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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2359a740-8ee2-47e2-98bc-499c9976dcad
|
zero-shot-generalizable-end-to-end-task
|
2303.16252
| null |
https://arxiv.org/abs/2303.16252v1
|
https://arxiv.org/pdf/2303.16252v1.pdf
|
Zero-Shot Generalizable End-to-End Task-Oriented Dialog System using Context Summarization and Domain Schema
|
Task-oriented dialog systems empower users to accomplish their goals by facilitating intuitive and expressive natural language interactions. State-of-the-art approaches in task-oriented dialog systems formulate the problem as a conditional sequence generation task and fine-tune pre-trained causal language models in the supervised setting. This requires labeled training data for each new domain or task, and acquiring such data is prohibitively laborious and expensive, thus making it a bottleneck for scaling systems to a wide range of domains. To overcome this challenge, we introduce a novel Zero-Shot generalizable end-to-end Task-oriented Dialog system, ZS-ToD, that leverages domain schemas to allow for robust generalization to unseen domains and exploits effective summarization of the dialog history. We employ GPT-2 as a backbone model and introduce a two-step training process where the goal of the first step is to learn the general structure of the dialog data and the second step optimizes the response generation as well as intermediate outputs, such as dialog state and system actions. As opposed to state-of-the-art systems that are trained to fulfill certain intents in the given domains and memorize task-specific conversational patterns, ZS-ToD learns generic task-completion skills by comprehending domain semantics via domain schemas and generalizing to unseen domains seamlessly. We conduct an extensive experimental evaluation on SGD and SGD-X datasets that span up to 20 unique domains and ZS-ToD outperforms state-of-the-art systems on key metrics, with an improvement of +17% on joint goal accuracy and +5 on inform. Additionally, we present a detailed ablation study to demonstrate the effectiveness of the proposed components and training mechanism
|
['A. B. Siddique', 'M. H. Maqbool', 'Adib Mosharrof']
|
2023-03-28
| null | null | null | null |
['response-generation']
|
['natural-language-processing']
|
[ 2.26877898e-01 5.90071142e-01 -1.53582647e-01 -7.08417356e-01
-6.06771469e-01 -8.22051585e-01 9.66961920e-01 -1.25473723e-01
-3.11201155e-01 1.10570478e+00 6.84438229e-01 -2.33263314e-01
1.00446798e-01 -6.22476935e-01 -2.92907298e-01 -2.33979642e-01
1.50165394e-01 1.29731762e+00 3.45565349e-01 -8.34995985e-01
-3.80797982e-02 -1.27103567e-01 -8.73624206e-01 4.05320674e-01
1.10314512e+00 7.01766908e-01 3.04907829e-01 6.26326203e-01
-4.45053577e-01 9.05240715e-01 -7.36597359e-01 -4.90883917e-01
-5.22579253e-02 -5.65411925e-01 -1.30822492e+00 2.35357359e-01
-2.82758363e-02 -6.67646706e-01 -4.67593074e-01 4.91755396e-01
4.95255768e-01 4.35112804e-01 6.36542976e-01 -1.20429027e+00
-6.74935341e-01 7.79079199e-01 1.03007287e-01 -2.46980175e-01
6.37121379e-01 6.33140385e-01 1.05475116e+00 -6.20081961e-01
6.16261363e-01 1.59398031e+00 3.05732131e-01 1.21073294e+00
-1.35291255e+00 -4.72378045e-01 5.08001268e-01 -3.57416570e-01
-5.74864626e-01 -5.50616741e-01 7.35399961e-01 -3.92899841e-01
1.18725574e+00 -2.24211782e-01 9.60281491e-02 1.82668805e+00
-1.49645776e-01 8.83016288e-01 8.91302466e-01 -2.07152575e-01
4.02459145e-01 2.54365265e-01 3.94520462e-01 5.18304110e-01
-2.16922864e-01 -2.66878735e-02 -6.09220982e-01 -3.11187267e-01
5.24860978e-01 -1.33415252e-01 -1.42454222e-01 -2.65487015e-01
-1.10233867e+00 1.03713858e+00 2.47695833e-01 9.22114030e-02
-3.80104929e-01 -2.56479800e-01 5.64339519e-01 4.64105666e-01
4.13371384e-01 8.04888666e-01 -6.60714686e-01 -3.40562463e-01
-3.35840076e-01 6.40140355e-01 1.45718443e+00 1.23157823e+00
4.03789014e-01 -3.99276055e-02 -5.18534184e-01 1.08587837e+00
-9.15954076e-03 3.56079400e-01 5.88074267e-01 -1.04444134e+00
8.20067942e-01 9.06838000e-01 3.03550065e-01 -7.87861496e-02
-4.34413642e-01 2.68733092e-02 -5.98255515e-01 -3.07242960e-01
6.49502277e-01 -8.05345714e-01 -8.24641407e-01 2.24396372e+00
2.98290700e-01 -1.92565903e-01 5.64762414e-01 7.79263675e-01
9.28379774e-01 6.48073792e-01 4.94222879e-01 -1.72759891e-01
1.37286341e+00 -9.91586864e-01 -5.69191337e-01 -6.90649748e-01
4.45840716e-01 -1.93580523e-01 1.51550186e+00 1.43349275e-01
-1.07546461e+00 -4.36845243e-01 -7.62935042e-01 -2.07267404e-01
-1.71326756e-01 -1.76608115e-01 6.63296700e-01 2.02400103e-01
-8.40972126e-01 2.99515247e-01 -5.78798771e-01 -6.50816262e-01
9.57726538e-02 3.88028055e-01 -1.91843316e-01 1.37588143e-01
-1.63710809e+00 9.24351275e-01 5.24498105e-01 -4.76003200e-01
-1.25576174e+00 -7.92051196e-01 -9.54444945e-01 2.62542665e-01
6.94450557e-01 -9.38971639e-01 2.01538444e+00 -4.22904283e-01
-2.12757277e+00 7.42280841e-01 -1.71272323e-01 -7.09651232e-01
4.42926824e-01 -4.55952078e-01 -1.28710285e-01 -4.81678993e-02
1.45651791e-02 6.76474690e-01 5.84560513e-01 -1.14616060e+00
-5.69627643e-01 -1.02184884e-01 5.05501390e-01 5.66391647e-01
-4.38345999e-01 -1.68558255e-01 -3.66363317e-01 -2.66882986e-01
-3.32620978e-01 -1.00865281e+00 -3.10025990e-01 -5.64926386e-01
-5.33858597e-01 -6.30686760e-01 7.98057497e-01 -3.72727454e-01
1.04964566e+00 -1.86925530e+00 3.47613096e-01 -4.34836835e-01
2.84999579e-01 2.95674413e-01 -1.06262363e-01 7.10713506e-01
2.89965540e-01 -1.96787670e-01 -3.31806958e-01 -7.07518399e-01
2.62053639e-01 4.15500551e-01 -7.87095308e-01 -2.98050314e-01
4.17352468e-01 9.25576568e-01 -1.08672786e+00 -3.06121081e-01
3.62020314e-01 -1.50642931e-01 -6.91170454e-01 1.02200758e+00
-1.08209014e+00 8.20732117e-01 -7.84962237e-01 8.54594335e-02
2.34791979e-01 -4.62246001e-01 5.75184584e-01 3.32444698e-01
2.89001197e-01 1.00665748e+00 -6.32805884e-01 2.09326124e+00
-7.81743765e-01 1.03057228e-01 8.75452980e-02 -8.87008011e-01
1.01610196e+00 6.46809816e-01 1.60587028e-01 -4.06481922e-01
1.47075914e-02 -1.69019058e-01 -1.99450910e-01 -5.07521093e-01
5.86977720e-01 -4.67014998e-01 -7.83053100e-01 7.05939770e-01
5.75755358e-01 -3.94976020e-01 -2.92083211e-02 6.85856462e-01
1.06707728e+00 5.25886752e-02 4.43805009e-01 -1.77910440e-02
5.65751612e-01 3.68055612e-01 3.80909353e-01 7.52060950e-01
-7.59752095e-02 2.00860769e-01 7.52382517e-01 -2.22937450e-01
-8.01550269e-01 -1.17168117e+00 3.48932087e-01 1.65415251e+00
9.78815034e-02 -3.04092526e-01 -7.51261771e-01 -9.72774804e-01
1.15117878e-02 1.29918981e+00 -3.97498876e-01 -3.24733734e-01
-6.06375337e-01 -4.71715748e-01 5.66032648e-01 5.02047122e-01
9.42894638e-01 -1.24241197e+00 -2.89975703e-01 3.29901159e-01
-5.27564168e-01 -1.44621992e+00 -5.20026684e-01 1.98147491e-01
-8.05388272e-01 -7.95818269e-01 -5.20145059e-01 -7.01776803e-01
3.17330390e-01 1.37207896e-01 1.38610816e+00 -2.81677723e-01
2.94778496e-01 3.55383426e-01 -2.40273625e-01 -2.06554651e-01
-9.32892442e-01 4.37469304e-01 1.09851584e-01 -2.93069482e-01
4.55288291e-01 -6.70297265e-01 -4.61415619e-01 4.62274402e-01
-4.60699588e-01 1.76119059e-01 3.90235960e-01 1.14862490e+00
-1.45382419e-01 -4.14483815e-01 1.01210368e+00 -1.27093756e+00
1.42416728e+00 -6.20248973e-01 -3.80158275e-01 2.47070596e-01
-4.55883622e-01 3.75846356e-01 1.06865573e+00 -4.87351567e-01
-1.82197690e+00 1.18312560e-01 5.64497942e-03 -1.45243630e-01
-4.16579127e-01 2.66046643e-01 -2.55725950e-01 7.66292751e-01
1.00397980e+00 3.56390685e-01 2.68994123e-02 -4.66544598e-01
8.18653941e-01 8.68892074e-01 8.72882009e-01 -1.11201274e+00
7.20295191e-01 5.29252253e-02 -5.55643022e-01 -7.59971559e-01
-1.01755261e+00 -6.02185309e-01 -4.93710369e-01 9.90601927e-02
1.05968404e+00 -1.04080629e+00 -9.84841108e-01 3.85195225e-01
-1.27092302e+00 -9.98379409e-01 -2.31818482e-01 -4.29130904e-02
-6.61226094e-01 9.27264318e-02 -6.41029537e-01 -7.88806856e-01
-6.28351212e-01 -8.71642113e-01 1.15248120e+00 2.00815573e-01
-6.39265120e-01 -1.25358343e+00 1.01134501e-01 4.25065994e-01
3.72983545e-01 -9.05960724e-02 1.06798494e+00 -1.35522556e+00
-3.08772713e-01 1.71448305e-01 2.71382481e-02 2.31927350e-01
1.49668321e-01 -7.77429044e-01 -1.00667560e+00 -1.89064011e-01
2.54021585e-01 -1.08211410e+00 4.90540832e-01 5.25324279e-03
6.26653314e-01 -5.20518661e-01 -4.06660259e-01 5.10685779e-02
6.90786362e-01 2.57199019e-01 1.72982290e-01 -1.21652551e-01
2.10794553e-01 8.76337826e-01 7.97629714e-01 4.76696074e-01
6.91650331e-01 8.99160624e-01 5.90415224e-02 1.34504080e-01
-7.16097094e-03 -7.37476885e-01 4.65297222e-01 3.21596414e-01
3.65296304e-01 -4.58823562e-01 -8.57663333e-01 6.15338326e-01
-2.08974028e+00 -7.95624375e-01 2.86068410e-01 1.94920695e+00
1.48144066e+00 3.37919801e-01 3.99876595e-01 -6.06155157e-01
4.85693216e-01 2.10153952e-01 -8.30052853e-01 -2.72350758e-01
2.96186507e-01 1.99261293e-01 -1.93646103e-01 7.07478285e-01
-8.96488607e-01 1.58637345e+00 5.71386671e+00 3.44839007e-01
-9.80360389e-01 1.39815882e-01 4.26532924e-01 6.88989311e-02
-1.02554344e-01 1.07395999e-01 -9.32173133e-01 2.38093108e-01
1.02858317e+00 -4.20271367e-01 4.48548734e-01 1.10899818e+00
2.05302835e-01 1.34608090e-01 -1.47162008e+00 3.74691188e-01
-2.40706474e-01 -1.07883692e+00 1.42963352e-02 -2.01754093e-01
5.10087430e-01 -8.72853249e-02 -8.92029554e-02 9.12457824e-01
1.28947115e+00 -8.64642084e-01 2.67167956e-01 -2.75161862e-02
8.08188617e-01 -1.31363586e-01 3.41478020e-01 7.66960084e-01
-7.48508632e-01 -1.29125804e-01 -2.65720218e-01 -2.05867738e-01
4.90340620e-01 3.05338539e-02 -1.78321981e+00 3.82989496e-01
2.15155780e-01 4.90826428e-01 3.28887887e-02 1.51429966e-01
-6.15412354e-01 6.12513304e-01 -1.60916701e-01 -2.35985890e-01
3.36291015e-01 3.17849778e-02 6.23274803e-01 1.24748206e+00
-1.60768390e-01 5.19314706e-01 4.67295736e-01 1.06267440e+00
-3.46863806e-01 -3.96562308e-01 -6.49125159e-01 -8.82474557e-02
7.44996905e-01 1.02559066e+00 1.60843860e-02 -5.53122759e-01
-3.62862706e-01 1.11370206e+00 4.78485763e-01 4.31082904e-01
-6.39091313e-01 -6.87440634e-02 8.60951722e-01 -1.35833472e-01
-9.56849679e-02 -3.48929316e-01 -3.98141086e-01 -1.32450652e+00
-3.60813588e-01 -1.05578589e+00 5.53930163e-01 -4.84424174e-01
-1.51159275e+00 6.87129021e-01 2.17803121e-01 -9.03521657e-01
-1.01195252e+00 -4.45119888e-01 -8.09233189e-01 1.07517350e+00
-1.16857100e+00 -1.19715118e+00 -5.36754966e-01 9.04958010e-01
1.22506392e+00 -4.47361916e-01 1.14361262e+00 -3.34418952e-01
-3.56957376e-01 5.38039744e-01 -3.08880568e-01 1.85645595e-01
1.05308557e+00 -1.52311409e+00 8.42750072e-01 5.03688633e-01
-3.82723719e-01 9.26235557e-01 9.05668139e-01 -6.98303759e-01
-1.25369382e+00 -1.14086723e+00 9.30141926e-01 -7.54565775e-01
7.26701915e-01 -8.59694660e-01 -9.25633013e-01 9.27964389e-01
5.17193079e-01 -6.31570697e-01 5.19769967e-01 5.27513325e-01
-3.46061349e-01 2.04360679e-01 -1.03201783e+00 8.13471854e-01
1.36770165e+00 -4.97606188e-01 -1.26632774e+00 5.89998901e-01
1.22359478e+00 -6.99940383e-01 -7.13301003e-01 8.98824856e-02
1.52953133e-01 -6.77523017e-01 9.25784945e-01 -1.11107254e+00
6.15918159e-01 2.47506663e-01 1.43501505e-01 -1.58899105e+00
9.45517619e-04 -1.18647242e+00 1.03293657e-02 1.25447273e+00
6.31914198e-01 -5.91619670e-01 6.32098496e-01 1.07508349e+00
-4.33150500e-01 -3.53811145e-01 -5.96033573e-01 -6.07640922e-01
3.24933827e-01 -1.86631128e-01 5.26535928e-01 7.98807800e-01
6.97166383e-01 1.30918920e+00 -6.13267958e-01 -7.12113157e-02
2.66672820e-01 3.66440743e-01 1.35012364e+00 -1.18950713e+00
-4.89305884e-01 -9.15788636e-02 5.92895031e-01 -1.78955328e+00
5.65875769e-01 -6.57178819e-01 3.39299828e-01 -1.37487090e+00
3.76297385e-02 -4.86977756e-01 2.96073794e-01 6.02884829e-01
-4.27260607e-01 -7.62347758e-01 1.26538947e-01 3.13384563e-01
-7.26776898e-01 8.98721635e-01 1.23349965e+00 -7.41543323e-02
-8.16837549e-01 2.64642775e-01 -1.08510041e+00 4.73709017e-01
6.07080638e-01 -1.44346356e-01 -1.07839334e+00 -3.46807450e-01
-4.59308922e-01 6.48374557e-01 1.90984130e-01 -5.92892289e-01
4.51918930e-01 -4.69276488e-01 -2.58818269e-01 -1.33429334e-01
6.98732436e-01 -5.03483176e-01 -3.99224222e-01 2.84283459e-01
-8.45401466e-01 -4.05424833e-01 2.56454915e-01 6.57218158e-01
-1.47781253e-01 -1.34131289e-03 6.45691216e-01 -2.78826326e-01
-7.77999938e-01 6.15220442e-02 -2.09728301e-01 5.54814994e-01
9.72187400e-01 2.11383849e-01 -7.83879995e-01 -9.66096103e-01
-6.96770310e-01 7.53557026e-01 1.97616294e-01 6.83443964e-01
4.33180481e-01 -8.10988605e-01 -6.37977540e-01 -1.09676430e-02
2.71488190e-01 2.80181319e-01 1.87538058e-01 2.67322183e-01
1.59456536e-01 7.40221977e-01 -9.05402079e-02 -5.27284861e-01
-9.56965625e-01 4.63236988e-01 3.00494671e-01 -7.84765065e-01
-5.22065639e-01 8.20647240e-01 7.78894424e-01 -7.21194923e-01
3.09883207e-01 -1.43863395e-01 -1.62292570e-01 -2.93414831e-01
2.60748893e-01 -2.21853420e-01 -2.57324100e-01 1.73999667e-02
-1.17182843e-01 -1.93745926e-01 -4.81228322e-01 -4.38788891e-01
1.17511272e+00 -1.51098967e-01 3.52353662e-01 2.07267940e-01
6.24752164e-01 -4.50220704e-01 -1.52790391e+00 -7.23945081e-01
1.32288635e-01 2.40125321e-02 -4.92731124e-01 -1.33723462e+00
-2.42836967e-01 9.22808409e-01 -1.12878606e-01 1.79551810e-01
9.77985620e-01 3.02571386e-01 1.06661415e+00 6.90389097e-01
4.46844488e-01 -9.14099932e-01 6.66371107e-01 1.08322430e+00
1.05894387e+00 -1.38819563e+00 -4.42269802e-01 -4.45395201e-01
-1.45262992e+00 8.11815798e-01 1.04772580e+00 6.64647818e-02
1.57245040e-01 1.26323581e-01 -7.65979514e-02 -2.00626984e-01
-1.33601105e+00 -6.24742247e-02 -1.19159641e-02 6.44135773e-01
3.94473761e-01 -9.74165462e-03 -9.13181752e-02 1.09823513e+00
-2.84244627e-01 9.16391537e-02 2.20315486e-01 8.16952586e-01
-4.38823581e-01 -1.33954906e+00 1.29931912e-01 1.81511715e-01
-1.48475841e-02 -1.15843974e-01 -8.97399843e-01 7.77773499e-01
-6.75556004e-01 1.33776140e+00 -2.94276327e-01 -5.36610007e-01
6.77484870e-01 6.81027770e-01 1.82465743e-02 -1.15246928e+00
-8.31663609e-01 -2.44283080e-01 6.99809551e-01 -4.58523959e-01
-1.53954327e-02 -5.27959883e-01 -1.45570815e+00 -3.00608993e-01
5.27929813e-02 2.77287900e-01 1.07882842e-01 1.00399756e+00
5.29722989e-01 4.49128002e-01 6.18385911e-01 -5.59900761e-01
-1.25114584e+00 -1.34808934e+00 8.85150209e-03 7.76933312e-01
1.62102297e-01 -7.96536028e-01 -1.14841811e-01 1.96691856e-01]
|
[12.822306632995605, 8.023255348205566]
|
4855786f-6cf5-44ed-b0dd-7c5406f9aa85
|
corefdre-document-level-relation-extraction
|
2202.10744
| null |
https://arxiv.org/abs/2202.10744v1
|
https://arxiv.org/pdf/2202.10744v1.pdf
|
CorefDRE: Document-level Relation Extraction with coreference resolution
|
Document-level relation extraction is to extract relation facts from a document consisting of multiple sentences, in which pronoun crossed sentences are a ubiquitous phenomenon against a single sentence. However, most of the previous works focus more on mentions coreference resolution except for pronouns, and rarely pay attention to mention-pronoun coreference and capturing the relations. To represent multi-sentence features by pronouns, we imitate the reading process of humans by leveraging coreference information when dynamically constructing a heterogeneous graph to enhance semantic information. Since the pronoun is notoriously ambiguous in the graph, a mention-pronoun coreference resolution is introduced to calculate the affinity between pronouns and corresponding mentions, and the noise suppression mechanism is proposed to reduce the noise caused by pronouns. Experiments on the public dataset, DocRED, DialogRE and MPDD, show that Coref-aware Doc-level Relation Extraction based on Graph Inference Network outperforms the state-of-the-art.
|
['Zhong Jiang', 'Qizhu Dai', 'Rongzhen Li', 'Zhongxuan Xue']
|
2022-02-22
| null | null | null | null |
['document-level-relation-extraction']
|
['natural-language-processing']
|
[ 2.57689327e-01 5.28668821e-01 -4.98504370e-01 -4.07355785e-01
-5.52476704e-01 -6.14810169e-01 7.23472416e-01 5.71530759e-01
-3.61554801e-01 8.41424227e-01 9.14159656e-01 -1.82924628e-01
-3.13781738e-01 -1.11947691e+00 -3.94037217e-01 -3.16894740e-01
1.62192285e-01 9.28804696e-01 4.61103618e-01 -7.29194224e-01
1.45215228e-01 2.46915564e-01 -1.24554694e+00 4.48196292e-01
9.47027564e-01 4.85018909e-01 2.68047065e-01 1.83103323e-01
-8.76272202e-01 6.81260884e-01 -1.00678372e+00 -1.08105075e+00
-3.26200515e-01 -2.67972708e-01 -1.41143203e+00 -1.16474189e-01
3.34714442e-01 2.63541669e-01 -5.24557173e-01 1.61248481e+00
4.48332399e-01 3.32913548e-01 5.21175325e-01 -9.16477263e-01
-5.64866185e-01 1.34761047e+00 -5.27835250e-01 6.15839243e-01
8.92066717e-01 -5.70886433e-01 1.65204108e+00 -4.22552139e-01
1.17371500e+00 1.91381752e+00 1.83495224e-01 5.33771932e-01
-1.06615806e+00 -4.93618727e-01 4.34143811e-01 6.32647932e-01
-1.28552461e+00 -3.64550412e-01 1.02064335e+00 3.54875368e-03
1.31584454e+00 6.27656043e-01 2.43384928e-01 1.37817025e+00
2.59021819e-01 6.72030091e-01 7.10454822e-01 -4.99255240e-01
-1.46969199e-01 -2.47060120e-01 8.15888345e-01 4.82911736e-01
4.18567449e-01 -4.03357506e-01 -6.45703375e-01 -1.60984427e-01
1.96537927e-01 -1.77533850e-01 -5.70698678e-01 1.23425052e-01
-1.00021553e+00 6.79769993e-01 3.68475735e-01 5.27996838e-01
-2.32826591e-01 -7.30056345e-01 5.55359781e-01 2.36062676e-01
3.09800804e-01 5.25505543e-01 -5.40592253e-01 9.25876349e-02
-3.49440217e-01 1.76011920e-01 1.09865034e+00 1.39863646e+00
6.32505000e-01 -8.01696479e-01 -3.64504337e-01 8.78063083e-01
2.46825412e-01 2.70159930e-01 3.47957939e-01 -6.88067794e-01
8.94810677e-01 1.11519444e+00 -2.48647183e-01 -1.26636243e+00
-6.55033767e-01 -4.63479817e-01 -9.37144041e-01 -8.11032772e-01
2.53294129e-02 -9.69055220e-02 -4.36692387e-01 1.48934150e+00
4.40419883e-01 -3.51664498e-02 3.86122435e-01 9.37442124e-01
1.64750063e+00 3.69133890e-01 1.88838333e-01 -8.85583282e-01
1.97753894e+00 -7.93248415e-01 -1.40441239e+00 -5.24405539e-01
5.04108131e-01 -7.01427281e-01 9.30649042e-01 -1.08648457e-01
-6.69075429e-01 -2.51858592e-01 -1.01326048e+00 -3.73262793e-01
-3.27571899e-01 -6.05007350e-01 7.54904389e-01 3.70702706e-02
-2.44652271e-01 6.66937053e-01 -4.28124100e-01 -4.07112479e-01
1.59172386e-01 3.19513917e-01 -5.64485073e-01 -6.12572543e-02
-1.98495352e+00 1.10504651e+00 7.58778512e-01 2.14748587e-02
-1.16149157e-01 -2.80625284e-01 -1.01090252e+00 2.26811454e-01
1.06962323e+00 -7.67728329e-01 1.06819320e+00 -5.16138852e-01
-1.11049521e+00 8.06769133e-01 -5.49697101e-01 -3.16115558e-01
-1.05972014e-01 -2.06338510e-01 -8.49125206e-01 -2.62418389e-02
2.77058899e-01 -1.44135905e-02 2.36330479e-01 -1.05319560e+00
-7.06198871e-01 -6.26860380e-01 5.22462845e-01 6.41669333e-01
-1.15560174e-01 2.48295382e-01 -6.99572861e-01 -1.71422914e-01
3.44812453e-01 -4.28084552e-01 6.91886693e-02 -1.20066345e+00
-9.33379829e-01 -8.76949191e-01 8.01594973e-01 -4.13758427e-01
1.58502436e+00 -1.99183214e+00 4.24499661e-01 1.36223147e-02
4.70201522e-01 1.01111114e-01 -1.66400984e-01 6.44164741e-01
6.02413639e-02 7.29453564e-02 5.67908585e-02 -8.86364952e-02
-9.75849945e-03 6.20453179e-01 -1.54053345e-01 6.05940409e-02
-1.22249700e-01 7.61168957e-01 -1.33651483e+00 -1.06738985e+00
-1.64485767e-01 1.99610740e-01 -1.08330950e-01 2.21252516e-01
-2.17281818e-01 8.79294872e-02 -7.85227299e-01 5.98369241e-01
6.78644896e-01 -1.98329613e-01 1.02328444e+00 -7.39642441e-01
2.42528871e-01 9.21851397e-01 -1.17031705e+00 1.59869421e+00
-1.76065341e-01 1.22382924e-01 -6.94959983e-03 -8.43110204e-01
9.50277030e-01 9.93447825e-02 9.33902059e-03 -6.54218793e-01
2.51315206e-01 -1.23782657e-01 2.61449963e-01 -6.36426926e-01
4.66718227e-01 6.10659681e-02 -2.26633489e-01 -2.25482345e-01
2.81268537e-01 7.18137762e-03 6.89644098e-01 7.08090842e-01
1.22996831e+00 -1.39391571e-01 5.54426014e-01 -3.65241617e-01
9.46540594e-01 5.22347391e-02 1.05268049e+00 4.11554515e-01
8.64552259e-02 1.14123344e-01 8.64497781e-01 3.69863771e-02
-2.27194369e-01 -1.05061209e+00 -1.15443379e-01 1.07433474e+00
7.15021729e-01 -1.04122818e+00 -7.03752875e-01 -1.16811955e+00
-9.49816555e-02 9.38660324e-01 -3.44641864e-01 -5.65403700e-02
-7.94463634e-01 -7.17492938e-01 2.89226085e-01 3.43690842e-01
6.64946079e-01 -1.04665244e+00 1.85388625e-01 2.59246171e-01
-7.13032663e-01 -1.45626652e+00 -5.68536639e-01 1.36566684e-01
-4.43658859e-01 -1.43625748e+00 2.26486474e-01 -1.00500059e+00
5.21550417e-01 4.62222286e-02 1.57804179e+00 1.82315439e-01
2.38989189e-01 -1.84622165e-02 -4.14720207e-01 -2.21542314e-01
-3.08144212e-01 3.93473923e-01 -3.26786190e-02 -4.62719709e-01
1.24413085e+00 -6.63883150e-01 -1.67512238e-01 -5.54851107e-02
-4.16983187e-01 -6.60649035e-03 4.26668316e-01 1.01393819e+00
5.97044945e-01 2.70549744e-01 3.03118646e-01 -1.49822688e+00
1.11934984e+00 -2.78598815e-01 -3.71349216e-01 4.22650754e-01
-6.39711559e-01 1.55708507e-01 4.70860958e-01 -2.00819775e-01
-1.58496642e+00 -2.76104689e-01 -7.99607709e-02 2.28216469e-01
-1.31756052e-01 8.92278612e-01 -8.67790043e-01 5.60188115e-01
4.19681996e-01 -1.58482522e-01 -3.66781563e-01 -4.89733458e-01
4.59681869e-01 8.34075630e-01 8.88281047e-01 -8.00832987e-01
6.06014252e-01 -1.88853331e-02 2.89843023e-01 -6.64981484e-01
-1.56207311e+00 -7.31297731e-01 -9.34937716e-01 5.11242151e-02
8.38778615e-01 -6.71911657e-01 -1.01356399e+00 -1.12482324e-01
-1.66089046e+00 5.44579029e-01 -8.22190940e-02 3.42067212e-01
3.88186984e-02 6.21627629e-01 -8.59957874e-01 -5.73133171e-01
-5.45701981e-01 -8.84927809e-01 9.61162448e-01 5.20841122e-01
-4.10649866e-01 -1.04294968e+00 3.60368304e-02 5.85452497e-01
-3.03883940e-01 5.97570464e-02 1.17371738e+00 -1.16617155e+00
-3.79962802e-01 7.55708963e-02 -3.15018028e-01 -4.41026874e-02
3.84595543e-01 -2.28681073e-01 -5.87923288e-01 1.52676878e-02
-1.81262493e-01 1.16598740e-01 6.30750299e-01 -1.56325325e-01
6.01535022e-01 -5.07126272e-01 -7.16734171e-01 2.00706124e-01
1.09287345e+00 1.33014277e-01 5.27624965e-01 3.85105968e-01
9.00159240e-01 7.12197781e-01 7.30078638e-01 -1.07742213e-01
8.30195487e-01 6.45045161e-01 3.40609521e-01 3.14517319e-01
-2.60303706e-01 -3.17276210e-01 4.82832640e-02 1.25028729e+00
-2.95271486e-01 -3.95453632e-01 -6.84958160e-01 4.00954425e-01
-2.02230668e+00 -1.11137474e+00 -6.32770240e-01 1.66805875e+00
1.36471903e+00 5.52217245e-01 -3.91238809e-01 -2.95751505e-02
1.06152725e+00 4.59847987e-01 -4.73050140e-02 -9.85522345e-02
-5.85845947e-01 7.91795552e-02 2.11414024e-01 8.96474004e-01
-1.14525676e+00 1.40409958e+00 5.16237926e+00 7.64988363e-01
-3.69426459e-01 1.18412655e-02 -1.59284756e-01 3.83489549e-01
-4.53495026e-01 2.80408233e-01 -1.21120334e+00 1.73002571e-01
6.14813030e-01 -5.77413499e-01 3.62323314e-01 5.45618415e-01
-2.48759061e-01 2.72110030e-02 -1.14022636e+00 7.55436003e-01
1.72077194e-01 -1.29306805e+00 3.62788051e-01 -1.87337622e-01
2.76992857e-01 -2.75695831e-01 -6.96280122e-01 4.55007643e-01
3.38139325e-01 -7.39323974e-01 7.11156353e-02 5.63333869e-01
2.92057574e-01 -8.82187366e-01 1.42175305e+00 4.14328754e-01
-1.37202120e+00 1.99015781e-01 -5.05547702e-01 -2.03748763e-01
2.54634976e-01 7.51939535e-01 -5.79641521e-01 1.12571287e+00
4.80233341e-01 6.75206363e-01 -4.56127435e-01 3.86940449e-01
-8.65939140e-01 3.69269550e-01 -6.50972649e-02 -4.09026146e-01
-1.34050369e-01 -2.76698798e-01 8.21923792e-01 1.53082681e+00
-2.33375505e-01 5.91642618e-01 4.09456104e-01 6.19959831e-01
-4.87641573e-01 4.27596509e-01 -3.86914134e-01 6.40512556e-02
1.06885338e+00 1.56240439e+00 -3.59852672e-01 -3.94623905e-01
-7.03018129e-01 9.23485994e-01 6.88653708e-01 2.48941258e-01
-3.28625381e-01 -6.37677133e-01 6.88287854e-01 -1.32325545e-01
-7.58043230e-02 3.62584158e-03 9.26132724e-02 -1.26354277e+00
8.08364060e-03 -8.04745495e-01 9.45501685e-01 -5.05682349e-01
-1.74220514e+00 6.12835228e-01 1.16468586e-01 -6.65457189e-01
-1.86992869e-01 -4.20135170e-01 -6.58977389e-01 8.93580496e-01
-1.43550146e+00 -9.98403788e-01 -2.38949433e-01 5.51773071e-01
4.63433653e-01 -6.41618446e-02 1.00446475e+00 3.06327194e-01
-7.67832637e-01 4.22103167e-01 -5.93385279e-01 5.37555456e-01
6.55507565e-01 -1.42705452e+00 2.95231223e-01 8.63518059e-01
2.46742859e-01 1.10675621e+00 9.37398434e-01 -1.00935996e+00
-1.48131037e+00 -8.16843987e-01 1.63829243e+00 -4.07253057e-01
8.39754105e-01 -3.73066723e-01 -1.13606822e+00 8.39556992e-01
5.89041471e-01 -2.06276119e-01 7.08823919e-01 9.05694067e-01
-3.48270535e-01 -2.20853649e-02 -8.80494118e-01 7.73117065e-01
1.77589917e+00 -6.21588945e-01 -1.69528890e+00 5.37909806e-01
1.01369095e+00 -8.83919656e-01 -9.57119823e-01 5.01475871e-01
-1.58047691e-01 -5.03970683e-01 9.14136767e-01 -1.04482138e+00
3.18357944e-01 -3.79854739e-01 -1.99848220e-01 -1.34942317e+00
-6.02521539e-01 -6.23982251e-01 -5.18256068e-01 1.88807154e+00
5.19414127e-01 -4.10426229e-01 4.72908974e-01 2.87645668e-01
-2.10499361e-01 -2.43768498e-01 -1.04845560e+00 -6.87459171e-01
-2.45764270e-01 6.55840989e-03 8.75381649e-01 1.17797685e+00
6.49534345e-01 1.53410482e+00 1.07630417e-01 3.95342708e-01
5.92304885e-01 4.49256420e-01 4.45380628e-01 -1.79549563e+00
-8.25199410e-02 -2.31274679e-01 -2.16662765e-01 -1.06459582e+00
9.10653472e-01 -1.14888012e+00 -7.40468949e-02 -1.84664583e+00
5.37443876e-01 1.74450710e-01 -2.00407833e-01 2.96184033e-01
-6.42993629e-01 -5.83381593e-01 -3.01549509e-02 3.02311461e-02
-8.43284309e-01 4.39309418e-01 1.56975436e+00 -5.56248069e-01
3.23872268e-02 -2.94302583e-01 -9.74316716e-01 1.00151432e+00
2.17216820e-01 -4.24120009e-01 -5.01817465e-01 -2.70152450e-01
2.67084599e-01 2.40698949e-01 -2.01799721e-01 -3.51088583e-01
6.15571618e-01 -4.43925202e-01 -6.11514151e-02 -5.31495750e-01
1.08635537e-01 -5.76655746e-01 -1.69992939e-01 -2.31053513e-02
-2.63110578e-01 8.32435936e-02 -2.00038806e-01 5.47753155e-01
-4.37384307e-01 -3.45191598e-01 1.99382037e-01 -2.96631753e-01
-9.29705977e-01 2.48244807e-01 9.69312117e-02 5.84492505e-01
4.39258605e-01 5.22140801e-01 -1.04578042e+00 -2.95163095e-02
-7.29987025e-01 4.11584139e-01 -1.14488840e-01 7.46574342e-01
4.88679200e-01 -1.29097033e+00 -7.80816436e-01 -3.23995620e-01
1.46584466e-01 1.86011672e-01 1.71824500e-01 4.05788213e-01
1.38087437e-01 2.56197691e-01 1.93884864e-01 -2.40533069e-01
-1.74251091e+00 7.94828773e-01 2.25007325e-01 -4.75996405e-01
-7.87624061e-01 8.52818131e-01 -1.65184528e-01 -3.60166609e-01
2.35511467e-01 1.76572189e-01 -9.74129140e-01 3.82681161e-01
5.95281482e-01 8.72235447e-02 1.12188026e-01 -8.36180687e-01
-8.84345233e-01 3.34300786e-01 -3.93294096e-01 3.09700161e-01
1.10395765e+00 -3.43638897e-01 -7.61184216e-01 2.91216165e-01
8.33630264e-01 4.54018027e-01 -3.05047691e-01 -5.65612495e-01
4.57619429e-01 -2.91823149e-01 1.72341391e-02 -6.69059694e-01
-8.24615777e-01 4.24111634e-01 -1.30204722e-01 4.11078066e-01
7.50089824e-01 4.48231339e-01 7.98754334e-01 8.09790671e-01
4.98663753e-01 -1.19223940e+00 -3.81677330e-01 1.01524663e+00
9.70161438e-01 -9.96918440e-01 1.78071097e-01 -1.28641570e+00
-5.64016938e-01 1.18711150e+00 9.88721550e-01 2.81448603e-01
4.97303635e-01 5.12675464e-01 1.27272725e-01 -5.26746035e-01
-7.36088693e-01 -5.01993060e-01 4.66283947e-01 6.91394687e-01
8.11071157e-01 2.04016611e-01 -9.91762459e-01 1.13916397e+00
-4.96636689e-01 -6.53279543e-01 4.98322129e-01 5.90836525e-01
-3.83926392e-01 -1.44526017e+00 -1.93082704e-03 4.51388121e-01
-4.27766383e-01 -5.15409172e-01 -7.94019401e-01 8.07313740e-01
1.80957079e-01 1.31725144e+00 1.27458513e-01 -4.12908137e-01
8.50453496e-01 -1.12629368e-03 4.55954492e-01 -9.02804732e-01
-5.17618895e-01 -1.82453945e-01 1.00386250e+00 -4.62729573e-01
-7.06522524e-01 -8.06340158e-01 -1.74407661e+00 -4.00317252e-01
-5.38485169e-01 5.76091647e-01 -4.95642647e-02 1.38425064e+00
1.99951500e-01 9.57050264e-01 2.96465307e-01 2.00407188e-02
-4.08478022e-01 -1.14949906e+00 -7.80674279e-01 7.84392238e-01
2.34194030e-03 -8.94414663e-01 -3.77226174e-01 -4.45348442e-01]
|
[9.30571174621582, 8.819731712341309]
|
e2243dbc-4248-4276-ac47-1fc5886086df
|
conformer-based-elderly-speech-recognition
|
2206.13232
| null |
https://arxiv.org/abs/2206.13232v1
|
https://arxiv.org/pdf/2206.13232v1.pdf
|
Conformer Based Elderly Speech Recognition System for Alzheimer's Disease Detection
|
Early diagnosis of Alzheimer's disease (AD) is crucial in facilitating preventive care to delay further progression. This paper presents the development of a state-of-the-art Conformer based speech recognition system built on the DementiaBank Pitt corpus for automatic AD detection. The baseline Conformer system trained with speed perturbation and SpecAugment based data augmentation is significantly improved by incorporating a set of purposefully designed modeling features, including neural architecture search based auto-configuration of domain-specific Conformer hyper-parameters in addition to parameter fine-tuning; fine-grained elderly speaker adaptation using learning hidden unit contributions (LHUC); and two-pass cross-system rescoring based combination with hybrid TDNN systems. An overall word error rate (WER) reduction of 13.6% absolute (34.8% relative) was obtained on the evaluation data of 48 elderly speakers. Using the final systems' recognition outputs to extract textual features, the best-published speech recognition based AD detection accuracy of 91.7% was obtained.
|
['Helen Meng', 'Xunying Liu', 'Zengrui Jin', 'Mingyu Cui', 'Yi Wang', 'Shoukang Hu', 'Zi Ye', 'Mengzhe Geng', 'Jiajun Deng', 'Tianzi Wang']
|
2022-06-23
| null | null | null | null |
['alzheimer-s-disease-detection']
|
['medical']
|
[ 5.57898462e-01 5.12615561e-01 2.96227545e-01 -7.95778394e-01
-1.54347849e+00 8.49107057e-02 6.05115950e-01 1.16004713e-01
-8.61278832e-01 8.08140159e-01 6.21469617e-01 -2.55467981e-01
-1.33492574e-01 -2.51839191e-01 -8.54031146e-02 -3.59223038e-01
-2.22782120e-01 8.85145485e-01 3.73590350e-01 -5.15634596e-01
-1.59635648e-01 1.59430653e-01 -1.47186863e+00 9.40439641e-01
9.38714862e-01 7.86121368e-01 3.18695128e-01 9.73729551e-01
5.11787012e-02 6.68915987e-01 -9.69839513e-01 -2.83697933e-01
-2.75887847e-01 -1.79677859e-01 -8.95796835e-01 2.15211529e-02
3.75432223e-01 -5.35302222e-01 -2.29215831e-01 6.97717249e-01
1.12912595e+00 -7.97880441e-02 4.75228459e-01 -4.00154561e-01
-6.57116055e-01 7.89535463e-01 2.58371383e-01 7.91940689e-01
2.45025739e-01 -3.68309245e-02 7.01401472e-01 -7.32807815e-01
2.23935395e-01 1.38376248e+00 6.78563476e-01 1.02901912e+00
-1.31623268e+00 -3.86014193e-01 -1.52613912e-02 6.41556501e-01
-1.31868470e+00 -1.12679708e+00 3.71053755e-01 -4.58861500e-01
1.74278951e+00 3.68474156e-01 7.27526248e-01 1.43480694e+00
-1.69388220e-01 5.04131317e-01 7.59786725e-01 -9.18770730e-01
5.71660638e-01 1.87228367e-01 5.86793542e-01 4.88573998e-01
-3.39581892e-02 5.29454015e-02 -3.97444457e-01 -6.61748767e-01
2.52250969e-01 -5.95532835e-01 -1.12470463e-01 4.14975286e-01
-1.03480852e+00 9.00649428e-01 2.31964085e-02 4.56999451e-01
-6.13670290e-01 -2.83512503e-01 7.55167007e-01 4.04739916e-01
6.12038136e-01 2.34088659e-01 -7.73220360e-01 -3.15290570e-01
-9.84531403e-01 2.40559042e-01 7.48895228e-01 8.12541127e-01
-1.05092280e-01 3.79653990e-01 -4.95176971e-01 1.50270092e+00
5.27436495e-01 5.37033558e-01 1.33263457e+00 -6.79437876e-01
5.97031534e-01 6.46495521e-01 -1.68000951e-01 1.07153736e-01
-7.02560127e-01 -4.91814882e-01 -5.82034826e-01 -1.64226636e-01
1.68898985e-01 -3.53867531e-01 -1.28505790e+00 1.58088887e+00
3.85620408e-02 -4.33438748e-01 4.06329900e-01 5.48458934e-01
5.23373544e-01 5.33083677e-01 4.26769108e-01 -2.79416770e-01
1.88071883e+00 -6.49326324e-01 -9.66528594e-01 -5.44703960e-01
9.52688456e-01 -5.66489935e-01 9.12519336e-01 3.94298136e-01
-9.55787063e-01 -4.25478041e-01 -1.32911277e+00 1.65142775e-01
-2.58496821e-01 5.33858895e-01 -1.91142727e-02 1.35748804e+00
-1.27658820e+00 8.11584443e-02 -9.98758018e-01 -6.63072288e-01
4.44715858e-01 7.88508773e-01 -3.17112714e-01 1.44738942e-01
-1.45159984e+00 1.11817503e+00 4.02189940e-01 -1.06429741e-01
-5.42768180e-01 -6.75543725e-01 -6.44354284e-01 -2.59444445e-01
-7.21927211e-02 -8.12755883e-01 1.67363441e+00 -7.85007179e-01
-1.64293373e+00 9.13752913e-01 -1.92212403e-01 -1.13885093e+00
3.78414154e-01 -5.23439348e-01 -1.08826387e+00 2.27842167e-01
-3.14877294e-02 3.64381939e-01 6.06263101e-01 -3.78005534e-01
-6.30352139e-01 -8.41251254e-01 -8.29502404e-01 6.37811571e-02
-5.91090620e-01 5.06061554e-01 2.96819776e-01 -7.27339149e-01
-2.20215499e-01 -7.69655645e-01 -4.73748706e-02 -2.36441687e-01
-4.16567355e-01 -2.16434792e-01 5.50925970e-01 -1.37261081e+00
1.63459253e+00 -1.84490514e+00 -2.54617512e-01 2.32966356e-02
8.66751149e-02 7.28580236e-01 -3.52040492e-02 1.30738646e-01
-2.29490981e-01 -1.11790270e-01 -3.74179900e-01 -3.43601733e-01
7.49680176e-02 1.16638198e-01 2.41074830e-01 2.36353338e-01
3.42864156e-01 5.54869413e-01 -2.26185143e-01 -1.87614068e-01
2.35301554e-01 8.03115189e-01 -6.45920515e-01 2.23835647e-01
4.40442935e-02 -8.00928921e-02 -3.20595086e-01 4.72637713e-01
3.06537864e-03 2.27329627e-01 5.72800115e-02 -2.97202375e-02
5.41773289e-02 7.77867615e-01 -8.15368354e-01 1.19626665e+00
-1.65049583e-01 3.06205779e-01 -6.75497800e-02 -9.15736258e-01
1.11071563e+00 5.68279028e-01 9.20468122e-02 -8.09204102e-01
2.75076509e-01 3.99128109e-01 3.53825420e-01 -7.94725060e-01
2.90690884e-02 3.46960761e-02 1.91471338e-01 -5.26859388e-02
4.61237431e-02 7.51525581e-01 -1.40413195e-01 -2.29383245e-01
1.31982183e+00 -4.37017769e-01 5.52224636e-01 -3.33661437e-01
8.04677546e-01 -1.09859340e-01 3.36346835e-01 6.45720720e-01
-5.00295937e-01 5.70944369e-01 -3.63168158e-02 -2.55174696e-01
-1.23661590e+00 -6.80575311e-01 -3.67945969e-01 1.17110837e+00
-1.26594949e+00 -5.10466337e-01 -1.32993495e+00 -4.14323747e-01
-3.84642154e-01 1.16065741e+00 -4.92872596e-01 -3.77448231e-01
-8.99181962e-01 -1.14476478e+00 9.69589949e-01 5.60671687e-01
5.17137527e-01 -8.88416529e-01 -4.23043698e-01 6.43110991e-01
-5.83932968e-03 -1.01493585e+00 -4.54113007e-01 5.35493493e-01
-9.18303847e-01 -7.97824264e-01 -1.23045790e+00 -1.00399959e+00
1.29015073e-01 -5.38054526e-01 6.45964265e-01 -5.16228557e-01
-3.61072898e-01 3.03363442e-01 -3.97819459e-01 -2.98420727e-01
-1.03045893e+00 4.11388993e-01 2.72179335e-01 -9.64027569e-02
8.48713517e-01 -4.69388962e-01 -4.38069791e-01 1.28602326e-01
-2.85109639e-01 -4.11317885e-01 8.64796281e-01 9.83153939e-01
2.46147931e-01 -5.24772048e-01 1.11578262e+00 -4.74566311e-01
7.99419999e-01 -1.58335134e-01 -1.99225500e-01 3.04932147e-01
-8.64538491e-01 3.41145098e-01 1.30611658e-01 -6.05671883e-01
-1.09108591e+00 2.26292461e-01 -7.31586158e-01 3.50681931e-01
-5.38660824e-01 1.35658249e-01 -5.10336757e-01 4.05453652e-01
1.14838839e+00 4.69243258e-01 3.61815333e-01 -7.64034152e-01
4.67060804e-02 1.45039511e+00 5.40999293e-01 -1.19312517e-01
-9.63830426e-02 -2.76368231e-01 -8.12728584e-01 -1.35081613e+00
-3.00695240e-01 -3.80122185e-01 -4.90184188e-01 1.37737393e-01
1.00041723e+00 -9.97457922e-01 -1.29454643e-01 6.56397402e-01
-1.07500041e+00 -3.91293675e-01 -1.58342138e-01 5.47783256e-01
-3.52446586e-01 2.47283682e-01 -4.95152473e-01 -1.02380395e+00
-1.12862420e+00 -1.17898166e+00 1.06091332e+00 -2.97019660e-01
-7.69176066e-01 -6.34790540e-01 2.06246048e-01 7.47974694e-01
5.32976031e-01 -5.37228167e-01 1.05360532e+00 -1.64373064e+00
4.44653392e-01 -3.16017151e-01 9.75930244e-02 7.70736694e-01
1.87138334e-01 -6.26247108e-01 -1.00583208e+00 -5.82909547e-02
5.24860844e-02 1.26981437e-01 4.68391776e-01 5.99577606e-01
2.19292343e-01 -5.51832557e-01 -2.59622157e-01 -6.68011531e-02
6.60692155e-01 6.17490530e-01 7.97799349e-01 8.50597680e-01
3.75265062e-01 3.62516344e-01 -2.45270357e-02 4.37173307e-01
4.28493798e-01 1.03732181e+00 -1.95175827e-01 4.17545974e-01
-4.76599872e-01 4.10688370e-01 7.40551889e-01 5.28423965e-01
7.61719793e-02 1.67997777e-02 -1.07298660e+00 4.10661697e-01
-1.45865774e+00 -7.70232737e-01 -2.25326359e-01 2.12313771e+00
9.73935366e-01 2.44666025e-01 6.48933530e-01 4.81389076e-01
1.02394056e+00 -4.04908031e-01 -4.85417873e-01 -3.08295608e-01
-2.10073441e-01 2.56807208e-01 4.66259748e-01 5.18468618e-01
-1.03115332e+00 6.91522002e-01 6.08026314e+00 4.09008861e-01
-6.95862710e-01 4.23621118e-01 4.90894377e-01 -3.16779792e-01
3.67429316e-01 -7.32681692e-01 -1.17831218e+00 5.34564435e-01
2.05773926e+00 1.08892128e-01 1.29855737e-01 9.08883393e-01
4.56846833e-01 2.66096771e-01 -8.94774795e-01 1.01114202e+00
2.39514366e-01 -8.95905912e-01 5.77547327e-02 2.13863373e-01
-3.04527096e-02 2.35287219e-01 -2.88824081e-01 3.57776791e-01
-1.09734029e-01 -5.07534266e-01 9.18251753e-01 4.93565083e-01
6.57868326e-01 -7.57246673e-01 7.90777802e-01 -1.03566855e-01
-7.03920245e-01 -3.54486108e-01 1.00827016e-01 3.49481612e-01
2.94309735e-01 3.69026482e-01 -1.43371344e+00 -3.08971144e-02
6.27204001e-01 6.52231053e-02 -6.57749176e-01 8.98742676e-01
1.49533376e-01 9.58324432e-01 -4.39322114e-01 -2.49245718e-01
-8.13982338e-02 5.87325215e-01 7.94102550e-01 1.52923512e+00
4.88767564e-01 -5.32378778e-02 -3.64491522e-01 3.48045856e-01
3.37157190e-01 2.34481081e-01 1.17425889e-01 -2.68625841e-02
6.49023175e-01 5.55469871e-01 1.02791227e-01 -5.30661285e-01
-1.41761899e-01 1.14025390e+00 1.12208448e-01 4.58374210e-02
-3.38583738e-01 -4.38576162e-01 5.94281256e-01 4.28703815e-01
3.91438276e-01 8.81417189e-03 -1.37089908e-01 -6.29369855e-01
1.81398317e-01 -1.15919363e+00 6.30573928e-01 -5.07826209e-01
-1.17213750e+00 1.11890137e+00 -9.61911529e-02 -6.85182691e-01
-7.46588111e-01 -6.60882354e-01 -2.45921060e-01 1.06777072e+00
-1.02161407e+00 -1.05954027e+00 1.72737941e-01 5.62989354e-01
9.09039855e-01 -1.01757503e+00 1.37235177e+00 6.96870863e-01
-8.36505473e-01 1.02370095e+00 -3.45223732e-02 -6.61722347e-02
8.22452128e-01 -1.20339191e+00 5.32069087e-01 5.30453086e-01
-4.48216289e-01 5.25494456e-01 7.36485243e-01 -8.00303698e-01
-8.02311599e-01 -1.32127905e+00 1.42959738e+00 -2.05580652e-01
6.85278654e-01 -3.92727524e-01 -1.01553512e+00 5.40368319e-01
4.95666191e-02 -6.34909391e-01 7.68165290e-01 -4.28581573e-02
-1.61608621e-01 -1.84512064e-01 -1.26072443e+00 5.12997746e-01
8.19829226e-01 -6.15693152e-01 -1.07811141e+00 3.61568391e-01
7.80100942e-01 -2.42814198e-02 -9.90692258e-01 2.34826490e-01
5.46839476e-01 -6.11235738e-01 1.15332437e+00 -6.47114873e-01
-2.68914193e-01 1.37612402e-01 -5.97176611e-01 -1.09855318e+00
-3.34497422e-01 -7.69648850e-01 -1.55997783e-01 1.40296471e+00
7.41999686e-01 -8.45656157e-01 4.21189904e-01 9.83153939e-01
-6.38691962e-01 -4.33015794e-01 -1.30642056e+00 -8.26124072e-01
-1.66070834e-01 -6.03434682e-01 3.13014537e-01 2.65290439e-01
1.88649252e-01 6.36986494e-01 -5.12949601e-02 2.54942209e-01
3.56195390e-01 -1.47318912e+00 -6.37319759e-02 -1.37101889e+00
-8.47437084e-02 -3.40836465e-01 -7.13351488e-01 -2.94592351e-01
-4.15400863e-02 -7.63547063e-01 -1.05307251e-02 -1.32111371e+00
-2.71895621e-02 4.15606834e-02 -1.27728164e-01 7.40541816e-01
-1.92765873e-02 -3.62708122e-01 -3.32088888e-01 7.34549686e-02
-4.14241292e-02 6.90472960e-01 1.98145390e-01 -2.76764989e-01
-6.89306676e-01 3.25716019e-01 -6.94954634e-01 4.08911198e-01
1.07782733e+00 -4.57656562e-01 -4.19261724e-01 -7.91595355e-02
-5.70268571e-01 -2.14283496e-01 4.95640606e-01 -1.24007308e+00
1.34518393e-03 6.51125669e-01 3.05698514e-01 -3.70318592e-01
5.52417517e-01 -3.34277630e-01 -6.55532554e-02 7.21519768e-01
-7.19133317e-01 -1.06243148e-01 3.99983555e-01 4.76447642e-01
3.16385001e-01 -2.43661895e-01 8.81152451e-01 5.54861687e-02
-5.56203485e-01 -2.99627006e-01 -1.14774048e+00 -2.02604253e-02
4.40714687e-01 -1.94428772e-01 -4.73099232e-01 -7.29103163e-02
-1.21782231e+00 -3.90671939e-01 -3.81473094e-01 4.27281231e-01
4.26676929e-01 -1.18080759e+00 -1.07801008e+00 2.46975422e-01
3.47675592e-01 -6.48786545e-01 3.38730961e-01 7.89990485e-01
-1.38053268e-01 7.53278732e-01 -4.12698127e-02 -2.79426455e-01
-1.66059029e+00 -2.71702781e-02 6.74296618e-01 -4.49786261e-02
-1.04705405e+00 7.63507545e-01 -4.39165384e-01 -7.32095912e-02
6.14308953e-01 -3.85149896e-01 -3.32104057e-01 1.26805872e-01
1.18048418e+00 6.22662663e-01 8.24881136e-01 -4.89214629e-01
-6.67237163e-01 -2.98240393e-01 -6.94038630e-01 -6.77696943e-01
1.37346673e+00 -2.05038562e-01 4.25728291e-01 3.16930026e-01
1.11875224e+00 -7.17361867e-01 -6.55996203e-01 -3.06349784e-01
4.34574813e-01 6.41229868e-01 6.31085932e-01 -1.39612269e+00
-4.54802096e-01 4.23133194e-01 1.62078667e+00 2.08443820e-01
8.62078130e-01 2.04545110e-01 9.17964339e-01 7.43038535e-01
-8.49412233e-02 -1.32991147e+00 -4.17827606e-01 3.61188054e-01
1.17791009e+00 -9.26024377e-01 -3.64838094e-01 5.23237325e-02
-6.24230683e-01 1.23296285e+00 1.41716257e-01 2.58282483e-01
5.53552091e-01 3.01884174e-01 1.50221571e-01 -6.08591847e-02
-6.96960568e-01 -2.14382246e-01 4.97398227e-01 9.53073919e-01
5.62229812e-01 1.08530685e-01 -2.29200393e-01 1.24602652e+00
-4.17189389e-01 -1.12724528e-01 7.25720674e-02 9.14427638e-01
-9.20615613e-01 -1.21596837e+00 -5.28424323e-01 7.55833924e-01
-5.12193739e-01 -4.36982632e-01 -4.75742549e-01 5.79898536e-01
-2.12170258e-01 1.10470057e+00 -1.43259419e-02 -2.87672549e-01
7.50392973e-01 8.03087473e-01 2.90585876e-01 -6.43097520e-01
-7.16941953e-01 3.18420261e-01 1.01756012e+00 -2.45007366e-01
-2.33204544e-01 -1.26206505e+00 -1.25256622e+00 3.94187182e-01
1.68389305e-02 -2.03542531e-01 8.25963974e-01 1.28751945e+00
5.95938623e-01 9.88685846e-01 -2.07377840e-02 -4.78212565e-01
-8.21246386e-01 -1.64842105e+00 -3.84754360e-01 -4.21196431e-01
5.49851418e-01 -3.86094987e-01 -2.09372073e-01 3.73691261e-01]
|
[13.96345043182373, 5.465603828430176]
|
56518bc6-3f7e-4134-9570-93723856b9f6
|
a-preliminary-study-on-environmental-sound
| null | null |
https://aclanthology.org/2021.rocling-1.14
|
https://aclanthology.org/2021.rocling-1.14.pdf
|
A Preliminary Study on Environmental Sound Classification Leveraging Large-Scale Pretrained Model and Semi-Supervised Learning
|
With the widespread commercialization of smart devices, research on environmental sound classification has gained more and more attention in recent years. In this paper, we set out to make effective use of large-scale audio pretrained model and semi-supervised model training paradigm for environmental sound classification. To this end, an environmental sound classification method is first put forward, whose component model is built on top a large-scale audio pretrained model. Further, to simulate a low-resource sound classification setting where only limited supervised examples are made available, we instantiate the notion of transfer learning with a recently proposed training algorithm (namely, FixMatch) and a data augmentation method (namely, SpecAugment) to achieve the goal of semi-supervised model training. Experiments conducted on bench-mark dataset UrbanSound8K reveal that our classification method can lead to an accuracy improvement of 2.4% in relation to a current baseline method.
|
['Berlin Chen', 'Shi-Yan Weng', 'Jiun-Ting Li', 'Tien-Hong Lo', 'You-Sheng Tsao']
| null | null | null | null |
rocling-2021-10
|
['environmental-sound-classification', 'sound-classification']
|
['audio', 'audio']
|
[ 5.51531136e-01 -5.89181669e-02 1.99466616e-01 -3.91082585e-01
-9.74821270e-01 -3.20327610e-01 5.17103553e-01 9.88503844e-02
-3.13785642e-01 3.33204240e-01 1.27079591e-01 -3.49995673e-01
2.11860642e-01 -9.45227861e-01 -6.55660808e-01 -4.76730168e-01
-4.07613488e-03 -6.37588277e-02 1.85637847e-01 -9.85341519e-03
-1.74181297e-01 6.16900437e-02 -1.65963662e+00 2.49872684e-01
9.26930726e-01 1.07601178e+00 2.14417055e-01 9.21131730e-01
5.10645360e-02 6.68988764e-01 -5.48063517e-01 -1.64745376e-01
2.12793693e-01 -4.89747822e-01 -8.88769269e-01 -1.30968124e-01
4.72934693e-01 -2.73135930e-01 -8.44712779e-02 9.29374516e-01
7.64427602e-01 3.93686324e-01 3.02233309e-01 -1.25551760e+00
-9.03925002e-02 9.29951727e-01 -6.94937482e-02 8.85637552e-02
1.26535252e-01 1.45790145e-01 1.20760834e+00 -6.73940539e-01
-3.30638029e-02 9.07694161e-01 8.93222332e-01 4.95303601e-01
-1.38748562e+00 -9.52101171e-01 1.63224369e-01 1.39655724e-01
-1.35937619e+00 -6.43952906e-01 1.00462484e+00 -3.40098500e-01
7.80173540e-01 3.16368520e-01 7.16295481e-01 1.08799410e+00
-4.37033325e-01 7.25315928e-01 1.34269893e+00 -7.04635978e-01
5.17152190e-01 2.50057399e-01 4.47438061e-02 2.89631784e-01
-1.65354580e-01 1.28456652e-01 -6.69026196e-01 -3.27206731e-01
3.33269894e-01 -3.38996679e-01 -1.81782544e-01 6.75779255e-03
-7.53704011e-01 5.98371148e-01 4.29611534e-01 3.51354957e-01
-2.54556209e-01 3.55171323e-01 4.31362063e-01 2.79048175e-01
8.44774365e-01 4.12813991e-01 -4.94794458e-01 -4.15568799e-01
-9.82777119e-01 1.96480617e-01 6.72481060e-01 6.75567806e-01
7.61584520e-01 2.55442649e-01 2.57835239e-01 1.09145129e+00
3.76325637e-01 5.46804309e-01 5.48080206e-01 -5.89439988e-01
4.43758488e-01 2.32988417e-01 -5.00870720e-02 -7.25513756e-01
-2.84275025e-01 -7.43090391e-01 -7.33466268e-01 -8.26692358e-02
5.54357953e-02 -1.50441363e-01 -8.09904933e-01 1.74157548e+00
3.90302241e-01 1.06415486e+00 1.35576993e-01 5.74221611e-01
7.09507704e-01 8.14062774e-01 1.34034365e-01 8.26915652e-02
9.45875585e-01 -1.22527611e+00 -4.27369922e-01 -2.61684537e-01
5.74057817e-01 -4.46745068e-01 1.48277295e+00 6.17550075e-01
-6.22751772e-01 -9.17850018e-01 -1.12969828e+00 2.99760044e-01
-4.81320083e-01 6.22296184e-02 5.54753602e-01 1.02424896e+00
-7.36330688e-01 4.91395682e-01 -9.04137909e-01 -2.42704868e-01
3.53849024e-01 2.60990679e-01 -1.25572845e-01 1.23957194e-01
-1.05610633e+00 1.88464910e-01 3.61636162e-01 1.86105505e-01
-9.57147062e-01 -1.00637984e+00 -6.20806813e-01 -2.19630580e-02
3.74480873e-01 -5.05750895e-01 1.46387279e+00 -8.79808664e-01
-1.87370229e+00 5.53203642e-01 3.24594915e-01 -5.53370416e-01
3.33298415e-01 -5.97522259e-01 -6.51557267e-01 7.22149387e-02
-1.03128394e-02 3.80343080e-01 1.01196289e+00 -1.35840297e+00
-7.38735974e-01 -7.98067376e-02 4.81830835e-02 -1.76380500e-01
-8.31391692e-01 9.04818401e-02 -1.48188174e-01 -7.17586219e-01
-2.79887050e-01 -1.04635894e+00 -2.29270816e-01 -5.19888341e-01
-5.63954830e-01 3.48914824e-02 5.14139056e-01 -6.32351100e-01
1.45676935e+00 -2.33692098e+00 -2.26355851e-01 4.72609192e-01
-6.05709851e-03 5.15646636e-01 -3.32576632e-01 3.21863472e-01
-1.92895085e-01 2.51987446e-02 -5.24942577e-01 -7.04350471e-01
3.55090834e-02 1.10779263e-01 -5.16950071e-01 5.21476381e-02
1.62319824e-01 5.01090944e-01 -1.10953069e+00 -1.89038306e-01
1.49634838e-01 3.95677716e-01 -9.53939617e-01 3.58260870e-01
-2.91053742e-01 6.08421743e-01 -5.03265440e-01 3.94492209e-01
6.34078622e-01 1.13861904e-01 -8.17088485e-02 8.00725818e-03
-1.68906584e-01 5.28464496e-01 -1.32561505e+00 1.81191444e+00
-1.06103647e+00 3.84644866e-01 3.98249030e-02 -8.47256720e-01
7.27016568e-01 3.44146430e-01 4.28195000e-01 -4.90893036e-01
6.79239482e-02 3.80094677e-01 -5.65364864e-03 -4.41716582e-01
5.07402003e-01 -1.52970240e-01 -4.46893834e-02 4.55829978e-01
3.39253433e-02 -2.45387495e-01 -3.04180473e-01 2.86680497e-02
1.14478159e+00 2.26635247e-01 -4.31312956e-02 -5.22387922e-02
5.43010056e-01 -2.33810589e-01 3.21474910e-01 6.32571518e-01
-7.30952993e-02 5.03316164e-01 -1.11995198e-01 -8.29925239e-02
-6.45495117e-01 -6.16002381e-01 -1.53906554e-01 1.48838675e+00
-2.33783767e-01 -7.79528975e-01 -1.05686259e+00 -7.05088198e-01
-7.52468333e-02 6.02345645e-01 -5.16165435e-01 -2.84081250e-01
-6.34854376e-01 -7.52745569e-01 9.15087640e-01 8.10084045e-01
7.31131196e-01 -1.09832013e+00 -7.69145668e-01 3.22432011e-01
-1.04341634e-01 -1.22227120e+00 -9.26322117e-02 4.86304343e-01
-5.84902287e-01 -6.57867074e-01 -3.29070270e-01 -5.22793233e-01
1.80982918e-01 1.08792603e-01 9.59587038e-01 4.91973013e-03
-1.66714862e-01 5.35209537e-01 -6.98041499e-01 -7.34723151e-01
-6.54933453e-01 4.32368159e-01 2.57364005e-01 4.19816941e-01
1.63913265e-01 -9.23093736e-01 -4.85246152e-01 2.63713021e-02
-1.01216996e+00 1.12390414e-01 4.73803282e-01 7.87766159e-01
5.30237317e-01 1.43999740e-01 1.00282788e+00 -8.21830988e-01
4.87430930e-01 -3.86273086e-01 -4.53462213e-01 -6.90195188e-02
-6.05425000e-01 -2.12724507e-01 9.30959702e-01 -3.98290455e-01
-1.12852621e+00 1.56499803e-01 -5.54997981e-01 -3.47471267e-01
-4.19585407e-01 6.86771154e-01 -4.52736795e-01 8.16462189e-02
7.00283885e-01 1.94045603e-01 -3.15906107e-01 -8.74986231e-01
4.60140467e-01 1.25629520e+00 6.02930844e-01 -5.68985701e-01
1.06744313e+00 2.83058405e-01 -2.16743633e-01 -8.58679354e-01
-9.34943020e-01 -4.96616721e-01 -7.42744803e-01 -1.99563757e-01
6.33410871e-01 -9.93422329e-01 -2.05935940e-01 6.82440400e-01
-6.94225430e-01 -9.81111765e-01 -3.46077055e-01 4.04197454e-01
-4.41231936e-01 1.01827726e-01 -2.76783139e-01 -1.02314961e+00
-3.56027246e-01 -8.32784534e-01 1.14703560e+00 -1.30636126e-01
-8.52792934e-02 -8.74634147e-01 4.44946975e-01 3.06029737e-01
6.05628014e-01 -1.16866725e-02 6.65276885e-01 -9.44520712e-01
-3.11364383e-01 -1.89922705e-01 8.17334503e-02 8.02112520e-01
2.70955890e-01 -1.86882690e-01 -1.71542132e+00 -6.93998486e-02
-1.84596896e-01 -4.39569205e-01 9.56181169e-01 -1.08558632e-01
1.29465616e+00 -2.70371288e-02 -1.57733753e-01 5.43431401e-01
1.15861428e+00 1.67342275e-01 3.81498694e-01 3.79015207e-01
6.79899156e-01 2.87725061e-01 6.21786892e-01 4.24748659e-01
2.23178580e-01 8.60145807e-01 3.14572841e-01 -1.45461019e-02
-3.81618798e-01 -5.72635829e-01 2.43607149e-01 1.09430659e+00
-6.19739108e-03 -1.57292694e-01 -9.66050208e-01 4.68169659e-01
-1.56722593e+00 -6.24574900e-01 1.03170961e-01 2.27892756e+00
9.25016999e-01 7.93513805e-02 2.89949834e-01 7.69694448e-01
3.15370619e-01 1.76314116e-01 -2.41607457e-01 -5.40616177e-03
3.78792793e-01 4.86803353e-01 -8.16060826e-02 3.43210608e-01
-1.43050456e+00 1.03890026e+00 5.95830202e+00 9.06745672e-01
-1.45692956e+00 1.06824830e-01 4.19513255e-01 1.73810735e-01
-1.31244212e-01 -2.65246779e-01 -6.96682513e-01 5.31812251e-01
1.37141502e+00 1.82969555e-01 5.24628043e-01 9.32891607e-01
3.00816447e-01 1.84891403e-01 -1.10073125e+00 9.51909900e-01
-9.89596695e-02 -7.76346684e-01 -1.50863647e-01 1.72655229e-02
5.63353360e-01 3.18251178e-02 1.68780267e-01 6.29209936e-01
1.15209572e-01 -7.96838522e-01 8.04378808e-01 1.43797815e-01
9.03930426e-01 -7.18169510e-01 6.07396424e-01 3.96159023e-01
-1.47331512e+00 -2.58973241e-01 -1.58742070e-01 -2.35087276e-01
3.90690565e-02 5.60343444e-01 -8.51051927e-01 7.50514269e-01
8.68273735e-01 7.20920384e-01 -5.25324523e-01 1.16128361e+00
-1.39570639e-01 1.50824094e+00 -4.36561793e-01 1.14118956e-01
1.26311123e-01 -2.10193992e-02 3.50945503e-01 1.38225543e+00
2.87268281e-01 8.03190172e-02 2.09139138e-01 5.53966582e-01
-1.42926365e-01 4.20828372e-01 -3.76849592e-01 1.07063726e-02
4.83104050e-01 1.27755606e+00 -4.13793594e-01 -4.24198776e-01
-3.24111491e-01 8.53881180e-01 8.72034803e-02 6.60514385e-02
-8.64165008e-01 -3.99060488e-01 5.59944212e-01 -6.78601637e-02
2.44914800e-01 6.82812259e-02 -1.25636801e-01 -9.37338650e-01
-7.86882490e-02 -9.48406279e-01 8.13848972e-02 -5.36204934e-01
-1.05640948e+00 7.99969018e-01 -4.65466790e-02 -1.37692571e+00
-2.55675912e-01 -3.53772193e-01 -8.51527572e-01 6.29629493e-01
-1.78554177e+00 -1.32493103e+00 -6.48459494e-01 3.13009381e-01
5.51775038e-01 -1.45038590e-01 1.17221713e+00 5.67757487e-01
-6.20791376e-01 7.72855878e-01 1.70512479e-02 1.99098438e-01
6.96633875e-01 -1.31194556e+00 4.63604182e-01 7.42951453e-01
5.20194709e-01 3.93103778e-01 6.91631198e-01 -3.17296624e-01
-1.18586600e+00 -1.62026584e+00 4.90258753e-01 -4.17878807e-01
8.78991485e-01 -6.55412436e-01 -9.77763176e-01 5.24986088e-01
-2.45748106e-02 2.81762660e-01 1.09196401e+00 4.06372361e-02
-5.68768620e-01 -5.24140477e-01 -9.68345225e-01 3.19652885e-01
1.08714163e+00 -7.86426485e-01 -4.12127465e-01 1.18493512e-01
8.22214246e-01 -2.12946579e-01 -8.21476460e-01 4.41787899e-01
5.61558545e-01 -5.84512115e-01 9.02898967e-01 -6.17491484e-01
2.97582567e-01 -2.13210851e-01 -3.15851182e-01 -1.63327360e+00
3.14140543e-02 -7.15929270e-01 -7.98529014e-02 1.58842087e+00
3.52214634e-01 -5.28765798e-01 7.71981001e-01 3.12742501e-01
-4.93994504e-01 -5.80424488e-01 -9.38030124e-01 -8.70507121e-01
1.01709202e-01 -1.09882998e+00 7.66470730e-01 8.37463558e-01
5.30894436e-02 4.43502843e-01 -4.11647171e-01 3.41251999e-01
3.06231886e-01 1.53618842e-01 1.13383472e+00 -1.40475559e+00
-6.50448620e-01 -9.15096104e-02 -2.91652232e-01 -1.25692725e+00
3.52335602e-01 -8.28810811e-01 4.49472725e-01 -1.12344372e+00
8.27456266e-02 -8.84776950e-01 -6.49809599e-01 7.59676218e-01
-2.61294723e-01 5.58603048e-01 -7.61982659e-03 -1.57413796e-01
-5.60747802e-01 7.60767460e-01 7.12586820e-01 -1.63716465e-01
-5.78324437e-01 3.29413176e-01 -6.50986671e-01 7.02619970e-01
8.44373703e-01 -4.15970296e-01 -5.34189343e-01 -2.90124059e-01
1.16305977e-01 -2.58311391e-01 4.53773707e-01 -1.40104926e+00
7.08900020e-02 1.46528080e-01 -3.00066471e-01 -1.22258708e-01
4.34225678e-01 -9.52232182e-01 1.13447765e-02 2.31584400e-01
-4.70234156e-01 -5.18507600e-01 3.77503276e-01 6.84654295e-01
-3.23536813e-01 -1.13264032e-01 5.03121972e-01 2.78676003e-01
-5.01930952e-01 1.25371680e-01 -3.35400581e-01 -2.10756153e-01
6.06893897e-01 1.02819033e-01 -5.67018874e-02 -2.51722515e-01
-6.62250698e-01 -1.41018063e-01 8.80429428e-03 4.38135892e-01
3.30202132e-01 -1.25981474e+00 -6.19312704e-01 3.28430206e-01
3.65414560e-01 -3.95569503e-02 1.45392910e-01 6.75501108e-01
-7.57170767e-02 3.29941273e-01 2.65589297e-01 -3.94926816e-01
-1.29623985e+00 2.56356418e-01 3.41579139e-01 -1.54956669e-01
-6.41268551e-01 7.84940481e-01 -3.29020731e-02 -7.43139148e-01
3.84277105e-01 -7.73754716e-01 -1.81076869e-01 -2.03269511e-01
6.56546652e-01 3.71533334e-01 3.19696188e-01 -5.92230260e-01
-1.32262722e-01 3.77144724e-01 3.94538462e-01 -1.77735686e-01
1.54062271e+00 1.87259302e-01 2.23011553e-01 7.07022369e-01
1.03717005e+00 4.76874143e-01 -1.01934373e+00 -4.35693890e-01
-1.34655312e-01 -3.97573113e-01 3.17254514e-01 -8.51418555e-01
-9.27789927e-01 1.04436922e+00 8.32318187e-01 4.83356923e-01
1.31829047e+00 -1.29989475e-01 9.58962739e-01 5.54713607e-01
5.07631898e-01 -9.33191597e-01 2.13111043e-01 3.75406861e-01
6.05116546e-01 -1.17777050e+00 -5.26195645e-01 -6.34821236e-01
-2.44386688e-01 7.19744861e-01 5.41850746e-01 8.27344283e-02
8.93191755e-01 4.58700210e-01 2.05378160e-01 1.83528423e-01
-6.39153004e-01 -2.80822724e-01 2.49077216e-01 6.24192774e-01
2.27530956e-01 1.20485641e-01 4.68868226e-01 1.16925704e+00
-5.07132649e-01 1.13990329e-01 1.10125124e-01 8.38237643e-01
-5.68288386e-01 -1.13592756e+00 -2.56448507e-01 4.15846616e-01
-4.85170513e-01 -2.57451922e-01 -1.87093392e-01 4.11524653e-01
2.25760058e-01 1.16804671e+00 2.97589097e-02 -7.96832979e-01
4.53525662e-01 2.63770878e-01 9.10576209e-02 -8.06716561e-01
-8.01671267e-01 2.30745733e-01 1.25139773e-01 -3.62383395e-01
-6.06215298e-01 -5.42846680e-01 -1.08920825e+00 4.09706980e-01
-5.65100908e-01 3.21329743e-01 5.88784814e-01 8.94372761e-01
2.52572745e-01 8.37815881e-01 8.74250591e-01 -9.16315079e-01
-6.51739717e-01 -1.15600598e+00 -5.96082509e-01 2.26933181e-01
3.30007583e-01 -4.69649971e-01 -5.01120806e-01 2.51062661e-01]
|
[15.169901847839355, 5.2015767097473145]
|
1efa69b6-4653-4914-9abf-ed5a23ff7d73
|
somoformer-multi-person-pose-forecasting-with
|
2208.14023
| null |
https://arxiv.org/abs/2208.14023v1
|
https://arxiv.org/pdf/2208.14023v1.pdf
|
SoMoFormer: Multi-Person Pose Forecasting with Transformers
|
Human pose forecasting is a challenging problem involving complex human body motion and posture dynamics. In cases that there are multiple people in the environment, one's motion may also be influenced by the motion and dynamic movements of others. Although there are several previous works targeting the problem of multi-person dynamic pose forecasting, they often model the entire pose sequence as time series (ignoring the underlying relationship between joints) or only output the future pose sequence of one person at a time. In this paper, we present a new method, called Social Motion Transformer (SoMoFormer), for multi-person 3D pose forecasting. Our transformer architecture uniquely models human motion input as a joint sequence rather than a time sequence, allowing us to perform attention over joints while predicting an entire future motion sequence for each joint in parallel. We show that with this problem reformulation, SoMoFormer naturally extends to multi-person scenes by using the joints of all people in a scene as input queries. Using learned embeddings to denote the type of joint, person identity, and global position, our model learns the relationships between joints and between people, attending more strongly to joints from the same or nearby people. SoMoFormer outperforms state-of-the-art methods for long-term motion prediction on the SoMoF benchmark as well as the CMU-Mocap and MuPoTS-3D datasets. Code will be made available after publication.
|
['Hamid Rezatofighi', 'Ehsan Adeli', 'Satyajit Kumar', 'Edward Vendrow']
|
2022-08-30
| null | null | null | null |
['multi-person-pose-forecasting', 'human-pose-forecasting']
|
['computer-vision', 'computer-vision']
|
[-3.04886997e-01 -2.03785107e-01 2.08188519e-02 -1.15292564e-01
-2.77568072e-01 -4.19346809e-01 7.39166796e-01 -4.93953586e-01
-4.83046293e-01 3.22132766e-01 9.10029233e-01 5.77071309e-01
1.61008626e-01 -4.43874091e-01 -5.64604282e-01 -5.81881225e-01
-1.08634569e-01 1.14036071e+00 2.85791218e-01 -3.98850083e-01
-2.97767222e-01 3.61700743e-01 -1.38819540e+00 2.35522151e-01
-5.73153086e-02 5.82066894e-01 -3.21301296e-02 9.11764741e-01
3.89682889e-01 6.17194057e-01 -7.24051058e-01 -5.00586092e-01
2.61220247e-01 -4.08511162e-01 -8.89104486e-01 2.73018271e-01
6.78258598e-01 -4.27618027e-01 -7.09526896e-01 3.11620772e-01
7.61191010e-01 5.65757990e-01 7.71207452e-01 -1.32350028e+00
-3.74245614e-01 5.17176725e-02 -5.11593699e-01 1.08492598e-01
8.11306715e-01 3.41867775e-01 8.71581614e-01 -7.42833018e-01
8.53731751e-01 1.75772583e+00 7.64926612e-01 8.60773623e-01
-1.06022310e+00 -2.00080156e-01 4.70579326e-01 3.89022559e-01
-1.10773647e+00 -1.21970408e-01 7.62258410e-01 -8.16155553e-01
1.06760287e+00 1.97560087e-01 1.23955595e+00 1.48485672e+00
5.08345485e-01 1.06554425e+00 3.71445924e-01 2.23366674e-02
-2.02264622e-01 -6.81406498e-01 6.57014027e-02 6.70919418e-01
-1.30179286e-01 -3.92664447e-02 -7.56729960e-01 -2.31765017e-01
7.70536482e-01 2.51639098e-01 -1.39688735e-03 -4.90983248e-01
-1.72297871e+00 5.65711677e-01 4.28632230e-01 -1.91606972e-02
-5.19026160e-01 6.11604512e-01 2.42944747e-01 1.08682752e-01
3.13243955e-01 8.04480463e-02 -3.91964763e-01 -2.70125657e-01
-7.15240359e-01 1.04519832e+00 8.59151900e-01 6.29046917e-01
2.93643534e-01 -2.42680773e-01 -3.15011859e-01 6.83247328e-01
3.50103289e-01 5.50991893e-01 5.91482162e-01 -1.07513273e+00
5.62389016e-01 3.14626068e-01 3.32677037e-01 -1.12987268e+00
-7.94060171e-01 -1.89542994e-01 -6.41582191e-01 -2.63207052e-02
7.48012722e-01 -2.22102419e-01 -9.60549235e-01 1.89183867e+00
5.87356746e-01 -1.90250808e-03 -3.64631116e-01 1.21102834e+00
5.43616593e-01 6.54639482e-01 2.36047152e-02 1.93855762e-01
1.44583952e+00 -1.27287233e+00 -4.22530472e-01 -4.93862569e-01
3.82991374e-01 -5.28722525e-01 5.08637428e-01 2.66690105e-01
-1.02889395e+00 -8.32797885e-01 -6.16584659e-01 -3.07870448e-01
-6.55452684e-02 1.22803329e-02 3.77238154e-01 2.25295022e-01
-8.38894188e-01 6.97552741e-01 -1.28687882e+00 -6.60289288e-01
-1.30610839e-01 4.21568871e-01 -4.07088578e-01 1.16956025e-01
-1.20635951e+00 1.15775061e+00 -1.70187265e-01 3.49836528e-01
-6.77946806e-01 -2.68059582e-01 -8.68643641e-01 -3.41481119e-01
2.34302044e-01 -1.38414061e+00 1.33827698e+00 -5.75635374e-01
-1.27174389e+00 7.43731618e-01 -2.95412719e-01 -3.23023945e-01
1.03999567e+00 -9.00464535e-01 -3.12210888e-01 7.33306631e-02
1.17668055e-01 9.56883371e-01 8.25969994e-01 -8.62392902e-01
-5.44100344e-01 -7.40018845e-01 -1.84353083e-01 7.00601161e-01
-1.26976028e-01 -7.51883984e-02 -9.07130718e-01 -8.29428017e-01
1.39512569e-01 -1.58283544e+00 -3.49345356e-01 2.54130781e-01
-4.83494461e-01 -4.40678149e-01 7.01663673e-01 -8.52780938e-01
1.03417814e+00 -1.83693683e+00 9.66306448e-01 4.91573364e-02
-7.39168469e-03 -1.01013266e-01 -1.53954864e-01 4.55956459e-01
1.31917924e-01 -3.69424462e-01 8.53657350e-02 -7.14985728e-01
1.41560540e-01 3.84296268e-01 1.75007731e-02 7.33939290e-01
-9.93084535e-02 1.10325682e+00 -7.93204546e-01 -3.95590156e-01
2.04739556e-01 6.98417783e-01 -5.20006955e-01 1.40527248e-01
-1.00649282e-01 6.73001170e-01 -3.32146883e-01 5.41771531e-01
-4.66261394e-02 -2.22074360e-01 4.08090167e-02 -2.32460827e-01
2.63805568e-01 -3.58022265e-02 -1.45874560e+00 1.80691552e+00
-7.09842518e-02 3.72681618e-01 -3.21949780e-01 -3.77722204e-01
5.54668903e-01 4.70004380e-01 9.50335860e-01 -2.02177241e-01
-4.01694067e-02 -1.91799060e-01 -7.52192065e-02 -6.50712132e-01
4.71716881e-01 -1.68413103e-01 -4.22530532e-01 4.22851890e-01
1.99607220e-02 3.11945140e-01 1.55245438e-01 -1.32056892e-01
8.75745773e-01 5.84147930e-01 1.51841789e-01 2.43274897e-01
2.14708582e-01 -2.81841215e-02 5.85235298e-01 5.22888422e-01
-3.95088792e-01 8.56025696e-01 1.78897560e-01 -8.05138171e-01
-1.17646396e+00 -1.24923408e+00 4.22302753e-01 1.32123256e+00
9.66272056e-02 -4.11699206e-01 -4.61453348e-01 -2.91089982e-01
4.16200519e-01 1.29369467e-01 -7.32843876e-01 -1.01112938e-02
-1.14520442e+00 -6.02301061e-01 4.57149982e-01 8.88085783e-01
2.42449880e-01 -1.04207671e+00 -8.44860554e-01 4.55831349e-01
-6.29001677e-01 -1.09541810e+00 -1.02953935e+00 -5.17836690e-01
-6.77555501e-01 -9.92697299e-01 -1.20345247e+00 -7.92260170e-01
2.44575337e-01 -1.06660500e-01 9.86248374e-01 -1.76967502e-01
-3.96272838e-01 7.22829640e-01 -9.64525715e-02 3.53054069e-02
-6.41633477e-03 -3.54118869e-02 5.00305593e-01 1.02935627e-01
2.27133676e-01 -4.80287403e-01 -8.64920378e-01 3.89131695e-01
-3.54336530e-01 1.82911918e-01 2.69662410e-01 6.44953966e-01
4.54801649e-01 -3.51040810e-01 -5.95195172e-03 -5.10015965e-01
3.26269567e-01 -4.38549876e-01 2.00059488e-01 1.98754162e-01
5.07863704e-03 -3.61723565e-02 2.05542997e-01 -7.68669307e-01
-7.59471774e-01 5.81922829e-01 -2.23017067e-01 -6.23618245e-01
-2.79203504e-01 1.62235737e-01 -3.80091257e-02 4.28759247e-01
4.94242162e-01 1.86888129e-01 2.05694571e-01 -7.09779859e-01
2.97747254e-01 7.60074332e-02 9.81503367e-01 -5.26383758e-01
7.98789024e-01 6.45635903e-01 1.78589091e-01 -7.98875630e-01
-5.90178668e-01 -6.09276354e-01 -1.12457025e+00 -6.10748470e-01
1.21047866e+00 -1.12141407e+00 -7.73239672e-01 8.43328595e-01
-1.32488692e+00 -2.23691955e-01 1.38982177e-01 5.03728509e-01
-6.94835961e-01 3.30493689e-01 -8.75228584e-01 -7.31136084e-01
-1.42324075e-01 -9.04330313e-01 1.44600821e+00 -1.86490208e-01
-9.48679745e-01 -9.82280254e-01 3.31541628e-01 5.86526096e-01
-2.17949897e-01 7.29632497e-01 4.93611336e-01 -3.49347234e-01
-3.76922697e-01 -4.94294554e-01 6.98947728e-01 -1.87340915e-01
2.47095451e-02 -2.52657622e-01 -3.99348497e-01 -3.82686257e-01
-1.89154014e-01 -1.64569169e-01 8.67757916e-01 4.36721057e-01
6.39213026e-01 -4.22316432e-01 -5.49615681e-01 4.74248528e-01
7.54018784e-01 -2.16578975e-01 4.71916348e-01 2.56658256e-01
1.17843032e+00 9.67649221e-01 4.09802228e-01 4.95534211e-01
8.42593491e-01 1.18814027e+00 2.20370054e-01 2.53144503e-01
-8.51082876e-02 -4.17721212e-01 7.06652641e-01 6.00972235e-01
-6.48573518e-01 -2.27361679e-01 -9.24144447e-01 4.94099259e-01
-2.06985569e+00 -1.28970206e+00 -9.49223265e-02 2.03614998e+00
4.09568101e-01 -5.16654886e-02 8.89684796e-01 -1.61378384e-01
6.98404551e-01 3.39074641e-01 -6.66784167e-01 1.24941245e-01
-4.04909775e-02 -3.40676844e-01 1.54326364e-01 4.71640944e-01
-1.28420269e+00 7.20507264e-01 5.91988325e+00 1.07246079e-01
-8.76144648e-01 -3.49647216e-02 2.47513890e-01 -7.15991795e-01
7.61076063e-02 -2.50336349e-01 -9.79825735e-01 4.54185694e-01
6.85770929e-01 1.76802322e-01 2.62582481e-01 5.11713028e-01
1.84898168e-01 1.17478453e-01 -1.36702895e+00 1.07000434e+00
2.09518000e-01 -9.00603414e-01 1.54813975e-01 1.43849090e-01
5.08535624e-01 -8.96776095e-02 4.35207097e-04 8.21126625e-02
1.82103902e-01 -9.20640409e-01 1.01415074e+00 9.33288813e-01
1.78618476e-01 -5.71588755e-01 3.89938504e-01 6.10430539e-01
-1.40322387e+00 -2.26311564e-01 9.39433649e-02 -2.67069995e-01
7.42163241e-01 9.03058052e-02 -5.09221673e-01 3.90908897e-01
8.28224838e-01 9.05335248e-01 -6.36464417e-01 8.94284129e-01
-1.39032438e-01 1.83271974e-01 -4.04922485e-01 6.23063371e-02
-3.23177315e-02 1.36311248e-01 8.42775762e-01 1.04751432e+00
3.55668843e-01 4.26610075e-02 5.93707144e-01 3.55955809e-01
4.74402994e-01 -2.96943903e-01 -3.75072271e-01 1.83337942e-01
8.07920173e-02 9.10808384e-01 -3.69527668e-01 -2.40950242e-01
-3.37042540e-01 1.42123020e+00 1.71786815e-01 4.36126977e-01
-7.08741546e-01 3.92931789e-01 1.06051683e+00 3.88557136e-01
3.07379514e-01 -7.28709698e-01 8.50894526e-02 -1.14787269e+00
2.02985525e-01 -6.45168662e-01 6.87264323e-01 -6.97241724e-01
-1.38593948e+00 3.46779257e-01 2.00029105e-01 -1.40115166e+00
-9.10892010e-01 -6.06513262e-01 -4.03564245e-01 6.67746365e-01
-5.53241909e-01 -1.34589219e+00 -8.13186690e-02 7.12480247e-01
5.67928016e-01 1.81769568e-03 7.21565247e-01 9.13528427e-02
-3.50894481e-01 2.79741198e-01 -2.70058483e-01 4.79116201e-01
7.42617071e-01 -1.11028564e+00 7.97907412e-01 5.22296965e-01
3.27723473e-01 5.64599991e-01 1.01361477e+00 -9.12680626e-01
-1.44740415e+00 -9.13138688e-01 1.27971148e+00 -1.03376317e+00
4.68148351e-01 -4.46351856e-01 -6.46153212e-01 9.84014392e-01
-1.86902702e-01 -3.39874402e-02 5.67455292e-01 1.27418414e-01
-1.13454953e-01 2.37707734e-01 -7.30464518e-01 8.64395916e-01
1.53028142e+00 -3.99270654e-01 -7.84054697e-01 3.36096764e-01
4.80164915e-01 -7.33977377e-01 -9.40350294e-01 3.34900588e-01
1.01324010e+00 -7.40521848e-01 1.38380754e+00 -8.20586860e-01
4.86922026e-01 -3.47522795e-01 -6.18049577e-02 -1.23176730e+00
-6.01173937e-01 -3.91680598e-01 -6.64454281e-01 6.36858106e-01
2.98296735e-02 -1.27459437e-01 1.16424072e+00 7.77242780e-01
1.39295340e-01 -8.20382893e-01 -1.09239471e+00 -6.31602645e-01
3.78048569e-02 -2.33030453e-01 4.73080218e-01 5.84142923e-01
-2.62268245e-01 3.47623259e-01 -1.12403321e+00 1.26065940e-01
6.51116610e-01 1.60622910e-01 1.10421193e+00 -1.20757270e+00
-7.32586265e-01 -2.86116749e-01 -8.57409418e-01 -1.43067491e+00
1.17189139e-01 -5.38622797e-01 1.16323307e-01 -1.54154658e+00
2.20492646e-01 4.17712778e-01 9.30534452e-02 4.45124179e-01
-3.10568839e-01 2.19010994e-01 6.36828780e-01 5.13040364e-01
-6.48547530e-01 6.72575414e-01 1.47453475e+00 -1.79910183e-01
-1.61368698e-01 4.14234579e-01 -1.38279330e-02 8.16872597e-01
3.07005644e-01 -2.35398039e-01 -2.56333262e-01 -5.51347017e-01
-2.46929172e-02 3.73315424e-01 9.92668390e-01 -1.43465006e+00
4.64225650e-01 -1.06766649e-01 1.03126240e+00 -8.28178942e-01
1.02695394e+00 -5.84998071e-01 6.87159836e-01 7.73281991e-01
-2.51128018e-01 4.62866783e-01 -2.87116438e-01 8.81562233e-01
4.70381938e-02 4.76512223e-01 3.06516945e-01 -4.06939924e-01
-9.37482715e-01 6.61440909e-01 -4.45596069e-01 -6.16949238e-02
8.30053329e-01 -4.66627896e-01 3.03014237e-02 -6.38561428e-01
-1.52932370e+00 5.19235194e-01 4.44191217e-01 9.71383989e-01
5.52741587e-01 -1.63235438e+00 -6.33028328e-01 -1.55293807e-01
7.72110969e-02 -5.35954013e-02 4.38601196e-01 7.03711092e-01
-3.24292958e-01 2.92743951e-01 -4.27564263e-01 -7.29919553e-01
-1.46720183e+00 2.01488465e-01 4.52129692e-01 -2.02287912e-01
-8.95297825e-01 8.52247655e-01 1.49055690e-01 -5.70661962e-01
2.78865129e-01 2.73991507e-02 -2.94580221e-01 1.69105813e-01
4.13181722e-01 5.28189182e-01 -4.42734510e-01 -1.35767996e+00
-5.21241426e-01 8.22886884e-01 1.42888337e-01 -4.05800581e-01
1.25841808e+00 -2.11118490e-01 1.00950666e-01 8.74170005e-01
1.22188580e+00 -4.17621136e-01 -1.67068589e+00 -2.89861143e-01
-8.22476596e-02 -4.70788538e-01 -7.15496004e-01 -5.68876803e-01
-8.92195582e-01 6.83957815e-01 5.45298517e-01 -3.47474307e-01
6.47561550e-01 3.05765450e-01 1.16453063e+00 3.81845832e-01
5.07782698e-01 -1.25859272e+00 5.62027633e-01 6.77824080e-01
1.11095536e+00 -9.56308722e-01 6.30831271e-02 -2.12965518e-01
-9.81836855e-01 1.08766127e+00 7.76706994e-01 -1.73666269e-01
6.49459660e-01 -1.45161480e-01 6.11569695e-02 -5.23073040e-02
-7.66774476e-01 -1.04022846e-01 6.22169614e-01 6.17584586e-01
5.22193789e-01 2.59069592e-01 1.66202467e-02 6.02074325e-01
-5.65160215e-01 -1.33231208e-01 5.77800497e-02 9.25852597e-01
-3.88662517e-01 -1.15940917e+00 -7.05430686e-01 3.30174088e-01
-1.83983520e-01 5.98070979e-01 -5.66737592e-01 7.25225747e-01
2.69738525e-01 5.74194133e-01 2.93762296e-01 -6.50331080e-01
6.22323990e-01 3.39546472e-01 7.29379654e-01 -6.18313193e-01
-5.14511883e-01 2.69259930e-01 1.07476600e-01 -7.51173913e-01
-4.99950528e-01 -1.32343829e+00 -1.32761991e+00 -4.85889971e-01
5.45197189e-01 -3.69554788e-01 1.12281181e-01 1.05847788e+00
1.37545407e-01 3.87811393e-01 7.67315179e-02 -1.42011845e+00
-4.98449951e-01 -8.97133231e-01 -5.24665833e-01 8.81102264e-01
5.09692013e-01 -9.51978564e-01 1.05550170e-01 2.75357515e-01]
|
[7.232895851135254, -0.361954927444458]
|
ab0b2bc9-5ce6-4cfd-a64d-4e2e47293933
|
openmixup-open-mixup-toolbox-and-benchmark
|
2209.04851
| null |
https://arxiv.org/abs/2209.04851v1
|
https://arxiv.org/pdf/2209.04851v1.pdf
|
OpenMixup: Open Mixup Toolbox and Benchmark for Visual Representation Learning
|
With the remarkable progress of deep neural networks in computer vision, data mixing augmentation techniques are widely studied to alleviate problems of degraded generalization when the amount of training data is limited. However, mixup strategies have not been well assembled in current vision toolboxes. In this paper, we propose \texttt{OpenMixup}, an open-source all-in-one toolbox for supervised, semi-, and self-supervised visual representation learning with mixup. It offers an integrated model design and training platform, comprising a rich set of prevailing network architectures and modules, a collection of data mixing augmentation methods as well as practical model analysis tools. In addition, we also provide standard mixup image classification benchmarks on various datasets, which expedites practitioners to make fair comparisons among state-of-the-art methods under the same settings. The source code and user documents are available at \url{https://github.com/Westlake-AI/openmixup}.
|
['Stan Z. Li', 'Di wu', 'Zicheng Liu', 'Zedong Wang', 'Siyuan Li']
|
2022-09-11
| null | null | null | null |
['semi-supervised-image-classification']
|
['computer-vision']
|
[-1.84824578e-02 -2.16527283e-01 -4.10626829e-01 -3.31393868e-01
-3.91510993e-01 -3.93605471e-01 6.85413122e-01 -3.10066372e-01
-2.70106196e-01 3.04297119e-01 -1.12747297e-01 -4.28105682e-01
3.74219745e-01 -4.68131602e-01 -4.67624128e-01 -7.66652107e-01
4.30717796e-01 3.75477731e-01 -1.53512478e-01 -1.76501900e-01
-1.69053078e-01 2.71582425e-01 -1.62347281e+00 2.99008161e-01
9.15458739e-01 1.09690094e+00 2.48024032e-01 6.01103961e-01
-1.60342202e-01 8.71654630e-01 -3.05427372e-01 -4.82968867e-01
3.05088788e-01 -2.55726635e-01 -6.98242426e-01 2.10707989e-02
7.91771173e-01 -3.40288371e-01 -6.98567867e-01 1.03800642e+00
5.43096483e-01 7.26602646e-03 4.26337630e-01 -1.72531939e+00
-9.55446780e-01 6.24269783e-01 -6.31302416e-01 4.43136096e-01
-4.35957760e-01 4.96034056e-01 7.17603147e-01 -1.09160531e+00
5.01639426e-01 8.60903442e-01 6.67998970e-01 8.66582215e-01
-1.23126602e+00 -1.03386748e+00 2.79188544e-01 3.19275290e-01
-1.12761712e+00 -7.39917874e-01 8.77330124e-01 -5.48531651e-01
6.33105278e-01 3.15895736e-01 7.38415003e-01 1.46635723e+00
-4.82761741e-01 1.12894905e+00 1.30834186e+00 -2.00731382e-01
-9.87801030e-02 1.15050755e-01 5.03059924e-01 8.17346394e-01
2.76903901e-02 -2.05709394e-02 -4.51074779e-01 9.10023153e-02
9.46354806e-01 2.40518063e-01 -3.06189626e-01 -5.46988845e-01
-1.21391308e+00 4.93625402e-01 8.18295777e-01 2.66333312e-01
-2.69800633e-01 3.20102364e-01 4.59895194e-01 1.25231072e-01
5.14012754e-01 2.42184460e-01 -5.57622552e-01 9.78912786e-02
-7.69677162e-01 1.91607460e-01 3.80833536e-01 1.06692719e+00
7.15830207e-01 5.58349609e-01 -1.68190718e-01 1.24059117e+00
2.71464556e-01 3.96904260e-01 5.85748374e-01 -1.02880812e+00
3.02280694e-01 8.29500616e-01 -2.26335973e-01 -6.13979578e-01
-4.09191072e-01 -8.06891382e-01 -1.25029600e+00 4.07553971e-01
4.97041553e-01 -4.43184450e-02 -1.04651761e+00 1.76369786e+00
2.77911901e-01 3.54418993e-01 -6.66477978e-02 7.14866042e-01
1.48483539e+00 4.44270730e-01 1.59390464e-01 2.63503776e-03
1.29817557e+00 -1.56897819e+00 -6.51756525e-01 -5.36267281e-01
5.46701729e-01 -6.60499394e-01 1.33956313e+00 4.23651099e-01
-9.73619163e-01 -8.47823918e-01 -1.05535793e+00 -1.80005223e-01
-5.75298011e-01 5.17066061e-01 8.62769425e-01 3.90172780e-01
-9.92603302e-01 4.75698739e-01 -9.50587928e-01 -3.04216057e-01
9.56011593e-01 1.03941478e-01 -5.02879441e-01 -3.38886738e-01
-6.21154606e-01 8.51919591e-01 3.93435121e-01 2.47497261e-01
-1.18690813e+00 -9.05045569e-01 -7.63182580e-01 -3.59352916e-01
4.22070056e-01 -7.08045363e-01 1.42723632e+00 -1.23675919e+00
-1.17707312e+00 1.14096510e+00 1.74249038e-01 -2.54045427e-01
5.32461762e-01 -4.04058725e-01 -3.41407090e-01 -9.54718590e-02
-1.41938865e-01 1.08040047e+00 9.08360362e-01 -1.45488715e+00
-2.43425846e-01 -4.34119672e-01 -8.52623805e-02 1.91328213e-01
-5.79008102e-01 5.26408777e-02 -8.19372535e-01 -7.14437246e-01
-2.30267063e-01 -7.88299382e-01 -3.56965899e-01 4.75116849e-01
-5.08805752e-01 -8.54657292e-02 7.39912033e-01 -5.61469674e-01
1.00763977e+00 -2.23859835e+00 3.70048056e-03 -3.60564202e-01
4.56947565e-01 8.15057099e-01 -3.89085710e-01 3.25418651e-01
-4.84011590e-01 3.01194079e-02 -4.46332783e-01 -7.85791397e-01
3.32500935e-02 1.48251668e-01 -2.72310436e-01 4.29293782e-01
1.43110052e-01 1.04389799e+00 -7.30954885e-01 -3.00712973e-01
6.16255224e-01 5.68816364e-01 -1.12868220e-01 3.98801953e-01
-2.86057115e-01 4.78063703e-01 -1.04800530e-01 8.11068833e-01
8.04542184e-01 -4.81290102e-01 -1.61977395e-01 -3.32486957e-01
-5.92785589e-02 -3.26436982e-02 -1.03993571e+00 1.83650780e+00
-2.23344639e-01 8.19137096e-01 2.07864806e-01 -9.53401029e-01
7.72083700e-01 9.35769379e-02 3.07093412e-01 -4.84829694e-01
4.82604891e-01 -2.79204771e-02 -2.67221555e-02 -4.38458115e-01
2.26395220e-01 4.04252082e-01 5.74328303e-01 2.79540092e-01
4.95689571e-01 5.14981933e-02 5.29696405e-01 1.16532587e-01
8.22031200e-01 3.82624030e-01 2.00382292e-01 -4.09398153e-02
9.32890847e-02 2.09000111e-01 4.42074865e-01 6.16716623e-01
-3.87124687e-01 7.75216401e-01 2.91281730e-01 -4.18895811e-01
-1.00752497e+00 -8.73797238e-01 -3.19302231e-01 1.30184591e+00
-1.24849319e-01 -6.47081792e-01 -8.32258284e-01 -4.81673926e-01
-1.30639464e-01 4.51570958e-01 -8.58181298e-01 -7.73475021e-02
-2.06099972e-01 -1.05773771e+00 7.00836539e-01 8.74699712e-01
9.48903203e-01 -1.18683004e+00 -8.59620795e-02 -1.54124916e-01
-8.67839456e-02 -1.16328263e+00 -2.36697719e-01 1.67489767e-01
-8.85186315e-01 -1.31208313e+00 -7.07013786e-01 -8.53320062e-01
8.01855147e-01 6.13811135e-01 1.18812573e+00 4.09750253e-01
-3.41185153e-01 3.41486275e-01 -3.09190422e-01 -5.90579689e-01
-2.41561517e-01 -6.29340559e-02 -1.59079749e-02 -3.41522098e-02
3.71351480e-01 -5.12162685e-01 -7.02700317e-01 4.74448144e-01
-9.87793565e-01 7.36167967e-01 5.03972352e-01 7.26236880e-01
5.80878973e-01 -4.39075559e-01 5.67083359e-01 -6.61245346e-01
4.33160365e-01 -4.73014951e-01 -5.91323078e-01 1.33345664e-01
-6.88335299e-01 -3.97464126e-01 5.28883755e-01 -5.56050122e-01
-7.10161746e-01 -7.56323859e-02 -3.01142633e-01 -7.37639487e-01
-3.89925420e-01 3.73007447e-01 -2.56382495e-01 -1.17632680e-01
8.07969451e-01 3.94562185e-01 1.98009863e-01 -6.95949137e-01
7.55988300e-01 6.45340741e-01 7.58938015e-01 -3.41458291e-01
7.12509930e-01 4.09939587e-01 -4.41096961e-01 -7.71585166e-01
-9.32745636e-01 -4.84783530e-01 -6.91012263e-01 -1.92230910e-01
6.98334336e-01 -1.11306942e+00 -2.77367920e-01 9.50486302e-01
-7.38659918e-01 -8.77490997e-01 -1.82626396e-01 2.04033956e-01
-4.76299077e-01 2.05594152e-01 -6.15371108e-01 -4.64337856e-01
-5.68321109e-01 -1.26895535e+00 4.51149702e-01 4.08878863e-01
1.08135179e-01 -8.51904333e-01 1.25661967e-02 6.67971015e-01
4.72875535e-01 1.57541007e-01 5.42693615e-01 -7.72040367e-01
-3.63703936e-01 -1.33332893e-01 -4.71401960e-01 9.03706789e-01
3.36786285e-02 3.56964260e-01 -1.46123266e+00 -3.50938410e-01
-4.73529428e-01 -6.09851897e-01 1.18320322e+00 4.17045325e-01
1.53894377e+00 -1.36230931e-01 -2.88232595e-01 1.11220026e+00
1.06681478e+00 -9.28021893e-02 5.77035844e-01 3.35483015e-01
1.00074017e+00 4.85409170e-01 2.76640087e-01 3.22195321e-01
3.87756675e-01 4.15966630e-01 8.17072868e-01 -5.05657315e-01
-3.42006385e-01 -6.46333173e-02 6.17961772e-02 7.12209344e-01
-1.56238005e-01 -4.39617820e-02 -1.06643128e+00 5.48998177e-01
-1.95877016e+00 -8.77657294e-01 -1.36207104e-01 2.00958347e+00
8.55434716e-01 -1.56722382e-01 2.76434034e-01 8.08977336e-02
6.01588011e-01 2.87662625e-01 -7.41663456e-01 2.37912685e-01
-2.08330095e-01 -5.02807274e-02 1.85027316e-01 5.64043373e-02
-1.38473725e+00 1.05754626e+00 5.99391603e+00 8.24266255e-01
-1.18996632e+00 2.43758276e-01 8.37543905e-01 -1.28701061e-01
8.33919272e-02 -2.39191622e-01 -6.61670148e-01 3.50947618e-01
4.98729646e-01 1.58349380e-01 4.77858961e-01 1.21024013e+00
-9.12508219e-02 2.00524062e-01 -9.92059529e-01 1.25046492e+00
1.46515384e-01 -1.67329288e+00 -7.94733539e-02 -1.38235942e-01
6.48350954e-01 6.92095578e-01 2.41497621e-01 3.94091487e-01
3.55043858e-01 -1.04252660e+00 5.37147284e-01 1.58003002e-01
8.12359571e-01 -3.64978135e-01 4.75313216e-01 1.75691918e-01
-1.06172168e+00 -3.95918414e-02 -4.05389160e-01 9.25661922e-02
-2.48420224e-01 3.80549848e-01 -3.03275466e-01 3.77020627e-01
9.49473798e-01 1.05585933e+00 -9.79600906e-01 1.20582533e+00
-3.54327679e-01 8.17423344e-01 -1.40734464e-01 4.41781849e-01
9.11420882e-02 -2.69157827e-01 2.11042717e-01 1.17168379e+00
-2.02461094e-01 -1.81176677e-01 2.10536197e-01 8.59193742e-01
-3.10161680e-01 7.26888925e-02 -6.09881818e-01 -1.74161553e-01
4.99585927e-01 1.75384188e+00 -5.97650886e-01 -4.76402432e-01
-5.16748846e-01 6.69618189e-01 5.22050500e-01 4.95068133e-01
-6.71298981e-01 1.21619347e-02 8.50464940e-01 -2.47718636e-02
-1.15187041e-01 -3.59524846e-01 -4.02444303e-01 -1.29472888e+00
-1.33220829e-05 -1.17885900e+00 3.38021874e-01 -1.16272628e+00
-1.33506298e+00 8.06252599e-01 1.43650293e-01 -1.25082779e+00
2.22412944e-01 -9.66460407e-01 -8.32299054e-01 5.98966300e-01
-1.47660792e+00 -1.61601138e+00 -9.58429277e-01 7.14110136e-01
7.09255338e-01 -4.66423333e-01 8.55695903e-01 4.06710654e-01
-1.31637561e+00 5.33937752e-01 1.11701190e-01 4.58187491e-01
7.24936724e-01 -1.18354428e+00 5.44574797e-01 9.41735387e-01
2.65252262e-01 6.88498855e-01 4.43234921e-01 -2.57616043e-01
-1.23651230e+00 -1.23258424e+00 -3.69243473e-02 -4.67425227e-01
8.17436278e-01 -4.84524816e-01 -1.12256837e+00 9.56473410e-01
6.39715374e-01 5.33347487e-01 8.84306431e-01 9.03002992e-02
-6.85726643e-01 -2.68204063e-01 -7.15759814e-01 7.79564977e-01
1.00900674e+00 -2.93427378e-01 -2.89154917e-01 4.02865440e-01
5.39008021e-01 -3.87373298e-01 -7.57351875e-01 4.23601419e-01
4.81186181e-01 -9.52214956e-01 9.91611898e-01 -5.74455857e-01
4.44299698e-01 -3.69579017e-01 -7.82797486e-02 -1.40162706e+00
-3.71523172e-01 -5.32717109e-01 -3.89888734e-01 1.36367238e+00
4.09495652e-01 -6.73729360e-01 6.10521257e-01 3.21313590e-01
-4.16989952e-01 -1.02751708e+00 -5.29657185e-01 -5.27400672e-01
1.05952762e-01 -6.41027808e-01 3.89290303e-01 1.15030026e+00
-3.57156783e-01 3.28501642e-01 -4.61652160e-01 -3.94825861e-02
7.48767555e-01 -1.29419759e-01 1.06872320e+00 -1.05718815e+00
-1.45854786e-01 -7.70066917e-01 -1.27744928e-01 -8.99867594e-01
9.82481893e-03 -9.92931306e-01 -3.44089121e-01 -1.74704480e+00
3.26629728e-01 -5.77153385e-01 -4.37804043e-01 1.05173194e+00
-2.91171402e-01 6.85016870e-01 3.67439181e-01 4.96473879e-01
-3.80806476e-01 7.49128938e-01 1.36669600e+00 -3.27654451e-01
-2.20199581e-02 -4.64292541e-02 -1.00211692e+00 9.18391287e-01
1.13471937e+00 -1.16717532e-01 -5.95077455e-01 -7.15880156e-01
-1.36017382e-01 -5.44455290e-01 6.47175372e-01 -1.22452927e+00
-9.33817774e-02 -1.39240414e-01 5.33826530e-01 -6.31994545e-01
5.24178624e-01 -6.30262494e-01 8.63595530e-02 1.87181011e-01
-3.06243747e-01 1.80094957e-01 4.89441782e-01 1.54690221e-01
-7.98783973e-02 -5.37962988e-02 1.14163888e+00 -1.04370579e-01
-9.23572838e-01 7.85540402e-01 -1.53080386e-03 2.83101965e-02
9.45746303e-01 -8.79873931e-02 -1.03523827e+00 -8.14350918e-02
-7.00895071e-01 3.67886811e-01 4.96411026e-01 7.33501971e-01
5.64632058e-01 -1.33639908e+00 -7.26972461e-01 2.59969443e-01
5.18759429e-01 2.63895661e-01 4.35847074e-01 1.01209569e+00
-4.83236253e-01 -7.23961294e-02 -5.57574749e-01 -5.56858718e-01
-1.36306322e+00 6.72697484e-01 4.86229241e-01 1.26582477e-03
-6.69930220e-01 9.55710590e-01 4.15596396e-01 -5.92575669e-01
4.09518421e-01 -3.19286287e-01 -3.11887741e-01 8.82964656e-02
9.20920312e-01 2.73692250e-01 -1.14285074e-01 -5.92715144e-01
-1.88585386e-01 3.74428302e-01 -2.58193344e-01 3.08810234e-01
1.26779914e+00 1.17953554e-01 -1.54522425e-02 5.63172936e-01
9.00446892e-01 -6.10943854e-01 -1.56635892e+00 -4.18681979e-01
-5.09992421e-01 -3.10950309e-01 1.58626080e-01 -8.73773217e-01
-1.45464599e+00 1.12585783e+00 8.13447416e-01 2.65613142e-02
1.20383775e+00 2.90447120e-02 3.45755696e-01 3.95307362e-01
-5.00856712e-02 -8.82656276e-01 2.55928993e-01 3.77065927e-01
1.23257351e+00 -1.54278934e+00 1.70123354e-02 -3.12354743e-01
-9.69997287e-01 7.40287125e-01 1.19603312e+00 -7.75383562e-02
6.37573063e-01 1.67712241e-01 6.53388858e-01 -5.17536402e-02
-7.73775518e-01 -4.14424390e-01 2.69498497e-01 9.70487833e-01
3.94878298e-01 -1.16990551e-01 2.70139843e-01 4.94427443e-01
-1.20107226e-01 5.01017198e-02 3.63591373e-01 6.66449845e-01
-1.85800657e-01 -1.04336321e+00 -2.21597761e-01 5.65454304e-01
-2.20122397e-01 -2.91164368e-01 -2.37606898e-01 8.88223588e-01
1.47016868e-01 8.22167218e-01 5.42974286e-03 -4.42506433e-01
1.20542556e-01 -4.65879776e-02 4.36614066e-01 -4.97270703e-01
-5.57499766e-01 -1.08936667e-01 4.16422375e-02 -4.80932206e-01
-4.97214347e-01 -3.53692949e-01 -9.09640729e-01 -2.34590277e-01
-1.96586236e-01 -3.13780218e-01 7.39477992e-01 6.29215002e-01
4.69792306e-01 6.83927298e-01 2.50989228e-01 -1.13261771e+00
-2.14989245e-01 -1.47092080e+00 -3.92491519e-01 3.33407044e-01
2.42927611e-01 -8.17021489e-01 -1.47097304e-01 1.92107603e-01]
|
[9.73228931427002, 1.80404531955719]
|
33ed29ee-71c4-4eba-8caa-2af93f558c9f
|
a-generalized-latent-factor-model-approach-to
|
2211.09272
| null |
https://arxiv.org/abs/2211.09272v1
|
https://arxiv.org/pdf/2211.09272v1.pdf
|
A Generalized Latent Factor Model Approach to Mixed-data Matrix Completion with Entrywise Consistency
|
Matrix completion is a class of machine learning methods that concerns the prediction of missing entries in a partially observed matrix. This paper studies matrix completion for mixed data, i.e., data involving mixed types of variables (e.g., continuous, binary, ordinal). We formulate it as a low-rank matrix estimation problem under a general family of non-linear factor models and then propose entrywise consistent estimators for estimating the low-rank matrix. Tight probabilistic error bounds are derived for the proposed estimators. The proposed methods are evaluated by simulation studies and real-data applications for collaborative filtering and large-scale educational assessment.
|
['Xiaoou Li', 'Yunxiao Chen']
|
2022-11-17
| null | null | null | null |
['matrix-completion']
|
['methodology']
|
[ 2.03254700e-01 8.96143094e-02 -3.43040675e-01 -4.15686697e-01
-9.86064017e-01 -4.79374737e-01 1.82090759e-01 3.79563570e-01
-2.95200795e-01 8.53414774e-01 4.27038610e-01 -4.11022991e-01
-9.35318112e-01 -4.26208019e-01 -9.71184552e-01 -5.96984982e-01
-3.13931435e-01 4.65925932e-01 -4.87205714e-01 1.11818343e-01
1.35116160e-01 7.02533573e-02 -1.32499230e+00 2.29197487e-01
1.19783318e+00 5.10164976e-01 5.65954077e-04 7.42471874e-01
2.98603863e-01 7.95991063e-01 -2.26862773e-01 -4.61080611e-01
3.31078172e-01 2.10499182e-01 -3.31583530e-01 3.37340742e-01
5.73237956e-01 -2.24101797e-01 -2.40751445e-01 1.33720863e+00
2.96062320e-01 3.64380658e-01 8.59588385e-01 -1.41959870e+00
-3.38523805e-01 8.49065423e-01 -9.53951597e-01 -6.41795387e-03
8.04732442e-01 -7.18647063e-01 1.20927489e+00 -1.47959960e+00
2.81988531e-01 1.49356663e+00 7.29550719e-01 -1.22675747e-01
-1.58863342e+00 -8.47766399e-01 1.92307457e-01 1.26171425e-01
-1.45383310e+00 -2.63070643e-01 4.58404541e-01 -1.00458407e+00
-5.89914322e-02 4.05006289e-01 1.51575714e-01 8.87131572e-01
3.11355382e-01 8.74189973e-01 1.65091491e+00 -2.56481498e-01
3.29449922e-01 1.00674532e-01 5.57070434e-01 5.58458149e-01
9.30680215e-01 -1.12093978e-01 -7.21779287e-01 -7.51317561e-01
5.81810355e-01 2.03749344e-01 -3.75282317e-02 -3.98406684e-01
-1.16661131e+00 8.68729234e-01 -4.26591277e-01 -3.56976837e-01
-6.67771816e-01 -2.00849041e-01 -3.47529724e-02 4.82172698e-01
2.69685686e-01 -2.09609225e-01 -5.07688642e-01 -4.28368710e-02
-7.86605835e-01 2.57510662e-01 9.29334581e-01 1.19847608e+00
7.22542226e-01 -9.95232612e-02 -1.75642908e-01 9.18006837e-01
6.34734273e-01 7.54883885e-01 -4.39553447e-02 -8.39107215e-01
8.19205523e-01 2.91192502e-01 5.09782851e-01 -1.08009708e+00
-4.64996338e-01 -5.53202391e-01 -1.20830166e+00 -1.80846274e-01
7.14532197e-01 -4.46845233e-01 -1.57244489e-01 1.85281062e+00
5.36843896e-01 7.21623838e-01 4.31480864e-03 7.58391976e-01
5.14485776e-01 4.42303628e-01 -3.07231247e-01 -5.62451959e-01
1.17196250e+00 -4.34562862e-01 -1.20874429e+00 -1.32638216e-01
1.93602279e-01 -9.02522743e-01 7.14096010e-01 9.91789818e-01
-1.10154295e+00 -3.87761265e-01 -7.82006204e-01 2.12631330e-01
2.08865777e-01 3.75248462e-01 8.25009465e-01 8.41784954e-01
-6.60357118e-01 4.96664584e-01 -7.56666839e-01 2.19111562e-01
-7.56188482e-02 5.97692311e-01 -5.92793167e-01 -3.88314664e-01
-1.01795006e+00 -2.11487408e-03 -2.92613387e-01 2.85499722e-01
-7.38503516e-01 -8.95773292e-01 -6.88557327e-01 8.29308704e-02
3.44088644e-01 -5.40455759e-01 1.01525104e+00 -3.33915710e-01
-1.09707999e+00 1.82381138e-01 -2.24483490e-01 -9.05243978e-02
4.54182923e-01 -5.18034697e-01 -3.58147919e-01 -2.41309747e-01
2.48097658e-01 -4.67376053e-01 1.04815686e+00 -8.95782232e-01
-6.68994129e-01 -5.00015140e-01 -1.49281323e-01 7.61617944e-02
-4.08063829e-01 -2.52359152e-01 -8.21388438e-02 -7.61360705e-01
6.16047561e-01 -1.07417619e+00 -6.98051631e-01 -3.60206664e-01
-2.21544102e-01 -1.08148016e-01 7.75760487e-02 -8.75252068e-01
1.41828036e+00 -2.15955710e+00 2.89613008e-01 7.14393437e-01
3.01351160e-01 -5.37718654e-01 -1.43734336e-01 7.17734694e-01
-2.13018760e-01 -1.29678726e-01 -3.84278335e-02 -4.33140606e-01
1.92024887e-01 -5.10647055e-03 -1.62425548e-01 9.57754910e-01
-4.80034530e-01 4.57161739e-02 -8.58585060e-01 -3.58930916e-01
1.50233367e-03 1.87707841e-01 -6.94260836e-01 1.19222589e-01
3.88217688e-01 3.49946022e-01 -4.62216794e-01 4.99134451e-01
1.06092775e+00 -3.40909243e-01 5.80835581e-01 -2.00916097e-01
-2.29430590e-02 -1.97445780e-01 -2.44268394e+00 1.43171155e+00
-4.87653911e-01 3.48328799e-01 6.86901689e-01 -8.52584839e-01
5.40695071e-01 4.55721110e-01 7.35386193e-01 1.82967514e-01
1.44147828e-01 9.87819582e-02 -2.01206319e-02 -3.14243287e-01
5.01386106e-01 9.55560058e-02 -2.33083010e-01 4.01900411e-01
-1.12674296e-01 5.17574131e-01 4.21550691e-01 4.65566516e-01
1.27112198e+00 -4.25243169e-01 5.60448349e-01 -5.63135922e-01
6.78402901e-01 -5.80310225e-01 1.05891001e+00 1.13038528e+00
1.43052951e-01 2.80316442e-01 5.35524726e-01 1.63699806e-01
-7.30608404e-01 -1.23851120e+00 -3.98278743e-01 1.27264929e+00
-1.91850409e-01 -6.01045370e-01 -2.71337867e-01 -1.68970957e-01
4.70212668e-01 4.43811864e-02 -5.92087150e-01 2.06097707e-01
5.02929240e-02 -7.85322666e-01 -9.36454535e-02 3.54819238e-01
-4.73427586e-02 1.03672609e-01 4.91656214e-01 3.21055263e-01
-1.29594788e-01 -1.19714546e+00 -5.45301318e-01 -1.05053447e-01
-1.04755509e+00 -1.37798238e+00 -3.88853371e-01 -8.66290748e-01
8.38795662e-01 4.77226228e-01 9.18731570e-01 -3.33677202e-01
1.23525456e-01 8.10056806e-01 -2.18750596e-01 -8.74603242e-02
-7.39117479e-03 -4.76276517e-01 7.00869322e-01 6.11365676e-01
1.54013947e-01 -6.73263609e-01 -5.08175015e-01 4.58853334e-01
-1.01573431e+00 -9.42765325e-02 7.59504914e-01 1.22171378e+00
8.18721712e-01 2.18954340e-01 5.70634544e-01 -1.27126765e+00
1.09266806e+00 -8.09847236e-01 -8.10281515e-01 3.76580626e-01
-7.94665158e-01 1.27285302e-01 3.54347497e-01 -9.80611920e-01
-1.02605987e+00 2.67093748e-01 4.86298889e-01 -2.91404128e-01
9.96419266e-02 1.11672056e+00 -4.30184990e-01 5.07812388e-02
3.18295211e-01 -4.21982892e-02 -9.10240784e-02 -7.74340630e-01
4.09446597e-01 8.52499664e-01 3.59554738e-01 -9.56602275e-01
9.08178508e-01 3.84140253e-01 4.26244587e-01 -9.09459949e-01
-8.42263758e-01 -7.99544156e-01 -6.41365409e-01 -8.88432562e-02
3.66235316e-01 -1.49309385e+00 -1.01641309e+00 1.46119282e-01
-5.61159253e-01 1.07893869e-01 7.29871541e-02 1.35302448e+00
-3.52791190e-01 4.69883323e-01 -9.12368834e-01 -1.10788560e+00
2.75299288e-02 -8.54285419e-01 5.72025180e-01 -2.46809255e-02
-4.16590758e-02 -1.06580436e+00 4.37232316e-01 6.10764325e-01
-2.95277983e-01 -8.44847877e-03 6.36943519e-01 -7.10554481e-01
-5.27595222e-01 -4.03088421e-01 4.60834168e-02 2.44174138e-01
-1.16037011e-01 -1.86764181e-01 -4.30208951e-01 -6.43360317e-01
-6.90992922e-02 -3.39799561e-02 2.21033946e-01 4.65458840e-01
1.00918508e+00 -7.72249341e-01 -1.13829315e-01 2.56883711e-01
1.14872587e+00 -2.75267631e-01 1.88513935e-01 -2.52822101e-01
5.01101136e-01 5.90161264e-01 1.05089259e+00 1.31718481e+00
5.03028512e-01 2.39630058e-01 -5.47050051e-02 3.44080359e-01
5.80760181e-01 -4.00462270e-01 3.52022797e-01 1.49807680e+00
5.03795087e-01 8.20176974e-02 -6.45155072e-01 5.12040198e-01
-2.22053099e+00 -8.25360537e-01 -8.38986158e-01 2.53353000e+00
6.98207259e-01 -3.01498264e-01 2.25387681e-02 2.28693143e-01
9.64681208e-01 -3.26118290e-01 -3.51518273e-01 -5.84366508e-02
-1.92202047e-01 2.59619772e-01 6.81208074e-01 6.93475187e-01
-1.07353497e+00 3.49390686e-01 6.59140205e+00 8.06075692e-01
-3.30122709e-02 2.41643563e-01 1.11015134e-01 1.20721497e-01
-3.20305407e-01 1.91480651e-01 -7.30178297e-01 3.26237202e-01
7.22197771e-01 -3.00762713e-01 5.18583894e-01 6.74950898e-01
4.46334362e-01 -1.96274385e-01 -1.11367893e+00 1.29574895e+00
-1.00111596e-01 -8.09316695e-01 -4.45085943e-01 3.94877046e-01
1.38537192e+00 -3.68483275e-01 2.95500994e-01 3.48575503e-01
7.47218549e-01 -6.69367433e-01 2.82811671e-01 8.05311620e-01
5.66167295e-01 -7.45627463e-01 4.75467354e-01 5.19674599e-01
-1.13357067e+00 -4.31903958e-01 -5.49852610e-01 -4.35872674e-01
-7.17491210e-02 1.09973431e+00 -3.30956995e-01 3.22637737e-01
3.60061944e-01 8.49972427e-01 -4.82253462e-01 1.47795367e+00
-1.22841120e-01 1.07464623e+00 -3.55864406e-01 1.71482429e-01
-4.05780554e-01 -7.28366256e-01 6.13688171e-01 7.95620322e-01
3.84586424e-01 3.95277351e-01 4.10156637e-01 2.18309715e-01
-1.34184048e-01 5.92796981e-01 -3.46357405e-01 3.23095471e-01
6.51077628e-01 1.44345474e+00 -2.76107907e-01 -1.12207435e-01
-8.25729728e-01 4.37433839e-01 2.29342118e-01 5.02588034e-01
-4.39374059e-01 -4.94765453e-02 7.85204887e-01 1.19224004e-01
1.44230410e-01 -3.46067727e-01 -4.25496012e-01 -1.50326681e+00
-1.36883169e-01 -1.13164115e+00 7.27164209e-01 -3.53046894e-01
-1.69378018e+00 -2.52937526e-01 5.98434024e-02 -1.35840440e+00
-1.50316685e-01 -4.10614252e-01 -3.58694494e-01 6.41939342e-01
-6.42109454e-01 -5.18879235e-01 -4.29324172e-02 8.51832449e-01
-8.79050698e-03 -3.31780314e-01 2.99716920e-01 7.12956369e-01
-8.00664485e-01 6.64779723e-01 1.00520790e+00 7.06483889e-03
7.48942912e-01 -1.43783760e+00 -3.71579915e-01 9.85028148e-01
2.04875320e-01 1.09893727e+00 1.01516557e+00 -9.23989177e-01
-1.97335339e+00 -8.80420148e-01 7.55493939e-01 -2.21237689e-01
1.00692332e+00 -6.22940123e-01 -6.82628453e-01 7.48125076e-01
-1.44169033e-01 -6.64326921e-03 1.27359700e+00 9.41801608e-01
-3.13786179e-01 -1.56592235e-01 -1.14861226e+00 4.33404803e-01
7.40491569e-01 -5.72668254e-01 -3.01506460e-01 6.64111555e-01
3.40972960e-01 -3.38915259e-01 -1.59078300e+00 3.76875192e-01
6.94143355e-01 -3.13038945e-01 1.01066887e+00 -1.00999343e+00
1.70585394e-01 -3.99125695e-01 -5.17184496e-01 -1.00584495e+00
-6.56835198e-01 -7.55500317e-01 -4.19368565e-01 1.19147062e+00
3.31937313e-01 -2.55939573e-01 1.12218094e+00 8.23869467e-01
3.02118480e-01 -4.17105138e-01 -9.69843209e-01 -7.46208191e-01
-1.72607094e-01 -5.65387011e-01 2.38921151e-01 1.05888653e+00
2.11568996e-01 5.41645229e-01 -1.02113056e+00 6.05777919e-01
1.25026047e+00 1.97148800e-01 1.06970060e+00 -1.47839153e+00
-6.65472209e-01 2.61167318e-01 -3.25657874e-01 -1.13303399e+00
3.56268227e-01 -8.00637424e-01 -2.64749736e-01 -1.28050172e+00
4.61466521e-01 -3.97359222e-01 -2.31811926e-01 -1.53575554e-01
-3.38803977e-01 -2.48474211e-01 -1.07518092e-01 2.80903764e-02
-8.95232677e-01 6.37326539e-01 8.26912940e-01 -5.71866594e-02
-1.18728429e-01 4.87996548e-01 -6.49803042e-01 6.55035079e-01
4.31942016e-01 -6.95772588e-01 -6.12556577e-01 3.22582982e-02
6.26325965e-01 8.72140646e-01 1.33528479e-03 -8.74146581e-01
3.96987855e-01 -4.00245070e-01 3.17160636e-01 -9.05563056e-01
1.79676622e-01 -9.99756813e-01 4.23008710e-01 2.41950527e-01
-6.56121373e-01 1.90470979e-01 -3.43604922e-01 1.20024157e+00
-2.12976709e-01 -2.10367545e-01 2.96290308e-01 5.78745365e-01
-1.07844599e-01 3.59957308e-01 -6.18999958e-01 2.43976340e-01
7.21565306e-01 2.29006529e-01 -1.45138754e-02 -7.82868445e-01
-1.34433007e+00 3.46773952e-01 -2.76485175e-01 2.31070638e-01
7.71708012e-01 -1.60904014e+00 -9.06885326e-01 1.72821820e-01
3.20082121e-02 -4.53666508e-01 6.09052896e-01 1.03650451e+00
2.36007482e-01 3.37492257e-01 1.93634808e-01 -4.66495305e-01
-1.51190615e+00 4.86111671e-01 -3.03838909e-01 -4.11251873e-01
6.18407391e-02 5.50644577e-01 4.23888899e-02 -6.75103128e-01
3.42454314e-01 -1.74629688e-01 -3.02400172e-01 1.13476031e-01
5.84306121e-01 9.73864675e-01 -1.21516019e-01 -4.68197256e-01
-1.02505162e-01 1.86841823e-02 6.25950797e-03 -3.46719384e-01
1.21769607e+00 -5.32444417e-01 -4.06808555e-01 7.13064611e-01
1.05741441e+00 6.50358021e-01 -9.87833560e-01 -8.56906414e-01
1.39726460e-01 -8.13377202e-01 -1.49782449e-01 -2.49044761e-01
-7.90867031e-01 4.59732115e-01 6.28165603e-01 -2.33755261e-01
8.84685457e-01 -4.30613607e-01 9.27478373e-02 5.68926692e-01
4.84525353e-01 -1.16072190e+00 -1.07722796e-01 4.56472874e-01
6.32886052e-01 -1.18000758e+00 4.26459938e-01 -5.94217718e-01
-3.41807425e-01 8.01981747e-01 4.94297355e-01 -1.84334025e-01
1.30537212e+00 9.47132036e-02 -5.16048133e-01 2.28390664e-01
-1.10654724e+00 4.85893153e-02 6.74178243e-01 4.07281786e-01
6.02803469e-01 4.82439637e-01 -8.65749300e-01 1.21412492e+00
-2.20990703e-01 -1.72914669e-01 1.07807839e+00 8.76763880e-01
-1.10928535e-01 -1.20729983e+00 -1.00999331e+00 1.00973701e+00
-7.69377351e-01 -1.08738877e-01 -3.97286676e-02 2.91772813e-01
-2.56574929e-01 1.56457090e+00 -3.47179860e-01 -4.98901486e-01
3.31958741e-01 -3.12671632e-01 4.19707686e-01 -5.72734118e-01
-3.54903042e-01 4.05553639e-01 -7.02047944e-02 -2.74267852e-01
-3.06488089e-02 -1.23909056e+00 -6.01860344e-01 -4.57690954e-01
-3.97771239e-01 6.23128593e-01 5.68731308e-01 6.54548585e-01
1.31975397e-01 1.87876791e-01 8.58231664e-01 -3.96194756e-02
-1.06664169e+00 -1.02440548e+00 -1.14021194e+00 4.96852279e-01
1.65639177e-01 -7.66097665e-01 -4.66270387e-01 -1.74353510e-01]
|
[7.064157962799072, 4.608391761779785]
|
025b767c-024f-4095-bdeb-4ec41bc03542
|
genres-parsers-and-bert-the-interaction
| null | null |
https://aclanthology.org/2021.adaptnlp-1.7
|
https://aclanthology.org/2021.adaptnlp-1.7.pdf
|
Genres, Parsers, and BERT: The Interaction Between Parsers and BERT Models in Cross-Genre Constituency Parsing in English and Swedish
|
Genre and domain are often used interchangeably, but are two different properties of a text. Successful parser adaptation requires both cross-domain and cross-genre sensitivity (Rehbein and Bildhauer, 2017). While the impact domain differences have on parser performance degradation is more easily measurable in respect to lexical differences, impact of genre differences can be more nuanced. With the predominance of pre-trained language models (LMs; e.g. BERT (Devlin et al., 2019)), there are now additional complexities in developing cross-genre sensitive models due to the infusion of linguistic characteristics derived from, usually, a third genre. We perform a systematic set of experiments using two neural constituency parsers to examine how different parsers behave in combination with different BERT models with varying source and target genres in English and Swedish. We find that there is extensive difficulty in predicting the best source due to the complex interactions between genres, parsers, and LMs. Additionally, the influence of the data used to derive the underlying BERT model heavily influences how best to create more robust and effective cross-genre parsing models.
|
['Daniel Dakota']
| null | null | null | null |
eacl-adaptnlp-2021-4
|
['constituency-parsing']
|
['natural-language-processing']
|
[ 1.46286711e-01 -2.66370386e-01 2.35009082e-02 -5.58541059e-01
-9.65590417e-01 -1.05303943e+00 5.24068952e-01 3.47718745e-01
-7.02764273e-01 5.63581228e-01 7.03826129e-01 -3.28506470e-01
-1.73238218e-01 -3.89840335e-01 -5.31956196e-01 -2.77905434e-01
3.27733427e-01 3.49856585e-01 2.90937245e-01 -3.73645693e-01
1.95575312e-01 1.57992393e-01 -1.25767934e+00 2.38766417e-01
6.34586275e-01 3.50810707e-01 4.84690011e-01 7.74484515e-01
-2.73726970e-01 6.00128710e-01 -9.71896648e-01 -6.21573865e-01
8.08118880e-02 -5.95331252e-01 -8.56100738e-01 -1.93374947e-01
1.67385504e-01 1.56247228e-01 1.05727382e-01 9.05987859e-01
6.42494857e-01 -3.80358147e-03 6.27564430e-01 -5.31110585e-01
-5.85575581e-01 1.23155224e+00 -3.67607027e-01 7.13342905e-01
3.71555537e-01 2.35267207e-01 1.15483785e+00 -4.04034197e-01
7.96909153e-01 1.53957796e+00 7.03141391e-01 5.87460876e-01
-1.54286885e+00 -7.49548197e-01 3.25857610e-01 -1.92554355e-01
-7.45685101e-01 -7.34550714e-01 7.45624483e-01 -4.75280553e-01
7.61640668e-01 6.88992515e-02 3.79815966e-01 1.61127543e+00
2.09684402e-01 3.12214196e-01 1.22344339e+00 -7.78690815e-01
1.96048245e-02 3.34795356e-01 2.36390650e-01 1.37500660e-02
2.56538898e-01 5.40248118e-02 -5.96085250e-01 -1.19885780e-01
4.07554507e-01 -7.59407461e-01 -2.36370921e-01 6.78092539e-02
-9.17335093e-01 8.68164837e-01 -1.26736104e-01 7.69384980e-01
3.62038948e-02 -2.62240440e-01 9.20994401e-01 5.96894085e-01
2.78857023e-01 8.87059450e-01 -8.07080626e-01 -7.27862775e-01
-6.60437047e-01 1.67481363e-01 8.92178833e-01 7.23176777e-01
3.00802514e-02 -8.82824790e-03 1.68628410e-01 1.33204031e+00
2.74521768e-01 2.08512634e-01 8.04192364e-01 -7.80133843e-01
6.99747741e-01 6.95475414e-02 -7.85806999e-02 -6.12355828e-01
-5.26858211e-01 -3.19192797e-01 1.62846819e-02 -1.20839871e-01
1.01288867e+00 -3.27255309e-01 -6.43409729e-01 2.30603814e+00
5.31121977e-02 -5.88560522e-01 2.32779250e-01 5.29985845e-01
6.18025661e-01 4.94669259e-01 7.68411875e-01 -3.45891923e-01
1.55850267e+00 -2.85688072e-01 -3.16151053e-01 -8.41845036e-01
8.95981908e-01 -1.02556396e+00 1.52814662e+00 1.98068514e-01
-1.20156610e+00 -5.16496420e-01 -1.05286837e+00 -1.79233268e-01
-2.35504940e-01 -2.81743556e-01 4.01833713e-01 8.22265506e-01
-6.85731947e-01 6.05861783e-01 -7.89289653e-01 -6.95208490e-01
-8.47564340e-02 1.73833057e-01 -2.91089833e-01 7.30382651e-02
-1.29487479e+00 1.17769420e+00 4.20433909e-01 -3.42939168e-01
-7.79145882e-02 -6.07761860e-01 -6.24088228e-01 -7.33900592e-02
7.71584660e-02 -3.49360377e-01 1.53841317e+00 -1.32609701e+00
-1.38364232e+00 9.25320864e-01 1.94344640e-01 -5.38058057e-02
3.64251107e-01 1.43170476e-01 -4.46051508e-01 -2.77629346e-01
3.25278997e-01 3.66373628e-01 5.01536846e-01 -9.73229468e-01
-9.46681559e-01 -5.62715709e-01 8.50391239e-02 2.95203060e-01
-4.06786531e-01 7.69989431e-01 -4.45314683e-03 -6.32966518e-01
-3.22117191e-03 -8.58798563e-01 2.55909532e-01 -7.07925320e-01
1.82852566e-01 -2.67503262e-01 3.23774129e-01 -8.04498196e-01
1.34777832e+00 -2.34053993e+00 7.99747109e-02 -3.16958636e-01
-1.49682626e-01 -1.30277977e-03 -1.32189900e-01 4.14752156e-01
-2.95191780e-02 4.07287091e-01 3.80557887e-02 -1.59390286e-01
3.07561224e-03 2.36319542e-01 6.56046197e-02 1.65851787e-01
2.71065652e-01 4.76031393e-01 -8.68146002e-01 -4.30399984e-01
-2.22236127e-01 4.18782741e-01 -3.02681059e-01 -2.53093451e-01
4.07040045e-02 4.88161385e-01 -1.68069839e-01 5.35382450e-01
1.33663893e-01 2.50316679e-01 3.92906547e-01 1.85959414e-01
-3.28496218e-01 1.00854063e+00 -1.01529348e+00 1.37209821e+00
-7.53079355e-01 7.22116888e-01 2.95343637e-01 -7.81794965e-01
7.76469946e-01 3.09084713e-01 1.03608213e-01 -7.08389044e-01
3.94530803e-01 5.37933826e-01 8.19740117e-01 -4.15436119e-01
5.86531937e-01 -6.11504793e-01 -4.91175026e-01 4.36155289e-01
9.64995101e-02 -9.99086052e-02 6.64782465e-01 -2.08022147e-01
1.23483527e+00 9.02128667e-02 2.87870109e-01 -5.16466975e-01
1.73639096e-02 2.54999369e-01 9.08901811e-01 6.65446520e-01
-4.45110589e-01 4.80628163e-01 7.58310795e-01 -3.55959088e-02
-1.04059017e+00 -8.29131961e-01 -7.39549875e-01 1.59863234e+00
-3.07701588e-01 -3.29195410e-01 -6.25752270e-01 -4.70497519e-01
-8.46895203e-02 1.10850942e+00 -4.82935399e-01 -2.80028731e-01
-8.15779865e-01 -8.80391240e-01 8.77925158e-01 5.88312447e-01
-8.01141188e-02 -1.22864890e+00 -8.31783175e-01 6.42317533e-01
-7.62543008e-02 -9.97369945e-01 -4.63486791e-01 5.57673097e-01
-9.03792679e-01 -7.95218468e-01 -3.03338468e-01 -9.38099802e-01
-3.05817127e-02 -7.21580163e-02 1.40013492e+00 -2.48240560e-01
1.29700139e-01 3.37293595e-01 -6.47249162e-01 -7.36421168e-01
-1.23940468e+00 4.61292535e-01 -1.69798940e-01 -6.39522612e-01
4.56547290e-01 -5.08756399e-01 8.42525512e-02 5.51594757e-02
-6.82045758e-01 -2.28592515e-01 5.40521383e-01 9.19708192e-01
2.03703225e-01 -3.69544178e-02 6.52607918e-01 -1.30398011e+00
1.00378108e+00 -5.79351306e-01 -3.69355589e-01 7.35197589e-02
-3.62143457e-01 1.43134907e-01 7.25440741e-01 -9.16881382e-01
-1.17746615e+00 -1.83854282e-01 -1.82615966e-01 2.93368012e-01
-3.46120477e-01 6.89415574e-01 -2.45340914e-01 2.91432589e-01
1.00767505e+00 -3.97713065e-01 -1.97980210e-01 -6.67718828e-01
8.89292657e-02 7.69799411e-01 2.69191116e-01 -9.88218367e-01
5.03883839e-01 -3.15627933e-01 -5.77560842e-01 -7.30245471e-01
-6.82634056e-01 -4.90576066e-02 -7.46949017e-01 -1.06671810e-01
7.56548941e-01 -7.45218515e-01 -3.11092623e-02 2.56459147e-01
-1.16347194e+00 -5.52700460e-01 -3.75538200e-01 4.74882126e-01
-2.59163707e-01 9.68948379e-02 -7.86382556e-01 -4.83918071e-01
-1.98624004e-02 -1.30516100e+00 7.72711694e-01 5.65771386e-02
-7.24475503e-01 -1.23072684e+00 7.60996267e-02 2.85945684e-01
2.77788818e-01 2.84136534e-01 1.47879922e+00 -8.44912052e-01
1.54601261e-01 -1.23754054e-01 1.22150205e-01 3.16052765e-01
3.23182881e-01 2.38779977e-01 -9.79697406e-01 -9.81320664e-02
3.12771469e-01 -2.09095761e-01 4.03887093e-01 2.63591528e-01
3.38571578e-01 -7.10462630e-02 -3.76287149e-03 3.54097724e-01
1.28811443e+00 4.88840193e-01 3.94842587e-02 7.67772019e-01
5.48175693e-01 7.74674714e-01 4.65081334e-01 -6.78548887e-02
5.27648449e-01 5.23306549e-01 -2.75973737e-01 3.79907310e-01
-1.54659361e-01 -5.54286093e-02 6.56950891e-01 1.10929370e+00
1.71600178e-01 -2.43391141e-01 -1.16961646e+00 4.64422852e-01
-1.33872879e+00 -8.41936350e-01 -1.40160233e-01 2.24696732e+00
1.19899011e+00 7.57701457e-01 4.40744102e-01 9.33276564e-02
6.78298354e-01 4.13998440e-02 -3.52052182e-01 -1.08605766e+00
-3.27258199e-01 2.40899548e-01 3.92093420e-01 4.06868279e-01
-8.56133163e-01 9.69360352e-01 6.40674639e+00 5.82881391e-01
-1.18872821e+00 1.16206639e-01 6.19874239e-01 -1.60640866e-01
-1.96235850e-01 -5.62349940e-03 -7.90135026e-01 6.56888068e-01
1.45457935e+00 -2.91411698e-01 3.99767816e-01 5.40210128e-01
9.70994681e-02 -2.01267183e-01 -1.27672434e+00 4.66754526e-01
-2.32402787e-01 -5.87741077e-01 -5.03789306e-01 -4.02757749e-02
3.90061766e-01 1.95647880e-01 -2.08191574e-01 4.92298722e-01
4.80625093e-01 -7.31178939e-01 1.21961164e+00 -8.31117406e-02
6.68648303e-01 -5.85066676e-01 5.83053827e-01 3.67819190e-01
-8.61810625e-01 -2.63229072e-01 -3.12922686e-01 -2.36402705e-01
2.42184699e-02 2.36550212e-01 -6.42537415e-01 3.83431017e-02
6.30318940e-01 2.04379231e-01 -6.40524685e-01 4.90097255e-01
-7.74892867e-02 1.05959439e+00 -3.79188865e-01 -1.13929115e-01
-1.97138954e-02 1.20391108e-01 5.68762720e-01 1.47039843e+00
8.43381230e-03 -6.08877875e-02 -9.93996561e-02 4.81366277e-01
5.08278199e-02 4.52907562e-01 -4.42786634e-01 -4.08921450e-01
7.70280957e-01 1.13475776e+00 -9.88944352e-01 2.02237386e-02
-8.03991675e-01 4.85666633e-01 4.02691483e-01 9.73931849e-02
-5.11094451e-01 -1.57533512e-01 7.42157757e-01 1.69423491e-01
1.63733348e-01 -1.55522719e-01 -6.15805924e-01 -8.54075372e-01
1.06638245e-01 -1.26169896e+00 5.35047591e-01 -3.48826826e-01
-1.50148094e+00 5.77684879e-01 1.54496223e-01 -6.60352647e-01
-3.68880361e-01 -6.94296360e-01 -4.45451915e-01 1.06612515e+00
-1.03758276e+00 -1.10598397e+00 2.03761891e-01 8.34898204e-02
7.91095972e-01 2.09023245e-02 8.96597266e-01 1.26718655e-01
-5.43171346e-01 7.72506654e-01 2.30566815e-01 2.41252199e-01
1.08108389e+00 -1.34518862e+00 6.11291707e-01 8.34097624e-01
1.50246531e-01 5.47999918e-01 8.08086991e-01 -5.75639009e-01
-1.11767805e+00 -6.35487735e-01 1.05734491e+00 -8.47778141e-01
9.53636229e-01 -4.93904918e-01 -9.90370929e-01 7.84018040e-01
1.29927993e-01 -6.04515851e-01 7.23746598e-01 6.45558000e-01
-4.60012794e-01 4.83876243e-02 -9.84628320e-01 6.81900024e-01
1.16687059e+00 -5.45908391e-01 -6.22232556e-01 1.27614615e-02
7.32442200e-01 -4.45736557e-01 -1.21823466e+00 1.37475818e-01
5.66882849e-01 -9.83371794e-01 4.70074952e-01 -5.73828220e-01
3.91845226e-01 3.52228731e-01 -1.81164294e-01 -1.51369691e+00
-6.52083039e-01 -3.04861933e-01 7.28717625e-01 1.96124339e+00
7.18901098e-01 -6.23828471e-01 1.41436696e-01 9.46344852e-01
-1.69341385e-01 -4.60672617e-01 -8.99098396e-01 -7.31762290e-01
7.84440577e-01 -6.08644605e-01 6.52853668e-01 8.62185538e-01
2.01526120e-01 7.82149851e-01 3.55107754e-01 -1.17665783e-01
-1.33991987e-02 -3.22667986e-01 4.56654906e-01 -1.53407407e+00
-5.84398627e-01 -7.78301716e-01 -3.64335239e-01 -2.97636241e-01
8.42630342e-02 -8.97311151e-01 1.94003999e-01 -1.03734505e+00
-1.64112430e-02 -6.70597970e-01 -2.54296958e-01 2.55899429e-01
-3.17571938e-01 -1.81881368e-01 5.62643111e-01 1.84137881e-01
-1.11880139e-01 -1.30642593e-01 8.82059038e-01 3.58649939e-01
-5.55420339e-01 4.89904778e-03 -1.14766467e+00 8.07352126e-01
9.52268541e-01 -6.83487594e-01 -3.49368721e-01 -8.93624365e-01
1.31107301e-01 1.01416856e-01 -2.63310343e-01 -8.27626526e-01
-1.24958776e-01 -7.85336643e-02 3.48857492e-01 3.36270422e-01
1.19356364e-01 -3.02220494e-01 4.07811031e-02 7.48533905e-02
-4.79855776e-01 5.81195295e-01 7.39681602e-01 9.35487077e-02
2.64036134e-02 -6.35635614e-01 8.63840044e-01 -3.58620971e-01
-4.52274024e-01 -3.58833551e-01 -4.73696530e-01 7.44172990e-01
4.90118474e-01 -4.12595779e-01 -1.19900540e-01 4.99211065e-03
-5.73866606e-01 -1.93629652e-01 6.94027066e-01 6.66482151e-01
-3.18432570e-01 -8.97287071e-01 -7.20899701e-01 1.32701755e-01
4.08895314e-02 -2.59050280e-01 -1.80828735e-01 5.77502549e-01
-2.09496617e-01 1.43527657e-01 -2.22393066e-01 -2.62127548e-01
-1.12330317e+00 1.84010953e-01 1.08072162e-01 -3.07979137e-01
-4.26238537e-01 1.05603004e+00 9.10045579e-02 -2.73574769e-01
3.08625842e-03 -2.43036911e-01 -1.99585468e-01 3.86129677e-01
1.36457533e-01 2.21474007e-01 1.10102639e-01 -7.57462204e-01
-3.44429135e-01 2.15606794e-01 -3.84872168e-01 -4.58307534e-01
1.32711041e+00 -8.72359499e-02 2.82673270e-01 1.00426877e+00
8.18564057e-01 3.03598285e-01 -1.13424730e+00 3.25259827e-02
4.33610618e-01 -1.92099720e-01 -7.38783628e-02 -9.68461215e-01
-6.67267442e-01 6.39504611e-01 2.77368098e-01 5.17591000e-01
1.04199731e+00 3.52804735e-02 6.30444169e-01 -2.34991342e-01
2.18785167e-01 -1.48390007e+00 -5.04615903e-01 5.66687346e-01
6.75504625e-01 -1.04707289e+00 -2.39367262e-01 -8.31857324e-02
-6.93772674e-01 1.13922870e+00 6.25766993e-01 1.97027177e-01
4.36218709e-01 4.96672004e-01 3.56470555e-01 1.83029667e-01
-8.27394426e-01 -2.16789879e-02 -2.24057525e-01 6.04840338e-01
1.07961464e+00 1.58532128e-01 -7.92612553e-01 8.87670040e-01
-8.25702310e-01 -5.04895926e-01 5.19687235e-01 8.50832462e-01
-2.90067494e-01 -1.61856532e+00 -5.98561347e-01 5.24138153e-01
-7.28875160e-01 -1.55123323e-01 -4.87229407e-01 1.25626171e+00
2.56907165e-01 1.05285728e+00 7.48373643e-02 -4.19494718e-01
4.73012745e-01 4.97113466e-01 4.99560952e-01 -9.94870842e-01
-9.10046816e-01 1.64606035e-01 5.48826396e-01 3.69232409e-02
-1.79960817e-01 -1.38531792e+00 -1.03903675e+00 -2.41199523e-01
-5.00313759e-01 5.64453714e-02 7.31475472e-01 1.08003259e+00
1.45835310e-01 5.08893192e-01 1.98585406e-01 -4.96708244e-01
-8.05402637e-01 -1.29905939e+00 -5.61711073e-01 4.83367920e-01
-3.68823968e-02 -6.53012753e-01 -4.44729239e-01 7.82736540e-02]
|
[10.672430992126465, 9.589637756347656]
|
27c30247-64a9-4954-a1d1-aa3e04a936ad
|
language-adaptive-weight-generation-for-multi-1
|
2306.04652
| null |
https://arxiv.org/abs/2306.04652v1
|
https://arxiv.org/pdf/2306.04652v1.pdf
|
Language Adaptive Weight Generation for Multi-task Visual Grounding
|
Although the impressive performance in visual grounding, the prevailing approaches usually exploit the visual backbone in a passive way, i.e., the visual backbone extracts features with fixed weights without expression-related hints. The passive perception may lead to mismatches (e.g., redundant and missing), limiting further performance improvement. Ideally, the visual backbone should actively extract visual features since the expressions already provide the blueprint of desired visual features. The active perception can take expressions as priors to extract relevant visual features, which can effectively alleviate the mismatches. Inspired by this, we propose an active perception Visual Grounding framework based on Language Adaptive Weights, called VG-LAW. The visual backbone serves as an expression-specific feature extractor through dynamic weights generated for various expressions. Benefiting from the specific and relevant visual features extracted from the language-aware visual backbone, VG-LAW does not require additional modules for cross-modal interaction. Along with a neat multi-task head, VG-LAW can be competent in referring expression comprehension and segmentation jointly. Extensive experiments on four representative datasets, i.e., RefCOCO, RefCOCO+, RefCOCOg, and ReferItGame, validate the effectiveness of the proposed framework and demonstrate state-of-the-art performance.
|
['Xi Li', 'Zheyang Li', 'Liang Qiao', 'Gaoang Wang', 'Huanzhang Dou', 'Peihan Miao', 'Wei Su']
|
2023-06-06
|
language-adaptive-weight-generation-for-multi
|
http://openaccess.thecvf.com//content/CVPR2023/html/Su_Language_Adaptive_Weight_Generation_for_Multi-Task_Visual_Grounding_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Su_Language_Adaptive_Weight_Generation_for_Multi-Task_Visual_Grounding_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['visual-grounding', 'referring-expression']
|
['computer-vision', 'computer-vision']
|
[ 1.60731986e-01 3.53992254e-01 -2.33877495e-01 -4.50825185e-01
-7.06732452e-01 -6.82071567e-01 4.78501052e-01 1.96914256e-01
-5.38826942e-01 2.74531454e-01 3.70091766e-01 7.12703392e-02
9.41463336e-02 -4.99404609e-01 -8.38656068e-01 -7.44688451e-01
4.69235569e-01 2.75357086e-02 1.64721444e-01 -5.00046134e-01
1.98661223e-01 1.38364639e-02 -1.57097900e+00 3.98272067e-01
1.23209226e+00 1.13283753e+00 5.27735293e-01 3.01326871e-01
-5.21856129e-01 7.82985389e-01 -5.50263047e-01 -5.69530249e-01
1.96534358e-02 -1.51369080e-01 -6.88742459e-01 3.26140344e-01
3.15130293e-01 -2.47811824e-01 -2.31597975e-01 1.15688264e+00
4.58172292e-01 1.96904942e-01 4.01927531e-01 -1.38146210e+00
-8.27608287e-01 6.49334133e-01 -7.53968477e-01 -1.75015286e-01
5.76891840e-01 6.19014561e-01 1.40373480e+00 -1.12782919e+00
6.10812306e-01 1.33379745e+00 1.59120694e-01 6.34035826e-01
-1.12368572e+00 -5.69143951e-01 8.76401186e-01 1.84007764e-01
-1.35312223e+00 -6.06994748e-01 1.15543532e+00 -2.82955140e-01
7.40122855e-01 2.11273462e-01 8.58521640e-01 1.26142919e+00
-3.25010836e-01 1.41873503e+00 1.11299384e+00 -2.01287270e-01
4.32015993e-02 6.74193054e-02 2.26451844e-01 8.20985973e-01
-2.10069254e-01 -5.40221296e-02 -1.00154018e+00 9.63391140e-02
5.77962577e-01 -1.00755662e-01 -6.05627358e-01 -5.39098203e-01
-1.24336541e+00 5.21427810e-01 7.15121031e-01 1.77558780e-01
-4.13102210e-01 1.10186055e-01 5.21893919e-01 7.33340979e-02
2.11036399e-01 1.74124002e-01 -3.01086545e-01 -1.20512992e-01
-6.92532063e-01 2.02373236e-01 1.92227066e-01 1.34212160e+00
9.79216754e-01 2.36967076e-02 -6.14517570e-01 9.95343566e-01
6.44321442e-01 4.78346646e-01 1.82500482e-01 -9.73211765e-01
7.25150645e-01 9.59350765e-01 6.95573017e-02 -1.16654265e+00
-3.55403781e-01 -3.79612058e-01 -7.50220239e-01 3.13194236e-04
4.10770327e-01 2.88447589e-02 -9.24419463e-01 2.18954372e+00
2.46629566e-01 -3.36770952e-01 5.84890358e-02 1.06915832e+00
1.31978774e+00 5.60326695e-01 3.68253171e-01 -3.11859343e-02
1.49223530e+00 -1.05708408e+00 -8.85126889e-01 -5.43359339e-01
5.20104825e-01 -5.60031891e-01 1.65092564e+00 2.72407681e-01
-1.09381795e+00 -6.64724588e-01 -9.91988659e-01 -2.69632936e-01
-3.71186793e-01 2.74267972e-01 9.02660370e-01 1.06935605e-01
-9.05295789e-01 -6.62153289e-02 -6.03593707e-01 -1.68495119e-01
5.29399872e-01 2.47603968e-01 -4.61972982e-01 8.48973636e-03
-1.11073971e+00 4.98889357e-01 6.00423753e-01 5.90781629e-01
-7.71917045e-01 -4.01756227e-01 -1.08663309e+00 6.68652877e-02
7.34850049e-01 -6.74155951e-01 1.12793112e+00 -1.30802667e+00
-1.46348679e+00 1.08843780e+00 -2.95501918e-01 1.34352436e-02
6.02956653e-01 -4.02811438e-01 -7.51909539e-02 4.28565979e-01
6.96302801e-02 1.09863544e+00 8.07439387e-01 -1.65086925e+00
-3.51015836e-01 -3.47098410e-01 5.33741832e-01 3.99867803e-01
-3.31899434e-01 -1.88171506e-01 -1.10318768e+00 -6.01926386e-01
2.39304706e-01 -8.53081584e-01 -1.99703281e-04 2.40494877e-01
-7.58420467e-01 -4.87287015e-01 5.04262447e-01 -6.46259308e-01
9.98273492e-01 -2.29729271e+00 4.10637438e-01 1.34749755e-01
5.71521282e-01 -1.77762061e-02 -2.25896105e-01 2.05385059e-01
4.75618541e-02 8.80410075e-02 -2.27153957e-01 -4.65489298e-01
1.72137320e-01 3.34382713e-01 -2.10909307e-01 9.91981179e-02
3.68494183e-01 1.37974083e+00 -1.03001249e+00 -8.09706509e-01
1.22116990e-01 3.23871106e-01 -5.44293821e-01 4.77481335e-01
-4.76579458e-01 5.08675396e-01 -6.17223144e-01 7.08038867e-01
7.83279657e-01 -3.37748438e-01 3.53306867e-02 -5.85813761e-01
8.20362121e-02 -1.35815486e-01 -7.66974926e-01 2.12669587e+00
-6.05198324e-01 4.37594116e-01 3.04104894e-01 -1.00849020e+00
1.01987982e+00 5.80810420e-02 1.14602298e-01 -1.12415600e+00
8.66664797e-02 -8.53608772e-02 -2.19530072e-02 -7.05297530e-01
5.01422405e-01 1.91495076e-01 -3.21898341e-01 2.05559712e-02
2.07856670e-01 -1.40556386e-02 7.59867346e-03 6.86224937e-01
5.66596806e-01 5.69371939e-01 1.54981956e-01 -4.38732021e-02
6.20154738e-01 -2.37084016e-01 7.57751167e-01 5.11783421e-01
-3.81601334e-01 4.39368725e-01 6.54957116e-01 1.60784408e-01
-4.74730849e-01 -1.11209249e+00 3.39906424e-01 1.49667621e+00
5.72054625e-01 -6.41376197e-01 -7.09011436e-01 -6.92424357e-01
-4.42075014e-01 6.14366293e-01 -7.39518106e-01 -2.26582393e-01
-4.61377531e-01 -3.88945788e-01 3.98317367e-01 6.49314940e-01
8.56736839e-01 -1.27345073e+00 -6.67484343e-01 -7.58648198e-03
-4.70961452e-01 -1.26010847e+00 -3.88739049e-01 9.51837525e-02
-4.87956405e-01 -8.63795817e-01 -8.18617463e-01 -7.27599382e-01
8.07175815e-01 2.79907912e-01 1.19475722e+00 3.08248192e-01
1.42424271e-01 6.15160704e-01 -4.77428943e-01 -2.12466404e-01
6.41467376e-03 -5.86301610e-02 -5.55096388e-01 2.12442949e-01
1.64294511e-01 -5.18404961e-01 -7.10165203e-01 4.96778451e-02
-7.77697802e-01 4.66509908e-01 7.31540978e-01 8.34972918e-01
7.15911508e-01 -6.01179004e-01 3.53285253e-01 -8.11978221e-01
7.67360210e-01 -2.79570520e-01 -3.20380539e-01 5.36721289e-01
-7.80413300e-02 1.73852369e-02 5.47088742e-01 -4.41087991e-01
-1.24163973e+00 1.44989222e-01 -1.19539328e-01 -5.80479860e-01
-9.20852721e-02 5.26994824e-01 -8.30795705e-01 7.27557465e-02
2.97903299e-01 3.85735631e-01 -2.67896146e-01 -2.44835868e-01
8.38679373e-01 3.77424717e-01 7.41288483e-01 -8.66510808e-01
6.31681085e-01 3.25347185e-01 -2.95199841e-01 -6.68900073e-01
-1.07076645e+00 -3.99876028e-01 -6.29381478e-01 -4.47166145e-01
1.06772578e+00 -1.02302957e+00 -9.54772353e-01 4.85915333e-01
-1.38647747e+00 -2.19242096e-01 -1.28692031e-01 6.46184385e-03
-5.27127147e-01 3.90549123e-01 -3.38188112e-01 -8.61539960e-01
-2.69611716e-01 -1.26959097e+00 1.36935639e+00 5.93609691e-01
-2.58763760e-01 -9.40591455e-01 -5.00616968e-01 5.24042010e-01
1.96149349e-01 3.60080928e-01 1.04546762e+00 -3.96164864e-01
-7.50187159e-01 1.58837363e-01 -7.25629270e-01 1.46506757e-01
-1.98182210e-01 2.62492355e-02 -1.21806860e+00 -2.24745292e-02
-2.89213359e-01 -8.06933284e-01 1.00011659e+00 -8.28617346e-03
1.36060166e+00 -2.10086465e-01 -2.34725773e-01 7.16150820e-01
1.21217585e+00 5.20487651e-02 5.57423115e-01 1.39121234e-01
9.73337054e-01 8.53642523e-01 7.94443190e-01 2.82865405e-01
8.27971876e-01 7.19278991e-01 7.03193545e-01 -4.04271692e-01
-5.49449399e-02 -5.15015006e-01 3.62673789e-01 9.61458504e-01
-1.36055440e-01 -1.34631932e-01 -8.89831662e-01 4.06912148e-01
-2.04524231e+00 -6.87328637e-01 5.09913824e-02 1.76953387e+00
9.15743589e-01 1.83789253e-01 -4.24795598e-02 -1.03954218e-01
4.20005143e-01 4.72262055e-01 -8.15817654e-01 -2.33247787e-01
-4.91668314e-01 -7.76389912e-02 1.34583665e-02 2.85225481e-01
-8.78457844e-01 1.29725170e+00 4.91403055e+00 8.74457300e-01
-1.04942143e+00 1.52429059e-01 4.79378700e-01 9.67534781e-02
-6.76597178e-01 1.58647880e-01 -4.83264387e-01 2.70000964e-01
-3.97823472e-03 3.28139663e-02 2.88624555e-01 7.14657068e-01
1.44507453e-01 -1.94588825e-01 -9.66016769e-01 1.43808413e+00
1.63049921e-01 -9.68253434e-01 2.43227527e-01 -8.47158208e-02
2.92230457e-01 -3.35644394e-01 1.08878940e-01 5.77098966e-01
4.70235422e-02 -8.81950617e-01 1.18261075e+00 7.50728965e-01
6.93102121e-01 -7.43863702e-01 5.66635489e-01 2.69667864e-01
-1.21672714e+00 8.76619741e-02 -1.79152951e-01 1.20286077e-01
2.25774497e-01 2.41361603e-01 -1.86602280e-01 6.80585146e-01
6.80924177e-01 8.93627405e-01 -8.51575613e-01 6.80199087e-01
-6.13819122e-01 5.78276336e-01 -6.14894461e-03 -8.70046914e-02
3.61604244e-01 -1.42370075e-01 5.73431253e-01 1.16758251e+00
-1.92095935e-01 4.43817526e-02 4.82016087e-01 1.15453577e+00
2.23692600e-02 3.94867837e-01 -2.64244974e-01 -1.16867259e-01
3.94282877e-01 1.35204697e+00 -5.96422970e-01 -2.51981944e-01
-5.35246313e-01 1.12865102e+00 6.61407530e-01 7.63202667e-01
-8.27297211e-01 -4.61980492e-01 3.74898076e-01 -1.47334486e-01
1.88796967e-01 -2.90733218e-01 -2.01036260e-01 -1.38612020e+00
2.79441237e-01 -9.27735507e-01 1.81374043e-01 -1.09816039e+00
-1.20824671e+00 6.45345926e-01 1.17056809e-01 -1.08369696e+00
2.82383226e-02 -5.72780311e-01 -6.63843691e-01 8.22721601e-01
-1.37526286e+00 -1.66247714e+00 -7.51654744e-01 8.02114069e-01
4.72068250e-01 7.16798902e-02 5.37070513e-01 -1.25499755e-01
-5.88295877e-01 8.04970682e-01 -7.38247037e-01 2.84220308e-01
5.25357008e-01 -1.18215144e+00 -3.04744504e-02 7.10294127e-01
2.51535565e-01 9.52781796e-01 6.63815618e-01 -4.62249815e-01
-1.43266320e+00 -6.99653804e-01 4.36447769e-01 -9.58316699e-02
5.69412947e-01 -8.01681995e-01 -1.01325977e+00 5.11672378e-01
4.56617594e-01 -4.59553115e-02 7.20865548e-01 1.73887834e-01
-5.36343336e-01 3.61410752e-02 -8.09369564e-01 9.27946687e-01
1.54633605e+00 -7.07669079e-01 -7.00528324e-01 1.93078309e-01
1.00376141e+00 -4.27705467e-01 -4.99788284e-01 2.01712504e-01
4.97524053e-01 -1.03380585e+00 9.42645907e-01 -5.67678392e-01
4.16495860e-01 -2.52528727e-01 -3.14010799e-01 -1.06604934e+00
-1.37081012e-01 -6.76141083e-01 1.64769366e-02 1.79973447e+00
3.71734917e-01 -4.48502183e-01 5.07169783e-01 6.71234310e-01
-1.57051295e-01 -8.96363497e-01 -7.65854895e-01 -4.86221433e-01
-1.47107735e-01 -7.19285727e-01 5.57706833e-01 8.31729472e-01
-1.34648860e-01 5.71228921e-01 -3.36218655e-01 -3.56181040e-02
5.09538651e-01 3.23011070e-01 8.48944426e-01 -1.02507901e+00
-3.94914329e-01 -5.06356835e-01 -2.09423944e-01 -1.61044788e+00
5.59732616e-01 -9.45026696e-01 1.46064013e-01 -1.55808771e+00
2.90316373e-01 -3.96286607e-01 -3.78063828e-01 8.24702740e-01
-5.27862251e-01 -1.11478381e-01 7.21990645e-01 1.41811162e-01
-9.34231520e-01 9.19906437e-01 1.64970124e+00 -3.85210544e-01
-1.59593448e-01 -3.58038634e-01 -1.02185690e+00 1.00617826e+00
6.22671664e-01 -5.73112182e-02 -6.80110931e-01 -5.80426097e-01
5.46929181e-01 -2.63624657e-02 5.69980919e-01 -4.94802177e-01
1.53867155e-01 -2.32078627e-01 3.30789387e-01 -5.90904832e-01
5.69793582e-01 -6.51097357e-01 -5.46423435e-01 -1.75487712e-01
-3.90483946e-01 1.85458828e-02 1.99911490e-01 5.77668130e-01
-4.81765866e-01 -9.73402038e-02 4.22337383e-01 -3.29658687e-01
-1.24520695e+00 2.05257982e-01 -8.22119266e-02 3.96467924e-01
6.99233532e-01 -3.75956625e-01 -3.46020132e-01 -5.20099103e-01
-8.08933020e-01 5.98482847e-01 4.44321364e-01 5.86681187e-01
8.18546951e-01 -1.36842096e+00 -3.79994929e-01 3.59164365e-03
7.22517014e-01 2.69504875e-01 5.11877716e-01 9.48516488e-01
9.10609309e-03 6.62485585e-02 -1.37009755e-01 -8.51174533e-01
-1.22337377e+00 6.29378557e-01 2.36587539e-01 -1.22855946e-01
-4.01389331e-01 1.16657460e+00 8.04963827e-01 -3.80400479e-01
1.97283521e-01 -3.83627355e-01 -4.14070010e-01 3.29385787e-01
3.67393672e-01 -2.37602815e-01 -3.40925217e-01 -9.75724995e-01
-3.69338214e-01 8.52347195e-01 -1.43181518e-01 -3.27666670e-01
1.05896699e+00 -3.78715277e-01 -3.99054214e-02 6.00810945e-01
9.91368651e-01 2.28392527e-01 -1.46298444e+00 -3.65499616e-01
-1.22671224e-01 -3.92156273e-01 -6.16298504e-02 -8.64269972e-01
-1.27883232e+00 1.30423522e+00 3.00653458e-01 -1.17728688e-01
1.45646930e+00 2.27420270e-01 5.24137735e-01 2.74403065e-01
3.92591655e-01 -9.93738115e-01 4.37774569e-01 3.19809973e-01
1.27175975e+00 -1.27355933e+00 -2.30285451e-01 -6.47482932e-01
-1.06972671e+00 8.37338507e-01 9.73273456e-01 1.46167427e-01
2.02642292e-01 -1.52947724e-01 2.62679636e-01 -2.87079543e-01
-7.76333570e-01 -5.74447870e-01 3.80545259e-01 6.96913600e-01
4.98565495e-01 5.43648861e-02 -6.90083727e-02 9.08629775e-01
-1.96475938e-01 -3.04375768e-01 2.16470123e-03 7.94235706e-01
-1.96946234e-01 -8.98144603e-01 -6.11541122e-02 1.54710189e-01
1.11955255e-02 -1.89590633e-01 -7.21968770e-01 8.92396092e-01
3.40500891e-01 9.61993575e-01 -1.04543440e-01 -2.12692559e-01
6.08132422e-01 -2.08925288e-02 4.68461424e-01 -3.41403335e-01
-7.01443076e-01 1.94221884e-01 7.24382326e-02 -8.26757848e-01
-5.98823845e-01 -4.45456296e-01 -1.58052623e+00 2.24393487e-01
-2.08966479e-01 -1.68215439e-01 2.31313929e-01 1.03653562e+00
3.55230510e-01 6.44904792e-01 2.10073128e-01 -7.42466986e-01
-1.85642242e-01 -9.05314445e-01 -2.84920633e-01 8.56933177e-01
2.01952189e-01 -8.77830207e-01 -1.05312057e-01 -9.13148373e-03]
|
[10.447237968444824, 1.3096363544464111]
|
365c36c1-b43e-4f69-852f-40159b0b105d
|
semantic-line-detection-using-mirror-1
|
2203.15285
| null |
https://arxiv.org/abs/2203.15285v1
|
https://arxiv.org/pdf/2203.15285v1.pdf
|
Semantic Line Detection Using Mirror Attention and Comparative Ranking and Matching
|
A novel algorithm to detect semantic lines is proposed in this paper. We develop three networks: detection network with mirror attention (D-Net) and comparative ranking and matching networks (R-Net and M-Net). D-Net extracts semantic lines by exploiting rich contextual information. To this end, we design the mirror attention module. Then, through pairwise comparisons of extracted semantic lines, we iteratively select the most semantic line and remove redundant ones overlapping with the selected one. For the pairwise comparisons, we develop R-Net and M-Net in the Siamese architecture. Experiments demonstrate that the proposed algorithm outperforms the conventional semantic line detector significantly. Moreover, we apply the proposed algorithm to detect two important kinds of semantic lines successfully: dominant parallel lines and reflection symmetry axes. Our codes are available at https://github.com/dongkwonjin/Semantic-Line-DRM.
|
['Chang-Su Kim', 'Jun-Tae Lee', 'Dongkwon Jin']
|
2022-03-29
|
semantic-line-detection-using-mirror
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3397_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123650120.pdf
|
eccv-2020-8
|
['line-detection']
|
['computer-vision']
|
[-9.00793374e-02 -8.01321268e-02 -1.42935768e-01 -3.49445105e-01
-5.96678197e-01 -4.60121065e-01 3.51055950e-01 -1.28456131e-01
-2.44589582e-01 2.43507221e-01 3.06592137e-01 -1.05882816e-01
-3.48940998e-01 -9.26451147e-01 -6.48744583e-01 -2.60843307e-01
1.32786080e-01 2.08601341e-01 4.48456973e-01 -2.26265073e-01
6.43461883e-01 6.71344280e-01 -1.16948211e+00 3.05299371e-01
9.30411875e-01 8.03842902e-01 1.06177822e-01 3.26597244e-01
-4.62224275e-01 5.55793881e-01 -3.20188046e-01 -3.83468300e-01
2.94012696e-01 -4.89125669e-01 -7.98390329e-01 -1.14021979e-01
4.59895462e-01 -2.09119573e-01 -5.65667689e-01 1.22839284e+00
4.81229722e-01 2.31580824e-01 5.60840726e-01 -1.51932132e+00
-6.83577418e-01 7.16869295e-01 -1.22385347e+00 1.80248976e-01
4.79647368e-01 -2.46349454e-01 1.37139738e+00 -1.50249779e+00
5.69067478e-01 1.32039177e+00 8.17911983e-01 1.24117188e-01
-6.51999652e-01 -7.66572237e-01 1.99660838e-01 3.64256263e-01
-1.62669218e+00 -1.79696202e-01 1.24365366e+00 -2.24195853e-01
6.77346349e-01 6.08462617e-02 5.31316340e-01 7.79865980e-01
-2.65333831e-01 1.17897582e+00 5.93543589e-01 -5.00509918e-01
-1.51129141e-01 -6.73038438e-02 3.96775663e-01 9.85374451e-01
3.47618848e-01 -1.94223925e-01 -5.96855760e-01 1.09224573e-01
8.58293831e-01 -5.50042139e-04 -3.97224844e-01 -4.89344120e-01
-1.19625700e+00 7.03358769e-01 8.79056096e-01 5.37823081e-01
4.31233309e-02 1.22009376e-02 4.43831801e-01 1.01770923e-01
1.99730784e-01 2.97271758e-01 -1.62704915e-01 3.68277907e-01
-8.91536355e-01 7.64104724e-02 4.37423885e-01 1.51828170e+00
8.47423196e-01 -3.90430570e-01 1.40731409e-02 1.24816442e+00
3.55990261e-01 3.29195231e-01 4.56771255e-01 -6.78104401e-01
6.32537782e-01 8.02163064e-01 -7.69984573e-02 -1.41572046e+00
-8.12678993e-01 -6.13829911e-01 -7.73121774e-01 -1.85338095e-01
3.22252125e-01 2.39178985e-02 -4.88784105e-01 1.23307240e+00
3.97342086e-01 1.57646492e-01 -1.20993786e-01 1.14165628e+00
1.07188749e+00 5.04837930e-01 -4.05043870e-01 4.61848915e-01
1.47678602e+00 -1.59796178e+00 -4.08763498e-01 -1.25284314e-01
7.53691018e-01 -9.90221560e-01 1.10938811e+00 1.23727851e-01
-1.02414787e+00 -6.74529791e-01 -1.22438037e+00 -5.22555828e-01
-3.82817954e-01 6.05920911e-01 4.27251041e-01 1.55905351e-01
-8.44793618e-01 4.74347621e-01 -3.66568595e-01 -5.93756139e-01
5.10215402e-01 2.06807226e-01 -1.63621500e-01 -4.09760624e-02
-1.14973319e+00 3.17429274e-01 2.85322547e-01 4.33973968e-01
-2.09898114e-01 -4.99575883e-01 -8.99542689e-01 1.81434929e-01
4.59904939e-01 -7.63324082e-01 1.11469877e+00 -1.10756755e+00
-1.30971229e+00 9.21520710e-01 -2.25829422e-01 -1.65347785e-01
7.67962635e-01 -6.83389127e-01 -4.15025592e-01 4.31098938e-01
4.56466258e-01 6.48097098e-01 4.40934658e-01 -1.38005686e+00
-7.54072309e-01 -3.55949491e-01 -1.80190980e-01 2.57362753e-01
-2.81535417e-01 4.46165316e-02 -9.51258421e-01 -8.05114090e-01
6.08819604e-01 -6.92306757e-01 1.14925550e-02 5.22982739e-02
-1.07466888e+00 -4.02205318e-01 8.74291956e-01 -6.39149785e-01
1.16241217e+00 -2.16946387e+00 -2.86138535e-01 8.29784930e-01
3.69765282e-01 -4.34266031e-02 -2.78624028e-01 5.41465104e-01
-1.55544117e-01 9.51872617e-02 -3.16788018e-01 -9.58059505e-02
-5.62346317e-02 -5.25412321e-01 -1.28762588e-01 5.02153397e-01
3.83006446e-02 8.27223361e-01 -7.96137869e-01 -5.26710749e-01
1.98715895e-01 2.02368364e-01 -5.13835132e-01 -7.18100071e-02
1.11960590e-01 1.07156835e-03 -6.14259481e-01 7.36164868e-01
8.77910674e-01 -3.85484040e-01 2.00544700e-01 -6.26316190e-01
-3.24918479e-01 8.64606872e-02 -1.22278547e+00 1.79739451e+00
-2.52251238e-01 6.45449042e-01 -2.48232394e-01 -8.62689197e-01
1.34934866e+00 -2.37601757e-01 6.41075134e-01 -1.03675997e+00
2.19005778e-01 4.13419694e-01 -1.70449123e-01 -5.82176924e-01
6.09082043e-01 6.03355229e-01 4.09818441e-02 2.86067635e-01
-1.27549559e-01 1.74666300e-01 2.70869583e-01 4.12793845e-01
7.25214839e-01 3.49887460e-01 1.48774818e-01 -4.76777375e-01
9.30018604e-01 6.81816880e-03 7.55263090e-01 6.38761818e-01
-1.63077936e-01 5.90824485e-01 5.08411884e-01 -4.01791245e-01
-1.17486644e+00 -1.04159188e+00 1.63074844e-02 8.88576448e-01
8.66582870e-01 -5.22709548e-01 -8.09242308e-01 -6.67256832e-01
1.14526562e-01 4.83282655e-01 -3.90445262e-01 -5.65131940e-02
-6.89516127e-01 -2.81157225e-01 4.22219485e-01 7.39912689e-01
8.69938791e-01 -1.04093122e+00 -3.13319743e-01 -7.65738124e-03
-8.34729671e-02 -9.77184355e-01 -7.60841131e-01 -3.40784997e-01
-5.08354425e-01 -1.35247588e+00 -9.56837773e-01 -1.21499026e+00
9.47070599e-01 7.29965031e-01 8.65821600e-01 1.98647782e-01
-3.81616145e-01 3.62663493e-02 -3.65068555e-01 4.68789190e-02
2.80450761e-01 2.17234597e-01 -3.46781552e-01 1.63093105e-01
5.67017078e-01 -3.25151056e-01 -8.89115572e-01 4.60944742e-01
-3.18791360e-01 3.93107057e-01 5.73194087e-01 7.82450020e-01
4.54372615e-01 -1.79362632e-02 3.19992959e-01 -9.18575585e-01
5.78583837e-01 -3.41354877e-01 -7.78354764e-01 1.67297259e-01
-1.98549360e-01 -1.94213595e-02 6.33510113e-01 1.73801228e-01
-1.05999970e+00 -8.40992555e-02 -1.37471929e-01 -4.88109678e-01
-9.70990136e-02 2.05455676e-01 -2.41821632e-01 5.15102968e-02
2.44301856e-01 -3.07479147e-02 -4.51616406e-01 -4.81036842e-01
4.21811581e-01 6.65903509e-01 6.58977389e-01 -5.33232570e-01
1.00400186e+00 7.34885693e-01 -1.13570936e-01 -9.94478166e-01
-9.04162884e-01 -8.53318274e-01 -8.88580322e-01 -2.26864949e-01
6.82082355e-01 -6.49785578e-01 -5.35904706e-01 5.89253664e-01
-1.12151158e+00 -3.16372067e-02 6.63798442e-03 4.03163940e-01
-4.25054818e-01 4.94311661e-01 -7.30900347e-01 -4.71015424e-01
-5.21440268e-01 -9.18518424e-01 1.18336511e+00 4.69497502e-01
-2.67323643e-01 -9.05121028e-01 -1.06286164e-02 4.53085423e-01
-4.51647155e-02 -1.53695662e-02 7.89880455e-01 -7.04858363e-01
-5.54010272e-01 4.92672399e-02 -9.20010269e-01 -9.70077813e-02
-2.00304613e-02 2.87537426e-01 -8.06235373e-01 1.58009157e-02
-6.48399711e-01 5.32331988e-02 1.07780921e+00 3.73982489e-01
1.34956145e+00 1.91546604e-02 -5.67484736e-01 8.76815379e-01
1.52348709e+00 2.53105849e-01 6.32555425e-01 7.49700427e-01
1.00022113e+00 7.91307867e-01 4.30435777e-01 3.59767497e-01
4.61297393e-01 4.62258905e-01 -1.55713013e-03 -5.36035299e-01
-1.03933021e-01 -4.21757370e-01 9.90278646e-02 8.61707926e-01
2.12332502e-01 -1.72515810e-01 -9.55523491e-01 4.79123920e-01
-2.03565192e+00 -9.77090001e-01 -5.61954260e-01 1.94873619e+00
5.15741259e-02 4.22888607e-01 1.82439402e-01 1.83184355e-01
1.07944810e+00 -1.22017153e-02 -5.03266215e-01 -1.78667948e-01
-2.94146299e-01 -5.15167043e-02 5.36230922e-01 4.92188543e-01
-1.16674089e+00 1.20759726e+00 4.75941372e+00 8.19211841e-01
-7.16803312e-01 -1.16188206e-01 4.80458200e-01 1.45912096e-01
-5.19149959e-01 5.48086129e-02 -6.79710567e-01 3.11249077e-01
4.06071283e-02 -4.75110151e-02 -1.01477141e-02 7.37647593e-01
3.33643615e-01 3.12564187e-02 -7.77167261e-01 1.08487070e+00
1.99689612e-01 -1.23120975e+00 2.30779588e-01 -3.96663874e-01
6.70773745e-01 -3.49174179e-02 -2.49503836e-01 3.99369523e-02
1.42204493e-01 -5.47602117e-01 8.62824261e-01 5.76104999e-01
5.75498641e-01 -9.75570440e-01 6.55339062e-01 -2.45534241e-01
-1.49183834e+00 6.28309250e-02 -4.00855690e-01 4.44968998e-01
1.88510725e-03 7.57874012e-01 -3.33393842e-01 8.61847579e-01
7.79275775e-01 1.10039389e+00 -4.68574405e-01 1.35244417e+00
-4.92139965e-01 2.72336721e-01 -3.25105220e-01 -4.69377786e-02
4.52492595e-01 -6.18498981e-01 5.13009608e-01 1.21520984e+00
4.44805473e-01 -3.05569649e-01 8.37248266e-02 1.05677629e+00
-3.02432716e-01 5.33497155e-01 -4.85923082e-01 3.79762769e-01
6.07813239e-01 1.45177150e+00 -1.13793766e+00 -2.49540597e-01
-7.70731807e-01 1.22496521e+00 4.23994631e-01 4.92663711e-01
-7.25307226e-01 -9.23125148e-01 4.03855473e-01 -9.06490628e-03
1.60771653e-01 -1.08101845e-01 -2.50940770e-01 -1.11416090e+00
1.11534916e-01 -4.50913727e-01 5.69048643e-01 -1.14601040e+00
-1.32914555e+00 4.90678608e-01 -4.28953499e-01 -1.16210818e+00
5.06213427e-01 -6.30060792e-01 -9.10383642e-01 5.33575773e-01
-1.58291876e+00 -1.17376590e+00 -6.21998608e-01 6.54309690e-01
6.78662121e-01 -3.49728204e-02 7.41373301e-02 3.44249547e-01
-7.85546601e-01 7.43779182e-01 3.37775379e-01 5.80113590e-01
6.30545914e-01 -1.01492763e+00 6.09686911e-01 6.77206337e-01
2.49741390e-01 6.31399214e-01 2.32902110e-01 -6.28256977e-01
-9.98479962e-01 -9.97112930e-01 8.35991681e-01 2.65932888e-01
8.34357619e-01 -4.48845625e-01 -7.85858572e-01 5.94529569e-01
1.04757547e-01 -3.41713995e-01 4.07008678e-01 9.17123854e-02
-5.40904760e-01 -1.47743389e-01 -7.72092044e-01 9.01930749e-01
1.42240489e+00 -4.10477102e-01 -4.96540964e-01 4.51040089e-01
6.68952346e-01 -4.04917859e-02 -6.05576396e-01 3.48234147e-01
6.09728396e-01 -1.22579527e+00 1.03945589e+00 -7.74685815e-02
5.04392445e-01 -3.85607332e-01 2.06185102e-01 -1.18183351e+00
-4.34103608e-01 -4.55808371e-01 4.77867812e-01 1.30885303e+00
4.26427990e-01 -7.30644882e-01 7.98270583e-01 -2.72355646e-01
-4.30920035e-01 -8.15287769e-01 -5.12292206e-01 -6.60027742e-01
-2.82251462e-02 -2.71718740e-01 6.94499612e-01 9.61075008e-01
2.65413746e-02 3.37140918e-01 -1.86457619e-01 2.92998821e-01
6.91556096e-01 5.47520339e-01 6.86442137e-01 -9.90678728e-01
1.28270371e-03 -9.67968285e-01 -3.60928744e-01 -1.31613958e+00
1.82866573e-01 -1.06399202e+00 -2.47305527e-01 -1.70708930e+00
2.10913435e-01 -4.32637066e-01 -4.57662314e-01 1.37916371e-01
3.94637790e-03 1.71713486e-01 1.76965714e-01 3.31925958e-01
-7.07911730e-01 6.91571653e-01 1.18109381e+00 -1.46345347e-01
-1.82122990e-01 -3.43806326e-01 -5.56324482e-01 1.03689623e+00
1.12759101e+00 -1.66628197e-01 -1.46085054e-01 -4.11329001e-01
2.90703535e-01 -2.77396798e-01 3.53481293e-01 -1.16257989e+00
4.38928008e-01 2.00082093e-01 5.71866572e-01 -9.01215672e-01
-1.16342887e-01 -6.04485571e-01 -4.79005277e-01 3.58938783e-01
-4.93084580e-01 1.85894772e-01 -7.98780192e-03 1.97133616e-01
-1.95522964e-01 -3.51247340e-01 6.54148400e-01 1.10381536e-01
-1.14756024e+00 3.47307950e-01 1.31584529e-03 1.39039397e-01
9.38071132e-01 -3.11272502e-01 -5.01451552e-01 -1.66710630e-01
-5.82094967e-01 7.03954518e-01 3.22851926e-01 6.06162131e-01
8.49543571e-01 -1.52944577e+00 -5.23432612e-01 3.38152707e-01
4.30827558e-01 -4.72239926e-02 2.45752946e-01 7.70817518e-01
-8.96217167e-01 3.33529443e-01 -1.70244306e-01 -3.27899069e-01
-1.14014077e+00 4.18480426e-01 4.04552430e-01 1.77910984e-01
-9.94659901e-01 9.44641352e-01 3.87519628e-01 -4.80653286e-01
-2.89210565e-02 -1.76744685e-01 -2.66900182e-01 -1.34859635e-02
2.17400566e-01 6.28549695e-01 -1.67284459e-01 -7.33393908e-01
-4.67910558e-01 1.06837022e+00 -9.71092805e-02 1.43881246e-01
1.16966665e+00 -1.94017097e-01 -1.38882905e-01 2.38143817e-01
1.34493804e+00 2.85989940e-01 -9.92079437e-01 -3.41532260e-01
2.32767731e-01 -3.54195684e-01 -4.75445800e-02 -2.15694979e-01
-1.27441454e+00 7.66995370e-01 2.82000124e-01 -4.48156036e-02
1.00535381e+00 -6.36381805e-02 9.78143573e-01 2.17366382e-01
1.95602000e-01 -1.34611356e+00 1.21920459e-01 3.56744617e-01
9.62048888e-01 -1.04647279e+00 -1.08731240e-01 -7.74959028e-01
-2.86900580e-01 1.32511604e+00 9.53855634e-01 -4.37680662e-01
5.52617073e-01 1.92004427e-01 7.09630921e-02 -5.05014718e-01
-2.03105375e-01 -3.24398011e-01 3.74069124e-01 1.72621772e-01
5.48187256e-01 -2.02668950e-01 -2.28516340e-01 2.79611260e-01
-3.87124270e-01 -2.26171404e-01 2.98336565e-01 6.72962546e-01
-4.74804461e-01 -7.13250577e-01 -4.85018879e-01 3.71637285e-01
8.56516212e-02 -2.22482875e-01 -6.11227334e-01 8.97374272e-01
-1.59849040e-02 7.53754258e-01 4.18330014e-01 -2.86278278e-01
4.98988718e-01 -1.63730785e-01 2.88879991e-01 -7.99474418e-02
-3.59334171e-01 2.51019120e-01 9.18439105e-02 -7.27666736e-01
-2.93088436e-01 -4.47319150e-01 -1.63682675e+00 -1.52171358e-01
-3.74327898e-01 1.97549671e-01 3.83301616e-01 5.67411423e-01
4.04938906e-01 5.82563937e-01 6.52924359e-01 -4.72158253e-01
-1.30517125e-01 -6.43535852e-01 -5.69583237e-01 6.41582727e-01
4.18883637e-02 -6.74966216e-01 -3.18207741e-01 -2.73702860e-01]
|
[8.346919059753418, -1.544702410697937]
|
3cbee7b9-09a7-4bea-8fcf-3c04c4e54cd1
|
diverse-and-faithful-knowledge-grounded
|
2306.01153
| null |
https://arxiv.org/abs/2306.01153v1
|
https://arxiv.org/pdf/2306.01153v1.pdf
|
Diverse and Faithful Knowledge-Grounded Dialogue Generation via Sequential Posterior Inference
|
The capability to generate responses with diversity and faithfulness using factual knowledge is paramount for creating a human-like, trustworthy dialogue system. Common strategies either adopt a two-step paradigm, which optimizes knowledge selection and response generation separately, and may overlook the inherent correlation between these two tasks, or leverage conditional variational method to jointly optimize knowledge selection and response generation by employing an inference network. In this paper, we present an end-to-end learning framework, termed Sequential Posterior Inference (SPI), capable of selecting knowledge and generating dialogues by approximately sampling from the posterior distribution. Unlike other methods, SPI does not require the inference network or assume a simple geometry of the posterior distribution. This straightforward and intuitive inference procedure of SPI directly queries the response generation model, allowing for accurate knowledge selection and generation of faithful responses. In addition to modeling contributions, our experimental results on two common dialogue datasets (Wizard of Wikipedia and Holl-E) demonstrate that SPI outperforms previous strong baselines according to both automatic and human evaluation metrics.
|
['Ying Nian Wu', 'Pascale Fung', 'Bo Pang', 'Ziwei Ji', 'Dehong Xu', 'Deqian Kong', 'Yan Xu']
|
2023-06-01
| null | null | null | null |
['dialogue-generation', 'response-generation', 'dialogue-generation']
|
['natural-language-processing', 'natural-language-processing', 'speech']
|
[ 3.45635377e-02 7.69932747e-01 -1.79016814e-01 -4.67378557e-01
-1.01991415e+00 -8.30212355e-01 1.09469807e+00 -1.44922867e-01
-4.92185622e-01 1.17623150e+00 5.70904553e-01 -4.94236685e-03
1.45169944e-01 -7.07207143e-01 -3.80704254e-01 -4.25774246e-01
4.74316537e-01 8.68356586e-01 -9.23938642e-04 -6.04726136e-01
3.58156472e-01 -1.85526446e-01 -1.02195442e+00 4.36096758e-01
9.78202224e-01 7.34773457e-01 -1.62166655e-01 9.34693515e-01
-1.32089421e-01 1.48637545e+00 -7.92352974e-01 -1.05881703e+00
-9.53257307e-02 -5.84316730e-01 -1.54240441e+00 -3.11832786e-01
-2.67158151e-02 -5.71853757e-01 -5.63748889e-02 8.54401052e-01
5.58911264e-01 4.65837151e-01 8.73338759e-01 -9.73223627e-01
-6.73891246e-01 1.15545392e+00 -1.29732862e-02 -4.19572324e-01
9.32548106e-01 3.09532315e-01 1.29320598e+00 -7.41544843e-01
6.63678348e-01 1.49106824e+00 6.38860166e-01 9.18431699e-01
-1.40227401e+00 -2.92522013e-01 3.98175512e-03 6.99734464e-02
-1.31739843e+00 -8.06308866e-01 7.45100856e-01 -4.17461157e-01
8.57555091e-01 4.11342531e-01 3.72990966e-01 1.61668146e+00
-2.27061242e-01 1.10336602e+00 1.16645682e+00 -4.18712050e-01
2.43789673e-01 7.36790657e-01 6.97282031e-02 4.83515143e-01
-2.93537080e-01 -8.88757184e-02 -1.14311922e+00 -6.51399732e-01
4.49779540e-01 -6.75308585e-01 -5.21442652e-01 4.91150878e-02
-1.31645370e+00 1.10582662e+00 -9.06865969e-02 -1.85671881e-01
-6.51019037e-01 -9.03146416e-02 4.27824169e-01 3.14256579e-01
4.78036106e-01 6.95829928e-01 -3.65826190e-01 -5.28353453e-01
-9.43675518e-01 7.34169304e-01 1.77504182e+00 9.75989521e-01
6.31783426e-01 -5.41893020e-02 -6.95345461e-01 8.73109341e-01
4.20944452e-01 3.59224975e-01 2.99311519e-01 -1.43232620e+00
4.12429959e-01 4.01267171e-01 7.76093483e-01 -9.35340643e-01
-8.55409577e-02 9.27018896e-02 -6.93744540e-01 -1.32403493e-01
7.34763801e-01 -5.33203721e-01 -1.61855802e-01 1.76329982e+00
5.97948909e-01 -3.80756766e-01 3.87271523e-01 9.63404775e-01
1.01875997e+00 5.75239003e-01 4.17705067e-02 -3.25815171e-01
1.30700684e+00 -9.03268456e-01 -8.84315372e-01 4.97114211e-02
2.61398941e-01 -7.48926461e-01 1.27039790e+00 4.88076150e-01
-1.31148350e+00 -1.88989937e-01 -7.53854275e-01 -2.55446762e-01
-4.86142561e-02 2.64241453e-02 6.25086725e-01 5.28069377e-01
-9.11578059e-01 5.40491045e-01 -4.01618510e-01 -6.17148802e-02
-2.24103089e-02 2.70296503e-02 -1.49980783e-01 3.13412637e-01
-1.72905195e+00 1.25714028e+00 3.54869217e-01 9.89344269e-02
-8.50057662e-01 -5.21276772e-01 -7.24351108e-01 -1.88988984e-01
6.32729471e-01 -9.90049541e-01 1.78642654e+00 -9.35732603e-01
-2.63462973e+00 6.85188055e-01 -7.51772821e-02 -4.48467225e-01
9.12062466e-01 -4.79273856e-01 1.48501515e-01 1.81492120e-01
-9.76099372e-02 6.02064431e-01 7.84260690e-01 -1.38787520e+00
-3.24002177e-01 1.97998449e-01 3.96224439e-01 4.89197940e-01
5.96247017e-02 -8.43689218e-02 -1.53919742e-01 -2.40591347e-01
-4.82989043e-01 -8.03549409e-01 -2.57276520e-02 -4.18663412e-01
-8.32682610e-01 -6.00739121e-01 1.05173789e-01 -7.81569660e-01
1.19712818e+00 -1.55394089e+00 3.26716840e-01 1.78505376e-01
2.53778577e-01 9.88553911e-02 1.64153814e-01 6.96125567e-01
6.51161969e-01 8.91568810e-02 -4.07660306e-02 -5.07858694e-01
4.68797475e-01 1.54135719e-01 -5.66086590e-01 1.61380395e-01
1.60048172e-01 9.41478848e-01 -1.11453378e+00 -6.85525775e-01
-5.82822077e-02 4.66740310e-01 -6.18880987e-01 6.96334600e-01
-6.38543785e-01 4.50100571e-01 -4.55414116e-01 2.79314786e-01
2.34977663e-01 -4.06903476e-01 5.57639599e-01 -1.70944557e-02
7.69734308e-02 6.59412563e-01 -9.88365769e-01 1.53476834e+00
-5.64959586e-01 2.49206990e-01 2.28223503e-01 -3.28824162e-01
7.50408292e-01 5.43944180e-01 -1.56307101e-01 -1.97463766e-01
5.80451563e-02 -5.96894184e-03 -3.08106333e-01 -5.00729322e-01
9.08124745e-01 -2.72924960e-01 -5.03817379e-01 9.99028206e-01
2.45738387e-01 -4.05967087e-01 2.06771195e-02 8.11844110e-01
5.90858936e-01 3.73372376e-01 3.75807703e-01 -1.80083901e-01
5.16388893e-01 -2.78480016e-02 4.66108471e-01 1.13513839e+00
-1.39617905e-01 3.08692247e-01 8.24308157e-01 -1.82549760e-01
-7.57979691e-01 -1.00567853e+00 2.20844433e-01 1.38786685e+00
-2.64086314e-02 -3.26867640e-01 -8.80973935e-01 -6.96771979e-01
-2.36310259e-01 1.36183178e+00 -5.88761985e-01 -1.09730698e-01
-4.16473210e-01 -5.21368504e-01 9.51571643e-01 1.19421393e-01
5.79618216e-01 -9.86680806e-01 -5.23117483e-01 2.59433776e-01
-1.03866482e+00 -1.03919959e+00 -5.41657567e-01 -2.53912956e-01
-2.01194167e-01 -1.05384684e+00 -3.66344035e-01 -2.16044679e-01
2.47166261e-01 -2.73718745e-01 1.49466121e+00 -6.32827431e-02
3.27017039e-01 7.01844692e-01 -4.42635119e-01 -2.41026744e-01
-8.93992066e-01 1.41372666e-01 8.08741152e-03 1.41991585e-01
3.77589077e-01 -3.96058887e-01 -5.76674223e-01 5.18798865e-02
-4.89739686e-01 2.81238526e-01 3.83846998e-01 1.12234926e+00
1.54432610e-01 -4.41141039e-01 8.55742276e-01 -1.23083556e+00
1.28013456e+00 -6.49499059e-01 -1.97409138e-01 5.38223803e-01
-6.42548740e-01 2.12993085e-01 7.03530431e-01 -2.84826994e-01
-1.75189543e+00 -1.52609900e-01 -8.71836916e-02 9.62816477e-02
7.13269273e-03 4.15845186e-01 -4.74719740e-02 2.94274002e-01
9.62549388e-01 4.01247710e-01 1.21516831e-01 -1.96408063e-01
9.06012297e-01 8.94545436e-01 5.31587958e-01 -1.10724342e+00
4.57862377e-01 2.62511432e-01 -6.38044000e-01 -5.19386530e-01
-1.09869850e+00 -1.12182081e-01 -4.85436231e-01 -3.48194152e-01
6.82245076e-01 -9.00708973e-01 -1.26184905e+00 3.03783476e-01
-1.48050392e+00 -5.80154479e-01 -1.58640400e-01 2.99195975e-01
-7.79762089e-01 5.03886700e-01 -8.77813518e-01 -1.25563073e+00
-7.13666141e-01 -7.01003075e-01 8.30343843e-01 2.45730475e-01
-8.83464694e-01 -1.17676473e+00 1.16773576e-01 7.43308544e-01
5.83464205e-01 2.08473817e-01 7.07718670e-01 -8.47664833e-01
-4.21729922e-01 -1.84699729e-01 5.17683700e-02 2.69295126e-01
-2.70759523e-01 1.84046999e-01 -1.12079811e+00 1.49039462e-01
7.20679834e-02 -1.12599671e+00 5.74205577e-01 -1.45873934e-01
5.38177609e-01 -1.03943431e+00 1.61187813e-01 1.09398942e-02
9.27169859e-01 -3.32635671e-01 4.94863898e-01 -1.05918497e-01
3.75300467e-01 1.02015150e+00 4.95230138e-01 8.47026229e-01
1.10937142e+00 6.15969956e-01 -8.15879256e-02 3.87212455e-01
2.25535944e-01 -6.41478598e-01 5.59747577e-01 8.04337323e-01
-8.29465389e-02 -2.09397733e-01 -5.73581994e-01 4.85701710e-01
-2.11879349e+00 -1.18578827e+00 4.96294796e-02 2.14845586e+00
1.82920253e+00 -8.50785598e-02 2.46514738e-01 -4.36892390e-01
4.85494792e-01 1.00815408e-01 -2.69887835e-01 -6.06445611e-01
-4.03128974e-02 4.05015536e-02 -4.02330868e-02 1.05443442e+00
-5.40992856e-01 1.16329575e+00 6.77013254e+00 8.24346542e-01
-5.82105815e-01 1.86805487e-01 5.23581326e-01 -2.43644103e-01
-7.36890078e-01 1.70281917e-01 -8.36303651e-01 3.92583609e-01
9.54137504e-01 -3.42384487e-01 6.05216742e-01 5.44879973e-01
2.79136479e-01 -2.81910598e-01 -1.24393404e+00 5.40958107e-01
-1.16067817e-02 -1.27245557e+00 9.65659618e-02 -3.79904002e-01
5.70302248e-01 -5.33059120e-01 -1.94821686e-01 6.00871682e-01
1.14825261e+00 -1.09837639e+00 7.77463913e-01 9.24583316e-01
5.05500793e-01 -5.51934600e-01 7.89053798e-01 7.43029833e-01
-3.90861124e-01 3.93819869e-01 -1.06896132e-01 -7.69448876e-02
3.35526377e-01 4.54662174e-01 -1.13129294e+00 4.59783375e-01
3.38659227e-01 8.90696049e-02 -3.29768471e-02 1.51507393e-01
-7.08371699e-01 7.08408654e-01 -2.44181052e-01 -3.58220249e-01
3.15120295e-02 -5.91346882e-02 5.25148749e-01 1.36873519e+00
-3.80533040e-01 2.65490055e-01 2.68899232e-01 1.32057309e+00
-2.34897375e-01 1.32436112e-01 -2.23766074e-01 -1.19562589e-01
9.87973988e-01 1.34255278e+00 1.41213462e-02 -5.49362183e-01
-4.59928103e-02 9.66338336e-01 6.51762009e-01 4.70463157e-01
-7.10983932e-01 -3.47471446e-01 3.90852630e-01 -3.00474465e-01
-4.75482307e-02 5.19258082e-02 -2.89939702e-01 -1.22182024e+00
-1.01580977e-01 -1.20622051e+00 3.30353945e-01 -5.66595256e-01
-1.49032748e+00 5.46321511e-01 1.91965476e-01 -6.39293551e-01
-9.55190778e-01 -1.60037115e-01 -6.05630100e-01 1.20887446e+00
-1.44555044e+00 -1.25111425e+00 -2.92664487e-02 7.36392140e-01
3.39404255e-01 3.98510210e-02 1.17708170e+00 -3.58686537e-01
-4.29072797e-01 8.43245089e-01 -2.79919535e-01 3.68262678e-02
9.94530678e-01 -1.55008841e+00 -2.59678140e-02 4.08432066e-01
-1.71353415e-01 1.00514412e+00 9.41694438e-01 -5.24997175e-01
-1.27070987e+00 -6.06689751e-01 1.20790529e+00 -1.02045512e+00
4.73050743e-01 -3.08823794e-01 -8.99621606e-01 4.86973852e-01
7.85473704e-01 -7.70813704e-01 1.13878083e+00 3.47487539e-01
-5.15736818e-01 2.86428839e-01 -1.18627477e+00 7.09998846e-01
5.50466597e-01 -6.62935257e-01 -7.95687079e-01 4.57705051e-01
7.49749422e-01 -6.19062066e-01 -1.08623719e+00 -4.29664627e-02
7.11358368e-01 -1.06623149e+00 6.29993141e-01 -6.00715756e-01
6.45586371e-01 -2.13936761e-01 -6.01093695e-02 -1.45672882e+00
-1.45382300e-01 -1.33359587e+00 -4.83261406e-01 1.35491514e+00
5.84148586e-01 -4.38039482e-01 5.55750251e-01 1.19095778e+00
2.69452006e-01 -6.06569469e-01 -5.24648488e-01 -2.34354794e-01
1.90097734e-01 -1.57449424e-01 4.90660250e-01 9.51965451e-01
4.83818293e-01 8.10943723e-01 -9.06034648e-01 -1.71766743e-01
7.04888940e-01 1.78437293e-01 1.02581918e+00 -8.30968320e-01
-6.54032469e-01 -4.95435089e-01 5.50024271e-01 -1.26669288e+00
3.81831884e-01 -5.43928266e-01 4.15266782e-01 -1.39088476e+00
1.69738039e-01 -7.81969652e-02 3.88417184e-01 4.32224035e-01
-5.41695833e-01 -1.94880396e-01 -2.25682445e-02 3.10396969e-01
-9.17808652e-01 8.79380643e-01 1.13852870e+00 1.37487650e-01
-2.80696243e-01 8.55333209e-02 -1.13290501e+00 6.25406563e-01
7.35416114e-01 -2.86437541e-01 -5.66322803e-01 -1.34738952e-01
4.87635970e-01 3.48491907e-01 4.10089433e-01 -8.32470804e-02
5.94912767e-01 -3.94095123e-01 3.37812155e-02 -2.62221366e-01
4.27740425e-01 -7.63782114e-02 -1.03042834e-02 -7.40256011e-02
-8.66893589e-01 -4.80776578e-01 -2.36127824e-01 6.69179201e-01
-2.19636843e-01 -2.61753559e-01 6.52563810e-01 -4.69651580e-01
-3.75334740e-01 -2.60983318e-01 -5.47055840e-01 4.26961392e-01
5.46735764e-01 -6.94186687e-02 -3.65724564e-01 -9.25279319e-01
-5.42720556e-01 5.46005845e-01 2.49599695e-01 2.46476769e-01
6.14835262e-01 -1.13818097e+00 -1.12255001e+00 -3.28053921e-01
1.08478405e-01 7.37527609e-02 3.09667259e-01 6.84200943e-01
-3.39907825e-01 2.03794271e-01 1.88793242e-01 -3.02477479e-01
-8.57221842e-01 3.66840810e-02 4.15108800e-01 -4.76009011e-01
-2.04244971e-01 1.18483210e+00 -1.71994075e-01 -8.64706516e-01
3.24728370e-01 4.25902158e-01 -3.02146792e-01 7.55022839e-02
5.86156964e-01 3.12053978e-01 -4.34260011e-01 -3.81577045e-01
-1.33227348e-01 -2.71192759e-01 -3.21511000e-01 -5.63600063e-01
8.79139423e-01 -3.49447966e-01 -3.02247316e-01 5.43928087e-01
6.44713461e-01 1.18450977e-01 -1.29046309e+00 -6.00371540e-01
-1.24511225e-02 -4.52395767e-01 -2.53287882e-01 -1.26102817e+00
-3.77176732e-01 6.45169377e-01 -3.91090214e-01 3.49324495e-01
5.33572257e-01 -2.55785257e-01 7.63596594e-01 6.96122169e-01
4.47014034e-01 -1.53157425e+00 3.61025542e-01 6.54754758e-01
1.06949317e+00 -1.34308290e+00 -9.47914124e-02 -2.38480031e-01
-1.48884046e+00 1.01836884e+00 8.15711498e-01 1.76511034e-01
2.27650195e-01 9.14058369e-03 3.37965190e-01 -5.94552942e-02
-1.44962084e+00 1.75341874e-01 1.50125280e-01 4.89813864e-01
7.35786259e-01 2.26491153e-01 -3.17580312e-01 9.52459931e-01
-5.14050603e-01 -1.62798792e-01 6.65637136e-01 6.14544749e-01
-3.32455337e-01 -1.08577836e+00 -1.82169154e-02 1.92734107e-01
-5.39418697e-01 -1.99703306e-01 -1.04325151e+00 4.30089980e-01
-4.52149481e-01 1.36001718e+00 -5.74172497e-01 -5.00636518e-01
2.44601011e-01 3.96311581e-01 3.89244407e-01 -5.52479744e-01
-1.02033019e+00 -2.78432041e-01 7.92155027e-01 -4.53793645e-01
-3.49647194e-01 -4.50519025e-01 -9.72219586e-01 -7.96416640e-01
-5.35234809e-01 5.11186361e-01 2.59267539e-01 9.38915074e-01
3.30658406e-01 -2.81165726e-02 6.96520627e-01 -5.27389288e-01
-1.43222320e+00 -1.13604510e+00 -3.26297551e-01 4.88997251e-01
5.31566963e-02 -2.61581182e-01 -2.86969006e-01 8.49475786e-02]
|
[12.591958045959473, 8.278257369995117]
|
db938a08-dacb-43fd-9396-0098510dd842
|
uncertainty-guided-multi-scale-residual-1
|
1906.11129
| null |
https://arxiv.org/abs/1906.11129v1
|
https://arxiv.org/pdf/1906.11129v1.pdf
|
Uncertainty Guided Multi-Scale Residual Learning-using a Cycle Spinning CNN for Single Image De-Raining
|
Single image de-raining is an extremely challenging problem since the rainy image may contain rain streaks which may vary in size, direction and density. Previous approaches have attempted to address this problem by leveraging some prior information to remove rain streaks from a single image. One of the major limitations of these approaches is that they do not consider the location information of rain drops in the image. The proposed Uncertainty guided Multi-scale Residual Learning (UMRL) network attempts to address this issue by learning the rain content at different scales and using them to estimate the final de-rained output. In addition, we introduce a technique which guides the network to learn the network weights based on the confidence measure about the estimate. Furthermore, we introduce a new training and testing procedure based on the notion of cycle spinning to improve the final de-raining performance. Extensive experiments on synthetic and real datasets to demonstrate that the proposed method achieves significant improvements over the recent state-of-the-art methods. Code is available at: https://github.com/rajeevyasarla/UMRL--using-Cycle-Spinning
|
['Rajeev Yasarla', 'Vishal M. Patel']
|
2019-06-12
|
uncertainty-guided-multi-scale-residual
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Yasarla_Uncertainty_Guided_Multi-Scale_Residual_Learning-Using_a_Cycle_Spinning_CNN_for_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Yasarla_Uncertainty_Guided_Multi-Scale_Residual_Learning-Using_a_Cycle_Spinning_CNN_for_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['single-image-deraining']
|
['computer-vision']
|
[-3.13819572e-02 -5.12979984e-01 1.75163493e-01 -5.97763479e-01
-6.22791886e-01 -2.71760672e-01 5.91322556e-02 -3.20712388e-01
-2.84458935e-01 9.84112442e-01 -1.81558594e-01 -1.61181390e-01
1.38921484e-01 -7.49709725e-01 -6.32362306e-01 -1.10242486e+00
-1.63263101e-02 -5.65772951e-02 4.14493740e-01 -1.06170572e-01
1.86725318e-01 5.01064956e-01 -1.31838441e+00 -1.43120900e-01
1.27006733e+00 6.82254255e-01 5.08787215e-01 8.33807886e-01
1.72401723e-02 9.46779847e-01 -5.55438817e-01 2.53470063e-01
3.60860825e-01 -6.09123886e-01 -2.36325245e-02 -9.31417290e-03
6.67294264e-01 -4.86010313e-01 6.34885347e-03 1.14046144e+00
4.64569360e-01 1.42645448e-01 4.58517611e-01 -7.55984604e-01
-3.56890976e-01 1.28049970e-01 -1.01732457e+00 6.42779887e-01
-1.51468009e-01 -1.03109062e-01 6.94194198e-01 -1.22688389e+00
2.47649834e-01 1.08955169e+00 7.03910887e-01 2.58750528e-01
-8.74106169e-01 -9.17808294e-01 2.95581222e-01 2.41731182e-01
-1.36905205e+00 -3.37373525e-01 5.13818741e-01 -3.29544991e-01
3.45459521e-01 1.86520830e-01 4.85888124e-01 4.55245942e-01
2.82517374e-01 6.12209082e-01 1.61331081e+00 -5.19994080e-01
2.30608612e-01 4.89498079e-02 3.96648407e-01 8.55823398e-01
6.76888406e-01 9.17649418e-02 -1.27962157e-01 2.63311733e-02
8.02915752e-01 2.29972214e-01 -7.33565092e-01 -8.48076269e-02
-6.47420704e-01 9.29248929e-01 6.12401247e-01 1.98670089e-01
-4.18381453e-01 -2.93874685e-02 -2.42176518e-01 2.37853825e-01
9.04313326e-01 1.13059901e-01 -2.68560231e-01 4.36478406e-01
-1.43387365e+00 3.14448297e-01 7.41875708e-01 3.46846879e-01
1.14743161e+00 3.80656183e-01 -1.26961589e-01 6.58416986e-01
4.37519848e-01 1.17423332e+00 -8.90429784e-03 -8.66228759e-01
4.21053529e-01 -1.81900132e-02 4.85876739e-01 -8.72114778e-01
-2.62596250e-01 -3.40475112e-01 -9.51796412e-01 8.08195055e-01
2.14114249e-01 -5.24749517e-01 -1.48447347e+00 1.36999762e+00
4.69124109e-01 7.02032506e-01 1.01454876e-01 1.23256969e+00
6.90163732e-01 9.78597820e-01 -2.01488972e-01 -4.93044764e-01
8.42899501e-01 -1.04928553e+00 -9.17956948e-01 -3.37438375e-01
-2.75364295e-02 -8.47047627e-01 7.61845708e-01 3.70175987e-01
-7.71555662e-01 -5.91818869e-01 -1.24747622e+00 4.90703970e-01
-1.22128718e-01 3.11773598e-01 4.23736304e-01 5.59580028e-01
-8.61855447e-01 6.37598395e-01 -8.70047271e-01 -1.94428295e-01
1.73623547e-01 -4.77083288e-02 1.85150146e-01 -3.95910680e-01
-1.10828984e+00 9.28368092e-01 2.69598484e-01 6.84982717e-01
-8.68650675e-01 -4.54028815e-01 -6.78432941e-01 -1.93533048e-01
2.14722246e-01 -4.30156052e-01 9.34574664e-01 -1.08869350e+00
-1.43041372e+00 2.05347851e-01 -4.36686218e-01 -4.01316553e-01
2.92127311e-01 -7.58083463e-01 -4.67276871e-01 2.52145410e-01
-3.08478735e-02 2.35506967e-01 1.36614358e+00 -1.61049199e+00
-9.76463735e-01 -1.87601820e-01 -5.56635223e-02 4.62573946e-01
2.31528565e-01 -2.53373504e-01 -3.62076193e-01 -8.07351470e-01
1.22534193e-01 -1.03032184e+00 -3.69196355e-01 -1.15783639e-01
-1.32806107e-01 3.22597742e-01 9.91114497e-01 -7.40381420e-01
1.23571599e+00 -2.04285789e+00 2.29885187e-02 7.98287243e-02
2.52809078e-01 7.18001008e-01 -4.45984527e-02 2.51411796e-01
1.79693848e-01 -1.25470728e-01 -6.29928827e-01 -3.29291970e-01
-5.72975814e-01 3.37126970e-01 -4.83260244e-01 7.30890810e-01
1.71792462e-01 2.40540400e-01 -7.48219967e-01 -3.76015484e-01
4.52539116e-01 6.36152625e-01 -3.05582639e-02 5.82724988e-01
-1.24947943e-01 4.76528555e-01 -3.14245850e-01 7.03030169e-01
1.39706290e+00 -4.91556339e-02 -6.68295398e-02 -1.30203694e-01
-2.89465070e-01 -2.01329425e-01 -1.46324360e+00 8.59869123e-01
-5.31245172e-01 5.40493488e-01 2.92378157e-01 -6.57262385e-01
7.68057942e-01 2.67365247e-01 -3.53295654e-02 -3.77376765e-01
-2.85082385e-02 2.56282538e-01 -2.73315787e-01 -6.26729906e-01
4.21267271e-01 -3.54796797e-01 5.24415493e-01 3.58653307e-01
-2.41366908e-01 -2.21742958e-01 1.26759365e-01 1.36795729e-01
7.51763642e-01 2.27848485e-01 1.96789905e-01 1.50389709e-02
4.75756973e-01 -2.63158530e-01 1.00460815e+00 9.62369680e-01
-2.78980047e-01 9.65985060e-01 -1.24017581e-01 -4.36763763e-01
-6.33544683e-01 -1.17525792e+00 -1.03079021e-01 8.23689044e-01
3.96409005e-01 1.52032092e-01 -5.66863477e-01 -6.10429823e-01
-2.84297504e-02 7.01931536e-01 -6.54482722e-01 3.93586844e-01
-6.24492526e-01 -1.27487969e+00 2.67495830e-02 2.55070388e-01
6.49019659e-01 -1.02921903e+00 -5.58120906e-01 7.26023167e-02
-3.02143157e-01 -1.13190889e+00 -1.29104510e-01 1.81295171e-01
-1.01053393e+00 -9.91934299e-01 -7.77920902e-01 -5.15367746e-01
8.31650913e-01 8.70316863e-01 1.03629208e+00 2.10611209e-01
-3.20603669e-01 3.56662273e-02 -6.52221441e-01 -5.57658494e-01
1.89121310e-02 -2.48641282e-01 -1.56408474e-01 4.78028227e-03
9.76478830e-02 -6.12733662e-01 -8.76028240e-01 -8.56163353e-03
-9.81974781e-01 -1.18532553e-01 7.46597886e-01 6.67777419e-01
6.28708124e-01 1.73583224e-01 5.21035492e-01 -1.13524628e+00
4.88632590e-01 -5.72883785e-01 -8.27432573e-01 2.48674050e-01
-6.62036240e-01 -2.61902735e-02 3.52717161e-01 -8.95815417e-02
-1.36521685e+00 8.96333829e-02 6.35766760e-02 -5.82705855e-01
-5.43932915e-02 4.12119269e-01 3.26940864e-01 -2.41608486e-01
4.11807030e-01 6.28803670e-02 -2.70526975e-01 -3.80364776e-01
3.69593829e-01 4.39989746e-01 2.70115197e-01 -1.09950500e-02
1.23757720e+00 7.08831966e-01 -1.39192581e-01 -9.07832265e-01
-1.44918334e+00 -7.09201217e-01 -4.67851579e-01 -3.04461122e-01
7.52938032e-01 -1.32838738e+00 -8.16370621e-02 6.95855379e-01
-9.03638303e-01 -4.89341706e-01 8.14624801e-02 5.95577180e-01
1.00718429e-02 4.43710119e-01 -6.25418723e-01 -1.21380448e+00
-6.72195315e-01 -7.39839196e-01 6.31945193e-01 6.65374339e-01
4.36782628e-01 -9.41780508e-01 3.59128773e-01 1.10589258e-01
6.54241920e-01 3.08386147e-01 4.18025315e-01 1.29942670e-01
-6.51957095e-01 4.52709422e-02 -4.06248152e-01 5.96025825e-01
4.56477791e-01 1.85484529e-01 -8.38762999e-01 -5.11633337e-01
3.02226692e-01 -2.10069522e-01 1.31690109e+00 7.57923543e-01
5.48953593e-01 -1.25900656e-01 -5.76542988e-02 6.71642423e-01
2.02294755e+00 -1.11007921e-01 6.19927883e-01 3.63392740e-01
7.09169209e-01 1.57429278e-01 1.02879190e+00 5.32882452e-01
2.52920806e-01 2.30441123e-01 6.52227938e-01 -3.74425888e-01
-2.22686246e-01 3.74063373e-01 3.67088318e-01 8.17992747e-01
-3.98098856e-01 -4.08301979e-01 -3.92150998e-01 7.14443028e-01
-2.04663301e+00 -9.81349587e-01 -1.75408393e-01 2.12516665e+00
7.45716214e-01 1.39438629e-01 -3.32525134e-01 -2.39455029e-01
6.90987647e-01 6.23865902e-01 -4.88295048e-01 -1.47398099e-01
-1.02763355e-01 4.36846316e-01 8.41187775e-01 1.02930546e+00
-1.25688124e+00 1.03385973e+00 5.83272505e+00 4.31872666e-01
-1.29216182e+00 1.55858889e-01 3.31201226e-01 -2.53739934e-02
-2.44099163e-02 3.78656834e-02 -1.07883573e+00 5.27237356e-01
6.71813548e-01 3.44739228e-01 4.70488101e-01 3.98358911e-01
7.06355929e-01 -6.50605142e-01 -1.07758954e-01 5.91838241e-01
2.02659771e-01 -9.70089376e-01 -1.81747213e-01 -5.36064863e-01
8.97067130e-01 4.50747788e-01 -1.69042394e-01 1.88797489e-01
5.70869625e-01 -8.81090403e-01 4.31818545e-01 9.97430146e-01
5.14789224e-01 -6.54392004e-01 1.00941288e+00 1.43482402e-01
-1.18363845e+00 7.89752752e-02 -5.39513111e-01 -1.46375269e-01
1.85896575e-01 1.27490580e+00 -6.93965554e-01 6.35324419e-01
9.91475463e-01 6.01173162e-01 -4.67270851e-01 1.20800149e+00
-8.37374628e-01 1.05439365e+00 -4.06569302e-01 3.84773344e-01
1.31077230e-01 -5.25705218e-01 5.22347450e-01 1.41572404e+00
5.03919601e-01 2.61547208e-01 2.23476261e-01 4.76080298e-01
1.09793141e-01 -2.63753355e-01 -3.60412389e-01 3.77835959e-01
3.97214562e-01 1.36081219e+00 -6.53980613e-01 -3.48211497e-01
-4.69911128e-01 9.56818044e-01 2.04281479e-01 7.24208832e-01
-9.31577146e-01 -5.78637362e-01 6.31716490e-01 9.95505485e-04
8.51617992e-01 -3.40658426e-01 -8.10763612e-02 -1.19316697e+00
2.15745419e-02 -7.54863620e-01 -6.21584430e-03 -8.22515666e-01
-1.11472273e+00 6.99964464e-01 -1.30173162e-01 -1.22266006e+00
-3.89465271e-03 -2.55253732e-01 -8.24832022e-01 1.07537568e+00
-2.21072483e+00 -1.08008432e+00 -8.54265451e-01 4.44677711e-01
7.14445710e-01 1.58230156e-01 5.12801051e-01 2.11051181e-01
-5.55908442e-01 8.71366356e-03 5.54802656e-01 -1.42089799e-01
1.09883320e+00 -1.32164979e+00 7.81759992e-02 1.44580162e+00
3.94952707e-02 2.06243560e-01 1.10091305e+00 -6.98856592e-01
-9.09345448e-01 -1.31867242e+00 5.02887607e-01 -3.03391814e-02
4.00631636e-01 -5.78178056e-02 -1.14853168e+00 6.68225646e-01
3.38578999e-01 4.72353965e-01 2.66213238e-01 -4.36701637e-04
-2.00013295e-01 -4.01824981e-01 -1.06797290e+00 2.90529311e-01
2.97021449e-01 -8.65821764e-02 -5.04286408e-01 2.61509299e-01
5.70336878e-01 -4.89489734e-01 -3.83595258e-01 7.18805671e-01
4.01828766e-01 -1.30935240e+00 6.17298782e-01 1.84449017e-01
3.29280198e-01 -8.27512801e-01 -1.86031759e-01 -1.59781313e+00
-3.80678475e-01 -2.21545845e-01 -3.33189726e-01 1.02108097e+00
2.44327739e-01 -6.74102724e-01 5.83210886e-01 -1.07307449e-01
1.51237682e-01 -6.61454797e-01 -5.62557101e-01 -4.72807318e-01
-2.24731833e-01 8.12904984e-02 6.41321912e-02 7.43525088e-01
-8.09376657e-01 1.95660412e-01 -8.08102429e-01 9.75203693e-01
1.02028656e+00 5.87441981e-01 6.33401036e-01 -1.10820866e+00
-1.32609934e-01 3.20235074e-01 1.28922492e-01 -8.18609416e-01
-1.76587954e-01 -1.96930155e-01 7.17403054e-01 -1.71864402e+00
1.88729376e-01 -4.23202723e-01 -5.96963704e-01 3.31158936e-01
-7.26371646e-01 4.06448573e-01 3.51170033e-01 5.20068824e-01
-6.24656737e-01 5.70104539e-01 1.16843891e+00 1.57712266e-01
-3.36989850e-01 3.50297987e-01 -2.86899179e-01 9.34543848e-01
1.20439577e+00 -8.20334613e-01 -3.46686602e-01 -6.90808773e-01
-4.33310792e-02 7.57177919e-02 1.65069312e-01 -1.30205262e+00
7.01721236e-02 -3.67907137e-01 4.13301349e-01 -7.62615979e-01
3.65562707e-01 -6.55599594e-01 -8.31357688e-02 3.52575362e-01
1.41861945e-01 -6.60899803e-02 7.47448057e-02 7.92892218e-01
-2.89842546e-01 -3.61655563e-01 1.23312914e+00 -2.15850636e-01
-6.67481184e-01 2.40038022e-01 -5.10285556e-01 -1.97604612e-01
7.95669794e-01 6.02504127e-02 -3.43827069e-01 -4.92799312e-01
-6.11695528e-01 2.04685047e-01 3.34064990e-01 1.08199053e-01
8.11899364e-01 -7.57204652e-01 -9.47596252e-01 5.24690486e-02
-2.29295105e-01 -5.01939692e-02 2.72348195e-01 6.92281187e-01
-7.75922716e-01 2.06541456e-02 -8.18246379e-02 -3.49608928e-01
-1.46587765e+00 8.17293003e-02 4.47997481e-01 -4.08222079e-01
-7.30485857e-01 7.22750723e-01 2.21781269e-01 -2.48879775e-01
7.05596954e-02 -3.16479385e-01 -2.74771661e-01 -8.42849165e-02
6.23600841e-01 2.01036230e-01 -5.00881411e-02 -3.58665764e-01
-1.13876358e-01 6.59108520e-01 -2.17249393e-01 -4.23686169e-02
1.51579869e+00 -4.41605270e-01 -1.86363786e-01 6.07281148e-01
6.47514224e-01 2.48120442e-01 -1.54826176e+00 -3.43494684e-01
-3.79564166e-01 -6.33298159e-01 3.27171981e-01 -8.15865457e-01
-1.44194317e+00 7.30293632e-01 1.14159620e+00 -1.18582398e-01
1.29820943e+00 -5.94977856e-01 7.27297187e-01 4.27712739e-01
1.23536780e-01 -8.88535440e-01 1.18040815e-02 7.00662374e-01
6.72629774e-01 -1.57602417e+00 5.44717312e-01 -3.15129310e-01
-5.87524891e-01 1.08189321e+00 5.26767313e-01 -5.59518933e-01
1.01735473e+00 3.88526469e-01 6.47587717e-01 -5.72773293e-02
-4.38730568e-01 -4.51769590e-01 -1.93821386e-01 3.51028800e-01
3.91828805e-01 1.20044447e-01 -4.02995795e-01 -8.78251567e-02
4.35939878e-01 2.67834544e-01 8.23007941e-01 1.13418114e+00
-1.05119574e+00 -8.16617131e-01 -8.04073870e-01 4.01864350e-01
-6.19885981e-01 -3.49036306e-01 2.55571961e-01 5.61324239e-01
1.85187638e-01 1.06658578e+00 -2.76065886e-01 -3.84057648e-02
4.78004925e-02 -3.51159066e-01 3.99332911e-01 -5.93596399e-01
-1.48437396e-01 1.95507973e-01 -2.29439005e-01 -2.60913402e-01
-9.77460682e-01 -3.99619997e-01 -1.11984658e+00 -1.49834573e-01
-5.68111837e-01 3.91603470e-01 5.42192876e-01 8.74826014e-01
1.65434539e-01 4.63807404e-01 9.03343379e-01 -1.12311387e+00
-2.22299576e-01 -1.08237541e+00 -9.21388566e-01 -4.98652384e-02
9.02832925e-01 -5.90852141e-01 -7.85402238e-01 1.32899493e-01]
|
[10.932170867919922, -3.27445125579834]
|
2109c26c-391a-4310-99f9-86e9ada2e18a
|
generating-equation-by-utilizing-operators
| null | null |
https://aclanthology.org/2020.coling-main.38
|
https://aclanthology.org/2020.coling-main.38.pdf
|
Generating Equation by Utilizing Operators : GEO model
|
Math word problem solving is an emerging research topic in Natural Language Processing. Recently, to address the math word problem-solving task, researchers have applied the encoder-decoder architecture, which is mainly used in machine translation tasks. The state-of-the-art neural models use hand-crafted features and are based on generation methods. In this paper, we propose the GEO (Generation of Equations by utilizing Operators) model that does not use hand-crafted features and addresses two issues that are present in existing neural models: 1. missing domain-specific knowledge features and 2. losing encoder-level knowledge. To address missing domain-specific feature issue, we designed two auxiliary tasks: operation group difference prediction and implicit pair prediction. To address losing encoder-level knowledge issue, we added an Operation Feature Feed Forward (OP3F) layer. Experimental results showed that the GEO model outperformed existing state-of-the-art models on two datasets, 85.1{\%} in MAWPS, and 62.5{\%} in DRAW-1K, and reached comparable performance of 82.1{\%} in ALG514 dataset.
