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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
31bfd71d-b5d8-4706-aade-90158de90824
|
make-an-animation-large-scale-text
|
2305.09662
| null |
https://arxiv.org/abs/2305.09662v1
|
https://arxiv.org/pdf/2305.09662v1.pdf
|
Make-An-Animation: Large-Scale Text-conditional 3D Human Motion Generation
|
Text-guided human motion generation has drawn significant interest because of its impactful applications spanning animation and robotics. Recently, application of diffusion models for motion generation has enabled improvements in the quality of generated motions. However, existing approaches are limited by their reliance on relatively small-scale motion capture data, leading to poor performance on more diverse, in-the-wild prompts. In this paper, we introduce Make-An-Animation, a text-conditioned human motion generation model which learns more diverse poses and prompts from large-scale image-text datasets, enabling significant improvement in performance over prior works. Make-An-Animation is trained in two stages. First, we train on a curated large-scale dataset of (text, static pseudo-pose) pairs extracted from image-text datasets. Second, we fine-tune on motion capture data, adding additional layers to model the temporal dimension. Unlike prior diffusion models for motion generation, Make-An-Animation uses a U-Net architecture similar to recent text-to-video generation models. Human evaluation of motion realism and alignment with input text shows that our model reaches state-of-the-art performance on text-to-motion generation.
|
['Sonal Gupta', 'Devi Parikh', 'Thomas Hayes', 'Akbar Shah', 'Samaneh Azadi']
|
2023-05-16
| null | null | null | null |
['video-generation', 'motion-synthesis', 'text-to-video-generation']
|
['computer-vision', 'computer-vision', 'natural-language-processing']
|
[ 1.28082618e-01 6.74045458e-02 -4.65646833e-02 -5.77265471e-02
-8.38914037e-01 -4.15946543e-01 1.10360205e+00 -4.27133441e-01
-4.17012244e-01 5.67193329e-01 8.94969046e-01 1.15643302e-02
3.71776313e-01 -6.51602566e-01 -6.66487515e-01 -4.06402260e-01
2.77954619e-02 5.75004697e-01 4.66264963e-01 -4.12390679e-01
1.14622220e-01 1.69865042e-01 -1.23954570e+00 2.53245622e-01
2.46448591e-01 3.43063354e-01 3.52502048e-01 1.32772422e+00
7.73322210e-02 1.02026761e+00 -5.39426804e-01 -1.02745242e-01
2.26293400e-01 -9.17846859e-01 -8.78332615e-01 -1.15064852e-01
6.94233835e-01 -8.09342384e-01 -7.68050730e-01 5.59418678e-01
8.51855636e-01 5.32495499e-01 7.46431470e-01 -1.42944944e+00
-7.77305841e-01 4.39093977e-01 -5.13255715e-01 3.55197452e-02
8.66251886e-01 8.22043717e-01 7.54247308e-01 -8.18572998e-01
1.45327020e+00 1.55375934e+00 6.81421995e-01 1.12198269e+00
-1.29869401e+00 -4.42441374e-01 -4.91774194e-02 -7.01578408e-02
-1.00855923e+00 -3.52302194e-01 5.81296623e-01 -5.82728326e-01
1.24873638e+00 -6.23809658e-02 1.07668638e+00 1.69111800e+00
2.81389207e-01 9.68143523e-01 4.89481628e-01 -1.63422137e-01
1.82122841e-01 -6.36834800e-01 -7.86025822e-01 5.42424560e-01
-1.48170173e-01 3.03395152e-01 -5.84116220e-01 -8.50844458e-02
1.32694137e+00 -3.56188089e-01 -2.85031110e-01 -4.15406823e-01
-1.75867772e+00 9.28049445e-01 3.68640840e-01 1.13408908e-01
-5.12024343e-01 9.15005088e-01 3.15138191e-01 -5.93592525e-02
3.69022310e-01 5.17815769e-01 -3.86282429e-02 -7.16435015e-01
-1.13478625e+00 1.00910258e+00 7.07986474e-01 1.00669420e+00
3.05803329e-01 4.76010621e-01 -3.61181200e-01 3.86645913e-01
2.25755289e-01 6.20430052e-01 8.27258587e-01 -1.36893749e+00
4.43491668e-01 8.04268569e-02 1.97655693e-01 -1.02141070e+00
-4.53264296e-01 2.27249160e-01 -6.57948554e-01 4.13677365e-01
4.54878837e-01 -5.61109364e-01 -1.13414168e+00 1.76076424e+00
3.92740309e-01 4.19278651e-01 -9.23554748e-02 1.15545404e+00
5.81802905e-01 8.75690579e-01 2.87631363e-01 1.95072606e-01
8.09324980e-01 -1.22315764e+00 -5.41484058e-01 -2.35740125e-01
6.62855208e-01 -8.79817486e-01 9.23560381e-01 1.00989588e-01
-1.24971974e+00 -6.72239184e-01 -7.05349624e-01 -2.12216184e-01
1.64961126e-02 -2.84235537e-01 2.48886392e-01 1.18300103e-01
-1.41771102e+00 8.12573910e-01 -9.39346015e-01 -6.48825288e-01
1.29478797e-01 1.61760315e-01 -4.65844363e-01 2.94505861e-02
-1.03779101e+00 8.87449622e-01 2.60941952e-01 -1.13399222e-01
-1.05822659e+00 -5.85404634e-01 -1.09826255e+00 -4.46531653e-01
9.53825414e-02 -1.37299669e+00 1.58397102e+00 -5.65756619e-01
-1.77807784e+00 3.48045081e-01 9.22994539e-02 -4.92014289e-01
1.05097950e+00 -4.42503363e-01 -6.04489259e-02 3.70853871e-01
1.97454065e-01 1.72988415e+00 1.06427884e+00 -1.06366265e+00
-6.05927408e-01 2.35167310e-01 -1.29471630e-01 4.09423858e-01
-6.29044250e-02 -2.15746745e-01 -5.44088721e-01 -1.12415802e+00
-4.14464891e-01 -1.22845256e+00 -5.45967579e-01 1.86224476e-01
-2.67416924e-01 -3.94205451e-02 1.20990825e+00 -5.50308287e-01
1.09123337e+00 -1.58897245e+00 6.21119738e-01 -3.05747956e-01
1.14680648e-01 3.25079143e-01 -5.54403961e-01 7.30201364e-01
3.04435566e-02 1.37117237e-01 3.47657092e-02 -5.48737884e-01
4.80259955e-02 8.86697695e-02 -2.68893689e-01 2.16547966e-01
4.56470430e-01 1.26623392e+00 -1.15956664e+00 -6.21081710e-01
5.76156795e-01 6.56581044e-01 -8.33797872e-01 3.20389450e-01
-6.92744374e-01 8.25609028e-01 -4.10332322e-01 2.33072609e-01
8.13433826e-02 -2.46673748e-01 -2.33993664e-01 -7.56471008e-02
-7.12642167e-03 -1.32537380e-01 -9.49133515e-01 2.32223439e+00
-3.12422305e-01 9.02373910e-01 -3.06844085e-01 -1.64905772e-01
4.82162923e-01 5.33771038e-01 8.21590245e-01 -4.54456896e-01
2.14417249e-01 9.49602723e-02 -7.17831999e-02 -6.34147346e-01
1.13978755e+00 -1.23879053e-02 -1.28227949e-01 6.96313083e-01
-2.46865936e-02 -7.65216172e-01 2.33016670e-01 5.33702195e-01
1.11318707e+00 9.68414128e-01 1.54868960e-02 2.88316071e-01
3.91610339e-02 4.54105347e-01 8.09303764e-03 4.11032706e-01
-1.31713241e-01 1.14181566e+00 1.12705797e-01 -3.53768170e-01
-1.59985542e+00 -7.43609309e-01 5.52687287e-01 8.04275632e-01
6.80428669e-02 -6.18463695e-01 -8.84975493e-01 -3.80007684e-01
-9.26371813e-02 6.06993258e-01 -7.09877431e-01 -1.68631002e-02
-1.06050539e+00 -4.70093638e-01 7.59326339e-01 7.61778414e-01
3.54239970e-01 -1.50496316e+00 -9.89168644e-01 5.36748707e-01
-3.09792221e-01 -1.29787755e+00 -8.93728137e-01 -7.01964200e-01
-8.41958404e-01 -8.16807270e-01 -1.39560652e+00 -6.42599702e-01
2.87448585e-01 9.40124169e-02 1.07615757e+00 -1.38879880e-01
-4.21103209e-01 6.10360861e-01 -4.19079304e-01 -1.12143613e-01
-8.37910831e-01 2.50818804e-02 1.19167827e-02 -3.80869716e-01
-1.71164304e-01 -4.67325479e-01 -8.64183068e-01 1.39144987e-01
-9.80320454e-01 4.22690928e-01 4.48372155e-01 7.11910486e-01
2.76868403e-01 -6.17976964e-01 4.89261776e-01 -3.63579273e-01
8.61590087e-01 -4.60164428e-01 -1.07291482e-01 -3.39082181e-01
-2.53728926e-01 8.82883519e-02 4.53122437e-01 -8.91443729e-01
-9.47594643e-01 2.28358731e-01 -1.59450173e-01 -8.22737694e-01
-8.91875401e-02 1.74881011e-01 4.76509333e-01 1.95362762e-01
8.41608107e-01 5.11757508e-02 -2.10297033e-02 -2.20710761e-03
9.40648317e-01 1.17055722e-01 9.62114513e-01 -5.07438779e-01
9.69994485e-01 4.26447421e-01 5.78263775e-02 -8.57490718e-01
-8.47516209e-03 -1.73819050e-01 -6.10323191e-01 -3.88053119e-01
1.31866848e+00 -9.07459795e-01 -4.78657544e-01 6.76563621e-01
-1.24387813e+00 -9.71716285e-01 -4.34266955e-01 6.55507028e-01
-9.92332697e-01 3.25761706e-01 -8.45301270e-01 -6.49220347e-01
-3.38860095e-01 -1.06189144e+00 1.48979676e+00 4.40540910e-02
-8.56903434e-01 -1.10332143e+00 5.06497025e-01 -5.52892797e-02
4.47068244e-01 6.92588091e-01 3.50121766e-01 -8.18454102e-02
-7.22477019e-01 -1.21518925e-01 1.34203479e-01 -1.51927710e-01
-5.16096503e-02 1.51885360e-01 -5.80226004e-01 -2.30019718e-01
-6.62563920e-01 -6.39871299e-01 6.06605589e-01 5.79388380e-01
5.21238267e-01 -2.68966615e-01 -3.07273924e-01 5.78350663e-01
1.04969299e+00 -9.39090177e-02 6.06180966e-01 2.66400248e-01
1.06780088e+00 4.44534183e-01 6.64798617e-01 5.42655826e-01
4.23470676e-01 8.66073668e-01 3.41056406e-01 -5.66811599e-02
-6.64227128e-01 -8.32237959e-01 4.70100552e-01 6.06720924e-01
-9.49403197e-02 -4.65562910e-01 -7.99905181e-01 8.07330310e-01
-2.11655259e+00 -1.31698453e+00 -1.88309625e-01 1.75549650e+00
7.72499859e-01 6.20391546e-03 6.20268166e-01 -3.56228918e-01
4.95461315e-01 4.81097311e-01 -5.59135377e-01 -8.70500430e-02
1.67824589e-02 8.87175500e-02 2.17924938e-01 6.74196005e-01
-8.79715085e-01 1.31862760e+00 6.28148174e+00 7.44568288e-01
-1.28086817e+00 -1.58455357e-01 3.01508278e-01 -4.05535281e-01
-4.01011497e-01 -1.24025792e-01 -4.63524014e-01 3.04585546e-01
7.87303686e-01 -3.10599744e-01 2.00329423e-01 7.29295373e-01
6.63426161e-01 -1.22046150e-01 -1.25150883e+00 8.81085515e-01
6.53980970e-02 -1.55818427e+00 3.15609276e-01 3.91174071e-02
1.13229561e+00 1.18694238e-01 3.97984646e-02 1.64552286e-01
8.30574572e-01 -1.08671343e+00 1.03129244e+00 4.00566310e-01
1.01247311e+00 -5.16136646e-01 2.83781707e-01 4.10305887e-01
-1.28044820e+00 3.17682832e-01 -7.20360577e-02 -5.47622964e-02
8.80384803e-01 -1.07903317e-01 -9.92951095e-01 4.00421619e-01
4.18049216e-01 7.81184435e-01 -3.54704201e-01 7.41919219e-01
-9.31107700e-02 2.05534816e-01 -1.58135176e-01 -1.77694067e-01
6.25902593e-01 1.54945627e-01 7.58962154e-01 1.39848936e+00
5.51892817e-01 6.57428056e-02 3.82855326e-01 8.53670657e-01
1.12524875e-01 -1.70682967e-01 -9.29737628e-01 -1.88352481e-01
3.23848277e-01 1.13830066e+00 -6.38067067e-01 -5.42993486e-01
-8.82601142e-02 1.33562851e+00 2.74072383e-02 3.93167257e-01
-8.23411703e-01 -6.74821883e-02 5.11801124e-01 1.65899515e-01
2.86844581e-01 -6.73530042e-01 2.18515873e-01 -9.69090104e-01
-4.29085553e-01 -8.14208746e-01 -4.13960852e-02 -1.11232173e+00
-9.36397076e-01 6.68577611e-01 2.40223318e-01 -1.35705841e+00
-1.23515046e+00 -2.57322669e-01 -6.02451980e-01 9.22347188e-01
-7.59602010e-01 -1.40256691e+00 -5.03445864e-01 6.50208414e-01
1.07904661e+00 5.13699912e-02 6.48731768e-01 -9.51085892e-03
-1.65792584e-01 1.68051139e-01 -3.72713685e-01 1.16394058e-01
1.00595820e+00 -1.14476502e+00 1.35320473e+00 8.68209898e-01
3.98010947e-02 3.31314534e-01 9.56329286e-01 -1.04120469e+00
-1.36712670e+00 -1.19031298e+00 5.49523473e-01 -9.23822999e-01
6.93033338e-01 -1.90746486e-02 -5.20365477e-01 7.01872647e-01
4.41062331e-01 -1.51550204e-01 3.56089100e-02 -7.98629045e-01
7.16354325e-02 8.04482162e-01 -8.25044215e-01 1.16279888e+00
1.29148257e+00 -3.64015907e-01 -4.31615829e-01 7.16986358e-02
8.57772887e-01 -7.41773486e-01 -7.70945609e-01 5.77184036e-02
7.99445450e-01 -6.72972381e-01 9.76088464e-01 -6.72426403e-01
1.04509759e+00 -2.89719135e-01 1.85673371e-01 -1.58063424e+00
-3.83064419e-01 -1.16457403e+00 -2.03127533e-01 7.39021420e-01
1.68938875e-01 6.20875508e-02 1.06161261e+00 6.60326004e-01
-6.70968462e-03 -5.19129038e-01 -5.59044719e-01 -5.37334681e-01
2.21158668e-01 -3.55374664e-01 3.41042250e-01 9.29347575e-01
-1.05001703e-02 5.87565482e-01 -8.39288414e-01 -4.38786119e-01
3.63809466e-01 -2.43226230e-01 1.44735718e+00 -7.39262044e-01
-3.88537675e-01 -5.55408418e-01 -3.16320628e-01 -1.44588590e+00
1.57250855e-02 -6.28328025e-01 2.67911136e-01 -1.85613716e+00
-8.01057890e-02 1.42623102e-02 7.57703364e-01 2.85490274e-01
-4.05686677e-01 4.07937825e-01 6.73735321e-01 3.86387795e-01
-4.02388275e-01 7.94721186e-01 1.76090991e+00 9.13883597e-02
-3.54421645e-01 -4.00283635e-01 -5.81795089e-02 7.26615429e-01
5.62760293e-01 -2.68842936e-01 -7.02287436e-01 -6.52265906e-01
3.61221805e-02 4.87675667e-01 5.05631387e-01 -1.20633233e+00
3.27388734e-01 -2.86273420e-01 5.24786711e-01 -6.22307658e-01
6.03017569e-01 -4.11899626e-01 4.10217881e-01 5.87543130e-01
-4.64478731e-01 5.87804914e-01 1.54047266e-01 7.68465042e-01
6.98986575e-02 1.50607839e-01 4.65063453e-01 -3.32411140e-01
-9.06170249e-01 5.59310436e-01 -5.20668149e-01 2.88845748e-01
9.26108420e-01 -3.66623938e-01 -2.68144906e-01 -1.02551699e+00
-6.54692531e-01 1.90659821e-01 7.36648738e-01 8.77421021e-01
7.36487985e-01 -1.54279041e+00 -9.20548916e-01 -1.46735549e-01
-7.41162747e-02 2.83355594e-01 2.19457015e-01 5.53757846e-01
-9.38740015e-01 3.56165826e-01 -3.05042654e-01 -7.63513744e-01
-1.04501545e+00 3.71230006e-01 2.72684242e-03 -2.13382378e-01
-9.66734290e-01 7.26788282e-01 2.48377368e-01 -1.96960136e-01
-7.54785016e-02 -3.11213404e-01 1.01212397e-01 -3.37863445e-01
4.45580184e-01 5.57966411e-01 -6.21470153e-01 -8.94844890e-01
3.71398181e-02 7.60141551e-01 1.48999631e-01 -1.03078794e+00
1.02241552e+00 5.71729010e-03 5.89064598e-01 1.59970999e-01
1.09198534e+00 -8.66215900e-02 -1.83911180e+00 1.73433676e-01
-2.67627597e-01 -3.93854946e-01 -3.31706852e-01 -5.19221485e-01
-8.56978655e-01 8.14988613e-01 1.70728356e-01 -2.89790541e-01
6.07475638e-01 -2.73789942e-01 1.25379550e+00 2.66281456e-01
3.89435798e-01 -1.09703994e+00 7.72009552e-01 5.64937174e-01
1.09378731e+00 -1.08037567e+00 -2.05870882e-01 3.16713899e-02
-9.42904711e-01 1.14528048e+00 8.03299010e-01 -2.54712701e-01
1.75457269e-01 2.71557778e-01 2.59301484e-01 -1.86714083e-02
-9.09472585e-01 4.88568395e-02 2.54072666e-01 8.80033493e-01
5.16651034e-01 -1.23613223e-01 -3.80348936e-02 -1.83293223e-01
-6.06141388e-01 3.66790384e-01 6.87336802e-01 9.88696575e-01
-3.00929636e-01 -1.08042276e+00 -3.81722420e-01 2.22922172e-02
-1.60165831e-01 -3.32840160e-02 -5.62900364e-01 1.09710312e+00
-3.32311541e-01 7.82076538e-01 8.10811576e-03 -5.22236526e-01
2.56857306e-01 -1.79430276e-01 6.73135936e-01 -4.96899188e-01
-4.77834493e-01 3.46685022e-01 2.20595643e-01 -8.35606754e-01
-4.57213938e-01 -7.11466551e-01 -1.49800193e+00 -6.29988909e-01
8.21603537e-02 -2.31852025e-01 5.15603781e-01 5.91016650e-01
4.44062233e-01 6.39782250e-01 1.15812421e-01 -1.72989070e+00
-4.06735033e-01 -1.13409722e+00 -2.32241843e-02 7.43003190e-01
4.10339028e-01 -3.15512687e-01 2.34690104e-02 4.37090009e-01]
|
[7.319580078125, -0.13089828193187714]
|
56029be0-f2c5-4ed7-b059-d8dfe8638644
|
universal-and-independent-multilingual
|
2210.13236
| null |
https://arxiv.org/abs/2210.13236v1
|
https://arxiv.org/pdf/2210.13236v1.pdf
|
Universal and Independent: Multilingual Probing Framework for Exhaustive Model Interpretation and Evaluation
|
Linguistic analysis of language models is one of the ways to explain and describe their reasoning, weaknesses, and limitations. In the probing part of the model interpretability research, studies concern individual languages as well as individual linguistic structures. The question arises: are the detected regularities linguistically coherent, or on the contrary, do they dissonate at the typological scale? Moreover, the majority of studies address the inherent set of languages and linguistic structures, leaving the actual typological diversity knowledge out of scope. In this paper, we present and apply the GUI-assisted framework allowing us to easily probe a massive number of languages for all the morphosyntactic features present in the Universal Dependencies data. We show that reflecting the anglo-centric trend in NLP over the past years, most of the regularities revealed in the mBERT model are typical for the western-European languages. Our framework can be integrated with the existing probing toolboxes, model cards, and leaderboards, allowing practitioners to use and share their standard probing methods to interpret multilingual models. Thus we propose a toolkit to systematize the multilingual flaws in multilingual models, providing a reproducible experimental setup for 104 languages and 80 morphosyntactic features. https://github.com/AIRI-Institute/Probing_framework
|
['Tatiana Shavrina', 'Viktoria Knyazkova', 'Ekaterina Voloshina', 'Vitaly Protasov', 'Oleg Serikov']
|
2022-10-24
| null | null | null | null |
['probing-language-models']
|
['natural-language-processing']
|
[-4.25389349e-01 1.95700377e-01 -5.74975789e-01 -2.83618063e-01
-3.64693165e-01 -1.05200863e+00 6.18321240e-01 2.49962106e-01
-1.06885344e-01 4.31860149e-01 4.93849635e-01 -1.00192118e+00
-2.24192277e-01 -5.05096018e-01 -3.99832964e-01 -2.11375147e-01
2.34890774e-01 7.42093801e-01 7.77706802e-02 -7.07726181e-01
1.66305289e-01 4.62350428e-01 -1.36435223e+00 5.75286508e-01
8.60664308e-01 3.11845690e-01 2.28699639e-01 2.50711590e-01
-5.64801037e-01 6.49771988e-01 -2.78265864e-01 -4.90215331e-01
-9.06917155e-02 -2.87573248e-01 -1.06191719e+00 -1.61519989e-01
3.79688382e-01 3.59607249e-01 3.11511576e-01 1.14484370e+00
8.31346661e-02 -5.64203680e-01 3.75415087e-01 -8.94241035e-01
-5.79872489e-01 1.13247490e+00 -2.53960758e-01 3.67001265e-01
7.20339298e-01 -1.89585518e-02 1.38340080e+00 -9.63834822e-01
1.02037871e+00 1.54872656e+00 5.36528587e-01 4.01399136e-01
-1.36458147e+00 -3.03819150e-01 3.52165878e-01 2.91538596e-01
-1.15250480e+00 -4.04476523e-01 6.79545224e-01 -7.98358560e-01
1.08912349e+00 5.49715996e-01 5.18889487e-01 1.23971188e+00
9.44257230e-02 4.26964164e-01 1.44053018e+00 -1.00807488e+00
-2.44848728e-01 7.58335531e-01 4.22757894e-01 7.33071029e-01
3.70318592e-01 2.34612986e-03 -3.90491724e-01 -5.05091511e-02
3.30247402e-01 -6.05563045e-01 -1.18961148e-01 -7.24451467e-02
-1.15210664e+00 8.37422848e-01 4.75747772e-02 1.19868994e+00
1.21538900e-01 -2.99335271e-01 5.49334466e-01 4.40087259e-01
4.56617504e-01 3.71273756e-01 -8.39506686e-01 -6.88338801e-02
-6.15501821e-01 3.08367997e-01 9.21057284e-01 7.67803192e-01
8.10580492e-01 -1.94819957e-01 3.38471562e-01 9.07766879e-01
7.46826828e-01 2.27582440e-01 6.24268711e-01 -8.68855715e-01
4.62682724e-01 9.44113791e-01 -1.08230235e-02 -9.94797826e-01
-6.77985251e-01 -5.52909076e-01 -2.93504000e-01 3.36736329e-02
7.47319102e-01 1.24882318e-01 -1.87796876e-01 1.86650360e+00
4.48217154e-01 -9.25871849e-01 -2.22526997e-01 6.31877303e-01
5.69893837e-01 3.12922686e-01 2.54880935e-01 -1.62102342e-01
1.81267130e+00 -6.98937356e-01 -7.02999890e-01 -4.95933861e-01
1.18280053e+00 -1.14886701e+00 1.77229488e+00 2.88767695e-01
-1.04793751e+00 -4.88225460e-01 -7.93388665e-01 -2.82732546e-01
-5.67228496e-01 1.36501536e-01 6.82997823e-01 9.05848026e-01
-1.01197982e+00 2.99063742e-01 -7.61046052e-01 -7.27892697e-01
-3.20141047e-01 -9.01059899e-03 -3.26080889e-01 2.52001584e-01
-1.20281160e+00 1.24028122e+00 6.78955853e-01 1.13052569e-01
-9.38321501e-02 -6.25587463e-01 -7.29179442e-01 -1.86934605e-01
2.71074235e-01 -3.58272433e-01 1.02957785e+00 -1.09792483e+00
-1.15113032e+00 1.22300160e+00 -3.08175147e-01 1.35995105e-01
6.99723959e-01 1.35424718e-01 -6.86871648e-01 -4.85999048e-01
2.53984362e-01 3.49658012e-01 -7.02441931e-02 -1.27261961e+00
-5.19910157e-01 -2.75489599e-01 2.44386196e-01 -1.44254684e-01
-2.14648321e-01 5.87998748e-01 -2.07517058e-01 -5.48333466e-01
2.06740096e-01 -8.65428686e-01 4.94333282e-02 -3.88171345e-01
-3.80136073e-01 -2.40550205e-01 3.08992475e-01 -8.74809325e-01
1.65918171e+00 -2.07739806e+00 2.71415979e-01 1.03123568e-01
8.80284142e-03 -2.67025642e-02 1.10733464e-01 9.40904200e-01
-3.12673569e-01 5.43487310e-01 -1.82018310e-01 -2.19366357e-01
5.29080629e-01 5.20147324e-01 -2.69708365e-01 3.94689679e-01
-6.56463131e-02 9.40641046e-01 -7.94614315e-01 -4.21085238e-01
3.51799101e-01 1.90244526e-01 -7.17655122e-01 -2.14024886e-01
-3.34425628e-01 6.43617690e-01 -1.49005651e-01 6.94643497e-01
5.77725291e-01 4.26149815e-02 7.75921404e-01 -2.13389963e-01
-6.10412121e-01 8.75583410e-01 -1.22215962e+00 1.50340927e+00
-6.23528659e-01 5.88172734e-01 -8.32807925e-03 -6.91144347e-01
7.13430285e-01 2.68223494e-01 -1.74203515e-01 -6.55824482e-01
-1.00603038e-02 8.76065731e-01 4.98179108e-01 -5.91202617e-01
3.70830238e-01 -2.54116625e-01 -2.97021180e-01 4.07782614e-01
1.66833296e-01 6.42092973e-02 4.92301911e-01 -1.82203546e-01
5.67076027e-01 2.49017894e-01 8.63119841e-01 -8.50167274e-01
8.29154491e-01 1.81123108e-01 5.11254907e-01 3.33618522e-01
-5.70588233e-03 2.53018051e-01 7.87754476e-01 -5.97678721e-01
-1.04381514e+00 -9.27895486e-01 -6.82154179e-01 1.26051164e+00
-3.71247828e-01 -7.23329782e-01 -8.32829416e-01 -5.37085533e-01
-2.52245873e-01 1.04986501e+00 -5.86714685e-01 3.31089765e-01
-7.71625102e-01 -7.01496363e-01 5.66342413e-01 9.40790474e-02
-1.35725979e-02 -1.23227465e+00 -3.70970130e-01 2.02930301e-01
-3.51697087e-01 -1.05569971e+00 1.10139787e-01 1.10763475e-01
-5.15773833e-01 -9.38653052e-01 1.83636039e-01 -7.32982993e-01
4.05263275e-01 -2.93186188e-01 1.43440306e+00 2.33682021e-01
1.99502066e-01 -6.61008134e-02 -3.24507862e-01 -2.97253966e-01
-1.12144756e+00 4.05421585e-01 -9.95513275e-02 -4.52987909e-01
5.63279152e-01 -6.01408899e-01 1.12323975e-02 3.33558142e-01
-9.29241061e-01 2.46621873e-02 2.01034725e-01 4.16280389e-01
1.58709273e-01 -3.74893874e-01 2.62052983e-01 -1.39717615e+00
7.81670511e-01 -5.61459243e-01 -5.45311332e-01 4.49968487e-01
-6.07241631e-01 8.21109712e-02 6.44227564e-01 -1.94516256e-01
-8.80126655e-01 -4.93293226e-01 -5.72309017e-01 4.38395232e-01
-4.59420472e-01 9.77160752e-01 -4.39228356e-01 2.14815289e-01
8.47759128e-01 -5.07743619e-02 -3.63609612e-01 -8.08051407e-01
5.13925493e-01 5.36071062e-01 1.53406575e-01 -9.65201259e-01
6.89114630e-01 2.19526470e-01 -4.31311667e-01 -7.63660014e-01
-7.41029799e-01 -2.39004537e-01 -9.03692186e-01 -4.99951132e-02
6.08994484e-01 -6.87025368e-01 -4.63297814e-01 4.84601073e-02
-1.51149583e+00 -2.66068876e-01 -3.39553773e-01 3.40455592e-01
-3.02145064e-01 5.49656928e-01 -6.76591039e-01 -5.97565651e-01
3.24081257e-02 -1.27312112e+00 7.77329326e-01 -2.80996114e-01
-9.86855924e-01 -1.64976180e+00 4.75299269e-01 2.71292716e-01
3.22436601e-01 6.90189600e-02 1.68313885e+00 -7.75635242e-01
-1.12769067e-01 1.98843077e-01 5.92806451e-02 1.63985237e-01
1.88644573e-01 2.59306550e-01 -8.01408768e-01 -3.76803130e-02
9.81830880e-02 -5.13912439e-02 2.07330346e-01 1.56112053e-02
4.04709309e-01 -5.05616546e-01 -5.35316914e-02 3.94980401e-01
1.42542267e+00 -2.14775428e-01 3.23626399e-01 6.55419648e-01
6.00744963e-01 1.23453987e+00 1.61677316e-01 9.12106335e-02
6.62827730e-01 8.88632238e-01 2.01076955e-01 5.39538935e-02
1.10050514e-02 -2.52084166e-01 5.64397335e-01 1.38806999e+00
1.20215110e-01 -2.57504564e-02 -1.44034481e+00 5.89145958e-01
-1.80704141e+00 -5.63459396e-01 -7.03068972e-01 1.92775261e+00
8.42761636e-01 2.44019464e-01 1.33891016e-01 2.32449640e-02
4.51936543e-01 2.51402795e-01 2.47364119e-01 -9.05365169e-01
-3.08225125e-01 -3.63073796e-02 1.20787043e-02 1.17213249e+00
-4.88296270e-01 1.21572661e+00 6.47988796e+00 7.27938175e-01
-1.10317028e+00 2.95010626e-01 2.72213876e-01 3.76245111e-01
-9.89300668e-01 3.45714927e-01 -8.95433843e-01 2.88963109e-01
1.16769767e+00 -1.28265843e-01 5.28073907e-01 5.38984358e-01
4.90607440e-01 8.13146010e-02 -1.17844260e+00 4.42866713e-01
-2.18324855e-01 -1.16845143e+00 1.04896605e-01 9.83239263e-02
3.34832013e-01 2.44593710e-01 -2.11694926e-01 1.85013726e-01
2.75326341e-01 -9.02345002e-01 1.37334335e+00 1.50519267e-01
6.00763977e-01 -2.27117583e-01 4.93350029e-01 5.32478333e-01
-1.06117809e+00 -2.27236804e-02 -2.78167099e-01 -3.21946800e-01
2.73531884e-01 4.90475327e-01 -5.16337156e-01 7.11900771e-01
5.06619096e-01 4.75835443e-01 -8.81646812e-01 4.41620231e-01
-6.04647815e-01 8.39069009e-01 -2.92256743e-01 1.73506945e-01
2.39102960e-01 -5.47256231e-01 6.94056511e-01 1.61280072e+00
1.37133211e-01 -6.47107542e-01 -2.50447206e-02 1.08404195e+00
4.08161461e-01 6.30370855e-01 -4.78102595e-01 -4.08346914e-02
3.53990763e-01 1.33308828e+00 -8.49386036e-01 -1.09450199e-01
-5.46394587e-01 4.18390900e-01 5.60664475e-01 1.52800873e-01
-6.84739828e-01 3.82957041e-01 4.80909944e-01 4.22045022e-01
-1.08555451e-01 -4.22102630e-01 -4.86947685e-01 -1.32768035e+00
2.99017966e-01 -1.38695502e+00 3.45288634e-01 -5.35505891e-01
-1.27253044e+00 6.88088417e-01 2.73745030e-01 -7.19952166e-01
-4.62386101e-01 -8.48233581e-01 -4.44257766e-01 9.27917719e-01
-1.36385643e+00 -1.48645890e+00 9.44027826e-02 3.14023107e-01
3.90795648e-01 4.29890491e-02 1.05987489e+00 4.23305959e-01
-5.68900347e-01 3.29551965e-01 -6.39097542e-02 -8.71357322e-03
5.05644500e-01 -1.31056082e+00 6.20159626e-01 7.07619727e-01
4.59908277e-01 1.00897026e+00 1.02690029e+00 -4.28249866e-01
-7.53767967e-01 -5.41686416e-01 1.77893496e+00 -7.16315567e-01
1.38365746e+00 -7.45576859e-01 -1.04926240e+00 1.03882611e+00
4.91094917e-01 -2.03869686e-01 7.48538673e-01 5.73477924e-01
-5.15620112e-01 2.19053552e-01 -7.95228779e-01 7.91388988e-01
1.13974535e+00 -7.58766115e-01 -7.82028317e-01 4.17625934e-01
5.74417651e-01 -2.46321514e-01 -7.65151024e-01 1.67796731e-01
6.53332889e-01 -1.32516658e+00 4.58982706e-01 -6.04121864e-01
2.44249493e-01 -2.36914024e-01 -4.01530027e-01 -1.26338339e+00
-3.31317306e-01 -5.34411728e-01 5.16687393e-01 1.66136968e+00
9.42515373e-01 -1.04796290e+00 1.87775254e-01 3.15193534e-01
-1.90835536e-01 -3.35434496e-01 -1.05637753e+00 -5.19557536e-01
6.79289639e-01 -8.88930619e-01 5.01936197e-01 1.15394902e+00
3.85115445e-01 4.47085381e-01 1.58082962e-01 -1.82896515e-03
1.41349971e-01 -1.28827402e-02 6.24737144e-01 -1.24053907e+00
-5.50723553e-01 -6.68095887e-01 -1.89977840e-01 -5.65122783e-01
4.01068807e-01 -1.20369136e+00 -5.53534091e-01 -1.32553411e+00
-1.04696184e-01 -5.31233311e-01 1.16435044e-01 3.61657828e-01
2.15971187e-01 5.51768951e-02 3.24992269e-01 4.41148847e-01
-2.17864886e-01 -1.19058266e-01 7.99780428e-01 1.52984545e-01
-1.59060374e-01 -3.79536301e-01 -8.89313996e-01 1.21237981e+00
7.74882257e-01 -5.46803892e-01 -6.21155649e-02 -6.85748041e-01
9.16475475e-01 -4.26434189e-01 2.44289100e-01 -7.38538682e-01
-9.04874578e-02 -2.83674985e-01 -1.58293113e-01 -1.48723707e-01
-2.24228308e-01 -9.08825159e-01 5.64081252e-01 5.74629068e-01
-1.45742342e-01 5.14688432e-01 3.72114360e-01 -2.57675678e-01
-2.30088249e-01 -4.60478663e-01 5.58710575e-01 -3.14491808e-01
-5.04782438e-01 -4.08948183e-01 -6.56956851e-01 2.60221511e-01
6.78575099e-01 -7.05244914e-02 -4.81300145e-01 2.36360639e-01
-7.71364629e-01 -1.85929030e-01 6.74823344e-01 5.66844404e-01
-2.44663417e-01 -1.16723931e+00 -7.49469817e-01 2.21686617e-01
3.09189081e-01 -5.77280700e-01 9.04712677e-02 9.18193460e-01
-7.80778289e-01 6.88585877e-01 -1.53054044e-01 -4.90897864e-01
-9.06371117e-01 5.47508240e-01 5.96930027e-01 -5.65108299e-01
-2.68066347e-01 3.40376616e-01 3.14039022e-01 -1.07130444e+00
-3.97534192e-01 -7.31686592e-01 -2.44825214e-01 3.73044521e-01
1.30656138e-01 1.78363219e-01 2.19092831e-01 -1.02526093e+00
-5.96347988e-01 6.32900476e-01 -1.66946538e-02 -2.15343624e-01
1.08463514e+00 -3.77469361e-01 -6.62273943e-01 1.06057525e+00
9.63012815e-01 7.92104363e-01 -4.49481934e-01 -9.47621763e-02
6.46967530e-01 -1.08358242e-01 -5.10403812e-01 -9.69885111e-01
-4.65452254e-01 7.88490593e-01 2.72271782e-01 5.73603630e-01
5.32422781e-01 3.61905575e-01 6.35506436e-02 -6.80517405e-02
3.82515728e-01 -9.34673548e-01 -7.57530034e-01 6.34456575e-01
1.07014740e+00 -8.46671522e-01 -2.66099840e-01 -6.56619966e-01
-4.05761600e-01 1.17305088e+00 2.90209353e-01 6.10327832e-02
5.67683399e-01 4.66365844e-01 5.33024848e-01 -3.61965001e-01
-7.41850257e-01 -1.03546731e-01 2.22205803e-01 4.02832210e-01
1.09421539e+00 3.18518817e-01 -1.08619487e+00 6.04525983e-01
-6.55111492e-01 -4.06842530e-01 4.86228168e-01 4.44605291e-01
-3.41450155e-01 -1.78364360e+00 -4.58674848e-01 -2.05239952e-01
-5.65936506e-01 -3.21048647e-01 -5.73628962e-01 1.24414384e+00
4.32499111e-01 8.66329134e-01 -8.65041316e-02 -1.64370313e-01
4.60056573e-01 3.56200486e-01 3.98392916e-01 -7.11873770e-01
-9.98405516e-01 2.05656454e-01 4.96434689e-01 -3.87333900e-01
-4.86384124e-01 -7.88903534e-01 -9.24776316e-01 -5.16229749e-01
-1.27046958e-01 1.25058681e-01 5.30519426e-01 1.22998226e+00
2.96884868e-02 2.31518686e-01 1.03432976e-01 -5.10770142e-01
-3.44629228e-01 -1.10016358e+00 -5.05000889e-01 2.82674044e-01
-1.03729349e-02 -5.12579203e-01 -5.70260406e-01 -1.03778772e-01]
|
[10.493786811828613, 9.768186569213867]
|
9adca49d-1033-4abf-875e-7852136e946a
|
a-bert-based-distractor-generation-scheme-1
| null | null |
https://aclanthology.org/2020.findings-emnlp.393
|
https://aclanthology.org/2020.findings-emnlp.393.pdf
|
A BERT-based Distractor Generation Scheme with Multi-tasking and Negative Answer Training Strategies.
|
In this paper, we investigate the following two limitations for the existing distractor generation (DG) methods. First, the quality of the existing DG methods are still far from practical use. There are still room for DG quality improvement. Second, the existing DG designs are mainly for single distractor generation. However, for practical MCQ preparation, multiple distractors are desired. Aiming at these goals, in this paper, we present a new distractor generation scheme with multi-tasking and negative answer training strategies for effectively generating \textit{multiple} distractors. The experimental results show that (1) our model advances the state-of-the-art result from 28.65 to 39.81 (BLEU 1 score) and (2) the generated multiple distractors are diverse and shows strong distracting power for multiple choice question.
|
['Yao-Chung Fan', 'Ying-Hong Chan', 'Ho-Lam Chung']
|
2020-11-01
| null | null | null |
findings-of-the-association-for-computational
|
['distractor-generation']
|
['natural-language-processing']
|
[-2.41410643e-01 -1.37101710e-01 -2.59188324e-01 1.32725835e-01
-1.39714992e+00 -7.07109749e-01 4.96762425e-01 -1.97949529e-01
-1.48442939e-01 1.27111018e+00 3.40235829e-01 -5.91267526e-01
1.16019905e-01 -4.07436132e-01 -3.28104377e-01 -6.13362312e-01
6.54375315e-01 6.01655722e-01 5.27142644e-01 -7.17460752e-01
5.59132993e-01 2.81736583e-01 -1.53763258e+00 3.87412012e-01
1.67817605e+00 6.96862817e-01 6.86402500e-01 7.86655068e-01
-5.00042200e-01 9.29116309e-01 -1.24984455e+00 -4.76212978e-01
-1.74875602e-01 -8.39969218e-01 -8.67721498e-01 -3.13975990e-01
3.67347777e-01 -4.48450267e-01 -4.00349014e-02 8.76395762e-01
1.11574268e+00 1.30623639e-01 7.76031077e-01 -1.31815195e+00
-1.18471241e+00 6.33922577e-01 -3.73868704e-01 4.41219807e-01
5.91950595e-01 1.83634087e-01 1.11411214e+00 -1.22826362e+00
3.30939084e-01 1.46826684e+00 7.29180276e-02 8.30411077e-01
-9.48382735e-01 -6.69924378e-01 1.01721175e-01 4.64339942e-01
-1.29965103e+00 -4.84231234e-01 5.85710585e-01 -2.11733505e-01
1.13365698e+00 3.07983428e-01 2.08903596e-01 1.43287778e+00
1.41289160e-01 1.14017487e+00 1.24619961e+00 -6.85245752e-01
1.59782723e-01 2.49913484e-01 1.97350457e-02 1.26212582e-01
4.88518655e-01 2.94229086e-03 -4.34116215e-01 -6.62605930e-03
5.17598748e-01 -4.79194283e-01 -3.51441532e-01 2.19107822e-01
-1.14518821e+00 1.00694370e+00 -1.87545605e-02 4.66116309e-01
-1.93417817e-01 -1.37663424e-01 1.01034328e-01 2.02548921e-01
5.25372803e-01 6.91344202e-01 -3.01218212e-01 -5.29481947e-01
-7.23365366e-01 5.75110316e-01 7.66616583e-01 1.26949918e+00
3.58990192e-01 3.91187787e-01 -9.68913198e-01 1.12167931e+00
1.64305121e-01 1.19102609e+00 6.61583066e-01 -8.56577218e-01
7.42201924e-01 2.01630950e-01 4.26172882e-01 -8.72211277e-01
-2.33407870e-01 -3.86938006e-01 -5.71007252e-01 -1.78452119e-01
4.50552076e-01 -1.49613813e-01 -7.65642583e-01 1.50689769e+00
-2.72126030e-02 -6.05208516e-01 -4.21705134e-02 6.94561005e-01
1.12919247e+00 7.21266568e-01 -1.69299915e-03 -3.74974579e-01
1.17178059e+00 -1.27558327e+00 -1.42634666e+00 -2.69688696e-01
7.12520123e-01 -1.27904630e+00 1.75786805e+00 7.03774095e-01
-1.34351122e+00 -8.06058407e-01 -8.62796366e-01 -3.38882627e-03
-2.30442554e-01 2.74183363e-01 2.56966293e-01 8.41497421e-01
-9.24385130e-01 2.09377240e-03 4.20938246e-02 -4.97006439e-02
9.43037216e-03 5.88189811e-03 3.54153514e-01 -3.69911671e-01
-1.52471936e+00 1.13101673e+00 2.38386616e-01 -2.88462847e-01
-8.10346842e-01 -3.43652934e-01 -4.89183217e-01 1.37013257e-01
5.62855780e-01 -5.88531137e-01 1.55102873e+00 -7.41961300e-01
-1.56966472e+00 3.06341529e-01 -2.81372160e-01 1.94914743e-01
2.58870214e-01 -4.92192239e-01 -7.38967419e-01 3.90889198e-02
4.31148320e-01 5.87263763e-01 7.42337406e-01 -1.54941833e+00
-7.85110235e-01 2.08087657e-02 7.08369911e-02 3.88197839e-01
-2.75515109e-01 2.38905326e-01 -3.35966647e-01 -1.02593851e+00
-1.98798284e-01 -7.54909575e-01 1.62822098e-01 -5.84399045e-01
-5.34787357e-01 -7.44421601e-01 4.68740374e-01 -4.19758767e-01
1.88182735e+00 -1.89015830e+00 -7.51209408e-02 -3.29016894e-01
2.40164146e-01 6.25206113e-01 -3.87951851e-01 5.51253855e-01
7.29083046e-02 2.98505425e-01 4.10096943e-01 -1.00519232e-01
2.60627329e-01 7.59878904e-02 -1.57340005e-01 -2.26506233e-01
2.65043020e-01 1.02173889e+00 -1.06237543e+00 -6.21361017e-01
-1.85730740e-01 -1.42752424e-01 -5.10216415e-01 5.30469060e-01
-3.87057573e-01 2.08022803e-01 -3.88959885e-01 7.26068676e-01
7.87557125e-01 -2.50096232e-01 -1.51422337e-01 1.11515388e-01
1.20967310e-02 7.52837956e-01 -1.12453103e+00 1.60213757e+00
-1.19255967e-01 2.06517309e-01 -3.15145671e-01 -3.87778431e-01
8.70704174e-01 2.65973538e-01 8.77194852e-02 -1.13008595e+00
-1.78966466e-02 5.10097921e-01 3.00188482e-01 -9.56902921e-01
9.35329258e-01 -3.37035507e-01 -7.58372769e-02 4.56722409e-01
4.46964763e-02 -2.17917815e-01 5.25636017e-01 3.82339656e-01
7.07552791e-01 4.45608161e-02 1.48892745e-01 -2.11804479e-01
6.13073587e-01 -1.75677296e-02 2.90442258e-01 7.10806429e-01
-4.13526654e-01 6.91358268e-01 4.06064212e-01 3.13785255e-01
-7.42952526e-01 -1.20498765e+00 2.30813727e-01 1.32269442e+00
6.34178817e-02 -4.02547151e-01 -5.94697952e-01 -9.66729164e-01
-2.31232002e-01 1.29154360e+00 -1.78375393e-01 -1.23340614e-01
-6.96821988e-01 -5.23132265e-01 6.50946736e-01 6.06177509e-01
2.11210281e-01 -1.17282212e+00 1.05338581e-01 3.36572111e-01
-1.08955693e+00 -9.20275569e-01 -9.16476429e-01 -1.48537815e-01
-6.32361650e-01 -7.95234382e-01 -8.87414217e-01 -6.81752443e-01
2.61860043e-01 9.40749526e-01 1.50977826e+00 4.59598340e-02
1.67738318e-01 1.56635180e-01 -7.82562494e-01 -6.95843816e-01
-6.15710139e-01 2.97012627e-01 -9.69749615e-02 -6.62947536e-01
4.70654309e-01 -1.62950799e-01 -6.31656289e-01 3.27498108e-01
-8.68920565e-01 -2.19158813e-01 8.98256004e-01 9.94832098e-01
2.77316034e-01 -3.39868784e-01 1.34778118e+00 -6.11760199e-01
1.39825916e+00 -6.36000872e-01 -1.34182915e-01 3.68715018e-01
-9.49299872e-01 9.83474255e-02 7.45493948e-01 -6.90020740e-01
-1.28657830e+00 -7.81789482e-01 -3.43218833e-01 8.45228806e-02
-1.27768084e-01 3.04368347e-01 -3.02413642e-01 4.18097705e-01
1.03276169e+00 3.81563634e-01 -8.74759135e-05 -3.83899719e-01
5.73266864e-01 8.84777308e-01 4.74162251e-02 -7.93341219e-01
5.89237928e-01 -3.00188124e-01 -3.78230214e-01 -3.14793766e-01
-9.74537969e-01 -4.68030423e-01 -2.25982666e-01 -2.38707364e-01
5.43525457e-01 -7.46700287e-01 -5.99643290e-01 2.47929618e-01
-1.19217563e+00 -8.35337937e-02 -2.40707844e-01 5.28400004e-01
-3.99821132e-01 6.39846027e-01 -5.49924731e-01 -1.01960301e+00
-4.46823120e-01 -1.18072331e+00 8.47434759e-01 2.23293304e-01
-2.45472491e-01 -7.50818789e-01 6.82954192e-02 6.97493196e-01
6.97205186e-01 -4.15065467e-01 1.02950048e+00 -7.99776316e-01
-5.31546712e-01 1.22622102e-01 -2.06374899e-01 4.67177898e-01
1.72855198e-01 -7.54966810e-02 -7.21826375e-01 -2.48808742e-01
6.93544885e-03 -7.10228920e-01 2.92758971e-01 2.70351619e-01
9.33896422e-01 -4.03883696e-01 -1.59657691e-02 -1.59710303e-01
1.18158317e+00 3.92898381e-01 9.10785139e-01 1.80820197e-01
4.99924332e-01 4.34993893e-01 1.02811217e+00 5.46683490e-01
6.78139150e-01 6.52320921e-01 3.17257851e-01 8.44305009e-02
-5.99176347e-01 -3.28470945e-01 6.69303477e-01 1.38877928e+00
1.16914339e-01 -7.01750755e-01 -5.33634901e-01 7.73638248e-01
-1.59643328e+00 -1.03974485e+00 -7.42587388e-01 2.26414990e+00
1.06113613e+00 9.62438509e-02 4.07689154e-01 7.50111267e-02
5.41416526e-01 -4.85415719e-02 -3.25952441e-01 -4.53751773e-01
-3.80857378e-01 2.81951874e-01 -6.51837792e-04 3.91453922e-01
-3.61053020e-01 9.25817192e-01 7.01208305e+00 1.40186584e+00
-7.05540717e-01 1.92583367e-01 3.36032480e-01 -2.25984320e-01
-7.70690203e-01 -2.34759644e-01 -1.13901544e+00 7.49487996e-01
8.75949085e-01 -4.13178295e-01 4.18172568e-01 5.66745639e-01
2.36062035e-01 -3.60154182e-01 -6.83041513e-01 9.69207585e-01
3.78185153e-01 -6.90877378e-01 3.11290056e-01 -3.86535116e-02
7.76277781e-01 -4.91747946e-01 2.70106703e-01 9.64158416e-01
2.92576253e-01 -1.18059027e+00 6.96853340e-01 3.78254801e-01
7.17383385e-01 -9.46563721e-01 6.78386450e-01 7.13111758e-01
-6.70893490e-01 4.39548194e-02 -4.03121680e-01 2.13992558e-02
1.38790175e-01 5.80927789e-01 -7.21354067e-01 7.04668939e-01
3.80553097e-01 -4.05938551e-02 -7.09217131e-01 8.75336230e-01
-5.08434951e-01 8.12186956e-01 5.79645708e-02 -6.66701376e-01
2.35951781e-01 -2.71301605e-02 2.99714983e-01 1.05659723e+00
6.17422283e-01 5.22163622e-02 7.03403875e-02 8.73480022e-01
-4.06386964e-02 4.67554539e-01 -5.10473847e-01 1.37911722e-01
8.10876667e-01 1.12671626e+00 -9.41512585e-02 -3.99053216e-01
-4.34315383e-01 9.86719310e-01 3.75919998e-01 4.05768603e-01
-1.06880534e+00 -4.05245245e-01 3.41286540e-01 8.20319951e-02
2.37929791e-01 -1.32180646e-01 -3.96917373e-01 -1.10811901e+00
9.81154889e-02 -1.42244220e+00 3.19692641e-01 -9.21204507e-01
-1.58479035e+00 4.51550573e-01 2.75333617e-02 -1.36375809e+00
-4.46432829e-01 -4.30999100e-01 -3.61007482e-01 1.16586959e+00
-1.64966679e+00 -8.86622846e-01 -1.62541971e-01 5.00367284e-01
8.27131391e-01 -1.87107190e-01 7.57399797e-01 4.60134327e-01
-3.74390900e-01 9.86519158e-01 1.96471304e-01 -3.35995287e-01
1.26365149e+00 -1.48499548e+00 8.97177607e-02 8.62083554e-01
-1.25327483e-01 6.35056674e-01 6.12744570e-01 -5.85815430e-01
-1.27011228e+00 -6.75148129e-01 1.36444879e+00 -5.96073687e-01
5.12773573e-01 -1.71914190e-01 -8.02107871e-01 8.12899023e-02
5.36701918e-01 -4.00298595e-01 8.82095337e-01 1.24645375e-01
-2.18650401e-01 -9.67394467e-03 -1.14100230e+00 7.39155889e-01
7.97180533e-01 -1.62269965e-01 -6.02783084e-01 4.50155258e-01
7.68385947e-01 -2.86234885e-01 -6.54265523e-01 2.11650431e-01
2.27830246e-01 -9.09666359e-01 8.63626063e-01 -6.23749137e-01
5.37553012e-01 -2.72613078e-01 -1.98940456e-01 -1.66538906e+00
-6.89130723e-01 -6.44801199e-01 -5.24306834e-01 1.41150296e+00
6.04980350e-01 -5.25783062e-01 3.42365742e-01 3.44829440e-01
-4.04837459e-01 -7.40571678e-01 -8.25787842e-01 -1.07638681e+00
6.25841320e-01 -1.11055426e-01 5.70034266e-01 7.94185877e-01
-1.73883364e-02 1.08502734e+00 -4.53408033e-01 -4.38937545e-01
2.11840630e-01 -8.34525824e-02 7.31847465e-01 -8.31460357e-01
-4.10118580e-01 -5.44086814e-01 5.75822353e-01 -1.75908566e+00
-1.94351926e-01 -6.57886982e-01 2.59378612e-01 -1.80301952e+00
3.03795755e-01 -3.15563440e-01 -9.24043059e-02 7.42058083e-02
-1.01159930e+00 -1.84964120e-01 3.40815872e-01 1.80027902e-01
-6.39731586e-01 8.43650043e-01 1.96711707e+00 8.84533301e-02
7.49544278e-02 -5.26760854e-02 -1.24138367e+00 1.47607371e-01
1.02708328e+00 -4.37446475e-01 -7.76252627e-01 -4.46640640e-01
1.67802542e-01 1.54322281e-01 -2.47088447e-01 -6.50643885e-01
-1.45588428e-01 -3.72373164e-01 1.52772695e-01 -1.00413382e+00
1.50263786e-01 -3.25541168e-01 -5.61656773e-01 2.67839968e-01
-1.80293486e-01 5.39966106e-01 1.70035750e-01 4.50376540e-01
-7.51791522e-02 -4.79274184e-01 6.00400150e-01 -1.13042518e-01
-2.56640017e-01 -7.77797401e-02 -7.62711704e-01 7.66882837e-01
7.49203980e-01 -6.64308593e-02 -9.19225156e-01 -6.59148037e-01
-1.95924014e-01 2.83849001e-01 2.06065327e-02 5.42350769e-01
7.04280853e-01 -1.71298134e+00 -1.02696860e+00 -2.30411336e-01
3.98122787e-01 -2.31642246e-01 1.80719480e-01 8.61702800e-01
-9.95043814e-02 7.88019717e-01 -3.01365107e-02 -3.09265345e-01
-1.18788779e+00 7.47784972e-01 -1.02269119e-02 -5.70513546e-01
-3.32721584e-02 7.77751207e-01 1.94059797e-02 -2.36333370e-01
1.47990361e-01 -1.80824474e-01 -4.63956505e-01 1.64145008e-01
4.55180198e-01 6.01157546e-01 1.67597115e-01 -4.90107656e-01
1.80260267e-03 -4.06851061e-02 -2.48974979e-01 -2.94393510e-01
6.77589178e-01 -2.12148190e-01 -4.57174331e-02 4.51272577e-01
7.82901347e-01 3.99778277e-01 -5.95989227e-01 -1.03486381e-01
2.36184578e-02 -6.94787800e-01 -3.66004229e-01 -1.32773888e+00
-6.78878725e-01 9.37078655e-01 2.75277972e-01 3.60018849e-01
1.25957143e+00 -9.21806842e-02 1.11062801e+00 2.93180019e-01
2.99044162e-01 -1.37027133e+00 7.72446930e-01 8.60561192e-01
1.04666746e+00 -1.05953670e+00 -1.90782353e-01 -4.09722835e-01
-9.19707358e-01 8.96839261e-01 1.12994266e+00 3.07740092e-01
2.23970085e-01 -8.25537601e-04 2.77373582e-01 -6.22326769e-02
-9.64272916e-01 -5.46950161e-01 4.72770572e-01 7.05489576e-01
9.20181334e-01 3.68997380e-02 -1.01222491e+00 8.29687119e-01
-7.94157311e-02 -2.10316643e-01 6.47901297e-01 7.98399389e-01
-8.29812765e-01 -1.38050556e+00 -5.46060801e-01 3.86873394e-01
-5.73389232e-01 -4.96305585e-01 -5.29986620e-01 6.47531867e-01
1.70075148e-02 1.59709156e+00 -5.84109366e-01 -2.08218247e-01
6.21804774e-01 1.77030325e-01 6.72236800e-01 -7.48039961e-01
-6.29628718e-01 2.72726029e-01 4.26790923e-01 -1.26440927e-01
-2.15165317e-01 -4.48609740e-01 -9.47322130e-01 -4.50826854e-01
-8.36292088e-01 1.79754511e-01 3.19800526e-01 8.44689429e-01
4.99602199e-01 7.06714928e-01 8.26176345e-01 -1.40317559e-01
-1.06592095e+00 -1.67655325e+00 -5.30430138e-01 4.47830826e-01
-9.61922854e-02 -7.74169326e-01 -3.41540366e-01 -2.12856889e-01]
|
[11.606990814208984, 8.355712890625]
|
9a4b1349-13b6-4988-8e4f-dc5fd9a05146
|
drone-based-rgbt-vehicle-detection-and
|
2003.02437
| null |
https://arxiv.org/abs/2003.02437v2
|
https://arxiv.org/pdf/2003.02437v2.pdf
|
Drone-based RGB-Infrared Cross-Modality Vehicle Detection via Uncertainty-Aware Learning
|
Drone-based vehicle detection aims at finding the vehicle locations and categories in an aerial image. It empowers smart city traffic management and disaster rescue. Researchers have made mount of efforts in this area and achieved considerable progress. Nevertheless, it is still a challenge when the objects are hard to distinguish, especially in low light conditions. To tackle this problem, we construct a large-scale drone-based RGB-Infrared vehicle detection dataset, termed DroneVehicle. Our DroneVehicle collects 28, 439 RGB-Infrared image pairs, covering urban roads, residential areas, parking lots, and other scenarios from day to night. Due to the great gap between RGB and infrared images, cross-modal images provide both effective information and redundant information. To address this dilemma, we further propose an uncertainty-aware cross-modality vehicle detection (UA-CMDet) framework to extract complementary information from cross-modal images, which can significantly improve the detection performance in low light conditions. An uncertainty-aware module (UAM) is designed to quantify the uncertainty weights of each modality, which is calculated by the cross-modal Intersection over Union (IoU) and the RGB illumination value. Furthermore, we design an illumination-aware cross-modal non-maximum suppression algorithm to better integrate the modal-specific information in the inference phase. Extensive experiments on the DroneVehicle dataset demonstrate the flexibility and effectiveness of the proposed method for crossmodality vehicle detection. The dataset can be download from https://github.com/VisDrone/DroneVehicle.
|
['QinGhua Hu', 'Pengfei Zhu', 'Bing Cao', 'Yiming Sun']
|
2020-03-05
| null | null | null | null |
['object-counting']
|
['computer-vision']
|
[-9.94308107e-03 -6.29552722e-01 -9.18988138e-02 -1.37648165e-01
-8.59396398e-01 -5.97979605e-01 5.14176965e-01 -2.67803758e-01
-3.18952769e-01 5.29917538e-01 -2.01260865e-01 -1.80142418e-01
-2.93116540e-01 -1.22755384e+00 -3.43979955e-01 -1.09901452e+00
1.62905648e-01 1.55208245e-01 2.15940773e-01 -5.07438004e-01
-1.98229440e-02 5.29826343e-01 -1.88666904e+00 -2.76289172e-02
1.07457817e+00 1.13836360e+00 4.33400840e-01 3.16117316e-01
2.29875848e-01 2.20424339e-01 -1.43572271e-01 -8.59347433e-02
4.23913300e-01 6.77365512e-02 -2.63657868e-01 7.18302056e-02
2.13073432e-01 -5.18999696e-01 -4.87316102e-01 1.31051207e+00
5.42029440e-01 2.44259447e-01 5.39333105e-01 -1.81780696e+00
-2.36004293e-01 2.11913958e-02 -8.37672830e-01 3.19974571e-01
1.52619079e-01 4.02508646e-01 6.09394073e-01 -9.40436184e-01
2.88288414e-01 1.10382628e+00 4.34865415e-01 1.84551671e-01
-7.48483717e-01 -9.32139814e-01 -1.66244358e-01 4.90850568e-01
-2.05836940e+00 -2.87460029e-01 8.57858062e-01 -3.75104517e-01
3.52848291e-01 4.74986076e-01 5.73872566e-01 5.19986093e-01
-1.71863869e-01 7.85487890e-01 1.02110052e+00 -8.64041075e-02
-1.21074699e-01 1.58017308e-01 -1.74711511e-01 6.73292637e-01
5.13130367e-01 4.26934123e-01 -1.73802316e-01 1.17103823e-01
2.69726843e-01 2.95099646e-01 -5.27915061e-01 -1.19945623e-01
-1.10779834e+00 6.58998847e-01 7.24558830e-01 -1.59589052e-02
-3.41975123e-01 -1.17721654e-01 1.19832180e-01 -2.33278602e-01
1.24445774e-01 -2.11844966e-01 -8.84230584e-02 3.07077110e-01
-7.51368463e-01 3.83915342e-02 -5.45030572e-02 1.02164376e+00
1.12260556e+00 -7.50966519e-02 -2.18845624e-02 6.48580790e-01
5.56385815e-01 1.23807693e+00 -2.31261626e-01 -6.72505915e-01
6.08407676e-01 8.45578909e-01 8.52114707e-02 -1.31966853e+00
-5.02828121e-01 -2.46023923e-01 -9.82408583e-01 2.03077540e-01
-2.28547379e-02 -1.05787300e-01 -7.40804791e-01 1.41434216e+00
6.95578277e-01 2.34042659e-01 7.66599104e-02 1.27778649e+00
1.07842231e+00 7.90419340e-01 9.59930047e-02 -2.47636735e-01
1.39397979e+00 -2.41570145e-01 -5.64546287e-01 -1.99789241e-01
4.41336870e-01 -6.08143032e-01 6.13951445e-01 1.11656532e-01
-3.62424910e-01 -3.74165297e-01 -1.02783024e+00 1.63122624e-01
-5.71326137e-01 4.86491203e-01 4.37838614e-01 5.35419226e-01
-5.37244797e-01 -2.44585603e-01 -4.22939897e-01 -1.27872288e-01
5.06473243e-01 2.59468406e-01 -3.43297303e-01 -5.43982267e-01
-1.49392271e+00 5.47049344e-01 7.29588151e-01 3.94321531e-01
-7.71147609e-01 -4.59493458e-01 -8.57752323e-01 -3.43560636e-01
7.08363652e-01 -3.36944103e-01 4.95781273e-01 -4.12512273e-01
-6.38773322e-01 6.09271228e-01 5.45288138e-02 -3.01771834e-02
2.70740956e-01 3.33683401e-01 -7.25507200e-01 1.93513110e-01
2.89599150e-01 7.07015097e-01 6.43200517e-01 -1.59822655e+00
-1.08708918e+00 -6.45949185e-01 6.59024948e-03 1.71459302e-01
-1.31416097e-01 -1.82817683e-01 -6.88027799e-01 -2.72664368e-01
2.17121348e-01 -9.60541189e-01 1.27644151e-01 -2.36796051e-01
-4.75229651e-01 -2.47752681e-01 1.00837564e+00 -4.86508220e-01
1.27022457e+00 -2.18247104e+00 -2.33894452e-01 4.23304021e-01
7.11632222e-02 3.94336373e-01 -1.42447397e-01 1.63819924e-01
1.98106393e-01 -9.60143059e-02 -4.80819404e-01 1.47007272e-01
-1.85903311e-02 9.04884338e-02 -2.05568939e-01 6.38208687e-01
1.31847396e-01 7.31486082e-01 -1.01716852e+00 -7.72530615e-01
7.29567170e-01 6.68629527e-01 1.85412969e-02 -9.78861451e-02
1.05500035e-01 2.14115337e-01 -6.24919116e-01 1.32142138e+00
1.43540657e+00 3.47073406e-01 -2.91779399e-01 -5.94098628e-01
-4.12489474e-01 -3.40761006e-01 -1.34621584e+00 1.19449115e+00
-3.12830359e-01 5.96881211e-01 1.17015690e-01 -7.40498126e-01
8.83125842e-01 2.19075363e-02 5.45387030e-01 -8.96395564e-01
3.86876822e-01 8.18429589e-02 -4.09204334e-01 -5.38728297e-01
7.04272091e-01 -5.53035662e-02 -1.24099910e-01 2.08268575e-02
-3.70502055e-01 -1.28713772e-01 2.58531988e-01 1.36533380e-01
7.02019036e-01 -8.76927376e-02 8.00783709e-02 4.00767215e-02
8.27774227e-01 3.83031040e-01 7.36800551e-01 2.38531515e-01
-4.93115962e-01 5.42205691e-01 -1.28968939e-01 -1.44537777e-01
-5.71774364e-01 -1.07795978e+00 -3.31708968e-01 6.18778169e-01
8.20607364e-01 -2.65859336e-01 -4.15647268e-01 -4.31605041e-01
1.27147332e-01 6.90803945e-01 -2.60099024e-01 -2.32646599e-01
-1.24030970e-01 -1.10079825e+00 5.19580543e-01 3.11388969e-01
9.85833347e-01 -4.50841814e-01 -4.68316287e-01 -1.25591084e-01
-5.90057194e-01 -1.23100042e+00 -1.66578114e-01 -3.25140297e-01
-4.02282208e-01 -1.27307332e+00 -3.39861065e-01 -4.30253446e-01
5.30532718e-01 1.01117206e+00 7.43352294e-01 1.93968102e-01
-5.58183312e-01 4.78936881e-01 -2.96866119e-01 -3.78436387e-01
1.79559574e-01 -3.69259596e-01 1.30623356e-01 1.20291106e-01
6.73465133e-01 -7.07876086e-02 -7.27679074e-01 6.59953713e-01
-9.47886288e-01 -1.83046743e-01 5.41047513e-01 5.63249946e-01
7.65363276e-01 7.35650599e-01 2.22837552e-01 3.34218070e-02
1.83592856e-01 -7.67321467e-01 -1.07120693e+00 2.37684846e-01
-3.68289083e-01 -4.06340569e-01 2.48635307e-01 1.84895732e-02
-9.73613977e-01 3.55616003e-01 1.38199866e-01 -4.30994421e-01
-2.40912929e-01 3.16766173e-01 -6.09907866e-01 -2.05445722e-01
3.09093088e-01 2.80660063e-01 -2.66067743e-01 7.70505890e-02
2.90161312e-01 1.05017269e+00 7.16697574e-01 -2.05587015e-01
1.16014409e+00 8.09776068e-01 2.05719560e-01 -1.01686108e+00
-2.83628464e-01 -8.17909360e-01 -3.68492275e-01 -7.00845480e-01
7.87797570e-01 -1.31482768e+00 -9.38770831e-01 3.58459175e-01
-9.71637309e-01 1.57253921e-01 2.67396569e-01 5.91654778e-01
-1.08430251e-01 3.59692454e-01 1.43802419e-01 -1.22771454e+00
-1.89457610e-01 -1.21929109e+00 1.22969854e+00 4.35344428e-01
6.55552983e-01 -5.63234508e-01 -2.03773960e-01 5.57585299e-01
-4.41585816e-02 3.31133455e-01 4.18612152e-01 1.31318167e-01
-1.12522149e+00 -2.77986258e-01 -7.51621783e-01 2.80774027e-01
9.40289348e-03 5.61289154e-02 -8.64236116e-01 -6.94409609e-02
-5.17866194e-01 -5.87346070e-02 1.16579461e+00 2.62585014e-01
8.59213889e-01 2.59747654e-01 -5.70902348e-01 6.19314015e-01
1.70833552e+00 1.66916966e-01 8.33552361e-01 3.88389587e-01
9.67746794e-01 5.24910390e-01 1.28384209e+00 6.34646595e-01
6.85070574e-01 7.01930821e-01 1.03322923e+00 -8.87028873e-02
-2.01499444e-02 1.62071157e-02 2.45157763e-01 2.20695466e-01
-1.30137697e-01 -5.40910840e-01 -1.04670477e+00 6.54374897e-01
-1.77166831e+00 -1.08089960e+00 -5.00209570e-01 2.31636214e+00
1.96399808e-01 -1.96249574e-01 4.58314382e-02 2.38437623e-01
1.03786302e+00 1.51318923e-01 -4.22436267e-01 3.36122990e-01
-3.67078662e-01 -4.72036183e-01 9.67100382e-01 3.30581844e-01
-1.44516432e+00 7.37727761e-01 4.70006180e+00 1.21108460e+00
-9.60183084e-01 7.23684430e-02 2.16359466e-01 1.35000810e-01
-4.78647411e-01 -7.43176192e-02 -9.57920611e-01 7.21147001e-01
3.02038401e-01 1.18902072e-01 4.78644609e-01 6.90252781e-01
4.19067144e-01 -5.33975780e-01 -3.19351792e-01 1.31059432e+00
1.67091072e-01 -9.61570859e-01 -1.11673832e-01 2.44516209e-01
5.74305594e-01 1.88663840e-01 8.76823142e-02 -3.76457423e-02
7.18377084e-02 -6.49387360e-01 3.59835595e-01 5.94914258e-01
9.55889881e-01 -1.20087337e+00 8.20679128e-01 4.30804640e-01
-1.74159098e+00 -3.54766846e-01 -4.59705621e-01 3.56631905e-01
1.60634547e-01 7.08990157e-01 -6.23164058e-01 9.65324461e-01
8.21555376e-01 5.93902707e-01 -4.41360861e-01 1.11500227e+00
-2.50901639e-01 1.97901219e-01 -6.20674193e-01 2.73709863e-01
2.02363238e-01 -5.99818110e-01 7.88836420e-01 7.82882035e-01
6.71773195e-01 5.18167555e-01 3.08854312e-01 8.84855032e-01
1.19425587e-01 -1.32067233e-01 -7.97968984e-01 1.44894317e-01
7.21728384e-01 1.55533278e+00 -6.40664816e-01 -5.92555441e-02
-3.43580335e-01 5.50478816e-01 -3.15157384e-01 3.75946194e-01
-1.12774777e+00 -5.36369205e-01 8.62948000e-01 4.39992175e-02
1.59576192e-01 -3.05517316e-01 -5.71205914e-02 -9.53341365e-01
-9.57682878e-02 -4.67735678e-01 3.66683900e-01 -1.05906069e+00
-8.60747218e-01 4.26930726e-01 2.41946295e-01 -1.67181122e+00
2.80088365e-01 -6.64802611e-01 -4.39782888e-01 5.78719795e-01
-1.88856733e+00 -1.35131550e+00 -8.73304248e-01 7.29975879e-01
2.09403500e-01 -1.61353536e-02 1.98155299e-01 7.09854901e-01
-9.76638794e-01 3.24083507e-01 1.91867590e-01 1.51084736e-01
5.36451876e-01 -6.30476177e-01 -3.46573353e-01 1.19930649e+00
-2.56845027e-01 2.10239291e-01 4.66858596e-01 -6.39896035e-01
-1.69841754e+00 -1.49167919e+00 3.72090757e-01 -1.83290035e-01
3.33363593e-01 3.97868380e-02 -3.97547871e-01 2.08660021e-01
-2.28136823e-01 2.76766062e-01 4.04687017e-01 -5.60554564e-01
-5.45228198e-02 -5.60459137e-01 -1.25377786e+00 3.67364585e-01
9.96699929e-01 -6.17000580e-01 -2.58568197e-01 3.36438090e-01
4.32887971e-01 -1.22396410e-01 -6.54682696e-01 7.99077213e-01
4.21225250e-01 -9.57347333e-01 1.28186738e+00 2.95285642e-01
-2.49124821e-02 -1.13582313e+00 -4.94868726e-01 -1.07988048e+00
-1.29621387e-01 -1.12293206e-01 2.93907017e-01 1.53168058e+00
-3.54235582e-02 -6.86705351e-01 3.64911765e-01 4.04499561e-01
-2.31335506e-01 -3.25721502e-01 -9.69200969e-01 -7.45577574e-01
-5.05508721e-01 -7.52715170e-01 1.00334001e+00 5.07677972e-01
-3.88563901e-01 -8.37386176e-02 -3.05016011e-01 8.70297492e-01
9.08769369e-01 2.82155871e-01 6.85309172e-01 -1.38049030e+00
6.09694541e-01 -2.09279329e-01 -6.28621459e-01 -6.81611955e-01
8.08291584e-02 -7.47266233e-01 2.83832163e-01 -1.60615671e+00
1.96027130e-01 -6.35131776e-01 -2.66189367e-01 4.30294454e-01
-1.55908287e-01 7.96032250e-01 3.16176899e-02 2.81719625e-01
-7.32394934e-01 7.06256628e-01 9.85903025e-01 -5.21397471e-01
1.79407578e-02 -1.18860886e-01 -4.59645510e-01 6.17899299e-01
8.26146781e-01 -3.55055034e-01 -4.13997382e-01 -2.67324418e-01
2.44910702e-01 -1.73750758e-01 7.45695055e-01 -1.15870297e+00
2.64492363e-01 -2.77089119e-01 2.92186767e-01 -1.29064882e+00
3.99048030e-01 -1.18192899e+00 2.17291966e-01 2.87271559e-01
5.30602634e-01 -2.14390889e-01 2.20285013e-01 6.13284945e-01
-2.84427315e-01 9.10819694e-02 6.82717204e-01 1.15246244e-01
-1.43346989e+00 5.40737271e-01 -3.28852862e-01 -1.64580420e-01
1.40172684e+00 -3.73674631e-01 -5.87362289e-01 -1.42411038e-01
2.22946312e-02 6.41553283e-01 5.32774448e-01 2.84922183e-01
8.78203750e-01 -1.67074156e+00 -6.36242926e-01 3.38715404e-01
6.37128413e-01 1.82054602e-04 7.32277632e-01 1.00042486e+00
-3.33129019e-01 4.11838859e-01 -1.54832989e-01 -8.34622145e-01
-1.39134753e+00 6.02841437e-01 1.80087894e-01 4.41100597e-01
-4.02199738e-02 5.67711055e-01 8.06543455e-02 -4.37166244e-01
-2.10387915e-01 -1.35332914e-02 -4.34182912e-01 3.94382536e-01
6.18301809e-01 8.34550619e-01 2.97675226e-02 -1.41272628e+00
-8.22490692e-01 9.93398070e-01 6.36965930e-01 2.05036387e-01
9.43531394e-01 -6.20114326e-01 -1.60709843e-01 -1.46034732e-01
1.05741453e+00 -1.13947488e-01 -1.02240169e+00 -5.25219552e-02
-5.55248499e-01 -7.14977503e-01 6.77950740e-01 -5.13117075e-01
-1.25534606e+00 7.65437305e-01 9.73237932e-01 5.59293553e-02
1.54280019e+00 -1.44065320e-01 9.27886069e-01 2.81105131e-01
5.50792694e-01 -1.15483105e+00 -4.17582482e-01 2.30529711e-01
5.75874031e-01 -1.73153377e+00 2.63601482e-01 -6.77820027e-01
-5.47940254e-01 1.01950026e+00 5.96932888e-01 2.72632658e-01
5.42815685e-01 8.71599466e-02 -8.04530680e-02 -2.80990422e-01
-8.01459849e-02 -1.00840294e+00 3.22050750e-01 7.48784661e-01
-3.22925240e-01 3.42339873e-01 -4.92959209e-02 3.16006690e-01
2.85531163e-01 -1.45030737e-01 2.09190995e-01 7.71986187e-01
-7.11542070e-01 -6.97704315e-01 -8.75110090e-01 1.78172991e-01
2.87707031e-01 -1.58505235e-02 -1.70126930e-02 8.45847368e-01
7.04254746e-01 1.49606144e+00 5.61103933e-02 -9.22502935e-01
1.69095322e-01 -5.23197174e-01 1.57100514e-01 -5.54113416e-03
2.69470245e-01 -4.83623818e-02 -1.05258889e-01 -4.92209941e-01
-5.58631539e-01 -5.41453242e-01 -1.42347229e+00 -4.57452804e-01
-8.13907385e-01 -1.69730559e-02 8.18983138e-01 8.32321823e-01
3.62720668e-01 2.86125481e-01 1.03246570e+00 -8.85374606e-01
-2.21377201e-02 -3.90191823e-01 -7.82250822e-01 3.82450148e-02
3.51528436e-01 -1.18268037e+00 -5.58749437e-01 -3.53785276e-01]
|
[8.501371383666992, -1.7676247358322144]
|
7d8ff90e-e8ca-43a0-9318-6b33e745ef66
|
neural-shape-compiler-a-unified-framework-for
|
2212.12952
| null |
https://arxiv.org/abs/2212.12952v2
|
https://arxiv.org/pdf/2212.12952v2.pdf
|
Neural Shape Compiler: A Unified Framework for Transforming between Text, Point Cloud, and Program
|
3D shapes have complementary abstractions from low-level geometry to part-based hierarchies to languages, which convey different levels of information. This paper presents a unified framework to translate between pairs of shape abstractions: $\textit{Text}$ $\Longleftrightarrow$ $\textit{Point Cloud}$ $\Longleftrightarrow$ $\textit{Program}$. We propose $\textbf{Neural Shape Compiler}$ to model the abstraction transformation as a conditional generation process. It converts 3D shapes of three abstract types into unified discrete shape code, transforms each shape code into code of other abstract types through the proposed $\textit{ShapeCode Transformer}$, and decodes them to output the target shape abstraction. Point Cloud code is obtained in a class-agnostic way by the proposed $\textit{Point}$VQVAE. On Text2Shape, ShapeGlot, ABO, Genre, and Program Synthetic datasets, Neural Shape Compiler shows strengths in $\textit{Text}$ $\Longrightarrow$ $\textit{Point Cloud}$, $\textit{Point Cloud}$ $\Longrightarrow$ $\textit{Text}$, $\textit{Point Cloud}$ $\Longrightarrow$ $\textit{Program}$, and Point Cloud Completion tasks. Additionally, Neural Shape Compiler benefits from jointly training on all heterogeneous data and tasks.
|
['Justin Johnson', 'Honglak Lee', 'Tiange Luo']
|
2022-12-25
| null | null | null | null |
['point-cloud-completion']
|
['computer-vision']
|
[ 4.88565527e-02 1.62902832e-01 1.97743312e-01 -5.42163908e-01
-9.26625490e-01 -1.03229427e+00 6.01482213e-01 1.66704759e-01
2.29223311e-01 2.30613485e-01 -5.67992330e-01 -1.02297759e+00
-8.12425390e-02 -1.43084311e+00 -1.15726161e+00 -3.43443036e-01
-1.87224343e-01 7.53316879e-01 1.11838393e-01 -3.60811472e-01
2.57543743e-01 8.78502905e-01 -1.97636056e+00 4.96716887e-01
9.35782313e-01 1.55199671e+00 6.28406182e-02 5.90619743e-01
-9.64159071e-01 2.97835529e-01 -3.89896691e-01 -7.39021659e-01
4.74866986e-01 1.91577777e-01 -5.50873101e-01 -1.96688384e-01
6.48656905e-01 -1.18879072e-01 1.97041780e-01 1.03126597e+00
3.32849324e-01 -1.04992919e-01 8.68750870e-01 -1.15019011e+00
-6.63223922e-01 3.66322994e-01 -6.41723752e-01 -5.32765329e-01
1.15739979e-01 5.52789450e-01 1.00619411e+00 -1.54038477e+00
5.78429341e-01 1.16843963e+00 9.50578153e-01 1.21691957e-01
-1.27286339e+00 -9.37435329e-01 1.42940268e-01 -9.16482806e-01
-1.52028716e+00 3.32754888e-02 7.34960496e-01 -8.19958568e-01
1.41389239e+00 4.72618937e-01 5.29911995e-01 4.03364062e-01
2.79704053e-02 6.74368620e-01 3.56308281e-01 4.25328948e-02
1.69281289e-01 -4.25227612e-01 8.07737038e-02 8.80858541e-01
-2.12280452e-03 4.16980922e-01 3.46956998e-02 -2.92019397e-01
1.18693590e+00 1.10024080e-01 2.45286450e-01 -5.45096278e-01
-9.59998071e-01 7.25179553e-01 6.74568057e-01 4.51003686e-02
8.37060437e-02 6.67917013e-01 3.56754601e-01 2.21681818e-01
5.28911829e-01 1.47121191e-01 -6.28105819e-01 -1.02778226e-01
-8.15575004e-01 5.81592977e-01 4.12287951e-01 2.06988621e+00
1.34840429e+00 6.57567680e-01 1.31787866e-01 8.85873675e-01
2.03327641e-01 1.41850734e+00 -3.03544223e-01 -9.99149561e-01
1.02361643e+00 1.14281809e+00 -1.67006075e-01 -7.58315086e-01
-3.97540689e-01 -2.64249176e-01 -1.00198627e+00 8.16011012e-01
-5.30543551e-02 -1.27615463e-02 -1.05593097e+00 1.26794147e+00
1.80925965e-01 -4.89007682e-01 -1.59063339e-01 4.74153221e-01
1.15995932e+00 7.63309300e-01 1.25770137e-01 4.24979925e-01
1.23625636e+00 -4.77068633e-01 3.44458550e-01 -1.74169745e-02
1.01786053e+00 -8.64866316e-01 1.59867847e+00 1.92477882e-01
-1.54257572e+00 -8.34137678e-01 -9.13621843e-01 -4.45458889e-01
-5.16595066e-01 2.42051825e-01 7.11631835e-01 7.00409532e-01
-9.49215710e-01 3.94920856e-01 -6.40801132e-01 1.88927203e-01
5.99408627e-01 5.12429297e-01 3.37811485e-02 1.06914319e-01
-4.48210448e-01 -2.40900982e-02 9.22491103e-02 -1.94537044e-01
-7.20104635e-01 -1.16836619e+00 -1.07555914e+00 1.90957844e-01
2.98622549e-01 -7.78144181e-01 1.14204812e+00 -3.02379847e-01
-1.03025031e+00 1.06536520e+00 -4.93258983e-02 -1.72611419e-02
2.85937250e-01 1.71628356e-01 -1.34172976e-01 -1.61607563e-01
1.88703239e-01 8.67062867e-01 9.25089300e-01 -1.37466478e+00
-5.91923594e-01 -6.24592662e-01 1.52290300e-01 -7.66082481e-02
4.23316032e-01 -1.99269652e-01 -4.25548255e-01 -9.25105989e-01
4.00277913e-01 -8.79843891e-01 9.46149975e-02 1.85249910e-01
-2.87344843e-01 -4.22203451e-01 8.59036386e-01 8.26294441e-03
9.57862437e-01 -2.23616791e+00 1.11381099e-01 3.33734393e-01
4.05418903e-01 6.48505092e-02 1.16876960e-01 4.11993116e-01
-1.81808189e-01 5.00890970e-01 -7.89509714e-01 -1.86959669e-01
3.47176641e-01 1.76696435e-01 -5.53857148e-01 -1.59496874e-01
2.15613201e-01 9.49099898e-01 -4.37781304e-01 -2.49980554e-01
2.58881658e-01 2.86451280e-01 -9.21680152e-01 -2.18666662e-02
-9.42027748e-01 4.35490534e-02 -8.13231647e-01 8.92183959e-01
1.08016372e+00 -1.90948829e-01 -4.85005170e-01 -4.96055298e-02
-4.51990664e-01 2.17314027e-02 -1.24338937e+00 1.99364924e+00
-3.97419214e-01 -1.82323828e-01 5.04899085e-01 -8.08376729e-01
1.45587134e+00 -2.14755014e-02 6.87229514e-01 -5.32851219e-01
1.49470996e-02 4.91287529e-01 -4.35305685e-01 7.39253238e-02
7.11474121e-01 -3.03814620e-01 -5.86185098e-01 7.06133783e-01
-2.74983764e-01 -1.63679373e+00 -2.96365786e-02 1.02559522e-01
6.97483301e-01 9.01683927e-01 -1.00350931e-01 -5.01627147e-01
4.93197113e-01 1.67416006e-01 4.73875590e-02 6.23919189e-01
5.87637424e-01 7.08200574e-01 6.76228702e-01 -9.03398156e-01
-1.32918882e+00 -1.59836221e+00 -2.12404475e-01 1.34669089e+00
-5.29614948e-02 -3.97846669e-01 -8.36458147e-01 -3.72672945e-01
3.01579356e-01 8.54779601e-01 -3.28702301e-01 1.84216976e-01
-7.99755156e-01 -2.70943671e-01 7.88319051e-01 8.52435470e-01
4.97119695e-01 -1.28714943e+00 -8.10783088e-01 -9.33154449e-02
3.38357836e-01 -6.84364915e-01 -6.13332570e-01 2.60682762e-01
-1.13809848e+00 -7.13144779e-01 -5.22606134e-01 -5.83924174e-01
9.58100379e-01 -1.98341280e-04 1.48381960e+00 5.08130431e-01
-3.09471101e-01 2.60291427e-01 -1.72347695e-01 -8.35216522e-01
-3.05662751e-01 1.78240106e-01 -4.47651923e-01 -3.98194075e-01
2.28386313e-01 -8.46696317e-01 -5.11514604e-01 2.81834543e-01
-1.38833892e+00 2.41492972e-01 2.93601573e-01 4.78853613e-01
1.27270710e+00 -3.38945925e-01 1.75717904e-03 -8.18770528e-01
2.83511549e-01 -4.05792780e-02 -1.02960312e+00 -1.72561869e-01
-2.90181845e-01 2.57037193e-01 9.69231009e-01 -4.23335508e-02
-7.17421174e-01 -1.16006069e-01 -4.46066409e-01 -8.02677333e-01
-2.64744431e-01 3.58818948e-01 -1.82360217e-01 4.05288100e-01
8.78975928e-01 5.52753210e-01 -2.32657015e-01 -5.12093484e-01
6.95177913e-01 2.33759478e-01 5.05789220e-01 -1.50165045e+00
9.40522373e-01 5.51502407e-01 2.11956695e-01 -6.43644631e-01
-1.51432440e-01 2.55553901e-01 -8.23930383e-01 1.65718347e-01
9.48657990e-01 -7.35581040e-01 -8.59547257e-01 2.32393414e-01
-1.12511647e+00 -5.07635057e-01 -6.45240903e-01 -2.36589298e-01
-9.83009696e-01 5.52143455e-02 -1.12691633e-01 -5.72643697e-01
-6.26572490e-01 -1.63953590e+00 1.92064011e+00 -2.93768793e-01
-1.56825751e-01 -5.24020791e-01 -6.73332572e-01 9.90496129e-02
2.63239980e-01 4.62944657e-01 1.59936464e+00 -1.73889264e-01
-1.11761820e+00 -3.57820898e-01 -6.96226180e-01 2.39473432e-01
-1.11289531e-01 1.98234692e-01 -6.78019226e-01 -2.43050054e-01
-2.23944128e-01 -1.41995385e-01 2.27782845e-01 1.87674746e-01
1.76681566e+00 -1.22763395e-01 -3.23855996e-01 1.19495785e+00
1.48792315e+00 4.65723276e-01 6.23028398e-01 -2.96485096e-01
8.62267673e-01 2.54681200e-01 -6.81990087e-02 4.24105823e-01
5.42234719e-01 5.10326505e-01 7.43016958e-01 -1.00133978e-01
-1.65638104e-01 -2.48191297e-01 -4.71707471e-02 5.85506022e-01
-3.42499375e-01 3.80636632e-01 -1.27608299e+00 1.93866313e-01
-1.31693828e+00 -6.82180941e-01 -3.36479247e-01 2.14935040e+00
5.59016168e-01 2.36050934e-01 -1.76814459e-02 -3.63117233e-02
1.84001788e-01 8.74432772e-02 -6.24778390e-01 -9.33974385e-01
4.17967848e-02 8.87177944e-01 2.56910086e-01 5.03909230e-01
-7.09581375e-01 1.02401698e+00 3.16291809e+00 8.11467767e-01
-1.25050056e+00 -2.26778477e-01 6.08336449e-01 -1.37934864e-01
-8.03726673e-01 9.09861699e-02 -6.94351673e-01 4.12391752e-01
3.00338626e-01 -6.80414587e-02 4.39739347e-01 1.19202900e+00
-2.76095361e-01 2.07482845e-01 -1.30214202e+00 1.19985509e+00
-9.44556519e-02 -1.71084499e+00 3.65239769e-01 2.18155384e-02
6.19976878e-01 6.53984770e-02 3.37125361e-01 6.51478946e-01
7.08855391e-01 -1.09701777e+00 1.27421105e+00 2.58230150e-01
1.78265953e+00 -6.54276013e-01 -1.08791418e-01 3.38086218e-01
-1.98239326e+00 -3.90519109e-03 -1.87548637e-01 -4.06723563e-03
-1.65964767e-01 3.44467312e-01 -6.02225602e-01 1.02138686e+00
9.20952499e-01 2.97743291e-01 -3.35273802e-01 5.50366156e-02
1.45973656e-02 -3.25485552e-03 -4.38720196e-01 -3.05312742e-02
2.19253749e-01 -3.24300557e-01 4.36882913e-01 9.24677432e-01
7.22836614e-01 5.26124358e-01 3.59393656e-01 1.55170310e+00
-1.15084015e-01 1.03709161e-01 -9.01769698e-01 9.96900424e-02
2.78013319e-01 6.14950061e-01 -8.98302138e-01 -4.06646878e-01
-3.58890533e-01 4.00823236e-01 1.81981236e-01 5.29683352e-01
-8.50023091e-01 -9.73252833e-01 6.32407129e-01 6.97366238e-01
5.70666254e-01 -5.86819589e-01 -1.21911275e+00 -9.62338269e-01
1.64500475e-01 -5.75676739e-01 2.76182055e-01 -1.31428623e+00
-7.38032043e-01 8.26482713e-01 4.04503614e-01 -1.43255389e+00
-1.69434994e-01 -7.67678320e-01 -2.75425464e-01 1.13114631e+00
-9.21777070e-01 -1.36792505e+00 -2.01573491e-01 5.65095127e-01
5.50597072e-01 -4.15416211e-01 8.88098240e-01 1.40985772e-01
1.73655942e-01 6.76397502e-01 -3.96460779e-02 2.10919261e-01
-1.55805841e-01 -1.07378268e+00 8.19104373e-01 2.69475579e-01
7.29992520e-04 7.81155229e-01 2.43325725e-01 -8.16846967e-01
-1.72058344e+00 -1.11354399e+00 3.15378278e-01 -7.13618577e-01
2.90705532e-01 -6.85313165e-01 -7.32028306e-01 9.35419559e-01
-2.58070469e-01 9.51037779e-02 1.69227779e-01 -2.30240420e-01
-6.77448153e-01 -2.26253167e-01 -1.34488130e+00 6.47754431e-01
1.35055435e+00 -5.87065637e-01 -4.46996152e-01 1.39503986e-01
1.03795552e+00 -8.86607230e-01 -1.27388179e+00 7.63523757e-01
4.05280024e-01 -1.04739213e+00 1.22870386e+00 -2.25165308e-01
7.02596664e-01 -2.05868870e-01 -5.27812839e-01 -9.05909717e-01
-8.38131979e-02 -5.54264724e-01 3.40302944e-01 7.18126059e-01
5.98592281e-01 -5.67205667e-01 9.17653203e-01 6.29439354e-01
-1.00650322e+00 -1.00042796e+00 -9.06450212e-01 -5.83733261e-01
1.03407812e+00 -1.06052291e+00 1.51649034e+00 6.29804492e-01
-4.98571962e-01 -1.29723579e-01 5.74835896e-01 -1.09176129e-01
5.79991043e-01 6.94103301e-01 1.17574298e+00 -1.17156267e+00
-5.46022132e-02 -1.02374506e+00 -1.81464888e-02 -1.21378553e+00
-2.44004130e-01 -1.52177334e+00 -3.79067242e-01 -1.50616801e+00
-3.68899405e-01 -1.16375959e+00 6.37702107e-01 7.05194056e-01
3.88267785e-01 8.17342289e-03 2.53297418e-01 3.18362601e-02
2.39277989e-01 3.64928573e-01 1.48085225e+00 -4.69152927e-01
-6.64851442e-02 -1.30723745e-01 -5.03262937e-01 8.64485145e-01
3.14789683e-01 -3.39204878e-01 -6.44621491e-01 -1.22914171e+00
6.38593614e-01 4.96235281e-01 4.38394576e-01 -7.97664285e-01
-6.34985790e-02 -2.47071803e-01 2.66263187e-01 -1.26163220e+00
6.36163354e-01 -8.31503689e-01 3.70539665e-01 1.08124554e-01
8.85479078e-02 5.91631651e-01 6.12020612e-01 -7.72694275e-02
1.69852674e-01 -1.60770938e-01 6.40317678e-01 -5.43332636e-01
-5.16576925e-03 7.18712628e-01 3.88566017e-01 8.15747902e-02
9.18837309e-01 -7.62710512e-01 -1.05030529e-01 6.63871691e-02
-9.04963017e-01 1.01661362e-01 6.77785397e-01 4.21925873e-01
6.93044960e-01 -1.55164635e+00 -5.24750888e-01 9.11541224e-01
2.44670987e-01 1.19076753e+00 3.46158803e-01 2.65137196e-01
-8.41195822e-01 4.76055503e-01 1.72440693e-01 -1.14361084e+00
-6.44365609e-01 5.12818635e-01 3.40890914e-01 1.19406335e-01
-6.18985474e-01 1.13660204e+00 5.14879525e-01 -1.20311439e+00
-2.89841704e-02 -1.15340281e+00 7.49142289e-01 -3.19712579e-01
-2.88868323e-02 2.25362077e-01 3.47112060e-01 -6.06847167e-01
-1.66891925e-02 1.14149070e+00 5.38315296e-01 6.14422783e-02
1.21715999e+00 5.22144318e-01 -4.69627619e-01 9.70064849e-02
1.35944414e+00 -7.71272182e-02 -1.11521983e+00 1.18762396e-01
-2.18857870e-01 -3.24833274e-01 -8.71575117e-01 -7.10427225e-01
-9.70081210e-01 1.36142671e+00 1.11450717e-01 3.80247831e-01
9.09665525e-01 1.32147849e-01 7.08224177e-01 2.41167411e-01
1.01034415e+00 -8.22966993e-01 -1.50223710e-02 8.34348500e-01
1.32775033e+00 -7.47825444e-01 2.45780032e-02 -4.56642538e-01
-2.92849809e-01 1.21662760e+00 5.61956584e-01 -2.10575625e-01
1.05043471e+00 5.70591748e-01 -2.63809055e-01 -6.82743013e-01
-4.42706376e-01 1.27384350e-01 2.64458954e-01 5.12110293e-01
3.46651077e-01 2.09562346e-01 -6.54120073e-02 8.80982637e-01
-8.31373930e-01 -1.29532322e-01 1.36195973e-01 9.08186376e-01
-3.46328080e-01 -1.11519706e+00 -4.72761691e-01 6.19641602e-01
-3.57290753e-03 -2.21690536e-01 -1.83013409e-01 8.70201349e-01
5.37542224e-01 1.48390040e-01 3.70559275e-01 -2.20616996e-01
7.13235021e-01 3.28063339e-01 4.01589543e-01 -7.87806392e-01
-8.81195486e-01 2.60089368e-01 -2.65395254e-01 -3.39204103e-01
3.11190724e-01 -5.21842897e-01 -2.02119756e+00 -3.23666453e-01
4.96879488e-01 1.67838395e-01 6.30429089e-01 3.75065863e-01
6.74819589e-01 4.19670314e-01 3.86365056e-01 -1.12593937e+00
-4.66811329e-01 -5.23443758e-01 -1.81191817e-01 6.28410757e-01
-9.27385464e-02 -5.71633160e-01 8.33320525e-03 2.43200719e-01]
|
[8.690433502197266, -3.636439561843872]
|
6658c844-0626-4051-a31e-20b8a3da2597
|
matching-the-blanks-distributional-similarity
|
1906.03158
| null |
https://arxiv.org/abs/1906.03158v1
|
https://arxiv.org/pdf/1906.03158v1.pdf
|
Matching the Blanks: Distributional Similarity for Relation Learning
|
General purpose relation extractors, which can model arbitrary relations, are a core aspiration in information extraction. Efforts have been made to build general purpose extractors that represent relations with their surface forms, or which jointly embed surface forms with relations from an existing knowledge graph. However, both of these approaches are limited in their ability to generalize. In this paper, we build on extensions of Harris' distributional hypothesis to relations, as well as recent advances in learning text representations (specifically, BERT), to build task agnostic relation representations solely from entity-linked text. We show that these representations significantly outperform previous work on exemplar based relation extraction (FewRel) even without using any of that task's training data. We also show that models initialized with our task agnostic representations, and then tuned on supervised relation extraction datasets, significantly outperform the previous methods on SemEval 2010 Task 8, KBP37, and TACRED.
|
['Tom Kwiatkowski', 'Livio Baldini Soares', 'Nicholas FitzGerald', 'Jeffrey Ling']
|
2019-06-07
|
matching-the-blanks-distributional-similarity-1
|
https://aclanthology.org/P19-1279
|
https://aclanthology.org/P19-1279.pdf
|
acl-2019-7
|
['few-shot-relation-classification', 'few-shot-relation-classification']
|
['methodology', 'natural-language-processing']
|
[ 2.19153404e-01 1.01598108e+00 -4.50435609e-01 -3.62911463e-01
-6.62255168e-01 -5.91941953e-01 1.13585222e+00 6.40131235e-01
-9.59331691e-02 1.06518495e+00 5.77235818e-01 -6.45818889e-01
-4.31987911e-01 -1.17017138e+00 -4.96227115e-01 -5.77054471e-02
-2.46784776e-01 1.07169580e+00 2.03998387e-01 -7.22868085e-01
-1.37710959e-01 4.71901059e-01 -1.01381826e+00 2.37168133e-01
6.49044394e-01 8.43527675e-01 -5.51726818e-01 6.19967639e-01
-1.81286573e-01 1.33910346e+00 -5.71193814e-01 -9.03687358e-01
-1.14221625e-01 -1.31780922e-01 -1.60825026e+00 -2.96099752e-01
2.23814324e-01 -1.02709001e-02 -8.28855932e-01 6.08823895e-01
1.18545733e-01 3.09996426e-01 1.24110329e+00 -1.24120343e+00
-1.12864578e+00 1.20935726e+00 -4.14597452e-01 4.76424962e-01
5.73845983e-01 -5.98111808e-01 1.60253274e+00 -7.58457839e-01
9.19731498e-01 1.23074090e+00 8.18716049e-01 3.27629834e-01
-1.45382154e+00 -5.59928358e-01 -2.98327152e-02 2.86241800e-01
-1.39235163e+00 -6.12272680e-01 5.72216392e-01 -3.11747104e-01
1.91033566e+00 3.47339600e-01 2.28330851e-01 9.62021112e-01
-7.23696500e-03 6.92453861e-01 6.75837696e-01 -8.58725131e-01
-3.35440010e-01 6.07562847e-02 5.51692665e-01 4.59368050e-01
8.45433593e-01 -1.64724007e-01 -3.37458909e-01 -1.76138490e-01
7.04905272e-01 -4.64105010e-01 -3.66933823e-01 -4.35442269e-01
-9.81669307e-01 8.47605824e-01 6.14074886e-01 5.36040664e-01
-2.51731008e-01 1.33514106e-01 4.57383722e-01 3.75420034e-01
7.85419345e-01 1.00189161e+00 -9.73100603e-01 2.35923067e-01
-5.07802129e-01 2.92531848e-01 1.38251936e+00 1.47786117e+00
9.69669282e-01 -4.07765687e-01 -2.51005918e-01 7.34463930e-01
1.05355084e-01 -1.21232949e-01 4.06371921e-01 -4.12479758e-01
8.53280127e-01 1.02084410e+00 -1.89503595e-01 -1.10464537e+00
-5.81155598e-01 -2.96920747e-01 -6.30816102e-01 -3.75916213e-01
5.58312349e-02 -3.30077410e-01 -9.74946618e-01 1.41704440e+00
1.65658712e-01 -1.52515089e-02 7.11655319e-01 1.69196755e-01
1.54878509e+00 2.42768347e-01 2.66721785e-01 -1.09092221e-01
1.46713138e+00 -7.59088695e-01 -8.77398968e-01 -3.69979203e-01
1.25511694e+00 -4.41913873e-01 3.96344841e-01 -7.76361823e-02
-7.15562940e-01 -1.99653566e-01 -1.21738291e+00 -4.41794336e-01
-1.04689157e+00 -7.91502595e-02 1.54298913e+00 5.58399916e-01
-9.19106305e-01 6.93875909e-01 -5.72865963e-01 -7.00654864e-01
6.68906808e-01 5.87265193e-01 -7.29581952e-01 1.93135589e-01
-1.57031202e+00 1.47551477e+00 9.40779567e-01 -7.22554028e-02
-3.50517571e-01 -4.87139165e-01 -1.33405411e+00 6.98207840e-02
8.92658889e-01 -7.45002568e-01 1.08839917e+00 -2.72986680e-01
-9.38619673e-01 9.79210198e-01 5.94655424e-02 -8.66261363e-01
-2.93502450e-01 -5.22245824e-01 -6.81001127e-01 -7.70673752e-02
2.04796329e-01 4.35225010e-01 9.68162790e-02 -1.28412235e+00
-4.18889165e-01 -1.25360176e-01 4.90822196e-01 2.35804170e-01
-3.10089767e-01 2.34931797e-01 -8.74844790e-02 -4.97357279e-01
1.44399758e-02 -7.01899409e-01 -2.79360324e-01 -7.31052816e-01
-1.06139266e+00 -8.36993277e-01 6.44208789e-01 -4.57661301e-01
1.14101589e+00 -1.64238536e+00 1.45380795e-01 2.97532827e-01
4.71406907e-01 2.58549213e-01 6.27759546e-02 7.39399552e-01
-5.02193749e-01 4.77509439e-01 -1.14110783e-01 -4.67068665e-02
5.42997867e-02 6.34714365e-01 -4.33151245e-01 9.83253471e-04
6.76474810e-01 1.32325006e+00 -1.07575572e+00 -6.89251900e-01
-1.88072488e-01 2.25328967e-01 -3.19643348e-01 1.60182014e-01
-2.92779088e-01 -1.18776679e-01 -5.51001251e-01 4.54987764e-01
2.25448161e-01 -3.54533970e-01 7.64726520e-01 -2.98737317e-01
4.66401994e-01 1.08783245e+00 -8.43272984e-01 1.46476722e+00
-4.53167439e-01 6.37176991e-01 -5.41517794e-01 -1.20817459e+00
1.01383507e+00 7.29813576e-01 4.03137445e-01 2.14929394e-02
1.70232847e-01 3.87506373e-02 1.96953654e-01 -4.09051299e-01
6.62567079e-01 -1.08439319e-01 -5.20282872e-02 2.86874980e-01
7.11900830e-01 -4.93876666e-01 5.27363598e-01 6.07429028e-01
1.42496157e+00 5.03354728e-01 9.67775166e-01 -1.53309807e-01
3.04829806e-01 5.69532365e-02 2.59438694e-01 4.25761461e-01
3.34721148e-01 3.51514816e-01 8.45880270e-01 -4.06497568e-01
-5.12968481e-01 -9.23764646e-01 -3.31873447e-01 9.41163421e-01
-2.31687903e-01 -1.17312717e+00 1.14610707e-02 -1.37036920e+00
1.52725816e-01 7.38992393e-01 -9.27015901e-01 -6.84082285e-02
-6.33517563e-01 -7.44589388e-01 7.84462810e-01 8.16013455e-01
1.47796258e-01 -1.23560131e+00 -2.40412634e-02 2.81548291e-01
3.32943201e-02 -1.38206470e+00 2.94511616e-01 8.52399051e-01
-6.18199170e-01 -1.37100303e+00 -5.00218794e-02 -8.56909931e-01
4.90849614e-01 -2.61969715e-01 1.67014134e+00 1.31071201e-02
-6.86604977e-02 1.90714389e-01 -7.05727696e-01 -5.18467009e-01
-3.35077524e-01 7.10268259e-01 -1.60229653e-01 -6.52536273e-01
9.14305508e-01 -7.33400583e-01 6.33282438e-02 -1.77164495e-01
-7.71374702e-01 -1.38947621e-01 5.20496309e-01 7.45877326e-01
1.91616893e-01 2.03963295e-01 7.11525679e-01 -1.85477257e+00
8.00232887e-01 -7.26274848e-01 1.65607229e-01 3.65236014e-01
-7.25875914e-01 3.47105622e-01 3.11339080e-01 -1.57573476e-01
-1.01367307e+00 -2.03382805e-01 -8.63799155e-02 8.16187859e-02
-1.61233634e-01 8.72632384e-01 -2.30170608e-01 2.09073484e-01
1.21358299e+00 -4.05960172e-01 -5.43734610e-01 -2.87199914e-01
9.46587145e-01 4.56213474e-01 5.37532270e-01 -9.43799913e-01
9.58057761e-01 3.77566554e-02 2.81118393e-01 -5.09323597e-01
-1.30142605e+00 -4.72541958e-01 -9.95425463e-01 6.93803728e-01
6.92424655e-01 -7.57845700e-01 -2.98865020e-01 -4.13455009e-01
-1.17192686e+00 -1.96612358e-01 -5.86904705e-01 3.62064630e-01
-4.45299655e-01 1.93377331e-01 -9.63059187e-01 -6.17957294e-01
-4.43504125e-01 -3.78585815e-01 8.21064115e-01 2.07064047e-01
-8.06196868e-01 -1.32473528e+00 2.60236859e-01 1.73158392e-01
9.01843607e-02 4.31787491e-01 1.06662476e+00 -1.36211061e+00
-2.30594426e-01 -3.16477984e-01 -4.98714983e-01 1.06559597e-01
7.30988920e-01 3.73746618e-04 -9.51943934e-01 1.36658311e-01
-5.61220586e-01 -6.75000727e-01 9.84845102e-01 -1.07559130e-01
6.94664299e-01 -6.06893301e-01 -9.71330285e-01 4.76288468e-01
1.17351270e+00 -7.38170817e-02 7.45963752e-01 4.65167850e-01
9.64235246e-01 8.37705374e-01 3.39737386e-01 -7.73875490e-02
6.17151499e-01 5.53136110e-01 6.42867461e-02 -3.80640067e-02
-3.17284077e-01 -3.44374686e-01 -2.02651158e-01 5.34770370e-01
-6.17148578e-01 -1.63384154e-01 -9.95264888e-01 8.14877927e-01
-1.85721004e+00 -8.68461013e-01 -3.72866899e-01 1.41986477e+00
1.51764762e+00 3.52671444e-01 -7.15530515e-02 2.26816058e-01
3.01959842e-01 5.79594150e-02 -1.04161821e-01 -4.56149429e-01
-2.56487489e-01 1.00480461e+00 4.08151060e-01 3.49139005e-01
-1.47889590e+00 1.50600886e+00 6.50563765e+00 5.89375675e-01
-2.34339416e-01 -1.11335166e-01 2.75086492e-01 4.34559137e-01
-5.57421327e-01 4.80384350e-01 -8.57080877e-01 -5.32160640e-01
9.80315208e-01 -1.86079264e-01 -3.06832064e-02 7.72369087e-01
-9.42166924e-01 2.39316031e-01 -1.62318397e+00 5.91386318e-01
8.27887654e-02 -1.42666054e+00 1.41839683e-01 5.94166443e-02
7.03651309e-01 -2.82219470e-01 -3.97443056e-01 8.47105443e-01
1.17617404e+00 -1.51050985e+00 -7.25495890e-02 2.82672524e-01
8.13589752e-01 -6.21106327e-01 1.11265934e+00 -2.23237082e-01
-1.37765419e+00 3.76407385e-01 -3.99068505e-01 -3.11236978e-01
-1.57116383e-01 5.04058242e-01 -1.26103723e+00 1.24405563e+00
4.20634627e-01 1.03520787e+00 -8.47234488e-01 3.59400123e-01
-8.68865669e-01 5.82013607e-01 -2.19589412e-01 -5.89089729e-02
-6.94717839e-02 1.90033615e-01 1.94368780e-01 1.51811469e+00
-2.91106790e-01 2.91292965e-01 4.10580710e-02 7.47272670e-01
-3.17496896e-01 7.78024644e-02 -1.09190702e+00 -2.59220749e-01
4.35473919e-01 1.43916392e+00 -6.27969682e-01 -5.34453869e-01
-6.01296782e-01 5.03245950e-01 9.86020148e-01 5.00344932e-01
-3.00098836e-01 -7.12442160e-01 5.43724000e-01 -1.62592962e-01
3.28727245e-01 -1.89426377e-01 -3.29175681e-01 -1.15815020e+00
-1.83491260e-01 -5.37096262e-01 7.63201237e-01 -5.37437260e-01
-1.54058838e+00 8.39378476e-01 3.25390995e-01 -5.58762133e-01
-6.29961491e-01 -6.22892082e-01 -3.20240676e-01 9.15937543e-01
-1.43463540e+00 -1.64901233e+00 1.56001464e-01 5.65103352e-01
2.81841364e-02 -2.08568856e-01 1.38698220e+00 3.18360934e-03
-5.44736624e-01 6.77667856e-01 -6.26115263e-01 7.93993473e-01
6.35340393e-01 -1.56464696e+00 7.14004040e-01 5.30518413e-01
8.40656459e-01 1.10628450e+00 6.34204328e-01 -8.05131853e-01
-9.96720612e-01 -9.59088147e-01 1.24087024e+00 -1.17852962e+00
1.08231258e+00 -3.92419696e-01 -9.64684069e-01 1.58803821e+00
4.32873189e-01 1.78906202e-01 9.48102236e-01 1.17105842e+00
-7.40371943e-01 1.33581340e-01 -8.27340841e-01 6.36195123e-01
1.45633543e+00 -6.27636850e-01 -1.35776389e+00 4.29217160e-01
7.85026848e-01 -3.25651079e-01 -1.32488072e+00 8.32659662e-01
1.36714503e-01 -3.29676747e-01 1.06236947e+00 -1.26846492e+00
6.20430291e-01 -7.58373551e-03 -1.88797280e-01 -1.32260108e+00
-4.52220231e-01 -6.45817220e-01 -7.05152214e-01 1.69435024e+00
1.01230419e+00 -6.40969634e-01 8.06378543e-01 5.99452138e-01
3.52553464e-02 -9.20252740e-01 -5.53994238e-01 -7.21408546e-01
3.19637448e-01 -3.46578389e-01 5.82272291e-01 1.28729868e+00
5.93393743e-01 1.25977373e+00 9.11240876e-02 4.38077636e-02
1.68040574e-01 9.69647691e-02 9.49163735e-01 -1.53847837e+00
-2.69847810e-01 -3.73046905e-01 -7.88592100e-01 -8.01122963e-01
6.09450996e-01 -1.19249487e+00 -1.95121258e-01 -2.11051679e+00
9.67016295e-02 -6.97131872e-01 -3.74885976e-01 1.01543713e+00
-4.26096797e-01 1.55417800e-01 -3.49225909e-01 1.69778749e-01
-3.80968958e-01 4.59859997e-01 9.70827878e-01 -3.35433871e-01
-2.01337002e-02 -1.13476761e-01 -1.28923440e+00 6.52546585e-01
6.44360662e-01 -4.63436991e-01 -6.54220581e-01 -5.88510185e-02
6.33711636e-01 -2.55048305e-01 3.65990512e-02 -5.83347678e-01
1.32600442e-01 -5.96894659e-02 4.65955019e-01 -2.38061428e-01
3.93400311e-01 -6.38646364e-01 -2.07472533e-01 -1.32422701e-01
-5.26182711e-01 -3.84195924e-01 1.50146216e-01 4.15365010e-01
-2.89167613e-01 -1.79856941e-01 2.48306952e-02 -5.11089861e-02
-6.06293023e-01 1.86031803e-01 9.10772830e-02 4.25678849e-01
8.67735028e-01 1.43795639e-01 -7.31133938e-01 -3.18571657e-01
-7.67320275e-01 1.25752121e-01 6.57036901e-02 3.57151896e-01
4.11395490e-01 -1.40482378e+00 -9.15278196e-01 -2.53521204e-01
3.98551017e-01 4.05635089e-01 -4.88240778e-01 5.23366809e-01
-2.26790503e-01 6.05761647e-01 1.42452940e-01 8.48291963e-02
-9.21712697e-01 9.13200080e-01 8.57385248e-02 -8.67848873e-01
-8.16792607e-01 1.06228781e+00 7.89363831e-02 -5.60757160e-01
-1.67975292e-01 -6.08659327e-01 -8.03339362e-01 1.67954504e-01
1.82836190e-01 -2.56143123e-01 2.43211180e-01 -4.71969604e-01
-6.51629865e-01 2.71418273e-01 -4.47025567e-01 2.12978050e-02
1.49230742e+00 4.06549454e-01 -1.94668412e-01 4.51888114e-01
8.50159109e-01 3.34301114e-01 -5.53931594e-01 -4.94549870e-01
6.43910885e-01 -5.06128743e-02 -3.59571576e-01 -7.16737807e-01
-6.98448896e-01 3.11938554e-01 -5.72693467e-01 7.10711777e-01
7.30964839e-01 7.44167149e-01 5.17293274e-01 7.34994233e-01
3.22836071e-01 -5.61023295e-01 -2.61308134e-01 7.86382675e-01
8.87946963e-01 -1.08398807e+00 6.00414693e-01 -1.06913126e+00
-5.49021482e-01 1.07126510e+00 6.67040884e-01 -3.55404198e-01
8.13069582e-01 5.22694528e-01 -2.12877259e-01 -6.30989015e-01
-9.45607066e-01 -8.85592103e-01 5.94657362e-01 1.21107781e+00
1.06451058e+00 2.18465671e-01 -2.06158891e-01 6.83542609e-01
-5.82231462e-01 -3.38274986e-01 3.31090420e-01 9.97788846e-01
1.74261443e-02 -1.36934483e+00 2.52944797e-01 9.21046615e-01
-5.68472624e-01 -6.58662140e-01 -9.05297816e-01 1.11799705e+00
1.59916598e-02 1.18731475e+00 -1.41986787e-01 -5.28731346e-01
4.07255113e-01 2.03643203e-01 7.59130657e-01 -1.36871421e+00
-5.18612564e-01 -6.26033247e-01 1.12417817e+00 -1.00142762e-01
-6.67691946e-01 -4.15992290e-01 -1.22601104e+00 -1.32817358e-01
-6.38425589e-01 4.00155693e-01 -1.90228559e-02 1.30464900e+00
9.11853984e-02 8.56713533e-01 -2.09232122e-02 -3.38721842e-01
-1.79625116e-02 -1.49741614e+00 -6.36632264e-01 4.98437881e-01
-3.67285088e-02 -8.87600958e-01 -3.26916352e-02 -9.02236924e-02]
|
[9.343609809875488, 8.534507751464844]
|
b99d5db7-3664-4a1c-a885-1fff13268856
|
deep-parametric-indoor-lighting-estimation
|
1910.08812
| null |
https://arxiv.org/abs/1910.08812v1
|
https://arxiv.org/pdf/1910.08812v1.pdf
|
Deep Parametric Indoor Lighting Estimation
|
We present a method to estimate lighting from a single image of an indoor scene. Previous work has used an environment map representation that does not account for the localized nature of indoor lighting. Instead, we represent lighting as a set of discrete 3D lights with geometric and photometric parameters. We train a deep neural network to regress these parameters from a single image, on a dataset of environment maps annotated with depth. We propose a differentiable layer to convert these parameters to an environment map to compute our loss; this bypasses the challenge of establishing correspondences between estimated and ground truth lights. We demonstrate, via quantitative and qualitative evaluations, that our representation and training scheme lead to more accurate results compared to previous work, while allowing for more realistic 3D object compositing with spatially-varying lighting.
