01 2.56618056e-02
-1.07418243e-02 5.17870545e-01 -2.89211869e-01 4.17155355e-01
8.27552736e-01 -3.93999606e-01 2.10057367e-02 2.88194150e-01
3.79390150e-01 -2.74319708e-01 -4.13569719e-01 1.86133730e+00
-2.33329646e-02 6.03291273e-01 -6.07140958e-01 -5.67120254e-01
1.23647809e+00 2.87316769e-01 2.40751266e-01 -8.60246539e-01
1.54098034e-01 6.05220422e-02 -7.54343569e-02 -2.53382385e-01
7.07533881e-02 -2.13323265e-01 -6.33747259e-04 5.85523963e-01
2.40642920e-01 4.50058818e-01 -6.47444308e-01 2.37153202e-01
1.40419233e+00 -5.46196818e-01 4.10682768e-01 -3.56237799e-01
4.21310574e-01 -3.22641134e-01 8.00180137e-01 1.05799282e+00
-3.64099801e-01 3.83751005e-01 2.48481557e-01 -8.24438870e-01
-6.00674987e-01 -1.36336052e+00 -2.92539358e-01 4.63271797e-01
9.12772398e-03 -6.12232089e-02 -5.90081275e-01 -7.07194865e-01
2.88642764e-01 8.48568320e-01 -7.33038127e-01 -1.20202971e+00
-5.39297648e-02 -8.61159325e-01 7.52919555e-01 1.95574239e-01
8.28910768e-01 -1.28267062e+00 -7.13625669e-01 4.63137627e-01
3.33693147e-01 -4.29247230e-01 -2.00736910e-01 3.37081611e-01
-1.20198417e+00 -1.00558472e+00 -7.87977457e-01 -7.09860325e-01
8.13320339e-01 6.08902462e-02 9.40935314e-01 8.88274387e-02
-3.84725094e-01 3.83116782e-01 3.14213498e-03 -9.11704063e-01
-2.94020683e-01 2.86010325e-01 4.05408204e-01 -3.31129655e-02
7.24784017e-01 -7.16588914e-01 -6.42058671e-01 -9.19369683e-02
-1.16649652e+00 -2.68650949e-01 8.77122760e-01 1.23409200e+00
7.61121869e-01 2.02286884e-01 1.42800891e+00 -1.12963283e+00
1.04249287e+00 -6.53448880e-01 -4.15327251e-01 4.61514771e-01
-1.05631757e+00 9.12152827e-02 4.63246703e-01 -6.74495578e-01
-9.62142289e-01 3.96164842e-02 2.25494102e-01 -5.20633042e-01
2.90537417e-01 4.29004431e-01 -2.57597733e-02 -1.43274171e-02
6.62916899e-01 8.11797261e-01 2.31204152e-01 -4.51146394e-01
-8.90458897e-02 6.09266579e-01 5.86471796e-01 -6.39470518e-02
2.57009476e-01 3.17714214e-01 -2.75570184e-01 -5.12797952e-01
-6.14630342e-01 2.58373529e-01 -3.54621738e-01 -4.38188240e-02
5.51534712e-01 -1.15494621e+00 -6.67157114e-01 7.65692532e-01
-8.22751462e-01 -2.71724015e-01 -3.63218397e-01 5.65687776e-01
-3.56596470e-01 -1.07720405e-01 -4.56578404e-01 -5.95890701e-01
-9.48654711e-01 -5.47260880e-01 2.35676244e-01 5.95865667e-01
-1.10604718e-01 -8.48823309e-01 3.09487671e-01 -3.07042629e-01
5.28448820e-01 1.55490831e-01 1.20576036e+00 -8.37782919e-01
-4.59304243e-01 -2.63207495e-01 6.61105737e-02 5.88528454e-01
2.95633316e-01 -6.49875224e-01 -1.04261863e+00 -9.38898087e-01
3.64570379e-01 -1.70074910e-01 1.13383937e+00 2.84573346e-01
1.37833130e+00 -4.50288415e-01 -5.62946022e-01 5.89878619e-01
1.56634760e+00 6.95261598e-01 7.81316400e-01 6.43744171e-02
2.39391938e-01 -5.24900556e-02 1.34727806e-01 5.06963670e-01
4.12571505e-02 -2.46861011e-01 8.78370851e-02 8.72142315e-02
-3.01392600e-02 -6.94828093e-01 3.08225870e-01 6.71452641e-01
7.08198249e-01 -2.41423592e-01 -8.61834705e-01 6.63382947e-01
-1.73402047e+00 -7.68501759e-01 4.39066827e-01 2.48397756e+00
1.25799561e+00 3.04206580e-01 -3.99265021e-01 -7.80614167e-02
5.86488545e-01 -3.25921565e-01 -1.44296205e+00 -3.21977973e-01
-1.52091935e-01 4.84959185e-01 6.65474594e-01 4.23239134e-02
-5.90062380e-01 8.13683510e-01 6.42192316e+00 5.63024402e-01
-1.39819789e+00 2.79204190e-01 8.86296690e-01 -3.23765397e-01
-2.40746871e-01 -3.96466166e-01 -7.48148322e-01 6.79378092e-01
1.11515391e+00 -6.73786581e-01 1.85043335e-01 7.83003569e-01
-2.70900756e-01 -3.28888953e-01 -1.02748227e+00 1.05892384e+00
-1.09412707e-03 -1.59278154e+00 2.52802640e-01 -9.62023959e-02
8.13861370e-01 2.41281167e-01 5.00070393e-01 4.63167220e-01
2.21507907e-01 -1.02349961e+00 -2.97982227e-02 1.08747029e+00
8.88292432e-01 -1.00706375e+00 8.20100665e-01 5.70048034e-01
-2.57349074e-01 -1.14734605e-01 -7.04724848e-01 3.09706628e-01
-2.97811747e-01 9.18502450e-01 -1.05387259e+00 2.74757355e-01
8.35614800e-01 4.76270378e-01 -6.91859484e-01 1.12734640e+00
-9.21394825e-02 7.13433385e-01 1.49160093e-02 -3.47312051e-03
-3.08988333e-01 4.27622288e-01 2.76437551e-01 8.79251182e-01
3.46832663e-01 -4.36707288e-02 -5.14438689e-01 7.66913891e-01
-5.91161489e-01 -1.10795468e-01 -7.63551474e-01 -4.01317589e-02
6.12812459e-01 5.89536369e-01 -3.96118313e-01 -2.12024242e-01
-1.03491902e-01 9.55879748e-01 2.69908667e-01 3.30774128e-01
-6.21912301e-01 -3.36717665e-01 4.04290229e-01 7.42925256e-02
2.86834806e-01 2.57912092e-02 -2.86348850e-01 -1.02201152e+00
-5.44733703e-02 -5.44633329e-01 5.08105814e-01 -5.73657274e-01
-1.38109386e+00 7.51359761e-01 -3.27006459e-01 -1.06861532e+00
2.31898695e-01 -2.22655609e-02 -2.99749762e-01 8.47814441e-01
-1.48591113e+00 -4.46971625e-01 -8.48086029e-02 9.48157787e-01
1.21543810e-01 -6.15195036e-01 1.10240757e+00 2.68653572e-01
-4.40018237e-01 1.11820936e+00 5.64074874e-01 -8.01460817e-02
9.17956531e-01 -5.88252008e-01 1.31610096e-01 5.78304112e-01
3.40865403e-02 8.83710504e-01 1.70679376e-01 -1.01636314e+00
-1.19337654e+00 -1.49567342e+00 8.67738008e-01 -1.88577890e-01
-4.05962169e-02 -3.24601203e-01 -1.38869381e+00 6.94298029e-01
-3.66025902e-02 1.78429157e-01 1.18847322e+00 -3.40651602e-01
-1.80851489e-01 -4.99545723e-01 -1.68573678e+00 2.18116879e-01
7.52526462e-01 -5.34338713e-01 -6.40208602e-01 -8.03319439e-02
8.62668335e-01 -1.96462780e-01 -6.99349940e-01 1.99333683e-01
5.75294137e-01 -6.07998908e-01 6.08524263e-01 -7.55794406e-01
-2.66947653e-02 -1.55055448e-01 1.65849447e-01 -1.38931429e+00
-2.70654470e-01 -6.80809081e-01 -3.43176425e-01 1.02793896e+00
3.31498146e-01 -9.67194974e-01 8.35916698e-01 6.41887009e-01
1.95724145e-01 -7.77663529e-01 -1.20723653e+00 -7.85258353e-01
1.65398300e-01 2.27768213e-01 6.80645168e-01 9.42462742e-01
-2.09074706e-01 2.76602149e-01 -5.30560434e-01 6.19979724e-02
6.34270370e-01 -3.92272800e-01 4.89968747e-01 -1.20386517e+00
-4.60616797e-02 2.64471203e-01 -3.82894754e-01 -4.39595431e-01
-3.95101756e-02 -1.08353102e+00 -3.68266761e-01 -1.18332827e+00
3.71519148e-01 -3.76053959e-01 -1.25284994e+00 7.30118394e-01
-1.20980009e-01 -2.17346266e-01 -8.68930593e-02 4.06850845e-01
-5.15501022e-01 8.22901428e-01 1.05234766e+00 -1.97292976e-02
-3.55147183e-01 2.86631167e-01 -8.63286376e-01 1.87043220e-01
1.18242466e+00 -1.06179595e+00 -9.56148505e-01 -3.99154067e-01
5.62382303e-02 1.07064784e-01 1.95624441e-01 -1.34431195e+00
4.90130842e-01 3.23817641e-01 9.72525179e-01 -6.91846550e-01
-3.37150842e-02 -7.27835000e-01 6.51649833e-01 1.03786778e+00
-5.69705248e-01 2.12046895e-02 5.70590615e-01 9.26244557e-01
3.70470015e-03 6.59765229e-02 9.17973280e-01 -2.02818364e-01
-4.87088948e-01 6.14195347e-01 -3.45483124e-01 -1.30511820e-02
1.20532751e+00 1.26795709e-01 -4.01608527e-01 -2.21766159e-01
-8.58063936e-01 5.08068353e-02 2.75646597e-01 4.96788025e-01
1.11391497e+00 -1.32652891e+00 -5.09565055e-01 7.25077391e-01
1.36455119e-01 -1.97742522e-01 3.73555183e-01 3.47527295e-01
-2.16488153e-01 4.40782458e-01 -5.22120357e-01 -3.10152978e-01
-9.53926265e-01 7.96977282e-01 3.38618428e-01 -1.58002496e-01
-7.18510926e-01 1.05002713e+00 1.81987748e-01 -1.62080199e-01
4.48755234e-01 -1.10480376e-01 1.37053087e-01 -1.97290018e-01
8.12863946e-01 -7.80452341e-02 2.66749948e-01 3.63237299e-02
-5.46319306e-01 -2.43150279e-01 -7.49590755e-01 2.06290632e-02
1.45250654e+00 -6.13662712e-02 4.93123122e-02 7.29460001e-01
1.35873568e+00 -4.70515907e-01 -1.20078003e+00 -7.65746593e-01
-2.60375500e-01 -2.84697592e-01 4.89207171e-02 -1.11576176e+00
-1.11152434e+00 8.12980354e-01 1.21562922e+00 -4.58661139e-01
1.17410457e+00 -3.92594814e-01 8.97598207e-01 5.06780207e-01
7.09563613e-01 -9.33736145e-01 1.37753978e-01 2.57981241e-01
7.20358074e-01 -9.14032936e-01 1.57629564e-01 4.85148549e-01
-5.39800286e-01 7.66815007e-01 5.89342475e-01 -7.41355717e-02
1.09518492e+00 1.17570616e-01 1.04451654e-02 -2.17401274e-02
-1.03130555e+00 7.39697278e-01 -2.58663923e-01 5.82773805e-01
-6.52295351e-02 -8.66856724e-02 -3.32468987e-01 1.00380909e+00
-1.55498618e-02 8.42273831e-01 4.69377041e-01 1.23455322e+00
-2.78186440e-01 -1.00788677e+00 2.10676357e-01 7.00820982e-01
-5.46622217e-01 -2.35981509e-01 -3.25363278e-01 4.47629720e-01
3.25554349e-02 5.11678636e-01 1.24565745e-02 -5.81162810e-01
1.53969288e-01 2.08545759e-01 1.90994307e-01 -4.52965826e-01
-6.92547560e-01 -5.18968642e-01 -6.62602425e-01 -1.70421928e-01
3.22753936e-02 -5.38395286e-01 -1.33710837e+00 -3.90591323e-01
-1.19335666e-01 1.19624428e-01 4.19384181e-01 5.14475405e-01
9.43396986e-01 5.30056715e-01 6.28423214e-01 1.77213505e-01
-4.58587497e-01 -6.08197451e-01 -8.07824492e-01 9.94809791e-02
6.56003118e-01 -4.47324932e-01 -1.91582069e-01 -8.41215700e-02] | [9.792078971862793, 3.4514923095703125] |
7d0eb565-9011-41f9-bbe4-a0168b87a8a2 | looking-into-your-speech-learning-cross-modal | 2104.02775 | null | https://arxiv.org/abs/2104.02775v1 | https://arxiv.org/pdf/2104.02775v1.pdf | Looking into Your Speech: Learning Cross-modal Affinity for Audio-visual Speech Separation | In this paper, we address the problem of separating individual speech signals from videos using audio-visual neural processing. Most conventional approaches utilize frame-wise matching criteria to extract shared information between co-occurring audio and video. Thus, their performance heavily depends on the accuracy of audio-visual synchronization and the effectiveness of their representations. To overcome the frame discontinuity problem between two modalities due to transmission delay mismatch or jitter, we propose a cross-modal affinity network (CaffNet) that learns global correspondence as well as locally-varying affinities between audio and visual streams. Given that the global term provides stability over a temporal sequence at the utterance-level, this resolves the label permutation problem characterized by inconsistent assignments. By extending the proposed cross-modal affinity on the complex network, we further improve the separation performance in the complex spectral domain. Experimental results verify that the proposed methods outperform conventional ones on various datasets, demonstrating their advantages in real-world scenarios. | ['Kwanghoon Sohn', 'Hong-Goo Kang', 'Sunok Kim', 'Soo-Whan Chung', 'Jiyoung Lee'] | 2021-03-25 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Lee_Looking_Into_Your_Speech_Learning_Cross-Modal_Affinity_for_Audio-Visual_Speech_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Lee_Looking_Into_Your_Speech_Learning_Cross-Modal_Affinity_for_Audio-Visual_Speech_CVPR_2021_paper.pdf | cvpr-2021-1 | ['audio-visual-synchronization', 'audio-visual-synchronization'] | ['audio', 'computer-vision'] | [ 1.61058828e-01 -5.44920266e-01 -1.81771368e-01 -3.79119486e-01
-8.93132567e-01 -5.39151192e-01 2.62182772e-01 3.03329360e-02
-3.67074579e-01 5.31660199e-01 8.14183801e-02 3.12078655e-01
-4.67066556e-01 -8.08000490e-02 -5.68217576e-01 -8.24214637e-01
-2.03859136e-01 -1.05533920e-01 4.29310381e-01 1.59820259e-01
1.13187723e-01 2.12149486e-01 -1.86747062e+00 3.97677302e-01
6.77447975e-01 1.28724086e+00 1.45932466e-01 5.51032722e-01
-8.36083442e-02 7.68666625e-01 -5.67919135e-01 8.72283708e-03
1.08404979e-01 -4.63189185e-01 -5.02618551e-01 1.30706400e-01
5.14376879e-01 -2.37409458e-01 -3.88117462e-01 1.28316987e+00
6.44228995e-01 3.94976079e-01 4.33003306e-01 -1.64373195e+00
-3.04504246e-01 5.12347281e-01 -7.01978028e-01 3.71942937e-01
4.32292014e-01 -1.48400605e-01 1.08952343e+00 -1.09329510e+00
2.92965025e-01 1.12011707e+00 8.62419844e-01 2.08083183e-01
-1.09087086e+00 -8.55524719e-01 4.39092219e-01 8.14850748e-01
-1.82859623e+00 -8.42797875e-01 1.05545890e+00 -5.37961125e-01
6.64687216e-01 2.86353499e-01 4.88769859e-01 9.04250801e-01
-2.12684616e-01 7.30001807e-01 7.61398613e-01 -2.19663039e-01
6.71177544e-03 -1.12010069e-01 -6.12370446e-02 3.62273395e-01
-3.60546947e-01 -6.28669932e-02 -1.04899633e+00 -1.79514512e-01
6.45640552e-01 -1.48867801e-01 -9.39235747e-01 -3.79003257e-01
-1.21766853e+00 3.51681113e-01 1.58044830e-01 6.82026565e-01
-1.10817961e-01 -5.74307106e-02 6.47934735e-01 3.72326344e-01
3.37037295e-01 -7.36105768e-03 -1.71576291e-01 -7.71779940e-02
-1.10725176e+00 -2.98570722e-01 3.14721018e-01 8.44749689e-01
6.36218727e-01 2.72117347e-01 -1.84484929e-01 9.46424365e-01
3.84087741e-01 2.31111348e-01 6.96788907e-01 -8.49157214e-01
4.08032507e-01 1.43994354e-02 9.33516622e-02 -1.45187140e+00
-2.98835754e-01 -2.76495874e-01 -8.22507381e-01 -2.00575382e-01
5.22527993e-01 5.83573245e-03 -4.82671767e-01 2.11799693e+00
4.37258899e-01 9.00336087e-01 -1.21908411e-01 1.40933549e+00
7.69689083e-01 7.70115197e-01 -1.60261393e-01 -7.03276813e-01
1.13381350e+00 -8.60225022e-01 -1.11530769e+00 1.06830776e-01
-7.33764172e-02 -8.38460743e-01 6.14190996e-01 3.88126254e-01
-1.07435262e+00 -9.46264565e-01 -1.14654040e+00 2.23997623e-01
3.95869613e-02 1.36670515e-01 2.34096777e-02 2.09031403e-01
-9.20658946e-01 4.52038854e-01 -6.46322072e-01 -1.92934886e-01
-8.68603736e-02 3.91269356e-01 -4.07923818e-01 2.15078339e-01
-1.49327362e+00 2.28050634e-01 5.00060976e-01 4.09918129e-01
-6.99239314e-01 -5.56746185e-01 -6.84634745e-01 2.33924508e-01
4.84358013e-01 -3.64394069e-01 1.06712162e+00 -1.30531120e+00
-1.46717191e+00 4.18493420e-01 -1.67149380e-01 -2.07088917e-01
2.74289250e-01 -1.40687212e-01 -7.85291195e-01 5.74427664e-01
-9.01857093e-02 4.05018300e-01 1.40467978e+00 -1.19399607e+00
-9.07233655e-01 -2.18685746e-01 -3.77720036e-02 4.16611195e-01
-6.63841546e-01 -9.13523212e-02 -7.36807168e-01 -7.67117739e-01
3.38328868e-01 -7.05214679e-01 4.47047472e-01 5.74769676e-02
-1.21573396e-01 -1.62869483e-01 8.95871222e-01 -6.41743541e-01
1.29899633e+00 -2.57963347e+00 3.79821450e-01 2.74076253e-01
1.64051563e-01 1.05911881e-01 -2.14902982e-01 2.95238614e-01
-2.73425877e-01 -4.17055309e-01 3.86702083e-02 -2.72222787e-01
-1.06207104e-02 -2.76857130e-02 -2.54963517e-01 7.46751666e-01
1.03035256e-01 3.06521326e-01 -8.33842397e-01 -8.05797219e-01
1.55568808e-01 6.95840597e-01 -3.18356156e-01 3.76466542e-01
3.38294417e-01 6.91080272e-01 -1.42259061e-01 4.66222256e-01
6.98373497e-01 -7.47403055e-02 2.89489388e-01 -6.67568982e-01
5.49342018e-03 -6.24106675e-02 -1.58898270e+00 2.03720737e+00
-2.99511760e-01 9.38411415e-01 5.14136732e-01 -1.13982975e+00
8.71606112e-01 7.41136968e-01 7.30928421e-01 -5.14653325e-01
8.91817808e-02 2.41922438e-01 -5.06840236e-02 -6.60317242e-01
3.64406377e-01 -5.81245162e-02 2.72971749e-01 1.48806468e-01
3.65756303e-01 3.96263957e-01 2.51889937e-02 -1.16399266e-01
7.26997137e-01 -4.04063463e-02 9.64339003e-02 -1.35477915e-01
1.03357697e+00 -7.50209272e-01 8.04716110e-01 5.23383796e-01
-5.76009214e-01 7.84272194e-01 1.90740749e-01 -5.02063818e-02
-6.29375160e-01 -9.80771005e-01 -8.99209306e-02 1.18501818e+00
5.30500531e-01 -4.52569336e-01 -6.62460983e-01 -3.83582264e-01
-1.83777511e-01 -6.17942698e-02 -3.01174015e-01 -2.17036396e-01
-5.33932388e-01 -3.24864239e-01 6.44249678e-01 4.83314395e-01
3.35292667e-01 -4.99775857e-01 -3.31419885e-01 1.72942296e-01
-6.74288034e-01 -1.39801407e+00 -7.73839056e-01 5.87610342e-03
-4.14739549e-01 -9.95015860e-01 -7.85982370e-01 -9.01029348e-01
3.62981349e-01 5.80436170e-01 6.67719483e-01 -7.43901432e-02
-6.13266677e-02 6.88635170e-01 -3.52270871e-01 2.50128895e-01
1.45940273e-03 -2.31877983e-01 3.11925441e-01 7.62396991e-01
1.28713340e-01 -8.56052935e-01 -5.46005428e-01 5.77359796e-01
-8.98011148e-01 -2.02602193e-01 1.23019181e-01 9.74158704e-01
4.74868536e-01 4.51917857e-01 8.81528854e-01 -6.67817667e-02
5.29704750e-01 -2.37807184e-01 -5.29071629e-01 2.79094577e-01
-2.23355219e-01 -1.59378782e-01 5.98340869e-01 -7.95396686e-01
-1.09139490e+00 2.63116732e-02 2.41040915e-01 -9.75629568e-01
-1.27322093e-01 5.04014492e-01 -2.58294344e-01 1.10526395e-03
1.88884497e-01 1.68880045e-01 -9.42818373e-02 -3.13304901e-01
6.93863183e-02 7.54263163e-01 8.52606535e-01 -5.02480626e-01
5.93144774e-01 4.57973450e-01 -2.12274998e-01 -9.77223277e-01
-4.05013949e-01 -6.80336773e-01 -5.32558262e-01 -5.90349793e-01
8.97655129e-01 -1.10566187e+00 -8.89111161e-01 4.77126747e-01
-1.23303413e+00 3.17462206e-01 -4.86153215e-02 9.64412808e-01
-5.27149975e-01 6.94477618e-01 -6.10444307e-01 -9.45034444e-01
8.80787671e-02 -1.13855910e+00 9.75244462e-01 1.33352473e-01
-9.26508978e-02 -7.29703248e-01 4.14770022e-02 1.31796360e-01
6.20524660e-02 -4.28332835e-02 6.03624701e-01 -5.00899494e-01
-4.65526432e-01 4.92969751e-02 -2.49711081e-01 1.96892992e-01
2.35123247e-01 1.90336764e-01 -1.23665869e+00 -4.00884241e-01
1.89792231e-01 -1.29226848e-01 7.17691541e-01 4.68961388e-01
9.50980246e-01 -5.41999564e-02 -1.41342893e-01 4.89830971e-01
1.13214958e+00 3.83451611e-01 2.21294180e-01 1.35111332e-01
7.19734907e-01 8.17560196e-01 6.55153275e-01 5.06831229e-01
1.37067974e-01 9.48070943e-01 4.35986936e-01 8.59845430e-02
-1.31248012e-01 -4.74981871e-03 5.49623489e-01 1.37743402e+00
1.30003884e-01 -1.85051411e-01 -5.45940757e-01 6.09185815e-01
-2.22968006e+00 -1.22911620e+00 1.85742434e-02 2.28044868e+00
7.77340829e-01 2.85690036e-02 1.71858981e-01 4.69969064e-01
1.35132468e+00 2.35847428e-01 -4.28406149e-01 5.81089258e-02
-3.65507841e-01 -2.32813418e-01 7.64023000e-03 4.39987570e-01
-1.28555679e+00 2.74055153e-01 5.44772291e+00 1.25267458e+00
-1.24040008e+00 2.77153552e-01 1.76980406e-01 -2.81887710e-01
2.14527808e-02 -2.59615451e-01 -3.79241467e-01 5.80883920e-01
6.88685060e-01 -8.54448751e-02 5.23640752e-01 2.94780880e-01
2.37938747e-01 4.51154672e-02 -1.31898713e+00 1.37113667e+00
2.36112297e-01 -9.22740340e-01 -2.96537489e-01 -4.16935325e-01
4.40952569e-01 -3.78578216e-01 9.42593217e-02 -8.00917298e-03
-4.74616438e-01 -7.25021839e-01 1.04193616e+00 5.41348577e-01
6.36073947e-01 -8.68936002e-01 6.09232545e-01 1.96934476e-01
-1.76021373e+00 -7.80360252e-02 -7.04534948e-02 2.83589602e-01
1.88718513e-01 4.07546043e-01 -4.66263384e-01 8.52491617e-01
9.81186390e-01 9.09578323e-01 -2.67461926e-01 1.02356088e+00
7.60439634e-02 2.63870358e-01 -3.29378843e-01 5.06428063e-01
4.44377102e-02 -9.33624059e-02 7.45389283e-01 1.23100197e+00
3.47914636e-01 -9.71211717e-02 2.10571334e-01 5.68965197e-01
1.95453465e-01 1.17768347e-01 -4.13494080e-01 3.61245945e-02
7.40615845e-01 1.15412760e+00 -6.85338140e-01 -9.17610377e-02
-5.70435643e-01 8.56781185e-01 5.00985347e-02 5.45864165e-01
-1.19503844e+00 -4.87882584e-01 6.51251435e-01 -4.02808875e-01
4.37218368e-01 -1.45200834e-01 2.73560524e-01 -1.14223182e+00
3.43012065e-01 -8.92082751e-01 4.37431335e-01 -9.19490933e-01
-1.27936637e+00 8.15245211e-01 -3.55864689e-02 -1.82319188e+00
-1.99347079e-01 -1.94832653e-01 -1.51690438e-01 5.12921810e-01
-1.47616446e+00 -8.72871339e-01 -2.72762209e-01 1.06905007e+00
5.49404085e-01 -1.37197047e-01 4.55383629e-01 9.22883987e-01
-6.45374477e-01 8.08328509e-01 1.84255600e-01 2.45065391e-02
1.23500121e+00 -6.76393867e-01 -4.97104913e-01 8.16369653e-01
4.01187211e-01 4.05071020e-01 6.12649322e-01 -2.82100677e-01
-1.19527781e+00 -8.40215683e-01 4.42802429e-01 2.42444187e-01
7.88213253e-01 -2.10945413e-01 -1.16072810e+00 3.41282427e-01
4.71181244e-01 1.02139406e-01 7.89459705e-01 -1.02822274e-01
-5.25991440e-01 -4.60614741e-01 -7.08960712e-01 2.34230518e-01
8.48281384e-01 -1.10967553e+00 -4.30322260e-01 8.63170344e-03
6.85335636e-01 -3.86496335e-01 -8.55060279e-01 4.32646990e-01
7.03824341e-01 -1.25991583e+00 9.19481277e-01 -3.12247574e-01
3.61288711e-02 -7.81328559e-01 -4.29724991e-01 -1.01727605e+00
-2.89794207e-01 -6.62121117e-01 -1.20554268e-01 1.67907393e+00
-6.74951226e-02 -4.31039482e-01 2.65496582e-01 1.56223923e-01
-3.13496105e-02 -2.14898705e-01 -1.35333860e+00 -9.02875066e-01
-7.26552129e-01 -4.37576890e-01 5.36476016e-01 1.34739411e+00
2.23918319e-01 3.82402688e-01 -7.03094780e-01 5.24823070e-01
4.63341087e-01 2.57839143e-01 3.59318733e-01 -1.15022635e+00
-5.72151601e-01 -4.27642435e-01 -6.43994153e-01 -1.02496529e+00
4.67430592e-01 -5.66141903e-01 2.65721172e-01 -7.04449892e-01
-4.64118645e-03 -5.22911437e-02 -6.88469887e-01 1.12033315e-01
-8.68572518e-02 3.86742771e-01 2.81518340e-01 5.40950179e-01
-8.33099663e-01 7.06491590e-01 7.84688771e-01 -2.13294700e-01
-1.52474955e-01 -3.05023789e-01 -1.03683680e-01 8.88962507e-01
4.34197515e-01 -3.25406402e-01 -6.81875467e-01 -6.13262177e-01
-2.15498246e-02 3.95525753e-01 3.04790229e-01 -1.18970954e+00
6.15328491e-01 7.04169506e-03 1.63695425e-01 -5.81791103e-01
6.41005158e-01 -1.30881369e+00 3.14894497e-01 -3.12382239e-03
-4.00566906e-01 3.69948037e-02 2.24412173e-01 9.30932760e-01
-8.11803699e-01 -1.17588647e-01 7.63752997e-01 3.81476760e-01
-7.43799448e-01 1.68326482e-01 -3.59792054e-01 -4.97656204e-02
1.02611411e+00 -4.03176248e-01 -2.85639763e-01 -7.36081719e-01
-8.99110913e-01 1.84330717e-01 9.56278145e-02 4.99165267e-01
5.66445053e-01 -1.74540019e+00 -2.97496408e-01 3.29689354e-01
1.36397675e-01 -3.09160173e-01 6.94109797e-01 1.24807882e+00
2.80229669e-05 2.66196817e-01 -2.83238053e-01 -1.04200387e+00
-1.40930498e+00 5.65644264e-01 3.90315205e-01 2.20382348e-01
-3.10003191e-01 8.26341152e-01 4.19860840e-01 1.73294187e-01
6.33709610e-01 -1.00430235e-01 -3.68738502e-01 4.48680729e-01
5.15591562e-01 3.37901056e-01 5.76924235e-02 -1.15785074e+00
-5.95835626e-01 8.01616549e-01 2.45459467e-01 -2.80171454e-01
1.02891874e+00 -6.56133771e-01 -6.86082765e-02 8.92026246e-01
1.36561799e+00 3.61709599e-03 -1.34318984e+00 -5.19819736e-01
-2.62821764e-01 -6.25510275e-01 6.13589846e-02 -1.97577208e-01
-1.23520422e+00 8.66501331e-01 8.86636972e-01 5.14152706e-01
1.51329732e+00 -2.88262695e-01 6.20779812e-01 7.75599107e-02
3.43088955e-02 -1.18858993e+00 2.03170016e-01 2.37585530e-01
7.09664702e-01 -1.04892516e+00 -1.01372048e-01 -3.67169797e-01
-4.55944657e-01 1.23551559e+00 5.92579067e-01 2.15733305e-01
6.33551598e-01 8.41687471e-02 9.47587043e-02 1.70110777e-01
-6.35881901e-01 -1.14169411e-01 4.99016464e-01 6.49472356e-01
2.54768580e-01 -3.54257911e-01 5.85979596e-03 6.02926731e-01
2.62658775e-01 -3.43846649e-01 1.97257027e-01 7.25738704e-01
-2.94076174e-01 -8.08213592e-01 -6.91740394e-01 -2.23039761e-01
-3.62168312e-01 1.55320406e-01 -1.64850995e-01 6.56499922e-01
1.00238055e-01 1.19085157e+00 3.15220356e-01 -5.43880880e-01
1.87014282e-01 9.73346904e-02 5.06003559e-01 -3.77042405e-02
-4.27509904e-01 9.25728083e-01 -8.34027454e-02 -5.91376722e-01
-1.06816745e+00 -5.93826711e-01 -1.20780313e+00 -7.39724934e-02
-5.24331570e-01 3.38175476e-01 3.01512271e-01 8.62814486e-01
5.21596670e-01 6.97411716e-01 7.90632308e-01 -1.04805744e+00
-4.29008961e-01 -6.71744466e-01 -8.29558790e-01 4.36707318e-01
7.62484252e-01 -8.48103404e-01 -4.96115357e-01 3.19853604e-01] | [14.114029884338379, 4.839443683624268] |
134dbd24-3376-45a7-8dce-f0798031d41a | the-paradox-of-choice-using-attention-in | 2201.09653 | null | https://arxiv.org/abs/2201.09653v1 | https://arxiv.org/pdf/2201.09653v1.pdf | The Paradox of Choice: Using Attention in Hierarchical Reinforcement Learning | Decision-making AI agents are often faced with two important challenges: the depth of the planning horizon, and the branching factor due to having many choices. Hierarchical reinforcement learning methods aim to solve the first problem, by providing shortcuts that skip over multiple time steps. To cope with the breadth, it is desirable to restrict the agent's attention at each step to a reasonable number of possible choices. The concept of affordances (Gibson, 1977) suggests that only certain actions are feasible in certain states. In this work, we model "affordances" through an attention mechanism that limits the available choices of temporally extended options. We present an online, model-free algorithm to learn affordances that can be used to further learn subgoal options. We investigate the role of hard versus soft attention in training data collection, abstract value learning in long-horizon tasks, and handling a growing number of choices. We identify and empirically illustrate the settings in which the paradox of choice arises, i.e. when having fewer but more meaningful choices improves the learning speed and performance of a reinforcement learning agent. | ['Doina Precup', 'Khimya Khetarpal', 'Andrei Nica'] | 2022-01-24 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [ 2.45224208e-01 2.40220740e-01 -6.35794818e-01 -6.16054051e-02
-4.98043597e-01 -5.95857024e-01 5.81619322e-01 1.55559555e-01
-8.68094206e-01 1.07535124e+00 3.68585289e-01 -5.01812577e-01
-5.15161216e-01 -7.65648067e-01 -4.31039929e-01 -6.21320367e-01
-4.45799202e-01 6.27668619e-01 2.64290094e-01 -4.39108521e-01
4.38302636e-01 6.22175694e-01 -1.69067621e+00 -9.17804539e-02
7.69592166e-01 8.85877669e-01 4.68534291e-01 6.08318210e-01
1.63330913e-01 8.23986232e-01 -3.33307594e-01 1.77937403e-01
4.78762120e-01 -3.26672822e-01 -1.07544172e+00 4.15490538e-01
-2.16217682e-01 -8.36930275e-01 -2.42327735e-01 9.15437281e-01
2.08850533e-01 7.69602001e-01 4.61420268e-01 -1.49298704e+00
-5.63733354e-02 8.57555270e-01 -2.00324655e-01 4.89922374e-01
3.11957240e-01 7.54933834e-01 1.16555333e+00 -1.73406616e-01
6.53025329e-01 1.40543461e+00 -2.72987783e-02 6.99005544e-01
-1.22316658e+00 -8.37665349e-02 7.04307020e-01 3.55948001e-01
-8.81321669e-01 -3.02500516e-01 5.64732492e-01 -4.87990379e-01
1.31777072e+00 2.78057188e-01 1.00498903e+00 8.64675522e-01
2.93015778e-01 8.22930038e-01 1.03811240e+00 -4.80586320e-01
7.55260944e-01 -5.71685195e-01 -1.90721333e-01 4.57531124e-01
-3.26581076e-02 7.28270173e-01 -5.03063560e-01 -1.15856841e-01
1.05486679e+00 2.19999939e-01 -2.01410465e-02 -5.15239120e-01
-1.22489357e+00 1.02639484e+00 4.19244885e-01 1.09775037e-01
-6.19659483e-01 2.99589157e-01 1.92353994e-01 6.86840236e-01
-2.59879917e-01 1.13911891e+00 -5.98735213e-01 -1.74422249e-01
-3.94445300e-01 5.74054480e-01 6.05356455e-01 9.88766074e-01
7.94578433e-01 -9.56624746e-02 -2.89633423e-01 4.17816252e-01
1.45893142e-01 -5.89610217e-03 2.29366258e-01 -1.49469876e+00
5.95916152e-01 2.77260959e-01 7.64148235e-01 -2.44854584e-01
-6.37419760e-01 1.21059222e-02 -1.37373582e-01 8.33702266e-01
7.05023527e-01 -4.86703306e-01 -1.09395838e+00 1.90369511e+00
4.93566811e-01 -3.68187249e-01 5.55085875e-02 1.15373981e+00
-1.22993417e-01 4.35375631e-01 3.32202017e-01 -4.36500102e-01
1.10843527e+00 -9.77740824e-01 -6.01667702e-01 -5.41326463e-01
5.37809432e-01 -2.24969044e-01 1.11320508e+00 4.51262146e-01
-1.24792302e+00 -1.59607559e-01 -9.33596432e-01 -1.78808831e-02
-1.02783116e-02 -6.58558965e-01 1.04313755e+00 4.57032397e-02
-8.30370486e-01 1.14231241e+00 -1.06561303e+00 -1.58105150e-01
4.06508356e-01 4.98971522e-01 -1.94119606e-02 -1.45008504e-01
-1.17990184e+00 1.19468725e+00 6.05418086e-01 -6.38886541e-02
-1.31184351e+00 -9.85868201e-02 -8.03598523e-01 1.89489856e-01
1.32112169e+00 -5.65697849e-01 1.71573436e+00 -1.02445030e+00
-1.54370582e+00 1.11019254e-01 2.57760823e-01 -5.25076687e-01
6.11049294e-01 -2.38137633e-01 2.62043297e-01 7.91849643e-02
1.93630129e-01 1.02395654e+00 9.25622046e-01 -7.39967883e-01
-9.00296152e-01 -3.45241159e-01 8.66465032e-01 8.94519806e-01
3.15901428e-01 -2.39095822e-01 -7.41842110e-03 -2.87890643e-01
1.12470642e-01 -1.18509614e+00 -9.56681669e-01 1.31278038e-02
-1.73622027e-01 -5.51483631e-01 -7.08483234e-02 -5.78965247e-02
8.51142824e-01 -1.87933111e+00 5.76689065e-01 2.91453605e-03
6.37482032e-02 -3.45090985e-01 -2.70141244e-01 6.59069657e-01
3.26629639e-01 6.43821508e-02 -9.95058119e-02 1.57023638e-01
1.99143410e-01 8.07550788e-01 -1.73228309e-01 1.76853940e-01
6.89274818e-02 7.04658628e-01 -1.18574595e+00 -3.30855161e-01
1.90389514e-01 -2.81094760e-01 -8.10775816e-01 3.45056772e-01
-8.21652949e-01 5.61200023e-01 -8.76773357e-01 4.67697531e-01
1.32322818e-01 4.99177910e-02 2.91480720e-01 5.05558372e-01
-4.57414389e-01 7.52317548e-01 -1.43535495e+00 1.76329148e+00
-2.11015001e-01 2.19848186e-01 -1.82602759e-02 -5.02004862e-01
3.50200742e-01 1.50142968e-01 3.10711831e-01 -6.74510002e-01
1.38391659e-01 1.00290112e-01 4.55290496e-01 -5.79281628e-01
4.70777810e-01 -4.23480511e-01 -1.32217512e-01 4.05398756e-01
-1.57660767e-01 -4.81653631e-01 5.28722584e-01 -1.66595325e-01
1.10632265e+00 5.12549520e-01 6.95971847e-01 -2.88471252e-01
-9.26509276e-02 3.11521292e-01 8.72553110e-01 1.14424467e+00
-3.63316536e-01 1.86554734e-02 8.93355072e-01 -7.07962930e-01
-1.14107990e+00 -8.35269630e-01 6.94637001e-02 1.54928768e+00
2.29124591e-01 -1.16831712e-01 -6.12646997e-01 -4.56763834e-01
3.41077819e-02 9.44518745e-01 -6.93508029e-01 -2.33458243e-02
-6.53794646e-01 -2.92855531e-01 -1.68555707e-01 6.11143887e-01
2.19456255e-01 -1.58465803e+00 -1.88077891e+00 4.00331497e-01
1.33112475e-01 -5.96367180e-01 -5.57429016e-01 6.89501345e-01
-9.54147875e-01 -9.52835917e-01 -5.26210606e-01 -3.63471389e-01
5.09063721e-01 2.30690800e-02 9.85180080e-01 2.73208261e-01
-5.38359582e-02 3.41362178e-01 -3.25556993e-01 -2.27237985e-01
-9.75905359e-02 3.70453522e-02 7.08458126e-02 -6.62577212e-01
1.03530161e-01 -4.07441527e-01 -6.57446682e-01 9.95008573e-02
-7.45116472e-01 4.53366995e-01 6.01092458e-01 8.88490260e-01
4.83753949e-01 1.91582412e-01 4.16693807e-01 -5.58735728e-01
7.11389184e-01 -4.52005088e-01 -8.81402194e-01 6.92585409e-02
-5.34309626e-01 7.10483134e-01 4.27467704e-01 -6.18279934e-01
-9.77167308e-01 -5.28690591e-03 1.33021399e-01 -1.28066078e-01
-2.60731995e-01 6.07009292e-01 -1.55737132e-01 2.22610429e-01
6.00113630e-01 -2.22490624e-01 -1.63786724e-01 -3.15505713e-01
3.30800682e-01 1.63255870e-01 -4.24773470e-02 -1.03060210e+00
1.49998486e-01 1.34092178e-02 1.41948402e-01 -6.68648541e-01
-7.88249969e-01 -1.79071128e-01 -5.04272938e-01 -1.96918249e-01
8.41280818e-01 -4.53227192e-01 -9.58347559e-01 9.76374894e-02
-8.66328299e-01 -1.06467390e+00 -6.48298562e-01 4.30995852e-01
-1.18387878e+00 1.84205845e-02 -4.40946400e-01 -1.02446532e+00
2.67448127e-01 -1.39987624e+00 5.98578334e-01 3.69729698e-01
-3.78308803e-01 -5.19232869e-01 -1.47633374e-01 -1.01289570e-01
1.26939803e-01 2.89013863e-01 8.61163378e-01 -3.97027671e-01
-8.86225343e-01 5.46299696e-01 2.06702143e-01 -4.23707843e-01
-9.11460295e-02 -5.54791510e-01 -4.18238431e-01 -2.66399354e-01
1.65368319e-02 -5.48875928e-01 7.55938888e-01 4.41612601e-01
8.99656713e-01 -7.96652317e-01 -1.71955898e-01 2.87584811e-01
1.32325494e+00 7.65999496e-01 4.58862036e-01 7.51657724e-01
2.53350198e-01 8.12700510e-01 9.37050581e-01 8.56043875e-01
3.99339437e-01 4.84266460e-01 8.39217126e-01 6.33750260e-01
4.67424750e-01 -2.64358670e-01 -3.93899083e-02 -2.40857348e-01
-3.21864158e-01 -2.47914448e-01 -9.13336039e-01 5.05267262e-01
-2.06068563e+00 -9.48492289e-01 6.66263640e-01 2.40595222e+00
1.02361441e+00 5.41171134e-01 3.11114967e-01 2.89332606e-02
3.14516962e-01 2.37703040e-01 -1.26133621e+00 -5.43376565e-01
5.12865663e-01 -1.37247965e-01 6.52930915e-01 8.82652581e-01
-9.92178082e-01 1.03281260e+00 6.41447020e+00 4.21969444e-01
-7.03258991e-01 -2.10131630e-01 4.86247122e-01 -4.26342100e-01
-3.67033452e-01 1.93022266e-01 -6.35124922e-01 2.18633696e-01
7.68555403e-01 4.41266689e-03 1.13489699e+00 8.44387889e-01
9.98826176e-02 -5.76629162e-01 -1.49937403e+00 4.75725263e-01
-6.62062824e-01 -1.01672995e+00 -3.15131277e-01 1.75892621e-01
6.36681616e-01 5.13052531e-02 -2.11015657e-01 4.26630288e-01
7.98414707e-01 -1.08720696e+00 1.05440354e+00 3.66510421e-01
4.87136573e-01 -8.91602874e-01 2.29745194e-01 8.34395826e-01
-9.07456040e-01 -7.58834660e-01 -2.43524611e-01 -8.06761026e-01
9.21370238e-02 -2.79112220e-01 -7.22879827e-01 -9.75041986e-02
4.18630749e-01 3.25730950e-01 4.86047007e-03 9.94269252e-01
-5.54112971e-01 -4.19355445e-02 -5.24161637e-01 -5.26229382e-01
7.63590932e-01 -1.92868114e-01 4.86960232e-01 4.82689083e-01
1.74191505e-01 7.19438672e-01 6.77949667e-01 8.31895769e-01
4.83343571e-01 -4.61708874e-01 -4.72536862e-01 -2.30099171e-01
6.49763703e-01 6.60504937e-01 -8.40071738e-01 -1.81778803e-01
-2.43269041e-01 5.59491992e-01 5.75860023e-01 5.09828568e-01
-4.49336588e-01 8.66056606e-02 6.79825187e-01 4.70424443e-02
2.48471856e-01 -5.10957837e-01 -2.09273770e-01 -9.42710042e-01
-2.97413558e-01 -8.98463309e-01 6.12875462e-01 -6.40978873e-01
-7.66813397e-01 2.83926964e-01 3.72770131e-01 -8.84554744e-01
-5.28933644e-01 -4.87509161e-01 -3.39091718e-01 7.14217603e-01
-1.57057154e+00 -5.81019104e-01 1.71259671e-01 5.03289759e-01
1.02952194e+00 2.23279312e-01 6.07673049e-01 -4.68714386e-01
-1.29239619e-01 -7.95443878e-02 -3.00877631e-01 -4.02192861e-01
7.68652302e-04 -1.42151570e+00 2.43214116e-01 5.09958744e-01
-2.78914958e-01 5.66665649e-01 1.03770578e+00 -5.88596284e-01
-1.39420140e+00 -3.26160580e-01 5.89753032e-01 -1.59335285e-01
5.13172328e-01 1.00513503e-01 -6.72796488e-01 8.80685449e-01
8.21011215e-02 -3.51825327e-01 3.06886911e-01 2.32889578e-01
6.64992258e-02 3.46485972e-01 -1.05828226e+00 1.14315069e+00
1.15190017e+00 -5.05170785e-02 -1.04664743e+00 1.24028549e-01
8.57115269e-01 -6.12790048e-01 -6.16094530e-01 1.94511756e-01
5.02803624e-01 -9.73453283e-01 7.57563949e-01 -9.14030492e-01
2.46446207e-01 -1.04204237e-01 -1.50111482e-01 -1.35361040e+00
-4.81194586e-01 -1.06354988e+00 -9.61445346e-02 3.06931257e-01
2.30677098e-01 -4.63570535e-01 7.23622084e-01 1.11354399e+00
-1.16922259e-01 -8.52652967e-01 -1.19668984e+00 -6.03479564e-01
2.40374297e-01 -2.89792836e-01 8.99830937e-01 5.92858970e-01
3.04348141e-01 2.49163851e-01 -3.23038071e-01 -1.11020990e-01
5.22780538e-01 2.30255350e-01 2.88846940e-01 -1.20624137e+00
-5.20116448e-01 -5.42633295e-01 2.45287031e-01 -1.34888244e+00
-1.69414476e-01 -2.86904037e-01 4.82051551e-01 -1.54462266e+00
-2.00439263e-02 -7.95200527e-01 -2.84280419e-01 6.94767594e-01
-3.50270420e-02 -8.63701224e-01 7.37601519e-01 1.57328010e-01
-7.72406697e-01 4.49902862e-01 1.81694698e+00 5.96090816e-02
-6.25349104e-01 1.70726344e-01 -6.27062082e-01 8.31171930e-01
9.47944164e-01 -2.47831285e-01 -6.24961197e-01 -5.05883336e-01
5.57228744e-01 8.47932696e-01 7.73426220e-02 -7.60473728e-01
1.72608718e-01 -1.24381912e+00 1.79920107e-01 -3.74441773e-01
3.82087976e-01 -9.57803607e-01 -9.92916673e-02 6.54465079e-01
-9.45377827e-01 1.72733158e-01 6.00277772e-03 5.65329909e-01
4.61252660e-01 -6.46265924e-01 5.55798590e-01 -6.93764091e-01
-1.05386686e+00 3.88891906e-01 -7.89917707e-01 1.40380055e-01
1.05021667e+00 -1.06249228e-01 1.25075474e-01 -2.71257281e-01
-1.14208841e+00 6.97270036e-01 3.95358533e-01 3.74075979e-01
4.89154100e-01 -1.06371117e+00 -2.39050582e-01 1.34699970e-01
-8.17772597e-02 3.36653769e-01 3.12839970e-02 4.42507505e-01
-1.97318152e-01 3.18567395e-01 -6.76405668e-01 -2.02843785e-01
-7.50084639e-01 9.68699455e-01 4.14633483e-01 -3.60334277e-01
-6.83042169e-01 6.51778698e-01 5.27641587e-02 -1.59853622e-01
4.35464203e-01 -6.54577553e-01 -4.20912802e-01 9.94824097e-02
4.63895708e-01 5.37515819e-01 -4.69307035e-01 -7.80214518e-02
-8.83048251e-02 2.20356062e-01 -2.17546709e-02 -5.66499054e-01
1.19720042e+00 -2.35581756e-01 2.64182597e-01 5.16925871e-01
2.54011989e-01 -7.31553972e-01 -2.26029897e+00 -1.88756347e-01
2.48078957e-01 -5.57517827e-01 -2.83576287e-02 -8.24165523e-01
-3.62830877e-01 7.05840766e-01 1.74468413e-01 4.77552325e-01
8.78841877e-01 -5.75860739e-02 5.14497221e-01 7.93538809e-01
7.98869193e-01 -1.43050897e+00 1.61158964e-01 8.03570807e-01
9.46741283e-01 -1.20464909e+00 -6.88696979e-03 1.81994453e-01
-8.77124369e-01 1.29263127e+00 7.25728929e-01 -1.68090403e-01
2.09122941e-01 -5.01230080e-03 -2.92335689e-01 -7.83172697e-02
-1.16086137e+00 -5.94981492e-01 -4.06049415e-02 5.72353542e-01
-9.96546820e-02 3.11115295e-01 -2.25025818e-01 1.85164317e-01
-2.27502882e-01 -9.20193642e-03 6.17953658e-01 1.33738387e+00
-1.06342614e+00 -9.09899592e-01 -3.01093131e-01 5.20889640e-01
-2.94005096e-01 6.74526989e-02 -5.62874712e-02 7.34224081e-01
1.38437644e-01 1.00276434e+00 1.40169650e-01 1.97162479e-01
1.87588036e-01 5.02884388e-02 7.53033817e-01 -8.04108381e-01
-2.85001576e-01 3.61632437e-01 3.01674336e-01 -8.14663291e-01
-3.02501708e-01 -9.52357411e-01 -1.49406445e+00 -1.12004325e-01
-9.22268108e-02 8.94959420e-02 2.52304137e-01 1.29390252e+00
-3.54421623e-02 6.91223443e-01 3.57215136e-01 -1.19089150e+00
-1.20864201e+00 -6.45413637e-01 -5.40443718e-01 1.72443256e-01
8.60487878e-01 -9.96465683e-01 -3.30169827e-01 -3.35423648e-01] | [4.10783052444458, 1.5081250667572021] |
f07622ac-498a-48eb-b692-114ff30bd3f2 | single-independent-component-recovery-and | 2110.05887 | null | https://arxiv.org/abs/2110.05887v3 | https://arxiv.org/pdf/2110.05887v3.pdf | Discovery of Single Independent Latent Variable | Latent variable discovery is a central problem in data analysis with a broad range of applications in applied science. In this work, we consider data given as an invertible mixture of two statistically independent components and assume that one of the components is observed while the other is hidden. Our goal is to recover the hidden component. For this purpose, we propose an autoencoder equipped with a discriminator. Unlike the standard nonlinear ICA problem, which was shown to be non-identifiable, in the special case of ICA we consider here, we show that our approach can recover the component of interest up to entropy-preserving transformation. We demonstrate the performance of the proposed approach in several tasks, including image synthesis, voice cloning, and fetal ECG extraction. | ['Ronen Talmon', 'Ori Katz', 'Jonathan Svirsky', 'Uri Shaham'] | 2021-10-12 | null | null | null | null | ['voice-cloning'] | ['speech'] | [ 6.00022733e-01 3.25089425e-01 -1.44842520e-01 -2.59196348e-02