|
['Gahgene Gweon', 'Bugeun Kim', 'Donggeon Lee', 'Kyung Seo Ki']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving']
|
['knowledge-base', 'reasoning', 'time-series']
|
[ 2.61647493e-01 -8.49740133e-02 -4.70036939e-02 -5.20279050e-01
-5.88105083e-01 -1.86983302e-01 2.19237119e-01 1.26577571e-01
-4.83227015e-01 9.27699745e-01 -1.85733940e-02 -5.38575232e-01
-1.00938581e-01 -1.17768347e+00 -1.03157389e+00 -2.34514475e-01
3.65425587e-01 3.73911262e-01 3.50769199e-02 -5.31106651e-01
5.55321872e-01 -1.15874365e-01 -1.34079695e+00 7.55470574e-01
1.32588577e+00 1.01388669e+00 4.02414501e-01 3.34130585e-01
-7.22268999e-01 9.83784378e-01 -5.75993598e-01 -8.44734788e-01
2.88849831e-01 -5.67375779e-01 -8.52072477e-01 -5.42632461e-01
2.89802641e-01 -4.39704150e-01 -3.77003461e-01 9.84180748e-01
4.61504757e-01 1.40252132e-02 5.94960690e-01 -1.22792006e+00
-1.39723587e+00 9.56335008e-01 -4.82805192e-01 2.01818332e-01
3.99267733e-01 4.34239358e-02 9.91883516e-01 -1.02936375e+00
3.00139427e-01 1.17518771e+00 5.17219543e-01 5.54896057e-01
-8.35402608e-01 -9.89190221e-01 2.08057277e-02 6.70747042e-01
-1.62432730e+00 -2.38021657e-01 6.83846056e-01 -3.43236178e-01
1.36306143e+00 8.62643495e-02 2.68481165e-01 8.38677227e-01
3.41582656e-01 7.90584803e-01 1.10155940e+00 -5.31553388e-01
-7.96824694e-03 1.25118554e-01 4.45651412e-01 9.85093951e-01
9.28481892e-02 -3.22026387e-02 -6.91370547e-01 1.59822598e-01
9.34119225e-01 -2.26329759e-01 -2.82467574e-01 2.57005274e-01
-1.25230980e+00 9.37510490e-01 3.12608331e-01 1.97307095e-01
-1.59534752e-01 7.21512958e-02 1.86336279e-01 5.48545659e-01
1.95685521e-01 7.11283267e-01 -7.83774257e-01 -3.49020243e-01
-7.83684850e-01 3.93635571e-01 7.42830038e-01 1.32297659e+00
8.71829927e-01 1.27255335e-01 -4.88444716e-01 8.68546546e-01
6.54461384e-02 2.74934769e-01 8.10989738e-01 -5.07151544e-01
1.13278556e+00 9.69849765e-01 -3.40847850e-01 -9.32738602e-01
-2.47407794e-01 -4.63460177e-01 -9.80545878e-01 -4.78737772e-01
2.77328610e-01 -1.76321343e-01 -9.58093047e-01 1.80426180e+00
-1.25112101e-01 2.31790826e-01 1.35123238e-01 8.85130823e-01
1.16471851e+00 1.01913869e+00 -2.51281619e-01 -1.89143926e-01
1.09558451e+00 -1.27757096e+00 -7.72740960e-01 -6.24199063e-02
8.31549942e-01 -9.12353754e-01 1.05530751e+00 4.05896425e-01
-1.33418429e+00 -8.95525694e-01 -1.00775409e+00 -4.41537857e-01
-5.69499612e-01 3.33582342e-01 6.69088185e-01 5.31496763e-01
-9.40109670e-01 7.36483157e-01 -2.45917618e-01 -8.12854618e-02
1.97579622e-01 5.33906519e-01 -2.91559905e-01 -2.31459931e-01
-1.62919915e+00 1.07130003e+00 6.30099118e-01 1.62622213e-01
-3.10819268e-01 -9.61376011e-01 -9.11090732e-01 2.32844427e-01
4.37778741e-01 -7.10358679e-01 1.22927511e+00 -6.49859846e-01
-1.75993335e+00 3.56385499e-01 -1.96451455e-01 -3.80324095e-01
1.31474242e-01 -3.99312347e-01 -6.37557149e-01 -1.78602695e-01
-1.20013513e-01 5.05573928e-01 6.00889087e-01 -8.42850983e-01
-4.89059180e-01 -1.47631854e-01 2.79060360e-02 2.09124446e-01
-4.33447838e-01 -8.95658210e-02 -4.00791883e-01 -8.94741535e-01
8.17788318e-02 -6.97425365e-01 1.15247369e-01 -2.03754976e-01
-2.16395780e-01 -4.69044417e-01 2.87552506e-01 -1.05750620e+00
1.59057844e+00 -1.74773347e+00 3.23776960e-01 5.77740222e-02
-5.77570722e-02 7.00304747e-01 -1.89622581e-01 4.89292353e-01
-1.38452098e-01 6.32443950e-02 -1.50128737e-01 -5.96546791e-02
7.57120773e-02 6.17943518e-02 -3.51031125e-01 -1.15900286e-01
6.28099680e-01 1.00823212e+00 -6.52046561e-01 -5.31226099e-01
1.43934429e-01 1.39481887e-01 -7.88740337e-01 3.58478189e-01
-3.74942929e-01 -8.42248946e-02 -4.61511523e-01 6.21554077e-01
8.99035156e-01 -8.74114111e-02 -7.32798800e-02 -1.38425112e-01
-1.72712833e-01 4.49060231e-01 -1.33392906e+00 2.14116788e+00
-5.78878462e-01 3.82678747e-01 -5.49489319e-01 -9.82291520e-01
1.24295807e+00 2.06442922e-01 -1.61095802e-02 -9.11590338e-01
3.22829813e-01 3.40801924e-01 2.94316828e-01 -6.37216508e-01
6.66057110e-01 8.45737830e-02 2.18325891e-02 2.49438971e-01
3.43600243e-01 -5.13843782e-02 1.52059719e-01 3.02335173e-02
1.20597780e+00 1.78303033e-01 1.15437075e-01 -2.83181101e-01
8.40312600e-01 -9.11147147e-02 6.16846263e-01 5.73286712e-01
1.55655801e-01 7.87198901e-01 2.85077751e-01 -5.53854585e-01
-6.28781736e-01 -8.03133845e-01 6.40880615e-02 9.72894192e-01
1.40683606e-01 -8.26558709e-01 -7.64738977e-01 -4.55767393e-01
-1.37300789e-01 9.77224708e-01 -4.51032698e-01 -4.69342202e-01
-8.45591664e-01 -7.57167935e-01 6.69806242e-01 6.01445436e-01
9.14312840e-01 -9.78179693e-01 -9.87017006e-02 5.06662488e-01
-3.89810175e-01 -1.16544199e+00 -2.41864353e-01 -8.58190749e-03
-7.11321950e-01 -7.47701466e-01 -7.42172658e-01 -9.93903756e-01
6.44630730e-01 -6.93623647e-02 1.10340118e+00 1.40864253e-01
-1.67378828e-01 -4.15089816e-01 -6.33054316e-01 -5.05154192e-01
-5.09039275e-02 3.84583920e-01 6.54228451e-03 -4.04491760e-02
5.72958231e-01 -5.35987556e-01 -1.19036317e-01 1.39198303e-01
-5.65641403e-01 5.70728660e-01 9.30709124e-01 8.82466435e-01
3.09237301e-01 -6.81989118e-02 6.12125278e-01 -7.26388395e-01
7.58880377e-01 -3.28570575e-01 -4.32885408e-01 6.02785230e-01
-5.30988574e-01 4.53540087e-01 9.94714916e-01 -3.57487381e-01
-1.00390399e+00 -1.86199263e-01 -1.79489136e-01 -4.08970416e-01
4.91847023e-02 6.82231426e-01 -2.84114271e-01 1.32716209e-01
3.56491148e-01 5.85747898e-01 -2.12990254e-01 -7.33874202e-01
1.68396398e-01 7.73050249e-01 4.39086497e-01 -1.00343537e+00
7.64952898e-01 -4.48335141e-01 -1.62085205e-01 -4.02534693e-01
-7.39388168e-01 6.07925393e-02 -2.99663275e-01 2.45458812e-01
8.15552711e-01 -9.96312857e-01 -6.99437499e-01 6.47440195e-01
-1.60410500e+00 -5.85019216e-02 4.87615876e-02 6.38338506e-01
-4.14132595e-01 8.24581534e-02 -6.41776502e-01 -5.67349613e-01
-6.21287048e-01 -1.30850589e+00 7.37262011e-01 4.63255316e-01
-9.52354968e-02 -4.80912328e-01 -3.78268152e-01 5.45638323e-01
6.59374774e-01 -1.56713620e-01 1.54253685e+00 -6.45395100e-01
-6.01335108e-01 9.07662585e-02 -5.15297234e-01 3.62019271e-01
6.70518121e-03 -2.81510055e-01 -6.52647614e-01 1.10568151e-01
-5.73794767e-02 -3.59208107e-01 7.52729595e-01 -1.28093421e-01
1.63611329e+00 -2.27184787e-01 1.07497480e-02 5.54660439e-01
1.21757400e+00 2.31160566e-01 8.15400362e-01 4.82573770e-02
6.51788831e-01 3.28173757e-01 3.54761213e-01 2.87064493e-01
7.90241182e-01 7.12372184e-01 1.22894317e-01 1.66409984e-01
-1.67055994e-01 -4.84833151e-01 2.85067767e-01 1.29416084e+00
-2.74259955e-01 -2.21302181e-01 -9.49104667e-01 4.49435711e-01
-1.96675670e+00 -5.41912913e-01 -2.77490169e-01 1.73615217e+00
1.33677483e+00 2.59375364e-01 -4.71681654e-01 2.68090665e-01
6.46072805e-01 -2.77377784e-01 -2.20922977e-01 -7.79972494e-01
6.66516870e-02 8.49760175e-01 3.66581559e-01 3.10641617e-01
-8.30436885e-01 1.15799820e+00 4.71905136e+00 1.12486374e+00
-9.46596026e-01 -1.37496561e-01 4.47874308e-01 6.29447550e-02
-9.45235565e-02 -1.43019453e-01 -1.00025380e+00 6.10147536e-01
8.01479220e-01 -2.32947797e-01 5.41490674e-01 4.87726301e-01
-1.06622197e-01 2.14784686e-02 -1.44956815e+00 1.21418333e+00
2.42374092e-01 -1.38506055e+00 3.31102222e-01 -2.21701235e-01
6.51146889e-01 -5.03299236e-01 -8.67162496e-02 8.94703031e-01
6.34909719e-02 -1.30694878e+00 4.85392869e-01 6.70517147e-01
7.48707533e-01 -7.61211216e-01 9.48735535e-01 5.06083846e-01
-1.13103068e+00 1.38019938e-02 -5.82852542e-01 -6.53352797e-01
8.49270150e-02 6.07999563e-01 -4.52154696e-01 9.11371350e-01
5.90786695e-01 6.19953752e-01 -6.44645810e-01 7.46751606e-01
-3.83417666e-01 4.75467265e-01 -2.17428759e-01 -1.87871337e-01
2.74269015e-01 -1.04168586e-01 -8.37995782e-02 9.77031112e-01
5.98209262e-01 3.98819149e-01 -1.10147372e-01 1.05091023e+00
-2.80299753e-01 2.05578119e-01 -1.59181789e-01 -1.11291699e-01
3.51649135e-01 9.75340486e-01 -1.73427343e-01 -2.88727582e-01
-4.99278426e-01 1.09905934e+00 5.69051087e-01 8.60950649e-02
-1.13032842e+00 -7.11800158e-01 5.11498272e-01 -1.65230166e-02
3.25394005e-01 -1.38853654e-01 -4.40357506e-01 -1.36077905e+00
2.36896381e-01 -1.00350463e+00 1.42170697e-01 -7.66985238e-01
-1.25718474e+00 4.45205986e-01 2.98863370e-02 -1.03852141e+00
-1.40690967e-01 -8.04051220e-01 -6.45147741e-01 1.19311166e+00
-1.61562157e+00 -1.24339032e+00 -1.26368001e-01 5.87702692e-01
7.41696000e-01 -5.56623578e-01 1.03706491e+00 5.95682442e-01
-7.62645006e-01 1.06644773e+00 -1.16753869e-01 4.01631564e-01
7.79960871e-01 -8.93210292e-01 3.98432016e-01 7.07417190e-01
-3.50493789e-02 8.52813125e-01 3.76172423e-01 -4.46802616e-01
-1.92480493e+00 -1.06694031e+00 1.54855764e+00 -1.21261887e-01
3.78182322e-01 -3.40619206e-01 -8.78992260e-01 5.63150346e-01
2.61733979e-01 -8.34717527e-02 7.68928707e-01 -4.63409200e-02
-2.01816142e-01 -1.22735158e-01 -1.09473157e+00 5.84926844e-01
1.11058807e+00 -7.22871870e-02 -9.60833073e-01 7.66311809e-02
9.88508105e-01 -7.45774090e-01 -8.13500583e-01 5.04184723e-01
3.60164255e-01 -4.74632233e-01 7.92127550e-01 -7.93909788e-01
1.12284398e+00 -2.43120119e-01 -2.95218855e-01 -1.40363240e+00
-4.84633029e-01 -4.21920478e-01 -4.52194303e-01 1.35310924e+00
6.21821284e-01 -4.26067203e-01 5.96737325e-01 7.42636442e-01
-2.82453686e-01 -1.16332448e+00 -7.49744475e-01 -7.50469387e-01
3.61097604e-01 -3.21956366e-01 9.89375412e-01 1.07209754e+00
7.42256716e-02 6.52046561e-01 -4.00378078e-01 -1.03781864e-01
1.31621743e-02 2.48593345e-01 6.19782627e-01 -8.88242722e-01
-4.85071421e-01 -2.52950758e-01 -2.07438231e-01 -1.10567641e+00
2.12346226e-01 -1.01076806e+00 -3.46447229e-01 -1.54557133e+00
1.17442040e-02 -3.13850343e-01 -4.00980175e-01 7.18327105e-01
-2.92738080e-01 -4.67249006e-02 3.88938844e-01 -2.64477849e-01
-2.63330966e-01 7.60282516e-01 1.25355935e+00 -5.52855320e-02
-8.82966891e-02 -3.48640800e-01 -8.98684442e-01 4.84381169e-01
1.06394434e+00 -1.91292673e-01 -4.20267969e-01 -1.25928855e+00
3.52552474e-01 4.10616882e-02 4.07791212e-02 -1.13579679e+00
4.22529370e-01 -4.05501425e-01 4.06180024e-01 -6.30916834e-01
5.05205989e-01 -6.81303918e-01 -1.50019541e-01 4.35746670e-01
-4.24380064e-01 5.04107535e-01 2.72448927e-01 9.87515301e-02
-1.83982149e-01 -3.43495578e-01 3.23657811e-01 -2.80048519e-01
-8.39302003e-01 7.27674738e-02 -1.40052706e-01 1.35262817e-01
1.01816130e+00 2.42744870e-02 -5.38383424e-01 -5.99383339e-02
5.88508043e-03 3.39584649e-01 -1.81501642e-01 7.04310179e-01
8.03846836e-01 -1.48931146e+00 -9.38553751e-01 4.31297898e-01
1.75799564e-01 1.88390180e-01 7.86494315e-02 5.20939350e-01
-6.91551983e-01 6.17058158e-01 -2.78161108e-01 1.00548174e-02
-9.17290986e-01 3.84423822e-01 4.85465489e-02 -5.02074003e-01
-2.18513995e-01 1.14093709e+00 -1.44700423e-01 -5.92401922e-01
1.71784610e-01 -6.11222267e-01 -1.12123765e-01 -2.24065199e-01
5.40970445e-01 3.26922596e-01 3.35390985e-01 -4.10631001e-01
-3.82619768e-01 4.73927855e-01 -4.11042869e-01 2.07496867e-01
1.35480070e+00 3.65345716e-01 -2.13150963e-01 1.80601358e-01
1.17436314e+00 -5.33515155e-01 -4.49453771e-01 -5.43592632e-01
-2.37118289e-01 -3.01179796e-01 1.88617036e-02 -1.15772438e+00
-1.14474499e+00 1.22620213e+00 3.00147384e-01 -4.74336267e-01
1.05648601e+00 -6.08128667e-01 1.23117197e+00 6.45666003e-01
4.31781203e-01 -1.26989901e+00 9.41456668e-03 1.08167779e+00
8.02794874e-01 -1.10193968e+00 3.74521874e-03 -7.65864134e-01
-4.22443777e-01 1.27251256e+00 1.23292053e+00 3.02484483e-02
5.26940227e-01 1.46703169e-01 -2.81355977e-01 2.30682552e-01
-8.62563550e-01 3.56790796e-02 3.83219063e-01 2.16850221e-01
6.81392193e-01 3.39172482e-02 -8.33033085e-01 1.56824028e+00
-6.62634194e-01 5.13139963e-01 3.55346352e-01 1.15405452e+00
-2.62719542e-01 -1.29332185e+00 -1.93893656e-01 7.68308401e-01
-2.87121892e-01 -6.44391894e-01 -4.24494237e-01 5.33497632e-01
6.50725067e-01 8.24897289e-01 3.00178993e-02 -6.50667131e-01
3.05534482e-01 3.16849381e-01 6.62685394e-01 -5.35898685e-01
-7.86043763e-01 -6.78580821e-01 6.08835816e-02 -2.92276323e-01
8.80489275e-02 -1.28333464e-01 -1.24013710e+00 -4.84982193e-01
-4.50466365e-01 2.35792864e-02 4.76492018e-01 1.12569833e+00
4.51244831e-01 7.67170489e-01 2.70539314e-01 -3.43389213e-01
-6.97272778e-01 -1.06644106e+00 -1.37970254e-01 1.32005230e-01
-1.74782217e-01 -4.97867972e-01 1.85376123e-01 2.63673123e-02]
|
[9.805419921875, 7.497879981994629]
|
67f5e0ca-5d9a-4537-aa21-674d243386ea
|
convolutional-monge-mapping-normalization-for
|
2305.18831
| null |
https://arxiv.org/abs/2305.18831v2
|
https://arxiv.org/pdf/2305.18831v2.pdf
|
Convolutional Monge Mapping Normalization for learning on biosignals
|
In many machine learning applications on signals and biomedical data, especially electroencephalogram (EEG), one major challenge is the variability of the data across subjects, sessions, and hardware devices. In this work, we propose a new method called Convolutional Monge Mapping Normalization (CMMN), which consists in filtering the signals in order to adapt their power spectrum density (PSD) to a Wasserstein barycenter estimated on training data. CMMN relies on novel closed-form solutions for optimal transport mappings and barycenters and provides individual test time adaptation to new data without needing to retrain a prediction model. Numerical experiments on sleep EEG data show that CMMN leads to significant and consistent performance gains independent from the neural network architecture when adapting between subjects, sessions, and even datasets collected with different hardware. Notably our performance gain is on par with much more numerically intensive Domain Adaptation (DA) methods and can be used in conjunction with those for even better performances.
|
['Alexandre Gramfort', 'Rémi Flamary', 'Théo Gnassounou']
|
2023-05-30
| null | null | null | null |
['eeg', 'eeg']
|
['methodology', 'time-series']
|
[ 2.43551224e-01 -1.29471883e-01 3.56346160e-01 -4.35032636e-01
-4.27560061e-01 -2.86132067e-01 1.48322985e-01 2.29231387e-01
-7.52982914e-01 1.14777493e+00 -7.69921616e-02 -4.80930880e-02
-4.85009789e-01 -2.89798170e-01 -7.75253117e-01 -7.74183512e-01
-4.26568151e-01 3.51755083e-01 -5.09979278e-02 -5.98344058e-02
1.03886433e-01 5.34966528e-01 -8.70553136e-01 -2.88486689e-01
7.72360742e-01 1.03908432e+00 2.18296517e-02 5.42530239e-01
2.38679096e-01 1.49388490e-02 -8.03059220e-01 -7.61470795e-02
3.04043174e-01 -4.95344043e-01 -5.63650012e-01 -1.95341781e-01
4.48865257e-02 8.70897546e-02 -1.57906562e-01 1.16175342e+00
7.47590363e-01 2.90979028e-01 6.59191906e-01 -1.10426199e+00
-4.18360174e-01 5.39575279e-01 -4.31104004e-01 5.30573070e-01
-2.08982453e-01 -1.59956411e-01 3.07477266e-01 -4.71433312e-01
1.70977980e-01 6.11434400e-01 1.00421083e+00 8.08463693e-01
-1.72956908e+00 -9.25096929e-01 -2.01630816e-01 3.85213971e-01
-1.48158813e+00 -3.13389361e-01 6.45222187e-01 -3.90256226e-01
7.30355442e-01 2.90052462e-02 6.91176951e-01 1.15443969e+00
5.91987967e-01 8.78388733e-02 8.05499196e-01 -1.99377194e-01
7.21134067e-01 2.01533690e-01 9.09122452e-02 2.41361529e-01
2.95053542e-01 -3.82652551e-01 -5.51455677e-01 -3.39162380e-01
7.54098833e-01 -2.78999805e-01 -6.58192396e-01 -3.99086148e-01
-1.22968340e+00 4.41020399e-01 2.37951830e-01 4.06175137e-01
-5.45884907e-01 1.12083778e-01 6.39300644e-01 4.60115999e-01
5.57320714e-01 6.82674050e-01 -7.75338531e-01 -3.06387961e-01
-8.88951719e-01 -1.11453542e-02 6.82523727e-01 7.41135657e-01
5.37508965e-01 1.08968824e-01 -1.57927006e-01 7.11344957e-01
-4.06357855e-01 4.49893981e-01 7.93733120e-01 -7.54275024e-01
2.20477000e-01 2.00102404e-01 -2.95899659e-02 -6.21434927e-01
-1.02956557e+00 -6.96071267e-01 -1.21789551e+00 -1.30633265e-01
3.99520546e-01 -5.12109995e-01 -6.02806866e-01 1.91572642e+00
2.26324070e-02 5.57130635e-01 -3.55859287e-02 6.83551431e-01
7.22123682e-02 3.51526707e-01 -1.71005297e-02 -3.38267654e-01
1.10732162e+00 -1.72041252e-01 -7.17126012e-01 -1.10904455e-01
5.08694410e-01 -3.32574189e-01 1.07114494e+00 6.46369576e-01
-1.12897801e+00 -2.93693811e-01 -1.25978541e+00 3.01784426e-01
-1.25436008e-01 6.58126846e-02 3.40973049e-01 5.96589327e-01
-1.18334639e+00 1.05189943e+00 -1.18958068e+00 -4.65645790e-01
4.79462862e-01 8.32319319e-01 -3.01008791e-01 4.31019485e-01
-9.93980765e-01 8.59004021e-01 3.40382516e-01 5.16355336e-02
-4.41209316e-01 -1.27845287e+00 -4.06190723e-01 2.78294861e-01
-2.38819659e-01 -6.03405237e-01 9.89624798e-01 -1.02358103e+00
-1.74342668e+00 2.27430880e-01 1.19362935e-01 -9.20067370e-01
3.78345102e-01 1.34062126e-01 -4.35729027e-01 8.20520669e-02
-1.58653364e-01 4.49090809e-01 9.45654750e-01 -4.91185933e-01
-2.40222350e-01 -5.25853336e-01 -5.71664512e-01 -1.05116412e-01
-9.33342814e-01 -1.98149890e-01 7.50207230e-02 -5.54361045e-01
3.27597857e-02 -9.02823508e-01 7.84751698e-02 -1.26504853e-01
-1.22303739e-01 -2.57778913e-03 4.97016877e-01 -6.82129323e-01
9.55096662e-01 -2.14113784e+00 4.97331083e-01 4.90625113e-01
6.97190166e-02 3.53228524e-02 -4.90809828e-02 4.34727594e-02
-2.40235031e-01 -3.22988808e-01 -4.30888742e-01 -2.74418056e-01
-1.48213178e-01 1.36613235e-01 -1.66138977e-01 8.30334663e-01
2.22486809e-01 5.86593330e-01 -6.31048620e-01 9.73046571e-02
-4.40620892e-02 7.85033286e-01 -5.64593434e-01 -3.69070880e-02
3.16481084e-01 9.65994298e-01 -8.93456116e-02 -1.87620409e-02
5.78782856e-01 -3.73356938e-01 1.63291916e-01 -2.54368752e-01
1.62856549e-01 7.07981959e-02 -9.57835257e-01 1.85441387e+00
-5.99755824e-01 9.06481862e-01 1.04747236e-01 -1.43309689e+00
8.07288706e-01 3.91635209e-01 8.22195113e-01 -7.29159057e-01
3.40311736e-01 2.17887461e-01 1.29953668e-01 -3.26017529e-01
2.82130316e-02 -6.86594024e-02 1.22178383e-01 4.64187175e-01
4.05223519e-01 2.17305884e-01 2.00531222e-02 -2.04802886e-01
1.28700078e+00 -2.98334658e-01 1.16265498e-01 -9.88381922e-01
4.64986026e-01 -4.85230505e-01 6.09169841e-01 4.71841067e-01
-2.55267452e-02 3.63501012e-01 4.56756473e-01 -3.26275647e-01
-1.14310813e+00 -1.11517084e+00 -6.94548905e-01 6.71837628e-01
-7.03270063e-02 -1.39496565e-01 -1.18186724e+00 -1.54113412e-01
-8.60128030e-02 5.22075891e-01 -7.42589355e-01 -6.67415738e-01
-5.52441359e-01 -1.20241058e+00 5.94397426e-01 7.22140670e-01
4.41369295e-01 -7.78800547e-01 -8.20913255e-01 3.82392734e-01
1.12196729e-01 -1.14760888e+00 -5.68924010e-01 5.47596216e-01
-1.06712747e+00 -7.66808808e-01 -8.11435580e-01 -4.79339868e-01
7.85433292e-01 -4.18238938e-01 6.82300270e-01 -3.45874608e-01
-4.83295143e-01 4.62675482e-01 -2.95250304e-02 -6.23843312e-01
-1.15240708e-01 3.08369428e-01 6.64951801e-01 1.52077407e-01
3.18216890e-01 -1.06723309e+00 -6.99005067e-01 2.94706851e-01
-8.22175980e-01 -2.92403907e-01 3.79158288e-01 8.84926975e-01
4.45422024e-01 2.46451229e-01 9.41972971e-01 -5.87574661e-01
9.53254282e-01 -3.95758033e-01 -7.69655228e-01 6.01872988e-02
-8.55395257e-01 2.74245918e-01 9.04831648e-01 -7.37779081e-01
-5.95712602e-01 -2.23201558e-01 1.96535423e-01 -3.15336496e-01
1.44183874e-01 1.05073318e-01 1.69398040e-01 -4.43948686e-01
9.00892973e-01 2.51939416e-01 -5.58988471e-03 -5.07304966e-01
-8.62383097e-02 5.75469971e-01 7.67498076e-01 -5.67097664e-01
5.70938349e-01 4.10161674e-01 2.31149003e-01 -9.75639045e-01
-3.47224206e-01 -1.62441611e-01 -8.00444186e-01 1.28644332e-01
8.57730508e-01 -5.74026525e-01 -9.38573539e-01 3.49192590e-01
-9.42637682e-01 -6.86251640e-01 -4.05469596e-01 8.33518803e-01
-4.99071896e-01 7.25801811e-02 -4.50626522e-01 -5.70181668e-01
-6.17622793e-01 -7.51713872e-01 6.26253784e-01 2.31436968e-01
-2.14359462e-01 -1.18882847e+00 -3.79911438e-02 -3.11824739e-01
7.23292410e-01 1.08389743e-02 1.02101970e+00 -5.39991796e-01
-4.50841896e-03 -3.04796726e-01 -6.35448247e-02 6.21899188e-01
2.36295924e-01 -4.24734503e-01 -9.59255338e-01 -6.94910824e-01
3.57322574e-01 1.27593100e-01 3.27254027e-01 6.94505453e-01
1.39641571e+00 -2.19151318e-01 -1.30873084e-01 9.23051059e-01
1.13772786e+00 2.21299931e-01 4.20713425e-01 2.89266229e-01
3.61069798e-01 2.75157243e-01 -1.97593868e-01 6.08826697e-01
1.11928329e-01 5.90499461e-01 -1.41319975e-01 7.15554431e-02
1.97475806e-01 3.27509701e-01 2.58631289e-01 9.59781826e-01
-3.55795361e-02 1.70148298e-01 -7.80279458e-01 3.00291598e-01
-1.54097426e+00 -6.22275174e-01 2.50890404e-01 2.52304411e+00
7.76296735e-01 1.01250432e-01 2.39356279e-01 2.70615608e-01
5.46040475e-01 -7.53377378e-01 -9.98491287e-01 -2.76478618e-01
1.29739977e-02 6.92722499e-01 9.40317094e-01 1.23808824e-01
-6.41580284e-01 1.98677704e-01 6.72663069e+00 4.76023972e-01
-1.40311790e+00 4.29361701e-01 3.56472701e-01 -4.47363317e-01
1.90536022e-01 -4.89476472e-01 -5.04707217e-01 6.97173595e-01
1.40609419e+00 -4.84109402e-01 9.64939475e-01 3.44830424e-01
2.46764421e-01 1.26814485e-01 -1.20100808e+00 1.22934806e+00
3.01876348e-02 -1.05265450e+00 -6.13418639e-01 3.49594913e-02
7.25057900e-01 2.06392184e-01 4.73725870e-02 1.55606329e-01
-3.12826067e-01 -7.63309240e-01 4.38424945e-01 6.38909519e-01
6.73740745e-01 -8.70410085e-01 6.52785838e-01 2.97641963e-01
-8.36030662e-01 -3.25579077e-01 -6.19511366e-01 9.35565010e-02
-6.91901222e-02 6.38052404e-01 -8.01560640e-01 2.20843568e-01
9.66257572e-01 5.89369118e-01 -4.26953226e-01 1.21385670e+00
2.31481478e-01 6.93166018e-01 -3.40294957e-01 -2.96071693e-02
-1.67029038e-01 -3.02130342e-01 5.09683847e-01 1.03866374e+00
5.40913343e-01 -8.66247714e-02 -3.44279617e-01 9.02544618e-01
-1.67155564e-01 1.36290610e-01 -1.69430360e-01 5.47754392e-02
3.84185553e-01 1.24664009e+00 -7.95398176e-01 -1.68552145e-01
-1.48692042e-01 1.10207188e+00 2.32292682e-01 5.52033663e-01
-8.45485330e-01 -6.30390108e-01 8.21714044e-01 2.05757946e-01
2.21869014e-02 -3.22292566e-01 -5.36089897e-01 -1.13272858e+00
2.25118831e-01 -3.82607460e-01 2.55918771e-01 -5.57231426e-01
-1.36412597e+00 8.32849979e-01 2.11059988e-01 -1.16404533e+00
-1.55320659e-01 -6.73962653e-01 -5.44072092e-01 8.17295074e-01
-1.28063059e+00 -2.37322509e-01 -1.55967116e-01 7.21561134e-01
2.36995853e-02 -2.22211972e-01 8.38088691e-01 7.22834587e-01
-6.84395850e-01 8.48671019e-01 4.87062305e-01 -9.59831774e-02
7.11580038e-01 -1.18828189e+00 2.76511341e-01 5.90055764e-01
-3.36848088e-02 5.59084654e-01 7.54281938e-01 -2.29173109e-01
-1.09021473e+00 -1.03276169e+00 2.57982969e-01 -2.43772104e-01
8.42180431e-01 -6.30465269e-01 -1.24622810e+00 5.85946143e-01
1.98980004e-01 6.59180665e-03 6.54551864e-01 1.16890512e-01
9.94639471e-02 -6.32221699e-01 -1.15694284e+00 3.73185456e-01
8.10294092e-01 -3.82122070e-01 -1.95177615e-01 4.96000350e-01
2.57885396e-01 -4.16967809e-01 -1.11352158e+00 3.51394087e-01
3.57690632e-01 -5.61918437e-01 7.04964757e-01 -5.78718841e-01
-3.63567621e-01 -3.49403918e-03 5.87859824e-02 -1.72518420e+00
-2.67335773e-01 -9.09239769e-01 7.01989383e-02 8.76062870e-01
3.37486058e-01 -9.96148288e-01 6.40845656e-01 9.85278726e-01
-1.40111074e-01 -6.19107306e-01 -1.28911698e+00 -1.04866147e+00
3.02986175e-01 -3.49032849e-01 7.55256355e-01 7.70233393e-01
2.81621248e-01 2.69525617e-01 -1.86162904e-01 2.24263817e-01
6.70113921e-01 -3.63242894e-01 3.94898415e-01 -1.41566694e+00
-3.01281244e-01 -4.02272403e-01 -9.33478057e-01 -5.71765661e-01
2.84525335e-01 -9.64320481e-01 8.74300255e-04 -1.02151787e+00
3.90007421e-02 -4.23233420e-01 -7.17076957e-01 5.14555156e-01
2.03399345e-01 3.87766927e-01 -2.64035642e-01 -5.41010536e-02
-1.64928854e-01 5.99486947e-01 7.95111716e-01 -4.12339419e-02
-6.44142210e-01 8.59538391e-02 -4.20042336e-01 6.69269383e-01
1.01011407e+00 -5.58483839e-01 -6.10203028e-01 -5.88382125e-01
1.02538176e-01 -1.05819546e-01 3.66737455e-01 -1.59128714e+00
3.89988631e-01 3.64782929e-01 6.68268740e-01 1.55488789e-01
2.89596051e-01 -7.53689110e-01 2.20339194e-01 3.92488182e-01
-3.66238624e-01 2.41524562e-01 6.54280007e-01 5.49415767e-01
2.52533823e-01 1.28087187e-02 9.87032890e-01 4.82116222e-01
-9.54018608e-02 3.38127494e-01 -3.53651524e-01 1.23045437e-01
8.57811868e-01 -7.18566552e-02 -9.30228680e-02 -2.74262607e-01
-9.61363494e-01 5.44895865e-02 1.92065895e-01 1.99178591e-01
3.58854204e-01 -1.32352912e+00 -5.98199487e-01 6.34528518e-01
-1.28191650e-01 -3.21996987e-01 2.93032110e-01 1.27191687e+00
-1.41386434e-01 3.33009362e-01 -4.92847711e-01 -6.15952194e-01
-9.00742471e-01 1.71582133e-01 6.28284097e-01 2.85098016e-01
-8.57611537e-01 7.95372784e-01 -2.65447404e-02 -1.30017951e-01
2.48505980e-01 -5.59074342e-01 1.32548094e-01 -1.92260459e-01
5.66780984e-01 3.87474537e-01 6.17471099e-01 -1.91731334e-01
-3.20523411e-01 4.58370268e-01 9.19543356e-02 -5.22099957e-02
1.54737031e+00 6.17028773e-02 -1.39330253e-01 5.09281516e-01
1.27326727e+00 -4.87633675e-01 -1.38379514e+00 -2.13607207e-01
-4.51403856e-02 -1.76899940e-01 1.67402580e-01 -5.49864948e-01
-1.21653390e+00 8.26688588e-01 1.06218743e+00 1.79625109e-01
1.41576147e+00 -2.58804053e-01 7.47098446e-01 6.06581032e-01
4.33099866e-01 -1.30881107e+00 -1.02681287e-01 1.70117483e-01
6.99436665e-01 -6.74687505e-01 -7.53938779e-02 3.54834914e-01
-3.86111706e-01 1.23090756e+00 3.32743049e-01 -3.08031648e-01
9.35934365e-01 3.98353547e-01 -9.68596190e-02 -4.63209040e-02
-5.13845205e-01 5.01546085e-01 3.28193635e-01 6.73854470e-01
1.56103745e-01 -3.65453921e-02 -2.20266536e-01 6.81375742e-01
-2.57135689e-01 7.22801089e-02 5.55497408e-01 5.45018315e-01
-9.67615321e-02 -9.79398131e-01 -2.12658793e-01 7.33644187e-01
-4.02924299e-01 -7.48699978e-02 2.11743996e-01 5.44451594e-01
-9.74336118e-02 5.58147430e-01 4.26849753e-01 -3.63356739e-01
4.47677106e-01 1.75468400e-01 7.91799307e-01 -4.29640114e-01
-3.21561068e-01 -1.27623826e-01 -6.17825627e-01 -5.62517524e-01
-2.18392536e-01 -7.80074656e-01 -1.54704762e+00 -2.53100097e-01
-2.92292595e-01 2.46411249e-01 9.96486008e-01 9.89311576e-01
6.56561375e-01 7.78120935e-01 4.97605860e-01 -8.61023009e-01
-6.01593494e-01 -1.02281487e+00 -9.16106999e-01 2.96003014e-01
3.90403777e-01 -6.66070282e-01 -3.59861612e-01 -5.14024170e-03]
|
[13.109780311584473, 3.448014497756958]
|
80d33001-0a15-4fc9-89a5-9a55a50913c4
|
multi-modal-learning-with-prior-visual
|
1812.09681
| null |
https://arxiv.org/abs/1812.09681v2
|
https://arxiv.org/pdf/1812.09681v2.pdf
|
Scene Graph Reasoning with Prior Visual Relationship for Visual Question Answering
|
One of the key issues of Visual Question Answering (VQA) is to reason with semantic clues in the visual content under the guidance of the question, how to model relational semantics still remains as a great challenge. To fully capture visual semantics, we propose to reason over a structured visual representation - scene graph, with embedded objects and inter-object relationships. This shows great benefit over vanilla vector representations and implicit visual relationship learning. Based on existing visual relationship models, we propose a visual relationship encoder that projects visual relationships into a learned deep semantic space constrained by visual context and language priors. Upon the constructed graph, we propose a Scene Graph Convolutional Network (SceneGCN) to jointly reason the object properties and relational semantics for the correct answer. We demonstrate the model's effectiveness and interpretability on the challenging GQA dataset and the classical VQA 2.0 dataset, remarkably achieving state-of-the-art 54.56% accuracy on GQA compared to the existing best model.
|
['Zhuoqian Yang', 'Yue Hu', 'Zengchang Qin', 'Jing Yu']
|
2018-12-23
| null | null | null | null |
['cross-modal-information-retrieval']
|
['miscellaneous']
|
[-4.27813306e-02 3.93912494e-01 -5.59675768e-02 -5.98224103e-01
-5.60137093e-01 -5.18770993e-01 5.66808760e-01 2.47195795e-01
3.06720417e-02 3.70068625e-02 4.95052636e-01 -6.18391216e-01
-4.26671058e-02 -8.60729814e-01 -9.82031286e-01 -1.60900101e-01
2.35893920e-01 5.63392699e-01 2.32626915e-01 -3.30309600e-01
1.11248724e-01 2.14936331e-01 -1.50810421e+00 8.12783241e-01
6.11691415e-01 1.03782547e+00 2.77136326e-01 7.56287754e-01
-6.27758682e-01 1.57526398e+00 -4.36515808e-01 -6.58145487e-01
-2.47322768e-01 -3.45988363e-01 -1.40299392e+00 1.99486047e-01
7.26096213e-01 -3.20961177e-01 -6.31736040e-01 7.96722710e-01
-1.60573334e-01 1.56081200e-01 5.12075305e-01 -1.57872045e+00
-1.53478181e+00 3.42283636e-01 -4.27086443e-01 1.70219496e-01
5.12363970e-01 3.44807833e-01 1.76331460e+00 -9.25419152e-01
7.15101719e-01 1.70088744e+00 2.15708390e-01 4.49331999e-01
-1.19146371e+00 -2.18240961e-01 5.00395954e-01 5.67721665e-01
-1.16678822e+00 -1.27008960e-01 8.03267419e-01 -5.36566079e-01
1.16849530e+00 1.92619473e-01 6.66669786e-01 9.15657878e-01
-2.83247113e-01 9.70692873e-01 8.08909714e-01 -2.78572828e-01
5.90125993e-02 -6.03262447e-02 4.28137571e-01 1.10330033e+00
1.81114078e-02 -4.00758624e-01 -5.60690701e-01 1.66790128e-01
7.33510077e-01 1.14583574e-01 -1.99512154e-01 -9.66689229e-01
-1.06830144e+00 9.89075243e-01 1.25816035e+00 -8.30313042e-02
-9.48496237e-02 7.81286895e-01 1.68507367e-01 4.59231213e-02
9.73885134e-02 4.04080480e-01 -2.33527318e-01 2.99905330e-01
-2.28048086e-01 2.94302940e-01 4.27189887e-01 1.10505688e+00
8.62541616e-01 1.62048079e-03 -4.42584008e-01 6.87486112e-01
7.81741738e-01 4.76992965e-01 -2.76036263e-01 -1.07815731e+00
7.10822225e-01 1.19772613e+00 -3.89827825e-02 -1.21280694e+00
-3.57900739e-01 -1.29400000e-01 -4.86227453e-01 1.58884481e-01
5.58993995e-01 6.11373663e-01 -1.20136774e+00 1.60333991e+00
3.46480787e-01 -1.46170944e-01 3.49354744e-01 1.15643692e+00
1.63243246e+00 8.13563466e-01 4.69902068e-01 4.90830392e-01
1.78409040e+00 -1.20171082e+00 -7.32399642e-01 -5.26486397e-01
5.20979762e-01 -4.04004723e-01 1.51390469e+00 -1.41483560e-01
-1.02006221e+00 -6.88436329e-01 -9.52811837e-01 -1.06859267e+00
-5.72043538e-01 2.46735383e-02 8.04164708e-01 8.49825889e-03
-1.10318279e+00 -1.92797005e-01 -5.19664466e-01 -5.36890686e-01
8.17587674e-01 -1.21692652e-02 -4.47926939e-01 -4.84936655e-01
-1.01955736e+00 7.23527312e-01 2.45898277e-01 2.21188188e-01
-1.16110110e+00 -6.24146461e-01 -1.24210453e+00 2.84092963e-01
5.97974777e-01 -1.19894981e+00 1.15566063e+00 -6.02892756e-01
-8.90517652e-01 1.17589700e+00 -3.88428390e-01 -3.02551180e-01
9.02733952e-02 -2.86503434e-01 -1.84493095e-01 4.67792600e-01
3.17202210e-01 8.41052651e-01 5.56400239e-01 -1.72097576e+00
-1.82063267e-01 -3.85671198e-01 7.02881157e-01 1.67395756e-01
2.42469028e-01 -1.35955751e-01 -9.49828506e-01 -3.38033944e-01
2.04812706e-01 -5.06809354e-01 -6.77605048e-02 4.00994301e-01
-4.12001520e-01 -5.52222073e-01 9.46057498e-01 -8.85592878e-01
6.72879219e-01 -2.11680746e+00 2.43915409e-01 8.33970159e-02
7.06686795e-01 5.59065305e-02 -3.75088483e-01 3.88669074e-01
-7.47155100e-02 -2.78092492e-02 -2.59996772e-01 -1.80036426e-01
1.07643493e-01 6.18348002e-01 -6.67164028e-01 1.72161654e-01
4.89080310e-01 1.66104019e+00 -1.14753497e+00 -4.09457147e-01
1.34046078e-01 6.02181554e-01 -6.53674126e-01 6.79288983e-01
-8.52413416e-01 2.96653211e-01 -5.90509653e-01 8.40829313e-01
4.66811955e-01 -1.14643550e+00 2.13787228e-01 -5.22501945e-01
3.99687916e-01 1.98802203e-01 -6.94206953e-01 1.97352350e+00
-3.37620497e-01 8.85648131e-01 -1.31662458e-01 -1.10800827e+00
1.20147002e+00 -1.38734907e-01 -1.06023520e-01 -1.22484541e+00
8.16252977e-02 -2.71021873e-01 -2.42082611e-01 -8.43020737e-01
4.05369580e-01 7.69901872e-02 -8.83997083e-02 2.49561459e-01
2.60417521e-01 -2.44874746e-01 -2.05795839e-01 9.92996931e-01
8.32335413e-01 3.30153257e-01 2.12692857e-01 4.80900109e-02
4.37796652e-01 1.32097796e-01 4.32247370e-02 4.93174613e-01
-1.19997248e-01 7.22372949e-01 9.78003442e-01 -6.04474127e-01
-8.53824794e-01 -1.31821656e+00 4.36499059e-01 1.18745708e+00
5.65488100e-01 -6.62692487e-01 -2.37816080e-01 -8.40290010e-01
2.49253988e-01 7.91053951e-01 -1.00332463e+00 -3.74260657e-02
-5.33530772e-01 2.28221733e-02 1.14812657e-01 8.28618467e-01
4.23236161e-01 -1.27245569e+00 -3.84922117e-01 -3.31138104e-01
-1.75181791e-01 -1.50766587e+00 -9.23694223e-02 -2.56222337e-01
-3.09041202e-01 -1.42700195e+00 -2.35832334e-01 -9.63370025e-01
6.66796267e-01 6.43990159e-01 1.60479140e+00 4.50260937e-01
-3.06029558e-01 9.34104741e-01 -3.29140216e-01 -8.29174295e-02
-2.24191323e-01 -4.53767180e-01 -8.29368114e-01 1.31157190e-01
1.90963119e-01 -1.95254609e-01 -7.80370355e-01 -3.08542512e-02
-7.58449674e-01 3.98009747e-01 2.63562411e-01 4.98649955e-01
7.33973801e-01 -5.59764564e-01 2.95769989e-01 -9.99857128e-01
3.65520567e-01 -5.34297168e-01 -5.54281890e-01 7.62065768e-01
-2.64005065e-01 3.04755419e-01 3.72112185e-01 1.59270123e-01
-1.02850235e+00 -1.68827400e-01 2.07212880e-01 -7.86666095e-01
-3.37653086e-02 2.83132851e-01 -5.42089283e-01 2.11379021e-01
3.47053617e-01 7.69030396e-03 -3.43037583e-02 -3.38776022e-01
1.43630910e+00 1.19698338e-01 8.40933621e-01 -5.99367082e-01
7.35225797e-01 7.32292533e-01 1.72811165e-01 -5.25444925e-01
-1.35032368e+00 -6.88840926e-01 -6.29953682e-01 -2.46867135e-01
1.56152129e+00 -1.08531845e+00 -9.98848438e-01 -2.62655526e-01
-1.45394421e+00 -3.99154097e-01 -1.93266705e-01 -9.93573815e-02
-6.15946531e-01 4.49480563e-01 -2.79441446e-01 -5.43733120e-01
-2.25928232e-01 -1.04604232e+00 1.24243712e+00 1.78358465e-01
-1.57626837e-01 -1.05308414e+00 -1.54280990e-01 1.12098336e+00
1.36352807e-01 3.39349002e-01 1.41203320e+00 -2.77674526e-01
-1.13318908e+00 3.39364588e-01 -1.13676262e+00 1.75620522e-02
-3.71900462e-02 -1.61363229e-01 -8.53295922e-01 -3.00130062e-03
-5.51661313e-01 -7.11152494e-01 1.02200460e+00 -1.66446287e-02
1.44533777e+00 -1.32325172e-01 -1.91281796e-01 7.17140496e-01
1.41393197e+00 -8.48331526e-02 5.96237063e-01 1.72483370e-01
1.54155302e+00 7.97005653e-01 4.42839742e-01 9.09129381e-02
1.07681370e+00 4.58415568e-01 1.16307735e+00 -2.48477310e-01
-6.48811698e-01 -7.31632411e-01 -1.79084122e-01 5.79444587e-01
2.24548608e-01 -2.69352555e-01 -1.11346686e+00 7.72194564e-01
-2.03561950e+00 -8.30934763e-01 -5.46552360e-01 1.49317789e+00
4.16357994e-01 -3.13102156e-01 -6.37305751e-02 -3.68492573e-01
3.82527173e-01 5.23575544e-01 -5.06235123e-01 -5.15436292e-01
-1.64810672e-01 -7.48854429e-02 5.38011789e-02 5.36446393e-01
-9.23107386e-01 1.10869980e+00 5.85865355e+00 2.50970453e-01
-6.15160942e-01 -1.92891825e-02 4.90780652e-01 3.17260265e-01
-8.85761559e-01 3.91595274e-01 -3.81309539e-01 -1.31366938e-01
3.82269770e-01 3.29737872e-01 4.16098088e-01 8.45466495e-01
-3.03439975e-01 1.31240264e-01 -1.14022493e+00 1.26661241e+00
5.14829397e-01 -1.68143761e+00 6.65336549e-01 -2.98192084e-01
5.27874947e-01 -4.13036704e-01 1.00931175e-01 5.21670520e-01
6.15667164e-01 -1.44092512e+00 7.79989719e-01 6.87894821e-01
7.43054569e-01 -4.52454597e-01 2.34849975e-01 -2.23137826e-01
-1.51674604e+00 -2.15295758e-02 -3.41922253e-01 -5.70620447e-02
2.47006714e-01 -1.34371832e-01 -8.71858239e-01 6.94870174e-01
9.03320789e-01 1.12015486e+00 -9.44035292e-01 6.49245679e-01
-6.43451631e-01 5.00495255e-01 3.79132897e-01 4.04190943e-02
4.35239315e-01 -1.46352246e-01 1.70013890e-01 8.98220479e-01
-1.51098117e-01 2.03717873e-01 -2.86760312e-02 1.31146944e+00
-3.07116091e-01 -1.16570361e-01 -6.86303079e-01 -1.41139627e-01
8.60045031e-02 1.13059938e+00 -6.40574694e-01 -3.47800046e-01
-9.19176280e-01 9.42526758e-01 1.00761902e+00 8.07476938e-01
-7.86475480e-01 -6.59091920e-02 6.34708941e-01 3.62336300e-02
5.43457031e-01 -1.39384970e-01 -2.64817029e-01 -1.12396955e+00
-1.54475472e-03 -7.50290930e-01 9.30173278e-01 -1.66534972e+00
-1.38533735e+00 2.95601189e-01 1.35237332e-02 -8.78770053e-01
1.61039248e-01 -9.79463041e-01 -3.97700310e-01 7.78121054e-01
-1.57540512e+00 -1.86500525e+00 -6.37629807e-01 8.41811359e-01
4.79578078e-01 2.90095098e-02 5.68568528e-01 -1.45684227e-01
-5.94416671e-02 1.87017426e-01 -5.63660920e-01 3.78822833e-01
2.72080541e-01 -1.45449603e+00 6.24205828e-01 6.60209239e-01
7.45091498e-01 5.85783780e-01 4.08992141e-01 -3.62023741e-01
-1.74175215e+00 -1.09841311e+00 9.33684945e-01 -9.75564599e-01
8.76753330e-01 -6.79044306e-01 -1.32095492e+00 9.93171334e-01
5.18759549e-01 4.32057321e-01 5.01619935e-01 2.28968292e-01
-1.26175082e+00 6.99677244e-02 -6.00317419e-01 6.49829566e-01
1.19334722e+00 -1.07657075e+00 -1.00622141e+00 2.81557053e-01
1.32267785e+00 -3.24758261e-01 -5.58923066e-01 2.89888650e-01
2.42610469e-01 -7.39188194e-01 1.46152270e+00 -1.10802400e+00
5.51482618e-01 -6.40679300e-01 -4.53338653e-01 -8.14882696e-01
-3.32222402e-01 -1.34016454e-01 -2.96566069e-01 1.15940905e+00
2.72642881e-01 -1.55742437e-01 4.94976789e-01 6.49276495e-01
-8.90656374e-03 -5.70325613e-01 -5.91070592e-01 -2.85517544e-01
-1.50232837e-01 -4.57240820e-01 5.59250116e-01 1.13883018e+00
-2.57042170e-01 9.27468598e-01 -3.10230345e-01 4.30078387e-01
5.77323139e-01 6.45565152e-01 1.05453455e+00 -1.04921937e+00
-3.30990016e-01 -2.79985279e-01 -4.62397814e-01 -1.21714616e+00
3.54772985e-01 -1.16549647e+00 -1.15903229e-01 -2.45374465e+00
4.63185042e-01 -1.14889301e-01 -1.98174536e-01 7.30685830e-01
-4.04723138e-01 2.00075313e-01 4.63771224e-01 -6.28832653e-02
-1.27020693e+00 7.57766664e-01 1.68253684e+00 -5.54489136e-01
2.08467036e-01 -7.25821316e-01 -8.46415937e-01 4.59733367e-01
4.36212927e-01 1.19228335e-03 -8.57000232e-01 -8.26492369e-01
7.01553345e-01 3.20902884e-01 1.04402161e+00 -2.81329066e-01
5.66844754e-02 -2.16055751e-01 2.27851450e-01 -7.05432653e-01
4.28574055e-01 -7.34113157e-01 -1.68408722e-01 2.34470759e-02
-5.43887436e-01 2.09859595e-01 1.56415328e-01 9.83479559e-01
-3.76118690e-01 3.23407352e-01 2.61758268e-01 -3.91449556e-02
-1.30808353e+00 3.13108087e-01 1.71191648e-01 5.03376663e-01
8.30653489e-01 4.25050072e-02 -9.52370822e-01 -5.54564595e-01
-7.93626010e-01 6.58870697e-01 2.66966194e-01 7.80645549e-01
1.06304491e+00 -1.35867047e+00 -4.56565857e-01 -7.87330419e-02
8.32273066e-01 2.03310758e-01 5.56586742e-01 2.24774167e-01
-6.83244586e-01 4.54564422e-01 -7.02358037e-02 -7.57491350e-01
-1.30664074e+00 1.02110696e+00 1.75476462e-01 7.61709064e-02
-9.19889033e-01 1.01415694e+00 8.26480865e-01 -4.18242484e-01
7.22766817e-02 -6.11405849e-01 -6.59943104e-01 -2.32054576e-01
3.90574694e-01 -9.07772705e-02 -3.49650919e-01 -9.37372386e-01
-5.25534511e-01 7.81669736e-01 1.99339166e-01 2.36078888e-01
1.04701555e+00 -2.14332670e-01 -2.51132011e-01 4.12391096e-01
1.36231005e+00 -1.58526540e-01 -1.26265788e+00 -4.39385027e-01
-5.96415512e-02 -6.13833725e-01 -2.08505064e-01 -7.41665304e-01
-1.07242322e+00 1.31325018e+00 1.85942486e-01 1.74000651e-01
7.19676137e-01 8.52645874e-01 3.27271670e-01 3.15945596e-01
-7.86642060e-02 -4.14228886e-01 8.64161372e-01 4.47527021e-01
1.31007791e+00 -1.51790619e+00 6.82509169e-02 -6.46518648e-01
-1.06037605e+00 9.86739278e-01 8.90120924e-01 -1.39417395e-01
3.65957290e-01 -5.32988966e-01 3.65749836e-01 -9.01993513e-01
-8.72915626e-01 -6.68202937e-01 7.74422407e-01 6.93066001e-01
1.92469686e-01 1.63275182e-01 3.29389602e-01 5.05880833e-01
-8.87443870e-03 -5.72501242e-01 1.42916724e-01 5.06637037e-01
-1.75447628e-01 -6.80649817e-01 -9.81742982e-03 -3.97049300e-02
-2.49525104e-02 -7.06673041e-02 -5.34918070e-01 9.99808311e-01
-1.39860854e-01 1.02093291e+00 2.89807588e-01 -2.37765908e-02
4.42689508e-01 -7.82673210e-02 4.34900612e-01 -5.49376547e-01
-2.41597876e-01 -4.00441259e-01 2.06354514e-01 -9.52166915e-01
-5.65972269e-01 -1.86787233e-01 -1.72671402e+00 2.11280599e-01
3.62061173e-01 -1.52020589e-01 3.33926797e-01 9.73977387e-01
3.78436685e-01 7.01435208e-01 5.30335605e-02 -6.23578131e-02
2.19341457e-01 -3.93491268e-01 -3.21729124e-01 1.03874850e+00
5.10036707e-01 -6.01015747e-01 -1.18220173e-01 1.31536856e-01]
|
[10.630026817321777, 1.7272694110870361]
|
2906f963-ca45-46db-9b5b-78c593c73048
|
text-independent-speaker-verification-using-1
|
1805.00604
| null |
http://arxiv.org/abs/1805.00604v3
|
http://arxiv.org/pdf/1805.00604v3.pdf
|
Text-Independent Speaker Verification Using Long Short-Term Memory Networks
|
In this paper, an architecture based on Long Short-Term Memory Networks has
been proposed for the text-independent scenario which is aimed to capture the
temporal speaker-related information by operating over traditional speech
features. For speaker verification, at first, a background model must be
created for speaker representation. Then, in enrollment stage, the speaker
models will be created based on the enrollment utterances. For this work, the
model will be trained in an end-to-end fashion to combine the first two stages.
The main goal of end-to-end training is the model being optimized to be
consistent with the speaker verification protocol. The end- to-end training
jointly learns the background and speaker models by creating the representation
space. The LSTM architecture is trained to create a discrimination space for
validating the match and non-match pairs for speaker verification. The proposed
architecture demonstrate its superiority in the text-independent compared to
other traditional methods.
|
['Mohammad Najarian', 'Aryan Mobiny']
|
2018-05-02
| null | null | null | null |
['text-independent-speaker-verification']
|
['speech']
|
[ 9.70780998e-02 -9.44685936e-02 -1.96080394e-02 -9.56553102e-01
-8.35115135e-01 -1.36131614e-01 5.46338081e-01 8.19400512e-03
-3.33275616e-01 2.61496097e-01 2.33596191e-01 -3.01010758e-01
8.51638541e-02 -3.00404757e-01 -1.40056193e-01 -7.49176741e-01
-1.82249665e-01 3.56852084e-01 -1.63251549e-01 -9.50096757e-04
2.67077088e-01 4.08701599e-01 -1.61125338e+00 3.75693142e-01
4.58660543e-01 1.01008844e+00 2.54292816e-01 8.04793060e-01
-3.63167673e-01 5.54195642e-01 -6.14876807e-01 -6.87649939e-03
-3.25870477e-02 -5.72173834e-01 -5.34825325e-01 2.92690452e-02
3.09287786e-01 -3.74112695e-01 -3.12603384e-01 9.24684465e-01
1.02296412e+00 4.10283417e-01 4.43229824e-01 -1.10150027e+00
-4.47159737e-01 8.17801416e-01 -2.75913715e-01 3.67819339e-01
2.94261783e-01 -3.93401179e-03 6.66487336e-01 -1.07492948e+00
1.82796791e-01 1.35064662e+00 2.88177341e-01 9.97600377e-01
-8.39266181e-01 -1.05136430e+00 5.07997394e-01 3.88753057e-01
-1.42875862e+00 -1.08811271e+00 1.03827524e+00 -3.48065227e-01
5.36801159e-01 1.40336365e-01 3.80848438e-01 1.12460339e+00
-3.87759581e-02 7.34345973e-01 8.62132847e-01 -7.30679870e-01
6.01491816e-02 4.18230921e-01 5.89760840e-01 4.39064711e-01
-5.01096129e-01 3.66986781e-01 -8.69408250e-01 -1.74944848e-02
3.25623482e-01 -2.01394688e-02 -1.45762891e-01 5.18339835e-02
-1.00753963e+00 5.69173872e-01 1.75470203e-01 7.47948289e-01
-4.91846055e-01 -3.29930007e-01 3.66521180e-01 2.10476920e-01
2.15431675e-01 -3.07375938e-01 -2.16931164e-01 3.27221602e-02
-1.36700904e+00 -1.31051585e-01 5.90895355e-01 5.79598725e-01
3.80940080e-01 3.56787741e-01 -6.36867523e-01 8.59532475e-01
1.04954815e+00 3.12923521e-01 1.12391615e+00 -8.60135928e-02
5.82707703e-01 3.74822080e-01 -1.48045987e-01 -5.15551329e-01
-2.50462711e-01 -9.04938698e-01 -8.94291162e-01 1.24360986e-01
5.51910400e-02 -8.16101804e-02 -1.27198374e+00 2.05991149e+00
4.21191931e-01 6.33551776e-01 5.54328799e-01 5.78387737e-01
1.21236598e+00 8.67078722e-01 2.82646537e-01 -1.60721079e-01
1.30506837e+00 -8.79582226e-01 -1.10612500e+00 -2.55885631e-01
3.66448671e-01 -8.36040556e-01 7.20501423e-01 4.53608809e-03
-8.85783613e-01 -8.97365808e-01 -1.21575081e+00 2.85380036e-01
-2.83298880e-01 3.36887717e-01 -1.62262708e-01 9.26599622e-01
-1.11705208e+00 1.70320496e-01 -6.30137265e-01 -2.70998776e-01
6.91538751e-02 4.79304582e-01 -1.25107259e-01 3.14919293e-01
-1.41428053e+00 8.04351211e-01 3.91966462e-01 7.07357168e-01
-1.30577970e+00 -3.01632971e-01 -9.47340548e-01 1.70024082e-01
-4.47583109e-01 -3.69728297e-01 1.50523722e+00 -1.20065296e+00
-2.04074740e+00 8.58538449e-01 -6.63870096e-01 -5.00982463e-01
3.89534891e-01 3.89081240e-01 -7.44543552e-01 -2.07406268e-01
-1.78065702e-01 3.67106915e-01 1.12329733e+00 -1.11353767e+00
-5.43678761e-01 -4.84863579e-01 -3.30419451e-01 3.28036964e-01
-4.85991329e-01 3.12171638e-01 -4.09376383e-01 -4.78171527e-01
4.58697885e-01 -7.76631892e-01 1.65942878e-01 -5.60157657e-01
-3.88465583e-01 -1.99959844e-01 1.10392308e+00 -1.09279835e+00
1.30047905e+00 -2.40070581e+00 -2.18485668e-01 3.95317972e-01
-1.77781656e-01 3.87766510e-01 -3.63736659e-01 2.67159283e-01
-5.04871309e-01 -1.15133010e-01 7.50385523e-02 -9.12229598e-01
-4.85776216e-02 -4.26107079e-01 -2.22837195e-01 5.56242526e-01
1.05769925e-01 5.76947391e-01 -3.02261323e-01 -6.40745759e-01
2.54802823e-01 5.43022275e-01 -2.14563310e-02 4.42704111e-01
1.81371346e-01 7.41302431e-01 -2.60972559e-01 4.82822955e-01
7.74620175e-01 2.16590434e-01 6.99274167e-02 1.53062180e-01
-1.35480091e-01 3.25867712e-01 -1.34069681e+00 1.66048098e+00
-5.07748187e-01 6.03772342e-01 4.58794653e-01 -9.04653609e-01
1.14979744e+00 9.52855706e-01 2.65276283e-01 -4.61017877e-01
4.27991897e-01 4.21412550e-02 2.98514932e-01 -4.25114214e-01
1.40122801e-01 -4.15927082e-01 1.53302878e-01 4.55698192e-01
7.43605942e-02 4.12181318e-01 -2.74691403e-01 -1.24502093e-01
4.01988745e-01 -2.23632887e-01 -2.53379941e-02 1.28560904e-02
1.25658369e+00 -5.57620287e-01 6.13861263e-01 4.49507356e-01
-4.99042660e-01 2.66553760e-01 -2.33170986e-01 -1.25380665e-01
-3.66020352e-01 -7.31743634e-01 -1.24747880e-01 1.02908075e+00
-8.57823417e-02 1.37885883e-02 -8.15342724e-01 -5.53757787e-01
-4.02311444e-01 7.81945527e-01 -4.76388216e-01 -2.60422170e-01
-4.31322068e-01 -3.02368641e-01 6.55195236e-01 3.22286606e-01
6.84896111e-01 -9.25654709e-01 1.94681473e-02 3.39296848e-01
-8.39938745e-02 -7.83516526e-01 -6.42431974e-01 1.90743506e-01
-8.63714278e-01 -5.16707301e-01 -8.80392611e-01 -1.18535650e+00
6.64359868e-01 1.67170949e-02 3.25450540e-01 1.12421624e-02
2.40448102e-01 5.90891503e-02 -1.29356772e-01 -6.31645203e-01
-6.64820611e-01 7.41240680e-02 2.36290902e-01 6.88491881e-01
5.28700173e-01 -4.45973933e-01 -4.03585643e-01 3.37699860e-01
-5.92407823e-01 -9.40673724e-02 6.64706171e-01 9.68160808e-01
2.77791530e-01 1.49915844e-01 8.66339803e-01 -2.39298895e-01
4.93105322e-01 -1.62631035e-01 -5.52557409e-01 4.37296003e-01
-3.72386336e-01 2.39256144e-01 3.75305802e-01 -6.51415646e-01
-1.28860748e+00 3.27820510e-01 -4.89461809e-01 -3.73844534e-01
-2.54807979e-01 7.20541716e-01 -5.80717146e-01 6.70266971e-02
2.36731425e-01 6.14284396e-01 6.80296198e-02 -4.25267458e-01
-6.82928087e-03 1.30727780e+00 2.36702010e-01 -3.00478965e-01
7.81456769e-01 -8.51079226e-02 -5.69255590e-01 -9.02411938e-01
-5.35715818e-01 -6.45279527e-01 -8.10083091e-01 -3.78899038e-01
8.88439059e-01 -1.05674827e+00 -6.66172385e-01 7.48408437e-01
-1.27613199e+00 -4.77964245e-02 5.02411425e-02 8.50319743e-01
-1.70815915e-01 2.18323246e-01 -3.69124740e-01 -1.41446555e+00
-6.47158563e-01 -1.14592218e+00 8.37968647e-01 4.83080685e-01
-4.06435430e-02 -8.55745673e-01 7.54350573e-02 4.52080011e-01
5.58298528e-01 -4.14005458e-01 6.63187742e-01 -1.19750834e+00
-4.04186487e-01 -5.33054590e-01 2.32892409e-01 4.89179313e-01
1.33853644e-01 -3.10048908e-02 -1.65429986e+00 -4.68229562e-01
4.25378084e-01 5.30733317e-02 6.67033970e-01 4.73424166e-01
8.27863812e-01 -2.06062078e-01 -2.26945236e-01 3.39032739e-01
1.01276958e+00 5.90872586e-01 4.13792580e-01 1.19768396e-01
3.35322917e-01 7.50946522e-01 5.13606369e-01 5.37763685e-02
2.15823621e-01 6.17093623e-01 7.72763640e-02 -9.50909182e-02
-1.72326773e-01 -2.94141501e-01 6.51736319e-01 9.53754425e-01
6.72111213e-01 -2.24979326e-01 -9.71174419e-01 4.46050704e-01
-1.63740110e+00 -1.30944455e+00 2.14363381e-01 2.42051053e+00
5.43484151e-01 1.71064317e-01 1.32475253e-02 3.44326258e-01
1.19114590e+00 3.44449788e-01 -5.17972887e-01 -4.32369262e-01
2.04991698e-01 5.38325273e-02 -8.16379860e-02 6.97827935e-01
-1.11671925e+00 7.54188418e-01 6.06387424e+00 5.50333202e-01
-1.89224982e+00 2.69843251e-01 4.97848958e-01 2.67623551e-02
-3.51182893e-02 -6.65993541e-02 -1.21780741e+00 3.27803999e-01
1.41783690e+00 -1.84729263e-01 -7.75372013e-02 7.58328021e-01
6.78406715e-01 2.66138166e-01 -1.17622304e+00 1.18419445e+00
3.96420002e-01 -8.57967615e-01 6.95018247e-02 -1.28160045e-01
2.89393067e-01 -2.53138453e-01 2.35594720e-01 5.27415931e-01
-2.38746867e-01 -9.75852668e-01 9.56533074e-01 5.15130460e-01
6.16885900e-01 -6.26496136e-01 8.85274231e-01 6.85403168e-01
-1.41690207e+00 -2.61680961e-01 1.66632906e-01 1.48887992e-01
2.84170330e-01 2.17817008e-01 -1.26593542e+00 4.97565687e-01
2.67910272e-01 3.42362702e-01 -3.36026371e-01 9.63235676e-01
-3.39262076e-02 6.26908660e-01 -3.49252112e-03 -7.22970068e-02
1.76146060e-01 1.47500187e-01 4.48164314e-01 1.22830677e+00
4.37067628e-01 -2.72845030e-01 1.94723248e-01 6.65797055e-01
-1.39036492e-01 3.49792153e-01 -4.81427938e-01 -5.52770793e-02
5.77665627e-01 1.00484383e+00 -2.39250198e-01 -3.87782097e-01
-1.56781554e-01 8.26427460e-01 1.76329922e-03 4.60251421e-01
-9.04958546e-01 -5.55834949e-01 3.45593899e-01 -2.58345306e-02
2.40562558e-01 -9.20054987e-02 -1.35988101e-01 -9.28473830e-01
5.45578711e-02 -9.99065697e-01 3.04419667e-01 -3.09208542e-01
-9.95336831e-01 9.10967171e-01 -2.49241665e-01 -1.23658812e+00
-5.06817639e-01 -2.14253739e-01 -1.11297464e+00 1.61209381e+00
-1.69164240e+00 -1.37084019e+00 -2.02757120e-01 7.95775414e-01
7.90818334e-01 -4.83692914e-01 9.83658671e-01 5.16128182e-01
-1.00948334e+00 9.81614053e-01 1.67267863e-02 4.98121083e-01
7.43354678e-01 -7.48803556e-01 1.56407848e-01 1.23930705e+00
1.94435686e-01 8.43249142e-01 5.76846838e-01 -5.18557966e-01
-1.01467717e+00 -1.03451025e+00 1.26785624e+00 7.31984898e-02
1.56239316e-01 -3.80753815e-01 -7.89234996e-01 6.33005321e-01
1.11407183e-01 -2.25831553e-01 9.63979244e-01 2.37993702e-01
-1.91097870e-01 -4.84519839e-01 -1.22203362e+00 2.66875148e-01
3.78244072e-01 -1.01723075e+00 -7.57038116e-01 -7.12402631e-03
3.00039172e-01 -3.56210232e-01 -5.58718622e-01 1.67871594e-01
6.70279682e-01 -5.47670186e-01 6.82398498e-01 -4.62125242e-01
-3.14520150e-01 -4.90627468e-01 -9.05555785e-02 -9.94576991e-01
-3.13458622e-01 -6.53155804e-01 3.34738009e-02 1.80524874e+00
7.95943558e-01 -4.87604231e-01 8.51463795e-01 6.35439515e-01
-7.37652779e-02 -2.86757290e-01 -1.14284360e+00 -8.01167548e-01
-2.54357249e-01 -4.16741848e-01 8.31589043e-01 7.65685201e-01
-2.21236557e-01 6.48789644e-01 -2.99914509e-01 6.61987007e-01
4.90719467e-01 9.63410828e-03 6.43138349e-01 -1.25654399e+00
-9.21761617e-02 -3.65113378e-01 -3.87928843e-01 -1.11669624e+00
5.38742840e-01 -1.05704522e+00 3.85279864e-01 -1.39002800e+00
4.10092771e-02 -3.70248526e-01 -7.49869347e-01 1.95464194e-01
-1.58724457e-01 -3.11781138e-01 6.76392838e-02 1.42527223e-01
-1.12790927e-01 6.53956175e-01 7.55296528e-01 -4.54042941e-01
-6.58502758e-01 4.67199683e-01 -4.91972357e-01 2.96876520e-01
9.73392248e-01 -3.65923703e-01 -5.18415749e-01 -2.51789510e-01
-9.04709578e-01 3.26596886e-01 1.18750118e-01 -1.11757112e+00
5.58489501e-01 1.63032547e-01 4.91492450e-01 -1.02855885e+00
6.10022068e-01 -8.39759052e-01 -2.10166369e-02 6.49416625e-01
-7.24759459e-01 3.90814655e-02 2.92743206e-01 5.64890563e-01
-6.43740416e-01 -3.59527171e-01 9.81274962e-01 2.10954025e-01
-4.52560961e-01 5.13013601e-01 -3.91489416e-01 -6.40689671e-01
1.04999673e+00 -2.76911616e-01 1.60138294e-01 -5.53197205e-01
-9.71458554e-01 2.10387617e-01 -2.57578135e-01 5.53500772e-01
7.45823979e-01 -1.38091850e+00 -8.72209013e-01 5.54409683e-01
1.90963581e-01 -4.50673640e-01 7.27208853e-01 7.19144523e-01
-1.02268465e-01 4.93761718e-01 -7.95874465e-03 -7.04897881e-01
-1.91767561e+00 5.10228992e-01 6.17178380e-01 -3.80149364e-01
-1.12023927e-01 9.63595152e-01 2.63658613e-02 -5.62261343e-01
9.71093178e-01 -2.21312091e-01 -4.78374004e-01 1.44659162e-01
8.29616129e-01 -1.91555977e-01 1.71491250e-01 -1.01489627e+00
-5.87367415e-01 3.95605654e-01 -2.17316076e-01 -7.06516266e-01
1.23717988e+00 -3.37863117e-01 1.77306533e-02 6.14730537e-01
1.15069711e+00 1.05968416e-01 -6.92989111e-01 -4.38052982e-01
4.15152013e-02 -3.95672210e-02 3.01113039e-01 -8.00597548e-01
-9.22859371e-01 1.19884479e+00 1.27438366e+00 -1.59603670e-01
1.02760196e+00 -4.98822272e-01 6.54168665e-01 2.14010760e-01
-9.86023545e-02 -9.29998040e-01 -2.81016886e-01 5.32517791e-01
8.63188744e-01 -1.29708052e+00 -5.16177535e-01 2.16642246e-01
-3.87802899e-01 1.21099496e+00 5.91737449e-01 5.19019067e-01
9.55061138e-01 -7.33584911e-02 4.34323221e-01 9.88537297e-02
-5.76109827e-01 -8.14971104e-02 4.71509725e-01 6.05293751e-01
6.38004959e-01 -8.63050371e-02 1.30818207e-02 6.96096718e-01
-1.92466453e-01 2.30203103e-02 -6.95045218e-02 7.62144446e-01
-4.10480022e-01 -1.48251784e+00 -7.12515116e-01 4.90206257e-02
-3.63876194e-01 -1.11177258e-01 -3.49502891e-01 2.15094700e-01
3.21000069e-02 1.42450273e+00 -1.34345636e-01 -5.73378980e-01
2.33340755e-01 7.31692076e-01 1.29612252e-01 -6.74664080e-01
-8.99558842e-01 -6.22364692e-02 2.24152599e-02 -1.17900439e-01
-2.65402496e-01 -5.60223520e-01 -1.18967700e+00 -2.60155108e-02
-5.37551582e-01 4.18157071e-01 1.25293672e+00 8.91329050e-01
4.10142303e-01 6.20117962e-01 1.02269673e+00 -6.97951436e-01
-7.65498936e-01 -1.26401341e+00 -3.54344100e-01 1.56766638e-01
7.98570275e-01 -2.95542657e-01 -4.26478326e-01 7.50895888e-02]
|
[14.413080215454102, 6.106123924255371]
|
76f66437-721a-4640-b4be-a74cfe09b752
|
playing-hard-exploration-games-by-watching
|
1805.11592
| null |
http://arxiv.org/abs/1805.11592v2
|
http://arxiv.org/pdf/1805.11592v2.pdf
|
Playing hard exploration games by watching YouTube
|
Deep reinforcement learning methods traditionally struggle with tasks where
environment rewards are particularly sparse. One successful method of guiding
exploration in these domains is to imitate trajectories provided by a human
demonstrator. However, these demonstrations are typically collected under
artificial conditions, i.e. with access to the agent's exact environment setup
and the demonstrator's action and reward trajectories. Here we propose a
two-stage method that overcomes these limitations by relying on noisy,
unaligned footage without access to such data. First, we learn to map unaligned
videos from multiple sources to a common representation using self-supervised
objectives constructed over both time and modality (i.e. vision and sound).
Second, we embed a single YouTube video in this representation to construct a
reward function that encourages an agent to imitate human gameplay. This method
of one-shot imitation allows our agent to convincingly exceed human-level
performance on the infamously hard exploration games Montezuma's Revenge,
Pitfall! and Private Eye for the first time, even if the agent is not presented
with any environment rewards.
|
['Ziyu Wang', 'Nando de Freitas', 'David Budden', 'Yusuf Aytar', 'Tom Le Paine', 'Tobias Pfaff']
|
2018-05-29
|
playing-hard-exploration-games-by-watching-1
|
http://papers.nips.cc/paper/7557-playing-hard-exploration-games-by-watching-youtube
|
http://papers.nips.cc/paper/7557-playing-hard-exploration-games-by-watching-youtube.pdf
|
neurips-2018-12
|
['montezumas-revenge']
|
['playing-games']
|
[ 4.05531339e-02 1.38864100e-01 -2.86276005e-02 2.80256439e-02
-7.24556744e-01 -7.89009869e-01 7.00304389e-01 -9.22791213e-02
-1.03924191e+00 1.00984800e+00 -1.44840479e-01 6.16620742e-02
4.49314853e-03 -3.53241444e-01 -9.72662389e-01 -5.62308729e-01
-5.38649261e-01 4.39431250e-01 -3.04738302e-02 -2.06463158e-01
2.62586981e-01 3.67019325e-01 -1.69859684e+00 -3.80139142e-01
6.95060492e-01 7.44193792e-01 6.49655521e-01 1.15978980e+00
6.22120678e-01 1.10832965e+00 -7.30014741e-01 3.55356769e-03
5.61965287e-01 -5.69412768e-01 -4.74936098e-01 1.17778800e-01
-2.60870047e-02 -1.09523380e+00 -5.80015361e-01 1.07404613e+00
2.51108855e-01 5.85834146e-01 3.36951077e-01 -1.56120944e+00
-1.34837389e-01 4.20396894e-01 -3.29008937e-01 1.66322425e-01
6.81499541e-01 9.75550890e-01 8.35216582e-01 -2.65515685e-01
1.00593638e+00 9.87371743e-01 1.83912978e-01 7.70481288e-01
-1.19032097e+00 -5.43058813e-01 6.18032776e-02 1.68329373e-01
-8.60792875e-01 -3.49850476e-01 6.60203099e-01 -3.85174632e-01
8.39307904e-01 -3.82309221e-02 1.05995059e+00 1.66068411e+00
-2.24596456e-01 8.56646895e-01 1.04219341e+00 -1.71527088e-01
5.98130047e-01 1.53562903e-01 -7.14374781e-01 6.43572032e-01
-2.37888902e-01 7.62460649e-01 -7.74625003e-01 -1.07159384e-01
1.03350878e+00 1.37071544e-02 -4.34187323e-01 -7.23368466e-01
-1.58719456e+00 7.15626001e-01 5.09885252e-01 4.67810035e-02
-7.18161583e-01 4.57405210e-01 2.19243601e-01 6.53529704e-01
-2.70157993e-01 1.00689960e+00 -1.59604356e-01 -8.91060650e-01
-5.28904557e-01 6.08445287e-01 6.90285385e-01 9.37928319e-01
5.53274572e-01 2.98088878e-01 2.13845402e-01 2.83022642e-01
1.55093357e-01 4.68851984e-01 5.19060373e-01 -1.43888295e+00
4.46003944e-01 8.90554115e-02 7.46992588e-01 -7.91082978e-01
-1.16249613e-01 -5.20738997e-02 -1.07255720e-01 1.09512353e+00
6.62080050e-01 -5.42229474e-01 -6.02281392e-01 1.97127008e+00
4.48164880e-01 1.73517108e-01 2.90052861e-01 1.34406292e+00
3.84089798e-01 4.93158042e-01 5.19631878e-02 2.81699952e-02
8.64349008e-01 -1.02717590e+00 -3.53973567e-01 -3.47706884e-01
4.90267098e-01 -2.96709150e-01 1.23516893e+00 5.06747603e-01
-1.08821237e+00 -4.36349362e-01 -1.09005594e+00 3.86440039e-01
-4.57757078e-02 -3.92668955e-02 4.78712052e-01 1.33891970e-01
-7.06164300e-01 1.02502120e+00 -1.16772461e+00 -4.48749512e-01
3.38495046e-01 3.58672082e-01 -5.98510981e-01 1.48124292e-01
-9.69921291e-01 9.56502795e-01 4.38558489e-01 -7.21447542e-02
-1.89577770e+00 -1.99150458e-01 -9.96841967e-01 -4.04251926e-02
7.15367615e-01 -3.02204400e-01 1.52428126e+00 -1.14892614e+00
-1.73628008e+00 6.24733329e-01 4.84755605e-01 -6.28538787e-01
9.25946832e-01 -2.61224121e-01 9.18073952e-02 4.24852282e-01
2.00145423e-01 1.05354762e+00 9.94465947e-01 -1.22686923e+00
-6.87777042e-01 -2.05964208e-01 6.10137761e-01 7.17826605e-01
-1.18790478e-01 -1.46293879e-01 -7.49629438e-02 -4.49661314e-01
-4.35789287e-01 -9.86086011e-01 -3.96398574e-01 3.02387834e-01
-3.19916576e-01 1.17764458e-01 7.68528342e-01 -3.47270042e-01
3.91363323e-01 -2.07022738e+00 5.74727297e-01 -8.89086947e-02
1.77718654e-01 6.82934970e-02 -2.02845082e-01 5.17737150e-01
2.17614859e-01 -3.09901386e-01 -4.69454303e-02 -5.41231751e-01
3.25978994e-01 2.95228779e-01 -1.47036031e-01 5.68985939e-01
-8.31467509e-02 8.59518170e-01 -1.49420798e+00 -2.70067811e-01
1.75973192e-01 2.07670152e-01 -6.05629981e-01 7.31723547e-01
-4.92050558e-01 1.04611671e+00 -5.32272995e-01 5.33486426e-01
5.53189106e-02 7.49359652e-02 6.34498745e-02 4.12935525e-01
-1.57354727e-01 1.24817565e-01 -1.10469854e+00 2.13844895e+00
-2.54579842e-01 6.92187786e-01 4.40509409e-01 -8.50428939e-01
5.84593832e-01 2.91392475e-01 4.47324008e-01 -5.30887485e-01
1.00540429e-01 3.34838480e-01 2.62316614e-02 -8.31583500e-01
6.35471523e-01 -9.84794647e-02 -9.68768001e-02 7.22297609e-01
2.07013249e-01 -4.93697077e-01 1.96813345e-01 3.08390647e-01
1.41061640e+00 6.76021218e-01 9.11457241e-02 2.90393233e-01
-7.58135542e-02 3.28440070e-01 4.30791467e-01 1.03120875e+00
-4.39343125e-01 3.93708080e-01 5.45387626e-01 -3.16336066e-01
-1.29105330e+00 -1.04699075e+00 5.51961541e-01 1.02909923e+00
3.07413280e-01 -2.33768210e-01 -7.91061282e-01 -6.79328918e-01
-6.69202656e-02 6.52662754e-01 -6.99722528e-01 -9.20019150e-02
-5.79484701e-01 1.12748340e-01 3.92589509e-01 3.60348940e-01
5.39449215e-01 -1.64107883e+00 -1.64954567e+00 1.42730832e-01
2.84245312e-02 -9.57180977e-01 -4.14431334e-01 3.64734888e-01
-4.84000772e-01 -9.38259721e-01 -7.88563013e-01 -4.81819034e-01
5.39189100e-01 2.61982400e-02 7.50755489e-01 -9.55014229e-02
-3.44669074e-01 8.42339218e-01 -3.68648618e-01 -1.78613499e-01
-3.66198242e-01 -2.96990097e-01 3.66727084e-01 -3.85678768e-01
-9.85404756e-03 -8.07252705e-01 -5.62399566e-01 1.66208580e-01
-7.22836673e-01 -2.54049245e-02 4.17775244e-01 9.57213461e-01
2.21414730e-01 -5.62656969e-02 3.74880642e-01 -1.45290732e-01
7.10399985e-01 -5.35855293e-01 -8.91088247e-01 -2.74921954e-01
-1.11397877e-01 -1.43554285e-01 5.87360740e-01 -8.62831652e-01
-6.83452249e-01 1.67894900e-01 3.73096257e-01 -7.88024902e-01
-3.96773636e-01 2.41474688e-01 1.11633204e-01 -9.55053121e-02
8.18697691e-01 1.60591125e-01 2.64059961e-01 -9.85091031e-02
2.78591961e-01 4.14634526e-01 9.08129513e-01 -7.29454756e-01
8.49273682e-01 3.57456088e-01 -2.83928901e-01 -6.35354757e-01
-1.72876030e-01 -8.89777169e-02 -4.11620229e-01 -6.18048131e-01
7.97372758e-01 -8.27229559e-01 -1.29524076e+00 4.65821594e-01
-9.41899478e-01 -7.83247173e-01 -6.25561118e-01 8.89147818e-01
-1.16656494e+00 2.31543295e-02 -3.56842637e-01 -9.86170471e-01
3.01111251e-01 -1.37054861e+00 7.90421128e-01 4.93939787e-01
-2.25766584e-01 -5.64616263e-01 1.41343310e-01 1.94147557e-01
3.07212889e-01 4.69199747e-01 1.17832609e-01 -5.06808698e-01
-8.64915669e-01 -1.39478862e-01 2.16450974e-01 1.47881731e-01
1.17106639e-01 -4.10750180e-01 -7.19432056e-01 -4.02796805e-01
7.47683942e-02 -1.07714546e+00 3.30802083e-01 1.18577965e-02
7.75569916e-01 -3.77854705e-01 -6.49501160e-02 5.03629625e-01
1.05622399e+00 4.82330859e-01 2.17284858e-01 7.68500865e-01
2.93223977e-01 6.20857120e-01 7.73803055e-01 6.25154316e-01
2.00459704e-01 6.91979945e-01 1.04474521e+00 1.57051712e-01
2.39998162e-01 -7.58941710e-01 5.68793535e-01 -8.44726432e-03
-1.81957006e-01 -1.05138227e-01 -4.11885589e-01 5.71224213e-01
-1.98028934e+00 -1.16041720e+00 6.55650377e-01 2.23725724e+00
6.27135456e-01 2.92700469e-01 4.22436327e-01 -3.11457336e-01
4.27355945e-01 2.99347155e-02 -1.20472634e+00 -1.70863807e-01
1.66767150e-01 -3.03092897e-01 3.95092785e-01 3.87517095e-01
-9.75283504e-01 8.52101862e-01 5.66914988e+00 4.08495396e-01
-8.93131614e-01 -2.31851608e-01 1.68496758e-01 -7.10090399e-01
-4.16811071e-02 -5.22956252e-02 -1.93710268e-01 5.89071333e-01
7.31988490e-01 -1.55515164e-01 1.06238127e+00 1.02961719e+00
3.20186496e-01 -6.54052913e-01 -1.42621517e+00 1.02597845e+00
-9.87513363e-02 -1.08526003e+00 -8.01985621e-01 1.97816476e-01
4.57138449e-01 1.55952722e-01 2.39008442e-01 5.57849944e-01
8.39740396e-01 -1.22952914e+00 9.41181898e-01 3.90602559e-01
7.09385395e-01 -4.88707960e-01 3.17541033e-01 8.41027915e-01
-6.91396534e-01 -3.33510280e-01 5.69086820e-02 -4.26674634e-01
3.10450137e-01 -5.21424115e-01 -8.71428728e-01 4.01222904e-04
6.92874432e-01 4.64132011e-01 5.83201386e-02 9.82264578e-01
-4.37209964e-01 2.58559704e-01 -5.75047314e-01 -3.59634459e-01
6.77835941e-01 -2.49309868e-01 1.00575721e+00 5.57238221e-01
3.85209829e-01 2.87051708e-01 3.91755760e-01 1.06430471e+00
-5.54656377e-03 -4.22304869e-01 -1.00115991e+00 -2.79964626e-01
3.19301873e-01 1.05197811e+00 -4.93317157e-01 -1.69677719e-01
4.58545350e-02 1.07573366e+00 5.55471122e-01 4.83413607e-01
-7.99318492e-01 -4.05561268e-01 7.12713420e-01 -2.04269573e-01
2.78676093e-01 -4.10034120e-01 3.39324743e-01 -1.08053219e+00
1.33380264e-01 -1.11577809e+00 9.79458634e-03 -1.10420704e+00
-6.68939769e-01 6.28653944e-01 4.31886017e-02 -1.37596238e+00
-6.77644074e-01 -1.63073108e-01 -6.11534655e-01 6.57132864e-01
-1.24965274e+00 -8.08265746e-01 -2.61059254e-01 6.31535590e-01
6.66420460e-01 -4.24975812e-01 7.15854704e-01 -7.83913061e-02
-3.69937301e-01 3.13737631e-01 5.20049920e-03 -9.35379881e-03
5.30638576e-01 -1.40298760e+00 2.16222797e-02 5.87930501e-01
1.85374752e-01 3.01190346e-01 1.13949430e+00 -5.08658469e-01
-1.47068882e+00 -3.76306593e-01 -4.30772342e-02 -3.00877780e-01
9.86481369e-01 -3.92780334e-01 -6.29752159e-01 8.17169845e-01
1.79761022e-01 1.78765692e-02 1.16942227e-01 -3.60696882e-01
4.40808432e-03 1.99209943e-01 -1.34490132e+00 9.08581734e-01
8.74287784e-01 -5.22791803e-01 -7.36629665e-01 3.58273268e-01
6.09800637e-01 -7.30046391e-01 -4.14148629e-01 -4.86465134e-02
6.02454901e-01 -9.16094184e-01 6.77472174e-01 -9.02532518e-01
6.82313323e-01 -3.38226587e-01 -1.06026061e-01 -1.64274192e+00
3.65502328e-01 -1.26228118e+00 8.00671056e-02 5.92992961e-01
1.17456317e-01 -4.30250794e-01 9.80363250e-01 6.39634669e-01
3.23656648e-02 -4.49024320e-01 -1.10096967e+00 -9.30851519e-01
-1.83033153e-01 -2.95298994e-01 4.05677050e-01 6.03948891e-01
4.54616934e-01 2.24939063e-02 -8.30438554e-01 -2.26223972e-02
8.60348642e-01 -2.41155922e-02 1.17210090e+00 -6.58907712e-01
-5.36926508e-01 -2.49465495e-01 -3.72042418e-01 -1.26093686e+00
3.46964002e-01 -4.80737776e-01 5.67107081e-01 -9.28462505e-01
3.37346569e-02 -4.74037558e-01 4.85970266e-02 2.74710178e-01
9.77202281e-02 -1.52914956e-01 4.39099669e-01 3.37881923e-01
-7.93008506e-01 8.34386170e-01 1.47534263e+00 -9.84997116e-03
-3.24431777e-01 -5.07011153e-02 -2.98600137e-01 7.77327359e-01
8.26716244e-01 -5.13677776e-01 -6.10744059e-01 -4.75543857e-01
-2.42585633e-02 6.13939464e-01 7.88768172e-01 -1.10736489e+00
5.67692935e-01 -4.12887841e-01 8.37630481e-02 -2.89489552e-02
7.93742776e-01 -8.14533055e-01 1.75669536e-01 5.41028798e-01
-7.34394550e-01 -2.22284230e-03 1.49878994e-01 1.02669680e+00
-7.72554427e-03 -4.13564444e-01 4.66221869e-01 -4.60222125e-01
-8.16657484e-01 8.05952474e-02 -5.94748378e-01 -2.00460963e-02
1.31081700e+00 -3.17153454e-01 -2.47232303e-01 -8.33913445e-01
-8.37413847e-01 6.27120733e-01 7.38219023e-01 2.69531876e-01
7.09829569e-01 -1.21806872e+00 -5.68966687e-01 2.66431719e-01
-1.22682959e-01 4.57668975e-02 3.34096134e-01 5.50115883e-01
-4.14947540e-01 -4.42937994e-03 -6.25810266e-01 -5.06359518e-01
-1.12815988e+00 6.47499979e-01 3.55256945e-01 -1.12467725e-02
-1.02372038e+00 8.67251575e-01 7.20421672e-02 -5.09102285e-01
6.05002463e-01 1.82402190e-02 -6.34001493e-02 -3.05155456e-01
4.07070160e-01 2.01067612e-01 -4.76335168e-01 -5.04877567e-01
2.62407884e-02 1.05661884e-01 -6.17858320e-02 -7.06542552e-01
1.42991400e+00 -3.66718173e-02 4.33700651e-01 4.91318375e-01
1.02734065e+00 -2.74548113e-01 -2.24638438e+00 -6.97964057e-02
-4.46852058e-01 -8.22204709e-01 -1.10117048e-01 -7.32751906e-01
-7.55229473e-01 5.71265638e-01 3.46387565e-01 1.42085299e-01
7.93331325e-01 -1.58738658e-01 7.36253500e-01 7.81508982e-01
7.64056981e-01 -1.29017341e+00 7.42160618e-01 2.08248302e-01
9.72823143e-01 -1.58464479e+00 -3.17134917e-01 5.64703405e-01
-1.01731896e+00 8.94330502e-01 7.30378509e-01 -2.75826097e-01
1.09764554e-01 1.09652266e-01 -3.31053957e-02 -1.99008897e-01
-8.13121438e-01 -1.81991756e-01 -3.28818619e-01 9.72434103e-01
-5.49505413e-01 1.21786252e-01 4.24372017e-01 2.00870752e-01
-4.46240872e-01 -5.27039170e-02 8.84744048e-01 1.18287015e+00
-6.20703161e-01 -6.70298100e-01 -3.85963440e-01 1.35811165e-01
-1.69608101e-01 3.34646791e-01 -2.22079024e-01 8.41910899e-01
-1.78498179e-01 8.88616443e-01 -6.02555200e-02 -2.78082371e-01
3.61650139e-01 -4.41475123e-01 5.18594444e-01 -5.52337229e-01
-4.92446929e-01 -1.98990963e-02 -5.83676547e-02 -8.33720446e-01
-2.52850205e-01 -9.16493058e-01 -1.11291468e+00 -1.84026346e-01
3.23514134e-04 2.93080717e-01 7.81312585e-01 7.26371348e-01
-6.19457513e-02 3.21080923e-01 7.55117178e-01 -1.48835731e+00
-9.05321777e-01 -7.10822105e-01 -7.42983103e-01 3.56684774e-01
8.16813171e-01 -8.32064569e-01 -6.45043731e-01 -9.47429389e-02]
|
[4.273669242858887, 1.3595255613327026]
|
411678b7-265e-4f25-988a-b90de2d2da6b
|
censnet-convolution-with-edge-node-switching
| null | null |
https://doi.org/10.24963/ijcai.2019/369
|
https://www.ijcai.org/proceedings/2019/0369.pdf
|
CensNet: Convolution with Edge-Node Switching in Graph Neural Networks
|
In this paper, we present CensNet, Convolution with Edge-Node Switching graph neural network, for semi-supervised classification and regression in graph-structured data with both node and edge features. CensNet is a general graph embedding framework, which embeds both nodes and edges to a latent feature space. By using line graph of the original undirected graph, the role of nodes and edges are switched, and two novel graph convolution operations are proposed for feature propagation. Experimental results on real-world academic citation networks and quantum chemistry graphs show that our approach has achieved or matched the state-of-the-art performance.
|
['Sheng Li', 'Pengsheng Ji', 'Xiaodong Jiang']
|
2019-08-10
| null | null | null |
proceedings-of-the-twenty-eighth
|
['graph-regression']
|
['graphs']
|
[ 2.56355740e-02 4.70376819e-01 -2.78983802e-01 -3.02422374e-01
5.08003592e-01 -4.61971641e-01 7.45397866e-01 3.79045606e-01
7.52926618e-02 7.07460225e-01 -1.64403260e-01 -6.49694026e-01
-3.16996783e-01 -1.23480392e+00 -3.71471822e-01 -5.81279993e-01
-8.12194705e-01 3.64496022e-01 5.97141162e-02 -3.03034820e-02
1.28572807e-01 7.49085844e-01 -7.47053444e-01 -7.58082494e-02
6.05304956e-01 6.47235990e-01 -2.95731127e-01 6.11631751e-01
-3.45479429e-01 8.47597957e-01 -2.70845771e-01 -3.59001517e-01
1.38518333e-01 -1.47496268e-01 -7.54413605e-01 -1.07584491e-01
2.80656517e-01 2.41512418e-01 -1.44245839e+00 1.24382174e+00
3.24188083e-01 -2.47260854e-02 8.80581021e-01 -1.65977752e+00
-1.47575867e+00 8.93671036e-01 -1.88320786e-01 1.76028803e-01
5.27447045e-01 -1.28998533e-02 1.39200878e+00 -6.49515688e-01
8.26180518e-01 1.32404578e+00 7.67874181e-01 1.89772293e-01
-1.34492874e+00 -6.46393538e-01 1.92939356e-01 2.84250081e-01
-1.29658377e+00 1.67249531e-01 1.30526674e+00 -6.41558230e-01
1.33229601e+00 -6.60541281e-02 9.20376956e-01 1.04862976e+00
7.85429299e-01 2.28012145e-01 8.46222460e-01 -3.12043637e-01
2.59970352e-02 -2.45416015e-02 7.88170338e-01 1.31488323e+00
4.77559388e-01 3.20768058e-01 -3.11774969e-01 -4.79295194e-01
8.23953688e-01 3.37575078e-01 -2.10104853e-01 -6.05935872e-01
-1.08645046e+00 1.22307312e+00 1.04869962e+00 3.20708692e-01
-2.68048137e-01 4.13199216e-01 3.01481158e-01 8.27555716e-01
6.95134759e-01 3.36901635e-01 -8.11686963e-02 5.43750405e-01
-4.70834374e-01 -3.51345599e-01 1.20883071e+00 1.18761539e+00
7.77469635e-01 3.98354530e-01 -1.99442908e-01 2.16291890e-01
5.90576172e-01 1.68163359e-01 1.65986523e-01 -3.87646228e-01
1.58352271e-01 1.42043710e+00 -5.65116405e-01 -1.47562444e+00
-7.17427135e-01 -6.45448983e-01 -1.29386020e+00 5.11873849e-02
-2.18141213e-01 2.95499377e-02 -1.21290314e+00 1.19805980e+00
-1.53253158e-03 5.68890631e-01 1.01894312e-01 4.65856194e-01
1.57911634e+00 4.79854286e-01 -5.19304723e-02 4.40023793e-03
1.02112257e+00 -1.19282389e+00 -9.29887295e-01 4.01672907e-02
6.05615318e-01 -1.86156884e-01 4.30239648e-01 -2.32214808e-01
-7.13935494e-01 -4.26729143e-01 -1.16328895e+00 4.31664288e-03
-1.09821761e+00 -3.15404326e-01 1.28438389e+00 3.99208307e-01
-1.28870118e+00 1.19791579e+00 -7.51171112e-01 -2.32103869e-01
4.05217141e-01 6.99953437e-01 -8.46891582e-01 -9.51490104e-02
-1.51531422e+00 3.36680025e-01 4.16999578e-01 1.76601157e-01
-6.74756467e-01 -4.49559182e-01 -1.37822127e+00 3.88618737e-01
1.02520332e-01 -7.28331923e-01 3.55994254e-01 -5.66920400e-01
-1.49209154e+00 6.73126578e-01 2.07173869e-01 -3.52111906e-01
7.17743933e-02 4.83910352e-01 -9.43340421e-01 6.61627948e-02
-8.34770650e-02 -2.77380971e-03 7.19794393e-01 -6.70671046e-01
2.57085413e-01 -4.52193975e-01 6.55599963e-03 -1.53065130e-01
-3.77098948e-01 -2.91965902e-01 -2.59115279e-01 -4.10853118e-01
3.73993695e-01 -9.14048254e-01 -3.25257927e-01 -6.04390725e-02
-6.34767473e-01 -7.35504746e-01 1.08456957e+00 -4.00231570e-01
1.24339330e+00 -2.05649877e+00 3.37078691e-01 6.17077351e-01
1.24784410e+00 3.64532880e-02 -2.26318941e-01 8.14988971e-01
-5.07432222e-01 2.98349619e-01 -1.82007045e-01 1.10158138e-02
1.52066965e-02 1.72098473e-01 1.40650541e-01 9.17287171e-01
1.96041539e-01 1.38111448e+00 -1.18571520e+00 -2.11908057e-01
3.09670776e-01 6.23141050e-01 -2.94155806e-01 2.35651419e-01
6.41146451e-02 2.18031049e-01 -6.96885347e-01 6.45966053e-01
5.86348593e-01 -9.18945909e-01 4.68561620e-01 -1.97771624e-01
3.54270369e-01 8.64473134e-02 -1.17918229e+00 1.56835580e+00
1.98221177e-01 7.80292988e-01 2.90868734e-03 -1.21739590e+00
1.20202720e+00 2.29403809e-01 4.59927112e-01 -3.63523185e-01
1.46575853e-01 -8.29814821e-02 1.15093648e-01 -2.21798763e-01
1.05478568e-02 5.10854363e-01 3.62330526e-02 1.99107811e-01
4.45589423e-01 1.42216310e-01 3.58699039e-02 6.14294946e-01
1.56925058e+00 -1.41841799e-01 2.22001687e-01 -4.09216672e-01
6.23255193e-01 -3.17417830e-01 8.76339152e-02 6.77440941e-01
-9.39562619e-02 1.07257642e-01 8.27483416e-01 -6.79863691e-01
-6.66925669e-01 -1.20725799e+00 1.76679622e-02 7.36053705e-01
-2.77624968e-02 -7.20704257e-01 -3.95590991e-01 -8.20814669e-01
3.98020297e-01 3.45222890e-01 -8.38684678e-01 -5.03377914e-01
-1.24512181e-01 -5.18172741e-01 3.11119646e-01 2.71737725e-01
2.97212243e-01 -1.19539583e+00 5.47601640e-01 4.62218940e-01
7.18253851e-01 -1.09350967e+00 -4.45967138e-01 3.25657517e-01
-7.70174265e-01 -1.36094666e+00 -1.88076749e-01 -1.22239292e+00
9.38850880e-01 1.19546019e-01 1.19681060e+00 3.46830159e-01
-5.04377306e-01 3.95599723e-01 -2.69204378e-01 1.51288837e-01
-2.16545969e-01 1.32220253e-01 -6.33269101e-02 -7.83264413e-02
3.77810121e-01 -1.01806331e+00 -4.60152090e-01 -1.47662804e-01
-6.74885929e-01 -1.70501307e-01 3.42245638e-01 1.09361267e+00
2.38213181e-01 -7.14509860e-02 4.79797363e-01 -1.45170665e+00
9.34269965e-01 -6.30834758e-01 -7.69686162e-01 2.23495826e-01
-1.18919218e+00 2.69785225e-01 8.80374849e-01 -1.79502755e-01
1.45851085e-02 -1.12768263e-01 3.95503581e-01 -6.42719448e-01
2.53269166e-01 1.00701666e+00 8.40635076e-02 -6.54109180e-01
4.43599463e-01 1.54878899e-01 7.88282752e-02 -3.42944175e-01
6.36330128e-01 7.32620358e-01 1.63969412e-01 -9.32404399e-02
8.02787721e-01 1.62519515e-01 6.13358438e-01 -6.51607633e-01
-2.04761058e-01 -3.38653177e-01 -1.00806379e+00 1.41996160e-01
7.25442708e-01 -8.22104096e-01 -8.49248588e-01 4.32644010e-01
-1.14320350e+00 7.38931028e-03 -9.27903876e-02 5.39862096e-01
-1.60674930e-01 5.06044447e-01 -9.21069443e-01 -2.12698534e-01
-5.45470059e-01 -9.02112305e-01 6.94785714e-01 1.34883940e-01
4.42479968e-01 -1.75490320e+00 1.02915011e-01 -3.51856947e-01
4.11515027e-01 6.29895985e-01 1.11385667e+00 -1.08629537e+00
-6.20594203e-01 -5.73435128e-01 -5.36175191e-01 1.23372540e-01
4.64122534e-01 5.57548925e-02 -6.21344805e-01 -7.79298604e-01
-6.84483945e-01 1.02485649e-01 8.96268606e-01 1.91291854e-01
1.20526373e+00 -1.84872717e-01 -7.90931702e-01 1.18755221e+00
1.47505665e+00 -1.56816483e-01 4.04334515e-01 -1.72248006e-01
1.23123538e+00 -5.74768595e-02 -4.76540476e-01 2.21828952e-01
4.11275744e-01 9.54937637e-02 4.22509849e-01 -3.08019638e-01
-1.88815325e-01 -2.28794292e-01 1.85231656e-01 1.36359382e+00
-1.19348280e-01 -5.58119476e-01 -7.92077720e-01 8.90717208e-02
-1.84605050e+00 -7.08743632e-01 -5.29270470e-01 1.65810323e+00
3.25310737e-01 2.86373734e-01 -2.94862747e-01 -2.20597520e-01
9.70431268e-01 5.48154056e-01 -6.12396598e-01 -4.19618100e-01
-1.31848902e-01 4.66452569e-01 8.30318213e-01 6.31081760e-01
-1.00199509e+00 1.18778265e+00 7.19440937e+00 1.12025164e-01
-1.12148976e+00 -2.34431569e-02 1.46304101e-01 6.34260952e-01
-4.76291955e-01 1.79733261e-01 -1.55543268e-01 2.00483233e-01
9.99551773e-01 -4.19683278e-01 8.49511325e-01 7.07298040e-01
-3.94329637e-01 9.17454839e-01 -1.31859863e+00 1.02560484e+00
-7.59918764e-02 -1.62032735e+00 1.73776790e-01 -1.01350050e-03
5.76331735e-01 3.89079124e-01 -2.10081831e-01 5.73008776e-01
7.64469147e-01 -1.46896982e+00 -2.43871689e-01 5.93554795e-01
9.86550987e-01 -4.69819993e-01 6.75643086e-01 -1.96940646e-01
-1.58785224e+00 3.26734573e-01 -4.59650367e-01 -3.29478860e-01
-3.84009153e-01 4.75861907e-01 -8.19669425e-01 1.08164930e+00
3.06256205e-01 1.53436649e+00 -7.59504616e-01 6.78547144e-01
-4.29656357e-01 7.99925625e-01 -1.47885848e-02 -2.52153695e-01
5.13446271e-01 -5.82126677e-01 7.28751302e-01 1.23233235e+00
-1.07332736e-01 -1.12904184e-01 4.83157009e-01 1.21093249e+00
-5.50905287e-01 -4.26345179e-03 -1.09618998e+00 -7.93944180e-01
4.74912643e-01 1.53674221e+00 -7.71008909e-01 -2.51228005e-01
-6.87551081e-01 1.12960041e+00 8.39270949e-01 8.48636508e-01
-6.42158389e-01 -1.07244205e+00 2.24185869e-01 -9.22807753e-02
2.93690264e-01 -4.06681329e-01 3.09176892e-01 -1.23229134e+00
-2.40827128e-01 -1.78574532e-01 5.54557204e-01 -5.58603704e-01
-1.73111355e+00 8.24678600e-01 -4.04423505e-01 -7.16399789e-01
2.36552328e-01 -1.09961927e+00 -9.90243971e-01 1.03836620e+00
-1.65894711e+00 -1.32295108e+00 -4.10897285e-01 8.88971269e-01
-3.48846376e-01 -6.29052699e-01 1.19428551e+00 1.19666249e-01
-5.42880177e-01 6.44134998e-01 3.84677798e-01 6.13706648e-01
2.41711602e-01 -1.57027233e+00 8.67457688e-01 2.58227199e-01
3.02101761e-01 7.83092439e-01 2.06230998e-01 -9.56636369e-01
-2.01396775e+00 -1.24937916e+00 5.70600092e-01 -9.63995531e-02
1.13356435e+00 -6.89854801e-01 -9.31958079e-01 1.16284919e+00
4.18878973e-01 8.46777797e-01 6.04442179e-01 2.64972985e-01
-4.58624065e-01 2.45594710e-01 -1.20375264e+00 4.01759177e-01
1.43250632e+00 -8.05989563e-01 -4.88263279e-01 7.24905252e-01
1.20846045e+00 -1.89205170e-01 -1.26498878e+00 3.22527468e-01
1.91334412e-01 -3.06539893e-01 9.85432744e-01 -1.21879435e+00
-2.51543447e-02 -9.71831977e-02 4.61129732e-02 -1.63218462e+00
-1.09332669e+00 -9.21253443e-01 -4.76857841e-01 7.20389068e-01
3.54401141e-01 -1.24946702e+00 8.03342700e-01 6.23717010e-02
-1.53884947e-01 -7.04581201e-01 -9.66688395e-01 -6.86401010e-01
-1.05941452e-01 2.09495291e-01 6.70204043e-01 1.50171220e+00
1.06719568e-01 8.25296521e-01 -1.27762273e-01 3.49369615e-01
7.77414560e-01 1.86564118e-01 3.71167064e-01 -1.73853254e+00
-1.16035249e-02 -4.17940080e-01 -1.33150506e+00 -5.08723736e-01
5.82920015e-01 -1.78678107e+00 -9.08427060e-01 -1.93899488e+00
-6.85312971e-02 -8.50665942e-02 -8.24005425e-01 5.20538211e-01
-7.84196034e-02 -1.74751639e-01 -1.89430505e-01 -9.37057845e-03
-5.17245114e-01 7.29403198e-01 1.39580441e+00 -5.79210460e-01
-2.18965322e-01 -3.14843088e-01 -3.73581439e-01 3.30921978e-01
5.66245735e-01 -5.66518843e-01 -3.08829814e-01 1.45129055e-01
1.57868013e-01 7.49567803e-03 3.30234319e-01 -6.84524298e-01
4.82559204e-01 1.52997121e-01 2.99794376e-01 -2.96119243e-01
4.80928682e-02 -8.77076626e-01 1.88957065e-01 6.07771158e-01
-3.04929107e-01 1.78559497e-01 -5.52892499e-02 1.03530467e+00
-1.19138688e-01 4.75422032e-02 2.89952248e-01 -1.22015521e-01
-5.31838715e-01 9.09975529e-01 8.73677433e-02 -1.40520722e-01
9.19228375e-01 -1.21215053e-01 -5.50722659e-01 -1.69267103e-01
-1.14355814e+00 4.50369626e-01 1.92073852e-01 4.97034371e-01
8.63266587e-01 -1.68346095e+00 -7.70460546e-01 5.98969102e-01
3.02532643e-01 -3.13523889e-01 -2.23081931e-01 5.29479623e-01
-5.61366320e-01 4.29621339e-01 -1.70188740e-01 -4.56024259e-01
-9.83110487e-01 8.35860014e-01 3.20779383e-01 -3.64443451e-01
-8.73516917e-01 6.50237381e-01 -1.76069602e-01 -9.48254764e-01
1.84335455e-01 -2.00951099e-01 -4.38083708e-01 -2.41292015e-01
-6.00679852e-02 2.57081330e-01 -2.25537211e-01 -4.80616301e-01
-4.69004303e-01 3.37065816e-01 -9.73009765e-02 6.33424997e-01
1.54051769e+00 3.16628963e-01 -6.99542165e-01 5.43921053e-01
1.57604349e+00 -2.47689679e-01 -6.48866117e-01 -5.23600936e-01
-1.40023932e-01 -9.27633345e-02 4.46152061e-01 -5.06487250e-01
-1.25058234e+00 6.90180004e-01 4.81007487e-01 6.41908705e-01
4.14189368e-01 1.32528981e-02 3.85072142e-01 6.78431213e-01
1.90346256e-01 -6.29405677e-01 -4.32070158e-02 6.35020494e-01
6.63518190e-01 -1.17166388e+00 2.97630429e-01 -6.11077487e-01
2.44429428e-02 1.58387673e+00 4.28125143e-01 -7.65157640e-01
1.40158236e+00 -2.01441631e-01 -5.02335608e-01 -1.03996170e+00
-5.38294494e-01 2.08255127e-01 6.52792990e-01 4.68788654e-01
5.01039684e-01 4.56412762e-01 -1.02009296e-01 4.73127604e-01
9.59520042e-02 -1.79467484e-01 4.92849767e-01 8.68496656e-01
-1.82205230e-01 -9.32186604e-01 2.82092482e-01 8.93153429e-01
-1.28209174e-01 -3.98987740e-01 -6.00807190e-01 8.73012006e-01
-5.22975802e-01 6.68885350e-01 -5.81836849e-02 -6.63246989e-01
1.56116098e-01 8.52131546e-02 4.35806155e-01 -8.57734680e-01
-4.54979122e-01 -4.75684434e-01 -4.12830450e-02 -4.98082280e-01
-2.11779073e-01 -6.19616285e-02 -1.37573683e+00 -4.67440873e-01
-6.05849504e-01 3.70521396e-01 5.78958333e-01 4.58096206e-01
7.99836397e-01 1.12626863e+00 8.83311927e-01 -4.65316951e-01
-3.95710170e-01 -1.16082346e+00 -1.13242757e+00 3.87328446e-01
3.75743061e-01 -7.88887084e-01 -6.73624516e-01 -4.02627081e-01]
|
[7.067370891571045, 6.243435382843018]
|
f2e13b17-f8e7-405b-8c3d-013d90655f63
|
cdpmsr-conditional-diffusion-probabilistic
|
2302.12831
| null |
https://arxiv.org/abs/2302.12831v1
|
https://arxiv.org/pdf/2302.12831v1.pdf
|
CDPMSR: Conditional Diffusion Probabilistic Models for Single Image Super-Resolution
|
Diffusion probabilistic models (DPM) have been widely adopted in image-to-image translation to generate high-quality images. Prior attempts at applying the DPM to image super-resolution (SR) have shown that iteratively refining a pure Gaussian noise with a conditional image using a U-Net trained on denoising at various-level noises can help obtain a satisfied high-resolution image for the low-resolution one. To further improve the performance and simplify current DPM-based super-resolution methods, we propose a simple but non-trivial DPM-based super-resolution post-process framework,i.e., cDPMSR. After applying a pre-trained SR model on the to-be-test LR image to provide the conditional input, we adapt the standard DPM to conduct conditional image generation and perform super-resolution through a deterministic iterative denoising process. Our method surpasses prior attempts on both qualitative and quantitative results and can generate more photo-realistic counterparts for the low-resolution images with various benchmark datasets including Set5, Set14, Urban100, BSD100, and Manga109. Code will be published after accepted.
|
['Yanning Zhang', 'In So Kweon', 'Yu Zhu', 'Jinqiu Sun', 'Trung X. Pham', 'Kang Zhang', 'Axi Niu']
|
2023-02-14
| null | null | null | null |
['conditional-image-generation']
|
['computer-vision']
|
[ 8.46473277e-01 5.36817238e-02 3.09082299e-01 -1.80836022e-01
-1.32945728e+00 -1.64079070e-01 6.93839073e-01 -7.04884827e-01
-1.42343998e-01 8.55500698e-01 3.21474403e-01 4.68237922e-02
-7.80881643e-02 -9.57495809e-01 -7.14421868e-01 -8.54252696e-01
3.47327769e-01 9.51134712e-02 4.67847556e-01 -2.78781384e-01
2.74124652e-01 4.62701619e-01 -1.49966311e+00 4.77036238e-01
9.25909877e-01 5.96448123e-01 6.71262383e-01 7.85967410e-01
4.95366799e-03 6.33306623e-01 -3.90596062e-01 -3.36824864e-01
2.91474134e-01 -7.85140514e-01 -8.41184437e-01 1.65866286e-01
4.09396648e-01 -3.48046482e-01 -1.13854930e-01 1.32039583e+00
6.50948644e-01 1.99316800e-01 7.06584096e-01 -4.51769263e-01
-1.31578040e+00 5.95697105e-01 -1.03711963e+00 3.64141643e-01
5.09632885e-01 3.90149131e-02 5.26037991e-01 -1.09294462e+00
9.84332681e-01 1.58827591e+00 5.35137355e-01 8.35635841e-01
-1.69198048e+00 -5.62242508e-01 -3.80450279e-01 6.70634657e-02
-1.43549216e+00 -4.60426629e-01 7.98227847e-01 -6.31081015e-02
3.71151417e-01 2.16749117e-01 5.40395863e-02 1.24169183e+00
1.76247120e-01 3.51944357e-01 1.78271735e+00 -4.42977220e-01
1.80636510e-01 6.39208630e-02 -3.32571477e-01 2.24894494e-01
1.77695788e-02 3.35461020e-01 -5.44685602e-01 5.06169461e-02
1.49699020e+00 -4.12782609e-01 -4.32014793e-01 6.71921074e-02
-1.09180105e+00 5.05913496e-01 5.18126190e-01 5.64710915e-01
-6.57781124e-01 -1.39936700e-01 -2.66257823e-01 5.94819039e-02
7.50537157e-01 3.56764287e-01 6.31471863e-03 2.01775968e-01
-1.32043552e+00 1.86962351e-01 1.40060857e-01 7.04991579e-01
7.22786069e-01 2.32562706e-01 -4.77769554e-01 1.01281953e+00
2.72130594e-02 4.29368675e-01 3.16008478e-01 -1.49342132e+00
5.04794829e-02 -1.01085566e-01 5.40089548e-01 -8.34312737e-01
1.11615457e-01 -5.40967703e-01 -1.28850651e+00 5.87915421e-01
-6.31363131e-03 1.00369275e-01 -1.24385786e+00 1.45580089e+00
2.73258358e-01 4.85039651e-01 2.79376835e-01 1.15659881e+00
8.69844615e-01 7.56448686e-01 -1.31359532e-01 -5.32566428e-01
1.04622138e+00 -7.91736186e-01 -8.54063094e-01 6.43721893e-02
-2.44008064e-01 -1.02522027e+00 1.04613018e+00 5.12643933e-01
-1.54869640e+00 -9.23287868e-01 -1.12709117e+00 -9.64068621e-02
2.80377686e-01 -1.02804162e-01 1.56005025e-02 4.78717893e-01
-1.47540462e+00 9.09216106e-01 -5.31323791e-01 -2.23550662e-01
6.72660708e-01 -2.41083071e-01 -3.86399060e-01 -5.75786471e-01
-1.14170206e+00 1.02233350e+00 3.78182083e-01 -8.46801233e-03
-1.17686522e+00 -6.41250432e-01 -6.75828636e-01 -3.22477281e-01
1.60044163e-01 -9.13703620e-01 8.11497390e-01 -8.79230618e-01
-1.85762537e+00 7.71474659e-01 -2.18593910e-01 -4.21078414e-01
5.25947154e-01 -6.89429641e-02 -4.25578117e-01 3.34079504e-01
1.46427631e-01 7.94975579e-01 1.27736938e+00 -1.93975830e+00
-4.42223072e-01 -1.77387595e-02 -1.12421796e-01 3.49925101e-01
2.67935455e-01 3.72437775e-01 -3.82638067e-01 -6.97528362e-01
3.56012881e-01 -4.91830111e-01 -3.97963077e-01 -3.13546538e-01
-3.90559912e-01 3.28769475e-01 6.74328864e-01 -9.69318092e-01
9.21663105e-01 -1.98280406e+00 3.54651779e-01 -5.49810603e-02
1.84876293e-01 3.36299986e-01 -4.73490775e-01 -6.62425160e-02
-2.06174821e-01 2.22506940e-01 -6.91511512e-01 -5.04988372e-01
-6.06398284e-01 1.99876741e-01 -1.09619521e-01 3.54670197e-01
4.64965940e-01 8.30323160e-01 -9.17176783e-01 -6.76474929e-01
4.37129855e-01 1.10376108e+00 -3.15178603e-01 2.84615457e-01
-4.16372493e-02 1.02954328e+00 -1.26501232e-01 4.27351326e-01
1.18937659e+00 -2.32936248e-01 -2.92358965e-01 -3.98726076e-01
-1.83772191e-01 -3.17776591e-01 -1.45834780e+00 1.65420473e+00
-5.73148668e-01 3.58672172e-01 2.70193070e-01 -5.26823878e-01
9.70694482e-01 9.21407193e-02 2.86153913e-01 -9.16756570e-01
-1.81196153e-01 1.01089396e-01 -4.47146058e-01 -2.56748170e-01
6.82662129e-01 -5.08318841e-01 4.83745307e-01 1.62233546e-01
8.93884972e-02 -3.59548926e-01 1.07148677e-01 2.13091776e-01
7.69792795e-01 5.67331553e-01 5.24035841e-02 -1.24913029e-01
8.32545877e-01 -2.48239085e-01 4.96230692e-01 8.63967299e-01
1.07983025e-02 1.60594153e+00 -8.94180674e-04 7.66417384e-02
-1.53051734e+00 -1.36754251e+00 -2.49629751e-01 5.63810706e-01
1.85578734e-01 1.23333232e-02 -9.88917708e-01 -1.05792724e-01
-7.33896434e-01 9.05393660e-01 -5.32691061e-01 1.77823886e-01
-4.23878938e-01 -9.23058808e-01 2.78724045e-01 1.79280788e-01
9.44697440e-01 -1.15907073e+00 -3.33202556e-02 1.77384257e-01
-5.77901304e-01 -1.30424190e+00 -2.51257658e-01 -2.41867021e-01
-6.72022998e-01 -6.30830050e-01 -1.27752256e+00 -7.90013969e-01
6.54267848e-01 4.06234354e-01 1.05885446e+00 -2.52152979e-01
-1.72928616e-01 1.73217580e-01 -3.08522284e-01 1.32420346e-01
-8.69694412e-01 -5.78127623e-01 -5.34438528e-02 2.83814996e-01
-3.28936845e-01 -7.62167931e-01 -7.82799900e-01 3.29545379e-01
-1.25013542e+00 4.65281934e-01 7.71863580e-01 8.69593382e-01
1.17589259e+00 5.11893749e-01 4.23525333e-01 -7.73382306e-01
6.92083478e-01 -1.94918975e-01 -3.98303896e-01 4.03478779e-02
-6.70105636e-01 4.42075767e-02 4.25318122e-01 -6.12376571e-01
-1.78731871e+00 -9.56672505e-02 -3.21133226e-01 -6.11928284e-01
-1.80889100e-01 1.25205368e-01 -2.07949072e-01 -2.57506490e-01
9.55263615e-01 6.62137926e-01 -7.56432042e-02 -5.49283087e-01
7.80776501e-01 4.82691556e-01 1.02592528e+00 -3.47766042e-01
1.34833312e+00 8.39203835e-01 8.28989893e-02 -7.51019537e-01
-9.43434536e-01 -1.58608913e-01 -6.66605234e-01 -1.23450331e-01
1.18093097e+00 -1.02413058e+00 -1.26077279e-01 4.64445889e-01
-1.14573753e+00 -3.46357554e-01 -3.53759438e-01 2.62509167e-01
-7.28801072e-01 5.79855859e-01 -8.45104456e-01 -7.64243841e-01
-4.03238416e-01 -1.10715568e+00 1.09041119e+00 4.82544869e-01
1.99983165e-01 -6.45162702e-01 -1.93444937e-02 6.40676379e-01
9.59125161e-01 1.85290962e-01 5.83903372e-01 2.14528680e-01
-7.51155019e-01 3.27781439e-01 -7.41826296e-01 8.47904444e-01
9.23922658e-02 -1.27027869e-01 -9.05248404e-01 -1.83053747e-01
2.98552752e-01 -7.05843642e-02 8.94733131e-01 6.46729171e-01
9.35929716e-01 -5.35446554e-02 7.96966627e-02 7.62792289e-01
1.81107283e+00 -2.44669437e-01 1.23041868e+00 3.17946434e-01
6.88140690e-01 3.67155105e-01 6.42493367e-01 7.56272897e-02
1.11181028e-01 6.35913253e-01 2.27828309e-01 -3.30895990e-01
-8.64876270e-01 -2.16764048e-01 3.86810303e-01 4.52425182e-01
-4.65571523e-01 5.46701215e-02 -2.95107901e-01 5.11057734e-01
-1.43034971e+00 -1.16070163e+00 -2.62838244e-01 2.12917948e+00
1.18588221e+00 4.12574597e-02 -2.95928687e-01 -7.21324161e-02
9.14599121e-01 2.07516328e-01 -3.76034319e-01 3.90665606e-02
-6.63167655e-01 4.23963368e-01 3.79446268e-01 8.24621916e-01
-8.99037659e-01 1.11075139e+00 6.18992376e+00 1.34913969e+00
-8.47806215e-01 5.11453509e-01 9.58597481e-01 1.34999514e-01
-5.15677273e-01 -1.18432939e-01 -6.27007365e-01 2.03225583e-01
9.52181578e-01 -5.94482571e-02 6.82818592e-01 3.92910540e-01
5.93149185e-01 -2.76433498e-01 -5.31998158e-01 1.21729386e+00
2.24140540e-01 -1.38177931e+00 2.85040528e-01 -1.91753104e-01
1.32287991e+00 -1.47515386e-01 2.49999151e-01 1.70038529e-02
4.72958893e-01 -1.31590784e+00 4.13125962e-01 9.44949389e-01
1.10294902e+00 -7.17437088e-01 6.68312192e-01 2.23102674e-01
-9.95085776e-01 1.33190572e-01 -6.20871246e-01 5.12146711e-01
4.54330504e-01 8.55916858e-01 -2.89991379e-01 9.30107594e-01
9.62840736e-01 5.30955851e-01 -4.88031209e-01 7.35127807e-01
-3.86443853e-01 5.53633153e-01 2.62978766e-02 9.59352672e-01
-2.11486220e-01 -4.63469356e-01 9.04395878e-01 1.01578903e+00
4.74827439e-01 2.29854554e-01 -2.80835599e-01 1.51418948e+00
9.25409198e-02 -9.18814633e-03 -3.32594633e-01 3.11692983e-01
1.18350968e-01 1.52188742e+00 -7.24885345e-01 -3.44457328e-01
-2.07708463e-01 1.43960428e+00 -6.75136223e-02 7.38869071e-01
-7.57532060e-01 1.07792504e-01 2.28327349e-01 2.87950844e-01
4.22489792e-01 -1.59220710e-01 -3.67312789e-01 -1.01750159e+00
-8.88946429e-02 -9.84397471e-01 4.38159071e-02 -1.51596165e+00
-1.25241804e+00 9.65876997e-01 -9.29827336e-04 -1.32710230e+00
-1.49366453e-01 4.49287929e-02 -4.11305755e-01 1.43352640e+00
-1.82934356e+00 -1.42708945e+00 -4.32689518e-01 5.84222674e-01
6.66233122e-01 9.68712196e-02 5.10034978e-01 1.87041879e-01
-2.50276715e-01 7.50212744e-02 9.79394764e-02 -2.26730242e-01
8.17706466e-01 -1.10627997e+00 3.50757897e-01 1.35037267e+00
-9.81720835e-02 4.27152693e-01 1.06652915e+00 -8.59353662e-01
-9.61945891e-01 -1.17380536e+00 4.51311558e-01 -5.19624114e-01
3.39910895e-01 7.75021641e-03 -1.20879340e+00 2.48553902e-01
3.45976979e-01 -3.40135582e-02 -1.15930013e-01 -5.69309950e-01
-5.84142283e-02 2.64698248e-02 -1.51238108e+00 7.60721624e-01
1.15138137e+00 -4.57558215e-01 -4.79409665e-01 -3.01833614e-03
9.73936439e-01 -4.73805726e-01 -1.08779562e+00 4.64745492e-01
6.27858788e-02 -1.10401165e+00 1.36587560e+00 1.26604186e-02
8.59962165e-01 -6.79581761e-01 -2.52011478e-01 -1.33114231e+00
-6.94557607e-01 -8.03875446e-01 1.32920995e-01 1.46493983e+00
2.41400719e-01 -4.20786858e-01 5.02264678e-01 3.79646927e-01
1.57027096e-01 -3.93289268e-01 -8.68476987e-01 -6.37664378e-01
2.23254915e-02 -3.03992361e-01 3.03131521e-01 7.74500728e-01
-8.94339144e-01 2.65582085e-01 -7.08523452e-01 4.42474097e-01
1.32905173e+00 -9.78422686e-02 5.33682823e-01 -9.04204428e-01
-2.86337823e-01 -2.89968848e-01 2.02780925e-02 -8.82129848e-01
-1.96020335e-01 -6.22393250e-01 1.89572185e-01 -1.81772220e+00
4.98380393e-01 -4.31562662e-02 -3.32379699e-01 -7.97913522e-02
-3.24164867e-01 7.74083674e-01 -7.35407844e-02 4.05774057e-01
-4.75426763e-01 5.23447514e-01 1.90446353e+00 9.20140967e-02
-1.20145783e-01 -1.05143838e-01 -8.64971399e-01 5.29504955e-01
5.29870212e-01 -4.10228163e-01 -3.48946422e-01 -8.88772681e-02
4.10433635e-02 3.41436505e-01 4.61820334e-01 -9.45055127e-01
-3.07518430e-02 -1.43409789e-01 6.57598257e-01 -6.22088253e-01
4.41568851e-01 -5.22093654e-01 3.69995713e-01 4.50621061e-02
-3.80759895e-01 -4.12013650e-01 -7.03589767e-02 6.56904995e-01
-2.68762946e-01 -2.10830554e-01 1.35553789e+00 -3.51886868e-01
-7.17380881e-01 2.23393962e-01 -6.80397823e-02 -1.85442001e-01
6.35428131e-01 -3.47520292e-01 -3.24202299e-01 -5.95325589e-01
-8.31838071e-01 -2.79227555e-01 7.51239181e-01 2.97306120e-01
9.50037420e-01 -1.27426684e+00 -1.29280806e+00 7.66442493e-02
-2.56004214e-01 1.97693184e-01 7.50639439e-01 7.50660121e-01
-3.13850433e-01 -5.49461991e-02 -3.20408612e-01 -6.31409526e-01
-1.32740808e+00 6.59108937e-01 3.95571887e-01 -2.69812346e-01
-9.61324155e-01 8.10904324e-01 3.14590663e-01 -1.05483145e-01
-2.27912962e-01 7.92808235e-02 -3.65062207e-01 -5.20535827e-01
8.38558137e-01 2.88543314e-01 -3.72306257e-01 -8.48857701e-01
9.25288186e-04 7.16276646e-01 -1.31918536e-02 -5.17142355e-01
1.53807938e+00 -6.31617486e-01 -1.93995610e-01 1.23502813e-01
8.44348431e-01 7.66503159e-03 -1.56270611e+00 -4.94844466e-01
-3.43200147e-01 -5.60412705e-01 3.86160254e-01 -8.49329472e-01
-1.13618302e+00 5.21778405e-01 9.40084040e-01 -1.57278091e-01
1.44011843e+00 2.13497654e-02 6.34967566e-01 -4.05504704e-01
4.28673714e-01 -9.34729278e-01 2.83045292e-01 1.32231534e-01
1.26569259e+00 -1.26581967e+00 2.23949835e-01 -4.65332091e-01
-6.52997196e-01 8.69278669e-01 3.93444628e-01 -7.78254345e-02
2.85814464e-01 3.40224892e-01 8.88625234e-02 1.56839326e-01
-4.62217510e-01 -3.57833862e-01 3.46509188e-01 9.64827061e-01
2.22136304e-01 -1.48223609e-01 -2.69110382e-01 3.88062835e-01
1.01717994e-01 3.19769681e-01 8.65127742e-01 3.81467342e-01
-3.54743093e-01 -9.54773188e-01 -7.84480929e-01 5.03128804e-02
-5.69999754e-01 -4.34105873e-01 8.73630270e-02 3.44111294e-01
4.35888022e-02 1.09109652e+00 -2.64903814e-01 -3.08960557e-01
8.62741023e-02 -2.27332041e-01 6.54158592e-01 -3.92846346e-01
5.15635349e-02 3.26202065e-01 -4.80779819e-02 -6.49396241e-01
-8.54312539e-01 -4.76765066e-01 -1.00573409e+00 -1.42186865e-01
-1.25625640e-01 -2.19673261e-01 6.12292707e-01 6.98317885e-01
3.71031731e-01 6.60286963e-01 5.31375229e-01 -1.23762822e+00
-3.48243028e-01 -1.21718431e+00 -5.46291232e-01 5.79774320e-01
1.61282346e-01 -4.46998566e-01 -4.64352667e-01 2.42696881e-01]
|
[11.16444206237793, -2.052915096282959]
|
059d3600-68cb-4410-869f-f247c0f59b21
|
learnable-triangulation-for-deep-learning
|
2109.11844
| null |
https://arxiv.org/abs/2109.11844v1
|
https://arxiv.org/pdf/2109.11844v1.pdf
|
Learnable Triangulation for Deep Learning-based 3D Reconstruction of Objects of Arbitrary Topology from Single RGB Images
|
We propose a novel deep reinforcement learning-based approach for 3D object reconstruction from monocular images. Prior works that use mesh representations are template based. Thus, they are limited to the reconstruction of objects that have the same topology as the template. Methods that use volumetric grids as intermediate representations are computationally expensive, which limits their application in real-time scenarios. In this paper, we propose a novel end-to-end method that reconstructs 3D objects of arbitrary topology from a monocular image. It is composed of of (1) a Vertex Generation Network (VGN), which predicts the initial 3D locations of the object's vertices from an input RGB image, (2) a differentiable triangulation layer, which learns in a non-supervised manner, using a novel reinforcement learning algorithm, the best triangulation of the object's vertices, and finally, (3) a hierarchical mesh refinement network that uses graph convolutions to refine the initial mesh. Our key contribution is the learnable triangulation process, which recovers in an unsupervised manner the topology of the input shape. Our experiments on ShapeNet and Pix3D benchmarks show that the proposed method outperforms the state-of-the-art in terms of visual quality, reconstruction accuracy, and computational time.
|
['Hamid Laga', 'Aladine Chetouani', 'Hedi Tabia', 'Tarek Ben Charrada']
|
2021-09-24
| null | null | null | null |
['3d-object-reconstruction', 'object-reconstruction']
|
['computer-vision', 'computer-vision']
|
[-1.39837101e-01 1.42832920e-01 3.42577994e-01 -1.86338633e-01
-4.64080691e-01 -4.30419207e-01 3.95137399e-01 4.35530916e-02
-8.99636894e-02 5.36516607e-01 -2.91208714e-01 -1.92913994e-01
1.74167618e-01 -1.33838534e+00 -1.34000409e+00 -6.02525949e-01
2.69665897e-01 1.06798971e+00 3.02522957e-01 7.80039802e-02
1.58202499e-01 1.04066920e+00 -1.55802619e+00 8.39279890e-02
5.69490671e-01 1.44792652e+00 1.35456041e-01 4.24779505e-01
-3.36018413e-01 5.01371801e-01 -1.02012955e-01 -1.75846502e-01
4.86446112e-01 -1.00032538e-01 -8.10458302e-01 4.67877686e-01
5.38477063e-01 -4.64213997e-01 -4.84136254e-01 8.21527004e-01
3.63230616e-01 -3.69138718e-02 7.18174219e-01 -9.69114959e-01
-5.46502292e-01 1.59122005e-01 -4.51219141e-01 -5.42983770e-01
3.22081566e-01 2.13703528e-01 6.39956772e-01 -1.18601167e+00
1.00892472e+00 1.39358377e+00 6.81111872e-01 3.40287507e-01
-1.41382813e+00 -5.00337005e-01 7.60207251e-02 -2.97871251e-02
-1.49767518e+00 -1.51610345e-01 1.01998627e+00 -5.52891612e-01
9.47697997e-01 -1.47178680e-01 1.04999852e+00 6.51355267e-01
2.01098427e-01 4.92705762e-01 1.05750835e+00 -2.36895427e-01
4.95432049e-01 -3.21178854e-01 -5.61278045e-01 1.04542673e+00
-2.94993855e-02 2.43667483e-01 -3.06418777e-01 -1.19996540e-01
1.42216313e+00 7.17047304e-02 -9.31871384e-02 -1.17732882e+00
-1.07366836e+00 7.37987280e-01 1.03518927e+00 -3.47566791e-02
-6.75090194e-01 4.86660719e-01 -2.15692408e-02 1.31474286e-01
5.76593637e-01 1.91020787e-01 -3.94111395e-01 2.75287330e-01
-7.75481701e-01 3.37482989e-01 7.64790714e-01 1.03645420e+00
1.11851776e+00 1.10511035e-01 2.21451640e-01 5.30026078e-01
4.92966443e-01 5.65867722e-01 -2.56487876e-01 -1.10695350e+00
1.52084768e-01 1.05847406e+00 3.17556947e-01 -8.59519780e-01
-4.88554001e-01 -3.00734609e-01 -8.57745111e-01 7.38629162e-01
1.10695764e-01 1.76631659e-01 -1.13368440e+00 1.34352958e+00
8.24686766e-01 2.82576442e-01 -1.87713504e-01 1.08365965e+00
1.07727802e+00 4.97588694e-01 -3.16517621e-01 1.70884699e-01
9.65821981e-01 -8.57141912e-01 -1.98285148e-01 8.20841789e-02
2.04977151e-02 -7.22361028e-01 7.06899524e-01 3.69132489e-01
-1.52831292e+00 -5.49629271e-01 -8.53084922e-01 -2.72152454e-01
-2.43065506e-01 -4.34839651e-02 3.34397376e-01 1.40391216e-01
-1.23865223e+00 8.30652356e-01 -8.65763485e-01 -1.57653987e-02
7.45567679e-01 4.05255347e-01 -3.29138428e-01 -2.50311702e-01
-5.01283705e-01 6.96480095e-01 2.17263192e-01 -5.37009351e-02
-1.07851923e+00 -8.39665413e-01 -1.08867240e+00 -4.48176451e-02
4.99199957e-01 -1.19590437e+00 1.02886057e+00 -8.55584860e-01
-1.68836677e+00 9.50439513e-01 -3.93429361e-02 -2.68496156e-01
6.37555540e-01 2.35616863e-01 4.49430376e-01 2.97336847e-01
1.11355877e-03 8.83436620e-01 1.07774246e+00 -1.71495247e+00
-5.08637846e-01 -4.51403141e-01 8.12377185e-02 3.14992219e-01
5.78865230e-01 -7.48340607e-01 -7.03066766e-01 -4.08408850e-01
5.39647579e-01 -7.71266341e-01 -2.49951601e-01 6.89575970e-01
-3.65833700e-01 -3.66477787e-01 7.41750419e-01 -5.83446443e-01
2.97678858e-01 -1.98784721e+00 6.40457690e-01 4.11657155e-01
2.91982651e-01 -1.24564320e-01 4.66909520e-02 2.40784183e-01
2.25927964e-01 -7.77916238e-02 -3.21642220e-01 -8.41032922e-01
-2.57357918e-02 2.20526993e-01 -8.95333514e-02 6.63564324e-01
2.07229137e-01 1.10647261e+00 -1.02277017e+00 -3.72791618e-01
7.63606548e-01 8.24446499e-01 -5.56829274e-01 4.25886869e-01
-6.22929513e-01 6.49104834e-01 -2.69989580e-01 7.61655390e-01
8.00558567e-01 -5.24470508e-01 -1.50594935e-02 -4.08555210e-01
-2.78588504e-01 1.26734838e-01 -1.43691909e+00 2.05649114e+00
-5.39493799e-01 1.05615489e-01 3.15137506e-01 -7.51247764e-01
1.04875612e+00 1.76237941e-01 6.35953605e-01 -6.14140272e-01
2.05020651e-01 3.01325113e-01 -5.06971896e-01 -1.12962879e-01
2.84325808e-01 -1.62990227e-01 3.24088186e-01 3.61469805e-01
-1.61243975e-02 -6.76572800e-01 -2.08585098e-01 3.92893888e-02
1.00035584e+00 5.04850507e-01 -2.88254060e-02 -1.55658405e-02
3.32006067e-01 3.82347568e-03 4.39144671e-01 2.80022532e-01
5.44429064e-01 7.61781216e-01 1.33038566e-01 -8.59017909e-01
-1.33047545e+00 -1.39661288e+00 -9.18728262e-02 1.82400048e-01
5.79453647e-01 -3.16497125e-02 -6.13455594e-01 -4.70476925e-01
3.62415016e-01 4.35901403e-01 -6.08537614e-01 1.24157652e-01
-6.80865645e-01 1.63499743e-01 -1.37583941e-01 4.43998694e-01
4.72406119e-01 -1.42436242e+00 -9.77147102e-01 3.63453090e-01
-1.89015586e-02 -1.27544308e+00 -3.87383640e-01 3.38894203e-02
-1.14846265e+00 -1.33957577e+00 -4.80817735e-01 -8.69351089e-01
9.98113334e-01 2.09667698e-01 1.17165339e+00 4.48993266e-01
-1.86982185e-01 5.16087055e-01 -9.75724608e-02 -1.55439720e-01
-3.40867579e-01 -6.67482987e-02 -1.98867321e-01 1.73544288e-01
-3.89445186e-01 -7.40481019e-01 -7.52066135e-01 2.34317765e-01
-8.51575911e-01 4.76433069e-01 4.75598991e-01 6.80318058e-01
1.26141977e+00 -4.68220636e-02 9.73700434e-02 -7.66425252e-01
-1.61434934e-01 -3.29813033e-01 -9.96197641e-01 2.18959693e-02
-4.24037486e-01 7.52394870e-02 6.16364598e-01 -2.56740808e-01
-6.28077984e-01 6.85088038e-01 -1.28210425e-01 -1.12757957e+00
-1.90298066e-01 2.63749629e-01 -1.14697842e-02 -3.76751840e-01
4.29724783e-01 3.61261755e-01 3.61977406e-02 -6.74929738e-01
2.97359407e-01 9.13897529e-02 5.41442513e-01 -5.40524960e-01
1.06285334e+00 8.48219573e-01 2.94494212e-01 -8.00388217e-01
-6.22518241e-01 -2.54750967e-01 -9.19342339e-01 -3.03184479e-01
8.13457668e-01 -8.56106758e-01 -8.77026856e-01 5.46448886e-01
-1.37282622e+00 -6.99914098e-01 -5.66735983e-01 2.34736577e-01
-1.07100666e+00 9.62797105e-02 -5.17434716e-01 -5.32109618e-01
-4.93885547e-01 -1.24918079e+00 1.56587946e+00 9.68450531e-02
2.55170405e-01 -7.76060700e-01 -9.28127319e-02 2.21694082e-01
1.90398753e-01 7.93111742e-01 9.82783318e-01 3.47847551e-01
-1.31929755e+00 1.96514443e-01 -3.60366106e-01 1.18303500e-01
6.19446710e-02 -1.37897298e-01 -7.17554271e-01 -3.92905712e-01
-2.30389342e-01 -3.44372272e-01 4.71241444e-01 5.08730292e-01
1.30617809e+00 -1.92308500e-01 -2.92349339e-01 1.07527483e+00
1.63755763e+00 -4.73392569e-02 7.49476016e-01 9.74050909e-02
9.96260345e-01 2.50345975e-01 3.28666240e-01 4.14085001e-01
6.19880080e-01 8.17316771e-01 1.15250731e+00 -2.91144699e-01
-4.93099064e-01 -4.13787365e-01 -1.43609941e-01 6.53431594e-01
-6.15807436e-02 7.20710307e-02 -8.46761763e-01 5.40783882e-01
-1.83728695e+00 -4.58974630e-01 -4.53419238e-03 2.09586787e+00
7.15766549e-01 3.99404764e-02 2.48181615e-02 8.33441168e-02
4.39866424e-01 -1.49426460e-01 -8.75151455e-01 -3.27010661e-01
1.16286956e-01 5.49664140e-01 2.49955624e-01 5.94611645e-01
-8.85630846e-01 1.18034494e+00 4.89154959e+00 3.25140774e-01
-1.22546935e+00 -6.13883100e-02 3.88521016e-01 1.19791962e-01
-3.00470442e-01 -8.16731825e-02 -4.52518731e-01 1.31628588e-01
3.11995447e-01 3.31287473e-01 8.46889913e-01 7.44045615e-01
-6.31156787e-02 4.64545982e-03 -1.22547674e+00 1.30133915e+00
4.48832288e-02 -1.74044907e+00 2.74831384e-01 2.08659321e-01
8.85465503e-01 2.09149316e-01 -2.64003724e-01 -1.22950941e-01
3.05147916e-01 -1.14154840e+00 1.13702750e+00 8.93338621e-01
9.65439141e-01 -9.02117550e-01 3.56369019e-01 4.63012397e-01
-1.34151042e+00 3.00412744e-01 -5.22013426e-01 5.63477129e-02
2.19588056e-02 6.99230969e-01 -9.50631440e-01 6.90920949e-01
7.88975954e-01 8.57890725e-01 -1.90720797e-01 1.18538725e+00
-3.96422058e-01 1.35321811e-01 -4.08540130e-01 1.84382185e-01
1.56569839e-01 -2.11272106e-01 4.78389561e-01 6.45796776e-01
2.41322681e-01 2.73682415e-01 4.57933664e-01 1.33888173e+00
-5.19427598e-01 3.94343883e-02 -7.10896909e-01 4.73186284e-01
5.02833366e-01 1.28679967e+00 -1.04193127e+00 -1.73203528e-01
-2.12416053e-01 1.00462568e+00 6.55564010e-01 1.46275386e-01
-5.31390607e-01 -2.50019506e-03 4.06258732e-01 4.44902956e-01
7.60849535e-01 -4.16730911e-01 -5.13704777e-01 -8.12356055e-01
2.12794349e-01 -6.27657175e-01 -7.52907991e-02 -1.00332618e+00
-1.18585336e+00 5.94839573e-01 -3.70420933e-01 -1.12683105e+00
9.86786783e-02 -3.96421492e-01 -2.84545720e-01 8.70680630e-01
-1.78921878e+00 -1.30740082e+00 -6.40495002e-01 7.27586448e-01
4.23975587e-01 1.42053068e-01 7.86584020e-01 1.06273688e-01
6.72620833e-02 1.10821709e-01 -2.75737137e-01 1.12824373e-01
8.39225948e-02 -1.23127198e+00 5.65693319e-01 3.29581141e-01
2.26158231e-01 1.73282437e-02 1.78673878e-01 -8.56169701e-01
-1.93994510e+00 -1.38546407e+00 5.42766750e-01 -2.93739408e-01
4.48534600e-02 -3.59137952e-01 -8.76916707e-01 5.52100718e-01
-1.79533094e-01 4.07934219e-01 -2.29260415e-01 -4.62850064e-01
-1.82629153e-01 -1.67670213e-02 -1.37431300e+00 4.45569545e-01
1.19263685e+00 -3.71846586e-01 -2.57231057e-01 4.00485873e-01
6.63742781e-01 -1.11382437e+00 -1.06690860e+00 3.82221431e-01
4.32900965e-01 -8.47560704e-01 1.13334966e+00 -2.32373700e-01
5.88009298e-01 -5.40874600e-01 -8.07596818e-02 -1.33872950e+00
-2.11429372e-01 -3.70400041e-01 -2.78376579e-01 5.98503351e-01
-7.51283159e-03 -3.67065996e-01 8.98177564e-01 3.02162766e-01
-3.47876161e-01 -1.17929351e+00 -1.10019433e+00 -5.61176717e-01
-6.86479658e-02 -2.85814047e-01 8.64479184e-01 7.38597214e-01
-7.89078534e-01 1.54773831e-01 -4.14260365e-02 4.10330445e-01
9.33968365e-01 6.61641836e-01 9.04377759e-01 -1.34687495e+00
3.31337767e-04 -2.20733508e-01 -4.68183279e-01 -1.26272786e+00
1.50603220e-01 -1.07989621e+00 2.28821740e-01 -1.81390107e+00
-2.31929109e-01 -8.50452185e-01 2.24117145e-01 5.04089355e-01
3.75026792e-01 4.61709917e-01 2.42232271e-02 1.84738394e-02
-4.77622330e-01 8.34864378e-01 1.65138447e+00 -2.00704381e-01
-2.60202706e-01 -1.09857708e-01 -2.08073124e-01 8.06920052e-01
6.35016739e-01 -5.01842618e-01 -1.29108295e-01 -7.70874918e-01
3.24112654e-01 3.90040308e-01 7.48541594e-01 -8.86819899e-01
2.60234982e-01 -1.33613214e-01 7.04911172e-01 -9.38531339e-01
7.12602019e-01 -1.06416035e+00 3.48149449e-01 6.20850444e-01
1.66090041e-01 6.36113510e-02 1.16344705e-01 5.21201432e-01
2.79025435e-01 -3.09285261e-02 9.76855040e-01 -3.68156940e-01
-1.57538950e-01 8.07780385e-01 1.04745604e-01 -8.26040208e-02
9.48669016e-01 -2.67774194e-01 8.98851231e-02 -2.05907091e-01
-7.23332345e-01 8.87735412e-02 1.08696854e+00 2.67966062e-01
1.23896062e+00 -1.59068263e+00 -8.10444832e-01 3.95948112e-01
-2.14488789e-01 8.38648558e-01 -7.59327039e-02 3.61378700e-01
-8.67166102e-01 4.44745384e-02 -1.43136859e-01 -9.29094493e-01
-8.74041378e-01 4.59792376e-01 6.30985856e-01 -1.74574237e-02
-1.09272027e+00 6.39443457e-01 1.29542783e-01 -7.13519514e-01
3.31215054e-01 -4.47400957e-01 2.69353181e-01 -4.42279279e-01
8.63010287e-02 3.55291873e-01 3.29638481e-01 -6.66953027e-01
-2.20416471e-01 9.18931782e-01 3.04391474e-01 5.30418009e-02
1.60621691e+00 1.17282182e-01 -2.10545421e-01 4.03601378e-01
1.01553679e+00 -3.44459862e-01 -1.49457741e+00 -4.82909113e-01
-4.23732996e-01 -6.39267981e-01 1.95724785e-01 -6.08037233e-01
-1.46745884e+00 8.48812759e-01 4.54756856e-01 -9.95423272e-02
8.40136945e-01 2.03922853e-01 9.43174183e-01 2.67515957e-01
8.65407228e-01 -6.76276565e-01 1.87876627e-01 6.02420568e-01
1.27654076e+00 -8.72129679e-01 9.87852365e-02 -5.16300797e-01
5.91274202e-02 1.18981218e+00 5.91225326e-01 -5.50024271e-01
8.67622435e-01 9.63211209e-02 -2.27619722e-01 -5.53849161e-01
-5.35933077e-01 -1.01156354e-01 4.07059431e-01 5.09139001e-01
-1.72466531e-01 1.64012805e-01 3.03626537e-01 3.20072728e-03
-2.45829895e-01 1.45362290e-02 2.05768123e-01 8.43251884e-01
-2.89056093e-01 -8.58928561e-01 -4.32771742e-01 4.58196372e-01
1.32953390e-01 2.40166202e-01 -4.74651188e-01 5.70564330e-01
2.90604532e-01 6.03102326e-01 2.91508853e-01 -2.06986696e-01
6.13104105e-01 -1.85778454e-01 8.62600863e-01 -8.12065363e-01
-5.06669044e-01 -3.91293243e-02 -3.95082533e-01 -8.36307645e-01
-2.51298666e-01 -5.34399390e-01 -1.56379390e+00 -3.76244754e-01
2.77226623e-02 -2.22608939e-01 8.88165891e-01 8.41023207e-01
4.37062293e-01 4.87828583e-01 7.80015290e-01 -1.47491348e+00
-1.00819640e-01 -4.96095806e-01 -3.32432956e-01 4.48096842e-01
2.59166628e-01 -9.36752260e-01 -1.15495976e-02 -5.40076010e-02]
|
[8.597366333007812, -3.5109875202178955]
|
26190b70-1135-41dc-bd9f-4708246fe7f7
|
why-can-t-discourse-parsing-generalize-a
|
2302.06488
| null |
https://arxiv.org/abs/2302.06488v1
|
https://arxiv.org/pdf/2302.06488v1.pdf
|
Why Can't Discourse Parsing Generalize? A Thorough Investigation of the Impact of Data Diversity
|
Recent advances in discourse parsing performance create the impression that, as in other NLP tasks, performance for high-resource languages such as English is finally becoming reliable. In this paper we demonstrate that this is not the case, and thoroughly investigate the impact of data diversity on RST parsing stability. We show that state-of-the-art architectures trained on the standard English newswire benchmark do not generalize well, even within the news domain. Using the two largest RST corpora of English with text from multiple genres, we quantify the impact of genre diversity in training data for achieving generalization to text types unseen during training. Our results show that a heterogeneous training regime is critical for stable and generalizable models, across parser architectures. We also provide error analyses of model outputs and out-of-domain performance. To our knowledge, this study is the first to fully evaluate cross-corpus RST parsing generalizability on complete trees, examine between-genre degradation within an RST corpus, and investigate the impact of genre diversity in training data composition.
|
['Amir Zeldes', 'Yang Janet Liu']
|
2023-02-13
| null | null | null | null |
['cross-corpus', 'discourse-parsing']
|
['computer-vision', 'natural-language-processing']
|
[ 2.86383688e-01 3.28433186e-01 -2.38197386e-01 -4.74610269e-01
-1.43672967e+00 -1.01254296e+00 5.92668176e-01 2.88475573e-01
-6.79446042e-01 6.98576868e-01 6.40401065e-01 -6.76096916e-01
9.34798047e-02 -4.37188268e-01 -8.92614543e-01 -2.40834221e-01
-1.85569689e-01 5.51070452e-01 3.64196151e-01 -3.22236389e-01
4.69907001e-02 -5.99587709e-02 -1.21900558e+00 7.73717880e-01
6.29985154e-01 5.28389752e-01 1.25579089e-01 8.26160789e-01
-1.56666353e-01 7.24258780e-01 -1.07082331e+00 -5.83270371e-01
-1.77458376e-02 -4.01571095e-01 -1.12889004e+00 -4.02275659e-03
5.73433280e-01 -3.05391133e-01 -8.02556500e-02 6.23688161e-01
5.68795323e-01 -6.20009042e-02 4.89972681e-01 -5.63903213e-01
-6.51988029e-01 1.36602736e+00 -2.76560426e-01 6.48991525e-01
4.37657982e-01 -8.78466852e-03 1.35705030e+00 -3.85470331e-01
1.15606523e+00 1.29957092e+00 7.51867771e-01 6.89662933e-01
-1.34664416e+00 -5.02957940e-01 5.09490550e-01 -1.34667754e-01
-7.13068187e-01 -8.05805385e-01 6.04886770e-01 -3.96577120e-01
1.38521171e+00 2.18655095e-01 5.38278185e-02 1.68000460e+00
1.63183972e-01 8.48308623e-01 1.10284185e+00 -7.50963748e-01
1.85973287e-01 -1.66595712e-01 5.03563643e-01 3.21534544e-01
2.74213970e-01 -2.23153040e-01 -7.23451078e-01 4.82800156e-02
8.40623006e-02 -9.79500353e-01 -3.08251798e-01 2.31104538e-01
-1.08334100e+00 9.32582557e-01 -2.30634380e-02 6.52365446e-01
-2.12936085e-02 -8.25221017e-02 1.03960443e+00 5.40012062e-01
7.17135012e-01 6.73018456e-01 -8.28001320e-01 -6.81056976e-01
-8.34937692e-01 3.20731789e-01 1.24393678e+00 8.73273194e-01
-6.73114657e-02 3.10013220e-02 6.03561066e-02 1.07631874e+00
-1.11637957e-01 2.63895571e-01 7.77457714e-01 -1.18226421e+00
1.20170522e+00 3.61065179e-01 -2.37258002e-01 -4.53722298e-01
-6.79225147e-01 -4.10271525e-01 -3.01364064e-01 -2.11618140e-01
9.36516821e-01 -5.29433846e-01 -6.17233396e-01 2.09264493e+00
1.02925159e-01 -6.83711469e-01 4.88527685e-01 4.77965057e-01
1.11555111e+00 4.47532296e-01 4.09106165e-01 -2.91571170e-01
1.54053771e+00 -6.49314523e-01 -4.60126281e-01 -6.24807298e-01
9.60277736e-01 -9.10409689e-01 1.31891370e+00 4.80064601e-01
-1.17914629e+00 -4.23351049e-01 -1.08935618e+00 -3.71969044e-01
-1.01388082e-01 1.86147645e-01 5.41292191e-01 8.54454696e-01
-6.76621020e-01 6.75189018e-01 -1.12714911e+00 -6.40537143e-01
1.82561532e-01 8.98732916e-02 -2.57012308e-01 1.37506947e-01
-1.13396299e+00 9.30996954e-01 4.28066313e-01 -2.46550828e-01
-4.65431184e-01 -6.84162676e-01 -6.41068995e-01 -8.67469311e-02
2.13437840e-01 -2.32975096e-01 1.94042087e+00 -1.06047726e+00
-1.44897318e+00 1.09511113e+00 -1.48802757e-01 -5.83847284e-01
5.12029648e-01 -3.08605790e-01 -3.07550371e-01 -1.72342777e-01
3.12765449e-01 3.65873188e-01 1.08191408e-01 -1.18595743e+00
-6.95273519e-01 -4.27490711e-01 2.25116238e-01 2.18219474e-01
-2.10585549e-01 3.43049079e-01 -6.37157485e-02 -5.71666300e-01
2.27581471e-01 -9.14136231e-01 2.36453906e-01 -8.68471920e-01
-4.07700390e-01 -3.44036609e-01 4.09200191e-01 -8.84680569e-01
1.40398657e+00 -2.14657187e+00 8.38863701e-02 -4.42987591e-01
-1.42748073e-01 -1.57457262e-01 -1.96999591e-02 4.88324165e-01
6.06821217e-02 4.43262666e-01 -2.15596661e-01 -5.82902372e-01
7.37239644e-02 4.22453880e-01 -2.85161197e-01 1.80148855e-01
3.20856273e-01 6.17832780e-01 -5.33367634e-01 -4.50937390e-01
-3.67852688e-01 1.01914003e-01 -6.73990786e-01 -9.32859406e-02
-5.01614034e-01 3.53394747e-01 -4.13953364e-01 3.66592914e-01
9.63064358e-02 -1.05979301e-01 5.43921411e-01 1.85404688e-01
-3.76878440e-01 1.06032097e+00 -7.46637344e-01 1.70036578e+00
-5.67001224e-01 7.35139072e-01 2.73906380e-01 -8.84148002e-01
5.55265367e-01 3.20903838e-01 2.58902520e-01 -8.30915809e-01
2.93782711e-01 4.88186210e-01 5.72943449e-01 -6.08176291e-01
7.43154049e-01 -2.81302214e-01 -4.96794701e-01 4.78734791e-01
9.47658345e-02 -3.76301259e-02 6.39747202e-01 -9.30923969e-02
1.27742517e+00 1.36096150e-01 6.47498518e-02 -5.22680819e-01
4.84811403e-02 5.56792617e-01 5.99270642e-01 7.25287259e-01
4.45016026e-02 7.01651454e-01 8.59959424e-01 -1.72366530e-01
-1.14586437e+00 -7.85659075e-01 -6.55409515e-01 1.72051144e+00
-3.83742452e-01 -4.66467768e-01 -8.94267380e-01 -6.39819145e-01
-1.94134668e-01 1.01143730e+00 -4.93534952e-01 3.63728553e-01
-1.11305618e+00 -1.23001969e+00 9.52011347e-01 4.88262832e-01
3.79875839e-01 -9.50250447e-01 -7.64005423e-01 6.19780123e-01
-3.26673597e-01 -1.47047102e+00 -2.99878493e-02 5.92146397e-01
-8.91621172e-01 -1.08420956e+00 -4.16802973e-01 -9.20600355e-01
-1.25926957e-01 -2.92898804e-01 1.49599051e+00 -1.47774696e-01
2.03980207e-01 2.23452151e-01 -6.98911309e-01 -5.00437021e-01
-1.10777020e+00 7.53653109e-01 -3.77604693e-01 -8.08461010e-01
4.14587051e-01 -3.81562978e-01 -1.11212470e-01 7.00884089e-02
-5.19597709e-01 -8.89063701e-02 3.09986472e-01 8.70540440e-01
2.75251538e-01 -2.92874631e-02 5.24972796e-01 -1.49168384e+00
7.95327067e-01 -5.85998178e-01 -1.99315682e-01 1.61357731e-01
-3.04607064e-01 2.06787586e-01 7.30196893e-01 -3.41058701e-01
-1.42156303e+00 -4.08736289e-01 -4.29429084e-01 5.50822556e-01
-2.56940395e-01 7.06160128e-01 5.85707487e-04 5.98439455e-01
1.04095447e+00 -2.84828365e-01 -3.51885587e-01 -7.21702516e-01
2.01839671e-01 6.73731267e-01 4.73640114e-01 -1.13868475e+00
1.23517759e-01 1.42681494e-01 -3.96751136e-01 -9.90334094e-01
-8.34980249e-01 -1.93442121e-01 -5.53632677e-01 2.33571142e-01
1.11844409e+00 -9.61407781e-01 -1.87935859e-01 2.73246914e-01
-1.38518107e+00 -7.31678307e-01 -2.29526132e-01 3.49611819e-01
-4.34045851e-01 2.25353763e-01 -1.07255244e+00 -4.99353498e-01
-1.88928396e-01 -1.18516719e+00 1.12809396e+00 -7.44771138e-02
-5.27714133e-01 -9.94757950e-01 2.92400002e-01 4.16096061e-01
8.02572891e-02 2.12251455e-01 1.06246495e+00 -1.06486833e+00
-1.18935399e-01 1.61782503e-01 1.44133255e-01 1.85406029e-01
-9.75938886e-02 1.53314695e-01 -1.08814716e+00 -1.94099680e-01
9.43613574e-02 -5.19151926e-01 8.54860187e-01 4.48295921e-01
5.51556945e-01 -2.13141695e-01 -2.12367728e-01 3.66410345e-01
1.09532714e+00 1.42340511e-01 3.11707288e-01 9.13247347e-01
3.63785386e-01 8.49278688e-01 4.18689191e-01 1.02981552e-01
6.28805041e-01 4.78885531e-01 -2.45761707e-01 1.79484129e-01
-2.42699221e-01 -4.54225345e-03 5.17626047e-01 1.02345002e+00
3.85475978e-02 -5.79663515e-01 -1.19922829e+00 5.06949186e-01
-1.49120152e+00 -8.10958982e-01 -2.26834834e-01 1.81837034e+00
1.10332513e+00 7.34574616e-01 2.17784241e-01 1.85356550e-02
4.80271846e-01 2.67535716e-01 -2.29095533e-01 -8.59402955e-01
-3.64735693e-01 3.12515438e-01 5.32813132e-01 5.18159926e-01
-1.15030956e+00 9.61023331e-01 6.71616411e+00 4.94889349e-01
-9.91199076e-01 4.06871140e-01 8.28205943e-01 -2.21270278e-01
-2.08681971e-01 -4.86499108e-02 -1.08157277e+00 3.12594622e-01
1.39091337e+00 7.71267153e-03 1.50862411e-02 8.57458055e-01
1.31664714e-02 -3.95735912e-02 -1.32929468e+00 2.48262599e-01
-6.76149279e-02 -1.03734767e+00 -5.13804436e-01 -2.48007383e-03
5.61824501e-01 5.81860363e-01 -2.16254815e-01 5.90772688e-01
6.28885329e-01 -8.47919583e-01 9.58096445e-01 -2.77138174e-01
6.60007358e-01 -3.51897597e-01 6.03673398e-01 3.33123326e-01
-9.12921906e-01 -3.45421880e-02 -1.84067160e-01 3.32268067e-02
3.19560558e-01 1.54583648e-01 -8.45028341e-01 4.07442778e-01
7.75569320e-01 3.70224893e-01 -7.92268753e-01 4.34032530e-01
-6.83512688e-02 1.03949869e+00 -5.06579697e-01 2.44759512e-03
2.47264892e-01 3.84755760e-01 5.69538116e-01 1.60427129e+00
-5.81543408e-02 1.03079550e-01 1.61118925e-01 2.41353855e-01
-1.44858792e-01 2.00809136e-01 -3.42074394e-01 -1.62738889e-01
5.03305256e-01 7.37111688e-01 -7.65157878e-01 -1.55084237e-01
-6.88128173e-01 5.11675835e-01 6.13724709e-01 2.12387159e-01
-5.50628543e-01 2.24116556e-02 3.67128193e-01 1.16864927e-01
3.22722584e-01 -4.20722961e-01 -6.05022073e-01 -1.08644474e+00
2.28331268e-01 -1.29647756e+00 7.25904822e-01 -2.64550537e-01
-1.39517581e+00 7.39560306e-01 2.28601724e-01 -7.14505196e-01
-4.72946972e-01 -8.32644165e-01 -4.54454958e-01 5.63867748e-01
-1.40719235e+00 -1.00709951e+00 1.42241865e-01 1.87219586e-02
9.81193781e-01 -2.16548935e-01 7.94582069e-01 1.31481037e-01
-7.41192937e-01 7.45872736e-01 3.53693575e-01 3.24584156e-01
7.51755893e-01 -1.37161696e+00 7.54438341e-01 9.18925941e-01
2.33853787e-01 6.54332280e-01 9.50773358e-01 -6.04572117e-01
-1.17043650e+00 -5.95497549e-01 9.06778395e-01 -6.82100654e-01
9.12787855e-01 -5.48332930e-01 -9.75654483e-01 1.05299056e+00
3.94661576e-01 -6.51626289e-01 6.64292336e-01 8.94252777e-01
-4.75472927e-01 2.81287551e-01 -6.84433818e-01 5.11640787e-01
1.22570169e+00 -4.34233814e-01 -8.53656948e-01 3.61379802e-01
8.51468861e-01 -8.57577682e-01 -1.20824969e+00 2.78716207e-01
5.23389220e-01 -1.01049423e+00 6.51152313e-01 -7.38445938e-01
8.40403736e-01 4.93345976e-01 -4.84232783e-01 -1.03826821e+00
-1.65119663e-01 -5.08491337e-01 3.98723423e-01 1.62908196e+00
9.88466740e-01 -4.80001152e-01 8.28728437e-01 6.60434842e-01
-5.41245341e-01 -5.29840946e-01 -1.30677533e+00 -7.36074269e-01
1.00770891e+00 -5.36643624e-01 4.10169214e-01 8.30060184e-01
2.46633336e-01 7.42934227e-01 3.27406496e-01 -1.90179069e-02
2.60506362e-01 1.15823280e-03 6.39395595e-01 -1.03345549e+00
-7.94989049e-01 -7.15561867e-01 7.80529603e-02 -8.98796201e-01
4.23293859e-01 -8.91118586e-01 1.82655886e-01 -1.31874657e+00
1.02794319e-02 -6.62640035e-01 1.80708751e-01 2.65939564e-01
-2.13545531e-01 -1.62948862e-01 3.48859936e-01 2.55531877e-01
-4.85475183e-01 2.21966431e-01 9.13243890e-01 2.08715871e-01
-2.97891676e-01 -2.07628682e-01 -8.70463133e-01 6.57269895e-01
9.79736567e-01 -5.82479119e-01 -3.14095318e-01 -9.89684224e-01
5.51852025e-02 3.61733325e-02 -1.05273291e-01 -8.46238852e-01
-1.00107901e-01 9.69088078e-02 1.45563886e-01 -2.26976037e-01
1.89668551e-01 -2.46564046e-01 -2.36302808e-01 1.23337686e-01
-5.37562191e-01 3.72644186e-01 6.55691683e-01 1.61986858e-01
-8.67732391e-02 -5.07137239e-01 5.12518466e-01 -3.21290493e-01
-5.70477843e-01 -4.21790391e-01 -4.12040263e-01 9.60480154e-01
5.05372226e-01 -9.41744149e-02 -7.68347621e-01 6.11498319e-02
-6.07261062e-01 7.91746378e-02 3.64299029e-01 4.72664952e-01
-2.97497034e-01 -7.25773215e-01 -1.04855514e+00 -1.13961332e-01
1.20841237e-02 3.87540720e-02 -5.18405857e-03 4.36891794e-01
-4.63443696e-01 4.38011438e-01 2.08826333e-01 -6.32262826e-01
-1.33086419e+00 9.20760483e-02 1.18313380e-01 -6.21237278e-01
-6.61615610e-01 9.67778385e-01 1.04419217e-01 -4.76659596e-01
1.32027611e-01 -5.01196027e-01 -1.23873852e-01 2.61331975e-01
1.48033440e-01 1.24779433e-01 3.20991725e-01 -5.43567061e-01
-3.36086631e-01 3.04872602e-01 -3.03434730e-01 -4.84808743e-01
1.49913669e+00 -4.29647639e-02 3.71570230e-01 7.71497369e-01
1.12450111e+00 1.28895596e-01 -1.11805022e+00 1.70376264e-02
3.43907624e-01 1.37500659e-01 -1.36489645e-01 -8.64003360e-01
-5.51160634e-01 6.92050159e-01 7.82107860e-02 6.10931277e-01
8.33135128e-01 2.24527866e-01 7.23159075e-01 3.89922142e-01
1.03826277e-01 -1.41651154e+00 -2.40878269e-01 8.40461195e-01
9.15046275e-01 -1.22885156e+00 4.94973101e-02 -2.89883524e-01
-8.76147330e-01 1.07964694e+00 5.31762719e-01 -1.00683719e-01
3.93317878e-01 4.94881243e-01 1.78426400e-01 -1.64968893e-01
-9.12212253e-01 8.45283195e-02 -1.00073829e-01 3.55401307e-01
1.21546364e+00 6.09740280e-02 -6.38058424e-01 8.91071200e-01
-9.97469842e-01 -5.09311318e-01 6.66624784e-01 1.06088614e+00
-3.78263712e-01 -1.36925387e+00 -2.45239288e-01 3.09384286e-01
-9.41232264e-01 -1.80939868e-01 -5.45021951e-01 1.40224552e+00
-1.50925547e-01 1.14531112e+00 1.68494448e-01 -9.97429267e-02
6.75593376e-01 4.97941196e-01 6.06072605e-01 -9.21471357e-01
-9.29623067e-01 1.72137350e-01 1.13378632e+00 -2.13287249e-01
-4.40921456e-01 -1.12626636e+00 -1.30827296e+00 -4.92100835e-01
-2.20855787e-01 1.51473269e-01 7.04293370e-01 9.06141639e-01
1.23955786e-01 7.30092049e-01 -2.10642125e-02 -6.12914205e-01
-7.08809674e-01 -1.15411401e+00 -2.17913985e-01 4.48799312e-01
2.68261790e-01 -4.61883485e-01 -4.37506229e-01 1.10608265e-01]
|
[10.646891593933105, 9.480335235595703]
|
d9baa06b-0380-4b67-bb11-e5f4594da245
|
confidence-aware-3d-gaze-estimation-and
|
2303.10062
| null |
https://arxiv.org/abs/2303.10062v1
|
https://arxiv.org/pdf/2303.10062v1.pdf
|
Confidence-aware 3D Gaze Estimation and Evaluation Metric
|
Deep learning appearance-based 3D gaze estimation is gaining popularity due to its minimal hardware requirements and being free of constraint. Unreliable and overconfident inferences, however, still limit the adoption of this gaze estimation method. To address the unreliable and overconfident issues, we introduce a confidence-aware model that predicts uncertainties together with gaze angle estimations. We also introduce a novel effectiveness evaluation method based on the causality between eye feature degradation and the rise in inference uncertainty to assess the uncertainty estimation. Our confidence-aware model demonstrates reliable uncertainty estimations while providing angular estimation accuracies on par with the state-of-the-art. Compared with the existing statistical uncertainty-angular-error evaluation metric, the proposed effectiveness evaluation approach can more effectively judge inferred uncertainties' performance at each prediction.
|
['Xiaoli Zhang', 'Amy Zhang', 'Jiucai Zhang', 'Qiaojie Zheng']
|
2023-03-17
| null | null | null | null |
['gaze-estimation']
|
['computer-vision']
|
[-1.90902174e-01 3.43291521e-01 1.17981203e-01 -8.80554676e-01
-6.19727850e-01 -1.31669939e-01 4.52203810e-01 1.92753181e-01
-4.52309579e-01 8.53714287e-01 -1.43442407e-01 -2.04750970e-01
-3.12997967e-01 1.44397952e-02 -7.14675426e-01 -5.46096623e-01
1.24589935e-01 -1.27269670e-01 1.70952603e-01 4.31647509e-01
6.70432925e-01 1.55729324e-01 -2.18478179e+00 -4.83266622e-01
1.37758279e+00 1.68487000e+00 -3.03550482e-01 3.97637248e-01
2.62408435e-01 5.27226508e-01 -6.83515191e-01 -6.20517671e-01
-4.87111509e-02 7.36081749e-02 -2.38274947e-01 -3.45196128e-01
9.43468153e-01 -8.72472167e-01 4.46578383e-01 1.20524013e+00
4.54518139e-01 -2.19362006e-02 7.88629651e-01 -1.68775833e+00
-5.14952898e-01 1.81076601e-01 -9.48969960e-01 2.48695761e-01
5.78758240e-01 7.21848384e-02 7.25944400e-01 -7.50100195e-01
2.36028567e-01 8.95061672e-01 6.44938827e-01 5.08516073e-01
-8.66889417e-01 -8.97818089e-01 4.15556222e-01 5.89053810e-01
-1.67255723e+00 -7.10674584e-01 5.77623308e-01 -5.55909872e-01
5.91263831e-01 2.58337706e-02 3.00110608e-01 9.83512759e-01
2.57644117e-01 6.86467826e-01 1.36469865e+00 -4.15297478e-01
3.64814013e-01 4.70363736e-01 1.39747992e-01 6.06094539e-01
5.83038270e-01 4.20398682e-01 -1.02844095e+00 -1.84021238e-02
4.81196731e-01 -5.51449776e-01 -3.23380888e-01 -3.08116943e-01
-6.62904978e-01 2.34526798e-01 3.70372653e-01 -3.55219454e-01
-2.82897562e-01 2.32135579e-01 -4.90911342e-02 -9.80892032e-02
7.47106612e-01 1.76517472e-01 -2.43274570e-01 -6.02583766e-01
-1.09229994e+00 1.42929748e-01 4.19570625e-01 1.12845337e+00
4.93176132e-01 1.56937063e-01 -2.04973832e-01 3.43723267e-01
1.04245770e+00 6.01112902e-01 -1.09180257e-01 -9.76762414e-01
-6.61614537e-02 4.20701206e-01 5.65770328e-01 -9.81319427e-01
-3.90223444e-01 -3.24067444e-01 -3.29705119e-01 8.92529666e-01
5.70593476e-01 -1.33883893e-01 -6.59792304e-01 1.87944376e+00
6.07330203e-01 2.93366998e-01 -4.97120231e-01 9.19865966e-01
6.17046237e-01 -1.54869959e-01 8.72949138e-02 -5.04504085e-01
1.29610729e+00 -3.50799471e-01 -8.81341159e-01 1.86775178e-01
9.41329747e-02 -6.99882388e-01 9.09743190e-01 7.61673689e-01
-8.37563813e-01 -2.88061023e-01 -1.39548969e+00 -2.03523394e-02
7.28748068e-02 9.13874209e-02 5.99188030e-01 1.10684991e+00
-1.16903973e+00 3.93251508e-01 -6.71140611e-01 -1.64266571e-01
3.99472624e-01 4.44078028e-01 -4.91653048e-02 4.28939462e-01
-7.60479748e-01 1.26902699e+00 -1.39728725e-01 2.77059168e-01
-5.04084527e-01 -6.98191047e-01 -8.36123943e-01 -1.37320384e-01
3.58285964e-01 -4.57875699e-01 1.35873902e+00 -5.76253712e-01
-2.07218814e+00 3.58832866e-01 -3.29662532e-01 -3.68909001e-01
6.32181108e-01 -5.82035422e-01 -4.54016060e-01 -1.49688259e-01
-3.60858709e-01 7.69092381e-01 1.30061209e+00 -1.27878869e+00
-7.59587169e-01 -6.56744301e-01 2.27299258e-01 2.75324374e-01
-9.27917361e-02 -8.41769502e-02 -1.59162357e-01 -1.11955870e-02
2.19405847e-04 -7.48789847e-01 3.65827650e-01 5.21725833e-01
-3.55100185e-01 -3.94389480e-01 6.06538117e-01 -3.34818214e-01
1.38319468e+00 -2.00420260e+00 -3.72157842e-01 1.19371288e-01
7.24814236e-01 -8.21978077e-02 4.13594991e-01 -3.63769710e-01
4.11638826e-01 -1.08334452e-01 1.93225980e-01 -8.48101199e-01
2.98462749e-01 -4.11127806e-01 1.26342922e-01 7.95322716e-01
1.05531141e-01 5.09267509e-01 -6.35606170e-01 -6.04364574e-01
4.96371150e-01 6.70097411e-01 -4.31667149e-01 1.97772026e-01
-5.78209125e-02 4.10447389e-01 -2.40986392e-01 7.91099072e-01
1.04340875e+00 -1.10842451e-01 -1.92872256e-01 -3.89012843e-01
-2.87261605e-01 9.68206003e-02 -1.14074576e+00 1.30634785e+00
-4.32733059e-01 1.02718031e+00 -2.18267411e-01 -1.69201925e-01
7.78116286e-01 -1.75111520e-04 -1.66484103e-01 -5.22729158e-01
6.21643722e-01 1.96517676e-01 8.30058125e-04 -4.29392457e-01
7.88778186e-01 2.79683977e-01 2.44539276e-01 3.18940192e-01
1.41142175e-01 4.32733707e-02 -5.55790961e-01 -1.98629990e-01
4.51521337e-01 5.90496182e-01 4.28827643e-01 -4.05306429e-01
3.94981176e-01 -7.56063521e-01 3.46705765e-01 4.43879157e-01
-8.15631926e-01 5.60078502e-01 5.20643175e-01 -5.96063361e-02
-6.50605023e-01 -9.48084474e-01 -5.63653350e-01 7.38774240e-01
4.72524911e-01 -4.12273183e-02 -8.67029130e-01 -7.46685982e-01
1.32533684e-01 1.10174143e+00 -9.32653189e-01 -7.75436014e-02
3.72245044e-01 -2.61875480e-01 4.14728671e-01 4.42321390e-01
3.68190408e-01 -3.04211289e-01 -8.62739146e-01 -4.58907396e-01
2.05830455e-01 -9.53098238e-01 -3.98500443e-01 -4.59773511e-01
-5.06738782e-01 -1.11624050e+00 -5.37793636e-01 7.98703134e-02
8.44172895e-01 -2.87116934e-02 8.54615033e-01 -1.18484646e-01
1.43552870e-01 4.76728380e-01 -1.42091513e-01 -9.61913347e-01
-8.00641924e-02 -3.46319377e-01 6.48646712e-01 1.21753789e-01
8.49767864e-01 -3.27735245e-01 -8.19666982e-01 4.11515057e-01
-2.70728558e-01 -2.65603989e-01 4.43049282e-01 5.11668146e-01
2.63007343e-01 -3.65182102e-01 5.05413890e-01 -3.97234529e-01
7.22870588e-01 -4.04599637e-01 -1.13946140e+00 3.03299308e-01
-1.61760116e+00 3.05453509e-01 -3.12059134e-01 -5.19163728e-01
-1.49360490e+00 -4.84420091e-01 1.47167310e-01 -5.44599771e-01
-2.19864532e-01 2.96265781e-01 1.25740796e-01 -4.88132983e-01
7.86168635e-01 -3.69288206e-01 1.82437360e-01 -2.64533103e-01
1.43676087e-01 1.06470430e+00 3.70105803e-01 -3.44646871e-01
3.02305400e-01 2.08938301e-01 7.68653899e-02 -6.11811399e-01
-6.79217041e-01 -1.35537088e-01 -4.22275126e-01 -8.09464574e-01
4.94641453e-01 -8.40756595e-01 -1.64336956e+00 8.07230413e-01
-1.17003405e+00 2.29825020e-01 1.63711995e-01 7.93655872e-01
-3.21865022e-01 4.95993495e-01 8.02330822e-02 -1.57397485e+00
-2.81048208e-01 -1.22121763e+00 1.17185950e+00 6.45219803e-01
-3.16140652e-01 -7.65762508e-01 -2.27666989e-01 2.80214041e-01
4.99578327e-01 2.84855664e-01 3.06952208e-01 -2.62218028e-01
-5.54689944e-01 -3.16386998e-01 -4.75475103e-01 3.12275589e-01
-1.13327332e-01 3.74435693e-01 -1.47399867e+00 -2.70034149e-02
5.13709150e-02 -3.17317665e-01 1.84866101e-01 7.67236829e-01
9.69589829e-01 3.27493250e-02 -9.67287943e-02 3.80623460e-01
1.25218999e+00 -3.65371592e-02 6.53411031e-01 2.39169165e-01
3.43062222e-01 6.92652643e-01 7.22556770e-01 8.40008140e-01
7.69090116e-01 7.18637168e-01 6.90300524e-01 5.36810517e-01
1.38679877e-01 -2.24844754e-01 1.54093593e-01 3.39363337e-01
-3.89487147e-01 -3.90040934e-01 -7.13790059e-01 2.12516665e-01
-1.66266966e+00 -5.86847365e-01 -7.41334409e-02 2.77076530e+00
7.05201685e-01 3.72086793e-01 -8.06319341e-03 -1.67282167e-04
7.08506346e-01 -8.83344933e-02 -9.47869956e-01 -4.50409859e-01
3.99608970e-01 -3.16199899e-01 4.10485655e-01 6.20590091e-01
-6.44883573e-01 4.83642220e-01 6.94347572e+00 7.22902596e-01
-1.10595834e+00 1.90486200e-02 4.97122675e-01 -2.96441406e-01
-3.83561224e-01 -1.59113675e-01 -1.11611509e+00 6.60421073e-01
9.14189696e-01 -1.57357395e-01 2.64249772e-01 7.99429417e-01
9.92850214e-02 -9.05088782e-01 -1.20699286e+00 1.29186010e+00
3.40310514e-01 -9.32352543e-01 -5.47110856e-01 2.28224546e-01
4.60940838e-01 -1.71009138e-01 5.31372964e-01 4.49340679e-02
-1.21129438e-01 -8.98465276e-01 1.04697859e+00 1.02107751e+00
1.23200607e+00 -7.19159305e-01 8.81287634e-01 1.55793682e-01
-6.75392568e-01 1.60803542e-01 -1.14033066e-01 -2.79677570e-01
-1.32484101e-02 6.26241982e-01 -8.98910105e-01 2.92043626e-01
8.79304767e-01 2.57110029e-01 -6.49037421e-01 1.13125718e+00
-5.01038730e-01 2.57427841e-01 -6.00580454e-01 -4.89454508e-01
-4.17860210e-01 4.65867743e-02 6.81456685e-01 4.74010646e-01
4.01605874e-01 -2.86820321e-03 -9.55843449e-01 1.23253727e+00
1.48359090e-01 -2.86052167e-01 -2.66361028e-01 2.82721937e-01
9.43809271e-01 1.02721965e+00 -1.27200007e-01 8.09315890e-02
-3.89317334e-01 6.19735599e-01 3.88575733e-01 2.72477835e-01
-9.14496243e-01 -2.08362952e-01 8.76547933e-01 2.59183571e-02
1.16159869e-02 -1.00986630e-01 -6.66069567e-01 -9.51421022e-01
1.66199252e-01 -4.55232918e-01 -1.18195280e-01 -1.14748168e+00
-1.07028770e+00 8.57462049e-01 5.04106104e-01 -1.49928510e+00
-5.25427699e-01 -5.76398432e-01 -3.42744410e-01 9.60794747e-01
-1.51820636e+00 -9.34067070e-01 -6.52953744e-01 1.79813385e-01
1.28584310e-01 -5.40189184e-02 6.35247707e-01 -1.59699306e-01
-6.82587862e-01 1.28921962e+00 -3.18717770e-02 -6.90390825e-01
8.97656262e-01 -1.12966871e+00 -9.74387005e-02 9.41968739e-01
-4.07805860e-01 7.57301450e-01 1.04862869e+00 -3.37693125e-01
-9.14102852e-01 -3.32137704e-01 6.19587302e-01 -7.61919379e-01
4.24334705e-01 -1.50306910e-01 -7.00651109e-01 2.55780429e-01
2.17706218e-01 -4.71887132e-03 6.58460200e-01 5.23672223e-01
-5.27200341e-01 -2.46272922e-01 -1.52983940e+00 5.62473595e-01
9.25723851e-01 -5.59854388e-01 -5.50119877e-01 -5.15292406e-01
5.95744073e-01 -7.64054060e-01 -8.61251950e-01 5.38386583e-01
1.24108636e+00 -1.40239120e+00 4.01440024e-01 1.36480808e-01
2.35406347e-02 -3.11193794e-01 -2.04393771e-02 -1.04584897e+00
2.22631427e-03 -5.78001618e-01 -5.66632330e-01 1.31087387e+00
4.63564962e-01 -8.21152627e-01 6.25037730e-01 1.38742054e+00
2.48052731e-01 -6.25360310e-01 -1.19012964e+00 -5.79392672e-01
-4.01697308e-01 -7.92055726e-01 7.99332559e-01 4.02413487e-01
1.96014091e-01 -8.06881934e-02 -4.31511223e-01 4.90936279e-01
1.16944754e+00 -4.23898697e-01 5.99657595e-01 -1.44576120e+00
5.56024127e-02 -4.78088289e-01 -7.57142305e-01 -8.60522389e-01
4.95236786e-03 3.13457668e-01 2.91623026e-01 -8.56143296e-01
-1.19207852e-01 -3.54042977e-01 -4.37559366e-01 -2.10514222e-03
-3.83959532e-01 2.32154027e-01 -1.15445897e-01 1.19784497e-01
-5.65208316e-01 6.42092168e-01 7.50190794e-01 2.27744803e-01
-1.15430497e-01 1.77275330e-01 -6.66912317e-01 7.37554789e-01
4.07095581e-01 -2.05194384e-01 -6.67748451e-01 -3.84908050e-01
6.30310774e-01 -7.72320852e-02 2.49820828e-01 -1.08723986e+00
4.09654379e-01 1.49822114e-02 3.51314038e-01 -5.85163176e-01
2.91784614e-01 -7.26696312e-01 -1.77838266e-01 -3.51192176e-01
-1.82140931e-01 -3.54660422e-01 5.19916952e-01 8.44802260e-01
9.66390893e-02 -1.80655256e-01 7.59415865e-01 6.46620154e-01
-6.50484264e-01 1.55011296e-01 -1.04845971e-01 -2.72812039e-01
9.38104868e-01 -5.39968431e-01 -5.08813262e-01 -6.64409399e-01
-3.29933763e-01 1.08649865e-01 7.24918008e-01 4.48026210e-01
7.14511991e-01 -1.02771521e+00 -4.60596889e-01 2.44727403e-01
5.52862942e-01 -1.11326575e-01 4.16086316e-01 1.23269475e+00
-1.89601898e-01 2.03052744e-01 -1.23895273e-01 -8.96066487e-01
-1.21571672e+00 3.72513354e-01 2.81177491e-01 4.92286325e-01
2.73498714e-01 1.16746652e+00 8.10053758e-03 1.68380648e-01
5.82620025e-01 -5.32370031e-01 -3.67732197e-01 -1.48163021e-01
7.60151744e-01 6.52049541e-01 1.23162128e-01 -4.36457127e-01
-5.46106756e-01 5.35242319e-01 8.58747512e-02 -2.21791670e-01
7.09697425e-01 -7.34014869e-01 1.56577244e-01 6.24120057e-01
7.67787397e-01 6.09874465e-02 -1.61197150e+00 -1.20710187e-01
-1.00530021e-01 -7.03596652e-01 5.74573457e-01 -9.86014605e-01
-5.57662845e-01 8.62454295e-01 1.11087501e+00 2.26992145e-02
1.08360028e+00 -2.42975637e-01 1.05127744e-01 2.90851556e-02
6.27665579e-01 -1.16194069e+00 -4.10188824e-01 -3.99600565e-02
8.57497096e-01 -1.89149880e+00 3.95106584e-01 -3.88171881e-01
-7.30717242e-01 8.15903664e-01 1.07204926e+00 3.63332689e-01
9.27328169e-01 2.27078006e-01 1.72159329e-01 -7.63502941e-02
-7.59077966e-01 -1.19706310e-01 9.95251000e-01 9.23464894e-01
3.27372581e-01 -1.21692188e-01 -2.58490175e-01 5.12539446e-01
7.54912570e-03 2.90235311e-01 4.20570225e-01 4.51396137e-01
-3.23194087e-01 -5.18809915e-01 -3.36167574e-01 4.75549906e-01
-1.49260908e-01 1.24594513e-02 -4.62024733e-02 6.86489105e-01
7.06679225e-02 1.23386765e+00 3.51946771e-01 -8.34555745e-01
9.12659094e-02 -6.20378666e-02 6.93103373e-01 -2.15513960e-01
1.60796091e-01 -1.57420963e-01 2.66116977e-01 -7.62797832e-01
-7.45538473e-01 -6.17929101e-01 -7.74617732e-01 -6.07048392e-01
-1.03371167e+00 -1.22472197e-01 9.51668739e-01 1.09736753e+00
6.68370306e-01 1.36098966e-01 5.56091905e-01 -9.25051391e-01
-7.24319279e-01 -1.26098037e+00 -6.37009382e-01 -1.25559971e-01
5.99850655e-01 -1.41058040e+00 -7.44581759e-01 -4.08092231e-01]
|
[14.116935729980469, 0.06662078201770782]
|
1e83730f-d64c-4477-904c-007de26c861e
|
singing-voice-synthesis-using-differentiable
|
2306.17252
| null |
https://arxiv.org/abs/2306.17252v1
|
https://arxiv.org/pdf/2306.17252v1.pdf
|
Singing Voice Synthesis Using Differentiable LPC and Glottal-Flow-Inspired Wavetables
|
This paper introduces GlOttal-flow LPC Filter (GOLF), a novel method for singing voice synthesis (SVS) that exploits the physical characteristics of the human voice using differentiable digital signal processing. GOLF employs a glottal model as the harmonic source and IIR filters to simulate the vocal tract, resulting in an interpretable and efficient approach. We show it is competitive with state-of-the-art singing voice vocoders, requiring fewer synthesis parameters and less memory to train, and runs an order of magnitude faster for inference. Additionally, we demonstrate that GOLF can model the phase components of the human voice, which has immense potential for rendering and analysing singing voices in a differentiable manner. Our results highlight the effectiveness of incorporating the physical properties of the human voice mechanism into SVS and underscore the advantages of signal-processing-based approaches, which offer greater interpretability and efficiency in synthesis. Audio samples are available at https://yoyololicon.github.io/golf-demo/.
|
['György Fazekas', 'Chin-Yun Yu']
|
2023-06-29
| null | null | null | null |
['singing-voice-synthesis']
|
['speech']
|
[-1.07930019e-01 -7.92381726e-03 1.58592850e-01 2.50755429e-01
-7.48729527e-01 -8.28813612e-01 2.94318736e-01 -4.74090993e-01
1.57832518e-01 4.74899232e-01 5.39051116e-01 -4.06985193e-01
-2.92368990e-04 -2.66755879e-01 -5.52203894e-01 -6.09173298e-01
-2.01999247e-01 -1.40789568e-01 7.84561560e-02 -4.14973140e-01
-1.49674341e-01 4.57830012e-01 -1.89136016e+00 2.26885334e-01
4.30261105e-01 7.18443274e-01 2.45902196e-01 1.50492680e+00
2.81299233e-01 7.05911696e-01 -9.32970583e-01 -1.09212093e-01
1.18568947e-04 -8.63393664e-01 -5.80374062e-01 -1.20552808e-01
3.70547831e-01 -2.01734602e-01 -4.09486204e-01 5.30298889e-01
7.58169472e-01 3.87073457e-01 4.23784435e-01 -8.71101618e-01
-3.98612022e-01 4.03288603e-01 2.19994053e-01 1.58126652e-01
6.53057277e-01 3.73566717e-01 1.41616106e+00 -8.95099461e-01
3.52164030e-01 1.38793135e+00 8.32981646e-01 7.94290662e-01
-1.08707201e+00 -5.54004848e-01 -4.72320169e-01 -3.31428535e-02
-1.03496182e+00 -7.62097120e-01 9.46785629e-01 -2.90125638e-01
8.68337572e-01 9.31105196e-01 9.91389096e-01 7.62474954e-01
1.97929293e-02 8.99121165e-01 8.84591699e-01 -6.82559907e-01
5.43022156e-02 -2.18866646e-01 -2.96227187e-01 6.34383380e-01
-5.58662295e-01 5.66473246e-01 -9.13148701e-01 -2.78897166e-01
8.93289387e-01 -4.42552745e-01 -6.28954291e-01 3.16425383e-01
-1.09660709e+00 5.45431077e-01 2.11232412e-03 3.27997744e-01
-3.26276541e-01 5.98849714e-01 5.54442465e-01 2.65219688e-01
3.39606822e-01 7.51778424e-01 -1.80273980e-01 -5.31047106e-01
-1.25972807e+00 5.86737692e-01 1.02528548e+00 5.79416633e-01
-6.63122162e-02 7.44882584e-01 -4.07981910e-02 9.10587251e-01
3.87802541e-01 6.31908178e-01 4.33571935e-01 -1.56773686e+00
-5.66952080e-02 -3.80914599e-01 -1.50804088e-01 -6.79257989e-01
-5.85290864e-02 -4.65691179e-01 -2.93221503e-01 2.13532910e-01
3.58433306e-01 -1.10873632e-01 -4.22066927e-01 1.56636846e+00
3.78570765e-01 4.17196155e-01 -8.23643357e-02 1.00027239e+00
7.80314028e-01 9.31188166e-01 -3.01544577e-01 -4.07499552e-01
1.36169267e+00 -1.02811074e+00 -1.12486959e+00 2.06733897e-01
1.55417874e-01 -1.18230367e+00 1.51618767e+00 6.13950789e-01
-1.56984544e+00 -6.64326429e-01 -1.00687766e+00 -2.23269492e-01
1.07982427e-01 6.52442276e-02 6.55744791e-01 7.80598342e-01
-1.11441708e+00 8.71906161e-01 -8.51829588e-01 3.58704954e-01
-9.36302766e-02 1.92969009e-01 6.82870522e-02 7.16672540e-01
-1.15865898e+00 4.01654452e-01 -1.26369908e-01 7.70257041e-02
-9.80995417e-01 -1.29284334e+00 -7.85078406e-01 1.39138669e-01
1.30003989e-01 -4.71217990e-01 2.02343631e+00 -6.82137311e-01
-2.15541816e+00 2.52076566e-01 -4.90599185e-01 -3.36206347e-01
2.08009690e-01 -2.89521068e-01 -6.21949613e-01 5.35124660e-01
-4.76894975e-01 1.77794069e-01 1.24976206e+00 -1.02658629e+00
-3.12814891e-01 2.47933000e-01 -2.80998379e-01 9.21147987e-02
-7.80203864e-02 1.85597375e-01 6.72946572e-02 -1.00359571e+00
-2.03339100e-01 -7.60996044e-01 2.13404641e-01 4.64338288e-02
-2.33022407e-01 -1.06947698e-01 8.14371288e-01 -1.07660890e+00
1.78687191e+00 -2.32584882e+00 2.00594813e-02 -8.30564871e-02
2.31289506e-01 6.05788648e-01 3.93122584e-02 8.22091997e-01
-1.83581449e-02 -1.51952691e-02 -3.03924233e-01 -3.43720406e-01
8.62048268e-02 1.07630394e-01 -7.51906633e-01 2.79146522e-01
5.85846752e-02 7.71084368e-01 -9.62367833e-01 -2.45921746e-01
2.45941132e-01 8.48166943e-01 -8.68243158e-01 4.83136177e-01
-1.29498914e-01 5.78912437e-01 -6.72950670e-02 4.08220619e-01
8.44472796e-02 5.80814660e-01 1.80381071e-02 -2.35428557e-01
-3.57170463e-01 1.07140732e+00 -1.19022965e+00 1.28578448e+00
-8.65872324e-01 7.92433977e-01 6.73675954e-01 -4.12137806e-01
8.33357632e-01 9.04226899e-01 1.21743619e-01 -1.07607678e-01
6.09134324e-04 5.69448888e-01 1.33457795e-01 -5.06648064e-01
4.72656310e-01 -6.01495564e-01 3.50609839e-01 3.09267104e-01
2.54359782e-01 -9.63409305e-01 8.10508057e-02 -1.06792882e-01
7.38918602e-01 2.46276215e-01 1.72158778e-01 -4.00389075e-01
6.96690142e-01 -4.68585730e-01 3.87753338e-01 2.55203068e-01
-7.45712444e-02 6.45158827e-01 1.32410154e-01 2.86781061e-02
-1.00407350e+00 -1.22938514e+00 -7.65679404e-02 9.77748930e-01
-6.42518342e-01 -7.45974779e-01 -1.06195736e+00 2.52743542e-01
5.72063308e-03 9.71213520e-01 -2.88064275e-02 3.15843672e-02
-8.54712963e-01 1.09196603e-01 1.01997507e+00 5.00975728e-01
-5.95377944e-02 -1.33353603e+00 -4.50767934e-01 4.22257453e-01
-1.03806190e-01 -7.03394055e-01 -1.15514064e+00 -1.20331511e-01
-8.55944335e-01 -7.82745659e-01 -6.34090960e-01 -7.59300172e-01
4.00729681e-04 -1.96810197e-02 9.68114853e-01 1.45473421e-01
-4.66383278e-01 7.16087699e-01 -9.90274325e-02 -5.30760348e-01
-9.99927461e-01 -3.79471421e-01 4.50527221e-01 -1.53731033e-01
-3.04743439e-01 -9.29568768e-01 -6.23831332e-01 1.43775672e-01
-8.86174977e-01 -1.43243864e-01 -2.32399091e-01 9.05877650e-01
4.37921643e-01 1.52523011e-01 7.95027852e-01 -4.73201513e-01
1.05803204e+00 4.45377752e-02 -3.42139423e-01 -2.51432478e-01
-2.57850170e-01 -1.70616776e-01 1.14832354e+00 -5.29122710e-01
-9.33555305e-01 -2.13354796e-01 -7.14449704e-01 -4.82470661e-01
4.60708849e-02 1.30328521e-01 -1.34665268e-02 1.63752973e-01
5.64798892e-01 1.34337336e-01 4.14373308e-01 -7.30672300e-01
5.76594532e-01 9.37907040e-01 9.97348487e-01 -6.38564646e-01
9.43157136e-01 2.03085631e-01 4.39637825e-02 -1.52419817e+00
-3.67442429e-01 -4.97928262e-01 -4.37094063e-01 -4.29670483e-01
5.67101657e-01 -6.61672473e-01 -1.19824219e+00 5.31995773e-01
-1.12936831e+00 -4.05919343e-01 -7.91483760e-01 6.82040155e-01
-9.81888473e-01 4.80757654e-01 -1.01954758e+00 -1.39165533e+00
-5.87934732e-01 -8.24516356e-01 9.55736995e-01 1.69250280e-01
-6.66940510e-01 -1.15200698e+00 -3.88470702e-02 4.16458488e-01
3.05860192e-01 1.51782915e-01 8.42475176e-01 -5.12541309e-02
-9.49379429e-02 7.36133829e-02 6.64940000e-01 7.94839263e-01
2.05659598e-01 3.52519423e-01 -1.36781371e+00 -1.76248372e-01
2.56356537e-01 -1.40907198e-01 4.14945275e-01 4.57637042e-01
9.15282667e-01 -6.56981170e-01 3.77033651e-01 5.28626561e-01
7.90173888e-01 1.29060119e-01 4.69856143e-01 -4.48755592e-01
5.20825684e-01 6.95337713e-01 4.64421064e-01 4.03804988e-01
-5.00674732e-02 4.84186023e-01 3.97226177e-02 -1.29779741e-01
-7.79571891e-01 -7.93466985e-01 6.00079060e-01 1.70940423e+00
-3.27042431e-01 -3.70194092e-02 -3.55042189e-01 6.79063082e-01
-1.14946473e+00 -1.11577737e+00 -2.33014628e-01 2.18480921e+00
1.19361401e+00 -1.69763163e-01 4.47136343e-01 8.65812182e-01
5.46084762e-01 3.28877658e-01 -1.93568870e-01 -1.04849625e+00
1.63135633e-01 8.84195685e-01 1.13487363e-01 1.09301269e+00
-6.11462355e-01 8.60103071e-01 7.01615477e+00 9.89927530e-01
-1.18412590e+00 -6.09306097e-02 -3.12118288e-02 -1.72820330e-01
-5.94000220e-01 -1.71628878e-01 -5.42296708e-01 2.32218057e-01
1.29771769e+00 -2.87063450e-01 1.25450480e+00 4.95059013e-01
8.34469438e-01 4.29210007e-01 -8.55635285e-01 7.99510300e-01
-1.49611861e-01 -1.12075353e+00 -1.95778474e-01 -1.95293412e-01
2.30235413e-01 -5.57437956e-01 1.95628926e-01 -2.16044169e-02
-4.35929328e-01 -1.01159835e+00 1.21265483e+00 4.29043591e-01
9.27992702e-01 -7.36957073e-01 5.13069257e-02 3.46073925e-01
-1.47918332e+00 4.11719121e-02 -2.36703064e-02 -2.63971269e-01
4.64587837e-01 3.18653971e-01 -1.04789591e+00 3.16497386e-01
4.77613688e-01 4.77395773e-01 6.59966618e-02 8.83187234e-01
-4.79720235e-01 1.54844153e+00 -3.99730116e-01 -1.01977043e-01
-2.72273868e-02 -3.92598063e-02 1.12730217e+00 1.39331055e+00
3.68871748e-01 2.37640977e-01 -3.12917173e-01 9.74541187e-01
1.06866889e-01 1.24366477e-01 -2.95154899e-01 -6.38400316e-01
5.29911697e-01 1.07828820e+00 -2.78349686e-02 -4.18584570e-02
-8.81908387e-02 7.28014052e-01 -5.49534678e-01 3.59671742e-01
-6.39133990e-01 -5.80685318e-01 9.03624415e-01 5.02618968e-01
2.78815478e-01 -5.62542260e-01 -7.54516572e-02 -7.42512763e-01
-1.01492628e-01 -1.30769897e+00 -4.85777073e-02 -8.41767490e-01
-8.96653473e-01 6.21301472e-01 -2.13006258e-01 -1.26506841e+00
-6.51794016e-01 -7.61562109e-01 -8.72541666e-01 1.05587506e+00
-1.24656999e+00 -8.64749730e-01 2.42870107e-01 2.90649533e-01
7.46175706e-01 9.47704166e-02 1.04770625e+00 3.35470438e-01
-5.56358732e-02 5.12886822e-01 1.28757462e-01 -2.54615247e-01
3.86727214e-01 -1.41613519e+00 7.74240494e-01 5.52351654e-01
2.28944868e-01 7.62405694e-01 1.03663266e+00 -2.56057411e-01
-1.72085547e+00 -5.67101181e-01 9.78604615e-01 -3.00046831e-01
8.52523386e-01 -4.28097308e-01 -1.01279986e+00 2.16257438e-01
2.35005200e-01 -2.97405124e-01 8.51567745e-01 -1.94395572e-01
-1.26274750e-01 6.08590990e-03 -9.70681310e-01 7.16966629e-01
8.84115160e-01 -9.48519230e-01 -8.08442771e-01 1.14903919e-01
8.72462988e-01 -5.38302481e-01 -9.95613337e-01 4.27297875e-02
8.59973431e-01 -7.91539729e-01 1.10556924e+00 -4.30827975e-01
2.48657301e-01 -4.99977589e-01 -5.68224527e-02 -1.41315877e+00
-1.88033670e-01 -1.56217158e+00 -4.86667395e-01 1.37760460e+00
2.91363537e-01 -7.76392817e-01 1.14245370e-01 1.77086994e-01
-4.65111554e-01 -7.09752202e-01 -8.73175800e-01 -9.91866648e-01
1.14663586e-01 -7.05441952e-01 5.84188282e-01 4.39231098e-01
2.86440760e-01 1.81717560e-01 -5.11415243e-01 8.91342685e-02
4.12435383e-01 -1.41400725e-01 4.88082767e-01 -8.59291553e-01
-9.11599696e-01 -3.78958255e-01 -1.48554426e-02 -1.17502046e+00
8.21593031e-02 -7.83170223e-01 2.26091281e-01 -1.12185192e+00
-8.85243833e-01 -5.09366277e-04 -5.29102758e-02 1.42654955e-01
-8.84085000e-02 3.45205426e-01 6.33555114e-01 5.50168157e-02
3.28335643e-01 5.99350095e-01 1.71339023e+00 1.58813700e-01
-4.67629313e-01 3.60082597e-01 -3.26176286e-01 8.30696166e-01
7.41587222e-01 -2.13667512e-01 -4.16204035e-01 1.93976518e-02
-3.59999001e-01 3.74062777e-01 4.37992215e-01 -1.01679873e+00
6.48744479e-02 2.25372434e-01 -5.02151176e-02 -4.36170727e-01
7.95099020e-01 -4.16531175e-01 2.82193005e-01 5.18293619e-01
-4.05736864e-01 -1.52477160e-01 4.79399711e-01 1.77312583e-01
-3.22850078e-01 -3.13326150e-01 8.08152914e-01 4.96951081e-02
-1.63382575e-01 -2.07391053e-01 -7.37274826e-01 1.03543840e-01
2.56746203e-01 -9.43360776e-02 2.81023979e-02 -7.34242320e-01
-6.67031586e-01 -3.12933087e-01 -2.38992572e-02 3.15888524e-01
5.66129267e-01 -1.20014858e+00 -6.09083712e-01 4.15996939e-01
-5.25455415e-01 -4.24436003e-01 2.24884629e-01 6.14828050e-01
-8.98124635e-01 5.00151098e-01 2.60124832e-01 -3.40507776e-01
-1.47279024e+00 2.29475141e-01 4.96300280e-01 2.61390388e-01
-8.21456611e-01 7.97461629e-01 2.14763600e-02 -4.61575985e-01
1.06551230e-01 -4.51353222e-01 1.73673630e-01 -2.26618424e-01
7.15329945e-01 7.73785651e-01 -1.62135288e-01 -6.14715040e-01
-1.37992069e-01 4.78556216e-01 6.17933095e-01 -5.16678631e-01
1.08011413e+00 -4.39346954e-02 -6.50492087e-02 7.48187304e-01
1.05318117e+00 8.09121907e-01 -1.09054506e+00 2.24412650e-01
-3.60877335e-01 -4.65439051e-01 2.98216283e-01 -7.38215327e-01
-6.64332092e-01 1.11662102e+00 1.14940874e-01 4.37235296e-01
1.30057907e+00 -2.59144962e-01 1.37504959e+00 -7.39418119e-02
5.63354231e-02 -1.01912630e+00 -1.25100195e-01 4.47092921e-01
1.25215364e+00 -1.88893706e-01 -2.44944200e-01 -7.70011127e-01
-6.00043356e-01 1.40957415e+00 -1.06230602e-01 -2.62444973e-01
7.30568767e-01 5.77343941e-01 3.50893080e-01 2.41702259e-01
-6.78138673e-01 -8.72355849e-02 6.27807200e-01 6.52229071e-01
8.94013822e-01 2.77694166e-01 -4.76181865e-01 6.28857136e-01
-1.00376964e+00 -1.38747454e-01 3.77944291e-01 5.08594036e-01
-4.08621550e-01 -1.28140926e+00 -6.45661950e-01 6.50549755e-02
-7.38315523e-01 -3.92433643e-01 -3.16018760e-01 3.75349700e-01
-1.60646170e-01 1.27133512e+00 -1.98863387e-01 -2.76672781e-01
5.61176360e-01 4.04897422e-01 5.53847909e-01 -4.50901568e-01
-1.11370218e+00 6.07858360e-01 3.32937360e-01 -3.44522774e-01
9.37488955e-03 -6.16700888e-01 -1.53348649e+00 -3.79840314e-01
-2.92523950e-01 4.91452217e-01 7.65060723e-01 5.60624897e-01
3.25179785e-01 9.47349012e-01 9.38845038e-01 -9.22674060e-01
-7.46270001e-01 -7.53728807e-01 -8.55927646e-01 1.40975535e-01
7.83226252e-01 -2.68135816e-01 -8.30733478e-01 2.57935554e-01]
|
[15.450562477111816, 6.1111531257629395]
|
589fca31-486a-4c0e-a822-b54502de9c7c
|
frsum-towards-faithful-abstractive-1
|
2211.00294
| null |
https://arxiv.org/abs/2211.00294v1
|
https://arxiv.org/pdf/2211.00294v1.pdf
|
FRSUM: Towards Faithful Abstractive Summarization via Enhancing Factual Robustness
|
Despite being able to generate fluent and grammatical text, current Seq2Seq summarization models still suffering from the unfaithful generation problem. In this paper, we study the faithfulness of existing systems from a new perspective of factual robustness which is the ability to correctly generate factual information over adversarial unfaithful information. We first measure a model's factual robustness by its success rate to defend against adversarial attacks when generating factual information. The factual robustness analysis on a wide range of current systems shows its good consistency with human judgments on faithfulness. Inspired by these findings, we propose to improve the faithfulness of a model by enhancing its factual robustness. Specifically, we propose a novel training strategy, namely FRSUM, which teaches the model to defend against both explicit adversarial samples and implicit factual adversarial perturbations. Extensive automatic and human evaluation results show that FRSUM consistently improves the faithfulness of various Seq2Seq models, such as T5, BART.
|
['Hua Wu', 'Sujian Li', 'Ziqiang Cao', 'Xinyan Xiao', 'Jiachen Liu', 'Wei Li', 'Wenhao Wu']
|
2022-11-01
| null | null | null | null |
['abstractive-text-summarization']
|
['natural-language-processing']
|
[ 3.92856747e-01 6.83866382e-01 -6.04159497e-02 -3.54456723e-01
-1.00528121e+00 -1.08408546e+00 1.06290483e+00 1.50416018e-02
-9.31014419e-02 1.20892763e+00 8.39516163e-01 -1.62257269e-01
2.52868354e-01 -8.44212830e-01 -9.14294243e-01 -3.20518732e-01
2.34014392e-01 2.38365367e-01 -8.58120620e-02 -7.84960568e-01
4.82605278e-01 2.78268427e-01 -9.55968976e-01 5.37979960e-01
1.43211210e+00 3.01623642e-01 -4.20739591e-01 9.22873974e-01
3.91018152e-01 1.40078163e+00 -1.34350359e+00 -1.14776659e+00
2.01326326e-01 -6.74704015e-01 -1.20838475e+00 -4.46899503e-01
1.03002894e+00 -5.00586689e-01 -3.99709731e-01 1.27625048e+00
7.75737703e-01 1.36455834e-01 6.16238236e-01 -1.15840626e+00
-6.82278872e-01 1.17267096e+00 6.50037602e-02 4.14245933e-01
8.10637057e-01 7.21128047e-01 1.04280257e+00 -1.50737494e-01
7.45865166e-01 1.52320468e+00 6.59609318e-01 1.09858167e+00
-8.88160646e-01 -6.13965154e-01 5.97961880e-02 4.23391676e-03
-7.36314833e-01 -8.15998137e-01 7.49478817e-01 -1.66434526e-01
8.82609487e-01 7.52639294e-01 2.63011396e-01 1.59726644e+00
7.44091690e-01 7.82711387e-01 8.96613300e-01 -4.38088328e-02
2.76945680e-01 1.50630353e-02 -8.63543153e-02 6.12176776e-01
2.44915903e-01 3.53409231e-01 -6.26332462e-01 -3.41286451e-01
2.11760879e-01 -7.30598807e-01 -6.34273589e-01 3.76583278e-01
-1.06903422e+00 9.04512942e-01 3.90658021e-01 1.44893050e-01
-3.31415147e-01 1.05619803e-01 7.17799067e-01 2.88951904e-01
4.79913741e-01 1.17862689e+00 -2.50692487e-01 -2.73390561e-01
-1.01117373e+00 5.29860675e-01 1.00678098e+00 7.91217327e-01
1.60436124e-01 5.72572470e-01 -6.88507438e-01 5.60694516e-01
-2.54556179e-01 6.66917980e-01 6.50631011e-01 -9.51704860e-01
7.57576168e-01 3.29207629e-01 -7.17192665e-02 -1.06612730e+00
-7.14107305e-02 -5.28907239e-01 -1.11845171e+00 -1.36931300e-01
1.42090276e-01 -4.22344297e-01 -5.81729412e-01 2.05351591e+00
-2.81004254e-02 7.77650811e-03 5.90899169e-01 7.13891745e-01
8.95916879e-01 8.59828293e-01 -1.15027111e-02 -2.21869588e-01
8.98733854e-01 -7.43957520e-01 -8.09946418e-01 -3.82929415e-01
3.62796336e-01 -6.15173995e-01 1.03811300e+00 3.96662533e-01
-1.43853688e+00 -4.09706980e-01 -1.18633616e+00 1.19973958e-01
-8.81580859e-02 -4.25241858e-01 4.43114847e-01 8.06954503e-01
-8.80034804e-01 8.70910645e-01 -4.25276548e-01 1.69022493e-02
5.52848220e-01 -7.54574239e-02 -5.71632922e-01 1.21717684e-01
-1.88738775e+00 1.32478452e+00 7.62692750e-01 -1.58248842e-01
-1.19657993e+00 -8.25681865e-01 -9.68786895e-01 8.36737454e-02
2.56268114e-01 -9.18090045e-01 1.45780611e+00 -1.06291795e+00
-1.70026791e+00 5.40615320e-01 2.82371223e-01 -1.05473661e+00
8.89026582e-01 -4.15313274e-01 -4.02343780e-01 3.77303094e-01
-3.12032942e-02 4.79469061e-01 8.90819907e-01 -1.33473086e+00
-8.27200338e-02 7.77216032e-02 3.69046688e-01 1.57430589e-01
-2.85032570e-01 -1.73122332e-01 3.13879430e-01 -1.02007115e+00
-4.61148471e-01 -7.14971721e-01 -2.77375653e-02 -7.24088550e-01
-1.11561596e+00 -8.21693912e-02 3.73671353e-01 -6.65659785e-01
1.37605917e+00 -1.59026921e+00 -2.76462324e-02 -1.96028113e-01
6.44894093e-02 8.51551533e-01 -3.89896393e-01 7.14047432e-01
-2.34240308e-01 5.61766744e-01 -4.56157506e-01 1.62419081e-01
8.95080343e-02 1.73074156e-01 -1.01018918e+00 1.82891414e-01
5.08874536e-01 1.16896486e+00 -1.29343510e+00 -4.53614950e-01
-2.30544973e-02 1.76521301e-01 -6.43193245e-01 4.27208066e-01
-3.80153984e-01 1.75464123e-01 -2.65847385e-01 2.40867600e-01
6.16907656e-01 2.32436419e-01 6.40411079e-02 -9.36061665e-02
1.92580283e-01 5.65296769e-01 -5.54487467e-01 1.50026715e+00
-2.70696193e-01 5.22256374e-01 -3.96962851e-01 -6.10417902e-01
7.03374088e-01 4.48422551e-01 -3.26581925e-01 -6.77052498e-01
1.19584441e-01 2.27923915e-02 8.63107666e-02 -4.87624377e-01
9.98632669e-01 -5.83602965e-01 -4.52737898e-01 5.15532792e-01
3.30492891e-02 -5.75684369e-01 2.29046479e-01 9.16929781e-01
1.09811389e+00 -4.66198064e-02 4.62198585e-01 -3.69388968e-01
5.82861662e-01 -5.56042232e-02 5.82162142e-01 1.03428936e+00
-3.47027719e-01 6.57635510e-01 8.08676779e-01 -2.92914450e-01
-8.65199566e-01 -1.19906032e+00 2.33598441e-01 7.66339242e-01
-1.76011890e-01 -4.19558346e-01 -1.24370396e+00 -1.21331978e+00
-1.97365254e-01 1.48627114e+00 -8.48874509e-01 -9.43544269e-01
-6.97379768e-01 -6.02933586e-01 1.37098849e+00 3.25093806e-01
7.67044127e-01 -1.15536976e+00 -5.59088171e-01 1.05686262e-01
-8.22153270e-01 -8.32021356e-01 -6.53910041e-01 -3.28056723e-01
-7.08755791e-01 -1.06500852e+00 -1.98122680e-01 -7.96670914e-02
2.84787595e-01 -1.19717166e-01 1.45780396e+00 -5.55476174e-03
1.21223919e-01 2.79624835e-02 -4.06155467e-01 -6.20638609e-01
-1.32859969e+00 1.22049391e-01 5.43914028e-02 -4.60791796e-01
-3.05450439e-01 -2.55957007e-01 -3.64934474e-01 2.34085806e-02
-1.35405064e+00 -5.91897853e-02 4.02432054e-01 8.30792427e-01
2.15219185e-02 1.22205734e-01 9.10662532e-01 -1.23290610e+00
1.17089486e+00 -4.70318675e-01 1.49576142e-01 3.84178162e-01
-2.65479535e-01 2.62652874e-01 1.30202675e+00 -3.39774936e-01
-1.20642745e+00 -6.81959808e-01 -4.73406553e-01 -1.02444142e-01
-1.09938368e-01 2.91580111e-01 -3.72228384e-01 2.23610640e-01
1.23858702e+00 3.77522409e-01 -9.36420858e-02 1.05394103e-01
6.78779006e-01 4.04590696e-01 1.01610339e+00 -7.29692876e-01
9.91946340e-01 2.02618733e-01 -3.11259210e-01 -5.55963278e-01
-1.23842967e+00 2.93997735e-01 -2.13242650e-01 -2.14819610e-01
3.70334148e-01 -6.13991678e-01 -7.68287361e-01 5.11104405e-01
-1.39053941e+00 -1.87623397e-01 -3.77261579e-01 -6.84774518e-02
-6.88367128e-01 6.75991297e-01 -5.74930549e-01 -6.31843925e-01
-9.47652757e-01 -6.84060514e-01 8.17838490e-01 4.80165631e-02
-6.47889555e-01 -1.11348319e+00 3.55118275e-01 4.65672076e-01
4.36281174e-01 8.36938202e-01 8.67222011e-01 -9.67786610e-01
-1.02468863e-01 -3.91959608e-01 2.24081203e-01 7.22375691e-01
8.20773374e-03 1.55493200e-01 -9.37404811e-01 -3.26790929e-01
2.46619076e-01 -6.04484200e-01 9.25828695e-01 -1.85307518e-01
7.03948319e-01 -1.24331713e+00 2.68979758e-01 3.36267799e-01
1.09073853e+00 -2.12346911e-01 1.12516761e+00 1.10663824e-01
4.96113569e-01 6.88373506e-01 6.30112350e-01 3.25908214e-01
1.98267415e-01 3.11336875e-01 6.74477816e-01 2.17995182e-01
-9.21536833e-02 -7.62639821e-01 7.50400901e-01 7.05661714e-01
1.14437148e-01 -7.93635070e-01 -6.55094385e-01 3.19371551e-01
-1.62283719e+00 -1.76565969e+00 1.48115069e-01 1.93975317e+00
1.30405319e+00 4.98545021e-01 3.40998732e-03 2.18265235e-01
5.86597800e-01 6.39481783e-01 -3.67755413e-01 -9.68639135e-01
-4.46475714e-01 -8.08219910e-02 1.54260755e-01 7.57653356e-01
-9.52959418e-01 1.30308461e+00 6.83187103e+00 1.03890383e+00
-1.05652547e+00 -1.73362240e-01 6.32676899e-01 -3.19234073e-01
-7.31435597e-01 -2.43881419e-01 -3.46818179e-01 6.53324187e-01
9.95662153e-01 -5.49683869e-01 2.47420520e-01 4.97748703e-01
1.82259873e-01 6.70569837e-02 -1.11225367e+00 2.00945944e-01
4.22533482e-01 -1.42662144e+00 8.38463604e-01 -4.44498032e-01
7.51007438e-01 -4.45766687e-01 1.10010959e-01 4.38389391e-01
8.01593542e-01 -1.23938119e+00 9.93458569e-01 6.40555143e-01
5.59919298e-01 -9.77252901e-01 8.37308705e-01 7.27096856e-01
-2.80434340e-01 7.84284547e-02 -3.15051049e-01 -5.18479049e-02
2.33896375e-01 4.11642194e-01 -1.03437734e+00 8.50615263e-01
8.75341445e-02 4.20090765e-01 -6.17707729e-01 2.79094309e-01
-6.48632109e-01 7.61701226e-01 6.98958561e-02 -5.31747229e-02
3.06885064e-01 3.34532797e-01 8.52696657e-01 1.49415886e+00
-1.07849441e-01 2.40165189e-01 -1.95355847e-01 8.09406042e-01
-3.28047216e-01 -2.18650147e-01 -8.63466322e-01 -1.31458923e-01
5.21380246e-01 1.03764689e+00 2.45573614e-02 -4.98371243e-01
4.60707098e-01 1.11233091e+00 5.06565928e-01 6.18539602e-02
-9.15426195e-01 -3.54076266e-01 3.84217203e-01 -2.35174328e-01
-2.78565735e-01 1.95260718e-01 -4.06002730e-01 -1.11810994e+00
-3.93673107e-02 -1.57122433e+00 5.23291945e-01 -8.30795586e-01
-1.35303187e+00 8.54944289e-01 -4.76954207e-02 -1.02882051e+00
-4.55819756e-01 -6.40530884e-02 -1.14518893e+00 6.73089445e-01
-1.31262422e+00 -1.08161926e+00 1.59802869e-01 5.00522554e-01
4.90173787e-01 -2.54552722e-01 7.72316456e-01 -4.09772307e-01
-6.52763903e-01 1.09857070e+00 -4.08007860e-01 9.30003151e-02
7.51157880e-01 -1.37580574e+00 8.10421050e-01 1.30111694e+00
-7.51356333e-02 9.74493206e-01 1.28003740e+00 -8.58571053e-01
-1.28180528e+00 -1.23090374e+00 9.81619120e-01 -7.55798519e-01
5.62144041e-01 -8.50091204e-02 -9.81443524e-01 5.58694601e-01
5.54322779e-01 -6.25268698e-01 6.33761525e-01 -1.93823352e-01
-7.74297059e-01 1.74527153e-01 -1.43115568e+00 7.85275638e-01
9.37483609e-01 -4.86055344e-01 -1.19726920e+00 3.75298440e-01
8.43430161e-01 -5.30400336e-01 -6.86804295e-01 5.33306837e-01
2.75668770e-01 -1.04099774e+00 8.96358550e-01 -1.13030553e+00
1.08768916e+00 2.89121009e-02 -1.30756661e-01 -1.77293408e+00
-4.01998997e-01 -1.04028893e+00 -7.12960064e-02 1.42363560e+00
2.87991613e-01 -6.16706073e-01 4.78286445e-01 4.49855506e-01
-3.39682639e-01 -4.53862578e-01 -9.27279532e-01 -9.31053817e-01
7.63970792e-01 -1.17439136e-01 6.72173917e-01 1.23187470e+00
3.71198684e-01 4.87977087e-01 -6.33389473e-01 2.14186966e-01
4.60804582e-01 -1.00103900e-01 8.27298880e-01 -6.33421481e-01
-2.35190451e-01 -4.61113423e-01 -1.96257606e-01 -4.93937463e-01
5.53866088e-01 -1.02609837e+00 2.85591006e-01 -1.38632810e+00
1.85315266e-01 2.35320672e-01 9.73657444e-02 4.86238360e-01
-8.11547697e-01 1.76078230e-01 2.89728701e-01 -2.29060464e-03
-5.58134675e-01 7.32396424e-01 1.30874670e+00 -2.15193436e-01
1.85498521e-01 -2.03662798e-01 -1.13552952e+00 7.33678639e-01
1.01095247e+00 -4.03731704e-01 -4.50901330e-01 -4.27823424e-01
2.96672940e-01 7.64785260e-02 5.13465345e-01 -8.85280788e-01
1.40386950e-02 -3.20606768e-01 4.44234870e-02 -9.08463225e-02
-7.50481039e-02 -1.76067650e-01 -1.34385610e-02 7.94443965e-01
-6.84484959e-01 -1.12300850e-02 3.90663743e-01 4.58131611e-01
-1.85060948e-01 -2.67273545e-01 9.26234722e-01 -2.14592412e-01
-2.72515714e-01 6.22147657e-02 -3.76041859e-01 7.46489286e-01
8.19331348e-01 1.04463875e-01 -1.10297167e+00 -5.98894477e-01
-2.00500295e-01 2.76910692e-01 5.01638472e-01 4.00551915e-01
6.50291800e-01 -1.17359233e+00 -1.39987671e+00 -4.71427441e-02
-3.76910940e-02 -4.57900405e-01 4.74430978e-01 1.65446192e-01
-5.05422473e-01 2.75346518e-01 -2.89253265e-01 -3.41496803e-02
-1.13259172e+00 5.84243894e-01 5.43473899e-01 -4.34423417e-01
-2.57355332e-01 8.28621089e-01 -6.66429400e-02 -5.11257172e-01
-1.40055612e-01 1.54126018e-01 -1.64728895e-01 -9.40408930e-02
6.22522414e-01 6.22080803e-01 3.12643275e-02 -6.00155771e-01
-2.38329649e-01 1.65181849e-02 -4.41549629e-01 -1.48803845e-01
8.03654313e-01 3.51888776e-01 -8.61733258e-02 4.57229652e-02
9.23292220e-01 3.25615555e-01 -1.20203125e+00 7.46039748e-02
-2.85491258e-01 -4.20183599e-01 -3.70427668e-01 -1.38072860e+00
-6.70452118e-01 9.16662991e-01 -2.86023110e-01 5.66389740e-01
9.67752695e-01 -4.69018698e-01 9.23891425e-01 4.69084412e-01
1.16124012e-01 -9.87534046e-01 2.23428965e-01 8.99409890e-01
1.44688523e+00 -9.86698687e-01 9.94172171e-02 -3.53436261e-01
-1.14025366e+00 8.23414087e-01 6.75650477e-01 -2.17134818e-01
-2.16461092e-01 5.74105494e-02 2.84194618e-01 2.35632304e-02
-1.03807104e+00 4.75136578e-01 3.44474643e-01 5.92645168e-01
4.15018380e-01 1.02973670e-01 -3.65443349e-01 6.55432165e-01
-1.10305941e+00 -4.02927369e-01 8.60289037e-01 5.19181371e-01
-4.78399515e-01 -6.78936362e-01 -2.98671961e-01 2.17961743e-01
-7.95139909e-01 -2.72617221e-01 -9.98853326e-01 6.38290107e-01
-2.59520829e-01 1.31669104e+00 -3.04761380e-01 -4.55207646e-01
5.52465618e-01 2.44964380e-02 4.52029437e-01 -5.09606063e-01
-1.31158745e+00 -7.00860560e-01 5.71407378e-01 -4.82607633e-01
-7.59655237e-02 -4.91195798e-01 -1.09211981e+00 -8.44486475e-01
-1.61022455e-01 2.75593579e-01 2.27651402e-01 9.41526234e-01
2.82496184e-01 5.84415197e-01 7.75192440e-01 -3.50305438e-01
-1.20391536e+00 -1.05928946e+00 -2.14718189e-03 8.17353249e-01
3.44333827e-01 8.44113231e-02 -4.80911434e-01 -2.10872777e-02]
|
[6.179298400878906, 8.184976577758789]
|
b51cd257-acc6-47d3-9b88-494296d9113c
|
inter-rater-uncertainty-quantification-in
|
2306.16556
| null |
https://arxiv.org/abs/2306.16556v1
|
https://arxiv.org/pdf/2306.16556v1.pdf
|
Inter-Rater Uncertainty Quantification in Medical Image Segmentation via Rater-Specific Bayesian Neural Networks
|
Automated medical image segmentation inherently involves a certain degree of uncertainty. One key factor contributing to this uncertainty is the ambiguity that can arise in determining the boundaries of a target region of interest, primarily due to variations in image appearance. On top of this, even among experts in the field, different opinions can emerge regarding the precise definition of specific anatomical structures. This work specifically addresses the modeling of segmentation uncertainty, known as inter-rater uncertainty. Its primary objective is to explore and analyze the variability in segmentation outcomes that can occur when multiple experts in medical imaging interpret and annotate the same images. We introduce a novel Bayesian neural network-based architecture to estimate inter-rater uncertainty in medical image segmentation. Our approach has three key advancements. Firstly, we introduce a one-encoder-multi-decoder architecture specifically tailored for uncertainty estimation, enabling us to capture the rater-specific representation of each expert involved. Secondly, we propose Bayesian modeling for the new architecture, allowing efficient capture of the inter-rater distribution, particularly in scenarios with limited annotations. Lastly, we enhance the rater-specific representation by integrating an attention module into each decoder. This module facilitates focused and refined segmentation results for each rater. We conduct extensive evaluations using synthetic and real-world datasets to validate our technical innovations rigorously. Our method surpasses existing baseline methods in five out of seven diverse tasks on the publicly available \emph{QUBIQ} dataset, considering two evaluation metrics encompassing different uncertainty aspects. Our codes, models, and the new dataset are available through our GitHub repository: https://github.com/HaoWang420/bOEMD-net .
|
['Hongwei Bran Li', 'JianGuo Zhang', 'Bjoern Menze', 'Benedikt Wiestler', 'Jan S. Kirschke', 'Zhiheng Zhangg', 'Yunhao Luo', 'Jing Luo', 'Hao Wang', 'Qingqiao Hu']
|
2023-06-28
| null | null | null | null |
['medical-image-segmentation']
|
['medical']
|
[ 3.83423924e-01 3.47385645e-01 -1.97437182e-01 -7.18160152e-01
-1.55333722e+00 -6.56204462e-01 1.91474870e-01 1.98458061e-01
-4.91330653e-01 5.81221879e-01 4.48987812e-01 -2.56225169e-01
-6.07054941e-02 -1.91081896e-01 -7.74096906e-01 -4.80030507e-01
1.44093335e-01 5.93096673e-01 2.30294079e-01 3.03870052e-01
-3.05453222e-02 8.94574523e-02 -7.29794860e-01 4.96001512e-01
8.50383222e-01 1.20697343e+00 4.47037429e-01 7.08705008e-01
1.65363014e-01 6.94880366e-01 -5.70800662e-01 -6.27751946e-01
3.63956057e-02 -1.66854367e-01 -1.01957357e+00 1.18963227e-01
1.17341302e-01 -4.71481085e-01 -3.63720238e-01 1.21209812e+00
7.37815917e-01 -2.63344437e-01 8.53806674e-01 -7.30148613e-01
-5.13215661e-01 1.18571723e+00 -7.11714029e-01 5.24261296e-01
1.10737197e-01 1.23387545e-01 7.85532534e-01 -5.27324498e-01
3.82920265e-01 1.03138351e+00 5.70626020e-01 5.96437752e-01
-1.12324584e+00 -3.93638849e-01 1.56079918e-01 -3.49422768e-02
-1.59644926e+00 -2.89033324e-01 3.71576548e-01 -5.25387943e-01
4.33573306e-01 2.03002855e-01 2.17468351e-01 1.31081402e+00
6.30808175e-01 1.05283141e+00 9.98529196e-01 -1.87769067e-02
2.31241450e-01 4.88554239e-02 5.64535707e-02 4.98415470e-01
1.07324742e-01 5.45566007e-02 -2.33400196e-01 2.25395039e-02
6.95180893e-01 -1.77634567e-01 -6.31536067e-01 -4.12337109e-02
-1.17857838e+00 6.07600689e-01 4.57767099e-01 1.74087450e-01
-2.40602791e-01 4.59166884e-01 4.54493016e-01 -2.04113171e-01
3.86084408e-01 3.04172516e-01 -5.04307687e-01 -3.96980673e-01
-1.07132125e+00 -2.54737347e-01 6.93176985e-01 1.06295896e+00
5.42837605e-02 -3.29233676e-01 -7.21768916e-01 8.82336974e-01
5.59849381e-01 3.10343862e-01 4.67944652e-01 -9.39655840e-01
4.55498517e-01 -1.75274932e-03 -1.02863446e-01 -6.08099222e-01
-5.36419749e-01 -7.15532899e-01 -8.38959098e-01 1.31136691e-02
3.83289665e-01 -3.33363950e-01 -1.31256175e+00 1.83192253e+00
9.88021195e-02 1.89881206e-01 -2.01239571e-01 9.85190332e-01
1.02316904e+00 3.70018154e-01 2.00522810e-01 -8.34282197e-04
1.72827542e+00 -8.65401089e-01 -6.82992339e-01 -2.98392743e-01
3.50762278e-01 -7.15026498e-01 7.21989751e-01 4.73115176e-01
-1.00950634e+00 -2.49246463e-01 -9.22109783e-01 8.06377381e-02
1.28717199e-01 2.68059939e-01 3.54838490e-01 7.50048995e-01
-9.73012745e-01 4.54848886e-01 -1.02799749e+00 2.06392571e-01
7.97288895e-01 3.53258669e-01 -9.16227698e-02 -1.84762001e-01
-1.33235681e+00 9.13853824e-01 5.12309849e-01 3.28214526e-01
-1.20765281e+00 -1.03314924e+00 -9.61149573e-01 -1.49696514e-01
6.60025954e-01 -6.91426933e-01 1.65782166e+00 -6.69626594e-01
-1.24349773e+00 7.62895584e-01 1.97695866e-02 -4.26241159e-01
9.23600912e-01 -1.63369149e-01 -2.64000773e-01 1.49728045e-01
1.14652619e-01 8.76881301e-01 7.72618830e-01 -1.33143854e+00
-4.78796721e-01 -1.73045874e-01 -1.06541105e-02 1.73640385e-01
3.81058782e-01 -1.07427850e-01 -1.05981827e+00 -9.43305910e-01
-3.59247923e-02 -1.02935553e+00 -5.18652737e-01 -9.10117999e-02
-6.73797309e-01 2.39308044e-01 9.33892617e-04 -7.25802541e-01
1.32194018e+00 -2.14391255e+00 1.36429042e-01 1.03875354e-01
5.03323019e-01 -6.62484244e-02 1.09593347e-01 -2.90739954e-01
-7.40663186e-02 3.99623305e-01 -8.04361463e-01 -3.93308580e-01
-2.04830498e-01 2.00777531e-01 3.93875986e-02 3.36341292e-01
3.24628770e-01 9.14438546e-01 -9.23162222e-01 -6.63337290e-01
2.37940893e-01 6.01989925e-01 -4.41587478e-01 3.11618056e-02
-1.84655026e-01 8.93345654e-01 -5.96847296e-01 6.63840413e-01
6.63047850e-01 -5.58887303e-01 6.49750436e-05 -5.71682215e-01
3.46220076e-01 4.52725925e-02 -1.16052353e+00 2.17477751e+00
-5.95289290e-01 3.54876995e-01 -2.93457173e-02 -6.50069058e-01
3.13833326e-01 5.24385631e-01 5.97299993e-01 -3.21012914e-01
4.92671818e-01 3.44602078e-01 1.66275159e-01 -3.24938655e-01
3.22839916e-01 -8.73033553e-02 -3.29706609e-01 2.01527670e-01
2.15288773e-01 -3.42253268e-01 1.19053751e-01 3.29193801e-01
1.14225459e+00 -1.20003283e-01 3.93602252e-01 -2.92502016e-01
2.95513630e-01 -3.70678633e-01 5.96154571e-01 9.64601219e-01
-5.96521676e-01 1.25161171e+00 6.73489928e-01 -1.47104681e-01
-6.68796122e-01 -1.23678875e+00 -7.33340561e-01 4.07823503e-01
3.20159853e-01 -1.93408370e-01 -7.36182272e-01 -9.73920643e-01
-2.00750396e-01 8.12268019e-01 -9.55546081e-01 -1.14040136e-01
-1.73640847e-01 -1.02640951e+00 6.16359532e-01 5.82765400e-01
2.11370736e-01 -7.56733835e-01 -7.68316746e-01 1.96990252e-01
-5.06439149e-01 -1.46798801e+00 -7.88348198e-01 2.36383244e-01
-7.73319840e-01 -9.81102705e-01 -9.27970767e-01 -1.55818075e-01
6.70068085e-01 -2.82898098e-01 1.39546227e+00 -2.99436808e-01
-6.26007795e-01 6.64398372e-01 -4.19663519e-01 -5.24447560e-01
-6.54338837e-01 1.14853278e-01 -4.17343408e-01 -2.38567099e-01
5.52333631e-02 -1.40634879e-01 -8.47384810e-01 3.75473529e-01
-1.17534208e+00 1.08522914e-01 7.77610660e-01 7.31455028e-01
8.56875062e-01 -1.12476312e-01 3.73528957e-01 -1.12006128e+00
6.44345999e-01 -8.00438344e-01 -4.27349895e-01 4.09154117e-01
-4.83099461e-01 3.61663103e-01 5.35588115e-02 -2.31988668e-01
-1.17752540e+00 1.38155311e-01 -3.87046754e-01 -2.07279876e-01
-1.65042594e-01 5.17268956e-01 9.66359377e-02 1.72507316e-01
5.98603368e-01 -7.65047297e-02 -1.39468178e-01 -1.52304828e-01
4.20100838e-01 6.89037800e-01 5.06429017e-01 -7.72603631e-01
1.49220973e-01 4.64945555e-01 -3.29251945e-01 -2.32538640e-01
-9.61240888e-01 -3.20529968e-01 -3.53740871e-01 -3.41262549e-01
1.05899918e+00 -1.04232442e+00 -3.92518878e-01 3.56286556e-01
-1.28066635e+00 -1.96678683e-01 -3.04739982e-01 7.04898179e-01
-4.92001772e-01 4.58999693e-01 -7.63859272e-01 -5.45477927e-01
-4.70613003e-01 -1.97626042e+00 1.26545489e+00 2.93083340e-01
-3.54717880e-01 -9.93496716e-01 -8.83562490e-02 4.75818604e-01
4.04596299e-01 1.96224377e-01 6.11086071e-01 -7.06037760e-01
-4.77180421e-01 4.17523049e-02 -3.41372788e-01 4.05013531e-01
1.90338269e-01 -2.12049320e-01 -9.40465689e-01 -1.10279903e-01
7.19167888e-02 -3.20068181e-01 1.02479970e+00 9.73734379e-01
1.60659659e+00 2.37370223e-01 -3.40852261e-01 5.92268050e-01
1.37222064e+00 4.54512388e-02 6.41893029e-01 -9.77786258e-02
5.51157415e-01 2.59961516e-01 4.06447947e-01 5.26003599e-01
4.67536628e-01 6.64849818e-01 6.14379406e-01 -3.29429731e-02
-1.88713104e-01 1.88966379e-01 6.95105046e-02 7.51702249e-01
7.04691038e-02 -4.77381706e-01 -1.09105468e+00 5.68223059e-01
-1.68011856e+00 -3.83530110e-01 1.28900602e-01 2.04499054e+00
1.27280521e+00 1.12607546e-01 -3.12917054e-01 -3.59673411e-01
7.15875208e-01 1.22500882e-01 -7.25023210e-01 -1.56914860e-01
2.67422289e-01 5.11635914e-02 7.58695602e-01 7.32474566e-01
-1.26234770e+00 6.59907043e-01 6.26321459e+00 1.09088957e+00
-9.31961298e-01 2.28100553e-01 1.31742120e+00 -2.05878869e-01
-3.90502661e-01 -3.95756632e-01 -6.86800778e-01 6.64845765e-01
8.95953000e-01 1.17663659e-01 7.97791183e-02 6.34984851e-01
2.95973979e-02 -3.19148242e-01 -1.23286057e+00 9.86866832e-01
1.17233120e-01 -1.20863676e+00 -1.23432554e-01 -2.14659899e-01
7.00954556e-01 3.00319791e-01 2.80047446e-01 8.51584002e-02
3.83601010e-01 -1.23141205e+00 7.52643883e-01 5.68492472e-01
1.05816007e+00 -4.60135311e-01 9.82605159e-01 9.52004734e-03
-8.55305612e-01 2.84663141e-01 -6.11170754e-02 8.42275441e-01
4.83683795e-01 9.40419674e-01 -8.76093745e-01 7.23586679e-01
7.46374726e-01 4.51148748e-01 -3.81557912e-01 1.09119952e+00
-4.04555947e-01 5.47565043e-01 -3.24021816e-01 3.74267101e-01
1.35509908e-01 2.79815078e-01 5.74668705e-01 1.52010798e+00
2.97423095e-01 2.51067072e-01 -1.63153931e-01 1.15025330e+00
-1.33002222e-01 -2.25847542e-01 -9.43899993e-03 2.65137851e-01
3.10934722e-01 1.33400202e+00 -8.89851153e-01 -1.86841831e-01
-3.22820425e-01 9.64860499e-01 4.04586010e-02 1.51634201e-01
-1.30804336e+00 9.91035253e-02 2.18975410e-01 -2.59906411e-01
2.66797036e-01 1.03572764e-01 -4.03502524e-01 -1.06749308e+00
-4.05179150e-02 -9.86396432e-01 6.05489016e-01 -6.76571429e-01
-1.29093504e+00 9.09980357e-01 2.84057081e-01 -1.03255010e+00
-2.65517473e-01 -6.00926280e-01 -4.33752015e-02 8.94534469e-01
-1.42085528e+00 -9.26056385e-01 -1.90398782e-01 3.14605772e-01
8.47462714e-01 1.71324626e-01 7.43855417e-01 4.87891853e-01
-4.97718215e-01 7.63507664e-01 -1.07330076e-01 1.26923740e-01
7.98196912e-01 -1.36916780e+00 3.11784685e-01 8.42709780e-01
1.12478316e-01 3.18826765e-01 6.53526723e-01 -6.28673792e-01
-9.05593395e-01 -1.00312650e+00 1.99373096e-01 -7.96540320e-01
4.79574651e-01 -1.24509446e-01 -7.28555262e-01 5.86999297e-01
-4.84099053e-02 2.25398988e-01 7.45199084e-01 -1.50089458e-01
-1.49595097e-01 1.70165718e-01 -1.20656288e+00 4.71276373e-01
6.55479670e-01 -3.11352223e-01 -4.03171510e-01 8.45602751e-02
9.19495463e-01 -1.09689999e+00 -1.16895163e+00 6.98073924e-01
5.44075906e-01 -7.68628061e-01 8.30154121e-01 -2.02125520e-01
8.54438722e-01 -3.13264072e-01 -1.78580791e-01 -1.40089869e+00
-1.41348124e-01 -2.78457314e-01 -1.04939215e-01 9.18276191e-01
9.81757700e-01 -3.19315583e-01 5.20061910e-01 1.04652154e+00
-2.63514102e-01 -1.02941608e+00 -1.17250025e+00 -2.92985976e-01
1.65545329e-01 -1.05410516e+00 2.85998195e-01 6.22372150e-01
-1.93689302e-01 3.27368546e-03 -4.24774259e-01 3.93799901e-01
5.83256006e-01 -4.34865057e-01 -1.48067083e-02 -6.16410255e-01
-7.29377687e-01 -4.59883690e-01 -3.30646753e-01 -1.07232237e+00
-2.38934085e-01 -8.70221138e-01 3.17118376e-01 -1.63675547e+00
5.51478565e-01 -4.05383080e-01 -5.84849834e-01 2.88308829e-01
-3.86244833e-01 1.65708646e-01 1.33461967e-01 4.67431769e-02
-7.17362821e-01 3.17069292e-01 1.42765141e+00 -2.60851562e-01
1.36047721e-01 2.14879349e-01 -9.16312873e-01 6.70020103e-01
6.77768648e-01 -7.79761314e-01 -4.41044867e-01 -8.06548774e-01
1.93963379e-01 2.00415865e-01 2.59687364e-01 -8.94881427e-01
1.88722908e-01 2.09229231e-01 2.29960576e-01 -4.58727717e-01
1.16903931e-01 -1.01558900e+00 1.16029173e-01 4.23685700e-01
-5.48597574e-01 -2.33442813e-01 5.68120539e-01 7.25463986e-01
-1.54400364e-01 -5.36608160e-01 9.44190264e-01 -1.52970389e-01
-5.59624195e-01 4.92112249e-01 -2.50154495e-01 4.95850861e-01
9.93785739e-01 1.81144819e-01 -9.06084925e-02 -4.10170645e-01
-9.98716295e-01 4.93913174e-01 -2.26583872e-02 4.34280455e-01
5.73758304e-01 -1.04102898e+00 -1.02791679e+00 -2.55677342e-01
2.31162190e-01 3.83821845e-01 6.80759847e-01 1.10617006e+00
-5.31200588e-01 3.53942484e-01 1.79609686e-01 -9.79748368e-01
-8.68863165e-01 2.42458373e-01 7.03113675e-01 -6.68068826e-01
-4.39078093e-01 1.17047405e+00 4.76715505e-01 -1.74189642e-01
2.19998121e-01 -7.14776874e-01 -6.75363541e-02 -1.57041505e-01
5.55027723e-01 1.63252223e-02 5.93169779e-02 -4.65803027e-01
-5.03970861e-01 4.65728581e-01 -4.12427008e-01 -3.56642157e-01
9.54022348e-01 -2.28757203e-01 2.70483881e-01 4.27359611e-01
9.95229363e-01 -1.62131265e-01 -1.48376501e+00 -3.40711892e-01
-2.74605900e-01 -2.32956693e-01 3.00965518e-01 -1.32019234e+00
-1.38324952e+00 7.90676057e-01 7.77265072e-01 -2.39179716e-01
1.05936241e+00 2.66288370e-01 5.17104983e-01 -2.71963656e-01
2.45988488e-01 -1.00950396e+00 -1.89308509e-01 2.30620191e-01
1.01414478e+00 -1.71718001e+00 1.91633112e-03 -4.91472542e-01
-1.16168594e+00 9.41102147e-01 2.96373993e-01 3.25366288e-01
8.64832044e-01 6.16042435e-01 1.82783678e-01 -1.88213766e-01
-5.19698381e-01 5.22224121e-02 7.03133225e-01 3.52043211e-01
7.21374810e-01 2.47616142e-01 -2.44033471e-01 9.99695837e-01
4.71604057e-02 8.90956223e-02 5.14466703e-01 5.85629642e-01
4.95769195e-02 -9.14226711e-01 -1.79157719e-01 4.81982738e-01
-1.08977771e+00 -3.24057609e-01 2.09673673e-01 4.89129901e-01
3.50372493e-02 8.54983151e-01 -9.02865231e-02 -1.73723638e-01
1.43145740e-01 -3.67897063e-01 5.07292628e-01 -7.18191326e-01
-3.93847078e-01 6.07218668e-02 1.02955788e-01 -6.38679266e-01
-2.03253418e-01 -5.34127414e-01 -1.37684655e+00 3.17600787e-01
-2.59575725e-01 -1.31516866e-02 7.47841179e-01 8.93438458e-01
2.30717748e-01 1.21495903e+00 1.95497841e-01 -4.62607473e-01
-6.38852656e-01 -8.28608215e-01 -4.30779159e-01 2.87345886e-01
2.47390717e-01 -5.17768681e-01 -2.95276374e-01 5.77913485e-02]
|
[14.539958953857422, -2.0592527389526367]
|
951b2af8-1a00-46c9-a992-e9beb0f97d99
|
scalable-knowledge-base-completion-with
|
2110.12341
| null |
https://arxiv.org/abs/2110.12341v1
|
https://arxiv.org/pdf/2110.12341v1.pdf
|
Scalable knowledge base completion with superposition memories
|
We present Harmonic Memory Networks (HMem), a neural architecture for knowledge base completion that models entities as weighted sums of pairwise bindings between an entity's neighbors and corresponding relations. Since entities are modeled as aggregated neighborhoods, representations of unseen entities can be generated on the fly. We demonstrate this with two new datasets: WNGen and FBGen. Experiments show that the model is SOTA on benchmarks, and flexible enough to evolve without retraining as the knowledge graph grows.