|
['Jean-François Lalonde', 'Christian Gagné', 'Yannick Hold-Geoffroy', 'Marc-André Gardner', 'Kalyan Sunkavalli']
|
2019-10-19
| null | null | null | null |
['lighting-estimation']
|
['computer-vision']
|
[ 2.32580036e-01 -1.29155591e-01 5.54082930e-01 -9.12053943e-01
-6.25095785e-01 -7.21485794e-01 6.83518112e-01 -1.00970015e-01
-3.67441922e-01 6.53470457e-01 2.07376525e-01 -1.68429002e-01
3.87749791e-01 -8.40690792e-01 -1.04945779e+00 -3.52155834e-01
1.44458875e-01 2.88442165e-01 -1.11536965e-01 9.28084776e-02
-6.54132515e-02 7.04104722e-01 -1.67995369e+00 6.03459543e-03
5.83286405e-01 1.03696537e+00 1.68970600e-01 1.04321098e+00
-1.81393567e-02 6.11515701e-01 -5.99561274e-01 -1.72092482e-01
6.92904830e-01 -2.36988217e-01 -4.28870469e-01 4.24427032e-01
1.06818962e+00 -6.80720866e-01 -2.20187113e-01 8.76354694e-01
3.86465967e-01 2.27569342e-01 6.19457841e-01 -1.02060020e+00
-7.23694742e-01 -3.93213034e-01 -3.78548563e-01 -3.00923347e-01
6.67552948e-01 2.70545125e-01 7.64029801e-01 -7.00999737e-01
4.56680983e-01 1.14972115e+00 9.89546537e-01 2.31164292e-01
-1.52813089e+00 -3.09623629e-01 2.89446414e-01 -4.07432824e-01
-1.41926968e+00 -4.55730826e-01 9.17461276e-01 -5.12733936e-01
9.14011419e-01 1.23517282e-01 8.04348469e-01 9.59106028e-01
-3.33291814e-02 3.24339092e-01 1.35671759e+00 -5.42284667e-01
3.65541935e-01 2.21880957e-01 -3.05488169e-01 9.28875268e-01
-1.01092450e-01 1.50949568e-01 -2.67416507e-01 8.85618851e-02
9.97314870e-01 -4.03853580e-02 -3.27124268e-01 -6.68527663e-01
-1.02443182e+00 5.16115844e-01 8.09949398e-01 -3.04671079e-01
-3.14494878e-01 5.10757864e-01 -1.38033986e-01 8.03652331e-02
6.72134221e-01 4.73222822e-01 -4.63873386e-01 1.27197266e-01
-7.85327733e-01 1.85470477e-01 6.51042283e-01 9.03587759e-01
1.35118246e+00 -1.26555607e-01 2.27830380e-01 5.78693986e-01
5.24360061e-01 8.19975615e-01 -1.36925370e-01 -1.57931554e+00
1.48772135e-01 3.12119693e-01 4.24314559e-01 -8.52689087e-01
-3.56158257e-01 -1.63078040e-01 -3.80946726e-01 6.78323209e-01
4.64910507e-01 -1.38735846e-01 -1.13533878e+00 1.73189950e+00
3.80896509e-01 3.18995774e-01 -2.38288537e-01 9.32186246e-01
5.27331591e-01 5.45651674e-01 -1.98755413e-01 3.13884795e-01
8.50466728e-01 -7.71755099e-01 -4.03036028e-01 -4.13773745e-01
1.38265431e-01 -7.93921173e-01 1.40610564e+00 4.01582509e-01
-1.20365906e+00 -4.58797932e-01 -9.95556176e-01 -5.23298979e-01
-4.95789260e-01 1.51233330e-01 6.03910565e-01 6.20902121e-01
-1.56444061e+00 3.28545779e-01 -8.69279146e-01 -4.59204942e-01
1.63191766e-01 2.57014394e-01 -2.44464844e-01 -1.50857419e-01
-7.79737651e-01 9.68809664e-01 -4.47170474e-02 1.78038806e-01
-1.05839896e+00 -7.14823484e-01 -1.13730228e+00 -2.53457338e-01
-8.22385997e-02 -8.83021653e-01 1.23788977e+00 -9.21170175e-01
-1.60478532e+00 8.94989967e-01 -4.41655040e-01 -1.87234268e-01
5.11408746e-01 -2.83544838e-01 8.06645751e-02 -1.05265297e-01
3.51921394e-02 8.53153527e-01 5.48642695e-01 -1.90453494e+00
-5.26079476e-01 -2.52330929e-01 6.10888779e-01 3.22658151e-01
-5.95324151e-02 -2.70009845e-01 -5.67922354e-01 -1.27808034e-01
1.33497387e-01 -8.11525583e-01 -3.33348006e-01 5.99895597e-01
-5.03553808e-01 4.44485724e-01 4.62592691e-01 -6.77321672e-01
5.01200378e-01 -1.97248745e+00 -6.57641366e-02 2.50698566e-01
2.54570395e-02 -4.04321462e-01 -8.53962377e-02 5.61859868e-02
1.68556660e-01 -5.76927960e-02 -3.74042511e-01 -1.12913275e+00
1.81817248e-01 3.21604937e-01 -4.12211597e-01 7.53527820e-01
1.00398719e-01 7.13607013e-01 -1.05311549e+00 -1.66378215e-01
7.92267323e-01 1.14093363e+00 -6.83339238e-01 3.70504051e-01
-4.40949321e-01 7.36941636e-01 -3.95545177e-02 4.76375878e-01
7.28274763e-01 -6.78748041e-02 -9.11063924e-02 -1.67554751e-01
-2.66237497e-01 3.39902192e-01 -1.19494224e+00 2.11034775e+00
-1.12064576e+00 9.15967524e-01 2.10969687e-01 -2.49832198e-01
8.83294344e-01 2.64528599e-02 4.28050101e-01 -6.48638308e-01
1.86360687e-01 4.65926751e-02 -7.38490343e-01 -2.41845623e-01
5.90770185e-01 -1.69125050e-01 9.91392285e-02 3.36091250e-01
-1.36158079e-01 -7.63506889e-01 -3.14234465e-01 -2.02405289e-01
9.48090255e-01 7.07863510e-01 -9.98844430e-02 4.29112911e-02
3.01738262e-01 -2.06735417e-01 3.32016498e-01 5.25655389e-01
-1.03151752e-02 9.62804377e-01 -8.90444592e-02 -6.29331350e-01
-1.34961843e+00 -1.36754179e+00 -2.17379540e-01 6.22441292e-01
2.40721405e-01 -1.15671277e-01 -7.88306832e-01 -3.51647526e-01
1.57420456e-01 7.45001972e-01 -6.29414678e-01 2.32035413e-01
-6.10359013e-01 -4.11491930e-01 1.06365204e-01 4.21758950e-01
6.36310041e-01 -5.44136822e-01 -8.07604611e-01 -1.93987876e-01
-1.90565646e-01 -1.46155000e+00 -2.90390193e-01 2.89600462e-01
-3.84178430e-01 -9.74999845e-01 -2.86769897e-01 -5.76444685e-01
9.79869962e-01 2.97999084e-01 1.51838470e+00 -8.41239020e-02
-5.25573611e-01 8.12578976e-01 8.79493952e-02 -5.42693973e-01
-2.91555613e-01 -3.86443615e-01 2.01472118e-02 -8.54896456e-02
1.20704062e-01 -6.91431820e-01 -8.08879256e-01 1.31085262e-01
-6.43228412e-01 2.47547105e-01 2.15872731e-02 3.43834460e-01
8.17067266e-01 -3.46452072e-02 -3.21627975e-01 -5.44410944e-01
2.45008364e-01 -1.37114286e-01 -1.09150720e+00 3.63689195e-03
-3.73701155e-01 -2.69638207e-02 6.75233245e-01 1.50860567e-02
-1.20855629e+00 5.32435238e-01 -1.21533677e-01 -4.82365042e-01
-5.68226874e-01 -2.07908019e-01 -1.95081055e-01 -3.25894892e-01
7.50846386e-01 -1.00315504e-01 -2.99795538e-01 -3.47412854e-01
6.28446221e-01 3.56741279e-01 7.62268543e-01 -8.33849311e-01
1.16257846e+00 9.52190280e-01 2.61472702e-01 -5.87854326e-01
-1.04694557e+00 -3.08173776e-01 -8.78630996e-01 -1.68054074e-01
1.01051939e+00 -1.14242887e+00 -7.52271175e-01 2.51592904e-01
-1.24244773e+00 -9.17839825e-01 -4.13264543e-01 4.08565462e-01
-8.17885101e-01 -8.62098951e-03 -5.06233752e-01 -9.86226201e-01
3.20620865e-01 -1.13672149e+00 1.64512062e+00 5.13792410e-02
-9.46667045e-02 -1.24393702e+00 2.77058661e-01 1.45301372e-01
3.38066339e-01 7.27048755e-01 6.20390892e-01 4.49096471e-01
-1.02542722e+00 -9.63869095e-02 -2.91790724e-01 4.95015860e-01
4.58880097e-01 1.93886310e-01 -1.57688332e+00 -1.14199959e-01
3.72248963e-02 -3.14919800e-01 6.29544735e-01 4.78884488e-01
1.24912095e+00 -3.58171165e-02 2.12986078e-02 1.25378156e+00
1.91854215e+00 -1.31421462e-01 3.56058985e-01 5.59071660e-01
9.85663414e-01 5.53094149e-01 2.27734342e-01 3.20010841e-01
8.13144565e-01 5.76643169e-01 8.03424060e-01 -5.99934876e-01
-3.96121830e-01 -4.49757308e-01 5.89034371e-02 3.44372511e-01
-1.76400747e-02 -1.74071923e-01 -8.72551918e-01 4.75026935e-01
-1.48005331e+00 -4.60148245e-01 6.76099733e-02 2.39939618e+00
7.69791782e-01 -3.45327109e-02 -1.27666786e-01 -1.72910318e-01
3.10453504e-01 1.38299435e-01 -6.14970386e-01 -4.28743690e-01
-1.49058118e-01 9.66924429e-02 8.65287006e-01 1.09164822e+00
-1.11134732e+00 7.71674216e-01 7.33929968e+00 -2.30803519e-01
-1.12578714e+00 -2.94612683e-02 6.79442942e-01 -2.31726274e-01
-7.27769613e-01 -1.06816314e-01 -5.59445322e-01 3.01733911e-01
8.32373619e-01 4.40717399e-01 1.02983093e+00 7.27961898e-01
4.67547894e-01 -1.19951688e-01 -1.29766881e+00 9.03216720e-01
1.28090352e-01 -1.12435007e+00 -3.49950194e-01 2.26484597e-01
1.03769994e+00 3.22057337e-01 1.12701304e-01 -8.39266032e-02
9.19850647e-01 -1.01148653e+00 1.01615655e+00 6.28442347e-01
8.21951568e-01 -4.40262914e-01 2.24516749e-01 1.83482185e-01
-1.11827874e+00 5.60501553e-02 -2.61738092e-01 -2.77051866e-01
2.40191698e-01 5.69357753e-01 -1.01021206e+00 2.41607741e-01
8.33528757e-01 5.28569579e-01 -5.50059080e-01 9.27618563e-01
-5.88878810e-01 1.73973441e-01 -5.62702537e-01 3.74021024e-01
-1.99755244e-02 -2.98418522e-01 1.72620371e-01 1.07048368e+00
2.63851613e-01 -3.69524688e-01 3.51870745e-01 1.17074788e+00
-1.76506057e-01 -3.17594707e-01 -8.15539956e-01 6.88295245e-01
4.52742755e-01 1.29552162e+00 -6.37737155e-01 -9.67964306e-02
-3.50049883e-01 1.10266304e+00 5.04639149e-01 7.33430982e-01
-8.08926582e-01 -2.03030899e-01 8.96399081e-01 2.16752127e-01
1.06235489e-01 -6.33504808e-01 -5.67124307e-01 -1.18219447e+00
1.56350762e-01 -3.03896666e-01 -4.30617511e-01 -1.24406242e+00
-1.12961173e+00 3.64769608e-01 -4.96018045e-02 -8.79583776e-01
-2.60988921e-01 -6.23753667e-01 -4.78795409e-01 1.04042768e+00
-2.04738021e+00 -1.24614620e+00 -8.03961039e-01 3.87158573e-01
3.02812666e-01 5.72784483e-01 9.77598906e-01 2.53926605e-01
-2.50495702e-01 2.06894070e-01 6.97252601e-02 9.04165674e-03
6.85727119e-01 -1.87450194e+00 7.96716332e-01 9.08436000e-01
9.94929895e-02 6.13157690e-01 9.57894742e-01 -1.72879934e-01
-1.33840215e+00 -1.22832358e+00 4.41719502e-01 -9.86904442e-01
2.99160779e-01 -8.02846372e-01 -5.59733927e-01 8.92344177e-01
1.11585461e-01 3.72354209e-01 3.41819853e-01 9.13636163e-02
-4.87822592e-01 -2.13088840e-01 -1.42204809e+00 5.88250220e-01
1.16693246e+00 -8.43061209e-01 -1.14079081e-01 4.16629851e-01
9.05819237e-01 -7.83996403e-01 -7.60820329e-01 1.41136274e-01
4.86601859e-01 -1.22304630e+00 1.31779802e+00 1.01937674e-01
4.71577078e-01 -6.57304347e-01 -5.60717642e-01 -1.50941432e+00
-3.41406390e-02 -3.05017143e-01 -1.66266486e-02 1.00320899e+00
1.61926270e-01 -5.07444620e-01 8.34775388e-01 1.02847111e+00
-2.21817806e-01 -5.50363719e-01 -6.13232970e-01 -4.98080581e-01
-1.09183090e-02 -4.83075351e-01 9.70478237e-01 7.29087412e-01
-9.82364953e-01 2.88548297e-03 -1.67499810e-01 5.47941029e-01
1.00538003e+00 5.00213467e-02 1.07301998e+00 -1.13898599e+00
-2.79842764e-01 -1.54260367e-01 -3.46730798e-01 -1.12674534e+00
3.90273154e-01 -5.02429068e-01 5.82618713e-01 -1.94566834e+00
-1.27382383e-01 -6.83803856e-01 -8.14709514e-02 3.91631752e-01
6.37495816e-02 6.98393822e-01 -4.77468893e-02 -1.06030136e-01
-4.85093445e-01 4.60100085e-01 1.10063004e+00 -2.38638092e-02
-1.70852467e-01 -3.66140753e-01 -3.69478285e-01 8.92446637e-01
6.29293740e-01 -2.93255057e-02 -6.03637040e-01 -1.07191837e+00
3.33287030e-01 -5.99883735e-01 6.02097631e-01 -1.03116882e+00
-1.45554845e-03 -2.29889110e-01 9.06776905e-01 -3.86974216e-01
8.75947833e-01 -1.09329534e+00 1.45316020e-01 -2.54203118e-02
-2.83341587e-01 -4.51833606e-02 1.96564361e-01 3.97225171e-01
1.87914729e-01 1.47543505e-01 6.02897346e-01 -2.99526125e-01
-6.88313186e-01 2.66851038e-01 1.73025936e-01 -2.74088591e-01
7.89516747e-01 -4.30683285e-01 -1.44883871e-01 -4.50215399e-01
-4.58407521e-01 6.53611347e-02 1.33806276e+00 1.45699501e-01
6.32055581e-01 -1.36914527e+00 -3.10662895e-01 4.60468680e-01
1.91226527e-02 5.12595832e-01 -1.56608045e-01 1.77347213e-01
-9.12554622e-01 1.62214324e-01 1.84448466e-01 -7.76758909e-01
-8.86818886e-01 3.40051442e-01 8.84109676e-01 2.72625566e-01
-5.70331335e-01 8.75400245e-01 4.00178432e-01 -1.03491676e+00
3.23169589e-01 -7.08889902e-01 4.25177753e-01 -5.59533000e-01
2.16991588e-01 4.24405485e-02 8.31489936e-02 -6.23972595e-01
-2.45653197e-01 8.40319335e-01 6.60143852e-01 -2.63009787e-01
1.36246276e+00 -5.81088066e-01 -1.46432236e-01 7.38655508e-01
1.53891289e+00 2.80743152e-01 -1.97337794e+00 -1.85232535e-01
-5.26843369e-01 -8.76300812e-01 3.49265605e-01 -8.95043612e-01
-8.30553591e-01 7.80444086e-01 7.74420321e-01 2.16675713e-03
1.04569733e+00 -2.62233794e-01 6.73853457e-01 3.69252086e-01
6.07567430e-01 -8.91242981e-01 -1.10447392e-01 4.91341323e-01
5.42941511e-01 -1.53264678e+00 2.14724112e-02 -2.12560162e-01
-2.92102471e-02 9.76900160e-01 5.11319697e-01 -1.46096155e-01
6.78389549e-01 5.16146958e-01 4.05320376e-01 -1.14154831e-01
-3.23630333e-01 -2.47750655e-01 2.14412987e-01 7.41400659e-01
4.37437296e-01 3.16943265e-02 7.67275333e-01 -4.25879061e-01
-5.01583755e-01 -5.60486317e-02 4.22856748e-01 8.02349746e-01
-4.69992429e-01 -9.03926253e-01 -5.96019149e-01 -1.45704467e-02
-1.17384017e-01 -5.97440731e-03 -2.35032991e-01 4.99608934e-01
3.65815997e-01 8.73771846e-01 3.65523964e-01 -2.68730044e-01
4.47748423e-01 -2.44018748e-01 7.83633292e-01 -6.46036386e-01
-1.50246933e-01 -5.87771498e-02 -3.71246561e-02 -8.06315839e-01
-5.14160872e-01 -4.94510412e-01 -1.10079014e+00 -1.99919552e-01
2.34047882e-02 -3.06376010e-01 1.36012900e+00 6.63656354e-01
1.37355492e-01 6.19280040e-01 9.77412283e-01 -1.37499118e+00
1.94417350e-02 -3.54903698e-01 -4.27951425e-01 4.20905143e-01
9.81676161e-01 -6.10607564e-01 -5.89625418e-01 1.72795862e-01]
|
[9.584202766418457, -2.9875102043151855]
|
a1239ecb-cbde-4075-87af-f401ead07d45
|
biodivtab-semantic-table-annotation-benchmark
| null | null |
https://ceur-ws.org/Vol-3324/om2022_LTpaper4.pdf
|
https://ceur-ws.org/Vol-3324/om2022_LTpaper4.pdf
|
BiodivTab: Semantic Table Annotation Benchmark Construction, Analysis, and New Additions
|
Systems that annotate tabular data semantically have witnessed increasing attention from the community in recent years; this process is commonly known as Semantic Table Annotation (STA). Its objective is to map individual table elements to their counterparts from a Knowledge Graph (KG). Individual cells and columns are assigned to KG entities and classes to disambiguate their meaning. STA-systems achieve high scores on the existing, synthetic benchmarks but often struggle on real-world datasets. Thus, realistic evaluation benchmarks are needed to enable the advancement of the field. In this paper, we detail the construction pipeline of BiodivTab, the first benchmark based on real-world data from the biodiversity domain. In addition, we compare it with the existing benchmarks. Moreover, we highlight common data characteristics and challenges in the field. BiodivTab is publicly available and has 50 tables as a mixture of real and augmented samples from biodiversity datasets. It has been applied during the SemTab 2021 challenge, and participants achieved F1-scores of at most ∼ 60% across individual annotation tasks. Such results show that domain-specific benchmarks are more challenging for state-of-the-art systems than synthetic datasets.
|
['Birgitta König-Ries', 'Sirko Schindler', 'Nora Abdelmageed']
|
2023-01-10
| null | null | null |
ontology-matching-iswc-2022-2023-1
|
['table-annotation', 'table-annotation']
|
['knowledge-base', 'natural-language-processing']
|
[-4.40286584e-02 3.86825562e-01 -3.77087653e-01 -3.13761294e-01
-6.60888612e-01 -9.82753694e-01 6.87403142e-01 8.83995771e-01
-2.25948006e-01 1.14507329e+00 4.19208139e-01 2.03429982e-01
-2.03715153e-02 -9.97236371e-01 -8.52636635e-01 -1.21558048e-01
-1.22375600e-01 1.00551391e+00 3.35819364e-01 -4.35587138e-01
-1.01569936e-01 -1.55157834e-01 -1.75712252e+00 8.37484419e-01
8.40362668e-01 1.27830267e+00 -6.82151541e-02 8.70739669e-02
-5.90598583e-01 9.54779267e-01 -7.59398520e-01 -1.10069263e+00
1.56459019e-01 -1.26882389e-01 -1.07795429e+00 -5.05629897e-01
6.63759649e-01 4.76403266e-01 1.10055343e-03 1.03466845e+00
4.25289094e-01 -1.67170465e-01 4.79159296e-01 -1.64044642e+00
-6.62533879e-01 1.19191706e+00 -6.63672015e-02 -1.58341393e-01
5.49473166e-01 -1.50622383e-01 1.38446176e+00 -6.42287910e-01
1.34439421e+00 1.31672239e+00 1.04776216e+00 3.58745813e-01
-1.40069723e+00 -5.45169592e-01 -7.24233780e-03 4.49682385e-01
-1.56728661e+00 -2.75784314e-01 1.16444483e-01 -5.09280205e-01
1.10998595e+00 3.87159258e-01 6.09944940e-01 1.08450079e+00
-8.67338479e-02 6.71825230e-01 8.95554781e-01 -1.61432903e-02
3.12581390e-01 1.98511168e-01 -3.18717137e-02 5.45539021e-01
9.15445447e-01 -7.56777883e-01 -9.67287898e-01 8.28851908e-02
9.98809859e-02 -5.89865267e-01 -1.89833671e-01 -8.95591617e-01
-1.64976203e+00 3.41059715e-01 6.53691471e-01 6.59747571e-02
-1.25818834e-01 -1.38383890e-02 9.27547097e-01 2.77386457e-02
3.69377762e-01 8.15180779e-01 -6.50889993e-01 -2.11404279e-01
-4.42303747e-01 6.95057869e-01 1.18147564e+00 1.46943986e+00
4.94779885e-01 -5.47136962e-01 -3.04509193e-01 8.59382927e-01
-1.57328412e-01 3.25574160e-01 9.70732421e-02 -7.71124005e-01
8.14935029e-01 1.17802882e+00 1.50561377e-01 -1.08092821e+00
-5.03765881e-01 -4.55581456e-01 -7.72651255e-01 -2.97903270e-01
7.89059699e-01 3.25228125e-01 -6.27135515e-01 1.57856715e+00
4.85438615e-01 -2.51884311e-01 3.26367110e-01 6.66182637e-01
1.36463940e+00 2.74421573e-01 2.72229820e-01 1.99281842e-01
1.69697452e+00 -6.10078752e-01 -1.05826735e+00 -2.12025180e-01
7.12547779e-01 -4.77633357e-01 1.27956140e+00 4.73288834e-01
-8.51211727e-01 -2.61446297e-01 -1.09906733e+00 -2.98257262e-01
-1.37471437e+00 -5.14580049e-02 7.08917379e-01 5.23247600e-01
-8.27687204e-01 3.71176302e-01 -5.47602654e-01 -8.86919260e-01
4.35642630e-01 -4.96109053e-02 -7.35482454e-01 1.25238135e-01
-1.51497126e+00 1.08693945e+00 9.94931877e-01 -1.54644087e-01
-5.81623554e-01 -1.12759995e+00 -8.98449779e-01 8.01410079e-02
7.69810915e-01 -7.00577855e-01 1.17426562e+00 -9.07596387e-03
-6.25822186e-01 1.36008465e+00 3.42005402e-01 -7.93817043e-01
5.59859991e-01 -1.78483814e-01 -5.94062626e-01 -1.86039075e-01
4.73212093e-01 7.90688515e-01 -1.29297554e-01 -1.24812818e+00
-8.47421527e-01 -2.27082610e-01 2.28094384e-01 -4.36407849e-02
-3.02059919e-01 -3.26112479e-01 -5.65562963e-01 -6.18893802e-01
-6.77402467e-02 -7.10417807e-01 3.89561653e-01 -1.26874875e-02
-6.68906629e-01 -3.83827418e-01 4.96336460e-01 -5.61442971e-01
1.45706630e+00 -1.84763288e+00 1.08636066e-01 -2.23183155e-01
2.35514835e-01 -1.54934436e-01 2.84887195e-01 8.20330977e-01
-3.15252952e-02 4.31570262e-01 -3.97558093e-01 -1.49823397e-01
4.48190629e-01 4.18235630e-01 -3.82481813e-01 3.65917310e-02
2.66317934e-01 8.94509017e-01 -1.19822681e+00 -6.52767599e-01
-5.73541373e-02 1.27373487e-01 -4.98144984e-01 -7.97677562e-02
-8.44450533e-01 1.04241036e-01 -1.20585756e-02 9.23157513e-01
5.70595920e-01 -3.64078045e-01 6.15522027e-01 -6.96112394e-01
7.37878829e-02 5.68089008e-01 -1.18324912e+00 1.98283076e+00
2.79111154e-02 5.19619107e-01 -1.92711562e-01 -6.52119935e-01
9.01036799e-01 -4.61661033e-02 5.27282894e-01 -6.89393759e-01
-2.09753662e-01 3.49500477e-01 -2.81421632e-01 -1.82834923e-01
8.67197037e-01 2.96874166e-01 -5.70630431e-01 -1.05486684e-01
2.82417834e-01 -2.33396292e-01 8.84568036e-01 4.04303849e-01
1.13887382e+00 4.12775218e-01 5.92404962e-01 -6.28042996e-01
5.76169908e-01 7.03576624e-01 5.49194753e-01 3.99796188e-01
-1.53587162e-02 3.34215283e-01 9.04252529e-01 -7.74335265e-01
-1.02199650e+00 -1.10024786e+00 -3.92369062e-01 8.68639052e-01
2.85903305e-01 -1.31720817e+00 -7.84494519e-01 -9.64846373e-01
4.90510851e-01 7.83946335e-01 -9.28871334e-01 -6.32596202e-03
-1.81737497e-01 -6.98294878e-01 8.93692315e-01 4.83307719e-01
7.00908840e-01 -1.08382595e+00 -4.78412390e-01 2.59669662e-01
-6.42413676e-01 -1.49468470e+00 6.67710006e-02 2.64402658e-01
-1.41334668e-01 -1.50218618e+00 2.70109549e-02 -3.19265455e-01
3.22519809e-01 -3.23397845e-01 1.83217359e+00 -3.12343568e-01
-2.26846069e-01 1.28493652e-01 -4.51380879e-01 -6.24013007e-01
-3.80106777e-01 5.51903188e-01 -2.84664240e-02 -3.27355146e-01
3.96566242e-01 -1.39455333e-01 -4.26316082e-01 5.89056134e-01
-9.35947359e-01 2.29206607e-01 9.12383571e-02 6.74152195e-01
9.61169899e-01 4.81323265e-02 6.60147667e-01 -1.11164081e+00
3.69424254e-01 -5.93886018e-01 -6.95199192e-01 5.83379984e-01
-7.06807852e-01 2.69933820e-01 6.61040127e-01 2.04188555e-01
-8.06755304e-01 -1.32323444e-01 8.34051967e-02 1.04593277e-01
1.84015498e-01 7.74115562e-01 -5.37953675e-01 4.08336669e-01
9.06557560e-01 -1.55526087e-01 -2.83090383e-01 -6.60506964e-01
4.85878199e-01 4.93978798e-01 1.05912888e+00 -9.61728990e-01
5.26234150e-01 3.97225440e-01 2.21771389e-01 -5.89535199e-02
-1.18594956e+00 -2.46272415e-01 -6.91182911e-01 -7.65377432e-02
7.94500649e-01 -1.10554695e+00 -8.34537864e-01 2.24924102e-01
-8.49571049e-01 -3.40290755e-01 -3.59674424e-01 -9.09874290e-02
-4.92370903e-01 -1.49355397e-01 -2.00665504e-01 -3.61510217e-01
-3.51050019e-01 -7.31719136e-01 1.22530031e+00 -1.68313682e-01
-3.56092960e-01 -9.69434679e-01 -1.20082445e-01 3.69228274e-01
3.61931980e-01 8.21477830e-01 1.07456386e+00 -9.31313396e-01
-4.57537830e-01 -5.54791875e-02 -2.61421293e-01 -8.76900703e-02
1.83954641e-01 5.60333692e-02 -8.24762344e-01 1.77338645e-02
-8.40041697e-01 -5.56562483e-01 8.47090602e-01 -3.89846414e-01
1.36820662e+00 -2.86314547e-01 -5.52490950e-01 5.56754887e-01
1.44266522e+00 -1.11497387e-01 6.31403863e-01 9.14434552e-01
7.54309237e-01 7.98477352e-01 1.00096166e+00 3.71438444e-01
9.50127363e-01 8.80993485e-01 7.52759337e-01 2.60658175e-01
-2.36049294e-01 -6.11645043e-01 -9.85293686e-02 4.24982667e-01
3.28350365e-01 -5.46580911e-01 -1.47273266e+00 6.52250111e-01
-2.07406020e+00 -8.32386613e-01 -3.78370851e-01 2.40721297e+00
1.16993451e+00 3.11419964e-01 1.66167706e-01 1.44943133e-01
6.26267374e-01 -1.79682508e-01 -5.15555501e-01 4.63004522e-02
-6.73528790e-01 -7.42224306e-02 4.99655306e-01 1.86238978e-02
-1.21927249e+00 1.09080350e+00 6.07457876e+00 8.98079336e-01
-5.96814394e-01 -7.91303962e-02 6.11696184e-01 6.22211769e-02
-2.42959902e-01 -1.15655698e-01 -1.12812638e+00 5.02809644e-01
9.17506576e-01 -5.29011786e-01 2.58897126e-01 8.13872397e-01
-5.16153097e-01 -4.96415757e-02 -1.29638910e+00 8.59652579e-01
-7.75160268e-02 -1.55885708e+00 3.29446346e-01 -1.40895367e-01
5.02780318e-01 -1.25708863e-01 -2.89065301e-01 5.68725586e-01
5.51191151e-01 -1.10429442e+00 1.04371738e+00 4.15146321e-01
1.17840755e+00 -6.02958739e-01 8.77576947e-01 -9.50626805e-02
-1.53846908e+00 2.77418882e-01 -3.71977746e-01 2.35335305e-01
-1.43725827e-01 4.70046222e-01 -9.96783912e-01 1.05449212e+00
1.26736867e+00 1.04294622e+00 -1.11606383e+00 9.57070351e-01
-1.58383906e-01 2.93300480e-01 -3.58404040e-01 1.04961380e-01
-7.44415820e-02 1.65821761e-01 1.92207962e-01 1.28978801e+00
2.08040476e-01 -3.39449018e-01 -1.04356315e-02 7.41608620e-01
-4.73040342e-01 2.68445313e-01 -6.02224469e-01 -2.30479509e-01
7.41982996e-01 1.39209318e+00 -8.26166749e-01 -5.66907108e-01
-1.20811209e-01 4.49097335e-01 5.35709620e-01 -2.12612316e-01
-9.21796739e-01 -4.04595226e-01 9.80901003e-01 2.24021450e-01
9.06054452e-02 2.13520467e-01 -5.25277674e-01 -1.12229335e+00
6.39984533e-02 -9.37536836e-01 1.00506163e+00 -9.70654368e-01
-1.50951469e+00 6.46143615e-01 2.82748103e-01 -1.27647769e+00
-9.95099321e-02 -6.97409391e-01 3.20737392e-01 5.67431927e-01
-1.20953333e+00 -1.20487618e+00 -9.02848065e-01 3.94948661e-01
1.77432179e-01 4.13934467e-03 1.07028341e+00 4.37698305e-01
-5.57262063e-01 6.55332863e-01 2.18495335e-02 1.73330829e-01
1.14578724e+00 -1.70778406e+00 8.94135356e-01 4.97189403e-01
9.11226403e-03 5.36672831e-01 7.57741809e-01 -9.63248134e-01
-1.42779124e+00 -1.34269488e+00 1.00204074e+00 -1.13245177e+00
1.01698112e+00 -9.75787938e-01 -1.05277073e+00 6.90379977e-01
3.30425650e-01 1.76531017e-01 5.80643475e-01 1.12164982e-01
-7.59380162e-01 -3.79795969e-01 -1.30413318e+00 4.78754818e-01
1.64133382e+00 -3.12761545e-01 -4.54515249e-01 5.09382665e-01
7.62546718e-01 -7.66637444e-01 -1.39307904e+00 7.30609477e-01
6.88874722e-01 -7.84501493e-01 1.02120197e+00 -6.38572752e-01
5.08059561e-01 -7.48511672e-01 -4.56961274e-01 -1.43111038e+00
7.27981329e-02 -3.40940714e-01 -1.98572561e-01 1.61147213e+00
5.53851545e-01 -4.84258622e-01 6.77099228e-01 2.60008782e-01
-2.17029676e-01 -4.34453785e-01 -7.34852850e-01 -1.01501226e+00
-6.93378672e-02 -2.37475350e-01 1.01587546e+00 1.34107459e+00
2.80439585e-01 3.91522467e-01 -1.70621593e-02 -1.06645055e-01
7.21018314e-01 -6.66149240e-03 9.17462647e-01 -1.52344692e+00
2.83298343e-01 -4.49403286e-01 -6.06743157e-01 -2.48127908e-01
-2.07563732e-02 -1.25543451e+00 -1.95077300e-01 -1.91579676e+00
4.60504219e-02 -5.16664267e-01 -2.35078007e-01 9.89575386e-01
-2.90070117e-01 3.47030967e-01 3.31512988e-02 1.58512756e-01
-9.02041614e-01 3.39106321e-01 8.57232094e-01 -3.71764868e-01
2.64163971e-01 -7.69372880e-01 -7.84301221e-01 2.16517583e-01
7.10787714e-01 -4.41370070e-01 -3.79780024e-01 -2.75746286e-01
8.69583845e-01 -3.24163049e-01 1.58088163e-01 -1.28132188e+00
2.43836761e-01 1.43903284e-03 2.94034898e-01 -8.14676166e-01
1.28095120e-01 -8.60661983e-01 8.06114614e-01 2.87186831e-01
-3.49692971e-01 2.21642092e-01 6.34787202e-01 3.73332620e-01
-3.78312558e-01 3.10116768e-01 4.07659978e-01 -7.77636319e-02
-9.82479453e-01 1.25892330e-02 2.25538269e-01 4.68590558e-01
9.28275168e-01 1.18015602e-01 -1.13064682e+00 3.03857382e-02
-6.17131948e-01 5.72146595e-01 6.49869978e-01 6.48431063e-01
4.62419949e-02 -1.40243518e+00 -6.96288526e-01 -1.86818480e-01
1.01949179e+00 1.73967808e-01 -1.82737246e-01 6.61404312e-01
-9.42260325e-01 6.24395132e-01 -6.57615781e-01 -6.16543114e-01
-9.81802344e-01 6.88665986e-01 3.20923179e-01 -4.78496701e-01
-4.54792798e-01 5.95363915e-01 -3.77348550e-02 -7.86171436e-01
2.22606167e-01 -4.85587806e-01 -2.19576046e-01 4.56724733e-01
4.47272360e-01 2.53376156e-01 5.70573568e-01 -3.06436211e-01
-6.54608905e-01 1.05290964e-01 7.65771493e-02 3.02617073e-01
1.21492529e+00 1.12690158e-01 -2.69285887e-01 5.45130014e-01
5.87393641e-01 -1.45903835e-02 -7.74919271e-01 -2.40777418e-01
6.66546404e-01 -3.85107279e-01 -5.79768836e-01 -1.50672507e+00
-7.32546210e-01 4.19108272e-01 1.53122574e-01 5.72655678e-01
8.73786092e-01 9.02095065e-03 3.90269637e-01 5.61544776e-01
8.02004576e-01 -9.04796362e-01 -2.75481403e-01 4.78060752e-01
1.10820711e+00 -1.10855734e+00 -3.16224843e-02 -9.61459517e-01
-6.16311371e-01 8.75916958e-01 9.28361297e-01 5.21811128e-01
1.04557417e-01 3.50885302e-01 -8.82627815e-02 -4.11020428e-01
-1.05181503e+00 -4.84404534e-01 2.70414382e-01 6.09769642e-01
6.36520863e-01 8.38269368e-02 -2.77260751e-01 7.79111087e-01
-6.41854286e-01 -1.72595665e-01 3.51869643e-01 8.07330549e-01
-1.33193865e-01 -1.24391115e+00 -2.10336983e-01 4.43874836e-01
-3.86880934e-01 -1.40762389e-01 -9.88498032e-01 1.23277426e+00
1.48643091e-01 6.72785699e-01 -1.31338090e-02 -4.90051508e-01
7.68028140e-01 2.36190140e-01 2.85499990e-01 -6.16773427e-01
-8.27343404e-01 -6.67855144e-01 7.56803393e-01 -6.86441004e-01
-2.36861959e-01 -6.01073980e-01 -1.33934402e+00 -5.68049431e-01
2.65284926e-01 4.51275021e-01 6.18807018e-01 3.06119680e-01
5.19132197e-01 6.14243209e-01 -2.65565127e-01 -1.82903334e-01
-3.98004800e-02 -1.03250027e+00 -3.45793366e-01 7.92650819e-01
-3.01901132e-01 -9.93721068e-01 -2.08870564e-02 2.41464511e-01]
|
[9.364334106445312, 7.9934539794921875]
|
7ef8d147-848f-4511-bc00-99ad7e34ceb6
|
low-power-neuromorphic-emg-gesture
|
2206.02061
| null |
https://arxiv.org/abs/2206.02061v1
|
https://arxiv.org/pdf/2206.02061v1.pdf
|
Low Power Neuromorphic EMG Gesture Classification
|
EMG (Electromyograph) signal based gesture recognition can prove vital for applications such as smart wearables and bio-medical neuro-prosthetic control. Spiking Neural Networks (SNNs) are promising for low-power, real-time EMG gesture recognition, owing to their inherent spike/event driven spatio-temporal dynamics. In literature, there are limited demonstrations of neuromorphic hardware implementation (at full chip/board/system scale) for EMG gesture classification. Moreover, most literature attempts exploit primitive SNNs based on LIF (Leaky Integrate and Fire) neurons. In this work, we address the aforementioned gaps with following key contributions: (1) Low-power, high accuracy demonstration of EMG-signal based gesture recognition using neuromorphic Recurrent Spiking Neural Networks (RSNN). In particular, we propose a multi-time scale recurrent neuromorphic system based on special double-exponential adaptive threshold (DEXAT) neurons. Our network achieves state-of-the-art classification accuracy (90%) while using ~53% lesser neurons than best reported prior art on Roshambo EMG dataset. (2) A new multi-channel spike encoder scheme for efficient processing of real-valued EMG data on neuromorphic systems. (3) Unique multi-compartment methodology to implement complex adaptive neurons on Intel's dedicated neuromorphic Loihi chip is shown. (4) RSNN implementation on Loihi (Nahuku 32) achieves significant energy/latency benefits of ~983X/19X compared to GPU for batch size as 50.
|
['Manan Suri', 'Ahmed Shaban', 'Sai Sukruth Bezugam']
|
2022-06-04
| null | null | null | null |
['gesture-recognition', 'emg-gesture-recognition']
|
['computer-vision', 'medical']
|
[ 7.51001656e-01 -5.79134345e-01 1.74473599e-01 1.51270509e-01
-4.59882170e-01 -1.16927877e-01 -1.08192839e-01 -5.71794748e-01
-8.65216255e-01 8.35425794e-01 -1.27807930e-01 1.68313339e-01
-8.56544226e-02 -3.59765887e-01 -8.45416248e-01 -9.52621877e-01
-1.54845119e-01 -1.90296564e-02 4.22259331e-01 -1.99036032e-01
3.36713523e-01 4.18340564e-01 -1.84574699e+00 4.86969501e-01
2.97577977e-01 1.31993425e+00 4.07834858e-01 8.17355812e-01
4.19122010e-01 1.50634348e-01 -5.69950521e-01 5.10216773e-01
1.35459051e-01 -5.87771952e-01 4.51613069e-02 -8.44927371e-01
-1.65604010e-01 -1.70701984e-02 -6.50612235e-01 7.52593577e-01
1.03368783e+00 -1.84696913e-01 6.30751789e-01 -1.06164002e+00
-3.75114083e-01 5.14595628e-01 -2.13601485e-01 6.73793077e-01
1.05086543e-01 7.44782388e-01 4.06182334e-02 -6.02154553e-01
7.04971492e-01 5.11800170e-01 1.17320836e+00 1.15565109e+00
-1.01114845e+00 -1.20086348e+00 -7.23183930e-01 2.58054584e-01
-1.25985289e+00 -3.15972596e-01 4.45095032e-01 2.38498539e-01
2.06315255e+00 2.34313399e-01 1.21159160e+00 1.65228128e+00
1.17926848e+00 5.51423907e-01 1.54472315e+00 1.52792618e-01
5.83771050e-01 -6.63119555e-01 3.82249922e-01 1.92300737e-01
2.83652186e-01 8.43071565e-02 -1.32905829e+00 7.52161723e-03
1.17374313e+00 1.72973827e-01 -3.39087471e-02 6.98498130e-01
-9.83982205e-01 2.11922508e-02 2.11347967e-01 4.23176020e-01
-9.41700578e-01 1.16388416e+00 6.71303213e-01 3.94927055e-01
-2.37002835e-01 2.59689480e-01 -2.35181704e-01 -1.15828860e+00
-7.71653771e-01 2.48968974e-01 5.81587076e-01 5.52178860e-01
-1.74541660e-02 6.70099318e-01 5.05741127e-02 6.72750652e-01
9.13634226e-02 8.20362389e-01 1.13735878e+00 -8.92134547e-01
9.47893690e-03 5.06011128e-01 -4.97591674e-01 -1.59464851e-01
-8.34332168e-01 -2.85781007e-02 -1.03130174e+00 5.44468939e-01
1.29369944e-01 -1.55833170e-01 -1.07693732e+00 1.45060599e+00
-3.10748667e-01 4.12459403e-01 1.05154172e-01 1.18407297e+00
6.78652465e-01 5.11652350e-01 7.32531548e-02 -1.65697977e-01
1.45675623e+00 -1.00228310e-01 -7.66085804e-01 3.38971615e-02
2.49286033e-02 -4.91017736e-02 9.40759897e-01 5.03362715e-01
-1.25578785e+00 -6.68668896e-02 -1.16199267e+00 1.17667608e-01
-1.74421415e-01 5.77156916e-02 6.03575647e-01 6.43109620e-01
-1.07244837e+00 9.55832779e-01 -1.72036254e+00 -6.65916622e-01
4.23488885e-01 1.36157131e+00 -4.01646569e-02 8.08352351e-01
-8.33828926e-01 5.56715429e-01 -7.66625777e-02 1.85948819e-01
-6.86249197e-01 -4.68795270e-01 -8.08518678e-02 -1.80110648e-01
-3.19690824e-01 -6.48383498e-01 8.38939428e-01 -7.09347308e-01
-2.04356956e+00 7.90763438e-01 -2.76563793e-01 -9.51486886e-01
-2.38936707e-01 1.56919315e-01 -2.75923342e-01 1.16066493e-01
-4.06315714e-01 6.46428466e-01 5.71745038e-01 -1.51734129e-01
-7.28902891e-02 -7.84587622e-01 -7.64977932e-01 -6.43637870e-03
-5.58798611e-01 2.98880994e-01 2.91536987e-01 -9.14533257e-01
2.26787299e-01 -1.20607710e+00 3.72809730e-02 1.99481666e-01
1.57144889e-01 -2.73449004e-01 7.89777100e-01 -3.60796392e-01
1.16214287e+00 -2.21154714e+00 9.76023078e-02 6.38670400e-02
-2.00001076e-01 3.92753392e-01 -6.48219604e-03 3.14213693e-01
1.07257694e-01 -5.28890967e-01 -3.31377238e-01 3.79038364e-01
-3.06145132e-01 4.18390185e-01 -1.15745395e-01 4.78023708e-01
1.30634457e-01 1.50879836e+00 -2.26781368e-01 1.18024804e-01
-9.58685651e-02 6.86272383e-01 -1.82579041e-01 -2.12593228e-01
3.69139791e-01 2.91572064e-01 -1.90178707e-01 1.12386727e+00
1.04154110e-01 1.57034457e-01 -3.92786227e-02 -2.92810589e-01
-2.36642525e-01 4.53122884e-01 -1.02079642e+00 2.07120490e+00
-5.95108531e-02 7.71522224e-01 5.59795387e-02 -7.17853129e-01
9.77443993e-01 5.02962291e-01 6.57765448e-01 -1.05701697e+00
4.99516994e-01 8.69255781e-01 1.98011503e-01 -5.86071014e-01
-2.11107116e-02 -2.42981449e-01 -1.31441861e-01 5.93266845e-01
3.33535165e-01 3.05558622e-01 -2.66655177e-01 -5.61295927e-01
1.84445012e+00 4.45985854e-01 -1.97998792e-01 -3.96979690e-01
-2.82986790e-01 4.06985916e-02 5.08958817e-01 6.49672925e-01
-3.73487711e-01 4.50623482e-01 -1.61728740e-01 -2.56829262e-01
-7.88704634e-01 -1.22195554e+00 -1.42493740e-01 7.91895688e-01
1.27836823e-01 8.08872432e-02 -8.73929143e-01 6.75162017e-01
-6.57463297e-02 5.78769445e-02 -3.14782768e-01 -2.22928002e-01
-1.18458509e+00 -9.37072933e-01 1.46678483e+00 1.04915571e+00
4.17172730e-01 -1.75173318e+00 -1.74265063e+00 7.92721272e-01
3.77122730e-01 -7.81705797e-01 -7.10119382e-02 9.92208540e-01
-1.47439098e+00 -5.20164907e-01 -7.32469440e-01 -8.39089155e-01
1.34988904e-01 -3.33578497e-01 1.37521654e-01 -4.30648506e-01
-9.04566586e-01 3.66041690e-01 6.73890486e-02 -5.36315739e-01
1.95136935e-01 -8.16645566e-03 5.44300199e-01 -6.63614392e-01
8.16752195e-01 -1.41015625e+00 -7.79805362e-01 -5.56270480e-02
-6.41353428e-01 -8.81498754e-02 7.71158695e-01 9.51206326e-01
8.92817259e-01 -7.13698268e-01 7.68796980e-01 -1.38741866e-01
9.39427555e-01 -1.72287896e-01 1.17242679e-01 -4.28280905e-02
-7.42661595e-01 1.44320995e-01 6.27035737e-01 -9.81391788e-01
-5.64248860e-01 1.89380035e-01 -2.12883607e-01 -3.54473472e-01
3.33624855e-02 1.77519754e-01 7.80235410e-01 -3.85510683e-01
9.20235515e-01 6.58324122e-01 4.45060879e-01 -1.61692977e-01
-5.92833221e-01 1.10109878e+00 9.84312892e-01 -3.97890896e-01
-1.55539632e-01 4.86291021e-01 2.69750386e-01 -7.99442708e-01
8.26421738e-01 -2.76348293e-01 -2.48179451e-01 -3.08148056e-01
6.57940686e-01 -7.89776325e-01 -1.41619265e+00 1.04577875e+00
-9.51111019e-01 -7.10759878e-01 -3.01959604e-01 6.29899621e-01
-8.85867417e-01 -1.32076874e-01 -1.42229712e+00 -1.28830159e+00
-1.31169426e+00 -8.67592812e-01 9.62589681e-01 5.39370418e-01
-6.27023399e-01 -1.05396986e-01 1.30136684e-01 -1.80466950e-01
8.10629964e-01 2.31637225e-01 3.67060453e-01 -1.09562464e-01
-2.05928043e-01 -4.19549756e-02 1.29969418e-01 -1.29165500e-01
-1.56121805e-01 -3.88193160e-01 -9.60772514e-01 -1.13467850e-01
2.99097896e-01 -4.86459702e-01 7.56519914e-01 6.44014537e-01
8.46791327e-01 9.55337062e-02 -4.74850208e-01 5.88542640e-01
1.53142464e+00 5.22284627e-01 9.17785227e-01 1.47925064e-01
2.82983512e-01 -1.00983523e-01 1.97355486e-02 3.01664144e-01
-9.45571885e-02 7.26518095e-01 4.96770926e-02 3.94743860e-01
-4.62370545e-01 1.49162143e-01 9.59239364e-01 1.24192548e+00
-7.46798277e-01 2.17041880e-01 -5.43883204e-01 3.82838517e-01
-1.76404536e+00 -8.25078487e-01 -3.55958670e-01 2.15918064e+00
8.78183007e-01 -8.88522528e-03 3.00392687e-01 2.35960960e-01
4.94460225e-01 -5.93942046e-01 -1.27417302e+00 -7.74575770e-01
-4.49820727e-01 1.38048720e+00 7.50778019e-01 -4.67028707e-01
-4.06736374e-01 6.40737712e-01 5.70363379e+00 8.10926855e-01
-1.54654408e+00 4.84246939e-01 -1.56759739e-01 -1.02383792e+00
2.12632701e-01 -5.94780505e-01 -7.92269886e-01 7.73435831e-01
1.78679550e+00 1.16644964e-01 8.74518037e-01 3.02003473e-01
2.01802433e-01 -1.19189516e-01 -6.89202130e-01 1.42167771e+00
-1.39095053e-01 -1.41355824e+00 -4.21001554e-01 1.87813267e-01
2.17996985e-01 6.47962153e-01 -6.64976910e-02 5.62599674e-02
-5.73363841e-01 -1.18613040e+00 4.87045884e-01 6.69498563e-01
1.25433409e+00 -5.88816762e-01 5.02567589e-01 9.89863276e-02
-1.30738306e+00 -2.31773719e-01 -3.08715910e-01 -3.41667950e-01
1.88396975e-01 -5.10892943e-02 -2.11334601e-02 -2.50434250e-01
1.20151258e+00 6.66590631e-01 9.67946723e-02 7.84652293e-01
4.60905105e-01 9.30114329e-01 -1.02072787e+00 -9.45201874e-01
2.18259804e-02 1.22708455e-02 5.61898828e-01 1.31534874e+00
7.01990545e-01 5.36104739e-01 -9.74502444e-01 9.70139623e-01
2.75848825e-02 -4.71521229e-01 -5.66151798e-01 -3.07370752e-01
4.07756031e-01 9.31947172e-01 -8.31146419e-01 1.42874852e-01
9.58074108e-02 1.35307455e+00 -1.76570937e-01 1.14093043e-01
-6.65720403e-01 -7.72606850e-01 7.98487723e-01 -8.57782550e-03
5.05766540e-04 -4.42912936e-01 -9.85989571e-01 -7.26906359e-01
4.79477823e-01 -5.89692354e-01 -3.41858603e-02 -7.60228038e-01
-7.79303193e-01 4.10412401e-01 -7.11447835e-01 -1.22252119e+00
-3.99873435e-01 -1.06874311e+00 -5.29097617e-01 5.54087877e-01
-7.42245078e-01 -7.62486637e-01 -9.81881469e-02 5.86023450e-01
6.53583229e-01 -9.38144550e-02 1.23897040e+00 1.92947522e-01
-4.17596042e-01 7.08084822e-01 1.83672696e-01 -2.41035208e-01
2.01480687e-01 -6.76305294e-01 4.99621242e-01 4.37453538e-01
-5.51546514e-02 8.28220308e-01 4.08672988e-01 -7.76602983e-01
-2.37314963e+00 -5.37857473e-01 4.23883706e-01 -3.27282073e-03
6.83558404e-01 -5.11584461e-01 -1.03786051e+00 3.72167602e-02
1.04862764e-01 -1.60382748e-01 6.02330685e-01 -5.90358138e-01
1.56768367e-01 -2.24891543e-01 -1.29336834e+00 7.72073448e-01
1.56220937e+00 -5.69291890e-01 -4.46716398e-01 -2.48474367e-02
-1.07867764e-02 -3.73934627e-01 -1.07773829e+00 5.48047841e-01
1.46745121e+00 -5.98907530e-01 6.03684187e-01 -2.02399120e-01
1.42733932e-01 -1.41291633e-01 5.20353951e-03 -6.79748654e-01
1.09577917e-01 -9.19384539e-01 -4.48321223e-01 4.89129692e-01
1.26914889e-01 -9.42878783e-01 1.00984263e+00 5.22923231e-01
-4.45979744e-01 -1.40998948e+00 -1.81932378e+00 -1.03199875e+00
-3.72514009e-01 -6.37601316e-01 -6.61251321e-02 -1.19200490e-01
7.24905193e-01 -6.42286316e-02 -9.26488414e-02 -4.44676459e-01
3.87039810e-01 -2.24798799e-01 -2.27047727e-02 -6.89327776e-01
-3.45982015e-01 -4.96880949e-01 -1.15008771e+00 -7.92644203e-01
-2.85016894e-01 -9.28815663e-01 3.49272192e-01 -1.13809073e+00
2.88487762e-01 -1.33348525e-01 -7.41345406e-01 7.75739610e-01
5.41439712e-01 9.61184323e-01 -2.06836477e-01 4.65717643e-01
-1.34751573e-01 2.59608507e-01 5.64449668e-01 1.56288683e-01
-3.00002396e-01 -2.20865801e-01 2.22510956e-02 2.37192631e-01
7.51405001e-01 -8.72355819e-01 1.34045318e-01 -2.60102600e-01
-2.08096560e-02 2.14693084e-01 5.04413784e-01 -1.74197137e+00
9.14368451e-01 3.41480434e-01 3.94006640e-01 -3.00342739e-01
7.08833337e-01 -4.30728734e-01 5.66035926e-01 1.15934908e+00
-8.73669684e-02 3.46535951e-01 3.67057264e-01 4.51642841e-01
1.96738988e-01 -4.51958142e-02 4.90688801e-01 -2.37937793e-02
-6.63448334e-01 -8.65151137e-02 -1.15895164e+00 -4.81256634e-01
8.25062692e-01 -1.26442206e+00 -3.53029042e-01 3.70412111e-01
-6.68060482e-01 -4.11202103e-01 3.37048650e-01 2.61241138e-01
9.75689292e-01 -1.22141433e+00 -2.90007204e-01 4.97540683e-01
-1.34614691e-01 -7.84378111e-01 2.20712632e-01 1.16838324e+00
-5.03859222e-01 5.92345476e-01 -1.09476435e+00 -7.32538819e-01
-1.25725126e+00 -3.55223209e-01 2.86441237e-01 2.05259860e-01
-9.96343195e-01 8.54303598e-01 -8.04828882e-01 5.65581247e-02
2.53903627e-01 -5.21848738e-01 3.08418363e-01 -4.95527953e-01
3.44348520e-01 6.49412572e-01 2.01759100e-01 -1.10389069e-01
-6.53988242e-01 7.67562807e-01 4.92775261e-01 -5.13343453e-01
1.85999167e+00 4.28029865e-01 -1.04065828e-01 1.10683060e+00
8.32136393e-01 -8.81712675e-01 -1.23965693e+00 4.92811888e-01
-1.18381761e-01 4.25001860e-01 -3.24497163e-01 -1.07996297e+00
-7.90373981e-01 9.41890776e-01 1.48283863e+00 -6.47909582e-01
1.47858751e+00 -4.42297816e-01 1.62944043e+00 6.45734489e-01
9.75826561e-01 -1.47881210e+00 -5.22485375e-02 2.22452521e-01
7.06475019e-01 -4.50486511e-01 -3.47692043e-01 1.46323100e-01
-4.27937716e-01 1.53653288e+00 5.08748412e-01 -9.17400420e-01
4.74766582e-01 9.72497642e-01 -1.31706998e-01 -3.79343033e-01
-8.70449543e-01 1.58318400e-01 -9.64592025e-02 6.63754880e-01
4.15690899e-01 2.52592355e-01 -8.90778840e-01 1.08421123e+00
9.99557897e-02 8.36863875e-01 3.08913201e-01 1.17634058e+00
-3.41141939e-01 -6.00618422e-01 7.22725177e-04 1.07619703e+00
-7.16976106e-01 -1.29923120e-01 -1.26283631e-01 2.35912621e-01
-2.03862190e-01 4.49516833e-01 2.50324190e-01 -7.77624011e-01
2.44115770e-01 2.33890533e-01 8.96496177e-01 -2.50537515e-01
-1.40599012e+00 2.10983172e-01 -3.00962120e-01 -9.18335378e-01
-2.22992092e-01 -6.81290805e-01 -2.27027750e+00 1.02791175e-01
-7.56558478e-02 -5.64111352e-01 1.43862796e+00 9.46172297e-01
8.91294420e-01 5.61670363e-01 -1.47351325e-01 -1.03496039e+00
-4.96660799e-01 -1.18172336e+00 -7.88264513e-01 4.06974778e-02
-3.41653734e-01 -4.87120688e-01 -1.29555359e-01 -1.32818863e-01]
|
[8.386691093444824, 2.3188066482543945]
|
b7329a88-d87e-4e33-b371-7fd135d371f4
|
interpretability-and-explainability-a-machine
|
2012.01805
| null |
https://arxiv.org/abs/2012.01805v2
|
https://arxiv.org/pdf/2012.01805v2.pdf
|
Interpretability and Explainability: A Machine Learning Zoo Mini-tour
|
In this review, we examine the problem of designing interpretable and explainable machine learning models. Interpretability and explainability lie at the core of many machine learning and statistical applications in medicine, economics, law, and natural sciences. Although interpretability and explainability have escaped a clear universal definition, many techniques motivated by these properties have been developed over the recent 30 years with the focus currently shifting towards deep learning methods. In this review, we emphasise the divide between interpretability and explainability and illustrate these two different research directions with concrete examples of the state-of-the-art. The review is intended for a general machine learning audience with interest in exploring the problems of interpretation and explanation beyond logistic regression or random forest variable importance. This work is not an exhaustive literature survey, but rather a primer focusing selectively on certain lines of research which the authors found interesting or informative.
|
['Julia E. Vogt', 'Ričards Marcinkevičs']
|
2020-12-03
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 5.52193403e-01 9.65974331e-01 -8.31753314e-01 -7.55929649e-01
-1.68042585e-01 -2.99908251e-01 4.09493357e-01 2.81489909e-01
4.15073521e-02 8.17691922e-01 3.32430989e-01 -6.85434759e-01
-6.53971255e-01 -4.37191367e-01 -3.59476715e-01 -5.20548582e-01
-1.18919052e-01 7.45472431e-01 -8.82474482e-01 1.82271842e-02
3.57067883e-01 2.75029331e-01 -1.37932897e+00 4.87632960e-01
1.03500044e+00 7.19632924e-01 -2.03214362e-01 4.34829414e-01
-2.51792729e-01 7.84683228e-01 -3.74713242e-01 -6.63081527e-01
-2.18718410e-01 -6.90342486e-01 -1.19974577e+00 4.77317795e-02
1.16232611e-01 -4.82054390e-02 -1.87902108e-01 4.39488173e-01
4.43060920e-02 -2.19312668e-01 8.51148188e-01 -1.53199494e+00
-1.49147475e+00 7.90231407e-01 -2.86176115e-01 3.17271680e-01
1.48288578e-01 -7.12225866e-03 1.53964436e+00 -7.23964930e-01
1.93843588e-01 1.29500926e+00 6.03769302e-01 8.06009889e-01
-1.06374586e+00 -3.76703411e-01 5.16012907e-01 4.98332858e-01
-7.19027936e-01 -7.55147310e-03 5.77758312e-01 -5.20718634e-01
9.08231735e-01 7.03238487e-01 9.00429010e-01 1.10146654e+00
4.42435533e-01 1.03115928e+00 1.12610304e+00 -6.41014457e-01
1.95743307e-01 -5.12362607e-02 7.51241684e-01 8.07958603e-01
6.14733875e-01 3.25106442e-01 -5.55455327e-01 1.91992018e-02
8.64711463e-01 3.34111035e-01 -1.70778751e-01 -3.23426992e-01
-1.28174818e+00 1.34694457e+00 5.74142992e-01 3.04499239e-01
-2.73763686e-01 3.07323277e-01 1.78580254e-01 3.70320767e-01
5.66328526e-01 8.67604017e-01 -7.05473840e-01 1.08007044e-01
-6.82757258e-01 7.31882453e-02 8.16951096e-01 6.17801368e-01
5.39168000e-01 2.35580400e-01 3.94014269e-02 3.33224177e-01
4.14722621e-01 5.41857183e-01 2.11825728e-01 -6.75582647e-01
1.87652305e-01 7.45225549e-01 -1.53450727e-01 -9.39276516e-01
-6.73318267e-01 -4.30750072e-01 -1.16254306e+00 6.24770671e-02
1.76719308e-01 7.37329423e-02 -8.62306058e-01 1.34779716e+00
-5.58443516e-02 -3.38173777e-01 -2.04679891e-01 8.99662316e-01
1.20065844e+00 2.02117056e-01 3.78800541e-01 -1.77492812e-01
1.36444008e+00 -1.07624698e+00 -1.06220853e+00 -4.24460173e-01
3.71155560e-01 -4.34510440e-01 1.16712141e+00 3.00106138e-01
-1.02354693e+00 -3.92371148e-01 -8.01158667e-01 -5.20187318e-01
-4.67629433e-01 -9.64296982e-02 1.48990059e+00 5.75469613e-01
-7.22179115e-01 5.48294902e-01 -9.16968882e-01 -3.01822126e-01
6.00063086e-01 5.07997394e-01 -3.58363003e-01 2.85721689e-01
-1.16413224e+00 1.24643743e+00 2.13440205e-03 1.40730619e-01
-3.34229857e-01 -4.91941482e-01 -9.08590615e-01 1.92251191e-01
1.84785053e-01 -1.15841341e+00 1.23691452e+00 -1.42181468e+00
-1.09228897e+00 1.12019050e+00 -6.34810805e-01 -6.04026198e-01
3.38982522e-01 -4.65323895e-01 -2.38589987e-01 -3.37627143e-01
-1.06893122e-01 4.71259892e-01 4.99443233e-01 -1.03335094e+00
-5.94992340e-01 -4.03508693e-01 2.47993618e-01 7.74335116e-02
6.53963089e-02 -1.25812367e-01 2.46638551e-01 -6.65587187e-01
2.13181674e-01 -8.79616737e-01 -1.66885197e-01 8.24523270e-02
-5.52253008e-01 -4.42266703e-01 3.26588035e-01 -4.84265268e-01
1.46695065e+00 -1.60541523e+00 3.31280291e-01 5.47450455e-03
8.52676034e-01 -1.44895688e-01 -7.02542998e-03 3.22654814e-01
-5.93068421e-01 5.88620126e-01 -4.17412579e-01 -2.42436275e-01
1.50384694e-01 4.34418380e-01 -4.05788630e-01 5.95795095e-01
3.15863848e-01 1.46400988e+00 -9.55510020e-01 -1.12723514e-01
4.56451595e-01 3.98238569e-01 -2.89750695e-01 -9.68834758e-02
2.95004062e-02 5.29211044e-01 -6.55203640e-01 6.94015265e-01
2.72492796e-01 -6.81747198e-01 1.94973662e-01 1.75620094e-01
1.01541512e-01 5.73571980e-01 -6.71905696e-01 9.95336473e-01
-4.52885389e-01 1.21911550e+00 -4.76963699e-01 -1.29362595e+00
7.16904223e-01 3.65495145e-01 1.71253413e-01 -3.70847136e-01
4.98350710e-02 4.46885020e-01 3.47044498e-01 -6.35038495e-01
2.29901731e-01 -4.74105418e-01 7.03963190e-02 5.74288130e-01
-4.63667542e-01 -8.09926689e-02 -2.48832449e-01 -4.33314621e-01
7.04832554e-01 -2.52077393e-02 1.22435927e+00 -1.88686311e-01
6.41317487e-01 -5.82792833e-02 1.17923617e-01 8.87694299e-01
-1.47423074e-01 5.96325159e-01 4.72956449e-01 -1.38021326e+00
-1.01973414e+00 -1.01128197e+00 -3.44231606e-01 9.84299719e-01
1.54769793e-02 -1.32921949e-01 -6.30439699e-01 -8.69408965e-01
3.00789326e-02 8.84938896e-01 -1.18931091e+00 -6.18711859e-02
-6.30051672e-01 -7.51458645e-01 1.35995179e-01 7.61577666e-01
1.67769834e-01 -1.35427332e+00 -6.91167831e-01 -2.06309840e-01
-3.78599524e-01 -6.27311170e-01 5.26879728e-02 4.66647804e-01
-1.14488602e+00 -1.21180403e+00 -3.18641692e-01 -6.23826385e-01
8.56326342e-01 2.73392916e-01 1.61624134e+00 5.81529319e-01
-1.11714885e-01 3.68300110e-01 -2.70247370e-01 -9.18891847e-01
-3.79183531e-01 5.42816699e-01 -7.42496550e-02 -5.21980226e-01
9.01496172e-01 -2.60034829e-01 -5.52294075e-01 1.98781386e-01
-9.72829819e-01 3.82809371e-01 6.27477169e-01 1.13007522e+00
5.00156462e-01 -4.87952143e-01 7.38766193e-01 -1.13861549e+00
5.57371914e-01 -6.62640035e-01 -3.56278419e-02 2.78586805e-01
-1.03845882e+00 1.72817215e-01 5.44599354e-01 -1.07604958e-01
-5.71354270e-01 -2.35875949e-01 -1.25291860e-02 3.45767140e-01
-3.45073611e-01 7.21998930e-01 -5.18689817e-03 2.72914201e-01
6.54281020e-01 -8.39886740e-02 2.60069389e-02 -3.57652307e-01
4.82371211e-01 7.90552020e-01 1.14844618e-02 -9.03952718e-02
7.99856901e-01 5.72598040e-01 1.53658003e-01 -5.54358661e-01
-1.27506840e+00 -9.18736011e-02 -8.76699746e-01 -8.28657392e-03
7.62473464e-01 -3.63276094e-01 -6.93634152e-01 -1.39517501e-01
-1.24748242e+00 1.30465925e-02 -5.78793705e-01 4.25287753e-01
-7.71021485e-01 1.82872102e-01 -2.76468247e-01 -7.44669378e-01
-5.37324429e-01 -1.20889115e+00 1.05309415e+00 2.03661636e-01
-1.16086483e+00 -1.65089333e+00 -1.21365495e-01 5.65604329e-01
2.19084084e-01 4.50938702e-01 1.31433654e+00 -9.32159185e-01
-6.08412683e-01 -3.09072107e-01 -4.86034714e-02 -1.26053980e-02
5.33439457e-01 -1.68548226e-01 -1.15724325e+00 2.76704967e-01
1.25988089e-02 2.25885883e-02 9.33153152e-01 8.73737097e-01
1.35023022e+00 -5.92292070e-01 -4.32707638e-01 5.43912172e-01
1.11865246e+00 1.24658169e-02 3.57164353e-01 5.36925077e-01
5.88018537e-01 8.00047457e-01 4.11050677e-01 1.42247900e-01
5.10031223e-01 3.08774889e-01 5.50544143e-01 -6.32364452e-01
-2.92666033e-02 -7.17579946e-02 -1.35806412e-01 3.92199218e-01
-5.32429278e-01 -3.89651775e-01 -9.38105702e-01 3.26091290e-01
-1.97543573e+00 -1.09111309e+00 -6.05345547e-01 1.98411119e+00
4.11960036e-01 -8.82425979e-02 -3.03494800e-02 3.15742522e-01
6.22712970e-01 1.42104954e-01 -5.98768830e-01 -8.56759906e-01
-2.16111615e-01 1.39017433e-01 1.89333364e-01 7.40824938e-01
-1.05922747e+00 6.29686832e-01 7.95430326e+00 3.22951823e-01
-1.03787065e+00 -2.72817284e-01 1.22257495e+00 1.16228759e-01
-8.62036526e-01 -1.01156741e-01 -2.02962548e-01 1.16508551e-01
7.20308483e-01 4.18428034e-02 4.98334378e-01 9.74939525e-01
6.29212677e-01 1.50195360e-01 -1.65776420e+00 7.45636523e-01
-7.42538646e-02 -1.42289746e+00 1.98922426e-01 2.15691045e-01
6.48592293e-01 -2.81752646e-01 5.09515822e-01 -4.09659855e-02
-4.26856168e-02 -1.90589738e+00 6.11248136e-01 4.97000575e-01
6.55085683e-01 -3.78878206e-01 9.02717412e-01 5.11100814e-02
-7.25669324e-01 -1.86999545e-01 -3.18981439e-01 -7.73168147e-01
-1.73610449e-02 6.80780828e-01 -6.30476177e-01 4.05207098e-01
5.36741316e-01 9.12224054e-01 -2.60788351e-01 8.61269057e-01
-6.85075581e-01 7.74378419e-01 1.12759151e-01 -2.70756602e-01
2.57457018e-01 -8.50056186e-02 4.85581309e-01 1.09541857e+00
-1.25226021e-01 6.16514497e-02 -4.29562598e-01 1.11180103e+00
2.28815436e-01 -8.53968486e-02 -5.62895179e-01 -8.89136046e-02
1.08594790e-01 1.01171124e+00 -7.43478298e-01 -1.89135864e-01
-6.14761114e-01 7.53117263e-01 2.67065316e-01 2.09587783e-01
-1.03764367e+00 6.60525262e-02 7.44584262e-01 2.16971904e-01
-8.89310166e-02 3.08371708e-02 -1.29576671e+00 -1.10678387e+00
-1.79116607e-01 -1.05683827e+00 5.14620721e-01 -5.94773114e-01
-1.40504742e+00 5.31402290e-01 1.86595097e-01 -8.22752178e-01
-3.78435493e-01 -8.68167341e-01 -5.13828754e-01 9.03763533e-01
-1.49502802e+00 -1.27089953e+00 -3.79190505e-01 -3.06034297e-01
7.79855192e-01 -8.22348595e-02 1.02565289e+00 -3.09087336e-01
-4.77760106e-01 3.64987314e-01 1.66001484e-01 -1.77721605e-01
2.12746486e-01 -1.51791644e+00 8.36205482e-01 1.23617604e-01
1.85541987e-01 1.01741064e+00 8.78754497e-01 -3.70124221e-01
-9.58366156e-01 -6.63661778e-01 1.54074478e+00 -9.28725660e-01
4.21967238e-01 -2.23653801e-02 -8.26769769e-01 1.14272475e+00
3.53726149e-01 -4.19051439e-01 1.10846615e+00 5.74273109e-01
-1.50843039e-01 1.11967474e-01 -1.01425290e+00 6.69855773e-01
1.02494669e+00 -2.17651457e-01 -8.17802072e-01 3.38056713e-01
4.80730385e-01 -3.65659148e-01 -5.40199518e-01 3.54311347e-01
9.85434532e-01 -9.20717895e-01 9.17162716e-01 -1.25662506e+00
8.16830993e-01 3.77935134e-02 3.20917100e-01 -1.09218609e+00
-7.24737108e-01 -5.98409176e-01 -1.90916106e-01 6.41596556e-01
6.66335523e-01 -7.51130521e-01 7.61982620e-01 9.48769569e-01
-7.58519247e-02 -1.25032508e+00 -7.62018263e-01 -2.62094766e-01
4.01108384e-01 -3.68797660e-01 8.82540226e-01 8.78056049e-01
2.79466122e-01 6.29631042e-01 -2.48519987e-01 -2.61657059e-01
3.08769941e-01 3.12435240e-01 5.56997359e-01 -1.38514972e+00
-2.93068346e-02 -7.20189154e-01 -3.07836682e-01 -9.65835512e-01
3.16077888e-01 -9.18891668e-01 -3.14481169e-01 -1.99144697e+00
5.72637320e-01 8.30439478e-02 -1.18394636e-01 6.85846508e-01
-6.01577818e-01 -1.61896124e-02 1.79976504e-02 2.88605690e-01
-3.44715029e-01 3.42722297e-01 1.31411636e+00 -3.15904886e-01
-1.48471389e-02 5.19839346e-01 -1.33360302e+00 1.06645370e+00
9.90511954e-01 -4.23072308e-01 -6.32082999e-01 -6.14686430e-01
5.12537241e-01 -2.73869097e-01 4.80454683e-01 -3.65119278e-01
-5.10482013e-01 -5.82860112e-01 5.57593465e-01 -2.37382218e-01
-7.50979632e-02 -9.28678274e-01 1.53002918e-01 6.98080361e-01
-8.27232957e-01 5.12601852e-01 6.07511178e-02 5.42351484e-01
-2.86340900e-02 -2.28589132e-01 6.78120136e-01 -1.56835198e-01
-4.71185505e-01 1.01875789e-01 -5.99587202e-01 -7.18536079e-02
9.08181608e-01 -4.59908634e-01 -2.17611507e-01 -7.38804877e-01
-7.30259061e-01 6.98087811e-02 1.62316605e-01 5.05209088e-01
6.67999804e-01 -9.70423996e-01 -7.03744888e-01 -1.09322235e-01
1.23063400e-01 -1.99034616e-01 1.91223864e-02 7.15133250e-01
-5.66587448e-01 1.07126725e+00 7.14609027e-02 -3.79015535e-01
-1.09964895e+00 5.54850161e-01 5.64793408e-01 -4.36132669e-01
-4.29094136e-01 5.28802335e-01 5.95909595e-01 -4.14625198e-01
1.62491426e-01 -7.22203553e-01 -3.65101248e-01 -3.69785279e-01
3.76717210e-01 5.90471327e-01 -2.76968509e-01 -3.45445335e-01
-4.50280428e-01 4.21626478e-01 1.73472285e-01 3.12170953e-01
1.44656599e+00 -3.26702833e-01 -2.51871586e-01 6.12696290e-01
7.85253227e-01 -3.75736117e-01 -6.10241294e-01 1.44605279e-01
1.03568070e-01 -1.01963833e-01 -2.66474783e-01 -8.95060003e-01
-8.75284076e-01 1.13652039e+00 2.68090099e-01 6.23590469e-01
9.99680936e-01 2.72412211e-01 4.14849043e-01 2.73680538e-01
-7.84995332e-02 -6.59729242e-01 -9.06442553e-02 4.96529222e-01
1.18675053e+00 -1.46498466e+00 4.38361406e-01 -4.76348281e-01
-6.61610484e-01 1.38010693e+00 4.27088141e-01 1.76917687e-01
6.43144250e-01 -9.66489464e-02 -5.30931391e-02 -3.67145419e-01
-5.23131371e-01 -3.09791882e-02 7.45352268e-01 7.54703164e-01
1.06600475e+00 2.29442939e-01 -5.84578633e-01 8.09752882e-01
-6.96395159e-01 4.11637574e-02 2.72576392e-01 3.59042019e-01
-4.67277825e-01 -9.53205943e-01 -3.77872497e-01 7.36767828e-01
-3.90288413e-01 -2.88411558e-01 -8.17234576e-01 1.19533503e+00
1.29858956e-01 9.73433018e-01 1.79301187e-01 -8.70328322e-02
-3.42723913e-02 -5.45950383e-02 2.75610685e-01 -4.57104892e-01
-6.18911564e-01 -4.25944984e-01 1.73963849e-02 -2.24606737e-01
-5.35947621e-01 -6.21244371e-01 -1.18927121e+00 -6.61376119e-01
-3.90262872e-01 -1.94730330e-02 4.17526364e-01 1.36807394e+00
1.00626431e-01 5.84537268e-01 2.46118188e-01 -3.25990021e-01
-3.77938002e-01 -8.61988366e-01 -4.34373677e-01 1.51652321e-01
8.55509102e-01 -5.03929138e-01 -4.65723157e-01 9.82092321e-02]
|
[8.761368751525879, 5.818778991699219]
|
ff1395e3-8168-4538-b4ed-481748df27f6
|
samplet-basis-pursuit
|
2306.10180
| null |
https://arxiv.org/abs/2306.10180v2
|
https://arxiv.org/pdf/2306.10180v2.pdf
|
Samplet basis pursuit
|
We consider kernel-based learning in samplet coordinates with l1-regularization. The application of an l1-regularization term enforces sparsity of the coefficients with respect to the samplet basis. Therefore, we call this approach samplet basis pursuit. Samplets are wavelet-type signed measures, which are tailored to scattered data. They provide similar properties as wavelets in terms of localization, multiresolution analysis, and data compression. The class of signals that can sparsely be represented in a samplet basis is considerably larger than the class of signals which exhibit a sparse representation in the single-scale basis. In particular, every signal that can be represented by the superposition of only a few features of the canonical feature map is also sparse in samplet coordinates. We propose the efficient solution of the problem under consideration by combining soft-shrinkage with the semi-smooth Newton method and compare the approach to the fast iterative shrinkage thresholding algorithm. We present numerical benchmarks as well as applications to surface reconstruction from noisy data and to the reconstruction of temperature data using a dictionary of multiple kernels.
|
['Helmut Harbrecht', 'Michael Multerer', 'Davide Baroli']
|
2023-06-16
| null | null | null | null |
['data-compression']
|
['time-series']
|
[ 4.09076869e-01 -7.11437762e-02 8.56914669e-02 -4.41696167e-01
-1.09552205e+00 -9.18635502e-02 3.73596758e-01 1.48053035e-01
-2.78103143e-01 5.84746897e-01 5.48483692e-02 3.15658927e-01
-3.33472461e-01 -6.39411986e-01 -6.52003229e-01 -1.24426425e+00
-2.41462648e-01 3.58626902e-01 4.90920171e-02 -2.37102151e-01
2.90834934e-01 4.78136510e-01 -1.42262197e+00 2.30583444e-01
6.86788917e-01 1.27311265e+00 1.56686336e-01 2.94468492e-01
9.12346542e-02 3.81053180e-01 -1.94509163e-01 4.41384465e-01
2.39607319e-01 -2.83439726e-01 -2.07165509e-01 2.94184178e-01
7.02502251e-01 2.04214275e-01 -9.76577625e-02 1.05929029e+00
3.10019016e-01 2.98521072e-01 6.75319493e-01 -7.37242103e-01
-3.80032867e-01 4.37421631e-03 -9.20543194e-01 9.49847475e-02
3.81799251e-01 -5.45577943e-01 7.73110092e-01 -1.48094594e+00
7.23442316e-01 1.15133321e+00 9.78493512e-01 -7.46328244e-03
-1.74719501e+00 1.20640390e-01 -3.21368098e-01 9.54523683e-02
-1.31854367e+00 -7.78140485e-01 1.07589293e+00 -2.75438994e-01
5.76225936e-01 7.05762506e-01 4.88082737e-01 5.46564519e-01
2.52287149e-01 6.48129046e-01 1.42149425e+00 -8.87875855e-01
6.24312222e-01 -8.42591077e-02 2.65152037e-01 7.92065084e-01
2.61690646e-01 -1.02412757e-02 -7.31211483e-01 -6.94726110e-01
8.19020927e-01 9.92976502e-02 -4.35618937e-01 -5.77887177e-01
-1.33850324e+00 1.17093205e+00 1.13901772e-01 6.97496772e-01
-6.78575337e-01 2.50782728e-01 4.20004487e-01 5.41346729e-01
1.08479691e+00 7.02303872e-02 -1.39144197e-01 3.21904987e-01
-1.35195696e+00 -8.06894619e-03 7.84275353e-01 5.64022362e-01
1.05664659e+00 4.92316306e-01 1.48126140e-01 7.64473021e-01
2.26383701e-01 1.06453407e+00 4.18779820e-01 -1.24965274e+00
-8.67653638e-02 1.21102326e-01 -7.33542368e-02 -1.22357857e+00
-1.75472602e-01 -2.10843667e-01 -1.07383537e+00 4.66205537e-01
3.75459969e-01 2.54419953e-01 -6.02228880e-01 1.39724731e+00
4.31491613e-01 5.59087753e-01 -1.66207440e-02 9.33336616e-01
5.31010091e-01 7.37267613e-01 -7.37559259e-01 -8.36487830e-01
1.02374041e+00 -2.13579535e-01 -8.47512782e-01 4.60422449e-02
5.11009276e-01 -9.97733414e-01 6.41342223e-01 6.29316509e-01
-1.14501250e+00 -3.68934333e-01 -7.97650278e-01 1.11062147e-01
8.26978311e-02 2.75206745e-01 5.32656074e-01 3.54761600e-01
-1.03837287e+00 8.00060868e-01 -9.98265803e-01 -1.68724880e-01
4.17444333e-02 -5.19083403e-02 -5.23048103e-01 -1.91865548e-01
-6.38771772e-01 9.42925692e-01 -2.24300921e-01 1.76788587e-02
-5.90938032e-01 -7.60811090e-01 -8.63295496e-01 -2.12975293e-01
8.43360871e-02 -6.49817362e-02 5.74004292e-01 -7.25308776e-01
-1.11870158e+00 7.55688965e-01 -6.43725812e-01 -2.23713353e-01
1.98445722e-01 1.30194187e-01 -1.22634985e-01 5.55386603e-01
2.83391446e-01 -1.85979754e-01 1.75679326e+00 -1.08984911e+00
-2.75970846e-01 -3.95129025e-01 -6.44433439e-01 -2.25171149e-01
-3.87765497e-01 -2.35863551e-01 1.76262647e-01 -9.38659608e-01
8.09777439e-01 -8.04717004e-01 -5.02340555e-01 8.61884952e-02
-2.21256427e-02 5.87717406e-02 1.20233059e+00 -7.88664997e-01
1.01953399e+00 -2.42242980e+00 5.05234718e-01 7.06331551e-01
2.28425324e-01 -3.94237190e-01 -5.34860753e-02 4.57230747e-01
-1.71223387e-01 -5.29052675e-01 -7.57041097e-01 -3.88749391e-01
-2.84981370e-01 3.37026864e-01 -5.41815877e-01 1.28504419e+00
-1.14682347e-01 3.46856147e-01 -6.97546005e-01 -4.12688285e-01
2.84781218e-01 7.06717789e-01 -2.10979894e-01 -1.14675343e-01
2.00532302e-01 3.57660174e-01 -6.00302219e-01 7.85045147e-01
7.91067839e-01 -3.20546553e-02 -1.63934946e-01 -5.48136532e-01
-4.52896535e-01 -3.73761892e-01 -1.49036419e+00 1.61817992e+00
-4.99593437e-01 6.49749160e-01 8.12983453e-01 -1.56601262e+00
1.21014798e+00 4.93439436e-01 9.58543837e-01 -1.18766539e-01
-1.66768447e-01 8.02904487e-01 -4.81023908e-01 -4.36971962e-01
1.07914709e-01 -5.06104946e-01 1.15273796e-01 3.62554103e-01
-6.90641776e-02 -4.38411176e-01 1.01835266e-01 4.54577692e-02
1.21548176e+00 -3.06118935e-01 4.25275922e-01 -8.05624366e-01
5.04047811e-01 -7.45979324e-02 4.46635455e-01 4.86068040e-01
1.70228854e-01 5.70993960e-01 7.93109983e-02 -6.15850687e-01
-1.10133839e+00 -6.27995491e-01 -6.73692107e-01 7.74070263e-01
-2.83323795e-01 -6.28919601e-02 -3.93872708e-01 -5.80160022e-02
2.67540783e-01 5.00695825e-01 -7.58784652e-01 5.79847172e-02
-6.84092999e-01 -7.80986130e-01 9.32093896e-03 1.37934700e-01
1.77544355e-01 -8.00451279e-01 -7.11236000e-01 2.14103758e-01
-1.32492436e-02 -9.07295585e-01 -5.07787704e-01 4.79196936e-01
-1.22380137e+00 -1.03481925e+00 -8.91081572e-01 -8.47110152e-01
8.74982774e-01 4.53105032e-01 8.36102188e-01 -2.06252798e-01
-5.24619043e-01 8.31666410e-01 -3.04240584e-01 -1.82952499e-03
-2.60590851e-01 -5.19586384e-01 3.55034113e-01 5.76579988e-01
5.95147647e-02 -8.88477743e-01 -5.99007197e-02 1.19206570e-01
-8.28089058e-01 -5.34837365e-01 3.07432085e-01 9.92212474e-01
1.08875155e+00 1.78230405e-01 3.75330716e-01 -5.05135179e-01
5.89182496e-01 -3.03146303e-01 -5.68532526e-01 2.03985378e-01
-2.62819737e-01 8.69916286e-03 5.37065566e-01 -4.74083513e-01
-7.91060269e-01 3.62531781e-01 3.23395431e-01 -7.19917774e-01
1.70052603e-01 7.49146461e-01 5.22940814e-01 -9.86432672e-01
8.84776890e-01 6.02676690e-01 2.66524345e-01 -6.29378676e-01
2.56660521e-01 2.50594109e-01 3.33671212e-01 -8.63960326e-01
7.98253059e-01 1.10481763e+00 5.75294018e-01 -1.71728182e+00
-6.69535697e-01 -8.57121289e-01 -5.72843671e-01 -1.88494891e-01
5.17889440e-01 -5.57131827e-01 -4.08228844e-01 1.52093127e-01
-9.90719438e-01 9.63942781e-02 -9.07184482e-01 6.85769916e-01
-1.00889289e+00 7.59894550e-01 -4.72940058e-01 -9.67629552e-01
-2.97969401e-01 -7.73514628e-01 1.30799651e+00 -3.87667149e-01
-1.00610321e-02 -1.20551682e+00 3.51896495e-01 -1.86239049e-01
6.09725237e-01 5.42764783e-01 6.72289670e-01 -3.27064753e-01
-2.29351148e-02 -5.50977707e-01 2.84247905e-01 5.71250081e-01
1.47709608e-01 -2.60966092e-01 -8.02520871e-01 -8.24488044e-01
1.03291476e+00 -3.96036416e-01 1.21333683e+00 1.00593519e+00
7.68674433e-01 -1.33877248e-01 -2.70375341e-01 7.35535383e-01
1.54716074e+00 -3.73170525e-01 1.34773865e-01 -2.29906105e-02
4.22904968e-01 6.63561225e-01 5.33499002e-01 5.57368159e-01
-2.35860154e-01 6.11882508e-01 2.63814628e-01 -6.08451441e-02
1.13380447e-01 4.41666007e-01 3.83986324e-01 1.13178325e+00
-4.17581528e-01 7.16650367e-01 -6.41625047e-01 6.09531224e-01
-1.87934291e+00 -1.07769084e+00 -5.76923430e-01 2.43623662e+00
6.45318925e-01 -1.84642136e-01 -1.47013098e-01 5.32502174e-01
7.48518229e-01 4.20066684e-01 -3.83260310e-01 -1.78120553e-01
-4.45164979e-01 5.60258329e-01 5.60297132e-01 8.70364010e-01
-1.07094073e+00 2.30328530e-01 6.92888546e+00 1.06368649e+00
-1.09049070e+00 2.47856468e-01 2.11363852e-01 1.50687978e-01
-3.17436010e-01 -9.46591236e-03 -4.71990615e-01 2.92797476e-01
7.00189769e-01 -1.72328830e-01 4.04103607e-01 9.25169051e-01
3.76878768e-01 -1.43746853e-01 -8.18336964e-01 9.82723415e-01
2.02950269e-01 -1.45303977e+00 -3.87169838e-01 7.01353475e-02
7.53679931e-01 9.73496139e-02 7.81766325e-02 -2.44742110e-01
-2.53938705e-01 -6.68975353e-01 6.39317155e-01 6.19769096e-01
8.76037240e-01 -4.79118705e-01 3.68329257e-01 3.37591738e-01
-1.30410099e+00 1.50633752e-01 -7.42101729e-01 3.46701480e-02
2.97802448e-01 1.34592354e+00 -4.01459873e-01 3.43157023e-01
6.47303760e-01 1.07742047e+00 2.58956961e-02 8.90073478e-01
1.80764779e-01 7.79879689e-01 -8.47645402e-01 3.01162511e-01
9.51157436e-02 -8.11422348e-01 1.00770581e+00 1.06479657e+00
8.52683365e-01 2.64771163e-01 6.92652643e-01 5.03379762e-01
4.00749236e-01 1.98790923e-01 -7.02550471e-01 3.19626749e-01
2.91867167e-01 1.23048246e+00 -6.45280361e-01 -3.75823259e-01
-5.59890866e-01 6.52667105e-01 -1.15313649e-01 5.58858693e-01
-2.00072629e-03 -2.64798164e-01 3.82246941e-01 4.47490692e-01
5.46866715e-01 -4.85131294e-01 -4.01738256e-01 -1.20467281e+00
2.00888410e-01 -6.96179092e-01 1.56829625e-01 -4.91943359e-01
-1.42052448e+00 4.81878221e-01 7.64704496e-02 -1.20829380e+00
9.47104096e-02 -5.53214073e-01 -7.88768291e-01 8.53043735e-01
-1.48875666e+00 -1.01829171e+00 -5.65027893e-02 7.90448785e-01
2.95454443e-01 -1.25989720e-01 1.02798045e+00 -1.10913694e-01
2.46505602e-04 -2.16553986e-01 7.69705057e-01 -3.78072143e-01
4.42533612e-01 -1.23767185e+00 -1.42454728e-01 6.39217436e-01
-7.82329217e-03 3.53689581e-01 9.73851264e-01 -6.19147897e-01
-1.56398094e+00 -6.64018035e-01 7.32593715e-01 -1.49236415e-02
8.28282177e-01 7.03346636e-03 -1.23110580e+00 6.23126745e-01
-1.76032662e-01 7.67200232e-01 5.65380692e-01 -1.83641732e-01
-2.81020224e-01 -2.00485975e-01 -1.32916021e+00 -2.20459908e-01
2.87070394e-01 -4.99631137e-01 -6.65992379e-01 7.28921056e-01
7.15883821e-02 -2.03952312e-01 -1.15310645e+00 2.99714565e-01
2.86638290e-01 -7.57605314e-01 1.09440494e+00 -2.75824338e-01
-6.83943704e-02 -3.86847734e-01 -7.04762816e-01 -1.32388902e+00
-5.58157563e-01 -8.40886474e-01 -2.67786711e-01 4.92841631e-01
-1.11289263e-01 -6.24884546e-01 8.39600444e-01 8.82050321e-02
-3.13697249e-01 -6.36914611e-01 -1.71606731e+00 -9.56612647e-01
-1.16159454e-01 -2.05652967e-01 -1.80611446e-01 1.24336576e+00
1.75567597e-01 -1.18159696e-01 -5.35282850e-01 8.19187537e-02
1.51021552e+00 4.15509164e-01 1.07197322e-01 -1.54364824e+00
-1.73838109e-01 -3.88247781e-02 -4.60526019e-01 -6.19069874e-01
1.57912374e-01 -9.88225579e-01 1.78205371e-01 -1.06909263e+00
-1.45160109e-01 -4.15974170e-01 3.25095616e-02 2.10197166e-01
3.96407932e-01 2.06465647e-01 -1.90242201e-01 5.72066307e-01
-1.15568943e-01 5.58920622e-01 1.15043545e+00 -7.51705244e-02
-4.11465578e-02 -6.48838133e-02 1.29496723e-01 8.53025079e-01
4.48030204e-01 -5.41494131e-01 -3.67184617e-02 -1.63445368e-01
1.74083471e-01 2.68366605e-01 2.53497869e-01 -9.50327218e-01
2.45467260e-01 -2.26464197e-01 4.23004963e-02 -2.67616302e-01
6.33999407e-01 -9.35065806e-01 3.78477946e-02 4.14842993e-01
-2.48037368e-01 -3.18217635e-01 -2.67278016e-01 7.91531563e-01
-5.10532022e-01 -5.47633290e-01 1.04555261e+00 -2.59489924e-01
-4.41443324e-01 6.76320791e-02 -3.98220718e-01 -1.01606831e-01
8.19466293e-01 -3.68203789e-01 1.51027113e-01 -4.75672632e-01
-9.51542199e-01 -3.08894277e-01 4.88849580e-01 -4.93965656e-01
9.83274639e-01 -1.64990258e+00 -1.12351835e+00 5.48504829e-01
-5.17541505e-02 -7.46675581e-02 8.92526507e-02 1.33418632e+00
-3.99103552e-01 1.44875631e-01 -4.28890996e-02 -7.83711493e-01
-1.15694702e+00 3.89809161e-01 1.62126645e-01 -1.70669798e-02
-9.37751353e-01 7.22139180e-01 -1.23465531e-01 -7.86082894e-02
-7.38720223e-02 -4.24674124e-01 1.09650493e-01 1.41427234e-01
6.33510828e-01 7.47230887e-01 2.78354794e-01 -8.52876306e-01
-4.28629905e-01 1.00109708e+00 2.66712874e-01 -5.29690534e-02
1.80059743e+00 1.61611006e-01 -7.92678416e-01 6.10085726e-01
1.53514910e+00 2.81468868e-01 -1.09010923e+00 -4.61074948e-01
1.59472153e-02 -5.57643652e-01 5.11860907e-01 -1.34620871e-02
-1.05529499e+00 6.07383251e-01 3.52134138e-01 7.23654747e-01
1.03375030e+00 -3.77917923e-02 5.32829523e-01 5.83757639e-01
4.41393197e-01 -1.09573662e+00 1.56224504e-01 3.09607595e-01
1.22787046e+00 -1.17524457e+00 3.84921461e-01 -6.52100921e-01
-8.21748842e-03 1.46699512e+00 -3.22854370e-01 -8.85961354e-01
1.00010490e+00 3.47765207e-01 -2.34898448e-01 -4.45065379e-01
-4.85542268e-01 1.64927896e-02 4.49714273e-01 5.10351241e-01
3.32642823e-01 5.69210760e-02 -4.97225612e-01 -1.04501508e-01
1.77273259e-01 -5.97323254e-02 3.34063530e-01 1.05412459e+00
-9.55880284e-01 -8.09578001e-01 -1.01607525e+00 6.11529469e-01
-1.20903835e-01 9.30133685e-02 1.40270159e-01 4.10666913e-01
-2.69984633e-01 7.41690814e-01 -2.76488543e-01 7.14079291e-02
2.93853074e-01 1.61056314e-02 5.63787222e-01 -5.22123933e-01
3.91551591e-02 6.37206197e-01 -3.00978012e-02 -5.87644398e-01
-7.39816248e-01 -1.07090557e+00 -1.07283521e+00 -1.45944402e-01
-2.23282322e-01 3.80196035e-01 7.83228099e-01 7.75975764e-01
-1.86779693e-01 7.58093894e-02 8.68330479e-01 -1.32171035e+00
-8.62084746e-01 -9.30315793e-01 -1.17496443e+00 2.90618688e-01
6.53708935e-01 -4.91558015e-01 -9.12496984e-01 1.05817564e-01]
|
[11.671968460083008, -2.310257911682129]
|
07cace9d-8416-4377-8540-3da08366d29e
|
a-conditional-splitting-framework-for
|
2106.15760
| null |
https://arxiv.org/abs/2106.15760v1
|
https://arxiv.org/pdf/2106.15760v1.pdf
|
A Conditional Splitting Framework for Efficient Constituency Parsing
|
We introduce a generic seq2seq parsing framework that casts constituency parsing problems (syntactic and discourse parsing) into a series of conditional splitting decisions. Our parsing model estimates the conditional probability distribution of possible splitting points in a given text span and supports efficient top-down decoding, which is linear in number of nodes. The conditional splitting formulation together with efficient beam search inference facilitate structural consistency without relying on expensive structured inference. Crucially, for discourse analysis we show that in our formulation, discourse segmentation can be framed as a special case of parsing which allows us to perform discourse parsing without requiring segmentation as a pre-requisite. Experiments show that our model achieves good results on the standard syntactic parsing tasks under settings with/without pre-trained representations and rivals state-of-the-art (SoTA) methods that are more computationally expensive than ours. In discourse parsing, our method outperforms SoTA by a good margin.