-5.33628166e-01 -5.27843237e-01 3.96614850e-01 -7.68066570e-02
-1.88164666e-01 8.35113943e-01 9.68058854e-02 -1.39169693e-01
-1.41091526e-01 -4.19193268e-01 -7.24551439e-01 -1.32384527e+00
-7.52566978e-02 3.95710856e-01 -5.59390545e-01 3.32051039e-01
-3.48153710e-01 3.50335687e-01 -1.12000096e+00 -4.80106361e-02
8.25574636e-01 9.85900700e-01 -1.53941721e-01 8.14601064e-01
2.40617976e-01 6.77794039e-01 -5.67765892e-01 -1.86279297e-01
2.68797815e-01 -9.75797713e-01 -7.02549815e-01 4.14603978e-01
-2.96377447e-02 -9.94572937e-02 -2.71940887e-01 1.13675630e+00
3.61195952e-01 2.23882511e-01 7.85830140e-01 -1.40023148e+00
-4.03372049e-01 5.10870695e-01 -4.31732088e-01 6.81009591e-02
-7.58872777e-02 -5.36049902e-01 8.02267373e-01 -5.04687726e-01
5.48709691e-01 9.44297850e-01 4.00976211e-01 6.09795392e-01
-1.54144871e+00 -7.33748078e-01 -1.77210391e-01 -5.26294410e-02
-1.25891757e+00 -7.25325465e-01 1.11738515e+00 -5.83404958e-01
1.97247177e-01 1.48266539e-01 3.85236144e-01 1.29690075e+00
8.33360851e-02 8.29932511e-01 1.21055424e+00 -6.15023851e-01
4.89056647e-01 2.91189551e-01 8.05686414e-02 4.26675409e-01
3.78342867e-01 -2.05520149e-02 -3.33829314e-01 -3.57043713e-01
7.32793510e-01 1.77259699e-01 -3.93155426e-01 -3.53420317e-01
-1.14873350e+00 1.07008827e+00 -7.18254521e-02 4.93320346e-01
-6.79605722e-01 1.51772246e-01 -9.85991955e-02 3.21583480e-01
5.24824321e-01 2.36253724e-01 -1.17930122e-01 8.36878493e-02
-1.22278214e+00 -1.21013865e-01 9.18879926e-01 6.58296943e-01
4.55809563e-01 5.39172709e-01 1.44820005e-01 4.82069045e-01
9.64350402e-02 4.41160589e-01 8.37992132e-01 -9.17920172e-01
1.90120831e-01 1.49982706e-01 -1.31751776e-01 -8.41192544e-01
-1.68021902e-01 -6.61432087e-01 -1.28241301e+00 1.87665507e-01
3.97838205e-01 -3.45466226e-01 -8.37235093e-01 1.97898436e+00
1.65808395e-01 6.06219769e-01 5.20228863e-01 5.24053037e-01
6.66902781e-01 5.64021945e-01 -1.85061395e-01 -7.50604153e-01
1.13244438e+00 -6.67305291e-01 -1.34095502e+00 -1.13798246e-01
-2.07838282e-01 -5.93099475e-01 8.23502392e-02 4.91864741e-01
-9.70538318e-01 -3.43676805e-01 -1.06923115e+00 -1.27248811e-02
1.39096370e-02 1.85594589e-01 5.07539272e-01 5.31950355e-01
-8.72674942e-01 4.29187357e-01 -9.53313947e-01 -1.76653862e-01
2.27030411e-01 3.99992913e-01 -6.78930223e-01 2.38537893e-01
-1.01294649e+00 4.76353228e-01 2.97814250e-01 2.11068857e-02
-9.36627030e-01 -8.65991488e-02 -8.36480498e-01 1.98275417e-01
3.19156855e-01 -7.17522264e-01 1.07514668e+00 -1.22141588e+00
-1.60370505e+00 4.51178670e-01 -4.68234509e-01 -5.92886388e-01
4.14655924e-01 8.91229697e-03 -5.58245480e-01 2.63128579e-01
-2.09070388e-02 1.55604586e-01 1.42097008e+00 -1.31793666e+00
-2.77484566e-01 -3.92238915e-01 -2.74922788e-01 -3.98867205e-02
-3.86179358e-01 -1.80885777e-01 -2.90294737e-01 -8.51500452e-01
5.95859647e-01 -1.22994995e+00 -7.36090764e-02 -3.00005168e-01
-5.63445032e-01 2.40950033e-01 8.41323853e-01 -1.00712371e+00
9.12990153e-01 -2.35071421e+00 5.83981037e-01 2.83307254e-01
5.63388526e-01 -1.55925676e-01 4.07790899e-01 2.32520282e-01
-4.98673320e-01 -6.73034880e-03 -7.19585419e-01 -7.31798708e-01
-1.83956295e-01 2.86688000e-01 -4.95215535e-01 8.71186316e-01
1.89581186e-01 6.48831725e-01 -4.86497670e-01 -3.51871371e-01
1.36943832e-01 9.11951780e-01 -3.92853677e-01 3.59628648e-01
2.72673219e-01 9.27290618e-01 -3.99539769e-01 4.82053131e-01
4.86382663e-01 -3.83850873e-01 3.28807592e-01 1.70219108e-01
5.43040037e-02 -4.06254083e-03 -1.34107482e+00 1.34384274e+00
-3.36784422e-01 9.03376997e-01 1.50147572e-01 -1.26591015e+00
7.04750240e-01 1.02993250e+00 6.02181315e-01 -7.74786323e-02
3.62813562e-01 9.00195017e-02 1.61800429e-01 -4.49000806e-01
1.20327018e-01 -6.69204891e-01 -5.12191430e-02 4.96308833e-01
4.02736247e-01 3.23505074e-01 -1.66736677e-01 1.03038000e-02
9.16838765e-01 -3.51875007e-01 8.68669450e-01 -5.89752086e-02
3.98745745e-01 -4.91923422e-01 6.86363578e-01 5.23948193e-01
3.85039039e-02 8.11198354e-01 5.58987439e-01 -2.23940108e-02
-8.15814674e-01 -7.48746216e-01 -2.02471957e-01 4.32837963e-01
-1.25266477e-01 1.15347639e-01 -9.25814986e-01 -3.82056713e-01
-2.24258676e-01 6.32304549e-01 -8.35250974e-01 -3.02963816e-02
-5.19286156e-01 -8.56659114e-01 5.34702122e-01 3.99627775e-01
4.85678792e-01 -9.41937685e-01 -5.60088277e-01 3.42410594e-01
-4.83813137e-01 -1.17202723e+00 -4.16288972e-01 4.59678560e-01
-1.07900679e+00 -8.27950597e-01 -9.08980310e-01 -6.75364673e-01
7.54902899e-01 -4.18886244e-02 8.60360265e-01 -2.70341158e-01
-5.96900880e-02 4.67184573e-01 -7.47388452e-02 -4.60229993e-01
-6.28425419e-01 -8.58711451e-02 1.98012859e-01 7.02726901e-01
-3.72537342e-03 -8.61843348e-01 -3.81800234e-01 -1.00110725e-01
-1.23941958e+00 -1.89831018e-01 5.93084455e-01 9.56607103e-01
6.76396132e-01 5.74992716e-01 4.43503022e-01 -9.55990016e-01
5.25904775e-01 -8.08910668e-01 -4.45711613e-01 2.21245870e-01
-5.00502527e-01 2.97902942e-01 8.39451492e-01 -7.51668632e-01
-8.25589120e-01 3.70329052e-01 9.48665813e-02 -6.97107434e-01
-3.62939686e-01 5.22533059e-01 -4.38372761e-01 9.21347365e-02
2.61537969e-01 4.16169375e-01 6.11476712e-02 -6.57007098e-01
2.92244609e-02 5.77340603e-01 7.36131430e-01 -2.19794080e-01
9.08360481e-01 6.52147532e-01 1.01147436e-01 -9.41159427e-01
-2.79746532e-01 -4.38722223e-01 -6.72576308e-01 9.41557065e-02
1.03956246e+00 -8.35004210e-01 -6.99819028e-01 1.50285453e-01
-1.14621615e+00 2.78688818e-01 -5.50342739e-01 8.50638807e-01
-7.04540372e-01 2.36962035e-01 -2.87022531e-01 -1.08853233e+00
-2.72062689e-01 -1.17494297e+00 7.31844187e-01 1.17664970e-01
-1.14170149e-01 -1.25988948e+00 4.07912284e-01 5.70522733e-02
8.69384110e-02 5.48132241e-01 7.78605521e-01 -8.75598431e-01
-3.46755147e-01 -4.74187881e-01 5.27150214e-01 5.68084180e-01
3.69567782e-01 -2.36997351e-01 -1.24327266e+00 -3.58788580e-01
7.42080033e-01 1.32894874e-01 8.14110994e-01 6.38951540e-01
8.44408572e-01 -5.09210706e-01 -2.45116875e-01 8.27687502e-01
1.23873234e+00 5.22471488e-01 4.99480158e-01 -9.34430435e-02
5.70207655e-01 6.04841352e-01 -1.47470906e-01 3.20649177e-01
1.84293315e-02 5.81812739e-01 1.18104659e-01 -4.27358985e-01
4.37136479e-02 -4.04591523e-02 1.92544028e-01 1.10460496e+00
-3.06924731e-01 -5.70024550e-01 -6.64131343e-01 5.11971653e-01
-1.77823186e+00 -1.03039718e+00 2.18604386e-01 2.19261742e+00
6.59171402e-01 -3.79117668e-01 -2.51450408e-02 5.58243752e-01
7.62661040e-01 -2.48893518e-02 -6.43809736e-01 -1.51474521e-01
-2.42257357e-01 4.07875836e-01 4.01845753e-01 5.64222395e-01
-1.20786059e+00 2.49244645e-01 6.54964542e+00 2.71840036e-01
-1.28420627e+00 2.24070847e-01 4.28244442e-01 2.04912484e-01
-2.10912943e-01 -1.14687346e-02 -3.25719863e-01 4.22155410e-01
9.10565197e-01 -2.69402862e-01 5.28144777e-01 5.99676251e-01
-9.02144536e-02 1.49603888e-01 -1.13221121e+00 9.05008852e-01
2.73790151e-01 -7.63561487e-01 -2.54979908e-01 2.68634617e-01
7.93121457e-01 -7.34245837e-01 4.22959894e-01 -7.83967152e-02
4.60595489e-02 -1.10404921e+00 5.76152027e-01 5.52904189e-01
7.38273323e-01 -6.22472227e-01 5.65245926e-01 5.40324330e-01
-8.83075953e-01 -9.45764631e-02 -4.29468509e-03 -1.71535630e-02
8.56135786e-02 6.96788669e-01 -9.06854451e-01 4.47663695e-01
3.08915436e-01 5.81859231e-01 -1.16427697e-01 9.08698142e-01
-3.64147276e-01 1.00777030e+00 -1.58771381e-01 6.36895001e-01
-2.40996376e-01 -4.03131485e-01 9.09362793e-01 9.60134268e-01
5.20200610e-01 3.27113986e-01 -4.26077321e-02 7.69861877e-01
-3.09727609e-01 -3.24096791e-02 -6.35243952e-01 -2.68688679e-01
1.44581363e-01 1.00144231e+00 -7.93990552e-01 -4.53267932e-01
-3.13116848e-01 1.12492609e+00 -7.84278512e-02 5.86758196e-01
-6.92939281e-01 -2.35039696e-01 3.47229421e-01 -2.05292195e-01
3.63781601e-01 -2.24217340e-01 -2.63196290e-01 -1.54708993e+00
-5.20676002e-02 -9.68436539e-01 4.11548972e-01 -3.17306757e-01
-9.15447116e-01 7.92961001e-01 6.04419075e-02 -1.43018651e+00
-6.43686652e-01 -6.91368580e-02 -4.03305143e-01 1.09466231e+00
-1.36128044e+00 -1.02532113e+00 -1.17639400e-01 7.60514379e-01
4.39761400e-01 -2.59793490e-01 1.11992419e+00 1.79693073e-01
-5.25006711e-01 4.04054135e-01 5.71437180e-01 1.66946277e-01
5.18760085e-01 -1.31285393e+00 8.23058337e-02 1.06693709e+00
4.70495522e-01 8.77769828e-01 9.04732406e-01 -7.07210422e-01
-1.20511138e+00 -8.05936873e-01 8.60721290e-01 -8.57777894e-03
4.14947689e-01 -3.92519683e-01 -9.50455248e-01 1.04958403e+00
2.73808479e-01 1.66062266e-01 8.50083530e-01 -2.28768587e-01
-2.99545787e-02 1.61649376e-01 -1.11039340e+00 1.11690544e-01
4.15556550e-01 -5.05080819e-01 -6.18573010e-01 1.13504879e-01
4.41743851e-01 -3.16566199e-01 -8.85996282e-01 2.98405766e-01
6.71478689e-01 -5.32251418e-01 8.71489227e-01 -6.34203434e-01
2.86181301e-01 -1.98373601e-01 -1.37747228e-01 -1.22614646e+00
-3.39340180e-01 -7.30097651e-01 -5.39887428e-01 1.13649440e+00
1.86962962e-01 -6.93479240e-01 6.77671254e-01 5.55546284e-01
2.76784211e-01 -9.69771594e-02 -1.17460871e+00 -6.71820641e-01
-6.30110651e-02 -1.95832387e-01 3.81344378e-01 1.10857570e+00
-1.48372561e-01 3.28553557e-01 -9.16978300e-01 4.91619557e-01
9.03634191e-01 3.40059459e-01 4.41757709e-01 -1.35069406e+00
-6.72862470e-01 2.08729375e-02 -3.88365388e-01 -8.59030664e-01
4.17138070e-01 -8.68607283e-01 2.62987614e-01 -1.17930400e+00
2.39509448e-01 -7.09981620e-02 -5.00237048e-01 3.09069276e-01
-1.55423850e-01 2.19111905e-01 4.50572856e-02 3.41013134e-01
-6.57701269e-02 4.97590363e-01 8.18570793e-01 -1.85090065e-01
-4.25271392e-01 5.37715554e-01 -8.27827096e-01 7.53996074e-01
7.51342356e-01 -8.15704167e-01 -4.37987655e-01 -1.74301639e-01
-2.80479491e-01 5.13415396e-01 3.33573729e-01 -9.50215042e-01
3.01049978e-01 2.18501277e-02 4.06920463e-01 -2.70872861e-01
6.04043007e-01 -1.17562497e+00 6.82255745e-01 3.24584931e-01
-4.37720984e-01 -4.15510796e-02 -9.60450619e-02 5.78831553e-01
-5.01010239e-01 -3.05960685e-01 6.08224154e-01 1.37547016e-01
-4.78541590e-02 2.66200721e-01 -4.03706044e-01 -3.32690328e-01
9.53987181e-01 4.27988134e-02 -2.02332642e-02 -8.66859078e-01
-1.09533226e+00 -3.07873636e-01 2.39014432e-01 1.03039376e-01
6.83559179e-01 -1.18949914e+00 -5.82880139e-01 4.30417418e-01
-2.52465844e-01 -1.82979837e-01 2.75242865e-01 9.31763828e-01
6.83902949e-02 3.68714094e-01 4.10512276e-02 -4.70612466e-01
-1.33553064e+00 7.13309169e-01 1.72019079e-01 -1.09547891e-01
-6.72518075e-01 4.20842558e-01 4.48025644e-01 8.74927193e-02
1.32734969e-01 -1.32996261e-01 -3.87850910e-01 1.60511464e-01
4.34611320e-01 4.09093171e-01 -7.30374157e-02 -9.47435439e-01
-2.14171469e-01 4.52321887e-01 3.84849459e-01 -4.28932905e-01
1.48607230e+00 -1.39794677e-01 -2.20558584e-01 8.53706419e-01
1.45626795e+00 2.21754760e-01 -9.93658185e-01 -3.01933199e-01
-1.89763233e-01 -1.70326725e-01 1.87565863e-01 -3.89508843e-01
-1.19314957e+00 9.62998569e-01 7.63558209e-01 4.61524546e-01
1.37967420e+00 8.93274918e-02 5.18937528e-01 1.25893041e-01
6.23447262e-02 -5.86702764e-01 -2.86009818e-01 -5.37414700e-02
7.44794130e-01 -1.14737475e+00 1.45092890e-01 -2.33882204e-01
-5.93756735e-01 1.14664948e+00 -2.56728619e-01 -3.71335819e-02
7.47375965e-01 3.51357430e-01 -3.70155126e-02 4.93541099e-02
-6.53660059e-01 -1.87321156e-01 5.54322004e-01 4.54643816e-01
2.82085061e-01 1.88127190e-01 -1.21685833e-01 5.74459851e-01
-2.46063158e-01 6.00248314e-02 6.18151724e-01 8.31856132e-01
3.30861993e-02 -1.00207412e+00 -7.22816110e-01 2.46743232e-01
-8.63154233e-01 -6.28216043e-02 -9.98435989e-02 6.88158512e-01
1.47787258e-01 9.69128251e-01 -6.24394827e-02 -9.72623155e-02
-5.88918552e-02 3.60460907e-01 2.11730123e-01 -3.88779581e-01
-1.51352227e-01 6.29188716e-01 -3.74078512e-01 -2.50645960e-03
-7.47353613e-01 -9.60488260e-01 -9.98586237e-01 1.68896213e-01
-3.66759926e-01 2.31991097e-01 9.60322976e-01 8.41489911e-01
-6.92608207e-02 7.45829940e-01 8.22360516e-01 -3.78263861e-01
-2.68298030e-01 -9.15055275e-01 -8.88010681e-01 3.54618162e-01
1.07031262e+00 -5.60072124e-01 -6.20610476e-01 6.38294160e-01] | [15.235723495483398, 5.711827754974365] |
4da2cb4e-7163-4c14-8207-f95cdf8b2acb | material-classification-with-thermal-imagery | null | null | http://openaccess.thecvf.com/content_cvpr_2015/html/Saponaro_Material_Classification_With_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Saponaro_Material_Classification_With_2015_CVPR_paper.pdf | Material Classification With Thermal Imagery | Material classification is an important area of research in computer vision. Typical algorithms use color and texture information for classification, but there are problems due to varying lighting conditions and diversity of colors in a single material class. In this work we study the use of long wave infrared (i.e. thermal) imagery for material classification. Thermal imagery has the benefit of relative invariance to color changes, invariance to lighting conditions, and can even work in the dark. We collect a database of 21 different material classes with both color and thermal imagery. We develop a set of features that describe water permeation and heating/cooling properties, and test several variations on these methods to obtain our final classifier. The results show that the proposed method outperforms typical color and texture features, and when combined with color information, the results are improved further. | ['Scott Sorensen', 'Philip Saponaro', 'Chandra Kambhamettu', 'Abhishek Kolagunda'] | 2015-06-01 | null | null | null | cvpr-2015-6 | ['material-classification'] | ['computer-vision'] | [ 3.59784752e-01 -1.20514119e+00 -2.46073008e-01 -2.99965322e-01
-2.94053346e-01 -6.92265332e-01 6.91244125e-01 -1.02873631e-01
-1.44339621e-01 6.66512549e-01 -2.71219909e-01 -9.85291693e-03
-2.18464836e-01 -9.97108340e-01 -1.21712409e-01 -1.25632298e+00
3.19758914e-02 1.49262741e-01 3.60218167e-01 -1.33284569e-01
3.99994284e-01 7.46169746e-01 -1.75701940e+00 5.80601871e-01
7.59747505e-01 1.28376353e+00 3.35406333e-01 7.98310816e-01
-1.89968571e-01 4.19856817e-01 -3.05166423e-01 2.17394501e-01
4.43457156e-01 -3.02735299e-01 -1.03504992e+00 4.55980748e-01
5.02009809e-01 -2.19834879e-01 9.27466527e-02 9.74150836e-01
2.90300846e-01 1.58502802e-01 9.90824342e-01 -1.08279860e+00
-8.69777024e-01 4.34545316e-02 -7.52649724e-01 -3.01332064e-02
4.61740941e-01 -2.22702324e-02 6.93714380e-01 -6.27136886e-01
4.41866398e-01 1.06477726e+00 3.94679725e-01 4.37569648e-01
-1.17055094e+00 -3.52916986e-01 2.50590630e-02 5.70866227e-01
-1.29102027e+00 -1.31146714e-01 1.00555444e+00 -4.16408330e-01
5.89089334e-01 6.39962137e-01 8.02977979e-01 8.27439547e-01
3.56087983e-01 8.01836252e-01 1.88649893e+00 -6.37532413e-01
2.72647560e-01 2.77813345e-01 2.63412833e-01 6.61268830e-01
2.86192536e-01 2.77458578e-01 -3.81317884e-02 -4.02884260e-02
4.89844531e-01 6.73203990e-02 -1.39529973e-01 -2.10676864e-01
-1.04089928e+00 5.25301039e-01 5.82490861e-01 3.29583675e-01
-2.46010080e-01 -2.24319220e-01 1.28842697e-01 5.15759528e-01
6.69355929e-01 4.08044636e-01 -4.37716484e-01 1.46807224e-01
-6.88419282e-01 9.43171531e-02 6.93507910e-01 6.84184849e-01
9.71361876e-01 -8.60138908e-02 9.86322984e-02 1.39214659e+00
1.58104494e-01 1.02257621e+00 1.91644818e-01 -5.24653912e-01
3.88167277e-02 2.66577601e-01 9.62542146e-02 -9.32137728e-01
-5.03855109e-01 2.51142949e-01 -7.87305892e-01 4.92441088e-01
3.53760064e-01 2.39165738e-01 -1.26244688e+00 9.31153119e-01
1.61224931e-01 -5.51989794e-01 2.45455168e-02 8.87436867e-01
8.13193083e-01 8.23764086e-01 8.93746018e-02 -2.74678439e-01
1.13802898e+00 -8.90820205e-01 -5.54567337e-01 -1.34245574e-01
2.51108706e-01 -1.13761914e+00 8.58940899e-01 6.91135347e-01
-5.66487014e-01 -5.03676176e-01 -9.07174110e-01 3.60670745e-01
-9.95795012e-01 2.17318296e-01 7.72885561e-01 9.41187382e-01
-7.37634718e-01 5.76787353e-01 -5.79751909e-01 -5.91334343e-01
1.03564840e-02 1.27254888e-01 -2.52676070e-01 -4.69272643e-01
-8.21465075e-01 8.12712133e-01 4.20242578e-01 3.43339503e-01
-1.17908709e-01 -2.88706303e-01 -4.54755247e-01 -7.10598230e-01
1.00333244e-01 -4.53012856e-03 6.93173766e-01 -1.18714392e+00
-1.50867915e+00 8.49039555e-01 1.44906323e-02 4.79859620e-01
5.40307343e-01 1.50056125e-03 -7.26423323e-01 3.14965338e-01
-2.97695369e-01 3.41963470e-01 8.78686428e-01 -1.71955550e+00
-4.03336555e-01 -2.85558373e-01 -4.38575894e-02 -1.14311755e-01
-2.72663653e-01 7.59577453e-02 -5.40969491e-01 -4.09564525e-01
3.48892689e-01 -1.26897907e+00 8.97034258e-02 6.68887123e-02
-5.40194571e-01 -4.87163141e-02 1.08166122e+00 -6.25715554e-01
6.54620051e-01 -2.02486372e+00 -1.26632273e-01 5.86983800e-01
-3.50194544e-01 1.78234175e-01 -2.54933655e-01 3.75044912e-01
-1.91786245e-01 1.30578831e-01 -1.88566655e-01 4.26846981e-01
-3.75514060e-01 2.04634726e-01 2.70291120e-01 6.43884659e-01
1.91055313e-01 5.59700429e-01 -6.96171761e-01 -5.35477817e-01
7.01406360e-01 4.08160537e-01 1.17413320e-01 -1.06389031e-01
-2.61755195e-02 2.94824630e-01 -5.41659236e-01 1.15669084e+00
1.16687441e+00 2.85043150e-01 -4.59544659e-02 -6.08980715e-01
-2.62953520e-01 -3.02542835e-01 -1.21609807e+00 1.15050709e+00
-5.06693959e-01 6.65995479e-01 8.75117257e-02 -1.00551128e+00
9.96963024e-01 9.23947096e-02 7.80451417e-01 -9.93644297e-01
1.29837349e-01 2.28524759e-01 -1.86479196e-01 -8.74217868e-01
5.65007031e-01 -1.31732121e-01 2.99465597e-01 4.33477938e-01
-4.86516833e-01 -3.80297691e-01 3.79148602e-01 -3.66614133e-01
3.99290830e-01 3.13487202e-01 -1.74206361e-01 -5.46997011e-01
6.06848896e-01 1.37104407e-01 1.32230133e-01 4.63884085e-01
6.34069592e-02 5.33496439e-01 -1.26991764e-01 -6.21966243e-01
-1.10330832e+00 -9.74424779e-01 -5.42581856e-01 1.15436220e+00
4.75939512e-01 1.90933317e-01 -3.87376577e-01 -2.10224390e-01
1.93753406e-01 2.55592465e-01 -6.30742550e-01 -3.28116342e-02
-4.62954372e-01 -1.21534443e+00 -1.45939693e-01 5.13566375e-01
8.24885249e-01 -1.03449082e+00 -3.47906172e-01 -4.00270447e-02
-1.33743703e-01 -9.52064574e-01 1.03643931e-01 6.04608580e-02
-8.74839187e-01 -1.33883893e+00 -7.08724082e-01 -6.44538760e-01
8.35255265e-01 7.43172228e-01 1.03042305e+00 1.15203999e-01
-8.43309760e-01 6.32968605e-01 -9.63793695e-01 -3.76202106e-01
-5.35494089e-01 -4.76256981e-02 -2.63862044e-01 1.71055093e-01
2.78536081e-01 1.64180025e-01 -5.86857796e-01 5.61563969e-01
-1.05236995e+00 5.15117422e-02 6.09988272e-01 7.52559245e-01
4.37106460e-01 6.39851868e-01 -1.22348644e-01 -9.37446654e-01
2.48545915e-01 -1.47172123e-01 -4.67240691e-01 5.90191543e-01
-7.06709802e-01 9.36138853e-02 5.29850900e-01 -4.14005071e-01
-1.27731752e+00 -5.73864579e-02 2.75276214e-01 2.04300955e-01
-4.40674961e-01 3.53031844e-01 -1.27596691e-01 -6.34694695e-01
6.57086730e-01 3.73752415e-02 -1.46228328e-01 -6.20276153e-01
1.27996653e-01 8.38896096e-01 1.26171097e-01 -9.04694378e-01
1.01944697e+00 5.12629509e-01 2.75596142e-01 -1.42294502e+00
-4.59377259e-01 -5.70650041e-01 -8.26702774e-01 -7.41429985e-01
8.23137343e-01 -4.57837492e-01 -4.01010245e-01 9.61168885e-01
-5.29713690e-01 -2.15035886e-01 2.40010649e-01 5.26751101e-01
-3.74171823e-01 5.10983825e-01 -6.11944675e-01 -1.04707193e+00
-2.23751590e-01 -1.06909943e+00 7.90839851e-01 1.37598425e-01
3.17438304e-01 -1.22100461e+00 -1.80355817e-01 2.79101789e-01
4.78292912e-01 4.14744794e-01 9.65954542e-01 4.44819927e-01
-4.13510829e-01 -3.38723034e-01 -4.20820445e-01 4.73658174e-01
7.31408715e-01 6.42278790e-01 -1.20590615e+00 -3.22119594e-01
-4.33610886e-01 -3.03827077e-01 1.34677505e+00 4.90498841e-01
1.28986418e+00 7.68602043e-02 -1.77565262e-01 5.69175482e-01
1.87568450e+00 4.58143353e-01 7.23958611e-01 6.51428580e-01
7.86955178e-01 8.68725002e-01 7.83826888e-01 1.35537013e-01
-2.59828120e-01 3.74755472e-01 2.92578667e-01 -6.17050111e-01
-8.99786353e-02 4.93828654e-01 1.33139774e-01 4.77761358e-01
-8.03928018e-01 -2.12680891e-01 -8.26889157e-01 2.05916747e-01
-1.38270068e+00 -1.12023365e+00 -4.09774393e-01 2.19985318e+00
7.16147006e-01 -1.48331270e-01 2.60423660e-01 5.84290326e-01
7.80403137e-01 9.46728215e-02 -2.98031062e-01 -5.18900454e-01
-2.59119362e-01 2.03249440e-01 9.31523442e-01 1.09430537e-01
-1.43764055e+00 5.86718440e-01 7.52050924e+00 5.70240021e-01
-1.47579956e+00 -4.98095185e-01 6.40221357e-01 4.14684176e-01
-1.52593374e-01 -2.69350648e-01 -2.36599699e-01 1.43099397e-01
4.55888718e-01 4.17092353e-01 6.41583562e-01 4.69060779e-01
1.78111702e-01 -7.00589716e-01 -8.14893961e-01 9.69501734e-01
1.41622573e-01 -7.09303200e-01 -1.34827286e-01 9.46085434e-03
8.85528207e-01 -2.18167692e-01 2.52823472e-01 -3.95971060e-01
6.84227943e-02 -7.75498569e-01 6.27975345e-01 5.01050293e-01
8.57943475e-01 -6.23694837e-01 7.79256344e-01 -2.60711104e-01
-1.47534311e+00 -1.38679206e-01 -5.25062442e-01 6.70196787e-02
-5.00859737e-01 5.25501192e-01 -3.64917278e-01 7.88763702e-01
8.13977122e-01 8.00880134e-01 -8.93735945e-01 1.16800475e+00
1.20838456e-01 1.80541605e-01 -4.57780719e-01 -3.84840220e-01
2.04549223e-01 -6.90617025e-01 -1.20171912e-01 1.30632412e+00
2.10615262e-01 -1.08056270e-01 5.00753284e-01 4.86526489e-01
5.33466935e-01 2.51480848e-01 -4.89428014e-01 -1.20315649e-01
9.93938819e-02 1.38797331e+00 -1.19383132e+00 -3.77090603e-01
-6.01213336e-01 9.08437908e-01 -4.05195564e-01 5.91808379e-01
-5.47904193e-01 -2.58588731e-01 3.90653521e-01 -2.65597254e-01
9.82125625e-02 -3.41151386e-01 -3.27613115e-01 -9.39052939e-01
1.29522076e-02 -3.50500673e-01 3.49554300e-01 -9.44951355e-01
-1.57583916e+00 2.79455572e-01 2.59310484e-01 -1.49761951e+00
4.42584276e-01 -1.22548985e+00 -5.12841046e-01 8.09129000e-01
-1.79949498e+00 -1.37396646e+00 -6.41249895e-01 6.45668626e-01
4.93250370e-01 5.16561382e-02 8.83845687e-01 6.81249723e-02
-5.84146500e-01 1.25999361e-01 6.26022696e-01 1.31818086e-01
8.61847103e-01 -1.35762179e+00 -2.45345443e-01 5.04975498e-01
-1.89837724e-01 1.53439477e-01 7.14684248e-01 -4.17053699e-01
-1.78975153e+00 -1.04363441e+00 5.99895902e-02 3.49265262e-02
2.88552105e-01 4.12610658e-02 -7.00489998e-01 2.50215773e-02
1.53116167e-01 -1.68321848e-01 7.89970934e-01 9.85350683e-02
-4.91474688e-01 -4.39129889e-01 -1.25633419e+00 1.63800180e-01
4.33680773e-01 -5.47448277e-01 -2.16965377e-01 4.71212268e-01
-2.34522209e-01 2.97985226e-02 -9.95601356e-01 2.33871102e-01
1.00198519e+00 -9.35192764e-01 1.13804984e+00 -9.30519949e-04
1.47839412e-01 -2.62210369e-01 -9.00468528e-02 -1.20310116e+00
-6.27236605e-01 1.16042003e-01 8.03493738e-01 1.00421917e+00
5.79333901e-01 -5.70616543e-01 6.52386487e-01 7.75695503e-01
2.79894602e-02 -4.06294055e-02 -3.59836251e-01 -1.09916270e+00
1.31572291e-01 -3.76139641e-01 3.75643104e-01 9.02854621e-01
-2.19401732e-01 -1.67446345e-01 -4.76161480e-01 5.61794564e-02
8.22255492e-01 6.37785375e-01 3.40433776e-01 -1.55501449e+00
1.23123102e-01 -4.98572409e-01 -3.71681809e-01 -1.91217974e-01
-8.65388066e-02 -6.44191086e-01 2.41103590e-01 -1.62645268e+00
4.50354546e-01 -6.78711474e-01 -4.91825908e-01 5.27494609e-01
-1.39253527e-01 7.27883756e-01 5.69896884e-02 1.55238241e-01
-1.62504062e-01 1.14453346e-01 1.48305392e+00 -7.22873986e-01
-3.75051349e-01 -1.92409884e-02 -1.30313396e-01 4.50672030e-01
1.10640001e+00 5.62733971e-02 -1.41675755e-01 -3.07133973e-01
3.01701799e-02 -4.86234456e-01 2.53935874e-01 -1.00899780e+00
-3.42109501e-01 -7.60336339e-01 9.24349010e-01 -5.40636301e-01
4.39129919e-01 -1.23379898e+00 2.00659528e-01 6.25039458e-01
-9.46293399e-02 -1.89256117e-01 3.82883102e-02 3.32659364e-01
-6.52053207e-02 -1.93693444e-01 1.16564047e+00 -3.17013085e-01
-1.23863029e+00 2.93209285e-01 -4.76871759e-01 -6.08261645e-01
1.06976652e+00 -6.54292405e-01 -5.71963608e-01 7.54052922e-02
-4.46548551e-01 -2.34572619e-01 8.28353882e-01 3.89768451e-01
4.96975392e-01 -1.47169054e+00 -4.92889136e-01 3.36268574e-01
4.94846195e-01 -5.33672750e-01 2.26990730e-01 6.34097874e-01
-1.11332726e+00 1.87036738e-01 -6.61106467e-01 -7.08795786e-01
-1.49734044e+00 5.96215487e-01 2.45736107e-01 3.53817582e-01
-4.67474163e-01 5.42807281e-01 -1.01505876e-01 1.06286742e-02
-2.30524361e-01 -3.65314215e-01 -3.94684970e-01 6.34695888e-02
4.68386501e-01 5.81195712e-01 2.13590220e-01 -6.89688861e-01
-4.72049654e-01 1.34301019e+00 1.54666930e-01 1.31338805e-01
1.43687189e+00 1.01699196e-02 -5.63232601e-01 6.53743744e-01
1.24860334e+00 -2.33898133e-01 -7.27904320e-01 -2.13468865e-01
-1.31746426e-01 -9.17312324e-01 3.51307184e-01 -8.46897602e-01
-1.35685313e+00 7.18832016e-01 1.05303693e+00 7.30113029e-01
1.55631649e+00 -2.37265691e-01 4.14857000e-01 5.50515890e-01
2.73426801e-01 -1.70408714e+00 -5.99220656e-02 3.09709102e-01
7.20588148e-01 -1.40701950e+00 4.39204872e-01 -7.86923826e-01
-5.13823688e-01 1.77362061e+00 4.40105915e-01 1.90984420e-02
7.26677895e-01 1.21822417e-01 3.72320950e-01 -1.60214771e-02
-3.47447127e-01 -4.31840777e-01 4.39966857e-01 9.05930758e-01
6.60106242e-01 3.88964176e-01 -2.46277973e-01 -7.62639880e-01
1.28975332e-01 -3.07865232e-01 3.76860827e-01 1.20061255e+00
-6.53783202e-01 -1.16386926e+00 -8.89705062e-01 5.80844939e-01
-1.84345201e-01 2.88153678e-01 -5.49537063e-01 6.46886468e-01
1.92146108e-01 1.34859550e+00 -3.82029191e-02 -6.36931717e-01
8.69631246e-02 5.65063916e-02 7.20608652e-01 -5.59023097e-02
-9.94654670e-02 4.36791539e-01 2.77639151e-01 -4.08924639e-01
-1.24633873e+00 -6.89754665e-01 -6.51175201e-01 -5.57962179e-01
-6.43380761e-01 -7.10245455e-03 1.07011187e+00 8.04871738e-01
-5.55474877e-01 3.66718948e-01 1.22677386e+00 -9.32825387e-01
9.15001780e-02 -9.66697812e-01 -1.22444534e+00 7.76868105e-01
3.90507340e-01 -8.18057775e-01 -2.66009361e-01 3.81589562e-01] | [10.304221153259277, -2.5051040649414062] |
3512dc73-bc03-4220-863a-c418a2e6e061 | outlining-and-filling-hierarchical-query | 2111.00732 | null | https://arxiv.org/abs/2111.00732v2 | https://arxiv.org/pdf/2111.00732v2.pdf | Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graphs | Query graph construction aims to construct the correct executable SPARQL on the KG to answer natural language questions. Although recent methods have achieved good results using neural network-based query graph ranking, they suffer from three new challenges when handling more complex questions: 1) complicated SPARQL syntax, 2) huge search space, and 3) locally ambiguous query graphs. In this paper, we provide a new solution. As a preparation, we extend the query graph by treating each SPARQL clause as a subgraph consisting of vertices and edges and define a unified graph grammar called AQG to describe the structure of query graphs. Based on these concepts, we propose a novel end-to-end model that performs hierarchical autoregressive decoding to generate query graphs. The high-level decoding generates an AQG as a constraint to prune the search space and reduce the locally ambiguous query graph. The bottom-level decoding accomplishes the query graph construction by selecting appropriate instances from the preprepared candidates to fill the slots in the AQG. The experimental results show that our method greatly improves the SOTA performance on complex KGQA benchmarks. Equipped with pre-trained models, the performance of our method is further improved, achieving SOTA for all three datasets used. | ['Tenggou Wang', 'Tianxing Wu', 'Guilin Qi', 'Huiying Li', 'Yongrui Chen'] | 2021-11-01 | null | null | null | null | ['graph-ranking'] | ['graphs'] | [-6.88821450e-02 2.96006918e-01 -2.34082699e-01 -6.01892054e-01
-1.19836080e+00 -4.71117824e-01 4.90656234e-02 2.77312934e-01
-2.13280186e-01 3.46685350e-01 3.23187053e-01 -4.22373921e-01
-2.30260044e-01 -1.33394217e+00 -9.59167957e-01 -1.94586605e-01
8.43247399e-02 9.48046625e-01 5.38278699e-01 -3.95511389e-01
-4.32614349e-02 1.29455417e-01 -1.52951264e+00 4.05335993e-01
1.31690228e+00 9.73211229e-01 2.62761235e-01 5.11656225e-01
-6.78180397e-01 1.01606894e+00 -5.12230337e-01 -5.64598441e-01
-4.04487513e-02 -3.48923832e-01 -1.07707524e+00 -3.73408318e-01
5.40703952e-01 -1.63945675e-01 -2.96472400e-01 1.07402432e+00
4.93002504e-01 9.34579223e-02 -3.34452614e-02 -1.12371349e+00
-5.29439449e-01 9.13535833e-01 5.67018576e-02 -8.84241909e-02
5.26315272e-01 -2.84260362e-01 1.66922009e+00 -6.63575470e-01
5.61045706e-01 1.33875513e+00 1.19190283e-01 3.76217127e-01
-8.22629094e-01 -3.90319675e-01 2.28176862e-01 4.31454360e-01
-1.41582072e+00 -1.71201080e-01 7.67411053e-01 -3.80989052e-02
1.19928372e+00 6.58362269e-01 5.78204453e-01 3.55732232e-01
1.72329575e-01 8.01850736e-01 4.55523074e-01 -3.45456183e-01
2.80645221e-01 -3.19269419e-01 5.11646271e-01 1.13359439e+00
6.94920868e-02 -2.70081162e-01 -2.68408835e-01 -4.93435770e-01
4.13237870e-01 -2.22201422e-01 -7.07005188e-02 -6.62992537e-01
-9.68089223e-01 1.05299342e+00 7.17051268e-01 4.91552912e-02
-3.78257841e-01 1.95562124e-01 3.13656151e-01 2.35448033e-01
8.85244459e-02 5.46096861e-01 -5.76743901e-01 3.88248742e-01
-3.80135357e-01 3.69391471e-01 1.03325427e+00 1.27935386e+00
1.09767008e+00 -2.30241567e-01 -6.30475640e-01 8.47728550e-01
3.49257022e-01 5.80705166e-01 2.11429596e-02 -7.73245215e-01
7.81675339e-01 1.20085108e+00 -1.77813992e-01 -1.05049431e+00
-5.27340889e-01 -5.72945952e-01 -4.48317677e-01 -5.43548584e-01
5.79079986e-02 1.81966096e-01 -9.14167225e-01 1.71481633e+00
5.30518293e-01 -1.29305944e-01 2.42077291e-01 9.06136453e-01
1.18195510e+00 8.28408599e-01 3.88948292e-01 2.22144034e-02
1.58983648e+00 -1.15236676e+00 -5.71851373e-01 -3.75589341e-01
1.03248906e+00 -2.73024708e-01 1.56441462e+00 1.98728338e-01
-8.87973666e-01 -2.69842058e-01 -9.35081720e-01 -5.28198779e-01
-3.02612334e-01 1.32535733e-02 8.81016076e-01 3.92242402e-01
-1.12869287e+00 -4.54997271e-02 -5.87594271e-01 -2.37041622e-01
4.74900715e-02 3.61059606e-01 -1.14588119e-01 -2.85122454e-01
-1.51478374e+00 4.43159461e-01 8.86982024e-01 -3.22686099e-02
-6.48398042e-01 -5.91122806e-01 -1.18450141e+00 4.24221277e-01
9.81546283e-01 -9.90357041e-01 1.14893389e+00 -1.95427611e-01
-1.20311201e+00 6.91138864e-01 -4.69541728e-01 -2.79211253e-01
-2.63686836e-01 -1.22778244e-01 -4.27472711e-01 -6.84237927e-02
1.71547681e-01 4.13212121e-01 3.97855967e-01 -9.32874918e-01
-8.06253672e-01 -6.63660347e-01 3.89234871e-01 3.54770511e-01
-1.14176702e-02 -6.53603347e-03 -1.29343379e+00 -2.69647747e-01
3.77088606e-01 -7.52228260e-01 -3.11011016e-01 -8.96137714e-01
-5.15089214e-01 -7.99558103e-01 3.22848350e-01 -6.51686847e-01
1.77232730e+00 -1.87795269e+00 2.79273868e-01 3.49249154e-01
4.60406899e-01 6.15780130e-02 -2.62333632e-01 4.65641975e-01
2.47796193e-01 6.26570806e-02 -6.53565302e-02 1.67775944e-01
2.91161031e-01 4.59842503e-01 -4.02206242e-01 -1.57910466e-01
8.10048953e-02 1.36759281e+00 -9.23420429e-01 -6.96743846e-01
-2.79899776e-01 4.38472107e-02 -9.76759017e-01 4.37800318e-01
-9.96943057e-01 -7.52317021e-03 -9.75603461e-01 7.39314437e-01
5.01748145e-01 -6.64494276e-01 3.74053806e-01 -3.34311694e-01
3.37364674e-01 6.71956420e-01 -1.03498125e+00 1.89833152e+00
-1.65068567e-01 -1.70380011e-01 -1.67233013e-02 -8.76537383e-01
1.01827300e+00 -1.19027555e-01 2.62618452e-01 -1.13921535e+00
-1.28913537e-01 2.47348711e-01 -2.45107919e-01 -7.67608643e-01
6.33692682e-01 1.58385754e-01 -7.04519391e-01 2.19708741e-01
5.34178987e-02 -2.50397414e-01 3.92363667e-01 4.93295491e-01
1.14761889e+00 -9.03952271e-02 8.84978250e-02 -1.17340691e-01
8.64574254e-01 2.07198918e-01 7.57873416e-01 7.43805170e-01
2.39590272e-01 1.57775447e-01 8.09687674e-01 -5.30972004e-01
-6.99942529e-01 -9.67963398e-01 3.54493678e-01 1.58332562e+00
1.46549135e-01 -8.39016676e-01 -7.56972611e-01 -9.51266587e-01
1.33167882e-03 8.73049498e-01 -1.81525260e-01 -2.22510144e-01
-9.06277597e-01 -5.88390291e-01 4.69082445e-01 2.78175712e-01
2.99226344e-01 -1.17354453e+00 -2.31681451e-01 2.43354231e-01
-4.62541550e-01 -1.15014625e+00 -5.83479106e-01 -4.90299240e-02
-6.51308119e-01 -1.23615980e+00 7.26129636e-02 -1.01300776e+00
5.23220599e-01 1.04326904e-01 1.63771689e+00 3.04487675e-01
9.73107740e-02 2.29434177e-01 -3.42804819e-01 -3.19328576e-01
-2.63161361e-01 3.70771050e-01 -5.00019610e-01 -3.00753266e-01
5.10408044e-01 -1.27457991e-01 -3.72236311e-01 1.81340560e-01
-1.13021350e+00 8.41563568e-03 6.38073564e-01 6.59750462e-01
1.18254030e+00 7.73264468e-02 6.05111659e-01 -1.35252726e+00
7.51715660e-01 -4.30691242e-01 -1.04130495e+00 7.68141747e-01
-8.51883054e-01 7.11665154e-01 6.84406042e-01 1.87001318e-01
-7.57887900e-01 -3.13881673e-02 -4.36413348e-01 -1.74053907e-01
2.69589186e-01 1.14422631e+00 -6.37471199e-01 1.69563159e-01
5.70992053e-01 1.25092849e-01 -2.39239931e-01 -5.37813067e-01
6.08226299e-01 3.12773675e-01 5.54340065e-01 -7.79186010e-01
6.81821883e-01 9.49150100e-02 -3.35009769e-02 -3.54711980e-01
-1.02065933e+00 -4.99219716e-01 -2.70871133e-01 1.04183286e-01
8.96539629e-01 -8.86618674e-01 -8.32128227e-01 -4.57296968e-02
-1.04965341e+00 -2.09171295e-01 -6.52038977e-02 2.09232330e-01
-4.23951775e-01 3.92962843e-01 -4.84024853e-01 -4.42628950e-01
-8.07233334e-01 -1.25967336e+00 1.16986394e+00 1.87681347e-01
1.87628731e-01 -6.83828235e-01 2.04574361e-01 6.53672695e-01
1.97918475e-01 -9.05177277e-03 1.69125378e+00 -8.38336766e-01
-1.07091546e+00 -7.76875019e-02 -2.89944082e-01 -1.84681192e-02
-2.98264354e-01 -4.56539512e-01 -4.27250922e-01 -3.30298662e-01
-1.79752097e-01 -3.72383922e-01 6.53230965e-01 -6.13359101e-02
1.19651437e+00 -4.19788867e-01 -2.61523873e-01 8.03946674e-01
1.60116732e+00 1.46795511e-01 6.02936506e-01 2.19224200e-01
9.16580975e-01 4.76326525e-01 7.08797336e-01 7.16690198e-02
9.56003904e-01 4.95121658e-01 7.33153045e-01 2.14159787e-01
1.51219875e-01 -7.92481840e-01 1.89520836e-01 1.06550884e+00
5.22349238e-01 -4.65067208e-01 -1.04635167e+00 3.90598178e-01
-1.91776860e+00 -3.90805453e-01 -2.93902248e-01 2.16369772e+00
6.66245699e-01 -1.08139195e-01 -8.37441757e-02 -3.44252676e-01
4.20851737e-01 3.87836426e-01 -5.63809514e-01 -2.56199926e-01
-1.95834666e-01 3.60438764e-01 3.31843287e-01 7.38442004e-01
-9.43156362e-01 1.43336058e+00 5.74290085e+00 6.61542773e-01
-8.25713038e-01 -9.92767215e-02 1.72672257e-01 4.67719175e-02
-8.81628335e-01 2.45410204e-01 -9.84422743e-01 2.23728374e-01
8.65684986e-01 -2.20272318e-01 6.68899953e-01 8.97066832e-01
-2.41994813e-01 1.65479288e-01 -7.82387912e-01 8.40552866e-01
1.54100984e-01 -1.24972355e+00 5.46959162e-01 -2.87540972e-01
2.72573382e-01 2.04729557e-01 -4.12153214e-01 9.58999097e-01
4.21489596e-01 -8.48256111e-01 4.49897915e-01 5.55597723e-01
7.21915483e-01 -8.38495553e-01 5.65762222e-01 4.29203182e-01
-1.33492637e+00 -1.85746714e-01 -6.80718720e-01 3.10303718e-01
-1.05243064e-01 3.87376517e-01 -9.33998585e-01 9.86879766e-01
5.53979516e-01 1.51248381e-01 -7.14245617e-01 8.27712953e-01
-5.09525061e-01 4.58399653e-01 -1.88755482e-01 -2.77379006e-01
1.57886162e-01 -4.03100282e-01 4.73133951e-01 8.32313061e-01
2.66079873e-01 2.90708035e-01 2.99984694e-01 9.69197929e-01
-3.80097985e-01 5.85191965e-01 -3.09395671e-01 -2.76400298e-01
5.67481279e-01 1.16963685e+00 -2.24918991e-01 -3.63798231e-01
-5.42586744e-01 6.53309047e-01 8.59721005e-01 5.31087637e-01
-7.67753661e-01 -4.28699940e-01 3.46222609e-01 4.46957052e-02
1.81990474e-01 -8.77442304e-03 1.44570813e-01 -1.27695537e+00
3.34009826e-01 -1.19608986e+00 1.23476291e+00 -8.50134730e-01
-9.24680531e-01 7.69722879e-01 -1.13062762e-01 -5.80299854e-01
-4.60275322e-01 -3.46744061e-01 -3.11160386e-01 8.74125838e-01
-1.39226902e+00 -1.20101202e+00 -4.66138273e-01 6.63616002e-01
2.83818394e-02 -3.73252295e-02 9.20370519e-01 5.33796012e-01
-5.79966843e-01 6.37436450e-01 -3.05842459e-01 2.03711316e-01
4.54365849e-01 -1.32517862e+00 4.34425950e-01 8.41148019e-01
3.00881505e-01 6.69434547e-01 3.69896501e-01 -7.64515400e-01
-2.04880810e+00 -1.15031648e+00 1.36859334e+00 -4.49950129e-01
4.70173508e-01 -5.02066016e-01 -1.10544252e+00 9.11773205e-01
-1.42996386e-01 -1.56183735e-01 5.13028562e-01 4.92196888e-01
-4.90829289e-01 -3.66861194e-01 -7.02791512e-01 3.81948382e-01
9.66321826e-01 -6.64547324e-01 -6.48060024e-01 3.97455037e-01
1.36921310e+00 -6.66131794e-01 -9.01072025e-01 6.57703221e-01
1.92304015e-01 -7.57506907e-01 8.64224136e-01 -9.54549849e-01
-6.96575595e-03 -4.90610868e-01 -2.51155168e-01 -8.82389724e-01
-3.79002601e-01 -5.10725021e-01 -2.80324399e-01 8.92405391e-01
6.52298570e-01 -5.12150466e-01 1.21178305e+00 7.01787591e-01
-4.02701497e-01 -1.04696929e+00 -8.10728133e-01 -3.48589629e-01
-1.99735612e-01 -5.96227586e-01 1.21207464e+00 7.04728246e-01
-1.82409361e-02 9.12827730e-01 -8.24137181e-02 4.80298609e-01
4.46663648e-01 6.75751448e-01 1.01221144e+00 -1.27074015e+00
-2.12825105e-01 -1.88675955e-01 -3.47139090e-02 -1.42774892e+00
2.48236701e-01 -1.19594729e+00 -2.56483275e-02 -2.01574063e+00