|
['Paul Smolensky', 'Eric Rosen', 'Matthias Lalisse']
|
2021-10-24
| null | null | null | null |
['knowledge-base-completion', 'knowledge-base-completion']
|
['graphs', 'knowledge-base']
|
[-2.84207642e-01 1.06220055e+00 -3.37226391e-01 -2.68841565e-01
-3.69287491e-01 -5.03705502e-01 3.39359671e-01 2.98837364e-01
-3.37286919e-01 1.15065157e+00 3.86413276e-01 1.02725439e-01
-4.06901300e-01 -1.47110653e+00 -1.04298592e+00 -2.41949007e-01
-7.05765843e-01 1.19185376e+00 2.78402805e-01 -4.00895268e-01
-4.28557247e-01 2.51627088e-01 -1.57107401e+00 2.32996181e-01
4.82181549e-01 4.66566354e-01 -2.51061141e-01 4.90290344e-01
-9.35218483e-02 1.01271820e+00 -6.61712170e-01 -8.43893647e-01
2.70596266e-01 7.80653581e-02 -1.27677715e+00 -5.09984136e-01
2.47274011e-01 -1.77869603e-01 -9.63631690e-01 8.13693285e-01
4.40339416e-01 8.82824957e-01 6.41341150e-01 -1.16378903e+00
-1.50797915e+00 1.31213534e+00 5.77041954e-02 1.60934210e-01
3.10829818e-01 -2.38328114e-01 1.38829792e+00 -1.04378045e+00
1.25451243e+00 1.25003326e+00 8.70676458e-01 6.77811086e-01
-1.37592554e+00 -4.42185462e-01 1.54772416e-01 4.37763870e-01
-1.79679608e+00 -4.43586379e-01 1.44924983e-01 -1.29429951e-01
1.69448352e+00 1.47238091e-01 6.19738936e-01 8.95252764e-01
-4.03582424e-01 5.66615462e-01 -7.73531944e-02 -3.59125227e-01
2.14468271e-01 6.70760497e-02 5.10963678e-01 7.97185779e-01
6.92640424e-01 -1.36043588e-02 -5.05160213e-01 -3.64569575e-01
5.64640343e-01 -2.19773114e-01 -2.40673646e-01 -4.35335129e-01
-9.08913851e-01 8.25122833e-01 8.95702779e-01 1.31779775e-01
-3.96062046e-01 2.51439363e-01 -5.64144291e-02 2.17940450e-01
1.90907493e-01 1.00378275e+00 -3.85362238e-01 2.96850622e-01
-7.22765386e-01 4.01503652e-01 1.16160750e+00 1.30905712e+00
9.42073226e-01 -8.33744705e-02 -2.98682302e-01 6.78318501e-01
-1.24762498e-01 3.50155115e-01 4.42576587e-01 -8.53583992e-01
1.89411014e-01 7.74903119e-01 1.98916107e-01 -1.12067759e+00
-5.31274855e-01 -3.68683070e-01 -7.86043227e-01 -4.26969141e-01
-1.62071481e-01 -4.20772314e-01 -1.18277586e+00 1.88518763e+00
1.87607557e-01 5.91208994e-01 4.70190018e-01 4.81509298e-01
1.49693084e+00 7.56923139e-01 4.16345090e-01 1.17012925e-01
8.58476758e-01 -8.97912681e-01 -6.68367505e-01 -3.15612495e-01
4.31901574e-01 1.39603198e-01 3.75420332e-01 -4.70692962e-02
-1.34606647e+00 -3.56827796e-01 -1.11964202e+00 -3.21192890e-01
-1.23072410e+00 -4.84088659e-01 8.85780931e-01 1.66688606e-01
-1.31396532e+00 8.19600463e-01 -8.65051091e-01 -3.95833433e-01
3.23303521e-01 7.40113080e-01 -4.94768441e-01 1.17369182e-01
-1.95571589e+00 1.19552219e+00 1.32661247e+00 2.69261688e-01
-8.68774593e-01 -6.45190895e-01 -1.21275723e+00 4.72706616e-01
2.29019314e-01 -1.13440192e+00 8.97890270e-01 -3.93261492e-01
-6.63050771e-01 6.63138092e-01 -2.59918839e-01 -8.05276394e-01
-2.99111307e-01 1.03784874e-01 -9.39773560e-01 -2.31353432e-01
-2.28361562e-01 9.70813692e-01 3.69192928e-01 -1.39647222e+00
-6.11313820e-01 -1.57242432e-01 3.91701132e-01 1.32383540e-01
-5.30603230e-01 -5.46255291e-01 -4.96935606e-01 -3.65618795e-01
-2.97841839e-02 -7.64554560e-01 -2.26486847e-01 -7.73769319e-01
-8.52244735e-01 -4.19325680e-01 4.93916094e-01 -6.74066186e-01
1.51608741e+00 -1.68198133e+00 5.55058956e-01 7.13406324e-01
4.59897995e-01 1.98847592e-01 -4.45445627e-01 4.27641064e-01
-2.38358498e-01 2.21506640e-01 -3.46108973e-02 -2.43613735e-01
3.32642466e-01 6.20604873e-01 -3.69457453e-01 -1.83605239e-01
2.12844744e-01 1.49759567e+00 -9.56135035e-01 -1.80784807e-01
-4.11559194e-01 5.89311540e-01 -5.97863793e-01 5.77362590e-02
-3.88198227e-01 -5.12113094e-01 -9.10421386e-02 5.10009527e-01
4.01239604e-01 -6.10063612e-01 5.21648526e-01 -3.08174103e-01
6.88295782e-01 2.39018574e-01 -1.26753676e+00 1.45846152e+00
-9.83375981e-02 3.96476746e-01 -3.12970519e-01 -6.71221733e-01
7.43414760e-01 3.81616950e-01 -3.20972712e-03 -3.36403131e-01
-3.25771809e-01 -5.98681942e-02 -2.30878085e-01 -3.37787420e-01
1.02549183e+00 1.67139798e-01 -8.29801559e-02 3.57648760e-01
6.96596861e-01 4.33560580e-01 6.10161483e-01 5.09472072e-01
1.21726179e+00 -2.86847651e-01 3.01171005e-01 2.08225884e-02
-1.21645536e-02 1.05659455e-01 6.09759510e-01 9.16356444e-01
3.82178098e-01 2.40886629e-01 2.42339179e-01 -9.25326526e-01
-8.84665370e-01 -1.43048155e+00 1.40017588e-02 1.25001812e+00
2.12253317e-01 -7.45968401e-01 -5.73280334e-01 -6.83908105e-01
4.42115873e-01 8.89608085e-01 -1.08056331e+00 -4.85088825e-01
-4.99100983e-01 -7.47322083e-01 6.21267021e-01 8.39519322e-01
7.88271949e-02 -1.57315111e+00 -9.52697843e-02 3.02304238e-01
-1.87845170e-01 -7.42626905e-01 -6.99812081e-04 2.65445471e-01
-4.89930749e-01 -1.17142844e+00 -3.13124686e-01 -1.07139754e+00
8.01703811e-01 -2.74720609e-01 1.86227989e+00 3.01930070e-01
-3.84524584e-01 1.52405784e-01 -1.22649157e-02 -2.13850424e-01
-2.07749516e-01 5.79719722e-01 2.15200499e-01 -5.96221626e-01
7.08111346e-01 -8.91222417e-01 -3.08636099e-01 -3.02287508e-02
-9.06397462e-01 -3.18723202e-01 3.10851812e-01 9.80923712e-01
7.84951687e-01 4.26913053e-01 8.00275862e-01 -1.49763203e+00
6.49210930e-01 -9.55374837e-01 -4.58762586e-01 7.80964494e-01
-5.01330256e-01 1.72387898e-01 3.84296924e-01 -2.00039223e-01
-1.01225221e+00 -6.68144822e-02 1.90720186e-01 -4.02852416e-01
-3.69617343e-02 9.80087340e-01 -8.42852071e-02 1.27967462e-01
1.04255521e+00 -2.24178713e-02 -6.86327875e-01 -2.39336371e-01
9.53278065e-01 1.51214138e-01 9.60676491e-01 -6.05257690e-01
8.72204244e-01 2.73453176e-01 -2.43442073e-01 -4.66215938e-01
-9.87224162e-01 -8.18285868e-02 -5.56100547e-01 4.29405987e-01
6.68146431e-01 -1.05580461e+00 -8.35350990e-01 -2.83899307e-01
-1.19553173e+00 -2.41885439e-01 -6.82098866e-01 1.33873791e-01
-7.65860453e-02 -2.31794879e-01 -8.76711309e-01 -4.68283713e-01
-5.51403522e-01 -3.99277031e-01 5.35014689e-01 5.04455984e-01
-5.11262834e-01 -1.31162810e+00 3.46427679e-01 6.29458874e-02
4.71354902e-01 3.42722684e-01 1.28440332e+00 -1.02731359e+00
-7.84000933e-01 -2.23631799e-01 -4.19706106e-03 -1.00289926e-01
-2.38403261e-01 -4.94921468e-02 -8.01878631e-01 -3.69179398e-01
-9.47350144e-01 -5.03973961e-01 1.08689570e+00 8.99579898e-02
8.52992475e-01 -7.33401477e-01 -8.26406658e-01 7.18019307e-01
1.26088190e+00 -3.17168906e-02 7.08488464e-01 1.80169612e-01
7.20158160e-01 2.03920051e-01 -5.09742722e-02 2.06026316e-01
7.90528357e-01 2.33931541e-01 3.08704376e-01 -9.97181330e-03
-6.27626851e-02 -6.20440543e-01 -1.38590127e-01 6.32998705e-01
-3.43524188e-01 -5.15960097e-01 -1.09029949e+00 9.57308292e-01
-2.02872562e+00 -1.22941780e+00 2.37955213e-01 1.85534012e+00
8.40774477e-01 4.71262597e-02 -1.01447605e-01 -5.83973944e-01
7.28836656e-01 -1.17060639e-01 -8.19053590e-01 -1.16054572e-01
-6.05027199e-01 6.56647563e-01 4.22427356e-01 6.57050014e-01
-1.17614555e+00 1.22670436e+00 7.90946198e+00 5.20186901e-01
-2.66432732e-01 -1.36259124e-01 4.68033552e-01 -3.47418875e-01
-6.75546944e-01 -1.51558220e-01 -8.68453681e-01 -5.46144545e-02
9.44836199e-01 -6.70955241e-01 7.25350976e-01 8.52311134e-01
-8.00531328e-01 3.58361006e-01 -1.34864557e+00 6.09940171e-01
-1.30299171e-02 -1.63387871e+00 4.33014512e-01 -1.36662543e-01
1.10133839e+00 2.37964690e-01 -7.92942941e-02 9.40689862e-01
1.17149007e+00 -1.48081410e+00 1.04951702e-01 8.33940327e-01
5.28761983e-01 -1.09631085e+00 8.12193871e-01 9.50260386e-02
-1.06543851e+00 -1.25841007e-01 -8.00241053e-01 2.06441835e-01
2.34913513e-01 2.79114544e-01 -1.09808505e+00 7.14314818e-01
7.68931150e-01 6.21507823e-01 -7.53280342e-01 7.61660337e-01
-4.89732713e-01 3.25672209e-01 -3.11502427e-01 2.22858697e-01
-1.05014868e-01 2.27743730e-01 3.82264674e-01 1.20466650e+00
2.26050615e-01 4.01720554e-01 2.10573282e-02 1.23000956e+00
-6.59211934e-01 -2.26213410e-01 -7.39116430e-01 -1.30447984e-01
1.02243817e+00 1.35143077e+00 -2.64041990e-01 -5.46684265e-01
-1.70749098e-01 6.55314147e-01 1.15056312e+00 8.44977140e-01
-6.28446519e-01 -6.14267290e-01 6.49510324e-01 -1.41388983e-01
5.32579005e-01 2.28056848e-01 1.78971916e-01 -1.13222492e+00
-2.25666866e-01 -5.79071522e-01 9.77593243e-01 -9.16668653e-01
-1.54668188e+00 9.24525797e-01 -8.05581957e-02 -3.37673903e-01
-5.28633952e-01 -3.27271283e-01 -3.97873551e-01 8.09679627e-01
-1.21206701e+00 -1.11792445e+00 -1.77308381e-01 6.14910901e-01
-2.06889421e-01 -1.76657081e-01 1.57134748e+00 2.44758144e-01
-6.76064312e-01 7.16368794e-01 -5.64350970e-02 5.62789738e-01
3.09252113e-01 -1.42428184e+00 1.05043507e+00 5.91012061e-01
5.10751307e-01 1.20898306e+00 5.15010238e-01 -7.44859338e-01
-1.12623513e+00 -1.22721267e+00 1.03890038e+00 -7.65336335e-01
5.99494934e-01 -3.11610252e-01 -1.20543468e+00 1.38912463e+00
4.25238311e-01 1.44463375e-01 9.25057709e-01 8.58872890e-01
-5.91164052e-01 1.40590861e-01 -9.55553651e-01 5.12217164e-01
1.50989544e+00 -5.23131192e-01 -9.77914333e-01 3.40300977e-01
9.03037310e-01 -5.12397349e-01 -1.39723468e+00 6.56732738e-01
3.11623812e-01 -5.40474474e-01 1.25608385e+00 -1.36523283e+00
1.46627486e-01 -2.86552101e-01 -5.31835556e-02 -1.61740458e+00
-6.97026134e-01 -4.91872758e-01 -1.12521064e+00 1.07398510e+00
1.25793219e+00 -2.50927508e-01 1.08858144e+00 1.14564967e+00
1.61695823e-01 -6.03547454e-01 -6.10087395e-01 -6.94123864e-01
-1.05353490e-01 1.59759745e-01 1.15624380e+00 1.39732468e+00
4.81779248e-01 7.02400982e-01 -2.46874124e-01 3.48155826e-01
3.99051905e-01 2.60980129e-01 4.06603605e-01 -1.42664731e+00
-4.39241379e-01 -1.62159726e-01 -4.87625867e-01 -5.46471059e-01
5.33954203e-01 -1.13953161e+00 -1.66416064e-01 -1.79212654e+00
4.04359102e-01 -2.81694651e-01 -6.82332277e-01 9.90959466e-01
-3.56329948e-01 6.95580915e-02 -1.62370369e-01 -1.39747471e-01
-1.10606289e+00 4.73439306e-01 6.91853821e-01 -4.30392116e-01
-2.32724056e-01 -4.36999083e-01 -7.55657256e-01 5.41514039e-01
5.99073470e-01 -3.30950588e-01 -6.72091424e-01 -6.94531739e-01
7.64557719e-01 -1.40391082e-01 1.39517322e-01 -7.86923647e-01
7.53591716e-01 2.57420063e-01 6.98753953e-01 -4.84157324e-01
4.56102401e-01 -4.50109273e-01 7.24686205e-01 -1.14895977e-01
-5.30398190e-01 2.42225453e-01 3.48226190e-01 6.82457745e-01
-3.74473155e-01 -3.44957560e-02 2.18375772e-01 -2.65627056e-01
-1.04463482e+00 4.47272152e-01 1.55793518e-01 4.09568399e-01
7.30240881e-01 1.13063026e-02 -7.97767937e-01 -4.97256249e-01
-1.49901998e+00 6.18027568e-01 2.11703196e-01 4.50786978e-01
7.04129577e-01 -1.54587042e+00 -6.19628549e-01 -3.61039750e-02
2.58877397e-01 3.70254219e-01 3.01863015e-01 5.16920052e-02
-2.10340559e-01 4.39305514e-01 -8.55914690e-03 1.91000059e-01
-8.48224699e-01 1.01712656e+00 5.80365419e-01 -4.33143079e-01
-3.39303732e-01 1.19910228e+00 -5.73421977e-02 -6.64139748e-01
5.77301443e-01 -4.32559438e-02 -4.19524670e-01 3.62413116e-02
5.58401287e-01 4.01586324e-01 1.08801119e-01 -5.65066457e-01
-2.79047877e-01 -9.97698456e-02 -3.62738013e-01 1.81100100e-01
1.64175785e+00 2.46691495e-01 -4.23979312e-01 1.11179277e-01
9.01320338e-01 -2.84228921e-01 -7.02638209e-01 -4.43319857e-01
4.08827573e-01 5.25133610e-02 -2.80628175e-01 -8.97101104e-01
-1.09958708e+00 5.07310852e-02 1.04800925e-01 3.97324234e-01
7.63700187e-01 1.37004986e-01 8.23349416e-01 1.31488943e+00
3.41735035e-01 -1.07875991e+00 -4.73479360e-01 7.11108387e-01
6.76999986e-01 -8.42898607e-01 -8.28870684e-02 -3.22543859e-01
-3.42400610e-01 6.51381612e-01 9.33218598e-01 -5.60125634e-02
7.25576103e-01 2.10453838e-01 -2.61616915e-01 -5.25637448e-01
-1.19901872e+00 -2.76187360e-01 6.00230873e-01 8.29955578e-01
2.46053576e-01 2.15450779e-01 3.47597480e-01 1.02400053e+00
-3.86956185e-01 -9.72619429e-02 2.90399551e-01 7.06200242e-01
-3.80520999e-01 -9.38683271e-01 1.11762926e-01 5.33353984e-01
-4.85625528e-02 -5.50031066e-01 -6.34935796e-01 9.61854100e-01
3.05771738e-01 5.86350679e-01 1.60175219e-01 -5.13908327e-01
6.80435956e-01 4.00430202e-01 3.42209399e-01 -8.24891567e-01
-4.63964760e-01 -7.81586468e-01 4.82206613e-01 -5.16844690e-01
-1.24037176e-01 -6.98663965e-02 -1.51947725e+00 -4.45051730e-01
-3.69800478e-01 5.35207093e-01 -5.86353019e-02 5.18672645e-01
8.90741646e-01 6.32825971e-01 -1.08679481e-01 -4.40931827e-01
-1.21883467e-01 -8.03930879e-01 -7.55425453e-01 7.74643958e-01
9.60066170e-02 -6.26385212e-01 -6.51538298e-02 5.76398969e-02]
|
[8.90749454498291, 8.054883003234863]
|
9aa9f7db-c006-42c9-b413-b07712a17b3a
|
colar-effective-and-efficient-online-action
|
2203.01057
| null |
https://arxiv.org/abs/2203.01057v2
|
https://arxiv.org/pdf/2203.01057v2.pdf
|
Colar: Effective and Efficient Online Action Detection by Consulting Exemplars
|
Online action detection has attracted increasing research interests in recent years. Current works model historical dependencies and anticipate the future to perceive the action evolution within a video segment and improve the detection accuracy. However, the existing paradigm ignores category-level modeling and does not pay sufficient attention to efficiency. Considering a category, its representative frames exhibit various characteristics. Thus, the category-level modeling can provide complimentary guidance to the temporal dependencies modeling. This paper develops an effective exemplar-consultation mechanism that first measures the similarity between a frame and exemplary frames, and then aggregates exemplary features based on the similarity weights. This is also an efficient mechanism, as both similarity measurement and feature aggregation require limited computations. Based on the exemplar-consultation mechanism, the long-term dependencies can be captured by regarding historical frames as exemplars, while the category-level modeling can be achieved by regarding representative frames from a category as exemplars. Due to the complementarity from the category-level modeling, our method employs a lightweight architecture but achieves new high performance on three benchmarks. In addition, using a spatio-temporal network to tackle video frames, our method makes a good trade-off between effectiveness and efficiency. Code is available at https://github.com/VividLe/Online-Action-Detection.
|
['Dingwen Zhang', 'Junwei Han', 'Le Yang']
|
2022-03-02
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Yang_Colar_Effective_and_Efficient_Online_Action_Detection_by_Consulting_Exemplars_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Yang_Colar_Effective_and_Efficient_Online_Action_Detection_by_Consulting_Exemplars_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['online-action-detection']
|
['computer-vision']
|
[-8.53018314e-02 -4.06088322e-01 -4.74959850e-01 -4.11236465e-01
-4.01705384e-01 -1.91298574e-01 6.64163113e-01 2.48081610e-01
-3.34002793e-01 2.77299464e-01 3.52870852e-01 1.38805479e-01
-1.18822731e-01 -7.75828958e-01 -4.11528915e-01 -6.31222427e-01
-3.31382930e-01 -1.83146149e-01 7.56407857e-01 -8.03974867e-02
2.23401070e-01 4.14955080e-01 -1.77193952e+00 5.05191982e-01
7.34582722e-01 1.13454711e+00 3.73948485e-01 5.35876751e-01
-1.31686196e-01 9.66396511e-01 -5.12365460e-01 -2.56941944e-01
2.20171928e-01 -6.73503160e-01 -5.31783879e-01 2.99144089e-01
2.78685004e-01 -4.73596573e-01 -4.69706863e-01 1.03380573e+00
3.56597394e-01 2.82822907e-01 2.82344609e-01 -1.39741457e+00
-4.29175973e-01 5.53872585e-01 -5.44616699e-01 5.33494890e-01
5.73926449e-01 3.66065621e-01 1.07689893e+00 -7.70104110e-01
5.52491367e-01 1.31473947e+00 3.50089133e-01 4.60057974e-01
-7.18498290e-01 -4.80305284e-01 8.21963847e-01 8.22836101e-01
-1.30288553e+00 -3.64622802e-01 7.93624043e-01 -4.54134107e-01
6.39507949e-01 2.28681058e-01 1.07192504e+00 8.52713227e-01
5.68456156e-03 1.18636668e+00 6.40150011e-01 -8.63321722e-02
2.24980921e-01 -2.07230434e-01 1.74516395e-01 3.82838517e-01
-1.51307613e-01 4.78615500e-02 -4.23500538e-01 7.35806450e-02
7.32985735e-01 4.70379025e-01 -3.38161826e-01 -1.57295913e-01
-1.28655064e+00 5.09879649e-01 5.08939505e-01 5.83145142e-01
-5.64364433e-01 2.76637942e-01 6.81838393e-01 2.06564441e-02
3.19575429e-01 3.35959308e-02 -1.49415463e-01 -4.97445434e-01
-9.90901589e-01 1.98984686e-02 4.10625488e-01 8.93626630e-01
7.44913459e-01 -7.24933436e-03 -5.08792818e-01 7.09921658e-01
1.49244472e-01 1.78527519e-01 4.66813594e-01 -1.14049077e+00
2.48271152e-01 8.28838170e-01 1.19245045e-01 -1.30147362e+00
-3.61791313e-01 -4.93530571e-01 -7.33628035e-01 1.71699673e-01
3.72613162e-01 1.41450986e-01 -5.55632055e-01 1.77657545e+00
5.16015232e-01 6.26311898e-01 -1.91533655e-01 1.07776344e+00
6.64093792e-01 6.61793470e-01 1.90352932e-01 -5.87499857e-01
1.31921935e+00 -1.07916331e+00 -7.64764667e-01 5.85062578e-02
5.81822634e-01 -6.16708100e-01 9.82307017e-01 2.52457023e-01
-1.04681790e+00 -8.31097901e-01 -9.91653919e-01 2.48985022e-01
-1.00960612e-01 2.28643566e-01 5.51222146e-01 2.47319862e-01
-9.00770128e-01 5.90816736e-01 -9.57005739e-01 -5.74100494e-01
2.90899575e-01 1.39267921e-01 -1.37739077e-01 7.92992786e-02
-1.13498306e+00 5.98221779e-01 5.48700571e-01 1.18664064e-01
-7.44272351e-01 -4.49570477e-01 -5.01405358e-01 2.80576311e-02
6.74276590e-01 -4.89156663e-01 1.14686430e+00 -1.16236711e+00
-1.28042734e+00 4.29468870e-01 -2.52681017e-01 -6.16227746e-01
6.17921889e-01 -2.69204587e-01 -6.01123273e-01 4.64555383e-01
-6.00956054e-03 4.53377932e-01 7.58049965e-01 -9.18842196e-01
-1.23389184e+00 -5.10050505e-02 6.87521696e-01 2.88911045e-01
-6.26165032e-01 1.48451015e-01 -7.77647853e-01 -7.81634390e-01
1.62753686e-01 -6.87802911e-01 -1.72415629e-01 1.62298396e-01
1.62223857e-02 -4.40925598e-01 8.59321356e-01 -2.95449704e-01
1.80451655e+00 -2.34222746e+00 1.51244327e-02 -1.15438424e-01
2.76510417e-01 3.49842906e-01 -7.41798505e-02 5.34609318e-01
-5.03637232e-02 -3.93151790e-02 3.69377248e-02 -8.40853825e-02
-5.03174141e-02 1.03977300e-01 -9.68135968e-02 4.05100465e-01
6.49311319e-02 6.59621477e-01 -1.13622212e+00 -6.91624165e-01
5.00659943e-01 3.14945847e-01 -5.74876785e-01 1.51213095e-01
-7.92436749e-02 3.33538264e-01 -6.14284813e-01 5.83377004e-01
4.48359072e-01 -2.34716952e-01 8.27293843e-02 -5.77686310e-01
-3.32387179e-01 -3.37832123e-02 -1.35847759e+00 1.55052948e+00
-3.50497723e-01 4.68976021e-01 -1.92618519e-01 -1.19849396e+00
8.05022717e-01 2.93635547e-01 9.29128170e-01 -6.40408039e-01
4.80619036e-02 1.36486977e-01 1.22798108e-01 -6.84581161e-01
3.59156191e-01 2.71330029e-01 1.34441584e-01 3.51971000e-01
-1.87751979e-01 5.43102562e-01 7.41548479e-01 2.94804454e-01
1.12192500e+00 4.87430215e-01 5.45557439e-01 1.18485335e-02
8.69131923e-01 -2.24611368e-02 1.01183403e+00 6.48054540e-01
-4.90263969e-01 4.34080988e-01 3.37032646e-01 -7.28834212e-01
-6.11708343e-01 -8.35018218e-01 3.91204283e-02 1.22306228e+00
6.24200702e-01 -7.52317250e-01 -6.36263847e-01 -6.71821833e-01
-1.90441161e-01 4.73749876e-01 -6.42764866e-01 -3.01312566e-01
-7.71314859e-01 -5.42354226e-01 1.43437013e-01 5.72015822e-01
7.66988039e-01 -1.07095325e+00 -9.01867330e-01 4.22786802e-01
-3.55498224e-01 -9.71314669e-01 -5.77034235e-01 -4.62240726e-01
-7.72957265e-01 -1.13136089e+00 -5.31585634e-01 -4.83387619e-01
4.10779089e-01 5.91197729e-01 1.05030417e+00 3.87950063e-01
-1.62479222e-01 5.35075903e-01 -8.34150910e-01 -8.16528946e-02
-2.42510468e-01 -2.67925262e-01 1.42993331e-01 4.38481033e-01
4.95076597e-01 -6.42417014e-01 -1.00617290e+00 5.51686525e-01
-7.71626055e-01 2.64371693e-01 4.84749824e-01 5.75034261e-01
6.84877872e-01 8.10709037e-03 5.16276360e-01 -2.89762497e-01
2.30514869e-01 -5.66343188e-01 -3.05861562e-01 4.07608658e-01
-2.44715363e-01 -3.29198450e-01 7.98760474e-01 -6.38471246e-01
-9.82538700e-01 5.92809804e-02 1.56462553e-03 -6.39336526e-01
-2.07159325e-01 4.72772598e-01 -1.44605473e-01 2.58661658e-01
4.48401898e-01 3.35038602e-01 -1.22966781e-01 -3.76222879e-01
3.45856786e-01 4.55071896e-01 4.90817010e-01 -5.41156411e-01
4.81992126e-01 6.82108402e-01 -1.14382610e-01 -5.99481940e-01
-7.94457018e-01 -6.98134303e-01 -7.06986725e-01 -8.17404509e-01
8.26505721e-01 -8.57690156e-01 -8.28374207e-01 4.16581482e-01
-1.10029614e+00 -7.19239935e-02 -3.66622508e-01 7.61019528e-01
-5.75411558e-01 5.39871335e-01 -5.49415767e-01 -8.77875149e-01
-3.08975782e-02 -1.11022043e+00 7.95368850e-01 2.72859395e-01
-1.30418584e-01 -7.95762181e-01 -2.72625893e-01 -6.14482500e-02
2.77089149e-01 3.48382026e-01 3.29504192e-01 -4.64940995e-01
-7.83935189e-01 -2.90311754e-01 -1.34341329e-01 2.33520448e-01
2.62826145e-01 3.76891494e-01 -5.41901171e-01 -1.93353638e-01
1.00379437e-01 1.61250934e-01 8.18024933e-01 5.42637348e-01
1.17635107e+00 -2.94387579e-01 -2.23096102e-01 4.06557441e-01
1.16287100e+00 5.34094214e-01 5.71361423e-01 4.29411381e-01
6.25274897e-01 5.20233035e-01 1.20604384e+00 8.21530700e-01
4.46175784e-01 9.81046259e-01 6.03690922e-01 6.26314618e-03
-1.99404642e-01 3.66467535e-02 4.40814137e-01 8.48620772e-01
-4.45256442e-01 -1.87193975e-01 -7.07120001e-01 4.36887681e-01
-2.35251641e+00 -1.61154723e+00 -1.98140278e-01 2.11964464e+00
4.89939570e-01 1.01652771e-01 6.12432182e-01 1.40499532e-01
1.09618187e+00 4.13467944e-01 -6.58681750e-01 5.57796620e-02
8.67589377e-03 -5.13650358e-01 -6.15406409e-03 1.16758123e-01
-1.24538577e+00 6.97211206e-01 5.51110840e+00 1.12215674e+00
-9.87043560e-01 1.78134695e-01 5.42788327e-01 -3.08201283e-01
1.46142766e-01 4.73823771e-02 -7.05945313e-01 8.39139163e-01
5.39847493e-01 -5.48670650e-01 1.65055856e-01 8.74623120e-01
6.22704327e-01 -1.31291613e-01 -1.18767047e+00 1.06195796e+00
-3.23503874e-02 -1.37615526e+00 2.13343292e-01 -1.00199863e-01
4.41320568e-01 -3.01162243e-01 -1.39928862e-01 2.22208530e-01
-7.39766657e-02 -4.14335728e-01 9.57683563e-01 8.29688966e-01
3.50252539e-01 -7.03100264e-01 5.47192037e-01 3.73266488e-01
-1.91390741e+00 -4.03618425e-01 -3.54211360e-01 -2.85953671e-01
3.81121188e-01 4.77543861e-01 -6.48023263e-02 7.56281197e-01
8.00827861e-01 1.28478253e+00 -5.11263311e-01 1.22023261e+00
-2.92455480e-02 4.49441224e-01 -1.96407795e-01 -6.24171384e-02
3.38645756e-01 -3.30112487e-01 5.62902987e-01 1.28836811e+00
4.26139772e-01 2.83992440e-01 6.09798074e-01 4.19694632e-01
3.39408189e-01 2.17776552e-01 -2.28411332e-01 1.62809998e-01
5.79438090e-01 1.22906148e+00 -8.25421929e-01 -7.27845848e-01
-7.62922108e-01 7.77170539e-01 1.44653752e-01 2.63778508e-01
-1.18361664e+00 -9.77806076e-02 7.32852042e-01 1.03951693e-01
2.54796684e-01 -1.53221220e-01 9.69936699e-02 -1.23551369e+00
1.84567079e-01 -7.75501192e-01 6.06029093e-01 -5.32309175e-01
-1.06514990e+00 5.58757722e-01 1.77411318e-01 -1.99586082e+00
-1.84660286e-01 -2.16594577e-01 -7.49270737e-01 3.19533795e-01
-1.11367965e+00 -9.03513372e-01 -6.11726642e-01 6.99273825e-01
8.39516997e-01 2.44908165e-02 3.96628976e-01 4.68759298e-01
-8.30792606e-01 4.24020141e-01 -1.81051537e-01 8.72660950e-02
4.71802145e-01 -1.00838017e+00 1.00020818e-01 1.10927069e+00
2.17885286e-01 5.64598739e-01 6.31312191e-01 -3.90302658e-01
-1.11853230e+00 -1.10205257e+00 5.38300872e-01 -1.64768115e-01
7.41904438e-01 1.15031399e-01 -1.01869178e+00 2.60862947e-01
7.65498355e-02 1.76434875e-01 5.58832407e-01 -9.10627022e-02
-2.18513831e-01 -3.95007104e-01 -7.60761023e-01 7.68137157e-01
1.53915071e+00 -3.78825724e-01 -4.75818008e-01 3.77908617e-01
5.42130888e-01 -1.57736510e-01 -9.95647609e-01 5.10115266e-01
6.63029075e-01 -1.42514133e+00 9.47021842e-01 -3.60584348e-01
4.13153082e-01 -6.61532342e-01 -7.03478754e-02 -8.84993315e-01
-5.92213333e-01 -5.73498905e-01 -5.14558554e-01 1.31773114e+00
-1.04951419e-01 -3.93660575e-01 6.15014613e-01 4.92212892e-01
-2.48888969e-01 -1.01831293e+00 -8.59049261e-01 -1.14899683e+00
-4.38649207e-01 -5.92657804e-01 7.39341557e-01 6.65834486e-01
7.52764568e-03 -4.94115427e-02 -6.06429100e-01 6.78290576e-02
4.23574656e-01 4.38971728e-01 8.13526571e-01 -9.85124052e-01
-3.87532413e-01 -7.98152804e-01 -7.81579196e-01 -1.34612477e+00
-1.32615879e-01 -5.79384148e-01 -1.25178099e-01 -1.60058713e+00
3.11205685e-01 -2.44414821e-01 -6.77682757e-01 4.24049914e-01
-3.56999755e-01 2.39254892e-01 5.64807832e-01 5.61066031e-01
-1.15155303e+00 5.39148450e-01 1.16104615e+00 2.65170485e-02
-1.83260888e-01 1.16977252e-01 -2.86760300e-01 9.69193876e-01
8.69107902e-01 -1.99159876e-01 -4.82028872e-01 -2.87047148e-01
-7.45547414e-02 6.93724379e-02 4.74191934e-01 -1.41168392e+00
4.75516081e-01 -4.07626122e-01 1.11106988e-02 -6.40468776e-01
4.10028428e-01 -9.00533080e-01 3.24090958e-01 6.08719528e-01
-2.94871509e-01 -1.95290521e-02 -1.74410179e-01 7.77002752e-01
-4.45956558e-01 -9.55573842e-02 7.35273957e-01 -2.14538455e-01
-1.36291778e+00 6.91491961e-01 -3.27824980e-01 -5.65843061e-02
1.30398262e+00 -6.65483654e-01 -1.96212739e-01 -3.93276989e-01
-8.56395483e-01 2.87022293e-01 5.02211273e-01 5.53618312e-01
5.61701298e-01 -1.61446393e+00 -5.86541414e-01 -1.26955897e-01
2.27393627e-01 -4.19845939e-01 5.61921656e-01 1.12139547e+00
-2.91704774e-01 2.57909130e-02 -1.84743598e-01 -7.90819764e-01
-1.33821392e+00 7.23696291e-01 2.75526047e-01 -8.87690261e-02
-6.90001249e-01 6.19218767e-01 5.00825465e-01 3.72382134e-01
3.10006440e-01 -1.36690021e-01 -5.30431032e-01 2.62635052e-01
7.93577611e-01 5.96342325e-01 -4.60432917e-01 -9.42953110e-01
-4.98711169e-01 7.27132797e-01 1.10549375e-01 2.20146269e-01
1.09388983e+00 -4.19895828e-01 1.08049385e-01 6.07003689e-01
1.05595529e+00 -2.97626257e-01 -1.58542013e+00 -4.09716666e-01
5.02609927e-03 -7.53762662e-01 -2.28898525e-01 -3.64644647e-01
-1.25226104e+00 7.49243796e-01 4.89260733e-01 3.77934635e-01
1.59414256e+00 -1.56104527e-02 6.60624623e-01 1.83738440e-01
5.69350004e-01 -1.26974535e+00 3.58812362e-01 3.61116827e-01
8.66418481e-01 -1.15934372e+00 5.19062877e-02 -4.68924582e-01
-6.64725006e-01 1.15330231e+00 7.69391000e-01 -2.77303874e-01
6.30249262e-01 -6.07265644e-02 -9.60744768e-02 -1.10686064e-01
-8.50460231e-01 -5.24290144e-01 2.68367529e-01 3.33585769e-01
3.57106030e-01 1.08407913e-02 -7.31358290e-01 5.12129366e-01
3.11884105e-01 2.45884415e-02 2.76470661e-01 8.23653996e-01
-5.05081952e-01 -9.33713615e-01 -8.12706724e-02 4.15701628e-01
-3.63961548e-01 1.28095403e-01 5.37466258e-02 6.89309061e-01
3.73641044e-01 1.06535029e+00 1.27818733e-01 -5.94734311e-01
4.41088498e-01 -3.76783460e-01 1.17996931e-01 -4.94749695e-01
-5.31368554e-01 1.73377112e-01 -4.18156646e-02 -9.97732282e-01
-9.88925278e-01 -7.30545521e-01 -1.20473802e+00 -3.25039983e-01
-2.12070838e-01 1.09002158e-01 1.34551704e-01 9.09803808e-01
5.35229385e-01 6.03466034e-01 8.50409508e-01 -9.66292739e-01
-2.54626751e-01 -7.78139710e-01 -4.64446038e-01 6.59138083e-01
3.91547829e-02 -8.86158943e-01 -3.88979316e-01 1.82266787e-01]
|
[8.47673511505127, 0.578626275062561]
|
4a12ee54-dee5-4409-a470-7efcd4b31fd7
|
regularizing-disparity-estimation-via-multi
|
2301.08140
| null |
https://arxiv.org/abs/2301.08140v1
|
https://arxiv.org/pdf/2301.08140v1.pdf
|
Regularizing disparity estimation via multi task learning with structured light reconstruction
|
3D reconstruction is a useful tool for surgical planning and guidance. However, the lack of available medical data stunts research and development in this field, as supervised deep learning methods for accurate disparity estimation rely heavily on large datasets containing ground truth information. Alternative approaches to supervision have been explored, such as self-supervision, which can reduce or remove entirely the need for ground truth. However, no proposed alternatives have demonstrated performance capabilities close to what would be expected from a supervised setup. This work aims to alleviate this issue. In this paper, we investigate the learning of structured light projections to enhance the development of direct disparity estimation networks. We show for the first time that it is possible to accurately learn the projection of structured light on a scene, implicitly learning disparity. Secondly, we \textcolor{black}{explore the use of a multi task learning (MTL) framework for the joint training of structured light and disparity. We present results which show that MTL with structured light improves disparity training; without increasing the number of model parameters. Our MTL setup outperformed the single task learning (STL) network in every validation test. Notably, in the medical generalisation test, the STL error was 1.4 times worse than that of the best MTL performance. The benefit of using MTL is emphasised when the training data is limited.} A dataset containing stereoscopic images, disparity maps and structured light projections on medical phantoms and ex vivo tissue was created for evaluation together with virtual scenes. This dataset will be made publicly available in the future.
|
['Stamatia Giannarou', 'Joseph Davids', 'Chi Xu', 'Joao Cartucho', 'Alistair Weld']
|
2023-01-19
| null | null | null | null |
['disparity-estimation']
|
['computer-vision']
|
[ 4.78426188e-01 3.96645427e-01 -3.39325033e-02 -4.70114797e-01
-7.87257195e-01 -2.23731771e-01 5.93406379e-01 5.17645441e-02
-7.71408975e-01 9.17441428e-01 1.82876945e-01 -5.28679788e-01
1.11279823e-02 -5.46716511e-01 -7.87730575e-01 -9.60248232e-01
2.75401622e-01 3.97625893e-01 1.83094636e-01 -1.57295261e-02
1.25953004e-01 3.80269319e-01 -1.42152488e+00 5.83748639e-01
6.25414550e-01 7.29236841e-01 6.34474218e-01 5.37900686e-01
-2.35569756e-02 6.60484910e-01 -3.94838274e-01 -1.60385400e-01
5.12724400e-01 -4.69861299e-01 -6.99503005e-01 -9.36276838e-03
7.08776236e-01 -4.04707372e-01 -2.64117289e-02 7.84650505e-01
1.00178349e+00 -2.58262008e-01 3.23795736e-01 -7.89975822e-01
4.08890098e-02 1.73065737e-01 -5.39050579e-01 2.44850814e-01
3.19058388e-01 4.06037599e-01 4.36762691e-01 -6.65512025e-01
9.37010169e-01 7.88383603e-01 6.92081392e-01 6.64642811e-01
-1.31474340e+00 -6.10536277e-01 -1.50633201e-01 -1.61456287e-01
-9.12743866e-01 -4.27096367e-01 7.32617319e-01 -5.15613794e-01
8.69696617e-01 1.90388411e-01 8.30278993e-01 1.10919750e+00
5.16129434e-01 5.58571517e-01 1.74272060e+00 -7.26913452e-01
1.13676168e-01 4.54704046e-01 -1.25824645e-01 9.61157203e-01
3.18277240e-01 6.81860805e-01 -4.98979539e-01 2.28664488e-01
9.66300488e-01 -1.66930333e-01 -4.52956080e-01 -6.69165492e-01
-1.18063200e+00 6.55937493e-01 6.16513669e-01 3.17228824e-01
-1.76596671e-01 9.09282938e-02 2.94579715e-01 2.58074731e-01
6.49331033e-01 5.19640207e-01 -1.95252076e-01 2.40087975e-02
-7.97407448e-01 -2.07317382e-01 7.37410128e-01 4.84010339e-01
7.25798190e-01 -2.56380122e-02 3.79316248e-02 6.64716184e-01
3.24563652e-01 3.64255756e-01 4.31164712e-01 -9.61899757e-01
3.75162065e-01 4.26238835e-01 -1.16404042e-01 -6.71567023e-01
-8.51399779e-01 -5.52931845e-01 -7.61852920e-01 1.01704776e+00
7.21946239e-01 -3.60417098e-01 -1.17349589e+00 1.44761157e+00
2.23844692e-01 1.70532390e-01 1.10473223e-01 1.01857269e+00
1.06813824e+00 1.20869942e-01 -1.88796058e-01 -1.15249708e-01
8.96077991e-01 -7.65363693e-01 -5.81838787e-01 -3.56064409e-01
1.01192343e+00 -8.65728438e-01 1.03559589e+00 5.90181112e-01
-1.02868581e+00 -3.45074952e-01 -9.66069162e-01 9.53162014e-02
-8.38861912e-02 5.85204139e-02 8.62879515e-01 9.57732022e-01
-1.32913232e+00 5.06254554e-01 -1.05145240e+00 -1.79316118e-01
6.71546340e-01 6.64589286e-01 -5.33495784e-01 -4.44439471e-01
-8.31927240e-01 1.16406465e+00 1.95093572e-01 1.99014187e-01
-8.63679707e-01 -7.73246050e-01 -1.02223325e+00 -5.26316524e-01
1.34388581e-01 -8.87206376e-01 9.78610277e-01 -9.94693816e-01
-1.54656732e+00 1.42753530e+00 1.42321199e-01 -5.30452549e-01
8.06828618e-01 -1.86760537e-02 4.06278186e-02 8.88097957e-02
8.43349192e-03 8.96048844e-01 4.05032396e-01 -1.59405541e+00
-3.24642628e-01 -4.39421594e-01 1.70942843e-01 2.68500119e-01
7.28895962e-02 -2.99192965e-01 -3.21150154e-01 -2.85423219e-01
1.08301498e-01 -1.15128863e+00 -4.91141826e-01 3.45774382e-01
-4.23994184e-01 4.70290363e-01 2.25640401e-01 -5.17713547e-01
5.70083797e-01 -2.07209373e+00 -1.74016610e-01 1.45375192e-01
8.46289620e-02 2.27603331e-01 2.14214530e-02 9.72722247e-02
-1.90466553e-01 -3.00924718e-01 -2.94532716e-01 -5.16059577e-01
-4.93994951e-01 3.78722280e-01 1.78499341e-01 6.88011348e-01
-1.38374060e-01 7.57648110e-01 -7.50984251e-01 -6.13530099e-01
5.58758438e-01 5.42398095e-01 -6.56250298e-01 -4.80012633e-02
6.95114955e-02 1.00514865e+00 -3.75813283e-02 3.44327092e-01
6.29095972e-01 -2.00623050e-01 -2.33501866e-02 -1.21658288e-01
-2.71330386e-01 1.53437674e-01 -9.38575566e-01 2.22267199e+00
-9.20333207e-01 8.61613870e-01 8.17788541e-02 -6.92305267e-01
8.29546988e-01 3.88176590e-01 6.54880226e-01 -1.02846980e+00
6.20650388e-02 5.12073815e-01 2.66770542e-01 -5.32283247e-01
-1.02378048e-01 -6.94822550e-01 4.65503931e-01 2.27696002e-01
-1.30190432e-01 -5.04760683e-01 -1.40852585e-01 -1.46741450e-01
8.44606042e-01 3.68870318e-01 7.27967247e-02 -2.86973268e-01
2.86022663e-01 2.80567914e-01 3.01229775e-01 5.71024656e-01
-9.46571752e-02 8.13107908e-01 3.10319006e-01 -5.91040790e-01
-7.83524454e-01 -9.09903884e-01 -6.06185913e-01 4.58555579e-01
2.14079127e-01 -6.13250136e-02 -4.25062746e-01 -5.96293449e-01
-1.72793493e-01 6.85460627e-01 -7.33128428e-01 3.01311165e-02
-5.27219117e-01 -1.02061856e+00 1.96724474e-01 2.82676101e-01
3.59107345e-01 -1.06938112e+00 -9.66887534e-01 -5.03978431e-02
-5.16243167e-02 -1.22103918e+00 1.78498358e-01 5.44201255e-01
-1.12429380e+00 -1.09712803e+00 -8.79121006e-01 -6.88760579e-01
7.71309793e-01 7.44487643e-02 9.88801718e-01 5.40164933e-02
-5.30514657e-01 3.82473141e-01 5.39024919e-02 -7.27474332e-01
-5.07260978e-01 -1.08758084e-01 -3.32949162e-01 -3.64906788e-01
-1.43935857e-02 -5.02842605e-01 -8.92382860e-01 1.75863191e-01
-7.53276944e-01 6.02344632e-01 8.62342536e-01 1.00339079e+00
6.59481406e-01 -4.19338137e-01 1.52858630e-01 -1.33714950e+00
1.62405238e-01 -2.40711700e-02 -5.55917799e-01 -1.40620679e-01
-7.14941263e-01 1.45289481e-01 2.57098585e-01 3.37430201e-02
-1.24397743e+00 3.24928910e-01 -3.52746516e-01 -3.45996588e-01
-2.98708230e-01 3.39850456e-01 3.67022783e-01 -4.00757670e-01
8.94593060e-01 -1.03637695e-01 3.09819669e-01 -2.94584036e-01
-1.57162711e-01 1.44214824e-01 2.46019587e-01 -2.83435971e-01
4.28093255e-01 9.02384818e-01 4.65335518e-01 -6.85330749e-01
-9.46425259e-01 -4.47288126e-01 -7.34046578e-01 -3.24284822e-01
9.14156973e-01 -9.62800205e-01 -2.77409613e-01 3.26240629e-01
-9.37531650e-01 -8.38217795e-01 -4.40075070e-01 8.96792173e-01
-7.32116401e-01 1.87189460e-01 -4.96279180e-01 -4.74910647e-01
-7.50560910e-02 -1.43164694e+00 1.11405993e+00 -7.26059526e-02
-4.16798815e-02 -1.39700520e+00 3.24661434e-01 6.26735270e-01
3.44865322e-01 6.11247361e-01 6.94207072e-01 -1.69885054e-01
-6.45590425e-01 -1.64456666e-01 -2.49569789e-01 2.57610470e-01
2.24431947e-01 -4.27053034e-01 -1.45074034e+00 -4.36929345e-01
2.06454068e-01 -4.09435183e-01 1.02062619e+00 8.94445121e-01
9.86513734e-01 2.86471516e-01 -3.23396891e-01 9.44343865e-01
1.67312956e+00 -1.22404238e-02 7.36761749e-01 6.59898221e-01
7.05189288e-01 7.44121850e-01 5.04706740e-01 1.89201668e-01
1.97265655e-01 7.23427057e-01 7.44458318e-01 -8.02970767e-01
-6.80238545e-01 6.80185854e-02 1.51245538e-02 3.21127504e-01
-3.31131548e-01 1.83071047e-01 -1.22399974e+00 2.59388894e-01
-1.27777791e+00 -6.32656753e-01 -3.79427940e-01 2.34885693e+00
7.70245016e-01 4.27524716e-01 -1.79289669e-01 1.72604457e-01
1.79630473e-01 -8.16228017e-02 -3.03113610e-01 -2.22587630e-01
-2.10738257e-02 3.19130450e-01 6.60457551e-01 8.26685488e-01
-9.25996006e-01 5.76054752e-01 5.98821211e+00 3.00333172e-01
-1.42591870e+00 1.06822126e-01 7.97932148e-01 -2.07239851e-01
-3.67931485e-01 -1.31337956e-01 -6.17136717e-01 2.89383411e-01
6.78425610e-01 2.70692140e-01 -1.13683967e-02 4.92908299e-01
4.57083732e-01 -6.83322549e-01 -1.13154614e+00 1.03500605e+00
1.70267612e-01 -1.36299062e+00 -2.74979949e-01 3.04701060e-01
9.41489697e-01 3.74734223e-01 8.06077048e-02 9.19383578e-03
1.28118917e-01 -1.13630617e+00 1.89611211e-01 5.16953766e-01
9.62304533e-01 -3.29295456e-01 1.09113133e+00 5.17007709e-01
-4.19697523e-01 1.23120755e-01 -1.48544192e-01 -3.24392729e-02
1.37316167e-01 7.01514304e-01 -1.20974147e+00 6.16069674e-01
5.24830401e-01 7.23710835e-01 -7.06105471e-01 1.36948526e+00
-1.47450909e-01 2.60636896e-01 -3.12010437e-01 2.13986576e-01
2.53814518e-01 -5.23752570e-02 2.70095646e-01 1.17574763e+00
9.89641547e-02 -1.72066897e-01 1.16658874e-01 4.60163265e-01
3.54474127e-01 1.35353163e-01 -8.62507403e-01 6.75782859e-01
-4.81896132e-01 1.05531597e+00 -8.87900829e-01 3.86416540e-02
-5.64995408e-01 8.47113669e-01 1.18247554e-01 1.77549005e-01
-4.80777264e-01 1.35758549e-01 7.08018467e-02 4.73902553e-01
-1.89411193e-01 1.71206072e-02 -6.40502334e-01 -8.97900522e-01
-8.51582922e-03 -6.30962551e-01 4.00547951e-01 -7.75843680e-01
-9.79862750e-01 7.20906198e-01 3.88112850e-03 -1.31340194e+00
-3.24057758e-01 -8.29233766e-01 -3.30390692e-01 9.73871946e-01
-1.99616754e+00 -9.54387903e-01 -7.10301280e-01 5.85081041e-01
4.15824115e-01 9.12977159e-02 9.15054798e-01 3.12410027e-01
-2.45845646e-01 1.88641265e-01 -6.43569008e-02 -1.11278363e-01
9.62977350e-01 -1.35620213e+00 -1.87877163e-01 5.45268297e-01
1.77483425e-01 2.17692211e-01 6.56928122e-01 -4.70754027e-01
-1.02624941e+00 -7.00914025e-01 5.33589303e-01 -6.15823269e-01
2.47863814e-01 -1.13963686e-01 -5.89979172e-01 5.46197355e-01
2.02793077e-01 4.15590107e-01 8.07605863e-01 -1.74318836e-03
1.72358453e-01 -1.11190528e-01 -1.18870509e+00 3.33065808e-01
8.73842776e-01 -2.62862444e-01 -3.09617668e-01 5.15564859e-01
3.06744665e-01 -9.27571237e-01 -6.00467443e-01 5.65607727e-01
4.64468271e-01 -1.57804656e+00 8.73753250e-01 -9.56486538e-02
6.53628707e-01 -5.18152080e-02 7.65781403e-02 -1.33317566e+00
2.44655073e-01 -1.88417599e-01 3.57629895e-01 4.97103810e-01
5.31848967e-01 -6.54071867e-01 1.20765626e+00 6.31685019e-01
-5.53304553e-01 -8.96021783e-01 -1.10221553e+00 -5.68047523e-01
2.71479249e-01 -4.78615880e-01 -2.35757992e-01 8.48727345e-01
-2.29096815e-01 2.03932241e-01 -1.08005285e-01 4.95349206e-02
7.34129190e-01 1.81812510e-01 6.67486370e-01 -1.14119542e+00
-5.44846714e-01 -2.91571409e-01 -4.85209405e-01 -7.69234300e-01
6.37640730e-02 -1.19232559e+00 9.47104245e-02 -1.94832003e+00
1.81223333e-01 -6.92934930e-01 -1.49170309e-01 4.86285061e-01
-6.02202863e-03 4.62334365e-01 1.10121936e-01 9.17240232e-02
-1.24891207e-01 1.93919614e-01 1.82482755e+00 7.47547299e-02
-2.73217827e-01 4.08678800e-01 -2.65205324e-01 7.20231533e-01
8.45407784e-01 -4.65874404e-01 -6.10665262e-01 -6.51014328e-01
6.91579878e-02 2.43243039e-01 6.11964405e-01 -1.14468408e+00
2.62599260e-01 1.80326954e-01 5.35768747e-01 -4.35388386e-01
6.40798807e-01 -1.01557708e+00 1.14836209e-01 7.79012740e-01
-2.14638144e-01 -2.35407621e-01 5.36154628e-01 4.81053978e-01
-1.84692517e-01 -2.20015123e-01 9.10244107e-01 -4.10223842e-01
-5.51682889e-01 4.28074449e-02 -7.08245188e-02 -1.31886706e-01
1.01458466e+00 -5.52842319e-01 -9.37107801e-02 -2.25979716e-01
-9.51520145e-01 -5.76799251e-02 5.11470139e-01 2.39735320e-02
6.46377206e-01 -9.19400930e-01 -7.40744054e-01 2.86426604e-01
2.01511271e-02 2.99312145e-01 2.02476338e-01 1.19569623e+00
-7.94508517e-01 4.78941172e-01 -3.99807513e-01 -9.41356242e-01
-1.35141766e+00 3.60626936e-01 6.83028102e-01 -3.70026708e-01
-8.31114173e-01 7.91843474e-01 4.24861282e-01 -5.58632016e-01
1.55285954e-01 -4.16519344e-01 -1.16416171e-01 -2.70208567e-01
2.13171735e-01 3.19608226e-02 3.68879795e-01 -2.96840429e-01
-2.70548314e-01 7.31147170e-01 4.50219326e-02 -2.29445755e-01
1.56932783e+00 -2.12745965e-02 1.28732964e-01 5.21748960e-01
1.17863023e+00 -1.12584069e-01 -1.55561674e+00 -2.32808385e-02
-1.12805642e-01 -6.42082334e-01 4.40494597e-01 -1.07252932e+00
-1.26703012e+00 1.11675727e+00 1.00169349e+00 -2.63200402e-01
1.10953796e+00 -1.22146919e-01 2.71267653e-01 1.95375293e-01
5.31318009e-01 -7.07262754e-01 1.70921013e-01 3.74775752e-02
7.15850353e-01 -1.81185484e+00 1.88824460e-01 -6.62373781e-01
-6.02029324e-01 1.04538655e+00 6.35211527e-01 8.20863247e-03
6.63609624e-01 6.07084930e-01 4.50161129e-01 -4.99687523e-01
-5.28141022e-01 -2.53858000e-01 2.63630271e-01 7.13160336e-01
6.85025632e-01 -3.14196497e-01 -1.00254998e-01 -1.87545747e-01
-2.17958644e-01 2.57216513e-01 5.52387536e-01 9.94669676e-01
-9.91031975e-02 -1.13287747e+00 -2.49174282e-01 3.98389459e-01
-4.48615938e-01 -1.65694758e-01 -9.90718529e-02 9.65833426e-01
2.05579981e-01 5.87198257e-01 -5.60142286e-02 7.82964900e-02
3.89551729e-01 -2.60962754e-01 7.63536453e-01 -8.41133118e-01
-6.41385972e-01 8.82681236e-02 2.55795777e-01 -5.75547397e-01
-7.08858550e-01 -6.18789852e-01 -1.10023975e+00 1.22891001e-01
-2.18226299e-01 -3.15429747e-01 8.92805457e-01 7.58798242e-01
-9.44287851e-02 6.77635312e-01 4.54762071e-01 -9.18415904e-01
-1.11957647e-01 -5.12056172e-01 -3.91361386e-01 3.57547373e-01
5.68181455e-01 -5.12980819e-01 -4.02076125e-01 4.51570144e-03]
|
[14.043069839477539, -2.960660219192505]
|
a10a159a-85c2-443f-b33b-362ec4b52bd4
|
dvqa-understanding-data-visualizations-via
|
1801.08163
| null |
http://arxiv.org/abs/1801.08163v2
|
http://arxiv.org/pdf/1801.08163v2.pdf
|
DVQA: Understanding Data Visualizations via Question Answering
|
Bar charts are an effective way to convey numeric information, but today's
algorithms cannot parse them. Existing methods fail when faced with even minor
variations in appearance. Here, we present DVQA, a dataset that tests many
aspects of bar chart understanding in a question answering framework. Unlike
visual question answering (VQA), DVQA requires processing words and answers
that are unique to a particular bar chart. State-of-the-art VQA algorithms
perform poorly on DVQA, and we propose two strong baselines that perform
considerably better. Our work will enable algorithms to automatically extract
numeric and semantic information from vast quantities of bar charts found in
scientific publications, Internet articles, business reports, and many other
areas.
|
['Brian Price', 'Kushal Kafle', 'Scott Cohen', 'Christopher Kanan']
|
2018-01-24
|
dvqa-understanding-data-visualizations-via-1
|
http://openaccess.thecvf.com/content_cvpr_2018/html/Kafle_DVQA_Understanding_Data_CVPR_2018_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2018/papers/Kafle_DVQA_Understanding_Data_CVPR_2018_paper.pdf
|
cvpr-2018-6
|
['chart-question-answering', 'chart-question-answering']
|
['computer-code', 'computer-vision']
|
[-1.00238122e-01 -9.99586284e-03 -2.12000296e-01 -2.71507382e-01
-1.35582876e+00 -1.25982857e+00 5.31784594e-01 7.76046336e-01
1.20918088e-01 5.59180379e-01 4.87747818e-01 -1.02893186e+00
-1.31477460e-01 -9.41727340e-01 -6.06448472e-01 9.20286924e-02
3.91958430e-02 4.61872101e-01 4.10366744e-01 -4.43882614e-01
7.34666109e-01 6.04729950e-01 -1.16684675e+00 9.15695429e-01
7.17963755e-01 1.10044694e+00 -3.68419707e-01 1.10215163e+00
-1.19195557e+00 1.28876817e+00 -1.20431280e+00 -9.70514178e-01
-1.24341831e-01 -4.49614882e-01 -9.34259832e-01 -1.72909811e-01
1.12052834e+00 -1.95937365e-01 -7.44343996e-02 9.24690008e-01
3.20372045e-01 -1.05259359e-01 9.40764725e-01 -1.39717996e+00
-1.56402576e+00 2.48419315e-01 -8.46637368e-01 6.22221410e-01
9.71220016e-01 5.02328798e-02 1.52406752e+00 -9.27631915e-01
8.46094131e-01 1.52697897e+00 5.05659640e-01 3.06175977e-01
-1.07852077e+00 -2.26512596e-01 1.96440220e-01 3.98608088e-01
-9.67612982e-01 -7.25545585e-02 7.86311388e-01 -5.99635601e-01
8.33310544e-01 6.21435463e-01 3.73711705e-01 8.46990168e-01
-5.12222052e-02 9.41899836e-01 1.06357467e+00 -3.50241542e-01
3.93591046e-01 -1.94037199e-01 3.61658096e-01 9.26156580e-01
2.46865302e-01 -6.31988645e-01 -5.49203753e-01 -1.66576996e-01
9.25207257e-01 -3.11185718e-01 4.28660326e-02 -5.56886315e-01
-1.24454749e+00 1.05009961e+00 5.87142229e-01 8.90826806e-03
8.57625529e-02 2.50940740e-01 6.84799910e-01 4.33797956e-01
1.16722666e-01 8.29488933e-01 -2.43989408e-01 -2.06837267e-01
-6.13728702e-01 5.34618318e-01 7.41319776e-01 1.06094325e+00
7.14843750e-01 -1.55601189e-01 -7.46685982e-01 5.10261118e-01
1.98392943e-01 7.16097116e-01 1.44169973e-02 -8.85960340e-01
8.97373438e-01 8.93637776e-01 1.68021306e-01 -1.38626266e+00
-4.27468508e-01 1.25427321e-01 -2.96625733e-01 3.20691526e-01
7.64278591e-01 1.80148363e-01 -1.46239841e+00 9.40322161e-01
1.84044898e-01 -6.49056852e-01 3.28326561e-02 8.45356822e-01
1.56094193e+00 7.99764395e-01 3.04644376e-01 2.98955798e-01
1.94630098e+00 -8.84581864e-01 -1.15387213e+00 -4.30702150e-01
6.18120313e-01 -8.81286621e-01 1.69547081e+00 1.63243636e-01
-9.74318743e-01 -3.73230577e-01 -1.20659411e+00 -7.48815238e-01
-9.98466611e-01 -1.60284221e-01 6.52611554e-01 8.17109585e-01
-9.75513458e-01 -6.91955388e-02 -2.05577001e-01 -1.52268678e-01
8.12261820e-01 -3.30241621e-01 -3.17056388e-01 -9.00663510e-02
-9.07434165e-01 9.73711312e-01 -1.44311503e-01 -2.38443822e-01
-3.54363561e-01 -1.05907524e+00 -1.24065995e+00 4.68805581e-02
5.92381179e-01 -7.28130460e-01 1.67478240e+00 -5.68993092e-01
-8.95558059e-01 8.95927846e-01 -4.00071144e-01 -4.41419959e-01
4.38385993e-01 -1.89223528e-01 -4.57209051e-01 6.86091721e-01
2.19222650e-01 6.40674353e-01 7.63894856e-01 -1.42931414e+00
-2.35355929e-01 -5.52393615e-01 5.14713943e-01 2.11926606e-02
1.71672076e-01 1.18167281e-01 -5.60997009e-01 -7.37501562e-01
-1.95422977e-01 -1.29774243e-01 5.75807281e-02 5.32359481e-01
-4.13858891e-01 -2.22307146e-01 1.10417807e+00 -9.16617095e-01
1.36673927e+00 -1.95149148e+00 -1.82028592e-01 2.54806221e-01
3.84675354e-01 -5.03444299e-02 -1.82450324e-01 4.95994091e-01
1.86294660e-01 5.27844191e-01 -2.13327855e-01 1.36902466e-01
2.99006402e-01 6.84039518e-02 -8.11623216e-01 -6.53298199e-02
4.55879331e-01 1.46657002e+00 -8.85743260e-01 -9.23967779e-01
2.33335897e-01 1.58416882e-01 -1.50291860e-01 3.64866108e-02
-7.72529364e-01 -1.22732081e-01 -5.15908122e-01 1.08229530e+00
4.42812890e-01 -6.38262868e-01 -2.18472779e-01 -7.99883306e-02
1.06264204e-01 -1.05956951e-02 -7.56028652e-01 1.67759478e+00
-1.94492400e-01 1.16624916e+00 -4.04329360e-01 -4.75749433e-01
1.12784982e+00 -6.01618085e-03 -8.24215263e-02 -1.32107985e+00
-1.32598713e-01 -1.41371652e-01 -3.86353970e-01 -6.44952476e-01
6.89946473e-01 1.09850101e-01 -2.61493206e-01 2.07867488e-01
-1.67766690e-01 -6.56853616e-01 5.83920598e-01 7.04968989e-01
1.05122089e+00 -2.36514494e-01 5.45866787e-01 9.26652700e-02
4.01485652e-01 6.07196689e-01 -2.96682835e-01 1.02782869e+00
-3.60500664e-01 8.33189249e-01 1.06460440e+00 -5.57762563e-01
-1.25578356e+00 -1.57307160e+00 6.56013787e-02 1.18772745e+00
2.39522755e-02 -6.97442353e-01 -5.64567804e-01 -9.26222146e-01
2.84042358e-01 7.70482481e-01 -1.08544052e+00 4.85029429e-01
-3.49237382e-01 -1.40160531e-01 4.04939115e-01 1.10109305e+00
2.01572329e-01 -1.06175983e+00 -6.38404131e-01 -2.18321964e-01
1.89062893e-01 -9.35817480e-01 -2.23358899e-01 -1.17852248e-01
-7.09306598e-01 -1.18236113e+00 -8.88179123e-01 -6.70344114e-01
5.62597036e-01 1.98641360e-01 1.63800228e+00 2.29426101e-02
-4.98191506e-01 6.96020544e-01 -3.65374207e-01 -9.37678576e-01
-1.76781774e-01 -1.16666131e-01 -1.10741091e+00 -7.02127993e-01
4.71961021e-01 1.93875849e-01 -5.37032247e-01 -6.02243096e-02
-9.94247735e-01 -1.55300289e-01 3.33961785e-01 3.93665791e-01
5.75936079e-01 -7.27864981e-01 6.82517946e-01 -1.29074502e+00
1.02505553e+00 -3.25550407e-01 -5.67518950e-01 6.51169658e-01
-2.02669218e-01 2.17148229e-01 5.65916896e-01 9.54943970e-02
-9.35215533e-01 -3.37382257e-01 -1.05807856e-01 -1.04895368e-01
-1.26994506e-01 5.10276496e-01 -2.80228332e-02 3.00012112e-01
1.02455866e+00 -3.50376874e-01 -1.57091349e-01 -3.60317022e-01
1.26387453e+00 2.02970132e-01 9.32602584e-01 -3.72599989e-01
9.05817688e-01 7.20525861e-01 -8.29969347e-02 -5.12920678e-01
-1.10155332e+00 -5.87246060e-01 -4.11020190e-01 -3.28453213e-01
1.16478467e+00 -5.92087090e-01 -6.77298427e-01 -2.26937667e-01
-1.24226213e+00 8.34081173e-02 -3.54597926e-01 -2.93845445e-01
-4.62151170e-01 3.24375719e-01 -3.11463594e-01 -7.15759099e-01
-2.82121927e-01 -6.60416722e-01 1.25738883e+00 4.00346130e-01
-3.66789997e-01 -1.14997435e+00 1.17609158e-01 5.17009079e-01
3.69427592e-01 4.57675815e-01 1.43358123e+00 -5.17329991e-01
-5.26643157e-01 -6.17002435e-02 -8.82897675e-01 -2.21391857e-01
-6.51777610e-02 3.98305535e-01 -8.49311590e-01 2.40264580e-01
-7.64834881e-01 -6.04579687e-01 8.81416082e-01 2.68616259e-01
1.56833541e+00 -3.90332639e-01 -1.16837062e-01 2.37275913e-01
1.37378311e+00 6.17941618e-01 8.62758756e-01 6.33282423e-01
8.78257751e-01 5.65685570e-01 4.98145908e-01 2.00170279e-01
6.16936803e-01 1.65621281e-01 6.06729805e-01 -4.47349906e-01
-1.98954523e-01 -4.63333666e-01 -1.92359418e-01 4.26671147e-01
3.08187902e-01 -2.91670799e-01 -1.22584951e+00 6.72510862e-01
-1.78543556e+00 -9.64188337e-01 -3.23378444e-01 1.61708283e+00
6.69794381e-01 2.17534423e-01 2.40061820e-01 7.29472116e-02
3.71526837e-01 5.33164084e-01 -3.80317241e-01 -9.63544965e-01
-2.90350288e-01 4.24485475e-01 3.34904999e-01 2.23173350e-01
-1.12805319e+00 6.75671637e-01 7.93502903e+00 6.53750181e-01
-5.73308945e-01 -3.47111970e-01 7.76699185e-01 2.84062535e-01
-9.27055597e-01 -1.77452385e-01 -2.57456839e-01 -7.58252591e-02
7.21514165e-01 -3.01748633e-01 1.42674232e-02 9.87687647e-01
-2.74661481e-01 -2.56624483e-02 -1.12676394e+00 1.24537933e+00
4.49722350e-01 -1.95647514e+00 6.42347634e-01 -6.39307320e-01
6.45085454e-01 -8.41052294e-01 5.33779442e-01 3.52670074e-01
4.06648964e-01 -1.63746262e+00 8.36691141e-01 6.12748861e-01
7.73312986e-01 -7.36932993e-01 5.35020530e-01 -5.88709831e-01
-1.12582695e+00 3.66923623e-02 -2.93048620e-01 -4.79792878e-02
4.38621193e-02 2.40303546e-01 -6.99222565e-01 5.90144396e-01
7.87691295e-01 3.50704312e-01 -1.36732483e+00 1.08988249e+00
-4.02879238e-01 4.37075078e-01 3.70380938e-01 -4.98844355e-01
4.90479678e-01 5.17275110e-02 6.80261478e-02 1.35091746e+00
2.03688920e-01 -5.21041676e-02 -2.99256384e-01 7.04892397e-01
-2.37068683e-01 5.06836236e-01 -8.79056275e-01 -4.44581151e-01
3.32190037e-01 9.02229071e-01 -9.52658951e-01 -6.80432081e-01
-8.96300435e-01 6.61195636e-01 4.88643020e-01 6.15128636e-01
-7.84578264e-01 -8.18310857e-01 3.98623884e-01 -1.88294470e-01
4.14798379e-01 -3.47194701e-01 -8.33169937e-01 -7.95727015e-01
1.67707443e-01 -1.11780560e+00 1.12856913e+00 -1.57720327e+00
-1.48872876e+00 3.24867398e-01 -1.32708430e-01 -9.24512148e-01
-2.59997725e-01 -1.11375856e+00 -3.73818398e-01 5.41645229e-01
-1.35896635e+00 -9.93884206e-01 -5.66100180e-01 4.42706794e-01
6.07364774e-01 -2.13873573e-02 8.07080567e-01 -1.32117569e-01
-3.82363014e-02 2.77226597e-01 1.50936946e-01 6.46847188e-01
8.09303939e-01 -1.91578269e+00 9.27399099e-01 5.32269895e-01
7.30693758e-01 4.42927092e-01 7.62517273e-01 -4.47419435e-01
-1.41057158e+00 -6.82869196e-01 5.25859714e-01 -1.09864044e+00
1.03344297e+00 -4.70158428e-01 -1.19516563e+00 7.08931625e-01
4.25598621e-01 -7.66974688e-02 7.48558164e-01 1.67431936e-01
-1.24337339e+00 2.38075983e-02 -8.87859106e-01 8.69013548e-01
7.13599741e-01 -9.80221629e-01 -1.26212633e+00 3.13076913e-01
9.05914426e-01 -5.55069566e-01 -5.64434588e-01 9.44703668e-02
4.37136143e-01 -6.29855871e-01 1.25976598e+00 -1.16855145e+00
8.43573272e-01 -5.36048055e-01 5.94507046e-02 -1.20730507e+00
-7.60946656e-03 -3.34902644e-01 -3.65657002e-01 8.84677708e-01
6.55538261e-01 -4.48039584e-02 7.91020155e-01 4.09235775e-01
2.47387774e-02 -3.31603169e-01 -7.86756516e-01 -5.80206335e-01
2.52211809e-01 -5.48414767e-01 7.08084404e-01 9.50819254e-01
1.93994865e-01 6.03799582e-01 7.50408918e-02 -2.45757028e-01
2.76815057e-01 4.11066055e-01 8.72912824e-01 -9.84339952e-01
2.60997474e-01 -6.98608160e-01 -6.24044657e-01 -9.48202550e-01
-2.15490416e-01 -6.30510092e-01 -2.56717980e-01 -2.50741911e+00
9.58087966e-02 1.85794771e-01 -7.89739937e-02 4.26343262e-01
-4.28045154e-01 4.01815951e-01 3.40025544e-01 -2.50241309e-01
-9.44380045e-01 1.02521330e-01 1.57874250e+00 -5.64806700e-01
8.67767707e-02 -6.65510893e-01 -9.49421585e-01 5.60597181e-01
5.66664696e-01 5.71096465e-02 -6.56584501e-01 -6.54825926e-01
7.16357410e-01 1.05261147e-01 5.14781654e-01 -5.28103113e-01
1.21852189e-01 -3.30783814e-01 1.00679529e+00 -1.04090595e+00
1.01224631e-01 -5.18952310e-01 -5.64250469e-01 -3.31126936e-02
-4.58769053e-01 8.14997017e-01 5.65243006e-01 6.13476336e-01
-3.74456882e-01 -1.10181674e-01 3.42545927e-01 -2.23036453e-01
-1.01850557e+00 -6.48653880e-02 -3.75076741e-01 6.34543538e-01
9.49451685e-01 -4.47337069e-02 -1.49228561e+00 -7.49490023e-01
-4.48154986e-01 4.35878813e-01 3.23918879e-01 6.32459760e-01
8.11529875e-01 -1.30795765e+00 -4.82270718e-01 -4.38617915e-01
6.50358140e-01 -1.21850356e-01 1.77318215e-01 1.59840420e-01
-1.24356735e+00 5.45074046e-01 -4.25329536e-01 -2.59328634e-01
-1.20462549e+00 1.08868051e+00 -1.63679570e-01 -4.65902723e-02
-6.15378082e-01 7.13919401e-01 1.48262158e-01 -6.57028332e-02
2.13900164e-01 -6.15494490e-01 -6.19670451e-01 4.03355122e-01
8.39108229e-01 2.93572217e-01 1.07109033e-01 -2.68877029e-01
-4.05533046e-01 5.24293005e-01 -9.20404866e-02 -3.17855626e-01
7.93961406e-01 1.60377800e-01 1.18309431e-01 7.08679318e-01
1.19586134e+00 3.37525427e-01 -8.09198141e-01 8.72576982e-02
4.27576542e-01 -8.05351377e-01 -4.20440882e-01 -9.54521298e-01
-5.30958593e-01 1.13978803e+00 1.97401389e-01 5.99120855e-01
9.12160456e-01 4.65941966e-01 4.36095744e-01 5.55149496e-01
-1.96228132e-01 -9.76849258e-01 4.94956225e-01 3.39613855e-01
1.14821756e+00 -1.34196961e+00 1.59812152e-01 -4.58736628e-01
-9.11332250e-01 1.49741447e+00 6.41414881e-01 2.38729298e-01
1.21223107e-01 1.96290523e-01 8.02619934e-01 -7.56171107e-01
-7.47657299e-01 -5.27902186e-01 9.00224447e-01 9.17968750e-01
5.07451475e-01 -8.87912363e-02 -5.07000834e-02 2.36056969e-01
-4.83069092e-01 -4.95312095e-01 5.76145470e-01 1.06152666e+00
-4.99776632e-01 -4.99970436e-01 -7.05695987e-01 6.21149302e-01
-4.28686708e-01 -1.78801149e-01 -9.11974609e-01 1.07391012e+00
-7.23381758e-01 9.21110570e-01 5.37904978e-01 1.78646535e-01
5.11582553e-01 3.03244710e-01 4.10311460e-01 -3.43889445e-01
-3.55618119e-01 -4.61537778e-01 2.40754187e-01 -5.45747221e-01
-2.10316509e-01 -3.41263652e-01 -1.50183010e+00 -2.38375872e-01
5.46359897e-01 1.68027580e-01 6.24623239e-01 4.90231782e-01
5.11460304e-02 6.66741073e-01 -1.38760969e-01 2.95717716e-01
-1.87572777e-01 -5.10765016e-01 -1.53853580e-01 8.18469048e-01
6.62893414e-01 -4.69048440e-01 -4.41903658e-02 2.09210753e-01]
|
[11.177488327026367, 2.051896810531616]
|
3d954392-92cd-46fe-9fcc-f075aea22c5f
|
is-one-annotation-enough-a-data-centric-image
|
2207.06214
| null |
https://arxiv.org/abs/2207.06214v3
|
https://arxiv.org/pdf/2207.06214v3.pdf
|
Is one annotation enough? A data-centric image classification benchmark for noisy and ambiguous label estimation
|
High-quality data is necessary for modern machine learning. However, the acquisition of such data is difficult due to noisy and ambiguous annotations of humans. The aggregation of such annotations to determine the label of an image leads to a lower data quality. We propose a data-centric image classification benchmark with ten real-world datasets and multiple annotations per image to allow researchers to investigate and quantify the impact of such data quality issues. With the benchmark we can study the impact of annotation costs and (semi-)supervised methods on the data quality for image classification by applying a novel methodology to a range of different algorithms and diverse datasets. Our benchmark uses a two-phase approach via a data label improvement method in the first phase and a fixed evaluation model in the second phase. Thereby, we give a measure for the relation between the input labeling effort and the performance of (semi-)supervised algorithms to enable a deeper insight into how labels should be created for effective model training. Across thousands of experiments, we show that one annotation is not enough and that the inclusion of multiple annotations allows for a better approximation of the real underlying class distribution. We identify that hard labels can not capture the ambiguity of the data and this might lead to the common issue of overconfident models. Based on the presented datasets, benchmarked methods, and analysis, we create multiple research opportunities for the future directed at the improvement of label noise estimation approaches, data annotation schemes, realistic (semi-)supervised learning, or more reliable image collection.
|
['Reinhard Koch', 'Nina Volkmann', 'Anna Valros', 'Jenny Stracke', 'Matti Pastell', 'Mariusz Oszust', 'Rainer Kiko', 'Sabine Dippel', 'Claudius Zelenka', 'Vasco Grossmann', 'Lars Schmarje']
|
2022-07-13
| null | null | null | null |
['noise-estimation']
|
['medical']
|
[ 4.79919940e-01 5.81288226e-02 -3.61441486e-02 -6.53697610e-01
-8.46423686e-01 -5.72537482e-01 5.80465734e-01 6.48989320e-01
-6.51191056e-01 5.48089147e-01 -1.35741979e-01 -3.23169008e-02
-3.00644547e-01 -5.72593510e-01 -6.37720466e-01 -7.78280973e-01
3.15590501e-01 6.67662740e-01 2.82142073e-01 3.07013333e-01
2.47973606e-01 2.69017488e-01 -1.89441001e+00 2.16932133e-01
6.23902380e-01 1.13132846e+00 3.31186533e-01 3.80114883e-01
-6.90416470e-02 6.14403009e-01 -7.66463101e-01 -4.37252462e-01
4.55657303e-01 -2.72824615e-01 -8.69980991e-01 5.23827553e-01
5.21716714e-01 3.62359919e-02 5.69033563e-01 1.12628961e+00
4.70426649e-01 -3.03703755e-01 6.16488755e-01 -1.32923198e+00
-1.11326911e-01 5.66697657e-01 -3.62560809e-01 -2.29968969e-02
7.03425631e-02 3.56206566e-01 8.59147251e-01 -4.14651334e-01
7.82787561e-01 9.77505922e-01 5.26553035e-01 2.24051669e-01
-1.41820586e+00 -4.79289263e-01 -1.61873728e-01 1.19341567e-01
-1.35231483e+00 -2.62508482e-01 6.04902327e-01 -7.76419520e-01
4.13912982e-02 2.70425141e-01 4.07362491e-01 1.09867346e+00
-3.55328441e-01 1.76816449e-01 1.60853994e+00 -7.20169485e-01
5.43704510e-01 7.32188463e-01 5.11541247e-01 4.16258246e-01
5.00883996e-01 1.20165542e-01 -3.06207746e-01 -5.41774370e-02
2.82257318e-01 -2.44716063e-01 -1.85847551e-01 -5.54682016e-01
-1.00858474e+00 6.44237518e-01 2.58104265e-01 5.81070840e-01
-1.97686151e-01 -1.89544037e-01 5.24023712e-01 3.46350372e-01
3.08402300e-01 8.46237600e-01 -5.19331396e-01 -8.46735686e-02
-1.07589114e+00 -1.38363140e-02 7.59472549e-01 6.06449842e-01
9.11877692e-01 -4.41146642e-01 -2.98763692e-01 6.83619201e-01
1.09547012e-01 1.44093916e-01 4.27567929e-01 -1.13263452e+00
1.81250736e-01 9.14160550e-01 2.29582682e-01 -9.36455548e-01
-3.74611199e-01 -7.04463422e-01 -6.16895616e-01 3.75224590e-01
9.79266465e-01 8.77012983e-02 -7.60558248e-01 1.56445599e+00
2.70599097e-01 -1.82472184e-01 -2.54925400e-01 8.28870475e-01
4.93468732e-01 1.93015516e-01 1.68809071e-01 -3.14122677e-01
1.43646336e+00 -6.96762919e-01 -6.19107604e-01 6.96384609e-02
9.74987090e-01 -7.54188776e-01 1.48233378e+00 7.06710517e-01
-6.00697100e-01 -7.08137870e-01 -1.03971994e+00 2.03483611e-01
-4.55183864e-01 3.56132060e-01 1.67776316e-01 9.18706834e-01
-7.73738503e-01 6.85672760e-01 -4.84118879e-01 -4.60872412e-01
3.63590479e-01 1.92959994e-01 -3.77267331e-01 -2.00756505e-01
-7.52952754e-01 8.97088945e-01 5.18562853e-01 -5.18938480e-03
-7.37797618e-01 -5.59050381e-01 -4.20861453e-01 -4.75268066e-02
5.58625698e-01 -2.90216744e-01 9.05513644e-01 -1.32878494e+00
-8.91380072e-01 1.18987119e+00 2.66426265e-01 -4.20716435e-01
7.18766093e-01 6.09800443e-02 -1.16349287e-01 -1.26593739e-01
2.10613366e-02 6.44331098e-01 6.91228211e-01 -1.74239445e+00
-4.81034040e-01 -5.44592142e-01 -7.16003776e-02 -1.73962727e-01
-3.38463902e-01 1.47444429e-02 -2.32809857e-01 -4.19737160e-01
-1.45144880e-01 -1.00702333e+00 -1.42432511e-01 -1.69386938e-01
-2.06705272e-01 -9.93182138e-03 5.02766669e-01 -3.49333316e-01
1.13735271e+00 -2.26592207e+00 -5.49024493e-02 2.24225134e-01
9.48814079e-02 2.99541742e-01 -1.62525684e-01 1.20025963e-01
-9.30143222e-02 3.92587095e-01 -1.83299065e-01 -5.60440898e-01
-9.06229839e-02 3.57939512e-01 1.03055157e-01 3.17664027e-01
1.58440292e-01 4.23121154e-01 -7.10874557e-01 -5.46003401e-01
2.11708844e-01 3.14575255e-01 -2.40355402e-01 2.76841730e-01
-2.45883986e-01 5.36866069e-01 -1.14065461e-01 3.87077570e-01
5.12407184e-01 -3.50254685e-01 1.76913664e-01 -4.46000814e-01
4.60821576e-02 1.62274595e-02 -1.57134044e+00 1.25815117e+00
-3.83137822e-01 4.95824307e-01 -1.70871019e-01 -9.38434720e-01
9.24325109e-01 2.20509589e-01 4.21610534e-01 -8.14142823e-01
3.60341012e-01 3.64210129e-01 2.74915472e-02 -4.76858944e-01
3.23451787e-01 1.44579813e-01 1.40631542e-01 4.64756787e-01
2.45668650e-01 -1.60374120e-01 4.53857720e-01 -1.37268513e-01
9.87557530e-01 -1.40731126e-01 1.39976710e-01 -4.22607332e-01
3.30941975e-01 1.09536938e-01 4.92517143e-01 8.35675359e-01
-2.21060336e-01 6.51276469e-01 6.13591731e-01 -5.85907757e-01
-1.22589481e+00 -4.25135225e-01 -2.96216577e-01 8.72392654e-01
4.92263294e-04 -3.56902063e-01 -1.03101003e+00 -8.14184189e-01
-3.17421824e-01 5.05729198e-01 -6.73986077e-01 -5.02749681e-02
-3.13687362e-02 -9.72313702e-01 4.40286398e-01 1.03236839e-01
3.32181126e-01 -8.53499651e-01 -8.25050592e-01 -5.99410199e-02
-1.75727203e-01 -1.30962598e+00 -5.45972995e-02 4.56218988e-01
-7.64221668e-01 -1.34861124e+00 -3.04165274e-01 -4.01068985e-01
8.26046944e-01 -7.85386339e-02 1.31478214e+00 4.96602952e-01
-1.33652374e-01 2.10222155e-01 -6.85914218e-01 -4.15510833e-01
-8.66841018e-01 1.55995071e-01 -1.82735994e-01 1.67779356e-01
3.47972751e-01 -2.30638996e-01 -3.92626822e-01 7.43933260e-01
-1.20921385e+00 -1.95688352e-01 6.35699093e-01 7.20452726e-01
6.20715141e-01 3.57952207e-01 3.35517734e-01 -1.07562947e+00
4.67717052e-01 -2.81060070e-01 -8.26821804e-01 4.23505872e-01
-1.07591903e+00 3.87476057e-01 4.04095471e-01 -5.38239956e-01
-7.92572260e-01 1.79241210e-01 1.86482202e-02 -3.93681563e-02
-4.91033792e-01 2.83941239e-01 -1.93341821e-01 -1.38836116e-01
1.02977395e+00 -2.23400980e-01 -1.03318371e-01 -5.24946511e-01
1.16664357e-01 6.81526065e-01 2.37382680e-01 -5.89312017e-01
5.76748252e-01 3.51147681e-01 3.06930635e-02 -5.66360593e-01
-1.01026607e+00 -5.78642607e-01 -7.03189373e-01 -2.24784732e-01
6.65589511e-01 -6.23069942e-01 -3.20567101e-01 4.68267173e-01
-1.01757491e+00 -3.52511168e-01 -5.12470603e-01 4.21424121e-01
-3.85001391e-01 3.91241461e-01 -2.24529743e-01 -6.68400824e-01
4.70488966e-02 -1.53233230e+00 1.02723157e+00 9.84145552e-02
-2.49723926e-01 -6.99715793e-01 -2.65399106e-02 8.40376794e-01
3.10222417e-01 1.58299252e-01 8.14922988e-01 -8.29485178e-01
-5.47346175e-01 -1.00845456e-01 -3.42480332e-01 7.61007011e-01
-7.35587627e-02 1.76998630e-01 -1.12837577e+00 -1.89538330e-01
1.42114282e-01 -5.56074560e-01 6.26847327e-01 3.99307534e-02
1.13409328e+00 -1.08414873e-01 6.13952987e-02 1.11175157e-01
1.63779449e+00 -1.02086015e-01 6.99334681e-01 4.93571818e-01
4.65894073e-01 9.17139530e-01 8.60479474e-01 2.63182342e-01
1.37687385e-01 9.12886381e-01 6.27179205e-01 -1.87710732e-01
-9.90445688e-02 1.44984111e-01 5.06644845e-02 3.93357217e-01
-8.04547220e-02 -2.08507314e-01 -1.01291406e+00 4.78184074e-01
-1.58777058e+00 -5.55449605e-01 -3.42667639e-01 2.63369298e+00
9.13952231e-01 3.93927813e-01 2.47240528e-01 6.68694019e-01
7.31910646e-01 -3.64292353e-01 -4.55896296e-02 -2.20659137e-01
-4.91847564e-03 -8.79681408e-02 5.93131602e-01 2.35195443e-01
-8.77399385e-01 3.40897530e-01 6.11103487e+00 8.20741951e-01
-1.05299950e+00 1.94887206e-01 8.76200616e-01 2.41222307e-01
-1.91471688e-02 6.97351098e-02 -7.79795945e-01 6.75631344e-01
9.08713818e-01 3.10007811e-01 1.91522509e-01 7.93017566e-01
2.43056118e-01 -5.08209467e-01 -1.20577550e+00 9.49280620e-01
1.39727024e-02 -1.03947079e+00 -1.21308133e-01 3.11283708e-01
7.63224006e-01 -9.22324061e-02 -2.27767542e-01 -1.24856345e-01
8.83061141e-02 -9.39475179e-01 8.17644775e-01 4.98940974e-01
4.41475332e-01 -4.81973708e-01 1.26354945e+00 5.50188780e-01
-6.40782058e-01 -2.63131738e-01 -1.46867454e-01 -5.28238192e-02
-2.34476030e-01 1.04395509e+00 -7.63040721e-01 4.39047396e-01
7.01161146e-01 5.68885282e-02 -1.08638966e+00 1.11740720e+00
-4.65740263e-02 6.49860024e-01 -3.06742877e-01 6.64277971e-02
-1.56780165e-02 -2.59297248e-02 1.59138888e-01 1.04342973e+00
7.32034817e-02 -3.45740885e-01 1.63867220e-01 8.03963661e-01
3.81761268e-02 2.40704864e-01 -3.38224560e-01 7.09734559e-02
4.60507184e-01 1.39352703e+00 -1.09142959e+00 -1.47529408e-01
-2.19923303e-01 5.08130252e-01 2.46830717e-01 2.26219688e-02
-5.74699879e-01 2.39246339e-01 9.30022635e-03 5.07939100e-01
-1.13825098e-01 -4.67556305e-02 -5.11918664e-01 -7.18019485e-01
1.61030024e-01 -1.10347414e+00 3.06622833e-01 -7.03535855e-01
-1.18494606e+00 5.63558519e-01 -9.02845350e-04 -1.05786180e+00
-8.32217261e-02 -5.50997436e-01 -1.08450785e-01 5.00852883e-01
-1.39396822e+00 -8.23465705e-01 -7.38384068e-01 1.60501033e-01
1.73150674e-01 1.64004147e-01 7.76325583e-01 5.63493967e-01
-3.93092901e-01 3.87352556e-01 -1.17794000e-01 4.67200279e-02
9.12173867e-01 -1.20142365e+00 -2.55853832e-01 6.66244805e-01
4.37491804e-01 1.08453713e-01 8.15405428e-01 -3.41049045e-01
-6.82644665e-01 -9.76872742e-01 6.59188688e-01 -7.39443600e-01
3.88636112e-01 -2.67694890e-01 -9.58097756e-01 2.08354101e-01
2.59452015e-02 2.02296674e-02 6.69269204e-01 8.55784118e-02
-2.61884332e-01 -3.04187804e-01 -1.23445296e+00 -2.64974907e-02
7.44627774e-01 -2.95475811e-01 -3.13492030e-01 1.82434544e-01
4.68649656e-01 -2.23215297e-02 -9.85194921e-01 4.99844790e-01
3.28215450e-01 -1.19895959e+00 4.91387993e-01 -1.46725982e-01
3.13757211e-01 -3.39869320e-01 -1.15168430e-01 -1.14534581e+00
4.35735434e-02 -5.31179532e-02 5.77202022e-01 1.50145888e+00
5.25297225e-01 -2.61431724e-01 6.41827404e-01 6.87943876e-01
3.72707456e-01 -5.90413690e-01 -7.09492624e-01 -7.60163665e-01
-2.57924378e-01 -5.00521600e-01 4.75297481e-01 9.87239003e-01
-5.47229171e-01 1.55403033e-01 -1.36977971e-01 7.23339841e-02
6.49255633e-01 -1.74131706e-01 8.69570673e-01 -1.56309175e+00
-1.73414022e-01 -2.55625665e-01 -6.90649092e-01 -3.09239298e-01
-9.30650905e-02 -4.33862031e-01 3.85673605e-02 -1.05604589e+00
3.51906806e-01 -9.04007077e-01 -1.48126006e-01 3.47036332e-01
-1.39250547e-01 4.38302487e-01 1.75748706e-01 2.31098101e-01
-7.55923867e-01 3.78816482e-03 1.00790799e+00 1.33560819e-03
3.86044197e-02 1.63948927e-02 -4.91189718e-01 7.17863262e-01
6.63386583e-01 -8.58288467e-01 -4.03720975e-01 -2.98079222e-01
5.50141752e-01 -4.81654584e-01 5.87702572e-01 -1.22164428e+00
1.40485689e-01 1.15269311e-01 6.68422505e-02 -1.46390602e-01
-3.43243256e-02 -1.27477849e+00 3.34679753e-01 3.20338190e-01
-4.25770134e-01 -1.61051750e-01 -1.95352629e-01 3.69482279e-01
-2.99068332e-01 -7.10032403e-01 9.69147563e-01 -2.83939391e-01
-5.40387571e-01 -1.02277525e-01 1.52942508e-01 1.12428993e-01
1.04095685e+00 -2.28609756e-01 -1.84960991e-01 -7.93817267e-02
-7.80458689e-01 8.54612663e-02 8.30053151e-01 2.97977537e-01
-6.54564500e-02 -9.22004342e-01 -6.62560105e-01 1.22933229e-02
3.87182117e-01 -3.52855865e-03 -8.75830129e-02 6.68903351e-01
-4.36044246e-01 3.32197137e-02 -2.27433309e-01 -9.06003833e-01
-1.44433308e+00 4.98003215e-01 3.68519574e-01 -5.85537612e-01
-4.24245633e-02 3.96372050e-01 -2.73940474e-01 -4.43953753e-01
5.78355730e-01 -3.71906608e-01 -2.05585018e-01 3.64319414e-01
4.41864818e-01 4.12494183e-01 5.07912755e-01 -5.10492921e-01
-4.80648838e-02 5.14452100e-01 2.07941577e-01 4.32786569e-02
1.17619169e+00 -2.24627987e-01 -2.79718488e-02 6.25236809e-01
8.69655609e-01 -1.56301215e-01 -1.11106467e+00 -1.78418517e-01
3.91317606e-01 -4.79137242e-01 5.26549555e-02 -9.32083607e-01
-1.00800371e+00 7.41872549e-01 9.37407553e-01 5.06724417e-01
1.28828311e+00 2.55154707e-02 3.47153582e-02 2.32995391e-01
5.20626724e-01 -1.32065141e+00 1.78714707e-01 -6.17033355e-02
6.14981472e-01 -1.63063562e+00 -2.39801724e-04 -6.08103096e-01
-6.26610935e-01 8.40881765e-01 4.35926586e-01 2.64057100e-01
4.74492162e-01 3.72317433e-01 3.31256449e-01 -2.80374527e-01
-5.40960968e-01 -2.51415700e-01 2.31708497e-01 4.40726787e-01
2.13667616e-01 -5.42087713e-03 -4.39665824e-01 5.48287928e-01
5.04912585e-02 3.08629543e-01 5.24450839e-01 5.77383101e-01
-3.25568229e-01 -1.49504972e+00 -5.38581550e-01 4.25302267e-01
-4.28065002e-01 3.14829648e-01 -4.54963446e-01 8.29858422e-01
7.74194479e-01 1.04743171e+00 -8.41165036e-02 -2.47872278e-01
4.13698196e-01 1.34706795e-01 4.69384164e-01 -7.31442153e-01
-6.21158600e-01 -2.41840541e-01 1.22116901e-01 -4.68660772e-01
-9.82077897e-01 -4.83963430e-01 -7.84672201e-01 5.97054549e-02
-7.12119937e-01 2.28310794e-01 1.01475716e+00 9.95798051e-01
3.73409450e-01 3.55533689e-01 5.44491589e-01 -5.24066567e-01
-6.00239515e-01 -9.99604583e-01 -4.74156976e-01 9.23730791e-01
-6.76791891e-02 -7.82003462e-01 -6.32893562e-01 3.01428974e-01]
|
[9.47148323059082, 3.852074384689331]
|
063b2911-f211-49df-8fb0-b0ce099283ab
|
3dcfs-fast-and-robust-joint-3d-semantic
|
2003.00535
| null |
https://arxiv.org/abs/2003.00535v1
|
https://arxiv.org/pdf/2003.00535v1.pdf
|
3DCFS: Fast and Robust Joint 3D Semantic-Instance Segmentation via Coupled Feature Selection
|
We propose a novel fast and robust 3D point clouds segmentation framework via coupled feature selection, named 3DCFS, that jointly performs semantic and instance segmentation. Inspired by the human scene perception process, we design a novel coupled feature selection module, named CFSM, that adaptively selects and fuses the reciprocal semantic and instance features from two tasks in a coupled manner. To further boost the performance of the instance segmentation task in our 3DCFS, we investigate a loss function that helps the model learn to balance the magnitudes of the output embedding dimensions during training, which makes calculating the Euclidean distance more reliable and enhances the generalizability of the model. Extensive experiments demonstrate that our 3DCFS outperforms state-of-the-art methods on benchmark datasets in terms of accuracy, speed and computational cost.
|
['Jianfeng Feng', 'xiangyang xue', 'Lili Chen', 'Liang Du', 'Hongkai Wen', 'Jiamao Li', 'Xiaolin Zhang', 'Jingang Tan']
|
2020-03-01
| null | null | null | null |
['3d-semantic-instance-segmentation']
|
['computer-vision']
|
[-1.09080382e-01 -2.45628938e-01 3.29089873e-02 -5.36072791e-01
-5.12917757e-01 -4.14227992e-01 5.54949880e-01 1.18907422e-01
-4.08758044e-01 1.32305939e-02 -2.52364427e-01 -1.31864205e-01
-2.99862504e-01 -9.08421457e-01 -5.71471512e-01 -5.88502407e-01
2.15890184e-01 4.06930506e-01 6.33485019e-01 4.08862904e-02
5.11736810e-01 6.25593603e-01 -1.60121500e+00 -1.78941056e-01
1.37911713e+00 1.21379542e+00 4.42712307e-01 4.23099399e-02
-3.06797594e-01 3.79782096e-02 -2.45172590e-01 -2.77696609e-01
5.31633258e-01 5.98299503e-02 -7.40335524e-01 2.57463306e-01
3.49073917e-01 -1.50579244e-01 -1.47095978e-01 1.10551822e+00
3.90808403e-01 2.90550798e-01 6.60696447e-01 -1.21407354e+00
-5.36817551e-01 1.45398766e-01 -5.86840570e-01 2.36766301e-02
1.74249172e-01 2.00874239e-01 1.05847168e+00 -1.35916519e+00
3.52468759e-01 1.30978429e+00 5.65765738e-01 1.24448106e-01
-1.22354424e+00 -7.75282741e-01 6.56356514e-01 -6.70726448e-02
-1.58597732e+00 6.38669415e-04 9.90565181e-01 -4.17261630e-01
8.08485329e-01 1.67821482e-01 7.23111451e-01 5.90637565e-01
-2.49143228e-01 1.01641273e+00 8.37125361e-01 -8.50291997e-02
2.69700587e-01 8.97992849e-02 1.58742413e-01 6.83817148e-01
1.62453294e-01 1.18486267e-02 -3.74580830e-01 -1.69892132e-01
9.40507114e-01 8.66866112e-02 -1.71017841e-01 -8.64672184e-01
-1.27324235e+00 8.27028453e-01 7.24836886e-01 1.28010511e-01
-2.05094129e-01 -9.50244889e-02 2.21095294e-01 8.94097164e-02
7.06050217e-01 5.42840421e-01 -5.57540715e-01 1.99977964e-01
-7.34171867e-01 3.93098325e-01 3.62829715e-01 1.03696465e+00
9.55300570e-01 -4.59477514e-01 -2.86208063e-01 1.00706148e+00
4.55507964e-01 4.76194143e-01 3.33482288e-02 -8.30247819e-01
4.64296967e-01 1.16670799e+00 -5.01498282e-02 -1.18025589e+00
-3.58507276e-01 -6.44442260e-01 -4.56242263e-01 4.01640795e-02
-1.20041743e-01 3.28233540e-01 -9.87204671e-01 1.50476384e+00
9.05935049e-01 5.36367536e-01 -2.22254604e-01 1.19934678e+00
8.29419971e-01 4.52800632e-01 5.77203296e-02 2.78697550e-01
9.29448366e-01 -1.11217988e+00 -1.15372062e-01 -2.88332045e-01
4.67659861e-01 -4.79066730e-01 1.26603699e+00 7.83354044e-02
-8.39160442e-01 -8.05468261e-01 -1.08442271e+00 -2.56993175e-01
-2.28278324e-01 1.08361058e-01 7.16727614e-01 2.49812871e-01
-5.52898169e-01 6.53325319e-01 -9.67691302e-01 -8.76095518e-02
6.67989552e-01 4.64035094e-01 -1.28405720e-01 2.85178646e-02
-8.43113422e-01 4.33342546e-01 4.64696497e-01 -4.04448286e-02
-5.10929108e-01 -8.47168446e-01 -9.57142949e-01 -8.55957270e-02
4.66015071e-01 -8.70107591e-01 8.59040856e-01 -3.63882422e-01
-1.45658064e+00 8.66411626e-01 3.83101404e-02 3.63289677e-02
5.53315520e-01 -4.09177005e-01 5.34551293e-02 2.54914433e-01
2.89896458e-01 9.65616465e-01 8.86406958e-01 -1.31065106e+00
-7.94596851e-01 -6.12410903e-01 -8.54955092e-02 4.90886271e-01
-3.28808427e-01 -3.98990929e-01 -9.96014118e-01 -5.91882408e-01
6.74779952e-01 -9.38030720e-01 -5.17056167e-01 2.66258091e-01
-4.82580125e-01 -6.37764215e-01 8.81507814e-01 -4.67840657e-02
1.08115494e+00 -2.40622854e+00 3.65086764e-01 5.25777042e-01
4.06546116e-01 6.01364067e-03 -1.36845216e-01 -2.28933953e-02
3.29994142e-01 2.01561213e-01 -5.76529860e-01 -5.07108569e-01
1.52528524e-01 1.32750854e-01 -1.80871189e-01 4.26561952e-01
4.96197969e-01 9.31387067e-01 -1.01841307e+00 -5.48512697e-01
4.77806658e-01 4.25198287e-01 -7.21544087e-01 4.82833296e-01
-2.02377960e-02 3.94787401e-01 -9.66479421e-01 6.11474931e-01
9.61152077e-01 -4.39823598e-01 -2.89543450e-01 -2.19300568e-01
-6.61176294e-02 3.25361878e-01 -1.15119672e+00 2.09077358e+00
-3.01621258e-01 -5.67391468e-03 -2.51965076e-01 -8.93872619e-01
1.15852261e+00 -2.16711968e-01 5.59479058e-01 -6.95498526e-01
4.95564379e-02 2.50733972e-01 -5.65244257e-01 -3.16683710e-01
4.38977569e-01 1.76654354e-01 -9.65565220e-02 -1.47147505e-02
3.68967056e-02 -5.44934869e-01 -8.91186818e-02 1.28212899e-01
7.16892242e-01 2.90403873e-01 6.67639961e-03 -4.26270723e-01
5.62668622e-01 -1.16125837e-01 6.79127991e-01 4.66782480e-01
-3.10886174e-01 7.11426735e-01 1.71119764e-01 -2.26681709e-01
-6.33679628e-01 -1.30923533e+00 -3.87299806e-01 8.74890089e-01
9.43554938e-01 -3.92912805e-01 -6.59329176e-01 -1.18835163e+00
5.18628478e-01 5.41662812e-01 -5.18437922e-01 -2.91222364e-01
-3.95450294e-01 -4.57428068e-01 6.33856934e-03 6.55684769e-01
5.31649709e-01 -6.26422524e-01 -4.60416585e-01 2.98601761e-02
7.87764266e-02 -1.29403651e+00 -5.27042866e-01 1.20103464e-01
-1.03264725e+00 -9.14714515e-01 -4.96284425e-01 -8.86125088e-01
7.84764588e-01 6.51158690e-01 8.82966220e-01 2.13974431e-01
1.70893129e-02 2.00559288e-01 -5.31737506e-01 -2.47553006e-01
2.25908369e-01 3.12250763e-01 -1.51344851e-01 -6.98470473e-02
4.28573608e-01 -5.28283000e-01 -8.42844248e-01 4.36449647e-01
-9.56442595e-01 1.63612530e-01 4.94915754e-01 5.30925751e-01
9.56721842e-01 -6.76790252e-02 3.26946139e-01 -8.25695872e-01
4.31918353e-01 -3.57013404e-01 -6.88183546e-01 7.52001405e-02
-8.14327300e-01 3.90159450e-02 3.67991835e-01 -2.23691449e-01
-6.73657835e-01 1.00786842e-01 -1.96321666e-01 -7.70866156e-01
-2.02608749e-01 2.90294707e-01 -2.51135379e-01 -5.24526358e-01
2.82329768e-01 1.94374725e-01 -1.81900114e-01 -6.48230076e-01
5.56660950e-01 6.63342357e-01 4.03870255e-01 -5.03809392e-01
1.05969465e+00 5.50817430e-01 -1.51780576e-01 -6.38759911e-01
-8.50262403e-01 -7.48537838e-01 -7.80184448e-01 3.03868745e-02
8.52118909e-01 -9.40450966e-01 -5.39502859e-01 5.14241219e-01
-1.03584146e+00 -4.04734835e-02 -1.75107881e-01 3.72950792e-01
-6.07119262e-01 3.25432569e-01 -2.96918631e-01 -5.02014279e-01
-2.08668262e-01 -1.43112814e+00 1.63238943e+00 2.58603245e-01
9.90941003e-02 -9.02405143e-01 -5.86189777e-02 2.42427662e-01
1.60748810e-01 2.57571787e-01 9.01493847e-01 -6.34749234e-01
-7.52888501e-01 -5.88824674e-02 -5.27389050e-01 4.72578377e-01
1.89964935e-01 8.20845738e-03 -8.63171875e-01 -3.25395525e-01
-9.36853588e-02 -1.56116337e-01 1.09927559e+00 1.93427339e-01
1.51635754e+00 1.67584866e-01 -5.42736769e-01 1.01464176e+00
1.36240673e+00 -2.92869098e-02 4.44027454e-01 3.60767066e-01
9.78231966e-01 3.92582238e-01 7.68018067e-01 3.88020396e-01
6.62692666e-01 5.53899229e-01 5.39639771e-01 -2.11752892e-01
2.07007214e-01 -3.46934706e-01 -2.32603684e-01 9.25180614e-01
1.95456401e-01 3.80050577e-02 -7.23868608e-01 5.54922640e-01
-1.86328375e+00 -3.49794865e-01 2.52036244e-01 2.08248258e+00
6.26336515e-01 3.48423809e-01 -5.10615297e-03 1.30243987e-01
6.04983449e-01 2.87697524e-01 -6.25052810e-01 -1.69809293e-02
1.72552645e-01 2.58481860e-01 3.44261110e-01 2.84701228e-01
-1.46048701e+00 1.15179670e+00 6.01951742e+00 9.78546500e-01
-1.15966976e+00 -1.38517469e-01 5.02660096e-01 9.00057629e-02
-5.48345149e-01 -2.32074056e-02 -6.92270756e-01 5.67709744e-01
2.14244053e-01 6.61702678e-02 2.01412052e-01 1.07357848e+00
-2.74328068e-02 1.89873025e-01 -1.00899720e+00 9.30146039e-01
-2.75587350e-01 -1.23582196e+00 3.52685124e-01 -2.05332469e-02
7.29697049e-01 1.35769755e-01 1.79509833e-01 2.18598187e-01
7.14217350e-02 -6.77477777e-01 8.68076146e-01 3.80428910e-01
4.66377676e-01 -8.37985992e-01 3.77891481e-01 2.74591982e-01
-1.46490407e+00 -1.68207422e-01 -2.91817844e-01 2.99313724e-01
3.34534422e-02 8.09619248e-01 -5.86707413e-01 6.88450754e-01
7.76817441e-01 9.63210762e-01 -6.40548527e-01 1.22427988e+00
-1.81075960e-01 3.09598237e-01 -4.01608855e-01 3.55720222e-02
4.22666878e-01 -4.41690952e-01 7.00495780e-01 1.02337492e+00
3.15549105e-01 7.08657056e-02 6.81602597e-01 1.23032892e+00
-4.54469882e-02 1.22599237e-01 -2.97271043e-01 4.05314751e-02
9.22951937e-01 1.25780737e+00 -8.54078770e-01 -1.24325864e-01
-2.89718568e-01 1.04194522e+00 4.75216955e-01 3.44689637e-01
-7.11380363e-01 -5.79862773e-01 9.21203971e-01 -5.96809909e-02
6.08529329e-01 -4.15619373e-01 -4.93280172e-01 -1.08309996e+00
2.26572871e-01 -3.55769187e-01 2.18810260e-01 -3.92696857e-01
-1.43728483e+00 6.16953731e-01 -5.39581664e-02 -1.31577051e+00
3.08984011e-01 -4.83935326e-01 -5.17094314e-01 5.55720687e-01
-1.81710684e+00 -1.22751474e+00 -4.87548262e-01 5.12212574e-01
5.38240612e-01 2.21715525e-01 2.96285391e-01 1.80985704e-01
-6.97197556e-01 5.37027895e-01 -2.67518103e-01 -9.52729210e-03
2.98498064e-01 -1.22892046e+00 7.72827625e-01 6.02416337e-01
3.50136086e-02 6.48421168e-01 2.09262639e-01 -4.76597995e-01
-1.19493127e+00 -1.40003550e+00 4.11591291e-01 -2.56676883e-01
2.77145535e-01 -4.44412142e-01 -9.86667871e-01 1.77115083e-01
-4.97234911e-01 2.33300269e-01 5.45619249e-01 7.88760372e-03
-4.43798721e-01 -1.65519267e-01 -1.22564697e+00 4.71826285e-01
1.45631087e+00 -5.55198491e-01 -6.01529658e-01 2.24058598e-01
1.33108985e+00 -5.09580314e-01 -8.35474193e-01 8.78434598e-01
2.38395259e-01 -9.16790843e-01 1.14066315e+00 -3.43824446e-01
1.43698439e-01 -5.28654993e-01 -3.11617106e-01 -1.18001342e+00
-6.43641651e-01 -2.55301982e-01 -9.21211466e-02 9.56934333e-01
3.67215186e-01 -6.45040572e-01 6.39644802e-01 4.38301772e-01
-3.69995266e-01 -1.24594057e+00 -9.98821914e-01 -7.69244134e-01
5.30952634e-03 -4.65435684e-01 1.03860807e+00 9.71211374e-01
-5.08355200e-01 1.19636126e-01 2.81331807e-01 4.00735229e-01
6.28724754e-01 5.63925803e-01 6.81413472e-01 -1.47295642e+00
6.40880689e-02 -6.14166141e-01 -6.84609592e-01 -1.40761089e+00
7.17858970e-02 -1.05858028e+00 3.27911042e-02 -1.53056896e+00
4.13212851e-02 -7.61591613e-01 -5.36379516e-01 3.48449081e-01
-5.79102099e-01 5.19416071e-02 2.96556711e-01 2.53917336e-01
-8.09574723e-01 1.04294193e+00 1.48937571e+00 2.05929168e-02
-4.79781419e-01 1.77157279e-02 -8.28107357e-01 6.67574465e-01
3.58947814e-01 -3.83979559e-01 -4.13180739e-01 -6.87329531e-01
-1.43046737e-01 -4.90871549e-01 3.35636646e-01 -1.08344078e+00
1.60482869e-01 -3.74190569e-01 4.08552438e-01 -6.68757379e-01
3.32707465e-01 -8.10759425e-01 -3.51596415e-01 1.88308023e-02
-1.57350719e-01 -1.93202078e-01 1.15650371e-02 6.46351099e-01
-1.37629911e-01 1.72804251e-01 7.67447591e-01 2.98745502e-02
-8.29020739e-01 7.43499756e-01 4.08853501e-01 -8.58367607e-02
1.06548202e+00 -4.69986260e-01 9.58092213e-02 2.85370678e-01
-3.51011872e-01 6.81954801e-01 6.84056044e-01 6.34613633e-01
9.09240901e-01 -1.44279277e+00 -3.52071017e-01 4.11548793e-01
3.56212288e-01 6.46785736e-01 -2.20340602e-02 8.09255719e-01
-4.30722892e-01 1.00969985e-01 1.55597478e-01 -1.04824865e+00
-8.10896218e-01 3.60368699e-01 2.42274448e-01 7.41818994e-02
-6.32090628e-01 1.08558631e+00 3.49764645e-01 -8.57033134e-01
2.66980886e-01 -5.50488293e-01 -8.55317488e-02 -7.39434958e-02
2.41586238e-01 3.29323590e-01 1.80821344e-01 -4.73878831e-01
-6.60137236e-01 9.27996337e-01 -1.06955454e-01 1.66573212e-01
1.33889449e+00 -1.96971029e-01 1.93575546e-02 2.44754583e-01
1.49779177e+00 -1.86699763e-01 -1.63859737e+00 -3.88013721e-01
-1.69126876e-02 -8.61993253e-01 3.35563749e-01 -4.16805953e-01
-1.23003745e+00 7.21962392e-01 6.21402800e-01 5.78051880e-02
1.23999584e+00 1.64859980e-01 1.07569516e+00 3.39424074e-01
3.95948738e-01 -1.06360769e+00 1.83916196e-01 4.80909824e-01
6.40158772e-01 -1.19921207e+00 -1.23043535e-02 -7.78632104e-01
-7.05404282e-01 8.81108999e-01 8.01510334e-01 -4.15704787e-01
8.16642523e-01 1.67088546e-02 2.61413772e-02 -3.46828490e-01
-3.66141796e-01 -3.71792287e-01 5.97772300e-01 3.63657027e-01
-2.12567523e-02 -3.43684927e-02 -2.11046189e-01 6.60121143e-01
-1.25792801e-01 -8.93835276e-02 -3.82123321e-01 9.30653334e-01
-6.68335319e-01 -8.57506812e-01 3.57069187e-02 3.57932121e-01
1.07356690e-01 2.37851202e-01 -4.18160677e-01 6.05752170e-01
3.61313313e-01 6.50746763e-01 1.88499853e-01 -8.13466966e-01
5.11772811e-01 -2.87137002e-01 2.46847644e-01 -6.45079255e-01
-4.10807967e-01 6.62994664e-03 -4.34600532e-01 -8.07271600e-01
-3.63784224e-01 -3.99265945e-01 -1.58883238e+00 8.58397260e-02
-6.96482837e-01 1.50559306e-01 6.34012997e-01 1.06055677e+00
7.08003759e-01 2.86093026e-01 1.21155775e+00 -1.10123336e+00
-5.74097037e-01 -5.74888885e-01 -4.27166283e-01 6.42414391e-01
2.81822741e-01 -1.03614354e+00 -3.45408648e-01 -5.02265453e-01]
|
[7.959840774536133, -3.2862131595611572]
|
27936281-9a5b-4528-b618-c5bcade79f72
|
neural-fine-tuning-search-for-few-shot
|
2306.09295
| null |
https://arxiv.org/abs/2306.09295v1
|
https://arxiv.org/pdf/2306.09295v1.pdf
|
Neural Fine-Tuning Search for Few-Shot Learning
|
In few-shot recognition, a classifier that has been trained on one set of classes is required to rapidly adapt and generalize to a disjoint, novel set of classes. To that end, recent studies have shown the efficacy of fine-tuning with carefully crafted adaptation architectures. However this raises the question of: How can one design the optimal adaptation strategy? In this paper, we study this question through the lens of neural architecture search (NAS). Given a pre-trained neural network, our algorithm discovers the optimal arrangement of adapters, which layers to keep frozen and which to fine-tune. We demonstrate the generality of our NAS method by applying it to both residual networks and vision transformers and report state-of-the-art performance on Meta-Dataset and Meta-Album.
|
['Timothy Hospedales', 'Da Li', 'Łukasz Dudziak', 'Panagiotis Eustratiadis']
|
2023-06-15
| null | null | null | null |
['architecture-search']
|
['methodology']
|
[ 3.37083191e-01 -1.19000509e-01 -7.34231025e-02 -4.49255615e-01
-2.84306556e-01 -6.27002776e-01 5.22803247e-01 -3.20148826e-01
-5.62632561e-01 4.40711170e-01 7.01277284e-03 -1.84888497e-01
-2.51958340e-01 -5.76592982e-01 -7.02170670e-01 -5.37012815e-01
1.12568446e-01 4.37267184e-01 4.09595460e-01 -2.77161986e-01
2.43657231e-01 5.90562582e-01 -1.77613366e+00 3.11917394e-01
5.08887827e-01 9.35520172e-01 7.56915957e-02 7.83609033e-01
5.38200662e-02 8.61461401e-01 -5.71152508e-01 -4.15766180e-01
3.88740122e-01 -5.52829504e-01 -8.80624354e-01 1.24946259e-01
6.18051589e-01 -1.84397846e-01 -9.76492390e-02 7.83051789e-01
4.96283144e-01 4.69602704e-01 6.68721437e-01 -8.74790668e-01
-8.06311846e-01 6.46224916e-01 -2.02275768e-01 6.01929545e-01
-2.98906922e-01 5.09803534e-01 9.50008392e-01 -9.59687054e-01
6.49923265e-01 7.68267155e-01 7.30777860e-01 9.66311455e-01
-1.47011900e+00 -5.59073627e-01 2.86716759e-01 2.09735185e-01
-1.35835564e+00 -9.04755771e-01 6.40637279e-01 -4.54694480e-01
1.24185765e+00 -6.56658262e-02 5.25416255e-01 1.22317326e+00
4.85083386e-02 3.24110806e-01 8.02701175e-01 -6.54306710e-01
5.58063269e-01 2.00293943e-01 1.92238420e-01 5.90060711e-01
2.30647072e-01 1.68118700e-01 -5.20055175e-01 8.68642703e-02
5.51287651e-01 -1.82765469e-01 -2.41757795e-01 -4.76165473e-01
-6.94009840e-01 6.57420397e-01 2.81072021e-01 4.47511375e-01
-2.46856004e-01 7.37858266e-02 3.28659624e-01 5.53108096e-01
1.73773929e-01 1.03138578e+00 -6.76056206e-01 -7.26156533e-02
-8.46602142e-01 -2.13096902e-01 9.07010078e-01 8.52989435e-01
7.65687346e-01 1.89871877e-01 -1.53411431e-02 1.07830524e+00
-1.59910709e-01 -1.22257531e-01 7.23813534e-01 -8.94850910e-01
8.66742283e-02 5.42386830e-01 -2.84181327e-01 -3.11593294e-01
-3.20661187e-01 -8.60849500e-01 -6.16722524e-01 5.06966352e-01
4.15620029e-01 -3.03846538e-01 -1.08886218e+00 1.79867756e+00
2.06784576e-01 3.09477478e-01 1.48648441e-01 7.25896060e-01
4.71113890e-01 2.80491233e-01 6.93742111e-02 -1.17765982e-02
9.24432814e-01 -1.04154134e+00 1.96911655e-02 -7.22072601e-01
4.66654807e-01 -4.78515446e-01 1.28652310e+00 3.13986361e-01
-9.74763691e-01 -7.04759419e-01 -1.38709295e+00 2.18139902e-01
-4.19789106e-01 -1.20982938e-01 3.80156040e-01 6.26057446e-01
-9.50039625e-01 8.27634931e-01 -6.57424331e-01 -7.12829173e-01
4.68933016e-01 5.15068710e-01 -2.56402314e-01 -1.28105476e-01
-8.52045178e-01 9.84002173e-01 4.76802289e-01 -1.14510342e-01
-1.08513081e+00 -6.66687787e-01 -2.20024869e-01 2.52446771e-01
5.08524835e-01 -9.99520838e-01 1.45302606e+00 -1.44365323e+00
-1.60335410e+00 8.28408360e-01 1.30988002e-01 -4.86578554e-01
8.32415670e-02 7.16915131e-02 -3.99476439e-01 -2.45551709e-02
-3.46477687e-01 4.53440219e-01 1.13987708e+00 -9.84645188e-01
-7.09677279e-01 -3.76325339e-01 2.25043491e-01 -3.61140370e-02
-6.29145145e-01 -3.94879617e-02 -4.17348444e-01 -4.38025236e-01
-9.94761661e-02 -1.00665557e+00 -1.64282620e-01 -1.62455514e-01
4.87152524e-02 -1.10742204e-01 3.44737262e-01 -1.35108277e-01
1.21398139e+00 -2.32901049e+00 2.96058327e-01 -7.19311312e-02
-3.46702188e-02 4.17048901e-01 -4.27294821e-01 1.74516484e-01
-2.57245332e-01 1.04852632e-01 -7.87318945e-02 -2.94049412e-01
-1.78705007e-01 2.43005440e-01 -2.98414081e-01 2.96933085e-01
3.22486341e-01 6.88476682e-01 -6.48787677e-01 -2.30482325e-01
-1.18062191e-01 2.71028668e-01 -7.32741952e-01 3.83021563e-01
-2.06664145e-01 1.02253906e-01 -3.04526657e-01 6.39946580e-01
1.89914584e-01 -4.51119900e-01 1.21279180e-01 -1.49871662e-01
-3.15050632e-02 -6.14888035e-03 -1.00538576e+00 1.60296595e+00
-4.57223922e-01 5.93509197e-01 -2.47307986e-01 -8.79294515e-01
9.78498638e-01 5.30117340e-02 8.44546184e-02 -4.94764537e-01
3.08142841e-01 1.01685941e-01 2.09619537e-01 -5.42302012e-01
2.40732804e-01 -3.85818005e-01 1.47111982e-01 5.26912987e-01
5.48813760e-01 2.53474265e-01 -2.56352127e-02 -2.79930770e-01
1.42647409e+00 -4.63003591e-02 3.80436242e-01 -1.21359974e-01
2.23586470e-01 1.42179266e-01 4.73942727e-01 1.06500220e+00
-2.80462533e-01 7.21385121e-01 1.61304384e-01 -7.18032062e-01
-1.16789484e+00 -8.91919136e-01 -8.15348029e-02 1.55412710e+00
-3.31823528e-01 -1.07362285e-01 -7.69550860e-01 -7.67999709e-01
-1.36562496e-01 8.57525945e-01 -8.06671917e-01 -5.65796316e-01
-5.74009180e-01 -6.58616662e-01 5.52536905e-01 5.75810611e-01
3.37664276e-01 -1.05480897e+00 -8.84863138e-01 2.59927601e-01
3.90851498e-01 -8.36083949e-01 -3.31109852e-01 6.37565553e-01
-9.69002306e-01 -1.08048844e+00 -4.25563782e-01 -7.51685917e-01
7.70225406e-01 2.08866194e-01 1.34855318e+00 2.32132882e-01
-1.88770175e-01 4.86215621e-01 -2.51077801e-01 -3.01038921e-01
-3.22060913e-01 6.58164561e-01 5.31605594e-02 8.46498832e-02
4.87580210e-01 -8.86968970e-01 -4.26565140e-01 3.48172754e-01
-9.20168877e-01 -1.71610504e-01 8.06883633e-01 9.90754426e-01
4.11555231e-01 6.54980168e-02 5.69285512e-01 -1.05055308e+00
5.98607004e-01 -4.75968599e-01 -6.15053713e-01 6.05515480e-01
-8.83364260e-01 3.09114516e-01 8.08525145e-01 -8.46601307e-01
-9.54317629e-01 2.11748496e-01 7.01533407e-02 -7.57824183e-01
-2.47816056e-01 3.93040568e-01 -1.79921910e-02 -2.03446120e-01
1.09981275e+00 1.04025707e-01 -1.61926836e-01 -5.12438416e-01
5.33889234e-01 5.78259885e-01 6.21253073e-01 -5.98413765e-01
6.89851463e-01 2.38650635e-01 -2.77311265e-01 -7.64601409e-01
-9.36200321e-01 -2.31654674e-01 -9.52948689e-01 -1.16292775e-01
6.45537436e-01 -7.41684258e-01 -3.10064703e-01 3.03993613e-01
-9.66152966e-01 -6.94049537e-01 -6.83468401e-01 1.10766925e-01
-5.30628383e-01 -1.97852686e-01 -1.97492167e-01 -5.22205591e-01
-2.76844859e-01 -9.65151608e-01 4.56684947e-01 4.46322232e-01
-3.69657576e-01 -7.96107769e-01 3.07410300e-01 1.02789648e-01
6.78268552e-01 -1.42960310e-01 1.06530857e+00 -8.65419388e-01
-6.82970762e-01 -1.32032499e-01 3.89603041e-02 4.94450718e-01
-6.17125779e-02 -5.55033311e-02 -1.14200389e+00 -4.68504816e-01
7.14378133e-02 -4.12577957e-01 1.01807630e+00 2.17747286e-01
1.08824611e+00 -2.83726960e-01 -3.07714790e-01 1.07328594e+00
1.49693871e+00 2.13262290e-01 6.15339696e-01 6.88985527e-01
3.56973410e-01 4.89288956e-01 1.95128117e-02 1.89301789e-01
6.30549565e-02 5.71213782e-01 1.68486193e-01 4.27838057e-01
-3.18105608e-01 -1.06910504e-01 3.94149087e-02 5.37626147e-01
-9.42951664e-02 -1.48547813e-01 -8.64635408e-01 5.84743202e-01
-1.72427762e+00 -9.04960692e-01 8.90485287e-01 2.11614490e+00
7.65779674e-01 3.64603102e-01 1.71095386e-01 -1.38644502e-01
4.12264913e-01 1.26160264e-01 -8.70098412e-01 -4.09856200e-01
3.00218873e-02 3.80045861e-01 3.81457031e-01 2.06897005e-01
-8.47845018e-01 9.87602353e-01 6.96677256e+00 6.28045559e-01
-1.25813043e+00 1.13967173e-01 5.67442477e-01 -4.07371014e-01
-2.64680162e-02 2.64947057e-01 -1.07008755e+00 1.75489932e-01
1.21698189e+00 -3.86552624e-02 9.19031680e-01 9.85540807e-01
-3.44012290e-01 2.05230907e-01 -1.34062469e+00 6.21124923e-01
2.36150712e-01 -1.43446648e+00 -1.67134497e-02 -1.68309525e-01
6.89498305e-01 3.10205787e-01 4.24831808e-02 5.61727941e-01
3.97313595e-01 -1.11840463e+00 6.01715684e-01 6.82835698e-01
7.64734924e-01 -4.63522881e-01 3.53408068e-01 2.91605681e-01
-7.34074652e-01 -5.28606832e-01 -4.97906893e-01 -4.39927867e-03
-3.80197108e-01 1.11231335e-01 -9.70339596e-01 -1.10089198e-01
6.05173051e-01 4.39184189e-01 -8.78683805e-01 1.19331312e+00
-1.79588705e-01 6.95171595e-01 -1.21765137e-01 -1.12566076e-01
1.37101382e-01 9.38642845e-02 3.61956239e-01 9.98593748e-01
2.95708895e-01 1.92405015e-01 -1.57749802e-01 7.78714418e-01
-7.48549551e-02 -2.43124515e-01 -5.58226109e-01 -2.05882177e-01
5.97895563e-01 1.16521704e+00 -6.72284126e-01 -1.93075433e-01
-4.06163275e-01 7.71725774e-01 8.19666862e-01 5.04708827e-01
-3.79656971e-01 -3.02870870e-01 7.02212930e-01 1.67098463e-01
6.58118606e-01 -1.06736675e-01 -2.45781377e-01 -1.14022040e+00
-1.88418582e-01 -1.02827775e+00 6.24741733e-01 -7.38106966e-01
-1.41188991e+00 8.20317805e-01 -1.76300347e-01 -9.87574518e-01
-3.41928810e-01 -6.65528595e-01 -7.87752211e-01 5.92103839e-01
-1.30949771e+00 -9.66544271e-01 -2.70866513e-01 4.99039173e-01
7.97374070e-01 -5.86806595e-01 9.09619153e-01 9.05260444e-02
-7.89568484e-01 7.97581077e-01 6.40463233e-02 7.92432427e-02
6.32083416e-01 -1.02968550e+00 3.29149038e-01 8.53540182e-01
4.23247337e-01 6.56311691e-01 7.72087812e-01 -2.30923697e-01
-1.42900717e+00 -9.05744076e-01 5.34293890e-01 -5.92670321e-01
6.19961023e-01 -1.10636562e-01 -1.19188583e+00 8.86198997e-01
1.03259675e-01 1.91314772e-01 7.32517779e-01 4.66030240e-01
-6.74024105e-01 -4.09275234e-01 -8.41565549e-01 6.27793968e-01
1.17640674e+00 -5.78503072e-01 -7.94070840e-01 8.39424282e-02
7.59664774e-01 -2.59582579e-01 -7.13459313e-01 1.94934532e-01
7.87396669e-01 -1.05963874e+00 9.32672620e-01 -1.10263860e+00
3.47038358e-01 -1.07174486e-01 -2.98123717e-01 -1.53454840e+00
-6.71324313e-01 -4.14681762e-01 -6.39411882e-02 1.19593656e+00
6.55548275e-01 -5.57231903e-01 7.68152773e-01 7.05158293e-01
-1.72924131e-01 -9.60475385e-01 -8.15329432e-01 -8.30248058e-01
9.68893990e-02 -2.93304354e-01 7.61042237e-01 8.41454089e-01
-2.54700363e-01 7.32543230e-01 -2.52263933e-01 8.44759271e-02
3.96175176e-01 5.86652644e-02 8.79962444e-01 -1.24860740e+00
-6.76794231e-01 -5.44749856e-01 -2.85625756e-01 -7.02035546e-01
1.50310755e-01 -6.95201993e-01 6.54152557e-02 -8.82124066e-01
2.12637082e-01 -4.87644166e-01 -6.45398796e-01 6.69773102e-01
3.46297724e-03 3.44634913e-02 2.28170857e-01 3.93188924e-01
-6.90648735e-01 2.70540476e-01 7.81340420e-01 -1.33304402e-01
-3.28840017e-01 5.75055145e-02 -1.01419365e+00 5.48054755e-01
7.58123994e-01 -5.98812819e-01 -4.54770207e-01 -6.30009830e-01
3.36087018e-01 -4.76312667e-01 2.13440135e-01 -1.25026011e+00
5.13465226e-01 -3.18943173e-01 5.54777622e-01 -5.02414331e-02
3.39961261e-01 -8.15975904e-01 1.88350081e-01 2.55574822e-01
-5.35046041e-01 2.87231952e-02 3.45633715e-01 5.59854448e-01
9.23621505e-02 -6.83012664e-01 9.98222113e-01 -3.15473944e-01
-1.06908762e+00 2.68758684e-01 -2.20521703e-01 2.13752031e-01
9.22184348e-01 -3.18228543e-01 -4.65918392e-01 -6.00859383e-03
-7.21227229e-01 3.08993347e-02 6.76398873e-01 5.00007153e-01
5.02070010e-01 -1.04221189e+00 -3.94711405e-01 3.85166943e-01
3.09654504e-01 -2.08083183e-01 3.55974957e-02 4.59384412e-01
3.05643529e-02 1.38605520e-01 -3.25695723e-01 -4.10112739e-01
-1.07631707e+00 7.56408215e-01 6.84864759e-01 1.72783360e-02
-5.89303434e-01 1.02790821e+00 -1.00709409e-01 -3.11921865e-01
3.16381663e-01 -1.42376497e-01 -9.96289477e-02 1.98193565e-02
6.25132382e-01 1.63935959e-01 1.47562757e-01 -2.58956432e-01
-2.58214146e-01 4.81862515e-01 -2.38663971e-01 6.32502958e-02
1.55836499e+00 -3.05269044e-02 1.86643824e-01 5.98248363e-01
1.13158214e+00 -4.40147638e-01 -1.63760579e+00 -3.26243669e-01
1.45325750e-01 -3.35157961e-01 8.29491913e-02 -9.15954590e-01
-1.00506890e+00 6.82704091e-01 6.59152746e-01 1.37271926e-01
1.30753887e+00 5.00110313e-02 4.43236083e-01 6.81909382e-01
1.67389423e-01 -1.17383039e+00 3.42221648e-01 6.66640222e-01
6.11867666e-01 -9.93480980e-01 -2.18757138e-01 3.23525131e-01
-5.47988772e-01 1.18291247e+00 8.32607508e-01 -1.87206760e-01
6.73607290e-01 1.66525200e-01 -3.54992189e-02 -2.51673102e-01
-1.14303720e+00 -2.63564557e-01 3.30207437e-01 5.26346028e-01
1.59666404e-01 -1.67475790e-01 3.76239896e-01 5.60852230e-01
-1.97325900e-01 2.85775453e-01 4.73185271e-01 9.78324354e-01
-6.71904862e-01 -9.51043367e-01 -1.51738808e-01 5.31008124e-01
-9.62871909e-02 -1.15519715e-02 -4.80241805e-01 6.41480207e-01
2.76068091e-01 7.95457661e-01 6.13192990e-02 -5.95949948e-01
5.90528190e-01 5.49703538e-01 6.33978605e-01 -9.46324587e-01
-7.19869554e-01 -4.46471691e-01 5.25360331e-02 -3.44809443e-01
-2.23899677e-01 -6.17151022e-01 -6.68634236e-01 -3.98212224e-02
-3.72114867e-01 -6.69575557e-02 4.24430609e-01 1.22079730e+00
5.35462677e-01 6.53299451e-01 4.60608214e-01 -5.25542378e-01
-9.74059880e-01 -9.10473168e-01 -3.79613191e-01 8.90794992e-02
4.38931197e-01 -5.54225862e-01 -5.00080764e-01 5.35696521e-02]
|
[9.196602821350098, 3.05815052986145]
|
8a707b50-c30e-4e53-a92b-199fc43860ea
|
frame-wise-action-representations-for-long
|
2203.14957
| null |
https://arxiv.org/abs/2203.14957v1
|
https://arxiv.org/pdf/2203.14957v1.pdf
|
Frame-wise Action Representations for Long Videos via Sequence Contrastive Learning
|
Prior works on action representation learning mainly focus on designing various architectures to extract the global representations for short video clips. In contrast, many practical applications such as video alignment have strong demand for learning dense representations for long videos. In this paper, we introduce a novel contrastive action representation learning (CARL) framework to learn frame-wise action representations, especially for long videos, in a self-supervised manner. Concretely, we introduce a simple yet efficient video encoder that considers spatio-temporal context to extract frame-wise representations. Inspired by the recent progress of self-supervised learning, we present a novel sequence contrastive loss (SCL) applied on two correlated views obtained through a series of spatio-temporal data augmentations. SCL optimizes the embedding space by minimizing the KL-divergence between the sequence similarity of two augmented views and a prior Gaussian distribution of timestamp distance. Experiments on FineGym, PennAction and Pouring datasets show that our method outperforms previous state-of-the-art by a large margin for downstream fine-grained action classification. Surprisingly, although without training on paired videos, our approach also shows outstanding performance on video alignment and fine-grained frame retrieval tasks. Code and models are available at https://github.com/minghchen/CARL_code.
|
['Deng Cai', 'Chong Li', 'Fangyun Wei', 'Minghao Chen']
|
2022-03-28
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Chen_Frame-Wise_Action_Representations_for_Long_Videos_via_Sequence_Contrastive_Learning_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Chen_Frame-Wise_Action_Representations_for_Long_Videos_via_Sequence_Contrastive_Learning_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['video-alignment']
|
['computer-vision']
|
[ 3.10416758e-01 -3.81906986e-01 -5.19071877e-01 -4.30514187e-01
-1.07138145e+00 -3.02598357e-01 7.34387219e-01 -1.51775062e-01
-3.98306072e-01 6.09649777e-01 6.70730770e-01 1.90361336e-01
2.22190679e-03 -4.30166870e-01 -9.60225880e-01 -7.13155866e-01
-1.75819263e-01 8.06456357e-02 1.68550193e-01 -5.98375686e-02
3.19681376e-01 3.10341090e-01 -1.58671927e+00 6.51812434e-01
4.30479109e-01 1.14221907e+00 2.38695949e-01 6.96791410e-01
1.65012464e-01 1.32893372e+00 -3.08812708e-01 -2.66384363e-01
4.60410863e-01 -7.78196394e-01 -9.20770168e-01 3.35523963e-01
6.92910433e-01 -6.65383935e-01 -6.87748432e-01 8.06746244e-01
3.88048381e-01 4.23752099e-01 4.62493300e-01 -1.46940172e+00
-6.36009395e-01 3.00552160e-01 -5.98439991e-01 5.89199245e-01
4.73366141e-01 2.62349069e-01 1.20709932e+00 -9.14918959e-01
5.88683248e-01 1.10837662e+00 4.74782407e-01 4.65171188e-01
-8.60801637e-01 -5.19396305e-01 3.81970108e-01 7.35161185e-01
-1.25841713e+00 -5.44594586e-01 6.34994209e-01 -5.51016390e-01
1.17515516e+00 2.20014509e-02 5.73963225e-01 1.30246699e+00
1.64594650e-01 1.09792233e+00 6.48285747e-01 -2.23302588e-01
8.42612907e-02 -4.80764180e-01 -2.34260336e-01 7.85432100e-01
-2.59892344e-01 5.66981137e-02 -8.30902219e-01 3.23938467e-02
9.42694545e-01 3.97473842e-01 -2.95131356e-01 -5.67074656e-01
-1.49480975e+00 7.92694688e-01 1.95828214e-01 4.52715009e-01
-5.46717346e-01 5.32312870e-01 7.66220391e-01 3.26427162e-01
5.19332767e-01 8.90781805e-02 -3.73402089e-01 -6.71549916e-01
-9.40927684e-01 1.49278611e-01 3.20145816e-01 9.07360375e-01
6.53019965e-01 8.85044262e-02 -3.96581262e-01 6.83366954e-01
1.13876484e-01 2.02877030e-01 8.79494131e-01 -1.22095001e+00
5.53871572e-01 3.50397706e-01 -8.27464536e-02 -9.36055720e-01
1.99622903e-02 2.82342806e-02 -6.87377334e-01 -1.93344466e-02
2.75525153e-01 1.21863298e-01 -5.55472851e-01 1.71228552e+00
1.53709695e-01 7.56156564e-01 -3.84447835e-02 1.05724108e+00
4.67294633e-01 7.32149959e-01 1.29852876e-01 -2.25478634e-01
1.02632821e+00 -1.39240694e+00 -6.47412419e-01 -9.82912332e-02
7.43921340e-01 -5.33509851e-01 9.72080469e-01 1.26234606e-01
-1.25735521e+00 -7.56204665e-01 -9.93282080e-01 -2.47124925e-01
-1.02734193e-01 1.54557779e-01 4.64865297e-01 3.57180052e-02
-9.41304803e-01 8.04931700e-01 -1.15519190e+00 -3.64223808e-01
6.25960886e-01 5.29667772e-02 -6.15512848e-01 -1.72077164e-01
-1.03727114e+00 6.05656981e-01 3.84436160e-01 -1.35602981e-01
-9.84691560e-01 -5.26995182e-01 -1.04541957e+00 9.75209102e-03
4.72643852e-01 -5.50966799e-01 1.18845332e+00 -1.30406106e+00
-1.53173387e+00 8.12958896e-01 -1.50962323e-01 -8.16635549e-01
3.91240150e-01 -6.87675297e-01 -2.54313797e-01 6.14877403e-01
2.58417428e-01 7.22099662e-01 1.04415190e+00 -5.67256510e-01
-6.13333106e-01 -2.10223556e-01 3.17105800e-01 2.80262887e-01
-3.48101169e-01 6.88694268e-02 -5.12947381e-01 -1.01016891e+00
-2.86549866e-01 -8.81895185e-01 -1.72580913e-01 3.18843216e-01
1.73049122e-01 -2.47235477e-01 8.15684855e-01 -7.00639248e-01
1.21355939e+00 -2.34078884e+00 4.79849607e-01 -3.60118836e-01
-2.64023580e-02 3.74174356e-01 -4.45322335e-01 6.11440957e-01
-2.97554851e-01 -2.97090560e-01 -2.28577584e-01 -3.00673515e-01
-6.59725349e-03 2.33004585e-01 -3.88371199e-01 6.31487727e-01
3.31681222e-01 1.02313972e+00 -1.13269413e+00 -4.41473156e-01
4.70273703e-01 4.28535253e-01 -7.59142339e-01 5.07309318e-01
-1.58287376e-01 4.40918446e-01 -4.14447486e-01 6.27551615e-01
2.33611628e-01 -5.15467703e-01 8.10054839e-02 -2.80309379e-01
9.73451883e-02 1.10618807e-01 -9.54329848e-01 2.35966420e+00
-3.59116167e-01 7.47368336e-01 -3.36535066e-01 -1.45209968e+00
6.31949365e-01 2.78068721e-01 9.94167447e-01 -7.26963699e-01
4.40927548e-03 -7.59067526e-03 -3.94845903e-01 -6.38188779e-01
5.12913644e-01 -3.13062519e-02 -6.15832619e-02 4.96316284e-01
3.19714457e-01 3.78882885e-01 3.81491423e-01 2.82230973e-01
1.17537594e+00 7.73946702e-01 5.98140955e-01 2.11728066e-01
6.92613006e-01 -3.07822526e-01 6.80408537e-01 4.67351377e-01
-4.45638061e-01 8.28388691e-01 3.26418996e-01 -4.95902479e-01
-7.91868031e-01 -8.94651413e-01 2.97696829e-01 1.40975523e+00
1.41218394e-01 -8.68090332e-01 -5.82285404e-01 -8.78596008e-01
-9.91243050e-02 3.08719575e-01 -5.47848940e-01 -2.49712363e-01
-8.49563539e-01 -1.59531072e-01 3.19640547e-01 9.43120241e-01
6.62017703e-01 -1.19719219e+00 -6.64839506e-01 1.78432405e-01
-4.08934623e-01 -1.30999184e+00 -8.68538797e-01 -1.98995933e-01
-7.89202631e-01 -1.16338027e+00 -7.60746002e-01 -6.67407215e-01
3.33453298e-01 6.18946671e-01 1.08978474e+00 2.23550983e-02
-3.52612853e-01 8.26330781e-01 -8.06889296e-01 2.04961464e-01
-4.88865701e-03 -1.60980374e-01 5.12350388e-02 2.28650063e-01
5.36069214e-01 -5.98144889e-01 -7.50436246e-01 3.72732371e-01
-9.60808396e-01 -9.05232783e-03 5.65930426e-01 8.64545941e-01
8.13440144e-01 -3.63538116e-01 4.79433537e-01 -5.17776191e-01
9.90127102e-02 -6.94720149e-01 -3.99573743e-01 2.65524119e-01
-1.51031449e-01 7.83813298e-02 5.60893297e-01 -3.41156453e-01
-8.35446119e-01 1.00543700e-01 -3.57703306e-02 -1.01915622e+00
-1.90306157e-01 2.39288986e-01 -2.50599477e-02 2.06687093e-01
3.81736606e-01 3.48208100e-01 7.33721331e-02 -3.42281461e-01
4.95653331e-01 4.87214297e-01 5.41248620e-01 -5.09581983e-01
5.78533709e-01 5.95020473e-01 -1.57873541e-01 -5.98408818e-01
-8.98191035e-01 -7.26492703e-01 -8.44663203e-01 -1.98196426e-01
1.13434100e+00 -1.14272320e+00 -4.99515414e-01 4.94188517e-01
-8.11925769e-01 -4.64987397e-01 -5.35454750e-01 7.24210382e-01
-1.24614120e+00 6.86435461e-01 -5.40582538e-01 -3.26387227e-01
-1.25834540e-01 -1.02461445e+00 1.37076211e+00 -1.19047552e-01
-2.12981373e-01 -8.96828711e-01 2.64751852e-01 6.34301066e-01
1.71169087e-01 2.58283317e-01 3.19630444e-01 -4.42580849e-01
-8.16132128e-01 -4.38750498e-02 -3.51768062e-02 6.61762059e-01
2.67748863e-01 -2.09030375e-01 -5.97447336e-01 -5.06145239e-01
-1.61790013e-01 -6.90160215e-01 1.04512799e+00 3.98837388e-01
1.51875138e+00 -3.77628803e-01 -6.92567155e-02 8.06866288e-01
1.32023001e+00 1.16166351e-02 7.78234720e-01 4.97277826e-01
7.97277391e-01 2.63278335e-01 1.12539339e+00 8.16175461e-01
2.51078427e-01 9.03402269e-01 4.15512443e-01 2.50226527e-01
-2.03869715e-01 -3.77518982e-01 7.01251030e-01 7.08771646e-01
-3.75405997e-01 -9.99924839e-02 -3.89564306e-01 5.25302887e-01
-2.36203980e+00 -1.55024624e+00 4.90050346e-01 2.10254002e+00
5.94807506e-01 -1.34318694e-01 2.61070520e-01 -7.25359917e-02
5.06608605e-01 7.98934102e-01 -5.58382988e-01 -1.23867139e-01
-5.15350187e-03 1.30459607e-01 3.89554203e-01 2.25241303e-01
-1.46188140e+00 9.67440188e-01 5.30052042e+00 8.60322475e-01
-9.86132443e-01 2.03309402e-01 3.92448813e-01 -4.68099058e-01
6.13588504e-02 -1.72412589e-01 -5.29532552e-01 5.93312800e-01
9.91292357e-01 -1.22407794e-01 2.88610965e-01 8.71254623e-01
2.13956743e-01 7.90567398e-02 -1.25517607e+00 1.31741416e+00
4.47191000e-01 -1.68724132e+00 1.74934000e-01 -6.34480342e-02
8.02115500e-01 1.10066064e-01 -1.32971793e-01 3.89258265e-01
-1.31714167e-02 -8.64296615e-01 6.54745281e-01 5.58571994e-01
6.92691922e-01 -4.94743735e-01 5.13860226e-01 5.51458821e-02
-1.47619462e+00 -3.22173834e-01 -3.27877045e-01 -1.73088163e-01
4.05279040e-01 2.08766423e-02 -2.46849611e-01 5.51502168e-01
8.21093023e-01 1.65598249e+00 -3.98307085e-01 7.59238243e-01
-1.18649274e-01 3.96171987e-01 1.99417889e-01 4.88712072e-01
5.00868440e-01 -2.86541283e-01 3.56747955e-01 1.22457004e+00
2.82032639e-01 2.96989590e-01 3.06586772e-01 1.92527190e-01
-1.51104972e-01 -7.86690600e-03 -6.53072238e-01 -2.84799844e-01
1.53918803e-01 1.05853570e+00 -2.60004789e-01 -3.95159215e-01
-8.75738382e-01 1.38687420e+00 5.16534746e-01 2.55182564e-01
-1.02118766e+00 3.09049971e-02 1.07663274e+00 1.70935020e-01
7.71443963e-01 -2.85344869e-01 4.35002714e-01 -1.40930176e+00
1.95326701e-01 -1.05433023e+00 7.02296317e-01 -6.76372111e-01
-1.23608518e+00 3.72102588e-01 4.25692350e-02 -1.88623762e+00
-7.42619336e-01 -4.44768012e-01 -4.30534095e-01 2.28875816e-01
-1.44665980e+00 -1.11688876e+00 -3.00358713e-01 1.03948128e+00
1.13748205e+00 -4.72797781e-01 7.17191517e-01 4.67297941e-01
-3.71175826e-01 5.30054510e-01 1.56726599e-01 2.82000333e-01
8.20431173e-01 -1.00195098e+00 2.73559093e-01 8.44947517e-01
4.62644130e-01 2.90820360e-01 3.83459061e-01 -3.13657761e-01
-1.44397914e+00 -1.21962357e+00 5.71982563e-01 -2.91971058e-01
7.77596831e-01 -4.33457568e-02 -8.96056533e-01 1.00973964e+00
2.87006110e-01 4.73267913e-01 8.51741970e-01 -2.32204914e-01
-4.72033352e-01 -1.70693025e-01 -8.05711389e-01 3.35828781e-01
1.31224060e+00 -7.24085391e-01 -6.31413579e-01 6.36305451e-01
5.80916405e-01 -2.88557678e-01 -9.62304890e-01 3.26684296e-01
6.30232155e-01 -1.16326118e+00 1.08445263e+00 -9.69784379e-01
6.97813928e-01 -3.42267752e-01 -5.23634911e-01 -1.00220597e+00
-5.09536624e-01 -6.21319056e-01 -4.70875859e-01 9.54606891e-01
-3.08480591e-01 -2.72539467e-01 7.86455154e-01 1.57634541e-01
-2.85543501e-01 -9.53697801e-01 -9.32030916e-01 -8.80827904e-01
-1.92641810e-01 -3.02296221e-01 2.85878509e-01 7.96606123e-01
-2.69499794e-02 2.88694296e-02 -7.99243808e-01 -2.94587296e-02
4.31137919e-01 3.13412428e-01 8.62545192e-01 -6.21457756e-01
-5.25848866e-01 -3.56722027e-01 -9.46163535e-01 -1.36768126e+00
5.04942000e-01 -7.59626031e-01 5.85868210e-03 -1.30826962e+00
3.44375610e-01 2.37000033e-01 -5.55029154e-01 4.38465178e-01
-1.70284972e-01 4.47257340e-01 3.77700239e-01 2.97166973e-01
-1.15870821e+00 9.04936314e-01 1.00391603e+00 -6.02406263e-02
1.84783220e-01 -1.77720502e-01 -3.65515947e-01 7.45710373e-01
6.72384501e-01 -1.66878805e-01 -5.78698754e-01 -4.83235776e-01
-1.86030120e-01 1.15263030e-01 3.72855574e-01 -1.08506250e+00
1.17222108e-02 -2.88100749e-01 2.63911545e-01 -5.73100448e-01
6.15058124e-01 -6.84730291e-01 -1.91276334e-02 2.60475367e-01
-5.25366902e-01 1.19646281e-01 -3.75605896e-02 7.58274436e-01
-6.72514617e-01 -5.58765233e-02 6.54604733e-01 -2.24221557e-01
-1.25357163e+00 6.83955133e-01 -2.53235072e-01 2.39859983e-01
1.26665366e+00 -2.09572613e-01 -2.07556754e-01 -4.70290571e-01
-6.29728973e-01 9.90317836e-02 4.52816248e-01 7.11086273e-01
7.60093629e-01 -1.66820204e+00 -7.66293824e-01 1.89423934e-01
2.84035683e-01 -4.33196217e-01 5.52267969e-01 8.75130534e-01
-3.15039814e-01 4.65125710e-01 -5.32517552e-01 -5.44686615e-01
-1.39973903e+00 6.09393954e-01 1.72854617e-01 -3.09366822e-01
-7.23013699e-01 8.58511388e-01 3.77600908e-01 -4.77486727e-04
2.22243533e-01 -1.70942917e-01 -7.77365193e-02 1.26938879e-01
8.45464766e-01 4.98246163e-01 -1.91898674e-01 -9.52534139e-01
-3.98822248e-01 5.99753022e-01 -1.72932774e-01 1.87408537e-01
1.45997357e+00 -1.31852835e-01 1.68994337e-01 4.42949355e-01
1.63962185e+00 -4.06669945e-01 -1.87749779e+00 -3.68676871e-01
-1.59376919e-01 -9.91718113e-01 -2.70076811e-01 -1.46297485e-01
-1.26539266e+00 8.04947078e-01 5.74707031e-01 -2.56651193e-01
1.29577363e+00 1.96857736e-01 9.06375706e-01 3.12381774e-01
3.02901179e-01 -1.12889612e+00 7.33688831e-01 5.00917971e-01
1.02199817e+00 -1.38016713e+00 1.15780234e-01 3.45944948e-02
-9.02003884e-01 1.03237760e+00 6.55928254e-01 -3.70210469e-01
4.97174591e-01 -1.50733545e-01 -1.79712743e-01 -5.35650291e-02
-8.82148623e-01 -2.48318225e-01 2.41081387e-01 4.85744387e-01
5.25025904e-01 -1.24347344e-01 -1.84821486e-01 2.61372864e-01
2.02252001e-01 2.61763543e-01 3.25430751e-01 1.13103688e+00
-3.94317180e-01 -1.08258617e+00 3.14335376e-02 3.93767744e-01
-4.11813140e-01 -4.72430177e-02 -4.51622009e-02 5.18980503e-01
-4.60324287e-02 6.13244951e-01 2.82588124e-01 -2.64817894e-01
1.28109947e-01 -5.66426739e-02 5.81409335e-01 -5.50119877e-01
-2.48142570e-01 -4.29394282e-02 1.36455419e-02 -1.25613761e+00
-9.64644313e-01 -9.84834373e-01 -1.11038637e+00 -1.10389099e-01
1.97417021e-01 -1.35478094e-01 1.03244871e-01 9.85715210e-01
4.44662184e-01 3.58503282e-01 6.99518800e-01 -1.12776375e+00
-5.88672459e-01 -8.16561460e-01 -6.32569373e-01 7.87670314e-01
4.38794017e-01 -7.30736673e-01 -2.01552570e-01 3.44010949e-01]
|
[8.65495491027832, 0.7047462463378906]
|
6dd8a36a-7736-43fa-a8e8-4581cd98e015
|
forecasting-with-economic-news
|
2203.15686
| null |
https://arxiv.org/abs/2203.15686v1
|
https://arxiv.org/pdf/2203.15686v1.pdf
|
Forecasting with Economic News
|
The goal of this paper is to evaluate the informational content of sentiment extracted from news articles about the state of the economy. We propose a fine-grained aspect-based sentiment analysis that has two main characteristics: 1) we consider only the text in the article that is semantically dependent on a term of interest (aspect-based) and, 2) assign a sentiment score to each word based on a dictionary that we develop for applications in economics and finance (fine-grained). Our data set includes six large US newspapers, for a total of over 6.6 million articles and 4.2 billion words. Our findings suggest that several measures of economic sentiment track closely business cycle fluctuations and that they are relevant predictors for four major macroeconomic variables. We find that there are significant improvements in forecasting when sentiment is considered along with macroeconomic factors. In addition, we also find that sentiment matters to explains the tails of the probability distribution across several macroeconomic variables.
|
['Sebastiano Manzan', 'Sergio Consoli', 'Luca Barbaglia']
|
2022-03-29
| null | null | null | null |
['aspect-based-sentiment-analysis']
|
['natural-language-processing']
|
[-4.85442966e-01 -1.88341096e-01 -9.04726505e-01 -3.71533781e-01
-6.53381288e-01 -8.02287161e-01 1.03246820e+00 5.44330776e-01
-3.73034328e-01 5.50908685e-01 1.25431252e+00 -7.06927717e-01
3.61154266e-02 -9.42051113e-01 -4.27325964e-01 -3.05741102e-01
4.05077189e-01 1.94412753e-01 -3.53212714e-01 -6.94777906e-01
7.74492264e-01 -1.58414710e-02 -1.08622587e+00 -6.98385164e-02
6.06530249e-01 1.15360630e+00 1.46553263e-01 2.43001387e-01
-3.61943007e-01 1.09978187e+00 -5.64870179e-01 -8.43336761e-01
6.94760457e-02 -2.15993509e-01 -6.53004587e-01 7.60098100e-02
-2.39000127e-01 2.11323366e-01 6.02193363e-02 1.07221472e+00
8.69001225e-02 2.47507989e-02 8.99909079e-01 -7.18537509e-01
-7.54321575e-01 7.56186903e-01 -6.46912932e-01 7.69681752e-01
1.88297555e-01 -7.44477734e-02 1.57618618e+00 -9.36110497e-01
7.88079441e-01 1.06709337e+00 5.72141826e-01 -1.31981492e-01
-7.71778286e-01 -3.99510741e-01 3.14428300e-01 -2.71119028e-01
-6.73380196e-01 -5.63251019e-01 6.80946350e-01 -7.28869855e-01
1.15586829e+00 8.18617940e-02 6.25583947e-01 6.96331799e-01
9.77668583e-01 8.43756795e-02 1.31024992e+00 -3.68079662e-01
1.67112350e-01 4.14440364e-01 4.16084886e-01 1.25290871e-01
7.40346432e-01 -1.99967787e-01 -6.66252613e-01 -3.61083508e-01
1.75525665e-01 4.54749875e-02 1.62210450e-01 4.33858544e-01
-1.22277248e+00 1.44526851e+00 -1.05065137e-01 2.91455626e-01
-7.49846399e-01 -1.03988208e-01 6.11496031e-01 4.07039642e-01
1.20971251e+00 6.25503302e-01 -1.09101963e+00 -4.69205230e-01
-5.65129340e-01 3.20317715e-01 1.10239649e+00 4.00096089e-01
6.71296656e-01 3.48030999e-02 2.12245867e-01 7.55400658e-01
3.68154854e-01 1.07551575e+00 8.38140965e-01 -8.08700740e-01
6.79794133e-01 4.76295471e-01 2.17002407e-01 -1.57815063e+00
-5.18369913e-01 -5.17362058e-01 -3.98044020e-01 -2.08406433e-01
1.18036687e-01 -4.93994474e-01 -5.91621220e-01 1.46634758e+00
-2.51664650e-02 -6.18692100e-01 2.75043815e-01 4.89807248e-01
7.94398963e-01 8.24010193e-01 2.88951606e-01 -3.63638163e-01
1.79970932e+00 -7.49879062e-01 -8.90611410e-01 -7.51710474e-01
6.13332212e-01 -9.47271347e-01 1.01643503e+00 6.52112514e-02
-8.94528389e-01 -1.13396786e-01 -7.80478656e-01 1.04056388e-01
-7.19409645e-01 -3.89004856e-01 9.14794981e-01 5.82041919e-01
-9.39734995e-01 1.18943579e-01 -5.92998266e-01 3.15631441e-05
-6.31508827e-02 -1.08415604e-01 -7.20947087e-02 5.46369791e-01
-1.16911149e+00 1.14165974e+00 -1.08138345e-01 -6.91640317e-01
-7.30354115e-02 -4.01725769e-01 -9.92508948e-01 2.52005816e-01
1.24825358e-01 -5.70691645e-01 1.26886427e+00 -1.08264232e+00
-1.17454970e+00 6.86525822e-01 -6.48972690e-01 -3.62219661e-01
-2.77122706e-01 -1.15275150e-02 -7.05864012e-01 1.41547307e-01
9.27298486e-01 -3.32744032e-01 4.01247114e-01 -9.31632996e-01
-9.78762448e-01 -4.57362443e-01 1.71455845e-01 1.82638228e-01
-5.87070763e-01 5.51228404e-01 -3.16929936e-01 -1.22870684e+00
-4.40045744e-02 -7.21250355e-01 -2.73122162e-01 -1.21713996e+00
-8.49469900e-02 -2.89458543e-01 1.80780292e-01 -7.66046762e-01
1.45411575e+00 -1.74944556e+00 -2.96797663e-01 3.78706604e-01
7.41286874e-02 -5.34594119e-01 1.41379192e-01 4.01394278e-01
-3.43481638e-02 6.30720854e-01 3.00345477e-02 -3.09812635e-01
1.24592967e-01 -2.86898073e-02 -6.58824801e-01 4.32575494e-01
1.17443562e-01 1.03509915e+00 -7.23825455e-01 -1.32734910e-01
-7.23058954e-02 8.98705050e-02 -4.98428494e-01 -5.10280609e-01
5.00763766e-02 3.90706211e-03 -8.14394474e-01 7.82884538e-01
2.79671550e-01 -4.86542851e-01 1.62116215e-01 1.04632437e-01
-4.36371177e-01 1.08446848e+00 -6.02052867e-01 8.61172974e-01
-5.48040152e-01 8.27748775e-01 -1.20770931e-01 -1.00451660e+00
8.01725626e-01 2.42736265e-01 4.16827083e-01 -9.70828712e-01
2.74688363e-01 1.27888575e-01 -2.56528050e-01 -2.09007487e-01
9.46051776e-01 -4.70810086e-01 -5.32150090e-01 8.41102064e-01
-2.84028858e-01 -4.64032322e-01 5.49205959e-01 2.45842919e-01
7.49089658e-01 -4.67450589e-01 7.08277941e-01 -8.11184347e-01
1.61739647e-01 3.40031385e-01 8.25424373e-01 4.11534220e-01
1.48227008e-03 2.72910684e-01 9.66820717e-01 -2.94325590e-01
-9.25499141e-01 -3.89242381e-01 -2.92155236e-01 1.13063574e+00
-2.51216978e-01 -6.43314183e-01 -3.57539028e-01 -3.83278459e-01
2.21959785e-01 8.17466915e-01 -8.73149097e-01 2.94432998e-01
-1.35827228e-01 -1.38536263e+00 -2.87841678e-01 4.10723060e-01
8.05109218e-02 -8.26579273e-01 -4.57309246e-01 1.43257037e-01
-4.21657205e-01 -1.07444680e+00 -3.80749136e-01 4.65429544e-01
-7.80959785e-01 -8.75879586e-01 -2.74642855e-01 -5.19855201e-01
3.48675400e-01 3.12683791e-01 1.69655609e+00 -2.94060737e-01
7.99207211e-01 3.40933532e-01 -6.93343818e-01 -9.82928693e-01
-2.09587470e-01 2.71436095e-01 1.02495998e-01 -1.42060816e-01
8.71839225e-01 -1.28371090e-01 -4.88370568e-01 -9.41214561e-02
-6.02613032e-01 -4.37882334e-01 1.68260574e-01 4.82768506e-01
4.16914850e-01 3.58333737e-01 9.52572405e-01 -1.05722797e+00
1.02319825e+00 -1.03948355e+00 -4.89376307e-01 -2.09660828e-01
-1.18660021e+00 -1.95014954e-01 3.55090231e-01 9.78000686e-02
-1.04366422e+00 -7.47973621e-01 -5.98366298e-02 6.81623757e-01
1.64279833e-01 1.42001998e+00 5.64060271e-01 4.08012867e-01
3.54296416e-01 -1.36974514e-01 -7.13694170e-02 -3.24450463e-01
-1.56624079e-01 7.31033087e-01 2.20739767e-01 -3.86402637e-01
6.47238672e-01 7.59742916e-01 -3.12355429e-01 -5.70632875e-01
-1.20662010e+00 -7.59051383e-01 8.88341442e-02 3.39808911e-02
7.89893389e-01 -1.49376488e+00 -4.47645575e-01 3.32253337e-01
-7.38456190e-01 -8.76135752e-02 -1.88558653e-01 9.34180200e-01
-3.14814776e-01 -5.02755940e-02 -8.92167747e-01 -8.09132576e-01
-3.92518848e-01 -1.16011095e+00 7.42153168e-01 2.35066891e-01
-5.16716182e-01 -1.41421223e+00 4.62082893e-01 4.07175630e-01
5.39254248e-01 2.48599753e-01 7.98038185e-01 -6.64447308e-01
2.32711673e-01 -2.52312601e-01 -3.51592042e-02 1.11324199e-01
3.43968421e-01 1.72561571e-01 -6.81791842e-01 1.12849064e-01
5.19598663e-01 -1.34794995e-01 1.08618581e+00 9.72372711e-01
3.86650473e-01 -4.62448835e-01 -1.28102854e-01 3.67542654e-01
1.50704145e+00 2.29757354e-01 3.11877042e-01 1.14601564e+00
2.67967880e-01 7.49735355e-01 6.60545826e-01 8.04785967e-01
9.41234291e-01 2.34820962e-01 1.35412991e-01 4.79584634e-02
5.51032841e-01 -9.00795683e-02 5.83558202e-01 1.31654358e+00
-2.06610814e-01 -1.35896310e-01 -9.43374813e-01 9.60812390e-01
-1.46976268e+00 -1.02497566e+00 -1.51336715e-01 1.53390598e+00
8.44573736e-01 5.46649933e-01 2.50356555e-01 -9.40700434e-03
3.70638043e-01 6.22753203e-01 -1.57354549e-01 -6.92847550e-01
-5.90951443e-01 3.37941885e-01 9.36341584e-01 5.92851400e-01
-1.02543819e+00 8.39609146e-01 7.27196169e+00 3.38290125e-01
-9.43911076e-01 3.38375606e-02 1.29400921e+00 1.12333260e-01
-9.06989872e-01 4.11412120e-02 -8.95979106e-01 5.17560422e-01
1.16323960e+00 -6.06527746e-01 2.20342577e-02 8.67792547e-01
6.83233559e-01 -4.22112703e-01 -2.13065550e-01 3.12026829e-01
9.42752957e-02 -1.48104858e+00 -3.27901430e-02 4.32065815e-01
1.29386675e+00 2.99882799e-01 4.36454147e-01 6.37057126e-02
5.05173862e-01 -8.57936203e-01 1.07356441e+00 3.48843098e-01
6.19474947e-01 -1.17454314e+00 1.07152426e+00 7.20702345e-03
-9.91821826e-01 -1.48797333e-01 -4.24563140e-01 -6.52938902e-01
2.65832871e-01 1.15606356e+00 -1.42713264e-01 1.43190846e-01
9.12694216e-01 1.04848826e+00 -3.85133237e-01 1.23805650e-01
-5.27202487e-02 9.61077213e-01 -1.96882337e-03 -1.19342685e-01
5.19468248e-01 -5.12836277e-01 2.65667140e-01 1.20270658e+00
3.39659840e-01 2.59704620e-01 -3.61246586e-01 2.86979318e-01
-1.87632069e-01 4.25426245e-01 -7.50530899e-01 -3.30093831e-01
2.86250114e-01 1.06814241e+00 -9.28064644e-01 -5.62866747e-01
-1.15538299e+00 2.24006414e-01 1.08468652e-01 3.16368699e-01
-2.17031196e-01 -4.61292028e-01 9.56652939e-01 -7.16353878e-02
4.61468786e-01 -8.56812596e-02 -9.00648892e-01 -1.48759317e+00
-4.32077646e-02 -9.77540314e-01 2.10908368e-01 -5.35837710e-01
-1.30427635e+00 3.09562922e-01 -3.90919685e-01 -6.39181077e-01
-4.72000927e-01 -7.21540451e-01 -6.81321681e-01 9.60808694e-01
-1.79169309e+00 -3.34057897e-01 5.22774696e-01 3.43006760e-01
5.47710359e-01 -3.65425974e-01 6.81480706e-01 -1.23485267e-01
-4.24611479e-01 -1.22282840e-01 5.08144677e-01 2.88607389e-01
6.29394233e-01 -1.39026976e+00 9.11645114e-01 7.46416569e-01
4.59920578e-02 8.72923911e-01 8.67843926e-01 -8.41920674e-01
-1.09560883e+00 -7.26237297e-01 1.86228597e+00 -7.90359080e-01
1.19768250e+00 1.86774552e-01 -2.27867901e-01 8.56294155e-01
3.69613647e-01 -6.15449965e-01 1.04361558e+00 4.82322693e-01
-4.23407525e-01 7.96304420e-02 -8.99645329e-01 3.58335406e-01
2.78717369e-01 -7.69686162e-01 -9.59835231e-01 4.25348401e-01
7.55451024e-01 7.38168461e-03 -1.08409154e+00 7.15145543e-02
6.11650467e-01 -8.77549946e-01 7.48696983e-01 -6.34211659e-01
9.51545775e-01 1.20555840e-01 -5.42118073e-01 -1.74667585e+00
-6.58351958e-01 -3.65800142e-01 1.05510496e-01 9.70841467e-01
7.96478510e-01 -1.08181107e+00 4.12745595e-01 6.01361275e-01
-4.16597240e-02 -6.25289500e-01 -4.48908567e-01 -3.15636665e-01
4.42421198e-01 -6.10387623e-01 9.56524432e-01 1.25111639e+00
5.33854485e-01 5.04904091e-01 9.93658677e-02 -2.66649127e-01
1.76980034e-01 4.52673346e-01 5.06975889e-01 -1.45325470e+00
-9.00664926e-02 -6.69829190e-01 -2.86339968e-03 -8.39017212e-01
2.80685335e-01 -5.49033463e-01 -5.77839732e-01 -1.46127224e+00
5.83961487e-01 -1.96154296e-01 -3.59283537e-01 3.28134932e-02
-2.41341636e-01 4.30827290e-01 -1.62724834e-02 2.15028584e-01
-1.10333227e-01 2.25361705e-01 1.08974969e+00 -1.12598948e-01
-1.29945418e-02 2.86122654e-02 -1.68601191e+00 9.77064550e-01
1.00844371e+00 -2.98535258e-01 -1.03268057e-01 -3.43347639e-01
1.13216352e+00 1.40314074e-02 -1.04289651e-01 -1.95647404e-01
-1.61722526e-01 -6.80537701e-01 3.13368201e-01 -6.37300968e-01
4.90962900e-03 -4.83783841e-01 -3.45627666e-01 2.71528840e-01
-1.44393459e-01 8.73235703e-01 6.07781224e-02 4.95068640e-01
-5.65091312e-01 -2.40657926e-01 3.32986444e-01 -2.99386650e-01
-4.15839791e-01 4.02424857e-02 -8.45269680e-01 5.34041166e-01
6.19587123e-01 1.97553024e-01 -3.94117355e-01 -8.81032169e-01
-2.75113314e-01 -1.72532097e-01 5.09601772e-01 3.47975582e-01
3.94885130e-02 -1.12708580e+00 -8.60308468e-01 -1.23471551e-01
1.24022245e-01 -6.78423643e-01 -4.99297142e-01 7.31306851e-01
-3.08608532e-01 1.12774038e+00 2.56911486e-01 2.83908844e-01
-5.39750099e-01 2.99170345e-01 -3.92377190e-02 -4.65732425e-01
-3.43922466e-01 5.18761694e-01 2.14128718e-01 -9.53647792e-02
-4.54613179e-01 -5.96094310e-01 -7.28394568e-01 7.72181034e-01
5.51107645e-01 3.06830823e-01 -1.80816371e-02 -1.32197380e+00
-3.46750110e-01 7.43913651e-01 1.44188657e-01 -5.70702791e-01
1.73187351e+00 -6.43507123e-01 -4.42887872e-01 8.19519997e-01
1.14153790e+00 6.66343272e-01 -5.73904157e-01 -1.59599110e-01
2.21562102e-01 -2.78800726e-01 4.19926882e-01 -6.73594773e-01
-1.27908075e+00 1.59703810e-02 -3.08703393e-01 6.77140772e-01
9.10319626e-01 9.75986496e-02 7.09387958e-01 1.62930340e-01
3.73971947e-02 -1.54931033e+00 -3.88957143e-01 8.86806786e-01
4.86194611e-01 -1.37486351e+00 4.15787369e-01 1.56096220e-01
-9.83647168e-01 9.19103801e-01 -2.25148737e-01 1.72045045e-02
1.29918718e+00 2.57178426e-01 3.76631886e-01 -6.79268181e-01
-9.12163496e-01 -1.41831741e-01 1.96016297e-01 2.20040917e-01
7.17420399e-01 3.62812966e-01 -9.52909708e-01 1.00411892e+00
-8.26441228e-01 -5.68196476e-01 7.57793605e-01 6.87333047e-01
-6.41100347e-01 -6.88501358e-01 -4.19709712e-01 9.83798504e-01
-1.49254286e+00 -4.99628305e-01 -2.60002643e-01 5.26211262e-01
-5.09621561e-01 1.52545261e+00 3.83805126e-01 -2.77062625e-01
5.05615883e-02 1.70864444e-02 -5.59925377e-01 -6.30538523e-01
-6.94909871e-01 1.97544262e-01 4.31793064e-01 -3.63869756e-01
-7.65589774e-01 -1.16906941e+00 -8.87372255e-01 -7.14304745e-01
-4.02370572e-01 4.83055830e-01 1.00649273e+00 1.12901223e+00
3.26436311e-01 2.65679955e-01 1.11492860e+00 -4.63256359e-01
-3.34662527e-01 -1.08447468e+00 -9.33569133e-01 2.16272399e-01
6.10355198e-01 -5.85544765e-01 -8.81176770e-01 2.31840491e-01]
|
[4.493825912475586, 4.401335716247559]
|
087114d8-7d06-4e0c-9e2b-a27ee7a355d1
|
video-object-segmentation-with-language
|
1803.08006
| null |
http://arxiv.org/abs/1803.08006v3
|
http://arxiv.org/pdf/1803.08006v3.pdf
|
Video Object Segmentation with Language Referring Expressions
|
Most state-of-the-art semi-supervised video object segmentation methods rely
on a pixel-accurate mask of a target object provided for the first frame of a
video. However, obtaining a detailed segmentation mask is expensive and
time-consuming. In this work we explore an alternative way of identifying a
target object, namely by employing language referring expressions. Besides
being a more practical and natural way of pointing out a target object, using
language specifications can help to avoid drift as well as make the system more
robust to complex dynamics and appearance variations. Leveraging recent
advances of language grounding models designed for images, we propose an
approach to extend them to video data, ensuring temporally coherent
predictions. To evaluate our method we augment the popular video object
segmentation benchmarks, DAVIS'16 and DAVIS'17 with language descriptions of
target objects. We show that our language-supervised approach performs on par
with the methods which have access to a pixel-level mask of the target object
on DAVIS'16 and is competitive to methods using scribbles on the challenging
DAVIS'17 dataset.
|
['Anna Rohrbach', 'Anna Khoreva', 'Bernt Schiele']
|
2018-03-21
| null | null | null | null |
['referring-expression-segmentation']
|
['computer-vision']
|
[ 3.25765163e-01 -7.20706163e-03 -4.24236327e-01 -4.85672981e-01
-7.98421562e-01 -7.17975199e-01 7.37564504e-01 1.55317396e-01
-6.01087391e-01 5.35212159e-01 -2.94341385e-01 -6.31490126e-02
3.09690863e-01 -4.41343576e-01 -8.28921854e-01 -5.25420725e-01
1.90018844e-02 6.49309397e-01 8.18031251e-01 6.58726841e-02
1.16447620e-01 5.53633153e-01 -1.49586034e+00 3.45207274e-01
5.59700966e-01 9.74903882e-01 2.11799726e-01 6.39713347e-01
-4.49058175e-01 7.81242847e-01 -4.58898395e-01 -2.63391614e-01
3.51227164e-01 -4.79227215e-01 -1.26721275e+00 9.35805559e-01
7.67243266e-01 -3.02052289e-01 -1.31323665e-01 1.04685009e+00
-7.85893649e-02 2.22710222e-01 5.87081373e-01 -1.32083130e+00
-1.01764910e-01 7.13994801e-01 -6.58570170e-01 1.23355679e-01
3.38273257e-01 1.78198919e-01 9.45354283e-01 -6.47997856e-01
1.12432003e+00 1.13386667e+00 4.58589852e-01 6.83122277e-01
-1.48205793e+00 -2.55685419e-01 7.40842521e-01 -2.99888226e-04
-1.46525192e+00 -6.07488096e-01 7.74166405e-01 -7.52646625e-01
7.43144214e-01 1.23237252e-01 5.99731505e-01 8.81910264e-01
-2.83569515e-01 1.16545272e+00 1.15316677e+00 -4.36791331e-01
2.37109035e-01 1.86118662e-01 2.69477218e-01 9.12302196e-01
-5.88907301e-03 -2.14608297e-01 -3.21449816e-01 1.00514561e-01
7.37515688e-01 -2.28334144e-01 -2.60121793e-01 -7.43604422e-01
-1.31983876e+00 4.39744025e-01 2.20521629e-01 3.34597230e-01
-2.76768148e-01 3.63799691e-01 4.57161814e-01 -1.38747931e-01
6.35895371e-01 5.30927442e-02 -4.76616561e-01 -1.38052076e-01
-1.52660620e+00 2.78325438e-01 7.57265747e-01 1.07614815e+00
8.91406059e-01 -7.89418593e-02 -2.03980401e-01 4.30831164e-01
3.48450363e-01 9.95803345e-03 1.23689964e-01 -1.22671533e+00
2.03428343e-01 5.83339334e-01 3.00632894e-01 -5.31332850e-01
-1.82949319e-01 -1.52468637e-01 -3.46963972e-01 3.51748824e-01
7.62258887e-01 1.90544739e-01 -1.25454795e+00 1.66399288e+00
3.48121762e-01 3.21417481e-01 7.87938908e-02 8.78012180e-01
6.09841526e-01 6.12090945e-01 1.74856067e-01 -3.55394214e-01
1.15692937e+00 -1.31145370e+00 -5.44269979e-01 -3.68985653e-01
6.41010225e-01 -6.03496432e-01 8.84268761e-01 4.82748240e-01
-1.03357947e+00 -5.86256981e-01 -7.07500577e-01 -7.33513907e-02
-3.20569247e-01 2.00404122e-01 6.02854848e-01 5.97755194e-01
-1.07020056e+00 6.83847964e-01 -1.12723207e+00 -5.82036018e-01
6.61582470e-01 4.46020544e-01 -3.62605929e-01 2.64402200e-02
-5.78218400e-01 6.22377396e-01 6.35981500e-01 -1.19292744e-01
-1.00073636e+00 -6.39883161e-01 -9.89199221e-01 -3.59767228e-01
7.31661916e-01 -3.72376531e-01 1.27313292e+00 -1.47939730e+00
-1.48650205e+00 1.36393690e+00 -3.09920162e-01 -8.61901045e-01
9.13458467e-01 -3.70582193e-01 -8.90711509e-03 5.11889279e-01
1.20301820e-01 1.36689806e+00 9.79868114e-01 -1.50141180e+00
-6.40507042e-01 -9.71903745e-03 9.05513316e-02 -1.07391506e-01
8.81347135e-02 2.49854386e-01 -1.09824455e+00 -5.85277259e-01
-4.05378044e-02 -1.00691593e+00 -3.94253254e-01 3.06537420e-01
-5.80092967e-01 -2.74604678e-01 1.03654754e+00 -5.97576320e-01
1.16313946e+00 -2.10994124e+00 1.90569341e-01 2.18320210e-02
9.51862335e-02 3.61680210e-01 -6.00603856e-02 5.72165325e-02
6.10972987e-03 2.69436866e-01 -5.69545090e-01 -8.25947583e-01
-1.81592181e-01 5.07159770e-01 -3.50496888e-01 6.03143275e-01
4.84143406e-01 8.78208041e-01 -8.79057348e-01 -9.89624798e-01
3.86733353e-01 2.60984093e-01 -4.77960974e-01 3.76780741e-02
-9.41518784e-01 4.61652249e-01 -3.55689049e-01 5.29113829e-01
3.48787755e-01 -3.04313123e-01 -8.43054354e-02 3.54317613e-02
-8.33647698e-02 8.69866386e-02 -1.25079477e+00 1.98627448e+00
-1.05815995e-02 7.70287216e-01 5.34791276e-02 -1.02826118e+00
8.18882823e-01 1.83632568e-01 7.45243013e-01 -1.95784882e-01
4.34566848e-02 2.31717303e-02 -2.41259098e-01 -3.32883328e-01
3.47980618e-01 3.47236581e-02 1.17298946e-01 4.02799934e-01
1.22580610e-01 -3.58942360e-01 6.97206318e-01 4.03148115e-01
7.45018780e-01 9.20777082e-01 8.81170481e-02 -4.94759113e-01
6.57556057e-01 2.61564732e-01 5.29145360e-01 7.03064442e-01
-3.61417741e-01 8.87857080e-01 4.93871540e-01 -3.30015808e-01
-8.23938251e-01 -8.74936342e-01 3.45638357e-02 7.75335908e-01
2.79911071e-01 -5.46037495e-01 -1.07892990e+00 -8.78537953e-01
-2.59029448e-01 6.29368067e-01 -5.93330026e-01 3.90785635e-01
-6.90017760e-01 -2.27679938e-01 3.21110725e-01 5.60506046e-01
4.54075038e-01 -8.14451456e-01 -7.60587811e-01 2.02570662e-01
-1.22563265e-01 -1.71266317e+00 -5.76155782e-01 2.22612619e-01
-8.74199927e-01 -9.54938650e-01 -7.59659410e-01 -7.41605818e-01
9.22902644e-01 7.45245516e-02 1.33467293e+00 1.38358787e-01
-3.27755004e-01 7.00179636e-01 -1.88123226e-01 -1.98092446e-01
-5.76892793e-01 6.43234476e-02 -4.67256568e-02 2.32339680e-01
8.15596506e-02 -1.17467418e-01 -2.69935101e-01 3.41793537e-01
-1.01680255e+00 3.70116353e-01 2.05514744e-01 3.02188694e-01
9.02195871e-01 3.74957211e-02 -2.57490464e-02 -8.93837035e-01
-1.32076934e-01 2.62621716e-02 -9.09912169e-01 2.81849682e-01
-2.96268821e-01 2.54666835e-01 4.62866992e-01 -5.41547537e-01
-8.22335243e-01 6.87707782e-01 1.18505716e-01 -7.13605344e-01
-4.59565461e-01 1.38190925e-01 -5.50671965e-02 -5.33305034e-02
4.24017876e-01 -3.33297439e-02 4.81447913e-02 -4.51306075e-01
4.97222006e-01 1.30314410e-01 7.74833620e-01 -7.75690079e-01
7.52866507e-01 7.81370938e-01 -2.33910745e-03 -7.82463133e-01
-9.53506589e-01 -7.47404933e-01 -1.06709623e+00 -2.87247300e-01
1.08763027e+00 -8.42595398e-01 -2.87279278e-01 3.54464948e-01
-1.25666249e+00 -6.75670862e-01 -2.98157901e-01 1.72152728e-01
-8.43940377e-01 4.86741394e-01 -4.55126554e-01 -7.65178859e-01
1.23781934e-01 -1.33964002e+00 1.37810385e+00 7.51198381e-02
-3.46814454e-01 -1.02526498e+00 -2.65107274e-01 4.29634660e-01
-1.48092071e-02 3.11574817e-01 4.37305570e-01 -6.47353709e-01
-8.29298496e-01 -2.66149864e-02 -2.23888606e-01 4.02239442e-01
1.62890404e-01 4.50808525e-01 -8.44385028e-01 -3.82937342e-02
-2.41536260e-01 -1.92804635e-01 1.03137147e+00 3.77011687e-01
1.06905341e+00 -7.77401626e-02 -4.47941422e-01 3.72014701e-01
1.39478433e+00 4.29009683e-02 5.07767856e-01 2.62016773e-01
7.55369604e-01 7.77811706e-01 8.19957554e-01 6.98720813e-02
2.68293172e-01 8.89603794e-01 3.12286168e-01 -1.89474195e-01
-2.72262275e-01 -1.27121687e-01 3.90382141e-01 1.90031305e-01
1.21048331e-01 -2.96633184e-01 -1.04520428e+00 7.99074113e-01
-1.93960857e+00 -7.31067896e-01 -1.84837013e-01 1.92612982e+00
8.68249774e-01 4.16284978e-01 5.40036619e-01 7.39359483e-02
5.89918554e-01 1.52379438e-01 -3.47421199e-01 -3.13935913e-02
-1.79577693e-01 -1.02480628e-01 5.23499906e-01 6.02330029e-01
-1.37301564e+00 1.35063779e+00 6.11382437e+00 6.92322493e-01
-1.17155647e+00 1.70825366e-02 9.00430739e-01 6.28021434e-02
3.75632080e-04 1.74961478e-01 -9.84269202e-01 1.43607959e-01
6.60220921e-01 2.12003291e-01 2.20905155e-01 6.24116480e-01
3.17139238e-01 -5.86938858e-01 -1.57461154e+00 9.42236423e-01
8.74157324e-02 -1.41144753e+00 3.69092077e-02 -2.10755244e-01
8.92094433e-01 -1.66313797e-02 -2.23361716e-01 -1.35699019e-01
1.06211744e-01 -9.17317092e-01 1.08926535e+00 4.73460257e-01
4.37745124e-01 -3.64238143e-01 4.02213752e-01 3.52164507e-01
-1.12327623e+00 4.08484280e-01 1.56324401e-01 5.11502326e-02
3.44896406e-01 1.77409291e-01 -6.98231041e-01 3.40527266e-01
6.15514755e-01 8.70863557e-01 -7.39430606e-01 8.96259069e-01
-1.25557825e-01 7.23289430e-01 -4.83325630e-01 3.39513659e-01
5.46549857e-01 -2.59645253e-01 4.83278751e-01 1.33702838e+00
-1.39643863e-01 -2.24641319e-02 7.13088632e-01 1.00309408e+00
1.13485545e-01 8.37304667e-02 -3.92742544e-01 -3.55708897e-01
-1.36245459e-01 1.02768266e+00 -1.48780012e+00 -5.73469520e-01
-3.94125551e-01 1.21217048e+00 6.57960847e-02 5.00933766e-01
-9.01827574e-01 2.08753675e-01 3.97484660e-01 2.18049884e-01
5.87709129e-01 -6.15809619e-01 -1.79486603e-01 -1.07962525e+00
1.38507515e-01 -8.32869351e-01 1.48021892e-01 -8.22051227e-01
-7.57036984e-01 6.05158746e-01 2.69329816e-01 -1.12663102e+00
-1.97208375e-01 -6.89555287e-01 -3.08098495e-01 4.07285541e-01
-1.47015154e+00 -1.29507113e+00 -2.01584771e-01 4.16183412e-01
9.24881577e-01 2.08498836e-01 5.09823382e-01 1.84987321e-01
-5.48328996e-01 3.00476477e-02 -4.52570647e-01 3.34649712e-01
5.22975326e-01 -1.25684083e+00 3.62346113e-01 1.12511849e+00
8.13255370e-01 4.47749674e-01 9.15974498e-01 -5.40051818e-01
-1.11766553e+00 -1.11604238e+00 6.53790116e-01 -5.55406928e-01
6.80854380e-01 -5.26944816e-01 -9.73052740e-01 8.69490743e-01
2.15343326e-01 3.27005833e-01 1.42101750e-01 -4.17829275e-01
-1.49663135e-01 7.80073032e-02 -8.22242379e-01 7.23462522e-01
1.05569124e+00 -4.80927169e-01 -4.30921316e-01 4.78164732e-01
7.85284042e-01 -6.76025152e-01 -4.94163305e-01 3.40337366e-01
1.31605074e-01 -8.76127183e-01 9.07602489e-01 -6.98505640e-01
3.29557210e-01 -6.39487863e-01 -1.86985470e-02 -5.92784762e-01
4.06184047e-01 -1.02870953e+00 6.67515211e-03 1.51079035e+00
3.01469922e-01 -9.22027454e-02 1.00007117e+00 7.59639084e-01
5.65177053e-02 -7.21992254e-01 -7.89113343e-01 -8.12044561e-01
-1.40474305e-01 -8.74869227e-01 5.57234325e-02 5.53900242e-01
-4.95490104e-01 -1.08833298e-01 -2.21650735e-01 1.31531745e-01
6.01369977e-01 2.11786032e-01 8.18873882e-01 -9.75484133e-01
-1.96057394e-01 -6.80891931e-01 -5.46336174e-01 -1.43441057e+00
6.27901256e-01 -6.02591634e-01 2.61757284e-01 -1.46381414e+00
9.74490121e-02 -5.01932561e-01 -3.82012874e-02 4.63616848e-01
8.98968726e-02 6.07934833e-01 2.23313138e-01 6.16913289e-02
-1.04381156e+00 6.13753609e-02 1.01829231e+00 -2.75418907e-01
-3.68623614e-01 -3.07444017e-02 -1.43246859e-01 1.06887698e+00
5.63230634e-01 -3.87663752e-01 -2.75402576e-01 -4.10926074e-01
-1.45658657e-01 -2.35963333e-02 6.48707867e-01 -7.77770281e-01
2.68149346e-01 -2.42309824e-01 -1.85769459e-03 -6.87400877e-01
3.41825962e-01 -8.12140048e-01 9.90280807e-02 3.59257042e-01
-3.70801538e-01 -1.11878842e-01 3.76143903e-01 5.40030837e-01
-3.30596238e-01 -3.53542924e-01 8.91536534e-01 -1.87005475e-01
-1.22421491e+00 3.43725145e-01 -3.72942060e-01 4.50083651e-02
1.25526857e+00 -3.90401155e-01 1.29530355e-01 -2.59765744e-01
-8.65528643e-01 3.00563931e-01 7.96175778e-01 5.47818244e-01
4.24911827e-01 -8.67930174e-01 -3.84854853e-01 4.70833480e-02
4.43600081e-02 1.71357438e-01 -6.40600175e-02 9.33983147e-01
-7.09512949e-01 4.14067447e-01 1.42903030e-01 -1.03118205e+00
-1.50336289e+00 7.05241680e-01 4.33511764e-01 -1.07493214e-01
-6.89520657e-01 9.35956776e-01 3.32075417e-01 1.35245651e-01
4.61538583e-01 -6.46681011e-01 -2.50131562e-02 2.30962932e-02
1.95508257e-01 -6.52619898e-02 -2.09792838e-01 -1.02639675e+00
-4.90997374e-01 8.26080024e-01 -3.05365343e-02 -3.13339442e-01
1.01534939e+00 -2.65140533e-01 -1.83745265e-01 6.88075721e-01
1.01540899e+00 -9.83064622e-02 -1.55919552e+00 -4.23176497e-01
5.06314337e-01 -2.95750976e-01 5.39526716e-02 -5.56039810e-01
-1.06126845e+00 8.36198330e-01 2.93853760e-01 1.18266329e-01
1.04477751e+00 2.54568368e-01 5.56856036e-01 1.92422196e-01
4.20628458e-01 -1.05056024e+00 1.61844581e-01 2.31813595e-01
6.12182915e-01 -1.43744719e+00 7.50702620e-02 -7.98793733e-01
-7.06304967e-01 1.16790473e+00 5.50385177e-01 -4.78867721e-03
4.31290269e-01 2.93114275e-01 2.06774592e-01 -7.37934858e-02
-5.61697423e-01 -5.06226540e-01 5.95154285e-01 4.37394440e-01
3.32195699e-01 -2.07023084e-01 -2.38834489e-02 1.51565090e-01
1.62045613e-01 1.30284876e-02 4.52694833e-01 1.02121234e+00
-2.55751610e-01 -1.26731777e+00 -2.62491673e-01 1.03367344e-01
-6.57073855e-01 7.82196224e-02 -4.73575026e-01 9.81716692e-01
1.68260202e-01 7.90149808e-01 1.67974368e-01 2.33855844e-01
-7.45377839e-02 1.97553799e-01 6.61310434e-01 -9.15157020e-01
-2.83346146e-01 3.98412764e-01 8.61396641e-02 -6.68572426e-01
-1.21727598e+00 -9.12706196e-01 -1.66655254e+00 4.05825138e-01
-2.17781156e-01 -1.12363830e-01 5.69621205e-01 1.35835183e+00
2.69141197e-02 3.59108299e-01 7.87897706e-02 -1.06436157e+00
-1.08045712e-01 -3.41393411e-01 -3.33388776e-01 5.96356273e-01
3.97008568e-01 -5.56534946e-01 -7.64109641e-02 7.73152411e-01]
|
[9.161745071411133, -0.16196516156196594]
|
a067e9d5-d5ff-4640-8383-168389eb9a62
|
case-base-neural-networks-survival-analysis
|
2301.06535
| null |
https://arxiv.org/abs/2301.06535v3
|
https://arxiv.org/pdf/2301.06535v3.pdf
|
Case-Base Neural Networks: survival analysis with time-varying, higher-order interactions
|
Neural network-based survival methods can model data-driven covariate interactions. While these methods can provide better predictive performance than regression-based approaches, not all can model time-varying interactions and complex baseline hazards. To address this, we propose Case-Base Neural Networks (CBNNs) as a new approach that combines the case-base sampling framework with flexible neural network architectures. Using a novel sampling scheme and data augmentation to naturally account for censoring, we construct a feed-forward neural network that may take time as an input. CBNNs predict the probability of an event occurring at a given moment to estimate the hazard function. We compare the performance of CBNNs to regression and neural network-based survival methods in a simulation and three case studies using two time-dependent metrics. First, we examine performance on a simulation involving a complex baseline hazard and time-varying interactions to assess all methods, with CBNN outperforming competitors. Then, we apply all methods to three real data applications, with CBNNs outperforming the competing models in two studies and showing similar performance in the third. Our results highlight the benefit of combining case-base sampling with deep learning to provide a simple and flexible modeling framework for data-driven, time-varying interaction modeling of single event survival outcomes. An R package is available at https://github.com/Jesse-Islam/cbnn.
|
['Sahir Bhatnagar', 'Robert Sladek', 'Maxime Turgeon', 'Jesse Islam']
|
2023-01-16
| null | null | null | null |
['survival-analysis']
|
['miscellaneous']
|
[-1.44749269e-01 -2.93741018e-01 -5.08955300e-01 -7.65360117e-01
-9.20624733e-01 4.45901938e-02 4.61578578e-01 3.44939053e-01
-4.62059110e-01 1.13949168e+00 3.97828668e-01 -8.44780922e-01
-4.09033775e-01 -9.29773986e-01 -6.59345567e-01 -5.75430393e-01
-6.67840540e-01 4.74390566e-01 -2.54629493e-01 -2.09388867e-01
-2.65418410e-01 6.41202211e-01 -9.96601224e-01 6.45961389e-02
5.95961094e-01 6.79684401e-01 -6.33290112e-01 6.88579321e-01
4.07154143e-01 6.78176582e-01 -5.07796884e-01 -1.84228197e-01
-1.33555699e-02 -1.34638444e-01 -3.30908149e-01 -7.69400597e-01
1.39835566e-01 -6.78930581e-01 -5.77496886e-01 1.25120683e-02
8.54130149e-01 2.14708913e-02 9.12158787e-01 -1.54581070e+00
-4.20967132e-01 6.36499643e-01 -2.24366874e-01 1.72059923e-01
-1.08192466e-01 2.31843337e-01 5.88953853e-01 -5.11595249e-01
1.06922522e-01 1.25289309e+00 1.46107423e+00 6.55035973e-01
-1.47212636e+00 -7.00238883e-01 -1.42843053e-01 -2.56130517e-01
-1.23602855e+00 -1.84569970e-01 5.67101598e-01 -5.66662133e-01
8.16148818e-01 3.64757866e-01 4.82544094e-01 1.54398322e+00
6.18195415e-01 6.49998724e-01 9.54798818e-01 -1.55418769e-01
2.65604228e-01 -3.61369044e-01 5.43223977e-01 2.04418585e-01
1.54443786e-01 1.04537606e+00 -1.35980891e-02 -6.60515428e-01
8.70255053e-01 7.81794250e-01 -9.95518714e-02 -3.71407233e-02
-9.30606067e-01 1.07347357e+00 5.54541886e-01 8.23326036e-03
-5.84700882e-01 6.67884111e-01 5.45222402e-01 1.39956653e-01
6.29845798e-01 -3.23735774e-02 -8.88172686e-01 1.16674848e-01
-1.19526339e+00 4.30646718e-01 8.54216754e-01 4.79375839e-01
7.16332421e-02 2.19618469e-01 -8.54089618e-01 7.44361103e-01
3.79885286e-02 2.21271828e-01 1.81474820e-01 -6.98331356e-01
1.03107765e-02 3.60433847e-01 2.25145534e-01 -4.26697493e-01
-1.19658685e+00 -7.83770263e-01 -1.42935431e+00 3.51665735e-01
8.18134546e-01 -5.60129702e-01 -1.07186556e+00 2.12036395e+00
1.00982696e-01 1.02399111e-01 -1.10708326e-01 3.60358298e-01
7.04435647e-01 4.89367038e-01 4.36511308e-01 -1.35082036e-01
1.22536397e+00 -4.73979622e-01 -5.41413426e-01 1.91145658e-01
9.40007269e-01 9.50047523e-02 8.07930470e-01 2.40755633e-01
-1.11683249e+00 -1.84579670e-01 -8.76030564e-01 1.01600084e-02
-4.40814942e-01 -1.18511796e-01 8.13223124e-01 5.95122874e-01
-1.10451210e+00 9.07728553e-01 -1.07289708e+00 -3.98490578e-01
5.64006150e-01 5.82442641e-01 -1.22323617e-01 3.40250880e-02
-1.65861845e+00 7.10303426e-01 1.47329643e-01 -5.60903996e-02
-1.16154087e+00 -1.21113038e+00 -7.37712324e-01 4.21997219e-01
-1.18501119e-01 -1.21408999e+00 1.25362349e+00 -6.95559740e-01
-9.19709682e-01 2.65837431e-01 1.00951642e-01 -7.99598813e-01
7.44577765e-01 6.06336212e-03 -3.98738593e-01 -5.28276503e-01
-1.18362106e-01 4.71096933e-01 1.35424687e-02 -9.28635478e-01
-4.88042772e-01 -2.92479217e-01 -2.19561875e-01 -2.49001816e-01
-2.23344967e-01 1.56889200e-01 1.34865209e-01 -9.63327467e-01
-6.44806147e-01 -6.01941407e-01 -5.08483589e-01 -1.07470369e-02
-3.75731826e-01 -2.97004789e-01 2.15342790e-01 -7.60971546e-01
1.47898316e+00 -2.01542401e+00 -3.91967624e-01 -1.09096475e-01
1.53797075e-01 -1.49173960e-01 -9.85087454e-02 7.61464596e-01
-4.61602896e-01 2.06255913e-01 -4.87400115e-01 -5.70862114e-01
-2.74947882e-01 1.04994379e-01 1.45918444e-01 3.23314577e-01
1.78214625e-01 1.10422075e+00 -5.11968911e-01 -1.54029161e-01
-6.81848675e-02 8.86418760e-01 -4.12235856e-01 2.34520063e-01
1.20234191e-01 3.84012818e-01 -2.21330021e-02 7.38271356e-01
4.74747509e-01 -2.97434568e-01 -1.52351469e-01 3.41373652e-01
4.03284840e-02 -1.55210623e-03 -6.92788064e-01 1.02367735e+00
-5.12033224e-01 4.98688012e-01 -2.83795834e-01 -1.06942546e+00
8.43827486e-01 5.54991961e-01 6.07350469e-01 -4.02946323e-01
2.08571792e-01 1.83528826e-01 1.05799939e-02 -7.25623891e-02
-1.27907962e-01 -5.98024070e-01 -2.84442782e-01 2.76769936e-01
-1.28479972e-01 6.04637623e-01 -2.05360055e-01 -9.10824463e-02
1.44969177e+00 -1.73424408e-01 5.00662744e-01 -2.52315015e-01
-2.99523156e-02 -1.56992376e-01 8.06480944e-01 1.12168193e+00
-7.77264237e-02 6.54718399e-01 7.80751884e-01 -1.00381052e+00
-8.30677390e-01 -1.18102419e+00 -6.37402534e-01 1.00758171e+00
-7.13100255e-01 2.13341728e-01 -4.60282624e-01 -5.74625075e-01
4.17972744e-01 1.05402839e+00 -1.32967961e+00 -3.74816149e-01
-3.05208713e-01 -1.44120514e+00 8.43733907e-01 8.94434810e-01
-7.50011131e-02 -1.03397346e+00 -4.08889055e-01 4.43747967e-01
8.24649632e-02 -1.00478925e-01 -3.11131954e-01 6.52287543e-01
-1.08708894e+00 -1.01329267e+00 -1.03643143e+00 -3.53209347e-01
2.04435766e-01 -4.51998264e-01 1.44453883e+00 2.21887410e-01
-1.86517879e-01 1.17054082e-01 8.68794844e-02 -7.28502929e-01
-5.27166009e-01 1.84216753e-01 -4.23148051e-02 -4.01386499e-01
2.71204114e-01 -7.67941236e-01 -9.94806945e-01 2.07382873e-01
-1.15795970e+00 -5.09714559e-02 4.19415087e-01 1.32324612e+00
2.31602356e-01 -3.43132615e-01 1.29097927e+00 -9.21462715e-01
5.42886496e-01 -1.15844226e+00 -5.15656829e-01 1.32367030e-01
-8.76680255e-01 -3.07401121e-01 6.34302378e-01 -6.20213687e-01
-6.51969969e-01 -8.63756612e-02 -1.01464264e-01 -1.49054781e-01
-2.15822503e-01 9.79416251e-01 9.63324606e-02 6.00129604e-01
7.30961263e-01 -3.72879766e-02 2.18232751e-01 -4.91605878e-01
-1.34423941e-01 4.36973631e-01 4.10113811e-01 -4.92670268e-01
1.91289365e-01 4.11767572e-01 3.11158478e-01 -3.54693606e-02
-4.86543417e-01 -2.18700264e-02 -3.84769648e-01 -2.27034781e-02
6.35474622e-01 -9.89333749e-01 -8.45277190e-01 8.39698315e-01
-9.78908479e-01 -9.86283481e-01 -3.20981354e-01 5.63540339e-01
-6.08052611e-01 -3.72569114e-01 -8.10605586e-01 -8.10735822e-01
-5.25961399e-01 -7.87870646e-01 7.98348784e-01 6.31573126e-02
-1.37754887e-01 -1.55583870e+00 2.87354857e-01 -2.80822724e-01
6.90125585e-01 1.05265033e+00 1.21213734e+00 -1.22930050e+00
5.80387656e-03 -7.88254619e-01 -1.56075180e-01 7.56234229e-02
1.78160891e-01 -8.09171721e-02 -7.51851201e-01 -5.34129798e-01
-3.74032825e-01 7.81263167e-04 8.40546668e-01 1.23901176e+00
1.42689478e+00 -2.93160200e-01 -5.34581244e-01 9.45682168e-01
1.25316453e+00 5.49683452e-01 5.98092973e-01 3.10839266e-01
4.35317904e-01 5.88150501e-01 1.82839260e-01 5.74306846e-01
7.36414969e-01 5.93185067e-01 4.82432455e-01 -7.51676977e-01
2.05966562e-01 -1.18020706e-01 1.33980513e-01 -9.82809141e-02
7.87051842e-02 -4.77547377e-01 -1.32459998e+00 7.75723517e-01
-1.86314666e+00 -8.18417490e-01 -4.73124683e-01 2.50428104e+00
9.71599579e-01 1.17654346e-01 6.15482211e-01 -1.01258792e-01
4.95965093e-01 -3.16257924e-01 -7.61401236e-01 -4.04426187e-01
-7.90278763e-02 2.29040295e-01 5.74299455e-01 3.24199885e-01
-1.09977722e+00 2.14654818e-01 6.74181461e+00 5.84483087e-01
-1.06899393e+00 1.47286430e-01 1.40186286e+00 -3.94355327e-01
2.87238532e-03 -1.43072411e-01 -6.15744650e-01 4.85761762e-01
1.54162419e+00 -2.38467753e-01 -7.45540438e-03 4.65045184e-01
5.79790950e-01 1.93113357e-01 -1.46832192e+00 3.45353723e-01
-5.84657192e-01 -1.27368867e+00 -2.74317920e-01 1.58769205e-01
4.53787118e-01 1.14459410e-01 1.08153515e-01 7.36820638e-01
8.28561068e-01 -1.46896672e+00 2.16103703e-01 9.23432827e-01
1.01934266e+00 -8.78973424e-01 1.09952748e+00 2.68448293e-01
-5.40948868e-01 -1.73590958e-01 1.66496634e-01 -2.17289358e-01
4.00956184e-01 7.43243992e-01 -6.95315361e-01 5.88645160e-01
8.30440879e-01 5.29551327e-01 -4.67415094e-01 1.24998236e+00
3.90585870e-01 1.07450533e+00 -3.58378500e-01 1.91685006e-01
-2.01338846e-02 5.13391376e-01 1.92646205e-01 1.27090168e+00
5.55148721e-01 -1.45745156e-02 -1.18037179e-01 7.33651757e-01
9.22070667e-02 -2.10180283e-01 -5.44095397e-01 5.00864923e-01
4.14029866e-01 9.75540340e-01 -3.74564260e-01 -2.77730495e-01
-4.54919994e-01 3.28115284e-01 7.61256963e-02 5.77632964e-01
-9.62258160e-01 -3.62087518e-01 4.58012342e-01 5.08206308e-01
-1.29038453e-01 1.09351210e-01 -4.98783112e-01 -6.78820252e-01
-4.29609329e-01 -6.95340872e-01 9.82502997e-01 -6.70709193e-01
-1.70447302e+00 4.43130076e-01 3.68292660e-01 -1.02482164e+00
-4.06155109e-01 -4.89606112e-01 -8.36308241e-01 1.15765572e+00
-1.21612811e+00 -1.27841222e+00 -2.19650447e-01 4.52906519e-01
1.82115152e-01 4.37407754e-02 8.53214264e-01 3.88040006e-01
-8.06119978e-01 9.97941613e-01 4.38314945e-01 2.11089730e-01
9.45589006e-01 -1.32404876e+00 5.94797075e-01 2.36363515e-01
-5.75734258e-01 5.40229201e-01 3.81037891e-01 -7.95766950e-01
-7.62196004e-01 -1.39553678e+00 7.57006586e-01 -5.52665591e-01
4.09009188e-01 -2.79392451e-01 -1.14912844e+00 7.83411920e-01
-1.19182374e-02 4.46470752e-02 9.57979143e-01 5.22745132e-01
-2.18489319e-01 -1.71130493e-01 -1.29092479e+00 5.64274311e-01
7.33730495e-01 3.04622073e-02 -1.02672130e-01 2.23470122e-01
6.48973465e-01 -2.18463719e-01 -1.25528669e+00 8.04224491e-01
8.29090238e-01 -9.47779357e-01 9.89268601e-01 -1.02624214e+00
6.85591698e-01 1.99673906e-01 3.86652797e-02 -1.40890145e+00
-4.92481351e-01 -4.36450630e-01 -1.11253999e-01 1.16378438e+00
6.01185679e-01 -9.26546037e-01 5.01358867e-01 9.29301679e-01
6.17579184e-02 -1.14585590e+00 -1.17472732e+00 -8.92046571e-01
9.20857668e-01 -4.28453386e-01 9.77574527e-01 9.52695727e-01
-3.16805989e-01 1.35528287e-02 -6.26542389e-01 8.26583952e-02
6.18146241e-01 -4.93293941e-01 5.38668394e-01 -1.47689879e+00
-1.95966303e-01 -4.59528506e-01 -6.35368899e-02 -2.01672852e-01
5.75565025e-02 -6.32382035e-01 -1.98624089e-01 -1.44561720e+00
4.78693217e-01 -5.91324985e-01 -9.22262371e-01 8.78363192e-01
-4.11289632e-01 -9.84450132e-02 -2.23875418e-01 -4.30683047e-02
1.38179570e-01 4.67729449e-01 5.46691656e-01 5.19805364e-02
-3.80460262e-01 5.68645477e-01 -7.47233808e-01 3.48913550e-01
1.07770729e+00 -6.55480444e-01 -1.31524220e-01 -4.14523110e-02
-2.83499379e-02 9.71002281e-01 8.16703260e-01 -9.49905217e-01
-3.10129859e-02 -2.48859748e-01 7.45119631e-01 -5.83344638e-01
-3.89352366e-02 -7.91133642e-01 6.22571290e-01 7.52176881e-01
-5.58865309e-01 3.68367612e-01 6.19051039e-01 5.80183566e-01
1.00632861e-01 3.33949745e-01 5.85142553e-01 2.90257335e-01
2.94114232e-01 6.16370738e-01 -5.51995456e-01 -1.91939771e-01
7.34000742e-01 -3.49329151e-02 -3.09583575e-01 -7.78508067e-01
-9.42023277e-01 5.40406644e-01 2.53248692e-01 3.95112485e-01
3.65326315e-01 -1.45153677e+00 -1.04204559e+00 1.96412832e-01
1.18780278e-01 -8.04755390e-02 4.50453013e-01 9.32894111e-01
-1.99051321e-01 2.62826115e-01 -1.47034988e-01 -3.40184450e-01
-9.81828034e-01 7.03177989e-01 8.48502755e-01 -5.92163742e-01
-2.07171977e-01 5.05780876e-01 5.54108441e-01 -6.79406285e-01
2.90046811e-01 -3.83889943e-01 -9.75392312e-02 -8.21229368e-02
3.85734171e-01 5.98498821e-01 -1.14191726e-01 -3.49503048e-02
-2.63846129e-01 -1.58245608e-01 -1.60420798e-02 -3.93836424e-02
1.53911126e+00 2.29557559e-01 8.03153738e-02 7.95089424e-01
1.08063066e+00 -6.88257694e-01 -1.42087841e+00 -1.63998734e-02
5.82787134e-02 1.31222084e-01 7.39416480e-02 -1.47764111e+00
-8.93553913e-01 8.31754327e-01 8.92447948e-01 2.71916091e-01
1.35304260e+00 -4.21405643e-01 4.43320334e-01 -1.52308360e-01
-1.24658877e-03 -3.70982170e-01 -4.41319913e-01 3.29743773e-01
9.89290833e-01 -1.27335393e+00 -2.16411009e-01 2.55131751e-01
-2.48147428e-01 9.06815052e-01 5.92158020e-01 -7.47701302e-02
1.05213046e+00 3.42059940e-01 9.52828526e-02 -2.64533106e-02
-1.06242800e+00 3.81181806e-01 2.56230354e-01 4.29811686e-01
7.38337874e-01 2.00254634e-01 -3.73358399e-01 1.13835096e+00
7.51773044e-02 5.90015411e-01 3.73076111e-01 5.49930751e-01
3.22936863e-01 -1.10741091e+00 -4.12067354e-01 9.19292748e-01
-8.82914603e-01 -3.74080688e-01 3.58601697e-02 1.26824903e+00
-2.06494927e-01 7.20483959e-01 1.93942294e-01 -1.78276792e-01
4.97449130e-01 2.45592549e-01 -1.82976604e-01 -2.33147502e-01
-8.89336586e-01 -1.51160479e-01 4.45651822e-02 -3.77504498e-01
-1.91136897e-02 -8.11372459e-01 -9.04506266e-01 -4.96774942e-01
-1.23565570e-01 -1.77376270e-02 4.27125633e-01 5.22349954e-01
4.26482528e-01 9.13456440e-01 6.53987885e-01 -7.28343546e-01
-7.10026443e-01 -1.08625019e+00 -5.54047287e-01 2.11735830e-01
7.84208179e-01 -6.08241260e-01 -5.19323349e-01 -3.09076697e-01]
|
[7.821594715118408, 5.654816150665283]
|
e6b01ecf-5026-4324-8efa-18464ea4d436
|
deepsat-a-learning-framework-for-satellite
|
1509.03602
| null |
http://arxiv.org/abs/1509.03602v1
|
http://arxiv.org/pdf/1509.03602v1.pdf
|
DeepSat - A Learning framework for Satellite Imagery
|
Satellite image classification is a challenging problem that lies at the
crossroads of remote sensing, computer vision, and machine learning. Due to the
high variability inherent in satellite data, most of the current object
classification approaches are not suitable for handling satellite datasets. The
progress of satellite image analytics has also been inhibited by the lack of a
single labeled high-resolution dataset with multiple class labels. The
contributions of this paper are twofold - (1) first, we present two new
satellite datasets called SAT-4 and SAT-6, and (2) then, we propose a
classification framework that extracts features from an input image, normalizes
them and feeds the normalized feature vectors to a Deep Belief Network for
classification. On the SAT-4 dataset, our best network produces a
classification accuracy of 97.95% and outperforms three state-of-the-art object
recognition algorithms, namely - Deep Belief Networks, Convolutional Neural
Networks and Stacked Denoising Autoencoders by ~11%. On SAT-6, it produces a
classification accuracy of 93.9% and outperforms the other algorithms by ~15%.
Comparative studies with a Random Forest classifier show the advantage of an
unsupervised learning approach over traditional supervised learning techniques.