|
['XiaoLi Li', 'Shafiq Joty', 'Xuan-Phi Nguyen', 'Thanh-Tung Nguyen']
|
2021-06-30
| null |
https://aclanthology.org/2021.acl-long.450
|
https://aclanthology.org/2021.acl-long.450.pdf
|
acl-2021-5
|
['discourse-segmentation', 'discourse-parsing', 'constituency-parsing']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 5.19394398e-01 8.75032306e-01 -2.72575676e-01 -5.23352742e-01
-1.39458251e+00 -9.17363226e-01 6.10286474e-01 3.45100611e-01
-3.10539216e-01 7.25990176e-01 8.20198357e-01 -8.33268166e-01
3.98972303e-01 -8.42375219e-01 -8.27590466e-01 -4.32470709e-01
6.74404427e-02 5.90267003e-01 3.51646453e-01 -2.44651556e-01
3.24794739e-01 -1.63761154e-01 -1.05442476e+00 6.51980579e-01
6.20129943e-01 5.79647839e-01 1.79352909e-01 1.20825076e+00
-6.09550416e-01 1.21788716e+00 -7.42822886e-01 -4.74334717e-01
-1.07825041e-01 -5.33399165e-01 -1.52340162e+00 1.14479065e-01
3.45637769e-01 -3.17600995e-01 1.21853046e-01 8.09041381e-01
2.48198614e-01 -1.06407106e-01 3.08643878e-01 -4.08830941e-01
-2.54769385e-01 1.06475699e+00 -3.86721492e-01 4.05714154e-01
4.72982973e-01 -2.89022058e-01 1.59168899e+00 -5.62341392e-01
8.19616735e-01 1.42862546e+00 6.38073742e-01 4.74847406e-01
-1.34325194e+00 -7.04944581e-02 6.75574958e-01 -1.04447462e-01
-5.10929704e-01 -5.23068011e-01 4.96446103e-01 -2.53759265e-01
1.34978211e+00 3.65741014e-01 3.82391870e-01 9.70934033e-01
5.74633740e-02 1.05663502e+00 8.35293531e-01 -6.41186535e-01
4.56667155e-01 -5.48387229e-01 6.80860043e-01 6.93110168e-01
-1.40716374e-01 -6.49217010e-01 -5.01638591e-01 -1.41096488e-01
4.16110754e-01 -7.86716878e-01 -1.99062917e-02 1.02331683e-01
-9.77058947e-01 9.91989851e-01 -1.49176568e-01 3.38971883e-01
-1.49078861e-01 2.36918285e-01 7.27099955e-01 2.37970486e-01
5.61029792e-01 3.42053652e-01 -5.08865535e-01 -4.75162655e-01
-9.06327128e-01 2.62113094e-01 1.15961373e+00 1.02940726e+00
3.46740425e-01 -2.24703029e-01 -3.84349108e-01 5.28999805e-01
1.13676049e-01 2.44912639e-01 9.10217986e-02 -1.49235547e+00
9.15670574e-01 3.09703648e-01 -5.20272367e-02 -5.66383004e-01
-4.51248914e-01 8.43598843e-02 -4.33291525e-01 -2.05526471e-01
7.74953723e-01 -4.19364452e-01 -6.70824945e-01 2.07173657e+00
4.20846373e-01 9.70548764e-02 1.53495401e-01 4.53941226e-01
5.96116185e-01 8.92978191e-01 3.54573458e-01 -5.70006788e-01
1.78627181e+00 -9.22984064e-01 -7.89798856e-01 -4.24486488e-01
9.57123280e-01 -5.70527494e-01 9.05542850e-01 2.75972545e-01
-1.67651057e+00 -2.43034810e-01 -8.10986876e-01 -5.27409196e-01
2.93174297e-01 -2.38611206e-01 8.76804113e-01 7.83927441e-01
-1.18601012e+00 4.95628059e-01 -1.21344328e+00 4.42123264e-02
3.37448925e-01 1.31696284e-01 -2.08605334e-01 1.82594419e-01
-9.35648680e-01 7.54110396e-01 4.16069686e-01 5.41062243e-02
-4.91333663e-01 -4.87590313e-01 -1.14656913e+00 1.62724033e-01
5.67239881e-01 -7.33752549e-01 1.92138505e+00 -8.09731543e-01
-1.91256702e+00 1.06981289e+00 -8.00650835e-01 -5.86014092e-01
2.46436298e-01 -3.58388811e-01 2.88880438e-01 3.66940737e-01
2.94188857e-01 4.89393026e-01 4.23698276e-01 -7.18198299e-01
-6.38233960e-01 -3.25550169e-01 5.59705496e-01 3.11297059e-01
1.82227552e-01 4.43928301e-01 -4.74703550e-01 -3.78127038e-01
2.87836194e-01 -6.79714322e-01 -3.28763962e-01 -5.98923326e-01
-8.06076884e-01 -5.42842805e-01 3.65010172e-01 -8.66361320e-01
1.29154563e+00 -1.72447884e+00 4.94630873e-01 -1.81564704e-01
1.46416813e-01 5.41979931e-02 -2.31781211e-02 4.37074661e-01
3.08730006e-02 4.38096881e-01 -5.02783775e-01 -7.55157650e-01
2.44555958e-02 4.10031825e-01 -4.51964259e-01 2.91128218e-01
4.86462772e-01 1.05537570e+00 -8.61477315e-01 -6.36133850e-01
-4.04487327e-02 -2.97385696e-02 -8.08124423e-01 3.67234081e-01
-7.97025621e-01 4.91552234e-01 -5.47644615e-01 4.53430176e-01
3.63621622e-01 -5.09168863e-01 1.02506459e+00 2.67692626e-01
-3.16806883e-01 1.23774385e+00 -9.55128729e-01 2.07375813e+00
-4.40061688e-01 5.72930157e-01 5.74244916e-01 -1.36359811e+00
3.00313950e-01 2.86126316e-01 -4.38048914e-02 -3.97366971e-01
2.75839895e-01 -1.50874659e-01 -1.44108057e-01 -5.03097951e-01
5.66577852e-01 -6.62219673e-02 -6.13529444e-01 6.49605572e-01
-6.66247588e-03 2.48587355e-02 3.24655056e-01 3.31115365e-01
1.28742707e+00 3.15291047e-01 4.49357599e-01 -4.33485836e-01
4.08248007e-01 2.07112923e-01 7.59050310e-01 8.57733428e-01
4.29415740e-02 4.27985847e-01 1.18687809e+00 -2.50921845e-01
-1.03559089e+00 -8.95707309e-01 -9.56541598e-02 1.64435399e+00
-1.39101237e-01 -6.36025190e-01 -1.15190291e+00 -6.65039361e-01
-5.54595768e-01 8.15288007e-01 -5.76973975e-01 7.24888623e-01
-1.49919605e+00 -8.50525856e-01 7.51731515e-01 8.62246931e-01
2.93327600e-01 -9.66068387e-01 -8.62078667e-01 3.88502717e-01
-2.15191275e-01 -1.29397273e+00 -7.57806823e-02 4.72336620e-01
-8.71604085e-01 -1.26200724e+00 -3.40183288e-01 -1.01131356e+00
3.56829464e-01 2.46043224e-03 1.48470712e+00 1.24426141e-01
2.27341093e-02 1.59033671e-01 -3.84522796e-01 -6.31586909e-02
-7.85675228e-01 3.16249251e-01 -7.73596764e-01 -7.39632726e-01
4.74271104e-02 -4.50524956e-01 -3.31931651e-01 -3.88652116e-01
-5.73833108e-01 4.22552586e-01 6.05805293e-02 9.65560496e-01
4.98381019e-01 -4.72046643e-01 5.18169641e-01 -1.56278336e+00
4.96945620e-01 -5.23996830e-01 -6.35980070e-01 3.90426427e-01
-1.33603299e-02 2.65329957e-01 5.76738894e-01 1.88645139e-01
-1.40814507e+00 -1.05292007e-01 -4.87388700e-01 3.20334375e-01
-2.91202903e-01 5.01566112e-01 -2.47757882e-01 7.41052330e-01
2.46802464e-01 -1.30312592e-01 -2.17638478e-01 -3.98498505e-01
7.11250186e-01 4.15617824e-01 6.69775844e-01 -9.28640783e-01
1.97276562e-01 3.41139168e-01 2.53962446e-03 -7.59665251e-01
-1.19099545e+00 -1.76993057e-01 -7.51811445e-01 3.36097747e-01
1.45672238e+00 -9.49228883e-01 -8.84432435e-01 1.60788655e-01
-1.63740134e+00 -5.80320895e-01 -2.12755606e-01 -6.53339922e-02
-5.91081560e-01 7.92507887e-01 -1.13411057e+00 -9.81268823e-01
-2.63173938e-01 -9.92105961e-01 1.39066923e+00 6.41746223e-02
-4.94592458e-01 -1.18552113e+00 1.15378657e-02 5.49393415e-01
2.68088188e-03 3.16415578e-01 1.17356384e+00 -7.11089611e-01
-6.33699775e-01 3.30370933e-01 -6.11148812e-02 -6.11275015e-03
-1.86566889e-01 -7.79944509e-02 -1.08760381e+00 9.32772011e-02
1.12365887e-01 -4.30048972e-01 8.92844319e-01 6.40375614e-01
1.00309765e+00 -3.09529841e-01 -1.10268064e-01 4.16585207e-01
8.36816609e-01 -1.11588627e-01 3.52949709e-01 2.86370546e-01
5.63595712e-01 7.47721791e-01 4.20206666e-01 2.73227483e-01
7.46891797e-01 3.73710394e-01 3.15916717e-01 1.67952657e-01
2.07732897e-02 -2.18532175e-01 3.22877288e-01 9.62621927e-01
1.78736150e-01 -5.46742737e-01 -9.23139751e-01 6.14874244e-01
-2.06151342e+00 -9.29749012e-01 -3.34523976e-01 1.68860030e+00
1.16349208e+00 2.94223249e-01 1.34028837e-01 1.21886618e-01
7.17312098e-01 5.91965199e-01 -2.76232749e-01 -8.44921112e-01
-1.72631815e-01 7.43123114e-01 3.28704506e-01 1.02771056e+00
-1.39494383e+00 1.18135202e+00 7.29036665e+00 4.53832358e-01
-6.15533650e-01 2.30702326e-01 9.11951721e-01 -2.54918020e-02
-5.51299334e-01 1.52251035e-01 -8.45798254e-01 2.31144622e-01
1.28336382e+00 1.40576184e-01 7.07997680e-02 8.01962972e-01
-1.20010257e-01 -2.77170390e-01 -1.28212166e+00 3.52856308e-01
-1.71967387e-01 -1.76614201e+00 -3.50154787e-01 -2.09586084e-01
4.15138870e-01 -2.45486535e-02 -4.08865213e-01 3.26515287e-01
7.83016086e-01 -1.13923585e+00 8.00727665e-01 -1.77365348e-01
5.61117232e-01 -4.43491697e-01 5.61654389e-01 6.44884825e-01
-1.06278825e+00 -2.53704395e-02 -9.31867063e-02 -4.60307896e-01
7.48103023e-01 3.50029439e-01 -7.52473712e-01 6.05045021e-01
3.94374311e-01 4.26403791e-01 2.43272632e-02 1.67444348e-01
-8.56230974e-01 1.13414299e+00 -3.44519258e-01 2.60906368e-02
4.24676895e-01 -4.52100672e-02 5.24769068e-01 1.62287426e+00
-1.69590399e-01 5.98506808e-01 4.01972085e-01 5.79265058e-01
-2.42191911e-01 -1.65907741e-02 -1.62870437e-01 1.36898801e-01
7.41173506e-01 8.75864208e-01 -8.10168505e-01 -6.34317994e-01
-3.54533762e-01 7.82861948e-01 6.63461506e-01 -6.45060018e-02
-5.28703272e-01 1.74877182e-01 3.65604669e-01 -2.55491883e-01
7.16742218e-01 -3.44694912e-01 -6.80516839e-01 -1.18309498e+00
3.98695990e-02 -7.92545915e-01 6.13340497e-01 -2.56278247e-01
-9.71144915e-01 5.21688402e-01 8.00880715e-02 -3.63645494e-01
-7.42250502e-01 -6.66068971e-01 -9.21379387e-01 8.48907948e-01
-1.61027634e+00 -1.08878887e+00 2.31560484e-01 4.13057804e-02
9.10295367e-01 3.71680588e-01 1.20276618e+00 -2.44662642e-01
-8.30741823e-01 3.05890322e-01 -3.35740656e-01 3.89725178e-01
1.60232872e-01 -1.59295261e+00 8.94733787e-01 1.14984012e+00
4.06313017e-02 5.61612606e-01 6.66247010e-01 -5.79491735e-01
-1.39277220e+00 -7.57472336e-01 1.26111054e+00 -4.72883493e-01
7.36913502e-01 -5.89523435e-01 -8.91143382e-01 9.49486136e-01
5.84415972e-01 -3.61097336e-01 8.67605746e-01 6.73834920e-01
-4.33534622e-01 6.80388510e-01 -5.92870653e-01 3.97554129e-01
1.13158238e+00 -5.86224914e-01 -1.09667671e+00 4.41491187e-01
8.85004461e-01 -9.86198068e-01 -7.82986224e-01 1.33355334e-01
2.87636340e-01 -9.54944134e-01 7.21977711e-01 -9.55052495e-01
7.25181282e-01 3.82680520e-02 -3.46772671e-01 -6.01278484e-01
-2.28941530e-01 -9.29016113e-01 -1.67448416e-01 1.16805744e+00
5.12959838e-01 -1.86175689e-01 9.38636839e-01 8.49676967e-01
-4.53031451e-01 -6.72658861e-01 -1.17826021e+00 -3.17186385e-01
5.43463707e-01 -6.88888133e-01 3.39186817e-01 7.35454857e-01
4.07205909e-01 7.42744505e-01 -5.06466888e-02 2.18658775e-01
3.59124810e-01 4.70770687e-01 5.60542107e-01 -9.68062937e-01
-7.00025618e-01 -4.91889924e-01 2.09321603e-01 -1.53537309e+00
7.08574414e-01 -8.27425718e-01 2.77797312e-01 -1.62700009e+00
7.71014765e-02 -3.70291263e-01 2.68013328e-01 4.71785367e-01
-4.13160801e-01 -2.07744136e-01 2.71664947e-01 -8.86906963e-03
-8.26095641e-01 2.14332357e-01 7.92872608e-01 4.11873870e-02
-1.90762654e-01 -1.59048468e-01 -8.97218406e-01 9.02851641e-01
6.47313237e-01 -3.42023104e-01 -2.68544763e-01 -7.84741640e-01
4.30809706e-01 6.48426235e-01 1.55746387e-02 -2.61075199e-01
2.36614436e-01 -7.65944868e-02 -1.11988015e-01 -5.50502539e-01
2.51668870e-01 -1.88543499e-02 -3.94816726e-01 1.63490415e-01
-7.54880667e-01 1.93524584e-01 2.05887079e-01 4.92462158e-01
-1.93177998e-01 -4.83757496e-01 4.21091855e-01 -3.63246888e-01
-4.00657535e-01 -3.41547728e-01 -6.23598337e-01 6.05806053e-01
7.88313925e-01 -3.10008377e-02 -5.52097559e-01 -2.75592744e-01
-8.28094482e-01 3.93016279e-01 2.33279243e-01 -9.66854170e-02
-6.18850626e-02 -6.58079743e-01 -7.20667243e-01 -2.03126177e-01
-6.32713437e-01 7.17041671e-01 1.62752569e-01 6.68240666e-01
-4.43889022e-01 6.14077866e-01 3.39332342e-01 -6.23500466e-01
-1.42771137e+00 7.31156319e-02 -1.40579179e-01 -8.64229739e-01
-7.84070909e-01 1.10702837e+00 2.65952438e-01 -2.45833069e-01
2.53823042e-01 -6.99067116e-01 -1.80844963e-01 6.72378913e-02
6.24360621e-01 7.53193200e-02 1.29693625e-02 -4.48756427e-01
-4.02729720e-01 2.42811725e-01 -1.46402389e-01 -1.87651739e-01
1.38523662e+00 -1.43059641e-01 -2.19830900e-01 3.99869293e-01
8.55238080e-01 2.19946921e-01 -1.25416970e+00 -1.20264851e-01
5.11967719e-01 1.49166599e-01 -3.31604593e-02 -5.77054262e-01
-3.75914395e-01 9.80093420e-01 -4.22122717e-01 4.88877535e-01
9.88779426e-01 3.05648118e-01 8.95009518e-01 5.65605521e-01
8.23192075e-02 -1.09494340e+00 -3.17280561e-01 1.03666568e+00
4.61209506e-01 -1.00972390e+00 -8.26314017e-02 -8.93398345e-01
-5.93833148e-01 1.15028584e+00 2.23513931e-01 -3.60792220e-01
3.69221270e-01 7.50804126e-01 -1.12788498e-01 -6.12296462e-02
-1.08907914e+00 -1.38494954e-01 -1.45783767e-01 3.94313723e-01
8.12896013e-01 8.17359909e-02 -3.91266674e-01 8.16999555e-01
-4.48939681e-01 -4.37000036e-01 4.27924573e-01 1.16055179e+00
-7.19400346e-01 -1.49778712e+00 -9.20357555e-02 1.69963781e-02
-1.00753868e+00 -3.55310321e-01 -2.17873961e-01 7.84939945e-01
-2.48178512e-01 1.06707573e+00 4.85360533e-01 1.12899147e-01
1.23034924e-01 3.64582479e-01 6.25149310e-01 -1.00615311e+00
-8.17222238e-01 1.67731225e-01 9.63945091e-01 -7.06888139e-01
-6.72258675e-01 -9.84943390e-01 -1.66286922e+00 -2.14899108e-01
-2.02989817e-01 1.10277817e-01 4.26680863e-01 1.31842136e+00
3.05042297e-01 6.55740440e-01 3.02681208e-01 -7.65997231e-01
-6.53011084e-01 -9.09009457e-01 2.75835004e-02 2.09976822e-01
2.86630541e-01 -4.24933732e-02 6.60041273e-02 2.63530254e-01]
|
[10.724250793457031, 9.493741989135742]
|
92eac626-6285-4796-803b-d00ef27016f0
|
ibot-image-bert-pre-training-with-online
|
2111.07832
| null |
https://arxiv.org/abs/2111.07832v3
|
https://arxiv.org/pdf/2111.07832v3.pdf
|
iBOT: Image BERT Pre-Training with Online Tokenizer
|
The success of language Transformers is primarily attributed to the pretext task of masked language modeling (MLM), where texts are first tokenized into semantically meaningful pieces. In this work, we study masked image modeling (MIM) and indicate the advantages and challenges of using a semantically meaningful visual tokenizer. We present a self-supervised framework iBOT that can perform masked prediction with an online tokenizer. Specifically, we perform self-distillation on masked patch tokens and take the teacher network as the online tokenizer, along with self-distillation on the class token to acquire visual semantics. The online tokenizer is jointly learnable with the MIM objective and dispenses with a multi-stage training pipeline where the tokenizer needs to be pre-trained beforehand. We show the prominence of iBOT by achieving an 82.3% linear probing accuracy and an 87.8% fine-tuning accuracy evaluated on ImageNet-1K. Beyond the state-of-the-art image classification results, we underline emerging local semantic patterns, which helps the models to obtain strong robustness against common corruptions and achieve leading results on dense downstream tasks, eg., object detection, instance segmentation, and semantic segmentation.
|
['Tao Kong', 'Alan Yuille', 'Cihang Xie', 'Wei Shen', 'Huiyu Wang', 'Chen Wei', 'Jinghao Zhou']
|
2021-11-15
| null | null | null | null |
['self-supervised-image-classification', 'semi-supervised-image-classification', 'unsupervised-image-classification']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 4.82621580e-01 7.29775071e-01 -3.79132926e-01 -5.07461667e-01
-9.41732228e-01 -3.28688204e-01 5.84149837e-01 8.75567943e-02
-6.67424262e-01 2.20343873e-01 7.10568279e-02 -3.47985119e-01
6.30617142e-01 -3.20754081e-01 -1.19444895e+00 -6.87616825e-01
8.85066837e-02 3.78179312e-01 2.78505325e-01 2.41247088e-01
-1.42494082e-01 -5.01463376e-02 -1.50400865e+00 9.01535928e-01
9.13413167e-01 1.14642859e+00 4.94507581e-01 4.71384138e-01
-5.00420392e-01 1.18024445e+00 -4.63034272e-01 -6.74806833e-01
6.47403225e-02 -7.55462348e-02 -9.37238932e-01 3.64662915e-01
6.05867684e-01 -2.27524236e-01 -8.78062099e-02 9.10874903e-01
2.17610970e-01 -2.12619081e-01 4.10822034e-01 -1.21325171e+00
-6.98926449e-01 1.30729473e+00 -5.52483797e-01 -1.44706413e-01
-2.81130195e-01 4.60732371e-01 1.23803067e+00 -1.33066857e+00
5.15006959e-01 1.46118581e+00 5.16222417e-01 5.24536133e-01
-1.33703375e+00 -6.30698800e-01 6.25522017e-01 2.85716027e-01
-1.42251706e+00 -3.98860872e-01 4.45679367e-01 -6.63751423e-01
1.16162312e+00 -9.47367549e-02 3.56211573e-01 9.52703178e-01
-4.89093438e-02 1.27222192e+00 1.04673254e+00 -5.65798402e-01
8.90637860e-02 4.47095931e-01 8.88323709e-02 8.75399053e-01
-2.46865094e-01 -3.49272452e-02 -7.66620755e-01 3.23788613e-01
2.13726684e-01 -2.63335735e-01 1.54025406e-01 -2.19480798e-01
-1.34187150e+00 6.24917686e-01 7.91182399e-01 1.61699012e-01
-2.35757738e-01 5.26777685e-01 4.11991835e-01 2.73671821e-02
8.85678828e-01 2.14540362e-01 -6.03645802e-01 1.87821373e-01
-1.50359082e+00 -1.58508018e-01 2.94854552e-01 9.35245335e-01
1.10918570e+00 3.33367199e-01 -4.45968479e-01 7.64698386e-01
3.77581567e-01 3.58345121e-01 4.73124892e-01 -6.44142151e-01
4.51216400e-01 6.23729467e-01 -2.38873452e-01 -4.12507772e-01
-2.70140022e-01 -5.91054082e-01 -5.37192047e-01 2.59037197e-01
3.68976295e-01 1.08307704e-01 -1.41766834e+00 1.80892360e+00
3.03473800e-01 4.51384872e-01 1.79447327e-02 7.06418574e-01
7.78715253e-01 5.31487048e-01 6.88820601e-01 1.99066833e-01
1.62061775e+00 -1.40609539e+00 -4.92238313e-01 -6.47539914e-01
8.36657584e-01 -7.38011837e-01 1.35673952e+00 5.31547546e-01
-1.14894426e+00 -7.31122613e-01 -9.87636089e-01 -3.31237465e-01
-4.48683828e-01 2.65041918e-01 3.77493978e-01 3.94442111e-01
-1.19481754e+00 5.12231886e-01 -1.04793477e+00 -1.06812604e-01
7.63513148e-01 2.43976966e-01 -1.09265596e-01 4.49289987e-03
-9.73758340e-01 9.17466402e-01 5.90707839e-01 9.15248916e-02
-1.23324132e+00 -9.79601979e-01 -8.53854001e-01 -1.78457219e-02
5.25245786e-01 -5.70592701e-01 1.22562945e+00 -1.49004972e+00
-1.30317378e+00 1.43719244e+00 -4.20933813e-01 -8.32709551e-01
8.05187106e-01 -1.71831608e-01 -8.94600824e-02 2.74542421e-01
2.99229234e-01 1.30683935e+00 1.10804963e+00 -1.30966043e+00
-5.95190763e-01 -1.03844199e-02 -2.47214675e-01 -5.14895888e-03
-3.24089259e-01 9.12620127e-02 -6.70475125e-01 -8.30039740e-01
1.45703673e-01 -6.70819640e-01 -2.88314939e-01 1.44228740e-02
-7.45324790e-01 -1.77546129e-01 6.16917014e-01 -8.89200509e-01
8.34181905e-01 -2.36239600e+00 -3.17297056e-02 4.82948171e-03
4.63607609e-01 1.92269385e-01 -2.54099429e-01 1.01572774e-01
-2.52591640e-01 1.63534239e-01 -2.90415317e-01 -1.17258894e+00
2.76670784e-01 1.58916935e-01 -6.34906888e-01 3.28371882e-01
6.27850771e-01 1.37406325e+00 -6.56894505e-01 -7.66752064e-01
2.07562387e-01 3.85459304e-01 -6.54995084e-01 1.22763850e-01
-7.13839293e-01 1.94357857e-01 8.07310045e-02 7.56056249e-01
5.75464547e-01 -5.88327229e-01 1.01584516e-01 -3.44566047e-01
-1.78048432e-01 5.90143263e-01 -9.46348965e-01 1.68154895e+00
-3.77113640e-01 5.62980950e-01 1.92026347e-01 -9.91466522e-01
5.29686987e-01 2.05720842e-01 3.22069377e-01 -8.11378598e-01
3.81831639e-02 9.21085700e-02 -4.23382789e-01 -3.51761103e-01
6.83667064e-01 -9.56450850e-02 1.70335453e-02 4.08993632e-01
2.24808916e-01 2.11769089e-01 -1.19095691e-01 4.40744966e-01
6.57738447e-01 2.59989530e-01 1.73141155e-02 -3.76895726e-01
2.05461472e-01 3.47031265e-01 3.38443249e-01 8.23972523e-01
-7.82782678e-03 6.30476654e-01 4.40633357e-01 -1.64574325e-01
-1.02273929e+00 -1.18611896e+00 -1.36457821e-02 1.48458397e+00
-4.68329061e-03 -4.39971954e-01 -9.80159223e-01 -6.38218701e-01
1.53561413e-01 5.57426214e-01 -5.84403872e-01 -1.52374387e-01
-4.93673831e-01 -6.93347454e-01 6.85813189e-01 6.41927779e-01
5.83392620e-01 -1.15875065e+00 -2.59076238e-01 2.56531030e-01
-1.05258629e-01 -1.46565711e+00 -4.42045212e-01 5.39073765e-01
-4.67966527e-01 -5.80686271e-01 -3.87619585e-01 -1.04364800e+00
6.20642483e-01 1.95383683e-01 1.11415422e+00 1.73428535e-01
-3.12746793e-01 7.85668418e-02 -9.95782763e-02 -2.86126286e-01
-6.37000799e-01 2.29630023e-01 -2.06490666e-01 3.15624207e-01
1.28811464e-01 -3.03539217e-01 -5.43890417e-01 1.22462139e-01
-8.77560019e-01 6.61595881e-01 6.64140403e-01 7.84236908e-01
8.12675774e-01 -3.13398510e-01 2.70898700e-01 -1.00186455e+00
-2.27045789e-01 -3.30422163e-01 -6.59694433e-01 2.47159183e-01
-6.22670531e-01 1.57037735e-01 3.12765658e-01 -5.84218562e-01
-8.61573040e-01 4.21149880e-01 -1.77462827e-02 -4.93425697e-01
2.78112907e-02 4.86763179e-01 -3.14193368e-01 1.02388985e-01
4.67671603e-01 3.43420148e-01 -1.06094152e-01 -6.07486904e-01
8.90062273e-01 5.23162246e-01 7.62331843e-01 -5.06186068e-01
8.16229045e-01 6.68966711e-01 -5.05069315e-01 -7.25129485e-01
-1.13454163e+00 -3.48056406e-01 -6.15090728e-01 1.06463255e-02
1.24641562e+00 -1.49148953e+00 -6.50816143e-01 6.37053609e-01
-1.10405457e+00 -9.28139389e-01 -4.32525218e-01 -2.47847848e-02
-3.93558830e-01 2.56534398e-01 -7.04277217e-01 -5.87038934e-01
-3.14268351e-01 -1.19150591e+00 1.16680396e+00 -2.05192387e-01
-1.14700012e-01 -9.10096943e-01 -7.17339396e-01 7.14358747e-01
3.88342589e-01 -1.08576111e-01 9.80363309e-01 -6.74664021e-01
-1.06605983e+00 2.23598182e-01 -5.92715681e-01 5.63921273e-01
-1.70256719e-01 -1.06953517e-01 -1.60982573e+00 -2.62322038e-01
-2.93351740e-01 -4.66333151e-01 1.36317205e+00 2.93972522e-01
1.29668796e+00 -5.36253870e-01 -3.39131325e-01 8.59279990e-01
1.24880481e+00 -4.34751689e-01 4.04883772e-01 3.26960891e-01
1.09676397e+00 6.45122647e-01 3.82391274e-01 1.77707210e-01
7.85740733e-01 5.00759661e-01 6.07568800e-01 -5.18909335e-01
-5.67546010e-01 -6.21610940e-01 5.18708467e-01 6.68506384e-01
6.94898069e-01 -8.64457339e-02 -1.01469505e+00 7.06812561e-01
-1.85326540e+00 -6.31989896e-01 5.67515865e-02 1.77469981e+00
1.25026917e+00 4.14999872e-01 -6.62863329e-02 -1.62334442e-01
5.95466197e-01 2.99948752e-01 -5.71720481e-01 -9.34618190e-02
-4.15753543e-01 2.73305833e-01 7.22249329e-01 8.31673324e-01
-1.25478494e+00 1.81509578e+00 5.89120293e+00 1.09606147e+00
-1.48048961e+00 5.58685720e-01 8.66286993e-01 -8.47080424e-02
-2.32213825e-01 1.85577288e-01 -9.74799275e-01 4.59142447e-01
7.97124326e-01 2.05009535e-01 3.82843822e-01 9.16365981e-01
2.52179801e-01 -1.01728208e-01 -1.07213449e+00 7.84347296e-01
-1.54985175e-01 -1.54372549e+00 2.18267828e-01 -1.59349129e-01
7.76832819e-01 4.95463908e-01 1.97740287e-01 2.99705863e-01
4.16415364e-01 -1.15394199e+00 1.39076030e+00 2.38213256e-01
8.45071852e-01 -3.72865587e-01 1.26132295e-01 3.86055917e-01
-1.23775232e+00 3.90043072e-02 -2.58992791e-01 7.43896589e-02
-1.03154220e-02 8.34448516e-01 -1.30077100e+00 1.99841723e-01
6.73426569e-01 7.56651759e-01 -6.71197951e-01 6.96319342e-01
-5.23594975e-01 9.33914065e-01 -2.37084508e-01 3.54702502e-01
4.36737210e-01 2.97603965e-01 1.07652262e-01 1.53385365e+00
-3.06544662e-01 -4.55329657e-01 5.79630911e-01 1.26101398e+00
-2.37279952e-01 4.89720814e-02 6.99544996e-02 1.60097033e-02
3.57287467e-01 1.27990115e+00 -7.58709192e-01 -8.01337779e-01
-3.28925163e-01 1.27218008e+00 4.88953054e-01 4.67876136e-01
-9.63655710e-01 8.34632143e-02 6.33758247e-01 1.63809866e-01
5.25415957e-01 -1.89559668e-01 -7.14147747e-01 -1.14683831e+00
1.24370895e-01 -7.59785712e-01 2.74075717e-01 -8.54176283e-01
-1.08475256e+00 5.27174532e-01 -1.94541097e-01 -8.43926907e-01
-4.47140746e-02 -5.94749808e-01 -4.86881673e-01 9.00744140e-01
-1.84687102e+00 -1.74666893e+00 -2.35750109e-01 4.96076405e-01
4.81131673e-01 1.96263656e-01 4.67557847e-01 2.67335594e-01
-6.76847637e-01 8.97119105e-01 -2.02922881e-01 2.97627389e-01
7.51072764e-01 -1.36255050e+00 7.04801738e-01 1.09551656e+00
2.51195371e-01 3.36209148e-01 5.27604282e-01 -6.28763855e-01
-9.92652774e-01 -1.57597494e+00 1.13412273e+00 -4.74064618e-01
8.70814681e-01 -9.68294740e-01 -9.25381184e-01 9.73376334e-01
1.81385294e-01 2.19267681e-01 2.22999662e-01 -2.50459194e-01
-6.66475892e-01 -8.87203515e-02 -8.31912756e-01 5.98237514e-01
9.60280776e-01 -7.77157366e-01 -3.21382344e-01 4.83793318e-01
1.21626258e+00 -4.67550874e-01 -3.46042007e-01 1.58312276e-01
2.01516345e-01 -5.56999922e-01 1.04799855e+00 -5.87638736e-01
4.77646559e-01 -3.26513171e-01 -1.92773655e-01 -8.98821056e-01
-1.90837551e-02 -6.20808899e-01 1.16429307e-01 1.31343436e+00
5.96152782e-01 -5.71939588e-01 9.90009069e-01 5.00598848e-01
-2.86591291e-01 -5.85900128e-01 -9.92167413e-01 -7.51050234e-01
6.91204295e-02 -8.70428622e-01 3.03031623e-01 9.82391179e-01
-8.86066929e-02 1.21586382e-01 -4.22342926e-01 4.01678473e-01
7.60859013e-01 -4.60944399e-02 6.34291887e-01 -6.23626947e-01
-3.70948553e-01 -4.59102064e-01 -1.09355487e-01 -1.27529919e+00
4.19021726e-01 -1.38814270e+00 2.31612772e-01 -1.35716569e+00
2.04844415e-01 -4.68623489e-01 -3.11704129e-01 1.12512743e+00
-4.08692896e-01 5.23881853e-01 2.60523647e-01 2.23952696e-01
-8.45943749e-01 4.90550756e-01 9.39686298e-01 -4.40644294e-01
-3.18755843e-02 -3.57887357e-01 -7.30518043e-01 7.32750773e-01
5.67411244e-01 -6.73688293e-01 -1.77222997e-01 -7.52407074e-01
3.08870394e-02 -3.20359617e-01 7.89865851e-01 -7.99238741e-01
2.76090324e-01 4.65025194e-02 2.27650002e-01 -5.86320460e-01
4.28651094e-01 -5.29088616e-01 -2.53578216e-01 4.79381859e-01
-5.31593382e-01 -1.48436487e-01 3.20417017e-01 2.23998457e-01
-1.28760844e-01 1.28428683e-01 9.18149769e-01 -1.49161950e-01
-9.61730301e-01 1.34985343e-01 -2.98237890e-01 1.46282062e-01
7.70807922e-01 -2.91889250e-01 -2.50529647e-01 1.07740853e-02
-1.08247530e+00 4.35218126e-01 4.60964799e-01 3.79594982e-01
4.08040047e-01 -9.75346386e-01 -6.43962741e-01 2.83142030e-01
1.13068260e-01 1.20695144e-01 2.78953344e-01 8.56508136e-01
-1.82537749e-01 1.88524991e-01 3.35784167e-01 -9.50254738e-01
-1.13500428e+00 7.25861132e-01 4.13159460e-01 -2.27798611e-01
-7.06385314e-01 1.15522623e+00 5.62679350e-01 -1.92432851e-01
4.85813946e-01 -6.81995153e-01 4.26739603e-01 1.40875563e-01
3.82241249e-01 -1.13184460e-01 2.90861487e-01 -5.20018816e-01
-4.96880025e-01 3.38750392e-01 -1.47338495e-01 -9.31124613e-02
1.14229536e+00 -2.00663924e-01 -2.65427768e-01 3.29972863e-01
1.17153966e+00 -9.73295644e-02 -1.59725332e+00 -7.24464953e-01
3.15851212e-01 7.25322366e-02 9.30038840e-02 -1.00717199e+00
-1.17043304e+00 9.72656965e-01 4.46193457e-01 -1.77674443e-01
9.05185342e-01 3.53456944e-01 6.44172311e-01 5.49426749e-02
6.36751726e-02 -1.03025615e+00 2.43437767e-01 3.89946282e-01
5.60982764e-01 -1.38136375e+00 -2.87741512e-01 -6.09664023e-01
-7.78550804e-01 6.69970989e-01 5.74012399e-01 2.83159558e-02
6.70907080e-01 4.50796545e-01 4.11631733e-01 -5.45659810e-02
-9.07887399e-01 -4.33188349e-01 4.42570895e-01 1.81037441e-01
3.26120824e-01 1.43447787e-01 3.43984962e-01 8.00505757e-01
-4.19847608e-01 -1.80895895e-01 8.25339630e-02 6.69415653e-01
-5.32561839e-01 -9.15922582e-01 -2.20425576e-01 3.90747078e-02
-4.11679089e-01 -7.50819921e-01 -2.50467420e-01 6.11835480e-01
4.07593995e-01 8.58765423e-01 1.37809291e-01 -3.81472498e-01
2.16560867e-02 4.39632088e-01 1.47823885e-01 -8.08088422e-01
-8.48962188e-01 9.05448496e-02 -5.18908650e-02 -7.09054947e-01
-2.11322293e-01 -4.27008420e-01 -1.47372854e+00 -1.09999470e-01
-4.38074693e-02 -1.10555999e-01 7.33905792e-01 1.16348898e+00
4.13005710e-01 6.45445466e-01 3.09742212e-01 -8.82971883e-01
-4.86940056e-01 -8.55707645e-01 -1.46995753e-01 3.35313857e-01
4.96526152e-01 -2.40350351e-01 -2.19304711e-01 4.52792883e-01]
|
[9.658617973327637, 0.851128876209259]
|
b78d88a7-45b6-4e0f-8298-ef3f6d17295e
|
benchmarking-scene-text-recognition-in
|
2104.04437
| null |
https://arxiv.org/abs/2104.04437v1
|
https://arxiv.org/pdf/2104.04437v1.pdf
|
Benchmarking Scene Text Recognition in Devanagari, Telugu and Malayalam
|
Inspired by the success of Deep Learning based approaches to English scene text recognition, we pose and benchmark scene text recognition for three Indic scripts - Devanagari, Telugu and Malayalam. Synthetic word images rendered from Unicode fonts are used for training the recognition system. And the performance is bench-marked on a new IIIT-ILST dataset comprising of hundreds of real scene images containing text in the above mentioned scripts. We use a segmentation free, hybrid but end-to-end trainable CNN-RNN deep neural network for transcribing the word images to the corresponding texts. The cropped word images need not be segmented into the sub-word units and the error is calculated and backpropagated for the the given word image at once. The network is trained using CTC loss, which is proven quite effective for sequence-to-sequence transcription tasks. The CNN layers in the network learn to extract robust feature representations from word images. The sequence of features learnt by the convolutional block is transcribed to a sequence of labels by the RNN+CTC block. The transcription is not bound by word length or a lexicon and is ideal for Indian languages which are highly inflectional. IIIT-ILST dataset, synthetic word images dataset and the script used to render synthetic images are available at http://cvit.iiit.ac.in/research/projects/cvit-projects/iiit-ilst
|
['CV Jawahar', 'Mohit Jain', 'Minesh Mathew']
|
2021-04-09
| null | null | null | null |
['scene-text-recognition']
|
['computer-vision']
|
[ 6.63143098e-01 -3.43084097e-01 1.37494177e-01 -4.57071215e-01
-7.66506195e-01 -8.23391140e-01 8.14994395e-01 -5.15118301e-01
-6.50795043e-01 5.60218573e-01 1.49793521e-01 -5.40951669e-01
4.99961436e-01 -5.94381154e-01 -7.05700159e-01 -7.35878348e-01
4.01274800e-01 7.09045053e-01 -1.61333695e-01 -1.96871564e-01
5.56119382e-01 3.65944445e-01 -1.04567242e+00 7.00302780e-01
6.75814807e-01 8.51687610e-01 4.21429008e-01 1.46789432e+00
-4.72071707e-01 9.85755801e-01 -9.61682439e-01 -3.31516534e-01
4.95206952e-01 -7.71320581e-01 -9.49817300e-01 4.79561538e-01
5.85719466e-01 -4.23557848e-01 -5.94756663e-01 8.20121348e-01
6.53277993e-01 1.85923222e-02 9.64814365e-01 -7.41607904e-01
-8.60518932e-01 7.12205887e-01 -5.65190852e-01 7.64635131e-02
3.16471547e-01 1.94758028e-01 6.84115171e-01 -1.18936813e+00
6.95052028e-01 1.23527062e+00 3.28792840e-01 4.61072415e-01
-7.17791021e-01 -4.45457578e-01 -2.99984664e-01 1.90580949e-01
-1.35366237e+00 -4.54093397e-01 3.88359606e-01 -4.54199076e-01
1.27983499e+00 5.10859251e-01 2.87991852e-01 1.41748464e+00
2.07630321e-01 1.00529528e+00 1.07394874e+00 -7.98407435e-01
2.85872594e-02 4.00220565e-02 1.07821256e-01 7.26628780e-01
-3.11133385e-01 -3.56654018e-01 -3.90969247e-01 5.42042553e-01
6.61931574e-01 -3.35421890e-01 -9.65784043e-02 2.91429192e-01
-1.28477693e+00 9.12646592e-01 1.59295246e-01 4.71574873e-01
-5.21680377e-02 1.81528747e-01 9.76939797e-01 4.48724002e-01
2.75189102e-01 1.15326948e-01 -4.14216757e-01 -3.35207880e-01
-1.27555895e+00 -9.57112536e-02 5.99719644e-01 1.13064945e+00
3.47758234e-01 8.12967598e-01 -3.14611971e-01 1.22364700e+00
3.43223996e-02 7.00908780e-01 9.68297243e-01 -3.03139985e-01
9.04022753e-01 3.75907332e-01 -3.44509244e-01 -6.74472332e-01
-8.72501731e-02 -1.06841013e-01 -1.00914216e+00 -5.61091267e-02
4.03319061e-01 -2.17469290e-01 -1.64223421e+00 9.90936935e-01
-1.74263135e-01 -3.03915620e-01 3.12150180e-01 8.12212169e-01
8.69120061e-01 1.35995650e+00 -2.32903227e-01 1.44794255e-01
1.34132516e+00 -1.24928510e+00 -5.56792915e-01 -2.82094866e-01
7.20276237e-01 -1.23410761e+00 1.28280592e+00 4.56351906e-01
-8.61974180e-01 -7.25432456e-01 -1.11746681e+00 -2.91529566e-01
-6.93043709e-01 6.90673470e-01 -1.46774889e-03 6.07998133e-01
-8.13248634e-01 5.96657731e-02 -4.46487010e-01 -6.52488708e-01
5.57143211e-01 2.83747166e-01 -2.89866656e-01 -8.93744901e-02
-1.07561874e+00 8.59098434e-01 7.53483295e-01 4.10329640e-01
-1.13291216e+00 5.79322129e-02 -8.82300079e-01 -2.28056923e-01
-2.24553198e-02 3.67423445e-02 9.94581044e-01 -1.61554801e+00
-1.73484921e+00 1.12727821e+00 2.82868128e-02 -4.01212245e-01
9.38798368e-01 -2.15835907e-02 -4.04089093e-01 1.28580913e-01
1.15749672e-01 8.01830888e-01 1.45762849e+00 -9.44150925e-01
-2.99122691e-01 -6.59486428e-02 -6.04673803e-01 2.96486288e-01
-1.52589336e-01 3.86435091e-01 -5.99795759e-01 -1.04449594e+00
-1.75784722e-01 -9.60270822e-01 8.58282596e-02 -4.45866406e-01
-9.71187770e-01 9.56676900e-03 1.23932922e+00 -1.03364170e+00
5.03647447e-01 -1.96671426e+00 2.96620186e-02 -6.89138472e-02
-4.03648734e-01 6.07074082e-01 -4.31930423e-01 5.84301293e-01
-2.37247258e-01 9.01812986e-02 -4.24460739e-01 -1.82520673e-01
7.52200782e-02 3.03764224e-01 -6.04280770e-01 5.53162813e-01
3.03637028e-01 1.12718141e+00 -3.46845180e-01 -5.06965995e-01
5.51087916e-01 4.27722394e-01 1.88609704e-01 2.71385759e-01
-3.50015372e-01 2.67950296e-01 -3.43874067e-01 4.15819824e-01
7.26832449e-01 2.34042391e-01 -2.04200447e-01 2.92657427e-02
-2.87960842e-02 1.21564671e-01 -9.15866673e-01 1.53841197e+00
-3.74557167e-01 1.13804579e+00 -3.24842423e-01 -1.47348642e+00
1.15565372e+00 3.82871896e-01 -2.38954097e-01 -8.52997065e-01
6.15178108e-01 3.42346311e-01 -6.02159500e-02 -6.77534461e-01
8.13822091e-01 1.12060733e-01 -3.13808292e-01 4.26768571e-01
3.43409866e-01 -5.54197848e-01 4.79595393e-01 1.56015502e-02
4.77143168e-01 3.44417304e-01 1.92120913e-02 -1.65963292e-01
6.15765691e-01 2.71057934e-01 -9.32968594e-03 8.07033718e-01
3.89248575e-03 1.04076731e+00 4.91650194e-01 -5.42786062e-01
-1.72965705e+00 -8.17342877e-01 -2.64930010e-01 1.31212568e+00
-4.70826149e-01 4.83264141e-02 -1.06101894e+00 -5.50092161e-01
-5.77221632e-01 8.69283497e-01 -6.00127935e-01 7.69855082e-02
-8.11823547e-01 -5.82304001e-01 1.12545884e+00 3.96124244e-01
9.70886886e-01 -1.43403411e+00 -5.29922783e-01 1.84751049e-01
3.29133589e-04 -1.37171602e+00 -7.52534866e-01 4.61065739e-01
-5.13752341e-01 -5.31050801e-01 -1.26628113e+00 -1.20329809e+00
7.13788688e-01 -2.11341664e-01 7.70188749e-01 -1.83778837e-01
-6.47684693e-01 2.23212987e-01 -5.77767015e-01 -4.58061755e-01
-6.75334334e-01 2.96970367e-01 -3.24049324e-01 1.55457258e-01
3.32516402e-01 9.79144722e-02 -2.03902721e-01 6.92915767e-02
-1.08538198e+00 3.43238950e-01 4.53546643e-01 9.96254921e-01
2.74301767e-01 -2.67891467e-01 2.92363495e-01 -7.69532382e-01
5.53718805e-01 -3.88936587e-02 -7.16548800e-01 2.71876961e-01
2.71652460e-01 -1.88712642e-01 1.02216887e+00 -5.68221092e-01
-9.53883231e-01 2.56076783e-01 -2.35767305e-01 -1.60400838e-01
-3.93640131e-01 3.46789330e-01 1.06532574e-02 2.39406750e-01
6.90665245e-01 9.06942189e-01 -5.26125789e-01 -5.36275990e-02
5.16811132e-01 1.34202659e+00 7.29298115e-01 -3.40847075e-01
7.93943524e-01 8.29713866e-02 -4.98228520e-01 -1.43113661e+00
-6.10607386e-01 -5.78106791e-02 -9.79634225e-01 -1.68406531e-01
1.22167563e+00 -7.16554523e-01 -2.30244651e-01 9.82798338e-01
-1.35805964e+00 -6.34702861e-01 -1.95155144e-01 2.80077726e-01
-6.33077621e-01 3.37837547e-01 -8.25346768e-01 -7.40651190e-01
-6.05851054e-01 -1.34640718e+00 1.19700456e+00 3.00987273e-01
-1.16459340e-01 -9.28054810e-01 -1.58191204e-01 5.05451977e-01
3.62488836e-01 7.20171779e-02 9.10586655e-01 -9.10860658e-01
-3.04455966e-01 -3.33859682e-01 -3.41052294e-01 7.88854241e-01
-4.63023186e-02 2.32445985e-01 -1.06949961e+00 -2.64912814e-01
-2.31821965e-02 -8.78018916e-01 8.31015468e-01 2.05668092e-01
1.15409350e+00 -3.92312735e-01 3.40972275e-01 6.23277247e-01
1.50014818e+00 2.67546445e-01 8.76678467e-01 3.97031456e-01
9.36749101e-01 4.23442632e-01 1.91549063e-01 1.55663222e-01
-6.73508793e-02 3.84293795e-01 -7.13179335e-02 -2.64387369e-01
-2.61919498e-01 -9.15549621e-02 5.69804490e-01 1.27863061e+00
4.85428691e-01 -7.34285593e-01 -1.18922174e+00 7.07674384e-01
-1.35336757e+00 -6.47720754e-01 -4.40921098e-01 1.85002697e+00
9.22214448e-01 8.39395523e-02 -1.71790093e-01 2.60086209e-01
7.30451286e-01 3.42359006e-01 -4.48544532e-01 -1.32982397e+00
-5.72451234e-01 4.28160578e-01 6.91333115e-01 4.61229295e-01
-1.01904285e+00 1.62857008e+00 5.50715160e+00 1.27814353e+00
-1.55516303e+00 -8.64162594e-02 9.58298564e-01 2.04605281e-01
2.83439159e-01 -3.27472359e-01 -5.36564887e-01 3.15418243e-01
1.21453774e+00 6.92070499e-02 3.54436994e-01 6.15133464e-01
3.41851324e-01 -1.26634300e-01 -7.43182421e-01 9.77019846e-01
3.73440444e-01 -1.24304748e+00 4.37206745e-01 -2.49250233e-01
8.81632745e-01 3.93495679e-01 1.89535201e-01 2.90107101e-01
2.12683082e-01 -1.61846459e+00 9.47649002e-01 1.61077023e-01
1.26849723e+00 -7.50879884e-01 7.24380970e-01 2.39828840e-01
-9.53518391e-01 2.47066572e-01 -8.26854825e-01 1.14297003e-01
-1.34250164e-01 1.84379518e-01 -1.08932865e+00 3.74790877e-01
3.61621827e-01 6.50791287e-01 -7.18445539e-01 4.25639838e-01
-1.52541935e-01 1.19101417e+00 -2.29766637e-01 -3.06048751e-01
8.67006004e-01 -5.01632929e-01 2.76129961e-01 1.89568710e+00
2.70518720e-01 -2.51011848e-01 -6.17379434e-02 7.65036464e-01
-2.98321515e-01 4.74118471e-01 -7.96422482e-01 -4.45095837e-01
-8.06869939e-02 1.20145786e+00 -1.04807115e+00 -6.86264157e-01
-1.40549928e-01 1.60688889e+00 -3.16740684e-02 5.38297236e-01
-7.39371717e-01 -8.52635264e-01 -3.37810591e-02 -5.05679786e-01
6.50393248e-01 -3.09750348e-01 -4.50178444e-01 -1.17542481e+00
1.95017681e-02 -1.27143228e+00 -3.27251963e-02 -1.07763088e+00
-9.75449026e-01 9.37983155e-01 -3.18272233e-01 -9.43844318e-01
-2.88191408e-01 -1.02766204e+00 -7.10267603e-01 1.09970188e+00
-1.00252879e+00 -1.33403599e+00 -1.37983292e-01 6.24928951e-01
1.40198982e+00 -6.81847274e-01 8.67115855e-01 3.72023731e-02
-7.35933006e-01 6.79161370e-01 6.12469435e-01 9.14794803e-01
4.75129962e-01 -1.21164250e+00 7.79153883e-01 8.85311723e-01
3.82759660e-01 1.08366281e-01 6.21489584e-01 -4.30844665e-01
-1.42105365e+00 -1.26058865e+00 6.95720851e-01 -4.33610499e-01
5.88000655e-01 -1.05643094e+00 -7.99488008e-01 7.73125052e-01
8.68724048e-01 -1.95496067e-01 3.67845714e-01 -9.94100511e-01
-4.29228008e-01 2.74008512e-01 -8.28728020e-01 8.03505719e-01
3.21505904e-01 -6.24850869e-01 -6.73556983e-01 6.80196285e-01
4.71311122e-01 -4.45375532e-01 -5.00084817e-01 -2.92222023e-01
4.29808676e-01 -6.06106400e-01 5.94823241e-01 -5.72804868e-01
7.14642882e-01 -8.32058787e-02 -3.66867334e-01 -1.05500734e+00
3.70377809e-01 -5.83230734e-01 7.30698466e-01 1.28285527e+00
6.36984289e-01 -4.72748846e-01 7.06430912e-01 -5.04512899e-02
-1.23607684e-02 -2.40860507e-01 -1.07504368e+00 -6.65769994e-01
6.33106411e-01 -6.03601694e-01 1.16863459e-01 1.00651717e+00
-4.73287702e-01 6.57387257e-01 -7.22386599e-01 -4.19731081e-01
4.47975367e-01 -1.55270219e-01 7.20580399e-01 -4.45148528e-01
6.09565638e-02 -2.88255781e-01 -3.97917569e-01 -1.10971773e+00
2.15545207e-01 -1.00169313e+00 2.43546739e-01 -1.42464280e+00
2.87215747e-02 7.11395964e-02 3.69190693e-01 4.41197753e-01
1.57579586e-01 5.70312202e-01 4.17968392e-01 9.76228341e-02
-2.89718390e-01 5.25377572e-01 1.24878347e+00 -5.55847764e-01
1.27852082e-01 -3.81328017e-01 1.08484346e-02 3.83794039e-01
1.06949723e+00 -5.01535356e-01 -2.24991992e-01 -7.18009591e-01
-1.17995434e-01 -5.52786477e-02 -4.69312258e-02 -6.52517855e-01
1.33369252e-01 2.22444646e-02 7.37660110e-01 -9.65735137e-01
2.36579522e-01 -5.78392863e-01 -2.86415130e-01 3.90740693e-01
-5.50480604e-01 4.69728149e-02 2.24201411e-01 -3.01922802e-02
-1.42641410e-01 -6.41189277e-01 1.06650925e+00 -2.86850035e-01
-5.13557196e-01 1.57438442e-02 -9.81019318e-01 1.68766648e-01
7.93951035e-01 -7.09277451e-01 -2.17117593e-01 -4.64783758e-01
-2.15702474e-01 -1.75551623e-02 2.51924545e-01 5.88184118e-01
7.06815481e-01 -1.09189522e+00 -1.27997971e+00 2.87716329e-01
1.01894572e-01 -9.23976600e-02 -3.53178494e-02 5.26650369e-01
-1.26024806e+00 8.03294837e-01 -3.84371400e-01 -8.02830100e-01
-1.17309344e+00 2.63349026e-01 5.14676869e-01 -5.06760404e-02
-5.23095071e-01 7.53458738e-01 1.02000877e-01 -8.61559927e-01
1.81978911e-01 -2.31396824e-01 -1.20644331e-01 -7.00671598e-02
4.11302418e-01 5.85738942e-02 2.90229190e-02 -1.09228206e+00
7.63118081e-03 8.56896818e-01 -2.94553101e-01 -3.54662061e-01
1.20728993e+00 -1.58303604e-02 -1.63633153e-01 6.16916120e-01
1.63872612e+00 -2.35258445e-01 -9.30979609e-01 1.49090700e-02
1.06529370e-01 -1.76712424e-01 -1.24971911e-01 -8.39047551e-01
-1.04027617e+00 1.31056881e+00 7.22685933e-01 -6.77516386e-02
9.99672592e-01 -5.07812083e-01 8.06929171e-01 5.86458087e-01
-2.09894821e-01 -1.38926065e+00 1.94262385e-01 1.09757030e+00
1.07893360e+00 -1.10833192e+00 -2.61719435e-01 2.67911822e-01
-1.04462683e+00 1.56751394e+00 4.20868337e-01 -2.71031857e-01
3.00315637e-02 3.84944350e-01 5.43067694e-01 1.19637161e-01
-4.36145931e-01 -3.35031450e-02 4.57609631e-02 7.19125867e-01
5.70393741e-01 -2.87034549e-02 -1.08916163e-01 -6.57276362e-02
-5.21767080e-01 -3.14169049e-01 9.70845044e-01 7.87226558e-01
-3.13144386e-01 -8.55414271e-01 -5.92838526e-01 5.57224214e-01
-5.93382061e-01 -3.80462438e-01 -7.13326752e-01 7.34637201e-01
-2.13789746e-01 7.00071156e-01 1.12782337e-01 -1.26679897e-01
2.14127019e-01 2.87673146e-01 2.21860811e-01 -4.85054016e-01
-8.06598067e-01 3.18735003e-01 2.72838194e-02 3.46964337e-02
2.79392209e-02 -3.97576272e-01 -1.25317001e+00 -1.87037528e-01
-2.18935609e-01 3.06274593e-02 8.69765937e-01 8.96580517e-01
-3.12257797e-01 4.98685658e-01 7.16951489e-01 -8.84652257e-01
-3.61800164e-01 -1.33922017e+00 -3.90008062e-01 1.88521579e-01
2.21882969e-01 2.52896935e-01 -2.04846904e-01 4.78646070e-01]
|
[11.881718635559082, 2.3476946353912354]
|
81d22d8a-2b6f-4c30-9158-7b682dcb5331
|
guided-generative-adversarial-neural-network
|
2003.02836
| null |
https://arxiv.org/abs/2003.02836v2
|
https://arxiv.org/pdf/2003.02836v2.pdf
|
Guided Generative Adversarial Neural Network for Representation Learning and High Fidelity Audio Generation using Fewer Labelled Audio Data
|
Recent improvements in Generative Adversarial Neural Networks (GANs) have shown their ability to generate higher quality samples as well as to learn good representations for transfer learning. Most of the representation learning methods based on GANs learn representations ignoring their post-use scenario, which can lead to increased generalisation ability. However, the model can become redundant if it is intended for a specific task. For example, assume we have a vast unlabelled audio dataset, and we want to learn a representation from this dataset so that it can be used to improve the emotion recognition performance of a small labelled audio dataset. During the representation learning training, if the model does not know the post emotion recognition task, it can completely ignore emotion-related characteristics in the learnt representation. This is a fundamental challenge for any unsupervised representation learning model. In this paper, we aim to address this challenge by proposing a novel GAN framework: Guided Generative Neural Network (GGAN), which guides a GAN to focus on learning desired representations and generating superior quality samples for audio data leveraging fewer labelled samples. Experimental results show that using a very small amount of labelled data as guidance, a GGAN learns significantly better representations.
|
['Björn Schuller', 'John H. L. Hansen', 'Rajib Rana', 'Kazi Nazmul Haque']
|
2020-03-05
| null | null | null | null |
['audio-generation']
|
['audio']
|
[ 8.51065218e-01 4.95573610e-01 1.37696594e-01 -4.37249362e-01
-8.22238445e-01 -3.87471795e-01 3.62449408e-01 -3.83984029e-01
1.83010861e-01 8.78407776e-01 2.64489651e-01 1.61924496e-01
3.28077525e-01 -1.12158012e+00 -7.07024217e-01 -9.06403124e-01
2.18694210e-01 4.41169500e-01 -4.19100016e-01 -3.48087043e-01
-2.22331151e-01 2.66800761e-01 -1.59618008e+00 6.06441259e-01
8.36467505e-01 1.08179486e+00 9.58342478e-02 7.41038620e-01
-1.49043024e-01 1.07532465e+00 -1.16100121e+00 -2.64596283e-01
2.04660028e-01 -9.76853132e-01 -6.47808373e-01 1.15108292e-03
2.35792119e-02 -1.67148814e-01 -1.55188680e-01 9.12316680e-01
8.00857425e-01 2.93034583e-01 8.40867877e-01 -1.47658348e+00
-6.31787121e-01 6.74455464e-01 -1.65965229e-01 -2.68797576e-01
1.79373339e-01 5.11331111e-02 8.16075861e-01 -5.14510810e-01
2.88947970e-01 1.11718142e+00 4.42160368e-01 1.23353851e+00
-1.12118363e+00 -1.04531157e+00 1.97648644e-01 3.56206670e-02
-1.06621814e+00 -4.85031039e-01 1.34245849e+00 -1.26646385e-01
4.27185327e-01 3.27676862e-01 7.19669878e-01 1.55184257e+00
-1.54626369e-01 9.28229272e-01 1.05996883e+00 -4.28462893e-01
4.97486800e-01 1.82247639e-01 -4.93706614e-01 4.25310619e-02
-2.82691151e-01 1.93664491e-01 -5.26066184e-01 4.32048319e-03
6.47593439e-01 1.87829465e-01 -5.50737023e-01 -2.84285605e-01
-6.58000946e-01 1.12404811e+00 6.18184149e-01 2.21539304e-01
-6.03532791e-01 2.16491446e-01 4.47544962e-01 6.36173964e-01
5.82923591e-01 5.01052558e-01 -2.97865778e-01 -4.44441557e-01
-8.33883166e-01 4.47263829e-02 6.75724685e-01 7.37922907e-01
7.75388300e-01 7.70255804e-01 -1.15273632e-01 9.82206166e-01
-8.66248179e-03 4.49498236e-01 9.74192023e-01 -5.95815539e-01
3.16127807e-01 6.08414292e-01 -4.07492906e-01 -7.91537344e-01
1.02215633e-01 -6.01866245e-01 -1.25428569e+00 3.54792714e-01
-4.35647517e-02 -3.48014951e-01 -1.15883458e+00 1.94859540e+00
9.53948423e-02 3.17224145e-01 5.37208736e-01 8.03391576e-01
8.16196501e-01 9.32994306e-01 -1.20131010e-02 -1.27683356e-01
7.99203336e-01 -8.08134198e-01 -7.15256214e-01 -4.50998306e-01
5.46022952e-01 -5.07847726e-01 9.64757204e-01 6.23612523e-01
-8.01488161e-01 -7.89798141e-01 -1.01251197e+00 4.13615108e-01
-3.09535086e-01 -6.46774843e-02 5.21834612e-01 7.72994816e-01
-8.84012163e-01 3.57299000e-01 -4.50914651e-01 5.64821670e-03
6.49750829e-01 2.45982051e-01 -2.44833484e-01 -4.57376182e-01
-1.21120536e+00 3.95123631e-01 4.89877909e-01 6.09641708e-02
-1.32531869e+00 -5.82042217e-01 -8.14740121e-01 2.62032449e-01
2.98242182e-01 -4.82338637e-01 1.25122392e+00 -1.79998648e+00
-1.70567095e+00 3.38521332e-01 1.67400286e-01 -4.90991384e-01
3.78214210e-01 -2.27326259e-01 -5.64816236e-01 9.34550762e-02
-1.81885630e-01 6.99217141e-01 1.38891292e+00 -1.57969677e+00
-3.64798188e-01 -1.52554139e-01 6.23818748e-02 1.26846179e-01
-4.31224734e-01 -4.03145313e-01 1.08556725e-01 -1.05786622e+00
-2.04398736e-01 -9.29336369e-01 -2.03981459e-01 -3.39197218e-01
-3.29532593e-01 2.93705687e-02 1.08947289e+00 -5.85709393e-01
7.70262957e-01 -2.36894751e+00 1.33389041e-01 3.12133461e-01
-2.76857819e-02 5.26650548e-01 -4.30798531e-01 4.34863627e-01
-4.67276454e-01 9.72790942e-02 -3.33513647e-01 -2.82984734e-01
-2.01639101e-01 4.97048408e-01 -7.20120966e-01 -9.73748341e-02
5.72494984e-01 9.47316110e-01 -9.50401366e-01 2.26550493e-02
5.74488975e-02 7.39952266e-01 -5.99494874e-01 7.53535211e-01
-2.52013117e-01 7.18904197e-01 -4.22194719e-01 4.29901659e-01
4.24764276e-01 2.64350027e-01 8.49113464e-02 -3.67810950e-02
6.31841481e-01 2.22281083e-01 -1.02750337e+00 1.62050402e+00
-9.18273091e-01 4.84888464e-01 -1.92843929e-01 -1.47606242e+00
1.37377095e+00 5.81623495e-01 3.39505732e-01 -6.81829691e-01
1.59444995e-02 4.73917462e-02 1.09152220e-01 -1.53103992e-01
1.75364375e-01 -5.27863562e-01 -7.30333105e-02 6.45001769e-01
1.40306041e-01 -3.39483082e-01 -3.58302712e-01 8.26856270e-02
1.03701234e+00 5.37242778e-02 2.97374278e-02 3.83408546e-01
3.16978574e-01 -3.94179404e-01 7.69932985e-01 3.95350575e-01
3.29963952e-01 9.27118897e-01 3.41305763e-01 -3.34012449e-01
-7.44529605e-01 -7.43940890e-01 4.03808624e-01 1.05514264e+00
-4.97159928e-01 -3.89626622e-01 -7.09954739e-01 -9.58748817e-01
-5.76189756e-01 8.63985479e-01 -8.55821311e-01 -9.18550253e-01
-3.72364730e-01 -5.48164785e-01 4.58217710e-01 7.20579803e-01
5.03540874e-01 -1.54403412e+00 -7.16744721e-01 3.57238889e-01
-1.33906469e-01 -6.89115524e-01 -3.63743119e-02 4.79838789e-01
-8.85403931e-01 -8.34121525e-01 -9.26733077e-01 -6.99408948e-01
8.71677279e-01 -1.39170960e-02 1.08098137e+00 -5.10957055e-02
1.16166966e-02 4.01427865e-01 -9.38993573e-01 -8.29657435e-01
-7.17266083e-01 1.11841723e-01 -1.19040735e-01 3.67422968e-01
2.26543784e-01 -8.99677932e-01 -2.83558577e-01 8.01886767e-02
-1.10660887e+00 1.01569230e-02 7.39145815e-01 1.20625293e+00
5.11678100e-01 3.66334230e-01 1.18613100e+00 -1.18656468e+00
7.88073778e-01 -5.32442868e-01 1.47203356e-01 4.01108749e-02
-3.78746510e-01 -1.38247311e-02 1.00174570e+00 -7.92759001e-01
-9.97956097e-01 3.47531475e-02 -3.54074925e-01 -8.06510210e-01
-3.38353783e-01 5.47500432e-01 -5.21574497e-01 2.08308294e-01
6.38451815e-01 5.15484750e-01 1.01652429e-01 -4.35178131e-01
3.50473344e-01 6.98951304e-01 3.94398570e-01 -6.26042783e-01
7.99496293e-01 6.88637495e-02 -1.99695498e-01 -5.09196222e-01
-7.68103719e-01 -5.33339269e-02 -3.36773902e-01 -1.94104031e-01
4.84438956e-01 -9.74766970e-01 -7.66853942e-03 2.87983000e-01
-8.17588449e-01 -5.81350744e-01 -9.94120181e-01 3.33253831e-01
-7.49416411e-01 -1.66703030e-01 -4.54699844e-02 -1.10454321e+00
-4.77782041e-01 -1.06925130e+00 9.42274749e-01 1.80872947e-01
-3.06472838e-01 -9.51344013e-01 1.94171369e-01 8.17905366e-02
6.61320210e-01 4.47961539e-01 9.08525705e-01 -9.78357911e-01
-1.25471652e-01 -4.21586365e-01 2.63166875e-01 9.86736417e-01
4.45636541e-01 -2.99946189e-01 -1.45815766e+00 -3.97373229e-01
2.18143940e-01 -7.39551842e-01 7.44267344e-01 8.54201838e-02
1.42193139e+00 -6.13201916e-01 1.29794151e-01 6.10018432e-01
1.12822092e+00 4.80275452e-01 9.77011621e-01 4.63007465e-02
7.05163598e-01 6.47659779e-01 6.61257863e-01 3.31661016e-01
-1.14393100e-01 4.68405426e-01 6.58998668e-01 -2.64307767e-01
-1.88715607e-01 -5.14064848e-01 6.59131944e-01 9.21679676e-01
-1.75267607e-01 -3.03361923e-01 -5.57435393e-01 5.58863521e-01
-1.72048247e+00 -8.96753371e-01 5.24839461e-01 2.08524823e+00
7.80706882e-01 -1.21531628e-01 9.92790088e-02 6.82450771e-01
5.18373072e-01 8.31684768e-02 -6.62113488e-01 -5.77597439e-01
1.16016865e-01 7.34205544e-01 -2.57269859e-01 -1.94261506e-01
-6.42774940e-01 7.43954182e-01 5.75155973e+00 9.18927312e-01
-1.43951154e+00 9.21487585e-02 7.43972957e-01 -1.03077747e-01
-3.85607243e-01 -2.24262580e-01 -2.41178066e-01 5.39709032e-01
1.14387631e+00 -1.99516878e-01 4.47364062e-01 1.06111825e+00
-1.07595153e-01 3.98636192e-01 -1.18234134e+00 1.20301998e+00
3.27213019e-01 -8.37149203e-01 5.08259237e-01 1.07110374e-01
7.87295163e-01 -5.07123351e-01 3.08040947e-01 8.41150224e-01
3.23778927e-01 -1.44719946e+00 4.19718921e-01 6.07817173e-01
8.97590935e-01 -1.29922831e+00 9.83734190e-01 4.11748260e-01
-9.03303444e-01 -5.64901121e-02 -5.02022982e-01 -9.78722796e-02
-1.00684591e-01 5.29907703e-01 -1.07860100e+00 6.45944357e-01
5.08588016e-01 7.28023767e-01 -4.73794609e-01 7.78945863e-01
-4.72951382e-01 9.88622010e-01 -3.36390547e-02 1.93916634e-01
1.48880154e-01 -1.07601598e-01 3.21476668e-01 8.30957651e-01
6.60040557e-01 -3.01083699e-02 6.20562909e-03 7.63859153e-01
-4.31490183e-01 1.44547433e-01 -9.36831117e-01 -3.08023125e-01
2.23875523e-01 1.20569623e+00 -3.33672047e-01 -3.48936886e-01
-7.60470331e-02 1.14314330e+00 2.27128044e-01 4.45054412e-01
-6.33294404e-01 -2.26175517e-01 5.55947006e-01 -9.35818106e-02
3.59674305e-01 3.43174428e-01 -7.89851043e-03 -1.02524674e+00
-1.06562026e-01 -1.34416294e+00 2.56003678e-01 -9.07553673e-01
-1.32571721e+00 8.35959017e-01 -3.68008375e-01 -1.51581788e+00
-8.82813811e-01 -2.44218677e-01 -9.53791380e-01 8.40683341e-01
-1.47088742e+00 -1.34662330e+00 -3.23633522e-01 8.42632413e-01
6.41220093e-01 -6.74728692e-01 1.19498336e+00 8.65843967e-02
-2.57880658e-01 8.25821638e-01 -8.87401849e-02 2.61421770e-01
7.20221281e-01 -1.23713553e+00 4.35912907e-02 6.72630489e-01
5.05398750e-01 2.57833421e-01 4.85029548e-01 -2.90183723e-01
-1.21399057e+00 -1.48975861e+00 2.19931901e-01 -1.21116735e-01
5.64947836e-02 -3.76494080e-01 -1.04899740e+00 6.88901603e-01
1.00827724e-01 4.43185214e-03 1.15859640e+00 9.68164057e-02
-3.16240966e-01 -2.01084673e-01 -1.35843658e+00 2.91321903e-01
7.43519485e-01 -5.74033320e-01 -5.68924069e-01 5.52114397e-02
6.70866847e-01 -1.12147868e-01 -7.53317833e-01 5.00162005e-01
3.93050790e-01 -7.57490695e-01 7.42872179e-01 -7.81831801e-01
6.33940637e-01 1.04536833e-02 -2.27709264e-01 -2.10353303e+00
2.88686082e-02 -4.99608934e-01 -3.67539883e-01 1.70588994e+00
1.30012497e-01 -5.64231277e-01 8.16272616e-01 3.62808555e-01
-2.20984563e-01 -7.91194797e-01 -8.57074738e-01 -7.48121917e-01
8.98628458e-02 -6.32897735e-01 1.02191710e+00 1.04105926e+00
-3.52497071e-01 5.10170519e-01 -7.38732636e-01 -6.16694354e-02
3.03477824e-01 1.74422964e-01 1.16954672e+00 -1.16407800e+00
-3.75363827e-01 -3.03108338e-02 -6.33411348e-01 -8.40173781e-01
2.40834266e-01 -8.64949465e-01 1.51117772e-01 -1.31112683e+00
-1.45563856e-01 -6.42632365e-01 -4.59185928e-01 7.63352334e-01
-5.37438169e-02 4.84311581e-01 3.02500099e-01 -3.60358804e-02
-1.82140395e-01 9.89023864e-01 1.25925517e+00 -3.97364497e-01
-2.10555747e-01 2.91000545e-01 -1.01528287e+00 4.74432766e-01
9.91230845e-01 -6.02672279e-01 -8.65940750e-01 -1.51268035e-01
9.83402580e-02 -8.93528312e-02 3.67489845e-01 -1.14212072e+00
-1.32101640e-01 6.75210282e-02 6.11701250e-01 -1.21170387e-01
5.78001916e-01 -1.07120812e+00 3.96627575e-01 2.61559874e-01
-4.28861916e-01 -3.61321628e-01 1.04818627e-01 5.87868094e-01
-6.81300759e-01 -4.00121331e-01 6.34636343e-01 -5.12405224e-02
-5.29306710e-01 2.98821181e-01 -1.86330095e-01 9.62323621e-02
9.88219440e-01 -2.47355983e-01 3.43907997e-02 -1.00179851e+00
-8.02281499e-01 -1.51117116e-01 3.29677016e-01 5.66039979e-01
8.83091509e-01 -1.67126882e+00 -8.30078602e-01 5.08691251e-01
2.21384272e-01 3.77152383e-01 3.11305821e-01 1.91343009e-01
1.33211404e-01 -7.70656914e-02 -4.08273429e-01 -3.93630534e-01
-1.02642047e+00 6.16656065e-01 1.10493675e-01 -3.14811230e-01
-6.06227160e-01 8.20287347e-01 4.01669085e-01 -4.38226044e-01
2.37500612e-02 -1.78225264e-01 -2.47930229e-01 1.53579324e-01
6.60686255e-01 -3.48037891e-02 7.33647421e-02 -5.56559503e-01
1.62828088e-01 2.48548493e-01 -1.49543151e-01 1.05377473e-01
1.58196795e+00 2.81289816e-01 2.91392267e-01 6.09249055e-01
1.21261430e+00 -3.29404920e-02 -1.24002206e+00 -1.73030660e-01
-6.30319893e-01 -4.94382918e-01 -7.48449378e-03 -8.65553916e-01
-1.69112062e+00 1.16035199e+00 6.84724450e-01 3.26808482e-01
1.57141936e+00 -1.41631335e-01 8.27730596e-01 2.34183431e-01
3.75445276e-01 -8.91122043e-01 3.99926007e-01 3.09012711e-01
1.19572973e+00 -9.82102394e-01 -4.09042925e-01 -1.55088440e-01
-8.99428606e-01 1.12849522e+00 5.87013781e-01 -2.06407726e-01
3.50168824e-01 1.25772402e-01 1.75061896e-01 -2.11825818e-02
-7.29333401e-01 -7.98085183e-02 3.13284248e-01 1.22997415e+00
2.77840525e-01 5.31641245e-02 4.48404163e-01 8.17550600e-01
-6.35398388e-01 -2.21181195e-02 4.11320150e-01 8.49497199e-01
-1.86801478e-01 -1.35720301e+00 -2.45841116e-01 6.15170658e-01
-2.79924989e-01 -1.26822479e-02 -5.12750030e-01 4.24554795e-01
1.25100672e-01 8.07726085e-01 1.70662757e-02 -5.85056603e-01
2.48665273e-01 4.18333203e-01 2.59690374e-01 -9.26520705e-01
-5.12520671e-01 -5.46207279e-02 -2.41656252e-03 -3.09188873e-01
-5.35514295e-01 -2.98222214e-01 -9.35091734e-01 1.18461266e-01
-2.29547605e-01 4.25986320e-01 5.00100672e-01 7.78798401e-01
2.54222035e-01 1.06661522e+00 9.53981578e-01 -8.39669347e-01
-4.20530260e-01 -1.10789263e+00 -6.72852039e-01 5.74824274e-01
2.06257746e-01 -5.59666991e-01 -4.25153583e-01 2.75946110e-01]
|
[11.784672737121582, 0.127290740609169]
|
6dc9b0c2-2c35-41ca-83f9-049dd0a0085e
|
character-based-neural-networks-for-sentence
|
1805.08297
| null |
http://arxiv.org/abs/1805.08297v1
|
http://arxiv.org/pdf/1805.08297v1.pdf
|
Character-based Neural Networks for Sentence Pair Modeling
|
Sentence pair modeling is critical for many NLP tasks, such as paraphrase
identification, semantic textual similarity, and natural language inference.
Most state-of-the-art neural models for these tasks rely on pretrained word
embedding and compose sentence-level semantics in varied ways; however, few
works have attempted to verify whether we really need pretrained embeddings in
these tasks. In this paper, we study how effective subword-level (character and
character n-gram) representations are in sentence pair modeling. Though it is
well-known that subword models are effective in tasks with single sentence
input, including language modeling and machine translation, they have not been
systematically studied in sentence pair modeling tasks where the semantic and
string similarities between texts matter. Our experiments show that subword
models without any pretrained word embedding can achieve new state-of-the-art
results on two social media datasets and competitive results on news data for
paraphrase identification.
|
['Wuwei Lan', 'Wei Xu']
|
2018-05-21
|
character-based-neural-networks-for-sentence-1
|
https://aclanthology.org/N18-2025
|
https://aclanthology.org/N18-2025.pdf
|
naacl-2018-6
|
['sentence-pair-modeling']
|
['natural-language-processing']
|
[ 2.48354629e-01 -2.73036152e-01 -4.33532357e-01 -5.46240509e-01
-5.59545577e-01 -3.98121864e-01 7.07481086e-01 8.75586689e-01
-5.91839790e-01 2.08906174e-01 7.41711557e-01 -5.75724781e-01
1.23255871e-01 -8.00313473e-01 -6.44388318e-01 -8.97730589e-02
3.09364229e-01 3.84403825e-01 1.50719538e-01 -6.61168396e-01
6.32227957e-01 1.44458516e-02 -1.36983740e+00 7.92663038e-01
8.56416285e-01 6.58128440e-01 2.13752866e-01 6.52877152e-01
-6.31265819e-01 6.48293197e-01 -3.70449215e-01 -7.68776596e-01
-9.94856283e-02 -6.53803766e-01 -1.03235149e+00 -4.69092607e-01
6.55567408e-01 -2.12872121e-02 -5.58798850e-01 1.23562622e+00
6.09672725e-01 1.07330911e-01 8.59317005e-01 -8.31136525e-01
-1.41780937e+00 9.51139152e-01 -3.39264244e-01 6.53334200e-01
8.53135884e-01 -1.40490040e-01 1.59617960e+00 -1.10786510e+00
3.44064474e-01 1.35125828e+00 9.62985933e-01 4.52269316e-01
-1.16613483e+00 -3.86504024e-01 -1.06874526e-01 8.49025011e-01
-8.03827763e-01 -1.08096808e-01 8.94118607e-01 -1.47615567e-01
1.48089659e+00 2.81581491e-01 3.54162097e-01 1.58501005e+00
3.99390072e-01 8.81303608e-01 7.73104250e-01 -7.46549129e-01
-1.05640076e-01 3.22557181e-01 8.53250980e-01 4.56280947e-01
1.13454528e-01 -5.56194365e-01 -5.77116549e-01 -2.84963816e-01
3.49894196e-01 2.06467867e-01 -2.73699373e-01 -1.01653092e-01
-1.22343540e+00 1.20469892e+00 3.16386700e-01 7.97353923e-01
-7.60501847e-02 -5.69580533e-02 9.45140064e-01 8.97708893e-01
6.65928543e-01 9.78805363e-01 -3.51968318e-01 -2.42620662e-01
-7.65422165e-01 1.69091269e-01 7.33583450e-01 8.12354565e-01
3.36407214e-01 -3.73053074e-01 -3.32026958e-01 1.33874130e+00
-7.67435730e-02 1.57529444e-01 1.30262887e+00 -4.50718284e-01
7.94103801e-01 6.38981104e-01 -3.38678926e-01 -1.24429142e+00
-1.62574634e-01 -3.99501055e-01 -7.64335394e-01 -6.93452299e-01
2.57342696e-01 1.74806759e-01 -2.51675129e-01 1.54407084e+00
-3.76396090e-01 3.78090143e-01 2.48358652e-01 6.81115329e-01
1.23274136e+00 6.48608387e-01 -2.17030451e-01 2.90941056e-02
1.53367078e+00 -1.33249342e+00 -4.85748380e-01 -7.30848014e-01
1.00925589e+00 -9.36761856e-01 1.55100679e+00 -2.46106982e-01
-1.26480460e+00 -8.07005942e-01 -9.93833840e-01 -4.75707054e-01
-5.36736429e-01 -2.89711565e-01 4.44833666e-01 3.61486107e-01
-7.10713685e-01 9.04823124e-01 -2.57084370e-01 -7.87551403e-01
1.35567427e-01 -9.40383747e-02 -3.87293577e-01 -2.71545500e-01
-1.73219395e+00 1.46632767e+00 7.96594694e-02 -3.84571463e-01
-2.11234629e-01 -8.10910523e-01 -1.00059235e+00 4.52895761e-01
-1.53958425e-01 -9.74867165e-01 1.20103681e+00 -8.25551510e-01
-1.19839394e+00 1.30130780e+00 -4.39358711e-01 -7.71726131e-01
3.04541234e-02 -1.65939450e-01 -3.48965019e-01 2.90988777e-02
5.48907779e-02 2.40596339e-01 6.66738272e-01 -4.42350507e-01
2.09212210e-02 -3.71047050e-01 2.26909891e-01 2.83436507e-01
-1.01558435e+00 5.11454642e-01 1.05786271e-01 -7.07921684e-01
-3.12571041e-02 -5.89226902e-01 -1.09734319e-01 -1.29320277e-02
-3.79487127e-02 -7.59120762e-01 3.85193676e-01 -7.07950056e-01
1.30226612e+00 -2.04503179e+00 2.61968613e-01 -4.49880630e-01
-4.41104062e-02 4.41674381e-01 -6.09241068e-01 1.08657467e+00
-1.86930776e-01 1.73375055e-01 -1.47611156e-01 -5.54235041e-01
1.47559017e-01 -1.48000553e-01 -6.34394825e-01 2.04084635e-01
6.37612566e-02 1.21865761e+00 -1.12072039e+00 -4.56763357e-01
6.73516169e-02 1.20844364e-01 -4.28300291e-01 3.29394013e-01
5.82003407e-02 -3.24545801e-01 -2.20318809e-01 1.88029647e-01
3.51649076e-01 -2.28468820e-01 9.87380370e-02 -6.76051155e-02
4.51660037e-01 9.12789166e-01 -3.96682650e-01 1.97869873e+00
-8.06497574e-01 7.99948752e-01 -5.18761575e-01 -1.40558577e+00
8.71861875e-01 1.47136986e-01 1.08849786e-01 -6.87854946e-01
1.68324053e-01 1.66594669e-01 2.09903643e-01 -7.37983227e-01
6.69518530e-01 -2.84241170e-01 -2.14232266e-01 8.15246642e-01
1.23505034e-01 -2.93253630e-01 1.53431132e-01 3.11841875e-01
1.15076363e+00 -4.90571260e-01 4.91339445e-01 -2.52669185e-01
7.19323337e-01 -2.16046408e-01 -6.50908574e-02 7.56803989e-01
-1.91172794e-01 8.07323039e-01 4.91698742e-01 -2.84776270e-01
-1.14608848e+00 -1.03906405e+00 -1.26783192e-01 1.30668914e+00
2.97328562e-01 -6.37476742e-01 -6.59763515e-01 -5.87704360e-01
1.48349121e-01 1.02139866e+00 -5.70760250e-01 -6.06766045e-01
-6.24597788e-01 -5.40419698e-01 6.86023176e-01 7.68284619e-01
7.42622986e-02 -1.04153144e+00 9.58518907e-02 2.58974642e-01
-2.86323786e-01 -1.13589621e+00 -8.26113880e-01 -1.11702897e-01
-1.08207071e+00 -9.59779322e-01 -6.36873186e-01 -1.29354501e+00
3.18029225e-01 7.32772529e-01 1.39668751e+00 3.46478559e-02
-1.85163274e-01 4.32166941e-02 -7.07899809e-01 -2.22169682e-01
-5.75768054e-01 2.66702920e-01 1.53385669e-01 -3.00164193e-01
1.04618359e+00 -7.80163586e-01 -3.75792116e-01 6.13552034e-02
-6.45443916e-01 -1.36008710e-01 3.63276958e-01 1.15312338e+00
5.07110953e-02 -7.54571915e-01 8.53741348e-01 -7.50516236e-01
1.51465988e+00 -6.89757824e-01 3.85688215e-01 6.04566813e-01
-3.98561865e-01 1.14871420e-01 1.10914004e+00 -5.72542429e-01
-3.93088102e-01 -7.74999082e-01 -3.15064758e-01 -3.15493971e-01
-8.84941593e-02 6.79206610e-01 3.61784458e-01 2.04976723e-01
8.05155933e-01 7.07730770e-01 2.45417312e-01 -4.92905110e-01
4.50068235e-01 1.16947603e+00 2.69471288e-01 -3.22998375e-01
4.48811620e-01 -4.54513840e-02 -3.43758255e-01 -7.83658803e-01
-1.14741898e+00 -8.60373259e-01 -4.92077023e-01 3.82330269e-01
6.94265187e-01 -6.66173756e-01 -3.69087249e-01 2.19666287e-01
-1.50139868e+00 3.16167980e-01 -1.87816396e-01 3.84576559e-01
-4.92694288e-01 9.97373879e-01 -7.95041859e-01 -1.84682012e-01
-6.69173300e-01 -8.66012752e-01 7.78320551e-01 -1.14034928e-01
-7.37330914e-01 -1.37227404e+00 2.75604904e-01 6.55687630e-01
6.63822889e-01 -5.63087046e-01 1.32000685e+00 -1.15526068e+00
1.86640620e-01 -4.44029093e-01 -3.15681458e-01 5.99452376e-01
2.28146195e-01 -4.42303747e-01 -7.57057011e-01 -1.78443357e-01
2.86020517e-01 -6.00723922e-01 1.12558460e+00 1.05632044e-01
1.31814981e+00 -2.34687373e-01 -4.37363200e-02 4.19120699e-01
1.01556551e+00 -5.06446660e-01 5.95915854e-01 3.30632061e-01
4.46723461e-01 6.62743866e-01 2.60035515e-01 1.16439313e-01
2.74698466e-01 6.38650239e-01 2.15204418e-01 4.10372972e-01
-1.93831936e-01 -4.77369070e-01 5.43448985e-01 1.23221254e+00
4.33451027e-01 -2.73403883e-01 -6.63661420e-01 6.97055161e-01
-2.02413344e+00 -1.36873627e+00 -3.38303894e-01 1.80945969e+00
1.12754977e+00 1.09092034e-01 -7.65665397e-02 1.62189409e-01
6.62683547e-01 6.25050604e-01 -3.24313462e-01 -9.90261972e-01
-2.98679739e-01 3.82141352e-01 1.89487282e-02 3.37819844e-01
-8.00487757e-01 8.74873400e-01 5.97098112e+00 9.79159355e-01
-7.88366497e-01 3.17295909e-01 3.00396293e-01 -3.88551466e-02
-7.10037410e-01 -1.86844036e-01 -6.48386419e-01 7.35815525e-01
9.81213987e-01 -5.29224277e-01 1.86531618e-01 8.26489925e-01
5.82302809e-02 2.95826346e-01 -1.58457816e+00 1.05582678e+00
7.62791097e-01 -1.48300552e+00 3.08426946e-01 -5.83384097e-01
5.87586105e-01 1.36627674e-01 -8.81092325e-02 5.92517316e-01
-2.20239326e-01 -1.16173398e+00 1.87444732e-01 2.63586968e-01
2.91789323e-01 -5.74971735e-01 1.04059923e+00 5.74774563e-01
-7.53843486e-01 -6.30500689e-02 -8.76304686e-01 -3.95753413e-01
2.68344015e-01 5.01615345e-01 -7.03784943e-01 3.86667341e-01
2.26169646e-01 1.11441791e+00 -8.58300984e-01 8.49937201e-01
-2.86722183e-01 6.51301861e-01 1.32663772e-01 -7.25309849e-01
3.20788383e-01 -1.68889716e-01 3.78344089e-01 1.35481369e+00
8.59421119e-02 -4.74040747e-01 -1.47719607e-01 7.70285547e-01
-3.20671082e-01 5.14113486e-01 -7.63094127e-01 -3.32747132e-01
3.57805938e-01 8.59308779e-01 -1.35929540e-01 -3.36275131e-01
-7.38304317e-01 1.28992116e+00 6.48117006e-01 -9.24316123e-02
-7.22301543e-01 -6.58930659e-01 9.74613607e-01 1.88225433e-01
-1.20982220e-02 -2.19642833e-01 -5.59280932e-01 -1.51027715e+00
2.06098393e-01 -7.63456643e-01 2.63340414e-01 -7.81787872e-01
-2.25464916e+00 5.06978691e-01 -2.36853704e-01 -1.07325315e+00
-4.40389037e-01 -7.76952803e-01 -1.23826087e+00 8.96874428e-01
-1.36792660e+00 -1.12547553e+00 1.15617141e-02 2.95424342e-01
1.14672422e+00 -5.31047642e-01 1.08881342e+00 2.95784865e-02
-2.48378977e-01 9.32091653e-01 4.26105529e-01 2.43743017e-01
9.03449297e-01 -1.02823091e+00 8.79599810e-01 5.66060364e-01
5.11161506e-01 1.05334210e+00 7.89814770e-01 -2.88670808e-01
-1.38189304e+00 -6.59493983e-01 1.75925291e+00 -5.55399954e-01
1.19434977e+00 -4.50656176e-01 -1.14010954e+00 3.03218275e-01
6.41740382e-01 -1.77519560e-01 9.25578892e-01 5.08290887e-01
-7.24070728e-01 1.42601311e-01 -8.07730258e-01 7.75088370e-01
1.29632580e+00 -1.18497670e+00 -1.45102584e+00 8.45819294e-01
1.03446209e+00 8.16442594e-02 -7.20573664e-01 1.58587173e-01
4.59452838e-01 -8.08534622e-01 1.27166164e+00 -1.48837757e+00
1.24083996e+00 4.21353430e-01 -2.16879413e-01 -1.56043005e+00
-3.57975483e-01 -5.52838594e-02 -3.56750637e-02 1.07484376e+00
4.35567021e-01 -6.53446615e-01 6.85106456e-01 2.49087617e-01
-3.45235586e-01 -1.13970745e+00 -7.66303718e-01 -1.09087574e+00
6.34683847e-01 -2.79042989e-01 4.65847343e-01 1.28404796e+00
7.74918318e-01 8.29963863e-01 -1.52088046e-01 -4.45175171e-01
1.71716169e-01 3.99707824e-01 4.13550854e-01 -9.22774315e-01
-4.66128945e-01 -8.47472847e-01 -5.80197871e-01 -1.15765536e+00
1.00209594e+00 -1.43581629e+00 -4.02701616e-01 -1.80935252e+00
6.68920815e-01 2.90204793e-01 -2.49447390e-01 -9.08951983e-02
-5.45051873e-01 5.04513131e-03 1.42242536e-02 2.88465451e-02
-4.16300327e-01 7.98032761e-01 9.46970522e-01 -4.96825427e-01
3.86029333e-01 7.74079487e-02 -7.16937065e-01 5.66520691e-01
9.37242031e-01 -3.98162365e-01 -4.56520051e-01 -9.78150427e-01
3.62525940e-01 -1.67281762e-01 3.50860864e-01 -4.91178840e-01
2.72979110e-01 -1.34363040e-01 -5.87892123e-02 -2.53271848e-01
4.35022265e-01 -4.37600046e-01 -6.27494335e-01 4.54804748e-01
-9.71324921e-01 5.59484661e-01 -2.03252599e-01 5.21865308e-01
-4.70805168e-01 -1.11441028e+00 6.58520281e-01 -3.10258776e-01
-4.02254075e-01 -9.26098495e-04 -3.32373649e-01 4.46442872e-01
7.55735397e-01 -3.28658849e-01 -3.49781275e-01 -5.08424997e-01
-3.01714361e-01 1.22273803e-01 4.96226937e-01 9.68587220e-01
7.80016422e-01 -1.38025653e+00 -9.80329216e-01 1.27982065e-01
4.54769284e-01 -7.14537799e-01 2.06053853e-01 5.65012932e-01
-2.61074126e-01 5.05594969e-01 -6.89820498e-02 -2.46696964e-01
-1.41421616e+00 6.01974249e-01 1.01980485e-01 -3.61724943e-01
-2.63774574e-01 1.20247924e+00 -2.16099799e-01 -6.85197055e-01
9.44687948e-02 -4.04233575e-01 -2.24935129e-01 2.40298748e-01
5.38058639e-01 1.28915504e-01 1.21487379e-01 -4.76453930e-01
-3.06879371e-01 6.56368136e-01 -3.82144243e-01 3.12097222e-01
1.37449276e+00 1.05939945e-02 -3.25247228e-01 5.43387890e-01
1.77177632e+00 -2.80728221e-01 1.46301389e-02 -4.83836293e-01
7.58284181e-02 -5.52247524e-01 -2.56684482e-01 -2.43534401e-01
-3.59880060e-01 1.39973891e+00 3.15059870e-02 3.41785163e-01
4.99990076e-01 1.04581296e-01 1.32676792e+00 7.31924772e-01
2.09774747e-01 -9.48271453e-01 4.14639741e-01 1.02339303e+00
1.05986714e+00 -1.43763506e+00 -1.54539481e-01 -2.01087549e-01
-5.72963417e-01 1.26379430e+00 5.85176408e-01 -4.62762207e-01
5.01632392e-01 -2.28190362e-01 -3.72802883e-01 1.02733344e-01
-7.99732983e-01 8.83175954e-02 3.69766682e-01 2.66907662e-01
8.20849419e-01 -1.94616556e-01 -7.17115879e-01 7.95547307e-01
-4.24739987e-01 -4.08404261e-01 5.15252531e-01 6.22410476e-01
-5.40575325e-01 -1.33500683e+00 1.86067656e-01 8.36637199e-01
-4.60744172e-01 -6.78113222e-01 -7.34267414e-01 1.05439007e-01
-4.52122837e-01 8.35461140e-01 1.57785460e-01 -4.82081562e-01
2.95633405e-01 3.98229450e-01 6.22858703e-01 -1.12061286e+00
-1.06276739e+00 -1.08650494e+00 4.23207551e-01 -2.32384667e-01
-7.24939778e-02 -5.57752013e-01 -9.12197411e-01 -4.58029985e-01
-3.42187732e-01 2.00362206e-01 5.00786185e-01 1.05877268e+00
4.25916702e-01 1.22740589e-01 6.60902500e-01 -4.99240398e-01
-1.31739378e+00 -1.47878695e+00 -2.82380253e-01 9.75354493e-01
-6.49969280e-02 -7.80477524e-02 -5.34786463e-01 -2.88133770e-01]
|
[11.089353561401367, 8.724143028259277]
|
82b88843-119d-4d91-9610-2c25825b9e3e
|
improving-knowledge-aware-dialogue-generation
|
1912.07491
| null |
https://arxiv.org/abs/1912.07491v1
|
https://arxiv.org/pdf/1912.07491v1.pdf
|
Improving Knowledge-aware Dialogue Generation via Knowledge Base Question Answering
|
Neural network models usually suffer from the challenge of incorporating commonsense knowledge into the open-domain dialogue systems. In this paper, we propose a novel knowledge-aware dialogue generation model (called TransDG), which transfers question representation and knowledge matching abilities from knowledge base question answering (KBQA) task to facilitate the utterance understanding and factual knowledge selection for dialogue generation. In addition, we propose a response guiding attention and a multi-step decoding strategy to steer our model to focus on relevant features for response generation. Experiments on two benchmark datasets demonstrate that our model has robust superiority over compared methods in generating informative and fluent dialogues. Our code is available at https://github.com/siat-nlp/TransDG.
|
['Xiaojiang Liu', 'Jian Wang', 'Junhao Liu', 'Ruifeng Xu', 'Min Yang', 'Kejing He', 'Wei Bi']
|
2019-12-16
| null | null | null | null |
['knowledge-base-question-answering']
|
['natural-language-processing']
|
[ 1.88013881e-01 7.65546739e-01 7.28142709e-02 -5.41679144e-01
-1.02861595e+00 -4.40699458e-01 9.07896876e-01 -2.78733134e-01
-1.99877203e-01 1.36629081e+00 9.01868045e-01 -2.62636274e-01
2.05490306e-01 -9.95195270e-01 -2.53947586e-01 -2.08995283e-01
5.51531434e-01 7.03941822e-01 3.36184772e-03 -9.76550341e-01
2.89161682e-01 -3.23726445e-01 -7.21513212e-01 7.25718081e-01
1.32421756e+00 7.72243500e-01 1.66897699e-01 8.34259450e-01
-4.32550460e-01 1.39375198e+00 -7.75577545e-01 -8.05848718e-01
-1.89918861e-01 -9.37351286e-01 -1.85233605e+00 -1.67707488e-01
-2.17627302e-01 -6.71174705e-01 -4.15608883e-01 8.19933832e-01
7.62077391e-01 7.02360690e-01 6.84893727e-01 -9.04783189e-01
-1.22887146e+00 1.14760923e+00 1.87848046e-01 1.80236518e-01
9.58898425e-01 5.29876113e-01 1.02216434e+00 -9.18404698e-01
5.36625206e-01 1.63584626e+00 2.12423757e-01 1.22309470e+00
-9.12139118e-01 -3.61766458e-01 -2.72980556e-02 6.25387669e-01
-1.01885581e+00 -7.16419399e-01 9.64295745e-01 -1.20878503e-01
1.04324317e+00 2.93647379e-01 3.35918248e-01 1.38596725e+00
-2.97090262e-01 1.14660466e+00 9.94190276e-01 -4.69888031e-01
1.96002629e-02 2.73832362e-02 2.47239947e-01 5.94181836e-01
-4.13142115e-01 -2.07184732e-01 -5.11093676e-01 -4.00178820e-01
7.09306180e-01 -7.51079381e-01 -6.79983079e-01 3.89875919e-01
-1.24020171e+00 1.34256077e+00 4.59450275e-01 9.24339443e-02
-5.73911965e-01 -1.51568070e-01 5.53297579e-01 4.26420510e-01
5.29142201e-01 8.83741200e-01 -3.74324828e-01 -3.74336600e-01
-1.06391355e-01 5.44744134e-01 1.28311670e+00 9.58478749e-01
6.04977667e-01 8.43803436e-02 -9.12885308e-01 1.40858352e+00
1.63879469e-01 3.72885674e-01 8.26177120e-01 -1.20252001e+00
6.74420834e-01 7.42874265e-01 2.66243577e-01 -9.34936643e-01
-2.84890413e-01 2.06356466e-01 -8.59369814e-01 -8.73863876e-01
2.53644317e-01 -6.91790998e-01 -4.12143886e-01 1.74194932e+00
4.80326235e-01 -4.63055260e-02 8.25837791e-01 1.10772407e+00
1.62156856e+00 1.00343430e+00 1.32514045e-01 -1.16210751e-01
1.51230109e+00 -1.16139126e+00 -9.49757695e-01 -2.25568578e-01
5.41329741e-01 -5.07682979e-01 1.19038427e+00 -4.48437827e-03
-1.13833857e+00 -2.90495992e-01 -5.30793130e-01 -4.02942151e-01
-1.27019688e-01 -9.72530469e-02 6.33370399e-01 8.28237310e-02
-8.63666713e-01 1.06790245e-01 -1.36618271e-01 -2.65407652e-01
2.60653615e-01 -8.24864954e-02 5.99808916e-02 8.53630342e-03
-2.22923923e+00 1.14232028e+00 8.28220963e-01 2.60370135e-01
-8.96061242e-01 -3.58340502e-01 -1.02738905e+00 -1.18453979e-01
5.35716712e-01 -1.06075084e+00 1.91684937e+00 -8.66647363e-01
-2.30068588e+00 5.95496953e-01 -1.30805641e-01 -5.57130456e-01
3.00225019e-01 -2.17816368e-01 -1.76236689e-01 2.38984793e-01
-4.30313498e-02 7.87394464e-01 3.97744924e-01 -1.04802215e+00
-3.01494718e-01 -3.33751887e-02 5.68157792e-01 7.47778475e-01
-1.75134197e-01 1.49665936e-03 -1.25291482e-01 -4.18732733e-01
-2.95209050e-01 -5.85579574e-01 -3.29659700e-01 -8.60195994e-01
-8.52838635e-01 -8.67776513e-01 2.28257045e-01 -8.99715960e-01
1.17353475e+00 -1.36382413e+00 3.88761342e-01 -2.98306048e-01
2.75081191e-02 4.06849712e-01 -3.11894089e-01 7.87712395e-01
4.00679201e-01 -2.99490783e-02 -3.10485244e-01 1.95980892e-01
2.21517473e-01 1.96681648e-01 -7.06002057e-01 -4.86580640e-01
5.67539811e-01 1.29391062e+00 -1.19084680e+00 -4.53914464e-01
-9.97193754e-02 6.30894303e-02 -3.89854878e-01 8.22030842e-01
-7.69749284e-01 5.17591953e-01 -9.34625387e-01 2.78785557e-01
3.09571445e-01 -4.18037564e-01 1.75187930e-01 -6.82640299e-02
4.67552245e-01 9.06279325e-01 -6.08426452e-01 1.74274838e+00
-5.20783186e-01 2.30844736e-01 -1.35712117e-01 -7.89323568e-01
1.14204371e+00 5.90114415e-01 -2.74409533e-01 -7.75178611e-01
3.21448863e-01 -8.12678114e-02 9.18406472e-02 -7.46269882e-01
1.06141269e+00 -2.00439841e-01 -4.17368978e-01 8.18084896e-01
2.17173055e-01 -4.19211686e-01 1.15064487e-01 6.31013930e-01
8.65985155e-01 7.70962536e-02 4.39648688e-01 -8.80503282e-02
8.32590699e-01 3.19790065e-01 5.07298231e-01 6.08454406e-01
-2.19237134e-01 1.20759442e-01 5.34864128e-01 -8.05785134e-02
-5.77115297e-01 -8.17886055e-01 2.41588175e-01 1.34829867e+00
-5.79356328e-02 -2.08252162e-01 -9.67963099e-01 -6.69050455e-01
-2.25442156e-01 1.18841231e+00 -5.72510958e-01 -4.08069223e-01
-5.70497990e-01 -6.40005469e-01 9.41283464e-01 3.15087736e-01
8.63475919e-01 -1.51073086e+00 -1.80349857e-01 4.24505085e-01
-1.04505932e+00 -9.83004332e-01 -4.58157003e-01 -2.92076617e-01
-5.07895350e-01 -1.01151156e+00 -6.27863586e-01 -6.42811716e-01
4.39945310e-01 2.73857936e-02 1.30245328e+00 6.76513091e-02
2.17031837e-01 4.34417218e-01 -7.80625999e-01 -2.71014988e-01
-8.79904330e-01 2.82933414e-01 -3.14194560e-01 -1.55578896e-01
5.04746556e-01 -2.59369850e-01 -6.12503707e-01 2.34501064e-01
-5.22986352e-01 5.44107974e-01 1.97864652e-01 1.24402356e+00
2.54003629e-02 -6.99031472e-01 1.42700732e+00 -9.39587772e-01
1.79623079e+00 -9.28028882e-01 -1.61600243e-02 4.13181275e-01
-3.19039494e-01 6.69813678e-02 7.38071322e-01 -9.84392911e-02
-1.71175790e+00 -4.59540963e-01 -3.13174933e-01 1.51529402e-01
-2.10718647e-01 6.68679774e-01 -2.41870508e-01 3.58794570e-01
8.83126855e-01 5.75756133e-01 -5.66430669e-03 -2.83042997e-01
9.26490963e-01 1.00435328e+00 4.83193189e-01 -1.02479005e+00
3.88571382e-01 -1.11509696e-01 -8.67638111e-01 -6.25904858e-01
-1.15322506e+00 -1.71052799e-01 -2.96646029e-01 -2.47284338e-01
8.04138601e-01 -7.77843714e-01 -7.52580464e-01 4.05050665e-01
-1.56723464e+00 -6.07910931e-01 -1.50039956e-01 1.37024270e-02
-7.95430422e-01 4.40581888e-01 -9.29512799e-01 -8.13293934e-01
-1.01096618e+00 -6.71949387e-01 4.90762264e-01 6.30143344e-01
-3.80264521e-01 -1.11548972e+00 1.34785384e-01 9.78624940e-01
6.13864183e-01 -1.07485406e-01 8.38998258e-01 -1.19461322e+00
-3.99581462e-01 2.10559994e-01 -2.74903804e-01 3.85572881e-01
1.82407379e-01 -4.89146411e-01 -8.38080287e-01 3.11249673e-01
-4.64455523e-02 -1.27663302e+00 7.80965924e-01 -1.08834460e-01
8.73872638e-01 -9.30717528e-01 2.63414621e-01 9.27995518e-02
7.19972074e-01 1.38791844e-01 6.47610962e-01 3.51794176e-02
5.57859838e-01 9.54951048e-01 8.97937357e-01 5.86305082e-01
1.15693390e+00 5.70945501e-01 -8.99507478e-02 3.24330032e-01
8.47071316e-03 -5.01231790e-01 3.69494021e-01 9.61280346e-01
3.32148597e-02 -3.37010950e-01 -9.58124399e-01 7.25281477e-01
-2.06523061e+00 -1.10920680e+00 -6.68481588e-02 1.51533043e+00
1.86779332e+00 -3.89463246e-01 -7.78228343e-02 -5.83930612e-01
7.04726696e-01 2.56896228e-01 -6.05517149e-01 -7.16284156e-01
-1.39347151e-01 1.89399738e-02 -3.06499660e-01 8.93954813e-01
-7.35610485e-01 1.39245343e+00 5.44532824e+00 9.68790710e-01
-5.74844837e-01 1.81071773e-01 7.12907255e-01 2.76249945e-01
-5.69754958e-01 -8.29934627e-02 -7.73116291e-01 2.83249348e-01
1.04288197e+00 -8.35648358e-01 3.72256607e-01 6.68747306e-01
1.77668557e-01 -1.53860301e-02 -8.41523290e-01 5.17218471e-01
2.05443308e-01 -1.37613177e+00 5.97181797e-01 -4.65692014e-01
5.24003088e-01 -2.41356939e-01 -5.00364244e-01 8.15158308e-01
9.09646332e-01 -1.06450140e+00 9.52764452e-02 6.74288094e-01
2.94094533e-01 -7.19912887e-01 6.30437613e-01 5.05927682e-01
-5.74237227e-01 9.23310593e-02 -4.75348771e-01 -2.06537753e-01
3.78033757e-01 1.22591183e-01 -1.55250752e+00 7.35204160e-01
-3.25554945e-02 2.98981458e-01 -3.14762026e-01 3.45065385e-01
-7.46261239e-01 7.25476682e-01 1.53250396e-01 -4.99105006e-01
3.74332279e-01 -1.03581183e-01 4.22496468e-01 1.26988268e+00
-1.77703053e-01 7.59749889e-01 2.38233075e-01 1.21725249e+00
-3.34738374e-01 3.36864561e-01 -4.11299855e-01 -1.65992856e-01
7.69113958e-01 1.21084750e+00 6.84013665e-02 -4.98130888e-01
-1.13404028e-01 1.09718812e+00 6.52951419e-01 3.58660609e-01
-5.66389382e-01 -5.12708068e-01 3.95358354e-01 -5.10657549e-01
-6.09254390e-02 2.47347459e-01 4.42820638e-02 -1.27878141e+00
-2.28713885e-01 -1.06348455e+00 5.31719625e-01 -8.40650260e-01
-1.61298525e+00 6.92758322e-01 -1.85875390e-02 -5.94101429e-01
-9.08300877e-01 -2.97274381e-01 -7.83694685e-01 1.12888885e+00
-1.58340049e+00 -1.23721278e+00 -1.82227343e-01 8.37369919e-01
8.86800110e-01 -1.29109547e-01 1.23664069e+00 -2.36774221e-01
-4.92455840e-01 5.26765823e-01 -2.67867744e-01 5.76692164e-01
6.79724336e-01 -1.22088349e+00 5.92870951e-01 3.83755267e-01
-3.88972461e-01 6.87458634e-01 4.84551460e-01 -5.77918708e-01
-1.24091935e+00 -9.85658944e-01 9.54692185e-01 -5.47040045e-01
6.57529354e-01 -2.24430170e-02 -1.19046378e+00 5.91749191e-01
8.54483008e-01 -7.89079368e-01 1.15177667e+00 2.26369321e-01
-1.60484195e-01 4.83372599e-01 -9.86155093e-01 7.22187638e-01
7.74119377e-01 -4.52093899e-01 -1.22624564e+00 5.51851153e-01
1.07100284e+00 -6.33869708e-01 -1.01498663e+00 1.44014314e-01
8.82032588e-02 -5.41669309e-01 7.41976380e-01 -1.11715209e+00
8.67461562e-01 1.10361233e-01 -5.11950962e-02 -1.56561100e+00
-1.63133785e-01 -8.34329844e-01 -3.33804339e-01 1.40333378e+00
6.30026817e-01 -5.59373021e-01 2.78275132e-01 9.39017296e-01
-3.00281286e-01 -8.62366736e-01 -8.83911729e-01 -2.33428702e-01
3.72894078e-01 -3.41584273e-02 6.92072272e-01 1.26965284e+00
6.20875776e-01 1.16779447e+00 -5.38285911e-01 -3.20982277e-01
1.14381067e-01 2.91288882e-01 8.87185872e-01 -6.90738499e-01
-3.55082035e-01 -3.97404343e-01 3.18107367e-01 -1.51256204e+00
5.18143535e-01 -8.49767327e-01 2.68971980e-01 -1.82416987e+00
2.69776642e-01 -1.37137845e-01 1.05199426e-01 5.64391077e-01
-7.33078301e-01 -2.08145335e-01 6.67042360e-02 -9.62749682e-03
-8.36631596e-01 1.20355380e+00 1.61660683e+00 -7.90538639e-02
-2.33386919e-01 -1.30525276e-01 -1.16624665e+00 5.83667457e-01
1.14195645e+00 -5.05652502e-02 -6.28516674e-01 -5.14713168e-01
9.69907492e-02 6.32008851e-01 3.74733508e-01 -2.06077144e-01
2.21028000e-01 -5.60979486e-01 -5.48880780e-03 -2.40389958e-01
5.18325627e-01 1.62949860e-01 -5.40913522e-01 1.11918658e-01
-9.99920130e-01 -2.90423065e-01 1.27164751e-01 3.92412513e-01
-2.38390625e-01 -4.32019502e-01 4.69972908e-01 -4.44756210e-01
-8.51006806e-01 -4.15889844e-02 -4.51118529e-01 7.90311515e-01
6.28533065e-01 1.60519540e-01 -8.45210075e-01 -9.56834674e-01
-4.94553506e-01 8.74344766e-01 -4.15432416e-02 6.40257418e-01
9.64661360e-01 -1.42266572e+00 -1.26895380e+00 -4.43097770e-01
2.26632431e-01 9.58593562e-02 6.03412509e-01 4.49547172e-01
-2.15447322e-01 6.66843235e-01 -7.42882788e-02 9.56845209e-02
-8.91703069e-01 5.86385839e-04 5.99130988e-01 -4.21574712e-01
-3.57421815e-01 1.21469522e+00 9.73830745e-02 -9.27577496e-01
-7.26543814e-02 2.37582743e-01 -6.65654123e-01 -1.09677531e-01
7.81095147e-01 1.65836155e-01 -3.71698409e-01 -5.43842793e-01
-2.31895577e-02 -3.63211691e-01 -4.96256202e-01 -2.44464070e-01
8.37080240e-01 -3.44484776e-01 -3.30172718e-01 3.40276271e-01
6.45733953e-01 -4.27594543e-01 -8.08520734e-01 -6.71486080e-01
-5.54665215e-02 -2.40626141e-01 -2.57852703e-01 -1.39308739e+00
-5.01006484e-01 6.47229671e-01 -2.30828673e-01 3.00939530e-01
9.15032089e-01 6.94309399e-02 1.06893706e+00 9.82642174e-01
2.25492835e-01 -1.15552592e+00 3.16347271e-01 1.16543031e+00
1.46055794e+00 -1.22998226e+00 -3.34318221e-01 -2.72654474e-01
-1.31259906e+00 9.44635749e-01 1.25842166e+00 6.54545799e-02
4.53762636e-02 -3.98659229e-01 3.49281400e-01 -2.21773028e-01
-1.36776495e+00 -1.77577481e-01 1.14970960e-01 6.03809714e-01
6.93201125e-01 1.87654987e-01 -6.00311160e-01 1.12272990e+00
-5.53044856e-01 -2.15089899e-02 7.42427289e-01 5.64435601e-01
-6.61692500e-01 -1.08937812e+00 -6.04755171e-02 3.85233641e-01
-1.34217411e-01 -4.69781369e-01 -1.07815874e+00 2.08446071e-01
-6.23720765e-01 1.32761776e+00 -3.79613847e-01 -4.72007066e-01
3.60608757e-01 5.43001950e-01 2.96918482e-01 -9.23037887e-01
-8.48079681e-01 -5.41179240e-01 9.46293950e-01 -3.88258547e-01
-3.47600609e-01 -3.13057125e-01 -1.41286147e+00 -3.50388199e-01
-3.66536498e-01 7.17937350e-01 5.26429452e-02 9.34837818e-01
5.54652512e-01 4.69857216e-01 6.58939660e-01 -2.74368048e-01
-1.05265975e+00 -1.59683156e+00 -9.77854729e-02 3.76341432e-01
-4.20851186e-02 -3.23953331e-01 -5.06958254e-02 -1.66273722e-03]
|
[12.444563865661621, 8.143044471740723]
|
03ea203d-0cf5-4e17-b205-6ebf5f0d2471
|
hiddensinger-high-quality-singing-voice
|
2306.06814
| null |
https://arxiv.org/abs/2306.06814v1
|
https://arxiv.org/pdf/2306.06814v1.pdf
|
HiddenSinger: High-Quality Singing Voice Synthesis via Neural Audio Codec and Latent Diffusion Models
|
Recently, denoising diffusion models have demonstrated remarkable performance among generative models in various domains. However, in the speech domain, the application of diffusion models for synthesizing time-varying audio faces limitations in terms of complexity and controllability, as speech synthesis requires very high-dimensional samples with long-term acoustic features. To alleviate the challenges posed by model complexity in singing voice synthesis, we propose HiddenSinger, a high-quality singing voice synthesis system using a neural audio codec and latent diffusion models. To ensure high-fidelity audio, we introduce an audio autoencoder that can encode audio into an audio codec as a compressed representation and reconstruct the high-fidelity audio from the low-dimensional compressed latent vector. Subsequently, we use the latent diffusion models to sample a latent representation from a musical score. In addition, our proposed model is extended to an unsupervised singing voice learning framework, HiddenSinger-U, to train the model using an unlabeled singing voice dataset. Experimental results demonstrate that our model outperforms previous models in terms of audio quality. Furthermore, the HiddenSinger-U can synthesize high-quality singing voices of speakers trained solely on unlabeled data.