1.53847545e-01 -6.44390345e-01 -2.15299234e-01 5.20689666e-01
-5.57939053e-01 -5.91804266e-01 8.53996500e-02 9.14644301e-02
-1.11414587e+00 5.25857031e-01 1.27503014e+00 -2.37505183e-01
-2.43834913e-01 -1.07362412e-01 -9.42532778e-01 1.83928862e-01
5.31784832e-01 -4.70872521e-01 -8.10844064e-01 -8.24601889e-01
9.42536950e-01 4.37090546e-01 1.62285283e-01 -7.06517041e-01
5.20021737e-01 -1.13632016e-01 -3.16070437e-01 -8.85885358e-01
4.59603779e-02 -6.47048831e-01 1.14744380e-01 3.19335431e-01
-3.25473547e-01 2.35666901e-01 -3.14845219e-02 6.21906877e-01
-4.88383442e-01 -1.08119093e-01 3.60472202e-01 -1.35274697e-02
-9.06879008e-01 6.48852289e-01 3.83651525e-01 3.89052331e-01
5.83553731e-01 4.15281594e-01 -4.66070741e-01 -4.60339755e-01
-6.95464194e-01 8.96793127e-01 2.27654219e-01 5.24615586e-01
5.55289805e-01 -1.32728875e+00 -7.58419275e-01 4.12051558e-01
3.99850696e-01 4.44136709e-01 2.45776996e-01 4.44304436e-01
-6.54035330e-01 9.66583729e-01 2.82487124e-01 -3.85952175e-01
-9.62035179e-01 8.41268957e-01 3.23560566e-01 -8.11479986e-01
-3.36380482e-01 9.06368613e-01 3.18525016e-01 -8.53816748e-01
2.91485250e-01 -4.92070824e-01 -3.66883844e-01 -2.41540939e-01
1.76154748e-01 6.25449745e-03 2.88641155e-01 -4.88132119e-01
-2.51687586e-01 2.20412344e-01 -8.40368867e-02 3.23555648e-01
1.10772300e+00 7.06921741e-02 -6.82887554e-01 -4.47022431e-02
1.14950895e+00 2.41026297e-01 -4.58639294e-01 -5.09987175e-01
4.44649875e-01 -2.31790617e-01 -1.69007286e-01 -6.22900665e-01
-1.00911522e+00 5.74029744e-01 7.01835603e-02 3.08916152e-01
1.27068031e+00 3.53344709e-01 1.08843136e+00 7.43141651e-01
5.82530618e-01 -7.91752219e-01 -1.45923316e-01 8.38974178e-01
8.95699024e-01 -9.18316543e-01 -2.09000260e-01 -6.72453403e-01
-4.91178691e-01 7.62723625e-01 7.61487365e-01 1.25356764e-01
2.52099127e-01 -2.21188113e-01 -3.98738049e-02 -6.58108354e-01
-8.99047613e-01 -4.29061562e-01 6.45146668e-01 1.62912145e-01
2.40704834e-01 2.33744178e-02 -4.90512699e-01 1.00927365e+00
-3.86885196e-01 -1.20794334e-01 -4.63628676e-03 6.45910740e-01
-6.81593537e-01 -1.29218531e+00 -2.74784435e-02 6.55402899e-01
-3.26858431e-01 -2.69347310e-01 -4.22013968e-01 7.44446993e-01
-1.77984074e-01 9.37176526e-01 -1.04269218e-02 -5.96079230e-01
8.47033918e-01 2.29218572e-01 2.17641070e-01 -8.34259927e-01
-4.56553817e-01 -1.00469306e-01 4.38171744e-01 -8.58683586e-01
1.91363484e-01 -8.57276022e-02 -1.66164887e+00 1.00450628e-02
-2.89165705e-01 7.42506206e-01 3.54841679e-01 7.22779393e-01
8.07086170e-01 4.22487706e-01 2.31279284e-01 2.56297052e-01
-5.70834041e-01 -7.57152736e-01 -3.68837118e-01 4.17305022e-01
-1.61923245e-02 -2.73759782e-01 7.89470598e-02 -3.77469391e-01] | [10.037015914916992, 7.82706356048584] |
afcca557-e1d0-48ad-9612-08ee9435a213 | seeking-salient-facial-regions-for-cross | 2111.15361 | null | https://arxiv.org/abs/2111.15361v3 | https://arxiv.org/pdf/2111.15361v3.pdf | Seeking Salient Facial Regions for Cross-Database Micro-Expression Recognition | Cross-Database Micro-Expression Recognition (CDMER) aims to develop the Micro-Expression Recognition (MER) methods with strong domain adaptability, i.e., the ability to recognize the Micro-Expressions (MEs) of different subjects captured by different imaging devices in different scenes. The development of CDMER is faced with two key problems: 1) the severe feature distribution gap between the source and target databases; 2) the feature representation bottleneck of ME such local and subtle facial expressions. To solve these problems, this paper proposes a novel Transfer Group Sparse Regression method, namely TGSR, which aims to 1) optimize the measurement and better alleviate the difference between the source and target databases, and 2) highlight the valid facial regions to enhance extracted features, by the operation of selecting the group features from the raw face feature, where each region is associated with a group of raw face feature, i.e., the salient facial region selection. Compared with previous transfer group sparse methods, our proposed TGSR has the ability to select the salient facial regions, which is effective in alleviating the aforementioned problems for better performance and reducing the computational cost at the same time. We use two public ME databases, i.e., CASME II and SMIC, to evaluate our proposed TGSR method. Experimental results show that our proposed TGSR learns the discriminative and explicable regions, and outperforms most state-of-the-art subspace-learning-based domain-adaptive methods for CDMER. | ['Mengting Wei', 'Jiateng Liu', 'Wenming Zheng', 'Yuan Zong', 'Xingxun Jiang'] | 2021-11-30 | null | null | null | null | ['micro-expression-recognition'] | ['computer-vision'] | [ 5.98500036e-02 -5.55875540e-01 -1.21457450e-01 -3.66522372e-01
-7.62004793e-01 -4.93820533e-02 8.68394300e-02 -6.67870224e-01
-2.40859706e-02 4.15978640e-01 3.09664935e-01 6.61315262e-01
-3.56424376e-02 -3.63986254e-01 -2.59076923e-01 -1.16981304e+00
2.08449915e-01 -9.20631811e-02 -2.02689573e-01 -2.84115106e-01
-1.32130221e-01 6.44370258e-01 -1.76377916e+00 4.47422415e-01
7.94562280e-01 1.50276232e+00 1.11304693e-01 -2.44383350e-01
-7.68065155e-02 7.43212104e-01 -4.53597158e-01 -1.30758598e-01
1.13774322e-01 -5.87923586e-01 -1.19578697e-01 3.05010736e-01
4.85050499e-01 -3.38680714e-01 -1.96314901e-01 1.18618333e+00
7.80176699e-01 4.54880148e-02 5.16024947e-01 -1.36215138e+00
-5.25764823e-01 -3.38375308e-02 -1.06290448e+00 6.77571818e-02
1.81720242e-01 -2.51081258e-01 4.59818840e-01 -1.51234746e+00
6.79915071e-01 1.36332881e+00 4.04186100e-01 6.65990114e-01
-1.03754044e+00 -1.33144629e+00 2.94416584e-02 3.93247366e-01
-1.96259618e+00 -1.00386047e+00 1.12198520e+00 -4.85618711e-01
4.60602015e-01 2.16323048e-01 5.74120939e-01 1.10625589e+00
3.98610309e-02 8.53568554e-01 1.17115188e+00 -1.62153661e-01
2.28220612e-01 2.99411774e-01 -1.52471751e-01 6.91464543e-01
-2.02015787e-01 -2.17563495e-01 -8.72458041e-01 -3.69101673e-01
7.71118760e-01 9.07302648e-02 -4.52389330e-01 -5.09888768e-01
-8.50226998e-01 6.86552644e-01 -8.21600780e-02 3.83887917e-01
-5.36384165e-01 -5.12701869e-01 4.64161009e-01 2.63329446e-01
4.49508250e-01 -2.50722110e-01 -4.46017444e-01 3.75213586e-02
-1.05969965e+00 -1.54083287e-02 4.48531896e-01 9.14040208e-01
9.50364411e-01 3.12285841e-01 -7.81599153e-03 1.24655950e+00
3.09393525e-01 5.95738232e-01 6.19898915e-01 -6.30124092e-01
4.22319949e-01 6.50335371e-01 -9.62394252e-02 -1.47376335e+00
-1.60021245e-01 -2.71206766e-01 -1.18095589e+00 -7.17682913e-02
4.71303873e-02 -2.76277602e-01 -5.08880615e-01 1.91844726e+00
4.58846956e-01 4.31640118e-01 2.08821625e-01 1.01569927e+00
1.04512775e+00 7.70457506e-01 6.28112480e-02 -6.96202755e-01
1.24296033e+00 -5.99198639e-01 -8.32840025e-01 -3.67071070e-02
5.34926116e-01 -7.04662323e-01 7.92204976e-01 3.19051445e-01
-6.78261042e-01 -5.49445093e-01 -7.91191459e-01 2.35928714e-01
4.62049693e-02 8.19659650e-01 4.18913841e-01 3.95329326e-01
-8.04317296e-01 -1.35498606e-02 -4.28626269e-01 -2.49853551e-01
6.02098405e-01 5.13921499e-01 -7.53041327e-01 -1.91139683e-01
-1.00875604e+00 4.26500291e-01 -5.33971302e-02 2.60874599e-01
-7.83445179e-01 -9.11822677e-01 -9.01562154e-01 1.60964072e-01
3.83427471e-01 -2.44849220e-01 5.98690987e-01 -1.77591193e+00
-1.65022838e+00 9.47329879e-01 -5.25180638e-01 2.83808708e-01
8.62761587e-02 -2.39690810e-01 -8.12817752e-01 2.38304034e-01
9.72571075e-02 4.22891766e-01 1.14909863e+00 -1.01856077e+00
-2.95049608e-01 -7.79486775e-01 -8.18251908e-01 1.99539989e-01
-6.08985066e-01 6.21874690e-01 -5.59262156e-01 -7.51893103e-01
8.43690410e-02 -7.91187882e-01 5.10232151e-02 2.05650717e-01
-8.36829990e-02 -1.87656343e-01 1.23033881e+00 -8.45145464e-01
1.01814640e+00 -2.71760917e+00 4.17154700e-01 3.86301428e-01
1.59610033e-01 3.40422302e-01 -2.98026502e-01 -1.13343894e-01
-4.72939670e-01 -4.05863404e-01 -1.35663301e-01 -3.19709361e-01
-4.81581569e-01 -4.40345630e-02 -2.53503531e-01 6.49682462e-01
2.55762309e-01 5.08288503e-01 -5.72929740e-01 -8.55845332e-01
5.92821576e-02 6.08992994e-01 -3.66896391e-01 3.94799501e-01
3.63669813e-01 5.34607112e-01 -7.90070891e-01 1.01437020e+00
1.15910709e+00 1.00413403e-02 1.23897061e-01 -6.13903224e-01
-6.23200685e-02 -5.87089717e-01 -1.51397681e+00 1.62712574e+00
-1.73291653e-01 3.19036543e-01 6.80459440e-01 -1.07596374e+00
1.29594672e+00 3.52331221e-01 9.89501536e-01 -8.20487618e-01
1.72624245e-01 2.18835950e-01 -4.16436702e-01 -4.30993110e-01
6.57273531e-02 -7.63174668e-02 4.96931486e-02 -2.94541959e-02
3.07429224e-01 5.68688095e-01 -4.52325255e-01 -8.01519603e-02
4.85903919e-01 -1.30268885e-02 3.75860691e-01 -4.08071160e-01
8.96222293e-01 -3.91323596e-01 1.33796549e+00 -1.76965654e-01
-3.54073673e-01 3.86037737e-01 3.91315341e-01 -2.49044016e-01
-4.50613528e-01 -9.42138851e-01 -1.61654115e-01 1.02372909e+00
3.04550946e-01 -3.70953977e-01 -7.01359570e-01 -4.76080298e-01
-1.83864027e-01 1.96199387e-01 -5.53373039e-01 -1.82412684e-01
-5.53398311e-01 -6.76326752e-01 4.18707788e-01 3.27516139e-01
8.54943335e-01 -8.59051347e-01 -2.74151981e-01 -3.05345487e-02
-3.07081521e-01 -1.00363052e+00 -6.23092830e-01 -3.41636747e-01
-6.47835553e-01 -8.05913568e-01 -8.30187321e-01 -1.05554461e+00
7.56625891e-01 5.10787189e-01 7.85662472e-01 -2.66185254e-01
-2.03592241e-01 2.69505441e-01 -2.62750328e-01 -2.00696647e-01
6.31943271e-02 -3.29515278e-01 2.47124687e-01 8.41559708e-01
5.97248495e-01 -6.19854927e-01 -5.70846617e-01 6.66807055e-01
-7.17730403e-01 3.18848304e-02 7.88502514e-01 1.03329504e+00
1.06626880e+00 -7.01668039e-02 6.77812219e-01 -5.72275937e-01
4.33420628e-01 -6.65412366e-01 -5.49248695e-01 3.49390358e-01
-3.32317114e-01 -3.46814066e-01 6.09337568e-01 -6.40080154e-01
-1.30726075e+00 3.50200146e-01 1.43123753e-02 -8.92598987e-01
-1.68560166e-02 4.76181298e-01 -9.39337432e-01 -3.70333642e-01
4.07918155e-01 6.53508663e-01 3.30600739e-01 -4.77870852e-01
6.08907305e-02 7.39246964e-01 4.71246213e-01 -6.48976445e-01
5.41017652e-01 4.15614933e-01 1.02342449e-01 -1.05094159e+00
-3.87579620e-01 -6.13765895e-01 -5.71632743e-01 -1.71373367e-01
6.73964918e-01 -1.42506742e+00 -4.02409703e-01 7.06111550e-01
-8.42668831e-01 1.69368401e-01 1.12159371e-01 3.98042321e-01
-4.23760951e-01 4.68953282e-01 -4.57719326e-01 -6.73466563e-01
-4.53490436e-01 -1.16049671e+00 1.32529151e+00 4.26134378e-01
-3.89403664e-02 -4.43448752e-01 -3.42724398e-02 2.83766478e-01
2.56112486e-01 4.51365829e-01 5.53827107e-01 -2.44547948e-01
-3.85979652e-01 1.27231017e-01 -2.47828037e-01 4.57333684e-01
3.19967061e-01 6.96328608e-03 -9.79980767e-01 -4.06527698e-01
2.07819462e-01 -3.70251387e-01 4.37278867e-01 2.78587461e-01
1.32380772e+00 -3.58963668e-01 -3.18354249e-01 9.01102960e-01
1.31063890e+00 2.52702892e-01 8.68980646e-01 -3.11709680e-02
6.58034325e-01 8.15065682e-01 1.00085807e+00 7.59917676e-01
1.40793815e-01 1.07768524e+00 3.43105160e-02 -3.41052949e-01
8.95939320e-02 -2.70198703e-01 7.81168282e-01 9.34402108e-01
1.03227429e-01 3.45321208e-01 -4.59373593e-01 6.10732973e-01
-1.91298664e+00 -9.45219398e-01 3.62717777e-01 1.91710210e+00
7.75142014e-01 -8.55117321e-01 1.27849787e-01 -2.39836529e-01
6.59882963e-01 2.36721963e-01 -7.19075143e-01 -2.28791498e-02
-4.72953111e-01 2.77302861e-01 -1.06415795e-02 -1.67413294e-01
-9.96734500e-01 9.06439900e-01 4.81929302e+00 1.26277435e+00
-1.51953757e+00 1.87485501e-01 7.00718105e-01 -2.15827435e-01
1.11591808e-01 -3.66331547e-01 -8.45140278e-01 6.29063368e-01
6.68124914e-01 -1.55351043e-01 3.51420403e-01 1.21507585e+00
2.99460262e-01 2.25299463e-01 -8.16935182e-01 1.72901285e+00
4.18716192e-01 -9.96452272e-01 3.64700519e-02 -3.56019661e-02
6.68234110e-01 -3.28336537e-01 1.80080786e-01 2.53576428e-01
-3.39468569e-01 -1.05774724e+00 4.65253353e-01 5.85188746e-01
1.07445157e+00 -8.35681379e-01 6.78934813e-01 1.78921521e-01
-1.35975111e+00 -9.03201476e-03 -5.18629491e-01 5.20862043e-01
-3.11813504e-01 5.03993452e-01 -2.97228307e-01 5.22370815e-01
9.23009396e-01 1.07232749e+00 -3.05641800e-01 5.26821911e-01
9.21821147e-02 5.19680262e-01 -1.22261368e-01 2.44164243e-01
-2.57796079e-01 -5.52144885e-01 5.13814151e-01 1.13032961e+00
6.36115074e-01 2.75333643e-01 1.42576963e-01 9.64103043e-01
9.40918475e-02 6.64366305e-01 -4.06808823e-01 1.37310416e-01
4.54015344e-01 1.43239582e+00 5.00078090e-02 3.83865717e-03
-5.96617520e-01 1.10779881e+00 1.78181395e-01 5.08646727e-01
-7.30581880e-01 -2.48250410e-01 1.19488311e+00 -7.81229557e-03
3.55996758e-01 -9.40853655e-02 1.42970964e-01 -1.48100007e+00
2.78322905e-01 -1.16160607e+00 4.82892931e-01 -7.42176771e-01
-1.28424585e+00 5.73254585e-01 -1.36053771e-01 -1.32344735e+00
-1.48920402e-01 -4.94821757e-01 -4.21489388e-01 1.05159211e+00
-1.49807656e+00 -1.37115502e+00 -5.72560132e-01 1.37304676e+00
3.49054724e-01 -6.17892861e-01 8.81394625e-01 6.47876501e-01
-9.52044725e-01 9.71997917e-01 2.03057319e-01 2.45887786e-01
9.84716654e-01 -3.11825991e-01 -6.40756309e-01 6.33677661e-01
-1.08991519e-01 6.24644220e-01 2.93269306e-01 -3.65264356e-01
-1.65061939e+00 -1.19389653e+00 4.87595111e-01 8.94551203e-02
2.84355819e-01 -4.75111604e-01 -9.67215180e-01 4.54535514e-01
-2.63433516e-01 3.20347250e-01 1.07695282e+00 -7.95480981e-03
-3.55556041e-01 -7.05365837e-01 -1.34177911e+00 5.08522689e-01
9.02119458e-01 -5.62531352e-01 -2.77043939e-01 5.88921458e-02
1.61489680e-01 -2.02905923e-01 -1.04289281e+00 5.27346671e-01
7.13389397e-01 -9.37801301e-01 8.96224022e-01 -3.56879771e-01
3.09417874e-01 -5.20832479e-01 -4.32576686e-01 -1.24269080e+00
-3.91611457e-01 -4.51938063e-01 -1.14438318e-01 1.67074096e+00
-1.35630175e-01 -6.71485782e-01 7.85830379e-01 4.60208535e-01
1.85403213e-01 -8.77878785e-01 -1.19894636e+00 -5.51800013e-01
-3.15863699e-01 8.39493871e-02 8.18097115e-01 1.07764459e+00
-2.38123715e-01 2.02044040e-01 -6.57439768e-01 2.79504091e-01
5.34885228e-01 3.71677846e-01 8.43885839e-01 -1.11469781e+00
2.70665204e-03 -1.94569424e-01 -6.25763118e-01 -9.21975195e-01
5.02507448e-01 -5.51494896e-01 -2.06319481e-01 -6.84222043e-01
5.16486406e-01 -3.56722832e-01 -4.05455112e-01 3.96794617e-01
6.31159246e-02 -9.20190811e-02 5.57840653e-02 4.42220181e-01
-5.83505392e-01 1.05507743e+00 1.13237262e+00 -1.09601401e-01
-2.55341023e-01 -2.33345240e-01 -9.56793308e-01 6.18268847e-01
4.56890225e-01 -2.53051162e-01 -5.00380397e-01 -1.45239040e-01
-3.14327031e-01 2.18283802e-01 3.48654151e-01 -8.51011276e-01
1.20842971e-01 -4.30904567e-01 6.40992045e-01 -4.44015652e-01
4.35095042e-01 -9.62286174e-01 2.59813845e-01 -4.51775752e-02
1.21357486e-01 -1.80728465e-01 2.88087100e-01 4.45970565e-01
-6.45151258e-01 3.85852605e-01 1.01668155e+00 1.31614298e-01
-1.21552849e+00 5.45946181e-01 -8.85644704e-02 -1.66130036e-01
1.18257773e+00 -4.14724052e-01 -6.00884445e-02 -3.85203153e-01
-4.69781816e-01 1.87611431e-01 3.05815130e-01 4.77732003e-01
9.00864720e-01 -1.56072342e+00 -8.92349184e-01 5.52832723e-01
4.38977510e-01 -2.19562069e-01 7.29393959e-01 8.65059555e-01
1.35849975e-02 1.70278102e-02 -5.91378570e-01 -6.92177176e-01
-1.73711669e+00 3.53355169e-01 4.77816164e-01 -6.24053292e-02
-2.90493339e-01 7.90876985e-01 7.33183563e-01 -1.47639170e-01
-1.00080632e-01 3.18653643e-01 -4.55008924e-01 2.13404045e-01
7.67094016e-01 2.29510307e-01 -1.26854733e-01 -1.25677371e+00
-6.26038134e-01 9.49189961e-01 -2.20277514e-02 3.64323318e-01
1.43195820e+00 -1.82326674e-01 -3.83653104e-01 9.79765654e-02
1.48071599e+00 2.25869954e-01 -1.19621062e+00 -4.71850842e-01
-3.51563126e-01 -7.97391176e-01 1.30136952e-01 -4.99966800e-01
-1.50107932e+00 6.41421914e-01 1.11464608e+00 -8.31074536e-01
1.78943038e+00 -2.68045980e-02 7.95778811e-01 -6.79486021e-02
6.23122215e-01 -1.06154585e+00 1.70298859e-01 4.57055308e-02
1.09675813e+00 -1.09366083e+00 -6.25292659e-02 -7.80256510e-01
-8.92193556e-01 1.08778775e+00 8.46396744e-01 5.95877394e-02
7.62299120e-01 1.93540916e-01 1.11767769e-01 -2.48473331e-01
-4.74676907e-01 1.79653719e-01 3.67796302e-01 6.22685611e-01
2.77004808e-01 9.38668475e-02 -5.69400229e-02 1.07931983e+00
1.27402753e-01 2.90005714e-01 -1.08246736e-01 5.02659857e-01
-1.58764184e-01 -8.90601695e-01 -5.34653962e-01 3.19164902e-01
-3.59582365e-01 1.58248618e-01 -4.55394566e-01 5.57932317e-01
3.49420726e-01 7.89065123e-01 -8.81692097e-02 -5.76732039e-01
2.98205763e-01 -1.06650397e-01 1.42778292e-01 -3.80031407e-01
-9.26924944e-02 2.26280928e-01 -1.24340631e-01 -7.57251024e-01
-5.25675237e-01 -8.66708338e-01 -9.29306090e-01 -8.19090530e-02
-1.77369684e-01 7.44922534e-02 2.46053800e-01 6.57274127e-01
8.46586108e-01 9.04800743e-03 9.88264620e-01 -6.68425083e-01
-4.97860938e-01 -6.78745508e-01 -1.12339306e+00 6.08895719e-01
1.88971102e-01 -9.39240396e-01 -1.28278047e-01 2.97272261e-02] | [13.617894172668457, 1.6194441318511963] |
91648906-a228-44b7-b0f0-05367e28783b | automatic-chord-recognition-with-higher-order | 1808.05341 | null | http://arxiv.org/abs/1808.05341v1 | http://arxiv.org/pdf/1808.05341v1.pdf | Automatic Chord Recognition with Higher-Order Harmonic Language Modelling | Common temporal models for automatic chord recognition model chord changes on
a frame-wise basis. Due to this fact, they are unable to capture musical
knowledge about chord progressions. In this paper, we propose a temporal model
that enables explicit modelling of chord changes and durations. We then apply
N-gram models and a neural-network-based acoustic model within this framework,
and evaluate the effect of model overconfidence. Our results show that model
overconfidence plays only a minor role (but target smoothing still improves the
acoustic model), and that stronger chord language models do improve recognition
results, however their effects are small compared to other domains. | ['Filip Korzeniowski', 'Gerhard Widmer'] | 2018-08-16 | null | null | null | null | ['chord-recognition'] | ['audio'] | [ 1.04685634e-01 -1.97992951e-01 -1.09105691e-01 -1.32851020e-01
-7.36933887e-01 -7.84489036e-01 7.35240877e-01 8.98361206e-02
-6.67253792e-01 2.42841288e-01 5.57517886e-01 -3.00595611e-01
-1.66871503e-01 -5.44995368e-01 -3.15092534e-01 -1.88290194e-01
-2.43500099e-01 2.14338854e-01 5.43380320e-01 -3.73066664e-01
4.59856302e-01 2.53902078e-01 -1.46936595e+00 2.77060986e-01
3.91961724e-01 6.55808449e-01 1.26850650e-01 1.15653741e+00
1.16726317e-01 1.10783100e+00 -9.25912321e-01 -1.31867141e-01
1.49365216e-01 -7.18540132e-01 -7.16366231e-01 -4.19869483e-01
3.83261025e-01 -2.19255626e-01 -3.72531176e-01 6.08293712e-01
4.83946532e-01 4.29526985e-01 7.53282189e-01 -7.00014055e-01
-3.28510284e-01 1.15872395e+00 2.90648732e-02 3.31762224e-01
5.09062469e-01 2.17904113e-02 1.21986496e+00 -6.11909032e-01
3.31541419e-01 1.10980546e+00 1.26829684e+00 2.71833867e-01
-1.34109974e+00 -6.07019246e-01 2.21811220e-01 2.87004411e-01
-1.36917675e+00 -5.52696347e-01 1.08724856e+00 -3.58891726e-01
1.30513155e+00 4.17253464e-01 8.97529840e-01 8.70217443e-01
1.83386371e-01 5.57331145e-01 9.39372122e-01 -6.66656077e-01
2.73496807e-02 -3.09070885e-01 5.78827113e-02 8.58126022e-03
-7.24355757e-01 5.85038304e-01 -8.80258083e-01 -1.13212906e-01
9.31611300e-01 -6.33628309e-01 4.73673716e-02 2.03119040e-01
-1.06165063e+00 6.48941755e-01 -1.44578563e-02 6.92579269e-01
-3.93335491e-01 3.21012527e-01 6.49761796e-01 5.13799012e-01
1.05734833e-01 7.95100689e-01 -6.01908326e-01 -9.40688729e-01
-1.43276525e+00 5.78111887e-01 8.03287148e-01 3.67717952e-01
1.17988437e-01 5.47105193e-01 -1.81645855e-01 1.11623418e+00
-7.13421311e-03 1.89455748e-01 8.45674455e-01 -1.34699726e+00
-1.68550462e-01 -1.03575870e-01 1.05197757e-01 -1.13788378e+00
-5.22690952e-01 -3.65037918e-01 -3.29453886e-01 1.42080486e-01
9.78984237e-01 2.32421026e-01 -7.15521634e-01 2.05780864e+00
-3.47009450e-01 3.01724315e-01 -2.19170630e-01 5.75733066e-01
3.82614136e-01 6.03227496e-01 1.29566669e-01 -5.13656557e-01
1.19672143e+00 -6.74115479e-01 -8.57493639e-01 -1.56573251e-01
6.69196010e-01 -1.05652809e+00 1.38715589e+00 8.66099417e-01
-1.38818216e+00 -9.25624549e-01 -9.66082811e-01 -9.64817181e-02
1.70627292e-02 1.51786879e-02 4.75369930e-01 7.35452533e-01
-8.56199980e-01 1.17282951e+00 -8.67747426e-01 -4.29125577e-02
-4.54474866e-01 1.92425162e-01 2.78817832e-01 6.35922730e-01
-1.45655119e+00 1.06534827e+00 5.03952146e-01 -1.41380131e-01
-5.70470273e-01 -7.81142771e-01 -7.83920407e-01 -3.70492935e-02
2.28931651e-01 6.76383451e-02 1.87990379e+00 -9.01126027e-01
-1.63761401e+00 4.65220362e-01 -1.04539596e-01 -6.21186018e-01
3.45503718e-01 -1.86142281e-01 -4.71730262e-01 -1.16354853e-01
-4.32143122e-01 5.90302348e-01 4.27083910e-01 -8.63233030e-01
-4.74984258e-01 -4.47741449e-02 4.82199304e-02 1.67628869e-01
-2.98257440e-01 1.98488042e-01 -3.76803637e-01 -1.10566127e+00
-8.50499049e-02 -1.19408500e+00 -2.58564472e-01 -5.05927622e-01
2.67151510e-03 -3.05218637e-01 1.65077537e-01 -1.10836208e+00
2.08628726e+00 -2.16520596e+00 -7.77743431e-03 2.95520633e-01
-4.38673675e-01 1.01092607e-01 -1.29953295e-01 3.45817536e-01
9.04668719e-02 -8.26754421e-03 -2.42464647e-01 -2.61726677e-01
4.96409349e-02 2.38995790e-01 -1.98225081e-01 -1.19679831e-01
-1.67094067e-01 7.14609563e-01 -5.04235268e-01 -2.17141271e-01
2.32846975e-01 5.39532483e-01 -7.56832838e-01 -1.54568553e-01
-2.34362558e-01 3.19197953e-01 4.01126415e-01 3.21707577e-01
8.31774771e-02 3.84169370e-01 4.61536884e-01 -1.27685189e-01
-3.61654848e-01 9.35365140e-01 -1.23606026e+00 1.79012525e+00
-4.06082094e-01 6.10679924e-01 -3.73418361e-01 -5.96245408e-01
9.85512733e-01 4.70324069e-01 2.10533664e-01 -8.09737682e-01
-2.86067184e-02 1.57894850e-01 7.76801348e-01 -5.93110062e-02
9.48361337e-01 -4.99233961e-01 -2.42050111e-01 3.24831694e-01
-2.57510930e-01 -5.90719879e-01 3.21215659e-01 -1.45123348e-01
8.77759397e-01 2.24859193e-01 3.40582907e-01 -7.80442655e-02
1.87592223e-01 -7.08705112e-02 8.11590195e-01 9.43454444e-01
-5.33251911e-02 6.59177959e-01 2.07818210e-01 -1.82093218e-01
-1.05687308e+00 -7.98114598e-01 2.88251359e-02 1.43494451e+00
-4.55858737e-01 -1.01568580e+00 -6.25577748e-01 -1.24295689e-01
-1.79416627e-01 1.15381157e+00 -5.09628177e-01 -3.33262980e-01
-8.92162979e-01 -3.60575169e-01 1.12008786e+00 9.00236607e-01
-3.13317120e-01 -1.40430391e+00 -6.59278691e-01 6.05581522e-01
-1.94340289e-01 -7.17941523e-01 -7.00809062e-01 3.08931023e-01
-9.14671957e-01 -8.20387721e-01 -4.32336986e-01 -5.91043115e-01
-5.38936436e-01 -2.11749479e-01 1.16265512e+00 -4.27122228e-02
-1.45778045e-01 2.29115590e-01 -3.16009790e-01 -6.12528503e-01
-6.04566932e-01 3.19006778e-02 1.87378258e-01 -6.20193601e-01
4.71713394e-01 -8.40256035e-01 -1.46521881e-01 2.15846643e-01
-6.24944627e-01 -2.87855834e-01 1.29975751e-01 6.44106984e-01
3.47209156e-01 4.53112349e-02 6.38885736e-01 -7.10271835e-01
9.42804575e-01 1.29684001e-01 -3.68370920e-01 -1.11364119e-01
-7.94387579e-01 -1.81929275e-01 4.44285125e-01 -1.07055700e+00
-8.74080420e-01 -1.66325077e-01 -3.56853247e-01 -4.12801474e-01
-9.21224132e-02 9.13931429e-01 7.03213364e-02 2.56802112e-01
7.96780765e-01 2.16602504e-01 -7.31982589e-02 -7.04618871e-01
2.67532527e-01 2.49317616e-01 7.39889026e-01 -6.54565573e-01
5.26417315e-01 -9.16575044e-02 -3.15979511e-01 -9.64715362e-01
-5.88236749e-01 -2.90450335e-01 -7.52943635e-01 -3.42671931e-01
4.42232877e-01 -6.73516273e-01 -5.46729803e-01 4.46106553e-01
-1.10126030e+00 -6.00533247e-01 -4.36143816e-01 8.66922557e-01
-9.51275766e-01 3.71705323e-01 -1.25274396e+00 -1.10071671e+00
8.18282515e-02 -8.18788826e-01 4.40153301e-01 4.47664084e-03
-1.11197507e+00 -9.20685530e-01 4.71142769e-01 3.28981467e-02
5.22847116e-01 -1.18078806e-01 1.06083071e+00 -5.98246753e-01
-1.03772826e-01 -8.37431848e-02 4.37632918e-01 5.04184306e-01
1.59759507e-01 2.23301440e-01 -9.33528602e-01 1.05482362e-01
3.16924192e-02 -1.52441606e-01 7.24355221e-01 7.16149926e-01
9.30981100e-01 -1.17674962e-01 2.62626499e-01 2.20122039e-01
8.14744949e-01 7.06326067e-01 7.07048476e-01 2.57889837e-01
4.58989739e-01 6.28337920e-01 7.75503337e-01 5.73449850e-01
3.36910903e-01 1.02748549e+00 -4.44196984e-02 2.27160260e-01
-2.63079792e-01 -3.82283926e-01 5.40212870e-01 1.38998103e+00
-3.73590231e-01 8.19735825e-02 -9.23167288e-01 8.01317513e-01
-1.59017324e+00 -1.18139994e+00 -1.51611328e-01 2.37626886e+00
1.20081854e+00 6.48652017e-01 6.55366302e-01 7.79231846e-01
1.96388736e-01 3.12250406e-01 -2.14546725e-01 -8.27533424e-01
-2.30432019e-01 3.64687830e-01 1.71085700e-01 7.93043613e-01
-7.92835832e-01 1.12717569e+00 8.09017467e+00 9.40059721e-01
-9.64727044e-01 -1.24025635e-01 6.37024716e-02 -2.60678858e-01
-3.83872390e-01 1.32365733e-01 -4.24801439e-01 2.69870818e-01
1.46125162e+00 -1.00019664e-01 5.36781549e-01 4.34920311e-01
4.71503735e-01 1.73402965e-01 -1.09726489e+00 7.45414376e-01
1.40565127e-01 -1.04357862e+00 -6.83503821e-02 -1.65778637e-01
2.54868567e-01 -2.99894243e-01 9.44681559e-03 5.54831505e-01
1.45447269e-01 -1.10187829e+00 1.19053817e+00 7.59498775e-01
5.27077079e-01 -1.03180921e+00 3.54500055e-01 3.67633909e-01
-1.22809350e+00 -8.33034068e-02 -2.15211123e-01 -7.21977293e-01
4.85519409e-01 1.94150031e-01 -7.23538101e-01 1.18742310e-01
5.18888295e-01 5.64272523e-01 -5.79186499e-01 1.00144255e+00
-2.00656042e-01 1.28302205e+00 -3.26748729e-01 1.91146493e-01
2.50469204e-02 3.08326352e-02 5.12079060e-01 1.50538504e+00
4.32533503e-01 3.04260850e-02 2.27984965e-01 5.53885579e-01
5.11364639e-01 2.50863940e-01 -3.83797497e-01 -2.18962386e-01
7.50473142e-01 4.84195143e-01 -6.26500547e-01 -2.30609491e-01
-3.32420707e-01 7.98097253e-01 5.73198609e-02 1.54509351e-01
-6.00127816e-01 -2.61290908e-01 6.07753694e-01 5.76483048e-02
3.50415975e-01 -4.87544239e-01 -3.72280270e-01 -7.31773198e-01
-2.66590595e-01 -1.18635046e+00 3.82058650e-01 -7.08436012e-01
-9.72824812e-01 3.16659361e-01 -4.17597741e-02 -9.34920073e-01
-9.55668747e-01 -4.40318316e-01 -5.86116970e-01 8.08828712e-01
-9.04042900e-01 -9.51823771e-01 4.07656401e-01 2.75087029e-01
6.48037851e-01 1.24962442e-01 1.19829261e+00 2.04109713e-01
1.22358821e-01 6.78334653e-01 -2.66560644e-01 2.20033452e-01
9.20290411e-01 -1.39574754e+00 4.97153878e-01 6.06214225e-01
7.20030665e-01 8.41922224e-01 1.04918659e+00 -6.26706421e-01
-6.61640346e-01 -4.78885353e-01 1.30412149e+00 -6.28888607e-01
6.42193139e-01 -2.35503182e-01 -1.18184257e+00 5.84716797e-01
7.91532174e-02 -6.28141463e-01 9.95973110e-01 7.81093538e-01
-6.31282032e-01 2.06636667e-01 -4.60629791e-01 6.44978821e-01
8.74205470e-01 -1.08650947e+00 -1.06692016e+00 -2.75581837e-01
7.56138146e-01 -1.67257041e-01 -1.07860732e+00 5.57385385e-01
1.09059918e+00 -1.03847885e+00 9.90379214e-01 -5.67884028e-01
1.61745176e-01 -2.53246546e-01 -2.28649825e-01 -1.29559183e+00
-6.01003587e-01 -5.58943033e-01 -1.99140847e-01 1.29857957e+00
5.41301072e-01 8.40855539e-02 6.76311076e-01 4.20093209e-01
-2.30733931e-01 -1.11364484e-01 -9.37805057e-01 -1.14868724e+00
3.31102490e-01 -1.22445703e+00 2.62538791e-01 1.05612087e+00
4.14319605e-01 3.39124113e-01 -7.59372592e-01 -2.74713159e-01
5.03322259e-02 -1.30822554e-01 4.30794418e-01 -1.38661945e+00
-8.78068209e-01 -7.87236035e-01 -3.52295280e-01 -9.78534579e-01
4.50390875e-02 -4.97390866e-01 1.52295351e-01 -7.66002297e-01
-2.34694675e-01 -4.00865465e-01 -7.25514531e-01 3.95741135e-01
8.71214550e-03 4.77857590e-01 5.98090649e-01 2.84538139e-02
-1.18751265e-01 3.56574059e-01 6.40161574e-01 2.51123086e-02
-6.40452325e-01 1.60492182e-01 -3.17760348e-01 9.92058635e-01
9.05284941e-01 -3.86946261e-01 -4.68562126e-01 -1.59000695e-01
1.87071726e-01 4.38803971e-01 5.35372272e-02 -1.16300070e+00
2.55749226e-01 -1.05268717e-01 2.68215001e-01 -5.22956014e-01
7.28368461e-01 -3.04403365e-01 2.40371406e-01 4.91303891e-01
-7.24441469e-01 3.28954667e-01 6.57487214e-01 3.14883083e-01
-3.46858233e-01 -3.37205559e-01 4.49211121e-01 -2.92066317e-02
-4.39284265e-01 -3.42026055e-01 -1.00721788e+00 -5.72160408e-02
1.70010120e-01 -4.17211443e-01 4.69150126e-01 -6.19570553e-01
-1.04688883e+00 -4.37498510e-01 4.17866051e-01 6.38747931e-01
1.26660466e-01 -1.43087327e+00 -6.51263297e-01 1.20300658e-01
2.35165413e-02 -7.76588500e-01 2.22934380e-01 7.49604106e-01
-4.77201045e-02 4.36212867e-01 4.24193256e-02 -3.86269987e-01
-1.71741498e+00 3.67352098e-01 1.44363135e-01 -3.22727352e-01
-3.63752663e-01 9.41234291e-01 9.98015609e-03 -5.74517488e-01
4.77189571e-01 -5.68886042e-01 -3.82969886e-01 3.18827778e-01
6.70682013e-01 4.03996348e-01 -1.04917437e-01 -6.67862713e-01
-3.13438207e-01 5.48613906e-01 -2.52054986e-02 -6.70242727e-01
9.11383808e-01 -8.49863607e-03 2.49404907e-01 1.45351040e+00
3.51037920e-01 4.46138650e-01 -1.13673747e+00 -2.92055756e-01
3.55970651e-01 -2.53881425e-01 -3.80327553e-03 -1.03212917e+00
-3.07965785e-01 8.26541960e-01 3.37121248e-01 1.50083274e-01
1.08244419e+00 -3.13044965e-01 4.82007444e-01 1.81073442e-01
1.76052481e-01 -1.37642443e+00 -1.02959968e-01 1.07662296e+00
7.06153393e-01 -4.54732090e-01 -3.30179363e-01 -1.93139598e-01
-6.82248712e-01 1.04418445e+00 4.31316793e-01 -4.62835990e-02
6.32934988e-01 4.44028765e-01 2.65062839e-01 2.68002927e-01
-9.28590059e-01 -1.34366512e-01 4.16531295e-01 4.93871778e-01
9.88455415e-01 1.55987933e-01 -5.64240336e-01 9.58321214e-01
-9.89086211e-01 -9.15769022e-03 4.54447269e-01 6.67952716e-01
-3.68116766e-01 -1.35168076e+00 -5.29429197e-01 1.79882705e-01
-7.15394855e-01 -3.85832161e-01 -5.23636460e-01 7.29476810e-01
1.65838167e-01 1.02434886e+00 2.99794137e-01 -6.54377103e-01
3.65510941e-01 5.06903052e-01 7.12555110e-01 -5.22404134e-01
-9.26168442e-01 7.07500935e-01 4.48415190e-01 -4.79121268e-01
-3.53244185e-01 -8.91705632e-01 -1.08038211e+00 -1.49578303e-01
-3.23917747e-01 3.16117674e-01 4.17903423e-01 7.88084149e-01
-1.75289363e-01 8.00675094e-01 1.58932224e-01 -7.41365433e-01
-5.91972470e-01 -1.40292335e+00 -7.49255776e-01 2.13210300e-01
-8.24351422e-03 -2.39319533e-01 -4.01673824e-01 1.20151341e-01] | [15.91412353515625, 5.364935874938965] |
1e59e422-1e98-4626-830a-2df8816472d6 | structure-aware-robustness-certificates-for | 2306.11915 | null | https://arxiv.org/abs/2306.11915v2 | https://arxiv.org/pdf/2306.11915v2.pdf | Structure-Aware Robustness Certificates for Graph Classification | Certifying the robustness of a graph-based machine learning model poses a critical challenge for safety. Current robustness certificates for graph classifiers guarantee output invariance with respect to the total number of node pair flips (edge addition or edge deletion), which amounts to an $l_{0}$ ball centred on the adjacency matrix. Although theoretically attractive, this type of isotropic structural noise can be too restrictive in practical scenarios where some node pairs are more critical than others in determining the classifier's output. The certificate, in this case, gives a pessimistic depiction of the robustness of the graph model. To tackle this issue, we develop a randomised smoothing method based on adding an anisotropic noise distribution to the input graph structure. We show that our process generates structural-aware certificates for our classifiers, whereby the magnitude of robustness certificates can vary across different pre-defined structures of the graph. We demonstrate the benefits of these certificates in both synthetic and real-world experiments. | ['Xiaowen Dong', 'Henry Kenlay', 'Pierre Osselin'] | 2023-06-20 | null | null | null | null | ['graph-classification', 'classification-1'] | ['graphs', 'methodology'] | [ 4.63507444e-01 3.88299286e-01 -5.98845771e-04 -1.72055364e-02
-6.28921747e-01 -1.23009169e+00 5.51877499e-01 5.34195244e-01
-3.05383861e-01 4.53938663e-01 -1.55561432e-01 -8.84336531e-01
-7.35118315e-02 -9.46870267e-01 -9.99642015e-01 -9.10744131e-01
-5.64411402e-01 5.57675920e-02 4.81297374e-01 -6.70614839e-02
1.68964431e-01 7.14222074e-01 -1.07127118e+00 -1.77578181e-02
2.87163913e-01 7.82193363e-01 -6.69369936e-01 9.03307080e-01
3.77199888e-01 4.73437339e-01 -7.55904138e-01 -7.15870321e-01
5.99859476e-01 -2.11519748e-01 -7.10123360e-01 -4.57092822e-02
6.92076862e-01 2.08550721e-01 -3.33899498e-01 1.36766863e+00
1.60043702e-01 -2.27665901e-01 7.07100034e-01 -1.74609172e+00
8.51897001e-02 7.65055060e-01 -4.72483337e-01 8.09295028e-02
2.41428286e-01 2.63156146e-01 1.26302195e+00 -7.41519332e-02
5.67954898e-01 8.99753630e-01 9.23464179e-01 3.89663398e-01
-1.37439549e+00 -5.21159470e-01 1.20088026e-01 -1.25841171e-01
-1.37407148e+00 -1.85308754e-01 7.77811408e-01 -3.71855289e-01
4.32175308e-01 6.98663652e-01 3.81621122e-01 7.64787078e-01
4.33558285e-01 9.76485983e-02 1.11078942e+00 -3.17098558e-01
5.64467609e-01 -1.66825615e-02 1.52985722e-01 7.87966669e-01
6.71214759e-01 7.88312107e-02 -1.45605594e-01 -6.17581308e-01
3.87631595e-01 -4.11036909e-01 -3.12860698e-01 -7.90747464e-01
-9.16021287e-01 7.87262142e-01 3.98835242e-01 -4.86663021e-02
2.41733983e-01 4.50015068e-01 4.20985460e-01 6.19674206e-01
-5.86833572e-03 2.57450342e-01 -4.09839779e-01 2.50638783e-01
-5.15718579e-01 -4.26253956e-03 1.00018036e+00 7.15901196e-01
3.86591941e-01 -3.48485634e-02 3.15306991e-01 -2.63484269e-02
8.94950628e-02 3.40884686e-01 -3.29492867e-01 -7.43558645e-01
3.89394194e-01 6.79824889e-01 -7.42732286e-02 -1.19352794e+00
-5.21399379e-01 -5.65396369e-01 -9.36286271e-01 4.87612277e-01
9.13836420e-01 -6.43858761e-02 -7.86201119e-01 1.97671235e+00
5.20004869e-01 3.39845479e-01 -2.50003878e-02 7.01965690e-01
2.88018405e-01 2.20338225e-01 1.53732613e-01 -1.68777093e-01
1.33335853e+00 -2.37897083e-01 -5.53522967e-02 -5.92826307e-02
9.53355372e-01 -4.57104027e-01 9.50736701e-01 3.41966510e-01
-6.41553402e-01 9.66833830e-02 -1.34503138e+00 4.44701284e-01
-2.91824371e-01 -3.95071357e-01 5.39338946e-01 1.25963593e+00
-9.93224859e-01 4.01797175e-01 -7.11211860e-01 5.34021519e-02
1.95357963e-01 6.80752337e-01 -8.47537279e-01 1.83990225e-02
-1.17547226e+00 8.05032015e-01 1.99782819e-01 -5.43674193e-02
-7.30109334e-01 -5.55679321e-01 -8.54396462e-01 8.87656510e-02
5.65945625e-01 -3.96749854e-01 8.42538774e-01 -8.00921619e-01
-7.28407323e-01 7.84579277e-01 2.33022973e-01 -6.42811179e-01
8.40150952e-01 4.78550494e-01 -2.47797042e-01 1.39477715e-01
-3.60064894e-01 1.36197582e-01 9.41432118e-01 -1.27395570e+00
-4.27906364e-01 -3.02655876e-01 4.85544354e-01 -1.80798396e-01
-1.23816147e-01 -8.23814608e-03 -1.31105959e-01 -5.96457541e-01
1.61963791e-01 -1.34724033e+00 -5.01178086e-01 -2.51841657e-02
-6.05792999e-01 8.16777796e-02 7.84400761e-01 -3.15320700e-01
1.07789624e+00 -2.19996643e+00 -2.51611590e-01 9.03054893e-01
3.44377071e-01 1.50168210e-01 -7.32206255e-02 4.11904007e-01
-2.96517670e-01 5.56057751e-01 -4.60250437e-01 7.85740018e-02
-5.88042066e-02 5.56099415e-02 -2.83082902e-01 1.06484842e+00
3.11988080e-03 4.24725801e-01 -8.31947803e-01 -3.52725387e-01
-2.91480310e-02 3.66529733e-01 -4.88741100e-01 -2.42625266e-01
-7.29331300e-02 1.58662289e-01 -3.41222584e-01 2.57180452e-01
7.44433165e-01 -1.04637206e-01 6.57436550e-01 1.53771490e-02
4.69753116e-01 1.57240808e-01 -1.66547763e+00 8.82296979e-01
-1.49136439e-01 3.28220725e-01 3.18911552e-01 -8.11204493e-01
6.25842273e-01 1.76600128e-01 3.64747643e-01 -1.52416006e-01
2.64396429e-01 -1.24689497e-01 2.11659461e-01 1.45657539e-01
2.45767698e-01 -1.50060877e-01 -3.98485899e-01 7.18966722e-01
-2.89551973e-01 -7.02245682e-02 -1.46568596e-01 5.71960449e-01
1.67123771e+00 -5.36296070e-01 1.58131361e-01 -5.42898893e-01
4.64457721e-01 -2.68779069e-01 7.18464196e-01 8.91228199e-01
-2.60332704e-01 6.39080584e-01 1.21627641e+00 -3.18025649e-01
-9.82417524e-01 -1.12632298e+00 1.12109773e-01 6.03792608e-01
1.53230265e-01 -5.95723748e-01 -1.16867435e+00 -1.28493416e+00
1.25220403e-01 4.77358133e-01 -6.58086658e-01 -6.67680979e-01
-4.44229811e-01 -6.75493598e-01 8.93018067e-01 2.71986276e-01
4.68395576e-02 -5.32039404e-01 -5.20281315e-01 -1.61787361e-01
1.42959863e-01 -1.21055996e+00 -5.69061399e-01 1.65991575e-01
-6.72500968e-01 -1.70516539e+00 1.97188675e-01 -6.20426893e-01
1.23671901e+00 1.01391070e-01 1.00256824e+00 5.58984995e-01
-1.55675203e-01 4.85328704e-01 -1.02795616e-01 -9.91077125e-02
-7.36280441e-01 5.04073463e-02 1.10696279e-01 9.60390121e-02
-2.03227744e-01 -5.48851669e-01 -4.65838939e-01 3.49032491e-01
-1.07261467e+00 -3.92647445e-01 1.67833164e-01 5.07888019e-01
3.41211826e-01 5.71916938e-01 2.45132372e-01 -1.04351640e+00
6.58893704e-01 -3.04199249e-01 -9.57199335e-01 4.10691470e-01
-5.20545244e-01 2.75503844e-01 8.61983716e-01 -4.26012993e-01
-2.02030614e-01 5.36192656e-01 9.09996629e-02 -1.50325865e-01
6.10846616e-02 3.53723526e-01 -5.03867209e-01 -6.34677470e-01
6.39037848e-01 -8.13170299e-02 -6.89979121e-02 -9.59622860e-02
4.35618401e-01 2.56132841e-01 4.48750883e-01 -6.94760382e-01
1.10624003e+00 4.39519107e-01 8.02107632e-01 -4.86237466e-01
-1.86601400e-01 4.84451652e-02 -5.30801535e-01 -2.26102680e-01
2.20633686e-01 -5.07610857e-01 -1.24494112e+00 3.86788458e-01
-7.31311262e-01 -3.23517531e-01 -7.96455517e-02 3.46361622e-02
-2.12955222e-01 6.28153443e-01 -4.56777424e-01 -8.44465673e-01