A statistical analysis based on Distribution Separability Criterion and
Intrinsic Dimensionality Estimation substantiates the effectiveness of our
approach in learning better representations for satellite imagery.
|
['Supratik Mukhopadhyay', 'Manohar Karki', 'Sangram Ganguly', 'Saikat Basu', 'Robert DiBiano', 'Ramakrishna Nemani']
|
2015-09-11
| null | null | null | null |
['satellite-image-classification']
|
['computer-vision']
|
[ 3.20792735e-01 -2.96217471e-01 -1.05256885e-01 -5.73766887e-01
-4.14251536e-01 -1.56614646e-01 5.64328134e-01 -4.53706719e-02
-4.74084407e-01 6.75694048e-01 -1.71249136e-02 -2.47565463e-01
-5.82456887e-01 -1.22557437e+00 -3.71807218e-01 -1.09414041e+00
-4.46783602e-01 3.80762249e-01 3.22228819e-02 -4.33515459e-01
8.58986527e-02 8.09174895e-01 -2.15648699e+00 2.07453728e-01
8.73772502e-01 1.37563705e+00 6.72403052e-02 4.66691107e-01
-1.05381332e-01 6.76364183e-01 -2.77432710e-01 7.62902573e-02
3.60775381e-01 -7.90865347e-02 -8.11688066e-01 2.94402957e-01
4.77770090e-01 -4.97935861e-01 -4.08729196e-01 1.10945320e+00
3.67757469e-01 2.08216131e-01 8.02372038e-01 -1.03738010e+00
-6.58942699e-01 4.41279978e-01 -4.18318510e-01 4.57635015e-01
-4.42252964e-01 -1.83805481e-01 8.16176772e-01 -6.90763056e-01
1.65145710e-01 9.81840551e-01 9.46221232e-01 1.24738902e-01
-1.23086810e+00 -5.57468593e-01 -8.61167312e-02 2.62592643e-01
-1.56579709e+00 -2.62409359e-01 3.99032116e-01 -6.88209236e-01
8.88876557e-01 3.65658283e-01 7.49205291e-01 4.89503592e-01
-1.59191899e-02 3.19638968e-01 1.54504848e+00 -4.36819553e-01
3.42454642e-01 1.18052132e-01 5.17823637e-01 3.44145983e-01
3.00311983e-01 3.37900549e-01 -5.73665202e-02 4.79629673e-02
6.77106321e-01 2.97426164e-01 -3.56476039e-01 -5.36945798e-02
-8.20343792e-01 1.16455829e+00 7.56631434e-01 6.97353840e-01
-7.11061478e-01 -3.68962526e-01 2.28514016e-01 4.00507778e-01
8.14295053e-01 5.51890843e-02 -3.71819645e-01 4.03613776e-01
-1.32406056e+00 1.07860118e-01 6.65617168e-01 1.85465559e-01
9.77191925e-01 4.14473951e-01 2.73665875e-01 9.18258131e-01
2.85611540e-01 7.92301476e-01 8.00379097e-01 -4.95764524e-01
3.20682209e-03 7.89547861e-01 -4.69915681e-02 -1.19929588e+00
-5.57713270e-01 -8.31975162e-01 -1.22997057e+00 4.61185127e-01
-2.39311531e-03 1.10302202e-01 -1.14103913e+00 1.03698182e+00
1.45357713e-01 -3.46618369e-02 3.84768456e-01 9.41321611e-01
1.00120366e+00 8.75654578e-01 -1.25091106e-01 7.36780018e-02
1.05451882e+00 -5.60799301e-01 -5.20580173e-01 -1.90422937e-01
2.83073068e-01 -3.02699119e-01 6.24567032e-01 4.97950166e-01
-1.84337422e-01 -6.12038791e-01 -1.08621883e+00 4.07027721e-01
-6.59459651e-01 4.37523007e-01 7.60220468e-01 7.82100737e-01
-1.01316226e+00 7.07081079e-01 -9.79639232e-01 -6.46417558e-01
4.68852401e-01 4.08635616e-01 -5.41146576e-01 2.12882590e-02
-9.37383115e-01 1.08619571e+00 7.54861116e-01 3.64361107e-01
-6.90685213e-01 -4.57352817e-01 -6.75295591e-01 1.95188224e-01
-2.45754734e-01 -1.30711779e-01 8.69110107e-01 -1.30923343e+00
-1.42504478e+00 7.22979784e-01 2.37363279e-01 -6.47729576e-01
-1.59958806e-02 -2.73256153e-02 -6.76522732e-01 8.67340788e-02
-2.07985123e-03 5.67520320e-01 7.17200041e-01 -1.15341842e+00
-9.10804272e-01 -6.00814402e-01 -2.63926119e-01 1.65958665e-02
-6.36588216e-01 -1.62999153e-01 3.36390167e-01 -4.56752002e-01
6.20638192e-01 -7.90915787e-01 -2.32360333e-01 -1.90536633e-01
-4.09764685e-02 5.50245419e-02 1.07196128e+00 -7.60759115e-01
9.48100924e-01 -2.14158845e+00 -6.80473521e-02 3.80340606e-01
-1.63252547e-01 4.50212955e-01 -9.34643671e-02 2.87595540e-01
-3.93684387e-01 6.58229068e-02 -4.40522134e-01 3.89110923e-01
-1.68975428e-01 4.71604973e-01 -4.89695281e-01 5.61162233e-01
2.08816558e-01 3.69302899e-01 -5.62314868e-01 -1.82687581e-01
4.35512215e-01 6.06922686e-01 -5.16875498e-02 4.55988757e-03
1.31963670e-01 1.03766039e-01 -1.58549219e-01 9.08041596e-01
9.22125936e-01 -6.30587786e-02 1.45853400e-01 -4.44409609e-01
-4.72226113e-01 -4.32454944e-02 -1.25418305e+00 1.03478098e+00
-1.07744291e-01 7.90883541e-01 -5.40441498e-02 -1.54918909e+00
1.27740157e+00 9.49221924e-02 3.26527715e-01 -6.46741390e-01
8.77407044e-02 3.45085412e-01 -9.77945402e-02 -6.19928777e-01
4.40672517e-01 -4.41706449e-01 3.27842921e-01 1.02715753e-01
2.87993938e-01 1.01440765e-01 1.18630208e-01 -4.25145060e-01
5.34849048e-01 -4.18819189e-02 2.05250084e-01 -4.37601149e-01
3.88500571e-01 3.20164174e-01 3.99534881e-01 6.90661371e-01
-1.15168609e-01 3.67430627e-01 2.58610398e-02 -1.05159080e+00
-8.61342847e-01 -7.02382147e-01 -7.31825829e-01 1.10423040e+00
-1.47573173e-01 2.11308226e-01 -5.29187679e-01 -3.25267792e-01
1.73519060e-01 5.21550775e-01 -6.72011018e-01 2.53954381e-01
5.02369255e-02 -1.52216351e+00 7.30049670e-01 4.76817071e-01
8.80910516e-01 -6.93621755e-01 -3.92648846e-01 6.07191250e-02
-1.69544175e-01 -7.71647155e-01 7.56219923e-01 5.20475328e-01
-1.45841885e+00 -8.95384252e-01 -3.81952226e-01 -6.95206702e-01
4.51154232e-01 4.73968327e-01 8.90072644e-01 9.93141904e-02
-1.55579269e-01 -6.78141564e-02 -4.53764856e-01 -2.87454486e-01
-2.06282675e-01 2.16600358e-01 9.47843045e-02 1.72524974e-01
6.12369776e-01 -7.79956698e-01 -3.63660723e-01 2.35342115e-01
-1.18565297e+00 -3.29796672e-01 8.97898436e-01 9.51383710e-01
5.43505132e-01 6.67704582e-01 2.10704684e-01 -4.37523246e-01
6.15509264e-02 -6.67191863e-01 -5.68219960e-01 1.69853941e-02
-5.96434534e-01 -7.82285184e-02 2.28941306e-01 -1.48315370e-01
-9.59360540e-01 3.22098583e-01 -1.55010328e-01 3.78742442e-02
-5.27744949e-01 1.04705811e+00 1.22886486e-01 -3.13604355e-01
9.98185635e-01 5.94070852e-01 3.24056059e-01 -6.80610359e-01
-5.03155366e-02 1.27686524e+00 4.59770411e-01 -1.82310864e-02
8.07217836e-01 6.53148115e-01 -1.63075700e-01 -1.38420904e+00
-9.20910239e-01 -4.81375992e-01 -8.68051410e-01 -2.16482028e-01
6.84521437e-01 -1.24059308e+00 -1.78296596e-01 7.97621429e-01
-6.03543639e-01 -8.16739127e-02 -1.38725117e-01 6.49665236e-01
-1.07286267e-01 2.35405684e-01 -2.99748063e-01 -8.11177194e-01
-4.01217312e-01 -8.15044880e-01 7.81633019e-01 2.93449014e-01
2.81018615e-01 -8.06076884e-01 1.93117440e-01 3.26983213e-01
7.14470983e-01 2.83583254e-01 6.21873736e-01 -8.02734554e-01
-2.39238843e-01 -4.10747111e-01 -3.93677324e-01 8.44690740e-01
2.37542957e-01 2.66455293e-01 -1.05106533e+00 -2.56359190e-01
-1.50089934e-01 -4.15744931e-01 1.23890114e+00 4.84164208e-01
9.64638889e-01 -2.31087834e-01 -8.02074447e-02 7.40741968e-01
1.80469573e+00 1.85375690e-01 8.71163607e-01 8.56832862e-01
2.92376429e-01 5.43518603e-01 2.95618266e-01 3.76669973e-01
2.45678678e-01 2.23269463e-01 8.13928664e-01 -1.58817977e-01
2.34634399e-01 3.06174546e-01 1.86208501e-01 7.13912427e-01
-5.39933562e-01 8.15683901e-02 -1.07535565e+00 5.09871364e-01
-1.72191298e+00 -1.22396028e+00 -5.08529127e-01 2.04958797e+00
4.94102269e-01 -1.89288169e-01 -8.96621495e-02 6.56843662e-01
6.55301571e-01 1.62459597e-01 -1.81065395e-01 -1.64875109e-02
-3.43598485e-01 3.45700473e-01 6.89122081e-01 3.01858604e-01
-1.84636354e+00 6.96654499e-01 6.56972170e+00 5.85317671e-01
-1.59498346e+00 -9.10149738e-02 4.47537333e-01 3.02251786e-01
2.50559926e-01 -2.49522582e-01 -7.78083861e-01 2.77373940e-01
1.12614560e+00 3.28927040e-01 1.69295341e-01 1.07475269e+00
5.62440306e-02 -2.40491167e-01 -4.62362975e-01 7.78176606e-01
1.76790506e-01 -1.24971688e+00 3.11191648e-01 1.49980560e-01
7.14008570e-01 5.15715778e-01 -1.84456538e-02 4.45960671e-01
2.71735996e-01 -1.18559301e+00 3.79967570e-01 8.03183854e-01
3.76503885e-01 -6.88884079e-01 1.20687377e+00 4.24497992e-01
-8.58397067e-01 -3.07586849e-01 -7.15556264e-01 -4.31331754e-01
-4.32559371e-01 7.81146049e-01 -4.50902939e-01 6.88021302e-01
1.26334393e+00 7.79289961e-01 -5.96376359e-01 1.10183084e+00
1.30153894e-01 9.05048132e-01 -5.56178510e-01 1.18235745e-01
3.77984494e-01 -3.60807478e-01 1.49793580e-01 1.26317286e+00
5.24495840e-01 3.14033270e-01 5.85721135e-02 3.74227196e-01
4.06704158e-01 6.25591576e-02 -5.45663834e-01 -2.25866392e-01
2.93413132e-01 1.31039035e+00 -6.86658263e-01 -3.22166145e-01
-2.82945186e-01 6.70163333e-01 -8.68806690e-02 1.91651404e-01
-4.32235479e-01 -4.59205359e-01 5.99145472e-01 -1.68602750e-01
6.19797826e-01 -1.57894805e-01 -2.66767621e-01 -1.16326356e+00
-2.86403626e-01 -9.30705786e-01 2.41429701e-01 -7.93404877e-01
-1.30596352e+00 9.74414110e-01 -6.89730272e-02 -1.46149063e+00
-4.58676033e-02 -9.60270286e-01 -3.31703961e-01 8.55992675e-01
-1.78195107e+00 -1.25094283e+00 -7.98479378e-01 4.06042337e-01
8.12706426e-02 -5.10999143e-01 1.31188774e+00 3.59594315e-01
-6.06430352e-01 -1.22907914e-01 8.19863558e-01 4.23090518e-01
1.52102217e-01 -1.01536298e+00 -1.28828853e-01 7.83130288e-01
6.63301796e-02 1.24555252e-01 5.25825918e-01 -2.28636950e-01
-1.17723954e+00 -1.19668388e+00 8.13462257e-01 1.26585588e-01
7.20891535e-01 2.05844611e-01 -1.15268087e+00 5.92871904e-01
-5.67504205e-02 3.48212242e-01 1.00981832e+00 -3.50934565e-02
-4.06659871e-01 -6.67152882e-01 -1.17737818e+00 -9.16269273e-02
3.33741456e-01 -4.08395410e-01 -8.19842160e-01 3.13865811e-01
-1.46381464e-02 2.58265100e-02 -1.12039101e+00 5.10239840e-01
7.45914936e-01 -1.17747235e+00 9.43981588e-01 -5.87315202e-01
5.38338363e-01 -3.87700707e-01 -6.66894376e-01 -1.38211370e+00
-6.88410401e-01 5.16852856e-01 3.25276583e-01 1.06900537e+00
2.02929407e-01 -5.76572359e-01 6.79077804e-01 7.27186352e-02
4.67167646e-02 -2.57207930e-01 -9.51917768e-01 -8.67510557e-01
9.67125818e-02 -3.76269341e-01 5.54009497e-01 1.21826553e+00
-6.04406238e-01 5.62907159e-02 -2.24093363e-01 5.75797260e-01
8.02447081e-01 3.23408097e-01 5.15215695e-01 -1.91830957e+00
5.24442382e-02 -3.84075344e-01 -7.96596229e-01 -4.82579917e-01
-7.06871077e-02 -7.09785759e-01 1.81283392e-02 -1.58916855e+00
1.74767107e-01 -3.88914555e-01 -3.14914972e-01 8.20194483e-01
3.34693015e-01 5.98353744e-01 -1.26519024e-01 4.90566790e-01
3.49947065e-02 6.32031739e-01 5.45749605e-01 -3.90870512e-01
-6.41988441e-02 2.06707101e-02 -4.76997674e-01 8.40647042e-01
9.19694722e-01 -4.84264046e-01 2.46525872e-02 -4.16534513e-01
-6.56306893e-02 -3.00162166e-01 5.81254482e-01 -1.36477888e+00
2.45838556e-02 -1.21724717e-01 6.18013740e-01 -8.50584686e-01
4.86376099e-02 -9.99056816e-01 3.75570774e-01 5.05128562e-01
1.13877319e-01 -4.44107383e-01 1.81141883e-01 3.78130883e-01
-6.07371747e-01 -4.06130493e-01 8.61249030e-01 -6.06888831e-02
-1.19073641e+00 2.63001751e-02 -7.23701954e-01 -5.88427007e-01
7.60721266e-01 -4.77223665e-01 -2.52621502e-01 -5.34489006e-02
-7.67872453e-01 -1.65739536e-01 -2.80593466e-02 1.67053059e-01
4.38482732e-01 -1.21915960e+00 -9.98709857e-01 3.55716586e-01
3.80932502e-02 -8.68439302e-02 2.95130223e-01 7.90999770e-01
-8.55917096e-01 3.69936466e-01 -6.71495855e-01 -9.19225276e-01
-1.23544800e+00 -5.88399321e-02 6.57277226e-01 -8.83353688e-03
-3.90517056e-01 5.66593468e-01 -4.16961402e-01 -6.61707520e-01
1.13372937e-01 -2.79525578e-01 -5.34828961e-01 5.38929284e-01
6.35174692e-01 4.75651115e-01 4.96509433e-01 -8.75412941e-01
-3.52832913e-01 4.56838697e-01 1.13126621e-01 1.34669125e-01
2.10200214e+00 1.62800863e-01 -4.32062924e-01 3.87781411e-01
9.41971838e-01 -7.09813476e-01 -9.20883775e-01 -3.41207176e-01
3.16449776e-02 -4.69881415e-01 8.54712486e-01 -8.45126688e-01
-1.30584145e+00 8.32203627e-01 1.27193439e+00 4.42812264e-01
1.37666714e+00 -4.90035683e-01 2.22175762e-01 8.28373015e-01
4.88323160e-02 -9.89016831e-01 -4.70975012e-01 7.43143737e-01
8.77069175e-01 -1.59174168e+00 1.98124379e-01 -4.47980687e-02
-3.19551229e-01 1.45152628e+00 2.57413626e-01 -1.82443425e-01
9.13209379e-01 1.61467105e-01 2.60382861e-01 -1.21018559e-01
-2.90557534e-01 -6.05065584e-01 2.96290070e-01 6.74198866e-01
4.61969495e-01 1.87177703e-01 -1.24453664e-01 5.38189650e-01
-1.65476412e-01 2.84339637e-01 3.07741582e-01 9.87814724e-01
-1.04714143e+00 -7.38936841e-01 -8.64536643e-01 5.99198639e-01
-3.09967279e-01 -1.98536187e-01 -2.54636675e-01 9.34704781e-01
3.58168542e-01 8.07587445e-01 3.50033104e-01 -6.14556611e-01
1.93320602e-01 1.78908393e-01 7.65567347e-02 -2.28572190e-01
-5.49994469e-01 -1.51149660e-01 -6.19318224e-02 -6.43887594e-02
-1.00458920e+00 -5.71708798e-01 -8.64253938e-01 -4.44154978e-01
-3.75989109e-01 2.12240219e-01 1.01881170e+00 9.53732848e-01
1.64840370e-01 2.64159292e-01 6.75352216e-01 -1.26301789e+00
-8.23437989e-01 -1.27864337e+00 -9.86086011e-01 1.05823763e-01
3.46931010e-01 -6.60576105e-01 -6.26976490e-01 2.32462034e-01]
|
[9.670945167541504, -1.4671649932861328]
|
81cd1083-0108-43a7-8d76-d389046bb68a
|
stickypillars-robust-feature-matching-on
|
2002.03983
| null |
https://arxiv.org/abs/2002.03983v3
|
https://arxiv.org/pdf/2002.03983v3.pdf
|
StickyPillars: Robust and Efficient Feature Matching on Point Clouds using Graph Neural Networks
|
Robust point cloud registration in real-time is an important prerequisite for many mapping and localization algorithms. Traditional methods like ICP tend to fail without good initialization, insufficient overlap or in the presence of dynamic objects. Modern deep learning based registration approaches present much better results, but suffer from a heavy run-time. We overcome these drawbacks by introducing StickyPillars, a fast, accurate and extremely robust deep middle-end 3D feature matching method on point clouds. It uses graph neural networks and performs context aggregation on sparse 3D key-points with the aid of transformer based multi-head self and cross-attention. The network output is used as the cost for an optimal transport problem whose solution yields the final matching probabilities. The system does not rely on hand crafted feature descriptors or heuristic matching strategies. We present state-of-art art accuracy results on the registration problem demonstrated on the KITTI dataset while being four times faster then leading deep methods. Furthermore, we integrate our matching system into a LiDAR odometry pipeline yielding most accurate results on the KITTI odometry dataset. Finally, we demonstrate robustness on KITTI odometry. Our method remains stable in accuracy where state-of-the-art procedures fail on frame drops and higher speeds.
|
['Horst-Michael Gross', 'Kai Fischer', 'Martin Simon', 'Florian Oelsner', 'Stefan Milz', 'Patrick Maeder']
|
2020-02-10
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Fischer_StickyPillars_Robust_and_Efficient_Feature_Matching_on_Point_Clouds_Using_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Fischer_StickyPillars_Robust_and_Efficient_Feature_Matching_on_Point_Clouds_Using_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['3d-feature-matching']
|
['computer-vision']
|
[-2.41503865e-01 -2.86899775e-01 8.62052143e-02 -4.14045185e-01
-1.15218031e+00 -2.64411718e-01 7.04070985e-01 3.55946034e-01
-7.07455695e-01 3.50242406e-01 -2.89189070e-01 -4.07659151e-02
-2.15089738e-01 -8.37000072e-01 -9.66138661e-01 -3.97694558e-01
-1.77949399e-01 1.38326943e+00 5.01844585e-01 -4.12615895e-01
1.99678406e-01 9.50967491e-01 -1.76659179e+00 -4.34378505e-01
5.89128494e-01 1.07105792e+00 1.79991767e-01 4.06411797e-01
-1.21724486e-01 -2.73633134e-02 -2.22003043e-01 -3.19688469e-01
6.62122309e-01 3.32998395e-01 -4.55980718e-01 -3.84015560e-01
1.27995682e+00 -1.52034312e-01 -2.74220258e-01 8.41420293e-01
8.69195461e-01 2.08499968e-01 3.47753584e-01 -1.42380524e+00
-2.48200670e-01 1.28419310e-01 -5.57461858e-01 4.36456017e-02
4.01698500e-01 1.50727123e-01 8.40485394e-01 -1.28526497e+00
7.09319115e-01 1.17793620e+00 1.35842299e+00 9.08115581e-02
-1.29822934e+00 -7.89357245e-01 -2.84234256e-01 1.57749102e-01
-1.72595787e+00 -5.00836790e-01 6.03499115e-01 -3.56135219e-01
1.47265422e+00 -2.20538393e-01 6.90892220e-01 6.87954247e-01
2.79283673e-01 1.16452828e-01 7.43626595e-01 -2.46943876e-01
1.78207219e-01 -2.40681380e-01 -1.73995733e-01 8.45386624e-01
2.29306445e-01 3.70447338e-01 -6.08240485e-01 -1.99465483e-01
7.18529940e-01 1.44015834e-01 1.42161906e-01 -8.39031518e-01
-1.29278791e+00 7.54074454e-01 9.95523572e-01 1.28181428e-01
-5.30132055e-01 5.45399308e-01 1.40270188e-01 1.76218286e-01
5.12384057e-01 1.61947221e-01 -4.64754581e-01 -6.36827424e-02
-1.26020360e+00 4.19508487e-01 6.37152970e-01 1.09203041e+00
1.29573298e+00 -2.00501367e-01 2.33986959e-01 5.20135880e-01
3.41080397e-01 8.45383644e-01 -3.91674452e-02 -1.04177976e+00
4.44656312e-01 6.50063932e-01 -7.86873102e-02 -1.05245304e+00
-7.70712793e-01 -3.62725854e-01 -4.89866525e-01 4.85082179e-01
1.79176658e-01 2.69319862e-01 -1.20482099e+00 1.24899912e+00
5.46869516e-01 4.44157332e-01 -2.43806243e-01 8.95482838e-01
8.15678596e-01 2.15512574e-01 -1.81306183e-01 3.70480359e-01
1.13995290e+00 -5.15038490e-01 -2.07425162e-01 -3.84472132e-01
5.84137261e-01 -9.00490820e-01 8.11962008e-01 1.52561128e-01
-1.01216924e+00 -4.92884696e-01 -1.08015239e+00 -4.77924198e-01
-4.81328905e-01 -1.75729364e-01 5.62857985e-01 3.71120244e-01
-1.45528138e+00 8.62283051e-01 -9.63453531e-01 -6.40182316e-01
5.37991703e-01 9.93600488e-01 -7.25042343e-01 -1.28804907e-01
-6.24822557e-01 1.31831169e+00 1.55330583e-01 1.00676522e-01
-5.04094183e-01 -9.43214715e-01 -1.13128340e+00 -2.26114348e-01
8.82987306e-02 -9.79469478e-01 1.05310500e+00 -2.60725886e-01
-1.39121032e+00 1.18858409e+00 -2.83510089e-01 -5.56717455e-01
5.99848509e-01 -4.05439943e-01 -8.80515901e-04 -8.38994756e-02
3.26397777e-01 9.92234409e-01 5.93613207e-01 -1.08192253e+00
-5.76533139e-01 -6.23113394e-01 -3.05111170e-01 1.25045806e-01
3.51165563e-01 -1.30839095e-01 -4.93234158e-01 2.86356956e-02
7.51985669e-01 -1.14766812e+00 -3.65539730e-01 2.85787165e-01
-1.45903140e-01 -2.44373366e-01 8.65095675e-01 -2.35030025e-01
2.55224824e-01 -1.98254812e+00 -6.63367361e-02 4.10215169e-01
1.53908238e-01 -4.60754149e-02 -1.48767188e-01 2.72466063e-01
9.45168361e-02 -3.92859161e-01 -3.23592842e-01 -8.62175524e-01
1.60984814e-01 2.64717966e-01 -1.62175670e-01 9.98392701e-01
2.93136656e-01 9.05808151e-01 -8.42380226e-01 -4.57818300e-01
6.26023054e-01 9.15309250e-01 -5.42857766e-01 3.08544952e-02
-9.89482254e-02 4.07371283e-01 6.64240792e-02 8.14299703e-01
8.55532587e-01 -2.73597036e-02 -4.59687769e-01 -3.14967185e-01
-3.60187024e-01 4.96192813e-01 -1.36262929e+00 2.41432166e+00
-6.13867164e-01 6.08455360e-01 1.69085562e-02 -6.73480928e-01
1.14257050e+00 -1.40739843e-01 7.17247367e-01 -6.62238479e-01
2.83079296e-01 6.19743049e-01 -2.27465227e-01 7.05374079e-03
8.40200961e-01 -7.43908361e-02 -4.02722061e-02 1.53691724e-01
3.11243922e-01 -7.64757633e-01 -2.29055062e-01 7.78782964e-02
1.11559892e+00 4.99396861e-01 1.21359184e-01 -3.74966085e-01
3.30186695e-01 2.59953856e-01 3.21444392e-01 5.09308159e-01
7.68602714e-02 9.67819214e-01 -1.57832280e-01 -7.21527457e-01
-1.05881917e+00 -1.02390122e+00 -3.23093683e-01 7.46831715e-01
3.34778398e-01 -3.30740333e-01 -2.70855010e-01 -3.23155373e-01
4.58281815e-01 4.07696575e-01 -4.75157320e-01 6.21359311e-02
-7.56724715e-01 -4.40915108e-01 5.35586178e-01 5.37898183e-01
3.06789994e-01 -9.05130029e-01 -5.76272786e-01 3.78920943e-01
1.76647112e-01 -1.29629064e+00 -2.07018331e-02 3.68596792e-01
-9.94055271e-01 -9.23544288e-01 -3.67976278e-01 -6.09555304e-01
6.04021668e-01 3.63837719e-01 1.41832221e+00 9.42404717e-02
-2.34602824e-01 2.81050175e-01 1.63964048e-01 -4.45621908e-01
-5.09640984e-02 4.56305832e-01 3.91407639e-01 -5.58594525e-01
6.87951207e-01 -8.56713414e-01 -4.66161489e-01 3.38694721e-01
-3.36930752e-01 -4.96862322e-01 3.92535955e-01 5.75388014e-01
8.80641103e-01 -6.74745739e-01 -5.40277362e-02 -2.73847789e-01
8.53692815e-02 -2.36519128e-01 -1.14350903e+00 -2.79586852e-01
-5.31920135e-01 1.86662778e-01 8.66805464e-02 -1.18562371e-01
-1.96764275e-01 6.14034832e-01 -4.77485210e-01 -6.95133865e-01
-1.83031484e-01 7.00937510e-02 1.07348114e-01 -8.03023279e-01
9.11706805e-01 3.48099158e-03 1.92905843e-01 -4.81684357e-01
4.14475471e-01 3.51800740e-01 7.38325894e-01 -5.13182700e-01
1.04283595e+00 8.79534483e-01 4.26916152e-01 -7.13168502e-01
-5.09601355e-01 -6.21565402e-01 -1.18330610e+00 1.49498051e-02
7.04406619e-01 -1.10193646e+00 -9.34762418e-01 3.63692522e-01
-1.34079134e+00 -3.06905448e-01 -2.84282982e-01 4.19342339e-01
-6.92569554e-01 1.64016768e-01 -3.13162744e-01 -4.89342183e-01
-4.56349432e-01 -1.34309804e+00 1.73331308e+00 4.08532210e-02
-3.26842046e-03 -7.97735274e-01 3.43019634e-01 -7.94273708e-03
3.81094038e-01 4.23308015e-01 1.96272612e-01 -4.30720448e-01
-1.06110370e+00 -5.56141496e-01 -3.54816377e-01 -2.01524556e-01
-1.05147466e-01 1.00670956e-01 -1.09500229e+00 -2.80572891e-01
-3.74250561e-01 -2.06100959e-02 7.50160396e-01 2.82512844e-01
6.61626756e-01 2.75556505e-01 -5.59457302e-01 1.08621109e+00
1.62156177e+00 -4.44713563e-01 4.76518899e-01 6.78418040e-01
9.36328888e-01 4.46750671e-01 5.07416546e-01 1.14853352e-01
7.58417964e-01 9.85096335e-01 8.77896249e-01 -1.23814970e-01
-1.18315905e-01 -9.86183509e-02 8.78192857e-02 6.33990884e-01
-1.40850171e-01 2.64466286e-01 -1.47475767e+00 6.55786872e-01
-1.81227767e+00 -6.15291178e-01 -4.13171202e-01 2.42674565e+00
3.49360108e-01 1.79577619e-01 -1.61781553e-02 -9.99737009e-02
4.26206976e-01 -2.48451680e-02 -3.54716480e-01 8.92816950e-03
-6.01634160e-02 5.56140006e-01 9.85466838e-01 7.82626331e-01
-1.15301144e+00 1.26602912e+00 5.77611876e+00 3.19823563e-01
-1.23510253e+00 3.56857240e-01 -2.27637798e-01 -2.34491020e-01
5.75913899e-02 1.04977958e-01 -1.08015215e+00 1.39626101e-01
8.82861376e-01 1.72302917e-01 2.77950943e-01 9.64184105e-01
-1.07856721e-01 -2.23309278e-01 -1.15254557e+00 1.26405776e+00
1.64150774e-01 -1.52361822e+00 -3.53429079e-01 3.13878983e-01
3.41516793e-01 1.10454679e+00 -2.33584642e-01 1.25813618e-01
4.78669852e-01 -7.96041012e-01 8.24260592e-01 2.51586735e-01
7.34571993e-01 -7.77992904e-01 8.12510192e-01 2.26760194e-01
-1.27082610e+00 3.36226791e-01 -6.21008515e-01 -1.18636593e-01
3.45782459e-01 7.11170316e-01 -1.01251495e+00 6.23940051e-01
8.32954347e-01 6.55057788e-01 -5.66742599e-01 1.31452167e+00
-1.07944034e-01 -1.86943159e-01 -1.21275294e+00 2.83995360e-01
2.69091547e-01 -9.30059552e-02 5.83307624e-01 1.04979420e+00
5.65898418e-01 -1.42646551e-01 2.53935844e-01 6.69132352e-01
-9.64833125e-02 -1.16142072e-01 -1.04277384e+00 8.13918948e-01
5.58288693e-01 1.25156152e+00 -7.48708904e-01 -2.04592779e-01
-1.20314158e-01 8.26332331e-01 4.93941009e-01 -1.79072931e-01
-5.99269152e-01 -1.94774181e-01 9.04863656e-01 3.35482568e-01
2.58642763e-01 -8.29014480e-01 -4.05241877e-01 -9.65106606e-01
4.96132709e-02 -2.77336091e-01 1.10607475e-01 -7.48073161e-01
-1.13981056e+00 7.49804735e-01 -7.86989927e-02 -1.23213303e+00
-2.59981483e-01 -5.51590860e-01 -4.70000744e-01 8.22052598e-01
-1.61004961e+00 -1.27923846e+00 -7.54263937e-01 6.88039362e-01
1.53900787e-01 5.30085228e-02 7.36407161e-01 7.45016575e-01
-1.67717651e-01 3.26902181e-01 -2.78185844e-01 -5.73916174e-02
7.82200694e-01 -1.15774286e+00 1.10095787e+00 7.55075455e-01
4.29609686e-01 5.04886150e-01 7.16130435e-01 -6.93093002e-01
-1.48936963e+00 -9.26111519e-01 1.11393881e+00 -8.39480042e-01
6.33560359e-01 -6.28066361e-01 -8.63313973e-01 7.29768276e-01
-1.68472052e-01 4.72428709e-01 1.39540285e-01 2.16975451e-01
-2.65879393e-01 -2.26046458e-01 -1.22022426e+00 2.05941826e-01
1.33559752e+00 -5.88081896e-01 -6.18434489e-01 5.77884972e-01
6.53468251e-01 -1.05472398e+00 -7.07334459e-01 5.09528816e-01
5.16390264e-01 -1.01849544e+00 1.30543983e+00 -1.19481511e-01
-1.28813908e-01 -5.67786396e-01 -2.70871401e-01 -1.03127992e+00
-2.14444607e-01 -5.62861025e-01 1.02377295e-01 8.21827173e-01
2.01144114e-01 -6.27104461e-01 9.53439415e-01 5.47701180e-01
-3.66639227e-01 -2.85432756e-01 -1.47814238e+00 -9.85887945e-01
-2.04976931e-01 -7.39829779e-01 8.69549394e-01 9.51163828e-01
-5.41393518e-01 1.29674271e-01 -1.97272971e-02 4.67975199e-01
1.08283985e+00 3.03032380e-02 1.34063137e+00 -1.66457379e+00
3.73989463e-01 -3.42950523e-01 -1.00585258e+00 -6.85315907e-01
2.37367049e-01 -1.04850006e+00 3.91727656e-01 -1.63299358e+00
-5.77944458e-01 -8.59419286e-01 1.02084227e-01 6.14753902e-01
1.93960115e-01 7.33961940e-01 1.63195327e-01 3.98664683e-01
-4.42726344e-01 5.68108320e-01 4.87090409e-01 -6.63579404e-02
-2.96982318e-01 -1.05023339e-01 5.96280620e-02 6.95427120e-01
6.20914876e-01 -9.23679292e-01 1.36303216e-01 -7.91171968e-01
3.14715952e-01 -4.40440685e-01 7.23729670e-01 -1.44755709e+00
5.95308125e-01 3.02633286e-01 2.32386306e-01 -1.03506505e+00
7.55106747e-01 -1.11044061e+00 3.92478734e-01 4.36436325e-01
4.18252289e-01 7.00318813e-01 3.35523754e-01 2.77234674e-01
4.06329259e-02 1.23014502e-01 7.79814065e-01 -5.80075942e-02
-7.49549150e-01 7.28909194e-01 2.83078790e-01 -2.75765002e-01
8.08481216e-01 -5.07446170e-01 -2.27884240e-02 -9.31277499e-03
-5.08126616e-01 1.30211815e-01 8.92463744e-01 5.06319463e-01
5.24220049e-01 -1.45199358e+00 -7.76176214e-01 4.19910550e-01
2.32675895e-01 6.32543504e-01 -1.35279253e-01 1.04694307e+00
-9.08446312e-01 1.66948631e-01 -2.60756850e-01 -1.40057254e+00
-9.34404731e-01 2.62580782e-01 5.99134445e-01 7.23066479e-02
-1.06013930e+00 8.44658613e-01 -4.69032675e-01 -8.59172940e-01
1.67884395e-01 -4.64771897e-01 3.05999100e-01 1.79994807e-01
1.38386503e-01 2.92086482e-01 6.69329464e-01 -9.24199164e-01
-8.59534621e-01 1.16608894e+00 2.44438842e-01 -1.88068345e-01
1.55401492e+00 1.18976735e-01 -9.15466174e-02 1.81746766e-01
1.23528099e+00 -1.75427169e-01 -1.18681860e+00 -3.00910681e-01
1.84563756e-01 -6.80266500e-01 2.77623177e-01 -3.08306307e-01
-1.00251722e+00 9.95542824e-01 8.84630859e-01 -1.83605880e-01
3.92978191e-01 1.24126345e-01 9.13648486e-01 5.30553699e-01
8.74417365e-01 -7.42714942e-01 -4.58960414e-01 7.38302171e-01
8.10772777e-01 -1.50837743e+00 4.05680507e-01 -1.71873599e-01
-7.23312944e-02 9.57181931e-01 5.01080155e-01 -6.80228472e-01
6.53166592e-01 5.33523738e-01 3.46683949e-01 -6.00445211e-01
-1.63582668e-01 -5.05508125e-01 3.03929776e-01 8.12337875e-01
1.00056473e-02 -2.60909855e-01 2.08987772e-01 -2.90783823e-01
-5.99572241e-01 -8.56512189e-02 6.97925836e-02 9.13617849e-01
-3.38151038e-01 -1.08500230e+00 -4.82823342e-01 1.79601297e-01
-1.77315667e-01 -1.64255053e-01 -1.86705083e-01 1.02663589e+00
8.93171653e-02 3.31961215e-01 3.99743736e-01 -4.11586434e-01
6.82238817e-01 -4.31877151e-02 6.83697701e-01 -5.61637104e-01
-7.80160248e-01 -1.37219340e-01 -2.75829852e-01 -1.01898706e+00
-5.17990887e-01 -7.41336763e-01 -1.30239463e+00 -4.79161739e-01
-4.55646545e-01 -2.16551021e-01 1.18878198e+00 9.60247695e-01
7.59211421e-01 6.88183233e-02 1.51457414e-01 -1.84830344e+00
-1.55192062e-01 -7.21467793e-01 -1.07691415e-01 8.24593231e-02
4.88049299e-01 -9.27712739e-01 -1.88725293e-01 -3.75139505e-01]
|
[7.557592868804932, -2.522120475769043]
|
1af913e9-0b08-4fb1-8131-03b7b6b9e2aa
|
y-net-multi-scale-feature-aggregation-network
|
2003.13912
| null |
https://arxiv.org/abs/2003.13912v1
|
https://arxiv.org/pdf/2003.13912v1.pdf
|
Y-net: Multi-scale feature aggregation network with wavelet structure similarity loss function for single image dehazing
|
Single image dehazing is the ill-posed two-dimensional signal reconstruction problem. Recently, deep convolutional neural networks (CNN) have been successfully used in many computer vision problems. In this paper, we propose a Y-net that is named for its structure. This network reconstructs clear images by aggregating multi-scale features maps. Additionally, we propose a Wavelet Structure SIMilarity (W-SSIM) loss function in the training step. In the proposed loss function, discrete wavelet transforms are applied repeatedly to divide the image into differently sized patches with different frequencies and scales. The proposed loss function is the accumulation of SSIM loss of various patches with respective ratios. Extensive experimental results demonstrate that the proposed Y-net with the W-SSIM loss function restores high-quality clear images and outperforms state-of-the-art algorithms. Code and models are available at https://github.com/dectrfov/Y-net.
|
['Yi-Chang James Tsai', 'Chao-Han Huck Yang', 'Hao-Hsiang Yang']
|
2020-03-31
| null | null | null | null |
['wavelet-structure-similarity-loss']
|
['computer-vision']
|
[ 1.56444609e-01 -3.90300035e-01 4.07773316e-01 -3.20003569e-01
-6.23675227e-01 5.10001667e-02 2.17397869e-01 -5.22859395e-01
-1.89318955e-01 7.64448643e-01 1.99477062e-01 7.35551119e-02
-3.07615489e-01 -7.42144763e-01 -6.37239099e-01 -1.02173603e+00
-1.68286953e-02 -4.15010482e-01 1.23587005e-01 -1.56541139e-01
2.43755281e-01 4.68403250e-01 -1.46907866e+00 1.59787327e-01
1.06508541e+00 1.36438966e+00 4.65662122e-01 4.37792420e-01
7.80030265e-02 7.48672307e-01 -4.82008129e-01 -1.24500684e-01
5.25776386e-01 -5.55804968e-01 -3.52300435e-01 8.15425143e-02
5.00983238e-01 -4.18438345e-01 -5.82603276e-01 1.26616836e+00
7.32179761e-01 2.33442694e-01 4.54146147e-01 -1.09328067e+00
-9.29115593e-01 1.23244599e-01 -8.05194020e-01 4.84701246e-01
-1.99575573e-02 1.52072817e-01 6.55185223e-01 -1.16317022e+00
2.45728552e-01 1.19304979e+00 9.49969769e-01 2.40211532e-01
-1.24504566e+00 -7.95923948e-01 -1.45837769e-01 5.26121557e-01
-1.23707676e+00 -2.95104474e-01 9.80393529e-01 -1.65438414e-01
6.22057915e-01 2.51662016e-01 3.19762409e-01 7.02454150e-01
3.78914595e-01 7.39807308e-01 1.40706730e+00 -2.99900055e-01
-4.31836918e-02 -2.64042050e-01 4.40335013e-02 6.38408661e-01
2.45194122e-01 1.56846672e-01 -3.63147527e-01 2.89601475e-01
8.73376787e-01 3.28558207e-01 -8.87368262e-01 -1.88277841e-01
-1.03351712e+00 7.27600336e-01 7.48918951e-01 4.55413193e-01
-4.77694839e-01 1.85055077e-01 2.51538366e-01 5.30547380e-01
5.96049428e-01 2.18530059e-01 -3.88990670e-01 2.86594748e-01
-8.61652553e-01 1.69491470e-01 3.58643442e-01 3.77055407e-01
6.37174487e-01 2.82266021e-01 -7.42108375e-02 1.17129159e+00
-1.18925767e-02 2.70859838e-01 5.69495082e-01 -1.14658296e+00
1.76782280e-01 1.25560880e-01 5.30359857e-02 -1.18898082e+00
-2.12636128e-01 -7.24792778e-01 -1.46485901e+00 6.23754501e-01
-6.72581121e-02 -3.15830149e-02 -7.30799317e-01 1.49446821e+00
1.79914981e-01 8.14745665e-01 2.29814768e-01 1.14861548e+00
1.02499950e+00 8.90511096e-01 -3.17316383e-01 -2.33688682e-01
1.18561995e+00 -1.00676787e+00 -9.31080461e-01 1.74594030e-01
-1.99692652e-01 -8.77644777e-01 9.60929096e-01 6.58361912e-01
-1.23013055e+00 -8.35644186e-01 -1.13908517e+00 -1.07012421e-01
4.08346690e-02 3.76422405e-01 1.95392996e-01 2.67617404e-01
-1.07054722e+00 9.12056804e-01 -6.62012279e-01 9.54700410e-02
5.90530455e-01 6.16612248e-02 -3.41276288e-01 -2.53553301e-01
-1.02419126e+00 6.89941704e-01 -1.41048878e-02 5.27476728e-01
-9.24182951e-01 -8.59181166e-01 -7.25564301e-01 2.87836462e-01
1.57674745e-01 -6.13074720e-01 9.09425378e-01 -9.32700038e-01
-1.53634834e+00 5.68135381e-01 4.46820892e-02 -3.84816647e-01
3.65195751e-01 -3.78746033e-01 -5.85486829e-01 4.60448205e-01
1.25873238e-01 2.50512213e-01 1.28313410e+00 -1.51643968e+00
-5.11050403e-01 -2.40345716e-01 -1.80666417e-01 1.31374493e-01
-2.01487422e-01 -8.11119005e-02 -1.93146646e-01 -1.01379097e+00
2.88473785e-01 -4.36003089e-01 -4.94048893e-02 3.62518847e-01
-3.41502845e-01 1.15503505e-01 1.00405169e+00 -9.83039856e-01
9.24272597e-01 -2.36581945e+00 1.50232717e-01 -1.16328500e-01
5.32729089e-01 3.67851108e-01 -3.68695736e-01 2.10352242e-01
-3.01898777e-01 -1.98513225e-01 -7.30896950e-01 -4.46073741e-01
-1.86177894e-01 -1.56769976e-01 -4.08155799e-01 7.44940221e-01
1.89940333e-01 5.89906216e-01 -4.96370584e-01 -7.95542151e-02
4.31393594e-01 1.04979336e+00 -1.75065666e-01 3.08096975e-01
3.55076432e-01 3.70684057e-01 -1.60371765e-01 4.66100484e-01
1.32628798e+00 -2.03469291e-01 -2.25363180e-01 -3.87416601e-01
-2.03393489e-01 -2.64449805e-01 -1.13134742e+00 1.50350559e+00
-6.26158893e-01 8.86412144e-01 5.33929706e-01 -1.42216372e+00
1.05720401e+00 1.69149980e-01 4.26089853e-01 -8.57434869e-01
1.37149915e-01 3.12450886e-01 -2.75936544e-01 -6.45390034e-01
1.35982081e-01 -3.80343407e-01 4.93988961e-01 9.93015394e-02
6.87815249e-02 -2.09793553e-01 -1.84667513e-01 -2.69480497e-01
9.77190137e-01 -5.55127300e-02 -5.27050346e-02 -1.15513347e-01
9.52784956e-01 -4.95466113e-01 7.73832440e-01 3.14607650e-01
-2.21679136e-01 1.02591360e+00 8.57855901e-02 -5.58257461e-01
-9.66415584e-01 -1.11170065e+00 -3.39217126e-01 5.43493152e-01
4.81136948e-01 1.85363576e-01 -6.47992492e-01 -2.26695165e-01
-1.18107118e-01 3.63070250e-01 -4.95286077e-01 -2.32594088e-01
-6.85110748e-01 -6.05772316e-01 3.62760752e-01 1.73644915e-01
1.01123726e+00 -1.20649815e+00 -6.38104320e-01 2.05963820e-01
-3.42067719e-01 -9.10573840e-01 -4.81309414e-01 -9.24396440e-02
-9.35293257e-01 -1.02180302e+00 -1.24858749e+00 -1.14800382e+00
6.55928493e-01 7.45570898e-01 9.31350946e-01 1.40752405e-01
-5.25519013e-01 7.92160723e-03 -4.53115195e-01 -5.50712273e-02
5.18961623e-02 -4.41298276e-01 -1.25075892e-01 3.12659740e-01
-3.48090529e-02 -1.03402996e+00 -9.86427784e-01 3.17682236e-01
-1.14880383e+00 -7.85923079e-02 7.04627693e-01 1.17032719e+00
7.03774989e-01 4.99266714e-01 5.24510622e-01 -3.78562182e-01
8.27139616e-01 -3.68473202e-01 -6.09582186e-01 8.98656473e-02
-5.17540991e-01 -1.14029601e-01 7.99182355e-01 -4.20210660e-01
-1.16708779e+00 -3.30722272e-01 -2.28405997e-01 -7.92058229e-01
-1.27710745e-01 3.70642513e-01 2.88079251e-02 -2.99461573e-01
3.07602286e-01 7.27394819e-01 1.42392546e-01 -8.41486156e-01
1.42407283e-01 4.41260904e-01 8.20492983e-01 -2.48737544e-01
9.90171134e-01 6.21533513e-01 -5.53375259e-02 -9.04628277e-01
-7.35306323e-01 -2.88022369e-01 -8.17681924e-02 -1.44980803e-01
7.47457385e-01 -1.08989573e+00 -7.16733217e-01 8.95393908e-01
-1.08472037e+00 -1.67225793e-01 -2.76296437e-01 5.95232427e-01
-5.25621712e-01 5.42705119e-01 -6.42812788e-01 -5.42277038e-01
-6.23200536e-01 -1.08598208e+00 7.83235848e-01 4.90150928e-01
5.39793611e-01 -7.95330286e-01 -1.53521849e-02 2.06149727e-01
7.01221943e-01 4.92848337e-01 6.06352329e-01 1.15408033e-01
-5.01409650e-01 1.11653768e-01 -5.21788299e-01 9.44081068e-01
2.15025142e-01 -3.85439843e-01 -8.51657629e-01 -5.10317326e-01
5.40469050e-01 -3.65300745e-01 1.24085355e+00 9.61533785e-01
1.52299941e+00 -3.48810971e-01 6.65638000e-02 1.17204046e+00
1.83329153e+00 1.30717710e-01 8.84668708e-01 4.28017408e-01
3.53125542e-01 4.65044022e-01 2.85183311e-01 4.00865674e-01
-2.17308081e-03 4.75774348e-01 5.19569635e-01 -4.90989834e-01
-4.75299180e-01 1.81997806e-01 3.09597552e-01 9.05719817e-01
-7.98740536e-02 -1.20896161e-01 -4.14355218e-01 5.79985619e-01
-1.54349601e+00 -1.13788390e+00 -6.90196902e-02 1.89208949e+00
6.75794482e-01 -1.45486034e-02 -4.11668122e-01 3.40375304e-01
7.90561914e-01 4.21347409e-01 -5.77492833e-01 -4.42106370e-03
-5.40436864e-01 4.67777789e-01 2.51840919e-01 6.46941721e-01
-1.02063835e+00 4.34260428e-01 5.20303106e+00 1.07167685e+00
-1.41752386e+00 1.51584312e-01 7.02932298e-01 8.02263170e-02
-2.00055525e-01 -2.45291024e-01 -1.94561645e-01 6.12655222e-01
3.76159698e-01 -1.71966985e-01 5.47546327e-01 4.82857734e-01
3.93626571e-01 2.04084255e-02 -3.94054323e-01 1.17998350e+00
1.55361071e-01 -1.39227808e+00 -7.82347694e-02 -2.60835111e-01
6.76498175e-01 2.19121631e-02 2.95028061e-01 -4.62640263e-02
1.40002435e-02 -1.06101131e+00 6.61253631e-01 7.18132973e-01
9.12508726e-01 -6.53781831e-01 8.04587901e-01 2.37372257e-02
-1.24945545e+00 -2.48841658e-01 -4.12947685e-01 9.56120417e-02
1.80391490e-01 8.35060835e-01 -2.10977308e-02 8.04765284e-01
1.23459029e+00 1.02829647e+00 -1.77107260e-01 1.28888345e+00
-1.56665251e-01 5.48163176e-01 -7.29678571e-02 4.70461518e-01
2.20433131e-01 -5.85355461e-01 7.03259230e-01 9.50096726e-01
6.49292827e-01 3.27658951e-01 -1.66969866e-01 8.57908428e-01
-1.56275615e-01 -2.71247894e-01 -2.51142234e-01 4.15685236e-01
2.79443771e-01 1.22548056e+00 -4.33145016e-01 -2.03881279e-01
-2.68587589e-01 1.03673792e+00 -7.54749551e-02 5.16810060e-01
-5.76394677e-01 -8.66776764e-01 7.53903270e-01 3.43476385e-02
6.64064884e-01 -6.35077804e-02 -2.53365815e-01 -1.19589150e+00
2.95191824e-01 -7.86242843e-01 1.12679236e-01 -1.13072968e+00
-1.50911975e+00 8.94310296e-01 -3.15483749e-01 -1.62710857e+00
5.04509449e-01 -4.93722796e-01 -8.19187105e-01 9.73822117e-01
-2.26832104e+00 -8.66286457e-01 -7.96179771e-01 6.58578694e-01
7.77481377e-01 -3.37749869e-01 4.88808334e-01 4.80348706e-01
-5.28538883e-01 3.29043567e-01 5.91603160e-01 7.15039298e-02
4.94844168e-01 -8.97022009e-01 1.60511076e-01 9.15945232e-01
-3.04573715e-01 4.74438906e-01 9.15699065e-01 -2.35575169e-01
-1.05964828e+00 -1.04832613e+00 4.31445062e-01 4.76568192e-01
3.70204389e-01 -1.27702225e-02 -1.06445789e+00 3.52899045e-01
5.01802802e-01 4.03706610e-01 3.57511491e-01 -6.86403096e-01
-4.98515338e-01 -5.11526585e-01 -1.42272270e+00 1.11899048e-01
6.73582256e-01 -3.34503025e-01 -5.35045683e-01 1.20891348e-01
8.21716130e-01 -2.44941324e-01 -7.53590405e-01 3.85065228e-01
4.73421901e-01 -1.26364994e+00 1.28309751e+00 -1.17426597e-01
8.36321592e-01 -4.72492129e-01 -1.89009532e-01 -1.48120642e+00
-5.61177552e-01 -4.89478290e-01 -3.98738086e-02 8.14722657e-01
2.01670658e-02 -8.18665028e-01 3.58836025e-01 -1.92434579e-01
-3.34216535e-01 -9.95029986e-01 -1.06316972e+00 -8.21781576e-01
-8.05275217e-02 -2.08164919e-02 4.74799335e-01 9.84090984e-01
-6.68021619e-01 -2.19326913e-02 -5.10984242e-01 3.17340195e-01
1.08850193e+00 1.94271162e-01 4.43443507e-01 -1.17909503e+00
-1.85675561e-01 -5.33049762e-01 -3.05150539e-01 -9.30539489e-01
8.70266333e-02 -5.85157037e-01 4.36248742e-02 -1.72136116e+00
6.52099401e-02 -3.30832124e-01 -5.30828655e-01 3.42282057e-01
-1.12483621e-01 5.09730995e-01 2.73865342e-01 3.61244828e-01
-2.16614738e-01 1.11194754e+00 1.44908857e+00 -2.23336861e-01
-1.10430524e-01 -1.20189741e-01 -6.43188715e-01 5.17999053e-01
1.05892515e+00 -4.26543474e-01 -2.70019293e-01 -6.69553638e-01
-2.66515136e-01 1.15807377e-01 5.46316445e-01 -1.28847134e+00
1.31410658e-01 -2.20723636e-02 5.67552030e-01 -5.70564330e-01
6.38706863e-01 -8.59296501e-01 2.57682115e-01 6.19996965e-01
-2.59861678e-01 -1.39837712e-01 1.42614618e-01 4.57078159e-01
-7.48302877e-01 1.06041625e-01 1.23630869e+00 -1.63793996e-01
-5.42881966e-01 4.27546471e-01 3.26069444e-02 -2.26315618e-01
9.41558540e-01 -3.35558146e-01 -3.53331715e-01 -4.18909073e-01
-5.71985960e-01 1.26475021e-02 1.56009302e-01 2.94016749e-01
1.23006296e+00 -1.33579099e+00 -1.04588032e+00 2.50381261e-01
-2.04171613e-01 -7.23898560e-02 7.08520830e-01 8.83213758e-01
-7.75863051e-01 -7.06821159e-02 -6.43827379e-01 -2.33862877e-01
-1.24692178e+00 2.56125122e-01 5.80205321e-01 -1.09047152e-01
-1.15952933e+00 9.31740403e-01 2.84292251e-01 -3.66572857e-01
2.30485260e-01 -1.63145155e-01 -2.51459748e-01 -3.07625532e-01
7.83553064e-01 5.03214359e-01 -8.04527998e-02 -6.36812747e-01
-7.00398088e-02 8.28557253e-01 7.89785162e-02 3.02089769e-02
2.07374597e+00 -2.63361633e-01 -3.96984339e-01 -4.12327163e-02
1.50967157e+00 -1.59969985e-01 -1.54515028e+00 -4.52395856e-01
-5.17182469e-01 -7.33636975e-01 4.18661267e-01 -6.94365680e-01
-1.60833788e+00 9.50078547e-01 9.56058204e-01 1.16432950e-01
1.81874800e+00 -4.18511599e-01 1.31918705e+00 7.09454715e-02
-1.68445602e-01 -7.99963355e-01 2.71624029e-01 1.96849719e-01
1.27490091e+00 -1.07238066e+00 1.21576652e-01 -2.90442914e-01
-3.93084943e-01 1.13942254e+00 4.72196788e-01 -5.94715297e-01
9.37138855e-01 1.67037666e-01 1.99682668e-01 -2.57860303e-01
-3.83853912e-01 -1.27602126e-02 -1.12190051e-03 6.19367421e-01
1.43643841e-01 -7.99793899e-02 -4.84199047e-01 5.35027981e-01
-6.72398806e-02 2.25974485e-01 5.93752563e-01 6.04756832e-01
-5.44260561e-01 -6.78321242e-01 -5.95089972e-01 4.23396140e-01
-5.54045439e-01 -7.69679025e-02 2.22022191e-01 3.34847659e-01
1.88554108e-01 9.92051899e-01 -3.64112817e-02 -3.97017956e-01
2.77136326e-01 -5.01378119e-01 2.05875292e-01 -4.98998985e-02
-2.32689917e-01 8.96408707e-02 -4.17049795e-01 -4.79355514e-01
-5.34071922e-01 -3.41744572e-01 -1.11374962e+00 -2.34333277e-01
-6.39149994e-02 1.58734113e-01 5.81611812e-01 3.96717429e-01
2.94908643e-01 6.86873794e-01 1.13321638e+00 -9.91934776e-01
-5.94556630e-01 -9.89840865e-01 -8.65617573e-01 3.77404869e-01
1.02463579e+00 -6.42087102e-01 -7.19792604e-01 2.27550313e-01]
|
[11.12200927734375, -2.342129945755005]
|
ff7ae8db-ba27-4faa-93cb-746ad5363f5a
|
listening-to-sounds-of-silence-for-speech
|
2010.12013
| null |
https://arxiv.org/abs/2010.12013v1
|
https://arxiv.org/pdf/2010.12013v1.pdf
|
Listening to Sounds of Silence for Speech Denoising
|
We introduce a deep learning model for speech denoising, a long-standing challenge in audio analysis arising in numerous applications. Our approach is based on a key observation about human speech: there is often a short pause between each sentence or word. In a recorded speech signal, those pauses introduce a series of time periods during which only noise is present. We leverage these incidental silent intervals to learn a model for automatic speech denoising given only mono-channel audio. Detected silent intervals over time expose not just pure noise but its time-varying features, allowing the model to learn noise dynamics and suppress it from the speech signal. Experiments on multiple datasets confirm the pivotal role of silent interval detection for speech denoising, and our method outperforms several state-of-the-art denoising methods, including those that accept only audio input (like ours) and those that denoise based on audiovisual input (and hence require more information). We also show that our method enjoys excellent generalization properties, such as denoising spoken languages not seen during training.
|
['Changxi Zheng', 'Carl Vondrick', 'Yuko Ishiwaka', 'Rundi Wu', 'Ruilin Xu']
|
2020-10-22
| null |
http://proceedings.neurips.cc/paper/2020/hash/6d7d394c9d0c886e9247542e06ebb705-Abstract.html
|
http://proceedings.neurips.cc/paper/2020/file/6d7d394c9d0c886e9247542e06ebb705-Paper.pdf
|
neurips-2020-12
|
['speech-denoising']
|
['speech']
|
[ 5.50573945e-01 1.31122479e-02 2.10610136e-01 -2.98482090e-01
-1.20597494e+00 -7.00762272e-01 4.91217226e-01 1.39160857e-01
-5.49459338e-01 4.16870862e-01 4.45421100e-01 -1.54593036e-01
1.34710595e-02 -4.44695711e-01 -8.04716825e-01 -1.03064728e+00
-2.16025546e-01 -7.32975006e-02 2.11062491e-01 -4.59580064e-01
-2.23345965e-01 9.87742990e-02 -1.50470459e+00 3.62476975e-01
5.76008260e-01 1.04741561e+00 1.55256778e-01 9.30266619e-01
1.84150599e-02 7.41984546e-01 -1.08224237e+00 -9.14338306e-02
1.21217035e-01 -5.20128846e-01 -1.92888662e-01 1.27531692e-01
9.91179645e-02 -2.08408281e-01 -7.35140324e-01 1.06132376e+00
6.70195878e-01 4.76293504e-01 2.82653689e-01 -7.62366652e-01
-1.90138295e-01 9.14964259e-01 -2.59862542e-01 5.27071655e-01
3.29224795e-01 4.13147882e-02 9.64694798e-01 -6.51173294e-01
2.98388153e-01 1.18968010e+00 9.89107311e-01 4.86777931e-01
-1.39105999e+00 -5.95529437e-01 4.13785994e-01 1.34879291e-01
-1.08971310e+00 -1.06719363e+00 1.09877682e+00 -1.56991452e-01
8.57915461e-01 4.39300984e-01 5.88410139e-01 1.57620239e+00
1.11604638e-01 9.22314167e-01 8.00747335e-01 -4.04360712e-01
3.92292053e-01 -3.67018640e-01 1.73709080e-01 1.03951924e-01
-5.96929073e-01 8.19023773e-02 -9.15768802e-01 -1.16394475e-01
3.80011588e-01 -1.45651683e-01 -5.51297426e-01 3.14181328e-01
-1.09907854e+00 5.97481489e-01 -9.89519358e-02 3.17157686e-01
-4.55379993e-01 1.70093894e-01 6.78294659e-01 8.60857666e-01
8.11069965e-01 -6.72704875e-02 -4.40564930e-01 -4.55845207e-01
-1.27092171e+00 1.33736700e-01 9.01908815e-01 5.00851989e-01
4.50129658e-01 4.49043483e-01 2.11576000e-03 1.02265048e+00
-1.17456257e-01 5.39839685e-01 4.73425746e-01 -9.83789146e-01
3.81321371e-01 -4.33142066e-01 -1.04900010e-01 -7.94287801e-01
-3.11549485e-01 -6.41900957e-01 -1.30792427e+00 -6.25598654e-02
3.65225494e-01 -6.78510740e-02 -9.85104918e-01 2.15239692e+00
5.35764694e-02 5.45177639e-01 -4.65236679e-02 7.48700500e-01
6.91434085e-01 8.66874039e-01 -2.37327024e-01 -7.45600879e-01
1.04964316e+00 -5.78175068e-01 -1.23945689e+00 -2.89747268e-01
-1.92679241e-02 -8.03680539e-01 8.98013651e-01 1.17592978e+00
-1.22360945e+00 -6.40364885e-01 -1.02156293e+00 3.24269496e-02
-1.12758689e-01 -3.50502789e-01 3.30728114e-01 6.07068717e-01
-1.06140518e+00 7.55715132e-01 -1.02007627e+00 4.21562009e-02
1.40345469e-01 2.12446928e-01 -2.29867503e-01 2.69708753e-01
-1.31733322e+00 3.59584481e-01 -2.27462545e-01 4.41018522e-01
-1.23788643e+00 -7.38254130e-01 -8.73730183e-01 5.79952188e-02
5.60658038e-01 -2.92821229e-01 1.70097268e+00 -1.03463209e+00
-1.65263534e+00 6.53806686e-01 -7.05653787e-01 -7.80094504e-01
7.09442675e-01 -4.28324640e-01 -6.29179955e-01 3.62862647e-01
-7.95167312e-02 -4.69396412e-02 1.62391651e+00 -1.10923839e+00
-4.15043622e-01 -2.26141259e-01 -1.99613810e-01 -7.73550570e-02
-2.38416269e-01 -3.82089764e-02 -4.29211557e-01 -1.16726363e+00
3.14500839e-01 -5.24910092e-01 -2.12924853e-01 -2.83874810e-01
-5.29930174e-01 -9.07580480e-02 8.58260751e-01 -9.32546914e-01
1.42357862e+00 -2.32235622e+00 3.78620416e-01 1.46674782e-01
3.10598016e-01 2.06811696e-01 -2.07300022e-01 5.90258539e-01
-2.03806564e-01 1.02370761e-01 -4.64830726e-01 -9.64047313e-01
-7.24202720e-03 3.40664983e-01 -9.09222364e-01 5.93712509e-01
4.50996961e-03 4.88260180e-01 -8.96111369e-01 9.96751487e-02
2.21327677e-01 8.52487504e-01 -3.36971849e-01 2.13877231e-01
-6.28212690e-02 6.69807434e-01 5.79040684e-02 3.40663135e-01
5.99738061e-01 3.69080544e-01 -6.72209486e-02 -1.06264986e-01
-4.99312095e-02 7.54737735e-01 -1.22547936e+00 1.67810190e+00
-6.60746098e-01 8.94517601e-01 8.99396598e-01 -1.31298470e+00
6.41145110e-01 7.66440392e-01 3.32001835e-01 -6.02043808e-01
4.28023376e-02 1.05113789e-01 -1.94704667e-01 -5.16734362e-01
1.72815323e-01 -1.90779597e-01 5.00355288e-02 3.71186107e-01
1.98936209e-01 -3.25920999e-01 6.10086992e-02 2.20684350e-01
1.47014105e+00 -4.34094220e-01 8.10191110e-02 -2.82921605e-02
4.19594377e-01 -7.80571818e-01 6.18217707e-01 1.19114983e+00
-2.52576768e-01 8.85392427e-01 5.98113835e-01 -1.62266627e-01
-8.97644699e-01 -1.32819951e+00 1.37489349e-01 1.31395984e+00
-1.79038018e-01 -5.47930419e-01 -8.22349787e-01 -2.83136785e-01
-2.33469397e-01 3.97246957e-01 -5.74070990e-01 -2.72108257e-01
-7.76207507e-01 -4.18341994e-01 6.17321372e-01 3.70946050e-01
-2.56832633e-02 -1.00093925e+00 -1.77917331e-01 5.38307786e-01
-5.01768112e-01 -9.78381097e-01 -7.16644704e-01 8.48777056e-01
-7.64002144e-01 -6.52599871e-01 -6.09860539e-01 -6.80947959e-01
1.20726243e-01 3.79369706e-01 1.15014207e+00 3.00353747e-02
-8.80432278e-02 5.03777564e-01 -2.59185851e-01 -5.15047967e-01
-6.81925654e-01 -1.13177381e-01 2.31716067e-01 1.97227627e-01
1.30121291e-01 -1.21324933e+00 -4.74612236e-01 9.65423940e-04
-1.16228497e+00 -4.70410973e-01 2.18768388e-01 1.06794190e+00
5.97934783e-01 4.77862984e-01 9.21724916e-01 -6.00332856e-01
8.89937937e-01 -3.90045524e-01 -4.39222544e-01 -2.47541800e-01
4.70695160e-02 -2.62639493e-01 7.69652307e-01 -7.99879730e-01
-7.76162028e-01 -8.02052990e-02 -6.03203356e-01 -4.71667826e-01
-2.99225807e-01 5.04125655e-01 -2.97235698e-01 3.77411366e-01
4.85190094e-01 5.76763034e-01 6.82906508e-02 -7.87508249e-01
1.86752796e-01 6.94880188e-01 1.04103565e+00 -2.19055802e-01
8.04271758e-01 7.55261421e-01 -3.71776313e-01 -1.53253973e+00
-9.76270437e-01 -7.40141094e-01 -4.13142502e-01 -6.75773695e-02
5.30027568e-01 -1.02539563e+00 -7.05966473e-01 6.41603351e-01
-1.35800362e+00 -3.45416427e-01 -3.97404015e-01 3.29480469e-01
-5.27533948e-01 5.24880290e-01 -9.60072517e-01 -1.29165149e+00
-1.50988206e-01 -9.74435806e-01 1.27514851e+00 -2.09484026e-01
-3.35803658e-01 -9.45658565e-01 -4.34965920e-03 -5.89742744e-03
3.43727589e-01 -4.77729253e-02 6.29753768e-01 -5.26769757e-01
-2.65427709e-01 -1.21147208e-01 5.09348214e-01 8.13091099e-01
2.72747457e-01 -1.36978939e-01 -1.54744828e+00 -3.19666445e-01
6.53905690e-01 -7.46735036e-02 1.50303018e+00 7.85148382e-01
1.28745854e+00 -3.11595589e-01 1.03552714e-01 7.06109226e-01
7.65998483e-01 1.63795248e-01 5.64034164e-01 -3.51888910e-02
3.40384722e-01 4.96058404e-01 1.47312686e-01 3.94263357e-01
-1.10141188e-01 3.93929273e-01 5.29666483e-01 -1.98169425e-01
-1.62944142e-02 -1.55285478e-01 6.18793428e-01 1.37556314e+00
1.21853590e-01 -3.66090059e-01 -4.81542200e-01 8.17711771e-01
-1.69634080e+00 -1.14386022e+00 -7.41670877e-02 2.32946181e+00
1.22378182e+00 4.38825488e-01 1.47388980e-01 9.03832316e-01
5.31174183e-01 6.07167065e-01 -4.90340263e-01 -1.77882776e-01
-4.85613048e-01 7.30551958e-01 1.39597028e-01 8.66329312e-01
-1.48574972e+00 6.06317699e-01 6.65971279e+00 8.84168625e-01
-1.30240786e+00 1.33635119e-01 4.77149487e-01 -1.95627615e-01
-2.18857333e-01 -5.47601700e-01 -3.14062685e-01 3.72348517e-01
1.27876115e+00 -1.69920400e-01 7.35891402e-01 4.11740720e-01
8.43163669e-01 1.41337827e-01 -1.19897938e+00 1.05238581e+00
-1.38560548e-01 -1.02895057e+00 -2.79980004e-01 -2.69785166e-01
3.81995648e-01 -6.95093069e-03 3.02948505e-01 2.99764007e-01
-9.64145139e-02 -1.12258959e+00 1.00900340e+00 4.00379300e-01
4.22195941e-01 -8.02902579e-01 4.36565459e-01 4.12211597e-01
-1.17186248e+00 -1.00732163e-01 -2.41081625e-01 -2.58551419e-01
3.33306164e-01 1.15796995e+00 -3.94679606e-01 3.76885086e-01
7.69575477e-01 8.99609387e-01 -1.68534834e-02 8.26553881e-01
-5.26922226e-01 1.30246830e+00 -4.76753503e-01 4.89125222e-01
7.36273220e-03 -1.40924469e-01 1.08794332e+00 1.42519569e+00
2.84776121e-01 -4.98187058e-02 -9.35799852e-02 5.10183215e-01
-1.34830266e-01 -3.63426000e-01 -6.63677931e-01 2.39257403e-02
5.50613999e-01 9.08764422e-01 -5.10428250e-01 -2.52877265e-01
-3.67869705e-01 1.06855404e+00 -2.41745129e-01 7.16879666e-01
-6.96145236e-01 -3.88016939e-01 9.87067819e-01 7.37423748e-02
6.01488590e-01 -3.59294415e-01 -2.30784044e-01 -1.06704831e+00
3.17533612e-01 -1.18114495e+00 1.07138328e-01 -4.58762825e-01
-1.38752329e+00 6.47591591e-01 -4.64127690e-01 -1.16071010e+00
-4.61589068e-01 -2.96867311e-01 -6.92250788e-01 8.12482595e-01
-1.47865856e+00 -7.27115870e-01 -1.16383703e-02 6.67842031e-01
9.83557522e-01 1.43192708e-01 8.29139471e-01 3.94296974e-01
-2.78099805e-01 4.44626451e-01 2.69503713e-01 4.40585166e-02
8.41976821e-01 -1.39268696e+00 6.89773262e-01 9.60133255e-01
4.85920429e-01 7.63359904e-01 1.10272145e+00 -3.45444918e-01
-1.46110368e+00 -9.61452305e-01 7.52140880e-01 -3.80413115e-01
8.53406072e-01 -9.83273029e-01 -1.27519357e+00 5.37251770e-01
3.65506262e-01 -1.00250512e-01 4.98064071e-01 1.26662120e-01
-4.48356807e-01 -4.45330173e-01 -6.82729185e-01 3.75887543e-01
9.31547582e-01 -1.03778374e+00 -8.37457180e-01 2.13526309e-01
1.09455705e+00 -4.10410404e-01 -3.23214203e-01 2.37217583e-02
4.12199527e-01 -1.11503899e+00 1.13735223e+00 -4.65181589e-01
1.79379150e-01 -8.39207694e-02 -1.52733743e-01 -1.56318045e+00
1.56653881e-01 -1.52207136e+00 -4.06817794e-01 1.23434138e+00
2.92077810e-01 -5.24748802e-01 4.50637817e-01 -1.23143017e-01
-3.99943233e-01 -3.36266667e-01 -1.27229357e+00 -8.71981204e-01
-7.78014809e-02 -9.61772442e-01 1.83769464e-01 6.76696777e-01
-3.34488839e-01 3.21429431e-01 -5.36723077e-01 4.99901682e-01
8.29604208e-01 -3.13687772e-01 5.59886813e-01 -1.29505086e+00
-3.82554203e-01 -2.52821743e-01 -1.56997621e-01 -1.53133535e+00
2.40005553e-01 -3.62443089e-01 5.23670614e-01 -1.05928779e+00
-2.96387166e-01 1.97893351e-01 -4.23894376e-01 1.50691286e-01
-1.46494925e-01 1.47477254e-01 -1.63482755e-01 -9.01641846e-02
-5.53164721e-01 7.13335812e-01 7.88088322e-01 -3.39145839e-01
-3.80150884e-01 3.89210224e-01 -4.84520108e-01 9.53564644e-01
4.10767347e-01 -5.55594206e-01 -4.32747096e-01 -4.23218071e-01
1.24868210e-02 2.57994235e-01 5.08360028e-01 -8.44422638e-01
3.49990427e-01 3.38635445e-01 5.34959435e-02 -6.79731965e-01
7.45222390e-01 -6.95828795e-01 -3.83845828e-02 1.99628592e-01
-4.34114635e-01 -2.92576432e-01 1.86619356e-01 9.18893933e-01
-6.52295947e-01 8.45786929e-02 6.85108244e-01 -5.99241257e-02
-3.37079465e-01 6.30247518e-02 -8.75845015e-01 1.49883330e-01
2.02279434e-01 9.26205888e-02 -6.38070256e-02 -1.11605120e+00
-1.24637818e+00 -2.15492785e-01 -1.74933314e-01 3.81029308e-01
4.96805400e-01 -9.72939670e-01 -6.46562874e-01 3.49221081e-01
-4.79754746e-01 4.35931198e-02 3.95234644e-01 9.38375950e-01
1.10829554e-01 1.20416440e-01 5.66426933e-01 -9.03262436e-01
-1.30601549e+00 3.47149462e-01 2.17991456e-01 -3.51677276e-02
-9.70588565e-01 1.09982157e+00 3.58681083e-01 -1.87215775e-01
9.57519829e-01 -7.72144973e-01 1.17004879e-01 3.58316064e-01
9.16949153e-01 1.54505759e-01 5.02666593e-01 -3.21045965e-01
-1.68812051e-01 3.23297441e-01 7.75861740e-02 -5.19343793e-01
1.64911783e+00 -4.26203161e-01 -1.05086185e-01 1.10641682e+00
1.13788462e+00 4.09054458e-01 -1.42775106e+00 -3.73932689e-01
-1.11580931e-01 -1.86873615e-01 3.68484944e-01 -5.96954465e-01
-9.67244148e-01 1.05770731e+00 4.04072165e-01 9.01281953e-01
1.57784534e+00 -1.62455589e-01 9.85837400e-01 4.23362374e-01
8.76285583e-02 -1.12474775e+00 2.19424337e-01 7.32968688e-01
1.05953193e+00 -1.04575813e+00 -4.83370960e-01 -3.58106822e-01
-2.27264583e-01 1.10573649e+00 -1.74733832e-01 -1.27302064e-02
8.54197085e-01 7.14279771e-01 4.42216307e-01 2.22065732e-01
-9.06335115e-01 -1.74373373e-01 -4.23065573e-02 8.34872663e-01
2.74507165e-01 -2.49356031e-01 1.04006588e-01 1.01100826e+00
-4.94256705e-01 -4.06481802e-01 5.59490263e-01 8.90067101e-01
-4.49169934e-01 -9.03710961e-01 -5.09997189e-01 1.57266304e-01
-7.06258953e-01 -2.92433411e-01 -4.71090972e-01 4.39955413e-01
-1.97446808e-01 1.43205392e+00 9.64840800e-02 -2.60957360e-01
4.53555197e-01 1.49320230e-01 2.14391381e-01 -4.55572486e-01
-5.97262859e-01 6.72775984e-01 3.02335843e-02 -6.41170084e-01
-3.01253289e-01 -6.91426158e-01 -9.46652591e-01 -2.43777975e-01
-1.55612797e-01 3.06656837e-01 7.16051877e-01 9.31662977e-01
-5.95286824e-02 1.04690146e+00 7.79478848e-01 -1.05513251e+00
-7.20025539e-01 -1.07816529e+00 -7.84449518e-01 2.90143281e-01
1.48833442e+00 -2.80985624e-01 -1.07323766e+00 5.38215518e-01]
|
[15.141100883483887, 5.810478210449219]
|
b0e0d881-1841-4b2a-a04a-5dffe3a8e54e
|
inferturbo-a-scalable-system-for-boosting
|
2307.00228
| null |
https://arxiv.org/abs/2307.00228v1
|
https://arxiv.org/pdf/2307.00228v1.pdf
|
InferTurbo: A Scalable System for Boosting Full-graph Inference of Graph Neural Network over Huge Graphs
|
GNN inference is a non-trivial task, especially in industrial scenarios with giant graphs, given three main challenges, i.e., scalability tailored for full-graph inference on huge graphs, inconsistency caused by stochastic acceleration strategies (e.g., sampling), and the serious redundant computation issue. To address the above challenges, we propose a scalable system named InferTurbo to boost the GNN inference tasks in industrial scenarios. Inspired by the philosophy of ``think-like-a-vertex", a GAS-like (Gather-Apply-Scatter) schema is proposed to describe the computation paradigm and data flow of GNN inference. The computation of GNNs is expressed in an iteration manner, in which a vertex would gather messages via in-edges and update its state information by forwarding an associated layer of GNNs with those messages and then send the updated information to other vertexes via out-edges. Following the schema, the proposed InferTurbo can be built with alternative backends (e.g., batch processing system or graph computing system). Moreover, InferTurbo introduces several strategies like shadow-nodes and partial-gather to handle nodes with large degrees for better load balancing. With InferTurbo, GNN inference can be hierarchically conducted over the full graph without sampling and redundant computation. Experimental results demonstrate that our system is robust and efficient for inference tasks over graphs containing some hub nodes with many adjacent edges. Meanwhile, the system gains a remarkable performance compared with the traditional inference pipeline, and it can finish a GNN inference task over a graph with tens of billions of nodes and hundreds of billions of edges within 2 hours.
|
['Jun Zhou', 'Zhiqiang Zhang', 'Lin Wang', 'Binbin Hu', 'Miao Tao', 'Yang Li', 'Zhiyang Hu', 'Xianzheng Song', 'Dalong Zhang']
|
2023-07-01
| null | null | null | null |
['philosophy']
|
['miscellaneous']
|
[-1.79790616e-01 9.27110761e-02 -1.36534736e-01 -1.75015181e-01
4.31565475e-03 -4.09774601e-01 2.02401340e-01 5.63123152e-02
1.53744295e-01 6.17854714e-01 -6.04395747e-01 -5.38098633e-01
-3.40261638e-01 -1.59985864e+00 -5.85595608e-01 -6.78665459e-01
-3.76349390e-01 1.03933299e+00 5.85555077e-01 -1.56727508e-01
-1.27139404e-01 3.07130933e-01 -1.36547875e+00 -1.65408745e-01
7.24037290e-01 9.94149268e-01 2.30443999e-01 6.34234905e-01
-2.18820706e-01 8.59296918e-01 -5.67805946e-01 -6.64197803e-01
2.25059003e-01 -2.46608943e-01 -7.80833662e-01 -3.95842269e-02
-1.25062922e-02 -3.72016788e-01 -4.01967466e-01 1.13558006e+00
5.24842322e-01 -9.21460800e-03 -1.88093692e-01 -1.72865164e+00
8.43143836e-02 1.09828532e+00 -8.12344491e-01 -2.93513872e-02
3.64203304e-01 5.21317601e-01 6.68472409e-01 -4.92886871e-01
5.32496691e-01 1.56011057e+00 7.24457145e-01 2.73749195e-02
-1.02638137e+00 -8.97906363e-01 3.69964004e-01 -1.16717659e-01
-1.47170949e+00 -2.22987339e-01 5.76842070e-01 -1.30568799e-02
7.77522862e-01 5.30480683e-01 8.30819607e-01 6.44445956e-01
3.91543806e-01 4.32978481e-01 5.21538258e-01 1.91129163e-01
4.55479205e-01 -4.00668591e-01 9.84335169e-02 7.99961388e-01
7.29946256e-01 -1.87450200e-01 -5.31549156e-01 -3.76628935e-01
9.26993608e-01 2.32713014e-01 2.11074531e-01 -1.95478853e-02
-1.23212159e+00 5.09994388e-01 7.03158677e-01 -1.86966196e-01
-6.10721767e-01 7.96591997e-01 9.61792469e-01 1.44238308e-01
4.14988011e-01 -2.12583378e-01 -6.21821642e-01 -6.11505434e-02
-8.76570582e-01 4.22967196e-01 1.35308623e+00 1.54099095e+00
1.21419966e+00 6.38413280e-02 -1.62626177e-01 2.78525621e-01
3.05001616e-01 5.92470288e-01 -2.27373123e-01 -6.70181274e-01
5.78381836e-01 7.61218727e-01 -4.81585830e-01 -1.35965562e+00
-4.20036942e-01 -3.86488408e-01 -1.65763915e+00 -1.35079026e-01
7.97187269e-04 -4.33323413e-01 -7.27614522e-01 1.40590310e+00
9.65933144e-01 4.81483102e-01 -3.30006421e-01 8.24596524e-01
7.51502991e-01 6.95194125e-01 5.99872926e-03 -3.36435825e-01
1.71605444e+00 -1.07580543e+00 -7.58066177e-01 -2.59701014e-01
5.02663493e-01 -6.31329656e-01 6.72957659e-01 4.17414844e-01
-1.02921546e+00 -6.13295674e-01 -6.07454658e-01 -2.56627295e-02
-3.30266058e-01 -2.54278421e-01 1.36017048e+00 4.51145113e-01
-1.08113134e+00 5.68311870e-01 -9.33591545e-01 -1.79612353e-01
1.10000677e-01 3.37583035e-01 9.24222842e-02 -2.59297013e-01
-1.04141617e+00 -2.22421065e-02 8.35313916e-01 7.18821228e-01
-1.14712346e+00 -7.39417851e-01 -7.79637098e-01 2.58436263e-01
1.06582403e+00 -1.31502199e+00 1.03325379e+00 -2.70092785e-01
-1.41796684e+00 1.14902295e-01 -1.50940642e-01 -1.22942269e-01
3.33172441e-01 3.22357595e-01 -4.41862285e-01 -2.97547191e-01
4.52192575e-02 -1.41542226e-01 5.22275627e-01 -7.81645477e-01
-5.93277752e-01 -7.16884971e-01 2.60038793e-01 -1.19594872e-01
8.60106423e-02 -1.42741248e-01 -9.82293069e-01 -2.05712274e-01
2.84865528e-01 -1.00238824e+00 -6.77369952e-01 -3.88719380e-01
-9.96818066e-01 -5.51795959e-01 8.01528215e-01 -2.48224512e-01
1.66672421e+00 -1.71437657e+00 -7.64954239e-02 7.62421906e-01
9.11008835e-01 5.83038889e-02 3.47496778e-01 8.06448638e-01
3.24657887e-01 -3.50215212e-02 5.56461364e-02 -2.56047640e-02
1.86255366e-01 4.06183243e-01 -9.12106968e-03 1.96877658e-01
-2.49648407e-01 1.00510693e+00 -1.36435950e+00 -8.77673268e-01
1.97475597e-01 3.09661403e-02 -4.74951982e-01 9.68530029e-02
-4.86362636e-01 8.18771049e-02 -8.28163326e-01 7.98550904e-01
1.27963877e+00 -7.40507960e-01 8.41957629e-01 -3.84004533e-01
6.40300363e-02 3.40171486e-01 -1.52883697e+00 1.48139977e+00
-5.28181612e-01 -1.23717412e-01 7.46771514e-01 -6.86552703e-01
9.37089443e-01 -1.04804516e-01 1.26717448e-01 -1.53916061e-01
1.48733318e-01 7.65561908e-02 -6.32498562e-02 -3.26751620e-01
5.07362127e-01 2.08353743e-01 -4.03752178e-01 3.75581324e-01
-2.28329018e-01 -2.69275606e-01 7.12849140e-01 6.37468040e-01
1.64717269e+00 -1.22440029e-02 3.16837020e-02 -3.02311987e-01
4.57492143e-01 -2.32506245e-02 8.68653536e-01 6.70148730e-01
2.42305592e-01 -3.12697917e-01 8.36853027e-01 -6.61635876e-01
-5.28185010e-01 -8.37912619e-01 4.04787302e-01 1.07535875e+00
4.10333544e-01 -1.22755814e+00 -6.14905775e-01 -6.98776722e-01
2.48604953e-01 4.02118772e-01 -1.46315977e-01 -4.50283550e-02
-5.01596570e-01 -8.28127027e-01 3.01719218e-01 4.48958844e-01
7.39791572e-01 -9.79723990e-01 -8.90401974e-02 4.45161194e-01
-6.09407835e-02 -1.15814555e+00 -2.54009545e-01 -7.02211261e-02
-9.37187731e-01 -1.26273096e+00 2.03664184e-01 -6.69003189e-01
8.41414690e-01 2.05475450e-01 1.54489899e+00 4.57659304e-01
-2.16321766e-01 -2.81544328e-01 -1.09619014e-02 -2.20275566e-01
-1.33199647e-01 2.67931610e-01 -1.33626848e-01 -3.47429782e-01
1.84568495e-01 -1.06064641e+00 -5.10430753e-01 3.17431092e-01
-8.53665471e-01 3.05213511e-01 6.52022004e-01 3.85207236e-01
7.16868877e-01 8.06310534e-01 3.16543907e-01 -1.42242610e+00
6.78334832e-01 -7.62170494e-01 -1.14020133e+00 9.75900143e-02
-7.12836444e-01 -1.73770070e-01 1.06277192e+00 -2.14273959e-01
-8.03567410e-01 -1.08005321e-02 -3.76001261e-02 -4.22531486e-01
4.54634845e-01 7.76469290e-01 -2.98846275e-01 1.08742438e-01
1.72945499e-01 -7.32086040e-03 -1.60049368e-02 -2.80894637e-01
3.19526643e-01 2.74396449e-01 2.70140499e-01 -8.52199376e-01
8.41278195e-01 3.93014997e-01 5.90986729e-01 -3.41619641e-01
-5.43559611e-01 -2.14188069e-01 -1.38578162e-01 -3.90908420e-01
2.94180602e-01 -9.07287717e-01 -1.82901967e+00 6.75598145e-01
-1.20984328e+00 -4.85742033e-01 4.81963605e-02 1.90578163e-01
-6.74054623e-02 2.66345561e-01 -1.14196181e+00 -8.27207685e-01
-7.05311716e-01 -1.19869900e+00 1.18720794e+00 1.42852440e-01
7.92202726e-02 -9.24622297e-01 -2.64990568e-01 2.37338594e-03
5.69259107e-01 4.19396460e-01 6.76887333e-01 -2.81572878e-01
-1.18421102e+00 -1.12404563e-01 -5.36428630e-01 -1.09459914e-01
9.07996073e-02 2.55753100e-01 -3.04984659e-01 -4.75517541e-01
-3.36446494e-01 -1.73396707e-01 3.72226387e-01 -4.47375663e-02
1.71673870e+00 -4.64262635e-01 -7.90918708e-01 8.32208455e-01
1.52158320e+00 -3.41067255e-01 5.13583124e-01 -2.24836618e-01
1.02919483e+00 5.73504865e-02 6.44370198e-01 5.76223135e-01
6.67784810e-01 2.25703076e-01 8.63081813e-01 -3.24063823e-02
3.26781794e-02 -4.24291790e-01 9.38276127e-02 1.37185979e+00
-1.80545688e-01 -6.01749122e-01 -6.58318043e-01 2.07932517e-01
-2.09776926e+00 -6.08682096e-01 -6.34818017e-01 1.98577118e+00
6.36703968e-01 4.18728769e-01 3.49010788e-02 -6.65058494e-02
9.73732352e-01 1.44397467e-01 -8.58829141e-01 -2.79705465e-01
5.51191509e-01 9.14491937e-02 6.45749211e-01 1.52584061e-01
-3.30025911e-01 9.13497388e-01 5.50353289e+00 1.15834081e+00
-6.43640459e-01 -9.75699276e-02 4.22608435e-01 3.31652582e-01
-2.45466143e-01 4.57304180e-01 -9.82914507e-01 9.44961011e-01
1.15043664e+00 -4.74406838e-01 8.62549841e-01 1.10871017e+00
3.94385718e-02 -1.54118329e-01 -9.01876926e-01 9.99817729e-01
-3.72173488e-01 -1.27565145e+00 6.75536245e-02 3.72650027e-01
5.94633937e-01 -4.81969342e-02 -6.86223388e-01 6.08744085e-01
1.18268335e+00 -5.58276236e-01 1.85636684e-01 3.21268171e-01
6.80888057e-01 -9.66455281e-01 7.77933776e-01 5.80451250e-01
-1.84185731e+00 1.52136043e-01 -6.42815292e-01 -3.13829780e-01
3.81799966e-01 1.57910120e+00 -9.16753113e-01 1.36350429e+00
8.09179187e-01 4.13989484e-01 -1.56631351e-01 6.25961781e-01
-2.15570763e-01 5.34487784e-01 -7.91885197e-01 -3.65145266e-01
-2.83800135e-03 -6.51819885e-01 3.24614674e-01 9.06300843e-01
2.79460490e-01 2.00013462e-02 7.00213969e-01 1.07505929e+00
-2.41257891e-01 -2.67358601e-01 -4.01053876e-01 1.10492297e-01
9.76490021e-01 1.90384996e+00 -9.57522690e-01 -7.62513459e-01
-9.06704664e-02 7.80575275e-01 3.99916708e-01 2.96609551e-01
-1.12902236e+00 -8.16734016e-01 5.48498452e-01 4.32868510e-01
2.15612710e-01 -1.42768979e-01 1.13930307e-01 -6.61712289e-01
1.21016279e-01 -8.77564192e-01 5.92374682e-01 -8.43793035e-01
-1.42211747e+00 6.18645728e-01 -3.42918150e-02 -7.00760961e-01
-2.17836335e-01 -3.19205880e-01 -8.07522178e-01 7.12332606e-01
-1.04119909e+00 -1.16919148e+00 -8.19503784e-01 8.07184935e-01
-6.02840260e-03 4.43427771e-01 4.26422924e-01 4.53964233e-01
-7.78531373e-01 2.67677844e-01 -4.33316469e-01 -2.94943783e-03
2.54197091e-01 -1.12010610e+00 7.42461503e-01 8.13720167e-01
-6.31829739e-01 8.89836609e-01 5.17038465e-01 -1.16116500e+00
-2.25546241e+00 -1.31139350e+00 5.45714080e-01 -3.03862263e-02
8.70960891e-01 -7.55413055e-01 -4.54665333e-01 1.06249833e+00
-1.51364878e-01 4.95147258e-01 8.47191215e-02 5.75772762e-01
7.65178427e-02 -7.52216458e-01 -1.07921124e+00 5.22931397e-01
1.44162774e+00 -7.26275295e-02 3.44233513e-01 8.08346152e-01
1.13569999e+00 -9.60367382e-01 -1.14915526e+00 4.29042965e-01
2.05368660e-02 -7.31918156e-01 8.93018842e-01 -2.96307743e-01
1.30493656e-01 -7.81400561e-01 3.96519899e-01 -1.01241124e+00
-4.49668974e-01 -1.17317259e+00 -4.70357627e-01 1.48707831e+00
7.43091106e-02 -1.03256357e+00 8.42175901e-01 2.78475493e-01
-2.02900201e-01 -7.11482584e-01 -6.75925672e-01 -6.25188887e-01
-8.50881457e-01 -4.03518051e-01 1.36485159e+00 6.14279091e-01
-1.29513785e-01 9.15495574e-01 -1.50431111e-01 4.83741581e-01
9.16789174e-01 4.70440447e-01 1.57524431e+00 -1.34635401e+00
-4.08585697e-01 1.30478218e-01 -2.83859491e-01 -1.24601233e+00
2.11255755e-02 -1.16329074e+00 -7.73644745e-02 -1.74242222e+00
2.49448225e-01 -7.71652758e-01 1.22105882e-01 5.73293328e-01
-1.75302953e-01 -2.55670309e-01 -4.72171195e-02 -3.05777881e-02
-1.07900739e+00 1.36334509e-01 1.51681983e+00 1.21950760e-01
8.60591382e-02 2.15127826e-01 -4.47454721e-01 5.72453737e-01
6.49934471e-01 -3.71210515e-01 -6.22106433e-01 -3.54784966e-01
9.03540134e-01 4.53916967e-01 5.86946189e-01 -8.58732343e-01
6.75802946e-01 -5.04877977e-02 1.95209846e-01 -1.07314825e+00
1.00161754e-01 -8.62048030e-01 8.74204159e-01 6.25787079e-01
5.30446470e-01 6.32535219e-01 1.14934243e-01 6.96420670e-01
-9.59196091e-02 2.82830566e-01 5.38967401e-02 -3.71715635e-01
-4.64912236e-01 7.69590735e-01 5.61365345e-03 1.32442400e-01
9.98026967e-01 1.63333379e-02 -6.00566328e-01 -2.07460716e-01
-5.04696131e-01 9.88268912e-01 2.96102792e-01 7.50674307e-03
2.95944124e-01 -1.32718945e+00 -6.21235073e-01 1.98363706e-01
-2.65025526e-01 1.01389325e+00 4.65508044e-01 1.01985514e+00
-7.78901756e-01 -6.48184717e-02 3.62883657e-01 -6.10492706e-01
-9.25134182e-01 8.44150543e-01 6.40492234e-03 -9.98260140e-01
-5.52356184e-01 7.70915627e-01 6.82241097e-02 -9.07559693e-01
-1.68185532e-01 -4.58499819e-01 7.55843222e-01 -4.49574977e-01
3.67346495e-01 9.84856188e-01 1.90462768e-01 2.11035252e-01
-4.59439218e-01 1.05371088e-01 7.93396756e-02 6.67617857e-01
1.12488508e+00 -2.23947749e-01 -1.02817822e+00 1.52490914e-01
6.90131128e-01 1.32514983e-01 -6.44060254e-01 -5.41460961e-02
-3.99412334e-01 -4.97293115e-01 6.28416985e-02 -5.59119463e-01
-1.58334637e+00 2.74252176e-01 -3.88784170e-01 7.02797055e-01
1.18048704e+00 -1.38798216e-02 1.22136366e+00 2.72882104e-01
9.03809607e-01 -8.49965572e-01 -3.63120466e-01 3.82248819e-01
2.41834983e-01 -7.63548195e-01 3.17103446e-01 -1.20874524e+00
-3.29570174e-02 1.04592931e+00 9.19690430e-01 -7.78410733e-02
8.76641214e-01 6.14120781e-01 -7.01459527e-01 -8.25124264e-01
-1.04989481e+00 2.50391513e-02 -5.61368644e-01 3.39268684e-01
-5.33271953e-02 3.86799663e-01 -2.22653478e-01 6.06437325e-01
-3.50940615e-01 3.48937243e-01 1.63735077e-01 6.81759357e-01
-3.70395184e-03 -1.04952669e+00 -5.87424338e-02 7.51176596e-01
-1.79227903e-01 -1.23808883e-01 4.08511348e-02 9.35026467e-01
1.47743061e-01 9.36340153e-01 3.17803144e-01 -5.34086704e-01
2.74837643e-01 -6.44613385e-01 1.29528582e-01 -5.13875484e-01
-7.69940495e-01 -6.71745613e-02 4.06715155e-01 -9.78768468e-01
2.32605681e-01 -1.65049151e-01 -1.41371512e+00 -1.26418960e+00
-5.71712852e-01 2.60440767e-01 6.28921211e-01 4.12773907e-01
7.28896260e-01 1.17173493e+00 5.66749096e-01 -7.34288812e-01
-4.90022749e-01 -8.39588523e-01 -8.44826460e-01 -4.90261242e-02
-1.71461612e-01 -2.53722638e-01 -4.25729066e-01 -4.42452371e-01]
|
[7.021779537200928, 5.846648216247559]
|
bdb2d339-498e-415c-8948-cd90ba5b660f
|
efficient-transformer-based-speech
|
2206.11703
| null |
https://arxiv.org/abs/2206.11703v1
|
https://arxiv.org/pdf/2206.11703v1.pdf
|
Efficient Transformer-based Speech Enhancement Using Long Frames and STFT Magnitudes
|
The SepFormer architecture shows very good results in speech separation. Like other learned-encoder models, it uses short frames, as they have been shown to obtain better performance in these cases. This results in a large number of frames at the input, which is problematic; since the SepFormer is transformer-based, its computational complexity drastically increases with longer sequences. In this paper, we employ the SepFormer in a speech enhancement task and show that by replacing the learned-encoder features with a magnitude short-time Fourier transform (STFT) representation, we can use long frames without compromising perceptual enhancement performance. We obtained equivalent quality and intelligibility evaluation scores while reducing the number of operations by a factor of approximately 8 for a 10-second utterance.
|
['Timo Gerkmann', 'Tal Peer', 'Danilo de Oliveira']
|
2022-06-23
| null | null | null | null |
['speech-separation']
|
['speech']
|
[ 3.06994826e-01 -5.80277182e-02 3.15560460e-01 -2.01446503e-01
-1.11431575e+00 -2.74552047e-01 4.13679391e-01 5.29143326e-02
-5.63676238e-01 6.52072430e-01 4.95549262e-01 -4.96130258e-01
-4.42072153e-02 -3.78077745e-01 -3.16663653e-01 -8.39664459e-01
-7.88136572e-02 -3.93857688e-01 2.54116505e-01 2.10769325e-02
-6.48138523e-02 2.27838337e-01 -1.59267437e+00 4.03901935e-01
6.97885156e-01 9.59813058e-01 5.48593760e-01 1.09769225e+00
2.65007257e-01 8.46064448e-01 -9.62121964e-01 -4.69224036e-01
2.10597754e-01 -7.05830157e-01 -5.32246530e-01 1.99711636e-01
4.70329285e-01 -5.73575258e-01 -6.06590986e-01 9.13993716e-01
7.91854262e-01 4.31556910e-01 4.02469695e-01 -7.68197656e-01
-2.56486446e-01 3.42876196e-01 -2.78804868e-01 4.21175182e-01
3.82839292e-01 -2.98144463e-02 9.82158482e-01 -9.83034074e-01
2.17016190e-01 1.17557514e+00 8.02528381e-01 2.08635509e-01
-1.31999612e+00 -4.75485116e-01 -3.02259684e-01 2.98100293e-01
-9.98929262e-01 -1.04556072e+00 5.95497429e-01 -4.28241119e-02
1.46668148e+00 3.73710871e-01 4.51347440e-01 8.65303934e-01
1.58334777e-01 6.01115763e-01 1.04974711e+00 -6.07896090e-01
-1.99451130e-02 -6.12428412e-02 -1.09780550e-01 2.48313606e-01
-2.31398627e-01 5.91044366e-01 -5.64798832e-01 2.58498073e-01
6.60250366e-01 -3.26195776e-01 -5.51661670e-01 1.43156469e-01
-9.70047414e-01 6.21457219e-01 1.39844179e-01 6.08293772e-01
-3.99131268e-01 1.72545373e-01 4.79118139e-01 7.49098718e-01
5.15564501e-01 4.27209169e-01 -2.85894096e-01 -5.09677112e-01
-1.36484134e+00 -1.93563811e-02 5.50640762e-01 5.73654711e-01
1.90487579e-01 6.17476702e-01 -6.18088134e-02 1.18025446e+00
2.68401783e-02 5.41736424e-01 3.79851967e-01 -1.20537233e+00
3.91580105e-01 -4.40839887e-01 6.08803444e-02 -1.05201757e+00
-3.20790261e-01 -7.68682837e-01 -6.28899992e-01 4.50955153e-01
4.36863959e-01 -2.94523776e-01 -8.34332705e-01 1.58435512e+00
-2.91957557e-01 6.03072681e-02 2.48962402e-01 9.34860349e-01
2.07738891e-01 9.59525943e-01 -4.96340618e-02 -5.87135196e-01
1.31024480e+00 -8.94567490e-01 -1.24035966e+00 -3.05606663e-01
1.14760764e-01 -1.29530883e+00 8.30005825e-01 7.50371575e-01
-1.45943141e+00 -7.91508496e-01 -1.18763471e+00 5.02025485e-02
-1.33603858e-02 2.24590808e-01 3.45223993e-01 9.20769334e-01
-1.30485165e+00 8.52705419e-01 -7.33080328e-01 -1.87982380e-01
-4.19973116e-03 3.04331958e-01 -3.13748658e-01 1.51786417e-01
-1.07821012e+00 1.07213902e+00 1.64308026e-01 -1.79130808e-01
-6.41804278e-01 -4.39355165e-01 -9.04750168e-01 5.76883435e-01
1.70999244e-01 -1.93636939e-01 1.48061395e+00 -7.72283316e-01
-1.62542140e+00 1.54594511e-01 -5.00683606e-01 -7.10494399e-01
1.37761414e-01 -3.29141408e-01 -8.83175135e-01 4.30148810e-01
-4.33954060e-01 4.49602157e-01 1.36744440e+00 -9.00755405e-01
-5.15242577e-01 7.27327690e-02 2.16816738e-02 1.76771387e-01
-6.19850278e-01 2.72553623e-01 -1.40607119e-01 -1.17075956e+00
1.61880136e-01 -6.18179142e-01 3.79687510e-02 -1.84852198e-01
2.95053571e-01 3.96198720e-01 8.69149566e-01 -1.12268519e+00
1.38542485e+00 -2.58114576e+00 -1.63631514e-01 -1.03123993e-01
7.82681033e-02 7.11391032e-01 -2.73220301e-01 4.27854449e-01
-2.77051151e-01 -9.83837172e-02 -7.75113925e-02 -4.12493467e-01
-1.17899410e-01 8.50146711e-02 -2.33912945e-01 2.80436397e-01
2.35573575e-01 2.60138005e-01 -7.86641061e-01 -2.39273965e-01
5.53931177e-01 9.72297668e-01 -7.35208452e-01 2.46757761e-01
4.82777208e-01 5.03906868e-02 4.22817498e-01 -7.98083143e-04
5.55539906e-01 1.97103649e-01 6.61532432e-02 -4.29349929e-01
-5.87387718e-02 7.00796723e-01 -1.01937973e+00 1.56484318e+00
-9.54755247e-01 1.43421566e+00 2.27876380e-01 -6.73545003e-01
8.24520588e-01 1.02032089e+00 4.24452096e-01 -8.75178039e-01
8.43975544e-02 2.27665737e-01 3.43815297e-01 -3.56358975e-01
3.67316842e-01 -5.81497729e-01 5.31726301e-01 2.18637392e-01
3.47275794e-01 -3.56927335e-01 3.31726521e-01 4.10637893e-02
1.11352122e+00 -3.69810849e-01 1.02008767e-01 3.33210430e-03
4.98449206e-01 -6.74662530e-01 3.26181978e-01 4.60744470e-01
-2.27528661e-01 7.61136353e-01 1.83942616e-01 1.04538634e-01
-1.03115880e+00 -1.31576025e+00 -1.80527672e-01 9.39814866e-01
-2.13213131e-01 -7.65715718e-01 -8.07490408e-01 -1.44540459e-01
-5.20575881e-01 9.29657400e-01 1.20426841e-01 -1.58382162e-01
-5.45635521e-01 -3.22829843e-01 5.56421041e-01 6.37291849e-01
2.38083884e-01 -7.73756802e-01 -7.64978886e-01 6.25784278e-01
-3.94800514e-01 -1.20546877e+00 -6.19718730e-01 5.41392386e-01
-7.33629882e-01 -7.32268393e-01 -9.11793649e-01 -7.16918111e-01
4.57689822e-01 2.65209079e-01 7.72472501e-01 -3.81418377e-01
-4.54019979e-02 1.08489506e-01 -4.98104155e-01 -1.57630816e-01
-4.59893346e-01 -5.37777543e-01 1.02301307e-01 -2.08394244e-01
1.58711478e-01 -6.96270168e-01 -3.51959318e-01 1.85250059e-01
-6.67838156e-01 -1.79197527e-02 5.51361144e-01 1.27773595e+00
-3.34170344e-03 6.75981641e-01 5.50611496e-01 -1.55890673e-01
8.96400332e-01 2.01760441e-01 -5.43016315e-01 -2.24217549e-01
-6.14585757e-01 -6.41424283e-02 7.31908441e-01 -3.67851764e-01
-1.28672564e+00 -3.40715319e-01 -5.83924294e-01 -4.57426548e-01
1.10003734e-02 4.25862223e-01 1.39840350e-01 -6.05360493e-02
5.02755880e-01 2.49968216e-01 1.16239652e-01 -6.73460484e-01
1.14700675e-01 9.85473514e-01 6.02925420e-01 -2.59683225e-02
7.05323279e-01 -1.49081454e-01 -2.55828023e-01 -9.67299104e-01
-4.33431506e-01 -3.43839914e-01 -9.69716012e-02 -1.85950875e-01
6.51104510e-01 -9.68648195e-01 -4.52207118e-01 2.89102137e-01
-1.09859157e+00 -6.22377209e-02 -3.77539843e-01 9.18605983e-01
-5.01015365e-01 3.58438075e-01 -9.74939704e-01 -1.06320500e+00
-1.69360295e-01 -1.29240108e+00 7.59835482e-01 -1.10708480e-03
-3.02806765e-01 -7.93451309e-01 -3.68515283e-01 5.08557111e-02
8.91531885e-01 -3.81840974e-01 7.32990086e-01 -3.04010987e-01
-2.72941235e-02 -7.30752945e-02 -6.83623031e-02 9.11082149e-01
3.67800802e-01 -3.08445841e-01 -1.28289306e+00 -5.47767222e-01
5.67471027e-01 -7.18856975e-02 8.38880956e-01 5.10741115e-01
9.51499820e-01 -2.51049727e-01 1.28728017e-01 4.55480248e-01
1.13550150e+00 7.25059509e-01 7.58831322e-01 -2.42823482e-01
9.09778550e-02 4.35999662e-01 5.38459182e-01 3.22893351e-01
-3.35112125e-01 7.82075703e-01 -1.40296862e-01 -4.28072214e-01
-8.05399179e-01 -1.57078639e-01 6.35258973e-01 1.20822775e+00
-8.36344585e-02 -4.35047209e-01 -5.90024114e-01 4.86520499e-01
-1.24704385e+00 -1.25298607e+00 2.24633127e-01 2.06972933e+00
7.02752292e-01 2.45554879e-01 6.93426356e-02 9.41499650e-01
4.55413222e-01 3.39789480e-01 -7.30522051e-02 -6.99634254e-01
-1.06881537e-01 6.41723514e-01 3.93834472e-01 8.21724117e-01
-8.33141983e-01 3.33787680e-01 7.54490137e+00 8.31112564e-01
-1.29410374e+00 2.83639789e-01 4.30716693e-01 -4.49034840e-01
1.25281066e-02 -2.55519986e-01 -1.40766785e-01 4.65010166e-01
1.44157445e+00 -3.01676631e-01 5.51516235e-01 4.07455206e-01
4.46228832e-01 -1.51005179e-01 -8.65474939e-01 1.33913434e+00
5.83599582e-02 -8.77387404e-01 -3.88251662e-01 -1.94931272e-02
2.93570876e-01 -2.08387539e-01 2.12150007e-01 1.35414442e-02
-2.07304746e-01 -1.05289960e+00 7.44041681e-01 2.68594623e-02
9.89614844e-01 -8.66023362e-01 6.66043282e-01 8.38718638e-02
-1.19292068e+00 -2.60717213e-01 -4.40987706e-01 -1.29911751e-01
3.40842515e-01 8.46263051e-01 -7.67042160e-01 3.19954783e-01
6.18052125e-01 3.59706134e-01 -1.57867283e-01 1.19460142e+00
-1.45660043e-01 8.04309607e-01 -2.44746268e-01 2.94369251e-01
1.04261726e-01 -1.86830051e-02 7.68684030e-01 1.44389796e+00
6.22371614e-01 1.13790803e-01 -3.70674729e-01 3.55660439e-01
2.01224416e-01 -2.32416555e-01 -4.57759917e-01 -6.33466318e-02
3.00042480e-01 9.08374608e-01 -2.72276253e-01 -1.60197705e-01
-5.28033018e-01 9.92351353e-01 -2.37380937e-01 5.09316087e-01
-7.69811988e-01 -8.92191768e-01 5.52524745e-01 4.98875342e-02
5.97650647e-01 -3.26980293e-01 1.20811373e-01 -7.91006327e-01
-5.37972078e-02 -1.13231075e+00 -4.03622501e-02 -8.33066940e-01
-8.15623164e-01 8.97013843e-01 -1.80387259e-01 -1.32977474e+00
-6.51877344e-01 -6.38894677e-01 -2.47327477e-01 1.12922561e+00
-1.33451879e+00 -4.67748016e-01 9.49639156e-02 4.43824053e-01
8.34556937e-01 -9.02845860e-02 9.41503286e-01 7.02188492e-01
-1.74094886e-01 7.69988298e-01 2.24347278e-01 -2.80392133e-02
6.69101298e-01 -1.16941381e+00 4.00435597e-01 1.23018491e+00
2.96321422e-01 5.72812200e-01 9.97390032e-01 -1.87212467e-01
-1.21497858e+00 -6.32768452e-01 1.09974122e+00 2.39959583e-01
4.08902228e-01 -2.85430461e-01 -7.99805284e-01 3.00371021e-01
6.49336576e-01 -1.67689264e-01 6.44254148e-01 1.12832859e-01
-2.62278497e-01 -1.78291097e-01 -1.05122209e+00 4.85132545e-01
8.46082389e-01 -8.69366586e-01 -6.30643547e-01 -5.86447343e-02
6.51029229e-01 -4.04007643e-01 -8.41490567e-01 1.92346498e-01
5.24580598e-01 -1.29758978e+00 9.96628344e-01 3.60729359e-02
1.86712846e-01 -3.48359257e-01 -2.54207671e-01 -1.67585039e+00
-5.15564919e-01 -8.89597595e-01 -2.72596031e-01 1.17654645e+00
5.08438230e-01 -5.58254719e-01 3.68639737e-01 1.92388758e-01
-1.72732189e-01 -5.22516489e-01 -1.00302505e+00 -1.13583386e+00
-3.12701583e-01 -5.57409585e-01 2.93698996e-01 3.64562899e-01
5.09234488e-01 4.43818152e-01 -6.22487187e-01 -4.09265533e-02
3.59990716e-01 -3.75354052e-01 1.07625462e-01 -6.77035928e-01
-6.58345461e-01 -3.90485823e-01 -4.84619051e-01 -1.23845196e+00
-9.56656262e-02 -4.44083571e-01 3.00684452e-01 -1.13067257e+00
-3.06879431e-01 -4.91964743e-02 -5.09122670e-01 2.74468154e-01
-1.07640862e-01 1.45812556e-01 4.78005379e-01 -3.97763699e-01
5.53970449e-02 4.76334274e-01 8.96365464e-01 -2.71971703e-01
-1.82982162e-02 -1.24019951e-01 -3.62202823e-01 6.05410814e-01
6.50547862e-01 -3.41690361e-01 -7.43382394e-01 -4.76161003e-01
-4.53611851e-01 4.82451618e-01 1.09560356e-01 -1.54293275e+00
6.33953214e-02 3.57125610e-01 5.21919250e-01 -3.95153046e-01
9.27976072e-01 -9.81492519e-01 3.06223392e-01 5.02299488e-01
-4.21448886e-01 1.36645868e-01 4.80996639e-01 2.27833614e-01
-7.09011793e-01 -2.42326766e-01 9.63655889e-01 2.76896834e-01
-3.45388323e-01 -3.48109752e-01 -6.76379740e-01 -3.63891304e-01
6.11949682e-01 -3.40665638e-01 2.82363896e-03 -8.33269477e-01
-8.41211736e-01 -4.67941493e-01 1.79219559e-01 1.72823206e-01
7.61092722e-01 -1.08985448e+00 -5.83889186e-01 4.13383216e-01
-4.49117839e-01 -7.30715156e-01 1.19305290e-01 9.07429636e-01
-2.14740500e-01 5.05557835e-01 -2.38367915e-01 -3.99738014e-01
-1.63083208e+00 3.97591591e-01 4.30556655e-01 -1.00390583e-01
-7.42861807e-01 8.15997243e-01 -4.40918095e-02 5.87872744e-01
4.10914719e-01 -1.59880921e-01 -3.15444022e-02 2.05952227e-02
8.28887761e-01 5.40800273e-01 3.34933639e-01 -5.77591479e-01
-1.72080770e-01 3.79306197e-01 2.56954413e-02 -5.96706986e-01
1.21494889e+00 -2.38781795e-01 4.19990152e-01 1.57263383e-01
1.44314897e+00 2.71238923e-01 -1.21002793e+00 -2.67478246e-02
-2.73770303e-01 -8.99488091e-01 6.00265086e-01 -7.79363573e-01
-1.11627603e+00 1.03724849e+00 9.24935281e-01 2.90163159e-01
1.80130529e+00 -4.31871444e-01 9.46769834e-01 1.14786528e-01
2.85806477e-01 -9.64116812e-01 4.89707887e-02 5.21918476e-01
8.65632415e-01 -8.60905170e-01 -1.40922859e-01 -4.06341136e-01
-2.75176764e-01 1.24800658e+00 1.45550430e-01 2.59658277e-01
5.35992146e-01 7.84943759e-01 1.73189446e-01 3.03295553e-01
-7.01261163e-01 -3.06786150e-01 1.77423298e-01 8.26285899e-01
7.11727500e-01 -5.08713648e-02 -3.23382020e-01 2.03790799e-01
-5.38528323e-01 -2.41669893e-01 4.35241967e-01 6.82616234e-01
-4.90005404e-01 -1.29124367e+00 -6.05066419e-01 5.01049221e-01
-6.67404771e-01 -3.54963988e-01 1.71575814e-01 2.95142263e-01
-8.14092010e-02 1.57806456e+00 2.29225636e-01 -6.05533540e-01
3.47095191e-01 1.26525894e-01 6.70613468e-01 -3.27940315e-01
-6.24824047e-01 5.92566311e-01 4.20351714e-01 -5.11756599e-01
-3.62761676e-01 -5.78115761e-01 -9.68488991e-01 -3.54201883e-01
-5.28271258e-01 1.94567382e-01 6.23833895e-01 7.00454235e-01
1.56012982e-01 9.50925112e-01 6.11010015e-01 -7.56858349e-01
-4.29700226e-01 -1.02865314e+00 -5.74894428e-01 3.58936578e-01
7.87734449e-01 -4.77508634e-01 -4.46788132e-01 3.34394425e-01]
|
[15.147876739501953, 5.953787803649902]
|
6ccc7787-a210-4dd6-b6f1-df2c145a31f8
|
semeval-2021-task-7-hahackathon-detecting-and
| null | null |
https://aclanthology.org/2021.semeval-1.9
|
https://aclanthology.org/2021.semeval-1.9.pdf
|
SemEval 2021 Task 7: HaHackathon, Detecting and Rating Humor and Offense
|
SemEval 2021 Task 7, HaHackathon, was the first shared task to combine the previously separate domains of humor detection and offense detection. We collected 10,000 texts from Twitter and the Kaggle Short Jokes dataset, and had each annotated for humor and offense by 20 annotators aged 18-70. Our subtasks were binary humor detection, prediction of humor and offense ratings, and a novel controversy task: to predict if the variance in the humor ratings was higher than a specific threshold. The subtasks attracted 36-58 submissions, with most of the participants choosing to use pre-trained language models. Many of the highest performing teams also implemented additional optimization techniques, including task-adaptive training and adversarial training. The results suggest that the participating systems are well suited to humor detection, but that humor controversy is a more challenging task. We discuss which models excel in this task, which auxiliary techniques boost their performance, and analyze the errors which were not captured by the best systems.
|
['Walid Magdy', 'Adam Lopez', 'Luis Chiruzzo', 'Steven Wilson', 'J. A. Meaney']
|
2021-08-01
| null | null | null |
semeval-2021
|
['humor-detection']
|
['natural-language-processing']
|
[-3.64026278e-01 1.23370670e-01 5.54173216e-02 1.07863313e-02
-6.20445311e-01 -6.28265977e-01 7.59774566e-01 2.25093573e-01
-3.64157081e-01 8.44727576e-01 6.97875857e-01 -2.57952243e-01
5.02451360e-01 -4.85971004e-01 -1.92032039e-01 -2.21757933e-01
2.31319457e-01 7.04396546e-01 1.38091400e-01 -7.38997221e-01
7.93193996e-01 3.70948873e-02 -1.18349743e+00 6.54902458e-01
8.91059339e-01 4.78800863e-01 -4.25437003e-01 1.00951266e+00
2.12716460e-01 1.93109477e+00 -9.17710960e-01 -8.50333571e-01
-7.89996702e-03 -5.44729888e-01 -1.06304812e+00 -2.93887109e-01
5.56219220e-01 -7.14526400e-02 -2.99166769e-01 7.99458802e-01
6.50197387e-01 3.19341928e-01 5.78718245e-01 -1.27451110e+00
-8.23452115e-01 9.43466127e-01 -3.33701164e-01 4.57615852e-01
7.34284699e-01 3.65828902e-01 1.19604993e+00 -1.26740861e+00
6.37142301e-01 1.18273723e+00 9.73934174e-01 5.80786645e-01
-1.11543822e+00 -6.40804410e-01 -7.46261358e-01 6.24294698e-01
-9.82259512e-01 -5.77700615e-01 8.10673892e-01 -1.06357539e+00
1.13908637e+00 2.52413571e-01 3.72520268e-01 1.20918608e+00
1.42162889e-01 7.76116371e-01 1.27542341e+00 -2.25712344e-01
4.50671352e-02 6.15195811e-01 4.93075341e-01 7.40988493e-01
-8.59884918e-02 -4.27288681e-01 -8.54677081e-01 -6.55722022e-01
1.65709783e-03 -5.55335462e-01 -3.75608802e-01 3.83900195e-01
-1.08368266e+00 1.24622750e+00 4.17483151e-01 3.58605981e-01
-1.55196220e-01 -2.79585540e-01 8.58719528e-01 4.62814689e-01
5.08316696e-01 1.17488968e+00 -1.58012882e-02 -2.90720671e-01
-1.28510463e+00 7.86688387e-01 1.38963425e+00 5.37066519e-01
3.89346778e-01 -2.18515173e-02 -3.88220429e-01 1.03833163e+00
-1.46542028e-01 3.09278280e-01 5.32490730e-01 -1.01144207e+00
5.47094166e-01 6.03040993e-01 3.92552793e-01 -1.37277150e+00
-7.07901716e-01 -2.46366233e-01 -4.53660786e-01 2.79723465e-01
6.76260889e-01 -1.79058269e-01 -3.16909850e-01 1.29147637e+00
-2.18895704e-01 -4.30971444e-01 -4.01032299e-01 1.06297958e+00
1.10384572e+00 5.35071850e-01 -1.38865829e-01 -2.28478745e-01
1.23613560e+00 -1.34066021e+00 -6.02905214e-01 -4.84719783e-01
6.42042160e-01 -1.09556818e+00 1.41593587e+00 5.46655297e-01
-1.50262463e+00 -1.69158787e-01 -1.08676147e+00 -6.20758474e-01
-3.42603922e-01 -1.45476237e-01 1.41258866e-01 4.49053824e-01
-6.38269424e-01 6.43975079e-01 -6.77407235e-02 -1.95759028e-01
2.94042796e-01 -1.84538290e-01 -6.00998960e-02 3.94790530e-01
-1.64307749e+00 1.62790871e+00 5.21697812e-02 -4.53241527e-01
-8.77273619e-01 -5.28309464e-01 -6.24002814e-01 -8.11519939e-03
-8.13243166e-02 -4.28493083e-01 1.38401473e+00 -9.51250017e-01
-1.08538735e+00 1.49163651e+00 6.57380745e-02 -7.06013501e-01
7.30270147e-01 -2.41709158e-01 -2.39222199e-01 -1.22464769e-01
4.25022155e-01 -3.38451155e-02 7.36582935e-01 -9.16510165e-01
-1.53559133e-01 -1.17225885e-01 -2.57498175e-01 1.12868361e-01
-5.35382092e-01 6.04772210e-01 3.86984527e-01 -4.17037427e-01
-5.77173829e-01 -8.04084241e-01 8.12921673e-02 -6.68345451e-01
-4.19937640e-01 -3.38083982e-01 5.06254196e-01 -1.12709820e+00
1.69104898e+00 -1.76650357e+00 1.45238757e-01 -9.56235379e-02
5.89613795e-01 2.73417056e-01 8.27744454e-02 5.53417444e-01
1.78016782e-01 1.46361589e-01 -7.47899637e-02 -3.72422457e-01
1.53722614e-01 -3.79462272e-01 -5.97070456e-01 4.06638503e-01
-2.12874651e-01 8.98553252e-01 -9.91462529e-01 -4.83267397e-01
-1.44424289e-01 1.77932605e-02 -3.73693526e-01 4.20847416e-01
-1.57296985e-01 -4.49412689e-02 1.55894548e-01 5.97952843e-01
2.62298048e-01 -3.12883675e-01 -2.73710310e-01 2.63938993e-01
-4.46829870e-02 8.76922011e-01 -5.04464567e-01 7.09969103e-01
-2.82719284e-01 1.19239163e+00 7.93832093e-02 -2.01998115e-01
1.18245947e+00 1.66739851e-01 2.12019950e-01 -3.37167561e-01
3.28861088e-01 4.10004646e-01 -3.71522345e-02 -6.04768097e-01
1.06387532e+00 -5.94787598e-01 -4.10272956e-01 5.75211823e-01
-2.19630152e-01 -4.51286644e-01 3.26363951e-01 5.56156397e-01
1.40141177e+00 -6.19667590e-01 5.20861208e-01 -3.10872167e-01
6.27670944e-01 3.73786926e-01 3.61461133e-01 9.26888347e-01
-6.92952991e-01 6.88469529e-01 8.71620715e-01 -6.15727782e-01
-1.41217756e+00 -7.09176898e-01 -2.56831199e-02 1.60970116e+00
-2.70658523e-01 -5.98110437e-01 -6.41348004e-01 -5.36578178e-01
1.10503033e-01 1.21097124e+00 -6.35966003e-01 9.55171604e-03
-5.31935751e-01 -8.44486952e-01 8.96294057e-01 2.51944274e-01
3.06580007e-01 -1.53781283e+00 -6.59638524e-01 3.47451866e-01
-5.89065135e-01 -8.22970569e-01 -6.35830164e-01 2.75105715e-01
-3.10473353e-01 -1.18849826e+00 -4.56719130e-01 -6.22467339e-01
-1.53110484e-02 1.00104347e-01 1.58323097e+00 4.12813485e-01
-1.85840577e-01 2.86319163e-02 -3.88319135e-01 -2.93562233e-01
-7.88965881e-01 2.27454320e-01 -1.05095498e-01 -5.87300777e-01
8.32985938e-01 -4.58410919e-01 -1.99112505e-01 2.79271960e-01
-2.98027217e-01 -1.28766177e-02 -1.37183249e-01 1.16956329e+00
-5.34079909e-01 -5.11930048e-01 7.30131865e-01 -9.91827905e-01
1.29161012e+00 -7.34354198e-01 1.55986294e-01 -2.49620765e-01
-7.30358124e-01 -5.20013273e-01 9.47653651e-01 -3.53772104e-01
-7.00539589e-01 -2.54302651e-01 4.85788211e-02 -7.59388208e-02
2.63161480e-01 4.32399988e-01 3.12561899e-01 1.51898086e-01
1.70819187e+00 -2.06028581e-01 3.70871164e-02 -2.66334742e-01
3.40265751e-01 9.19941604e-01 7.91357696e-01 -3.78746003e-01
7.52699792e-01 -1.49466693e-02 -4.27152067e-01 -6.32938147e-01
-1.27193904e+00 -7.60118484e-01 -5.37792385e-01 -1.93673924e-01
6.43002450e-01 -9.15861070e-01 -6.74920559e-01 5.37187934e-01
-1.61719692e+00 -4.34839547e-01 -9.94137600e-02 -1.21753849e-01
-5.67181528e-01 3.81600738e-01 -1.13048518e+00 -9.52959239e-01
-8.71330202e-01 -6.32767320e-01 1.92904770e-01 1.75562605e-01
-9.02958333e-01 -8.41756821e-01 5.25233448e-01 1.08525610e+00
4.56555307e-01 1.62338480e-01 8.53592694e-01 -1.04951811e+00
2.88435727e-01 -3.84465545e-01 -2.51942843e-01 3.68209064e-01
-5.34607589e-01 -1.10640779e-01 -9.12196517e-01 -8.48460346e-02
1.39586806e-01 -1.15537453e+00 1.05054212e+00 -1.80540383e-01
6.70585096e-01 -7.29014337e-01 1.74526289e-01 2.88715046e-02
6.62487805e-01 -3.29551458e-01 7.23324537e-01 8.18441153e-01
5.68942785e-01 6.12199783e-01 1.82945907e-01 8.19155514e-01
5.62403619e-01 5.00890911e-01 3.68086278e-01 2.86535829e-01
2.97820836e-04 -4.59069043e-01 8.66146326e-01 8.31516802e-01
2.98681036e-02 -4.38037179e-02 -1.16599095e+00 8.37494671e-01
-1.89562190e+00 -1.55824435e+00 -9.22363281e-01 1.82449460e+00
1.17552960e+00 3.00419658e-01 6.95254862e-01 9.96305719e-02
6.08477771e-01 4.89697158e-01 -2.40834117e-01 -8.94077122e-01
-4.00076210e-01 -2.79993452e-02 1.50884241e-01 9.80447590e-01
-9.45915520e-01 1.11456513e+00 7.00890732e+00 5.44281244e-01
-6.70240581e-01 3.93763423e-01 5.42060375e-01 -4.70361531e-01
-2.93350190e-01 -2.02835854e-02 -5.93222797e-01 5.87349415e-01
8.34482312e-01 -4.04593527e-01 7.34991014e-01 1.12546432e+00
3.14927876e-01 -4.76877205e-02 -8.19766164e-01 7.94240415e-01
7.08914399e-01 -1.09270883e+00 -4.84770775e-01 -3.17134321e-01
9.09900427e-01 2.11511105e-01 -2.44881213e-01 8.07948828e-01
4.79872853e-01 -1.47110808e+00 7.94089794e-01 5.52778363e-01
1.17771104e-01 -3.99137020e-01 8.19349647e-01 6.07341111e-01
-2.50009805e-01 -2.78536737e-01 -3.73502702e-01 -7.20834851e-01
3.03249687e-01 7.92973340e-01 -1.16856861e+00 -5.40152848e-01
4.03458059e-01 6.38780773e-01 -1.02528763e+00 7.84154117e-01
-5.84616065e-01 9.22820330e-01 -3.41289267e-02 -5.20336330e-01
1.83687508e-02 3.57993335e-01 8.85749876e-01 1.75699174e+00
-1.49698123e-01 -4.99064066e-02 1.37796119e-01 1.15434158e+00
-3.74526083e-01 2.42153183e-01 -4.10817981e-01 -3.01285274e-02
5.54218531e-01 1.64309037e+00 -1.56616315e-01 -3.60657662e-01
-4.64398004e-02 8.98974895e-01 6.80015147e-01 -1.88014686e-01
-8.08925867e-01 -4.63796914e-01 1.55869588e-01 5.42856157e-01
-2.92863131e-01 -8.21626335e-02 -1.11621213e+00 -1.29119921e+00
-3.48004669e-01 -1.31267369e+00 6.00454152e-01 -9.45584893e-01
-1.73725867e+00 5.00432611e-01 -5.70290387e-01 -5.16377151e-01
-2.71247178e-01 -6.19321525e-01 -1.13941443e+00 9.14293110e-01
-8.83183122e-01 -1.03924072e+00 -4.87006336e-01 3.97431046e-01
4.95659053e-01 -4.27178532e-01 3.92147034e-01 1.47817001e-01
-3.56037617e-01 4.95684445e-01 -1.85091555e-01 2.66375452e-01
1.18894756e+00 -1.47635067e+00 8.87598321e-02 3.83543819e-01
-1.69594020e-01 4.93525177e-01 1.33511353e+00 -6.95985615e-01
-6.65069997e-01 -6.14006698e-01 1.64658320e+00 -1.18965590e+00
1.24495256e+00 -2.66508132e-01 -1.13665020e+00 5.26005208e-01
4.75549161e-01 -5.05899251e-01 5.83187282e-01 4.86230195e-01
-7.60891080e-01 5.49797654e-01 -1.03140950e+00 3.23306352e-01
5.80817401e-01 -8.17396343e-01 -1.03206253e+00 6.66316092e-01
1.27375185e-01 -4.02358711e-01 -5.85164368e-01 -5.84331974e-02
4.83551949e-01 -1.19312727e+00 4.75219280e-01 -1.01390326e+00
1.46780610e+00 -3.26297171e-02 -2.21788175e-02 -1.21535385e+00
-7.42335975e-01 -6.27496362e-01 -2.19882578e-01 1.16060865e+00
4.54238385e-01 -2.95395292e-02 6.70521557e-01 6.77662432e-01
7.37176165e-02 -5.31099737e-01 -6.60905540e-01 -4.10727859e-01
7.74459898e-01 -9.35060903e-02 -6.34340569e-02 1.30298352e+00
9.40266013e-01 1.09364665e+00 -9.41249967e-01 -5.28172493e-01
4.73531544e-01 -3.03517189e-02 9.31202173e-01 -9.76563573e-01
-3.32870424e-01 -9.55184102e-01 -1.67544976e-01 -6.55587971e-01
4.10490811e-01 -1.17705047e+00 1.88564643e-01 -1.20569813e+00
8.74489069e-01 2.46429697e-01 -2.34550517e-02 6.55858934e-01
-5.05878091e-01 4.23752904e-01 4.93375421e-01 4.21159327e-01
-7.48003721e-01 4.99491215e-01 8.05927098e-01 -2.24354789e-01
-2.68610656e-01 -2.82952338e-01 -9.02146280e-01 7.95355141e-01
7.80030549e-01 -3.87134582e-01 1.96626529e-01 1.46518394e-01
6.78655148e-01 -4.95882966e-02 9.03472662e-01 -7.88515329e-01
4.20443803e-01 -1.55181631e-01 2.41003811e-01 -5.70660293e-01
6.64977431e-02 -2.18079183e-02 -2.84804165e-01 2.14504823e-01
-6.65426850e-01 -1.12354457e-01 -4.62787375e-02 -6.99322000e-02
7.49208033e-02 -5.78616679e-01 1.22036421e+00 -2.89596289e-01
-1.79392874e-01 -5.75626016e-01 -7.59443283e-01 6.21211827e-01
7.19482183e-01 1.79090172e-01 -9.79038835e-01 -7.67677963e-01
-6.43895328e-01 2.43748054e-01 5.89617014e-01 4.38037395e-01
3.27522814e-01 -1.13531637e+00 -1.24453783e+00 -4.63899046e-01
-2.67874338e-02 -7.59003758e-01 -1.02462053e-01 1.00320637e+00
-3.20261270e-01 7.22668916e-02 -2.93535620e-01 1.17191792e-01
-1.35081494e+00 4.92026538e-01 5.02815187e-01 -4.96265948e-01
-4.24509376e-01 8.23188186e-01 -4.19848979e-01 -5.59802771e-01
-1.02745630e-01 8.14850569e-01 -2.82438487e-01 3.35551322e-01
7.03450143e-01 9.85351264e-01 -7.47169135e-03 -7.78320312e-01
-2.57392794e-01 -2.36100584e-01 -2.56713837e-01 -1.85213938e-01
1.12918544e+00 3.38638164e-02 -4.92548853e-01 7.29504824e-01
8.89388442e-01 3.31522524e-01 -5.15565693e-01 -6.78058118e-02
2.23284379e-01 -3.92522901e-01 2.68545374e-02 -1.36926937e+00
-2.76623487e-01 8.79884064e-01 -1.92397952e-01 5.98278224e-01
5.43495774e-01 -4.07174304e-02 9.38576937e-01 2.89621919e-01
1.50530532e-01 -1.41933703e+00 5.36282659e-01 1.27590585e+00
1.31460702e+00 -1.22476923e+00 2.01011375e-01 7.41501600e-02
-1.01053584e+00 1.22226000e+00 9.74274695e-01 -2.99157798e-01
5.29698096e-02 3.52469534e-02 1.64336562e-02 -4.51518804e-01
-1.00540960e+00 1.86037466e-01 3.99689466e-01 3.55682895e-02
9.58731830e-01 7.18484074e-02 -8.04071069e-01 1.06595290e+00
-7.76194096e-01 -1.43097505e-01 1.02897167e+00 2.72930354e-01
-9.73782122e-01 -3.58484298e-01 -6.22262776e-01 6.04925513e-01
-4.57034528e-01 -2.76771396e-01 -1.33348930e+00 4.40309614e-01
-3.82356346e-02 1.20292187e+00 -4.92290556e-01 -8.25213850e-01
3.07529777e-01 3.41517150e-01 9.87425372e-02 -5.19624889e-01
-1.37123168e+00 -4.71266896e-01 7.16040194e-01 -1.88017160e-01
2.35076711e-01 -7.28964508e-01 -1.06195009e+00 -1.02026272e+00
-2.98372835e-01 2.84771591e-01 1.08675718e-01 9.14599180e-01
-1.87522173e-02 -8.81426260e-02 7.60711789e-01 -6.70796096e-01
-9.95042384e-01 -1.28911924e+00 -3.83920133e-01 8.11294913e-01
1.40896529e-01 -2.82986611e-01 -9.84079480e-01 1.99249201e-02]
|
[8.876900672912598, 11.075698852539062]
|
7d7c8261-234f-43e1-bea6-c7342b47c0e6
|
distributed-deep-reinforcement-learning-learn
|
1801.02852
| null |
http://arxiv.org/abs/1801.02852v2
|
http://arxiv.org/pdf/1801.02852v2.pdf
|
Distributed Deep Reinforcement Learning: Learn how to play Atari games in 21 minutes
|
We present a study in Distributed Deep Reinforcement Learning (DDRL) focused
on scalability of a state-of-the-art Deep Reinforcement Learning algorithm
known as Batch Asynchronous Advantage ActorCritic (BA3C). We show that using
the Adam optimization algorithm with a batch size of up to 2048 is a viable
choice for carrying out large scale machine learning computations. This,
combined with careful reexamination of the optimizer's hyperparameters, using
synchronous training on the node level (while keeping the local, single node
part of the algorithm asynchronous) and minimizing the memory footprint of the
model, allowed us to achieve linear scaling for up to 64 CPU nodes. This
corresponds to a training time of 21 minutes on 768 CPU cores, as opposed to 10
hours when using a single node with 24 cores achieved by a baseline single-node
implementation.
|
['Adam Jędrych', 'Igor Adamski', 'Kamil Kaczmarek', 'Henryk Michalewski', 'Tomasz Grel', 'Robert Adamski']
|
2018-01-09
| null | null | null | null |
['2048']
|
['playing-games']
|
[-6.13025129e-01 3.33911568e-01 9.00969803e-02 -8.59919935e-02
-5.83516896e-01 -3.61554712e-01 5.57762861e-01 4.01519299e-01
-1.28288317e+00 9.66755450e-01 -2.25895017e-01 -6.32423341e-01
1.27072290e-01 -6.62404120e-01 -7.94142008e-01 -9.72930491e-01
-5.03811955e-01 6.85762346e-01 2.55063772e-02 -2.08841175e-01
-6.24231994e-03 5.82173407e-01 -1.23385429e+00 -1.49376076e-02
2.12161615e-01 1.05062723e+00 -1.73972815e-01 1.25803185e+00
-5.92452250e-02 1.31474519e+00 -1.00563848e+00 -1.27460271e-01
5.22430658e-01 -4.43752140e-01 -8.65805745e-01 -2.27207229e-01
1.35844573e-01 -9.25639689e-01 -4.32397425e-01 6.97252274e-01
7.64891148e-01 2.15236157e-01 3.06106638e-02 -1.33721197e+00
3.59566122e-01 8.65148365e-01 -5.09181321e-01 2.69059747e-01
-2.27196664e-01 1.82833850e-01 8.92937779e-01 -4.41154689e-01
3.84821206e-01 9.86257553e-01 6.47952020e-01 5.52672744e-01
-1.25059187e+00 -5.57040751e-01 1.06882125e-01 -9.95314214e-03
-1.19927144e+00 -6.12229824e-01 3.86899114e-01 1.08426064e-01
1.52385414e+00 -2.15215191e-01 1.13481498e+00 7.51324058e-01
4.60782260e-01 7.08141506e-01 1.01964259e+00 -5.41136622e-01
9.50343311e-01 -1.40356511e-01 -3.77904624e-01 9.51158822e-01
1.45352885e-01 7.66392648e-02 -6.63013160e-01 -4.32896525e-01
8.57749999e-01 -3.13489109e-01 4.33411181e-01 -1.70622751e-01
-1.06849289e+00 1.03272080e+00 3.64353210e-01 2.06464510e-02
-7.31309772e-01 1.05140817e+00 9.98678744e-01 9.10681069e-01
7.33879328e-01 2.80882627e-01 -8.28914404e-01 -5.98406613e-01
-1.00791883e+00 4.80032653e-01 1.13331246e+00 6.01304293e-01
1.04168355e+00 6.82774782e-01 3.59744698e-01 3.48825544e-01
2.49902248e-01 5.03693640e-01 5.66742122e-01 -1.38503408e+00
2.36474723e-01 1.22543342e-01 -2.06969418e-02 -3.61122400e-01
-7.27046609e-01 -2.84034520e-01 -8.02834630e-01 8.62073541e-01
6.28150344e-01 -1.09794736e+00 -5.21605551e-01 1.51844501e+00
8.83023143e-01 1.13663599e-01 2.58165449e-01 7.05174387e-01
6.97583184e-02 6.85842097e-01 -1.49539551e-02 -1.82764813e-01
1.14163375e+00 -1.21795535e+00 -1.88070565e-01 -6.08563721e-02
1.20034170e+00 -5.24051249e-01 4.65037614e-01 4.89161640e-01
-1.16406226e+00 -1.90416351e-01 -1.23850644e+00 1.34958491e-01
-9.28958133e-02 -5.60517609e-02 1.02469969e+00 6.37515485e-01
-1.60074186e+00 9.61360335e-01 -1.49830651e+00 9.40372348e-02
3.63509893e-01 7.21887469e-01 -2.82000065e-01 4.27006364e-01
-8.35690022e-01 8.41060042e-01 3.75430673e-01 -1.09953634e-01
-1.43829417e+00 -7.86038697e-01 -4.53852743e-01 1.03723094e-01
3.51274908e-01 -5.28103828e-01 1.65802884e+00 -1.35386205e+00
-2.22585583e+00 4.38033402e-01 7.52177611e-02 -9.92257178e-01
7.12778151e-01 -1.72297984e-01 1.97708741e-01 2.99031377e-01
-3.84214014e-01 6.47611260e-01 1.00441992e+00 -5.79450846e-01
-6.26435757e-01 -4.14948583e-01 2.34874487e-01 3.67830992e-01
-3.99099082e-01 1.65273219e-01 5.49850464e-02 -2.31566191e-01
-3.53525192e-01 -1.05826974e+00 -5.91028571e-01 -2.68432591e-03
3.62089962e-01 -3.97979259e-01 6.03490412e-01 -5.87638199e-01
6.82950199e-01 -1.94719613e+00 1.04273997e-01 3.41049463e-01
5.15609622e-01 1.49690181e-01 -1.46580771e-01 7.60481656e-01
2.10333407e-01 -3.39370310e-01 1.34795696e-01 -6.01117373e-01
6.32456094e-02 4.06070948e-01 -4.17918637e-02 8.84810984e-01
-3.00109982e-01 6.20921850e-01 -8.65256250e-01 -4.68244135e-01
-7.30953887e-02 2.46597201e-01 -5.76184988e-01 3.50472033e-01
-4.05337572e-01 2.10660577e-01 -4.01157796e-01 2.28289679e-01
3.40235770e-01 -2.75795966e-01 6.69998765e-01 5.03677607e-01
-2.39889491e-02 2.77772814e-01 -1.37685394e+00 1.89937735e+00
-7.15792120e-01 6.30407929e-01 6.55683815e-01 -1.06516790e+00
6.70814574e-01 6.56678855e-01 9.09742773e-01 -6.10874712e-01
6.48802668e-02 4.05550927e-01 -1.00809604e-01 3.29198758e-03
2.14324415e-01 1.41290963e-01 2.54655838e-01 1.12142372e+00
3.12153757e-01 6.19232878e-02 1.60515398e-01 2.80757427e-01
1.58662570e+00 1.78569391e-01 6.83115795e-02 -6.38608456e-01
7.79251233e-02 1.97917983e-01 4.65872079e-01 9.83146727e-01
-2.48262003e-01 -3.92957777e-01 8.35872054e-01 -1.01257181e+00
-1.52398467e+00 -4.13143128e-01 2.60897458e-01 1.46574759e+00
-4.13345426e-01 -6.49688363e-01 -1.06381238e+00 -7.45489597e-01
4.57146168e-02 1.13295339e-01 -5.66714525e-01 2.49098763e-01
-9.28285658e-01 -7.48574138e-01 9.24155891e-01 4.92629200e-01
7.44452417e-01 -1.17874920e+00 -1.31421554e+00 4.71143693e-01
6.23297751e-01 -6.56543195e-01 -1.69922877e-02 6.67247772e-01
-1.12342703e+00 -7.36740589e-01 -7.29078054e-01 -4.11885530e-01
5.45706749e-01 -3.06343049e-01 1.27301311e+00 3.90265614e-01
-1.70694247e-01 1.46671474e-01 -1.86618567e-01 -2.58602172e-01
-4.39031959e-01 4.48029399e-01 2.32640326e-01 -4.01942432e-01
-1.47284538e-01 -6.95453942e-01 -7.50631690e-01 -3.60739768e-01
-8.97884667e-01 9.30369049e-02 3.86040390e-01 9.06812549e-01
1.74102396e-01 -1.63273096e-01 5.57542682e-01 -8.60502005e-01
5.38067818e-01 -3.84044498e-01 -1.19073033e+00 7.43456781e-02
-1.05639946e+00 4.24688250e-01 1.08212125e+00 -4.10242409e-01
-7.93033123e-01 2.37105533e-01 -3.79020125e-01 -3.25897098e-01
1.54050872e-01 3.77107948e-01 6.11931443e-01 -4.89844084e-01
6.50446653e-01 -4.61491644e-02 5.69370508e-01 -2.28237301e-01
3.23938757e-01 3.62216622e-01 -2.22986102e-01 -8.39704752e-01
2.45061129e-01 3.53173018e-01 2.12209582e-01 -7.14700222e-01
-3.67295057e-01 -5.82934693e-02 -3.59128028e-01 -1.24716081e-01
4.48194116e-01 -1.10496318e+00 -1.27209115e+00 6.86942995e-01
-1.03561568e+00 -1.24422741e+00 -5.06339431e-01 3.63907486e-01
-5.95433950e-01 -3.55822816e-02 -1.01222014e+00 -7.83186555e-01
-7.31681228e-01 -1.05234754e+00 8.62180829e-01 1.60549536e-01
8.48927125e-02 -1.26926899e+00 5.18537223e-01 -1.18211582e-01
9.26310480e-01 -1.95275769e-02 5.68099976e-01 -7.68179417e-01
-1.52628049e-01 -5.45671443e-03 1.67243466e-01 2.24481270e-01
-5.74758112e-01 -2.06829488e-01 -8.48980129e-01 -8.15723777e-01
-1.10507615e-01 -9.01460528e-01 5.62908590e-01 1.66254222e-01
8.73818636e-01 -3.70435029e-01 8.90377760e-02 5.94755113e-01
1.68030334e+00 -1.68048367e-01 3.02260280e-01 4.23971176e-01
5.58494329e-01 -3.19559164e-02 2.73791939e-01 1.18115938e+00
6.79330826e-02 4.10359293e-01 4.51479435e-01 -1.50778100e-01
3.64297718e-01 -4.35150675e-02 4.16501075e-01 7.21381128e-01
-6.68147057e-02 1.14869751e-01 -9.68648016e-01 2.19159916e-01
-2.07814908e+00 -4.68718678e-01 3.92157674e-01 2.08877301e+00
1.06871009e+00 -8.23099613e-02 4.55361784e-01 -1.95279166e-01
6.58457726e-02 2.87354261e-01 -6.68156028e-01 -8.59094322e-01
2.95263618e-01 7.83214211e-01 1.04405177e+00 6.39542699e-01
-7.19822526e-01 1.21570265e+00 7.04571056e+00 9.18451488e-01
-1.29467547e+00 3.08825761e-01 7.62540460e-01 -4.16199088e-01
1.67161673e-01 1.59584314e-01 -8.33651483e-01 1.92610130e-01
1.73087442e+00 -5.57246991e-02 1.03980291e+00 1.26267517e+00
-8.74045957e-03 -3.88393074e-01 -8.26046467e-01 8.26012909e-01
-2.28983641e-01 -1.40366948e+00 -5.54907799e-01 2.97222078e-01
8.21565688e-01 6.89777076e-01 -5.65217733e-01 4.87928987e-01
8.78308535e-01 -1.04597378e+00 6.09812796e-01 -4.09634784e-03
4.88393754e-01 -1.15748477e+00 6.35188222e-01 4.62604761e-01
-8.79377961e-01 -2.27133054e-02 -6.70113862e-01 -3.85325849e-01
-2.09778368e-01 4.33656812e-01 -6.99511647e-01 1.60465196e-01
8.05518806e-01 2.66929686e-01 -2.98271060e-01 5.05231798e-01
-1.18790455e-01 7.84797966e-01 -8.47108245e-01 -2.77879238e-01
6.74136758e-01 7.20846048e-03 3.15124579e-02 8.64721060e-01
-9.86547768e-02 -1.16730973e-01 1.73544869e-01 2.17654452e-01
-3.06675553e-01 1.61534399e-01 -2.82312810e-01 8.36961195e-02
3.55557948e-01 1.48752308e+00 -6.76822543e-01 -6.19325161e-01
-2.66105622e-01 9.59623277e-01 1.07588625e+00 2.92326123e-01
-7.72938013e-01 -3.82344872e-01 6.32385015e-01 -2.75622249e-01
3.84642750e-01 -7.19525039e-01 -1.15218617e-01 -9.93951321e-01
-1.18974775e-01 -1.24881089e+00 2.50509232e-01 -1.00611471e-01
-7.23628938e-01 4.67179239e-01 -1.80774868e-01 -4.34612215e-01
-6.46728277e-01 -5.84266603e-01 -6.28825545e-01 5.96085012e-01
-1.19763505e+00 -7.53604054e-01 1.46316856e-01 7.48165250e-01
2.47535616e-01 -3.43693405e-01 1.09165740e+00 4.12836019e-03
-4.63801950e-01 8.14139843e-01 5.41196465e-01 -2.03812681e-03
3.15787256e-01 -1.32210314e+00 3.91438127e-01 5.21353245e-01
2.40581319e-01 2.84880370e-01 7.16713488e-01 -3.66807520e-01
-1.99842238e+00 -5.66759169e-01 7.03001976e-01 2.38284066e-01
8.79533887e-01 -3.11830431e-01 -5.47357440e-01 7.25361884e-01
6.89384997e-01 4.41011280e-01 2.41668209e-01 2.02231213e-01
3.82426195e-02 -4.55252022e-01 -9.22115088e-01 4.72717375e-01
4.88536447e-01 -1.39796838e-01 2.73147255e-01 6.23624325e-01
3.69089395e-01 -7.30341852e-01 -1.13401389e+00 -1.80288762e-01
4.06466097e-01 -7.72721827e-01 5.48528790e-01 -7.93190837e-01
2.72331715e-01 1.22570448e-01 2.69186646e-01 -1.36310041e+00
1.53383523e-01 -1.11821067e+00 -4.33068603e-01 4.72079456e-01
3.14699769e-01 -7.83339500e-01 1.19045258e+00 6.49554610e-01
6.09068684e-02 -1.03816104e+00 -1.35385716e+00 -4.65271533e-01
2.93012023e-01 -1.58819556e-01 4.06158358e-01 4.80962753e-01
6.38507456e-02 1.47705570e-01 -4.41201150e-01 -5.43649197e-02
5.36174357e-01 -8.68478045e-02 9.47906554e-01 -7.03546047e-01
-7.94932961e-01 -1.62769020e-01 -1.06317617e-01 -9.72545087e-01
5.04251778e-01 -6.02177680e-01 -1.23859502e-01 -1.09762311e+00
-2.71467656e-01 -6.92697465e-01 -2.27793410e-01 8.84284914e-01
3.16366374e-01 -3.08433399e-02 1.13798767e-01 6.71284720e-02
-8.34871411e-01 5.96029758e-01 7.76848257e-01 2.35229447e-01
-1.59188017e-01 -1.78017020e-01 -1.22557282e-01 5.50817311e-01
9.79809523e-01 -7.62732029e-01 -2.10451663e-01 -7.29003787e-01
5.22178888e-01 4.10353988e-01 1.21803470e-01 -8.45156610e-01
5.11738300e-01 -1.97522506e-01 4.68225300e-01 1.98969126e-01
3.30021203e-01 -6.46875083e-01 -1.35872379e-01 9.28686261e-01
-5.34165919e-01 6.84578180e-01 2.18747094e-01 3.29560518e-01
1.19649293e-02 -3.67762357e-01 8.84328127e-01 -3.67375910e-01
-3.83319169e-01 2.03561068e-01 -7.39084482e-01 1.43762439e-01
1.19212615e+00 5.41454494e-01 -2.38323122e-01 -3.75167936e-01
-4.50121671e-01 2.87844211e-01 3.29480320e-01 -3.01453739e-01
3.29580098e-01 -9.13837969e-01 -6.40601158e-01 2.13497028e-01
-6.08576596e-01 -9.43392143e-02 6.50290921e-02 7.98341990e-01
-1.32159114e+00 1.19800612e-01 -2.87882119e-01 -2.36298814e-01
-1.19459915e+00 2.01504126e-01 6.44049585e-01 -7.36886203e-01
-9.83094513e-01 8.91446352e-01 -7.62763560e-01 -2.05245987e-01
4.05108601e-01 1.73441410e-01 3.82773340e-01 -2.20928892e-01
4.09217447e-01 5.19711256e-01 3.93659383e-01 2.08777502e-01
-3.18866938e-01 -1.67218849e-01 -1.89048856e-01 -6.42372489e-01
1.44713724e+00 3.16629589e-01 -2.98832506e-01 2.15161279e-01
1.12912858e+00 -2.14853033e-01 -1.60635924e+00 -2.45155752e-01
-1.86808944e-01 4.06787135e-02 5.53547502e-01 -2.96297550e-01
-1.25893056e+00 5.94563842e-01 6.10444665e-01 -1.35613410e-02
7.32382774e-01 -4.64377224e-01 8.66008341e-01 1.06902254e+00
6.38607562e-01 -1.63566029e+00 3.46222818e-02 6.91253603e-01
4.21138346e-01 -1.15919173e+00 4.36303914e-01 7.00946391e-01
-5.37681818e-01 1.41589987e+00 6.41506076e-01 -7.95680583e-01
7.78116286e-01 7.41914988e-01 8.27222317e-02 -1.08551301e-01
-1.27980745e+00 2.78405935e-01 -6.42274380e-01 -6.29429892e-02
1.70774356e-01 2.01300494e-02 -1.23075143e-01 -2.34383225e-01
-6.25002310e-02 -2.75347888e-01 3.25431645e-01 1.34752250e+00
-5.20726204e-01 -1.17658436e+00 3.53840999e-02 2.50692040e-01
-7.05411017e-01 -1.36867344e-01 1.97116852e-01 7.42875814e-01
-4.24173176e-01 7.37482667e-01 2.48554930e-01 -5.80949895e-02
-3.91700387e-01 1.20502099e-01 6.18534207e-01 -3.10717881e-01
-1.21980333e+00 9.45406593e-03 1.64003342e-01 -8.23810041e-01
-2.23591998e-01 -3.30553532e-01 -1.50860417e+00 -9.29163039e-01
7.28317164e-03 3.29367012e-01 1.26272511e+00 1.15499258e+00
4.69744861e-01 2.62272567e-01 6.95511758e-01 -1.03786576e+00
-1.30140197e+00 -8.91742706e-01 -7.56768882e-01 -1.80468023e-01
2.36909628e-01 -9.46328789e-02 -4.33409929e-01 -4.09433454e-01]
|
[4.0098042488098145, 1.7035598754882812]
|
1dc3be7a-8f61-48da-b7a2-17438663cad2
|
embedding-synthetic-off-policy-experience-for
|
2212.01375
| null |
https://arxiv.org/abs/2212.01375v1
|
https://arxiv.org/pdf/2212.01375v1.pdf
|
Embedding Synthetic Off-Policy Experience for Autonomous Driving via Zero-Shot Curricula
|
ML-based motion planning is a promising approach to produce agents that exhibit complex behaviors, and automatically adapt to novel environments. In the context of autonomous driving, it is common to treat all available training data equally. However, this approach produces agents that do not perform robustly in safety-critical settings, an issue that cannot be addressed by simply adding more data to the training set - we show that an agent trained using only a 10% subset of the data performs just as well as an agent trained on the entire dataset. We present a method to predict the inherent difficulty of a driving situation given data collected from a fleet of autonomous vehicles deployed on public roads. We then demonstrate that this difficulty score can be used in a zero-shot transfer to generate curricula for an imitation-learning based planning agent. Compared to training on the entire unbiased training dataset, we show that prioritizing difficult driving scenarios both reduces collisions by 15% and increases route adherence by 14% in closed-loop evaluation, all while using only 10% of the training data.
|
['Shimon Whiteson', 'Payam Nikdel', "Matthew O'Kelly", 'Aman Sinha', 'Supratik Paul', 'Sirish Srinivasan', 'Eli Bronstein']
|
2022-12-02
| null | null | null | null |
['motion-planning']
|
['robots']
|
[ 1.99530482e-01 5.77958286e-01 5.52776121e-02 -3.36110294e-01
-6.84270799e-01 -6.70678794e-01 5.81186593e-01 8.33584741e-02
-7.85293102e-01 8.72065246e-01 -1.89553034e-02 -4.57403183e-01
-5.80640957e-02 -9.11044002e-01 -1.05492306e+00 -4.03527677e-01
-1.31077051e-01 9.90236521e-01 5.69983482e-01 -7.19259381e-01
1.17132626e-01 6.77202404e-01 -2.06378603e+00 -4.75743823e-02
1.00291061e+00 1.91745326e-01 5.51352978e-01 6.55604959e-01
4.27648962e-01 7.92054117e-01 -4.80274171e-01 1.03562705e-01
5.65255225e-01 -3.10882211e-01 -7.50921190e-01 3.43006067e-02
3.60646904e-01 -5.72776616e-01 -5.49039245e-01 4.17022049e-01
4.01280433e-01 5.40377557e-01 6.59995914e-01 -1.50839484e+00
3.63839537e-01 3.16497803e-01 2.39567995e-01 9.16200280e-02
3.93720865e-01 8.01274478e-01 4.56152886e-01 -2.91887313e-01
9.15857673e-01 9.89103615e-01 4.89683986e-01 5.99804878e-01
-1.07571733e+00 -5.33127069e-01 5.67232817e-02 1.64374650e-01
-1.19206202e+00 -3.56916308e-01 4.66446608e-01 -3.62010717e-01
1.15486789e+00 -1.15915768e-01 8.94759655e-01 9.16204274e-01
4.03090119e-01 7.58086741e-01 8.00648391e-01 -2.69416701e-02
3.12997222e-01 1.05249718e-01 -3.46391112e-01 7.67443597e-01
2.53297716e-01 7.10452020e-01 -2.28573442e-01 2.05305427e-01
2.71997541e-01 -4.70777094e-01 1.53510064e-01 -7.35189617e-01
-1.19040513e+00 8.47483099e-01 3.74373078e-01 -1.39895096e-01
-4.90733624e-01 2.78455377e-01 3.07244807e-01 5.45333564e-01
-7.31787682e-02 9.49877501e-01 -3.79183620e-01 -5.88082075e-01
-5.31289041e-01 1.08909822e+00 8.37383151e-01 1.18716073e+00
1.07841992e+00 2.88693249e-01 1.59585938e-01 3.75062406e-01
-1.32783920e-01 6.22702956e-01 8.07694793e-02 -1.53068078e+00
5.58283508e-01 7.15546608e-01 4.64445144e-01 -6.52549565e-01
-8.19978237e-01 -7.75339231e-02 1.14176519e-01 8.55172932e-01
4.23964560e-01 -7.45167673e-01 -8.82187307e-01 1.80165482e+00
5.02853751e-01 -7.86264706e-03 2.28827924e-01 7.24311888e-01
2.29209274e-01 6.22134984e-01 1.61113217e-01 5.85483201e-02
6.76857591e-01 -1.04549205e+00 -2.53533065e-01 -6.45545125e-01
1.19406438e+00 -3.72426003e-01 9.50394630e-01 3.25980812e-01
-1.11249006e+00 -6.52934372e-01 -1.30085993e+00 2.73998797e-01
-3.94497633e-01 -4.32961047e-01 5.66175342e-01 5.36449075e-01
-1.07523644e+00 7.83014297e-01 -7.95513988e-01 -4.39654440e-01
3.13054770e-01 5.22323847e-01 -4.58089232e-01 -2.70430356e-01
-9.48825717e-01 1.64990997e+00 3.37379932e-01 -3.53458941e-01
-1.60168839e+00 -6.99609995e-01 -1.11162472e+00 -2.71555185e-01
3.52597594e-01 -5.25842607e-01 1.50180030e+00 -5.99912882e-01
-1.50185823e+00 2.70775050e-01 4.21023928e-03 -3.64151895e-01
6.85323834e-01 -9.22111124e-02 -4.42448743e-02 -2.12055579e-01
3.43630880e-01 1.21788371e+00 3.44856322e-01 -1.38288534e+00
-1.10345161e+00 4.40071784e-02 4.68110025e-01 7.09465623e-01
2.16789767e-01 -4.36539561e-01 -5.39497361e-02 1.58375889e-01
-4.36521977e-01 -1.45115352e+00 -7.17838883e-01 -2.49358848e-01
1.27047241e-01 -1.18067160e-01 8.34447622e-01 -3.38690072e-01
5.30148685e-01 -1.77549386e+00 2.92831212e-01 3.48971516e-01
-1.21013671e-01 1.56628534e-01 -5.90363145e-01 7.21570194e-01
3.90456796e-01 -2.78190792e-01 -3.41890901e-01 -7.63787702e-02
2.43949927e-02 5.08616626e-01 -4.05577905e-02 2.05306977e-01
2.78960139e-01 8.82418811e-01 -1.26157713e+00 -2.81761914e-01
4.67413545e-01 6.74548075e-02 -8.01280677e-01 1.35382548e-01
-3.21282387e-01 7.42137969e-01 -3.87075156e-01 4.54572029e-03
3.48256409e-01 5.05308390e-01 1.55706972e-01 4.39131886e-01
-2.26186961e-01 2.95051217e-01 -1.06179643e+00 1.61680102e+00
-5.88727057e-01 8.78302574e-01 -2.43893623e-01 -7.14750767e-01
8.44743073e-01 1.36054203e-01 5.33763587e-01 -1.09026432e+00
7.92962238e-02 2.42007881e-01 5.00993669e-01 -8.32750499e-01
6.61822259e-01 -1.53176934e-01 -3.92656595e-01 5.35939515e-01
-1.42925963e-01 -7.72521019e-01 4.45379287e-01 -1.14139631e-01
1.58477664e+00 2.63700932e-01 -2.15442374e-01 -1.37542620e-01
2.34098658e-01 8.85895848e-01 4.06248182e-01 7.74692774e-01
-4.86553550e-01 1.84070781e-01 2.39411488e-01 -5.91703415e-01
-1.51203620e+00 -8.45973432e-01 3.27220678e-01 1.00966787e+00
3.93884957e-01 -1.57391027e-01 -1.08537543e+00 -7.30174720e-01
1.20603256e-01 1.29598570e+00 -5.57682931e-01 -3.95639688e-01
-8.83838832e-01 -2.68682957e-01 7.03187168e-01 4.18643385e-01
2.97326297e-01 -1.40409994e+00 -1.24228787e+00 3.12027276e-01
-5.39896078e-02 -1.12260771e+00 -2.92427301e-01 1.17684692e-01
-5.06989479e-01 -1.03312051e+00 -2.08719119e-01 -8.76775444e-01
8.28787446e-01 3.75045210e-01 8.59769762e-01 2.82250732e-01
-1.41670316e-01 3.34835976e-01 -2.59166092e-01 -4.89336014e-01
-9.96720016e-01 2.50484020e-01 2.18886629e-01 -6.82136357e-01
2.25043982e-01 -3.82762372e-01 -4.29237604e-01 5.97947121e-01
-4.33818370e-01 3.00739616e-01 6.50864601e-01 5.81842482e-01
1.01300739e-01 2.51217365e-01 6.35062993e-01 -5.67224383e-01
6.69130027e-01 -4.93309706e-01 -6.11480653e-01 -2.58060843e-01
-7.04075873e-01 2.32198834e-01 6.15613341e-01 -5.64134121e-01
-9.20703471e-01 4.52380568e-01 -1.44425910e-02 -1.28018439e-01
-5.39802969e-01 4.17011738e-01 5.84290479e-04 -2.90382236e-01
8.71326745e-01 1.17001295e-01 4.31618363e-01 2.01115936e-01
3.47288311e-01 2.64627457e-01 2.67110705e-01 -6.87001884e-01
1.03093350e+00 2.60974348e-01 1.17946759e-01 -6.07149124e-01
-1.76269725e-01 -1.94139421e-01 -6.27198935e-01 -5.86491466e-01
7.85812616e-01 -6.52726471e-01 -7.20828533e-01 1.73884392e-01
-8.83386970e-01 -1.18887711e+00 -4.36477363e-01 5.74444950e-01
-1.16176343e+00 -1.31390944e-01 -3.29755060e-02 -5.87931037e-01
3.36422622e-01 -1.60635126e+00 7.95748174e-01 1.28276229e-01
-4.69413936e-01 -6.94024265e-01 3.05536360e-01 4.04994309e-01
4.57117736e-01 2.48148516e-01 7.27567434e-01 -4.43702996e-01
-5.23147643e-01 -3.32437962e-01 2.52371520e-01 -4.07883432e-03
-5.97240478e-02 -2.19338447e-01 -5.37529707e-01 -3.82030457e-01
-4.63925064e-01 -5.31938076e-01 2.79074162e-01 9.76261646e-02
4.86281842e-01 -2.56039631e-02 -4.68783110e-01 1.32813342e-02
1.06264758e+00 5.00457287e-01 6.99806094e-01 6.21547520e-01
4.90509570e-01 1.16544819e+00 1.03551781e+00 7.77026862e-02
8.91489387e-01 7.77699411e-01 5.47650874e-01 7.87908509e-02
-6.28204197e-02 -3.91109735e-01 5.14017463e-01 2.39984021e-01
-7.97394291e-02 -3.93716358e-02 -1.30886972e+00 7.79253066e-01
-1.96535158e+00 -1.14002860e+00 -1.01927243e-01 2.11472678e+00
4.73080218e-01 4.41825747e-01 4.30962831e-01 8.74214619e-03
3.27969581e-01 -2.19855174e-01 -6.51350677e-01 -5.65352261e-01
3.23351711e-01 2.45655887e-02 8.87121677e-01 7.90133715e-01
-7.57282734e-01 1.17474484e+00 7.24983788e+00 4.82883483e-01
-8.70836318e-01 -5.89795671e-02 2.54695535e-01 -1.91615313e-01
-3.76014709e-01 1.63116351e-01 -5.29356360e-01 3.80241007e-01
1.38781142e+00 -2.63882160e-01 7.32626975e-01 8.81426215e-01
7.37567782e-01 -3.58337373e-01 -1.21675491e+00 2.52515912e-01
-5.91780618e-02 -1.20239329e+00 -1.76436886e-01 2.81157583e-01
8.90246809e-01 2.36785293e-01 -1.00348584e-01 8.16596866e-01
7.52557576e-01 -1.26831353e+00 9.29232895e-01 2.28191569e-01
2.96571165e-01 -1.07847631e+00 7.55156875e-01 9.18616533e-01
-9.36985552e-01 -2.63996154e-01 -2.10524559e-01 -4.38008666e-01
3.89468163e-01 -3.21487814e-01 -1.43969750e+00 3.44506353e-01
3.81402910e-01 2.22005710e-01 -4.82732922e-01 1.14993310e+00
-9.03751105e-02 3.71848822e-01 -3.57504040e-01 -2.80977815e-01
5.69979906e-01 2.64172833e-02 6.19333208e-01 7.43157387e-01
2.25445658e-01 7.61634484e-02 3.87376726e-01 5.76647997e-01
4.06424642e-01 -2.83780962e-01 -1.22054052e+00 2.70114481e-01
5.18995523e-01 8.07018399e-01 -3.36728126e-01 -2.60975927e-01
-2.03494340e-01 5.51377356e-01 4.61036533e-01 3.20834577e-01
-8.76232266e-01 -2.35975713e-01 7.83313751e-01 2.81599402e-01
2.81954676e-01 -5.34466922e-01 -2.31859431e-01 -2.85675496e-01
-2.37923250e-01 -9.37624395e-01 -1.85516879e-01 -9.04547930e-01
-4.84241873e-01 4.65952277e-01 3.74805838e-01 -1.55887139e+00
-6.73357010e-01 -4.22774374e-01 -6.03292704e-01 6.90193415e-01
-1.57957506e+00 -8.11837018e-01 -6.44681871e-01 2.82838196e-01
8.19030344e-01 -4.00879622e-01 7.38544106e-01 2.34771118e-01
-3.83707374e-01 2.85505444e-01 -2.34356880e-01 -3.22793752e-01
3.55703890e-01 -9.28344369e-01 6.89285040e-01 6.03905439e-01
-3.90778273e-01 2.49380186e-01 1.21198142e+00 -7.14448750e-01
-1.39430964e+00 -1.25636005e+00 5.91648400e-01 -7.67637730e-01
3.80458236e-01 -3.65546532e-02 -6.94136620e-01 7.24163890e-01
1.48599252e-01 -3.67155522e-01 2.37540632e-01 -2.08790228e-01
2.51099944e-01 9.68098938e-02 -1.28419733e+00 1.00744081e+00
1.30578089e+00 -9.21624526e-02 -5.46378136e-01 2.70892024e-01
6.51871622e-01 -7.00688243e-01 -6.66642189e-01 4.89056855e-01
4.74490404e-01 -6.30767763e-01 6.65848076e-01 -7.78602183e-01
3.55802387e-01 -4.29843485e-01 1.45440577e-02 -1.70138478e+00
-3.20360154e-01 -4.89954144e-01 3.66220921e-01 4.14836645e-01
8.13518047e-01 -5.65008461e-01 1.16558957e+00 6.98081791e-01
-5.74608922e-01 -6.29888892e-01 -1.03644788e+00 -8.79887760e-01
3.93159270e-01 -6.80348992e-01 5.69331229e-01 4.65771139e-01
9.37882736e-02 9.75510105e-02 -2.01936841e-01 2.95091569e-01
4.15757686e-01 -4.08296198e-01 1.40674460e+00 -9.57349896e-01
8.41687024e-02 -3.75516385e-01 -5.19080997e-01 -8.93991470e-01
4.98277098e-01 -6.99333608e-01 6.74803674e-01 -1.83586228e+00
-2.44437769e-01 -9.19398606e-01 2.87173390e-01 4.43663657e-01
1.19169191e-01 -1.45636886e-01 2.09739864e-01 5.61954156e-02
-4.88440335e-01 4.71549124e-01 1.46968436e+00 -1.08668216e-01
-2.79404193e-01 -3.69223952e-02 -3.77886564e-01 5.29115498e-01
9.91721928e-01 -3.96322578e-01 -9.41688001e-01 -4.78325397e-01
5.54789044e-02 9.73693356e-02 3.06277812e-01 -1.45183086e+00
2.98664242e-01 -7.43433535e-01 3.21409814e-02 -3.60060841e-01
5.09868145e-01 -9.57922339e-01 3.44076723e-01 7.21111953e-01
-3.48414510e-01 2.30438262e-01 7.14586735e-01 4.19343054e-01
2.45228052e-01 -4.09634680e-01 4.68760401e-01 -6.90615624e-02
-1.15279245e+00 9.40192714e-02 -1.02823174e+00 -4.54785116e-02
1.55761027e+00 -5.21526515e-01 -3.48066181e-01 -4.91166383e-01
-4.17065084e-01 7.06001163e-01 7.38681912e-01 5.50262809e-01
5.47887981e-01 -1.16652894e+00 -8.01731229e-01 3.31919014e-01
1.42745957e-01 1.29364371e-01 2.45631263e-01 5.74621081e-01
-7.37963200e-01 3.85632336e-01 -6.89850569e-01 -4.98379409e-01
-9.43301022e-01 3.73343408e-01 4.92247015e-01 1.38574596e-02
-7.06237435e-01 4.17318106e-01 -1.07625760e-02 -9.58970428e-01
-2.71907784e-02 -3.98076139e-02 -2.67121106e-01 -3.19389522e-01
1.53967559e-01 3.35077941e-01 4.87316325e-02 -7.05320954e-01
-2.57689267e-01 3.43936443e-01 1.44146889e-01 -5.98054707e-01
1.05720580e+00 9.75039899e-02 6.99728131e-01 2.64267445e-01
7.87809551e-01 -3.23976815e-01 -1.76550913e+00 3.45477134e-01
-9.13319513e-02 -4.13622260e-01 -1.54048517e-01 -6.62476242e-01
-5.86240590e-01 4.69099224e-01 4.01012540e-01 9.55634639e-02
5.08105874e-01 -2.42942229e-01 6.98546767e-01 6.76020741e-01
8.79981160e-01 -1.45531964e+00 1.55206323e-01 6.49726808e-01
9.07238841e-01 -1.29873848e+00 -2.45827675e-01 -1.96393654e-01
-1.10004902e+00 7.66215026e-01 1.01923501e+00 -2.73259312e-01
2.33547732e-01 2.33178988e-01 9.09954980e-02 -1.02318674e-01
-1.00251746e+00 -4.17065471e-01 -1.66185364e-01 1.20890749e+00
-2.35536471e-01 3.68763991e-02 1.05623351e-02 -1.76957369e-01
-6.48060977e-01 -1.01392306e-01 9.65014577e-01 1.30538464e+00
-8.37081671e-01 -9.80816722e-01 -3.76253389e-02 5.13844371e-01
5.61565697e-01 5.15337467e-01 -1.17640555e-01 1.03409755e+00
2.54178882e-01 1.06682658e+00 2.63616920e-01 -7.43858278e-01
8.25594723e-01 4.24376801e-02 4.55133736e-01 -7.43927717e-01
-3.92627090e-01 -6.75956249e-01 7.00164378e-01 -6.08234286e-01
-1.11116819e-01 -9.68251288e-01 -1.68804181e+00 -6.59170926e-01
1.33994877e-01 1.27795592e-01 7.84765005e-01 1.03958893e+00
3.85143340e-01 7.50402272e-01 7.01472521e-01 -1.30917788e+00
-4.86706495e-01 -6.66611791e-01 1.00257054e-01 3.59047413e-01
3.13572079e-01 -1.12290001e+00 -2.65356272e-01 -2.15202078e-01]
|
[5.015110492706299, 1.1888312101364136]
|
1b088b97-30d2-4f35-bec2-0e57dfcc0eef
|
deep-label-distribution-learning-with-label
|
1611.01731
| null |
http://arxiv.org/abs/1611.01731v2
|
http://arxiv.org/pdf/1611.01731v2.pdf
|
Deep Label Distribution Learning with Label Ambiguity
|
Convolutional Neural Networks (ConvNets) have achieved excellent recognition
performance in various visual recognition tasks. A large labeled training set
is one of the most important factors for its success. However, it is difficult
to collect sufficient training images with precise labels in some domains such
as apparent age estimation, head pose estimation, multi-label classification
and semantic segmentation. Fortunately, there is ambiguous information among
labels, which makes these tasks different from traditional classification.
Based on this observation, we convert the label of each image into a discrete
label distribution, and learn the label distribution by minimizing a
Kullback-Leibler divergence between the predicted and ground-truth label
distributions using deep ConvNets. The proposed DLDL (Deep Label Distribution
Learning) method effectively utilizes the label ambiguity in both feature
learning and classifier learning, which help prevent the network from
over-fitting even when the training set is small. Experimental results show
that the proposed approach produces significantly better results than
state-of-the-art methods for age estimation and head pose estimation. At the
same time, it also improves recognition performance for multi-label
classification and semantic segmentation tasks.
|
['Xin Geng', 'Chen-Wei Xie', 'Bin-Bin Gao', 'Chao Xing', 'Jianxin Wu']
|
2016-11-06
| null | null | null | null |
['head-pose-estimation']
|
['computer-vision']
|
[ 7.54966512e-02 5.81927858e-02 -3.59266967e-01 -1.01691806e+00
-6.22002780e-01 -2.55263537e-01 1.75442925e-04 7.65251368e-02
-7.20266521e-01 6.39065742e-01 -1.41063586e-01 2.32991651e-01
-5.49425595e-02 -3.48404557e-01 -5.05953610e-01 -9.56298232e-01
3.75747949e-01 7.53181517e-01 -2.73605622e-03 5.39131343e-01
5.67480624e-02 4.09059435e-01 -1.77635896e+00 -3.87786865e-01
9.90178764e-01 1.51618278e+00 1.41987175e-01 8.24063048e-02
-9.10829082e-02 5.45616269e-01 -4.40486193e-01 -4.62930471e-01
-1.76754355e-01 -1.69033073e-02 -6.53328001e-01 2.75624573e-01
7.34032452e-01 -5.09006917e-01 -1.29510403e-01 1.21105874e+00
6.61394417e-01 -3.85913551e-02 1.10410631e+00 -1.41052735e+00
-4.53822106e-01 2.25305915e-01 -1.03509784e+00 -2.33489305e-01
1.73081961e-02 -4.74568546e-01 7.96033919e-01 -7.30048895e-01
1.34946495e-01 1.31848550e+00 8.15020621e-01 7.36396670e-01
-8.12040448e-01 -1.15246928e+00 2.52593249e-01 2.28429928e-01
-1.52046895e+00 -3.39855582e-01 6.55468762e-01 -4.20974016e-01
1.95420548e-01 -6.32156432e-02 4.30470109e-01 8.55919719e-01
-3.23520787e-02 1.14551711e+00 1.18542457e+00 -3.78440261e-01
-2.01033596e-02 2.86451820e-02 3.50754052e-01 9.24605131e-01
2.67486542e-01 -1.95902079e-01 -4.72685456e-01 -8.14429969e-02
5.90345621e-01 9.63069424e-02 1.93114821e-02 -4.19431329e-01
-6.66250885e-01 7.50289977e-01 3.80263209e-01 3.01536936e-02
-1.58535033e-01 4.50732671e-02 4.40499008e-01 -3.20111543e-01
6.03342712e-01 -2.70522505e-01 -4.40130502e-01 2.38890171e-01
-9.68254387e-01 2.64127553e-01 5.25245368e-01 8.04018319e-01
7.56117344e-01 -7.90508389e-02 -1.52874570e-02 1.40840316e+00
6.85965776e-01 7.62862265e-01 5.86161733e-01 -7.87117600e-01
2.50897676e-01 5.24927258e-01 -9.01713893e-02 -8.12531054e-01
-8.69079709e-01 -4.41671729e-01 -8.42124641e-01 3.96908820e-02
9.35964108e-01 -1.91585839e-01 -1.15085363e+00 2.00431204e+00
4.38042790e-01 -9.79117700e-04 -2.52158970e-01 8.99807990e-01
1.02181613e+00 2.06591293e-01 4.82182175e-01 -3.74437213e-01
1.39808249e+00 -8.55656564e-01 -6.91080153e-01 -5.93018115e-01
6.18773162e-01 -5.61225355e-01 6.09559178e-01 3.21189910e-01
-6.69582367e-01 -4.68326777e-01 -9.82197285e-01 3.77156287e-02
-9.46734995e-02 6.63630188e-01 8.19402039e-01 7.06003845e-01
-7.17041433e-01 2.07125500e-01 -7.41590858e-01 -1.69735655e-01
6.93694293e-01 6.96865380e-01 -4.36642796e-01 -3.76011789e-01
-9.92879272e-01 7.26316571e-01 4.12082940e-01 2.24394277e-01
-5.54349005e-01 -3.27473730e-01 -1.02552497e+00 -1.70125991e-01
2.66744763e-01 -2.57974803e-01 1.41943979e+00 -9.33778346e-01
-1.01876485e+00 1.31601131e+00 -2.90639490e-01 -2.71228198e-02
3.95110160e-01 -1.57756075e-01 -1.95827872e-01 -6.01703413e-02
3.05914581e-01 1.02589571e+00 8.41391504e-01 -1.14501381e+00
-7.06343949e-01 -1.11392272e+00 -4.36656803e-01 3.60793114e-01
-3.72928798e-01 1.48338020e-01 -5.10823846e-01 -3.03817362e-01
5.86683750e-01 -1.00290823e+00 6.89964890e-02 1.42431945e-01
-3.30746979e-01 -7.75922835e-01 7.95279324e-01 -7.99484313e-01
9.51090991e-01 -2.04052377e+00 -1.45680517e-01 1.61403000e-01
3.71105075e-01 1.61127299e-01 1.55002341e-01 -4.40129340e-01
2.94214841e-02 -3.05366009e-01 1.77240581e-03 -7.12976575e-01
-2.12318264e-03 3.06296170e-01 3.85492176e-01 7.86191642e-01
-3.86327356e-01 6.65931761e-01 -5.49384415e-01 -9.82824504e-01
1.15026258e-01 4.27478135e-01 -1.44300312e-01 1.59967571e-01
2.19045524e-02 7.60761797e-01 -5.92279255e-01 9.18150961e-01
9.18499172e-01 -1.67377248e-01 7.77767822e-02 -3.28230381e-01
2.50602812e-01 -3.34710538e-01 -1.29921591e+00 1.53149986e+00
-4.74584550e-01 3.83265078e-01 6.13107793e-02 -1.21047437e+00
1.13136554e+00 2.08998471e-01 5.73114395e-01 -6.65968657e-01
5.59296012e-01 3.76625657e-01 -2.06181303e-01 -4.97380048e-01
1.24016747e-01 -2.52187133e-01 -1.50806591e-01 3.33418965e-01
1.51181653e-01 2.92895019e-01 -4.88995668e-03 -2.67021626e-01
3.36156875e-01 -1.41710922e-01 1.62077263e-01 -4.43638340e-02
5.27959228e-01 -6.66714728e-01 1.09707463e+00 4.26684737e-01
-4.59728092e-01 6.05055213e-01 3.42772603e-01 -4.86526787e-01
-8.14992905e-01 -8.44886541e-01 -4.53320116e-01 1.37925565e+00
2.52713084e-01 -1.57740861e-02 -1.10519016e+00 -8.42780173e-01
1.10294916e-01 2.87859589e-01 -5.59389174e-01 -1.05178028e-01
-4.25523698e-01 -8.54120791e-01 6.88838184e-01 8.98624599e-01
8.09011161e-01 -9.81704116e-01 -1.96674794e-01 -5.40826432e-02
-1.82526693e-01 -1.17482424e+00 -5.27438462e-01 7.67984763e-02
-7.11517990e-01 -1.17218697e+00 -9.77876425e-01 -1.26784086e+00
1.00027907e+00 -1.03781424e-01 6.24474764e-01 1.20981997e-02
-3.48156720e-01 1.18460275e-01 -3.89299802e-02 -4.26116616e-01
-1.89653840e-02 2.32347742e-01 4.23965901e-01 1.58492655e-01
7.54499853e-01 -4.68838304e-01 -5.95526755e-01 5.24633408e-01
-5.24121225e-01 -6.26500994e-02 5.49570918e-01 9.28793550e-01
6.42661989e-01 6.75681308e-02 9.21956420e-01 -8.09211612e-01
2.57704228e-01 -8.18108022e-02 -6.10422194e-01 4.70120966e-01
-7.97919869e-01 2.14005560e-02 3.07089359e-01 -6.30559921e-01
-1.06082177e+00 3.85628015e-01 -3.77984643e-01 -2.02398658e-01
-4.81270522e-01 3.11436892e-01 -4.32706684e-01 -1.43195510e-01
2.58533388e-01 7.92698190e-02 1.42902419e-01 -6.87020779e-01
-2.30055442e-03 1.10165107e+00 5.87895513e-01 -5.33559620e-01
2.41134077e-01 6.63642511e-02 1.39208123e-01 -6.41370475e-01
-1.40750194e+00 -5.04551709e-01 -8.80120933e-01 -3.47583979e-01
9.86901224e-01 -9.66709316e-01 -9.61521447e-01 1.26183474e+00
-8.99295747e-01 2.04833634e-02 4.60746139e-01 6.29261494e-01
-3.80172312e-01 3.15319031e-01 -6.09059632e-01 -8.54029894e-01
-5.00602305e-01 -1.21271992e+00 1.21028888e+00 8.62591803e-01
2.43469253e-02 -9.60412681e-01 -5.15900195e-01 6.84800446e-01
9.08827707e-02 2.32198108e-02 8.02352011e-01 -7.97878087e-01
-5.35461456e-02 -6.09610856e-01 -6.09140456e-01 4.71089900e-01
-1.04460272e-03 -3.36533010e-01 -1.17814112e+00 -4.15759504e-01
-2.62687355e-01 -8.36644948e-01 8.28509271e-01 8.68493617e-01
1.48478031e+00 1.05522119e-01 -4.83404666e-01 5.95569313e-01
9.97195780e-01 1.75507993e-01 2.69488543e-01 6.34684414e-02
1.08532596e+00 6.67194545e-01 7.91207790e-01 4.43277985e-01
7.72359014e-01 5.85588932e-01 3.55146319e-01 -6.40780255e-02
3.39800753e-02 -2.40330413e-01 -2.08365902e-01 5.90955138e-01
1.60498679e-01 8.89208764e-02 -8.79331470e-01 2.87471205e-01
-1.80989408e+00 -3.22110564e-01 1.32399216e-01 2.17628503e+00
7.86765933e-01 6.76221699e-02 2.09897310e-01 2.62558132e-01
1.02499735e+00 5.57484664e-02 -9.15384054e-01 -6.81569218e-04
1.68089867e-01 -9.41944346e-02 7.19063818e-01 1.40455469e-01
-1.45485103e+00 8.07584047e-01 5.57902861e+00 9.71101880e-01
-1.19336021e+00 1.80702329e-01 8.56633782e-01 3.22322428e-01
3.56573254e-01 -5.57029366e-01 -1.31686652e+00 4.83628839e-01
5.49997807e-01 2.31228307e-01 2.50936393e-02 1.18354285e+00
-1.42763257e-01 -3.18649352e-01 -1.04804397e+00 1.52533150e+00
4.08174664e-01 -6.19476438e-01 -5.12548208e-01 1.06277745e-02
4.91460472e-01 -2.16813892e-01 3.10626477e-01 3.54059190e-01
2.05939878e-02 -1.14640081e+00 7.31267273e-01 3.12501311e-01
1.04082787e+00 -9.74618435e-01 8.69582772e-01 5.56292415e-01
-1.17137361e+00 -2.74146736e-01 -4.52686667e-01 9.83336419e-02
7.34479055e-02 6.30555689e-01 -5.85256934e-01 3.44989449e-02
5.51257372e-01 6.31342888e-01 -6.90151036e-01 1.23427343e+00
-3.27733397e-01 4.56062019e-01 -4.30258960e-01 -3.31653543e-02
-1.97475664e-02 -7.35528544e-02 -1.11699998e-01 7.06262589e-01
1.61649674e-01 -1.50880143e-01 4.75816399e-01 3.83909762e-01
-3.56378168e-01 2.29545966e-01 -2.76212215e-01 1.91673130e-01
6.43773317e-01 1.26886559e+00 -8.01443279e-01 -4.53997031e-02
-4.24836099e-01 7.53367066e-01 5.23488641e-01 1.70019135e-01
-7.37444937e-01 -1.30018950e-01 3.51789027e-01 -1.56446904e-01
-5.55745997e-02 -4.33830172e-03 -4.50761408e-01 -9.12698746e-01
6.41583698e-03 -4.08625662e-01 4.98952061e-01 -5.67988098e-01
-1.22900283e+00 2.78000116e-01 6.65980279e-02 -9.30969238e-01
-3.25691313e-01 -7.21106291e-01 -2.49966711e-01 7.40453601e-01
-1.23314142e+00 -1.56388509e+00 -4.70637172e-01 2.34751984e-01
4.17972296e-01 -1.10499226e-01 7.80230343e-01 6.59969151e-01
-8.34291041e-01 1.19917786e+00 2.03941822e-01 3.43572170e-01
8.44910741e-01 -1.16609895e+00 -2.57425189e-01 3.42513859e-01
-1.86165988e-01 1.43228278e-01 4.21214849e-01 -4.98409808e-01
-8.27319980e-01 -9.43237245e-01 7.23767042e-01 8.19861442e-02
9.21511278e-02 -2.59447694e-01 -7.83215284e-01 6.09375000e-01
-4.68471795e-01 2.20853865e-01 7.24431753e-01 3.93248558e-01
-2.96840847e-01 -3.69211733e-01 -1.21136439e+00 1.23599157e-01
8.14215302e-01 -3.43344808e-01 -2.10047856e-01 3.87205690e-01
2.00496525e-01 -7.22132564e-01 -8.33900809e-01 6.27757251e-01
8.52843285e-01 -6.18418455e-01 8.61962020e-01 -1.85163245e-01
3.14307883e-02 4.04233038e-02 -5.67113869e-02 -1.21968520e+00
3.70268547e-03 3.65649104e-01 4.42894734e-02 1.43334711e+00
3.29674274e-01 -5.30706644e-01 1.15541697e+00 9.25553381e-01
2.35700645e-02 -9.90804493e-01 -8.75291646e-01 -5.59362173e-01
1.03512190e-01 -1.43707588e-01 4.68975872e-01 7.06139266e-01
-4.71376091e-01 4.50105995e-01 -4.90137041e-01 2.20044836e-01
1.09028685e+00 -1.20963547e-02 3.87808979e-01 -1.69363594e+00
1.54811621e-01 -1.76884189e-01 -6.11745954e-01 -1.12743711e+00
7.83339441e-01 -7.29508042e-01 2.61298120e-01 -1.56456530e+00
4.87944901e-01 -8.40401292e-01 -2.66435146e-01 6.87180459e-01
-2.22254306e-01 3.23320091e-01 -1.40550062e-01 1.27175748e-01
-8.28598082e-01 5.13799667e-01 1.23805249e+00 -1.37635872e-01
8.83807167e-02 4.26471680e-01 -5.50157070e-01 1.07734442e+00
7.33630955e-01 -4.94928360e-01 -3.56225520e-01 -3.73956710e-01
-1.54127732e-01 -4.34795842e-02 1.10506117e-02 -9.48838353e-01
3.28089535e-01 -3.21000069e-02 7.66172469e-01 -7.76487112e-01
4.55347985e-01 -6.43798828e-01 -1.98176190e-01 2.40395829e-01
-2.81547755e-01 -3.55744720e-01 -6.82148933e-02 3.37754071e-01
-1.21866435e-01 -4.99565899e-01 1.08376908e+00 -1.86998080e-02
-8.24762583e-01 6.38346076e-01 5.07517830e-02 1.11336887e-01
8.41144502e-01 -2.43542597e-01 -2.16803044e-01 -3.31006110e-01
-9.40884233e-01 4.83970374e-01 3.12841177e-01 4.05036688e-01
5.56212127e-01 -1.40306127e+00 -4.41687882e-01 3.50004941e-01
1.75233409e-01 3.98609459e-01 3.47921520e-01 8.85369539e-01
-2.93593943e-01 2.56429881e-01 -9.91044417e-02 -7.88222373e-01
-1.61252034e+00 1.23109259e-01 4.34621096e-01 -1.13390878e-01
-1.98137149e-01 1.07099330e+00 4.11032230e-01 -6.09872103e-01
7.60403156e-01 1.07101358e-01 -5.25941670e-01 3.22550684e-01
7.00452447e-01 2.88249850e-01 -7.43042976e-02 -9.98648345e-01
-3.89649123e-01 9.31370258e-01 -5.04671633e-01 3.18225533e-01
1.18210733e+00 -2.85881221e-01 -1.04156666e-01 2.94244409e-01
1.40995717e+00 -5.52422106e-01 -1.27489483e+00 -4.04741436e-01
-7.64694437e-02 -3.45631540e-01 2.34089568e-01 -7.88668454e-01
-1.30893517e+00 9.84943271e-01 9.77612138e-01 -2.71357000e-01
1.03563333e+00 2.26636201e-01 9.12998021e-01 1.44072190e-01
5.07878542e-01 -1.44359732e+00 1.14267692e-01 3.92425746e-01
2.87164688e-01 -1.68927491e+00 -3.76698747e-02 -4.93645489e-01
-5.30117452e-01 8.76886368e-01 1.01451468e+00 2.86245614e-01
7.53500879e-01 -2.26657037e-02 2.74809241e-01 -2.51506884e-02
8.41129571e-02 -1.57444164e-01 4.00202513e-01 6.16162896e-01
3.77670556e-01 3.10516566e-01 -4.68338907e-01 8.71651232e-01
-1.57954171e-02 -1.05980858e-01 -8.19437951e-03 6.50369525e-01
-6.72559023e-01 -1.07512331e+00 -4.74252671e-01 8.29453945e-01
-6.36789382e-01 3.58641356e-01 5.96453995e-02 5.49100339e-01
3.78226072e-01 8.61860037e-01 5.77513501e-02 -2.86920875e-01
7.40734115e-02 2.16680825e-01 6.80007994e-01 -5.19867241e-01
8.93010423e-02 5.71520850e-02 2.07974557e-02 -1.32008761e-01
-3.99222881e-01 -7.34928787e-01 -1.41641045e+00 -1.75476342e-01
-7.09816396e-01 5.12316711e-02 9.04973865e-01 1.28566062e+00
-2.37572744e-01 2.56628066e-01 5.63015163e-01 -8.49618256e-01
-5.47815561e-01 -1.23001027e+00 -1.12211168e+00 4.23954099e-01
7.03215823e-02 -1.29312134e+00 -2.01047823e-01 -1.04437225e-01]
|
[13.581055641174316, 0.8592618107795715]
|
a42faa4c-18f8-4a67-9d74-be9279418c6f
|
alibabas-submission-for-the-wmt-2020-ape
| null | null |
https://aclanthology.org/2020.wmt-1.84
|
https://aclanthology.org/2020.wmt-1.84.pdf
|
Alibaba’s Submission for the WMT 2020 APE Shared Task: Improving Automatic Post-Editing with Pre-trained Conditional Cross-Lingual BERT
|
The goal of Automatic Post-Editing (APE) is basically to examine the automatic methods for correcting translation errors generated by an unknown machine translation (MT) system. This paper describes Alibaba’s submissions to the WMT 2020 APE Shared Task for the English-German language pair. We design a two-stage training pipeline. First, a BERT-like cross-lingual language model is pre-trained by randomly masking target sentences alone. Then, an additional neural decoder on the top of the pre-trained model is jointly fine-tuned for the APE task. We also apply an imitation learning strategy to augment a reasonable amount of pseudo APE training data, potentially preventing the model to overfit on the limited real training data and boosting the performance on held-out data. To verify our proposed model and data augmentation, we examine our approach with the well-known benchmarking English-German dataset from the WMT 2017 APE task. The experiment results demonstrate that our system significantly outperforms all other baselines and achieves the state-of-the-art performance. The final results on the WMT 2020 test dataset show that our submission can achieve +5.56 BLEU and -4.57 TER with respect to the official MT baseline.
|
['Yu Zhao', 'Yangbin Shi', 'Xin Ge', 'Jun Lu', 'Yuqi Zhang', 'Kai Fan', 'Ke Wang', 'Jiayi Wang']
| null | null | null | null |
wmt-emnlp-2020-11
|
['automatic-post-editing', 'automatic-post-editing']
|
['computer-vision', 'natural-language-processing']
|
[ 3.90410781e-01 2.57674873e-01 -5.55343702e-02 -2.95506030e-01
-1.51692545e+00 -6.28049135e-01 7.97000229e-01 -1.88839629e-01
-6.78234279e-01 9.93281364e-01 6.06295057e-02 -6.33555830e-01
6.49455845e-01 -1.22447275e-01 -1.35462248e+00 -2.80282944e-01
2.89557487e-01 8.01748216e-01 -5.64322956e-02 -3.24866921e-01
3.73388045e-02 -1.20620780e-01 -8.46904457e-01 5.38675129e-01
1.31029391e+00 4.04422790e-01 5.32876790e-01 7.88891971e-01
1.63179770e-01 3.34255785e-01 -6.18325293e-01 -8.52170646e-01
5.12667179e-01 -4.92868513e-01 -8.62671614e-01 -1.46122679e-01
4.31602895e-01 -1.71230778e-01 -2.21344709e-01 1.04674411e+00
6.95630431e-01 -2.93914497e-01 5.35277724e-01 -8.65471661e-01
-9.16123867e-01 9.03309405e-01 -4.84986275e-01 6.79956526e-02
1.58190444e-01 2.79543817e-01 7.18201339e-01 -1.55137682e+00
6.91392004e-01 9.78711784e-01 7.03334928e-01 7.87755728e-01
-1.13285983e+00 -5.17855346e-01 -1.82430089e-01 3.57448794e-02
-1.30280709e+00 -8.26695859e-01 1.51988998e-01 -1.89374745e-01
1.31542027e+00 1.59263492e-01 9.18915272e-02 1.52211750e+00
3.90317857e-01 9.27329719e-01 1.17841268e+00 -7.07577348e-01
-1.73468977e-01 3.36445123e-01 -3.50887597e-01 5.85276067e-01
-3.55565059e-03 1.26684055e-01 -5.34623206e-01 1.35995910e-01
2.25745648e-01 -7.06766903e-01 -2.89833426e-01 9.68443602e-02
-1.77022159e+00 3.23600262e-01 8.67161304e-02 2.78920621e-01
-4.47517008e-01 2.04917677e-02 4.31623548e-01 6.12320840e-01
7.45920122e-01 5.65978885e-01 -9.17542815e-01 -3.58787298e-01
-1.14160979e+00 3.24830937e-04 5.47551692e-01 1.27106547e+00
5.88528812e-01 -1.08976990e-01 -5.79019785e-01 1.11727142e+00
7.51826763e-02 7.12996900e-01 7.10512400e-01 -5.63428998e-01
1.26573944e+00 2.28606880e-01 2.75487781e-01 -1.88772887e-01
1.76081762e-01 -5.45788527e-01 -6.65472806e-01 -1.96748301e-01
4.51278090e-01 -4.35781091e-01 -9.86872315e-01 1.80744314e+00
-4.64842729e-02 -1.16540760e-01 2.88330257e-01 6.83997989e-01
3.63875270e-01 8.98568094e-01 -1.81654304e-01 -2.74855584e-01
1.03739846e+00 -1.57052886e+00 -8.42994452e-01 -6.06011391e-01
1.04633284e+00 -1.30088317e+00 1.31231809e+00 1.02565974e-01
-1.28628051e+00 -5.24095297e-01 -1.08652508e+00 -1.06327087e-01
-2.00755090e-01 7.42813289e-01 -8.38949010e-02 2.92632759e-01
-1.13010907e+00 6.62865639e-01 -8.85570288e-01 -5.27062416e-01
8.79654065e-02 2.43627340e-01 -4.40847695e-01 -2.07576945e-01
-1.34530532e+00 1.43720317e+00 3.16589981e-01 3.65442604e-01
-8.88320863e-01 -6.81333780e-01 -5.68030655e-01 -2.78470159e-01
8.08740258e-02 -5.44243753e-01 1.57990837e+00 -1.11273241e+00
-1.86166012e+00 1.04105318e+00 -3.86433601e-01 -6.07665420e-01
1.02719843e+00 -4.76802945e-01 -4.86945510e-01 -3.84671301e-01
2.93711334e-01 7.86500216e-01 5.89996457e-01 -9.93065476e-01
-5.67329407e-01 6.61973730e-02 -4.09935564e-01 2.94768095e-01
-2.35971794e-01 2.72182822e-01 -5.68725228e-01 -8.60255182e-01
-2.28959128e-01 -1.27325273e+00 -7.99193755e-02 -4.68467116e-01
-5.44182301e-01 4.67270985e-02 4.98149008e-01 -1.22024059e+00
1.19842803e+00 -1.89633119e+00 3.22545260e-01 -2.30462044e-01
-3.53472650e-01 5.34755290e-01 -5.73102534e-01 5.49971581e-01
-1.05389692e-02 1.65602475e-01 -5.31152785e-01 -8.26485455e-01
-6.64194897e-02 1.18092604e-01 -2.81201422e-01 3.05794150e-01
4.40389574e-01 1.15780795e+00 -8.71000350e-01 -1.60060346e-01
-1.40651748e-01 1.50814369e-01 -3.69982958e-01 3.62585098e-01
-2.80725151e-01 7.57668793e-01 1.18659303e-01 5.03765404e-01
6.55089378e-01 3.73023041e-02 1.79530218e-01 1.87964022e-01
-1.97194651e-01 9.05716956e-01 -6.82243824e-01 2.00725460e+00
-5.49848616e-01 8.30649555e-01 -1.74518779e-01 -6.30051494e-01
7.58641839e-01 5.52811682e-01 -8.67363513e-02 -6.96540952e-01
7.16760159e-02 8.54978859e-01 2.52257228e-01 -3.03683460e-01
6.48301125e-01 1.72492191e-01 -9.96476561e-02 5.78640103e-01
2.51298368e-01 -4.84620556e-02 1.48154750e-01 -5.85084595e-02
9.97722924e-01 4.78758931e-01 5.00878207e-02 -4.98183787e-01
5.79344690e-01 1.06294893e-01 6.52471662e-01 6.92941368e-01
-1.68728560e-01 7.71131456e-01 -2.81244814e-02 -1.71111196e-01
-1.46429396e+00 -8.16278100e-01 2.39297654e-02 9.20196891e-01
-2.84954071e-01 -4.97152179e-01 -1.19901359e+00 -8.89047563e-01
-4.02564734e-01 9.75330412e-01 -4.62173581e-01 -1.66807190e-01
-9.38223660e-01 -8.89905512e-01 8.34946454e-01 2.89002359e-01
7.26132095e-01 -9.91357446e-01 1.30750656e-01 4.59086150e-01
-7.82291234e-01 -1.34680235e+00 -9.43614900e-01 7.52114505e-02
-7.46176660e-01 -5.21037757e-01 -8.47986817e-01 -1.09272933e+00
6.10231936e-01 4.72305380e-02 1.12349856e+00 -1.06128842e-01
4.48094696e-01 -1.65041447e-01 -3.28520209e-01 -2.69246459e-01
-1.13672149e+00 5.62075198e-01 2.67122269e-01 -7.33080059e-02
4.75558341e-01 -2.63919294e-01 -2.74258554e-01 3.74766380e-01
-3.07274014e-01 4.17508543e-01 8.77624035e-01 1.06548381e+00
5.95420718e-01 -7.48496950e-01 4.70129102e-01 -6.24822617e-01
5.44420004e-01 -2.57772535e-01 -5.89725494e-01 6.12007141e-01
-6.40011489e-01 2.15232775e-01 6.29420757e-01 -6.41319394e-01
-9.23569858e-01 1.38343489e-02 -1.61684111e-01 -2.88855731e-01
3.34340185e-01 4.57996815e-01 -3.73855352e-01 1.30787686e-01
5.54534912e-01 5.02754927e-01 -2.34701455e-01 -6.95987225e-01
2.89800614e-01 1.11715722e+00 7.15935230e-01 -6.86005592e-01
8.30777407e-01 -4.49634582e-01 -4.96832311e-01 -2.76630670e-01
-6.37813210e-01 9.06936973e-02 -8.14140856e-01 1.66352659e-01
6.90761030e-01 -1.12199056e+00 4.09313031e-02 6.53421819e-01
-1.57358730e+00 -7.17828631e-01 8.34934339e-02 5.68769038e-01
-6.51301086e-01 1.18815593e-01 -9.60259974e-01 -3.30335617e-01
-7.15798199e-01 -1.46028531e+00 1.07074475e+00 -3.12666357e-01
-2.78849602e-01 -7.02142656e-01 3.24795634e-01 4.83967662e-01
4.24794406e-01 -3.63576382e-01 6.69718802e-01 -7.99608052e-01
-4.93925273e-01 -2.78112181e-02 8.71066377e-02 7.08950698e-01
3.73340249e-02 -3.60783152e-02 -8.75898063e-01 -4.43169206e-01
-5.82259819e-02 -2.29790539e-01 6.44995809e-01 -1.58981234e-01
7.04611599e-01 -4.45533365e-01 -1.66827798e-01 5.64505100e-01
1.00734019e+00 -1.49434730e-01 5.97102880e-01 5.68807304e-01
6.12929940e-01 3.88864040e-01 8.32079351e-01 -1.67864293e-01
4.09938604e-01 8.59304607e-01 4.47047502e-02 -3.06237191e-02
-3.29843789e-01 -5.97603798e-01 9.36163425e-01 1.48197830e+00
1.45877087e-02 -4.35233265e-01 -9.81961608e-01 6.77326858e-01
-1.84968615e+00 -6.03181541e-01 -3.58670652e-01 2.40476680e+00
1.23913682e+00 1.05419986e-01 -2.26316959e-01 -2.30242103e-01
9.70173776e-01 -3.35324287e-01 -2.43427783e-01 -6.90065503e-01
-3.56629521e-01 1.90726504e-01 6.50777161e-01 7.51315653e-01
-9.92275000e-01 1.48185766e+00 6.22184610e+00 8.31150353e-01
-1.08086240e+00 5.51867545e-01 4.61055100e-01 5.03555425e-02
-7.10224658e-02 -4.54318859e-02 -7.73190856e-01 6.74006641e-01
1.50475609e+00 -3.05677176e-01 7.32728481e-01 4.57604587e-01
3.42615843e-01 2.66458005e-01 -1.37344110e+00 6.76963806e-01
-3.06313168e-02 -1.07814825e+00 -5.62730385e-03 2.26124778e-01
1.02375340e+00 7.19945610e-01 -1.09437853e-02 5.98899961e-01
2.97666460e-01 -7.77173519e-01 1.03013361e+00 1.57335564e-01
1.09546137e+00 -4.84471768e-01 8.33697498e-01 5.33544362e-01
-6.77965939e-01 2.76281327e-01 -3.92665058e-01 1.32652866e-02
1.25961557e-01 4.67335790e-01 -1.18601811e+00 6.28873229e-01
4.37254369e-01 6.54583871e-01 -6.73033476e-01 8.28777492e-01
-6.49022460e-01 8.97291005e-01 -2.50473589e-01 2.25942448e-01
2.35831141e-01 -1.02098361e-01 7.04754293e-01 1.51397192e+00
6.86069489e-01 -4.09931481e-01 -2.99780756e-01 7.48014927e-01
-7.09210932e-01 2.24389404e-01 -4.85898852e-01 -1.34521738e-01
6.29062891e-01 9.77490366e-01 2.06535123e-02 -5.41107893e-01
-3.97373378e-01 1.73995316e+00 6.36513233e-01 2.70429075e-01
-9.29766774e-01 -1.03922494e-01 6.37488484e-01 -2.37987459e-01
2.28850812e-01 -2.41483122e-01 -2.95406610e-01 -1.47495651e+00
3.99337888e-01 -1.25679386e+00 -1.53269291e-01 -8.51465762e-01
-1.10594141e+00 9.15812850e-01 -4.99391198e-01 -1.48207092e+00
-4.81278688e-01 -5.27814209e-01 -4.32464361e-01 1.40738559e+00
-1.38405597e+00 -1.29256952e+00 3.61091346e-01 2.64232427e-01
8.49484563e-01 -3.33759964e-01 9.52648401e-01 5.36375761e-01
-6.96373224e-01 1.13864470e+00 5.72325885e-01 2.73557007e-01
1.29226804e+00 -1.16186261e+00 1.14883232e+00 1.42103517e+00
1.21204235e-01 5.68157554e-01 6.67230666e-01 -7.15242982e-01
-1.38321757e+00 -1.37727726e+00 1.71940434e+00 -9.78950143e-01
6.70620918e-01 -7.65870810e-01 -1.00186074e+00 9.69887972e-01
5.31126738e-01 -1.08476274e-01 2.27361962e-01 -1.43876895e-01
-2.03027233e-01 1.51104018e-01 -9.92573559e-01 7.50652671e-01
9.87784088e-01 -6.05358183e-01 -9.14222300e-01 4.53090340e-01
8.68607819e-01 -6.67109489e-01 -7.65320003e-01 4.51832086e-01
4.33255792e-01 -2.84527600e-01 5.62092125e-01 -6.79589033e-01
6.98471308e-01 -2.38468051e-01 -3.13970983e-01 -1.84013236e+00
-8.16717595e-02 -1.01292598e+00 9.21874195e-02 1.42713523e+00
1.02806377e+00 -4.96203363e-01 2.53441781e-01 2.72516161e-01
-4.29092675e-01 -5.37720561e-01 -1.01471043e+00 -1.03026462e+00
6.10259056e-01 -2.33738109e-01 5.07683396e-01 1.02710772e+00
-4.40789424e-02 6.20234728e-01 -6.53398871e-01 2.21600190e-01
3.92971516e-01 -3.42033297e-01 9.10181761e-01 -5.14476776e-01
-4.71446902e-01 -2.53647268e-01 2.51266629e-01 -1.21267092e+00
9.76740494e-02 -1.27528465e+00 4.19212073e-01 -1.28651929e+00
2.44865745e-01 -2.12872148e-01 -1.03005074e-01 5.51791787e-01
-6.45962596e-01 3.74867558e-01 2.07739174e-01 4.87648189e-01
-2.65211374e-01 5.26567698e-01 1.25975931e+00 -2.37765253e-01
-9.20116007e-02 -1.55417055e-01 -4.20479685e-01 2.70883143e-01
8.38402867e-01 -7.51959622e-01 1.88913476e-02 -1.23609221e+00
-1.94655359e-02 -1.68687239e-01 -4.54772785e-02 -7.17994452e-01
9.87585187e-02 1.10420629e-01 7.54617751e-02 -4.56378043e-01
-5.00294492e-02 -5.24506032e-01 1.73554774e-02 4.25538480e-01
-3.88891935e-01 4.90255535e-01 3.82836372e-01 1.87521372e-02
-1.02942668e-01 -1.04600169e-01 6.72809720e-01 6.23300150e-02
-9.32386667e-02 1.12341285e-01 -4.96475965e-01 1.57274604e-01
4.91670132e-01 9.41797346e-02 -4.37900394e-01 -1.06534988e-01
-4.28461403e-01 3.08079809e-01 5.67713976e-01 7.56683588e-01
2.02421434e-02 -1.43152773e+00 -1.31545103e+00 2.34925076e-01
3.40382427e-01 -3.78358424e-01 -2.54381657e-01 9.71372187e-01
-3.63725841e-01 4.54541445e-01 -6.88170195e-02 -5.87212384e-01
-1.03318560e+00 4.84404445e-01 4.75581378e-01 -4.74730670e-01
-4.11695987e-01 6.52444899e-01 -1.42296195e-01 -9.97382879e-01
-6.12280518e-02 -1.56929821e-01 5.09084940e-01 -5.34360588e-01
4.11129475e-01 2.49218896e-01 5.27018249e-01 -6.84910536e-01
-3.99149090e-01 2.67894000e-01 -2.67767459e-01 -5.87842762e-01
1.16667914e+00 -3.24389100e-01 -1.42742425e-01 3.52094591e-01
1.19498897e+00 2.20755741e-01 -9.80040193e-01 -4.81863528e-01
2.77631104e-01 -1.63636550e-01 -2.04096362e-01 -1.44469738e+00
-4.48565066e-01 1.11755121e+00 4.48852003e-01 -5.10059476e-01
8.38347435e-01 -3.64167154e-01 8.76404583e-01 5.44504941e-01
4.63973552e-01 -1.20519805e+00 -3.33421767e-01 9.53325450e-01
1.11202633e+00 -1.37859106e+00 -5.39202094e-01 -2.05188736e-01
-6.21825516e-01 9.38019276e-01 5.96745312e-01 1.20687753e-01
-6.10248446e-02 2.74750084e-01 3.58603716e-01 6.04274392e-01
-9.21670854e-01 2.15336934e-01 3.54069084e-01 3.69888872e-01
6.60310447e-01 1.33743256e-01 -6.75768435e-01 6.55853689e-01
-4.30605859e-01 8.94733369e-02 5.68666875e-01 7.34033287e-01
-1.29266247e-01 -1.54513085e+00 -1.75933748e-01 -5.31613789e-02
-6.73794031e-01 -6.62288427e-01 -6.04470074e-01 6.32876515e-01
-9.91169810e-02 9.45344090e-01 -1.48306876e-01 -5.18156707e-01
3.73040736e-01 4.55318451e-01 6.28732622e-01 -6.09182537e-01
-8.66246402e-01 1.34914264e-01 3.63075584e-01 -3.58430833e-01
-1.88208353e-02 -6.50682449e-01 -8.45188737e-01 -3.74503225e-01
-2.40069672e-01 2.37093732e-01 8.65803540e-01 1.01616931e+00
6.36530459e-01 3.06528658e-01 6.33629203e-01 -7.48367369e-01
-7.95527220e-01 -1.59928167e+00 2.66716033e-01 1.63391605e-01
2.42598668e-01 -1.00163721e-01 -2.88283229e-01 1.77822620e-01]
|
[11.649972915649414, 10.283778190612793]
|
39e2777e-7a8b-468d-a743-202490015910
|
a-medical-semantic-assisted-transformer-for
|
2208.10358
| null |
https://arxiv.org/abs/2208.10358v1
|
https://arxiv.org/pdf/2208.10358v1.pdf
|
A Medical Semantic-Assisted Transformer for Radiographic Report Generation
|
Automated radiographic report generation is a challenging cross-domain task that aims to automatically generate accurate and semantic-coherence reports to describe medical images. Despite the recent progress in this field, there are still many challenges at least in the following aspects. First, radiographic images are very similar to each other, and thus it is difficult to capture the fine-grained visual differences using CNN as the visual feature extractor like many existing methods. Further, semantic information has been widely applied to boost the performance of generation tasks (e.g. image captioning), but existing methods often fail to provide effective medical semantic features. Toward solving those problems, in this paper, we propose a memory-augmented sparse attention block utilizing bilinear pooling to capture the higher-order interactions between the input fine-grained image features while producing sparse attention. Moreover, we introduce a novel Medical Concepts Generation Network (MCGN) to predict fine-grained semantic concepts and incorporate them into the report generation process as guidance. Our proposed method shows promising performance on the recently released largest benchmark MIMIC-CXR. It outperforms multiple state-of-the-art methods in image captioning and medical report generation.
|
['Luping Zhou', 'Xiu Li', 'Lei Wang', 'Mingkang Tang', 'Zhanyu Wang']
|
2022-08-22
| null | null | null | null |
['medical-report-generation']
|
['medical']
|
[ 4.76170510e-01 2.77255058e-01 -2.35869139e-01 -5.15505970e-01
-1.39661539e+00 -4.22006473e-02 5.45469224e-01 2.42624223e-01
-3.93058360e-02 8.61026406e-01 8.36958885e-01 -5.23749851e-02
-1.66922826e-02 -7.16891110e-01 -8.39674473e-01 -6.42691255e-01
2.41320640e-01 3.43575537e-01 4.96238023e-02 -1.40289351e-01
3.12753171e-01 1.14794768e-01 -1.43091869e+00 7.47008502e-01
9.63250816e-01 1.02613163e+00 5.59576750e-01 4.90672708e-01
-2.51847148e-01 1.16712809e+00 -6.34569228e-01 -2.88406640e-01
-2.71495849e-01 -7.54216254e-01 -1.08528686e+00 2.72249699e-01
3.70260715e-01 -3.00901681e-01 -3.49938542e-01 1.07755947e+00
6.04377925e-01 6.02055416e-02 8.36962938e-01 -8.56884897e-01
-1.21052849e+00 5.25995910e-01 -6.58126116e-01 2.25145102e-01
4.47244078e-01 1.33736327e-01 8.61831427e-01 -7.77098298e-01
8.27181876e-01 1.16736889e+00 2.49272212e-01 7.36021280e-01
-9.35224891e-01 -6.38376296e-01 1.65403888e-01 1.88433334e-01
-1.21957088e+00 -9.52733029e-03 8.36254656e-01 -5.41027486e-01
5.25621176e-01 4.32732463e-01 4.98466522e-01 1.17071867e+00
3.46436203e-01 8.92148852e-01 1.13008344e+00 -1.63685620e-01
3.28246830e-03 7.10023269e-02 -1.92768827e-01 8.46086800e-01
6.87862560e-02 -1.66526109e-01 -3.38169694e-01 -1.92278922e-01
9.03534591e-01 1.76658958e-01 -5.34059942e-01 4.48242277e-02
-1.59119999e+00 1.06972063e+00 1.02685308e+00 4.98783827e-01
-6.49532676e-01 1.08983129e-01 2.25933805e-01 -2.27471784e-01
5.42254508e-01 6.13863468e-01 1.22587338e-01 2.50819713e-01
-9.01222050e-01 4.95859861e-01 2.60546535e-01 7.45593250e-01
5.27546823e-01 -2.11432159e-01 -9.62330043e-01 9.31490302e-01
7.85959288e-02 2.65005291e-01 8.10166955e-01 -5.47146201e-01
7.72278666e-01 6.40754104e-01 -2.17953753e-02 -1.15556479e+00
-3.87928605e-01 -5.98702431e-01 -1.18942320e+00 -1.52393296e-01
-6.51904419e-02 2.38269806e-01 -1.34727514e+00 1.53736329e+00
1.15908772e-01 1.01709880e-01 4.05525863e-02 1.29045093e+00
1.39959538e+00 7.60061204e-01 3.83828610e-01 -1.04281001e-01
1.62405872e+00 -1.22011304e+00 -8.46503258e-01 -1.77991465e-01
3.89436781e-01 -8.50559831e-01 1.02826107e+00 -7.15638474e-02
-1.25049114e+00 -5.91521859e-01 -9.07251716e-01 -1.61944360e-01
-1.13242619e-01 1.81479290e-01 8.03572834e-01 -4.30642553e-02
-9.82344389e-01 8.31409916e-02 -6.16196394e-01 -6.30519465e-02
6.86797917e-01 2.04733774e-01 -4.01769280e-01 -5.68161130e-01
-1.19016802e+00 7.45253205e-01 5.05126595e-01 1.08555786e-01
-9.50083673e-01 -8.78358305e-01 -1.12087905e+00 1.06620513e-01
1.99519053e-01 -1.13407528e+00 1.28278434e+00 -7.40305901e-01
-1.00835860e+00 1.05586660e+00 -6.80063143e-02 -2.62953192e-01
3.68674397e-01 1.45519346e-01 -1.87242478e-01 3.32881778e-01
4.87831056e-01 1.17037416e+00 6.22545838e-01 -1.43938160e+00
-3.94757986e-01 -2.63074160e-01 8.75524804e-02 3.77812415e-01
-1.26418620e-01 -9.48160365e-02 -5.84526658e-01 -1.20874047e+00
3.43663506e-02 -8.73010337e-01 -6.25526249e-01 -6.11989647e-02
-6.77377403e-01 -2.63904124e-01 3.54111671e-01 -7.53815353e-01
1.11449230e+00 -1.91273260e+00 1.15921006e-01 -6.86419457e-02
3.72694373e-01 2.31245384e-02 -1.37171924e-01 1.64073333e-01
-2.30247110e-01 -3.09280981e-03 -6.25872612e-01 -4.23008651e-01
-3.93511921e-01 1.62468683e-02 -2.69580543e-01 3.22858617e-02
5.01079500e-01 1.16197562e+00 -1.03308749e+00 -8.54995251e-01
1.01025179e-01 5.30573487e-01 -5.63979566e-01 6.04142725e-01
-2.46362224e-01 9.00092244e-01 -7.63095677e-01 7.02653408e-01
5.49820244e-01 -7.72491097e-01 -2.93302178e-01 -4.38946217e-01
2.32372701e-01 2.35041097e-01 -5.43357134e-01 2.21411085e+00
-7.29847431e-01 2.32733563e-01 -3.42307329e-01 -9.99484658e-01
8.07827771e-01 3.85157973e-01 6.35600865e-01 -8.22046638e-01
4.96870503e-02 2.60227650e-01 -1.25232607e-01 -6.69530511e-01
4.05906171e-01 -4.20546055e-01 -1.88794807e-01 4.85878587e-01
7.14997994e-03 -4.13147807e-01 1.45180582e-03 3.98262501e-01
9.74549234e-01 -3.04789633e-01 3.29232812e-01 1.61047921e-01
6.46189451e-01 2.27849737e-01 2.96002030e-01 6.52696729e-01
9.88133997e-02 1.42807162e+00 2.25735962e-01 -4.08444405e-01
-7.82449543e-01 -9.72889483e-01 -4.85707596e-02 6.48934305e-01
2.54304290e-01 -1.61395401e-01 -7.13079870e-01 -8.99835944e-01
-2.14576244e-01 3.46874893e-01 -9.14828360e-01 -2.67210901e-01
-5.38947821e-01 -9.58989680e-01 1.62716180e-01 7.27163672e-01
4.60038334e-01 -1.52134907e+00 -2.26972982e-01 3.44667107e-01
-6.94233656e-01 -1.19349360e+00 -8.54635656e-01 -2.83562094e-01
-6.62595212e-01 -9.91095364e-01 -1.38783753e+00 -1.01856089e+00
1.01582062e+00 2.50514507e-01 1.35432208e+00 3.12949121e-01
-6.58974409e-01 1.26123443e-01 -4.59811300e-01 -4.02973980e-01
-4.95385975e-01 1.41235486e-01 -5.72132051e-01 -6.02132035e-03
-7.74431825e-02 -5.35624772e-02 -1.03971004e+00 -1.19198173e-01
-1.18436539e+00 5.92297494e-01 1.01636004e+00 1.19005239e+00
8.94785583e-01 -2.20560580e-01 5.21707356e-01 -1.02027500e+00
7.60391891e-01 -7.53942311e-01 4.13992330e-02 2.67547756e-01
-3.48772764e-01 2.26194620e-01 4.87205207e-01 -1.60799176e-01
-1.16839540e+00 -4.67356928e-02 -4.48707968e-01 -2.13229090e-01
-1.87154979e-01 5.55708230e-01 2.09967077e-01 7.96391740e-02
3.36546957e-01 6.40227377e-01 -5.80360107e-02 -2.95168728e-01
3.00064385e-01 5.53399742e-01 6.61861777e-01 -4.02110994e-01
5.98600626e-01 6.08708501e-01 -1.38917893e-01 -2.04592243e-01
-1.52306235e+00 -4.65619564e-01 -1.83357388e-01 -6.83528930e-02
1.31842268e+00 -1.06156361e+00 -3.83217663e-01 2.26339325e-01
-1.37431824e+00 2.78861314e-01 -1.64929524e-01 4.75529015e-01
-7.54133821e-01 1.43091559e-01 -6.72165453e-01 -2.82630801e-01
-7.55362928e-01 -1.73109496e+00 1.56195581e+00 3.07228804e-01
-1.00563943e-01 -8.20606411e-01 4.61752824e-02 7.31391072e-01
4.29565489e-01 4.74081665e-01 1.01906943e+00 -1.72324151e-01
-6.82161629e-01 1.28849000e-01 -6.96632445e-01 2.56419659e-01
3.18633646e-01 -6.07627690e-01 -7.32704461e-01 -2.09759444e-01
-6.35991693e-02 -4.46113586e-01 1.13056493e+00 6.01727426e-01
1.74878037e+00 -4.40528691e-01 -4.17702436e-01 6.40524983e-01
1.24383724e+00 1.37400106e-01 6.29346013e-01 1.91518351e-01
9.06829596e-01 6.04698479e-01 7.11634815e-01 3.14850628e-01
7.15517759e-01 7.26537943e-01 4.94812071e-01 -7.55422771e-01
-4.35420126e-01 -4.14676547e-01 -1.49636045e-01 8.57022583e-01
1.06539629e-01 -7.10710734e-02 -7.77693152e-01 7.80352414e-01
-1.79418314e+00 -6.64840758e-01 2.14875713e-02 1.65387368e+00
1.17727768e+00 -1.35641783e-01 -2.33606756e-01 -2.20628500e-01
6.02058053e-01 3.16545039e-01 -3.68877113e-01 -6.44489527e-02
4.46652099e-02 3.09316546e-01 2.39567503e-01 1.94493011e-01
-1.14359343e+00 6.44629300e-01 5.42894793e+00 8.78862143e-01
-1.18960094e+00 2.54296869e-01 1.11353695e+00 6.34152219e-02
-5.91966629e-01 -4.40900266e-01 -4.98474211e-01 5.74077964e-01
4.39068168e-01 -1.50137199e-02 -2.46443778e-01 7.16156781e-01
7.92471766e-02 5.17356433e-02 -8.36125851e-01 1.19677782e+00
5.75542688e-01 -1.81279004e+00 5.45481503e-01 -1.06617786e-01
1.12407315e+00 -2.84785748e-01 3.24512988e-01 1.09754108e-01
2.06664488e-01 -1.26938200e+00 4.49349910e-01 6.67060792e-01
1.00673103e+00 -6.95302904e-01 1.03788459e+00 -3.94318253e-02
-9.91613328e-01 1.33495599e-01 -2.77500898e-01 3.92838180e-01
2.91399091e-01 5.65143883e-01 -9.85831439e-01 6.28572822e-01
5.75317264e-01 6.12783492e-01 -5.48340380e-01 1.18883240e+00
-1.40928730e-01 3.78019094e-01 1.83862254e-01 1.69012189e-01
6.09796643e-01 2.31647775e-01 1.67550966e-01 1.13472927e+00
4.94985193e-01 3.07341754e-01 2.56960899e-01 1.07132900e+00
-2.39632383e-01 2.88429350e-01 -4.07440960e-01 1.66323371e-02
-1.30346686e-01 1.13426578e+00 -7.72948027e-01 -5.09162426e-01
-4.35409099e-01 1.03760791e+00 2.02410892e-01 1.88618124e-01
-7.54802644e-01 -1.19543318e-02 4.75352883e-01 3.22872400e-01
6.67146519e-02 2.18747795e-01 -2.18791544e-01 -1.19960880e+00
6.75310120e-02 -8.53511691e-01 5.58270633e-01 -9.24689054e-01
-1.44888353e+00 8.99332881e-01 -2.27250427e-01 -1.46236730e+00
-3.55325311e-01 -2.54004002e-01 -5.03568649e-01 8.02815378e-01
-1.74139094e+00 -1.43142700e+00 -6.80988193e-01 6.67022586e-01
8.52694333e-01 2.39603985e-02 8.61371577e-01 3.60878378e-01
-1.48107320e-01 5.10681748e-01 -2.90648699e-01 1.74487054e-01
7.17893958e-01 -1.16155910e+00 2.20366970e-01 4.79307026e-01
1.02273099e-01 4.76622134e-01 4.88548100e-01 -6.23024464e-01
-8.61293137e-01 -1.33759606e+00 7.21368372e-01 -2.47246623e-01
3.49001497e-01 -1.78237721e-01 -9.31267619e-01 3.16042334e-01
1.17068350e-01 2.72158265e-01 6.55021369e-01 -4.19581175e-01
-1.84141621e-01 9.14564133e-02 -9.84992146e-01 4.83089805e-01
8.82275939e-01 -3.99856478e-01 -5.38768351e-01 6.34797037e-01
1.05931497e+00 -6.78071856e-01 -8.89740467e-01 5.27942598e-01
2.23611131e-01 -6.86474323e-01 1.13739693e+00 -4.95195210e-01
1.01849461e+00 -3.24370950e-01 7.56936744e-02 -1.28625023e+00
-2.68023908e-01 -1.28826469e-01 2.30949238e-01 9.75179970e-01
3.50621253e-01 -1.85525835e-01 7.30235338e-01 2.02601075e-01
-3.68323803e-01 -1.14910018e+00 -6.87366545e-01 -1.34881318e-01
4.99609858e-02 -2.79186606e-01 6.63958788e-01 8.66457760e-01
-3.26633155e-01 2.26250783e-01 -4.76038873e-01 -2.60354262e-02
4.70672607e-01 4.20061141e-01 3.17818522e-01 -8.80213976e-01
-2.86150932e-01 -3.48943502e-01 -4.45662141e-01 -8.61076832e-01
7.97531083e-02 -1.09796727e+00 1.56798646e-01 -2.12859964e+00
8.33702922e-01 -5.02077639e-01 -5.23882329e-01 4.18112606e-01
-7.44861245e-01 5.95601082e-01 2.05056652e-01 2.43203357e-01
-6.72264576e-01 7.53156066e-01 1.85076368e+00 -5.77379644e-01
1.88836575e-01 8.26713908e-03 -1.00088608e+00 3.90802771e-01
3.96974027e-01 -6.01635456e-01 -3.61288965e-01 -3.31593484e-01
-4.85454276e-02 4.80800390e-01 5.32356501e-01 -9.80503619e-01
1.86837446e-02 -1.55211821e-01 4.39721882e-01 -7.77096927e-01
1.63527951e-01 -4.59965348e-01 -6.39052540e-02 4.30369079e-01
-5.29272258e-01 8.27197656e-02 -3.00941523e-02 5.15616000e-01
-8.60226214e-01 4.90791984e-02 7.37887681e-01 -5.62245071e-01
-5.46166599e-01 6.71475470e-01 2.73431111e-02 1.03819244e-01
9.62672889e-01 1.48836628e-01 -3.15590411e-01 -3.55597645e-01
-6.75572634e-01 1.83991566e-01 1.57011375e-01 8.34106624e-01
1.00479305e+00 -1.44273722e+00 -1.14508033e+00 6.71658590e-02
6.41874254e-01 3.11716378e-01 8.70682359e-01 8.07650506e-01
-4.80083615e-01 6.13419414e-01 -9.60910097e-02 -6.28136575e-01
-1.06680250e+00 5.13278365e-01 1.79803029e-01 -8.22218299e-01
-6.78269327e-01 1.04197538e+00 8.54279697e-01 -1.94746211e-01
-3.53192650e-02 -4.25861895e-01 -3.38753253e-01 -1.91521525e-01
7.87542760e-01 -4.24903095e-01 1.49512991e-01 -6.43271327e-01
-2.66247809e-01 6.58233881e-01 -5.02126575e-01 1.73033893e-01
1.37995696e+00 -2.30592005e-02 -7.93606415e-02 5.21713197e-02
1.36046493e+00 -3.77552122e-01 -9.60459054e-01 -2.07669452e-01
-2.68818349e-01 -3.21245313e-01 5.04742227e-02 -9.23171520e-01
-1.31545913e+00 8.79740596e-01 6.12026453e-01 -2.49902681e-01
1.25724089e+00 4.87449646e-01 1.12903452e+00 -1.73021525e-01
1.73581168e-01 -7.01734781e-01 3.91733706e-01 7.04012997e-03
1.34193766e+00 -1.66696835e+00 -8.24136734e-02 -4.97319818e-01
-9.38145995e-01 7.92764723e-01 7.41481721e-01 -1.15941338e-01
3.44261169e-01 4.66626622e-02 1.61108479e-01 -4.55356002e-01
-6.25617504e-01 -2.67613381e-01 7.23045290e-01 5.45788109e-01
7.17521727e-01 7.53876343e-02 -4.98878926e-01 6.53299868e-01
-1.26125798e-01 2.12458149e-02 4.47704285e-01 6.44811153e-01
-1.64189681e-01 -8.43121290e-01 -4.78784084e-01 7.01394141e-01
-8.25621963e-01 -2.89188564e-01 -2.91619729e-02 5.09167016e-01
1.01188630e-01 7.11090207e-01 -4.75150123e-02 -1.55137256e-01
2.97394574e-01 -4.85597581e-01 2.76088655e-01 -1.01306200e+00
-4.76520717e-01 -1.84702396e-01 -2.37772420e-01 -5.87351799e-01
-3.93457204e-01 -3.72489750e-01 -1.27131140e+00 3.28190982e-01
6.64920136e-02 1.12047225e-01 6.38309300e-01 9.27408934e-01
3.01095009e-01 1.06843364e+00 4.76415515e-01 -5.94340622e-01
-3.80489945e-01 -1.00135398e+00 -2.94526547e-01 8.90018642e-01
4.34255630e-01 -7.79083490e-01 -1.68548860e-02 1.84720576e-01]
|
[15.030987739562988, -1.4269849061965942]
|
fd4578de-d370-4776-a669-aba24638923c
|
unveiling-the-three-dimensional-spin-texture
|
2101.12630
| null |
https://arxiv.org/abs/2101.12630v1
|
https://arxiv.org/pdf/2101.12630v1.pdf
|
Unveiling the three-dimensional spin texture of skyrmion tubes
|
Magnetic skyrmions are stable topological solitons with complex non-coplanar spin structures. Their nanoscopic size and the low electric currents required to initiate and control their motion has opened a new field of research, skyrmionics, that aims at using skyrmions as information carriers for data storage and manipulation. Recent advances in skyrmionics prompt for a thorough understanding of the detailed three-dimensional (3D) spin texture of a skyrmion including skyrmion-skyrmion interactions and their coupling to surfaces and interfaces. These properties crucially affect application-related aspects such as the stability and mobility of skyrmions in confined structures. To date, however, experimental techniques to measure the three-dimensional spin texture with nanometer resolution are largely missing. We therefore adapt holographic vector field electron tomography to the problem and report on the first quantitative reconstruction of the 3D spin texture of skyrmions with sub-10 nanometer resolution. The reconstructed textures reveal a variety of previously unseen local deviations from a homogeneous Bloch character within the skyrmion tubes (SkTs), details of the collapse of the skyrmion texture at surfaces, and a correlated modulation of the SkT in FeGe along their tube axes. The quantitative 3D data of the magnetic induction also allow to experimentally confirm some principles of skyrmion formation by deriving spatially resolved maps of the magnetic energy density across these magnetic solitons.
|
['Axel Lubk', 'Bernd Rellinghaus', 'Bernd Büchner', 'Rafal E. Dunin-Borkowski', 'Marcus Schmidt', 'András Kovács', 'Ulrich K. Rößler', 'Sebastian Schneider', 'Daniel Wolf']
|
2021-01-29
| null | null | null | null |
['electron-tomography']
|
['medical']
|
[ 2.58181363e-01 -4.24682721e-02 -4.21516076e-02 -1.42839879e-01
2.83785313e-02 -2.97972083e-01 7.26487637e-01 -5.23814797e-01
-4.57572252e-01 9.08475578e-01 -2.44841608e-03 -2.11215064e-01
-2.28550911e-01 -8.11974466e-01 -6.35118246e-01 -1.50046909e+00
-4.47007269e-02 1.11068869e+00 4.79389220e-01 -4.85489786e-01
8.09314549e-01 3.85529697e-01 -1.72815144e+00 3.29780996e-01
5.97835541e-01 9.33889151e-01 4.48525399e-01 6.51675045e-01
-1.79228242e-02 3.49349678e-01 -3.21828097e-01 -6.39078319e-02
-1.52038738e-01 -5.77243030e-01 -5.10365069e-01 -2.05221966e-01
1.33982062e-01 1.27976269e-01 -2.01513991e-01 9.44178343e-01
3.07263166e-01 -3.61389630e-02 7.07286298e-01 -4.10100490e-01
-7.06240058e-01 2.34482795e-01 -3.53239864e-01 1.61080778e-01
2.28159484e-02 1.05360828e-01 4.92371231e-01 -6.16476536e-01
1.32464910e+00 7.23971188e-01 2.12550908e-01 7.53038466e-01
-8.90666962e-01 -3.62523347e-01 -9.68580484e-01 3.24014932e-01
-1.02109468e+00 -6.27277792e-01 6.88460052e-01 -3.54949147e-01
1.10673869e+00 6.22148395e-01 5.55232882e-01 5.88103294e-01
5.84324002e-01 -2.39571765e-01 1.76550257e+00 -8.32571745e-01
4.40842181e-01 2.90648699e-01 1.00932643e-01 3.69479269e-01
7.53607810e-01 -2.17887592e-02 -6.45253539e-01 -1.74777620e-02
7.38002777e-01 -2.64896840e-01 -2.12299153e-01 -7.28717983e-01
-1.02221191e+00 1.10593006e-01 5.86559847e-02 9.43723738e-01
-4.03651863e-01 1.93266287e-01 -8.78669098e-02 4.98651080e-02
1.46301761e-01 8.04801345e-01 -2.79873051e-02 -5.02551615e-01
-5.93095422e-01 1.18812613e-01 6.02544785e-01 3.46966177e-01
5.58693290e-01 -2.39205703e-01 2.55370826e-01 4.23554242e-01
4.20193896e-02 1.32951391e+00 4.59894657e-01 -4.55072582e-01
3.12848836e-01 3.55078429e-01 4.67480600e-01 -6.16925478e-01
-4.36586559e-01 -4.78472896e-02 -6.93626225e-01 6.05618283e-02
1.09827111e-03 1.11337282e-01 -1.02101517e+00 1.18630576e+00
3.77557188e-01 -2.56123155e-01 1.49156615e-01 9.86774385e-01
7.30496645e-01 4.88899797e-01 -7.79993296e-01 -4.59178478e-01
1.11989725e+00 -1.09162003e-01 -5.47267973e-01 4.00067747e-01
4.95806932e-01 -2.77271450e-01 7.29644716e-01 -4.43794467e-02
-1.32016158e+00 3.24367911e-01 -9.54787076e-01 6.02474093e-01
-1.17174834e-01 -2.37240329e-01 5.54467320e-01 7.03130305e-01
-6.07063890e-01 8.90700758e-01 -1.08694553e+00 -3.61514956e-01
-3.05727590e-03 6.70841992e-01 -4.59243059e-01 2.59633631e-01
-8.24181855e-01 7.30272353e-01 -3.49655226e-02 -9.45898667e-02
-2.72104368e-02 -1.99601799e-01 3.67973782e-02 -4.46899772e-01
-1.58091456e-01 -4.05044079e-01 9.88325477e-01 -3.17410350e-01
-1.87056148e+00 1.12556612e+00 -3.61060679e-01 -3.88635159e-01
-3.01378399e-01 4.71713632e-01 -6.58295631e-01 5.39769113e-01
2.22088099e-01 -2.68841267e-01 4.71655607e-01 -1.08099735e+00
1.63138226e-01 -5.48321545e-01 -8.93864036e-01 -1.88004047e-01
-2.03320578e-01 4.66509834e-02 1.85922325e-01 3.08002979e-01
6.42798007e-01 -1.02958333e+00 1.07304463e-02 -1.15261447e+00
-5.09912074e-01 6.08928315e-02 1.08466887e+00 2.27934808e-01
7.15466738e-01 -1.76566112e+00 8.14192370e-02 2.55462736e-01
2.98594147e-01 5.95600784e-01 3.04744393e-01 8.73772442e-01
1.94660366e-01 1.43719688e-01 -1.99300081e-01 2.32828438e-01
-2.11004168e-01 2.38996302e-03 -2.25985393e-01 8.23549867e-01
-3.58424038e-01 1.07878566e+00 -6.26370668e-01 -1.10026104e-02
1.98943526e-01 3.89921337e-01 -5.90501010e-01 -1.42462671e-01
-9.45548192e-02 7.45386004e-01 -8.66887748e-01 4.50996399e-01
1.17070615e+00 -5.85600972e-01 5.63620627e-01 2.19096430e-02
-9.76710081e-01 6.77423894e-01 -4.79599625e-01 4.69947249e-01
5.91292232e-02 5.43023646e-01 5.22899628e-01 -9.92642999e-01
6.06636226e-01 2.90058374e-01 4.58726168e-01 -1.73056197e+00
-1.27163872e-01 8.19303215e-01 4.02250439e-02 -6.47373617e-01
6.47029638e-01 -1.06104231e+00 -7.09149987e-02 6.40079081e-01
-1.44991294e-01 -4.93267775e-01 1.44094718e-03 -1.54345453e-01
1.15239799e+00 -5.51209092e-01 -2.78667003e-01 -6.86112165e-01
1.75267741e-01 -8.10138583e-02 9.20111239e-02 8.76440465e-01
5.62036395e-01 6.22363746e-01 3.93875569e-01 -6.54103816e-01
-1.25975466e+00 -9.32252824e-01 -7.78910935e-01 2.45226905e-01
6.74853265e-01 -3.34918499e-01 -7.61259377e-01 1.82610199e-01
-6.31160960e-02 3.82650048e-01 -4.07553822e-01 -1.88151449e-02
-7.91039705e-01 -1.40716481e+00 9.94847342e-02 -2.10696936e-01
6.49810612e-01 -1.26239371e+00 -9.65038896e-01 1.64297298e-02
5.51472371e-03 -1.05170834e+00 5.83710313e-01 1.17375448e-01
-8.17561269e-01 -1.08854687e+00 -4.53699946e-01 -6.49215281e-02
7.45842874e-01 8.91047418e-02 1.01376891e+00 1.48756459e-01
-4.78967667e-01 3.92320722e-01 -1.85592726e-01 -1.03426374e-01
-4.16613042e-01 -1.35282695e-01 8.29384863e-01 -1.65664643e-01
2.68775046e-01 -9.28159118e-01 -5.74446440e-01 8.05037081e-01
-6.94129586e-01 3.76707092e-02 3.84746850e-01 2.91573912e-01
5.32141626e-01 -1.13601357e-01 4.52000409e-01 -6.35620415e-01
2.89154559e-01 -3.67969215e-01 -9.15825784e-01 5.08424230e-02
-1.67950064e-01 2.52493203e-01 6.77819610e-01 1.72633111e-01
-9.88773048e-01 -7.78878987e-01 -3.08090318e-02 3.47985208e-01
-1.88597403e-02 7.45718256e-02 3.50228906e-01 -4.47231263e-01
4.14491653e-01 3.89422268e-01 -1.19134679e-01 -3.82943809e-01
-7.32540339e-02 9.68906939e-01 4.11235988e-01 -3.98302466e-01
6.03671432e-01 1.14808929e+00 5.41928530e-01 -1.84653544e+00
-5.19554794e-01 -3.66395593e-01 -3.13505769e-01 -4.01498266e-02
8.50957572e-01 -2.28041932e-01 -1.19137633e+00 5.48538387e-01
-8.82315099e-01 -3.22795004e-01 -3.73840839e-01 5.60415745e-01
-5.75148523e-01 9.91143435e-02 -5.02871335e-01 -9.12835598e-01
-2.37317845e-01 -7.32303202e-01 1.01451230e+00 2.34504282e-01
9.77137983e-02 -9.15184081e-01 4.78744775e-01 5.08664310e-01
1.04232097e+00 1.66381240e-01 1.18977952e+00 9.12008062e-02
-1.07906413e+00 -2.45472297e-01 1.42630950e-01 -1.43736720e-01
-1.46840168e-02 -4.33252186e-01 -5.17333686e-01 -3.49540152e-02
6.05688512e-01 -2.64085293e-01 1.10928440e+00 6.37557268e-01
3.63044232e-01 -6.37929216e-02 -8.06195796e-01 5.48709154e-01
1.22161674e+00 6.08629780e-03 8.24609876e-01 3.84688139e-01
5.61485112e-01 1.62429854e-01 3.03096980e-01 2.52330035e-01
-7.24100545e-02 9.81524408e-01 3.08009386e-01 3.88702482e-01
1.02686852e-01 2.38214761e-01 3.20543736e-01 1.36654663e+00
-6.83687031e-01 -1.97342858e-01 -8.88027966e-01 2.36938968e-01
-1.42641044e+00 -1.21275532e+00 -6.35955930e-01 2.62077284e+00
4.20482665e-01 -1.32351671e-03 -3.32267106e-01 -2.52237737e-01
8.73690665e-01 1.57427698e-01 -4.30619389e-01 -3.98371547e-01
-5.21973491e-01 4.57277566e-01 8.96572411e-01 5.53921402e-01
-3.81959319e-01 9.64809895e-01 6.97424221e+00 4.20673370e-01
-1.82283151e+00 2.46840324e-02 -1.32506907e-01 -3.24178308e-01
-9.04330432e-01 1.69252694e-01 -9.64135945e-01 7.16334939e-01
8.78029943e-01 -4.54225652e-02 3.35018754e-01 1.38740733e-01
1.41913399e-01 -5.35880029e-01 -2.39218578e-01 8.62247288e-01
-2.96353489e-01 -2.06713152e+00 -6.76689371e-02 5.28781474e-01
1.08641040e+00 6.60617352e-01 1.97521165e-01 -4.29417551e-01
-4.27647352e-01 -8.78522813e-01 2.45326504e-01 6.93397820e-01
1.19240355e+00 -3.05576742e-01 3.85221422e-01 3.39784235e-01
-5.84613025e-01 2.79682070e-01 -6.78495109e-01 -3.54508132e-01
5.60346246e-01 1.16321445e+00 -8.23457360e-01 2.60978397e-02
7.15432227e-01 2.59247005e-01 -1.36795670e-01 5.93609095e-01
2.38688156e-01 6.48818016e-01 -6.92517102e-01 -5.39843559e-01
1.67606860e-01 -8.65814030e-01 1.02485633e+00 5.76821804e-01
3.71808559e-01 4.09833521e-01 -7.83390403e-01 9.75385606e-01
-1.05075603e-02 -3.66915077e-01 -6.78531170e-01 -5.66626310e-01
8.23421255e-02 1.35718417e+00 -1.21199882e+00 -1.55734569e-01
2.36807793e-01 6.00018978e-01 -1.06954902e-01 1.05897024e-01
-1.64441034e-01 -5.42565584e-01 4.61688817e-01 1.09962642e+00
8.03358674e-01 -6.17294908e-01 -1.88887175e-02 -1.34997749e+00
4.03197736e-01 -9.21143666e-02 -5.92249632e-01 -4.38766271e-01
-7.91780353e-01 4.10841852e-01 -1.33022666e-01 -4.65106368e-01
-1.99220255e-01 -7.27263272e-01 -8.48087847e-01 4.77686763e-01
-1.09772491e+00 -6.24581099e-01 2.14955404e-01 3.02677721e-01
-7.30566323e-01 1.01674855e-01 7.63575137e-01 -6.33575916e-02
5.13094291e-03 -3.53543460e-01 1.11325276e+00 -6.08740032e-01
2.42728174e-01 -1.01631486e+00 4.88487303e-01 1.54903039e-01
-1.43082887e-01 8.22564900e-01 1.02490592e+00 -8.46178353e-01
-2.15620732e+00 -5.96270621e-01 7.88426280e-01 -8.20832312e-01
4.41519916e-01 -7.94975281e-01 -5.24834692e-01 1.53248519e-01
1.68762635e-02 1.84481710e-01 5.17449498e-01 -1.44459650e-01
4.53698188e-01 2.10379615e-01 -1.15507245e+00 2.65609324e-01
1.10539365e+00 -6.00794077e-01 -3.67710292e-01 5.97658873e-01
1.06255203e-01 -7.46600851e-02 -3.02109957e-01 6.55207157e-01
6.00323856e-01 -1.68933392e+00 3.96875709e-01 -5.91351569e-01
1.78851724e-01 -1.75255388e-01 -1.55507892e-01 -1.00355756e+00
1.50971130e-01 -9.32749629e-01 4.13975924e-01 7.12664664e-01
1.60268232e-01 -8.57679427e-01 1.03758144e+00 2.83469886e-01
-1.06193155e-01 -6.80249274e-01 -1.37901914e+00 -7.46775806e-01
-1.23196433e-03 1.46424145e-01 2.67918020e-01 7.42902517e-01
4.18728381e-01 3.65653098e-01 -2.48301089e-01 -5.05413339e-02
6.61342382e-01 6.38748229e-01 2.83555239e-01 -1.16708910e+00
-1.50845021e-01 1.13339871e-01 -5.63384175e-01 -7.27118552e-01
6.06621057e-02 -9.52044010e-01 -1.02763489e-01 -1.40529394e+00
9.98561531e-02 -7.62525976e-01 -2.79672481e-02 -3.73498499e-01
9.44773138e-01 5.07422745e-01 -2.83332825e-01 5.07871628e-01
-1.27598500e+00 7.68757999e-01 1.18558419e+00 4.90519613e-01
-4.49106634e-01 2.21049096e-02 -2.34412804e-01 6.62973523e-01
6.78682208e-01 -5.66819549e-01 1.47216529e-01 -7.94102773e-02
9.01120663e-01 1.19964391e-01 3.73131275e-01 -1.36760914e+00
2.70745724e-01 -6.32953644e-02 -9.00703073e-02 -5.48305690e-01
4.35719043e-01 -5.88716902e-02 2.96099812e-01 2.92406052e-01
3.90886784e-01 -4.83048737e-01 -3.40583444e-01 2.64754355e-01
-1.86576888e-01 -2.14942724e-01 1.12989092e+00 -4.09289211e-01
2.10851561e-02 3.86233814e-02 -7.88526237e-01 9.45573524e-02
9.95597184e-01 -2.16730282e-01 -6.96860969e-01 -1.48242950e-01
-3.85087430e-01 -2.90333509e-01 1.18230617e+00 -1.04216352e-01
5.18529475e-01 -1.11446536e+00 8.19117725e-02 4.49313939e-01
-2.64047801e-01 -2.86000401e-01 3.79850328e-01 1.00498974e+00
-5.51094413e-01 1.01746953e+00 -3.40719819e-01 -8.22357178e-01
-1.14760339e+00 9.69648585e-02 6.38403296e-01 -1.54113039e-01
-6.95477009e-01 5.59238970e-01 1.87344500e-03 -4.70549703e-01
-8.96461308e-01 -4.31315377e-02 -9.36442334e-03 -4.57316071e-01
4.47849929e-01 3.28070968e-01 4.33029026e-01 -8.35110486e-01
-5.51712573e-01 1.14158118e+00 -1.03266062e-02 -2.26766497e-01
1.73415005e+00 -9.21332985e-02 -8.65478814e-01 4.75859880e-01
1.02942228e+00 4.58639592e-01 -7.06850350e-01 2.38224834e-01
-1.06587093e-02 -4.00783718e-01 -1.92741215e-01 -2.96094686e-01
-4.65473831e-01 8.73814821e-01 1.03536367e-01 4.76625204e-01
4.38463837e-01 8.06082785e-01 7.97438204e-01 7.89017856e-01
7.92903781e-01 -1.19418621e+00 -8.65216404e-02 5.66327453e-01
5.39270759e-01 -5.85744739e-01 -2.65070975e-01 -1.50717139e-01
-3.16382736e-01 9.80372608e-01 1.74976736e-01 -2.02836469e-01
7.44116843e-01 1.79644704e-01 -4.24752802e-01 -9.24831688e-01
-7.55740345e-01 1.84450567e-01 -1.96087763e-01 5.36416531e-01
2.90812850e-01 3.15232635e-01 -4.16892499e-01 3.60902846e-01
-3.48405331e-01 -6.41526431e-02 8.88859332e-01 1.06237924e+00
-9.31501448e-01 -1.34255838e+00 -3.79297435e-01 6.36764705e-01
-5.07500172e-01 1.50296256e-01 -2.77230710e-01 3.95564646e-01
-3.54484618e-02 5.31783164e-01 3.11809987e-01 -3.13818634e-01
3.56289297e-01 -2.40706187e-02 8.15946877e-01 -5.86906672e-01
2.53855586e-01 -5.06477654e-01 -1.92054175e-02 -3.99678707e-01
-5.78475237e-01 -4.76234943e-01 -1.49360716e+00 -3.18107903e-01
-4.82362598e-01 8.64156425e-01 1.16464365e+00 1.14600980e+00
6.71442330e-01 -3.44691396e-01 6.41749084e-01 -1.01308608e+00
-3.78346562e-01 -5.70412695e-01 -1.19707632e+00 2.66176224e-01
3.82582009e-01 -7.04637706e-01 -5.51984012e-01 -8.35003197e-01]
|
[5.498984336853027, 4.828834056854248]
|
a51eb797-d75b-4244-b3c3-560a39817b0b
|
biomedical-named-entity-recognition-at-scale
|
2011.06315
| null |
https://arxiv.org/abs/2011.06315v1
|
https://arxiv.org/pdf/2011.06315v1.pdf
|
Biomedical Named Entity Recognition at Scale
|
Named entity recognition (NER) is a widely applicable natural language processing task and building block of question answering, topic modeling, information retrieval, etc. In the medical domain, NER plays a crucial role by extracting meaningful chunks from clinical notes and reports, which are then fed to downstream tasks like assertion status detection, entity resolution, relation extraction, and de-identification. Reimplementing a Bi-LSTM-CNN-Char deep learning architecture on top of Apache Spark, we present a single trainable NER model that obtains new state-of-the-art results on seven public biomedical benchmarks without using heavy contextual embeddings like BERT. This includes improving BC4CHEMD to 93.72% (4.1% gain), Species800 to 80.91% (4.6% gain), and JNLPBA to 81.29% (5.2% gain). In addition, this model is freely available within a production-grade code base as part of the open-source Spark NLP library; can scale up for training and inference in any Spark cluster; has GPU support and libraries for popular programming languages such as Python, R, Scala and Java; and can be extended to support other human languages with no code changes.