|
['Seong-Whan Lee', 'Sang-Hoon Lee', 'Ji-Sang Hwang']
|
2023-06-12
| null | null | null | null |
['speech-synthesis', 'singing-voice-synthesis']
|
['speech', 'speech']
|
[ 7.39067867e-02 1.01103351e-01 3.73342521e-02 1.66335046e-01
-1.15141976e+00 -4.36870605e-01 1.65903822e-01 -8.60845625e-01
4.30271626e-01 3.91438127e-01 6.09268129e-01 1.26206696e-01
-1.73163131e-01 -7.66640365e-01 -6.11031294e-01 -9.68248487e-01
8.78614709e-02 2.69953042e-01 -1.46459833e-01 3.32963541e-02
-2.97874391e-01 1.54040426e-01 -1.82804382e+00 2.30635926e-01
8.88652503e-01 7.64575660e-01 4.61113989e-01 1.18259132e+00
1.16887398e-01 8.76912892e-01 -7.67188787e-01 1.33000001e-01
1.43623158e-01 -1.15397763e+00 -8.74276087e-02 3.79910946e-01
1.78888053e-01 -5.56526184e-01 -6.78975582e-01 9.00334060e-01
7.46418774e-01 2.30713248e-01 8.31471562e-01 -1.00082719e+00
-1.04895079e+00 8.22101474e-01 2.78566957e-01 -3.01067263e-01
1.95130676e-01 2.39354119e-01 1.09325600e+00 -1.10947967e+00
5.84565222e-01 1.34191632e+00 5.20669818e-01 9.07247186e-01
-1.17891431e+00 -8.35363925e-01 -2.24913403e-01 -6.19952716e-02
-1.25875235e+00 -1.03047574e+00 1.22501230e+00 -4.95749593e-01
7.71243036e-01 2.28145376e-01 8.66956830e-01 1.18434274e+00
-5.39884903e-02 1.04606307e+00 8.57428789e-01 -3.37506860e-01
5.10803282e-01 -2.72474229e-01 -6.72505438e-01 5.33389986e-01
-6.30924761e-01 4.29508269e-01 -9.75320637e-01 -2.20883235e-01
9.73374188e-01 -2.36716673e-01 -3.84290755e-01 -1.35773852e-01
-1.16377044e+00 9.90732312e-01 6.78455606e-02 4.27234381e-01
-6.33126795e-01 2.28619024e-01 8.11054260e-02 5.05807102e-01
5.36184490e-01 2.71887392e-01 2.04584390e-01 -3.80793005e-01
-1.53367043e+00 1.67761996e-01 9.29388762e-01 8.11515987e-01
-1.97253432e-02 1.21162343e+00 -3.30820791e-02 1.13380373e+00
5.98839760e-01 7.88160563e-01 8.73646140e-01 -1.32094443e+00
3.52763027e-01 -1.79828465e-01 -1.11440361e-01 -7.59088993e-01
3.76213908e-01 -7.08860934e-01 -1.06197381e+00 2.14581378e-02
-1.70967430e-01 -2.53986061e-01 -7.61507928e-01 1.64325666e+00
2.32118651e-01 6.07432663e-01 4.55623150e-01 1.04083276e+00
6.36503160e-01 1.42060566e+00 -4.86070186e-01 -7.32307494e-01
5.79638839e-01 -1.30657423e+00 -1.29512012e+00 4.04213257e-02
4.59814211e-03 -9.04690742e-01 1.06878006e+00 6.75496399e-01
-1.43560147e+00 -7.93535352e-01 -1.20211351e+00 7.50542358e-02
3.59737128e-01 2.81197935e-01 2.86896639e-02 5.92190921e-01
-1.09057200e+00 7.10163295e-01 -8.41984868e-01 2.88468778e-01
-1.07471365e-02 2.35672086e-01 1.17916591e-01 1.81538969e-01
-1.18889713e+00 9.12272111e-02 -7.12481588e-02 8.23939778e-03
-1.74797332e+00 -6.27176046e-01 -7.38404870e-01 1.93049237e-01
9.60110128e-02 -6.96738362e-01 1.40070653e+00 -7.05375850e-01
-2.40102196e+00 -8.48050341e-02 -1.63979277e-01 -4.52768594e-01
2.03749195e-01 -9.53974053e-02 -6.74658358e-01 3.01763237e-01
-2.68376451e-02 3.50249052e-01 1.67849839e+00 -1.22138369e+00
-2.70694315e-01 7.91812092e-02 -5.87183535e-01 1.28949627e-01
-7.53654838e-01 -3.14503372e-01 -1.58463374e-01 -1.26221991e+00
3.43176454e-01 -8.76247525e-01 -1.09894358e-01 -2.39017427e-01
-2.34344080e-01 -1.13013410e-03 1.19806731e+00 -7.41562665e-01
1.45967042e+00 -2.39489055e+00 7.66980469e-01 -3.23517844e-02
9.87416953e-02 2.57796913e-01 -3.23452741e-01 5.58555543e-01
9.71339568e-02 -1.96032092e-01 -4.29801226e-01 -6.03072226e-01
1.00941718e-01 2.84709007e-01 -9.29354668e-01 7.19476566e-02
1.34646893e-01 6.24705315e-01 -9.05171037e-01 -3.33126873e-01
-1.60261076e-02 8.66711140e-01 -9.50376093e-01 5.93410432e-01
-3.33477139e-01 6.33136451e-01 -2.81295210e-01 6.53810859e-01
8.11178461e-02 7.52039552e-02 -7.55594075e-02 1.90798685e-01
-1.01821646e-01 3.16343933e-01 -1.34126556e+00 1.85285735e+00
-7.06890762e-01 4.39245343e-01 5.84378600e-01 -6.73604786e-01
1.26772738e+00 9.88783240e-01 4.36188310e-01 -1.93982556e-01
-7.28678331e-02 5.64134657e-01 -1.27452090e-01 -5.49402416e-01
4.83565241e-01 -6.63353026e-01 9.38243791e-02 7.02526629e-01
4.74074066e-01 -8.20984423e-01 -2.57263362e-01 -8.33145692e-04
8.03325117e-01 -5.25357537e-02 -2.59358674e-01 2.67514288e-01
3.70974571e-01 -4.56126690e-01 5.02370775e-01 3.81343246e-01
9.43835899e-02 8.45764041e-01 4.40204926e-02 3.46865237e-01
-1.16306949e+00 -1.08490145e+00 2.02972934e-01 8.75679076e-01
-6.16264284e-01 -5.15488863e-01 -1.00409400e+00 9.22892336e-03
-1.94308951e-01 6.86531007e-01 -1.06613144e-01 -2.65985161e-01
-5.47321379e-01 -1.66879162e-01 7.70637691e-01 3.27352643e-01
1.78691342e-01 -1.14406729e+00 1.16700575e-01 7.64613926e-01
-2.76485473e-01 -6.35326028e-01 -9.17077839e-01 -8.44065621e-02
-1.01806736e+00 -2.33283773e-01 -1.21806920e+00 -1.07802999e+00
3.75559106e-02 5.57244718e-02 5.00472724e-01 -4.29028034e-01
3.62743177e-02 5.38477778e-01 -1.72751129e-01 -2.24371389e-01
-1.07062137e+00 -5.30285798e-02 6.73724473e-01 4.19909060e-01
-1.97681293e-01 -1.02441621e+00 -4.20189857e-01 2.45263889e-01
-1.09821594e+00 -1.91274554e-01 3.84493738e-01 1.05945146e+00
8.61748815e-01 3.42305034e-01 1.22116649e+00 -2.77825296e-01
1.17019236e+00 -4.81177837e-01 -4.30598438e-01 -1.84559047e-01
-5.53697586e-01 -5.80544285e-02 7.98322380e-01 -9.99478400e-01
-1.00900042e+00 1.46332756e-01 -4.75884348e-01 -1.12693346e+00
2.52357155e-01 5.48293173e-01 -1.93895295e-01 3.97831798e-01
6.13853335e-01 7.74988770e-01 3.07205915e-01 -6.62951708e-01
6.72210813e-01 1.23975074e+00 6.81903601e-01 -4.71751869e-01
1.03731894e+00 1.81285337e-01 -4.18273956e-01 -1.24723160e+00
-6.27084553e-01 -2.03375161e-01 -4.38523114e-01 -3.53626728e-01
7.97702670e-01 -1.08760810e+00 -3.39798063e-01 5.94153702e-01
-9.58459318e-01 -3.60077500e-01 -8.51367593e-01 8.78131568e-01
-1.02162182e+00 2.94109732e-01 -1.04218841e+00 -1.24602854e+00
-3.67713392e-01 -1.20799422e+00 9.84842062e-01 -1.19385086e-01
-2.00366467e-01 -8.08637083e-01 3.87697488e-01 3.91564369e-01
5.30772448e-01 -1.72552168e-01 6.78942442e-01 -1.25373721e-01
-8.34946036e-01 -1.51322097e-01 8.13020527e-01 8.84877682e-01
2.96640933e-01 -1.93727389e-02 -1.09900475e+00 -4.07048285e-01
6.32493615e-01 -5.87248445e-01 7.60534286e-01 4.86870080e-01
8.61016035e-01 -5.86255908e-01 1.87921762e-01 5.98482847e-01
8.06568027e-01 3.02979439e-01 2.78174579e-01 -4.71894652e-01
6.07625067e-01 5.08849859e-01 3.97190958e-01 4.45114642e-01
-4.42624427e-02 4.08292145e-01 7.84771219e-02 2.08043069e-01
-6.42974019e-01 -9.36964035e-01 8.82794917e-01 2.47571397e+00
-1.43595142e-02 -9.45861563e-02 -3.40898067e-01 7.65484989e-01
-1.44638252e+00 -1.14037371e+00 2.52617985e-01 1.96542645e+00
9.90100622e-01 -1.90607086e-01 1.98318943e-01 7.37876236e-01
5.82746208e-01 3.97624880e-01 -7.71849036e-01 -1.51408061e-01
-5.35952374e-02 1.99100241e-01 -2.72920638e-01 5.78057528e-01
-6.14448845e-01 9.33398187e-01 6.73737049e+00 1.10824859e+00
-1.29855394e+00 2.46398866e-01 4.34463657e-02 -3.94305199e-01
-5.16570866e-01 -2.56493986e-01 -7.50953078e-01 2.72801191e-01
1.41819763e+00 -3.13617110e-01 9.95713651e-01 8.32137525e-01
5.22144794e-01 9.50924516e-01 -8.18468273e-01 9.43071663e-01
3.21049899e-01 -1.32499564e+00 1.64945886e-01 1.20351650e-01
1.01200593e+00 -3.68401915e-01 5.91939867e-01 3.70420963e-01
1.14488646e-01 -1.02623701e+00 9.71776307e-01 6.11374021e-01
9.92621243e-01 -6.80114627e-01 -8.10961872e-02 7.27827430e-01
-1.18668568e+00 -1.98703513e-01 -3.37133259e-01 2.54625250e-02
4.00349826e-01 4.72948402e-01 -8.41633081e-01 1.07757710e-01
3.48440677e-01 8.85519505e-01 3.22395504e-01 6.67619109e-01
-2.76326746e-01 1.32749903e+00 -1.32403627e-01 2.88715679e-02
1.44026890e-01 -3.37976813e-01 8.36839616e-01 8.33545446e-01
1.04135978e+00 1.39987186e-01 -1.12892807e-01 1.17853582e+00
-1.44502521e-01 2.10301220e-01 -7.41569042e-01 -7.42467284e-01
4.97849524e-01 8.55822742e-01 -1.53522268e-02 -3.49262983e-01
-5.65674417e-02 9.91864622e-01 -2.57222950e-01 5.39939642e-01
-4.82358992e-01 -2.71504253e-01 4.27629977e-01 1.19640477e-01
5.37118018e-01 -6.04826450e-01 1.85150370e-01 -1.19791400e+00
-1.98191032e-01 -1.18461514e+00 -2.17282906e-01 -9.33888018e-01
-1.17193913e+00 7.65000045e-01 -5.86479247e-01 -1.66760898e+00
-8.66476476e-01 -7.52414912e-02 -4.35701638e-01 8.93820822e-01
-1.31099319e+00 -1.04852200e+00 2.14363188e-01 7.35568643e-01
1.09447277e+00 -7.69097686e-01 1.09334612e+00 3.05693805e-01
-2.70495981e-01 3.46874893e-01 4.99309957e-01 -1.32214144e-01
5.45421302e-01 -1.05644202e+00 2.46435389e-01 5.22077441e-01
5.73090911e-01 5.33115804e-01 6.57897294e-01 -5.58231235e-01
-1.57024074e+00 -1.19546783e+00 8.43375564e-01 -6.62711263e-02
7.55729079e-01 -3.87858868e-01 -1.01114631e+00 3.72219115e-01
2.58015990e-01 -1.92096382e-01 9.46214974e-01 -3.55894119e-01
-4.46372777e-02 -2.47007702e-02 -7.97935009e-01 4.89450157e-01
9.20539916e-01 -1.10541153e+00 -7.54203081e-01 2.78242439e-01
1.06757987e+00 -3.12509865e-01 -1.31307495e+00 -8.64376798e-02
3.92950863e-01 -5.37642896e-01 9.43718851e-01 -3.06908876e-01
6.14100993e-01 -2.99409628e-01 -2.67225802e-01 -1.60935140e+00
-2.35864043e-01 -1.23460209e+00 -6.33910596e-01 1.46948147e+00
3.62796754e-01 -1.96597502e-01 6.18948519e-01 -1.15138724e-01
-2.91562527e-01 -3.53541464e-01 -1.01623571e+00 -9.23281014e-01
2.64090151e-01 -5.40790737e-01 7.16830075e-01 9.39871907e-01
-8.12684223e-02 5.63788593e-01 -9.50736523e-01 2.92397439e-01
6.78325236e-01 2.96203285e-01 6.36590064e-01 -9.38528001e-01
-7.86826372e-01 -5.19403890e-02 6.78306520e-02 -1.64232922e+00
3.44691098e-01 -9.27756429e-01 3.58849049e-01 -1.34886014e+00
-4.02700752e-01 -3.42454225e-01 -5.85376844e-02 -3.64651158e-02
2.55303383e-01 1.28354222e-01 2.07542866e-01 6.32798374e-01
1.29859084e-02 1.26680112e+00 1.74625456e+00 -3.42010230e-01
-5.69840908e-01 3.25655431e-01 -1.43939853e-01 6.92325890e-01
5.44653952e-01 -6.04912877e-01 -7.86656916e-01 -2.72286624e-01
-2.23690301e-01 9.17958379e-01 -3.19597237e-02 -1.22650170e+00
5.41534901e-01 -5.36415689e-02 -6.84058741e-02 -5.55097163e-01
9.34345126e-01 -6.85980916e-01 2.40989208e-01 3.82721454e-01
-6.45475864e-01 -4.54378992e-01 -2.46369943e-01 7.97759891e-01
-7.51926363e-01 -2.63905019e-01 5.73731899e-01 1.72094535e-02
4.88217324e-02 3.77819419e-01 -7.64464617e-01 -1.45492092e-01
6.44764543e-01 -1.17261307e-02 2.94807881e-01 -9.51231182e-01
-1.19610703e+00 -4.00365680e-01 3.00254822e-02 6.44490778e-01
1.04115403e+00 -1.78752804e+00 -7.69781232e-01 7.10091710e-01
-1.83620125e-01 -2.32458897e-02 3.76960874e-01 3.09276313e-01
-1.09735072e-01 4.28084821e-01 2.01690763e-01 -6.29813671e-01
-1.04207242e+00 5.27816832e-01 2.17230782e-01 2.11072639e-01
-6.20524883e-01 6.86065912e-01 2.50860602e-02 -5.18742502e-01
4.08738345e-01 -3.76098245e-01 -2.10845526e-02 6.37138635e-02
4.42052454e-01 2.72923321e-01 -2.23825336e-01 -7.92745650e-01
4.45915550e-01 5.62641621e-01 5.40326416e-01 -7.25169957e-01
1.48202682e+00 -7.59159997e-02 1.26390085e-01 8.11663926e-01
1.09379101e+00 5.11751115e-01 -1.55798721e+00 -2.13207752e-01
-5.19657254e-01 -3.21952015e-01 4.89546061e-01 -2.79401630e-01
-1.02952468e+00 1.17834890e+00 3.88220727e-01 2.66243249e-01
9.61183667e-01 -1.63355261e-01 1.19716048e+00 2.45950535e-01
7.10889623e-02 -1.14423096e+00 5.61882079e-01 3.63551050e-01
1.28057992e+00 -5.94089031e-01 -5.70844471e-01 -1.46883637e-01
-7.63518929e-01 1.07369995e+00 3.34232263e-02 -2.03190714e-01
8.67777586e-01 5.20483255e-01 3.20182562e-01 1.43925339e-01
-9.96597946e-01 1.48250520e-01 3.32640588e-01 6.92992151e-01
2.83912331e-01 1.46774545e-01 1.87023595e-01 7.12614954e-01
-7.15002477e-01 2.65889674e-01 3.94430131e-01 4.84400243e-01
-5.85406244e-01 -1.11072242e+00 -6.32480323e-01 6.15111217e-02
-2.48627439e-01 -8.89804587e-02 -3.41993868e-01 -5.02066165e-02
-2.11549595e-01 1.15659773e+00 -1.67487293e-01 -6.49174690e-01
2.57784605e-01 3.16056818e-01 2.04327404e-01 -5.76938808e-01
-3.00223589e-01 8.65251839e-01 -1.38249949e-01 -1.42522929e-02
-3.48430336e-01 -5.72883785e-01 -1.05202734e+00 -2.71673482e-02
-4.33966190e-01 4.89526629e-01 7.45228946e-01 5.80490887e-01
3.16967130e-01 9.22789633e-01 1.06361425e+00 -8.06885302e-01
-1.12804437e+00 -1.12152541e+00 -1.09244621e+00 -2.95119118e-02
7.18999386e-01 -2.52483189e-01 -6.90225780e-01 4.90695417e-01]
|
[15.500258445739746, 6.177810192108154]
|
4dd72201-08e4-4b90-9ee8-fcbe2a3e1138
|
neural-coreference-resolution-based-on
|
2212.09028
| null |
https://arxiv.org/abs/2212.09028v1
|
https://arxiv.org/pdf/2212.09028v1.pdf
|
Neural Coreference Resolution based on Reinforcement Learning
|
The target of a coreference resolution system is to cluster all mentions that refer to the same entity in a given context. All coreference resolution systems need to solve two subtasks; one task is to detect all of the potential mentions, and the other is to learn the linking of an antecedent for each possible mention. In this paper, we propose a reinforcement learning actor-critic-based neural coreference resolution system, which can achieve both mention detection and mention clustering by leveraging an actor-critic deep reinforcement learning technique and a joint training algorithm. We experiment on the BERT model to generate different input span representations. Our model with the BERT span representation achieves the state-of-the-art performance among the models on the CoNLL-2012 Shared Task English Test Set.
|
['Hongxia Jin', 'Yu Wang']
|
2022-12-18
| null | null | null | null |
['coreference-resolution']
|
['natural-language-processing']
|
[-1.93360418e-01 7.49524176e-01 -3.52070540e-01 -4.30604607e-01
-1.39456630e+00 -3.74381244e-01 3.49112749e-01 1.37795001e-01
-4.69526947e-01 8.89557660e-01 5.95257223e-01 -1.56903908e-01
-1.47473395e-01 -7.00321198e-01 -7.26537406e-01 -5.27427793e-01
-9.08300057e-02 1.31761420e+00 1.39246270e-01 -4.41692442e-01
-8.28784704e-02 4.03268814e-01 -1.03040540e+00 4.01257575e-01
8.98423612e-01 4.47752953e-01 9.86698717e-02 4.53505993e-01
-1.89806566e-01 1.11292589e+00 -6.81100190e-01 -7.87124515e-01
-1.30030647e-01 -2.53522754e-01 -1.47645581e+00 -6.62127554e-01
3.93027544e-01 2.69979704e-02 -6.16956115e-01 1.04421711e+00
5.88792562e-01 4.91197079e-01 4.86529678e-01 -1.18767822e+00
-6.18610799e-01 1.63183069e+00 -6.64625227e-01 4.19420570e-01
4.66509193e-01 -1.93963066e-01 1.53414178e+00 -3.36933315e-01
7.42681086e-01 1.59334815e+00 3.85409921e-01 8.23794782e-01
-1.15043938e+00 -8.60240877e-01 3.58264357e-01 7.87276745e-01
-1.28296053e+00 -4.40424860e-01 7.83623636e-01 6.59878040e-03
1.56367934e+00 -6.18727924e-03 3.01043689e-01 1.01394117e+00
-1.73212722e-01 9.28178430e-01 2.36609593e-01 -3.96261334e-01
4.88023227e-03 -5.60825765e-01 5.35827041e-01 3.62681091e-01
1.20671228e-01 2.63646036e-01 -1.40149280e-01 -1.41701624e-01
2.82876372e-01 -3.79892945e-01 -2.22958669e-01 -8.17135051e-02
-8.42817843e-01 1.07851887e+00 9.05221283e-01 4.91343617e-01
-5.43326259e-01 2.23320052e-01 5.74177921e-01 2.77167618e-01
2.22930107e-02 8.87798667e-01 -6.17487848e-01 2.16691822e-01
-7.03098536e-01 3.55704933e-01 9.70531046e-01 9.94330227e-01
4.96342778e-01 -2.56271690e-01 -3.26255977e-01 8.41388941e-01
4.28996414e-01 2.26350740e-01 1.54687405e-01 -1.24307323e+00
7.81512797e-01 7.24836826e-01 1.22754470e-01 -4.94962484e-01
-8.45099926e-01 -2.09577993e-01 -6.25962853e-01 -5.89999966e-02
4.08216506e-01 -5.07013440e-01 -5.56245029e-01 2.27126694e+00
4.52704549e-01 5.17616928e-01 5.46262503e-01 9.90345657e-01
1.28549409e+00 6.15296185e-01 5.61922073e-01 -4.38004255e-01
1.59633255e+00 -1.04240859e+00 -7.54896343e-01 -3.10434192e-01
5.58579266e-01 -4.33249742e-01 3.60241562e-01 -7.72854164e-02
-1.14009035e+00 -2.22072691e-01 -9.89548028e-01 -3.38882983e-01
1.57175101e-02 -8.35495442e-02 4.91992712e-01 -6.80717081e-02
-6.00897431e-01 5.01442254e-01 -7.78877914e-01 -2.94110805e-01
-3.10717663e-03 5.08971214e-01 -2.86177427e-01 1.52035996e-01
-1.87557018e+00 1.22651827e+00 8.91399622e-01 -1.33813828e-01
-7.70108759e-01 -6.39865160e-01 -1.08371305e+00 6.54670417e-01
3.99326980e-01 -6.51419342e-01 1.49348450e+00 -8.30307603e-01
-1.25976288e+00 9.72587883e-01 -8.04856494e-02 -8.18081260e-01
-1.58609357e-02 -2.78577745e-01 -7.73176551e-01 9.06450748e-02
1.92127585e-01 8.41934323e-01 2.02298269e-01 -1.25032747e+00
-8.99614334e-01 -1.73614338e-01 2.70295322e-01 2.75139183e-01
4.10925567e-01 2.68454373e-01 -3.69397163e-01 -2.25163370e-01
-2.40274385e-01 -9.02074039e-01 -1.04290612e-01 -8.26519370e-01
-6.14573121e-01 -1.05646122e+00 5.23487747e-01 -6.01807892e-01
1.10421622e+00 -1.93681800e+00 4.15588617e-01 -1.37183562e-01
-5.11637442e-02 6.13393545e-01 -4.29247260e-01 4.22861218e-01
-5.90539932e-01 -1.48073271e-01 1.24090560e-01 -2.94487983e-01
5.54467626e-02 1.42633781e-01 -4.72071886e-01 2.76143312e-01
3.52244884e-01 8.42101455e-01 -1.23451018e+00 -7.97335804e-01
-6.21471331e-02 2.70065755e-01 -6.07639194e-01 6.14754498e-01
-5.19174874e-01 1.95160106e-01 -3.46667141e-01 1.60430014e-01
5.63977420e-01 -9.69862193e-02 7.56959260e-01 -4.56865907e-01
5.81250265e-02 8.03945303e-01 -1.26388037e+00 1.85361707e+00
-2.25643039e-01 3.79748255e-01 8.12166706e-02 -1.12576675e+00
1.00885999e+00 7.30990469e-01 3.91700059e-01 -7.68217921e-01
2.56230682e-02 3.73484157e-02 2.53153354e-01 -4.24021989e-01
5.39181411e-01 -1.96618631e-01 -4.85938936e-01 5.36076725e-01
5.14800370e-01 4.46451753e-01 2.82890499e-01 5.43429434e-01
1.17002702e+00 1.53553367e-01 3.43820989e-01 -1.14952616e-01
6.34288430e-01 -3.60519402e-02 1.26478410e+00 4.89743680e-01
-2.63911664e-01 1.91422552e-01 7.72130191e-01 -4.10144567e-01
-7.89572597e-01 -9.35437918e-01 4.83591110e-02 1.25918603e+00
2.63754159e-01 -2.77420998e-01 -6.62873268e-01 -9.68410850e-01
-4.15366329e-02 1.21945977e+00 -6.37154400e-01 -2.70706862e-01
-1.05735648e+00 -4.33703601e-01 8.29808474e-01 7.18961179e-01
3.26450765e-01 -1.69532716e+00 -6.25649869e-01 4.37957138e-01
-7.35636532e-01 -9.70748305e-01 -6.13225043e-01 3.47047001e-01
-3.18480790e-01 -1.53762412e+00 -1.67106301e-01 -1.18674481e+00
1.39996946e-01 -1.70311868e-01 1.50059533e+00 2.63617635e-01
1.01879165e-01 -1.10734671e-01 -2.18997806e-01 -2.24352539e-01
-4.13925469e-01 5.28144956e-01 -3.14667746e-02 -3.65522385e-01
8.47841859e-01 -4.14452285e-01 -3.56885076e-01 1.55645683e-02
-2.63997823e-01 -2.23747104e-01 3.63453746e-01 9.89269376e-01
4.23839003e-01 -5.58212936e-01 9.28660691e-01 -1.16030550e+00
5.14315307e-01 -6.39874160e-01 -4.07599926e-01 7.02424109e-01
-4.40075845e-01 4.59543049e-01 5.75312555e-01 -3.25769454e-01
-1.44264317e+00 1.45244628e-01 -3.52954686e-01 -4.57592189e-01
-2.59881526e-01 5.87554812e-01 -5.47786236e-01 7.32850552e-01
6.92779541e-01 -4.16963249e-01 -6.19784296e-01 -4.64727968e-01
7.17395425e-01 4.23605800e-01 1.03171206e+00 -9.18227792e-01
3.34639549e-01 -2.00504869e-01 -3.25945705e-01 -1.45684719e-01
-1.24925053e+00 -6.10648274e-01 -8.72696817e-01 2.25819498e-01
9.22199130e-01 -1.00112355e+00 -1.27272820e+00 -1.29494995e-01
-1.68935072e+00 -3.62305760e-01 -2.23274708e-01 4.37908322e-01
-6.83558047e-01 2.45401546e-01 -9.39934254e-01 -3.55342239e-01
-9.32688057e-01 -1.00008166e+00 6.96668088e-01 5.22226453e-01
-4.78035539e-01 -6.79874778e-01 7.84250200e-01 2.02595934e-01
-1.22545749e-01 1.89414397e-01 1.16656017e+00 -1.29889250e+00
-2.22026870e-01 2.88963258e-01 -3.38995904e-01 -5.43931603e-01
-2.43507653e-01 2.23697871e-02 -7.88165331e-01 -4.44966674e-01
-5.49125850e-01 -6.18753493e-01 9.81366336e-01 4.40925658e-01
4.64045256e-01 -4.29600388e-01 -6.93567276e-01 7.09009826e-01
1.23547173e+00 2.67350674e-01 6.76882327e-01 5.72212815e-01
5.13884783e-01 5.27826965e-01 8.06867838e-01 3.00232917e-02
5.85803628e-01 8.44020784e-01 5.31786799e-01 1.65675446e-01
-2.02113941e-01 -3.56010795e-01 1.37536041e-02 5.17858565e-01
3.28279734e-02 -1.49487883e-01 -1.02591789e+00 8.30483913e-01
-2.39350438e+00 -1.45346904e+00 -4.03665602e-02 1.48152339e+00
1.15816414e+00 3.89474854e-02 8.42603445e-02 -1.96063653e-01
1.10146296e+00 4.83101793e-02 -7.21546292e-01 -5.13389230e-01
4.96739782e-02 1.53053589e-02 6.81411624e-02 8.07812393e-01
-1.25683367e+00 1.39808023e+00 5.80894184e+00 3.13389689e-01
-7.79621661e-01 7.64806494e-02 1.52689397e-01 6.81310594e-02
-1.22483328e-01 2.10689321e-01 -1.02725911e+00 5.00436798e-02
1.26401591e+00 -2.42842659e-01 3.76336962e-01 6.72164798e-01
-3.17943752e-01 2.26816058e-01 -1.29931819e+00 6.36365950e-01
-8.31074864e-02 -1.18834949e+00 8.18616524e-03 -4.67599303e-01
6.07405543e-01 2.48763189e-01 -5.83813608e-01 8.98740232e-01
1.15287662e+00 -9.42989111e-01 4.79049295e-01 4.44023997e-01
6.72071815e-01 -1.21693397e+00 8.70254159e-01 2.70106465e-01
-1.25188458e+00 -1.84649497e-01 -4.55989122e-01 2.97117174e-01
5.11342287e-01 1.03880614e-01 -8.67855728e-01 5.95420539e-01
4.98736024e-01 5.39186478e-01 4.60457988e-02 1.14987898e+00
-9.68429565e-01 4.87134576e-01 2.80468375e-03 1.28852949e-01
7.64969960e-02 2.39114791e-01 5.29433191e-01 1.56264651e+00
-2.09138602e-01 7.33149469e-01 1.71093464e-01 1.08714116e+00
-5.04602671e-01 -1.43688545e-01 -1.28911436e-01 4.75158364e-01
1.27290750e+00 1.48644257e+00 -2.55796492e-01 -2.38112032e-01
-3.70896846e-01 5.76207101e-01 1.29174638e+00 2.24401549e-01
-1.02051461e+00 -6.25642598e-01 7.00957894e-01 -5.66392899e-01
4.45551604e-01 2.95150816e-01 5.30113950e-02 -1.02820766e+00
-7.46388853e-01 -1.05169046e+00 1.22560930e+00 -6.55724168e-01
-1.25797784e+00 6.18218184e-01 2.19088960e-02 -6.36640429e-01
-4.96644616e-01 -2.10502967e-01 -1.15612674e+00 8.61248553e-01
-1.55826759e+00 -1.16666496e+00 2.34274380e-02 8.99417043e-01
3.62888187e-01 -2.31130362e-01 1.16757536e+00 2.68671781e-01
-8.89545679e-01 8.82320940e-01 -2.88000196e-01 8.58481288e-01
9.23820198e-01 -1.43884051e+00 5.42640567e-01 7.69624710e-01
1.79350600e-01 6.19873703e-01 7.63474703e-01 -6.24455273e-01
-9.96300340e-01 -1.12827682e+00 1.24489617e+00 -2.02753767e-01
4.16232586e-01 1.24697842e-01 -1.11346459e+00 1.29984272e+00
6.91745162e-01 -2.11852744e-01 4.53801423e-01 7.43380070e-01
-6.22565866e-01 -7.46977255e-02 -1.07539535e+00 2.82167256e-01
8.76644492e-01 -3.34281743e-01 -1.52334881e+00 5.09854704e-02
1.03708363e+00 -6.98822320e-01 -1.15444946e+00 3.15218300e-01
-1.72736384e-02 -4.72201288e-01 1.04570699e+00 -1.17554605e+00
3.54055524e-01 -7.36269280e-02 -5.38919233e-02 -1.64623213e+00
-9.07069743e-01 -5.29148579e-01 -6.01351798e-01 1.70658040e+00
6.75673544e-01 3.84143041e-03 3.83813173e-01 7.86116660e-01
-3.63890529e-01 -2.47518763e-01 -1.19876838e+00 -3.67059439e-01
3.90209615e-01 6.29429966e-02 9.16543901e-01 1.19957352e+00
8.02877426e-01 1.22609258e+00 -5.26814051e-02 4.12646562e-01
6.60748243e-01 5.42477012e-01 3.70714933e-01 -1.41721332e+00
-1.60976022e-01 -3.81901920e-01 2.47719541e-01 -7.74333298e-01
1.06660640e+00 -1.15143764e+00 1.70786411e-01 -1.55716622e+00
3.98370892e-01 -4.51542735e-01 -4.88496512e-01 9.34602499e-01
-3.92341942e-01 -5.69770455e-01 2.79912919e-01 1.42252892e-01
-1.03085387e+00 4.13941205e-01 7.11580634e-01 -5.49467504e-01
-2.08022520e-01 -1.18897751e-01 -9.02631640e-01 5.63464463e-01
5.41917264e-01 -6.49770319e-01 3.47009045e-03 -5.40659547e-01
2.68224448e-01 7.29965150e-01 -1.25331089e-01 -5.38821518e-01
6.98853791e-01 -2.68162221e-01 2.17528760e-01 -6.69174075e-01
1.37339994e-01 -5.91759086e-01 -1.16248555e-01 4.60968733e-01
-7.20349371e-01 1.38785303e-01 2.30605334e-01 3.54132801e-01
-1.55213773e-01 -3.57576340e-01 8.80233943e-01 -1.14217564e-01
-9.41622555e-01 9.68914032e-02 2.71030106e-02 6.33673966e-01
9.91004765e-01 8.38376462e-01 -8.37929130e-01 -1.34118170e-01
-9.51473951e-01 8.44467938e-01 -5.37024140e-02 6.36851847e-01
4.17600840e-01 -1.39419842e+00 -1.11826777e+00 -2.54613876e-01
-9.05647874e-02 7.84990862e-02 9.72159803e-02 4.37450588e-01
-4.28805351e-02 5.21687746e-01 -2.66336828e-01 -1.16426103e-01
-1.28719592e+00 1.11882377e+00 7.89155483e-01 -7.52400279e-01
-7.33556449e-01 8.69751930e-01 2.26725196e-03 -6.35419071e-01
5.50119936e-01 7.13006631e-02 -8.49639714e-01 1.97074458e-01
6.79251194e-01 2.60167003e-01 -9.74113196e-02 -7.59460151e-01
-6.37962818e-01 1.74651891e-01 -4.64737713e-01 1.68817401e-01
1.49024868e+00 1.02096729e-01 6.44903183e-02 4.57990915e-03
8.86553049e-01 -3.45376462e-01 -1.02077091e+00 -4.75450844e-01
4.05398369e-01 2.57423341e-01 -1.74749330e-01 -9.19329464e-01
-1.33679533e+00 7.16352522e-01 2.36968830e-01 -2.83428393e-02
7.52998829e-01 4.09691393e-01 7.84663320e-01 5.72079837e-01
2.52753973e-01 -9.79722798e-01 -1.63100049e-01 9.28397238e-01
9.17137802e-01 -1.04481387e+00 -1.90929174e-01 -5.65608256e-02
-8.51881981e-01 1.10284173e+00 1.11024737e+00 -3.66175592e-01
3.43194962e-01 3.97003353e-01 -2.43543740e-02 -3.03529412e-01
-1.22037995e+00 -3.48643392e-01 9.29282978e-02 5.83147347e-01
7.77812541e-01 1.82125568e-01 -4.41557258e-01 1.04592204e+00
-8.31055120e-02 -3.25854659e-01 3.84952158e-01 2.53061116e-01
-5.66363335e-01 -1.10527265e+00 -2.99536645e-01 4.86438200e-02
-6.14415407e-01 -5.73976971e-02 -4.26169515e-01 6.49452806e-01
-8.18643868e-02 1.15413845e+00 2.52938092e-01 -1.59200996e-01
7.35156476e-01 2.50136793e-01 4.54119921e-01 -6.42702162e-01
-8.31039250e-01 -2.33528525e-01 4.31711525e-01 -4.60335165e-01
-5.67217588e-01 -8.96867573e-01 -2.01936603e+00 -4.89718616e-01
-6.38064206e-01 7.09909678e-01 -3.42264250e-02 1.17731202e+00
2.18873590e-01 7.32664824e-01 4.24797356e-01 -4.82929766e-01
-5.44031382e-01 -1.21082735e+00 -4.49472934e-01 8.30329418e-01
1.81060702e-01 -6.94881976e-01 1.41109943e-01 -2.96238869e-01]
|
[9.299397468566895, 9.533286094665527]
|
d5d63d24-ddf6-46a9-97d2-ab533cd194f2
|
a-method-for-detection-of-small-moving
| null | null |
https://www.mdpi.com/2072-4292/13/4/653
|
https://www.mdpi.com/2072-4292/13/4/653/pdf
|
A Method for Detection of Small Moving Objects in UAV Videos
|
Detection of small moving objects is an important research area with applications including monitoring of flying insects, studying their foraging behavior, using insect pollinators to monitor flowering and pollination of crops, surveillance of honeybee colonies, and tracking movement of honeybees. However, due to the lack of distinctive shape and textural details on small objects, direct application of modern object detection methods based on convolutional neural networks (CNNs) shows considerably lower performance. In this paper we propose a method for the detection of small moving objects in videos recorded using unmanned aerial vehicles equipped with standard video cameras. The main steps of the proposed method are video stabilization, background estimation and subtraction, frame segmentation using a CNN, and thresholding the segmented frame. However, for training a CNN it is required that a large labeled dataset is available. Manual labelling of small moving objects in videos is very difficult and time consuming, and such labeled datasets do not exist at the moment. To circumvent this problem, we propose training a CNN using synthetic videos generated by adding small blob-like objects to video sequences with real-world backgrounds. The experimental results on detection of flying honeybees show that by using a combination of classical computer vision techniques and CNNs, as well as synthetic training sets, the proposed approach overcomes the problems associated with direct application of CNNs to the given problem and achieves an average F1-score of 0.86 in tests on real-world videos.
|
['and Zdenka Babić', 'Nikola Kezić', 'Janja Filipi', 'Vedran Jovanović', 'Mario Muštra', 'Vladimir Risojević', 'Vladan Stojnić']
|
2021-02-11
| null | null | null | null |
['video-stabilization', 'small-object-detection', 'segmentation-of-remote-sensing-imagery']
|
['computer-vision', 'computer-vision', 'miscellaneous']
|
[ 5.65771163e-01 -2.91842490e-01 2.20063299e-01 -8.15240294e-02
3.02218914e-01 -6.39391303e-01 4.00001585e-01 1.11935131e-01
-8.23079407e-01 6.86403871e-01 -7.21507013e-01 -2.57765979e-01
1.44230556e-02 -9.12128270e-01 -5.67717433e-01 -7.80286729e-01
-3.29667956e-01 1.39737949e-01 8.46009195e-01 -7.99359456e-02
9.54807699e-02 6.90406978e-01 -1.83006811e+00 8.68775174e-02
3.76834899e-01 1.21096551e+00 4.04650122e-01 7.76274264e-01
-5.19818403e-02 4.00241852e-01 -9.57646430e-01 -5.73283583e-02
4.74974960e-01 -4.31201398e-01 -5.52538276e-01 5.57923436e-01
3.12276155e-01 -4.57127064e-01 7.36566558e-02 1.22155809e+00
3.45543697e-02 2.59410709e-01 3.40472609e-01 -1.15040207e+00
-9.28357095e-02 2.63974965e-01 -7.47969151e-01 5.48619509e-01
3.39456648e-02 1.30985871e-01 2.82320857e-01 -1.91642314e-01
4.75756764e-01 9.83427584e-01 6.10048711e-01 6.35382056e-01
-1.08589494e+00 -6.34669006e-01 -2.78217662e-02 2.47303154e-02
-1.31310689e+00 -3.38687688e-01 5.38375080e-01 -2.65790641e-01
5.23739934e-01 1.88792333e-01 7.43495941e-01 8.39176595e-01
5.47613297e-03 5.71115792e-01 7.82736361e-01 -4.99608845e-01
4.53330964e-01 2.66011134e-02 1.14960253e-01 9.04484987e-01
5.93385279e-01 8.66365880e-02 3.14490683e-02 1.98940234e-03
7.91310608e-01 1.67812914e-01 -4.98959422e-01 -4.50082093e-01
-1.19244361e+00 6.75488472e-01 5.77433467e-01 6.39991581e-01
-2.60076612e-01 -1.12865761e-01 2.56907940e-01 5.58606051e-02
2.67272055e-01 1.49594083e-01 -5.75416148e-01 2.92505920e-01
-1.13394558e+00 3.38021331e-02 7.89529979e-01 8.32563877e-01
5.91935635e-01 1.43914297e-01 2.77037799e-01 5.06944180e-01
1.04162022e-01 6.07587695e-01 6.97292447e-01 -8.34623098e-01
6.83865324e-02 7.90651798e-01 3.51691812e-01 -1.20858443e+00
-5.76862812e-01 -7.33475462e-02 -9.00132000e-01 4.78891224e-01
9.30126905e-01 -3.20732296e-01 -1.00759661e+00 1.46910214e+00
5.01223624e-01 8.59614983e-02 1.81820512e-01 8.78976226e-01
7.12319374e-01 8.04355979e-01 -1.67005181e-01 -4.79081303e-01
1.38019347e+00 -7.71070302e-01 -6.77119613e-01 -3.38305891e-01
3.12681556e-01 -5.63457191e-01 6.97199166e-01 6.63792193e-01
-7.18657494e-01 -7.55805433e-01 -1.07115984e+00 4.72239554e-01
-7.28975594e-01 6.69704974e-01 5.70307732e-01 8.06455672e-01
-8.41691196e-01 4.84883428e-01 -1.10688531e+00 -7.76398540e-01
4.38632518e-01 6.58244669e-01 -3.85468513e-01 -2.28626691e-02
-5.75794578e-01 4.91223007e-01 7.67598867e-01 6.20885909e-01
-9.68003750e-01 9.20977667e-02 -8.47986817e-01 5.56823350e-02
5.71582198e-01 -2.60491610e-01 1.05176413e+00 -1.48993778e+00
-1.42971063e+00 6.45409942e-01 6.98062852e-02 -6.13658488e-01
3.62417996e-01 -2.64776573e-02 -3.28872353e-01 1.76501840e-01
-6.48623630e-02 7.87591398e-01 1.21334922e+00 -1.01389265e+00
-1.06621051e+00 -5.22867918e-01 2.83796549e-01 -2.85181254e-01
-3.37867767e-01 5.50735369e-02 -2.28658244e-01 -4.87440437e-01
-1.00387640e-01 -9.96423542e-01 -3.12155128e-01 1.51848733e-01
1.96890580e-03 2.14131385e-01 1.38367701e+00 -5.22342563e-01
7.42626369e-01 -2.11525607e+00 1.69011325e-01 -6.78524375e-03
-1.82005152e-01 8.50742280e-01 -1.21972978e-01 -1.48924530e-01
2.85421938e-01 -2.70516634e-01 -5.23226082e-01 1.60245001e-01
-5.76202273e-01 2.26112127e-01 6.25009611e-02 4.49444860e-01
3.66940051e-01 3.51019174e-01 -9.69519138e-01 -4.03778434e-01
4.36875552e-01 5.13280809e-01 -2.86167324e-01 1.04672045e-01
-3.34772825e-01 4.22409177e-01 -2.90204763e-01 9.12875652e-01
6.84492588e-01 9.26075056e-02 2.13794589e-01 -1.77192762e-01
-2.33415425e-01 -3.00809771e-01 -1.43651295e+00 1.24666262e+00
-2.66507119e-01 8.23869348e-01 6.30391359e-01 -1.16967463e+00
7.21630335e-01 7.04228207e-02 3.21715862e-01 -8.53635222e-02
6.02390170e-01 1.46947011e-01 1.57641172e-01 -5.01292765e-01
4.41389322e-01 1.12579256e-01 4.09714282e-01 2.07752869e-01
1.99884057e-01 8.05243105e-03 7.53155947e-01 -2.80896425e-01
1.15388930e+00 1.24540612e-01 3.50509286e-01 -1.39771536e-01
7.39946008e-01 2.07064673e-01 5.32642961e-01 4.25996542e-01
-3.29158396e-01 2.53020287e-01 7.54774660e-02 -7.12061286e-01
-5.59251726e-01 -4.15014923e-01 -1.15504391e-01 7.16929674e-01
3.22139680e-01 -1.03629045e-01 -1.04838562e+00 -5.66274524e-01
-3.65309775e-01 2.63219684e-01 -6.68940187e-01 1.25519447e-02
-5.08931041e-01 -1.34638500e+00 2.31520161e-01 2.81773478e-01
7.98150957e-01 -1.53059733e+00 -1.54898226e+00 2.63485670e-01
1.42386165e-02 -1.42832601e+00 2.11862668e-01 3.51993650e-01
-1.05648768e+00 -1.39708948e+00 -8.44633102e-01 -9.41609263e-01
9.18938994e-01 7.50245929e-01 7.27068901e-01 2.63226658e-01
-6.81227744e-01 4.95320521e-02 -5.64068377e-01 -4.42879945e-01
-1.84914261e-01 -5.92497997e-02 2.41349228e-02 1.19911991e-01
3.30209643e-01 -1.96079060e-01 -4.76822376e-01 5.19267321e-01
-1.32468140e+00 -2.73530930e-01 5.82599223e-01 7.28701651e-01
2.56861120e-01 7.92341411e-01 -5.98469116e-02 -4.59846318e-01
2.21040651e-01 7.50092715e-02 -1.27584326e+00 2.33194619e-01
1.58459961e-01 -3.80145282e-01 7.84528673e-01 -7.44552910e-01
-7.39235759e-01 5.84875464e-01 3.58022094e-01 -1.39097556e-01
-7.26662219e-01 1.48983523e-02 -1.71890289e-01 -3.99281025e-01
5.05059659e-01 5.41592687e-02 -1.45450896e-02 -1.18335485e-01
-7.07418472e-03 6.60021186e-01 5.05969346e-01 1.46564096e-01
8.00224602e-01 7.19762921e-01 1.09282002e-01 -1.43910742e+00
-5.60874760e-01 -4.77495432e-01 -7.70406425e-01 -1.71590582e-01
9.89030898e-01 -5.20976901e-01 -7.11329222e-01 7.27429569e-01
-1.11613059e+00 -3.62369388e-01 -7.08563402e-02 4.73441362e-01
-2.96534866e-01 4.81219947e-01 -3.24630469e-01 -8.49124551e-01
-2.40577906e-01 -1.30829680e+00 1.03840983e+00 5.90849340e-01
1.59753576e-01 -7.83028424e-01 -3.50589335e-01 2.22384453e-01
3.20402443e-01 3.76297474e-01 4.28096712e-01 -3.00897986e-01
-6.87678695e-01 -5.35814464e-01 -2.15009868e-01 3.80385965e-01
4.49237049e-01 3.80306333e-01 -8.25697243e-01 -3.52458745e-01
1.63558155e-01 -1.59347802e-01 8.11508238e-01 6.82093680e-01
9.45009410e-01 1.69549137e-02 -4.99175876e-01 4.88085866e-01
1.47704160e+00 7.20487595e-01 3.50169241e-01 2.75004804e-01
3.05468887e-01 7.99120724e-01 8.61432076e-01 3.18472564e-01
-3.55328947e-01 4.83446270e-01 7.39004433e-01 -1.18522078e-01
1.70815766e-01 5.22532761e-01 4.74285632e-01 9.65416506e-02
-2.21490085e-01 -5.04997313e-01 -5.57121754e-01 5.42780399e-01
-1.54318714e+00 -8.68738770e-01 -2.54442275e-01 2.34594107e+00
1.82054654e-01 1.92899048e-01 1.89545348e-01 4.28966612e-01
8.84120226e-01 -1.91860750e-01 -3.85725945e-01 1.29454494e-01
-3.62815149e-02 2.40978912e-01 7.08738565e-01 3.32304448e-01
-1.59463871e+00 1.05036867e+00 5.40759325e+00 5.33441782e-01
-1.20004439e+00 -1.33336395e-01 2.63882041e-01 7.13053346e-02
9.11030591e-01 -2.52389401e-01 -1.00465739e+00 4.59715754e-01
5.57088554e-01 5.40185571e-01 5.87230027e-01 6.73723280e-01
6.81955889e-02 -5.59666991e-01 -5.53650320e-01 8.38905931e-01
1.25524908e-01 -8.35549295e-01 -6.10402748e-02 -3.60670313e-02
5.24231493e-01 -3.79032671e-01 -2.70214349e-01 -6.82621077e-02
-5.26637537e-03 -5.39329946e-01 3.11720967e-01 3.52768525e-02
7.56813809e-02 -6.40356064e-01 1.09190106e+00 3.89594376e-01
-1.24820280e+00 -3.00548553e-01 -8.70533288e-01 -3.78151685e-02
-3.53788547e-02 2.96394169e-01 -7.40756929e-01 1.72256261e-01
8.88691187e-01 5.06038904e-01 -6.76460385e-01 1.28537118e+00
4.51273918e-02 5.24439156e-01 -5.82034171e-01 -2.11431563e-01
4.28137869e-01 -4.70855594e-01 4.45605993e-01 1.04329538e+00
4.53009278e-01 1.58062786e-01 2.91453525e-02 7.30790615e-01
1.43502444e-01 1.32662607e-02 -7.47687876e-01 -3.37059349e-01
-1.44085377e-01 1.62616634e+00 -1.76794815e+00 -3.91246021e-01
-3.78267765e-01 1.04428661e+00 -2.32388422e-01 1.20485917e-01
-6.99113369e-01 -6.83488905e-01 2.72836745e-01 -3.55569050e-02
8.14048052e-01 -3.86996120e-01 4.79641110e-01 -1.04465735e+00
-2.33772725e-01 -7.85032511e-01 1.74643785e-01 -6.56970263e-01
-6.69359863e-01 7.96215057e-01 7.07157478e-02 -1.23156190e+00
-6.91328943e-02 -1.17453587e+00 -5.36308587e-01 1.18430682e-01
-1.28745604e+00 -1.00929916e+00 -9.17613864e-01 5.96669734e-01
8.26121926e-01 -3.43442887e-01 7.27861047e-01 9.28043798e-02
-7.17049897e-01 -4.16046456e-02 7.07037896e-02 1.32963896e-01
3.57608408e-01 -9.47009087e-01 1.13108985e-01 1.07462370e+00
3.74565840e-01 2.65032202e-01 6.01719677e-01 -4.35730934e-01
-1.14814210e+00 -1.08795857e+00 1.74776077e-01 -1.30780935e-01
5.08158207e-01 -2.50881851e-01 -7.83371270e-01 3.43509763e-01
2.42012411e-01 7.10008591e-02 4.13205296e-01 -5.88496029e-01
5.91708899e-01 -1.96548223e-01 -1.17106926e+00 2.55755097e-01
6.08203709e-01 3.00446033e-01 -3.39855611e-01 4.95654434e-01
2.83566982e-01 -1.05965562e-01 -2.61666238e-01 5.80973923e-01
4.93047655e-01 -9.53429639e-01 9.36761260e-01 -4.00612652e-01
-3.16853784e-02 -7.38745987e-01 -1.22014917e-01 -1.11262345e+00
1.15951590e-01 -4.72441047e-01 2.23708689e-01 1.04682839e+00
1.51087791e-02 -1.43985569e-01 8.81313682e-01 3.24595347e-02
2.70892531e-01 2.39852928e-02 -4.94130552e-01 -7.05296099e-01
-6.60937309e-01 -4.76734489e-02 1.12089649e-01 6.72867298e-01
-5.93454599e-01 1.70785472e-01 -2.59037286e-01 4.53960717e-01
6.29851997e-01 2.25109041e-01 9.11995292e-01 -1.55681467e+00
-9.68372077e-02 -1.30739510e-01 -6.98339164e-01 -6.10667229e-01
6.93794340e-02 -2.54018545e-01 2.27340728e-01 -1.30523598e+00
-3.11594397e-01 -1.53790548e-01 3.12859416e-02 4.62909698e-01
7.77403861e-02 7.50190675e-01 4.55959998e-02 -2.72857338e-01
-4.54158604e-01 1.29167587e-01 8.94630373e-01 -3.89602810e-01
-4.65259105e-01 5.16271412e-01 4.63115387e-02 1.11398792e+00
8.71328413e-01 -3.61600846e-01 -3.08643997e-01 -3.35057795e-01
-3.90028238e-01 -1.46128356e-01 4.60968971e-01 -1.42084789e+00
6.62278831e-02 1.39209908e-02 6.62359059e-01 -5.06515324e-01
5.21924555e-01 -1.32672119e+00 1.17465250e-01 9.43735838e-01
-1.69386882e-02 4.09641787e-02 4.95921761e-01 6.31622136e-01
-1.89799681e-01 -7.58127034e-01 9.94724989e-01 -4.56603318e-01
-8.89474690e-01 -8.16275626e-02 -8.96645784e-01 -3.92110705e-01
1.34770072e+00 -3.91495168e-01 -5.48687652e-02 -1.03771277e-01
-6.55820012e-01 -1.66576266e-01 4.32743609e-01 3.19381177e-01
5.23531556e-01 -6.91937447e-01 -1.62128493e-01 6.17594361e-01
-1.47434562e-01 2.17099667e-01 -1.91750288e-01 6.41020238e-01
-1.06061351e+00 3.54611009e-01 -7.87988484e-01 -5.89619994e-01
-1.67557681e+00 7.71376133e-01 2.03022107e-01 1.25810713e-01
-2.27150828e-01 8.35079849e-01 1.41336530e-01 -5.98206930e-02
5.25686860e-01 -9.23839033e-01 -3.82961899e-01 1.30971253e-01
7.17267931e-01 3.17180544e-01 1.84882939e-01 -7.31266499e-01
-2.48822525e-01 6.86956048e-01 1.58253908e-01 2.63803959e-01
1.27975261e+00 1.36226147e-01 -2.65343875e-01 1.97227985e-01
6.64917231e-01 -2.97056735e-01 -1.02534342e+00 1.97854340e-01
4.06799130e-02 -5.04165292e-01 2.57146835e-01 -4.69234079e-01
-1.39169204e+00 1.03785777e+00 9.85293567e-01 6.58098876e-01
1.36325204e+00 -4.87568647e-01 7.08226681e-01 8.00593495e-01
6.36854053e-01 -7.67630577e-01 -1.08064242e-01 1.65721148e-01
3.84200782e-01 -1.37268806e+00 -1.36453196e-01 -3.64430487e-01
-2.91228712e-01 1.31005216e+00 7.82109499e-01 -1.26658618e-01
6.19060874e-01 3.28001767e-01 2.49374911e-01 5.71354032e-02
-4.60679829e-01 -5.22923291e-01 1.84889964e-03 7.02433467e-01
1.62797384e-02 -1.97559625e-01 -2.89588124e-01 2.87824739e-02
1.75013766e-01 1.91831097e-01 6.94678187e-01 1.20727181e+00
-7.01003432e-01 -9.26292002e-01 -6.88677549e-01 2.63754249e-01
-6.09119296e-01 2.57313967e-01 -6.74295843e-01 1.02401161e+00
5.19706726e-01 1.10495305e+00 2.14430660e-01 -1.79223895e-01
1.59072712e-01 -2.51500279e-01 6.23173952e-01 -3.97243649e-01
-4.96117026e-01 2.18980983e-01 -3.18971783e-01 -2.76974082e-01
-1.06211877e+00 -4.24676120e-01 -7.87832916e-01 1.26209632e-01
-7.75689363e-01 1.29889727e-01 8.73670936e-01 7.15117753e-01
-1.54579163e-01 3.62006396e-01 2.79480934e-01 -1.33288515e+00
-1.25295997e-01 -9.91379142e-01 -7.54450560e-01 1.55805677e-01
3.32544118e-01 -8.30978990e-01 -3.63892972e-01 5.34307003e-01]
|
[8.702872276306152, -0.8157316446304321]
|
e156f9de-68a9-4ea5-97e3-ae25dd249fa6
|
context-based-roman-urdu-to-urdu-script
|
2109.14197
| null |
https://arxiv.org/abs/2109.14197v1
|
https://arxiv.org/pdf/2109.14197v1.pdf
|
Context based Roman-Urdu to Urdu Script Transliteration System
|
Now a day computer is necessary for human being and it is very useful in many fields like search engine, text processing, short messaging services, voice chatting and text recognition. Since last many years there are many tools and techniques that have been developed to support the writing of language script. Most of the Asian languages like Arabic, Urdu, Persian, Chains and Korean are written in Roman alphabets. Roman alphabets are the most commonly used for transliteration of languages, which have non-Latin scripts. For writing Urdu characters as an input, there are many layouts which are already exist. Mostly Urdu speaker prefer to use Roman-Urdu for different applications, because mostly user is not familiar with Urdu language keyboard. The objective of this work is to improve the context base transliteration of Roman-Urdu to Urdu script. In this paper, we propose an algorithm which effectively solve the transliteration issues. The algorithm work like, convert the encoding roman words into the words in the standard Urdu script and match it with the lexicon. If match found, then display the word in the text editor. The highest frequency words are displayed if more than one match found in the lexicon. Display the first encoded and converted instance and set it to the default if there is not a single instance of the match is found and then adjust the given ambiguous word to their desire location according to their context. The outcome of this algorithm proved the efficiency and significance as compare to other models and algorithms which work for transliteration of Raman-Urdu to Urdu on context.
|
['Muhammad Waheed', 'Rashid Khan', 'H Muhammad Shakeel']
|
2021-09-29
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[ 7.43068457e-02 -5.42905509e-01 -9.92730334e-02 -2.88815707e-01
9.27238837e-02 -1.11585367e+00 4.00207192e-01 -2.87665427e-02
-3.32176536e-01 8.63438368e-01 2.13327840e-01 -8.79514515e-01
7.72877187e-02 -8.23554754e-01 -1.61606036e-02 -1.93314552e-01
6.56568468e-01 7.25506723e-01 3.96425426e-01 -6.25355482e-01
5.70171177e-01 6.81982040e-01 -1.32166743e+00 1.50326967e-01
8.70743811e-01 2.43827105e-01 6.39367878e-01 7.25823283e-01
-7.34770715e-01 4.10270870e-01 -8.33289206e-01 -1.49483562e-01
2.42409125e-01 -8.00977647e-01 -1.00712466e+00 -1.81978658e-01
-3.71774286e-01 -4.35425609e-01 2.93072648e-02 1.04727399e+00
4.76498157e-01 -4.18075174e-03 9.07111108e-01 -6.32324159e-01
-7.88442552e-01 7.34274447e-01 -4.40455526e-01 3.84200454e-01
8.16519797e-01 -6.25854194e-01 4.34608549e-01 -7.73376644e-01
6.65458918e-01 1.22507298e+00 2.47564465e-01 3.54111016e-01
-6.10198259e-01 -6.16700470e-01 -4.13064957e-01 -2.17802338e-02
-1.56634498e+00 8.80441144e-02 1.92166969e-01 -3.03568840e-01
9.92161989e-01 6.36301756e-01 2.13745564e-01 5.11473894e-01
2.38954604e-01 4.25965562e-02 1.27704215e+00 -1.03332174e+00
-3.56469989e-01 6.14325762e-01 4.85482246e-01 4.10317838e-01
2.13040650e-01 -5.33648491e-01 -2.26128865e-02 1.41767725e-01
8.96991611e-01 -4.81930152e-02 -4.20706123e-02 6.89292967e-01
-9.76293445e-01 6.43411398e-01 -3.90290737e-01 9.55129802e-01
-5.21989912e-02 -4.16652620e-01 4.15996104e-01 4.62699831e-01
-1.54721841e-01 1.55529022e-01 -4.56314176e-01 -6.93935394e-01
-6.49998844e-01 8.04203376e-02 9.11246359e-01 1.25553823e+00
6.74597144e-01 -5.57161309e-02 3.16245854e-02 1.11732268e+00
3.91120970e-01 6.00957811e-01 8.25850964e-01 -1.82769164e-01
3.27731818e-01 8.12211812e-01 1.10940270e-01 -6.18469119e-01
-1.30578890e-01 3.62053633e-01 -3.58038843e-01 3.57551090e-02
4.89436656e-01 -3.32575023e-01 -9.64447677e-01 1.03718340e+00
3.72722626e-01 -5.24124861e-01 1.71720639e-01 6.56233788e-01
9.16303158e-01 1.21136725e+00 -1.68872371e-01 -3.45594287e-01
1.73598981e+00 -4.90957171e-01 -1.02380490e+00 1.76775351e-01
4.40293312e-01 -1.77420843e+00 1.29538751e+00 3.97872299e-01
-6.20289326e-01 -5.62126040e-01 -1.13876116e+00 -1.74679354e-01
-7.65030563e-01 4.74199474e-01 3.15805376e-01 1.19077718e+00
-7.38870203e-01 3.91873151e-01 -2.67484605e-01 -9.47707653e-01
-5.68259537e-01 2.40531161e-01 -1.09120779e-01 1.03681058e-01
-1.29498994e+00 1.09877515e+00 8.26419234e-01 -9.83486176e-02
-4.42868657e-02 2.11817399e-01 -5.01518607e-01 -4.22636658e-01
4.97799329e-02 1.08604096e-01 1.14239407e+00 -1.20237195e+00
-1.71451962e+00 9.70505059e-01 -3.30897868e-01 2.43353188e-01
5.04919291e-01 1.98306844e-01 -8.82798672e-01 -2.83060759e-01
-4.76877321e-04 2.21865997e-02 5.06902456e-01 -7.82131433e-01
-8.46260786e-01 -2.70889312e-01 -1.98751971e-01 2.82604009e-01
-1.33505370e-02 9.96072054e-01 -7.44422138e-01 -6.34014368e-01
1.58556357e-01 -7.82526314e-01 4.17942971e-01 -8.07828903e-01
-8.22345093e-02 -5.29253960e-01 1.07551777e+00 -1.18588257e+00
1.99493277e+00 -1.89659166e+00 -2.09825739e-01 5.79522491e-01
-5.80621839e-01 4.14895773e-01 4.72416997e-01 7.94627011e-01
-6.77335188e-02 3.21158618e-01 9.03256685e-02 4.06902820e-01
-5.08577265e-02 5.82954764e-01 -1.79374456e-01 -2.34267600e-02
-1.92304522e-01 3.86838913e-01 -5.42561769e-01 -6.54854774e-01
1.17659261e-02 3.23391080e-01 -1.01816701e-02 1.90428272e-01
-6.67105988e-02 2.40800828e-01 -4.02726680e-01 8.19858909e-01
6.69519126e-01 4.29654658e-01 3.66371185e-01 -6.96998462e-02
-7.88320720e-01 2.37738237e-01 -1.57899141e+00 9.74658847e-01
-5.44631362e-01 6.19068861e-01 -4.27206963e-01 -5.26021421e-01
1.18643987e+00 5.43392122e-01 -1.55841306e-01 -4.76531029e-01
5.26559293e-01 8.68949652e-01 1.44635618e-01 -8.65700185e-01
8.22521329e-01 1.67791396e-01 1.33777065e-02 7.19830394e-01
-1.51856780e-01 -2.72972018e-01 8.42721522e-01 -3.01452782e-02
3.76506984e-01 5.87760746e-01 7.65173614e-01 -2.12614894e-01
8.42317104e-01 -4.46936600e-02 1.61380351e-01 5.06933928e-01
3.26883823e-01 6.05455220e-01 4.86835659e-01 -6.33488670e-02
-1.26654172e+00 -7.32558072e-01 -5.41513622e-01 1.17644012e+00
-2.56902009e-01 -3.12379062e-01 -9.02328193e-01 -3.76768589e-01
-4.77399260e-01 8.66843343e-01 -1.18985511e-01 5.00456035e-01
-8.07325065e-01 -4.63726163e-01 6.36424482e-01 7.14278221e-02
6.43096089e-01 -1.25336611e+00 -5.47074854e-01 3.29386979e-01
-2.59819683e-02 -8.46399903e-01 -5.71443558e-01 7.16010928e-02
-7.41865277e-01 -6.56000197e-01 -8.74672294e-01 -1.17264986e+00
7.43893921e-01 -7.39628030e-03 4.28309947e-01 2.32680231e-01
-3.10119539e-01 4.29796688e-02 -8.40591788e-01 -4.76027578e-01
-8.55113924e-01 2.23916262e-01 -6.42589256e-02 -3.92153561e-01
6.44794941e-01 -2.72482216e-01 3.33601683e-02 2.80184567e-01
-1.12511885e+00 -2.31401157e-03 3.09899986e-01 2.98796207e-01
1.57198161e-01 -2.40891978e-01 4.63751644e-01 -9.28566813e-01
9.43994761e-01 -5.13484597e-01 -5.77644825e-01 4.61189508e-01
-5.91668725e-01 1.56852871e-01 8.20426583e-01 -3.98536921e-01
-8.98159802e-01 -1.56275094e-01 -4.11304295e-01 4.12830800e-01
-1.89172775e-01 5.63798130e-01 -2.94859886e-01 1.29721850e-01
6.51272297e-01 5.22517920e-01 -1.27256781e-01 -6.56134009e-01
-3.52083123e-03 1.60345685e+00 2.21655995e-01 -6.28906369e-01
5.23953199e-01 -2.92281240e-01 -3.63884062e-01 -1.01898587e+00
9.44072679e-02 -4.22158659e-01 -5.08211792e-01 -2.27390036e-01
1.05938017e+00 -1.77827060e-01 -4.30044323e-01 5.07443368e-01
-1.57010329e+00 1.71142057e-01 3.10718209e-01 4.29935426e-01
-4.36146781e-02 6.54720783e-01 -4.23585743e-01 -9.98179734e-01
-4.44090515e-01 -1.16546798e+00 4.21340227e-01 7.92560279e-01
-4.27134544e-01 -9.05673683e-01 -2.19251569e-02 -2.06731528e-01
3.12536687e-01 -1.85670763e-01 1.35085225e+00 -7.34147251e-01
-1.92943700e-02 -1.15466505e-01 -2.26157740e-01 3.54456574e-01
6.42410100e-01 4.31528926e-01 -3.83371919e-01 2.20585361e-01
-3.36778820e-01 4.31903861e-02 1.67657956e-02 -4.32776600e-01
6.73460126e-01 -2.70783752e-01 9.44759510e-03 3.23085308e-01
1.47680342e+00 1.08347964e+00 8.11180174e-01 4.76579934e-01
4.43820536e-01 3.30938637e-01 6.71099067e-01 3.42031479e-01
1.44937247e-01 5.68653524e-01 -2.04171956e-01 2.60029286e-01
-3.61822322e-02 -2.76807427e-01 6.12399757e-01 1.04513323e+00
-3.06627125e-01 -2.06448317e-01 -9.96796787e-01 1.96695775e-01
-1.50256944e+00 -6.86837494e-01 -5.10188878e-01 2.49202204e+00
1.25744164e+00 -4.72448161e-03 8.26466456e-02 2.37935185e-01
7.31005132e-01 -3.14093679e-01 1.90361574e-01 -1.34364176e+00
-2.04102963e-01 6.38172686e-01 5.52461207e-01 8.62733245e-01
-6.13353252e-01 1.31316078e+00 5.61786556e+00 8.17496419e-01
-1.41052341e+00 1.47911627e-02 3.17469031e-01 5.73206663e-01
-4.20780629e-01 1.59768209e-01 -1.25870287e+00 5.63451946e-01
6.47103548e-01 -3.89245510e-01 7.16741204e-01 3.91643703e-01
5.18867970e-01 -5.39624333e-01 -9.66077745e-01 1.12693250e+00
2.38259405e-01 -8.74526381e-01 4.13353473e-01 -2.53446579e-01
6.27034843e-01 -4.81558204e-01 -3.98679078e-01 1.54369190e-01
-3.77553925e-02 -1.05303633e+00 6.82690501e-01 2.11463213e-01
8.70886266e-01 -8.29797804e-01 7.64354050e-01 2.46030793e-01
-1.01334417e+00 3.48885000e-01 -5.05289376e-01 -1.71681091e-01
6.57000169e-02 -6.05962574e-02 -1.12841952e+00 5.53026378e-01
4.88849461e-01 4.71292809e-02 -4.84394908e-01 7.92902231e-01
-1.35206014e-01 6.26962662e-01 -7.01760590e-01 -7.02702880e-01
3.27091902e-01 -1.00201571e+00 4.58536655e-01 1.64508402e+00
9.85496581e-01 1.98108062e-01 -1.62983134e-01 4.54812050e-01
3.62264067e-01 1.20842648e+00 -5.72877288e-01 -4.03797120e-01
5.43310523e-01 1.00254261e+00 -9.91305828e-01 -3.91915500e-01
-3.19189876e-01 1.45439899e+00 -3.15248847e-01 3.40755999e-01
-7.14198112e-01 -1.08312118e+00 2.16413036e-01 1.92138478e-01
-4.61597741e-02 -5.09730458e-01 -3.07561457e-01 -6.62809908e-01
-1.15471795e-01 -1.32581770e+00 3.47115308e-01 -6.98898971e-01
-5.31690896e-01 7.53443003e-01 2.05566451e-01 -1.05438876e+00
-3.87754798e-01 -9.45194006e-01 -5.44586003e-01 1.36880898e+00
-9.43872750e-01 -9.83225405e-01 -3.25931720e-02 6.10089004e-01
8.08959603e-01 -3.68000329e-01 1.03739989e+00 6.22513592e-01
-5.16372979e-01 4.80168402e-01 2.29565471e-01 3.15905690e-01
8.94720733e-01 -1.05845428e+00 -2.37226069e-01 1.00096905e+00
1.29549906e-01 1.11292696e+00 8.33953261e-01 -9.58188355e-01
-1.09251797e+00 -1.76158398e-01 1.49146879e+00 -1.82210952e-02
6.18172586e-01 -2.24169433e-01 -6.27990484e-01 6.72999859e-01
6.51714563e-01 -7.68030643e-01 6.87231541e-01 -4.31730092e-01
1.46109641e-01 -5.12567610e-02 -1.08278406e+00 1.09103322e+00
4.89150822e-01 -2.83478439e-01 -6.99276686e-01 3.80377173e-01
3.91544372e-01 -5.56075752e-01 -4.69254494e-01 -2.81374782e-01
6.73607290e-01 -6.40838921e-01 3.23177963e-01 -5.51464319e-01
1.09246582e-01 -8.33230317e-01 -2.08157793e-01 -1.04451239e+00
-4.08526547e-02 -1.00553083e+00 6.24049425e-01 1.70526981e+00
7.85388052e-01 -6.50556386e-01 6.54189438e-02 2.81709313e-01
5.83411939e-02 -1.22686945e-01 -7.62706816e-01 -3.57923001e-01
-1.01569928e-01 -1.55855954e-01 5.79927027e-01 8.99751246e-01
1.17200471e-01 3.69664788e-01 -3.67894888e-01 -1.32032767e-01
-1.06927209e-01 -2.21839175e-01 3.88074934e-01 -8.73641849e-01
-3.47098440e-01 -2.67026484e-01 -2.51712173e-01 -1.08547294e+00
-3.42007965e-01 -8.52936625e-01 -3.03695887e-01 -1.65991402e+00
-2.27749318e-01 -4.88286942e-01 3.46864522e-01 4.06029552e-01
-1.05930373e-01 2.58479834e-01 2.64440417e-01 3.85247357e-02
2.08994806e-01 -3.85995537e-01 9.48271394e-01 3.44093829e-01
-5.58963478e-01 1.79773018e-01 -3.25573564e-01 4.64717805e-01
8.51279497e-01 -3.66411954e-01 -2.97075391e-01 -4.07337934e-01
5.86850882e-01 -6.39686435e-02 -5.68102419e-01 -7.51962543e-01
6.56465441e-02 -4.71240699e-01 3.68976235e-01 -5.82138121e-01
-3.38190079e-01 -1.08600879e+00 2.47446775e-01 3.64848942e-01
2.78056562e-01 8.19676757e-01 6.16467409e-02 -4.28757042e-01
9.87531319e-02 -1.17554188e+00 7.74261773e-01 -1.78258315e-01
-8.37340355e-01 -2.44059429e-01 -1.05764437e+00 -1.67313784e-01
1.19632292e+00 -7.96554089e-01 1.21564522e-01 -4.10487443e-01
-3.47968966e-01 -1.40519679e-01 2.94187248e-01 5.41753292e-01
3.11220437e-01 -1.06409037e+00 -5.02751470e-01 2.20034346e-01
-1.45611241e-01 -5.68071723e-01 -3.35926235e-01 4.80175853e-01
-1.56196904e+00 5.14741421e-01 -5.78516901e-01 4.10770290e-02
-1.61599159e+00 1.64642885e-01 3.40063065e-01 -1.36067271e-01
-1.00842021e-01 3.53397787e-01 -5.98612607e-01 -4.33518648e-01
9.91009921e-02 -4.16210681e-01 -8.09937119e-01 2.00135291e-01
5.92945933e-01 5.23108244e-01 1.42632231e-01 -7.77479589e-01
-2.96955854e-01 8.81242812e-01 -7.27095902e-02 -6.85379028e-01
7.05060959e-01 -2.66994685e-01 -7.27808714e-01 7.35116303e-01
9.25973952e-01 7.81866014e-01 -3.16489190e-01 1.81769148e-01
1.31664604e-01 -3.49384576e-01 -5.53712428e-01 -8.97166848e-01
-2.97023952e-01 6.02324843e-01 7.31513143e-01 1.65819392e-01
8.06785345e-01 -4.22600269e-01 5.49096346e-01 5.03600478e-01
4.26674992e-01 -1.57066262e+00 -4.60598767e-01 1.31248105e+00
6.30384922e-01 -8.70200992e-01 -1.66762903e-01 -2.57817864e-01
-6.91285670e-01 1.81467760e+00 5.00667036e-01 7.22352043e-02
6.84609711e-01 4.67559248e-01 4.76183623e-01 3.63105685e-01
3.02918032e-02 -4.91676964e-02 4.89981733e-02 7.06878304e-01
1.11252689e+00 -4.19451967e-02 -1.61668670e+00 6.13973558e-01
-6.14896894e-01 1.60435557e-01 7.87014127e-01 1.00726771e+00
-7.25664198e-01 -1.73871338e+00 -9.21955585e-01 2.70971149e-01
-8.71244371e-01 -3.64395738e-01 -4.76555496e-01 7.05262542e-01
4.38412309e-01 1.01238978e+00 1.43117070e-01 -5.67078851e-02
2.62318850e-01 3.81803006e-01 5.68250060e-01 -6.22115433e-01
-9.92442071e-01 6.89972267e-02 2.38168046e-01 3.03613901e-01
1.66987449e-01 -5.91374159e-01 -1.75967920e+00 -4.76096690e-01
1.49353147e-01 3.59424084e-01 8.04035425e-01 9.67133164e-01
-3.69474381e-01 -6.05907924e-02 5.15555978e-01 -1.37106493e-01
-4.71458375e-01 -1.11115694e+00 -4.19696599e-01 8.81873518e-02
-2.58985728e-01 -1.15604781e-01 2.10492942e-03 9.99565572e-02]
|
[10.65872859954834, 10.498636245727539]
|
938df728-440a-45f6-8dc4-67a9ea1877b2
|
the-effect-of-the-loss-on-generalization
|
2108.04815
| null |
https://arxiv.org/abs/2108.04815v1
|
https://arxiv.org/pdf/2108.04815v1.pdf
|
The Effect of the Loss on Generalization: Empirical Study on Synthetic Lung Nodule Data
|
Convolutional Neural Networks (CNNs) are widely used for image classification in a variety of fields, including medical imaging. While most studies deploy cross-entropy as the loss function in such tasks, a growing number of approaches have turned to a family of contrastive learning-based losses. Even though performance metrics such as accuracy, sensitivity and specificity are regularly used for the evaluation of CNN classifiers, the features that these classifiers actually learn are rarely identified and their effect on the classification performance on out-of-distribution test samples is insufficiently explored. In this paper, motivated by the real-world task of lung nodule classification, we investigate the features that a CNN learns when trained and tested on different distributions of a synthetic dataset with controlled modes of variation. We show that different loss functions lead to different features being learned and consequently affect the generalization ability of the classifier on unseen data. This study provides some important insights into the design of deep learning solutions for medical imaging tasks.
|
['Julia A. Schnabel', 'Ben Glocker', 'Sujal Desai', 'Arjun Nair', 'Kyriaki-Margarita Bintsi', 'Octavio E. Martinez Manzanera', 'Sam Ellis', 'Loic Le Folgoc', 'Vasileios Baltatzis']
|
2021-08-10
| null | null | null | null |
['lung-nodule-classification']
|
['medical']
|
[ 2.78046906e-01 -1.83760509e-04 -4.26122844e-01 -6.37946188e-01
-6.51334226e-01 -3.72970730e-01 3.28617245e-01 3.34335059e-01
-6.83134675e-01 6.98729992e-01 -2.25079626e-01 -3.58368635e-01
-2.57407188e-01 -6.54447556e-01 -5.52129209e-01 -9.34699297e-01
-5.97720817e-02 2.52849102e-01 2.49611527e-01 2.18538821e-01
-9.22294185e-02 6.59726799e-01 -1.39876592e+00 2.36294851e-01
7.06453383e-01 1.31086326e+00 -8.47544894e-02 5.21634698e-01
2.17481166e-01 7.42439270e-01 -5.91083825e-01 -3.62854630e-01
1.91563472e-01 -5.98931551e-01 -8.51071239e-01 -1.88773666e-02
9.35723186e-02 -7.60117127e-03 -3.49311203e-01 1.04512775e+00
5.55900693e-01 -1.35115888e-02 9.29160655e-01 -8.48662436e-01
-2.96848863e-01 3.67303371e-01 -1.97377857e-02 5.13516247e-01
-2.53647327e-01 1.36727601e-01 9.21134889e-01 -4.03944939e-01
4.12143350e-01 6.50550067e-01 7.67881393e-01 6.01275742e-01
-1.09953916e+00 -4.39908206e-01 -3.79423201e-01 8.00501406e-02
-1.09774029e+00 -1.16257273e-01 5.50385058e-01 -4.71744180e-01
2.37155065e-01 4.39206034e-01 4.32341278e-01 1.13731682e+00
5.92210770e-01 6.60729527e-01 1.11351490e+00 -3.49727511e-01
2.13909879e-01 5.49228072e-01 2.66145710e-02 5.72776437e-01
3.41924846e-01 1.97252348e-01 -1.07721463e-01 -1.45829037e-01
5.99040449e-01 2.11229697e-02 -4.94937152e-01 -3.64709646e-01
-1.03455818e+00 9.94648814e-01 6.75354064e-01 7.16763914e-01
-2.20859259e-01 1.21263564e-02 6.35643899e-01 2.66892493e-01
5.80201626e-01 6.84559524e-01 -4.54855621e-01 1.47827100e-02
-7.08438098e-01 2.60945261e-01 7.79731989e-01 2.41827890e-01
4.04709876e-01 -2.41448939e-01 -4.56107140e-01 9.20244277e-01
-1.07266635e-01 1.50278255e-01 7.72369862e-01 -3.62840742e-01
5.42499162e-02 5.09098470e-01 -2.18637645e-01 -8.37687075e-01
-5.99431813e-01 -9.56399798e-01 -9.33951437e-01 1.41635478e-01
6.60622060e-01 -2.51123488e-01 -8.60826850e-01 1.62006783e+00
8.36913288e-03 -2.19183102e-01 -5.22934236e-02 9.86387134e-01
7.77714670e-01 3.83713618e-02 8.97455886e-02 6.94528744e-02
1.16526568e+00 -5.38034499e-01 -4.03188676e-01 -1.38124108e-01
7.31405973e-01 -5.47511697e-01 1.06217527e+00 2.56909132e-01
-9.56374884e-01 -3.89506817e-01 -1.06579852e+00 1.80521742e-01
-3.55610639e-01 5.09926118e-02 4.80515003e-01 6.85462236e-01
-6.68929040e-01 8.50190759e-01 -9.59308445e-01 -2.93755084e-01
8.44639122e-01 3.55666786e-01 -2.53156513e-01 -1.70702204e-01
-1.07106447e+00 9.52788234e-01 3.67258012e-01 9.58596468e-02
-8.32189560e-01 -8.37561071e-01 -4.63153392e-01 3.18641871e-01
7.70835131e-02 -6.96286142e-01 1.32240403e+00 -1.22590673e+00
-1.12567127e+00 1.04010558e+00 4.40237373e-01 -7.02213705e-01
8.35754097e-01 1.59113318e-01 -1.66490331e-01 -1.19962841e-01
-1.23447396e-01 4.55854118e-01 6.95390463e-01 -9.97281909e-01
-3.77368033e-01 -2.56914318e-01 -6.77295104e-02 -1.94150120e-01
-4.18438464e-01 -2.73179203e-01 1.15103163e-01 -6.71852171e-01
-1.99839380e-02 -9.90698457e-01 -1.96696058e-01 1.10482581e-01
-5.82656145e-01 9.27433744e-03 6.82012022e-01 -2.92855442e-01
1.03366876e+00 -2.24876618e+00 -5.31971827e-02 2.63099998e-01
2.32148156e-01 3.90097320e-01 7.96784759e-02 1.45556882e-01
-1.23164289e-01 2.25912973e-01 -4.34290946e-01 1.61486119e-01
-2.61210650e-01 1.81844637e-01 1.05164580e-01 5.17698586e-01
5.17978728e-01 8.50026846e-01 -7.66592383e-01 -2.91392893e-01
1.28058463e-01 7.71211147e-01 -4.70021755e-01 2.36635998e-01
-9.53157768e-02 7.78693616e-01 -5.22820294e-01 3.72023523e-01
2.83172995e-01 -4.63760287e-01 7.05672577e-02 -1.80405915e-01
3.08812529e-01 9.58548486e-02 -4.22977000e-01 1.11426270e+00
-5.87604165e-01 8.61158848e-01 -3.94228995e-01 -1.32562125e+00
7.79815018e-01 3.40488225e-01 4.61474299e-01 -5.38048089e-01
3.82141858e-01 3.40837538e-01 5.49077213e-01 -7.23030090e-01
-4.61679623e-02 -3.76530141e-01 1.27827883e-01 1.31705090e-01
8.99306163e-02 -2.30950676e-03 6.62086010e-02 -4.16230977e-01
1.29755437e+00 -4.44188058e-01 3.28459799e-01 -4.02219236e-01
3.62627596e-01 -6.18276894e-02 3.26377779e-01 7.18049407e-01
-3.00484240e-01 8.27731788e-01 7.71937966e-01 -6.95215583e-01
-1.05315280e+00 -8.69584799e-01 -8.30384851e-01 6.06581211e-01
-1.21552378e-01 2.39228442e-01 -7.05136299e-01 -9.25091147e-01
1.25723686e-02 4.44980651e-01 -8.88426542e-01 -5.67267597e-01
-3.70635599e-01 -1.16820538e+00 6.00459695e-01 3.76341581e-01
3.85493726e-01 -1.19816387e+00 -1.03236401e+00 6.33781850e-02
6.77858144e-02 -9.87542868e-01 -1.77513227e-01 5.78534305e-01
-9.96057272e-01 -1.36385679e+00 -8.28678906e-01 -6.68107808e-01
7.33815849e-01 -2.51803339e-01 1.23093712e+00 1.92951754e-01
-6.51627004e-01 2.70337015e-01 -3.71446282e-01 -6.06930315e-01
-6.78626299e-01 3.90176237e-01 -3.94106358e-01 1.34096280e-01
3.07531416e-01 -3.50763589e-01 -8.31664741e-01 2.52168238e-01
-1.11782372e+00 -3.14859301e-01 7.97609329e-01 1.23321581e+00
4.85938400e-01 6.06383085e-02 6.11017168e-01 -1.12217700e+00
7.81044900e-01 -5.66944659e-01 -2.19714373e-01 1.62363291e-01
-5.50373673e-01 1.97445527e-01 7.20705330e-01 -5.56740105e-01
-6.16023779e-01 -2.30117872e-01 -8.33989084e-02 -3.43064100e-01
-1.53554648e-01 7.12813377e-01 2.87362933e-01 -2.45863125e-01
9.35020447e-01 5.47390468e-02 2.13670403e-01 -1.76758125e-01
-4.23286587e-01 5.04756331e-01 1.40353233e-01 -2.84609318e-01
3.99098814e-01 2.53088892e-01 2.77708948e-01 -7.15413928e-01
-9.94095504e-01 -2.45221034e-01 -2.61812210e-01 -2.54905075e-01
9.21801388e-01 -4.03682739e-01 -5.53054392e-01 2.39976883e-01
-7.05230534e-01 -3.74840736e-01 -5.09817541e-01 6.50570095e-01
-4.34125572e-01 -7.18158036e-02 -3.92338932e-01 -6.89775586e-01
-2.33540326e-01 -1.46051633e+00 7.85914183e-01 2.86021292e-01
-2.99695373e-01 -1.25477648e+00 -3.87662053e-02 8.03368986e-02
7.27355301e-01 5.34912825e-01 1.36328709e+00 -9.50064540e-01
-3.18208128e-01 -6.28550112e-01 -2.02345073e-01 8.38554800e-01
1.80315018e-01 -9.34270769e-02 -1.12621963e+00 -3.65599960e-01
7.66830370e-02 -5.34517705e-01 9.87621963e-01 7.91682303e-01
1.65332413e+00 -1.20116755e-01 -3.18550289e-01 6.25180542e-01
1.43285656e+00 2.31836855e-01 7.08827376e-01 3.12024027e-01
1.98316589e-01 5.55024385e-01 1.45565078e-01 1.64637163e-01
-3.04538876e-01 4.33053821e-01 6.61241055e-01 -2.03572407e-01
5.40036969e-02 1.51164770e-01 -1.93660602e-01 2.98724294e-01
-5.19103259e-02 -3.71922761e-01 -9.71768081e-01 3.95422637e-01
-1.39788818e+00 -6.91510141e-01 2.76783705e-01 2.30841279e+00
8.09245110e-01 2.21033916e-01 4.50037345e-02 3.49950820e-01
5.67237079e-01 -2.45189667e-02 -6.33331597e-01 -2.33209729e-01
-2.08356902e-02 3.43475878e-01 5.33884287e-01 9.81575698e-02
-1.30028987e+00 1.96235850e-01 6.21836090e+00 6.67565465e-01
-1.67091262e+00 -1.21532448e-01 1.32463872e+00 6.51084483e-02
-4.38862592e-02 -4.53873038e-01 -2.21401289e-01 4.46662813e-01
7.35467970e-01 7.18348846e-02 -1.21481776e-01 8.39481294e-01
2.33208053e-02 -1.44782752e-01 -1.23781693e+00 8.87438893e-01
-6.14850856e-02 -1.15481901e+00 -1.88917264e-01 1.05627924e-01
7.64810085e-01 7.25168213e-02 4.73485261e-01 1.94378331e-01
-2.58588880e-01 -1.38727379e+00 3.51572514e-01 4.30462539e-01
7.30994761e-01 -6.72030747e-01 1.19995546e+00 2.59740442e-01
-4.30920094e-01 -1.67838112e-01 -3.46013367e-01 2.39005446e-01
-2.78318554e-01 8.78618240e-01 -1.12539780e+00 2.08166987e-01
6.91400111e-01 3.19437683e-01 -6.01332366e-01 1.38881230e+00
1.68525219e-01 8.49826396e-01 -4.26777564e-02 -3.50462139e-01
3.04798454e-01 1.66341051e-01 3.45869958e-01 1.05772758e+00
2.65324086e-01 -1.73508242e-01 -7.21884007e-03 7.75356412e-01
-2.76532859e-01 1.34187892e-01 -5.53890646e-01 -9.51925367e-02
7.94782713e-02 1.17880380e+00 -8.22980881e-01 3.00387945e-02
-1.44421905e-01 5.06099522e-01 1.16256833e-01 2.89219111e-01
-7.34036148e-01 -2.32905626e-01 5.61166883e-01 4.82926250e-01
2.26668254e-01 2.93439269e-01 -3.66613448e-01 -8.52881193e-01
2.10808799e-01 -7.10391641e-01 2.87681729e-01 -3.02796990e-01
-1.53829551e+00 7.79882848e-01 6.94668153e-03 -1.28344321e+00
-1.72319517e-01 -8.10078442e-01 -6.72475934e-01 6.19153619e-01
-1.51173735e+00 -6.45329475e-01 -3.86047035e-01 1.35872513e-01
2.63099641e-01 -2.36612245e-01 8.00375879e-01 3.49288553e-01
-4.45807785e-01 7.03083634e-01 3.48332971e-01 4.65806067e-01
4.97502893e-01 -1.09260035e+00 -7.92004019e-02 2.56671518e-01
-2.66119000e-03 1.89307004e-01 7.61253595e-01 -9.43646803e-02
-6.96866632e-01 -1.08087242e+00 4.97014165e-01 -1.93215162e-01
3.39507908e-01 -6.23940639e-02 -9.75073278e-01 3.82021964e-01
-1.48539126e-01 5.87571859e-01 6.60184264e-01 -1.66441530e-01
-2.13723511e-01 -1.63626909e-01 -1.37098897e+00 3.63325685e-01
5.91013670e-01 -3.25087607e-01 4.77164648e-02 3.71046662e-01
1.28799558e-01 -4.20244992e-01 -1.02704072e+00 8.31815660e-01
6.72049463e-01 -1.22527075e+00 8.47990394e-01 -9.07179356e-01
6.34327650e-01 3.06746930e-01 7.76157454e-02 -1.47121537e+00
-5.20598069e-02 1.41580775e-01 3.63716513e-01 5.08863688e-01
6.70694172e-01 -7.82515407e-01 9.40687120e-01 3.67571801e-01
2.41531935e-02 -1.47931540e+00 -9.63084757e-01 -5.71111083e-01
3.16162139e-01 -2.66748726e-01 1.90093130e-01 9.53642845e-01
-3.69537354e-01 -2.94479076e-02 7.66546428e-02 -2.49599576e-01
3.37054223e-01 -2.87372947e-01 3.03936273e-01 -1.42195570e+00
-2.58032858e-01 -6.02446735e-01 -9.01952207e-01 -2.98915714e-01
4.70162667e-02 -1.10311699e+00 4.82287607e-04 -1.04470444e+00
3.20415616e-01 -6.19885564e-01 -3.99117649e-01 2.61319906e-01
-9.37852338e-02 2.92573214e-01 5.25059178e-04 3.11879870e-02
-1.58629015e-01 3.27804357e-01 1.33785903e+00 -4.02513087e-01
9.01425928e-02 4.53376561e-01 -5.08708060e-01 5.81521273e-01
8.87801170e-01 -6.73420727e-01 -4.52960849e-01 -2.55326182e-01
1.32489815e-01 -7.88298622e-02 6.23871386e-01 -1.25680482e+00
-1.30168721e-01 6.84895739e-02 6.55410230e-01 6.47987798e-03
1.60364956e-02 -7.86447763e-01 2.08327398e-02 8.08146834e-01
-7.41480410e-01 -5.82254939e-02 6.68251589e-02 4.42502052e-01
-5.30123889e-01 -4.57533568e-01 1.08202076e+00 -2.10866988e-01
-1.15252011e-01 4.65213120e-01 -1.32430613e-01 3.07103723e-01
9.46694911e-01 -1.29177779e-01 -2.10552849e-02 -3.39977711e-01
-7.76876628e-01 -3.00208926e-01 2.00876459e-01 1.67142421e-01
4.38611209e-01 -1.09992611e+00 -7.38448083e-01 1.54924929e-01
1.98966742e-01 1.47318080e-01 8.65789652e-02 1.00855911e+00
-6.62681639e-01 3.78537208e-01 -1.34074688e-01 -9.29600120e-01
-1.03446317e+00 1.87863544e-01 9.60381269e-01 -5.74201107e-01
-3.54133517e-01 7.71179974e-01 3.03432316e-01 -2.48352319e-01
2.42191747e-01 -4.53978151e-01 -2.17202771e-03 -1.53673008e-01
2.92492628e-01 1.83386013e-01 4.70541418e-01 -1.41791597e-01
-1.22340992e-01 2.29353398e-01 -1.86880037e-01 2.65475869e-01
1.29371858e+00 5.67766905e-01 1.17721014e-01 5.68944991e-01
1.54867554e+00 -5.99080265e-01 -1.04124904e+00 -1.27937645e-01
1.51954651e-01 -2.92020410e-01 1.02439389e-01 -8.15091789e-01
-1.34344447e+00 1.00721359e+00 1.05306697e+00 5.50468862e-01
1.13513911e+00 8.72835889e-02 4.56109732e-01 4.09031272e-01
6.61050007e-02 -8.18536818e-01 2.87864536e-01 2.33684495e-01
7.71616936e-01 -1.53555548e+00 -2.49949008e-01 -2.44323775e-01
-4.48222011e-01 1.16956258e+00 4.04027253e-01 -1.99284315e-01
8.54579985e-01 3.27514052e-01 1.27467051e-01 -1.44790232e-01
-5.00440478e-01 -1.22061506e-01 3.13889414e-01 5.70492685e-01
7.58824587e-01 5.99070378e-02 -3.13162178e-01 4.65994775e-01
-1.07810810e-01 1.13235690e-01 3.83668840e-01 8.45115185e-01
-1.47355840e-01 -8.78684342e-01 -2.55856607e-02 9.46097255e-01
-9.73718166e-01 1.44709796e-01 -1.33331373e-01 9.70044017e-01
5.67444973e-02 4.10095572e-01 2.37320125e-01 -2.52577484e-01
2.39481464e-01 1.19182944e-01 5.12501955e-01 -6.89085662e-01
-7.40495324e-01 -4.88073051e-01 -2.34615728e-02 -1.77576005e-01
-4.02452201e-01 -6.52694345e-01 -8.17466736e-01 6.65159710e-03
-4.76849258e-01 6.20419383e-02 6.62899375e-01 9.13090527e-01
-9.31361169e-02 8.51327717e-01 6.01817667e-01 -6.20464146e-01
-8.83372664e-01 -1.01138461e+00 -5.15041530e-01 5.90116799e-01
5.27289271e-01 -6.79568887e-01 -5.06738365e-01 -2.08853856e-01]
|
[14.966662406921387, -2.4610869884490967]
|
36d2bcc3-02e2-4d4b-ae42-53f0883cd127
|
partial-matrix-completion
|
2208.12063
| null |
https://arxiv.org/abs/2208.12063v1
|
https://arxiv.org/pdf/2208.12063v1.pdf
|
Partial Matrix Completion
|
In the matrix completion problem, one wishes to reconstruct a low-rank matrix based on a revealed set of (possibly noisy) entries. Prior work considers completing the entire matrix, which may be highly inaccurate in the common case where the distribution over entries is non-uniform. We formalize the problem of Partial Matrix Completion where the goal is to complete a large subset of the entries, or equivalently to complete the entire matrix and specify an accurate subset of the entries. Interestingly, even though the distribution is unknown and arbitrarily complex, our efficient algorithm is able to guarantee: (a) high accuracy over all completed entries, and (b) high coverage, meaning that it covers at least as much of the matrix as the distribution of observations.
|
['Adam Tauman Kalai', 'Elad Hazan', 'Varun Kanade']
|
2022-08-25
| null | null | null | null |
['matrix-completion']
|
['methodology']
|
[ 3.37234974e-01 2.06259429e-01 -6.59678951e-02 8.38509873e-02
-1.30306852e+00 -1.18750191e+00 2.11944625e-01 -4.61191051e-02
-1.23375133e-01 8.81328464e-01 5.18784225e-01 -2.02005535e-01
-3.88242036e-01 -4.21104133e-01 -9.82636333e-01 -8.67984474e-01
-2.46513322e-01 1.01961899e+00 -4.10971671e-01 -1.09141298e-01
-1.93051875e-01 3.53400767e-01 -9.42491174e-01 6.89878762e-02
5.20899177e-01 7.35169053e-01 1.52195096e-01 7.43633270e-01
3.42430830e-01 6.15911663e-01 -3.17680180e-01 -2.68900067e-01
7.15309024e-01 -9.15295333e-02 -7.33257115e-01 7.19371796e-01
-2.49258000e-02 -2.65086979e-01 -5.10104537e-01 1.46039927e+00
4.90293391e-02 -1.14302762e-01 5.08908808e-01 -1.00292051e+00
-3.39471549e-01 7.15480030e-01 -6.93166971e-01 -1.98717266e-01
8.50578904e-01 -1.33000195e-01 1.32278025e+00 -1.08036602e+00
8.42784047e-01 7.03809917e-01 5.53598225e-01 -4.13169004e-02
-1.79990947e+00 -4.90677088e-01 -1.84136201e-02 -3.76679033e-01
-1.92438400e+00 -7.69804180e-01 4.58161533e-01 -5.86874664e-01
-5.97958937e-02 3.16253573e-01 3.26498359e-01 8.62755239e-01
-1.47711277e-01 7.45736122e-01 9.92934585e-01 -2.02036917e-01
4.57766980e-01 7.95659497e-02 2.72623867e-01 2.15887353e-01
8.89173865e-01 2.36982089e-02 -6.46673799e-01 -9.06306088e-01
4.11655188e-01 1.99378550e-01 -6.25913799e-01 -4.47723866e-01
-1.59619808e+00 7.37776458e-01 -3.07764888e-01 -4.34047021e-02
-6.30733132e-01 4.63649258e-02 -2.58663315e-02 6.17029428e-01
5.79230599e-02 4.29510921e-01 -4.74672973e-01 3.44133526e-02
-1.01072121e+00 4.80739027e-01 1.36133611e+00 1.32272303e+00
1.23504043e+00 -7.84934536e-02 6.96795061e-02 3.54670197e-01
-1.79092005e-01 8.92201424e-01 -3.77469927e-01 -9.74088490e-01
7.99434483e-01 1.69685841e-01 8.16799521e-01 -1.03558958e+00
-8.65922123e-02 -5.42614698e-01 -1.12290323e+00 -1.09686358e-02
6.58071876e-01 -3.94167155e-01 -5.38747966e-01 1.97548735e+00
1.90934777e-01 -7.66594931e-02 7.65526369e-02 1.01622212e+00
4.14042547e-02 5.70641100e-01 -7.33089268e-01 -5.23977876e-01
1.07874417e+00 -1.25520557e-01 -6.94288194e-01 -6.03507519e-01
2.20674261e-01 -7.80181289e-01 6.44884586e-01 6.56486690e-01
-9.70009148e-01 1.72650218e-02 -1.06912887e+00 1.50885433e-01
3.63093942e-01 2.43151307e-01 7.51105130e-01 6.60804629e-01
-1.06199026e+00 1.48956791e-01 -6.80787444e-01 7.95191899e-02
7.13433921e-02 6.28265560e-01 -9.08758223e-01 -8.36693466e-01
-6.29807234e-01 1.19388647e-01 1.34710103e-01 1.30105257e-01
-1.08380318e+00 -6.39828742e-01 -5.89848578e-01 3.68120223e-02
7.57765472e-01 -7.16462970e-01 9.16021824e-01 -6.85247481e-01
-5.86782098e-01 6.34542942e-01 -4.67535287e-01 -1.75329536e-01
5.09213746e-01 -6.95121139e-02 -2.68418267e-02 1.01380795e-01
3.08147579e-01 -5.84491082e-02 8.23014319e-01 -1.44521081e+00
-4.93467271e-01 -5.51611900e-01 1.70664087e-01 2.08315730e-01
1.36100315e-02 -5.71021199e-01 -8.04844081e-01 -5.11714101e-01
6.95831716e-01 -1.22247827e+00 -6.73061252e-01 -2.23271340e-01
-9.19174910e-01 4.72751826e-01 2.95138061e-01 -8.35528135e-01
1.12723088e+00 -2.20245957e+00 4.57582116e-01 7.94741988e-01
6.68327153e-01 -5.01699269e-01 -3.03619951e-01 1.10230601e+00
-3.97285074e-02 -6.16062917e-02 -4.82269913e-01 -6.07586861e-01
-7.63556510e-02 5.18170118e-01 -6.95348382e-01 1.11767972e+00
-4.16255683e-01 3.92957121e-01 -7.88293898e-01 2.10811898e-01
-1.40189379e-01 1.44756481e-03 -7.07385182e-01 -6.43287739e-03
-1.17844373e-01 5.59465110e-01 -6.93792522e-01 7.55781829e-01
1.00149608e+00 -3.84247750e-01 6.00078523e-01 6.74000978e-02
3.27183902e-01 -2.74132527e-02 -2.18115473e+00 1.64957142e+00
5.16148806e-02 2.60506362e-01 8.86912763e-01 -7.25278735e-01
4.94986773e-01 5.58412135e-01 9.42841113e-01 1.00138880e-01
-6.85008764e-02 2.90715754e-01 -1.80225417e-01 -4.59475927e-02
6.86668932e-01 -2.02473313e-01 -2.62578189e-01 9.56446290e-01
-1.71706766e-01 3.72585505e-01 1.65260598e-01 7.04742551e-01
1.56108403e+00 -6.36436999e-01 4.23852056e-01 -3.32958311e-01
7.75416642e-02 3.67166772e-02 7.18807936e-01 1.07439697e+00
4.00678486e-01 9.72592652e-01 7.23608255e-01 1.32793384e-02
-1.34816074e+00 -1.09290743e+00 -3.39735672e-02 5.19373298e-01
1.00482970e-01 -4.45523351e-01 -4.33751553e-01 -1.46235958e-01
1.43309519e-01 2.06675306e-01 -6.82451010e-01 2.77190775e-01
-6.88044280e-02 -7.73136735e-01 1.31977469e-01 5.47083654e-02
-8.91102329e-02 -4.86526936e-02 3.51851821e-01 3.01499486e-01
-4.86044079e-01 -1.31829202e+00 -9.30072963e-01 1.74632937e-01
-7.61299253e-01 -1.31536877e+00 -2.90719092e-01 -4.89028215e-01
1.15118945e+00 6.10491276e-01 9.28580403e-01 -6.40563443e-02
1.54625788e-01 5.85631609e-01 -2.84898847e-01 9.00175497e-02
-7.95155093e-02 -1.82223648e-01 3.78643453e-01 5.43930233e-01
-8.41673987e-04 -9.24527168e-01 -3.00797433e-01 1.43805653e-01
-1.09705698e+00 -3.21003795e-01 5.61836302e-01 7.62956440e-01
7.81895280e-01 4.25110459e-01 3.40244710e-01 -1.45905912e+00
7.72735715e-01 -8.38053644e-01 -6.58041358e-01 4.66259047e-02
-3.18317205e-01 9.72637385e-02 5.06766319e-01 -5.20521760e-01
-4.80911463e-01 6.28711820e-01 2.43375883e-01 -5.00024736e-01
2.56334901e-01 8.57398033e-01 -3.40323061e-01 6.92242384e-02
7.22248912e-01 4.16505009e-01 -4.27326523e-02 -7.23604798e-01
4.89763886e-01 3.54692221e-01 7.05507934e-01 -9.13123488e-01
1.24014306e+00 7.85738945e-01 2.51408756e-01 -8.21424425e-01
-6.98771715e-01 -8.55816841e-01 -6.55886650e-01 3.34556192e-01
-6.68574348e-02 -1.47583711e+00 -7.96134531e-01 -1.30503461e-01
-7.53795743e-01 -2.26749126e-02 -4.85085011e-01 6.01326168e-01
-4.81116116e-01 5.16668439e-01 -2.99524963e-01 -1.05110359e+00
1.08066581e-01 -1.07852292e+00 9.52187896e-01 -6.14143848e-01
-2.69786060e-01 -8.52725625e-01 1.54040828e-01 1.15556329e-01
4.29092534e-02 3.86718154e-01 6.19895101e-01 -8.54146361e-01
-1.00703299e+00 -8.37672234e-01 -1.43111125e-02 5.60256094e-03
-3.22366841e-02 -5.44004142e-01 -6.68602407e-01 -7.49259949e-01
1.62581012e-01 -1.99300811e-01 7.05843568e-01 3.91801745e-01
7.34097242e-01 -6.79609120e-01 -3.73873800e-01 4.33037937e-01
1.44209850e+00 -3.32604319e-01 4.13395494e-01 -2.77845532e-01
3.49396586e-01 4.99161601e-01 5.39194226e-01 1.27129841e+00
4.17429268e-01 4.18872982e-01 3.15634370e-01 2.00029224e-01
3.61036479e-01 -4.54479158e-01 2.48017594e-01 7.52419233e-01
3.79828602e-01 -1.92794487e-01 -5.95306993e-01 7.32143402e-01
-1.93849826e+00 -8.28804016e-01 -1.75723746e-01 2.64078927e+00
9.08522010e-01 -1.76520482e-01 5.36332801e-02 2.84493327e-01
5.68524897e-01 1.71097681e-01 -5.75108826e-01 3.48661602e-01
-4.78459448e-01 -1.61559641e-01 8.71548951e-01 8.82834554e-01
-7.90288568e-01 3.41476947e-01 7.62955713e+00 7.33350217e-01
-3.38428438e-01 7.30833504e-04 2.85699904e-01 -1.84926346e-01
-8.77492130e-01 6.46638930e-01 -8.03848088e-01 2.27802902e-01
5.17251790e-01 -3.34151000e-01 1.08901525e+00 6.80683672e-01
1.09000579e-01 -3.86932969e-01 -1.45524609e+00 1.08742177e+00
-6.67382032e-02 -1.16248834e+00 -1.13632726e-02 7.74864018e-01
1.21432292e+00 -1.06197938e-01 -2.78648529e-02 -8.52315202e-02
8.01890314e-01 -9.70815718e-01 6.63513958e-01 6.69328094e-01
1.24886036e+00 -7.19680071e-01 4.97157186e-01 8.63295555e-01
-1.04321718e+00 -1.66479319e-01 -6.03072762e-01 -1.79298535e-01
2.34992832e-01 1.34112132e+00 -6.66821420e-01 4.80500251e-01
3.00748885e-01 5.65785408e-01 3.19056436e-02 9.13237274e-01
-4.70970431e-03 8.94403815e-01 -8.74978542e-01 4.87913191e-01
-3.28966603e-02 -7.44983554e-01 8.31719160e-01 7.48193443e-01
4.03168738e-01 4.84675735e-01 9.01587427e-01 5.40415823e-01
-3.78893226e-01 -6.06089085e-02 -7.15544820e-01 -2.53541142e-01
8.25987697e-01 1.03919756e+00 -2.20238551e-01 -1.98878899e-01
-4.68589485e-01 8.76335740e-01 3.22860926e-01 8.40474188e-01
-8.99375752e-02 1.18040694e-02 9.13238704e-01 2.05700785e-01
3.63391250e-01 -5.98311484e-01 -6.44047201e-01 -1.31825817e+00
3.79513055e-01 -1.06467068e+00 5.12268364e-01 -3.41317266e-01
-1.27463043e+00 1.25624374e-01 -3.94065768e-01 -1.22915554e+00
-3.00536454e-01 -1.87362000e-01 -1.17460884e-01 9.71717656e-01
-8.39472711e-01 -9.09117818e-01 2.00998634e-01 8.22668314e-01
-8.78720451e-03 -2.29951784e-01 6.72644854e-01 3.95811051e-01
-2.29421362e-01 2.59950757e-01 5.85119069e-01 -1.92521387e-04
3.91499907e-01 -1.23843277e+00 -7.46557266e-02 1.08284295e+00
2.56266266e-01 9.86225009e-01 1.10592139e+00 -8.38517666e-01
-2.32221746e+00 -9.87879157e-01 7.36974120e-01 -6.60570323e-01
7.32308209e-01 -4.37997878e-01 -5.60235322e-01 1.28997850e+00
-4.39624369e-01 1.58537403e-01 7.63612151e-01 3.59621048e-01
-4.10846233e-01 -1.79415452e-03 -1.05496430e+00 4.97411877e-01
7.31534004e-01 -6.98957086e-01 -2.86437213e-01 5.99683583e-01
4.16848421e-01 -5.00378788e-01 -7.99585819e-01 1.50299221e-01
4.79311913e-01 -5.42647898e-01 9.13757145e-01 -4.60099608e-01
-7.60639533e-02 -6.47158563e-01 -9.76251364e-01 -8.98020387e-01
-3.73589218e-01 -9.87002373e-01 -3.68832082e-01 8.87087107e-01
4.46973324e-01 -4.31495458e-01 1.07542920e+00 7.27058589e-01
1.97021976e-01 -3.67009610e-01 -1.09711564e+00 -5.78318954e-01
-4.33467358e-01 -7.24447250e-01 5.22545159e-01 8.13362300e-01
-2.12966129e-01 2.76125580e-01 -1.23679554e+00 7.60137200e-01
9.10219789e-01 3.39804441e-01 1.16900444e+00 -1.13550615e+00
-6.67391002e-01 3.50347430e-01 1.04719196e-02 -1.58038914e+00
-6.85231760e-02 -9.35437202e-01 -7.63495192e-02 -1.33124316e+00
6.69725895e-01 -7.07801938e-01 1.19158149e-01 2.15365484e-01
6.43384531e-02 2.56791770e-01 -1.14974692e-01 5.68482816e-01
-8.29926729e-01 2.80550480e-01 1.11180627e+00 5.44083677e-03
-1.16443709e-01 5.47577858e-01 -1.28631794e+00 3.09384704e-01
1.59003973e-01 -7.12787092e-01 -4.47074890e-01 -2.12003529e-01
6.92179024e-01 7.71965802e-01 7.24344924e-02 -7.16272235e-01
5.05848885e-01 -2.54909813e-01 2.21461475e-01 -8.71737301e-01
5.97317278e-01 -1.01325619e+00 8.23399127e-01 3.22011828e-01
-1.70806929e-01 -2.06540421e-01 -3.94457370e-01 1.22774172e+00
-1.88572675e-01 -2.28527501e-01 3.02823663e-01 -6.96849003e-02
-1.27513573e-01 7.73008585e-01 -3.76947522e-01 3.17478120e-01
7.09817410e-01 -2.55996734e-01 9.71536934e-02 -9.58867073e-01
-1.05859351e+00 2.89457291e-01 7.76418090e-01 -8.17409456e-02
5.20883381e-01 -1.32078946e+00 -7.57958412e-01 2.10719362e-01
2.84981132e-01 4.71823253e-02 2.73064137e-01 7.36273587e-01
-6.94188699e-02 4.10419166e-01 5.10085464e-01 -3.94257396e-01
-8.00196528e-01 7.08787560e-01 -1.81295067e-01 -2.80800283e-01
-6.10900879e-01 5.01184583e-01 2.26108119e-01 -4.51963902e-01
2.55685449e-01 1.54491320e-01 4.28902209e-01 -7.10442290e-02
7.27917433e-01 3.66228610e-01 -1.04040310e-01 -6.10849619e-01
-7.54358172e-02 2.99304098e-01 1.41741503e-02 -3.89653414e-01
1.25749576e+00 -5.48138499e-01 -3.30197185e-01 3.15369308e-01
1.09069252e+00 8.83206129e-01 -1.31140113e+00 -9.05701280e-01
-1.67578444e-01 -8.16292286e-01 -2.98892200e-01 -5.15491068e-01
-9.37185884e-01 3.67407203e-01 -1.78809628e-01 -8.94670486e-02
9.99198496e-01 -4.16551076e-04 5.80275834e-01 7.23809898e-01
9.62499857e-01 -5.76203227e-01 -2.65390426e-01 3.74804974e-01
8.89045298e-01 -1.06664109e+00 3.96759421e-01 -4.79953349e-01
-6.17835462e-01 8.43560278e-01 -2.48975515e-01 -6.94992840e-02
8.64766002e-01 4.77149397e-01 -3.69715035e-01 -4.05643284e-02
-1.08681607e+00 -8.16603675e-02 1.32141829e-01 6.82932556e-01
-1.54973835e-01 3.71673822e-01 -1.09156400e-01 8.49328339e-01
-3.55431646e-01 -2.91352183e-01 1.01000202e+00 7.71291554e-01
-5.24337828e-01 -1.16876638e+00 -7.95480371e-01 8.73542905e-01
-5.05860448e-01 -3.34387459e-02 -2.58911282e-01 4.28840578e-01
-3.10508460e-01 1.28883576e+00 -5.17190814e-01 -4.14243996e-01
2.65505314e-01 -4.05682534e-01 7.89843276e-02 -8.81226838e-01
7.79223591e-02 3.35079730e-01 2.86499232e-01 -5.65200388e-01
1.28912613e-01 -1.02249634e+00 -7.83722341e-01 -7.01771796e-01
-4.68154773e-02 4.04962450e-01 3.71586531e-01 9.97948647e-01
1.84445113e-01 -2.41268545e-01 8.78932655e-01 -4.38419193e-01
-1.03990054e+00 -8.07266057e-01 -1.48655331e+00 2.60200799e-01
7.91180909e-01 -3.83938700e-01 -4.26715940e-01 3.93548086e-02]
|
[6.935223579406738, 4.7225518226623535]
|
26e49297-4c42-48fc-a2a3-f31cc371e4d2
|
transferring-hierarchical-structure-with-dual
| null | null |
https://openreview.net/forum?id=t3E10H8UNz
|
https://openreview.net/pdf?id=t3E10H8UNz
|
Transferring Hierarchical Structure with Dual Meta Imitation Learning
|
Hierarchical Imitation learning (HIL) is an effective way for robots to learn sub-skills from long-horizon unsegmented demonstrations. However, the learned hierarchical structure lacks the mechanism to transfer across multi-tasks or to new tasks, which makes them have to learn from scratch when facing a new situation. Transferring and reorganizing modular sub-skills require fast adaptation ability of the whole hierarchical structure. In this work, we propose Dual Meta Imitation Learning (DMIL), a hierarchical meta imitation learning method where the high-level network and sub-skills are iteratively meta-learned with model-agnostic meta-learning. DMIL uses the likelihood of state-action pairs from each sub-skill as the supervision for the high-level network adaptation, and use the adapted high-level network to determine different data set for each sub-skill adaptation. We theoretically prove the convergence of the iterative training process of DMIL and establish the connection between DMIL and the Expectation-Maximization algorithm. Empirically, we achieve state-of-the-art few-shot imitation learning performance on the meta-world benchmark.