-1.11535244e-01 -1.28829777e+00 6.85378850e-01 -2.12212443e-01
-1.54502630e-01 -1.15469778e+00 -1.33819908e-01 -6.35079220e-02
3.74192968e-02 7.11843252e-01 1.18885553e+00 -9.04262662e-01
-3.61166507e-01 -8.36924136e-01 1.91661399e-02 3.12411129e-01
-3.56082506e-02 2.98797250e-01 -8.09230030e-01 -6.11238182e-01
-7.71054178e-02 -8.79093483e-02 7.50747740e-01 -1.18616007e-01
9.85249341e-01 -7.08079100e-01 -2.34547228e-01 2.95738786e-01
1.41024268e+00 -3.38119596e-01 4.79003638e-01 1.11203529e-01
6.96074843e-01 4.44495887e-01 2.41088465e-01 1.47420973e-01
1.97373554e-01 5.01124501e-01 6.82340264e-01 3.68484892e-02
6.00484759e-02 -2.71429002e-01 3.12605500e-01 1.97688267e-01
2.53227532e-01 -2.69453615e-01 -1.05151904e+00 3.22552681e-01
-1.38350034e+00 -6.39493763e-01 -2.02012002e-01 2.68836188e+00
6.63987517e-01 7.02940047e-01 1.59501493e-01 6.40619874e-01
8.31662238e-01 1.88406289e-01 -2.52591521e-01 -5.22202134e-01
-1.51432574e-01 8.68750960e-02 1.05351913e+00 7.78447270e-01
-1.03651571e+00 5.24802506e-01 6.72219086e+00 7.08236158e-01
-1.14799011e+00 -1.11484356e-01 5.81942856e-01 1.59417853e-01
-4.51491773e-01 4.18054491e-01 -4.58007783e-01 5.34861624e-01
9.01797831e-01 -1.13697983e-01 2.27386907e-01 7.62861848e-01
-1.64161146e-01 -1.45055816e-01 -1.07042623e+00 5.31356990e-01
-1.11369930e-01 -1.07587254e+00 -9.47417989e-02 2.68971294e-01
4.18056518e-01 -1.77437693e-01 1.22413769e-01 -1.89657494e-01
7.87474334e-01 -1.05975974e+00 6.56533778e-01 2.30390821e-02
8.57871711e-01 -1.07862687e+00 5.39039731e-01 3.54277939e-01
-1.20150340e+00 -9.21986550e-02 9.36085824e-04 1.07997917e-01
-3.17566812e-01 7.60658145e-01 -9.96355474e-01 5.14016569e-01
3.00456583e-01 -9.97681990e-02 -7.81597018e-01 8.09578359e-01
-3.58529299e-01 8.68135035e-01 -5.99632978e-01 2.21647754e-01
-1.76716726e-02 -5.54380454e-02 7.53239691e-01 1.19437134e+00
-7.81470463e-02 1.12434365e-01 1.55813098e-01 2.56438196e-01
-2.73684323e-01 -2.70001382e-01 -8.07070673e-01 1.80337697e-01
6.02122903e-01 1.36604559e+00 -1.19432580e+00 4.05767672e-02
-1.67426422e-01 4.87503141e-01 1.94995880e-01 2.26469964e-01
-6.92724168e-01 -3.14100981e-01 7.16398060e-01 2.86185950e-01
1.98760062e-01 -4.10962343e-01 -3.23483914e-01 -9.33831871e-01
1.75113499e-01 -1.20512271e+00 7.35520422e-01 -1.06830984e-01
-9.43894923e-01 2.51334488e-01 -3.53737384e-01 -1.17434859e+00
9.09431353e-02 -6.13430083e-01 -7.47785151e-01 4.79625553e-01
-9.16572928e-01 -1.10147536e+00 3.08045354e-02 5.40377915e-01
-5.03633499e-01 1.71950564e-01 8.04252923e-01 -6.80996403e-02
-4.34493214e-01 1.03206551e+00 -3.09437782e-01 3.62798572e-01
4.71848726e-01 -1.27584481e+00 7.57878959e-01 1.25278187e+00
3.03722143e-01 6.83245361e-01 1.15142941e+00 -5.85077226e-01
-1.75369835e+00 -1.16712260e+00 6.05163574e-01 -7.30492294e-01
7.34847963e-01 -7.12000489e-01 -7.76443183e-01 8.22397947e-01
-3.38089228e-01 5.94375849e-01 3.93521667e-01 5.42539023e-02
-1.02062881e+00 -1.05857179e-01 -1.43813241e+00 7.36317873e-01
1.03081656e+00 -6.89686596e-01 -8.55351761e-02 1.99342191e-01
4.83058095e-01 -3.47239614e-01 -9.63390231e-01 5.84283769e-01
5.35097718e-01 -9.08892632e-01 7.78749049e-01 -7.93388069e-01
-2.44408160e-01 -5.82898676e-01 -5.51523939e-02 -1.19452345e+00
1.96830649e-02 -8.49130332e-01 -7.51369516e-04 1.06965244e+00
5.81284404e-01 -8.29623222e-01 9.51878250e-01 6.70714259e-01
2.45495886e-01 -6.27235472e-01 -1.26314652e+00 -8.83410633e-01
1.89336613e-01 -5.57466984e-01 5.80879986e-01 1.01863444e+00
2.38352910e-01 -1.57478396e-02 -1.44211158e-01 5.05754590e-01
6.85785770e-01 -2.60185480e-01 8.62159371e-01 -1.25351608e+00
-1.91161335e-01 -4.89460468e-01 -9.80708718e-01 -2.32153952e-01
2.79843867e-01 -1.01613450e+00 -2.43355572e-01 -9.70384300e-01
-1.13725737e-02 -6.52705491e-01 -3.21435601e-01 6.34434640e-01
-1.74339563e-02 3.22812080e-01 1.55025735e-01 -2.15399966e-01
-4.70810562e-01 -2.24461511e-01 6.83711171e-01 -2.05011070e-01
8.12941194e-02 1.45542875e-01 -7.45754719e-01 6.90209150e-01
9.70386386e-01 -7.39427984e-01 -3.62249851e-01 1.06553942e-01
6.16029322e-01 -6.66704327e-02 6.10171735e-01 -8.37064922e-01
2.83642381e-01 -9.14060026e-02 6.83909804e-02 1.37665421e-01
2.22775172e-02 -9.79475319e-01 4.85434145e-01 8.61846328e-01
-4.82637733e-01 1.47719949e-01 9.52154249e-02 7.01637447e-01
1.44026652e-01 4.90109883e-02 8.33169699e-01 1.57361224e-01
-1.47128841e-02 3.42354834e-01 -1.34632111e-01 1.34670213e-01
1.13508642e+00 -6.31951839e-02 -5.38424134e-01 -4.80035603e-01
-4.36697870e-01 7.63632134e-02 9.26681876e-01 2.32340917e-01
2.42079213e-01 -1.08278263e+00 -5.45789063e-01 2.75285661e-01
1.07545279e-01 -2.93464124e-01 -7.73580894e-02 7.78598249e-01
-4.41692978e-01 7.20690638e-02 3.34431008e-02 -3.74521285e-01
-1.65896618e+00 7.65328944e-01 5.04822552e-01 -2.03214571e-01
-5.41942239e-01 6.47215068e-01 -5.39960228e-02 -3.00742298e-01
3.50230813e-01 -1.55806512e-01 2.30001494e-01 -1.91851154e-01
2.46068969e-01 2.11798579e-01 4.10650969e-01 -7.26714015e-01
-5.94962120e-01 4.58234906e-01 3.19090970e-02 -1.00453854e-01
7.41877377e-01 2.29397446e-01 -1.07866354e-01 9.45712347e-03
1.08820891e+00 5.14698148e-01 -1.27720582e+00 7.80485645e-02
1.42562598e-01 -3.01517844e-01 -1.71202913e-01 -4.43127155e-01
-1.23783386e+00 6.40516043e-01 2.91509390e-01 5.48769474e-01
9.11912918e-01 -1.88601300e-01 3.83282900e-01 3.18925887e-01
5.72239935e-01 -8.50181758e-01 -3.83169740e-01 2.29401663e-01
4.90136892e-01 -8.33997905e-01 2.39060879e-01 -6.63431168e-01
-3.69704902e-01 9.77254808e-01 2.18560517e-01 -2.84135044e-01
6.93088889e-01 4.95089859e-01 7.90742338e-02 7.40803927e-02
-7.63826013e-01 1.54465944e-01 1.29770681e-01 6.25228584e-01
1.95585005e-02 1.79368928e-01 -2.53400952e-01 4.63371396e-01
-4.01792407e-01 -6.21853530e-01 7.98433661e-01 8.54612112e-01
-3.36327910e-01 -1.15098906e+00 -4.89777535e-01 3.14567715e-01
-7.27374136e-01 8.60385820e-02 -6.92152798e-01 8.48880529e-01
-1.03415258e-01 9.87119079e-01 -1.87116459e-01 -6.02014840e-01
2.05357134e-01 -6.02670312e-02 4.82154042e-01 -2.30940223e-01
-6.05344534e-01 -3.61258328e-01 1.94287106e-01 -3.82021457e-01
1.95116132e-01 -5.81616580e-01 -1.31466055e+00 -4.95646894e-01
-3.45105946e-01 2.74274319e-01 7.05111742e-01 8.05324793e-01
3.26405019e-01 1.13010317e-01 8.15177798e-01 -1.93371847e-01
-7.94630885e-01 -4.03527439e-01 -6.28721774e-01 5.20141900e-01
3.98498118e-01 -3.76825154e-01 -9.63078856e-01 -7.19157308e-02] | [5.94571590423584, 7.406943321228027] |
3cb3c406-08a3-434f-af70-e81ea31f9f60 | weisfeiler-and-leman-go-relational | 2211.17113 | null | https://arxiv.org/abs/2211.17113v1 | https://arxiv.org/pdf/2211.17113v1.pdf | Weisfeiler and Leman Go Relational | Knowledge graphs, modeling multi-relational data, improve numerous applications such as question answering or graph logical reasoning. Many graph neural networks for such data emerged recently, often outperforming shallow architectures. However, the design of such multi-relational graph neural networks is ad-hoc, driven mainly by intuition and empirical insights. Up to now, their expressivity, their relation to each other, and their (practical) learning performance is poorly understood. Here, we initiate the study of deriving a more principled understanding of multi-relational graph neural networks. Namely, we investigate the limitations in the expressive power of the well-known Relational GCN and Compositional GCN architectures and shed some light on their practical learning performance. By aligning both architectures with a suitable version of the Weisfeiler-Leman test, we establish under which conditions both models have the same expressive power in distinguishing non-isomorphic (multi-relational) graphs or vertices with different structural roles. Further, by leveraging recent progress in designing expressive graph neural networks, we introduce the $k$-RN architecture that provably overcomes the expressiveness limitations of the above two architectures. Empirically, we confirm our theoretical findings in a vertex classification setting over small and large multi-relational graphs. | ['Miguel Romero Orth', 'Christopher Morris', 'Mikhail Galkin', 'Pablo Barcelo'] | 2022-11-30 | null | null | null | null | ['logical-reasoning'] | ['reasoning'] | [ 1.04106873e-01 5.37429810e-01 -4.80959862e-01 -2.91782737e-01
-2.31481701e-01 -7.73773074e-01 4.05571282e-01 2.98876345e-01
-4.54728641e-02 5.77541411e-01 7.59782940e-02 -7.74187148e-01
-6.78818345e-01 -1.37744415e+00 -9.38756645e-01 -4.34799850e-01
-4.97341007e-01 6.81836009e-01 2.35086486e-01 -3.34957778e-01
-2.37923592e-01 5.66296160e-01 -1.45636833e+00 2.47452661e-01
6.81284904e-01 9.33209300e-01 -3.86052966e-01 7.58981168e-01
-9.00943875e-02 1.30585325e+00 -1.29817471e-01 -7.49243021e-01
2.75797814e-01 -2.39189699e-01 -1.11602437e+00 -2.06806302e-01
4.55858320e-01 -9.65406150e-02 -8.36410999e-01 1.08638012e+00
1.44260406e-01 8.38459134e-02 4.31947052e-01 -1.61731791e+00
-1.13695955e+00 1.32202625e+00 -2.98799723e-01 1.72298551e-01
2.09886923e-01 -1.31646723e-01 1.85374320e+00 -2.99635649e-01
5.74676394e-01 1.24026573e+00 9.34820771e-01 4.36069667e-01
-1.08401179e+00 -4.09155160e-01 2.74372160e-01 1.28487870e-01
-1.41854084e+00 -2.56862789e-01 8.75482261e-01 -1.51144877e-01
1.06396985e+00 4.14028227e-01 5.93174398e-01 1.00284779e+00
-2.10010737e-01 7.54317760e-01 7.99666584e-01 -4.80802268e-01
-1.23960607e-01 -1.70784295e-02 3.92587960e-01 1.08396959e+00
6.26393735e-01 5.20534553e-02 -2.71505743e-01 -1.20717637e-01
8.04498851e-01 -1.87123194e-01 -3.39738965e-01 -6.69155121e-01
-7.90868640e-01 1.01564038e+00 7.31534421e-01 6.26693070e-01
9.46871042e-02 6.22210562e-01 5.87820828e-01 6.45969212e-01
1.99574932e-01 4.94007289e-01 -3.29555005e-01 3.72446150e-01
-3.61900538e-01 2.71904916e-02 1.05897319e+00 1.10879958e+00
6.44887388e-01 5.56809120e-02 1.55976832e-01 4.41541523e-01
1.87674060e-01 9.57187489e-02 1.39557332e-01 -5.76896369e-01
5.27001262e-01 1.05990422e+00 -5.32445252e-01 -1.26248765e+00
-6.94206715e-01 -2.96795696e-01 -1.09044790e+00 -4.17477369e-01
4.39405769e-01 1.52308971e-01 -5.83790183e-01 1.99805582e+00
1.78682953e-01 -4.82692197e-02 6.78319409e-02 6.30304992e-01
8.25012624e-01 3.29905599e-01 -7.86871761e-02 -8.50416496e-02
1.21337283e+00 -7.24704683e-01 -3.42477202e-01 -8.00736770e-02
1.05606341e+00 2.14372054e-01 1.07346523e+00 1.96340263e-01
-1.11494362e+00 -2.45850563e-01 -1.02290773e+00 -4.43767130e-01
-6.01339459e-01 -3.63001168e-01 1.33992827e+00 7.38537610e-01
-1.36622393e+00 6.16567850e-01 -5.74629903e-01 -3.42381358e-01
3.44073832e-01 5.55438161e-01 -6.78099275e-01 -1.91091478e-01
-1.45528674e+00 6.54983640e-01 6.81333780e-01 3.10604215e-01
-5.04076898e-01 -5.28194845e-01 -9.74593103e-01 2.29796454e-01
9.33559835e-01 -8.54013324e-01 9.29693043e-01 -9.59957898e-01
-1.16755509e+00 9.57389712e-01 3.92738909e-01 -6.20447636e-01
2.24668711e-01 3.12732458e-01 -5.14934897e-01 2.72738039e-01
-3.27795178e-01 5.04723266e-02 1.54559299e-01 -1.10840428e+00
-2.53744513e-01 -6.41821325e-01 8.60857666e-01 -4.44872398e-03
-3.46572816e-01 -1.16670169e-01 -3.54540706e-01 -3.32664251e-01
4.30569239e-02 -9.15893495e-01 -2.24060521e-01 -9.56468582e-02
-6.29921675e-01 -3.89234155e-01 3.40521365e-01 1.11245669e-01
1.24869084e+00 -1.92477620e+00 3.15042108e-01 6.79430246e-01
9.62202311e-01 1.45410106e-01 -2.92338520e-01 6.49204075e-01
-2.16243625e-01 3.92838895e-01 -1.50690665e-02 1.65009096e-01
3.15685689e-01 4.78393227e-01 -3.31065536e-01 4.26164061e-01
6.94653243e-02 1.39224291e+00 -8.21052551e-01 -4.93594080e-01
-1.17448866e-01 1.40033752e-01 -5.84820688e-01 1.20131977e-01
-3.86297196e-01 -2.45313987e-01 -4.68215168e-01 6.57420337e-01
3.94320726e-01 -8.61355007e-01 7.87713647e-01 -7.00973943e-02
4.83613968e-01 1.82607800e-01 -1.16660464e+00 1.28212023e+00
-3.02594692e-01 5.72656393e-01 8.08499381e-02 -1.39220750e+00
6.80818200e-01 1.24223888e-01 2.96054959e-01 -6.23361945e-01
5.89589998e-02 1.24154411e-01 2.55120039e-01 -4.30342913e-01
5.29540956e-01 -3.78647089e-01 -7.21409218e-03 5.56241632e-01
8.80652592e-02 2.43230209e-01 3.09619069e-01 4.69737113e-01
1.27787626e+00 -2.70576656e-01 3.47252607e-01 -2.92938888e-01
4.11894709e-01 -2.98727721e-01 2.56966114e-01 1.01548553e+00
-1.81166381e-02 1.70566201e-01 1.16955507e+00 -5.14054775e-01
-7.95377314e-01 -9.03671861e-01 8.36270228e-02 1.41134667e+00
2.85867844e-02 -7.77773857e-01 -4.10362244e-01 -7.39412606e-01
1.80901140e-01 1.97015047e-01 -8.39519322e-01 -5.03275275e-01
-6.90074742e-01 -6.88494980e-01 1.15536058e+00 8.00885797e-01
-2.85014622e-02 -9.08729434e-01 -2.59769231e-01 -1.13041744e-01
1.71416834e-01 -1.34637702e+00 -5.70637099e-02 4.06305701e-01
-7.58855879e-01 -1.61308587e+00 -5.44606298e-02 -7.41046369e-01
4.89486694e-01 2.07404658e-01 1.43666053e+00 5.86761355e-01
1.07425675e-01 4.38794523e-01 -2.14228824e-01 1.11413440e-02
-4.94192630e-01 4.46046531e-01 7.11863860e-02 -2.32629046e-01
2.11198419e-01 -8.98010671e-01 -1.53524846e-01 1.74982771e-01
-1.21097934e+00 3.33065018e-02 6.39505446e-01 5.35349905e-01
3.84077340e-01 3.00872803e-01 5.33670723e-01 -1.47095180e+00
8.94140184e-01 -5.62221706e-01 -7.21546590e-01 6.01967633e-01
-7.99264431e-01 4.64206785e-01 9.63532388e-01 -4.20665503e-01
-4.35609311e-01 -4.45412338e-01 7.67949875e-03 -4.10837054e-01
2.23935738e-01 1.01880002e+00 -3.68226826e-01 -3.19338441e-01
6.70748413e-01 1.45195052e-02 -2.14721009e-01 6.68974444e-02
7.56670892e-01 2.45808765e-01 4.20239598e-01 -1.11086822e+00
8.51340771e-01 2.86295474e-01 4.67787832e-01 -6.87474668e-01
-8.06753993e-01 -1.42065763e-01 -5.24020612e-01 1.05874151e-01
5.55868387e-01 -5.23343086e-01 -1.29562140e+00 2.33521774e-01
-9.09196913e-01 -4.99072015e-01 -1.76594496e-01 3.85213457e-02
-5.21330893e-01 5.90493619e-01 -9.58826363e-01 -7.94202626e-01
-2.45131344e-01 -1.01126254e+00 7.03165650e-01 -2.04978585e-01
4.22614478e-02 -1.30471230e+00 -2.26996597e-02 1.72580928e-01
3.84314895e-01 3.61152530e-01 1.75810802e+00 -9.96745050e-01
-7.94826627e-01 -3.70401843e-03 -5.50355434e-01 9.78600979e-03
-3.76665443e-01 -1.75422922e-01 -6.65351510e-01 -2.00490505e-01
-4.44567740e-01 -6.84715033e-01 8.43816996e-01 9.09841135e-02
1.42174578e+00 -5.09598017e-01 -3.16525966e-01 7.25624740e-01
1.50795567e+00 -2.36289918e-01 4.97919410e-01 1.05523653e-01
1.10381043e+00 6.17888689e-01 -3.32537949e-01 -1.82278436e-02
7.62739658e-01 4.11623567e-01 6.36287987e-01 2.81099260e-01
1.55357391e-01 -4.87281978e-01 -6.75265864e-02 9.30720150e-01
-4.19001311e-01 -5.10398269e-01 -1.06307840e+00 3.29366535e-01
-1.92303240e+00 -8.99106562e-01 -2.34238118e-01 2.03225660e+00
7.86441386e-01 3.08563620e-01 2.36187398e-01 8.69428441e-02
5.63777745e-01 4.88570750e-01 -4.77093488e-01 -3.03903788e-01
-3.84905934e-01 2.81884342e-01 6.12809300e-01 4.62757945e-01
-8.31307054e-01 9.53959405e-01 6.33128452e+00 5.36181748e-01
-7.84448266e-01 -1.38585523e-01 4.56473142e-01 2.53225584e-02
-7.74911404e-01 9.47587006e-03 -4.41005409e-01 -1.59522876e-01
1.28348792e+00 -2.12423474e-01 8.94664466e-01 8.45928967e-01
-6.79820597e-01 4.80636120e-01 -1.63828075e+00 9.01405752e-01
4.40390557e-02 -1.58514202e+00 2.62913167e-01 2.92324185e-01
3.40564102e-01 6.15124702e-02 -7.65061453e-02 7.48164415e-01
7.35451937e-01 -1.57376051e+00 4.64424312e-01 2.35399738e-01
8.17327738e-01 -7.56286144e-01 4.35693890e-01 2.19782546e-01
-1.38603258e+00 -3.07575107e-01 -3.64221036e-01 -1.04862675e-01
-3.56357276e-01 3.73146296e-01 -3.71740460e-01 9.22542036e-01
3.64830554e-01 4.51159120e-01 -6.32718980e-01 2.28724316e-01
-1.32350251e-01 2.82680094e-01 -3.42244923e-01 -1.68550815e-02
3.59680295e-01 -1.58444628e-01 1.95365846e-01 1.02130294e+00
-7.17476085e-02 3.31155986e-01 1.25663530e-03 1.01471853e+00
-6.09735012e-01 -9.72185060e-02 -1.04283476e+00 -5.11976719e-01
3.69180948e-01 1.19399190e+00 -7.02210307e-01 -6.84734955e-02
-5.21605372e-01 4.73199040e-01 1.12091267e+00 3.89466673e-01
-7.97248721e-01 -3.51895481e-01 3.73619258e-01 1.15642898e-01
2.12434530e-01 -1.65406525e-01 -4.15011905e-02 -1.31517780e+00
2.02898785e-01 -1.03119671e+00 8.22625399e-01 -6.06926858e-01
-1.40931368e+00 7.03051209e-01 9.08367336e-02 -4.53031272e-01
-1.99943930e-01 -1.06929088e+00 -4.16359186e-01 4.36858624e-01
-1.42955697e+00 -1.45347512e+00 -7.05519468e-02 7.82027483e-01
-4.54603642e-01 4.49446216e-02 8.10354471e-01 3.03703010e-01
-5.01372755e-01 9.09418166e-01 -2.57863790e-01 5.02903581e-01
1.16814874e-01 -1.48962688e+00 2.84490585e-01 7.54213095e-01
5.05655289e-01 8.93727779e-01 4.21190947e-01 -3.09054852e-01
-2.14706516e+00 -9.79663372e-01 6.61991715e-01 -5.24447381e-01
1.18195736e+00 -5.48259795e-01 -1.11420691e+00 1.24620008e+00
-1.88491166e-01 4.90697622e-01 7.24238575e-01 9.31605697e-01
-9.41220105e-01 -1.54242560e-01 -8.38311672e-01 6.65301085e-01
1.60472202e+00 -1.02116835e+00 -3.64518493e-01 3.30800503e-01
1.05233562e+00 -3.85179460e-01 -1.16793358e+00 4.37579125e-01
5.10113537e-01 -1.06649292e+00 9.43105221e-01 -1.24195695e+00
6.60518706e-01 5.24312444e-02 -3.88553172e-01 -1.01843429e+00
-3.63741577e-01 -5.36197245e-01 -5.70563614e-01 8.60156238e-01
3.38048786e-01 -8.41340303e-01 8.72867286e-01 6.21518195e-01
-1.20024934e-01 -1.04800594e+00 -8.38004947e-01 -7.98699617e-01
3.26338679e-01 -5.63836217e-01 8.02974284e-01 1.26224172e+00
3.05057526e-01 7.29843974e-01 -2.47265309e-01 3.44051659e-01
3.25366914e-01 4.36019272e-01 6.91130400e-01 -1.47582769e+00
-6.45012856e-01 -8.71123850e-01 -6.27184689e-01 -9.64222431e-01
6.08324349e-01 -1.41811454e+00 -4.98290837e-01 -1.53663313e+00
2.78358042e-01 -5.64739287e-01 -4.03551966e-01 6.89028084e-01
1.39432758e-01 -3.07104420e-02 1.77533478e-02 -1.22822687e-01
-8.31453860e-01 2.92735666e-01 1.09849608e+00 -1.78894758e-01
2.42010746e-02 -2.11952597e-01 -1.14245999e+00 4.72989947e-01
4.87435997e-01 -1.67730391e-01 -8.29224586e-01 -5.07885277e-01
1.07281995e+00 2.46200487e-01 4.27330852e-01 -5.04643619e-01
3.82002056e-01 -2.32437134e-01 -1.49766430e-01 -1.15350462e-01
-8.13048426e-03 -9.15269256e-01 1.77398071e-01 2.40055785e-01
-5.83710790e-01 2.22545475e-01 -7.15695834e-03 8.01141262e-01
-5.67388237e-02 -1.64917242e-02 4.00504738e-01 -2.81880707e-01
-4.20401752e-01 7.01667786e-01 1.75323188e-01 4.86760885e-01
8.53330672e-01 -2.27894947e-01 -8.12739074e-01 -3.39742035e-01
-4.23502654e-01 2.92380273e-01 2.19968408e-01 3.46168369e-01
4.82016385e-01 -1.32227540e+00 -4.39310998e-01 1.36072382e-01
4.16722506e-01 2.13409394e-01 1.69093639e-01 9.56193209e-01
-4.58330423e-01 5.13398528e-01 5.00240326e-02 -1.67802304e-01
-9.00228858e-01 1.06669974e+00 6.00745618e-01 -5.61786711e-01
-6.28771663e-01 6.01877630e-01 4.04825240e-01 -6.60240531e-01
1.89971715e-01 -6.96770132e-01 6.02280200e-02 -2.26245090e-01
1.72221109e-01 2.43496135e-01 3.37010473e-02 -4.16864157e-01
-2.99646288e-01 1.91793263e-01 -1.56759992e-01 4.08091903e-01
1.34538782e+00 2.32197195e-01 -4.04062808e-01 5.18311739e-01
1.16527593e+00 4.16630581e-02 -5.14916658e-01 -3.92492741e-01
2.99478143e-01 -4.94222753e-02 -2.92976975e-01 -4.22251284e-01
-1.16243315e+00 7.40334630e-01 -3.40593338e-01 9.21961010e-01
9.95243371e-01 4.53868061e-01 5.72817981e-01 9.33146060e-01
2.68705428e-01 -7.77163863e-01 -1.12267599e-01 5.80940127e-01
5.92870176e-01 -1.13639164e+00 1.05385277e-02 -4.28550720e-01
-3.01581919e-01 1.22511148e+00 5.96739769e-01 2.14460511e-02
5.57007909e-01 3.03013008e-02 -4.47650880e-01 -7.00183570e-01
-9.00875151e-01 -2.20110804e-01 3.52781415e-01 5.38284719e-01
3.12188476e-01 2.38132402e-01 1.58601776e-01 7.03888476e-01
-3.30726892e-01 -1.83102146e-01 4.10538435e-01 3.94272894e-01
-2.00278193e-01 -9.25372839e-01 2.93138504e-01 5.42611003e-01
-4.56101179e-01 -2.64611989e-01 -5.87784767e-01 1.27930784e+00
-2.88764924e-01 6.74126685e-01 -6.70335665e-02 -4.34753090e-01
2.34708786e-01 -1.95877805e-01 7.33749449e-01 -4.38425153e-01
-4.31276828e-01 -6.22451842e-01 4.49733436e-01 -5.06640017e-01
-3.69193614e-01 1.46876592e-02 -1.13371682e+00 -7.55336523e-01
-3.62627089e-01 8.34379196e-02 3.43147442e-02 9.21523571e-01
3.58726382e-01 4.10076648e-01 2.84145385e-01 -9.37720388e-02
-7.27277994e-01 -6.29637837e-01 -8.30564976e-01 3.75747710e-01
2.75613755e-01 -3.58427644e-01 -3.49492222e-01 -6.11322045e-01] | [7.02721643447876, 6.332330703735352] |
584a7cc9-ea1b-48fa-a595-d22dcfc8c003 | zoomnas-searching-for-whole-body-human-pose | 2208.11547 | null | https://arxiv.org/abs/2208.11547v1 | https://arxiv.org/pdf/2208.11547v1.pdf | ZoomNAS: Searching for Whole-body Human Pose Estimation in the Wild | This paper investigates the task of 2D whole-body human pose estimation, which aims to localize dense landmarks on the entire human body including body, feet, face, and hands. We propose a single-network approach, termed ZoomNet, to take into account the hierarchical structure of the full human body and solve the scale variation of different body parts. We further propose a neural architecture search framework, termed ZoomNAS, to promote both the accuracy and efficiency of whole-body pose estimation. ZoomNAS jointly searches the model architecture and the connections between different sub-modules, and automatically allocates computational complexity for searched sub-modules. To train and evaluate ZoomNAS, we introduce the first large-scale 2D human whole-body dataset, namely COCO-WholeBody V1.0, which annotates 133 keypoints for in-the-wild images. Extensive experiments demonstrate the effectiveness of ZoomNAS and the significance of COCO-WholeBody V1.0. | ['Xiaogang Wang', 'Ping Luo', 'Wanli Ouyang', 'Chen Qian', 'Wentao Liu', 'Sheng Jin', 'Lumin Xu'] | 2022-08-23 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [-5.80032408e-01 2.66059786e-01 -1.23812808e-02 -2.50185013e-01
-3.94105643e-01 1.04348827e-02 8.78069270e-03 -4.00379092e-01
-3.94056708e-01 2.53855824e-01 3.16395402e-01 7.55012035e-01
3.20513062e-02 -3.96163821e-01 -6.44104004e-01 -1.48468599e-01
-1.56163275e-01 9.27103281e-01 2.61764973e-01 -2.66149580e-01
-5.36002457e-01 3.65011483e-01 -1.23882687e+00 -2.81310290e-01
3.59223545e-01 1.06116128e+00 1.98463537e-03 2.60114193e-01
6.10895216e-01 2.65292048e-01 -5.81773460e-01 -6.77157283e-01
4.15109128e-01 -2.75781333e-01 -7.21893668e-01 1.45542964e-01
7.54484236e-01 -2.75603294e-01 -4.77293700e-01 8.99091125e-01
9.15784419e-01 2.14194879e-01 3.91554028e-01 -1.30588770e+00
2.55392082e-02 2.12181926e-01 -8.39981437e-01 -8.93282369e-02
3.19944948e-01 2.60291994e-01 7.74270415e-01 -9.98026252e-01
6.92796648e-01 1.62264633e+00 1.10470378e+00 6.91819072e-01
-8.52553546e-01 -6.36568189e-01 1.68418393e-01 -1.72639400e-01
-1.82233822e+00 -4.06520724e-01 7.03489363e-01 -4.34894651e-01
9.51594710e-01 -1.58722416e-01 1.14193010e+00 9.31614995e-01
4.27371621e-01 1.07356369e+00 3.52554411e-01 -3.23537052e-01
-1.76827461e-01 -7.84159064e-01 -1.23488918e-01 1.37544167e+00
4.08325374e-01 6.64579794e-02 -7.62194514e-01 -8.66753310e-02
1.18869710e+00 -1.89094245e-01 -4.88437600e-02 -9.47179675e-01
-1.36547029e+00 4.98120368e-01 8.14295650e-01 -1.79805741e-01
-4.70876127e-01 5.73675036e-01 5.18180192e-01 -2.21493483e-01
9.08609629e-02 3.06456119e-01 -6.39174402e-01 2.54537314e-01
-8.76472592e-01 4.16882485e-01 5.65777421e-01 9.75334585e-01
3.26986372e-01 -1.84371695e-01 -7.23279193e-02 9.66378570e-01
6.69857800e-01 6.17511094e-01 4.53372180e-01 -9.05660629e-01
5.12179196e-01 8.05663049e-01 -6.54696301e-02 -1.05914128e+00
-9.96898770e-01 -3.50675434e-01 -9.65105355e-01 -4.87304572e-03
3.77077460e-01 -2.38162830e-01 -1.01735127e+00 1.81333935e+00
1.02209651e+00 -2.39085451e-01 -4.14584935e-01 1.35949099e+00
1.11056280e+00 1.19616397e-01 2.45385617e-01 4.92346168e-01
1.58924842e+00 -1.52497852e+00 -2.49065757e-01 -5.81212878e-01
3.14445198e-01 -2.08521754e-01 7.37280846e-01 9.70926806e-02
-1.26710737e+00 -9.55196679e-01 -9.68686223e-01 -1.77635118e-01
-1.06017627e-01 6.31129384e-01 5.09733975e-01 1.74928010e-01
-8.46233666e-01 4.02174979e-01 -1.27401686e+00 -4.08782542e-01
2.60273039e-01 6.59614801e-01 -8.59757900e-01 2.97247320e-01
-1.15112066e+00 8.81256700e-01 5.64651549e-01 6.78774536e-01
-8.46633077e-01 -2.44775876e-01 -1.52798688e+00 -1.53942481e-01
5.45333624e-01 -1.41396987e+00 1.14506829e+00 -4.65176284e-01
-1.49203622e+00 1.27287471e+00 8.27373266e-02 -2.71812409e-01
5.96683621e-01 -9.07944918e-01 -8.75378922e-02 1.68331996e-01
1.62311554e-01 1.16757274e+00 8.12057018e-01 -7.32884049e-01
-4.29676533e-01 -6.76405013e-01 -3.21253419e-01 4.08594877e-01
5.88677898e-02 -1.13941118e-01 -1.31896639e+00 -7.95788527e-01
2.89355576e-01 -1.08386946e+00 -3.78562212e-01 2.68758118e-01
-7.65582502e-01 -2.94074237e-01 1.28023267e-01 -7.98721373e-01
1.05918562e+00 -1.85648143e+00 7.58824468e-01 4.62276101e-01
3.46237332e-01 -1.22398682e-01 -1.57259360e-01 -2.49455255e-02
1.04051515e-01 -6.18759215e-01 2.22603101e-02 -6.63493693e-01
1.66565657e-01 1.55631885e-01 5.58457375e-01 8.32988858e-01
-1.17892630e-01 1.43283641e+00 -4.68021005e-01 -9.90582526e-01
2.29954690e-01 4.76877242e-01 -4.66993421e-01 9.80731323e-02
2.76074093e-03 2.60074258e-01 -4.10650641e-01 1.07635462e+00
2.84502536e-01 -5.75203001e-01 1.90861717e-01 -5.35625875e-01
4.94170427e-01 -2.01225474e-01 -1.28854930e+00 2.38479877e+00
5.72218336e-02 -1.50825724e-03 3.27093810e-01 -3.01104635e-01
7.19237387e-01 1.49624139e-01 6.60492182e-01 -6.41812146e-01
3.39129776e-01 -1.10465243e-01 -2.50604063e-01 -1.77000925e-01
1.59236372e-01 5.89383245e-02 -3.24834794e-01 9.71809253e-02
4.93199527e-01 2.05739304e-01 2.48200595e-02 -9.81297195e-02
7.43855953e-01 5.26922584e-01 7.29228139e-01 -3.44874680e-01
4.54904288e-01 -1.93693325e-01 4.87715304e-01 4.01674479e-01
-5.40563643e-01 6.71807110e-01 -1.77584123e-02 -7.21742213e-01
-7.78551042e-01 -1.12349319e+00 3.05480212e-01 1.30822718e+00
3.90814483e-01 -6.87133491e-01 -9.46798205e-01 -7.85177886e-01
4.07175392e-01 -3.95919293e-01 -8.49694729e-01 -2.27304220e-01
-8.83010685e-01 -4.56552535e-01 6.56386435e-01 9.00607884e-01
8.15501750e-01 -1.16831887e+00 -1.21397889e+00 4.84362133e-02
-4.39302713e-01 -9.59090173e-01 -8.30296278e-01 -6.11947551e-02
-7.99150765e-01 -1.31867218e+00 -9.83110547e-01 -8.80123258e-01
5.80197692e-01 -4.93406087e-01 1.38675427e+00 1.03463857e-02
-7.45548069e-01 4.28977787e-01 9.87104625e-02 -8.27168077e-02
3.55868220e-01 3.19948256e-01 3.09229583e-01 -3.17784280e-01
9.01381671e-02 -2.79775113e-01 -7.21143305e-01 6.99601948e-01
-1.27393212e-02 9.48397517e-02 6.46951854e-01 6.28034115e-01
7.36779749e-01 -1.99616849e-01 -2.95535326e-02 -2.08341748e-01
2.00437605e-01 9.54782870e-03 -4.63567615e-01 3.70669544e-01
-1.28167406e-01 -1.05416570e-02 -1.16265394e-01 -2.61350363e-01
-6.45137429e-01 6.01566434e-01 -4.01829541e-01 -6.84789240e-01
-1.08453929e-01 3.79868150e-01 -4.78635579e-01 -2.15329111e-01
5.01890123e-01 -4.33274247e-02 2.56643534e-01 -6.27020895e-01
2.71919250e-01 -9.03675184e-02 1.17458451e+00 -6.60522401e-01
6.51922286e-01 3.44735295e-01 1.36853635e-01 -4.85653639e-01
-9.11709487e-01 -6.81313455e-01 -1.26920569e+00 -3.69906157e-01
1.00484049e+00 -1.34226632e+00 -8.84671628e-01 5.76767743e-01
-9.15953100e-01 -4.95094389e-01 -1.73142925e-01 4.40539539e-01
-6.86407387e-01 3.43472689e-01 -8.97540212e-01 -3.39778274e-01
-8.92777145e-01 -7.93662965e-01 1.79153287e+00 5.67569546e-02
-7.30104685e-01 -8.44280660e-01 3.54788035e-01 2.02515423e-01
-2.73050457e-01 5.87227941e-01 3.64652455e-01 -2.85899252e-01
-1.83916558e-02 -4.03610677e-01 5.50777428e-02 1.85842328e-02
-1.71038464e-01 -5.10865927e-01 -4.50046659e-01 -6.51032567e-01
-4.63855028e-01 -6.76687300e-01 6.27277315e-01 7.37118840e-01
7.86259532e-01 -1.54084682e-01 -4.89590913e-01 8.75457525e-01
8.91037166e-01 -4.70091403e-01 3.20792973e-01 3.96259129e-01
9.71018493e-01 6.32625520e-01 7.48956919e-01 5.20781040e-01
5.64652801e-01 9.11077142e-01 5.03062785e-01 -3.68799955e-01
-1.97433770e-01 -5.67000687e-01 1.87654749e-01 6.42394125e-01
-2.86748469e-01 1.07020266e-01 -8.54392588e-01 4.52497125e-01
-1.82416713e+00 -5.70989609e-01 4.08593059e-01 1.83733177e+00
6.58090591e-01 1.29479513e-01 6.25057340e-01 -3.86849016e-01
7.98692822e-01 1.10017426e-01 -6.86355770e-01 4.54212695e-01
1.02557570e-01 1.18566237e-01 2.25367457e-01 2.50504941e-01
-1.60720229e+00 1.02567911e+00 6.60602951e+00 5.59397697e-01
-7.23293602e-01 -2.03845222e-02 7.06928000e-02 -3.67986768e-01
7.48659492e-01 -5.74537158e-01 -1.03560662e+00 2.72161484e-01
3.03660899e-01 3.13895166e-01 8.37778673e-03 1.24260271e+00
-2.10690394e-01 2.00850263e-01 -1.30273986e+00 1.36487651e+00
3.20123255e-01 -7.13672400e-01 -9.52740386e-02 2.60569919e-02
4.09218907e-01 9.46789458e-02 -3.28666270e-01 2.54014432e-01
1.68655232e-01 -9.56034422e-01 9.44376647e-01 4.83556300e-01
7.06811726e-01 -7.20671594e-01 6.27209246e-01 4.07716393e-01
-1.62508082e+00 1.06684022e-01 -3.24124664e-01 1.27439201e-01
3.34533095e-01 1.99436754e-01 -4.38093334e-01 3.75001341e-01
1.12031567e+00 7.40673780e-01 -8.67140651e-01 1.14171958e+00
-4.42179590e-01 7.64821246e-02 -6.31632090e-01 2.63652921e-01
-9.45924819e-02 2.63438404e-01 4.01059091e-01 1.03821456e+00
6.07700236e-02 -3.61205190e-02 6.43319190e-01 4.82034802e-01
-1.16214156e-01 6.11019693e-02 -1.08879879e-01 3.90745997e-01
1.32898718e-01 1.21277666e+00 -6.44401193e-01 -3.65269750e-01
-1.29614305e-02 1.31820297e+00 4.07939166e-01 8.95736739e-02
-8.28677893e-01 -2.05521688e-01 6.33216441e-01 3.57378989e-01
1.87203765e-01 -2.34745905e-01 2.60616601e-01 -1.20961380e+00
1.33818671e-01 -9.89651263e-01 9.15615320e-01 -8.30600441e-01
-1.18913329e+00 4.10450190e-01 1.83682323e-01 -8.51868212e-01
-2.27073476e-01 -6.48454130e-01 -1.14827603e-01 6.27711356e-01
-5.50430119e-01 -1.58993971e+00 -5.58414996e-01 1.03443551e+00
5.30292213e-01 1.37800485e-01 6.49417937e-01 2.33798966e-01
-6.16286874e-01 7.45603979e-01 -6.21004879e-01 7.37527788e-01
8.08156729e-01 -1.04288530e+00 9.27304447e-01 5.85142434e-01
1.02776445e-01 8.72814715e-01 2.91403443e-01 -1.06844962e+00
-1.19018435e+00 -1.04128432e+00 7.34353125e-01 -6.70851827e-01
1.29617259e-01 -4.93605494e-01 -3.52146775e-01 9.31323349e-01
-2.14626446e-01 1.76167458e-01 4.49217081e-01 4.01804954e-01
-2.62228400e-01 1.04681633e-01 -9.30667341e-01 6.01089418e-01
1.48067915e+00 -2.91816115e-01 -9.00068164e-01 1.44252688e-01
5.29403567e-01 -9.64199245e-01 -9.77671444e-01 6.91829681e-01
1.04153061e+00 -6.27757668e-01 1.50785983e+00 -7.17195868e-01
2.34003872e-01 -3.31319511e-01 -7.42077008e-02 -1.01921725e+00
-5.85108280e-01 -2.14353159e-01 -6.53521657e-01 5.48727930e-01
3.16253752e-02 -1.86000258e-01 1.16657066e+00 5.52468061e-01
4.68375348e-02 -7.89925218e-01 -1.03605855e+00 -5.90888202e-01
-1.49382651e-01 -3.04273516e-02 3.93999249e-01 5.33793867e-01
-1.50791720e-01 4.59810704e-01 -8.37844193e-01 2.31241375e-01
9.51748013e-01 1.35383487e-01 1.21110129e+00 -1.31551230e+00
-3.94466013e-01 -1.12134248e-01 -8.71345758e-01 -1.31132996e+00
4.74792421e-02 -5.71317136e-01 2.30272606e-01 -1.35311520e+00
2.85856366e-01 1.52662799e-01 -1.82718158e-01 9.00919974e-01
-2.15440020e-01 5.20222664e-01 2.42335424e-01 3.57944936e-01
-1.08517277e+00 5.99847376e-01 1.34103930e+00 -1.12344123e-01
-1.37516394e-01 2.72572376e-02 -2.18625695e-01 1.09160340e+00
3.72097790e-01 -2.57591963e-01 -2.02442214e-01 -4.07946169e-01
4.12001461e-02 -7.57951895e-03 8.60121846e-01 -1.48707235e+00
3.96015137e-01 2.20042273e-01 1.17889297e+00 -9.94947135e-01
5.72880328e-01 -7.57507384e-01 4.13478762e-01 7.20739543e-01
-1.47334769e-01 3.37227523e-01 1.81449413e-01 4.67774957e-01
-1.61853924e-01 4.37672406e-01 7.80305088e-01 -5.88293910e-01
-9.39587295e-01 6.24265909e-01 3.25588673e-01 2.16546848e-01
9.29834247e-01 -2.33978286e-01 2.93489814e-01 -5.60355559e-02
-1.20624924e+00 7.62363553e-01 4.33543950e-01 5.35606742e-01
7.04498470e-01 -1.35593808e+00 -3.84852171e-01 2.86687285e-01
1.47098973e-01 2.86781281e-01 3.81729245e-01 8.41786563e-01
-6.92664146e-01 4.12642747e-01 -3.59310776e-01 -8.56306255e-01
-1.42779207e+00 4.90601569e-01 8.96172822e-01 -3.66967112e-01
-1.01031315e+00 1.13101864e+00 4.46457595e-01 -8.58551919e-01
4.76740092e-01 -2.03856736e-01 7.00478181e-02 -1.65982187e-01
2.55252451e-01 3.83531332e-01 -1.86622888e-01 -1.01851869e+00
-7.01573908e-01 1.04671681e+00 3.35520804e-01 5.19980351e-03
8.26955914e-01 -1.86224639e-01 -9.69320014e-02 1.67450141e-02
1.04127288e+00 -3.30732346e-01 -1.30357146e+00 -2.19945639e-01
-2.01799989e-01 -2.83300072e-01 -4.43761140e-01 -9.13980842e-01
-1.20929563e+00 7.71597505e-01 7.23125994e-01 -9.00448918e-01
9.50512767e-01 2.22329661e-01 8.55964005e-01 5.43730557e-01
8.12092185e-01 -1.24722493e+00 3.10905218e-01 4.01831418e-01
1.08817899e+00 -1.06979752e+00 3.20660174e-01 -3.25415611e-01
-6.21180058e-01 8.71520221e-01 1.20660603e+00 -1.43222645e-01
6.62468135e-01 1.15420269e-02 1.12706847e-01 -6.11773193e-01
-2.37065971e-01 -1.95845455e-01 7.54697144e-01 4.47470367e-01
1.24647036e-01 8.09759181e-03 7.62007311e-02 7.27779031e-01
-3.44717532e-01 3.12962048e-02 -7.24694610e-01 1.06500781e+00
-2.79324353e-01 -6.48878098e-01 -6.41581953e-01 9.39380750e-02
-3.15959841e-01 2.49473989e-01 -6.79890633e-01 1.21570861e+00
3.25894415e-01 3.09632927e-01 -2.25337014e-01 -4.65023488e-01
7.24685073e-01 2.16049150e-01 6.23476267e-01 -4.52991843e-01
-5.83839953e-01 3.70104671e-01 -7.36054704e-02 -1.20717490e+00
-4.03330207e-01 -4.25687879e-01 -1.30324364e+00 -1.21258430e-01
-2.36205488e-01 -1.31260161e-03 3.12372893e-01 8.73504162e-01
4.14212883e-01 4.26767707e-01 -4.13714349e-01 -1.26776826e+00
-5.56215227e-01 -1.03928483e+00 -6.74488962e-01 2.48876512e-01
1.83185712e-01 -1.13578641e+00 3.01695228e-01 4.73158434e-02] | [7.028871536254883, -0.8862003087997437] |
5d6e42d2-d73b-4253-9162-da52950da4ba | learning-to-exploit-temporal-structure-for | 2301.04558 | null | https://arxiv.org/abs/2301.04558v2 | https://arxiv.org/pdf/2301.04558v2.pdf | Learning to Exploit Temporal Structure for Biomedical Vision-Language Processing | Self-supervised learning in vision-language processing exploits semantic alignment between imaging and text modalities. Prior work in biomedical VLP has mostly relied on the alignment of single image and report pairs even though clinical notes commonly refer to prior images. This does not only introduce poor alignment between the modalities but also a missed opportunity to exploit rich self-supervision through existing temporal content in the data. In this work, we explicitly account for prior images and reports when available during both training and fine-tuning. Our approach, named BioViL-T, uses a CNN-Transformer hybrid multi-image encoder trained jointly with a text model. It is designed to be versatile to arising challenges such as pose variations and missing input images across time. The resulting model excels on downstream tasks both in single- and multi-image setups, achieving state-of-the-art performance on (I) progression classification, (II) phrase grounding, and (III) report generation, whilst offering consistent improvements on disease classification and sentence-similarity tasks. We release a novel multi-modal temporal benchmark dataset, MS-CXR-T, to quantify the quality of vision-language representations in terms of temporal semantics. Our experimental results show the advantages of incorporating prior images and reports to make most use of the data. | ['Fernando Pérez-García', 'Ozan Oktay', 'Javier Alvarez-Valle', 'Aditya Nori', 'Matthew P. Lungren', 'Maria Wetscherek', 'Anton Schwaighofer', 'Anja Thieme', 'Kenza Bouzid', 'Harshita Sharma', 'Benedikt Boecking', 'Daniel C. Castro', 'Maximilian Ilse', 'Qianchu Liu', 'Stephanie Hyland', 'Shruthi Bannur'] | 2023-01-11 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Bannur_Learning_To_Exploit_Temporal_Structure_for_Biomedical_Vision-Language_Processing_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Bannur_Learning_To_Exploit_Temporal_Structure_for_Biomedical_Vision-Language_Processing_CVPR_2023_paper.pdf | cvpr-2023-1 | ['phrase-grounding'] | ['natural-language-processing'] | [ 5.89233875e-01 9.59054567e-03 -3.26356798e-01 -5.80821395e-01
-1.63635242e+00 -6.04379892e-01 9.51291740e-01 3.71756375e-01
-4.76744205e-01 6.12772584e-01 5.86160123e-01 -1.91836685e-01
-1.14841163e-01 -2.94086456e-01 -5.88834405e-01 -4.57295179e-01
4.73576263e-02 5.09431362e-01 1.16977192e-01 6.93249255e-02
-1.09923899e-01 -1.47581948e-02 -1.06315219e+00 9.25174177e-01
2.34278053e-01 1.00504148e+00 5.11654675e-01 9.82438922e-01
6.62821755e-02 9.43811893e-01 -1.68299258e-01 -5.99086642e-01
-1.10596091e-01 -5.44088900e-01 -1.04132819e+00 4.11603570e-01
6.99640632e-01 -2.29132965e-01 -4.03471082e-01 5.37267029e-01
8.69820297e-01 -2.63218999e-01 6.54723942e-01 -8.88848901e-01
-7.46015012e-01 4.78208810e-01 -7.12664843e-01 7.82821715e-01
2.82270938e-01 5.54482222e-01 9.54075336e-01 -6.42889738e-01
1.15844190e+00 8.51303279e-01 7.13178635e-01 5.72170675e-01
-1.48356259e+00 -1.38677523e-01 -9.46870968e-02 2.15719357e-01