|
['David Talby', 'Veysel Kocaman']
|
2020-11-12
| null | null | null | null |
['medical-named-entity-recognition']
|
['natural-language-processing']
|
[-2.36253858e-01 1.93998173e-01 -1.44228652e-01 -3.97222102e-01
-9.93907630e-01 -2.66144007e-01 1.39504477e-01 9.15702105e-01
-9.21429694e-01 8.97270441e-01 3.75101537e-01 -5.41230321e-01
3.96674536e-02 -9.40751791e-01 -5.10759234e-01 -6.05615795e-01
-2.29411557e-01 5.17791092e-01 2.24154428e-01 9.11395811e-03
-1.85084373e-01 5.12650490e-01 -1.01473486e+00 6.21252894e-01
5.14253139e-01 9.91729915e-01 -1.34881839e-01 7.78879344e-01
-2.85045534e-01 1.10709858e+00 -5.63054800e-01 -4.28915650e-01
-1.84535801e-01 1.31932631e-01 -9.79692996e-01 -8.61901164e-01
-2.74074376e-01 -1.37731463e-01 -4.06023175e-01 7.82029748e-01
1.01015139e+00 -1.64871246e-01 1.22891389e-01 -8.23967397e-01
-5.99660635e-01 5.82965851e-01 -2.36194953e-01 4.04976040e-01
1.95981741e-01 -7.93069378e-02 7.59705186e-01 -8.95181298e-01
7.88701355e-01 6.75740480e-01 1.05077696e+00 5.48872948e-01
-8.76717269e-01 -7.35917211e-01 -5.65692246e-01 2.26167858e-01
-1.58268583e+00 -3.73348087e-01 1.29653856e-01 -3.03606927e-01
1.46408689e+00 3.88132006e-01 2.30512917e-01 9.19743001e-01
4.83330071e-01 6.79326177e-01 6.50627732e-01 -9.93029773e-02
2.70002812e-01 4.24182303e-02 4.44007635e-01 6.49460912e-01
1.12195283e-01 -2.03279242e-01 -3.65523994e-01 -6.75849736e-01
2.65397042e-01 2.17428923e-01 -1.72863767e-01 4.45966750e-01
-1.51807702e+00 7.71166265e-01 6.14166975e-01 4.21257615e-01
-7.86853313e-01 -5.33133484e-02 9.20558631e-01 6.28731176e-02
5.87081671e-01 5.47110915e-01 -8.58182609e-01 -1.87163055e-01
-8.71554911e-01 7.13803545e-02 1.14737582e+00 7.31400311e-01
4.82463449e-01 -3.05918992e-01 -4.46977258e-01 8.46226871e-01
-1.72916383e-01 5.71839176e-02 7.54770339e-01 -4.35022652e-01
6.38594851e-02 7.84838676e-01 -1.52800053e-01 -7.53802955e-01
-9.65904236e-01 -3.86989594e-01 -1.11975074e+00 -4.18535501e-01
7.11300895e-02 -4.18591708e-01 -9.48980451e-01 1.48609293e+00
6.34481192e-01 3.25593621e-01 2.56562322e-01 5.79760551e-01
1.51994276e+00 6.72330499e-01 5.41426241e-01 -6.14229916e-03
2.01348567e+00 -8.10674906e-01 -7.69106090e-01 4.89060991e-02
9.28535581e-01 -7.82052040e-01 3.60504389e-01 2.38452524e-01
-7.96956956e-01 -1.74352765e-01 -7.02129185e-01 -6.82073116e-01
-9.00572717e-01 1.89178020e-01 7.29090989e-01 3.59279811e-01
-1.12273371e+00 3.73759806e-01 -1.16198123e+00 -4.02072370e-01
6.09799027e-01 5.08140564e-01 -5.53549647e-01 -1.97210506e-01
-1.40955245e+00 9.52003300e-01 7.86933720e-01 1.30386233e-01
-6.58704937e-01 -1.22718990e+00 -8.10190737e-01 2.28753254e-01
1.29121557e-01 -9.49245512e-01 9.78858650e-01 -3.31809744e-02
-1.11087573e+00 1.05515015e+00 -1.50369346e-01 -6.88111305e-01
1.61530077e-01 -1.75550848e-01 -6.24387205e-01 -3.03614140e-03
2.06641853e-01 5.80896258e-01 -2.21328646e-01 -2.30280578e-01
-5.23899496e-01 -5.21074831e-01 -2.72496372e-01 -6.90756664e-02
-2.47840971e-01 4.92628664e-01 -4.01364893e-01 -5.02714336e-01
-3.52042109e-01 -7.49225080e-01 -4.99872893e-01 -4.22822051e-02
-6.84025407e-01 -6.94843113e-01 4.07876581e-01 -1.20373976e+00
1.08192575e+00 -2.07902217e+00 -3.21311653e-01 -1.05671652e-01
6.05271399e-01 5.32064974e-01 1.46157846e-01 5.35363436e-01
-3.22210342e-01 8.83606970e-02 -2.28221282e-01 -5.29671088e-02
-2.73334861e-01 -1.37956440e-01 8.61191470e-03 3.25856060e-01
3.46174330e-01 1.08142555e+00 -8.04404914e-01 -4.84562486e-01
-7.99317434e-02 9.05925095e-01 -3.25093776e-01 2.74959564e-01
1.32538930e-01 4.85224836e-02 -5.42107344e-01 5.28933287e-01
6.11892998e-01 -5.76632082e-01 6.54146001e-02 -1.27536699e-01
-1.88448653e-01 6.44197583e-01 -9.76020694e-01 1.75327647e+00
-4.76182729e-01 4.23631042e-01 9.97323021e-02 -1.05451965e+00
1.03455949e+00 6.53516531e-01 6.15520477e-01 -2.36513227e-01
4.03594971e-02 2.30073392e-01 -1.49822474e-01 -6.44966006e-01
3.38328958e-01 9.28868167e-03 -5.54411188e-02 2.84859240e-01
1.53512254e-01 4.40337479e-01 6.95407689e-02 2.72376120e-01
1.58646381e+00 -3.99365395e-01 7.46323228e-01 -3.75472337e-01
4.23293471e-01 1.92523465e-01 8.28887343e-01 4.54107344e-01
-7.66903087e-02 2.45918721e-01 5.22697926e-01 -7.59299517e-01
-9.21694696e-01 -8.78644764e-01 -5.57779908e-01 1.24686170e+00
-5.10118425e-01 -4.36875701e-01 -4.71640885e-01 -5.47054231e-01
-1.92367882e-02 6.27758443e-01 -6.53195322e-01 -1.17987588e-01
-5.33957243e-01 -1.28713357e+00 1.03851998e+00 6.36800051e-01
5.22466600e-01 -1.37928581e+00 -5.49029052e-01 5.44574380e-01
-1.42917112e-01 -1.13232887e+00 -3.15828621e-01 4.85842407e-01
-6.29730761e-01 -1.13602340e+00 -7.86744833e-01 -1.03078425e+00
3.45378280e-01 -3.01776916e-01 1.17113793e+00 -2.75738966e-02
-8.45765769e-01 -1.17138319e-01 -1.53397724e-01 -6.71999156e-01
-2.33288229e-01 3.80242914e-01 -2.08115876e-01 -3.96300703e-01
8.63354981e-01 -2.36024484e-01 -6.94190502e-01 -5.17206229e-02
-9.25582230e-01 -3.32684442e-02 5.10293245e-01 1.06068516e+00
8.34200561e-01 -2.36745805e-01 7.60276556e-01 -1.01439154e+00
5.17308891e-01 -8.46510231e-01 -2.26508185e-01 2.76825994e-01
-5.43850660e-01 4.50266711e-03 6.46218657e-01 -2.53525469e-02
-9.36022341e-01 1.42718121e-01 -7.45719373e-01 6.32762164e-02
-3.69101226e-01 7.58999050e-01 -4.97983061e-02 5.34476519e-01
7.98051596e-01 3.09977770e-01 -1.26808211e-01 -5.95924437e-01
1.53678328e-01 1.08159626e+00 5.87988377e-01 -2.10526474e-02
-2.84385569e-02 3.42885822e-01 -3.00477087e-01 -8.58007133e-01
-1.00987458e+00 -8.07146728e-01 -7.40706548e-02 6.43582225e-01
1.15817606e+00 -1.18266129e+00 -1.00535965e+00 2.47924969e-01
-1.24025249e+00 -9.54211503e-02 -8.69664922e-02 5.10862648e-01
-1.01649769e-01 5.08654602e-02 -1.21179914e+00 -2.37914473e-01
-1.29412806e+00 -9.64591742e-01 9.95957792e-01 3.40990752e-01
-2.02568725e-01 -7.71734476e-01 2.78386384e-01 2.35867336e-01
5.77770352e-01 3.54780346e-01 8.44394386e-01 -1.37323868e+00
8.49409867e-03 -4.26721036e-01 -4.59591031e-01 1.93617672e-01
-1.14556208e-01 -2.00388089e-01 -9.61459339e-01 -7.46715888e-02
-1.59984007e-01 -2.42666975e-01 1.03139174e+00 4.07344311e-01
1.37429881e+00 -3.37196350e-01 -8.07137311e-01 7.45900810e-01
1.15314007e+00 1.46423861e-01 6.34573579e-01 3.36729020e-01
4.75613177e-01 4.67784494e-01 1.22964531e-01 3.91732603e-01
6.01681828e-01 4.16867435e-01 1.21917203e-01 -2.97230035e-01
-3.04688904e-02 8.48790407e-02 -7.68756270e-02 8.51197362e-01
1.44813910e-01 7.80278295e-02 -1.39086056e+00 7.20249474e-01
-1.74849308e+00 -7.93563128e-01 -3.60979170e-01 1.85318935e+00
1.28119111e+00 -2.24211425e-01 -6.45843968e-02 -3.26525241e-01
7.99095511e-01 -2.53620505e-01 -6.26036465e-01 -4.58218306e-01
2.79046055e-02 6.22163832e-01 4.07383442e-01 -1.80758592e-02
-1.29009330e+00 8.08233619e-01 5.34446907e+00 7.43865609e-01
-1.12454081e+00 4.66931492e-01 9.79473829e-01 9.73286927e-02
2.20342264e-01 -4.94567513e-01 -9.21092570e-01 3.52728724e-01
1.43400145e+00 -4.36541975e-01 -3.73105868e-03 1.04121673e+00
-3.25026596e-03 1.98086515e-01 -1.01500428e+00 9.76645172e-01
-3.96331608e-01 -1.90799558e+00 -5.37095606e-01 1.05729796e-01
3.66149604e-01 7.60452867e-01 -3.92041683e-01 5.47700524e-01
3.97019416e-01 -1.21094465e+00 -1.38041332e-01 3.89148355e-01
1.01017725e+00 -8.58439863e-01 1.47084188e+00 1.94293991e-01
-9.42726731e-01 1.80144727e-01 -6.17240906e-01 3.34262669e-01
8.91586691e-02 1.25664604e+00 -1.19537222e+00 6.45512760e-01
1.09472430e+00 4.90190774e-01 -2.39885017e-01 1.12173486e+00
-1.02382369e-01 7.80819237e-01 -5.56321859e-01 -2.15688974e-01
-8.62450302e-02 5.04783094e-01 1.20797135e-01 1.70753956e+00
1.27437353e-01 2.18271956e-01 1.71442583e-01 6.41315639e-01
-5.34504354e-01 5.35475731e-01 -2.23742634e-01 -2.84466222e-02
4.34081405e-01 1.51560056e+00 -7.06372321e-01 -5.43019831e-01
-1.15703031e-01 7.06563294e-01 4.45372194e-01 5.22513166e-02
-9.39964056e-01 -1.03666818e+00 7.23951042e-01 -3.40049028e-01
3.93390715e-01 2.04697385e-01 -1.52080804e-01 -1.07367337e+00
-2.06266314e-01 -7.38655269e-01 7.79307067e-01 -4.36832130e-01
-1.50097132e+00 1.07448530e+00 -5.20299733e-01 -5.99415004e-01
-1.72130823e-01 -6.92936122e-01 -5.15420437e-01 8.85980785e-01
-1.40539765e+00 -9.44117844e-01 -2.78363168e-01 5.39141834e-01
1.80171598e-02 -6.03340417e-02 1.47102678e+00 6.51162624e-01
-7.87855685e-01 6.26308143e-01 1.65722340e-01 5.78698277e-01
8.12535107e-01 -1.15327644e+00 4.53448504e-01 3.06815952e-01
-2.40802877e-02 8.13889027e-01 2.63507783e-01 -4.25793231e-01
-1.17253745e+00 -1.40225911e+00 1.51172924e+00 -2.83345580e-01
7.56797433e-01 -4.08463299e-01 -1.05395269e+00 5.65867901e-01
1.61711261e-01 4.66582984e-01 1.29810560e+00 3.88579667e-01
-4.74027365e-01 -1.90002307e-01 -1.31496167e+00 1.54650286e-01
5.63542724e-01 -4.58505362e-01 -5.39850891e-01 8.46370280e-01
9.67547834e-01 -6.23569548e-01 -1.54326129e+00 2.41976053e-01
3.19560945e-01 -5.61909139e-01 9.25955474e-01 -9.61338580e-01
4.49498683e-01 -8.93685967e-02 1.31005332e-01 -8.22837353e-01
-3.20865691e-01 -4.86702055e-01 -8.94230455e-02 1.02956235e+00
6.20704055e-01 -8.50990772e-01 6.86304629e-01 5.88261306e-01
-3.07382196e-01 -1.07593513e+00 -1.15447736e+00 -3.12718421e-01
1.41328707e-01 -3.73103201e-01 6.70219064e-01 1.15689874e+00
2.72756934e-01 4.34199929e-01 4.06513363e-02 1.96976796e-01
1.89980552e-01 4.14071605e-02 3.46679121e-01 -1.13963783e+00
-2.05545455e-01 -1.60277009e-01 -4.78278458e-01 -5.35853684e-01
1.39151560e-02 -1.17603219e+00 5.78673817e-02 -1.79101598e+00
3.79970610e-01 -4.63035226e-01 -7.00693011e-01 1.14026380e+00
-3.07092428e-01 2.67317027e-01 -1.27945766e-01 4.78740446e-02
-4.79483068e-01 1.43183723e-01 4.34845030e-01 -1.56807274e-01
-1.95407376e-01 -5.06176949e-02 -8.00496697e-01 4.90374684e-01
9.21167135e-01 -9.95439053e-01 3.76070410e-01 -2.83778608e-01
2.55499452e-01 8.97431523e-02 2.54216254e-01 -8.18187416e-01
4.45548028e-01 3.61656368e-01 3.54841501e-01 -5.01060843e-01
-9.32132304e-02 -3.11764300e-01 6.14933185e-02 7.04596043e-01
-4.16150659e-01 8.97676051e-02 5.23283124e-01 4.40419585e-01
-1.61183968e-01 -7.39225671e-02 4.64100629e-01 -1.84514485e-02
-5.06229818e-01 4.27427083e-01 -1.61490738e-01 2.12224573e-01
1.08055997e+00 3.15680563e-01 -5.74663341e-01 9.57031846e-02
-8.70814979e-01 3.54904324e-01 -1.98172614e-01 2.95346856e-01
3.50634456e-01 -9.61763561e-01 -1.18759573e+00 -1.59022108e-01
2.66064107e-01 2.20625490e-01 4.92789388e-01 1.06534803e+00
-7.33124018e-01 6.99600339e-01 1.56690210e-01 -4.29699868e-01
-1.10422850e+00 7.01221228e-01 1.74046084e-01 -7.74103820e-01
-8.11963141e-01 1.02691793e+00 8.98746923e-02 -7.80209005e-01
7.20364526e-02 -2.72551864e-01 -4.45969664e-02 -8.66689831e-02
1.01926267e+00 3.94374490e-01 6.54987395e-01 -1.13993824e-01
-9.50223863e-01 -2.14826226e-01 -3.34882021e-01 4.79436189e-01
1.75578403e+00 6.59226060e-01 -5.91516733e-01 1.53762639e-01
1.40138173e+00 -8.85755122e-02 -4.19172019e-01 2.66590621e-02
1.65882230e-01 5.29096127e-01 2.45760053e-01 -1.13457215e+00
-9.97259915e-01 7.95828342e-01 5.05662858e-01 7.98015222e-02
1.02373326e+00 2.36001417e-01 9.71826196e-01 6.33000195e-01
1.07771069e-01 -6.48660779e-01 -5.98509431e-01 5.05584419e-01
6.11182928e-01 -1.02623737e+00 5.37265688e-02 -1.33751646e-01
-4.75299001e-01 1.14207578e+00 1.26067355e-01 -4.42023302e-04
9.43982244e-01 6.27833128e-01 -6.19754568e-02 -2.63291866e-01
-9.34972882e-01 5.69569506e-02 2.17206255e-01 3.42657149e-01
9.25801992e-01 1.84944510e-01 -4.87389535e-01 9.46668983e-01
-2.14939073e-01 4.60157901e-01 3.46749038e-01 7.25224555e-01
-8.13333467e-02 -7.71378040e-01 -5.04960865e-02 7.68715858e-01
-1.19601560e+00 -6.62913501e-01 1.39292628e-01 5.28792858e-01
1.57049492e-01 7.87671804e-01 1.10187918e-01 -1.07983954e-01
2.50797778e-01 2.88896531e-01 -2.94920236e-01 -7.17949152e-01
-1.03185594e+00 -1.74298882e-01 3.42874706e-01 -5.66246390e-01
-1.14289612e-01 -2.82916367e-01 -1.67219174e+00 -2.80266255e-01
-2.55641699e-01 4.62779105e-01 8.23610485e-01 6.65024281e-01
1.17794156e+00 7.02958107e-01 -4.40852493e-02 -8.37264955e-02
-3.28162551e-01 -1.04473531e+00 -2.53699422e-01 3.42843272e-02
2.09630448e-02 -1.23813182e-01 -8.57682303e-02 -3.21598649e-02]
|
[8.505779266357422, 8.826972007751465]
|
c76839f5-f3f4-454e-9699-3d51280483ca
|
domain-independent-abstract-generation-for
| null | null |
https://aclanthology.org/P13-1137
|
https://aclanthology.org/P13-1137.pdf
|
Domain-Independent Abstract Generation for Focused Meeting Summarization
| null |
['Lu Wang', 'Claire Cardie']
|
2013-08-01
| null | null | null |
acl-2013-8
|
['meeting-summarization']
|
['natural-language-processing']
|
[-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]
|
[-7.2955522537231445, 3.680513381958008]
|
0db46695-abd6-4924-83b6-9b182babef55
|
scalable-bilevel-optimization-for-generating
|
2304.10912
| null |
https://arxiv.org/abs/2304.10912v1
|
https://arxiv.org/pdf/2304.10912v1.pdf
|
Scalable Bilevel Optimization for Generating Maximally Representative OPF Datasets
|
New generations of power systems, containing high shares of renewable energy resources, require improved data-driven tools which can swiftly adapt to changes in system operation. Many of these tools, such as ones using machine learning, rely on high-quality training datasets to construct probabilistic models. Such models should be able to accurately represent the system when operating at its limits (i.e., operating with a high degree of ``active constraints"). However, generating training datasets that accurately represent the many possible combinations of these active constraints is a particularly challenging task, especially within the realm of nonlinear AC Optimal Power Flow (OPF), since most active constraints cannot be enforced explicitly. Using bilevel optimization, this paper introduces a data collection routine that sequentially solves for OPF solutions which are ``optimally far" from previously acquired voltage, power, and load profile data points. The routine, termed RAMBO, samples critical data close to a system's boundaries much more effectively than a random sampling benchmark. Simulated test results are collected on the 30-, 57-, and 118-bus PGLib test cases.
|
['Samuel Chevalier', 'Ignasi Ventura Nadal']
|
2023-04-21
| null | null | null | null |
['bilevel-optimization']
|
['methodology']
|
[-2.61450291e-01 -3.49607795e-01 -5.74025214e-01 -2.89431930e-01
-6.82675064e-01 -7.86950588e-01 2.07457051e-01 2.72367865e-01
2.80246586e-01 1.34444416e+00 -3.05323780e-01 -4.52552021e-01
-6.68890238e-01 -8.38255167e-01 -1.92694709e-01 -9.74610209e-01
-3.64092231e-01 8.39208722e-01 -2.12254092e-01 -2.57915556e-02
2.71477044e-01 5.94474494e-01 -1.24787354e+00 -4.01700377e-01
1.37789142e+00 8.31680834e-01 1.60568953e-01 3.34578544e-01
1.26164302e-01 1.80363283e-01 -1.13187909e+00 3.86595726e-01
-4.22353037e-02 -1.78079396e-01 -4.36895549e-01 -2.33269081e-01
-5.09061158e-01 -1.57514215e-01 -1.22239366e-02 1.24150491e+00
3.47303361e-01 2.25599200e-01 6.42927110e-01 -1.60488486e+00
-1.25424951e-01 6.06956244e-01 -5.03513277e-01 3.80860358e-01
3.43806356e-01 2.10490882e-01 9.45302188e-01 -4.36628819e-01
2.81789359e-02 6.34743333e-01 2.76090115e-01 1.84116036e-01
-1.76215899e+00 -5.51370680e-01 9.76925716e-02 5.11510134e-01
-1.49106419e+00 -5.04840761e-02 1.08445168e+00 -3.92864555e-01
1.03682804e+00 6.61849558e-01 7.77962446e-01 6.07374549e-01
3.04500312e-01 5.28764248e-01 9.89665270e-01 -3.01074207e-01
6.38346672e-01 1.29782498e-01 6.04413673e-02 -1.30744666e-01
3.21123332e-01 -6.57441374e-03 -2.50916809e-01 -3.55272830e-01
-1.80504168e-03 -3.16358000e-01 -7.03684747e-01 -5.84105074e-01
-6.72919810e-01 7.02766299e-01 2.72185862e-01 4.40388113e-01
-2.21879303e-01 -1.77831888e-01 1.36069164e-01 -1.09348908e-01
1.88520372e-01 7.90210485e-01 -7.98610151e-01 -3.03785533e-01
-1.27572513e+00 1.50502026e-01 8.52511883e-01 9.71672595e-01
5.50322533e-01 6.69100046e-01 -3.26858163e-02 7.28197038e-01
2.49620080e-01 5.47106385e-01 3.47682893e-01 -4.31981474e-01
4.29196239e-01 6.66939676e-01 2.92044759e-01 -5.52294374e-01
-5.14481068e-01 -7.42885113e-01 -9.07258391e-01 3.26868534e-01
3.96589071e-01 -4.30648476e-01 -5.61547101e-01 1.34249055e+00
1.54274851e-01 -1.15898266e-01 -6.22766875e-02 5.87323427e-01
-6.38626292e-02 1.28998816e+00 -2.19222143e-01 -8.29761446e-01
9.53317463e-01 -1.50174320e-01 -8.85382235e-01 -1.93633318e-01
3.70352864e-01 -7.57170022e-01 7.39581168e-01 5.23082137e-01
-9.43310440e-01 -1.88824087e-01 -1.49266362e+00 6.61135733e-01
-3.66545707e-01 2.08014593e-01 3.56975913e-01 8.20919931e-01
-5.33890367e-01 6.20001197e-01 -6.68809533e-01 3.02208602e-01
4.69921172e-01 1.72876254e-01 1.99525446e-01 2.83010956e-02
-1.14303744e+00 1.39682591e+00 6.61934674e-01 5.18940985e-01
-8.30731869e-01 -1.02836359e+00 -6.74484313e-01 4.52472746e-01
7.11582184e-01 -1.59179106e-01 8.87650728e-01 -4.88815129e-01
-1.52950895e+00 -2.58915097e-01 -1.89085901e-02 -2.19372690e-01
7.69222528e-02 1.41066313e-01 -5.85674345e-01 -1.96012259e-01
-2.45026112e-01 -3.11595768e-01 7.27688372e-01 -1.31668246e+00
-4.87484902e-01 -1.95272103e-01 -2.40913495e-01 -1.63315851e-02
-2.20505819e-01 -2.33989209e-01 2.28139848e-01 -4.91929799e-01
-1.72045320e-01 -6.26112878e-01 -3.66769552e-01 -4.25544590e-01
-7.83065379e-01 -4.29977745e-01 1.10266268e+00 -7.98870981e-01
1.42112279e+00 -1.69860709e+00 3.41315508e-01 8.75597179e-01
-4.05505329e-01 4.02936459e-01 5.73201954e-01 6.47119880e-01
-3.17326397e-01 2.27553204e-01 -3.99521619e-01 1.26071692e-01
5.01634359e-01 5.29497027e-01 -1.80412427e-01 4.95712340e-01
2.59037107e-01 3.80543768e-01 -7.23463535e-01 -7.25336969e-02
6.62036180e-01 2.86466219e-02 -3.19995791e-01 1.03699289e-01
-3.85310769e-01 1.90496728e-01 -3.98353577e-01 4.15505230e-01
6.13897860e-01 -1.47035331e-01 5.46174526e-01 -3.38498086e-01
-1.77096665e-01 -1.92955241e-01 -1.42943633e+00 1.21251523e+00
-7.94989288e-01 6.49126172e-01 1.07435696e-01 -1.36903822e+00
8.17528188e-01 2.75274038e-01 7.68226564e-01 -5.34124076e-01
4.84047532e-02 5.52128702e-02 1.28542140e-01 -1.12274781e-01
8.59272107e-02 1.49021804e-01 -1.10147521e-01 3.53063226e-01
1.21203326e-01 -6.21415973e-01 6.73844039e-01 -4.94491421e-02
6.95425987e-01 -2.22989008e-01 4.87847418e-01 -8.16306412e-01
7.80058444e-01 -1.16056884e-02 1.22343397e+00 9.21976492e-02
1.38987958e-01 1.47340134e-01 9.46261823e-01 -1.23021103e-01
-9.19450462e-01 -1.09817970e+00 -7.89980590e-01 1.20664284e-01
-2.85013970e-02 -4.08236295e-01 -4.37625945e-01 -5.10382831e-01
8.59678537e-02 1.47115314e+00 -1.38732508e-01 -2.22500086e-01
-3.79198074e-01 -1.33205843e+00 -3.42153102e-01 3.18419367e-01
7.90699124e-02 -4.74865168e-01 -5.25461435e-01 4.69748557e-01
1.29828200e-01 -8.89819682e-01 -1.23119041e-01 5.78751981e-01
-6.01426005e-01 -1.29510558e+00 -4.26554382e-01 -3.04776877e-01
9.02149916e-01 -6.24557972e-01 1.20130992e+00 1.53099880e-01
-5.57948172e-01 -1.84654102e-01 -1.22357853e-01 -2.67464012e-01
-3.22036803e-01 3.62316035e-02 4.89386506e-02 -3.24832708e-01
-1.52402207e-01 -7.18318164e-01 -1.26429334e-01 5.54767191e-01
-6.15206599e-01 -1.33344844e-01 3.10151517e-01 9.86665189e-01
6.10396385e-01 1.13328946e+00 9.95136201e-01 -4.10897285e-01
5.69401026e-01 -7.06644416e-01 -1.48062289e+00 5.73939145e-01
-8.85151327e-01 1.68854237e-01 1.40992188e+00 -2.56192088e-01
-8.89371455e-01 -1.72390804e-01 2.37863790e-02 -3.31483096e-01
8.30649806e-04 5.15229583e-01 -7.05695391e-01 8.89073536e-02
1.54039651e-01 8.98394063e-02 -4.03991461e-01 -4.13313925e-01
4.54811525e-04 5.74568272e-01 3.28083992e-01 -1.00200236e+00
1.35087562e+00 1.08031603e-02 2.81733334e-01 -6.90085828e-01
-4.27082241e-01 -3.94869559e-02 -4.56887335e-01 -3.59746635e-01
2.37248257e-01 -2.94117868e-01 -8.91533077e-01 3.76040936e-01
-6.31160498e-01 -2.28013754e-01 -4.27764565e-01 4.04078394e-01
-2.74640173e-01 9.37272385e-02 1.06978185e-01 -9.63537216e-01
-2.65400350e-01 -1.28446972e+00 3.48226815e-01 7.51621902e-01
-2.02073172e-01 -1.18227351e+00 -3.70639190e-02 -6.68750890e-03
5.21415532e-01 5.11860669e-01 1.36694443e+00 -3.46018672e-01
-4.93437171e-01 -2.94450402e-01 2.73931623e-01 6.00041509e-01
3.25546563e-01 5.63042939e-01 -4.96481657e-01 -7.58895695e-01
1.04117438e-01 1.96480826e-01 -3.79247546e-01 4.27022040e-01
1.29451025e+00 -4.37806875e-01 -4.89969343e-01 4.59182799e-01
1.70281768e+00 5.81926346e-01 3.57840657e-01 -1.35510564e-01
1.02885135e-01 2.01473251e-01 4.40263003e-01 7.52971649e-01
3.02563399e-01 8.21136951e-01 4.73715156e-01 1.98489308e-01
5.85634589e-01 1.20784387e-01 5.75293042e-02 6.82531357e-01
3.39062840e-01 -2.92142004e-01 -9.23308790e-01 4.83437181e-01
-1.44314313e+00 -7.50125349e-01 3.20639201e-02 2.14336777e+00
7.90133893e-01 5.39193273e-01 -1.58312574e-01 8.03510070e-01
6.63491130e-01 2.44244114e-02 -7.73072720e-01 -5.47126055e-01
5.24287531e-03 3.65825862e-01 3.92388165e-01 5.06884217e-01
-6.85589254e-01 -3.91991101e-02 5.65958166e+00 1.03066802e+00
-9.17879760e-01 -4.11651611e-01 7.08359957e-01 -2.06704065e-01
-3.21983486e-01 1.15465112e-01 -8.03988039e-01 1.14162660e+00
1.28305018e+00 -9.91912186e-01 8.69194210e-01 7.65819311e-01
5.57637691e-01 -7.29168713e-01 -1.13872671e+00 7.38777459e-01
-1.02333851e-01 -1.35576189e+00 -5.39113224e-01 -2.00235397e-02
1.10120308e+00 -3.29600841e-01 -4.05499578e-01 2.30966255e-01
3.01477045e-01 -9.61311400e-01 6.23394072e-01 5.36846936e-01
3.40282023e-01 -1.13318110e+00 8.54109228e-01 5.60947120e-01
-9.69002604e-01 -3.84161443e-01 2.83462051e-02 2.88933188e-01
4.99804914e-01 1.28421032e+00 -6.22827709e-01 1.03785622e+00
8.11902761e-01 5.03650665e-01 -3.24520409e-01 1.31564415e+00
-4.41819042e-01 6.14654958e-01 -7.36328781e-01 -2.69365937e-01
-2.50277758e-01 -3.34501356e-01 5.15525401e-01 2.62112558e-01
3.60148132e-01 -1.26097035e-02 3.26744735e-01 1.01937795e+00
2.31674969e-01 -3.39135259e-01 -1.12178162e-01 2.36314148e-01
8.96093249e-01 1.40136814e+00 -4.51491058e-01 1.04112094e-02
-7.20134079e-02 -1.85065828e-02 -1.94681123e-01 5.23612380e-01
-9.68406558e-01 -4.44145679e-01 7.49736607e-01 -2.54322886e-01
-2.38038257e-01 -3.78700972e-01 -4.47876453e-01 -8.24758172e-01
2.36085713e-01 -8.23724985e-01 4.29407090e-01 -7.75787354e-01
-1.38989806e+00 2.41589233e-01 4.98304874e-01 -1.06167352e+00
-7.78449357e-01 -4.28406596e-01 -9.13486481e-01 1.26592886e+00
-1.45441496e+00 -3.20406079e-01 8.24761689e-02 3.82240385e-01
5.61681747e-01 3.49802710e-02 8.02052498e-01 4.28961247e-01
-1.17601454e+00 1.54832780e-01 6.36592388e-01 -1.27239496e-01
-2.59851217e-01 -1.44620025e+00 -3.50113928e-01 9.93114531e-01
-4.19976711e-02 6.21478520e-02 1.00028193e+00 -1.47980511e-01
-1.70409286e+00 -6.58820152e-01 2.79009879e-01 -6.80998638e-02
7.97624290e-01 -3.93921971e-01 -9.74704325e-01 3.96531522e-01
3.53087544e-01 2.04948902e-01 4.96631354e-01 7.79906362e-02
5.34286380e-01 -5.22201717e-01 -1.37339866e+00 3.66023004e-01
3.87114361e-02 -2.33641222e-01 -5.97278535e-01 4.85483140e-01
-3.28852087e-01 -4.32461888e-01 -1.35770798e+00 6.65144444e-01
-6.79935813e-02 -4.26797092e-01 9.81023788e-01 -3.42246175e-01
-3.36648762e-01 -7.24686444e-01 6.41433406e-04 -2.21070766e+00
1.06228836e-01 -9.52710390e-01 -5.44337094e-01 1.38867021e+00
4.13402498e-01 -9.07799184e-01 6.69383228e-01 5.96497178e-01
-1.42423481e-01 -9.43538427e-01 -1.59193587e+00 -9.23916578e-01
2.24098623e-01 -3.14310461e-01 1.20305943e+00 1.02760100e+00
4.11426067e-01 -4.35750932e-02 1.20937772e-01 7.18100071e-01
9.61010337e-01 2.55561084e-01 3.23195338e-01 -9.62354183e-01
-1.00554422e-01 -6.38859332e-01 -8.04065093e-02 -2.58698225e-01
2.48554140e-01 -5.47680914e-01 -1.33858025e-01 -1.54658318e+00
-3.11309606e-01 -7.76661992e-01 -9.64208469e-02 3.62822175e-01
-7.67417997e-02 -3.66209298e-01 1.15200147e-01 -1.19940765e-01
4.50436808e-02 8.86979282e-01 8.03654194e-01 -2.66258836e-01
8.91307071e-02 2.84410357e-01 -6.09908476e-02 5.53503573e-01
1.13071859e+00 -2.55226910e-01 -3.91986072e-01 -9.62965190e-02
1.64480224e-01 4.08652991e-01 -3.35759111e-02 -1.36699104e+00
2.28148192e-01 -5.14519215e-01 4.65960771e-01 -8.94016087e-01
1.83980033e-01 -1.30199885e+00 6.20402992e-01 4.38177764e-01
2.26875216e-01 2.81805038e-01 4.50343370e-01 1.82179734e-01
-1.89998910e-01 -4.62454289e-01 7.70598829e-01 3.34116995e-01
-5.87280452e-01 6.88775629e-03 -4.90184724e-01 1.66131467e-01
1.48718023e+00 4.47318591e-02 -5.25984168e-01 -1.86153501e-01
-8.36456299e-01 1.14088130e+00 1.46524936e-01 4.63293612e-01
1.00162491e-01 -1.17040229e+00 -5.85259080e-01 2.55911559e-01
-3.53758276e-01 1.54554546e-01 2.19656497e-01 3.77221942e-01
-2.35970601e-01 5.27921319e-01 1.21159159e-01 -7.82023787e-01
-6.40983224e-01 2.80886292e-01 6.72291696e-01 -4.78281796e-01
-3.58161688e-01 5.09594530e-02 -7.06417799e-01 -1.41447276e-01
-1.59742489e-01 -4.59844261e-01 -5.94611540e-02 2.86485583e-01
1.56307250e-01 5.94750881e-01 3.84617239e-01 -2.27972344e-01
-4.14041132e-01 3.39700729e-01 3.78882647e-01 4.34468001e-01
1.41759849e+00 1.12612404e-01 1.70125104e-02 3.31256658e-01
9.23808038e-01 -2.53499895e-01 -1.28423393e+00 1.90481082e-01
1.54182300e-01 -6.85392141e-01 3.82556558e-01 -1.15821695e+00
-1.55335748e+00 6.09835684e-01 3.78833473e-01 7.49185562e-01
1.19269013e+00 -5.07045090e-01 1.92641765e-01 3.10092866e-02
6.98059499e-01 -1.52822697e+00 -4.50757861e-01 1.66333824e-01
7.32503116e-01 -7.70159245e-01 3.68197918e-01 -6.52232096e-02
1.21214790e-02 1.19270563e+00 5.05514681e-01 -1.60391759e-02
7.98848748e-01 7.74134576e-01 -3.96112770e-01 2.69586474e-01
-9.23595488e-01 4.38793659e-01 3.15869570e-01 4.96074975e-01
-1.65609673e-01 2.66564637e-01 -2.14022383e-01 4.49386328e-01
-1.91280007e-01 -1.39281884e-01 6.99818552e-01 1.01653218e+00
-1.21460266e-01 -9.27407920e-01 -6.93367541e-01 7.56254315e-01
-2.90256619e-01 3.04425240e-01 5.03023624e-01 8.91115189e-01
-2.77707931e-02 1.01273930e+00 9.62388963e-02 3.65352154e-01
5.72940469e-01 -6.59959018e-02 1.79514840e-01 -3.75798911e-01
-1.90877080e-01 -2.26039186e-01 1.08217873e-01 -4.49198544e-01
2.85160393e-01 -8.90134573e-01 -1.27600956e+00 -2.81307995e-01
-6.79729640e-01 6.82882786e-01 9.57601309e-01 1.05779016e+00
-2.37709060e-01 9.17717814e-01 1.11401534e+00 -8.43363285e-01
-8.77269506e-01 -6.54231131e-01 -6.67054832e-01 -1.74608827e-01
1.49669480e-02 -8.31509948e-01 -7.24620163e-01 -5.10956824e-01]
|
[5.772492408752441, 2.625206470489502]
|
9c1240a4-b90e-4382-a016-a1a9c22224a9
|
cross-domain-contract-element-extraction-with
|
2105.06083
| null |
https://arxiv.org/abs/2105.06083v1
|
https://arxiv.org/pdf/2105.06083v1.pdf
|
Cross-Domain Contract Element Extraction with a Bi-directional Feedback Clause-Element Relation Network
|
Contract element extraction (CEE) is the novel task of automatically identifying and extracting legally relevant elements such as contract dates, payments, and legislation references from contracts. Automatic methods for this task view it as a sequence labeling problem and dramatically reduce human labor. However, as contract genres and element types may vary widely, a significant challenge for this sequence labeling task is how to transfer knowledge from one domain to another, i.e., cross-domain CEE. Cross-domain CEE differs from cross-domain named entity recognition (NER) in two important ways. First, contract elements are far more fine-grained than named entities, which hinders the transfer of extractors. Second, the extraction zones for cross-domain CEE are much larger than for cross-domain NER. As a result, the contexts of elements from different domains can be more diverse. We propose a framework, the Bi-directional Feedback cLause-Element relaTion network (Bi-FLEET), for the cross-domain CEE task that addresses the above challenges. Bi-FLEET has three main components: (1) a context encoder, (2) a clause-element relation encoder, and (3) an inference layer. To incorporate invariant knowledge about element and clause types, a clause-element graph is constructed across domains and a hierarchical graph neural network is adopted in the clause-element relation encoder. To reduce the influence of context variations, a multi-task framework with a bi-directional feedback scheme is designed in the inference layer, conducting both clause classification and element extraction. The experimental results over both cross-domain NER and CEE tasks show that Bi-FLEET significantly outperforms state-of-the-art baselines.
|
['Maarten de Rijke', 'Hongsong Li', 'Xiaozhong Liu', 'Zhumin Chen', 'Pengjie Ren', 'Zhaochun Ren', 'Hongye Song', 'Zihan Wang']
|
2021-05-13
| null | null | null | null |
['cross-domain-named-entity-recognition']
|
['natural-language-processing']
|
[ 2.86495447e-01 1.65162906e-01 -4.91681427e-01 -6.83627784e-01
-1.04017055e+00 -9.06028390e-01 4.62611973e-01 -2.28470396e-02
-4.73005205e-01 7.93695211e-01 4.76398855e-01 -5.40680647e-01
1.04966305e-01 -8.82416904e-01 -7.23213851e-01 -2.36925468e-01
9.37854573e-02 8.27859759e-01 1.79652900e-01 -4.04245883e-01
-7.88460746e-02 2.19768703e-01 -7.98124194e-01 6.41250849e-01
1.06544888e+00 8.25176179e-01 5.35026267e-02 5.05409352e-02
-5.55194557e-01 9.10287440e-01 -7.40448415e-01 -8.78182769e-01
1.86143264e-01 -3.75604004e-01 -1.08276224e+00 -8.61039981e-02
-6.90869838e-02 -1.25228226e-01 -2.55842894e-01 1.13684678e+00
3.14754367e-01 -1.49371326e-01 7.29461312e-01 -1.07895660e+00
-6.80054426e-01 1.06324339e+00 -5.98143220e-01 -1.46204278e-01
3.44154626e-01 -4.44129892e-02 1.43285310e+00 -7.86937535e-01
1.09769940e+00 1.15969837e+00 5.38663328e-01 5.90855658e-01
-1.13256788e+00 -9.13328350e-01 1.85044602e-01 7.49997124e-02
-1.49495089e+00 -1.68235272e-01 9.24975991e-01 -6.08717144e-01
1.22556174e+00 2.73908004e-02 2.86389709e-01 1.21655345e+00
5.58921993e-02 8.53639066e-01 7.57423222e-01 -2.47238442e-01
9.56667960e-02 1.52646052e-02 3.77827771e-02 4.20385659e-01
2.54768133e-01 4.52502631e-02 -1.38484627e-01 -8.07413906e-02
6.08962297e-01 -1.69098094e-01 -2.85352379e-01 -2.57747829e-01
-8.81646454e-01 8.92230988e-01 2.53804892e-01 6.67395592e-01
-3.38607579e-01 -1.48048118e-01 9.45738494e-01 2.60182261e-01
2.80634522e-01 3.79526734e-01 -8.41468692e-01 -1.19317934e-01
-6.30897462e-01 3.82101208e-01 1.12438333e+00 1.49675238e+00
7.33379960e-01 -4.13549125e-01 -1.85281754e-01 9.62508082e-01
8.36853981e-02 1.39058873e-01 2.27179810e-01 -4.14571166e-01
1.32519507e+00 9.68510568e-01 -3.71461250e-02 -7.01926589e-01
-2.14168444e-01 -4.49708611e-01 -9.07420278e-01 -1.47607118e-01
9.15502608e-02 -3.92441273e-01 -8.68126988e-01 1.83780682e+00
2.68394589e-01 -1.85079485e-01 4.96883765e-02 7.72543490e-01
8.58742714e-01 4.96078461e-01 1.98608130e-01 -8.24292377e-02
1.65547407e+00 -9.27124381e-01 -8.28892052e-01 -5.90343833e-01
8.13596904e-01 -4.58528399e-01 6.98728263e-01 -4.36043628e-02
-7.70805538e-01 -3.62479210e-01 -9.07718539e-01 -3.61612141e-01
-6.26022995e-01 5.87876141e-03 6.97388411e-01 3.03423733e-01
-1.82803780e-01 2.79203653e-01 -4.68811214e-01 2.05195054e-01
5.55341482e-01 2.09617466e-01 -4.26321000e-01 -3.06167573e-01
-1.75660050e+00 7.06134498e-01 9.85260785e-01 6.14217669e-02
-2.79825509e-01 -6.20007455e-01 -1.37028027e+00 5.00974178e-01
8.80473256e-01 -4.43831354e-01 1.27553415e+00 -8.89263690e-01
-1.04803157e+00 9.02983725e-01 -1.77639872e-01 -3.05507869e-01
3.03259254e-01 -5.55303879e-02 -8.52829993e-01 -2.71028757e-01
3.80206525e-01 3.11486214e-01 3.75936002e-01 -1.12789404e+00
-9.06090856e-01 -2.93292105e-01 1.62792385e-01 -1.23865129e-02
2.77058721e-01 2.16169611e-01 -7.15094984e-01 -8.38178992e-01
-1.58868238e-01 -9.23370600e-01 -5.02232127e-02 -7.02790201e-01
-4.45078164e-01 -7.32987285e-01 6.40380502e-01 -7.50139296e-01
1.64435422e+00 -2.24528527e+00 -1.66004803e-02 2.95404792e-01
2.57207721e-01 3.81826311e-01 -1.78528100e-01 6.37855232e-01
-3.98232788e-01 2.81391650e-01 -4.27512825e-01 -4.37258296e-02
3.01267028e-01 4.84349459e-01 -1.44092515e-01 -1.00455940e-01
6.71036363e-01 1.05770695e+00 -9.27553296e-01 -5.59358180e-01
-2.24419236e-01 1.37710437e-01 -5.36113203e-01 1.13074362e-01
-5.16498148e-01 3.05439204e-01 -5.61167479e-01 6.82525933e-01
7.34024704e-01 -2.71675348e-01 7.54016936e-01 -2.20055550e-01
1.90070868e-02 8.40554655e-01 -1.29752100e+00 1.70676494e+00
-5.35409331e-01 3.54128391e-01 1.51918694e-01 -1.16237962e+00
7.96948075e-01 5.74044287e-01 5.15717685e-01 -8.37141037e-01
1.53361633e-01 4.90899503e-01 1.34753779e-01 -4.59118068e-01
3.64507765e-01 -3.09384465e-01 -6.95331454e-01 2.25964129e-01
1.75216109e-01 2.30193391e-01 6.70448661e-01 2.00678349e-01
1.23451269e+00 3.62993926e-02 6.12891138e-01 5.00323018e-03
5.26191115e-01 3.12856212e-02 1.46818912e+00 3.16853285e-01
1.15678702e-02 2.04112932e-01 8.14130127e-01 -4.87075448e-01
-7.27996171e-01 -7.70792961e-01 -2.78161243e-02 9.84298825e-01
1.36850089e-01 -6.91107213e-01 -5.30110896e-01 -1.30378652e+00
1.52989313e-01 4.94562536e-01 -3.93753707e-01 2.53162473e-01
-1.00003850e+00 -2.31282085e-01 6.21346653e-01 8.53501081e-01
5.26702046e-01 -1.25473070e+00 -8.47614408e-02 5.89708328e-01
-5.23392320e-01 -1.48654723e+00 -9.74719524e-01 4.24960375e-01
-4.80872393e-01 -1.25817001e+00 -1.77114278e-01 -1.11752975e+00
2.79618710e-01 -3.25340092e-01 1.61220181e+00 -1.50696263e-01
1.49293855e-01 -4.02194560e-01 -4.58230704e-01 -4.89201337e-01
-3.20314020e-01 4.01637733e-01 -5.39041102e-01 -1.80684611e-01
1.03218353e+00 -4.62467670e-01 -2.76541114e-01 2.83182174e-01
-1.06991959e+00 -3.19447890e-02 8.30356181e-01 9.64852512e-01
7.55082190e-01 2.54409403e-01 6.70790255e-01 -1.87232268e+00
6.80119634e-01 -4.19120461e-01 -7.54737020e-01 5.22903204e-01
-6.73855424e-01 1.59056738e-01 6.50989413e-01 -1.11017853e-01
-1.26634395e+00 1.50471941e-01 -3.34159523e-01 -7.29526356e-02
-6.69407099e-02 8.58706594e-01 -8.24782431e-01 5.52412152e-01
3.71894956e-01 -8.16978738e-02 -7.19302475e-01 -5.31868339e-01
4.28687185e-01 1.01036239e+00 7.54910171e-01 -8.10391843e-01
9.26667035e-01 -8.06962326e-02 -4.45512831e-01 -2.81641006e-01
-7.66179025e-01 -6.78185403e-01 -6.13217413e-01 3.73883098e-01
1.02811432e+00 -9.80350077e-01 -6.12462282e-01 2.87718296e-01
-1.51530325e+00 -1.42216027e-01 -2.31845602e-01 3.48722577e-01
-3.16524953e-01 1.75311536e-01 -9.45449412e-01 -4.76850927e-01
-4.40235823e-01 -1.07162690e+00 1.08566165e+00 1.96159557e-01
-2.58608550e-01 -8.64969015e-01 5.12841009e-02 6.35078847e-01
-1.60992876e-01 3.34767044e-01 1.43560266e+00 -9.13358212e-01
-6.20167494e-01 6.65729959e-03 -4.20062006e-01 2.47715443e-01
2.03013554e-01 -4.59244430e-01 -5.26403248e-01 -1.90054119e-01
-2.04617217e-01 -2.28100315e-01 6.57748044e-01 -4.04246897e-02
9.73213613e-01 -5.27760267e-01 -5.48836768e-01 6.15063310e-01
1.50762820e+00 6.46811247e-01 7.64450967e-01 3.26967835e-01
7.72748470e-01 5.27804136e-01 6.66077256e-01 2.55015314e-01
6.40361369e-01 6.70688033e-01 -4.63608950e-02 -2.17779294e-01
-2.53115278e-02 -4.15770382e-01 1.13838807e-01 7.83756316e-01
1.77357346e-02 -3.29685211e-01 -1.05848193e+00 7.29712605e-01
-1.75864422e+00 -1.09232295e+00 1.23402104e-01 1.69677520e+00
1.24825644e+00 3.44285667e-01 4.71703662e-03 1.43384689e-03
8.39879155e-01 1.22777410e-01 -5.97522616e-01 -3.29214334e-01
1.19702471e-02 4.06735361e-01 4.79205906e-01 2.05843791e-01
-1.28534985e+00 9.83950973e-01 4.64877748e+00 9.76747692e-01
-8.00233066e-01 5.95825538e-02 1.74394488e-01 3.93983245e-01
-4.23215985e-01 2.42920205e-01 -8.59912872e-01 6.54001653e-01
4.87839848e-01 1.83574986e-02 3.15319598e-01 8.95549953e-01
-3.18746299e-01 4.13915634e-01 -1.40035582e+00 8.75934422e-01
-3.06023329e-01 -1.36587429e+00 -1.61145374e-01 2.58851320e-01
6.72144771e-01 2.61037886e-01 -5.42246282e-01 9.02418256e-01
5.87130547e-01 -8.36255789e-01 6.48459435e-01 -3.48879956e-02
1.22962856e+00 -9.25688326e-01 1.11805260e+00 2.89735287e-01
-1.64273918e+00 -7.39713088e-02 -3.07656098e-02 3.54655713e-01
4.59718466e-01 6.43521905e-01 -6.60512626e-01 1.04300320e+00
3.89038175e-01 6.74028099e-01 9.19148419e-03 6.41877532e-01
-5.62236726e-01 4.92741138e-01 -8.36940780e-02 2.89168686e-01
3.25542480e-01 -3.94847631e-01 3.72380465e-01 1.71138358e+00
-4.09833528e-02 4.13991153e-01 3.29636157e-01 9.66533303e-01
-5.90125322e-01 -7.44138658e-02 -4.24451143e-01 -3.41370672e-01
7.47753501e-01 1.03269827e+00 -2.83880025e-01 -3.03960323e-01
-9.44340110e-01 9.22611356e-01 5.43646276e-01 4.80580688e-01
-8.50951433e-01 -9.78958607e-01 7.54752874e-01 -8.87517929e-02
8.33349288e-01 -5.88073172e-02 -1.24919638e-01 -1.32859385e+00
2.48433545e-01 -1.15380263e+00 7.70079017e-01 -2.69879282e-01
-1.45125830e+00 6.85435712e-01 -2.17094570e-01 -1.14148617e+00
-4.86895025e-01 -5.90629995e-01 -3.74651700e-01 8.85564506e-01
-1.66855335e+00 -1.20773268e+00 2.35722020e-01 5.83125949e-01
5.20180941e-01 -1.08437397e-01 7.08511233e-01 8.94642413e-01
-5.42712569e-01 8.88630450e-01 -1.44089296e-01 1.25094879e+00
5.44718504e-01 -1.31090081e+00 5.42517900e-01 8.14495206e-01
3.48736167e-01 9.42764521e-01 -2.62340289e-02 -9.09776568e-01
-1.07083750e+00 -1.28238308e+00 1.46100152e+00 -2.35760748e-01
5.42132378e-01 -8.36383104e-01 -1.06005323e+00 1.04110956e+00
1.37496730e-02 -1.24858320e-01 8.29773724e-01 5.17436445e-01
-7.62349904e-01 -7.94779137e-02 -9.07506406e-01 4.14747357e-01
1.38305616e+00 -9.61137772e-01 -9.78736043e-01 1.43962055e-01
8.26648235e-01 -5.86669505e-01 -1.12500703e+00 4.27229524e-01
3.66091937e-01 -6.39706433e-01 6.43890440e-01 -6.77816808e-01
4.93680507e-01 -2.36731485e-01 -1.39426095e-02 -1.11199379e+00
-4.08677936e-01 -6.13836348e-01 -8.48420337e-02 1.76979899e+00
7.44742393e-01 -5.03997982e-01 6.38668120e-01 7.01077342e-01
-2.14876503e-01 -7.81514168e-01 -9.01680171e-01 -9.34505284e-01
2.75121368e-02 -3.36903185e-01 1.01658165e+00 1.19835818e+00
1.70298323e-01 1.07760179e+00 -7.16821328e-02 1.40807003e-01
1.69806629e-01 5.17864823e-01 3.97926390e-01 -1.32963276e+00
-4.50533450e-01 -3.24057162e-01 -2.16801703e-01 -1.17253172e+00
4.18420821e-01 -1.10241199e+00 1.16779581e-01 -1.58869183e+00
1.95090532e-01 -6.24052286e-01 -1.98505059e-01 6.99577987e-01
-1.49219155e-01 -4.03056145e-01 2.08250776e-01 1.89347357e-01
-4.63713169e-01 3.30027878e-01 1.15486264e+00 -3.26210767e-01
-1.13066122e-01 -1.33461252e-01 -7.88720965e-01 5.33969641e-01
4.35278177e-01 -7.69774139e-01 -2.85221130e-01 -6.52291477e-01
3.37919921e-01 3.39489073e-01 -3.59428257e-01 -4.93222743e-01
3.06814671e-01 -2.31312841e-01 8.66328105e-02 -6.37161314e-01
-1.73892140e-01 -8.93773317e-01 -3.81468721e-02 9.90197361e-02
-1.96948946e-01 -6.48043677e-02 -3.69311459e-02 6.93798602e-01
-6.68249786e-01 -2.95871526e-01 3.62471998e-01 -2.35447928e-01
-8.28581631e-01 3.55976790e-01 -9.85438153e-02 6.82254314e-01
6.78916454e-01 -5.09935245e-02 -4.22012299e-01 -1.95020679e-02
-4.56279248e-01 4.60146546e-01 2.23508716e-01 5.31699896e-01
2.56467402e-01 -1.41290402e+00 -8.06643248e-01 2.20429435e-01
4.25843805e-01 5.00569940e-01 1.15000069e-01 3.46772552e-01
-1.20093435e-01 5.29943228e-01 7.50192925e-02 -1.25549093e-01
-1.13584375e+00 5.61801314e-01 1.44805968e-01 -1.02326787e+00
-4.09250170e-01 7.81569362e-01 3.19315404e-01 -7.48857141e-01
8.02608728e-02 -4.56104875e-01 -2.85449505e-01 2.04327703e-02
2.34651610e-01 -1.41838476e-01 2.30801880e-01 -5.21317840e-01
-5.52499056e-01 3.78061563e-01 -4.75837409e-01 1.05312377e-01
1.48029351e+00 2.17658490e-01 -2.43227169e-01 -1.34754404e-01
1.19580197e+00 1.88791931e-01 -9.72983122e-01 -5.81472516e-01
6.37223840e-01 -2.85415530e-01 -1.82878509e-01 -1.04003978e+00
-1.16838503e+00 4.88337874e-01 -2.62822770e-02 -4.57688794e-02
1.23474205e+00 1.89481854e-01 1.26543581e+00 2.60956436e-01
4.48989332e-01 -1.34236860e+00 -3.96038145e-01 9.07579303e-01
7.63891816e-01 -1.19607568e+00 -3.84620965e-01 -8.19659114e-01
-7.68320441e-01 7.79254019e-01 6.12928748e-01 1.81970730e-01
3.95505071e-01 5.88402450e-01 2.25258395e-02 -4.92625862e-01
-5.45525670e-01 -3.92278850e-01 3.47031325e-01 5.86768866e-01
6.89959705e-01 1.78180873e-01 -7.52518773e-01 1.19810843e+00
-9.52976644e-02 2.03381866e-01 1.13954946e-01 1.00135708e+00
1.29832879e-01 -1.65463233e+00 2.07724214e-01 2.76415378e-01
-6.09954596e-01 -2.09613159e-01 -4.68310148e-01 9.31214452e-01
5.60230255e-01 9.19705689e-01 -4.17768657e-02 -4.54964697e-01
5.94301701e-01 1.63308427e-01 1.77658796e-01 -1.01062238e+00
-9.47430730e-01 -3.92415486e-02 5.75137794e-01 -5.25086522e-01
-2.19563156e-01 -4.69691843e-01 -1.51415527e+00 -1.16123594e-01
-5.47027349e-01 4.37989056e-01 4.02124673e-01 1.10572553e+00
4.61870700e-01 7.03777075e-01 4.35605824e-01 8.03830475e-02
-4.64592427e-01 -8.29146326e-01 -7.78849840e-01 8.45415771e-01
8.96369200e-03 -5.84553838e-01 1.88189328e-01 -8.14362019e-02]
|
[9.337051391601562, 8.881819725036621]
|
51f50f2f-3ef4-46e3-9043-f5b728a757f9
|
deep-speech-2-end-to-end-speech-recognition
|
1512.02595
| null |
http://arxiv.org/abs/1512.02595v1
|
http://arxiv.org/pdf/1512.02595v1.pdf
|
Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
|
We show that an end-to-end deep learning approach can be used to recognize
either English or Mandarin Chinese speech--two vastly different languages.
Because it replaces entire pipelines of hand-engineered components with neural
networks, end-to-end learning allows us to handle a diverse variety of speech
including noisy environments, accents and different languages. Key to our
approach is our application of HPC techniques, resulting in a 7x speedup over
our previous system. Because of this efficiency, experiments that previously
took weeks now run in days. This enables us to iterate more quickly to identify
superior architectures and algorithms. As a result, in several cases, our
system is competitive with the transcription of human workers when benchmarked
on standard datasets. Finally, using a technique called Batch Dispatch with
GPUs in the data center, we show that our system can be inexpensively deployed
in an online setting, delivering low latency when serving users at scale.
|
['Sherjil Ozair', 'Billy Jun', 'Jesse Engel', 'Erich Elsen', 'Jingdong Chen', 'Jared Casper', 'Greg Diamos', 'David Seetapun', 'Bo Xiao', 'Awni Hannun', 'Andrew Ng', 'Sharan Narang', 'Eric Battenberg', 'Christopher Fougner', 'Adam Coates', 'Zhiqian Wang', 'Zhenyao Zhu', 'Tony Han', 'Patrick LeGresley', 'Mike Chrzanowski', 'Linxi Fan', 'Libby Lin', 'Jun Zhan', 'Jonathan Raiman', 'Dario Amodei', 'Carl Case', 'Yi Wang', 'Shubho Sengupta', 'Sanjeev Satheesh', 'Ryan Prenger', 'Rishita Anubhai', 'Dani Yogatama', 'Chong Wang', 'Bryan Catanzaro']
|
2015-12-08
| null | null | null | null |
['noisy-speech-recognition', 'accented-speech-recognition']
|
['speech', 'speech']
|
[-3.23258311e-01 -3.75989974e-01 4.50718671e-01 -4.05211210e-01
-1.23438776e+00 -8.83179426e-01 3.28856975e-01 -2.91632473e-01
-4.94328886e-01 2.35110179e-01 1.88910246e-01 -8.03321242e-01
3.65860939e-01 -3.26845467e-01 -5.51109731e-01 -4.99139279e-01
-9.10499468e-02 5.87563694e-01 2.57568300e-01 -1.06556527e-01
2.71505285e-02 5.46392977e-01 -1.47938323e+00 5.25224090e-01
1.86213881e-01 5.91065586e-01 3.48476559e-01 1.35583794e+00
2.56480537e-02 8.09175014e-01 -7.35571742e-01 -2.46350214e-01
4.14480388e-01 -9.90928262e-02 -9.75623250e-01 -1.07741512e-01
5.83937943e-01 -4.96193439e-01 -2.94679105e-01 6.23273551e-01
1.12937677e+00 8.05316865e-02 -1.23195171e-01 -9.37177718e-01
-2.71689802e-01 4.92168754e-01 -3.41658384e-01 2.85358757e-01
4.25328612e-01 4.49913830e-01 9.20461357e-01 -8.34340036e-01
2.85988986e-01 1.08204532e+00 9.19867873e-01 3.76404524e-01
-1.16302693e+00 -6.15692794e-01 -4.79348227e-02 -1.66739047e-01
-1.27201450e+00 -9.68736887e-01 2.59819180e-01 -2.26077348e-01
1.53804171e+00 4.35170025e-01 4.51850772e-01 9.99351799e-01
-5.76222390e-02 7.41221011e-01 9.52266097e-01 -3.64176393e-01
1.97214141e-01 -4.77899574e-02 -1.40416063e-02 6.63957298e-01
-2.36076713e-01 -8.78323391e-02 -6.45121038e-01 -3.36585313e-01
2.69916326e-01 -6.04455620e-02 -1.74933463e-01 3.18620503e-01
-1.37769067e+00 3.56053770e-01 3.11892200e-03 3.63155335e-01
-1.80019036e-01 2.11763307e-01 6.77986979e-01 3.50626558e-01
5.27212799e-01 4.09423441e-01 -9.67837691e-01 -7.74289906e-01
-1.15561664e+00 2.57635981e-01 1.37627184e+00 9.30800736e-01
5.02077281e-01 5.09484261e-02 2.54674822e-01 7.39431024e-01
-6.46084175e-02 4.77889866e-01 6.16680384e-01 -1.27732396e+00
3.04006934e-01 -6.98670521e-02 -1.08841531e-01 -4.07420367e-01
-5.90538442e-01 -5.36447108e-01 -6.67218089e-01 1.91507444e-01
5.47161877e-01 -5.60579479e-01 -7.60506332e-01 1.45811033e+00
2.33217850e-01 3.56893778e-01 -6.35427684e-02 9.10933435e-01
4.09968138e-01 7.07611561e-01 -1.56422257e-01 1.57461405e-01
1.60385919e+00 -1.07414675e+00 -2.66347647e-01 -2.43903518e-01
7.83555567e-01 -1.09986067e+00 1.42683804e+00 6.72053695e-01
-1.18668902e+00 -4.79402959e-01 -8.06365490e-01 -3.29091251e-01
-2.85418272e-01 -6.41520619e-02 5.88234961e-01 7.73247838e-01
-1.58603573e+00 6.35640979e-01 -1.25846505e+00 -4.37854081e-01
8.24487135e-02 5.51046252e-01 -1.92050770e-01 2.09598124e-01
-6.17512465e-01 5.37130713e-01 5.53930402e-02 -1.25697136e-01
-7.23620355e-01 -8.42575490e-01 -2.68721223e-01 2.73870796e-01
4.73924913e-02 -5.45421720e-01 1.96228158e+00 -9.50539887e-01
-1.69442284e+00 7.83792377e-01 -2.38413259e-01 -3.28064919e-01
3.73215526e-01 -2.69880533e-01 -2.79813170e-01 -1.07911611e-02
-1.71549872e-01 3.01997513e-01 6.22749567e-01 -5.46172798e-01
-8.24026525e-01 -2.08127305e-01 -1.61568701e-01 2.29707941e-01
-4.08042759e-01 5.22523344e-01 -7.15414226e-01 -3.86377186e-01
-2.24225163e-01 -1.06665611e+00 -1.49526045e-01 -3.65378946e-01
-2.00292394e-01 -1.50356546e-01 7.32859671e-01 -8.50220382e-01
7.74139166e-01 -2.48955011e+00 -2.02633321e-01 -5.02011552e-02
4.40374315e-01 3.41380298e-01 2.01632455e-03 4.53947365e-01
1.48605525e-01 1.76939398e-01 -4.17549685e-02 -6.82936251e-01
1.75898135e-01 1.13243856e-01 -1.97512880e-01 2.90509194e-01
3.23817953e-02 5.20691812e-01 -7.88950026e-01 -2.53105134e-01
-2.25893348e-01 5.72012961e-01 -6.92941368e-01 3.89919788e-01
-8.77674855e-03 4.51270133e-01 -1.35333732e-01 4.65267479e-01
5.39294720e-01 -2.77560353e-01 3.19642901e-01 3.71720433e-01
-3.29297274e-01 7.04834998e-01 -1.01765943e+00 1.92421579e+00
-7.82829106e-01 1.03149092e+00 7.02665746e-01 -7.00303316e-01
5.03902555e-01 4.97015536e-01 1.99127197e-01 -5.57938278e-01
-1.37207553e-01 5.02890527e-01 1.53072625e-01 -3.92339975e-01
4.42034006e-01 1.73852578e-01 2.11198274e-02 7.93158650e-01
6.68850318e-02 -2.09781349e-01 -5.19333128e-03 1.64052978e-01
1.48587310e+00 -5.03748432e-02 -1.63188681e-01 -4.26807910e-01
7.27103576e-02 7.72825330e-02 4.57225859e-01 8.39364588e-01
-3.35250467e-01 6.91753626e-01 5.36603689e-01 -7.77303815e-01
-1.30405259e+00 -8.93666983e-01 6.28953874e-02 1.82599962e+00
-7.82155752e-01 -4.55915451e-01 -1.08477771e+00 -3.68752837e-01
-2.87038088e-01 3.10265422e-01 9.25735384e-02 4.05924082e-01
-8.88531923e-01 -6.10282004e-01 9.50076759e-01 6.85669422e-01
3.06679517e-01 -9.29352582e-01 -6.21509016e-01 2.27038041e-01
1.61790356e-01 -1.24021387e+00 -5.77998400e-01 4.83542770e-01
-5.28006375e-01 -5.73351979e-01 -6.88171208e-01 -1.02005696e+00
1.91690072e-01 2.95639306e-01 1.50542891e+00 3.07558894e-01
-2.97806203e-01 1.95730641e-01 -1.51893079e-01 -3.43478709e-01
-6.83789134e-01 4.63918358e-01 2.19434530e-01 -3.39531183e-01
3.70008379e-01 -7.34450758e-01 -5.93344390e-01 -5.93487993e-02
-5.46721697e-01 1.83632940e-01 4.47230399e-01 7.10689127e-01
1.12457596e-01 -8.35047066e-02 2.13776037e-01 -9.23045516e-01
6.93971813e-01 -4.33275521e-01 -8.21209490e-01 9.70172957e-02
-4.20456976e-01 5.00608012e-02 1.02853882e+00 -2.87473619e-01
-7.73474276e-01 3.39129865e-01 -7.06955016e-01 -1.92812636e-01
-4.16137099e-01 2.42339894e-01 1.76130444e-01 5.77963702e-02
6.73157752e-01 1.35562882e-01 6.70194402e-02 -6.56845152e-01
3.34379971e-01 1.09635806e+00 6.63386643e-01 -6.26443326e-01
2.57434666e-01 1.60405159e-01 -4.40668344e-01 -9.68761444e-01
-4.63244170e-01 -4.70142037e-01 -4.66332078e-01 1.85604393e-01
7.44881272e-01 -1.25382984e+00 -1.07443321e+00 6.49269819e-01
-1.35226440e+00 -7.17981160e-01 1.63537934e-01 4.24361855e-01
-1.71314016e-01 2.42544502e-01 -1.11908162e+00 -7.04818308e-01
-5.81010938e-01 -1.30425048e+00 1.40621066e+00 1.18306786e-01
-1.64661229e-01 -8.44447911e-01 1.03157297e-01 3.40935111e-01
7.26268649e-01 -2.98211724e-01 4.83957827e-01 -9.87715304e-01
-4.72223580e-01 -7.66934976e-02 -1.71836436e-01 4.27869797e-01
-2.43221030e-01 2.10629016e-01 -1.39272165e+00 -6.57223165e-01
6.61590546e-02 -3.98482472e-01 5.15905857e-01 -1.52916620e-02
1.23717189e+00 -2.50804454e-01 2.22836807e-02 7.71311820e-01
9.51007068e-01 4.89619039e-02 2.01908529e-01 7.23685175e-02
7.45318890e-01 3.55840474e-01 -3.71417813e-02 4.02304292e-01
3.66908044e-01 5.19948959e-01 -1.67200729e-01 -3.87180179e-01
2.00663805e-02 2.22128928e-01 3.97938311e-01 1.35276008e+00
3.73474956e-02 -1.34318620e-01 -1.49622858e+00 4.47828352e-01
-1.59902060e+00 -8.30827177e-01 -6.59625381e-02 2.04696059e+00
7.84017026e-01 1.87893480e-01 5.38441360e-01 -1.00523822e-01
3.62750560e-01 -9.44616273e-02 -4.78800356e-01 -7.09760010e-01
2.61066139e-01 6.54304862e-01 5.15986204e-01 5.35523176e-01
-1.01992619e+00 9.09506917e-01 7.43640089e+00 6.05015934e-01
-1.52530909e+00 3.62763047e-01 8.37738037e-01 -3.47950965e-01
6.98939115e-02 -5.84795587e-02 -6.77641213e-01 5.27075589e-01
1.59779906e+00 -2.45934408e-02 9.90177155e-01 1.01491189e+00
2.41864294e-01 2.04569906e-01 -1.01361728e+00 9.58587706e-01
-2.12731600e-01 -1.20368207e+00 -8.28480899e-01 1.96783438e-01
5.26855230e-01 8.58849823e-01 -4.29449528e-02 3.88569295e-01
6.82124436e-01 -9.82994676e-01 6.76194370e-01 4.88900999e-03
6.85804367e-01 -8.04745018e-01 6.11129344e-01 6.79799974e-01
-1.02569032e+00 8.43721554e-02 -3.41905087e-01 -3.09076816e-01
-6.71601295e-03 5.69906175e-01 -1.15365565e+00 2.12559685e-01
9.08000827e-01 -6.60899580e-02 -4.42137301e-01 6.78606749e-01
8.30266476e-02 9.61109757e-01 -6.38879836e-01 -1.65423334e-01
2.62906700e-01 2.37166300e-01 1.29601836e-01 1.83332717e+00
4.70655471e-01 -3.74312140e-02 3.31084937e-01 3.97479862e-01
-1.31837860e-01 -1.00783482e-01 -6.88278973e-01 -7.38740712e-02
5.46741188e-01 1.43102944e+00 -6.94797993e-01 -4.56553429e-01
-6.04360640e-01 1.17272043e+00 5.51605761e-01 2.83227086e-01
-7.62946784e-01 -6.59447074e-01 8.73281598e-01 -1.74792022e-01
3.49719554e-01 -7.03366041e-01 -2.71829605e-01 -1.12035453e+00
2.19173264e-02 -1.45418942e+00 -4.37968075e-02 -5.29609799e-01
-1.14696062e+00 9.19382215e-01 -6.01014793e-01 -6.67035818e-01
-5.41411340e-01 -7.64581323e-01 -6.53490305e-01 1.13520873e+00
-9.62299943e-01 -8.15412283e-01 -1.35272175e-01 4.11260009e-01
5.61213493e-01 -3.06795835e-01 1.04833996e+00 6.18246377e-01
-6.00657701e-01 5.77206433e-01 4.14270610e-01 2.57785648e-01
7.19488263e-01 -1.47240567e+00 1.20103276e+00 1.07950389e+00
4.93791193e-01 7.78463483e-01 5.59257686e-01 -2.03006238e-01
-1.77220404e+00 -7.70912349e-01 9.69772100e-01 -4.53856826e-01
7.73128510e-01 -9.89967525e-01 -8.52504790e-01 5.98697066e-01
6.82631910e-01 1.31044790e-01 7.49483764e-01 4.84522671e-01
-3.45535994e-01 -1.15278803e-01 -5.96538186e-01 5.24527609e-01
1.00552499e+00 -8.39275599e-01 -2.26023737e-02 7.05783367e-01
9.68507290e-01 -8.26317489e-01 -6.45278871e-01 -2.16472968e-01
6.83767438e-01 -1.10049343e+00 6.12260401e-01 -7.41331100e-01
1.13654524e-01 -1.99711621e-01 -1.82061464e-01 -1.26371777e+00
-3.62384856e-01 -1.19182730e+00 1.03228427e-01 1.14572787e+00
5.67743123e-01 -6.01485670e-01 6.58196568e-01 7.79222071e-01
-3.09876412e-01 -5.21619201e-01 -8.49462569e-01 -7.16066301e-01
4.68088314e-02 -7.56452560e-01 8.57231975e-01 8.13629329e-01
-2.02699587e-01 3.87352020e-01 -2.19839752e-01 2.63892859e-01
2.46566728e-01 -6.72440827e-02 9.01877761e-01 -8.57499540e-01
-8.94934595e-01 -3.95974696e-01 -1.51365459e-01 -1.07241225e+00
2.14269146e-01 -9.43746984e-01 2.69692630e-01 -1.05554473e+00
9.48854014e-02 -4.37439322e-01 -1.94762021e-01 6.87972009e-01
-9.95193422e-02 3.94358128e-01 3.02623481e-01 2.30632126e-01
-5.38224518e-01 -1.49731487e-01 7.16920674e-01 2.03838989e-01
-1.04535230e-01 -9.63180289e-02 -7.32616782e-01 7.03632116e-01
9.04480934e-01 -4.56510365e-01 -1.79744616e-01 -1.02609730e+00
1.17008768e-01 -1.13952518e-01 1.31294087e-01 -1.15100801e+00
5.05091667e-01 2.61723936e-01 2.03448072e-01 -2.93631405e-01
1.71860144e-01 -5.59665561e-01 -1.12827241e-01 3.13333869e-01
-1.92390487e-01 5.62822342e-01 3.45941424e-01 5.99826425e-02
-4.09665965e-02 5.85232191e-02 6.86177015e-01 -1.14799492e-01
-4.42421079e-01 -2.52312440e-02 -6.72482491e-01 2.56068379e-01
4.95178312e-01 3.74205858e-01 -4.82550025e-01 -4.21562344e-01
-5.63099384e-01 6.27408177e-02 4.33986247e-01 1.14530973e-01
6.70125261e-02 -8.85624051e-01 -8.29068720e-01 3.78853858e-01
-3.78839552e-01 -2.97246259e-02 2.55989600e-02 8.47977579e-01
-1.21892881e+00 3.10455918e-01 1.69477805e-01 -7.46588111e-01
-1.31915486e+00 2.95803636e-01 3.44750732e-01 -1.73981071e-01
-6.45351052e-01 1.07804668e+00 -1.59502819e-01 -7.65560925e-01
4.73654836e-01 -2.53860980e-01 7.38019288e-01 -4.41126972e-01
7.49569714e-01 1.75317049e-01 5.95513523e-01 -1.82264477e-01
-5.44419348e-01 -6.58256561e-02 5.86538874e-02 -6.10674381e-01
1.41510904e+00 2.74208009e-01 -2.35869586e-01 3.44227225e-01
1.43225431e+00 3.55624527e-01 -1.21832502e+00 1.54302977e-02
8.69495124e-02 -1.85823634e-01 1.31178588e-01 -7.22657204e-01
-8.67148995e-01 9.66311336e-01 7.06245601e-01 3.79759371e-01
1.33696890e+00 -1.01973869e-01 1.08543146e+00 8.60255361e-01
3.07868928e-01 -1.08733749e+00 -3.66103917e-01 8.25483203e-01
4.56855416e-01 -1.11204851e+00 -2.94392347e-01 5.75507954e-02
-5.16255498e-01 1.25555730e+00 1.89005673e-01 1.14766017e-01
5.45410216e-01 1.04677844e+00 3.65408331e-01 1.26961157e-01
-1.14498401e+00 -2.48136669e-02 -2.02702984e-01 4.43445683e-01
9.76406574e-01 3.21938068e-01 1.10905066e-01 3.81768107e-01
-5.66513240e-01 -2.60521442e-01 3.69300991e-01 1.04709899e+00
-2.61135161e-01 -1.31943035e+00 -3.85836363e-01 2.38491714e-01
-8.17467511e-01 -3.31433773e-01 -2.39671752e-01 6.04691267e-01
4.41605272e-03 1.07632494e+00 2.20313311e-01 -5.10589421e-01
2.14541137e-01 3.15879434e-01 1.99519813e-01 -6.08891606e-01
-1.17758203e+00 3.49102944e-01 2.82071710e-01 -6.92603290e-01
1.63883746e-01 -7.03750074e-01 -1.18149495e+00 -7.87609696e-01
3.26322131e-02 4.70302776e-02 1.11793292e+00 8.23488653e-01
9.19382930e-01 3.83841783e-01 7.30710983e-01 -1.01614392e+00
-6.13064349e-01 -8.75286400e-01 -4.73693848e-01 2.12292075e-01
2.69592613e-01 2.39968881e-01 -2.91651309e-01 7.55643696e-02]
|
[14.289448738098145, 6.491180419921875]
|
cf6ff13e-fdff-480d-985b-149cfaf47551
|
hyphen-hyperbolic-hawkes-attention-for-text
| null | null |
https://aclanthology.org/2022.acl-short.69
|
https://aclanthology.org/2022.acl-short.69.pdf
|
HYPHEN: Hyperbolic Hawkes Attention For Text Streams
|
Analyzing the temporal sequence of texts from sources such as social media, news, and parliamentary debates is a challenging problem as it exhibits time-varying scale-free properties and fine-grained timing irregularities. We propose a Hyperbolic Hawkes Attention Network (HYPHEN), which learns a data-driven hyperbolic space and models irregular powerlaw excitations using a hyperbolic Hawkes process. Through quantitative and exploratory experiments over financial NLP, suicide ideation detection, and political debate analysis we demonstrate HYPHEN’s practical applicability for modeling online text sequences in a geometry agnostic manner.
|
['Sudheer Chava', 'Ritesh Soun', 'Sanchit Ahuja', 'Ramit Sawhney', 'Shivam Agarwal']
| null | null | null | null |
acl-2022-5
|
['stock-price-prediction']
|
['time-series']
|
[-6.20617509e-01 2.28976727e-01 1.53639823e-01 5.33468127e-02
-8.82982194e-01 -8.85092974e-01 1.06171262e+00 4.49430615e-01
-3.02018672e-01 5.24359763e-01 7.30692685e-01 -8.81471753e-01
-4.28772897e-01 -8.36762249e-01 -4.54514980e-01 -5.90656638e-01
-3.51081103e-01 8.32873642e-01 9.44363773e-02 -4.35296774e-01
4.53355134e-01 3.56769919e-01 -3.61195356e-01 1.87379509e-01
7.54342914e-01 6.71550810e-01 -6.50116384e-01 1.11165166e+00
-9.46170092e-02 1.24127221e+00 -4.84432578e-01 -6.20397449e-01
2.34379008e-01 -3.54072005e-01 -8.35085511e-01 -4.05889720e-01
-5.20429164e-02 6.17087148e-02 -1.04420567e+00 7.36200035e-01
2.48368010e-01 2.20972613e-01 9.78454471e-01 -1.16062415e+00
-1.09745789e+00 9.40459549e-01 -9.02499676e-01 9.60771084e-01
-8.67366567e-02 1.77196980e-01 1.25197554e+00 -6.46992207e-01
5.88477492e-01 1.24422967e+00 1.10076022e+00 -1.32130692e-02
-8.64850819e-01 -3.22814673e-01 -3.36116850e-02 2.69994915e-01
-1.11523306e+00 5.45686722e-01 8.75285745e-01 -5.01003087e-01
6.77544296e-01 2.38542378e-01 8.68371725e-01 1.44093227e+00
6.98842168e-01 7.71137357e-01 7.35194087e-01 -3.46761905e-02
4.08445388e-01 -7.71006525e-01 3.44211668e-01 7.19019413e-01
-1.18343577e-01 -2.24138543e-01 -4.74357754e-01 -6.44381702e-01
5.72769642e-01 -1.38419449e-01 3.60335298e-02 3.78962189e-01
-1.20957530e+00 1.24246395e+00 2.08414346e-01 2.98073918e-01
-2.76127160e-01 2.60900706e-01 6.06939733e-01 2.11042523e-01
9.37248409e-01 6.55209124e-01 -4.02232200e-01 -5.37428677e-01
-7.96265483e-01 4.90372628e-01 9.38903332e-01 5.79979658e-01
1.92872986e-01 -2.00170502e-01 -6.94161057e-02 1.63652942e-01
6.30392432e-02 5.08794367e-01 4.83901203e-01 -9.28253293e-01
4.35729653e-01 5.29235244e-01 1.34090438e-01 -1.37120354e+00
-1.00042403e+00 -2.51465261e-01 -1.10689712e+00 -4.26395983e-01
8.51634026e-01 -4.61518317e-01 -4.05194789e-01 1.46076751e+00
5.32505810e-01 1.59389883e-01 -2.03999743e-01 8.18380892e-01
6.64003074e-01 1.10616958e+00 -5.62390015e-02 -2.90935665e-01
1.36701941e+00 -6.79780185e-01 -4.72910941e-01 4.90904242e-01
6.04255855e-01 -2.82429457e-01 1.03083694e+00 2.05271527e-01
-1.25810182e+00 1.04162544e-01 -4.77214336e-01 -4.35237229e-01
-3.77296984e-01 -5.89637995e-01 2.74581969e-01 1.30179718e-01
-7.02745080e-01 8.08007002e-01 -8.75460267e-01 -2.36453682e-01
3.19528908e-01 -3.74808550e-01 2.59007454e-01 6.48839295e-01
-1.29096472e+00 3.93801302e-01 -1.91649303e-01 3.98434773e-02
-4.07651514e-01 -1.14655149e+00 -5.31250000e-01 2.18804762e-01
2.05310985e-01 -5.42031288e-01 1.47655952e+00 -3.63059014e-01
-1.32250738e+00 5.72568297e-01 4.86092091e-01 -8.33818078e-01
1.14028203e+00 -1.26089249e-02 -2.36566365e-01 4.20421064e-01
-1.91191584e-02 -2.52792776e-01 6.48640454e-01 -3.91193837e-01
-1.09674975e-01 -3.47044259e-01 -5.93985282e-02 1.77228916e-02
-3.35863739e-01 2.71193348e-02 1.21809825e-01 -1.01779115e+00
-1.64464280e-01 -1.02623487e+00 -2.39309683e-01 -1.79407015e-01
-6.43382430e-01 -5.95035255e-01 9.71098542e-01 -9.06221807e-01
1.31141853e+00 -1.80950904e+00 1.21500462e-01 1.33880079e-01
6.32731616e-01 -3.55522066e-01 5.87593243e-02 9.58268702e-01
2.39143506e-01 4.38361734e-01 -1.93814084e-01 -2.60045025e-02
2.79067576e-01 -1.36031583e-01 -7.60957181e-01 1.05126071e+00
-1.24779955e-01 1.23411787e+00 -9.54784513e-01 -2.57809877e-01
-3.09522450e-01 2.58045107e-01 -5.98474622e-01 -1.75987929e-01
-5.13655484e-01 2.90594876e-01 -6.61201298e-01 2.63771832e-01
2.27591977e-01 -9.62825716e-01 1.54505000e-02 1.12229235e-01
-4.56435502e-01 2.37230230e-02 -5.58743715e-01 9.78625238e-01
-2.63537437e-01 1.34206343e+00 -2.21537918e-01 -9.50033188e-01
6.54197454e-01 1.76537052e-01 6.81850076e-01 -8.77082348e-01
4.62815583e-01 -3.78955424e-01 -8.98255482e-02 -4.70104694e-01
6.83482528e-01 -9.22923461e-02 -5.52949965e-01 1.14475954e+00
-3.47596675e-01 1.46653682e-01 -8.57346319e-03 7.41728306e-01
1.50918603e+00 -5.22940159e-01 2.39774719e-01 -6.95127785e-01
-1.22622497e-01 1.30869225e-01 3.01910669e-01 9.40130889e-01
-1.85654342e-01 8.26752603e-01 1.14057076e+00 -1.18264222e+00
-1.92220759e+00 -9.67738867e-01 2.21272577e-02 1.09195685e+00
-1.04744092e-01 -6.07759118e-01 -8.01700056e-01 -3.91286463e-01
-1.60190627e-01 5.87577999e-01 -1.23140931e+00 1.05489798e-01
-9.58445311e-01 -1.20244575e+00 6.81587398e-01 3.37720811e-01
4.34105724e-01 -9.76550639e-01 -4.78106618e-01 2.84511626e-01
-2.58961290e-01 -1.04097271e+00 -9.94623959e-01 -3.21026981e-01
-3.03310692e-01 -1.44985950e+00 -8.22472095e-01 -3.20524186e-01
2.34378934e-01 -2.38204077e-02 9.72331882e-01 -1.38357654e-01
-6.56385362e-01 5.34281254e-01 -2.01290160e-01 -4.54819977e-01
-4.54550892e-01 3.53316694e-01 3.82181145e-02 2.28913262e-01
1.05071274e-04 -5.90877831e-01 -8.65797758e-01 1.07832611e-01
-9.47499394e-01 1.41265453e-03 -1.97022468e-01 6.02650702e-01
2.55912654e-02 -6.22342676e-02 3.33132565e-01 -5.97040176e-01
1.07825303e+00 -8.74921858e-01 -8.19983184e-01 7.83479661e-02
-3.08949590e-01 8.15212280e-02 1.00869620e+00 -6.56575918e-01
-8.76770198e-01 -7.58502305e-01 3.25041234e-01 -1.86128929e-01
2.51984179e-01 6.26258016e-01 8.23914111e-01 2.39616171e-01
9.90043342e-01 1.18980668e-01 -9.03208256e-02 -2.70632327e-01
4.55296040e-01 2.31536686e-01 4.93180156e-01 -5.50564885e-01
1.03923953e+00 1.00667083e+00 6.52083308e-02 -1.26049435e+00
-9.92958546e-01 -3.63558590e-01 -3.36089760e-01 -1.68012902e-01
1.19873285e+00 -5.28780639e-01 -1.47473335e+00 4.96486396e-01
-1.38082278e+00 -7.83426881e-01 -4.50717360e-01 2.28592440e-01
-6.35891378e-01 4.94738013e-01 -1.31750810e+00 -8.41931045e-01
-5.56375682e-01 -2.60308713e-01 8.76622021e-01 2.09245369e-01
-4.55259204e-01 -1.44696856e+00 8.44676614e-01 1.44874349e-01
1.91940725e-01 6.03753686e-01 1.27191126e+00 -7.13566363e-01
-4.16374803e-01 5.11855111e-02 -3.50745559e-01 -6.95487916e-01
-5.56988418e-01 6.45667553e-01 -2.63994813e-01 5.64108137e-03
-6.11453839e-02 -6.57723323e-02 5.60582876e-01 6.58592939e-01
1.05607772e+00 -1.04286528e+00 1.11953557e-01 6.29456043e-01
9.48488653e-01 -1.46579504e-01 3.37394595e-01 5.81177294e-01
4.54903752e-01 3.00023794e-01 -1.39064744e-01 9.99899566e-01
5.81782162e-01 1.81814596e-01 1.92346498e-01 8.96268431e-03
4.56021249e-01 -3.91573668e-01 8.90101418e-02 1.03274608e+00
-1.39307752e-01 -5.21700621e-01 -1.40698183e+00 4.73807126e-01
-2.06415081e+00 -1.51373136e+00 -4.21827555e-01 1.28488529e+00
4.81585711e-01 1.68969985e-02 6.73375607e-01 -1.73564553e-01
7.59847403e-01 3.54266971e-01 -6.05322361e-01 -5.94130933e-01
-4.10110265e-01 -1.14597112e-01 5.73268116e-01 4.10499483e-01
-9.38282669e-01 4.42100942e-01 6.75243187e+00 6.39019489e-01
-8.95814657e-01 2.03592539e-01 9.63621855e-01 -1.06612749e-01
-4.96702343e-01 -3.01785082e-01 -2.02079222e-01 5.93508899e-01
1.10190344e+00 -8.65908623e-01 4.11125451e-01 4.51658517e-01
5.51450670e-01 3.92898142e-01 -3.07996958e-01 7.77266800e-01
-3.11490178e-01 -1.95994532e+00 -2.76229203e-01 2.64043957e-01
8.48285198e-01 4.82855111e-01 2.63303101e-01 1.64559439e-01
9.49806392e-01 -8.77643168e-01 8.45851421e-01 6.06057465e-01
3.40987414e-01 -7.29625344e-01 2.43564472e-01 6.00490630e-01
-1.03426778e+00 -3.34875017e-01 -1.63916320e-01 -8.83464143e-02
3.89152855e-01 4.87589329e-01 -7.95825124e-01 1.70659378e-01
8.02634716e-01 6.33145988e-01 -2.58155614e-01 7.58922338e-01
1.28594279e-01 1.21207881e+00 -5.87633252e-01 -4.27412182e-01
8.51759970e-01 -4.37176704e-01 8.00148726e-01 1.14357710e+00
4.50535297e-01 8.34615767e-01 -9.99101549e-02 8.62165570e-01
-2.80554563e-01 2.23704740e-01 -3.04984123e-01 -4.24206525e-01
1.19672596e-01 1.15851712e+00 -1.12902009e+00 -3.60540375e-02
-4.12511498e-01 5.77072084e-01 4.46822017e-01 5.73498189e-01
-1.16806054e+00 -1.22391351e-01 2.12947816e-01 5.20735264e-01
1.52799845e-01 -7.48536050e-01 -3.74965698e-01 -1.26477134e+00
-2.93385029e-01 -5.31652331e-01 9.18342531e-01 -7.91574001e-01
-1.60150361e+00 5.46374202e-01 -2.92702112e-02 -8.49219799e-01
-1.29950210e-01 -4.38928485e-01 -1.20898795e+00 3.05734754e-01
-7.94913530e-01 -8.82871449e-01 -7.54626021e-02 6.14521623e-01
5.38605154e-01 -6.78538978e-02 2.77176559e-01 -1.49105251e-01
-4.54947591e-01 2.10554153e-01 5.99923849e-01 4.87799555e-01
2.50821039e-02 -1.43270826e+00 9.72724199e-01 3.12118262e-01
1.54216871e-01 1.79253101e-01 9.98777270e-01 -6.90010726e-01
-1.33388364e+00 -1.11480355e+00 8.20083559e-01 -5.62929213e-01
1.68042171e+00 -4.08555180e-01 -9.69899952e-01 6.17658198e-01
3.99386913e-01 2.94408463e-02 7.15940773e-01 -9.22582448e-02
-5.56316912e-01 4.85297084e-01 -7.50047803e-01 9.42444801e-01
8.30953896e-01 -4.94194239e-01 -5.60905576e-01 9.97714758e-01
7.58823156e-01 -2.07424104e-01 -7.34193325e-01 -4.71319258e-02
3.01768541e-01 -5.75442851e-01 6.30563617e-01 -1.18701124e+00
6.67936683e-01 3.08253765e-01 3.50990772e-01 -1.23250461e+00
-6.26081824e-01 -1.56303108e+00 -3.94483656e-01 5.70808411e-01
2.56893724e-01 -4.01371419e-01 8.10392499e-01 4.57067192e-01
1.96980268e-01 -5.67350984e-01 -1.14772701e+00 -6.62124157e-01
7.70291448e-01 -2.40472928e-01 3.68375242e-01 1.12388504e+00
3.76887649e-01 3.28859866e-01 -3.21191221e-01 1.10008940e-01
8.13097298e-01 2.26485327e-01 5.02019048e-01 -1.22686756e+00
-1.50183067e-01 -5.47860801e-01 -1.37405336e-01 -7.32361734e-01
1.12647764e-01 -5.85031688e-01 -2.72243708e-01 -1.03150511e+00
3.31464708e-01 -5.54218963e-02 4.34908569e-01 -2.53366351e-01
2.13650793e-01 2.13104844e-01 5.99598996e-02 4.21440125e-01
-8.52255464e-01 7.23309755e-01 1.03643084e+00 -1.88404843e-01
-4.02575321e-02 -1.40766039e-01 -2.49053270e-01 8.08891237e-01
7.91785300e-01 -4.00932103e-01 7.08639547e-02 -8.43902975e-02
1.07693303e+00 3.61847252e-01 4.89229649e-01 -4.67370063e-01
5.04895270e-01 -2.88068503e-01 -1.62046589e-02 -7.61720181e-01
-1.92279026e-01 -1.11523718e-01 -1.48095936e-01 3.12479138e-01
-7.34840691e-01 7.05940962e-01 -4.53236401e-02 9.63791609e-01
2.23895237e-01 3.64519536e-01 7.11642921e-01 -1.85793310e-01
3.20391327e-01 8.66043210e-01 -8.80057514e-01 7.95366347e-01
9.65744436e-01 3.49123180e-01 -6.81046784e-01 -1.05627084e+00
-7.84075737e-01 2.96126783e-01 1.79644391e-01 2.38049954e-01
1.27198637e-01 -1.12998605e+00 -1.06470978e+00 -3.61024529e-01
-2.81457633e-01 -9.46233124e-02 5.77628791e-01 9.29814935e-01
-7.40328491e-01 2.91015387e-01 3.30322802e-01 -3.30791175e-01
-7.27690220e-01 6.92803741e-01 3.89528722e-01 -6.02590680e-01
-1.15574098e+00 2.75770932e-01 1.18536048e-01 -3.97149414e-01
-2.12255925e-01 -3.43701512e-01 -1.00953013e-01 2.12397426e-01
5.38193583e-01 6.51787996e-01 -1.94924563e-01 -4.82921958e-01
1.76585555e-01 3.18207443e-01 1.04427645e-02 -3.88260126e-01
1.56574965e+00 -1.89777747e-01 8.33426267e-02 6.41153991e-01
1.24736333e+00 -1.49983019e-01 -1.46563828e+00 -3.45065862e-01
1.79748803e-01 1.78822413e-01 -2.27043256e-01 -1.33000985e-01
-6.70356572e-01 6.97777867e-01 -3.34502786e-01 1.42412686e+00
2.66862094e-01 4.52148557e-01 1.05230439e+00 3.35162073e-01
-1.71645463e-01 -1.42792058e+00 5.11875391e-01 9.96018529e-01
9.45957601e-01 -7.48896956e-01 -2.59364158e-01 2.04739109e-01
-4.98855859e-01 1.39509773e+00 1.55099660e-01 -5.12171388e-01
9.62449551e-01 4.40875798e-01 -3.52991283e-01 -4.84422803e-01
-1.04155087e+00 5.26419699e-01 7.29267448e-02 1.04921788e-01
3.85366417e-02 -1.70352906e-01 -3.36554766e-01 8.14974666e-01
-5.03179312e-01 -6.23756230e-01 9.53252792e-01 4.87439871e-01
-2.95915782e-01 -8.58980492e-02 -4.20259506e-01 2.65279293e-01
-8.13882828e-01 -1.46967649e-01 -5.54165423e-01 8.26172769e-01
-7.89862216e-01 3.46315354e-01 6.95544243e-01 1.64149970e-01
-1.53926253e-01 -1.06439278e-01 1.54030733e-02 -2.48503722e-02
-6.19200230e-01 -1.28171980e-01 -1.31211698e-01 -3.22776020e-01
2.16144919e-01 -8.56215239e-01 -1.43137717e+00 -1.01265085e+00
2.23943949e-01 4.31986421e-01 4.89742905e-01 1.14870882e+00
4.47561145e-01 2.51212955e-01 7.88160801e-01 -4.68965113e-01
-8.71348798e-01 -8.46375823e-01 -5.15068233e-01 5.01631796e-01
5.69222093e-01 2.34922674e-02 -6.75051987e-01 -2.27318272e-01]
|
[6.91347599029541, 3.4296388626098633]
|
910cefb6-9832-4c47-b67b-b2a40f57e7be
|
polarmix-a-general-data-augmentation
|
2208.00223
| null |
https://arxiv.org/abs/2208.00223v1
|
https://arxiv.org/pdf/2208.00223v1.pdf
|
PolarMix: A General Data Augmentation Technique for LiDAR Point Clouds
|
LiDAR point clouds, which are usually scanned by rotating LiDAR sensors continuously, capture precise geometry of the surrounding environment and are crucial to many autonomous detection and navigation tasks. Though many 3D deep architectures have been developed, efficient collection and annotation of large amounts of point clouds remain one major challenge in the analytic and understanding of point cloud data. This paper presents PolarMix, a point cloud augmentation technique that is simple and generic but can mitigate the data constraint effectively across different perception tasks and scenarios. PolarMix enriches point cloud distributions and preserves point cloud fidelity via two cross-scan augmentation strategies that cut, edit, and mix point clouds along the scanning direction. The first is scene-level swapping which exchanges point cloud sectors of two LiDAR scans that are cut along the azimuth axis. The second is instance-level rotation and paste which crops point instances from one LiDAR scan, rotates them by multiple angles (to create multiple copies), and paste the rotated point instances into other scans. Extensive experiments show that PolarMix achieves superior performance consistently across different perception tasks and scenarios. In addition, it can work as plug-and-play for various 3D deep architectures and also performs well for unsupervised domain adaptation.
|
['Ling Shao', 'Shijian Lu', 'Kaiwen Cui', 'Dayan Guan', 'Jiaxing Huang', 'Aoran Xiao']
|
2022-07-30
| null | null | null | null |
['lidar-semantic-segmentation']
|
['computer-vision']
|
[ 2.35525981e-01 -4.31982368e-01 1.92697309e-02 -7.14465082e-01
-5.82806706e-01 -8.03955913e-01 6.79874659e-01 3.07351023e-01
-3.45676988e-01 2.09027693e-01 -4.43880051e-01 -2.99547791e-01
3.79113629e-02 -9.23614204e-01 -9.61347342e-01 -5.14450967e-01
1.27750158e-01 1.17311323e+00 5.60494661e-01 -2.45573267e-01
3.42241287e-01 1.15366960e+00 -1.69605088e+00 -1.01944268e-01
9.93824959e-01 8.87848914e-01 4.68625456e-01 3.68727744e-01
-6.40852511e-01 -3.00020665e-01 -3.90200615e-01 -2.87168950e-01
8.03308904e-01 5.30430496e-01 -2.28091180e-01 6.21516816e-02
1.03254199e+00 -2.79720157e-01 1.01097561e-02 1.07880056e+00
4.91223991e-01 3.33676413e-02 4.02891755e-01 -1.52330709e+00
-4.99094337e-01 -3.57576869e-02 -1.05469799e+00 -1.34212673e-01
-1.45067066e-01 1.53045088e-01 5.30828714e-01 -1.25530827e+00
5.20011783e-01 1.35569274e+00 8.78612459e-01 1.01182722e-01
-1.21728694e+00 -9.74327385e-01 1.36502221e-01 -8.84508640e-02
-1.50854695e+00 -6.09569810e-02 8.27714920e-01 -5.37446320e-01
9.28890347e-01 1.80760518e-01 7.10602582e-01 6.91622972e-01
-2.62141138e-01 2.19493404e-01 6.46152854e-01 -1.32239699e-01
2.24751681e-01 -1.58398584e-01 -1.43645406e-01 3.26090634e-01
4.41075474e-01 -7.01579824e-02 -5.96365929e-01 -1.19627245e-01
9.05357718e-01 3.14412981e-01 1.83772575e-02 -1.15348566e+00
-1.27490771e+00 6.40232205e-01 6.62782252e-01 -2.60215282e-01
-3.28212708e-01 1.90244555e-01 1.32577047e-01 9.49476734e-02
3.43158007e-01 3.48920643e-01 -4.82274503e-01 3.40621993e-02
-8.45173776e-01 6.58872008e-01 1.64704472e-01 1.48916292e+00
9.59704995e-01 -1.54323131e-01 3.88512582e-01 8.77073407e-01
2.16719404e-01 1.12435365e+00 -5.00757769e-02 -1.16736877e+00
7.77179599e-01 6.55875444e-01 1.45448163e-01 -8.96689117e-01
-3.61782551e-01 -3.58026147e-01 -8.25105250e-01 5.34256697e-01
1.72082465e-02 2.45157987e-01 -1.15121233e+00 1.30998194e+00
7.21305490e-01 1.49321914e-01 -3.06229174e-01 9.07206118e-01
5.86282313e-01 4.37369078e-01 -1.63071156e-01 4.10934776e-01
1.24004483e+00 -5.01763761e-01 -1.81487575e-01 -4.89758730e-01
2.75268465e-01 -8.83721709e-01 1.12578034e+00 2.62241930e-01
-9.46866632e-01 -7.38635123e-01 -1.17462587e+00 -5.87227941e-01
-5.02909184e-01 -9.02351439e-02 4.23932701e-01 1.80352181e-01
-7.52220690e-01 2.66488791e-01 -9.40296113e-01 -1.55007541e-01
5.83707929e-01 2.47445166e-01 -4.56385136e-01 -3.43444139e-01
-6.23423457e-01 6.76176071e-01 1.43011689e-01 -1.86273567e-02
-3.75048667e-01 -1.05519640e+00 -8.41181099e-01 -1.62025332e-01
2.70444363e-01 -8.89973283e-01 1.20368373e+00 -3.00023377e-01
-1.00224316e+00 9.76603508e-01 -3.29094976e-01 -4.06887501e-01
6.03943825e-01 -5.56360424e-01 -1.95076078e-01 -6.09382540e-02
5.00779450e-01 1.21669054e+00 8.29938352e-01 -1.59582961e+00
-8.03446531e-01 -7.94047117e-01 -3.52380216e-01 5.55406511e-01
3.30130726e-01 -4.81647432e-01 -8.50003183e-01 -3.68547529e-01
9.34500456e-01 -9.84956324e-01 -1.59698889e-01 4.74518180e-01
-2.92715520e-01 -6.91557303e-02 1.28621233e+00 -1.73949942e-01
3.99426073e-01 -2.39151049e+00 -1.50092021e-01 3.40359896e-01
1.03573255e-01 1.72570720e-01 -1.32296458e-01 3.45288157e-01
-1.18552610e-01 1.46209821e-01 -5.10349035e-01 -6.23375118e-01
-2.50098724e-02 4.77099210e-01 -5.79782724e-01 3.58143657e-01
3.68075073e-01 6.76298380e-01 -7.73628950e-01 -2.76566148e-01
5.61508477e-01 4.59955215e-01 -5.39389729e-01 -1.64674312e-01
-4.37176585e-01 3.87766719e-01 -2.00520873e-01 8.28218937e-01
1.52151442e+00 7.16049550e-03 -2.45666921e-01 -1.69167697e-01
-3.29892308e-01 3.43920320e-01 -1.35117042e+00 2.03520179e+00
-2.05994189e-01 7.25013614e-01 1.65644765e-01 -4.99888837e-01
1.20141530e+00 -3.03856462e-01 4.48366970e-01 -6.64237261e-01
-2.54206210e-01 2.41469890e-01 -3.11638862e-01 -2.00021356e-01
9.39516187e-01 -8.21509510e-02 2.45698951e-02 1.49079591e-01
-3.55233848e-01 -1.10715663e+00 -2.30452985e-01 6.16916120e-02
6.74956620e-01 9.98193026e-02 -6.52592257e-02 1.90003738e-01
9.99096259e-02 4.40861225e-01 4.64013755e-01 6.92173898e-01
4.94194739e-02 9.42940950e-01 5.33382595e-02 -4.97271389e-01
-1.27152467e+00 -1.50632572e+00 -3.52602303e-01 7.45245755e-01
4.99873489e-01 7.18911961e-02 -2.12317213e-01 -3.94435912e-01
6.32840514e-01 7.60550499e-01 -1.29201367e-01 -8.86039436e-03
-5.82292259e-01 -2.24393427e-01 3.71594757e-01 7.63140440e-01
6.39662325e-01 -5.96528590e-01 -6.50920153e-01 -7.28468075e-02
-1.95133537e-01 -1.15484118e+00 -1.32079124e-01 2.86049932e-01
-9.83685911e-01 -9.93846714e-01 -1.94947913e-01 -7.17704535e-01
5.62238336e-01 1.09773076e+00 1.12605107e+00 -8.00623223e-02
2.24719830e-02 9.14575756e-02 -2.66377836e-01 -8.15973580e-01
7.76028484e-02 1.03126764e-01 2.86179543e-01 -4.21030521e-01
6.68071210e-01 -7.70008147e-01 -3.46590102e-01 4.24483299e-01
-8.37533772e-01 -3.80402729e-02 5.68901360e-01 2.74018586e-01
1.25130701e+00 -2.81044506e-02 -9.02326629e-02 -6.81460798e-01
2.23781139e-01 -3.48442167e-01 -9.07872379e-01 -2.79656380e-01
-3.26528519e-01 -1.38163343e-01 -3.84552777e-02 -6.39965087e-02
-6.15008950e-01 3.06342602e-01 -5.48221134e-02 -1.01339293e+00
-3.22913766e-01 2.30424702e-01 -3.68077576e-01 -5.51896617e-02
6.77228391e-01 1.00571960e-01 1.34850934e-01 -7.29706287e-01
6.73306525e-01 5.45069337e-01 1.08708572e+00 -4.64863598e-01
1.32872808e+00 9.96570885e-01 1.85752630e-01 -1.11217427e+00
-4.44744051e-01 -7.85101056e-01 -1.05942130e+00 8.57542828e-02
7.90450454e-01 -1.09222007e+00 -3.97950172e-01 5.70975542e-01
-1.28525627e+00 -1.14184603e-01 -3.91056538e-01 3.00693184e-01
-1.90094694e-01 3.54433239e-01 1.50171995e-01 -5.05769432e-01
-1.38035089e-01 -1.07558465e+00 1.57570052e+00 1.19995706e-01
9.24255624e-02 -3.31048220e-01 1.26191765e-01 1.05974324e-01
-4.65204678e-02 2.60409653e-01 8.99113536e-01 -3.73665869e-01
-1.07793355e+00 -2.60694683e-01 -4.53758687e-01 2.16593221e-01
4.55277823e-02 2.50104159e-01 -8.39444041e-01 -1.72768012e-01
-1.87049806e-01 -1.25546558e-02 6.53244853e-01 3.23055029e-01
1.44440556e+00 2.81673104e-01 -6.40127480e-01 1.02070081e+00
1.25021124e+00 1.21725209e-01 5.28052330e-01 3.91777545e-01
1.00233746e+00 4.49822277e-01 7.68619120e-01 1.14844114e-01
6.43914223e-01 7.64289737e-01 1.04075408e+00 -3.05670779e-02
1.82949170e-01 -3.70950282e-01 -2.48865783e-01 5.64572930e-01
5.01851588e-02 -1.64307803e-02 -1.18055975e+00 4.60489959e-01
-1.54014957e+00 -7.86922753e-01 -5.75778127e-01 2.29826570e+00
3.82467568e-01 2.65269548e-01 -2.24492475e-01 -9.41698719e-03
6.82039618e-01 1.40279785e-01 -9.06934142e-01 -1.13062449e-01
-9.47180912e-02 4.60369378e-01 9.00351644e-01 4.23761994e-01
-9.98921752e-01 9.69868541e-01 5.53750324e+00 3.75313282e-01
-1.26047039e+00 -8.56358111e-02 -1.75703272e-01 -2.29351029e-01
-2.87728459e-01 -2.97277644e-02 -7.32161045e-01 2.97727883e-01
2.44855195e-01 3.86388183e-01 1.61768392e-01 1.14478385e+00
6.42134696e-02 -7.13116229e-02 -1.14470172e+00 1.21785426e+00
-2.61991173e-01 -1.35764074e+00 1.49127632e-01 3.50237966e-01
5.95403373e-01 6.67958498e-01 1.69330820e-01 4.93993163e-02
3.27393800e-01 -6.40064299e-01 9.66268659e-01 3.46509933e-01
9.80601609e-01 -7.15390265e-01 5.57708621e-01 5.33088565e-01
-1.19577360e+00 1.67552143e-01 -6.10818684e-01 5.59726097e-02
2.14320078e-01 7.28593230e-01 -1.14397109e+00 5.54144263e-01
1.12227952e+00 5.14834642e-01 -4.91479963e-01 1.16178870e+00
-1.29401386e-01 -1.01832161e-02 -7.87041366e-01 3.88235927e-01
7.97330737e-02 -4.16596353e-01 8.39441359e-01 8.65288615e-01
6.42443776e-01 -1.53238386e-01 9.90750715e-02 8.90318155e-01
-2.11550847e-01 -4.54222530e-01 -8.79167795e-01 3.25675398e-01
1.42736328e+00 1.01882219e+00 -4.79685307e-01 -2.89923906e-01
-1.55941531e-01 6.93208456e-01 2.00452909e-01 1.33253798e-01
-6.48240924e-01 -4.22178358e-01 1.27529156e+00 4.46315795e-01
4.61601794e-01 -8.98739159e-01 -8.37917209e-01 -7.08480358e-01
3.34501594e-01 -4.36555654e-01 -1.02057129e-01 -1.15984643e+00
-1.18048239e+00 2.15843320e-01 2.62921095e-01 -1.69603920e+00
-2.66690627e-02 -4.63466465e-01 -4.82397676e-01 1.04633188e+00
-1.65690303e+00 -1.20738101e+00 -9.91200149e-01 5.36854386e-01
5.86397648e-01 1.19303249e-01 3.60474586e-01 2.61531711e-01
-5.79616502e-02 2.37782046e-01 8.30050856e-02 -2.29820386e-02
7.22547114e-01 -1.30449200e+00 1.13718510e+00 8.18043053e-01
2.09877431e-01 6.97777569e-01 4.85619694e-01 -8.26335251e-01
-1.18422067e+00 -1.23788059e+00 6.01202548e-01 -6.61605120e-01
2.21008494e-01 -5.44561028e-01 -1.22576725e+00 5.25878251e-01
-2.22743109e-01 -1.61425937e-02 3.45851928e-01 -5.88060655e-02
-5.07323503e-01 -2.98119813e-01 -1.19002235e+00 4.03378934e-01
1.27980208e+00 -4.55812573e-01 -6.26273990e-01 2.99528003e-01
9.86178458e-01 -9.14269269e-01 -4.23458934e-01 6.17413580e-01
3.82570386e-01 -1.11651397e+00 1.35594666e+00 -3.37275565e-01
3.00243318e-01 -8.25892806e-01 -3.86865944e-01 -1.34685981e+00
-2.16811091e-01 -1.37955517e-01 1.02479771e-01 1.09064603e+00
7.89047182e-02 -6.95380986e-01 9.35542822e-01 3.89142752e-01
-6.59231663e-01 -3.02572608e-01 -1.10145009e+00 -6.60595894e-01
3.64808179e-02 -7.80936539e-01 1.20509911e+00 1.14324820e+00
-8.54798257e-01 1.24757379e-01 2.40464285e-01 7.62839854e-01
7.21571267e-01 2.24543050e-01 1.42469907e+00 -1.69714737e+00
3.58436942e-01 -5.51084638e-01 -4.33471024e-01 -1.28229201e+00
-3.08688283e-01 -7.69609809e-01 1.21104665e-01 -1.49391115e+00
-5.75802743e-01 -1.03167725e+00 3.80598545e-01 3.33264351e-01
8.40375572e-02 1.34170592e-01 2.69850999e-01 5.95393777e-01
1.01639256e-01 4.62449878e-01 9.59796786e-01 -1.85530588e-01
-3.87382358e-01 1.31394103e-01 -4.82721537e-01 8.81843746e-01
8.08580816e-01 -4.09009635e-01 -3.65621626e-01 -1.12654006e+00
2.56629586e-01 -4.21599716e-01 5.80218554e-01 -1.25562775e+00
2.14699581e-01 -2.89904594e-01 5.78322291e-01 -1.62242699e+00
7.12346852e-01 -1.12298453e+00 2.25088343e-01 1.00283861e-01
2.25589424e-01 5.37016392e-01 3.75699908e-01 6.64227188e-01
-2.65412237e-02 3.99737246e-02 8.70649517e-01 -2.38166004e-01
-7.70752192e-01 4.90004420e-01 1.66862577e-01 -1.44519404e-01
9.18462515e-01 -6.66468978e-01 -2.97240287e-01 9.03169736e-02
-3.61102402e-01 5.29353499e-01 9.80177760e-01 7.20770597e-01
7.47759163e-01 -1.40069199e+00 -4.31892037e-01 5.91741621e-01
3.58961791e-01 1.32193995e+00 9.53731164e-02 4.59693253e-01
-9.87282634e-01 2.28423312e-01 -1.74798667e-01 -1.36101592e+00
-1.21400845e+00 2.87546366e-01 3.04095924e-01 3.14566225e-01
-7.45912611e-01 1.11634648e+00 6.60168007e-02 -9.04385090e-01
6.71598911e-02 -6.75628543e-01 2.96079040e-01 -6.47092462e-02
2.21788779e-01 2.52425641e-01 4.57021773e-01 -6.26334548e-01
-4.78189081e-01 9.23290014e-01 7.56311789e-02 -1.72966585e-01
1.27671385e+00 -6.11555949e-02 -9.75127816e-02 3.55797499e-01
8.64114106e-01 1.05381630e-01 -1.35982835e+00 -3.50344896e-01
-1.92820385e-01 -9.35308754e-01 -6.38246462e-02 -5.61368942e-01
-8.69304001e-01 1.25881016e+00 7.66192913e-01 5.38597703e-02
7.47067451e-01 2.03530584e-02 6.07567132e-01 5.26275277e-01
5.13918221e-01 -9.84290183e-01 -1.88451678e-01 6.14562929e-01
9.02001739e-01 -1.24484444e+00 1.60082176e-01 -5.50256550e-01
-6.29437447e-01 8.05581748e-01 7.85992801e-01 -1.69915576e-02
5.27630508e-01 2.21180961e-01 8.61119330e-02 -4.43935543e-01
-1.96696460e-01 -9.99503955e-02 1.46368355e-01 1.20899618e+00
-2.32328206e-01 1.61547929e-01 4.77086514e-01 -1.16571551e-03
-5.72371066e-01 -2.76135027e-01 1.53411508e-01 1.07294691e+00
-7.41762519e-01 -9.54204559e-01 -6.85053289e-01 4.19297010e-01
5.44620633e-01 2.07717922e-02 -5.25270045e-01 9.34188008e-01
4.94808316e-01 4.60722178e-01 7.19254315e-01 -4.64033335e-01
6.21491492e-01 -2.19487939e-02 2.97033399e-01 -7.34899938e-01
-7.51486197e-02 -1.14838280e-01 -3.28970641e-01 -4.79106784e-01
-2.10476428e-01 -8.48789573e-01 -1.50751209e+00 -3.50350589e-01
-1.56440139e-01 -1.31580770e-01 1.23094022e+00 7.22458839e-01
7.66635299e-01 1.97354376e-01 4.86804247e-01 -1.35132980e+00
-4.79889423e-01 -8.58687162e-01 -3.60232860e-01 2.00626373e-01
4.68717456e-01 -1.08773112e+00 -1.33525416e-01 4.14342284e-02]
|
[7.970095157623291, -2.9776296615600586]
|
6068e0be-5164-4b47-aa96-d74560faf841
|
cp3-unifying-point-cloud-completion-by
|
2207.05359
| null |
https://arxiv.org/abs/2207.05359v2
|
https://arxiv.org/pdf/2207.05359v2.pdf
|
CP3: Unifying Point Cloud Completion by Pretrain-Prompt-Predict Paradigm
|
Point cloud completion aims to predict complete shape from its partial observation. Current approaches mainly consist of generation and refinement stages in a coarse-to-fine style. However, the generation stage often lacks robustness to tackle different incomplete variations, while the refinement stage blindly recovers point clouds without the semantic awareness. To tackle these challenges, we unify point cloud Completion by a generic Pretrain-Prompt-Predict paradigm, namely CP3. Inspired by prompting approaches from NLP, we creatively reinterpret point cloud generation and refinement as the prompting and predicting stages, respectively. Then, we introduce a concise self-supervised pretraining stage before prompting. It can effectively increase robustness of point cloud generation, by an Incompletion-Of-Incompletion (IOI) pretext task. Moreover, we develop a novel Semantic Conditional Refinement (SCR) network at the predicting stage. It can discriminatively modulate multi-scale refinement with the guidance of semantics. Finally, extensive experiments demonstrate that our CP3 outperforms the state-of-the-art methods with a large margin.
|
['Tong He', 'Yu Qiao', 'Yihao Liu', 'Yali Wang', 'Mingye Xu']
|
2022-07-12
| null | null | null | null |
['point-cloud-completion', 'point-cloud-generation']
|
['computer-vision', 'computer-vision']
|
[ 3.98850322e-01 2.22414643e-01 -6.63758293e-02 -4.19920653e-01
-1.08389080e+00 -5.84795237e-01 7.91690528e-01 -7.97730535e-02
-7.46554807e-02 2.80162454e-01 2.71527059e-02 -8.96074995e-02
8.16620737e-02 -8.63761604e-01 -1.16353738e+00 -3.93838227e-01
4.97352749e-01 8.39446604e-01 3.60985726e-01 -1.98347494e-01
3.59816730e-01 4.72636729e-01 -1.56194556e+00 3.12589794e-01
1.23818505e+00 8.48005831e-01 6.39730036e-01 2.76820391e-01
-2.60615259e-01 4.10632968e-01 -2.35057592e-01 -2.89265901e-01
2.85740256e-01 1.89730719e-01 -6.83148980e-01 3.22064102e-01
4.72701877e-01 -2.07107633e-01 -1.45935237e-01 1.06672859e+00
1.80014998e-01 -2.01173127e-02 6.71807230e-01 -1.36599302e+00
-9.02172148e-01 2.95393825e-01 -6.52161062e-01 -4.28633571e-01
4.44940388e-01 2.42752358e-01 1.04857576e+00 -1.32722962e+00
5.27078867e-01 1.27678466e+00 6.04983151e-01 5.70411444e-01
-1.15698862e+00 -9.26099300e-01 7.56949246e-01 -5.83698153e-02
-1.54438543e+00 -1.66342601e-01 9.86618340e-01 -5.16929626e-01
6.41486764e-01 1.28460810e-01 4.73098814e-01 9.97120976e-01
-5.34609914e-01 1.09022105e+00 7.08450019e-01 3.12350094e-02
5.22439852e-02 -1.91722870e-01 -2.28568256e-01 7.08933711e-01
-2.05244236e-02 3.51861030e-01 -3.62053066e-01 -1.45751208e-01
1.05111706e+00 3.10613155e-01 -2.88751841e-01 -4.76773471e-01
-1.49581122e+00 6.24761045e-01 7.83618331e-01 -1.84451446e-01
-3.89544874e-01 4.70777266e-02 -1.53053299e-01 3.34958360e-02
6.14699841e-01 4.56059873e-01 -7.80992687e-01 3.01600635e-01
-1.07005179e+00 4.86838311e-01 5.01064420e-01 1.56387269e+00
1.16441560e+00 -2.55890042e-01 -1.79783329e-01 4.88542080e-01
4.81750429e-01 7.21363902e-01 6.29317835e-02 -6.16265953e-01
7.90645480e-01 8.94380689e-01 1.92440838e-01 -6.48101389e-01
-3.24702799e-01 -4.84293967e-01 -9.29474711e-01 1.81568459e-01
-5.42872995e-02 8.54890142e-03 -1.27124560e+00 1.52167070e+00
6.30924463e-01 8.26087177e-01 -5.39247356e-02 1.01419473e+00
8.82793903e-01 6.27392888e-01 2.10943028e-01 2.90879458e-02
1.25303710e+00 -1.25460899e+00 -2.40662113e-01 -2.58853376e-01
2.96579480e-01 -7.03889430e-01 1.20003712e+00 3.95408243e-01
-9.26602304e-01 -8.03587675e-01 -8.56920660e-01 -3.53613257e-01
-1.11469805e-01 2.30571762e-01 5.70062459e-01 -4.25111279e-02
-8.06962430e-01 6.79286838e-01 -1.02875590e+00 3.22806230e-03
6.71811640e-01 2.88397312e-01 -4.11722064e-01 -1.98790982e-01
-6.79282069e-01 4.50167447e-01 3.49116057e-01 9.93505418e-02
-8.26737165e-01 -1.21904147e+00 -8.03764880e-01 -5.90509437e-02
5.61253846e-01 -1.26914012e+00 1.20806694e+00 -7.44382262e-01
-1.25904524e+00 7.87026644e-01 -3.31796378e-01 -1.56956002e-01
5.67133307e-01 -5.92572927e-01 -1.27020001e-01 4.64468449e-02
4.69685942e-01 1.12134421e+00 1.08715999e+00 -1.75854170e+00
-1.04246068e+00 -3.39974523e-01 2.21912693e-02 3.66262734e-01
3.08169723e-01 -3.63991022e-01 -9.77160931e-01 -7.83697128e-01
5.39176762e-01 -8.25748324e-01 -4.43701267e-01 1.63904533e-01
-5.89651465e-01 -4.87551630e-01 8.22600782e-01 -3.08729619e-01
7.46180534e-01 -2.08157849e+00 3.17923814e-01 2.60791957e-01
3.56053352e-01 -4.75413986e-02 -3.68972600e-01 2.29426369e-01
-2.16911390e-01 6.70516267e-02 -5.57985067e-01 -7.59938002e-01
1.75642505e-01 2.87796468e-01 -7.81901419e-01 2.32307702e-01
6.60201967e-01 1.03536737e+00 -1.03980076e+00 -3.79155189e-01
3.82040799e-01 4.46356684e-01 -8.30747128e-01 4.31304008e-01
-8.14024508e-01 8.53950858e-01 -7.25719631e-01 8.55571806e-01
9.69782710e-01 -6.68613076e-01 -4.13200021e-01 -3.74882281e-01
-2.03222260e-01 5.25319800e-02 -1.03554296e+00 2.38671517e+00
-3.72552305e-01 -1.25147909e-01 -4.80231196e-02 -7.03248024e-01
9.47503865e-01 1.53020233e-01 4.77534920e-01 -3.94405186e-01
-3.60192269e-01 2.48692885e-01 -6.28502309e-01 -2.55616993e-01
6.37379766e-01 -2.60254145e-01 2.03826074e-02 7.08722100e-02
8.16499162e-03 -4.73330855e-01 -2.77406245e-01 2.23757997e-01
8.73007119e-01 6.67674780e-01 -3.97578701e-02 1.77081317e-01
5.86429238e-01 1.10434636e-01 7.15846062e-01 5.12634575e-01
1.14530459e-01 1.33605397e+00 1.99218944e-01 -3.08131516e-01
-8.99217367e-01 -1.06129444e+00 8.17750841e-02 8.57212782e-01
7.73938119e-01 -4.32729244e-01 -4.73195344e-01 -1.02603900e+00
1.72413081e-01 7.21661210e-01 -3.69730413e-01 2.00027768e-02
-5.49274683e-01 -4.09062773e-01 4.10577394e-02 6.85815394e-01
3.49780411e-01 -1.14431953e+00 -2.51986831e-03 2.38137320e-02
-2.88948178e-01 -1.31241584e+00 -4.88250375e-01 -1.27345845e-01
-9.72449899e-01 -1.23853290e+00 -7.04698980e-01 -8.16606522e-01
9.21017706e-01 4.53223348e-01 1.30154693e+00 3.52381170e-01
2.98429370e-01 2.51070201e-01 -5.84876657e-01 -2.83913434e-01
-6.38273433e-02 2.99642444e-01 1.91340651e-02 -1.76428575e-02
3.06779534e-01 -8.17032397e-01 -7.41444230e-01 2.56061047e-01
-9.45442677e-01 5.73686540e-01 9.55576122e-01 4.86428231e-01
1.16170239e+00 -1.14273481e-01 2.98279792e-01 -1.16899478e+00
4.49749045e-02 -5.43403924e-01 -6.56475365e-01 1.90039918e-01
-6.16949260e-01 2.25941569e-01 5.89320302e-01 -2.35524222e-01
-9.65197623e-01 4.03266072e-01 -5.51829278e-01 -1.05834854e+00
-4.74942207e-01 2.93419480e-01 -4.72212553e-01 -1.13546334e-01
3.70165050e-01 3.61488342e-01 -4.29874808e-01 -8.72227013e-01
6.50648475e-01 1.41007870e-01 9.16852295e-01 -6.88850462e-01
1.52590060e+00 7.56456077e-01 -9.80333239e-02 -3.37039351e-01
-1.18436706e+00 -6.40475512e-01 -8.36594045e-01 1.08961627e-01
7.90151358e-01 -1.31890452e+00 -5.74117422e-01 3.59421253e-01
-1.47827458e+00 -2.33483940e-01 -3.61678153e-01 5.50124049e-02
-5.84272385e-01 4.22056437e-01 -2.03540266e-01 -4.07998681e-01
-3.63088906e-01 -9.85225022e-01 1.86742890e+00 1.44315898e-01
2.26086929e-01 -6.26106620e-01 -8.87092948e-02 3.06571752e-01
-1.08915329e-01 2.41502866e-01 5.92899263e-01 -5.82321584e-01
-1.18392241e+00 4.19468898e-03 -5.57800710e-01 1.22908443e-01
1.66258737e-02 -2.07001865e-01 -8.80817533e-01 -2.57993579e-01
-8.55465606e-02 -1.81974500e-01 9.96310472e-01 -4.07593958e-02
1.47518063e+00 -3.66633713e-01 -4.20374066e-01 1.12657893e+00
1.43473351e+00 -5.12721241e-01 6.50557756e-01 1.89113304e-01
1.29302108e+00 3.73424381e-01 8.64247859e-01 4.33178723e-01
8.17964375e-01 4.84869540e-01 8.52968693e-01 -1.62959367e-01
-1.89921856e-01 -8.70321989e-01 3.75855528e-02 7.23516047e-01
-6.27026483e-02 2.70303097e-02 -9.06209528e-01 3.57813507e-01
-1.97434771e+00 -5.98145068e-01 -2.71445304e-01 1.93262041e+00
7.31727600e-01 7.59341046e-02 -1.84084758e-01 -1.77618295e-01
6.17939591e-01 2.23602265e-01 -6.37014687e-01 6.76853120e-01
6.36916533e-02 1.98782131e-01 4.13490236e-01 5.80544353e-01
-1.19614339e+00 1.43654096e+00 5.15621090e+00 1.06374252e+00
-9.50471163e-01 8.64739046e-02 2.79074371e-01 2.43076146e-01
-6.54560268e-01 3.09776932e-01 -8.14043581e-01 5.27744234e-01
2.14642733e-01 3.09752971e-01 3.06474745e-01 1.05252516e+00
1.48068871e-02 4.20955598e-01 -1.26743591e+00 1.18691587e+00
1.01586636e-02 -1.48258317e+00 4.00843590e-01 -2.63223171e-01
8.88025403e-01 2.85419285e-01 -7.01226294e-02 4.87212479e-01
2.63498783e-01 -8.33352029e-01 9.61992383e-01 8.59131694e-01
9.68803525e-01 -6.07060730e-01 4.04227257e-01 6.45700276e-01
-1.54906666e+00 1.84391364e-01 -4.65910196e-01 8.85447413e-02
3.59408408e-01 6.20923340e-01 -4.51340795e-01 1.09752154e+00
4.58323687e-01 1.04167128e+00 -6.10597968e-01 1.05537260e+00
-6.58862829e-01 2.21809715e-01 -3.36599678e-01 5.05512774e-01
1.01585448e-01 -3.28647554e-01 6.93242788e-01 8.21835160e-01
3.76036346e-01 4.41852182e-01 4.96985823e-01 1.23890424e+00
-9.68687162e-02 -2.10854664e-01 -2.03521535e-01 3.34169358e-01
5.84564149e-01 1.36770260e+00 -4.76195663e-01 -3.42384368e-01
-4.54133898e-01 1.13679266e+00 5.35020471e-01 4.16938931e-01
-6.99070513e-01 -1.37477845e-01 6.16655707e-01 1.52470902e-01
5.16391873e-01 -1.77983105e-01 -5.77377558e-01 -1.47544622e+00
2.19530478e-01 -6.18510544e-01 1.02566667e-01 -1.16328406e+00
-1.57011712e+00 5.31996727e-01 -1.12211347e-01 -1.58341837e+00
1.05377898e-01 -4.14562047e-01 -7.26308286e-01 9.05501783e-01
-2.02609754e+00 -1.77986944e+00 -5.95725477e-01 6.25089467e-01
8.74231577e-01 8.73636454e-02 4.63920742e-01 1.96497664e-01
-2.58338779e-01 2.98495650e-01 -3.96033883e-01 -5.24022952e-02
5.62066734e-01 -1.37735415e+00 7.48843849e-01 9.27142084e-01
1.76072255e-01 4.53250229e-01 4.63864863e-01 -9.26945090e-01
-1.22068620e+00 -1.68841887e+00 8.16243172e-01 -7.25225031e-01
4.09488231e-01 -3.94937932e-01 -1.06279933e+00 6.59838736e-01
-3.67966294e-01 1.18989259e-01 -4.33036825e-03 2.11611584e-01
-4.71788734e-01 9.02916268e-02 -8.59471321e-01 5.31247199e-01
1.32039857e+00 -5.33974648e-01 -6.93096340e-01 6.34183943e-01
1.32607687e+00 -7.74530649e-01 -5.74753881e-01 8.13009262e-01
5.99700771e-02 -7.61776149e-01 1.20792294e+00 -3.60820323e-01
6.11200392e-01 -6.43841147e-01 -7.19252229e-02 -1.02100980e+00
-4.69040334e-01 -7.06679761e-01 -3.63989115e-01 1.29583371e+00
4.16679382e-01 -2.35743806e-01 1.18957293e+00 4.40876335e-01
-6.44040644e-01 -7.07689762e-01 -6.22228622e-01 -5.67398608e-01
1.36701260e-02 -7.12132871e-01 1.15512717e+00 8.99257958e-01
-4.88688678e-01 5.16356766e-01 -2.15189591e-01 8.00227761e-01
6.49965048e-01 5.23920596e-01 1.08764672e+00 -1.51405680e+00
-2.43721142e-01 -1.93102211e-01 -9.42039341e-02 -1.74144852e+00
1.16394788e-01 -9.65918303e-01 2.60223836e-01 -1.66963136e+00
1.08875796e-01 -8.56970668e-01 -2.10246649e-02 6.24019504e-01
-6.07256830e-01 4.27490547e-02 2.78260976e-01 6.81198537e-01
-8.11881483e-01 1.05466008e+00 1.49468744e+00 -1.97167564e-02
-2.61970639e-01 3.17718863e-01 -8.24565947e-01 8.85348380e-01
4.18754041e-01 -4.66584533e-01 -4.33951139e-01 -6.39089882e-01
3.59067380e-01 5.91948815e-02 6.57670438e-01 -1.07237446e+00
3.56383443e-01 -2.12946549e-01 2.38189474e-01 -1.22573578e+00
3.62347811e-01 -9.73732650e-01 6.98764101e-02 -1.41795829e-01
-7.55543858e-02 2.74605700e-03 -8.11312273e-02 1.03283715e+00
-1.21126443e-01 6.29802123e-02 3.86203229e-01 -1.41379058e-01
-7.97599852e-01 1.10346818e+00 5.02719223e-01 2.69769207e-02
7.04279721e-01 -1.06623068e-01 -1.59821719e-01 1.06580749e-01
-6.97047830e-01 6.54674709e-01 7.18417883e-01 6.63100362e-01
8.15680087e-01 -1.41836607e+00 -7.21314728e-01 4.41337794e-01
4.39714760e-01 1.09113908e+00 3.54774147e-01 7.06032634e-01
-3.87336135e-01 1.55215919e-01 3.57030302e-01 -9.13608372e-01
-8.64678562e-01 7.66472578e-01 -1.64205637e-02 -2.22449273e-01
-1.01450062e+00 9.95645285e-01 7.61697829e-01 -8.75428617e-01
1.29887193e-01 -5.09118617e-01 -8.26476291e-02 -3.84700239e-01
2.61466324e-01 -1.81733668e-02 3.36948112e-02 -5.47255337e-01
-2.03929439e-01 8.42122138e-01 -4.87157218e-02 -5.89801744e-02
1.23439884e+00 -1.34570166e-01 -1.01391897e-02 1.40989631e-01
8.99461508e-01 -8.29672590e-02 -1.70833337e+00 -4.85998511e-01
-9.18387845e-02 -5.37687182e-01 -1.07480742e-01 -7.32373714e-01
-9.64889467e-01 7.74904847e-01 6.19874001e-02 -1.22454934e-01
1.04692721e+00 2.43460193e-01 8.85695457e-01 3.10461760e-01
3.73505026e-01 -7.94030964e-01 -8.03904608e-02 5.34946680e-01
1.14927733e+00 -1.34983897e+00 -7.98582435e-02 -9.00633931e-01
-7.46394277e-01 8.76950741e-01 8.56981635e-01 -3.21569800e-01
7.42982864e-01 -4.00851890e-02 -1.63189396e-01 -3.67477685e-01
-7.03684688e-01 -3.57724041e-01 7.03153014e-01 7.32716501e-01
-7.66366720e-02 5.54683991e-02 2.12897182e-01 8.97856295e-01
-3.26866835e-01 4.25473861e-02 -4.76468578e-02 6.04002237e-01
-4.23326761e-01 -1.02845728e+00 -3.26342940e-01 3.34287703e-01
1.79839820e-01 -2.57535517e-01 -2.91898459e-01 7.22550571e-01
2.87039965e-01 6.97886884e-01 1.37201980e-01 -6.19840026e-01
5.02471149e-01 -2.53164709e-01 1.70054093e-01 -9.98556137e-01
-4.32852209e-01 9.74025354e-02 -3.48988593e-01 -8.09570789e-01
-3.54303420e-01 -6.36814952e-01 -1.34847569e+00 -1.23610295e-01
-4.31432903e-01 5.05348668e-02 3.95103812e-01 1.19408607e+00
6.15462005e-01 5.43998301e-01 7.27519453e-01 -1.26518142e+00
-4.67844456e-01 -8.74911785e-01 -2.99801528e-01 6.31006658e-01
4.23067421e-01 -6.86385870e-01 -2.47257829e-01 4.91001233e-02]
|
[8.243673324584961, -3.4876935482025146]
|
78a0f5c4-53e2-4012-b88e-fee96efad54c
|
multi-scale-2d-representation-learning-for
|
2111.02741
| null |
https://arxiv.org/abs/2111.02741v1
|
https://arxiv.org/pdf/2111.02741v1.pdf
|
Multi-scale 2D Representation Learning for weakly-supervised moment retrieval
|
Video moment retrieval aims to search the moment most relevant to a given language query. However, most existing methods in this community often require temporal boundary annotations which are expensive and time-consuming to label. Hence weakly supervised methods have been put forward recently by only using coarse video-level label. Despite effectiveness, these methods usually process moment candidates independently, while ignoring a critical issue that the natural temporal dependencies between candidates in different temporal scales. To cope with this issue, we propose a Multi-scale 2D Representation Learning method for weakly supervised video moment retrieval. Specifically, we first construct a two-dimensional map for each temporal scale to capture the temporal dependencies between candidates. Two dimensions in this map indicate the start and end time points of these candidates. Then, we select top-K candidates from each scale-varied map with a learnable convolutional neural network. With a newly designed Moments Evaluation Module, we obtain the alignment scores of the selected candidates. At last, the similarity between captions and language query is served as supervision for further training the candidates' selector. Experiments on two benchmark datasets Charades-STA and ActivityNet Captions demonstrate that our approach achieves superior performance to state-of-the-art results.
|
['Wensheng Zhang', 'Zhizhong Zhang', 'Yongqiang Tang', 'Rui Wu', 'Ding Li']
|
2021-11-04
| null | null | null | null |
['moment-retrieval']
|
['computer-vision']
|
[-1.53492436e-01 -4.97059643e-01 -6.73709571e-01 -3.85157138e-01
-1.11453187e+00 -5.89952826e-01 7.63123751e-01 2.90144205e-01
-5.34686208e-01 3.11221510e-01 3.91774356e-01 3.27477872e-01
-3.96868177e-02 -4.12900388e-01 -7.24550426e-01 -5.82164347e-01
-4.01941508e-01 3.11492264e-01 7.07858443e-01 8.82129967e-02
3.02649379e-01 3.37374210e-01 -1.28804529e+00 3.23983997e-01
4.88351703e-01 1.29123175e+00 2.66794503e-01 2.66725302e-01
-1.83412895e-01 8.11780870e-01 -3.78896683e-01 -1.15660913e-01
4.85869870e-02 -4.02818650e-01 -7.13887334e-01 1.66128054e-01
5.04286468e-01 -3.23485941e-01 -7.97109246e-01 9.61156726e-01
4.34703887e-01 3.90535802e-01 4.79039371e-01 -1.05910671e+00
-3.13166201e-01 4.23361868e-01 -7.11665213e-01 6.46027505e-01
6.75211072e-01 -1.39141798e-01 1.14536798e+00 -1.04649043e+00
8.92643690e-01 9.89648044e-01 3.56939077e-01 4.16555315e-01
-8.26489985e-01 -3.50656301e-01 4.17174071e-01 5.48261940e-01
-1.54843235e+00 -3.32229167e-01 1.16580129e+00 -3.98211867e-01
6.09585762e-01 6.60750493e-02 6.16788089e-01 1.05408812e+00
-7.09111243e-02 1.20732677e+00 6.64155722e-01 -2.53920462e-02
8.23298097e-02 -1.71886757e-01 -1.67898640e-01 7.61100948e-01
-5.70303142e-01 -4.29591596e-01 -7.28009343e-01 -1.55772284e-01
7.73083866e-01 7.93752894e-02 -1.93343446e-01 -5.60793161e-01
-1.67887998e+00 6.05338931e-01 4.72092330e-01 5.45563281e-01
-4.70387369e-01 2.99790859e-01 6.19269192e-01 1.53291985e-01
6.59502387e-01 2.77605131e-02 -2.48932168e-01 -2.15748176e-01
-1.22617888e+00 6.58444911e-02 4.27334011e-01 7.30203331e-01
6.71631873e-01 -4.28731173e-01 -5.33496499e-01 7.96959519e-01
2.62347639e-01 1.52435526e-01 5.42942762e-01 -7.23636568e-01
7.02697515e-01 5.58829248e-01 3.03073704e-01 -1.19585228e+00
-3.36161971e-01 -3.97248566e-01 -4.42781270e-01 -5.39470196e-01
3.23672682e-01 2.77265936e-01 -8.42779577e-01 1.61870289e+00
5.07247865e-01 6.66519582e-01 -1.95136622e-01 1.30152822e+00
6.44618630e-01 1.01533306e+00 4.75151166e-02 -3.67706120e-01
1.35637271e+00 -1.13699424e+00 -5.11012495e-01 -2.10267697e-02
5.73331356e-01 -7.99607337e-01 1.05782020e+00 2.41651703e-02
-9.87637997e-01 -3.99308920e-01 -7.71370113e-01 -8.04319382e-02
-1.95883453e-01 4.50144023e-01 6.04839146e-01 -9.13086310e-02
-8.68713975e-01 4.66921270e-01 -9.41752434e-01 -3.35753500e-01
1.41308278e-01 1.79193780e-01 -3.28013748e-01 2.41126791e-02
-1.34950662e+00 6.40117705e-01 3.34079057e-01 8.94058868e-02
-9.95038569e-01 -3.10837448e-01 -8.81460369e-01 -1.16201669e-01
4.68941003e-01 -2.16535226e-01 1.27605808e+00 -9.66174126e-01
-1.32154739e+00 9.26395118e-01 -3.46986592e-01 -4.97622281e-01
6.25339925e-01 -2.66972482e-01 -4.94184643e-01 7.16321111e-01
4.32307363e-01 8.56905520e-01 8.47801208e-01 -8.74045074e-01
-8.42323780e-01 2.40758657e-02 2.95327395e-01 4.57012981e-01
-4.23615724e-01 4.28554177e-01 -1.37071586e+00 -7.30131328e-01
3.67319643e-01 -9.66786027e-01 -1.10329002e-01 5.69432527e-02
-9.53085199e-02 -6.98203027e-01 9.56995130e-01 -5.43915153e-01
1.43975127e+00 -2.18137765e+00 2.38571033e-01 -9.59657691e-03
1.05099343e-02 -2.87394058e-02 -2.58981049e-01 2.91254252e-01
1.13440841e-01 -2.70063668e-01 1.71347797e-01 -3.72744739e-01
-7.96947628e-02 -8.26884434e-02 -3.42870653e-01 7.05366671e-01
1.34644404e-01 8.54145586e-01 -1.26374996e+00 -8.44241679e-01
3.24173033e-01 4.24486816e-01 -2.92727441e-01 2.79764295e-01
-4.06462610e-01 5.16876459e-01 -7.31807768e-01 7.68016756e-01
1.78730264e-01 -4.83597040e-01 -1.73981756e-01 -3.43752801e-01
-2.87981689e-01 3.61734152e-01 -9.83396232e-01 2.19143677e+00
-2.53333956e-01 6.55213416e-01 -2.93606311e-01 -9.76227283e-01
7.08506584e-01 4.54256326e-01 9.95716453e-01 -6.87564611e-01
-2.96305884e-02 2.34050661e-01 -4.12650675e-01 -7.48935401e-01
4.39946651e-01 2.93461651e-01 -3.01719218e-01 2.74679542e-01
-5.23002185e-02 1.83488578e-01 4.98053998e-01 3.52391213e-01
1.04828990e+00 4.17098135e-01 -1.21003062e-01 1.46174923e-01
7.84343898e-01 -2.11998671e-01 6.59326255e-01 4.72067386e-01
-4.57712561e-01 8.28746736e-01 5.69794059e-01 -7.15642393e-01
-9.38338935e-01 -8.58267784e-01 2.24601880e-01 1.31108415e+00
4.87989455e-01 -5.83061576e-01 -5.15453935e-01 -9.34762836e-01
-4.16541547e-01 1.49811432e-01 -6.26766801e-01 -3.68711799e-02
-8.73813808e-01 -3.93762320e-01 3.08989853e-01 3.45309526e-01
4.89147902e-01 -1.02430010e+00 -6.45436764e-01 2.39527598e-01
-7.00387001e-01 -1.40937757e+00 -9.83668149e-01 -1.18888661e-01
-7.40188360e-01 -8.84360433e-01 -1.14911187e+00 -1.10069895e+00
6.86263621e-01 4.29893404e-01 9.46059346e-01 -8.82845223e-02
9.54386368e-02 3.73054326e-01 -4.90232259e-01 2.48552233e-01
1.78060427e-01 1.98263377e-01 1.06000774e-01 3.51605773e-01
3.58410627e-01 -3.57208431e-01 -9.88557398e-01 5.26493609e-01
-8.69847775e-01 8.61591399e-02 5.18487692e-01 5.27016044e-01
9.09091711e-01 -2.78655827e-01 4.31807846e-01 -1.94700539e-01
3.79810274e-01 -5.20660639e-01 -6.01820886e-01 3.78564477e-01
-6.55869544e-02 1.21110447e-01 4.94514316e-01 -8.00918341e-01
-6.56752646e-01 3.92482340e-01 2.06142291e-01 -8.08188915e-01
-1.50926039e-02 6.84331000e-01 5.30411042e-02 1.39629051e-01
2.91723996e-01 4.49977338e-01 -5.33184826e-01 -3.23666066e-01
2.89516836e-01 3.83301079e-01 6.09224856e-01 -5.71546495e-01
7.45111585e-01 7.05121040e-01 -2.31827840e-01 -5.98926604e-01
-1.06442440e+00 -8.54742467e-01 -6.93054497e-01 -5.39363325e-01
8.56975913e-01 -1.10516500e+00 -2.87588209e-01 1.76552728e-01
-1.21400380e+00 -9.43499207e-02 4.92150672e-02 8.34391177e-01
-6.10289037e-01 4.86412615e-01 -6.45832121e-01 -4.67448682e-01
-2.13973537e-01 -1.10105920e+00 1.48027778e+00 2.72083193e-01
-1.09492816e-01 -7.18610823e-01 2.65130609e-01 2.37871319e-01
8.07475597e-02 2.20259488e-01 5.50041080e-01 -5.48728347e-01
-7.76737690e-01 -5.87683260e-01 -2.96196818e-01 -1.36761710e-01
-2.46927496e-02 -6.32476509e-02 -6.49834037e-01 -1.60574809e-01
-1.73313901e-01 -3.51492077e-01 8.82865131e-01 4.44059253e-01
1.16248739e+00 -1.77309096e-01 -5.30311644e-01 5.59357285e-01
1.06233156e+00 6.74727783e-02 2.87875831e-01 4.97135252e-01
5.53422272e-01 4.78796601e-01 9.90587473e-01 4.97023016e-01
3.69410098e-01 9.02082503e-01 2.88927019e-01 1.72155440e-01
1.42424345e-01 -3.62243652e-01 5.28011739e-01 9.06433105e-01
-6.74834475e-02 -1.90534562e-01 -7.51867712e-01 7.53341317e-01
-2.14924955e+00 -1.08640838e+00 3.35662216e-01 2.16066647e+00
7.35115528e-01 3.42540920e-01 1.48068458e-01 -1.52575538e-01
8.20607901e-01 6.93054795e-01 -4.92689490e-01 3.73290420e-01
-9.38512757e-02 -4.99548554e-01 1.62241399e-01 1.39946923e-01
-1.58183348e+00 1.07452190e+00 5.07763577e+00 8.95868421e-01
-1.29468822e+00 3.33163142e-02 5.51798582e-01 -4.32894021e-01
-2.45617293e-02 1.18115045e-01 -6.32772863e-01 5.42409539e-01
7.82923281e-01 -5.09807803e-02 2.43819639e-01 7.83849061e-01
5.77649295e-01 -8.11080635e-02 -1.19980013e+00 1.29271007e+00
2.58266151e-01 -1.24525023e+00 -6.46288842e-02 -3.32612008e-01
8.18905473e-01 3.11189234e-01 -6.48074877e-03 3.63334686e-01
-4.40086812e-01 -4.55543518e-01 8.40701759e-01 6.41568363e-01
5.66252053e-01 -7.63833106e-01 4.71532017e-01 3.22843432e-01
-1.70879877e+00 1.21661715e-01 -3.53762388e-01 3.08795452e-01
3.08349729e-01 4.73873556e-01 -5.93642712e-01 2.82818645e-01
6.30032420e-01 9.19465959e-01 -6.49112284e-01 1.41024792e+00
-2.33089298e-01 3.67006034e-01 -4.39299077e-01 -3.86367142e-02
6.27573371e-01 -6.95971698e-02 6.24506831e-01 1.30746675e+00
4.82863069e-01 -2.99050398e-02 4.57808018e-01 2.41259843e-01
-2.57434785e-01 2.45107159e-01 -4.46827501e-01 -1.16838753e-01
2.89759099e-01 1.36460078e+00 -1.21440053e+00 -4.66255188e-01
-5.27338922e-01 1.33450043e+00 3.49299580e-01 3.88630331e-01
-1.16784918e+00 -2.66062677e-01 2.40134507e-01 9.56873372e-02
2.93661654e-01 -4.51587468e-01 4.75082517e-01 -1.50536704e+00
3.49045098e-01 -4.76387054e-01 5.95191479e-01 -7.54481256e-01
-9.94046092e-01 5.68794012e-01 9.20509547e-02 -1.79112065e+00
-3.08445662e-01 -8.02994072e-02 -4.90457326e-01 3.59835297e-01
-1.46000886e+00 -1.14892805e+00 -2.11826473e-01 7.21238434e-01
9.04558897e-01 6.99237064e-02 3.74719471e-01 5.80441713e-01
-4.11578923e-01 3.39137733e-01 1.28681242e-01 3.36525649e-01
8.59453022e-01 -1.00113845e+00 2.18664765e-01 8.09982657e-01
5.40686309e-01 4.22464281e-01 5.09524524e-01 -6.60478294e-01
-1.31533515e+00 -1.02636135e+00 1.13985741e+00 -3.31569791e-01
8.94587159e-01 -3.42656881e-01 -8.53007853e-01 3.76168847e-01
-6.94902521e-03 4.45144415e-01 2.94448406e-01 -2.83459723e-01
-3.39434683e-01 -1.19276457e-01 -5.54055750e-01 7.34168053e-01
1.00146520e+00 -8.90077174e-01 -5.17983496e-01 7.43846774e-01
6.91348135e-01 -4.75179464e-01 -5.81809521e-01 4.11742449e-01
4.73156780e-01 -6.58486187e-01 9.78865385e-01 -3.78253847e-01
3.23806584e-01 -5.31802177e-01 2.81951670e-03 -8.01224113e-01
1.48607860e-03 -6.94667995e-01 -3.50100845e-01 1.13627684e+00
2.14053825e-01 2.08720803e-01 9.04347718e-01 3.65903050e-01
-7.17780143e-02 -9.14040923e-01 -1.09850752e+00 -7.48893678e-01
-5.07351100e-01 -5.09848535e-01 2.17785478e-01 9.06651318e-01
7.45218098e-02 3.53948951e-01 -5.94860137e-01 2.46854588e-01
4.27999020e-01 4.20532972e-01 5.47414064e-01 -9.96731341e-01
5.74854054e-02 -4.92591649e-01 -4.71320301e-01 -1.50267184e+00
2.32901827e-01 -7.76138604e-01 2.27541983e-01 -1.54948246e+00
3.14498842e-01 -2.12344378e-01 -6.28660977e-01 2.05947146e-01
-9.83274058e-02 3.16682845e-01 4.23681214e-02 6.79466486e-01
-1.42046165e+00 6.09120488e-01 1.15662098e+00 -1.85370460e-01
-3.75427544e-01 -5.04777618e-02 9.93326232e-02 8.79444420e-01
5.56383610e-01 -5.24156034e-01 -4.25568074e-01 -5.44427693e-01
2.54533648e-01 2.28787854e-01 2.91419297e-01 -1.15653181e+00
4.18679297e-01 -2.67779797e-01 3.49925071e-01 -9.54944253e-01
4.88542676e-01 -6.76162124e-01 -1.15587525e-01 1.49832115e-01
-5.68937659e-01 7.41379037e-02 -1.65079027e-01 7.85686195e-01
-4.98395890e-01 -1.90574020e-01 5.70088804e-01 -4.30478603e-02
-1.01004279e+00 7.84041524e-01 -1.34046957e-01 -6.28518835e-02
1.05085945e+00 8.33572075e-02 7.76402354e-02 -5.04200041e-01
-7.20842898e-01 3.67986649e-01 3.28356385e-01 7.32355952e-01
6.93131447e-01 -1.53610933e+00 -4.73397911e-01 -2.29869559e-01
3.30963820e-01 -5.19004352e-02 1.89299569e-01 1.07107890e+00
-3.80030364e-01 5.27569890e-01 1.89232618e-01 -8.45818937e-01
-1.14549172e+00 6.12360239e-01 5.32139875e-02 -3.00954849e-01
-5.64705431e-01 8.73515904e-01 1.30985573e-01 1.56657487e-01
4.87917513e-01 -3.89287442e-01 -4.31352496e-01 4.36305642e-01
5.13701975e-01 -3.30189653e-02 -2.33863235e-01 -9.95613158e-01
-4.90689129e-01 7.93369234e-01 -1.17095262e-01 -2.70500034e-01
1.18714821e+00 -3.50010306e-01 -2.01807451e-03 5.98096728e-01
1.51634955e+00 -8.34357068e-02 -1.52879381e+00 -5.45242488e-01
3.78762901e-01 -3.68269503e-01 -1.65767208e-01 -1.25142127e-01
-1.06629276e+00 7.75047183e-01 5.79919219e-01 6.61289468e-02
1.06633520e+00 3.87574077e-01 9.76712346e-01 4.54145700e-01
2.33681798e-01 -1.30513656e+00 4.71157432e-01 5.59545875e-01
7.67827809e-01 -1.29109454e+00 -8.51063058e-02 -1.19763874e-01
-5.07350624e-01 1.05826819e+00 5.94026148e-01 -1.06902733e-01
4.21208739e-01 -4.16708201e-01 -5.54872444e-03 -2.17191845e-01
-7.02829421e-01 -2.76518375e-01 6.29693568e-01 1.07763693e-01
3.78927469e-01 -3.33492935e-01 -4.66852754e-01 2.18610555e-01
2.67326742e-01 2.11576093e-02 5.61517626e-02 9.59965467e-01
-4.62525934e-01 -9.13882494e-01 -2.33862147e-01 1.89528644e-01
-6.27056420e-01 5.03790975e-02 -1.82400569e-01 4.49595034e-01
-4.80620340e-02 6.07742429e-01 9.25309062e-02 -3.10990185e-01
1.72566935e-01 8.20324942e-02 2.74157465e-01 -4.42473739e-01
-2.42904812e-01 6.77979112e-01 -1.45265147e-01 -7.18456507e-01
-7.52301693e-01 -7.48904049e-01 -1.35122693e+00 3.13772768e-01
-2.33662784e-01 4.03337628e-01 4.82494861e-01 9.16522443e-01
2.59388804e-01 1.31528839e-01 7.74603724e-01 -9.97935832e-01
-1.86243027e-01 -8.00034285e-01 -2.32569426e-01 5.90932608e-01
2.99771041e-01 -6.49604440e-01 -2.85643280e-01 2.11224794e-01]
|
[10.029834747314453, 0.6922467350959778]
|
02ef1705-613f-444f-a114-51a556ca78e8
|
using-motif-transitions-for-temporal-graph
|
2306.11190
| null |
https://arxiv.org/abs/2306.11190v1
|
https://arxiv.org/pdf/2306.11190v1.pdf
|
Using Motif Transitions for Temporal Graph Generation
|
Graph generative models are highly important for sharing surrogate data and benchmarking purposes. Real-world complex systems often exhibit dynamic nature, where the interactions among nodes change over time in the form of a temporal network. Most temporal network generation models extend the static graph generation models by incorporating temporality in the generation process. More recently, temporal motifs are used to generate temporal networks with better success. However, existing models are often restricted to a small set of predefined motif patterns due to the high computational cost of counting temporal motifs. In this work, we develop a practical temporal graph generator, Motif Transition Model (MTM), to generate synthetic temporal networks with realistic global and local features. Our key idea is modeling the arrival of new events as temporal motif transition processes. We first calculate the transition properties from the input graph and then simulate the motif transition processes based on the transition probabilities and transition rates. We demonstrate that our model consistently outperforms the baselines with respect to preserving various global and local temporal graph statistics and runtime performance.
|
['A. Erdem Sarıyüce', 'Penghang Liu']
|
2023-06-19
| null | null | null | null |
['benchmarking', 'benchmarking']
|
['miscellaneous', 'robots']
|
[ 1.73703209e-01 -3.02451421e-02 -2.94122189e-01 1.04801014e-01
-2.59557292e-02 -7.50372469e-01 1.15357971e+00 2.42387041e-01
1.66394666e-01 7.34308362e-01 1.04330599e-01 -5.00072598e-01
-2.10288286e-01 -1.30872822e+00 -6.58165216e-01 -3.96939009e-01
-7.79357016e-01 6.57525778e-01 6.72042370e-01 -1.60411060e-01
-1.54840767e-01 3.28284174e-01 -9.68083680e-01 -1.27057344e-01
7.70767868e-01 1.60182193e-01 3.39186639e-02 8.10428143e-01
-2.53475420e-02 7.87203789e-01 -6.33037031e-01 -3.54625463e-01
2.11563379e-01 -8.97603273e-01 -4.92027521e-01 2.10335068e-02
-4.28110242e-01 -7.52906799e-02 -9.24109519e-01 7.29807734e-01
2.62940466e-01 3.77071112e-01 5.70562243e-01 -1.86541080e+00
-3.67048740e-01 9.19142723e-01 -5.22920728e-01 2.26717502e-01
4.61657315e-01 4.60243970e-01 1.09201574e+00 -3.40279907e-01
8.93513918e-01 1.15938413e+00 7.21058607e-01 4.23187733e-01
-1.65738404e+00 -6.01516247e-01 3.56971979e-01 4.49775644e-02
-1.32116342e+00 2.05068514e-02 9.05366659e-01 -2.86319762e-01
8.15290093e-01 1.62109867e-01 9.81145561e-01 1.44415045e+00
5.17524123e-01 4.21327919e-01 7.01716721e-01 -1.41944334e-01
1.66319460e-01 -6.11151159e-01 -1.30919786e-02 6.66814268e-01
4.80656624e-02 2.39184424e-01 -5.93493879e-01 -5.76125920e-01
9.42183375e-01 3.18027616e-01 1.08124033e-01 -1.68581635e-01
-1.39766490e+00 6.90909863e-01 3.27077538e-01 2.21104190e-01
-3.88377041e-01 7.54150391e-01 3.42561930e-01 3.87420088e-01
3.01072568e-01 -1.22855179e-01 -4.31878911e-03 -3.39628309e-01
-9.47930217e-01 3.85733515e-01 1.04023182e+00 1.06209552e+00
7.53868401e-01 1.58528671e-01 -5.08302689e-01 4.67133224e-01
2.85871506e-01 4.00986105e-01 7.23341182e-02 -6.23310745e-01
3.01844150e-01 5.83256841e-01 2.37399489e-02 -1.37438703e+00
-2.30470151e-01 -3.41039360e-01 -1.18262613e+00 -5.13499856e-01
4.23215955e-01 -1.32821918e-01 -1.07666242e+00 1.97662163e+00
3.02218288e-01 7.78614640e-01 -2.79355973e-01 2.27759466e-01
4.76088732e-01 9.14750040e-01 3.92042175e-02 -2.86944628e-01
8.08543682e-01 -7.98669517e-01 -6.31758511e-01 1.13829605e-01
4.36759651e-01 -5.38941085e-01 9.32202041e-01 -1.73411191e-01
-9.56844211e-01 -6.40115812e-02 -5.76063335e-01 6.71171129e-01
-1.15050890e-01 -4.37637866e-01 7.24624336e-01 4.20059949e-01
-1.23661566e+00 8.56385708e-01 -1.20898545e+00 -6.00152552e-01
-2.35843305e-02 7.09646642e-02 -1.21394545e-02 -4.43125777e-02
-1.31615150e+00 1.70382842e-01 2.82199830e-01 6.64044768e-02
-1.32862568e+00 -6.24298036e-01 -7.95308709e-01 -3.81809212e-02
3.96358401e-01 -8.70755613e-01 1.09590626e+00 -6.04524076e-01
-1.36285913e+00 1.74515188e-01 -2.75332421e-01 -7.71146595e-01
6.44326150e-01 4.53843266e-01 -6.26382411e-01 -9.25626084e-02
1.52509078e-01 2.24632829e-01 6.73630118e-01 -9.58643794e-01
-1.47255883e-01 3.52221042e-01 5.04557490e-02 -6.70744628e-02
-3.14500555e-02 -2.94580370e-01 -7.66952813e-01 -9.45032835e-01
-1.08602092e-01 -1.23727596e+00 -7.38659143e-01 -4.90467250e-01
-8.34914207e-01 -2.19095528e-01 8.27542841e-01 -1.75821438e-01
1.65164244e+00 -2.01053882e+00 9.14404076e-03 6.37808561e-01
3.05767030e-01 -2.78898120e-01 -3.24603558e-01 1.32422841e+00
6.89900145e-02 4.17041093e-01 -2.01384157e-01 -3.26147109e-01
6.29165024e-02 3.66793066e-01 -3.03019464e-01 1.80176631e-01
-9.06579494e-02 1.16575444e+00 -1.30575335e+00 -3.28902185e-01
4.02450189e-02 1.95140511e-01 -4.40837741e-01 -8.09429865e-03
-5.69493532e-01 4.69941020e-01 -3.40821892e-01 2.31370658e-01
2.50520498e-01 -7.65361011e-01 5.63359618e-01 2.78071016e-01
2.83235997e-01 3.41006756e-01 -1.13840222e+00 1.41316903e+00
-1.52592689e-01 3.86169195e-01 -5.02501786e-01 -5.82891285e-01
7.53820956e-01 3.71033937e-01 7.35432565e-01 -4.71052349e-01
-2.76458859e-01 -1.03740305e-01 3.34012806e-01 1.00232840e-01
4.93647158e-01 1.94120347e-01 -2.02928215e-01 1.07665777e+00
-2.88748443e-01 -2.71002855e-03 7.39898503e-01 8.17078710e-01
1.78727162e+00 -1.19423404e-01 8.31118003e-02 2.46399157e-02
-7.57971257e-02 3.22338417e-02 5.84381282e-01 8.21118593e-01
1.21817231e-01 5.47725201e-01 1.07205784e+00 -5.77928901e-01
-1.15350986e+00 -1.46485281e+00 6.20463908e-01 6.51757061e-01
6.26072586e-02 -9.56818104e-01 -4.17668283e-01 -6.04787588e-01
-6.30098283e-02 4.48750854e-01 -7.50135243e-01 -3.82761955e-01
-6.30086601e-01 -9.36146557e-01 5.39623618e-01 2.27983758e-01
1.93182036e-01 -1.33758485e+00 -5.89363463e-02 5.76434493e-01
-5.84793910e-02 -1.02280247e+00 -9.79696810e-01 -2.96917111e-01
-9.13370550e-01 -1.12966728e+00 -3.27379882e-01 -5.74409664e-01
7.84339070e-01 2.27290496e-01 1.24891138e+00 1.90454081e-01
-2.32578352e-01 3.80663365e-01 -2.95851797e-01 1.46983951e-01
-6.83096647e-01 2.89777219e-01 -2.20590070e-01 1.18108749e-01
-2.11145520e-01 -1.27153981e+00 -7.55033910e-01 3.76637071e-01
-1.20001733e+00 3.89307916e-01 3.38040292e-01 6.89432263e-01
4.96488452e-01 2.02462748e-01 5.57542205e-01 -1.18278491e+00
1.03506148e+00 -7.44860649e-01 -4.80471611e-01 1.54634804e-01
-6.88233495e-01 7.91757647e-03 7.75380731e-01 -7.61938095e-01
-5.65174997e-01 -2.74022043e-01 4.94684577e-01 -4.18679833e-01
2.25454912e-01 8.39446783e-01 1.31878570e-01 3.03978443e-01
5.92970848e-01 5.82490742e-01 3.71675676e-04 1.53971329e-01
4.19110388e-01 -3.91993225e-01 1.06979676e-01 -7.38350272e-01
1.13606977e+00 5.05206108e-01 3.56047988e-01 -4.62550521e-01
-1.28248453e-01 -2.17631489e-01 -2.00839728e-01 -4.44621176e-01
1.54191256e-01 -5.46217501e-01 -6.11124396e-01 6.00723445e-01
-1.04277647e+00 -8.50320041e-01 -3.94332051e-01 1.23959325e-01
-3.21328729e-01 3.07414711e-01 -9.33516681e-01 -5.85401893e-01
-2.57963389e-01 -8.00337672e-01 8.52715075e-01 -1.30958902e-02
-3.70258987e-01 -1.36490166e+00 4.33215082e-01 -7.00648904e-01
5.43768883e-01 7.80856729e-01 9.64948237e-01 -2.59094566e-01
-9.89683092e-01 -2.82335877e-01 -8.56085271e-02 -4.56291586e-01
5.01469731e-01 5.16129911e-01 -9.81015936e-02 -3.13645005e-01
-8.48987818e-01 4.62759912e-01 5.58536947e-01 4.65512544e-01
8.71779025e-01 -4.43529099e-01 -6.88847601e-01 4.02852714e-01
1.29678380e+00 2.66274542e-01 6.94851041e-01 -1.07031010e-01
8.04316521e-01 3.70458394e-01 2.04642236e-01 5.10135412e-01
4.40080523e-01 4.62103426e-01 2.83699214e-01 1.05752863e-01
-5.22959919e-04 -7.97320426e-01 3.91243815e-01 1.01482177e+00
-9.09266919e-02 -6.57929838e-01 -1.26631439e+00 8.63820493e-01
-2.23154759e+00 -1.17183912e+00 -3.45846951e-01 2.22716022e+00
6.10181630e-01 4.13844049e-01 5.23345232e-01 -1.13095239e-01
9.20238674e-01 4.22863126e-01 -3.91846716e-01 2.21642759e-03
6.52565360e-02 6.09362237e-02 4.59170997e-01 4.18232054e-01
-5.34437597e-01 9.44008827e-01 6.62586641e+00 7.46720314e-01
-8.49165976e-01 -6.50833175e-02 6.58936977e-01 -3.51939090e-02
-5.87558329e-01 4.27192658e-01 -4.98217016e-01 6.69173002e-01
1.33971477e+00 -8.92381251e-01 6.79747999e-01 3.80484998e-01
6.11618400e-01 1.97887927e-01 -1.04484248e+00 7.78310120e-01
-5.00919223e-01 -1.50610483e+00 4.09615278e-01 2.58913517e-01
9.91488099e-01 -4.58648279e-02 1.27405319e-02 1.73197523e-01
1.12286663e+00 -1.05271029e+00 3.99164110e-01 7.57410765e-01
5.67319751e-01 -8.45587075e-01 1.05141558e-01 3.24321985e-01
-1.77415752e+00 3.64074856e-01 1.88995093e-01 3.90628949e-02
5.21208942e-01 8.58241022e-01 -1.05403090e+00 7.09206581e-01
4.24561679e-01 1.07129633e+00 -4.21401083e-01 9.76050675e-01
-1.37391046e-01 1.06983507e+00 -6.54315710e-01 -1.20514192e-01
3.34312171e-01 -5.97630024e-01 7.26912141e-01 8.81520629e-01
3.58416080e-01 -3.71855706e-01 3.45434666e-01 9.85115230e-01
-1.17145531e-01 -1.13448367e-01 -9.46987629e-01 -4.96311009e-01
7.67854750e-01 1.17199886e+00 -1.15938294e+00 -1.63035586e-01
-2.82362729e-01 7.85664499e-01 1.74666300e-01 7.19069302e-01
-1.12916458e+00 -2.11613268e-01 4.19757098e-01 4.99209493e-01
2.51477920e-02 -6.76123738e-01 3.39099169e-01 -9.54917371e-01
3.96540426e-02 -8.24987948e-01 3.94729078e-01 -3.52776319e-01
-1.37223256e+00 6.06371462e-01 9.57590044e-02 -1.24389005e+00
-4.83269274e-01 2.14265496e-01 -1.17992878e+00 6.86098278e-01
-8.79343450e-01 -1.17818630e+00 -2.84091711e-01 6.96399629e-01
2.03847885e-01 2.24343061e-01 4.96490180e-01 2.49797136e-01
-5.01217663e-01 4.35520887e-01 -5.06666787e-02 1.97166070e-01
3.80087227e-01 -1.26505232e+00 1.17394841e+00 1.12186313e+00
2.63877630e-01 7.19586194e-01 8.77195239e-01 -1.03971255e+00
-1.27762413e+00 -1.61480653e+00 7.55042851e-01 -3.00955623e-01
1.09892046e+00 -6.21693909e-01 -6.68234110e-01 8.94768834e-01
1.13506958e-01 9.28238779e-02 2.99712420e-01 -7.62722865e-02
-9.20567214e-02 4.69636619e-02 -7.25415349e-01 1.17385709e+00
1.47170913e+00 -4.42930102e-01 3.39128584e-01 4.73324001e-01
9.49797451e-01 -2.40989029e-01 -7.69078374e-01 2.29247525e-01
3.16735208e-01 -5.93496561e-01 6.33616209e-01 -5.07847846e-01
2.42496073e-01 -5.13602138e-01 3.62019688e-01 -1.35356247e+00
-3.45009685e-01 -1.43002224e+00 -3.65000367e-01 1.29818034e+00
7.42953300e-01 -7.97793090e-01 9.97048378e-01 2.34581396e-01
3.26524496e-01 -4.93945360e-01 -7.85933495e-01 -1.07768548e+00
-5.26009083e-01 -3.39406669e-01 9.18413699e-01 9.09717083e-01
-1.85136467e-01 4.18921739e-01 -5.73127270e-01 4.81162816e-02
5.72516501e-01 2.05565885e-01 1.06035602e+00 -1.06287479e+00
-4.55367446e-01 -4.42812264e-01 -3.88394624e-01 -7.99395561e-01
-9.47373435e-02 -9.50682640e-01 -1.03077367e-01 -1.66409254e+00
3.40686530e-01 -3.98819447e-01 -1.96774416e-02 1.62682071e-01
-9.24215019e-02 -1.61745057e-01 -7.70548955e-02 2.06504136e-01
-6.65442050e-01 7.20590889e-01 1.24822056e+00 4.43827100e-02
-4.56829637e-01 2.21203834e-01 -3.51369716e-02 2.71083802e-01
7.91657150e-01 -5.99936366e-01 -9.17791903e-01 3.99970599e-02
5.71693778e-01 4.70665574e-01 3.32034737e-01 -7.94427574e-01
4.73594964e-01 -3.31521869e-01 -3.07455271e-01 -6.47863686e-01
4.09270003e-02 -5.50182819e-01 1.13408196e+00 5.59324086e-01
-1.87247336e-01 6.75277710e-01 3.98175754e-02 1.24886096e+00
-1.74518436e-01 5.17145753e-01 2.69257426e-01 -4.72296216e-02
-2.73191065e-01 1.08296001e+00 -5.58700085e-01 1.59847066e-01
1.16175568e+00 -4.53334562e-02 -3.73262376e-01 -8.79053950e-01
-9.17590737e-01 3.33839715e-01 6.58122122e-01 3.94117177e-01
5.61368227e-01 -1.51453662e+00 -4.95493889e-01 -1.25982150e-01
-1.27391806e-02 -1.99266989e-02 9.38549042e-02 8.77656519e-01
-6.09120011e-01 -4.81322780e-03 1.83832005e-01 -6.33100748e-01
-1.03578162e+00 5.71200252e-01 4.36161846e-01 -8.79451513e-01
-7.37611294e-01 2.50553280e-01 3.78214754e-02 -4.01218832e-01
-1.28180668e-01 -2.59012789e-01 1.78740963e-01 -2.49884382e-01
5.66576682e-02 2.90154040e-01 -3.85077953e-01 -2.42347300e-01
-8.19075331e-02 1.41946256e-01 2.00101919e-02 -2.99642026e-01
1.25469768e+00 1.09038323e-01 -3.06214601e-01 7.27230191e-01
7.15009809e-01 -6.94109276e-02 -1.06319189e+00 -4.19593662e-01
1.88013926e-01 -3.06258589e-01 -6.37018502e-01 -1.28711209e-01
-1.20984805e+00 1.95395410e-01 -8.31961855e-02 7.27374077e-01
9.09875751e-01 -1.33366203e-02 9.49926376e-01 5.71688311e-03
6.11284614e-01 -6.37182236e-01 2.98059106e-01 5.06566644e-01
5.40837646e-01 -6.62056804e-01 -2.35111460e-01 -6.82925999e-01
-3.19850743e-01 7.36470461e-01 4.57571328e-01 -4.73648787e-01
9.63214219e-01 1.84150919e-01 -5.86126506e-01 -2.95710653e-01
-1.25860834e+00 1.43328207e-02 9.96989608e-02 5.75973570e-01
3.08112800e-01 2.93813139e-01 -3.55827659e-01 7.20710978e-02
-1.62640125e-01 -2.76974052e-01 6.76479280e-01 8.13804567e-01
1.93999976e-01 -1.42478812e+00 1.69586942e-01 7.35199571e-01
-3.53008151e-01 -2.81857789e-01 -4.06805068e-01 8.64295304e-01
-4.62558031e-01 8.20465684e-01 8.64476115e-02 -3.75580788e-01
1.75942615e-01 -2.08823949e-01 2.22855031e-01 -7.55028367e-01
-4.75968093e-01 3.51838320e-02 2.03717977e-01 -5.70990980e-01
-1.79178670e-01 -8.12427461e-01 -9.76304829e-01 -8.36758137e-01
-8.47723261e-02 3.10018569e-01 1.44838214e-01 5.00173271e-01
5.99434674e-01 9.09429550e-01 7.08132088e-01 -4.59595591e-01
1.19567133e-01 -9.40224230e-01 -6.93623781e-01 5.34390152e-01
1.04898378e-01 -4.48894083e-01 -4.00024951e-01 1.77291825e-01]
|
[7.218002796173096, 5.876262187957764]
|
dab6228e-38f7-4d51-9bf4-c18bb8708040
|
deep-multi-task-model-for-sarcasm-detection
|
2106.12488
| null |
https://arxiv.org/abs/2106.12488v1
|
https://arxiv.org/pdf/2106.12488v1.pdf
|
Deep Multi-Task Model for Sarcasm Detection and Sentiment Analysis in Arabic Language
|
The prominence of figurative language devices, such as sarcasm and irony, poses serious challenges for Arabic Sentiment Analysis (SA). While previous research works tackle SA and sarcasm detection separately, this paper introduces an end-to-end deep Multi-Task Learning (MTL) model, allowing knowledge interaction between the two tasks. Our MTL model's architecture consists of a Bidirectional Encoder Representation from Transformers (BERT) model, a multi-task attention interaction module, and two task classifiers. The overall obtained results show that our proposed model outperforms its single-task counterparts on both SA and sarcasm detection sub-tasks.
|
['Ahmed Khoumsi', 'Ismail Berrada', 'Nabil El Mamoun', 'Kabil Essefar', 'Abdellah El Mekki', 'Abdelkader El Mahdaouy']
|
2021-06-23
| null |
https://aclanthology.org/2021.wanlp-1.42
|
https://aclanthology.org/2021.wanlp-1.42.pdf
|
eacl-wanlp-2021-4
|
['arabic-sentiment-analysis']
|
['natural-language-processing']
|
[-2.55359292e-01 1.26576900e-01 -5.65218478e-02 -5.59344351e-01
-8.43928277e-01 -2.76901990e-01 7.65774310e-01 6.13148883e-02
-4.66188520e-01 2.78052658e-01 4.79181081e-01 -1.21243797e-01
5.91108620e-01 -3.36191863e-01 -5.74169338e-01 -4.33026582e-01
4.14598018e-01 4.46734905e-01 -1.71903670e-01 -8.19104254e-01
2.81512082e-01 -3.03303808e-01 -8.11737061e-01 9.76064682e-01
6.30075932e-01 1.12710321e+00 -6.23833500e-02 7.16231108e-01
1.01948239e-01 1.90256250e+00 -8.32044780e-01 -1.17056525e+00
-5.30666411e-01 -2.92947680e-01 -1.28035498e+00 -6.22277111e-02
1.23954169e-01 -2.47847453e-01 4.35682178e-01 7.64370620e-01
7.54470587e-01 -6.73205312e-03 5.29856503e-01 -1.21277344e+00
-1.20078993e+00 1.13007486e+00 -9.03203428e-01 -3.51580232e-02
2.60133266e-01 -2.48816505e-01 1.39546335e+00 -1.13970566e+00
2.49618627e-02 1.47552848e+00 1.00831556e+00 4.94543701e-01
-6.99606895e-01 -4.61256981e-01 6.87685162e-02 3.51209372e-01
-7.34960198e-01 -2.48128414e-01 1.27778089e+00 -5.35141528e-01
1.24917090e+00 -2.08400667e-01 4.46355075e-01 1.51676190e+00
4.40343857e-01 1.52873027e+00 8.69230211e-01 -3.82170141e-01
-2.99631685e-01 2.79425293e-01 5.51435590e-01 7.94419646e-01
-3.73591483e-01 -7.19774425e-01 -9.03373599e-01 7.02308044e-02
1.38943881e-01 -1.86001241e-01 2.85540909e-01 1.61566406e-01
-8.41491580e-01 1.08239067e+00 5.99359632e-01 3.36481780e-01
-1.58509359e-01 3.76285702e-01 1.05173397e+00 4.22062695e-01
7.22457588e-01 1.74440384e-01 -4.67533529e-01 -1.80964366e-01
-6.29582345e-01 7.49527737e-02 6.42213225e-01 5.24659276e-01
1.84451818e-01 1.18435986e-01 -1.50324777e-01 1.24362183e+00
5.04201889e-01 4.32726920e-01 7.75385797e-01 -8.93549994e-02
4.62344587e-01 6.77441895e-01 9.15178210e-02 -1.01346731e+00
-7.70206332e-01 -5.57020903e-01 -6.56269610e-01 -9.29847732e-02
9.52138007e-02 -3.81081164e-01 -1.86513633e-01 1.71472585e+00
-7.37752095e-02 -3.68335247e-01 3.14797223e-01 1.05652654e+00
1.38578594e+00 6.76186919e-01 1.82998165e-01 1.11412637e-01
1.74681306e+00 -1.82442486e+00 -9.69392419e-01 -6.54656649e-01
9.36303258e-01 -1.03821838e+00 1.63238883e+00 4.42431509e-01
-1.42296541e+00 -6.71160519e-01 -1.20262432e+00 -8.12395513e-01
-2.06978753e-01 9.62868631e-01 3.59377921e-01 3.57824355e-01
-6.66265488e-01 -5.12303822e-02 -7.06253886e-01 1.95414610e-02
4.26920742e-01 6.13894723e-02 1.48814976e-01 6.23648882e-01
-1.20326304e+00 1.52836251e+00 -3.14508639e-02 5.51737845e-01
-1.01177239e+00 -3.22545379e-01 -1.03170121e+00 1.12017222e-01
7.83142224e-02 -5.64201772e-01 1.59705293e+00 -1.48691320e+00
-1.86865342e+00 1.31391740e+00 8.31659585e-02 -4.53746140e-01
1.53733432e-01 -9.14257705e-01 -2.82468289e-01 -1.92793593e-01
-1.14862993e-02 2.39199534e-01 1.13766551e+00 -1.13539410e+00
-2.19494075e-01 -2.82103270e-01 4.38260704e-01 3.40573311e-01
-6.53006613e-01 7.58067191e-01 1.87547475e-01 -6.45940602e-01
-5.46095967e-01 -6.14575744e-01 2.46984094e-01 -5.05954444e-01
-5.41851461e-01 -5.91621637e-01 1.00262475e+00 -8.05531800e-01
1.20701766e+00 -2.15466833e+00 3.40322077e-01 -7.00407028e-01
2.59554058e-01 1.64082050e-01 -3.05590212e-01 4.20596898e-01
-6.52918369e-02 -3.51537347e-01 6.78799227e-02 -9.36383307e-01
1.54249042e-01 -1.28845304e-01 -3.27384174e-01 4.17474180e-01
4.58335280e-01 1.28086388e+00 -8.24679017e-01 -3.59045982e-01
-5.86464768e-04 5.84659159e-01 -3.02924544e-01 3.58589888e-01
-1.25428215e-01 1.46981701e-01 -2.37303525e-01 6.65617704e-01
4.51386333e-01 -7.09893823e-01 2.59483159e-01 -5.09832859e-01
4.47702706e-02 7.34209359e-01 -3.42082262e-01 1.69117773e+00
-1.11516702e+00 5.51435947e-01 2.99108833e-01 -1.02197754e+00
1.15914869e+00 4.55914468e-01 8.90812352e-02 -7.59247363e-01
6.83768392e-01 2.13269398e-01 -3.74829434e-02 -4.94353652e-01
4.18167531e-01 -4.70841885e-01 -6.26512229e-01 8.28435481e-01
1.94441363e-01 -2.51726300e-01 -1.23630494e-01 3.05204600e-01
6.16332173e-01 6.21852428e-02 4.83763307e-01 -4.45869058e-01
9.78735089e-01 -1.75108835e-01 2.72651970e-01 1.79455429e-01
-2.42640138e-01 1.55298173e-01 6.75399542e-01 -8.95844042e-01
-7.80607224e-01 -8.25430751e-01 2.46598683e-02 1.92357934e+00
1.99467704e-01 -5.59993207e-01 -4.69438165e-01 -8.42156291e-01
-1.49083182e-01 5.30415952e-01 -8.27733696e-01 -2.52067357e-01
-7.34522641e-01 -1.04119110e+00 6.49914205e-01 7.11486340e-01
6.66535139e-01 -1.44518566e+00 -1.10333157e+00 1.89636886e-01
-4.44149554e-01 -1.27432895e+00 -3.91839534e-01 3.64587665e-01
-4.47539240e-01 -1.05549884e+00 -4.58970219e-01 -1.30814838e+00
1.30911395e-01 2.05963757e-02 1.56416464e+00 1.26309037e-01
1.32207602e-01 -1.73117761e-02 -6.80884480e-01 -6.71401978e-01
-4.77646053e-01 9.27195400e-02 -4.55887198e-01 2.17440590e-01
4.98364210e-01 -2.16017708e-01 -1.86228007e-01 3.92397605e-02
-5.52810133e-01 4.93845910e-01 4.39444333e-01 1.38774276e+00
-1.15945928e-01 -6.67741537e-01 1.05661881e+00 -9.92846549e-01
1.00848401e+00 -3.46152276e-01 -1.45574600e-01 3.19610983e-01
-2.66889364e-01 -2.67624736e-01 8.35057557e-01 -2.42422715e-01
-1.15544391e+00 -8.36245250e-03 -3.54951710e-01 1.03586086e-03
3.66129607e-01 7.93785870e-01 2.46615745e-02 3.75947475e-01
5.58039188e-01 -4.12448309e-02 1.28235102e-01 -4.98272896e-01
4.25040036e-01 1.13704300e+00 5.48928916e-01 -4.47539419e-01
-1.44454716e-02 2.91675985e-01 -4.39200640e-01 -4.90204453e-01
-1.76967621e+00 -3.62101644e-01 -6.49186313e-01 -2.76077777e-01
9.58901167e-01 -1.31012571e+00 -1.01532423e+00 8.82728338e-01
-1.53691924e+00 -5.57572842e-01 7.71478638e-02 -6.73664510e-02
-6.45307124e-01 2.97453582e-01 -1.29028738e+00 -7.76073515e-01
-1.04252720e+00 -1.25734711e+00 1.38850367e+00 4.44060080e-02
-3.41381490e-01 -1.27304292e+00 1.10584877e-01 9.66283143e-01
3.43891680e-01 -6.83436031e-03 1.00345635e+00 -6.51039183e-01
4.98247594e-01 -2.52643414e-02 -3.42199594e-01 5.65574586e-01
-4.69960645e-02 -2.72656530e-01 -1.16885614e+00 -4.36518520e-01
4.92055118e-01 -1.46560836e+00 8.82197857e-01 3.15384328e-01
7.39405572e-01 -4.33406889e-01 2.45834529e-01 2.24277511e-01
1.03560770e+00 -1.19573459e-01 3.53277534e-01 5.25199652e-01
8.87162209e-01 4.36694682e-01 4.47164625e-01 6.17411554e-01
1.04323494e+00 8.00567031e-01 6.83797061e-01 -4.10873652e-01
-2.06114158e-01 1.93430245e-01 1.13060784e+00 1.23953736e+00
3.20373088e-01 -2.45535314e-01 -8.81056666e-01 5.69577992e-01
-2.05907130e+00 -8.10051441e-01 -3.02875161e-01 1.31895542e+00
1.04467356e+00 5.91781514e-04 4.64820772e-01 2.02730760e-01
3.33533674e-01 5.43755293e-01 -4.29897159e-01 -1.31421292e+00
-4.28612471e-01 6.88981488e-02 -2.85859585e-01 5.92565238e-01
-1.52482831e+00 1.18140447e+00 6.15195131e+00 5.34845889e-01
-1.22417212e+00 6.97313905e-01 7.01925218e-01 -1.13259621e-01
-3.63163501e-02 -4.67324495e-01 -5.82696915e-01 3.27759147e-01
7.77299643e-01 1.06524564e-01 -5.93931749e-02 1.00692976e+00
1.37134358e-01 1.32624432e-01 -8.48339498e-01 8.22174191e-01
7.49178946e-01 -9.65040743e-01 5.95141836e-02 -7.15214550e-01
4.94323313e-01 9.88348722e-02 2.16902360e-01 5.32737255e-01
4.29063171e-01 -1.00558293e+00 1.11878049e+00 -6.63223788e-02
5.69230795e-01 -6.37157261e-01 1.16912401e+00 1.65577516e-01
-9.64919329e-01 -4.92030919e-01 -3.05038225e-02 -5.26907563e-01
4.78838950e-01 4.92804080e-01 -4.67408091e-01 1.91852942e-01
5.87052763e-01 1.46099806e+00 -5.61000109e-01 1.46710888e-01
-6.48829401e-01 7.73093104e-01 1.85831919e-01 -3.16870362e-01
5.77703536e-01 -1.45526469e-01 4.45154533e-02 1.47672307e+00
-2.58916229e-01 -3.92834604e-01 2.20243350e-01 6.55783415e-01
-1.25698686e-01 4.13499415e-01 -7.07895830e-02 8.39880202e-03
-1.85005710e-01 1.57013381e+00 -4.04252797e-01 -3.47782433e-01
-7.87469745e-01 1.15774107e+00 7.40154326e-01 -2.17206568e-01
-1.08342516e+00 -9.22433585e-02 3.74953985e-01 -5.31473577e-01
1.88635960e-01 -4.45701145e-02 -7.47950613e-01 -1.27212834e+00
-5.43651655e-02 -9.03878152e-01 6.90766454e-01 -9.60376799e-01
-1.50367785e+00 7.94021189e-01 -6.58298910e-01 -8.59234154e-01
-1.19577199e-01 -8.87219012e-01 -8.12979877e-01 7.11965561e-01
-1.53402448e+00 -2.29341197e+00 -9.26539823e-02 4.46892828e-01
1.02193749e+00 -2.00044513e-01 9.08298194e-01 2.52200335e-01
-7.35891521e-01 7.60007083e-01 -3.19641858e-01 3.34423244e-01
6.69859648e-01 -1.40972817e+00 3.12356859e-01 6.92989349e-01
8.09136108e-02 1.12890393e-01 4.62017417e-01 -2.69067407e-01
-1.10670340e+00 -6.96460307e-01 1.22232878e+00 -5.78646362e-01
9.43433225e-01 -4.79821950e-01 -6.26119256e-01 7.58266389e-01
7.24423885e-01 -4.24051255e-01 7.49551415e-01 5.89741528e-01
-6.70691133e-01 1.60558462e-01 -5.59495568e-01 2.59596050e-01
3.44077021e-01 -7.11963236e-01 -8.10838759e-01 5.46105027e-01
3.66176486e-01 -2.37388939e-01 -5.93301058e-01 4.18193281e-01
6.04567409e-01 -9.08278942e-01 8.60517919e-01 -8.41449916e-01
1.41059899e+00 1.38820887e-01 6.19034246e-02 -1.48941362e+00
-1.67470351e-01 -3.57295394e-01 -3.49026352e-01 9.05806303e-01
4.73591626e-01 7.05617443e-02 4.07509029e-01 -9.55288857e-02
-5.67761242e-01 -9.76199925e-01 -6.04552388e-01 -2.21540347e-01
3.73898327e-01 -2.01824233e-01 1.82079241e-01 1.03625941e+00
6.29729331e-01 1.59292650e+00 -9.72454786e-01 -2.97892481e-01
-3.45377102e-02 6.66927516e-01 6.03250086e-01 -9.49712873e-01
-3.66933554e-01 -8.46890926e-01 1.83284760e-01 -1.20728040e+00
3.64034861e-01 -9.95579898e-01 -5.43260388e-02 -1.42831957e+00
4.87138182e-01 -5.20562660e-03 -3.25337559e-01 8.21317196e-01
-2.34168708e-01 4.68463540e-01 4.47787285e-01 4.53121625e-02
-1.07755947e+00 8.37122619e-01 1.24568534e+00 -2.51093864e-01
-6.70815334e-02 -2.50066608e-01 -8.04828584e-01 9.45117533e-01
8.02201211e-01 -1.79444075e-01 -2.09711701e-01 -1.00817633e+00
8.99684608e-01 -2.25967970e-02 2.35387295e-01 -4.34629560e-01
-5.31879207e-03 2.92790085e-01 -1.21406950e-01 -6.48842454e-01
6.27229512e-01 -3.52494478e-01 -7.62075543e-01 5.22504270e-01
-5.96783042e-01 4.93615597e-01 1.57730937e-01 -1.35966554e-01
-4.28019941e-01 -2.24962026e-01 1.04329562e+00 8.46912116e-02
-3.62357229e-01 -4.92436439e-01 -5.75355887e-01 1.30846992e-01
7.11206973e-01 3.70411932e-01 -6.01156235e-01 -4.71917987e-01
-7.29137480e-01 3.10774565e-01 -2.02149805e-02 7.21483648e-01
5.91341197e-01 -1.07230365e+00 -1.08766234e+00 2.44293511e-02
9.15764570e-02 -1.43418118e-01 6.70019388e-02 1.14993119e+00
-3.03839117e-01 3.91683541e-02 -3.84162307e-01 -3.22089612e-01
-1.37972975e+00 3.68470937e-01 4.30026472e-01 -5.14960289e-01
-4.78482276e-01 1.22517359e+00 1.41249999e-01 -5.32774210e-01
1.83002859e-01 -2.74048150e-02 -6.94233119e-01 4.13994521e-01
5.23290157e-01 2.53373891e-01 1.00133732e-01 -9.54716980e-01
-3.83645624e-01 2.87203193e-01 -4.48839843e-01 1.98287725e-01
1.35428464e+00 -2.21204311e-01 -3.74412954e-01 8.51977885e-01
1.13135684e+00 -1.41805828e-01 -7.79676676e-01 -1.55671895e-01
1.41777605e-01 1.27840802e-01 9.37349051e-02 -1.30462062e+00
-9.17933941e-01 1.24496818e+00 5.96416183e-02 1.26500770e-01
1.18829131e+00 1.38806880e-01 1.16757858e+00 3.72930229e-01
-1.09001145e-01 -1.23417568e+00 8.73653173e-01 9.89908755e-01
1.21279049e+00 -1.48023021e+00 -1.73047543e-01 -1.56335279e-01
-1.36485040e+00 1.06944859e+00 8.76200736e-01 -2.85987675e-01
6.31458580e-01 4.96294230e-01 5.79592407e-01 -6.47175550e-01
-1.03102672e+00 7.15492815e-02 8.20791721e-02 4.04529795e-02
1.10940254e+00 -6.33734912e-02 -5.83427191e-01 1.37335896e+00
-4.79371011e-01 -1.34619236e-01 7.07119107e-01 6.98352158e-01
-1.11558393e-01 -7.96154737e-01 8.22440684e-02 2.16334358e-01
-7.25330234e-01 -2.47862831e-01 -5.78334391e-01 3.15352887e-01
-1.02547087e-01 1.00186253e+00 1.89878181e-01 -5.01565337e-01
1.59875080e-01 2.46420484e-02 2.67331302e-01 -5.80588818e-01
-1.52459407e+00 9.63075235e-02 4.21061754e-01 -3.53762299e-01
-8.08151364e-01 -3.27191234e-01 -1.12272441e+00 -1.00471601e-01
-4.54821914e-01 1.26615524e-01 4.83190894e-01 1.19240952e+00
1.43300608e-01 4.85325843e-01 7.51952648e-01 -6.75009251e-01
-7.57691145e-01 -1.36403191e+00 -5.31180143e-01 6.90278411e-01
2.78044611e-01 -6.28822088e-01 -7.76372151e-03 2.65282422e-01]
|
[9.166670799255371, 10.507304191589355]
|
5efddd97-016f-493e-8114-ecf75f593cab
|
consistent-rank-logits-for-ordinal-regression
|
1901.07884
| null |
https://arxiv.org/abs/1901.07884v7
|
https://arxiv.org/pdf/1901.07884v7.pdf
|
Rank consistent ordinal regression for neural networks with application to age estimation
|
In many real-world prediction tasks, class labels include information about the relative ordering between labels, which is not captured by commonly-used loss functions such as multi-category cross-entropy. Recently, the deep learning community adopted ordinal regression frameworks to take such ordering information into account. Neural networks were equipped with ordinal regression capabilities by transforming ordinal targets into binary classification subtasks. However, this method suffers from inconsistencies among the different binary classifiers. To resolve these inconsistencies, we propose the COnsistent RAnk Logits (CORAL) framework with strong theoretical guarantees for rank-monotonicity and consistent confidence scores. Moreover, the proposed method is architecture-agnostic and can extend arbitrary state-of-the-art deep neural network classifiers for ordinal regression tasks. The empirical evaluation of the proposed rank-consistent method on a range of face-image datasets for age prediction shows a substantial reduction of the prediction error compared to the reference ordinal regression network.
|
['Sebastian Raschka', 'Vahid Mirjalili', 'Wenzhi Cao']
|
2019-01-20
| null | null | null | null |
['age-and-gender-classification', 'gender-prediction']
|
['computer-vision', 'computer-vision']
|
[ 2.23297656e-01 3.72106761e-01 -7.63269961e-01 -1.12820041e+00
-5.50833464e-01 -1.60606995e-01 6.63173556e-01 3.98820460e-01
-3.49811286e-01 9.96681154e-01 -9.67050809e-03 1.53025938e-02
-8.68655205e-01 -6.65894806e-01 -5.39625585e-01 -4.63149846e-01
-3.36260617e-01 7.24591792e-01 -3.65367174e-01 2.70707846e-01
2.20297620e-01 2.37734735e-01 -1.96313941e+00 3.09691191e-01
9.93368089e-01 1.72334516e+00 -6.00134254e-01 3.27948034e-02
-3.51200476e-02 9.30289328e-01 -1.23499259e-01 -9.36868370e-01
7.87551478e-02 2.20487081e-02 -7.58957803e-01 -5.28336346e-01
9.82730985e-01 -4.94355619e-01 2.33047921e-02 1.00590611e+00
2.35906139e-01 -2.98887491e-01 1.32269323e+00 -1.74436533e+00
-8.26661646e-01 8.65329504e-01 -6.48064673e-01 -2.84541994e-01
3.75570618e-02 -4.92372513e-01 1.62512970e+00 -5.13293982e-01
2.75766347e-02 1.39869165e+00 9.74587083e-01 6.16587877e-01
-1.34373510e+00 -1.11908209e+00 1.34907395e-01 4.91791725e-01
-1.09213483e+00 -3.62543136e-01 7.38611341e-01 -6.09831095e-01
3.56293291e-01 2.90609747e-01 1.86410278e-01 1.34001815e+00
3.68218198e-02 4.24629927e-01 1.62187994e+00 -3.36614609e-01
9.40132737e-02 9.91511792e-02 6.51180685e-01 9.49428439e-01
4.46489155e-01 3.00891340e-01 -5.61620593e-01 -2.32083291e-01
5.44904888e-01 9.53396317e-03 4.21373606e-01 -3.95648271e-01
-6.90129459e-01 1.00025475e+00 5.42318940e-01 -4.27506538e-03
-1.54686734e-01 5.31982243e-01 5.29360592e-01 3.24045151e-01
8.65830362e-01 1.76575527e-01 -6.14103973e-01 1.54088512e-01
-8.78576636e-01 7.14546219e-02 6.14050150e-01 4.63440686e-01
5.53245842e-01 -3.21363062e-01 -2.98235595e-01 1.24032164e+00
3.95970196e-01 -1.39237583e-01 4.03078556e-01 -1.38952434e+00
1.83243454e-01 5.55012226e-01 -2.53005862e-01 -9.84800696e-01
-6.05238855e-01 -5.54238558e-01 -1.22860765e+00 4.07997489e-01
7.00458646e-01 2.30391905e-01 -6.50407434e-01 2.34491158e+00
1.05017990e-01 1.75103590e-01 -3.87118787e-01 5.46241879e-01
8.84827316e-01 -1.46317994e-02 5.23688555e-01 -2.15190738e-01
1.39706290e+00 -5.13975561e-01 -4.79244918e-01 -4.41964641e-02
3.86971593e-01 5.83372526e-02 7.97408342e-01 6.17463052e-01
-8.39222372e-01 -4.73873317e-01 -1.08059871e+00 -5.46263009e-02
-3.24998528e-01 9.63353962e-02 1.12051177e+00 9.99426782e-01
-1.08550847e+00 8.40831220e-01 -4.35951024e-01 -3.04963347e-02
6.76272810e-01 8.25876117e-01 -5.38485706e-01 8.67067873e-02
-1.38349175e+00 7.39159644e-01 3.12729865e-01 2.98033990e-02
-3.16192716e-01 -7.37208903e-01 -8.63563240e-01 1.99531928e-01
3.19470987e-02 -6.75727129e-01 1.19810939e+00 -1.45560634e+00
-1.31547129e+00 1.15177310e+00 3.86488880e-03 -4.02357429e-01
7.56515741e-01 -1.94797471e-01 -5.95598668e-02 -2.34206185e-01
1.90302506e-01 9.76098299e-01 7.99855471e-01 -1.02618265e+00
-7.55261421e-01 -6.12116814e-01 1.89014867e-01 -9.11705121e-02
-6.95851564e-01 -3.33547336e-03 1.42490536e-01 -4.21984941e-01
7.92028457e-02 -7.52305508e-01 6.04377240e-02 1.91468745e-01
-2.35451639e-01 -9.45864141e-01 2.41255060e-01 -4.45264280e-01
1.22678030e+00 -1.91740966e+00 -3.95899937e-02 1.25000730e-01
4.02727515e-01 -4.82424885e-01 -1.96422059e-02 -2.39108220e-01
-5.04582226e-01 2.43509516e-01 -1.94514051e-01 -6.55276597e-01
3.87863815e-01 1.62840530e-01 -9.16325450e-02 5.45980513e-01
1.71227023e-01 5.54860473e-01 -6.26934707e-01 -8.43844354e-01
-2.78038867e-02 1.75212637e-01 -6.25426829e-01 5.44373170e-02
1.29621942e-02 9.46791768e-02 -2.88339883e-01 8.48682404e-01
6.26936376e-01 -2.16449425e-01 3.16404969e-01 -4.91629019e-02
1.56570345e-01 1.81308270e-01 -7.71586478e-01 1.16842830e+00
-5.24594307e-01 4.78259593e-01 -3.10110062e-01 -1.35724795e+00
9.66139495e-01 1.22236326e-01 7.55305588e-01 -6.45413637e-01
3.47457044e-02 3.72567624e-01 -9.02694277e-03 1.54789304e-02
1.67445391e-01 -2.29472429e-01 -3.71586084e-01 2.73851335e-01
5.06356843e-02 4.76455897e-01 4.38075699e-02 -3.71554494e-01
9.00111020e-01 1.48742929e-01 4.79867071e-01 -3.48336518e-01
5.34882307e-01 -4.61058587e-01 6.72811508e-01 8.90069425e-01
-3.86781096e-01 4.72108513e-01 8.90156507e-01 -7.32887566e-01
-9.26225483e-01 -1.12674963e+00 -8.23762655e-01 1.59168863e+00
-1.75797418e-01 3.26902829e-02 -6.21397257e-01 -1.04974449e+00
4.69995946e-01 3.92723083e-01 -1.07779765e+00 -1.35563210e-01
-3.67307872e-01 -9.99073267e-01 7.36941934e-01 6.23655677e-01
3.42113435e-01 -9.65255499e-01 -1.24238975e-01 -1.67579390e-02
-1.20316900e-01 -8.90193820e-01 1.14428990e-01 4.20107126e-01
-8.68914604e-01 -1.17181838e+00 -3.18953633e-01 -6.79126501e-01
5.41947544e-01 -6.25323176e-01 1.41411448e+00 5.73530570e-02
-2.21678498e-03 7.03464821e-02 -1.53099418e-01 -2.08861440e-01
-2.14438856e-01 1.93662405e-01 3.34342510e-01 1.72096953e-01
5.29120147e-01 -7.38936543e-01 -6.10630512e-01 1.44536853e-01
-4.86194313e-01 -2.19935384e-02 6.74035192e-01 1.13531446e+00
2.91914344e-01 -6.76009580e-02 1.22132146e+00 -9.79563296e-01
5.28645217e-01 -7.12433875e-01 -4.86763656e-01 1.77925810e-01
-1.18010819e+00 2.99284577e-01 6.35060608e-01 -4.91576850e-01
-8.32056999e-01 -4.74127606e-02 -8.90429914e-02 2.53304951e-02
-1.40947372e-01 5.09192705e-01 7.97876567e-02 1.51523009e-01
3.51104081e-01 -2.94398099e-01 -8.60950351e-02 -4.17796850e-01
2.52012193e-01 8.10991287e-01 4.36140984e-01 -8.88594508e-01
4.10905987e-01 2.00977057e-01 5.24526000e-01 -2.40440369e-01
-1.21678078e+00 -3.26365642e-02 -7.56468236e-01 -2.99157351e-01
7.51068294e-01 -7.93619454e-01 -1.17228889e+00 4.78328079e-01
-1.06871676e+00 -1.19051278e-01 1.49355322e-01 2.95266151e-01
-6.31733000e-01 2.13601485e-01 -6.75159812e-01 -9.95612800e-01
-3.28382283e-01 -1.04050279e+00 1.00053108e+00 3.19911093e-02
-4.10272211e-01 -8.86062860e-01 -2.27328673e-01 4.99948084e-01
1.60154000e-01 4.87825722e-01 1.46043146e+00 -9.01644170e-01
-1.30069584e-01 -2.62588501e-01 -7.58478463e-01 4.33243632e-01
2.13118214e-02 -7.51823187e-02 -1.03697777e+00 3.54203209e-02
-5.59309900e-01 -6.67594850e-01 1.14611042e+00 6.26170993e-01
1.94693244e+00 -4.38183755e-01 -3.54573652e-02 6.34363711e-01
1.21168363e+00 -1.87798411e-01 3.64587575e-01 4.21264440e-01
6.29702449e-01 1.04324245e+00 5.41422606e-01 5.42012453e-01
6.38723314e-01 6.81116641e-01 7.69755125e-01 4.73619252e-02
3.05968702e-01 8.77870321e-02 1.55628353e-01 3.52527857e-01
-2.15315744e-01 1.70977026e-01 -7.07010150e-01 2.75248885e-01
-1.89603567e+00 -8.69560957e-01 -3.54499221e-02 2.16385293e+00
1.17823517e+00 2.14210257e-01 1.73042670e-01 3.84226292e-01
7.96931148e-01 1.28697366e-01 -5.90456843e-01 -7.24401534e-01
2.25590721e-01 2.50032336e-01 4.99908209e-01 2.40281597e-01
-1.51941097e+00 4.75763649e-01 6.19303226e+00 9.13835406e-01
-9.10176218e-01 1.45100534e-01 1.35922217e+00 -2.39775218e-02
6.41578138e-02 -3.61680388e-01 -6.78709805e-01 5.57796717e-01
1.01911688e+00 1.13012128e-01 2.30221510e-01 1.18352485e+00
-2.20727697e-01 2.35876575e-01 -1.72252524e+00 9.52932715e-01
-1.02372557e-01 -8.42857540e-01 -2.67124534e-01 2.21181795e-01
5.49761534e-01 -3.33863974e-01 5.42880356e-01 6.55941784e-01
5.83621025e-01 -1.51327419e+00 7.51327753e-01 5.50999641e-01
1.27019525e+00 -9.52517927e-01 7.50441492e-01 1.58114627e-01
-8.85279059e-01 -7.10229576e-01 -3.03355604e-01 -3.85919601e-01
-4.68308479e-01 8.44099820e-01 -4.91087079e-01 2.09710851e-01
9.16041434e-01 7.09177852e-01 -7.66277909e-01 5.82401812e-01
9.92836431e-02 5.24332047e-01 -2.01651111e-01 1.41094159e-02
1.49240196e-01 -1.94057241e-01 -1.66873112e-01 9.33208048e-01
2.92917162e-01 -3.17131549e-01 -1.72473103e-01 5.90304017e-01
-5.49549162e-01 3.67278643e-02 -5.48864782e-01 2.56137222e-01
3.70294511e-01 1.23271883e+00 -3.22285593e-01 -5.26024848e-02
-4.16196823e-01 4.88781661e-01 7.56384134e-01 -8.26689452e-02
-9.91744637e-01 -1.77227997e-03 7.50582457e-01 1.56085268e-01
-2.55647123e-01 1.94285005e-01 -8.12703729e-01 -7.20404685e-01
-9.64651555e-02 -6.69653416e-01 6.62087619e-01 -4.04918849e-01
-1.96821320e+00 3.24921966e-01 3.02077204e-01 -9.92585361e-01
-4.36575413e-01 -9.70288575e-01 -2.47586593e-01 4.95832413e-01
-1.58690035e+00 -1.39211094e+00 -2.84838855e-01 1.08799219e-01
2.59864181e-01 -3.22820067e-01 7.48536050e-01 5.02950370e-01
-5.87194502e-01 1.11615026e+00 1.19903378e-01 2.20682025e-01
9.19513762e-01 -1.63239551e+00 -1.59241974e-01 1.02715902e-01
-2.03628182e-01 4.89894390e-01 4.53170031e-01 -2.77510703e-01
-6.16772652e-01 -1.04834151e+00 1.00122511e+00 -5.65710306e-01
6.64950788e-01 -2.35259220e-01 -6.37585998e-01 4.93771434e-01
-1.69563472e-01 1.97084084e-01 7.83104539e-01 7.27983952e-01
-9.05761898e-01 -6.58120275e-01 -1.20903563e+00 3.75245124e-01
1.16503644e+00 -4.55014855e-01 -3.96886230e-01 1.83799118e-01
5.48502564e-01 1.57393351e-01 -1.02220047e+00 1.07404149e+00
1.35689628e+00 -1.21490622e+00 1.22887838e+00 -6.92965508e-01
1.16401815e+00 3.87651324e-01 -2.04169437e-01 -8.86167943e-01
-4.40745145e-01 1.42597958e-01 -2.95670867e-01 1.23960567e+00
3.60333413e-01 -6.80961132e-01 8.59327972e-01 8.55471551e-01
1.03275992e-01 -1.10224640e+00 -1.38239610e+00 -7.02161372e-01
4.76243138e-01 -4.20646697e-01 5.96429825e-01 1.09259462e+00
-1.54303119e-01 1.43017739e-01 -5.19119680e-01 -1.35399982e-01
1.06365407e+00 -9.97959822e-02 2.63925672e-01 -2.18493080e+00
-2.12326944e-01 -8.92461240e-01 -5.77190101e-01 -3.14593136e-01
8.57540786e-01 -1.03082013e+00 -1.98678657e-01 -1.22056913e+00
5.63768148e-01 -7.42365718e-01 -9.79523778e-01 6.69218838e-01
-5.55696748e-02 4.65382516e-01 -2.71320343e-01 6.15411205e-03
-7.02231586e-01 4.69986141e-01 5.81195593e-01 -3.35318446e-01
2.34277382e-01 2.05400586e-01 -8.67198825e-01 1.01184690e+00
7.82983005e-01 -8.01386654e-01 -2.45815933e-01 -4.25207280e-02
4.78127629e-01 1.35493279e-01 4.17192370e-01 -9.26342249e-01
-2.37429533e-02 -2.98007369e-01 7.10240483e-01 -3.36308926e-01
3.76031011e-01 -9.75317895e-01 -5.86604141e-02 4.76525247e-01
-8.93736303e-01 -2.91805007e-02 -3.85758102e-01 5.63286185e-01
-1.79397956e-01 -2.47442737e-01 9.39784467e-01 2.48343140e-01
-4.88949418e-01 5.01335621e-01 2.24201143e-01 -1.07845277e-01
7.81370997e-01 -2.08304286e-01 -4.61809546e-01 -2.52721727e-01
-6.49934769e-01 2.80693155e-02 7.21314326e-02 4.23675776e-01
3.45083684e-01 -1.58690333e+00 -8.83276343e-01 -8.08403194e-02
2.56149292e-01 -4.39083993e-01 -5.76143041e-02 7.81963110e-01
-1.77417651e-01 3.70634079e-01 -3.35550547e-01 -5.76816440e-01
-1.38529134e+00 2.66822100e-01 4.15804923e-01 -6.95574045e-01
9.60545912e-02 6.72242105e-01 4.35861856e-01 -6.88354313e-01
4.72795725e-01 -1.54196680e-01 -6.44022524e-01 4.69561547e-01
1.36566177e-01 4.36958075e-01 -1.57272592e-02 -4.49635535e-01
-5.72226763e-01 4.59146887e-01 8.92019495e-02 9.82957892e-03
1.34106827e+00 1.07039072e-01 -4.88147974e-01 5.52116394e-01
1.15082133e+00 -4.87569809e-01 -9.93409753e-01 -1.35758802e-01
5.64242065e-01 -1.57522827e-01 -6.27196133e-02 -9.30410743e-01
-8.88398051e-01 8.19422722e-01 7.75376916e-01 1.71818405e-01
1.22305465e+00 -1.16323791e-01 2.03454211e-01 4.65211987e-01
3.97149622e-01 -1.07056963e+00 5.95859252e-02 3.94804388e-01
7.97873080e-01 -1.65377605e+00 1.97099879e-01 -2.90634811e-01
-2.68323779e-01 1.09757435e+00 8.74176502e-01 -1.44419931e-02
8.04973125e-01 -2.08481759e-01 -3.61980945e-01 2.26067498e-01
-9.34937119e-01 -5.82053922e-02 7.24780381e-01 5.19423306e-01
8.12369347e-01 3.65578860e-01 -7.53189445e-01 1.11008883e+00
-2.59774357e-01 -1.27130738e-02 1.74230054e-01 1.93360731e-01
-2.62672216e-01 -1.18644726e+00 -1.53699160e-01 1.01370084e+00
-8.62068772e-01 -1.57840580e-01 -3.33346605e-01 7.43782640e-01
3.12490493e-01 8.72741342e-01 3.36117685e-01 -4.29388642e-01
-6.51678070e-02 2.73054749e-01 5.31164944e-01 -3.56972307e-01
-4.79543239e-01 -4.90196109e-01 3.14189106e-01 -4.45579410e-01
-4.53661710e-01 -6.27801776e-01 -8.85012507e-01 -3.99606317e-01
-1.69131737e-02 -1.50309518e-01 7.45254099e-01 9.34998035e-01
-5.77914380e-02 3.95968169e-01 7.15803742e-01 -5.44789433e-01
-8.69705379e-01 -1.02560449e+00 -5.82037628e-01 5.58051348e-01
3.48154426e-01 -1.14546382e+00 -4.37757820e-01 -2.08330661e-01]
|
[9.177706718444824, 4.004767417907715]
|
78329304-3910-41f3-9417-c85f065387f1
|
temporal-variation-measure-analysis-an
|
2212.08369
| null |
https://arxiv.org/abs/2212.08369v1
|
https://arxiv.org/pdf/2212.08369v1.pdf
|
Temporal Variation Measure Analysis: An Improved Second-Order Difference Plot
|
In this study, an improved second-order difference plot is proposed to analyze the variability of heart rate variability. Although the variation of physiological status of cardiovascular system can be shown graphically by the second-order difference plot, the descriptive ability of existing indicators for this plot is insufficient. As a result, the physiological information contained in the second-order difference plot cannot be extracted adequately. Addressing the problem, the temporal variation measure analysis is presented to describe distribution patterns of scatter points in the second-order difference plot quantitatively and extract the acceleration information for variation of heart rate variability. Experiment results demonstrate the effectiveness of the temporal variation measure analysis. As a quantitative indicator, the temporal variation entropy is properly designed and successfully applied in the recognition and classification of the physiological statuses of the heart.
|
['Ning Cai', 'Chen Diao']
|
2022-12-16
| null | null | null | null |
['heart-rate-variability']
|
['medical']
|
[-2.37105876e-01 -4.19931859e-01 -1.13852561e-01 -2.90504456e-01
2.10714340e-01 -2.50279933e-01 -7.25991875e-02 2.82148659e-01
3.53625938e-02 6.34375036e-01 -7.73139969e-02 -4.66987997e-01
-5.33308864e-01 -4.28973377e-01 2.60246068e-01 -7.78465450e-01
-2.99859762e-01 -2.83354878e-01 -7.82794785e-03 -1.51619166e-01
3.52478266e-01 6.38733804e-01 -9.61939812e-01 -2.36969978e-01
9.81381416e-01 1.23645818e+00 -2.27433935e-01 6.26452565e-01
-9.60372686e-02 1.93919986e-01 -9.66582775e-01 4.96343851e-01
1.18053488e-01 -1.17979097e+00 -4.55230139e-02 -6.99438900e-02
-2.39928618e-01 -1.80293798e-01 -2.31485590e-01 8.94119263e-01
4.98144537e-01 -1.30120823e-02 8.86465490e-01 -1.29501474e+00
-2.91647017e-01 2.21252888e-01 -5.75642526e-01 8.01934481e-01
2.41108894e-01 2.69558989e-02 3.68717283e-01 -5.58354616e-01
1.57348216e-01 8.79835069e-01 6.33973420e-01 1.13763653e-01
-1.02424574e+00 -6.14562869e-01 -2.10758924e-01 -1.10724280e-02
-1.45345175e+00 1.35270953e-01 1.23959446e+00 -6.53012276e-01
1.24890842e-01 6.45385325e-01 1.23055637e+00 2.71551967e-01
7.92191327e-01 3.92565191e-01 1.35612583e+00 -2.10475966e-01
-1.83890373e-01 3.43745291e-01 5.65384328e-01 7.37599373e-01
3.71807754e-01 2.19010726e-01 -1.17083259e-01 -8.59300345e-02
9.15398419e-01 3.00558686e-01 -5.31555057e-01 -2.33872250e-01
-1.01905811e+00 4.06584591e-01 4.05626923e-01 8.59981418e-01
-1.68514892e-01 -3.01087558e-01 5.28412819e-01 2.87128597e-01
4.78837222e-01 3.47491562e-01 -2.95445383e-01 -4.12761688e-01
-9.86822546e-01 -1.52096242e-01 6.53110802e-01 3.70270163e-01
2.78107733e-01 3.78645867e-01 -3.01940769e-01 3.71045172e-01
4.56569761e-01 8.05735767e-01 6.85486495e-01 -6.36296570e-01
3.57916243e-02 8.22211444e-01 -9.53354016e-02 -1.51159072e+00
-8.37499380e-01 -4.53480601e-01 -1.19885457e+00 1.81718722e-01
6.15302861e-01 -2.66100466e-01 -3.90361607e-01 1.14217734e+00
4.73610401e-01 -2.19229862e-01 5.84946945e-02 9.88609314e-01
7.99001098e-01 6.28975689e-01 6.86310977e-02 -7.92824745e-01
1.52123499e+00 -1.57722667e-01 -1.18776035e+00 4.14583653e-01
1.42920002e-01 -2.97912925e-01 7.72583663e-01 -1.06108204e-01
-6.00213170e-01 -7.86196172e-01 -1.16525447e+00 6.56353772e-01
-7.51275569e-02 2.32491121e-01 1.90999717e-01 6.02044761e-01
-3.26547325e-01 9.19312477e-01 -8.54235530e-01 -2.07355917e-01
1.04433380e-01 -4.93817538e-01 -1.19589388e-01 6.00439131e-01
-1.40419531e+00 7.79718518e-01 3.70332241e-01 5.01914382e-01
1.41651884e-01 -9.71932232e-01 -7.75260687e-01 2.77786285e-01
-4.65655595e-01 -1.88056603e-01 6.00197017e-01 -3.85389775e-01
-1.43023276e+00 3.19457352e-01 -6.21291064e-02 -5.24956696e-02
6.82278693e-01 7.78065175e-02 -8.88960838e-01 3.52157801e-01
-2.88560539e-01 -3.61652613e-01 6.18811667e-01 -1.02492595e+00
-9.54922438e-02 -4.72458959e-01 -8.61165583e-01 6.35828124e-03
-1.48230568e-01 -3.66398841e-01 2.64306545e-01 -4.76200581e-01
5.72663128e-01 -6.58666730e-01 4.95056733e-02 5.78597933e-02
-1.31029114e-01 8.42362046e-02 9.92607713e-01 -8.60779762e-01
2.06440496e+00 -2.55116940e+00 -5.09468853e-01 4.88032013e-01
4.82619226e-01 4.44414586e-01 6.74235523e-01 3.50580096e-01
-1.10748217e-01 3.53019804e-01 -3.66946846e-01 6.78986967e-01
-2.43183777e-01 -1.11771777e-01 -6.59911633e-02 6.66542709e-01
1.77899480e-01 4.84849185e-01 -8.53862107e-01 -5.27354300e-01
3.74117374e-01 4.43414629e-01 -5.94845158e-04 1.78380817e-01
6.70242012e-01 7.20191360e-01 -6.23495460e-01 3.50918263e-01
7.90546417e-01 -9.18933526e-02 2.40802124e-01 -5.40434003e-01
-3.18572193e-01 -1.16402090e-01 -9.52894568e-01 8.25768709e-01
3.50289121e-02 1.08747351e+00 -5.33400238e-01 -9.50131953e-01
1.58454049e+00 4.61328745e-01 9.50541496e-01 -7.77765930e-01
1.58830896e-01 1.35889202e-01 3.87121528e-01 -8.88176560e-01
-7.46833459e-02 -4.64150399e-01 -2.02465672e-02 1.62458241e-01
-4.46628183e-01 3.57331000e-02 3.44857201e-02 -1.72944292e-01
4.81969059e-01 -7.74077177e-02 8.26167643e-01 -6.27473712e-01
6.49572611e-01 -3.02873522e-01 6.16490066e-01 2.41425008e-01
-8.02559912e-01 2.75199473e-01 8.92397881e-01 -7.02394605e-01
-7.86553681e-01 -9.99437749e-01 -6.99938238e-01 -9.66201574e-02
3.43387276e-01 -2.45229617e-01 -4.25383747e-01 -4.22286302e-01
2.71724224e-01 4.04089749e-01 -5.52853465e-01 -5.34911811e-01
-4.84973997e-01 -7.67964840e-01 5.92917442e-01 5.61972380e-01
8.30191195e-01 -7.22080529e-01 -1.19708800e+00 1.43857718e-01
-1.11621208e-01 -6.53296292e-01 -4.02830899e-01 -9.16628242e-02
-1.19784725e+00 -1.31920671e+00 -5.95747411e-01 -3.35947216e-01
4.52216864e-01 -4.21961024e-02 7.76394844e-01 2.42876704e-04
-6.89718902e-01 1.81764737e-01 -8.74416009e-02 -5.35199106e-01
-3.48702103e-01 -5.01846671e-01 -2.58060638e-02 -1.49545208e-01
2.19007149e-01 -5.82750916e-01 -9.33911860e-01 6.11998260e-01
-4.65354145e-01 -5.34425080e-01 2.97974259e-01 6.34578645e-01
3.09880614e-01 2.50366986e-01 8.07718515e-01 -3.73388588e-01
8.19324732e-01 -4.55777287e-01 -5.56053400e-01 5.01667382e-03
-1.12759650e+00 7.85812810e-02 6.03755176e-01 -4.66543823e-01
-8.17246497e-01 -4.12610114e-01 4.05602098e-01 -3.90798092e-01
1.98355149e-02 5.62832475e-01 1.45075068e-01 2.85132140e-01
7.40166426e-01 4.98887867e-01 7.16465652e-01 -2.46142432e-01
-1.30638495e-01 6.40508711e-01 4.17398989e-01 -9.83315483e-02
5.08740425e-01 5.18364180e-03 5.17078996e-01 -9.76290405e-01
-3.19007963e-01 -6.02884412e-01 -4.85471398e-01 -6.73904836e-01
8.79132569e-01 -4.68562484e-01 -1.18851066e+00 3.12917113e-01
-7.65084565e-01 2.72562075e-02 -2.92836636e-01 8.76480758e-01
-3.43207270e-01 6.73371732e-01 -5.03942013e-01 -1.35101128e+00
-3.88074666e-01 -4.77513045e-01 5.27723849e-01 4.90646839e-01
-3.25094521e-01 -1.28618610e+00 2.13947862e-01 -1.57481194e-01
4.04257536e-01 8.94685686e-01 9.29461956e-01 -3.00059468e-01
1.34196773e-01 -5.02113938e-01 8.64652321e-02 3.08610141e-01
4.40157264e-01 5.70816159e-01 -3.95045906e-01 -1.09911002e-02
4.88424987e-01 5.43215990e-01 3.31521481e-01 6.78478956e-01
9.33847487e-01 -4.07166183e-01 -3.47239554e-01 3.86298984e-01
1.41037083e+00 9.21488225e-01 7.31323719e-01 -2.96814591e-02
3.66231024e-01 4.82721090e-01 5.39478779e-01 7.04320669e-01
8.98253731e-03 3.50770503e-01 8.26534722e-03 -2.69771904e-01
2.33477592e-01 -6.50492534e-02 2.27452308e-01 9.69353914e-01
-2.83059627e-01 1.22873247e-01 -8.43679011e-01 -3.21125076e-03
-1.48183870e+00 -1.30811012e+00 -6.58958018e-01 2.32914901e+00
5.63706040e-01 -7.57960901e-02 3.86945277e-01 4.81522679e-01
9.14748490e-01 2.16316119e-01 -4.10121441e-01 -4.89379108e-01
1.53439969e-01 -4.20282692e-01 1.50220290e-01 3.03919107e-01
-7.82417715e-01 -1.17679343e-01 7.29543686e+00 2.57958680e-01
-1.41812539e+00 -5.78943670e-01 7.14444458e-01 3.80822390e-01
-7.16331750e-02 -1.28007650e-01 -3.40227604e-01 9.35624361e-01
8.90885353e-01 -5.53639889e-01 -8.19887891e-02 5.97302079e-01
4.79382843e-01 -2.09986076e-01 -6.60478711e-01 1.39230621e+00
-1.97492972e-01 -6.27265513e-01 -3.98211777e-01 1.59959923e-02
2.60799587e-01 -7.91740716e-01 5.28409295e-02 1.03669949e-01
-7.56013989e-01 -7.15847611e-01 1.31628990e-01 1.07885849e+00
7.46727467e-01 -4.68850553e-01 7.53934979e-01 2.70455331e-01
-1.37076628e+00 -1.23904645e-01 -1.91113055e-01 -1.62964135e-01
1.16981968e-01 1.13242006e+00 -7.24788368e-01 6.84709132e-01
2.87080735e-01 6.11481309e-01 -3.31889272e-01 1.15068185e+00
1.03252262e-01 7.18419790e-01 -2.45477185e-01 -3.27043325e-01
-2.79777825e-01 -7.17017412e-01 7.38592565e-01 9.42868054e-01
5.52296877e-01 3.94791812e-01 -8.99003074e-02 9.32019174e-01
8.24308932e-01 2.66785979e-01 -6.00535810e-01 -2.98727483e-01
3.77418607e-01 1.13635838e+00 -7.71219313e-01 -3.79590601e-01
-2.71362979e-02 3.90256137e-01 -5.77767968e-01 2.49420017e-01
-9.17454243e-01 -7.68808544e-01 5.17949104e-01 2.77080983e-01
-1.40814573e-01 -2.80750215e-01 -7.12082863e-01 -8.98109734e-01
3.35218817e-01 -3.38209748e-01 4.64594215e-01 -5.09203553e-01
-1.05280304e+00 6.73862040e-01 3.11214000e-01 -1.69612527e+00
-2.37104952e-01 -2.97612697e-01 -9.59999681e-01 1.07495749e+00
-8.57670903e-01 -2.13393457e-02 -5.76208353e-01 3.86637896e-01
1.08342342e-01 7.08179697e-02 7.77095258e-01 2.15791941e-01
-7.97261119e-01 3.91372591e-01 1.33558318e-01 4.05987017e-02
2.81723827e-01 -1.34570062e+00 -2.99334854e-01 5.57860136e-01
-4.48445141e-01 5.33355832e-01 9.04181361e-01 -6.80568933e-01
-7.87033856e-01 -4.59222734e-01 6.71633422e-01 -1.16834104e-01
4.52839375e-01 6.10807501e-02 -9.69147503e-01 -1.59876525e-01
-2.08967596e-01 3.92520845e-01 7.95262575e-01 -3.08915615e-01
1.08683951e-01 -4.47338998e-01 -1.21746314e+00 3.60638559e-01
3.41365963e-01 -2.96853960e-01 -6.68017447e-01 -1.30904093e-01
1.23054676e-01 -1.18799530e-01 -1.49677408e+00 6.84971154e-01
9.48844016e-01 -8.22005808e-01 6.79522812e-01 -4.63327616e-01
2.53972232e-01 -4.89386141e-01 1.58743545e-01 -1.41204464e+00
-3.54733258e-01 -5.86374164e-01 -1.21519528e-01 1.12093759e+00
2.58559525e-01 -1.04854882e+00 2.95419484e-01 4.98512059e-01
2.86080688e-01 -9.28973496e-01 -7.29710579e-01 -9.92451370e-01
-2.17480198e-01 1.84689879e-01 2.02916458e-01 8.66116166e-01
8.10329556e-01 2.56992102e-01 -1.50414005e-01 -1.96061015e-01
3.85138541e-01 8.51336941e-02 4.53957677e-01 -1.31155241e+00
5.93630336e-02 -5.67747533e-01 -8.75285029e-01 -6.09101236e-01
-5.08264065e-01 -4.22480404e-01 -4.42539379e-02 -1.50519383e+00
-1.08640091e-02 -2.98776150e-01 -7.29491234e-01 -1.86622500e-01
-6.03896201e-01 -1.90596819e-01 2.84152001e-01 3.78519297e-01
3.54577571e-01 5.96000493e-01 1.56332147e+00 1.24614164e-01
-6.08674645e-01 1.46594867e-01 -2.59727091e-01 3.67194921e-01
1.09967172e+00 -3.11547905e-01 -7.05610812e-01 7.11835980e-01
-2.86254942e-01 6.84980512e-01 2.21937627e-01 -1.23436546e+00
-3.48878324e-01 -1.18071824e-01 6.74655199e-01 -6.22014940e-01
-1.79151148e-01 -8.38733017e-01 3.18552643e-01 9.62376535e-01
-1.90970659e-01 3.02177727e-01 1.78578198e-01 5.23871958e-01
-4.88367319e-01 2.14315027e-01 7.95892835e-01 2.15941593e-01
-1.29628643e-01 3.26871797e-02 -4.74376619e-01 8.44187215e-02
1.00234556e+00 -6.61891758e-01 -3.50126594e-01 -2.17499018e-01
-6.92394555e-01 -3.72689590e-02 -7.19234943e-02 -6.00596741e-02
5.90886831e-01 -1.61240613e+00 -4.90163177e-01 2.67617941e-01
-8.78928415e-03 -5.26511252e-01 8.59895766e-01 1.41226375e+00
-6.87144339e-01 4.06034768e-01 -5.96870244e-01 -7.37164021e-01
-1.12991202e+00 7.10505784e-01 5.97332597e-01 -2.35535860e-01
-8.24982882e-01 2.60272082e-02 1.39801979e-01 2.27334529e-01
-2.49648556e-01 -5.62708676e-01 -6.95764899e-01 1.47280931e-01
4.92408067e-01 6.76823080e-01 -4.24482435e-01 -3.14915836e-01
-5.68287849e-01 7.99805760e-01 6.95094168e-01 5.06908149e-02
7.35931337e-01 -3.59073400e-01 -3.24334241e-02 1.13368034e+00
1.39979529e+00 -1.03578210e-01 -1.26443231e+00 3.44476759e-01
-1.22010961e-01 -4.38462198e-01 -4.35755908e-01 -7.51728356e-01
-1.13768053e+00 9.20862138e-01 9.97514665e-01 8.78535807e-01
1.34936976e+00 -5.94285131e-01 5.70388734e-01 -1.77874729e-01
-1.79709464e-01 -8.57903004e-01 1.01408437e-01 3.59325036e-02
8.46735418e-01 -9.50963259e-01 1.91752940e-01 -5.55027604e-01
-9.09271240e-01 1.55760503e+00 4.25468445e-01 -1.31877929e-01
1.19764805e+00 1.93113029e-01 3.95020425e-01 -2.06028596e-01
-2.62279391e-01 2.60645807e-01 5.26277184e-01 5.22571325e-01
6.67613685e-01 2.47272983e-01 -1.20633304e+00 2.72688538e-01
-1.02005601e-01 -2.07572743e-01 3.58565003e-01 6.69036508e-01
-4.77471173e-01 -2.95980752e-01 -2.89690793e-01 2.67539144e-01
-5.38057745e-01 3.09158742e-01 -2.22460423e-02 9.62380230e-01
-3.01719397e-01 9.46088374e-01 1.90169320e-01 -4.19863760e-01
5.16146362e-01 3.88209105e-01 2.10198328e-01 1.28807798e-01
-2.68521428e-01 1.23327315e-01 -2.75680721e-01 -4.57124382e-01
-2.02135697e-01 -2.70848215e-01 -1.37453532e+00 -2.47298442e-02
-2.50432491e-01 3.85175616e-01 8.35410893e-01 6.83310390e-01
1.84385866e-01 8.76807868e-01 1.00873196e+00 -1.52467459e-01
-5.00637233e-01 -1.04657793e+00 -1.02653527e+00 4.01146531e-01
6.03932619e-01 -6.53046906e-01 -8.02436948e-01 -9.82619822e-02]
|
[14.02756404876709, 3.0535356998443604]
|
a72eb98b-798b-4ecb-8982-9b9c65c2b395
|
improving-arabic-diacritization-through
| null | null |
https://aclanthology.org/D15-1152
|
https://aclanthology.org/D15-1152.pdf
|
Improving Arabic Diacritization through Syntactic Analysis
| null |
['Salam Khalifa', 'Anas Shahrour', 'Nizar Habash']
|
2015-09-01
| null | null | null |
emnlp-2015-9
|
['morphological-tagging']
|
['natural-language-processing']
|
[-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]
|
[-7.2474775314331055, 3.6305480003356934]
|
86dfc48c-b248-4f1d-bfd7-c6aff9c0efbd
|
multi-task-deep-neural-networks-in-automated
|
1705.04802
| null |
http://arxiv.org/abs/1705.04802v2
|
http://arxiv.org/pdf/1705.04802v2.pdf
|
Multi-task Deep Neural Networks in Automated Protein Function Prediction
|
In recent years, deep learning algorithms have outperformed the state-of-the
art methods in several areas thanks to the efficient methods for training and
for preventing overfitting, advancement in computer hardware, the availability
of vast amount data. The high performance of multi-task deep neural networks in
drug discovery has attracted the attention to deep learning algorithms in
bioinformatics area. Here, we proposed a hierarchical multi-task deep neural
network architecture based on Gene Ontology (GO) terms as a solution to protein
function prediction problem and investigated various aspects of the proposed
architecture by performing several experiments. First, we showed that there is
a positive correlation between performance of the system and the size of
training datasets. Second, we investigated whether the level of GO terms on GO
hierarchy related to their performance. We showed that there is no relation
between the depth of GO terms on GO hierarchy and their performance. In
addition, we included all annotations to the training of a set of GO terms to
investigate whether including noisy data to the training datasets change the
performance of the system. The results showed that including less reliable
annotations in training of deep neural networks increased the performance of
the low performed GO terms, significantly. We evaluated the performance of the
system using hierarchical evaluation method. Mathews correlation coefficient
was calculated as 0.75, 0.49 and 0.63 for molecular function, biological
process and cellular component categories, respectively. We showed that deep
learning algorithms have a great potential in protein function prediction area.