|
['Feng Chen', 'Yizhou Jiang', 'Chongkai Gao']
|
2021-09-29
| null | null | null | null |
['few-shot-imitation-learning']
|
['methodology']
|
[-8.76857638e-02 1.43419221e-01 -8.05082768e-02 -1.04618460e-01
-4.40088451e-01 -1.83462992e-01 5.94266474e-01 -3.85139972e-01
-6.00938320e-01 8.94608736e-01 -7.16721406e-03 2.10387364e-01
-1.52902797e-01 -5.65961063e-01 -1.14484739e+00 -7.67155111e-01
-2.38134131e-01 5.41123629e-01 6.63962603e-01 -2.57235855e-01
1.45906359e-01 2.23639518e-01 -1.53753448e+00 3.20769101e-01
9.84908402e-01 4.91640359e-01 1.02188206e+00 7.83321798e-01
5.60523570e-02 1.10395956e+00 -4.14011031e-01 1.55674607e-01
3.25051576e-01 -6.91231549e-01 -9.42226112e-01 1.18307844e-01
-6.51142821e-02 -5.48991740e-01 -6.82536006e-01 1.07994044e+00
2.94807255e-01 5.76098502e-01 7.71286607e-01 -1.30552006e+00
-2.53016651e-01 6.27124786e-01 -2.77774423e-01 1.19268030e-01
1.00094073e-01 5.57161868e-01 5.53494513e-01 -6.55008912e-01
6.94291353e-01 1.41151440e+00 4.58194405e-01 8.84734154e-01
-9.12766755e-01 -5.80260396e-01 2.60642022e-01 4.70408291e-01
-8.77644837e-01 -3.26456651e-02 4.52123404e-01 -3.63725752e-01
1.22880352e+00 -7.04727173e-01 9.62163150e-01 1.26367927e+00
6.83924437e-01 1.02524209e+00 1.23400033e+00 -1.62827775e-01
3.91731232e-01 -2.03603446e-01 -3.48709434e-01 1.01119864e+00
-1.02432139e-01 4.98372376e-01 -5.71680844e-01 3.53642479e-02
1.13429606e+00 2.27972254e-01 1.28319606e-01 -6.82587206e-01
-1.49545050e+00 5.10896802e-01 6.18778408e-01 1.85025096e-01
-7.46141255e-01 5.61573029e-01 5.29007852e-01 8.82407844e-01
-1.37706712e-01 5.53161561e-01 -6.36864424e-01 -2.35280305e-01
-4.63515759e-01 -1.01203695e-01 7.34685540e-01 1.08185399e+00
9.87378716e-01 1.88131228e-01 1.59931816e-02 7.00137317e-01
1.78059533e-01 1.76705107e-01 1.01259899e+00 -1.45003355e+00
4.22102779e-01 5.43245912e-01 3.57414745e-02 -1.75359339e-01
-3.75656128e-01 -2.23051950e-01 -7.19820440e-01 5.06726861e-01
9.43990424e-02 -3.47862542e-01 -1.01743257e+00 1.88995707e+00
3.89239520e-01 4.69251305e-01 3.93719733e-01 5.81447303e-01
3.48372579e-01 8.49410355e-01 -1.20005660e-01 -3.44865263e-01
7.12917328e-01 -1.71418560e+00 -2.17368156e-01 -3.43731582e-01
5.18080711e-01 -8.81830081e-02 9.92373466e-01 2.73251742e-01
-1.14523423e+00 -9.93032575e-01 -1.16257155e+00 2.41425157e-01
-8.68092254e-02 -1.73575699e-01 1.96964741e-01 -3.80341232e-01
-8.86304557e-01 1.02423799e+00 -9.73497808e-01 -4.38488662e-01
1.88131735e-01 3.79182339e-01 -3.91488254e-01 2.88996547e-02
-1.04156220e+00 1.19461060e+00 9.47807431e-01 -1.96742773e-01
-2.01306844e+00 -3.01262170e-01 -7.52232671e-01 6.26505837e-02
5.76542735e-01 -9.75801229e-01 1.60477602e+00 -1.07478833e+00
-2.11687803e+00 4.11937565e-01 1.43128246e-01 -5.01412570e-01
4.80971277e-01 -1.54568315e-01 9.65442210e-02 2.33476803e-01
2.36996666e-01 9.98808146e-01 1.36792243e+00 -1.22032821e+00
-1.08686912e+00 -9.31095108e-02 1.44019112e-01 5.85431755e-01
-8.19563344e-02 -5.79597294e-01 -2.04384804e-01 -3.78122479e-01
1.66634060e-02 -1.07773066e+00 -3.34268212e-01 -3.95322859e-01
8.56721625e-02 -4.25845832e-01 7.32230484e-01 -3.96086276e-01
6.20855093e-01 -2.09435987e+00 9.03415918e-01 -2.71746516e-01
9.43192989e-02 1.76481977e-01 -4.35505956e-01 6.00822985e-01
2.92676210e-01 -5.38922131e-01 -2.02506557e-01 -2.39988282e-01
3.41925509e-02 5.40635943e-01 -9.03979093e-02 1.46410137e-01
-1.57300234e-01 1.12619722e+00 -1.45858419e+00 -5.38135946e-01
2.55198151e-01 1.81818858e-03 -3.51507843e-01 6.17742300e-01
-3.93130600e-01 8.03472817e-01 -4.73396331e-01 3.28086942e-01
-3.62578891e-02 -2.45977938e-01 2.02791899e-01 1.56706810e-01
-4.19028569e-03 3.89832072e-02 -7.37654209e-01 2.21109653e+00
-7.07051098e-01 3.85755062e-01 3.51370536e-02 -1.12803948e+00
7.89063036e-01 2.71726161e-01 3.74331087e-01 -5.65991759e-01
7.71007016e-02 1.98856920e-01 2.83445239e-01 -7.19389796e-01
2.37928301e-01 -1.85234860e-01 -2.08159134e-01 5.17426670e-01
6.96811676e-01 -5.03649890e-01 8.24057981e-02 -2.08257530e-02
1.13426149e+00 6.55451059e-01 5.69642425e-01 8.00565034e-02
4.43442315e-01 1.52771652e-01 5.60947597e-01 1.03040719e+00
-3.86308998e-01 -4.25211303e-02 1.43064305e-01 -4.03548092e-01
-1.23717189e+00 -1.24122870e+00 5.80484390e-01 1.38507926e+00
2.40267888e-01 -1.08240899e-02 -7.47897327e-01 -8.68315816e-01
4.23258357e-02 5.53244531e-01 -7.36544490e-01 -5.83504140e-01
-6.31712973e-01 -1.21566787e-01 2.93090135e-01 5.11492968e-01
8.85526359e-01 -1.73404753e+00 -1.14497757e+00 5.92009664e-01
5.03099971e-02 -8.62452567e-01 -3.73537153e-01 4.00848776e-01
-1.24784565e+00 -1.00168014e+00 -8.08359027e-01 -1.23622799e+00
6.84916556e-01 3.40021253e-01 7.31673419e-01 -7.28182197e-02
1.44286687e-02 6.08718693e-01 -2.83179432e-01 -1.41308084e-01
-8.58240306e-01 2.12255582e-01 4.62701023e-01 -3.91325593e-01
-1.48023248e-01 -1.05017030e+00 -4.08868492e-01 4.98235673e-01
-6.07609630e-01 3.23941082e-01 1.16433454e+00 1.26480818e+00
6.06712699e-01 -1.87784601e-02 8.33211720e-01 -2.24082738e-01
6.92824304e-01 -4.05889720e-01 -5.87741733e-01 3.24480891e-01
-4.96445596e-01 3.48517776e-01 8.29190373e-01 -9.50228512e-01
-1.08081591e+00 1.35588795e-01 1.81347236e-01 -7.53391623e-01
-9.44206193e-02 3.86549205e-01 1.86240003e-01 2.67452858e-02
4.65994120e-01 7.33059049e-01 2.29610801e-01 -3.42346251e-01
5.03663540e-01 4.79267985e-01 7.06640601e-01 -6.71265066e-01
6.03370488e-01 3.46774727e-01 -6.12564608e-02 -5.83749294e-01
-8.50830436e-01 -4.10500765e-01 -1.05387032e+00 -3.36670130e-01
8.10451329e-01 -9.04229760e-01 -7.94865191e-01 8.70039463e-01
-1.01026618e+00 -1.14608526e+00 -5.41224480e-01 5.17513990e-01
-1.31469786e+00 3.48502040e-01 -9.13188517e-01 -4.70922112e-01
-2.00142950e-01 -1.28093171e+00 7.15165854e-01 2.77518809e-01
1.73029944e-01 -1.01226163e+00 3.93602252e-01 -7.45173618e-02
1.96475685e-01 -8.28662887e-02 7.17065811e-01 -3.46566796e-01
-4.38341022e-01 9.60687548e-02 1.37122005e-01 5.25248408e-01
1.83248341e-01 -4.97074068e-01 -5.38086355e-01 -8.04979563e-01
2.39265278e-01 -9.90740061e-01 8.32570195e-01 2.78461844e-01
9.12798941e-01 -3.39409471e-01 -2.82938838e-01 3.49906296e-01
1.08655047e+00 1.22113094e-01 4.96420383e-01 6.06110632e-01
6.31531835e-01 5.08789241e-01 8.36304963e-01 3.75627011e-01
4.33259934e-01 5.10701537e-01 5.97813368e-01 5.82040846e-01
5.19586774e-03 -6.43182337e-01 8.78787398e-01 1.24605215e+00
-8.47698376e-02 4.30094898e-01 -5.25443733e-01 4.90171283e-01
-2.27729774e+00 -1.04602945e+00 6.08107746e-01 1.99229586e+00
8.35246563e-01 3.08487803e-01 2.11630747e-01 -5.97960949e-01
6.88106656e-01 -7.45932851e-03 -1.18304551e+00 -2.22832859e-01
3.48261148e-01 -2.46599063e-01 1.73439354e-01 3.86071563e-01
-8.33627999e-01 1.28176010e+00 6.37293434e+00 8.25230360e-01
-8.45971763e-01 3.05553615e-01 -8.25584754e-02 -8.31686854e-02
5.23934603e-01 2.30175927e-02 -5.91902375e-01 4.62912530e-01
7.30739772e-01 -3.05939198e-01 9.56576765e-01 1.08156645e+00
1.34603241e-02 -2.36226857e-01 -1.34502721e+00 7.22961307e-01
-1.24615997e-01 -1.00473332e+00 4.44088392e-02 -2.38834042e-02
1.13456237e+00 5.44191241e-01 -1.00300141e-01 1.00747621e+00
7.39901006e-01 -7.07410812e-01 5.60992539e-01 5.66686630e-01
4.65936363e-01 -5.61805665e-01 4.69557554e-01 1.05935895e+00
-1.20030320e+00 -5.86128294e-01 -7.80841887e-01 -2.10272729e-01
1.50201783e-01 -2.81542301e-01 -7.67383933e-01 4.00721967e-01
5.95411241e-01 8.63284409e-01 -2.71320969e-01 8.69916141e-01
-6.60058498e-01 2.28952616e-01 -2.96891369e-02 -1.90226614e-01
6.42665863e-01 -1.53937444e-01 6.96834207e-01 8.56037080e-01
3.33658248e-01 -9.27463267e-03 4.49656099e-01 6.62336409e-01
-6.18485771e-02 -4.78935838e-01 -6.97534382e-01 3.25731598e-02
4.67953920e-01 1.15011311e+00 -2.83862203e-01 -7.41814733e-01
-2.19447628e-01 1.32960713e+00 8.91436815e-01 3.62220556e-01
-7.40024149e-01 -1.59485474e-01 3.95249575e-01 -5.05709171e-01
4.10371810e-01 -4.12456661e-01 3.83741766e-01 -1.11308670e+00
-3.20805222e-01 -9.72601771e-01 3.03093046e-01 -1.02341294e+00
-1.21411598e+00 2.65018761e-01 1.80162475e-01 -1.57223439e+00
-7.46818304e-01 -4.08997983e-01 -7.49143958e-01 4.06550378e-01
-1.33895671e+00 -9.21863079e-01 -3.64938825e-01 6.91442192e-01
1.11546254e+00 -5.55113196e-01 5.91658652e-01 -3.74102056e-01
-3.69352251e-01 3.88973773e-01 3.69066507e-01 -9.28384438e-02
4.82389838e-01 -1.22994101e+00 3.51712853e-01 4.60803390e-01
-9.41801444e-02 3.25862736e-01 6.50905430e-01 -7.39146411e-01
-1.33803821e+00 -1.00330293e+00 4.98135500e-02 -2.08728433e-01
8.91255736e-01 -1.08354323e-01 -9.24699068e-01 8.68984878e-01
2.63666540e-01 -3.65760207e-01 -3.19414809e-02 -2.67409325e-01
-1.15291081e-01 5.19909300e-02 -9.94409859e-01 7.78817654e-01
1.25703120e+00 -5.22208810e-01 -1.41165066e+00 2.28325456e-01
9.26826656e-01 -3.64610761e-01 -9.29754019e-01 3.97602201e-01
6.85613930e-01 -8.47962856e-01 8.16883087e-01 -8.59769583e-01
5.75467348e-01 -1.77283391e-01 1.60551846e-01 -1.95523274e+00
-5.74449122e-01 -5.96570492e-01 -5.25601864e-01 4.38881546e-01
2.17770070e-01 -7.07412302e-01 5.89482129e-01 -2.96994269e-01
-5.73200464e-01 -7.74862945e-01 -1.01163113e+00 -1.38579857e+00
1.54706866e-01 1.35605127e-01 3.00372750e-01 7.39859700e-01
5.92026353e-01 5.18328011e-01 -5.46024859e-01 1.87487267e-02
7.28096187e-01 2.14171544e-01 1.11010337e+00 -8.95389915e-01
-8.02260697e-01 -4.07506913e-01 -1.54791072e-01 -1.39970684e+00
5.13585329e-01 -8.15424740e-01 4.38416034e-01 -1.72396922e+00
3.98309678e-01 -2.61567868e-02 -4.81895655e-01 5.22470772e-01
-1.43671319e-01 -4.58453685e-01 3.23581845e-01 6.56095505e-01
-1.06405568e+00 1.04808772e+00 1.84725738e+00 -2.30410486e-01
-5.02614260e-01 -1.17468201e-02 1.66098289e-02 8.35729599e-01
9.84171212e-01 -6.99301660e-01 -6.43800676e-01 -3.86661738e-01
-1.51016891e-01 1.79816246e-01 2.33319327e-01 -1.26376903e+00
5.26926637e-01 -3.58000278e-01 3.14337134e-01 -3.22280467e-01
4.65982169e-01 -6.60095215e-01 -1.65840253e-01 1.13184237e+00
-3.30435932e-01 2.01692775e-01 6.13816865e-02 8.44412744e-01
-1.06501013e-01 -5.68492889e-01 8.80713344e-01 -7.14624047e-01
-1.19704592e+00 3.21101159e-01 -6.35130763e-01 1.02446705e-01
1.18231094e+00 -3.11670065e-01 -2.24731714e-01 -2.53397554e-01
-9.44902897e-01 7.74993777e-01 6.71080172e-01 5.81086278e-01
7.39586651e-01 -1.20848846e+00 -4.58519012e-01 -9.71810967e-02
1.66847222e-02 1.18359484e-01 2.89729655e-01 7.93579757e-01
-9.68621820e-02 -5.93883619e-02 -8.54278266e-01 -6.79791808e-01
-7.99124658e-01 8.40768039e-01 6.17253065e-01 -4.29096878e-01
-8.16664875e-01 6.66046143e-01 3.32093120e-01 -7.31356382e-01
2.81135887e-01 -2.57809579e-01 -2.12358939e-03 -3.08012456e-01
2.32160047e-01 6.63144946e-01 -6.60668612e-01 -2.29476586e-01
9.93935317e-02 4.64559793e-01 -1.37216389e-01 -4.09280509e-01
1.28136027e+00 -8.04412663e-02 -8.88231471e-02 9.54566061e-01
1.02453637e+00 -9.88360226e-01 -2.01661491e+00 -3.51144642e-01
6.22525869e-04 -1.51285683e-04 -4.64742124e-01 -5.91086686e-01
-4.33804780e-01 9.61494803e-01 3.13907236e-01 -2.68847674e-01
6.57054543e-01 1.21606901e-01 7.76014686e-01 1.16447961e+00
9.76261556e-01 -1.66513050e+00 1.04259956e+00 9.67144966e-01
9.31313217e-01 -1.14553738e+00 -7.59760439e-02 3.13467473e-01
-7.32634664e-01 9.29920733e-01 1.03329134e+00 -3.92693162e-01
3.38410825e-01 -1.19785465e-01 -2.91270703e-01 -9.23752412e-02
-1.01088738e+00 -8.07641894e-02 -1.90081506e-03 6.80360138e-01
-4.65855211e-01 -6.14077896e-02 1.32218197e-01 1.78247884e-01
6.66554272e-02 1.23242058e-01 4.05949086e-01 1.22214878e+00
-1.07411659e+00 -8.45086217e-01 6.10026270e-02 4.48333323e-01
2.77246505e-01 3.37617129e-01 -1.68516204e-01 7.91336656e-01
-7.06140995e-02 5.77147067e-01 -1.62397206e-01 -4.86715376e-01
2.75437355e-01 1.46840900e-01 1.03232205e+00 -9.02190626e-01
-4.38917547e-01 -1.65004477e-01 -3.08876425e-01 -6.57245815e-01
-4.07498777e-01 -5.69729924e-01 -1.57893634e+00 -4.78544012e-02
6.67930394e-03 2.15872705e-01 3.83210808e-01 1.32194138e+00
3.45416576e-01 7.28391290e-01 7.44691014e-01 -1.30420446e+00
-1.14462709e+00 -1.34278548e+00 -4.50674623e-01 3.07137311e-01
3.48364711e-01 -9.28545237e-01 -3.53440017e-01 -5.58028966e-02]
|
[4.366684913635254, 1.166095495223999]
|
a018e0a2-e8f1-47be-ac04-3ed00988b4c7
|
csvideonet-a-real-time-end-to-end-learning
|
1612.05203
| null |
http://arxiv.org/abs/1612.05203v5
|
http://arxiv.org/pdf/1612.05203v5.pdf
|
CSVideoNet: A Real-time End-to-end Learning Framework for High-frame-rate Video Compressive Sensing
|
This paper addresses the real-time encoding-decoding problem for
high-frame-rate video compressive sensing (CS). Unlike prior works that perform
reconstruction using iterative optimization-based approaches, we propose a
non-iterative model, named "CSVideoNet". CSVideoNet directly learns the inverse
mapping of CS and reconstructs the original input in a single forward
propagation. To overcome the limitations of existing CS cameras, we propose a
multi-rate CNN and a synthesizing RNN to improve the trade-off between
compression ratio (CR) and spatial-temporal resolution of the reconstructed
videos. The experiment results demonstrate that CSVideoNet significantly
outperforms the state-of-the-art approaches. With no pre/post-processing, we
achieve 25dB PSNR recovery quality at 100x CR, with a frame rate of 125 fps on
a Titan X GPU. Due to the feedforward and high-data-concurrency natures of
CSVideoNet, it can take advantage of GPU acceleration to achieve three orders
of magnitude speed-up over conventional iterative-based approaches. We share
the source code at https://github.com/PSCLab-ASU/CSVideoNet.
|
['Kai Xu', 'Fengbo Ren']
|
2016-12-15
| null | null | null | null |
['video-compressive-sensing']
|
['computer-vision']
|
[ 2.94730932e-01 -2.61330783e-01 2.95703225e-02 -1.44359946e-01
-7.41993725e-01 -1.91709548e-01 2.64565915e-01 -4.45597053e-01
-4.43593502e-01 3.51029992e-01 4.00663525e-01 -5.05314410e-01
1.36560932e-01 -4.73161399e-01 -1.01874089e+00 -4.55279261e-01
-2.27892026e-01 -2.80562967e-01 8.28457549e-02 -1.26393944e-01
3.04597765e-01 1.73950449e-01 -1.39284194e+00 5.30588448e-01
5.38812399e-01 1.29938233e+00 6.50621235e-01 9.99732852e-01
3.56857508e-01 1.24887490e+00 -9.74599868e-02 -1.52950451e-01
6.20441675e-01 -5.62183797e-01 -5.19541323e-01 1.06526881e-01
2.37219021e-01 -1.04375744e+00 -1.22772014e+00 1.02012718e+00
5.45459032e-01 -5.20064719e-02 -5.58543764e-02 -8.08758914e-01
-6.08824313e-01 4.38470036e-01 -5.85075676e-01 2.22972795e-01
5.33857226e-01 3.13476771e-01 6.95018291e-01 -1.14127815e+00
5.30847132e-01 1.13287652e+00 5.85688114e-01 3.52724165e-01
-8.83976579e-01 -8.31639230e-01 -3.40663731e-01 3.80573332e-01
-1.65414262e+00 -9.24791455e-01 4.82739955e-01 -4.50628139e-02
8.00043344e-01 6.14740662e-02 7.22804844e-01 9.25374925e-01
1.57670230e-01 7.06428111e-01 8.59614670e-01 -2.76142865e-01
1.69261366e-01 -6.24544084e-01 -6.83993816e-01 6.90187812e-01
6.13353867e-03 3.79708797e-01 -8.41378510e-01 3.02893072e-01
1.47047853e+00 1.53432608e-01 -7.88010299e-01 6.38807788e-02
-1.49982786e+00 6.32088065e-01 5.46624780e-01 1.49474323e-01
-4.55544412e-01 7.27633595e-01 3.79822284e-01 4.70053256e-01
1.60508454e-01 1.10804476e-01 -8.27010572e-02 -2.97495633e-01
-1.29205298e+00 -1.01694822e-01 6.01543963e-01 1.17253697e+00
4.64901060e-01 5.61729610e-01 2.72937845e-02 6.12245858e-01
3.37736964e-01 5.72022796e-01 5.19203365e-01 -1.60840046e+00
5.61930418e-01 -1.97223201e-01 -1.64342690e-02 -1.03568673e+00
1.85875058e-01 -6.02976799e-01 -1.13986254e+00 3.79567780e-02
-2.22022101e-01 -1.88179895e-01 -7.37308025e-01 1.25562239e+00
1.29970178e-01 6.53610528e-01 7.72126541e-02 1.40328944e+00
6.63182735e-01 1.01319242e+00 -4.99973089e-01 -4.06133622e-01
9.55036998e-01 -1.19789660e+00 -6.37504041e-01 -2.80145705e-01
3.00448179e-01 -8.71557474e-01 6.80068433e-01 4.87776518e-01
-1.26466322e+00 -5.84676921e-01 -1.27553523e+00 -2.37204701e-01
6.21311724e-01 3.05505365e-01 4.31455195e-01 1.69246659e-01
-1.25318956e+00 7.84841359e-01 -1.22746885e+00 7.67893121e-02
5.33001184e-01 2.30934471e-01 -2.38303542e-01 -4.80607659e-01
-7.80050397e-01 2.82782048e-01 3.32136184e-01 3.62244174e-02
-1.26092052e+00 -8.67716491e-01 -6.73529804e-01 1.91956222e-01
4.76894468e-01 -6.19184613e-01 1.15190351e+00 -1.09090781e+00
-1.81488764e+00 3.56872648e-01 -2.44437590e-01 -5.24197578e-01
5.95677972e-01 -4.34102446e-01 -2.28776187e-01 6.56466305e-01
-1.10403128e-01 6.07926726e-01 9.22777474e-01 -1.00761116e+00
-4.06321764e-01 2.78266557e-02 -1.07935570e-01 2.70405740e-01
-2.21493557e-01 -3.72150093e-02 -9.51883256e-01 -7.80753851e-01
3.77530515e-01 -8.48835468e-01 -2.58972555e-01 4.97656107e-01
-1.68846786e-01 6.42380893e-01 7.04627216e-01 -8.59505653e-01
1.07046854e+00 -2.44486117e+00 1.67966023e-01 -3.31174761e-01
4.36327726e-01 4.53374624e-01 -2.23421067e-01 5.32274008e-01
5.27671427e-02 -2.45891690e-01 -3.24452639e-01 -4.91101146e-01
-3.66018742e-01 8.95746499e-02 -3.58997494e-01 7.19415843e-01
-1.48422003e-01 7.95089126e-01 -8.80749762e-01 -1.07990161e-01
2.45112360e-01 9.30463016e-01 -8.22608650e-01 4.71170485e-01
-3.92062142e-02 7.68059075e-01 -2.69086063e-01 6.24124169e-01
8.42248678e-01 -5.64706743e-01 4.69984561e-01 -3.62009853e-01
-2.72133380e-01 3.66100341e-01 -1.13043213e+00 2.33606815e+00
-4.36288387e-01 9.80476797e-01 4.98369604e-01 -9.36429143e-01
5.19516945e-01 3.78578842e-01 4.67970729e-01 -9.50487792e-01
3.47583324e-01 5.17716110e-01 -3.36943209e-01 -3.62405211e-01
6.33825779e-01 -5.65587953e-02 4.93106127e-01 2.58665025e-01
4.06427728e-03 1.00965805e-01 1.91744901e-02 3.88641596e-01
1.39440024e+00 3.86935651e-01 2.24003911e-01 -1.00589566e-01
2.76859552e-01 -3.55142474e-01 6.16778374e-01 5.72620273e-01
2.49168314e-02 1.02839482e+00 9.86292437e-02 -2.08350480e-01
-1.42020762e+00 -9.40842807e-01 1.96859278e-02 8.00321877e-01
3.26476663e-01 -6.13637865e-01 -4.73493695e-01 1.79022461e-01
-3.63073736e-01 3.79155576e-01 -3.08954762e-03 1.00255616e-01
-7.40663886e-01 -2.32009202e-01 5.30784667e-01 3.37748855e-01
1.01944840e+00 -6.86938465e-01 -8.82826447e-01 3.22297275e-01
-4.50665027e-01 -1.67202330e+00 -5.04310191e-01 -1.52302861e-01
-1.04590225e+00 -8.40751827e-01 -8.77274394e-01 -7.08564997e-01
4.84696984e-01 7.63979733e-01 8.96589875e-01 3.86822760e-01
-4.05200571e-02 1.67755038e-01 -5.40188551e-01 2.58835256e-01
-3.40820730e-01 -2.93980718e-01 -1.74066499e-02 3.62867564e-02
-2.56031543e-01 -9.15476441e-01 -1.11485815e+00 7.64188319e-02
-1.16488779e+00 5.02970040e-01 6.01645708e-01 7.46822417e-01
6.70525849e-01 -1.26792982e-01 2.14858919e-01 -5.45456231e-01
3.30629386e-02 -4.86573786e-01 -7.34245121e-01 -1.29598424e-01
-5.28601170e-01 -1.54869169e-01 7.68507361e-01 -4.07357454e-01
-7.77270675e-01 1.79785773e-01 -2.55412608e-01 -1.16711855e+00
3.00687730e-01 4.67539012e-01 2.89505631e-01 -2.20624745e-01
4.33559865e-01 7.07845747e-01 2.34234750e-01 -2.45515168e-01
1.58021480e-01 6.42944217e-01 8.83273900e-01 -3.65310848e-01
8.19177508e-01 6.79382205e-01 -1.72436476e-01 -7.94094026e-01
-5.46211481e-01 -3.52200300e-01 -1.07860111e-01 -2.99160898e-01
6.70523524e-01 -1.71007848e+00 -8.10198367e-01 5.19533217e-01
-1.08095086e+00 -6.73187613e-01 -1.84560999e-01 8.60298753e-01
-6.41396880e-01 7.14211583e-01 -1.10041130e+00 -5.01592755e-01
-4.69712734e-01 -1.18839264e+00 1.01817691e+00 -2.27620266e-03
3.91783386e-01 -5.44310868e-01 -3.01874280e-01 4.74409640e-01
7.97321439e-01 -5.94191812e-02 1.57872379e-01 2.89862126e-01
-1.21163964e+00 1.89422637e-01 -5.08825123e-01 4.28634971e-01
-2.44053274e-01 -4.75910574e-01 -8.27781975e-01 -8.72161508e-01
2.65267044e-01 -3.37656468e-01 8.92134964e-01 3.62110704e-01
1.32275367e+00 -5.68795919e-01 1.17662311e-01 1.40978384e+00
1.89838803e+00 7.75829256e-02 1.10573280e+00 1.15479670e-01
7.81084597e-01 -2.14084744e-01 4.47973192e-01 8.21196735e-01
4.24584389e-01 5.04159629e-01 6.97669804e-01 6.70422763e-02
-4.68231052e-01 -3.19652438e-01 6.21678472e-01 1.30698538e+00
-1.91191792e-01 -4.55658466e-01 -4.71296281e-01 3.93290371e-01
-1.71447766e+00 -8.92880797e-01 -4.53003682e-03 2.18530130e+00
6.89713478e-01 -1.87176421e-01 -3.67508203e-01 5.63188680e-02
5.43810427e-01 4.76038277e-01 -5.50296724e-01 9.54307392e-02
-1.32717207e-01 1.73344210e-01 8.10597301e-01 5.93439043e-01
-7.27991641e-01 8.00444782e-01 5.55835867e+00 8.59254420e-01
-1.41490376e+00 2.41240308e-01 5.29823661e-01 -4.97846395e-01
-1.91003352e-01 1.24513060e-01 -1.90657064e-01 4.92718637e-01
1.15657139e+00 1.44886598e-01 1.04580998e+00 5.56855619e-01
2.94732422e-01 -3.84104811e-02 -9.69549596e-01 1.39025843e+00
1.77862212e-01 -1.79858780e+00 -2.03532100e-01 2.27880031e-01
7.50290751e-01 5.22292554e-01 -5.61987869e-02 -5.93894199e-02
8.88375118e-02 -9.18519437e-01 9.76179600e-01 3.64678651e-01
1.48948932e+00 -5.10383844e-01 4.41969573e-01 2.42145494e-01
-1.31699693e+00 -5.99463172e-02 -5.61143637e-01 -1.86018005e-01
4.06680048e-01 7.31428266e-01 -3.70421350e-01 6.38521254e-01
8.17920387e-01 1.08543456e+00 -3.49761918e-02 9.52789187e-01
-1.00430898e-01 6.67244673e-01 -3.64078075e-01 4.81376797e-01
1.12802535e-01 -1.63543731e-01 5.85694492e-01 1.04116094e+00
9.30978537e-01 5.35700321e-01 4.13675979e-02 6.34788036e-01
-3.89815181e-01 -3.62113863e-01 -4.02794540e-01 -6.07358105e-02
5.80581844e-01 8.41081917e-01 -3.58466685e-01 -3.18742424e-01
-4.19861645e-01 1.42947900e+00 -2.43045576e-02 3.82802963e-01
-9.72643375e-01 -1.51898906e-01 4.76136237e-01 2.34830335e-01
7.70267189e-01 -6.54262841e-01 -1.14564650e-01 -1.58119726e+00
1.52076557e-01 -1.06733692e+00 -2.05103308e-01 -1.04794586e+00
-6.47989690e-01 5.07919848e-01 -4.17173117e-01 -1.53269970e+00
1.76475495e-02 -2.92721987e-01 -2.75690973e-01 4.98087555e-01
-1.67883897e+00 -8.74115944e-01 -5.35181642e-01 6.08893812e-01
7.60514081e-01 -1.16630569e-01 4.58031952e-01 6.38193607e-01
-3.22000861e-01 4.00982678e-01 3.90482426e-01 1.28379628e-01
4.19565558e-01 -5.11302650e-01 3.83874118e-01 1.30237651e+00
-1.28768325e-01 4.89440233e-01 6.09896779e-01 -6.22716606e-01
-2.22887301e+00 -1.15185797e+00 3.68886650e-01 4.28447425e-01
6.08592629e-01 -1.11780420e-01 -8.19475889e-01 6.81139886e-01
5.50754249e-01 5.10877788e-01 3.53474706e-01 -7.27800667e-01
-5.32059014e-01 -7.53596276e-02 -9.31044459e-01 4.88112688e-01
1.28269613e+00 -6.65309370e-01 1.01552084e-01 2.47723043e-01
8.58397067e-01 -7.86483467e-01 -8.38116467e-01 2.36200854e-01
4.94000196e-01 -1.31005323e+00 1.07456577e+00 3.23170066e-01
1.06609797e+00 -4.28066999e-01 -6.49813175e-01 -9.98800278e-01
-3.66018057e-01 -8.68617535e-01 -3.60428154e-01 5.99305570e-01
-1.14842817e-01 -4.29480791e-01 6.64794326e-01 -1.74634382e-01
-4.09309298e-01 -7.51592457e-01 -1.03611827e+00 -6.29660368e-01
-3.88785422e-01 -6.41258895e-01 3.46604705e-01 8.70918453e-01
-2.24490300e-01 2.43734345e-01 -8.27371597e-01 4.19427872e-01
8.03634584e-01 1.08188473e-01 6.32716060e-01 -4.81087029e-01
-8.50601971e-01 6.55640513e-02 -4.21510011e-01 -1.89577210e+00
-3.02989125e-01 -6.84528232e-01 1.21955700e-01 -1.31552422e+00
5.17253168e-02 -2.73542911e-01 1.25063118e-02 1.99142769e-01
3.16309839e-01 5.82774580e-01 5.12751639e-01 6.29863977e-01
-6.95281863e-01 7.47104824e-01 1.38785994e+00 1.10205159e-01
1.88306659e-01 -5.60794175e-01 -6.01375639e-01 3.69117051e-01
5.19455135e-01 -4.24741447e-01 -4.35524374e-01 -1.12220657e+00
8.95900428e-02 8.81256044e-01 4.46349919e-01 -1.37138975e+00
5.73097587e-01 -5.27331270e-02 3.18751752e-01 -3.77453059e-01
6.55306220e-01 -7.38096297e-01 3.67246091e-01 6.76355362e-01
-1.90444916e-01 2.73755612e-03 -9.22664702e-02 7.35649765e-01
-2.50962734e-01 9.31734815e-02 8.25922847e-01 -1.63103655e-01
-6.10612631e-01 4.97733355e-01 -3.88404578e-01 -8.65384489e-02
7.30951607e-01 -1.30915523e-01 -4.23799545e-01 -6.77017093e-01
-2.33374834e-01 -6.20217733e-02 7.01935470e-01 6.21335693e-02
1.11950612e+00 -1.20814717e+00 -8.60827506e-01 3.51261675e-01
-3.07283103e-01 1.51856720e-01 4.67826366e-01 9.38499153e-01
-1.23863828e+00 2.94181496e-01 -1.78282142e-01 -6.30626380e-01
-8.78977895e-01 2.99223036e-01 2.38515720e-01 1.10047683e-01
-1.05267310e+00 8.26720655e-01 -1.90242976e-02 3.38779669e-03
9.69523415e-02 8.83191451e-02 3.41204524e-01 -6.84619069e-01
7.51581430e-01 4.09363985e-01 -6.25949651e-02 -5.67435384e-01
-7.26573542e-02 3.27983141e-01 1.70672640e-01 -1.89057365e-01
1.50969124e+00 -3.13224137e-01 3.39931808e-02 -6.49627671e-02
1.44793510e+00 -6.63695997e-03 -1.74107492e+00 -2.87946612e-01
-6.20427608e-01 -9.70573068e-01 5.68094373e-01 -3.68373215e-01
-1.70368481e+00 6.85592473e-01 7.13693678e-01 -2.39204913e-01
1.43361092e+00 -3.13883752e-01 1.30412352e+00 2.34168336e-01
5.50995350e-01 -7.19054818e-01 1.53910607e-01 5.01079798e-01
9.63395894e-01 -1.18490505e+00 3.05663824e-01 -4.98855829e-01
-4.85207826e-01 1.15376747e+00 2.03108564e-01 -3.74428302e-01
5.16075552e-01 4.86061037e-01 -1.08330689e-01 -7.26358267e-03
-7.82214344e-01 1.54831186e-01 -1.13589607e-01 2.82082498e-01
4.06577587e-01 6.10885769e-03 -1.07044406e-01 -1.76649839e-01
-1.17564537e-01 4.33237523e-01 8.96532595e-01 1.05845881e+00
-3.22711796e-01 -7.56265223e-01 -2.29998767e-01 2.62775272e-01
-5.39818347e-01 -3.61184686e-01 4.67870981e-01 2.87343264e-01
-2.69694954e-01 1.06606245e+00 1.13705367e-01 -7.82679081e-01
-1.35006890e-01 -6.30175829e-01 4.82357115e-01 -3.15614223e-01
-3.29287529e-01 4.91644651e-01 -2.69934395e-03 -1.04697168e+00
-4.78847742e-01 -3.93295139e-01 -1.39248991e+00 -8.41588557e-01
-3.69762369e-02 -1.97024792e-01 7.44016588e-01 7.95931995e-01
7.11861491e-01 5.22163451e-01 8.90323460e-01 -1.08363140e+00
-3.78644347e-01 -6.97050989e-01 -4.26688731e-01 -4.48451960e-04
5.91960967e-01 -9.24404040e-02 -3.43255520e-01 1.62532941e-01]
|
[11.143889427185059, -2.035737991333008]
|
5e37c71e-6484-4e35-94df-9c97abd2880f
|
probabilistic-metamodels-for-an-efficient
|
2110.02892
| null |
https://arxiv.org/abs/2110.02892v3
|
https://arxiv.org/pdf/2110.02892v3.pdf
|
Probabilistic Metamodels for an Efficient Characterization of Complex Driving Scenarios
|
To validate the safety of automated vehicles (AV), scenario-based testing aims to systematically describe driving scenarios an AV might encounter. In this process, continuous inputs such as velocities result in an infinite number of possible variations of a scenario. Thus, metamodels are used to perform analyses or to select specific variations for examination. However, despite the safety criticality of AV testing, metamodels are usually seen as a part of an overall approach, and their predictions are not questioned. This paper analyzes the predictive performance of Gaussian processes (GP), deep Gaussian processes, extra-trees, and Bayesian neural networks (BNN), considering four scenarios with 5 to 20 inputs. Building on this, an iterative approach is introduced and evaluated, which allows to efficiently select test cases for common analysis tasks. The results show that regarding predictive performance, the appropriate selection of test cases is more important than the choice of metamodels. However, the choice of metamodels remains crucial: Their great flexibility allows BNNs to benefit from large amounts of data and to model even the most complex scenarios. In contrast, less flexible models like GPs convince with higher reliability. Hence, relevant test cases are best explored using scalable virtual test setups and flexible models. Subsequently, more realistic test setups and more reliable models can be used for targeted testing and validation.
|
['Steffen Müller', 'Hadj Hamma Tadjine', 'Mike Kohlhoff', 'Max Winkelmann']
|
2021-10-06
| null | null | null | null |
['probabilistic-deep-learning']
|
['computer-vision']
|
[-3.59828249e-02 -4.02414501e-02 1.76454857e-02 -2.50747025e-01
-2.73475498e-01 -4.76280451e-01 7.72432983e-01 2.50599474e-01
-1.48435324e-01 8.31939697e-01 -5.83055198e-01 -7.02114344e-01
-6.95922315e-01 -1.14165843e+00 -5.65385938e-01 -7.95197785e-01
-1.13270506e-01 7.99827635e-01 6.57783091e-01 -4.83076572e-02
8.20909664e-02 9.11930680e-01 -2.14569449e+00 -1.03857610e-02
1.06428742e+00 8.76947045e-01 2.96707392e-01 4.08366650e-01
-8.54465887e-02 4.14759457e-01 -5.46123803e-01 -4.64790910e-01
7.44616264e-04 -2.33848482e-01 -3.52351964e-01 -1.63989156e-01
-3.09007525e-01 -1.34570733e-01 2.50275105e-01 8.00992787e-01
3.74619007e-01 2.44224906e-01 7.73170888e-01 -1.66967499e+00
2.86123514e-01 4.56177711e-01 2.97287166e-01 -2.90716439e-01
2.56230295e-01 6.28223658e-01 5.43911219e-01 -5.10281265e-01
3.45621020e-01 1.19247270e+00 5.18872023e-01 2.40536079e-01
-1.49386382e+00 -5.70653796e-01 -4.72042244e-03 3.74970496e-01
-1.38034511e+00 -1.24371558e-01 7.59467840e-01 -6.15364134e-01
7.98266828e-01 3.87848645e-01 9.85421360e-01 1.43365073e+00
4.31085736e-01 3.10558319e-01 1.09302270e+00 -1.33740276e-01
7.85261035e-01 5.18198371e-01 -3.58930044e-02 -6.82045147e-02
8.43453765e-01 3.84815991e-01 7.08352402e-03 -2.21667260e-01
2.41023600e-01 -2.42709666e-01 -8.70394558e-02 -6.92874491e-01
-7.35993087e-01 7.12049782e-01 -3.44815612e-01 1.49482116e-01
-6.14845753e-01 7.08309375e-03 3.74627620e-01 7.20987841e-02
-1.09143987e-01 3.93534362e-01 -4.27141964e-01 -5.31186402e-01
-8.15131247e-01 6.29762411e-01 1.05785501e+00 7.59945750e-01
6.55651033e-01 2.32877761e-01 -3.32128555e-01 5.16589940e-01
4.86750156e-01 7.34672070e-01 5.48499636e-02 -7.13743567e-01
3.05761218e-01 7.49087512e-01 2.40599662e-01 -8.00583839e-01
-4.72322017e-01 -4.93158609e-01 -4.39314276e-01 5.76057673e-01
2.37407148e-01 -2.72610545e-01 -7.29494750e-01 1.31532669e+00
3.34416717e-01 6.93927184e-02 2.14923918e-01 3.12635362e-01
2.88357496e-01 5.10922194e-01 4.05795872e-01 -6.51018545e-02
1.07069683e+00 -2.62667313e-02 -4.55254316e-01 -1.45211414e-01
6.86689198e-01 -4.30902779e-01 8.43131542e-01 6.20282233e-01
-8.91784608e-01 -6.56115949e-01 -1.06527030e+00 8.97770047e-01
-4.92345482e-01 -3.30742419e-01 2.85839498e-01 1.07375824e+00
-8.89318049e-01 6.77113056e-01 -7.95873523e-01 -1.96377248e-01
8.79282132e-02 4.08806950e-01 -8.77128243e-02 -1.35819092e-01
-1.48141301e+00 1.25511801e+00 7.10767865e-01 3.78090620e-01
-1.05390573e+00 -5.99093139e-01 -7.89902747e-01 1.80948392e-01
3.40115726e-01 -4.92391586e-01 1.02950382e+00 -2.72420228e-01
-1.46860576e+00 -4.19381075e-03 8.29962566e-02 -5.74402928e-01
1.00552940e+00 1.22130901e-01 -7.06050336e-01 5.12394011e-02
-2.70359427e-01 2.96593606e-01 5.80810070e-01 -1.49450648e+00
-5.31427324e-01 1.96913123e-01 7.27792457e-02 -3.70915025e-01
3.11507553e-01 4.26739976e-02 -1.42849594e-01 1.65434331e-01
-2.07785442e-01 -1.01453340e+00 -3.62652600e-01 -7.23477066e-01
-3.03232312e-01 1.08114127e-02 7.41818368e-01 -4.99476373e-01
1.22557497e+00 -2.01604223e+00 -2.84779012e-01 8.13723803e-01
-2.92412549e-01 2.97597438e-01 2.59994805e-01 7.09281147e-01
1.77044235e-02 1.62430912e-01 -1.54146641e-01 7.49178901e-02
4.05289054e-01 4.22974020e-01 -6.55171787e-03 6.97326586e-02
4.44752693e-01 7.51575589e-01 -6.41849041e-01 -5.28189361e-01
6.82780981e-01 4.43115354e-01 -2.48927236e-01 -1.16457991e-01
-3.37520182e-01 5.26770890e-01 -7.42207587e-01 4.18409735e-01
6.42890751e-01 2.46866062e-01 4.16575335e-02 1.63006097e-01
-1.19233146e-01 -7.56336600e-02 -1.34781563e+00 4.51444060e-01
-6.82108581e-01 3.40116918e-01 -3.31279069e-01 -8.58632147e-01
1.33282423e+00 4.04493570e-01 3.05474967e-01 -7.17501819e-01
2.41505742e-01 3.07503581e-01 3.67924064e-01 -6.42764390e-01
2.89440811e-01 -2.72596870e-02 -1.19155822e-02 -5.85983396e-02
-2.37361521e-01 -6.08713567e-01 5.42852581e-01 -4.18354094e-01
1.00787938e+00 4.48422164e-01 7.97741413e-02 -7.00397417e-02
6.48881674e-01 4.22142111e-02 6.57028019e-01 6.74658298e-01
7.98269361e-03 2.13083580e-01 1.02120829e+00 -3.00620407e-01
-9.70559061e-01 -9.76451635e-01 -4.56032485e-01 1.97438434e-01
-7.28253350e-02 -1.48935989e-01 -6.24764502e-01 -4.32919800e-01
5.91048971e-02 1.55600488e+00 -4.24889296e-01 -3.59537303e-01
-3.45124245e-01 -6.60226643e-01 3.22553694e-01 3.52767050e-01
2.57848531e-01 -1.14271593e+00 -1.10059845e+00 3.34775567e-01
2.37207577e-01 -9.66824591e-01 7.46714056e-01 3.43032748e-01
-8.94398034e-01 -1.10955501e+00 -1.71822414e-01 -2.28008889e-02
4.54224825e-01 -2.93850332e-01 9.97334778e-01 -1.11252740e-01
1.85127392e-01 2.88339734e-01 -4.21349764e-01 -5.66985786e-01
-1.00215244e+00 -3.31758261e-01 -1.89277694e-01 -9.47656631e-02
3.02065432e-01 -6.73417270e-01 -2.70366907e-01 8.47274184e-01
-1.04693484e+00 -2.00799704e-01 7.14515209e-01 5.24935305e-01
6.14795625e-01 3.86352092e-01 6.62991464e-01 -6.28143728e-01
7.42840052e-01 -4.00046855e-01 -1.06041718e+00 4.54791844e-01
-8.06244731e-01 9.57040861e-02 7.95190930e-01 -5.22987485e-01
-1.07142234e+00 -2.45430052e-01 -3.36878538e-01 -4.06517267e-01
-7.36763775e-01 7.43793368e-01 -6.74480438e-01 -1.03230793e-02
5.60376585e-01 2.45716167e-03 -1.12703376e-01 -5.77413216e-02
-1.36600614e-01 2.44171143e-01 -1.34424791e-01 -8.07813287e-01
7.23047495e-01 -3.29790041e-02 6.02529049e-01 -7.24864423e-01
2.86535323e-01 1.67809576e-01 -6.21003509e-01 -8.13664675e-01
6.52649164e-01 -2.97964036e-01 -7.70131588e-01 1.41583920e-01
-9.07755196e-01 -5.33884823e-01 -3.90039891e-01 7.17545211e-01
-7.51154423e-01 1.22350045e-01 1.14447199e-01 -1.20377481e+00
4.05724615e-01 -1.77310097e+00 7.57236123e-01 -3.89923565e-02
-5.39078236e-01 -9.84005332e-01 -1.63081765e-01 1.75267283e-03
4.01268721e-01 5.58360577e-01 1.18306935e+00 -1.15788376e+00
-6.48391962e-01 -7.59615719e-01 3.23161364e-01 4.56495255e-01
-2.20623747e-01 4.66219455e-01 -8.71067524e-01 7.62493834e-02
-6.44656718e-02 2.76532590e-01 2.00250506e-01 2.38933906e-01
1.01999235e+00 2.83496026e-02 -3.96295160e-01 1.35712460e-01
1.35102665e+00 8.00656676e-01 8.89108896e-01 4.63589042e-01
1.71025693e-01 1.06101716e+00 9.01960909e-01 3.12018394e-01
5.35653830e-02 6.87427461e-01 6.78899765e-01 5.01815796e-01
4.03691679e-01 -9.31700766e-02 2.58501679e-01 1.36903778e-01
-1.80335581e-01 -2.80120969e-01 -1.29563224e+00 2.17596814e-01
-1.69521630e+00 -1.05919850e+00 -5.96139491e-01 2.67053413e+00
2.37945080e-01 7.36557782e-01 1.10678978e-01 5.44172347e-01
6.78571701e-01 -2.87041932e-01 -3.12635928e-01 -5.81791043e-01
4.42743208e-03 4.03451407e-03 3.66944432e-01 1.19476259e-01
-6.40977979e-01 2.93544829e-01 5.69029617e+00 9.79660809e-01
-1.05603814e+00 -2.13491827e-01 4.53006536e-01 1.92923695e-02
-4.54832286e-01 1.67510852e-01 -8.65539551e-01 7.35258698e-01
1.50707376e+00 -2.73692131e-01 2.87434203e-03 9.37408924e-01
7.39813328e-01 -5.15488923e-01 -1.07490718e+00 4.62078065e-01
-5.31275153e-01 -9.45542514e-01 -5.46845496e-02 2.89750099e-01
5.63463211e-01 -2.34792873e-01 -3.81710380e-02 5.17474234e-01
2.41485208e-01 -9.91200626e-01 9.25244808e-01 8.13699603e-01
2.08248407e-01 -1.02105331e+00 1.20999694e+00 3.53383303e-01
-7.69385457e-01 -2.42367804e-01 -2.78536081e-01 3.18087608e-01
4.42351222e-01 5.65144598e-01 -9.14620876e-01 8.80865216e-01
5.74043930e-01 -9.88005251e-02 -6.42970562e-01 1.51269150e+00
-3.23407799e-01 8.85618865e-01 -4.25074816e-01 -3.42966765e-01
1.84961826e-01 -2.38094330e-01 5.68503797e-01 1.09140682e+00
7.45191395e-01 -5.56985736e-01 -1.24912679e-01 9.62738276e-01
9.18291688e-01 7.33592883e-02 -8.32122624e-01 2.00578973e-01
5.41014612e-01 1.09457684e+00 -8.31205904e-01 -1.24556050e-02
-3.02449346e-01 1.55703090e-02 -4.22289401e-01 6.52302563e-01
-1.10404170e+00 -9.19527411e-02 6.42580807e-01 4.07413006e-01
4.58843619e-01 -1.71189204e-01 -3.55725825e-01 -2.06315249e-01
-1.94200000e-03 -7.56870031e-01 -6.72924817e-02 -8.88933718e-01
-9.13374424e-01 6.49096787e-01 7.73432255e-01 -1.60290051e+00
-8.00803483e-01 -7.75365651e-01 -8.67442429e-01 1.00664365e+00
-1.21652508e+00 -9.28768516e-01 -1.87497914e-01 1.55314133e-01
1.01589486e-01 8.34704749e-03 6.65038705e-01 4.83259559e-02
-4.11375016e-01 1.49598718e-01 6.16511963e-02 -5.17525375e-01
1.25187159e-01 -9.89705265e-01 2.11201057e-01 7.99687862e-01
-4.50487703e-01 4.63848054e-01 1.18001413e+00 -9.25435424e-01
-1.14259911e+00 -1.18008423e+00 5.88035822e-01 -3.71076643e-01
8.01337659e-01 -3.25352728e-01 -1.10430014e+00 1.68726161e-01
-1.16862185e-01 -2.95805305e-01 5.03972650e-01 7.14041013e-03
3.15953255e-01 -2.21254736e-01 -1.11735940e+00 8.32128227e-01
5.27638733e-01 -1.83137536e-01 -3.02308917e-01 -2.44809270e-01
4.32823896e-01 -7.22158775e-02 -9.92612183e-01 8.33259284e-01
5.25184512e-01 -1.16619337e+00 7.02700317e-01 -2.96078563e-01
6.92822933e-02 -4.28513259e-01 3.45927104e-02 -1.24032974e+00
-3.31884958e-02 -3.35473627e-01 1.02498010e-02 1.34753239e+00
8.42982471e-01 -9.86486971e-01 5.71865559e-01 1.16505325e+00
-2.13227719e-01 -8.11512053e-01 -1.03398359e+00 -1.10690463e+00
-5.54325208e-02 -1.34056449e+00 1.14528704e+00 3.91840369e-01
-2.67971486e-01 -2.91804910e-01 2.43668705e-02 3.97957087e-01
3.90958399e-01 -2.54477948e-01 8.33949149e-01 -1.60954189e+00
-1.85018897e-01 -5.51322937e-01 -8.45791340e-01 -4.10054512e-02
1.72377646e-01 -3.89284968e-01 3.56289931e-02 -1.63171637e+00
-3.65880042e-01 -5.20901620e-01 -7.86365196e-02 6.77501038e-02
3.14667150e-02 -2.83701986e-01 5.45368046e-02 -2.18736887e-01
-1.55581638e-01 4.55870718e-01 9.12872434e-01 -1.94446910e-02
-2.52234370e-01 6.55337036e-01 -4.23479453e-02 5.52873015e-01
9.75375891e-01 -3.48378897e-01 -7.35903680e-01 4.13412869e-01
6.17603242e-01 2.33191282e-01 7.54173279e-01 -1.30080748e+00
9.62937847e-02 -4.11106229e-01 2.94862479e-01 -8.75843287e-01
3.54672432e-01 -1.22386634e+00 1.03065562e+00 3.24869752e-01
1.05186857e-01 1.41731873e-01 5.07823706e-01 4.62951005e-01
-3.64893466e-01 -7.74828434e-01 3.60729665e-01 2.72025883e-01
-6.92378521e-01 4.73226830e-02 -7.50314772e-01 -5.20230591e-01
1.49587250e+00 -9.10336733e-01 1.07210450e-01 -2.20520526e-01
-8.61678064e-01 2.69351035e-01 4.66726065e-01 2.25683853e-01
4.40031648e-01 -9.90802765e-01 -4.55681771e-01 2.14210659e-01
2.35893026e-01 1.38525441e-01 5.65399766e-01 9.79688883e-01
-4.55628067e-01 5.72549045e-01 -2.06421956e-01 -7.20595956e-01
-8.74044538e-01 6.29007757e-01 3.53203237e-01 -1.56880498e-01
-2.89791971e-01 1.19814098e-01 3.83284166e-02 -1.89655900e-01
-1.56862989e-01 -4.51591492e-01 -5.04955411e-01 1.42286286e-01
2.43783414e-01 5.92331588e-01 4.39585865e-01 -5.10260701e-01
-1.83808506e-01 2.46472016e-01 6.07427716e-01 -1.06483705e-01
1.11162031e+00 5.36702611e-02 1.92528695e-01 5.15308499e-01
4.58370030e-01 -5.91040924e-02 -1.48634434e+00 6.06636405e-01
3.12682420e-01 -2.41097808e-01 -2.04863593e-01 -7.75062323e-01
-5.69280684e-01 7.69847155e-01 3.43635738e-01 6.42891765e-01
1.07483387e+00 -1.69347897e-01 -1.74267516e-01 2.27093056e-01
6.93285286e-01 -1.00775909e+00 -5.45289218e-01 3.06582510e-01
8.54356647e-01 -5.83319604e-01 -4.15604204e-01 -3.57387990e-01
-7.30955601e-01 1.13715601e+00 4.98442143e-01 2.93947101e-01
6.21051788e-01 4.73881543e-01 -3.94020915e-01 -4.85822465e-03
-8.41178119e-01 -1.80953443e-01 1.64128691e-01 8.43145907e-01
-7.98669830e-02 9.68058780e-02 -2.13823333e-01 6.89605951e-01
-2.34110773e-01 6.07771575e-02 3.95164996e-01 7.54010379e-01
-2.54145682e-01 -1.15749037e+00 -6.78667903e-01 6.33445323e-01
7.44768884e-03 2.44835734e-01 1.23617172e-01 1.24750149e+00
3.44735682e-01 1.20485985e+00 -2.27883935e-01 -4.35147822e-01
6.75740540e-01 5.06281376e-01 1.73227400e-01 -3.08115393e-01
-4.50526416e-01 -3.47985089e-01 5.56529939e-01 -4.11700457e-01
-3.47118489e-02 -9.67248678e-01 -8.50377560e-01 -2.62557000e-01
-2.96870768e-01 4.37588751e-01 1.01176953e+00 1.06763709e+00
2.01976433e-01 8.01340520e-01 3.38716179e-01 -6.74701333e-01
-5.91078043e-01 -8.87298942e-01 -5.52144527e-01 2.80724326e-03
-3.01914603e-01 -1.11526692e+00 -3.40589613e-01 -2.75270283e-01]
|
[5.7231597900390625, 1.4301674365997314]
|
8856739d-a7ce-4e0c-b140-b1da8d55aa4d
|
screen-content-image-segmentation-using-least
|
1501.03755
| null |
http://arxiv.org/abs/1501.03755v2
|
http://arxiv.org/pdf/1501.03755v2.pdf
|
Screen Content Image Segmentation Using Least Absolute Deviation Fitting
|
We propose an algorithm for separating the foreground (mainly text and line
graphics) from the smoothly varying background in screen content images. The
proposed method is designed based on the assumption that the background part of
the image is smoothly varying and can be represented by a linear combination of
a few smoothly varying basis functions, while the foreground text and graphics
create sharp discontinuity and cannot be modeled by this smooth representation.
The algorithm separates the background and foreground using a least absolute
deviation method to fit the smooth model to the image pixels. This algorithm
has been tested on several images from HEVC standard test sequences for screen
content coding, and is shown to have superior performance over other popular
methods, such as k-means clustering based segmentation in DjVu and shape
primitive extraction and coding (SPEC) algorithm. Such background/foreground
segmentation are important pre-processing steps for text extraction and
separate coding of background and foreground for compression of screen content
images.
|
['Yao Wang', 'Shervin Minaee']
|
2015-01-15
| null | null | null | null |
['foreground-segmentation']
|
['computer-vision']
|
[ 7.37317860e-01 -3.75006348e-01 -1.67270988e-01 -7.45600015e-02
-2.70894796e-01 -6.52154088e-01 4.83324438e-01 -2.30023399e-01
7.68373534e-02 4.75339323e-01 -2.08039716e-01 -4.83766258e-01
2.34511480e-01 -5.75662136e-01 -2.40511462e-01 -9.33684886e-01
3.97155851e-01 4.61804956e-01 1.02513528e+00 3.84544954e-02
4.76740330e-01 7.12409377e-01 -1.28726721e+00 4.73883480e-01
8.49395275e-01 7.56818116e-01 1.30094782e-01 1.27567077e+00
-8.38512540e-01 1.13239288e+00 -7.31867611e-01 -6.86531961e-02
3.30854088e-01 -8.59817445e-01 -6.42293751e-01 8.83074224e-01
4.66714233e-01 -4.64010030e-01 -3.24238092e-01 1.32177508e+00
1.72059953e-01 -3.47420499e-02 9.03653920e-01 -8.83951068e-01
-2.76368558e-01 1.51678264e-01 -1.17330170e+00 5.01818180e-01
6.41101971e-02 -3.50680947e-01 1.37136847e-01 -6.14511430e-01
7.17898667e-01 1.15551186e+00 4.88144755e-01 2.99260348e-01
-1.27555239e+00 -3.75219405e-01 1.91034656e-02 -6.72060400e-02
-1.52614224e+00 -4.78567600e-01 8.29609215e-01 -6.00345373e-01
5.18273950e-01 8.02416325e-01 5.46766400e-01 1.47773758e-01
3.14832002e-01 9.68167841e-01 7.66347051e-01 -6.29903018e-01
2.72865266e-01 2.29338750e-01 5.07957339e-01 6.55718267e-01
3.11148971e-01 -5.79845428e-01 2.35492319e-01 -2.71990031e-01
1.34608102e+00 -5.67884967e-02 -6.70432031e-01 -2.80117601e-01
-7.69023478e-01 4.58971024e-01 -3.11798424e-01 5.25804460e-01
-9.01953280e-02 8.30548033e-02 2.02015892e-01 -3.75567451e-02
2.63212323e-01 -3.89001131e-01 -1.81992397e-01 -7.75320530e-02
-1.70587802e+00 -1.55657038e-01 7.70048261e-01 1.34161413e+00
3.73064280e-01 4.78842109e-01 -1.38950720e-01 7.52153039e-01
3.34237039e-01 7.81898320e-01 3.88068706e-01 -9.16733146e-01
2.10158348e-01 5.73997915e-01 2.04653189e-01 -1.05358970e+00
2.98336804e-01 4.27078068e-01 -6.74332559e-01 5.65353096e-01
4.29386318e-01 -2.02448353e-01 -1.30564928e+00 6.61552131e-01
4.16250437e-01 3.21301609e-01 -1.76191777e-01 4.50622231e-01
9.63366687e-01 1.24853170e+00 -4.50271428e-01 -5.45499206e-01
9.59486008e-01 -8.37841868e-01 -1.18015850e+00 2.05113545e-01
5.78675158e-02 -1.20934582e+00 5.46492219e-01 7.09167123e-01
-1.46791053e+00 -7.01693177e-01 -9.57434177e-01 -5.86779462e-03
7.73810297e-02 3.68929982e-01 2.02698246e-01 1.04548979e+00
-1.04754090e+00 2.78356493e-01 -8.25593174e-01 -1.12190455e-01
3.88487987e-02 4.66287613e-01 1.57083526e-01 1.55926153e-01
-2.86706835e-01 3.30244124e-01 4.09021646e-01 3.99189219e-02
-3.33967179e-01 -1.05220422e-01 -4.45248336e-01 3.53798154e-03
3.15747887e-01 -9.33412537e-02 8.91662240e-01 -1.71688759e+00
-1.68108416e+00 9.23355520e-01 -5.37225008e-01 -9.24330428e-02
6.55127764e-01 7.74477348e-02 -5.08796930e-01 5.96795261e-01
-5.65853953e-01 1.06452545e-02 1.49872172e+00 -1.55006933e+00
-8.56648266e-01 2.23787744e-02 -5.86083770e-01 -7.95517955e-03
2.71870881e-01 3.60435277e-01 -1.16230428e+00 -7.67076731e-01
4.44691181e-01 -6.53855801e-01 5.74915744e-02 4.09575133e-03
-2.74975121e-01 4.15539622e-01 1.61700964e+00 -1.31952262e+00
1.54198229e+00 -2.33903956e+00 -4.38852608e-02 5.53476512e-01
1.29458904e-01 4.28298503e-01 4.07679886e-01 -1.01980567e-01
6.00434747e-03 5.82853593e-02 -3.66389990e-01 2.24919423e-01
-4.89956617e-01 8.51055384e-02 -1.39699787e-01 7.28959680e-01
-1.11413607e-02 3.13447088e-01 -2.61285484e-01 -9.55548346e-01
4.40134376e-01 6.24847233e-01 -1.91263318e-01 -2.20373441e-02
-1.25084206e-01 1.18429363e-01 -3.50466549e-01 8.56926143e-01
1.31011379e+00 7.92830735e-02 7.50592947e-02 -6.81575909e-02
-1.63015693e-01 -4.46179777e-01 -1.86884606e+00 5.81920981e-01
3.60717237e-01 1.09136057e+00 6.66685998e-01 -7.62354732e-01
9.30326700e-01 3.56023192e-01 5.66672325e-01 -1.38849780e-01
5.48670888e-01 2.12828573e-02 -3.61684151e-03 -3.93109351e-01
6.57071173e-01 2.28547916e-01 5.63613296e-01 -1.06746458e-01
-3.02863419e-01 -4.91019905e-01 2.28036702e-01 7.66690150e-02
6.39941573e-01 1.03853390e-01 2.90476501e-01 -4.88241404e-01
9.99845207e-01 5.75029515e-02 5.56043446e-01 4.07855064e-01
-3.03135961e-01 8.14496577e-01 4.20005441e-01 7.53785893e-02
-9.76397038e-01 -8.72945666e-01 -3.15600544e-01 5.35480559e-01
6.41122878e-01 -8.35248977e-02 -1.29454350e+00 -9.92976725e-02
-2.59116083e-01 6.59355462e-01 -2.31915385e-01 3.21690261e-01
-8.79362047e-01 -6.36636615e-01 2.19817296e-01 2.98678786e-01
7.33661056e-01 -7.03942120e-01 -3.85851383e-01 2.06400692e-01
-7.15048835e-02 -1.10876930e+00 -5.81702590e-01 2.05490381e-01
-1.17243421e+00 -1.07917094e+00 -1.03838515e+00 -1.13709950e+00
8.20434391e-01 6.73817754e-01 8.72260094e-01 3.45271021e-01
-4.46060628e-01 5.94152570e-01 -3.03130984e-01 -1.85100377e-01
-6.81443393e-01 -8.47596884e-01 -6.41859651e-01 3.77004176e-01
2.20092237e-01 5.60243875e-02 -3.00329030e-01 4.16666240e-01
-9.54975009e-01 2.67524123e-01 2.01402996e-02 3.81735981e-01
1.00560534e+00 7.84067571e-01 -6.17844939e-01 -9.17077303e-01
3.01005810e-01 -1.23009160e-01 -9.48619425e-01 3.08350146e-01
-2.70460639e-02 -4.45681602e-01 4.59239900e-01 -6.38680518e-01
-1.45908928e+00 4.54242602e-02 2.56571680e-01 -3.23643416e-01
-2.82455504e-01 -2.32417420e-01 -4.38141286e-01 -3.49008024e-01
2.19860181e-01 5.43409705e-01 -2.61603355e-01 -5.11891901e-01
2.44589359e-01 9.10367727e-01 7.42512047e-01 -3.49561244e-01
8.84479284e-01 8.35610867e-01 -6.83648735e-02 -1.74687076e+00
2.87633538e-01 -8.52843761e-01 -7.07906961e-01 -3.49164069e-01
1.23864603e+00 -6.29702330e-01 -1.12185441e-01 7.65467286e-01
-1.04952860e+00 -4.92970079e-01 -1.68111891e-01 2.68095434e-01
-4.32696581e-01 8.42088461e-01 -9.35288072e-01 -1.18533146e+00
-6.92113489e-02 -1.17650712e+00 8.89266372e-01 4.98130351e-01
9.64504015e-03 -1.07944250e+00 -2.78032303e-01 1.64745107e-01
9.91595834e-02 4.61060017e-01 1.04475784e+00 -2.87205517e-01
-7.90627241e-01 -2.66502500e-01 -1.88813344e-01 4.76036638e-01
3.56477648e-01 1.12655830e+00 -7.90372968e-01 -1.47154862e-02
5.68116248e-01 4.93807495e-01 6.87797666e-01 1.06010401e+00
9.44380879e-01 6.61896244e-02 -4.36445445e-01 9.54276741e-01
1.82878530e+00 1.01230848e+00 1.08604705e+00 1.86598331e-01
8.56528699e-01 2.97665983e-01 4.71313685e-01 3.34166378e-01
-3.77063394e-01 2.92559832e-01 3.05209607e-02 -5.79667151e-01
-2.40606859e-01 5.05528331e-01 3.47634017e-01 8.15197408e-01
-2.95108616e-01 -4.24897432e-01 -7.99153566e-01 2.74896294e-01
-1.58650148e+00 -1.21131992e+00 -1.08103156e+00 1.95512164e+00
7.42863476e-01 3.74241501e-01 1.08607009e-01 5.22679090e-01
1.17414629e+00 -1.53350919e-01 -2.26894021e-01 -7.10661411e-01
-3.38102877e-01 3.25645953e-01 7.25130737e-01 7.62645602e-01
-1.28916168e+00 9.94565070e-01 6.77937031e+00 1.16071212e+00
-1.22754991e+00 -2.80742705e-01 7.95455515e-01 2.55461514e-01
-4.61936966e-02 7.70413317e-03 -9.45048749e-01 6.18032753e-01
6.35631621e-01 -1.40201285e-01 3.10351878e-01 8.08738410e-01
2.74827123e-01 -6.34448588e-01 -6.94032729e-01 1.19481575e+00
9.41637158e-02 -1.14802396e+00 5.08509129e-02 -1.10674255e-01
1.08499014e+00 -6.80261731e-01 -7.97931179e-02 -1.62743315e-01
-6.12481609e-02 -7.29156971e-01 8.94012094e-01 3.76198232e-01
5.81728518e-01 -6.02978289e-01 4.87879664e-01 3.66086185e-01
-1.43931985e+00 2.67749220e-01 -6.11954153e-01 2.60715246e-01
2.77954340e-02 4.16044116e-01 -4.12956595e-01 8.38881135e-02
4.29748416e-01 2.54974335e-01 -4.81891900e-01 1.21907508e+00
4.33941573e-01 9.05361712e-01 -2.57717729e-01 1.15570612e-01
1.92904085e-01 -9.27946270e-01 4.36525553e-01 1.58528066e+00
2.13283524e-01 3.71005625e-01 1.33504987e-01 8.06473911e-01
3.55456382e-01 4.22385871e-01 -1.64485827e-01 -1.90909952e-01
2.87745912e-02 9.16065693e-01 -1.66633677e+00 -7.96240985e-01
-7.07480907e-01 1.27061069e+00 -7.94092119e-01 6.31693602e-01
-1.04708898e+00 -5.53617418e-01 1.26959398e-01 3.63697410e-01
7.87572742e-01 -2.40354717e-01 -6.14275515e-01 -1.08038414e+00
-2.58569241e-01 -1.02865195e+00 2.03922354e-02 -7.05083251e-01
-5.31748414e-01 5.21756530e-01 8.51964131e-02 -1.14254916e+00
2.57933587e-01 -7.47415185e-01 -7.61999488e-01 1.06444585e+00
-1.16260934e+00 -8.09181571e-01 -3.43976110e-01 9.80411112e-01
9.59009349e-01 -2.58713603e-01 2.89238155e-01 9.22066793e-02
-5.73260844e-01 4.54528332e-02 8.94670010e-01 1.44363046e-01
3.06682974e-01 -1.26901245e+00 2.22386211e-01 1.23805916e+00
-1.58285931e-01 4.98330444e-01 7.47840166e-01 -1.07965231e+00
-1.08366334e+00 -7.55398631e-01 3.48818153e-01 -2.39652887e-01
2.09416956e-01 -1.94467738e-01 -1.16967654e+00 4.85576987e-01
2.96196163e-01 -2.01452717e-01 5.72136223e-01 -8.10294509e-01
4.16557610e-01 3.12220931e-01 -1.27862370e+00 6.52896643e-01
2.82329321e-01 -6.89883903e-02 -5.88097513e-01 1.23526767e-01
1.63138628e-01 -4.63131100e-01 -4.16622609e-01 -7.94564083e-04
1.60427824e-01 -1.13407636e+00 9.45504665e-01 8.80485773e-02
1.65128767e-01 -6.51346922e-01 -2.92091370e-01 -2.43615970e-01
-4.71871704e-01 -8.64019454e-01 -2.10413471e-01 1.45428216e+00
-7.04236142e-03 1.98858567e-02 9.29599404e-01 7.14392602e-01
2.90335834e-01 -2.01153353e-01 -4.59890991e-01 -5.95981717e-01
-2.11599231e-01 -2.80580800e-02 9.56000686e-02 8.86156082e-01
-5.00345826e-01 -3.63782272e-02 -2.79681236e-01 -9.75584090e-02
6.58300757e-01 8.81797150e-02 8.59674394e-01 -1.20130908e+00
-3.39524060e-01 -5.32340884e-01 -3.66209865e-01 -1.01896214e+00
-3.25297326e-01 -3.83545250e-01 1.34951957e-02 -1.70755076e+00
1.12099953e-01 -1.78648278e-01 3.26603651e-01 -4.04905587e-01
-2.99949050e-01 4.03313227e-02 1.62367001e-01 2.92613119e-01
-1.72267094e-01 -1.05898447e-01 1.22759271e+00 -1.54871970e-01
-5.14912546e-01 2.17687368e-01 -1.40904889e-01 1.30055845e+00
4.49964315e-01 -2.55884051e-01 -5.95609307e-01 -7.74768144e-02
-3.92448187e-01 2.98313230e-01 -2.17565522e-01 -8.75029027e-01
-1.13309048e-01 -4.45366949e-01 8.18755448e-01 -9.45697188e-01
1.23185664e-01 -1.31531250e+00 4.11742508e-01 3.34670782e-01
1.25571281e-01 -2.54890859e-01 3.63162160e-01 4.90030497e-01
-2.17348889e-01 -7.82089710e-01 1.27683485e+00 -2.18953937e-01
-8.97201896e-01 -7.85003453e-02 -9.80954528e-01 6.58181682e-03
1.24444878e+00 -1.15819895e+00 9.81072411e-02 -3.75346363e-01
-7.19253123e-01 -3.58658820e-01 8.35524619e-01 -4.32355523e-01
6.55340314e-01 -1.06911314e+00 -4.38233078e-01 3.22599351e-01
-6.83045626e-01 -1.25757471e-01 3.22230235e-02 7.58260310e-01
-1.65202737e+00 1.46517009e-01 -7.21007809e-02 -7.18948126e-01
-1.95703423e+00 6.24462903e-01 2.90251642e-01 7.18354285e-02
-8.12583864e-01 8.81814837e-01 3.86578947e-01 5.95173657e-01
4.65341926e-01 -8.86036217e-01 -3.31665158e-01 -2.53275454e-01
7.25724936e-01 8.10352862e-01 -3.12675416e-01 -9.37238693e-01
-1.92806125e-01 9.67815995e-01 1.68962717e-01 -1.03032880e-01
7.72469521e-01 -2.66454548e-01 -2.60512561e-01 5.60610235e-01
1.04828835e+00 6.34114206e-01 -1.20757842e+00 3.46043929e-02
6.88754302e-03 -8.02273154e-01 2.42358893e-01 -2.28525937e-01
-1.04314601e+00 8.83326948e-01 6.80164754e-01 5.61446548e-01
1.38805890e+00 -5.98977685e-01 8.13634932e-01 3.10622063e-02
1.79313757e-02 -1.46302295e+00 -1.05511598e-01 3.13708782e-01
5.65504432e-01 -8.03250134e-01 2.41376862e-01 -8.87869120e-01
-7.31844366e-01 1.55841982e+00 3.61601830e-01 -4.58159536e-01
7.40800381e-01 8.47354531e-01 6.53796420e-02 2.54997432e-01
-2.15265661e-01 -9.01990607e-02 3.94277364e-01 7.07297981e-01
6.37770891e-01 -1.69641614e-01 -3.62425774e-01 1.63309574e-01
3.32705289e-01 -1.67154610e-01 8.16192627e-01 1.22252464e+00
-9.69333470e-01 -6.73887908e-01 -1.10065973e+00 3.62986803e-01
-9.57020581e-01 -5.61279692e-02 -6.60780966e-01 9.45964277e-01
2.38139555e-01 8.99676025e-01 2.28615731e-01 7.88366571e-02
7.80420080e-02 6.89000338e-02 6.13210917e-01 -2.67782688e-01
-5.26787758e-01 1.28421760e+00 -1.86779514e-01 -2.22921297e-01
-3.88206989e-01 -6.47128403e-01 -1.44458640e+00 -3.94474268e-01
-6.86446249e-01 -3.26587074e-02 5.10341465e-01 4.67311054e-01
-5.39995611e-01 6.52855515e-01 3.18588823e-01 -8.24246883e-01
1.04008012e-01 -4.80309814e-01 -9.78975892e-01 4.25104082e-01
3.10721904e-01 -2.12931886e-01 -3.86239260e-01 1.11489606e+00]
|
[8.974955558776855, -0.8172535300254822]
|
af0740b4-11e3-45ad-b7f9-33c47d6aea70
|
conveying-the-predicted-future-to-users-a
|
2302.09122
| null |
https://arxiv.org/abs/2302.09122v1
|
https://arxiv.org/pdf/2302.09122v1.pdf
|
Conveying the Predicted Future to Users: A Case Study of Story Plot Prediction
|
Creative writing is hard: Novelists struggle with writer's block daily. While automatic story generation has advanced recently, it is treated as a "toy task" for advancing artificial intelligence rather than helping people. In this paper, we create a system that produces a short description that narrates a predicted plot using existing story generation approaches. Our goal is to assist writers in crafting a consistent and compelling story arc. We conducted experiments on Amazon Mechanical Turk (AMT) to examine the quality of the generated story plots in terms of consistency and storiability. The results show that short descriptions produced by our frame-enhanced GPT-2 (FGPT-2) were rated as the most consistent and storiable among all models; FGPT-2's outputs even beat some random story snippets written by humans. Next, we conducted a preliminary user study using a story continuation task where AMT workers were given access to machine-generated story plots and asked to write a follow-up story. FGPT-2 could positively affect the writing process, though people favor other baselines more. Our study shed some light on the possibilities of future creative writing support systems beyond the scope of completing sentences. Our code is available at: https://github.com/appleternity/Story-Plot-Generation.
|
["Ting-Hao 'Kenneth' Huang", 'Kavya Laalasa Karanam', 'Saniya Naphade', 'Chieh-Yang Huang']
|
2023-02-17
| null | null | null | null |
['story-continuation', 'story-generation']
|
['computer-vision', 'natural-language-processing']
|
[ 1.10346138e-01 5.06849647e-01 -1.48101896e-03 -3.36327642e-01
-8.65100145e-01 -8.58653069e-01 1.05933809e+00 -1.65658891e-01
8.68456215e-02 8.57224822e-01 8.44784617e-01 -1.14768185e-01
1.62773624e-01 -5.42920649e-01 -4.36048031e-01 1.66391637e-02
6.25571668e-01 6.89763188e-01 -6.31467327e-02 -4.88250285e-01
7.08575964e-01 -2.07084231e-02 -1.27901411e+00 8.25197935e-01
7.99421430e-01 1.93077967e-01 4.67708081e-01 9.94160831e-01
-2.36835092e-01 1.27337837e+00 -1.02297914e+00 -8.38950932e-01
5.13751693e-02 -9.01585460e-01 -8.54701519e-01 3.90786864e-02
5.10928571e-01 -4.85584825e-01 -2.28742301e-01 4.79492515e-01
5.65446019e-01 1.12134807e-01 5.72933316e-01 -1.37029934e+00
-1.24283230e+00 1.22888279e+00 -3.54537129e-01 3.08132879e-02
1.02890134e+00 5.47266543e-01 1.07710469e+00 -9.95172441e-01
1.19736433e+00 1.24910438e+00 7.08851755e-01 8.08056295e-01
-1.34271419e+00 -7.28446126e-01 -7.31565729e-02 -5.19647216e-03
-1.12213004e+00 -7.38870382e-01 7.19181359e-01 -7.02049911e-01
1.08239317e+00 4.94371265e-01 8.71844649e-01 1.56380391e+00
2.87220441e-02 8.97933364e-01 1.16424239e+00 -4.35586870e-01
6.60988614e-02 3.52516711e-01 6.22683624e-03 5.47942460e-01
-9.82050970e-02 -2.73093015e-01 -1.29199219e+00 -1.31675303e-01
7.60614991e-01 -6.32370532e-01 -3.55145752e-01 3.82356077e-01
-1.47918952e+00 8.87873650e-01 6.00048192e-02 4.90162164e-01
-2.78776646e-01 3.01467597e-01 2.88513482e-01 2.11006314e-01
5.59892714e-01 1.17185056e+00 2.59916782e-01 -7.99434960e-01
-1.16440177e+00 1.14715123e+00 1.01550114e+00 1.14868724e+00
-1.96118057e-01 -1.18282430e-01 -9.24925745e-01 8.98914874e-01
-2.49532722e-02 4.42292660e-01 5.97842872e-01 -1.04247725e+00
6.09723032e-01 3.65469098e-01 5.38294137e-01 -1.11472225e+00
-1.77976072e-01 -1.91259533e-01 -2.79136330e-01 2.37259418e-01
6.05597794e-01 -2.53018349e-01 -3.55747372e-01 1.49478269e+00
-3.35073411e-01 -4.48100924e-01 -1.98202819e-01 9.75463390e-01
1.04660618e+00 8.50874543e-01 -2.24365115e-01 -2.25268736e-01
1.36113894e+00 -1.12216198e+00 -1.01452112e+00 -2.53470570e-01
4.32113022e-01 -1.13464725e+00 1.75147259e+00 4.74906147e-01
-1.51168561e+00 -5.84433079e-01 -1.07553613e+00 -4.28712398e-01
9.53554362e-02 2.93531626e-01 5.25220454e-01 4.20436084e-01
-1.12831926e+00 7.26224065e-01 -5.56999445e-01 -6.16670787e-01
5.05299389e-01 -4.22092646e-01 -1.76974237e-01 2.12907389e-01
-8.17156851e-01 1.05239677e+00 1.92582607e-01 -3.72280657e-01
-4.34955060e-01 -6.62182570e-01 -5.49028575e-01 -1.93893611e-01
1.75280601e-01 -8.67460907e-01 2.06637955e+00 -7.20560431e-01
-1.47043407e+00 7.54872143e-01 -3.19547266e-01 -3.71574312e-01
1.07528317e+00 -4.92650151e-01 -8.11050981e-02 -4.32259776e-02
5.92014849e-01 8.96842897e-01 5.71060538e-01 -1.13053858e+00
-3.69865835e-01 2.45803326e-01 -2.77079940e-02 3.61206740e-01
-1.40611976e-01 2.69795060e-01 1.52675584e-01 -1.05266201e+00
-2.05859870e-01 -7.68277884e-01 2.30735153e-01 -7.51649365e-02
-7.07116544e-01 -4.69926953e-01 6.37804925e-01 -8.84846628e-01
1.42169285e+00 -1.79159713e+00 -1.02942199e-01 -2.79894143e-01
2.38360971e-01 -2.01845497e-01 4.47063483e-02 9.89856482e-01
3.44379932e-01 5.85088134e-01 -2.66849454e-02 -7.13181138e-01
2.38920927e-01 -3.29232633e-01 -5.93347073e-01 -1.93477988e-01
9.59339514e-02 1.13561189e+00 -1.04957008e+00 -5.11798680e-01
-1.47520006e-01 2.85462290e-01 -2.67252713e-01 1.06386147e-01
-6.66871548e-01 2.48482659e-01 -2.03294456e-01 3.48130643e-01
1.10250756e-01 -4.52935994e-01 -1.70233801e-01 4.17709142e-01
-3.32382172e-01 6.15369618e-01 -7.11665630e-01 1.87136829e+00
-1.32634267e-01 1.24121225e+00 -6.65675879e-01 3.33732635e-01
9.61381853e-01 3.90061826e-01 -2.69098163e-01 -5.68955779e-01
-2.03581769e-02 7.29760900e-02 -1.48803949e-01 -7.61776447e-01
1.18647599e+00 -3.34692448e-01 -1.49431601e-01 1.27436757e+00
-4.20260310e-01 -7.44192123e-01 4.55580294e-01 6.38604701e-01
1.17263627e+00 3.91661584e-01 2.02771246e-01 9.58831385e-02
-3.26577604e-01 5.49341738e-01 -1.02459617e-01 1.21463370e+00
9.16604400e-02 9.80074167e-01 6.11784756e-01 -2.87187755e-01
-1.43930590e+00 -1.04059446e+00 2.89041370e-01 9.08671916e-01
-1.68162927e-01 -9.52152193e-01 -7.34188914e-01 -4.24324632e-01
-2.61150360e-01 1.58548105e+00 -3.38412315e-01 2.53505498e-01
-5.22426069e-01 -1.30198240e-01 8.21524560e-01 5.12780845e-01
5.26967108e-01 -1.46672726e+00 -9.78069782e-01 2.90567935e-01
-7.23325849e-01 -7.86362529e-01 -7.84863234e-01 -4.75993901e-01
-3.34992051e-01 -6.04249358e-01 -7.65439451e-01 -5.16166508e-01
3.52361858e-01 2.76036173e-01 9.86905217e-01 9.32801664e-02
1.03970766e-02 4.02139910e-02 -5.87905288e-01 -6.61368728e-01
-7.92780757e-01 6.92863762e-02 -1.11130670e-01 -6.17250979e-01
-1.39472438e-02 -5.62970519e-01 -2.08158195e-01 2.86081359e-02
-5.78367949e-01 1.24669003e+00 1.49366692e-01 7.44076073e-01
-6.89485148e-02 -5.85062169e-02 5.61056376e-01 -9.84955728e-01
1.57599306e+00 -1.67289361e-01 1.11939155e-01 1.65803507e-01
-5.40490091e-01 -7.60471821e-02 5.32238722e-01 -6.33296609e-01
-1.32972527e+00 -1.70296565e-01 1.04440033e-01 2.13146745e-03
1.83141857e-01 3.72264355e-01 3.46451223e-01 6.16651356e-01
1.06182432e+00 1.10723868e-01 -4.77445051e-02 -2.65778393e-01
5.71331918e-01 6.71853483e-01 7.88783669e-01 -5.93475103e-01
9.57733631e-01 1.84759106e-02 -7.56515145e-01 -5.03842711e-01
-6.25119150e-01 1.66289493e-01 -1.60729110e-01 -6.59843624e-01
6.26045346e-01 -6.76646113e-01 -6.22609556e-01 3.13818038e-01
-1.55543220e+00 -7.44745433e-01 -4.85642105e-01 1.40196055e-01
-7.14257479e-01 -1.34263724e-01 -7.18936563e-01 -9.42573249e-01
-4.79278684e-01 -7.36548305e-01 9.32122052e-01 4.76346940e-01
-1.50660932e+00 -4.86436576e-01 2.24882871e-01 7.11203992e-01
3.70769918e-01 3.30618322e-01 7.58995175e-01 -5.02146780e-01
-3.26827645e-01 -1.67468980e-01 -2.68445909e-01 -3.30793530e-01
-5.75431883e-02 2.66813755e-01 -9.89177823e-01 3.19862336e-01
-2.73052126e-01 -5.06342351e-01 4.48460013e-01 1.41749889e-01
5.90430737e-01 -8.37595999e-01 -2.39455149e-01 9.45495293e-02
7.92689264e-01 2.28213131e-01 4.62025464e-01 3.46644014e-01
4.93984282e-01 5.60777307e-01 6.49825633e-01 7.07938254e-01
5.05907953e-01 7.51358986e-01 -3.09971303e-01 3.20995390e-01
-3.79522383e-01 -9.92818475e-01 5.15572846e-01 3.48167539e-01
-1.55003399e-01 -8.20877194e-01 -9.83366013e-01 4.10127997e-01
-2.07702613e+00 -1.44011176e+00 -3.52197081e-01 1.54952312e+00
9.57921267e-01 4.22697723e-01 3.97836477e-01 7.25848526e-02
6.24122143e-01 2.43947878e-01 -2.51926184e-01 -6.21653020e-01
-2.30555028e-01 1.76805496e-01 -1.59987226e-01 6.17569685e-01
-5.04020631e-01 1.02645385e+00 6.08375549e+00 6.08425438e-01
-8.40214550e-01 1.53036237e-01 5.81700087e-01 -5.77302337e-01
-5.50196111e-01 1.58386976e-02 -5.22769511e-01 5.42606056e-01
6.37568116e-01 -7.60356605e-01 4.80851978e-01 7.20970094e-01
8.18559706e-01 -2.67592251e-01 -1.24320817e+00 9.67173159e-01
3.48888695e-01 -1.62965536e+00 9.42132249e-02 -2.15929762e-01
8.18361402e-01 -6.00366354e-01 7.06075281e-02 2.38316640e-01
4.72202182e-01 -1.25126183e+00 1.55102968e+00 7.74405539e-01
8.31350803e-01 -4.18583393e-01 4.40934300e-01 5.20923615e-01
-5.83587945e-01 2.28938729e-01 8.43906850e-02 -7.19465792e-01
6.76233649e-01 1.81954756e-01 -1.57266903e+00 -5.19862771e-02
3.42057973e-01 5.61477184e-01 -8.84338081e-01 7.30215073e-01
-6.38526797e-01 5.81659377e-01 -1.83708295e-02 -5.75834811e-01
-1.57609656e-01 2.76976228e-01 7.26588130e-01 1.28857362e+00
5.26722372e-01 4.61388171e-01 -1.87457502e-02 1.40757048e+00
-2.89220572e-01 1.29636437e-01 -6.80876851e-01 -8.00423265e-01
6.94931746e-01 1.01113915e+00 -9.94274020e-01 -4.42427158e-01
2.30338782e-01 1.25951731e+00 2.27958351e-01 1.41924888e-01
-8.17561090e-01 -2.99613923e-01 -1.72386393e-02 6.40874565e-01
-2.79631346e-01 -1.54228732e-01 -1.03544211e+00 -9.86361682e-01
2.65485048e-01 -8.75732839e-01 -1.19839564e-01 -1.80191624e+00
-1.28054917e+00 7.00177073e-01 1.09371161e-02 -9.05066252e-01
-5.37200093e-01 7.97194615e-02 -9.92721140e-01 8.77988935e-01
-3.40792030e-01 -1.19667232e+00 -4.89869863e-01 -7.54719153e-02
1.07228243e+00 -1.29328921e-01 7.51689017e-01 -4.12680566e-01
-1.58020571e-01 4.41105992e-01 -4.52247977e-01 -2.02730913e-02
8.96348834e-01 -1.25977647e+00 1.01423109e+00 8.12110305e-01
3.89721662e-01 6.50301874e-01 1.30314183e+00 -1.22259653e+00
-6.82965040e-01 -6.17277801e-01 1.46612394e+00 -8.32583666e-01
6.88457489e-01 -4.34441239e-01 -5.82438111e-01 6.70906723e-01
7.70260632e-01 -9.10514832e-01 5.45731425e-01 4.68656421e-02
-2.67529309e-01 6.81325436e-01 -1.01824212e+00 9.48872983e-01
1.36338317e+00 -5.74069858e-01 -8.33888829e-01 6.11617744e-01
7.91718304e-01 -4.80810016e-01 -3.90923381e-01 -3.41011405e-01
9.10678744e-01 -8.79642785e-01 3.81181568e-01 -3.88372451e-01
1.40516603e+00 -2.66501546e-01 2.87599236e-01 -1.36532748e+00
-3.53966743e-01 -1.12863493e+00 8.94902870e-02 1.54526055e+00
5.97324073e-01 -1.20273292e-01 5.97246826e-01 9.28479493e-01
-3.20377499e-01 -4.57894593e-01 -3.59020978e-01 -6.33622408e-01
-2.51848370e-01 -5.50712168e-01 4.37666714e-01 9.24321890e-01
6.03331506e-01 6.83434129e-01 -4.34423894e-01 -5.32091439e-01
4.28410470e-02 -1.23167843e-01 1.02219772e+00 -8.99325550e-01
-5.25599420e-01 -8.23324680e-01 1.70958832e-01 -6.05168879e-01
-1.93474293e-01 -1.04886258e+00 2.30676323e-01 -1.97825444e+00
4.99676347e-01 -3.29876691e-02 6.83430016e-01 7.03557432e-01
-1.15560040e-01 3.71358097e-01 7.91981220e-01 4.51279551e-01
-4.83744174e-01 3.03842723e-01 1.35412729e+00 -8.58004242e-02
-6.48838520e-01 -3.82930674e-02 -1.12134171e+00 4.73287314e-01
8.22860301e-01 -2.86698848e-01 -6.33337140e-01 -2.29852974e-01
6.14353478e-01 2.75028884e-01 4.92527515e-01 -1.03480160e+00
2.46807516e-01 -2.24253282e-01 5.26507974e-01 -6.40446186e-01
3.09453070e-01 1.03009142e-01 6.28076434e-01 1.05072372e-01
-8.49960625e-01 3.57821375e-01 6.73054680e-02 -1.08322769e-01
2.75597960e-01 -3.01767558e-01 3.38358194e-01 -2.31643081e-01
-1.60416409e-01 -4.84760314e-01 -6.66152775e-01 -7.58060627e-03
9.04026389e-01 -5.09060681e-01 -7.96559572e-01 -1.04248655e+00
-5.64444304e-01 1.05423145e-01 7.45045006e-01 5.80759168e-01
7.89402187e-01 -1.37624049e+00 -1.08182955e+00 -2.75956631e-01
2.72729039e-01 -1.14363045e-01 -7.09386244e-02 3.00080687e-01
-7.58291662e-01 2.90320188e-01 -2.47141212e-01 -3.97446901e-02
-1.32238722e+00 1.53589323e-01 -1.38710812e-01 -1.26883134e-01
-1.04299533e+00 1.10166419e+00 -1.59777820e-01 1.71442911e-01
6.98362365e-02 -2.25086719e-01 2.23853774e-02 9.95579660e-02
9.51239347e-01 4.30293381e-01 -4.52131122e-01 -3.51170838e-01
1.10351250e-01 -1.34529620e-01 -3.06950450e-01 -9.55471218e-01
1.28987765e+00 -8.61059204e-02 2.10398629e-01 8.40099812e-01
3.96153659e-01 1.91522211e-01 -1.06331551e+00 2.19842240e-01
5.45946285e-02 -5.47760129e-01 -5.09763837e-01 -1.42482781e+00
-1.87160999e-01 4.50083911e-01 -1.09566011e-01 3.79002273e-01
5.47497690e-01 1.63498580e-01 9.74578261e-01 3.95492405e-01
3.18188459e-01 -1.01433384e+00 4.26061809e-01 3.76819044e-01
1.61709559e+00 -9.86186743e-01 9.22152176e-02 -1.19739592e-01
-1.23420703e+00 1.21466970e+00 6.90697193e-01 6.63127378e-02
-2.35376865e-01 2.37881854e-01 -5.46179935e-02 -3.15791249e-01
-1.19297004e+00 2.76173770e-01 1.86606258e-01 5.55273235e-01
9.77239311e-01 2.85552919e-01 -5.24521589e-01 1.00205290e+00
-1.35595250e+00 3.21887940e-01 9.73634183e-01 8.06393981e-01
-4.04038280e-01 -8.03097129e-01 -4.84264374e-01 5.91093600e-01
-3.57661471e-02 -7.07319081e-02 -1.21548247e+00 7.34427214e-01
-3.56716514e-02 1.13971543e+00 -1.44503772e-01 -4.89739686e-01
2.86352217e-01 2.57753968e-01 5.75624287e-01 -8.69159818e-01
-8.51449072e-01 -2.24237978e-01 7.25348055e-01 -2.40541905e-01
7.26354197e-02 -9.08931017e-01 -1.29513407e+00 -8.34894240e-01
1.16405096e-02 5.65763973e-02 4.37619478e-01 8.30540657e-01
1.28904015e-01 4.05903220e-01 1.29991129e-01 -7.83774555e-01
-5.98720014e-02 -1.37663507e+00 -1.72992468e-01 3.89872670e-01
-8.77321959e-02 -2.86807269e-01 4.84337993e-02 4.37888891e-01]
|
[11.750041961669922, 8.773791313171387]
|
96aa59ec-a3e0-4d2b-aaaf-c75e1d954a78
|
cardiac-mr-image-segmentation-techniques-an
|
1502.04252
| null |
http://arxiv.org/abs/1502.04252v1
|
http://arxiv.org/pdf/1502.04252v1.pdf
|
Cardiac MR Image Segmentation Techniques: an overview
|
Broadly speaking, the objective in cardiac image segmentation is to delineate
the outer and inner walls of the heart to segment out either the entire or
parts of the organ boundaries. This paper will focus on MR images as they are
the most widely used in cardiac segmentation -- as a result of the accurate
morphological information and better soft tissue contrast they provide. This
cardiac segmentation information is very useful as it eases physical
measurements that provides useful metrics for cardiac diagnosis such as
infracted volumes, ventricular volumes, ejection fraction, myocardial mass,
cardiac movement, and the like. But, this task is difficult due to the
intensity and texture similarities amongst the different cardiac and background
structures on top of some noisy artifacts present in MR images. Thus far,
various researchers have proposed different techniques to solve some of the
pressing issues. This seminar paper presents an overview of representative
medical image segmentation techniques. The paper also highlights preferred
approaches for segmentation of the four cardiac chambers: the left ventricle
(LV), right ventricle (RV), left atrium (LA) and right atrium (RA), on short
axis image planes.
|
['Tizita Nesibu Shewaye']
|
2015-02-14
| null | null | null | null |
['cardiac-segmentation']
|
['medical']
|
[ 5.89427948e-02 5.07007986e-02 -3.27187926e-02 -3.12265046e-02
4.26870845e-02 -6.46111965e-01 -6.70163473e-03 3.46790791e-01
-1.84971124e-01 7.99819648e-01 -4.27695699e-02 -2.93307453e-01
4.00246643e-02 -4.55439806e-01 2.34932810e-01 -8.82366776e-01
-1.30459264e-01 7.12421000e-01 3.49742949e-01 2.74489492e-01
2.95841485e-01 7.71189392e-01 -9.73235190e-01 -2.51775861e-01
7.77906299e-01 6.66044176e-01 9.32659805e-02 9.39601123e-01
-1.78586602e-01 6.91715062e-01 -4.65566695e-01 -9.65102911e-02
1.16548397e-01 -9.31738198e-01 -9.38916206e-01 4.43146706e-01
-1.17634848e-01 -3.46946687e-01 2.26877034e-01 8.73335600e-01
5.05681157e-01 -1.79618388e-01 7.42779672e-01 -6.84028804e-01
2.06334993e-01 4.72637266e-01 -7.12074935e-01 6.30448878e-01
-1.51618823e-01 -2.29052991e-01 3.95286113e-01 -6.46507323e-01
6.96739554e-01 5.90367734e-01 5.97220600e-01 4.20449585e-01
-1.08917654e+00 -2.33878776e-01 -2.46562511e-01 -4.14238781e-01
-1.09558630e+00 -6.00057095e-02 5.91315448e-01 -7.79598296e-01
5.25339305e-01 5.15191495e-01 8.98077071e-01 -4.20369506e-02
5.10623932e-01 4.37354445e-01 1.22972119e+00 -5.53184867e-01
8.11398104e-02 1.47766352e-01 3.41191322e-01 7.25768566e-01
4.18743521e-01 -1.63014278e-01 1.60077691e-01 -1.75134078e-01
1.09366977e+00 1.48143843e-01 -4.52497602e-01 -5.83513081e-01
-1.32466102e+00 4.30282831e-01 -8.09476525e-02 9.17114139e-01
-2.77435303e-01 -1.15344845e-01 6.88352883e-01 -9.14185494e-02
1.26718968e-01 1.90820217e-01 -3.27236891e-01 2.62065344e-02
-1.17328882e+00 -8.48739594e-03 6.68154061e-01 3.02692294e-01
1.70791730e-01 1.39149189e-01 -2.80138776e-02 8.67358148e-01
5.40060401e-01 3.92711669e-01 4.44400042e-01 -8.58908057e-01
3.31269920e-01 4.95635420e-01 -2.08260879e-01 -7.34972894e-01
-5.04776239e-01 -4.67326194e-01 -1.15752828e+00 3.71859580e-01
5.73718071e-01 -2.29277864e-01 -9.83690500e-01 1.08718574e+00
5.70014477e-01 -2.93896616e-01 -1.97619438e-01 1.10660005e+00
9.85276818e-01 3.68669391e-01 3.08306843e-01 -4.78727788e-01
1.77303731e+00 -6.65620923e-01 -8.28453839e-01 -1.90092221e-01
5.87427020e-01 -9.11508083e-01 4.07822698e-01 5.49379885e-02
-1.25083244e+00 -4.89745617e-01 -1.05613720e+00 4.21168566e-01
2.34331880e-02 2.25820288e-01 4.07462746e-01 8.32204998e-01
-7.78768003e-01 6.35802746e-01 -9.70610261e-01 -1.31042391e-01
2.41040796e-01 2.59580165e-01 -3.39745015e-01 2.48403117e-01
-6.20026112e-01 9.06059325e-01 4.20581698e-01 2.41597354e-01
-1.96015731e-01 -5.26608169e-01 -6.42109811e-01 -2.41307318e-01
1.86620444e-01 -8.74770880e-01 8.31919551e-01 -7.26974547e-01
-1.24715781e+00 1.40537345e+00 1.08477794e-01 -1.44691035e-01
5.39609373e-01 3.39518219e-01 7.93246925e-03 5.37810445e-01
4.34936285e-02 3.40764701e-01 6.92395210e-01 -1.29635072e+00
-4.62862343e-01 -6.81010783e-01 -3.09319645e-01 2.86285192e-01
6.94113910e-01 1.74513832e-01 -3.71056348e-01 -9.10559118e-01
7.47135878e-01 -9.92936254e-01 -3.69639993e-01 -2.99211949e-01
-4.48394895e-01 3.82721782e-01 7.72640467e-01 -1.06364131e+00
1.37039423e+00 -2.22482514e+00 1.01652898e-01 4.16846126e-01
6.34716988e-01 4.50347215e-01 7.90735960e-01 -8.99391342e-03
-1.68449402e-01 3.41646969e-01 -6.20861292e-01 1.45416811e-01
-5.75527966e-01 3.48031610e-01 1.72198772e-01 6.71696424e-01
-2.65991539e-01 7.55060077e-01 -5.39921165e-01 -1.02617836e+00
4.17452574e-01 4.23370928e-01 1.47470087e-01 -1.25140503e-01
4.44642305e-01 1.10792923e+00 -6.11682355e-01 6.24689817e-01
7.56266534e-01 -9.71596912e-02 3.82463157e-01 1.74916416e-01
-2.24871799e-01 -1.59730673e-01 -1.24280477e+00 1.16666424e+00
1.91303287e-02 3.57011139e-01 6.22305512e-01 -1.13167560e+00
8.47449958e-01 6.92248285e-01 9.48160946e-01 -5.76216340e-01
2.85636634e-01 3.54144663e-01 2.97408849e-01 -4.28639740e-01
-5.32202376e-03 -7.01211035e-01 3.12364399e-01 5.39042234e-01
-3.83472800e-01 -3.74004096e-01 3.50343049e-01 8.54814276e-02
4.42505032e-01 -7.65901655e-02 6.10007107e-01 -5.16474605e-01
9.97172952e-01 -6.65579960e-02 7.97569096e-01 2.88394541e-01
-5.08343637e-01 8.27450395e-01 5.81468523e-01 -6.98503613e-01
-9.92817700e-01 -1.03621268e+00 -5.92032015e-01 1.28871128e-01
1.96341395e-01 5.38388640e-02 -9.10370827e-01 -4.14533854e-01
-2.48769104e-01 -3.48218456e-02 -2.25794226e-01 4.81642723e-01
-9.18080389e-01 -1.03464425e+00 2.76128322e-01 3.96320790e-01
4.56450373e-01 -1.07166338e+00 -1.31445336e+00 4.50976789e-01
-2.30156854e-01 -9.72864866e-01 -2.17264354e-01 -7.19687622e-03
-1.74891806e+00 -1.09690309e+00 -1.31401992e+00 -9.11939800e-01
8.82214546e-01 -1.54966965e-01 1.29533625e+00 7.15063870e-01
-7.66169012e-01 1.29582286e-01 -1.17122404e-01 -3.90011102e-01
-6.28306091e-01 -3.58771533e-01 -2.95476377e-01 -3.29933524e-01
-3.70561123e-01 -4.14507329e-01 -6.58385813e-01 3.97129834e-01
-6.07210279e-01 5.75927235e-02 2.33275637e-01 5.91559708e-01
9.18801069e-01 7.43691698e-02 2.15783492e-01 -1.10064304e+00
3.48051071e-01 6.90080896e-02 -4.92031008e-01 1.96937874e-01
-5.86235523e-01 -3.78499836e-01 3.71688485e-01 1.76182821e-01
-8.57696593e-01 -1.10705964e-01 7.57766291e-02 3.79448757e-02
-3.91442597e-01 1.60451010e-01 3.12710404e-01 -6.64374009e-02
2.86653101e-01 2.01905668e-01 9.20597464e-02 -2.03010842e-01
-8.67487937e-02 4.29619730e-01 5.87690949e-01 -3.77103835e-01
3.45845252e-01 5.94801128e-01 4.23113883e-01 -1.02023637e+00
-4.51072246e-01 -6.98839247e-01 -1.24330974e+00 -3.96214187e-01
1.21937370e+00 -2.58377880e-01 -3.74711514e-01 2.61495054e-01
-8.93157005e-01 -8.56379569e-02 -5.07183433e-01 7.08917320e-01
-3.45963478e-01 8.61623526e-01 -8.73717189e-01 -8.49333465e-01
-7.36485720e-01 -1.35341001e+00 5.95993102e-01 4.40239340e-01
-2.03147858e-01 -1.49560487e+00 1.10949734e-02 4.38115656e-01
3.73727590e-01 8.86122346e-01 1.12343061e+00 -2.44652107e-01
-3.37405354e-01 -1.15366146e-01 -5.13632298e-02 5.38296402e-01
8.37823376e-02 2.06851631e-01 -5.06522477e-01 -6.95582032e-02
2.66824722e-01 2.50145078e-01 6.64296031e-01 1.24498367e+00
7.32989788e-01 3.13922077e-01 -4.16340768e-01 5.25088906e-01
1.38714349e+00 6.65586233e-01 6.77928746e-01 5.62040061e-02
4.14547890e-01 6.31442547e-01 7.33892143e-01 4.00842190e-01
-7.80829713e-02 2.77518094e-01 2.16377884e-01 -7.31627584e-01
-2.21413702e-01 5.89180887e-01 -2.36422315e-01 1.11364114e+00
-5.50073683e-01 1.99146256e-01 -9.55383599e-01 4.40867871e-01
-1.32510054e+00 -8.94885778e-01 -5.76377034e-01 2.32455564e+00
5.54459751e-01 5.79930693e-02 2.84770340e-01 4.42800373e-01
8.09684038e-01 -1.50695473e-01 -1.42798483e-01 -5.63586831e-01
9.55785252e-03 3.76651496e-01 4.73899871e-01 4.33528721e-01
-1.20208299e+00 1.87954947e-01 7.02022982e+00 2.28060842e-01
-1.19036126e+00 -7.45129362e-02 1.08019567e+00 4.48311448e-01
1.76232994e-01 1.54367954e-01 -6.86379194e-01 4.21067774e-01
3.43343377e-01 2.38241240e-01 -3.24577577e-02 3.52249503e-01
1.96727172e-01 -7.97390461e-01 -7.56199896e-01 9.61866140e-01
-1.14230387e-01 -1.20206773e+00 -4.70681310e-01 1.52199328e-01
4.25814629e-01 -3.62000525e-01 -3.07448238e-01 -1.55935690e-01
-7.26672590e-01 -9.58568394e-01 4.92525697e-01 4.50608701e-01
8.87251496e-01 -4.99835521e-01 1.02521443e+00 3.76455218e-01
-1.26752841e+00 3.29785496e-01 -1.97609037e-01 1.54587865e-01
2.94547349e-01 7.78115690e-01 -8.34007323e-01 4.40803915e-01
4.33664203e-01 2.26387441e-01 -1.99000806e-01 1.21890581e+00
7.38846362e-02 5.58643758e-01 -2.28158653e-01 3.93253654e-01
1.10715091e-01 -8.38254213e-01 6.13410711e-01 1.08841121e+00
-5.79747930e-02 4.33347791e-01 3.84613603e-01 8.52927446e-01
2.95397133e-01 5.79554319e-01 -2.71850348e-01 1.32544354e-01
4.42522578e-02 1.41713321e+00 -1.80298603e+00 -5.82907856e-01
-2.80950159e-01 4.49375033e-01 -4.53791022e-01 1.76898167e-01
-6.33392215e-01 -2.11340994e-01 -6.39098287e-02 4.87360567e-01
-1.93651214e-01 -5.29297665e-02 -6.16535962e-01 -8.63018513e-01
-9.19729024e-02 -7.26738334e-01 6.81322277e-01 -1.80186436e-01
-5.69736123e-01 3.63586307e-01 3.09012700e-02 -1.01184320e+00
-1.49879336e-01 -3.47865492e-01 -6.53102815e-01 1.24921227e+00
-1.11284685e+00 -7.49726713e-01 -1.85910031e-01 1.10266827e-01
4.54783857e-01 9.96405408e-02 7.13412881e-01 3.38720322e-01
-4.85608310e-01 -6.89787716e-02 -1.10233776e-01 5.32689214e-01
2.37118185e-01 -1.52188492e+00 -6.96514025e-02 7.25212634e-01
-2.31724709e-01 5.70371568e-01 4.94048476e-01 -6.49463654e-01
-6.08239651e-01 -3.90175760e-01 7.01426744e-01 -2.32677795e-02
-3.41945663e-02 3.14029634e-01 -7.09633768e-01 4.32983965e-01
8.12946334e-02 2.63186753e-01 8.43311489e-01 -4.76659566e-01
6.17253184e-01 2.08897635e-01 -1.20449185e+00 2.13290200e-01
1.41265899e-01 1.22903161e-01 -4.82973248e-01 1.05055906e-01
-1.99086413e-01 -8.21279168e-01 -1.21650398e+00 4.24317092e-01
6.20871067e-01 -1.27664566e+00 1.12164354e+00 -1.52823597e-01
1.19109541e-01 -4.66149241e-01 5.58628559e-01 -6.90869808e-01
-1.40332952e-01 -2.81871945e-01 2.96451122e-01 9.69965696e-01
3.31151545e-01 -6.26703978e-01 8.23586106e-01 6.64665759e-01
-5.40019609e-02 -7.86644042e-01 -8.63679111e-01 -2.99781144e-01
-1.14719681e-01 1.45418704e-01 2.70742886e-02 1.09324789e+00
-1.66693881e-01 3.02922189e-01 -9.19431541e-03 -2.45077297e-01
7.88122892e-01 3.63761306e-01 3.28341156e-01 -1.53823781e+00
-1.90650314e-01 -5.22333980e-01 -4.03330803e-01 -4.15092647e-01
-5.28288186e-01 -7.47762561e-01 -3.83385152e-01 -1.88389051e+00
4.26470697e-01 -8.24443221e-01 -5.43602556e-02 3.03954575e-02
-3.28676224e-01 2.45973766e-01 3.08947235e-01 3.44801873e-01
2.99788383e-03 -3.56799334e-01 1.99599993e+00 2.99744517e-01
-4.02916104e-01 2.72419512e-01 -2.51956761e-01 9.71721768e-01
8.68276358e-01 -4.65512723e-01 -3.94625813e-01 -5.93840703e-02
-1.82849243e-01 7.14546442e-01 7.21524432e-02 -9.39924419e-01
-2.00021341e-01 4.59362082e-02 6.01306736e-01 -8.94101918e-01
-2.32319757e-02 -7.35498846e-01 3.48836273e-01 9.33859766e-01
8.23050365e-02 3.22285056e-01 -2.69092191e-02 4.26357938e-03
-3.80586803e-01 -6.60731792e-01 1.38706005e+00 -8.47193480e-01
-3.24664176e-01 9.89188105e-02 -6.19725347e-01 3.16737592e-01
1.31214619e+00 -6.54626667e-01 3.42355728e-01 -1.13701038e-01
-1.30035710e+00 7.90919960e-02 2.92675734e-01 -3.09015721e-01
4.46629137e-01 -6.64296746e-01 -8.16597641e-01 1.02610841e-01
-3.89309645e-01 3.55064631e-01 3.31606120e-01 1.72103250e+00
-1.44257724e+00 3.95843238e-01 -2.49628037e-01 -1.02685356e+00
-1.87277484e+00 2.36745715e-01 8.47676575e-01 -6.46172523e-01
-1.06756520e+00 6.71839595e-01 2.94838130e-01 -1.42310590e-01
-4.38732244e-02 -4.74031359e-01 -4.87650961e-01 -1.25421494e-01
2.86515236e-01 5.12076020e-01 -6.75425455e-02 -9.22836185e-01
-4.30286467e-01 9.15864289e-01 2.56846189e-01 1.22024469e-01
7.94038653e-01 -4.28057402e-01 -4.82661426e-01 4.28389490e-01
5.75686812e-01 2.18977779e-01 -6.30470395e-01 1.41764745e-01
1.00604735e-01 -4.24086720e-01 2.41693839e-01 -7.46508181e-01
-1.12611127e+00 1.13689744e+00 6.48001492e-01 3.29658955e-01
1.10518074e+00 -7.59752840e-02 7.42493749e-01 -6.06396973e-01
1.43590942e-01 -1.25167692e+00 -3.02139252e-01 9.81105193e-02
4.55841124e-01 -8.66108239e-01 4.59137291e-01 -7.82111883e-01
-8.90657485e-01 1.42019963e+00 1.26683637e-01 -3.12708947e-03
6.67325258e-01 4.75113809e-01 6.05930388e-01 -1.72079355e-01
-3.06323189e-02 -1.93760227e-02 2.78799772e-01 3.99034858e-01
9.90907967e-01 6.30793050e-02 -7.36711979e-01 -3.34342793e-02
1.65524691e-01 -3.05673510e-01 4.07839328e-01 1.04603839e+00
-6.19067311e-01 -1.10733104e+00 -7.92498171e-01 5.75896919e-01
-1.22781467e+00 2.26918906e-01 -9.39766169e-02 9.58480954e-01
3.14276159e-01 5.89013934e-01 -2.29571387e-01 6.06169820e-01
1.63512349e-01 -1.63198821e-03 6.66903913e-01 -4.83481616e-01
-6.65369272e-01 6.77633464e-01 -1.14569314e-01 -1.34334378e-02
-6.08578146e-01 -8.21897507e-01 -1.79472649e+00 1.20345447e-02
-3.41249466e-01 2.30954662e-01 7.43470490e-01 9.59991097e-01
-3.50686073e-01 6.47183120e-01 3.99323493e-01 -5.69076955e-01
-1.91897869e-01 -7.36276150e-01 -1.15128410e+00 4.52327907e-01
1.55712008e-01 -4.66954350e-01 -1.19822063e-01 3.12044293e-01]
|
[14.21518611907959, -2.5337109565734863]
|
d464de57-b3f5-4dc7-ba4f-c8c013c3a8ff
|
memex-detecting-explanatory-evidence-for
|
2305.15913
| null |
https://arxiv.org/abs/2305.15913v2
|
https://arxiv.org/pdf/2305.15913v2.pdf
|
MEMEX: Detecting Explanatory Evidence for Memes via Knowledge-Enriched Contextualization
|
Memes are a powerful tool for communication over social media. Their affinity for evolving across politics, history, and sociocultural phenomena makes them an ideal communication vehicle. To comprehend the subtle message conveyed within a meme, one must understand the background that facilitates its holistic assimilation. Besides digital archiving of memes and their metadata by a few websites like knowyourmeme.com, currently, there is no efficient way to deduce a meme's context dynamically. In this work, we propose a novel task, MEMEX - given a meme and a related document, the aim is to mine the context that succinctly explains the background of the meme. At first, we develop MCC (Meme Context Corpus), a novel dataset for MEMEX. Further, to benchmark MCC, we propose MIME (MultImodal Meme Explainer), a multimodal neural framework that uses common sense enriched meme representation and a layered approach to capture the cross-modal semantic dependencies between the meme and the context. MIME surpasses several unimodal and multimodal systems and yields an absolute improvement of ~ 4% F1-score over the best baseline. Lastly, we conduct detailed analyses of MIME's performance, highlighting the aspects that could lead to optimal modeling of cross-modal contextual associations.