-1.12537086e+00 -6.39042079e-01 2.64049411e-01 -6.01902783e-01
1.37540352e+00 2.59227931e-01 3.60313177e-01 1.47515798e+00
4.85965192e-01 7.37672746e-01 1.29328728e+00 -4.03078347e-01
-1.95815712e-01 1.05917893e-01 1.32245729e-02 7.89659739e-01
-4.12865371e-01 -1.40741142e-03 -8.76766860e-01 5.43310307e-02
5.25480211e-01 -4.08323891e-02 -4.59117711e-01 1.25567555e-01
-1.81030762e+00 6.53363943e-01 3.34717363e-01 5.10504901e-01
-3.81090492e-01 9.16912630e-02 7.24488854e-01 2.50816703e-01
6.03541255e-01 1.14386596e-01 -3.16885829e-01 6.05467185e-02
-1.33815157e+00 3.10693346e-02 3.70891660e-01 8.00698161e-01
1.36817545e-01 -3.65264267e-01 -1.00213885e+00 9.01730120e-01
4.96612489e-02 3.33762378e-01 8.05668175e-01 -8.21824312e-01
7.53573596e-01 2.59822160e-01 -3.81905526e-01 -6.46284223e-01
-6.13498032e-01 -4.72182572e-01 -9.21052337e-01 -2.46182084e-01
2.57935911e-01 1.52055904e-01 -1.29985857e+00 1.84929371e+00
1.57059617e-02 4.58104275e-02 1.19118422e-01 6.55499518e-01
1.13540745e+00 2.70984113e-01 1.26329362e-01 -4.56637442e-01
1.60656881e+00 -9.83583868e-01 -8.12695563e-01 -2.34432220e-01
7.14823306e-01 -8.89170647e-01 8.64210486e-01 -1.66295432e-02
-1.31352472e+00 -3.81948143e-01 -7.04406738e-01 -3.54930848e-01
-1.70648828e-01 2.32167646e-01 4.99473184e-01 9.81779546e-02
-1.54638994e+00 5.89139640e-01 -9.23525751e-01 -6.33178115e-01
6.58880591e-01 2.97388345e-01 -7.27131128e-01 -4.62999016e-01
-8.96769345e-01 1.03783309e+00 3.28019291e-01 -1.43929288e-01
-8.30556810e-01 -1.11628950e+00 -8.55674982e-01 -4.83338684e-01
2.24267974e-01 -1.16585791e+00 1.27082086e+00 -6.29767954e-01
-9.89881277e-01 1.59491551e+00 -2.76246697e-01 -7.51507163e-01
6.33566916e-01 2.95721292e-01 -3.90868574e-01 6.77717984e-01
5.73396921e-01 1.15839744e+00 9.91443038e-01 -6.09099627e-01
-5.02884746e-01 -3.81690413e-01 -2.11607754e-01 2.51023233e-01
-9.55228731e-02 1.03044927e-01 -7.08482683e-01 -9.04030263e-01
-1.97093636e-01 -9.87439632e-01 -1.48033470e-01 2.60892659e-01
-4.42373216e-01 -8.85428116e-02 6.05494499e-01 -8.90049458e-01
9.38040674e-01 -1.97344971e+00 1.27398729e-01 -2.55617201e-01
3.19553286e-01 -1.85645759e-01 -2.06783220e-01 3.31096202e-01
-4.44278538e-01 9.05438438e-02 -2.48328164e-01 -8.85383427e-01
-4.76386875e-01 2.60522842e-01 -2.65608400e-01 5.42646527e-01
5.72973847e-01 1.23480320e+00 -8.22510242e-01 -8.94950092e-01
1.00569122e-01 4.19790894e-01 -4.95051593e-01 4.15852368e-02
-4.90296967e-02 8.41561139e-01 -9.47605446e-02 7.38956928e-01
3.99418101e-02 -8.55415225e-01 2.43520308e-02 -6.95663214e-01
-3.64751853e-02 5.51524758e-01 -3.99412662e-01 2.45504189e+00
-6.89321518e-01 6.28946722e-01 -1.48032472e-01 -9.39603448e-01
2.29716629e-01 7.36912727e-01 8.76182795e-01 -9.67579126e-01
-1.63401797e-01 2.58985728e-01 -9.23517793e-02 -6.87024832e-01
1.78848028e-01 -3.84963721e-01 9.80932042e-02 4.88835424e-01
4.97300714e-01 -2.35925958e-01 3.21784854e-01 5.87376714e-01
1.40289366e+00 3.62409838e-02 1.52796298e-01 4.14942764e-02
3.56942594e-01 1.59352884e-01 2.02742502e-01 9.29452002e-01
-1.89751312e-01 9.91562665e-01 3.52427512e-01 -9.29361656e-02
-1.20123398e+00 -1.20226133e+00 -3.44390631e-01 7.47670889e-01
-4.55990821e-01 -5.83269477e-01 -2.01161042e-01 -7.06834972e-01
-2.12485105e-01 5.25253892e-01 -7.95429587e-01 -2.76691411e-02
-4.63496566e-01 -8.97769153e-01 6.46895170e-01 6.13067389e-01
1.09591097e-01 -9.34504628e-01 -6.10951841e-01 3.35250854e-01
-4.96771574e-01 -1.64466894e+00 -8.29035878e-01 2.92952150e-01
-9.14603710e-01 -8.77496719e-01 -1.15546811e+00 -5.77125669e-01
4.17697370e-01 4.73163873e-02 1.28834522e+00 -3.08097005e-01
-7.57894397e-01 1.05356681e+00 -3.06137085e-01 -3.40911388e-01
-4.76232082e-01 -7.45527679e-03 -2.15779006e-01 -9.74119827e-03
-1.09634943e-01 -3.98984075e-01 -6.58378005e-01 -2.68273652e-01
-9.79250610e-01 3.64762634e-01 8.43283296e-01 1.31083453e+00
8.68509293e-01 -4.25886005e-01 2.96881318e-01 -6.05282128e-01
3.43226165e-01 -4.05554384e-01 5.80709614e-02 4.23260212e-01
-5.75594962e-01 2.09945500e-01 9.02779847e-02 -2.96400636e-01
-8.83866489e-01 -1.08910970e-01 -5.05380007e-03 -5.81306159e-01
-3.50078046e-02 6.22309327e-01 6.22798443e-01 1.13782309e-01
4.64354247e-01 6.13792896e-01 3.35779876e-01 -2.20079720e-01
4.68127757e-01 4.70133066e-01 9.37495530e-01 -3.27141494e-01
3.85105401e-01 8.41596127e-01 1.65292412e-01 -5.58235943e-01
-1.27773571e+00 -7.29586422e-01 -6.58358157e-01 -8.73491019e-02
1.09830725e+00 -1.14572442e+00 -3.72303039e-01 1.67537242e-01
-1.25877225e+00 -1.31335735e-01 -3.61175060e-01 3.95348638e-01
-7.91317165e-01 2.21114144e-01 -8.25919330e-01 -2.17704877e-01
-7.23149121e-01 -1.50349534e+00 1.41149974e+00 -1.32176980e-01
-3.32750410e-01 -1.19790375e+00 1.70163885e-02 5.50738692e-01
3.39556664e-01 2.24821016e-01 7.93691337e-01 -4.12531167e-01
-5.34970820e-01 9.10440087e-02 -4.28323925e-01 2.15284213e-01
1.98049754e-01 -3.58508766e-01 -1.10029638e+00 -4.24716830e-01
-9.32406411e-02 -7.40802348e-01 1.28765225e+00 5.67020297e-01
1.06622314e+00 -2.31912851e-01 -2.57976770e-01 6.57997787e-01
1.31774223e+00 -2.25977242e-01 4.26679254e-01 2.11783975e-01
6.63814843e-01 6.77308857e-01 3.26126933e-01 2.61005968e-01
6.20730579e-01 8.04655790e-01 1.50247574e-01 -2.86786944e-01
-5.08834124e-01 6.88010827e-02 1.75536439e-01 9.13631380e-01
2.59725928e-01 -8.21581930e-02 -1.09827590e+00 9.68531370e-01
-1.70813513e+00 -9.40843046e-01 5.89973778e-02 1.74052453e+00
1.34657991e+00 8.16351622e-02 -1.15992669e-02 -1.27177864e-01
2.93326318e-01 2.69064516e-01 -3.85938615e-01 -9.02509987e-02
-2.57624775e-01 4.49570000e-01 5.58136046e-01 2.43562698e-01
-1.26566124e+00 4.97676820e-01 5.93744040e+00 7.60276377e-01
-1.23946428e+00 6.94917440e-01 8.12779129e-01 -5.51930726e-01
-1.78249612e-01 -3.08658600e-01 -6.86803937e-01 2.50447989e-01
1.20428538e+00 2.39827354e-02 6.92325179e-03 7.05467314e-02
2.75663882e-01 -1.02125123e-01 -1.56063831e+00 1.22274816e+00
2.57787466e-01 -1.96333170e+00 2.50085056e-01 -1.09820865e-01
5.45419633e-01 4.15487200e-01 4.07707363e-01 -4.62352894e-02
-1.06470279e-01 -1.18461001e+00 7.79109836e-01 6.80171013e-01
1.32247901e+00 -2.00891227e-01 6.55276060e-01 -2.14610416e-02
-1.05963445e+00 2.91607916e-01 2.06317708e-01 4.89549786e-01
2.50068635e-01 4.72797781e-01 -1.11399984e+00 8.36488783e-01
7.04748631e-01 1.15291727e+00 -7.07033336e-01 8.82889450e-01
1.78697988e-01 4.88544941e-01 -2.94433177e-01 5.28166175e-01
4.43546653e-01 3.94159496e-01 5.74640632e-01 1.57509971e+00
1.52586982e-01 -8.02620947e-02 2.95556277e-01 8.96348298e-01
-2.48705465e-02 -7.35806255e-03 -7.61541605e-01 -2.81372726e-01
-1.43043529e-02 1.18858910e+00 -7.27062225e-01 -3.94121945e-01
-7.18032777e-01 1.14878941e+00 2.01110646e-01 9.19838920e-02
-6.56314492e-01 1.57909870e-01 2.62987554e-01 9.90856960e-02
3.20964545e-01 -9.68855917e-02 -3.89322400e-01 -1.08599365e+00
1.82542115e-01 -6.68089092e-01 8.72243583e-01 -9.29308951e-01
-1.47160232e+00 5.48543513e-01 -1.33290932e-01 -1.23862135e+00
-4.80353862e-01 -4.83413309e-01 -2.99633205e-01 1.02167010e+00
-1.71240354e+00 -1.59339619e+00 4.17842269e-02 7.65715241e-01
6.46114647e-01 -1.02915131e-01 9.37481880e-01 4.02526736e-01
-3.48194957e-01 7.07527280e-01 -1.11579865e-01 9.10982955e-03
1.09158695e+00 -1.18590260e+00 5.54306470e-02 7.04005539e-01
2.18613818e-01 5.26426435e-01 4.41451371e-01 -4.86078531e-01
-1.22992063e+00 -1.24537277e+00 8.97322416e-01 -7.96628416e-01
8.82351935e-01 3.63593921e-02 -5.54508150e-01 6.62962258e-01
4.04967546e-01 2.81545460e-01 8.06939960e-01 -1.94060251e-01
-4.04299110e-01 -1.21050373e-01 -9.41733122e-01 5.18246531e-01
8.23246896e-01 -1.09703910e+00 -5.44968605e-01 8.28189790e-01
9.73511517e-01 -7.30716050e-01 -1.34490073e+00 3.66398126e-01
3.06214094e-01 -7.00820267e-01 1.11204004e+00 -6.53823674e-01
8.69149625e-01 -1.40596041e-02 -1.71168685e-01 -9.27254140e-01
-1.16318762e-01 -5.22077084e-01 1.82555001e-02 1.22140205e+00
5.57333171e-01 -1.77509531e-01 3.86761665e-01 3.04566354e-01
-2.74433434e-01 -9.78980243e-01 -1.19798994e+00 -4.26391721e-01
-8.43182281e-02 -6.88694298e-01 -2.80776560e-01 9.42873895e-01
-1.33935004e-01 4.14579779e-01 -1.70248792e-01 3.74002084e-02
6.13379776e-01 4.30934802e-02 3.23565230e-02 -5.08469880e-01
-6.62362218e-01 -4.17912871e-01 -3.11184645e-01 -3.68955284e-01
-3.08382604e-02 -1.38536894e+00 -1.67892233e-01 -1.64730453e+00
6.30360484e-01 -6.28857985e-02 -3.97367865e-01 8.00656974e-01
-1.15766369e-01 6.51035964e-01 1.77310377e-01 4.42299336e-01
-7.65169144e-01 2.07681462e-01 1.31082606e+00 -5.19687831e-01
-2.24659573e-02 -2.90299118e-01 -6.47914946e-01 4.12280738e-01
2.78279275e-01 -5.14372468e-01 -3.63564819e-01 -6.04330838e-01
-4.58123721e-03 4.57637727e-01 7.90410876e-01 -8.43168557e-01
2.88965344e-01 1.42516643e-01 3.93861502e-01 -7.58853137e-01
3.83536100e-01 -4.62875664e-01 -6.42952994e-02 4.31137025e-01
-7.04296291e-01 2.14125186e-01 4.65991974e-01 6.78697705e-01
-3.20111454e-01 5.22167720e-02 8.40555251e-01 -5.23941934e-01
-4.91364926e-01 4.31525379e-01 2.52246168e-02 4.13692892e-01
7.56405354e-01 1.18731536e-01 -4.47085142e-01 -9.34704170e-02
-9.61798549e-01 3.37151915e-01 1.20035075e-01 4.86238360e-01
6.95697427e-01 -1.08561754e+00 -9.41189468e-01 -1.53840303e-01
4.93368804e-01 -7.01076612e-02 5.42148173e-01 1.58785903e+00
-1.33696884e-01 8.73141348e-01 1.24395102e-01 -1.23395002e+00
-1.34982336e+00 3.52911979e-01 3.25731814e-01 -8.93588722e-01
-1.05092394e+00 9.28273022e-01 1.85857341e-01 -8.72653052e-02
1.51435867e-01 -3.92215610e-01 -5.03476104e-03 2.06000179e-01
6.59584701e-01 -4.16059703e-01 4.77355659e-01 -5.72344184e-01
-5.17606020e-01 4.49145347e-01 -8.05964291e-01 -4.69847709e-01
1.41235995e+00 -3.14825296e-01 -3.89148220e-02 6.01037562e-01
1.40396273e+00 -5.26293635e-01 -9.18700695e-01 -5.98730206e-01
-5.34527702e-03 7.39235058e-02 4.74882931e-01 -1.16469014e+00
-8.66726756e-01 8.90449822e-01 7.71080971e-01 -3.58414173e-01
1.08692253e+00 3.27083468e-01 8.70268226e-01 1.50323749e-01
8.69669169e-02 -8.31092715e-01 4.50562805e-01 3.22416872e-01
1.02692962e+00 -1.48486400e+00 1.13089606e-01 -5.54883443e-02
-9.27059710e-01 9.58401859e-01 1.87228143e-01 3.33407402e-01
2.99585968e-01 3.00262481e-01 5.81580140e-02 -2.78422445e-01
-1.17753363e+00 -2.87158072e-01 7.42703021e-01 6.66894197e-01
6.16543591e-01 -1.56972557e-01 -2.79532969e-02 3.19478124e-01
1.51431292e-01 2.27440432e-01 4.26346540e-01 1.00528693e+00
3.15915853e-01 -9.54781651e-01 -1.68751046e-01 7.40104914e-01
-9.64856029e-01 -5.08584619e-01 1.31856874e-01 5.30850708e-01
5.60164377e-02 6.63064241e-01 -1.39103802e-02 -1.48545012e-01
1.01486444e-01 7.08303601e-02 8.24624956e-01 -9.16511536e-01
-4.91809368e-01 2.25204065e-01 1.41702637e-01 -7.02190578e-01
-7.83556581e-01 -8.22061419e-01 -1.02611494e+00 2.69501716e-01
1.66419610e-01 -4.23522413e-01 6.07039750e-01 1.21729159e+00
3.58979404e-01 1.08539033e+00 2.15546235e-01 -7.15904176e-01
-5.54225087e-01 -8.50842774e-01 -3.06891948e-01 5.04402459e-01
6.81336761e-01 -5.26848316e-01 1.11210860e-01 6.08584523e-01] | [14.932223320007324, -1.820898413658142] |
0fa6b82a-2997-4bea-bb05-fa030fe9ebc1 | local-gaussian-process-extrapolation-for-bart | 2204.10963 | null | https://arxiv.org/abs/2204.10963v2 | https://arxiv.org/pdf/2204.10963v2.pdf | Local Gaussian process extrapolation for BART models with applications to causal inference | Bayesian additive regression trees (BART) is a semi-parametric regression model offering state-of-the-art performance on out-of-sample prediction. Despite this success, standard implementations of BART typically provide inaccurate prediction and overly narrow prediction intervals at points outside the range of the training data. This paper proposes a novel extrapolation strategy that grafts Gaussian processes to the leaf nodes in BART for predicting points outside the range of the observed data. The new method is compared to standard BART implementations and recent frequentist resampling-based methods for predictive inference. We apply the new approach to a challenging problem from causal inference, wherein for some regions of predictor space, only treated or untreated units are observed (but not both). In simulation studies, the new approach boasts superior performance compared to popular alternatives, such as Jackknife+. | ['P. Richard Hahn', 'Jingyu He', 'Meijiang Wang'] | 2022-04-23 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 3.82339120e-01 2.29518577e-01 -4.23762858e-01 -5.32664716e-01
-8.77586186e-01 3.08155045e-02 7.51357436e-01 2.57275134e-01
-2.02464178e-01 1.48444784e+00 1.20019995e-01 -7.20949829e-01
-4.96863276e-01 -9.30756330e-01 -7.42323816e-01 -7.72578001e-01
-1.43156216e-01 8.76230538e-01 2.69807369e-01 2.45112643e-01
2.53598481e-01 3.60661358e-01 -1.19949830e+00 5.20653836e-02
1.02551305e+00 6.22854710e-01 -1.47275716e-01 6.77307427e-01
2.09351763e-01 9.03437853e-01 -4.00984049e-01 -2.40013197e-01
-1.17439404e-01 -2.48078823e-01 -3.00760120e-01 -3.47205788e-01
9.07225832e-02 -3.75319153e-01 -1.30890012e-01 5.18868208e-01
3.89403760e-01 2.63424069e-01 1.17753565e+00 -1.19919181e+00
-5.30471504e-01 8.01692784e-01 -8.56608868e-01 1.95078298e-01
3.27331960e-01 -2.39267901e-01 6.19116724e-01 -9.03218746e-01
1.18561335e-01 1.44985938e+00 1.26211226e+00 9.54115167e-02
-2.07074308e+00 -7.28391528e-01 -8.13969523e-02 1.75688714e-01
-1.52596879e+00 -4.79469746e-01 1.71484038e-01 -7.06657290e-01
8.67600203e-01 1.51558563e-01 4.34034258e-01 1.24990880e+00
7.10328937e-01 4.91883457e-01 1.25834036e+00 -5.49027920e-01
7.73084760e-01 1.09664693e-01 3.44119519e-01 4.65917997e-02
4.03726429e-01 6.11777782e-01 -4.36375439e-01 -1.03358245e+00
8.48486125e-01 2.77142912e-01 -1.68324858e-02 -3.08335632e-01
-6.46141946e-01 1.17568851e+00 -8.58957171e-02 -4.08559263e-01
-7.13171601e-01 4.18839306e-01 2.21351653e-01 -1.39117315e-01
9.85523343e-01 -2.72735238e-01 -5.29126227e-01 -5.02058193e-02
-1.25039876e+00 5.81521392e-01 8.57641220e-01 8.18001509e-01
3.52662981e-01 2.84174029e-02 -5.72760344e-01 8.76280725e-01
4.39797640e-01 3.21271181e-01 2.76324302e-02 -7.46666074e-01
2.55764812e-01 1.11388035e-01 6.58052385e-01 -4.44971532e-01
-3.82675618e-01 -3.45890373e-01 -9.38988268e-01 1.07274987e-01
6.42683923e-01 -3.58927637e-01 -9.15007710e-01 1.28492641e+00
5.11905551e-01 5.33003867e-01 -3.05723011e-01 2.19444752e-01
1.12756073e-01 7.88365960e-01 5.66711247e-01 -7.29520261e-01
9.24939394e-01 -4.39157754e-01 -7.01742053e-01 -6.50693700e-02
3.56963694e-01 -5.17909110e-01 5.87863922e-01 6.16214573e-01
-9.29056048e-01 -3.64972204e-01 -5.69899142e-01 3.50782931e-01
6.14866316e-02 -1.39055848e-01 6.75561190e-01 8.80778790e-01
-6.92224920e-01 9.59238708e-01 -9.93076980e-01 -5.73256351e-02
4.66321945e-01 2.54971564e-01 1.46328464e-01 -3.42525691e-01
-9.87024784e-01 8.16958308e-01 1.36304557e-01 9.92668569e-02
-9.17723596e-01 -1.31721008e+00 -4.96635437e-01 1.85528547e-01
3.97184700e-01 -7.35062838e-01 1.45958769e+00 -4.09054130e-01
-1.34954011e+00 2.19672203e-01 -4.78926063e-01 -7.97679067e-01
8.30482960e-01 -5.43643117e-01 -2.27282971e-01 -5.28810203e-01
8.71608257e-02 -2.23868378e-02 8.51422787e-01 -9.58568752e-01
-6.19461238e-01 -5.21652222e-01 -8.47099304e-01 -1.71835870e-01
4.21289951e-01 1.30073711e-01 3.14617753e-01 -6.76267147e-01
-5.32186590e-02 -7.50307918e-01 -7.53730178e-01 -3.58247161e-01
-5.39035916e-01 -3.63073528e-01 4.37323421e-01 -6.18142009e-01
1.44541609e+00 -1.85047436e+00 -3.93445015e-01 3.64627659e-01
-3.41831595e-02 -2.99600810e-01 6.48116767e-01 7.12581098e-01
-3.37949157e-01 9.17840302e-02 -3.31024736e-01 -2.39074200e-01
-2.02366516e-01 3.76338065e-01 -5.90885818e-01 7.33275473e-01
-4.35160585e-02 3.83294880e-01 -8.90886486e-01 -4.77756858e-01
2.91859090e-01 3.80109698e-01 -3.56475681e-01 1.11570969e-01
-3.57070528e-02 3.67416561e-01 -2.29128957e-01 4.11003739e-01
6.04838490e-01 -8.30631703e-02 2.46077403e-02 5.42867362e-01
-1.53970286e-01 4.95726839e-02 -1.19909394e+00 5.74508488e-01
-1.79701984e-01 2.95388550e-01 -3.33958685e-01 -1.08831644e+00
1.01065493e+00 5.88232398e-01 4.34013903e-01 3.03728729e-01
-2.41159990e-01 3.57088633e-02 -1.62133604e-01 1.00378118e-01
2.62103975e-01 -5.61745465e-01 1.04035594e-01 1.74538955e-01
-1.81605980e-01 7.13647064e-03 -1.86659634e-01 -2.26597518e-01
1.29233229e+00 3.59520972e-01 1.05005074e+00 -3.44092906e-01
-1.38163287e-02 4.14814092e-02 8.63724828e-01 1.37990379e+00
3.97308730e-02 2.53757417e-01 5.58757305e-01 -4.12541300e-01
-1.08686614e+00 -1.49852347e+00 -6.75198078e-01 1.25235772e+00
-4.81811523e-01 -2.05157310e-01 -2.84066677e-01 -3.07599008e-01
3.91845375e-01 1.51087832e+00 -8.15033436e-01 1.25644669e-01
-1.27262831e-01 -1.29920888e+00 3.06511521e-01 6.85023546e-01
5.74222617e-02 -7.18900681e-01 -3.63436341e-01 6.98939800e-01
2.90236264e-01 -6.48609340e-01 2.38684788e-01 5.62461674e-01
-1.32917917e+00 -8.71605456e-01 -5.48610985e-01 5.67424968e-02
3.22945058e-01 -8.25790763e-02 1.08022070e+00 -5.54797769e-01
-7.25618303e-02 -6.24342859e-02 -1.66952163e-01 -8.51721525e-01
-5.81569195e-01 -3.91953290e-01 1.00077325e-02 -3.13113570e-01
5.32549560e-01 -5.80728650e-01 -4.34983402e-01 4.14725542e-01
-1.43938750e-01 -1.28032103e-01 3.09358269e-01 1.14446616e+00
5.88823438e-01 1.15120158e-01 8.70227516e-01 -1.31850219e+00
5.95739901e-01 -8.68828356e-01 -6.46552742e-01 1.91563427e-01
-7.96845198e-01 -1.66619405e-01 6.12585604e-01 -5.19420207e-01
-1.46196878e+00 -2.57970206e-02 2.41562247e-01 -2.42056623e-01
-3.04320663e-01 6.67489171e-01 3.78596187e-01 5.26229322e-01
9.48749363e-01 -1.24648459e-01 -9.43108648e-02 -6.03007793e-01
5.33377491e-02 6.25891030e-01 4.30023193e-01 -5.29210925e-01
2.50303596e-01 2.49744162e-01 3.00615162e-01 -6.28237665e-01
-7.17311382e-01 -4.31657791e-01 -5.45586646e-01 -1.21318512e-02
3.64262998e-01 -8.63446355e-01 -5.91698229e-01 9.87901017e-02
-7.51617968e-01 -5.64605057e-01 -5.25069296e-01 8.12601984e-01
-1.02973247e+00 9.69917551e-02 -5.78059494e-01 -1.66786098e+00
-2.17767328e-01 -6.49153233e-01 9.82747018e-01 -8.13316274e-03
-6.08256817e-01 -1.07026172e+00 2.49945775e-01 3.80621786e-04
2.21419744e-02 3.24114621e-01 1.06395984e+00 -1.03108835e+00
8.49119872e-02 -6.03726864e-01 -5.18167466e-02 -9.37110260e-02
-1.81796551e-01 4.04418588e-01 -8.11243773e-01 -9.18751433e-02
-2.33250819e-02 1.15806885e-01 7.62219727e-01 1.20940781e+00
1.25426424e+00 -2.65617222e-01 -8.92563939e-01 2.23404020e-01
1.26191700e+00 4.00997102e-01 7.20264733e-01 9.09121484e-02
1.23004235e-01 4.76598829e-01 8.10469866e-01 1.15703642e+00
1.70818970e-01 6.19380713e-01 1.03947982e-01 2.58016169e-01
4.85465050e-01 -5.31907737e-01 2.89052039e-01 2.58969460e-02
-2.70467818e-01 4.86768372e-02 -9.63029981e-01 4.90188062e-01
-2.29946709e+00 -1.37996984e+00 -6.84129596e-01 2.70840192e+00
1.12475526e+00 1.86229408e-01 2.80825824e-01 9.72537324e-02
1.00015843e+00 -4.78678018e-01 -4.76443797e-01 -6.85988665e-01
3.70790273e-01 3.48026425e-01 8.12408924e-01 4.59929109e-01
-1.08548069e+00 4.98258322e-01 7.85482931e+00 1.08369446e+00
-2.58583069e-01 2.07293943e-01 7.83474684e-01 1.37041792e-01
1.00799516e-01 2.88357258e-01 -1.24802983e+00 5.25286198e-01
1.54179037e+00 -2.78230250e-01 9.60764736e-02 1.04061580e+00
7.96206772e-01 -6.00568593e-01 -1.37434220e+00 3.58627617e-01
-5.20485699e-01 -1.05051243e+00 -5.85627079e-01 1.08365491e-01
8.45135391e-01 -1.44169629e-01 -1.66198209e-01 5.96282840e-01
1.23984981e+00 -1.33299994e+00 3.67441267e-01 8.67005467e-01
6.63580239e-01 -7.96426892e-01 8.63225102e-01 8.52386475e-01
-6.20378852e-01 -2.85095185e-01 -6.74483955e-01 -3.63490671e-01
3.35372597e-01 1.21513200e+00 -1.31682360e+00 2.56824613e-01
7.38945842e-01 5.74445367e-01 -1.93218738e-01 1.43478119e+00
-2.74177134e-01 1.45269835e+00 -5.85928857e-01 7.95789436e-02
-8.68237615e-02 -1.92727402e-01 3.48657668e-01 1.19954324e+00
5.86916625e-01 8.43654871e-02 4.14404972e-03 8.20229590e-01
5.71483612e-01 -5.66952415e-02 -8.01255107e-01 6.22298062e-01
8.56576920e-01 4.26147640e-01 -3.95909727e-01 -5.31401038e-01
-3.60173166e-01 1.57762781e-01 2.10212395e-01 5.39912999e-01
-8.43110502e-01 1.01166569e-01 1.19995847e-01 2.17526644e-01
3.37270916e-01 6.36923090e-02 -8.36185813e-01 -5.26175439e-01
-6.84296787e-01 -3.65155756e-01 6.55277431e-01 -7.84631252e-01
-1.68382120e+00 -3.33733886e-01 6.83031380e-01 -8.72065783e-01
-7.26662338e-01 -2.88057745e-01 -6.65163815e-01 1.23343277e+00
-6.87848210e-01 -7.33757496e-01 1.06955506e-01 2.76539177e-01
5.83092630e-01 3.65896523e-02 1.07450044e+00 -3.39431167e-01
-4.89109933e-01 2.88484156e-01 9.40010726e-01 -5.36577880e-01
7.36759722e-01 -1.36841094e+00 2.77323246e-01 3.76663059e-01
-3.67120028e-01 6.13409340e-01 1.24054396e+00 -1.05552590e+00
-6.51102602e-01 -1.10612321e+00 6.91459656e-01 -4.25555915e-01
7.38047123e-01 -4.88695763e-02 -9.80933428e-01 1.01619720e+00
-2.32174426e-01 -1.22119859e-01 1.01447976e+00 8.05312574e-01
-8.23479369e-02 -1.16646767e-01 -1.29313743e+00 6.28244281e-01
4.66497064e-01 5.96139254e-03 -6.42461002e-01 4.54727322e-01
2.74866730e-01 5.36765195e-02 -1.13358212e+00 6.12706482e-01
7.90656686e-01 -9.66225207e-01 9.75793540e-01 -7.99795866e-01
5.80418348e-01 1.16263971e-01 -1.96324751e-01 -1.49111927e+00
-5.12354255e-01 -6.72383785e-01 -4.05617118e-01 1.08955371e+00
4.64477360e-01 -7.46324778e-01 7.99120665e-01 7.40576923e-01
1.30779341e-01 -7.94826865e-01 -1.20302379e+00 -7.14011610e-01
3.11415523e-01 -6.39163852e-01 2.63919145e-01 6.59402847e-01
3.22673768e-02 3.33068073e-01 -6.59119546e-01 7.81832039e-02
1.20765686e+00 -1.24737248e-01 7.92962909e-01 -1.65423727e+00
-3.88031006e-01 -5.85767217e-02 -2.57051677e-01 -6.29149079e-01
-2.04767901e-02 -1.71938851e-01 1.78997859e-01 -1.34988415e+00
4.32021379e-01 -6.04586124e-01 -8.82038176e-02 3.85684967e-01
-4.45933193e-01 -1.84374452e-01 -4.27443832e-01 1.37056231e-01
2.08641455e-01 4.73621279e-01 7.55694449e-01 2.40644932e-01
-4.55872893e-01 7.20131278e-01 -3.51143211e-01 9.49281871e-01
7.97074080e-01 -7.83236623e-01 -4.10767704e-01 4.34588879e-01
1.01973815e-02 9.00419593e-01 7.03257620e-01 -6.14901483e-01
1.41070068e-01 -5.88820755e-01 8.48463893e-01 -1.11611629e+00
1.93800479e-01 -7.46025205e-01 8.46306741e-01 3.21569264e-01
-4.92630959e-01 -2.03478515e-01 3.85369472e-02 1.18343234e+00
2.22400531e-01 -4.88241792e-01 6.87746227e-01 8.18675160e-02
-1.50273591e-01 -1.19266659e-01 -6.70708537e-01 -3.01763892e-01
1.12132812e+00 -3.89057964e-01 -3.44784379e-01 -5.11472821e-01
-1.12410784e+00 1.13078102e-01 3.67131643e-02 -2.53738344e-01
3.97059858e-01 -9.06745732e-01 -9.67183232e-01 -1.67368576e-01
-1.71899617e-01 -2.25523561e-02 1.54566690e-01 1.18656600e+00
-2.14582533e-01 3.24156672e-01 1.97668001e-01 -6.45550430e-01
-1.13737071e+00 6.52812839e-01 3.33338156e-02 -4.64494944e-01
-6.73810780e-01 5.36785483e-01 1.86254323e-01 -2.50190675e-01
2.45165136e-02 -1.47680342e-01 -1.03774175e-01 -3.86578515e-02
5.35922527e-01 1.06587660e+00 -2.18414485e-01 -5.53994402e-02
-5.20364493e-02 -1.11745246e-01 -4.97456044e-02 -1.65441260e-01
1.39656746e+00 -9.07747746e-02 -6.38421029e-02 1.22829497e+00
4.01621491e-01 -2.25179017e-01 -1.42173243e+00 -1.35187745e-01
1.98626831e-01 -6.32217228e-01 3.36259902e-02 -8.05510640e-01
5.18990345e-02 7.77945101e-01 3.14489156e-01 3.22230339e-01
7.09277451e-01 -1.94233671e-01 -2.24267036e-01 7.78623000e-02
4.72563237e-01 -9.85693455e-01 -7.39298165e-01 1.79492831e-01
7.44958460e-01 -9.99725223e-01 6.48903549e-01 -5.73702812e-01
-4.75004554e-01 9.36795890e-01 1.57173052e-01 -3.99377465e-01
9.06357706e-01 3.87462527e-01 -6.32567346e-01 2.50292182e-01
-1.34480572e+00 4.47818488e-01 3.25423889e-02 8.43447208e-01
6.78506315e-01 5.40262938e-01 -4.05170709e-01 7.61055648e-01
-2.21374959e-01 3.83064002e-01 5.74157178e-01 7.11246133e-01
-5.04908264e-01 -5.74170530e-01 -9.06931818e-01 1.13292515e+00
-7.42157996e-01 -2.36487031e-01 1.18453518e-01 9.75004315e-01
-3.43538880e-01 1.16475153e+00 1.28321245e-01 2.15803146e-01
2.39287511e-01 3.19333404e-01 1.96487725e-01 -6.48969293e-01
-3.67232203e-01 3.86485249e-01 3.42328012e-01 -2.68319756e-01
-1.11533098e-01 -1.08166087e+00 -7.62319088e-01 -7.11531937e-01
-6.72528684e-01 2.23430037e-01 3.89888406e-01 7.71620870e-01
-6.90382197e-02 4.54138607e-01 5.71217716e-01 -8.42010677e-01
-1.24424541e+00 -1.31753194e+00 -9.52668607e-01 -2.74391413e-01
1.29626155e-01 -9.76448417e-01 -3.76419336e-01 1.47074386e-01] | [7.415706157684326, 4.348930835723877] |
9a5c032e-bd9f-4076-afe3-80fd13190a61 | how-useful-are-educational-questions | 2304.06638 | null | https://arxiv.org/abs/2304.06638v1 | https://arxiv.org/pdf/2304.06638v1.pdf | How Useful are Educational Questions Generated by Large Language Models? | Controllable text generation (CTG) by large language models has a huge potential to transform education for teachers and students alike. Specifically, high quality and diverse question generation can dramatically reduce the load on teachers and improve the quality of their educational content. Recent work in this domain has made progress with generation, but fails to show that real teachers judge the generated questions as sufficiently useful for the classroom setting; or if instead the questions have errors and/or pedagogically unhelpful content. We conduct a human evaluation with teachers to assess the quality and usefulness of outputs from combining CTG and question taxonomies (Bloom's and a difficulty taxonomy). The results demonstrate that the questions generated are high quality and sufficiently useful, showing their promise for widespread use in the classroom setting. | ['Iulian Serban', 'Jackie C. K. Cheung', 'Ekaterina Kochmar', 'Sabina Elkins'] | 2023-04-13 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [-1.73728690e-02 5.13548851e-01 3.00751757e-02 -6.62195534e-02
-1.02376926e+00 -8.73639345e-01 5.10330081e-01 4.69541699e-01
-1.13604695e-01 6.95324838e-01 5.86499572e-01 -8.41648579e-01
-2.78943390e-01 -9.01893377e-01 -4.75569308e-01 -2.74544895e-01
6.49350524e-01 4.38079387e-01 4.40560609e-01 -4.55371499e-01
4.09053773e-01 2.72264183e-01 -2.10491681e+00 2.65419453e-01
1.42907381e+00 1.38131902e-01 3.61008674e-01 8.71178806e-01
-5.94694316e-01 1.22175717e+00 -1.29511511e+00 -3.43476027e-01
-1.87365010e-01 -7.03067541e-01 -9.09325302e-01 1.00885555e-02
9.80286121e-01 -5.70951521e-01 6.60053513e-04 8.28312457e-01
6.04794741e-01 3.49104196e-01 2.93086618e-01 -1.14236975e+00
-9.70169723e-01 6.44683719e-01 -1.00741126e-02 -2.53996905e-02
6.92975461e-01 2.66226739e-01 8.06058288e-01 -4.82841164e-01
5.00434816e-01 1.13335407e+00 3.09931666e-01 6.03987396e-01
-1.07495117e+00 -6.36394799e-01 -3.87306735e-02 -1.87250860e-02
-6.73344195e-01 -2.44824871e-01 2.82199115e-01 -6.34230793e-01
7.54321814e-01 3.92743826e-01 1.24992692e+00 7.27067292e-01
2.87082762e-01 5.82588434e-01 1.21133721e+00 -5.61234295e-01
2.52876371e-01 4.34633583e-01 5.65251932e-02 5.00277698e-01
5.81834197e-01 -2.46653289e-01 -5.68712711e-01 1.41709492e-01
5.20341575e-01 -3.55287671e-01 -3.51880699e-01 1.53145194e-01
-9.29792643e-01 8.87906432e-01 -1.05677001e-01 4.57989246e-01
-5.25692850e-02 9.32141021e-02 -9.29603800e-02 5.92056453e-01
4.27661628e-01 1.01878428e+00 -2.65168995e-01 -8.39520574e-01
-9.31459785e-01 8.40817571e-01 9.68075991e-01 1.15685511e+00
2.40023449e-01 2.63435125e-01 -4.62227315e-01 8.19877505e-01
1.94292262e-01 7.35703409e-01 6.42585635e-01 -1.29268539e+00
1.72933087e-01 8.82019341e-01 2.99438864e-01 -7.05988586e-01
-7.00782314e-02 -3.48453730e-01 5.27997725e-02 4.39821780e-01
7.70739615e-01 -4.10023689e-01 -7.13649690e-01 1.78135145e+00
3.79065096e-01 -2.97678798e-01 -1.49179533e-01 2.81213790e-01
1.39176583e+00 7.26102769e-01 2.32356146e-01 -2.99921464e-02
1.25388908e+00 -7.88738906e-01 -8.89235020e-01 -9.32097510e-02
9.80290532e-01 -1.35256672e+00 1.51446319e+00 5.00328183e-01
-1.64591014e+00 -6.45835638e-01 -6.99377537e-01 -2.43437454e-01
-1.97308674e-01 -2.98717290e-01 2.85135686e-01 1.09495342e+00
-1.29243362e+00 2.92153627e-01 -2.47540489e-01 -2.43976079e-02
2.35970959e-01 2.20240485e-02 3.73838879e-02 -2.36468777e-01
-8.85063231e-01 7.85994053e-01 -5.71562536e-03 -6.38093412e-01
-7.33429909e-01 -1.04583728e+00 -6.82369053e-01 4.16532874e-01
3.37528706e-01 -4.53437179e-01 1.93636775e+00 -5.05293667e-01
-1.65037978e+00 5.49311280e-01 1.54718652e-01 2.03595638e-01
3.31762522e-01 -3.17049623e-01 1.23415612e-01 1.85201481e-01
1.44148991e-01 7.53868520e-01 4.47973937e-01 -8.69786859e-01
-8.89057040e-01 -9.91417468e-02 4.20413166e-01 3.19429338e-01
-7.51225054e-01 -2.77338736e-02 3.15609068e-01 -5.90670526e-01
1.49408400e-01 -6.33345485e-01 -1.43512443e-01 -2.57956654e-01
9.05735120e-02 -8.32888782e-01 6.69868231e-01 -6.14390790e-01
1.22232330e+00 -1.66796875e+00 -5.32755971e-01 -9.24562216e-02
4.85187352e-01 4.24184322e-01 -1.88021600e-01 7.99243033e-01
1.10125206e-01 5.21562994e-01 6.61115706e-01 3.15577328e-01
2.58185357e-01 -6.99398667e-02 -1.55712008e-01 -2.60146260e-01
-2.26372391e-01 8.70970607e-01 -1.35418975e+00 -3.74634624e-01
1.30354673e-01 1.31741196e-01 -7.41928518e-01 6.39305651e-01
-4.84519720e-01 -4.03609984e-02 -3.94621611e-01 7.65859634e-02
1.56298324e-01 -2.27002889e-01 -1.45985141e-01 6.85146332e-01
-3.37427169e-01 9.52943444e-01 -9.05268371e-01 1.13366365e+00
-7.78442264e-01 9.90209699e-01 -1.64808258e-01 -2.32194647e-01
7.92607963e-01 4.25309300e-01 2.10482016e-01 -6.76248372e-01
-1.78128675e-01 1.09080330e-01 2.90492982e-01 -9.74479139e-01
9.31606591e-01 -1.33580655e-01 3.32914114e-01 1.05143225e+00
1.48983240e-01 -1.08478522e+00 6.65203333e-01 5.61488211e-01
1.03732455e+00 1.14560485e-01 -2.24453673e-01 -3.51667106e-01
-1.24255672e-01 8.35432038e-02 -1.83165357e-01 8.82298768e-01
5.65756299e-02 2.65474588e-01 4.80614692e-01 2.71514952e-01
-8.80051851e-01 -1.08615732e+00 2.62583464e-01 1.43895531e+00
-4.94588047e-01 -4.99249220e-01 -1.12321687e+00 -4.55731541e-01
-2.68055946e-01 1.26287878e+00 9.85684767e-02 -7.78854638e-02
-2.10733965e-01 1.26091957e-01 2.96988070e-01 4.38010573e-01
8.13076273e-02 -1.22146523e+00 -8.80808830e-01 3.85697782e-01
-4.35031086e-01 -8.74325871e-01 -4.67264712e-01 -3.15403551e-01
-7.92578876e-01 -8.37497771e-01 -5.40336251e-01 -9.11798775e-01
6.42522991e-01 6.93932176e-01 1.35248685e+00 3.17436606e-01
1.47431185e-02 9.32560742e-01 -5.20728111e-01 -7.15484083e-01
-9.19013739e-01 -4.20724005e-02 -3.96145493e-01 -9.99846756e-01
2.82479405e-01 -5.39846897e-01 -4.37562168e-01 2.71888226e-02
-1.03913867e+00 3.53539079e-01 2.42934749e-01 5.71388483e-01
-8.06890130e-02 1.39263138e-01 1.02548218e+00 -8.21229756e-01
1.29281271e+00 -1.86795294e-01 -8.17938328e-01 1.44450516e-01
-8.81678283e-01 -8.98058563e-02 5.93716264e-01 -5.02791047e-01
-1.08533084e+00 -6.96792424e-01 -4.91050422e-01 1.50074318e-01
-2.37008840e-01 3.94408733e-01 2.95768352e-03 -3.46962847e-02
7.53856122e-01 3.51384655e-02 -2.16127336e-01 -2.40375564e-01
4.01303828e-01 5.99936783e-01 3.52592021e-01 -8.72964680e-01
8.16003859e-01 -3.44216466e-01 -1.32289276e-01 -9.71345961e-01
-7.79069960e-01 -2.33094916e-01 7.36320242e-02 -5.48870146e-01
5.58619797e-01 -7.84401178e-01 -9.60229754e-01 2.39999563e-01
-9.16012704e-01 -6.87462270e-01 -8.20852578e-01 5.85657179e-01
-3.62916410e-01 1.62176281e-01 -6.47005677e-01 -7.44613886e-01
-3.12299491e-03 -1.29479730e+00 5.41528702e-01 9.43728209e-01
-4.29697514e-01 -1.08153236e+00 -1.00447662e-01 9.41918135e-01
5.73857009e-01 -5.25091030e-02 1.25010526e+00 -7.35029876e-01
-7.73103535e-01 -1.45842284e-01 2.22350687e-01 2.96265781e-01
7.70152882e-02 2.54542440e-01 -9.49006677e-01 5.08604608e-02
1.29818439e-01 -7.51816869e-01 1.11626387e-01 2.22228467e-01
8.79473865e-01 -6.67507350e-01 3.05715591e-01 -1.26941800e-01
1.10266614e+00 1.89761028e-01 4.17421907e-01 9.76055786e-02
3.15081120e-01 1.05756104e+00 4.06096786e-01 3.15556318e-01
4.94150430e-01 2.98544079e-01 1.62565969e-02 2.37346575e-01
-5.63163579e-01 -7.61247754e-01 4.40746963e-01 1.30078936e+00
2.99896032e-01 -5.37570715e-01 -9.29078698e-01 9.14678514e-01
-1.25425649e+00 -1.11078060e+00 -5.98728597e-01 2.18338108e+00
1.15906847e+00 2.41554290e-01 -1.35717019e-01 1.37583718e-01
2.05200300e-01 -2.19308794e-01 5.80631979e-02 -7.60654628e-01
3.25825632e-01 6.49954259e-01 -1.26399830e-01 6.38022184e-01
9.73593742e-02 7.86076248e-01 7.00640678e+00 9.23619449e-01
-7.04470277e-01 -9.73562226e-02 5.32262027e-01 -1.59178331e-01
-1.17668474e+00 -4.93858941e-02 -8.98536146e-01 3.27194691e-01
1.24601460e+00 -8.03420424e-01 2.27819666e-01 6.45172596e-01
5.25681496e-01 -2.82510787e-01 -9.29549694e-01 3.72653902e-01
-1.61919788e-01 -1.30347359e+00 8.69807005e-02 7.79541135e-02
1.19432569e+00 -5.36899984e-01 2.36750558e-01 4.28065389e-01
9.15020406e-01 -1.04259932e+00 6.61604583e-01 2.36616924e-01
5.44884264e-01 -6.89022243e-01 3.81383061e-01 5.20511031e-01
-6.56232715e-01 3.33049297e-02 -2.81541944e-01 -4.02135640e-01
-2.82666385e-01 3.48589897e-01 -1.41084003e+00 -3.42944637e-02
3.71989459e-01 -1.13200948e-01 -8.95811021e-01 1.16485643e+00
-7.33582795e-01 1.32329452e+00 -1.97967574e-01 -6.27962172e-01
2.31889322e-01 -5.06039523e-02 3.55088323e-01 8.71234179e-01
5.19996703e-01 3.90050948e-01 1.33285642e-01 7.55833030e-01
-1.68423504e-01 4.89994168e-01 -5.72661519e-01 -3.87137592e-01
7.19993651e-01 1.15196383e+00 -5.53765059e-01 -3.64343047e-01
-2.78758764e-01 1.16327383e-01 -3.00120637e-02 2.82228529e-01
-3.72557312e-01 -3.44551027e-01 4.06234026e-01 4.06227887e-01
-1.68481380e-01 -3.10808495e-02 -4.44379389e-01 -7.33453572e-01
-1.58605829e-01 -1.30144548e+00 -1.00074463e-01 -1.27532935e+00
-8.79772007e-01 8.63513499e-02 1.68210473e-02 -8.45931232e-01
-6.25422060e-01 -4.10896838e-01 -5.94542742e-01 9.56535101e-01
-1.05044401e+00 -7.55165339e-01 -5.11212587e-01 1.91624835e-01
7.78877556e-01 1.19851522e-01 5.78704715e-01 -3.85059975e-03
-2.20973957e-02 5.94236553e-01 1.20724693e-01 -4.06523645e-01
7.64688849e-01 -1.45424294e+00 2.36021399e-01 6.29132628e-01
1.51277453e-01 5.36249280e-01 7.87441790e-01 -6.01529479e-01
-1.45812786e+00 -7.51396120e-01 1.27245224e+00 -7.37055719e-01
4.73018765e-01 -7.68102407e-02 -9.23996866e-01 3.50346893e-01
5.78605890e-01 -8.38659465e-01 8.12197268e-01 -1.04263015e-02
-8.27650130e-02 1.30511254e-01 -1.07696116e+00 7.48939335e-01
7.34179020e-01 -2.91128993e-01 -8.68278861e-01 6.43442214e-01
1.04258859e+00 -4.15754825e-01 -1.06557333e+00 -1.56154782e-01
4.20136273e-01 -1.02024913e+00 6.13813937e-01 -6.58179820e-01
1.06242514e+00 2.33713090e-01 2.79737979e-01 -1.40472019e+00
-5.44931702e-02 -7.35137463e-01 3.09985697e-01 1.61467552e+00
3.65934879e-01 -4.85192955e-01 7.95748353e-01 9.93901908e-01
-5.46028674e-01 -7.29534209e-01 -3.21965605e-01 -5.67309022e-01
5.74779451e-01 -4.64547336e-01 8.18070650e-01 8.61089647e-01
3.07178497e-01 4.96997029e-01 3.24402720e-01 -3.64682823e-01
2.94353902e-01 1.45668864e-01 9.24831331e-01 -1.27749920e+00
-1.72853231e-01 -8.50982666e-01 -1.05232812e-01 -9.96305704e-01
-8.98855999e-02 -6.64785385e-01 1.99187741e-01 -1.95846736e+00
4.34400933e-03 -3.92400086e-01 5.16661525e-01 3.89573872e-01
-6.00769997e-01 -7.95446336e-02 4.14787829e-01 -2.89075226e-01
-4.20545012e-01 2.03137144e-01 1.88089705e+00 1.40593067e-01
-1.55203760e-01 1.56388938e-01 -1.10613251e+00 6.16947174e-01
7.21507788e-01 -2.83697039e-01 -9.03152943e-01 -3.74584407e-01
5.92388749e-01 3.10192049e-01 -1.41381353e-01 -9.47516084e-01
1.86589971e-01 -5.63843250e-01 2.68528700e-01 -3.53361726e-01
-1.79893360e-01 -5.41742027e-01 -2.58208901e-01 3.78684402e-01
-7.43718028e-01 2.52245486e-01 3.95723462e-01 -1.51980266e-01
-2.86441878e-03 -9.01431620e-01 5.26180685e-01 -8.55682120e-02
-4.55230959e-02 -1.79775968e-01 -8.73618066e-01 7.08275676e-01
7.17878222e-01 -2.74485946e-01 -5.86693525e-01 -1.05104434e+00
-2.68566340e-01 1.92929134e-01 3.78268838e-01 3.51293266e-01
6.09742105e-01 -1.00905418e+00 -8.17440927e-01 1.13249540e-01
-7.27532655e-02 -1.39289992e-02 -3.46366242e-02 1.96263850e-01
-5.39618731e-01 5.80642104e-01 -3.83098610e-02 -3.18745315e-01
-1.32935023e+00 7.70013928e-02 8.20280090e-02 -3.38908643e-01
-2.84138232e-01 9.58401442e-01 6.71825558e-02 -2.62509763e-01
2.27308452e-01 -6.99817181e-01 -2.43506685e-01 4.87661473e-02
7.95542836e-01 7.09642887e-01 5.78866340e-02 -4.22851299e-04
8.05947781e-01 1.06596343e-01 5.43596298e-02 -2.41370767e-01
1.06933093e+00 -2.49880292e-02 6.63903132e-02 3.04110467e-01
7.33112276e-01 5.35264611e-01 -8.26147377e-01 1.70199081e-01
-2.05762923e-01 -6.11967146e-01 -3.56959552e-02 -1.05223250e+00
-6.08441055e-01 9.95019972e-01 2.31573015e-01 7.63786018e-01
9.75968301e-01 -9.93539020e-02 7.80545354e-01 3.02744448e-01
5.21634147e-02 -9.14035499e-01 5.09296000e-01 4.54994828e-01
6.83220565e-01 -7.79479563e-01 -1.42991796e-01 -3.38083267e-01
-1.70310482e-01 1.16574752e+00 1.01199257e+00 4.73326266e-01
2.33760059e-01 1.52122468e-01 1.21824741e-01 -1.92050114e-01
-1.13250089e+00 -8.52327198e-02 1.91718757e-01 7.35992432e-01