We plan to further improve the DEEPred by including other types of annotations
from various biological data sources. We plan to construct DEEPred as an open
access online tool.
|
[]
|
2017-05-28
| null | null | null | null |
['protein-function-prediction']
|
['medical']
|
[-1.82843357e-01 -1.23603776e-01 1.60111949e-01 -4.75955874e-01
-3.64495575e-01 -2.14452431e-01 -8.32602680e-02 3.36398035e-01
-3.80953014e-01 1.16317952e+00 -5.41678891e-02 -2.92009741e-01
-3.74533594e-01 -9.96577561e-01 -9.53674495e-01 -1.08615756e+00
-2.06527710e-01 4.70377803e-01 3.39273006e-01 -3.52411062e-01
8.10879394e-02 3.97947669e-01 -1.64034545e+00 4.89193976e-01
5.88171363e-01 9.85287905e-01 2.12851241e-01 4.33152884e-01
-1.77791715e-01 3.99793059e-01 -6.49767935e-01 1.11267418e-01
-3.45635153e-02 -2.12729827e-01 -9.69561636e-01 -4.35702831e-01
-5.69181815e-02 1.06097512e-01 8.48110914e-02 9.56688464e-01
9.75180805e-01 9.78595272e-05 5.28321445e-01 -1.04524648e+00
-6.89018309e-01 3.47336948e-01 -1.98285565e-01 1.95633486e-01
1.50456250e-01 1.04059298e-02 9.17106211e-01 -6.71142280e-01
5.97355485e-01 1.15759540e+00 8.80570173e-01 2.62435764e-01
-1.05281079e+00 -6.34019494e-01 -3.34304571e-01 3.09127212e-01
-1.45168924e+00 -1.18263930e-01 2.92587489e-01 -5.15829682e-01
1.44625449e+00 -2.35554297e-02 3.60035330e-01 7.25817680e-01
5.99802196e-01 1.43653199e-01 8.77342463e-01 -4.29168880e-01
6.82626804e-03 -1.39813200e-01 3.69870722e-01 1.02992725e+00
3.34629953e-01 -1.15728550e-01 -1.46190941e-01 -3.11657488e-01
3.35720956e-01 -2.38375962e-02 -1.43273234e-01 1.19163901e-01
-7.62726843e-01 8.80552888e-01 4.84147578e-01 9.20250475e-01
-4.54585046e-01 1.45771384e-01 6.97138608e-01 2.93561518e-01
3.62970114e-01 5.12306690e-01 -1.04700756e+00 3.48934174e-01
-4.06996846e-01 -6.93095801e-03 7.53329039e-01 4.33520406e-01
7.73945034e-01 -8.97263810e-02 -1.16059870e-01 1.00357389e+00
2.08373532e-01 -3.35912943e-01 1.02275789e+00 -4.08580482e-01
-2.68502414e-01 9.90503550e-01 -3.29623431e-01 -9.79661644e-01
-8.59971464e-01 -5.16331375e-01 -9.60706234e-01 -4.09853794e-02
4.83663589e-01 -1.54987067e-01 -9.41074967e-01 1.63341856e+00
1.91392690e-01 -5.39928637e-02 1.23796379e-02 7.76280224e-01
1.14442909e+00 5.41399837e-01 3.81379515e-01 8.17093924e-02
1.63780057e+00 -5.57533562e-01 -7.90813029e-01 4.48643208e-01
1.16326320e+00 -5.81332684e-01 1.05523503e+00 5.37196755e-01
-5.20998597e-01 -7.11656868e-01 -1.08875000e+00 -2.71810353e-01
-9.82033253e-01 1.68414429e-01 9.16611016e-01 6.07148528e-01
-1.03409052e+00 9.90071058e-01 -5.95032275e-01 -5.11829078e-01
4.49149966e-01 9.02012229e-01 -6.24008715e-01 -5.01914248e-02
-1.65449214e+00 8.14820945e-01 8.40609133e-01 -2.61923559e-02
-9.09838855e-01 -3.46136898e-01 -4.66843307e-01 3.95609170e-01
-4.16254513e-02 -8.23812008e-01 6.19947612e-01 -8.15596104e-01
-1.14329660e+00 9.36913729e-01 3.75160649e-02 -2.75048405e-01
-5.43411672e-02 -4.25392622e-03 -3.42539817e-01 -7.30335563e-02
6.38803989e-02 7.00968266e-01 -2.93170661e-01 -6.29822731e-01
-6.24615014e-01 -6.17984951e-01 -1.39872372e-01 -1.24354742e-01
-4.22584236e-01 -8.13547373e-02 -1.64309978e-01 -1.72623292e-01
4.78292592e-02 -8.42853725e-01 -2.25198865e-01 -3.89558971e-01
-2.35041365e-01 -6.83207870e-01 6.14554346e-01 -5.77694237e-01
1.01752889e+00 -1.84732735e+00 -7.24681094e-02 7.59656206e-02
1.41649768e-01 2.72364229e-01 -1.42392054e-01 3.16540956e-01
-4.77313310e-01 4.82815236e-01 2.80733734e-01 4.11926657e-01
-3.01275611e-01 3.27269316e-01 4.32184428e-01 4.18791980e-01
9.47299153e-02 7.34321833e-01 -4.76124614e-01 -3.10056120e-01
-1.72044173e-01 5.94061673e-01 -4.29252982e-01 6.61395267e-02
-2.57112294e-01 3.20905119e-01 -3.86813700e-01 8.79300475e-01
5.80892742e-01 -4.23978239e-01 4.96895015e-02 -3.27662021e-01
2.56144665e-02 1.80976942e-01 -6.88132823e-01 1.43070698e+00
-1.61400661e-02 3.02045643e-01 -2.67616898e-01 -1.44620621e+00
1.23091924e+00 6.70656145e-01 6.98133767e-01 -5.78195274e-01
3.57848614e-01 2.50208646e-01 5.43254614e-01 -7.40544140e-01
-6.45238385e-02 -2.05652177e-01 3.16298008e-01 -2.66216487e-01
4.21339631e-01 5.46692073e-01 2.79234201e-01 -3.83542120e-01
1.30781758e+00 1.07037477e-01 4.85498637e-01 -6.09059393e-01
6.52887583e-01 1.65402636e-01 6.44833267e-01 5.03681064e-01
-2.53075331e-01 1.99591681e-01 7.14006066e-01 -9.46266294e-01
-9.39631820e-01 -1.84726641e-01 -3.75607610e-01 1.45028150e+00
-3.59203815e-01 -2.04460815e-01 -6.55495346e-01 -3.79520833e-01
-1.40802383e-01 1.33200943e-01 -5.72080970e-01 -2.42352918e-01
-2.13445827e-01 -1.58720446e+00 9.68618572e-01 9.10153762e-02
6.63021326e-01 -8.53974521e-01 -7.40364343e-02 3.72616142e-01
-4.93172444e-02 -8.84607971e-01 2.95194298e-01 7.58556783e-01
-1.09951854e+00 -1.17066038e+00 -5.53276360e-01 -1.04369116e+00
3.53721827e-01 -1.69869661e-01 9.18460786e-01 3.89957845e-01
-3.33887607e-01 -6.27275467e-01 -3.17776173e-01 -6.97644055e-01
-3.23086977e-01 2.97033727e-01 -5.86515013e-03 -4.88497853e-01
7.12701917e-01 -5.78246653e-01 -4.59633321e-01 4.50642258e-01
-9.52270150e-01 -1.62243769e-01 6.66369915e-01 1.02747679e+00
7.04782367e-01 3.15736860e-01 9.07195210e-01 -8.98152471e-01
8.03550422e-01 -5.40266752e-01 -5.35261035e-01 1.90520987e-01
-6.61854804e-01 4.69564080e-01 5.00804722e-01 -1.47754937e-01
-5.28452396e-01 3.23767751e-01 -7.58490682e-01 3.34796049e-02
-3.83501887e-01 8.02010655e-01 -3.17511380e-01 -1.95588857e-01
7.25254714e-01 -4.36100326e-02 -6.19635843e-02 -4.83994991e-01
-3.33511144e-01 6.80310845e-01 -1.02285229e-01 -5.29200852e-01
-2.38954604e-01 -7.67666427e-03 5.80780745e-01 -5.64905941e-01
-6.57621682e-01 -5.83388031e-01 -4.49152201e-01 1.45205691e-01
9.81909692e-01 -7.63458312e-01 -1.24054003e+00 3.97138417e-01
-1.15591240e+00 4.79439460e-02 6.71094418e-01 4.23173636e-01
-1.57237023e-01 3.27766329e-01 -7.14358985e-01 -3.95965815e-01
-7.25305498e-01 -1.34807909e+00 8.03242564e-01 -2.62189396e-02
4.91583981e-02 -9.61538732e-01 9.46885422e-02 3.84796977e-01
3.06887984e-01 4.06368047e-01 1.36281037e+00 -1.18556106e+00
-1.53162435e-01 -2.31179148e-01 -2.65086710e-01 2.69530594e-01
1.86270088e-01 -1.36067763e-01 -1.09535849e+00 7.46098068e-03
-9.57518891e-02 -3.26920986e-01 9.72116470e-01 5.24002492e-01
1.47582972e+00 -1.05783045e-01 -5.36197484e-01 6.02365613e-01
1.66846669e+00 5.45455396e-01 9.46630478e-01 5.56645870e-01
5.32339573e-01 5.84117651e-01 3.59700620e-01 -3.37915099e-03
-8.83745998e-02 5.34724653e-01 5.53293169e-01 -2.65565544e-01
1.75743595e-01 2.94846326e-01 1.67961884e-02 4.41886455e-01
-3.25559765e-01 -4.55598801e-01 -1.21734560e+00 3.84049773e-01
-1.78375340e+00 -5.41639924e-01 -6.64320827e-01 1.87779415e+00
8.48789573e-01 5.34638613e-02 -1.42598331e-01 1.50809959e-01
6.11723781e-01 -7.30128706e-01 -3.42463493e-01 -5.67544222e-01
-2.43529290e-01 2.82393515e-01 5.58999896e-01 1.74575746e-01
-1.13316953e+00 9.19306219e-01 6.16898489e+00 9.25079048e-01
-1.17787039e+00 1.24843962e-01 9.26280439e-01 2.34004587e-01
4.41121966e-01 -2.91978687e-01 -9.89622414e-01 4.16805595e-01
1.35790431e+00 1.14604019e-01 -5.63367978e-02 8.45694005e-01
3.86676997e-01 -1.62937924e-01 -9.29453135e-01 4.77333426e-01
-3.40230912e-01 -1.37807214e+00 2.26805080e-02 3.10736686e-01
4.12907183e-01 3.49509150e-01 -5.59643805e-01 4.59371954e-01
3.40837300e-01 -1.20475781e+00 -3.69240105e-01 4.05987978e-01
3.60313058e-01 -6.53205633e-01 1.58277154e+00 3.11901122e-01
-8.42517555e-01 8.01925287e-02 -6.18092120e-01 -1.65759742e-01
-3.67804259e-01 8.10873508e-01 -1.17085469e+00 6.80187047e-01
8.92022729e-01 2.56351262e-01 -6.27502620e-01 1.08179593e+00
1.92780271e-01 5.71555555e-01 -2.73932278e-01 -2.16994941e-01
3.14779311e-01 6.85135871e-02 -8.48964751e-02 1.20856798e+00
3.77844542e-01 1.27451450e-01 2.63767809e-01 5.11333585e-01
-6.99287057e-02 4.76417750e-01 -5.42924404e-01 -1.58626869e-01
-4.28050272e-02 1.13278341e+00 -7.48392761e-01 -3.66252661e-01
-3.53677332e-01 5.73545396e-01 2.55219847e-01 9.27152783e-02
-7.82944143e-01 -6.41211510e-01 5.55213928e-01 9.80070382e-02
-7.01171607e-02 1.61549047e-01 -4.00086939e-01 -6.24581873e-01
-4.70416427e-01 -6.53117239e-01 7.50755787e-01 -8.41916323e-01
-1.16021323e+00 6.76654160e-01 -3.94223928e-01 -7.87198007e-01
1.86259165e-01 -1.01806998e+00 -1.84393600e-01 1.02802563e+00
-1.33752728e+00 -9.23698783e-01 -2.15227082e-01 3.44997615e-01
2.54761040e-01 -3.43501508e-01 1.29662466e+00 7.54209220e-01
-7.19948947e-01 3.87088299e-01 2.76519656e-01 8.19728747e-02
8.03651214e-01 -9.94400442e-01 -1.71424210e-01 2.83682764e-01
-3.80213201e-01 7.46629179e-01 6.93716645e-01 -5.63372195e-01
-9.32046056e-01 -1.08119857e+00 9.34220970e-01 5.37379086e-02
4.74667430e-01 -5.85054681e-02 -1.21703780e+00 3.61422956e-01
4.46866453e-02 -2.94630211e-02 1.20402324e+00 3.26399624e-01
5.64149320e-02 -4.18258868e-02 -1.25800920e+00 1.73129942e-02
6.39651895e-01 1.66728497e-02 -3.10218781e-01 5.70892811e-01
8.24455440e-01 -8.31275955e-02 -1.08909655e+00 6.62511170e-01
6.20609283e-01 -7.42967427e-01 8.62106264e-01 -8.59316885e-01
5.50691068e-01 -2.66521811e-01 -2.01101884e-01 -1.14842796e+00
-6.98310494e-01 2.50062585e-01 3.20511639e-01 8.95033181e-01
6.44608200e-01 -7.00500131e-01 7.32273936e-01 2.95856386e-01
-4.55600977e-01 -1.11503875e+00 -7.31868386e-01 -4.76233989e-01
1.71483025e-01 6.86809272e-02 3.84543866e-01 1.01161873e+00
-1.39814705e-01 6.03993893e-01 -1.78804263e-01 8.02131668e-02
-5.40126823e-02 -2.28673860e-01 4.06584442e-01 -1.61998498e+00
-2.90785998e-01 -1.95344552e-01 -7.19816089e-01 -3.48577082e-01
3.44707593e-02 -1.07902157e+00 -1.94664404e-01 -1.73536241e+00
1.27276435e-01 -8.23653564e-02 -7.67050445e-01 1.01053476e+00
-1.50100216e-01 2.31958926e-01 -5.53631127e-01 1.21645495e-01
-4.27852333e-01 1.87377796e-01 1.10627091e+00 1.27174050e-01
-1.43448636e-01 -1.92688182e-01 -8.24099839e-01 6.05388224e-01
1.00384653e+00 -6.88064277e-01 -5.20440377e-02 -2.95689821e-01
2.70019144e-01 -2.30757549e-01 5.14941886e-02 -1.11282265e+00
1.80365399e-01 1.23284988e-01 6.37524903e-01 -4.50512618e-01
9.68661979e-02 -7.39787579e-01 3.01162839e-01 8.35331440e-01
-1.87119469e-01 -4.05491069e-02 4.85841006e-01 2.90352553e-01
-3.18766892e-01 -3.09631944e-01 7.51177609e-01 -4.85383719e-01
-6.15066826e-01 5.49985953e-02 -5.18133521e-01 -6.76049232e-01
8.36412370e-01 -3.34913403e-01 -1.54355675e-01 3.01476181e-01
-9.95007038e-01 1.87428281e-01 -5.22038713e-02 2.43275851e-01
1.48847878e-01 -1.07755351e+00 -4.39640194e-01 3.48951481e-02
1.15236215e-01 -1.34440437e-01 1.04609348e-01 8.33206236e-01
-8.02177966e-01 9.28870738e-01 -6.61707163e-01 -5.72689772e-01
-1.69489312e+00 3.90868872e-01 6.47363305e-01 -4.62128907e-01
7.67828599e-02 7.51597524e-01 2.38343194e-01 -5.69920123e-01
2.39460915e-01 -5.13177156e-01 -5.31015635e-01 -4.08348553e-02
2.52004206e-01 2.80867964e-01 6.98919058e-01 -3.30302626e-01
-4.73275930e-01 3.83286834e-01 -8.10644999e-02 5.29354334e-01
1.68897712e+00 2.22892091e-01 -5.81340432e-01 2.91368663e-01
1.34712744e+00 -5.90262294e-01 -2.86003202e-01 2.15406045e-01
3.36790264e-01 1.12884961e-01 2.67970234e-01 -1.16071439e+00
-9.05153811e-01 7.27913558e-01 1.00258517e+00 1.29795656e-01
1.17795050e+00 -2.14141876e-01 4.90371436e-01 7.96353519e-01
2.85979569e-01 -8.72449756e-01 -2.22674713e-01 8.23772490e-01
4.96989459e-01 -1.34742272e+00 -9.30825695e-02 -3.82358462e-01
-8.62864926e-02 1.37253821e+00 6.76400542e-01 8.85977969e-02
6.27199709e-01 1.40617669e-01 -2.60920972e-01 -5.81495166e-01
-9.18655634e-01 -1.20367631e-01 2.08382100e-01 4.22139853e-01
1.25653839e+00 -4.81452644e-02 -9.09225762e-01 8.39628339e-01
4.74753790e-02 6.54854238e-01 2.15060100e-01 5.41643083e-01
-8.69758427e-01 -1.24416637e+00 -2.82457769e-01 4.52263921e-01
-1.01596868e+00 -2.89352983e-01 -4.29905176e-01 6.63483202e-01
6.37284219e-01 8.89587224e-01 -2.22289547e-01 -4.11409199e-01
2.05169633e-01 5.23677051e-01 1.52100921e-01 -6.15549982e-01
-6.38395190e-01 1.12693705e-01 2.65852809e-01 -3.97072852e-01
-2.45896056e-01 -5.88082708e-02 -1.59902418e+00 -9.69891101e-02
-5.25842130e-01 4.06860173e-01 8.83136153e-01 9.01190400e-01
7.71709919e-01 8.82205069e-01 -3.30200382e-02 -3.52103442e-01
2.83788983e-02 -1.32993758e+00 -3.68215054e-01 2.20469058e-01
-1.56326994e-01 -6.59985721e-01 -9.05165598e-02 5.58220856e-02]
|
[4.879788875579834, 5.649763107299805]
|
8ea73c2b-4b19-4b48-8b6f-748200dd01df
|
unsupervised-domain-adaptation-for-semantic-2
|
2109.08912
| null |
https://arxiv.org/abs/2109.08912v1
|
https://arxiv.org/pdf/2109.08912v1.pdf
|
Unsupervised Domain Adaptation for Semantic Segmentation via Low-level Edge Information Transfer
|
Unsupervised domain adaptation for semantic segmentation aims to make models trained on synthetic data (source domain) adapt to real images (target domain). Previous feature-level adversarial learning methods only consider adapting models on the high-level semantic features. However, the large domain gap between source and target domains in the high-level semantic features makes accurate adaptation difficult. In this paper, we present the first attempt at explicitly using low-level edge information, which has a small inter-domain gap, to guide the transfer of semantic information. To this end, a semantic-edge domain adaptation architecture is proposed, which uses an independent edge stream to process edge information, thereby generating high-quality semantic boundaries over the target domain. Then, an edge consistency loss is presented to align target semantic predictions with produced semantic boundaries. Moreover, we further propose two entropy reweighting methods for semantic adversarial learning and self-supervised learning, respectively, which can further enhance the adaptation performance of our architecture. Comprehensive experiments on two UDA benchmark datasets demonstrate the superiority of our architecture compared with state-of-the-art methods.
|
['Bo Du', 'Yonghao Xu', 'Chen Wu', 'Hongruixuan Chen']
|
2021-09-18
| null | null | null | null |
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 7.02035189e-01 4.89991397e-01 -2.23926693e-01 -5.28850198e-01
-7.20007956e-01 -4.12459046e-01 4.21869844e-01 -1.92986771e-01
-2.91070819e-01 7.49193549e-01 4.82804440e-02 6.30190670e-02
3.11027050e-01 -1.06201315e+00 -9.84608114e-01 -5.37113607e-01
3.41186613e-01 3.90124023e-01 5.05848408e-01 -1.45811498e-01
-1.08738922e-01 -4.70545478e-02 -1.02680802e+00 2.52245814e-01
1.39240503e+00 1.20029318e+00 1.32629871e-01 2.32582867e-01
-3.37081999e-01 5.07143319e-01 -5.01345038e-01 -5.80821812e-01
1.68555737e-01 -8.35484684e-01 -8.15256774e-01 6.90213144e-02
1.63351536e-01 -1.74725145e-01 -4.50195670e-01 1.44490373e+00
5.20469606e-01 1.95479065e-01 9.46993828e-01 -1.40691626e+00
-1.09401691e+00 3.51079971e-01 -3.48350942e-01 -1.63560063e-01
1.76906005e-01 1.72531605e-01 8.12193990e-01 -6.61579251e-01
9.02595997e-01 1.14219582e+00 7.43701160e-01 9.51322556e-01
-1.06511521e+00 -8.85536134e-01 3.62165719e-01 3.56855571e-01
-1.08888769e+00 -1.61706075e-01 1.37588835e+00 -2.21057042e-01
6.48280799e-01 -3.16243798e-01 5.48303008e-01 1.52024329e+00
2.42322031e-02 9.19364572e-01 1.04618800e+00 -3.33297104e-01
4.61944848e-01 2.03556716e-01 -4.21547920e-01 4.76783752e-01
1.09970331e-01 3.55668515e-01 -2.60016143e-01 1.81118354e-01
6.98670685e-01 -2.90642262e-01 -2.08127543e-01 -8.80813301e-01
-9.52532172e-01 9.37036991e-01 7.46194124e-01 1.55978456e-01
-2.23610476e-01 -1.29264742e-01 7.04010308e-01 1.58514291e-01
6.04375362e-01 2.24588409e-01 -5.36736310e-01 1.03886098e-01
-5.67987323e-01 8.23609531e-02 5.61524868e-01 1.26337385e+00
6.50916219e-01 3.02355438e-01 -2.36815885e-01 9.91914332e-01
3.23631138e-01 4.21593219e-01 7.33398259e-01 -7.78933406e-01
4.22082365e-01 5.97271979e-01 -1.00998394e-01 -7.77158976e-01
-1.43416211e-01 -3.94169003e-01 -8.76055717e-01 3.25064987e-01
4.24497783e-01 -2.95337379e-01 -1.21780872e+00 2.00949478e+00
4.96086329e-01 5.46597183e-01 5.24045050e-01 9.14186001e-01
8.81979167e-01 5.05472898e-01 4.37743932e-01 8.62675309e-02
1.20441663e+00 -1.18328381e+00 -6.92531347e-01 -6.52560532e-01
3.21737885e-01 -4.12744552e-01 1.20670164e+00 -1.98601112e-01
-8.56563032e-01 -7.52349496e-01 -1.17813718e+00 4.07500751e-02
-4.70825493e-01 -2.75863618e-01 3.76171917e-01 5.50110579e-01
-6.32084131e-01 5.42779446e-01 -7.84559190e-01 -2.18590260e-01
8.81723464e-01 1.48959801e-01 -2.46531337e-01 -1.13853298e-01
-1.73098922e+00 7.05831170e-01 8.77263129e-01 -4.87701774e-01
-7.43083358e-01 -7.89793968e-01 -1.23812807e+00 -5.26944771e-02
3.17151248e-01 -9.04551685e-01 1.14843535e+00 -1.50630641e+00
-1.71902943e+00 1.09137654e+00 1.09105274e-01 -5.44889688e-01
6.85807765e-01 1.10906646e-01 -6.86774671e-01 2.66326964e-01
1.74206436e-01 1.05438268e+00 1.08432686e+00 -1.42914784e+00
-5.96299827e-01 -3.91697556e-01 -1.00667633e-01 4.59322751e-01
-4.82469589e-01 -4.78242755e-01 -5.11273324e-01 -1.15115547e+00
-4.40523028e-02 -8.06559384e-01 -3.16046715e-01 5.54260202e-02
-1.81969285e-01 3.41862738e-02 8.01097155e-01 -7.67967284e-01
8.86359274e-01 -2.17301345e+00 2.37867355e-01 1.08437538e-01
-6.18723258e-02 3.06318343e-01 -1.84984401e-01 -1.56594008e-01
-1.64893702e-01 -1.01775944e-01 -9.35185909e-01 -9.66891572e-02
7.35539347e-02 3.01483512e-01 -2.33212680e-01 3.94525006e-02
4.27581728e-01 1.23154151e+00 -1.11531186e+00 -5.41851759e-01
1.59924179e-01 3.51064354e-01 -6.36802733e-01 3.21893156e-01
-2.81462967e-01 7.16682017e-01 -7.33977437e-01 5.54884315e-01
7.89263606e-01 -2.77176294e-02 -4.91529740e-02 -1.78070515e-01
6.68219328e-01 -8.21465335e-04 -8.16923082e-01 2.15013766e+00
-5.87766767e-01 3.03988278e-01 -1.72132641e-01 -1.33894813e+00
1.10992432e+00 1.11942060e-01 4.70166326e-01 -9.56559062e-01
1.26302302e-01 2.91661561e-01 -1.54189602e-01 -2.24369124e-01
2.31688485e-01 -4.20958638e-01 -3.92241329e-01 5.82022108e-02
2.47904941e-01 -4.12854850e-01 -2.21452013e-01 -7.13851452e-02
7.42819488e-01 5.36807597e-01 2.12340415e-01 -2.49056201e-02
6.63833141e-01 1.48026049e-01 9.92556334e-01 3.18636507e-01
-6.87374651e-01 8.79466474e-01 3.14680964e-01 -1.25138193e-01
-1.24278915e+00 -1.26254189e+00 -1.54764056e-01 7.44832754e-01
7.81700850e-01 1.71976238e-01 -1.32580268e+00 -1.34239352e+00
1.99367460e-02 1.04506516e+00 -6.87089443e-01 -9.67013657e-01
-3.96193504e-01 -6.31106257e-01 4.92700875e-01 9.00065362e-01
9.90134418e-01 -1.34849632e+00 -1.97128668e-01 2.83301890e-01
-2.93325752e-01 -1.30555487e+00 -7.79895842e-01 -6.52849078e-02
-9.14640248e-01 -8.15392315e-01 -1.02228963e+00 -1.14249158e+00
8.56134653e-01 -2.50710428e-01 1.12610924e+00 -5.17650843e-01
-3.32425870e-02 2.48016685e-01 -4.77546751e-01 -3.49536866e-01
-6.51731491e-01 9.09587443e-02 -1.44098625e-01 7.28776082e-02
6.75861716e-01 -6.16332710e-01 -6.42095089e-01 3.61734271e-01
-1.08036327e+00 1.43412396e-01 5.76387107e-01 1.15879118e+00
8.73867452e-01 -2.40998436e-03 1.00554633e+00 -1.14748883e+00
4.69469875e-01 -4.90302771e-01 -2.93612987e-01 1.61354721e-01
-7.23879695e-01 1.70993716e-01 9.83627141e-01 -5.81974864e-01
-1.45378351e+00 1.96669430e-01 -3.37422013e-01 -6.29337966e-01
-3.44220281e-01 9.80275199e-02 -7.28053033e-01 -6.69382960e-02
7.75776744e-01 4.20719594e-01 -5.55739179e-02 -2.03008085e-01
5.66907883e-01 5.49817979e-01 7.09998190e-01 -5.56857049e-01
8.84414256e-01 4.34585094e-01 -3.13170999e-01 -3.02639455e-01
-8.57804537e-01 -9.75855142e-02 -6.49873316e-01 2.93513127e-02
1.01020288e+00 -9.66496766e-01 -3.73916142e-02 8.88557911e-01
-8.38706553e-01 -5.58060408e-01 -7.43491650e-01 3.06539446e-01
-1.09028733e+00 3.45846027e-01 -3.79174471e-01 -2.41938725e-01
-3.33539248e-01 -1.02976608e+00 8.86755824e-01 3.72746468e-01
-5.44378795e-02 -1.43287480e+00 -2.72996705e-02 4.16942775e-01
2.25288421e-01 5.45242846e-01 6.23252451e-01 -8.59063625e-01
-1.64247915e-01 -5.43832518e-02 -3.37646544e-01 6.93009496e-01
1.76364467e-01 -6.24287486e-01 -9.01204467e-01 -2.16694668e-01
3.59712318e-02 -4.87978697e-01 9.48569655e-01 4.32062298e-01
1.41029537e+00 -1.59960330e-01 -4.58079010e-01 1.00539279e+00
1.35094666e+00 3.31046313e-01 7.09504426e-01 3.66121382e-01
8.47604454e-01 4.18701917e-01 8.54026973e-01 1.25670388e-01
4.21763510e-01 6.16708577e-01 3.53819937e-01 -3.41662377e-01
-4.94526982e-01 -7.91836858e-01 3.05196136e-01 5.33778071e-01
3.59802872e-01 -1.74112961e-01 -7.46297657e-01 7.14204788e-01
-1.75102961e+00 -5.93549848e-01 3.59841555e-01 1.96692395e+00
9.75091934e-01 5.07148027e-01 -1.47642381e-02 -2.12325558e-01
9.42460179e-01 1.85661420e-01 -1.29730141e+00 -2.82825440e-01
-3.08950618e-02 1.57740444e-01 6.94332659e-01 3.42807680e-01
-1.40405273e+00 1.28133523e+00 5.61184597e+00 1.12366617e+00
-8.40093791e-01 1.59322485e-01 6.91686273e-01 3.85863870e-01
-5.01169503e-01 -1.99981213e-01 -4.43759948e-01 7.99271584e-01
6.07885838e-01 -4.43797916e-01 3.25601548e-01 1.17277944e+00
-2.89645761e-01 4.87129629e-01 -8.94923151e-01 7.74741948e-01
6.89233020e-02 -1.01957119e+00 2.12245762e-01 -3.03465992e-01
1.15188169e+00 -2.59489208e-01 1.70447215e-01 4.54329342e-01
5.80380321e-01 -8.38900328e-01 5.88484883e-01 2.83723593e-01
1.22894573e+00 -9.05670345e-01 5.71065843e-01 1.78605467e-01
-1.19782495e+00 6.59927130e-02 -3.96910995e-01 3.75870705e-01
2.75558203e-01 3.43128294e-01 -5.63580275e-01 7.36018479e-01
6.60633802e-01 1.02600586e+00 -2.17186496e-01 6.71843410e-01
-5.00707269e-01 5.02372801e-01 -8.40189457e-02 3.24676663e-01
2.48964444e-01 -2.73398221e-01 6.09414458e-01 1.09314871e+00
2.64403373e-01 4.18763049e-02 7.04131797e-02 1.05377662e+00
-4.03838426e-01 4.88199592e-02 -6.85154796e-01 1.39891997e-01
5.20788491e-01 9.07675505e-01 -6.31382108e-01 -4.25170541e-01
-4.99725193e-01 1.48637295e+00 2.35035658e-01 4.73679990e-01
-1.09701908e+00 -6.65116727e-01 8.87643397e-01 -1.33974895e-01
3.22248042e-01 2.34137222e-01 -7.28404343e-01 -1.28502190e+00
1.35009900e-01 -6.83070004e-01 4.49547440e-01 -5.85288525e-01
-1.57099652e+00 4.88791734e-01 -2.64331877e-01 -1.47802198e+00
-2.60826916e-01 -4.84985709e-01 -6.72967672e-01 8.85940015e-01
-1.73799586e+00 -1.39315331e+00 -3.54390740e-01 6.49144471e-01
7.63681650e-01 -3.89408797e-01 7.93051660e-01 2.18644664e-01
-3.27454299e-01 1.15876603e+00 3.87841821e-01 4.10292596e-01
8.50597441e-01 -1.29130161e+00 8.04086566e-01 7.68417656e-01
-1.34681717e-01 -1.13776058e-01 4.42974508e-01 -7.10301459e-01
-5.56179345e-01 -1.53038633e+00 3.83315742e-01 -3.41075480e-01
4.84381557e-01 -2.88840979e-01 -1.22403550e+00 6.60958946e-01
-6.55328296e-03 1.98082179e-01 5.41937888e-01 -3.41941774e-01
-2.81173915e-01 -5.45475148e-02 -1.56406629e+00 6.52204573e-01
1.44094610e+00 -4.25055206e-01 -8.11310768e-01 5.91525771e-02
1.05503690e+00 -5.87067366e-01 -1.01153624e+00 6.40454412e-01
2.74644107e-01 -7.20182657e-01 1.22950053e+00 -7.04449713e-01
6.99346781e-01 -1.51485980e-01 1.33517191e-01 -1.74570131e+00
-3.13562036e-01 -2.67475843e-01 -9.06412452e-02 1.37025893e+00
3.13951582e-01 -7.77411044e-01 9.77197230e-01 4.75828826e-01
-3.30692887e-01 -5.24919748e-01 -8.78738701e-01 -9.65557754e-01
6.11154616e-01 -2.72137761e-01 6.92703068e-01 1.14405990e+00
-8.02609846e-02 2.79939711e-01 -1.13034815e-01 9.85404570e-03
7.23758042e-01 1.02488674e-01 4.81011391e-01 -9.83928502e-01
-1.39448628e-01 -4.29794669e-01 -5.55946052e-01 -1.02843893e+00
6.16815627e-01 -1.19699168e+00 1.38514787e-01 -1.44211411e+00
2.19179783e-02 -5.40723741e-01 -5.28972089e-01 4.01609957e-01
-5.57156444e-01 2.85459995e-01 -1.69255174e-04 5.37946401e-03
-3.88714761e-01 1.12798285e+00 1.70604694e+00 -4.09857810e-01
-1.55136347e-01 -8.06476325e-02 -8.19746077e-01 8.40969741e-01
8.86781096e-01 -4.79015976e-01 -6.49240196e-01 -3.44366282e-01
-4.10659999e-01 -1.68480128e-01 2.83868045e-01 -1.14462137e+00
-4.08973359e-02 -1.67078272e-01 4.83506799e-01 7.56827071e-02
1.37767032e-01 -9.58637536e-01 -1.38230577e-01 3.54911178e-01
-4.29130912e-01 -7.56116152e-01 3.74560922e-01 7.83682764e-01
-4.75657701e-01 -2.61667252e-01 1.23113620e+00 -2.09953287e-03
-1.21035445e+00 5.97083688e-01 2.47711599e-01 6.93984151e-01
1.43823254e+00 -3.91989678e-01 1.12457506e-01 -1.69470519e-01
-9.25021172e-01 3.37048709e-01 8.29096496e-01 7.22649932e-01
5.55922747e-01 -1.78340936e+00 -6.78169072e-01 3.33738387e-01
4.51374918e-01 3.39493513e-01 5.35295963e-01 1.66642442e-01
-2.97088832e-01 -5.61382510e-02 -6.36181474e-01 -4.03527617e-01
-7.06016362e-01 8.69114101e-01 4.14346367e-01 -3.08802366e-01
-6.33047998e-01 9.74042475e-01 6.59709334e-01 -7.01616347e-01
-8.62791836e-02 1.42893745e-02 -1.07127167e-01 -1.52686849e-01
1.67404264e-01 9.84427109e-02 -2.54874051e-01 -5.92434764e-01
-2.89982527e-01 7.49926388e-01 -7.62030631e-02 -8.22114293e-03
9.94770765e-01 -1.72109067e-01 5.37178159e-01 7.65641034e-02
1.24465525e+00 -3.90729070e-01 -1.98219442e+00 -5.31368911e-01
-1.30574659e-01 -4.31042016e-01 -1.06878780e-01 -9.35217917e-01
-1.25708699e+00 9.53998923e-01 6.95036173e-01 -2.55605429e-01
1.50422633e+00 1.99136399e-02 1.29080546e+00 -1.70374602e-01
1.76235959e-01 -1.53073561e+00 2.15609506e-01 2.98343152e-01
6.82555199e-01 -1.55953407e+00 -5.77423096e-01 -6.15552485e-01
-1.09751630e+00 6.79258049e-01 9.76275146e-01 -2.47972846e-01
4.02301759e-01 6.34310907e-03 2.63142429e-04 2.83552647e-01
-2.26361766e-01 -2.53256470e-01 3.06118429e-01 1.23009014e+00
-6.47519380e-02 7.80670419e-02 -2.44183347e-01 9.71307695e-01
-1.95174534e-02 1.76438585e-01 7.17797875e-02 4.84279662e-01
-2.75047034e-01 -1.19705284e+00 -1.56898677e-01 2.71804541e-01
-3.44985425e-01 -8.47189799e-02 -3.69843870e-01 6.83670461e-01
1.59900457e-01 5.61241329e-01 1.23887554e-01 -3.80924404e-01
5.78574657e-01 1.92539111e-01 2.91321158e-01 -4.06449318e-01
-6.93457201e-02 -3.57197732e-01 -1.16004065e-01 -4.06069726e-01
-1.18822627e-01 -5.62091291e-01 -1.56189966e+00 1.52467072e-01
3.55393216e-02 -5.94321229e-02 4.27925974e-01 7.62762606e-01
5.04281163e-01 9.60948825e-01 6.54256761e-01 -5.24024844e-01
-5.99183083e-01 -8.13145638e-01 -4.74693030e-01 1.01873159e+00
2.84881294e-02 -8.25389504e-01 -1.73117250e-01 5.09073913e-01]
|
[9.76917552947998, 1.4436546564102173]
|
ef5326dc-088c-418e-bda5-76f7edc64d81
|
using-natural-language-processing-and-2
|
2306.09737
| null |
https://arxiv.org/abs/2306.09737v1
|
https://arxiv.org/pdf/2306.09737v1.pdf
|
Using Natural Language Processing and Networks to Automate Structured Literature Reviews: An Application to Farmers Climate Change Adaptation
|
The fast-growing number of research articles makes it problematic for scholars to keep track of the new findings related to their areas of expertise. Furthermore, linking knowledge across disciplines in rapidly developing fields becomes challenging for complex topics like climate change that demand interdisciplinary solutions. At the same time, the rise of Black Box types of text summarization makes it difficult to understand how text relationships are built, let alone relate to existing theories conceptualizing cause-effect relationships and permitting hypothesizing. This work aims to sensibly use Natural Language Processing by extracting variables relations and synthesizing their findings using networks while relating to key concepts dominant in relevant disciplines. As an example, we apply our methodology to the analysis of farmers' adaptation to climate change. For this, we perform a Natural Language Processing analysis of publications returned by Scopus in August 2022. Results show that the use of Natural Language Processing together with networks in a descriptive manner offers a fast and interpretable way to synthesize literature review findings as long as researchers back up results with theory.
|
['Tatiana Filatova', 'Sofia Gil-Clavel']
|
2023-06-16
| null | null | null | null |
['text-summarization']
|
['natural-language-processing']
|
[ 4.02188867e-01 1.39553964e-01 -5.97586334e-01 1.84751943e-01
1.79316834e-01 -8.44877243e-01 7.34968901e-01 1.15290654e+00
-5.09862363e-01 6.03642166e-01 5.64415216e-01 -1.02447522e+00
-5.69392383e-01 -1.06632054e+00 -4.89290416e-01 -5.21203242e-02
4.86831255e-02 2.50742841e-03 -1.59804404e-01 -3.24640900e-01
6.97574973e-01 6.48924530e-01 -1.50078654e+00 7.13005215e-02
1.06143367e+00 2.68297464e-01 3.40814173e-01 1.80224687e-01
-7.03641295e-01 7.35691428e-01 -5.61518788e-01 -4.20938373e-01
-5.91463894e-02 -4.27132130e-01 -8.74094129e-01 -3.88246775e-01
2.25577466e-02 2.13515460e-01 1.44489959e-01 1.12040091e+00
1.49936512e-01 -1.48037881e-01 4.02072638e-01 -1.11670613e+00
-7.70421863e-01 9.09120560e-01 -6.76931381e-01 4.07696992e-01
6.05593860e-01 -3.85961048e-02 9.89107609e-01 -4.24804628e-01
1.11682606e+00 1.47048533e+00 5.91181397e-01 -2.59413153e-01
-1.17723799e+00 -6.09414697e-01 3.76799732e-01 1.37397096e-01
-1.12331855e+00 -2.07528740e-01 6.15083694e-01 -7.77610660e-01
1.06081200e+00 2.72807598e-01 7.58529782e-01 8.62122536e-01
5.01535892e-01 -1.64656833e-01 1.16406381e+00 -6.60104036e-01
1.08024739e-01 1.92734681e-03 2.51399219e-01 2.30075538e-01
1.10102677e+00 -3.08406681e-01 -5.37982225e-01 -7.53850490e-03
3.01473856e-01 1.28247529e-01 -4.30231504e-02 3.18699390e-01
-1.48053455e+00 6.44730210e-01 3.86373878e-01 8.07594657e-01
-6.59527540e-01 -2.05495253e-01 5.27184427e-01 4.06975359e-01
6.65698588e-01 7.83498824e-01 -3.22866946e-01 -5.75069934e-02
-9.56499398e-01 2.18690842e-01 1.11445057e+00 7.06283271e-01
6.09165668e-01 -2.35087574e-01 4.14327681e-01 4.92166996e-01
4.29417133e-01 3.85403961e-01 2.76510030e-01 -8.54693055e-01
4.61861879e-01 1.11115611e+00 -1.69278651e-01 -1.72791696e+00
-5.12902439e-01 -4.05287534e-01 -7.62724280e-01 8.69243592e-02
3.23031902e-01 -2.57894605e-01 -3.88279080e-01 1.40251362e+00
4.71296236e-02 -5.07489085e-01 1.40389547e-01 5.55422246e-01
9.95379388e-01 7.37584114e-01 5.73576927e-01 -5.78748107e-01
1.61810839e+00 -2.93446541e-01 -1.14287186e+00 -9.71394405e-02
4.82761830e-01 -8.85659218e-01 7.81527340e-01 3.40400219e-01
-8.31421912e-01 -1.01868674e-01 -8.69349003e-01 -1.90980658e-01
-1.10012984e+00 -3.57148409e-01 5.67007840e-01 1.94167271e-01
-9.31215763e-01 6.50345445e-01 -6.21413767e-01 -9.68280375e-01
4.34054762e-01 -1.07293487e-01 -2.65874356e-01 7.44459629e-02
-1.26401412e+00 1.43748224e+00 6.24721289e-01 1.17933311e-01
1.77134022e-01 -9.90921736e-01 -5.26029468e-01 7.10343644e-02
7.63406396e-01 -7.80108392e-01 7.69537687e-01 -7.28018105e-01
-9.65116799e-01 7.22039819e-01 -2.37299472e-01 -3.60393733e-01
3.50240231e-01 2.30146721e-02 -5.92731953e-01 2.28122398e-01
4.40448642e-01 1.86799541e-01 7.32729584e-02 -1.13691688e+00
-5.89917004e-01 -5.73213816e-01 6.70187697e-02 -1.00609310e-01
-6.02465272e-01 5.44539452e-01 1.31886214e-01 -6.99720979e-01
1.62652358e-01 -4.64285761e-01 -3.96720141e-01 -2.75503732e-02
-1.93143398e-01 -2.11546153e-01 6.59062386e-01 -6.94996357e-01
1.80823231e+00 -1.75339711e+00 8.65236670e-02 2.85809010e-01
6.16047442e-01 -6.83774129e-02 2.32034370e-01 1.06766045e+00
-4.70033772e-02 8.71271968e-01 -2.45568916e-01 3.94666344e-01
-1.60478219e-01 3.17393690e-01 -3.46103400e-01 2.15839967e-01
2.44395986e-01 7.72823453e-01 -1.14822567e+00 -5.06719172e-01
2.09281325e-01 1.90765306e-01 -1.52730867e-01 -3.88389796e-01
3.75281386e-02 -1.18913420e-01 -6.41340971e-01 5.79982698e-01
3.66025209e-01 -8.12681243e-02 5.38811088e-01 -2.46699914e-01
-1.01595986e+00 4.41810012e-01 -8.41584146e-01 1.34163988e+00
-4.70496625e-01 9.75525320e-01 5.84902726e-02 -1.13915300e+00
1.02722490e+00 2.46754661e-01 5.28281510e-01 -5.81166923e-01
1.28426731e-01 2.11150676e-01 2.92942077e-01 -6.93935633e-01
4.34572965e-01 -3.81425828e-01 -2.22510751e-02 3.98401439e-01
-1.67397022e-01 -2.98099250e-01 6.01140141e-01 2.40760833e-01
8.71120572e-01 2.76696216e-02 7.92438626e-01 -7.06826150e-01
2.23587751e-01 4.79638726e-01 3.42293888e-01 2.96972275e-01
1.70306444e-01 -1.84431635e-02 7.75328398e-01 -3.65111232e-01
-9.83453333e-01 -6.68904006e-01 -2.24521309e-01 5.63644767e-01
-3.67337972e-01 -6.69850469e-01 -3.50387067e-01 1.50788024e-01
2.58587509e-01 7.17296362e-01 -6.27797246e-01 1.68601662e-01
-1.80911869e-01 -5.94912231e-01 2.67238349e-01 4.45027696e-03
4.73671019e-01 -1.10008562e+00 -9.03962970e-01 4.60706472e-01
-7.68746287e-02 -9.77842450e-01 4.85865265e-01 2.02447653e-01
-9.76816356e-01 -1.12827301e+00 -5.74599028e-01 -4.31061596e-01
5.98807156e-01 2.53156304e-01 1.01414442e+00 -5.56515902e-02
-2.67539173e-01 6.84898347e-02 -4.91326630e-01 -9.82033491e-01
-7.32911050e-01 3.28295022e-01 -1.02677144e-01 -7.80116975e-01
6.18420422e-01 -7.35693753e-01 -1.27871752e-01 -1.28227666e-01
-1.23293281e+00 -1.41620934e-01 6.27648711e-01 1.37023538e-01
1.79385215e-01 2.69959867e-01 1.09363985e+00 -7.80464172e-01
1.14387155e+00 -1.08656728e+00 -4.88697201e-01 4.57903862e-01
-1.14206898e+00 -3.27605903e-02 5.43698192e-01 -1.12823367e-01
-7.61787653e-01 -5.42276680e-01 3.05876970e-01 2.64514208e-01
-2.45315477e-01 1.50784743e+00 1.62598625e-01 3.77847463e-01
7.74878442e-01 -4.48036134e-01 3.35436761e-01 -4.16703671e-01
5.31486750e-01 6.83443546e-01 4.38548297e-01 -5.60917974e-01
6.94286525e-01 2.95257509e-01 1.58153459e-01 -1.08503318e+00
-3.86498749e-01 -3.94946039e-01 -6.50447488e-01 -4.99574393e-01
8.23764265e-01 -6.99721992e-01 -8.81018460e-01 -1.57706171e-01
-1.16091716e+00 1.61263347e-01 -2.89894253e-01 5.46115100e-01
6.53817654e-02 2.79387563e-01 -2.12238971e-02 -5.30557573e-01
-2.98081458e-01 -7.05668390e-01 2.96598941e-01 2.80962616e-01
-8.11372638e-01 -1.32109010e+00 7.56209865e-02 2.81123102e-01
4.93207604e-01 6.63877487e-01 1.05407274e+00 -4.45326716e-01
-4.65002246e-02 -8.32054671e-03 -2.63771176e-01 -1.24454409e-01
6.75432026e-01 6.22107565e-01 -4.54922587e-01 2.09877938e-01
-2.18265966e-01 3.91199291e-01 6.13618851e-01 3.08770627e-01
7.81366229e-01 -6.55638814e-01 -3.09142709e-01 -8.84374231e-02
1.54098916e+00 2.64998645e-01 4.65854824e-01 9.24830139e-01
5.59566855e-01 1.41104293e+00 3.47492278e-01 2.17578784e-01
4.81265157e-01 1.63881019e-01 3.34412940e-02 6.94111437e-02
2.77114958e-01 -1.07238732e-01 1.79251395e-02 9.13956881e-01
-3.16222578e-01 -1.39665350e-01 -1.50254905e+00 6.84481084e-01
-1.78613174e+00 -1.16081154e+00 -4.26797330e-01 1.79385293e+00
9.02325153e-01 2.37366319e-01 5.82268722e-02 -8.92966893e-03
5.36624968e-01 1.13075204e-01 -1.92764133e-01 -7.46115327e-01
-2.17895895e-01 1.58554852e-01 6.18015945e-01 2.38796055e-01
-4.00263518e-01 7.61844218e-01 6.22361994e+00 2.66814619e-01
-1.11187112e+00 -3.79421026e-01 3.46645862e-01 1.11111514e-01
-7.42577016e-01 3.53954941e-01 -1.98834583e-01 1.71972066e-01
1.11664748e+00 -9.23681736e-01 1.14168473e-01 3.24742436e-01
1.01501679e+00 -5.48534453e-01 -6.03062034e-01 4.31541741e-01
-3.12560558e-01 -1.68483698e+00 2.28628069e-01 -4.07442227e-02
5.71359992e-01 -1.38755128e-01 -3.63764077e-01 -3.00989956e-01
4.46560949e-01 -9.33840930e-01 6.49055600e-01 6.54969573e-01
3.97757292e-01 -4.56922114e-01 6.94372654e-01 1.43726319e-01
-9.49032187e-01 -5.67398258e-02 -3.09580714e-01 -7.10168064e-01
9.37125385e-02 8.04565489e-01 -7.11640537e-01 1.11015379e+00
7.54706383e-01 9.98710334e-01 -4.08635467e-01 8.31624687e-01
-1.63708091e-01 7.66341567e-01 -2.24900261e-01 -2.12038860e-01
1.23139597e-01 -4.62367862e-01 5.35597563e-01 1.32371688e+00
3.01854044e-01 1.63578659e-01 -2.33369440e-01 1.03759360e+00
-1.07910462e-01 5.39230287e-01 -9.79869068e-01 -7.74267077e-01
6.93041742e-01 1.21005702e+00 -1.21296549e+00 -1.63656950e-01
-5.24706423e-01 1.05044216e-01 -2.27416769e-01 4.29999739e-01
3.84687744e-02 -4.85015869e-01 4.22630012e-01 3.80778164e-01
-2.59553552e-01 -4.10424471e-01 -6.68568730e-01 -8.36928785e-01
-9.63225309e-03 -8.50955248e-01 2.00203046e-01 -8.14604700e-01
-1.24625373e+00 1.91993162e-01 5.31422973e-01 -9.50028777e-01
5.47132222e-03 -3.22268158e-01 -6.42199039e-01 1.02176070e+00
-1.34473956e+00 -8.50928605e-01 9.56623331e-02 -1.20964520e-01
1.66159466e-01 6.57610819e-02 6.52064741e-01 8.01123679e-02
-6.01445615e-01 -3.04346234e-01 2.92237885e-02 -2.66873658e-01
6.26015842e-01 -9.08641934e-01 3.23176265e-01 7.80164540e-01
-3.15906465e-01 1.05750382e+00 8.68443608e-01 -1.10846901e+00
-1.21427751e+00 -6.23369873e-01 1.31956553e+00 -1.05703019e-01
1.46863401e+00 7.75169069e-03 -1.09968305e+00 2.88267553e-01
7.34306276e-01 -5.88409960e-01 8.81211281e-01 3.12483788e-01
-1.30419910e-01 -1.55674160e-01 -9.69380081e-01 9.03055847e-01
8.27423513e-01 -3.02569687e-01 -1.00246239e+00 8.95230025e-02
7.35116780e-01 2.20708176e-01 -1.19125903e+00 8.76520500e-02
6.08312488e-01 -2.27247253e-01 6.28759742e-01 -7.22997427e-01
8.34774077e-01 -3.02154064e-01 1.75712913e-01 -1.14450836e+00
-1.67569578e-01 -5.63414812e-01 6.44731998e-01 1.48648906e+00
6.37283742e-01 -7.86161542e-01 7.63876438e-02 6.93220317e-01
-1.41703989e-02 -5.34476519e-01 -5.31213105e-01 -5.34773767e-01
4.76814300e-01 -3.12027097e-01 3.99131000e-01 1.60829413e+00
5.99105239e-01 2.99707621e-01 3.34932655e-01 -2.83362150e-01
4.54757184e-01 -3.11695300e-02 5.00889480e-01 -1.92029274e+00
6.74051285e-01 -9.43044126e-01 -2.99588174e-01 -1.85788027e-03
6.01199791e-02 -9.57599521e-01 -4.94559914e-01 -2.20266509e+00
5.40805832e-02 -1.61456630e-01 6.49822354e-02 6.06477320e-01
-3.27127911e-02 -3.17503721e-01 3.75285983e-01 3.42280239e-01
9.18906108e-02 -5.72485775e-02 1.02317595e+00 -3.86896506e-02
-4.55695391e-01 -3.68631274e-01 -1.30431294e+00 7.52268076e-01
9.44744825e-01 -5.47952771e-01 -3.46928418e-01 -9.89059210e-02
1.14347696e+00 -2.33715802e-01 4.73739684e-01 -5.55862486e-01
3.20297867e-01 -6.26119733e-01 -9.72402915e-02 -4.91664410e-01
-7.92689621e-01 -1.06069314e+00 4.68570292e-01 4.42539543e-01
-3.97411019e-01 5.30215859e-01 7.13852584e-01 2.21600041e-01
-3.25178206e-01 -2.88934439e-01 -2.98262807e-03 -1.88096106e-01
-5.59567213e-01 -4.42621350e-01 -8.82352054e-01 -1.15171231e-01
8.66772532e-01 -3.47611248e-01 -6.79407358e-01 -2.88690627e-01
-5.59505224e-01 5.22522211e-01 5.25430083e-01 7.02486277e-01
3.14306229e-01 -8.46501052e-01 -7.43853509e-01 -5.18144906e-01
-7.01466650e-02 5.01741767e-02 8.35792422e-02 7.47348845e-01
-8.21739554e-01 5.05796492e-01 -4.59565252e-01 -6.72706664e-02
-1.06229484e+00 5.98606288e-01 -1.61784157e-01 -9.98189226e-02
-5.96098065e-01 1.72560681e-02 -2.60250241e-01 -1.32387698e-01
-1.30930722e-01 -8.32108438e-01 -5.82426131e-01 9.76560891e-01
2.63642788e-01 5.98974705e-01 -1.53620407e-01 -5.44483006e-01
-5.50962865e-01 5.01430154e-01 2.35570014e-01 -2.81872749e-01
1.78504753e+00 -3.17861766e-01 -8.73964012e-01 9.10308540e-01
1.01340401e+00 1.99447021e-01 -3.35572779e-01 -6.84745386e-02
5.46626091e-01 -6.07469976e-02 -5.95341856e-03 -8.99060011e-01
-7.01560616e-01 5.97526491e-01 1.56361014e-02 7.17733681e-01
1.10153222e+00 -7.03779235e-03 6.73983886e-05 5.81317484e-01
-1.66735396e-01 -1.12035775e+00 -5.52408040e-01 1.83868423e-01
1.14785790e+00 -9.31940556e-01 6.85985208e-01 -4.49301779e-01
-2.59947658e-01 1.58635378e+00 4.73153479e-02 2.81930715e-01
7.91216791e-01 3.65084499e-01 9.48141515e-02 -5.01969516e-01
-4.83473301e-01 -2.13919669e-01 1.34392828e-01 2.81780213e-01
7.66763628e-01 4.77395207e-02 -1.08236778e+00 2.73189485e-01
-3.32352430e-01 1.64175972e-01 8.15243006e-01 1.01692140e+00
-4.37090933e-01 -1.22927082e+00 -6.73001170e-01 6.01446033e-01
-8.83822322e-01 -1.56130880e-01 -8.54273081e-01 1.19877088e+00
-4.57330458e-02 1.11407757e+00 9.64558870e-02 -6.66975677e-02
4.23712432e-01 -2.50540432e-02 -7.87382647e-02 -6.22143626e-01
-8.26101363e-01 1.85315590e-02 2.70874798e-01 -4.11548615e-02
-1.14667284e+00 -9.87162411e-01 -1.34023070e+00 -5.58057010e-01
-9.74390879e-02 1.97213888e-01 9.31495070e-01 1.08678830e+00
5.18694520e-01 8.29697251e-01 1.56669974e-01 -3.20254236e-01
9.53275412e-02 -9.37863827e-01 -1.26210108e-01 -1.33640468e-01
1.91547170e-01 -6.21210992e-01 -1.84913009e-01 2.76608318e-01]
|
[9.51965618133545, 8.125582695007324]
|
00143df5-d47f-489f-a7a2-45bb3a62eb9c
|
quantization-aware-and-tensor-compressed
|
2306.01076
| null |
https://arxiv.org/abs/2306.01076v2
|
https://arxiv.org/pdf/2306.01076v2.pdf
|
Quantization-Aware and Tensor-Compressed Training of Transformers for Natural Language Understanding
|
Fine-tuned transformer models have shown superior performances in many natural language tasks. However, the large model size prohibits deploying high-performance transformer models on resource-constrained devices. This paper proposes a quantization-aware tensor-compressed training approach to reduce the model size, arithmetic operations, and ultimately runtime latency of transformer-based models. We compress the embedding and linear layers of transformers into small low-rank tensor cores, which significantly reduces model parameters. A quantization-aware training with learnable scale factors is used to further obtain low-precision representations of the tensor-compressed models. The developed approach can be used for both end-to-end training and distillation-based training. To improve the convergence, a layer-by-layer distillation is applied to distill a quantized and tensor-compressed student model from a pre-trained transformer. The performance is demonstrated in two natural language understanding tasks, showing up to $63\times$ compression ratio, little accuracy loss and remarkable inference and training speedup.
|
['Zheng Zhang', 'Siegfried Kunzmann', 'Samridhi Choudhary', 'Zi Yang']
|
2023-06-01
| null | null | null | null |
['quantization']
|
['methodology']
|
[ 6.56293556e-02 1.38843851e-02 -2.19168603e-01 -7.14041829e-01
-8.80500555e-01 -3.34452927e-01 3.11990678e-01 1.75279915e-01
-4.01614547e-01 2.46792987e-01 4.31896038e-02 -7.13854432e-01
-2.66848300e-02 -9.20424402e-01 -9.47283864e-01 -4.48052377e-01
-8.48356113e-02 8.00851166e-01 1.64009005e-01 -8.53598956e-03
2.09911719e-01 2.02308729e-01 -1.65794694e+00 6.19572520e-01
9.66725111e-01 1.33739972e+00 4.09189254e-01 7.50272870e-01
-3.32799077e-01 8.67750168e-01 -1.26019895e-01 -7.53910422e-01
7.27929920e-02 2.35948876e-01 -7.09011495e-01 -3.63453329e-01
7.54483402e-01 -7.80091643e-01 -3.85568738e-01 8.85748506e-01
2.15188608e-01 -1.15936227e-01 2.91172445e-01 -9.26645100e-01
-3.99327427e-01 9.42565680e-01 -1.06171422e-01 6.67673200e-02
1.34108299e-02 4.54843864e-02 1.29182744e+00 -9.35467184e-01
1.67043760e-01 1.39259088e+00 6.72186792e-01 2.12410912e-01
-1.18618524e+00 -9.02428627e-01 -6.15401864e-02 2.92672217e-01
-1.29813635e+00 -3.52686435e-01 5.72309077e-01 -2.13089988e-01
1.48999619e+00 2.62846708e-01 8.71183991e-01 6.78417742e-01
1.72971517e-01 8.32650363e-01 9.14790630e-01 -1.65120617e-01
2.86088675e-01 1.59104019e-01 1.08124930e-02 1.25114751e+00
2.06698060e-01 -1.38698414e-01 -7.75171757e-01 -2.02369720e-01
5.85686088e-01 1.57616660e-01 3.07935297e-01 -1.57786533e-01
-1.18068540e+00 8.30765128e-01 8.45108151e-01 2.07135484e-01
-3.23491424e-01 5.48028827e-01 7.90781796e-01 3.04504365e-01
4.91399556e-01 3.67731482e-01 -6.96928740e-01 -5.66338837e-01
-1.17567623e+00 3.09717447e-01 6.86164916e-01 9.55120504e-01
8.30509067e-01 3.56094003e-01 -3.86102982e-02 8.83178830e-01
2.11216480e-01 7.49457359e-01 6.95517182e-01 -7.37605274e-01
8.59177232e-01 8.49437952e-01 -4.65860248e-01 -9.94220018e-01
-1.44071579e-02 -5.37518799e-01 -1.09282959e+00 -5.74575782e-01
-2.03163605e-02 3.80684197e-01 -8.59941363e-01 1.34088254e+00
4.19621378e-01 2.77171850e-01 -8.03497806e-02 6.86073244e-01
3.79365414e-01 8.22873473e-01 1.04098268e-01 3.26786101e-01
1.67735028e+00 -9.66350675e-01 -5.11898041e-01 -1.78388387e-01
1.26994669e+00 -6.37844443e-01 1.42884135e+00 6.48681223e-01
-1.15278053e+00 -5.37596762e-01 -1.15998650e+00 -8.98263156e-01
-4.34830040e-01 4.81853783e-01 1.00676560e+00 4.30397302e-01
-7.69906819e-01 8.33753049e-01 -1.31023371e+00 9.86293554e-02
5.42534292e-01 6.14861071e-01 -1.27875552e-01 -3.52709770e-01
-9.79414880e-01 7.35530198e-01 5.78258693e-01 2.17541024e-01
-1.04713809e+00 -1.09256971e+00 -7.32452095e-01 3.69744599e-01
-2.27903724e-02 -7.09709585e-01 1.19633591e+00 -1.83546189e-02
-1.60396326e+00 5.24316192e-01 -1.65739834e-01 -8.63441110e-01
1.24997482e-01 -4.42513764e-01 1.25736013e-01 9.63963568e-02
-2.44643301e-01 5.07922947e-01 9.50184286e-01 -3.17488551e-01
-5.67168176e-01 -5.29234886e-01 2.09994942e-01 5.62049337e-02
-9.64686334e-01 -3.87531102e-01 -4.18905497e-01 -4.76415753e-01
3.69672090e-01 -8.84412766e-01 -8.26476589e-02 9.04257894e-02
-2.82411933e-01 -5.04807711e-01 7.43436575e-01 -6.58061624e-01
1.21419299e+00 -2.04910207e+00 2.22117081e-01 -6.24330118e-02
4.02148873e-01 3.33392441e-01 8.22815299e-03 2.08626330e-01
4.09766197e-01 -6.89051300e-02 -3.25229466e-02 -6.78470969e-01
2.71887481e-01 6.92301452e-01 -6.28729522e-01 1.76441088e-01
8.11756253e-02 8.98599863e-01 -7.81725049e-01 -5.41116536e-01
3.07017356e-01 6.46406174e-01 -1.11712205e+00 3.97361338e-01
-5.29015601e-01 -1.68965653e-01 -5.01518965e-01 4.97090787e-01
6.06479824e-01 -3.84647518e-01 1.24888778e-01 -7.44557917e-01
1.97748974e-01 9.48576570e-01 -7.43611097e-01 2.01999664e+00
-1.02832234e+00 3.56942177e-01 -1.49407864e-01 -1.18188560e+00
8.70756745e-01 1.54816374e-01 1.49082363e-01 -8.19469035e-01
8.38818029e-02 4.95349109e-01 -2.40201101e-01 -2.95495868e-01
6.54269695e-01 -8.63313824e-02 -1.22943304e-01 4.91806924e-01
2.04220816e-01 -5.27212501e-01 1.15212969e-01 3.30122113e-01
8.14277589e-01 1.39043182e-01 -2.83878177e-01 -1.84254915e-01
3.97902727e-01 -8.16782638e-02 2.15571493e-01 2.55532861e-01
4.29936469e-01 -1.52719662e-01 3.88632834e-01 -7.43033230e-01
-1.43912578e+00 -8.33439589e-01 -7.67049491e-02 1.62398362e+00
-4.15514022e-01 -9.10745561e-01 -7.02189684e-01 -1.70960218e-01
9.04122293e-02 6.25298083e-01 -3.49060565e-01 -3.17729890e-01
-7.37793624e-01 -6.01369262e-01 7.88716137e-01 6.57372355e-01
7.29081810e-01 -1.91966385e-01 -6.01548553e-01 2.40341708e-01
-1.14749856e-01 -1.42561877e+00 -1.75273135e-01 2.30640814e-01
-1.51928532e+00 -5.90739667e-01 -1.45415932e-01 -6.34224176e-01
5.69983482e-01 -1.23330466e-01 9.82020676e-01 1.17622823e-01
-3.63845490e-02 -3.30554485e-01 2.20709876e-03 -1.27057180e-01
-1.59242392e-01 4.52159852e-01 1.99492410e-01 -2.64624983e-01
4.07382637e-01 -7.94832826e-01 -4.05246764e-01 -1.07505940e-01
-7.53667295e-01 5.12644768e-01 7.98149765e-01 9.39069867e-01
6.76904023e-01 5.30766062e-02 -9.36207846e-02 -6.87584877e-01
4.86028761e-01 3.23468680e-03 -8.91777873e-01 1.81449279e-01
-8.45954299e-01 7.39866853e-01 1.01951957e+00 -5.53631067e-01
-4.92604464e-01 -2.15920676e-02 -1.49788752e-01 -8.03257763e-01
5.96449792e-01 5.96875846e-01 2.64883548e-01 -1.95112620e-02
4.88591522e-01 4.99447793e-01 -1.43835679e-01 -8.03301036e-01
4.37105119e-01 6.43437564e-01 3.77372146e-01 -9.40010726e-01
8.25975120e-01 1.47532970e-01 7.53677115e-02 -7.15417802e-01
-1.13836586e+00 -1.18255265e-01 -4.84702021e-01 5.70152819e-01
5.92635095e-01 -1.30355144e+00 -8.14049363e-01 9.51521248e-02
-1.08386946e+00 -3.02491933e-01 -1.55097872e-01 6.52087331e-01
-3.06865722e-01 -6.71163946e-02 -9.78101790e-01 -5.17780125e-01
-9.20604229e-01 -1.32028973e+00 1.30994248e+00 -2.29286805e-01
2.33604729e-01 -8.67681682e-01 -1.77887708e-01 7.22600281e-01
7.36169159e-01 -3.40675414e-01 1.18652749e+00 -5.60103118e-01
-1.05650342e+00 -3.79308581e-01 -4.39886957e-01 6.56748235e-01
-5.15311360e-01 -3.73486996e-01 -1.10692525e+00 -4.58528280e-01
-4.85494435e-02 -6.59703672e-01 6.24716282e-01 -1.26834437e-01
1.52846885e+00 -6.70062184e-01 -2.91190028e-01 1.01466811e+00
1.29907310e+00 -4.10472900e-01 2.58587241e-01 -1.73190907e-01
1.13683188e+00 -1.55669838e-01 3.68574053e-01 3.58522177e-01
6.79887474e-01 5.79216719e-01 1.22399703e-01 2.13176355e-01
4.94035892e-02 -7.39240825e-01 2.72494793e-01 1.70727336e+00
1.04376115e-01 3.43238473e-01 -8.64635050e-01 5.59630930e-01
-1.33234990e+00 -3.92589301e-01 2.23579630e-01 2.07430863e+00
9.71673965e-01 3.80450904e-01 -1.85486317e-01 3.27265918e-01
-7.95132741e-02 1.60567537e-02 -6.55014336e-01 -7.41309285e-01
4.96194124e-01 5.55668831e-01 4.86517847e-01 5.86469710e-01
-6.25686347e-01 1.17861795e+00 6.00328207e+00 1.03088880e+00
-1.47261667e+00 3.48688632e-01 4.78751987e-01 -2.45040715e-01
-2.52077073e-01 1.21556921e-02 -9.37102258e-01 9.39250439e-02
1.58058417e+00 -2.87782490e-01 6.32939219e-01 1.26876545e+00
-3.71154621e-02 3.45931143e-01 -1.43982494e+00 1.17571998e+00
-2.85746008e-01 -1.42334068e+00 4.51936305e-01 1.64240718e-01
3.17866057e-01 3.38628829e-01 1.26186132e-01 8.12136769e-01
1.59763262e-01 -9.38047528e-01 4.23877299e-01 1.10837452e-01
1.05334496e+00 -7.15322673e-01 5.99391103e-01 4.39625412e-01
-1.43475175e+00 -1.88789845e-01 -8.34447324e-01 -2.30321959e-01
-2.90971667e-01 7.34048188e-01 -1.17361689e+00 2.43016988e-01
8.23982000e-01 6.33528650e-01 -6.15882814e-01 1.53357327e-01
1.04143806e-01 7.96775162e-01 -8.55474055e-01 -2.98680931e-01
4.28748578e-01 -2.45143503e-01 -1.80347621e-01 1.15158927e+00
3.07966739e-01 3.53878699e-02 1.74695343e-01 9.04151082e-01
-1.75792396e-01 -2.23647710e-02 -4.04041737e-01 -4.51832384e-01
6.11705899e-01 1.20918572e+00 -1.22482337e-01 -9.65707600e-01
-1.37474239e-01 8.60247731e-01 6.18057847e-01 -1.13013685e-01
-8.98328304e-01 -3.00880522e-01 5.23145258e-01 2.83075213e-01
4.97106820e-01 -5.56740165e-01 -4.41896260e-01 -1.46113718e+00
2.45517731e-01 -8.46822858e-01 1.09294958e-01 -6.05798900e-01
-7.49680996e-01 6.99851155e-01 1.34633914e-01 -9.51466262e-01
-4.78085071e-01 -6.31869197e-01 1.62827075e-02 8.37379694e-01
-1.39229834e+00 -1.38597310e+00 -2.34713301e-01 5.19830585e-01
2.86350995e-01 -1.06712848e-01 1.02023995e+00 5.74887097e-01
-4.97427940e-01 9.22090530e-01 1.05278403e-01 1.31142139e-01
-2.05486398e-02 -1.12248421e+00 4.45332289e-01 6.03916407e-01
2.12142915e-02 8.92543375e-01 4.83156472e-01 -1.39380246e-01
-2.20798683e+00 -1.10721755e+00 1.16860831e+00 -6.79289773e-02
9.25984800e-01 -8.50729406e-01 -8.89527023e-01 7.80456424e-01
-3.08337599e-01 3.06254566e-01 5.46020985e-01 2.76679128e-01
-7.65162647e-01 -6.80023074e-01 -1.09968317e+00 4.30610687e-01
8.90418172e-01 -1.11357498e+00 -4.02061164e-01 7.14226305e-01
9.67862189e-01 -6.18608773e-01 -1.46543145e+00 1.78411469e-01
7.22516716e-01 -6.48369193e-01 1.03685224e+00 -4.97281998e-01
6.52438939e-01 5.54272272e-02 -4.12366241e-01 -8.76603603e-01
-6.47470132e-02 -4.16432649e-01 -4.97119099e-01 7.44389296e-01
2.53189266e-01 -5.20128310e-01 9.62077141e-01 6.59463704e-01
-1.00196548e-01 -1.19051123e+00 -9.39514458e-01 -5.45062542e-01
1.64580747e-01 -5.52404523e-01 7.92548478e-01 6.08759701e-01
1.03003994e-01 7.59643912e-01 -2.20516056e-01 1.37505252e-02
7.20003903e-01 2.21152708e-01 8.61961424e-01 -9.80792224e-01
-5.14774263e-01 -6.93599582e-02 -4.42045301e-01 -1.59165847e+00
-6.65278137e-02 -1.12855721e+00 -2.60668755e-01 -1.22684026e+00
-1.66845452e-02 -9.46629345e-01 -6.26225211e-03 6.03719294e-01
2.20319986e-01 1.09474756e-01 1.83385402e-01 5.68449162e-02
-3.79151046e-01 7.58878112e-01 1.36025679e+00 -2.12357864e-01
2.78297246e-01 -3.90643030e-01 -2.38621056e-01 4.11555290e-01
5.92627883e-01 -5.04042566e-01 -7.80178130e-01 -9.69148338e-01
3.89924765e-01 7.61419013e-02 1.37271434e-01 -1.24092925e+00
3.51351321e-01 1.26603231e-01 9.64345858e-02 -7.10765183e-01
7.89196789e-01 -8.26284766e-01 -1.38519496e-01 7.32448936e-01
-5.57354689e-01 3.71392488e-01 3.32085371e-01 1.89406767e-01
-1.49466872e-01 1.99691579e-02 6.68056369e-01 -9.05680954e-02
-1.59275562e-01 4.79524225e-01 1.47174463e-01 -9.33272690e-02
4.90863174e-01 1.19034961e-01 -2.20207814e-02 9.47968364e-02
-4.59094971e-01 -1.10879049e-01 2.24768855e-02 2.04394996e-01
6.98823154e-01 -1.22596765e+00 -5.60466945e-01 6.17010117e-01
-9.07088742e-02 5.59256554e-01 8.00968111e-02 7.41399944e-01
-8.13676536e-01 8.31129193e-01 1.87109748e-03 -9.43008363e-01
-1.07727003e+00 5.56481361e-01 3.11228484e-01 -8.59416008e-01
-7.38425016e-01 9.78679478e-01 -2.52760679e-01 -7.21130013e-01
2.75559664e-01 -1.24804890e+00 4.81406957e-01 -4.97854143e-01
6.09391451e-01 2.02496767e-01 3.01422775e-01 -2.41605118e-01
-2.43879706e-01 5.35818279e-01 -3.36273044e-01 5.33966310e-02
1.43255734e+00 3.51295173e-01 -2.36928001e-01 5.02512217e-01
1.65401804e+00 -4.91805673e-01 -8.06040466e-01 -5.69213510e-01
-1.68988720e-01 -3.33991885e-01 6.06136620e-01 -3.67909342e-01
-1.11175060e+00 1.40074968e+00 6.08012676e-01 -1.52790993e-01
9.57654834e-01 -3.75603378e-01 1.32084978e+00 9.83706594e-01
4.69462723e-01 -8.19433272e-01 7.65106305e-02 6.25043273e-01
6.05894983e-01 -9.75660980e-01 3.32577080e-01 -2.91292548e-01
-1.32271752e-01 9.86600399e-01 3.76226991e-01 -3.14636081e-01
7.72051752e-01 4.77836460e-01 -5.11218011e-01 -1.96567565e-01
-1.39191449e+00 4.48051453e-01 3.63813609e-01 2.16451392e-01
4.72565591e-01 2.89287835e-01 1.60269573e-01 1.20622024e-01
-9.33063567e-01 3.60286564e-01 -3.76739763e-02 7.19090462e-01
-3.59179586e-01 -8.71089160e-01 -5.05556688e-02 7.80687213e-01
-1.99065641e-01 -4.78190362e-01 3.27773005e-01 3.27020675e-01
-6.03214977e-03 4.81203139e-01 3.08782071e-01 -8.23830247e-01
1.69562817e-01 1.96349114e-01 6.74157321e-01 -3.56405526e-01
-7.12600291e-01 -2.85557628e-01 7.36725982e-03 -8.02124083e-01
9.36757624e-02 -1.33000091e-01 -1.20800352e+00 -8.73545587e-01
-1.95932746e-01 2.09113151e-01 1.12678432e+00 9.36617792e-01
6.17158592e-01 4.16386575e-01 2.91983992e-01 -7.05869198e-01
-1.13422740e+00 -1.33659470e+00 -1.60513684e-01 2.33376816e-01
2.19464645e-01 -4.31390971e-01 -2.44498760e-01 1.08463727e-02]
|
[8.727134704589844, 3.5949342250823975]
|
66281cc1-9f78-4e5c-b9bc-66920761fc83
|
action-recognition-with-multi-stream-motion
|
2306.07576
| null |
https://arxiv.org/abs/2306.07576v1
|
https://arxiv.org/pdf/2306.07576v1.pdf
|
Action Recognition with Multi-stream Motion Modeling and Mutual Information Maximization
|
Action recognition has long been a fundamental and intriguing problem in artificial intelligence. The task is challenging due to the high dimensionality nature of an action, as well as the subtle motion details to be considered. Current state-of-the-art approaches typically learn from articulated motion sequences in the straightforward 3D Euclidean space. However, the vanilla Euclidean space is not efficient for modeling important motion characteristics such as the joint-wise angular acceleration, which reveals the driving force behind the motion. Moreover, current methods typically attend to each channel equally and lack theoretical constrains on extracting task-relevant features from the input. In this paper, we seek to tackle these challenges from three aspects: (1) We propose to incorporate an acceleration representation, explicitly modeling the higher-order variations in motion. (2) We introduce a novel Stream-GCN network equipped with multi-stream components and channel attention, where different representations (i.e., streams) supplement each other towards a more precise action recognition while attention capitalizes on those important channels. (3) We explore feature-level supervision for maximizing the extraction of task-relevant information and formulate this into a mutual information loss. Empirically, our approach sets the new state-of-the-art performance on three benchmark datasets, NTU RGB+D, NTU RGB+D 120, and NW-UCLA. Our code is anonymously released at https://github.com/ActionR-Group/Stream-GCN, hoping to inspire the community.
|
['Kui Ren', 'Zhibo Wang', 'Shuang Wu', 'Beibei Zhang', 'Yingda Lyu', 'Zhenguang Liu', 'Haipeng Chen', 'Yuheng Yang']
|
2023-06-13
| null | null | null | null |
['action-recognition-in-videos']
|
['computer-vision']
|
[ 2.89564699e-01 -3.13225299e-01 -4.62266654e-01 -2.18762368e-01
-6.33433342e-01 -3.96024227e-01 7.00695515e-01 -2.77600914e-01
-5.67322493e-01 4.55154955e-01 6.72323823e-01 3.71744037e-02
-1.01749681e-01 -4.69330519e-01 -7.59176910e-01 -8.72212350e-01
-2.61884868e-01 5.12815006e-02 5.98472543e-02 -3.43787044e-01
2.99914747e-01 3.59841615e-01 -1.43132222e+00 2.42455840e-01
7.28528440e-01 1.16241968e+00 1.71368942e-01 7.19618142e-01
1.40206665e-01 9.10485029e-01 -2.01722383e-01 -1.48357645e-01
3.09499592e-01 -6.07641697e-01 -7.88926423e-01 2.89777040e-01
1.73726454e-01 -4.29540247e-01 -7.49769211e-01 8.87738764e-01
3.97717088e-01 2.82600880e-01 5.41368783e-01 -1.36607027e+00
-4.23354745e-01 2.48169050e-01 -6.26728654e-01 3.13223720e-01
3.09860677e-01 6.14418983e-01 1.09062409e+00 -6.67764604e-01
6.06035531e-01 1.16846132e+00 2.62952179e-01 6.53154492e-01
-9.44667876e-01 -3.05930018e-01 4.67441112e-01 6.26447201e-01
-1.05563402e+00 -3.62489849e-01 9.51169431e-01 -4.62475717e-01
7.76884377e-01 2.16336370e-01 7.97266841e-01 1.53616655e+00
1.14512499e-02 1.31493545e+00 6.48510754e-01 4.61289585e-02
2.87951559e-01 -5.95261455e-01 -3.69770080e-02 4.73486781e-01
-7.41288513e-02 -3.83820161e-02 -7.58917153e-01 1.94089770e-01
9.11299944e-01 1.78920791e-01 -5.02434671e-01 -5.01668572e-01
-1.46176195e+00 7.48161018e-01 4.76444393e-01 1.96695819e-01
-4.88630772e-01 4.57605451e-01 5.38984358e-01 8.92072264e-03
3.04167271e-01 2.69182414e-01 -4.94440973e-01 -8.88832688e-01
-4.52084750e-01 4.31277484e-01 3.69029701e-01 7.75371909e-01
5.60577869e-01 4.06125374e-02 -1.13562010e-01 4.87850487e-01
3.16076100e-01 3.56609821e-01 5.61215222e-01 -1.09512675e+00
7.56660938e-01 6.33128881e-01 1.13828763e-01 -9.81576681e-01
-4.10623610e-01 -3.23452175e-01 -9.32464719e-01 8.86710212e-02
5.80403030e-01 -1.71270698e-01 -6.74706340e-01 1.80776155e+00
4.02686894e-01 2.96898514e-01 -3.16673443e-02 1.41700935e+00
4.48074192e-01 5.16475856e-01 -4.98310737e-02 6.29056022e-02
1.23965645e+00 -1.15663993e+00 -6.06582761e-01 -3.91303003e-01
7.16560066e-01 -3.01213175e-01 1.08325636e+00 3.31892461e-01
-9.96007621e-01 -5.90689421e-01 -8.82142127e-01 -2.43282646e-01
-1.16232701e-01 2.43963391e-01 7.32909441e-01 1.49833262e-01
-6.33180797e-01 8.02878201e-01 -1.33210909e+00 -2.66616911e-01
5.25533199e-01 1.56996489e-01 -4.64126498e-01 -1.83476597e-01
-1.12780392e+00 5.52147329e-01 1.06734119e-01 3.19053352e-01
-8.48887205e-01 -3.98916602e-01 -9.98942673e-01 -1.75902277e-01
5.56338847e-01 -6.98719978e-01 1.11394382e+00 -9.45551991e-01
-1.53062522e+00 3.96563381e-01 -2.34157175e-01 -2.65784800e-01
8.43119681e-01 -6.21805787e-01 -7.03192130e-02 3.28567713e-01
8.12379569e-02 5.57921946e-01 8.81780446e-01 -9.00860727e-01
-6.05978668e-01 -5.25363684e-01 2.31366992e-01 3.94833654e-01
-3.48028898e-01 -2.50937015e-01 -6.33988023e-01 -8.61740470e-01
9.07657593e-02 -9.97421980e-01 -3.23639870e-01 3.08666080e-01
-3.54547292e-01 -1.66961402e-01 7.57162154e-01 -5.52503049e-01
1.15002692e+00 -2.29877877e+00 6.94985569e-01 -3.21234584e-01
1.39601275e-01 3.24426085e-01 -3.99574488e-02 3.61391574e-01
-3.08348797e-02 -8.14550221e-02 -2.56336331e-01 -4.54700708e-01
1.19803607e-01 3.67923737e-01 -1.39532804e-01 6.62202299e-01
4.83390629e-01 1.06562054e+00 -1.07392478e+00 -1.71132758e-01
3.32030356e-01 6.49932861e-01 -6.35563850e-01 9.22081098e-02
-1.37552753e-01 8.54634821e-01 -8.59903574e-01 7.06836224e-01
2.77956545e-01 -3.51672381e-01 -1.50396377e-01 -2.61879802e-01
-3.72019932e-02 2.39225715e-01 -1.18772113e+00 2.11153030e+00
-1.83992594e-01 5.83589792e-01 5.74424267e-02 -1.14858878e+00
6.06880307e-01 1.44240543e-01 8.15970540e-01 -5.65994143e-01
2.46490628e-01 4.16795909e-02 3.19672674e-02 -7.80296326e-01
4.07738984e-01 1.80739224e-01 -6.19350979e-03 3.36547196e-01
-5.97597957e-02 2.47954637e-01 9.56625566e-02 1.19659156e-01
1.26115251e+00 4.83558089e-01 1.64396808e-01 9.48210359e-02
4.21107113e-01 -1.81172490e-01 6.95994318e-01 4.39344049e-01
-6.06517136e-01 7.66226888e-01 6.34746432e-01 -4.31011647e-01
-8.70926023e-01 -7.14245558e-01 1.55147731e-01 8.51962268e-01
2.14085057e-01 -5.10829568e-01 -6.33409262e-01 -7.10500002e-01
6.76968182e-03 3.29797626e-01 -8.32521617e-01 -2.01728716e-01
-7.18151867e-01 -5.45759737e-01 3.50894362e-01 8.24507833e-01
5.83076000e-01 -1.02625942e+00 -9.34970498e-01 1.96128875e-01
-4.57610637e-01 -1.25389910e+00 -6.45745158e-01 3.28054503e-02
-8.85909915e-01 -1.18743110e+00 -8.28612506e-01 -1.55697420e-01
3.95700455e-01 3.58883351e-01 7.24906027e-01 -8.39875340e-02
-3.02407980e-01 5.42488396e-01 -7.34293997e-01 -1.87234268e-01
2.35270903e-01 6.38270602e-02 8.16518515e-02 4.45919007e-01
3.19071293e-01 -6.00201070e-01 -9.28771377e-01 3.72174740e-01
-8.96146417e-01 6.00577965e-02 6.05671704e-01 7.17216790e-01
5.91517091e-01 -1.53043240e-01 2.79678881e-01 -3.52028102e-01
1.76794559e-01 -5.63122869e-01 -2.75106989e-02 -2.40793094e-01
2.26891618e-02 3.85701004e-03 5.82598388e-01 -4.72559005e-01
-6.87376142e-01 2.40611792e-01 -1.56521946e-01 -7.77332067e-01
-2.68176407e-01 3.55922639e-01 -3.68394107e-01 1.35297805e-01
3.82942885e-01 3.97802621e-01 6.74932674e-02 -4.60831910e-01
3.54875028e-01 3.48722875e-01 5.39008915e-01 -6.00742221e-01
6.51419461e-01 8.84812415e-01 5.95916137e-02 -9.22018230e-01
-7.06521153e-01 -6.37565076e-01 -8.79392564e-01 -3.93972933e-01
1.07082653e+00 -8.85441184e-01 -7.37753034e-01 7.81974196e-01
-1.02244902e+00 -5.51321805e-01 -3.61370057e-01 6.74550295e-01
-9.30770755e-01 5.27282298e-01 -5.78555763e-01 -8.02431047e-01
-2.21043956e-02 -1.31918705e+00 1.37064457e+00 1.03458233e-01
-1.46875992e-01 -7.16715276e-01 1.85835455e-02 5.23087680e-01
2.22555339e-01 5.16740561e-01 4.31530952e-01 -2.13109508e-01
-6.19581521e-01 -1.12904571e-01 -1.86428502e-01 3.35984558e-01
1.57878712e-01 -8.81807059e-02 -8.76030385e-01 -1.24689318e-01
-2.23121587e-02 -3.82063597e-01 1.11012816e+00 4.29209799e-01
1.26341331e+00 -2.63844699e-01 -4.64206412e-02 7.39371777e-01
1.10265362e+00 4.67684306e-02 7.34404206e-01 4.15589750e-01
1.09796059e+00 6.02176428e-01 7.86669672e-01 7.99719453e-01
5.08972764e-01 7.72169173e-01 7.14282393e-01 1.70877293e-01
3.05869132e-02 -2.43988529e-01 4.77256417e-01 6.85040236e-01
-4.81065482e-01 -1.01900794e-01 -7.43134022e-01 5.16912341e-01
-2.22069526e+00 -9.67242897e-01 -2.16709465e-01 1.98385119e+00
5.58383524e-01 1.26661494e-01 2.90527701e-01 2.47062415e-01
3.27423394e-01 6.24800146e-01 -8.51295471e-01 9.84489396e-02
-9.52854529e-02 -2.04558507e-01 3.76386881e-01 2.26630047e-01
-1.31032717e+00 7.90083468e-01 4.61186028e+00 6.26945019e-01
-1.20090759e+00 -1.23128369e-01 5.56764781e-01 -4.04371977e-01
-1.16238436e-02 -2.37379655e-01 -6.58465266e-01 5.93591213e-01
6.33460581e-01 6.18494768e-03 3.05074066e-01 7.46819794e-01
4.73879218e-01 -8.04375336e-02 -1.15747857e+00 1.09194231e+00
1.14888605e-02 -1.04603410e+00 -7.48189241e-02 2.73640066e-01
4.55447197e-01 2.03596294e-01 -1.53732486e-02 1.92135364e-01
-1.06672071e-01 -9.06080127e-01 9.03954744e-01 6.19035006e-01
6.05842829e-01 -6.91031039e-01 4.27479863e-01 3.84149432e-01
-1.39161992e+00 -1.60406098e-01 -1.94832206e-01 -3.59188974e-01
3.63904148e-01 3.55131924e-01 -5.84611259e-02 6.66720688e-01
7.16701388e-01 1.33393836e+00 -3.47015947e-01 8.93137515e-01
-2.82367527e-01 4.86037374e-01 -2.12756157e-01 1.97898056e-02
6.26653016e-01 -2.80945808e-01 7.12818444e-01 1.00673199e+00
2.33659372e-01 2.97771394e-01 1.53896138e-01 6.26502216e-01
1.36079639e-01 -1.45634949e-01 -6.08314514e-01 -2.30196759e-01
-9.37011614e-02 1.01306832e+00 -5.68392873e-01 -1.45975754e-01
-5.51798522e-01 1.19012570e+00 2.83262819e-01 4.61634576e-01
-8.49794626e-01 -1.95990074e-02 1.25551605e+00 -5.47049195e-02
5.75919509e-01 -5.93002200e-01 -9.51157734e-02 -1.48451948e+00
4.08445925e-01 -8.45964730e-01 3.18049997e-01 -5.02486348e-01
-1.10746479e+00 3.02227288e-01 -1.86180040e-01 -1.54950976e+00
-3.31438601e-01 -8.03329706e-01 -5.53336799e-01 5.77482045e-01
-1.56593275e+00 -9.64210689e-01 -3.89554530e-01 7.88567305e-01
7.88288593e-01 1.99510396e-01 5.19257784e-01 4.11940038e-01
-7.75720835e-01 3.84901226e-01 -2.01019291e-02 3.79338682e-01
4.81055111e-01 -1.15923107e+00 5.12461245e-01 8.26931655e-01
1.00904047e-01 3.12630624e-01 5.85500300e-01 -4.44536954e-01
-1.91212678e+00 -1.06311631e+00 5.98125100e-01 -6.43000305e-01
9.03571546e-01 -3.83863240e-01 -8.74899566e-01 6.52748406e-01
-2.51476347e-01 4.09439594e-01 4.71742600e-01 -1.99018329e-01
-2.89772838e-01 1.69454575e-01 -6.16672039e-01 6.19932413e-01
1.42239141e+00 -5.18687129e-01 -2.25969732e-01 2.59614229e-01
5.89454055e-01 -3.95165443e-01 -7.76517689e-01 2.95852721e-01
6.58872128e-01 -9.95330513e-01 9.29382741e-01 -8.20705354e-01
7.83848882e-01 -2.82587975e-01 -2.69735247e-01 -1.19032419e+00
-1.64022967e-01 -6.81420207e-01 -4.34646547e-01 7.83275127e-01
1.44001413e-02 -3.23430657e-01 9.50644195e-01 5.08529842e-01
-1.22011252e-01 -1.17550182e+00 -1.06188989e+00 -7.59068191e-01
-8.76576602e-02 -8.82356048e-01 3.05538297e-01 9.49900031e-01
2.67555434e-02 1.83657795e-01 -6.85312092e-01 5.58359548e-03
5.99359810e-01 5.90807816e-04 9.32199240e-01 -8.49450648e-01
-4.81742442e-01 -6.19130969e-01 -7.73365557e-01 -1.52429771e+00
4.13257256e-02 -5.10999739e-01 -3.33289104e-03 -1.41616333e+00
3.31550129e-02 -6.86911270e-02 -1.73096806e-01 4.28309560e-01
-3.98101956e-01 5.45203686e-02 3.81836802e-01 2.80908018e-01
-7.04592884e-01 1.09670556e+00 1.53145540e+00 -1.37037352e-01
-4.85309698e-02 1.26966044e-01 -5.27041912e-01 7.83838093e-01
6.57849610e-01 -1.52776480e-01 -4.20423090e-01 -6.09789670e-01
-9.90250483e-02 9.50838327e-02 6.75575435e-01 -1.02292144e+00
2.27978900e-01 -2.49855563e-01 3.17017347e-01 -4.11981612e-01
5.75984895e-01 -7.52124846e-01 -2.52781034e-01 3.47813159e-01
-3.03565800e-01 -1.50875105e-02 -3.57435457e-02 8.77548635e-01
-3.26580167e-01 1.46247745e-01 5.12755156e-01 -1.85321480e-01
-1.01329684e+00 6.88926756e-01 -1.57132745e-01 2.34726638e-01
9.70540762e-01 -2.79472828e-01 -1.79251984e-01 -6.02552354e-01
-6.22632086e-01 3.06364000e-01 3.91299188e-01 6.74719632e-01
5.24619401e-01 -1.45037329e+00 -7.27135062e-01 1.88092470e-01
2.26220116e-01 1.35422528e-01 4.78667021e-01 1.13799036e+00
-2.62741864e-01 3.60832870e-01 -2.56485313e-01 -6.50995374e-01
-9.63885963e-01 3.89219224e-01 2.02935427e-01 -1.41538680e-01
-9.24816847e-01 8.17578018e-01 3.69502604e-01 -1.18249148e-01
3.86861324e-01 -4.20776069e-01 -2.15003848e-01 1.70940787e-01
6.80870175e-01 4.63889211e-01 -1.34251148e-01 -9.17390227e-01
-4.30079281e-01 7.42156267e-01 1.06070064e-01 -3.48396674e-02
1.30766976e+00 -9.75690410e-02 3.44394267e-01 5.71547151e-01
1.41198289e+00 -5.57728469e-01 -1.89040065e+00 -2.47132182e-01
-1.35677576e-01 -7.39213765e-01 -3.01652215e-02 -3.35278302e-01
-1.28551590e+00 1.10848236e+00 2.94183433e-01 -1.74056552e-02
1.19960570e+00 -2.32972428e-02 8.97574902e-01 2.78683364e-01
3.35687011e-01 -1.09487987e+00 2.99949288e-01 5.84377766e-01
9.41679597e-01 -1.39471233e+00 -5.86262979e-02 -1.60744235e-01
-7.81840324e-01 1.04798090e+00 6.05357826e-01 -1.19255625e-01
5.48605800e-01 -2.86081340e-02 -1.85939297e-02 -2.02656522e-01
-7.60911226e-01 -3.93724531e-01 2.85966188e-01 4.62646186e-01
3.37470651e-01 4.54243831e-02 -2.50476807e-01 5.32296240e-01
5.86023778e-02 -3.29855345e-02 3.01356852e-01 1.17173612e+00
-1.69208542e-01 -9.08751428e-01 -9.78359878e-02 2.83813983e-01
-3.26710701e-01 1.98146909e-01 -3.43106598e-01 7.87203670e-01
-1.33348759e-02 7.78954983e-01 7.53477402e-03 -5.31199157e-01
5.14053166e-01 -1.34756565e-01 2.92847455e-01 -2.20526561e-01
-2.22391635e-01 8.83863047e-02 5.17408807e-05 -1.25795710e+00
-5.60318887e-01 -1.09897172e+00 -1.27700555e+00 -1.72321215e-01
1.88938692e-01 -2.20407039e-01 6.70074463e-01 1.00463784e+00
4.54310834e-01 6.26164854e-01 5.50030947e-01 -1.29672980e+00
-6.03356242e-01 -8.57819498e-01 -5.85316658e-01 6.63118839e-01
6.00683689e-01 -8.72036755e-01 -4.65447426e-01 4.28472459e-02]
|
[7.922214984893799, 0.25923222303390503]
|
28223de2-87a1-4456-8d62-280cd68a50bf
|
dcp-nas-discrepant-child-parent-neural
|
2306.15390
| null |
https://arxiv.org/abs/2306.15390v1
|
https://arxiv.org/pdf/2306.15390v1.pdf
|
DCP-NAS: Discrepant Child-Parent Neural Architecture Search for 1-bit CNNs
|
Neural architecture search (NAS) proves to be among the effective approaches for many tasks by generating an application-adaptive neural architecture, which is still challenged by high computational cost and memory consumption. At the same time, 1-bit convolutional neural networks (CNNs) with binary weights and activations show their potential for resource-limited embedded devices. One natural approach is to use 1-bit CNNs to reduce the computation and memory cost of NAS by taking advantage of the strengths of each in a unified framework, while searching the 1-bit CNNs is more challenging due to the more complicated processes involved. In this paper, we introduce Discrepant Child-Parent Neural Architecture Search (DCP-NAS) to efficiently search 1-bit CNNs, based on a new framework of searching the 1-bit model (Child) under the supervision of a real-valued model (Parent). Particularly, we first utilize a Parent model to calculate a tangent direction, based on which the tangent propagation method is introduced to search the optimized 1-bit Child. We further observe a coupling relationship between the weights and architecture parameters existing in such differentiable frameworks. To address the issue, we propose a decoupled optimization method to search an optimized architecture. Extensive experiments demonstrate that our DCP-NAS achieves much better results than prior arts on both CIFAR-10 and ImageNet datasets. In particular, the backbones achieved by our DCP-NAS achieve strong generalization performance on person re-identification and object detection.
|
['Guodong Guo', 'Tian Wang', 'Baochang Zhang', "Li'an Zhuo", 'Xianbin Cao', 'Sheng Xu', 'Yanjing Li']
|
2023-06-27
| null | null | null | null |
['person-re-identification', 'architecture-search']
|
['computer-vision', 'methodology']
|
[ 9.43092406e-02 -2.15023220e-01 -2.54560530e-01 -4.65514779e-01
-2.50359148e-01 -3.91329318e-01 4.11553979e-02 -3.51345628e-01
-7.41378605e-01 2.97930270e-01 -2.98332423e-01 -3.06400537e-01
-1.56464159e-01 -8.12463462e-01 -6.90068424e-01 -7.56970406e-01
2.08774254e-01 1.35801390e-01 7.25436881e-02 -2.20745474e-01
-1.90303959e-02 6.81193709e-01 -1.32744110e+00 4.78681400e-02
6.84457481e-01 1.49066341e+00 1.58558771e-01 4.43421960e-01
-6.09199051e-03 1.49751157e-01 -4.39233065e-01 -9.19869244e-01
5.13220668e-01 1.31527074e-02 -6.08025551e-01 -5.04282653e-01
7.13938594e-01 -3.20707142e-01 -5.17675817e-01 1.37553084e+00
6.21666133e-01 1.07004501e-01 3.78148854e-01 -1.30057311e+00
-6.81293666e-01 7.12505460e-01 -2.04410210e-01 2.66790420e-01
-2.58643657e-01 3.67174506e-01 1.03173423e+00 -8.96841347e-01
8.71530455e-03 1.31011879e+00 9.13365483e-01 8.65684748e-01
-1.12463868e+00 -8.57118726e-01 4.32224393e-01 3.71059507e-01
-1.54145741e+00 -3.89210731e-01 8.09504390e-01 -9.61099640e-02
1.08143806e+00 1.02561273e-01 7.47515619e-01 1.15581536e+00
-2.39489973e-01 6.88301444e-01 4.98328090e-01 -2.53390044e-01
2.53774107e-01 -6.09688871e-02 3.47420454e-01 9.09878612e-01
4.67198104e-01 4.06056037e-03 -6.00396276e-01 -7.30247572e-02
9.07106102e-01 2.36649230e-01 -3.74865860e-01 -4.41238470e-02
-9.46899414e-01 6.01686180e-01 9.22976911e-01 2.68297285e-01
-3.92346084e-01 4.84499723e-01 2.58331925e-01 1.89808339e-01
-6.88161775e-02 4.19694692e-01 -5.53889930e-01 -5.91928735e-02
-7.11469412e-01 1.99677736e-01 4.98727888e-01 7.27067530e-01
5.98340452e-01 2.73432702e-01 -3.35086852e-01 9.59764302e-01
3.14681083e-01 3.10289681e-01 6.04672611e-01 -6.72803342e-01
5.70623398e-01 8.15158844e-01 -2.51616240e-01 -9.79411364e-01
-3.92565221e-01 -8.40207279e-01 -1.27325237e+00 1.21261083e-01
4.36878085e-01 -8.38398263e-02 -8.88865650e-01 1.95147538e+00
5.49097881e-02 2.60211110e-01 6.48568477e-03 9.57877576e-01
6.95143044e-01 4.78923529e-01 7.83564597e-02 3.92066598e-01
1.55985904e+00 -1.14558113e+00 -2.39784643e-01 -2.78909981e-01
3.63420278e-01 -1.22282885e-01 1.01957560e+00 3.05055857e-01
-1.15866780e+00 -8.64607036e-01 -1.26602030e+00 -1.22770473e-01
-5.55483937e-01 5.01821816e-01 5.82348943e-01 9.23021138e-01
-1.22495592e+00 5.35422027e-01 -8.49968433e-01 -4.38619358e-03
6.80540144e-01 6.92820787e-01 -6.02524504e-02 2.12964073e-01
-1.04018188e+00 5.11426449e-01 5.78481793e-01 5.25932670e-01
-5.34652114e-01 -7.37330675e-01 -7.28644133e-01 5.84423304e-01
6.58226088e-02 -6.90227032e-01 1.09621739e+00 -8.92183840e-01
-1.53743088e+00 4.60391551e-01 -7.64812082e-02 -7.67322898e-01
2.89269745e-01 -2.38509849e-02 -4.96163428e-01 9.98254046e-02
-3.62300247e-01 9.75998104e-01 9.03341830e-01 -6.44152462e-01
-6.39376223e-01 -4.18748379e-01 -1.73319187e-02 -5.08025810e-02
-1.18212438e+00 -1.53516904e-01 -7.12274849e-01 -8.10295284e-01
2.33217791e-01 -8.89990151e-01 -2.64342576e-01 3.00772518e-01
-3.91638875e-01 -3.38074565e-01 3.75630081e-01 -3.63467366e-01
1.40422094e+00 -2.17842412e+00 1.46923199e-01 4.31638211e-01
2.66243786e-01 6.58048749e-01 -3.15951705e-01 -3.90250862e-01
-5.61941834e-03 1.45841554e-01 -7.26763234e-02 -3.74413639e-01
3.14037912e-02 4.01633792e-02 -2.44083270e-01 1.10442936e-01
4.27369088e-01 1.17716897e+00 -6.09741628e-01 -2.94928700e-01
-1.87220454e-01 7.59183288e-01 -7.81968832e-01 -1.81452841e-01
4.55504581e-02 -8.00815970e-02 -4.74700421e-01 8.75972450e-01
6.08930469e-01 -5.16546667e-01 1.19335353e-01 -7.28457928e-01
-7.87064359e-02 3.02625477e-01 -1.04284847e+00 1.51947927e+00
-2.31597438e-01 5.67359567e-01 -3.24191488e-02 -1.07719600e+00
1.03381050e+00 1.45096555e-01 1.85722664e-01 -8.06970775e-01
2.26449654e-01 3.62047851e-01 2.54432540e-02 -1.15535416e-01
3.53693187e-01 4.12998617e-01 1.57166913e-01 -7.03023281e-03
6.82949200e-02 5.22698581e-01 -1.23041615e-01 -2.99652785e-01
8.37107897e-01 -2.61917531e-01 -7.62368366e-02 -1.90538734e-01
6.34388506e-01 -2.94180930e-01 6.43687844e-01 8.74335885e-01
-4.36318308e-01 4.64191914e-01 1.58742636e-01 -8.92259717e-01
-7.86232889e-01 -9.23843980e-01 -2.79461205e-01 9.29471850e-01
1.00519240e-01 -2.33853787e-01 -8.79705727e-01 -5.12681067e-01
-1.12024061e-01 2.11877778e-01 -3.88523072e-01 -2.60689080e-01
-1.01009691e+00 -8.51543963e-01 1.01459563e+00 7.40617096e-01
1.03739250e+00 -8.63869905e-01 -9.42461967e-01 2.06507623e-01
3.77656939e-03 -1.11031926e+00 -6.31759226e-01 3.44159037e-01
-9.66985703e-01 -7.13905394e-01 -1.03438163e+00 -1.14274073e+00
6.98767364e-01 1.85408313e-02 8.84022832e-01 2.80360639e-01
-2.62966931e-01 5.77817261e-02 2.23348532e-02 -1.95580840e-01
1.93735629e-01 5.14798224e-01 1.23008855e-01 2.24040911e-01
3.22864920e-01 -5.72547853e-01 -9.75364327e-01 2.41097346e-01
-7.07302451e-01 -1.35338018e-02 7.95768678e-01 9.92555261e-01
6.42306566e-01 4.95598018e-02 4.73976284e-01 -3.12867075e-01
5.70903063e-01 -1.36775300e-01 -7.87788928e-01 5.78335881e-01
-7.97675908e-01 2.65478849e-01 6.67473316e-01 -7.92509437e-01
-6.91339612e-01 2.45852292e-01 -1.94721177e-01 -5.89090168e-01
1.13307200e-01 3.92134637e-01 -1.22064784e-01 -3.83154005e-01
5.96394837e-01 2.92166680e-01 -1.86894029e-01 -6.16247773e-01
5.69091327e-02 4.59881574e-01 8.79837751e-01 -5.65889299e-01
6.61912978e-01 2.52292603e-01 -4.93432209e-02 -4.68132675e-01
-4.65567976e-01 -5.33635169e-02 -4.32908326e-01 8.25076997e-02
8.49106312e-01 -8.86797786e-01 -1.15109074e+00 7.52888381e-01
-1.40912807e+00 -2.82862604e-01 5.48690446e-02 4.39059585e-01
1.17590897e-01 1.46813050e-01 -6.29952848e-01 -6.45866454e-01
-8.07545304e-01 -1.33040559e+00 7.50889897e-01 5.94436467e-01
1.10152364e-01 -6.87884212e-01 -3.74309003e-01 -1.83407273e-02
7.31940329e-01 -2.57821381e-01 1.11536014e+00 -5.61314464e-01
-6.84535146e-01 -2.21822977e-01 -5.01550376e-01 3.33738148e-01
-1.62695765e-01 -3.81733894e-01 -9.94759798e-01 -3.62225264e-01
-8.02972466e-02 -5.41817136e-02 9.97804999e-01 2.81295747e-01
1.67554045e+00 -5.27566612e-01 -2.72089005e-01 1.10083628e+00
1.37643206e+00 1.61747992e-01 5.28649867e-01 3.74752671e-01
6.77314222e-01 1.65538654e-01 -1.44120216e-01 2.82205611e-01
3.94540101e-01 8.77897322e-01 5.99420965e-01 4.00959002e-03
-1.94685876e-01 -1.99526042e-01 2.01508090e-01 5.98655641e-01
-1.73673123e-01 -4.55671363e-02 -9.13836777e-01 4.12936479e-01
-1.74285722e+00 -7.84926772e-01 2.73677140e-01 2.16511178e+00
9.15119231e-01 2.65082777e-01 8.53376277e-03 4.14358638e-02
6.42505705e-01 -7.31054470e-02 -8.79477262e-01 -8.49726796e-02
-1.66714355e-01 4.73023266e-01 6.31421089e-01 9.32760313e-02
-1.00440621e+00 7.63950467e-01 5.99870682e+00 7.31952846e-01
-1.32889426e+00 -4.04965319e-02 9.40920651e-01 -2.24299252e-01
-1.44143971e-02 -4.10464704e-01 -1.40111685e+00 6.27538204e-01
7.02300429e-01 2.80118793e-01 6.90270245e-01 1.14750504e+00
-3.69834483e-01 6.46365821e-01 -1.12665284e+00 1.61522377e+00
-9.71974656e-02 -1.44768965e+00 2.63590485e-01 -8.02626181e-03
5.51722050e-01 8.13995674e-02 3.54823470e-01 3.54669034e-01
-1.55317903e-01 -1.16451132e+00 9.00998116e-01 3.67222100e-01
7.87169993e-01 -5.81258357e-01 6.06547534e-01 6.92159384e-02
-1.49479413e+00 -4.21746850e-01 -5.03012538e-01 2.04783931e-01
-1.94650590e-01 3.26793015e-01 -4.41127807e-01 8.20571557e-02
1.12129676e+00 4.37593192e-01 -6.68022931e-01 1.27442908e+00
1.35118933e-02 3.42761010e-01 -3.30473483e-01 -3.47836584e-01
1.80020288e-01 -8.02161470e-02 2.09496602e-01 1.17686284e+00
5.15047491e-01 -8.34756792e-02 -1.76913708e-01 1.22214162e+00
-4.18120414e-01 -3.01564872e-01 -7.80516714e-02 1.04118370e-01
8.29077184e-01 1.15554237e+00 -5.05925834e-01 -3.16788167e-01
-2.31223598e-01 9.81624961e-01 5.41588783e-01 5.42659700e-01
-9.14971292e-01 -5.35960853e-01 9.05531943e-01 -2.83022642e-01
4.93362725e-01 -2.06584707e-01 -6.02363110e-01 -9.98678505e-01
3.54012370e-01 -8.86445701e-01 3.58905792e-01 -4.19176728e-01
-1.13801336e+00 8.81754637e-01 -2.26609558e-01 -1.06038105e+00
2.66627339e-03 -1.03406620e+00 -4.77294505e-01 8.69004548e-01
-1.93692243e+00 -9.42125142e-01 -4.00127083e-01 7.55437553e-01
4.95843172e-01 -3.41144860e-01 7.58289576e-01 5.61141193e-01
-1.03512502e+00 1.38204527e+00 -4.23866540e-01 4.78759140e-01
2.37595856e-01 -7.96078980e-01 6.26818180e-01 8.35013151e-01
1.60683036e-01 8.46415877e-01 -5.16605042e-02 -1.75379515e-01
-1.65839362e+00 -1.04127395e+00 7.31573462e-01 8.18752348e-02
3.95500034e-01 -2.49230206e-01 -8.45822871e-01 4.83765274e-01
-1.98280647e-01 1.99833617e-01 3.09617877e-01 -3.74743529e-03
-7.09137380e-01 -4.31424946e-01 -9.11940277e-01 9.79607582e-01
1.24385571e+00 -4.59493041e-01 -2.10173547e-01 -1.66216847e-02
8.60646307e-01 -3.39775234e-01 -5.51545799e-01 4.41613883e-01
7.94972360e-01 -8.15223634e-01 1.46293271e+00 -4.70690638e-01
1.02693126e-01 -2.18137905e-01 -1.30228326e-01 -8.47170353e-01
-3.71918827e-01 -5.69863081e-01 -2.21280038e-01 9.32761967e-01
5.71314752e-01 -1.02144468e+00 9.15379047e-01 9.12939072e-01
-2.02768698e-01 -1.12264836e+00 -1.15453446e+00 -8.23063612e-01
-9.54007283e-02 -4.39774752e-01 1.05083466e+00 5.36943674e-01
-4.07558620e-01 3.89025360e-03 -7.88244680e-02 4.26000118e-01
5.68390787e-01 -1.27170473e-01 3.84111345e-01 -1.29259372e+00
-5.13266981e-01 -1.09247434e+00 -4.61243302e-01 -1.52956879e+00
8.90678074e-03 -8.03102016e-01 -6.96947202e-02 -9.46841240e-01
-9.82305929e-02 -7.58785367e-01 -7.81981587e-01 8.42318654e-01
-1.55078052e-02 3.71545017e-01 2.84087926e-01 3.87034327e-01
-3.41347456e-01 5.56158245e-01 8.22260380e-01 -4.37606037e-01
-2.73329854e-01 5.14712296e-02 -6.16986454e-01 7.71221280e-01
8.32441092e-01 -3.26426566e-01 -2.15626553e-01 -9.23124611e-01
4.02786106e-01 -3.99268746e-01 7.79554129e-01 -1.26659334e+00
7.79942930e-01 2.47480690e-01 5.25992274e-01 -3.91832173e-01
6.19993031e-01 -8.30404639e-01 -7.56351501e-02 7.20496893e-01
-4.08712596e-01 3.56213212e-01 2.08248556e-01 3.52570862e-01
-1.50016308e-01 -4.25551772e-01 7.71709144e-01 4.32755835e-02
-7.32067764e-01 5.67115068e-01 2.14553699e-01 -2.40115076e-01
5.43280244e-01 -4.39412564e-01 -3.04765016e-01 1.28521577e-01
-3.87685120e-01 1.12945914e-01 -4.97671477e-02 3.14798445e-01
8.45556974e-01 -1.46042359e+00 -3.32077712e-01 5.40223300e-01
-7.70402700e-02 1.28897533e-01 1.26070445e-02 6.72951341e-01
-3.37747216e-01 6.12333775e-01 -2.77559519e-01 -5.72760344e-01
-9.66444194e-01 2.07339346e-01 6.22104824e-01 -2.02156126e-01
-4.31470215e-01 1.17527366e+00 9.47220176e-02 -2.43302315e-01
7.99100757e-01 -4.93412822e-01 -1.18534565e-01 -2.57929713e-01
7.42253840e-01 2.91117072e-01 1.44060805e-01 -2.52145857e-01
-3.26659560e-01 7.08926499e-01 -7.14571700e-02 1.33349344e-01
1.35072291e+00 2.26576760e-01 -1.57128312e-02 -2.69916564e-01
1.17345500e+00 -5.06448686e-01 -1.41417658e+00 -3.39754552e-01
6.29079109e-03 -1.69819564e-01 1.46830022e-01 -6.26453757e-01
-1.60783172e+00 8.10352206e-01 9.68173444e-01 -6.01625107e-02
1.26197922e+00 -2.25967184e-01 1.03922296e+00 7.89026141e-01
2.59557158e-01 -9.50468659e-01 1.93125546e-01 4.79186356e-01
6.68206930e-01 -1.05201542e+00 -4.54442233e-01 -2.63064533e-01
-1.66341588e-01 1.47616303e+00 8.19580376e-01 1.19900340e-02
7.14358568e-01 1.15681261e-01 -2.33665869e-01 -1.95191309e-01
-3.90531540e-01 -3.96143384e-02 4.59283471e-01 4.40202296e-01
2.28766389e-02 -2.26545837e-02 1.81473508e-01 9.52288628e-01
-2.82324195e-01 -1.28647145e-02 -2.92207569e-01 5.08739471e-01
-2.08208159e-01 -1.11575425e+00 -1.98850140e-01 2.34711260e-01
-3.32004160e-01 -4.98842984e-01 -4.61436473e-02 3.88559014e-01
1.12125307e-01 6.35250628e-01 1.99395001e-01 -5.62940240e-01
2.78551906e-01 1.26138493e-01 2.76682079e-01 -1.68284655e-01
-8.72720420e-01 -3.85598212e-01 -4.04053092e-01 -5.56257427e-01
-8.16686898e-02 -2.43728921e-01 -1.25219083e+00 -1.95528805e-01
-2.62446165e-01 -3.60012472e-01 9.88962650e-01 6.54201508e-01
7.70485461e-01 4.58640903e-01 3.91882658e-01 -8.81019533e-01
-9.49268699e-01 -4.96113420e-01 -3.07107326e-02 1.57876566e-01
2.92447507e-01 -4.20791298e-01 -1.33614406e-01 -1.98819637e-01]
|
[8.629400253295898, 2.962913751602173]
|
c906fa3a-2c49-40ea-b8e7-b5919e46166b
|
classification-of-brain-tumours-in-mr-images
|
2105.14071
| null |
https://arxiv.org/abs/2105.14071v2
|
https://arxiv.org/pdf/2105.14071v2.pdf
|
Classification of Brain Tumours in MR Images using Deep Spatiospatial Models
|
A brain tumour is a mass or cluster of abnormal cells in the brain, which has the possibility of becoming life-threatening because of its ability to invade neighbouring tissues and also form metastases. An accurate diagnosis is essential for successful treatment planning and magnetic resonance imaging is the principal imaging modality for diagnostic of brain tumours and their extent. Deep Learning methods in computer vision applications have shown significant improvement in recent years, most of which can be credited to the fact that a sizeable amount of data is available to train models on, and the improvements in the model architectures yielding better approximations in a supervised setting. Classifying tumours using such deep learning methods has made significant progress with the availability of open datasets with reliable annotations. Typically those methods are either 3D models, which use 3D volumetric MRIs or even 2D models considering each slice separately. However, by treating the slice spatial dimension separately, spatiotemporal models can be employed as spatiospatial models for this task. These models have the capabilities of learning specific spatial and temporal relationship, while reducing computational costs. This paper uses two spatiotemporal models, ResNet (2+1)D and ResNet Mixed Convolution, to classify different types of brain tumours. It was observed that both these models performed superior to the pure 3D convolutional model, ResNet18. Furthermore, it was also observed that pre-training the models on a different, even unrelated dataset before training them for the task of tumour classification improves the performance. Finally, Pre-trained ResNet Mixed Convolution was observed to be the best model in these experiments, achieving a macro F1-score of 0.93 and a test accuracy of 96.98\%, while at the same time being the model with the least computational cost.
|
['Oliver Speck', 'Andreas Nürnberger', 'Faraz Ahmed Nizamani', 'Soumick Chatterjee']
|
2021-05-28
| null | null | null | null |
['tumour-classification']
|
['medical']
|
[ 2.69617811e-02 1.42299771e-01 2.57590767e-02 -2.42865067e-02
-5.03151298e-01 -1.69112295e-01 8.66044581e-01 2.03605562e-01
-6.38123333e-01 8.02470803e-01 8.80068913e-02 -3.65893841e-01
-3.37840408e-01 -6.81933999e-01 -2.59637654e-01 -9.27312613e-01
-4.38217968e-01 5.69698870e-01 5.90219855e-01 3.82152461e-02
7.13508502e-02 8.79698455e-01 -1.29928851e+00 3.09079736e-01
6.31869435e-01 1.20776117e+00 5.47030866e-01 6.10455930e-01
-2.38348633e-01 9.60078239e-01 -4.42535132e-01 1.25115737e-01
1.06367871e-01 -1.40121982e-01 -9.89824355e-01 -9.66281146e-02
-1.54074758e-01 -5.75475357e-02 -5.38372815e-01 6.96384966e-01
4.15822715e-01 -3.73383537e-02 8.03106785e-01 -6.56267405e-01
-1.84758559e-01 2.36336723e-01 -4.51945186e-01 6.56904638e-01
1.43483415e-01 5.44192642e-02 4.10754770e-01 -6.84765458e-01
5.62611938e-01 5.83267272e-01 5.89488983e-01 3.79101276e-01
-9.75351334e-01 -5.16620040e-01 -2.17274621e-01 2.67613798e-01
-1.32673550e+00 -1.71818465e-01 3.14975560e-01 -6.98366821e-01
1.26371622e+00 3.07042181e-01 9.61797655e-01 1.02524972e+00
7.01167822e-01 5.09263873e-01 1.39655960e+00 -3.23271573e-01
3.01628858e-01 7.69970343e-02 -1.10681817e-01 4.44211453e-01
-1.10593498e-01 3.18257391e-01 -5.75096384e-02 6.08112141e-02
8.84410322e-01 2.30386287e-01 -3.15940768e-01 -2.08285466e-01
-1.24559927e+00 6.98650897e-01 8.22388828e-01 9.21134472e-01
-4.65358019e-01 1.92067951e-01 5.29915631e-01 1.45880774e-01
6.32917523e-01 2.87425876e-01 -4.34325606e-01 4.19093929e-02
-1.05694687e+00 3.58127654e-02 4.98578638e-01 2.74251759e-01
1.43017516e-01 -4.68720943e-02 -4.26398218e-02 8.57140362e-01
2.26772279e-02 2.79290993e-02 1.15434217e+00 -2.22242460e-01
6.36523888e-02 5.95826626e-01 -3.16885978e-01 -5.41292667e-01
-1.06944108e+00 -8.31924498e-01 -1.23488343e+00 3.30750048e-01
5.60386240e-01 5.73835857e-02 -1.35140109e+00 1.43400538e+00
1.00141294e-01 3.34630162e-01 4.24071699e-02 6.89078331e-01
8.54259074e-01 5.33588290e-01 1.01725608e-01 -2.42037270e-02
1.37472606e+00 -7.37240136e-01 -4.41746354e-01 -8.12426209e-02
1.07520533e+00 -5.79849005e-01 4.76101726e-01 2.10365579e-01
-9.07567978e-01 -2.16389745e-01 -8.42551351e-01 3.21499467e-01
-7.34497130e-01 -1.55471191e-01 6.39864862e-01 5.65663874e-01
-1.46071637e+00 5.88492990e-01 -1.09150541e+00 -5.80406845e-01
6.12442791e-01 6.04924917e-01 -5.77447951e-01 -3.71654555e-02
-1.16044235e+00 1.33380270e+00 5.07357895e-01 -1.32118672e-01
-8.81799579e-01 -8.23873878e-01 -6.00896537e-01 -1.04358539e-01
-1.59534421e-02 -5.45229793e-01 9.72034872e-01 -8.24782908e-01
-1.13880396e+00 9.45605338e-01 1.27523113e-02 -7.79046059e-01
7.96442270e-01 4.55029786e-01 -4.12260681e-01 5.87994941e-02
5.20557016e-02 7.54474998e-01 4.32670772e-01 -7.83233464e-01
-7.61992335e-01 -5.71930647e-01 4.90320250e-02 7.96468332e-02
-1.82840645e-01 -2.18346491e-02 -1.89337537e-01 -6.16913259e-01
1.48249730e-01 -9.72409844e-01 -4.18891788e-01 -2.24582523e-01
-1.60917133e-01 -1.09305903e-01 9.18861866e-01 -8.15144420e-01
6.62171781e-01 -1.84375155e+00 1.47100508e-01 1.79988682e-01
4.73131835e-01 2.28799522e-01 2.29815558e-01 7.37441406e-02
-4.57535774e-01 1.01768151e-01 -3.53992581e-01 1.01334685e-02
-5.32032013e-01 1.95982218e-01 2.29318649e-01 6.82718456e-01
7.64726326e-02 9.76543605e-01 -6.97942555e-01 -2.64721513e-01
4.11293775e-01 6.55098557e-01 -4.97320816e-02 -1.92167342e-01
2.01061338e-01 8.34871411e-01 -2.53160447e-01 3.16017509e-01
4.14242089e-01 -1.96451932e-01 -1.76830232e-01 1.51996315e-01
-9.94628072e-02 -1.16990292e-02 -7.06853509e-01 1.48707700e+00
-6.70985997e-01 8.66495669e-01 -3.86401750e-02 -1.45233929e+00
6.52635217e-01 7.75985301e-01 1.02962947e+00 -1.02666390e+00
1.87826216e-01 4.81330991e-01 3.89107585e-01 -5.22517502e-01
-5.73683605e-02 -2.07494393e-01 2.63839811e-01 2.77205527e-01
-4.61072549e-02 -1.15062118e-01 9.94896218e-02 -1.37483269e-01
1.47273743e+00 -2.16072321e-01 3.71897072e-01 -5.27787447e-01
6.10478044e-01 -5.17796874e-02 1.46593347e-01 4.43411201e-01
-2.14809790e-01 5.20606697e-01 3.26406628e-01 -7.15370059e-01
-9.92182136e-01 -7.37581491e-01 -7.21953988e-01 4.74656224e-01
-1.95991561e-01 2.39704862e-01 -6.51514947e-01 -5.31168520e-01
-2.48842075e-01 4.06872600e-01 -8.89037132e-01 -5.38828261e-02
-6.37567341e-01 -9.96141016e-01 5.45946777e-01 6.48586452e-01
6.41436756e-01 -1.12206364e+00 -9.47343409e-01 2.80997694e-01
-1.37822879e-02 -9.39809799e-01 1.44781008e-01 6.99222386e-01
-1.03642654e+00 -1.12476242e+00 -1.19632256e+00 -8.34855437e-01
6.47094131e-01 4.29709740e-02 7.28602707e-01 1.94390982e-01
-5.88372469e-01 -3.27037424e-02 -3.02447051e-01 -4.27188665e-01
-3.60165775e-01 2.56874710e-01 -2.48476733e-02 -2.12003544e-01
2.09015802e-01 -6.39114559e-01 -5.43563247e-01 2.84702688e-01
-1.06723392e+00 2.19197273e-01 6.00939512e-01 1.03012347e+00
4.26689446e-01 3.02574933e-01 7.00856626e-01 -6.34688318e-01
5.47307730e-01 -6.56507254e-01 -2.76585877e-01 -8.16648826e-02
-3.64630699e-01 -1.22531019e-01 7.20605373e-01 -3.26303273e-01
-6.60145164e-01 4.55909818e-02 -4.04217303e-01 -2.20060498e-01
-5.82159340e-01 6.98514700e-01 4.23499703e-01 -3.77097100e-01
6.94148302e-01 3.95993233e-01 1.30614147e-01 -9.82304588e-02
-2.51984000e-01 5.06550610e-01 2.35720336e-01 -6.80420846e-02
3.51767749e-01 6.56844378e-01 2.83727258e-01 -9.84513462e-01
-5.08537531e-01 -5.37036538e-01 -8.80852938e-01 -2.80522615e-01
9.81217682e-01 -5.21095276e-01 -2.24819511e-01 6.00692987e-01
-9.20231700e-01 -4.90546107e-01 -3.12118441e-01 6.80495799e-01
-4.41259742e-01 9.36922207e-02 -5.48106194e-01 -4.88392651e-01
-6.34665489e-02 -1.47815239e+00 6.38770998e-01 1.44266873e-03
-1.02475569e-01 -1.39847875e+00 -1.01953000e-01 6.11314625e-02
8.04898143e-01 4.63082910e-01 9.41009581e-01 -9.04953599e-01
-2.96903402e-01 -5.70139825e-01 -3.36213171e-01 1.13546997e-02
2.24892765e-01 -3.61025125e-01 -1.14095104e+00 -2.88425505e-01
1.31421939e-01 -3.21467258e-02 9.50897038e-01 9.03850555e-01
1.28524387e+00 1.47534996e-01 -5.99699080e-01 5.80703855e-01
1.33814526e+00 6.77632332e-01 6.43989742e-01 5.35134614e-01
4.30027157e-01 5.73514640e-01 -1.22220581e-02 1.55645758e-01
9.68488231e-02 7.85122991e-01 6.87168896e-01 -3.78088981e-01
-2.99827456e-01 4.96516585e-01 -1.30172268e-01 5.59401393e-01
-3.66259933e-01 -3.91581096e-02 -1.32149696e+00 6.25516176e-01
-1.61803353e+00 -1.00048125e+00 -1.60184607e-01 2.13228440e+00
3.15316975e-01 2.55500972e-01 1.30959377e-01 4.94559616e-01
4.66287613e-01 -1.05100125e-01 -4.36481893e-01 -2.28492543e-01
-4.78843860e-02 3.07883590e-01 6.63196921e-01 1.86319336e-01
-1.15516782e+00 4.58117098e-01 6.16202307e+00 8.10605168e-01
-1.56267202e+00 2.60784686e-01 8.66949260e-01 -7.75333270e-02
2.10161731e-01 -3.89750123e-01 -2.91200459e-01 5.28577089e-01
1.11030674e+00 3.93049382e-02 2.18141764e-01 4.29113388e-01
2.50405371e-01 -3.97801787e-01 -1.03492951e+00 9.50055003e-01
-1.94959611e-01 -1.38559580e+00 -3.02873105e-01 4.39582080e-01
5.81330240e-01 3.73491943e-01 -5.83321303e-02 2.18586475e-01
2.64781769e-02 -1.52641094e+00 5.87037742e-01 5.34403741e-01
7.63144970e-01 -6.80533409e-01 1.20759392e+00 6.75712824e-01
-1.09234071e+00 -1.08782269e-01 -1.28982186e-01 -6.84817657e-02
8.79103914e-02 4.55605388e-01 -1.05769432e+00 5.39714932e-01
7.47981071e-01 6.33958519e-01 -5.52342832e-01 1.40452158e+00
2.92830318e-01 4.38595712e-01 -4.18398559e-01 -7.55093098e-02
5.55818439e-01 1.48005188e-01 4.31222320e-01 1.25393307e+00
4.71500695e-01 -5.95064322e-03 1.86333388e-01 6.07848167e-01
4.11644340e-01 1.17264643e-01 -7.32896745e-01 2.87537992e-01
9.62461997e-03 1.21213496e+00 -1.18370926e+00 -9.83939841e-02
-4.29377466e-01 7.30866671e-01 2.33841896e-01 1.38271138e-01
-8.09046209e-01 -1.38479382e-01 2.65822083e-01 4.21878397e-01
6.19029962e-02 -1.41038805e-01 -3.51086259e-01 -7.61187613e-01
-2.35726997e-01 -2.84402162e-01 3.33749801e-01 -5.47457933e-01
-9.47553277e-01 9.36395705e-01 6.46554157e-02 -9.65434670e-01
-3.08227718e-01 -8.31311703e-01 -5.53106964e-01 9.40555811e-01
-1.54246175e+00 -9.80223000e-01 -3.24111372e-01 6.29098177e-01
5.77942312e-01 -2.04633176e-01 9.73006606e-01 1.94155455e-01
-3.55367631e-01 2.09298521e-01 5.94424754e-02 8.52905214e-02
1.76372528e-01 -1.33581340e+00 -2.28424296e-02 4.82684106e-01
-1.53713241e-01 1.16240241e-01 3.71164501e-01 -3.88912827e-01
-7.88626254e-01 -1.18932343e+00 6.02600992e-01 -1.39965326e-01
6.67889714e-01 -1.19014874e-01 -1.01971114e+00 4.46149886e-01
1.57512486e-01 3.09508920e-01 5.51716387e-01 -3.25246215e-01
2.05953941e-01 9.97417793e-02 -1.34855938e+00 3.93077880e-01
8.84480357e-01 -3.71726424e-01 -3.85249674e-01 5.39569318e-01
2.77765840e-01 -5.72770655e-01 -1.05848050e+00 5.16508579e-01
2.86298364e-01 -1.05871427e+00 9.60087955e-01 -2.25514114e-01
4.04971868e-01 -9.45725664e-02 1.05024479e-01 -1.56003106e+00
-4.71207738e-01 2.35508785e-01 1.66060209e-01 5.27890623e-01
5.83669007e-01 -9.11835611e-01 7.96621621e-01 4.03051257e-01
-3.18077296e-01 -1.23860002e+00 -1.24757099e+00 -8.80054712e-01
3.43380272e-01 -5.20957649e-01 4.20797706e-01 8.77195001e-01
4.80520679e-03 -5.84407263e-02 1.19149037e-01 -5.35112433e-02
2.41271839e-01 -4.19138044e-01 1.59595773e-01 -1.43265593e+00
5.79198562e-02 -9.19698536e-01 -9.90980208e-01 -5.53417325e-01
1.49234235e-01 -1.24016845e+00 -3.22950155e-01 -1.74397516e+00
1.22800872e-01 -7.06443548e-01 -4.13769454e-01 3.69380355e-01
3.11970294e-01 3.60570014e-01 -2.74306625e-01 4.41145003e-01
8.55186880e-02 4.28751577e-03 1.37513578e+00 -2.89080173e-01
-1.02427579e-01 9.81880724e-02 -2.11487874e-01 7.31069922e-01
8.90106738e-01 -1.96830884e-01 -4.04583007e-01 -2.10263073e-01
-1.66438356e-01 2.69607872e-01 5.21743298e-01 -1.36404979e+00
3.53233367e-01 4.45337780e-02 7.48592198e-01 -4.96199667e-01
4.87896860e-01 -9.74903166e-01 4.36735034e-01 7.96747983e-01
-2.49795467e-01 1.87041257e-02 4.00201291e-01 3.14553767e-01
-3.07230473e-01 -2.39653394e-01 1.07202947e+00 -4.94416833e-01
-7.13708758e-01 4.21067089e-01 -7.24824786e-01 -3.50605816e-01
1.49058247e+00 -6.11865461e-01 3.98463719e-02 -1.95815355e-01
-9.84656096e-01 -1.10193394e-01 2.44750097e-01 2.27732927e-01
5.04056633e-01 -1.04908228e+00 -6.41760170e-01 1.15603961e-01
-3.84698808e-02 1.89697921e-01 3.24541628e-01 1.46535528e+00
-7.79229164e-01 8.61381114e-01 -3.22779924e-01 -8.72483194e-01
-1.01511502e+00 4.03723210e-01 9.62114096e-01 -5.70727050e-01
-1.00808358e+00 8.35485399e-01 3.04729342e-01 -3.91961604e-01
2.78269500e-01 -5.12059271e-01 -5.68046451e-01 -7.98061639e-02
5.03325820e-01 1.62494972e-01 6.27619863e-01 -7.79855371e-01
-4.83449847e-01 3.90569955e-01 -2.47296333e-01 -8.05636123e-03
1.47798717e+00 2.65762806e-01 -1.31269336e-01 2.44341508e-01
1.17499399e+00 -4.61799651e-01 -8.63086879e-01 -1.24501154e-01
1.28986582e-01 -1.24438532e-01 5.73100090e-01 -9.32624578e-01
-1.32908785e+00 7.89698958e-01 8.73238385e-01 6.08206928e-01
1.07478607e+00 1.01325460e-01 5.94055772e-01 -1.08590662e-01
4.02582854e-01 -6.79409206e-01 -2.00856075e-01 5.09043932e-01
7.87443995e-01 -1.12205291e+00 -3.61161202e-01 -1.93836704e-01
-3.67593765e-01 1.32860410e+00 4.19785738e-01 -1.55577272e-01
9.33622479e-01 5.02307713e-01 -3.80267054e-02 -4.36591923e-01
-5.67459643e-01 -1.22226343e-01 1.94795355e-01 6.66782200e-01
6.73242092e-01 1.83710501e-01 -2.01187178e-01 2.61411607e-01
-1.78009629e-01 1.65736541e-01 2.87542313e-01 9.26484227e-01
-3.00605923e-01 -8.12038064e-01 -3.09438676e-01 8.18284929e-01
-6.79185212e-01 -3.16508510e-03 -1.24111839e-01 9.84452367e-01
1.54576600e-01 6.23849452e-01 1.71331152e-01 -2.27155194e-01
1.69038102e-01 6.56871796e-02 5.51279426e-01 -4.09835905e-01
-6.77081764e-01 -1.24593958e-01 -2.24234879e-01 -3.14433336e-01
-4.49731767e-01 -6.10843003e-01 -1.36011732e+00 -7.09489286e-02
-3.53781283e-01 3.43479700e-02 8.31964374e-01 1.25806653e+00
3.56808268e-02 9.28852618e-01 4.68186229e-01 -1.05274510e+00
-2.25143448e-01 -1.11496556e+00 -7.05448925e-01 1.42664105e-01
3.27506214e-01 -8.61590147e-01 -2.39901945e-01 -1.94313884e-01]
|
[14.716777801513672, -2.5765342712402344]
|
ea812357-96c7-4b1c-9fa3-2d7624beb179
|
openeds2020-open-eyes-dataset
|
2005.03876
| null |
https://arxiv.org/abs/2005.03876v1
|
https://arxiv.org/pdf/2005.03876v1.pdf
|
OpenEDS2020: Open Eyes Dataset
|
We present the second edition of OpenEDS dataset, OpenEDS2020, a novel dataset of eye-image sequences captured at a frame rate of 100 Hz under controlled illumination, using a virtual-reality head-mounted display mounted with two synchronized eye-facing cameras. The dataset, which is anonymized to remove any personally identifiable information on participants, consists of 80 participants of varied appearance performing several gaze-elicited tasks, and is divided in two subsets: 1) Gaze Prediction Dataset, with up to 66,560 sequences containing 550,400 eye-images and respective gaze vectors, created to foster research in spatio-temporal gaze estimation and prediction approaches; and 2) Eye Segmentation Dataset, consisting of 200 sequences sampled at 5 Hz, with up to 29,500 images, of which 5% contain a semantic segmentation label, devised to encourage the use of temporal information to propagate labels to contiguous frames. Baseline experiments have been evaluated on OpenEDS2020, one for each task, with average angular error of 5.37 degrees when performing gaze prediction on 1 to 5 frames into the future, and a mean intersection over union score of 84.1% for semantic segmentation. As its predecessor, OpenEDS dataset, we anticipate that this new dataset will continue creating opportunities to researchers in eye tracking, machine learning and computer vision communities, to advance the state of the art for virtual reality applications. The dataset is available for download upon request at http://research.fb.com/programs/openeds-2020-challenge/.
|
['Cristina Palmero', 'Sachin S. Talathi', 'Kapil Krishnakumar', 'Karsten Behrendt', 'Abhishek Sharma', 'Oleg V. Komogortsev']
|
2020-05-08
| null | null | null | null |
['eye-tracking']
|
['computer-vision']
|
[ 3.01263571e-01 2.21669912e-01 -1.02186464e-01 -4.33663189e-01
-4.06183034e-01 -5.66653192e-01 2.22665414e-01 -4.63782936e-01
-3.41878384e-01 5.27555525e-01 -2.48372741e-02 -1.46725371e-01
2.55852491e-01 4.47482802e-02 -6.67788446e-01 -4.87336278e-01
6.65137619e-02 -2.30380014e-01 3.98567140e-01 4.05784696e-02
5.29305160e-01 1.33754209e-01 -2.22626853e+00 9.67123061e-02
5.58478236e-01 1.27493370e+00 2.24994108e-01 8.11879396e-01
4.94579971e-01 5.31304240e-01 -4.14526671e-01 -3.82320017e-01
2.66126424e-01 -2.73442924e-01 -8.03541660e-01 1.66723847e-01
9.26981688e-01 -2.91246146e-01 -1.11536540e-01 9.10560310e-01
4.62074637e-01 1.38219148e-01 1.07005700e-01 -1.75537431e+00
-7.51473904e-01 -3.44268739e-01 -9.27142501e-01 8.44467998e-01
8.65401089e-01 6.11524463e-01 7.43689775e-01 -7.48333156e-01
1.00951815e+00 9.60050344e-01 4.59489137e-01 8.07564855e-01
-1.07950723e+00 -1.05011177e+00 2.47284845e-02 3.81520927e-01
-1.47162580e+00 -1.01621008e+00 4.13610309e-01 -7.82997310e-01
7.71396995e-01 3.53564858e-01 7.43904948e-01 1.46511030e+00
-3.30599584e-02 6.92887664e-01 1.26706493e+00 -3.19285601e-01
-2.10334640e-02 2.21860602e-01 4.48412657e-01 4.43950266e-01
3.50898653e-02 3.85390729e-01 -1.13551438e+00 2.06110999e-01
4.90789950e-01 -3.20058286e-01 -8.24192703e-01 -1.49675742e-01
-1.17093539e+00 5.98515570e-01 2.61526376e-01 -1.24597393e-01
-1.49404585e-01 -2.07541138e-01 1.61200225e-01 1.78308263e-01
8.24484646e-01 1.67713359e-01 -5.56853235e-01 -5.00289261e-01
-9.12599027e-01 2.72468388e-01 3.51185977e-01 1.17530310e+00
6.34960949e-01 -3.23253989e-01 -8.60667676e-02 5.72932959e-01
5.67091286e-01 6.41981483e-01 3.59762937e-01 -1.38318002e+00
3.42341006e-01 4.20868427e-01 3.80988002e-01 -7.70351589e-01
-6.65066659e-01 1.18008375e-01 -2.56785423e-01 2.74607509e-01
6.08149469e-01 -3.97049516e-01 -6.95046008e-01 1.73884463e+00
4.84925896e-01 6.20446980e-01 -4.53497380e-01 1.22375453e+00
1.02339661e+00 2.85311967e-01 -2.29842737e-01 -4.07167673e-01
1.59939897e+00 -8.06024373e-01 -7.89438307e-01 4.04201448e-02
4.61233586e-01 -9.33808863e-01 1.23785436e+00 5.74453533e-01
-1.10574353e+00 -6.77006900e-01 -7.33815610e-01 -1.42608523e-01
-1.34767503e-01 -2.87091304e-02 3.75794172e-01 1.10838783e+00
-1.59384823e+00 8.18746984e-02 -6.71578705e-01 -6.13012016e-01
7.10692465e-01 4.66290534e-01 -2.19497725e-01 3.03707391e-01
-8.90660346e-01 5.85259736e-01 -6.17514178e-02 -6.66934326e-02
-3.39447469e-01 -9.29378867e-01 -9.03038621e-01 -3.83076727e-01
3.92349184e-01 -5.76634467e-01 1.33938622e+00 -1.18569541e+00
-1.43520796e+00 1.39270902e+00 -9.09284055e-01 -3.87084574e-01
3.87077779e-01 -3.37841630e-01 -7.50423849e-01 2.05279693e-01
1.05692424e-01 1.07206678e+00 1.11712670e+00 -1.01798189e+00
-9.02506828e-01 -5.17286897e-01 -8.21614824e-03 2.05150098e-01
-1.47009298e-01 7.01870203e-01 -8.35666597e-01 -3.58175933e-01
-3.51400763e-01 -1.35221577e+00 4.95970875e-01 -8.25316906e-02
-5.55173099e-01 -3.50073010e-01 8.73685181e-01 -5.78833818e-01
1.14278853e+00 -2.15240884e+00 -7.12727383e-02 -6.23541139e-02
5.56051731e-01 2.19083577e-01 -8.72032791e-02 -3.03222120e-01
-4.00104195e-01 -2.14540362e-02 1.17099985e-01 -7.98271537e-01
-3.67488384e-01 -4.24627602e-01 -2.83978015e-01 6.52671218e-01
8.01695511e-02 8.15542996e-01 -7.97707200e-01 -3.72820795e-01
3.14574808e-01 4.33639020e-01 -4.78485912e-01 9.28754453e-03
1.33451289e-02 7.98047125e-01 -1.49068058e-01 6.20450199e-01
8.09387326e-01 -5.57575226e-01 -4.60206956e-01 7.71724507e-02
-2.20711514e-01 3.24152224e-02 -9.74127114e-01 1.53031850e+00
4.00237776e-02 1.56404805e+00 -1.95998460e-01 -1.15467824e-01
5.98375022e-01 1.87270269e-01 5.22066057e-01 -8.00672352e-01
3.67026538e-01 -2.73934364e-01 -2.46417686e-01 -7.52477407e-01
8.07298601e-01 6.07512116e-01 2.30320558e-01 4.53344733e-01
9.48155820e-02 4.18179989e-01 4.46296066e-01 1.98260456e-01
8.13427925e-01 2.72864342e-01 -2.23599881e-01 -9.22903940e-02
6.19848788e-01 -9.74406078e-02 3.77429247e-01 2.97740817e-01
-8.29486609e-01 9.97016966e-01 5.39555728e-01 -2.80992568e-01
-7.80976057e-01 -9.95583057e-01 -4.84898686e-01 1.16756761e+00
4.19336408e-01 -4.33552951e-01 -1.11413085e+00 -3.84636074e-01
-3.58948827e-01 5.37575305e-01 -7.87462950e-01 1.99057832e-01
-2.26102471e-01 -4.55825210e-01 3.33329231e-01 1.69662401e-01
3.82815152e-01 -1.18033183e+00 -8.41244102e-01 -5.83408415e-01
-5.09976268e-01 -1.29857934e+00 -8.74476969e-01 -7.28939593e-01
-3.62252563e-01 -1.68068242e+00 -8.15020978e-01 -4.30577844e-01
4.73753810e-01 6.07055783e-01 1.11188924e+00 4.26684693e-02
-2.30233356e-01 6.69471741e-01 -4.20046329e-01 -4.86674488e-01
1.01277277e-01 4.30059694e-02 2.44662762e-01 3.27441156e-01
8.88731122e-01 -4.34812084e-02 -9.29279685e-01 6.20369434e-01
-2.68473506e-01 1.13293424e-01 -2.74187773e-01 5.10675550e-01
3.77177179e-01 -5.72144568e-01 1.18803412e-01 -6.40976131e-01
3.01684529e-01 -6.38714492e-01 -9.71233308e-01 1.32261524e-02
-4.23318505e-01 -6.21981502e-01 -9.90270972e-02 -3.22787344e-01
-1.06904757e+00 -1.78421825e-01 2.84834743e-01 -8.12024295e-01
-4.79550600e-01 -2.74874449e-01 3.22814703e-01 -1.88699514e-01
8.06676269e-01 1.32832721e-01 2.57174104e-01 -2.03237325e-01
2.10815474e-01 1.05533111e+00 7.54511535e-01 7.49124587e-03
3.50602001e-01 3.80531788e-01 -2.61190057e-01 -1.00856650e+00
-8.52519929e-01 -8.72756541e-01 -4.39981490e-01 -5.59792459e-01
1.04032135e+00 -1.17450643e+00 -1.34863973e+00 9.73496795e-01
-9.74779844e-01 -4.53605235e-01 3.78354825e-02 5.06421268e-01
-7.10630059e-01 8.52417797e-02 -3.16836715e-01 -7.72104621e-01
-1.64919779e-01 -1.29634929e+00 1.22154450e+00 6.50009036e-01
-4.33353454e-01 -6.18715525e-01 3.74505110e-02 9.36163127e-01
6.50053918e-02 1.12679899e-01 -2.15010360e-01 -3.11435282e-01
-5.84644198e-01 2.42955729e-01 -3.73373419e-01 2.40639493e-01
-1.03952356e-01 4.36895579e-01 -1.30042803e+00 -3.64404738e-01
-1.01419203e-01 -2.44842485e-01 4.96322483e-01 6.99581027e-01
1.12026370e+00 2.21873507e-01 -4.23072398e-01 6.67399108e-01
9.12854731e-01 2.01065019e-01 7.33199596e-01 3.91279191e-01
7.34166503e-01 8.10000598e-01 7.91409373e-01 4.44270700e-01
7.38761485e-01 9.05550361e-01 4.71932828e-01 1.39304206e-01
-2.45442808e-01 2.12283388e-01 2.41941929e-01 1.57019809e-01
-4.05582845e-01 -3.45639765e-01 -1.14804220e+00 4.05782342e-01
-1.44594634e+00 -1.07646763e+00 -5.93238533e-01 2.36088729e+00
6.81940854e-01 -2.09541544e-01 6.16628945e-01 -3.22337747e-02
1.07626331e+00 1.70987219e-01 -6.82059407e-01 -5.27173951e-02
-1.97764323e-03 7.60149211e-02 4.07792509e-01 2.71573544e-01
-1.17070377e+00 8.67184520e-01 6.10498667e+00 4.85216767e-01
-1.29679406e+00 1.79412141e-01 7.34462678e-01 -8.16350222e-01
1.84812829e-01 -2.72744000e-01 -1.12004805e+00 1.28859258e+00
1.41277277e+00 5.22906967e-02 5.54749906e-01 4.63335156e-01
5.52404106e-01 -5.67570388e-01 -8.09876800e-01 1.36306536e+00
3.28653812e-01 -1.34554565e+00 -8.39120209e-01 2.51303345e-01
6.78125024e-01 5.77154279e-01 5.35049796e-01 -2.09348217e-01
-1.15260042e-01 -9.18342352e-01 9.71042693e-01 7.67292082e-01
1.22998893e+00 -5.00249326e-01 4.15020347e-01 1.04439497e-01
-9.29376125e-01 5.78335002e-02 1.56436726e-01 6.77538663e-02
2.58966208e-01 -7.51986206e-02 -4.94272172e-01 9.72043648e-02
1.44286394e+00 1.07114267e+00 -9.04846191e-01 1.04982638e+00
4.62173708e-02 5.57028711e-01 -4.00604606e-01 1.87749326e-01
-1.96240664e-01 -2.79587030e-01 6.55921757e-01 6.93456173e-01
1.75387830e-01 1.88957036e-01 -4.91354883e-01 7.20716715e-01
-8.97815302e-02 -3.70181143e-01 -5.05936682e-01 4.66804683e-01
8.39972317e-01 1.08079123e+00 -6.44791961e-01 -6.19723052e-02
-6.63999081e-01 7.94762433e-01 -1.07205682e-01 6.53287351e-01
-1.15237153e+00 -2.38491312e-01 1.04533279e+00 4.32952672e-01
2.47809127e-01 4.27691527e-02 -2.99556345e-01 -1.00771666e+00
1.21627055e-01 -7.89173901e-01 1.79446489e-01 -1.56974077e+00
-8.36750329e-01 7.79700339e-01 -3.52711044e-02 -1.33814204e+00
-3.01994532e-01 -4.91521895e-01 -2.65744805e-01 1.13949656e+00
-1.51916146e+00 -9.36077833e-01 -7.36062169e-01 1.11106336e+00
5.27434528e-01 -3.43809694e-01 3.84777337e-01 2.98945218e-01
-9.67810750e-01 8.45661223e-01 -1.35929301e-01 -8.70025679e-02
1.08139110e+00 -9.49677527e-01 5.41785359e-01 8.37791264e-01
7.38723651e-02 5.24177432e-01 8.51330578e-01 -2.91589916e-01
-9.36454713e-01 -9.30552483e-01 9.46548700e-01 -1.06655216e+00
5.82622051e-01 -3.39344263e-01 -7.68191338e-01 9.68047976e-01
4.03840721e-01 4.90240604e-01 6.56480312e-01 1.21322259e-01
-8.49284977e-02 2.35955402e-01 -1.05867767e+00 5.58213234e-01
1.29147458e+00 -5.79714656e-01 -4.79908675e-01 2.25232869e-01
6.43434107e-01 -9.60453391e-01 -7.86287844e-01 4.36372422e-02
5.38090467e-01 -1.40309179e+00 8.52091849e-01 -2.34851822e-01
1.51691824e-01 -3.04856420e-01 2.19346672e-01 -8.70306194e-01
1.13868505e-01 -1.02914584e+00 -3.32795411e-01 1.10946572e+00
1.58967108e-01 -8.77598941e-01 8.92353952e-01 7.59163737e-01
-1.22710494e-02 -5.57047546e-01 -8.36708665e-01 -4.97828722e-01
-5.37077963e-01 -6.11218810e-01 4.72403228e-01 8.14837694e-01
-3.63934010e-01 2.92373747e-01 -1.94766417e-01 9.18424875e-02
6.02193952e-01 -1.59903497e-01 1.10516155e+00 -1.34694302e+00
3.17631394e-01 -4.22913522e-01 -5.86846232e-01 -1.20638919e+00
2.82029539e-01 -1.88608274e-01 -3.06228429e-01 -6.21388733e-01
-8.74403939e-02 -1.43950954e-01 -1.21904211e-02 2.68073529e-01
-1.78390652e-01 8.47149789e-01 2.01880440e-01 4.88021404e-01
-8.62295985e-01 2.95135707e-01 1.15309668e+00 3.80005389e-01
-3.18259269e-01 2.32773051e-01 -6.35367513e-01 8.07401121e-01
5.80731809e-01 -2.79622078e-01 -5.60429454e-01 -1.16062470e-01
1.20370269e-01 -3.41767445e-02 6.52502835e-01 -8.80115271e-01
4.01538283e-01 -7.75579959e-02 2.91302145e-01 -7.85752535e-01
5.08330286e-01 -5.29925704e-01 1.81304663e-01 -5.12965173e-02
-4.05417308e-02 5.30022196e-02 2.99602956e-01 5.04897296e-01
-1.08800620e-01 1.26968116e-01 7.51842260e-01 4.52104688e-01
-1.12045169e+00 4.41684067e-01 -8.57609808e-02 4.63946193e-01
1.47268820e+00 -7.65913904e-01 -6.10178947e-01 -1.87088087e-01
-8.29044163e-01 3.20609540e-01 8.59941006e-01 8.84689391e-01
4.06718135e-01 -9.67998028e-01 -4.94387835e-01 5.12508869e-01
3.73980075e-01 -9.06452686e-02 5.87854326e-01 1.17448592e+00
-3.77700478e-01 5.65121412e-01 -3.82639438e-01 -1.14325941e+00
-1.72679043e+00 4.87042427e-01 1.60405651e-01 5.91428280e-01
-6.38585687e-01 1.15371299e+00 1.84923247e-01 2.05918252e-01
1.44890398e-01 -2.07830653e-01 -5.48914373e-01 2.69596934e-01
8.64936292e-01 5.88088930e-01 4.10274399e-04 -1.35694206e+00
-4.26600486e-01 7.10352838e-01 2.32169572e-02 3.20120119e-02
9.86037374e-01 -9.70968604e-01 1.83285117e-01 4.16376501e-01
1.21262813e+00 -1.94780633e-01 -1.74493110e+00 -2.44412124e-01
-2.00592712e-01 -7.64607131e-01 1.56909913e-01 -5.44535697e-01
-1.32837331e+00 6.18599594e-01 9.63437974e-01 2.43059590e-01
1.32061684e+00 8.00410733e-02 7.56011248e-01 -2.75576830e-01
2.82646149e-01 -7.50134528e-01 -5.51414490e-02 4.35602665e-01
6.84604704e-01 -1.65108478e+00 -3.16915542e-01 -4.11952972e-01
-8.41611981e-01 7.53238380e-01 7.07976818e-01 7.36645833e-02
5.79063833e-01 -8.30761790e-02 2.42082670e-01 -3.86334091e-01
-7.22933590e-01 -3.56241852e-01 5.42722821e-01 8.56582880e-01
2.99565583e-01 -2.09743410e-01 3.12676430e-01 3.68454099e-01
-6.16844952e-01 3.55538636e-01 6.05204582e-01 4.57319140e-01
-2.70955358e-02 -5.82614362e-01 -6.13287985e-01 4.17792201e-01
-5.07813632e-01 -1.57196864e-01 2.80572306e-02 7.38443494e-01
1.92609787e-01 1.26043880e+00 6.72283471e-01 -2.77159274e-01
1.82362869e-01 -7.82864243e-02 4.47101772e-01 -3.67232919e-01
-3.06172639e-01 -3.30483228e-01 -2.41603274e-02 -1.08340049e+00
-6.53837323e-01 -1.12280488e+00 -8.09600949e-01 -7.30670989e-01
-2.34128758e-01 -4.04573560e-01 4.68144476e-01 6.96254373e-01
7.60954440e-01 3.18360716e-01 5.28124273e-01 -1.27037108e+00
8.71129707e-02 -9.10099149e-01 -5.37442923e-01 3.93883973e-01
6.91754401e-01 -1.00967455e+00 -4.10123140e-01 6.49083078e-01]
|
[14.106330871582031, 0.0832081064581871]
|
6a7a2062-acb2-4b2f-b0bd-5ae585c21c25
|
learning-video-instance-segmentation-with
|
2012.03911
| null |
https://arxiv.org/abs/2012.03911v1
|
https://arxiv.org/pdf/2012.03911v1.pdf
|
Learning Video Instance Segmentation with Recurrent Graph Neural Networks
|
Most existing approaches to video instance segmentation comprise multiple modules that are heuristically combined to produce the final output. Formulating a purely learning-based method instead, which models both the temporal aspect as well as a generic track management required to solve the video instance segmentation task, is a highly challenging problem. In this work, we propose a novel learning formulation, where the entire video instance segmentation problem is modelled jointly. We fit a flexible model to our formulation that, with the help of a graph neural network, processes all available new information in each frame. Past information is considered and processed via a recurrent connection. We demonstrate the effectiveness of the proposed approach in comprehensive experiments. Our approach, operating at over 25 FPS, outperforms previous video real-time methods. We further conduct detailed ablative experiments that validate the different aspects of our approach.
|
['Michael Felsberg', 'Martin Danelljan', 'Emil Brissman', 'Joakim Johnander']
|
2020-12-07
| null | null | null | null |
['video-instance-segmentation']
|
['computer-vision']
|
[ 3.14067602e-01 8.78814384e-02 -3.41680825e-01 -2.13664234e-01
-7.06784189e-01 -5.50086498e-01 4.26227361e-01 4.07146886e-02
-3.56092930e-01 7.34423637e-01 -2.00534746e-01 -3.17720592e-01
-1.75637290e-01 -5.60061157e-01 -9.62534487e-01 -4.32803214e-01
1.64806750e-02 2.95199364e-01 5.71161807e-01 2.05460548e-01
2.04473689e-01 3.84983450e-01 -1.29469264e+00 2.94896394e-01
8.35613430e-01 1.02961135e+00 4.45999771e-01 9.57082212e-01
-2.30650023e-01 1.32366574e+00 -5.75377524e-01 -3.23935062e-01
3.29230249e-01 -4.45762396e-01 -1.00482309e+00 9.32613194e-01
3.32910359e-01 -2.18923077e-01 -3.87408584e-01 6.67359293e-01
5.93335964e-02 2.93046355e-01 2.57834017e-01 -1.14580059e+00
1.87883526e-02 7.08720505e-01 -5.68856537e-01 4.73589242e-01
3.27802181e-01 2.12231115e-01 7.50800431e-01 -4.75302219e-01
7.68265486e-01 8.65064859e-01 4.42050457e-01 4.40574974e-01
-9.71308529e-01 -2.02177510e-01 8.78756940e-01 4.48360890e-01
-1.21078396e+00 -4.18749958e-01 7.94298112e-01 -3.64435256e-01
8.76545429e-01 1.77008100e-02 9.74797428e-01 6.47020638e-01
8.60365406e-02 1.25131226e+00 7.59432673e-01 -3.26154709e-01
1.80424497e-01 9.08898711e-02 1.82424471e-01 7.32188821e-01
4.19871658e-02 -3.26779544e-01 -5.03675342e-01 3.55644792e-01
8.96988213e-01 1.62689341e-03 -2.98637897e-01 -3.04544449e-01
-1.00540757e+00 3.63870412e-01 1.22105658e-01 9.69326422e-02
-4.95090246e-01 4.63804781e-01 3.20178986e-01 2.18998805e-01
5.00668466e-01 4.78236601e-02 -3.06569278e-01 -2.85891265e-01
-1.36900020e+00 1.65556058e-01 9.83437717e-01 1.21907640e+00
6.70883834e-01 1.32644460e-01 -2.33731896e-01 4.87552315e-01
2.61439055e-01 -1.11614168e-01 1.38364345e-01 -1.25288725e+00
4.66210783e-01 4.87571716e-01 1.44668624e-01 -7.37495363e-01
-1.96122751e-01 -2.88941741e-01 -4.54505771e-01 -1.21948652e-01
3.58374089e-01 -5.30114293e-01 -8.86815071e-01 1.57025874e+00
4.56705153e-01 8.49005580e-01 8.33116174e-02 7.44751930e-01
5.21606386e-01 9.26385760e-01 1.32277653e-01 -6.66117251e-01
8.80340397e-01 -1.45559680e+00 -8.80856812e-01 -4.54953015e-02
2.56867766e-01 -5.57724237e-01 5.61129808e-01 4.21274692e-01
-1.60969627e+00 -7.04812586e-01 -1.00538480e+00 1.84256777e-01
-1.84151363e-02 1.68826550e-01 5.39870560e-01 2.44343147e-01
-1.39611721e+00 7.28261232e-01 -1.08229578e+00 -4.37421501e-01
2.25971684e-01 4.96096224e-01 1.34780675e-01 1.15414143e-01
-6.72477543e-01 3.72289002e-01 5.71820021e-01 4.91487235e-01
-8.81107926e-01 -3.30869198e-01 -6.35965466e-01 6.46385178e-02
1.11082995e+00 -9.53063488e-01 1.54583871e+00 -1.20130002e+00
-1.72525275e+00 4.26427275e-01 -3.38830858e-01 -7.20637381e-01
6.42350137e-01 -3.97067130e-01 -8.27197358e-02 5.05747974e-01
-3.00674349e-01 6.21159375e-01 1.06124711e+00 -1.27008343e+00
-1.02541149e+00 -1.12809306e-02 6.05126917e-01 4.77623314e-01
-5.38134500e-02 -6.22514561e-02 -1.32136238e+00 -6.21133387e-01
-5.23839816e-02 -9.47181404e-01 -4.56775039e-01 -3.17342430e-01
-2.67519683e-01 -8.49005431e-02 7.47794151e-01 -5.83549738e-01
1.62600100e+00 -1.86677217e+00 6.85618758e-01 3.74172851e-02
1.28264964e-01 3.15701336e-01 1.75863028e-01 4.59639668e-01
1.80196270e-01 7.65466765e-02 -6.76285923e-02 -5.23344278e-01
-2.15656698e-01 2.06309617e-01 -1.19450606e-01 1.61177248e-01
3.10277164e-01 8.41381133e-01 -7.99893379e-01 -6.73186660e-01
2.42279768e-01 3.26102197e-01 -5.72476804e-01 4.33541954e-01
-6.43446863e-01 3.75040740e-01 -4.55647916e-01 6.72581911e-01
2.53954411e-01 -4.51362878e-01 4.47867662e-01 -2.47480795e-01
-1.61238149e-01 -1.52863309e-01 -1.33057833e+00 1.83950508e+00
-1.62160262e-01 4.73600745e-01 1.91948459e-01 -1.12809420e+00
4.29077327e-01 4.32686299e-01 7.59282291e-01 -3.21063787e-01
7.95332715e-03 -1.98467717e-01 -3.88546795e-01 -7.41686642e-01
7.15090156e-01 1.01544544e-01 2.18945056e-01 3.59092236e-01
1.35330424e-01 1.14995807e-01 6.60027444e-01 4.18993086e-01
1.06701314e+00 6.56269252e-01 1.69904679e-01 1.47685483e-01
6.57023787e-01 3.35686743e-01 7.38492429e-01 8.39446187e-01
-3.20283741e-01 4.18356806e-01 7.39983082e-01 -4.15948480e-01
-7.10120499e-01 -7.43511617e-01 4.13834542e-01 6.77779794e-01
5.11687338e-01 -5.33868849e-01 -1.10713708e+00 -7.53811300e-01
-4.33140814e-01 3.34148824e-01 -2.26214409e-01 1.77899480e-01
-7.07591832e-01 -5.73955715e-01 8.71937945e-02 6.41138494e-01
5.59830785e-01 -1.05413115e+00 -6.89778328e-01 5.44919014e-01
-2.08547622e-01 -1.46852505e+00 -4.77986962e-01 -2.41589341e-02
-1.08311212e+00 -1.06361485e+00 -4.69098628e-01 -7.15875745e-01
4.37184513e-01 4.72929269e-01 1.08846235e+00 2.66185045e-01
-1.76459998e-01 7.73050547e-01 -3.78858775e-01 -4.09400724e-02
-2.08793372e-01 1.89206541e-01 -3.97467434e-01 3.10975343e-01
-6.53588623e-02 -3.09588999e-01 -4.28192735e-01 -6.65675774e-02
-9.72728074e-01 4.65359718e-01 5.23291051e-01 6.44806564e-01
9.50824738e-01 2.06910387e-01 4.94668007e-01 -1.18685782e+00
4.80952293e-01 -4.22139227e-01 -7.02215254e-01 4.99547362e-01
-2.03815639e-01 -1.42604142e-01 4.54238832e-01 -3.87405545e-01
-1.16148412e+00 4.98235136e-01 1.29287943e-01 -7.61716962e-01
-1.31724015e-01 6.92880392e-01 -9.40200090e-02 2.15312298e-02
-6.85166791e-02 2.04251587e-01 -9.04337466e-02 -2.27761433e-01
4.19226587e-01 3.35830539e-01 6.37935638e-01 -5.76333880e-01
6.00213587e-01 3.50498825e-01 -1.59386203e-01 -7.66416073e-01
-7.56948233e-01 -5.98880708e-01 -8.15482497e-01 -7.69438803e-01
9.75389123e-01 -1.07713139e+00 -7.92503417e-01 5.52018881e-01
-1.17846668e+00 -5.28885901e-01 -2.48995528e-01 3.55702281e-01
-9.04646277e-01 3.98862183e-01 -9.15446818e-01 -1.00996029e+00
-1.25995472e-01 -1.26968038e+00 9.37809944e-01 4.70636934e-01
5.98739237e-02 -1.06340206e+00 -3.32366489e-02 4.83530939e-01
5.42472117e-02 2.27862656e-01 4.92763370e-01 -3.52507263e-01
-1.32659876e+00 -1.01205252e-01 -1.39586851e-01 1.57563791e-01
-1.37417847e-02 3.60014051e-01 -6.79989100e-01 -2.04713210e-01
1.03420325e-01 -5.07939681e-02 9.85204816e-01 5.23543596e-01
1.29530942e+00 -2.52456397e-01 -3.42503399e-01 4.93190199e-01
1.66856265e+00 4.19003338e-01 5.62277675e-01 2.92778790e-01
8.31916332e-01 3.30296308e-01 8.45930099e-01 5.95159411e-01
5.60762763e-01 7.16119349e-01 3.93036157e-01 -5.08997068e-02
-2.10521240e-02 5.39240912e-02 4.22860235e-01 1.00414085e+00
-4.37192678e-01 -7.60574996e-01 -6.67789400e-01 4.31449354e-01
-2.46203327e+00 -1.20588982e+00 1.91846430e-01 1.92000115e+00
4.37739432e-01 3.79059702e-01 3.35377872e-01 6.89290464e-02
7.24636495e-01 2.18252599e-01 -5.08818626e-01 -3.03939283e-01
2.70888597e-01 7.34639242e-02 3.40403676e-01 5.42364955e-01
-1.37630653e+00 1.17014205e+00 6.84069586e+00 4.93425280e-01
-1.07672644e+00 -2.12723777e-01 6.13036811e-01 -3.01157504e-01
7.25378469e-02 1.30329847e-01 -7.17912853e-01 3.46573025e-01
1.03395653e+00 -1.28085077e-01 5.69920361e-01 5.62190056e-01
5.63596070e-01 -3.15674186e-01 -1.19712031e+00 1.04047823e+00
1.66630939e-01 -1.55725825e+00 2.20376819e-01 -3.21268439e-01
6.23154342e-01 -3.03268433e-01 -1.40156046e-01 2.12455973e-01
1.28636628e-01 -6.17689967e-01 9.49944019e-01 8.88222158e-01
1.66668817e-01 -7.78550923e-01 3.70213121e-01 3.74876291e-01
-1.54151154e+00 -1.33495957e-01 7.46256635e-02 -4.18246277e-02
5.37354529e-01 2.46249393e-01 -7.13222206e-01 9.47375655e-01
5.11283219e-01 1.17988682e+00 -5.06867290e-01 1.28156722e+00
-1.10470660e-01 6.17903233e-01 -1.36352628e-01 3.06591004e-01
3.80866587e-01 -3.17839622e-01 4.70803171e-01 1.23548901e+00
1.27272442e-01 1.98184073e-01 6.77170455e-01 6.46420717e-01
-4.13538888e-03 1.61630269e-02 -3.39208692e-01 -2.37341091e-01
1.95491970e-01 1.31497216e+00 -1.14688993e+00 -7.26366222e-01
-6.16371632e-01 1.01486337e+00 4.57119018e-01 6.73667312e-01
-1.12175083e+00 -7.44304135e-02 2.06667572e-01 -4.92720790e-02
5.64233720e-01 -4.75063115e-01 4.43260148e-02 -1.36405575e+00
2.10237101e-01 -7.05535710e-01 4.65680689e-01 -7.17340589e-01
-7.87313938e-01 5.61071634e-01 1.35266408e-02 -1.25579894e+00
-5.35887241e-01 -2.76550829e-01 -5.87419271e-01 4.21791226e-01
-1.73874354e+00 -9.51933742e-01 -3.10914695e-01 6.61272705e-01
1.07170630e+00 6.90877140e-02 2.00038612e-01 3.88013095e-01
-1.18406558e+00 1.56788558e-01 -3.34352583e-01 -6.37881011e-02
2.66263574e-01 -1.15897751e+00 5.86244613e-02 1.14023352e+00
2.48548478e-01 2.55076498e-01 5.63089132e-01 -8.00839186e-01
-1.64195573e+00 -1.26815176e+00 5.63512206e-01 -7.47954175e-02
6.46916986e-01 2.37334054e-02 -8.93789172e-01 9.88245070e-01
3.82570982e-01 -1.13322638e-01 4.04384226e-01 -3.29311520e-01
2.32396960e-01 -9.24828947e-02 -6.51865125e-01 4.92517442e-01
1.03044379e+00 -3.02914798e-01 -4.23489064e-01 3.59384090e-01
8.51424634e-01 -6.72774971e-01 -7.40154445e-01 2.75441825e-01
2.26799786e-01 -8.97152126e-01 6.85432911e-01 -6.06301010e-01
3.15182954e-01 -4.77526724e-01 9.27493274e-02 -8.51909518e-01
-5.67495823e-02 -8.56050074e-01 -6.71966434e-01 1.13060808e+00
2.66187876e-01 8.98370296e-02 8.51429284e-01 6.06347203e-01
-2.80948013e-01 -8.92886221e-01 -5.43706000e-01 -6.31381571e-01
-6.52187824e-01 -5.64874828e-01 2.02357292e-01 4.83975977e-01
-2.41429098e-02 2.76547253e-01 -5.41709244e-01 3.62449259e-01
5.86703181e-01 3.23109657e-01 7.07640171e-01 -7.52866685e-01
-5.28082728e-01 -2.46611908e-01 -4.22257841e-01 -1.30783832e+00
2.30578303e-01 -5.04839122e-01 1.27880007e-01 -1.68297660e+00
2.07811669e-01 -1.07368536e-01 -4.08552766e-01 2.78792590e-01
-2.34035358e-01 -7.66614377e-02 5.43401182e-01 1.86161458e-01
-1.27434385e+00 1.58636242e-01 1.11118519e+00 -1.62206262e-01
-4.64921415e-01 4.23919261e-02 -3.71520370e-01 7.77720094e-01
6.52551591e-01 -1.31049290e-01 -6.89755559e-01 -6.92452669e-01
-1.38762578e-01 6.67401731e-01 1.68795809e-01 -1.32198060e+00
7.18674839e-01 -3.49270284e-01 1.58379331e-01 -7.23225176e-01
4.42849249e-01 -8.34332108e-01 3.78042579e-01 2.74184853e-01
-2.87307471e-01 2.28456870e-01 2.13392675e-01 8.31611335e-01
-4.76093054e-01 -1.16410330e-01 4.71665055e-01 -3.65994930e-01
-1.14390385e+00 6.07447207e-01 -5.23349345e-01 -1.55482903e-01
1.39970422e+00 -3.87015790e-01 1.53040424e-01 -3.84472430e-01
-1.12304556e+00 5.11769474e-01 3.73580664e-01 2.73173034e-01
4.84971017e-01 -9.40055251e-01 -2.80945748e-01 -1.69258043e-01
-3.56350720e-01 8.00112858e-02 2.74847925e-01 8.67766440e-01
-5.57390928e-01 3.21121931e-01 -1.07998988e-02 -7.33732998e-01
-1.12856758e+00 8.01657856e-01 3.46401244e-01 -3.18630666e-01
-6.31474197e-01 5.09317517e-01 -1.00911468e-01 4.17973220e-01
4.97428209e-01 -3.36225122e-01 -5.31584799e-01 2.10469052e-01
3.91990066e-01 3.65527302e-01 -1.43846408e-01 -5.01982868e-01
1.37466295e-02 5.47718048e-01 -1.90662459e-01 -4.20598686e-02
1.27157390e+00 -5.79901755e-01 6.70710020e-03 6.58269107e-01
5.87629318e-01 -4.25861835e-01 -1.62732756e+00 -1.86549127e-01
1.61375672e-01 -2.14051202e-01 -9.59177837e-02 -4.86208797e-01
-1.39117491e+00 5.26626825e-01 2.30924875e-01 2.68541336e-01
1.32813811e+00 -1.97800234e-01 8.29313934e-01 2.80144095e-01
3.73963177e-01 -1.31122351e+00 9.09230933e-02 3.88447553e-01
3.49995077e-01 -1.04520166e+00 2.01222524e-02 -6.94903851e-01
-5.13014734e-01 1.19159913e+00 7.83824980e-01 -1.17943026e-01
4.34538245e-01 3.65894169e-01 1.12856077e-02 -5.98716959e-02
-1.19392967e+00 -2.47268036e-01 8.70850384e-02 1.50071695e-01
3.70857418e-01 -2.89705575e-01 -3.02965581e-01 4.91558611e-01
4.63041991e-01 5.00606358e-01 6.79469645e-01 1.32708323e+00
-3.95090193e-01 -1.07079005e+00 -4.70171012e-02 2.90925086e-01
-4.32527333e-01 3.45368296e-01 -2.41387516e-01 7.46115625e-01
1.24630583e-02 9.18085158e-01 5.12315556e-02 -3.09387892e-01
2.70660609e-01 9.86790136e-02 5.49331784e-01 -5.92749715e-01
-5.57563901e-01 4.63674486e-01 4.48222533e-02 -8.35824311e-01
-1.08937407e+00 -6.76882505e-01 -1.41612875e+00 -1.22908667e-01
-1.82145715e-01 -4.70557995e-02 4.13371801e-01 1.11757350e+00
2.72632271e-01 9.02558386e-01 4.52968836e-01 -1.01564860e+00
-8.85975361e-02 -4.07466650e-01 -3.82297724e-01 2.44092330e-01
2.09859520e-01 -4.91610587e-01 -2.00356804e-02 5.37285984e-01]
|
[8.96119499206543, -0.02590038999915123]
|
622fabcc-b3ad-4edf-81b3-5a2a86e2c9ba
|
time-series-forecasting-using-manifold
|
2110.03625
| null |
https://arxiv.org/abs/2110.03625v4
|
https://arxiv.org/pdf/2110.03625v4.pdf
|
Time Series Forecasting Using Manifold Learning
|
We address a three-tier numerical framework based on manifold learning for the forecasting of high-dimensional time series. At the first step, we embed the time series into a reduced low-dimensional space using a nonlinear manifold learning algorithm such as Locally Linear Embedding and Diffusion Maps. At the second step, we construct reduced-order regression models on the manifold, in particular Multivariate Autoregressive (MVAR) and Gaussian Process Regression (GPR) models, to forecast the embedded dynamics. At the final step, we lift the embedded time series back to the original high-dimensional space using Radial Basis Functions interpolation and Geometric Harmonics. For our illustrations, we test the forecasting performance of the proposed numerical scheme with four sets of time series: three synthetic stochastic ones resembling EEG signals produced from linear and nonlinear stochastic models with different model orders, and one real-world data set containing daily time series of 10 key foreign exchange rates (FOREX) spanning the time period 03/09/2001-29/10/2020. The forecasting performance of the proposed numerical scheme is assessed using the combinations of manifold learning, modelling and lifting approaches. We also provide a comparison with the Principal Component Analysis algorithm as well as with the naive random walk model and the MVAR and GPR models trained and implemented directly in the high-dimensional space.
|
['Ioannis Kevrekidis', 'Constantinos Siettos', 'Ronen Talmon', 'Panagiotis Papaioannou']
|
2021-10-07
| null | null | null | null |
['gpr', 'gpr']
|
['computer-vision', 'miscellaneous']
|
[-2.76947647e-01 -8.45214948e-02 4.40682203e-01 1.41732305e-01
-6.43068016e-01 -3.30285668e-01 9.49698091e-01 -1.27855614e-01
-2.58419424e-01 6.41079307e-01 2.23900944e-01 -3.90313178e-01
-5.57416201e-01 -5.30087411e-01 -4.82642263e-01 -8.94190788e-01
-7.71336019e-01 4.57040668e-01 -4.69051898e-01 -9.87312049e-02
1.22063383e-01 6.91726446e-01 -1.17798853e+00 -3.42944890e-01
7.63377368e-01 9.32326138e-01 -4.66855019e-02 6.87699378e-01
1.52998596e-01 3.91068459e-01 -2.88335830e-01 7.53491521e-02
1.34564996e-01 -1.58556148e-01 -3.29648465e-01 9.23381224e-02
-2.37095341e-01 1.47284672e-01 -5.24352491e-01 8.07258010e-01
3.42987061e-01 4.16790783e-01 9.47439969e-01 -1.09630477e+00
-8.76839995e-01 1.30224943e-01 -3.72748584e-01 4.36882406e-01
7.30190575e-02 -1.17922194e-01 4.35502142e-01 -1.18078935e+00
3.87535065e-01 1.19834328e+00 7.62820661e-01 2.80764878e-01
-1.80384409e+00 -3.06234390e-01 -1.64746165e-01 4.53383289e-02
-1.43471241e+00 -2.22448736e-01 1.26521909e+00 -1.00498354e+00
6.11929536e-01 1.24272704e-01 4.86059308e-01 1.02089679e+00
5.06439447e-01 1.07825063e-01 1.34384596e+00 -2.94867605e-01
5.53537190e-01 4.53359149e-02 3.80570501e-01 4.87532467e-01
-2.89870739e-01 2.53448665e-01 -1.91747934e-01 -7.06184626e-01
7.14315414e-01 1.67218402e-01 -2.49727026e-01 -3.06453049e-01
-1.41504276e+00 9.92013931e-01 9.60057676e-02 6.49992704e-01
-9.68312621e-01 -7.55911246e-02 1.39455065e-01 5.14310062e-01
9.88074481e-01 1.74770296e-01 -3.91552359e-01 2.04307064e-02
-1.02416134e+00 7.00412244e-02 9.11691070e-01 3.14307660e-01
5.36621451e-01 4.09639716e-01 2.00128764e-01 4.25111562e-01
4.44167495e-01 5.53772807e-01 8.80607903e-01 -8.60570073e-01
3.26744556e-01 9.05504748e-02 2.61826515e-01 -1.45361865e+00
-5.72953939e-01 -2.75927424e-01 -1.31407177e+00 1.03803836e-01
2.43304923e-01 -4.53768373e-01 -2.13370755e-01 1.58287489e+00
2.32290402e-01 9.20255303e-01 3.08860034e-01 6.28006458e-01
1.80007845e-01 1.16511869e+00 -1.43339336e-01 -7.54163265e-01
1.02827501e+00 -5.60925186e-01 -7.01394916e-01 6.77595973e-01
5.30683339e-01 -5.19992054e-01 8.56380105e-01 2.68844247e-01
-1.06234634e+00 -7.71503687e-01 -7.78783917e-01 4.61116612e-01
-3.03414851e-01 1.97556391e-01 -4.46183840e-03 3.77868354e-01
-1.15148008e+00 1.08199346e+00 -1.20149386e+00 -7.63273612e-02
-3.51819955e-02 3.19129750e-02 -3.21028411e-01 3.51842046e-01
-1.28528249e+00 7.48984993e-01 -2.02955276e-01 1.56975031e-01
-7.89605081e-01 -8.63474607e-01 -7.38320172e-01 -8.55397880e-02
-6.81633413e-01 -4.38162714e-01 4.40730661e-01 -4.54559058e-01
-1.61827314e+00 3.64319652e-01 -2.59631336e-01 -5.68715215e-01
4.34802979e-01 -4.91339266e-02 -8.81674409e-01 7.54949078e-02
-2.63047040e-01 6.78633600e-02 1.48053885e+00 -8.43065202e-01
6.50041625e-02 -5.60365260e-01 -4.78438497e-01 -8.97016004e-02
-4.55108941e-01 -1.14230074e-01 5.37692964e-01 -9.33712780e-01
3.35553199e-01 -1.07987964e+00 -2.95609117e-01 -3.67897063e-01
-1.42341644e-01 -2.15105161e-01 1.03758562e+00 -1.25925195e+00
1.13936615e+00 -2.24220753e+00 7.72095919e-01 3.64321232e-01
5.83805218e-02 -2.13809595e-01 -5.29635996e-02 5.83190858e-01
-5.46876788e-01 -3.78681049e-02 -3.92558962e-01 -6.32480741e-01
-1.51366308e-01 -3.43890041e-02 -7.16696620e-01 8.30517590e-01
1.07216850e-01 5.80728531e-01 -7.80975938e-01 8.39413628e-02
1.76091671e-01 9.46442068e-01 -2.71198422e-01 -8.47230386e-03
1.91240683e-01 9.85900044e-01 -2.35109866e-01 -8.00922140e-02
3.96654248e-01 -1.31301150e-01 -2.86210269e-01 -2.97507077e-01
-2.96205878e-01 -1.16529815e-01 -1.19738686e+00 1.62991750e+00
-6.20133519e-01 5.71595728e-01 -7.81687498e-02 -1.31788731e+00
1.17429817e+00 7.00003326e-01 7.64661610e-01 -3.05600613e-01
-2.03928411e-01 5.85478768e-02 -2.02091575e-01 -5.09344757e-01
2.10213125e-01 -1.50393605e-01 1.03956372e-01 5.73955774e-01
-9.84391198e-04 -1.79072693e-02 -2.64696598e-01 -1.25378162e-01
8.30620587e-01 2.22287998e-01 -1.49501801e-01 -5.68982124e-01
8.07017386e-01 -3.38376880e-01 1.00080915e-01 -9.61116850e-02
7.68115968e-02 3.02586049e-01 4.52167809e-01 -5.69755793e-01
-1.13102412e+00 -1.10184479e+00 -4.11899984e-01 4.99316722e-01
-5.19827068e-01 -5.46906404e-02 -7.49223113e-01 2.30326914e-04
-5.32704145e-02 1.01233649e+00 -6.84357166e-01 -9.69508663e-02
-4.41805601e-01 -1.09573984e+00 1.56811625e-01 2.37148628e-02
3.58162597e-02 -9.20191765e-01 -3.55451286e-01 3.45088571e-01
-4.87339906e-02 -8.46209228e-01 -3.33303332e-01 -1.17909886e-01
-1.24980664e+00 -5.66920936e-01 -9.88314688e-01 -6.81925595e-01
5.42546272e-01 -3.85021359e-01 6.28270864e-01 -7.07456052e-01
-1.90233707e-01 7.20576584e-01 9.34543163e-02 -4.42379974e-02
-4.90946889e-01 -3.97629261e-01 7.19078839e-01 6.77079260e-01
1.14224747e-01 -1.14712965e+00 -5.27609587e-01 2.97977805e-01
-7.48807013e-01 -2.21334115e-01 -8.69040191e-02 6.99246764e-01
5.54475963e-01 4.04016584e-01 7.58906007e-01 -8.82562324e-02
1.04218554e+00 -9.26183462e-01 -7.65056014e-01 -1.32735148e-01
-6.00153327e-01 1.01571731e-01 6.35021985e-01 -7.47586489e-01
-6.37315929e-01 -1.64959908e-01 2.17339471e-01 -8.70395243e-01
-3.51727381e-02 5.67476571e-01 3.79682064e-01 1.83458235e-02
7.41627753e-01 5.80133200e-01 3.43286306e-01 -8.09053183e-01
6.03212893e-01 5.23047268e-01 4.38182354e-01 -4.07857120e-01
9.59825993e-01 6.32456243e-01 1.17477193e-01 -1.25896895e+00
-7.89298713e-02 -9.77626964e-02 -1.07667232e+00 -8.88420269e-02
8.56128454e-01 -7.81414926e-01 -5.00667751e-01 5.26333630e-01
-1.09446311e+00 -3.70189011e-01 -4.97824758e-01 8.65047753e-01
-1.01568103e+00 2.90067643e-01 -1.00727510e+00 -9.64118302e-01
-3.03839564e-01 -7.89408624e-01 7.53318191e-01 -3.77104968e-01
-2.97146618e-01 -1.62402511e+00 4.92424697e-01 -2.56810725e-01
5.12113392e-01 5.74587166e-01 1.23444176e+00 -5.64682484e-01
8.82188752e-02 -3.16956699e-01 3.87792319e-01 5.35854220e-01
1.41209468e-01 -6.67631105e-02 -7.84534395e-01 -3.39131892e-01
8.28680754e-01 3.54505867e-01 2.73322910e-01 5.86954296e-01
7.60169804e-01 -2.98091114e-01 -1.55001596e-01 4.83784288e-01
1.22938383e+00 1.33198097e-01 4.47442025e-01 3.35768424e-02
6.19946480e-01 8.32054913e-01 6.91243336e-02 4.39772993e-01
4.10198390e-01 5.08690536e-01 -4.25947905e-02 2.05108508e-01
3.72955084e-01 -1.81282893e-01 4.92650270e-01 1.57444394e+00
-2.57786185e-01 5.52842021e-01 -1.02032471e+00 4.15816396e-01
-1.59794593e+00 -1.05512738e+00 -3.37055959e-02 2.46314883e+00
1.81810886e-01 -2.04349652e-01 3.10412288e-01 4.92812514e-01
7.69921958e-01 -1.01673948e-02 -4.21881735e-01 -1.25558674e-01
-1.43453345e-01 -1.22694686e-01 -2.80031413e-02 6.15277469e-01
-1.03065825e+00 3.35250616e-01 5.91153908e+00 4.74429131e-01
-1.29338324e+00 3.27077717e-01 7.47373044e-01 1.68617308e-01
-2.04187706e-01 -2.38665119e-01 -2.49755397e-01 6.82922900e-01
1.72644782e+00 -4.72860962e-01 8.31262112e-01 5.38144886e-01
6.85730278e-01 5.57675064e-01 -1.03684533e+00 1.20419323e+00
-2.04849970e-02 -1.15018058e+00 -1.04208782e-01 2.46671841e-01
7.25704312e-01 1.24691270e-01 3.16025645e-01 2.83754826e-01
-1.40140563e-01 -9.16849911e-01 5.25084734e-01 1.25591183e+00
5.62239408e-01 -6.47212744e-01 2.66668200e-01 8.04833293e-01
-1.03589845e+00 -1.16602644e-01 -4.10934091e-01 -1.77451327e-01
4.70879823e-01 6.76563680e-01 -2.00222433e-01 4.00074095e-01
4.63374704e-01 9.51797724e-01 -8.28019828e-02 6.77187443e-01
3.46051484e-01 7.82909572e-01 -2.65254766e-01 3.20198596e-01
1.45807892e-01 -9.42286849e-01 9.78008807e-01 6.69942975e-01
8.19047987e-01 1.57669172e-01 -2.32329518e-01 9.81143415e-01
4.05915082e-01 2.35558376e-01 -7.19818830e-01 -6.20419905e-02
1.51203603e-01 1.30750144e+00 -4.49763805e-01 -5.77427223e-02
-4.06893790e-01 7.81989396e-01 9.30071622e-02 8.18818450e-01
-6.73923969e-01 -5.20897657e-02 5.01097620e-01 2.58002996e-01
2.34208032e-01 -7.26455450e-01 1.78259671e-01 -1.28004587e+00
-5.68496622e-02 -3.88521016e-01 1.26608148e-01 -7.70862937e-01
-1.57261193e+00 9.14830387e-01 9.14249346e-02 -1.40591967e+00
-5.06901741e-01 -3.60733837e-01 -7.16175914e-01 1.33566868e+00
-1.13868237e+00 -5.56829214e-01 3.25414389e-01 8.25733900e-01
2.36377224e-01 -5.02579033e-01 1.24121964e+00 2.68690914e-01
-3.43890578e-01 -6.27004057e-02 7.09749401e-01 -1.84811085e-01
-3.97075824e-02 -1.08251297e+00 5.81711769e-01 3.84610236e-01
1.10915661e-01 5.12744904e-01 7.15083718e-01 -3.09812009e-01
-1.27772164e+00 -1.27630329e+00 6.76398218e-01 -4.94360536e-01
1.21677470e+00 -2.34069303e-01 -1.24393082e+00 6.45546257e-01
-8.51035938e-02 2.63400584e-01 5.71927428e-01 -3.26933861e-01
1.45712286e-01 1.60981804e-01 -1.20834434e+00 4.78667855e-01
3.47850621e-01 -7.65824974e-01 -8.58744800e-01 6.33835196e-01
6.71872556e-01 2.07609057e-01 -1.54854345e+00 1.40328154e-01
2.02791840e-01 -4.15271252e-01 1.07625115e+00 -7.30213463e-01
7.22033232e-02 -1.62936360e-01 -3.52945805e-01 -1.74715889e+00
-5.93695283e-01 -1.21196616e+00 -5.67810953e-01 9.69691992e-01
3.27918649e-01 -1.00267768e+00 4.06954169e-01 5.59409082e-01
2.71586418e-01 -7.85623014e-01 -1.25614643e+00 -8.17665517e-01
5.79724431e-01 -4.83165801e-01 3.60357553e-01 1.05368829e+00
2.22738340e-01 2.28615403e-01 -4.18213814e-01 4.17104304e-01
8.69390070e-01 3.58227156e-02 4.18934911e-01 -1.48292422e+00
-2.49098107e-01 -2.58938909e-01 -6.65463805e-01 -6.89602673e-01
4.99450445e-01 -8.71612132e-01 -5.28723776e-01 -1.01443923e+00
-4.67300624e-01 -3.89560372e-01 -3.33354622e-01 -4.05314118e-01
2.27733701e-01 -1.04356438e-01 1.15488797e-01 7.70974100e-01
1.92229033e-01 8.57508719e-01 8.11870873e-01 1.74561739e-01
-4.14964378e-01 2.36514494e-01 1.03215806e-01 8.06670904e-01
6.15224123e-01 -2.56457537e-01 -4.61457580e-01 -2.75611151e-02
-9.36591476e-02 7.09636509e-01 5.40445089e-01 -1.12001634e+00
1.90059200e-01 3.15071702e-01 2.67188579e-01 -4.19083446e-01
5.21347642e-01 -8.33626986e-01 3.97148073e-01 3.87153924e-01
-3.15495849e-01 6.79652333e-01 1.41472489e-01 9.08247948e-01
-2.53441870e-01 2.09316552e-01 5.41259646e-01 3.64536732e-01
-1.92411572e-01 3.46048146e-01 -3.67289215e-01 -2.69968450e-01
1.16861641e+00 1.59666926e-01 1.43236846e-01 -5.22562027e-01
-1.73202825e+00 -1.74086094e-01 7.30924904e-02 3.90606493e-01
6.07494771e-01 -1.56412911e+00 -8.59452903e-01 5.25955200e-01
-3.43245894e-01 -6.66382790e-01 3.72131944e-01 1.32535601e+00
-1.26349553e-01 3.43863249e-01 -1.64916500e-01 -6.90799177e-01
-5.27559102e-01 1.04504490e+00 4.38693374e-01 -1.51586726e-01
-8.27836812e-01 1.85232148e-01 -9.44714844e-02 -6.36615396e-01
-9.08282995e-02 -3.51577818e-01 -4.77253795e-01 1.34688571e-01
4.74151909e-01 7.32860446e-01 -1.25005156e-01 -1.11805224e+00
-4.52060252e-02 7.79657722e-01 6.75661504e-01 -4.33152765e-01
1.42859113e+00 -4.26329166e-01 -3.30772698e-01 1.16094387e+00
1.57513082e+00 -2.64078200e-01 -1.12857640e+00 -3.31406325e-01
2.39500210e-01 1.41927987e-01 1.52945712e-01 3.82495709e-02
-6.96735084e-01 1.00149930e+00 8.87751639e-01 7.02445447e-01
7.79762328e-01 -3.03150296e-01 4.32882458e-01 1.83198065e-01
4.13357675e-01 -6.64727926e-01 -2.31190920e-01 2.73157239e-01
1.11119413e+00 -7.99962878e-01 -4.34424460e-01 3.28506380e-02
-3.90853643e-01 1.20291138e+00 -4.70590442e-01 -7.50688314e-01
1.51268578e+00 -8.15035701e-02 -1.54538780e-01 -1.67061225e-01
-7.81383276e-01 5.08657813e-01 5.16531765e-01 4.72650588e-01
3.25957328e-01 1.93783224e-01 -1.16315097e-01 4.73784566e-01
-2.67327070e-01 -2.41680462e-02 3.35335404e-01 3.09642583e-01
6.00543730e-02 -4.69403535e-01 -6.78847909e-01 3.25936586e-01
-2.50268936e-01 -8.13509896e-03 5.40033698e-01 4.89220709e-01
-4.37620878e-01 8.49869609e-01 3.98466319e-01 -1.93930328e-01
2.55150318e-01 6.94718003e-01 2.21253429e-02 -1.51847035e-01
1.24546811e-01 2.30528504e-01 -3.99913013e-01 -3.67293775e-01
-3.19887906e-01 -1.10619998e+00 -9.83812869e-01 -1.15030251e-01
7.00108558e-02 3.99056673e-01 1.00375152e+00 7.86032379e-01
5.00039577e-01 3.63223284e-01 9.56576765e-01 -1.40904331e+00
-8.95682573e-01 -1.11425352e+00 -9.47256744e-01 4.50620830e-01
5.26491702e-01 -5.35649955e-01 -7.55358815e-01 2.36227602e-01]
|
[6.605464935302734, 3.534614086151123]
|
e87869f2-1b8b-4262-9899-0aa93a1113c9
|
bridging-the-language-gap-knowledge-injected
|
2304.03159
| null |
https://arxiv.org/abs/2304.03159v1
|
https://arxiv.org/pdf/2304.03159v1.pdf
|
Bridging the Language Gap: Knowledge Injected Multilingual Question Answering
|
Question Answering (QA) is the task of automatically answering questions posed by humans in natural languages. There are different settings to answer a question, such as abstractive, extractive, boolean, and multiple-choice QA. As a popular topic in natural language processing tasks, extractive question answering task (extractive QA) has gained extensive attention in the past few years. With the continuous evolvement of the world, generalized cross-lingual transfer (G-XLT), where question and answer context are in different languages, poses some unique challenges over cross-lingual transfer (XLT), where question and answer context are in the same language. With the boost of corresponding development of related benchmarks, many works have been done to improve the performance of various language QA tasks. However, only a few works are dedicated to the G-XLT task. In this work, we propose a generalized cross-lingual transfer framework to enhance the model's ability to understand different languages. Specifically, we first assemble triples from different languages to form multilingual knowledge. Since the lack of knowledge between different languages greatly limits models' reasoning ability, we further design a knowledge injection strategy via leveraging link prediction techniques to enrich the model storage of multilingual knowledge. In this way, we can profoundly exploit rich semantic knowledge. Experiment results on real-world datasets MLQA demonstrate that the proposed method can improve the performance by a large margin, outperforming the baseline method by 13.18%/12.00% F1/EM on average.
|
['Jianyong Wang', 'Ning Liu', 'Zhenyu Li', 'Zhengyan Zhang', 'Xiuxing Li', 'Zhichao Duan']
|
2023-04-06
| null | null | null | null |
['cross-lingual-transfer']
|
['natural-language-processing']
|
[-1.20851271e-01 3.39972042e-02 -9.82714221e-02 -3.67996275e-01
-1.11407554e+00 -8.48815501e-01 6.62081301e-01 1.20327778e-01
-4.76670831e-01 9.14556265e-01 2.45435953e-01 -5.71362793e-01
-1.18968189e-02 -9.52560544e-01 -9.11947668e-01 -1.01833425e-01
4.84659046e-01 6.62249565e-01 5.27112722e-01 -7.88000643e-01
-1.15280189e-01 -9.75976735e-02 -1.09258902e+00 5.60006201e-01
1.54625976e+00 9.76826191e-01 1.36562452e-01 6.30778521e-02
-8.48121345e-01 1.00499463e+00 -4.46600497e-01 -9.70895231e-01
8.87523070e-02 -5.00921667e-01 -1.28007317e+00 -4.82123852e-01
4.67742711e-01 -6.06942177e-02 -1.45756945e-01 9.98099864e-01
3.51318061e-01 -3.02647986e-02 2.70918816e-01 -1.13226879e+00
-1.00874484e+00 8.03454161e-01 -2.71301359e-01 1.16285078e-01
7.93324292e-01 1.42304718e-01 1.27643800e+00 -9.65009272e-01
4.56603706e-01 1.61718047e+00 4.21261966e-01 4.31589335e-01
-7.72563040e-01 -6.59294724e-01 2.27778122e-01 6.14190102e-01
-1.19858193e+00 -1.27411604e-01 6.87672734e-01 -8.97361711e-02
9.36592996e-01 2.69855648e-01 1.58834085e-01 7.81857789e-01
-3.26514021e-02 9.73556638e-01 1.45945442e+00 -5.29689908e-01
-1.82276994e-01 2.34473497e-01 2.87283003e-01 8.84533823e-01
-7.61614228e-03 -5.36209524e-01 -5.84579289e-01 -9.39730834e-03
2.16154203e-01 -3.35150748e-01 -4.93441701e-01 -3.78223136e-02
-1.26583183e+00 8.28531504e-01 5.48431516e-01 2.69913882e-01
-1.04317591e-01 -2.05296144e-01 4.82100844e-01 6.84635162e-01
2.15840980e-01 6.06721759e-01 -8.87955129e-01 1.01743236e-01
-2.63125837e-01 3.22483629e-01 1.03914285e+00 1.04104042e+00
9.48120534e-01 -4.74924833e-01 -2.92750388e-01 8.66332889e-01
2.44712934e-01 7.78603137e-01 5.43035924e-01 -8.09438109e-01
1.11037648e+00 1.13296258e+00 -4.92577255e-02 -8.41141522e-01
-5.81311919e-02 -4.11165982e-01 -6.77429438e-01 -5.10457218e-01
5.70287824e-01 -2.53179878e-01 -6.35279417e-01 1.80468535e+00
4.83752519e-01 -1.87999308e-01 5.05588770e-01 7.39319026e-01
9.74200904e-01 7.67322361e-01 1.91575691e-01 -4.42109220e-02
1.84291780e+00 -1.15129209e+00 -8.12626898e-01 -4.37122434e-01
6.94265246e-01 -8.58757436e-01 1.46134233e+00 6.91966861e-02
-7.90633261e-01 -5.46798885e-01 -8.26727152e-01 -4.71459895e-01
-7.14785159e-01 3.48426066e-02 5.05258143e-01 4.47144717e-01
-6.90386474e-01 -1.52380034e-01 -3.80613476e-01 -3.88324529e-01
2.56365865e-01 3.62193920e-02 -4.09112751e-01 -7.71789670e-01
-1.83363771e+00 1.06249261e+00 7.07877457e-01 7.19238743e-02
-5.22698343e-01 -7.80759692e-01 -8.48254085e-01 -6.06544036e-03
8.67487013e-01 -1.04987490e+00 1.20602143e+00 -7.32491195e-01
-1.35122156e+00 7.10019410e-01 -2.86926001e-01 -3.42228830e-01
3.12283874e-01 -5.07069051e-01 -5.59454322e-01 1.51833698e-01
3.70041192e-01 5.20196795e-01 3.45748901e-01 -1.09428430e+00
-7.12801695e-01 -4.58283007e-01 6.74411654e-01 4.46257323e-01
-3.02901328e-01 1.52586445e-01 -5.88135421e-01 -3.65346968e-01
-6.79033697e-02 -8.41707885e-01 2.28214219e-01 -2.84160227e-01
-1.93766281e-01 -7.60841250e-01 6.64684117e-01 -1.04491663e+00
1.13041985e+00 -1.58084750e+00 1.64876446e-01 -1.69791460e-01
-1.79446198e-03 3.93284738e-01 -1.33658424e-01 6.43244326e-01
3.39309692e-01 1.55703992e-01 -5.08518219e-01 -2.19567008e-02
3.23251069e-01 5.28358161e-01 -5.31275451e-01 -3.87606561e-01
4.03122753e-01 1.33445549e+00 -9.73413706e-01 -7.22592592e-01
-3.58215362e-01 5.62688373e-02 -5.10795116e-01 3.03257048e-01
-6.54133141e-01 5.51646233e-01 -7.15389907e-01 6.75757647e-01
4.92322177e-01 -3.36590409e-01 1.14504069e-01 -3.03205550e-01
2.45517388e-01 5.24917781e-01 -8.20056796e-01 2.03571773e+00
-8.06138933e-01 2.02716962e-01 -1.34541705e-01 -8.02996099e-01
7.54003465e-01 3.90858442e-01 -1.34854421e-01 -9.66283679e-01
-1.77409410e-01 5.63384593e-01 1.26681224e-01 -7.75958896e-01
3.18548352e-01 -4.12400663e-01 -2.61776596e-01 2.90994793e-01
1.48675531e-01 -2.33037114e-01 3.33734453e-01 4.03988242e-01
9.39262211e-01 2.23015547e-01 1.94188043e-01 -1.85187936e-01
1.14695513e+00 1.68614417e-01 4.28286284e-01 4.30355668e-01
-6.18139375e-03 2.20746570e-03 3.38617772e-01 -7.43813142e-02
-3.64581645e-01 -1.07305503e+00 1.43856004e-01 1.13187528e+00
2.59293884e-01 -4.64950085e-01 -6.96726203e-01 -1.08631790e+00
3.72490212e-02 8.12258184e-01 -3.18649918e-01 -1.85348064e-01
-9.66479361e-01 -5.48904300e-01 7.46517420e-01 3.21375817e-01
1.08277047e+00 -1.06822252e+00 -1.67413205e-01 1.28844962e-01
-7.95470536e-01 -1.60363674e+00 -3.72368753e-01 -2.56956190e-01
-7.24336207e-01 -1.06079018e+00 -5.46835661e-01 -9.73526955e-01
3.65794599e-01 2.15880005e-04 1.35387039e+00 4.27850448e-02
3.00225586e-01 5.89030385e-01 -6.18774295e-01 -2.81018078e-01
-2.78153896e-01 3.91941369e-01 -1.64479926e-01 2.11339578e-01
5.15674055e-01 -3.39678019e-01 -5.19371092e-01 3.01551789e-01
-1.07611048e+00 -5.41635230e-02 6.58790946e-01 6.53363466e-01
4.77984488e-01 -1.14302441e-01 1.00000787e+00 -9.55596745e-01
9.13064718e-01 -7.48007953e-01 -3.05591494e-01 7.71768749e-01
-3.91097784e-01 4.54382032e-01 9.50744748e-01 -1.81491584e-01
-1.51196814e+00 -5.65632463e-01 -2.42904663e-01 2.23419033e-02
-6.46400601e-02 9.99861419e-01 -6.08498275e-01 2.08928157e-02
4.85940874e-01 3.77185345e-01 -2.48358011e-01 -5.35567760e-01
6.84509277e-01 4.59661931e-01 4.89702553e-01 -1.04532087e+00
9.95384634e-01 1.36780292e-01 -3.58808458e-01 -4.47308272e-01
-1.15017509e+00 -2.95002013e-01 -3.67453247e-01 5.21561541e-02
9.21213984e-01 -9.10064042e-01 -6.59432411e-01 3.81154448e-01
-1.25658381e+00 -1.35634005e-01 7.70784914e-02 2.74784416e-01
-1.28772045e-02 5.37209511e-01 -5.72174847e-01 -3.39557171e-01
-6.26228333e-01 -1.10546362e+00 9.21476901e-01 3.75020832e-01
1.20497040e-01 -1.11725926e+00 9.39902514e-02 1.03286445e+00
4.93375152e-01 -6.53807074e-02 1.57212710e+00 -7.27924824e-01
-8.12796295e-01 2.88162604e-02 -3.57124567e-01 5.03665030e-01
3.02872241e-01 -7.48134732e-01 -6.38233781e-01 -7.81281143e-02
9.26800296e-02 -7.26853907e-01 6.32286966e-01 -4.80947942e-01
7.80613422e-01 -4.90088403e-01 -4.22289371e-02 1.50217012e-01
1.32142711e+00 1.51722496e-02 5.82932532e-01 3.82189423e-01
8.44941735e-01 9.20440018e-01 6.74648941e-01 -3.40259552e-01
1.24516320e+00 5.40397584e-01 2.34729111e-01 2.77723163e-01
-2.76581973e-01 -5.85981727e-01 4.38053757e-01 1.39200926e+00
1.53767318e-01 5.95391914e-03 -1.18191147e+00 6.35258675e-01
-1.66790342e+00 -5.31800568e-01 -1.00828096e-01 1.88042545e+00
1.36868465e+00 -4.20636646e-02 -3.35827321e-01 -3.32211256e-01
3.84893477e-01 -7.55292475e-02 -5.96374214e-01 -1.12663597e-01
-2.31757402e-01 4.27384406e-01 -4.02070917e-02 5.89920342e-01
-7.55932808e-01 1.23925138e+00 4.40108490e+00 1.07133150e+00
-7.97050774e-01 1.83913335e-01 1.97890103e-01 6.34625912e-01
-6.08633101e-01 3.28277647e-01 -9.00181890e-01 3.78751397e-01
6.54710770e-01 -2.78225034e-01 3.10725063e-01 3.35820824e-01
-3.59129101e-01 2.53406912e-02 -9.53794241e-01 7.86764681e-01
2.59240896e-01 -8.81101191e-01 5.04247248e-01 -3.41631413e-01
6.07569098e-01 -4.84720208e-02 -9.50421691e-02 1.01028478e+00
2.98110545e-01 -8.49736333e-01 3.87721032e-01 4.75614786e-01
4.97121572e-01 -7.04483688e-01 8.71386290e-01 6.28371596e-01
-1.28994417e+00 1.97667927e-02 -2.16583565e-01 -3.36386040e-02
3.63143325e-01 4.31209773e-01 -5.73621094e-01 1.33146667e+00
7.85483539e-01 3.38619173e-01 -8.58535171e-01 6.61891162e-01
-7.28858352e-01 7.47285724e-01 -2.49671906e-01 -6.95046037e-02
2.56664366e-01 -3.48755479e-01 3.99281770e-01 9.19046819e-01
2.76566446e-01 2.80250758e-01 2.35166773e-01 7.72516191e-01
-4.87381428e-01 5.35465002e-01 -4.43087518e-01 9.65912417e-02
3.84015411e-01 1.10994875e+00 -5.68011962e-02 -4.14676756e-01
-6.81320369e-01 9.89219189e-01 5.68605542e-01 4.50467825e-01
-6.86509728e-01 -5.31009793e-01 2.06319660e-01 -1.30406305e-01
7.31691197e-02 -2.50422299e-01 1.29661798e-01 -1.43070531e+00
5.20383239e-01 -1.36814916e+00 6.94309652e-01 -8.60027969e-01
-1.59111106e+00 7.01460719e-01 1.05280146e-01 -8.68365943e-01
-1.59164116e-01 -5.47056317e-01 -2.53212810e-01 1.04561830e+00
-2.05777526e+00 -1.47247338e+00 -1.99665859e-01 8.96946251e-01
4.64542776e-01 -1.16125032e-01 7.49207318e-01 4.89038140e-01
-2.31569052e-01 6.01270378e-01 -2.33278170e-01 3.35642785e-01
9.72086012e-01 -1.32424796e+00 1.94767699e-01 7.60639489e-01
2.38863900e-01 8.95718873e-01 3.29984069e-01 -5.50095499e-01
-1.72873449e+00 -1.12785602e+00 1.34891534e+00 -7.35715568e-01
8.60577524e-01 -2.29305267e-01 -1.35409188e+00 8.30193520e-01
7.04651535e-01 -1.96532309e-01 6.43889010e-01 1.80264771e-01
-7.33802021e-01 -3.69753689e-01 -9.06960428e-01 7.63255477e-01
8.40020478e-01 -8.28325152e-01 -1.19061732e+00 3.34618837e-01
1.19419467e+00 -3.59397203e-01 -1.08297539e+00 6.16558492e-01
1.26267478e-01 -5.34196496e-01 8.66821408e-01 -6.92052901e-01
3.54474545e-01 -5.63029051e-01 -2.24305958e-01 -1.30286646e+00
2.64492363e-01 -3.90204638e-01 -6.59876764e-02 1.43255913e+00
6.36934936e-01 -1.01024115e+00 1.92990884e-01 4.07220453e-01
-1.79739162e-01 -7.25657344e-01 -1.06509149e+00 -6.69207931e-01
5.59473336e-01 -2.77227074e-01 7.65382051e-01 1.15680957e+00
-4.83372696e-02 9.76982415e-01 -7.64551163e-02 2.46386722e-01
4.25793469e-01 4.36764210e-01 7.67237425e-01 -1.09966433e+00
-2.80475050e-01 -1.25588194e-01 8.10112581e-02 -1.47490227e+00
4.42980349e-01 -1.32911217e+00 -1.87917441e-01 -1.72412694e+00
-3.99041083e-03 -4.38932657e-01 -1.51712447e-01 6.71054006e-01
-6.61887467e-01 -1.01353593e-01 1.49700060e-01 4.91379797e-02
-8.13893855e-01 8.89334679e-01 1.47058499e+00 -2.63892084e-01
2.23152325e-01 -3.74804169e-01 -7.79158056e-01 5.18550873e-01
7.70436704e-01 -3.29683304e-01 -5.75601637e-01 -1.00950265e+00
5.40946484e-01 6.75521716e-02 2.05672771e-01 -7.79175043e-01
3.87357861e-01 5.97041585e-02 -2.56775975e-01 -2.53559619e-01
1.77009046e-01 -8.08025539e-01 -3.38440955e-01 2.66032606e-01
-1.60888553e-01 4.09180373e-01 3.39497894e-01 4.55687195e-01
-7.28484511e-01 -9.52568427e-02 2.00198278e-01 -2.13687360e-01
-8.87330949e-01 2.17591584e-01 1.06982693e-01 7.40043044e-01
6.36217773e-01 3.92198533e-01 -7.20919013e-01 -2.90135682e-01
-2.53559858e-01 7.90628910e-01 -4.33214009e-02 6.21080339e-01
4.18444604e-01 -1.43113351e+00 -1.01081181e+00 -2.13364094e-01
4.97480452e-01 9.03453603e-02 3.93176198e-01 9.21603262e-01
-4.12275523e-01 7.74104416e-01 -4.39138897e-03 -3.66477937e-01
-8.63256216e-01 4.16273654e-01 4.02682632e-01 -7.44660974e-01
-1.36195362e-01 5.93919814e-01 1.67672798e-01 -1.14154112e+00
-2.13966250e-01 -3.94466341e-01 -3.86011720e-01 3.21370363e-03
2.93539971e-01 9.20195803e-02 1.46460295e-01 -5.62995732e-01
-2.89776832e-01 6.94030285e-01 -1.84342146e-01 -1.05816359e-02
8.10157120e-01 -2.05740660e-01 -5.24290919e-01 3.32284451e-01
1.07075787e+00 2.45241195e-01 -4.57604676e-01 -7.75747240e-01
3.04966509e-01 -8.61137658e-02 -3.88844132e-01 -1.18218100e+00
-8.05608034e-01 9.67547417e-01 1.97311550e-01 1.40886709e-01
1.07896757e+00 3.00697863e-01 1.23949599e+00 7.02256501e-01
6.66730523e-01 -7.78352737e-01 6.76868036e-02 9.60505366e-01
1.08810782e+00 -1.43175197e+00 -4.94990915e-01 -5.61590254e-01
-5.75004399e-01 7.01267123e-01 8.57477069e-01 3.29820395e-01
2.38989666e-01 -3.45301300e-01 3.19186956e-01 -2.74908274e-01
-5.86118340e-01 -4.75330055e-01 4.95251000e-01 3.36225212e-01
4.67898309e-01 7.90336076e-03 -3.94842476e-01 8.07277977e-01
-3.50223243e-01 -9.97313410e-02 -4.50659432e-02 8.39799166e-01
-3.45133305e-01 -1.46608496e+00 -2.86347568e-01 2.72600591e-01
-4.62888122e-01 -4.60039973e-01 -3.53571802e-01 9.35818672e-01
9.29227993e-02 1.23456728e+00 -5.66561162e-01 -9.74012539e-02
6.80836320e-01 4.80878502e-01 5.41458905e-01 -6.01393640e-01
-7.18137205e-01 -6.35281205e-01 2.69830793e-01 -2.64507264e-01
-4.44166034e-01 -3.23062539e-01 -1.38926673e+00 1.37269525e-02
-1.99370772e-01 5.67299545e-01 2.85710067e-01 1.39166713e+00
3.10932636e-01 4.20792848e-01 1.64341316e-01 3.06655943e-01
-5.48362613e-01 -9.76980865e-01 -2.93692015e-02 3.43457639e-01
1.15496412e-01 -4.47987497e-01 -1.81616962e-01 -8.76815766e-02]
|
[10.749554634094238, 8.08152961730957]
|
f6339727-539f-4e4e-ae63-a4f12481c272
|
customer-lifetime-value-prediction-using
|
1703.02596
| null |
http://arxiv.org/abs/1703.02596v3
|
http://arxiv.org/pdf/1703.02596v3.pdf
|
Customer Lifetime Value Prediction Using Embeddings
|
We describe the Customer LifeTime Value (CLTV) prediction system deployed at
ASOS.com, a global online fashion retailer. CLTV prediction is an important
problem in e-commerce where an accurate estimate of future value allows
retailers to effectively allocate marketing spend, identify and nurture high
value customers and mitigate exposure to losses. The system at ASOS provides
daily estimates of the future value of every customer and is one of the
cornerstones of the personalised shopping experience. The state of the art in
this domain uses large numbers of handcrafted features and ensemble regressors
to forecast value, predict churn and evaluate customer loyalty. Recently,
domains including language, vision and speech have shown dramatic advances by
replacing handcrafted features with features that are learned automatically
from data. We detail the system deployed at ASOS and show that learning feature
representations is a promising extension to the state of the art in CLTV
modelling. We propose a novel way to generate embeddings of customers, which
addresses the issue of the ever changing product catalogue and obtain a
significant improvement over an exhaustive set of handcrafted features.
|
['Roberto Pagliari', 'C. H. Bryan Liu', 'Benjamin Paul Chamberlain', 'Marc Peter Deisenroth', 'Angelo Cardoso']
|
2017-03-07
| null | null | null | null |
['value-prediction']
|
['computer-code']
|
[-4.99003798e-01 -1.12369314e-01 -5.08484244e-01 -9.52153265e-01
-4.82790232e-01 -5.40633559e-01 4.13550287e-01 6.65305853e-01
-3.57796878e-01 1.91989332e-01 3.90710145e-01 1.54067948e-01
-1.77493691e-01 -1.06173146e+00 -5.37737906e-01 -3.95793676e-01
-1.90994740e-01 1.03957009e+00 -4.63710874e-01 -8.37867141e-01
2.37100661e-01 4.43957031e-01 -1.58419263e+00 4.11154002e-01
1.11769646e-01 1.67316341e+00 7.62493759e-02 7.03041196e-01
-2.48217806e-01 7.31141448e-01 -1.92377433e-01 -9.26354706e-01
3.51174921e-01 1.37411475e-01 -4.08182919e-01 1.16767012e-01
-7.65789077e-02 -3.42359096e-01 -3.11801195e-01 5.32194138e-01
3.62987041e-01 1.08613461e-01 3.83659303e-01 -1.38136482e+00
-1.09452462e+00 1.01297581e+00 -9.16732848e-02 1.48270994e-01
4.44874652e-02 -1.87600017e-01 1.65278292e+00 -5.94541252e-01
5.86340070e-01 9.77978766e-01 7.90340900e-01 8.06330144e-02
-1.32302749e+00 -3.45090508e-01 2.36312360e-01 3.27854425e-01
-8.52532268e-01 -2.20722213e-01 8.00895870e-01 -4.05402660e-01
1.24192595e+00 1.77904248e-01 9.80925381e-01 1.09727371e+00
3.74682814e-01 1.15510976e+00 4.45746958e-01 -1.79746315e-01
1.69021264e-01 7.74594545e-01 2.24084601e-01 1.60327509e-01
-9.83893797e-02 1.60103515e-01 -4.38931286e-01 -2.93370243e-02
5.34005284e-01 7.08174467e-01 4.55901206e-01 -4.86459225e-01
-8.52829337e-01 1.51122582e+00 5.40408611e-01 2.37449661e-01
-6.45674825e-01 1.41821325e-01 5.95766664e-01 7.19258606e-01
4.80220973e-01 4.46328700e-01 -9.58951235e-01 -4.11632895e-01
-6.94391131e-01 4.10283566e-01 1.02613902e+00 1.09954870e+00
4.42511827e-01 -7.69051025e-03 1.76615268e-01 9.18694019e-01
2.17664987e-01 2.68699616e-01 9.18853462e-01 -6.38680458e-01
3.32451146e-03 5.98304212e-01 1.27188429e-01 -1.04501843e+00
-3.40025902e-01 -7.59212971e-01 -1.95672214e-01 -3.08278084e-01
5.85938282e-02 7.30694234e-02 -6.94076061e-01 1.10569549e+00
-7.93988332e-02 -3.66592199e-01 2.66647842e-02 8.24146390e-01
5.86339891e-01 7.33018994e-01 3.59911919e-02 -1.29019069e-02
1.07119644e+00 -1.02761889e+00 -5.39245427e-01 -3.08684915e-01
6.53079212e-01 -7.45181799e-01 6.23364091e-01 7.64174819e-01
-7.50068069e-01 -5.72220683e-01 -8.02709043e-01 1.67482474e-03
-5.80419362e-01 -2.91145027e-01 1.33044720e+00 4.31358963e-01
-4.79445189e-01 8.89808595e-01 -5.68386972e-01 -3.07656556e-01
3.32676113e-01 6.76592231e-01 -2.92075425e-01 -2.37270966e-01
-1.16777754e+00 9.65721250e-01 2.53337622e-01 1.53487353e-02
-6.97480440e-01 -6.07388794e-01 -1.11639190e+00 3.02794307e-01
7.23785302e-03 -2.23173529e-01 1.46257603e+00 -1.20554101e+00
-1.08801281e+00 6.05158448e-01 3.06431144e-01 -9.39803898e-01
1.96514651e-01 -4.26097989e-01 -8.34018350e-01 -5.77079952e-01
-1.82984561e-01 5.43664813e-01 6.43963575e-01 -9.85544860e-01
-8.59288216e-01 -5.35275221e-01 -3.57986361e-01 -1.27276361e-01
-4.87445265e-01 -3.68722081e-01 -1.06189929e-01 -4.93729979e-01
-1.23550922e-01 -8.80644560e-01 -4.16822463e-01 -5.55200517e-01
1.12335876e-01 -2.46577993e-01 5.55509150e-01 -7.05798388e-01
1.05861330e+00 -2.09686947e+00 -1.42293260e-01 2.69272506e-01
9.49476436e-02 -4.69062850e-02 -2.11204946e-01 8.18536520e-01
1.41685009e-01 -2.80554414e-01 4.12362486e-01 -1.55937269e-01
5.04406393e-01 5.48217833e-01 -2.99593508e-01 8.34014490e-02
5.14211766e-02 1.20931506e+00 -7.92224288e-01 5.19261844e-02
3.86111885e-01 5.02377093e-01 -7.84211695e-01 2.15874732e-01
-4.84139949e-01 2.61991266e-02 -4.05569702e-01 8.89189780e-01
4.97528583e-01 -6.57901019e-02 3.91920328e-01 -3.43984276e-01
5.02532199e-02 -3.68106216e-02 -9.34128702e-01 1.39835024e+00
-9.05970931e-01 4.49151754e-01 -2.02363893e-01 -1.08138871e+00
1.41731334e+00 -2.15886578e-01 7.64041960e-01 -9.58131254e-01
5.99323869e-01 1.09837338e-01 -2.70567864e-01 -4.70831811e-01
9.45255339e-01 -3.80385935e-01 -6.23028994e-01 1.21429913e-01
3.97590011e-01 9.39958319e-02 -2.29629442e-01 1.04609236e-01
8.90357733e-01 -1.46413743e-01 2.39697576e-01 2.26057649e-01
4.15978301e-03 1.54834762e-01 5.18271923e-01 2.59630531e-01
-2.14892387e-01 2.84500122e-01 3.32628965e-01 -9.27059233e-01
-1.33567786e+00 -9.19682264e-01 -1.63428634e-01 1.55209172e+00
-1.63830027e-01 -3.26316744e-01 -2.99795061e-01 -5.19279957e-01
9.22880828e-01 9.41926181e-01 -5.26972890e-01 -1.71170682e-01
-3.65073055e-01 -5.74292123e-01 -2.62228101e-01 9.95543122e-01
-1.36829630e-01 -9.64295745e-01 -1.81500211e-01 7.03230441e-01
1.08852834e-01 -9.50080216e-01 -5.21191359e-01 5.48275352e-01
-8.63553464e-01 -5.87882340e-01 -2.50708640e-01 -9.57903504e-01
9.17522237e-02 -1.21287711e-01 1.38719594e+00 -4.08748031e-01
-3.90783429e-01 4.69280958e-01 -5.57726860e-01 -5.44636309e-01
-2.44904265e-01 1.25851680e-03 2.21213058e-01 2.40260512e-01
9.98953581e-01 -3.41190279e-01 -6.16514206e-01 2.62144029e-01
-5.63891470e-01 -6.84347570e-01 6.33215606e-01 1.10372412e+00
8.23923767e-01 3.05481315e-01 8.48972261e-01 -1.22923291e+00
7.98320353e-01 -9.41074193e-01 -3.02221894e-01 -1.36127368e-01
-1.27244949e+00 1.33486181e-01 5.81244826e-01 -2.43137106e-01
-7.13876307e-01 2.88749367e-01 -5.55325866e-01 -3.16323042e-01
2.22491205e-01 5.29411495e-01 5.45371026e-02 2.51253814e-01
1.33437112e-01 -1.22972894e-02 7.31365904e-02 -8.18552136e-01
8.62539589e-01 1.03198135e+00 4.86248136e-01 -2.76162047e-02
1.14530742e-01 1.82665452e-01 -4.20413882e-01 -5.47279239e-01
-7.27429986e-01 -8.75584245e-01 -7.83164620e-01 -2.40223464e-02
3.37187827e-01 -8.84865761e-01 -1.01566529e+00 -8.17054957e-02
-6.93500221e-01 2.23454952e-01 -8.03352237e-01 5.76803923e-01
-7.84736872e-01 -3.57106596e-01 -7.56639004e-01 -7.01879323e-01
-5.60247481e-01 -6.98153734e-01 8.75836968e-01 1.12902664e-01
-3.58313918e-01 -9.83636200e-01 4.42333110e-02 6.59308136e-01
5.10610461e-01 3.30184460e-01 9.64226782e-01 -1.13761163e+00
-2.58748800e-01 -9.28352177e-01 7.26999491e-02 7.48543978e-01
-2.35327780e-01 -3.87117386e-01 -7.54703045e-01 -4.37814474e-01
4.02080314e-03 -2.23986939e-01 8.90853167e-01 6.16209567e-01
7.94656336e-01 -3.26997101e-01 -2.39769295e-01 3.32973778e-01
1.56973207e+00 2.77331769e-01 3.17847043e-01 5.01763046e-01
2.99003571e-01 7.72978067e-01 9.13754344e-01 9.92271721e-01
6.79616988e-01 5.53377271e-01 8.08635592e-01 3.27295870e-01
4.45243031e-01 -4.97871101e-01 2.81305522e-01 1.00507140e+00
1.48761347e-01 -1.49564624e-01 -3.48692030e-01 9.26958978e-01
-1.94573641e+00 -9.43414748e-01 -2.93816403e-02 1.92496955e+00
2.36952633e-01 3.68031442e-01 4.66338545e-01 2.82990247e-01
4.92904186e-01 -5.65675795e-02 -7.70981014e-01 -1.23781824e+00
2.07606286e-01 1.70198470e-01 9.62474823e-01 1.68280587e-01
-9.21968043e-01 8.22312176e-01 6.55865431e+00 4.56075400e-01
-9.01294827e-01 2.62537867e-01 5.87291598e-01 -4.89008069e-01
-6.26344383e-01 -3.35804254e-01 -9.53335881e-01 3.32326889e-01
1.42224181e+00 -1.58881724e-01 5.27168930e-01 1.47372472e+00
3.44754159e-02 3.72214526e-01 -1.28958571e+00 1.08422267e+00
2.18226925e-01 -1.38337123e+00 -2.43105978e-01 2.11534739e-01
5.52119255e-01 1.56068996e-01 4.64383006e-01 7.53802657e-01
4.81507510e-01 -1.17086470e+00 6.73498511e-01 6.43837750e-01
3.35626692e-01 -1.35024154e+00 1.18081141e+00 8.44187737e-02
-9.16736841e-01 -8.41636419e-01 -3.52217168e-01 -7.66801089e-02
6.56349421e-01 7.19774187e-01 -9.62312639e-01 -4.90221828e-02
7.20522106e-01 9.83512282e-01 -1.50736779e-01 6.58821285e-01
5.30194581e-01 2.52163172e-01 -5.17079271e-02 -3.20030600e-01
4.36698467e-01 -1.47709563e-01 5.04920073e-02 1.16879296e+00
3.76385063e-01 -2.98812479e-01 -4.79146792e-03 4.99682218e-01
-3.62217039e-01 2.63899416e-01 -4.76278871e-01 -6.60346389e-01
7.02408180e-02 1.31045735e+00 -2.46014178e-01 1.60340294e-01
-4.65793937e-01 1.05854368e+00 1.18511572e-01 -3.41457069e-01
-6.41611755e-01 -3.01588088e-01 9.72028017e-01 5.44485986e-01
7.82281816e-01 -2.34718665e-01 -8.58521610e-02 -9.29339707e-01
-2.21140072e-01 -6.66390479e-01 2.52356589e-01 -2.15347737e-01
-1.68528724e+00 5.30185044e-01 -6.59593403e-01 -1.15530729e+00
-4.23655093e-01 -6.67857647e-01 -2.15752929e-01 4.25932616e-01
-1.50885010e+00 -1.40351677e+00 9.31368023e-02 3.54272097e-01
8.25126708e-01 -4.06232804e-01 9.59397614e-01 4.51793492e-01
-9.34128091e-03 5.51224172e-01 7.94901550e-01 3.83634642e-02
3.84053439e-01 -1.24162054e+00 4.13354367e-01 -2.09892169e-01
2.68206298e-01 3.34626347e-01 7.68347442e-01 -4.47336465e-01
-1.75817263e+00 -9.56687987e-01 1.34108770e+00 -6.96074903e-01
8.38434458e-01 -6.17392778e-01 -3.29101145e-01 1.02358675e+00
-7.72941709e-02 -4.40885536e-02 1.13603020e+00 6.53077841e-01
-1.29303381e-01 -7.67822444e-01 -1.47704327e+00 -1.93289116e-01
6.56740546e-01 -1.11878157e-01 -3.96826386e-01 1.96613997e-01
8.58767152e-01 7.64991269e-02 -1.55400229e+00 -9.49504506e-03
1.18868208e+00 -7.92596102e-01 8.78798127e-01 -7.98836410e-01
3.76259685e-01 5.68211913e-01 -6.72338426e-01 -1.41225076e+00
-5.28445780e-01 -3.36265266e-01 -2.99890429e-01 1.50491309e+00
3.36243242e-01 -4.28211331e-01 1.17287683e+00 9.06089723e-01
1.11820973e-01 -9.78321195e-01 -6.38526678e-01 -5.67680895e-01
-2.43409902e-01 -5.62483072e-01 1.00029945e+00 7.48466015e-01
3.25417742e-02 2.89959520e-01 -6.48190260e-01 -3.09006214e-01
4.72037137e-01 3.79566133e-01 5.22583008e-01 -1.50185788e+00
-4.36242640e-01 -1.14381816e-02 -8.98674846e-01 -8.39859605e-01
-1.54414773e-01 -1.04492950e+00 -3.84762049e-01 -1.36696613e+00
-8.72766003e-02 -5.42584538e-01 -7.44756341e-01 -3.66639197e-02
9.26949978e-01 1.20834984e-01 3.98640007e-01 3.40793654e-02
-3.95526975e-01 3.95285219e-01 1.09716642e+00 -2.79130071e-01
-3.22061241e-01 3.76636744e-01 -1.17936075e+00 4.12903935e-01
6.12669408e-01 -3.52042288e-01 -2.09073022e-01 -3.17552418e-01
6.43648565e-01 1.07027292e-01 5.50260320e-02 -4.23845738e-01
-9.27919745e-02 -3.20633091e-02 6.59570277e-01 -6.27341986e-01
6.28091872e-01 -1.19839704e+00 2.23284796e-01 2.32888773e-01
-4.52855796e-01 2.74977744e-01 -1.63976058e-01 9.34373140e-01
-3.62702876e-01 -2.29339287e-01 4.62761432e-01 -3.52369756e-01
-1.13336349e+00 2.32096389e-01 -9.65128243e-02 -6.45939529e-01
1.20380187e+00 -6.03855290e-02 3.11838347e-03 -5.67927122e-01
-1.18218124e+00 4.29645509e-01 7.97008350e-02 9.50826228e-01
6.86281145e-01 -1.49528396e+00 -5.02015889e-01 5.35833240e-01
2.96862990e-01 -5.85853815e-01 2.06673905e-01 2.81745851e-01
-3.51630867e-01 6.10011816e-01 -3.98980469e-01 -2.62391090e-01
-8.93773019e-01 8.64785194e-01 -7.79501423e-02 -4.11693245e-01
-3.68448377e-01 9.66335118e-01 -5.51845014e-01 -5.14479458e-01
6.50329739e-02 -1.72627509e-01 -6.49846256e-01 6.03015482e-01
2.68890947e-01 3.22644114e-01 3.28978926e-01 -9.19663727e-01
-2.19550252e-01 2.81124145e-01 -6.18481338e-01 3.19869608e-01
2.24456978e+00 -3.01622182e-01 3.92857969e-01 5.08073747e-01
1.53889298e+00 -4.00241464e-01 -1.16640949e+00 -2.69080281e-01
1.09690756e-01 -6.85386121e-01 4.85848874e-01 -9.50023293e-01
-1.23327172e+00 7.76612282e-01 9.00009334e-01 4.34456736e-01
9.81634617e-01 1.44899502e-01 1.62756765e+00 2.18213499e-01
4.88971412e-01 -1.65677178e+00 -2.72077452e-02 2.63826847e-01
6.92411065e-01 -1.53551805e+00 -2.72040397e-01 1.99724138e-01
-1.41850579e+00 1.08619797e+00 8.85042623e-02 -4.42375273e-01
1.06577289e+00 1.51827261e-01 -5.37218899e-02 -2.01115340e-01
-8.12727571e-01 -1.86163679e-01 -9.51049402e-02 6.73189342e-01
5.12493014e-01 5.55441856e-01 -2.04807609e-01 1.39667940e+00
-5.33864975e-01 8.85465816e-02 2.92479455e-01 8.87128592e-01
-5.60705662e-01 -1.37526309e+00 1.15333319e-01 9.57287669e-01
-5.00649691e-01 -1.00027479e-01 -4.85108122e-02 4.86743569e-01
3.23662728e-01 9.82092798e-01 2.44269013e-01 -8.06767404e-01
6.76921368e-01 7.50377716e-04 3.58687967e-01 -5.01400769e-01
-8.24783862e-01 1.93222426e-02 1.86247572e-01 -3.92105311e-01
6.95305988e-02 -7.21446097e-01 -1.00687230e+00 -7.49293566e-01
-4.03412074e-01 2.41036668e-01 1.21243131e+00 4.73532230e-01
4.11429107e-01 4.12084877e-01 1.11731064e+00 -8.47420096e-01
-8.16717923e-01 -7.06520200e-01 -1.25963151e+00 6.98457599e-01
2.08053067e-01 -5.46968818e-01 1.84513554e-02 8.97170380e-02]
|
[9.763537406921387, 5.999914646148682]
|
2a2e1b5a-52b2-4c2c-b8e4-f270e146e347
|
odim-an-efficient-method-to-detect-outliers
|
2301.04257
| null |
https://arxiv.org/abs/2301.04257v1
|
https://arxiv.org/pdf/2301.04257v1.pdf
|
ODIM: an efficient method to detect outliers via inlier-memorization effect of deep generative models
|
Identifying whether a given sample is an outlier or not is an important issue in various real-world domains. This study aims to solve the unsupervised outlier detection problem where training data contain outliers, but any label information about inliers and outliers is not given. We propose a powerful and efficient learning framework to identify outliers in a training data set using deep neural networks. We start with a new observation called the inlier-memorization (IM) effect. When we train a deep generative model with data contaminated with outliers, the model first memorizes inliers before outliers. Exploiting this finding, we develop a new method called the outlier detection via the IM effect (ODIM). The ODIM only requires a few updates; thus, it is computationally efficient, tens of times faster than other deep-learning-based algorithms. Also, the ODIM filters out outliers successfully, regardless of the types of data, such as tabular, image, and sequential. We empirically demonstrate the superiority and efficiency of the ODIM by analyzing 20 data sets.