|
['Ramaneswaran S', 'Tanmoy Chakraborty', 'Md. Shad Akhtar', 'Udit Arora', 'Shivam Sharma']
|
2023-05-25
| null | null | null | null |
['common-sense-reasoning']
|
['reasoning']
|
[-7.80301094e-02 -5.12411594e-02 -2.96129491e-02 -1.41891256e-01
-6.05361462e-01 -6.01414621e-01 1.15425754e+00 5.93390405e-01
-4.97040480e-01 6.47305727e-01 8.00391614e-01 -2.42367193e-01
4.36868519e-02 -6.23617411e-01 -6.31151736e-01 -2.31999218e-01
4.54880565e-01 1.53386280e-01 5.48903793e-02 -7.81926751e-01
4.31795567e-01 -2.89092779e-01 -1.54929650e+00 1.03866673e+00
6.98614359e-01 7.67417848e-01 4.49566722e-01 5.27596533e-01
-1.00760889e+00 9.42308128e-01 -7.68202722e-01 -6.86605453e-01
-6.19377971e-01 -7.10973859e-01 -1.10899436e+00 -1.76855966e-01
4.77539986e-01 2.02366352e-01 -5.63756585e-01 7.66507626e-01
1.78369135e-01 9.26719606e-02 7.30343282e-01 -9.09929156e-01
-9.89226818e-01 1.11289358e+00 -3.30870092e-01 3.51753592e-01
5.31142592e-01 -2.36846939e-01 1.02108324e+00 -1.24115753e+00
9.64764118e-01 1.28348696e+00 4.55574542e-01 5.10332584e-01
-1.05201900e+00 -3.27816606e-01 1.49342299e-01 1.33612275e-01
-1.27546215e+00 -3.73041779e-01 6.14907444e-01 -3.99790645e-01
8.34536910e-01 5.00967741e-01 5.85247576e-01 1.50641394e+00
4.32504207e-01 6.35579407e-01 1.22861910e+00 -4.63712424e-01
-1.06517360e-01 3.57246101e-01 2.22350419e-01 5.62482476e-01
-2.67770648e-01 -4.50408101e-01 -1.09587657e+00 -2.11722218e-02
-4.31384332e-02 6.67360658e-03 -2.43596524e-01 5.15279233e-01
-1.43432593e+00 5.47014415e-01 4.75332260e-01 5.51371992e-01
-3.90606970e-01 9.32618231e-02 4.15677428e-01 3.05422038e-01
5.48754334e-01 5.51159739e-01 3.22124735e-02 -9.92142558e-02
-8.09806347e-01 3.17813382e-02 1.02061844e+00 7.55019426e-01
5.39964318e-01 -6.03231072e-01 -1.31758541e-01 1.07694221e+00
2.99643546e-01 5.16248703e-01 3.50303918e-01 -5.21919906e-01
6.43239319e-01 8.66807401e-01 1.05675511e-01 -1.70696700e+00
-6.14046276e-01 -2.87533462e-01 -7.25179434e-01 -6.99085057e-01
-4.84433584e-03 -9.15825814e-02 -5.13019502e-01 1.80135643e+00
2.41669104e-01 -1.28504336e-01 1.75090387e-01 8.25014949e-01
1.44207788e+00 1.17865860e+00 3.52787614e-01 -4.80761603e-02
1.38417339e+00 -7.40886688e-01 -9.32902813e-01 -4.53545690e-01
3.55248064e-01 -1.17930794e+00 1.17488754e+00 -2.38300651e-01
-7.79079318e-01 -3.60029578e-01 -9.51492071e-01 -3.50849926e-01
-9.51351941e-01 1.58735260e-01 4.18295711e-01 1.06239572e-01
-7.97653317e-01 2.74633586e-01 -4.33185607e-01 -8.56010854e-01
3.04010580e-04 -5.03921658e-02 -4.35993552e-01 1.01766944e-01
-1.49887073e+00 9.55152690e-01 7.02286124e-01 2.01257512e-01
-4.47813541e-01 -4.97162670e-01 -7.50570476e-01 -8.16624686e-02
4.75730121e-01 -8.12610328e-01 7.96117425e-01 -7.37847269e-01
-1.24332547e+00 1.07566297e+00 -3.39259475e-01 -3.17449793e-02
5.34037277e-02 -9.99491960e-02 -9.12867367e-01 8.76557454e-02
4.29203026e-02 8.32348347e-01 5.73832512e-01 -1.45243120e+00
-6.57083154e-01 -8.60153213e-02 1.83519676e-01 9.45133865e-02
-7.15560317e-01 1.10213935e-01 -5.88603079e-01 -6.60689414e-01
1.35154501e-01 -8.95291150e-01 4.22169924e-01 -5.08269489e-01
-6.65972471e-01 -2.05950290e-01 6.93264544e-01 -7.89324105e-01
1.74377394e+00 -2.29244375e+00 5.51012754e-01 -3.55521291e-02
5.13316035e-01 -2.38054544e-01 -2.47042894e-01 1.28324258e+00
3.74132931e-01 3.20682287e-01 -3.29449773e-01 -4.38872904e-01
3.54118139e-01 1.66709036e-01 -4.77730185e-01 3.48612145e-02
7.70795122e-02 9.85675454e-01 -8.23443532e-01 -4.69414502e-01
-2.56044954e-01 6.03257477e-01 -2.97236443e-01 3.72451842e-02
-3.77522111e-01 4.78430718e-01 -2.44543090e-01 5.04926741e-01
3.02732646e-01 -6.44626081e-01 5.25700510e-01 -4.06254768e-01
-1.33678377e-01 3.33594114e-01 -8.31423402e-01 1.67829859e+00
-6.83672011e-01 1.10631776e+00 -2.11356636e-02 -3.14024657e-01
8.99915516e-01 1.31643876e-01 1.68095976e-01 -8.55642915e-01
2.54488319e-01 2.03675523e-01 -4.55183625e-01 -7.70781517e-01
1.19622254e+00 -2.07910374e-01 -4.68580723e-01 7.40338027e-01
-6.57247826e-02 1.53907493e-01 1.52679816e-01 7.43224382e-01
7.56781876e-01 -1.81795791e-01 2.74942428e-01 -2.37601101e-01
5.53191364e-01 6.89753518e-02 1.13963895e-01 6.28945470e-01
1.42918319e-01 3.32226604e-01 4.49469060e-01 -3.19950104e-01
-7.16008782e-01 -7.73428977e-01 -5.83747663e-02 1.41841042e+00
6.89988673e-01 -8.67103338e-01 -4.98711020e-01 -4.82227117e-01
-3.95704918e-02 7.43947208e-01 -8.64838421e-01 -1.19664976e-02
-4.50007439e-01 -8.71224999e-01 4.35383618e-01 1.70441255e-01
5.29717982e-01 -9.50112760e-01 -2.32369766e-01 2.53433526e-01
-9.64865446e-01 -1.24424505e+00 -4.90780145e-01 -2.17901707e-01
-3.53058696e-01 -9.38086867e-01 -2.21800700e-01 -5.61934769e-01
3.70988727e-01 3.38879198e-01 1.36450863e+00 3.25981110e-01
1.48099020e-01 6.18214488e-01 -4.56232816e-01 -1.66699976e-01
-6.70737743e-01 5.45785069e-01 -9.16895643e-02 6.32450879e-02
5.56115329e-01 -3.51887852e-01 -4.72491801e-01 2.54618764e-01
-1.11780941e+00 4.81369704e-01 3.44225317e-01 8.57009292e-01
3.83291543e-01 -3.54054272e-01 6.26205087e-01 -1.02702963e+00
7.85571158e-01 -1.25669634e+00 1.16332792e-01 5.09851217e-01
-9.92147699e-02 -1.28575146e-01 4.10401016e-01 -3.88806343e-01
-1.06647825e+00 -5.28778136e-01 5.18439300e-02 3.01502526e-01
2.22977772e-01 1.08674777e+00 7.02369809e-02 3.41508180e-01
6.69519782e-01 2.28990793e-01 -2.20224962e-01 -4.39723074e-01
5.48590660e-01 1.00938869e+00 6.14600837e-01 -6.61120653e-01
3.39733392e-01 4.65652794e-01 -2.24332914e-01 -9.71966505e-01
-1.16001666e+00 -3.42034489e-01 -4.11312550e-01 -4.52033132e-01
8.84521246e-01 -7.90815532e-01 -8.16242516e-01 9.00447518e-02
-1.28648341e+00 5.34218028e-02 3.47938269e-01 1.73661605e-01
-1.97107762e-01 -1.12950690e-02 -8.26303363e-01 -5.29867947e-01
-1.13133706e-01 -5.98550856e-01 9.29670215e-01 2.11833209e-01
-6.40785992e-01 -1.18698442e+00 3.27958949e-02 6.14343882e-01
5.33808172e-01 3.94560724e-01 1.11939204e+00 -6.99196160e-01
-4.72470671e-01 2.07775027e-01 -3.63894939e-01 -2.35944420e-01
5.96093424e-02 9.28075016e-02 -9.35324073e-01 -1.24807745e-01
-4.05184746e-01 -3.09070855e-01 1.00273597e+00 -2.01566890e-01
7.85555363e-01 -5.01191020e-01 -4.28639829e-01 1.81427643e-01
1.26458859e+00 3.60876089e-03 3.60101938e-01 5.69856226e-01
6.99322402e-01 1.00567293e+00 2.43654415e-01 3.29068869e-01
9.67755556e-01 5.95519602e-01 3.00606430e-01 1.43148780e-01
-1.88881606e-01 -5.19056797e-01 3.34540278e-01 1.62785995e+00
2.15432853e-01 -5.41456342e-01 -1.07488418e+00 4.71284986e-01
-1.87430227e+00 -1.02196205e+00 -1.70795947e-01 1.52954876e+00
7.47786283e-01 -1.79134339e-01 1.71622541e-02 -4.04367447e-01
8.05518508e-01 4.54320073e-01 -1.13254234e-01 -3.19418252e-01
-5.96242547e-01 -4.79506254e-01 -1.67163566e-01 6.56728208e-01
-8.73060524e-01 9.80915725e-01 6.10140944e+00 6.29871607e-01
-1.18245220e+00 3.50420296e-01 2.69587964e-01 -4.52502891e-02
-6.75634742e-01 -1.25593424e-01 -6.85612082e-01 7.84969568e-01
1.00521839e+00 -7.25544244e-02 4.94342774e-01 1.41024038e-01
-7.89759029e-03 -1.68418765e-01 -1.06487536e+00 9.46018100e-01
4.39622492e-01 -1.77799332e+00 1.79841116e-01 -1.07229844e-01
7.55542636e-01 -1.87401455e-02 1.62108481e-01 3.90767843e-01
-3.00621539e-01 -9.04071689e-01 1.01417756e+00 8.62747610e-01
5.82830489e-01 -5.41876137e-01 7.13927150e-01 2.90776014e-01
-9.69006062e-01 8.54454637e-02 -2.38984495e-01 1.09606117e-01
1.89640686e-01 3.01711529e-01 -5.17153621e-01 5.06433249e-01
4.89068985e-01 8.08795929e-01 -7.78243065e-01 4.56457257e-01
-1.17130145e-01 4.28755730e-01 -1.27528951e-01 -4.83550519e-01
4.06608969e-01 1.80925224e-02 7.90310621e-01 1.85531175e+00
2.99046338e-01 2.77806610e-01 4.74897698e-02 1.01151037e+00
-4.06790406e-01 3.80694687e-01 -4.60898638e-01 -6.65950000e-01
8.18602026e-01 1.18903327e+00 -7.50969231e-01 -2.26011112e-01
-4.24484640e-01 1.18892467e+00 3.81554276e-01 5.25885105e-01
-6.57250881e-01 -2.09025621e-01 3.48903924e-01 -1.25567809e-01
-1.22089587e-01 -2.86086202e-01 -1.89679161e-01 -1.17919230e+00
5.33255972e-02 -7.66660690e-01 4.89272714e-01 -8.22455227e-01
-1.48614049e+00 8.87141943e-01 -1.88523173e-01 -8.49837422e-01
-3.26256342e-02 -4.18181002e-01 -4.02302831e-01 8.15268397e-01
-1.26308572e+00 -1.49538577e+00 -3.15542966e-01 2.07306683e-01
4.97011870e-01 -2.20030621e-01 8.06446254e-01 4.17747676e-01
-6.41774952e-01 2.95463949e-01 3.50928232e-02 -2.30495911e-02
8.57084513e-01 -1.06365645e+00 1.12406217e-01 6.62928998e-01
3.86204720e-01 1.05851638e+00 9.67621565e-01 -8.01544070e-01
-1.49433219e+00 -7.93497443e-01 1.38983274e+00 -9.34508145e-01
1.10528052e+00 -3.35073113e-01 -1.05375636e+00 6.09109402e-01
7.89586246e-01 -4.83258367e-01 9.87491488e-01 4.33060050e-01
-4.74497974e-01 1.63921282e-01 -6.78287446e-01 8.55781138e-01
1.03024971e+00 -1.08268046e+00 -9.89967048e-01 1.78098470e-01
9.17132914e-01 -4.23536956e-01 -9.42801654e-01 -1.55444443e-01
5.36523461e-01 -8.59384418e-01 6.15112662e-01 -7.66625881e-01
9.81139243e-01 -3.98578107e-01 -5.68872452e-01 -1.45997488e+00
4.75436822e-03 -6.41290069e-01 -2.38952830e-01 1.71356821e+00
7.12621927e-01 -3.68411064e-01 9.24049616e-02 2.86027908e-01
-1.44660845e-01 -6.56383514e-01 -7.01808751e-01 -3.13514352e-01
-2.83354074e-02 -5.20060718e-01 7.10239947e-01 1.37537217e+00
4.12974745e-01 6.62817836e-01 -5.38190424e-01 5.97296208e-02
1.64865792e-01 2.09847525e-01 5.08869946e-01 -9.45797205e-01
2.14781519e-02 -5.45012355e-01 -1.28377303e-01 -8.60819757e-01
4.22081381e-01 -1.18323565e+00 -1.43276185e-01 -1.66692007e+00
6.66737616e-01 -1.84957296e-01 -2.72865921e-01 2.94256955e-01
-2.57188171e-01 3.49794865e-01 4.55353111e-01 3.90607297e-01
-9.32020128e-01 5.47256231e-01 1.07995582e+00 -4.48999733e-01
-2.22517997e-01 -7.23894954e-01 -9.11170781e-01 4.80829537e-01
4.73687142e-01 -4.77421075e-01 -8.15456882e-02 -5.27651429e-01
1.10373616e+00 6.51884452e-03 4.28357005e-01 -4.49614763e-01
5.40906847e-01 -2.87479460e-01 9.11453664e-02 -5.98669291e-01
3.47890913e-01 -6.79250717e-01 2.72680163e-01 -9.35345888e-02
-6.19160116e-01 3.07239622e-01 3.51409674e-01 5.39602697e-01
-5.80482483e-01 -3.80673446e-02 3.40147018e-01 5.65815903e-02
-9.56093252e-01 -1.12523153e-01 -6.65753961e-01 2.87640482e-01
5.27376473e-01 1.15416721e-01 -8.88883293e-01 -4.57002252e-01
-8.45218837e-01 2.15522587e-01 3.55339110e-01 8.37993443e-01
6.05223000e-01 -1.50230491e+00 -8.30308020e-01 -4.17474210e-01
5.63636065e-01 -7.88396180e-01 4.94287491e-01 9.04362381e-01
-2.24041328e-01 3.66046488e-01 -7.54793407e-03 -3.92459452e-01
-9.92176473e-01 3.25222611e-01 2.50497580e-01 2.74932534e-01
-7.14632392e-01 6.23037517e-01 -3.36393043e-02 -5.30077577e-01
-2.21966226e-02 1.89852889e-03 -5.20045102e-01 5.84259450e-01
8.86781931e-01 2.95879096e-01 -1.51564702e-01 -1.04008555e+00
-3.11782658e-01 4.37460005e-01 1.50658965e-01 -2.87885934e-01
1.30107677e+00 -9.63809013e-01 -6.51661873e-01 1.04701614e+00
1.19626033e+00 3.01811606e-01 -5.56723595e-01 -3.46948534e-01
4.33340281e-01 -3.48407686e-01 -6.36153370e-02 -1.12146139e+00
-5.67246974e-01 7.70816207e-01 2.49014258e-01 4.94065255e-01
7.61293709e-01 3.66611391e-01 1.02618313e+00 4.72769648e-01
3.85248587e-02 -1.24567604e+00 1.35921508e-01 8.86093497e-01
1.22853696e+00 -1.43897927e+00 -2.43112981e-01 -2.73415744e-01
-7.61857092e-01 1.29419732e+00 4.65794206e-01 6.84616685e-01
4.98594403e-01 -8.85589197e-02 2.73667306e-01 -6.77878499e-01
-8.82995009e-01 -1.39318332e-01 7.79015005e-01 -3.04723717e-02
6.79930031e-01 1.11182690e-01 -4.31493551e-01 8.38653684e-01
-2.97664493e-01 -4.33321208e-01 5.30961990e-01 5.62945664e-01
-5.08252621e-01 -7.15324521e-01 -3.61113816e-01 2.64921151e-02
-4.62265253e-01 -8.46847743e-02 -8.49446535e-01 8.07845712e-01
1.28598586e-01 1.18258715e+00 2.62141854e-01 -6.39900446e-01
4.68306653e-02 2.09913239e-01 4.69679311e-02 -5.67714691e-01
-7.45918751e-01 -1.88077316e-01 3.98844898e-01 -4.88707125e-01
-6.98154628e-01 -3.46768916e-01 -1.40344584e+00 -7.72456229e-01
9.48527381e-02 1.22853011e-01 9.38722253e-01 1.25705719e+00
4.72034454e-01 5.38418233e-01 4.72551227e-01 -3.70742053e-01
3.06553483e-01 -9.36112642e-01 -2.72041857e-01 4.80660290e-01
1.31126493e-01 -5.67434967e-01 -2.24919245e-01 -4.32885848e-02]
|
[8.516389846801758, 10.69361686706543]
|
5b48f174-bfb9-4d9a-8fc5-9247227e9473
|
task-specific-fine-tuning-via-variational
|
2303.08446
| null |
https://arxiv.org/abs/2303.08446v1
|
https://arxiv.org/pdf/2303.08446v1.pdf
|
Task-specific Fine-tuning via Variational Information Bottleneck for Weakly-supervised Pathology Whole Slide Image Classification
|
While Multiple Instance Learning (MIL) has shown promising results in digital Pathology Whole Slide Image (WSI) classification, such a paradigm still faces performance and generalization problems due to challenges in high computational costs on Gigapixel WSIs and limited sample size for model training. To deal with the computation problem, most MIL methods utilize a frozen pretrained model from ImageNet to obtain representations first. This process may lose essential information owing to the large domain gap and hinder the generalization of model due to the lack of image-level training-time augmentations. Though Self-supervised Learning (SSL) proposes viable representation learning schemes, the improvement of the downstream task still needs to be further explored in the conversion from the task-agnostic features of SSL to the task-specifics under the partial label supervised learning. To alleviate the dilemma of computation cost and performance, we propose an efficient WSI fine-tuning framework motivated by the Information Bottleneck theory. The theory enables the framework to find the minimal sufficient statistics of WSI, thus supporting us to fine-tune the backbone into a task-specific representation only depending on WSI-level weak labels. The WSI-MIL problem is further analyzed to theoretically deduce our fine-tuning method. Our framework is evaluated on five pathology WSI datasets on various WSI heads. The experimental results of our fine-tuned representations show significant improvements in both accuracy and generalization compared with previous works. Source code will be available at https://github.com/invoker-LL/WSI-finetuning.
|
['Lin Yang', 'Sunyi Zheng', 'Wenwei Kuang', 'Zhongyi Shui', 'Yuxuan Sun', 'Yunlong Zhang', 'Chenglu Zhu', 'Honglin Li']
|
2023-03-15
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Task-Specific_Fine-Tuning_via_Variational_Information_Bottleneck_for_Weakly-Supervised_Pathology_Whole_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Task-Specific_Fine-Tuning_via_Variational_Information_Bottleneck_for_Weakly-Supervised_Pathology_Whole_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['multiple-instance-learning']
|
['methodology']
|
[ 6.42942011e-01 1.97702259e-01 -5.48825443e-01 -4.83230412e-01
-1.28799379e+00 -2.42565930e-01 2.58876622e-01 2.26809829e-01
-5.52774549e-01 8.70077193e-01 1.27557993e-01 -2.52686381e-01
-5.36554873e-01 -6.52870119e-01 -5.44710696e-01 -1.09927297e+00
1.54543340e-01 4.99687970e-01 2.06324607e-01 -1.45806015e-01
1.80727199e-01 3.30860674e-01 -1.19045186e+00 5.63471854e-01
9.59794044e-01 1.01687837e+00 5.09898245e-01 5.44858158e-01
-4.53128278e-01 5.25943339e-01 -5.78865707e-01 1.50362495e-02
1.49753034e-01 -2.35510662e-01 -1.08236766e+00 -4.58266698e-02
1.96289822e-01 2.40617990e-02 -3.58036816e-01 1.03913975e+00
5.93183517e-01 -5.11648618e-02 5.40474832e-01 -1.12658501e+00
-3.43037665e-01 5.04208386e-01 -6.89721048e-01 5.00711858e-01
-3.36453885e-01 1.84010729e-01 9.63767707e-01 -5.68362057e-01
8.60537350e-01 8.66311789e-01 5.23792684e-01 7.66689718e-01
-1.21707976e+00 -8.26012611e-01 4.51721177e-02 2.69580066e-01
-1.50610721e+00 -3.74303401e-01 5.81670046e-01 -1.48581162e-01
7.68088162e-01 3.62119079e-01 3.02108437e-01 8.82296026e-01
1.09093361e-01 8.84892225e-01 1.25462496e+00 -5.03243089e-01
-2.44259238e-02 1.48780748e-01 4.18781221e-01 7.04226434e-01
1.59988180e-01 -2.45631248e-01 -5.45347393e-01 -2.70856656e-02
7.54199445e-01 1.13884211e-01 -3.11001986e-01 -5.79481833e-02
-1.25161505e+00 7.53550887e-01 6.92907214e-01 5.41667759e-01
1.03707463e-02 -1.46989673e-01 6.59085333e-01 4.22274441e-01
6.24341846e-01 5.27297854e-01 -6.68217421e-01 1.27060995e-01
-9.97512460e-01 -1.26474798e-01 4.10760134e-01 6.09178901e-01
7.46286929e-01 -3.42708141e-01 -4.35965627e-01 1.12252378e+00
-1.51657850e-01 7.99101070e-02 8.56993735e-01 -5.16793907e-01
4.96677935e-01 9.18582201e-01 -4.85997260e-01 -6.16680682e-01
-7.46671021e-01 -8.31078172e-01 -1.15688872e+00 -1.22032724e-01
5.94897866e-01 2.12700680e-01 -9.91572797e-01 1.71874344e+00
2.63943344e-01 3.45039129e-01 -9.53599587e-02 8.58270168e-01
1.04572880e+00 4.80816334e-01 1.49921194e-01 -2.34909803e-01
1.44819665e+00 -9.69724059e-01 -5.82142651e-01 -1.27890214e-01
1.10760844e+00 -4.37227190e-01 1.11897099e+00 6.28909618e-02
-8.64119709e-01 -4.58466113e-01 -9.58149016e-01 -1.18772969e-01
-4.65251386e-01 2.80015737e-01 7.44151413e-01 3.86547834e-01
-1.13306844e+00 4.46647257e-01 -7.79139102e-01 -3.36499929e-01
8.86018097e-01 7.15561271e-01 -5.16776562e-01 -2.99286336e-01
-1.20334268e+00 6.15009069e-01 4.59274441e-01 5.41109145e-02
-6.79104984e-01 -1.09181702e+00 -6.29461586e-01 1.10732101e-01
4.04561162e-01 -7.01566458e-01 8.97356629e-01 -8.39654565e-01
-1.34499943e+00 1.19686258e+00 -1.58825263e-01 -2.96701461e-01
4.84262794e-01 4.02410567e-01 -1.26323149e-01 2.19333217e-01
1.53024271e-01 6.59719706e-01 6.37961984e-01 -8.09270620e-01
-7.00704455e-01 -5.99672258e-01 7.07420334e-02 2.60833293e-01
-6.82709754e-01 -3.46700400e-01 -7.49855340e-01 -5.82369745e-01
1.91672444e-01 -9.15389359e-01 -5.13407528e-01 -1.14432797e-01
-5.55944681e-01 -2.70915031e-01 4.66187537e-01 -4.01310444e-01
1.11957240e+00 -2.23732257e+00 -2.44571492e-02 2.03185722e-01
2.98797399e-01 4.06748891e-01 -5.56517363e-01 6.43718839e-02
-3.29203129e-01 2.21947536e-01 -9.18218419e-02 -2.36558795e-01
-4.81568724e-01 1.45186320e-01 1.17228113e-01 5.42505503e-01
3.17356557e-01 8.87041509e-01 -8.39365005e-01 -8.14639688e-01
1.52521916e-02 3.00893933e-01 -4.62260872e-01 2.09077522e-01
7.33585581e-02 6.33976698e-01 -5.84912419e-01 6.85258567e-01
7.68201649e-01 -7.52775669e-01 2.78709829e-01 -4.14301336e-01
1.64019153e-01 3.08228731e-01 -9.34849620e-01 1.95211828e+00
-4.42244411e-01 2.03688174e-01 1.04421496e-01 -1.50662684e+00
7.19079435e-01 8.03464428e-02 7.07540214e-01 -7.43141055e-01
-1.48686823e-02 2.67898858e-01 5.56254275e-02 -4.84848112e-01
-1.27144586e-02 -1.30242333e-01 -1.09524755e-02 2.10464954e-01
2.32864574e-01 2.42968783e-01 3.74011368e-01 2.47874394e-01
1.26060879e+00 -4.83942404e-02 3.84616375e-01 -4.89566118e-01
7.82438815e-01 1.65915251e-01 7.75738776e-01 8.07118356e-01
-3.04133028e-01 4.29663122e-01 4.90440190e-01 -3.07052374e-01
-7.55482912e-01 -8.39310050e-01 -5.68111956e-01 1.25697017e+00
9.89085287e-02 -2.47626752e-01 -6.60982370e-01 -8.94273281e-01
-1.53522894e-01 2.36617491e-01 -8.44663501e-01 -2.48309612e-01
-5.18428147e-01 -1.42269540e+00 5.31274736e-01 3.68617952e-01
3.67037743e-01 -7.57391453e-01 -1.46390110e-01 9.41734463e-02
-2.93785602e-01 -8.78244519e-01 -3.13043267e-01 3.91661108e-01
-1.04084849e+00 -9.72565651e-01 -7.59654880e-01 -8.03619385e-01
1.06841385e+00 4.36058491e-01 7.16930211e-01 3.29818577e-01
-8.28039050e-01 -9.17593576e-03 -3.12638551e-01 -3.67339760e-01
-4.68562692e-01 5.28402805e-01 -2.05339313e-01 -8.41083154e-02
3.23563576e-01 -3.58054549e-01 -9.00706232e-01 2.79393524e-01
-1.16352105e+00 3.07239592e-01 9.31904674e-01 1.34619677e+00
8.38743269e-01 6.58424348e-02 8.15416932e-01 -1.19324470e+00
3.62398535e-01 -3.94857645e-01 -3.23800206e-01 4.25208867e-01
-6.02238595e-01 2.03518733e-01 5.54093361e-01 -3.34951311e-01
-1.10540390e+00 -7.92938545e-02 -1.90410376e-01 -8.28327537e-02
3.25385528e-03 5.54235578e-01 -4.42529917e-02 -1.97256327e-01
6.36056185e-01 2.78070807e-01 3.93317699e-01 -2.41649404e-01
2.07420457e-02 6.04709268e-01 8.96796137e-02 -4.54360008e-01
4.87959355e-01 5.12732029e-01 3.23137939e-02 -6.51598990e-01
-1.23068833e+00 -8.03658843e-01 -5.66202879e-01 -3.12948972e-02
4.91826117e-01 -7.94551373e-01 -7.27932215e-01 3.29108536e-01
-7.38974571e-01 -3.64581198e-01 -4.27448422e-01 2.73783654e-01
-4.61806685e-01 2.97300905e-01 -9.64490175e-01 -2.19375372e-01
-5.53474367e-01 -1.48316932e+00 1.02014232e+00 2.11464792e-01
4.59594280e-02 -9.56243694e-01 4.34661545e-02 6.43949091e-01
4.94106442e-01 -5.40012121e-02 1.23528814e+00 -8.27877462e-01
-4.63704079e-01 -2.95370132e-01 -5.92734218e-01 2.35557750e-01
2.60256261e-01 -3.72835040e-01 -1.02304912e+00 -5.14339089e-01
-1.97246268e-01 -5.54061413e-01 1.20944238e+00 5.98097622e-01
1.71999097e+00 -9.92427990e-02 -7.16357529e-01 9.35839772e-01
1.61448777e+00 -2.22741812e-01 4.63550329e-01 4.71506625e-01
4.92324919e-01 6.16483271e-01 6.63770139e-01 3.85718733e-01
2.13166922e-01 5.30658007e-01 2.02823296e-01 -4.36368585e-01
-4.85593677e-01 -2.59007644e-02 -1.41529694e-01 9.33019698e-01
-1.22363754e-01 -2.41728071e-02 -8.20731997e-01 4.80757445e-01
-1.66521406e+00 -6.60423756e-01 2.18598455e-01 2.13542342e+00
1.13257921e+00 2.03217529e-02 -2.37609208e-01 1.97088718e-01
5.54187953e-01 7.65263960e-02 -7.46402085e-01 -2.86962721e-03
1.49032343e-02 2.64084816e-01 7.21676588e-01 3.02669078e-01
-1.09581888e+00 9.04526651e-01 5.30697393e+00 1.47777963e+00
-1.19471908e+00 3.86496425e-01 1.10115612e+00 -2.36645445e-01
-9.06595513e-02 -3.18164825e-01 -1.11662817e+00 3.12384307e-01
9.13493097e-01 -4.97488081e-02 1.82026714e-01 6.08188450e-01
1.71705335e-01 2.77583711e-02 -8.43542159e-01 9.79494095e-01
-1.77593157e-01 -1.82156217e+00 1.22994594e-01 1.47641435e-01
5.76488554e-01 1.66425318e-01 1.07676812e-01 5.10380983e-01
-1.80401606e-03 -9.02201235e-01 -6.09221049e-02 3.15632969e-01
1.25052822e+00 -5.40487468e-01 9.04653907e-01 2.68604457e-01
-1.01601434e+00 -8.31863210e-02 -5.47277331e-01 2.85681218e-01
-3.78523141e-01 7.29543447e-01 -1.27347159e+00 6.11071229e-01
6.11477911e-01 5.70384622e-01 -7.64540911e-01 8.69341791e-01
1.18402719e-01 6.66566312e-01 -1.44946143e-01 6.98683932e-02
2.67929643e-01 -1.03613265e-01 2.13336319e-01 1.47558141e+00
1.80379584e-01 8.54335055e-02 1.78899378e-01 5.16323209e-01
-1.41126126e-01 2.94715673e-01 -2.53985912e-01 -2.19663177e-02
3.10228318e-01 1.58047974e+00 -9.34090316e-01 -3.16997200e-01
-2.56430894e-01 8.31174254e-01 7.09340513e-01 3.60780656e-01
-5.30660212e-01 -2.38722786e-01 5.60097694e-01 2.07269952e-01
-3.15707140e-02 1.47196710e-01 -4.37431425e-01 -1.09387636e+00
-1.87280238e-01 -7.24527597e-01 7.56152570e-01 -7.24110454e-02
-1.42441022e+00 6.34203494e-01 -2.72659838e-01 -1.25941467e+00
6.79248944e-02 -7.21045911e-01 -4.19517428e-01 5.70369065e-01
-1.97455931e+00 -1.33833647e+00 -3.02017301e-01 6.67902291e-01
5.92626512e-01 -8.99242535e-02 9.47311044e-01 3.11414897e-01
-8.36543024e-01 1.07143140e+00 3.98221463e-01 1.10739470e-01
9.95666683e-01 -9.99843180e-01 -2.48644426e-01 3.64743590e-01
-2.94248134e-01 5.61735392e-01 2.08710834e-01 -3.70821446e-01
-1.25780976e+00 -1.06203330e+00 7.47256815e-01 -6.73835129e-02
8.05284202e-01 -1.92159548e-01 -1.03122842e+00 3.79931957e-01
-2.96563923e-01 4.17947024e-01 1.18027294e+00 2.36155808e-01
-2.39404798e-01 -4.40311193e-01 -1.16303861e+00 3.87242079e-01
1.07172811e+00 -3.36915791e-01 -6.62880689e-02 6.61789715e-01
6.33819222e-01 -3.69979531e-01 -1.12616146e+00 4.86942977e-01
2.78128177e-01 -5.68117440e-01 1.04411340e+00 -7.25582004e-01
2.62734532e-01 -9.10382792e-02 7.80667514e-02 -1.14785993e+00
-4.66862261e-01 -2.15916410e-01 3.79629552e-01 1.13675630e+00
6.15244806e-01 -6.66068077e-01 9.02785301e-01 4.47610885e-01
-2.63403237e-01 -1.20149171e+00 -1.15269256e+00 -5.92648149e-01
1.16656788e-01 -3.33522111e-01 4.37360078e-01 8.90161812e-01
1.85389429e-01 1.32463425e-01 -6.04792982e-02 7.32328147e-02
5.81237733e-01 2.84434497e-01 4.02358532e-01 -1.09256577e+00
-2.61834025e-01 -4.91611153e-01 -3.80968034e-01 -6.29770994e-01
2.21927986e-01 -1.47393179e+00 -6.40299693e-02 -1.36890662e+00
7.08560526e-01 -8.55076671e-01 -8.79486442e-01 6.64071679e-01
-4.09495473e-01 3.24758649e-01 8.15462768e-02 4.84833241e-01
-7.37411737e-01 2.20654398e-01 1.58509111e+00 -3.80079597e-01
-1.56802550e-01 1.13497898e-02 -8.07789743e-01 4.65632230e-01
8.17029357e-01 -4.69740868e-01 -4.34667557e-01 -2.78049439e-01
5.57646565e-02 1.35414749e-01 5.27606383e-02 -8.15060973e-01
3.45196724e-01 -1.62184164e-01 4.76567239e-01 -2.68510967e-01
2.76312441e-01 -4.44500208e-01 -3.09916258e-01 5.70528686e-01
-7.32566178e-01 -7.15765178e-01 1.71326190e-01 5.20644307e-01
-3.52515012e-01 -4.03491288e-01 1.03119040e+00 -2.36114785e-01
-6.90181077e-01 7.46114910e-01 -1.68299749e-01 -1.69956200e-02
9.05440211e-01 -1.33853033e-01 -4.86929119e-01 1.33760899e-01
-7.82711685e-01 2.91093498e-01 1.12268306e-01 1.32590383e-01
5.69348454e-01 -1.01902974e+00 -8.15590203e-01 3.72195363e-01
3.74527901e-01 1.63641855e-01 8.11032057e-01 1.13279903e+00
-3.33991796e-01 5.87124646e-01 -1.97582468e-01 -7.17000186e-01
-1.09265983e+00 5.05398452e-01 1.25128075e-01 -1.10567534e+00
-4.94950175e-01 1.11511540e+00 6.51421130e-01 -5.82552254e-01
1.75698683e-01 -2.18515515e-01 -1.50612384e-01 -1.26043241e-02
7.39404619e-01 1.40911132e-01 2.99165249e-01 -2.33105138e-01
-4.23828572e-01 4.12709683e-01 -6.31017566e-01 4.34828550e-01
1.35363066e+00 -1.07389405e-01 -1.24695018e-01 7.89351612e-02
1.47982752e+00 -4.71055478e-01 -1.11013699e+00 -5.68956792e-01
7.16582760e-02 -2.86850214e-01 3.16575408e-01 -7.13525534e-01
-1.11395943e+00 8.65901113e-01 8.10795069e-01 -2.84108192e-01
1.29336095e+00 2.01126158e-01 8.14554513e-01 2.98629314e-01
3.03836137e-01 -1.03849363e+00 7.72154629e-02 2.36115173e-01
6.33066475e-01 -1.49801791e+00 -3.71116139e-02 -5.23175418e-01
-4.97740120e-01 1.10007715e+00 6.52735591e-01 6.51772395e-02
6.56658530e-01 4.47144657e-01 -2.54232772e-02 -1.74335390e-01
-7.08299637e-01 -1.94827422e-01 2.24755004e-01 4.98153865e-01
4.89264965e-01 1.63999796e-01 -4.26062375e-01 6.89326882e-01
1.67485639e-01 2.14173108e-01 2.03261316e-01 6.08961761e-01
-2.82022595e-01 -1.22039366e+00 -6.86161444e-02 8.59597921e-01
-4.77875203e-01 -6.30874485e-02 1.80830270e-01 7.80681849e-01
1.21397175e-01 6.33656740e-01 4.28387783e-02 -1.23083346e-01
1.00706875e-01 1.75490491e-02 3.90515924e-01 -6.11079752e-01
-4.44085866e-01 1.67978123e-01 -1.27944335e-01 -6.78145528e-01
-4.30385649e-01 -3.61810714e-01 -1.49235737e+00 -3.38886045e-02
-2.91206360e-01 1.50653231e-03 5.86103797e-01 9.50547218e-01
3.71584326e-01 8.02468300e-01 6.25341535e-01 -6.79944992e-01
-7.03963339e-01 -9.73508060e-01 -6.11572146e-01 5.10311186e-01
2.98541486e-01 -5.36459327e-01 -2.72176415e-01 -2.14972913e-01]
|
[15.097735404968262, -2.7876036167144775]
|
ff7f31dd-c7da-45d7-8251-d16659019924
|
decomposed-cross-modal-distillation-for-rgb
|
2303.17285
| null |
https://arxiv.org/abs/2303.17285v1
|
https://arxiv.org/pdf/2303.17285v1.pdf
|
Decomposed Cross-modal Distillation for RGB-based Temporal Action Detection
|
Temporal action detection aims to predict the time intervals and the classes of action instances in the video. Despite the promising performance, existing two-stream models exhibit slow inference speed due to their reliance on computationally expensive optical flow. In this paper, we introduce a decomposed cross-modal distillation framework to build a strong RGB-based detector by transferring knowledge of the motion modality. Specifically, instead of direct distillation, we propose to separately learn RGB and motion representations, which are in turn combined to perform action localization. The dual-branch design and the asymmetric training objectives enable effective motion knowledge transfer while preserving RGB information intact. In addition, we introduce a local attentive fusion to better exploit the multimodal complementarity. It is designed to preserve the local discriminability of the features that is important for action localization. Extensive experiments on the benchmarks verify the effectiveness of the proposed method in enhancing RGB-based action detectors. Notably, our framework is agnostic to backbones and detection heads, bringing consistent gains across different model combinations.
|
['Hyeran Byun', 'Dongyoon Wee', 'Minho Shim', 'Taeoh Kim', 'Pilhyeon Lee']
|
2023-03-30
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Lee_Decomposed_Cross-Modal_Distillation_for_RGB-Based_Temporal_Action_Detection_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Lee_Decomposed_Cross-Modal_Distillation_for_RGB-Based_Temporal_Action_Detection_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['action-localization']
|
['computer-vision']
|
[ 2.61139154e-01 -1.63008779e-01 -6.26699328e-01 -1.51936084e-01
-6.53961420e-01 -5.81309021e-01 7.56856203e-01 -2.18702987e-01
-3.84025156e-01 5.29335856e-01 5.27298927e-01 3.27399187e-02
-1.31197974e-01 -5.35945117e-01 -5.88218987e-01 -9.54515636e-01
-6.62236139e-02 -2.12975502e-01 3.18780363e-01 -3.10528018e-02
1.63393065e-01 4.53206301e-01 -1.55155814e+00 5.44481277e-01
7.11988688e-01 1.29339480e+00 6.43934831e-02 6.68456554e-01
1.58244282e-01 1.46417809e+00 -1.29410252e-01 -1.98320597e-01
4.07573283e-01 -6.08819485e-01 -7.17111111e-01 2.93586075e-01
4.37602103e-01 -6.35096252e-01 -8.52852941e-01 6.62254930e-01
3.32374901e-01 2.70356536e-01 2.40027264e-01 -1.54949653e+00
-3.10621142e-01 3.52596164e-01 -5.49210846e-01 3.09938818e-01
4.68538910e-01 6.25199914e-01 1.25093627e+00 -6.63042247e-01
5.25005877e-01 1.16541660e+00 4.16879654e-01 5.26611507e-01
-9.72791195e-01 -4.12631720e-01 4.36489731e-01 5.81799328e-01
-1.10100341e+00 -5.56061149e-01 9.45808887e-01 -3.55076700e-01
6.67060256e-01 1.18458845e-01 7.96166897e-01 1.24416137e+00
-7.61851743e-02 1.25759673e+00 1.07127976e+00 -5.59826940e-02
2.24153295e-01 -1.52520612e-01 -1.89851969e-01 9.91024494e-01
-1.21607460e-01 8.54308903e-02 -1.03337586e+00 2.15572238e-01
9.03731525e-01 3.07159841e-01 -5.52967250e-01 -5.37022829e-01
-1.54086030e+00 5.68564057e-01 7.01410651e-01 7.96276554e-02
-4.26653326e-01 4.36574161e-01 4.26325321e-01 -1.14345998e-01
2.04103351e-01 1.18662730e-01 -2.97383189e-01 -4.19669658e-01
-8.17587137e-01 6.61028847e-02 4.25996900e-01 7.64220417e-01
8.30837905e-01 -1.28744483e-01 -7.11517274e-01 1.83888122e-01
3.11121911e-01 4.07364577e-01 3.90433580e-01 -1.43672025e+00
7.36344755e-01 8.31889451e-01 -1.41755156e-02 -1.10524416e+00
-3.49248886e-01 -3.17162395e-01 -6.34442210e-01 1.95467770e-02
4.73093659e-01 6.28071576e-02 -7.34981298e-01 1.82522511e+00
3.81343395e-01 6.36478782e-01 -5.02926223e-02 1.08834064e+00
4.30607617e-01 3.31550777e-01 1.82676300e-01 7.87039399e-02
1.16180336e+00 -1.08674955e+00 -4.97229517e-01 -2.35742286e-01
7.89362431e-01 -3.81229639e-01 8.97281826e-01 6.12764545e-02
-1.06335413e+00 -5.70469201e-01 -9.60205853e-01 -2.82439768e-01
-1.99974597e-01 2.61565685e-01 8.65375042e-01 4.58548516e-01
-7.36209333e-01 5.54607809e-01 -1.26167715e+00 -1.46509424e-01
6.86056972e-01 2.55534351e-01 -4.84699011e-01 -1.20400392e-01
-8.93323541e-01 6.27538443e-01 4.66178596e-01 2.07227662e-01
-9.25329447e-01 -6.23239338e-01 -9.07417536e-01 4.69316915e-02
4.31547016e-01 -7.95131922e-01 9.62741613e-01 -9.52708960e-01
-1.65179265e+00 4.59599435e-01 -2.60661691e-01 -4.65844691e-01
7.04811037e-01 -2.37239018e-01 -7.65850162e-03 7.74122655e-01
3.89481522e-02 9.31368053e-01 7.51169562e-01 -1.01243043e+00
-1.04943299e+00 -3.74758631e-01 6.29480004e-01 3.65130782e-01
-6.74618125e-01 -2.96332806e-01 -7.41030157e-01 -4.92717743e-01
1.55542552e-01 -8.46889496e-01 -9.77037996e-02 3.49279732e-01
-4.20197278e-01 -2.37134397e-02 8.34252775e-01 -4.88399476e-01
1.26636171e+00 -2.15434170e+00 4.51360434e-01 -3.87022644e-02
2.45188117e-01 1.18007138e-01 -1.93644434e-01 2.38811761e-01
1.70156360e-01 -3.16980273e-01 -1.36082739e-01 -5.67960560e-01
1.00385182e-01 4.18139249e-01 -1.68282315e-01 8.16821098e-01
5.67209899e-01 1.02860188e+00 -1.15034544e+00 -6.92502499e-01
5.39992809e-01 7.12645531e-01 -7.91202307e-01 8.29875395e-02
-1.20912746e-01 7.93346465e-01 -7.10536957e-01 8.74690115e-01
3.12166274e-01 -2.82040417e-01 1.19679265e-01 -4.83395010e-01
2.40572896e-02 3.45493019e-01 -1.14106965e+00 2.12615466e+00
-3.18425119e-01 4.95805860e-01 -1.07078459e-02 -1.07592952e+00
3.72621417e-01 1.34563416e-01 9.73706245e-01 -8.74192357e-01
4.30716239e-02 1.66002307e-02 -1.64718747e-01 -4.69656587e-01
2.51073569e-01 5.18957675e-02 8.19137618e-02 2.93805838e-01
6.14698743e-03 4.83357370e-01 2.26717010e-01 2.69110501e-01
1.24842680e+00 7.48892903e-01 1.52559459e-01 2.11247683e-01
6.63837075e-01 -1.86280370e-01 7.93288291e-01 4.87678915e-01
-5.99345267e-01 5.86922526e-01 4.95556563e-01 -3.20339650e-01
-4.69547182e-01 -9.82997537e-01 1.57311231e-01 1.15709412e+00
4.85129029e-01 -5.67111611e-01 -4.18826878e-01 -9.29776669e-01
-3.30794342e-02 3.04806679e-01 -7.00824678e-01 -3.66784036e-01
-7.29933619e-01 -5.37756145e-01 5.84603012e-01 9.44502711e-01
7.33158946e-01 -7.81763434e-01 -1.14999139e+00 -7.37712998e-03
-4.55653012e-01 -1.32591414e+00 -4.39092308e-01 1.05520189e-01
-7.72492230e-01 -1.21006238e+00 -5.40392697e-01 -3.57405454e-01
3.78628165e-01 4.85716492e-01 7.53459930e-01 -1.09488711e-01
-2.31704712e-01 6.88538194e-01 -4.77716327e-01 1.86695665e-01
5.65646850e-02 -2.65603643e-02 -1.46352157e-01 4.59086210e-01
2.14968815e-01 -6.09528840e-01 -1.15057015e+00 1.51896104e-01
-9.15671110e-01 1.28416136e-01 6.79549932e-01 6.94859684e-01
3.51515263e-01 -2.62923568e-01 8.23641270e-02 -3.22047353e-01
-1.30585851e-02 -4.22334373e-01 -1.52980939e-01 2.75169253e-01
-2.21988752e-01 2.39885703e-01 5.22893548e-01 -2.43241951e-01
-1.09791934e+00 4.73964930e-01 2.95847386e-01 -6.53691888e-01
-1.35642514e-01 6.68867603e-02 -2.07081199e-01 -9.43027362e-02
1.92809790e-01 3.16699564e-01 -8.52939188e-02 -3.14998180e-01
7.10593998e-01 2.64104873e-01 5.68122268e-01 -5.37437081e-01
5.71511090e-01 1.03706455e+00 1.44766703e-01 -4.89363641e-01
-7.64718831e-01 -6.75806582e-01 -7.97619998e-01 -4.28718239e-01
1.03543901e+00 -1.04742372e+00 -1.04297650e+00 3.47499430e-01
-1.02080369e+00 -2.16123283e-01 -2.97468632e-01 6.79896951e-01
-7.68884599e-01 5.00584304e-01 -6.91300869e-01 -7.01012433e-01
-1.48747480e-02 -1.13293374e+00 1.42524469e+00 1.93071485e-01
1.26136571e-01 -9.29609120e-01 4.52760160e-02 4.34073925e-01
1.70282245e-01 3.28482658e-01 5.01371086e-01 -1.84784949e-01
-8.45317543e-01 -1.46550044e-01 -4.48452473e-01 2.13956833e-01
2.06135839e-01 -3.91345508e-02 -1.09225631e+00 -9.62201878e-02
-3.56444716e-01 -3.71528238e-01 1.29356778e+00 2.20215261e-01
1.05332899e+00 -1.13413528e-01 -2.79829502e-01 8.50985348e-01
1.28224349e+00 -1.67361721e-01 6.34095073e-01 4.10419136e-01
1.03795278e+00 5.49728215e-01 6.23118222e-01 8.31730425e-01
5.79350173e-01 8.14781189e-01 6.97626531e-01 -1.72381207e-01
-2.83956558e-01 -2.75266171e-01 7.25121617e-01 4.32336390e-01
-4.62105215e-01 -2.07114052e-02 -6.14242017e-01 3.01815718e-01
-2.11609817e+00 -1.18749821e+00 1.57123789e-01 2.05387330e+00
7.36526966e-01 -2.72406526e-02 3.77019018e-01 1.28271863e-01
3.46639961e-01 4.74576324e-01 -5.02250373e-01 1.67956471e-01
-2.41328672e-01 1.25635713e-01 6.80761278e-01 2.85639077e-01
-1.35276556e+00 8.00058126e-01 5.45403767e+00 5.61492622e-01
-1.05505407e+00 5.16184345e-02 3.72188032e-01 -4.22039270e-01
-9.46474168e-03 1.08427919e-01 -6.42809153e-01 3.96891266e-01
5.41065574e-01 8.61738622e-02 2.26671949e-01 4.33180034e-01
4.94950831e-01 -7.94999003e-02 -1.16839731e+00 8.65363777e-01
-1.61732715e-02 -1.26419020e+00 5.69372787e-04 -3.20684667e-05
5.68946540e-01 -1.88682348e-01 9.47076604e-02 2.51492471e-01
-2.45031528e-03 -6.17872596e-01 9.06541467e-01 7.09504902e-01
3.38633746e-01 -6.31134629e-01 3.83839369e-01 1.82808369e-01
-1.43000102e+00 -4.96644258e-01 -1.35568660e-02 -2.26588026e-01
2.43966147e-01 1.53953761e-01 -4.17476028e-01 6.91221178e-01
5.40524721e-01 1.29988587e+00 -6.71685696e-01 8.97318244e-01
-3.23763430e-01 3.66888314e-01 -1.92015231e-01 3.66489530e-01
5.10141850e-01 -5.86365126e-02 3.26180488e-01 1.14195001e+00
1.51300818e-01 -1.22367144e-02 4.93498683e-01 7.91432023e-01
1.47634432e-01 -2.39060715e-01 -4.44314688e-01 -1.15426511e-01
2.29451194e-01 1.27132845e+00 -5.78071237e-01 -2.21623257e-01
-7.02533722e-01 1.26338732e+00 3.17843348e-01 5.26141167e-01
-1.21201718e+00 3.33095156e-02 8.82281423e-01 -4.55971668e-03
5.79628944e-01 -3.63194108e-01 -1.37826964e-01 -1.33262515e+00
1.54088795e-01 -6.75571740e-01 4.46588367e-01 -3.93655062e-01
-9.47987199e-01 8.99071768e-02 -7.57724568e-02 -1.58681667e+00
-1.99952468e-01 -6.80394948e-01 -4.79890853e-01 5.10274768e-01
-1.82372057e+00 -1.55290675e+00 -4.93779421e-01 9.94458377e-01
4.00162339e-01 2.84850478e-01 4.89541143e-01 4.44580108e-01
-8.57792258e-01 5.17994583e-01 -1.73585877e-01 2.60143638e-01
6.97524667e-01 -1.18717122e+00 -3.05522740e-01 1.01589203e+00
1.13914907e-01 4.27472919e-01 2.75063187e-01 -2.74078995e-01
-1.76361728e+00 -1.14060485e+00 3.09877425e-01 -3.66635293e-01
7.66427517e-01 -5.10256141e-02 -5.17219901e-01 6.62657917e-01
-1.40744280e-02 3.99161816e-01 4.49028850e-01 -2.15760082e-01
-5.95719576e-01 -3.43637794e-01 -6.86742723e-01 4.28533852e-01
1.25449252e+00 -7.36513615e-01 -2.91169643e-01 2.29893073e-01
5.26251554e-01 -2.11311072e-01 -7.57569492e-01 3.54688674e-01
7.38116682e-01 -1.33013141e+00 1.19906390e+00 -7.39229977e-01
7.57039845e-01 -5.11039495e-01 -3.12293142e-01 -6.45877600e-01
-3.01147074e-01 -4.83380347e-01 -6.04464889e-01 1.18722618e+00
-9.89522934e-02 -3.80338103e-01 8.74178648e-01 5.52837133e-01
-1.18772641e-01 -7.86207139e-01 -1.02125561e+00 -7.00606763e-01
-2.89407343e-01 -4.05804664e-01 1.48733661e-01 7.68420339e-01
1.89331129e-01 8.10906067e-02 -6.20135128e-01 1.90997988e-01
5.39475024e-01 3.19146693e-01 7.27477252e-01 -6.09430969e-01
-7.30800986e-01 -6.51888669e-01 -7.76903510e-01 -1.33394790e+00
1.19640239e-01 -7.58420229e-01 9.26886350e-02 -1.24051619e+00
2.75048792e-01 -1.97555259e-01 -7.58405566e-01 7.18125463e-01
-3.65667582e-01 4.04463381e-01 4.75954175e-01 2.85971284e-01
-1.02907014e+00 7.91936994e-01 1.31988859e+00 -1.63955972e-01
-2.28314981e-01 -1.47742838e-01 -4.37455982e-01 6.69288039e-01
6.15828693e-01 -1.71212912e-01 -6.15867674e-01 -5.21401584e-01
-1.05594888e-01 7.93521702e-02 9.37664688e-01 -1.13124299e+00
3.94793481e-01 -2.44002178e-01 3.29398632e-01 -3.19751352e-01
5.61012924e-01 -7.02454090e-01 -3.29242259e-01 4.60646570e-01
-4.29716796e-01 -1.84858888e-01 -8.09865370e-02 9.29511547e-01
-2.87996262e-01 3.90596360e-01 6.18214726e-01 -2.82738507e-02
-1.11333811e+00 4.70615208e-01 7.64808729e-02 -5.05002448e-03
1.23308289e+00 -2.33373597e-01 -2.91858345e-01 -3.21518123e-01
-5.35253465e-01 1.97080806e-01 5.01564205e-01 3.92604351e-01
5.21750867e-01 -1.31788909e+00 -4.02907282e-01 9.69523042e-02
2.16443673e-01 -2.42703706e-01 3.15864682e-01 1.51438189e+00
-3.28238666e-01 3.77658159e-01 -3.47175002e-01 -6.97611213e-01
-9.76537645e-01 4.98250127e-01 3.82459819e-01 -2.43243903e-01
-7.49468923e-01 7.85824656e-01 3.78055334e-01 1.30143046e-01
4.57945704e-01 -3.33340287e-01 -9.25858691e-02 1.32801294e-01
5.57202756e-01 6.16477430e-01 -2.65939564e-01 -7.53038943e-01
-5.82950473e-01 5.13853014e-01 2.75914103e-01 -8.63694400e-02
1.14982641e+00 -3.32742423e-01 1.46993950e-01 2.97029734e-01
1.28352773e+00 -2.61290908e-01 -1.84817743e+00 -2.87325680e-01
-5.96722402e-02 -6.38641357e-01 1.31985560e-01 -4.69106644e-01
-1.37194431e+00 1.06713760e+00 4.45783645e-01 -1.38321534e-01
1.48018157e+00 1.16435550e-02 8.66810203e-01 2.18836144e-01
1.98032498e-01 -9.77212787e-01 3.00458074e-01 1.85854018e-01
4.25510406e-01 -1.35775244e+00 -9.80782043e-03 -4.05055910e-01
-5.99022985e-01 1.18654037e+00 5.77549934e-01 6.21788390e-02
2.47816280e-01 -8.00379959e-04 -2.70001680e-01 -1.06114089e-01
-7.28805721e-01 -5.86717963e-01 2.90407151e-01 4.64287400e-01
3.45079690e-01 -1.46060318e-01 -1.04357786e-01 2.88242728e-01
4.42663074e-01 6.24159612e-02 1.00738488e-01 1.18233943e+00
-2.19228908e-01 -9.47828114e-01 -4.85334285e-02 -2.38193534e-02
-1.91809297e-01 1.25289604e-01 -3.33187222e-01 8.11487913e-01
3.07588905e-01 9.28943992e-01 -5.13138957e-02 -4.39164490e-01
2.60281891e-01 -7.34811872e-02 5.98692119e-01 -2.36149132e-01
-3.74023706e-01 -3.01833060e-02 -1.98679164e-01 -1.30619121e+00
-9.20360327e-01 -6.76336825e-01 -1.39588737e+00 -2.18847290e-01
2.41127540e-03 -3.89002562e-01 2.40284905e-01 1.04095006e+00
5.27130663e-01 5.21667540e-01 6.78025842e-01 -9.68691647e-01
-5.24435818e-01 -5.22522807e-01 -4.26844925e-01 7.16166198e-01
5.35333574e-01 -9.23370063e-01 -3.77650768e-01 2.40983829e-01]
|
[8.33189582824707, 0.435358464717865]
|
868d211a-ac43-41b5-abbc-d5a6fc641e6c
|
a2dele-adaptive-and-attentive-depth-distiller
| null | null |
http://openaccess.thecvf.com/content_CVPR_2020/html/Piao_A2dele_Adaptive_and_Attentive_Depth_Distiller_for_Efficient_RGB-D_Salient_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Piao_A2dele_Adaptive_and_Attentive_Depth_Distiller_for_Efficient_RGB-D_Salient_CVPR_2020_paper.pdf
|
A2dele: Adaptive and Attentive Depth Distiller for Efficient RGB-D Salient Object Detection
|
Existing state-of-the-art RGB-D salient object detection methods explore RGB-D data relying on a two-stream architecture, in which an independent subnetwork is required to process depth data. This inevitably incurs extra computational costs and memory consumption, and using depth data during testing may hinder the practical applications of RGB-D saliency detection. To tackle these two dilemmas, we propose a depth distiller (A2dele) to explore the way of using network prediction and attention as two bridges to transfer the depth knowledge from the depth stream to the RGB stream. First, by adaptively minimizing the differences between predictions generated from the depth stream and RGB stream, we realize the desired control of pixel-wise depth knowledge transferred to the RGB stream. Second, to transfer the localization knowledge to RGB features, we encourage consistencies between the dilated prediction of the depth stream and the attention map from the RGB stream. As a result, we achieve a lightweight architecture without use of depth data at test time by embedding our A2dele. Our extensive experimental evaluation on five benchmarks demonstrate that our RGB stream achieves state-of-the-art performance, which tremendously minimizes the model size by 76% and runs 12 times faster, compared with the best performing method. Furthermore, our A2dele can be applied to existing RGB-D networks to significantly improve their efficiency while maintaining performance (boosts FPS by nearly twice for DMRA and 3 times for CPFP).
|
[' Huchuan Lu', ' Weisong Ren', ' Miao Zhang', ' Zhengkun Rong', 'Yongri Piao']
|
2020-06-01
| null | null | null |
cvpr-2020-6
|
['rgb-d-salient-object-detection']
|
['computer-vision']
|
[ 3.36942017e-01 3.73244464e-01 -4.72811945e-02 -2.07611576e-01
-3.87391537e-01 -1.73724473e-01 2.74619550e-01 -5.29399179e-02
-3.60147417e-01 3.11804920e-01 4.89643076e-03 -4.19938266e-01
2.78945833e-01 -8.95497322e-01 -8.19670260e-01 -6.29864872e-01
1.69830233e-01 -9.75932628e-02 9.71533239e-01 -2.09556252e-01
2.51670569e-01 5.76948166e-01 -1.79269457e+00 3.16819131e-01
7.18150675e-01 1.69333744e+00 4.99096125e-01 4.59511727e-01
-2.14081600e-01 6.44172132e-01 -3.14114600e-01 -1.02268368e-01
6.16498411e-01 -2.87584543e-01 -6.33888245e-01 -3.60029796e-03
3.61031383e-01 -7.16263592e-01 -6.11559451e-01 9.49896574e-01
5.48363984e-01 -1.76446177e-02 5.47409169e-02 -1.32734931e+00
-5.20463884e-01 3.35624695e-01 -9.46837664e-01 3.32524627e-01
1.86998948e-01 5.13042390e-01 6.66802824e-01 -9.74069238e-01
4.25070882e-01 1.17178524e+00 4.49751407e-01 6.82693303e-01
-9.15165067e-01 -6.67814732e-01 6.18371427e-01 1.43832952e-01
-1.03761530e+00 -1.88908041e-01 1.08289599e+00 6.73322231e-02
9.75639880e-01 1.37992427e-01 1.19813716e+00 7.29812145e-01
-2.29518905e-01 1.25041819e+00 8.95202816e-01 -1.77943036e-01
4.26528811e-01 -1.21529795e-01 -1.30447462e-01 9.20278072e-01
7.14815035e-02 2.57166922e-02 -9.27817047e-01 3.62664402e-01
1.24809659e+00 5.25708124e-02 -3.52222264e-01 -6.09633684e-01
-1.21098304e+00 5.25127888e-01 1.08618784e+00 -1.33653849e-01
-2.40691856e-01 3.50332052e-01 2.24145159e-01 3.47624198e-02
4.60656583e-01 3.01442564e-01 -5.50592005e-01 -1.27857462e-01
-8.95453870e-01 3.84240858e-02 3.23689789e-01 1.03853607e+00
9.64499831e-01 -1.41116664e-01 -1.45994956e-02 2.45264590e-01
4.60762352e-01 5.33921003e-01 4.27183032e-01 -9.55380797e-01
6.89471126e-01 1.11610067e+00 -3.18434425e-02 -8.06213021e-01
-4.67404425e-01 -3.10235173e-01 -5.87658465e-01 4.87624913e-01
4.70015764e-01 8.85780007e-02 -1.08678079e+00 1.43381274e+00
5.18985808e-01 2.71399647e-01 -1.34318665e-01 1.32511985e+00
7.83226252e-01 4.06940401e-01 -9.17004719e-02 1.51274875e-01
1.05301332e+00 -1.08366001e+00 -3.49076241e-02 -7.59373665e-01
5.38875520e-01 -2.76453942e-01 1.36796379e+00 1.10902660e-01
-1.28011096e+00 -5.69886744e-01 -1.21223330e+00 -5.53252697e-01
-2.56334454e-01 4.76613529e-02 8.91578853e-01 3.14756602e-01
-1.15010047e+00 5.48253715e-01 -1.12463796e+00 2.55688047e-03
7.71920979e-01 5.67500770e-01 -1.52733624e-01 -2.26149466e-02
-1.01636350e+00 5.17157793e-01 2.08730251e-01 2.93572694e-01
-8.28877807e-01 -9.03329909e-01 -8.52176964e-01 8.44814852e-02
1.58417344e-01 -5.89980602e-01 1.01137173e+00 -9.46493804e-01
-1.45135701e+00 7.59924114e-01 -2.46832728e-01 -2.83718437e-01
5.63721299e-01 -3.12864810e-01 1.24793939e-01 4.71696973e-01
2.03483328e-02 1.09153128e+00 7.39034295e-01 -1.07466805e+00
-1.00756705e+00 -4.94389147e-01 2.89736301e-01 3.26039225e-01
-4.73349512e-01 -5.22434294e-01 -8.29196692e-01 -2.92877465e-01
7.50839472e-01 -6.42087400e-01 -2.63389289e-01 7.78785884e-01
-3.56564611e-01 -6.63053915e-02 8.45088303e-01 -2.73902923e-01
7.59585857e-01 -2.34409332e+00 1.79340973e-01 7.93218240e-02
4.76604044e-01 1.55361503e-01 5.91655374e-02 -3.21848899e-01
1.28222734e-01 -3.82007658e-02 -2.39315078e-01 -6.71830654e-01
-2.86501229e-01 3.22812311e-02 -5.51977217e-01 4.64720011e-01
5.10704577e-01 1.06361341e+00 -1.10771406e+00 -4.91610467e-01
2.79433966e-01 4.32482004e-01 -7.10646391e-01 1.68633014e-01
-2.32134297e-01 2.63530314e-01 -6.41295135e-01 8.41312051e-01
6.13583744e-01 -4.14922655e-01 -1.58729210e-01 -2.43076101e-01
-2.34766185e-01 6.48806036e-01 -8.74827921e-01 2.06353283e+00
-2.67338216e-01 6.99879825e-01 -1.27538219e-01 -7.18080461e-01
9.28878188e-01 -2.65211165e-01 3.70054662e-01 -1.01975393e+00
2.65887678e-01 2.39869982e-01 -2.51855016e-01 -1.04399681e-01
3.63770664e-01 2.00355053e-01 1.34011835e-01 3.83092612e-01
-1.60655156e-01 -2.10389927e-01 -2.43022248e-01 6.35924190e-02
1.10336828e+00 4.08628523e-01 -2.35478058e-01 5.22775240e-02
2.20591500e-01 -1.28378230e-03 6.24095798e-01 5.61399996e-01
-4.80936348e-01 6.52187407e-01 8.28589141e-01 -5.18102169e-01
-8.56726885e-01 -1.16582036e+00 1.72495857e-01 9.46965754e-01
8.63843858e-01 3.87293696e-02 -6.01362169e-01 -8.83594215e-01
5.12384661e-02 3.55658591e-01 -8.00208926e-01 -3.96757305e-01
-6.19477451e-01 -3.78747314e-01 2.90852666e-01 9.56722617e-01
9.40855503e-01 -8.60341132e-01 -1.40972900e+00 8.13225657e-02
4.60536405e-02 -1.15003955e+00 -1.61985472e-01 5.36354482e-01
-1.14164758e+00 -9.54365075e-01 -6.68732107e-01 -8.44622433e-01
7.84557462e-01 6.80791795e-01 8.56230676e-01 8.61624405e-02
-1.98920608e-01 1.36010647e-02 -2.07551032e-01 -3.85472745e-01
3.15254807e-01 2.13457167e-01 -3.57182741e-01 -2.78564394e-01
3.34673375e-01 -7.28447378e-01 -1.17205834e+00 2.29164302e-01
-7.59792089e-01 6.27784073e-01 7.04671502e-01 5.14223814e-01
8.16243649e-01 -3.87115210e-01 2.35966131e-01 -2.48924077e-01
-1.76111966e-01 -3.12650234e-01 -7.08112895e-01 -1.02728978e-01
-4.87544775e-01 1.16699822e-02 2.17554912e-01 -4.02803808e-01
-7.58329213e-01 4.19736534e-01 9.20626614e-03 -7.02711582e-01
1.41386345e-01 -2.33058613e-02 -1.73940659e-01 -2.18271613e-01
3.65966797e-01 2.74309665e-01 1.12169705e-01 -3.53206068e-01
3.79860848e-01 3.59432012e-01 4.62996453e-01 -1.13993883e-01
7.56248415e-01 8.60372782e-01 7.25290701e-02 -3.52705568e-01
-1.00748706e+00 -3.71524870e-01 -5.90360761e-01 -2.63512731e-01
7.54864872e-01 -1.07345295e+00 -7.97922850e-01 5.45194983e-01
-1.15189886e+00 -7.59569764e-01 -4.92758751e-01 1.80596381e-01
-4.21262681e-01 -2.46055033e-02 -5.04968941e-01 -5.62824070e-01
-2.83992171e-01 -1.19477844e+00 1.41914678e+00 4.44070756e-01
1.57584488e-01 -5.81844687e-01 -4.51467216e-01 6.50391653e-02
3.51063907e-01 1.68929592e-01 7.72106230e-01 6.86597452e-02
-9.65909660e-01 4.89110267e-03 -7.24155128e-01 4.31293659e-02
2.71562189e-02 -4.66941863e-01 -1.15956008e+00 -5.56656979e-02
-5.59836328e-02 -3.49153012e-01 1.04018486e+00 1.96447492e-01
1.55430329e+00 2.35078800e-02 -4.48511332e-01 8.78755867e-01
1.29280603e+00 -1.90941676e-01 5.99890172e-01 5.66664875e-01
1.02709651e+00 4.83423442e-01 6.31736577e-01 4.17130858e-01
6.14552140e-01 4.65777099e-01 1.06510556e+00 -4.87160683e-01
-4.08289105e-01 -5.16507626e-01 1.60212830e-01 3.50437284e-01
1.33150846e-01 2.51343325e-02 -8.74208212e-01 7.14734614e-01
-1.81489205e+00 -5.59064507e-01 5.27132265e-02 2.12853026e+00
1.06320894e+00 5.15981436e-01 2.75261495e-02 3.45834523e-01
2.76800573e-01 1.24160744e-01 -9.86374974e-01 7.67953973e-03
-2.13212650e-02 2.13275895e-01 7.00622380e-01 2.84933239e-01
-9.94795978e-01 9.79942441e-01 5.69849205e+00 4.14339334e-01
-1.55203259e+00 -4.51966301e-02 8.44106853e-01 -8.36273789e-01
-4.02152002e-01 -5.59624434e-02 -8.89488876e-01 4.67765629e-01
5.02150774e-01 2.33354568e-01 2.92843014e-01 1.09581864e+00
1.36080340e-01 -3.58820379e-01 -1.14333451e+00 1.10913336e+00
-1.15145348e-01 -1.33934104e+00 -1.12373680e-01 1.69741847e-02
7.64507115e-01 2.97429889e-01 1.30401090e-01 1.21617236e-03
-9.21568647e-03 -6.51588678e-01 1.11730182e+00 2.23330811e-01
7.20481277e-01 -7.59113491e-01 5.35462558e-01 1.72151893e-01
-1.15874326e+00 -2.10097179e-01 -5.79809368e-01 -2.81417310e-01
5.98105788e-02 8.52473021e-01 -5.86607575e-01 1.14164397e-01
1.05500698e+00 7.80575097e-01 -6.80896521e-01 1.04297566e+00
-5.24469674e-01 1.90900385e-01 -4.72791642e-01 -1.70566335e-01
4.00969952e-01 2.51317352e-01 3.16367537e-01 7.44425595e-01
1.84989765e-01 -4.19343298e-04 7.89452717e-03 1.09318209e+00
-1.78057492e-01 -4.08591807e-01 -3.24717969e-01 4.33295012e-01
4.88486856e-01 1.00154483e+00 -9.27776039e-01 -2.55534142e-01
-3.27442050e-01 1.17187214e+00 4.45675164e-01 2.90924579e-01
-1.07740819e+00 -5.15808642e-01 8.88190866e-01 3.38374674e-01
6.01997316e-01 -1.13944724e-01 -7.82111049e-01 -8.24295044e-01
2.92145997e-01 -2.66633511e-01 -1.41032070e-01 -9.39499140e-01
-7.98268497e-01 5.34578383e-01 -3.44760358e-01 -1.32632375e+00
1.99287623e-01 -6.37072384e-01 -3.58148605e-01 9.54038680e-01
-2.03245687e+00 -1.07594693e+00 -8.41146946e-01 5.78612149e-01
4.00206685e-01 4.31892276e-01 3.37467849e-01 1.56441525e-01
-4.41753298e-01 5.48404038e-01 -5.08372545e-01 2.41955221e-02
3.39695811e-01 -1.19354260e+00 7.77641654e-01 8.93816113e-01
-2.04758674e-01 2.85151541e-01 2.93125927e-01 -4.19118226e-01
-1.57254231e+00 -1.21243906e+00 4.22062635e-01 -3.04091692e-01
4.86415833e-01 -5.27438462e-01 -7.81059206e-01 3.60017061e-01
-3.89758587e-01 5.06089330e-01 1.92824617e-01 -3.11185688e-01
-3.35672230e-01 -3.06176364e-01 -9.63093102e-01 7.11302698e-01
1.45423985e+00 -5.68386257e-01 -3.47155988e-01 -9.03403852e-03
1.23069263e+00 -7.51604080e-01 -5.31434119e-01 3.24068248e-01
5.56252837e-01 -1.29152679e+00 1.07179141e+00 -1.43259674e-01
8.29934180e-01 -5.20375252e-01 -8.39595199e-02 -9.52969313e-01
-8.72347951e-02 -3.84075552e-01 -4.71218795e-01 9.29567933e-01
3.98715705e-01 -5.12985885e-01 1.13747966e+00 7.47363806e-01
-4.05727863e-01 -1.30173576e+00 -1.09033799e+00 -3.43661636e-01
-2.20242918e-01 -6.37001455e-01 7.15224981e-01 4.84684229e-01
-7.23687634e-02 -4.76475134e-02 1.28804758e-01 2.98892587e-01
5.73733568e-01 3.08529288e-01 6.32309437e-01 -9.90355313e-01
-2.64874130e-01 -6.74974978e-01 -6.10215306e-01 -1.83964837e+00
-2.23249316e-01 -5.65617561e-01 2.92861551e-01 -1.62470841e+00
-1.33984685e-01 -7.38062322e-01 -3.29811394e-01 8.25299799e-01
-3.34760636e-01 4.99506295e-01 2.09982723e-01 1.58450261e-01
-5.85727751e-01 7.49735117e-01 1.58495033e+00 -1.02619939e-01
-5.00185728e-01 -2.51018345e-01 -8.21275651e-01 8.63299549e-01
6.04266822e-01 -2.59508401e-01 -6.15503728e-01 -7.64487326e-01
2.64939576e-01 -1.85593233e-01 6.68396652e-01 -1.06642520e+00
3.70909333e-01 -1.98581219e-02 7.80311465e-01 -7.53214836e-01
4.85560387e-01 -5.90123415e-01 -6.25724673e-01 3.67787182e-01
-8.30335617e-02 -2.40533561e-01 4.18661833e-01 5.54831564e-01
-4.62743156e-02 2.72267997e-01 8.03821743e-01 -1.18179232e-01
-1.02962530e+00 4.28343683e-01 1.82666734e-01 8.97168461e-03
1.07619095e+00 -6.19260311e-01 -4.66511190e-01 -3.08890804e-03
-2.95631319e-01 3.79371077e-01 7.01846302e-01 4.68023926e-01
1.00609875e+00 -1.16906190e+00 -1.25657350e-01 6.21118128e-01
-1.37961656e-02 7.65982866e-01 1.13235578e-01 9.69123721e-01
-5.64767778e-01 2.22304851e-01 -2.28078514e-01 -8.79778683e-01
-7.38223732e-01 4.05098051e-01 3.11980277e-01 2.43180245e-02
-8.78108501e-01 1.29677153e+00 4.06048656e-01 -1.54235831e-03
4.77259248e-01 -6.26165450e-01 2.06337169e-01 -1.23263404e-01
6.20236039e-01 2.62316972e-01 6.91337213e-02 -9.35847536e-02
-5.37105441e-01 5.58978558e-01 8.99041742e-02 -4.97655272e-02
1.40756154e+00 -2.20975727e-01 -2.75637209e-02 3.52062583e-01
1.15105784e+00 -3.54871899e-01 -2.12607503e+00 -3.09200846e-02
-3.01849872e-01 -5.59548318e-01 4.56499577e-01 -5.42267859e-01
-1.45624769e+00 9.34212029e-01 7.10819483e-01 3.53106819e-02
1.43847108e+00 2.02756554e-01 9.45236564e-01 5.56600578e-02
4.85495716e-01 -8.49438846e-01 2.82285213e-01 2.85275877e-01
7.21993208e-01 -1.30593634e+00 -5.93832023e-02 -5.69861829e-01
-4.87904847e-01 9.18644428e-01 1.00120795e+00 -1.74930438e-01
5.44751167e-01 2.47089654e-01 -8.87041166e-02 -1.98078722e-01
-5.67973793e-01 -2.94862509e-01 8.83657560e-02 6.13449275e-01
-1.51926372e-02 -2.62667239e-01 4.06632513e-01 4.03473824e-01
-1.10308133e-01 -9.72016603e-02 2.37049088e-01 1.07833147e+00
-5.80361128e-01 -6.60883367e-01 -3.17717940e-02 3.90552938e-01
-1.08404653e-02 -2.20916227e-01 -3.37743312e-01 7.87384689e-01
2.18002394e-01 7.10553706e-01 2.84478098e-01 -4.63917911e-01
4.30448264e-01 -2.24438921e-01 3.69493216e-01 -5.36578536e-01
-3.94145250e-01 -6.57075644e-02 -3.42786461e-01 -1.21532536e+00
-3.66338134e-01 -3.42627764e-01 -1.59497106e+00 -2.13378653e-01
-3.47083747e-01 -4.97674108e-01 7.43041396e-01 8.89040768e-01
6.11611307e-01 7.28738248e-01 6.82413697e-01 -1.29858422e+00
-2.54426390e-01 -6.41052067e-01 -2.72347510e-01 -3.24558653e-03
5.69436789e-01 -7.27748394e-01 -4.05004442e-01 -2.50786811e-01]
|
[9.648853302001953, -0.8406720757484436]
|
dd630f06-4ed9-421b-b8a9-d29b65bcd843
|
visual-chirality-1
|
2006.09512
| null |
https://arxiv.org/abs/2006.09512v1
|
https://arxiv.org/pdf/2006.09512v1.pdf
|
Visual Chirality
|
How can we tell whether an image has been mirrored? While we understand the geometry of mirror reflections very well, less has been said about how it affects distributions of imagery at scale, despite widespread use for data augmentation in computer vision. In this paper, we investigate how the statistics of visual data are changed by reflection. We refer to these changes as "visual chirality", after the concept of geometric chirality - the notion of objects that are distinct from their mirror image. Our analysis of visual chirality reveals surprising results, including low-level chiral signals pervading imagery stemming from image processing in cameras, to the ability to discover visual chirality in images of people and faces. Our work has implications for data augmentation, self-supervised learning, and image forensics.
|
['Abe Davis', 'Zhiqiu Lin', 'Jin Sun', 'Noah Snavely']
|
2020-06-16
|
visual-chirality
|
http://openaccess.thecvf.com/content_CVPR_2020/html/Lin_Visual_Chirality_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Lin_Visual_Chirality_CVPR_2020_paper.pdf
|
cvpr-2020-6
|
['image-forensics']
|
['computer-vision']
|
[ 8.09553266e-01 1.42454207e-01 3.86700071e-02 -4.09980625e-01
1.74763575e-01 -6.67520702e-01 1.21029353e+00 -1.67858839e-01
-2.69103140e-01 2.30581731e-01 7.67556012e-01 -3.66204083e-01
6.07878938e-02 -5.28674841e-01 -7.66983986e-01 -6.89882398e-01
-1.77414745e-01 2.24548906e-01 -1.70262128e-01 -2.25473836e-01
6.20111644e-01 4.67601985e-01 -1.68410385e+00 5.50587475e-01
2.78422892e-01 4.22314733e-01 -1.63707569e-01 6.09787881e-01
2.43103057e-01 6.86802864e-01 -2.09179282e-01 -5.40467143e-01
5.47934830e-01 -5.74934661e-01 -6.69681549e-01 4.54657286e-01
1.25965619e+00 -3.56920481e-01 -4.88972664e-01 1.36321676e+00
1.88889652e-01 -4.23714846e-01 8.13811839e-01 -1.22474694e+00
-1.06747568e+00 2.15127051e-01 -1.01582158e+00 2.55491376e-01
1.22329205e-01 1.96688816e-01 9.91810501e-01 -7.15852380e-01
1.02899361e+00 1.28849196e+00 7.23062396e-01 5.31428337e-01
-1.54272521e+00 -5.87760568e-01 -4.24649656e-01 2.06244603e-01
-9.96833563e-01 -8.34344447e-01 6.24159575e-01 -9.80817676e-01
4.20819908e-01 2.64485657e-01 6.89188778e-01 9.49874222e-01
8.82512033e-02 2.13092193e-01 1.59515464e+00 -5.58237731e-01
-3.16309750e-01 1.21449623e-02 -2.95286715e-01 6.14454091e-01
5.24391115e-01 3.66241038e-01 -8.44442785e-01 -1.49605006e-01
7.94863820e-01 -2.13424757e-01 -1.96208805e-01 -4.96489882e-01
-1.23287380e+00 5.70111394e-01 1.80662215e-01 -5.53100370e-02
-7.06424117e-02 -4.57723327e-02 1.90909535e-01 1.90326259e-01
3.58862847e-01 7.08837152e-01 -4.97858636e-02 1.45175442e-01
-6.84935629e-01 1.10966787e-01 3.76716167e-01 6.13268256e-01
7.01937258e-01 1.82307109e-01 2.52441794e-01 6.70557797e-01
1.11365736e-01 7.80138135e-01 1.70789242e-01 -1.07073259e+00
1.46230400e-01 4.23980027e-01 -2.98868716e-01 -1.30340695e+00
-4.04538959e-01 -9.80125591e-02 -9.24843371e-01 5.24824798e-01
5.81612349e-01 2.57679343e-01 -7.16461003e-01 1.79268754e+00
7.30374502e-03 7.29003176e-02 -4.40545343e-02 8.94162595e-01
6.96660936e-01 7.68735483e-02 -3.95992130e-01 6.64811730e-02
1.53956592e+00 -4.24025841e-02 -2.16754019e-01 -2.81706870e-01
1.39309078e-01 -1.14509702e+00 9.49114561e-01 4.58254099e-01
-7.25129247e-01 -2.67068803e-01 -1.11712122e+00 4.77876365e-02
-1.81290135e-01 -1.18011303e-01 7.32166767e-01 8.66781235e-01
-8.24317873e-01 5.71594417e-01 -1.40109256e-01 -6.23992443e-01
6.66796565e-01 1.88324332e-01 -6.99795902e-01 -5.49051799e-02
-4.45749760e-01 7.45816648e-01 -1.06677182e-01 -2.88429797e-01
-6.65161669e-01 -6.81406617e-01 -5.50408840e-01 -6.09957576e-01
4.00670208e-02 -5.91143489e-01 8.12179267e-01 -9.92437065e-01
-7.69835770e-01 1.55555904e+00 -2.80900121e-01 -2.37235755e-01
2.45017052e-01 2.24069968e-01 -5.61457813e-01 1.79303959e-01
6.26342073e-02 5.36730945e-01 1.33940911e+00 -1.63421822e+00
-3.78811151e-01 -8.15542817e-01 -2.37585753e-01 1.15994617e-01
-3.36323619e-01 -6.65477365e-02 -3.45895290e-02 -5.05095065e-01
5.63201308e-01 -1.07956731e+00 1.00124747e-01 -1.27425030e-01
-4.99567509e-01 5.53167045e-01 8.48490357e-01 -6.23123050e-01
6.94263935e-01 -2.06187844e+00 -4.34627473e-01 4.25152153e-01
6.08371735e-01 -2.81804167e-02 -4.03663546e-01 5.04300416e-01
-4.96231556e-01 3.52154523e-01 -1.79436039e-02 3.62201095e-01
-4.03956294e-01 5.64865321e-02 -5.87436140e-01 1.09441936e+00
3.24006736e-01 8.39692593e-01 -7.36054420e-01 -2.59095758e-01
1.92936748e-01 3.16155225e-01 -7.94772685e-01 -2.40931079e-01
1.02302574e-01 7.52710462e-01 2.20088884e-01 6.70768857e-01
1.12538898e+00 -1.26684874e-01 3.91493618e-01 -6.10146821e-01
-3.14491481e-01 2.53607184e-01 -7.19813466e-01 8.93619418e-01
1.96915939e-01 1.14545572e+00 -3.97330895e-02 -7.51250148e-01
7.93701947e-01 -2.94620037e-01 6.28533542e-01 -1.26924419e+00
-2.26702020e-01 -1.28149167e-01 4.31752890e-01 -6.12888813e-01
9.10856187e-01 -5.18633246e-01 3.42310667e-01 6.01005137e-01
-1.62282780e-01 -2.65404254e-01 8.11413452e-02 2.12445810e-01
9.27631915e-01 -1.56854652e-02 1.71548828e-01 -2.70983875e-01
1.74551010e-02 1.30440980e-01 2.44819567e-01 6.58215165e-01
6.90663308e-02 8.59839380e-01 6.07985854e-01 -3.48802745e-01
-1.82492626e+00 -1.39214110e+00 -5.46435714e-01 8.06106389e-01
1.13270074e-01 -5.70879042e-01 -4.80856836e-01 8.74339789e-02
1.56449556e-01 7.00525045e-02 -7.42963254e-01 -4.57213372e-01
-4.45360899e-01 -1.21719849e+00 8.90516043e-01 -9.20953900e-02
4.12061691e-01 -7.65244722e-01 -5.95193684e-01 -4.65301991e-01
-1.59691200e-01 -1.34674692e+00 -3.12697053e-01 -1.70977890e-01
-9.70399618e-01 -1.59000647e+00 -4.05614346e-01 -2.92108327e-01
8.79026175e-01 6.22325480e-01 1.27077878e+00 4.26990315e-02
-7.87925899e-01 7.50267267e-01 -3.99209291e-01 -4.50937986e-01
-6.40383959e-01 -5.57740986e-01 3.74418259e-01 1.28373042e-01
4.86792386e-01 -5.95983148e-01 -6.11664057e-01 2.45707124e-01
-9.54978585e-01 -3.66369002e-02 7.52101541e-01 6.77874386e-01
1.58850998e-01 -1.27785310e-01 -2.95159161e-01 -1.00461018e+00
7.10377157e-01 -2.27768824e-01 -3.70533556e-01 4.52889837e-02
-6.04561150e-01 3.57050270e-01 2.39665300e-01 -1.12612464e-01
-7.98077404e-01 -1.64546758e-01 5.01196325e-01 -1.03252567e-01
-1.42882988e-01 1.17067352e-01 1.35261700e-01 -3.29291165e-01
1.03110349e+00 2.60874212e-01 6.04125679e-01 -2.30343327e-01
5.00898719e-01 4.88699645e-01 1.03937018e+00 -5.91883481e-01
1.09212172e+00 1.20966446e+00 4.40997332e-01 -1.84586251e+00
-5.52711070e-01 -4.32412505e-01 -7.96261430e-01 -2.50154406e-01
8.35481107e-01 -7.68869281e-01 -9.44772720e-01 4.38344538e-01
-1.01821065e+00 5.52654564e-02 -4.40357029e-01 3.65610778e-01
-5.26594102e-01 7.29004681e-01 -2.00625047e-01 -7.52943993e-01
2.31604025e-01 -6.98520362e-01 7.57856309e-01 3.76617834e-02
-4.29883420e-01 -6.71772778e-01 3.89714509e-01 6.07352555e-01
1.20680807e-02 1.96638748e-01 1.16676688e+00 -2.30916347e-02
-5.46246350e-01 1.61877692e-01 -4.80389386e-01 4.45560962e-01
1.69305891e-01 2.03345105e-01 -1.38197625e+00 -8.54819342e-02
-2.35281438e-01 -1.84638798e-01 9.35682952e-01 2.63370067e-01
7.96796978e-01 -5.03275752e-01 2.35308483e-02 8.46948624e-01
1.15857112e+00 -9.00477245e-02 1.11567867e+00 4.90333766e-01
7.44443059e-01 1.11131096e+00 1.26860976e-01 6.19149745e-01
-4.11239564e-02 4.89728987e-01 6.53427720e-01 -1.34317115e-01
-2.80266047e-01 -3.64022672e-01 4.03595805e-01 4.45657998e-01
-5.41926205e-01 3.06917250e-01 -8.76620352e-01 1.79473460e-01
-1.20731354e+00 -1.54328978e+00 -5.85483909e-01 2.50111771e+00
4.78317350e-01 -2.96918362e-01 1.72860697e-01 -5.87942153e-02
6.18932724e-01 3.41529399e-01 -4.92777497e-01 -3.18680793e-01
-7.41033077e-01 1.19015304e-02 8.53092253e-01 2.51903534e-01
-1.02267480e+00 7.29255259e-01 7.71842194e+00 1.58192113e-01
-1.17113948e+00 -2.91270405e-01 4.88251120e-01 1.96094718e-03
-4.70092922e-01 4.17205215e-01 -3.34732682e-01 8.43418911e-02
3.05168003e-01 1.09831937e-01 7.14162588e-01 3.43100667e-01
2.80212820e-01 -1.65353879e-01 -9.99306917e-01 1.22526622e+00
4.94997054e-01 -1.34756649e+00 3.50463271e-01 7.68541574e-01
7.31772184e-01 2.28844225e-01 7.03348756e-01 -6.00255430e-01
2.46889010e-01 -1.29857945e+00 7.16990113e-01 5.75073898e-01
9.44957495e-01 -4.14762855e-01 1.72001749e-01 -3.28813680e-02
-7.53583908e-01 -1.11368679e-01 -4.94214088e-01 -1.26819968e-01
-2.07750484e-01 5.82711935e-01 -1.19291508e+00 4.04249653e-02
7.54051864e-01 5.98827839e-01 -1.07938945e+00 8.10836613e-01
1.92614272e-01 2.79314190e-01 -3.96522582e-02 4.70180720e-01
-1.05701134e-01 -7.99080729e-01 6.64964199e-01 8.97250354e-01
1.30559087e-01 -9.95315239e-02 -4.49238420e-01 8.84773493e-01
-2.42207106e-02 -1.74638063e-01 -1.23879981e+00 -4.74244326e-01
1.51976526e-01 1.28895378e+00 -6.49601221e-01 -1.92418955e-02
-4.61476892e-01 9.69160736e-01 -9.88622457e-02 2.45686501e-01
-2.12475389e-01 2.02992961e-01 1.05267060e+00 7.72031724e-01
-9.54300724e-03 -4.37149197e-01 -3.60086650e-01 -1.07484412e+00
1.83711767e-01 -1.18478894e+00 -9.31275412e-02 -1.05153608e+00
-1.49810374e+00 6.55303970e-02 -2.43306816e-01 -1.31131983e+00
2.35195439e-02 -1.02313232e+00 -6.13689065e-01 4.60542411e-01
-8.57430935e-01 -1.03405225e+00 -3.89794439e-01 5.34830153e-01
-3.73952359e-01 -4.49164808e-01 5.77397645e-01 1.73326936e-02
1.34029076e-01 2.67452180e-01 4.56005126e-01 2.17616960e-01
1.03646278e+00 -1.09611225e+00 3.91562819e-01 7.19358385e-01
6.35110140e-01 7.55218387e-01 9.17732775e-01 -7.11780012e-01
-1.45413673e+00 -5.65287769e-01 4.21280593e-01 -7.83241391e-01
5.74687958e-01 -2.27419987e-01 -4.66878086e-01 4.32765782e-01
2.48370320e-01 -1.95716754e-01 7.37038076e-01 2.99507052e-01
-1.03338695e+00 3.82322036e-02 -8.60091507e-01 9.69215333e-01
1.29197609e+00 -9.00033951e-01 -3.42617989e-01 1.51958093e-01
-7.79935569e-02 2.33646035e-01 -4.43628252e-01 4.21488196e-01
8.80310476e-01 -1.46341026e+00 1.26814008e+00 -1.01627898e+00
6.13126338e-01 -2.84671336e-01 -4.22804564e-01 -1.16327405e+00
-2.18752071e-01 -4.17223543e-01 5.46310544e-01 1.00245595e+00
1.39220193e-01 -4.31636363e-01 1.07985640e+00 3.44170779e-01
1.55693948e-01 -2.29742564e-02 -7.67676592e-01 -6.97756171e-01
2.11075336e-01 -3.74915332e-01 3.27918589e-01 1.41512203e+00
-4.07640189e-02 4.03485864e-01 -5.24414718e-01 -2.42301524e-02
1.03857172e+00 -6.08579442e-02 1.11639881e+00 -1.46898234e+00
-1.00077018e-01 -5.85419774e-01 -7.49257863e-01 -6.48460865e-01
-1.69292569e-01 -1.11763012e+00 -2.73920149e-01 -8.18173230e-01
8.50311399e-01 -1.64347708e-01 3.84082764e-01 8.75978097e-02
4.10591274e-01 8.08603764e-01 4.01077151e-01 4.75676417e-01
-1.81860477e-01 3.24248970e-01 1.42713630e+00 -2.74135709e-01
1.67713895e-01 -4.59521502e-01 -9.88114357e-01 1.11470640e+00
6.68957531e-01 -3.46166998e-01 -1.05509251e-01 -1.74977109e-01
9.35832202e-01 -6.64124906e-01 1.01166606e+00 -9.11786795e-01
-7.91453570e-02 -1.39364421e-01 7.30900109e-01 -4.44254786e-01
2.79769897e-01 -6.47555828e-01 -4.22692858e-02 2.69640684e-01
-1.30595729e-01 3.28725845e-01 -2.39957981e-02 4.71811920e-01
-1.00368887e-01 -2.14925393e-01 8.33350480e-01 -3.98530632e-01
-7.53950477e-01 1.77810550e-01 -5.95767438e-01 4.59180415e-01
6.76167905e-01 -4.26972836e-01 -9.52538729e-01 -5.23488879e-01
-4.79255050e-01 -3.66233528e-01 9.64290619e-01 4.07236487e-01
7.06352890e-01 -1.41327894e+00 -7.27007985e-01 4.55531031e-01
4.29319054e-01 -6.92152083e-01 2.10829675e-01 9.38792646e-01
-5.65652847e-01 2.09312141e-01 -5.23918271e-01 -7.62194276e-01
-1.67558181e+00 4.07226205e-01 1.70234784e-01 4.24117327e-01
-6.06198072e-01 4.33332264e-01 4.57458586e-01 -5.68898618e-01
-4.21400040e-01 9.44985077e-02 9.58020519e-03 3.86396974e-01
3.68038893e-01 4.79034901e-01 -1.24098420e-01 -1.12540781e+00
-2.76732564e-01 9.43768919e-01 1.18684098e-01 -5.44491649e-01
1.27941191e+00 -9.98220891e-02 -6.26505554e-01 4.29875851e-01
1.21973014e+00 4.26673353e-01 -1.12143314e+00 -3.68428975e-02
-1.45609574e-02 -7.72994697e-01 -3.98023427e-01 -5.07951021e-01
-7.13479400e-01 9.13541436e-01 7.18408227e-01 2.80322522e-01
6.95922554e-01 2.02019051e-01 -4.02956866e-02 4.15746450e-01
1.52073309e-01 -1.08511925e+00 4.28470790e-01 6.14618421e-01
9.59570110e-01 -1.32721233e+00 4.66637969e-01 -2.44249254e-01
-8.05288076e-01 1.16841280e+00 3.11191618e-01 -1.36681020e-01
5.97895920e-01 2.39248499e-01 1.50192931e-01 -4.23127234e-01
-3.03182602e-01 -3.66082937e-01 3.36435169e-01 1.04697740e+00
5.05854666e-01 -4.98107728e-03 2.45059624e-01 -2.13042244e-01
-7.64540732e-01 -3.25506091e-01 9.42650855e-01 4.97908771e-01
-3.99122924e-01 -1.10035098e+00 -9.59721446e-01 6.69565797e-01
-3.27008188e-01 -4.43353832e-01 -1.11167133e+00 8.24450433e-01
3.85233939e-01 7.97679484e-01 5.38026273e-01 -8.67170393e-01
2.04734374e-02 1.92093533e-02 1.05258369e+00 -5.10391891e-01
1.11212479e-02 -1.73886478e-01 1.71754092e-01 -4.77086306e-01
-8.15078437e-01 -1.00171661e+00 -7.78215468e-01 -6.58573389e-01
8.81626457e-02 -3.43586236e-01 7.92587340e-01 6.70979619e-01
2.06910446e-01 6.72154799e-02 3.90224874e-01 -6.85283720e-01
-3.07862699e-01 -7.96390772e-01 -8.21933925e-01 7.53417850e-01
4.35755938e-01 -3.73028070e-01 -3.85520428e-01 4.62736726e-01]
|
[11.500053405761719, 0.7838843464851379]
|
0033e3d6-f9d8-4726-b7c9-0e8e400594e8
|
191111236
|
1911.11236
| null |
https://arxiv.org/abs/1911.11236v3
|
https://arxiv.org/pdf/1911.11236v3.pdf
|
RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds
|
We study the problem of efficient semantic segmentation for large-scale 3D point clouds. By relying on expensive sampling techniques or computationally heavy pre/post-processing steps, most existing approaches are only able to be trained and operate over small-scale point clouds. In this paper, we introduce RandLA-Net, an efficient and lightweight neural architecture to directly infer per-point semantics for large-scale point clouds. The key to our approach is to use random point sampling instead of more complex point selection approaches. Although remarkably computation and memory efficient, random sampling can discard key features by chance. To overcome this, we introduce a novel local feature aggregation module to progressively increase the receptive field for each 3D point, thereby effectively preserving geometric details. Extensive experiments show that our RandLA-Net can process 1 million points in a single pass with up to 200X faster than existing approaches. Moreover, our RandLA-Net clearly surpasses state-of-the-art approaches for semantic segmentation on two large-scale benchmarks Semantic3D and SemanticKITTI.