1.05960679e+00 7.44321346e-02 -7.26417422e-01 1.40312493e-01
-9.41592097e-01 9.93319377e-02 1.41238773e+00 8.70462477e-01
-1.09693980e+00 -1.08305061e+00 -6.33073151e-01 7.63346851e-01
-6.18752420e-01 -1.30144507e-01 -5.00624061e-01 6.25533164e-01
2.53143013e-02 1.57461858e+00 -3.01380813e-01 -8.25272948e-02
2.99036026e-01 3.05120528e-01 8.08178484e-01 -1.13648915e+00
-1.16857517e+00 -1.02467164e-01 9.98369083e-02 -8.44059959e-02
1.47480637e-01 -3.56321990e-01 -1.17212725e+00 -5.54621875e-01
-6.86248064e-01 6.17504716e-01 6.80223405e-01 7.40908265e-01
9.80563089e-02 6.41616642e-01 3.75179559e-01 2.85070445e-02
-9.28240955e-01 -1.05950952e+00 -2.76805431e-01 2.50936687e-01
1.98631480e-01 -1.63116723e-01 -2.69621491e-01 1.07204616e-01] | [10.87680435180664, 7.707876682281494] |
71da6519-d9e6-4c7b-a13c-a7af24257117 | interpretable-3d-human-action-analysis-with | 1704.04516 | null | http://arxiv.org/abs/1704.04516v1 | http://arxiv.org/pdf/1704.04516v1.pdf | Interpretable 3D Human Action Analysis with Temporal Convolutional Networks | The discriminative power of modern deep learning models for 3D human action
recognition is growing ever so potent. In conjunction with the recent
resurgence of 3D human action representation with 3D skeletons, the quality and
the pace of recent progress have been significant. However, the inner workings
of state-of-the-art learning based methods in 3D human action recognition still
remain mostly black-box. In this work, we propose to use a new class of models
known as Temporal Convolutional Neural Networks (TCN) for 3D human action
recognition. Compared to popular LSTM-based Recurrent Neural Network models,
given interpretable input such as 3D skeletons, TCN provides us a way to
explicitly learn readily interpretable spatio-temporal representations for 3D
human action recognition. We provide our strategy in re-designing the TCN with
interpretability in mind and how such characteristics of the model is leveraged
to construct a powerful 3D activity recognition method. Through this work, we
wish to take a step towards a spatio-temporal model that is easier to
understand, explain and interpret. The resulting model, Res-TCN, achieves
state-of-the-art results on the largest 3D human action recognition dataset,
NTU-RGBD. | ['Austin Reiter', 'Tae Soo Kim'] | 2017-04-14 | null | null | null | null | ['action-analysis', 'multimodal-activity-recognition', '3d-human-action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.10611993e-01 1.79454803e-01 -3.56338710e-01 -3.15643966e-01
-2.18487233e-01 -1.60264209e-01 8.33296657e-01 -3.95041108e-01
-1.81360662e-01 2.53570020e-01 7.23217309e-01 -4.35345262e-01
-1.24952674e-01 -3.91771793e-01 -5.92362165e-01 -3.90191317e-01
-1.55511707e-01 3.51928174e-01 1.06623005e-02 -8.07951167e-02
8.53868127e-02 9.71201897e-01 -1.40122771e+00 5.30926585e-01
1.65433586e-01 1.12648380e+00 -3.83810997e-01 8.84557426e-01
-1.34229228e-01 1.48657870e+00 -4.61302370e-01 1.16686687e-01
1.92808032e-01 -7.25442350e-01 -1.23468482e+00 5.08097649e-01
2.23208696e-01 -5.23636222e-01 -7.12105215e-01 3.25206041e-01
2.74746060e-01 3.92802626e-01 7.63326645e-01 -1.13708615e+00
-7.53931224e-01 1.07198432e-01 -3.37851226e-01 3.75153273e-01
4.62043345e-01 3.31991971e-01 1.01515138e+00 -6.21089935e-01
6.77944064e-01 1.42817402e+00 7.22826123e-01 8.25323939e-01
-1.00435114e+00 3.09350807e-03 4.60777909e-01 3.67298126e-01
-7.60488272e-01 -4.12422478e-01 6.70908332e-01 -4.20435727e-01
1.81381631e+00 1.78608000e-01 1.28268218e+00 1.55656862e+00
3.26845944e-01 1.55649281e+00 8.84843826e-01 -5.44143319e-01
1.47217855e-01 -8.19043159e-01 -6.06418774e-03 7.00863361e-01
-2.76123106e-01 2.00409994e-01 -8.58099401e-01 3.17234397e-01
1.26502156e+00 4.38724697e-01 1.43653318e-01 -6.01991653e-01
-1.19893038e+00 4.67358053e-01 5.96057892e-01 4.50417310e-01
-5.18834174e-01 8.20302844e-01 5.12702942e-01 1.21523410e-01
8.03954720e-01 2.60458380e-01 -4.98171866e-01 -7.30282664e-01
-6.99614942e-01 2.11211890e-01 3.59048903e-01 5.17453551e-01
1.72963828e-01 4.08778578e-01 -1.51225448e-01 5.66294074e-01
3.30988228e-01 3.30026746e-01 6.03926539e-01 -1.07651758e+00
2.41919696e-01 8.78835917e-01 -6.37545213e-02 -7.41372347e-01
-5.55840373e-01 -1.89839572e-01 -7.46876776e-01 5.90830147e-01
3.69626254e-01 4.24149334e-01 -1.37121248e+00 1.36298776e+00
1.09766789e-01 4.99010831e-02 1.76673740e-01 9.09648061e-01
3.81840587e-01 4.07109797e-01 1.68985337e-01 4.68667924e-01
1.03885317e+00 -1.06313086e+00 -5.20987272e-01 -3.45251262e-01
1.02118349e+00 -1.00901663e-01 8.64501595e-01 2.45560572e-01
-9.95482087e-01 -6.12222195e-01 -1.00815868e+00 -3.40048254e-01
-4.43483114e-01 4.55685332e-02 8.90721440e-01 4.75568265e-01
-9.34809327e-01 8.44022572e-01 -1.57809913e+00 -7.50076532e-01
7.40339756e-01 2.31408909e-01 -6.70692682e-01 2.36938685e-01
-9.50742006e-01 1.38705134e+00 3.33040059e-01 3.70356679e-01
-9.57979679e-01 -1.76602453e-01 -9.74842668e-01 -3.87189031e-01
2.20832363e-01 -9.11344945e-01 1.55192971e+00 -1.14561963e+00
-1.62413406e+00 1.08822727e+00 -2.24143833e-01 -8.71790648e-01
5.05073845e-01 -5.07142901e-01 -1.71833262e-01 1.33801714e-01
-1.14169501e-01 6.36276245e-01 6.68161869e-01 -7.63863266e-01
-4.92151499e-01 -5.47292650e-01 1.97538495e-01 2.10434824e-01
3.93419899e-02 -4.16680202e-02 -1.50185302e-01 -7.21862316e-01
2.71937430e-01 -1.01918101e+00 -3.31559658e-01 4.05979514e-01
-1.95443124e-01 -3.44510198e-01 9.17112350e-01 -4.46761370e-01
9.32222724e-01 -1.88311553e+00 4.29865718e-01 -2.01781929e-01
2.12064028e-01 4.67286915e-01 -4.92295884e-02 3.13522816e-01
-4.13680136e-01 9.98457819e-02 -1.67781502e-01 -6.54327989e-01
2.75402904e-01 8.12892973e-01 -2.67655522e-01 5.30065536e-01
5.72395205e-01 1.39025223e+00 -9.78613436e-01 -2.37418100e-01
6.16376936e-01 6.63598537e-01 -2.83095360e-01 1.11276142e-01
-2.96720237e-01 5.85787833e-01 -7.91039586e-01 7.88095772e-01
-1.25205517e-01 -4.12610620e-01 -1.06380895e-01 8.24908316e-02
-1.24210209e-01 4.68739450e-01 -8.01735759e-01 2.16648626e+00
-3.29874516e-01 8.22723031e-01 -6.46610975e-01 -1.29838097e+00
8.22755218e-01 4.92539704e-01 7.74507880e-01 -9.18643534e-01
8.51251632e-02 7.54146725e-02 -2.76596576e-01 -6.56662345e-01
3.62587214e-01 -2.25347191e-01 2.02589668e-04 8.87310207e-01
1.13555819e-01 9.10906047e-02 -2.42998406e-01 -1.38167694e-01
1.42257309e+00 9.71028447e-01 5.28734088e-01 2.59743631e-01
2.69994169e-01 7.90442526e-02 2.66163498e-01 6.33931220e-01
-5.49322665e-01 6.12296700e-01 4.30652410e-01 -1.11837101e+00
-1.04276752e+00 -9.10585880e-01 2.53877521e-01 8.86313915e-01
-2.76768029e-01 -3.39101255e-01 -5.08477747e-01 -8.62388432e-01
-8.12794864e-02 6.04614139e-01 -1.02486801e+00 -2.28512615e-01
-8.37246299e-01 -1.48543924e-01 9.70014989e-01 1.17465496e+00
7.13130116e-01 -1.24266469e+00 -1.30273235e+00 2.35160127e-01
3.86671722e-02 -1.06089139e+00 -7.92170390e-02 3.99936736e-01
-1.34160304e+00 -9.67696428e-01 -7.56080270e-01 -2.76172072e-01
4.41232443e-01 2.17168212e-01 1.12980592e+00 -9.77958292e-02
-2.62879252e-01 7.62567639e-01 -7.18808591e-01 -4.64007497e-01
-3.05753559e-01 -1.27556130e-01 9.60928351e-02 -7.87057579e-02
5.20241141e-01 -6.92930758e-01 -4.54501003e-01 2.62305230e-01
-1.05600500e+00 3.40700150e-01 5.94708920e-01 8.04612517e-01
6.23227358e-01 -2.23172247e-01 3.50713916e-02 -4.28393424e-01
1.70115650e-01 -1.73644125e-01 1.11323982e-01 1.59759730e-01
-2.99302876e-01 4.22958612e-01 3.40889782e-01 -3.98059309e-01
-9.96855378e-01 2.44054377e-01 -2.28649274e-01 -7.23648429e-01
-5.04896343e-01 4.89982784e-01 1.14685275e-01 1.56800836e-01
7.15266347e-01 2.96349347e-01 1.07429475e-01 -7.04435647e-01
4.55606997e-01 4.24932122e-01 4.71484333e-01 -3.56546879e-01
4.26044285e-01 7.83966899e-01 2.69689679e-01 -7.71436930e-01
-1.01511919e+00 -2.54408121e-01 -1.20120502e+00 -4.31720048e-01
1.25290477e+00 -7.15146840e-01 -5.32920957e-01 8.48236203e-01
-1.32148981e+00 -7.74109185e-01 -6.06487155e-01 3.35770428e-01
-1.04630184e+00 2.93438077e-01 -6.19325936e-01 -1.08368361e+00
-7.93129019e-03 -8.80994797e-01 1.42013144e+00 -1.87949389e-01
-6.79722309e-01 -1.31856298e+00 1.11852929e-01 5.07164478e-01
2.40018085e-01 7.81113744e-01 7.60281205e-01 -5.09829283e-01
-5.88144600e-01 -3.11192960e-01 4.78191115e-02 4.09060001e-01
2.68977702e-01 -2.73123354e-01 -9.29033399e-01 2.59878367e-01
3.69901806e-02 -4.66080308e-01 8.74160051e-01 4.95937854e-01
1.24823058e+00 -3.98755595e-02 -2.03322217e-01 4.85606790e-01
7.07396984e-01 9.64223519e-02 8.62838626e-01 5.38901746e-01
8.02329659e-01 3.67219269e-01 4.83056933e-01 3.82409215e-01
2.91865528e-01 8.54312301e-01 5.19726455e-01 -3.02681655e-01
-3.46736282e-01 -5.91878355e-01 4.98251617e-01 4.65916932e-01
-6.61936224e-01 -2.06004322e-01 -1.19206357e+00 2.95870841e-01
-2.25587821e+00 -1.19504070e+00 1.42729163e-01 1.67463994e+00
3.63658309e-01 3.02646101e-01 2.33728528e-01 4.34443653e-01
6.27775788e-02 6.47136331e-01 -7.36862004e-01 -4.73589003e-01
-8.40011388e-02 2.62839586e-01 2.63488293e-01 2.01927513e-01
-1.22700942e+00 9.86582756e-01 6.61233139e+00 5.39479613e-01
-1.02425766e+00 -1.70487851e-01 5.26255906e-01 -1.49907798e-01
-2.91311368e-03 -8.00206065e-02 -5.03566027e-01 -9.05081034e-02
8.94143522e-01 2.92975992e-01 1.00874640e-01 8.19865823e-01
6.01041913e-01 -4.01079580e-02 -1.47907948e+00 1.11398602e+00
1.49487674e-01 -1.41766357e+00 2.49937832e-01 1.36765867e-01
3.81927252e-01 4.71144207e-02 -3.99867706e-02 2.68941164e-01
2.39050448e-01 -1.41605902e+00 9.40470636e-01 7.90010691e-01
4.66019392e-01 -3.54954541e-01 3.11106443e-01 3.80236596e-01
-1.21185899e+00 -8.72257501e-02 3.36216129e-02 -5.54899573e-01
4.39175308e-01 8.81576389e-02 -4.73028094e-01 4.61612374e-01
7.74228454e-01 1.55586731e+00 -4.72796053e-01 6.38300538e-01
-4.72170770e-01 2.96779931e-01 -1.54822439e-01 -3.20520215e-02
7.39540458e-01 1.45708248e-01 5.14887333e-01 9.81012583e-01
1.15752637e-01 3.27106655e-01 -6.35321736e-02 8.27479243e-01
1.51312709e-01 -5.70073068e-01 -1.01024234e+00 -5.21290421e-01
-3.93172443e-01 4.55320150e-01 -7.43182182e-01 -3.69830310e-01
-3.01650435e-01 1.41205263e+00 4.14634556e-01 4.81916547e-01
-8.34207773e-01 2.13473544e-01 9.34854805e-01 7.63052050e-03
3.50941837e-01 -7.89541066e-01 -3.70703220e-01 -1.14568162e+00
8.05820376e-02 -9.59265232e-01 4.05953944e-01 -1.09412313e+00
-9.80359018e-01 3.56227756e-01 1.02154374e-01 -1.32381666e+00
-5.45807183e-01 -1.05949926e+00 -3.91534984e-01 5.03995776e-01
-1.10911143e+00 -1.41721082e+00 -1.10984609e-01 5.44651806e-01
9.02012110e-01 4.17399555e-02 1.06455970e+00 -1.08633183e-01
-3.85884047e-01 1.27599701e-01 -3.21630001e-01 3.50542545e-01
9.38020498e-02 -1.12732661e+00 1.02477121e+00 7.62278795e-01
6.24236286e-01 4.74271178e-01 4.75570947e-01 -4.38723892e-01
-1.54294431e+00 -8.19851100e-01 9.79743838e-01 -1.11646414e+00
6.48186088e-01 -2.05001652e-01 -5.57819486e-01 1.10396326e+00
-1.20715417e-01 8.05985928e-02 6.10270739e-01 2.40546048e-01
-4.22860622e-01 1.52551800e-01 -6.85570836e-01 7.76795745e-01
1.63185644e+00 -9.24322128e-01 -9.75498199e-01 2.78659612e-01
4.99951184e-01 -5.66233456e-01 -7.10371971e-01 4.22155231e-01
9.15663302e-01 -1.15588236e+00 1.14565206e+00 -1.21564710e+00
5.75492084e-01 -2.31009081e-01 -4.43925932e-02 -1.10967231e+00
-2.05813304e-01 -4.80928391e-01 -5.38428128e-01 3.11414480e-01
1.61222354e-01 -3.64618778e-01 1.10291576e+00 6.93007588e-01
-4.95382845e-01 -9.22045946e-01 -1.03146195e+00 -8.91598821e-01
-1.17613614e-01 -9.22828436e-01 2.66972959e-01 5.85170984e-01
2.12878007e-02 1.47689015e-01 -5.99718392e-01 -3.76911938e-01
3.32885116e-01 1.26863997e-02 9.06428635e-01 -1.04208755e+00
-3.17089617e-01 -5.34381866e-01 -1.00379610e+00 -1.76021767e+00
2.91218877e-01 -6.62486732e-01 -1.12516798e-01 -1.85661793e+00
-1.18888341e-01 5.35383858e-02 -3.24011743e-01 1.03054941e+00
3.64630342e-01 1.33586615e-01 1.83221668e-01 2.10301727e-01
-7.62281060e-01 7.32971191e-01 1.35166073e+00 -2.56211817e-01
-1.61495790e-01 8.61855373e-02 -1.31842077e-01 8.24916005e-01
6.03372991e-01 -1.99551255e-01 -4.23822522e-01 -8.53364646e-01
-1.35389537e-01 1.12330336e-02 8.22558880e-01 -1.15627778e+00
1.68827504e-01 -9.46346819e-02 7.50233054e-01 -7.15365946e-01
7.37594247e-01 -7.53138542e-01 1.31370753e-01 5.71308970e-01
-5.00608325e-01 -5.68765309e-03 2.70766318e-01 6.79041326e-01
-2.95171052e-01 3.37565422e-01 3.76620114e-01 -4.22083586e-01
-1.16384935e+00 4.21664238e-01 -6.89172029e-01 -1.97394878e-01
8.08916330e-01 -9.99138892e-01 3.83121371e-02 -4.37391609e-01
-1.00752103e+00 -1.34253174e-01 3.28198254e-01 7.61013269e-01
7.06325769e-01 -1.52693892e+00 -2.76952207e-01 1.91028655e-01
1.30696326e-01 -2.11896330e-01 2.79012382e-01 8.23641837e-01
-4.59849894e-01 8.36707234e-01 -5.36080658e-01 -6.68540895e-01
-1.08249545e+00 1.84975490e-01 7.34389186e-01 -2.84004003e-01
-1.10056877e+00 7.10958600e-01 -1.16553567e-01 -3.37273449e-01
3.73128414e-01 -7.70514369e-01 -4.96702120e-02 -2.47824281e-01
3.77938360e-01 3.10712993e-01 -1.83426499e-01 -7.50004232e-01
-4.86704379e-01 4.51033652e-01 3.38063478e-01 -1.46539003e-01
1.52158809e+00 1.16740189e-01 2.86939144e-01 8.93407524e-01
1.09220743e+00 -8.96318257e-01 -1.66261673e+00 -1.03444755e-01
2.66891420e-01 -5.19854367e-01 1.87925436e-03 -6.85054660e-01
-9.66363668e-01 1.14921212e+00 5.17439127e-01 1.60618559e-01
1.00268340e+00 1.24687150e-01 9.31645691e-01 5.75664401e-01
3.65716130e-01 -1.01825452e+00 5.47697306e-01 8.63721728e-01
8.77255023e-01 -1.17943621e+00 1.30966201e-01 2.79697061e-01
-6.11900270e-01 1.31106448e+00 3.37501615e-01 -1.49202356e-02
5.02886057e-01 -1.08775884e-01 7.78710991e-02 -6.08009577e-01
-6.73649669e-01 -3.89023632e-01 3.68571341e-01 6.50283992e-01
4.73287910e-01 -6.14531785e-02 3.25197697e-01 2.94253916e-01
1.02911234e-01 1.90527529e-01 4.14524190e-02 1.30752361e+00
-3.25931698e-01 -1.13338637e+00 2.05078553e-02 6.81218952e-02
-2.43666947e-01 2.67107785e-01 -7.18540192e-01 9.53752160e-01
-1.23728767e-01 6.28731966e-01 -1.26461778e-02 -4.49346930e-01
6.76728606e-01 3.43951821e-01 7.02664733e-01 -4.20652211e-01
-3.64094019e-01 -3.53278458e-01 2.64083326e-01 -1.11105728e+00
-1.01950455e+00 -6.18149757e-01 -1.32765007e+00 -1.88952684e-01
2.03269213e-01 -4.56185579e-01 4.60570931e-01 1.42553914e+00
4.03490752e-01 4.79399800e-01 1.03809729e-01 -1.30404806e+00
-3.13591182e-01 -8.28906536e-01 -5.56129217e-01 3.63070905e-01
4.40814614e-01 -8.93805027e-01 -1.66203603e-01 2.76448607e-01] | [7.954879283905029, 0.4834100306034088] |
c65dffdc-e85d-4cf5-bfc7-4d274f3cf076 | methodological-aspects-of-developing-and | null | null | https://aclanthology.org/2020.lrec-1.392 | https://aclanthology.org/2020.lrec-1.392.pdf | Methodological Aspects of Developing and Managing an Etymological Lexical Resource: Introducing EtymDB-2.0 | Diachronic lexical information is not only important in the field of historical linguistics, but is also increasingly used in NLP, most recently for machine translation of low resource languages. Therefore, there is a need for fine-grained, large-coverage and accurate etymological lexical resources. In this paper, we propose a set of guidelines to generate such resources, for each step of the life-cycle of an etymological lexicon: creation, update, evaluation, dissemination, and exploitation. To illustrate the guidelines, we introduce EtymDB 2.0, an etymological database automatically generated from the Wiktionary, which contains 1.8 million lexemes, linked by more than 700,000 fine-grained etymological relations, across 2,536 living and dead languages. We also introduce use cases for which EtymDB 2.0 could represent a key resource, such as phylogenetic tree generation, low resource machine translation or medieval languages study. | ['Beno{\\^\\i}t Sagot', "Cl{\\'e}mentine Fourrier"] | 2020-05-01 | null | null | null | lrec-2020-5 | ['cognate-prediction'] | ['natural-language-processing'] | [ 1.71146184e-01 1.11419909e-01 -5.52113891e-01 -2.41523888e-02
-2.63206601e-01 -9.84471023e-01 7.66005039e-01 6.70709968e-01
-6.91910923e-01 1.42614436e+00 5.20559967e-01 -4.53735501e-01
-6.21612146e-02 -7.95648336e-01 -2.65563041e-01 -1.15708739e-01
3.78972381e-01 1.11605251e+00 -2.01934204e-02 -4.87119049e-01
5.59601009e-01 3.56955916e-01 -1.24508047e+00 -2.20918804e-02
8.98505747e-01 3.49743038e-01 2.66464740e-01 7.01557398e-02
-5.89641869e-01 1.54235885e-01 -6.48004591e-01 -9.37803030e-01
-2.06163719e-01 -6.14685059e-01 -1.12923908e+00 -5.97057879e-01
5.25615294e-04 2.23561347e-01 2.39135921e-01 1.02427518e+00
5.75862765e-01 -2.40603656e-01 6.83724284e-01 -7.12883890e-01
-6.15360260e-01 1.31618512e+00 -3.03513288e-01 2.88490087e-01
4.23197448e-01 -2.38459319e-01 1.34379303e+00 -7.10555553e-01
1.49199653e+00 1.07146049e+00 3.39344978e-01 5.42932928e-01
-8.84966671e-01 -6.40277863e-01 -3.54480892e-01 5.39128959e-01
-1.33085692e+00 -4.86665964e-01 5.22827446e-01 -4.62426990e-01
1.18770540e+00 1.67541847e-01 9.26752746e-01 1.07604909e+00
4.54296410e-01 1.66268408e-01 1.03102100e+00 -7.49660313e-01
4.96540032e-02 4.41377833e-02 -1.54069871e-01 3.28553945e-01
6.53699577e-01 -1.00644268e-01 -1.04623079e+00 -5.09347975e-01
5.99646330e-01 -7.79641926e-01 -3.86866555e-02 3.55590791e-01
-1.36830986e+00 6.31289661e-01 -1.21297330e-01 7.49901235e-01
-2.78246343e-01 -3.12063485e-01 8.05033565e-01 3.16114306e-01
5.49307466e-01 6.38316691e-01 -6.27935350e-01 -4.32807833e-01
-7.40307510e-01 5.97304814e-02 8.19052935e-01 1.03353763e+00
5.44472933e-01 -2.21850991e-01 4.36461091e-01 1.06261468e+00
2.11316705e-01 3.79608005e-01 8.84705961e-01 -3.17305744e-01
1.85485229e-01 6.49229646e-01 -2.57002473e-01 -6.77181304e-01
-6.18561029e-01 9.98440683e-02 -4.77020264e-01 -6.64672375e-01
9.12830383e-02 1.67796001e-01 -4.69211131e-01 1.80003166e+00
6.70234382e-01 -3.07276040e-01 2.50052869e-01 5.22466898e-01
1.06412375e+00 5.43058932e-01 2.51981109e-01 -6.84922397e-01
1.63414991e+00 -3.73579025e-01 -7.49453425e-01 -1.16407193e-01
4.99955475e-01 -1.14642799e+00 1.04708362e+00 7.96855167e-02
-1.27089238e+00 -8.66632611e-02 -8.46022427e-01 -3.23267400e-01
-6.65446818e-01 -1.47599712e-01 6.37153924e-01 5.58837175e-01
-7.57008612e-01 5.36555827e-01 -5.81765652e-01 -8.81997168e-01
-2.47882515e-01 3.02551743e-02 -4.22028363e-01 4.11703318e-01
-1.58004224e+00 1.53645718e+00 8.84397447e-01 -9.20812637e-02
-1.79932803e-01 -5.63802063e-01 -4.63410795e-01 -4.31523234e-01
2.40566105e-01 -5.31953871e-01 9.72161472e-01 -3.72947425e-01
-1.33089459e+00 1.68383133e+00 2.49590371e-02 -2.32480958e-01
2.04306871e-01 -8.37653726e-02 -7.56173313e-01 -7.18801618e-02
4.71773267e-01 4.55067784e-01 2.11701304e-01 -4.42063898e-01
-6.51978672e-01 -3.47019613e-01 -3.42380017e-01 1.59842566e-01
-3.19880605e-01 6.06075943e-01 -2.74478287e-01 -7.83677995e-01
1.45112835e-02 -7.06735611e-01 5.72425313e-02 -2.46867731e-01
-2.90758252e-01 -3.75155121e-01 1.15411483e-01 -9.87837911e-01
1.49187624e+00 -1.63714278e+00 3.74847800e-01 1.90120026e-01
-4.15924489e-02 1.13055827e-02 1.57887399e-01 8.51775944e-01
1.74658507e-01 4.18906063e-01 -3.89050752e-01 3.50026011e-01
-8.37429464e-02 4.28266823e-01 -1.46049321e-01 4.05674070e-01
-4.08928059e-02 9.60691571e-01 -1.19663227e+00 -7.00543344e-01
5.35003096e-02 2.34997377e-01 -1.25810534e-01 -3.12944621e-01
-2.38905713e-01 4.47111189e-01 -4.55526829e-01 7.76482880e-01
2.25933746e-01 1.99795365e-01 7.36687720e-01 1.23649083e-01
-5.19827425e-01 1.00702345e+00 -6.13142371e-01 1.83941829e+00
-5.70330381e-01 5.54815471e-01 -2.55848080e-01 -2.68158644e-01
1.07538688e+00 3.66075516e-01 1.97345823e-01 -6.63509190e-01
3.64793062e-01 8.13379109e-01 2.85357714e-01 -1.70224592e-01
1.00242674e+00 -4.15775478e-01 -3.67232621e-01 4.82450545e-01
2.59122193e-01 -3.28763604e-01 6.97386384e-01 1.13993064e-02
7.02992320e-01 3.86901021e-01 1.19755554e+00 -6.60400689e-01
6.62602484e-01 3.96945775e-01 8.71920466e-01 2.23512128e-02
6.92660809e-02 9.73306298e-02 3.43553960e-01 -5.70650220e-01
-1.37279570e+00 -7.50716925e-01 -8.49551737e-01 7.20513403e-01
-8.17548931e-02 -8.80822778e-01 -5.50620556e-01 -1.06141865e-01
-3.83309990e-01 9.10195947e-01 -4.28959131e-01 -2.54107416e-01
-7.73819029e-01 -1.10611641e+00 1.02175748e+00 6.52465150e-02
2.16580987e-01 -1.50122142e+00 -7.36679912e-01 5.82590103e-01
-6.69873595e-01 -1.01949704e+00 -2.04830304e-01 -3.93624865e-02
-5.22340238e-01 -9.68789697e-01 -4.16345417e-01 -8.02397430e-01
1.81651041e-01 -4.06196773e-01 1.28912187e+00 1.35451939e-03
-2.80801713e-01 -4.47419584e-01 -6.26968145e-01 -4.96167272e-01
-8.92232001e-01 5.84650159e-01 4.07505661e-01 -7.17298806e-01
6.70945227e-01 -6.75187767e-01 1.48406580e-01 -6.34256098e-03
-8.58142793e-01 5.47153354e-02 1.79318622e-01 7.54100561e-01
6.55894518e-01 -5.04724145e-01 9.03499603e-01 -1.11533952e+00
7.12717950e-01 -5.93105853e-01 -6.88466191e-01 5.34552932e-01
-6.69236779e-01 1.08193710e-01 6.51568234e-01 -3.09510618e-01
-8.98800731e-01 -4.63518143e-01 -7.86390901e-01 6.79756343e-01
1.23697862e-01 1.01957774e+00 -2.61293858e-01 2.42662355e-01
8.65341485e-01 1.17023870e-01 -4.20884639e-01 -6.20533347e-01
6.70509696e-01 8.90707195e-01 5.84883809e-01 -9.57433581e-01
5.04584372e-01 4.92993742e-02 -3.35238241e-02 -9.96254623e-01
-3.70726079e-01 -3.62432510e-01 -1.09482992e+00 -9.34640225e-03
6.81803167e-01 -6.25866234e-01 -4.42451894e-01 1.22071542e-01
-1.20114899e+00 -1.20329961e-01 -3.38087142e-01 4.48862046e-01
-4.25085872e-01 1.21426992e-01 -6.27414703e-01 -2.15283424e-01
-6.84476256e-01 -8.05862069e-01 7.66808569e-01 8.88818800e-02
-8.10214162e-01 -1.09687865e+00 6.28662169e-01 6.09469190e-02
-1.26876663e-02 3.94526124e-01 1.44530666e+00 -6.74262702e-01
1.10343970e-01 2.67798364e-01 1.88682631e-01 -3.30145299e-01
3.64911258e-01 2.54915774e-01 -4.46401358e-01 2.21052274e-01
-2.22454354e-01 -3.62064838e-01 3.90926927e-01 -1.55259997e-01
1.90048024e-01 -3.51532429e-01 -1.38018161e-01 4.88579363e-01
1.18779111e+00 1.72955006e-01 5.76608241e-01 8.20863783e-01
5.62709987e-01 8.20831656e-01 6.65040314e-01 2.95875013e-01
4.95450288e-01 7.16037333e-01 -1.95921630e-01 4.10003304e-01
-1.16766945e-01 -3.32117468e-01 3.13324809e-01 1.43154299e+00
-3.66853118e-01 -1.32748494e-02 -1.37218869e+00 7.85356522e-01
-1.52115154e+00 -7.27869689e-01 -2.49234274e-01 2.15849209e+00
1.49715447e+00 -8.91871601e-02 1.25684917e-01 -7.37143606e-02
6.93182886e-01 -3.70034501e-02 -4.46594328e-01 -6.62017643e-01
-4.56014276e-01 5.94412327e-01 4.04716194e-01 5.63573778e-01
-2.66712636e-01 1.80727041e+00 6.39492893e+00 9.53066647e-01
-1.00850415e+00 9.15960371e-02 1.38124660e-01 4.43550460e-02
-5.62165022e-01 2.93989807e-01 -1.01214755e+00 2.61936724e-01
1.17693758e+00 -8.56416643e-01 2.10445955e-01 5.91927767e-01
6.82183653e-02 1.23317195e-02 -8.62591922e-01 6.67293608e-01
8.31290558e-02 -1.69083607e+00 1.18920639e-01 9.77646336e-02
4.48011458e-01 2.82024294e-01 -6.58724308e-01 -6.43686056e-02
3.14931482e-01 -7.54321992e-01 9.70419705e-01 -4.74806614e-02
1.43338668e+00 -8.74453604e-01 6.30536020e-01 1.51131272e-01
-1.14155865e+00 6.72667921e-01 -8.01518738e-01 5.26853390e-02
8.03041339e-01 6.72304451e-01 -7.55952597e-01 7.80305803e-01
4.24656540e-01 6.62341774e-01 -5.38845658e-01 7.39257038e-01
-8.38067770e-01 6.91606224e-01 -4.27897930e-01 -4.77782488e-01
7.79720023e-02 -3.97287905e-01 6.46444976e-01 1.31888723e+00
2.62178570e-01 9.48048085e-02 -6.49136379e-02 5.46765506e-01
-8.03240016e-02 9.58425879e-01 -3.92097592e-01 -3.99078131e-01
9.30173635e-01 1.34148681e+00 -1.23250759e+00 -4.57082957e-01
-1.12996586e-01 1.06133223e+00 4.58431602e-01 -2.90570050e-01
-5.41361928e-01 -3.08217138e-01 6.51103735e-01 6.13835305e-02
-4.10691798e-01 -2.29101062e-01 -3.52890253e-01 -1.15450227e+00
-1.75681144e-01 -1.04834509e+00 4.19452161e-01 -4.72186357e-01
-1.37285137e+00 1.00032353e+00 1.23589672e-01 -7.62474954e-01
-6.18322611e-01 -6.21400714e-01 -2.54530102e-01 9.66157138e-01
-1.02742386e+00 -1.30505431e+00 3.42685103e-01 1.64852798e-01
3.13926429e-01 -1.95779443e-01 1.01450658e+00 3.10014278e-01
-4.60000336e-01 3.61180454e-01 1.26074880e-01 -2.05938444e-01
9.08365965e-01 -9.88422275e-01 9.43644643e-01 4.74681675e-01
1.98028684e-01 7.33136117e-01 6.09019279e-01 -1.13645887e+00
-7.23086536e-01 -8.33072901e-01 1.75215983e+00 -4.35930878e-01
1.23053849e+00 -5.77001274e-01 -7.77592480e-01 5.93365550e-01
2.08812922e-01 -9.54588592e-01 1.06247270e+00 2.78473854e-01
-1.59217551e-01 3.93013090e-01 -1.15495694e+00 9.79381621e-01
1.38693035e+00 -6.45504653e-01 -8.67796600e-01 4.93486345e-01
7.59246290e-01 -3.19469273e-01 -1.20366108e+00 1.70730010e-01
7.08185852e-01 -2.96456754e-01 6.15266323e-01 -5.48599303e-01
4.56803828e-01 -2.66722649e-01 8.77666548e-02 -1.26583350e+00
-2.83733279e-01 -8.08267891e-01 2.54783660e-01 1.44052887e+00
8.35777700e-01 -6.72250271e-01 2.83855975e-01 4.88556027e-02
-2.03788340e-01 -3.45717221e-01 -1.12636018e+00 -5.56795418e-01
4.38106358e-01 -1.98400542e-01 8.85424733e-01 1.43007040e+00
5.41084111e-01 9.60544884e-01 -3.60823393e-01 -5.21334231e-01
2.34443545e-01 -1.33852333e-01 3.17161292e-01 -1.40861106e+00
-8.37604329e-02 -5.67808270e-01 -4.91727740e-01 -3.18040937e-01
3.69069427e-01 -1.32756650e+00 -1.42599791e-01 -1.52575576e+00
2.12101378e-02 -4.19293821e-01 2.42173776e-01 5.38794458e-01
-1.62328735e-01 3.64442915e-01 -6.62555471e-02 5.47180474e-01
2.15129167e-01 3.97139877e-01 7.92016506e-01 5.44129550e-01
-3.72267216e-01 -5.98397136e-01 -6.36444926e-01 7.61848152e-01
9.78825331e-01 -7.27081537e-01 -2.58762967e-02 -2.29882374e-01
1.02846396e+00 -3.58892083e-01 -3.78189176e-01 -4.18990791e-01
3.05353217e-02 -6.01876974e-01 -1.56316683e-02 -5.40768504e-01
-1.07706510e-01 -2.30437815e-01 4.36764061e-01 6.84919953e-01
-3.05962488e-02 5.45743704e-01 2.38495424e-01 -2.54564136e-01
-1.36876509e-01 -7.74242818e-01 5.63853025e-01 -3.91857058e-01
-7.81851709e-01 5.23253158e-02 -5.20437479e-01 3.18653733e-01
7.43435681e-01 -1.37340814e-01 -5.81908286e-01 3.44035953e-01
-3.95219654e-01 1.90861039e-02 7.36694634e-01 4.74613219e-01
2.99342155e-01 -1.15813923e+00 -1.02087653e+00 -6.25645816e-02
4.46643084e-01 -4.24154967e-01 -1.05107881e-01 4.23645705e-01
-9.50484991e-01 2.85857797e-01 -7.26578772e-01 -2.29537971e-02
-9.59691644e-01 4.66426402e-01 -2.86009371e-01 -2.38151625e-01
-6.32650912e-01 4.36175257e-01 -2.11350501e-01 -7.71741450e-01
-4.14684653e-01 1.32917076e-01 -5.38935125e-01 4.28796113e-01
5.46023011e-01 3.42751980e-01 1.82941839e-01 -1.24601626e+00
-5.99166274e-01 6.22453451e-01 -2.91752648e-02 -6.52984917e-01
1.30199909e+00 -2.65609175e-01 -8.38737905e-01 1.00160193e+00
4.88001823e-01 5.20087719e-01 -1.26750201e-01 -6.28803894e-02
5.97023070e-01 -4.69289608e-02 -4.66026038e-01 -9.11085904e-01
-3.30921113e-01 4.52359617e-01 -1.46629736e-01 -2.74374962e-01
8.01143885e-01 1.47098109e-01 7.96789110e-01 2.09781662e-01
5.65837026e-01 -1.28234601e+00 -6.18691623e-01 9.30314004e-01
9.26448286e-01 -6.31779909e-01 4.06730105e-04 -4.67435718e-01
-4.65646029e-01 1.07822955e+00 3.11511785e-01 3.26274782e-01
4.35861945e-01 2.48577535e-01 2.23982081e-01 -3.24035257e-01
-7.26017296e-01 -1.81008235e-01 1.57114819e-01 5.70011258e-01
1.04473722e+00 1.59945056e-01 -1.65524924e+00 4.04475600e-01
-1.06233346e+00 -3.08792144e-01 5.69917977e-01 8.04699779e-01
-3.15657675e-01 -1.89042068e+00 -1.55023649e-01 2.52159268e-01
-5.49799919e-01 -7.83346951e-01 -1.04355884e+00 7.45445192e-01
2.73070842e-01 5.41362822e-01 9.16676149e-02 -2.88660556e-01
5.97946942e-02 1.38002470e-01 5.53241134e-01 -7.83830941e-01
-8.11994255e-01 1.47397757e-01 6.97238803e-01 1.71956897e-01
-5.07048607e-01 -6.80030704e-01 -1.38235056e+00 -6.58290923e-01
-4.23132330e-01 5.01052022e-01 8.09895337e-01 9.54671502e-01
4.44901846e-02 2.11168468e-01 -5.29526360e-02 -4.14522707e-01
9.36860293e-02 -1.06313074e+00 -4.54578727e-01 -3.05939876e-02
-6.51968420e-01 -5.14415503e-01 1.39394403e-02 1.10980585e-01] | [10.322925567626953, 10.120329856872559] |
6baefd82-2066-4ccb-9be8-772551dffbdb | quantifying-context-overlap-for-training-word | null | null | https://aclanthology.org/D18-1057 | https://aclanthology.org/D18-1057.pdf | Quantifying Context Overlap for Training Word Embeddings | Most models for learning word embeddings are trained based on the context information of words, more precisely first order co-occurrence relations. In this paper, a metric is designed to estimate second order co-occurrence relations based on context overlap. The estimated values are further used as the augmented data to enhance the learning of word embeddings by joint training with existing neural word embedding models. Experimental results show that better word vectors can be obtained for word similarity tasks and some downstream NLP tasks by the enhanced approach. | ['Yinhe Zheng', 'Yimeng Zhuang', 'Xuan Zhu', 'Jinghui Xie'] | 2018-10-01 | null | null | null | emnlp-2018-10 | ['learning-word-embeddings'] | ['methodology'] | [-2.02729642e-01 6.54368028e-02 -5.77330351e-01 -4.77008909e-01
-1.89380348e-01 -1.79940924e-01 6.73664391e-01 6.78649843e-01
-1.01347077e+00 4.52453107e-01 8.80345583e-01 -3.40850741e-01
-1.91731751e-01 -7.78989553e-01 -1.87658670e-03 -4.85857010e-01
-2.64528453e-01 2.74939984e-01 1.06888317e-01 -4.22901005e-01
4.59638983e-01 2.54683256e-01 -1.21866035e+00 1.80286020e-01
7.19123065e-01 6.76503897e-01 2.88186371e-01 2.90495545e-01
-7.61842906e-01 1.25589013e-01 -3.60558629e-01 -3.04847181e-01
-9.17781293e-02 -2.02913687e-01 -5.37150562e-01 -6.18126512e-01
-3.66697699e-01 -4.12105694e-02 -3.98019284e-01 9.49405491e-01
4.10818338e-01 5.95756829e-01 7.60336220e-01 -9.67482448e-01
-1.22924078e+00 7.63343036e-01 -1.51636496e-01 5.69081724e-01
2.73763508e-01 -2.21898228e-01 1.78008378e+00 -1.46423388e+00
2.66730219e-01 1.22535527e+00 4.44524646e-01 1.83704302e-01
-8.50114286e-01 -6.40517890e-01 3.59196514e-01 7.72494674e-01
-1.66281784e+00 1.41124010e-01 7.28926182e-01 -6.58481196e-02
1.81957948e+00 -1.25034422e-01 5.90357900e-01 8.18919539e-01
2.62441188e-01 2.82035708e-01 4.61583555e-01 -7.43891478e-01
-1.44587621e-01 1.67549253e-01 5.87026536e-01 4.05560762e-01
3.97144914e-01 9.64243636e-02 -3.87988597e-01 -1.62851289e-01
4.99955833e-01 2.36968443e-01 -1.63401350e-01 8.94806162e-02
-9.99561787e-01 1.28794360e+00 6.22609317e-01 1.02648401e+00
-2.94231653e-01 3.25346440e-02 5.22040129e-01 9.06011015e-02
8.39258969e-01 7.67284751e-01 -6.47324979e-01 -1.17305927e-01
-2.99332321e-01 -5.14536873e-02 3.92693818e-01 7.35323310e-01
9.70011532e-01 -1.89305499e-01 -2.60603935e-01 1.24851918e+00
7.07007468e-01 -2.41830829e-03 1.14474070e+00 3.42065059e-02
3.92957926e-01 8.39558423e-01 -3.19956213e-01 -1.15133607e+00
-2.90766209e-01 -1.94918364e-01 -4.40418750e-01 -5.10732055e-01
-3.38835567e-01 -5.63231409e-02 -9.24383104e-01 1.58498621e+00
2.88077056e-01 6.64968967e-01 2.82273561e-01 6.98114157e-01
9.16881382e-01 1.04701507e+00 4.26624030e-01 -2.26790011e-01
1.54642940e+00 -1.02806139e+00 -1.26600420e+00 -4.32068557e-01
1.09021139e+00 -7.65521526e-01 1.18861699e+00 -1.65593073e-01
-3.38120908e-01 -7.35219717e-01 -1.06242168e+00 -2.26503581e-01
-1.17176735e+00 -1.82409063e-01 6.93213224e-01 3.78629118e-01
-6.68403685e-01 5.01734257e-01 -4.30900961e-01 -3.66316766e-01
3.17262322e-01 2.03043997e-01 -3.95658404e-01 -3.22386064e-02
-1.88212788e+00 1.38084173e+00 1.04231715e+00 -5.64420223e-02
-6.17539883e-02 -8.39995265e-01 -1.25218177e+00 2.31638521e-01
-2.55109161e-01 -1.20384626e-01 6.30717635e-01 -2.14654803e-01
-9.86140907e-01 6.35624766e-01 -3.35642964e-01 -2.53379345e-01
-5.86010277e-01 -4.21080232e-01 -8.04696679e-01 -1.99724481e-01
-1.27934873e-01 4.18527752e-01 1.96282536e-01 -8.66972208e-01
-6.34125888e-01 -2.33517155e-01 -4.92390171e-02 4.04189080e-01
-1.21721029e+00 8.98410827e-02 -2.88891256e-01 -8.42789412e-01
-1.04971349e-01 -5.91725707e-01 -3.47329706e-01 -1.31074250e-01
7.57472813e-02 -1.13437688e+00 8.44751716e-01 -5.56125164e-01
1.80648196e+00 -2.10216427e+00 5.53349257e-02 3.14569354e-01
3.87336463e-02 8.58972847e-01 -6.05289102e-01 7.74034321e-01
-3.12922269e-01 2.03680843e-01 -4.36216928e-02 -1.33622587e-01
-4.01332835e-03 5.73677421e-01 -4.00877595e-01 6.30487734e-03
5.31677186e-01 9.97133791e-01 -1.12689757e+00 -4.62170213e-01
1.75508857e-01 5.75205922e-01 -3.52606118e-01 5.49235702e-01
1.95175782e-01 -2.97302276e-01 -4.26301360e-01 6.13209419e-03
2.77595967e-01 1.40105769e-01 4.09924924e-01 -3.12391698e-01
2.18582436e-01 6.73254550e-01 -6.96803510e-01 1.62813938e+00
-9.76345897e-01 6.06556535e-01 -8.85532856e-01 -1.27688420e+00
1.23258054e+00 4.53088045e-01 1.99578062e-01 -5.01507282e-01
2.68986136e-01 6.50501018e-03 4.20843571e-01 -7.82105386e-01
6.60572886e-01 -2.23596141e-01 7.99613371e-02 3.19262147e-01
4.17612851e-01 1.22011229e-02 1.37695268e-01 1.85502529e-01
9.76055443e-01 -2.66294330e-01 7.03279972e-01 -2.92654902e-01
5.86920321e-01 -3.88605088e-01 4.05539602e-01 -2.55851559e-02
-1.29989102e-01 2.31514469e-01 2.43474156e-01 -3.22310746e-01
-9.50338900e-01 -1.04339075e+00 -3.57146382e-01 1.16857958e+00
5.29732145e-02 -9.71705794e-01 -1.96017146e-01 -9.19989467e-01
6.67522550e-02 9.10892010e-01 -7.93155372e-01 -7.10293412e-01
-5.84125698e-01 -6.88182235e-01 3.21309209e-01 1.00311625e+00
-2.25661054e-01 -1.04619014e+00 1.43758968e-01 4.01837498e-01
1.66000828e-01 -1.03203952e+00 -6.95983291e-01 3.39311093e-01
-8.23181450e-01 -7.24900842e-01 -5.10490894e-01 -1.22323585e+00
6.16885304e-01 1.86315328e-01 7.45022535e-01 2.59621859e-01
-2.02114627e-01 2.63161995e-02 -9.36430037e-01 -3.17898422e-01
1.28012404e-01 7.20997974e-02 3.58639359e-01 -2.35731885e-01
1.09200132e+00 -6.63193822e-01 -2.78027236e-01 -1.05757654e-01
-9.97898459e-01 -5.82812488e-01 5.15013754e-01 1.12002337e+00
3.48306119e-01 -1.51667893e-01 7.69184113e-01 -7.07256496e-01
1.24981809e+00 -6.77351058e-01 3.85750495e-02 3.08025748e-01
-1.08087885e+00 4.63184655e-01 4.81241524e-01 -9.50105131e-01
-9.62926924e-01 -5.02176821e-01 -3.80395710e-01 -2.22069353e-01
5.88596426e-02 9.67883885e-01 -5.43945394e-02 2.10794494e-01
3.12473953e-01 9.39247850e-03 -4.42122877e-01 -5.39165318e-01
8.43993366e-01 8.63547087e-01 -1.80856213e-01 -4.49541867e-01
5.53354263e-01 -2.02341154e-01 -3.09359640e-01 -8.00739348e-01
-7.65419006e-01 -8.01819146e-01 -6.78041995e-01 2.63225675e-01
9.53885317e-01 -6.36061430e-01 -8.58015046e-02 -1.37391627e-01
-1.53872633e+00 3.35509658e-01 -7.96282887e-02 9.79051948e-01
1.48476645e-01 3.72697830e-01 -5.10876417e-01 -6.73251390e-01
-3.42292935e-01 -6.95542157e-01 6.62996233e-01 2.62126148e-01
-4.03652281e-01 -1.55365586e+00 5.91398656e-01 -1.65297300e-01
5.09162724e-01 -4.74421322e-01 1.37362790e+00 -1.29578960e+00
3.00631195e-01 -4.21160549e-01 -3.37226480e-01 5.38466036e-01
5.24743080e-01 -1.04570009e-01 -6.87561929e-01 1.46798268e-01
-3.78355265e-01 -4.59547807e-03 9.85482812e-01 -5.27559838e-04
1.02901554e+00 -2.59478927e-01 -6.79819465e-01 2.63573140e-01
1.38419402e+00 3.53503138e-01 6.27824187e-01 1.36206830e-02
8.89760196e-01 7.77626812e-01 8.12850177e-01 1.95230976e-01
2.80964822e-01 5.64898670e-01 6.36792332e-02 1.60327569e-01
4.27216403e-02 -4.28841174e-01 1.96995318e-01 1.43058074e+00
3.30973901e-02 -1.88476712e-01 -1.03408206e+00 9.43219244e-01
-1.84447789e+00 -4.93770868e-01 -1.86599806e-01 1.73775089e+00
1.04258251e+00 1.11607321e-01 -4.67088401e-01 1.07054017e-01
7.80587614e-01 4.32884932e-01 -6.72498792e-02 -9.08280015e-01
3.30716163e-01 7.77665496e-01 1.02631092e-01 7.11862624e-01
-8.02317500e-01 1.28391540e+00 6.34547234e+00 1.01702595e+00
-7.11464405e-01 2.97770768e-01 1.65084615e-01 2.17577174e-01
-5.70642650e-01 6.59353239e-03 -8.11967134e-01 3.89059901e-01
9.47390914e-01 -4.41792250e-01 -1.86431453e-01 7.85223246e-01
-5.37589341e-02 2.69589663e-01 -9.79657531e-01 9.10202384e-01
2.73559451e-01 -1.05581355e+00 2.42805898e-01 -3.51063237e-02
5.70925474e-01 -1.79600418e-01 -9.74643528e-02 5.88080823e-01
3.49089921e-01 -1.05652952e+00 -2.36813769e-01 4.03194577e-01
6.11500680e-01 -9.31603491e-01 1.18415058e+00 5.59724122e-02
-1.50088263e+00 -3.63623649e-02 -7.40778446e-01 -2.64717966e-01
3.78045499e-01 7.57416368e-01 -8.86299372e-01 4.97743905e-01
2.26136222e-01 9.96277630e-01 -4.54655647e-01 6.42297804e-01
-7.58672714e-01 7.23231733e-01 -1.54387593e-01 -7.07177937e-01
4.42425191e-01 -2.63389021e-01 1.56560123e-01 1.48193967e+00
2.59474754e-01 2.85967052e-01 -4.60025407e-02 5.57239890e-01
7.18526356e-03 6.22669220e-01 -7.73777366e-01 -5.04271626e-01
6.71209037e-01 1.44667888e+00 -3.12851101e-01 -3.91461551e-01
-5.40859818e-01 8.84716928e-01 7.89229155e-01 2.34092325e-01
-7.60056019e-01 -8.53111863e-01 1.15897548e+00 -2.23335549e-01
3.50039512e-01 -5.39553523e-01 -5.46997786e-02 -9.42635357e-01
-2.47379825e-01 -7.89751783e-02 4.02096719e-01 -6.04658961e-01
-1.88338625e+00 6.77454591e-01 1.59615904e-01 -8.68037403e-01
-2.18768075e-01 -1.02599621e+00 -1.04788136e+00 1.03398323e+00
-1.65585208e+00 -1.05151308e+00 1.74282134e-01 1.66788653e-01