|
['Yongdai Kim', 'Kunwoong Kim', 'Jongjin Lee', 'Jaesung Hwang', 'Dongha Kim']
|
2023-01-11
| null | null | null | null |
['memorization']
|
['natural-language-processing']
|
[-3.01343977e-01 -3.81467462e-01 2.88024962e-01 -2.73101032e-01
-6.39470279e-01 -1.00243457e-01 4.47772682e-01 4.73255575e-01
-3.59377712e-01 5.26844680e-01 4.29285131e-02 9.30207819e-02
-1.01480372e-02 -7.18647480e-01 -1.08241570e+00 -7.37924576e-01
-2.35898942e-02 5.95591784e-01 -4.11873870e-02 2.17008740e-01
4.89277780e-01 5.46295702e-01 -1.39705062e+00 -1.15364879e-01
1.23165834e+00 1.02074134e+00 -1.09526142e-01 1.82659686e-01
-8.44802558e-02 6.41568542e-01 -9.02555048e-01 -4.70042378e-02
4.10301685e-01 -4.56952840e-01 -9.63559151e-02 2.89262056e-01
4.69227910e-01 -5.24721384e-01 -3.66969705e-01 1.23757923e+00
5.03980994e-01 3.13546568e-01 5.75355828e-01 -1.51176989e+00
-4.53064829e-01 4.34209883e-01 -7.22203255e-01 2.11877137e-01
1.56556055e-01 -5.62902279e-02 7.24946678e-01 -1.45523596e+00
3.37224126e-01 9.60667014e-01 8.16018343e-01 -8.49157944e-02
-9.30598497e-01 -8.09321225e-01 2.89834559e-01 1.57525823e-01
-1.39364779e+00 -2.47479677e-01 8.16175163e-01 -4.86747891e-01
5.04018843e-01 1.32097363e-01 6.73345268e-01 1.00069046e+00
3.79377365e-01 9.60600853e-01 4.71535712e-01 4.83771861e-02
4.68260705e-01 -4.94668901e-01 2.13003010e-01 4.99596715e-01
8.82008374e-01 5.37270270e-02 -6.83764875e-01 -3.81428123e-01
6.86511457e-01 7.50570416e-01 -3.41959625e-01 -3.60232353e-01
-1.22569633e+00 8.10169578e-01 4.61194575e-01 1.30349770e-01
-3.13760996e-01 1.80454180e-01 5.38347185e-01 6.32962048e-01
5.67782879e-01 5.61356306e-01 -1.93445966e-01 -5.12065478e-02
-1.18306661e+00 1.27931178e-01 6.47149801e-01 9.85125244e-01
8.99560690e-01 3.11582744e-01 1.55560747e-01 6.51569843e-01
4.23384905e-01 3.53769809e-01 7.71188855e-01 -5.78987062e-01
3.98248374e-01 7.75658250e-01 2.49463350e-01 -1.39477038e+00
-5.63543320e-01 -7.06006229e-01 -1.19289994e+00 4.57714535e-02
2.83411860e-01 -1.34527460e-02 -1.07447743e+00 1.40432358e+00
2.59191185e-01 6.31517768e-01 -1.46110475e-01 9.23969090e-01
8.56107473e-01 6.06387138e-01 -4.65878069e-01 -3.56428444e-01
6.23193383e-01 -7.68550694e-01 -7.35370278e-01 -1.88771039e-01
7.20193505e-01 -8.23645949e-01 8.03738534e-01 7.52791643e-01
-6.49879992e-01 -5.81603765e-01 -1.09676862e+00 1.82105508e-02
-2.14236483e-01 3.63959938e-01 5.03564179e-01 1.65839881e-01
-6.22219563e-01 6.79208398e-01 -1.00624526e+00 -3.42207849e-01
3.11668068e-01 3.94283026e-01 -4.87644732e-01 -1.51076064e-01
-7.23289847e-01 3.10769379e-01 5.63872218e-01 4.06648129e-01
-1.03751457e+00 -4.64898646e-01 -1.03789735e+00 1.22193068e-01
4.40435469e-01 -4.32153225e-01 9.89069581e-01 -1.06608558e+00
-1.14941740e+00 6.58026159e-01 -3.00469011e-01 -5.02037704e-01
8.55335295e-01 -8.19096446e-01 -5.08323908e-01 -3.39152843e-01
3.51031661e-01 -1.28663287e-01 1.29291344e+00 -1.33273745e+00
-5.03813088e-01 -5.22853136e-01 -5.62162638e-01 -1.02351196e-01
-4.70675863e-02 -1.98892653e-01 -4.77131337e-01 -9.77016926e-01
9.14033890e-01 -8.40600133e-01 -2.38661110e-01 -2.34586760e-01
-8.68509889e-01 -2.09068611e-01 8.92150104e-01 -4.36464548e-01
1.20704794e+00 -2.61590815e+00 -2.41370156e-01 7.12779284e-01
6.40195251e-01 5.26035354e-02 2.45945472e-02 5.16385436e-01
-3.24409008e-01 -1.53958395e-01 -3.01925451e-01 -4.08154547e-01
7.48946965e-02 3.11350554e-01 -7.34717131e-01 9.12318170e-01
-3.33737172e-02 5.58537304e-01 -9.94409859e-01 -1.64086819e-01
2.02240050e-01 -5.28766289e-02 -5.77819407e-01 4.60597426e-01
1.22358352e-01 7.05570221e-01 -2.21546039e-01 7.76869357e-01
1.02284729e+00 -2.20969155e-01 -3.71146232e-01 -2.21263189e-02
5.10638161e-03 -1.22958273e-01 -1.58781326e+00 1.58671260e+00
-7.66591877e-02 6.92705631e-01 -4.78746980e-01 -9.12090778e-01
1.20744693e+00 1.74268335e-02 5.50026476e-01 -6.08399332e-01
1.26591444e-01 7.72209525e-01 -1.07032761e-01 -4.40681249e-01
5.79667389e-01 1.55276552e-01 -1.74701344e-02 4.53124195e-01
2.01325398e-02 3.66313815e-01 3.87303293e-01 1.86699167e-01
1.22132659e+00 -1.82290435e-01 2.35479146e-01 -5.33157252e-02
2.02591836e-01 -4.02099550e-01 1.07644987e+00 1.08044434e+00
-2.75810361e-02 1.03041530e+00 6.05488956e-01 -8.09220016e-01
-8.22747767e-01 -1.23946571e+00 -9.57700089e-02 7.15561986e-01
3.08536440e-01 -5.03409088e-01 -3.33964616e-01 -5.83914220e-01
3.77946049e-01 5.11236906e-01 -4.67400908e-01 -4.24494207e-01
-5.91045499e-01 -7.99619079e-01 2.20986545e-01 4.64387298e-01
5.98027408e-01 -1.00578928e+00 -3.08502257e-01 1.98569372e-01
6.61654910e-03 -8.30875397e-01 -3.51129144e-01 1.44622207e-01
-1.15376759e+00 -1.30050588e+00 -5.57630897e-01 -5.44159710e-01
1.20146847e+00 5.76007776e-02 1.17000973e+00 2.90962726e-01
-3.61864977e-02 -2.01981999e-02 -3.65206182e-01 -5.32923222e-01
-1.12664676e-03 -2.02483565e-01 3.18983674e-01 2.84095496e-01
5.90745389e-01 -5.31594157e-01 -5.73315501e-01 2.37114787e-01
-1.14455771e+00 -4.12717015e-01 5.91036677e-01 9.65727329e-01
6.75234437e-01 2.49678001e-01 3.82951647e-01 -8.79910648e-01
6.71482623e-01 -7.62449145e-01 -7.53164411e-01 -3.22248459e-01
-2.37971753e-01 1.02404682e-02 7.95255363e-01 -2.25452617e-01
-5.05296111e-01 2.61821114e-02 -1.59853652e-01 -6.01313293e-01
-2.58445203e-01 7.77957320e-01 -2.07304820e-01 1.51418045e-01
4.67183590e-01 2.15515539e-01 -3.44179720e-01 -6.27350152e-01
-1.20705754e-01 3.73287827e-01 6.26007795e-01 -5.09671926e-01
1.01907742e+00 6.78202212e-01 -1.10359021e-01 -7.83582926e-01
-8.91466796e-01 -7.20223010e-01 -6.29559159e-01 -3.49532366e-02
2.91948825e-01 -9.20507669e-01 -4.64020312e-01 8.12456846e-01
-1.15775216e+00 1.41995594e-01 -4.19702768e-01 7.26012170e-01
-4.17656779e-01 5.59438944e-01 -4.81096506e-01 -5.83458304e-01
-1.28910944e-01 -1.07552922e+00 9.72653747e-01 4.46371250e-02
-1.63415432e-01 -7.23684371e-01 2.49879420e-01 -2.89316833e-01
5.54759204e-02 6.90532029e-01 6.79292381e-01 -1.31040645e+00
-7.14810431e-01 -6.00312054e-01 -2.19936937e-01 3.77023548e-01
1.34968400e-01 1.71380877e-01 -7.11648285e-01 -7.10325241e-01
2.31516272e-01 -3.49553898e-02 1.06670082e+00 2.94592857e-01
1.42557728e+00 -2.83642590e-01 -1.57178849e-01 1.03202188e+00
1.31328380e+00 1.91314504e-01 6.06363237e-01 6.16354823e-01
7.92724609e-01 -2.01873034e-02 6.34595215e-01 9.01677489e-01
-3.86878140e-02 1.17303640e-01 7.07382977e-01 -2.34366328e-01
5.76783776e-01 -3.22697461e-01 3.33503217e-01 1.09194493e+00
9.79581252e-02 -3.27397764e-01 -1.03244627e+00 6.21247053e-01
-2.23303723e+00 -7.13794827e-01 -2.65362054e-01 2.54544234e+00
3.42543304e-01 1.78310797e-01 -2.12394074e-01 2.59830862e-01
8.18667531e-01 1.22992858e-01 -8.69494736e-01 6.04976574e-03
-6.07458837e-02 -6.40845448e-02 3.33070010e-01 1.22112587e-01
-1.33280921e+00 7.27281690e-01 5.98317432e+00 4.66217846e-01
-9.80445445e-01 -2.38517329e-01 4.57071334e-01 3.08448845e-03
-8.26632306e-02 -7.97053650e-02 -4.93295103e-01 8.84615898e-01
5.76946616e-01 -6.13457114e-02 3.17977443e-02 1.08269179e+00
1.81427419e-01 -2.12447479e-01 -1.35547376e+00 1.31225741e+00
4.22063440e-01 -9.90610003e-01 2.41407946e-01 -8.46182778e-02
8.17499638e-01 1.97638914e-01 1.41252086e-01 3.74750048e-01
2.58999050e-01 -9.17840838e-01 5.03714502e-01 7.06270218e-01
2.47904405e-01 -1.02787793e+00 1.20177901e+00 2.83783346e-01
-8.37540567e-01 -2.48422831e-01 -8.02575529e-01 -3.28907929e-02
-6.57083169e-02 1.34022760e+00 -4.80530083e-01 6.71721339e-01
9.27148759e-01 1.14262748e+00 -6.67528808e-01 1.71532083e+00
-2.94193506e-01 4.89210129e-01 -4.54815269e-01 5.18800974e-01
2.37870947e-01 -5.85549295e-01 7.09451020e-01 9.03689742e-01
9.54100132e-01 -3.16008359e-01 3.49683911e-01 7.88662851e-01
-4.53792453e-01 2.03203991e-01 -7.54427016e-01 1.44168913e-01
3.92141998e-01 8.32388401e-01 -7.13464022e-01 -3.69617164e-01
-3.16501439e-01 1.11929202e+00 2.24182531e-01 6.75956011e-01
-7.83031702e-01 -5.10883093e-01 6.30320132e-01 -4.56183776e-02
1.83325648e-01 -3.54144007e-01 -3.24078500e-01 -1.50494969e+00
3.43465328e-01 -1.01618540e+00 3.65161389e-01 -6.87845469e-01
-1.49872434e+00 2.98875481e-01 -5.09659708e-01 -1.79310858e+00
-8.19614902e-02 -5.17870545e-01 -9.08003688e-01 2.53160179e-01
-1.26963317e+00 -5.47222853e-01 -8.75337720e-01 6.03622556e-01
3.41428995e-01 -3.89292955e-01 3.28675479e-01 4.03365403e-01
-9.13661838e-01 5.36668241e-01 7.72505999e-01 4.61649835e-01
1.03599906e+00 -1.29402363e+00 3.97890955e-01 1.35644329e+00
3.82860929e-01 8.27074945e-01 8.12441349e-01 -8.19858909e-01
-1.14135075e+00 -1.33916974e+00 6.34327590e-01 -2.51360685e-01
6.24606431e-01 -2.39722446e-01 -1.13167346e+00 1.00530052e+00
-1.86848029e-01 2.57788777e-01 6.42316580e-01 8.45778286e-02
-1.95628628e-01 -2.79776752e-01 -9.74112630e-01 4.66533393e-01
7.73422539e-01 -2.32509539e-01 -7.12533712e-01 4.85589594e-01
5.50720155e-01 -7.86736310e-01 -5.37299037e-01 4.76933032e-01
-1.52918920e-02 -1.37994039e+00 6.31263554e-01 -4.89811003e-01
3.39070797e-01 -6.50642037e-01 -2.96882559e-02 -1.49891365e+00
-2.03224778e-01 -6.62465274e-01 -4.44574982e-01 9.78659153e-01
5.51923504e-03 -9.65223789e-01 6.67723596e-01 2.96165347e-01
-6.47239268e-01 -4.98417616e-01 -9.34639156e-01 -9.38447297e-01
-3.23977649e-01 -5.23333490e-01 9.02463198e-01 1.13042927e+00
-4.92952496e-01 -3.03263783e-01 -5.02159297e-01 5.11833310e-01
6.16007090e-01 1.55669421e-01 1.26797462e+00 -1.65662062e+00
8.38913769e-02 -1.35167062e-01 -8.24627161e-01 -1.05261683e+00
4.69684005e-02 -7.32157648e-01 1.50171921e-01 -1.36297715e+00
-3.79811525e-02 -1.87326059e-01 -7.80188918e-01 3.61687064e-01
-2.38065913e-01 1.01556443e-01 -8.05451795e-02 5.75412750e-01
-8.65571260e-01 9.02617216e-01 7.15649307e-01 1.02000377e-04
-1.60802409e-01 1.35939837e-01 -3.51225853e-01 1.17357111e+00
7.16203749e-01 -7.06060410e-01 -6.75184950e-02 -5.44087172e-01
4.86540139e-01 -5.02432048e-01 3.02353352e-01 -1.45146251e+00
3.91119272e-01 1.76688239e-01 8.34022045e-01 -1.06862247e+00
-9.10453051e-02 -9.93864238e-01 -6.13965131e-02 5.11520684e-01
7.07723796e-02 5.89091539e-01 1.70403067e-02 8.88303041e-01
-6.79261923e-01 -1.27675742e-01 6.25761330e-01 6.45175716e-03
-8.86917770e-01 5.99449635e-01 -1.69310495e-01 2.00069442e-01
1.00164557e+00 -1.50476769e-01 -1.46678492e-01 -4.43534970e-01
-6.57165468e-01 2.41245031e-01 6.85491443e-01 3.13455433e-01
1.08150840e+00 -1.62674153e+00 -6.27013266e-01 8.52183044e-01
4.14464682e-01 3.94652307e-01 4.11419757e-02 9.97071922e-01
-8.51594746e-01 -4.48844582e-02 4.47628535e-02 -9.28398013e-01
-8.09222221e-01 5.95184743e-01 2.41601124e-01 -1.04768850e-01
-9.04646218e-01 7.74438143e-01 2.12136671e-01 -5.34464538e-01
5.07531404e-01 -5.86352468e-01 2.36992061e-01 6.53252602e-02
3.64308238e-01 5.10146141e-01 2.60420501e-01 -4.91754562e-01
-1.48117200e-01 4.24354255e-01 -1.69256017e-01 4.58149374e-01
1.24185705e+00 2.22139984e-01 -5.28231680e-01 8.06638598e-01
1.14749980e+00 5.25835603e-02 -1.06499612e+00 -3.93025458e-01
4.06597584e-01 -8.76542985e-01 -3.54704499e-01 -8.25198069e-02
-1.19265687e+00 6.58292115e-01 4.99305069e-01 1.21227913e-01
1.09937370e+00 -3.79809260e-01 1.04500127e+00 7.19390154e-01
2.04998970e-01 -1.28556395e+00 3.76242638e-01 7.73090243e-01
9.19107080e-01 -1.58879709e+00 1.08354487e-01 9.00094435e-02
-3.01193088e-01 1.12790751e+00 7.70767987e-01 -6.88795447e-01
6.82896078e-01 5.49881766e-03 1.00062959e-01 -3.11281115e-01
-2.33458668e-01 -3.18703987e-02 1.83278531e-01 1.09698370e-01
2.71127135e-01 -2.64322490e-01 -5.94125055e-02 3.69078934e-01
-2.63551801e-01 -2.77133703e-01 6.10159338e-01 8.22373748e-01
-5.63271582e-01 -6.18325531e-01 -5.97935498e-01 7.59554923e-01
-2.69850731e-01 -5.27361557e-02 -2.68720359e-01 8.68068993e-01
1.34326622e-01 7.19282448e-01 4.56115723e-01 -2.84472018e-01
4.06581670e-01 -1.68700948e-01 -2.49412075e-01 -5.94449520e-01
-2.64157414e-01 9.10195336e-02 -6.03395641e-01 -8.55568171e-01
-6.20157719e-02 -4.17683065e-01 -1.25521314e+00 -3.10231596e-01
-3.29588681e-01 3.44659984e-01 3.31532300e-01 1.08703017e+00
4.02215540e-01 5.10355473e-01 7.47882605e-01 -7.00298667e-01
-3.69895428e-01 -8.54924560e-01 -8.30065250e-01 8.12707424e-01
7.57567048e-01 -6.94990695e-01 -8.21403086e-01 -2.81383038e-01]
|
[7.663820266723633, 2.453977346420288]
|
aa3478e2-cee7-4221-ac31-6ebfd1ade8b9
|
data-driven-real-time-short-term-prediction
|
2211.09814
| null |
https://arxiv.org/abs/2211.09814v1
|
https://arxiv.org/pdf/2211.09814v1.pdf
|
Data-driven Real-time Short-term Prediction of Air Quality: Comparison of ES, ARIMA, and LSTM
|
Air pollution is a worldwide issue that affects the lives of many people in urban areas. It is considered that the air pollution may lead to heart and lung diseases. A careful and timely forecast of the air quality could help to reduce the exposure risk for affected people. In this paper, we use a data-driven approach to predict air quality based on historical data. We compare three popular methods for time series prediction: Exponential Smoothing (ES), Auto-Regressive Integrated Moving Average (ARIMA) and Long short-term memory (LSTM). Considering prediction accuracy and time complexity, our experiments reveal that for short-term air pollution prediction ES performs better than ARIMA and LSTM.
|
['Sabri Pllana', 'Iryna Talamanova']
|
2022-11-16
| null | null | null | null |
['air-pollution-prediction', 'time-series-prediction']
|
['miscellaneous', 'time-series']
|
[-2.40126979e-02 -5.18119931e-01 1.67509973e-01 -1.47688761e-01
-4.53660816e-01 -9.23089162e-02 3.94745797e-01 2.25613624e-01
-4.33263242e-01 1.06401777e+00 2.60169923e-01 -7.53660500e-01
-4.12336946e-01 -1.30671322e+00 -2.17751861e-01 -7.14432657e-01
3.74974906e-02 -1.64750755e-01 2.99162030e-01 1.48258001e-01
-1.04925834e-01 6.37387812e-01 -1.26187944e+00 3.27867977e-02
1.24536645e+00 1.00914431e+00 -1.62338186e-02 8.58847737e-01
-3.00960690e-01 4.14393544e-01 -6.77418053e-01 -1.01440795e-01
-4.30572070e-02 -1.58701032e-01 -4.88138050e-01 -6.03438437e-01
-2.74040252e-02 1.69104457e-01 -1.86465666e-01 8.55793417e-01
4.01666552e-01 7.66625345e-01 5.04210353e-01 -8.04358840e-01
-3.98584008e-01 -1.04394041e-01 5.12221977e-02 5.89971900e-01
-1.63800120e-01 3.58328342e-01 2.53140152e-01 -4.65754360e-01
-2.41405264e-01 1.05153060e+00 1.11295795e+00 3.44768375e-01
-8.58546615e-01 -6.87132776e-01 3.59402239e-01 2.65511990e-01
-1.16020024e+00 -7.66512156e-02 2.58967042e-01 -6.15979433e-01
1.05913675e+00 6.08219922e-01 7.43633151e-01 8.04829717e-01
9.23950732e-01 -7.51699135e-02 1.21899271e+00 -1.12837605e-01
2.25244284e-01 7.36160725e-02 3.48303556e-01 1.78158432e-01
1.91419706e-01 6.72908366e-01 1.71828941e-01 -3.85214627e-01
2.49764666e-01 8.17772985e-01 4.55587506e-02 1.16043139e+00
-7.94536173e-01 5.43161273e-01 4.95397508e-01 6.18294358e-01
-9.11384881e-01 1.94983140e-01 2.10278451e-01 2.85132587e-01
1.11480296e+00 1.21404994e-02 -7.50503421e-01 1.44080846e-02
-8.48643720e-01 1.67717114e-01 4.38051134e-01 3.74461144e-01
5.65479159e-01 2.63815850e-01 -6.38848364e-01 6.56563818e-01
5.05725622e-01 1.34162915e+00 4.14084584e-01 -5.33291399e-01
2.37250552e-01 1.64358824e-01 3.69781584e-01 -1.26847196e+00
-4.91279602e-01 -4.75505501e-01 -1.17532218e+00 1.15771607e-01
3.42776477e-01 -5.20537555e-01 -5.98157048e-01 1.06282341e+00
2.06149548e-01 6.48643494e-01 -2.04322800e-01 2.42683083e-01
5.58412850e-01 1.45552409e+00 6.73032701e-01 -4.21982467e-01
1.26626587e+00 -8.31061244e-01 -1.19273281e+00 1.31157607e-01
9.24770907e-02 -3.80674750e-01 6.63473725e-01 3.40183169e-01
-7.51611412e-01 -9.38181221e-01 -2.51934588e-01 4.93213117e-01
-1.10290790e+00 -1.17596433e-01 1.73840091e-01 1.12603188e+00
-5.55468798e-01 1.00547290e+00 -9.83337700e-01 -1.07477680e-02
1.49850771e-01 2.16299519e-01 3.46321762e-01 3.33961219e-01
-1.56759393e+00 8.34885657e-01 -3.71378139e-02 2.46160895e-01
-4.08192396e-01 -9.03185248e-01 -4.43814278e-01 -1.16837010e-01
-2.55679488e-01 -8.08446646e-01 1.23434043e+00 -4.52263027e-01
-1.36259437e+00 -1.23532258e-01 -6.57531619e-01 -6.50366187e-01
1.25708297e-01 -4.58664805e-01 -1.35646880e+00 -2.84452379e-01
-2.97487915e-01 -1.96293488e-01 8.56260657e-01 -5.28035164e-01
-9.17681873e-01 -4.00769621e-01 -6.19994462e-01 -5.70804775e-01
-5.16337216e-01 3.10269415e-01 4.78375018e-01 -5.59135318e-01
-5.27087331e-01 -9.26543772e-01 -6.58447802e-01 -3.41975689e-01
-1.17949784e-01 -7.37470984e-01 4.49069917e-01 -1.39555287e+00
1.95570350e+00 -1.81325209e+00 -8.25079262e-01 2.45363012e-01
-1.90552160e-01 8.20063531e-01 2.72677630e-01 2.69314289e-01
-7.51532465e-02 1.73083737e-01 -1.33779287e-01 -4.71167229e-02
-1.18614063e-01 2.08156228e-01 -6.28209889e-01 5.25067151e-01
1.27097145e-01 8.76767159e-01 -9.31637228e-01 -1.47973195e-01
3.99277210e-01 7.94143796e-01 1.23402402e-01 2.71891624e-01
-3.87200058e-01 6.90681934e-01 -4.95874137e-01 1.41248748e-01
5.86377144e-01 -3.68190371e-02 -5.77104032e-01 2.00432882e-01
-8.34996879e-01 4.61332381e-01 -7.79337406e-01 8.48150909e-01
-9.44648504e-01 7.35604644e-01 -5.76541305e-01 -3.07455301e-01
8.73787344e-01 1.02066743e+00 2.34085843e-01 -8.71257663e-01
2.78810915e-02 2.09790483e-01 -3.92134905e-01 -9.66156125e-01
2.22536996e-01 -2.83627212e-01 4.56803501e-01 2.60036469e-01
-8.88546228e-01 5.28789222e-01 -1.23229161e-01 -6.36867940e-01
6.58117056e-01 -3.41570377e-01 1.77940369e-01 -3.41663361e-02
7.91173577e-01 -1.35402620e-01 5.12944996e-01 4.26782936e-01
-1.90128222e-01 -1.46205351e-01 -3.44053417e-01 -6.64445519e-01
-7.13474035e-01 -9.40792978e-01 -3.02335262e-01 1.08099258e+00
-6.09463692e-01 -1.35768736e-02 -4.26656872e-01 -3.69298160e-01
-1.92772761e-01 1.22050786e+00 -3.81806195e-01 -6.18118746e-03
-5.79611838e-01 -1.00289524e+00 3.19606394e-01 4.15716678e-01
5.18322885e-01 -9.87090945e-01 -4.63867307e-01 4.53152835e-01
4.16071080e-02 -4.03256327e-01 -2.68741637e-01 -3.54655832e-01
-1.04784203e+00 -5.29632032e-01 -8.01695645e-01 -1.63722232e-01
4.06932443e-01 2.50370413e-01 9.81398642e-01 -9.46849883e-02
7.18325302e-02 3.83568347e-01 8.02950561e-03 -1.15574777e+00
-4.54905033e-01 -2.00987309e-01 -7.03798607e-03 1.03900895e-01
5.28347373e-01 -6.88509464e-01 -8.20201576e-01 1.73114583e-01
-7.30597794e-01 -4.92170721e-01 3.05722535e-01 1.18380271e-01
6.96963251e-01 8.39486122e-01 8.83136749e-01 -6.51168942e-01
6.98867798e-01 -8.39422166e-01 -7.25581110e-01 3.17386150e-01
-1.01909053e+00 -3.16132069e-01 8.12627971e-01 -3.80215168e-01
-1.25560760e+00 -2.92462587e-01 -5.47298729e-01 -9.96881723e-02
-6.54921174e-01 3.07508796e-01 2.28222117e-01 3.24238092e-01
4.26406413e-01 3.74531597e-01 -5.24458051e-01 -1.03310454e+00
-1.99769456e-02 7.99048901e-01 2.38442987e-01 -5.19143306e-02
7.30654120e-01 3.87996525e-01 1.16856769e-01 -1.19863689e+00
-1.03092563e+00 -5.87495506e-01 -5.96563816e-01 -5.10523558e-01
1.37074387e+00 -9.09445286e-01 -8.80722046e-01 3.70819896e-01
-1.45607281e+00 -3.04075360e-01 -1.17613249e-01 6.12676322e-01
2.03695655e-01 -1.67054579e-01 -4.04397696e-01 -1.38076150e+00
-6.77009463e-01 -4.29817259e-01 5.32493353e-01 3.02791119e-01
-2.24075556e-01 -1.85419881e+00 5.60982406e-01 -6.00737780e-02
9.35970962e-01 5.14586985e-01 7.39749014e-01 -3.93091530e-01
-2.33264655e-01 -2.98871845e-01 -1.37818128e-01 5.23634017e-01
2.35521346e-01 2.25410834e-02 -9.94959891e-01 -3.02635338e-02
2.65352815e-01 6.43892527e-01 7.51229286e-01 6.71127200e-01
1.33030522e+00 -8.09467673e-01 -4.55071539e-01 1.70667842e-01
1.26154518e+00 6.46715224e-01 7.66487241e-01 1.72369435e-01
4.58069444e-01 6.28652930e-01 4.79358137e-01 1.89181805e-01
2.45606884e-01 1.20379418e-01 1.17108837e-01 -1.61400884e-01
1.13577977e-01 -1.38198957e-01 4.58103240e-01 9.65376616e-01
-3.49508733e-01 -3.21676254e-01 -1.02883816e+00 6.83344841e-01
-1.56486964e+00 -1.40996480e+00 -1.01608431e+00 2.26872420e+00
6.48024201e-01 5.94086088e-02 3.67253155e-01 2.07222655e-01
6.88318193e-01 5.03088608e-02 -1.92411453e-01 -5.56593478e-01
4.14357334e-01 4.68750656e-01 7.29394853e-01 5.10405838e-01
-1.18358195e+00 2.37031534e-01 7.09862614e+00 7.67282009e-01
-1.21750450e+00 5.91165006e-01 6.15222454e-01 1.43096805e-01
-1.67280942e-01 -3.29688847e-01 -1.02903163e+00 9.53174889e-01
1.90459120e+00 -2.93308735e-01 2.85519034e-01 4.71326888e-01
1.09100199e+00 8.55024904e-02 -3.53468210e-01 5.24519265e-01
-6.29648089e-01 -9.78858292e-01 -1.94873184e-01 1.33881480e-01
9.40986872e-01 1.79991335e-01 2.38385767e-01 2.02545404e-01
1.05256535e-01 -1.18781924e+00 1.89761639e-01 1.61140943e+00
3.79912943e-01 -8.25590968e-01 5.62998652e-01 6.22743309e-01
-1.38274920e+00 -2.39927158e-01 -5.38370430e-01 -4.04551268e-01
4.11426574e-01 1.26913702e+00 -4.59249049e-01 4.48293298e-01
6.04116380e-01 6.38818204e-01 -5.71827352e-01 1.24449074e+00
-1.95700988e-01 1.23141325e+00 -2.75389314e-01 -1.87484175e-01
1.73978955e-01 -2.68994868e-01 5.75944304e-01 1.27116001e+00
7.09790587e-01 2.99241364e-01 -5.00335023e-02 9.01682854e-01
5.39125621e-01 1.50103876e-02 -5.32486260e-01 9.76601988e-03
1.92688644e-01 7.61061072e-01 -1.82805538e-01 -5.01640201e-01
-6.97364509e-01 3.95664096e-01 -5.71863115e-01 4.84096318e-01
-1.00214648e+00 -4.16616023e-01 6.11954927e-01 2.89329290e-01
1.53218672e-01 -3.25498313e-01 -3.39655131e-01 -3.90945256e-01
-2.75661618e-01 -3.16987425e-01 5.21741629e-01 -4.45875198e-01
-1.43830585e+00 5.41437864e-01 2.63266731e-03 -1.19380951e+00
9.77969542e-02 -5.16152918e-01 -1.37385774e+00 1.26751184e+00
-1.84593141e+00 -6.35496497e-01 -6.34557307e-02 5.53116798e-01
4.57839310e-01 1.51861712e-01 9.84511018e-01 5.94286859e-01
-8.28968465e-01 -1.57152668e-01 3.92877847e-01 -1.98276997e-01
2.16826662e-01 -1.11515999e+00 6.31224155e-01 7.44714141e-01
-3.81334454e-01 6.51536584e-01 8.36611450e-01 -9.53186333e-01
-8.20273399e-01 -1.96348679e+00 1.56871402e+00 -3.87153387e-01
6.47805989e-01 3.87871832e-01 -1.02074599e+00 4.46844548e-01
1.79879591e-01 -1.76435873e-01 1.00444806e+00 -1.85284447e-02
2.59229213e-01 -4.00192797e-01 -1.03811491e+00 2.81990409e-01
2.80060679e-01 -4.48051155e-01 -5.28864682e-01 5.49900055e-01
9.59858358e-01 4.80134159e-01 -1.20119786e+00 2.61412770e-01
5.02480686e-01 -8.49675953e-01 9.69351649e-01 -6.80737197e-01
-3.72616909e-02 -2.48145893e-01 1.43268824e-01 -1.33190894e+00
-6.73928499e-01 -2.99579740e-01 -1.91248730e-01 1.16359079e+00
5.00773489e-01 -1.01271057e+00 2.50347286e-01 1.04356325e+00
-5.24872867e-03 -4.47970808e-01 -7.87839234e-01 -1.12601483e+00
3.07460606e-01 -7.96225846e-01 1.01303387e+00 4.96442944e-01
-5.03215671e-01 1.99964549e-02 -5.75660825e-01 5.07318735e-01
3.88172030e-01 -6.90900162e-03 3.33968967e-01 -1.53530574e+00
7.34603480e-02 -2.61324197e-01 3.00479472e-01 -7.83761680e-01
7.03882501e-02 -5.70097685e-01 -7.23407343e-02 -1.79085112e+00
-3.21000904e-01 -2.08493948e-01 -7.80868948e-01 2.87422389e-01
-2.49134123e-01 -7.31709749e-02 -2.69129932e-01 -3.30965042e-01
-3.33350003e-01 5.79821885e-01 9.83133137e-01 -2.24245310e-01
-5.80809772e-01 9.61147189e-01 -8.77045467e-02 7.50159264e-01
1.17918193e+00 -8.52304041e-01 -1.37010351e-01 -3.70745838e-01
4.79994535e-01 7.75866881e-02 5.38141787e-01 -1.26917613e+00
3.47614497e-01 -5.71155012e-01 4.93506193e-01 -9.64922547e-01
1.01999290e-01 -1.15054929e+00 7.22140729e-01 1.04004979e+00
-2.73657948e-01 2.38883898e-01 2.89360732e-01 8.73183846e-01
-1.51100140e-02 2.06472389e-02 7.01266289e-01 -3.48220058e-02
-1.57576591e-01 5.77684641e-01 -1.13440573e+00 -5.51017225e-01
7.74564564e-01 2.18746021e-01 -3.13654602e-01 -1.92837000e-01
-9.22878802e-01 1.96848795e-01 -2.84333348e-01 2.99808770e-01
4.85566437e-01 -1.51067257e+00 -4.22005832e-01 1.43644735e-01
-3.84408742e-01 -3.53617519e-01 6.09372616e-01 1.11617744e+00
-1.60899282e-01 9.77269828e-01 2.89574206e-01 -1.57980412e-01
-1.33683848e+00 6.92683637e-01 6.07302666e-01 -4.15292025e-01
-5.03657043e-01 4.91690159e-01 -2.50378579e-01 -1.18389852e-01
1.64496198e-01 -5.64137042e-01 -5.09877801e-01 -3.52906026e-02
1.08591437e+00 1.46590829e+00 -5.72165363e-02 -5.63161433e-01
-1.11918256e-01 3.43245268e-01 6.37870014e-01 2.04317614e-01
1.38205636e+00 -3.12149435e-01 -5.13693511e-01 1.07855642e+00
1.06058741e+00 1.34813577e-01 -7.99381375e-01 -1.42717093e-01
2.86176234e-01 -3.55568916e-01 4.05034721e-01 -7.64618695e-01
-7.24510431e-01 1.11250722e+00 1.01423717e+00 7.64316678e-01
1.11613178e+00 -6.11809373e-01 1.68807757e+00 4.50674981e-01
-1.49147317e-01 -9.89208519e-01 -3.70040148e-01 6.32470131e-01
5.76912165e-01 -7.99931467e-01 -1.76797464e-01 5.04652096e-04
4.14211340e-02 1.03680456e+00 1.04083613e-01 -1.92054547e-02
1.45521915e+00 -2.54657865e-01 2.15313472e-02 -1.21226706e-01
-9.29602921e-01 -3.51868659e-01 6.09398901e-01 5.21040916e-01
4.91384119e-01 4.57293481e-01 -1.89809144e-01 7.36014664e-01
-4.86820899e-02 -1.08450570e-03 -3.40665132e-01 3.37543428e-01
-9.06381309e-01 -9.57231164e-01 -9.18741047e-01 6.35680258e-01
-8.60923827e-01 -1.04451664e-01 5.27525768e-02 -9.72295627e-02
5.86384892e-01 1.54763162e+00 8.50408152e-02 -7.03981817e-02
4.34343398e-01 4.32447284e-01 -1.76333189e-01 -3.57214421e-01
-7.33214259e-01 -2.35351101e-01 -4.36914042e-02 -2.03979746e-01
-5.98825097e-01 -4.56404448e-01 -1.06235909e+00 -5.14441550e-01
-2.21250340e-01 2.55179048e-01 7.46525228e-01 1.16839314e+00
1.45464957e-01 8.76582205e-01 8.27643156e-01 -4.44828928e-01
-2.41624117e-01 -9.54982877e-01 -6.03192151e-01 1.21633001e-01
6.74182355e-01 -1.88065410e-01 -4.39558029e-01 8.85217935e-02]
|
[6.278886795043945, 2.598992109298706]
|
07624be1-53d5-4821-88db-23a7378303f2
|
champion-solution-for-the-wsdm2023-toloka-vqa
|
2301.09045
| null |
https://arxiv.org/abs/2301.09045v2
|
https://arxiv.org/pdf/2301.09045v2.pdf
|
Champion Solution for the WSDM2023 Toloka VQA Challenge
|
In this report, we present our champion solution to the WSDM2023 Toloka Visual Question Answering (VQA) Challenge. Different from the common VQA and visual grounding (VG) tasks, this challenge involves a more complex scenario, i.e. inferring and locating the object implicitly specified by the given interrogative question. For this task, we leverage ViT-Adapter, a pre-training-free adapter network, to adapt multi-modal pre-trained Uni-Perceiver for better cross-modal localization. Our method ranks first on the leaderboard, achieving 77.5 and 76.347 IoU on public and private test sets, respectively. It shows that ViT-Adapter is also an effective paradigm for adapting the unified perception model to vision-language downstream tasks. Code and models will be released at https://github.com/czczup/ViT-Adapter/tree/main/wsdm2023.
|
['Tong Lu', 'Wenhai Wang', 'Guo Chen', 'Zhe Chen', 'Shengyi Gao']
|
2023-01-22
| null | null | null | null |
['visual-grounding']
|
['computer-vision']
|
[-4.24132682e-02 2.13783532e-01 9.29451361e-02 -4.63564754e-01
-1.35375977e+00 -1.16674757e+00 6.13988698e-01 -1.52091354e-01
-3.89738590e-01 3.85838121e-01 4.06470507e-01 -6.11279011e-01
2.40745485e-01 -3.51762891e-01 -1.07627547e+00 -3.53797495e-01
6.18004382e-01 6.48901284e-01 2.72203714e-01 -2.19815552e-01
2.83516031e-02 -1.81112811e-02 -1.58517253e+00 9.26208317e-01
5.56496799e-01 1.06330180e+00 5.38753808e-01 9.72746730e-01
-1.75068289e-01 9.86177444e-01 -4.86976117e-01 -5.43330908e-01
1.73383623e-01 -3.27456266e-01 -1.40956032e+00 -5.56509346e-02
1.17306876e+00 -2.65061170e-01 -3.06327790e-01 9.51011837e-01
5.25264144e-01 1.01747878e-01 4.10597593e-01 -1.36798728e+00
-1.13519955e+00 3.75453353e-01 -2.95681477e-01 2.09364712e-01
6.57349288e-01 7.20800519e-01 1.35809922e+00 -1.17876923e+00
8.52082849e-01 1.39808846e+00 2.30702594e-01 7.89886773e-01
-1.20666814e+00 -4.47053343e-01 3.60694617e-01 7.37714231e-01
-1.33852708e+00 -7.24332035e-01 4.36637968e-01 -5.40126801e-01
1.12621236e+00 4.50596213e-01 3.11783701e-01 1.47310996e+00
-2.98275471e-01 1.13394213e+00 1.20150304e+00 -2.20950469e-01
-3.20028961e-02 1.65604576e-01 2.34373868e-01 7.10223019e-01
-3.05965431e-02 -9.83455181e-02 -6.65992796e-01 1.18778639e-01
2.81065434e-01 -4.81585383e-01 -6.08112454e-01 -4.09249008e-01
-1.30478609e+00 6.92571878e-01 6.39674127e-01 -6.50116131e-02
-2.59981245e-01 3.89455616e-01 4.07329649e-01 3.82002264e-01
1.53997943e-01 5.98910511e-01 -5.89976668e-01 -1.77766867e-02
-4.56924230e-01 3.22689652e-01 4.75035757e-01 1.12825143e+00
7.97480047e-01 -1.89813063e-01 -6.10761523e-01 6.02007687e-01
5.74790597e-01 6.06182098e-01 -7.50142634e-02 -1.35115290e+00
5.91023088e-01 4.25398141e-01 2.00424001e-01 -4.48447078e-01
-3.19570720e-01 -3.52103233e-01 -3.47287208e-01 2.36877188e-01
7.49206245e-01 4.97869849e-02 -1.18539619e+00 1.90231681e+00
3.86498988e-01 -8.43432993e-02 1.74844772e-01 1.29615271e+00
1.43167126e+00 7.49993682e-01 3.52117598e-01 3.84256214e-01
1.82753849e+00 -1.28423512e+00 -4.36158121e-01 -4.64219421e-01
5.14453053e-01 -7.93361545e-01 1.72786236e+00 3.27840567e-01
-1.20533609e+00 -7.49876857e-01 -8.42929959e-01 -7.97437906e-01
-3.91289890e-01 6.41457438e-02 4.89310265e-01 1.70589700e-01
-1.49386895e+00 -2.63445884e-01 -4.37367886e-01 -7.10335374e-01
4.86824661e-01 -1.36609659e-01 -4.32748616e-01 -4.82618898e-01
-9.30932283e-01 8.22844684e-01 8.98969397e-02 2.25039516e-02
-1.56195605e+00 -6.74966276e-01 -7.99285591e-01 -1.61522612e-01
5.62510729e-01 -1.20181370e+00 1.55256975e+00 -9.97056782e-01
-1.24221408e+00 1.39932752e+00 -4.00799334e-01 -3.75751793e-01
3.71965766e-01 -1.92308024e-01 -3.30629408e-01 2.88122237e-01
2.95314580e-01 9.67099726e-01 7.73247898e-01 -1.42800295e+00
-5.03241539e-01 -3.89393151e-01 6.13418102e-01 2.47509137e-01
4.19283897e-01 -9.71618295e-02 -7.81862676e-01 -8.64399318e-03
-3.83683443e-02 -6.85850322e-01 2.09829986e-01 4.68830317e-02
-5.47945797e-01 -6.16916776e-01 4.98766780e-01 -7.76451409e-01
5.13802171e-01 -2.30416489e+00 2.64764965e-01 -2.63189882e-01
4.25184667e-01 1.28884867e-01 -6.74644113e-01 4.79020029e-01
1.33164495e-01 -6.75931796e-02 -1.55841529e-01 -6.33781314e-01
3.02097976e-01 2.03420863e-01 -6.82350814e-01 3.70693743e-01
3.26690137e-01 1.33959711e+00 -8.95652592e-01 -2.89944828e-01
1.32880360e-01 3.25667858e-01 -4.75699395e-01 4.69868124e-01
-8.39620113e-01 5.76220930e-01 -1.48012623e-01 8.11629236e-01
7.18719244e-01 -6.31110430e-01 3.67799699e-02 -4.46548969e-01
-6.09804727e-02 4.16351706e-01 -7.36201227e-01 2.21614170e+00
-3.72257769e-01 9.03660715e-01 2.05613434e-01 -6.49373829e-01
6.30997539e-01 1.59310818e-01 -1.87815800e-01 -1.11040175e+00
5.46701588e-02 -1.31302714e-01 -3.28966945e-01 -8.06617141e-01
5.45268297e-01 2.29850292e-01 -1.60469189e-01 1.49728134e-01
4.82111216e-01 -1.24778435e-01 -6.90451544e-03 5.43307781e-01
1.01639462e+00 3.96245092e-01 1.10076785e-01 -1.89297199e-02
4.21397239e-01 2.70006984e-01 2.17357442e-01 8.68466437e-01
-5.50263941e-01 7.87253261e-01 4.84815359e-01 -1.76767007e-01
-7.12729275e-01 -1.55336785e+00 1.51746929e-01 1.41806948e+00
2.28729010e-01 -5.26856303e-01 -4.84060585e-01 -7.34956145e-01
1.24537423e-02 1.08731771e+00 -5.98046243e-01 6.86879307e-02
-1.59713447e-01 9.69802365e-02 6.81691587e-01 3.37794155e-01
3.55426937e-01 -1.11869323e+00 -5.66993237e-01 -2.89239734e-01
-5.96104741e-01 -1.23132336e+00 -3.31817150e-01 -2.09937945e-01
-1.66323423e-01 -1.20156908e+00 -4.94320393e-01 -7.42034137e-01
1.99648127e-01 4.46510166e-01 1.51522338e+00 2.87817288e-02
-4.34906110e-02 9.79645073e-01 -3.70253503e-01 -3.79783601e-01
-2.15473488e-01 2.11406723e-01 -4.21055764e-01 -7.94727430e-02
4.42809582e-01 -7.33254328e-02 -7.51535416e-01 1.13644861e-01
-6.56334639e-01 3.10319602e-01 3.63485217e-01 5.90207219e-01
7.99341619e-01 -1.00537443e+00 3.27976733e-01 -6.46312952e-01
4.79654402e-01 -5.37616491e-01 -8.23873222e-01 4.69346732e-01
-3.79501194e-01 -1.37223929e-01 2.14708939e-01 -1.84596181e-01
-9.42879617e-01 -3.00763845e-02 -3.55150431e-01 -4.61249322e-01
-5.08248031e-01 2.13134825e-01 -3.45308185e-01 1.81985959e-01
7.70034611e-01 2.52832800e-01 -2.32843608e-01 -4.78098720e-01
9.96574700e-01 5.24353206e-01 8.89116824e-01 -4.73686188e-01
6.15298629e-01 5.89861572e-01 -3.30602765e-01 -5.77972591e-01
-1.09676766e+00 -6.52445734e-01 -1.63717568e-01 -2.04006329e-01
1.37493944e+00 -1.13466072e+00 -1.25643539e+00 1.99596539e-01
-1.56588292e+00 -7.60803640e-01 -3.00543249e-01 1.48182705e-01
-5.98395407e-01 1.55528605e-01 -3.47001374e-01 -6.08542085e-01
-2.69529015e-01 -1.18489432e+00 1.26820910e+00 2.18180969e-01
-5.77639565e-02 -6.66752815e-01 2.14494497e-01 1.05731440e+00
3.34059894e-01 -1.10416748e-01 6.04849219e-01 -5.66498101e-01
-1.01829076e+00 3.54141861e-01 -6.91637695e-01 2.85843581e-01
-4.53122467e-01 -2.48343319e-01 -1.45916092e+00 -3.30604672e-01
-2.96052277e-01 -8.95064771e-01 1.17269957e+00 1.65734708e-01
1.16398656e+00 -7.48236850e-02 -9.05967355e-02 9.10901666e-01
1.36668408e+00 -2.03499660e-01 6.52162433e-01 2.81293243e-01
8.74469876e-01 6.26591265e-01 5.15732110e-01 8.34108815e-02
1.13783240e+00 6.49490297e-01 1.00644481e+00 -1.06960513e-01
-6.20782375e-01 -2.66849011e-01 4.96014476e-01 3.49653751e-01
4.68556881e-02 -5.65347552e-01 -1.18394864e+00 7.35838175e-01
-1.75414073e+00 -9.24081028e-01 -3.09785664e-01 1.73776436e+00
6.70308769e-01 -3.60998690e-01 9.52470824e-02 -6.49907053e-01
2.45076686e-01 3.72274667e-01 -6.03855073e-01 -3.19228858e-01
-3.78614664e-01 8.28382745e-02 2.82266527e-01 7.60678351e-01
-9.24509704e-01 1.27131462e+00 5.58443642e+00 5.31594515e-01
-9.00966704e-01 6.21009767e-01 3.36677134e-01 -2.11577624e-01
-6.16043270e-01 8.93891975e-02 -5.98583579e-01 1.30563036e-01
8.86909187e-01 3.90537679e-01 6.57139599e-01 5.25756419e-01
-2.44070917e-01 -9.90436822e-02 -9.95711505e-01 1.10031319e+00
3.36213797e-01 -1.47298801e+00 1.65472835e-01 -3.00439417e-01
3.79640669e-01 7.06755161e-01 2.62736589e-01 5.03407240e-01
3.32654864e-01 -1.18132794e+00 1.04742479e+00 7.69552529e-01
9.69115496e-01 -3.08444053e-01 5.42087078e-01 -2.70144437e-02
-9.94491398e-01 1.51855484e-01 -3.39252293e-01 1.19636938e-01
2.30268151e-01 4.06640619e-02 -9.02845740e-01 5.62480211e-01
9.42775309e-01 3.73293102e-01 -8.09431791e-01 9.16579902e-01
-6.45092905e-01 6.66612267e-01 -2.86784978e-03 1.37733012e-01
2.13507965e-01 1.38546437e-01 7.13973999e-01 1.12091398e+00
2.28783563e-02 -7.10447058e-02 -7.43236095e-02 1.08643270e+00
-1.97422475e-01 -5.82771711e-02 -5.47783494e-01 1.29456207e-01
2.80969888e-01 1.22653639e+00 -4.00194004e-02 -3.07110608e-01
-4.88431722e-01 1.26962149e+00 7.17884421e-01 8.68556023e-01
-1.01176226e+00 -1.19434774e-01 9.34161484e-01 -1.02628157e-01
4.81052488e-01 -2.93087274e-01 9.30403359e-03 -1.19419050e+00
-8.72047544e-02 -1.05353355e+00 6.62209094e-01 -1.48535740e+00
-1.35282671e+00 7.12030709e-01 -2.53537655e-01 -7.22022355e-01
-1.90014064e-01 -1.01611185e+00 -2.14318469e-01 8.91273081e-01
-1.63051033e+00 -1.72157514e+00 -6.48568153e-01 9.89862740e-01
4.27902013e-01 -8.86300765e-03 8.23347211e-01 4.61739190e-02
-2.23551244e-01 6.71226382e-01 -3.11248124e-01 5.83688915e-02
9.89528418e-01 -1.30481255e+00 4.86917704e-01 8.96489859e-01
5.25807261e-01 4.24989641e-01 7.67706037e-01 -2.65421122e-01
-1.72624600e+00 -1.00873137e+00 8.32733989e-01 -1.17161691e+00
7.25352049e-01 -6.96754098e-01 -9.56190526e-01 1.08623779e+00
8.15317571e-01 -9.86494869e-03 4.93264705e-01 2.70511359e-01
-1.08548558e+00 -1.34394422e-01 -8.60113859e-01 7.07048416e-01
1.33039773e+00 -1.11707592e+00 -6.77640080e-01 4.70195860e-01
1.18801415e+00 -5.32289088e-01 -5.99036992e-01 1.48708507e-01
3.04814100e-01 -9.54088628e-01 1.14252877e+00 -8.25871110e-01
2.21213788e-01 -6.56341791e-01 -6.74701691e-01 -1.00142670e+00
-3.30804586e-01 -5.56756496e-01 -9.84021928e-03 1.18377197e+00
5.83185613e-01 -6.78331733e-01 2.75641114e-01 9.87816676e-02
-3.08018774e-01 -3.26194644e-01 -1.15657258e+00 -4.37862068e-01
-1.21984631e-01 -8.77167821e-01 4.66019899e-01 8.98914754e-01
-4.86184746e-01 7.23478317e-01 -2.05083624e-01 5.01085639e-01
6.13084435e-01 2.25304291e-01 1.10942495e+00 -6.55804276e-01
-6.65628195e-01 -2.90084511e-01 -2.07101613e-01 -1.28535974e+00
1.48941711e-01 -1.23344791e+00 1.55460238e-01 -2.05875707e+00
6.06679767e-02 -3.04806530e-02 -3.62899631e-01 6.35923028e-01
3.37633101e-04 3.44523042e-01 4.29857612e-01 7.05593154e-02
-1.11692572e+00 4.95777309e-01 1.33834004e+00 -3.68293494e-01
1.81960449e-01 -2.47695789e-01 -9.28515553e-01 3.95979732e-01
9.67666388e-01 -1.49990961e-01 -4.94062454e-01 -1.08077466e+00
4.95114267e-01 -8.93114954e-02 1.00140691e+00 -6.87366188e-01
2.42051929e-01 -3.50141525e-02 1.02987386e-01 -6.78254426e-01
6.39427483e-01 -4.86887544e-01 -7.29702264e-02 1.77853301e-01
-3.28027010e-01 2.52393514e-01 4.73104924e-01 3.90770435e-01
-8.23654458e-02 1.21406265e-01 4.36403394e-01 -1.29327327e-01
-1.31853414e+00 1.99531257e-01 -1.63325131e-01 4.41600710e-01
8.00901890e-01 1.27767786e-01 -1.28194225e+00 -4.43177611e-01
-7.67656505e-01 4.48292494e-01 5.34231305e-01 5.01548946e-01
7.71241665e-01 -1.02020681e+00 -9.28595185e-01 -7.72945583e-02
8.61135006e-01 -7.41389170e-02 5.96819818e-01 9.04557943e-01
-3.69093806e-01 5.17729223e-01 -1.25953749e-01 -8.24589789e-01
-1.17324042e+00 6.97592080e-01 4.44921941e-01 2.35797882e-01
-3.22556138e-01 1.33711350e+00 4.94664133e-01 -4.74218518e-01
1.65694878e-01 -3.68398160e-01 5.10784201e-02 -3.44739892e-02
5.87432027e-01 9.54331011e-02 -1.52882412e-01 -7.05572426e-01
-6.98562622e-01 3.88481230e-01 1.34584218e-01 -3.76279831e-01
9.11320686e-01 -3.84378374e-01 -2.94574082e-01 6.13577247e-01
1.07089448e+00 1.84072964e-02 -1.23054218e+00 -2.92056590e-01
-1.95606276e-01 -2.28278115e-01 -1.67846888e-01 -1.28987098e+00
-7.09040701e-01 9.79070127e-01 7.17640579e-01 3.79113406e-02
1.01537693e+00 9.00620341e-01 3.79004925e-01 4.05759692e-01
2.44536176e-01 -6.72913373e-01 1.20696798e-01 5.86690366e-01
1.40009999e+00 -1.48718953e+00 -4.45559502e-01 -1.26308307e-01
-8.54642749e-01 5.25438130e-01 1.00641406e+00 1.53409064e-01
7.18746558e-02 -3.57052118e-01 4.97077107e-01 -4.26047564e-01
-1.08116889e+00 -6.53740823e-01 7.02127337e-01 8.87544274e-01
2.70304471e-01 2.19609693e-01 4.31817681e-01 4.22681749e-01
-1.42788485e-01 -3.43703121e-01 2.84128517e-01 4.05264348e-01
-2.32698470e-01 -6.74935937e-01 -2.71745145e-01 2.23258547e-02
-1.59810688e-02 -3.14754874e-01 -5.11136353e-01 7.72885203e-01
2.57168174e-01 1.22764170e+00 1.81837782e-01 -3.22159559e-01
4.67932284e-01 1.47750035e-01 7.29948997e-01 -6.49109900e-01
-4.90414321e-01 -2.88862497e-01 4.20110375e-01 -1.12226200e+00
-2.76364982e-01 -4.34225410e-01 -1.09857142e+00 -1.44087493e-01
2.46167928e-01 -2.46457011e-01 5.84006369e-01 7.17788219e-01
7.20269680e-01 4.96435344e-01 -1.33685321e-01 -3.95622641e-01
-3.79835874e-01 -8.33858013e-01 -3.79928425e-02 5.52410245e-01
6.63618088e-01 -4.89720345e-01 -2.38942564e-01 -7.87689686e-02]
|
[10.865164756774902, 1.704764723777771]
|
eef6c159-f48f-436f-b90b-032e4b2c17fb
|
designing-accurate-emulators-for-scientific
|
2005.02328
| null |
https://arxiv.org/abs/2005.02328v1
|
https://arxiv.org/pdf/2005.02328v1.pdf
|
Designing Accurate Emulators for Scientific Processes using Calibration-Driven Deep Models
|
Predictive models that accurately emulate complex scientific processes can achieve exponential speed-ups over numerical simulators or experiments, and at the same time provide surrogates for improving the subsequent analysis. Consequently, there is a recent surge in utilizing modern machine learning (ML) methods, such as deep neural networks, to build data-driven emulators. While the majority of existing efforts has focused on tailoring off-the-shelf ML solutions to better suit the scientific problem at hand, we study an often overlooked, yet important, problem of choosing loss functions to measure the discrepancy between observed data and the predictions from a model. Due to lack of better priors on the expected residual structure, in practice, simple choices such as the mean squared error and the mean absolute error are made. However, the inherent symmetric noise assumption made by these loss functions makes them inappropriate in cases where the data is heterogeneous or when the noise distribution is asymmetric. We propose Learn-by-Calibrating (LbC), a novel deep learning approach based on interval calibration for designing emulators in scientific applications, that are effective even with heterogeneous data and are robust to outliers. Using a large suite of use-cases, we show that LbC provides significant improvements in generalization error over widely-adopted loss function choices, achieves high-quality emulators even in small data regimes and more importantly, recovers the inherent noise structure without any explicit priors.
|
['Peer-Timo Bremer', 'Jayaraman J. Thiagarajan', 'Jim Gaffney', 'Gemma Anderson', 'Brian Spears', 'Rushil Anirudh', 'Bindya Venkatesh']
|
2020-05-05
| null | null | null | null |
['small-data']
|
['computer-vision']
|
[-1.27847821e-01 -4.05593038e-01 -1.72717616e-01 -4.02946949e-01
-1.03021419e+00 -4.97141153e-01 5.61436892e-01 4.84807551e-01
-3.33829015e-01 9.91985440e-01 -2.81640708e-01 -6.00350142e-01
-5.76698422e-01 -7.89639950e-01 -9.18929517e-01 -9.11020696e-01
-1.02124937e-01 7.41970062e-01 -3.65236431e-01 1.08651035e-01
2.64167726e-01 6.82389438e-01 -1.30074430e+00 -4.29505408e-01
1.01396453e+00 1.22155404e+00 -2.90179551e-01 4.73622918e-01
2.30984688e-02 4.83595937e-01 -6.78156376e-01 -3.59604865e-01
3.37517828e-01 -2.34558463e-01 -1.05810143e-01 -1.80779248e-01
1.26892775e-01 -4.45538461e-02 -2.08280683e-01 8.82174671e-01
6.56372726e-01 3.24800789e-01 8.49727333e-01 -1.19948721e+00
-3.38210285e-01 3.42490405e-01 -6.36228323e-01 1.41354874e-01
-1.33207321e-01 3.70676666e-01 8.04578841e-01 -4.82404232e-01
1.47608802e-01 9.91817176e-01 1.15908420e+00 8.04030076e-02
-1.71166301e+00 -7.25469112e-01 -1.24105275e-01 -2.12417364e-01
-1.37627649e+00 -2.15460628e-01 6.44774020e-01 -5.86934865e-01
3.41675758e-01 1.11155780e-02 1.67248875e-01 1.33068216e+00
5.12605190e-01 3.25047612e-01 1.04670286e+00 -1.58400521e-01
6.10971630e-01 2.00638562e-01 4.17710580e-02 3.80101390e-02
5.60585856e-01 3.22068781e-01 -2.01325223e-01 -4.57404375e-01
8.49546969e-01 -6.42172098e-02 -3.09285462e-01 -5.04381418e-01
-9.17267919e-01 9.69217300e-01 -3.61554250e-02 3.50049473e-02
-4.65002626e-01 2.91695654e-01 4.91429746e-01 4.55292404e-01
6.61337137e-01 7.81252146e-01 -6.37223601e-01 -1.89243838e-01
-1.09262085e+00 5.79905152e-01 9.69476044e-01 6.19590461e-01
5.12274027e-01 2.37627447e-01 -2.82013595e-01 7.24132657e-01
3.68246175e-02 4.89429772e-01 2.57888645e-01 -1.09863567e+00
3.98654103e-01 3.00322771e-01 4.74266797e-01 -1.00567019e+00
-6.24134481e-01 -9.27301407e-01 -1.07707238e+00 3.17337483e-01
7.00937808e-01 -3.79919231e-01 -7.52489388e-01 1.91809583e+00
2.66420573e-01 2.27093577e-01 -1.50281817e-01 6.13640964e-01
6.82005212e-02 5.02526402e-01 -7.25749135e-02 -2.03869790e-01
9.17844534e-01 -2.22532809e-01 -4.65109438e-01 -4.94433232e-02
5.27856946e-01 -6.35804832e-01 1.20634687e+00 5.52081287e-01
-1.05614591e+00 -4.57399786e-01 -1.20481634e+00 3.68557483e-01
-2.62501538e-01 -1.87992677e-01 4.68748987e-01 6.36946499e-01
-6.94491267e-01 1.01458848e+00 -9.16706085e-01 1.50251510e-02
2.26690501e-01 3.04626971e-01 -8.48845020e-02 8.39662924e-02
-1.05813038e+00 6.88789487e-01 1.66043192e-01 9.70417187e-02
-7.21724808e-01 -1.35499406e+00 -6.50350809e-01 3.20476890e-01
4.86061364e-01 -7.42527008e-01 1.25120246e+00 -8.66548479e-01
-1.48290277e+00 3.59700859e-01 2.96889514e-01 -6.98109090e-01
1.05389965e+00 -1.17390156e-01 -3.82841587e-01 -2.31188506e-01
-2.58797348e-01 1.06063478e-01 7.78813541e-01 -1.19773006e+00
-1.93916157e-01 -2.57350951e-01 -4.10077721e-01 -1.87328547e-01
-4.95229661e-02 -2.28634074e-01 -1.61907524e-01 -8.24474037e-01
-2.52296835e-01 -8.41842175e-01 -3.06181550e-01 1.37608245e-01
-3.63457918e-01 -2.58560814e-02 5.37693083e-01 -5.20577073e-01
1.05232465e+00 -2.10778666e+00 -2.75669545e-01 6.44671142e-01
5.83798103e-02 2.02637464e-01 1.03082791e-01 5.30902624e-01
-1.39300838e-01 7.88198486e-02 -5.39122641e-01 -5.84345400e-01
2.16314763e-01 2.27777034e-01 -3.71347129e-01 7.28806734e-01
2.23683968e-01 4.94688094e-01 -6.82944357e-01 -1.78924829e-01
2.72672355e-01 5.34916580e-01 -3.66727442e-01 2.94022471e-01
-3.83093387e-01 6.68114603e-01 -2.73356289e-01 2.22160578e-01
6.04943097e-01 -6.32993877e-01 -1.11080118e-01 1.31560609e-01
1.40349463e-01 1.18623845e-01 -1.53355205e+00 1.13322508e+00
-7.20502615e-01 6.26258552e-01 8.98841023e-02 -1.30359912e+00
1.04904842e+00 2.98059523e-01 6.31055355e-01 -5.84512830e-01
1.72581941e-01 3.21901947e-01 3.38835195e-02 -2.50013560e-01
1.64609086e-02 -3.62501115e-01 3.05398889e-02 3.29804122e-01
-2.28964135e-01 -1.57119259e-01 1.92950189e-01 -2.65901506e-01
1.03757906e+00 2.17574075e-01 3.63493383e-01 -3.90223026e-01
2.61832267e-01 -1.51257306e-01 7.55203784e-01 9.69540238e-01
4.18751910e-02 7.45188117e-01 7.50125110e-01 -3.41446191e-01
-1.24221623e+00 -1.11804664e+00 -5.96262395e-01 6.50648534e-01
-3.05859059e-01 9.29720476e-02 -4.19833004e-01 -2.12873682e-01
5.29068351e-01 1.07167792e+00 -6.46085024e-01 -3.20992529e-01
-4.05645013e-01 -1.12841547e+00 4.88061249e-01 6.80547655e-01
1.67087898e-01 -7.09743500e-01 -3.19431156e-01 5.07559478e-01
2.57554740e-01 -1.01203513e+00 -3.04045081e-01 4.59420919e-01
-8.35528433e-01 -9.87796426e-01 -6.19381189e-01 -5.46566695e-02
3.80297661e-01 -3.39050680e-01 1.43348467e+00 -3.49428803e-01
-1.30291849e-01 1.64912596e-01 1.66882217e-01 -5.28039157e-01
-4.90088075e-01 -6.96834028e-02 1.32813960e-01 -5.58334664e-02
2.23500490e-01 -8.14211786e-01 -5.89641690e-01 2.26509169e-01
-8.81112814e-01 -4.58412021e-01 5.76963067e-01 9.55478966e-01
5.53492844e-01 1.16112486e-01 9.25126970e-01 -8.97783875e-01
6.25754237e-01 -6.92535162e-01 -1.13474679e+00 1.59227327e-01
-9.75710630e-01 4.22543317e-01 1.07731664e+00 -5.60415387e-01
-9.08500731e-01 -5.26088119e-01 9.71655641e-03 -5.70242345e-01
-2.21614137e-01 7.84376860e-01 -2.82288697e-02 1.39747960e-02
7.53750503e-01 -7.43139163e-02 6.60897046e-02 -5.22950053e-01
-8.98344070e-02 4.61937547e-01 6.67830467e-01 -8.68789971e-01
5.61845899e-01 2.23314196e-01 5.59269905e-01 -5.07468283e-01
-8.73640358e-01 -2.46294141e-01 -2.04817042e-01 1.17584147e-01
2.98514187e-01 -8.94272029e-01 -8.32730174e-01 3.70222330e-01
-6.66544139e-01 -5.14076710e-01 -2.88709819e-01 5.95974863e-01
-7.63699591e-01 1.52948797e-01 -5.04042923e-01 -8.94245088e-01
-4.16513532e-01 -1.07743084e+00 8.70629191e-01 1.41549557e-01
-3.36909354e-01 -1.39932525e+00 2.34216765e-01 -2.21122503e-02
5.95880389e-01 7.04356611e-01 1.14999235e+00 -9.09262896e-01
-3.13289911e-01 -4.42996562e-01 -2.33927280e-01 5.29539287e-01
1.36658950e-02 3.11391026e-01 -9.01369274e-01 -3.92934412e-01
1.04027197e-01 -2.46351585e-01 5.17848492e-01 8.41894388e-01
1.60688365e+00 -1.40117928e-01 -4.92753685e-02 9.54297900e-01
1.39261866e+00 2.42106900e-01 4.52366501e-01 3.32073450e-01
2.37055436e-01 4.46321011e-01 2.51183093e-01 7.46230483e-01
2.61965953e-02 5.13553381e-01 1.23729609e-01 -1.42137110e-01
4.46149796e-01 -1.19493417e-01 -4.72346358e-02 5.08097410e-01
2.65133679e-01 -2.67165750e-01 -1.01411557e+00 3.24090540e-01
-1.76441658e+00 -6.58303201e-01 6.71349838e-02 2.76973963e+00
1.09414721e+00 4.92141187e-01 1.04784034e-03 9.98617411e-02
4.76751775e-01 -1.63454011e-01 -9.17429090e-01 -3.35203409e-01
-1.67424262e-01 3.31304133e-01 7.92428970e-01 2.94161737e-01
-1.05063879e+00 1.16158776e-01 5.89517975e+00 9.56972897e-01
-1.28337657e+00 -2.31458679e-01 1.00445259e+00 6.29152963e-03
-1.21752843e-01 -2.92267561e-01 -6.36530340e-01 6.60839796e-01
1.41207874e+00 -4.75775480e-01 2.80869782e-01 8.57595325e-01
5.56534588e-01 -9.44141857e-03 -1.37533486e+00 9.38715160e-01
-4.22032654e-01 -1.28958547e+00 -2.84708053e-01 1.16452053e-02
8.94281089e-01 -1.23094790e-01 3.06464672e-01 4.12830502e-01
5.51106572e-01 -1.33182001e+00 4.26078141e-01 6.50852382e-01
6.85776830e-01 -1.06339681e+00 1.04278266e+00 2.76499152e-01
-6.90302730e-01 1.47565445e-02 -2.40519211e-01 -3.86175476e-02
-1.28624886e-01 1.22201383e+00 -5.81494510e-01 5.46446323e-01
5.61457396e-01 4.40257967e-01 -2.67957121e-01 1.46206558e+00
2.21761957e-01 9.10353005e-01 -5.89456737e-01 1.76842928e-01
1.99225515e-01 -3.73907149e-01 4.09343272e-01 1.00740039e+00
4.92543638e-01 -5.11421680e-01 7.44554326e-02 1.02412140e+00
-1.65016681e-01 2.33351309e-02 -5.80174088e-01 5.20233922e-02
5.74821532e-01 9.74296629e-01 -4.57314134e-01 -1.49543703e-01
-3.46209526e-01 1.18395843e-01 1.35673568e-01 6.48126364e-01
-8.43129039e-01 -4.42353070e-01 7.26370871e-01 1.02218941e-01
2.12527275e-01 -3.20820749e-01 -7.98996866e-01 -8.13432753e-01
1.12513326e-01 -1.08154476e+00 4.04708624e-01 -4.34937358e-01
-1.73138297e+00 1.07683994e-01 8.68705288e-02 -1.17056406e+00
-5.72080493e-01 -6.34985924e-01 -8.27681303e-01 9.15508389e-01
-1.29532182e+00 -4.51638579e-01 -1.64053604e-01 1.96881935e-01
1.60313323e-01 7.64230639e-02 5.36998868e-01 2.56910831e-01
-7.13892698e-01 6.28428221e-01 9.19317126e-01 -9.41360146e-02
9.39803541e-01 -1.32141697e+00 2.05217451e-01 5.13830006e-01
-1.62068069e-01 6.46621406e-01 1.26804018e+00 -4.36236590e-01
-1.22496867e+00 -1.10137641e+00 3.23133051e-01 -2.90020794e-01
9.20236051e-01 -3.80115062e-01 -1.22142529e+00 4.72128570e-01
-2.78382242e-01 9.26092044e-02 7.02106297e-01 1.93372220e-01
-2.07628667e-01 -4.07007486e-01 -1.23817003e+00 5.33966362e-01
3.66681486e-01 -2.72915751e-01 -1.86864048e-01 3.34237963e-01
4.22030658e-01 -2.85387129e-01 -1.24893165e+00 6.60977960e-01
3.31949949e-01 -7.55921423e-01 8.96404028e-01 -7.34486461e-01
3.80110681e-01 -1.24179438e-01 -9.02673975e-02 -1.48693323e+00
1.42555073e-01 -8.28224719e-01 -2.11744919e-01 1.30557299e+00
4.32784200e-01 -7.69182980e-01 7.23176479e-01 8.91945064e-01
9.47126970e-02 -8.44892919e-01 -9.41506028e-01 -1.03084421e+00
4.45360959e-01 -5.45820951e-01 5.33873379e-01 9.66799915e-01
-5.62904239e-01 -6.05903789e-02 -3.27387005e-01 2.21582785e-01
8.96988809e-01 1.70203507e-01 8.16555381e-01 -1.48002267e+00
-5.45183778e-01 -6.39450967e-01 -2.10526198e-01 -6.62406027e-01
1.23037003e-01 -3.12876582e-01 1.16602123e-01 -9.38673079e-01
-1.62473157e-01 -6.60107255e-01 -4.42325413e-01 4.13660742e-02
-2.25853041e-01 -1.28573284e-01 -2.67639309e-01 -3.72254476e-02
-1.48070395e-01 7.74156928e-01 8.58523428e-01 8.18731040e-02
-2.73128659e-01 4.67567593e-01 -5.46622396e-01 7.31225550e-01
8.95354629e-01 -5.26404500e-01 -2.84942865e-01 -1.80082526e-02
3.71956199e-01 2.04802603e-01 4.01163459e-01 -1.24009311e+00
2.02112541e-01 -3.52539748e-01 5.79203069e-01 -3.89503211e-01
3.41146141e-02 -6.79365039e-01 3.83346289e-01 1.74041167e-01
-4.68939811e-01 2.53551275e-01 4.36970860e-01 7.36415565e-01
-1.82375044e-01 -2.02425703e-01 1.05142605e+00 1.52507037e-01
-3.24852243e-02 2.93622285e-01 -1.58453062e-01 4.20198977e-01
7.79450893e-01 2.42145315e-01 -1.93800882e-01 -6.73150718e-01
-4.31619048e-01 3.12180698e-01 4.44241852e-01 8.68841186e-02
9.54474434e-02 -1.06050920e+00 -7.70671785e-01 1.50853559e-01
-2.00300887e-02 2.21006885e-01 -3.93872568e-03 9.25333679e-01
-3.93920958e-01 2.13078111e-01 1.60069898e-01 -5.60117781e-01
-6.45635903e-01 5.39551795e-01 5.85421324e-01 -4.83274847e-01
-6.25207365e-01 5.33415139e-01 2.03694582e-01 -4.12919700e-01
5.06240845e-01 -4.07553911e-01 3.39625329e-01 -2.14894742e-01
3.62029135e-01 5.27601600e-01 3.46682876e-01 1.02364108e-01
2.15676464e-02 3.14384192e-01 1.55622989e-01 1.64717302e-01
1.33891857e+00 9.57628861e-02 2.97607243e-01 7.92173862e-01
1.14823472e+00 -1.11306332e-01 -1.66751158e+00 -3.13355207e-01
2.62408167e-01 -3.00893962e-01 1.63479909e-01 -7.36219347e-01
-1.03169048e+00 9.10728633e-01 4.53483731e-01 2.71344125e-01
9.16745901e-01 -4.51191664e-01 5.05632520e-01 5.29902875e-01
1.12659417e-01 -1.17520726e+00 -5.87792508e-02 3.09342891e-01
6.70466661e-01 -1.28674614e+00 1.24806695e-01 1.78809464e-01
-3.35811108e-01 9.84645009e-01 4.24947441e-01 -1.82343945e-01
7.09964931e-01 5.15561819e-01 3.92086022e-02 1.46376774e-01
-7.80456841e-01 3.10195118e-01 2.05992386e-01 4.84724104e-01
2.57815540e-01 -1.19867109e-01 -4.63349223e-02 5.78579545e-01
-1.93601504e-01 9.79653522e-02 4.86615330e-01 7.59300530e-01
-5.96053079e-02 -8.48622859e-01 -5.02573609e-01 7.99431324e-01
-6.68489814e-01 8.59602541e-02 2.68450588e-01 1.11071289e+00
-2.98498660e-01 7.98894882e-01 2.39909172e-01 2.03643948e-01
3.34682167e-01 1.48688601e-02 2.93923430e-02 -1.60956696e-01
-4.37041938e-01 -4.85810228e-02 -5.18241636e-02 -3.09061050e-01
1.92192830e-02 -6.73093796e-01 -8.82998884e-01 -7.00240076e-01
-1.19857386e-01 1.45812213e-01 7.06689835e-01 1.06647527e+00
3.79960388e-01 5.63110948e-01 6.75162613e-01 -6.77705050e-01
-1.33401477e+00 -9.04365897e-01 -7.28873312e-01 4.89141047e-01
4.98097152e-01 -9.18031275e-01 -7.35701919e-01 -3.22309166e-01]
|
[7.136688709259033, 3.899250030517578]
|
44f74b93-e3b5-46cb-a0e6-e819ae1cd3b5
|
evaluating-semantic-models-with-word-sentence
|
1603.07253
| null |
http://arxiv.org/abs/1603.07253v2
|
http://arxiv.org/pdf/1603.07253v2.pdf
|
Evaluating semantic models with word-sentence relatedness
|
Semantic textual similarity (STS) systems are designed to encode and evaluate
the semantic similarity between words, phrases, sentences, and documents. One
method for assessing the quality or authenticity of semantic information
encoded in these systems is by comparison with human judgments. A data set for
evaluating semantic models was developed consisting of 775 English
word-sentence pairs, each annotated for semantic relatedness by human raters
engaged in a Maximum Difference Scaling (MDS) task, as well as a faster
alternative task. As a sample application of this relatedness data,
behavior-based relatedness was compared to the relatedness computed via four
off-the-shelf STS models: n-gram, Latent Semantic Analysis (LSA), Word2Vec, and
UMBC Ebiquity. Some STS models captured much of the variance in the human
judgments collected, but they were not sensitive to the implicatures and
entailments that were processed and considered by the participants. All text
stimuli and judgment data have been made freely available.
|
['Kimberly Glasgow', 'Mark Chevillet', 'Matthew Roos', 'Amy Haufler', 'Michael Wolmetz']
|
2016-03-23
| null | null | null | null |
['implicatures']
|
['natural-language-processing']
|
[ 2.11690560e-01 4.99646366e-02 1.38084158e-01 -6.75796688e-01
-5.85724175e-01 -7.14342058e-01 8.00962210e-01 1.03969908e+00
-8.30404699e-01 1.64573789e-01 1.07193458e+00 -1.35826483e-01
-3.43400747e-01 -4.60067928e-01 7.11764321e-02 -1.49297729e-01
4.57167566e-01 5.93007088e-01 9.35327560e-02 -4.60981935e-01
9.13817823e-01 -4.81039569e-05 -1.54153693e+00 6.43243253e-01
8.42297971e-01 9.07008827e-01 4.27064657e-01 2.87860870e-01
-3.76272172e-01 6.30509257e-01 -5.81835687e-01 -6.07696593e-01
-1.73956305e-01 -4.16591018e-01 -1.12870812e+00 -2.96986759e-01
2.89654791e-01 1.46949574e-01 1.74821056e-02 9.11329508e-01
6.12765431e-01 4.76481974e-01 8.59293997e-01 -1.00419402e+00
-1.17522871e+00 7.29849935e-01 2.48784810e-01 5.45418739e-01
9.55510020e-01 -3.27531338e-01 1.26867557e+00 -1.09046292e+00
6.90865636e-01 1.74143147e+00 5.81236303e-01 2.17705011e-01
-1.04768705e+00 -4.47955519e-01 -4.62192804e-01 5.59341371e-01
-1.12708247e+00 -4.84362274e-01 3.55226070e-01 -5.67864239e-01
1.40952623e+00 2.77909636e-01 3.33581328e-01 1.39333093e+00
1.15709655e-01 2.98572183e-01 1.05836749e+00 -6.23947203e-01
4.76125419e-01 4.38983887e-01 5.56880116e-01 3.66649241e-04
1.00322224e-01 -2.05694258e-01 -6.03175163e-01 -4.36766803e-01
5.09990826e-02 -3.03583562e-01 -1.34664550e-01 1.42913277e-03
-1.34971786e+00 1.01455128e+00 3.71890247e-01 6.87923372e-01
-2.97241926e-01 -3.78701806e-01 7.82291889e-01 4.79558587e-01
6.60752594e-01 8.61810446e-01 -4.45857882e-01 -2.69780248e-01
-4.85482454e-01 2.61921674e-01 7.87754536e-01 5.82678616e-01
4.81953561e-01 -2.54303485e-01 -3.54653478e-01 1.15233898e+00
4.06915843e-01 4.56765771e-01 1.36836624e+00 -9.47072804e-01
4.34621781e-01 5.36149859e-01 9.36722383e-02 -1.61149490e+00
-3.37983578e-01 9.29693952e-02 -5.24470508e-02 -5.66683829e-01
-7.48984814e-02 3.49764675e-01 -4.34869707e-01 1.61811101e+00
8.99095833e-02 -2.89927036e-01 1.75548241e-01 8.93008173e-01
9.82465863e-01 4.33211207e-01 5.90051055e-01 -1.23488195e-01
1.63207233e+00 -6.17403507e-01 -8.02180588e-01 -6.07523322e-01
1.12767959e+00 -1.17727804e+00 1.36881506e+00 -1.81380108e-01
-8.89213800e-01 -7.98872530e-01 -9.46569026e-01 -3.22253972e-01
-6.54372871e-01 -3.96522492e-01 3.07550788e-01 5.60290575e-01
-1.11305225e+00 6.68527305e-01 -4.62586321e-02 -9.62637842e-01
-2.50320081e-02 -2.63083071e-01 -4.79466289e-01 5.47050238e-02
-1.64136100e+00 1.59626234e+00 5.60510337e-01 -4.30560768e-01
-2.04421654e-01 -4.36404794e-01 -1.18194044e+00 1.05696581e-01
1.93666723e-02 -5.90746343e-01 1.19233239e+00 -1.02321470e+00
-8.96160722e-01 1.26457691e+00 -3.21658254e-01 -1.24413460e-01
-2.45910302e-01 -6.95606470e-02 -8.33018422e-01 1.45600080e-01
6.40837789e-01 5.07044971e-01 5.05784571e-01 -8.45410764e-01
-1.20618656e-01 -7.03613102e-01 -3.50641161e-01 4.23472345e-01
-5.62503278e-01 7.18545794e-01 3.59101176e-01 -1.00446761e+00
2.30592906e-01 -6.70778215e-01 1.54301301e-01 -2.42987230e-01
5.69419004e-02 -5.36644995e-01 3.35033596e-01 -9.43761349e-01
1.27288294e+00 -2.28283238e+00 -4.20019440e-02 1.58993959e-01
-8.22430402e-02 1.67940646e-01 -4.90442753e-01 8.91172469e-01
-1.48845151e-01 3.33255351e-01 -2.60260869e-02 -3.98327351e-01
3.37851167e-01 -4.35046945e-03 -1.80819288e-01 1.22495264e-01
-3.12651306e-01 1.02619779e+00 -1.09934366e+00 -5.73993504e-01
1.06939115e-01 2.31360540e-01 -3.22965264e-01 -4.83442061e-02
1.49578765e-01 -1.31467611e-01 -1.56073004e-01 1.78704083e-01
5.85127017e-03 -3.77116092e-02 1.76215932e-01 -3.74091804e-01
2.42251754e-01 8.54318798e-01 -6.66700244e-01 1.80024493e+00
-4.65581715e-01 7.95827568e-01 -6.14052892e-01 -7.52707899e-01
1.04958296e+00 4.18399453e-01 3.85469012e-02 -1.13848078e+00
3.79657775e-01 1.25497907e-01 -2.09025033e-02 -7.79684901e-01
8.92862976e-01 -3.66660595e-01 -3.19109976e-01 8.08726013e-01
-3.37037258e-02 -2.69288659e-01 2.35122293e-01 3.85810763e-01
8.84924829e-01 -3.98492604e-01 6.14612758e-01 -5.75037599e-01
3.80137086e-01 1.18303828e-01 1.27010539e-01 3.48929048e-01
-1.62528411e-01 3.12062770e-01 1.90205529e-01 -1.96993247e-01
-1.15976906e+00 -1.07191360e+00 -2.14945748e-01 1.09587741e+00
2.57846177e-01 -7.91071653e-01 -7.08937526e-01 -1.59419402e-01
-6.96420595e-02 1.48953986e+00 -6.24392629e-01 -6.48995042e-01
1.22932754e-01 -1.86476544e-01 4.50588256e-01 6.77509606e-01
3.10794618e-02 -1.21494758e+00 -5.68975031e-01 2.11943388e-01
-7.44040430e-01 -1.26912868e+00 -6.63028240e-01 -2.62029946e-01
-6.16662562e-01 -8.16864789e-01 -6.07822876e-05 -6.89656258e-01
2.14029789e-01 3.80436033e-01 1.22325110e+00 -1.51298240e-01
-2.46849522e-01 4.01264369e-01 -7.97376990e-01 -2.19423965e-01
-5.68889141e-01 -4.75374013e-01 3.01971704e-01 -3.86254936e-01
1.06694341e+00 -2.14737386e-01 -3.59424621e-01 3.53287518e-01
-9.68508959e-01 -2.76526362e-01 9.53566208e-02 5.77962637e-01
5.16760312e-02 -5.89297898e-02 5.44051886e-01 -2.95851260e-01
1.47014117e+00 -7.66004682e-01 3.60362619e-01 3.75663072e-01
-8.19655120e-01 -7.97356106e-03 1.69918776e-01 -2.63591260e-01
-8.93896103e-01 -7.35265434e-01 1.99411601e-01 7.68270204e-03
-1.98423907e-01 6.36956275e-01 1.36761114e-01 4.56883043e-01
1.03752005e+00 -2.07728595e-01 1.68261454e-01 -3.82928938e-01
5.49862087e-01 1.12786758e+00 4.15792286e-01 -4.81557697e-01
2.03269884e-01 6.09324686e-02 -6.57148182e-01 -6.77476645e-01
-9.55802739e-01 -8.18961740e-01 -5.52594721e-01 -7.68415928e-02
1.06976473e+00 -6.88695431e-01 -4.96570706e-01 9.14546475e-02
-1.10080194e+00 3.14576551e-02 -3.54562163e-01 6.73814774e-01
-4.10470724e-01 6.46086156e-01 -5.84206522e-01 -4.25804615e-01
-4.47817296e-01 -7.80703485e-01 8.39220583e-01 -2.87797332e-01
-1.41029668e+00 -1.21973324e+00 9.63495076e-02 7.21336901e-01
5.97145677e-01 -1.54311433e-01 1.33783281e+00 -1.25599694e+00
6.47046745e-01 -3.01834255e-01 -3.06259334e-01 3.00084889e-01
3.46707880e-01 -2.75391370e-01 -8.13738883e-01 -1.41921982e-01
4.97475296e-01 -7.40805447e-01 5.30180097e-01 1.20092951e-01
6.63731754e-01 -3.88354480e-01 -7.28011727e-02 -1.54650718e-01
1.21021605e+00 1.88689992e-01 4.34533656e-01 2.88176835e-01
3.99235666e-01 1.09793162e+00 6.58489585e-01 4.08109903e-01
5.34124136e-01 7.77194858e-01 -2.88002372e-01 5.26769400e-01
6.30046502e-02 -3.51925820e-01 3.53106350e-01 1.07122338e+00
5.52460194e-01 -2.84546763e-01 -1.12307143e+00 4.46925044e-01
-1.58459377e+00 -1.04903686e+00 -1.45568267e-01 2.02257133e+00
6.09188795e-01 -5.70200235e-02 -1.32690161e-01 4.74940032e-01
7.81784832e-01 1.59250796e-01 -1.16506509e-01 -8.79892230e-01
-2.39748612e-01 4.22057733e-02 -4.95554805e-02 5.13556123e-01
-4.51718390e-01 9.44477141e-01 6.91072321e+00 9.19611454e-01
-5.67793310e-01 1.69280302e-02 4.00300384e-01 1.69371575e-01
-8.25437129e-01 1.37425428e-02 -1.09757192e-01 5.51861286e-01
1.24234307e+00 -5.45597672e-01 2.10032582e-01 6.49071455e-01
3.03112268e-01 -3.72551084e-01 -1.17600513e+00 9.30525959e-01
6.12174690e-01 -8.59822750e-01 2.13917509e-01 -4.85126406e-01
2.68099576e-01 -1.72455028e-01 -1.04179747e-01 2.30046451e-01
7.26293772e-02 -9.29855287e-01 7.90161431e-01 3.03131133e-01
6.19281828e-01 -3.64737064e-01 8.44456196e-01 2.46489778e-01
-7.77630806e-01 -2.35243104e-02 -4.35027391e-01 -4.24997658e-01
2.22825304e-01 1.89638287e-01 -6.85604215e-01 1.31675795e-01
7.68552840e-01 8.19877863e-01 -9.25854027e-01 3.48880887e-01
-1.66856483e-01 2.13765070e-01 8.08164254e-02 -4.14118230e-01
8.25224221e-02 -8.97998810e-02 3.56596142e-01 1.22257531e+00
2.21295863e-01 7.95931891e-02 -2.93941677e-01 9.33706105e-01
4.22011971e-01 5.36199272e-01 -5.07682741e-01 -5.54253101e-01
1.06988001e+00 1.03880370e+00 -8.78753066e-01 -3.85460198e-01
-4.12838399e-01 1.11608529e+00 3.52980167e-01 2.21985579e-01
-4.08463150e-01 -3.29576671e-01 5.68863928e-01 -2.35077105e-02
-2.38799080e-01 -2.42918427e-03 -5.11579156e-01 -9.15689886e-01
-4.54034917e-02 -7.40441382e-01 5.07442534e-01 -1.45020378e+00
-1.98997664e+00 5.75370789e-01 2.03544378e-01 -8.06187153e-01
-2.13951811e-01 -4.62269872e-01 -2.89506674e-01 9.78052914e-01
-7.25331604e-01 -4.74993855e-01 -4.62305933e-01 4.73989159e-01
5.41842461e-01 -1.73509121e-01 1.12786853e+00 4.00687233e-02
-4.98218164e-02 2.37664178e-01 -5.93033805e-02 -1.60237059e-01
1.07800758e+00 -9.15516078e-01 6.85192168e-01 2.45058179e-01
8.92716721e-02 9.06518519e-01 8.04239750e-01 -8.94162118e-01
-5.44616878e-01 -4.42270309e-01 1.72109294e+00 -6.71081066e-01
1.07013261e+00 -5.75677305e-02 -1.00022161e+00 4.40254062e-01
3.43362212e-01 -7.38742352e-01 1.25731623e+00 8.20338875e-02
-6.56814575e-01 3.57796639e-01 -1.12505484e+00 6.16590440e-01
1.09564161e+00 -1.22442031e+00 -1.52182937e+00 2.50167727e-01
1.02601159e+00 1.40106246e-01 -8.63722205e-01 2.21071303e-01
3.51408064e-01 -7.16254830e-01 1.15209293e+00 -8.86839986e-01
6.93896949e-01 1.26431480e-01 -7.47584164e-01 -1.59405446e+00
-6.65210307e-01 5.69931865e-02 5.41926622e-01 1.19858408e+00
2.42484450e-01 -5.02686679e-01 4.41948101e-02 1.06206536e+00
-1.52057514e-01 -2.28103220e-01 -6.43107533e-01 -7.33188331e-01
1.63597856e-02 -5.59345305e-01 5.44534266e-01 1.31870151e+00
6.53424919e-01 6.82070971e-01 3.21339875e-01 -3.54143083e-01
3.04496914e-01 -2.06025794e-01 -4.88765314e-02 -1.11939454e+00
1.45334259e-01 -5.64338982e-01 -7.34811366e-01 -2.56882906e-01
5.77745914e-01 -1.29315805e+00 -2.44057938e-01 -1.68647969e+00
2.92850852e-01 -1.28785282e-01 -1.40437216e-01 1.62738949e-01
-4.64872152e-01 4.98925857e-02 2.16170877e-01 2.94726938e-01
-5.14032245e-01 6.44747078e-01 7.93884516e-01 1.81660742e-01
1.31078318e-01 -6.10343575e-01 -8.36934865e-01 6.96318626e-01
9.05130386e-01 -4.85517561e-01 -6.87267721e-01 -4.95803148e-01
4.48304623e-01 -9.69123933e-03 4.70635772e-01 -7.81331539e-01
2.11008653e-01 -6.02025613e-02 1.52706921e-01 -1.58306256e-01
3.47382754e-01 -6.61257029e-01 1.61684409e-01 2.89137542e-01
-9.95563149e-01 5.81274569e-01 1.27798691e-01 3.58151644e-01
-4.28697228e-01 -5.51897347e-01 4.19501096e-01 -1.14621848e-01
-9.28473353e-01 -3.31386834e-01 -5.80062330e-01 5.92629254e-01
8.79698932e-01 -4.41404402e-01 -4.37763721e-01 -5.47247350e-01
-6.51363432e-01 -1.03314273e-01 4.34214592e-01 1.00907910e+00
7.08599806e-01 -1.61547375e+00 -7.45452166e-01 -1.33613735e-01
6.75345063e-01 -9.75331843e-01 2.49384135e-01 5.04838645e-01
-2.04730183e-01 6.23312414e-01 -3.06712598e-01 -9.84518901e-02
-1.19382513e+00 8.69010091e-01 -1.07353248e-01 3.84827614e-01
-1.71084553e-01 6.91577435e-01 -1.37624264e-01 -4.33365524e-01
-8.01476464e-02 -3.95333245e-02 -5.07716775e-01 5.30755699e-01
5.33589005e-01 5.85036635e-01 2.82774977e-02 -1.05225754e+00
-3.39391351e-01 5.18086374e-01 1.55612394e-01 -4.24027234e-01
8.68335128e-01 -4.99453723e-01 -3.76163900e-01 9.49312866e-01
1.47943425e+00 -4.70195353e-01 -1.76742613e-01 -5.52591085e-01
4.78337795e-01 -5.82208037e-01 -7.66591821e-03 -6.74171090e-01
-2.27443308e-01 6.62966669e-01 2.41246775e-01 1.60061359e-01
6.28799498e-01 1.17893390e-01 7.95352280e-01 3.52256000e-01
2.11136475e-01 -1.52958083e+00 3.68293285e-01 5.05611360e-01
9.61285532e-01 -1.26260316e+00 -1.91522583e-01 -3.31880897e-01
-9.97155786e-01 8.65631580e-01 4.63933200e-01 2.77626544e-01
5.89230299e-01 -1.75523043e-01 8.60545412e-02 -3.05596560e-01
-7.60406196e-01 1.73821300e-02 5.58640361e-01 4.70902026e-01
7.57361293e-01 4.68771756e-02 -8.05110097e-01 8.12018037e-01
-5.41132629e-01 -3.83384883e-01 9.00849625e-02 6.75126374e-01
-5.23962080e-01 -7.93384194e-01 -1.70264125e-01 4.14258629e-01
-1.13738783e-01 -4.88617182e-01 -7.81706691e-01 2.51713753e-01
-2.57967353e-01 1.38082421e+00 2.87267774e-01 -4.28971291e-01
2.37420887e-01 2.77933478e-01 2.46674225e-01 -6.70344353e-01
-7.50414014e-01 -5.70303857e-01 4.92451459e-01 -8.05182278e-01
-3.96054417e-01 -6.73936009e-01 -1.24715626e+00 -3.31790119e-01
-1.97386041e-01 3.00273508e-01 7.73535669e-01 1.29981422e+00
2.90626407e-01 1.20208092e-01 3.29280198e-01 -3.39462578e-01
-5.47624052e-01 -1.27479780e+00 -5.09455144e-01 1.20483148e+00
-4.24250662e-01 -4.74050015e-01 -6.99606359e-01 -1.24803726e-02]
|
[10.613570213317871, 9.063237190246582]
|
88f10a62-3c92-42e1-a55e-f56b9c8865d0
|
a-universality-individuality-integration
|
2204.06185
| null |
https://arxiv.org/abs/2204.06185v1
|
https://arxiv.org/pdf/2204.06185v1.pdf
|
A Universality-Individuality Integration Model for Dialog Act Classification
|
Dialog Act (DA) reveals the general intent of the speaker utterance in a conversation. Accurately predicting DAs can greatly facilitate the development of dialog agents. Although researchers have done extensive research on dialog act classification, the feature information of classification has not been fully considered. This paper suggests that word cues, part-of-speech cues and statistical cues can complement each other to improve the basis for recognition. In addition, the different types of the three lead to the diversity of their distribution forms, which hinders the mining of feature information. To solve this problem, we propose a novel model based on universality and individuality strategies, called Universality-Individuality Integration Model (UIIM). UIIM not only deepens the connection between the clues by learning universality, but also utilizes the learning of individuality to capture the characteristics of the clues themselves. Experiments were made over two most popular benchmark data sets SwDA and MRDA for dialogue act classification, and the results show that extracting the universalities and individualities between cues can more fully excavate the hidden information in the utterance, and improve the accuracy of automatic dialogue act recognition.
|
['Ma Yinglong', 'Gao Pengfei']
|
2022-04-13
| null | null | null | null |
['dialog-act-classification', 'dialogue-act-classification']
|
['natural-language-processing', 'natural-language-processing']
|
[-3.36702108e-01 -1.57043740e-01 -3.91816527e-01 -5.48300266e-01
8.51643011e-02 -4.26716983e-01 8.08400095e-01 7.87975267e-02
-9.58359614e-02 4.55619693e-01 9.16518748e-01 -6.40950054e-02
1.12667114e-01 -3.99634451e-01 4.78521824e-01 -7.22511828e-01
3.39244664e-01 2.95083135e-01 4.47451413e-01 -8.24884772e-01
3.32861185e-01 8.25501010e-02 -1.10838556e+00 4.82755840e-01
9.76812005e-01 8.79727423e-01 1.63939089e-01 2.32712477e-01
-8.84909630e-01 9.84765053e-01 -7.66840756e-01 -1.94683179e-01
-9.59055573e-02 -8.49253893e-01 -7.36459851e-01 3.01755726e-01
-6.05372131e-01 -5.68572462e-01 -3.40575159e-01 7.57485151e-01
4.11608636e-01 7.26078451e-02 7.98770428e-01 -1.04061723e+00
-4.28691119e-01 8.11759412e-01 -3.48941207e-01 3.46064329e-01
6.30388498e-01 1.13766961e-01 1.31565845e+00 -6.92160249e-01
1.17429197e-01 1.61991251e+00 3.10095489e-01 7.17650294e-01
-6.29189610e-01 -5.86168587e-01 2.03648955e-01 2.27219895e-01
-1.03549838e+00 -2.94982970e-01 1.11624241e+00 -3.85445446e-01
6.43720686e-01 2.57216454e-01 4.86330569e-01 1.07151425e+00
-1.95575543e-02 1.44770491e+00 1.24343646e+00 -4.97716099e-01
-2.01105908e-01 4.82846916e-01 9.04946446e-01 6.90657794e-01
-3.41695756e-01 -2.91954547e-01 -5.34585059e-01 -2.14598745e-01
6.35167778e-01 2.52177566e-01 -4.18638319e-01 3.29810232e-01
-8.83462608e-01 9.33988988e-01 1.99385047e-01 6.86232865e-01
5.74140549e-02 -8.69594693e-01 5.66895485e-01 1.75661996e-01
2.28830069e-01 3.42545927e-01 -4.48793888e-01 -3.80992889e-01
2.57365666e-02 7.00213984e-02 1.07116902e+00 5.93154490e-01
1.09334385e+00 -5.87032996e-02 -3.70061398e-01 1.08683276e+00
4.23906356e-01 1.08671822e-01 1.20351970e+00 -3.55100185e-01
4.35323447e-01 1.41381407e+00 -2.33304426e-01 -1.01998317e+00
-5.61076105e-01 2.42916513e-02 -6.50898457e-01 -4.24571365e-01
2.87682831e-01 -4.65775460e-01 -3.91043067e-01 1.56634998e+00
3.16317886e-01 -1.30333543e-01 1.33549139e-01 7.92673051e-01
1.22382057e+00 8.28442812e-01 -4.14873213e-02 -3.77213120e-01
1.44936621e+00 -8.07095647e-01 -1.14805818e+00 -2.05522045e-01
5.86099982e-01 -7.31337428e-01 1.27563846e+00 1.72622919e-01
-3.37721705e-01 -5.81029058e-01 -1.14624798e+00 -4.41147909e-02
-3.24095786e-01 1.86333105e-01 8.28502476e-01 6.19928777e-01
-1.54566377e-01 5.33175133e-02 -3.28369349e-01 -1.31226316e-01
-2.90525164e-02 1.77812442e-01 -9.46540907e-02 1.26086518e-01
-1.54670203e+00 8.68310869e-01 6.06907368e-01 8.99752378e-02
-2.28717804e-01 -1.85664624e-01 -7.17255175e-01 5.54931648e-02
5.97809494e-01 2.47726776e-02 1.20496976e+00 -8.40089083e-01
-1.75446510e+00 3.07808846e-01 -1.51558056e-01 -9.21851099e-02
1.20399455e-02 -8.14582407e-03 -6.12394392e-01 -2.37078652e-01
-1.40232950e-01 2.68564135e-01 5.43531954e-01 -1.05297542e+00
-7.58235753e-01 -5.41681468e-01 1.55950278e-01 4.04716194e-01
-6.50512934e-01 2.69128233e-01 -3.16534281e-01 -2.90670186e-01
2.91824162e-01 -6.32557392e-01 1.60813749e-01 -6.42813981e-01
-3.07168663e-01 -9.90098894e-01 9.84799325e-01 -6.86667085e-01
1.61247015e+00 -2.23958349e+00 1.51597545e-01 -9.79999453e-03
3.75188500e-01 2.98436105e-01 3.58126283e-01 7.18153059e-01
4.40051764e-01 -2.59021670e-01 4.09160666e-02 -1.42585054e-01
9.13281366e-02 8.21335793e-01 -4.20281172e-01 -8.80815610e-02
2.01645074e-03 5.55915713e-01 -6.60108268e-01 -6.77361012e-01
1.72911957e-01 -2.01503590e-01 -3.42124999e-01 6.57167077e-01
-2.74024159e-01 5.22105634e-01 -8.91027868e-01 4.16848689e-01
2.80061394e-01 5.67523353e-02 2.21674889e-01 -2.18456566e-01
-1.18185189e-02 6.07036531e-01 -9.98144865e-01 1.18799067e+00
-2.76231319e-01 4.96693939e-01 -1.72317401e-02 -7.72695780e-01
1.27091789e+00 4.14688647e-01 2.88790584e-01 -4.35470849e-01
3.39960098e-01 2.64384300e-01 5.88042438e-01 -7.28942633e-01
3.57873112e-01 -3.12975019e-01 -2.26966769e-01 5.04575610e-01
5.47582880e-02 1.91475078e-01 5.05111068e-02 1.90081134e-01
6.70885444e-01 -3.80427599e-01 6.44374788e-01 -1.31882861e-01
9.58464682e-01 -2.68059611e-01 7.73300827e-01 1.54078797e-01
-2.54122734e-01 1.21185303e-01 6.24251366e-01 -4.22691077e-01
-3.42327327e-01 -5.56779265e-01 -2.33386025e-01 1.21396828e+00
4.07595068e-01 -6.16820514e-01 -4.74219203e-01 -9.86304164e-01
-2.44298697e-01 6.31033123e-01 -4.67220426e-01 -2.19356135e-01
-6.24121666e-01 -7.36839950e-01 5.91379821e-01 5.04476309e-01
9.80615199e-01 -1.11648953e+00 -8.45807642e-02 1.07227243e-01
-3.77442211e-01 -1.02524936e+00 -4.76835549e-01 1.09899543e-01
-5.58900356e-01 -1.06335962e+00 -1.99302986e-01 -5.97265661e-01
3.57169658e-01 4.60629523e-01 6.23740196e-01 2.30373487e-01
8.32886621e-02 1.95238009e-01 -6.41207933e-01 -3.95370156e-01
-4.86080229e-01 1.93294976e-02 1.78379759e-01 2.81997949e-01
9.80898023e-01 -4.25122231e-01 -1.61478594e-01 5.85425675e-01
-7.00499833e-01 -1.32377163e-01 4.83309567e-01 9.81536627e-01
-3.13644886e-01 1.40671730e-01 5.88991404e-01 -8.39639962e-01
1.36262500e+00 -5.50098658e-01 5.91840819e-02 2.41903305e-01
-6.14835620e-01 3.17012459e-01 6.46563232e-01 -4.24907029e-01
-1.42727399e+00 -3.43466371e-01 -1.92348510e-01 5.23150191e-02
-3.18469793e-01 7.14849055e-01 -6.48062587e-01 4.58110303e-01
3.52150172e-01 6.19546711e-01 2.86903709e-01 -5.72088003e-01
7.52964616e-02 1.12004387e+00 1.46521598e-01 -6.83640003e-01
2.02867076e-01 -6.09826595e-02 -4.11149204e-01 -1.14378679e+00
-7.52713919e-01 -8.35207283e-01 -6.11997247e-01 -2.22400159e-01
8.90698195e-01 -5.38882792e-01 -8.21731627e-01 5.06944776e-01
-1.35017693e+00 2.23907858e-01 1.76115558e-01 4.45354670e-01
-7.33859167e-02 7.97883809e-01 -7.34216452e-01 -1.04259336e+00
-6.02748096e-02 -1.19519126e+00 4.71697032e-01 5.69772005e-01
-4.03868437e-01 -1.06619871e+00 5.98211512e-02 5.07171750e-01
7.53225759e-02 -4.08872813e-01 1.06839466e+00 -1.38592875e+00
-3.22160542e-01 -2.13741034e-01 -8.63476656e-03 4.65872169e-01
6.43762827e-01 -7.22842813e-02 -9.85880196e-01 3.17049384e-01
6.41961336e-01 -4.30934310e-01 6.45070910e-01 -9.17400420e-02
5.74618280e-01 -6.16439879e-01 -1.41787708e-01 -9.09241438e-02
6.39824092e-01 8.03125203e-01 4.69270259e-01 4.03697193e-02
5.18384874e-01 9.66914237e-01 6.83088541e-01 5.90997279e-01
5.65179169e-01 8.49066496e-01 1.13831408e-01 1.60001725e-01
1.33220091e-01 -2.15174705e-01 6.06464326e-01 1.29928744e+00
-8.40521827e-02 5.23444638e-02 -8.61541748e-01 2.02011794e-01
-1.93465459e+00 -9.29675877e-01 -1.36144504e-01 1.53536749e+00
1.16768014e+00 1.32354006e-01 3.31726521e-01 7.59072304e-02
6.08136773e-01 3.67079884e-01 -2.53100872e-01 -2.14576617e-01
-2.42250577e-01 -4.69750196e-01 -2.53152966e-01 5.43556929e-01
-8.43300462e-01 9.41796601e-01 5.02487516e+00 9.85012054e-01
-9.74192917e-01 -3.18590328e-02 5.66838741e-01 7.00647950e-01
-1.86212316e-01 -1.45684600e-01 -1.29050589e+00 5.55034995e-01
4.58964109e-01 -2.27009073e-01 1.75939620e-01 9.20572340e-01
-7.19691487e-03 1.17538935e-02 -1.01913989e+00 8.45533550e-01
2.16172338e-01 -8.77327740e-01 1.02486365e-01 1.13819897e-01
2.57429928e-01 -3.98421049e-01 -2.08867759e-01 5.65671444e-01
4.33576494e-01 -8.12566936e-01 -7.35526485e-03 3.32908303e-01
-1.50752962e-01 -5.02013087e-01 1.19555187e+00 8.34054828e-01
-1.15305507e+00 -9.74479467e-02 -4.93285060e-01 -2.72380888e-01
-2.43872106e-01 1.28589064e-01 -1.36873949e+00 7.36744821e-01
1.45994216e-01 7.80038536e-01 -4.07143563e-01 4.95600373e-01
-2.89871693e-01 7.19021440e-01 -3.06815088e-01 -5.81353962e-01
4.45919216e-01 -4.27016288e-01 4.29897428e-01 1.25742781e+00
-1.34735480e-01 5.67614913e-01 6.15015686e-01 5.52462399e-01
3.11855525e-01 3.98472637e-01 -3.96716416e-01 -1.20568469e-01
4.86716032e-01 1.03303003e+00 -4.49901342e-01 -3.09636220e-02
-5.17208040e-01 6.23983145e-01 1.52736023e-01 -2.57681627e-02
-4.10822093e-01 -2.10259318e-01 5.72010458e-01 -2.12333664e-01
-4.75579463e-02 -3.14219445e-01 -2.83345610e-01 -1.14664769e+00
-2.66974941e-02 -1.04005241e+00 5.96912563e-01 -2.78739154e-01
-1.53120744e+00 7.06830263e-01 2.73877662e-02 -1.21052468e+00
-5.01677394e-01 -8.24121892e-01 -1.11523700e+00 9.29976642e-01
-1.08817971e+00 -1.05756855e+00 -3.23975950e-01 7.24307954e-01
1.15816844e+00 -7.45336354e-01 9.00671601e-01 -8.24637264e-02
-9.00029182e-01 4.02026862e-01 -1.65958449e-01 5.96350372e-01
7.17469454e-01 -1.05920959e+00 -3.52963030e-01 4.79181319e-01
1.10867701e-01 9.88818228e-01 5.88347852e-01 -4.68537450e-01
-1.12083614e+00 -2.63202548e-01 6.50759995e-01 -4.14444268e-01
7.76038051e-01 -1.29821956e-01 -1.18698871e+00 4.57007229e-01
4.22599196e-01 -7.23759711e-01 1.06953251e+00 6.50866389e-01
-4.06683475e-01 -1.12084754e-01 -5.59428751e-01 7.04155743e-01
4.34870273e-01 -5.64879894e-01 -1.10769176e+00 1.56075224e-01
7.58876860e-01 -6.44837096e-02 -5.04454255e-01 2.46721327e-01
4.11601096e-01 -1.12870717e+00 7.18119323e-01 -6.85115993e-01
4.30786997e-01 -9.17985961e-02 -2.41686031e-01 -1.18642628e+00
-6.33572116e-02 -4.33778942e-01 -1.89852323e-02 1.70264316e+00
3.31781328e-01 -6.45097136e-01 4.99620825e-01 7.39767790e-01
-3.57024521e-01 -8.33324254e-01 -8.84347200e-01 -4.87650275e-01
-1.41709834e-01 -1.61583945e-01 7.97199845e-01 1.11253667e+00
7.99273193e-01 1.03987265e+00 -4.93657559e-01 -2.14267671e-01
-1.91695035e-01 8.18318650e-02 8.99814725e-01 -1.30229235e+00
-2.93852895e-01 -6.70051813e-01 -3.71893287e-01 -1.82180500e+00
3.18544298e-01 -5.61541617e-01 -3.90393995e-02 -1.22373068e+00
1.50517374e-01 -5.24526477e-01 -2.99701523e-02 4.21747714e-01
-6.68037057e-01 -7.43852794e-01 -1.82592459e-02 6.74621940e-01
-4.70709831e-01 1.08636105e+00 1.22185779e+00 -1.62457213e-01
-4.37510163e-01 3.55287313e-01 -6.07028544e-01 9.00903821e-01
9.03363705e-01 7.53606483e-02 -7.11433589e-01 -1.76485647e-02
-4.17957902e-01 2.52462089e-01 -1.44792885e-01 -7.15208828e-01
4.77255523e-01 -4.37668681e-01 3.05292439e-02 -6.08517349e-01
6.68641567e-01 -7.70601392e-01 -7.67143488e-01 3.61963987e-01
-3.85312051e-01 -1.16588615e-01 -1.27617940e-01 6.03217304e-01
-5.86070001e-01 -6.04042292e-01 2.97049105e-01 -2.95248240e-01
-8.42729926e-01 -6.31663352e-02 -3.76174062e-01 1.40581548e-01
6.57873869e-01 -1.68242067e-01 -4.72944230e-01 -6.03713512e-01
-2.56217629e-01 3.62569779e-01 -1.70185342e-01 4.90737855e-01
6.44032538e-01 -1.10727453e+00 -6.23463571e-01 3.49932343e-01
9.87387002e-02 -2.22702995e-01 2.24253207e-01 5.52456856e-01
-1.22166783e-01 4.54263598e-01 -2.08075911e-01 -5.77663243e-01
-1.40225887e+00 2.58487076e-01 2.96345145e-01 -4.30091530e-01
-2.53192604e-01 6.74256146e-01 4.83628452e-01 -4.93149400e-01
2.71248221e-01 -4.33779545e-02 -8.99980664e-01 2.60979295e-01
5.36848009e-01 1.56463027e-01 -4.47109133e-01 -6.84503257e-01
-5.13031185e-02 1.56962246e-01 -5.95889688e-01 -7.80784190e-02
9.02467847e-01 -2.54728407e-01 -3.97348225e-01 9.19555664e-01
8.18599284e-01 3.32062811e-01 -7.98082292e-01 -7.55287051e-01
2.46931270e-01 -4.36192274e-01 -2.69893318e-01 -6.33982122e-01
-6.41163290e-01 1.07396197e+00 9.41433907e-02 7.42685556e-01
7.83601105e-01 1.08444020e-02 9.09229636e-01 5.13365746e-01
2.94661492e-01 -1.13417637e+00 4.47673261e-01 9.42543566e-01
8.17100763e-01 -1.51069605e+00 -4.90722619e-02 -7.03517973e-01
-1.32665813e+00 1.43258655e+00 1.11310923e+00 3.17353070e-01
7.60921955e-01 1.23528400e-02 2.50229657e-01 -1.44751400e-01
-5.83610296e-01 -1.94091022e-01 3.64903957e-01 3.98748279e-01
5.26645005e-01 -5.32517992e-02 -6.42643929e-01 1.24715281e+00
-1.41739190e-01 -7.15390801e-01 2.95174897e-01 7.13619053e-01
-9.07400370e-01 -1.41999042e+00 -1.12740792e-01 1.25503734e-01
-1.20469101e-01 1.15162609e-02 -9.21630979e-01 6.22375548e-01
7.62888342e-02 1.26633573e+00 -1.99526191e-01 -8.66515875e-01
9.30247456e-02 4.48786765e-01 -2.39183545e-01 -6.75808549e-01
-6.08273089e-01 1.23121381e-01 3.05372804e-01 8.92523900e-02
-4.81834292e-01 -4.24843639e-01 -1.36544085e+00 -2.41068184e-01
-4.08743620e-01 6.15308046e-01 2.34050423e-01 1.39101529e+00
-7.58742765e-02 3.23957384e-01 1.10496426e+00 -1.14915147e-01
-8.76446247e-01 -1.45452976e+00 -5.04129231e-01 3.46376717e-01
2.79567093e-02 -8.49095821e-01 -4.29298937e-01 -3.46276909e-01]
|
[12.707733154296875, 7.748104095458984]
|
1ee26e62-7687-4c2a-9165-79e37d0b0986
|
capabilities-of-gpt-4-on-medical-challenge
|
2303.13375
| null |
https://arxiv.org/abs/2303.13375v2
|
https://arxiv.org/pdf/2303.13375v2.pdf
|
Capabilities of GPT-4 on Medical Challenge Problems
|
Large language models (LLMs) have demonstrated remarkable capabilities in natural language understanding and generation across various domains, including medicine. We present a comprehensive evaluation of GPT-4, a state-of-the-art LLM, on medical competency examinations and benchmark datasets. GPT-4 is a general-purpose model that is not specialized for medical problems through training or engineered to solve clinical tasks. Our analysis covers two sets of official practice materials for the USMLE, a three-step examination program used to assess clinical competency and grant licensure in the United States. We also evaluate performance on the MultiMedQA suite of benchmark datasets. Beyond measuring model performance, experiments were conducted to investigate the influence of test questions containing both text and images on model performance, probe for memorization of content during training, and study probability calibration, which is of critical importance in high-stakes applications like medicine. Our results show that GPT-4, without any specialized prompt crafting, exceeds the passing score on USMLE by over 20 points and outperforms earlier general-purpose models (GPT-3.5) as well as models specifically fine-tuned on medical knowledge (Med-PaLM, a prompt-tuned version of Flan-PaLM 540B). In addition, GPT-4 is significantly better calibrated than GPT-3.5, demonstrating a much-improved ability to predict the likelihood that its answers are correct. We also explore the behavior of the model qualitatively through a case study that shows the ability of GPT-4 to explain medical reasoning, personalize explanations to students, and interactively craft new counterfactual scenarios around a medical case. Implications of the findings are discussed for potential uses of GPT-4 in medical education, assessment, and clinical practice, with appropriate attention to challenges of accuracy and safety.
|
['Eric Horvitz', 'Dean Carignan', 'Scott Mayer McKinney', 'Nicholas King', 'Harsha Nori']
|
2023-03-20
| null | null | null | null |
['memorization']
|
['natural-language-processing']
|
[ 1.81312293e-01 6.62564218e-01 -4.49390352e-01 -5.00505507e-01
-1.41345119e+00 -6.58023894e-01 1.60525605e-01 4.61368859e-01
-5.27448058e-01 8.03710461e-01 4.19697225e-01 -1.28325522e+00
-6.82622313e-01 -5.86917818e-01 -1.11484790e+00 -1.56022862e-01
3.31671506e-01 7.44290531e-01 -1.79730237e-01 -1.82389483e-01
1.47845402e-01 2.33961925e-01 -1.10238254e+00 7.69043148e-01
1.38343978e+00 1.89591825e-01 9.08392668e-02 8.72408271e-01
1.42572656e-01 9.92366672e-01 -5.87574244e-01 -9.67530012e-01
-1.78793594e-01 -4.93162870e-01 -9.66314793e-01 -2.81360477e-01
8.67376626e-01 -2.80331701e-01 -1.09137997e-01 5.63113511e-01
6.45325541e-01 1.72709465e-01 4.11441863e-01 -8.07223737e-01
-9.08553600e-01 8.02881718e-01 -1.98481843e-01 3.56843591e-01
7.15429306e-01 5.96525013e-01 7.57230937e-01 -1.84444606e-01
6.27861857e-01 1.33816230e+00 7.35704720e-01 9.43194687e-01
-1.19635046e+00 -8.30347061e-01 2.05742642e-02 2.52860673e-02
-7.98764169e-01 2.49325726e-02 -2.96908747e-02 -4.82549489e-01
7.25208163e-01 3.48855406e-01 6.55751705e-01 1.20300508e+00
7.57455647e-01 8.97835791e-01 1.47455454e+00 -4.35713083e-01
9.80935246e-02 3.71589571e-01 3.25752407e-01 8.37891996e-01
4.81336474e-01 2.06905201e-01 -6.05889976e-01 -3.47824544e-01
6.80192530e-01 -3.18479806e-01 -4.83005047e-01 7.83178583e-02
-1.14835107e+00 7.24031031e-01 2.03298435e-01 6.20850585e-02
-4.09634590e-01 3.84318084e-02 -2.20624015e-01 9.96732339e-02
-7.70432130e-02 9.86392438e-01 -6.59215510e-01 -4.54679281e-01
-9.33076441e-01 5.61960816e-01 9.41300571e-01 5.55944622e-01
-1.29571050e-01 -2.47292936e-01 -8.54277909e-01 5.16140163e-01
2.31405661e-01 4.96023089e-01 6.49296641e-01 -1.01819944e+00
5.25546491e-01 5.21396339e-01 8.10822845e-02 -4.25828010e-01
-4.38286453e-01 -7.34086454e-01 -5.44876568e-02 -9.14584175e-02
5.32261848e-01 -2.72568226e-01 -1.07053268e+00 1.80262303e+00
5.37623130e-02 4.74910319e-01 2.52171189e-01 4.65903699e-01
1.19300187e+00 2.91602939e-01 7.39092588e-01 6.59045652e-02
1.57246244e+00 -6.40684247e-01 -6.69703364e-01 -3.68969351e-01
1.07798886e+00 -7.94878125e-01 1.39095891e+00 6.66013300e-01
-1.56303155e+00 -5.17295539e-01 -6.06271088e-01 4.96563443e-04
7.44456649e-02 -2.17519253e-01 5.67689836e-01 9.71759021e-01
-9.74143803e-01 5.56922495e-01 -7.27139711e-01 1.10732295e-01
5.12084305e-01 1.49841294e-01 -1.89919248e-01 -5.98336041e-01
-1.37178504e+00 1.13867152e+00 8.99149030e-02 -2.69051224e-01
-1.02988815e+00 -1.68811715e+00 -1.04519093e+00 3.88751686e-01
1.07054166e-01 -9.31169450e-01 1.74642503e+00 -3.27374309e-01
-1.23566449e+00 8.39652002e-01 -1.02145188e-01 -5.94405949e-01
6.69619739e-01 -2.11979911e-01 -4.45349067e-01 1.11042045e-01
1.12951830e-01 8.15022171e-01 -1.20304719e-01 -9.91856396e-01
-4.86734152e-01 3.53045166e-02 1.63570195e-01 2.58037269e-01
2.85774264e-02 -4.45410907e-01 -1.91885084e-01 -4.45606202e-01
-2.65663624e-01 -9.62424517e-01 -6.26381516e-01 -5.73093176e-01
-2.98654050e-01 -2.40800649e-01 2.91299112e-02 -6.80254042e-01
1.36813271e+00 -1.78687501e+00 -5.97886562e-01 2.56085485e-01
1.19467773e-01 2.97575146e-01 -2.86422104e-01 1.72934122e-02
-4.03723031e-01 5.27998507e-01 -5.78526780e-02 6.03566617e-02
-1.47599027e-01 1.38911128e-01 -1.20111309e-01 -1.06748000e-01
1.23429507e-01 1.23402989e+00 -1.11424577e+00 -7.27941155e-01
1.89431995e-01 2.59183615e-01 -1.10457659e+00 1.43291846e-01
-1.48451120e-01 5.58223486e-01 -4.26309079e-01 3.81542295e-01
3.11077952e-01 -6.36097729e-01 2.04749450e-01 2.30343491e-01
2.73305833e-01 6.10968411e-01 -7.68511355e-01 1.66749239e+00
-6.12911701e-01 2.53822654e-01 -4.30274844e-01 -4.33856040e-01
2.98665583e-01 5.01262248e-01 2.84421593e-01 -5.69563329e-01
-5.98098896e-02 1.34851858e-01 5.62071800e-01 -8.84399831e-01
3.23843360e-01 -4.74525750e-01 1.50332868e-01 5.22877693e-01
1.32126287e-01 -4.95695025e-01 1.43934593e-01 4.81081128e-01
1.18281054e+00 -6.08887188e-02 1.02263547e-01 -3.91883045e-01
2.63107955e-01 2.66634732e-01 3.53559107e-01 1.24228156e+00
-8.53678659e-02 5.27061582e-01 4.57416385e-01 -4.25800458e-02
-4.06698674e-01 -1.16622710e+00 -4.94804502e-01 8.43225718e-01
-2.27540776e-01 -2.29910254e-01 -7.96956837e-01 -8.01186860e-01
1.64220512e-01 1.65439153e+00 -7.53989160e-01 -4.61761028e-01
-2.75518477e-01 -7.40165412e-01 5.30550301e-01 5.02633214e-01
9.63900983e-02 -1.10198653e+00 -6.66616082e-01 2.22056806e-01
-4.98073995e-01 -9.65062022e-01 -5.10467947e-01 -2.32828051e-01
-1.04949105e+00 -1.31008768e+00 -6.36219680e-01 -5.41032672e-01
7.52790093e-01 -1.44638523e-01 1.54847038e+00 2.27358893e-01
-3.27572376e-01 1.19563055e+00 7.47136623e-02 -7.67266095e-01
-8.90255392e-01 -1.79249987e-01 -3.52066338e-01 -7.38384068e-01
6.33893490e-01 6.56519905e-02 -7.51129329e-01 1.79119125e-01
-1.09270799e+00 2.47712553e-01 8.76597285e-01 1.04516971e+00
5.05986273e-01 -3.63151431e-01 4.90831107e-01 -1.48391199e+00
1.06276906e+00 -4.94649351e-01 -2.19614282e-01 7.92044759e-01
-8.50274563e-01 1.14443898e-01 4.83496450e-02 -6.69650137e-01
-1.41115725e+00 -5.38634121e-01 -2.59660721e-01 -1.54475510e-01
-2.74890810e-02 7.85179436e-01 3.18018168e-01 -1.04178272e-01
1.03487182e+00 4.70359325e-02 -1.23300500e-01 -1.64711311e-01
6.98562041e-02 2.02862546e-01 5.60872197e-01 -1.10141706e+00
5.28746665e-01 -1.21715426e-01 -2.13389307e-01 -1.80549771e-01
-1.39975977e+00 -2.25383583e-02 9.88997072e-02 -7.48716518e-02
8.12961519e-01 -1.05689347e+00 -9.90990222e-01 -1.93103164e-01
-6.41738892e-01 -6.45631909e-01 -4.81540531e-01 1.03043973e+00
-3.83733243e-01 -2.49458756e-03 -8.51588845e-01 -4.87435311e-01
-7.12529942e-02 -1.45856559e+00 6.01631761e-01 5.64196229e-01
-5.62890232e-01 -1.35820997e+00 1.37300462e-01 1.15555871e+00
2.98025042e-01 2.29134131e-02 1.26795685e+00 -9.48289752e-01
-4.78487104e-01 -2.46881276e-01 3.24832618e-01 1.68375984e-01
-1.39108971e-01 -2.40081638e-01 -8.41849267e-01 -1.64964586e-01
6.17717281e-02 -6.48790658e-01 5.83738208e-01 8.30596447e-01
1.65581393e+00 -1.70435131e-01 -3.18560004e-01 3.82385731e-01
1.11425650e+00 2.88710803e-01 5.18305540e-01 2.26550207e-01
2.88784981e-01 6.46292746e-01 6.87983394e-01 8.27712715e-02
6.70044661e-01 -3.62034664e-02 5.62495328e-02 -7.30730370e-02
-9.58351791e-02 -7.00647056e-01 5.54511473e-02 7.02128470e-01
7.06624091e-02 -2.34503523e-01 -1.26860631e+00 6.75650716e-01
-1.39037740e+00 -7.00052381e-01 -1.20973393e-01 2.19848204e+00
1.30642045e+00 3.72238487e-01 -6.32369041e-01 -4.65230286e-01
1.86213404e-01 -4.42905188e-01 -4.40329105e-01 -5.93305111e-01
6.68857321e-02 7.16604948e-01 4.28197652e-01 6.55428827e-01
-4.00004506e-01 5.75409234e-01 7.58001089e+00 8.48397315e-01
-5.51856697e-01 1.11091204e-01 1.26067674e+00 -1.40555039e-01
-9.47157562e-01 -1.03666559e-01 -7.45053768e-01 1.61156729e-01
1.50457895e+00 -3.26552570e-01 -5.15681729e-02 6.67558134e-01
2.37613440e-01 -2.17755362e-01 -1.32104027e+00 4.97541577e-01
-4.07338813e-02 -1.65654707e+00 3.86124641e-01 1.34719871e-02
1.31974828e+00 -5.60995281e-01 5.87565601e-01 7.69561589e-01
8.54694247e-01 -1.56209314e+00 3.30330312e-01 4.73585457e-01
7.93843627e-01 -5.12833118e-01 8.78536284e-01 2.93974042e-01
-3.12033087e-01 4.40953881e-04 -1.01146661e-01 1.35510191e-01
-3.77443396e-02 3.42879832e-01 -1.29894936e+00 3.94385666e-01
4.98789549e-01 1.17426075e-01 -5.79097211e-01 1.03632045e+00
-4.87292320e-01 1.20657492e+00 1.91626668e-01 2.22966239e-01
3.47413629e-01 3.54203910e-01 -6.96196267e-03 1.29451776e+00
2.98123658e-01 7.49312758e-01 -8.47359970e-02 8.16313088e-01
-1.53215766e-01 -1.76188909e-03 -1.45097762e-01 -1.89243928e-02
5.22011995e-01 9.16173935e-01 -2.07919121e-01 -5.36879599e-01
-3.01102728e-01 2.62877822e-01 6.27049198e-03 3.96229297e-01
-8.88204455e-01 1.57220230e-01 4.15012449e-01 3.83577138e-01
-3.27305794e-01 5.23133218e-01 -4.65700120e-01 -8.99655581e-01
-4.89140838e-01 -1.41192174e+00 6.36359215e-01 -1.05992746e+00
-1.20130181e+00 2.25646853e-01 2.32529461e-01 -1.02715325e+00
-2.19283938e-01 -7.07934380e-01 -4.64104325e-01 1.09850109e+00
-1.51522481e+00 -7.44043052e-01 -3.02287012e-01 7.05169201e-01
4.34544355e-01 3.39871198e-02 8.27015579e-01 1.36551604e-01
-3.23719949e-01 1.01109290e+00 -1.94997162e-01 1.09537886e-02
1.09471297e+00 -1.31333244e+00 1.91936836e-01 5.68205595e-01
-8.96308646e-02 1.06314576e+00 7.03479290e-01 -9.36459243e-01
-1.02890229e+00 -7.73383141e-01 8.42297912e-01 -9.81232941e-01
3.08818340e-01 3.69971603e-01 -1.09963560e+00 1.06105733e+00
2.49992386e-02 -3.95623446e-01 1.31063521e+00 1.90956727e-01
-1.16235800e-02 2.27226958e-01 -1.39140880e+00 8.11647773e-01
6.74255848e-01 -3.21909726e-01 -1.02800059e+00 5.67119837e-01
9.13426518e-01 -1.15696859e+00 -1.20756590e+00 4.63061482e-01
5.78473926e-01 -7.24652052e-01 1.04067719e+00 -1.29388750e+00
1.03959453e+00 3.69308382e-01 3.60561520e-01 -1.36688638e+00
-3.29759151e-01 -4.89922464e-01 2.10962430e-01 5.76334894e-01
8.05873454e-01 -6.31433427e-01 1.00225687e+00 1.29463923e+00
-2.82939643e-01 -1.08927596e+00 -5.59703648e-01 -2.75864750e-01
4.52557683e-01 -5.13563514e-01 4.51841950e-01 1.13706017e+00
-4.04737815e-02 -1.58274218e-01 7.74176717e-02 3.12888294e-01
5.07066011e-01 -5.45239486e-02 5.30234456e-01 -1.07510126e+00
-6.35173023e-01 -4.03773963e-01 1.51515394e-01 -9.56861198e-01
-4.17372165e-03 -7.63828456e-01 -1.67212002e-02 -1.77666116e+00
4.06322062e-01 -4.12710518e-01 -4.86828685e-01 6.02864742e-01
-8.18480134e-01 -4.19998497e-01 1.17606223e-01 -2.26276860e-01
-1.52714580e-01 -1.48575017e-02 1.84131265e+00 1.92392822e-02
-2.22907647e-01 1.19463190e-01 -1.34991372e+00 4.75015730e-01
5.20948470e-01 -5.13927102e-01 -7.29024172e-01 -3.92241120e-01
1.75402254e-01 5.85771203e-01 3.37795287e-01 -9.41101074e-01
2.23325208e-01 -5.56238115e-01 6.57776535e-01 -3.29452991e-01
1.46234212e-02 -5.15425384e-01 1.28108412e-01 8.88506293e-01
-9.53974247e-01 1.67865798e-01 8.86690676e-01 3.26507926e-01
-5.55006266e-02 -4.75122869e-01 6.26494288e-01 -4.72856641e-01
-2.23823830e-01 -5.87882809e-02 -4.02479857e-01 4.48388904e-01
9.31140482e-01 -2.34133354e-03 -5.71370065e-01 -5.70118845e-01
-8.22604358e-01 7.04785824e-01 -4.26186249e-02 3.62629771e-01
6.18822217e-01 -8.62154245e-01 -1.01540434e+00 -2.43275017e-02
-3.53632905e-02 1.61036938e-01 9.43931699e-01 8.14312100e-01
-6.32993221e-01 8.48794878e-01 3.12019624e-02 -5.25788784e-01
-1.04270422e+00 2.63358504e-01 6.13660693e-01 -1.02536058e+00
-2.30826631e-01 1.14695215e+00 5.42049825e-01 -6.43251359e-01
1.28482714e-01 -6.25955820e-01 -8.05218816e-02 -5.43865323e-01
6.63217545e-01 5.50112221e-03 -3.95765379e-02 1.88477486e-01
-3.65667939e-02 1.87165007e-01 -2.95747966e-01 -1.74238294e-01
1.05237997e+00 2.54541665e-01 4.48287398e-01 2.96136975e-01
4.02574807e-01 1.15600564e-01 -9.57665443e-01 -4.41222377e-02
-8.74794647e-02 -1.94528326e-01 -2.76616644e-02 -1.72420299e+00
-5.59343100e-01 7.41120875e-01 5.65782130e-01 -5.21150470e-01
8.47845554e-01 -2.09973603e-01 5.06007552e-01 2.90661037e-01
4.11407985e-02 -6.27375722e-01 2.96060562e-01 6.74342066e-02
5.90563774e-01 -1.33937132e+00 6.85357153e-02 -1.23009443e-01
-8.81118298e-01 7.76905596e-01 1.12211597e+00 3.25274140e-01
4.80366021e-01 -3.78946774e-02 2.39011392e-01 -2.75825173e-01
-9.60460305e-01 5.31756043e-01 7.16658175e-01 2.48813644e-01
8.36185157e-01 2.65665114e-01 -3.55902225e-01 9.51956451e-01
-5.35926223e-01 2.88845152e-01 8.14313650e-01 8.99936140e-01
-8.36089775e-02 -9.76417243e-01 -6.06886268e-01 7.79100716e-01
-8.19548726e-01 -3.92268002e-01 6.49309754e-02 9.93816435e-01
2.15019602e-02 8.32358837e-01 -1.85957938e-01 -7.79206604e-02
4.81524050e-01 3.62386912e-01 5.33141375e-01 -1.01151848e+00
-1.14238584e+00 -3.65609795e-01 -1.56694949e-02 -4.65359598e-01
-1.63625762e-01 -4.09969568e-01 -1.20932424e+00 -3.76312494e-01
-8.77886042e-02 5.13088346e-01 4.51139063e-01 9.24003541e-01
2.60987788e-01 1.15258110e+00 -2.65537620e-01 1.93913952e-01
-7.90626884e-01 -9.61586595e-01 -1.36988819e-01 4.49279100e-01
2.06743211e-01 -3.56732875e-01 -1.07916661e-01 -1.37419701e-01]
|
[8.822016716003418, 8.482512474060059]
|
6b89b40f-29f2-462b-a50c-3194f3f4a381
|
the-golden-ratio-of-learning-and-momentum
|
2006.04751
| null |
https://arxiv.org/abs/2006.04751v1
|
https://arxiv.org/pdf/2006.04751v1.pdf
|
The Golden Ratio of Learning and Momentum
|
Gradient descent has been a central training principle for artificial neural networks from the early beginnings to today's deep learning networks. The most common implementation is the backpropagation algorithm for training feed-forward neural networks in a supervised fashion. Backpropagation involves computing the gradient of a loss function, with respect to the weights of the network, to update the weights and thus minimize loss. Although the mean square error is often used as a loss function, the general stochastic gradient descent principle does not immediately connect with a specific loss function. Another drawback of backpropagation has been the search for optimal values of two important training parameters, learning rate and momentum weight, which are determined empirically in most systems. The learning rate specifies the step size towards a minimum of the loss function when following the gradient, while the momentum weight considers previous weight changes when updating current weights. Using both parameters in conjunction with each other is generally accepted as a means to improving training, although their specific values do not follow immediately from standard backpropagation theory. This paper proposes a new information-theoretical loss function motivated by neural signal processing in a synapse. The new loss function implies a specific learning rate and momentum weight, leading to empirical parameters often used in practice. The proposed framework also provides a more formal explanation of the momentum term and its smoothing effect on the training process. All results taken together show that loss, learning rate, and momentum are closely connected. To support these theoretical findings, experiments for handwritten digit recognition show the practical usefulness of the proposed loss function and training parameters.
|
['Stefan Jaeger']
|
2020-06-08
| null | null | null | null |
['handwritten-digit-recognition']
|
['computer-vision']
|
[ 4.60083969e-02 -6.58285394e-02 -3.68146226e-02 -6.41317368e-01
2.88085639e-01 6.24129586e-02 3.07970196e-01 3.03913444e-01
-1.13796067e+00 8.30654383e-01 -3.90797943e-01 -3.01201105e-01
-4.31280315e-01 -8.97236288e-01 -4.60921586e-01 -8.48414302e-01
-9.84388068e-02 9.81526673e-02 4.55518246e-01 -3.23940277e-01
6.07628167e-01 7.18411744e-01 -1.34574497e+00 -1.32782415e-01
7.33106971e-01 1.01758480e+00 2.81595677e-01 5.61267853e-01
-3.47847819e-01 5.94123185e-01 -5.37705779e-01 -4.31331962e-01
2.08282173e-01 -5.96887767e-01 -6.45107388e-01 -1.13215223e-01
-9.55899730e-02 -1.00102223e-01 -8.03347006e-02 1.10117817e+00
3.97455186e-01 2.62127936e-01 6.94462657e-01 -7.66912997e-01
-2.99470991e-01 5.16514599e-01 -2.02025548e-01 5.34107327e-01
-1.02334484e-01 -1.57780647e-01 7.37573683e-01 -8.78275692e-01
3.07632893e-01 7.54652560e-01 1.02076650e+00 5.27362287e-01
-9.17272925e-01 -3.46622765e-01 1.15181588e-01 2.74189502e-01
-1.16497290e+00 3.28374803e-02 8.52962136e-01 -3.36341470e-01
9.66914356e-01 1.15964584e-01 8.09353888e-01 2.69042611e-01
4.65639204e-01 4.97793078e-01 7.20221817e-01 -9.39448357e-01
3.73418152e-01 6.69314682e-01 3.64683181e-01 7.30142415e-01
3.92783135e-01 8.77792090e-02 -2.90924132e-01 1.51521146e-01
7.57818878e-01 -6.74410984e-02 -3.44799787e-01 -2.34049797e-01
-3.65114152e-01 8.16699624e-01 6.52242780e-01 7.28306055e-01
-4.14145380e-01 1.27900630e-01 3.33560050e-01 5.03234863e-01
4.12472606e-01 2.89979666e-01 -2.66777962e-01 -9.99241397e-02
-9.71162200e-01 5.09976931e-02 8.72646093e-01 1.50795355e-01
8.34135354e-01 3.18179309e-01 8.65387470e-02 1.02526116e+00
6.08848631e-01 1.28775701e-01 8.29200983e-01 -5.05654216e-01
3.94166827e-01 6.94790602e-01 -1.07441634e-01 -1.19236100e+00
-3.80613804e-01 -6.86098158e-01 -6.42573178e-01 9.43592191e-01
8.63577902e-01 -4.22445238e-01 -6.19632661e-01 1.57491958e+00
1.72011554e-01 -1.53543532e-01 -8.62316638e-02 1.09109926e+00
2.42757648e-01 5.73019922e-01 -5.69230616e-02 -3.96674037e-01
8.17313194e-01 -6.76862895e-01 -6.88402414e-01 -2.95734107e-01
3.39287221e-01 -6.52781308e-01 8.45534742e-01 3.93664628e-01
-1.19129431e+00 -4.48210984e-01 -1.26550555e+00 2.94895887e-01
-4.46885705e-01 6.67130500e-02 4.41387951e-01 6.33992672e-01
-9.16368902e-01 1.24289727e+00 -8.20581973e-01 -2.32103527e-01
1.13836795e-01 4.72730666e-01 6.70924112e-02 5.12203276e-01
-1.14203596e+00 1.26090884e+00 6.18829310e-01 5.85696220e-01
-2.97290981e-02 -3.65193456e-01 -3.65524769e-01 1.78806722e-01
-1.30075291e-01 -4.47791666e-01 9.97806191e-01 -1.20739794e+00
-2.00861979e+00 6.72363460e-01 -9.23926011e-03 -8.80043268e-01
7.23678350e-01 -2.51796514e-01 -3.53754133e-01 -3.78322303e-02
-6.19013488e-01 3.60466391e-01 9.12410975e-01 -8.95357311e-01
-5.41629016e-01 -1.34869784e-01 -2.34692603e-01 3.31296831e-01
-6.87873065e-01 -1.68905899e-01 -9.82592925e-02 -4.15967852e-01
4.32613522e-01 -4.99956965e-01 -1.78630516e-01 2.23907739e-01
-6.18537515e-02 -6.95837215e-02 5.43486953e-01 -3.91880184e-01
1.24245131e+00 -2.12619400e+00 -1.12606183e-01 5.59310436e-01
-1.48920342e-01 4.09005016e-01 2.56058246e-01 2.90779084e-01
-1.89291432e-01 -3.12274456e-01 -4.13859814e-01 -4.28363234e-02
-2.64310211e-01 2.00057298e-01 -1.36418819e-01 3.46416116e-01
8.27889070e-02 5.07668257e-01 -7.67050505e-01 -1.65382609e-01
2.25889072e-01 9.25641000e-01 -4.56639498e-01 -1.10562660e-01
2.32768357e-01 -1.25839427e-01 -1.43663317e-01 -1.41528055e-01
4.57917690e-01 6.45472035e-02 6.46156305e-03 -9.31791961e-02
-3.66748929e-01 2.92092502e-01 -1.41032243e+00 9.39813077e-01
-2.53987432e-01 8.48150492e-01 -7.21211731e-02 -1.27891076e+00
1.46531916e+00 2.47594327e-01 3.18962336e-01 -4.22076136e-01
4.16165203e-01 6.01349533e-01 1.34985268e-01 -3.45871866e-01
4.49553996e-01 -5.49394667e-01 8.18867028e-01 2.27881163e-01
2.46926006e-02 2.66304668e-02 4.45237637e-01 -3.47054362e-01
5.56212485e-01 2.11455598e-02 2.07752347e-01 -1.75150529e-01
8.30038846e-01 -2.36656591e-01 4.67004597e-01 4.49099511e-01
-8.98440704e-02 4.96883720e-01 3.27825606e-01 -5.54971635e-01
-8.69579077e-01 -6.57548666e-01 -4.61186498e-01 8.54035199e-01
-7.22026899e-02 1.25084177e-01 -6.37353480e-01 -4.76272136e-01
1.41853467e-02 5.76730013e-01 -4.68447059e-01 -2.74186760e-01
-5.82691014e-01 -9.23398674e-01 4.77595627e-01 1.56713009e-01
6.55770957e-01 -1.23928261e+00 -9.35673714e-01 3.81850809e-01
4.14630324e-01 -3.99213016e-01 -5.02346493e-02 6.20906949e-01
-1.46937132e+00 -9.98206735e-01 -1.03303230e+00 -7.85519540e-01
9.48794067e-01 -2.47954383e-01 7.18133092e-01 3.75135869e-01
-2.06904545e-01 7.74256065e-02 -5.71435019e-02 -4.78977054e-01
-4.68279511e-01 1.19841415e-02 5.50557934e-02 -5.91360889e-02
3.30309778e-01 -7.25525320e-01 -5.37366033e-01 4.42379490e-02
-7.54994869e-01 -3.58199537e-01 5.72575152e-01 8.41676772e-01
4.32307243e-01 1.74459010e-01 5.13426483e-01 -9.34012234e-01
9.30803537e-01 -1.64501771e-01 -6.20167494e-01 2.34017894e-02
-1.10997057e+00 2.98716009e-01 6.53762162e-01 -4.52080697e-01
-9.97966170e-01 -2.34025806e-01 -5.43171763e-01 8.82335752e-03
1.41601162e-02 6.91339254e-01 4.90855604e-01 -3.66597772e-01
8.02066386e-01 3.44607800e-01 1.94538966e-01 -4.51469898e-01
1.09398449e-02 2.21575931e-01 3.93670738e-01 -1.49870232e-01
4.68006253e-01 1.31285906e-01 -1.12500126e-02 -6.93853676e-01
-5.01423717e-01 -3.25443685e-01 -5.32521486e-01 -4.15598869e-01
2.96659499e-01 -1.45101890e-01 -6.81729376e-01 6.74740374e-01
-1.02668905e+00 -1.10612288e-01 -6.52065277e-01 8.82043719e-01
-4.57808793e-01 2.50265151e-01 -7.18319356e-01 -9.79659259e-01
-5.84950984e-01 -6.36528492e-01 -1.64321825e-01 4.29798186e-01
-2.09474824e-02 -1.50702918e+00 4.77488302e-02 -5.44059396e-01
6.44500434e-01 1.99187919e-02 7.73049235e-01 -4.93248850e-01
1.46098301e-01 -4.63191718e-01 4.40356992e-02 9.11624849e-01
1.10018551e-01 1.06083535e-01 -7.37193584e-01 -2.63143122e-01
4.50545520e-01 6.36000633e-02 1.00747406e+00 5.61414778e-01
6.91477001e-01 -2.99703062e-01 -3.48297283e-02 6.55119896e-01
1.81555176e+00 3.92831236e-01 6.27686799e-01 6.32325292e-01
1.82572752e-01 4.13459152e-01 1.61526710e-01 5.31274438e-01
-2.97258496e-01 2.85249680e-01 4.18603897e-01 1.23038599e-02
-6.83023483e-02 -4.11642045e-02 2.28824824e-01 1.14001107e+00
-3.36631060e-01 3.14828426e-01 -7.47951269e-01 1.79214448e-01
-1.51770151e+00 -1.16822743e+00 -1.67416394e-01 2.62829208e+00
1.06411338e+00 8.78158510e-01 -1.19230524e-02 7.71927238e-01
6.19271815e-01 -8.94324183e-02 -4.94448543e-01 -7.69354463e-01
8.20976589e-03 2.15049341e-01 5.99092841e-01 8.89022589e-01
-7.39158630e-01 4.19394314e-01 6.52280998e+00 4.55692053e-01
-1.73640072e+00 -2.85173297e-01 2.70589113e-01 4.61819060e-02
-9.66328904e-02 -1.00402258e-01 -1.04369724e+00 5.19467235e-01
7.95535445e-01 -2.52588928e-01 3.45791042e-01 9.05091345e-01
1.60278767e-01 -1.96031108e-01 -8.33332002e-01 8.17023218e-01
-1.63657859e-01 -1.20306516e+00 -3.30317169e-02 -2.33573049e-01
5.07777035e-01 -5.73324040e-02 7.10309893e-02 1.40810683e-01
-3.26534152e-01 -6.90926969e-01 6.72049165e-01 6.13091588e-01
9.89557058e-02 -8.08718026e-01 8.82090628e-01 4.49286342e-01
-7.40844905e-01 -2.15834931e-01 -6.06550395e-01 -3.83838952e-01
1.25349209e-01 1.06291997e+00 -8.67595434e-01 7.80420080e-02
3.20274889e-01 4.19686735e-01 -1.74064621e-01 1.64537239e+00
-3.71836275e-01 6.58123016e-01 -6.17624700e-01 -4.54510868e-01
4.04583722e-01 -4.30447072e-01 6.52503252e-01 1.44442475e+00
2.95597970e-01 -2.67970145e-01 -4.58608717e-01 9.28463995e-01
8.05824474e-02 3.12948853e-01 -2.26136103e-01 1.39942914e-01
5.18279731e-01 9.87982452e-01 -8.35786581e-01 -4.44372781e-02
-1.49446324e-01 8.30003798e-01 2.56167471e-01 4.64681029e-01
-4.92211163e-01 -9.50575829e-01 3.82465959e-01 3.24495643e-01
3.68126094e-01 -1.94962621e-01 -5.48451662e-01 -5.32428384e-01
2.54994541e-01 -1.70358881e-01 3.58903334e-02 -2.69048154e-01
-1.01377273e+00 6.86572611e-01 -1.33005202e-01 -1.10281265e+00
-3.85812759e-01 -8.38004529e-01 -8.10607612e-01 1.21504891e+00
-1.68807817e+00 -2.11825565e-01 -2.51550991e-02 4.32261825e-01
3.07912856e-01 -2.02495307e-01 7.94518709e-01 4.57775205e-01
-6.00741088e-01 5.79721391e-01 3.42694163e-01 6.90675303e-02
2.70163536e-01 -1.12964463e+00 -1.05437711e-01 7.10476577e-01
-3.65857268e-03 7.88881958e-01 1.03997958e+00 -2.59450644e-01
-6.46481395e-01 -4.22771335e-01 1.09350073e+00 4.19266999e-01
4.78154868e-01 2.13501468e-01 -1.23816025e+00 2.62613207e-01
-1.21312171e-01 -5.79028241e-02 3.82227391e-01 2.78874990e-02
2.46000011e-02 -3.42429221e-01 -1.14326870e+00 3.12719882e-01
3.10839981e-01 -2.69394904e-01 -6.75921142e-01 -7.62324035e-02
-3.64769585e-02 -2.77039200e-01 -6.87259018e-01 1.71792537e-01
6.90737426e-01 -1.29364109e+00 6.92952752e-01 -1.99249938e-01
1.81702435e-01 -1.08985804e-01 3.38233083e-01 -1.43009782e+00
-2.35798687e-01 -2.81655937e-01 -5.47683239e-02 1.03783643e+00
6.05074286e-01 -9.40445423e-01 1.05084407e+00 6.44568622e-01
6.52358355e-03 -1.31147289e+00 -8.72913897e-01 -8.60449553e-01
1.07138492e-01 -5.32786131e-01 3.83337513e-02 6.45061076e-01
5.87698221e-02 2.63280589e-02 5.33780716e-02 -1.99165285e-01
5.13330340e-01 -4.55659240e-01 1.06198363e-01 -1.39055657e+00
-3.65142673e-01 -9.01562393e-01 -6.28824532e-01 -1.05420971e+00
-2.92595178e-01 -7.45310426e-01 4.90893275e-02 -1.61907506e+00
-3.76697689e-01 -5.91284752e-01 -6.97626173e-01 3.34959537e-01
-8.27486590e-02 2.29415335e-02 2.01724917e-01 4.26251531e-01
2.30922908e-01 4.24087882e-01 9.42336679e-01 1.17781773e-01
-6.54156864e-01 4.77518976e-01 -2.35795602e-01 1.17786098e+00
9.64697599e-01 -5.95370293e-01 -5.31089008e-01 -4.90269452e-01
2.99103469e-01 -4.52436328e-01 1.41776807e-03 -1.26579344e+00
5.20301819e-01 5.79868294e-02 3.71710300e-01 -1.73441455e-01
2.91749209e-01 -1.02989674e+00 -7.34040663e-02 8.54215443e-01
-4.27912861e-01 -1.54106617e-01 1.38970882e-01 2.67393649e-01
-4.20981139e-01 -1.08451796e+00 1.08849025e+00 1.76680628e-02
-8.26237857e-01 -8.70552808e-02 -3.23321521e-01 -1.80322990e-01
7.40038753e-01 -7.93560028e-01 2.43576944e-01 -1.94118738e-01
-8.74509335e-01 -1.48209855e-01 2.76530534e-02 1.27911150e-01
8.87995422e-01 -1.13052034e+00 -5.95789313e-01 4.43622887e-01
-4.16308612e-01 -4.41258937e-01 -1.82399064e-01 8.73067141e-01
-8.66568983e-01 2.13622421e-01 -3.99834067e-01 -2.75171340e-01
-1.23806834e+00 1.01552986e-01 8.69192421e-01 -2.11560130e-01
-6.17056608e-01 1.25622523e+00 -6.34146750e-01 -7.94886947e-02
6.28542602e-01 -4.41669464e-01 -3.97176027e-01 -4.31706943e-02
6.73129916e-01 4.81165886e-01 1.84476852e-01 -1.06902868e-01
-2.46173367e-01 5.87957799e-01 -1.46962814e-02 -1.99899420e-01
1.33185863e+00 -3.65319215e-02 -1.45119250e-01 8.61514807e-01
1.09104872e+00 -2.62490094e-01 -1.21883285e+00 -2.27651492e-01
2.99121231e-01 -1.96720317e-01 1.84059516e-01 -7.40057528e-01
-1.05090654e+00 9.06164110e-01 8.54252398e-01 4.51906770e-01
1.13493216e+00 -7.33196437e-01 5.98467827e-01 6.28293753e-01
1.88927516e-01 -1.45138764e+00 -1.69456854e-01 7.48785734e-01
7.37510860e-01 -9.62754428e-01 4.06033322e-02 -1.50849581e-01
-2.47305214e-01 1.72986639e+00 4.12167460e-01 -4.33227539e-01
8.78935456e-01 2.79888928e-01 1.55122444e-01 1.41782120e-01
-4.21899050e-01 5.56494109e-02 2.77973771e-01 2.42885828e-01
8.19816768e-01 -3.25030059e-01 -1.06503451e+00 1.87628612e-01
-1.37877703e-01 3.79189819e-01 1.80476055e-01 8.74813616e-01
-1.00462997e+00 -1.10300565e+00 -4.11760569e-01 3.14359397e-01
-5.32074749e-01 -1.77105274e-02 6.89054839e-03 6.64255083e-01
2.20821917e-01 6.81973338e-01 1.47869557e-01 -2.14688882e-01
5.01781344e-01 1.80691332e-01 5.66229701e-01 -2.97196269e-01
-6.95711315e-01 -4.12514478e-01 -3.54408413e-01 3.38439457e-02
-3.22933137e-01 -1.81530371e-01 -1.73423028e+00 -3.54185134e-01
-5.59648514e-01 5.38325429e-01 1.24161804e+00 1.06695664e+00
-2.04829827e-01 5.33516824e-01 5.35833597e-01 -7.10396409e-01
-9.35398936e-01 -9.33390677e-01 -7.11872637e-01 3.27115446e-01
1.87775820e-01 -4.61248726e-01 -5.70683599e-01 1.77647627e-03]
|
[7.99333381652832, 3.4375720024108887]
|
128bc24c-c645-4db4-9cc8-8840459295ac
|
combining-contrastive-and-non-contrastive
|
2211.01964
| null |
https://arxiv.org/abs/2211.01964v1
|
https://arxiv.org/pdf/2211.01964v1.pdf
|
Combining Contrastive and Non-Contrastive Losses for Fine-Tuning Pretrained Models in Speech Analysis
|
Embedding paralinguistic properties is a challenging task as there are only a few hours of training data available for domains such as emotional speech. One solution to this problem is to pretrain a general self-supervised speech representation model on large amounts of unlabeled speech. This pretrained model is then finetuned to a specific task. Paralinguistic properties however have notoriously high class variance, making the finetuning ineffective. In this work, we propose a two step approach to this. First we improve the embedding space, then we train an adapter to bridge the gap from the embedding space to a classification task. In order to improve the class invariance we use a combination of contrastive and non-contrastive losses to explicitly optimize for class invariant, yet discriminative features. Our approach consistently outperforms baselines that are finetuned end-to-end on multiple tasks and surpasses a benchmark on state-of-the-art emotion classification.
|
['Ngoc Thang Vu', 'Ching-Yi Chen', 'Florian Lux']
|
2022-10-21
| null | null | null | null |
['emotion-classification', 'emotion-classification']
|
['computer-vision', 'natural-language-processing']
|
[ 2.85225362e-01 1.91862434e-01 1.09831719e-02 -8.87688696e-01
-1.17229760e+00 -6.33553267e-01 6.32061899e-01 -3.66739370e-02
-5.65002501e-01 4.87031609e-01 3.21016967e-01 -8.01861286e-02
2.62882918e-01 -3.48271012e-01 -4.90240276e-01 -6.14974916e-01
9.98004079e-02 4.70358044e-01 -5.84272854e-02 -2.96757102e-01
-1.28292069e-01 2.78852075e-01 -1.47588217e+00 2.88913012e-01
7.76035607e-01 1.22495532e+00 -4.88888510e-02 6.91932440e-01
-1.54724702e-01 4.44966584e-01 -4.87293571e-01 -3.86960059e-01
1.91318274e-01 -5.23143113e-01 -8.09625685e-01 1.91326514e-01
4.21900958e-01 -5.95059991e-02 -3.64354163e-01 1.05887544e+00
6.12402856e-01 5.88490129e-01 6.10331416e-01 -1.20302749e+00
-5.42713225e-01 3.63255858e-01 -2.23905727e-01 2.75679797e-01
4.91779037e-02 -2.84070432e-01 1.27752531e+00 -1.05463171e+00
3.30680370e-01 1.24918258e+00 4.34425205e-01 8.02612484e-01
-1.21797216e+00 -5.82221091e-01 2.47462809e-01 1.03308514e-01
-1.11477077e+00 -8.28620195e-01 1.05108666e+00 -2.80801833e-01
1.30294156e+00 2.19544888e-01 2.60265231e-01 1.34816706e+00
-1.21347405e-01 7.95200884e-01 1.23093474e+00 -5.21199048e-01
4.07113075e-01 3.67405057e-01 2.08591111e-02 5.67628443e-01
-6.40170097e-01 2.57841408e-01 -5.17733455e-01 -1.62541613e-01
3.65561545e-01 -2.11126208e-01 -1.07293382e-01 -4.42632824e-01
-9.50560868e-01 1.01931369e+00 3.78518015e-01 3.56294781e-01
-2.27134824e-01 -1.13147236e-01 8.18090856e-01 6.29245877e-01
7.97978580e-01 6.20427787e-01 -7.27255285e-01 -5.10063052e-01
-8.74023139e-01 -2.30781183e-01 7.80157506e-01 4.39809203e-01
8.97362769e-01 1.42230451e-01 -6.77683055e-02 1.41883934e+00
7.24411011e-02 1.60943434e-01 7.89426982e-01 -7.79412806e-01
4.31978613e-01 2.59725422e-01 -3.51057023e-01 -6.29208684e-01
-1.94341004e-01 -4.69377100e-01 -7.09011614e-01 1.29227713e-01
1.44686967e-01 -1.74638331e-01 -9.88109708e-01 1.93489945e+00
3.25677365e-01 5.73338084e-02 2.64911711e-01 8.87223840e-01
4.88874376e-01 9.30906057e-01 1.03302278e-01 -6.07812963e-02
1.38352478e+00 -1.45868862e+00 -7.95594215e-01 -5.33493459e-01
4.16704029e-01 -6.80982947e-01 1.44535518e+00 3.45438063e-01
-8.53119075e-01 -4.90283310e-01 -1.08098972e+00 -1.89008296e-01
-5.25958836e-01 1.44415781e-01 5.10930479e-01 5.68538070e-01
-7.73779035e-01 5.89229167e-01 -9.55774665e-01 -1.85945183e-01
3.12189937e-01 3.16403896e-01 -6.26062155e-01 8.89766682e-03
-1.37492287e+00 1.00345802e+00 4.54033285e-01 -3.92960012e-02
-7.18285322e-01 -5.66842020e-01 -1.13326156e+00 2.05347076e-01
4.13082838e-01 -3.68710935e-01 1.27512956e+00 -1.26189470e+00
-2.13477707e+00 9.62455034e-01 -1.23655804e-01 -2.83658177e-01
2.52550960e-01 -1.30418852e-01 -5.55909991e-01 2.21039176e-01
-9.58845168e-02 6.54745102e-01 1.14245546e+00 -1.12127161e+00
-4.14701641e-01 -2.86832124e-01 -5.38636521e-02 2.10358158e-01
-8.58157456e-01 2.08048478e-01 -3.38658214e-01 -8.13343942e-01
-8.08143914e-02 -9.91690278e-01 -1.63285270e-01 -1.75397426e-01
-2.02185795e-01 -4.00357425e-01 1.02875364e+00 -5.55703759e-01
9.68610168e-01 -2.37642384e+00 4.11979526e-01 6.64703771e-02
-5.80754541e-02 3.43780428e-01 -4.39882189e-01 3.47427696e-01
-2.96951383e-01 -4.43191342e-02 -3.70293856e-01 -7.58592427e-01
2.77215064e-01 4.08005148e-01 -4.14423525e-01 2.41783440e-01
5.09373486e-01 6.00253761e-01 -9.54730392e-01 -1.91136092e-01
2.20125496e-01 6.87058926e-01 -5.75690150e-01 4.69192624e-01
-1.15890637e-01 3.68206918e-01 -2.10934475e-01 3.36788386e-01
5.38332582e-01 -2.10707895e-02 2.64945235e-02 -1.59941822e-01
2.70963162e-01 7.48641253e-01 -8.68907809e-01 1.79918242e+00
-9.30321932e-01 5.99497080e-01 1.25852272e-01 -1.51656282e+00
1.01973510e+00 4.57896829e-01 3.65326136e-01 -4.17028904e-01
1.22714862e-01 2.56369531e-01 1.44231191e-03 -4.07502711e-01
2.40268201e-01 -6.20931149e-01 -2.64905274e-01 2.92542040e-01
5.96853614e-01 -3.17099392e-01 -9.41826701e-02 2.61372030e-02
1.16396618e+00 6.19975058e-03 1.64404154e-01 -1.13451496e-01
5.28160095e-01 -3.72172356e-01 6.76772296e-01 2.22139105e-01
-2.26989195e-01 6.01143837e-01 5.68244457e-01 -2.46895745e-01
-9.57970679e-01 -9.40298438e-01 -1.86136559e-01 1.30920255e+00
-3.08247209e-01 -5.37515581e-01 -7.44527459e-01 -1.16777980e+00
-1.84000239e-01 6.31275117e-01 -7.43038535e-01 -6.15305364e-01
-3.32772732e-01 -5.06874621e-01 5.08846819e-01 6.39880836e-01
4.01651204e-01 -1.09259510e+00 -2.52313167e-01 2.29874283e-01
-1.37796521e-01 -1.39411438e+00 -5.80429077e-01 7.53882825e-01
-4.71512228e-01 -3.33000213e-01 -5.96298516e-01 -7.03912795e-01
6.67240977e-01 -1.41085088e-01 1.25248730e+00 -1.80961803e-01
-8.43937248e-02 3.11963350e-01 -4.80364323e-01 -2.71288961e-01
-5.07595241e-01 1.79586485e-01 8.24526772e-02 2.75845885e-01
5.32673538e-01 -5.61641753e-01 -3.32347900e-01 1.22605868e-01
-7.82112122e-01 -4.76384342e-01 5.40247262e-01 1.34921741e+00
5.37059128e-01 1.92721069e-01 6.47973955e-01 -6.47725523e-01
6.67771459e-01 -3.07605535e-01 -3.72050256e-01 5.59950918e-02
-4.00027037e-01 2.35726699e-01 8.92376006e-01 -6.21868312e-01
-8.25157881e-01 2.97182530e-01 -3.35029870e-01 -7.51147568e-01
-2.66653657e-01 4.65214789e-01 -3.36164027e-01 8.78834501e-02
3.69912088e-01 -7.07206056e-02 9.51353833e-02 -4.14216042e-01
3.19947302e-01 8.81873608e-01 4.32660133e-01 -6.66281164e-01
7.29894996e-01 1.26214713e-01 -3.87744397e-01 -1.07227743e+00
-1.01987100e+00 -5.33253253e-01 -5.48151433e-01 1.63714200e-01
7.86905646e-01 -8.74380469e-01 -2.41962448e-01 1.19548984e-01
-9.76253927e-01 -4.97726411e-01 -4.04322803e-01 6.40160799e-01
-5.19359887e-01 2.89438725e-01 -5.88633120e-01 -8.22407544e-01
-3.04703623e-01 -1.03077066e+00 1.28253746e+00 -7.24929646e-02
-2.66557753e-01 -1.04437244e+00 3.09602797e-01 3.37823033e-01
4.76370901e-01 -1.21237017e-01 7.62650490e-01 -8.51401389e-01
1.29102841e-01 -2.53692061e-01 -7.64184818e-02 9.93049145e-01
2.08947599e-01 -8.55815932e-02 -1.31431472e+00 -3.77792448e-01
8.27814862e-02 -7.92712629e-01 1.04560912e+00 -3.62288840e-02
1.27047229e+00 -2.40898639e-01 -5.43378806e-03 5.64605653e-01
9.76273239e-01 7.32969344e-02 5.24958074e-01 1.71645626e-01
5.11586785e-01 7.33288646e-01 5.37830710e-01 2.41900250e-01
3.44917148e-01 9.36106563e-01 1.50085077e-01 -1.58788323e-01
2.81107835e-02 -2.74778694e-01 5.41773558e-01 1.18042493e+00
3.31108540e-01 -1.70184299e-01 -6.63861156e-01 6.27977610e-01
-1.65264738e+00 -8.40891600e-01 6.68225169e-01 2.07826471e+00
9.85797346e-01 1.20234981e-01 1.72545999e-01 2.20283866e-01
4.32518840e-01 3.09316695e-01 -4.37672377e-01 -6.49222016e-01
5.19437380e-02 3.18459719e-01 -5.42276502e-02 5.68366528e-01
-1.36084139e+00 1.25250530e+00 5.66879416e+00 7.73925364e-01
-1.61481977e+00 3.12164277e-01 6.16854548e-01 7.91638866e-02
-2.86698699e-01 -1.84358545e-02 -5.05470037e-01 3.36762130e-01
1.11411119e+00 1.50505632e-01 6.21391714e-01 9.78565514e-01
-1.02399036e-01 3.05895120e-01 -1.27268600e+00 1.08955526e+00
2.71351397e-01 -8.57183874e-01 -4.01539296e-01 -1.29697934e-01
5.26922464e-01 1.71593457e-01 4.20495123e-02 8.24071169e-01
2.52487659e-01 -9.37264740e-01 4.48553681e-01 -4.23754603e-02
7.23281682e-01 -7.10050941e-01 6.22497976e-01 1.68504879e-01
-8.31438839e-01 3.85411493e-02 -5.12165666e-01 1.00656793e-01
1.89413935e-01 4.91883188e-01 -7.14382231e-01 2.20551163e-01
7.49171138e-01 5.02019286e-01 -2.72865951e-01 6.37447417e-01
-3.03317755e-01 5.73666394e-01 -3.19155097e-01 9.92922764e-03
3.78678828e-01 -9.24523920e-02 4.29323792e-01 1.34518528e+00
2.29315192e-01 -2.68646833e-02 2.25964263e-01 5.53290963e-01
-3.00869286e-01 1.73463643e-01 -4.79584992e-01 -4.35360461e-01
3.30528110e-01 1.47080994e+00 -4.30867225e-01 -3.73876840e-01
-4.79431659e-01 1.48467672e+00 8.01330388e-01 3.62917721e-01
-6.80488825e-01 -5.58704436e-01 8.83893192e-01 -3.72661620e-01
5.68896174e-01 -1.99523017e-01 1.11539021e-01 -1.48511219e+00
1.14662305e-01 -1.09518528e+00 3.05725813e-01 -4.48028207e-01
-1.56848729e+00 1.04308152e+00 -3.15916836e-01 -1.07750285e+00
-5.60324550e-01 -7.69354880e-01 -6.46016717e-01 7.09224284e-01
-1.66502833e+00 -1.13610744e+00 -7.60054737e-02 5.21141410e-01
6.60151362e-01 -3.53381276e-01 1.21832454e+00 3.27218980e-01
-6.38552845e-01 8.59317958e-01 -2.77979448e-02 1.41188264e-01
8.99894953e-01 -1.44479191e+00 1.95384040e-01 6.75557733e-01
4.18008715e-01 3.79155427e-01 7.52349198e-01 -1.86953843e-02
-1.15191543e+00 -7.97183156e-01 1.00742126e+00 -3.52051020e-01
8.16534936e-01 -9.61273015e-01 -1.16965330e+00 5.06408036e-01
3.37233633e-01 5.16747177e-01 8.74480903e-01 4.69162852e-01
-8.42287362e-01 -2.87501335e-01 -8.86796236e-01 4.03503269e-01
7.22866178e-01 -9.68639731e-01 -6.66180193e-01 3.08393925e-01
7.78283656e-01 -2.35698804e-01 -8.26056421e-01 3.40789944e-01
4.11968201e-01 -5.96583128e-01 8.03267181e-01 -8.71715128e-01
3.97105455e-01 3.30231152e-02 -3.80781144e-01 -1.85278726e+00
-1.05003826e-01 -9.21095848e-01 -1.18139917e-02 1.39483035e+00
6.06517255e-01 -6.31138206e-01 5.74815750e-01 5.93583584e-01
-3.85445535e-01 -8.82797241e-01 -1.09394479e+00 -9.73390877e-01
2.89868057e-01 -2.85911620e-01 2.98685968e-01 1.06582355e+00
3.35220098e-01 8.27156723e-01 -3.60263199e-01 2.86433622e-02
3.59175205e-01 4.08187769e-02 6.62286043e-01 -1.07047141e+00
-3.99331957e-01 -4.74600047e-01 -5.33805907e-01 -9.30367410e-01
8.14485788e-01 -8.45038176e-01 1.65480658e-01 -9.96796310e-01
-4.15892266e-02 -5.10695517e-01 -6.26367331e-01 6.24353647e-01
-2.11068973e-01 1.08794332e-01 4.51203138e-02 -2.51653016e-01
-5.14663458e-01 1.00850964e+00 8.07152212e-01 -1.83659211e-01
-1.84645981e-01 -2.01341763e-01 -5.13760269e-01 4.44193661e-01
8.49359274e-01 -5.33051729e-01 -5.52194417e-01 -3.48272592e-01
-1.60709750e-02 -1.21888243e-01 1.66555569e-01 -7.43082166e-01
-3.42168324e-02 3.12332716e-02 5.62960692e-02 -1.29225001e-01
8.63150239e-01 -8.35308433e-01 -5.83919883e-01 -6.69158176e-02
-5.79508543e-01 -2.41079032e-01 3.00170630e-01 5.35785377e-01
-7.41318047e-01 -1.73943907e-01 1.01143217e+00 2.13987723e-01
-6.46632016e-01 3.34718496e-01 -2.29208350e-01 3.47085685e-01
8.06805849e-01 2.13139325e-01 -1.12290002e-01 -5.39818227e-01
-6.60090446e-01 -4.08763997e-02 1.27603769e-01 7.85515368e-01
4.79825884e-01 -1.43561769e+00 -6.34283960e-01 4.05133516e-01
3.26718897e-01 -3.82380456e-01 2.94932693e-01 6.98357522e-01
2.06178993e-01 2.17731610e-01 -1.42203256e-01 -3.90706003e-01
-1.28468192e+00 6.62128270e-01 4.18903083e-01 -2.65399218e-01
-5.57040989e-01 1.01425314e+00 3.42383504e-01 -9.29522932e-01
3.57833326e-01 -2.32517958e-01 5.07183671e-02 6.82603866e-02
4.19203997e-01 -1.74609601e-01 3.46960604e-01 -8.71379316e-01
-5.08662820e-01 5.13647974e-01 -2.27958009e-01 -4.19644028e-01
1.44878721e+00 -1.24241158e-01 1.32512540e-01 5.50609231e-01
1.89043653e+00 7.23374113e-02 -1.43861628e+00 -3.67508769e-01
-1.59959570e-01 -3.15070599e-01 3.03951025e-01 -7.21609235e-01
-1.03898537e+00 1.17241120e+00 4.05100793e-01 3.56574506e-01
1.20065212e+00 1.72266990e-01 8.60359251e-01 4.07692820e-01
-1.30536169e-01 -1.45645344e+00 4.33033347e-01 7.76309431e-01
1.06071711e+00 -1.55792344e+00 -4.78990883e-01 -3.77074838e-01
-8.84436309e-01 1.12325561e+00 3.46772760e-01 -3.26519549e-01
7.39547253e-01 3.12977970e-01 3.06283265e-01 -1.77968249e-01
-9.71493065e-01 -1.45937160e-01 4.62840825e-01 4.25727993e-01
5.38296878e-01 4.93878722e-02 8.30689743e-02 6.40547156e-01
-3.51902038e-01 -2.21255437e-01 -9.35499556e-03 6.47107840e-01
-1.41136318e-01 -1.28438544e+00 1.47868656e-02 1.64039344e-01
-6.08646750e-01 -2.97265146e-02 -4.35290933e-01 4.54170853e-01
-2.43439540e-01 1.04343998e+00 8.63574967e-02 -4.54231918e-01
4.04450059e-01 3.19810539e-01 3.33788186e-01 -6.73242927e-01
-3.68284792e-01 2.71139652e-01 3.40165377e-01 -4.90544349e-01
-1.71797916e-01 -5.85492790e-01 -1.02230644e+00 2.43901297e-01
-3.55949283e-01 3.28323603e-01 8.87750626e-01 9.07624245e-01
3.71948689e-01 6.20045543e-01 9.44258153e-01 -9.00046587e-01
-9.49818552e-01 -9.78596687e-01 -4.51708525e-01 6.55156553e-01
5.04512966e-01 -7.34810174e-01 -7.52160311e-01 -1.49110183e-01]
|
[13.79356575012207, 5.946377277374268]
|
991ff2e1-241e-4815-bdf9-3939603caa9e
|
self-supervised-facial-action-unit-detection
|
2303.05708
| null |
https://arxiv.org/abs/2303.05708v1
|
https://arxiv.org/pdf/2303.05708v1.pdf
|
Self-supervised Facial Action Unit Detection with Region and Relation Learning
|
Facial action unit (AU) detection is a challenging task due to the scarcity of manual annotations. Recent works on AU detection with self-supervised learning have emerged to address this problem, aiming to learn meaningful AU representations from numerous unlabeled data. However, most existing AU detection works with self-supervised learning utilize global facial features only, while AU-related properties such as locality and relevance are not fully explored. In this paper, we propose a novel self-supervised framework for AU detection with the region and relation learning. In particular, AU related attention map is utilized to guide the model to focus more on AU-specific regions to enhance the integrity of AU local features. Meanwhile, an improved Optimal Transport (OT) algorithm is introduced to exploit the correlation characteristics among AUs. In addition, Swin Transformer is exploited to model the long-distance dependencies within each AU region during feature learning. The evaluation results on BP4D and DISFA demonstrate that our proposed method is comparable or even superior to the state-of-the-art self-supervised learning methods and supervised AU detection methods.
|
['Zhilei Liu', 'Juan Song']
|
2023-03-10
| null | null | null | null |
['action-unit-detection', 'facial-action-unit-detection']
|
['computer-vision', 'computer-vision']
|
[ 2.10068852e-01 1.11625984e-01 -3.15068841e-01 -3.79645795e-01
-7.17220187e-01 2.36840406e-03 3.40493411e-01 3.17903459e-02
-1.88702986e-01 4.23881978e-01 2.91196167e-01 5.60111225e-01
2.12927774e-01 -6.65183783e-01 -5.15083969e-01 -1.07868099e+00
-2.32957061e-02 -7.85060301e-02 4.54986513e-01 -3.29309195e-01
8.20013974e-03 4.67560858e-01 -1.79325438e+00 2.30730951e-01
7.34951556e-01 1.48149955e+00 -6.00098353e-03 2.14430154e-03
1.13816010e-02 7.39045739e-01 -1.63568586e-01 -1.83338970e-02
2.60601074e-01 -6.25092685e-01 -5.32217681e-01 2.57943988e-01
4.07719016e-01 -5.70906818e-01 -4.50036407e-01 1.08673298e+00
6.11861467e-01 -1.71949563e-03 9.36512411e-01 -1.12984991e+00
-4.83636260e-01 3.02005261e-01 -1.03448594e+00 3.49739999e-01
1.70756355e-01 1.61382616e-01 1.23008478e+00 -1.07515585e+00
3.57983708e-01 1.15675783e+00 2.85784364e-01 7.27367759e-01
-7.58825362e-01 -9.89101231e-01 2.90019393e-01 4.71917361e-01
-1.75062084e+00 -5.35303831e-01 1.00525081e+00 -1.56860486e-01
7.52776086e-01 -1.85414538e-01 6.68075204e-01 9.15611506e-01
-1.08334176e-01 1.16705978e+00 8.12729180e-01 -4.95643437e-01
-7.84259755e-03 1.24699093e-01 -2.96308715e-02 1.32481241e+00
-2.36178696e-01 1.69407502e-01 -6.88916624e-01 -9.52969417e-02
6.81129158e-01 -5.93315512e-02 7.55154118e-02 -1.92556784e-01
-3.84363472e-01 7.52155364e-01 7.53559053e-01 3.45976084e-01
-1.80724680e-01 -7.46133700e-02 4.72012430e-01 -4.97966483e-02
8.04270744e-01 -1.51254147e-01 -6.41855001e-02 2.67676413e-02
-5.35984278e-01 -2.21546352e-01 7.31741786e-02 9.31008756e-01
1.27795267e+00 5.72135150e-02 -4.29000258e-01 8.84479642e-01
4.51398730e-01 2.65802026e-01 3.70550334e-01 -4.74687368e-01
3.03961843e-01 1.13711607e+00 -1.97260141e-01 -1.01171803e+00
-5.73718429e-01 -2.73846269e-01 -6.43523037e-01 2.14734763e-01
1.86350733e-01 -2.57303447e-01 -5.85125566e-01 1.57959116e+00
5.41528285e-01 4.88266945e-01 6.08238801e-02 1.03054321e+00
1.14834726e+00 4.99933273e-01 7.65952766e-02 -4.19652522e-01
1.14816260e+00 -1.16158736e+00 -5.86736977e-01 -1.93000481e-01
1.01039243e+00 -5.15772223e-01 7.48099983e-01 3.61718610e-02
-7.75444210e-01 -7.15101540e-01 -8.34930480e-01 1.04508519e-01
-1.59058869e-01 5.17319024e-01 5.92473805e-01 6.42716587e-01
-6.60580277e-01 1.64203778e-01 -6.46095395e-01 -5.40069878e-01
1.01621914e+00 3.89187425e-01 -3.93243700e-01 -1.54638916e-01
-1.20441794e+00 5.10832667e-01 3.09630960e-01 4.36729312e-01
-1.07019770e+00 -3.00215065e-01 -1.07145536e+00 -1.11461341e-01
3.97786021e-01 6.34002090e-02 8.47589195e-01 -1.18577981e+00
-1.39166093e+00 9.17053640e-01 -2.94309258e-01 -2.15078264e-01
1.48495033e-01 -1.39221057e-01 -3.02119166e-01 4.00522679e-01
5.24959266e-02 7.88486719e-01 1.01314354e+00 -1.02807748e+00
-9.25329268e-01 -6.30630016e-01 -6.80292770e-02 5.92762053e-01
-9.23397064e-01 3.03253502e-01 -5.10550439e-01 -3.49685937e-01
3.55164409e-02 -6.70173347e-01 1.16263017e-01 3.75525922e-01
-1.90780610e-01 -8.00183356e-01 1.20151365e+00 -1.84071973e-01
1.30393815e+00 -2.26601171e+00 4.69456688e-02 2.06589699e-01
3.23533326e-01 3.04251403e-01 -2.84161776e-01 2.19308123e-01
1.76683202e-01 -5.19962788e-01 -9.59046483e-02 -4.92538393e-01
-3.94817770e-01 2.30530143e-01 2.72939414e-01 8.93010795e-01
6.71944022e-01 7.58189142e-01 -9.07395065e-01 -9.99887168e-01
2.37765029e-01 1.98892996e-01 -3.80493313e-01 4.37393337e-01
-5.80475368e-02 4.78102326e-01 -7.47356355e-01 1.14673579e+00
6.26245677e-01 -6.15890883e-03 -3.10506940e-01 -3.72679949e-01
-8.43573436e-02 -3.18039596e-01 -9.09443259e-01 1.73812425e+00
-2.69627869e-01 4.14650142e-01 6.26002252e-02 -1.02536607e+00
1.19362795e+00 1.43990576e-01 7.48014152e-01 -7.17480659e-01
5.56433916e-01 1.49735697e-02 7.28041492e-03 -6.45037413e-01
-1.98740829e-02 1.70183808e-01 4.11338687e-01 4.84081686e-01
8.47780630e-02 4.44715500e-01 6.98947115e-04 4.99169566e-02
9.08251226e-01 3.54056507e-01 3.80832165e-01 -2.08138004e-01
8.02455306e-01 -2.52604723e-01 6.36583388e-01 1.81720644e-01
-6.21113300e-01 5.03656507e-01 3.31713915e-01 -3.66662592e-01
-3.29422623e-01 -7.18164384e-01 -3.28445196e-01 1.23263764e+00
4.90785033e-01 -4.39172179e-01 -1.01851618e+00 -1.15806329e+00
-1.16311491e-01 8.91257152e-02 -9.73138809e-01 -5.63632131e-01
-2.23733768e-01 -7.76498258e-01 5.68236768e-01 7.19207644e-01
7.76186347e-01 -1.15859485e+00 -4.83291209e-01 6.65428266e-02
1.03951141e-01 -1.11317849e+00 -5.27044952e-01 -1.09485872e-01
-5.51729739e-01 -1.22359061e+00 -7.26259530e-01 -1.03085482e+00
9.47157741e-01 4.73614186e-01 3.09260249e-01 1.84714407e-01
-7.02453494e-01 2.34760150e-01 -6.65755928e-01 -4.35448110e-01
-1.83057459e-03 -2.40699053e-02 2.80325741e-01 8.09944332e-01
8.87023151e-01 -3.11516076e-01 -8.08270633e-01 5.77982843e-01
-5.98269403e-01 -2.83416778e-01 7.01821625e-01 9.34478283e-01
6.09636843e-01 -1.34481266e-01 5.62407315e-01 -6.15241885e-01
2.43799910e-01 -4.25458431e-01 -3.17203373e-01 2.16757923e-01
-4.13727552e-01 -1.98572829e-01 3.56188029e-01 -3.31667364e-01
-1.24167311e+00 4.45279390e-01 -1.72300935e-01 -5.30384421e-01
-3.26977819e-01 -9.25726909e-03 -4.08250928e-01 -5.53028166e-01
7.50145972e-01 2.93499798e-01 2.99970269e-01 -1.26232296e-01
1.73819259e-01 1.02769303e+00 -4.38531451e-02 -4.39657152e-01
5.60252607e-01 8.09473276e-01 -1.89163256e-02 -1.03028882e+00
-1.23888671e+00 -8.39961886e-01 -9.88817155e-01 -4.55163658e-01
8.79139125e-01 -1.08094358e+00 -5.92306376e-01 5.66870332e-01
-9.10475910e-01 -2.27371350e-01 -1.69118255e-01 3.41032445e-01
-4.75670278e-01 5.33471048e-01 -6.27124846e-01 -1.08264267e+00
-5.70972502e-01 -9.69571769e-01 1.46190608e+00 4.40363884e-01
6.68629706e-02 -7.46259749e-01 -1.03554107e-01 2.89145291e-01
-1.49342055e-02 4.76844283e-03 3.33241552e-01 -3.20287466e-01
-2.21361116e-01 -3.10889870e-01 -8.09477031e-01 5.07102787e-01
6.45270944e-01 -9.64946151e-02 -1.13688529e+00 -1.66655630e-01
-2.26004899e-01 -7.07602203e-01 8.06994200e-01 7.13454559e-02
1.10895860e+00 -1.81990370e-01 -4.93851572e-01 4.69521105e-01
9.10409808e-01 -8.63399059e-02 4.54815924e-01 -5.78688309e-02
8.91191602e-01 7.90007293e-01 1.23605251e+00 7.01674879e-01
1.19998388e-01 7.16409981e-01 7.20983565e-01 -2.94506133e-01
-4.87148724e-02 -1.76490471e-01 4.42095786e-01 1.73974752e-01
-2.63085067e-01 2.62987256e-01 -5.40992141e-01 4.08888787e-01
-1.96356213e+00 -8.08879912e-01 9.12869722e-02 1.76645744e+00
6.02101982e-01 7.78128952e-02 4.16328788e-01 -7.48177171e-02
5.92715561e-01 2.71834403e-01 -5.90354085e-01 3.55706923e-02
-2.33295664e-01 -9.01471153e-02 2.77440131e-01 1.76865920e-01
-1.32629919e+00 1.42067671e+00 5.49891043e+00 1.10943532e+00
-1.00769889e+00 3.03757936e-01 6.10447526e-01 3.45799699e-02
2.71890134e-01 -3.88387710e-01 -1.21304691e+00 1.65002704e-01
1.69351742e-01 1.84637219e-01 -1.04156353e-01 1.00107813e+00
3.68345261e-01 -8.25385451e-02 -8.82644355e-01 1.19244838e+00
6.07017994e-01 -8.95094097e-01 -5.80182709e-02 -2.19522025e-02
7.88146973e-01 -1.02897711e-01 -5.12539372e-02 4.61682789e-02
-9.11063775e-02 -8.44247222e-01 2.55930543e-01 3.31742883e-01
8.97400439e-01 -8.91621828e-01 8.26130509e-01 1.71476021e-01
-1.58146620e+00 -3.34100544e-01 -6.72336400e-01 -1.79090902e-01
-2.08878249e-01 5.05072176e-02 -6.39653385e-01 2.33506098e-01
8.38954210e-01 1.38004649e+00 -7.41509080e-01 7.03880012e-01
-4.38626409e-01 6.36857748e-01 -4.04330164e-01 -2.25518912e-01
5.58106601e-01 -3.30169261e-01 2.68723279e-01 1.03739083e+00
4.59191352e-02 2.69047171e-01 2.41096407e-01 7.20975518e-01
1.38514161e-01 7.11878777e-01 -6.47504747e-01 2.22820386e-01
-5.59504107e-02 1.47254574e+00 -4.63168412e-01 1.48207709e-01
-7.86291897e-01 1.15595472e+00 7.22971916e-01 7.43235573e-02
-7.13914394e-01 -3.82831842e-02 7.03112364e-01 2.11836219e-01
1.53781235e-01 4.80052233e-02 2.97768801e-01 -8.14940691e-01
-1.43192679e-01 -4.40609425e-01 6.95192933e-01 -5.19185901e-01
-1.37272656e+00 6.18310273e-01 -1.77198574e-01 -1.56911027e+00
7.39271343e-02 -5.82571566e-01 -7.36684918e-01 3.78455281e-01
-1.73433399e+00 -1.52592647e+00 -4.48039263e-01 9.19468343e-01
5.76278627e-01 -4.22784507e-01 7.86445141e-01 2.11248308e-01
-8.56815755e-01 1.08423185e+00 -8.49623159e-02 4.55445498e-01
7.21664548e-01 -6.63688600e-01 -3.24451029e-01 6.94151700e-01
2.78016150e-01 1.06257774e-01 1.45402089e-01 -5.11942148e-01
-1.20799458e+00 -1.22582138e+00 2.22083732e-01 -1.79157630e-01
7.59482741e-01 -3.77224326e-01 -7.06576467e-01 3.69109571e-01
-1.28344567e-02 7.57565856e-01 6.21455371e-01 -4.27467413e-02
-3.94697160e-01 -5.77441871e-01 -9.96647835e-01 5.57521641e-01
1.46230972e+00 -4.03443635e-01 1.02761444e-02 4.74511355e-01
2.80486733e-01 -1.84936702e-01 -5.94627261e-01 4.15815562e-01
4.25072342e-01 -9.29657340e-01 4.11020309e-01 -4.73575354e-01
1.90799132e-01 -2.51550883e-01 -7.88534526e-03 -9.72104311e-01
-1.22892931e-01 -6.46459341e-01 -2.72577167e-01 1.38884079e+00
1.86447755e-01 -4.25069988e-01 1.27359819e+00 1.00115642e-01
-4.32685763e-02 -9.97353077e-01 -1.04991364e+00 -4.98957962e-01
-3.90603632e-01 -2.58371800e-01 2.02792257e-01 6.32609725e-01
2.32108548e-01 4.57263976e-01 -4.32947129e-01 2.85655111e-01
7.99776793e-01 1.35490403e-01 6.82723522e-01 -1.04432249e+00
1.01014368e-01 -3.23242307e-01 -8.76115322e-01 -1.07723534e+00
4.58219826e-01 -6.63301110e-01 1.99796692e-01 -1.32504618e+00
2.69234866e-01 -5.42573869e-01 -3.04651499e-01 6.80014312e-01
-2.48074874e-01 5.31434178e-01 -2.49939486e-01 1.82555586e-01
-9.27908361e-01 1.07448304e+00 1.23025024e+00 -2.71938294e-01
-2.63701439e-01 1.19007371e-01 -4.22869980e-01 7.59152114e-01
6.94897354e-01 -2.02965096e-01 -3.67328078e-01 -8.53460208e-02
-2.99855232e-01 -3.04957926e-01 2.27755904e-01 -1.14612901e+00
2.92146504e-01 -5.42592593e-02 1.38237596e-01 -4.36236471e-01
4.51504469e-01 -9.46952283e-01 -9.14906442e-01 3.42224956e-01
-8.35337862e-02 -4.90863830e-01 1.38410136e-01 7.31735587e-01
-4.07455713e-01 -1.48860440e-01 9.14675534e-01 -2.40761451e-02
-1.12208891e+00 8.77100766e-01 -2.09068775e-01 -2.05561236e-01
1.51977479e+00 -2.88606256e-01 1.20531227e-02 -3.54102075e-01
-5.33353806e-01 3.94239515e-01 1.13695860e-01 5.66514969e-01
9.84436452e-01 -1.34840953e+00 -8.70756447e-01 3.91915411e-01
8.75161350e-01 8.66535008e-02 5.22982001e-01 8.61099005e-01
-2.01907113e-01 4.82900143e-02 -2.12815836e-01 -7.88971066e-01
-1.64135218e+00 4.99325722e-01 4.17171925e-01 2.50276625e-01
-5.28341413e-01 1.05692852e+00 5.19580305e-01 -4.17852551e-02
3.94145548e-01 3.82942647e-01 -6.51308894e-01 2.82834947e-01
8.10401440e-01 1.55453742e-01 -1.30808800e-01 -1.20483267e+00
-2.97431886e-01 8.72856319e-01 -2.58815736e-01 5.45512199e-01
1.03943324e+00 -3.05043310e-01 -8.72310326e-02 1.10499546e-01
1.34500372e+00 -1.98422328e-01 -1.56638443e+00 -6.77686691e-01
-3.11995536e-01 -4.36854303e-01 1.09762691e-01 4.83804476e-03
-1.28388774e+00 1.02194381e+00 9.09633994e-01 -3.52550000e-01
1.22133160e+00 3.33617061e-01 6.34950578e-01 2.48935819e-01
4.28663760e-01 -1.15769434e+00 5.53248227e-01 1.73907325e-01
6.53506219e-01 -1.59131277e+00 -7.86148459e-02 -7.74673760e-01
-7.62198508e-01 1.07234490e+00 1.35039723e+00 -1.31752104e-01
8.62089097e-01 1.44499913e-01 1.87381938e-01 -2.91982800e-01
-2.49856547e-01 -8.74238312e-01 3.72567564e-01 6.54361963e-01
2.37685397e-01 -4.64671075e-01 -2.58511484e-01 4.06890899e-01
4.11363870e-01 -2.33471468e-01 1.22699095e-03 7.77953804e-01
-6.24805570e-01 -7.64916360e-01 -1.53376162e-01 5.18342495e-01
-1.32130817e-01 2.04622913e-02 -4.65205193e-01 6.70543849e-01
2.14120299e-01 9.90309775e-01 1.73434511e-01 -5.10736644e-01
1.00318380e-01 -1.95298716e-01 5.19494534e-01 -9.34657872e-01
-1.85093999e-01 2.24040493e-01 -1.83914274e-01 -6.77470148e-01
-6.91935480e-01 -7.22817183e-01 -1.41764510e+00 3.06105018e-01
-6.34283662e-01 1.02265440e-02 1.55349210e-01 1.09468281e+00
2.53667057e-01 1.35057136e-01 9.79009807e-01 -7.65831530e-01
-1.77565053e-01 -1.14629364e+00 -8.94362748e-01 3.66537482e-01
1.12804398e-01 -9.83114481e-01 -2.54122943e-01 -3.20374936e-01]
|
[13.662312507629395, 1.5478315353393555]
|
3e9bcf76-24f1-43df-8041-f5296bc01419
|
look-beneath-the-surface-exploiting
|
2306.04220
| null |
https://arxiv.org/abs/2306.04220v3
|
https://arxiv.org/pdf/2306.04220v3.pdf
|
Look Beneath the Surface: Exploiting Fundamental Symmetry for Sample-Efficient Offline RL
|
Offline reinforcement learning (RL) offers an appealing approach to real-world tasks by learning policies from pre-collected datasets without interacting with the environment. However, the performance of existing offline RL algorithms heavily depends on the scale and state-action space coverage of datasets. Real-world data collection is often expensive and uncontrollable, leading to small and narrowly covered datasets and posing significant challenges for practical deployments of offline RL. In this paper, we provide a new insight that leveraging the fundamental symmetry of system dynamics can substantially enhance offline RL performance under small datasets. Specifically, we propose a Time-reversal symmetry (T-symmetry) enforced Dynamics Model (TDM), which establishes consistency between a pair of forward and reverse latent dynamics. TDM provides both well-behaved representations for small datasets and a new reliability measure for OOD samples based on compliance with the T-symmetry. These can be readily used to construct a new offline RL algorithm (TSRL) with less conservative policy constraints and a reliable latent space data augmentation procedure. Based on extensive experiments, we find TSRL achieves great performance on small benchmark datasets with as few as 1% of the original samples, which significantly outperforms the recent offline RL algorithms in terms of data efficiency and generalizability.
|
['Li Jiang', 'Youfang Lin', 'Han Wang', 'Shoucheng Song', 'Wenjia Zhang', 'Zhihao Wu', 'Xianyuan Zhan', 'Peng Cheng']
|
2023-06-07
| null | null | null | null |
['offline-rl']
|
['playing-games']
|
[-1.22119859e-01 -6.93058819e-02 -6.54722154e-01 1.06019378e-01
-5.51958740e-01 -6.93565488e-01 6.32466137e-01 -5.40222526e-02
-2.36026078e-01 9.09496546e-01 1.59447566e-01 -4.32038307e-01
-4.01244491e-01 -4.46962297e-01 -7.18203545e-01 -8.75257134e-01
-4.72943097e-01 3.13168943e-01 -1.28491849e-01 -2.41242319e-01
-4.76773120e-02 4.12446082e-01 -1.40019393e+00 -3.50843787e-01
8.55009913e-01 9.89801526e-01 2.13995278e-01 3.86713117e-01
3.37997437e-01 8.65728498e-01 -5.13352394e-01 5.05941093e-01
6.57587290e-01 -3.32547694e-01 -4.33603555e-01 7.77775794e-02
-2.11988375e-01 -6.81039989e-01 -6.99090481e-01 6.39689982e-01
5.04277468e-01 4.02982831e-01 1.98521510e-01 -1.49169743e+00
-4.57087398e-01 4.17212784e-01 -4.37992871e-01 3.24272737e-02
2.78364420e-01 5.43357968e-01 8.28066826e-01 -3.54334801e-01
4.63025182e-01 1.22312534e+00 4.41650540e-01 5.98490953e-01
-1.43369949e+00 -8.73561144e-01 7.05820441e-01 2.37532239e-02
-1.00571275e+00 -4.72275764e-01 7.77020395e-01 -2.34403223e-01
8.60516310e-01 1.07982770e-01 6.91726208e-01 1.42788482e+00
7.18318895e-02 9.62444425e-01 1.36711717e+00 -6.88694790e-02
6.58142626e-01 -1.00129105e-01 -1.62014648e-01 5.54285705e-01
2.47896358e-01 5.02566457e-01 -5.11392057e-01 -2.16048807e-01
1.05925608e+00 8.70300233e-02 -8.34451318e-02 -7.37578392e-01
-1.24163675e+00 7.59842157e-01 1.93880230e-01 -1.13410160e-01
-3.45823348e-01 2.29273140e-01 6.01807892e-01 5.63723326e-01
1.55494854e-01 5.31326056e-01 -4.92057770e-01 -5.11453986e-01
-3.03517461e-01 3.66243243e-01 6.03197217e-01 1.01811218e+00
5.40375412e-01 5.09184957e-01 -2.67243445e-01 5.03266275e-01
-2.59064175e-02 5.96454322e-01 7.59512782e-01 -1.04696798e+00
6.09567761e-01 5.97670794e-01 5.10376990e-01 -8.88804376e-01
-4.18173522e-01 -4.62305427e-01 -8.00054550e-01 5.09409606e-02
3.38454425e-01 -2.25358069e-01 -6.23906732e-01 2.21771693e+00
5.54198503e-01 2.41115212e-01 2.82255292e-01 7.41375089e-01
-1.07925192e-01 7.05689669e-01 -2.29681939e-01 -6.22375488e-01
7.26982832e-01 -8.72563720e-01 -8.82417560e-01 -2.80572385e-01
4.76733446e-01 -6.78752512e-02 1.52939224e+00 4.00576890e-01
-7.52987742e-01 -3.42900127e-01 -1.01489401e+00 3.34615469e-01
-3.45069431e-02 1.88726574e-01 9.55861449e-01 4.44245338e-01
-7.01436281e-01 6.38286829e-01 -1.28285718e+00 -1.37887806e-01
1.98869914e-01 5.12637675e-01 -2.22143084e-01 7.20530823e-02
-9.48516548e-01 6.74899876e-01 2.43304312e-01 2.08024696e-01
-1.35163093e+00 -6.39459074e-01 -8.38994384e-01 -2.19752595e-01
9.58101392e-01 -1.71051890e-01 1.38933229e+00 -3.89385134e-01
-2.01049447e+00 3.14309485e-02 -2.07185326e-03 -5.65737188e-01
7.18357801e-01 -3.38717788e-01 -3.67667377e-01 2.95508094e-02
1.10295340e-02 8.88919681e-02 1.07156110e+00 -1.17420650e+00
-4.43719715e-01 -3.17198575e-01 1.05847508e-01 3.31047684e-01
-5.01252174e-01 -4.26087886e-01 -1.63369268e-01 -6.01473451e-01
-8.91251713e-02 -1.06489825e+00 -3.79142344e-01 -1.86772227e-01
-2.44871199e-01 -1.47724450e-01 1.05239892e+00 -4.94504869e-01
1.40231168e+00 -2.00316668e+00 1.69457614e-01 1.26450479e-01
1.95142031e-01 2.34300524e-01 -2.93180496e-01 8.09632897e-01
9.40232277e-02 -2.27398574e-01 4.68529761e-02 -1.71501338e-01
2.29740262e-01 5.37171960e-01 -9.90859568e-01 6.17202759e-01
9.42011829e-03 8.69638920e-01 -1.14932215e+00 3.18527520e-02
2.33048126e-01 -1.31598368e-01 -3.88013303e-01 5.44827342e-01
-3.52212787e-01 7.74949014e-01 -6.58002615e-01 5.92546344e-01
2.32347310e-01 -2.36573756e-01 5.23637950e-01 2.07143247e-01
-1.75801039e-01 3.08163106e-01 -1.22971904e+00 1.36424065e+00
-6.26493454e-01 3.76817584e-01 9.72774066e-03 -1.14367259e+00
9.38144982e-01 2.00568736e-01 8.68326902e-01 -9.70422149e-01
4.18999344e-02 -2.61922032e-02 -1.44483507e-01 -4.23401505e-01
2.20746636e-01 1.58023685e-01 -3.63015503e-01 8.69192064e-01
-2.03310341e-01 7.12037534e-02 1.90074742e-01 -1.26454076e-02
1.22248840e+00 2.77348101e-01 4.96227175e-01 -1.82019502e-01
1.06141679e-01 -2.17919320e-01 9.07942176e-01 8.86495173e-01
-4.24741775e-01 -2.24961653e-01 6.52734101e-01 -5.01942992e-01
-9.61931705e-01 -1.02097404e+00 2.25325853e-01 9.52712893e-01
1.53185308e-01 -3.07326913e-01 -4.64150190e-01 -6.52462065e-01
3.17261636e-01 5.67968726e-01 -7.10526109e-01 -5.66364586e-01
-6.18596077e-01 -4.64855015e-01 4.07154709e-01 6.14244223e-01
4.86666054e-01 -9.52943265e-01 -7.30341554e-01 3.46830308e-01
8.00694898e-02 -1.20427442e+00 -5.96705317e-01 2.53119886e-01
-9.43340182e-01 -1.05186546e+00 -3.19287747e-01 -3.27537417e-01
6.96154475e-01 4.51559603e-01 4.39852238e-01 -3.46614718e-01
-1.87321320e-01 5.40661395e-01 -2.38378108e-01 -2.36826017e-01
-3.85966539e-01 -1.35501742e-01 8.15157771e-01 -4.29288521e-02
-1.24583192e-01 -5.74448347e-01 -4.11448389e-01 4.38721418e-01
-8.02804708e-01 -1.34682003e-03 4.86261368e-01 1.03861523e+00
6.04105473e-01 1.96720541e-01 8.20034266e-01 -5.32575369e-01
8.54268134e-01 -4.06968474e-01 -1.04643667e+00 2.65130877e-01
-9.35165167e-01 3.80380273e-01 1.14964712e+00 -9.91452932e-01
-7.90213406e-01 -3.85529436e-02 4.60981667e-01 -7.99998879e-01
1.85316280e-01 3.87638330e-01 -7.25760013e-02 1.35435611e-01
4.16972816e-01 5.40478349e-01 3.51686388e-01 -3.68442148e-01
2.92498052e-01 5.00761151e-01 3.65428805e-01 -9.60846603e-01
8.66106987e-01 4.89687145e-01 -4.23424952e-02 -5.69742680e-01
-8.92334163e-01 -2.78535575e-01 -3.97914767e-01 -1.48337096e-01
9.32959914e-02 -8.72794271e-01 -1.10366988e+00 3.88806671e-01
-2.95538992e-01 -9.44581151e-01 -5.30537069e-01 3.81653726e-01
-9.72899377e-01 2.59754390e-01 -4.51730222e-01 -1.23409438e+00
-1.50604695e-01 -1.17580497e+00 7.90990651e-01 -2.80905049e-02
5.20223901e-02 -7.64234126e-01 2.59798765e-01 8.94157737e-02
3.51879984e-01 3.12524855e-01 7.84792423e-01 -2.62539744e-01
-4.22964931e-01 -1.33766413e-01 1.44637108e-01 3.18131417e-01
3.43259871e-01 -2.79309332e-01 -7.34819293e-01 -9.54139471e-01
8.18853006e-02 -7.32363462e-01 5.66101789e-01 1.71938896e-01
1.44097710e+00 -6.79681420e-01 -1.11972943e-01 4.17021424e-01
1.14917207e+00 4.05863822e-01 2.59788394e-01 3.67429078e-01
5.33038735e-01 2.76554108e-01 1.05458736e+00 1.02749753e+00
3.98764849e-01 5.56328952e-01 3.49163443e-01 1.33308858e-01
4.27732587e-01 -8.92333865e-01 9.08059299e-01 9.23603714e-01
1.87364757e-01 2.79216357e-02 -6.30843163e-01 4.01131094e-01
-2.16228318e+00 -9.21040773e-01 5.08910775e-01 2.64097023e+00
9.17705059e-01 4.30396795e-02 4.08915639e-01 1.27709970e-01
3.42432737e-01 1.41255736e-01 -1.28958809e+00 -1.34349048e-01
1.82960838e-01 2.13025026e-02 6.23219609e-01 2.45264396e-01
-9.11009014e-01 8.57863605e-01 6.86765575e+00 8.07582855e-01
-1.24142742e+00 -1.88331865e-02 3.58398616e-01 -2.82284170e-01
-5.48408069e-02 1.19804516e-02 -8.25416446e-01 5.71278095e-01
9.80033934e-01 -3.76794189e-01 1.01691103e+00 9.47278380e-01
6.66214466e-01 5.91171579e-03 -1.17322743e+00 9.38773751e-01
-3.68607432e-01 -1.10837221e+00 -2.01508969e-01 3.49978000e-01
7.77978361e-01 -3.46046984e-02 1.97079703e-01 7.45852828e-01
6.21750116e-01 -8.41546774e-01 6.77896261e-01 2.29410291e-01
9.07024741e-01 -7.63467312e-01 1.36266872e-01 7.04417586e-01
-1.24932885e+00 -7.56042778e-01 -3.60878944e-01 -2.48576313e-01
-1.75921679e-01 2.09540308e-01 -6.77489161e-01 3.95267367e-01
4.87990916e-01 8.72864366e-01 -2.12598905e-01 5.57209015e-01
-3.12381923e-01 6.51678801e-01 -4.24241126e-01 -9.85641330e-02
2.49628365e-01 -3.51944387e-01 4.76437002e-01 4.13696706e-01
1.67168140e-01 -8.25195760e-02 6.51725113e-01 6.95501685e-01
1.26455233e-01 -2.76762605e-01 -7.68164337e-01 -5.15431404e-01
6.59376383e-01 1.07229233e+00 -4.06293273e-01 -1.69069022e-01
-1.18927665e-01 6.44761562e-01 6.18646860e-01 5.35682499e-01
-9.32119966e-01 -2.26810053e-02 8.62643778e-01 -3.28017473e-01
1.29811242e-01 -6.90004647e-01 1.01143718e-01 -1.36049819e+00
1.08434796e-01 -1.21608043e+00 3.85022104e-01 -1.72845826e-01
-1.16013992e+00 2.11092085e-01 3.87707353e-03 -1.62300408e+00
-5.03142893e-01 -4.08370584e-01 -3.52353990e-01 3.42350066e-01
-1.42816591e+00 -7.68786967e-01 -1.11508161e-01 7.28493929e-01
4.88300860e-01 -2.64772773e-01 8.83665979e-01 -4.78578061e-02
-1.12291634e+00 6.61631584e-01 5.85474253e-01 -2.60454416e-01
4.86698717e-01 -1.26081264e+00 2.10250974e-01 6.95855796e-01
-2.31898241e-02 7.17821836e-01 7.28395700e-01 -6.03053570e-01
-2.13292456e+00 -1.17205536e+00 -1.14832208e-01 -2.72150397e-01
9.92912054e-01 -6.82624340e-01 -6.56096995e-01 7.38311768e-01
-3.16224575e-01 6.59729540e-02 4.10424352e-01 -2.27181287e-03
-2.79822290e-01 -3.15992713e-01 -9.09459174e-01 7.12974131e-01
1.05278063e+00 -5.42999029e-01 -2.41586834e-01 4.24160898e-01
8.29497933e-01 -5.02806485e-01 -8.91181588e-01 3.19609761e-01
5.04677415e-01 -4.90888119e-01 7.50532210e-01 -8.94131064e-01
-1.97758898e-02 -2.06700236e-01 -4.43771482e-02 -1.27743936e+00
-4.47034575e-02 -1.18565452e+00 -7.33585119e-01 8.08444083e-01
-2.54821926e-02 -8.59587848e-01 5.71197987e-01 5.68284154e-01
1.20350704e-01 -9.61081266e-01 -7.89868534e-01 -1.45607054e+00
-1.38229817e-01 -4.34002936e-01 6.07656896e-01 9.66725469e-01
2.53334254e-01 -1.53983552e-02 -7.07675397e-01 2.58734614e-01
5.64001679e-01 5.15503228e-01 1.15008271e+00 -8.24283838e-01
-6.04834676e-01 -1.56033009e-01 7.70295784e-02 -1.27718651e+00
3.33453000e-01 -4.83585507e-01 7.67379701e-02 -1.13596749e+00
-1.23483744e-02 -7.36659050e-01 -5.03384590e-01 7.59584308e-01
5.79934716e-02 -3.39595318e-01 5.38348183e-02 4.11521792e-01
-7.38633871e-01 1.21322823e+00 1.26776290e+00 1.92996234e-01
-6.63041770e-01 1.94073394e-01 -4.76402342e-01 3.65999997e-01
8.85547936e-01 -2.86822140e-01 -9.34347928e-01 -1.63463205e-01
-7.69204460e-03 3.51899713e-01 1.54792488e-01 -7.11217582e-01
-8.48114043e-02 -6.80039465e-01 -5.48594678e-03 -3.81960750e-01
3.04660350e-01 -7.52491534e-01 -1.31093651e-01 7.70049214e-01
-4.89315271e-01 8.08477700e-02 3.11067015e-01 1.04625094e+00
1.06835246e-01 3.42186421e-01 4.99656022e-01 1.29262254e-01
-5.84182203e-01 5.78638017e-01 -2.83002257e-01 1.67211518e-01
1.29267597e+00 -3.97224538e-02 -3.35941523e-01 -5.41871369e-01
-4.20735627e-01 6.72477365e-01 3.00621927e-01 7.18442976e-01
4.73280907e-01 -1.31404781e+00 -7.54819810e-02 4.11309004e-01
6.29409552e-02 -1.06058009e-01 1.59010172e-01 8.41026485e-01
7.82919824e-02 5.60985565e-01 -2.37127885e-01 -4.58217472e-01
-7.04985440e-01 7.19405472e-01 9.68613774e-02 -4.22912568e-01
-9.12726521e-01 1.94446206e-01 5.30047491e-02 -5.53435028e-01
4.38121885e-01 -6.43585324e-01 8.33253264e-02 -1.16390377e-01
4.90122318e-01 4.29139465e-01 -2.03124017e-01 -2.08778623e-02
-5.96269332e-02 7.15201572e-02 -1.74353451e-01 -4.77810502e-02
1.44525456e+00 -8.40501860e-02 3.05977345e-01 6.74865544e-01
8.75334382e-01 -2.21801311e-01 -2.04975796e+00 -3.09508085e-01
-9.63228345e-02 -5.96721292e-01 6.11485764e-02 -5.79474270e-01
-8.86742949e-01 5.45807064e-01 4.63481277e-01 2.29053870e-01
1.08723354e+00 -3.49585772e-01 7.04244673e-01 7.55781651e-01
8.48255634e-01 -1.44185400e+00 4.45069790e-01 4.62726951e-01
8.50769281e-01 -1.07289934e+00 6.95142001e-02 9.44994986e-02
-8.32690954e-01 8.06737185e-01 7.14125037e-01 -2.61799991e-01
3.18174005e-01 2.42359087e-01 -3.20225060e-01 1.44189328e-01
-1.20906425e+00 -6.75362498e-02 -5.86449355e-02 5.29459000e-01
-2.63459980e-01 2.51701236e-01 -1.55227343e-02 5.97960770e-01
-1.61675103e-02 -3.54722440e-02 5.19444108e-01 1.14256525e+00
-3.32190812e-01 -1.10785699e+00 -8.72029662e-02 3.94323468e-01
2.04538018e-01 5.02962112e-01 -1.36610627e-01 9.64730859e-01
-5.71083724e-01 6.80791616e-01 -8.93791765e-02 -4.07441586e-01
2.15387702e-01 -2.58939922e-01 4.14689660e-01 -3.81442666e-01
-9.73671675e-02 8.48665312e-02 -1.56309664e-01 -1.09112096e+00
1.66936330e-02 -7.63757825e-01 -1.37499261e+00 -2.70262063e-01
-2.64427722e-01 -6.84373826e-02 5.18293142e-01 1.01932609e+00
7.17570484e-01 3.92432809e-01 1.19880366e+00 -7.24216223e-01
-1.50800443e+00 -8.08699489e-01 -6.30002558e-01 2.44287804e-01
5.04524827e-01 -1.15098131e+00 -3.36315364e-01 -3.21433336e-01]
|
[4.12369441986084, 2.1486754417419434]
|
1f97c6b6-e824-429d-ad13-d923082f43f8
|
unsupervised-multi-target-domain-adaptation-2
|
2007.07077
| null |
https://arxiv.org/abs/2007.07077v4
|
https://arxiv.org/pdf/2007.07077v4.pdf
|
Unsupervised Multi-Target Domain Adaptation Through Knowledge Distillation
|
Unsupervised domain adaptation (UDA) seeks to alleviate the problem of domain shift between the distribution of unlabeled data from the target domain w.r.t. labeled data from the source domain. While the single-target UDA scenario is well studied in the literature, Multi-Target Domain Adaptation (MTDA) remains largely unexplored despite its practical importance, e.g., in multi-camera video-surveillance applications. The MTDA problem can be addressed by adapting one specialized model per target domain, although this solution is too costly in many real-world applications. Blending multiple targets for MTDA has been proposed, yet this solution may lead to a reduction in model specificity and accuracy. In this paper, we propose a novel unsupervised MTDA approach to train a CNN that can generalize well across multiple target domains. Our Multi-Teacher MTDA (MT-MTDA) method relies on multi-teacher knowledge distillation (KD) to iteratively distill target domain knowledge from multiple teachers to a common student. The KD process is performed in a progressive manner, where the student is trained by each teacher on how to perform UDA for a specific target, instead of directly learning domain adapted features. Finally, instead of combining the knowledge from each teacher, MT-MTDA alternates between teachers that distill knowledge, thereby preserving the specificity of each target (teacher) when learning to adapt to the student. MT-MTDA is compared against state-of-the-art methods on several challenging UDA benchmarks, and empirical results show that our proposed model can provide a considerably higher level of accuracy across multiple target domains. Our code is available at: https://github.com/LIVIAETS/MT-MTDA
|
['Atif Belal', 'Louis-Antoine Blais-Morin', 'Madhu Kiran', 'Le Thanh Nguyen-Meidine', 'Jose Dolz', 'Eric Granger']
|
2020-07-14
| null | null | null | null |
['multi-target-domain-adaptation']
|
['computer-vision']
|
[ 1.99265182e-01 9.65032578e-02 -2.13955402e-01 -4.14172679e-01
-8.80434752e-01 -8.79223228e-01 5.39511800e-01 1.20274909e-01
-4.36767876e-01 5.99452496e-01 -1.96370795e-01 -2.23086700e-01
-1.30956873e-01 -8.08696032e-01 -7.08650172e-01 -7.96672761e-01
4.71987337e-01 7.63879001e-01 5.26238561e-01 -6.58258498e-02
-2.99565613e-01 4.11018878e-01 -1.22623718e+00 1.06549375e-01
1.08339965e+00 7.40493059e-01 3.15480173e-01 4.23191071e-01
-8.38841721e-02 4.91243333e-01 -6.98731959e-01 -4.38619584e-01
3.39587361e-01 -5.46514571e-01 -9.78941381e-01 2.53301829e-01
6.80265248e-01 -2.07810268e-01 -1.75725281e-01 1.04593861e+00
3.91479939e-01 2.50417203e-01 8.40882242e-01 -1.19019389e+00
-7.81882524e-01 2.80855954e-01 -6.57465398e-01 2.67985225e-01
-1.15420669e-02 1.62051454e-01 7.90230513e-01 -7.17784286e-01
5.01321375e-01 1.08427238e+00 5.00122547e-01 8.77502441e-01
-1.33564186e+00 -7.74804890e-01 3.56551349e-01 8.90750811e-02
-1.20896482e+00 -1.48128733e-01 7.84021378e-01 -5.96710265e-01
5.63803434e-01 -2.35077336e-01 3.73662859e-01 1.16885448e+00
-1.38199195e-01 9.62280631e-01 1.17578101e+00 -3.09677690e-01
2.72218019e-01 2.59641737e-01 1.95538387e-01 4.19124007e-01
1.95187137e-01 -9.77150276e-02 -3.60181838e-01 -1.73174366e-02
7.70685196e-01 -7.81991461e-04 -1.91356793e-01 -7.24935472e-01
-9.98833656e-01 7.50701368e-01 3.57497811e-01 1.48168191e-01
-3.56882602e-01 -3.51768404e-01 3.69309396e-01 5.72036207e-01
6.64945602e-01 4.54807758e-01 -7.33137548e-01 2.42913254e-02
-7.93304503e-01 3.70941162e-01 6.17230415e-01 1.04772842e+00
1.00194740e+00 5.38366772e-02 -1.78790212e-01 1.03639078e+00
1.47498861e-01 4.89259422e-01 5.84114134e-01 -7.11618721e-01
3.22477520e-01 8.09815586e-01 7.32781887e-02 -5.77278554e-01
-1.38705671e-01 -4.76292133e-01 -7.03628898e-01 3.70963007e-01
7.11849451e-01 -2.45407909e-01 -1.05731046e+00 1.87813568e+00
7.77789891e-01 4.75461513e-01 3.21544558e-01 8.38379502e-01
7.76190937e-01 6.20165765e-01 1.37446225e-01 -3.03887725e-02
1.13021624e+00 -1.14289868e+00 -1.58791900e-01 -4.39295769e-01
5.07444382e-01 -6.60604060e-01 1.05863988e+00 5.15490115e-01
-7.77564645e-01 -7.20421731e-01 -7.83115208e-01 -2.69624237e-02
-4.22696471e-01 -1.89599656e-02 6.25907164e-03 4.64049816e-01
-9.23529029e-01 2.22366527e-01 -7.26645529e-01 -6.96444750e-01
6.27430618e-01 4.23681915e-01 -4.12530720e-01 -4.18353528e-01
-1.01313174e+00 9.74284112e-01 5.66148818e-01 -4.65133220e-01
-1.33851290e+00 -8.72908831e-01 -8.87659848e-01 8.38402882e-02
5.23606420e-01 -5.95270872e-01 1.63462937e+00 -1.25692165e+00
-1.74088252e+00 8.94143045e-01 -3.13633941e-02 -4.72257137e-01
3.92058641e-01 -2.90747643e-01 -2.74063230e-01 -2.26358394e-03
1.75588012e-01 7.55636573e-01 1.01924098e+00 -1.20396817e+00
-7.92135239e-01 -2.69378066e-01 2.68430114e-01 3.09757918e-01
-5.67683637e-01 -2.41109937e-01 -3.20937157e-01 -6.63419306e-01
-2.46244907e-01 -9.75508869e-01 -1.48351893e-01 -7.94088989e-02
-3.29329446e-02 -5.80615401e-01 1.24342549e+00 -2.65967876e-01
1.05322421e+00 -2.20103693e+00 2.29961306e-01 -7.77956024e-02
3.09735298e-01 7.83519566e-01 -4.21444058e-01 3.00981015e-01
-1.53955832e-01 -4.03789312e-01 -5.50725877e-01 -2.90194988e-01
-1.26304865e-01 3.84267360e-01 -1.92924887e-01 3.39770883e-01
4.34346288e-01 6.71939433e-01 -1.14936888e+00 -3.46656978e-01
6.00059852e-02 3.68344665e-01 -6.35136366e-01 4.99499589e-01
-5.73844194e-01 7.28284836e-01 -6.20863497e-01 4.74751502e-01
7.58267462e-01 -4.05137718e-01 2.20946878e-01 1.00043200e-01
3.26903500e-02 2.73817241e-01 -1.01739109e+00 1.63752222e+00
-5.29851854e-01 4.98567790e-01 7.49770328e-02 -1.36726725e+00
1.17033577e+00 3.07554275e-01 3.62584442e-01 -5.43941498e-01
5.45806251e-02 2.37887815e-01 1.02929540e-01 -2.69180208e-01
2.88948655e-01 -2.07108796e-01 -1.02845736e-01 4.96897489e-01
3.77770215e-01 -1.16335958e-01 6.05356917e-02 7.36193359e-02
1.04620361e+00 2.73663729e-01 4.46219712e-01 -2.98689276e-01
4.95307922e-01 4.02551472e-01 7.92441130e-01 5.06073833e-01
-4.03323382e-01 4.32599932e-01 1.18990116e-01 -4.60128963e-01
-8.82205546e-01 -1.20879257e+00 1.30323455e-01 1.39284337e+00
1.26685038e-01 5.10715097e-02 -8.39572549e-01 -1.22785592e+00
9.87517387e-02 5.71101129e-01 -5.88389754e-01 -3.30159575e-01
-5.43737292e-01 -3.87594998e-01 4.27411675e-01 5.01628339e-01
6.56083167e-01 -9.96340990e-01 -6.44934833e-01 1.68676585e-01
-2.62662359e-02 -9.87608194e-01 -6.39843524e-01 3.53530943e-01
-8.37016165e-01 -1.07689583e+00 -9.04047370e-01 -1.00893354e+00
7.85488665e-01 3.87831569e-01 1.14564741e+00 -3.24002594e-01
1.67774618e-01 6.63906276e-01 -3.88199240e-01 -4.73064631e-01
-6.76070213e-01 4.38458085e-01 1.35736525e-01 1.35677069e-01
7.57787704e-01 -5.50638556e-01 -4.49441135e-01 4.56704617e-01
-1.04466867e+00 -7.05481172e-02 6.92093849e-01 9.33754742e-01
5.75130999e-01 -4.69339229e-02 8.59563649e-01 -1.13061857e+00
5.30665696e-01 -8.03263128e-01 -8.27036798e-01 2.00209200e-01
-5.28355002e-01 -8.13192725e-02 9.39696848e-01 -9.40630376e-01
-1.21199334e+00 3.13736945e-01 8.13336670e-02 -8.35013568e-01
-6.14152074e-01 3.87144953e-01 -3.43723863e-01 -1.37671232e-02
8.35165560e-01 2.88584799e-01 1.40600717e-02 -5.44451714e-01
3.04866761e-01 6.86432004e-01 5.93185544e-01 -8.14622641e-01
9.96339023e-01 1.34489655e-01 -3.36803108e-01 -6.33737564e-01
-1.03982806e+00 -6.64438665e-01 -8.20830762e-01 -7.77956024e-02
7.49157667e-01 -1.06167459e+00 -2.17877403e-01 6.30371034e-01
-9.69010949e-01 -7.33534038e-01 -4.77112770e-01 4.22728717e-01
-4.28407848e-01 1.50975004e-01 -1.58338636e-01 -2.39591032e-01
-3.24838497e-02 -1.27840364e+00 6.98926866e-01 5.59700608e-01
-2.29385719e-01 -1.34275556e+00 2.62331963e-01 3.64944279e-01
3.32360476e-01 3.71246785e-02 8.52900922e-01 -1.21163976e+00
-3.46601337e-01 3.72954309e-02 -1.86033659e-02 5.84966421e-01
5.13278902e-01 -3.83966953e-01 -1.07837939e+00 -5.88793695e-01
-9.09606274e-03 -5.63465595e-01 5.82021475e-01 2.62388200e-01
9.11973000e-01 -1.97839439e-01 -3.18478465e-01 3.31438661e-01
1.22128189e+00 2.40980089e-01 1.37037322e-01 5.71422338e-01
7.16257930e-01 4.76169199e-01 8.73560369e-01 2.88728714e-01
5.92700839e-01 6.76519513e-01 3.68074864e-01 -2.03479201e-01
-4.12877172e-01 -2.22848371e-01 5.32070220e-01 7.67633677e-01
3.62449259e-01 -1.58172220e-01 -1.15843582e+00 1.07263196e+00
-1.68538654e+00 -5.35930634e-01 1.43414125e-01 2.14793873e+00
1.03976631e+00 -7.64398500e-02 5.35574138e-01 -2.32997701e-01
7.54321694e-01 -3.10625676e-02 -9.64759290e-01 -3.70811552e-01
1.23309843e-01 2.51645237e-01 3.68700802e-01 3.71449471e-01
-1.21109188e+00 1.02259207e+00 5.51873732e+00 9.15988743e-01
-1.22928631e+00 3.66944700e-01 4.47680354e-01 1.13567501e-01
-7.87349194e-02 -6.65165931e-02 -9.16310251e-01 2.96269983e-01
8.08604181e-01 -3.53645116e-01 2.27549121e-01 1.16189909e+00
-6.11061826e-02 3.54642011e-02 -1.27007985e+00 6.38441443e-01
-9.99552682e-02 -9.87117052e-01 1.84419349e-01 -2.37925220e-02
1.05485249e+00 2.74938568e-02 2.18395054e-01 6.09136641e-01
7.31789410e-01 -6.01513147e-01 3.89574915e-01 -1.87229563e-03
6.63570940e-01 -7.06767499e-01 4.41591173e-01 5.98610699e-01
-1.08876598e+00 -9.06671435e-02 -3.63549411e-01 7.99320042e-02
-3.59101027e-01 4.47310627e-01 -1.17354012e+00 6.32871568e-01
8.13133001e-01 8.30696464e-01 -4.84404474e-01 8.99040639e-01
-2.31157795e-01 8.23565245e-01 -1.87112421e-01 2.64892429e-01
3.91030580e-01 -1.57153040e-01 5.72145224e-01 1.22539902e+00
2.86146730e-01 1.41291425e-01 4.71845120e-01 7.88099468e-01
-2.29600191e-01 -1.62553787e-02 -7.22403705e-01 9.52115580e-02
5.48567951e-01 1.05408072e+00 -4.25009817e-01 -4.24658656e-01
-7.36970127e-01 1.01237607e+00 5.60221136e-01 4.65514213e-01
-6.70768499e-01 -1.47171006e-01 9.00096714e-01 1.54567748e-01
4.75288302e-01 -4.76012006e-02 -1.42604029e-02 -1.03625333e+00
-1.31710827e-01 -1.13041604e+00 7.33774722e-01 -3.93487483e-01
-1.60115993e+00 5.05931914e-01 2.55194575e-01 -1.52197254e+00
-1.83306828e-01 -4.70647454e-01 -6.72352433e-01 9.35149610e-01
-1.79082143e+00 -1.19999993e+00 -2.73491085e-01 9.30380404e-01
9.17478144e-01 -3.45109016e-01 7.95035303e-01 2.71352857e-01
-5.64451218e-01 7.67243147e-01 1.89529195e-01 1.80295363e-01
1.10826504e+00 -1.36672962e+00 3.94103229e-01 8.61189783e-01
-5.73186316e-02 4.88662630e-01 4.98228073e-01 -4.70043451e-01
-1.12477756e+00 -1.46765697e+00 6.15588605e-01 -4.29826349e-01
5.89682758e-01 -2.09291354e-01 -1.34003973e+00 9.08311307e-01
3.55661273e-01 2.72289962e-01 6.59810960e-01 4.52099554e-02
-6.17133021e-01 -2.77927756e-01 -1.29479218e+00 3.92151147e-01
4.66706157e-01 -4.22218353e-01 -7.40221024e-01 1.18611693e-01
6.25310540e-01 -4.24702287e-01 -9.88859594e-01 2.89653748e-01
2.14652434e-01 -7.91139066e-01 7.46562064e-01 -5.00217795e-01
3.30867410e-01 -5.66761434e-01 1.73740044e-01 -1.82025981e+00
-3.22708577e-01 -3.02876383e-01 -1.28274933e-01 1.41053891e+00
2.47233614e-01 -7.22722292e-01 7.43516743e-01 3.71652603e-01
-2.92589545e-01 -7.06805944e-01 -9.64309335e-01 -1.16741252e+00
5.94950974e-01 -2.41642520e-02 5.95185399e-01 1.38921630e+00
-2.59029955e-01 4.14444417e-01 -7.84684941e-02 4.89507973e-01
5.30141115e-01 1.99733689e-01 1.04763591e+00 -1.31480134e+00
-3.98801833e-01 -3.40938658e-01 -9.21779796e-02 -1.31376219e+00
2.47662544e-01 -9.40687180e-01 8.98909494e-02 -1.37695551e+00
1.72645152e-01 -5.55089414e-01 -4.11833167e-01 7.97170699e-01
-2.88641989e-01 -1.35953929e-02 1.56607300e-01 1.54274270e-01
-5.60650408e-01 5.92112005e-01 1.36110604e+00 -3.29201877e-01
-2.78594464e-01 8.78277346e-02 -5.80966711e-01 6.97361708e-01
8.86051238e-01 -7.82769442e-01 -7.83154905e-01 -6.37273014e-01
-5.04581034e-01 -1.68260559e-01 2.85866380e-01 -9.54098940e-01
2.77922571e-01 -3.79822165e-01 2.33118013e-01 -3.11379075e-01
1.08335562e-01 -9.35200632e-01 -2.13287786e-01 2.91260600e-01
-8.57003033e-02 -1.42102510e-01 4.98888373e-01 5.23104906e-01
-3.90921205e-01 -2.04204842e-01 1.26278603e+00 -5.61408363e-02
-9.38120425e-01 3.59587401e-01 -2.60882974e-01 2.58363843e-01
1.21336818e+00 -1.34514108e-01 -4.03626651e-01 -1.26951382e-01
-6.90667331e-01 4.04190540e-01 5.60492933e-01 4.93546098e-01
4.34585601e-01 -1.24854267e+00 -7.02611923e-01 1.34380192e-01
3.24223489e-01 6.52352095e-01 2.07815424e-01 6.29869401e-01
-5.22992834e-02 2.07505450e-01 -2.91289449e-01 -8.01073253e-01
-1.18517423e+00 6.45658135e-01 3.51524860e-01 -4.85360324e-01
-4.60440040e-01 8.50357592e-01 7.17166841e-01 -7.70775378e-01
1.61597103e-01 -1.98867843e-01 -1.76375732e-01 -4.56081964e-02
3.37548852e-01 1.93794340e-01 -2.80610677e-02 -5.28203487e-01
-2.42730483e-01 6.55789554e-01 -5.24631917e-01 2.74634838e-01
1.29199266e+00 -1.05324626e-01 3.31730932e-01 3.59043717e-01
9.41140234e-01 -2.92266816e-01 -1.63968325e+00 -6.32742822e-01
7.44314492e-02 -1.70243770e-01 -2.54720151e-01 -9.55196559e-01
-1.04280269e+00 9.37538087e-01 5.18573225e-01 -3.58230844e-02
1.48600948e+00 8.13624263e-02 7.17873156e-01 3.56422305e-01
2.19219908e-01 -8.75493050e-01 4.14944559e-01 6.85711682e-01
5.16147316e-01 -1.44602609e+00 -2.16304764e-01 -1.90326318e-01
-8.28484297e-01 9.14542615e-01 1.18192172e+00 -1.50567591e-01
5.14017045e-01 -1.50883310e-02 3.99163336e-01 -8.64308774e-02
-6.18615925e-01 -2.08453164e-01 2.28930950e-01 8.38274896e-01
2.24035352e-01 7.57794827e-02 1.42876983e-01 7.00057805e-01
1.42356649e-01 -4.59641255e-02 4.29522157e-01 9.69468892e-01
-3.52911770e-01 -1.40697193e+00 -4.60184306e-01 1.20552756e-01
-2.61795163e-01 1.40043110e-01 -3.75467956e-01 9.29346442e-01
2.69782156e-01 7.85530031e-01 -6.31208718e-02 -2.36179829e-01
4.69834298e-01 2.03379244e-01 3.37076694e-01 -1.06631708e+00
-6.82941198e-01 8.62149671e-02 -3.04982960e-01 -1.76228046e-01
-6.05105340e-01 -7.26126015e-01 -1.04935730e+00 3.18990909e-02
-3.26050483e-02 2.36883253e-01 2.31228650e-01 9.85459149e-01
3.37317318e-01 4.44573492e-01 5.83050668e-01 -5.21951258e-01
-5.03930867e-01 -9.95907009e-01 -3.72726709e-01 3.76064211e-01
5.20416856e-01 -8.63589048e-01 -7.25436062e-02 3.41296017e-01]
|
[10.235393524169922, 2.8150651454925537]
|
1a5fab11-6823-4ca4-aead-9c73deefab55
|
option-optimization-algorithm-benchmarking-1
|
2211.11332
| null |
https://arxiv.org/abs/2211.11332v1
|
https://arxiv.org/pdf/2211.11332v1.pdf
|
OPTION: OPTImization Algorithm Benchmarking ONtology
|
Many optimization algorithm benchmarking platforms allow users to share their experimental data to promote reproducible and reusable research. However, different platforms use different data models and formats, which drastically complicates the identification of relevant datasets, their interpretation, and their interoperability. Therefore, a semantically rich, ontology-based, machine-readable data model that can be used by different platforms is highly desirable. In this paper, we report on the development of such an ontology, which we call OPTION (OPTImization algorithm benchmarking ONtology). Our ontology provides the vocabulary needed for semantic annotation of the core entities involved in the benchmarking process, such as algorithms, problems, and evaluation measures. It also provides means for automatic data integration, improved interoperability, and powerful querying capabilities, thereby increasing the value of the benchmarking data. We demonstrate the utility of OPTION, by annotating and querying a corpus of benchmark performance data from the BBOB collection of the COCO framework and from the Yet Another Black-Box Optimization Benchmark (YABBOB) family of the Nevergrad environment. In addition, we integrate features of the BBOB functional performance landscape into the OPTION knowledge base using publicly available datasets with exploratory landscape analysis. Finally, we integrate the OPTION knowledge base into the IOHprofiler environment and provide users with the ability to perform meta-analysis of performance data.