|
['Andrew Markham', 'Zhihua Wang', 'Niki Trigoni', 'Bo Yang', 'Stefano Rosa', 'Yulan Guo', 'Qingyong Hu', 'Linhai Xie']
|
2019-11-25
|
randla-net-efficient-semantic-segmentation-of
|
http://openaccess.thecvf.com/content_CVPR_2020/html/Hu_RandLA-Net_Efficient_Semantic_Segmentation_of_Large-Scale_Point_Clouds_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Hu_RandLA-Net_Efficient_Semantic_Segmentation_of_Large-Scale_Point_Clouds_CVPR_2020_paper.pdf
|
cvpr-2020-6
|
['lidar-semantic-segmentation']
|
['computer-vision']
|
[ 1.19688243e-01 -2.78986245e-02 7.62692988e-02 -4.60734546e-01
-9.07447577e-01 -6.09905958e-01 4.45321560e-01 4.00039673e-01
-5.84926426e-01 2.10730523e-01 -3.33429486e-01 -3.37403804e-01
1.54244587e-01 -1.13062024e+00 -1.15266788e+00 -3.01440686e-01
5.91978319e-02 8.99713814e-01 8.62618327e-01 -8.59337598e-02
4.34153855e-01 1.15014529e+00 -1.83948755e+00 6.25334829e-02
7.20313013e-01 1.14422417e+00 3.51963341e-01 5.74869752e-01
-6.26109600e-01 1.40476832e-02 -4.38092798e-01 -1.61222667e-01
5.61440110e-01 2.32499599e-01 -9.84755278e-01 2.93342695e-02
9.02715266e-01 -4.54150558e-01 4.33500558e-02 1.03045869e+00
4.24421489e-01 2.99225956e-01 4.07909065e-01 -1.06279349e+00
-3.29959244e-01 2.81973928e-01 -5.91008425e-01 -2.69790571e-02
4.16014381e-02 4.47043404e-02 9.10884440e-01 -1.11272240e+00
5.13148427e-01 1.42357814e+00 1.10018325e+00 3.59826535e-01
-1.21865773e+00 -4.89450514e-01 2.67696589e-01 -3.34571153e-01
-1.46480024e+00 -9.14455801e-02 8.10355902e-01 -9.40018743e-02
1.22100568e+00 3.00844789e-01 8.50393593e-01 3.32252383e-01
-2.91071653e-01 8.59710813e-01 7.67475128e-01 -1.31746575e-01
4.30018157e-01 -3.48081410e-01 1.33055449e-01 7.11184740e-01
2.35888347e-01 -1.67541325e-01 -2.63023227e-01 -3.88460904e-01
1.14170790e+00 3.70240152e-01 5.74771315e-02 -8.07848990e-01
-1.37703395e+00 8.01057458e-01 8.85900736e-01 7.83398226e-02
-4.20748353e-01 5.32265723e-01 3.11911941e-01 1.87752396e-01
7.55724967e-01 4.08266246e-01 -6.70636177e-01 8.62933770e-02
-1.37602246e+00 5.42190313e-01 5.62332392e-01 1.07599115e+00
1.14895785e+00 -2.66474962e-01 1.50722578e-01 7.08746850e-01
1.44047409e-01 6.24453962e-01 1.65038127e-02 -1.28029776e+00
2.70320266e-01 9.04764831e-01 1.53618768e-01 -8.38705361e-01
-5.30004084e-01 -1.45950973e-01 -5.87823153e-01 5.30297875e-01
3.26899171e-01 3.57800186e-01 -1.42913127e+00 1.05852747e+00
6.23637974e-01 3.82004201e-01 -2.99062729e-01 8.68847370e-01
7.84374535e-01 6.20963931e-01 -1.09646711e-02 4.20639694e-01
1.23292351e+00 -9.99920070e-01 1.95970923e-01 -2.39686668e-01
5.46212792e-01 -6.92851841e-01 1.23454392e+00 1.84490398e-01
-1.33974838e+00 -4.05845761e-01 -9.42404687e-01 -5.18630803e-01
-4.86549526e-01 -1.72670767e-01 9.84427929e-01 4.61320966e-01
-1.12218678e+00 1.01475763e+00 -1.21093512e+00 -3.49205971e-01
1.02291417e+00 6.98261082e-01 -1.50267199e-01 -1.49819925e-01
-3.62810045e-01 3.35811764e-01 3.68377000e-01 -1.21342272e-01
-5.56720436e-01 -1.07855678e+00 -7.46059954e-01 1.12525381e-01
4.61457729e-01 -9.81941342e-01 1.35196936e+00 -4.87404197e-01
-1.30884433e+00 1.09287870e+00 -2.89341748e-01 -5.66631615e-01
4.71240848e-01 -5.92477441e-01 3.70439202e-01 4.31735784e-01
2.49405906e-01 1.21700323e+00 7.68219411e-01 -1.47067928e+00
-7.35298038e-01 -5.83780408e-01 -4.36798064e-03 2.47277394e-01
8.13157633e-02 -1.05831362e-01 -8.37461650e-01 -4.95112211e-01
6.11098886e-01 -8.98656309e-01 -6.94526494e-01 3.42294037e-01
-3.60297412e-01 -4.98439580e-01 9.20302153e-01 -7.41391033e-02
3.77340823e-01 -2.07213545e+00 -1.88776121e-01 3.00486177e-01
4.48024839e-01 2.90234387e-01 2.60022990e-02 -2.53490657e-02
3.14107627e-01 2.49863788e-01 -5.91420114e-01 -7.31548309e-01
1.07154094e-01 2.75278002e-01 -4.24479306e-01 4.50728714e-01
2.23864987e-01 1.13750565e+00 -8.77764106e-01 -4.53741759e-01
6.95755601e-01 6.13918960e-01 -8.04524481e-01 -1.91203341e-01
-6.12291515e-01 1.33533254e-01 -7.01055825e-01 8.63029361e-01
1.11206639e+00 -5.63477695e-01 -5.19829273e-01 -1.51325583e-01
-1.05974674e-01 4.03827637e-01 -1.00666714e+00 2.37657118e+00
-4.39810663e-01 3.17002177e-01 4.32129093e-02 -8.29568088e-01
9.02982712e-01 -1.49557978e-01 6.79395437e-01 -3.41226846e-01
1.59268767e-01 4.93632823e-01 -7.19744444e-01 3.82432848e-01
7.43797779e-01 8.49552974e-02 -7.16297776e-02 2.98713863e-01
-9.69625190e-02 -8.62746239e-01 -1.90756172e-01 1.15928575e-01
1.02938306e+00 2.39160284e-01 -6.26290068e-02 -3.04113746e-01
2.62340963e-01 4.50926244e-01 2.56438166e-01 9.86666977e-01
-5.00289723e-03 9.31241453e-01 2.07344353e-01 -7.86974192e-01
-1.13302851e+00 -1.20115960e+00 -1.46041632e-01 9.67539251e-01
5.08203626e-01 -2.91031539e-01 -7.62536108e-01 -7.87262619e-01
4.18526232e-01 5.64016640e-01 -2.46019736e-01 2.55123347e-01
-6.95216477e-01 -4.55954313e-01 3.31594259e-01 7.26139545e-01
4.83590811e-01 -9.87391055e-01 -7.65774488e-01 2.42885962e-01
3.02817792e-01 -1.16676152e+00 -2.16767788e-01 3.00370842e-01
-1.45883942e+00 -7.26570666e-01 -7.19141722e-01 -7.75769591e-01
8.45159948e-01 6.01354182e-01 1.60405588e+00 2.65804589e-01
-1.31435260e-01 8.95419270e-02 -2.25304767e-01 -4.05251622e-01
6.33885041e-02 4.29791600e-01 -3.00672382e-01 -6.38034701e-01
7.67586291e-01 -7.64391661e-01 -8.33096385e-01 2.56788075e-01
-8.05774093e-01 -2.21298821e-02 6.30582988e-01 5.03494501e-01
1.13810635e+00 -3.89495976e-02 5.59782721e-02 -8.06173980e-01
1.50377080e-01 -5.84572926e-02 -7.82753408e-01 -7.85200372e-02
-2.50541091e-01 -2.44718092e-03 3.95755798e-01 -1.03328027e-01
-5.09387612e-01 4.41179723e-01 -5.84134579e-01 -8.55772197e-01
-4.41841960e-01 -8.74201581e-02 2.21668575e-02 -5.60858369e-01
3.74675006e-01 1.38980120e-01 -1.87290207e-01 -6.98095679e-01
5.79880714e-01 3.34274948e-01 5.56724966e-01 -7.26909161e-01
9.16940749e-01 1.14442253e+00 1.17481232e-01 -9.04826462e-01
-8.13924313e-01 -6.79627419e-01 -9.41996753e-01 2.90080339e-01
9.15616035e-01 -9.80433166e-01 -7.85189152e-01 5.51832080e-01
-1.26520693e+00 -3.14776719e-01 -6.10446632e-01 1.17616773e-01
-6.80624187e-01 2.40590021e-01 -5.39197564e-01 -4.74834532e-01
-4.97653723e-01 -1.12756550e+00 1.88027620e+00 1.43910553e-02
5.35207950e-02 -6.65366948e-01 -2.11703554e-01 2.44097456e-01
2.46144399e-01 1.89592287e-01 6.22386456e-01 -5.96816897e-01
-1.20157957e+00 -2.82513261e-01 -6.61316216e-01 1.40695229e-01
-2.00612232e-01 -1.98638633e-01 -9.38948214e-01 -1.26943290e-01
-1.30850628e-01 -3.03887248e-01 1.08267522e+00 4.90929544e-01
1.73450148e+00 1.26990154e-01 -5.40174127e-01 1.06175363e+00
1.52936447e+00 -2.79665232e-01 4.48280007e-01 3.06455582e-01
1.05412745e+00 1.38218433e-01 5.50134301e-01 2.57277966e-01
4.01739180e-01 4.05172110e-01 6.68627143e-01 -2.93079764e-01
5.18416092e-02 -3.48265886e-01 -3.79455090e-01 5.73265970e-01
-1.80790678e-01 -8.27552155e-02 -9.77649808e-01 7.48559058e-01
-1.68245983e+00 -4.78648812e-01 -1.88642487e-01 2.15355539e+00
4.35736686e-01 3.41900736e-01 1.40159167e-02 2.80385558e-03
4.72390980e-01 3.03865761e-01 -6.52404428e-01 -2.33857065e-01
9.26816538e-02 6.26052141e-01 1.06833255e+00 3.27158660e-01
-1.33472812e+00 1.31742370e+00 6.09748125e+00 9.78776872e-01
-1.02276707e+00 5.59186786e-02 5.57582140e-01 -2.95764595e-01
-3.49124134e-01 -7.31268376e-02 -8.60140026e-01 2.18821555e-01
5.63291669e-01 3.34539950e-01 1.47905543e-01 1.24997449e+00
-1.89920440e-01 -5.61604016e-02 -9.62539911e-01 1.13439488e+00
-2.07461894e-01 -1.67204463e+00 2.33869776e-01 1.01274520e-01
7.18821645e-01 7.87302673e-01 -2.11259812e-01 1.18772034e-02
5.28703511e-01 -1.03657877e+00 6.63739860e-01 1.34179860e-01
6.51635885e-01 -9.72593427e-01 5.85585475e-01 1.92975685e-01
-1.21292210e+00 3.47364366e-01 -7.38114178e-01 1.54770583e-01
2.00345114e-01 9.44066882e-01 -6.14335895e-01 4.17011708e-01
1.06939781e+00 6.60860896e-01 -4.68005836e-01 1.08098972e+00
3.46887037e-02 3.38476032e-01 -1.13323963e+00 -3.48914154e-02
6.14186406e-01 1.08994925e-02 5.72038233e-01 9.62990344e-01
5.33733368e-01 4.82624993e-02 3.78205210e-01 1.05163515e+00
-2.83074737e-01 -7.63140842e-02 -5.80804825e-01 6.28474429e-02
6.63932979e-01 1.02810836e+00 -1.35456383e+00 -4.35693443e-01
-3.51990610e-01 1.15419030e+00 3.34225565e-01 3.42389531e-02
-5.15733480e-01 -4.25087839e-01 7.88762271e-01 2.00161397e-01
7.51500785e-01 -6.41127586e-01 -7.95982957e-01 -9.33664322e-01
8.95766839e-02 -2.89863378e-01 -4.57604192e-02 -7.12764561e-01
-1.11923230e+00 5.83384931e-01 -1.03172004e-01 -1.02031946e+00
1.40024751e-01 -5.67662537e-01 -4.29863662e-01 7.21347868e-01
-1.76174974e+00 -1.34000790e+00 -3.79981428e-01 5.86709440e-01
6.37221813e-01 3.32464397e-01 5.89647889e-01 1.60038263e-01
1.07331619e-01 1.93342105e-01 9.81269255e-02 -3.72160897e-02
1.82474479e-01 -1.38156855e+00 1.30371237e+00 5.99439263e-01
1.61577612e-01 6.19477153e-01 3.72667730e-01 -7.28867054e-01
-1.32777464e+00 -1.12839651e+00 7.83924997e-01 -5.07417977e-01
4.50772136e-01 -6.22718751e-01 -1.00162804e+00 5.70779204e-01
-3.66959810e-01 4.13847297e-01 2.45402426e-01 7.43746907e-02
-2.57903725e-01 9.40975398e-02 -1.30280864e+00 5.16497970e-01
1.44618201e+00 -3.06584597e-01 -6.90481782e-01 4.40401137e-01
1.17835021e+00 -6.83996916e-01 -7.42616117e-01 5.89879513e-01
1.05621733e-01 -1.02707410e+00 1.66155660e+00 -1.89663842e-01
3.14705223e-01 -4.16032732e-01 -1.51899710e-01 -9.96266305e-01
-1.35819301e-01 -4.94247228e-01 -7.52977803e-02 7.75455117e-01
8.29931572e-02 -5.39478600e-01 1.34699702e+00 4.13693756e-01
-4.72749025e-01 -7.91503310e-01 -1.08422303e+00 -7.34406233e-01
2.34514162e-01 -7.74597108e-01 1.16753101e+00 5.83968043e-01
-8.71836603e-01 -1.95285454e-01 2.64823735e-01 4.37459886e-01
8.99339437e-01 4.85317826e-01 9.15941596e-01 -1.51514328e+00
1.97318137e-01 -5.55797875e-01 -6.33733392e-01 -1.51447082e+00
2.26344094e-02 -8.30473721e-01 1.93549126e-01 -1.70831299e+00
-2.64555663e-01 -8.71447206e-01 -1.28586069e-01 5.42035401e-01
-1.33370236e-01 7.30560720e-01 9.27258804e-02 3.26794595e-01
-7.42323399e-01 4.94762719e-01 1.16306698e+00 1.10528795e-02
-3.70021015e-01 3.36543843e-02 -4.46220011e-01 1.03626895e+00
7.07033277e-01 -4.96221364e-01 -2.72676140e-01 -6.93494678e-01
1.49464384e-01 -4.79511380e-01 7.56316304e-01 -1.30047989e+00
1.43727139e-01 -3.67040336e-02 4.08076674e-01 -1.34692156e+00
6.37090743e-01 -9.18449521e-01 -2.94515163e-01 1.42312884e-01
1.13735825e-01 -1.10054336e-01 3.50022197e-01 4.97296840e-01
-4.16402332e-02 -1.03766292e-01 7.27878034e-01 -5.60919285e-01
-8.62730801e-01 5.95906377e-01 1.92193225e-01 -1.14174969e-01
9.41343904e-01 -3.86879355e-01 -3.19134407e-02 2.45140076e-01
-5.41026235e-01 2.27277607e-01 1.07062006e+00 2.63354927e-01
7.71554410e-01 -1.09104967e+00 -3.11637193e-01 3.29812557e-01
-3.20656337e-02 1.08215177e+00 1.90508127e-01 3.79120290e-01
-1.06687760e+00 3.80527705e-01 -1.03383549e-02 -1.10361123e+00
-1.05262852e+00 4.85682249e-01 1.60418332e-01 -3.16706896e-02
-1.17815328e+00 1.19215024e+00 1.95796773e-01 -8.12933147e-01
9.12483484e-02 -7.29526639e-01 3.91697705e-01 -3.94766688e-01
1.77602842e-01 2.47730017e-01 3.45559716e-01 -4.14832413e-01
-4.93116349e-01 9.31474447e-01 -1.88719884e-01 1.40878484e-01
1.58125651e+00 -1.04329977e-02 -6.71536252e-02 3.20464849e-01
1.20035326e+00 -1.32126465e-01 -1.39888763e+00 -2.32719779e-01
-1.08323283e-01 -6.88386858e-01 3.85045439e-01 -4.26839918e-01
-1.08324993e+00 9.02835488e-01 3.95321846e-01 1.88650027e-01
8.74896824e-01 3.48033696e-01 1.36939681e+00 5.84572792e-01
7.09998906e-01 -9.55799818e-01 -4.09327835e-01 5.26758373e-01
5.20435989e-01 -1.31554663e+00 2.17323840e-01 -8.06777656e-01
-1.08786531e-01 9.61620212e-01 3.40979427e-01 -6.60473764e-01
6.35100603e-01 1.88948154e-01 -2.90230047e-02 -5.28952420e-01
-2.26159573e-01 -2.46107310e-01 2.72661567e-01 5.17709672e-01
-4.38378714e-02 -1.36335357e-03 2.12457269e-01 3.53336632e-02
-4.66061026e-01 7.02506900e-02 -1.20422691e-02 1.10648942e+00
-7.00787127e-01 -9.15074468e-01 -4.91379708e-01 6.03952408e-01
-3.11833262e-01 -8.61940980e-02 -2.64203221e-01 7.37975061e-01
-2.22129915e-02 4.32358265e-01 5.91042459e-01 -8.37688446e-02
3.17975819e-01 -4.03957441e-02 3.62065166e-01 -5.60733616e-01
-5.11001945e-01 -9.68070142e-03 -3.95926028e-01 -1.04565966e+00
-5.29293120e-01 -7.12382376e-01 -1.60072267e+00 -3.91251028e-01
-1.72892734e-01 -1.17751375e-01 9.51271951e-01 8.11077833e-01
6.70792341e-01 3.75209689e-01 3.45365465e-01 -1.56480277e+00
-2.80157745e-01 -4.65971619e-01 -3.89193386e-01 3.69724363e-01
3.30913126e-01 -5.96640229e-01 -3.57521266e-01 -3.10795337e-01]
|
[7.965831279754639, -3.3206868171691895]
|
d73458a4-fca9-4345-947b-bcca14ed6f6e
|
gnn3dmot-graph-neural-network-for-3d-multi
| null | null |
http://openaccess.thecvf.com/content_CVPR_2020/html/Weng_GNN3DMOT_Graph_Neural_Network_for_3D_Multi-Object_Tracking_With_2D-3D_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Weng_GNN3DMOT_Graph_Neural_Network_for_3D_Multi-Object_Tracking_With_2D-3D_CVPR_2020_paper.pdf
|
GNN3DMOT: Graph Neural Network for 3D Multi-Object Tracking With 2D-3D Multi-Feature Learning
|
3D Multi-object tracking (MOT) is crucial to autonomous systems. Recent work uses a standard tracking-by-detection pipeline, where feature extraction is first performed independently for each object in order to compute an affinity matrix. Then the affinity matrix is passed to the Hungarian algorithm for data association. A key process of this standard pipeline is to learn discriminative features for different objects in order to reduce confusion during data association. In this work, we propose two techniques to improve the discriminative feature learning for MOT: (1) instead of obtaining features for each object independently, we propose a novel feature interaction mechanism by introducing the Graph Neural Network. As a result, the feature of one object is informed of the features of other objects so that the object feature can lean towards the object with similar feature (i.e., object probably with a same ID) and deviate from objects with dissimilar features (i.e., object probably with different IDs), leading to a more discriminative feature for each object; (2) instead of obtaining the feature from either 2D or 3D space in prior work, we propose a novel joint feature extractor to learn appearance and motion features from 2D and 3D space simultaneously. As features from different modalities often have complementary information, the joint feature can be more discriminate than feature from each individual modality. To ensure that the joint feature extractor does not heavily rely on one modality, we also propose an ensemble training paradigm. Through extensive evaluation, our proposed method achieves state-of-the-art performance on KITTI and nuScenes 3D MOT benchmarks. Our code will be made available at https://github.com/xinshuoweng/GNN3DMOT
|
[' Kris M. Kitani', ' Yunze Man', ' Yongxin Wang', 'Xinshuo Weng']
|
2020-06-01
| null | null | null |
cvpr-2020-6
|
['3d-multi-object-tracking']
|
['computer-vision']
|
[ 3.83705385e-02 -4.43338424e-01 -5.60053475e-02 -2.35904768e-01
-4.94209796e-01 -5.29311299e-01 5.31911314e-01 -6.00099415e-02
-4.04720098e-01 4.30908978e-01 -1.96901575e-01 5.46445660e-02
-7.65400305e-02 -7.26410747e-01 -7.11769879e-01 -9.70710397e-01
1.09140880e-01 6.73344493e-01 7.94969440e-01 1.18341066e-01
1.75369196e-02 8.45296800e-01 -1.74308705e+00 -7.74225146e-02
3.10196131e-01 1.07913685e+00 3.21470827e-01 6.26327813e-01
6.25715926e-02 3.98162395e-01 -4.70683724e-01 1.41442373e-01
5.20412505e-01 -5.33671379e-01 -4.95136589e-01 2.12181509e-01
6.75627649e-01 -3.43512714e-01 -3.89845133e-01 1.17756164e+00
3.97908479e-01 2.00267151e-01 7.94228077e-01 -1.33926272e+00
-2.16300771e-01 1.52729675e-01 -6.71502292e-01 2.00489640e-01
1.30364209e-01 3.00863236e-01 1.06277335e+00 -8.42072606e-01
6.66651487e-01 1.14801109e+00 3.00988019e-01 5.76731741e-01
-1.16372550e+00 -7.09938347e-01 3.01636428e-01 2.60208160e-01
-1.41964662e+00 -2.96024919e-01 9.30860817e-01 -4.06513959e-01
5.55866480e-01 2.65700370e-01 1.00358880e+00 8.28428090e-01
1.19293638e-01 9.02951419e-01 8.70836318e-01 -1.29737243e-01
-9.58778244e-03 7.36840069e-02 9.80879366e-02 8.42207849e-01
4.95283604e-01 3.15193683e-01 -4.79304969e-01 -5.39209694e-02
4.82889265e-01 3.04941118e-01 2.35594083e-02 -8.35367560e-01
-1.23097181e+00 6.05625272e-01 7.16321170e-01 2.94446170e-01
-3.00051332e-01 2.12835252e-01 2.48848628e-02 2.45861545e-01
9.59586911e-03 -4.71883863e-02 -3.36453646e-01 9.02080834e-02
-6.14095807e-01 4.01683241e-01 5.25394380e-01 1.01928663e+00
1.24867308e+00 -3.11051816e-01 -3.21838498e-01 4.17484671e-01
7.10073888e-01 7.49446988e-01 2.48055860e-01 -7.54762292e-01
2.04044849e-01 9.16702569e-01 -6.27420028e-04 -8.59004617e-01
-4.89112318e-01 -2.70615458e-01 -5.89373469e-01 6.29134834e-01
5.92349529e-01 5.34512587e-02 -9.47972417e-01 1.57961261e+00
8.66446972e-01 1.21438496e-01 -7.37662092e-02 1.18500400e+00
9.06836569e-01 2.86747456e-01 -2.59583741e-01 1.17077649e-01
1.49809480e+00 -9.47974086e-01 -2.44518951e-01 -2.93640226e-01
6.33881211e-01 -6.89843893e-01 4.97980326e-01 1.43245071e-01
-7.22085714e-01 -7.61070907e-01 -1.01353264e+00 2.06394047e-02
-3.92510355e-01 1.10497652e-03 4.98504549e-01 3.49951506e-01
-6.99837267e-01 4.22108203e-01 -9.61437941e-01 -5.40555418e-01
4.23900127e-01 5.31607985e-01 -4.34560597e-01 -6.08934015e-02
-7.05244005e-01 8.13249707e-01 5.80262244e-01 -5.84269762e-02
-8.99944246e-01 -2.55635828e-01 -8.29427719e-01 -2.49709904e-01
5.35889745e-01 -8.59341443e-01 9.75467026e-01 -7.53031373e-01
-1.36031485e+00 7.01028049e-01 -2.08078340e-01 -9.67667773e-02
3.86629552e-01 -3.29180300e-01 -2.13018119e-01 -3.12833115e-02
7.28424937e-02 6.57112062e-01 1.04118276e+00 -1.29656136e+00
-1.07776785e+00 -5.24479389e-01 -1.21621579e-01 3.22634578e-01
-1.11818932e-01 -8.52466002e-03 -7.87159801e-01 -1.96955815e-01
2.95288891e-01 -1.25270736e+00 4.49602585e-03 3.28708082e-01
-3.74696165e-01 -4.24004138e-01 1.26078415e+00 -1.14497907e-01
7.61072695e-01 -2.18353295e+00 3.85399222e-01 3.18426847e-01
4.95505005e-01 2.25687280e-01 -8.47441703e-02 -2.82889828e-02
3.77139986e-01 -3.72722059e-01 2.48708203e-02 -3.35353434e-01
-7.24353045e-02 3.29249322e-01 2.58587211e-01 7.53342688e-01
5.10955751e-01 8.87173116e-01 -9.84722137e-01 -6.69181466e-01
3.83053154e-01 3.55341166e-01 -4.30737644e-01 1.57807708e-01
-1.51287526e-01 6.51673257e-01 -8.28033507e-01 9.01352167e-01
6.37755752e-01 -3.44717503e-01 -7.19197169e-02 -4.38007355e-01
-2.35125571e-01 1.42533481e-01 -1.39138818e+00 1.50448787e+00
5.97542897e-03 3.91039819e-01 -5.87679893e-02 -7.70458519e-01
1.03513086e+00 -6.42652288e-02 7.01280653e-01 -4.84430611e-01
2.73980796e-01 1.40153483e-01 2.97202379e-01 -3.23319763e-01
4.19061005e-01 1.70389339e-01 -5.60809448e-02 3.59236926e-01
2.05895722e-01 -1.00333067e-02 7.91909024e-02 5.09287864e-02
1.12782705e+00 4.05753583e-01 2.17258528e-01 7.74350539e-02
5.16467214e-01 -5.85702509e-02 7.78968513e-01 8.15658808e-01
-4.38220888e-01 4.27185178e-01 4.90671843e-02 -2.96838343e-01
-8.63641679e-01 -1.06872308e+00 -8.71849358e-02 9.54573214e-01
6.77813768e-01 -2.43115306e-01 3.28710265e-02 -1.16246045e+00
4.11990881e-01 1.58515990e-01 -5.84599733e-01 -2.67488897e-01
-5.68767130e-01 -2.64618069e-01 2.29996800e-01 5.32881141e-01
3.81358236e-01 -8.70422721e-01 -8.58058810e-01 9.07042399e-02
1.85291797e-01 -9.37567830e-01 -4.91430730e-01 4.91453946e-01
-7.57917404e-01 -1.04683781e+00 -3.87331545e-01 -5.62826693e-01
7.05810845e-01 6.59923494e-01 7.63691962e-01 3.70040536e-01
-1.57880381e-01 3.64300609e-01 -5.42791486e-01 -3.16897631e-01
-1.34329632e-01 -1.32640330e-02 3.07359248e-01 2.24724367e-01
6.05686665e-01 -2.68406451e-01 -5.87789178e-01 5.35977066e-01
-6.21572554e-01 -1.63545348e-02 8.39044511e-01 8.04372728e-01
8.34112406e-01 -1.46620333e-01 -7.66213462e-02 -3.42344910e-01
-1.62501693e-01 -4.41439867e-01 -7.62776732e-01 6.80404380e-02
-2.16485247e-01 1.76999018e-01 3.93750221e-01 -6.77626193e-01
-6.00787044e-01 6.25975192e-01 1.62303910e-01 -8.32170308e-01
-4.40519571e-01 1.31293207e-01 -3.35583836e-01 -2.88674861e-01
3.85537058e-01 2.42185622e-01 7.06113949e-02 -4.98147875e-01
2.13406235e-01 4.90516543e-01 2.74660438e-01 -4.06535119e-01
1.22932136e+00 6.53603077e-01 3.03218514e-01 -7.09282279e-01
-7.44434953e-01 -7.23482907e-01 -9.64570045e-01 -4.92607057e-01
9.72411036e-01 -8.35399985e-01 -6.38740182e-01 3.97031754e-01
-8.42811227e-01 -1.16873011e-01 -2.00032145e-01 8.93170536e-01
-2.49660760e-01 3.23599398e-01 -1.65403694e-01 -7.40768909e-01
-9.60543156e-02 -1.18843508e+00 1.23419964e+00 5.81784904e-01
-4.14997973e-02 -7.71814406e-01 1.60314322e-01 7.07264468e-02
1.41072810e-01 1.00010693e-01 2.92500645e-01 -9.25250828e-01
-8.05435121e-01 -2.90308982e-01 -3.06786567e-01 4.29797843e-02
4.22925442e-01 1.60615593e-01 -8.10311854e-01 -5.33507109e-01
-1.22245342e-01 -2.25598156e-01 8.59897792e-01 2.11628214e-01
6.95178986e-01 1.29452601e-01 -7.42315590e-01 6.48550808e-01
1.19571757e+00 7.47953281e-02 2.22914249e-01 4.36434060e-01
9.72513437e-01 3.26806724e-01 8.66193652e-01 2.80688733e-01
4.05258209e-01 9.41614807e-01 6.03234947e-01 1.07991770e-01
-2.93118030e-01 -2.61399373e-02 5.74659884e-01 4.46868360e-01
-1.25949681e-01 -5.45697212e-02 -7.13683724e-01 4.36576128e-01
-2.04312348e+00 -8.47749650e-01 -1.78330541e-01 2.17747593e+00
3.29984248e-01 2.26603553e-01 4.60859239e-01 -2.56856948e-01
6.66981399e-01 -5.46196885e-02 -7.63916016e-01 2.76359022e-01
-1.00699626e-01 -1.97669089e-01 5.59616268e-01 9.86579880e-02
-1.13426793e+00 8.31998348e-01 4.48892784e+00 5.35450041e-01
-1.04876304e+00 -1.07981600e-01 -1.32486790e-01 -1.84024930e-01
8.24245140e-02 2.78494924e-01 -1.32430851e+00 4.65050936e-01
3.54880333e-01 8.16731676e-02 1.07524499e-01 7.60138929e-01
-2.76789457e-01 -2.00442344e-01 -1.17055154e+00 9.17375624e-01
2.18402129e-02 -8.54044318e-01 4.37397556e-03 3.04202199e-01
3.35696667e-01 2.79528767e-01 -2.00588107e-01 2.39703536e-01
2.75510371e-01 -4.88418281e-01 7.83278346e-01 5.73410630e-01
2.93016762e-01 -4.43268657e-01 6.42314613e-01 4.15192604e-01
-1.61303532e+00 6.38515800e-02 -4.64265317e-01 8.42644796e-02
1.32515179e-02 4.56418306e-01 -6.52126670e-01 9.25959826e-01
8.67360473e-01 9.92058814e-01 -6.77633405e-01 1.25711799e+00
-1.73259467e-01 2.67657667e-01 -6.69761181e-01 -1.19831078e-01
1.38951227e-01 -8.56182873e-02 9.43442822e-01 9.71117079e-01
3.77894521e-01 -1.33425295e-01 5.57104528e-01 7.85975039e-01
8.12856033e-02 -1.22502774e-01 -7.00175226e-01 2.14898318e-01
4.42935795e-01 1.53372478e+00 -7.64295757e-01 -3.43608350e-01
-7.46154368e-01 9.61170435e-01 4.12812501e-01 7.13093728e-02
-8.69913459e-01 -1.66570559e-01 7.87874758e-01 -1.49642110e-01
9.08001661e-01 -3.41870546e-01 1.37605920e-01 -1.06079352e+00
3.69979292e-02 -4.10485119e-01 5.95966101e-01 -3.93581450e-01
-1.34081769e+00 4.47179317e-01 -1.14545956e-01 -1.67799687e+00
7.48326723e-03 -6.52422607e-01 -5.20481169e-01 8.41534436e-01
-1.46663749e+00 -1.34534407e+00 -5.34162700e-01 7.75062263e-01
1.78709760e-01 -6.13384843e-02 3.69989246e-01 2.45664075e-01
-4.49308902e-01 4.54342544e-01 -2.72619314e-02 1.84461847e-01
8.15913975e-01 -1.14131904e+00 1.39160469e-01 7.89437890e-01
2.85224885e-01 2.71119446e-01 4.80526537e-01 -8.79770160e-01
-1.75163114e+00 -1.18265760e+00 5.28274298e-01 -5.42786539e-01
6.16626143e-01 -3.07370752e-01 -9.14739668e-01 5.68403602e-01
-2.03550503e-01 3.95292014e-01 4.51019287e-01 -1.21417539e-02
-1.50230646e-01 -1.24079838e-01 -9.33816016e-01 3.93045425e-01
1.25369143e+00 -2.57437199e-01 -5.32229602e-01 1.28250495e-01
4.11517650e-01 -4.70238715e-01 -8.36478949e-01 5.05743802e-01
7.28258431e-01 -8.38267863e-01 8.87680769e-01 -4.59001780e-01
-1.90533355e-01 -1.06071401e+00 -2.37738848e-01 -1.01983500e+00
-6.37142062e-01 -2.12693259e-01 -5.89204550e-01 1.08746064e+00
1.75910115e-01 -5.29260159e-01 6.96838975e-01 2.38486782e-01
-1.97754934e-01 -5.07446706e-01 -1.11356223e+00 -1.07423615e+00
-3.27271849e-01 -2.47288883e-01 5.49025953e-01 6.59250438e-01
-5.03304660e-01 4.25732106e-01 -4.22390729e-01 4.84500825e-01
8.20994914e-01 5.69069326e-01 1.25555301e+00 -1.44196534e+00
-3.41242254e-01 -4.09313202e-01 -7.90215552e-01 -1.39907598e+00
-3.34370211e-02 -1.06979573e+00 2.71915436e-01 -1.31051815e+00
4.17959690e-01 -5.86043835e-01 -4.52435851e-01 6.63333654e-01
-2.86276281e-01 4.14969444e-01 3.88428956e-01 3.51192474e-01
-9.08442080e-01 6.47931099e-01 1.36671245e+00 -2.36302763e-01
-3.42414916e-01 8.83667171e-02 -3.37410569e-01 5.17237604e-01
7.92128325e-01 -7.51764953e-01 1.53141603e-01 -3.55335802e-01
-1.63204744e-01 -2.87336677e-01 8.38553965e-01 -1.25458205e+00
4.33355629e-01 -1.39811978e-01 6.98990762e-01 -9.05378103e-01
4.70658362e-01 -1.21414661e+00 2.80691296e-01 6.61447942e-01
2.78892130e-01 -1.69359833e-01 1.34048760e-01 5.57592809e-01
3.90390605e-02 -1.69149891e-01 7.27846742e-01 1.14332967e-01
-9.47674870e-01 5.13236046e-01 -2.35787421e-01 -3.25006306e-01
1.20779920e+00 -4.87060457e-01 -2.16825068e-01 2.12622751e-02
-5.37965357e-01 5.00087857e-01 8.06210756e-01 5.82813323e-01
6.06449425e-01 -1.54864323e+00 -7.41069436e-01 3.57695669e-01
3.78025740e-01 1.39413625e-01 3.32933404e-02 1.10196388e+00
1.29497141e-01 9.79811847e-02 -2.70253569e-01 -1.16239166e+00
-1.65653944e+00 5.64374864e-01 3.36238772e-01 -6.26872927e-02
-8.08639050e-01 6.81601584e-01 2.24068284e-01 -3.48123223e-01
3.23661678e-02 -2.17451379e-01 -1.62819326e-01 6.16607629e-02
3.97935063e-01 1.81180894e-01 -7.25914612e-02 -9.13435996e-01
-7.33694077e-01 9.12999094e-01 -3.48759651e-01 1.66229397e-01
1.20227456e+00 -9.55619141e-02 1.32230282e-01 5.03727973e-01
1.04751563e+00 2.15744339e-02 -1.47515535e+00 -4.98500794e-01
-1.17600031e-01 -7.01666057e-01 -1.36813603e-03 -3.88013572e-01
-1.10262656e+00 5.06983042e-01 9.20403004e-01 1.67836770e-01
9.26642895e-01 5.43744981e-01 5.03742099e-01 3.95543575e-01
3.71602178e-01 -8.77437651e-01 1.93073869e-01 6.92768276e-01
4.57139164e-01 -1.44386446e+00 6.93065003e-02 -3.90826948e-02
-5.50564647e-01 1.02657497e+00 8.00233006e-01 -3.03930696e-02
4.98357832e-01 1.36141643e-01 5.60869463e-02 -4.99583155e-01
-4.39233214e-01 -7.55752146e-01 6.59190774e-01 4.86943811e-01
5.48583977e-02 -1.16000220e-01 1.06180489e-01 2.06564546e-01
2.32893810e-01 -2.58046091e-01 -2.36568972e-02 1.20737398e+00
-6.26708269e-01 -1.22740626e+00 -5.16934156e-01 5.74170291e-01
-2.93231346e-02 3.77528489e-01 -6.09218061e-01 7.60414600e-01
2.69411683e-01 7.33692586e-01 7.69155920e-02 -6.87268913e-01
3.57866794e-01 -3.47925313e-02 6.78406537e-01 -5.27392209e-01
-4.49623346e-01 2.13533446e-01 -1.34592175e-01 -6.35209918e-01
-7.36175954e-01 -9.56506312e-01 -1.41545701e+00 -4.26790342e-02
-6.06467843e-01 -7.61819184e-02 5.24101555e-01 9.15726423e-01
4.06620413e-01 5.23899913e-01 6.25306666e-01 -1.17616987e+00
-2.09619612e-01 -7.58266389e-01 -3.82056177e-01 4.95565385e-01
4.52547908e-01 -1.20381558e+00 -2.87737966e-01 -2.23356098e-01]
|
[6.439352035522461, -2.2244467735290527]
|
19863a09-b6db-4579-986b-01bc20f300bd
|
grammatical-error-annotation-for-korean
| null | null |
https://aclanthology.org/L12-1035
|
https://aclanthology.org/L12-1035.pdf
|
Grammatical Error Annotation for Korean Learners of Spoken English
|
The goal of our research is to build a grammatical error-tagged corpus for Korean learners of Spoken English dubbed Postech Learner Corpus. We collected raw story-telling speech from Korean university students. Transcription and annotation using the Cambridge Learner Corpus tagset were performed by six Korean annotators fluent in English. For the annotation of the corpus, we developed an annotation tool and a validation tool. After comparing human annotation with machine-recommended error tags, unmatched errors were rechecked by a native annotator. We observed different characteristics between the spoken language corpus built in this study and an existing written language corpus.
|
['Hae-Ri Kim', 'Soo-Ok Kweon', 'Kyusong Lee', 'Gary Geunbae Lee', 'Hongsuck Seo']
|
2012-05-01
| null | null | null |
lrec-2012-5
|
['grammatical-error-detection']
|
['natural-language-processing']
|
[-1.05050534e-01 3.00823718e-01 1.47736877e-01 -5.76591432e-01
-1.43851388e+00 -7.42070019e-01 1.37579501e-01 4.52546328e-01
-8.34715366e-01 1.16680741e+00 6.27821565e-01 -2.82298565e-01
3.13236922e-01 -4.12096500e-01 -6.57540262e-01 -2.43617594e-02
2.50542015e-01 4.57668155e-01 5.04129231e-01 -1.71610624e-01
3.93286377e-01 -3.61135006e-01 -1.12138569e+00 3.59467477e-01
1.37541914e+00 3.88547003e-01 5.27610421e-01 8.00192118e-01
-4.11465645e-01 1.21898365e+00 -8.81514192e-01 -7.06399679e-01
-4.36891019e-01 -6.92217588e-01 -1.37444794e+00 -5.37189431e-02
1.68646097e-01 -1.30832419e-01 1.64212231e-02 9.16228592e-01
5.18152595e-01 3.70524377e-01 2.44671538e-01 -7.68866718e-01
-6.75863206e-01 1.64162111e+00 5.00942767e-01 1.70648962e-01
8.92590702e-01 -5.89657187e-01 8.32881033e-01 -1.27659237e+00
1.12316108e+00 7.17723608e-01 7.33196437e-01 8.40918362e-01
-8.00867140e-01 -6.73133552e-01 -1.74734399e-01 2.33257726e-01
-1.74182618e+00 -6.18558347e-01 5.95278919e-01 -4.34498161e-01
1.02629077e+00 1.84124243e-02 7.57919550e-01 1.36850142e+00
-4.21341509e-03 8.53743911e-01 9.40113187e-01 -1.01539576e+00
1.83227628e-01 5.10940552e-01 4.17670250e-01 6.42383456e-01
-3.31371516e-01 -3.68059784e-01 -1.00183356e+00 2.09387630e-01
1.28142312e-01 -7.39015758e-01 -6.99902475e-01 2.38841653e-01
-1.10727072e+00 7.07185388e-01 -5.61400294e-01 7.31142223e-01
1.10381469e-01 -1.90489411e-01 6.85426831e-01 5.24765909e-01
5.65053761e-01 4.80668247e-01 -6.80998862e-01 -8.41652393e-01
-8.57638538e-01 -2.05937684e-01 8.99825215e-01 1.57381010e+00
1.66429996e-01 5.09810299e-02 2.27368042e-01 1.17648196e+00
3.93306464e-01 3.10624659e-01 1.11049736e+00 -8.02398741e-01
4.32333678e-01 3.96681070e-01 8.67057871e-03 -5.68024635e-01
-1.58466756e-01 -1.09460361e-01 -2.25975186e-01 -5.61229885e-01
4.54826027e-01 -4.41279352e-01 -5.02098262e-01 1.34722996e+00
2.26527289e-01 -4.16933298e-02 6.24210596e-01 4.11190599e-01
1.19228661e+00 6.62381709e-01 2.74360418e-01 -5.07050931e-01
1.20917499e+00 -1.09973383e+00 -1.27843630e+00 1.59667253e-01
1.31356883e+00 -1.00507569e+00 1.37064493e+00 6.10124707e-01
-9.99947071e-01 -5.20640194e-01 -1.07816088e+00 -2.83386618e-01
-3.08093786e-01 1.20090291e-01 -2.02029899e-01 7.44132102e-01
-7.40726590e-01 2.49618709e-01 -5.65645993e-01 -5.24995565e-01
-2.12670267e-01 -3.01279604e-01 -5.61767399e-01 3.52451168e-02
-1.50652361e+00 7.97076166e-01 7.86912382e-01 -2.28438988e-01
-9.12028074e-01 -5.30840158e-01 -1.06787324e+00 -3.03322345e-01
3.63472670e-01 2.60935187e-01 1.89721811e+00 -1.07477701e+00
-1.91005540e+00 1.27620089e+00 -1.95115693e-02 -3.02064508e-01
3.33442211e-01 -4.87527788e-01 -9.02390480e-01 -5.22454083e-02
1.25857562e-01 4.56847399e-02 2.79331267e-01 -1.12763536e+00
-9.31911230e-01 -6.61734492e-02 -5.75095892e-01 2.93889821e-01
-3.07510078e-01 3.75012457e-01 -3.36302310e-01 -1.09759176e+00
1.59841761e-01 -5.57638884e-01 3.73675048e-01 -9.59094048e-01
-1.51530191e-01 -6.11107945e-01 4.76462245e-01 -1.17748833e+00
1.79634309e+00 -2.13789773e+00 -1.43993199e-01 2.18697749e-02
-9.70112979e-02 2.35600844e-01 -6.00187704e-02 8.51055801e-01
1.28047720e-01 4.58710760e-01 -1.25715092e-01 -5.02434790e-01
-9.45710763e-02 4.39451903e-01 -3.67929816e-01 1.40332446e-01
-2.68643975e-01 5.64766049e-01 -1.39089799e+00 -8.14368188e-01
-2.07229272e-01 1.45212501e-01 -2.59759039e-01 7.22798586e-01
7.97056109e-02 6.11421406e-01 -1.37151256e-01 6.62179172e-01
-3.54829729e-02 7.64840364e-01 1.94808871e-01 5.43553770e-01
-4.90316600e-01 1.06626725e+00 -1.03993535e+00 2.02523756e+00
-4.97642964e-01 6.33862913e-01 3.59645841e-04 -7.65115917e-01
1.04392743e+00 1.11290920e+00 -1.67703792e-01 -4.66795832e-01
1.53719813e-01 6.21583402e-01 -1.58318922e-01 -9.19197977e-01
9.20140445e-01 -5.43405265e-02 -5.15437365e-01 4.89785433e-01
7.56394625e-01 -5.99149883e-01 1.00850619e-01 1.08952954e-01
8.63196373e-01 1.90580472e-01 5.81260741e-01 -2.42929071e-01
6.31698132e-01 4.17554408e-01 5.37220657e-01 6.96011841e-01
-2.67202646e-01 4.64600980e-01 1.07952774e-01 -1.90430805e-01
-9.08224583e-01 -7.02072978e-01 -2.71752387e-01 1.52560163e+00
-4.37322736e-01 -9.29371536e-01 -1.16383350e+00 -8.62557590e-01
-9.03575778e-01 1.15752459e+00 -5.32880537e-02 1.88031822e-01
-7.31036842e-01 3.19745004e-01 1.10104275e+00 4.13550407e-01
4.22012419e-01 -1.16254961e+00 -1.92788228e-01 7.29410350e-01
-4.76441771e-01 -1.22318482e+00 -8.06067944e-01 2.51949728e-01
-2.82774985e-01 -1.02530789e+00 -2.34215721e-01 -1.39962649e+00
3.12139601e-01 -2.57737964e-01 1.09491658e+00 2.27403015e-01
4.13636297e-01 4.21051949e-01 -1.01730537e+00 -7.19587326e-01
-8.96162629e-01 2.94068605e-01 1.39030412e-01 -5.38718998e-01
4.42968696e-01 -3.39765429e-01 2.77911693e-01 2.56798834e-01
-6.97117329e-01 5.60412409e-05 -3.42388377e-02 7.92368650e-01
4.27878529e-01 -2.27876622e-02 6.18879616e-01 -1.02626562e+00
6.89948320e-01 -1.44598112e-01 -4.76552904e-01 2.03397140e-01
-3.28646183e-01 -8.68602470e-02 4.94285703e-01 -6.33077979e-01
-1.28944850e+00 3.76624130e-02 -7.22886324e-01 2.86205500e-01
-2.88276494e-01 8.71186793e-01 -4.64297235e-01 3.15403640e-01
4.17070597e-01 2.00296521e-01 -6.30970895e-01 -5.44277489e-01
1.28760949e-01 1.18444824e+00 1.04612327e+00 -8.03194940e-01
2.19107240e-01 -8.31281960e-01 -9.44641471e-01 -9.85016942e-01
-8.80228519e-01 -3.53298515e-01 -1.00201464e+00 -6.66508377e-01
1.04596484e+00 -1.01097667e+00 -1.89440191e-01 7.46076226e-01
-1.30390728e+00 -4.77253884e-01 -3.06896925e-01 7.53333986e-01
-2.31391832e-01 1.33390754e-01 -7.79027700e-01 -7.87260294e-01
-4.28408803e-03 -8.85792732e-01 5.25371850e-01 -8.15218687e-02
-7.51526058e-01 -1.07327724e+00 5.06307900e-01 5.95396221e-01
-7.60206506e-02 -3.17038149e-01 7.33398914e-01 -1.33237135e+00
9.05478299e-02 -1.88610733e-01 6.54072404e-01 4.52439576e-01
-3.24023485e-01 1.98066711e-01 -9.13095772e-01 1.74232572e-01
7.86475241e-02 -8.90125573e-01 1.65675074e-01 -2.94591784e-01
6.70758009e-01 -5.62144995e-01 5.91012947e-02 2.18957700e-02
9.70964849e-01 6.03800297e-01 3.83922607e-01 1.31542027e-01
6.16461396e-01 6.74964726e-01 5.66853523e-01 1.96582787e-02
6.50439441e-01 4.52285409e-01 -4.30679023e-01 8.96099925e-01
2.91694611e-01 -9.63470399e-01 6.60309911e-01 2.16279650e+00
1.34500369e-01 -7.50410140e-01 -1.31751323e+00 8.49977851e-01
-1.70055079e+00 -5.93182981e-01 -4.15015817e-01 1.98236454e+00
1.48332369e+00 1.42713979e-01 -2.65377551e-01 2.77601391e-01
6.31131232e-01 -1.85013667e-01 5.98084092e-01 -5.41405916e-01
4.87598069e-02 4.23530519e-01 1.19616427e-01 1.02880251e+00
-8.20525646e-01 1.32714295e+00 6.92425060e+00 1.04075265e+00
-7.69099891e-01 5.71008682e-01 5.46207428e-02 3.32595050e-01
-2.82203674e-01 -1.29678264e-01 -8.72786105e-01 4.68842417e-01
1.51288724e+00 -2.00308323e-01 3.87732089e-02 9.13677692e-01
1.95869312e-01 -2.72158325e-01 -1.01184404e+00 6.29422128e-01
3.13488692e-01 -1.00200689e+00 -2.82252878e-01 -4.13215488e-01
6.83457136e-01 -2.52788842e-01 -7.58540750e-01 7.94648945e-01
5.59429109e-01 -8.49638522e-01 1.23514903e+00 5.89038670e-01
7.65581489e-01 -9.44642723e-01 1.09684718e+00 6.08354032e-01
-9.04973626e-01 4.36536670e-01 -4.10362780e-02 -1.20376095e-01
2.03674763e-01 2.55876333e-01 -9.44830954e-01 1.91801399e-01
5.85511863e-01 2.82270819e-01 -3.44466299e-01 5.72899759e-01
-9.16323364e-01 1.44872832e+00 -4.68149260e-02 -2.53736526e-01
4.27797884e-02 -3.58577259e-03 6.80850863e-01 1.50544453e+00
6.39800489e-01 4.82935935e-01 3.19574088e-01 3.98325711e-01
-5.23953140e-01 5.67424238e-01 -4.85012829e-01 -3.68456453e-01
1.08670151e+00 8.91713858e-01 -5.03899276e-01 -4.25608099e-01
-3.56627285e-01 1.24241769e+00 4.71895486e-01 1.32783622e-01
-4.37792927e-01 -6.24865830e-01 1.90444022e-01 7.70585537e-02
-2.03856193e-02 -6.15175664e-01 -1.55299708e-01 -1.10000014e+00
-1.87409326e-01 -7.76230335e-01 2.45311812e-01 -9.37757909e-01
-1.09889948e+00 7.66678452e-01 -8.57212171e-02 -9.87999082e-01
-4.86871541e-01 -3.39270383e-01 -5.06364048e-01 5.56434274e-01
-7.63334453e-01 -7.73393989e-01 -1.15997829e-01 3.30444157e-01
9.39707220e-01 -2.96721935e-01 1.21951056e+00 3.90864670e-01
-6.81179345e-01 6.40924931e-01 -1.27010018e-01 6.32581532e-01
9.80157077e-01 -1.41861534e+00 -1.53075039e-01 8.77023220e-01
3.35226327e-01 4.74083453e-01 7.92136550e-01 -9.53009546e-01
-8.03437889e-01 -9.60994363e-01 1.95902193e+00 -5.09505033e-01
9.15156782e-01 -5.17529249e-01 -1.26257443e+00 9.57059443e-01
7.53208995e-01 -1.63969606e-01 1.15024102e+00 -1.85945764e-01
6.73187748e-02 3.54680032e-01 -8.26069236e-01 2.39334628e-01
9.48437512e-01 -8.97171855e-01 -1.31466973e+00 1.74921453e-02
1.00363040e+00 -8.49363387e-01 -1.02376854e+00 -8.64621401e-02
3.65885496e-01 -3.22248667e-01 8.68437514e-02 -4.91777390e-01
4.83960092e-01 -2.02147454e-01 -1.27485290e-01 -1.68592727e+00
3.15437675e-01 -6.93528056e-01 3.37600142e-01 2.01083851e+00
5.44551253e-01 -1.26849473e-01 -8.99162702e-03 5.29406190e-01
-9.37971294e-01 -1.99934751e-01 -1.06805205e+00 -5.74227393e-01
1.42261162e-01 -9.02478993e-01 2.93027639e-01 1.33516061e+00
8.82683694e-01 6.36423230e-01 -1.04686052e-01 3.72554474e-02
4.70552370e-02 -8.48361909e-01 5.13874769e-01 -1.00093770e+00
2.96608806e-01 -3.82564170e-03 -7.91787431e-02 -6.92730069e-01
6.40137196e-01 -8.98241699e-01 6.33524060e-01 -1.01449287e+00
-2.74408042e-01 -1.82945162e-01 2.62253165e-01 6.35359228e-01
-1.39113799e-01 2.28325471e-01 -2.16224208e-01 -7.56155001e-03
-6.59797192e-01 6.08941734e-01 6.32327735e-01 2.45377257e-01
-4.59132791e-01 -3.72358769e-01 -1.29062131e-01 9.42881525e-01
6.04530990e-01 -9.13454711e-01 -4.86047193e-02 -3.65555167e-01
3.40576291e-01 1.36050239e-01 -1.67027012e-01 -1.08806801e+00
3.11981201e-01 -1.26064599e-01 -3.61657329e-02 -5.69120586e-01
-4.54038292e-01 -8.18195462e-01 8.54933634e-02 -1.24586225e-01
-6.34792864e-01 1.99822873e-01 2.05082700e-01 -2.56263703e-01
-6.28199279e-01 -9.04150724e-01 7.08134413e-01 -9.27790180e-02
-8.85695815e-01 -4.52960908e-01 -1.38938057e+00 5.48043668e-01
1.00700986e+00 -2.57381439e-01 -1.17768809e-01 -6.53170645e-01
-7.67943978e-01 2.51904339e-01 2.42851198e-01 3.05880576e-01
6.00525618e-01 -1.43220782e+00 -6.77077174e-01 1.37725919e-01
3.41715634e-01 2.34600343e-02 -1.66159049e-01 4.21719640e-01
-6.83446229e-01 3.91011983e-01 -2.50940844e-02 -1.12995900e-01
-1.40008914e+00 5.51350676e-02 1.13963030e-01 -3.56819853e-02
-4.06675637e-01 9.84165132e-01 -9.24830735e-01 -7.48411357e-01
5.39958358e-01 -5.58881819e-01 -3.59878778e-01 2.00366214e-01
6.58404827e-01 4.59577054e-01 2.55624026e-01 -8.13688099e-01
-1.40320495e-01 -5.17362915e-02 3.11735064e-01 -6.22120023e-01
1.14978349e+00 -5.76885283e-01 -3.12438197e-02 1.18547392e+00
8.85647595e-01 9.97203290e-01 -4.06673431e-01 -1.52182862e-01
5.94581008e-01 3.12618986e-02 2.39845756e-02 -8.17972124e-01
-2.60792345e-01 4.79295850e-01 7.01791942e-02 2.14638934e-01
6.76376045e-01 -6.14917744e-03 1.01836789e+00 5.48949361e-01
3.17430258e-01 -1.75464892e+00 -4.49685194e-02 1.41301799e+00
8.86406064e-01 -1.20941889e+00 -5.93503535e-01 -3.37972999e-01
-8.62880886e-01 1.08387673e+00 7.72280157e-01 3.48571658e-01
5.20060539e-01 4.43568677e-01 5.03821790e-01 -4.96679395e-02
-5.52671850e-01 7.23201921e-03 4.71425235e-01 5.83086431e-01
1.08433890e+00 -2.77720648e-03 -9.37922895e-01 1.29230225e+00
-9.15137470e-01 6.28694370e-02 7.89980233e-01 1.15473390e+00
-6.35287821e-01 -1.24465644e+00 -4.14140075e-01 -2.76322782e-01
-6.50857925e-01 -2.67548621e-01 -6.67148471e-01 8.61129344e-01
3.76555532e-01 1.22550666e+00 2.52004713e-01 -4.09860522e-01
3.82086575e-01 9.09696460e-01 2.53545612e-01 -1.05734229e+00
-7.25423574e-01 -9.78883952e-02 7.31102884e-01 -2.73477346e-01
-4.79783028e-01 -5.91873050e-01 -1.59827185e+00 1.35710940e-01
-4.28610772e-01 8.93649161e-01 5.16246438e-01 1.26983571e+00
-4.04121697e-01 3.85050476e-01 2.64596492e-01 -1.90440983e-01
4.58224006e-02 -1.52075243e+00 -6.90020025e-01 -1.66678671e-02
5.88537082e-02 -1.62477747e-01 -3.18830371e-01 5.41417003e-01]
|
[10.930191993713379, 10.484824180603027]
|
068c10d0-b26e-49a5-9e27-05bdf8795a69
|
learning-to-infer-counterfactuals-meta
|
2208.06748
| null |
https://arxiv.org/abs/2208.06748v1
|
https://arxiv.org/pdf/2208.06748v1.pdf
|
Learning to Infer Counterfactuals: Meta-Learning for Estimating Multiple Imbalanced Treatment Effects
|
We regularly consider answering counterfactual questions in practice, such as "Would people with diabetes take a turn for the better had they choose another medication?". Observational studies are growing in significance in answering such questions due to their widespread accumulation and comparatively easier acquisition than Randomized Control Trials (RCTs). Recently, some works have introduced representation learning and domain adaptation into counterfactual inference. However, most current works focus on the setting of binary treatments. None of them considers that different treatments' sample sizes are imbalanced, especially data examples in some treatment groups are relatively limited due to inherent user preference. In this paper, we design a new algorithmic framework for counterfactual inference, which brings an idea from Meta-learning for Estimating Individual Treatment Effects (MetaITE) to fill the above research gaps, especially considering multiple imbalanced treatments. Specifically, we regard data episodes among treatment groups in counterfactual inference as meta-learning tasks. We train a meta-learner from a set of source treatment groups with sufficient samples and update the model by gradient descent with limited samples in target treatment. Moreover, we introduce two complementary losses. One is the supervised loss on multiple source treatments. The other loss which aligns latent distributions among various treatment groups is proposed to reduce the discrepancy. We perform experiments on two real-world datasets to evaluate inference accuracy and generalization ability. Experimental results demonstrate that the model MetaITE matches/outperforms state-of-the-art methods.
|
['Liming Zhu', 'Chen Wang', 'Xiwei Xu', 'Lina Yao', 'Guanglin Zhou']
|
2022-08-13
| null | null | null | null |
['counterfactual-inference']
|
['miscellaneous']
|
[ 5.08273184e-01 3.50248590e-02 -1.36946273e+00 -5.72099447e-01
-9.81167853e-01 -2.87537184e-02 4.67750400e-01 5.90884462e-02
-4.78970826e-01 1.50225198e+00 6.21963859e-01 -4.59043622e-01
-4.65128392e-01 -8.07264030e-01 -9.00292575e-01 -6.62841558e-01
-3.39515544e-02 4.53512520e-01 -6.56951010e-01 1.42759219e-01
3.46841037e-01 -9.65398476e-02 -1.43357158e+00 5.03477454e-01
1.53863311e+00 6.14968419e-01 -1.31848395e-01 -1.22521020e-01
-1.53934032e-01 6.82299912e-01 -4.67433989e-01 -5.40857553e-01
8.71996731e-02 -5.70441246e-01 -5.04143715e-01 -1.21861115e-01
4.26488906e-01 -4.48497504e-01 -1.60171747e-01 1.02544487e+00
9.15502548e-01 1.20447939e-02 9.57205594e-01 -1.47362435e+00
-7.07986236e-01 1.09647250e+00 -1.06026757e+00 5.11039793e-03
2.13911682e-01 2.75053978e-01 7.34883666e-01 -2.58246183e-01
2.66944945e-01 1.63836956e+00 6.69221222e-01 6.39353693e-01
-1.42320549e+00 -1.13800049e+00 3.31006080e-01 3.98727477e-01
-8.45156312e-01 -4.07422483e-01 8.98108780e-01 -3.22106361e-01
4.32415128e-01 1.22894093e-01 4.77954596e-01 1.55107689e+00
3.38790536e-01 8.66882145e-01 1.40118897e+00 -4.03323621e-01
6.00290954e-01 2.87235435e-02 -8.68117660e-02 1.74275544e-02
7.23225236e-01 5.07363081e-01 -6.19672500e-02 -4.98342007e-01
5.78803599e-01 3.42430800e-01 -4.28753525e-01 -6.14664376e-01
-1.14399266e+00 1.24289417e+00 4.22395259e-01 -7.50546679e-02
-6.99818790e-01 6.43120632e-02 7.89800167e-01 2.03201741e-01
6.71829700e-01 1.22399017e-01 -6.36019588e-01 1.30464301e-01
-8.04956496e-01 7.28644490e-01 5.30264258e-01 5.05620241e-01
2.26792321e-01 -1.21075720e-01 -4.76468027e-01 1.04866827e+00
1.29161075e-01 5.95668375e-01 1.03343463e+00 -9.07781959e-01
1.07248425e+00 5.31372905e-01 4.35413688e-01 -5.91535985e-01
-3.29155773e-01 -3.10536474e-01 -1.35143948e+00 3.00930366e-02
5.17945409e-01 -4.92976815e-01 -6.40108705e-01 2.03919792e+00
4.82631534e-01 4.61808562e-01 5.09890839e-02 7.57213771e-01
4.26726133e-01 3.53620082e-01 4.13773268e-01 -9.13854301e-01
1.27869391e+00 -4.95853305e-01 -9.53541279e-01 -1.47574201e-01
7.88814068e-01 -3.85191351e-01 9.83577549e-01 3.25195968e-01
-1.03518462e+00 -4.18187022e-01 -8.87091637e-01 2.78107107e-01
-1.72910899e-01 7.20482618e-02 9.18922842e-01 9.21557188e-01
-1.89157411e-01 7.70664871e-01 -3.11651558e-01 7.79854208e-02
9.43319440e-01 2.36716121e-01 2.25800835e-02 -2.79561371e-01
-1.60247815e+00 6.93426311e-01 5.43289185e-01 -1.91230908e-01
-6.79766178e-01 -1.45288670e+00 -9.54659581e-01 -2.12773941e-02
5.41385055e-01 -1.30619252e+00 1.32402670e+00 -1.01543272e+00
-1.44468224e+00 6.68380618e-01 -1.16402589e-01 -8.31147909e-01
1.00105977e+00 -1.53478920e-01 -4.58178937e-01 -3.78474653e-01
3.86135876e-01 4.10530061e-01 6.37100399e-01 -1.05310309e+00
-8.03502738e-01 -7.98397303e-01 1.05605563e-02 2.80167431e-01
3.56869511e-02 -3.66633594e-01 5.76940238e-01 -7.71063983e-01
-3.23653013e-01 -5.39957941e-01 -4.55857068e-01 -2.37925649e-01
-4.53087002e-01 -3.55112463e-01 3.95830899e-01 -2.98727483e-01
1.29631376e+00 -1.67977321e+00 -2.32230693e-01 -2.39335358e-01
-5.10331523e-03 1.05326593e-01 -4.83624525e-02 1.56787708e-01
-6.47615433e-01 1.97728321e-01 -2.74299204e-01 2.79340446e-02
4.93686460e-02 -1.75898802e-02 -6.86011612e-01 5.79295278e-01
-2.74117917e-01 7.59862006e-01 -1.04532993e+00 -5.34771740e-01
3.61931354e-01 3.87956873e-02 -7.94659138e-01 1.14412151e-01
-2.64290839e-01 2.41204932e-01 -6.17973745e-01 2.10131586e-01
1.25334477e+00 -6.59282506e-02 5.25423944e-01 -2.54868925e-01
-7.74314925e-02 5.67496955e-01 -1.24101448e+00 1.52701545e+00
-6.55101180e-01 -1.83021277e-01 -4.79902297e-01 -1.68516707e+00
4.60984856e-01 4.80488718e-01 4.74467665e-01 -7.60988057e-01
1.87003508e-01 2.52821594e-01 2.19419241e-01 -7.27179945e-01
-1.05684176e-01 -8.37034464e-01 -6.62346482e-02 4.03876245e-01
-3.05755138e-01 8.27585608e-02 1.11617856e-01 -2.67667472e-01
4.05329257e-01 1.36202440e-01 8.22188675e-01 -9.61340889e-02
3.10841352e-01 -3.30480516e-01 9.11991477e-01 1.00118458e+00
-1.37389287e-01 3.66455883e-01 5.48226237e-01 -4.49915230e-01
-8.95664036e-01 -1.12077117e+00 -4.31456834e-01 7.25927770e-01
-1.30368426e-01 3.02858174e-01 -4.66642469e-01 -9.12198961e-01
4.11797106e-01 1.12234581e+00 -8.48005533e-01 -4.65886354e-01
-3.65262538e-01 -1.48851192e+00 2.83977389e-01 5.91165125e-01
7.75499403e-01 -1.13102090e+00 -2.88280606e-01 2.06676796e-01
-4.21958387e-01 -2.89728373e-01 -4.36246812e-01 -2.90147513e-01
-1.33226562e+00 -1.23728442e+00 -9.99450386e-01 -3.69086891e-01
1.94006398e-01 1.46421105e-01 1.08442485e+00 -7.37751901e-01
-1.29760310e-01 -2.12681293e-01 1.40267700e-01 -7.57713974e-01
-1.97431311e-01 -2.56512254e-01 2.71249473e-01 -3.54528911e-02
6.49348795e-01 -8.06936204e-01 -9.85416174e-01 6.18363405e-03
-7.00579286e-01 1.48976799e-02 7.55583227e-01 1.05898774e+00
4.11140800e-01 3.08978632e-02 1.28074813e+00 -1.34611130e+00
7.84453690e-01 -9.21785355e-01 -4.38947588e-01 2.55670398e-01
-7.53366292e-01 6.21132776e-02 6.46228552e-01 -7.18740821e-01
-1.45350671e+00 -6.43676996e-01 2.55467772e-01 -2.46396527e-01
-1.11750096e-01 6.46892309e-01 -5.69878876e-01 8.10209632e-01
6.44412220e-01 6.90897405e-02 7.56851137e-02 -4.85121906e-01
4.49881315e-01 9.04311895e-01 1.64379418e-01 -7.25269318e-01
1.84183300e-01 5.04848480e-01 -7.45991319e-02 -2.20935374e-01
-1.21894526e+00 -8.85114893e-02 1.02801234e-01 3.90919238e-01
3.87276143e-01 -9.95219588e-01 -1.13276923e+00 3.82647514e-01
-8.32920253e-01 -3.14631104e-01 -3.87389362e-01 1.02313912e+00
-9.34095502e-01 2.17829019e-01 -1.39923200e-01 -9.04411614e-01
-3.37455302e-01 -8.70218694e-01 7.66543984e-01 2.11489558e-01
-1.65542588e-01 -1.14217150e+00 4.03344601e-01 4.50223744e-01
5.59195764e-02 4.48924273e-01 1.15636790e+00 -3.81524354e-01
5.83139174e-02 2.90606059e-02 -3.42030436e-01 1.72042564e-01
6.46478653e-01 -5.65236688e-01 -7.55890429e-01 -3.76994669e-01
2.48358786e-01 -2.85992831e-01 9.68435824e-01 1.26106238e+00
1.84066582e+00 -7.46369481e-01 -6.93146586e-01 3.57973486e-01
1.38291049e+00 3.24408710e-01 7.61390150e-01 1.76408499e-01
2.92177111e-01 6.98891342e-01 7.03130662e-01 6.62242115e-01
4.57193762e-01 6.53971791e-01 3.60263497e-01 6.48547560e-02
2.16655403e-01 -4.48587507e-01 1.47257999e-01 -3.69907618e-02
-1.71100438e-01 -1.48093626e-01 -4.23019141e-01 7.03222632e-01
-2.02880287e+00 -1.41789770e+00 3.81859504e-02 2.67845917e+00
1.28144920e+00 1.11814417e-01 3.60107571e-01 6.27637506e-02
1.01129055e+00 1.28912330e-01 -9.24669743e-01 -3.45822603e-01
9.27467495e-02 1.72711536e-01 5.22587895e-01 1.93214208e-01
-1.24879038e+00 1.60121128e-01 5.35860300e+00 1.10239458e+00
-9.73005652e-01 1.00348927e-01 1.08911467e+00 -2.18871132e-01
-4.09669489e-01 -1.16283037e-01 -5.01609147e-01 9.59452808e-01
9.93411541e-01 -5.65702379e-01 2.72468720e-02 6.79334223e-01
7.94705212e-01 -2.02883277e-02 -1.43235254e+00 1.04308510e+00
-1.83220014e-01 -1.42652023e+00 2.97619641e-01 9.08498392e-02
1.15976012e+00 -4.05960262e-01 2.43470445e-01 6.77498996e-01
5.73653519e-01 -1.05765104e+00 3.74619246e-01 5.67189276e-01
9.57844555e-01 -8.60088110e-01 9.28659916e-01 4.91962641e-01
-4.70847249e-01 -3.27838272e-01 -5.44363022e-01 -3.46485704e-01
6.81357831e-02 8.89884531e-01 -7.20575511e-01 8.15457344e-01
2.97802240e-01 7.57755756e-01 6.87933788e-02 1.28302467e+00
-2.19869446e-02 6.55812383e-01 -2.07322538e-02 7.61410668e-02
-8.27256143e-02 -2.09688872e-01 1.22792155e-01 7.77752161e-01
3.60306740e-01 -1.20137259e-02 8.53496045e-02 8.74764383e-01
-3.56610030e-01 1.90749183e-01 -7.74449050e-01 3.09990197e-01
6.17046952e-01 5.93432486e-01 3.80150452e-02 -5.08503199e-01
-2.95725882e-01 4.28292602e-01 7.68367425e-02 3.71983737e-01
-9.55545902e-01 -6.86118603e-02 6.42196000e-01 1.89599156e-01
-3.65301669e-02 8.85759354e-01 -5.57662845e-01 -1.30064261e+00
-4.07583825e-02 -1.07520497e+00 8.59126866e-01 -4.19312626e-01
-1.87996471e+00 -3.66670430e-01 3.95813912e-01 -1.45975792e+00
-3.93980682e-01 -3.83969516e-01 -6.83034897e-01 7.75695264e-01
-1.55640626e+00 -8.53397489e-01 2.37831056e-01 2.89074689e-01
4.88938689e-01 5.40569499e-02 6.18448257e-01 3.13076347e-01
-5.59748113e-01 7.90715992e-01 3.19733411e-01 -1.13224395e-01
9.14693415e-01 -1.17475474e+00 -3.46987098e-01 6.12086207e-02
-5.20139933e-01 7.26225019e-01 6.78372502e-01 -6.60395682e-01
-8.49834740e-01 -1.16643786e+00 7.78101265e-01 -1.63449228e-01
3.38058114e-01 4.08955477e-02 -6.63984001e-01 7.60744631e-01
1.10133782e-01 -2.67416209e-01 8.54856491e-01 4.54788536e-01
-3.64029020e-01 -3.77864867e-01 -1.55249691e+00 8.70157182e-01
9.57905054e-01 1.37108797e-02 -1.13024402e+00 3.84082317e-01
6.01671576e-01 -2.26591289e-01 -5.80885351e-01 7.22008169e-01
6.30059600e-01 -8.10709655e-01 1.17169285e+00 -1.27764952e+00
9.29114759e-01 -1.25440229e-02 1.15949787e-01 -1.54746711e+00
-4.16715555e-02 -3.29006374e-01 -2.42408961e-02 1.14707398e+00
1.75214380e-01 -9.52439904e-01 7.83701122e-01 4.44026381e-01
3.11146736e-01 -7.16739893e-01 -1.06727242e+00 -6.22559190e-01
6.13006711e-01 -2.28486285e-01 9.49581742e-01 1.24129891e+00
1.98311657e-01 3.91265094e-01 -5.97081661e-01 -1.52676627e-01
1.00775683e+00 7.22376287e-01 5.98742783e-01 -1.17099190e+00
-2.44047403e-01 -4.38363910e-01 5.61426580e-02 -6.71864271e-01
3.15126568e-01 -7.66119540e-01 -4.10867929e-01 -1.39175272e+00
8.50241601e-01 -3.55319858e-01 -4.97713387e-01 3.09076935e-01
-5.60023248e-01 -2.07065970e-01 -2.26739049e-01 -2.30969727e-01
-2.50703067e-01 9.58441734e-01 1.29533637e+00 -3.89727712e-01
-2.99749434e-01 5.11713088e-01 -1.09485972e+00 8.30963135e-01
8.60783339e-01 -5.60968995e-01 -6.56051517e-01 1.14109561e-01
4.18359705e-04 3.23433489e-01 4.30866927e-01 -5.03801584e-01
-3.04872721e-01 -6.65734231e-01 4.50727433e-01 -4.95384961e-01
-2.10142538e-01 -5.53289950e-01 4.13620882e-02 6.46694362e-01
-6.75010204e-01 -3.99479598e-01 1.14775673e-01 6.87464595e-01
3.93493585e-02 -2.01854974e-01 6.84443533e-01 -1.85797706e-01
-1.96196720e-01 2.70240098e-01 7.99243376e-02 4.54372972e-01
7.58683741e-01 3.13171037e-02 -3.52482617e-01 -3.64715517e-01
-5.03733397e-01 4.43496287e-01 -6.20119609e-02 4.13867831e-01
4.10464287e-01 -1.63740480e+00 -1.11566734e+00 -2.62173861e-01
4.03782912e-02 -1.34555653e-01 7.73413658e-01 8.97708237e-01
1.32109314e-01 4.97599602e-01 -1.67933211e-01 -1.56772986e-01
-8.25280011e-01 1.00260544e+00 3.80816638e-01 -6.81685150e-01
-3.01444173e-01 2.43344843e-01 6.89692795e-01 -6.40798450e-01
3.24638039e-02 -4.01244849e-01 -2.72155344e-01 2.78463662e-01
7.24163830e-01 6.78427398e-01 -2.83295244e-01 1.84593067e-01
-1.17568471e-01 1.79853499e-01 -3.50457430e-01 2.07922786e-01
1.41051936e+00 -5.16165122e-02 1.78054236e-02 4.84874606e-01
1.09875667e+00 -2.16776982e-01 -1.16992247e+00 -2.02699110e-01
-1.54971555e-01 -6.70150876e-01 -7.93107972e-02 -1.01547873e+00
-7.63571799e-01 7.29695380e-01 8.52824032e-01 -1.47866225e-02
1.11998963e+00 -3.17906976e-01 3.71127546e-01 1.32096810e-02
3.40093017e-01 -1.04019499e+00 -2.10174158e-01 -1.81378409e-01
9.28814828e-01 -1.40020037e+00 2.42895991e-01 -1.31311551e-01
-4.47568804e-01 5.36649823e-01 4.67584699e-01 -2.40437344e-01
5.12813270e-01 -1.51920378e-01 -1.68294415e-01 1.88652068e-01
-7.08549380e-01 2.29063407e-01 -2.04980206e-02 7.47892737e-01
6.37839735e-01 4.36403185e-01 -1.02641678e+00 9.18102443e-01
-1.84130743e-01 7.52197564e-01 3.99889171e-01 3.98027450e-01
8.48825425e-02 -1.22789049e+00 -4.69575137e-01 9.04335558e-01
-6.16982996e-01 -4.99962196e-02 2.29287475e-01 9.62416947e-01
3.09651375e-01 6.94692552e-01 1.12849273e-01 4.12388265e-01
5.16112447e-01 2.47669294e-02 6.64237857e-01 -4.27446753e-01
-5.64668737e-02 -1.45501643e-01 -3.68064120e-02 -4.08037722e-01
-7.39340246e-01 -8.99460077e-01 -7.86131918e-01 -3.57239813e-01
-2.24931955e-01 3.51599604e-01 1.52882621e-01 9.81090188e-01
1.75248653e-01 6.49171054e-01 8.81982028e-01 -6.09903753e-01
-1.31612194e+00 -1.21430135e+00 -6.33954585e-01 5.52450180e-01
3.38254064e-01 -8.70168984e-01 -4.33664858e-01 -2.29649484e-01]
|
[8.081954002380371, 5.454188823699951]
|
f4da690f-9679-4a9d-b82a-ffa1f9009e98
|
a-benchmark-of-pdf-information-extraction
|
2303.09957
| null |
https://arxiv.org/abs/2303.09957v1
|
https://arxiv.org/pdf/2303.09957v1.pdf
|
A Benchmark of PDF Information Extraction Tools using a Multi-Task and Multi-Domain Evaluation Framework for Academic Documents
|
Extracting information from academic PDF documents is crucial for numerous indexing, retrieval, and analysis use cases. Choosing the best tool to extract specific content elements is difficult because many, technically diverse tools are available, but recent performance benchmarks are rare. Moreover, such benchmarks typically cover only a few content elements like header metadata or bibliographic references and use smaller datasets from specific academic disciplines. We provide a large and diverse evaluation framework that supports more extraction tasks than most related datasets. Our framework builds upon DocBank, a multi-domain dataset of 1.5M annotated content elements extracted from 500K pages of research papers on arXiv. Using the new framework, we benchmark ten freely available tools in extracting document metadata, bibliographic references, tables, and other content elements from academic PDF documents. GROBID achieves the best metadata and reference extraction results, followed by CERMINE and Science Parse. For table extraction, Adobe Extract outperforms other tools, even though the performance is much lower than for other content elements. All tools struggle to extract lists, footers, and equations. We conclude that more research on improving and combining tools is necessary to achieve satisfactory extraction quality for most content elements. Evaluation datasets and frameworks like the one we present support this line of research. We make our data and code publicly available to contribute toward this goal.
|
['Bela Gipp', 'Jelena Mitrović', 'Timo Spinde', 'Apurva Jagdale', 'Norman Meuschke']
|
2023-03-17
| null | null | null | null |
['table-extraction']
|
['miscellaneous']
|
[-3.48856360e-01 -6.53312877e-02 -5.16439855e-01 3.48412544e-02
-1.32499707e+00 -1.26153612e+00 8.50706041e-01 4.85091865e-01
-3.88444930e-01 1.00144911e+00 4.05934721e-01 -5.44831455e-01
-5.84483564e-01 -7.90319860e-01 -7.57717490e-01 -9.96426120e-02
3.84545535e-01 5.87347567e-01 4.15231496e-01 2.38813281e-01
1.14025223e+00 6.43639743e-01 -1.39161551e+00 2.00432837e-01
1.04393423e+00 6.80771410e-01 7.68986270e-02 5.61914384e-01
-1.05162024e+00 6.51337147e-01 -8.48890543e-01 -8.16557169e-01
1.42921116e-02 -1.32574081e-01 -1.13306355e+00 -6.64966524e-01
1.01864874e+00 -7.13303015e-02 -3.22044045e-01 1.00757134e+00
5.66628754e-01 -4.08095628e-01 7.90442288e-01 -1.27640676e+00
-6.40722573e-01 1.06435394e+00 -4.15234864e-01 5.69245338e-01
6.44198835e-01 -2.01708317e-01 9.89935338e-01 -8.25225115e-01
1.17942739e+00 1.08046961e+00 6.17011964e-01 3.76878008e-02
-8.48123491e-01 -8.36050928e-01 -4.15522128e-01 2.87104845e-01
-1.34186769e+00 -8.20405006e-01 2.79756606e-01 -5.94355106e-01
8.59151900e-01 4.37586397e-01 2.60870636e-01 1.13203418e+00
1.99756280e-01 4.28799421e-01 1.17415762e+00 -4.68546540e-01
3.39737311e-02 3.94879878e-01 5.92836320e-01 4.79554057e-01
1.01175427e+00 -7.84691751e-01 -7.11287558e-01 -2.35350907e-01
4.41517174e-01 -3.17494035e-01 -1.33184597e-01 -5.68875894e-02
-1.41234303e+00 2.10843340e-01 -4.25901473e-01 6.05392635e-01
-2.14096159e-01 -2.65251875e-01 4.80690151e-01 1.92471430e-01
3.78498733e-02 8.86499763e-01 -4.54586029e-01 -6.25513256e-01
-1.43359768e+00 6.72057211e-01 1.51259875e+00 1.55564642e+00
3.74172270e-01 -6.27533853e-01 -4.02616680e-01 5.14508247e-01
4.40283231e-02 4.70428795e-01 1.71823516e-01 -1.54709780e+00
8.75204802e-01 7.74512470e-01 1.38689816e-01 -1.12826478e+00
-3.61797482e-01 -4.22116458e-01 -3.28167468e-01 -1.88426927e-01
9.10022795e-01 4.34293635e-02 -4.31132972e-01 9.84241545e-01
9.62179825e-02 -5.14042974e-01 -7.89348260e-02 3.93733025e-01
1.59397376e+00 4.77828801e-01 1.56733505e-02 -4.77558794e-03
1.63255417e+00 -5.17085075e-01 -1.08149147e+00 2.47049510e-01
6.82279766e-01 -1.44882214e+00 7.56121457e-01 4.86956686e-01
-1.55631256e+00 -1.91267729e-01 -7.69671917e-01 -6.05157375e-01
-9.85922933e-01 3.35796326e-02 4.89393741e-01 5.55828273e-01
-8.99570346e-01 6.99581385e-01 -4.18850869e-01 -3.54503214e-01
5.34438789e-01 -1.56205431e-01 -2.94391096e-01 -1.78488404e-01
-8.81965995e-01 9.17549491e-01 1.92365617e-01 -6.30603373e-01
1.94334798e-03 -1.26496208e+00 -4.77599978e-01 2.05697954e-01
5.68400383e-01 -5.84897101e-01 1.17680252e+00 1.85754716e-01
-9.39986587e-01 1.12170208e+00 7.60200694e-02 -2.73782849e-01
4.75119323e-01 -2.26677552e-01 -4.60449040e-01 2.01461583e-01
4.11791474e-01 2.14041427e-01 9.73476693e-02 -6.86530769e-01
-5.59895575e-01 -4.31958258e-01 -2.81790376e-01 -2.58958310e-01
-6.02462947e-01 4.98710483e-01 -1.02230036e+00 -6.52182341e-01
-2.14282572e-01 -4.95265365e-01 3.98102492e-01 -3.02278101e-01
-6.22814238e-01 -5.03192365e-01 4.88594204e-01 -1.06918788e+00
1.69335973e+00 -1.73175633e+00 1.21323377e-01 8.08031783e-02
4.11737174e-01 -7.62370601e-02 1.94780782e-01 6.59474611e-01
4.12576944e-01 7.76945531e-01 2.05000177e-01 7.16781020e-02
5.15152514e-01 -3.69286150e-01 -4.81896400e-01 1.17763020e-01
-1.28501266e-01 8.02345514e-01 -6.44105732e-01 -1.11697388e+00
-8.85651186e-02 3.31773520e-01 -4.76376683e-01 -2.37627819e-01
-2.62880921e-01 -1.17162578e-01 -6.70099914e-01 1.04953480e+00
6.22966230e-01 -5.14731765e-01 -2.29370054e-02 -2.57320642e-01
-5.65637171e-01 8.46045077e-01 -1.55363023e+00 1.67390561e+00
-1.85454801e-01 8.77002001e-01 8.98816809e-02 -4.53006655e-01
8.28045607e-01 1.10528000e-01 5.01242757e-01 -6.40033305e-01
-4.84553091e-02 5.91933370e-01 -8.84851888e-02 -4.55876291e-01
8.62119675e-01 8.55659842e-01 -2.58273482e-01 3.99250925e-01
3.13937634e-01 -4.38338757e-01 1.04305446e+00 9.23192620e-01
1.45796609e+00 3.27728331e-01 1.43206611e-01 -5.59278786e-01
5.03481567e-01 2.42299482e-01 2.83757567e-01 1.00412142e+00
1.06163532e-01 3.00103933e-01 9.00195003e-01 -6.89737499e-02
-1.34834242e+00 -8.57518315e-01 -8.17419767e-01 8.07684064e-01
-3.74731213e-01 -1.14087689e+00 -8.71750176e-01 -3.89381737e-01
2.39732742e-01 7.41422594e-01 -1.67598248e-01 5.07313907e-01
-4.99825716e-01 -5.10133684e-01 6.79827988e-01 2.79648572e-01
2.34520778e-01 -9.52901065e-01 -3.17311555e-01 1.57855034e-01
-3.17682996e-02 -1.31902719e+00 -1.09259963e-01 5.08345924e-02
-6.53339565e-01 -1.15255284e+00 -8.30556870e-01 -3.48836005e-01
1.37693569e-01 -1.99248686e-01 1.58361399e+00 -2.28792708e-02
-3.56803983e-01 5.56568444e-01 -1.35076702e-01 -6.91437066e-01
-3.43275666e-01 7.17803121e-01 -2.97976315e-01 -1.26266313e+00
7.02015400e-01 -3.44461769e-01 -1.22747459e-01 -9.28676575e-02
-8.21423888e-01 -2.33196139e-01 5.55493295e-01 3.29253167e-01
5.59628367e-01 -1.12466082e-01 5.78551650e-01 -9.89077032e-01
6.50650203e-01 -6.63586378e-01 -9.18033063e-01 3.61672640e-01
-9.03694630e-01 1.38110846e-01 6.65825903e-01 2.13504389e-01
-8.91489565e-01 -5.29816389e-01 6.62208721e-02 1.89914808e-01
-1.45564705e-01 5.66756248e-01 -2.54561275e-01 1.36499330e-01
5.41060090e-01 -1.71698675e-01 -4.08760607e-01 -1.12189400e+00
3.27552289e-01 9.23296571e-01 8.10414135e-01 -9.40462649e-01
7.77743697e-01 -7.93715045e-02 2.16056377e-01 -5.23924053e-01
-8.73280466e-01 -4.77591127e-01 -7.00789154e-01 -6.21801068e-04
4.58374143e-01 -7.93078959e-01 -7.70291269e-01 -6.49112277e-04
-1.05610621e+00 1.54074609e-01 -1.35320827e-01 3.57175469e-01
-8.91683921e-02 3.08337957e-01 -7.65264452e-01 -2.23240599e-01
-3.74042511e-01 -1.06266034e+00 1.03151429e+00 4.77627814e-01
-4.30538327e-01 -8.04511011e-01 5.02694286e-02 5.51304281e-01
3.48544449e-01 1.43009543e-01 9.40205753e-01 -9.27114964e-01
-9.24857616e-01 -2.25238621e-01 -5.02655864e-01 -2.17040196e-01
-2.41933838e-01 8.97676766e-01 -5.30566752e-01 2.02896252e-01
-5.85016549e-01 -2.18671069e-01 8.68880332e-01 2.00182572e-01
1.73999119e+00 -4.14038718e-01 -6.72716022e-01 5.37895977e-01
1.31236053e+00 1.00287879e-02 6.91967368e-01 1.08965600e+00
6.67571664e-01 6.41252041e-01 4.04400378e-01 5.10552585e-01
5.24689794e-01 4.41483557e-01 -6.64272010e-02 5.88146925e-01
-2.28157490e-01 1.10992625e-01 8.99788141e-02 1.14722204e+00
2.96991132e-02 -3.56917858e-01 -1.49299729e+00 4.91700619e-01
-1.24633098e+00 -1.02237606e+00 -6.18851125e-01 1.86585116e+00
1.41383803e+00 4.65128511e-01 4.61899377e-02 2.57424951e-01
2.55857557e-01 -2.04268798e-01 -1.46242872e-01 -2.43103787e-01
-2.12493464e-01 6.20863557e-01 7.10231543e-01 1.73895627e-01
-7.84434974e-01 7.63786793e-01 6.23410892e+00 9.01701629e-01
-6.16363704e-01 -2.08102420e-01 2.10349485e-01 -2.15815023e-01
-4.53010648e-01 4.36929464e-02 -1.60527599e+00 7.89370835e-01
1.59539497e+00 -6.67801082e-01 4.79629114e-02 8.77827823e-01
-1.70754522e-01 -2.35568032e-01 -1.14895344e+00 7.68243849e-01
-1.91591963e-01 -1.97721875e+00 -6.16741478e-02 2.14177370e-01
4.47627008e-01 -1.02905087e-01 2.29415251e-03 4.13311809e-01
4.60039467e-01 -1.01413536e+00 7.22122073e-01 9.63097036e-01
6.70986295e-01 -6.81872904e-01 7.65400231e-01 -6.63338602e-02
-5.60216725e-01 3.59638065e-01 -3.78476739e-01 2.36501649e-01
-2.61666536e-01 1.01572084e+00 -3.05349559e-01 6.87880158e-01
1.15051651e+00 7.98841238e-01 -1.11948919e+00 1.40011382e+00
7.80686140e-02 5.65952241e-01 -1.18476592e-01 -1.75903097e-01
-1.22193120e-01 -2.18634009e-01 5.26008189e-01 1.78882825e+00
4.66720879e-01 -1.01690412e-01 -2.03791142e-01 9.11226690e-01
-6.07356429e-01 4.05121833e-01 -5.01158834e-01 -4.79168653e-01
1.14305818e+00 1.54702485e+00 -9.17012751e-01 -6.58354223e-01
-4.77056116e-01 2.14566857e-01 1.93911791e-01 -5.21824183e-03
-6.31046057e-01 -1.13261700e+00 3.79451513e-01 1.12246923e-01
1.65418267e-01 -2.57564187e-01 -7.47668684e-01 -1.10297441e+00
1.65487573e-01 -1.18390596e+00 5.22041559e-01 -7.80445099e-01
-1.27028382e+00 1.99809253e-01 8.36346671e-02 -8.05087388e-01
-3.81791562e-01 -7.11212754e-01 -8.41558948e-02 1.02560532e+00
-1.25039470e+00 -4.35559273e-01 -4.10889626e-01 1.35575384e-01
1.02777116e-01 -2.83284485e-01 7.16754973e-01 7.19991267e-01
-8.50358486e-01 5.09017766e-01 5.32464445e-01 3.65087479e-01
1.37356353e+00 -1.50246382e+00 2.94967979e-01 5.46198964e-01
1.22788563e-01 1.02908087e+00 5.38562894e-01 -8.54286551e-01
-1.80712759e+00 -6.42437577e-01 1.23118448e+00 -1.17176807e+00
1.04758441e+00 -2.04787418e-01 -1.14209461e+00 5.61832666e-01
5.78408837e-01 -3.82377416e-01 8.56835186e-01 1.64589331e-01
-3.62966567e-01 -1.79461893e-02 -9.72298563e-01 4.98407632e-01
1.02088487e+00 -3.41912210e-01 -8.07348490e-01 5.59579134e-01
5.71663558e-01 -6.64612532e-01 -1.67250717e+00 1.21345706e-01
5.42015076e-01 -5.75034261e-01 1.07143986e+00 -5.15893996e-01
9.86241221e-01 -1.11418620e-01 8.98391660e-03 -7.56941080e-01
-2.43527696e-01 -4.72108722e-01 -5.91213942e-01 2.05477023e+00
3.79497051e-01 -2.58072019e-01 6.02181852e-01 7.58328259e-01
-1.81667134e-01 -4.06942010e-01 -4.52569544e-01 -9.07442808e-01
4.76561904e-01 -2.91166157e-01 7.36614823e-01 1.09373009e+00
6.10897802e-02 3.00949663e-01 4.21280921e-01 -3.71926099e-01
8.06843579e-01 3.19668531e-01 9.10511136e-01 -1.64625037e+00
3.25753808e-01 -9.24592912e-01 -6.69732764e-02 -2.45219916e-01
2.02658027e-01 -1.24778247e+00 -6.45723045e-01 -1.96358263e+00
4.27725792e-01 -5.47170281e-01 1.90270902e-03 5.14783204e-01
-3.51618789e-02 9.26003680e-02 5.01125976e-02 6.16655767e-01
-7.56606042e-01 -2.04892009e-01 1.03603446e+00 5.83327897e-02
4.95920144e-02 -5.71936667e-01 -9.72866356e-01 6.93430245e-01
5.12184739e-01 -6.48439705e-01 2.57506520e-01 -1.64817259e-01
5.95168412e-01 -5.94725385e-02 2.58952290e-01 -1.28289521e+00
5.38859069e-01 -2.87555873e-01 8.59952629e-01 -1.00765455e+00
-3.24221879e-01 -4.05936658e-01 1.48092424e-02 -6.85829744e-02
-2.75555521e-01 1.27911791e-01 4.46920812e-01 -7.22096637e-02
-3.77456956e-02 -5.57837427e-01 3.53973031e-01 -3.34229290e-01
-4.24590260e-01 3.90651543e-03 -3.01226139e-01 7.45146215e-01
4.83951747e-01 1.46481216e-01 -9.60228384e-01 1.43445060e-01
-2.24629790e-01 1.54498562e-01 8.14056158e-01 3.45653385e-01
-5.54690138e-02 -1.04801500e+00 -6.50562882e-01 -5.07123828e-01
-1.11612789e-01 -2.32565731e-01 -3.34856093e-01 7.35542059e-01
-6.67046487e-01 1.02755022e+00 -5.37613988e-01 -1.64554000e-01
-9.59228039e-01 5.18230557e-01 -4.12408382e-01 -3.57612312e-01
-6.06376767e-01 3.92714888e-01 -7.22179711e-01 -1.90941438e-01
4.97392595e-01 -3.21751922e-01 -5.06805897e-01 4.74144757e-01
8.20415437e-01 9.09971893e-01 5.05573988e-01 -1.79629669e-01
-4.69493210e-01 2.93419987e-01 -2.13481024e-01 -5.38254939e-02
1.57051575e+00 3.57344002e-02 -4.11538780e-01 4.58451271e-01
9.74837720e-01 7.36056089e-01 -4.39278305e-01 -1.20100908e-01
7.29662001e-01 -3.05126429e-01 1.58419371e-01 -1.00110424e+00
-7.01739550e-01 5.72938502e-01 1.37051698e-04 2.95445949e-01
7.56632268e-01 -3.80362794e-02 5.22316694e-01 3.75686884e-01
1.20187446e-01 -1.24883068e+00 -2.65156955e-01 5.23473740e-01
6.81608081e-01 -9.72630262e-01 6.16730690e-01 -3.30814213e-01
1.13161035e-01 1.43912494e+00 5.51118076e-01 3.30424160e-01
4.51906532e-01 7.86973476e-01 -3.75439823e-01 -4.58064497e-01
-5.74483454e-01 1.47135392e-01 6.59831822e-01 2.21730843e-01
1.05943024e+00 -3.96125466e-01 -7.35077560e-01 8.15589905e-01
-6.44628584e-01 2.41220474e-01 8.53713572e-01 1.04267704e+00
-2.57269442e-01 -1.33897448e+00 -8.24689507e-01 9.08864379e-01
-1.29734695e+00 -3.44663888e-01 -6.80674016e-01 1.08642900e+00
-2.38354579e-01 4.77203041e-01 1.43261522e-01 2.72832543e-01
1.96471810e-01 3.62122625e-01 6.97058737e-01 -4.55573201e-01
-7.18894124e-01 -1.53326586e-01 2.23452747e-01 -4.38460410e-01
-2.57383913e-01 -9.58242536e-01 -1.26396191e+00 -9.39764202e-01
1.64326161e-01 3.17418784e-01 9.95388031e-01 5.85731208e-01
7.17418253e-01 7.98441589e-01 -3.06660742e-01 -3.52548927e-01
-2.28990003e-01 -8.75014663e-01 -2.97054917e-01 1.20422788e-01
-9.62969661e-02 -6.41357064e-01 -2.05809325e-01 2.41660520e-01]
|
[9.587217330932617, 8.165563583374023]
|
3d33e5a3-78f1-4f2f-b58d-bd46d1adc9b7
|
tell-me-why-you-feel-that-way-processing
|
2103.05815
| null |
https://arxiv.org/abs/2103.05815v1
|
https://arxiv.org/pdf/2103.05815v1.pdf
|
Tell Me Why You Feel That Way: Processing Compositional Dependency for Tree-LSTM Aspect Sentiment Triplet Extraction (TASTE)
|
Sentiment analysis has transitioned from classifying the sentiment of an entire sentence to providing the contextual information of what targets exist in a sentence, what sentiment the individual targets have, and what the causal words responsible for that sentiment are. However, this has led to elaborate requirements being placed on the datasets needed to train neural networks on the joint triplet task of determining an entity, its sentiment, and the causal words for that sentiment. Requiring this kind of data for training systems is problematic, as they suffer from stacking subjective annotations and domain over-fitting leading to poor model generalisation when applied in new contexts. These problems are also likely to be compounded as we attempt to jointly determine additional contextual elements in the future. To mitigate these problems, we present a hybrid neural-symbolic method utilising a Dependency Tree-LSTM's compositional sentiment parse structure and complementary symbolic rules to correctly extract target-sentiment-cause triplets from sentences without the need for triplet training data. We show that this method has the potential to perform in line with state-of-the-art approaches while also simplifying the data required and providing a degree of interpretability through the Tree-LSTM.
|
['S. Wermter', 'S. Magg', 'T. Hellström', 'S. Bensch', 'A. Sutherland']
|
2021-03-10
| null | null | null | null |
['aspect-sentiment-triplet-extraction']
|
['natural-language-processing']
|
[ 6.46643519e-01 4.10089076e-01 6.64220154e-02 -9.71890330e-01
-7.81241059e-01 -8.00079226e-01 8.52185488e-01 5.18294811e-01
-2.56509483e-01 8.31279814e-01 5.56012213e-01 -5.20633817e-01
-5.93112521e-02 -5.33630013e-01 -6.77478373e-01 -4.59405690e-01
1.23624668e-01 4.11508232e-01 6.55468106e-02 -4.07420307e-01
2.26914853e-01 8.61204974e-03 -1.46614742e+00 6.90156579e-01
3.94407868e-01 1.05021513e+00 2.52984837e-02 5.76153576e-01
-4.65597242e-01 1.23931730e+00 -8.12487602e-01 -7.38231063e-01
-2.12029904e-01 -3.95621538e-01 -9.23497736e-01 3.31741795e-02
3.01457077e-01 8.02822337e-02 5.79909563e-01 8.70785952e-01
7.22276866e-02 -1.08398467e-01 7.53253281e-01 -8.67503643e-01
-3.58457237e-01 8.49079669e-01 -2.42011145e-01 1.58632696e-01
3.52368236e-01 -2.53214687e-01 1.27596259e+00 -6.01049960e-01
6.30110443e-01 1.36639738e+00 7.52285182e-01 4.67549056e-01
-1.21045756e+00 -4.15587544e-01 3.23924333e-01 -1.58807829e-01
-5.28747976e-01 -7.36218750e-01 6.47328973e-01 -4.94650006e-01
1.46081543e+00 1.77211136e-01 2.60981053e-01 1.27390742e+00
1.52417541e-01 5.38892567e-01 1.01400781e+00 -7.40687549e-01
1.16711468e-01 4.55470264e-01 3.03090304e-01 3.95175368e-01
1.31305456e-01 -1.78546414e-01 -6.07734501e-01 -2.79052947e-02
3.05028651e-02 -4.68360960e-01 1.56229392e-01 -9.33103412e-02
-9.34755921e-01 8.79472733e-01 3.03682894e-01 5.55108488e-01
-3.06089669e-01 1.60083473e-01 8.12040746e-01 1.17645502e-01
6.39369488e-01 6.74472392e-01 -1.09920835e+00 -2.38184363e-01
-8.92445743e-01 3.54466408e-01 1.07231331e+00 5.56722939e-01
5.50182819e-01 -6.63379356e-02 2.41567478e-01 6.28941119e-01
3.95684659e-01 1.22519873e-01 4.45976853e-01 -8.70635331e-01
6.41996205e-01 7.42414176e-01 1.36686385e-01 -1.07576907e+00
-4.60827917e-01 -2.68228322e-01 -8.83661434e-02 -4.03285474e-02
4.64610130e-01 -3.66751283e-01 -9.43048060e-01 1.89746106e+00
1.91921234e-01 -3.13095659e-01 3.56569469e-01 4.74092960e-01
5.22607684e-01 5.29973090e-01 3.88804585e-01 -2.08705604e-01
1.33153284e+00 -4.33955401e-01 -6.98934615e-01 -7.92791605e-01
1.04600096e+00 -8.64402592e-01 9.22956824e-01 3.32966805e-01
-9.04236078e-01 -2.29096144e-01 -1.12943077e+00 -1.44709662e-01
-6.46629274e-01 -6.39648363e-02 8.03922355e-01 6.86125934e-01
-7.02142000e-01 6.52765512e-01 -7.41621137e-01 -2.66989946e-01
1.81876123e-01 5.55968583e-01 -2.25177184e-01 2.30650648e-01
-1.37498665e+00 1.43561339e+00 4.44483072e-01 2.11120740e-01
-1.02531321e-01 -5.51288843e-01 -9.89301205e-01 -1.80354081e-02
2.36487493e-01 -4.37050402e-01 1.44326031e+00 -1.44393277e+00
-1.14666772e+00 7.37355590e-01 -4.73448247e-01 -2.81521618e-01
1.06378868e-01 -2.50179261e-01 -4.18025434e-01 -3.66411716e-01
4.51398462e-01 4.40217733e-01 6.33817375e-01 -1.25151372e+00
-9.48437035e-01 -5.13315380e-01 1.27588123e-01 1.68435887e-01
-4.01568890e-01 5.45884848e-01 -1.15617074e-01 -2.89220691e-01
-1.90719783e-01 -9.41496909e-01 -1.58574030e-01 -8.30794096e-01
-3.44789356e-01 -3.08466524e-01 8.80542934e-01 -6.54187858e-01
1.20103979e+00 -1.99028313e+00 6.98594153e-02 1.08902656e-01
-2.22969219e-01 1.62019029e-01 2.16041937e-01 4.64642674e-01
-3.42101306e-01 2.46121913e-01 -2.39260271e-01 -5.61349630e-01
-6.71045035e-02 5.32505870e-01 -7.30555356e-01 5.01258858e-02
7.10947514e-01 5.50762057e-01 -8.20286155e-01 -4.48089600e-01
2.86976565e-02 3.97259086e-01 -2.21907005e-01 -2.21966151e-02
-4.90716398e-01 2.27068305e-01 -4.19307679e-01 2.93456823e-01
3.18692322e-03 -5.56558892e-02 4.60083544e-01 -1.13339804e-01
-2.64554601e-02 1.03915358e+00 -9.68987346e-01 1.10280359e+00
-6.33258402e-01 5.16990244e-01 5.05474210e-02 -9.95822370e-01
9.17279124e-01 4.34835613e-01 1.21122964e-01 -4.78186905e-01
3.10293257e-01 6.53601170e-01 9.62744579e-02 -6.65754914e-01
5.59760451e-01 -8.19216371e-01 -5.09274125e-01 4.50232685e-01
7.64924241e-03 -1.87107280e-01 4.65315163e-01 9.13401172e-02
8.89137208e-01 5.17056942e-01 1.65140152e-01 -1.36171266e-01
5.62962890e-01 3.43577504e-01 6.19178355e-01 3.07949752e-01
2.13237047e-01 1.63967639e-01 9.38763618e-01 -3.95387381e-01
-9.64885235e-01 -3.56239855e-01 -1.08452201e-01 1.12882590e+00
-4.02771205e-01 -3.61651599e-01 -5.87338746e-01 -8.12155962e-01
-2.89421558e-01 1.32533181e+00 -8.32961202e-01 1.51462089e-02
-6.81681991e-01 -8.61612022e-01 4.70324337e-01 5.38627386e-01
-1.84335843e-01 -1.19441235e+00 -6.63332403e-01 3.96871507e-01
-3.55734169e-01 -1.33326817e+00 1.71849608e-01 8.65648687e-01
-7.12981343e-01 -9.82898474e-01 -3.36368829e-02 -6.11386001e-01
7.37046838e-01 -4.04227734e-01 1.10652912e+00 -1.54044762e-01
2.38588572e-01 -8.63495544e-02 -3.97323996e-01 -8.14944804e-01
-8.57565343e-01 5.33737130e-02 -1.45423099e-01 -2.65339892e-02
6.73855782e-01 -4.17774439e-01 -7.17462301e-02 -1.39425814e-01
-9.40472722e-01 -2.29546521e-02 3.87752146e-01 7.51421034e-01
7.92075172e-02 1.36150599e-01 6.21894956e-01 -1.37243342e+00
7.98913121e-01 -5.33933938e-01 -2.63505608e-01 1.47445664e-01
-6.04429722e-01 2.89383709e-01 8.91758621e-01 -1.52371243e-01
-1.27007806e+00 2.43063837e-01 -1.36830524e-01 2.32585788e-01
-3.14882338e-01 9.73827064e-01 -8.20455626e-02 5.30861676e-01
7.11307108e-01 -1.86107442e-01 -8.33645388e-02 -2.16242000e-01
2.43841246e-01 7.52242148e-01 3.63361388e-01 -5.44337153e-01
3.07059258e-01 2.83662409e-01 6.97212964e-02 -5.12664914e-01
-1.52816904e+00 -3.12366337e-01 -7.02602983e-01 -2.04500761e-02
8.49165201e-01 -8.35216582e-01 -1.45094439e-01 2.92637676e-01
-1.32314408e+00 -2.67635882e-01 -2.28255782e-02 1.01124950e-01
-3.80253911e-01 1.63889155e-01 -3.12769145e-01 -8.08593512e-01
-1.42021343e-01 -1.10732186e+00 9.98062193e-01 -1.33081213e-01
-9.32257235e-01 -1.24254477e+00 -5.96244745e-02 5.34471393e-01
1.70778662e-01 4.04570341e-01 1.27565730e+00 -9.78120029e-01
8.96439776e-02 -5.93872726e-01 -1.33689806e-01 5.48790276e-01
2.52977669e-01 2.01106846e-01 -9.10859406e-01 1.01874039e-01
2.85593271e-01 -5.02086341e-01 5.36998868e-01 2.15015471e-01
4.07805681e-01 -3.97374809e-01 -2.55614430e-01 -3.52641866e-02
1.27824020e+00 1.13026477e-01 1.50861368e-01 5.67243338e-01
6.20522022e-01 1.28468132e+00 6.27343893e-01 -7.28685930e-02
6.01494372e-01 5.74964106e-01 3.12505603e-01 -8.77394378e-02
1.37382582e-01 -2.75286306e-02 6.02734387e-01 5.01329541e-01
3.70151490e-01 -1.73322439e-01 -9.44538593e-01 8.15491378e-01
-1.90044379e+00 -7.51541555e-01 -5.71179032e-01 1.78266871e+00
1.13198698e+00 5.36038518e-01 -3.51593345e-02 3.12163025e-01
4.72874731e-01 1.01661064e-01 -2.87959397e-01 -1.02085567e+00
2.56591174e-03 3.32622789e-02 4.64127511e-01 7.38389611e-01
-1.08147120e+00 1.04643154e+00 6.04267693e+00 3.55301023e-01
-1.22805369e+00 -1.02141812e-01 7.42220640e-01 3.10072899e-02
-3.85428220e-01 3.99365991e-01 -1.00797367e+00 3.16792786e-01
1.33447230e+00 3.15002054e-01 1.05522536e-01 6.74344361e-01
4.41965133e-01 -3.74908239e-01 -1.23862326e+00 2.45609239e-01
1.61710948e-01 -1.06143749e+00 -1.23186223e-01 -1.03821762e-01
5.22833467e-01 4.10038792e-02 -2.10704610e-01 2.34687045e-01
4.00144517e-01 -9.56382692e-01 9.33998048e-01 2.34854087e-01
4.38037395e-01 -6.33055925e-01 8.99604738e-01 4.01526242e-01
-6.11670971e-01 -1.85810015e-01 -1.71256792e-02 -5.37571132e-01
2.64910042e-01 6.75904274e-01 -1.16473472e+00 3.63915145e-01
5.33484876e-01 6.69021070e-01 -6.02173030e-01 1.14849746e-01
-5.27888894e-01 6.06200755e-01 -3.51348668e-01 -4.06298757e-01
3.96993250e-01 1.48761451e-01 3.23025137e-01 1.44613981e+00
4.47604917e-02 -1.43233359e-01 -1.63948938e-01 5.63978553e-01
2.75768369e-01 3.93850096e-02 -5.46506286e-01 -1.86017677e-01
1.24889322e-01 1.29455006e+00 -9.08802867e-01 -4.50542212e-01
-4.19613868e-01 5.52542925e-01 3.52884114e-01 1.50110170e-01
-4.41868395e-01 -2.08776683e-01 4.52258229e-01 -6.36864733e-03
4.99079913e-01 -1.00250818e-01 -8.24218452e-01 -9.96198356e-01
1.42747536e-01 -1.04021478e+00 2.26222441e-01 -8.48128676e-01
-1.14701021e+00 6.34659946e-01 -1.10466659e-01 -5.36837876e-01
-7.21151292e-01 -8.40298653e-01 -6.02563083e-01 9.62717652e-01
-1.43127728e+00 -1.33310878e+00 3.80001426e-01 1.98003694e-01
4.13801253e-01 1.20128259e-01 1.10865319e+00 5.27389646e-02
-3.78022134e-01 1.35059044e-01 -3.76435310e-01 7.90573955e-02
6.71672642e-01 -1.38394010e+00 4.39279169e-01 1.02561855e+00
4.70918156e-02 7.64182389e-01 1.20718825e+00 -7.09192276e-01
-1.03815472e+00 -8.94152820e-01 1.83691180e+00 -9.38757718e-01
1.10197198e+00 -3.85373622e-01 -6.96829855e-01 7.48919249e-01
1.79901272e-01 -5.06108642e-01 9.16110754e-01 6.15191340e-01
-4.07180786e-01 4.68319729e-02 -7.94488788e-01 5.38503349e-01
4.86863166e-01 -4.70744103e-01 -9.90857661e-01 1.32274166e-01
6.61058068e-01 -2.48380348e-01 -6.18466735e-01 4.17751878e-01
3.47262621e-01 -1.00617611e+00 3.86047244e-01 -6.76407754e-01
1.00694168e+00 -2.61962891e-01 -2.36631766e-01 -1.24456239e+00
-7.10813189e-03 -2.25888118e-01 2.48619735e-01 1.60394073e+00
1.11919880e+00 -4.58690852e-01 5.55141330e-01 1.14387918e+00
-2.17334852e-01 -7.94546783e-01 -8.25711787e-01 -2.23926410e-01
-3.50825004e-02 -8.75033319e-01 4.41258371e-01 9.47700202e-01
2.06181169e-01 9.56944168e-01 -1.57545730e-01 4.01570313e-02
2.71163881e-01 7.46158361e-02 5.04081249e-01 -1.07456422e+00
-1.88563719e-01 -5.01150906e-01 -7.50735998e-02 -4.82828975e-01
3.31879318e-01 -7.17344224e-01 1.38211712e-01 -1.69171154e+00
-1.85702056e-01 -4.34253365e-01 -4.87164557e-02 8.49832594e-01
-1.58469856e-01 -6.03119005e-03 1.37724951e-01 5.80023713e-02
-3.18472356e-01 4.84696738e-02 7.26790786e-01 1.43801719e-01
-1.07787758e-01 -7.95513019e-02 -1.20477438e+00 9.40357566e-01
7.37434328e-01 -6.72405124e-01 -4.18633163e-01 -4.96801794e-01
9.72233355e-01 -4.10783887e-02 2.05424681e-01 -5.72829783e-01
1.43583149e-01 -3.78344092e-03 3.95327747e-01 -3.36053997e-01
3.03543180e-01 -7.82553971e-01 -8.30268189e-02 1.71937212e-01
-6.74117208e-01 -8.95545632e-03 3.61092240e-01 2.64834255e-01
-3.50417674e-01 -4.86764103e-01 4.05001193e-01 -2.10929006e-01
-4.73573089e-01 -5.04698396e-01 -4.83970046e-01 -8.21966007e-02
7.23015964e-01 -2.59940445e-01 -7.68410265e-02 -2.57285595e-01
-7.16649890e-01 1.25439212e-01 3.21822882e-01 6.60246491e-01
1.95346802e-01 -6.92241073e-01 -5.41892111e-01 1.28250867e-01
-1.42083699e-02 1.24097113e-02 -1.90426663e-01 6.50025368e-01
-1.20358668e-01 6.21178925e-01 1.84337288e-01 -2.46432930e-01
-1.22908902e+00 3.18857670e-01 2.25399315e-01 -5.43122292e-01
-1.90330431e-01 9.41030025e-01 -2.91998815e-02 -4.66898829e-01
1.30636385e-02 -4.11760479e-01 -5.16261935e-01 5.30177295e-01
1.39871150e-01 -2.55060643e-01 3.39442044e-01 -8.39014232e-01
-3.94929796e-01 2.33587340e-01 -2.54331440e-01 -2.61067271e-01
1.53149676e+00 -2.25949600e-01 -4.50983375e-01 7.92982459e-01
1.14713645e+00 1.60929218e-01 -9.46721911e-01 2.33613960e-02
5.49163699e-01 3.55402427e-03 1.04057834e-01 -1.20545733e+00
-4.45071489e-01 6.61198556e-01 -1.73074141e-01 4.58730429e-01
9.00913417e-01 6.05830848e-02 7.55209386e-01 1.17270492e-01
-4.29493152e-02 -1.17959750e+00 -3.98822039e-01 7.80064642e-01
7.38870203e-01 -1.14731622e+00 -2.09984719e-03 -3.56257409e-01
-7.99379468e-01 1.31197739e+00 2.84762293e-01 -6.31560162e-02
4.74534333e-01 3.58704090e-01 4.49649006e-01 -4.54210460e-01
-1.01453424e+00 -1.00319847e-01 9.91162509e-02 2.44708434e-01
8.67076635e-01 -4.92248684e-02 -2.45098069e-01 4.98948425e-01
-4.44970161e-01 -1.82635874e-01 7.04659700e-01 8.14265370e-01
-1.35567054e-01 -1.44289684e+00 -3.42850387e-01 5.09233177e-01
-9.70794141e-01 -3.13387305e-01 -7.26271033e-01 5.17188072e-01
3.14485341e-01 1.16115355e+00 -1.30554825e-01 -2.19105095e-01
3.40175271e-01 3.98884237e-01 2.29315668e-01 -9.45358038e-01
-8.85878205e-01 2.44078875e-01 8.85405600e-01 -1.79005250e-01
-7.69938231e-01 -9.66552615e-01 -1.31611335e+00 1.45940647e-01
-4.05532151e-01 1.10322192e-01 1.10423934e+00 1.61143041e+00
2.48289794e-01 6.83934391e-01 3.91633153e-01 -7.84693539e-01
-5.43009460e-01 -9.44618404e-01 -1.14618041e-01 4.26507056e-01
4.09968674e-01 -3.94433796e-01 -3.20843309e-01 3.50294322e-01]
|
[11.223186492919922, 7.065085411071777]
|
0f0500d1-ed3d-43ea-a69a-b10df5af20c4
|
boundarycam-a-boundary-based-refinement
|
2303.07853
| null |
https://arxiv.org/abs/2303.07853v1
|
https://arxiv.org/pdf/2303.07853v1.pdf
|
BoundaryCAM: A Boundary-based Refinement Framework for Weakly Supervised Semantic Segmentation of Medical Images
|
Weakly Supervised Semantic Segmentation (WSSS) with only image-level supervision is a promising approach to deal with the need for Segmentation networks, especially for generating a large number of pixel-wise masks in a given dataset. However, most state-of-the-art image-level WSSS techniques lack an understanding of the geometric features embedded in the images since the network cannot derive any object boundary information from just image-level labels. We define a boundary here as the line separating an object and its background, or two different objects. To address this drawback, we propose our novel BoundaryCAM framework, which deploys state-of-the-art class activation maps combined with various post-processing techniques in order to achieve fine-grained higher-accuracy segmentation masks. To achieve this, we investigate a state-of-the-art unsupervised semantic segmentation network that can be used to construct a boundary map, which enables BoundaryCAM to predict object locations with sharper boundaries. By applying our method to WSSS predictions, we were able to achieve up to 10% improvements even to the benefit of the current state-of-the-art WSSS methods for medical imaging. The framework is open-source and accessible online at https://github.com/bharathprabakaran/BoundaryCAM.