5.02828896e-01 -3.11020702e-01 1.23959720e+00 3.32418144e-01
-4.67226893e-01 7.49677122e-01 -8.79738256e-02 2.73784190e-01
7.16519237e-01 -1.11387599e+00 3.24986875e-01 5.65964460e-01
7.72503138e-01 8.62382650e-01 4.26102668e-01 -6.25249505e-01
-8.91679227e-01 -1.14193511e+00 1.68738127e+00 -2.49054492e-01
9.80258465e-01 -3.25281292e-01 -1.23298395e+00 5.56491375e-01
4.54496533e-01 2.65151620e-01 1.32026899e+00 5.92572987e-01
-6.13188922e-01 -1.08590715e-01 -7.23867655e-01 5.61210155e-01
1.12907076e+00 -8.58033299e-01 -1.34727311e+00 2.36280248e-01
1.36030614e+00 2.20412716e-01 -9.42860007e-01 4.02902752e-01
4.38791096e-01 -5.94460592e-02 1.06650639e+00 -1.10663009e+00
4.21578377e-01 -8.44438151e-02 -4.41557020e-01 -1.69691098e+00
-4.53463733e-01 1.82138920e-01 -3.06718528e-01 1.55349362e+00
7.14794099e-01 -7.80488193e-01 2.79778391e-01 2.38691315e-01
-6.65291622e-02 -1.05508709e+00 -9.08295691e-01 -9.36610281e-01
2.11829230e-01 -4.59684014e-01 5.45744300e-01 1.25431657e+00
5.92076659e-01 7.26696908e-01 9.10845175e-02 8.02080706e-02
-5.45861423e-02 -2.82452732e-01 -8.47246572e-02 -9.80583549e-01
-1.17094971e-01 -2.63643235e-01 -7.08105445e-01 -9.26949024e-01
6.77841961e-01 -1.20165265e+00 1.39866337e-01 -1.60062075e+00
-1.13896161e-01 -2.19743311e-01 -1.12204516e+00 4.17201072e-01
-8.60967398e-01 2.06266969e-01 -1.74310014e-01 -3.09649765e-01
-2.57644534e-01 9.56433415e-01 8.04259181e-01 5.19334823e-02
2.21807342e-02 -4.94123220e-01 -5.18897235e-01 4.93905216e-01
9.51303780e-01 -5.41918516e-01 -5.50355017e-01 -5.67913890e-01
1.43973559e-01 -7.55057812e-01 -2.34054238e-01 -5.63135564e-01
7.00685382e-02 -2.61128128e-01 1.83613688e-01 -3.00375491e-01
4.31186765e-01 -9.14533675e-01 -4.69068676e-01 3.53049695e-01
-7.47484922e-01 3.89588475e-01 1.06315315e-01 6.20804906e-01
-4.21994865e-01 -6.09749973e-01 2.47166276e-01 2.61296004e-01
-8.88127744e-01 3.12222183e-01 -2.50262290e-01 9.30650812e-03
9.60419297e-01 6.79657683e-02 4.00508493e-02 -1.12169676e-01
-5.25323272e-01 1.31235525e-01 -2.75394827e-01 1.01599205e+00
9.24276888e-01 -1.96325672e+00 -5.06157935e-01 1.70296550e-01
6.46360338e-01 -4.88759041e-01 -1.68051019e-01 2.30130538e-01
2.93297432e-02 4.25770193e-01 9.41395983e-02 9.93215367e-02
-1.28186274e+00 7.61238396e-01 -1.98548585e-01 -5.89536607e-01
-2.21124873e-01 1.12496281e+00 5.82654364e-02 -4.59281772e-01
-7.82793686e-02 -3.98188859e-01 -7.62789845e-01 3.52200001e-01
7.91444242e-01 -1.77918375e-02 -4.32866476e-02 -6.57659769e-01
-5.28591335e-01 6.16308689e-01 -1.89441949e-01 -1.33393824e-01
1.54835355e+00 8.61739516e-02 -2.55102366e-01 5.12607098e-01
1.77320981e+00 -2.36306041e-01 -4.15124774e-01 -7.24097013e-01
4.65674579e-01 -6.39507651e-01 1.79310039e-01 -3.93505365e-01
-8.25044274e-01 1.27073061e+00 4.72400248e-01 1.24400690e-01
7.88530529e-01 1.71005294e-01 9.57318425e-01 3.59638661e-01
-1.26114070e-01 -1.18987203e+00 2.10977018e-01 7.68618166e-01
7.47513294e-01 -9.46193099e-01 -8.59020427e-02 -3.37036371e-01
-4.89231944e-01 1.32428801e+00 8.12372744e-01 -3.76980901e-01
1.24117947e+00 2.10176349e-01 -4.83753011e-02 -1.15337856e-01
-7.70205915e-01 -5.98842800e-01 7.14309990e-01 8.76418531e-01
8.83836210e-01 -5.60130738e-02 -1.21957672e+00 9.18326020e-01
1.25707209e-01 -5.68575025e-01 -8.50559548e-02 6.60309434e-01
-6.25319123e-01 -1.69805861e+00 1.82152361e-01 6.88109875e-01
-1.74393609e-01 -7.26622462e-01 -2.85033941e-01 2.88326293e-01
3.42649102e-01 9.72815692e-01 4.30922091e-01 -5.66234052e-01
1.82509720e-01 3.75075161e-01 2.20310137e-01 -1.11632812e+00
-3.75276119e-01 -4.22352076e-01 2.53213257e-01 -4.18585271e-01
-2.30593219e-01 -6.68548048e-02 -1.43640673e+00 1.34003252e-01
-8.63759398e-01 3.12314123e-01 6.79180145e-01 1.29953277e+00
3.26118559e-01 6.42789602e-01 6.57739580e-01 -5.02468348e-01
-3.89543712e-01 -1.40731478e+00 -4.43904340e-01 6.03939176e-01
-8.59910548e-02 -6.95199788e-01 -4.13667232e-01 -2.58355558e-01] | [10.39834213256836, 8.69887638092041] |
61a879ba-3e2d-4d20-8dcb-cc07e74a458a | flare7k-a-phenomenological-nighttime-flare | 2210.06570 | null | https://arxiv.org/abs/2210.06570v1 | https://arxiv.org/pdf/2210.06570v1.pdf | Flare7K: A Phenomenological Nighttime Flare Removal Dataset | Artificial lights commonly leave strong lens flare artifacts on images captured at night. Nighttime flare not only affects the visual quality but also degrades the performance of vision algorithms. Existing flare removal methods mainly focus on removing daytime flares and fail in nighttime. Nighttime flare removal is challenging because of the unique luminance and spectrum of artificial lights and the diverse patterns and image degradation of the flares captured at night. The scarcity of nighttime flare removal datasets limits the research on this crucial task. In this paper, we introduce, Flare7K, the first nighttime flare removal dataset, which is generated based on the observation and statistics of real-world nighttime lens flares. It offers 5,000 scattering and 2,000 reflective flare images, consisting of 25 types of scattering flares and 10 types of reflective flares. The 7,000 flare patterns can be randomly added to flare-free images, forming the flare-corrupted and flare-free image pairs. With the paired data, we can train deep models to restore flare-corrupted images taken in the real world effectively. Apart from abundant flare patterns, we also provide rich annotations, including the labeling of light source, glare with shimmer, reflective flare, and streak, which are commonly absent from existing datasets. Hence, our dataset can facilitate new work in nighttime flare removal and more fine-grained analysis of flare patterns. Extensive experiments show that our dataset adds diversity to existing flare datasets and pushes the frontier of nighttime flare removal. | ['Chen Change Loy', 'Ruicheng Feng', 'Shangchen Zhou', 'Chongyi Li', 'Yuekun Dai'] | 2022-10-12 | null | null | null | null | ['flare-removal'] | ['computer-vision'] | [ 4.74409729e-01 -1.16201341e+00 6.51964009e-01 -3.36876541e-01
-5.00078201e-01 -1.17608213e+00 5.27713954e-01 -6.31790459e-01
1.32853642e-01 1.01795769e+00 3.33742559e-01 8.50062668e-02
-2.61299312e-01 -6.32368982e-01 -6.88192248e-01 -1.08499050e+00
9.12287012e-02 -1.68384001e-01 1.64307266e-01 -6.51753962e-01
9.56851318e-02 6.53669953e-01 -2.09876084e+00 1.15623295e-01
1.26059210e+00 7.23942459e-01 5.50133288e-01 9.47917342e-01
1.23357870e-01 8.68557930e-01 -7.05689728e-01 2.77970340e-02
6.63322330e-01 -5.66508830e-01 2.36615911e-01 1.27314955e-01
1.47109783e+00 -3.73889059e-01 -5.68900406e-01 1.33275294e+00
6.95631742e-01 2.46249452e-01 2.29041412e-01 -1.21043825e+00
-1.00380099e+00 -1.40352130e-01 -5.90494275e-01 5.90261817e-01
2.54499346e-01 7.78806865e-01 4.45522159e-01 -7.88176179e-01
1.87441871e-01 8.47810626e-01 8.28508198e-01 3.93642247e-01
-8.61568272e-01 -8.92554760e-01 -2.44098231e-01 2.85490155e-01
-1.00625241e+00 -7.77165771e-01 6.85582280e-01 -4.95883197e-01
8.49985242e-01 5.28219879e-01 7.58430481e-01 7.38430023e-01
3.82882416e-01 3.91387701e-01 1.58617043e+00 -1.21196620e-01
-1.84770271e-01 -6.49764240e-01 3.73837531e-01 4.38532501e-01
6.78149521e-01 6.09930456e-01 -2.80597627e-01 1.85451254e-01
3.79945427e-01 4.13033038e-01 -1.25862408e+00 1.79445863e-01
-8.96249712e-01 4.99048859e-01 6.83435559e-01 3.38730179e-02
4.39686514e-02 -1.45324823e-02 -2.48233944e-01 5.23772895e-01
2.32909754e-01 2.76657790e-01 -5.64022183e-01 2.23750517e-01
-7.03109145e-01 5.50482333e-01 4.42901611e-01 7.01005101e-01
1.03605092e+00 3.95164043e-01 -1.75871715e-01 7.26199567e-01
1.24790370e-01 1.76558459e+00 1.90015358e-03 -6.57324374e-01
8.51682480e-03 2.93641835e-01 5.25175273e-01 -1.06019568e+00
-6.13502920e-01 -1.42384052e-01 -8.98205638e-01 4.86099869e-01
1.47898257e-01 -2.87579805e-01 -1.40127718e+00 1.39783847e+00
-2.65488654e-01 5.32795608e-01 6.34301752e-02 1.54608846e+00
1.19793200e+00 4.69879866e-01 -5.66198945e-01 -6.69889212e-01
8.58128607e-01 -9.08587098e-01 -1.30320835e+00 -8.02250087e-01
6.39151931e-02 -1.10053015e+00 1.03048050e+00 3.03450733e-01
-8.72462153e-01 -4.48217154e-01 -8.29977036e-01 -3.65026370e-02
-2.58759975e-01 -2.44502604e-01 9.76852179e-01 2.41384730e-01
-1.03353620e+00 1.54696986e-01 -4.29164171e-01 -7.30058923e-02
3.27751338e-01 -2.32819080e-01 -2.32248455e-01 -4.95467097e-01
-1.15685356e+00 6.77516401e-01 -2.24623024e-01 6.44186676e-01
-8.10763240e-01 -8.41753125e-01 -7.64656842e-01 -3.18553269e-01
2.78564274e-01 -6.93454444e-01 8.81180048e-01 -9.69905913e-01
-7.55971193e-01 8.23565722e-01 -3.25128555e-01 -3.47304314e-01
-3.71010303e-02 -2.30277523e-01 -1.12030125e+00 -1.08864069e-01
1.05276987e-01 -9.28533673e-02 1.16096234e+00 -1.53876710e+00
-9.56694901e-01 -3.44461292e-01 4.53520082e-02 2.05484301e-01
5.99703550e-01 -7.25583658e-02 -2.56325781e-01 -5.18062592e-01
-1.90263569e-01 -9.08235788e-01 1.22260742e-01 -2.84747720e-01
-6.33521974e-02 4.55921620e-01 1.02614212e+00 -4.78418320e-01
1.18456006e+00 -2.28590035e+00 -2.74139583e-01 -5.61033309e-01
2.35935479e-01 6.34554863e-01 -2.15508446e-01 1.14042498e-01
-8.48637596e-02 -3.45634639e-01 -6.35208130e-01 1.82509780e-01
-9.22051221e-02 4.91005898e-01 -7.80665874e-01 5.76234102e-01
2.06797078e-01 9.29596424e-01 -9.80663002e-01 2.16005355e-01
4.66673851e-01 4.49245870e-01 -1.81659997e-01 2.16415197e-01
-2.43300065e-01 6.14464581e-01 -1.05073273e-01 8.33947122e-01
1.53182554e+00 3.82171333e-01 -6.23040318e-01 -3.81024003e-01
-4.59982187e-01 -1.81570992e-01 -9.02653813e-01 1.44564104e+00
-3.08753192e-01 9.94648159e-01 2.19308570e-01 2.17175514e-01
9.21916187e-01 -3.60475749e-01 -1.03053503e-01 -1.11876345e+00
8.98070633e-02 1.90998778e-01 -8.38064179e-02 -8.72004807e-01
5.21896780e-01 -4.13638741e-01 9.23273936e-02 3.87317427e-02
-3.89050812e-01 -3.65909904e-01 3.79861265e-01 -2.29148626e-01
1.40320897e+00 -6.41354546e-02 -2.91023970e-01 -8.46823677e-02
3.81365806e-01 -2.77515780e-02 7.40098119e-01 9.13184285e-01
-3.80599678e-01 1.02657926e+00 -5.25861681e-01 -6.82886064e-01
-3.73318762e-01 -1.57334328e+00 -4.22002077e-01 9.30993617e-01
2.43243471e-01 2.91893054e-02 -2.13784516e-01 -2.96256065e-01
1.06223211e-01 5.74636340e-01 -3.07664990e-01 -2.37500355e-01
-4.49282616e-01 -1.18127394e+00 1.36012077e-01 1.72424644e-01
9.66650248e-01 -1.38824844e+00 -5.94016731e-01 -2.96632171e-01
-3.38913888e-01 -8.83641005e-01 -3.33232760e-01 4.00714055e-02
-2.65603185e-01 -1.52976358e+00 -5.34676574e-02 -5.10091007e-01
2.74805039e-01 1.24967194e+00 1.46682084e+00 2.27382973e-01
-9.09302354e-01 2.18059734e-01 -5.11127174e-01 -7.30495572e-01
1.38287619e-01 -9.37435567e-01 8.82307887e-02 -7.89728463e-02
6.53999686e-01 -5.73270917e-01 -9.28597569e-01 -1.48476690e-01
-1.14358425e+00 -3.63552272e-01 2.60586590e-01 5.20923615e-01
7.04261303e-01 7.44485319e-01 2.70827949e-01 -7.65527964e-01
3.50117296e-01 -1.70913890e-01 -7.29677916e-01 -1.24795094e-01
-1.70578226e-01 -6.04064584e-01 7.67841458e-01 -2.39178184e-02
-1.36686802e+00 -1.11780465e-01 2.85750151e-01 -5.47531903e-01
-3.48508149e-01 1.67434290e-02 -3.17183048e-01 -4.53485638e-01
9.58404183e-01 1.29531836e-02 -4.57376003e-01 -5.10880232e-01
3.45591694e-01 7.26112545e-01 1.11661315e+00 2.46276259e-01
1.20426738e+00 1.00248051e+00 -2.79231295e-02 -8.88351560e-01
-1.53780305e+00 -5.48653305e-01 -2.08901942e-01 -2.29682475e-01
6.39251888e-01 -1.02970886e+00 -5.06491184e-01 1.05471718e+00
-5.51826358e-01 -5.23629129e-01 -3.07782024e-01 2.79545605e-01
-3.56914312e-01 1.79722384e-01 -1.53130457e-01 -9.12036180e-01
-3.02938610e-01 -7.46526361e-01 1.12732899e+00 9.18561518e-01
6.99811935e-01 -7.37929285e-01 3.29293966e-01 3.64550143e-01
3.31964821e-01 4.61656451e-01 3.73844594e-01 5.24201035e-01
-8.53949785e-01 4.37274352e-02 -4.37395990e-01 3.45964283e-01
7.81421483e-01 4.35437821e-02 -1.27321947e+00 -5.66779852e-01
3.89242113e-01 -1.21187791e-03 1.53208768e+00 6.47859156e-01
5.71947515e-01 5.83546124e-02 -4.26168479e-02 1.24749362e+00
1.78857684e+00 3.08331758e-01 9.96273279e-01 5.41881919e-01
1.09245002e+00 2.90857524e-01 8.68707359e-01 4.65478480e-01
6.67419508e-02 1.39765054e-01 7.22791374e-01 -4.56390470e-01
-6.24818206e-01 3.44445318e-01 2.88319319e-01 2.87279963e-01
1.53508514e-01 -5.42629719e-01 -7.04133391e-01 6.53245032e-01
-1.72643495e+00 -1.36922622e+00 -9.61526692e-01 2.20619011e+00
7.53974855e-01 -3.56521696e-01 -6.00207329e-01 -4.49376181e-02
5.43127298e-01 4.92627889e-01 -5.58818340e-01 -8.44606608e-02
-1.08906281e+00 3.88237178e-01 7.85524368e-01 7.56292164e-01
-1.32980144e+00 9.42982972e-01 5.81946230e+00 1.01133153e-01
-1.21968198e+00 -2.39659458e-01 -4.34612721e-01 -4.97797102e-01
-5.69314659e-01 -2.38002464e-02 -8.37825537e-01 6.96515918e-01
5.96239269e-01 -4.99183536e-02 1.15655804e+00 -1.42299801e-01
5.83557487e-01 -4.57681149e-01 -4.76644188e-01 1.29717958e+00
6.92649901e-01 -9.14091587e-01 -7.71115646e-02 -3.85857493e-01
1.07061982e+00 4.32377487e-01 -1.22307934e-01 1.04004405e-01
4.09523875e-01 -1.00142539e+00 6.51713237e-02 1.56520331e+00
5.97655773e-01 -8.40316355e-01 5.08989394e-01 1.29437402e-01
-8.24949861e-01 -2.09866032e-01 -5.43531358e-01 -3.81323159e-01
3.42549741e-01 1.08308315e+00 -9.77300480e-02 5.00037491e-01
1.30483711e+00 1.20617211e+00 -8.86268079e-01 1.35975921e+00
-4.45263743e-01 4.02044445e-01 -2.54930049e-01 7.81779587e-01
-7.56040812e-02 -7.27635324e-01 9.09286439e-01 1.04127133e+00
2.07885906e-01 1.18457235e-01 2.36425802e-01 4.44147497e-01
2.10641354e-01 -6.99637115e-01 -9.96740401e-01 8.90642181e-02
1.51810914e-01 1.51321304e+00 -4.79206331e-02 2.02539161e-01
-4.55085218e-01 8.71284962e-01 -4.27855879e-01 5.45558274e-01
-8.47706795e-01 -3.61579269e-01 1.28669143e+00 1.28857806e-01
2.08578393e-01 -2.11800352e-01 2.46068031e-01 -1.12610173e+00
1.48967549e-01 -7.35391080e-01 2.44972587e-01 -1.69394684e+00
-1.75677264e+00 5.87421000e-01 -3.58783036e-01 -1.22158229e+00
7.30853975e-02 -6.33134127e-01 -6.11528516e-01 1.06053269e+00
-2.28678298e+00 -1.22572231e+00 -1.22267914e+00 9.05740917e-01
3.93623829e-01 6.12193942e-02 6.92517996e-01 4.37707990e-01
-3.97895634e-01 -1.47634670e-01 5.71223676e-01 -2.37098604e-01
1.32942176e+00 -1.31279910e+00 1.95002064e-01 1.38651681e+00
-4.34032045e-02 2.57872909e-01 1.14279389e+00 -7.49494791e-01
-1.64046311e+00 -1.74924004e+00 6.00441635e-01 -6.94603980e-01
9.48845744e-01 -8.14076811e-02 -1.14232624e+00 7.03591526e-01
-7.68172517e-02 4.20625865e-01 4.15584117e-01 -3.87560129e-01
-1.52587965e-02 -3.83635402e-01 -9.06634271e-01 6.74437881e-01
1.33261681e+00 -5.32142580e-01 -7.51477122e-01 7.35432982e-01
5.34913778e-01 -3.80273312e-01 -1.02779649e-01 9.30339634e-01
1.34957001e-01 -1.55797088e+00 8.83572519e-01 -1.27342463e-01
9.15384218e-02 -9.33803678e-01 -3.72801781e-01 -1.54544997e+00
-6.12124026e-01 -7.92581558e-01 -1.34206936e-01 1.33866572e+00
-3.04761499e-01 -4.91007984e-01 2.19386332e-02 -2.42840238e-02
-6.34473562e-01 2.84386545e-01 -2.70404220e-01 -5.42718649e-01
-2.65743524e-01 -2.50864953e-01 5.78810275e-01 8.99595261e-01
-1.32584023e+00 2.06529886e-01 -3.89853507e-01 6.95207596e-01
7.87656903e-01 8.61186206e-01 9.20917332e-01 -1.57770562e+00
-4.49828506e-02 1.81521103e-02 -1.72513738e-01 -4.45291013e-01
2.53420383e-01 -7.47738183e-01 5.23711801e-01 -1.81692457e+00
4.25314754e-02 -6.99517317e-03 -2.17093483e-01 4.99176085e-01
-4.42396134e-01 9.26147103e-01 1.13947302e-01 3.87497455e-01
-4.50112551e-01 3.52080673e-01 1.49709523e+00 -2.78194118e-02
-2.95769632e-01 -3.95155817e-01 -6.94786966e-01 1.16142213e+00
6.09810114e-01 3.92625406e-02 -4.52970475e-01 -8.48004878e-01
4.69285280e-01 -2.05605477e-01 7.36963212e-01 -1.22104514e+00
8.16189423e-02 -4.10266608e-01 4.69698310e-01 -7.49207556e-01
1.78571478e-01 -8.53775382e-01 3.87507170e-01 -3.98408063e-02
5.51566422e-01 -4.69144076e-01 4.46912020e-01 6.32256091e-01
-1.01213157e-01 -8.76426995e-02 7.75231242e-01 -2.15412721e-01
-1.16372132e+00 4.15665805e-01 -1.72244191e-01 1.70151293e-01
6.63477302e-01 -3.13699931e-01 -1.21525121e+00 -6.91734254e-02
-5.43001294e-01 2.14278981e-01 8.91269922e-01 6.14209712e-01
4.17919993e-01 -1.07364070e+00 -8.34142447e-01 8.72003436e-01
3.96934062e-01 -8.21561888e-02 8.19059789e-01 7.56514907e-01
-5.81107497e-01 1.62819505e-01 -4.17291045e-01 -3.23638767e-01
-1.22112608e+00 5.23269057e-01 6.05531096e-01 4.94334787e-01
-9.62903380e-01 6.27247691e-01 7.71551907e-01 1.49550945e-01
-2.82680690e-01 -9.01901424e-02 -4.26772356e-01 2.51150094e-02
1.10178936e+00 3.34264517e-01 3.05330068e-01 -7.36513793e-01
-1.50276348e-01 8.91313195e-01 1.74845830e-01 7.89691746e-01
1.28473282e+00 -7.25282252e-01 -4.23202455e-01 7.99768686e-01
6.42030120e-01 4.88643318e-01 -1.49498546e+00 3.54774557e-02
-8.54933977e-01 -1.08348083e+00 5.37126482e-01 -1.27977419e+00
-1.29253864e+00 4.18003201e-01 1.03364527e+00 2.32633770e-01
2.00477600e+00 -4.81382102e-01 9.95124340e-01 3.39881033e-01
4.16204304e-01 -4.76053238e-01 -4.32921618e-01 9.32983041e-01
9.36757743e-01 -1.36628675e+00 -3.80850792e-01 -2.28013739e-01
-3.37384641e-01 7.14447021e-01 4.81018931e-01 -4.59933907e-01
3.10784310e-01 6.89203203e-01 5.69421053e-01 -4.36353594e-01
-5.83684981e-01 -9.88829970e-01 1.38741508e-01 1.08983028e+00
6.03027865e-02 -3.26176882e-02 -8.35698247e-02 4.64364678e-01
-5.28024793e-01 7.52959028e-02 7.83846915e-01 8.20057392e-01
-8.84221613e-01 -4.57791030e-01 -6.76149547e-01 6.13960564e-01
-1.36632159e-01 -4.25726295e-01 -5.73140323e-01 4.03301179e-01
6.03038430e-01 1.40345538e+00 3.45408738e-01 2.88181333e-03
5.46592891e-01 -1.68185085e-01 7.05920458e-01 -3.44798356e-01
-5.22683620e-01 1.74946517e-01 -1.38909325e-01 -6.45444334e-01
-5.89672923e-01 -2.82718927e-01 -1.06897032e+00 -4.28018093e-01
-1.91098601e-01 -2.41297528e-01 2.32028499e-01 6.40358388e-01
2.84772426e-01 6.95620298e-01 7.53498673e-01 -8.61745894e-01
3.93612504e-01 -7.77754962e-01 -1.27926040e+00 8.32615435e-01
1.04317212e+00 -9.17575538e-01 -1.30024636e+00 3.37648213e-01] | [10.751118659973145, -3.1057324409484863] |
c96bb7de-acd5-48f6-bc48-ee17de880015 | siatrans-siamese-transformer-network-for-rgb | 2207.04224 | null | https://arxiv.org/abs/2207.04224v1 | https://arxiv.org/pdf/2207.04224v1.pdf | SiaTrans: Siamese Transformer Network for RGB-D Salient Object Detection with Depth Image Classification | RGB-D SOD uses depth information to handle challenging scenes and obtain high-quality saliency maps. Existing state-of-the-art RGB-D saliency detection methods overwhelmingly rely on the strategy of directly fusing depth information. Although these methods improve the accuracy of saliency prediction through various cross-modality fusion strategies, misinformation provided by some poor-quality depth images can affect the saliency prediction result. To address this issue, a novel RGB-D salient object detection model (SiaTrans) is proposed in this paper, which allows training on depth image quality classification at the same time as training on SOD. In light of the common information between RGB and depth images on salient objects, SiaTrans uses a Siamese transformer network with shared weight parameters as the encoder and extracts RGB and depth features concatenated on the batch dimension, saving space resources without compromising performance. SiaTrans uses the Class token in the backbone network (T2T-ViT) to classify the quality of depth images without preventing the token sequence from going on with the saliency detection task. Transformer-based cross-modality fusion module (CMF) can effectively fuse RGB and depth information. And in the testing process, CMF can choose to fuse cross-modality information or enhance RGB information according to the quality classification signal of the depth image. The greatest benefit of our designed CMF and decoder is that they maintain the consistency of RGB and RGB-D information decoding: SiaTrans decodes RGB-D or RGB information under the same model parameters according to the classification signal during testing. Comprehensive experiments on nine RGB-D SOD benchmark datasets show that SiaTrans has the best overall performance and the least computation compared with recent state-of-the-art methods. | ['Yanjun Peng', 'Dongye Changlei', 'Xingzhao Jia'] | 2022-07-09 | null | null | null | null | ['rgb-d-salient-object-detection'] | ['computer-vision'] | [ 4.12564158e-01 2.50468627e-02 -2.31220275e-01 -1.94006562e-01
-7.04894066e-01 -1.02587594e-02 1.33415639e-01 -1.30467594e-01
-2.95114249e-01 3.27941686e-01 1.95074737e-01 -7.02390680e-04
2.18924545e-02 -8.06314111e-01 -6.38502479e-01 -8.69855046e-01
4.06738013e-01 -2.23052263e-01 8.83427918e-01 -4.71729517e-01
4.68515009e-01 1.81233615e-01 -2.02236938e+00 4.87372637e-01
1.06854212e+00 1.66861892e+00 6.90064609e-01 4.48779345e-01
-1.03346951e-01 8.17954659e-01 -5.60716152e-01 -2.68394768e-01
4.30206746e-01 -3.98034155e-01 -5.64758062e-01 -1.55581623e-01
1.70120671e-01 -4.05018806e-01 -5.47405005e-01 1.09558475e+00
8.66197824e-01 -2.54361868e-01 1.78970978e-01 -1.60160661e+00
-5.39664984e-01 3.84281814e-01 -7.33171463e-01 3.77896577e-01
4.43211198e-01 2.26973563e-01 7.20658243e-01 -9.24725533e-01
3.86270344e-01 1.02057564e+00 3.81560236e-01 4.42551315e-01
-6.72835052e-01 -6.65717542e-01 7.59864822e-02 6.24721646e-01
-1.26417184e+00 -1.60192698e-01 1.27987683e+00 1.45591557e-01
6.65667713e-01 3.39269370e-01 9.69661832e-01 7.39917457e-01
3.87793779e-01 1.33736479e+00 1.17149270e+00 -1.98412061e-01
3.25958937e-01 2.84535438e-03 -2.87953019e-01 7.85100996e-01
5.10748066e-02 6.20706566e-02 -1.20710695e+00 4.33305025e-01
7.09624827e-01 9.90718529e-02 -3.95072877e-01 -4.16430146e-01
-1.38731444e+00 4.93565440e-01 1.01023889e+00 2.76874781e-01
-3.18023235e-01 -5.79774082e-02 1.87001437e-01 5.59343472e-02
2.32664227e-01 6.27083182e-02 -4.04120207e-01 -1.16012812e-01
-9.76367176e-01 -1.18608773e-01 1.31032512e-01 9.25198197e-01
8.43829751e-01 3.66784707e-02 -2.20007434e-01 4.44296926e-01
4.20723647e-01 8.77960801e-01 8.19467306e-01 -6.58714592e-01
6.64120197e-01 9.47089493e-01 -2.47706503e-01 -9.37152863e-01
-5.51142812e-01 -4.76182789e-01 -5.87703049e-01 3.92374992e-01
-2.34702256e-05 2.32026145e-01 -1.13911605e+00 1.50786185e+00
3.90759140e-01 7.07729384e-02 3.01140994e-01 1.44154310e+00
1.11496890e+00 4.31138843e-01 -1.99454471e-01 1.02960318e-01
1.19879127e+00 -7.44358540e-01 -6.29820049e-01 -5.36407113e-01
4.06061828e-01 -6.72319055e-01 1.02399981e+00 9.97246802e-02
-1.06524241e+00 -6.80135131e-01 -1.48738408e+00 -4.15747046e-01
-3.70263964e-01 3.39419335e-01 7.07870424e-01 5.39065897e-01
-1.07887721e+00 3.54375094e-01 -8.03820491e-01 -1.32941930e-02
4.39431250e-01 5.00794768e-01 -2.15963200e-01 -1.24298096e-01
-1.41212928e+00 1.03591633e+00 3.37313205e-01 2.13687271e-01
-8.28901589e-01 -5.74150741e-01 -9.08971369e-01 -8.83570760e-02
2.68850833e-01 -3.56297523e-01 1.09558678e+00 -1.14834321e+00
-1.35919678e+00 6.33385599e-01 -1.36804298e-01 -1.64824992e-01
2.57231474e-01 -1.31015563e-02 -4.50154394e-01 5.38535893e-01
2.79659659e-01 1.00384569e+00 9.83986735e-01 -1.08627272e+00
-1.13549030e+00 -5.61662138e-01 1.29911629e-02 5.54530919e-01
-2.41198435e-01 -4.51599896e-01 -5.79949141e-01 -4.94527519e-01
8.88229609e-01 -6.69248104e-01 2.41886437e-01 1.59630418e-01
-3.69241714e-01 4.27656434e-02 1.07709599e+00 -4.31546628e-01
1.11722720e+00 -2.27366710e+00 3.37031931e-01 -2.46297102e-02
2.96280593e-01 2.33587310e-01 1.04446867e-02 -1.42117187e-01
8.08449686e-02 -3.39905620e-01 -3.10619771e-01 -2.95591027e-01
-1.25420019e-01 1.47006541e-01 -1.19246736e-01 5.32337129e-01
4.84311968e-01 1.06669211e+00 -9.85154927e-01 -7.51155794e-01
3.98988813e-01 3.30281138e-01 -3.07161182e-01 1.11556888e-01
5.18972576e-02 1.72015607e-01 -5.66763043e-01 1.15501881e+00
8.01369369e-01 -2.71685809e-01 -2.62198329e-01 -5.27388513e-01
-1.41851917e-01 4.29801434e-01 -1.08253312e+00 2.13191295e+00
-2.20908090e-01 6.21055961e-01 -2.58597672e-01 -8.25937629e-01
8.54029775e-01 -3.85385263e-03 5.67611873e-01 -1.43008673e+00
2.79497504e-01 4.82377023e-01 -1.06021672e-01 -2.94816732e-01
6.11767352e-01 8.78150240e-02 -2.04621449e-01 1.88692197e-01
5.43733053e-02 -3.20667505e-01 -2.33158007e-01 2.14384943e-01
8.83959115e-01 2.43304372e-01 -8.93028751e-02 1.10880680e-01
4.60369647e-01 4.39091697e-02 7.18934953e-01 3.91917050e-01
-4.12030488e-01 7.28449702e-01 3.09569120e-01 1.60638727e-02
-8.89450967e-01 -1.21737361e+00 -1.77594926e-02 7.64098585e-01
9.91981149e-01 -4.32332307e-02 -5.00357926e-01 -7.73514807e-01
-5.81854582e-02 4.98328656e-01 -6.69218242e-01 -6.77108884e-01
-2.12083548e-01 -5.58844030e-01 3.59570384e-01 4.18043077e-01
1.14340258e+00 -8.87740076e-01 -1.23879623e+00 9.45182070e-02
-3.90364677e-01 -1.09983599e+00 -2.84011722e-01 5.10498047e-01
-8.73278260e-01 -9.22957897e-01 -8.13183010e-01 -8.41298342e-01
4.76469666e-01 7.71156490e-01 6.16811872e-01 -1.23254694e-01
-9.14929658e-02 4.80899811e-02 -5.43376863e-01 -4.14921373e-01
4.96272333e-02 5.80607913e-02 -2.74793923e-01 -8.27848017e-02
3.75226378e-01 -3.44722331e-01 -8.39978039e-01 3.52986068e-01
-1.07378447e+00 6.33919775e-01 8.15853536e-01 6.70784652e-01
5.56610346e-01 2.63795108e-02 4.90837634e-01 5.92247993e-02
-1.53269069e-02 -3.19971472e-01 -4.80249524e-01 1.79003716e-01
-6.28983259e-01 1.33710533e-01 1.72842115e-01 -2.18496278e-01
-8.56154025e-01 7.05823079e-02 5.03753275e-02 -5.66129684e-01
3.30191553e-01 2.04372942e-01 -3.34713995e-01 -2.98790663e-01
3.04746658e-01 6.48947299e-01 8.57219622e-02 -1.51750728e-01
4.97282296e-02 7.28154540e-01 6.09405100e-01 1.99192718e-01
5.52246511e-01 5.07373035e-01 1.39613837e-01 -2.79721826e-01
-9.02117252e-01 -3.91780287e-01 -3.55953962e-01 -5.49979031e-01
8.68039429e-01 -1.22261214e+00 -7.21383333e-01 8.93411458e-01
-7.93045163e-01 -5.29215150e-02 -1.54552624e-01 4.76878911e-01
-3.70477200e-01 1.39364809e-01 -3.93267393e-01 -6.33648634e-01
-2.60733306e-01 -1.50650239e+00 1.51132822e+00 5.75217426e-01
4.31298733e-01 -4.61656183e-01 -6.33341014e-01 4.14646804e-01
3.14113110e-01 5.64069115e-02 6.12125874e-01 6.53852150e-02
-7.53459513e-01 1.27338981e-02 -5.61232626e-01 2.96667397e-01
2.38329470e-01 -5.41540802e-01 -1.16609967e+00 1.25006080e-01
1.15616672e-01 -2.14323208e-01 8.83193254e-01 3.50734681e-01
9.24693882e-01 1.18629456e-01 -2.03696355e-01 5.50129294e-01
1.57781494e+00 1.50873765e-01 7.88208187e-01 6.45153999e-01
8.83195341e-01 2.00885460e-01 1.10796058e+00 3.75168115e-01
7.78467596e-01 7.30166376e-01 9.01221693e-01 -3.87191206e-01
-4.66611922e-01 -2.30195791e-01 6.11286283e-01 6.56684518e-01
3.67077440e-01 -1.36452690e-01 -7.36093402e-01 5.46886861e-01
-1.73159313e+00 -8.50678980e-01 -1.05596714e-01 2.04615664e+00
1.08309293e+00 4.07572299e-01 -1.01763010e-01 6.18095160e-01
4.91530329e-01 9.90498513e-02 -7.78089106e-01 -1.50008257e-02
-5.49131095e-01 9.16011855e-02 9.38286662e-01 8.95203501e-02
-9.20476854e-01 8.01244199e-01 5.13364267e+00 9.21994448e-01
-1.56700063e+00 2.11802468e-01 5.26523292e-01 -5.00037313e-01
-4.57504392e-01 -1.30562216e-01 -6.92751706e-01 7.54267633e-01
4.64287251e-01 2.51616746e-01 2.41010219e-01 7.29206085e-01
1.09704398e-01 -9.13399160e-01 -7.76244402e-01 1.22331166e+00
3.07406306e-01 -1.08718848e+00 -1.54321641e-01 -2.53421217e-01
7.17749417e-01 1.87739938e-01 3.55192661e-01 2.03250442e-02
-2.27035895e-01 -6.58912957e-01 1.30907953e+00 5.84086835e-01
6.66272819e-01 -6.84331357e-01 1.01111627e+00 1.14594437e-01
-1.14697373e+00 -3.20380211e-01 -2.99999803e-01 5.14684953e-02
8.03401619e-02 8.77003312e-01 -5.71912587e-01 8.01399231e-01
1.01179504e+00 9.18917656e-01 -9.28654611e-01 1.02679884e+00
-3.43347311e-01 2.10610852e-01 -4.20001358e-01 -9.72767174e-02
3.52219045e-01 8.66933241e-02 5.18647373e-01 6.14652753e-01
4.53911632e-01 5.34673780e-02 -1.12229571e-01 6.82122350e-01
3.03423673e-01 -2.98485905e-01 -1.33233190e-01 2.86675692e-01
3.42627376e-01 9.38995600e-01 -7.11023033e-01 -4.13139284e-01
-3.70752990e-01 1.19299841e+00 -3.80775705e-02 1.12664625e-01
-1.19032466e+00 -4.27391917e-01 5.82610667e-01 -1.22699216e-01
6.07673705e-01 -1.32360114e-02 -8.40181351e-01 -9.14874673e-01
2.98252761e-01 -6.58578694e-01 1.24808609e-01 -1.32011783e+00
-6.06235862e-01 4.93321717e-01 -3.23855817e-01 -1.72023404e+00
2.71433294e-01 -5.04269600e-01 -9.62076709e-02 8.78408670e-01
-2.00801110e+00 -1.17722738e+00 -6.43375337e-01 7.25640833e-01
3.16369206e-01 3.56019475e-02 1.00739971e-01 2.80517250e-01
-5.15134692e-01 5.31474888e-01 -1.89146906e-01 -1.57585427e-01
6.03246272e-01 -1.11731076e+00 -5.82523309e-02 1.01469326e+00
-3.13858926e-01 6.31459465e-04 5.71920156e-01 -6.66470408e-01
-1.73035753e+00 -9.64501739e-01 5.98658144e-01 1.64299496e-02
2.98961729e-01 -2.21618220e-01 -6.35959804e-01 -1.10467121e-01
-1.30542204e-01 1.09578274e-01 2.77600825e-01 -5.81774771e-01
-4.62929420e-02 -4.02930319e-01 -1.26016462e+00 4.61723179e-01
1.11867118e+00 -6.73163116e-01 -6.04347050e-01 -4.24328111e-02
1.00566614e+00 -7.44427860e-01 -6.98204517e-01 6.23940170e-01
4.99345660e-01 -1.31920063e+00 9.39618051e-01 2.64492095e-01
8.26805949e-01 -8.18082333e-01 -4.07787681e-01 -1.03726494e+00
-6.04232354e-03 2.25503910e-02 -7.37668425e-02 1.01148772e+00
2.48944208e-01 -4.23081130e-01 7.49405146e-01 3.33514184e-01
-3.72690767e-01 -9.33278084e-01 -1.35389376e+00 -4.43781406e-01
-7.08862245e-01 -6.75110281e-01 8.30874205e-01 4.64951873e-01
2.70553380e-02 4.88896742e-02 -1.56220749e-01 4.08355981e-01
5.93421459e-01 3.51049513e-01 3.81244570e-01 -9.22050655e-01
1.00193165e-01 -5.11275053e-01 -8.64377260e-01 -1.06775808e+00
-2.11864531e-01 -7.52003670e-01 2.43754715e-01 -1.60935926e+00
-3.74034494e-02 -6.06940508e-01 -5.53612232e-01 7.30997503e-01
-3.38457555e-01 6.08775377e-01 2.30422541e-01 1.27529174e-01
-7.48714268e-01 9.82884645e-01 1.65730679e+00 -3.88295352e-01
-2.49561042e-01 -3.05931151e-01 -6.54594064e-01 2.55411923e-01
6.23553514e-01 -3.59732240e-01 -5.71942389e-01 -4.89663363e-01
2.38039732e-01 1.31878585e-01 6.64265692e-01 -1.39261031e+00
4.09410894e-01 1.97666362e-02 7.85362244e-01 -8.94604862e-01
4.98007208e-01 -8.85263503e-01 -2.69706100e-01 4.76936638e-01
1.42947668e-02 -8.21968690e-02 2.33013824e-01 2.75746286e-01
-3.92768741e-01 1.39355659e-01 8.18814695e-01 2.21761748e-01
-1.18681538e+00 1.30746678e-01 -1.04093559e-01 -2.25393385e-01
1.08122909e+00 -7.68520594e-01 -4.28610027e-01 -1.60690010e-01
-2.78132290e-01 1.22498453e-01 6.33432269e-01 5.89961290e-01
1.04434240e+00 -1.47070241e+00 -2.16050193e-01 4.67814088e-01
3.32509845e-01 1.13709345e-01 4.02835190e-01 1.11582720e+00
-2.71363139e-01 2.75243223e-01 -5.61630964e-01 -9.41391051e-01
-9.60582316e-01 4.73306477e-01 2.87915796e-01 1.08266532e-01
-3.50250453e-01 9.49261010e-01 4.71442612e-03 2.23309305e-02
1.10429704e-01 -4.73477960e-01 -1.67074606e-01 9.13120955e-02
4.35713440e-01 1.99827269e-01 3.66923630e-01 -8.06336761e-01
-6.44976318e-01 4.86024529e-01 2.04014584e-01 -1.43970519e-01
1.13489330e+00 -4.72966373e-01 3.10700256e-02 4.47124302e-01
1.21811426e+00 -3.93407315e-01 -1.46104741e+00 -2.51412302e-01
-4.23112571e-01 -6.06109321e-01 7.84558475e-01 -9.09087360e-01
-1.44351673e+00 7.59676099e-01 1.20105493e+00 -1.72158360e-01
1.79361093e+00 -4.16821465e-02 1.00112867e+00 -2.31918201e-01
4.95507747e-01 -1.27129459e+00 2.78630167e-01 2.52611816e-01
7.24591434e-01 -1.50567448e+00 6.61827400e-02 -4.12453800e-01
-9.09057796e-01 8.78925323e-01 7.34285176e-01 1.49854481e-01
4.01334822e-01 6.99686781e-02 -1.24764536e-02 -2.13753492e-01
-4.47755754e-01 -5.35007834e-01 4.11399633e-01 5.54447651e-01
-1.10399656e-01 -6.77430704e-02 -5.10159740e-03 4.73178715e-01
-2.41513997e-01 1.47996179e-03 3.96857917e-01 1.20542562e+00
-5.55536926e-01 -6.43411875e-01 -4.24470127e-01 4.59138751e-01
-9.77547467e-02 -2.03245953e-01 -4.11568508e-02 4.36162382e-01
4.39132333e-01 1.10872459e+00 4.17747945e-02 -7.44562447e-01
2.27026910e-01 -3.15319479e-01 3.82156253e-01 -1.22683331e-01
-5.14173687e-01 -2.72758286e-02 -3.02664429e-01 -9.25162256e-01
-7.14661002e-01 -6.59701705e-01 -1.52912903e+00 -1.76082835e-01
-4.86670196e-01 -1.02848634e-01 9.04072762e-01 1.13534153e+00
4.23262447e-01 7.76521802e-01 8.95659268e-01 -9.57965553e-01
-1.03563875e-01 -7.12814689e-01 -6.65690184e-01 1.45780176e-01
6.58351541e-01 -1.01382887e+00 -2.77831107e-01 -2.39040121e-01] | [9.666224479675293, -0.8335204720497131] |
cffae4bf-9007-448b-a79e-5bc323c8f848 | classification-of-social-media-toxic-comments | 2304.06934 | null | https://arxiv.org/abs/2304.06934v1 | https://arxiv.org/pdf/2304.06934v1.pdf | Classification of social media Toxic comments using Machine learning models | The abstract outlines the problem of toxic comments on social media platforms, where individuals use disrespectful, abusive, and unreasonable language that can drive users away from discussions. This behavior is referred to as anti-social behavior, which occurs during online debates, comments, and fights. The comments containing explicit language can be classified into various categories, such as toxic, severe toxic, obscene, threat, insult, and identity hate. This behavior leads to online harassment and cyberbullying, which forces individuals to stop expressing their opinions and ideas. To protect users from offensive language, companies have started flagging comments and blocking users. The abstract proposes to create a classifier using an Lstm-cnn model that can differentiate between toxic and non-toxic comments with high accuracy. The classifier can help organizations examine the toxicity of the comment section better. | ['S. Ayyasamy', 'M. Arun Kuamr Reddy', 'A. Sai Charish', 'K. Poojitha'] | 2023-04-14 | null | null | null | null | ['blocking'] | ['natural-language-processing'] | [-3.56406599e-01 2.39291847e-01 -9.19347107e-02 1.16533246e-02
1.04439147e-01 -9.58872616e-01 5.36179781e-01 5.67496836e-01
-2.83862203e-01 8.13167453e-01 5.43895304e-01 -5.08805633e-01
4.85783279e-01 -4.97513175e-01 1.44384906e-01 -4.62080181e-01
5.46679020e-01 -3.86931866e-01 -3.15590888e-01 -6.26428306e-01
1.01726568e+00 3.21485490e-01 -8.09982121e-01 7.07241893e-01
7.03864038e-01 5.26127875e-01 -7.37175047e-01 5.91623724e-01
-3.53387117e-01 1.36612260e+00 -1.16905582e+00 -7.68551230e-01
-2.51159489e-01 -4.36730117e-01 -6.90790772e-01 -1.80179849e-02
2.95182675e-01 -5.24524748e-01 1.52284399e-01 1.52676868e+00
3.98512006e-01 -2.31535196e-01 2.65747666e-01 -1.23045588e+00
-1.32983291e+00 6.83382332e-01 -6.11663222e-01 2.50651389e-01
6.83728397e-01 3.21521908e-02 4.13750768e-01 -3.42427254e-01
5.46992362e-01 1.60404038e+00 9.31250095e-01 8.42331767e-01
-1.01068759e+00 -1.10013354e+00 -9.33798496e-03 -4.59415793e-01
-1.01889944e+00 -1.79037929e-01 7.48008907e-01 -1.09873176e+00
5.32274306e-01 5.58749676e-01 7.90783823e-01 1.68731415e+00
8.90958786e-01 3.29378277e-01 1.15361285e+00 -6.72091171e-02
5.61687760e-02 8.27969074e-01 4.10787165e-01 4.92043138e-01
3.70969325e-01 -3.04029286e-01 -3.29459488e-01 -8.11747372e-01
-8.82073715e-02 1.16673544e-01 -1.87011376e-01 6.49851918e-01
-5.21899104e-01 1.26772392e+00 6.04962230e-01 5.84702849e-01
1.48037700e-02 -2.98684806e-01 9.96864080e-01 4.45728213e-01
1.01214290e+00 6.51878297e-01 2.00413428e-02 -8.54421481e-02
-3.42285663e-01 3.66549790e-02 1.07932389e+00 1.45619765e-01
3.80017698e-01 1.21635005e-01 -2.39111871e-01 1.08354390e+00
3.50612372e-01 5.08344233e-01 5.05348504e-01 -5.13400972e-01
2.02226192e-01 9.22211528e-01 9.91748646e-02 -1.88216150e+00
-4.43118155e-01 -3.76512706e-01 -5.94868779e-01 2.36733764e-01
-4.89842519e-03 -8.07787001e-01 -3.07404667e-01 1.23791182e+00
5.21682240e-02 -4.05154824e-01 -3.27404499e-01 7.20244527e-01
9.10329938e-01 7.17747808e-01 4.11005080e-01 -1.84153482e-01
9.10276413e-01 -5.15661716e-01 -1.23154104e+00 8.07372034e-02
1.15819895e+00 -8.21514189e-01 9.23352003e-01 5.43693900e-01
-7.49379039e-01 -1.20651789e-01 -9.42160428e-01 -1.04580782e-01
-1.03862476e+00 -4.83149320e-01 2.97989964e-01 1.22790682e+00
-7.37695694e-01 6.34847283e-01 3.15890796e-02 -3.67712200e-01
4.17009383e-01 2.67266124e-01 -3.20862859e-01 6.04500651e-01
-1.42382228e+00 1.02587867e+00 1.81302667e-01 6.39514551e-02
-2.54665464e-01 -4.40657079e-01 -7.03785002e-01 -3.33770484e-01
4.55358066e-03 5.67685366e-02 1.06550813e+00 -1.60783970e+00
-1.21675670e+00 1.27089250e+00 4.47144836e-01 -2.87863761e-01
6.19661152e-01 -5.31057119e-01 -8.52740347e-01 -3.24993789e-01
1.16014928e-01 1.99253678e-01 9.07506168e-01 -1.19299984e+00
1.29903331e-01 -5.31282604e-01 1.12916090e-01 -2.69590616e-01
-9.43763793e-01 7.22476900e-01 8.72274339e-01 -4.96412635e-01
-4.62986976e-01 -9.97329354e-01 2.28494793e-01 -2.89741993e-01
-1.20540333e+00 -3.49590003e-01 1.23196578e+00 -8.22826624e-01
1.41513014e+00 -2.25074434e+00 -5.46938717e-01 -3.38475183e-02
5.11701286e-01 8.43397319e-01 3.61524701e-01 8.38072896e-01
-2.15282813e-02 1.23706591e+00 3.83959055e-01 2.36313745e-01
-1.61483869e-01 -2.07870051e-01 -3.28948855e-01 6.68950975e-01