|
['Tome Eftimov', 'Panče Panov', 'Saso Džeroski', 'Carola Doerr', 'Diederick Vermetten', 'Ana Kostovska']
|
2022-11-21
| null | null | null | null |
['data-integration']
|
['knowledge-base']
|
[-3.93596381e-01 -3.09116513e-01 -1.42710358e-01 -3.97286773e-01
-3.42594266e-01 -7.18554974e-01 2.33469814e-01 9.84517813e-01
-4.59864318e-01 5.09349108e-01 2.26692453e-01 -1.16472512e-01
-6.68973505e-01 -9.37272251e-01 -5.66511691e-01 -3.92896950e-01
1.23784505e-01 7.32435703e-01 3.44998866e-01 -4.07320172e-01
2.72029966e-01 4.89807069e-01 -2.17752385e+00 4.26403850e-01
9.62961316e-01 1.02242494e+00 3.41328263e-01 3.35662752e-01
-2.89517254e-01 3.33612263e-01 -7.04374254e-01 -2.32573181e-01
1.70015946e-01 -4.25129198e-02 -9.28469062e-01 -4.94326323e-01
-3.86186950e-02 4.54757571e-01 4.44745481e-01 8.84618819e-01
6.91772938e-01 5.95301390e-02 2.13842556e-01 -1.43587863e+00
-2.12811247e-01 3.94567966e-01 5.05643487e-01 -3.06881469e-04
5.43118715e-01 1.60991058e-01 9.71533298e-01 -3.90366405e-01
1.26970720e+00 8.11440170e-01 2.41665065e-01 2.00079322e-01
-1.15099049e+00 -2.06926540e-01 -3.28815371e-01 5.11488259e-01
-1.23281717e+00 -3.08181405e-01 2.19369844e-01 -6.55590057e-01
1.12292027e+00 8.81817639e-01 8.14849973e-01 6.90662026e-01
-7.97869936e-02 2.81807333e-01 6.59188628e-01 -4.77308780e-01
4.75143164e-01 4.21465188e-01 3.17402035e-01 6.44579887e-01
3.88169944e-01 -8.23220834e-02 -7.30933726e-01 -2.56859124e-01
-9.02586356e-02 -2.08985195e-01 -2.20636413e-01 -7.06387281e-01
-1.06812918e+00 3.51363778e-01 3.15795332e-01 4.01788801e-01
-1.28541216e-01 -7.01292083e-02 5.05808592e-01 4.55806345e-01
1.19967580e-01 9.66084898e-01 -6.19135499e-01 -4.76334393e-01
-2.35045776e-01 5.06105244e-01 1.15970767e+00 9.35363650e-01
9.06824172e-01 -5.39425254e-01 -1.35630861e-01 7.68964887e-01
2.67382443e-01 3.54572609e-02 6.74515605e-01 -9.17523146e-01
3.47828746e-01 1.40077460e+00 3.70602518e-01 -6.63754404e-01
-7.29887307e-01 -3.71108830e-01 -5.72586171e-02 3.46752554e-01
2.41820946e-01 4.06862229e-01 -4.79554862e-01 1.44194484e+00
4.82889265e-01 -5.12752116e-01 3.62619609e-01 9.42260504e-01
9.51849818e-01 2.43850708e-01 1.24639854e-01 1.38241962e-01
1.46669793e+00 -8.61925244e-01 -4.87851024e-01 3.19391936e-01
1.14138556e+00 -7.59681046e-01 1.39909673e+00 3.45973372e-01
-8.24225426e-01 -3.46623152e-01 -1.16727364e+00 -1.25056326e-01
-1.09789717e+00 -3.74717832e-01 6.09044909e-01 5.01845181e-01
-7.16005564e-01 7.74302721e-01 -6.11728668e-01 -4.78726685e-01
4.15314771e-02 2.80355901e-01 -4.66457844e-01 -2.87637860e-03
-1.12023675e+00 1.10159266e+00 9.40040708e-01 -4.49069947e-01
-4.95268732e-01 -1.00317037e+00 -5.89330435e-01 2.08222076e-01
4.33683366e-01 -7.33251035e-01 7.99261689e-01 -7.64836729e-01
-1.29852617e+00 7.38262475e-01 2.10260645e-01 -2.76100457e-01
3.18520516e-01 -2.57010143e-02 -6.07262850e-01 -4.29744631e-01
7.08684251e-02 3.18021357e-01 -8.52082893e-02 -6.51706696e-01
-4.98428464e-01 -5.33572316e-01 1.68426290e-01 1.31980613e-01
-3.37890893e-01 1.89869523e-01 -7.49874890e-01 -2.98459440e-01
-3.09728622e-01 -6.58952773e-01 -1.73275560e-01 -3.15050125e-01
3.17113660e-02 -3.19361463e-02 4.72184211e-01 -5.64826131e-01
1.65233600e+00 -2.08363008e+00 6.40000999e-01 3.39864314e-01
7.69092292e-02 -3.35367559e-03 1.53991599e-02 6.32634580e-01
3.24698612e-02 5.00738025e-01 -1.24299146e-01 1.52526855e-01
2.92626977e-01 3.89747560e-01 4.03907627e-01 4.53739725e-02
4.81048748e-02 5.59164464e-01 -7.14521885e-01 -3.96858722e-01
3.29017252e-01 4.08751033e-02 -7.83076346e-01 3.23783636e-01
-6.57778919e-01 4.98538375e-01 -5.24116576e-01 7.21657217e-01
2.06731245e-01 -2.84622490e-01 3.27435672e-01 -2.63235331e-01
-5.69527864e-01 2.71391898e-01 -1.39475727e+00 2.00731802e+00
-7.17796326e-01 3.10492724e-01 -2.82575607e-01 -7.57865071e-01
8.91936123e-01 2.63953358e-01 6.29168928e-01 -9.16952074e-01
2.44641602e-02 3.98767412e-01 5.88762946e-02 -7.41303444e-01
5.00724018e-01 4.60783154e-01 -1.85405403e-01 3.80122393e-01
2.43752450e-01 -1.35729775e-01 9.98481095e-01 -2.46645764e-01
1.05892527e+00 3.19573134e-01 4.48352009e-01 -5.01156390e-01
6.09453142e-01 3.40664834e-01 5.78752816e-01 2.89174557e-01
2.32897371e-01 2.66243786e-01 5.54431438e-01 -7.78184831e-01
-1.13467860e+00 -8.70296776e-01 -5.89569211e-01 1.20929110e+00
8.17024410e-02 -1.09653294e+00 -6.58893764e-01 -1.68019950e-01
8.75079185e-02 6.23296559e-01 -4.11736339e-01 -4.98049259e-02
-1.93581894e-01 -9.15599644e-01 3.51736367e-01 3.29404697e-02
3.64942342e-01 -1.04752743e+00 -8.69685292e-01 3.39189857e-01
-3.15064609e-01 -8.83565187e-01 1.59319714e-01 1.71418250e-01
-6.07199848e-01 -1.39833534e+00 1.57009721e-01 -2.88862288e-01
1.66533381e-01 -5.02196193e-01 1.49892402e+00 2.78257549e-01
-6.22212350e-01 8.24663565e-02 -4.65338618e-01 -7.63643086e-01
-6.88529730e-01 4.15336102e-01 -2.96046674e-01 -3.96737963e-01
1.25421733e-01 -2.76205361e-01 -2.97941923e-01 6.68670774e-01
-1.07027721e+00 2.18867302e-01 1.31071284e-01 5.34323871e-01
7.93672502e-01 -5.54334782e-02 3.15238774e-01 -5.84891498e-01
5.33226967e-01 -5.27582645e-01 -1.26961219e+00 6.08257413e-01
-1.00323212e+00 5.68084061e-01 5.33627331e-01 2.29888916e-01
-6.64492130e-01 -2.00593099e-01 -1.67704388e-01 1.10167235e-01
-1.15984075e-01 8.19673002e-01 -6.79112971e-01 -1.79357417e-02
8.28293443e-01 -3.58464837e-01 -2.90826820e-02 -9.45647418e-01
2.45653391e-01 7.95139074e-01 2.28329182e-01 -8.40361238e-01
1.39184937e-01 2.26215139e-01 2.80678663e-02 -4.80182827e-01
-4.41080898e-01 -6.71711981e-01 -3.44896406e-01 -1.02873579e-01
8.53849709e-01 -5.07539272e-01 -8.71371031e-01 1.35655746e-01
-8.88643980e-01 -2.28953287e-01 -5.42668939e-01 1.95308268e-01
-4.64382559e-01 -2.83000227e-02 8.09987709e-02 -2.72628009e-01
-3.30393314e-01 -1.64978468e+00 8.25200617e-01 2.37083688e-01
-3.66636723e-01 -9.32574928e-01 3.22630912e-01 6.59960449e-01
5.50800860e-01 4.68030185e-01 1.13803375e+00 -7.60973454e-01
-5.18971026e-01 -1.81792259e-01 2.99591441e-02 1.90223321e-01
-2.62294918e-01 2.36478418e-01 -6.27935648e-01 6.89881295e-02
-4.10802543e-01 3.92317958e-02 2.98229814e-01 -2.92492747e-01
1.39308500e+00 1.08957075e-01 -2.93126047e-01 7.93959200e-01
1.56175518e+00 -1.16445400e-01 5.85186124e-01 1.04005587e+00
3.27701837e-01 6.98997796e-01 5.47828376e-01 3.87065709e-01
3.37635338e-01 1.24296570e+00 6.61758542e-01 7.88335875e-02
2.72505462e-01 2.41398454e-01 5.08000776e-02 7.54395604e-01
-3.73021334e-01 -4.54793543e-01 -1.31340480e+00 1.05130441e-01
-2.04634047e+00 -5.32515407e-01 -2.87191808e-01 2.33810115e+00
6.99031472e-01 -8.83847252e-02 -7.96641856e-02 8.23926181e-02
4.22146291e-01 -3.94665062e-01 -3.19232643e-01 -7.79208779e-01
-6.59566075e-02 3.61060590e-01 4.27629113e-01 2.48776689e-01
-6.86652899e-01 5.72477818e-01 5.75466919e+00 6.26406789e-01
-8.87603700e-01 2.70872563e-01 6.99034706e-02 -2.72080034e-01
-2.54019797e-01 1.72667876e-01 -5.80152571e-01 5.29293120e-01
1.48431373e+00 -5.89964390e-01 8.13095868e-01 8.15479696e-01
3.39777082e-01 -1.71437889e-01 -1.08046246e+00 6.77059114e-01
-2.26258785e-01 -1.87892580e+00 -2.03046724e-01 6.79264441e-02
4.17812645e-01 3.81467313e-01 -5.47758520e-01 5.65807242e-03
3.00792664e-01 -1.02266562e+00 8.86174381e-01 7.33451068e-01
5.07933319e-01 -6.15006208e-01 1.02329659e+00 5.53097017e-02
-9.36389446e-01 -1.45936698e-01 -1.47225544e-01 2.17915513e-02
-2.21325845e-01 6.34374201e-01 -6.18383527e-01 1.31981480e+00
1.09545696e+00 4.76684839e-01 -8.89344037e-01 1.15776300e+00
7.45825395e-02 7.21005574e-02 -2.63087332e-01 -1.04929991e-02
-4.15525347e-01 -3.82979840e-01 4.51283157e-01 8.22940946e-01
2.77585030e-01 -5.37891388e-01 1.64774016e-01 8.06751013e-01
-1.15159415e-01 7.14614093e-01 -1.36480495e-01 -1.43646523e-01
3.22956294e-01 1.17429161e+00 -3.04871738e-01 -2.41732493e-01
-3.28709304e-01 4.20355320e-01 4.24383968e-01 5.32235093e-02
-8.28108490e-01 -4.64456528e-01 9.68039274e-01 1.22035421e-01
-2.55278736e-01 5.26249483e-02 -1.51238844e-01 -1.04363477e+00
1.70076877e-01 -1.03042948e+00 7.15496480e-01 -8.03777158e-01
-9.55108643e-01 3.99468511e-01 1.58647493e-01 -9.99260008e-01
-9.63872671e-02 -7.33748496e-01 -1.66434079e-01 9.14537132e-01
-1.10059643e+00 -6.60989583e-01 -8.54139626e-01 2.14215338e-01
-4.25724573e-02 -8.56609643e-02 1.09759820e+00 6.31775141e-01
-7.89212823e-01 -8.05789977e-03 3.46404642e-01 -2.61959732e-01
6.39639199e-01 -1.26590014e+00 1.52842194e-01 2.74752706e-01
5.02066463e-02 4.48583573e-01 6.73936248e-01 -2.57855862e-01
-1.44429171e+00 -9.89133656e-01 5.22023439e-01 -5.72758317e-01
6.96163535e-01 -2.68831581e-01 -9.41004157e-01 1.66403726e-01
-1.05555966e-01 -8.80109444e-02 1.10149217e+00 3.42394739e-01
-1.80287704e-01 -2.67332226e-01 -9.81368661e-01 4.63327289e-01
1.07175398e+00 -1.70304194e-01 -4.84931767e-01 4.40196097e-01
5.63459635e-01 -4.49910551e-01 -1.59526694e+00 3.18459749e-01
6.23296738e-01 -1.22097826e+00 8.70689332e-01 -9.23085511e-01
8.36553127e-02 -5.42821527e-01 -3.75156909e-01 -1.11550605e+00
1.65693179e-01 -2.56657332e-01 1.87220886e-01 1.14975405e+00
6.51697397e-01 -8.05264115e-01 2.18419448e-01 5.91113627e-01
-3.65083396e-01 -5.73999047e-01 -8.72707546e-01 -8.26318502e-01
-1.76851392e-01 -7.08158374e-01 1.39358747e+00 7.82427073e-01
1.49052531e-01 -9.17575657e-02 3.33496779e-01 1.15809761e-01
9.23330709e-02 2.21045047e-01 1.02454233e+00 -1.54691637e+00
-4.47107136e-01 -5.41156650e-01 -8.13154280e-01 1.31618246e-01
-1.91648439e-01 -1.28631091e+00 -4.41483915e-01 -1.54082179e+00
5.62807806e-02 -7.56311834e-01 -3.82626146e-01 4.88027573e-01
1.33577690e-01 6.20539896e-02 1.51665837e-01 4.37910765e-01
-7.05872178e-01 2.33679086e-01 8.40535402e-01 2.10230291e-01
-2.14995697e-01 -5.77492595e-01 -3.36843491e-01 5.30553579e-01
7.32526720e-01 -6.35131598e-01 -1.67535283e-02 -3.99807930e-01
7.23076522e-01 -3.40605021e-01 3.47255111e-01 -1.28201246e+00
-9.77071449e-02 -2.56976575e-01 -8.51270184e-02 -6.19378500e-02
-3.39788720e-02 -7.98243999e-01 9.16699171e-01 4.40051258e-01
-1.69769168e-01 3.27846199e-01 2.48423129e-01 4.90583852e-02
-2.74011016e-01 -5.53668976e-01 3.79516453e-01 -6.20363913e-02
-7.56049156e-01 4.50121015e-02 -5.69702014e-02 2.58780003e-01
1.04056525e+00 1.12570480e-01 -6.31147265e-01 4.76191193e-01
-8.94991219e-01 3.52287591e-01 1.14482021e+00 6.12170935e-01
-2.89837837e-01 -9.10525084e-01 -4.18873399e-01 3.27339649e-01
8.32828760e-01 -8.99747089e-02 -7.33496845e-02 6.92511618e-01
-1.07379878e+00 2.83363968e-01 -4.48592842e-01 -5.85363507e-01
-1.05251503e+00 5.13537228e-01 4.90636200e-01 -5.02722383e-01
-2.43792713e-01 3.23743448e-02 -5.04279137e-01 -8.13771009e-01
-1.00389071e-01 -2.93194652e-01 -1.36705279e-01 6.14468269e-02
4.16242599e-01 4.68115747e-01 8.17279339e-01 -3.23260665e-01
-5.06539464e-01 2.43739307e-01 5.74219942e-01 3.44927870e-02
1.41197646e+00 3.07053000e-01 -4.90431756e-01 4.96404201e-01
8.41517210e-01 1.50800779e-01 -4.92722481e-01 3.86296451e-01
6.36409760e-01 -4.20894355e-01 -1.24037303e-01 -1.01608038e+00
-7.36924410e-01 3.96936864e-01 7.44722724e-01 3.10739845e-01
1.05920649e+00 9.81964357e-03 -6.59159794e-02 4.36800510e-01
4.13291156e-01 -1.24482501e+00 -4.78240758e-01 2.48197287e-01
9.79947209e-01 -1.01023817e+00 1.65547967e-01 -4.28599983e-01
-2.74830997e-01 1.20313990e+00 4.49222028e-01 5.67539811e-01
4.29338723e-01 2.56823689e-01 1.76854417e-01 -4.48384583e-01
-9.23761606e-01 -4.18003857e-01 3.66713643e-01 4.30792660e-01
5.02851546e-01 7.93858320e-02 -6.14974201e-01 7.41353095e-01
-3.23602676e-01 2.70891964e-01 2.82918543e-01 1.08048928e+00
-3.55732352e-01 -1.69438922e+00 -3.98574054e-01 4.10321683e-01
-2.53191739e-01 -5.47661446e-02 -2.71187335e-01 8.50921154e-01
5.59834719e-01 6.88930631e-01 7.93206617e-02 -9.93447155e-02
7.46355295e-01 3.25230360e-01 2.73311645e-01 -5.72256863e-01
-1.01297367e+00 -5.38635612e-01 5.38211823e-01 -8.30159307e-01
-1.03675745e-01 -3.36812049e-01 -1.25849330e+00 -1.82577744e-01
-2.94910967e-01 5.47263026e-01 1.36057162e+00 7.68727839e-01
9.16974366e-01 7.48742104e-01 -1.29931718e-02 -4.11240906e-01
-8.21472853e-02 -6.23506665e-01 -4.94684279e-02 7.19129801e-01
-6.12724841e-01 -7.95855761e-01 -8.73088464e-02 -4.89971526e-02]
|
[9.166831016540527, 7.918874740600586]
|
09f481af-491f-4bfb-9d51-c94a3d8b4e4b
|
deep-class-incremental-learning-a-survey
|
2302.03648
| null |
https://arxiv.org/abs/2302.03648v1
|
https://arxiv.org/pdf/2302.03648v1.pdf
|
Deep Class-Incremental Learning: A Survey
|
Deep models, e.g., CNNs and Vision Transformers, have achieved impressive achievements in many vision tasks in the closed world. However, novel classes emerge from time to time in our ever-changing world, requiring a learning system to acquire new knowledge continually. For example, a robot needs to understand new instructions, and an opinion monitoring system should analyze emerging topics every day. Class-Incremental Learning (CIL) enables the learner to incorporate the knowledge of new classes incrementally and build a universal classifier among all seen classes. Correspondingly, when directly training the model with new class instances, a fatal problem occurs -- the model tends to catastrophically forget the characteristics of former ones, and its performance drastically degrades. There have been numerous efforts to tackle catastrophic forgetting in the machine learning community. In this paper, we survey comprehensively recent advances in deep class-incremental learning and summarize these methods from three aspects, i.e., data-centric, model-centric, and algorithm-centric. We also provide a rigorous and unified evaluation of 16 methods in benchmark image classification tasks to find out the characteristics of different algorithms empirically. Furthermore, we notice that the current comparison protocol ignores the influence of memory budget in model storage, which may result in unfair comparison and biased results. Hence, we advocate fair comparison by aligning the memory budget in evaluation, as well as several memory-agnostic performance measures. The source code to reproduce these evaluations is available at https://github.com/zhoudw-zdw/CIL_Survey/
|
['Ziwei Liu', 'De-Chuan Zhan', 'Han-Jia Ye', 'Zhi-Hong Qi', 'Qi-Wei Wang', 'Da-Wei Zhou']
|
2023-02-07
| null | null | null | null |
['class-incremental-learning']
|
['computer-vision']
|
[ 1.34454325e-01 -3.00153732e-01 -1.75436348e-01 -3.74605805e-01
-2.46823773e-01 -3.33445519e-01 5.37401140e-01 2.87356406e-01
-5.19037008e-01 7.30717719e-01 -2.74207771e-01 -2.28603169e-01
-2.47230032e-03 -7.92240620e-01 -7.65327334e-01 -7.41590023e-01
2.40322515e-01 3.31683785e-01 4.01022941e-01 1.05278246e-01
3.59693408e-01 4.94641274e-01 -1.87214482e+00 1.65882424e-01
9.24796939e-01 1.02817798e+00 5.14939666e-01 5.04323959e-01
-4.54374142e-02 9.21265244e-01 -5.83025932e-01 -4.54655290e-01
1.06113143e-01 -2.43126780e-01 -7.75314867e-01 8.92572477e-02
4.46554542e-01 -3.41870636e-01 -6.05328977e-01 1.22417104e+00
4.59708750e-01 2.11821236e-02 3.89984995e-01 -1.33172536e+00
-6.61267757e-01 4.57548648e-01 -4.96289581e-01 2.98461109e-01
-1.46428287e-01 5.38238347e-01 5.21307588e-01 -1.12834692e+00
5.63912272e-01 9.76335049e-01 6.53944492e-01 7.14548469e-01
-8.37042868e-01 -7.61515915e-01 4.72000808e-01 8.02125633e-01
-1.12848270e+00 -5.22344530e-01 8.32354248e-01 -3.57219547e-01
8.82539690e-01 -3.29059996e-02 6.66716099e-01 1.17101073e+00
4.26165134e-01 1.03854299e+00 1.01916122e+00 -3.41657639e-01
4.33663994e-01 1.57667130e-01 5.76659381e-01 7.57390261e-01
5.06145597e-01 2.40582719e-01 -7.55184770e-01 1.47436917e-01
4.03978735e-01 4.60716486e-01 -3.03716153e-01 -4.91428316e-01
-1.01952529e+00 5.04681945e-01 4.52210397e-01 2.68194556e-01
-3.28678310e-01 2.44282894e-02 5.04762173e-01 5.02052605e-01
2.91847169e-01 2.77728558e-01 -7.92392433e-01 -1.95469528e-01
-7.45008767e-01 -2.87130345e-02 6.19288981e-01 1.00512159e+00
9.69967246e-01 1.90990925e-01 -8.50841496e-03 7.80202448e-01
-1.02450967e-01 2.73117840e-01 8.31073523e-01 -8.96612585e-01
2.74681710e-02 5.73374569e-01 -1.88747972e-01 -9.81479883e-01
-4.07203346e-01 -8.96670818e-01 -1.17836332e+00 1.28515288e-01
1.76833957e-01 4.33695316e-02 -1.08268988e+00 1.81115568e+00
1.52871802e-01 1.61893725e-01 1.22024014e-03 5.15162706e-01
7.44993210e-01 4.77757871e-01 -4.84702960e-02 -4.18550313e-01
1.11584473e+00 -1.30543733e+00 -6.47129297e-01 -3.75511736e-01
2.82154620e-01 -4.64109212e-01 1.21208751e+00 6.61440313e-01
-9.28297102e-01 -9.44132686e-01 -1.11635661e+00 1.64414853e-01
-4.78590101e-01 6.89525753e-02 6.02291286e-01 2.17522383e-01
-9.39096153e-01 6.66175723e-01 -9.92872715e-01 -5.05437195e-01
7.20890522e-01 1.33035794e-01 -7.88655207e-02 -3.61580551e-01
-1.04176700e+00 1.01018846e+00 5.08037806e-01 -2.47726645e-02
-1.27290988e+00 -6.65253282e-01 -4.41486180e-01 2.90003959e-02
3.73517841e-01 -7.73215115e-01 1.57721853e+00 -1.17519438e+00
-1.33896828e+00 6.33388937e-01 -4.22630638e-01 -6.46470189e-01
4.97168869e-01 -3.49840790e-01 -4.64811802e-01 -2.11964831e-01
-1.75702795e-01 6.07750356e-01 1.06266415e+00 -1.28917015e+00
-9.07009184e-01 -3.86199921e-01 2.30885386e-01 7.86527470e-02
-6.51056707e-01 -5.17672062e-01 -4.03181344e-01 -4.58244473e-01
1.66557372e-01 -8.94281626e-01 -1.51928604e-01 7.03762919e-02
-6.20713755e-02 -2.66295850e-01 8.30841839e-01 -1.75599650e-01
1.37042618e+00 -2.31990290e+00 3.08976620e-02 -4.08424228e-01
4.86874074e-01 4.88380343e-01 -1.03753492e-01 1.84515104e-01
-1.12505712e-01 -2.35978946e-01 -2.99058497e-01 -3.54245991e-01
-2.41666809e-01 3.73044223e-01 -4.75892395e-01 1.46993309e-01
5.70574962e-02 8.86547327e-01 -1.03954661e+00 -1.37447834e-01
1.59326896e-01 3.32405120e-01 -3.31627667e-01 2.21370421e-02
-1.87434077e-01 2.06512928e-01 -1.12154044e-01 7.11741388e-01
6.03615522e-01 -4.22499180e-01 -4.01831381e-02 -3.09294701e-01
-1.52336247e-02 -1.27049675e-02 -7.38168836e-01 1.72799897e+00
-4.16443259e-01 7.57439077e-01 -3.50581557e-01 -1.09880805e+00
6.34075403e-01 1.73784643e-02 4.71250303e-02 -7.49587834e-01
8.63698050e-02 1.41054064e-01 1.76570848e-01 -4.13323849e-01
4.99624103e-01 9.69273821e-02 3.05751622e-01 2.58002877e-01
1.99439839e-01 1.50118008e-01 1.75093710e-01 1.23549744e-01
1.19221175e+00 -1.98491827e-01 2.73213416e-01 -1.89407989e-02
4.17922974e-01 1.28408179e-01 7.58445740e-01 9.37304318e-01
-5.09673655e-01 4.20354545e-01 8.62779841e-02 -1.01311564e+00
-7.22941875e-01 -1.22594714e+00 -5.82490116e-02 1.07278955e+00
2.12848023e-01 -2.23507792e-01 -4.71452087e-01 -7.32581139e-01
-5.41389026e-02 7.66280770e-01 -5.43970883e-01 -7.94542491e-01
-3.59209269e-01 -8.75159621e-01 1.48974627e-01 4.96107131e-01
7.56803036e-01 -1.13102019e+00 -8.66387606e-01 2.19424561e-01
-6.33129105e-02 -7.82030165e-01 -8.72483924e-02 3.75591934e-01
-1.24541450e+00 -1.16124725e+00 -5.09897351e-01 -8.28625977e-01
8.40383112e-01 6.51308477e-01 1.20419455e+00 2.31878191e-01
-2.50286102e-01 4.81211692e-01 -3.75970483e-01 -7.56267130e-01
-2.86955386e-01 2.45426103e-01 3.49293798e-01 -1.21245429e-01
5.72207868e-01 -5.38387775e-01 -6.30353332e-01 1.08824797e-01
-8.48488688e-01 1.91445723e-01 8.64778519e-01 9.28907275e-01
6.27329707e-01 4.14952457e-01 6.35674655e-01 -9.21269953e-01
4.19757664e-01 -4.68138009e-01 -5.25256753e-01 3.17330837e-01
-9.03289974e-01 -3.30738090e-02 6.09280407e-01 -7.18455315e-01
-8.40088665e-01 2.71464959e-02 7.78575912e-02 -5.48502147e-01
-2.09598616e-01 6.20077848e-01 -7.38252252e-02 5.28379120e-02
6.10624075e-01 6.84625685e-01 -1.46765023e-01 -4.07752514e-01
2.84186661e-01 5.09902716e-01 5.75376987e-01 -4.29762006e-01
7.33821273e-01 3.90661627e-01 -4.76809531e-01 -7.65408218e-01
-9.99693871e-01 -1.95360631e-01 -5.86556435e-01 -3.39314729e-01
2.78605431e-01 -9.16397750e-01 -3.24921191e-01 1.14293027e+00
-1.33494878e+00 -4.16911274e-01 -5.21563649e-01 3.06822687e-01
-4.20480698e-01 1.41292170e-01 -6.76477194e-01 -4.80029494e-01
-2.99352258e-01 -1.09222841e+00 4.40968841e-01 5.36367536e-01
3.96385789e-02 -6.97565734e-01 -1.73318312e-02 3.16511504e-02
6.68134511e-01 -1.29479945e-01 1.08362222e+00 -5.09949803e-01
-6.11581266e-01 -1.68278649e-01 -2.11725846e-01 5.77021003e-01
2.66071260e-01 -4.28061672e-02 -1.21091771e+00 -6.97841465e-01
3.08037937e-01 -3.65064651e-01 1.20737433e+00 2.09911153e-01
1.47470725e+00 -2.70460933e-01 -4.19918895e-01 4.71828669e-01
1.31801760e+00 4.49813128e-01 5.24436533e-01 3.70159447e-01
5.34383893e-01 3.55718464e-01 4.02954876e-01 4.17941242e-01
4.91281331e-01 2.33737156e-01 4.41364706e-01 3.83232236e-02
-3.67348015e-01 -5.20157963e-02 3.60593110e-01 1.17214644e+00
1.55410349e-01 -1.63314149e-01 -8.89340401e-01 5.99748075e-01
-1.80137897e+00 -7.35049486e-01 2.65120387e-01 2.33220553e+00
9.41926181e-01 4.37999547e-01 -3.15975875e-01 1.53504357e-01
5.92337132e-01 -9.50827673e-02 -1.20447290e+00 8.26348644e-03
-2.60996073e-02 4.25043367e-02 2.77113229e-01 2.99646467e-01
-9.79107738e-01 9.03632402e-01 5.70558023e+00 7.58714020e-01
-1.33568692e+00 2.75661528e-01 8.14559639e-01 -1.75863996e-01
3.80219258e-02 -5.42816781e-02 -8.22765410e-01 4.29289281e-01
8.21927547e-01 -4.67435718e-01 3.59195650e-01 9.21921551e-01
-1.17353976e-01 -2.74033785e-01 -1.24923193e+00 1.19971180e+00
1.44259036e-01 -1.31547499e+00 2.73931861e-01 -2.79389918e-01
7.97844172e-01 4.04062420e-01 4.10835803e-01 4.59500015e-01
1.96269915e-01 -5.27888536e-01 7.15451777e-01 6.66202188e-01
5.80546081e-01 -6.32593334e-01 6.69633269e-01 4.54267979e-01
-8.82428229e-01 -2.90254593e-01 -4.98764724e-01 -2.83576071e-01
-2.95297295e-01 9.26697254e-01 -4.32488769e-01 3.55161935e-01
1.00600469e+00 8.89929235e-01 -8.55008304e-01 1.27296436e+00
-1.66747570e-01 5.96949279e-01 -8.26065335e-03 1.30671978e-01
3.49673792e-03 2.70832092e-01 3.85758162e-01 8.11143696e-01
3.00155699e-01 -6.87780827e-02 -9.02228355e-02 5.91678739e-01
-1.89729482e-01 -3.19921613e-01 -5.49430966e-01 1.40075445e-01
5.88975370e-01 1.06399858e+00 -7.04148769e-01 -6.49052083e-01
-5.98536015e-01 1.02452421e+00 5.35486877e-01 4.01136279e-01
-8.00841391e-01 -3.15958172e-01 6.80788815e-01 -8.32487419e-02
1.84724152e-01 -3.73103291e-01 -2.27757439e-01 -1.26307321e+00
4.22322601e-02 -9.21001017e-01 2.47501463e-01 -6.72881484e-01
-1.28716755e+00 7.83818364e-01 -1.89888462e-01 -1.20798063e+00
4.67677750e-02 -6.01923466e-01 -4.70376164e-01 3.27363998e-01
-1.64664423e+00 -6.43799186e-01 -5.76643527e-01 4.60402876e-01
9.90547776e-01 -3.46001506e-01 5.85915744e-01 4.00428832e-01
-7.29750752e-01 7.37819970e-01 2.44084343e-01 -8.76365528e-02
7.62016416e-01 -8.02287519e-01 3.18433017e-01 8.65705788e-01
2.17822164e-01 7.18243599e-01 6.51816428e-01 -4.19255763e-01
-1.30311084e+00 -1.33947539e+00 7.36955583e-01 -6.09911680e-01
5.49031138e-01 -2.61828065e-01 -1.26005828e+00 5.38976908e-01
8.38460028e-02 -3.84034105e-02 4.11523491e-01 -2.31742356e-02
-5.18255353e-01 -4.31803495e-01 -7.92469501e-01 5.93584776e-01
1.04160595e+00 -3.56800526e-01 -4.94792521e-01 2.74582684e-01
8.98354769e-01 -2.52128035e-01 -3.79977971e-01 5.76128006e-01
6.08534634e-01 -1.15811527e+00 6.25979900e-01 -5.79548001e-01
2.66823232e-01 -3.57497126e-01 -6.02650307e-02 -1.30369282e+00
-4.99657691e-01 -1.82625845e-01 -5.65948606e-01 9.35178101e-01
4.07137036e-01 -8.69941294e-01 6.95639729e-01 3.07117164e-01
-2.61274338e-01 -9.13809538e-01 -7.84447849e-01 -9.54750896e-01
1.27386257e-01 -4.89553541e-01 5.49041152e-01 9.27541196e-01
-3.46543133e-01 3.33344787e-01 -2.33981818e-01 1.72848761e-01
5.12807012e-01 1.38333276e-01 6.06590033e-01 -1.18086076e+00
-2.69976914e-01 -5.90555906e-01 -2.96433687e-01 -1.02207005e+00
-8.15960765e-02 -6.19227588e-01 9.21978056e-02 -1.33604395e+00
6.06922448e-01 -3.68972838e-01 -8.27010274e-01 6.46696985e-01
-1.77180603e-01 -1.73086561e-02 3.11610013e-01 5.88116527e-01
-9.47079182e-01 6.90090060e-01 1.07682633e+00 -4.36653376e-01
-1.17190249e-01 1.29252210e-01 -6.56991184e-01 9.11786079e-01
9.58389103e-01 -4.65597034e-01 -6.04197800e-01 -8.55305433e-01
2.55709052e-01 -4.83786255e-01 2.98935980e-01 -1.56262231e+00
7.10681915e-01 -1.33265220e-02 4.08655912e-01 -5.01038432e-01
3.27873155e-02 -8.07037711e-01 5.39692258e-03 8.92193317e-01
-1.48888767e-01 2.42121115e-01 3.56478810e-01 6.39138162e-01
-2.81578004e-01 -1.91353470e-01 9.47665215e-01 -2.38890782e-01
-1.19262373e+00 5.87714791e-01 -3.98774058e-01 -6.97512627e-02
9.39391971e-01 -2.31731147e-01 -5.94914675e-01 -2.69550383e-01
-6.67162359e-01 1.09912552e-01 5.71866512e-01 7.14729965e-01
8.41977656e-01 -1.14946353e+00 -4.99719054e-01 3.87826681e-01
3.25446904e-01 1.96703315e-01 4.71455514e-01 6.98904574e-01
-2.64337122e-01 2.37581342e-01 -1.74169764e-01 -6.04982257e-01
-9.82834697e-01 8.27107549e-01 3.08085710e-01 -1.41315535e-01
-3.47720206e-01 8.92263174e-01 3.84375840e-01 -3.52310002e-01
5.72138369e-01 -3.56694907e-01 1.83367878e-02 4.06123735e-02
7.58307934e-01 2.76501566e-01 2.44893536e-01 -1.75468296e-01
-2.83453971e-01 4.08801675e-01 -5.95842063e-01 4.63452309e-01
1.18896699e+00 -1.83076441e-01 -8.58463347e-02 8.36125433e-01
9.36458766e-01 -5.95062613e-01 -1.33372891e+00 -5.64422846e-01
-4.94154990e-02 -1.30721465e-01 7.73640524e-04 -9.94909048e-01
-1.20774722e+00 1.04563093e+00 9.64027822e-01 -2.04453431e-02
1.42058289e+00 -1.47564843e-01 7.75182545e-01 7.01680124e-01
7.25686967e-01 -1.12160492e+00 3.41510892e-01 9.95541930e-01
7.59886622e-01 -1.31703186e+00 7.83053983e-04 1.34057283e-01
-3.39124709e-01 1.02649999e+00 9.54265833e-01 1.59167781e-01
7.87852228e-01 -5.28352195e-03 -1.18147396e-02 -6.07147701e-02
-1.09240937e+00 -3.38850021e-02 -1.55944109e-01 6.26356244e-01
1.64164364e-01 -7.90134668e-02 -4.39714529e-02 4.60541338e-01
-5.92911020e-02 2.23664537e-01 5.07630467e-01 1.07693672e+00
-6.14289105e-01 -9.20495927e-01 1.38842277e-02 6.14815772e-01
4.33357842e-02 -2.35269457e-01 -1.45150244e-01 5.62240481e-01
3.10178310e-01 7.83368170e-01 6.93928450e-02 -6.36821151e-01
3.25081885e-01 9.62356925e-02 4.60406899e-01 -5.82974017e-01
-6.18008375e-02 -5.66010714e-01 -4.90524590e-01 -4.88139540e-01
-4.34890747e-01 -5.38141966e-01 -1.06502914e+00 -4.19855982e-01
-3.06965977e-01 -7.17695206e-02 4.82075423e-01 7.93954790e-01
5.94852209e-01 6.58943176e-01 5.84307253e-01 -6.42953455e-01
-5.70534110e-01 -8.65125358e-01 -1.47548601e-01 1.39402583e-01
4.31905538e-01 -8.16407323e-01 -4.23826098e-01 1.58725023e-01]
|
[9.824945449829102, 3.383453845977783]
|
6a76052e-0d35-496a-a19c-060a1b020d76
|
distributional-perturbation-for-efficient
| null | null |
https://openreview.net/forum?id=rGg-Qcyplgq
|
https://openreview.net/pdf?id=rGg-Qcyplgq
|
Distributional Perturbation for Efficient Exploration in Distributional Reinforcement Learning
|
Distributional reinforcement learning aims to learn distribution of return under stochastic environments. Since the learned distribution of return contains rich information about the stochasticity of the environment, previous studies have relied on descriptive statistics, such as standard deviation, for optimism in face of uncertainty. These prior works are divided into risk-seeking or averse methods, which can be considered as having a one-sided tendency on risk. Unexpectedly, such approaches hinder convergence. In this paper, we propose a novel distributional reinforcement learning that explores by randomizing the risk criterion to reach a risk-neutral optimal policy. First, we provide a perturbed distributional Bellman optimality operator by distorting the risk measure in action selection. Second, we prove the convergence and optimality of the proposed method by using weaker contraction property. Our theoretical results support that the proposed method does not fall into biased exploration and converges to an optimal return distribution. Finally, we empirically show that our method outperforms other existing distribution-based algorithms in various environments including Atari games.
|
['Jungwoo Lee', 'Kyungjae Lee', 'Heesoo Lee', 'Sungyeob Han', 'Tae Hyun Cho']
|
2021-09-29
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[-2.67176390e-01 3.87280822e-01 -4.26528335e-01 -2.08475217e-01
-8.30436110e-01 -5.17387569e-01 4.45713699e-01 -7.82365501e-02
-8.82063508e-01 1.17399526e+00 1.78598881e-01 -5.47981203e-01
-5.50913215e-01 -8.75898421e-01 -6.05372667e-01 -1.11748147e+00
-2.71284074e-01 3.29148322e-01 -2.21110120e-01 -1.49653986e-01
6.36111438e-01 6.04258105e-02 -1.19580817e+00 -4.51561570e-01
1.23536503e+00 1.24695194e+00 3.07407707e-01 3.55073571e-01
6.08722679e-02 7.53957033e-01 -6.75608158e-01 -3.00236553e-01
5.05260587e-01 -5.36451399e-01 -4.38817650e-01 -2.28202134e-01
-4.81710553e-01 -5.86304009e-01 7.31060728e-02 1.57656455e+00
6.04742646e-01 4.42953169e-01 9.49672997e-01 -1.12324595e+00
-5.03425539e-01 1.08095825e+00 -9.09020722e-01 2.42032975e-01
2.20593195e-02 2.21374378e-01 9.57587242e-01 -4.26937491e-01
2.70918339e-01 1.38458979e+00 2.28921905e-01 6.81956589e-01
-1.03092861e+00 -5.86418509e-01 4.32025939e-01 -6.02623448e-02
-1.12899065e+00 8.35505798e-02 7.27741778e-01 -1.35736629e-01
4.73034143e-01 2.32648179e-01 5.92925072e-01 1.02107728e+00
4.57463801e-01 8.82563055e-01 1.56922662e+00 -2.57543445e-01
8.45256746e-01 1.50837749e-01 -6.00859225e-01 2.68872947e-01
7.10794806e-01 6.80653036e-01 -1.41251698e-01 -3.78951281e-01
5.68780005e-01 -1.67477503e-01 -1.26102149e-01 -5.70505083e-01
-7.22446442e-01 1.10176098e+00 1.31352141e-01 -1.67040870e-01
-7.45142400e-01 2.81433105e-01 2.40135729e-01 5.46337187e-01
3.38585466e-01 4.68500286e-01 -1.84817761e-01 -5.20515621e-01
-4.53020394e-01 6.01852417e-01 8.00094783e-01 6.45926476e-01
4.37333375e-01 3.52689415e-01 -3.21877182e-01 4.06355947e-01
4.81591851e-01 7.42834926e-01 5.57612777e-01 -1.13146186e+00
6.93964183e-01 -7.49285668e-02 6.88197374e-01 -9.52873051e-01
-4.69496064e-02 -4.23089713e-01 -5.36527932e-01 6.11213207e-01
4.24166739e-01 -7.65457451e-01 -5.31143844e-01 2.04009748e+00
3.37611258e-01 -2.82082576e-02 3.06449354e-01 9.39314187e-01
-9.22898725e-02 5.15430331e-01 -5.72096594e-02 -7.00242996e-01
5.75678229e-01 -4.27769989e-01 -7.91479468e-01 1.05792031e-01
3.61969262e-01 -2.05717772e-01 1.30608511e+00 6.17979527e-01
-1.16864133e+00 1.19287148e-01 -9.41028178e-01 8.52285683e-01
2.37607360e-01 -5.51234305e-01 4.17418212e-01 1.00967515e+00
-7.36058414e-01 8.56288612e-01 -8.04117024e-01 4.21176152e-03
5.18926203e-01 4.62503582e-02 3.74404818e-01 4.16404456e-01
-1.07027483e+00 7.92182565e-01 5.21261871e-01 -1.48348987e-01
-1.51508331e+00 -4.49869931e-01 -5.27360499e-01 7.20745772e-02
1.09547472e+00 -3.34969103e-01 1.49236000e+00 -7.91642547e-01
-1.94159436e+00 1.62425026e-01 3.60871613e-01 -7.39622533e-01
1.01391399e+00 -3.92157763e-01 1.98636681e-01 -7.42575005e-02
9.05054957e-02 3.84051725e-02 9.17167246e-01 -1.21939659e+00
-7.40285516e-01 -3.24027091e-01 2.52601177e-01 7.26991773e-01
-4.02772248e-01 -2.93143362e-01 4.58081633e-01 -6.68265283e-01
-3.08544576e-01 -7.63332248e-01 -6.94598854e-01 -4.48296845e-01
-4.05782521e-01 -2.55570859e-01 2.19087958e-01 1.47421956e-02
1.35739625e+00 -2.09533858e+00 1.07415495e-02 4.40278441e-01
-3.73739228e-02 -1.26062468e-01 9.58829820e-02 4.73395050e-01
2.76816994e-01 2.62537360e-01 -2.62499958e-01 2.41675116e-02
4.74737227e-01 2.53944576e-01 -6.98660016e-01 6.40815318e-01
-2.05347210e-01 5.30340791e-01 -1.07972014e+00 -3.21794450e-01
-1.69733405e-01 -2.57977992e-01 -7.05703974e-01 4.09725517e-01
-2.96938568e-01 2.47699782e-01 -8.43647361e-01 4.39879000e-01
6.78684592e-01 3.83243650e-01 1.70589387e-01 5.44122934e-01
-2.25880012e-01 5.91273680e-02 -1.40679443e+00 1.16482484e+00
-3.60621214e-01 -8.69418830e-02 8.83801877e-02 -1.27390480e+00
9.68330741e-01 -7.85923079e-02 3.87005240e-01 -4.68224734e-01
4.47277009e-01 3.42927545e-01 8.61925334e-02 -3.04052502e-01
5.54574490e-01 -5.71203649e-01 -2.33188078e-01 8.85522723e-01
-3.71760577e-01 -2.70354539e-01 -1.14634290e-01 -6.31762370e-02
8.09820652e-01 2.29957074e-01 4.66074198e-01 -7.27656066e-01
1.56613454e-01 -4.21712011e-01 8.76044154e-01 1.27121615e+00
-4.40584332e-01 -7.52232820e-02 1.01041722e+00 -1.02755360e-01
-7.25613296e-01 -1.32182550e+00 1.83813134e-03 8.93011272e-01
3.21376473e-01 -3.10135838e-02 -7.67380893e-01 -9.10960197e-01
2.34592795e-01 1.16976547e+00 -7.45409429e-01 -4.58859503e-01
-1.42960444e-01 -9.54535723e-01 3.29843849e-01 3.75790089e-01
6.34258926e-01 -9.87450719e-01 -1.12581778e+00 7.52602192e-03
1.11893833e-01 -1.51138991e-01 -4.57151771e-01 3.73342246e-01
-8.06943238e-01 -9.29950416e-01 -8.19419920e-01 -9.32907015e-02
5.02656758e-01 -1.73861057e-01 8.25122893e-01 -5.59614599e-01
1.43995687e-01 5.59402287e-01 -2.35264719e-01 -9.52994704e-01
-1.50194272e-01 -2.04898700e-01 5.27093410e-01 -1.12488307e-01
3.72372031e-01 -5.69929063e-01 -9.16375756e-01 2.04053625e-01
-7.38566101e-01 -7.13136137e-01 5.79178095e-01 9.00091648e-01
6.79740787e-01 5.02954721e-01 9.12524641e-01 -7.01615870e-01
1.38393867e+00 -7.12690055e-01 -8.75619590e-01 7.77954906e-02
-1.00184989e+00 5.39701641e-01 6.08282208e-01 -5.88417292e-01
-1.53572345e+00 -4.03718114e-01 3.58929068e-01 -1.92355856e-01
1.70800626e-01 5.50553858e-01 -2.76695341e-01 3.16541493e-01
6.85430169e-01 2.93319732e-01 1.91010162e-01 -2.19113380e-01
3.56159031e-01 6.74606085e-01 1.77814126e-01 -1.15853679e+00
7.13037193e-01 4.27224189e-01 1.22303978e-01 -3.89559746e-01
-8.87741923e-01 1.22788936e-01 1.49837285e-01 -2.46797323e-01
4.89492208e-01 -5.48687160e-01 -1.08471215e+00 2.10140079e-01
-4.23528105e-01 -3.81505847e-01 -6.59827769e-01 6.90550268e-01
-1.34797657e+00 3.38787526e-01 -1.65990710e-01 -1.68057930e+00
-1.72201872e-01 -8.94286931e-01 4.47875708e-01 2.93780953e-01
1.76711842e-01 -8.91676188e-01 3.76178503e-01 -2.26607323e-01
4.71503109e-01 3.62668455e-01 5.72009623e-01 -6.16794288e-01
-2.78340161e-01 2.89572954e-01 2.42938668e-01 3.19525421e-01
8.69574845e-02 -4.83750314e-01 -4.34421152e-01 -4.94187146e-01
5.14076293e-01 -6.52287900e-01 8.38096380e-01 5.83485365e-01
1.19036698e+00 -7.47272909e-01 4.78507765e-02 6.30052090e-01
1.38852310e+00 6.20156229e-01 4.48292285e-01 7.80674338e-01
5.04843378e-03 6.31167293e-01 1.31338656e+00 1.23841000e+00
8.96642506e-02 1.11857437e-01 8.89551878e-01 5.74800074e-01
9.49725091e-01 -5.90364993e-01 5.16698301e-01 6.99165910e-02
-1.71471253e-01 -7.15021193e-02 -5.85911453e-01 4.60239291e-01
-2.07877398e+00 -1.29178298e+00 7.85400987e-01 2.63234663e+00
9.71578717e-01 3.01123917e-01 5.34321964e-01 -2.75889426e-01
6.31096005e-01 4.38219421e-02 -1.12627959e+00 -6.66843712e-01
-1.24958158e-02 -1.59406513e-01 8.66563261e-01 4.69743192e-01
-8.09206009e-01 7.18615532e-01 6.53085518e+00 1.03781557e+00
-8.16435099e-01 -2.04599902e-01 8.12603831e-01 -3.75242770e-01
-8.20788026e-01 -1.93137191e-02 -5.98931670e-01 6.14148974e-01
6.90002263e-01 -8.41727614e-01 4.87441510e-01 1.20957184e+00
3.73742789e-01 -3.09472919e-01 -7.69095659e-01 8.31874132e-01
-3.78240168e-01 -7.78288662e-01 -1.52809188e-01 2.89928526e-01
7.47734487e-01 -3.96575600e-01 5.78751206e-01 5.09761810e-01
9.58470762e-01 -1.22450447e+00 8.96020830e-01 7.41029978e-01
4.28120941e-01 -1.53789556e+00 7.99573660e-01 6.37835503e-01
-5.98384321e-01 -5.48308492e-01 -6.67339742e-01 -2.91714042e-01
2.20944472e-02 4.15058345e-01 -5.54124594e-01 3.25821161e-01
6.42411053e-01 2.46050969e-01 3.22405510e-02 8.49507451e-01
-2.65017211e-01 6.74299121e-01 -3.09974581e-01 -6.21542871e-01
6.19644225e-01 -7.06419766e-01 6.90062582e-01 5.58307946e-01
6.14527762e-01 6.25169501e-02 2.81959563e-01 9.94062364e-01
2.76859313e-01 1.60960034e-01 -9.18601394e-01 -9.17324647e-02
5.82634628e-01 8.05118084e-01 -4.99228060e-01 3.44061814e-02
-5.12853786e-02 4.66360986e-01 3.20823103e-01 3.39528024e-01
-8.87121677e-01 -6.08821988e-01 6.57715023e-01 -2.08938509e-01
3.25659305e-01 7.22316746e-03 -2.79678315e-01 -8.75401258e-01
1.34189114e-01 -8.29327345e-01 5.81073523e-01 2.02552485e-03
-1.27049458e+00 3.61738682e-01 2.81443954e-01 -1.08377969e+00
-5.13139129e-01 -3.65431070e-01 -6.74760222e-01 6.69494629e-01
-1.28415239e+00 -2.37401038e-01 3.91742676e-01 5.14331877e-01
3.35950851e-01 -3.95486623e-01 3.09297174e-01 -3.74073982e-01
-4.00347918e-01 6.39983594e-01 6.28602386e-01 -3.68212879e-01
5.88237345e-01 -1.73887730e+00 -3.31417114e-01 5.17144561e-01
-4.20408189e-01 5.26388168e-01 1.07655752e+00 -7.38481522e-01
-1.30093825e+00 -8.59846592e-01 -1.09902963e-01 -1.33425742e-01
8.24905038e-01 6.90597668e-02 -4.62796152e-01 4.39358771e-01
2.07513914e-01 -3.33706200e-01 5.04741251e-01 8.84342864e-02
1.85275357e-02 -1.35815471e-01 -1.42263699e+00 9.43043590e-01
9.50704932e-01 6.80166408e-02 -7.12679148e-01 -4.76711988e-02
6.28755033e-01 -2.27385938e-01 -5.01427174e-01 2.61773944e-01
3.52754086e-01 -1.07337976e+00 5.67683816e-01 -8.19701970e-01
2.44557112e-01 1.48691013e-02 -3.32587034e-01 -1.75467098e+00
-8.47178325e-02 -1.18786049e+00 -4.05595116e-02 8.95573497e-01
1.64174646e-01 -8.58205318e-01 8.53887737e-01 4.37957108e-01
1.16705805e-01 -1.05139816e+00 -1.10066044e+00 -1.31914377e+00
6.76742315e-01 -2.71705389e-01 6.23182356e-01 4.82156008e-01
3.11854035e-01 -1.64228246e-01 -6.31396115e-01 -1.44854397e-01
1.14726877e+00 2.18707949e-01 5.48785865e-01 -7.87712455e-01
-5.25927961e-01 -6.54513359e-01 4.58983518e-02 -9.10959780e-01
2.64551163e-01 -4.54869121e-01 3.43517333e-01 -9.93280351e-01
1.78425550e-01 -3.38416249e-01 -6.00354016e-01 -2.39123311e-02
-2.48787120e-01 -5.75611651e-01 4.55140769e-02 -1.08807333e-01
-7.70952880e-01 1.20899332e+00 1.29221535e+00 6.11814260e-02
-4.20919150e-01 3.07987720e-01 -1.08370578e+00 8.90884697e-01
1.11414647e+00 -4.15736705e-01 -9.30184126e-01 5.93290552e-02
5.35295725e-01 2.74242103e-01 -2.01928943e-01 -5.80381274e-01
-2.11083859e-01 -9.51572537e-01 -6.06851801e-02 -3.27268034e-01
-1.27075002e-01 -5.18353343e-01 -3.00696909e-01 7.62458324e-01
-6.64067030e-01 4.42804135e-02 -1.97991952e-01 1.07542372e+00
-3.58166620e-02 -5.66582322e-01 8.40033412e-01 -3.26104462e-01
-1.92358881e-01 3.21499795e-01 -5.66993356e-01 5.20587981e-01
1.27153325e+00 6.83444515e-02 2.24359352e-02 -8.61157179e-01
-4.45782423e-01 5.71961105e-01 2.02801034e-01 3.84830795e-02
6.50015771e-01 -1.31490028e+00 -6.62599266e-01 -2.72111475e-01
-1.42654866e-01 -1.04627252e-01 5.26004918e-02 6.79875314e-01
-1.21920742e-01 1.89361572e-01 -2.23222539e-01 -1.57139987e-01
-6.86491072e-01 7.61061788e-01 4.06492323e-01 -2.89323419e-01
-3.24788153e-01 6.61698878e-01 3.54018241e-01 -2.67572761e-01
4.80933905e-01 -8.42007846e-02 -1.31884873e-01 6.21095635e-02
5.84369779e-01 5.85583389e-01 -5.62062681e-01 1.08209431e-01
-1.47498161e-01 1.75571844e-01 -2.34256517e-02 -6.48423374e-01
1.26304328e+00 -4.02576268e-01 2.19264641e-01 5.64392567e-01
5.92501223e-01 -3.76494974e-03 -1.64570045e+00 -9.07938853e-02
2.33366966e-01 -8.31829488e-01 -6.63427711e-02 -6.12390637e-01
-7.23711073e-01 5.73252738e-01 5.07286489e-01 2.80928761e-01
1.02302003e+00 -3.46079409e-01 3.21207911e-01 6.84382319e-01
7.44948149e-01 -1.70522404e+00 1.46290466e-01 4.03200746e-01
8.00797045e-01 -1.17366922e+00 -1.37326166e-01 2.58137107e-01
-1.15771747e+00 7.30732203e-01 8.24526191e-01 -3.33292127e-01
6.71812892e-01 2.08145827e-01 -2.87120212e-02 1.76485643e-01
-8.94089282e-01 -2.96840698e-01 -3.19602966e-01 8.78255665e-01
-2.00975873e-02 2.58143961e-01 -7.97903776e-01 8.83603394e-01
-4.65536118e-01 -3.14565569e-01 7.20918596e-01 9.94399965e-01
-7.83839643e-01 -9.25064445e-01 -4.63012666e-01 4.43599641e-01
-6.63233519e-01 7.84017891e-02 -1.06691614e-01 6.15926623e-01
-4.00945365e-01 8.94405663e-01 -2.62815375e-02 -5.05810194e-02
1.26196668e-01 -3.53933454e-01 5.02883136e-01 -3.44515324e-01
-8.26598778e-02 1.43320963e-01 -1.58095747e-01 -6.29707456e-01
-1.22624889e-01 -8.09385538e-01 -1.00914371e+00 -1.54499620e-01
-1.28876686e-01 6.25444829e-01 2.75384814e-01 7.74387419e-01
8.30031335e-02 1.51703626e-01 9.83935893e-01 -2.67380178e-01
-1.92114472e+00 -7.97828794e-01 -1.11546028e+00 7.62390569e-02
2.23462403e-01 -9.30908799e-01 -7.42649972e-01 -7.26796627e-01]
|
[4.247776031494141, 2.5876049995422363]
|
73dd2a3d-6bf7-4bd4-b792-a050eb6ebd46
|
meta-learning-siamese-network-for-few-shot
|
2302.03507
| null |
https://arxiv.org/abs/2302.03507v2
|
https://arxiv.org/pdf/2302.03507v2.pdf
|
Meta-Learning Siamese Network for Few-Shot Text Classification
|
Few-shot learning has been used to tackle the problem of label scarcity in text classification, of which meta-learning based methods have shown to be effective, such as the prototypical networks (PROTO). Despite the success of PROTO, there still exist three main problems: (1) ignore the randomness of the sampled support sets when computing prototype vectors; (2) disregard the importance of labeled samples; (3) construct meta-tasks in a purely random manner. In this paper, we propose a Meta-Learning Siamese Network, namely, Meta-SN, to address these issues. Specifically, instead of computing prototype vectors from the sampled support sets, Meta-SN utilizes external knowledge (e.g. class names and descriptive texts) for class labels, which is encoded as the low-dimensional embeddings of prototype vectors. In addition, Meta-SN presents a novel sampling strategy for constructing meta-tasks, which gives higher sampling probabilities to hard-to-classify samples. Extensive experiments are conducted on six benchmark datasets to show the clear superiority of Meta-SN over other state-of-the-art models. For reproducibility, all the datasets and codes are provided at https://github.com/hccngu/Meta-SN.
|
['Aoying Zhou', 'Ming Gao', 'Minghui Qiu', 'Xiang Li', 'Yingnan Fu', 'Yuhe Wang', 'Chengcheng Han']
|
2023-02-05
| null | null | null | null |
['few-shot-text-classification']
|
['natural-language-processing']
|
[ 8.93730298e-02 -1.48055911e-01 -4.49886948e-01 -2.38844469e-01
-6.48847997e-01 -3.31023112e-02 5.70312798e-01 6.55684397e-02
-5.69472134e-01 8.13727140e-01 4.88867797e-03 5.15962653e-02
-2.53214508e-01 -8.66480529e-01 -2.90423840e-01 -7.22864747e-01
2.86724448e-01 5.22070169e-01 2.77852237e-01 -1.34710386e-01
4.83566463e-01 -8.17124918e-02 -1.81622922e+00 3.53640050e-01
9.57200289e-01 8.69581223e-01 1.47103772e-01 3.25185627e-01
-5.94659865e-01 5.15613914e-01 -5.70708454e-01 -4.12561238e-01
-5.67364180e-03 -6.52272224e-01 -6.76344395e-01 4.98416461e-02
4.44888100e-02 -1.08378775e-01 -3.57508808e-01 1.42783058e+00
4.99352455e-01 4.57898349e-01 8.67988706e-01 -1.52521515e+00
-6.91382408e-01 8.84451509e-01 -6.05957448e-01 4.40522321e-02
5.19354530e-02 9.96422321e-02 1.27170467e+00 -1.26854014e+00
5.39630711e-01 1.20713532e+00 5.40927887e-01 6.80825591e-01
-1.05753410e+00 -8.12197924e-01 2.34269407e-02 7.44462967e-01
-1.50858665e+00 -4.60494995e-01 1.08755410e+00 -1.75041556e-01
4.18063343e-01 1.58515602e-01 4.81775045e-01 1.45195055e+00
-1.16817176e-01 1.02359033e+00 1.12681115e+00 -6.00009918e-01
7.63273478e-01 3.76074612e-01 6.18450522e-01 4.74070251e-01
3.69100720e-01 -1.73249960e-01 -3.39585334e-01 -4.95224088e-01
1.72668785e-01 4.11316335e-01 -2.77473390e-01 -4.32310879e-01
-1.12995791e+00 1.00270283e+00 2.68774182e-01 5.34996152e-01
-2.31452793e-01 -2.98942447e-01 7.22954690e-01 -7.60272667e-02
6.43458068e-01 2.93900788e-01 -3.47304642e-01 -1.27195148e-02
-7.92040944e-01 1.96353316e-01 7.56442070e-01 1.11287260e+00
8.52580190e-01 -1.07527606e-01 -3.04820478e-01 1.23046088e+00
1.73532382e-01 1.91668689e-01 1.10896754e+00 -6.64833546e-01
7.42915034e-01 6.55872405e-01 5.91331767e-03 -9.97523308e-01
-3.22204590e-01 -2.64122307e-01 -1.07115197e+00 -3.14572126e-01
3.13460469e-01 -2.24189654e-01 -7.77733803e-01 1.52936471e+00
4.86816883e-01 5.42616189e-01 1.11201167e-01 7.14610219e-01
8.91816854e-01 7.62007833e-01 -5.95785119e-03 -1.57829225e-01
1.27726865e+00 -1.16731048e+00 -7.08511233e-01 -5.94002753e-02
8.08770835e-01 -2.83276737e-01 1.22126102e+00 2.49396935e-01
-6.46793604e-01 -5.71011245e-01 -1.07419538e+00 2.94098794e-01
-7.30686903e-01 -3.64288501e-02 3.51765871e-01 5.83818436e-01
-4.83531952e-01 8.16254497e-01 -3.77200425e-01 -2.73983330e-01
6.79235995e-01 -7.39156008e-02 -1.84789062e-01 -4.33220863e-01
-1.50104237e+00 5.41442513e-01 8.90854120e-01 -3.06184217e-02
-6.05259418e-01 -6.25824809e-01 -7.31126308e-01 2.55477339e-01
6.24481678e-01 -4.06381249e-01 1.17049849e+00 -8.73161793e-01
-1.33568478e+00 4.95471597e-01 -1.46276146e-01 -9.88141596e-02
5.32389522e-01 2.39514336e-01 -5.24608612e-01 1.29487664e-01
7.67898187e-02 3.76164943e-01 8.49642158e-01 -1.23039770e+00
-7.91463137e-01 -3.31674933e-01 -2.88612872e-01 2.18920648e-01
-7.97405422e-01 -3.18180323e-01 -3.49243760e-01 -5.86483359e-01
-3.47298943e-02 -9.49473500e-01 -1.45727783e-01 -2.77112070e-02
-5.86159885e-01 -8.19456935e-01 7.91325688e-01 -7.16882944e-02
1.32905912e+00 -2.16936064e+00 3.45801003e-04 -4.50224876e-02
4.24115598e-01 6.27962589e-01 -2.53078550e-01 4.70088333e-01
4.75763008e-02 1.28694102e-01 -3.25248182e-01 -3.33963364e-01
5.27919382e-02 2.38723576e-01 -8.87964517e-02 4.30891961e-01
-2.06437215e-01 7.46612251e-01 -1.21392822e+00 -7.87139118e-01
2.08430946e-01 2.25911975e-01 -1.72272816e-01 3.24440263e-02
-2.83323586e-01 -2.13530302e-01 -5.46720386e-01 5.65765977e-01
6.40723586e-01 -3.67483824e-01 1.71648875e-01 -1.69842839e-01
1.44860476e-01 1.07317239e-01 -1.29365456e+00 1.48395967e+00
-2.81071395e-01 2.60572076e-01 -2.76547700e-01 -1.21904385e+00
8.29062879e-01 3.01589429e-01 2.30609775e-01 -4.42052335e-01
3.83699834e-01 2.64450550e-01 -4.67852280e-02 -5.27300954e-01
4.43041116e-01 -2.26954669e-01 8.42859074e-02 6.59612119e-01
2.27385029e-01 2.51221478e-01 5.37466824e-01 1.74088389e-01
8.37233663e-01 -1.87740564e-01 3.86089563e-01 -1.52855247e-01
5.47230422e-01 9.21830349e-03 7.35689878e-01 7.74301231e-01
-5.18321753e-01 5.87283313e-01 4.30187851e-01 -3.18868101e-01
-1.00697255e+00 -7.47410715e-01 -2.22943515e-01 1.12038147e+00
2.53873289e-01 -4.27932680e-01 -6.16534233e-01 -9.74686384e-01
-1.64810382e-02 1.01300788e+00 -7.89038301e-01 -2.84966648e-01
-2.70961940e-01 -9.06467736e-01 3.20054501e-01 3.81157815e-01
4.96971846e-01 -1.08846879e+00 -4.03267860e-01 2.65127212e-01
-2.68334985e-01 -6.96159840e-01 -4.52928782e-01 1.64310515e-01
-8.73357356e-01 -1.23984051e+00 -9.23968196e-01 -7.90330827e-01
6.97745621e-01 6.74971521e-01 7.20295608e-01 4.20498326e-02
-2.41451517e-01 -5.06060943e-02 -7.71842718e-01 -2.58775949e-01
-2.68960655e-01 2.36808047e-01 3.30530629e-02 2.06260905e-01
7.53357410e-01 -4.16398406e-01 -3.46472204e-01 3.18207085e-01
-9.17395115e-01 3.97231355e-02 4.20074642e-01 1.28839052e+00
5.35124660e-01 4.53869522e-01 7.78215945e-01 -1.58953047e+00
8.14071774e-01 -9.08219457e-01 -1.68419868e-01 3.99328291e-01
-7.92000175e-01 -1.76772177e-01 1.05635202e+00 -6.50170207e-01
-8.97063255e-01 -3.61476183e-01 7.50563145e-02 -5.50598621e-01
-1.11538060e-01 5.41792333e-01 -8.92729908e-02 3.12533885e-01
7.02497959e-01 4.44645703e-01 2.03947261e-01 -4.22245383e-01
3.16479206e-01 1.17301309e+00 9.91353840e-02 -5.42774737e-01
5.95138431e-01 3.46430063e-01 -1.78012475e-01 -9.30326104e-01
-1.21855330e+00 -7.08439469e-01 -4.69748557e-01 -4.38341387e-02
3.61683905e-01 -5.89626670e-01 -2.23641664e-01 4.41701353e-01
-8.58743787e-01 -7.09422678e-02 -3.05966914e-01 5.18760562e-01
-3.45735312e-01 4.23588514e-01 -5.99323452e-01 -7.69939542e-01
-4.07028139e-01 -9.48916078e-01 5.31991839e-01 4.07727540e-01
-9.50900614e-02 -1.06770015e+00 2.69520264e-02 2.32772186e-01
3.80881906e-01 -5.72727993e-04 1.08803296e+00 -1.14930689e+00
5.52456081e-02 -2.70389795e-01 -2.88924932e-01 3.24513018e-01
3.10285985e-01 3.00344899e-02 -1.08687854e+00 -3.56529504e-01
-5.08222841e-02 -4.51845646e-01 7.83157825e-01 1.88176811e-01
1.36965907e+00 -2.90133566e-01 -4.36725140e-01 2.71016598e-01
1.32065487e+00 1.89277679e-01 3.78374219e-01 2.43674457e-01
6.10481024e-01 7.35293150e-01 7.13628054e-01 6.65573239e-01
3.81773770e-01 3.90093833e-01 1.53329924e-01 3.56566012e-01
-4.55659628e-02 -4.35750127e-01 -1.07050017e-02 1.22431052e+00
3.50816309e-01 -2.39928246e-01 -8.65976930e-01 4.79403377e-01
-1.97484255e+00 -1.01271439e+00 -9.17962268e-02 2.04224586e+00
8.55193853e-01 2.36773536e-01 1.88915387e-01 4.45115358e-01
1.16817117e+00 2.60691732e-01 -7.13329256e-01 6.69242535e-03
5.11545315e-02 -4.74431813e-02 7.28432015e-02 1.18989982e-01
-1.12761855e+00 7.15756536e-01 5.10317850e+00 1.22951901e+00
-8.15811098e-01 3.92148525e-01 5.78855753e-01 -1.13633662e-01
-2.12945133e-01 -5.42260483e-02 -9.60580826e-01 9.26469684e-01
8.21127355e-01 -4.29618090e-01 2.77291834e-01 1.06382978e+00
-2.55906582e-01 9.00576264e-02 -1.00678074e+00 1.01675606e+00
3.50290835e-01 -1.30410779e+00 1.00379825e-01 -1.57181025e-01
9.05516326e-01 -4.62636389e-02 -1.35725886e-01 8.21621180e-01
1.57384992e-01 -5.43883562e-01 5.49440503e-01 3.06570590e-01
7.58749843e-01 -9.13669586e-01 8.56166363e-01 6.44388914e-01
-9.88063395e-01 -2.95752227e-01 -9.86843407e-01 1.82616279e-01
-1.77460179e-01 8.80707979e-01 -6.87607586e-01 5.96055627e-01
4.95101571e-01 8.25108826e-01 -5.94156027e-01 1.02582526e+00
-1.19096711e-01 6.01757646e-01 -1.09570205e-01 -7.07192957e-01
2.86872923e-01 -1.72405362e-01 2.79511511e-01 1.05245471e+00
1.92792624e-01 -5.24374843e-02 1.89721823e-01 9.60853040e-01
-2.66571760e-01 2.63812900e-01 -5.58269143e-01 -8.51936936e-02
9.56864476e-01 1.33526039e+00 -6.57884121e-01 -5.20995736e-01
-4.61230516e-01 6.91610694e-01 4.01112020e-01 3.23120475e-01
-5.53565383e-01 -9.99918938e-01 3.04196537e-01 -2.28123754e-01
2.72874236e-01 1.45895660e-01 -1.88117623e-01 -1.22295082e+00
-1.54860824e-01 -7.46102273e-01 4.37599421e-01 -5.55544734e-01
-1.74520588e+00 4.55869645e-01 1.05643108e-01 -1.41735864e+00
-9.34297442e-02 -4.44883227e-01 -6.27218902e-01 6.88765526e-01
-1.42941380e+00 -6.78501248e-01 -3.81331921e-01 3.70526373e-01
7.85179734e-01 -3.70386362e-01 6.74302697e-01 2.18140587e-01
-8.19287896e-01 6.21516466e-01 5.36801398e-01 2.86050558e-01
6.93076730e-01 -1.13209748e+00 1.76516265e-01 4.82685715e-01
1.49835736e-01 5.18869042e-01 4.78063852e-01 -4.94009554e-01
-1.39118171e+00 -1.15861678e+00 7.78500259e-01 -2.57222116e-01
6.89650178e-01 -3.55584413e-01 -1.05264592e+00 2.98130631e-01
-9.04396176e-02 2.80862600e-01 8.83015990e-01 1.34492040e-01
-4.42156762e-01 -1.91930592e-01 -1.18687284e+00 7.66271234e-01
8.65543544e-01 -2.73719370e-01 -6.77949071e-01 3.83328527e-01
6.59421146e-01 7.09744021e-02 -4.56636369e-01 1.98505700e-01
3.78076911e-01 -8.82531404e-01 6.77793562e-01 -6.40988827e-01
6.14691496e-01 -1.54704481e-01 -1.32489443e-01 -1.67175484e+00
-4.64096636e-01 -8.99563357e-03 -3.88855338e-01 1.39831567e+00
3.24101269e-01 -8.15088630e-01 8.36905241e-01 3.91879737e-01
1.23583265e-01 -1.02128446e+00 -9.95182395e-01 -9.17499661e-01
1.33888975e-01 -2.22418711e-01 6.67881787e-01 1.30022049e+00
2.12513030e-01 4.63724613e-01 -3.26484352e-01 -3.97554219e-01
8.92487824e-01 2.33183861e-01 5.08143187e-01 -1.50450397e+00
-1.66465908e-01 -5.09798646e-01 -1.22388273e-01 -7.03616977e-01
4.69999135e-01 -1.10056973e+00 1.27555966e-01 -1.38639784e+00
4.69831586e-01 -6.75448656e-01 -4.65486497e-01 4.35750842e-01
-4.84955072e-01 -1.09065831e-01 7.71522820e-02 3.72115612e-01
-8.31919909e-01 8.25142860e-01 1.23533392e+00 -2.30847001e-01
-9.90612879e-02 -6.28347322e-02 -8.10807467e-01 5.83384514e-01
1.05182552e+00 -7.93503404e-01 -5.85586965e-01 -5.28981462e-02
-4.20337096e-02 -5.29842414e-02 9.87058878e-03 -9.35084224e-01
4.28237379e-01 -2.72582054e-01 3.15925688e-01 -5.34603596e-01
1.23210683e-01 -7.42361546e-01 -1.94446847e-01 5.30380487e-01
-5.62381208e-01 -2.36333907e-01 -3.48826349e-01 8.55781913e-01
-1.68620497e-01 -9.99784470e-01 8.86784554e-01 -2.31793672e-01
-6.60664678e-01 4.33771878e-01 -1.31904885e-01 5.59244990e-01
1.07992291e+00 -9.04753059e-02 -4.78960454e-01 5.36483638e-02
-2.23451227e-01 3.42205524e-01 3.24627876e-01 4.51649666e-01
5.51444530e-01 -1.44987905e+00 -6.77611351e-01 4.85369414e-02
4.98277128e-01 1.69607215e-02 4.83786196e-01 7.13132501e-01
-1.92007095e-01 2.56002128e-01 8.32123235e-02 -3.72448027e-01
-8.60117912e-01 8.60361695e-01 1.17831836e-02 -1.12342894e-01
-7.50759780e-01 5.90536475e-01 -2.17735693e-01 -6.84292197e-01
3.22559983e-01 1.87134370e-02 -3.58418971e-01 3.91725361e-01
7.50669122e-01 6.61525726e-01 -7.84687027e-02 -2.62150675e-01
-1.80718899e-01 1.54245794e-01 -4.10007894e-01 9.82953683e-02
1.30709398e+00 2.26548426e-02 7.13980049e-02 9.31854069e-01
1.37842751e+00 -5.08388221e-01 -1.04031181e+00 -7.05558300e-01
1.56345934e-01 -5.58759511e-01 3.36825941e-03 -4.34990883e-01
-9.38635528e-01 1.00372326e+00 3.12202871e-01 3.29739481e-01
6.40186548e-01 -2.64487088e-01 9.06853080e-01 5.16166925e-01
4.61779505e-01 -1.44728684e+00 1.06984347e-01 4.74127591e-01
3.17319274e-01 -1.32469893e+00 3.80654186e-02 -7.85340592e-02
-7.47971058e-01 1.14882278e+00 7.51355290e-01 -4.84334752e-02
8.65146756e-01 -2.36164868e-01 -2.03403950e-01 1.02702305e-01
-9.79083180e-01 -4.72546704e-02 -8.44852813e-03 4.47786152e-01
2.70982057e-01 1.38764828e-01 -4.12845165e-01 7.96720207e-01
2.36086816e-01 6.56847060e-02 5.13601482e-01 1.08882499e+00
-5.51599026e-01 -1.03684163e+00 -1.40660226e-01 9.29687858e-01
-1.65973961e-01 -5.02547063e-03 -1.18609883e-01 6.09698594e-01
3.65500376e-02 8.79624605e-01 -1.72205389e-01 -5.14961064e-01
2.62294829e-01 3.84739637e-01 1.04602218e-01 -8.72780859e-01
-3.58432174e-01 -3.06325436e-01 -9.80990380e-02 -2.15011891e-02
-2.31765479e-01 -4.66013223e-01 -9.47212875e-01 -3.19566816e-01
-5.00811577e-01 5.14257073e-01 4.67008144e-01 9.19095755e-01
3.09882313e-01 4.03640479e-01 9.13413882e-01 -9.31588888e-01
-1.19297075e+00 -1.17412043e+00 -9.73922312e-01 4.89149660e-01
1.80561915e-02 -8.20955753e-01 -7.74751663e-01 -4.65137571e-01]
|
[10.166346549987793, 3.4900448322296143]
|
a56d87f1-84ab-41b2-b08f-9f9b8ef29428
|
explore-the-power-of-synthetic-data-on-few
|
2303.13221
| null |
https://arxiv.org/abs/2303.13221v2
|
https://arxiv.org/pdf/2303.13221v2.pdf
|
Explore the Power of Synthetic Data on Few-shot Object Detection
|
Few-shot object detection (FSOD) aims to expand an object detector for novel categories given only a few instances for training. The few training samples restrict the performance of FSOD model. Recent text-to-image generation models have shown promising results in generating high-quality images. How applicable these synthetic images are for FSOD tasks remains under-explored. This work extensively studies how synthetic images generated from state-of-the-art text-to-image generators benefit FSOD tasks. We focus on two perspectives: (1) How to use synthetic data for FSOD? (2) How to find representative samples from the large-scale synthetic dataset? We design a copy-paste-based pipeline for using synthetic data. Specifically, saliency object detection is applied to the original generated image, and the minimum enclosing box is used for cropping the main object based on the saliency map. After that, the cropped object is randomly pasted on the image, which comes from the base dataset. We also study the influence of the input text of text-to-image generator and the number of synthetic images used. To construct a representative synthetic training dataset, we maximize the diversity of the selected images via a sample-based and cluster-based method. However, the severe problem of high false positives (FP) ratio of novel categories in FSOD can not be solved by using synthetic data. We propose integrating CLIP, a zero-shot recognition model, into the FSOD pipeline, which can filter 90% of FP by defining a threshold for the similarity score between the detected object and the text of the predicted category. Extensive experiments on PASCAL VOC and MS COCO validate the effectiveness of our method, in which performance gain is up to 21.9% compared to the few-shot baseline.
|
['Rui Zhao', 'Xingyu Zeng', 'Kun Wang', 'Shaobo Lin']
|
2023-03-23
| null | null | null | null |
['few-shot-object-detection']
|
['computer-vision']
|
[ 6.00951314e-01 9.24401730e-02 9.18359458e-02 -1.80349499e-01
-7.99255371e-01 -2.01176986e-01 6.82571471e-01 -1.64747193e-01
-3.64406198e-01 5.38050950e-01 -1.80686578e-01 3.17513824e-01
2.54599899e-01 -8.19484353e-01 -9.49434459e-01 -7.54552066e-01
4.99953628e-01 2.73892403e-01 8.75048220e-01 -8.87181833e-02
3.91021520e-01 2.21716031e-01 -2.19829059e+00 5.14843047e-01
1.15506303e+00 8.68295014e-01 8.83533835e-01 5.42679071e-01
-2.29072869e-01 4.97535616e-01 -1.10142994e+00 -1.17198735e-01
3.07567656e-01 -9.57095146e-01 -3.18279356e-01 3.28670114e-01
4.38339263e-01 -2.91756064e-01 4.49028909e-02 1.03450978e+00
6.46578193e-01 1.71406806e-01 7.09234238e-01 -1.44712627e+00
-4.46078420e-01 4.65782255e-01 -5.90766311e-01 2.58491844e-01
1.00871839e-01 5.47491610e-01 6.32138491e-01 -1.29928446e+00
9.84490514e-01 1.15237260e+00 1.47309363e-01 7.05630541e-01
-1.18703389e+00 -7.15912879e-01 -2.08514974e-01 1.86292350e-01
-1.32979858e+00 -4.26399320e-01 5.38960814e-01 -3.66565883e-01
5.45076549e-01 3.10553908e-01 5.43163538e-01 1.19414735e+00
-6.30251318e-02 9.63451326e-01 9.85218287e-01 -7.41677940e-01
4.75569457e-01 5.10082126e-01 -5.07950895e-02 3.22412372e-01
3.63863528e-01 5.09855300e-02 -6.36862457e-01 2.13926509e-01
4.91027594e-01 -8.57127532e-02 -2.05805451e-01 -3.96496117e-01
-1.27928746e+00 7.02442288e-01 4.36062098e-01 1.96686506e-01
-2.75255889e-01 -1.59342960e-01 1.62064925e-01 -2.05384061e-01
2.73431093e-01 6.69631362e-01 -5.89732863e-02 6.93696588e-02
-1.22398055e+00 5.45200109e-01 4.88939255e-01 1.17942107e+00
7.01189995e-01 1.32124245e-01 -9.22704816e-01 9.96184647e-01
-2.79420882e-01 6.91632390e-01 5.56317091e-01 -7.30171621e-01
2.75531054e-01 7.71230340e-01 8.30827877e-02 -7.05578327e-01
1.20746516e-01 -3.38535339e-01 -3.77525568e-01 2.86732167e-01
2.74534345e-01 -1.16840154e-01 -1.24145103e+00 1.43592060e+00
4.52936292e-01 1.23146005e-01 -3.78855281e-02 9.71134424e-01
8.29705179e-01 8.23706567e-01 -1.20105259e-01 -2.16799855e-01
1.20406532e+00 -1.16725874e+00 -5.73004127e-01 -4.05676574e-01
4.94175762e-01 -7.06712067e-01 1.33151233e+00 5.45600951e-02
-7.71683753e-01 -6.89433277e-01 -1.07386911e+00 2.24582374e-01
-4.03309196e-01 4.52630669e-01 -7.76224071e-03 4.14432555e-01
-6.22962773e-01 3.06398004e-01 -4.36002910e-01 -5.10435641e-01
5.54663002e-01 -3.44163366e-02 6.83310926e-02 -2.11586460e-01
-9.87631977e-01 6.92721963e-01 7.79119492e-01 -3.83969188e-01
-1.19818664e+00 -6.60888851e-01 -6.77193463e-01 2.82976776e-02
6.41467690e-01 -4.62946892e-01 1.00562227e+00 -1.13395524e+00
-1.02436423e+00 7.06566930e-01 -9.47018489e-02 -4.18332696e-01
4.57077026e-01 3.43740359e-02 -1.97822273e-01 2.81901091e-01
5.71680367e-01 1.16301966e+00 1.19961131e+00 -1.35763371e+00
-8.56251538e-01 -3.99284102e-02 -2.21408397e-01 1.49168283e-01
-1.76596344e-01 1.18002586e-01 -5.50107837e-01 -8.20488214e-01
-2.36707836e-01 -8.83065701e-01 -1.10406704e-01 9.19165164e-02
-4.52138305e-01 -6.76302761e-02 9.98694897e-01 -3.73499930e-01
1.08888578e+00 -2.47861242e+00 -2.86742270e-01 -2.31688827e-01
-1.65161505e-01 5.94183922e-01 -2.97977120e-01 2.75812060e-01
6.46047369e-02 5.85999899e-02 -2.90838391e-01 -1.91771701e-01
-2.05273256e-01 -8.02474469e-02 -3.86486650e-01 -1.51141226e-01
6.54590845e-01 9.19240117e-01 -9.71405447e-01 -6.28489256e-01
2.92277932e-01 4.80827428e-02 -3.40449005e-01 3.86148632e-01
-5.19539058e-01 8.50436091e-02 -2.90828466e-01 6.00756705e-01
7.42051244e-01 -1.24059081e-01 -3.09609473e-01 -7.06004128e-02
-5.19499630e-02 -3.04032844e-02 -1.28391075e+00 1.41349435e+00
-3.51472609e-02 5.35336733e-01 -4.56324100e-01 -6.27718925e-01
1.12320971e+00 -2.65038401e-01 -1.57016516e-02 -7.22414792e-01
7.77789950e-02 2.17475623e-01 1.05596833e-01 -4.97962981e-01
6.46117985e-01 -3.96148413e-02 3.82870436e-02 4.09508884e-01
2.02487990e-01 -3.02668273e-01 7.04739928e-01 4.43860680e-01
9.00767565e-01 2.62736529e-01 1.73451260e-01 -1.14341654e-01
1.84034824e-01 4.11985755e-01 5.84933877e-01 9.62661088e-01
-7.65809491e-02 1.14791942e+00 6.08465612e-01 2.70001311e-02
-1.23409963e+00 -9.15746391e-01 -1.74893349e-01 1.05324256e+00
4.21944410e-01 -7.71443844e-02 -1.14826000e+00 -8.10095966e-01
-1.41794577e-01 1.29458582e+00 -6.87982678e-01 -3.64297003e-01
-1.81155398e-01 -7.14614689e-01 4.00830746e-01 2.69897997e-01
5.08882105e-01 -1.46628225e+00 -9.67517316e-01 1.73501849e-01
-1.20323807e-01 -1.05964005e+00 -5.70609808e-01 3.26279663e-02
-4.92157876e-01 -1.00852442e+00 -8.80611062e-01 -9.01459932e-01
8.07456076e-01 7.23845184e-01 7.63151884e-01 -1.41409606e-01
-5.43291032e-01 -7.78016448e-02 -6.94631577e-01 -7.20720708e-01
-6.88372493e-01 -1.49954051e-01 -1.80020556e-01 2.71086633e-01
3.09658915e-01 6.71987887e-03 -5.52642643e-01 5.59431374e-01
-1.06094158e+00 4.13311303e-01 6.60328150e-01 8.71097088e-01
5.81900358e-01 -6.09899051e-02 6.42061889e-01 -6.59251630e-01
4.23991770e-01 -2.59644061e-01 -4.47111666e-01 2.41636634e-01
-3.79679471e-01 1.79325771e-02 5.80844820e-01 -7.70744979e-01
-1.07367611e+00 6.12519495e-02 3.70350778e-01 -7.88272321e-01
-1.72597021e-01 -9.90283340e-02 -1.57404214e-01 2.71219492e-01
1.09534180e+00 5.30321956e-01 1.63126305e-01 -1.37448341e-01
4.22580004e-01 8.80224347e-01 4.38687891e-01 -2.36704990e-01
6.99693501e-01 4.39039975e-01 -4.26347971e-01 -8.77257645e-01
-7.83494234e-01 -3.58215928e-01 -4.21393961e-01 -3.04392815e-01
7.71401882e-01 -8.60375106e-01 1.44856408e-01 5.40426433e-01
-1.05320537e+00 -3.14688921e-01 -6.58988059e-01 3.01239580e-01
-4.12419289e-01 1.13483869e-01 -1.32590204e-01 -7.96348810e-01
-4.31069255e-01 -1.27907121e+00 1.29733264e+00 4.17749852e-01
-2.65163109e-02 -1.48100168e-01 -3.05969417e-01 2.50067323e-01
2.11259216e-01 3.24956775e-01 6.26848578e-01 -7.55259752e-01
-7.12591171e-01 -1.84201211e-01 -3.19328964e-01 5.31848848e-01
-8.47504884e-02 1.33556232e-01 -9.88906384e-01 -1.89701185e-01
-1.53455392e-01 -3.27430904e-01 9.21595573e-01 2.47553095e-01
1.11725235e+00 -1.40639916e-01 -5.83684027e-01 2.52716303e-01
1.26730180e+00 3.41484785e-01 6.66653216e-01 2.20742911e-01
4.87245113e-01 6.65033221e-01 1.19062483e+00 3.36841702e-01
-9.09038037e-02 7.76369870e-01 2.62379795e-01 -5.52249365e-02
-6.11352921e-01 -4.10887241e-01 3.79917860e-01 2.56405085e-01
4.22649831e-01 -3.92260194e-01 -6.94707870e-01 7.69056559e-01
-1.68345475e+00 -1.02910304e+00 -9.03733224e-02 2.36913300e+00
7.47569323e-01 3.51253867e-01 1.73200503e-01 2.96044592e-02
1.19113410e+00 9.61843692e-03 -5.77404797e-01 -5.73612973e-02
-2.16335982e-01 6.92325011e-02 1.68114826e-01 -1.25346735e-01
-8.31715524e-01 1.08404171e+00 5.27671146e+00 1.35617769e+00
-1.16679025e+00 1.29554898e-01 6.20468199e-01 -4.00810570e-01
-6.81997240e-02 1.10209085e-01 -1.23097134e+00 8.30732405e-01
6.34687662e-01 -3.69047761e-01 1.94496870e-01 1.06240523e+00
3.13936591e-01 -5.23733675e-01 -8.81805003e-01 8.98325682e-01
5.34774005e-01 -1.26292872e+00 3.31006020e-01 -1.16988257e-01
1.01977611e+00 -9.32453498e-02 -1.42917335e-01 3.32763940e-01
-1.10771030e-01 -6.52663231e-01 9.94632781e-01 3.66128683e-01
9.93504286e-01 -5.58017373e-01 5.26381969e-01 6.21778190e-01
-8.63145292e-01 -1.60905153e-01 -5.58620930e-01 2.59348601e-01
1.11237548e-01 8.40277731e-01 -1.31128204e+00 2.48211950e-01
6.77743256e-01 3.70960116e-01 -9.61085200e-01 1.18109167e+00
-2.46862888e-01 5.42151928e-01 -2.66422808e-01 -3.62091810e-01
1.31781921e-01 -2.96341348e-02 7.48869002e-01 1.16172874e+00
5.28233826e-01 -1.74509883e-01 -6.12534210e-02 1.41547632e+00
6.07901365e-02 1.02462493e-01 -6.62027121e-01 -5.27471080e-02
6.13543391e-01 1.28820765e+00 -9.18617249e-01 -6.11565292e-01
-1.59475341e-01 9.52786863e-01 1.20385952e-01 2.43257254e-01
-9.04757798e-01 -6.61343515e-01 1.06464639e-01 4.38142031e-01
6.30732596e-01 3.62077743e-01 -2.06177771e-01 -9.58358109e-01
2.64657646e-01 -1.00292265e+00 1.08699054e-01 -1.16844106e+00
-9.33271945e-01 5.02648652e-01 1.99349001e-01 -1.51440382e+00
-1.41907603e-01 -2.52842218e-01 -8.22766840e-01 7.12389112e-01
-9.23637331e-01 -1.05792105e+00 -6.43135846e-01 1.41170338e-01
9.88454282e-01 -1.74040854e-01 4.22179341e-01 -2.22766344e-02
-5.98893404e-01 4.89398420e-01 9.83578190e-02 4.01346795e-02
7.00402081e-01 -9.91306007e-01 5.69056928e-01 1.23268402e+00
1.81312978e-01 2.84101754e-01 6.70338452e-01 -7.92299449e-01
-9.78832901e-01 -1.42539990e+00 5.97865820e-01 -3.84572536e-01
1.67658329e-01 -7.03648329e-01 -1.02905548e+00 1.14017636e-01
-3.95784788e-02 9.35543850e-02 7.25052357e-02 -5.91650903e-01
2.57781763e-02 -1.46428451e-01 -1.13331425e+00 7.74990201e-01
9.50532436e-01 2.21089423e-02 -7.10526168e-01 2.32272521e-01
9.76887465e-01 -2.33746350e-01 -2.30088726e-01 4.23480600e-01
2.96373755e-01 -1.05767429e+00 7.11557150e-01 -1.32351547e-01
6.60477161e-01 -6.61690354e-01 1.00838043e-01 -1.35382128e+00
-1.15148768e-01 -3.63115370e-01 1.04359165e-01 1.45374846e+00
4.29588407e-01 -2.08309382e-01 5.22095621e-01 2.23198950e-01
-6.58107176e-02 -7.37144053e-01 -6.52327716e-01 -8.64125192e-01
-4.15702701e-01 -1.78101107e-01 4.77484822e-01 5.96652091e-01
-3.40852857e-01 4.14691895e-01 -1.71880141e-01 -1.94659427e-01
5.73206246e-01 1.24001846e-01 1.21396685e+00 -7.63568938e-01
-1.83281735e-01 -1.91101283e-01 -3.52834970e-01 -6.87934279e-01
-3.28402936e-01 -7.60931015e-01 3.79459351e-01 -1.40820682e+00
5.51319718e-01 -2.88207561e-01 4.59930748e-02 3.48559707e-01
-5.46880484e-01 2.78027415e-01 4.78865236e-01 1.58115849e-01
-6.33848071e-01 7.23242879e-01 1.40705776e+00 -1.60445899e-01
-2.98357576e-01 -1.22893028e-01 -4.96827692e-01 3.33325714e-01
6.06303930e-01 -6.61299646e-01 -3.76506597e-01 2.92290691e-02
-3.08158129e-01 -1.80073768e-01 4.20255601e-01 -1.16626728e+00
1.56394746e-02 -2.55025506e-01 3.61491024e-01 -8.00023198e-01
1.90682963e-01 -3.31013948e-01 9.02116373e-02 4.71555114e-01
-1.57133922e-01 -4.95432496e-01 1.26719639e-01 5.50394475e-01
-6.83602989e-02 -6.13371313e-01 1.17763507e+00 -1.68660030e-01
-8.91722441e-01 -2.78167017e-02 -1.77060977e-01 1.99471608e-01
1.45940161e+00 -4.22529578e-01 -3.62869591e-01 -3.97721082e-02
-2.13776648e-01 1.08911410e-01 6.16739929e-01 6.75463021e-01
6.90957963e-01 -1.16357875e+00 -8.30158472e-01 4.16108429e-01
6.71659052e-01 1.76476911e-01 1.51874274e-01 6.13243163e-01
-3.95158499e-01 6.53327778e-02 -1.75549582e-01 -8.47371101e-01
-1.10769415e+00 8.28339994e-01 1.95535883e-01 1.22720391e-01
-6.07038796e-01 6.79547548e-01 4.48377579e-01 -1.81005925e-01
4.23673913e-02 -1.23254269e-01 -9.00896825e-03 1.24119349e-01
7.25609004e-01 4.27874058e-01 -3.58865294e-03 -3.95877808e-01
-2.20727473e-01 1.63580775e-01 -2.47534066e-01 -1.63090587e-01
1.09446836e+00 1.53478548e-01 1.40534848e-01 5.12854695e-01
8.18201840e-01 -3.07228416e-01 -1.37405694e+00 -3.76707837e-02
-2.18513295e-01 -6.04225159e-01 -1.71869844e-01 -7.92199254e-01
-6.86148703e-01 7.25037634e-01 6.99546516e-01 -7.93980956e-02
1.01343608e+00 6.47188053e-02 6.92192733e-01 9.44596156e-03
3.69785935e-01 -1.29227710e+00 4.35434520e-01 2.15047181e-01
9.99106467e-01 -1.24825752e+00 -1.28351882e-01 -4.88599747e-01
-9.50319827e-01 7.08670199e-01 9.94169533e-01 -6.08680993e-02
6.71987832e-02 1.22698739e-01 -1.00020848e-01 4.48456109e-02
-7.48785436e-01 -3.42488229e-01 2.07704559e-01 6.57966077e-01
-1.50713757e-01 -6.10627495e-02 -4.49274749e-01 5.09854734e-01
-1.19054273e-01 9.74701494e-02 7.80569911e-01 8.49667311e-01
-8.14914405e-01 -6.89030170e-01 -4.88988429e-01 7.59481907e-01
-1.01260699e-01 -1.48733050e-01 -4.00802493e-01 5.59419990e-01
3.91535699e-01 8.81909192e-01 1.70536205e-01 -2.87442207e-01
3.77568334e-01 2.74285451e-02 8.83937478e-02 -1.02565169e+00
-3.50286722e-01 4.54600975e-02 -1.79251328e-01 -2.44605139e-01
-1.17135927e-01 -6.29840791e-01 -1.10797238e+00 3.10834378e-01
-7.75297821e-01 1.66067318e-03 6.35084033e-01 7.17449307e-01
5.78854799e-01 5.32376409e-01 7.20461249e-01 -9.75933433e-01
-5.65534234e-01 -1.24535465e+00 -5.85745513e-01 6.36310160e-01
-9.77372527e-02 -8.47263277e-01 -5.93562186e-01 1.87266156e-01]
|
[9.586894989013672, 1.719116449356079]
|
4b198f67-632c-4c27-acb6-3d5cb51a8c6a
|
cnn-based-density-estimation-and-crowd
|
2003.12783
| null |
https://arxiv.org/abs/2003.12783v1
|
https://arxiv.org/pdf/2003.12783v1.pdf
|
CNN-based Density Estimation and Crowd Counting: A Survey
|
Accurately estimating the number of objects in a single image is a challenging yet meaningful task and has been applied in many applications such as urban planning and public safety. In the various object counting tasks, crowd counting is particularly prominent due to its specific significance to social security and development. Fortunately, the development of the techniques for crowd counting can be generalized to other related fields such as vehicle counting and environment survey, if without taking their characteristics into account. Therefore, many researchers are devoting to crowd counting, and many excellent works of literature and works have spurted out. In these works, they are must be helpful for the development of crowd counting. However, the question we should consider is why they are effective for this task. Limited by the cost of time and energy, we cannot analyze all the algorithms. In this paper, we have surveyed over 220 works to comprehensively and systematically study the crowd counting models, mainly CNN-based density map estimation methods. Finally, according to the evaluation metrics, we select the top three performers on their crowd counting datasets and analyze their merits and drawbacks. Through our analysis, we expect to make reasonable inference and prediction for the future development of crowd counting, and meanwhile, it can also provide feasible solutions for the problem of object counting in other fields. We provide the density maps and prediction results of some mainstream algorithm in the validation set of NWPU dataset for comparison and testing. Meanwhile, density map generation and evaluation tools are also provided. All the codes and evaluation results are made publicly available at https://github.com/gaoguangshuai/survey-for-crowd-counting.
|
['Junyu. Gao', 'Qi. Wang', 'Qingjie Liu', 'Yunhong Wang', 'Guangshuai Gao']
|
2020-03-28
| null | null | null | null |
['object-counting']
|
['computer-vision']
|
[-3.84611964e-01 -4.94198620e-01 -2.83486992e-01 -3.65223467e-01
-1.89589322e-01 -1.14635609e-01 4.99188930e-01 8.00422803e-02
-6.84491992e-01 8.72266173e-01 1.14504591e-01 -4.31224614e-01
2.01918557e-01 -1.20037258e+00 -2.31341049e-01 -7.49544799e-01
8.87187570e-02 4.42732841e-01 5.72852552e-01 -9.08648893e-02
2.55271852e-01 2.36930549e-01 -1.68795216e+00 -4.14784789e-01
1.04812193e+00 8.80254924e-01 3.44460130e-01 4.17428970e-01
-3.64814103e-01 9.94756520e-01 -7.17691064e-01 -7.20744371e-01
-2.19664052e-01 -1.92166224e-01 -4.72361535e-01 -3.16570818e-01
2.51899302e-01 -8.55216384e-01 -5.30326068e-01 1.20646584e+00
7.06624448e-01 2.82506365e-02 6.62123084e-01 -1.31943393e+00
-6.46817684e-01 2.97391176e-01 -9.29436803e-01 6.84305131e-01
1.03993215e-01 2.09706858e-01 4.30937886e-01 -8.14719737e-01
-4.51488383e-02 1.00569463e+00 5.00781715e-01 4.68548387e-01
-7.80018568e-02 -1.12841117e+00 8.97710770e-02 3.75478804e-01
-1.58986783e+00 -3.42342287e-01 4.53159124e-01 -6.06199861e-01
6.07033014e-01 1.73815668e-01 8.45402122e-01 5.74418783e-01
-2.91373700e-01 1.02489293e+00 6.84167922e-01 -1.33013770e-01
1.90474302e-01 4.36606891e-02 2.07661107e-01 7.85787702e-01
8.74246299e-01 -1.63000882e-01 -1.19272001e-01 7.79237673e-02
5.90671182e-01 2.38543198e-01 -1.66608375e-02 2.77519494e-01
-1.02854180e+00 8.87199998e-01 5.10518670e-01 3.26132506e-01
1.79085452e-02 1.55718341e-01 4.67796266e-01 -4.22083199e-01
8.40134978e-01 -3.06842953e-01 1.19678266e-01 -4.04432714e-01
-9.31383371e-01 4.46031004e-01 4.94877726e-01 1.16490006e+00
8.23294818e-01 6.99875429e-02 -4.22251105e-01 6.70180976e-01
1.13682188e-01 9.80198741e-01 1.30256370e-01 -7.70968795e-01
8.62019479e-01 6.65329099e-01 2.66234785e-01 -1.43007696e+00
-4.49702144e-01 6.53600320e-02 -1.15840292e+00 -3.15193325e-01
4.46715474e-01 -3.91122013e-01 -4.31773603e-01 1.25271749e+00
5.05047083e-01 3.74515384e-01 -4.35017258e-01 8.65534186e-01
1.25620055e+00 7.47855484e-01 3.04898679e-01 -1.70310065e-01
1.55460572e+00 -7.37891376e-01 -6.91733599e-01 -1.55406147e-01
5.99873304e-01 -4.74517554e-01 9.10482883e-01 -1.53702840e-01
-7.84657776e-01 -4.86552387e-01 -8.46293032e-01 -4.44411822e-02
-4.40210640e-01 3.41639310e-01 9.51482117e-01 8.20113122e-01
-6.38262928e-01 6.78976774e-02 -8.82644951e-01 -4.71143782e-01
1.00716150e+00 2.12355256e-01 1.12174273e-01 -2.14320138e-01
-1.17358041e+00 8.62190962e-01 2.37989902e-01 1.96936592e-01
-4.42703605e-01 -2.76077002e-01 -8.29941273e-01 -8.99865478e-02
1.86863899e-01 -6.10133111e-01 1.26063156e+00 -1.39819697e-01
-8.85477960e-01 6.97929800e-01 -5.84225178e-01 -1.70733973e-01
6.98079824e-01 7.96605796e-02 -4.71398473e-01 -8.67896825e-02
6.03626609e-01 5.54111004e-01 -5.80473896e-03 -9.18211937e-01
-1.11473072e+00 -2.87111402e-01 1.36527747e-01 -4.27130163e-02
-6.68568611e-01 2.54659146e-01 -5.79700351e-01 -2.43303165e-01
-3.45637769e-01 -3.92766237e-01 -2.23918840e-01 -2.30299383e-01
-3.67597997e-01 -5.46672881e-01 8.44047129e-01 -6.96234703e-01
1.58948362e+00 -1.88535929e+00 -7.16794610e-01 -7.48241916e-02
5.66268802e-01 5.75579464e-01 1.62936449e-01 1.02393314e-01
7.08525062e-01 2.24258885e-01 -2.09986299e-01 -4.66606975e-01
-2.08320752e-01 -8.65431204e-02 -4.47795987e-02 6.88672423e-01
1.53988272e-01 1.14634573e+00 -1.19885707e+00 -1.02098238e+00
5.88273823e-01 5.39299726e-01 -2.16702461e-01 6.51453063e-02
9.18192640e-02 6.29806161e-01 -7.86694825e-01 8.96733582e-01
1.12008893e+00 -2.75048494e-01 -3.57171506e-01 2.06967276e-02
-3.76509011e-01 -1.20290443e-01 -1.01353192e+00 9.40774441e-01
-3.50915819e-01 5.84137678e-01 -3.97724211e-01 -8.56517375e-01
9.81954932e-01 3.15701701e-02 3.01458180e-01 -8.44311535e-01
5.60345471e-01 5.08128285e-01 -1.50117844e-01 -6.68082654e-01
7.58925378e-01 7.02964291e-02 -1.31966978e-01 2.76215255e-01
-3.04708391e-01 2.43551083e-04 7.35138357e-01 1.41860038e-01
7.47300148e-01 -3.98040593e-01 6.32526100e-01 -1.21597126e-01
6.65416837e-01 8.32170993e-02 4.25037086e-01 5.25748193e-01
-6.93669975e-01 4.52677041e-01 2.47923493e-01 -5.96972108e-01
-1.01230109e+00 -6.92327142e-01 -2.63316065e-01 7.53193021e-01
4.57530409e-01 -7.58796756e-04 -7.94920921e-01 -4.35308605e-01
-9.37126130e-02 4.17631030e-01 -4.63585049e-01 2.01945513e-01
-7.78584719e-01 -9.22501266e-01 7.48577595e-01 7.48769224e-01
1.23573411e+00 -1.09904921e+00 -4.06061590e-01 -8.27917010e-02
-4.97748554e-01 -1.24473584e+00 -4.83332276e-01 -6.16203725e-01
-6.26845777e-01 -1.17930496e+00 -1.16065180e+00 -8.45150471e-01
5.36537707e-01 9.24995005e-01 1.11136377e+00 6.83305562e-01
3.16368081e-02 4.23790962e-02 -2.38435507e-01 -9.87173021e-01
4.03623246e-02 4.19861704e-01 8.85268599e-02 -3.90621066e-01
1.09475207e+00 -4.00347292e-01 -8.13857734e-01 5.19798398e-01
-5.33326030e-01 -1.11259505e-01 2.16420442e-01 2.64025599e-01
2.60995597e-01 1.13847375e-01 7.43343472e-01 -6.43219292e-01
5.79865217e-01 -7.97334492e-01 -8.58627439e-01 1.17017829e-03
-6.41538352e-02 -5.75007260e-01 5.50693333e-01 -2.33796686e-01
-1.03452778e+00 -3.20434421e-01 -3.48879486e-01 4.57499735e-03
-1.14451401e-01 1.30056888e-01 -2.74969369e-01 6.30908608e-02
3.78444105e-01 2.91196734e-01 -2.84498960e-01 -1.12688437e-01
-9.85869914e-02 8.43431890e-01 2.83811092e-01 -3.44818771e-01
7.60607183e-01 7.21169889e-01 -6.82929903e-02 -8.95777464e-01
-9.91919935e-01 -7.78849661e-01 -3.31262529e-01 -4.33446288e-01
9.54110980e-01 -1.05689240e+00 -1.20921516e+00 7.27409244e-01
-1.47058952e+00 5.31770401e-02 8.85769725e-02 5.30078471e-01
-1.01747088e-01 5.13384700e-01 -5.00240088e-01 -1.30728018e+00
-2.59635806e-01 -1.05443454e+00 1.08063769e+00 8.52945387e-01
1.06611438e-01 -1.13261116e+00 1.43446133e-01 3.66704017e-01
6.38708115e-01 -4.03503105e-02 4.11640018e-01 -3.24232519e-01
-7.46607721e-01 -3.21696728e-01 -6.83264613e-01 1.96738601e-01
-1.03330821e-01 4.69739176e-02 -1.08307230e+00 9.35266092e-02
-2.93585569e-01 -9.01397914e-02 9.95203376e-01 7.31769264e-01
1.41638911e+00 -1.77591592e-01 -5.61191320e-01 4.89445537e-01
1.27775466e+00 1.25406846e-01 9.47599649e-01 3.56166929e-01
9.14513052e-01 6.09568834e-01 6.52004540e-01 5.63605547e-01
9.76141751e-01 3.53057444e-01 4.31675494e-01 -1.61398411e-01
1.10481516e-01 -5.91889143e-01 -8.09305608e-02 9.01974320e-01
-7.95481563e-01 -3.94119531e-01 -1.04887414e+00 5.74745417e-01
-1.75193167e+00 -1.42093718e+00 -4.89387691e-01 1.96387529e+00
3.99650097e-01 -1.82001159e-01 5.74139714e-01 1.42322674e-01
1.25571835e+00 4.16501880e-01 -3.37166756e-01 2.53376126e-01
-1.20042570e-01 -1.01921730e-01 6.78399146e-01 3.24227720e-01
-1.20664823e+00 8.75320137e-01 5.42864180e+00 1.23384035e+00
-7.53594756e-01 2.45198920e-01 9.86784697e-01 6.33315146e-02
1.58733521e-02 -3.97859067e-01 -1.31674159e+00 9.91491139e-01
5.21178186e-01 -1.84927925e-01 2.23374158e-01 9.24390554e-01
3.98807049e-01 -5.48412085e-01 -4.61757034e-01 1.21201241e+00
-4.11147587e-02 -1.38944411e+00 -9.24537629e-02 9.97302234e-02
7.84844875e-01 -2.24091783e-02 -1.34835929e-01 5.48016250e-01
2.36725405e-01 -1.13836610e+00 5.39288998e-01 4.27020729e-01
7.82821774e-01 -8.46615613e-01 1.32255185e+00 8.26701760e-01
-1.59379554e+00 1.17845051e-01 -8.99329185e-01 -4.72112387e-01
4.75395709e-01 9.84530568e-01 -4.62863654e-01 3.33060831e-01
6.69002175e-01 5.88966906e-01 -4.38119054e-01 1.37752032e+00
-2.08219349e-01 6.25112712e-01 -2.99618542e-01 -7.32586682e-01
-4.05292474e-02 -2.28717834e-01 1.90489646e-02 1.22840822e+00
6.59844518e-01 1.52898327e-01 1.32446662e-01 7.43307769e-01
-1.33328184e-01 2.41778538e-01 -7.35386729e-01 2.86032289e-01
8.62961471e-01 1.34888864e+00 -8.64026427e-01 -4.70216095e-01
-5.68663538e-01 3.28047961e-01 3.56668323e-01 1.88430607e-01
-1.17625773e+00 -3.08272630e-01 4.66391534e-01 4.42333251e-01
-2.76937950e-02 -2.21654713e-01 -4.33772743e-01 -1.15518439e+00
1.73514783e-01 -2.28942961e-01 1.14836723e-01 -3.40145469e-01
-1.29762650e+00 3.17924321e-01 3.89898717e-01 -1.21622872e+00
1.52865395e-01 -4.71557200e-01 -1.09281874e+00 7.96937704e-01
-1.68767917e+00 -9.51411188e-01 -9.35486495e-01 2.63443559e-01
3.02717060e-01 -1.98912144e-01 3.26760322e-01 7.57209897e-01
-7.00659811e-01 5.25606751e-01 -1.33153975e-01 6.93956614e-01
1.45254910e-01 -6.48487806e-01 6.71518445e-01 7.98403442e-01
-2.44649827e-01 3.36988688e-01 3.12154621e-01 -6.05695248e-01
-7.87179112e-01 -1.32768607e+00 1.14421511e+00 -6.83158159e-01
4.18292552e-01 -3.00705194e-01 -6.44346416e-01 4.06330734e-01
-8.66373107e-02 2.76105046e-01 3.60348225e-01 -1.20866187e-01
2.01482624e-01 4.77682799e-02 -1.10848844e+00 4.60748613e-01
1.20303941e+00 -8.73216093e-02 3.28079313e-02 3.35021049e-01
4.87470120e-01 -2.68273979e-01 -2.95577675e-01 3.01077724e-01
4.24045265e-01 -1.34779739e+00 8.47769141e-01 -5.70322797e-02
5.94065309e-01 -4.49694514e-01 -4.19129943e-03 -8.52122724e-01
-2.25876227e-01 8.49792510e-02 -2.13179752e-01 1.51052713e+00
4.70639728e-02 -8.06944489e-01 1.05632782e+00 2.90743232e-01
4.05477099e-02 -7.77002692e-01 -9.56051350e-01 -6.45489991e-01
1.80372894e-01 -5.20961821e-01 1.07781529e+00 6.01823807e-01
-1.95700020e-01 3.00480425e-01 -4.67347592e-01 4.71481830e-02
5.13448894e-01 -3.28639507e-01 1.15847027e+00 -1.03279829e+00
3.08015168e-01 -4.68655109e-01 -5.74939132e-01 -1.43447983e+00
3.06646656e-02 -4.88095552e-01 -9.71471816e-02 -1.86162317e+00
5.99067330e-01 -5.68395495e-01 4.01811957e-01 -1.33755738e-02
-6.17369652e-01 3.93820465e-01 9.68809798e-02 3.91131133e-01
-9.35571671e-01 6.46035373e-01 1.51591969e+00 -3.02358091e-01
2.22634122e-01 3.41904163e-01 -7.15533376e-01 8.85560215e-01
1.17953598e+00 -3.14795882e-01 -2.12153748e-01 -5.76538742e-01
1.92247987e-01 -1.59341738e-01 5.01728356e-01 -1.19445324e+00
3.85137379e-01 -2.27773800e-01 3.63711655e-01 -9.07063067e-01
1.70797065e-01 -4.35314566e-01 -3.27099919e-01 4.38535064e-01
4.67595279e-01 7.85947070e-02 -7.68238530e-02 4.03945416e-01
-3.66035402e-01 -2.65840381e-01 6.39458597e-01 -2.04015404e-01
-8.97887707e-01 7.82675922e-01 -1.57569185e-01 3.32296371e-01
1.17196488e+00 -4.03360218e-01 -7.35515416e-01 -5.02577960e-01
1.44836038e-01 5.55631161e-01 2.98449337e-01 -2.37405784e-02
3.61766219e-01 -1.51997590e+00 -9.47948277e-01 -4.11678590e-02
2.19450012e-01 3.66621733e-01 5.28059721e-01 9.14628267e-01
-7.34218538e-01 5.72917223e-01 6.51313066e-02 -6.47504866e-01
-8.64856541e-01 3.85732800e-01 2.04143092e-01 -2.81601220e-01
-1.91617489e-01 7.24741459e-01 2.77423799e-01 -4.40338701e-01
1.42629161e-01 -3.10431272e-01 -5.68806052e-01 -9.48779210e-02
1.06318283e+00 9.53217208e-01 -1.13534972e-01 -8.55020583e-01
-4.58744437e-01 5.43746412e-01 2.93092340e-01 4.59225476e-01
1.00554621e+00 -1.44432455e-01 5.82591332e-02 2.37537131e-01
1.04803467e+00 1.12165287e-01 -1.05842733e+00 4.06711741e-04
-3.19277108e-01 -7.52999663e-01 -3.39688122e-01 -1.78522933e-02
-1.24418783e+00 1.16116607e+00 3.93291593e-01 3.31599414e-01
7.73089528e-01 1.87147900e-01 8.74442518e-01 2.31571034e-01
8.48151267e-01 -8.69972348e-01 -1.25060514e-01 7.38442659e-01
2.81160355e-01 -1.76647687e+00 1.46799847e-01 -5.45631170e-01
-5.29492199e-01 7.72405148e-01 8.74128222e-01 -1.54675901e-01
7.44704247e-01 4.48037177e-01 -6.66876510e-02 -2.08900541e-01
-5.71778379e-02 -4.96932745e-01 -7.57151842e-02 1.02190924e+00
6.16608679e-01 2.64405876e-01 -3.85971755e-01 7.13951349e-01
-3.91389698e-01 1.56248361e-01 2.54537970e-01 4.94780213e-01
-8.53829563e-01 -5.93145132e-01 -5.08226871e-01 8.31911027e-01
-1.80495217e-01 -1.48929596e-01 2.63161540e-01 9.60455716e-01
4.97895479e-01 1.11516821e+00 2.47280702e-01 -3.39509577e-01
3.23715985e-01 -6.80013597e-01 2.78218091e-01 -3.84910762e-01
-5.09631932e-02 -6.01854980e-01 6.16805181e-02 3.35689485e-02
-7.62801111e-01 -5.71550846e-01 -1.09222317e+00 -1.26872313e+00
-5.44258118e-01 9.23113376e-02 3.36693168e-01 9.25137401e-01
-2.03343540e-01 2.71756351e-01 4.13398981e-01 -1.03002357e+00
-2.02937543e-01 -1.13541090e+00 -6.72271192e-01 -5.71563728e-02
-1.74989343e-01 -1.02616036e+00 -3.03966045e-01 -4.30713803e-01]
|
[8.424327850341797, -0.33595335483551025]
|
1098a6ff-c3ed-4a0a-b90f-30a5522de70b
|
sad-segment-any-rgbd
|
2305.14207
| null |
https://arxiv.org/abs/2305.14207v1
|
https://arxiv.org/pdf/2305.14207v1.pdf
|
SAD: Segment Any RGBD
|
The Segment Anything Model (SAM) has demonstrated its effectiveness in segmenting any part of 2D RGB images. However, SAM exhibits a stronger emphasis on texture information while paying less attention to geometry information when segmenting RGB images. To address this limitation, we propose the Segment Any RGBD (SAD) model, which is specifically designed to extract geometry information directly from images. Inspired by the natural ability of humans to identify objects through the visualization of depth maps, SAD utilizes SAM to segment the rendered depth map, thus providing cues with enhanced geometry information and mitigating the issue of over-segmentation. We further include the open-vocabulary semantic segmentation in our framework, so that the 3D panoptic segmentation is fulfilled. The project is available on https://github.com/Jun-CEN/SegmentAnyRGBD.
|
['Qifeng Chen', 'Ziwei Liu', 'Lingdong Kong', 'Yixuan Pei', 'Jingkang Yang', 'Xingyi Li', 'Kewei Wang', 'Yizheng Wu', 'Jun Cen']
|
2023-05-23
| null | null | null | null |
['panoptic-segmentation']
|
['computer-vision']
|
[ 2.46494204e-01 2.14192078e-01 1.40950203e-01 -4.75382864e-01
-5.41716695e-01 -7.27759957e-01 4.80550587e-01 7.41371661e-02
-1.78286195e-01 -1.02779523e-01 9.24656540e-02 -3.99684280e-01
1.48620248e-01 -1.09074545e+00 -3.08888882e-01 -5.54685116e-01
4.10922378e-01 3.24350484e-02 3.36291343e-01 -1.66671753e-01
4.11563873e-01 8.14594269e-01 -1.68391478e+00 2.57540464e-01
6.76233351e-01 1.25448060e+00 3.15335363e-01 3.61664742e-01
-4.88559365e-01 3.09661061e-01 -3.49934489e-01 -5.99487834e-02
5.20735979e-01 -2.85097420e-01 -7.96218693e-01 2.68948406e-01
4.05540109e-01 -5.15176058e-01 -1.02952324e-01 1.03130054e+00
3.20911884e-01 1.66272260e-02 3.71569216e-01 -1.15024221e+00
-3.82499963e-01 6.70765191e-02 -6.16342068e-01 -1.37938365e-01
7.02864707e-01 8.81754905e-02 9.00564551e-01 -9.61572289e-01
4.78015810e-01 1.06331778e+00 3.12032521e-01 2.52883375e-01
-1.00871372e+00 -5.60951173e-01 2.99140871e-01 -1.40965655e-01
-1.50503039e+00 -9.19149965e-02 1.12448430e+00 -4.24197882e-01
7.25090921e-01 7.29828358e-01 1.08246481e+00 7.31199622e-01
-2.87398517e-01 8.74954700e-01 1.38194513e+00 -3.71719569e-01
2.54847944e-01 8.44766498e-02 5.14439531e-02 5.21414101e-01
-3.48632112e-02 2.42494931e-03 -4.71586734e-01 1.91518217e-01
1.12847447e+00 1.85652480e-01 -3.65321755e-01 -3.92736077e-01
-1.00071132e+00 5.30518413e-01 6.25579000e-01 4.51989435e-02
-3.25769603e-01 -1.26664564e-01 -8.65364447e-02 -2.57238839e-02
6.71081245e-01 3.93486708e-01 -3.64626080e-01 -8.91242176e-02
-9.52108383e-01 2.83893459e-02 2.79851288e-01 9.45275784e-01
1.19171238e+00 -2.52789855e-01 1.12176165e-01 5.14758646e-01
4.94986624e-01 6.45798802e-01 1.01417415e-01 -1.30639529e+00
1.38329983e-01 1.16766608e+00 -5.07896505e-02 -9.32942271e-01
-5.63309371e-01 9.27945524e-02 -3.96378934e-01 5.01740575e-01
2.32806608e-01 3.19762111e-01 -1.18589437e+00 1.15598285e+00
8.05997014e-01 -6.09832644e-01 -2.59397358e-01 1.30210173e+00
1.04945397e+00 3.44386488e-01 2.27638651e-02 4.16156590e-01
1.46668649e+00 -6.37140512e-01 -4.94205296e-01 -3.51380020e-01
3.09935063e-01 -7.47395396e-01 1.54522145e+00 5.49575448e-01
-7.38059342e-01 -3.98467213e-01 -9.17600334e-01 -5.08068800e-01
-7.40754604e-01 -2.48178523e-02 7.80202687e-01 7.25490093e-01
-9.58789289e-01 3.51945400e-01 -1.04738724e+00 -4.85842526e-01
2.49670774e-01 5.80649711e-02 -3.07665378e-01 -2.11295225e-02
-8.93287182e-01 5.99834919e-01 4.37806368e-01 1.63912132e-01
-3.50767195e-01 -4.94401515e-01 -8.85574162e-01 -3.80630940e-01
4.77006644e-01 -4.19728696e-01 1.06030178e+00 -9.16362584e-01
-1.50351334e+00 9.69229221e-01 -2.74611656e-02 3.71553242e-01
5.29173911e-01 -3.41962606e-01 -5.71882278e-02 7.13929951e-01
-7.82356262e-02 9.51870143e-01 7.23572075e-01 -1.51034439e+00
-3.94589752e-01 -7.50118136e-01 5.12080491e-01 4.19118345e-01
-1.62671044e-01 -1.21441387e-01 -8.20052922e-01 -5.95684886e-01
7.66157448e-01 -8.29970658e-01 -2.77599134e-02 3.18066686e-01
-6.22702062e-01 -3.67101729e-02 9.70176339e-01 -8.11610103e-01
1.10616744e+00 -2.47064304e+00 7.41377026e-02 2.68495440e-01
1.87339365e-01 -1.06798954e-01 1.70507729e-01 3.55492651e-01
1.05557963e-01 2.64544815e-01 -5.58328867e-01 -4.45436478e-01
1.05473630e-01 2.85472721e-01 7.44162593e-04 2.97965229e-01
1.21471278e-01 7.24233091e-01 -6.89857364e-01 -5.20659626e-01
5.92056930e-01 6.00430906e-01 -3.47089738e-01 3.22679460e-01
-3.05110693e-01 7.43243158e-01 -6.45424962e-01 1.07749343e+00
1.10824144e+00 -1.56970158e-01 -1.76978037e-01 -3.40933502e-01
-4.47228670e-01 3.01445067e-01 -1.31077218e+00 2.09579682e+00
-1.82317078e-01 4.38987285e-01 1.45227700e-01 -4.54982042e-01
8.91902447e-01 7.07635060e-02 6.99148178e-01 -1.00768507e+00
1.16054080e-01 -1.11673982e-03 -6.61433876e-01 -5.72012603e-01
7.31730878e-01 2.19344184e-01 -1.00301087e-01 5.74007690e-01
-4.78636593e-01 -8.95560861e-01 -2.10928783e-01 2.46554151e-01
5.29556811e-01 7.05881894e-01 -1.24274520e-02 -8.24155379e-03
6.68753460e-02 3.77248049e-01 2.88452059e-01 4.31613773e-01
-7.28588551e-02 1.21790826e+00 2.91671216e-01 -1.99445754e-01
-7.43676841e-01 -1.22174394e+00 -2.92897314e-01 8.17141593e-01
6.06239259e-01 -5.00148177e-01 -8.40884626e-01 -3.80240560e-01
-7.76089951e-02 7.16804802e-01 -6.48760378e-01 2.59368062e-01
-2.79053152e-01 -4.78905827e-01 2.19322085e-01 3.39052260e-01
7.95418978e-01 -8.17588449e-01 -1.25834298e+00 -3.47514331e-01
-3.02327126e-01 -9.35964048e-01 -2.21494243e-01 1.80610657e-01
-8.24488342e-01 -1.17000377e+00 -6.52267337e-01 -2.49727473e-01
7.45937884e-01 5.60261309e-01 8.96814466e-01 -8.59856382e-02
-4.32661057e-01 7.73882270e-01 -7.97747374e-01 -3.78543764e-01
1.67565554e-01 1.04483433e-01 -5.65237403e-01 -1.69539422e-01
4.49241638e-01 -3.19765866e-01 -1.05577743e+00 4.00601029e-01
-1.32703030e+00 7.78758466e-01 2.60973424e-01 -8.42337236e-02
8.46501291e-01 -2.09195495e-01 -2.93519378e-01 -6.40219152e-01
2.38438219e-01 -2.49818161e-01 -5.02534151e-01 7.86544010e-02
-5.24734557e-01 -2.94670254e-01 -9.99083929e-03 -4.32586260e-02
-1.13147211e+00 1.67684421e-01 -6.91866577e-02 -2.12197080e-01
-4.64823782e-01 3.56496006e-01 -2.41700858e-01 -1.60436630e-01
1.97736144e-01 -2.14979192e-03 5.33720180e-02 -8.38670075e-01
4.68928486e-01 7.55479515e-01 2.19315916e-01 -4.73111302e-01
4.69964266e-01 9.08616006e-01 -1.83124185e-01 -8.49466443e-01
-7.84389138e-01 -6.61535323e-01 -9.86710668e-01 -4.48146492e-01
1.01301837e+00 -7.23566413e-01 -3.32450360e-01 4.87193704e-01
-9.36820686e-01 -5.99459708e-01 -3.21531832e-01 3.75923842e-01
-3.75269562e-01 4.15435880e-01 -4.11168098e-01 -7.43213654e-01
-2.90695764e-02 -1.18906236e+00 1.39231813e+00 4.14007008e-01
-2.24251837e-01 -6.65473819e-01 -2.43753299e-01 4.42707330e-01
2.09098309e-01 5.60509145e-01 7.31753290e-01 -1.71160504e-01
-7.83757091e-01 -5.22085615e-02 -5.39582491e-01 3.77149165e-01
2.19395459e-01 8.78709257e-02 -1.23432398e+00 2.57704556e-01
3.21203358e-02 3.64719592e-02 5.53658962e-01 1.78847492e-01
1.14120913e+00 8.48134607e-02 -4.45505790e-02 8.54790807e-01
1.51720405e+00 1.89278558e-01 6.66428447e-01 6.65221870e-01
1.04273021e+00 8.81093144e-01 7.79936492e-01 5.19724607e-01
6.80062175e-01 6.38711631e-01 7.22410381e-01 -7.14637756e-01
-2.60107994e-01 -1.58395544e-01 -7.12296274e-03 6.24659061e-01
-2.33084083e-01 -7.50849694e-02 -1.12234163e+00 1.75591081e-01
-1.66177392e+00 -1.61229536e-01 -5.28620720e-01 2.08176351e+00
6.86735451e-01 -9.51272547e-02 5.69312163e-02 2.03737482e-01
3.40820312e-01 1.53812274e-01 -4.73402172e-01 -4.06551898e-01
-3.14292818e-01 1.37208879e-01 5.87338269e-01 3.34955931e-01
-1.00053263e+00 9.49975848e-01 5.81787682e+00 6.63642526e-01
-1.25458407e+00 -3.90828326e-02 5.86093366e-01 -1.56594917e-01
-5.59277236e-01 1.66588828e-01 -3.05192351e-01 3.23884636e-01
3.50691557e-01 4.33233261e-01 3.18941146e-01 6.80573106e-01
4.24431294e-01 -8.40067208e-01 -5.93704045e-01 8.43251646e-01
-1.02476679e-01 -6.63956940e-01 8.97292644e-02 1.36158615e-01
4.92878228e-01 -1.25667647e-01 7.49678761e-02 -2.60007620e-01
-1.13677651e-01 -8.68274510e-01 1.03549719e+00 5.83226860e-01
9.12674308e-01 -4.90589738e-01 4.34432745e-01 9.27012563e-02
-1.16446745e+00 2.39534244e-01 -1.74724266e-01 -1.43748462e-01
1.74926624e-01 6.83592379e-01 -6.16216242e-01 6.09703064e-01
8.99627328e-01 4.76229072e-01 -9.45614338e-01 9.58998978e-01
-4.69865352e-01 2.19139129e-01 -6.55859292e-01 3.68481964e-01
2.90939301e-01 -6.51629388e-01 2.21768767e-01 1.09405434e+00
4.03666347e-01 2.61068761e-01 1.57177404e-01 9.23819065e-01
5.04978299e-01 6.85154721e-02 -4.58481550e-01 1.04286872e-01
3.65054846e-01 1.25301814e+00 -1.31510472e+00 -9.42404866e-02
-4.38786238e-01 1.35570967e+00 -2.19870925e-01 6.16861045e-01
-7.16619611e-01 -3.82606924e-01 4.54134256e-01 3.45815241e-01
2.29696050e-01 -5.15882075e-01 -8.16862226e-01 -8.43488395e-01
-3.16337347e-02 -5.63452840e-01 3.07485938e-01 -1.37979591e+00
-7.36487448e-01 4.53204662e-01 3.26271862e-01 -1.13391411e+00
3.19634616e-01 -7.08414197e-01 -7.95820132e-02 8.18193316e-01
-1.49759424e+00 -1.50250292e+00 -7.75457501e-01 7.24582553e-01
3.06369543e-01 7.82511055e-01 7.66518474e-01 1.85608789e-01
-5.01575887e-01 -7.88838193e-02 -3.30501139e-01 4.86376174e-02
5.27807236e-01 -1.39073074e+00 2.76322991e-01 7.07408607e-01
-1.12875141e-01 5.51542580e-01 5.36670566e-01 -6.21006727e-01
-1.40565610e+00 -6.99191093e-01 3.66668761e-01 -5.16425729e-01
2.04586849e-01 -4.34383661e-01 -7.56670594e-01 4.81386334e-01
-1.08805701e-01 -2.38036156e-01 6.24531507e-01 -2.88606733e-01
-4.58878428e-01 1.79474726e-01 -1.21663427e+00 5.85858345e-01
1.12704849e+00 -8.05918276e-01 -4.21945661e-01 2.39847302e-02
8.65836918e-01 -7.19804227e-01 -9.03912842e-01 5.03647387e-01
5.44872940e-01 -1.40062153e+00 1.15828741e+00 2.81162590e-01
3.56502444e-01 -5.67595363e-01 -2.87084162e-01 -9.15956378e-01
1.71998143e-02 -2.21269518e-01 1.69677004e-01 1.21046090e+00
2.77407855e-01 -5.62732041e-01 7.60040104e-01 9.37228978e-01
-2.58692294e-01 -5.40295482e-01 -8.31523359e-01 -4.25529838e-01
-3.15610349e-01 -8.18857491e-01 7.94932663e-01 8.86976898e-01
-1.70853138e-01 -3.72398615e-01 1.97653696e-01 2.68905908e-01
3.93131167e-01 5.21326721e-01 6.57708645e-01 -1.10952091e+00
-2.94798473e-03 -6.15717232e-01 -1.86668932e-01 -1.01598263e+00
-5.01519084e-01 -7.00248063e-01 -1.05086006e-01 -1.79560232e+00
-6.43063709e-02 -5.78518391e-01 -1.02905691e-01 5.67385375e-01
1.62045155e-02 7.85418808e-01 2.96503693e-01 3.73530656e-01
-5.62945306e-01 3.57320637e-01 1.60394490e+00 1.07649930e-01
-2.58563191e-01 -1.48454309e-01 -7.40461051e-01 7.58595586e-01
9.75804090e-01 -2.38995776e-01 -3.12486619e-01 -6.20986581e-01
2.24181741e-01 -3.00262213e-01 5.00581622e-01 -7.48183966e-01
-2.59607341e-02 -2.58964211e-01 3.96399349e-01 -8.36243153e-01
5.69260299e-01 -9.23191071e-01 2.53343731e-01 9.83092636e-02
1.00629829e-01 -2.03774888e-02 2.67581433e-01 1.61782920e-01
-1.27238154e-01 -1.14510901e-01 3.78480673e-01 -4.45480198e-01
-9.99648154e-01 1.90012589e-01 -2.04818159e-01 -1.93522424e-01
8.38684261e-01 -7.51877904e-01 -1.30487308e-01 -1.90104172e-01
-7.49724269e-01 8.97918120e-02 1.11316097e+00 2.08795458e-01
7.20796108e-01 -1.04167342e+00 -1.91740226e-02 5.15480161e-01
3.43836635e-01 6.28721356e-01 3.92831296e-01 7.89312482e-01
-9.84972358e-01 2.19484299e-01 -1.50679410e-01 -7.20397830e-01
-1.00037980e+00 2.83452123e-01 1.49551734e-01 3.58325839e-01
-8.96450043e-01 7.71937609e-01 3.69879574e-01 -4.95076239e-01
1.58879697e-01 -5.61855972e-01 -5.89631796e-02 2.54064143e-01
2.74219096e-01 4.10593539e-01 2.23830447e-01 -5.48813522e-01
-3.66438925e-01 9.29070413e-01 4.34798479e-01 -3.75588745e-01
1.17201793e+00 -5.59533894e-01 -2.26665542e-01 6.91038609e-01
1.08912110e+00 1.72833443e-01 -1.55427110e+00 -2.80006789e-02
-1.01618566e-01 -6.93365216e-01 3.00210059e-01 -9.04312670e-01
-1.08051407e+00 9.84393537e-01 6.30540311e-01 4.22215402e-01
1.41100109e+00 4.91158627e-02 6.19876981e-01 -2.08491534e-01
5.54097176e-01 -1.06970048e+00 1.07977688e-01 3.47569615e-01
6.98160172e-01 -1.18915141e+00 1.94250271e-01 -6.50388181e-01
-6.93510115e-01 1.24291778e+00 5.62852740e-01 1.79997191e-01
6.04750633e-01 1.14691801e-01 4.17600781e-01 -6.30313218e-01
1.76327363e-01 -5.01242936e-01 3.33079278e-01 6.82367325e-01
2.98851788e-01 1.75982773e-01 3.60338055e-02 2.49211907e-01
-3.40391695e-01 -3.11063260e-01 2.95682162e-01 1.18261433e+00
-3.17870498e-01 -9.54290271e-01 -6.38094723e-01 5.51910810e-02
-7.48150349e-02 -1.49226129e-01 -7.20650733e-01 7.89411366e-01
2.49669716e-01 8.97732675e-01 4.44732934e-01 -3.41076523e-01
2.79360831e-01 -4.66395952e-02 4.84865606e-01 -6.19004488e-01
-3.55576336e-01 2.70753503e-01 -1.70894176e-01 -8.79484177e-01
-6.49234056e-01 -5.60779929e-01 -1.32682216e+00 -1.62397861e-01
-4.23618704e-02 -1.65293992e-01 1.04572701e+00 5.77205837e-01
3.53678554e-01 3.60370785e-01 4.60247993e-01 -1.11453557e+00
2.84304380e-01 -7.26540327e-01 -6.85566068e-01 4.08522516e-01
2.07317561e-01 -6.43242419e-01 -4.62351203e-01 -5.01633547e-02]
|
[8.851123809814453, -2.8830795288085938]
|
026aa436-b7d1-4f1b-9207-0aacd4097c09
|
not-all-languages-are-created-equal-in-llms
|
2305.07004
| null |
https://arxiv.org/abs/2305.07004v1
|
https://arxiv.org/pdf/2305.07004v1.pdf
|
Not All Languages Are Created Equal in LLMs: Improving Multilingual Capability by Cross-Lingual-Thought Prompting
|
Large language models (LLMs) demonstrate impressive multilingual capability, but their performance varies substantially across different languages. In this work, we introduce a simple yet effective method, called cross-lingual-thought prompting (XLT), to systematically improve the multilingual capability of LLMs. Specifically, XLT is a generic template prompt that stimulates cross-lingual and logical reasoning skills to enhance task performance across languages. We conduct comprehensive evaluations on 7 typical benchmarks related to reasoning, understanding, and generation tasks, covering both high-resource and low-resource languages. Experimental results show that XLT not only remarkably enhances the performance of various multilingual tasks but also significantly reduces the gap between the average performance and the best performance of each task in different languages. Notably, XLT brings over 10 points of average improvement in arithmetic reasoning and open-domain question-answering tasks.
|
['Furu Wei', 'Yan Xia', 'Ting Song', 'Wayne Xin Zhao', 'Dongdong Zhang', 'Tianyi Tang', 'Haoyang Huang']
|
2023-05-11
| null | null | null | null |
['open-domain-question-answering', 'arithmetic-reasoning', 'logical-reasoning']
|
['natural-language-processing', 'reasoning', 'reasoning']
|
[-4.65112627e-01 -5.32771721e-02 -2.22781137e-01 -2.35605597e-01
-1.50224102e+00 -6.76189482e-01 9.24031138e-01 2.14340642e-01
-4.69808936e-01 7.33173013e-01 5.18698275e-01 -7.47989655e-01
6.38701320e-02 -7.79663801e-01 -7.59353220e-01 -8.25416744e-02
3.76026988e-01 5.02199173e-01 -8.80894717e-03 -6.63460612e-01
7.61058927e-02 -3.26850601e-02 -8.62565577e-01 7.28777170e-01
1.77864683e+00 5.05590498e-01 3.20523113e-01 3.12929630e-01
-8.04438531e-01 1.37094414e+00 -7.18986630e-01 -8.96997511e-01
-2.96610463e-02 -2.53762692e-01 -1.02085733e+00 -3.35410267e-01
4.67567861e-01 -2.03209054e-02 -1.02378815e-01 8.74100149e-01
5.13694823e-01 4.02987748e-02 4.58370060e-01 -1.01965415e+00
-1.21905804e+00 1.30063665e+00 -4.91905183e-01 7.52066225e-02
8.18494976e-01 2.60652959e-01 1.05797219e+00 -1.02159131e+00
5.98375022e-01 1.61169469e+00 5.89225650e-01 3.36170673e-01
-1.10205281e+00 -6.81458235e-01 2.30306759e-01 1.09972417e-01
-1.37348294e+00 -3.20365876e-01 4.78536189e-01 -3.79166454e-01
1.19295442e+00 -5.37656806e-02 1.40710935e-01 9.67404664e-01
1.41136661e-01 1.00516498e+00 1.53040004e+00 -6.38832569e-01
-1.18913651e-01 2.28018641e-01 1.33157134e-01 7.89815068e-01
5.74924462e-02 -3.18338126e-01 -8.84738147e-01 1.11705028e-01
4.77398127e-01 -5.62790930e-01 -2.66868383e-01 2.66782254e-01
-1.64695787e+00 6.87987149e-01 2.21965596e-01 4.92296427e-01
-3.03007215e-01 -2.63258249e-01 5.02554178e-01 6.30879045e-01
4.54472512e-01 8.20975959e-01 -6.68912530e-01 -1.26494840e-01
-4.08109337e-01 3.72871608e-01 8.77783418e-01 1.08256948e+00
5.88380456e-01 1.33342177e-01 -6.81713581e-01 1.21177924e+00
-7.85811543e-02 9.66706097e-01 6.68121874e-01 -9.21593547e-01
1.01454782e+00 8.78519118e-01 -8.22720826e-02 -6.07480407e-01
-2.21852213e-01 -3.72904539e-01 -5.98766506e-01 -5.63627541e-01
6.99827790e-01 -3.01610589e-01 -2.79944897e-01 2.03110456e+00
8.80101100e-02 -4.31295544e-01 5.89329422e-01 5.22905767e-01
1.09278786e+00 7.31568158e-01 3.93820196e-01 -5.75681143e-02
1.56800830e+00 -1.27150548e+00 -8.06736171e-01 -5.53787172e-01
1.06305742e+00 -1.03542435e+00 1.74894798e+00 2.39652470e-01
-1.24160683e+00 -6.49538457e-01 -7.06769288e-01 -5.16778350e-01
-5.22598326e-01 2.83390462e-01 7.58367479e-01 3.34784746e-01
-8.76748919e-01 -7.27384314e-02 -2.53261805e-01 -2.34539047e-01
-2.27487862e-01 -2.98140496e-01 -2.58782566e-01 -6.31386876e-01
-1.63492298e+00 1.04336441e+00 5.05273163e-01 -3.61458004e-01
-5.26646435e-01 -1.01466727e+00 -9.81507719e-01 -1.03714187e-02
4.84157830e-01 -5.28251708e-01 1.48637092e+00 -4.23419148e-01
-1.38319540e+00 9.20214295e-01 -1.70153350e-01 -3.96667600e-01
4.46241736e-01 -5.34773588e-01 -4.57997561e-01 -2.79981434e-01
5.38090408e-01 8.85523200e-01 1.99726924e-01 -7.54452705e-01
-4.39735979e-01 2.13133078e-03 3.96189541e-01 4.46714759e-01
-3.89457911e-01 2.24367827e-01 -5.28460324e-01 -8.85399103e-01
-4.11539413e-02 -6.30344212e-01 4.44122143e-02 -5.74124455e-01
-2.66835153e-01 -9.22818124e-01 1.20738976e-01 -9.62341666e-01
1.18832517e+00 -1.90304446e+00 -4.94161919e-02 -4.22867030e-01
-1.44609302e-01 8.67860466e-02 -5.68719804e-01 6.07308567e-01
1.03789590e-01 6.73877373e-02 1.33553609e-01 5.46021201e-02
3.82136375e-01 2.31071651e-01 -6.25711381e-01 -3.14787567e-01
1.14627905e-01 1.41090572e+00 -9.96515989e-01 -4.63620037e-01
-1.39979199e-01 4.80804481e-02 -5.50558805e-01 2.96513736e-01
-7.04482615e-01 4.34535384e-01 -3.85093302e-01 5.96233368e-01
4.69954699e-01 -3.79495233e-01 3.70578825e-01 2.37451196e-01
-1.42215237e-01 8.02127123e-01 -7.32506096e-01 2.07329345e+00
-9.81346607e-01 4.39520448e-01 -4.78147358e-01 -3.70265812e-01
9.91825342e-01 3.18280309e-01 -6.15717135e-02 -1.46299350e+00
-3.79649967e-01 5.31371057e-01 1.81472957e-01 -6.33491397e-01
5.95706761e-01 7.39632398e-02 -4.45901483e-01 6.58762813e-01
-9.59395170e-02 -1.96548507e-01 6.12857103e-01 4.89019692e-01
7.80447423e-01 8.43121037e-02 4.54554141e-01 -6.00904763e-01
1.00596058e+00 5.49849123e-04 3.69560987e-01 7.59031296e-01
8.78419280e-02 -2.92552918e-01 4.89723682e-01 -1.83408022e-01
-8.09768915e-01 -1.15615547e+00 4.91509140e-02 1.67547834e+00
-2.39964589e-01 -7.87829280e-01 -7.91072011e-01 -5.52196503e-01
7.92799518e-02 1.08296001e+00 1.93644837e-02 8.01426396e-02
-6.85643852e-01 -7.01887608e-01 9.39271331e-01 6.08705223e-01
9.54429150e-01 -1.18510079e+00 1.32635208e-02 1.53960228e-01
-9.24319267e-01 -1.89997780e+00 -3.10433209e-01 -2.60115057e-01
-5.22487044e-01 -6.95179999e-01 -5.80407083e-01 -8.67171288e-01
4.91613150e-01 2.52789170e-01 1.57604563e+00 -7.87491426e-02
2.12062150e-01 2.17527449e-01 -3.27683628e-01 -1.86031520e-01
-5.03956258e-01 2.46016160e-01 5.04720677e-03 -5.41794837e-01
5.35009921e-01 1.96293113e-03 5.84580973e-02 6.57930970e-02
-4.74567562e-01 3.54703814e-01 6.40075862e-01 6.02556586e-01
3.79526168e-01 -1.11071259e-01 8.69194925e-01 -7.62213588e-01
1.19609094e+00 -3.23877543e-01 -4.24855322e-01 8.19312990e-01
-4.95466471e-01 3.94302726e-01 8.78365040e-01 -2.55127192e-01
-1.33934200e+00 -7.72378862e-01 -4.08711694e-02 2.49702245e-01
3.38282697e-02 8.49467039e-01 -1.00325532e-01 5.43494783e-02
7.02826738e-01 4.97488886e-01 -4.71904784e-01 -4.98306870e-01
5.97460747e-01 5.95518887e-01 6.53144777e-01 -1.43686187e+00
6.55843198e-01 -1.50139600e-01 -5.33691525e-01 -5.31203985e-01
-1.16221631e+00 -1.12363689e-01 -2.56592900e-01 -7.09952414e-02
7.81870186e-01 -1.41734409e+00 -6.70476973e-01 5.19290447e-01
-1.35413456e+00 -7.76373208e-01 1.41248897e-01 4.42016870e-01
-2.99209177e-01 2.36015961e-01 -9.66941595e-01 -4.93475437e-01
-5.71868360e-01 -1.24753237e+00 9.72858667e-01 1.15540758e-01
-2.67493278e-01 -1.30510879e+00 -1.36186391e-01 8.60897779e-01
4.95833397e-01 -3.27288777e-01 1.53572285e+00 -7.16613054e-01
-6.91754937e-01 2.62770057e-01 -4.54469055e-01 2.63868600e-01
-1.05930358e-01 -5.46345234e-01 -5.55933654e-01 -9.93093476e-02
-1.82268903e-01 -9.07045841e-01 4.34257537e-01 -1.27506256e-03
8.79096746e-01 -2.33383849e-01 5.07863723e-02 3.96026790e-01
1.22190833e+00 6.04088046e-03 3.96009535e-01 5.16836822e-01
6.05233252e-01 6.67423904e-01 7.05002248e-01 1.94718614e-02
1.17899680e+00 5.57314396e-01 -5.34401774e-01 -3.60393664e-03
-2.58118927e-01 -6.04439318e-01 7.25567639e-01 1.41281486e+00
2.15580314e-01 4.02684249e-02 -1.31646681e+00 4.66478348e-01
-1.72728491e+00 -4.95485723e-01 -2.54524529e-01 2.02892900e+00
1.36041582e+00 -4.21543866e-02 -3.21590215e-01 -3.97752881e-01
2.85190761e-01 1.33884475e-01 -3.40299964e-01 -5.47240496e-01
-3.80468249e-01 2.08595946e-01 2.95864195e-02 6.06665909e-01
-5.31107605e-01 1.54580343e+00 6.56851673e+00 1.12800276e+00
-8.93091381e-01 2.82476395e-01 3.29990983e-01 3.56202781e-01
-6.26483500e-01 -1.19652711e-01 -8.83178711e-01 1.87423617e-01
9.47793365e-01 -5.68598747e-01 6.03692770e-01 6.76212013e-01
3.72578539e-02 -1.45001754e-01 -1.10125625e+00 8.79403412e-01
2.09994465e-01 -1.03382635e+00 4.15431052e-01 -3.65236223e-01
1.06400669e+00 9.80731696e-02 -2.90211774e-02 1.01382494e+00
5.62039912e-01 -8.88702571e-01 6.55482948e-01 3.31317455e-01
7.88840413e-01 -7.48846114e-01 5.04849732e-01 6.45937622e-01
-1.23452342e+00 1.77170098e-01 -2.03608990e-01 -2.58575797e-01
1.27928659e-01 3.27002048e-01 -5.72511375e-01 6.40239537e-01
4.38022763e-01 3.93538147e-01 -8.45983624e-01 4.22055662e-01
-8.85618567e-01 4.08449173e-01 1.24704026e-01 -1.86883379e-02
4.32361811e-01 -1.19109377e-01 1.20685354e-01 1.26937139e+00
1.59970194e-01 -8.63786489e-02 6.04027212e-01 7.42470086e-01
-2.80341506e-01 6.74140036e-01 -5.75544834e-01 -2.71422386e-01
6.71090841e-01 1.02257192e+00 -2.02571809e-01 -6.86139107e-01
-7.02632427e-01 5.93366325e-01 8.71241331e-01 4.00342554e-01
-8.68552327e-01 -2.69391149e-01 4.08267170e-01 -1.39752969e-01
-2.38111034e-01 -4.52636063e-01 -3.96259516e-01 -1.45933926e+00
1.01754546e-01 -1.50783730e+00 5.40813744e-01 -1.09874904e+00
-1.45184839e+00 6.78351045e-01 -6.59801960e-02 -7.44506299e-01
-4.34487641e-01 -6.37458682e-01 -1.64543033e-01 1.11888742e+00
-1.71628034e+00 -1.28952992e+00 -2.41392195e-01 7.10895658e-01
8.57906938e-01 -4.37552333e-01 8.02058995e-01 4.69573557e-01
-5.65579653e-01 7.65486300e-01 -1.54604837e-01 3.23064715e-01
1.08740234e+00 -1.29968345e+00 5.75680375e-01 9.03016686e-01
1.86874092e-01 9.87738550e-01 2.01868758e-01 -6.61514223e-01
-1.36764324e+00 -1.07577062e+00 1.67923892e+00 -4.83826965e-01
1.12991381e+00 -3.36286724e-01 -1.07166862e+00 9.10192013e-01
7.24127650e-01 -6.37338996e-01 6.71876907e-01 4.84521508e-01
-7.51547277e-01 -5.91473514e-03 -5.87395191e-01 1.02237141e+00
8.49865496e-01 -1.09745467e+00 -8.98901403e-01 8.89698982e-01
9.69869673e-01 -6.47011936e-01 -1.19571590e+00 3.55886996e-01
2.28536770e-01 -7.47872829e-01 9.81244266e-01 -6.67489111e-01
8.10181499e-01 -3.51265669e-02 -1.94052011e-01 -1.41690564e+00
-2.00171620e-01 -5.53333700e-01 2.52006948e-01 1.36748600e+00
6.15159452e-01 -9.11788523e-01 -1.29517734e-01 4.40411121e-01
-6.81035966e-02 -6.97178483e-01 -4.13133442e-01 -8.67170632e-01
6.79007232e-01 -6.07202828e-01 7.40720868e-01 1.29539323e+00
3.17199528e-01 7.59700775e-01 -5.00394218e-02 -5.29206283e-02
4.53247547e-01 3.20723891e-01 8.29067588e-01 -7.72990942e-01
-3.72880369e-01 -5.08066714e-01 3.89437139e-01 -1.31194806e+00
7.87923515e-01 -1.44774652e+00 -1.90209970e-01 -1.59124494e+00
2.35652298e-01 -4.37348187e-01 6.98966486e-03 5.91682494e-01
-6.38985932e-01 -1.90898135e-01 1.82317168e-01 3.57999429e-02
-6.98717892e-01 4.15753603e-01 1.49841237e+00 -6.36461517e-03
9.15475860e-02 -4.51081395e-01 -9.63476300e-01 7.39057839e-01
6.22824311e-01 -2.28378490e-01 -4.74627137e-01 -1.20451760e+00
4.74646658e-01 2.34193578e-01 6.89216778e-02 -8.42299461e-01
2.81942755e-01 -1.56132653e-01 1.26979863e-02 -4.75923955e-01
-8.61797631e-02 1.80023871e-02 -4.03904676e-01 3.75484407e-01
-6.29627049e-01 4.96673226e-01 3.64449680e-01 -1.10841766e-02
-4.08676416e-01 -6.96339458e-02 4.33017939e-01 -2.35140786e-01
-9.29233253e-01 -8.68095905e-02 -9.04484540e-02 7.63784289e-01
5.87910473e-01 5.29558659e-01 -9.50600863e-01 -2.94240415e-01
-1.40855849e-01 5.52745998e-01 8.85512605e-02 7.93181837e-01
8.68308023e-02 -1.47866249e+00 -9.73688841e-01 4.95150611e-02
4.56905812e-01 -2.19821721e-01 2.23744586e-01 9.28356111e-01
-4.27789390e-01 1.21243834e+00 -9.16157216e-02 -3.88415098e-01
-7.84661770e-01 4.80159014e-01 3.32864016e-01 -7.36635566e-01
-4.66258198e-01 8.47937822e-01 2.72342592e-01 -1.06114244e+00
6.06280789e-02 -3.97344887e-01 -1.13872863e-01 -1.80228621e-01
5.49994171e-01 2.48497918e-01 4.44137864e-02 -3.65366846e-01
-9.19149294e-02 4.10565436e-01 -3.07301849e-01 -2.05396712e-01
6.32281125e-01 4.64459211e-02 -5.33329785e-01 4.72321630e-01
7.09580958e-01 3.93485457e-01 -6.19518518e-01 -7.71885395e-01
2.49938756e-01 -1.56987160e-01 -2.54097164e-01 -1.13318443e+00
-7.15448260e-01 7.06461668e-01 -3.13655972e-01 -3.95905934e-02
8.46484423e-01 4.07206267e-02 1.02106106e+00 7.53776252e-01
6.77351236e-01 -1.06502473e+00 2.24255189e-01 1.16457057e+00
1.09728682e+00 -1.28190315e+00 -2.62852132e-01 -3.40636641e-01
-9.45150971e-01 7.98779070e-01 1.18603373e+00 3.06816399e-01
6.26756698e-02 2.80194134e-01 3.19188535e-01 2.23401003e-02
-1.10398722e+00 -2.32236564e-01 3.50471497e-01 2.64725059e-01
1.09328616e+00 3.96832138e-01 -6.96108878e-01 7.27777243e-01
-7.18159258e-01 -4.68564928e-02 2.52775848e-02 6.62496030e-01
-1.97721854e-01 -1.18989205e+00 -4.04569507e-01 1.46286367e-02
-3.52078378e-01 -7.27649570e-01 -3.78866732e-01 9.24154520e-01
6.03410639e-02 9.16599154e-01 -6.95417598e-02 -1.43604362e-02
3.64239782e-01 4.45986986e-01 6.45417571e-01 -4.73868251e-01
-6.90867901e-01 -2.36949340e-01 3.05943519e-01 -4.70193595e-01
-5.38698174e-02 -4.01585162e-01 -1.36736083e+00 -4.50952530e-01
1.69837937e-01 1.52277902e-01 3.67413193e-01 1.33108842e+00
1.82266518e-01 5.98658442e-01 1.05475821e-01 2.13633820e-01
-7.32124627e-01 -1.18092215e+00 -1.55245975e-01 2.55364597e-01
-2.47830704e-01 -4.56185371e-01 -8.68552644e-03 -1.02338277e-01]
|
[10.988234519958496, 9.34571647644043]
|
38b0a0f8-bec2-49bc-a384-2c042381dcd7
|
on-the-optimality-of-batch-policy
|
2104.02293
| null |
https://arxiv.org/abs/2104.02293v1
|
https://arxiv.org/pdf/2104.02293v1.pdf
|
On the Optimality of Batch Policy Optimization Algorithms
|
Batch policy optimization considers leveraging existing data for policy construction before interacting with an environment. Although interest in this problem has grown significantly in recent years, its theoretical foundations remain under-developed. To advance the understanding of this problem, we provide three results that characterize the limits and possibilities of batch policy optimization in the finite-armed stochastic bandit setting. First, we introduce a class of confidence-adjusted index algorithms that unifies optimistic and pessimistic principles in a common framework, which enables a general analysis. For this family, we show that any confidence-adjusted index algorithm is minimax optimal, whether it be optimistic, pessimistic or neutral. Our analysis reveals that instance-dependent optimality, commonly used to establish optimality of on-line stochastic bandit algorithms, cannot be achieved by any algorithm in the batch setting. In particular, for any algorithm that performs optimally in some environment, there exists another environment where the same algorithm suffers arbitrarily larger regret. Therefore, to establish a framework for distinguishing algorithms, we introduce a new weighted-minimax criterion that considers the inherent difficulty of optimal value prediction. We demonstrate how this criterion can be used to justify commonly used pessimistic principles for batch policy optimization.
|
['Dale Schuurmans', 'Csaba Szepesvari', 'Lihong Li', 'Jincheng Mei', 'Bo Dai', 'Tor Lattimore', 'Yifan Wu', 'Chenjun Xiao']
|
2021-04-06
| null | null | null | null |
['value-prediction']
|
['computer-code']
|
[ 5.82141094e-02 1.04046933e-01 -9.71020758e-01 -2.94019967e-01
-8.99594545e-01 -9.55180228e-01 3.57117563e-01 1.44390807e-01
-5.48782289e-01 1.14555025e+00 1.34393886e-01 -9.05679584e-01
-7.37061322e-01 -4.62919474e-01 -7.93654263e-01 -8.92173469e-01
-1.13503471e-01 6.10548079e-01 -2.12136477e-01 -3.33372094e-02
3.77395630e-01 6.64747953e-01 -1.31525469e+00 -2.23348569e-02
7.58999050e-01 1.51406693e+00 -6.69008270e-02 6.34468496e-01
7.11681843e-02 7.73715913e-01 -5.37206829e-01 -4.79080081e-01
6.73129499e-01 -4.52224553e-01 -7.03561187e-01 6.03112914e-02
-1.84984300e-02 -4.25746202e-01 1.67645872e-01 1.26590502e+00
3.73343378e-01 3.60121697e-01 4.47756231e-01 -1.24235368e+00
-2.70247787e-01 8.96298766e-01 -3.97391140e-01 3.78844976e-01
7.79684111e-02 2.18934223e-01 1.32927525e+00 1.40704131e-02
3.38905513e-01 1.18497837e+00 3.52561533e-01 4.47110474e-01
-1.30271757e+00 -2.78583229e-01 5.45716763e-01 8.11376199e-02
-7.89338410e-01 -5.38268149e-01 5.84260404e-01 -3.57170224e-01
4.90666270e-01 7.82123983e-01 8.45125854e-01 9.91224647e-01
6.12680614e-03 1.10428333e+00 1.36608148e+00 -6.10551178e-01
5.90390682e-01 4.18090045e-01 1.79613620e-01 2.47089371e-01
6.61451697e-01 6.85951769e-01 -2.59399265e-01 -4.93248105e-01
5.31777859e-01 -7.63058141e-02 -2.32540950e-01 -5.61273813e-01
-8.77141237e-01 9.56786275e-01 -5.57335764e-02 1.35472745e-01
-5.34802914e-01 -2.74537667e-03 2.50427514e-01 5.20026326e-01
6.89395905e-01 6.63029909e-01 -4.99108702e-01 -3.64138305e-01
-9.35636282e-01 5.87174118e-01 1.00963819e+00 6.64229095e-01
1.72527328e-01 4.19147089e-02 -6.45733714e-01 4.49860245e-01
6.03975020e-02 4.32249784e-01 2.22550899e-01 -1.25275993e+00
5.33304930e-01 -3.07936341e-01 1.08733010e+00 -6.62477016e-01
-2.45173916e-01 -7.05768108e-01 -2.68089503e-01 2.60838985e-01
7.41160095e-01 -2.41278842e-01 -3.58023256e-01 1.80856788e+00
4.01341438e-01 -8.94395635e-02 9.03082341e-02 8.43928218e-01
-4.15524304e-01 4.78725404e-01 -3.03843617e-01 -1.05312467e+00
7.96140552e-01 -7.80855000e-01 -6.35029376e-01 -4.51970734e-02
4.92701590e-01 -4.45203334e-01 9.52508032e-01 6.61668062e-01
-1.44042313e+00 1.74749434e-01 -9.46315706e-01 5.30055761e-01
4.59494032e-02 -4.17314708e-01 7.67867863e-01 1.09059203e+00
-8.27500820e-01 7.90511250e-01 -7.49032497e-01 -1.64135531e-01
2.97098488e-01 1.73120856e-01 3.04237962e-01 2.95808256e-01
-8.64038289e-01 1.03240228e+00 5.59875965e-01 2.80449633e-02
-9.72816646e-01 -5.35150588e-01 -3.06816697e-01 3.13598901e-01
1.00969934e+00 -6.19147658e-01 1.90146911e+00 -1.26781690e+00
-1.83089542e+00 3.82080764e-01 -1.00679167e-01 -8.83579195e-01
1.03544664e+00 -1.08548976e-01 -1.31333113e-01 2.62122173e-02
-8.21567625e-02 -2.73027897e-01 8.70876372e-01 -1.12213814e+00
-9.49860096e-01 -4.28046227e-01 5.75416625e-01 3.69063973e-01
-2.11957917e-01 2.16716215e-01 1.24603316e-01 -5.90993226e-01
-1.97369263e-01 -9.24804032e-01 -6.02439702e-01 -4.21833426e-01
-4.34489161e-01 -1.07373871e-01 -3.68733592e-02 8.37154016e-02
1.42662740e+00 -1.86278725e+00 -2.46572837e-01 5.28161347e-01
-2.41455555e-01 -1.37417704e-01 2.75104463e-01 3.43412310e-01
8.92878845e-02 4.02107775e-01 -7.09477440e-02 -2.53806889e-01
3.20741832e-01 3.30722749e-01 -7.56836832e-01 8.03130984e-01
-4.80869055e-01 5.80468297e-01 -9.29143965e-01 -2.62382746e-01
4.09529209e-02 -5.65687120e-01 -7.91267276e-01 5.05775847e-02
-4.96682644e-01 3.96292716e-01 -7.26859629e-01 5.74188232e-01
4.43252057e-01 -1.78252906e-01 3.99757326e-01 3.84161949e-01
-4.88213778e-01 1.95621520e-01 -1.23295605e+00 9.90714610e-01
-3.28718662e-01 1.40612632e-01 3.49327266e-01 -1.67638731e+00
2.05479935e-01 2.02838436e-01 6.66704297e-01 -3.15794855e-01
2.62468755e-01 2.52076805e-01 -2.74297029e-01 -3.25926304e-01
4.47979271e-01 -4.51443613e-01 1.34461690e-02 7.45434403e-01
-4.59138274e-01 8.56074318e-02 1.42096937e-01 -2.43981704e-01
7.25851238e-01 -1.70460790e-01 5.30576289e-01 -4.41098273e-01
2.38519117e-01 6.40556077e-03 6.28178954e-01 1.58834791e+00
-4.13994163e-01 -4.75391094e-03 7.49959409e-01 -3.44760776e-01
-8.48184943e-01 -9.27444994e-01 -9.93309990e-02 1.47783399e+00
1.26551360e-01 5.77242188e-02 -5.06822288e-01 -6.74332082e-01
3.58850896e-01 1.07827020e+00 -6.58131659e-01 1.45869568e-01
-1.70278698e-01 -8.93483877e-01 4.47163396e-02 2.10814834e-01
1.72269225e-01 -5.11607289e-01 -7.72671759e-01 4.42709148e-01
-3.35868709e-02 -6.73243761e-01 -4.09612656e-01 1.83205217e-01
-9.81537104e-01 -9.71073031e-01 -5.37560463e-01 2.47440249e-01
3.79458398e-01 3.31582457e-01 1.03999186e+00 -2.76868522e-01
4.58727568e-01 8.22221339e-01 -2.15705067e-01 -8.70483637e-01
-4.22996998e-01 4.61363699e-03 3.19790423e-01 1.60593510e-01
-1.11386413e-03 -3.54660720e-01 -7.95166910e-01 2.61903524e-01
-7.14185774e-01 -2.55030155e-01 3.41447562e-01 7.74699926e-01
6.24808371e-01 1.23914462e-02 5.41054010e-01 -8.21789503e-01
7.53109396e-01 -6.13469839e-01 -1.19205678e+00 5.47650456e-01
-9.92891133e-01 2.69984424e-01 5.49194753e-01 -2.99129456e-01
-1.20256710e+00 -1.71800762e-01 1.16350450e-01 -2.10701898e-01
1.96807116e-01 6.52276158e-01 1.00097954e-01 1.28833801e-01
6.31996930e-01 2.19570369e-01 2.44554672e-02 -5.70263267e-01
4.92110163e-01 7.04823554e-01 3.11848581e-01 -1.24230516e+00
3.45477313e-01 7.29158998e-01 6.90257847e-02 -4.88900423e-01
-1.43157220e+00 -3.82610291e-01 3.74190472e-02 -2.44564086e-01
3.95985842e-01 -4.31111157e-01 -1.17165148e+00 -1.94814671e-02
-7.93038845e-01 -3.47612977e-01 -5.39817691e-01 5.82283139e-01
-1.27563632e+00 2.68495470e-01 -1.21579561e-02 -1.65717185e+00
-4.03776206e-02 -1.24779713e+00 4.63976979e-01 -3.88209964e-03
7.27610961e-02 -1.02168655e+00 4.47626226e-02 2.55140513e-01
2.41172567e-01 2.04072684e-01 6.26434863e-01 -9.31447625e-01
-4.66830909e-01 -8.21613893e-02 1.96191266e-01 3.63159627e-01
-1.79591730e-01 -3.46413767e-03 -6.37331009e-01 -5.47325492e-01
5.31430423e-01 -2.31777459e-01 7.11951494e-01 9.06297326e-01
1.44452238e+00 -1.06159329e+00 -3.05272937e-01 5.53936958e-01
1.42070782e+00 4.27222699e-01 -2.26712171e-02 7.66138017e-01
-5.92450574e-02 5.44843614e-01 7.28673637e-01 9.59829688e-01
2.89072637e-02 5.03974855e-01 5.78992724e-01 5.01306355e-01
1.00243223e+00 -6.71427697e-02 2.81659365e-01 1.09088095e-02
-3.66119325e-01 -1.65727302e-01 -4.42518204e-01 5.45039058e-01
-2.18387008e+00 -1.41946185e+00 4.88287836e-01 2.77247810e+00
1.09255266e+00 4.29367095e-01 6.03239596e-01 -6.05022907e-02
7.79107869e-01 1.86480492e-01 -8.94266725e-01 -7.73685277e-01
-2.40786243e-02 -4.05309908e-03 1.00198090e+00 5.53602695e-01
-9.50814843e-01 5.72979152e-01 7.07449484e+00 1.00622821e+00
-9.32829201e-01 1.67978823e-01 9.33247030e-01 -6.10595047e-01
-3.72617275e-01 6.85698539e-02 -8.86254847e-01 6.51939332e-01
1.13547349e+00 -7.89676368e-01 7.43100882e-01 1.42183423e+00
5.83501637e-01 -2.27192193e-01 -1.21963799e+00 7.09565699e-01
-4.65266287e-01 -1.53790832e+00 -4.00686383e-01 1.61011726e-01
1.02746928e+00 -1.87553659e-01 3.50512326e-01 2.92230189e-01
9.60623264e-01 -8.21847677e-01 1.06492031e+00 4.74983603e-01
4.45096493e-01 -1.00783229e+00 3.30818057e-01 6.96367383e-01
-3.86301696e-01 -6.09695852e-01 -2.77796924e-01 -3.42309535e-01
4.19356339e-02 6.23590529e-01 -4.35435116e-01 5.50346971e-01
3.84841114e-01 1.10099934e-01 2.44455650e-01 1.22187686e+00
3.15479040e-02 6.27685308e-01 -6.63507938e-01 -1.20974332e-01
6.69527888e-01 -3.37915391e-01 6.96463108e-01 9.12832022e-01
3.58815521e-01 1.32237554e-01 3.79185826e-01 5.64145565e-01
1.51733458e-01 2.34145641e-01 -4.56639171e-01 -3.89924310e-02
4.76997972e-01 6.46922350e-01 -5.44029117e-01 -4.49007899e-01
-3.57948035e-01 4.70562607e-01 2.72264212e-01 4.49327826e-01
-8.24729145e-01 2.96118647e-01 9.42811549e-01 -2.40542695e-01
3.55637401e-01 -8.64010770e-03 -6.47946596e-01 -1.13281524e+00
9.57627594e-02 -9.33318615e-01 7.31608093e-01 -1.98426038e-01
-1.19563770e+00 -3.71603593e-02 4.02852058e-01 -9.66685116e-01
-5.56804121e-01 -4.79927778e-01 -4.22500044e-01 5.80346525e-01
-1.37458777e+00 -3.51493776e-01 4.89154339e-01 5.01875997e-01
4.38711911e-01 8.49132314e-02 4.45751488e-01 -3.91624957e-01
-5.83027244e-01 5.94970882e-01 1.12109303e+00 -5.63751638e-01
3.56032223e-01 -1.36897683e+00 -2.94863701e-01 8.35417211e-01
-3.40882801e-02 6.56061172e-01 1.16567087e+00 -2.97273427e-01
-1.38850212e+00 -8.00388515e-01 3.04962933e-01 -3.18675756e-01
9.73202765e-01 1.96811870e-01 -3.36962849e-01 7.48915553e-01
-1.61969885e-01 -8.45075846e-02 6.50088191e-01 4.25840527e-01
-1.45486295e-01 -3.20409715e-01 -1.16776609e+00 7.80253589e-01
8.90044808e-01 -1.21250235e-01 -5.18015444e-01 7.73970127e-01
6.45608664e-01 -4.14463729e-01 -6.83306813e-01 3.16229612e-01
8.27888012e-01 -1.22986627e+00 8.07406068e-01 -1.25843537e+00
2.45306219e-05 2.45469764e-01 -3.29142243e-01 -1.31146002e+00
-1.32719412e-01 -1.38021076e+00 -3.90623450e-01 6.34719908e-01
3.02295715e-01 -1.00554645e+00 8.03452611e-01 9.35952663e-01
-2.73349253e-03 -1.02358747e+00 -1.19134343e+00 -1.40978527e+00
5.18797517e-01 -9.04172361e-01 7.60002434e-01 7.02177584e-01
2.26817682e-01 -4.30211693e-01 -5.48831582e-01 5.22248708e-02
7.38237381e-01 5.02194881e-01 5.51164806e-01 -9.48640287e-01
-6.40792549e-01 -9.32686031e-01 2.44664043e-01 -1.35822856e+00
1.77441940e-01 -4.59583312e-01 3.71351652e-02 -1.14720106e+00
5.16449735e-02 -6.42397642e-01 -7.98856020e-01 1.95583895e-01
-1.05230972e-01 -4.04941678e-01 3.68830204e-01 2.96079218e-01
-6.66758716e-01 3.11208606e-01 9.79194403e-01 -2.33167205e-02
-3.62118721e-01 7.30302572e-01 -1.16314673e+00 7.43095636e-01
9.56580877e-01 -4.94230211e-01 -4.64967728e-01 -3.10364664e-02
4.79356676e-01 4.67185766e-01 3.95431519e-02 -5.36032736e-01
2.06042640e-02 -9.62777197e-01 -5.41199706e-02 -5.32846093e-01
-3.70654054e-02 -8.02491009e-01 1.07353497e-02 5.52303672e-01
-6.71524942e-01 -2.27675676e-01 -1.15907721e-01 8.32624435e-01
2.73783773e-01 -7.19898641e-01 8.66324544e-01 -4.59610075e-01
-2.30074838e-01 3.08548301e-01 -4.18317348e-01 1.37364671e-01
1.12205875e+00 -1.33909792e-01 -1.32081375e-01 -7.22585320e-01
-8.43012691e-01 3.97300869e-01 2.95600533e-01 -1.77542359e-01
7.15868771e-02 -1.03681028e+00 -5.18691003e-01 -2.62576878e-01
-1.18578508e-01 -4.69480276e-01 7.63183162e-02 1.07661998e+00
-4.40515503e-02 6.81920826e-01 1.62867427e-01 -3.30821544e-01
-8.34651232e-01 9.61188793e-01 3.82000923e-01 -5.52809656e-01
-1.11718960e-01 6.59739554e-01 8.79920498e-02 2.99966931e-01
3.65466744e-01 -4.45258558e-01 1.98093906e-01 1.18981794e-01
6.08163178e-01 5.63027859e-01 -8.04158896e-02 -1.16018824e-01
3.71183231e-02 -1.25755593e-01 -6.77475631e-02 -6.19109750e-01
1.17345035e+00 -4.23077285e-01 1.15555264e-01 5.65022349e-01
7.32638240e-01 5.85322902e-02 -1.44952071e+00 -3.05186450e-01
2.36424834e-01 -6.80331886e-01 1.82726979e-01 -8.32241833e-01
-6.68075204e-01 3.03427607e-01 2.15648517e-01 8.57269704e-01
1.04825866e+00 -3.46872240e-01 3.00301671e-01 6.92064106e-01
6.52076781e-01 -1.42934608e+00 -5.85437894e-01 4.94277179e-01
8.85480464e-01 -1.21737492e+00 1.10579185e-01 2.52635270e-01
-5.87586343e-01 9.52438891e-01 -5.51937222e-02 1.01731114e-01
6.35704517e-01 -1.52600601e-01 -3.68575484e-01 3.22190881e-01
-8.70990098e-01 -4.78212446e-01 8.84540193e-03 4.30283457e-01
-1.03312820e-01 3.49726498e-01 -6.35671854e-01 7.65245438e-01
-4.11527812e-01 2.43952181e-02 5.14162958e-01 9.46136594e-01
-7.97660768e-01 -8.20331454e-01 -6.67198777e-01 5.75106859e-01
-1.01969039e+00 -8.27947445e-03 -8.50117356e-02 5.80470264e-01
-4.24311757e-01 1.11450100e+00 -2.46220455e-02 1.95868984e-01
1.08610220e-01 6.93418384e-02 6.61777020e-01 -3.56094152e-01
-2.82086968e-01 1.43629819e-01 3.52815032e-01 -5.67755640e-01
-4.77067769e-01 -7.61864483e-01 -4.17348176e-01 -5.02008140e-01
-2.35052183e-01 5.63323617e-01 6.02033556e-01 1.14177036e+00
2.92695966e-02 -7.41238520e-02 1.00668085e+00 -6.47629619e-01
-1.68937707e+00 -6.91956699e-01 -7.89047539e-01 -2.27876678e-02
5.65810740e-01 -8.27484906e-01 -5.92273712e-01 -3.14073086e-01]
|
[4.516161918640137, 3.234903335571289]
|
76d1b3ff-64df-4528-babd-e34c00f2e134
|
multi-source-domain-adaptation-using-gradient
|
2109.01503
| null |
https://arxiv.org/abs/2109.01503v1
|
https://arxiv.org/pdf/2109.01503v1.pdf
|
Multi-source Domain Adaptation Using Gradient Reversal Layer for Mitotic Cell Detection
|
This is a write-up of our method submitted to Mitosis Domain Generalization (MIDOG 2021) Challenge held in MICCAI2021 conference.
|
['Satoshi Kondo']
|
2021-09-02
| null | null | null | null |
['cell-detection']
|
['computer-vision']
|
[ 6.04133368e-01 5.67221761e-01 -3.37317705e-01 -8.31362545e-01
-5.27650476e-01 -3.77932131e-01 8.91927958e-01 -9.05352924e-03
-2.86217749e-01 1.39650691e+00 7.20919073e-02 -1.56581581e-01
1.42084241e-01 -1.81430817e-01 -7.55255997e-01 -7.18280524e-02
-4.06130821e-01 6.92140102e-01 3.58762681e-01 -4.58141029e-01
2.84338053e-02 3.15070361e-01 -1.52702069e+00 1.16600835e+00
5.71481347e-01 1.08784044e+00 -3.49646211e-01 6.34018540e-01
6.53895974e-01 7.27061629e-01 -7.86171854e-01 9.62813348e-02
2.86748409e-01 -4.37043399e-01 -1.49506176e+00 -2.03739986e-01
9.11568522e-01 1.18849926e-01 -3.98298442e-01 9.90172803e-01
3.40473562e-01 4.93533701e-01 1.05580401e+00 -1.57595670e+00
-2.89480865e-01 9.92810309e-01 -3.68643403e-01 9.26264286e-01
5.11713743e-01 -3.13906640e-01 7.81058431e-01 -9.29620802e-01
8.61969411e-01 1.43804681e+00 7.66141415e-01 1.06396270e+00
-1.23636043e+00 -9.26050305e-01 3.31513435e-01 8.78582716e-01
-1.13039792e+00 -3.55332375e-01 5.44793248e-01 -4.21351552e-01
1.44795215e+00 2.62161672e-01 1.66734442e-01 1.83865905e+00
-2.12921202e-01 1.68333924e+00 1.28214502e+00 -2.08024740e-01
4.71383631e-01 1.28647331e-02 5.78526020e-01 9.78590548e-02
3.51282656e-01 3.13564420e-01 -5.34058571e-01 -4.51766253e-01
4.25103456e-01 -8.07111621e-01 -1.15608439e-01 -1.30158067e-01
-1.11668611e+00 8.74631047e-01 4.41592544e-01 4.34511453e-01
-9.18282121e-02 -1.34027928e-01 8.35216343e-01 6.29469156e-01
4.06722099e-01 5.41185141e-01 -6.24145031e-01 -1.21900305e-01
-1.00329292e+00 7.01183915e-01 6.02648497e-01 1.20537543e+00
1.85371056e-01 -2.09412843e-01 9.71142128e-02 9.66644049e-01
-2.88583934e-01 -1.18492544e-01 1.04516506e+00 -9.00240660e-01
6.17306590e-01 1.48209110e-01 -1.02125414e-01 -4.77000415e-01
-6.22260094e-01 9.54969227e-02 -8.59816134e-01 -3.46228480e-01
3.09277117e-01 -2.45155215e-01 -1.12693477e+00 1.53980160e+00
-2.24661052e-01 6.28960013e-01 2.25818992e-01 7.61654019e-01
1.09692073e+00 4.81265008e-01 4.12883639e-01 4.30918708e-02
7.18157291e-01 -1.22661602e+00 -3.23655307e-01 -2.86047459e-01
1.02293849e+00 -2.20073402e-01 4.78229403e-01 8.72867048e-01
-8.56502116e-01 -8.68921936e-01 -1.34025145e+00 6.58948272e-02
-7.37607062e-01 -1.63297325e-01 5.07850885e-01 2.46180356e-01
-6.87662840e-01 8.95445108e-01 -8.02122533e-01 -7.27403045e-01
5.48701525e-01 3.75828326e-01 -3.16539168e-01 -3.61844450e-02
-1.73204327e+00 7.10474908e-01 9.39708352e-01 -5.14063656e-01
-1.02836728e+00 -9.78990078e-01 -7.41783619e-01 -8.34543288e-01
2.39808373e-02 -7.76512399e-02 1.66954684e+00 -8.83464575e-01
-1.14014757e+00 1.29531288e+00 1.17764026e-01 -1.21069407e+00
5.03217101e-01 -4.32985604e-01 -1.20452487e+00 -5.73693365e-02
1.62877440e-01 1.13210618e+00 5.62738955e-01 -8.84786785e-01
-1.15840209e+00 -5.02342343e-01 8.51100534e-02 -5.23760766e-02
-3.94840926e-01 -3.22556555e-01 2.64503360e-01 -1.02966762e+00
-1.18007012e-01 -8.87844384e-01 -8.20006281e-02 -6.59137785e-01
-4.78096068e-01 -8.70546460e-01 1.64756966e+00 -6.29152000e-01
1.20262170e+00 -2.07031584e+00 1.51274875e-01 -6.46396652e-02
-6.07527941e-02 4.73070025e-01 -2.61576146e-01 2.68430591e-01
-7.54306078e-01 -3.33002359e-02 -4.88066912e-01 -3.19763035e-01
-2.01911226e-01 1.90666869e-01 -6.76458359e-01 5.53501129e-01
-1.63052946e-01 5.46868563e-01 -1.13008058e+00 -3.16872865e-01
-5.11738658e-02 5.78185841e-02 -2.02559456e-01 -8.30845311e-02
-4.71037298e-01 4.67230082e-02 -5.59918761e-01 5.73386848e-01
1.08415616e+00 -8.91483873e-02 2.28775650e-01 -2.61801958e-01
3.20866525e-01 5.89051008e-01 -1.03059745e+00 2.29141045e+00
-9.65790302e-02 6.59503818e-01 -1.78039134e-01 -1.81870496e+00
8.33055377e-01 6.19390249e-01 6.32570505e-01 -7.05939472e-01
-4.73850295e-02 4.23137516e-01 -1.90563276e-01 -3.47061343e-02
5.00740588e-01 5.01228012e-02 -1.40607387e-01 3.68905589e-02
4.29267585e-01 -3.53638649e-01 5.61175466e-01 4.93109338e-02
1.06192780e+00 3.96425605e-01 7.27127552e-01 -1.08269000e+00
7.46647060e-01 2.56763548e-01 4.23750043e-01 8.48211229e-01
-5.02826214e-01 6.12836540e-01 -1.63523108e-01 -9.13272500e-01
-8.12962770e-01 -9.29303229e-01 -6.43572092e-01 1.15480089e+00
2.15052180e-02 -3.83363068e-01 -7.55519986e-01 -1.56688738e+00
2.23931789e-01 9.11652803e-01 -1.07793474e+00 -3.27858150e-01
-8.65516543e-01 -4.43933696e-01 1.01996756e+00 8.12213778e-01
9.19565976e-01 -1.29595709e+00 -4.31777835e-01 1.91268742e-01
-3.26044187e-02 -1.27318645e+00 -3.50428909e-01 5.25976419e-01
-1.28061295e+00 -1.12986422e+00 -7.30586410e-01 -1.03318620e+00
4.18321006e-02 -2.38175809e-01 1.28626037e+00 -4.31370974e-01
-5.32391727e-01 -1.12849481e-01 -5.23720264e-01 -7.28051543e-01
-4.37737465e-01 9.04141843e-01 4.03334856e-01 -4.86517251e-01
9.80047703e-01 -6.84223831e-01 -2.90804148e-01 4.87361014e-01
-7.12808311e-01 -2.00176924e-01 1.89112008e-01 9.81823802e-01
5.10230064e-01 -2.22084537e-01 9.99078274e-01 -1.14117610e+00
7.99390316e-01 -6.14029467e-01 -6.83571517e-01 1.50530472e-01
-7.37349987e-01 -3.60862225e-01 9.36901569e-01 -3.12180161e-01
-8.67846489e-01 3.67306471e-01 1.45622760e-01 -1.28234237e-01
-6.17504835e-01 -1.18368462e-01 -3.01702380e-01 2.23544717e-01
1.33694887e+00 4.35416847e-01 -2.84503996e-01 -9.02984977e-01
1.49232909e-01 1.07521617e+00 1.20821917e+00 -8.83615792e-01
2.52005786e-01 3.04488659e-01 -4.23600078e-02 -8.19670796e-01
-9.36764181e-01 -5.81790328e-01 -6.30399466e-01 4.48191732e-01
6.45558298e-01 -8.98110986e-01 -1.80805907e-01 2.65455663e-01
-8.95910203e-01 -3.07984054e-01 -2.09892347e-01 2.04272568e-01
-8.23099673e-01 -5.58244996e-04 -1.05595574e-01 -2.52268612e-02
-3.03769469e-01 -7.04975784e-01 8.90320539e-01 3.77159715e-01
-5.27659655e-01 -1.33665776e+00 4.05751020e-01 3.12197775e-01
2.46714458e-01 2.28612334e-01 6.16467953e-01 -1.40954900e+00
4.86477733e-01 -3.71348619e-01 -1.27589464e-01 6.46323204e-01
2.84902126e-01 -5.21480560e-01 -1.45925689e+00 -5.12399495e-01
-4.27078754e-01 -7.31806278e-01 1.19765806e+00 3.93822879e-01
1.70103395e+00 -1.45204782e-01 -1.07665348e+00 5.21420121e-01
9.42743301e-01 3.43409657e-01 4.88914967e-01 6.37842715e-01
3.84014964e-01 1.40391260e-01 8.05657685e-01 3.16740185e-01
1.99283928e-01 8.99752915e-01 8.69390815e-02 9.65599045e-02
-3.67288113e-01 -1.09123424e-01 2.50756085e-01 -7.80037418e-02
1.36755005e-01 -3.63687575e-01 -1.32985401e+00 7.83915281e-01
-1.83376861e+00 -7.82727361e-01 -6.36638105e-02 1.56323099e+00
1.07505584e+00 1.89258337e-01 7.55138278e-01 3.66653800e-01
7.51364887e-01 2.50657424e-02 -6.84882998e-01 -6.42339528e-01
-3.22908968e-01 3.59547496e-01 3.27844262e-01 3.89172733e-01
-1.68252897e+00 1.01568389e+00 7.86813641e+00 1.21841574e+00
-9.54925060e-01 -8.29852968e-02 5.52388906e-01 2.20624715e-01
2.52856255e-01 -2.34496653e-01 -9.26813483e-01 4.21377480e-01
1.41256022e+00 -5.47013581e-01 1.34723395e-01 1.27284074e+00
-5.12937963e-01 9.85360593e-02 -1.93207192e+00 6.50517106e-01
1.80691004e-01 -1.42285359e+00 4.45253626e-02 9.30390656e-02
9.41359282e-01 5.12272358e-01 -1.49653792e-01 7.23345757e-01
2.42377639e-01 -1.45287490e+00 1.63766801e-01 2.26543471e-02
7.08567858e-01 -8.03304970e-01 7.32438684e-01 1.76917702e-01
-8.89213741e-01 -5.15438259e-01 -3.42664480e-01 1.32877827e-01
-2.34692037e-01 3.22066635e-01 -1.24157619e+00 5.87183952e-01
7.46286273e-01 1.35211372e+00 -7.15039849e-01 7.79650629e-01
-6.64693117e-02 1.00855756e+00 -3.35076690e-01 2.57414401e-01
1.39379531e-01 6.74415648e-01 8.64829302e-01 1.74818110e+00
-5.01410663e-01 1.62072629e-01 2.78659940e-01 2.23128527e-01
-5.47201633e-01 -6.64636672e-01 -4.76368725e-01 -4.29634042e-02
4.73000228e-01 7.52951026e-01 -1.30683839e-01 -7.44796634e-01
-2.68039614e-01 1.04074728e+00 -5.95553592e-02 3.16612661e-01
-7.09893942e-01 -4.72150564e-01 6.32358074e-01 1.84022486e-01
1.23031363e-01 4.24249381e-01 -4.58076894e-01 -8.43223572e-01
-6.34847283e-02 -1.31710756e+00 9.36163008e-01 -3.00381333e-01
-1.67090774e+00 7.55088806e-01 6.08589470e-01 -1.33738399e+00
-5.31895816e-01 -8.32298100e-01 -1.04034148e-01 6.15115404e-01
-1.24107981e+00 -1.06350636e+00 -2.49932647e-01 5.90830684e-01
7.58160114e-01 -8.39667797e-01 1.24569416e+00 2.68043131e-01
-4.53151576e-02 1.35163534e+00 2.00351223e-01 -1.95073672e-02
7.85068035e-01 -1.36043787e+00 1.15835106e+00 3.39919716e-01
1.11313567e-01 1.43558204e-01 6.44916475e-01 -2.23724455e-01
-5.64588070e-01 -1.33528996e+00 9.07650471e-01 -5.43472409e-01
5.98615408e-01 -3.11630428e-01 -7.49736071e-01 1.02757013e+00
2.24409178e-01 2.08104074e-01 2.48220384e-01 4.57915198e-03
-3.81618559e-01 2.13862002e-01 -1.36892056e+00 -8.76072198e-02
1.05770910e+00 -1.04887180e-01 -1.03461063e+00 5.99088609e-01
5.93791544e-01 -7.15274632e-01 -8.81302476e-01 8.65808010e-01
1.01834424e-01 -3.61779988e-01 1.03476381e+00 -1.32198906e+00
5.16739964e-01 1.16653115e-01 -5.48654199e-01 -1.44364715e+00
-2.11541474e-01 -5.18959165e-01 -3.70121181e-01 6.73070490e-01
5.41186571e-01 -2.00165547e-02 1.11129117e+00 -1.46588773e-01
-1.67849511e-01 -8.45494092e-01 -1.35986686e+00 -1.34990835e+00
9.05089974e-01 -3.76313180e-02 6.41382813e-01 1.24323499e+00
2.05396637e-01 1.02982134e-01 2.81162206e-02 -5.02079725e-04
8.43810499e-01 -8.37027878e-02 5.44378400e-01 -1.43156457e+00
-3.73313546e-01 -6.10583305e-01 -9.63856220e-01 -1.00243688e+00
2.72121906e-01 -1.24341547e+00 -1.97119355e-01 -1.11230183e+00
-8.02943669e-03 2.96555668e-01 -6.60892010e-01 6.83735609e-01
-1.05944211e-02 3.98324698e-01 1.75407380e-01 1.22709237e-01
-7.46620774e-01 3.26935381e-01 5.88383079e-01 -6.42581344e-01
-6.45683110e-02 2.24424437e-01 -6.98318362e-01 5.10452986e-01
6.69963062e-01 -5.47385097e-01 -5.14328361e-01 -4.30720806e-01
-7.71911502e-01 -1.97194725e-01 3.74210835e-01 -1.49941838e+00
-2.88822830e-01 -4.27869767e-01 4.62682784e-01 -4.18154508e-01
4.19989288e-01 -4.19086277e-01 -6.10542178e-01 3.82845640e-01
-7.56396711e-01 -2.94614136e-01 8.94823432e-01 2.88387120e-01
-4.95624363e-01 9.47413296e-02 1.14886367e+00 -2.10167468e-03
-1.12454367e+00 1.97078586e-01 -1.05192706e-01 8.12325358e-01
1.27576637e+00 -1.75085887e-01 -4.70251590e-01 1.70550421e-01
-1.28673065e+00 5.87962270e-01 -7.27867484e-02 6.35195434e-01
3.29761833e-01 -1.38133693e+00 -1.01750779e+00 8.66805390e-03
5.13825059e-01 -2.76548505e-01 -1.66945755e-01 3.66630435e-01
-2.30966836e-01 8.94381344e-01 -3.36005986e-01 -4.78454769e-01
-1.03085446e+00 4.68877971e-01 4.10904527e-01 -3.79304469e-01
-7.01267600e-01 1.51590157e+00 1.07605904e-01 -5.29577374e-01
4.90594149e-01 -8.84638280e-02 -1.29487589e-01 -4.11892980e-01
8.69025946e-01 6.48022294e-01 4.94494051e-01 -1.74189121e-01
-1.12145233e+00 9.00762621e-04 -5.84149420e-01 -3.51638012e-02
1.13513803e+00 4.38622624e-01 4.36370730e-01 1.86396569e-01
1.53207159e+00 -1.03456867e+00 -8.62355709e-01 -1.76604614e-01
7.33264506e-01 1.27185702e-01 -8.20672438e-02 -1.30855846e+00
-5.87424636e-01 9.60654736e-01 9.36273277e-01 -1.13502651e-01
1.13720596e+00 -1.31870300e-01 9.47224617e-01 9.71217275e-01
2.65402436e-01 -1.74374676e+00 -3.29698145e-01 6.65454626e-01
1.15011525e+00 -1.11770558e+00 8.53708312e-02 9.84679163e-03
-6.04780376e-01 1.21707606e+00 1.13220739e+00 -4.65174288e-01
8.84702623e-01 -6.84630424e-02 -1.71741903e-01 1.80129334e-01
-8.69210124e-01 2.71229148e-01 5.11530399e-01 1.31650460e+00
5.67513704e-01 1.07670113e-01 -5.33627748e-01 8.55136454e-01
-2.64739752e-01 5.70946813e-01 3.61929059e-01 8.94127965e-01
-3.70021015e-01 -1.07670903e+00 -3.33180316e-02 -7.32430741e-02
-6.17874786e-02 -5.78291193e-02 -8.59331369e-01 9.30969238e-01
2.89523125e-01 7.83467472e-01 -1.90715551e-01 -5.92982352e-01
5.78402579e-01 9.28503722e-02 5.77877402e-01 -8.87127638e-01
-1.66543335e-01 -7.34992743e-01 3.39867026e-01 -9.16512787e-01
-3.47983807e-01 -6.98692858e-01 -1.56332719e+00 8.43328387e-02
6.20811939e-01 5.96389532e-01 6.82299614e-01 8.99576306e-01
4.41908002e-01 4.50541079e-01 5.00375271e-01 -7.56166101e-01
-9.06183481e-01 -1.38147867e+00 -8.40687037e-01 5.69863260e-01
3.12885433e-01 -8.10459495e-01 -3.63878608e-01 5.46018600e-01]
|
[10.203283309936523, 2.9693894386291504]
|
86b09a6b-2903-4250-b486-b007559eb42e
|
real-time-spatio-temporal-action-localization
| null | null |
https://openaccess.thecvf.com/content/ACCV2020W/MMHAU/papers/Liu_Real-time_spatio-temporal_action_localization_via_learning_motion_representation_ACCVW_2020_paper.pdf
|
https://openaccess.thecvf.com/content/ACCV2020W/MMHAU/papers/Liu_Real-time_spatio-temporal_action_localization_via_learning_motion_representation_ACCVW_2020_paper.pdf
|
Real-time Spatio-temporal Action Localization via Learning Motion Representation
|
Abstract. Most state-of-the-art spatio-temporal (S-T) action localization methods explicitly use optical flow as auxiliary motion information. Although the combination of optical flow and RGB significantly improves the performance, optical flow estimation brings a large amount of computational cost and the whole network is not end-to-end trainable. These shortcomings hinder the interactive fusion between motion information and RGB information, and greatly limit its real-world applications. In this paper, we exploit better ways to use motion information in a unified end-to-end trainable network architecture. First, we use knowledge distillation to enable the 3D-Convolutional branch to learn motion information from RGB inputs. Second, we propose a novel motion cue called short-range-motion (SRM) module to enhance the 2DConvolutional branch to learn RGB information and dynamic motion information. In this strategy, flow computation at test time is avoided. Finally, we apply our methods to learn powerful RGB-motion representations for action classification and localization. Experimental results show that our method significantly outperforms the state-of-the-arts on
dataset benchmarks J-HMDB-21 and UCF101-24 with an impressive improvement of ∼ 8% and ∼ 3%.
|
['and Qianqing Qin', 'Xing Xie', 'Liyu Lin', 'Zhigang Tu', 'Yuanzhong Liu']
|
2020-11-30
| null | null | null |
accv-2020-11
|
['spatio-temporal-action-localization']
|
['computer-vision']
|
[ 1.99702736e-02 -5.09938598e-01 -6.00125849e-01 -1.54019982e-01
-5.91593504e-01 -3.74106914e-01 4.54790890e-01 -5.82048059e-01
-7.44734526e-01 8.63210618e-01 2.09075928e-01 -1.04219720e-01
3.78549099e-02 -5.76658487e-01 -7.32000053e-01 -8.72934878e-01
-4.80149873e-02 -1.61172181e-01 5.55568039e-01 2.91251857e-02
1.76071078e-01 3.44765306e-01 -1.28300369e+00 3.90139878e-01
7.08471239e-01 1.28234720e+00 3.01004469e-01 6.72589362e-01
-1.52806416e-01 1.45217168e+00 -3.40065390e-01 2.54567899e-02
2.96143651e-01 -6.63228929e-01 -9.22920406e-01 -4.61323671e-02
5.18272698e-01 -7.54946887e-01 -9.01086271e-01 7.22589910e-01
6.17622733e-01 4.86994982e-01 1.25436798e-01 -1.39255869e+00
-6.97257459e-01 2.01630905e-01 -5.50169587e-01 4.73491937e-01
4.06010449e-01 7.84917295e-01 6.40270650e-01 -7.18141973e-01
7.60227978e-01 1.25949585e+00 3.45780015e-01 7.53886163e-01
-8.08369398e-01 -4.46804464e-01 3.00419211e-01 7.85064459e-01
-1.13594854e+00 -3.89079392e-01 9.76579726e-01 -2.25431621e-01
9.60255563e-01 6.68340400e-02 9.05146301e-01 1.29251873e+00
2.54106838e-02 1.35977173e+00 9.20126736e-01 -1.99000351e-02
9.77954715e-02 -4.92002517e-01 -2.65619516e-01 1.09876215e+00
-3.02536666e-01 2.49901727e-01 -8.30694914e-01 4.41487342e-01
1.18341148e+00 1.51985019e-01 -5.54959476e-01 -3.71148616e-01
-1.53820825e+00 5.48297167e-01 1.00551975e+00 3.06137145e-01
-2.04870984e-01 7.85750389e-01 4.60099548e-01 1.27379254e-01
2.90019661e-01 8.25790390e-02 -4.73088890e-01 -7.42770255e-01
-8.17185283e-01 -1.61513034e-02 3.13709617e-01 7.32350469e-01
7.23075569e-01 1.24561086e-01 -3.45116138e-01 4.66278434e-01
2.48450011e-01 4.24704820e-01 6.81528151e-01 -1.37848675e+00
7.59340584e-01 6.51812077e-01 1.40902236e-01 -8.39507103e-01
-4.29702342e-01 -2.41243199e-01 -8.59068751e-01 2.35707000e-01
6.99885666e-01 3.53939719e-02 -9.85288560e-01 1.50636530e+00
2.64195591e-01 7.61272252e-01 -4.19847155e-03 1.43211150e+00
7.37272382e-01 4.36194628e-01 3.74302231e-02 -4.34355550e-02
7.90709078e-01 -1.51230049e+00 -7.54426539e-01 -3.18245232e-01
9.68707204e-01 -4.13985968e-01 1.16694713e+00 1.84116557e-01
-1.11459327e+00 -7.71677911e-01 -9.65055764e-01 -4.10089105e-01
-3.59918177e-01 3.07027787e-01 9.05048609e-01 2.66794443e-01
-8.63204837e-01 8.40009749e-01 -1.33878028e+00 -1.18844248e-01
7.38291323e-01 3.72821152e-01 -5.71434975e-01 -2.98384637e-01
-1.12163818e+00 7.07403004e-01 4.61654127e-01 4.37309861e-01
-9.25285220e-01 -6.35068834e-01 -9.05390203e-01 -3.37420434e-01
3.90185118e-01 -8.37552369e-01 9.72586274e-01 -9.52292502e-01
-1.79184878e+00 3.32561314e-01 -1.47075102e-01 -2.35305011e-01
8.50017726e-01 -4.90794718e-01 -1.53560966e-01 5.97275138e-01
1.50801223e-02 8.68510544e-01 5.26602268e-01 -8.30991924e-01
-8.06902051e-01 -2.26137936e-01 3.31067652e-01 1.54706672e-01
-3.53767455e-01 -3.05636466e-01 -8.15232456e-01 -5.91681421e-01
1.09116122e-01 -9.61214602e-01 -1.67605385e-01 5.33491850e-01
-1.66419730e-01 -1.20093226e-01 1.16236675e+00 -5.69751263e-01
1.10646987e+00 -1.89369440e+00 3.07010770e-01 -3.66828769e-01
8.47881008e-03 4.99799013e-01 -2.59823769e-01 4.24965983e-03
1.48955584e-01 -1.46958396e-01 -1.50220275e-01 -4.50220555e-01
-1.27927482e-01 4.48132336e-01 -3.81588787e-02 7.39048183e-01
3.78955603e-01 1.32149327e+00 -1.31534410e+00 -5.07271230e-01
7.25428879e-01 7.88065970e-01 -6.44391000e-01 1.48740649e-01
-1.25440493e-01 1.01718390e+00 -6.87343299e-01 7.52705097e-01
3.76955062e-01 -2.59780198e-01 -3.12701404e-01 -2.24154875e-01
-2.32985243e-03 3.51374924e-01 -1.06734264e+00 2.61803579e+00
-5.67067742e-01 7.22153068e-01 -4.07956004e-01 -7.77444899e-01
5.36247432e-01 4.55885679e-02 7.68868804e-01 -9.46075737e-01
1.41930684e-01 7.68220052e-02 -1.64771363e-01 -7.55934417e-01
1.71536028e-01 3.40480834e-01 2.47772634e-01 2.84990042e-01
2.27324054e-01 3.63412470e-01 2.11281553e-01 -2.32646503e-02
1.26541400e+00 9.39343274e-01 -1.57739878e-01 1.66603237e-01
7.63831854e-01 -2.64996469e-01 7.77578652e-01 4.47680950e-01
-6.47452593e-01 6.16153300e-01 4.60607976e-01 -6.72196746e-01
-5.80285549e-01 -9.11668181e-01 3.66840631e-01 1.06284630e+00
4.53125775e-01 -2.24141940e-01 -4.89872098e-01 -1.08446527e+00
-1.03506118e-01 3.30427140e-01 -7.45301366e-01 -2.87790149e-01
-1.04668486e+00 -4.46392059e-01 4.95030284e-01 1.07847571e+00
1.15143692e+00 -1.10620356e+00 -9.45633054e-01 1.59137607e-01
-4.96455908e-01 -1.39358354e+00 -6.76699460e-01 -1.12425439e-01
-9.31973696e-01 -1.11771977e+00 -8.59491408e-01 -4.57347602e-01
3.99123371e-01 3.53117198e-01 7.27913678e-01 -1.12271802e-02
-2.36992553e-01 3.26733083e-01 -5.94197690e-01 3.62233669e-01
1.56721175e-01 -1.86133943e-02 -2.56539583e-01 1.67462453e-01
2.34770477e-01 -6.27259672e-01 -1.14234197e+00 3.29003036e-01
-8.60240042e-01 -4.01071133e-03 5.70802212e-01 8.35735977e-01
4.67661262e-01 -4.14748043e-01 1.88849762e-01 -1.31442308e-01
-5.91478273e-02 -1.06428221e-01 -3.24230462e-01 2.03243420e-01
-1.59913465e-01 2.50318974e-01 4.99060810e-01 -5.85853100e-01
-9.03071702e-01 5.41675568e-01 -2.94343140e-02 -9.72151637e-01
-5.06404303e-02 2.49568582e-01 -1.85143992e-01 -3.29064250e-01
4.09740984e-01 4.68738914e-01 -1.08207107e-01 -4.16801631e-01
5.67622185e-01 2.49274015e-01 7.82299042e-01 -2.08003476e-01
5.19303679e-01 7.90676653e-01 9.26950723e-02 -4.83294606e-01
-8.12637568e-01 -6.11257672e-01 -9.65559065e-01 -4.60990459e-01
1.25716269e+00 -9.28119004e-01 -9.02346432e-01 6.37145877e-01
-1.24487007e+00 -7.17403293e-01 -1.25949591e-01 8.32191348e-01
-8.40433002e-01 4.44731146e-01 -6.60487056e-01 -5.55872023e-01
-1.71813965e-01 -1.38340867e+00 1.19049144e+00 2.03080222e-01
2.93312550e-01 -9.58709717e-01 -5.90162650e-02 4.94866699e-01
3.12459022e-01 4.60646212e-01 2.08339572e-01 8.98542106e-02
-1.00164926e+00 -5.88635877e-02 -3.75669181e-01 3.02806824e-01
1.77422538e-01 -2.64656425e-01 -9.00942206e-01 -2.07759947e-01
-2.57989079e-01 -5.49337864e-01 1.35373282e+00 2.32775763e-01
1.37659979e+00 -1.69206679e-01 -2.39410296e-01 1.05451977e+00
1.26204586e+00 4.60646339e-02 1.00018322e+00 4.65886801e-01
1.27090764e+00 2.84113616e-01 7.34991193e-01 3.78124803e-01
4.38574553e-01 7.74379075e-01 6.18861973e-01 -5.30115888e-02
-5.38841486e-01 -2.78187335e-01 6.44853473e-01 6.38268709e-01
-4.57217842e-01 -1.26631573e-01 -6.92990065e-01 4.16936100e-01
-2.27410722e+00 -1.04698360e+00 -7.22905323e-02 1.97341406e+00
7.87404001e-01 6.32213056e-02 1.25752762e-01 1.26973927e-01
3.46568763e-01 4.85020101e-01 -6.61494434e-01 3.36659461e-01
-9.46575329e-02 3.77238542e-02 5.94386756e-01 4.61067110e-01
-1.38104880e+00 1.21022737e+00 4.81382418e+00 6.86627924e-01
-1.33104694e+00 1.79491282e-01 5.27431309e-01 -3.53425115e-01
1.47271991e-01 -1.10024363e-01 -4.73169267e-01 5.50459027e-01
6.13065064e-01 4.17573065e-01 4.15126920e-01 6.90159619e-01
4.47807014e-01 -2.93853819e-01 -1.02309239e+00 1.20647347e+00
2.30471580e-03 -1.36749852e+00 -2.18931228e-01 -4.15255427e-02
6.65813863e-01 1.62504330e-01 -7.78753608e-02 2.71025777e-01
7.72731602e-02 -9.42290545e-01 6.21230900e-01 7.60418117e-01
7.11076081e-01 -7.50302494e-01 5.88785291e-01 2.00765491e-01
-1.47141111e+00 -2.14640439e-01 -3.58064383e-01 -9.28744525e-02
3.29909176e-01 2.34328195e-01 -1.78890958e-01 6.66586637e-01
6.58629000e-01 1.28938532e+00 -6.02992833e-01 1.11987543e+00
-5.41079223e-01 3.54151249e-01 -1.24903351e-01 -3.92909870e-02
7.04954803e-01 -1.30738150e-02 2.39077076e-01 1.08879137e+00
1.55608803e-01 2.52883788e-02 3.63379031e-01 6.60740197e-01
3.70234996e-02 -2.84394920e-01 -3.98716509e-01 -6.50323834e-03
-2.73491386e-02 1.06380546e+00 -6.38101816e-01 -2.69728243e-01
-5.10391593e-01 1.44770908e+00 3.86835039e-01 5.51396549e-01
-1.14823377e+00 -2.88229614e-01 7.48490632e-01 -1.12468399e-01
4.72080767e-01 -5.66470444e-01 1.32403657e-01 -1.35948670e+00
2.07846776e-01 -4.59849715e-01 2.27481961e-01 -7.26839304e-01
-9.26483095e-01 3.46072435e-01 -3.44083488e-01 -1.51769865e+00
-3.43508840e-01 -7.22818136e-01 -4.29342419e-01 5.62870979e-01
-1.75107896e+00 -1.34213090e+00 -6.11637056e-01 9.15862322e-01
4.93177176e-01 6.83280528e-02 5.11532068e-01 4.42474186e-01
-6.63460851e-01 5.91408610e-01 -5.77265210e-02 5.50850153e-01
7.79381990e-01 -1.17784095e+00 1.53893456e-01 9.42095697e-01
2.01608300e-01 1.31280333e-01 7.24836364e-02 -4.44418579e-01
-1.72114241e+00 -1.36091924e+00 4.03443158e-01 -5.42896569e-01
8.21792722e-01 -1.38554618e-01 -6.89610898e-01 4.91477489e-01
7.09004998e-02 9.25144792e-01 2.75204688e-01 -5.17497003e-01
-3.02502304e-01 -2.24243134e-01 -7.74485290e-01 5.55505276e-01
1.48217857e+00 -6.24713957e-01 -1.79179400e-01 2.62925178e-01
7.74188459e-01 -4.45584625e-01 -7.37754762e-01 3.74131650e-01
6.55540705e-01 -1.05422938e+00 1.00113153e+00 -6.88415408e-01
6.94079578e-01 -5.77554882e-01 -1.09508224e-01 -1.08445585e+00
-1.08676657e-01 -8.39161694e-01 -6.06749296e-01 8.45595956e-01
-2.54602339e-02 -3.18065703e-01 9.79190052e-01 3.41091335e-01
-2.30945155e-01 -1.08227956e+00 -1.12509286e+00 -8.14426839e-01
-1.07975297e-01 -5.19521296e-01 1.51694119e-01 9.58930850e-01
-2.77112313e-02 -2.84334961e-02 -5.30813456e-01 -2.14845285e-01
3.48797888e-01 1.07673757e-01 8.14352453e-01 -5.41611791e-01
-4.94702876e-01 -5.64144790e-01 -7.61663079e-01 -1.65016580e+00
2.44062424e-01 -6.29992068e-01 7.52328560e-02 -1.56982958e+00
-1.66327864e-01 -7.75913671e-02 -5.58744669e-01 6.64756238e-01
-3.69054168e-01 4.37006861e-01 4.65786308e-01 9.79510397e-02
-9.87116396e-01 9.45668995e-01 1.83204031e+00 -2.31365114e-01
-2.87119716e-01 -2.38216951e-01 -2.33379845e-02 6.88080668e-01
6.75101101e-01 -2.18093917e-01 -5.62876999e-01 -6.85876310e-01
-3.01573277e-01 1.57810912e-01 7.89592206e-01 -1.20453143e+00
3.52361828e-01 -2.57881761e-01 6.73842251e-01 -5.29027760e-01
5.54454148e-01 -6.24544561e-01 -4.90564317e-01 6.00208461e-01
-2.55893588e-01 -8.60363990e-02 7.23621994e-02 7.16685236e-01
-2.73770690e-01 2.17367500e-01 6.23537421e-01 -2.38779247e-01
-1.16520858e+00 6.09676659e-01 -3.83704863e-02 1.37577325e-01
1.04047978e+00 -2.12772429e-01 -5.35436749e-01 -2.93919146e-01
-5.77430844e-01 2.50642598e-01 2.68763363e-01 5.46794832e-01
7.31809855e-01 -1.52476406e+00 -3.66117209e-01 -2.31541377e-02
1.11837070e-02 -1.03372587e-02 3.04341257e-01 1.34803724e+00
-7.01136172e-01 4.12677228e-01 -4.25991684e-01 -7.46817350e-01
-8.39100063e-01 4.77714777e-01 3.76119584e-01 -2.23512113e-01
-6.82949483e-01 9.82339799e-01 3.61612476e-02 2.17376519e-02
5.10220945e-01 -4.43107843e-01 -1.07097523e-02 -1.60926133e-01
7.01933980e-01 5.96426487e-01 -3.36749136e-01 -6.12078428e-01
-6.05050802e-01 6.07967079e-01 3.03041130e-01 -1.20060846e-01
1.23333526e+00 -8.80544111e-02 1.53724119e-01 3.74415547e-01
1.63689530e+00 -5.90033531e-01 -1.90868342e+00 -8.13108832e-02
-1.51413694e-01 -8.43690455e-01 2.02448994e-01 -7.07082331e-01
-1.58510888e+00 1.22629368e+00 7.24856436e-01 -3.24329406e-01
1.41887474e+00 -1.88694865e-01 1.02791703e+00 5.80335855e-01
3.88217032e-01 -1.04375219e+00 5.71380377e-01 4.69345152e-01
8.05135071e-01 -1.42307198e+00 -7.08738491e-02 -1.90782607e-01
-6.42977178e-01 1.22914517e+00 8.79894197e-01 -1.43701866e-01
4.08079147e-01 -2.60947794e-02 5.75513877e-02 1.81518853e-01
-6.07821643e-01 -5.24987400e-01 4.02242869e-01 5.19584596e-01
3.49989206e-01 -3.55649441e-01 -9.58407298e-02 1.59245715e-01
2.88176775e-01 4.06805843e-01 1.58063248e-01 1.20385289e+00
-2.57995009e-01 -9.80218649e-01 1.38080925e-01 -8.11590627e-02
-2.72097409e-01 1.19851112e-01 -2.97646791e-01 9.53605175e-01
3.48594427e-01 7.89636374e-01 -3.47470492e-02 -5.79932928e-01
1.27335027e-01 -8.59128088e-02 7.32074320e-01 -4.78060283e-02
-2.85542995e-01 1.24014512e-01 -3.24952193e-02 -1.42212522e+00
-8.74587178e-01 -4.38945472e-01 -1.52163982e+00 -2.16523945e-01
-1.46209911e-01 -4.53706771e-01 4.92797524e-01 1.11215568e+00
4.15926188e-01 6.91754758e-01 4.48592931e-01 -1.05378604e+00
-1.98439091e-01 -9.33864534e-01 -1.77400574e-01 3.71386707e-01
5.78777850e-01 -8.08508277e-01 -2.29790092e-01 1.97647035e-01]
|
[8.431056022644043, 0.3152414560317993]
|
f2080d93-dd19-4165-97a7-b31336062ad4
|
crackformer-transformer-network-for-fine
| null | null |
http://openaccess.thecvf.com//content/ICCV2021/html/Liu_CrackFormer_Transformer_Network_for_Fine-Grained_Crack_Detection_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Liu_CrackFormer_Transformer_Network_for_Fine-Grained_Crack_Detection_ICCV_2021_paper.pdf
|
CrackFormer: Transformer Network for Fine-Grained Crack Detection
|
Cracks are irregular line structures that are of interest in many computer vision applications. Crack detection (e.g., from pavement images) is a challenging task due to intensity in-homogeneity, topology complexity, low contrast and noisy background. The overall crack detection accuracy can be significantly affected by the detection performance on fine-grained cracks. In this work, we propose a Crack Transformer network (CrackFormer) for fine-grained crack detection. The CrackFormer is composed of novel attention modules in a SegNet-like encoder-decoder architecture. Specifically, it consists of novel self-attention modules with 1x1 convolutional kernels for efficient contextual information extraction across feature-channels, and efficient positional embedding to capture large receptive field contextual information for long range interactions. It also introduces new scaling-attention modules to combine outputs from the corresponding encoder and decoder blocks to suppress non-semantic features and sharpen semantic cracks. The CrackFormer is trained and evaluated on three classical crack datasets. The experimental results show that CrackFormer achieves ODS values of 0.871, 0.877 and 0.881, respectively, on the three datasets and outperforms the state-of-the-art methods.
|
['Hui Kong', 'Chengzhong Xu', 'Christoph Mertz', 'Xiangyu Miao', 'Huajun Liu']
|
2021-01-01
| null | null | null |
iccv-2021-1
|
['crack-segmentation']
|
['computer-vision']
|
[ 1.67577311e-01 -9.48671550e-02 5.01771688e-01 -1.63576871e-01
-1.02960885e+00 7.65198795e-03 3.22753079e-02 2.71249890e-01
-2.34784976e-01 2.77047038e-01 1.89559966e-01 2.63383061e-01
2.52373870e-02 -1.02086377e+00 -7.94019938e-01 -1.03644812e+00
6.42879680e-02 -8.48772302e-02 9.28563297e-01 -2.69054264e-01
7.51512170e-01 3.27470958e-01 -1.62265825e+00 6.35076940e-01
9.30940211e-01 1.07374179e+00 4.64521468e-01 5.57945967e-01
3.28159809e-01 6.90098166e-01 -2.10127622e-01 -1.24642001e-02
-4.38502669e-01 -2.69022100e-02 -4.67573404e-01 -6.65535405e-02
3.71452451e-01 -3.36145014e-01 -4.99527484e-01 8.40704978e-01
5.81274867e-01 -1.45554826e-01 8.95498753e-01 -5.25794327e-01
-9.04030800e-01 4.80784267e-01 -9.39781189e-01 3.15237761e-01
3.58683616e-02 1.13272294e-01 1.33161569e+00 -1.61936617e+00
2.17223108e-01 1.33979487e+00 8.91065955e-01 1.38982579e-01
-9.40152526e-01 -5.32140672e-01 7.81954080e-02 4.10959810e-01
-1.43246853e+00 -1.27829626e-01 1.07867837e+00 -5.67647159e-01
1.00293255e+00 -7.54523799e-02 2.57135540e-01 8.17203939e-01
5.20614445e-01 5.31798065e-01 5.69307268e-01 -3.38688731e-01
2.78021008e-01 -5.06454885e-01 3.66528273e-01 1.01460278e+00
5.44321597e-01 -1.27765760e-01 -4.87098306e-01 1.81962505e-01
1.08196568e+00 3.07137351e-02 -3.13993156e-01 1.08920723e-01
-9.36478913e-01 1.06827044e+00 1.05308735e+00 2.84269482e-01
-4.17392582e-01 5.66530883e-01 2.73917913e-01 -1.25484942e-02
2.63663173e-01 2.71715522e-01 -2.57546417e-02 4.28665459e-01
-3.57383579e-01 2.40369335e-01 3.57963964e-02 5.03021061e-01
9.42868531e-01 3.55016366e-02 2.28961241e-02 1.18583226e+00
8.11505735e-01 6.42641544e-01 1.75192431e-01 -7.04278469e-01
4.63837594e-01 6.82832599e-01 -2.58518577e-01 -1.31142223e+00
-5.94207287e-01 -5.10596037e-01 -8.25265110e-01 2.18453154e-01
6.08071350e-02 1.41415710e-03 -9.62726891e-01 1.05181599e+00
-1.10874670e-02 2.10594460e-01 -1.22993037e-01 9.57629204e-01
1.07904506e+00 7.12342143e-01 -1.35762274e-01 3.32143426e-01
1.41172552e+00 -1.07871234e+00 -5.17383337e-01 -6.57402098e-01
3.78283024e-01 -8.22166502e-01 1.26308846e+00 1.76636308e-01
-7.32198536e-01 -6.23209238e-01 -1.38173532e+00 -1.01582587e-01
-1.37038663e-01 3.71633619e-01 1.15917899e-01 9.18150991e-02
-5.53532124e-01 3.56623024e-01 -8.94885004e-01 -1.75368011e-01
6.66313589e-01 3.45333405e-02 -5.73048554e-02 -4.62993830e-01
-1.02883768e+00 4.59254205e-01 -1.02034248e-01 5.10659993e-01
-9.57600594e-01 -4.04293418e-01 -1.05677521e+00 2.19412342e-01
-9.91634503e-02 -1.36873141e-01 8.19951475e-01 4.04764414e-02
-9.95787203e-01 4.97327834e-01 3.77463065e-02 5.98301068e-02
-1.77146181e-01 -4.02405858e-01 -1.04559891e-01 5.46699345e-01
5.05400300e-01 2.84186840e-01 9.50849891e-01 -1.49771297e+00
-4.67653483e-01 -3.02848786e-01 -1.59357995e-01 1.05888844e-02
-8.91298354e-02 -1.30328804e-01 -1.76112473e-01 -9.35778677e-01
5.18263280e-01 -5.78099251e-01 -3.26893032e-01 9.82952118e-02
-6.45075381e-01 -1.58141866e-01 1.07892263e+00 -5.61805487e-01
1.25650096e+00 -2.19416189e+00 -3.40741612e-02 2.20393419e-01
2.55261987e-01 -8.91385823e-02 -2.19780952e-01 7.09580243e-01
3.48978885e-03 -1.24695458e-01 -9.86619055e-01 -1.06605433e-01
-3.32564980e-01 1.21983320e-01 1.23884849e-01 6.23816848e-01
7.66455531e-01 7.06818521e-01 -6.18026972e-01 -4.93542284e-01
1.27701506e-01 5.51310182e-01 -7.35576034e-01 2.50462770e-01
1.34927824e-01 1.93248406e-01 -5.77376544e-01 1.04782522e+00
7.91883826e-01 -2.88388014e-01 -6.63678527e-01 -4.50199902e-01
-3.27377796e-01 -2.74623901e-01 -1.04337084e+00 1.61487472e+00
-4.46940303e-01 7.32792854e-01 1.33090153e-01 -9.08398211e-01
1.16515386e+00 2.51026273e-01 1.39785260e-01 -6.88496232e-01
2.24998534e-01 5.94951928e-01 -8.95002931e-02 -1.01154697e+00
1.15443751e-01 -1.23271249e-01 -2.09014580e-01 2.50732362e-01
-2.92621136e-01 -1.91342264e-01 -5.79922460e-03 -1.01778388e-01
1.29418790e+00 -4.79731917e-01 -5.51176071e-01 -4.98518556e-01
6.28647864e-01 -3.01274598e-01 3.30389470e-01 3.63174587e-01
1.14118889e-01 1.13471317e+00 1.45400941e-01 -4.09190089e-01
-8.07795107e-01 -1.09454322e+00 -2.93901712e-01 7.75205016e-01
5.88125706e-01 -6.21292256e-02 -6.66419148e-01 -3.96847993e-01
1.08727545e-01 1.94884494e-01 -8.45965326e-01 -4.69041079e-01
-8.72133434e-01 -9.46103930e-01 5.97865522e-01 1.00866139e+00
7.11765110e-01 -1.10842562e+00 -8.45886469e-01 3.81842166e-01
-2.85163552e-01 -9.09899056e-01 -4.43115860e-01 1.66529283e-01
-6.43511474e-01 -1.32212615e+00 -7.40126073e-01 -1.26122165e+00
6.97208881e-01 3.48998427e-01 7.05265820e-01 4.60270464e-01
-6.91442132e-01 -2.79462300e-02 -5.96083581e-01 -1.20460637e-01
2.42483616e-01 9.31195319e-02 -6.35900736e-01 3.72733504e-01
-3.11919712e-02 -3.74805242e-01 -9.56214070e-01 3.32207918e-01
-8.85097802e-01 -2.47863680e-01 9.50734317e-01 1.40576053e+00
4.90909755e-01 2.51465052e-01 7.91132689e-01 -6.41290843e-01
5.98963857e-01 -5.14825225e-01 -9.34084952e-02 -2.63067216e-01
-5.25373816e-01 5.82210766e-03 2.55917877e-01 9.51078609e-02
-1.27007222e+00 -7.77816176e-02 -6.25207067e-01 -9.36437696e-02
-8.00508484e-02 7.85802186e-01 -7.94956684e-02 -8.66241157e-02
6.84492290e-01 -3.09433718e-03 -2.90724397e-01 -5.56159794e-01
-4.23574261e-02 7.61149526e-01 6.26751125e-01 -6.86713219e-01
7.83800483e-01 5.93983293e-01 -1.42301857e-01 -1.27332425e+00
-7.44103134e-01 -4.95920658e-01 -5.68269074e-01 -4.87874568e-01
1.12976789e+00 -9.64997053e-01 -4.44206744e-01 1.10019732e+00
-1.30989420e+00 -2.13637829e-01 1.56437643e-02 4.52710301e-01
-9.75970253e-02 4.31066155e-01 -1.01550984e+00 -6.61529660e-01
-4.40985709e-01 -1.34502470e+00 1.46527171e+00 4.36061621e-01
3.09552372e-01 -6.10334456e-01 7.51669751e-03 3.99981797e-01
4.71306592e-01 4.26057369e-01 1.07385957e+00 1.79746732e-01
-6.24052286e-01 -2.17640415e-01 -7.30904937e-01 3.02809477e-01
7.42547540e-03 7.24941958e-03 -1.18770611e+00 -9.78139266e-02
-3.34328413e-01 -3.68548810e-01 1.47715890e+00 4.92768258e-01
7.44961798e-01 2.53041536e-01 -3.57771188e-01 8.16522837e-02
1.65539837e+00 1.40134990e-01 7.14309275e-01 1.25316679e-01
1.07540858e+00 5.66462457e-01 5.44052482e-01 3.69520158e-01
4.03989136e-01 2.89425611e-01 1.00800908e+00 -3.45175058e-01
-1.71798453e-01 6.14911430e-02 4.06025499e-02 8.17390501e-01
-1.99373752e-01 -3.12607996e-02 -1.13424909e+00 9.47585702e-01
-1.82284355e+00 -9.64046657e-01 -8.71710539e-01 1.59046507e+00
4.43889320e-01 3.59048992e-01 -4.33783472e-01 5.92771232e-01
7.65761316e-01 2.55819768e-01 -5.60020864e-01 -2.12736174e-01
-1.78177923e-01 3.89634579e-01 2.88558632e-01 6.24129295e-01
-1.09176028e+00 7.51875758e-01 4.93120003e+00 6.13344967e-01
-9.04167354e-01 1.97556317e-01 5.04035294e-01 5.74353933e-01
-2.93455780e-01 -2.91541725e-01 -3.75772387e-01 3.06666046e-01
2.64347017e-01 6.79375768e-01 -1.89825252e-01 6.07974648e-01
1.02328904e-01 -9.61167589e-02 -6.08972967e-01 8.37827325e-01
7.34048188e-02 -1.33177984e+00 -3.30730289e-01 -2.31912315e-01
7.26501346e-01 2.09984481e-01 -1.27613489e-02 -8.91653523e-02
-1.31234944e-01 -1.02623379e+00 8.82944047e-01 3.92642915e-01
7.06640840e-01 -9.34463084e-01 1.08988059e+00 -4.20161933e-02
-1.68910241e+00 -5.40006280e-01 -4.24753010e-01 -1.20175533e-01
1.65523514e-01 8.62453759e-01 -1.00154415e-01 2.74178654e-01
1.12726319e+00 1.23151243e+00 -4.14448917e-01 1.09568989e+00
-3.32290560e-01 7.24843442e-01 -2.64467597e-01 1.89342037e-01
5.27041972e-01 1.42311066e-01 1.59941077e-01 1.44993818e+00
3.68607432e-01 1.34531289e-01 1.94767993e-02 7.68144071e-01
-2.70388685e-02 -8.11577663e-02 -3.48579019e-01 6.05342448e-01
6.27157867e-01 1.03826380e+00 -7.68141747e-01 1.27940819e-01
-4.98963714e-01 5.98142982e-01 1.08871445e-01 2.41708905e-01
-8.31862748e-01 -9.64079916e-01 6.31347537e-01 2.85320789e-01
5.80320954e-01 -1.24770448e-01 -5.53825796e-01 -7.88749635e-01
1.25939315e-02 -4.27769184e-01 1.80634752e-01 -7.78784215e-01
-1.32084584e+00 5.97864151e-01 -3.37757975e-01 -9.07491624e-01
7.31398880e-01 -6.33768499e-01 -1.17956817e+00 5.47953606e-01
-1.73095083e+00 -1.40247667e+00 -7.41347075e-01 3.89893234e-01
9.23102438e-01 3.59197944e-01 3.58804107e-01 6.07432127e-01
-8.73442590e-01 5.12663007e-01 9.43969265e-02 4.46909010e-01
4.64590907e-01 -1.01421344e+00 2.88773358e-01 1.00039542e+00
-3.98451328e-01 1.95941806e-01 4.07190174e-01 -5.60838461e-01
-1.21460116e+00 -1.38940346e+00 5.86961806e-01 9.38110575e-02
4.14992392e-01 -1.40456170e-01 -1.34196508e+00 2.79724687e-01
2.64934003e-01 1.60553128e-01 2.35671714e-01 -2.87912667e-01
-4.75737393e-01 -8.22491273e-02 -8.53565156e-01 1.14261687e-01
7.63346374e-01 -4.33268845e-01 -5.17027378e-01 -5.99761531e-02
3.99699420e-01 -3.48497361e-01 -1.00784576e+00 4.79070187e-01
6.85565531e-01 -1.02367687e+00 1.03203213e+00 3.76094282e-01
8.84777486e-01 -3.07302415e-01 -3.84338766e-01 -1.05727148e+00
-6.56655788e-01 6.55940548e-02 2.73984998e-01 9.01443601e-01
4.34734702e-01 -5.77700198e-01 7.32838035e-01 -2.72238795e-02
-1.01925361e+00 -1.08849967e+00 -1.02502382e+00 -1.72259048e-01
8.48213509e-02 -3.79205883e-01 4.58223760e-01 6.59577668e-01
-1.74811810e-01 6.95131779e-01 2.64132314e-06 7.11749136e-01
9.07818377e-01 2.69950777e-01 7.79708773e-02 -1.66991460e+00
1.05945036e-01 -4.31570321e-01 -2.66117185e-01 -1.02131248e+00
-3.56079102e-01 -3.88906747e-01 5.51819205e-01 -1.84944344e+00
2.90375967e-02 -5.51354170e-01 -4.09292132e-01 5.24673164e-01
-2.40392312e-01 5.50433695e-01 -3.79687726e-01 2.51923978e-01
-1.48399413e-01 8.58214557e-01 1.35693824e+00 -5.01284540e-01
1.30716175e-01 -2.22424179e-01 -3.79188985e-01 7.98640490e-01
8.40827584e-01 -5.58285356e-01 -2.52734601e-01 -7.55328655e-01
2.00863779e-01 -1.63286954e-01 5.52563190e-01 -1.42002380e+00
2.31686562e-01 1.24289408e-01 1.87706560e-01 -8.74590039e-01
3.73995900e-01 -6.82205200e-01 -2.50612795e-01 9.33968067e-01
-2.00889468e-01 -8.29030126e-02 3.48445848e-02 8.46300900e-01
-5.08269012e-01 -3.02238435e-01 1.21326280e+00 -4.64191139e-02
-6.18259907e-01 6.95953816e-02 -6.75803244e-01 2.98335552e-02
8.63549829e-01 -2.76345313e-01 -4.37772691e-01 1.73392937e-01
-5.26724577e-01 9.61144269e-02 2.72502124e-01 2.28172645e-01
1.30655897e+00 -1.30039537e+00 -9.84011948e-01 5.37039161e-01
4.58227992e-01 5.07899463e-01 5.84940135e-01 7.17792451e-01
-9.60901916e-01 5.49592897e-02 -1.10414691e-01 -7.88086355e-01
-1.07088554e+00 -1.32086337e-01 3.15489024e-01 -1.20965295e-01
-7.91612506e-01 1.27175510e+00 4.29741651e-01 1.70696117e-02
-3.69532332e-02 -8.73954356e-01 -3.54069859e-01 -3.79771646e-03
3.08405221e-01 6.44593775e-01 2.09833384e-01 -7.72455037e-01
-3.78506154e-01 1.46754169e+00 -3.51999365e-02 4.78289694e-01
1.69526434e+00 -1.17175221e-01 -1.48619309e-01 2.03306124e-01
1.17440414e+00 -2.13600919e-01 -1.27781022e+00 -2.65814215e-01
-1.31664500e-01 -3.13448519e-01 2.25347355e-01 -2.71774918e-01
-1.49792051e+00 1.33300281e+00 7.01128185e-01 1.28114522e-01
1.13900626e+00 2.79815793e-01 1.13513613e+00 2.36733183e-01
6.50420338e-02 -1.06579196e+00 7.61477530e-01 5.27734578e-01
1.13622618e+00 -1.22405648e+00 -2.02504814e-01 -5.79491138e-01
-3.28765780e-01 1.17592275e+00 8.50513279e-01 -5.31386614e-01
9.64029312e-01 3.86936098e-01 1.36338093e-03 -9.93452728e-01
-5.36225021e-01 -1.40984178e-01 -8.71547610e-02 4.80025291e-01
3.49900365e-01 -2.66104192e-01 -1.44371334e-02 6.89050794e-01
6.20933712e-01 -5.59177876e-01 5.17521381e-01 1.12559676e+00
-1.11746573e+00 -5.56119680e-01 -5.09157598e-01 5.32368541e-01
-2.37741157e-01 -5.61704114e-03 1.74977072e-02 5.26616395e-01
3.97169381e-01 1.17168832e+00 -4.97686826e-02 -5.91436267e-01
6.12519741e-01 -5.83854258e-01 1.55947907e-02 -7.62625456e-01
-4.67530012e-01 1.72328219e-01 -5.47381826e-02 -3.86658043e-01
-2.77366787e-01 -5.47661960e-01 -1.65092599e+00 3.61645877e-01
-6.69380248e-01 1.58548336e-02 3.29301506e-01 7.70028591e-01
2.47709289e-01 8.68075430e-01 7.96777248e-01 -1.00015664e+00
-1.27236694e-01 -1.10276508e+00 -6.70583189e-01 1.91031843e-01
3.97086173e-01 -1.03079152e+00 -3.82221729e-01 2.52143532e-01]
|
[7.522474765777588, 1.4770376682281494]
|
5e78c6a6-c478-4537-99b3-da4f5efea1c1
|
pippi2021-an-approach-to-automated-diagnosis
|
2211.02639
| null |
https://arxiv.org/abs/2211.02639v1
|
https://arxiv.org/pdf/2211.02639v1.pdf
|
PIPPI2021: An Approach to Automated Diagnosis and Texture Analysis of the Fetal Liver & Placenta in Fetal Growth Restriction
|
Fetal growth restriction (FGR) is a prevalent pregnancy condition characterised by failure of the fetus to reach its genetically predetermined growth potential. We explore the application of model fitting techniques, linear regression machine learning models, deep learning regression, and Haralick textured features from multi-contrast MRI for multi-fetal organ analysis of FGR. We employed T2 relaxometry and diffusion-weighted MRI datasets (using a combined T2-diffusion scan) for 12 normally grown and 12 FGR gestational age (GA) matched pregnancies. We applied the Intravoxel Incoherent Motion Model and novel multi-compartment models for MRI fetal analysis, which exhibit potential to provide a multi-organ FGR assessment, overcoming the limitations of empirical indicators - such as abnormal artery Doppler findings - to evaluate placental dysfunction. The placenta and fetal liver presented key differentiators between FGR and normal controls (decreased perfusion, abnormal fetal blood motion and reduced fetal blood oxygenation. This may be associated with the preferential shunting of the fetal blood towards the fetal brain. These features were further explored to determine their role in assessing FGR severity, by employing simple machine learning models to predict FGR diagnosis (100\% accuracy in test data, n=5), GA at delivery, time from MRI scan to delivery, and baby weight. Moreover, we explored the use of deep learning to regress the latter three variables. Image texture analysis of the fetal organs demonstrated prominent textural variations in the placental perfusion fractions maps between the groups (p$<$0.0009), and spatial differences in the incoherent fetal capillary blood motion in the liver (p$<$0.009). This research serves as a proof-of-concept, investigating the effect of FGR on fetal organs.
|
['Andrew Melbourne', 'Anna David', 'Sebastien Ourselin', 'Rosalind Aughwane', 'Kasia Maksym', 'Nada Mufti', 'Dimitra Flouri', 'Zhanchong Ou', 'Ashay Patel', 'Paula Ramirez Gilliland', 'Aya Mutaz Zeidan']
|
2022-11-01
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[ 1.09189022e-02 1.01902440e-01 -1.88509390e-01 -4.13899630e-01
-2.71523237e-01 -5.54830432e-01 2.45885670e-01 2.46450469e-01
-1.42745376e-01 2.84386873e-01 3.76653462e-03 -5.71261227e-01
-5.30732810e-01 -7.55256712e-01 -5.88195562e-01 -9.32121277e-01
-8.51232469e-01 5.71123600e-01 -4.46285754e-02 1.38024732e-01
1.69146031e-01 9.64339793e-01 -6.28917634e-01 5.32282114e-01
7.76186883e-01 7.81242788e-01 8.10760781e-02 9.09387410e-01
-2.83033729e-01 7.82785118e-01 9.63088721e-02 -1.53278172e-01
1.59093872e-01 -5.95304787e-01 -5.27990758e-01 -4.25194383e-01
3.25328112e-01 -5.64334035e-01 -1.80740416e-01 4.31801200e-01
8.39049935e-01 1.40857697e-03 9.38413322e-01 -6.86023355e-01
-4.41766620e-01 5.24099708e-01 -8.43319178e-01 9.57508206e-01
2.52416190e-02 -1.73789382e-01 -5.25754206e-02 -8.31030250e-01
4.62296128e-01 8.60323966e-01 7.73392737e-01 5.84627926e-01
-1.02982414e+00 -7.25352943e-01 -5.56550860e-01 8.31447840e-02
-1.05092824e+00 -6.83864713e-01 3.39290440e-01 -9.80480909e-01
9.01337445e-01 3.71319920e-01 7.41801083e-01 3.49323064e-01
5.27759612e-01 -2.75887698e-02 9.85371172e-01 -4.73093539e-01
-1.89750433e-01 -3.03954154e-01 -5.64987779e-01 1.06871021e+00
3.25731099e-01 8.35475624e-02 -1.80764571e-01 -1.82825670e-01
1.14333177e+00 -3.15861136e-01 -2.76782531e-02 -2.39864349e-01
-6.97856188e-01 6.64526761e-01 -1.58422336e-01 7.79664218e-01
-3.53317857e-01 1.74009591e-01 5.06338775e-01 4.01898146e-01
6.57565832e-01 2.39636704e-01 -4.13233817e-01 -1.32093772e-01
-9.26484466e-01 -1.11699224e-01 4.36126709e-01 3.33648503e-01
4.40739393e-01 1.64476827e-01 -1.37410641e-01 9.82253492e-01
6.86030030e-01 5.78552783e-01 6.87761545e-01 -1.17657304e+00
4.19249207e-01 2.99522430e-01 -2.50626564e-01 -1.15541220e+00
-7.93556452e-01 1.13916239e-02 -8.89326572e-01 2.12624986e-02
5.94542682e-01 -3.19914788e-01 -7.54859984e-01 1.35486615e+00
5.46348274e-01 3.19787085e-01 -1.55805513e-01 8.58346462e-01
9.90893483e-01 2.70744026e-01 2.60613143e-01 -6.06453717e-01
9.99662876e-01 -5.18819213e-01 -4.91749674e-01 1.82337090e-01
9.47762132e-01 -4.14656729e-01 2.18451425e-01 6.15098923e-02
-1.59131086e+00 -4.52221967e-02 -6.02680326e-01 3.02689552e-01
1.26367226e-01 -9.60986242e-02 7.77936161e-01 1.10964596e+00
-1.04977870e+00 1.00928175e+00 -1.54525065e+00 -2.65628725e-01
6.63163066e-01 3.79798025e-01 -7.66418993e-01 -2.39325732e-01
-7.95126855e-01 1.42637277e+00 -2.51722455e-01 3.71422172e-01
-7.97861516e-01 -1.29129899e+00 -9.71144915e-01 6.63654655e-02
-3.84826690e-01 -2.50377864e-01 5.03268838e-01 -5.74035585e-01
-1.15354908e+00 9.70978498e-01 -1.75171182e-01 -2.38344166e-02
6.55453384e-01 2.38914102e-01 -4.41353619e-01 8.32373261e-01
3.97663340e-02 -6.24140091e-02 4.52392876e-01 -8.97840321e-01
-9.70727503e-02 -6.68888569e-01 -6.48061275e-01 -3.57559994e-02
2.30522797e-01 7.67870724e-01 2.32977360e-01 -2.66728759e-01
6.65948093e-01 -5.84602356e-01 -3.02663684e-01 1.57756433e-01
5.06934404e-01 3.87936532e-01 1.77962095e-01 -1.16375184e+00
8.57028425e-01 -1.77280831e+00 -4.97362942e-01 4.81830984e-01
5.30865252e-01 4.01573956e-01 -3.06517184e-01 1.74790516e-01
-4.03009474e-01 5.34599066e-01 -4.07804269e-03 3.56159389e-01
-4.29128796e-01 -8.50122347e-02 6.77520812e-01 1.33888388e+00
2.97503769e-01 1.16441727e+00 -9.21095967e-01 -6.78368211e-01
2.22336680e-01 4.68208909e-01 -2.64956415e-01 1.21254951e-01
5.73581338e-01 7.61632025e-01 -5.31684041e-01 5.54839134e-01
8.81520152e-01 1.62521914e-01 4.64289039e-01 7.23647997e-02
-5.03156960e-01 -1.32608935e-01 -3.24863911e-01 1.39239001e+00
-1.70073181e-01 7.50877619e-01 2.39611879e-01 -1.35280776e+00
9.87721205e-01 8.11756968e-01 9.22611773e-01 -1.04283524e+00
2.86592960e-01 4.75438058e-01 7.34287500e-01 -1.26200092e+00
-5.11878908e-01 -6.38163030e-01 8.73940945e-01 3.03315282e-01
7.55530447e-02 1.35003150e-01 3.63811105e-01 -4.70419466e-01
1.10830832e+00 2.72284746e-01 7.63462782e-02 -7.45567262e-01
4.75885123e-01 -5.38672447e-01 4.08413351e-01 5.88371098e-01
-3.33954155e-01 9.22988057e-01 1.01979768e+00 -7.31824279e-01
-9.71826732e-01 -8.70214760e-01 -5.40662766e-01 8.60303581e-01
-2.76897013e-01 5.44343054e-01 -3.13277185e-01 -2.98703611e-01
2.49614045e-01 1.85449049e-01 -8.63419592e-01 7.66325602e-03
-1.03678656e+00 -1.26605368e+00 9.80196178e-01 3.70097995e-01
-2.28867605e-01 -7.43768394e-01 -8.27928782e-01 3.65304112e-01
7.36468704e-03 -6.25081182e-01 2.04929411e-01 -1.53233618e-01
-9.54263210e-01 -1.16818774e+00 -1.22528219e+00 -6.22816503e-01
7.85152137e-01 -2.83267051e-01 7.87465632e-01 3.96514326e-01
-6.31864846e-01 1.17061228e-01 -3.76622826e-01 -8.32995772e-02
-8.21369648e-01 -4.66806471e-01 3.47935371e-02 -1.80018097e-01
-1.62070304e-01 -9.24755454e-01 -1.05375504e+00 3.01140010e-01
-6.16845965e-01 -1.37334570e-01 6.17944717e-01 2.56214440e-01
1.15388699e-01 -6.56403899e-01 6.85164094e-01 -8.55111480e-01
-3.46210087e-03 -1.08686411e+00 -3.19414377e-01 4.01285321e-01
-6.04289651e-01 -5.52574754e-01 1.84178337e-01 -4.59576130e-01
-1.06831634e+00 -9.61143732e-01 -1.43922105e-01 -2.53780246e-01
-1.89930886e-01 7.16427028e-01 4.38387156e-01 -5.38752258e-01
6.12892687e-01 1.64101079e-01 2.80942798e-01 -3.88916790e-01
-1.97178021e-01 2.58784890e-01 2.05187231e-01 -8.44006240e-01
2.37218559e-01 3.54526639e-01 6.76361084e-01 -8.67414176e-01
-1.22774756e-02 -2.84510612e-01 -6.66730046e-01 -3.15892071e-01
6.97188973e-01 -5.05948544e-01 -8.00696015e-01 9.60527286e-02
-8.94373715e-01 -7.22529888e-01 2.29053482e-01 1.12209094e+00
-4.96417016e-01 4.19886291e-01 -7.75214612e-01 -6.82850361e-01
-4.83166158e-01 -8.94704223e-01 8.47073272e-02 1.08977459e-01
6.22434281e-02 -1.38762355e+00 4.16354924e-01 4.06744123e-01
9.52636957e-01 1.00469661e+00 1.48120534e+00 -7.36754656e-01
7.45603070e-02 -7.64679983e-02 -4.00600553e-01 6.69153631e-02
1.23536371e-01 3.79721552e-01 -5.28696060e-01 -2.33986676e-01
6.11867756e-02 1.13549747e-01 3.36063772e-01 9.28266644e-01
6.24769509e-01 -1.26354888e-01 -1.03186511e-01 7.48936832e-01
1.45608473e+00 7.74509788e-01 5.08886039e-01 8.79562199e-02
5.17434120e-01 8.51314783e-01 3.35378170e-01 6.40169442e-01
4.08931613e-01 -1.20536545e-02 2.46842161e-01 -2.92028278e-01
-3.16769153e-01 4.40943003e-01 -2.85127670e-01 6.20815396e-01
-9.48551476e-01 -6.06416538e-02 -1.45520639e+00 7.23332167e-01
-1.18835723e+00 -6.60222411e-01 -4.31670278e-01 1.94423473e+00
6.10428214e-01 -4.32336360e-01 -1.59497276e-01 -5.61565042e-01
7.23171592e-01 -2.09489375e-01 -1.14315055e-01 -8.21313143e-01
1.10459700e-01 4.38737899e-01 4.73262995e-01 1.58407316e-01
-4.10997659e-01 2.42174208e-01 6.65396929e+00 1.81862205e-01
-1.42486572e+00 2.67842978e-01 1.10460377e+00 -9.34098884e-02
-3.01357627e-01 -9.02517117e-04 1.58826206e-02 3.67380470e-01
1.17616701e+00 3.90770398e-02 2.63291836e-01 3.07766609e-02
5.94918370e-01 -2.35564500e-01 -8.65913332e-01 2.63832480e-01
-1.03567764e-01 -1.33068371e+00 -5.29670179e-01 -6.06822819e-02
5.45522273e-01 2.08304673e-01 -2.41480842e-01 -3.48570570e-02
-6.12512052e-01 -1.25656199e+00 2.99345046e-01 7.45666862e-01
1.49228835e+00 -2.40604490e-01 6.62786067e-01 -1.22758038e-02
-8.72650206e-01 1.11209333e-01 -2.55764872e-01 -4.97002490e-02
-1.43717259e-01 2.00495586e-01 -1.09691477e+00 4.01343256e-01
6.42127216e-01 2.54198648e-02 -1.29830629e-01 9.28875029e-01
4.28579181e-01 7.37975359e-01 -1.08386382e-01 4.23549175e-01
-6.57545477e-02 -3.52821648e-01 2.98830479e-01 1.51935172e+00
4.92326856e-01 6.01182222e-01 -7.04124629e-01 8.90826643e-01
3.91991735e-01 4.57232028e-01 -2.91176945e-01 -1.22213878e-01
7.95857385e-02 1.40625656e+00 -1.11050677e+00 -2.72609945e-03
-8.79926980e-01 6.67444915e-02 1.35025904e-01 3.35832864e-01
-5.04362106e-01 -1.13949358e-01 2.96787247e-02 5.44388175e-01
-1.36739030e-01 6.47641197e-02 -2.92891741e-01 -8.22535157e-01
-2.89993495e-01 -1.92298025e-01 1.71496987e-01 -4.10833806e-01
-5.49810529e-01 6.22117855e-02 1.60852492e-01 -4.52284575e-01
-9.61604416e-02 -2.66557842e-01 -9.45102990e-01 1.27609134e+00
-1.68302989e+00 -1.08563554e+00 -3.93006951e-02 -1.54402301e-01
-1.64549038e-01 -8.08545351e-02 8.17011893e-01 6.36765480e-01
-4.58983809e-01 5.48710704e-01 -8.49815756e-02 1.88085228e-01
2.52535522e-01 -1.28748620e+00 2.87241694e-02 6.09450936e-01
-9.03655469e-01 7.99282134e-01 3.60364348e-01 -9.05041635e-01
-1.28874063e+00 -8.82525086e-01 7.96452224e-01 -3.52209508e-02
3.92285645e-01 1.66143343e-01 -8.90926600e-01 6.42733634e-01
-1.65672123e-01 8.12249184e-01 1.22591460e+00 -4.41009790e-01
2.37171352e-01 -2.99640242e-02 -1.56032634e+00 1.85758993e-01
5.19968688e-01 3.15477431e-01 -2.48275369e-01 -6.58469722e-02
6.46750182e-02 -6.81685865e-01 -1.75601673e+00 1.00056982e+00
1.25415909e+00 -8.34918261e-01 7.85418451e-01 -8.09919655e-01
6.33064866e-01 2.06850827e-01 2.55813241e-01 -6.99309230e-01
-4.56089824e-01 -6.26607835e-01 -1.22012280e-01 1.20744824e+00
5.28711498e-01 -5.94854474e-01 8.12118530e-01 1.20304251e+00
-2.88744599e-01 -1.02805269e+00 -9.97012615e-01 5.26905917e-02
5.15569866e-01 -1.02293976e-01 3.33600879e-01 1.17416501e+00
3.13458778e-02 -7.61405528e-01 3.10970426e-01 2.09012106e-01
2.41011247e-01 -8.51965621e-02 -8.56779739e-02 -7.81714797e-01
1.39566809e-01 -4.12242800e-01 -3.86411786e-01 1.39545083e-01
-1.26607135e-01 -9.88750160e-01 -2.55386770e-01 -1.59108984e+00
2.48943627e-01 -1.03480542e+00 -4.61749136e-01 4.39074755e-01
1.60089672e-01 2.38364965e-01 -1.06180064e-01 -2.57753581e-01
2.90944558e-02 -8.29325691e-02 1.62138522e+00 4.07635033e-01
-1.95519954e-01 2.91883230e-01 -4.66105819e-01 6.38724685e-01
7.08424330e-01 -5.66539288e-01 -2.07890972e-01 -5.04906297e-01
2.22060218e-01 1.09677362e+00 5.02654128e-02 -5.32345593e-01
-8.64399672e-02 -3.67837280e-01 8.96729827e-01 -9.69687030e-02
-1.34340957e-01 -5.40393412e-01 2.36574575e-01 5.83265305e-01
-1.58460110e-01 -1.03586547e-01 2.87285894e-01 -3.99659753e-01
3.30843300e-01 -3.96716088e-01 1.00517118e+00 -3.82006228e-01
5.38996905e-02 5.67450881e-01 -8.93586636e-01 5.82074784e-02
7.26255357e-01 -3.05891067e-01 -1.01702183e-01 1.28615931e-01
-1.15095556e+00 1.62951559e-01 9.01354328e-02 9.32406113e-02
6.12793028e-01 -7.06159651e-01 -1.23245239e+00 3.18319529e-01
-4.73979741e-01 1.49699743e-03 7.66063035e-01 1.75511992e+00
-1.64562166e+00 3.17702055e-01 -5.25669634e-01 -5.23696184e-01
-9.84461725e-01 7.99119100e-02 8.41167331e-01 2.93525234e-02
-5.24472952e-01 9.17868435e-01 1.86123073e-01 -3.29549223e-01
-4.27639097e-01 -2.46605277e-01 -4.25747395e-01 -2.61541922e-02
5.05594254e-01 1.00449741e+00 2.06643954e-01 -7.65696228e-01
-3.60450536e-01 7.79484630e-01 1.76590294e-01 2.69019663e-01
1.38272023e+00 -5.01865268e-01 -6.72523439e-01 1.24080583e-01
1.31179523e+00 -1.78195331e-02 -9.94978309e-01 1.87198192e-01
-1.46275774e-01 -5.99778652e-01 -1.75477237e-01 -8.82568896e-01
-1.41639400e+00 8.64354670e-01 9.05162930e-01 2.88692892e-01
8.37297261e-01 -4.56788689e-02 6.64461255e-02 -2.97782123e-01
-6.53217435e-02 -3.32402349e-01 -4.62693810e-01 -1.34741947e-01
8.06220829e-01 -1.06460011e+00 3.42639983e-02 -2.23391891e-01
-2.02236801e-01 1.43172109e+00 5.72661936e-01 -9.14584994e-02
5.89874089e-01 5.06737471e-01 4.12742436e-01 -1.72652066e-01
-5.83309293e-01 4.72980917e-01 3.16286117e-01 6.35796785e-01
8.93187702e-01 -5.50748548e-03 -7.99341798e-01 5.40682435e-01
2.28454039e-01 -6.67573810e-02 5.43606043e-01 7.80217350e-01
-3.99793983e-01 -6.03041649e-01 -4.42622267e-02 7.72109270e-01
-1.37557352e+00 -2.72528548e-03 5.28469384e-01 8.65395427e-01
4.23301786e-01 5.97513020e-01 4.64429408e-02 5.77241719e-01
8.51107240e-02 8.98672417e-02 8.23376000e-01 -6.49553716e-01
-5.38013935e-01 3.28476906e-01 -7.87581131e-02 -4.42322135e-01
-4.16416466e-01 -1.07102954e+00 -1.42335272e+00 -1.50119185e-01
-2.66350597e-01 1.43074496e-02 1.08773184e+00 1.16072905e+00
-5.76125570e-02 4.33941275e-01 6.63393915e-01 -6.91924453e-01
3.26614112e-01 -1.09538543e+00 -1.05801129e+00 -2.53398538e-01
5.48459470e-01 -5.03652394e-01 -2.08180904e-01 -9.02870577e-03]
|
[14.032720565795898, -2.374577522277832]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.