-1.95982561e-01 6.51372075e-01 -9.16573882e-01 -4.01259184e-01
-2.42372572e-01 3.10088456e-01 -3.99210811e-01 -1.02712438e-01
1.51688501e-01 4.10072833e-01 -8.09719205e-01 3.51380974e-01
8.65992367e-01 -2.90668495e-02 3.41390483e-02 3.56321365e-01
-5.26872337e-01 3.86076242e-01 -1.11075431e-01 4.54888076e-01
-3.51989865e-01 1.07469451e+00 2.37770677e-01 -2.50465184e-01
1.13871908e+00 1.85036987e-01 -5.67940325e-02 -2.74240643e-01
1.06931460e+00 1.67872101e-01 1.60071731e-01 -9.53321636e-01
7.08488703e-01 -3.14703941e-01 -3.38930994e-01 8.42595816e-01
-5.31251729e-01 3.72042954e-01 -3.10781598e-01 2.80488521e-01
6.75346911e-01 -3.78198951e-01 3.35750103e-01 -3.68169665e-01
6.51827037e-01 -3.52866024e-01 2.85112828e-01 4.25525188e-01
-6.14771545e-01 9.98549908e-02 8.28218937e-01 -5.88373363e-01
-6.93829954e-01 -6.44394636e-01 2.68623441e-01 1.34896708e+00
1.44715622e-01 -1.11000404e-01 -9.51080501e-01 -8.54337692e-01
1.14108205e-01 8.50494802e-01 -6.07062936e-01 -6.47339284e-01
-1.56350851e-01 -3.21429133e-01 7.74304509e-01 -1.67270944e-01
9.42756057e-01 -9.49467182e-01 -2.26216748e-01 -4.53740992e-02
-1.18346676e-01 -6.47819877e-01 -5.54984689e-01 -2.75423437e-01
-3.00284177e-01 -1.12175345e+00 -3.79704177e-01 -3.80850196e-01
4.98363286e-01 3.07776749e-01 3.65674138e-01 5.97655952e-01
1.76299572e-01 -1.18116312e-01 -5.24209976e-01 -7.61573970e-01
-8.01909089e-01 2.32459053e-01 2.27411408e-02 3.58822972e-01
6.90049946e-01 -1.69831991e-01 -1.70259222e-01 -5.42225316e-02
-9.27911818e-01 -3.65290254e-01 -2.74777830e-01 2.80385226e-01
-9.25001502e-01 -6.67594850e-01 6.40458584e-01 -1.45082796e+00
1.36941361e+00 -7.88270235e-01 1.48349136e-01 -1.76943406e-01
-2.37315878e-01 -6.30249023e-01 9.00937796e-01 -7.53604352e-01
-1.00776839e+00 -6.67057395e-01 -2.12557971e-01 9.69899446e-02
-2.29561254e-01 2.77280986e-01 2.63682961e-01 -4.30862010e-01
9.64016616e-01 -2.25124449e-01 6.39079930e-03 -2.78318912e-01
-2.05948010e-01 1.34369445e+00 -2.95133948e-01 1.54321864e-01
8.30837190e-01 5.39142132e-01 -5.87113142e-01 -1.10288250e+00
-1.35902572e+00 -4.22350943e-01 -2.18375564e-01 -8.09812903e-01
1.00717974e+00 -4.24211740e-01 -1.25106859e+00 5.25506973e-01
-1.52630103e+00 1.68641523e-01 5.86386621e-01 1.05991490e-01
3.15111309e-01 6.77776873e-01 -1.23722172e+00 -1.28082991e+00
-5.18897176e-01 -2.09003150e-01 2.57971436e-01 3.04223001e-01
-9.45779324e-01 -1.15028048e+00 1.27428845e-01 6.70161664e-01
5.78604400e-01 7.39737332e-01 7.16479778e-01 -1.25688219e+00
3.81069899e-01 -4.53605652e-01 -1.14117134e-02 7.30659723e-01
1.87370852e-01 3.62295598e-01 -9.32309628e-01 -9.71594751e-02
3.64454985e-01 -6.63493097e-01 4.20083642e-01 -3.22028399e-01
8.40134859e-01 -1.16864824e+00 -2.25461856e-01 -2.72605871e-03
9.80981350e-01 3.52401346e-01 7.95644343e-01 4.23151761e-01
5.75292110e-01 8.44184279e-01 3.62291485e-01 4.87028062e-01
-2.64427066e-01 -2.24750996e-01 6.52533889e-01 -2.71252859e-02
5.81327438e-01 -3.71596396e-01 6.75941825e-01 8.36184204e-01
3.01136523e-01 -4.61077064e-01 -9.12531793e-01 1.72692582e-01
-1.37564874e+00 -1.04395866e+00 -7.20481873e-01 1.64228928e+00
4.27695274e-01 3.44427049e-01 4.34822105e-02 1.96838334e-01
1.36163855e+00 5.18160820e-01 -2.44297221e-01 -1.68708682e+00
2.28161782e-01 -4.62567449e-01 5.83528817e-01 5.98949730e-01
-1.09141302e+00 8.38839054e-01 6.27928925e+00 3.88160676e-01
-1.66037154e+00 3.34414989e-01 8.76375377e-01 8.28517973e-02
-2.57192343e-01 -3.78056377e-01 -5.51970363e-01 6.83299005e-01
1.01793897e+00 -2.09369317e-01 8.70844573e-02 9.97365952e-01
6.80113733e-01 7.34371841e-02 -4.60986525e-01 7.76584327e-01
6.07695103e-01 -1.17933631e+00 2.20415574e-02 5.16853556e-02
7.02827990e-01 -1.89560965e-01 3.57852072e-01 2.88936973e-01
3.58069807e-01 -1.00143039e+00 5.72935045e-01 1.99041411e-01
3.21308166e-01 -5.90229928e-01 8.63076031e-01 5.60523450e-01
4.38151369e-03 -4.62650299e-01 -2.04175517e-01 -8.12920570e-01
2.82896362e-04 4.65103447e-01 -6.11323237e-01 -1.05861440e-01
5.00747204e-01 8.47412825e-01 -5.02397299e-01 5.57056665e-01
-2.56522715e-01 6.67088091e-01 2.62317538e-01 -8.05387497e-01
5.27470529e-01 -2.45067403e-01 5.94289720e-01 1.17040110e+00
-8.26443955e-02 1.54816378e-02 7.59348795e-02 9.80752170e-01
-7.85891637e-02 2.98575461e-01 -1.07562447e+00 -6.09721303e-01
5.84233850e-02 1.19693446e+00 -6.33742690e-01 2.80982107e-02
-1.84814811e-01 7.27416039e-01 2.13303134e-01 2.81697601e-01
-7.75375783e-01 -5.26296794e-01 3.77733767e-01 5.80377460e-01
-6.56724751e-01 -8.39921460e-02 -5.22054613e-01 -8.01444590e-01
-3.93693745e-01 -7.92337477e-01 2.75887728e-01 -8.33328307e-01
-1.49446917e+00 3.88117403e-01 -5.83893418e-01 -1.01103878e+00
3.67851615e-01 -5.87089598e-01 -8.47078145e-01 6.01264954e-01
-8.83024216e-01 -8.71571124e-01 -1.10979434e-02 1.57941028e-01
4.09645170e-01 2.01322585e-02 4.99068797e-01 2.99127132e-01
-6.72080934e-01 3.82714182e-01 -2.87305564e-01 5.18568754e-01
7.60531306e-01 -6.19162381e-01 -3.77240807e-01 8.41741338e-02
-8.75931144e-01 7.52740085e-01 7.45526671e-01 -9.17520583e-01
-5.79894423e-01 -1.02289057e+00 1.39439499e+00 -4.17766273e-01
1.08808243e+00 -3.12373281e-01 -7.17773795e-01 7.19097018e-01
7.05053866e-01 -7.18970716e-01 1.39436352e+00 1.12857081e-01
-5.15317440e-01 2.93934107e-01 -1.32233417e+00 1.08410001e+00
6.46811843e-01 -7.37240195e-01 -5.88290274e-01 8.37991714e-01
6.33303285e-01 2.71448344e-01 -3.13901097e-01 -6.96385562e-01
3.78582805e-01 -1.19044650e+00 2.52715021e-01 -1.13592720e+00
9.07138586e-01 4.00168687e-01 4.81617957e-01 -1.18793750e+00
-5.04625916e-01 -9.66705263e-01 5.97051561e-01 1.33098435e+00
4.21455115e-01 -1.01286304e+00 4.87751305e-01 7.10322022e-01
-9.77093354e-02 -5.04422151e-02 -4.39130276e-01 -5.19201815e-01
4.90221769e-01 1.40754893e-01 1.24567404e-01 1.91168928e+00
7.21898973e-01 6.11847460e-01 -6.61404192e-01 -1.60357803e-01
3.58149484e-02 -5.06618857e-01 5.29296041e-01 -1.50195551e+00
6.81924343e-01 -3.74638319e-01 -1.16578668e-01 -5.16908705e-01
4.18617517e-01 -7.82631874e-01 -4.56618994e-01 -1.04392958e+00
1.07308626e-01 2.63393641e-01 -2.36662738e-02 2.62951195e-01
3.91009808e-01 4.38628465e-01 3.34987819e-01 1.19765505e-01
-6.32573664e-01 2.20019251e-01 1.29611194e+00 -3.04787040e-01
-2.46906921e-01 -2.45359406e-01 -1.14084244e+00 1.11714566e+00
1.08372605e+00 -7.01646745e-01 2.10739337e-02 3.85220572e-02
9.65351760e-01 -2.06774682e-01 2.49193341e-01 -5.23166776e-01
-1.50498807e-01 -4.53789115e-01 1.38560878e-02 -2.05279574e-01
1.39780641e-01 -6.02978230e-01 -3.01373243e-01 1.09773374e+00
-9.42889929e-01 4.53999564e-02 -1.03042647e-01 3.41533244e-01
-1.09604076e-01 -3.87392193e-01 1.16286457e+00 -3.94035220e-01
2.02124715e-01 -2.17060611e-01 -1.20924795e+00 8.34836736e-02
1.21438348e+00 -3.43797863e-01 -8.55920017e-01 -1.08632123e+00
-8.15867722e-01 9.32055563e-02 2.88387835e-01 6.80966914e-01
3.31026912e-01 -1.13096952e+00 -8.30863535e-01 -3.22973967e-01
-1.68425366e-02 -1.04699898e+00 2.44130507e-01 7.06125081e-01
-7.43249953e-01 1.99101955e-01 -2.85520285e-01 2.33278647e-02
-1.45940101e+00 6.91952169e-01 3.10532033e-01 1.10498868e-01
-2.32443064e-01 5.23278117e-01 -5.00108004e-02 -4.25466388e-01
-3.05249728e-03 1.92017332e-01 -7.08159804e-01 2.80146092e-01
8.29588711e-01 7.85513461e-01 -3.42800170e-01 -1.05924726e+00
-2.38021597e-01 -2.78569479e-03 -4.29374009e-01 4.66524661e-01
6.90320671e-01 -6.18106611e-02 -6.64402604e-01 7.62405396e-01
1.63872325e+00 3.51466566e-01 -2.55488396e-01 4.45416301e-01
7.24047720e-02 -6.84786022e-01 -2.19390139e-01 -9.04992282e-01
-7.83398390e-01 8.39441299e-01 8.83437395e-02 1.52238905e+00
1.93987980e-01 -4.63002592e-01 9.55718756e-01 4.31868732e-01
-1.28168032e-01 -1.46477532e+00 5.13947666e-01 7.32296288e-01
1.22117531e+00 -9.63895977e-01 -1.55382514e-01 -6.66131496e-01
-7.52860606e-01 1.19150507e+00 7.46688008e-01 -3.66189331e-01
8.56280982e-01 6.22381344e-02 4.53439891e-01 -2.84472555e-01
-5.24561226e-01 6.51642084e-01 -2.23128632e-01 5.75113595e-01
7.44069815e-01 5.17796837e-02 -1.06171119e+00 3.85396898e-01
-3.00104946e-01 -2.42120042e-01 1.28943074e+00 6.16279423e-01
-7.14047909e-01 -5.44067025e-01 -7.32745409e-01 4.76222783e-01
-1.06166017e+00 1.12814248e-01 -1.65543544e+00 6.70352876e-01
4.09104466e-01 1.51886368e+00 -1.65953994e-01 -7.99522460e-01
6.91553066e-03 1.78856328e-01 -7.09855855e-01 -7.76480794e-01
-1.60045838e+00 -3.07208717e-01 6.41473651e-01 -1.33066475e-01
-2.31385455e-01 -2.25629479e-01 -9.00199473e-01 -1.09636712e+00
-4.60540950e-01 3.09554100e-01 5.68580866e-01 8.35362792e-01
3.37817729e-01 9.11248103e-02 8.56985092e-01 -3.44080895e-01
-5.76038301e-01 -1.24421453e+00 -6.03663564e-01 7.21923232e-01
5.84504426e-01 -4.39536929e-01 -8.53817344e-01 -4.48459923e-01] | [8.695298194885254, 10.57834243774414] |
ce356143-f2b1-43d2-9f91-afdd180b73fe | rapid-rl-a-reconfigurable-architecture-with | 2109.08231 | null | https://arxiv.org/abs/2109.08231v1 | https://arxiv.org/pdf/2109.08231v1.pdf | RAPID-RL: A Reconfigurable Architecture with Preemptive-Exits for Efficient Deep-Reinforcement Learning | Present-day Deep Reinforcement Learning (RL) systems show great promise towards building intelligent agents surpassing human-level performance. However, the computational complexity associated with the underlying deep neural networks (DNNs) leads to power-hungry implementations. This makes deep RL systems unsuitable for deployment on resource-constrained edge devices. To address this challenge, we propose a reconfigurable architecture with preemptive exits for efficient deep RL (RAPID-RL). RAPID-RL enables conditional activation of DNN layers based on the difficulty level of inputs. This allows to dynamically adjust the compute effort during inference while maintaining competitive performance. We achieve this by augmenting a deep Q-network (DQN) with side-branches capable of generating intermediate predictions along with an associated confidence score. We also propose a novel training methodology for learning the actions and branch confidence scores in a dynamic RL setting. Our experiments evaluate the proposed framework for Atari 2600 gaming tasks and a realistic Drone navigation task on an open-source drone simulator (PEDRA). We show that RAPID-RL incurs 0.34x (0.25x) number of operations (OPS) while maintaining performance above 0.88x (0.91x) on Atari (Drone navigation) tasks, compared to a baseline-DQN without any side-branches. The reduction in OPS leads to fast and efficient inference, proving to be highly beneficial for the resource-constrained edge where making quick decisions with minimal compute is essential. | ['Kaushik Roy', 'Arijit Raychowdhury', 'Priyadarshini Panda', 'Malik Aqeel Anwar', 'Adarsh Kumar Kosta'] | 2021-09-16 | null | null | null | null | ['drone-navigation'] | ['computer-vision'] | [-1.55448750e-01 1.02163538e-01 -2.56139547e-01 -2.12126732e-01
-3.37693810e-01 -5.39202213e-01 4.57421869e-01 -3.32835734e-01
-9.09014344e-01 8.33444834e-01 -4.60837930e-01 -7.20824420e-01
-7.38873286e-03 -1.04213238e+00 -9.20539439e-01 -6.55070961e-01
-3.55573803e-01 5.20850003e-01 3.62621307e-01 -4.68067616e-01
8.82058740e-02 6.40670896e-01 -1.82184958e+00 -1.08947232e-02
7.83259988e-01 1.21976948e+00 4.08931106e-01 1.02808607e+00
3.49762112e-01 1.04380095e+00 -7.04469085e-01 -1.78041935e-01
6.49841130e-01 -4.71505672e-02 -3.55885357e-01 -5.28079748e-01
2.21823618e-01 -8.51294875e-01 -4.11686122e-01 7.01024115e-01
6.35145366e-01 2.69950628e-01 3.11916798e-01 -1.44392085e+00
2.28325188e-01 5.28662264e-01 -4.09410179e-01 3.64261776e-01
-3.04358989e-01 4.85350758e-01 9.67574298e-01 -2.94392079e-01
2.91261971e-01 9.70685661e-01 4.52244699e-01 6.82568848e-01
-8.07901382e-01 -8.07123840e-01 1.48286387e-01 1.67517483e-01
-1.14911354e+00 -5.02671599e-01 2.93458223e-01 7.92997107e-02
1.47888374e+00 -1.76964954e-01 6.80258095e-01 1.07402623e+00
5.03211200e-01 7.06878543e-01 8.37682366e-01 8.25638045e-03
8.70007336e-01 -3.06744218e-01 -4.30478096e-01 1.05217993e+00
3.51340055e-01 4.39291567e-01 -4.04654384e-01 1.29837617e-01
1.00625527e+00 -1.43686295e-01 2.29065225e-01 -2.31045067e-01
-7.44537175e-01 7.87358105e-01 6.61830842e-01 -1.86867997e-01
-5.75243890e-01 8.77944469e-01 5.58910847e-01 4.06164676e-01
-3.23874176e-01 4.65043277e-01 -7.31912315e-01 -5.54022014e-01
-6.56267166e-01 5.26643157e-01 7.48615742e-01 1.00197077e+00
7.06504881e-01 6.22954428e-01 -6.11872971e-02 5.07753432e-01
1.49732381e-01 5.38391232e-01 3.58744383e-01 -1.25074065e+00
5.61103582e-01 4.52422738e-01 1.29407510e-01 -5.94380558e-01
-8.31834018e-01 -9.20669556e-01 -8.30181062e-01 7.75412261e-01
1.25994608e-01 -6.40141428e-01 -9.65692043e-01 1.81748605e+00
2.18454584e-01 1.94477305e-01 3.52351189e-01 8.91224504e-01
3.11476856e-01 8.37593734e-01 6.39354587e-02 2.11979419e-01
1.41281497e+00 -1.02748489e+00 -1.75386831e-01 -4.30838794e-01
6.31488979e-01 -8.74413401e-02 1.09118319e+00 8.75707924e-01
-1.09966302e+00 -6.80655718e-01 -1.54818571e+00 3.13140787e-02
-1.86445266e-01 2.74024516e-01 8.64101410e-01 8.60166252e-01
-1.23730826e+00 4.30872679e-01 -1.09680128e+00 2.00016841e-01
3.28723758e-01 1.05644751e+00 2.06780538e-01 1.66870221e-01
-1.08131623e+00 7.24002481e-01 4.57513720e-01 9.78039876e-02
-1.52188969e+00 -6.82795703e-01 -5.76836288e-01 3.14173907e-01
7.32917964e-01 -7.06516743e-01 1.53810096e+00 -9.45086122e-01
-2.23970318e+00 1.59093499e-01 4.51338768e-01 -1.11945629e+00
5.24563134e-01 -2.41264224e-01 -3.37709963e-01 2.23149702e-01
-3.63355875e-01 1.20494294e+00 9.03978944e-01 -8.26527953e-01
-1.12057281e+00 -2.31645226e-01 7.41662741e-01 3.42661113e-01
-1.33571252e-01 -6.00837171e-01 -3.89453679e-01 -2.21875697e-01
-4.73437220e-01 -1.26835477e+00 -4.82441604e-01 -1.73807174e-01
2.60472655e-01 -1.52469382e-01 6.23884141e-01 -2.24982962e-01
9.42848206e-01 -1.82481301e+00 -1.22341342e-01 8.77652168e-02
2.19745830e-01 4.81649160e-01 -2.04644635e-01 -2.58532614e-02
6.12535715e-01 -3.15928698e-01 -3.80353332e-02 2.25154907e-02
3.40475105e-02 5.48547089e-01 -1.27862230e-01 1.98380753e-01
1.68132603e-01 9.43137228e-01 -8.74145210e-01 -1.10486887e-01
1.01130284e-01 3.25597584e-01 -1.12746823e+00 2.31846452e-01
-5.29946089e-01 8.99242833e-02 -3.87337446e-01 5.75726986e-01
4.01783079e-01 -1.09079974e-02 4.23643023e-01 -7.11245388e-02
-2.20690534e-01 4.11690503e-01 -1.01805675e+00 1.63176203e+00
-1.05345416e+00 5.96498549e-01 2.60651886e-01 -8.59234154e-01
1.01392400e+00 -9.00067668e-03 1.73426822e-01 -1.32791591e+00
3.70551109e-01 2.42944866e-01 3.75507742e-01 -1.03608156e-02
7.55203068e-01 1.86855599e-01 -3.32862496e-01 1.58347458e-01
1.30173743e-01 6.18611462e-02 2.17728242e-01 -1.65888429e-01
1.52284622e+00 3.05060536e-01 7.37459436e-02 -3.53643209e-01
4.79589403e-01 -6.58714399e-02 6.81177258e-01 8.78475964e-01
-2.85707384e-01 -3.81557941e-01 6.37118518e-01 -7.29241252e-01
-1.02877569e+00 -1.05674839e+00 2.71504641e-01 1.38623297e+00
1.38284460e-01 -2.52319872e-01 -6.85835063e-01 -5.79114616e-01
-1.75726235e-01 8.10938835e-01 -3.97889197e-01 -2.50378698e-01
-7.46352732e-01 -7.52063036e-01 8.05503428e-01 7.98778534e-01
9.24163997e-01 -1.00625598e+00 -1.56801653e+00 3.86095494e-01
3.79806995e-01 -1.01760268e+00 1.17218040e-01 6.37002707e-01
-1.04903316e+00 -5.35499990e-01 -3.31779033e-01 -5.73697746e-01
4.78586704e-01 3.01951580e-02 1.34616780e+00 -5.82897104e-02
-1.33051917e-01 -3.39256003e-02 -1.88964322e-01 -1.28047660e-01
-2.84345388e-01 4.33524042e-01 6.46092072e-02 -6.12204492e-01
7.98956975e-02 -4.89478558e-01 -9.19573605e-01 3.86968374e-01
-8.87832642e-01 1.20909631e-01 9.84541237e-01 9.70586777e-01
4.70962137e-01 1.16916381e-01 7.68150032e-01 -5.49375951e-01
4.78230268e-01 -3.68823946e-01 -1.38612604e+00 -2.24961266e-02
-5.91597676e-01 5.20428836e-01 1.09249461e+00 -2.66345978e-01
-1.02031171e+00 6.52677789e-02 -4.36634779e-01 -2.73799393e-02
1.78856850e-01 7.44949132e-02 4.41821367e-02 -1.10257000e-01
6.13996983e-01 -9.45407376e-02 -6.54660165e-02 2.30493799e-01
1.65098071e-01 5.22747517e-01 5.03195405e-01 -6.73950613e-01
2.38867104e-01 2.10842639e-01 3.37201953e-01 -5.58457434e-01
-3.27238649e-01 1.94732025e-01 2.62068715e-02 -2.24967301e-01
5.42963505e-01 -1.27258337e+00 -1.51883411e+00 6.93179965e-02
-6.38755023e-01 -1.00592685e+00 -5.58080785e-02 2.34308749e-01
-7.44573593e-01 -3.34653616e-01 -5.86969674e-01 -7.68840969e-01
-6.77227199e-01 -1.42823410e+00 7.47766972e-01 5.76468527e-01
1.27793938e-01 -5.27765870e-01 -1.40015021e-01 2.56987959e-01
6.19322300e-01 -1.87529046e-02 8.21840167e-01 -1.26636237e-01
-8.34239483e-01 1.03494093e-01 -9.12241861e-02 2.38854006e-01
-4.47126180e-01 -2.23657012e-01 -8.60976398e-01 -4.92784113e-01
-4.83094156e-01 -6.21420503e-01 6.17937982e-01 3.71903867e-01
1.23486364e+00 -1.16493098e-01 1.98262278e-02 7.72690833e-01
1.66476262e+00 5.78436911e-01 6.11634493e-01 4.72955495e-01
3.32319826e-01 -9.76761281e-02 7.95622647e-01 7.99078345e-01
3.52751702e-01 7.61514366e-01 9.12337780e-01 1.08135574e-01
2.08051309e-01 -2.45680556e-01 7.36494780e-01 3.48881066e-01
-1.60311505e-01 -5.31065524e-01 -8.59978318e-01 2.80557543e-01
-1.75904524e+00 -4.69474196e-01 4.73563582e-01 2.19131827e+00
5.09694636e-01 6.87871814e-01 2.73205370e-01 -8.19009379e-04
5.25651872e-02 -1.61151648e-01 -1.08414316e+00 -9.43758607e-01
3.71070921e-01 6.59107983e-01 1.01069963e+00 3.79620612e-01
-6.86937749e-01 1.17114639e+00 5.25477266e+00 7.67814219e-01
-1.17912829e+00 -4.98310216e-02 6.70748472e-01 -3.48078966e-01
6.12098798e-02 -3.55753243e-01 -1.11567962e+00 2.17401147e-01
1.58071423e+00 2.30210900e-01 7.46157110e-01 1.26506341e+00
5.46737276e-02 -3.70745212e-01 -8.99069309e-01 9.33098614e-01
-2.92667329e-01 -1.33363521e+00 -6.54211193e-02 7.64966384e-02
6.02353573e-01 4.20540154e-01 1.11655086e-01 9.09797847e-01
7.36133754e-01 -9.41001415e-01 6.22742176e-01 6.72311261e-02
8.25643957e-01 -1.52361381e+00 8.30868244e-01 2.91780561e-01
-9.60524082e-01 -5.40119946e-01 -6.61288738e-01 -2.89331377e-01
-1.76404431e-01 8.07924047e-02 -1.16314793e+00 1.50590524e-01
7.81774163e-01 1.67388067e-01 -2.76925951e-01 5.09214461e-01
-3.54438394e-01 4.88855273e-01 -3.72988790e-01 -4.52462554e-01
6.52245939e-01 -5.28310351e-02 8.61466527e-02 8.99112105e-01
2.21022710e-01 1.75892457e-01 -8.20061639e-02 5.21783054e-01
-2.70679593e-01 -3.90922248e-01 -2.97713131e-01 6.13549240e-02
5.20909667e-01 1.36269510e+00 -9.24805880e-01 -3.84690046e-01
-1.07363246e-01 9.41115320e-01 4.34230238e-01 -2.06796527e-02
-1.22680199e+00 -3.91714334e-01 1.00046611e+00 -8.36016238e-02
4.38380241e-01 -5.06661654e-01 -1.99225381e-01 -6.71230137e-01
-3.45616907e-01 -9.59560394e-01 8.40949342e-02 -4.76768196e-01
-5.18227220e-01 8.82213593e-01 -3.42816979e-01 -9.76592243e-01
-5.44467032e-01 -9.16254163e-01 -1.12111390e-01 3.13068181e-01
-1.69295347e+00 -5.56088388e-01 -4.31445867e-01 4.41458941e-01
6.08960807e-01 -4.09055948e-01 7.26824880e-01 4.13497180e-01
-6.85768247e-01 8.49594533e-01 -1.99890733e-01 -1.39280245e-01
1.54024377e-01 -1.16324282e+00 7.19757795e-01 6.76933467e-01
-2.49140739e-01 3.20903301e-01 6.15398347e-01 -3.57640177e-01
-1.88547993e+00 -9.76920784e-01 -3.78065221e-02 1.61930770e-01
3.41646880e-01 -5.14213860e-01 -2.48505443e-01 3.41758966e-01
2.48828992e-01 2.00485270e-02 4.63436067e-01 -6.23791777e-02
-3.08581907e-02 -6.68826044e-01 -1.19135380e+00 1.03450358e+00
1.14808691e+00 1.64560284e-02 -1.21203385e-01 -1.54710576e-01
6.14684284e-01 -5.58033109e-01 -5.52398086e-01 3.52499455e-01
7.34196782e-01 -1.16912079e+00 8.20391178e-01 -3.30899388e-01
4.44813788e-01 -3.93575966e-01 -9.07726511e-02 -1.18295312e+00
-1.80950448e-01 -7.48488486e-01 -2.78313845e-01 3.55187058e-01
3.67985934e-01 -5.68814099e-01 1.20456207e+00 3.83344740e-01
-3.78019124e-01 -9.29357648e-01 -8.15450668e-01 -7.23725796e-01
-1.98897243e-01 -4.41160113e-01 5.45172453e-01 1.49300322e-01
-3.91860813e-01 3.59810680e-01 -2.84797758e-01 4.25866991e-01
3.32291871e-01 -1.07072406e-01 9.05752063e-01 -9.08990741e-01
-6.78712606e-01 -3.48099470e-01 -3.51950198e-01 -1.36023772e+00
1.23689650e-02 -4.64724600e-01 2.87775934e-01 -1.13191962e+00
-4.47811425e-01 -8.20620894e-01 -3.53396147e-01 6.08203113e-01
2.67055601e-01 1.03350013e-01 1.14005841e-01 -3.15117806e-01
-7.68180490e-01 5.25928438e-01 9.05011356e-01 1.63152605e-01
-4.06520039e-01 3.31912041e-02 -3.33824724e-01 6.15833580e-01
1.16029346e+00 -2.41406217e-01 -9.18803155e-01 -6.13176286e-01
8.04850459e-01 4.23331320e-01 1.74918875e-01 -1.62937963e+00
2.40502000e-01 -8.55642371e-03 3.84821087e-01 -6.13541245e-01
3.37298721e-01 -8.42452526e-01 -6.70959204e-02 9.97500181e-01
-1.82670712e-01 5.91335714e-01 5.64065158e-01 4.28579271e-01
1.17636465e-01 -9.20636430e-02 8.62178087e-01 1.71569567e-02
-1.02335238e+00 1.87038064e-01 -6.97861850e-01 -5.44876792e-02
1.10637760e+00 -1.02322541e-01 -2.27287486e-01 -1.41816407e-01
-1.84819326e-01 3.95298213e-01 1.86992124e-01 3.12682062e-01
5.96584797e-01 -8.23725879e-01 -2.16320261e-01 4.48976308e-01
-2.38018334e-01 4.75116856e-02 3.76635253e-01 4.35188055e-01
-1.13676488e+00 5.27759314e-01 -9.24151182e-01 -3.65351826e-01
-7.78605461e-01 3.25158238e-01 4.90209788e-01 -5.78572273e-01
-4.86768365e-01 8.80203545e-01 -1.61297731e-02 -2.84440190e-01
3.50652426e-01 -6.73761785e-01 3.48638445e-02 -2.18613818e-01
4.41438973e-01 6.06115222e-01 2.92835295e-01 2.22424105e-01
-3.31221342e-01 1.10868877e-03 -1.55270755e-01 -2.46720195e-01
1.29954827e+00 1.88686326e-01 3.89657378e-01 -1.56668350e-01
6.71417236e-01 -2.09292695e-01 -1.85648870e+00 2.33651459e-01
-1.23297200e-01 1.08304903e-01 6.37219191e-01 -1.02520442e+00
-1.36923218e+00 7.04621136e-01 7.84401774e-01 -3.26631755e-01
1.36876929e+00 -6.95382953e-01 9.51612890e-01 9.43575799e-01
9.30087030e-01 -1.24472749e+00 1.08972609e-01 6.53484225e-01
3.62594306e-01 -9.93185759e-01 -8.81886259e-02 1.69182345e-01
-6.46657526e-01 1.14882481e+00 9.80921805e-01 -4.79876369e-01
2.95031995e-01 7.13505805e-01 -1.89756393e-01 6.69780821e-02
-1.23948443e+00 -2.09972501e-01 -4.23537403e-01 4.63493258e-01
-1.05489396e-01 2.65904337e-01 -1.11692652e-01 4.16923285e-01
-2.17839390e-01 1.15443729e-01 5.76033533e-01 9.80773926e-01
-5.28774083e-01 -9.77262795e-01 1.54550284e-01 3.85504663e-01
-2.43927449e-01 2.40586000e-03 2.49698967e-01 8.23679745e-01
3.31895620e-01 7.73136139e-01 4.52518910e-01 -4.99219179e-01
1.63668513e-01 -3.43447179e-01 4.84647006e-01 -2.11452529e-01
-8.35951507e-01 -2.67476231e-01 1.98379233e-01 -8.20816934e-01
7.59143522e-03 -1.67340979e-01 -1.56886435e+00 -4.83617693e-01
1.25813514e-01 -1.07384317e-01 8.62674534e-01 5.88203251e-01
6.86015069e-01 8.65967512e-01 3.46016049e-01 -7.95498371e-01
-6.28010988e-01 -4.48901236e-01 -3.07097644e-01 -5.52970707e-01
2.86883146e-01 -8.41887236e-01 6.36756942e-02 -3.97323459e-01] | [3.8459224700927734, 1.4765948057174683] |
6815cc5b-10e9-4f64-bd9e-88ca9ad7af8b | multimodal-engagement-analysis-from-facial | 2101.04215 | null | https://arxiv.org/abs/2101.04215v2 | https://arxiv.org/pdf/2101.04215v2.pdf | Multimodal Engagement Analysis from Facial Videos in the Classroom | Student engagement is a key construct for learning and teaching. While most of the literature explored the student engagement analysis on computer-based settings, this paper extends that focus to classroom instruction. To best examine student visual engagement in the classroom, we conducted a study utilizing the audiovisual recordings of classes at a secondary school over one and a half month's time, acquired continuous engagement labeling per student (N=15) in repeated sessions, and explored computer vision methods to classify engagement levels from faces in the classroom. We trained deep embeddings for attentional and emotional features, training Attention-Net for head pose estimation and Affect-Net for facial expression recognition. We additionally trained different engagement classifiers, consisting of Support Vector Machines, Random Forest, Multilayer Perceptron, and Long Short-Term Memory, for both features. The best performing engagement classifiers achieved AUCs of .620 and .720 in Grades 8 and 12, respectively. We further investigated fusion strategies and found score-level fusion either improves the engagement classifiers or is on par with the best performing modality. We also investigated the effect of personalization and found that using only 60-seconds of person-specific data selected by margin uncertainty of the base classifier yielded an average AUC improvement of .084. 4.Our main aim with this work is to provide the technical means to facilitate the manual data analysis of classroom videos in research on teaching quality and in the context of teacher training. | ['Enkelejda Kasneci', 'Ulrich Trautwein', 'Peter Gerjets', "Sidney D'Mello", 'Patricia Goldberg', 'Ömer Sümer'] | 2021-01-11 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [ 3.91207188e-02 3.03820729e-01 -1.79453529e-02 -5.34523249e-01
-8.35576892e-01 -5.06619513e-01 1.06576972e-01 6.61715090e-01
-4.74342555e-01 1.71368316e-01 1.71322942e-01 -2.78232187e-01
-4.57557201e-01 -4.13717628e-01 -6.42927587e-01 -6.40856206e-01
9.99794975e-02 -1.41960412e-01 -1.33342773e-01 1.19837530e-01
3.66102636e-01 9.61687624e-01 -2.36233664e+00 3.75070304e-01
8.03096414e-01 8.17457199e-01 -3.55366915e-01 8.79573226e-01
-9.97226834e-02 8.43354940e-01 -7.02990770e-01 -3.78598392e-01
-2.38979310e-01 -1.27113611e-01 -6.25548899e-01 4.81655076e-02
1.20356989e+00 -3.79475743e-01 -1.56289503e-01 7.72290826e-01
8.36956084e-01 7.70575821e-01 3.75657469e-01 -1.43959081e+00
-4.04825985e-01 6.20560301e-03 -5.09533405e-01 4.52487022e-01
7.57653296e-01 1.66037396e-01 4.14738327e-01 -9.74156737e-01
1.55609012e-01 1.06039786e+00 4.71337914e-01 4.95322764e-01
-1.03234935e+00 -1.00775039e+00 1.93869725e-01 4.85586047e-01
-1.07668209e+00 -4.75653708e-01 5.59653044e-01 -7.56752551e-01
8.61931980e-01 1.81704566e-01 1.14943838e+00 1.16089380e+00
7.76304305e-02 7.12462604e-01 1.54972470e+00 -7.39726365e-01
-6.17326908e-02 5.98881304e-01 7.50326931e-01 6.81704700e-01
-6.88165948e-02 -6.31209870e-04 -8.93414140e-01 1.28572449e-01
2.56021529e-01 -5.57298446e-03 -1.39843032e-01 -8.10535252e-02
-5.75948179e-01 9.76154983e-01 5.72783314e-02 1.85013890e-01
-2.47215971e-01 -1.84521168e-01 2.21274123e-01 4.15214926e-01
5.04668891e-01 3.52798074e-01 -1.19545661e-01 -6.92774713e-01
-7.75771141e-01 4.26716404e-03 6.30327463e-01 5.77425957e-01
4.04624164e-01 3.46755935e-03 -3.29198509e-01 7.20987618e-01
3.73452634e-01 2.21677721e-01 4.83686060e-01 -9.67438042e-01
-9.74080116e-02 6.18700266e-01 -3.54286402e-01 -7.24653959e-01
-5.20069420e-01 -2.21479058e-01 5.99017963e-02 6.20654106e-01
5.99158466e-01 -2.77237087e-01 -6.57392323e-01 1.70433402e+00
8.33214700e-01 2.10182294e-01 -2.64391959e-01 4.90000039e-01
1.38392627e+00 2.79678077e-01 3.20330203e-01 -4.02523726e-01
1.62287426e+00 -9.04513776e-01 -1.05920351e+00 1.74187109e-01
7.64210522e-01 -8.82335663e-01 1.07653594e+00 7.19428420e-01
-1.17514956e+00 -7.79625595e-01 -8.30248952e-01 1.45615235e-01
-4.78538454e-01 -1.60563737e-01 4.48329628e-01 1.15996230e+00
-1.10965765e+00 3.54971409e-01 -7.31902421e-01 -1.60039485e-01
5.42270660e-01 4.86437023e-01 -5.21512210e-01 1.83052883e-01
-6.44919693e-01 1.09568071e+00 -5.06162763e-01 -3.61220986e-01
-7.77089536e-01 -1.05939186e+00 -8.75679970e-01 2.95526236e-01
-1.29412219e-01 1.49045605e-02 1.38986659e+00 -1.21343660e+00
-1.83137524e+00 1.07700217e+00 5.40348366e-02 9.13854912e-02
5.36135957e-02 -4.50950086e-01 -1.43221751e-01 2.99537063e-01
-4.21804070e-01 7.74346292e-01 6.79980457e-01 -4.90732044e-01
-3.13898593e-01 -8.29870820e-01 -3.92309204e-02 3.98509562e-01
-7.51014948e-01 2.69251794e-01 2.74490237e-01 -1.54924452e-01
-1.45529851e-01 -8.30730498e-01 3.68552685e-01 -1.93330601e-01
4.23865795e-01 -8.39340150e-01 9.80371416e-01 -7.82765031e-01
1.23539591e+00 -2.25132537e+00 -3.04359555e-01 1.98948309e-01
1.83663577e-01 2.58034378e-01 -3.79600860e-02 4.40218598e-02
-6.86991811e-01 -1.68411836e-01 7.33428717e-01 -2.36969903e-01
1.08707689e-01 -2.51613498e-01 1.09457761e-01 7.69393802e-01
-2.87493728e-02 5.38020730e-01 -6.26883209e-01 -5.36214113e-01
6.11652732e-01 7.64460623e-01 -6.56240702e-01 6.47297680e-01
5.27369916e-01 2.61614293e-01 1.39474481e-01 5.44366181e-01
3.74276787e-01 3.85789603e-01 -2.64139444e-01 9.19595435e-02
-3.24551642e-01 1.61279619e-01 -1.08655512e+00 1.49188137e+00
-5.27395904e-01 1.17158139e+00 2.18959227e-01 -8.62243772e-01
9.42520261e-01 5.36744177e-01 5.39592683e-01 -6.36660993e-01
2.91212231e-01 -3.48361433e-01 4.15578112e-02 -9.19347882e-01
2.56826073e-01 1.51114017e-01 3.01262528e-01 4.47687656e-01
4.39784914e-01 -1.01042032e-01 -3.38802814e-01 -9.24000293e-02
1.03374457e+00 1.58605590e-01 -2.07278281e-02 -1.69879198e-01
2.47025385e-01 -5.19823611e-01 -2.44552810e-02 3.02430660e-01
-8.69031131e-01 2.08606154e-01 4.29424554e-01 -1.63975745e-01
-1.38597369e-01 -8.56578588e-01 -3.35669577e-01 1.91009772e+00
-6.84767902e-01 -2.23492607e-02 -1.14900374e+00 -7.80002594e-01
-2.41245061e-01 9.23873127e-01 -6.74358785e-01 -2.87405312e-01
-1.82368219e-01 -1.50477350e-01 2.18642354e-01 4.30309683e-01
-1.34516686e-01 -1.14943302e+00 -9.93415236e-01 -2.03879789e-01
2.27393195e-01 -6.94333613e-01 -2.56493777e-01 3.50031286e-01
-5.44056773e-01 -1.03571725e+00 -3.24930608e-01 -6.80770636e-01
5.90732813e-01 5.57528548e-02 9.47027564e-01 2.89573465e-02
-4.49097753e-01 1.37419522e+00 -9.28816199e-02 -9.63398099e-01
1.31090313e-01 -2.59664208e-01 1.95010900e-01 -3.40005130e-01
9.71428692e-01 -5.80302000e-01 -4.58783597e-01 -2.94909120e-01
-6.15068257e-01 -3.94962817e-01 2.93677092e-01 5.93813837e-01
1.45572588e-01 -2.97171861e-01 6.86541915e-01 -3.04415643e-01
5.69603980e-01 -2.39781305e-01 -5.65405488e-01 7.36002252e-02
-6.93771780e-01 -4.87101108e-01 -1.76090717e-01 -7.16807544e-01
-9.90383327e-01 1.36559242e-02 -1.77511156e-01 -8.13835859e-01
-7.20571935e-01 1.43014088e-01 -3.30256298e-02 -4.66813087e-01
5.77312946e-01 -4.53465700e-01 2.74050415e-01 -1.45505160e-01
-7.21848458e-02 7.34859943e-01 3.06886196e-01 -6.98588789e-01
1.55261114e-01 -1.86694652e-01 -6.30183518e-02 -1.01995099e+00
-1.05896413e+00 -4.21152413e-01 -6.71997070e-01 -9.62401927e-01
1.03613901e+00 -1.14997017e+00 -1.51145113e+00 4.01701033e-01
-8.15138459e-01 -4.25724119e-01 -3.54195565e-01 8.64678741e-01
-3.45293164e-01 -1.44052595e-01 -3.73368472e-01 -1.16430509e+00
-1.88709185e-01 -1.27710760e+00 6.46994948e-01 7.33932078e-01
-5.19514263e-01 -1.03537261e+00 3.87060314e-01 9.73866701e-01
1.40818879e-01 3.09521407e-01 4.70189542e-01 -1.08596921e+00
2.09733192e-02 -1.81284547e-01 2.38860652e-01 3.66473019e-01
-2.78562546e-01 5.01861811e-01 -1.82488048e+00 -2.71010756e-01
-3.38361487e-02 -8.27463269e-01 3.85215878e-01 5.84481418e-01
1.58954823e+00 -1.01381913e-01 2.30063692e-01 2.26254255e-01
9.57079470e-01 4.12718743e-01 4.11984950e-01 2.46792182e-01
5.87930739e-01 1.19147623e+00 5.85676730e-01 2.96344668e-01
2.48041481e-01 6.40898645e-01 4.94480312e-01 1.72836363e-01
-7.85655901e-03 -1.35666309e-02 7.97751606e-01 6.34242892e-01
9.24675986e-02 2.27883294e-01 -9.53854442e-01 6.49019361e-01
-1.14251518e+00 -1.04918754e+00 -9.29309875e-02 2.29065251e+00
7.71408319e-01 -1.79442614e-01 3.41713578e-01 2.96776772e-01
4.34527457e-01 -1.74842015e-01 4.05452633e-03 -1.21515095e+00
7.88556039e-01 8.71585250e-01 -3.89048085e-02 5.20228922e-01
-9.88359809e-01 3.67026508e-01 5.93431950e+00 6.45928621e-01
-1.15222871e+00 1.54615864e-01 6.87944412e-01 -6.72976971e-01
-9.01907496e-03 -6.75796270e-01 -8.49611402e-01 2.64849603e-01
1.61595201e+00 3.39496583e-01 1.20534912e-01 8.53967905e-01
1.92577451e-01 -2.24492058e-01 -1.15669239e+00 8.81798923e-01
2.48986468e-01 -6.79615736e-01 -6.45242095e-01 7.90647790e-02
7.61342704e-01 -4.77093995e-01 4.95824754e-01 8.08251917e-01
3.17161828e-02 -1.42607069e+00 3.86240959e-01 6.58783317e-01
5.30979872e-01 -1.19060600e+00 5.52873969e-01 8.20822120e-02
-6.05669320e-01 -1.35607019e-01 3.62741202e-01 -4.64663535e-01
-8.30629230e-01 1.59761965e-01 -9.35540915e-01 -1.58839032e-01
1.03434682e+00 2.17991024e-01 -6.71654880e-01 7.10543871e-01
-1.86725482e-01 9.20334399e-01 -1.89569533e-01 -2.10308388e-01
2.26610780e-01 2.69836816e-03 2.10503832e-01 1.20516562e+00
2.65835673e-01 3.73673767e-01 -1.40818385e-02 2.52339542e-01
7.43079185e-02 2.61252463e-01 -6.80354893e-01 6.40493929e-02
5.53098440e-01 1.72560990e+00 -4.58756983e-01 2.14354116e-02
-7.37068534e-01 2.94356138e-01 4.44247693e-01 2.01242015e-01
-8.72419477e-01 -3.89649510e-01 8.17624152e-01 3.16998690e-01
-3.51864472e-02 2.64986008e-01 -1.90899476e-01 -5.01144588e-01
-4.49802220e-01 -1.02869833e+00 3.97538722e-01 -9.05988872e-01
-8.74508739e-01 -4.77349795e-02 2.84682542e-01 -7.15609252e-01
-1.80246100e-01 -7.03289211e-01 -9.93964791e-01 7.98422813e-01
-8.61553490e-01 -7.37086177e-01 -7.86693633e-01 5.27006030e-01
4.70136851e-01 1.86536238e-02 8.37693512e-01 2.71850199e-01
-1.08947742e+00 9.77561712e-01 -2.63621926e-01 -6.50408715e-02
1.05394757e+00 -1.31528580e+00 -7.69418657e-01 4.27395105e-01
1.41750276e-01 4.35917556e-01 6.80392027e-01 -1.99416168e-02
-1.31878591e+00 -7.46698201e-01 6.56888187e-01 -6.33976519e-01
3.10752749e-01 1.58006623e-01 -9.91372943e-01 8.12791526e-01
7.64529228e-01 -1.50653139e-01 1.36352086e+00 3.39254856e-01
-7.00702816e-02 -1.22325204e-01 -1.37194204e+00 2.47849688e-01
4.09892291e-01 -6.44150436e-01 -7.39578128e-01 8.54316428e-02
4.12502915e-01 -7.36594796e-01 -1.32363200e+00 3.27665329e-01
8.42486858e-01 -1.11856842e+00 9.09944117e-01 -6.21219397e-01
5.45122385e-01 4.31656420e-01 6.27424568e-02 -1.26431179e+00
-1.40570149e-01 -2.89297849e-01 -1.98662579e-01 1.39978564e+00
-1.18079312e-01 -3.52007776e-01 9.05431569e-01 9.80030179e-01
-3.74174416e-01 -1.19405568e+00 -8.95800531e-01 -4.65431344e-03
2.25457698e-01 -4.63225037e-01 3.07095498e-01 1.13472939e+00
1.92711577e-01 2.11324468e-01 4.87829089e-01 4.23052944e-02
4.30886209e-01 -1.65622503e-01 8.21516871e-01 -1.19532728e+00
1.74407527e-01 -7.32458353e-01 -5.81431746e-01 -1.42081365e-01
6.49185419e-01 -6.43559039e-01 -2.79810399e-01 -1.00944114e+00
2.66445458e-01 1.39540046e-01 -3.18810076e-01 6.73914790e-01
-2.48451546e-01 6.53147027e-02 -5.76902553e-02 -8.96708429e-01
-4.18647289e-01 3.06280613e-01 1.06350410e+00 2.52661020e-01
-2.66699821e-01 4.05572169e-03 -6.50364935e-01 8.30666184e-01
6.55356884e-01 -1.73220336e-01 -5.23962080e-01 1.41560629e-01
-1.75313130e-01 -3.80776031e-03 3.77784252e-01 -1.02932787e+00
8.01250860e-02 -2.55881548e-01 8.25031519e-01 -5.63607395e-01
3.95833194e-01 -1.01869404e+00 -2.45668218e-01 1.91847637e-01
-5.67319512e-01 -9.07459483e-02 7.79667795e-01 -1.35198981e-01
-9.41710472e-02 -5.88165522e-01 1.00444615e+00 3.57991517e-01
-2.56631583e-01 1.60858016e-02 -6.77512407e-01 -1.00040182e-01
1.34752572e+00 -3.37297082e-01 -1.03912756e-01 -6.07400775e-01
-9.54486787e-01 1.74968690e-01 9.27613750e-02 4.12911505e-01
5.62738299e-01 -1.30676866e+00 -3.17000389e-01 3.91674787e-01
-7.59935752e-02 -1.06686518e-01 5.00189602e-01 1.15325880e+00
-9.85136777e-02 3.03138465e-01 -3.14628720e-01 -8.19499135e-01
-2.28764963e+00 2.94066608e-01 3.25650185e-01 9.98730771e-03
9.42500029e-03 1.09106410e+00 -1.71077671e-03 -4.46257472e-01
8.06807697e-01 -2.81744570e-01 -7.17064917e-01 7.70248294e-01
7.79721856e-01 8.24896336e-01 2.95000583e-01 -4.49923009e-01
-1.71188965e-01 2.77568012e-01 1.59199104e-01 -5.89342602e-02
1.26846135e+00 1.63203746e-01 3.22879165e-01 6.56811178e-01
1.22293544e+00 1.15600906e-01 -1.04138386e+00 3.23828101e-01
-2.48810530e-01 -4.00174767e-01 5.43653369e-01 -6.56969845e-01
-9.17035162e-01 1.27556002e+00 1.26936769e+00 -1.68101210e-02
1.26342905e+00 -2.83673316e-01 -6.30683601e-02 1.78237081e-01
-3.67253363e-01 -1.14180648e+00 5.90290368e-01 4.29895282e-01
6.81902885e-01 -1.27444339e+00 7.53266513e-02 1.27314642e-01
-3.92762542e-01 1.19541812e+00 1.25860715e+00 1.95575446e-01
6.40220940e-01 1.92419916e-01 -4.92562801e-02 -5.48715770e-01
-9.07886207e-01 5.17517887e-02 7.48250604e-01 4.34903920e-01
1.11872149e+00 1.99800551e-01 1.01826593e-01 7.56423354e-01
-3.30022573e-01 -1.40667364e-01 5.35958767e-01 9.90637958e-01
-7.83994019e-01 -4.94633496e-01 -8.82799685e-01 5.67642093e-01
-8.00948501e-01 2.55526394e-01 -2.66206503e-01 8.03852201e-01
3.16902965e-01 1.04624200e+00 5.84994972e-01 -2.48483837e-01
4.79237586e-01 8.63972485e-01 9.94035184e-01 -7.49336004e-01
-1.22653294e+00 -2.04409510e-01 -3.10334694e-02 -5.91644526e-01
-3.86097133e-01 -1.04914498e+00 -1.06314349e+00 -4.41593826e-01
-5.22392511e-01 1.52769148e-01 7.91839361e-01 6.89873874e-01
1.15317283e-02 8.69514883e-01 6.47819281e-01 -7.91517556e-01
-4.21874374e-01 -1.30653155e+00 -4.16932940e-01 5.92352599e-02
4.20533240e-01 -8.71606469e-01 -7.22368598e-01 -2.50250161e-01] | [13.496199607849121, 2.4262328147888184] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.