|
['Muhammad Shafique', 'Erik Ostrowski', 'Bharath Srinivas Prabakaran']
|
2023-03-14
| null | null | null | null |
['unsupervised-semantic-segmentation']
|
['computer-vision']
|
[ 6.47762477e-01 5.46611071e-01 -5.15823811e-02 -5.14578342e-01
-8.69457662e-01 -3.86846870e-01 3.77239048e-01 2.63338327e-01
-4.30977076e-01 2.97226071e-01 -2.94183522e-01 -2.37238079e-01
1.17073722e-01 -8.04780126e-01 -7.98511803e-01 -6.46326900e-01
1.00233369e-01 6.64276421e-01 1.03494477e+00 -2.67262489e-01
1.76015601e-01 5.35198867e-01 -1.48496938e+00 5.58970571e-01
1.01601291e+00 1.10350478e+00 5.95864594e-01 5.51865160e-01
-4.11182910e-01 2.44042844e-01 -3.22179139e-01 -1.28280193e-01
3.24425966e-01 -4.43970442e-01 -1.04404294e+00 1.29381880e-01
4.22275543e-01 -7.35257640e-02 1.71586201e-01 1.23667800e+00
1.82666928e-01 -2.09146872e-01 5.38117409e-01 -9.30421293e-01
-1.47076041e-01 5.26008844e-01 -4.29765046e-01 5.27968742e-02
-1.41659603e-02 2.09926054e-01 7.35339344e-01 -5.55735946e-01
5.97095966e-01 7.29838431e-01 5.40072322e-01 6.17199183e-01
-1.23080337e+00 -3.27043235e-01 1.27505928e-01 1.05962917e-01
-1.36895275e+00 -2.91148394e-01 9.50171173e-01 -4.60767359e-01
5.81851900e-01 3.83091152e-01 6.19213104e-01 7.73710072e-01
-7.31550381e-02 9.39964414e-01 1.38212812e+00 -5.65687180e-01
2.49974430e-01 1.58132717e-01 1.86444536e-01 8.06174517e-01
1.01160988e-01 -2.41821349e-01 -1.62447035e-01 1.99736148e-01
9.21776116e-01 -5.58749922e-02 -2.46944845e-01 -4.78483438e-01
-1.10357070e+00 5.59844673e-01 8.83941531e-01 6.49309814e-01
-3.23967814e-01 -5.71376905e-02 -4.85235396e-05 -2.76806563e-01
6.91507578e-01 4.83615607e-01 -4.27927434e-01 3.70688170e-01
-1.46817994e+00 2.42472775e-02 5.73829293e-01 6.28605664e-01
8.51137221e-01 -3.98687869e-01 -1.81592733e-01 6.72108710e-01
2.96642572e-01 1.36482522e-01 3.40666741e-01 -8.98142815e-01
2.72340149e-01 9.01178539e-01 -2.12413855e-02 -7.82858074e-01
-6.57327712e-01 -5.08121431e-01 -5.42844296e-01 3.58114004e-01
6.94297850e-01 1.80698484e-01 -1.48348808e+00 1.33446503e+00
4.79374528e-01 4.31338638e-01 -1.56404153e-01 1.04943764e+00
8.90453577e-01 4.29665327e-01 -1.03377486e-02 1.45842105e-01
1.48904836e+00 -1.13806951e+00 -4.16960806e-01 -4.87084210e-01
7.79994130e-01 -5.18331170e-01 1.14530623e+00 2.95868695e-01
-1.13303542e+00 -4.42272246e-01 -1.03388965e+00 6.06885995e-04
-4.92238820e-01 7.42288232e-02 4.98342156e-01 6.23635828e-01
-1.24582601e+00 6.06078148e-01 -1.24728978e+00 -3.04756880e-01
8.46468806e-01 4.46895570e-01 -2.47055024e-01 1.75220761e-02
-9.44219947e-01 8.35197330e-01 4.26834226e-01 1.63407654e-01
-7.83674121e-01 -7.64812768e-01 -8.53406012e-01 8.53797272e-02
6.87549233e-01 -6.33943379e-01 1.08703470e+00 -1.20548391e+00
-1.23777485e+00 1.20327401e+00 -1.72110230e-01 -4.45100635e-01
7.79620588e-01 1.15438811e-01 -5.33516631e-02 5.28812826e-01
2.17641026e-01 9.59774315e-01 6.85332477e-01 -1.53833592e+00
-5.48701823e-01 -5.54032862e-01 -1.99382305e-01 9.83353928e-02
-5.13788238e-02 -8.19212645e-02 -7.57575989e-01 -6.01530612e-01
4.10925478e-01 -8.64931524e-01 -5.46696842e-01 5.70576862e-02
-7.40973055e-01 -6.86383322e-02 8.25130820e-01 -6.11046672e-01
9.35843110e-01 -1.96755469e+00 1.83304120e-02 2.40426674e-01
1.40726835e-01 4.78312671e-01 -8.81615803e-02 -7.34345987e-02
1.51012853e-01 1.51408195e-01 -8.62347305e-01 -5.47930479e-01
-3.40274721e-01 7.97864199e-02 1.31089047e-01 3.37339431e-01
2.95330614e-01 8.69243979e-01 -8.29852700e-01 -6.33371413e-01
4.67365056e-01 4.35251713e-01 -4.67481792e-01 1.41130164e-01
-5.01687169e-01 7.78004706e-01 -5.45855701e-01 6.41566455e-01
7.11738408e-01 -3.12136531e-01 6.15835600e-02 -2.57116616e-01
-1.00073151e-01 -2.55545601e-02 -1.13117325e+00 1.94120371e+00
-1.69527456e-01 3.56643766e-01 1.22470528e-01 -1.28025901e+00
6.82973206e-01 1.46964103e-01 6.96451724e-01 -5.87036312e-01
3.16084266e-01 4.52783823e-01 -1.24544196e-01 -4.21923757e-01
2.01416194e-01 -1.77143097e-01 8.31644144e-03 3.55164334e-02
9.17844996e-02 -3.78443003e-01 2.51708806e-01 2.24437900e-02
8.31685245e-01 2.69833714e-01 -3.29652503e-02 -3.77580643e-01
6.13467872e-01 3.15728575e-01 4.52000707e-01 9.28507388e-01
-1.56115964e-01 1.27941096e+00 4.49736059e-01 -3.76361996e-01
-8.82605255e-01 -1.05651510e+00 -2.97510833e-01 7.87622213e-01
4.30866569e-01 -1.30783901e-01 -1.39567578e+00 -9.25927579e-01
-2.91966230e-01 5.42838395e-01 -6.16307437e-01 1.02557756e-01
-4.90537435e-01 -6.62834942e-01 4.03798550e-01 4.12308544e-01
6.38475955e-01 -1.17950749e+00 -7.61950135e-01 2.55180091e-01
-2.95894742e-01 -1.40887976e+00 -2.29595587e-01 1.96305037e-01
-8.77345443e-01 -1.17223895e+00 -9.43373978e-01 -8.98613393e-01
1.02131569e+00 -5.32430522e-02 1.01740944e+00 3.62774104e-01
-5.18206775e-01 8.22977163e-03 -3.78290743e-01 -3.93615603e-01
-4.46382940e-01 2.90283233e-01 -5.21444380e-01 -1.32029792e-02
-2.76973769e-02 -3.17899615e-01 -9.73782539e-01 3.99634689e-01
-1.13423753e+00 4.73461717e-01 6.80867255e-01 5.94082594e-01
9.06489909e-01 -6.96011037e-02 3.79676431e-01 -1.22995949e+00
6.23883009e-02 -1.80704936e-01 -6.89035177e-01 3.47037375e-01
-4.78722841e-01 -2.04484221e-02 4.66086745e-01 -1.58071503e-01
-9.54292893e-01 4.19396788e-01 -4.50843036e-01 -1.34435922e-01
-6.61058962e-01 3.03016692e-01 1.17048845e-02 -1.69077575e-01
6.25793099e-01 2.45940641e-01 1.02588654e-01 -3.61540765e-01
4.35461670e-01 6.75447524e-01 6.39488220e-01 -3.52296710e-01
3.99917036e-01 9.43827808e-01 -8.66658911e-02 -7.22406864e-01
-1.15936661e+00 -7.96063006e-01 -1.03466082e+00 -2.47722149e-01
1.22707248e+00 -5.15189588e-01 -2.59087265e-01 6.58286750e-01
-9.95354354e-01 -8.71503949e-01 -3.82001966e-01 9.66378450e-02
-6.71877086e-01 2.19292268e-01 -5.70352972e-01 -4.32602376e-01
-2.96430081e-01 -1.38711810e+00 1.34053707e+00 3.02683145e-01
-6.09278865e-02 -9.47651923e-01 -2.42611513e-01 8.84745419e-01
3.51147383e-01 3.70672762e-01 6.43308938e-01 -6.21194184e-01
-7.90662169e-01 -8.01642612e-02 -2.36676946e-01 4.44372118e-01
1.01664633e-01 -1.53429419e-01 -9.46413219e-01 7.86952227e-02
-4.02963944e-02 -2.46008839e-02 1.01547825e+00 8.29468131e-01
1.43883860e+00 -3.60990800e-02 -6.20663762e-01 7.15833902e-01
1.34786355e+00 3.26154120e-02 6.23920441e-01 3.90381277e-01
9.80216622e-01 6.66428983e-01 5.21935523e-01 1.64392162e-02
4.41092163e-01 8.03342283e-01 4.06718403e-01 -7.30264366e-01
-4.38348174e-01 -9.34441388e-02 -2.07952484e-01 3.34057122e-01
9.64567065e-02 -1.04416840e-01 -1.14484251e+00 7.08141685e-01
-1.77970326e+00 -3.34586561e-01 -4.55397189e-01 1.94269097e+00
9.06609595e-01 4.82265294e-01 4.47189882e-02 9.38085914e-02
6.64160728e-01 9.61852744e-02 -6.11247182e-01 -1.22724168e-01
1.22739099e-01 2.78930902e-01 6.95221841e-01 5.07494092e-01
-1.22991908e+00 1.38283968e+00 5.64997244e+00 9.53919590e-01
-1.30473804e+00 2.09435821e-01 1.05006373e+00 1.73033386e-01
-1.46288544e-01 -4.01515514e-02 -6.44869864e-01 4.48485821e-01
6.62329197e-01 6.44027948e-01 1.38354421e-01 6.29386425e-01
2.71759361e-01 -4.58230525e-01 -8.86374295e-01 6.21916175e-01
1.12230271e-01 -1.54221272e+00 -8.29423070e-02 -1.38352036e-01
8.82218599e-01 2.37461492e-01 -1.62383735e-01 -1.26125157e-01
5.46870343e-02 -9.31317627e-01 7.72950053e-01 4.15558845e-01
6.75669074e-01 -2.86985308e-01 8.02788556e-01 4.72573251e-01
-1.04489172e+00 2.23320991e-01 -8.71745497e-02 2.01765403e-01
2.97907233e-01 7.48980582e-01 -8.10454309e-01 5.41952193e-01
7.62905717e-01 4.18819636e-01 -6.93354011e-01 1.11955106e+00
-3.36814225e-01 5.31769037e-01 -4.37595606e-01 3.15675110e-01
4.57236975e-01 -1.56965300e-01 2.73155570e-01 1.16653883e+00
2.09092736e-01 1.18619762e-01 3.32744539e-01 1.04288626e+00
3.57123576e-02 1.56413186e-02 -2.09069073e-01 2.73423046e-01
8.52079391e-02 1.33664823e+00 -1.63795769e+00 -5.20170271e-01
-1.95025161e-01 1.11865985e+00 1.60830766e-01 3.80497843e-01
-6.92279696e-01 -1.10537000e-01 1.79618910e-01 5.39951205e-01
3.02134931e-01 -2.16262549e-01 -7.45086789e-01 -9.19682741e-01
-9.85920653e-02 -4.68605161e-01 3.61716747e-01 -7.74871826e-01
-1.01228094e+00 7.16418564e-01 7.71765709e-02 -8.96201611e-01
2.09439516e-01 -6.53725088e-01 -5.67644715e-01 6.20336890e-01
-1.66812897e+00 -1.28694272e+00 -4.89956588e-01 4.45389092e-01
6.22765839e-01 2.55441576e-01 6.52161360e-01 1.04495034e-01
-3.09582114e-01 1.93367913e-01 -1.80442393e-01 1.61604285e-01
4.08334404e-01 -1.30049455e+00 3.31778407e-01 8.09308827e-01
2.83440143e-01 1.98408723e-01 5.93218744e-01 -6.50791705e-01
-6.81418419e-01 -9.88702297e-01 5.69781482e-01 -5.33334136e-01
4.80495721e-01 -4.81556535e-01 -9.66770589e-01 4.18430775e-01
-1.79950178e-01 5.13634264e-01 3.39134544e-01 -1.78445086e-01
1.54268369e-01 -3.02149523e-02 -1.19384050e+00 6.54311478e-01
1.06853306e+00 -2.95083195e-01 -4.81045157e-01 3.39679569e-01
5.75663030e-01 -7.20123649e-01 -5.63971937e-01 7.01783419e-01
2.05809340e-01 -1.15636897e+00 9.93236959e-01 -9.77137685e-02
4.11057740e-01 -4.68111753e-01 3.27838033e-01 -1.17063630e+00
6.58589378e-02 -1.47935927e-01 2.53587544e-01 1.02223122e+00
5.29184163e-01 -6.37880504e-01 1.13802433e+00 5.91580510e-01
-5.29113352e-01 -9.98383582e-01 -9.46939707e-01 -6.69563890e-01
-6.55582314e-03 -5.62745035e-01 3.02078724e-01 8.52824211e-01
-3.19146484e-01 -1.42699420e-01 1.71401456e-01 3.13984394e-01
5.56914270e-01 9.40515175e-02 4.09363598e-01 -1.12258601e+00
-3.91666181e-02 -6.50841773e-01 -4.67979729e-01 -8.97975266e-01
1.20802402e-01 -1.07165182e+00 2.86320210e-01 -2.07688546e+00
9.17199031e-02 -7.86267817e-01 -4.86778736e-01 6.15656376e-01
-2.05863670e-01 7.79679775e-01 1.51361465e-01 1.71362892e-01
-6.43208504e-01 2.84084022e-01 1.45463967e+00 -1.27171546e-01
-2.39635795e-01 1.72694102e-02 -5.59969306e-01 9.95341420e-01
8.36231649e-01 -6.28274560e-01 -4.00947809e-01 -2.90209949e-01
-3.69322181e-01 -2.93130223e-02 6.00492895e-01 -1.01249981e+00
3.02504689e-01 3.72168496e-02 3.23684454e-01 -4.38133001e-01
1.91078901e-01 -8.40130270e-01 6.45001559e-03 5.07832170e-01
-2.72683769e-01 -7.31883883e-01 1.73165500e-01 2.38038823e-01
-2.14554295e-01 -3.42550218e-01 7.89266944e-01 -4.03007776e-01
-7.04438925e-01 2.81351596e-01 -1.71584785e-01 -5.49716502e-02
1.22480226e+00 -4.32070583e-01 -1.78469464e-01 9.97396782e-02
-9.53210294e-01 2.24017635e-01 5.46341896e-01 2.55552024e-01
5.36606550e-01 -7.43069589e-01 -5.30404866e-01 3.35762501e-01
1.44998923e-01 6.54513836e-01 4.45020974e-01 1.00431454e+00
-7.43713856e-01 2.69534916e-01 -1.25094801e-01 -1.07546914e+00
-1.13774371e+00 2.46456593e-01 5.19585669e-01 -3.95621359e-01
-8.03491712e-01 8.96589637e-01 4.51313943e-01 -4.55012053e-01
-8.67508948e-02 -6.60021126e-01 -1.73394084e-01 -1.53545365e-01
2.31656283e-01 -6.78125173e-02 3.29253614e-01 -6.42818332e-01
-4.51275676e-01 7.15144992e-01 3.31801437e-02 -2.86313351e-02
1.18468118e+00 -7.24527100e-03 1.00316122e-01 3.40227276e-01
9.52822089e-01 -3.39756817e-01 -1.65844762e+00 -7.34234750e-02
2.05876470e-01 -3.90387386e-01 2.91714072e-01 -9.35064077e-01
-1.30226386e+00 7.50016928e-01 7.73283064e-01 2.49913335e-01
1.05871332e+00 4.04381573e-01 9.44412172e-01 -1.72939092e-01
5.16268611e-01 -1.07423067e+00 3.42099704e-02 1.62015066e-01
5.27463436e-01 -1.53347743e+00 -2.06294179e-01 -9.02289450e-01
-5.90472400e-01 8.61982167e-01 4.08729285e-01 -1.14417516e-01
6.46389842e-01 2.91377336e-01 2.79696405e-01 -4.63587552e-01
-2.53975540e-02 -5.39292455e-01 6.95684552e-01 5.25124907e-01
2.63684213e-01 1.95857108e-01 -1.73903674e-01 5.41072130e-01
-6.59112185e-02 1.10675909e-01 3.37546259e-01 7.51572371e-01
-3.68739933e-01 -1.12932003e+00 -2.59157747e-01 4.21990007e-01
-5.36113799e-01 -1.49912313e-01 -2.57126838e-01 6.09299660e-01
3.04249495e-01 8.23829591e-01 8.88877735e-02 -7.84801766e-02
4.06833500e-01 -1.71366781e-01 4.21437502e-01 -9.28009748e-01
-4.61276799e-01 2.55755603e-01 -1.30961299e-01 -7.16252625e-01
-5.50497115e-01 -5.64147294e-01 -1.84355712e+00 3.49178523e-01
-2.10241333e-01 -1.58958942e-01 8.51771533e-01 1.18197858e+00
1.83327749e-01 6.64142728e-01 3.07945013e-01 -1.08603358e+00
1.49714053e-01 -7.29357481e-01 -4.30434972e-01 6.90156281e-01
1.89063430e-01 -5.33509910e-01 -1.23869479e-01 2.45153993e-01]
|
[9.596844673156738, 0.2791537940502167]
|
a4569bbc-28f4-41f2-93cf-883215facbda
|
lg-hand-advancing-3d-hand-pose-estimation
|
2211.03151
| null |
https://arxiv.org/abs/2211.03151v1
|
https://arxiv.org/pdf/2211.03151v1.pdf
|
LG-Hand: Advancing 3D Hand Pose Estimation with Locally and Globally Kinematic Knowledge
|
3D hand pose estimation from RGB images suffers from the difficulty of obtaining the depth information. Therefore, a great deal of attention has been spent on estimating 3D hand pose from 2D hand joints. In this paper, we leverage the advantage of spatial-temporal Graph Convolutional Neural Networks and propose LG-Hand, a powerful method for 3D hand pose estimation. Our method incorporates both spatial and temporal dependencies into a single process. We argue that kinematic information plays an important role, contributing to the performance of 3D hand pose estimation. We thereby introduce two new objective functions, Angle and Direction loss, to take the hand structure into account. While Angle loss covers locally kinematic information, Direction loss handles globally kinematic one. Our LG-Hand achieves promising results on the First-Person Hand Action Benchmark (FPHAB) dataset. We also perform an ablation study to show the efficacy of the two proposed objective functions.
|
['Thanh-Hai Tran', 'Thi Ngoc Hien Doan', 'Trung Tran-Quang', 'Tu Le-Xuan']
|
2022-11-06
| null | null | null | null |
['3d-hand-pose-estimation', '3d-hand-pose-estimation']
|
['computer-vision', 'graphs']
|
[-3.65620792e-01 -1.74582094e-01 -4.37432766e-01 -7.00525865e-02
-5.03344178e-01 -4.78627831e-01 3.75150204e-01 -4.45461810e-01
-6.87585413e-01 6.18847251e-01 2.76499182e-01 -7.64099881e-02
-7.45978355e-02 -4.07290697e-01 -5.55903018e-01 -6.42535627e-01
-1.28458794e-02 4.61333543e-01 9.63556916e-02 -1.40897140e-01
6.12271801e-02 9.13168371e-01 -1.04218960e+00 -3.16811770e-01
3.97732884e-01 9.18610632e-01 1.57140717e-01 6.41816854e-01
2.44168833e-01 7.65793622e-01 -4.54506755e-01 -2.12676913e-01
5.06793797e-01 -3.70180786e-01 -1.03264308e+00 -3.68378572e-02
3.20667982e-01 -7.70482004e-01 -5.04867733e-01 6.53059006e-01
1.08657932e+00 1.54765919e-01 5.90973675e-01 -1.43457806e+00
-4.16491516e-02 4.35831010e-01 -7.69641399e-01 -2.27799620e-02
4.21318382e-01 4.88846421e-01 1.18359625e+00 -6.07630551e-01
8.75158310e-01 1.27656603e+00 7.94430077e-01 4.36652124e-01
-9.13989365e-01 -3.63370538e-01 4.39246893e-01 3.44090611e-01
-1.41986334e+00 -3.13925208e-04 1.10834825e+00 -4.72065359e-01
1.16439247e+00 -4.30068746e-02 8.65350723e-01 1.33225858e+00
-2.12264452e-02 1.22913754e+00 9.98791635e-01 -4.74585056e-01
-1.02343298e-01 -4.29189146e-01 1.18975118e-01 7.52052248e-01
1.36808589e-01 -4.49298285e-02 -6.22174919e-01 6.11268654e-02
1.18402648e+00 -6.93961531e-02 -4.27732706e-01 -5.88171124e-01
-1.22957432e+00 3.18078548e-01 7.44000971e-01 1.84467360e-01
-5.46908975e-01 6.70892060e-01 4.25490499e-01 3.87254506e-02
2.85643578e-01 5.30583560e-02 -5.24328649e-01 -6.19015038e-01
-5.28852880e-01 4.64850873e-01 5.60327888e-01 9.72135663e-01
2.39099279e-01 -2.70373166e-01 -4.32182491e-01 4.37382698e-01
5.11453211e-01 4.38687503e-01 -2.00961918e-01 -7.82240391e-01
7.52966762e-01 6.06815875e-01 2.35507444e-01 -8.66222620e-01
-8.43717277e-01 -4.23410952e-01 -6.12744331e-01 4.09047484e-01
9.72273171e-01 1.12828845e-03 -9.85753298e-01 1.87583840e+00
3.25005025e-01 -4.87168372e-01 -5.15822947e-01 1.40082443e+00
4.69194055e-01 -3.47089805e-02 9.29970592e-02 1.23063892e-01
1.23950934e+00 -1.01626003e+00 -7.28186488e-01 -2.98124664e-02
1.36234149e-01 -5.90959370e-01 1.20847082e+00 4.57411706e-01
-1.05577159e+00 -4.36644495e-01 -8.98946345e-01 -4.49523211e-01
-7.66919628e-02 4.06969517e-01 8.36830616e-01 5.95307291e-01
-7.97603905e-01 6.63277447e-01 -1.21811271e+00 -1.92687169e-01
4.55185533e-01 4.48733240e-01 -3.59245062e-01 1.80504680e-01
-9.32591617e-01 8.04701090e-01 5.90273142e-02 4.92628992e-01
-4.84293789e-01 -3.05334777e-01 -5.18757820e-01 -1.76749572e-01
4.81704295e-01 -7.82300830e-01 1.16883326e+00 -2.76265144e-01
-1.74667370e+00 6.84688270e-01 -1.80924594e-01 5.92985116e-02
1.29585254e+00 -5.80258906e-01 4.12678719e-01 2.46337324e-01
-1.60059795e-01 4.58108336e-01 8.64879131e-01 -1.16150713e+00
-3.06351036e-01 -9.02670205e-01 2.47656122e-01 1.11966267e-01
-2.34591514e-01 -2.37077817e-01 -7.95964003e-01 -7.85481393e-01
1.49169758e-01 -1.04522550e+00 9.08597410e-02 2.81775922e-01
-6.82359934e-01 -6.12001836e-01 4.47498351e-01 -9.18243051e-01
1.00767565e+00 -1.54964471e+00 7.51589358e-01 3.42568249e-01
2.04131633e-01 1.42225683e-01 1.06130326e-04 2.32749850e-01
3.44215512e-01 -5.54837435e-02 3.42715234e-02 -6.02795124e-01
1.63219094e-01 1.16129592e-01 -7.73206577e-02 4.98252183e-01
1.91233441e-01 1.09864438e+00 -7.96843290e-01 -4.48511571e-01
4.58589084e-02 1.04038250e+00 -5.52152216e-01 1.39758795e-01
-2.00488076e-01 7.86050677e-01 -7.51543820e-01 7.40778923e-01
4.57297355e-01 -2.11733550e-01 1.18327320e-01 -5.01975417e-01
6.09443104e-03 2.73189008e-01 -1.09665871e+00 2.25195312e+00
-4.05447453e-01 3.24855894e-01 -2.97406863e-04 -3.73349816e-01
3.97824466e-01 4.94073093e-01 6.39016867e-01 -4.50048208e-01
6.24911904e-01 1.42814085e-01 -1.09081246e-01 -4.07030344e-01
2.13887826e-01 8.13779011e-02 3.47504467e-01 6.76257730e-01
-2.58452445e-02 9.39025432e-02 -1.21552184e-01 -8.71771201e-02
8.76943886e-01 8.56180906e-01 1.76567078e-01 6.00496717e-02
1.99050173e-01 -1.72081575e-01 3.10857594e-01 5.07989585e-01
-4.74177003e-01 7.65001893e-01 7.89548695e-01 -4.11534458e-01
-1.00340068e+00 -9.41944480e-01 2.35371083e-01 8.42637599e-01
1.08574562e-01 -2.56971568e-01 -7.02667952e-01 -9.39408362e-01
2.90948391e-01 3.37104499e-02 -8.61026824e-01 1.63199127e-01
-9.85714018e-01 -4.86666799e-01 6.51223242e-01 1.26341593e+00
6.35148346e-01 -9.45203841e-01 -9.62870002e-01 1.61475595e-02
-3.07502627e-01 -1.05362308e+00 -7.30986714e-01 5.41604832e-02
-7.81418860e-01 -1.21474195e+00 -1.20383048e+00 -4.09784853e-01
3.43061745e-01 6.06806502e-02 7.97643483e-01 -1.90749571e-01
-4.28691238e-01 5.07183909e-01 -4.91769731e-01 -2.09682554e-01
1.76681787e-01 6.07922137e-01 1.07679836e-01 -2.64151424e-01
6.03051186e-02 -8.13904166e-01 -8.89414549e-01 5.02520800e-01
-1.73427016e-01 -9.25682634e-02 6.17511690e-01 7.53733814e-01
4.53340203e-01 -4.50335652e-01 3.80200386e-01 -1.74573243e-01
5.26936293e-01 3.25475007e-01 -4.81338412e-01 2.48161867e-01
-4.19729978e-01 3.25940400e-01 1.62199005e-01 -3.99925679e-01
-1.05972648e+00 3.97653431e-01 -4.37622875e-01 -4.65246230e-01
1.91045031e-01 2.87851006e-01 -4.20027554e-01 -1.21414937e-01
4.15343195e-01 2.58251783e-02 1.58012167e-01 -8.03821802e-01
4.13821161e-01 5.49058259e-01 3.12175781e-01 -8.06205809e-01
5.88222146e-01 6.65431619e-01 2.77384967e-01 -4.82132137e-01
-4.90111619e-01 -4.12715912e-01 -1.15674365e+00 -3.57223272e-01
8.92000377e-01 -6.08071268e-01 -1.34254777e+00 8.27487051e-01
-1.48798025e+00 -4.95337278e-01 -6.86990703e-03 5.63176334e-01
-7.52408385e-01 4.83759582e-01 -5.86654246e-01 -1.11351788e+00
-5.49531996e-01 -1.29177773e+00 1.27084601e+00 -3.05275917e-01
-2.29457036e-01 -6.01520061e-01 -5.29763624e-02 2.69262165e-01
5.47479540e-02 4.54419285e-01 6.65651441e-01 -7.89219663e-02
-6.53950632e-01 -3.96876067e-01 -3.94104123e-01 2.99147040e-01
3.06608319e-01 -3.77821505e-01 -1.00840092e+00 -5.69829643e-01
-3.34421188e-01 -4.27866518e-01 8.80607665e-01 3.37884307e-01
9.55421448e-01 2.04940164e-03 -1.24514483e-01 5.07211685e-01
1.02443266e+00 -1.15445890e-01 4.98064160e-01 3.32480013e-01
1.25918376e+00 5.40771306e-01 3.27856034e-01 5.87782741e-01
4.29689735e-01 1.03189993e+00 5.26152194e-01 1.98413491e-01
-5.01245797e-01 -2.48827592e-01 5.53229973e-02 4.85067189e-01
-1.20640922e+00 -1.96865484e-01 -9.97335732e-01 3.24307054e-01
-1.87352610e+00 -5.74920714e-01 9.52534378e-02 2.03096628e+00
9.31514204e-01 9.38007981e-02 7.00877070e-01 5.24439573e-01
3.93761069e-01 1.53056130e-01 -7.59037912e-01 3.99832428e-01
2.36997623e-02 2.71305323e-01 4.63300169e-01 4.76829112e-01
-9.42209959e-01 1.03919458e+00 5.54871368e+00 5.71706176e-01
-1.14669168e+00 -6.70058206e-02 -2.69118138e-02 -2.70254880e-01
1.03495233e-01 -2.89923668e-01 -7.65385211e-01 1.89770712e-04
-1.08792901e-01 5.39169550e-01 4.88809705e-01 7.40774870e-01
2.45302007e-01 7.02153966e-02 -1.23372781e+00 1.05054641e+00
-2.16867134e-01 -6.11724496e-01 1.15865348e-02 2.50371903e-01
3.15451831e-01 -1.55182138e-01 -1.63724534e-02 -2.06948921e-01
-1.24434993e-01 -9.63302672e-01 1.02406597e+00 4.39158261e-01
8.77560794e-01 -7.05014169e-01 5.05042613e-01 2.23712578e-01
-1.38735843e+00 1.99978754e-01 2.02225000e-01 -3.57386917e-01
3.12238008e-01 2.43980914e-01 -8.01131546e-01 5.59275925e-01
6.09069407e-01 6.08211637e-01 -3.43103558e-01 9.09179509e-01
-7.03495502e-01 2.24257067e-01 -4.87459958e-01 -3.99784036e-02
6.24609403e-02 2.66409695e-01 6.56483352e-01 8.19969177e-01
-1.48391172e-01 5.89342006e-02 4.33703586e-02 9.10641432e-01
5.88804930e-02 -1.33765042e-01 -1.00417249e-01 -1.68843865e-01
9.86399278e-02 9.30665970e-01 -7.58096576e-01 1.44699916e-01
-1.67289183e-01 1.25249588e+00 4.26381707e-01 4.48277324e-01
-6.57046080e-01 -5.06462038e-01 7.34552264e-01 -1.53257586e-02
2.84098685e-01 -7.55228758e-01 -5.04203439e-01 -1.16329527e+00
5.45869708e-01 -6.30046606e-01 1.83559582e-01 -4.83147264e-01
-1.08183110e+00 4.51385945e-01 -1.11999273e-01 -9.75687563e-01
-3.43158752e-01 -1.04905534e+00 -1.42907217e-01 1.13373756e+00
-1.43588424e+00 -1.49359906e+00 -7.12602139e-01 7.70872831e-01
3.44553024e-01 2.27432862e-01 5.56359172e-01 1.15318991e-01
-5.36597848e-01 9.09153879e-01 -5.86386263e-01 5.69751978e-01
5.79671085e-01 -1.32741320e+00 5.93319774e-01 5.01454592e-01
5.73843755e-02 7.57692695e-01 5.11374056e-01 -5.95127702e-01
-1.76786935e+00 -4.55736279e-01 7.22946167e-01 -7.68222690e-01
3.22179824e-01 -3.29875261e-01 -4.65623200e-01 6.75049543e-01
-1.99786797e-01 1.75320674e-02 8.02602023e-02 2.83977836e-01
-6.74576283e-01 1.57990515e-01 -1.11788726e+00 4.53936011e-01
1.67070138e+00 -5.63577235e-01 -4.17223096e-01 6.97723776e-02
4.48712617e-01 -5.26465118e-01 -8.68731141e-01 3.90443474e-01
1.18567073e+00 -7.56486833e-01 1.10369039e+00 -7.94206917e-01
3.06296587e-01 -1.56961635e-01 4.17117216e-02 -1.07983470e+00
-7.31045157e-02 -5.78544855e-01 -3.78162980e-01 7.44165957e-01
-2.34317593e-02 -4.54312682e-01 1.07923615e+00 4.66535509e-01
3.90703321e-01 -8.35686803e-01 -9.33694363e-01 -1.04159284e+00
2.14770645e-01 -5.46362221e-01 5.36758721e-01 4.93978560e-01
5.33816516e-02 2.32851133e-01 -5.08437634e-01 -4.41814512e-02
7.80306220e-01 1.94554061e-01 1.03478706e+00 -1.17526829e+00
-4.35780108e-01 -6.89086437e-01 -4.61895198e-01 -1.44257331e+00
8.25777352e-02 -5.74703217e-01 2.55525503e-02 -1.54115844e+00
2.18211830e-01 -3.21849078e-01 -2.83477545e-01 7.18827963e-01
-2.25186005e-01 2.91978449e-01 3.91606748e-01 3.26011449e-01
-2.31108531e-01 6.37651324e-01 1.68834078e+00 -7.25090131e-02
-3.54326844e-01 1.19166933e-01 -7.54438043e-02 6.13574266e-01
7.64041424e-01 -5.99987321e-02 -2.59859204e-01 -6.88526630e-01
2.31300190e-01 1.01143174e-01 8.01772892e-01 -7.28212953e-01
1.26398250e-01 -4.33679558e-02 3.50730389e-01 -7.22021580e-01
4.17592406e-01 -7.01662898e-01 -2.70652682e-01 5.85126460e-01
-2.25418285e-01 -2.41327174e-02 -1.49656245e-02 4.74733084e-01
1.03663802e-01 3.81073505e-01 4.84797060e-01 -1.40429318e-01
-6.62411869e-01 5.17756879e-01 1.64859444e-01 -7.10200593e-02
5.64137578e-01 -2.88855046e-01 1.17868468e-01 -3.51036698e-01
-9.30660188e-01 1.09019175e-01 2.06442446e-01 5.33941567e-01
4.40384150e-01 -1.27349532e+00 -5.35151362e-01 -1.21741267e-02
3.32026966e-02 1.98901340e-01 1.31491125e-02 1.19363499e+00
-5.98750353e-01 6.90967202e-01 -3.33711296e-01 -6.15391433e-01
-1.25610590e+00 3.27303976e-01 3.21431726e-01 -2.10195810e-01
-7.81745553e-01 1.09406841e+00 -2.57042825e-01 -2.92362332e-01
8.06237757e-01 -5.43572366e-01 1.42881334e-01 2.27213930e-02
4.55092698e-01 7.36350417e-01 6.55731931e-02 -6.03066742e-01
-6.00949287e-01 1.02173924e+00 1.36537313e-01 -2.44792774e-01
1.30761027e+00 8.28529671e-02 5.23620173e-02 2.23548129e-01
1.30019391e+00 -1.62249357e-01 -1.64209759e+00 -1.86727822e-01
-1.98020518e-01 -5.72130978e-01 6.25319555e-02 -1.09177125e+00
-1.26715100e+00 1.23160541e+00 5.95742762e-01 -3.01100463e-01
1.01722395e+00 1.34476960e-01 8.86352897e-01 5.88421762e-01
7.95303166e-01 -9.87964571e-01 3.02183628e-01 4.56844538e-01
1.18586314e+00 -1.22742331e+00 4.51149978e-02 -6.29426420e-01
-2.47983307e-01 1.15578735e+00 4.97680068e-01 5.56655601e-02
4.44682956e-01 1.22210950e-01 3.13041136e-02 -8.35964680e-02
3.94480824e-02 -4.62462425e-01 5.76336622e-01 4.82422084e-01
5.03395319e-01 5.43109179e-02 -4.22419399e-01 6.50136113e-01
-1.21089749e-01 4.00192171e-01 -2.42730051e-01 1.28338003e+00
-4.74826992e-02 -1.29245639e+00 -2.82821417e-01 -1.23611704e-01
-3.58465016e-01 2.54568219e-01 -6.68478072e-01 1.10540509e+00
5.16059715e-03 4.58549947e-01 -3.49601656e-01 -5.55610120e-01
5.94613254e-01 1.41070932e-01 1.36359942e+00 -1.00362949e-01
-4.43780869e-01 1.03808641e-01 -1.83268175e-01 -8.09744418e-01
-4.92635787e-01 -4.85207319e-01 -1.25107622e+00 -2.91812420e-01
-2.63987452e-01 -4.34024930e-01 7.01012552e-01 9.13734794e-01
2.32901692e-01 5.47868788e-01 1.81335196e-01 -1.42527413e+00
-8.87625217e-01 -1.05956018e+00 -6.82279170e-01 3.39492679e-01
4.76233631e-01 -1.22734678e+00 -9.92271081e-02 -3.23143750e-01]
|
[6.611468315124512, -0.7106606364250183]
|
71fd30ef-3663-4882-86a2-ea2b01896198
|
analysis-of-overfitting-in-the-regularized
|
1904.06632
| null |
https://arxiv.org/abs/1904.06632v2
|
https://arxiv.org/pdf/1904.06632v2.pdf
|
Analysis of overfitting in the regularized Cox model
|
The Cox proportional hazards model is ubiquitous in the analysis of time-to-event data. However, when the data dimension p is comparable to the sample size $N$, maximum likelihood estimates for its regression parameters are known to be biased or break down entirely due to overfitting. This prompted the introduction of the so-called regularized Cox model. In this paper we use the replica method from statistical physics to investigate the relationship between the true and inferred regression parameters in regularized multivariate Cox regression with L2 regularization, in the regime where both p and N are large but with p/N ~ O(1). We thereby generalize a recent study from maximum likelihood to maximum a posteriori inference. We also establish a relationship between the optimal regularization parameter and p/N, allowing for straightforward overfitting corrections in time-to-event analysis.
|
['M Sheikh', 'A. C. C. Coolen']
|
2019-04-14
| null | null | null | null |
['l2-regularization']
|
['methodology']
|
[ 1.47136658e-01 2.22390257e-02 -3.64460081e-01 -3.49983811e-01
-1.06894672e+00 3.29252961e-03 4.75419134e-01 5.22205055e-01
-5.93284369e-01 1.02822840e+00 -2.52934173e-02 -3.38129252e-01
-4.05622065e-01 -8.34839582e-01 -6.30951405e-01 -8.96360338e-01
-5.98643005e-01 4.73890156e-01 2.17214704e-01 8.89424235e-02
2.39443213e-01 4.65886384e-01 -1.13822520e+00 -5.98227799e-01
8.37305963e-01 5.70141613e-01 -5.42489409e-01 7.18048930e-01
2.70524859e-01 3.16048294e-01 -8.40924755e-02 -9.19566080e-02
7.17292130e-02 -5.18712938e-01 -5.60535908e-01 -4.10196871e-01
-3.57382111e-02 -6.86871856e-02 -4.82375950e-01 9.48893368e-01
4.53053206e-01 3.42463166e-01 1.12593210e+00 -1.14280760e+00
-7.53149837e-02 4.33742672e-01 -9.04158294e-01 2.69628465e-01
-2.46108454e-02 -2.77200639e-01 4.64518815e-01 -6.76497877e-01
5.13383031e-01 1.28297925e+00 1.02295530e+00 3.32242101e-01
-1.59938073e+00 -8.12540114e-01 -4.10891324e-01 -2.54768908e-01
-1.85393763e+00 -4.33085889e-01 3.98930639e-01 -6.85202539e-01
6.00837469e-01 1.63897991e-01 3.26530099e-01 8.97922635e-01
6.76562011e-01 -1.96636230e-01 1.16881180e+00 -7.88733006e-01
4.81089324e-01 5.37453927e-02 4.10411507e-01 6.48264647e-01
6.78520918e-01 4.87865031e-01 -2.62856334e-01 -7.15390623e-01
8.18588078e-01 -1.41810343e-01 5.35917059e-02 -1.69085026e-01
-9.67041731e-01 1.12118220e+00 2.37489697e-02 3.19435626e-01
-2.30302364e-01 5.66606820e-01 4.62919533e-01 9.67094079e-02
5.56709409e-01 -5.08578084e-02 -2.22499549e-01 2.09888667e-01
-7.82991588e-01 4.61188585e-01 5.87022722e-01 6.78351283e-01
3.54056299e-01 -2.96659619e-01 -1.02094077e-01 5.08918941e-01
3.17557931e-01 5.14384449e-01 -1.16216078e-01 -9.33217824e-01
2.74613351e-01 7.83419162e-02 3.19475055e-01 -4.70858991e-01
-7.26491392e-01 -2.95624405e-01 -1.04144597e+00 1.93156764e-01
8.80627513e-01 -1.76449701e-01 -4.74558383e-01 2.00703025e+00
3.60609293e-01 -1.15064658e-01 -2.60951906e-01 4.06791955e-01
2.15420518e-02 2.84579545e-01 5.64461589e-01 -8.93160105e-01
1.38493550e+00 -7.12215677e-02 -7.89287984e-01 6.06981181e-02
8.38222563e-01 -5.10802925e-01 8.36104333e-01 2.61468768e-01
-1.02111506e+00 -1.61527302e-02 -8.13742399e-01 2.12183036e-02
-5.37101366e-02 -3.95510137e-01 3.00819635e-01 6.68101013e-01
-7.26859510e-01 7.97521949e-01 -9.19768214e-01 -5.07502019e-01
2.29016945e-01 3.37145597e-01 -1.59868196e-01 -8.64581317e-02
-1.13814235e+00 8.67022157e-01 2.91156203e-01 1.17358165e-02
-5.61084807e-01 -7.34799623e-01 -4.03826177e-01 -7.33953044e-02
1.22262374e-01 -7.78167069e-01 9.76239800e-01 -2.31411472e-01
-8.89114857e-01 6.72199547e-01 -3.56631637e-01 -3.34777981e-01
6.46082342e-01 1.60866454e-01 -1.48946732e-01 -6.95677623e-02
9.70605612e-02 4.43261713e-02 5.34557879e-01 -8.47675264e-01
-1.84742853e-01 -7.30745912e-01 -2.83426851e-01 -2.62141436e-01
2.26614714e-01 2.62068897e-01 1.98221523e-02 -3.60473484e-01
4.19804782e-01 -9.86073732e-01 -6.25746131e-01 2.29051150e-02
-2.91197032e-01 -4.33343910e-02 1.47858813e-01 -6.03578269e-01
1.12645376e+00 -2.22880006e+00 -2.15105221e-01 4.60111648e-01
2.62538493e-01 -6.22187495e-01 3.69088292e-01 4.88733351e-01
-3.81927818e-01 1.50489882e-01 -5.05927026e-01 -1.50666967e-01
-2.82663256e-01 4.19447683e-02 6.20858371e-02 1.12309778e+00
-1.49478287e-01 3.85742188e-01 -7.50207186e-01 -7.81567752e-01
-1.19915433e-01 6.42508090e-01 -5.41469932e-01 -1.69921219e-01
2.87823290e-01 5.98198295e-01 -6.68678761e-01 2.60152400e-01
8.14910352e-01 -2.87544250e-01 2.68070102e-01 1.23659782e-01
-5.44629633e-01 -1.19304406e-02 -8.92296314e-01 1.12233174e+00
-1.86532766e-01 4.94310856e-01 4.75484356e-02 -1.15211332e+00
6.17095232e-01 5.61957121e-01 5.66862106e-01 -2.93055624e-01
3.12159747e-01 2.99021810e-01 -1.72973499e-01 -3.22726101e-01
8.15834180e-02 -1.15665424e+00 -2.79479027e-01 3.68758999e-02
-3.00509602e-01 1.71906993e-01 1.13098480e-01 -8.33139662e-03
1.10480797e+00 -3.32138687e-01 7.60302484e-01 -7.72903085e-01
3.32214683e-01 -1.98740922e-02 5.49277604e-01 9.31605816e-01
-1.72689676e-01 4.45847690e-01 1.06375790e+00 -6.67406842e-02
-1.10436618e+00 -1.26768816e+00 -1.20790029e+00 5.47333062e-01
-2.61234045e-01 -1.75631225e-01 -5.03469169e-01 -3.58313739e-01
-1.70902500e-03 9.32178259e-01 -7.97432661e-01 -2.50924647e-01
-5.53309083e-01 -1.82656646e+00 6.17124021e-01 2.95362949e-01
1.74651965e-02 -4.68345731e-01 -1.11883290e-01 2.19518691e-01
9.42468494e-02 -6.28549993e-01 -1.96057960e-01 4.01234388e-01
-1.34801793e+00 -1.24115908e+00 -6.15859509e-01 -2.30099261e-01
7.96111941e-01 -3.52592885e-01 8.57627392e-01 -1.19803824e-01
-2.35380217e-01 1.70079842e-01 -3.38781402e-02 -3.56248140e-01
-6.93990707e-01 -2.56178498e-01 2.03499600e-01 -4.53237385e-01
1.31756037e-01 -5.63222110e-01 -6.06895566e-01 2.62992263e-01
-7.59771526e-01 -4.67612535e-01 2.64119446e-01 7.35485315e-01
5.89144468e-01 1.34861559e-01 7.16845274e-01 -9.56490040e-01
2.47782558e-01 -7.15750337e-01 -9.37326610e-01 7.15020224e-02
-9.06277478e-01 2.56352305e-01 3.61381829e-01 -2.94439822e-01
-9.36244011e-01 -3.38119030e-01 -9.66185927e-02 1.24926455e-01
3.93619463e-02 5.98254502e-01 1.53484613e-01 -9.50006172e-02
7.26808906e-01 -3.07961464e-01 -9.19756144e-02 -4.69563425e-01
-7.58795887e-02 2.50123739e-01 1.50760114e-01 -4.94121224e-01
3.89007151e-01 5.66479921e-01 8.59028876e-01 -8.38703454e-01
-6.49502456e-01 -1.82217479e-01 -5.66334724e-01 1.32933602e-01
8.18513930e-01 -7.38925695e-01 -8.49994063e-01 1.39611706e-01
-8.48689377e-01 -2.91073740e-01 -5.71533918e-01 1.13439023e+00
-8.00657392e-01 4.13804144e-01 -6.37569368e-01 -1.09585106e+00
7.09739551e-02 -8.99199545e-01 6.02392137e-01 -1.13257349e-01
-1.91553041e-01 -1.22592664e+00 3.75963151e-01 -1.86625406e-01
2.10518539e-01 4.54822868e-01 1.27337432e+00 -5.43172598e-01
-1.76221803e-01 -6.44502878e-01 -3.22157323e-01 -2.07037538e-01
-2.09881008e-01 -1.15942648e-02 -6.53606892e-01 -3.24051917e-01
4.18655246e-01 2.45138481e-01 6.97356403e-01 1.11491561e+00
1.07860041e+00 -1.68442115e-01 -7.33381331e-01 4.43593293e-01
1.59208584e+00 1.85591117e-01 6.10135376e-01 -6.23725615e-02
2.51402676e-01 6.16142571e-01 4.50357765e-01 6.32041574e-01
1.73202559e-01 6.97333932e-01 -1.03331462e-01 2.01669261e-01
3.09369117e-01 -1.51796713e-01 3.59207951e-02 6.70947134e-01
-2.19726637e-01 -7.68418908e-02 -8.93729091e-01 2.32506782e-01
-1.59711444e+00 -6.77222908e-01 -7.25311756e-01 3.07566857e+00
8.69674563e-01 4.31671858e-01 1.66972235e-01 -8.39487556e-03
7.27137923e-01 -1.98305011e-01 -2.78019369e-01 -1.20997190e-01
1.36161456e-02 1.78567618e-01 1.10395145e+00 8.23223233e-01
-6.53094828e-01 1.10119760e-01 7.85647964e+00 9.74137187e-01
-4.98026013e-01 5.11813641e-01 6.87973261e-01 -9.07024145e-02
-1.59691647e-01 5.16543031e-01 -8.27108502e-01 5.08976460e-01
1.54325867e+00 -4.51851785e-01 6.73038289e-02 2.98189759e-01
7.25489318e-01 -5.83163023e-01 -8.92251611e-01 4.68840390e-01
-3.54966789e-01 -7.92636037e-01 -4.40109581e-01 4.62386101e-01
3.65153551e-01 -1.94733329e-02 -3.99549335e-01 2.07364053e-01
7.97439553e-03 -9.53104198e-01 2.52379388e-01 7.26159990e-01
1.06725502e+00 -7.78197169e-01 6.43876672e-01 5.03418863e-01
-1.00347292e+00 1.32740125e-01 -4.26987410e-01 -8.32761526e-02
3.65306437e-01 1.01062405e+00 -5.89933157e-01 1.52954325e-01
2.65654624e-01 2.63765395e-01 -2.13249072e-01 1.02287269e+00
2.45779887e-01 6.01949513e-01 -7.13416934e-01 2.87223876e-01
-1.45665959e-01 -4.58161831e-01 5.28314650e-01 8.37292552e-01
5.83365381e-01 3.39945853e-01 -3.73683929e-01 7.27171302e-01
3.32820654e-01 1.47665828e-01 -4.22508866e-01 9.26168412e-02
3.12541097e-01 7.27449775e-01 -9.39213037e-01 -1.24181278e-01
-5.11054039e-01 2.42187679e-01 8.48631039e-02 3.51462781e-01
-6.60349369e-01 -5.12327366e-02 2.07852513e-01 6.72646165e-01
-6.86462745e-02 -3.48841786e-01 -2.86103070e-01 -9.41034853e-01
-2.49123335e-01 -7.21470313e-03 6.53108418e-01 -2.36802667e-01
-1.35813236e+00 8.56578872e-02 7.34565198e-01 -9.05062854e-01
-1.45510942e-01 -5.68657935e-01 -5.34100056e-01 1.01217043e+00
-1.03941500e+00 -6.32347643e-01 3.98780137e-01 2.37611368e-01
-2.43826527e-02 3.27924132e-01 7.65700579e-01 2.76357144e-01
-7.73776770e-01 4.82301414e-01 6.17552102e-01 -3.02620023e-01
6.55084074e-01 -9.57787991e-01 1.35355741e-01 5.35353124e-01
-7.86403000e-01 7.01411068e-01 1.28219426e+00 -9.26586211e-01
-1.03558779e+00 -6.28170371e-01 1.21864128e+00 -2.80386984e-01
8.84102881e-01 -2.64333040e-01 -1.00201595e+00 6.09134793e-01
-3.46129417e-01 2.26676717e-01 7.29121745e-01 3.72913867e-01
-2.54924633e-02 -8.50784257e-02 -1.39782190e+00 5.10296941e-01
7.86303818e-01 -3.55389357e-01 -2.11654618e-01 4.66534555e-01
4.27041531e-01 -4.46827598e-02 -1.16030908e+00 4.12131995e-01
6.48135543e-01 -8.15239072e-01 1.00051737e+00 -6.10878050e-01
-6.72338009e-02 8.61934572e-02 -1.12248801e-01 -7.62776256e-01
-1.32421613e-01 -5.12454331e-01 4.15562272e-01 8.88524413e-01
4.50860053e-01 -8.43997359e-01 5.05828619e-01 8.08140635e-01
2.65040636e-01 -4.32458222e-01 -1.69988906e+00 -7.44888186e-01
5.67582548e-01 -5.30821979e-01 -1.86764196e-01 8.98470700e-01
3.15228701e-01 3.96687202e-02 -2.94281691e-01 1.07761875e-01
1.26612914e+00 -4.56095606e-01 1.16961785e-01 -1.46858835e+00
-1.99192092e-01 -6.02232553e-02 -1.36041760e-01 -5.74355006e-01
5.74311651e-02 -7.07180679e-01 8.98467153e-02 -8.41731191e-01
5.93018234e-01 -7.89674461e-01 -1.56054914e-01 -1.24448046e-01
-1.11726634e-01 -9.91754979e-02 -4.00357246e-01 1.25105694e-01
-1.26122078e-02 3.71755272e-01 8.46861005e-01 4.06670302e-01
-2.54263252e-01 4.00003225e-01 -3.39457929e-01 9.56271827e-01
7.62238801e-01 -1.17117846e+00 -9.51795802e-02 3.78681242e-01
5.13969064e-01 9.44517612e-01 5.87003529e-01 -7.23996401e-01
-1.04237441e-02 -3.45165759e-01 2.53208607e-01 -4.65119481e-01
1.82037741e-01 -7.31788933e-01 4.82392430e-01 4.75673527e-01
-5.93664348e-01 -1.40460089e-01 1.12157866e-01 8.15305352e-01
1.71439514e-01 -7.90434539e-01 8.50843608e-01 -6.99874833e-02
3.11202586e-01 1.69181094e-01 -6.18648410e-01 1.05529957e-01
6.73012912e-01 2.37326726e-01 -2.40444407e-01 -2.33033016e-01
-9.57514167e-01 4.11994057e-03 3.48136216e-01 -3.20366859e-01
2.52301216e-01 -1.37559938e+00 -7.67100811e-01 7.05663115e-02
-2.59067416e-02 -2.99643308e-01 6.66609526e-01 1.45146012e+00
-5.43365002e-01 3.86399806e-01 2.79523551e-01 -4.13508385e-01
-9.63977218e-01 6.68630064e-01 3.96820962e-01 -3.19675505e-01
-5.83241165e-01 3.58171701e-01 4.69185114e-01 3.67114134e-02
-1.34084061e-01 -1.72516972e-01 2.26728618e-01 -4.16728258e-02
4.85989958e-01 9.02365386e-01 -2.38150433e-01 -4.10063893e-01
-3.63876879e-01 5.31076014e-01 2.20867231e-01 -3.68748933e-01
1.03300774e+00 -3.74558568e-01 -3.65718395e-01 8.81967008e-01
1.28385210e+00 2.01373309e-01 -1.23103392e+00 -5.07997833e-02
1.77204698e-01 -2.65175134e-01 1.48321152e-01 -8.61812010e-02
-4.67901528e-01 6.03040338e-01 7.31124580e-01 2.90373355e-01
7.40538895e-01 1.82869405e-01 1.66608170e-01 -4.58326051e-03
2.54139185e-01 -8.57008636e-01 -6.04186356e-01 1.49187371e-01
7.91084588e-01 -1.19542396e+00 3.75699401e-01 -5.77299654e-01
-8.92607421e-02 1.00171947e+00 5.77207729e-02 -3.39470983e-01
1.19493484e+00 4.54461813e-01 -4.06056136e-01 -1.93212584e-01
-6.28683269e-01 2.83006430e-01 1.56860322e-01 3.91667485e-01
6.82122588e-01 1.69110417e-01 -1.00021076e+00 5.40822327e-01
1.14469901e-01 2.22768560e-01 6.05631590e-01 7.88188457e-01
-3.86641741e-01 -1.07497990e+00 -5.42661488e-01 5.17419755e-01
-6.02468133e-01 -2.41813809e-02 2.84935772e-01 1.15320241e+00
-1.30517080e-01 7.98354745e-01 2.70134181e-01 3.46528143e-01
2.99903721e-01 3.64651769e-01 5.33200622e-01 -8.21084380e-02
-1.30149141e-01 4.28431392e-01 6.93813860e-02 -2.37041920e-01
-5.01321316e-01 -1.12537992e+00 -1.27408719e+00 -5.86240530e-01
-4.89230067e-01 3.42615217e-01 4.68170345e-01 1.08873844e+00
-2.77776271e-01 3.70258182e-01 5.37006557e-01 -4.92061019e-01
-7.56819069e-01 -8.44591856e-01 -1.20251274e+00 -2.43504241e-01
5.20566344e-01 -6.72401249e-01 -8.57251227e-01 -2.28454456e-01]
|
[7.722892761230469, 4.821357727050781]
|
45bc3e66-5f6e-4e14-94d0-d7c0a4692979
|
instance-segmentation-of-fibers-from-low
|
1901.01034
| null |
http://arxiv.org/abs/1901.01034v1
|
http://arxiv.org/pdf/1901.01034v1.pdf
|
Instance Segmentation of Fibers from Low Resolution CT Scans via 3D Deep Embedding Learning
|
We propose a novel approach for automatic extraction (instance segmentation)
of fibers from low resolution 3D X-ray computed tomography scans of short glass
fiber reinforced polymers. We have designed a 3D instance segmentation
architecture built upon a deep fully convolutional network for semantic
segmentation with an extra output for embedding learning. We show that the
embedding learning is capable of learning a mapping of voxels to an embedded
space in which a standard clustering algorithm can be used to distinguish
between different instances of an object in a volume. In addition, we discuss a
merging post-processing method which makes it possible to process volumes of
any size. The proposed 3D instance segmentation network together with our
merging algorithm is the first known to authors knowledge procedure that
produces results good enough, that they can be used for further analysis of low
resolution fiber composites CT scans.
|
['Jürgen Hesser', 'Thorben Kröger', 'Tomasz Konopczyński', 'Lei Zheng']
|
2019-01-04
| null | null | null | null |
['3d-instance-segmentation-1']
|
['computer-vision']
|
[ 3.83389413e-01 5.11587083e-01 1.98693901e-01 -4.15812969e-01
-4.89166081e-01 -3.12751710e-01 3.10472429e-01 4.80735362e-01
-4.04295772e-01 2.62337059e-01 -1.96531221e-01 -2.21361995e-01
-5.10336876e-01 -1.27988183e+00 -9.86507714e-01 -7.37568736e-01
-5.63989043e-01 1.17705131e+00 4.72362161e-01 -3.51692252e-02
2.14093462e-01 1.11420202e+00 -1.54850066e+00 3.77482027e-01
4.04174656e-01 9.57181633e-01 6.49722517e-01 7.10338414e-01
-2.73579985e-01 1.49334908e-01 -2.58821845e-01 1.41974138e-02
3.71557087e-01 2.06744194e-01 -1.20579398e+00 5.02938509e-01
3.39090943e-01 -3.69845033e-01 2.51653399e-02 9.05931234e-01
9.20476913e-02 -2.69376915e-02 9.90273535e-01 -7.07768202e-01
-3.35951269e-01 9.48978305e-01 -8.85668024e-02 -1.63929284e-01
1.54527366e-01 -8.68216306e-02 8.22352350e-01 -7.02800751e-01
9.08256888e-01 1.05693340e+00 4.08918381e-01 4.17225480e-01
-1.32131612e+00 4.09937501e-02 -1.05323374e-01 -1.74119830e-01
-8.39575768e-01 3.52422774e-01 9.08023953e-01 -8.24009597e-01
8.68635297e-01 3.22988242e-01 1.03300667e+00 6.96106493e-01
2.23722965e-01 3.79118234e-01 1.38712776e+00 -3.39097977e-01
4.97293741e-01 -6.15962036e-03 6.22430921e-01 9.10234392e-01
2.54629582e-01 2.17241365e-02 2.89668888e-01 3.22298884e-01
1.25594687e+00 9.66045111e-02 9.33740940e-03 -2.72786856e-01
-1.13386238e+00 7.64470339e-01 9.18380201e-01 5.15069544e-01
-4.46170151e-01 4.24190342e-01 2.71107584e-01 1.86490089e-01
6.56439483e-01 5.12697458e-01 -4.70291972e-01 5.16628206e-01
-1.06068563e+00 1.84019461e-01 9.43059206e-01 5.24895012e-01
9.31177437e-01 -2.34022930e-01 1.89744309e-01 3.87980938e-01
4.95898038e-01 1.55958951e-01 2.20130548e-01 -1.01285326e+00
-1.13183193e-01 8.92372847e-01 -2.76405483e-01 -6.77168012e-01
-5.28565407e-01 -1.89129010e-01 -3.30496103e-01 8.43563914e-01
2.83318967e-01 2.40873009e-01 -1.21981335e+00 9.28401113e-01
4.56111968e-01 -3.75130698e-02 -1.15453184e-01 1.00479686e+00
8.49635124e-01 4.26071435e-01 -9.71981287e-02 2.81111121e-01
1.57446325e+00 -5.98580360e-01 -3.38964820e-01 2.85982490e-01
3.54008913e-01 -5.57512701e-01 7.84550190e-01 4.83678758e-01
-8.34867895e-01 -5.24285913e-01 -1.30392718e+00 -9.37420204e-02
-7.23286331e-01 8.53371471e-02 8.21136117e-01 5.75690687e-01
-9.56160367e-01 1.34835422e+00 -1.16519034e+00 -5.78892641e-02
7.45641172e-01 6.95316195e-01 -5.29597342e-01 2.05569014e-01
-6.03174329e-01 6.25066519e-01 8.44263971e-01 9.94990170e-02
-9.25685167e-01 -4.51836050e-01 -6.51233792e-01 -6.46920949e-02
3.42918217e-01 -6.88406527e-01 7.19574153e-01 -8.00173104e-01
-1.37192857e+00 1.22114778e+00 2.25098535e-01 -4.73849982e-01
4.16398406e-01 -1.44301310e-01 -1.44471675e-01 6.74913108e-01
-6.58277422e-02 7.66568720e-01 1.09815073e+00 -1.63779724e+00
-1.38655320e-01 -6.10619009e-01 3.69526476e-01 -3.52653772e-01
3.57585959e-02 -4.01127756e-01 7.02435300e-02 -6.25497222e-01
5.87379336e-01 -7.92239964e-01 -6.34528399e-01 -2.43200853e-01
-8.82308006e-01 -2.41627604e-01 9.36745822e-01 -8.34184408e-01
5.60407221e-01 -1.85677505e+00 3.48795652e-01 6.22830987e-01
5.63968062e-01 -1.96618110e-01 9.55309570e-02 2.18494996e-01
-4.19618845e-01 3.38830024e-01 -6.83698058e-01 -3.78985107e-02
-8.07793736e-02 2.69014865e-01 3.54024693e-02 3.71814787e-01
3.16161007e-01 6.20664895e-01 -7.10321188e-01 -4.46679384e-01
5.66070259e-01 4.72860992e-01 -4.93840009e-01 1.01732358e-01
-6.66908622e-01 4.56505328e-01 -4.79761809e-01 5.09625912e-01
7.68975019e-01 -1.92083463e-01 -8.69911984e-02 -5.13074458e-01
-1.35982722e-01 -1.34300776e-02 -1.23350501e+00 1.76684868e+00
-3.38275164e-01 3.06709349e-01 9.51837450e-02 -9.77309048e-01
8.61740470e-01 3.22493911e-01 9.69920278e-01 -1.15937442e-01
4.41045612e-01 2.25071698e-01 -1.13628954e-01 -5.59016883e-01
4.07747597e-01 -4.09657598e-01 1.80710331e-01 6.02211118e-01
2.86902696e-01 -5.55446446e-01 2.91760474e-01 7.14682862e-02
9.05760884e-01 8.28759596e-02 -5.00494778e-01 -4.81440991e-01
5.71551800e-01 6.83461726e-02 -9.77978483e-02 2.30560273e-01
3.67024958e-01 6.27532244e-01 4.45936888e-01 -6.98885679e-01
-1.45825481e+00 -1.36125505e+00 -4.49976027e-01 4.45903510e-01
1.17667280e-01 -1.85421646e-01 -1.03618920e+00 -6.08888030e-01
9.92090851e-02 1.88352033e-01 -6.33085251e-01 4.22089726e-01
-7.20428228e-01 -5.78211784e-01 -2.02131733e-01 4.51717615e-01
2.56480277e-01 -1.03127384e+00 -5.70849180e-01 4.85549033e-01
4.29052413e-01 -1.13829565e+00 1.37177095e-01 4.94254917e-01
-1.55452228e+00 -1.42288697e+00 -5.36850512e-01 -1.16732311e+00
8.74618232e-01 -3.02087188e-01 1.02802956e+00 2.38024414e-01
-4.65487212e-01 5.93492568e-01 -4.46413100e-01 -1.98028401e-01
-6.95178449e-01 2.78003097e-01 -3.13511938e-01 -1.97276413e-01
1.50541589e-01 -5.80399394e-01 -6.89488649e-01 -1.39960364e-01
-1.20864975e+00 -1.11788787e-01 5.58272481e-01 2.93478161e-01
9.64628816e-01 3.59019965e-01 1.33875385e-01 -1.09530175e+00
4.28521603e-01 -1.60287455e-01 -5.61142445e-01 -5.62804341e-02
-4.83642101e-01 2.33650342e-01 4.71281290e-01 4.14203666e-02
-4.96579468e-01 2.05196455e-01 -6.33118689e-01 -2.37856120e-01
-6.77433848e-01 2.57322907e-01 1.59116313e-01 -7.45850205e-02
4.15178299e-01 -2.01314747e-01 2.26873860e-01 -7.40340650e-01
5.13184249e-01 5.67116618e-01 2.36092731e-01 -5.22184253e-01
6.73178136e-01 7.52105474e-01 2.40443021e-01 -1.02658570e+00
-2.73776025e-01 -3.71535420e-01 -1.11496758e+00 -4.43385154e-01
1.46923923e+00 -5.62868714e-01 -6.26788735e-01 2.25404114e-01
-1.11396015e+00 -3.61965090e-01 -6.95618451e-01 5.61389685e-01
-9.08679783e-01 3.83340418e-01 -9.86107409e-01 -5.05007744e-01
-2.45960891e-01 -1.48335540e+00 1.31185651e+00 -5.39717497e-03
2.76469868e-02 -1.12760782e+00 1.22956848e-02 5.05423307e-01
-1.08764648e-01 5.56296766e-01 1.28236425e+00 -4.34808671e-01
-8.55797172e-01 -5.25652617e-02 9.39683095e-02 6.72750711e-01
-4.89854924e-02 2.28530928e-01 -9.66002405e-01 -1.69999748e-01
1.35873452e-01 1.85182989e-01 1.19320953e+00 6.71559215e-01
1.46193671e+00 -6.72910213e-02 -4.34643269e-01 3.17474842e-01
1.94006968e+00 -4.15906161e-02 8.89609993e-01 4.01062578e-01
7.67823100e-01 6.28687024e-01 1.48227111e-01 -8.13396648e-03
-1.57023609e-01 4.78311032e-01 7.16437757e-01 -3.48176271e-01
-1.81690246e-01 3.40185821e-01 7.42470473e-02 7.94611096e-01
-6.14331126e-01 1.71714380e-01 -7.04920113e-01 4.65569973e-01
-1.26338983e+00 -8.48445415e-01 -5.35627067e-01 1.94427252e+00
4.99057263e-01 3.86061966e-01 1.94754645e-01 5.89125872e-01
6.77105784e-01 -4.47171479e-02 -2.06054062e-01 -6.98480010e-01
2.37016171e-01 1.03683805e+00 6.09442174e-01 6.56899214e-01
-1.24258149e+00 7.63231695e-01 6.29383087e+00 4.66690868e-01
-9.45670784e-01 1.18752584e-01 3.22967470e-01 3.39364022e-01
-3.98557991e-01 -3.32364850e-02 -4.04103160e-01 3.30681115e-01
8.56257856e-01 5.55691719e-01 2.11692527e-01 7.33708739e-01
7.24903196e-02 -1.79499928e-02 -1.09421504e+00 5.45750201e-01
-3.68804336e-01 -1.57276726e+00 2.83987552e-01 1.77719831e-01
4.83635932e-01 1.59659795e-02 -1.87053293e-01 -2.39883557e-01
3.07309836e-01 -1.21577728e+00 5.89169741e-01 6.39849663e-01
4.93501604e-01 -7.25030601e-01 6.88296080e-01 -6.19462654e-02
-1.16860938e+00 -9.98188704e-02 -2.38865733e-01 2.37154186e-01
3.49835128e-01 1.00155413e+00 -9.87004340e-01 8.03635359e-01
5.55909514e-01 5.14764786e-01 -3.58130604e-01 8.27426732e-01
-1.22367233e-01 3.45856667e-01 -3.79520983e-01 2.20020249e-01
3.94452065e-01 -4.12332714e-01 6.21460974e-01 9.46789980e-01
3.41113836e-01 -1.17561802e-01 3.62675548e-01 1.15827584e+00
-6.14218740e-03 5.60553484e-02 -5.72043180e-01 -1.73242018e-01
-3.24525014e-02 1.26023161e+00 -1.59260237e+00 -4.28322405e-01
-9.35457926e-03 8.03064048e-01 4.25080806e-02 -1.28484210e-02
-5.99525571e-01 -2.63722062e-01 3.76919925e-01 5.41261315e-01
7.36435533e-01 -6.00294828e-01 -4.38446969e-01 -6.81556165e-01
-1.09818086e-01 -1.57357365e-01 2.70565436e-03 -7.16279924e-01
-1.30350137e+00 6.21398628e-01 -5.27569838e-02 -9.67968643e-01
8.60124081e-02 -1.23933434e+00 -4.45756495e-01 4.89696681e-01
-1.26051652e+00 -1.13706470e+00 -4.55661789e-02 2.96183109e-01
3.86654794e-01 -3.08221579e-02 8.28381062e-01 2.43336000e-02
-2.58810103e-01 -3.34856868e-01 -6.61770329e-02 2.98672944e-01
-1.55289024e-01 -1.76711476e+00 -1.67815965e-02 3.80013138e-01
3.20632160e-01 5.21932900e-01 6.19035602e-01 -6.90721393e-01
-1.13654232e+00 -1.11757290e+00 4.43777680e-01 -3.17438185e-01
2.76599884e-01 -3.58196735e-01 -8.85972083e-01 7.33867168e-01
1.33076504e-01 -3.52790117e-01 7.50588775e-01 2.46489956e-03
3.36817913e-02 3.51515003e-02 -1.29937971e+00 2.65772551e-01
6.98732078e-01 -4.74306166e-01 -8.55153501e-01 7.27178752e-01
7.70781577e-01 -3.61416519e-01 -1.55733597e+00 2.77858257e-01
3.33841890e-01 -9.38857079e-01 1.37666249e+00 -4.62566018e-01
8.36493433e-01 -2.64338076e-01 2.29017041e-03 -9.47609961e-01
-2.28936642e-01 -1.74875613e-02 3.49806063e-02 8.04899454e-01
2.38743529e-01 -5.14941156e-01 9.20060039e-01 3.44324857e-01
-3.99013191e-01 -8.90850902e-01 -9.33776677e-01 -7.48101950e-01
1.68256253e-01 -4.30112600e-01 5.80084503e-01 6.34663582e-01
-5.88878751e-01 -6.02757223e-02 5.95082462e-01 3.28861833e-01
7.95665264e-01 4.89668369e-01 3.15501988e-01 -1.65833545e+00
-1.82271987e-01 -4.13000047e-01 -9.40459847e-01 -6.76333845e-01
2.21338600e-01 -1.42337203e+00 -2.41212845e-01 -1.79204655e+00
-7.32003078e-02 -6.26059115e-01 -2.01873451e-01 1.62695795e-01
5.19361377e-01 3.82452637e-01 1.92804374e-02 6.48086965e-02
1.21303990e-01 2.31653988e-01 1.54615521e+00 -4.26030695e-01
-5.84438592e-02 -1.26462534e-01 -1.43706903e-01 5.91291547e-01
7.61137247e-01 -6.05807960e-01 -2.18889024e-02 -3.16376090e-01
9.54595804e-02 -1.80960149e-01 7.81391144e-01 -1.22054434e+00
-2.50176013e-01 4.61759090e-01 5.46602130e-01 -7.91683495e-01
2.39370063e-01 -1.44972718e+00 3.35860342e-01 6.64416075e-01
-1.79942295e-01 -3.43968451e-01 -1.77214816e-02 4.76274073e-01
-2.09841430e-02 -7.23889410e-01 6.66130185e-01 -7.45824397e-01
-7.17747390e-01 4.17936087e-01 -3.93901676e-01 -6.53742492e-01
1.14323831e+00 -5.01146197e-01 9.23908055e-02 3.92637938e-01
-1.48074234e+00 -3.14439505e-01 7.10250676e-01 -7.19336467e-03
7.46352673e-01 -1.21949077e+00 -4.20675337e-01 2.13476807e-01
-3.11197877e-01 1.96243986e-01 3.37450773e-01 5.14701724e-01
-1.21642828e+00 1.26621783e-01 -4.79947984e-01 -9.72476244e-01
-1.01987219e+00 5.24913430e-01 2.31517285e-01 -3.10843438e-01
-9.15719092e-01 4.99476969e-01 -4.46050316e-01 -5.07351339e-01
-2.99437523e-01 -9.31182504e-01 -3.71029794e-01 4.99920957e-02
4.70697805e-02 3.58697653e-01 3.38471889e-01 -5.21957994e-01
-7.91728124e-02 8.77591372e-01 -1.33201841e-03 -9.69141349e-02
1.85365891e+00 3.01743656e-01 -3.79747480e-01 5.11557817e-01
1.38821650e+00 -3.17286789e-01 -1.11289954e+00 2.69492835e-01
4.81670238e-02 -2.46392906e-01 2.91897357e-01 -4.41752821e-01
-1.39802206e+00 8.88776660e-01 9.88447964e-01 7.17841387e-01
8.09206426e-01 3.69683713e-01 9.83750403e-01 1.58640817e-01
4.59745973e-01 -1.29778242e+00 5.59554808e-02 3.11146468e-01
8.56543362e-01 -1.11764848e+00 2.21553668e-01 -6.99860513e-01
1.62209183e-01 1.83699071e+00 5.01579121e-02 -6.91310942e-01
1.03974319e+00 3.99064034e-01 -1.84832290e-01 -9.47717369e-01
-1.14115901e-01 -3.88569295e-01 2.53783375e-01 6.83265150e-01
2.99327493e-01 3.27129960e-01 -3.29376966e-01 2.53196895e-01
-2.40405008e-01 -4.84497212e-02 4.57417518e-01 9.35015917e-01
-6.49126232e-01 -1.32923758e+00 -2.93110222e-01 6.79264307e-01
-2.09779665e-01 3.43552321e-01 -2.52835810e-01 9.37102735e-01
3.48679096e-01 3.65362912e-01 3.79439741e-01 -3.96839380e-01
1.96695372e-01 8.36013928e-02 9.60685730e-01 -8.22437942e-01
-7.71838844e-01 8.61716643e-02 -1.65702760e-01 -5.75098753e-01
-7.37285316e-01 -3.58271956e-01 -1.64697671e+00 6.44146204e-02
-2.61156410e-01 -1.15639172e-01 1.08971918e+00 8.57360661e-01
-6.59496412e-02 7.66535103e-01 6.89841509e-01 -1.17018819e+00
7.94848800e-02 -7.08591461e-01 -1.08795261e+00 5.87790132e-01
1.54143706e-01 -6.54038131e-01 -1.46306902e-01 3.77216041e-01]
|
[14.163503646850586, -2.739170551300049]
|
ff9b9ae7-85e5-4032-a317-dfbfd9b5522b
|
modular-visual-question-answering-via-code
|
2306.05392
| null |
https://arxiv.org/abs/2306.05392v1
|
https://arxiv.org/pdf/2306.05392v1.pdf
|
Modular Visual Question Answering via Code Generation
|
We present a framework that formulates visual question answering as modular code generation. In contrast to prior work on modular approaches to VQA, our approach requires no additional training and relies on pre-trained language models (LMs), visual models pre-trained on image-caption pairs, and fifty VQA examples used for in-context learning. The generated Python programs invoke and compose the outputs of the visual models using arithmetic and conditional logic. Our approach improves accuracy on the COVR dataset by at least 3% and on the GQA dataset by roughly 2% compared to the few-shot baseline that does not employ code generation.
|
['Dan Klein', 'Trevor Darrell', 'Andy Zeng', 'Cordelia Schmid', 'Arsha Nagrani', 'Kevin Yang', 'Kushal Khangaonkar', 'Medhini Narasimhan', 'Sanjay Subramanian']
|
2023-06-08
| null | null | null | null |
['code-generation', 'visual-question-answering-1']
|
['computer-code', 'computer-vision']
|
[ 1.37240112e-01 4.54725891e-01 2.29250982e-01 -4.18351620e-01
-1.41013694e+00 -8.43247533e-01 8.74715030e-01 1.35762304e-01
-1.44256577e-01 4.46035922e-01 1.59456253e-01 -7.37359226e-01
6.45042181e-01 -8.45730722e-01 -1.17515707e+00 2.81477664e-02
4.23134059e-01 5.19880891e-01 4.55895782e-01 -4.08342987e-01
2.65895545e-01 -2.15962648e-01 -1.49787664e+00 8.18455637e-01
9.34859574e-01 8.88932168e-01 4.13327292e-02 1.11864758e+00
-6.70948267e-01 1.49572635e+00 -8.45790625e-01 -9.02759790e-01
1.87714770e-02 -6.46110773e-01 -1.02225816e+00 -3.20872068e-02
1.01899433e+00 -2.49749780e-01 -1.09874755e-01 8.48847151e-01
8.68892372e-02 -8.08909386e-02 7.56999016e-01 -1.30073118e+00
-1.44098520e+00 4.32696521e-01 -4.18759763e-01 4.28150520e-02
7.04155445e-01 7.97601521e-01 1.34754205e+00 -1.15471613e+00
9.34440494e-01 1.32683969e+00 5.75696826e-01 9.75677133e-01
-1.60768843e+00 -3.12813997e-01 -7.33997226e-02 1.19638637e-01
-1.21483326e+00 -4.04566497e-01 3.81335795e-01 -7.82715440e-01
1.48862231e+00 -2.66719349e-02 7.11942196e-01 1.07407784e+00
-3.73097323e-02 9.11579669e-01 9.33369100e-01 -5.21719813e-01
3.46429735e-01 1.83205843e-01 1.19510062e-01 1.08797204e+00
-6.10743836e-02 -3.11474688e-02 -3.00601393e-01 -1.76627561e-01
4.26547647e-01 -5.24245501e-01 5.73477149e-02 -6.78145766e-01
-8.71444941e-01 1.13246906e+00 4.09551263e-01 -8.49680081e-02
-2.12168060e-02 7.50770807e-01 6.14871800e-01 3.02575499e-01
9.76980478e-02 7.27330685e-01 -1.12986527e-01 -5.38349450e-02
-9.95866537e-01 3.67839128e-01 5.56602597e-01 1.23208189e+00
8.60876679e-01 3.57324362e-01 -6.41993880e-01 3.04702580e-01
4.40530539e-01 6.54557467e-01 1.61537766e-01 -1.08519721e+00
8.27868819e-01 8.19288254e-01 9.32118371e-02 -3.85924935e-01
2.33710259e-01 1.33453161e-01 2.96662524e-02 6.45256162e-01
4.01456654e-01 2.71257535e-02 -1.31091464e+00 1.34122157e+00
-1.07390270e-01 -2.57969111e-01 3.62748861e-01 5.06210744e-01
1.23497784e+00 1.03092003e+00 6.64467454e-01 3.19781184e-01
1.34562564e+00 -1.23813748e+00 -4.24579322e-01 -3.74829352e-01
6.31535172e-01 -6.21227622e-01 1.73439622e+00 9.62801129e-02
-1.50534225e+00 -7.84909785e-01 -1.26304340e+00 -5.16445339e-01
-3.53316009e-01 -5.88153023e-03 4.10694718e-01 6.98016167e-01
-1.70755041e+00 -4.28513587e-02 -4.21871096e-01 -2.50762373e-01
8.76058817e-01 -8.98690075e-02 -1.68200031e-01 -1.83763295e-01
-6.71863198e-01 8.96554768e-01 1.63311183e-01 -6.13914967e-01
-1.70116532e+00 -9.76316810e-01 -1.33367968e+00 -2.83901338e-02
2.48492762e-01 -1.09041297e+00 1.73853183e+00 -1.51076937e+00
-1.29250073e+00 1.08526492e+00 -2.46704042e-01 -6.14808619e-01
5.30972183e-01 -1.16429947e-01 7.95274898e-02 5.28688073e-01
2.27283955e-01 1.14779806e+00 9.02522087e-01 -1.40931416e+00
-1.11457646e-01 1.12974718e-01 8.60221326e-01 2.91665196e-02
2.81786263e-01 1.32860854e-01 -5.55996895e-01 -4.03138787e-01
-8.64914775e-01 -5.24383068e-01 -2.16436341e-01 3.21838796e-01
2.12365091e-02 -2.64432192e-01 6.64466619e-01 -9.18566108e-01
7.55766511e-01 -2.00931740e+00 1.50294751e-01 -6.84102699e-02
1.90172374e-01 1.86718136e-01 -5.22617400e-01 3.39581698e-01
5.70214391e-02 2.22926319e-01 -3.92247200e-01 -2.78562486e-01
-2.11233199e-02 3.45558412e-02 -7.07373738e-01 -1.09364703e-01
7.34036982e-01 1.62778640e+00 -1.11045492e+00 -8.57997298e-01
2.14524373e-01 3.05530369e-01 -8.11567903e-01 5.50769567e-01
-1.08651054e+00 -6.53348267e-02 -1.01343408e-01 6.93288445e-01
2.39338845e-01 -5.90546548e-01 9.84892622e-02 3.54828313e-02
1.11626729e-01 2.85253048e-01 -5.07638454e-01 2.02343011e+00
-7.11718738e-01 9.79558229e-01 -3.66021365e-01 -7.61456490e-01
8.76583397e-01 4.01062161e-01 -3.29309404e-01 -1.01066494e+00
-1.34467185e-01 -1.85868487e-01 -4.25645739e-01 -7.56032884e-01
5.16324997e-01 -4.49335650e-02 -2.32503802e-01 4.95778739e-01
5.28450370e-01 -6.22758389e-01 3.55510384e-01 9.50518787e-01
1.13002944e+00 7.46864557e-01 3.27150583e-01 2.10881755e-02
5.20880938e-01 6.25826240e-01 -1.29802436e-01 7.45650530e-01
-1.08288959e-01 6.60123825e-01 7.73307383e-01 -3.75659019e-01
-1.52174282e+00 -1.33583367e+00 3.99698377e-01 1.31268525e+00
-4.09044139e-03 -7.38615394e-01 -8.86767030e-01 -9.76636410e-01
-2.45622560e-01 1.19594049e+00 -9.66870606e-01 8.94262493e-02
-2.99215317e-01 2.18503922e-02 7.69698083e-01 8.54192734e-01
2.65176833e-01 -1.32802224e+00 -6.91086769e-01 -5.98240718e-02
-7.85148293e-02 -9.86766040e-01 -2.50689089e-01 -1.60026252e-01
-6.17128134e-01 -1.13327003e+00 -6.45919025e-01 -9.40131068e-01
6.21538699e-01 9.98076200e-02 1.94944000e+00 2.94685274e-01
-5.05455971e-01 9.23368990e-01 -4.50302839e-01 -4.24068063e-01
-7.93139100e-01 -3.09647322e-01 -1.03430820e+00 -6.31411970e-01
3.92568260e-01 -7.07836300e-02 -5.22734463e-01 -3.62933785e-01
-8.82477582e-01 3.74080807e-01 4.40865934e-01 7.72850871e-01
4.10904735e-01 -1.14139068e+00 3.89187336e-01 -8.82904291e-01
5.38191319e-01 -3.67812932e-01 -9.26074028e-01 8.06262732e-01
-2.48244807e-01 3.39524597e-01 4.36533689e-01 -2.01211169e-01
-1.30863667e+00 2.79248238e-01 -1.77058391e-03 -4.55096811e-01
-1.66214228e-01 1.72030509e-01 -2.41161510e-02 -5.31822667e-02
1.14737225e+00 2.00442567e-01 -2.51651376e-01 1.26654029e-01
1.29208970e+00 3.42230797e-01 5.63533187e-01 -6.56239986e-01
1.01899409e+00 3.83889258e-01 -4.24356937e-01 -5.50977349e-01
-4.94151235e-01 -2.74297912e-02 -3.94679576e-01 -2.73342609e-01
1.44522119e+00 -1.11772871e+00 -4.00054336e-01 -1.58995479e-01
-1.49210668e+00 -6.00650549e-01 -4.80333149e-01 -3.67037535e-01
-8.08533788e-01 3.14494789e-01 -4.66570973e-01 -7.76875973e-01
-6.83135986e-01 -1.25953507e+00 1.13185048e+00 1.29373103e-01
-3.56866747e-01 -7.27187514e-01 2.86157995e-01 7.11788535e-01
4.55871671e-01 1.62167579e-01 1.22579098e+00 -2.31257737e-01
-1.06444550e+00 1.72309935e-01 -5.31195045e-01 2.55199790e-01
-5.02089977e-01 2.75566280e-01 -9.69862878e-01 8.45936835e-02
-6.00749016e-01 -1.23279524e+00 7.48749912e-01 -8.59474465e-02
1.00766146e+00 -5.15965782e-02 -1.03231512e-01 3.04701060e-01
1.72652721e+00 3.13032120e-01 8.65915418e-01 1.55174226e-01
6.97223365e-01 3.95970374e-01 3.70674729e-01 1.34599298e-01
6.38716578e-01 4.06202078e-01 7.13010550e-01 1.08383736e-03
-4.56038713e-01 -4.78374660e-01 5.28837442e-01 3.59680206e-01
1.41888559e-01 6.08536899e-02 -1.19359457e+00 1.14033556e+00
-1.72899663e+00 -1.11667275e+00 -1.06190048e-01 1.69553208e+00
9.69915032e-01 -2.14447841e-01 1.12278171e-01 -5.91155291e-01
2.53707439e-01 2.57189274e-01 -3.50946069e-01 -9.80612040e-01
1.75391361e-02 8.56365323e-01 1.41943768e-01 6.62425399e-01
-9.63611543e-01 1.25726998e+00 7.59017754e+00 3.39348465e-01
-6.10895574e-01 3.09466720e-01 5.09013832e-01 -1.10970840e-01
-8.35750461e-01 3.58005464e-01 -3.44877124e-01 -3.31216529e-02
1.06197906e+00 -1.36467755e-01 5.61984420e-01 1.16745222e+00
-3.75341356e-01 2.67842505e-02 -1.09367144e+00 9.16381478e-01
6.34599209e-01 -1.60076356e+00 4.46163118e-01 -4.49429631e-01
9.83082473e-01 2.77766846e-02 2.90328637e-02 8.59570265e-01
9.60643709e-01 -1.25242496e+00 9.98283744e-01 4.98525262e-01
9.97882247e-01 -5.33594251e-01 2.63815582e-01 -1.85875595e-01
-1.07529688e+00 7.85910264e-02 -3.78539890e-01 7.37597793e-02
1.26659989e-01 -2.07249582e-01 -9.75313723e-01 3.25742900e-01
6.72694802e-01 3.38028789e-01 -1.15879190e+00 9.25544739e-01
-7.35708237e-01 6.94700599e-01 4.83008921e-01 -2.89632440e-01
4.28340852e-01 7.55547211e-02 1.31807342e-01 1.25826609e+00
1.20339375e-02 -8.09794888e-02 -1.13983944e-01 1.28824639e+00
-3.09392601e-01 3.16944718e-01 -8.29195499e-01 -3.90079796e-01
-3.65193561e-03 1.09661984e+00 -1.44471273e-01 -9.27033126e-01
-1.10906744e+00 1.13255000e+00 5.63877821e-01 5.65778136e-01
-9.18967247e-01 -7.36320674e-01 2.64477730e-01 -1.58872567e-02
7.61592150e-01 -8.69185701e-02 -2.27140352e-01 -1.27779984e+00
-9.33566689e-02 -1.26248336e+00 4.33918059e-01 -1.63543761e+00
-1.14370823e+00 6.24148726e-01 -4.49866951e-02 -8.56368780e-01
-6.94034576e-01 -7.54090667e-01 -7.82571495e-01 9.12151515e-01
-1.16334105e+00 -1.43660522e+00 -5.21104932e-01 5.91313720e-01
7.63025522e-01 -2.07821190e-01 1.01918554e+00 -1.17308296e-01
2.13827789e-01 4.82199550e-01 -5.61692834e-01 3.79533947e-01
4.82289970e-01 -1.53380060e+00 1.03633142e+00 1.00876153e+00
4.88956630e-01 5.41276753e-01 6.65392756e-01 -5.46341419e-01
-1.26043332e+00 -1.07774186e+00 6.62018657e-01 -1.10274923e+00
8.31588507e-01 -6.31531835e-01 -9.55185592e-01 9.35600460e-01
1.05968928e+00 -1.14913732e-02 8.06943893e-01 -2.13277012e-01
-1.17583847e+00 2.36961231e-01 -9.05262589e-01 6.39436841e-01
9.06405449e-01 -1.00542927e+00 -1.27971995e+00 1.60495579e-01
1.03891826e+00 -1.61022708e-01 -4.47441816e-01 -3.76350470e-02
3.75564456e-01 -6.15604639e-01 9.51372385e-01 -9.81149077e-01
9.65255857e-01 -5.28339505e-01 -1.98912755e-01 -9.42989469e-01
1.17492490e-01 -2.90162683e-01 -1.72785550e-01 9.88363981e-01
6.55108035e-01 -2.21081115e-02 5.92934012e-01 5.66327512e-01
-6.35612607e-02 -2.55969018e-01 -6.64254665e-01 -4.97461140e-01
2.06047028e-01 -5.58006465e-01 2.55912781e-01 6.93926096e-01
1.81240831e-02 7.06998050e-01 -2.19236851e-01 -1.95933759e-01
5.00305295e-01 3.33425432e-01 1.24013734e+00 -6.21853173e-01
-6.85441375e-01 -2.29749680e-01 -3.76738995e-01 -9.67803776e-01
1.36634886e-01 -1.11475420e+00 2.04083011e-01 -2.14945459e+00
4.36328948e-01 2.24730223e-01 1.20542943e-01 7.52833009e-01
-2.67465323e-01 5.35360098e-01 3.89834762e-01 -1.50728047e-01
-1.20029283e+00 4.28469598e-01 9.16203320e-01 -7.01531470e-01
1.52693037e-02 -7.52068281e-01 -8.40707123e-01 3.79204690e-01
5.57428479e-01 -2.03793257e-01 -6.89957082e-01 -7.43479848e-01
6.42540216e-01 5.87051362e-02 6.80568933e-01 -9.19412732e-01
-1.39161691e-01 3.38426325e-03 4.50585812e-01 -4.86546069e-01
2.55266547e-01 -2.95285285e-01 -2.21449375e-01 3.01431209e-01
-5.31491101e-01 3.09536874e-01 5.92477977e-01 3.15996349e-01
-1.06687620e-01 -4.98446882e-01 5.29867649e-01 -6.20403290e-01
-9.75236058e-01 -1.07334368e-01 -5.70580959e-01 4.85093057e-01
1.20850110e+00 -4.13898118e-02 -9.17778611e-01 -6.86893702e-01
-5.14244139e-01 3.65443647e-01 6.14550650e-01 5.31554043e-01
8.52476001e-01 -1.18131542e+00 -5.36918163e-01 -2.49565169e-01
7.88407087e-01 -3.14594716e-01 1.16955481e-01 1.86677039e-01
-9.86197948e-01 3.81925106e-01 -2.80659258e-01 -4.92853969e-01
-1.28192377e+00 9.18511033e-01 2.83864498e-01 -7.63648190e-03
-3.33707780e-01 8.44269395e-01 3.96539688e-01 -3.18927139e-01
-2.73719206e-02 -1.69749334e-01 -8.31809491e-02 -1.95548326e-01
8.32113206e-01 -7.96664134e-02 -2.44647309e-01 -6.16753399e-01
-2.80977607e-01 4.21485633e-01 -4.19234820e-02 -5.74335992e-01
1.00004613e+00 3.12444627e-01 6.58681095e-02 4.89489377e-01
1.11137295e+00 -4.81915474e-02 -1.25523114e+00 -9.13010817e-03
-1.20664448e-01 -2.23909765e-01 -3.78610134e-01 -1.08546150e+00
-6.14903450e-01 1.45877278e+00 5.82569957e-01 -1.04842514e-01
7.31861472e-01 5.83926797e-01 3.52951080e-01 4.53785717e-01
2.57915944e-01 -8.43785405e-01 8.50930333e-01 4.17528033e-01
1.11359644e+00 -1.10813427e+00 -1.88793421e-01 -3.11107356e-02
-1.12511468e+00 1.00748777e+00 9.34184790e-01 -3.15764636e-01
-5.44033572e-02 1.82984322e-02 5.03407001e-01 -3.45556855e-01
-1.12545884e+00 -4.95127767e-01 4.21293139e-01 9.48105872e-01
4.36230659e-01 -2.15652615e-01 -4.86063100e-02 3.41266930e-01
6.04420528e-02 9.11836699e-02 4.47147131e-01 1.07075238e+00
-3.84998977e-01 -1.01057291e+00 -9.13187861e-02 1.52076721e-01
-3.31939071e-01 -5.29081047e-01 -5.59616029e-01 9.21878219e-01
-2.57522501e-02 8.52051258e-01 3.12479496e-01 -1.46630704e-01
3.28383327e-01 4.81427103e-01 7.14482427e-01 -9.01643038e-01
-7.39489496e-01 -4.45906132e-01 2.08621591e-01 -8.17960560e-01
-3.06998909e-01 -5.11961102e-01 -1.27987683e+00 2.77948678e-01
8.20207447e-02 -8.91461521e-02 5.41808128e-01 7.71925330e-01
3.84790629e-01 4.67159033e-01 -1.03274301e-01 -4.11596179e-01
-3.46773714e-01 -7.41237879e-01 4.43012893e-01 6.67750776e-01
2.21455947e-01 -9.92653668e-02 -1.04951501e-01 4.57527339e-01]
|
[10.864350318908691, 1.8184727430343628]
|
5c0bcef3-3c1a-45e8-8538-2693fc3ce48e
|
when-newer-is-not-better-does-deep-learning
|
2305.01801
| null |
https://arxiv.org/abs/2305.01801v1
|
https://arxiv.org/pdf/2305.01801v1.pdf
|
When Newer is Not Better: Does Deep Learning Really Benefit Recommendation From Implicit Feedback?
|
In recent years, neural models have been repeatedly touted to exhibit state-of-the-art performance in recommendation. Nevertheless, multiple recent studies have revealed that the reported state-of-the-art results of many neural recommendation models cannot be reliably replicated. A primary reason is that existing evaluations are performed under various inconsistent protocols. Correspondingly, these replicability issues make it difficult to understand how much benefit we can actually gain from these neural models. It then becomes clear that a fair and comprehensive performance comparison between traditional and neural models is needed. Motivated by these issues, we perform a large-scale, systematic study to compare recent neural recommendation models against traditional ones in top-n recommendation from implicit data. We propose a set of evaluation strategies for measuring memorization performance, generalization performance, and subgroup-specific performance of recommendation models. We conduct extensive experiments with 13 popular recommendation models (including two neural models and 11 traditional ones as baselines) on nine commonly used datasets. Our experiments demonstrate that even with extensive hyper-parameter searches, neural models do not dominate traditional models in all aspects, e.g., they fare worse in terms of average HitRate. We further find that there are areas where neural models seem to outperform non-neural models, for example, in recommendation diversity and robustness between different subgroups of users and items. Our work illuminates the relative advantages and disadvantages of neural models in recommendation and is therefore an important step towards building better recommender systems.
|
['Tobias Schnabel', 'Jundong Li', 'Yushun Dong']
|
2023-05-02
| null | null | null | null |
['memorization']
|
['natural-language-processing']
|
[-3.81733738e-02 -5.05328059e-01 -6.03517830e-01 -4.27047968e-01
-3.56685400e-01 -6.09179854e-01 5.18348396e-01 -9.49475318e-02
-4.71395731e-01 6.68326139e-01 6.16924345e-01 -6.07929111e-01
-6.96102381e-01 -7.61787891e-01 -6.98095739e-01 -5.11129200e-01
-1.17731623e-01 3.97662252e-01 2.67963838e-02 -5.85353434e-01
6.36770010e-01 3.66637677e-01 -2.01385331e+00 4.38186854e-01
8.36365700e-01 6.98811412e-01 -5.51397502e-02 2.20723987e-01
1.31890878e-01 4.07559097e-01 -4.31462318e-01 -5.82128406e-01
2.04295278e-01 -2.02002347e-01 -5.61371982e-01 -5.90023160e-01
8.03807497e-01 -4.83542711e-01 -3.93914521e-01 7.16572583e-01
7.54279196e-01 6.79981112e-01 7.78800905e-01 -6.69409633e-01
-1.47412503e+00 1.20628870e+00 -1.72814250e-01 3.97854567e-01
1.42464519e-01 -9.28592831e-02 1.28786898e+00 -9.15015697e-01
2.65760005e-01 1.07740390e+00 1.02186847e+00 5.93561888e-01
-1.18478644e+00 -7.83633471e-01 6.14198029e-01 1.70435347e-02
-1.18078423e+00 -4.47348028e-01 1.97476983e-01 -1.47979781e-01
1.09824514e+00 3.59665722e-01 4.06627029e-01 1.42195594e+00
1.62714303e-01 6.06283486e-01 9.21571076e-01 -2.41413817e-01
4.84995171e-02 2.66668826e-01 7.45410681e-01 2.58899778e-01
8.97171319e-01 4.34349626e-01 -5.70701957e-01 -3.20834696e-01
7.53766477e-01 1.84872344e-01 -5.45666397e-01 -5.80909736e-02
-8.93769562e-01 9.40778017e-01 3.97813469e-01 4.88856822e-01
-5.28572559e-01 1.05174355e-01 2.21180916e-01 4.45864409e-01
6.26968563e-01 9.55275357e-01 -5.78011036e-01 -7.21641853e-02
-9.90283787e-01 5.67425191e-01 8.86128545e-01 5.83393872e-01
2.38857716e-01 2.38515586e-01 -3.39761078e-01 1.20818114e+00
2.47070387e-01 3.89225185e-01 7.18609571e-01 -9.28029895e-01
1.83131084e-01 1.75346881e-01 1.36323705e-01 -1.17677355e+00
-4.55909222e-01 -1.03796721e+00 -8.94786596e-01 -1.57612860e-01
4.26883578e-01 -1.11201100e-01 -5.15324891e-01 1.78366959e+00
-3.44862252e-01 7.39684477e-02 -1.26130924e-01 9.41102922e-01
9.95978713e-01 5.12437344e-01 -3.50394920e-02 -1.49212584e-01
1.04145133e+00 -1.02800906e+00 -4.34147686e-01 -2.65959173e-01
6.65019929e-01 -6.45059705e-01 1.10509241e+00 6.74724996e-01
-1.21161091e+00 -6.70218349e-01 -9.58471894e-01 2.64690310e-01
-5.09591401e-01 3.29138711e-02 1.07726312e+00 1.01534176e+00
-1.01804316e+00 9.85388517e-01 -3.16962034e-01 -5.90721607e-01
5.21206893e-02 6.86665356e-01 -1.23651908e-03 -1.71986490e-01
-1.34214747e+00 1.01280332e+00 1.02592044e-01 1.18281044e-01
-3.79940063e-01 -6.34659708e-01 -1.60112649e-01 1.78319633e-01
3.40604395e-01 -8.73149455e-01 1.29690659e+00 -9.58890975e-01
-1.35963523e+00 9.19131264e-02 1.87597647e-01 -4.43312466e-01
-1.98418826e-01 -4.53654855e-01 -6.93707645e-01 -6.79205060e-01
-5.28558791e-01 3.64062965e-01 2.93796211e-01 -1.29859638e+00
-6.32121265e-01 -2.12175220e-01 3.16822141e-01 3.37884068e-01
-7.30995595e-01 4.85567413e-02 -4.51335996e-01 -9.35409009e-01
-1.06131211e-01 -1.06795347e+00 -2.22448751e-01 -5.87106347e-01
1.85530484e-02 -4.34943259e-01 -7.12942258e-02 -3.36955428e-01
1.88084531e+00 -1.75779843e+00 -4.47763726e-02 4.16069448e-01
5.87539785e-02 4.34233725e-01 -5.33319056e-01 4.04849201e-01
1.27221435e-01 3.75570089e-01 4.81929898e-01 -4.21603397e-02
6.26980364e-02 1.97548375e-01 -5.20978630e-01 2.20094323e-01
-5.37177980e-01 9.35224593e-01 -6.95966005e-01 2.27564186e-01
-1.64121911e-01 7.03588963e-01 -9.46268916e-01 -3.60166132e-02
-7.35284463e-02 -9.21947043e-03 -4.28318202e-01 4.56713140e-01
3.68794352e-01 -3.92848730e-01 2.67345876e-01 -1.33963510e-01
-4.21340317e-02 8.21407735e-01 -1.04906523e+00 1.48304093e+00
-2.14819312e-01 2.28505969e-01 -4.10250723e-01 -7.78844595e-01
7.28861034e-01 2.16957018e-01 1.05656281e-01 -9.40087020e-01
1.03590712e-01 6.06792085e-02 3.72329682e-01 -9.85390842e-02
1.25020325e+00 1.45461500e-01 2.05635518e-01 9.94780123e-01
-1.01793565e-01 5.54903090e-01 6.27176836e-02 3.54449488e-02
7.13158250e-01 -7.13897822e-03 -5.28793447e-02 -3.77031326e-01
-3.92883411e-03 -4.03606236e-01 3.10487270e-01 1.41465724e+00
4.10733700e-01 5.63470423e-01 -1.54420689e-01 -5.53082108e-01
-7.20953584e-01 -6.64932191e-01 3.16781104e-02 1.85466135e+00
-1.06495850e-01 -4.39991236e-01 -4.31206524e-01 -6.39372945e-01
7.48783201e-02 9.88343656e-01 -7.72663116e-01 -3.41172963e-01
-5.47377527e-01 -1.19751108e+00 5.83779871e-01 7.11212873e-01
-6.10708306e-03 -1.02352571e+00 -1.32689521e-01 1.87247440e-01
-3.12506519e-02 -5.29677331e-01 -5.22216380e-01 1.51342943e-01
-1.33783162e+00 -6.55850053e-01 -6.98975801e-01 -4.34078932e-01
3.98795784e-01 1.05919290e+00 1.57556117e+00 4.88740116e-01
6.37723148e-01 2.92877555e-01 -5.67550957e-01 -1.29269674e-01
-4.67259772e-02 5.03592789e-01 5.22684395e-01 -4.77226615e-01
5.40974617e-01 -6.90529466e-01 -8.63947332e-01 7.31723845e-01
-7.63269007e-01 -3.46354485e-01 7.90798247e-01 7.43952513e-01
3.14007610e-01 -1.06328741e-01 9.41595554e-01 -1.19573736e+00
1.37293148e+00 -8.09968829e-01 -1.13139987e-01 4.37354267e-01
-1.44088888e+00 2.19968874e-02 6.32419705e-01 -7.62588322e-01
-9.59085882e-01 -8.26687872e-01 -1.89779341e-01 -5.87208336e-03
-2.64767036e-02 9.26511467e-01 3.83428663e-01 5.96688315e-02
1.23242640e+00 1.48812011e-01 -2.52142102e-01 -8.05244267e-01
5.84943891e-01 7.82783329e-01 3.29134375e-01 -9.56590295e-01
3.91786903e-01 -1.02915689e-01 -6.75461233e-01 -4.56123203e-01
-9.43226099e-01 -2.48785779e-01 -4.24450338e-02 -2.94280481e-02
1.31111771e-01 -8.27221513e-01 -5.35830855e-01 -2.66171359e-02
-6.02348626e-01 -4.00593311e-01 -1.01524308e-01 7.59698391e-01
-2.87239756e-02 2.56328255e-01 -8.31846714e-01 -6.17972493e-01
-7.41266668e-01 -1.06872582e+00 2.32118770e-01 2.96478271e-01
-4.16820943e-01 -1.01751423e+00 1.62223116e-01 3.94697070e-01
1.13859987e+00 -7.61265159e-01 1.01494730e+00 -1.03349507e+00
-1.77535743e-01 -9.44622755e-02 -1.83605716e-01 3.47937018e-01
-1.32279560e-01 -5.07740080e-02 -8.44583809e-01 -4.54272568e-01
-4.48904276e-01 -3.04652065e-01 9.47420180e-01 6.61022365e-01
1.27274883e+00 -2.60573417e-01 -2.70899355e-01 4.75944787e-01
1.08434856e+00 2.76519865e-01 6.52342498e-01 4.33899850e-01
4.42284882e-01 3.69641453e-01 3.65410119e-01 1.81195319e-01
3.89337122e-01 7.97212183e-01 1.21530518e-01 2.59062946e-01
-5.57857938e-02 -1.76390946e-01 5.06713688e-01 1.07711112e+00
-7.99690366e-01 -7.15586782e-01 -3.30608398e-01 1.72401562e-01
-1.86356878e+00 -1.12459874e+00 6.77244514e-02 2.63896179e+00
5.30441940e-01 1.43157139e-01 4.30896640e-01 -1.03684053e-01
4.88655925e-01 -9.03821662e-02 -5.45823693e-01 -5.94869554e-01
-1.39255136e-01 2.37039417e-01 5.03657699e-01 6.22411706e-02
-7.12938607e-01 7.57334054e-01 7.52932930e+00 7.62355030e-01
-1.04252207e+00 -7.26641268e-02 5.62887132e-01 -4.66861635e-01
-5.71722269e-01 -3.33203942e-01 -1.05788016e+00 3.13374609e-01
1.33435810e+00 -4.95247729e-02 8.56710017e-01 6.99597716e-01
-3.58756706e-02 2.86504775e-01 -1.13693345e+00 6.27141654e-01
2.93418974e-01 -1.44450021e+00 4.65159982e-01 3.13822359e-01
8.90885949e-01 3.89989287e-01 4.58178610e-01 8.88258517e-01
5.48946917e-01 -1.14003754e+00 3.93403053e-01 5.14598191e-01
5.00636637e-01 -7.20479667e-01 8.76626134e-01 2.43061945e-01
-6.06474042e-01 -1.65303275e-01 -7.88100779e-01 -2.63630778e-01
-1.35892227e-01 2.99034506e-01 -1.91348150e-01 5.70847690e-01
7.09771633e-01 6.92591846e-01 -5.59852004e-01 1.20954132e+00
7.28551447e-02 9.98631716e-01 -1.21411577e-01 -3.00005704e-01
1.64288327e-01 -2.13354632e-01 7.65319243e-02 1.22851682e+00
5.92988133e-01 1.51292026e-01 -1.86379209e-01 4.69182342e-01
-3.04092467e-01 3.57255399e-01 -6.17215633e-01 7.01873899e-02
6.24004841e-01 1.01327717e+00 -5.91742575e-01 -7.95822814e-02
-7.03825653e-01 4.36708629e-01 4.33755934e-01 4.87454176e-01
-7.32736409e-01 -4.33398760e-04 7.04960823e-01 3.13817710e-02
3.91121000e-01 2.05356479e-02 -2.89801568e-01 -1.12008417e+00
-3.77217919e-01 -1.25651610e+00 4.63449597e-01 -6.25436604e-01
-1.68632281e+00 6.30390108e-01 9.01011229e-02 -1.00040042e+00
-2.00844750e-01 -6.26116216e-01 -3.73814166e-01 6.42373443e-01
-1.22491813e+00 -7.05829442e-01 -6.45262844e-05 3.44187796e-01
4.91486996e-01 -4.66955781e-01 1.09849346e+00 5.51032186e-01
-7.30129302e-01 1.20569718e+00 5.65418661e-01 -3.67338747e-01
9.07173634e-01 -8.00250232e-01 4.55095977e-01 4.92650598e-01
4.57786918e-01 1.46042573e+00 6.25078499e-01 -4.25066292e-01
-1.52497303e+00 -8.51388931e-01 6.56712830e-01 -5.68465173e-01
3.19743127e-01 8.31901133e-02 -9.90809619e-01 7.07831085e-01
2.07610160e-01 -5.03111839e-01 1.12981904e+00 1.09620142e+00
-7.47144818e-01 -1.85214411e-02 -6.72016919e-01 7.64821827e-01
1.33011401e+00 -1.73440382e-01 -6.88674331e-01 -9.51716397e-03
5.94476938e-01 -5.67192398e-03 -9.45515513e-01 4.02168572e-01
1.29242980e+00 -1.22338247e+00 1.31382763e+00 -8.92373562e-01
4.35614496e-01 7.50119090e-02 -3.35244685e-01 -1.44716167e+00
-1.03159726e+00 -3.23682159e-01 -3.61723036e-01 9.81119990e-01
7.46981859e-01 -7.20577240e-01 7.01787293e-01 7.39577293e-01
-1.50030598e-01 -1.00945127e+00 -2.34165788e-01 -8.00497949e-01
4.94210422e-01 -5.76429844e-01 6.62289679e-01 1.10964131e+00
-1.00697726e-01 6.37989104e-01 -7.01747894e-01 -1.46199152e-01
4.04367559e-02 1.77037492e-01 6.83800101e-01 -1.52183330e+00
-5.45153916e-01 -8.65302563e-01 2.59509593e-01 -1.50835097e+00
1.69969667e-02 -9.29787993e-01 -3.13865155e-01 -1.56032765e+00
3.30750406e-01 -6.92615628e-01 -1.06942642e+00 3.79031330e-01
-1.24332584e-01 4.20970917e-01 -3.66139449e-02 3.36357802e-01
-6.21908307e-01 2.44601503e-01 1.09915578e+00 1.63329870e-01
-3.59478921e-01 2.28849038e-01 -1.65542185e+00 5.37616253e-01
9.23576117e-01 -4.22495335e-01 -7.12026536e-01 -6.77851498e-01
6.77110374e-01 -3.22086722e-01 -2.45082229e-01 -7.70820498e-01
2.91865706e-01 -2.27732003e-01 3.13780040e-01 -4.42482352e-01
1.83405593e-01 -4.76883769e-01 1.25520989e-01 1.09938383e-01
-6.87590122e-01 3.53942573e-01 1.64517432e-01 5.15770197e-01
3.83669645e-01 -2.04409674e-01 2.39331141e-01 5.95300179e-03
-4.10658181e-01 2.53913254e-01 -4.38825011e-01 -1.30410001e-01
9.90242362e-02 -3.53588730e-01 -6.23038173e-01 -6.14746511e-01
-2.63720006e-01 -1.26991078e-01 3.79633278e-01 8.39393854e-01
4.84944195e-01 -1.13259923e+00 -7.11260259e-01 1.95684619e-02
1.86884314e-01 -7.45678186e-01 2.70673007e-01 7.35222042e-01
4.69389074e-02 6.73275054e-01 -6.00978434e-02 7.28319283e-04
-1.09041798e+00 5.87089360e-01 2.14511603e-01 -2.56617635e-01
-3.81307960e-01 8.47188175e-01 4.68280492e-03 -6.12387419e-01
6.83077574e-01 -4.34107214e-01 -5.46126902e-01 5.26856817e-02
7.10783362e-01 5.56617558e-01 1.31145939e-01 -3.50091577e-01
-1.23048970e-03 4.56047058e-01 -5.37607074e-01 1.54062942e-01
1.50863671e+00 8.09065998e-02 9.45532694e-02 2.97125548e-01
5.78466892e-01 3.27746607e-02 -5.33682585e-01 -3.23184997e-01
-1.91290572e-01 -3.10313880e-01 3.38747770e-01 -1.20231926e+00
-1.21478903e+00 4.54429477e-01 5.60427070e-01 3.07277828e-01
1.00759768e+00 -4.70478296e-01 7.10866511e-01 8.57907236e-01
4.36403036e-01 -9.88937497e-01 -3.40962887e-01 7.42731094e-01
7.05586493e-01 -9.80285525e-01 3.11118454e-01 1.18138500e-01
-4.99456704e-01 9.47723389e-01 5.67358136e-01 -1.96604222e-01
7.81580389e-01 -1.91415355e-01 -6.33839965e-02 -1.48756936e-01
-1.16405773e+00 -1.45609200e-03 6.82320416e-01 3.44376117e-01
1.10552454e+00 5.67165501e-02 -6.53175950e-01 1.05255687e+00
-1.68514669e-01 2.14280009e-01 3.05857688e-01 5.00123620e-01
-4.80871081e-01 -1.33304071e+00 -1.24088675e-01 9.49294090e-01
-6.35572493e-01 -4.25846070e-01 -2.96740174e-01 7.49000847e-01
-1.60054192e-01 1.17298591e+00 -2.25092158e-01 -9.60586250e-01
4.83024448e-01 4.09303345e-02 4.35058862e-01 -5.32248616e-01
-1.20843446e+00 -1.92609280e-01 4.38585967e-01 -4.63615417e-01
-3.63268822e-01 -4.27059144e-01 -5.73689878e-01 -7.23972261e-01
-7.29470074e-01 4.90605474e-01 5.64420760e-01 7.72605479e-01
6.09263122e-01 4.23598319e-01 2.87489027e-01 -8.90330434e-01
-9.16735411e-01 -1.13710237e+00 -5.20415723e-01 3.20882738e-01
-2.28989303e-01 -7.28205442e-01 -3.33155483e-01 -5.48254609e-01]
|
[10.096728324890137, 5.6659016609191895]
|
72f619d6-30c0-4687-b958-e17da33e8c8d
|
a-gromov-wasserstein-geometric-view-of
|
2306.08854
| null |
https://arxiv.org/abs/2306.08854v1
|
https://arxiv.org/pdf/2306.08854v1.pdf
|
A Gromov--Wasserstein Geometric View of Spectrum-Preserving Graph Coarsening
|
Graph coarsening is a technique for solving large-scale graph problems by working on a smaller version of the original graph, and possibly interpolating the results back to the original graph. It has a long history in scientific computing and has recently gained popularity in machine learning, particularly in methods that preserve the graph spectrum. This work studies graph coarsening from a different perspective, developing a theory for preserving graph distances and proposing a method to achieve this. The geometric approach is useful when working with a collection of graphs, such as in graph classification and regression. In this study, we consider a graph as an element on a metric space equipped with the Gromov--Wasserstein (GW) distance, and bound the difference between the distance of two graphs and their coarsened versions. Minimizing this difference can be done using the popular weighted kernel $K$-means method, which improves existing spectrum-preserving methods with the proper choice of the kernel. The study includes a set of experiments to support the theory and method, including approximating the GW distance, preserving the graph spectrum, classifying graphs using spectral information, and performing regression using graph convolutional networks. Code is available at https://github.com/ychen-stat-ml/GW-Graph-Coarsening .
|
['Jie Chen', 'Yun Yang', 'Rentian Yao', 'Yifan Chen']
|
2023-06-15
| null | null | null | null |
['graph-classification']
|
['graphs']
|
[ 1.44940332e-01 2.50811100e-01 -9.05348957e-02 -1.55599609e-01
-1.67502418e-01 -3.72009754e-01 2.21857920e-01 4.59189832e-01
-2.42227808e-01 4.81941491e-01 -3.62919234e-02 -1.75154090e-01
-3.90195012e-01 -1.17560959e+00 -6.28793776e-01 -7.84215569e-01
-4.57738131e-01 2.50382662e-01 1.55810192e-01 -1.92851260e-01
1.83440387e-01 6.13677740e-01 -1.18297637e+00 -2.88579971e-01
8.84723306e-01 7.32303500e-01 2.85835844e-02 7.29290307e-01
5.48848324e-02 3.12424719e-01 -2.48082429e-01 -4.30122077e-01
5.78916073e-01 -6.56240165e-01 -9.05661285e-01 7.72387758e-02
5.53716123e-01 4.15973485e-01 -6.03636205e-01 1.58785784e+00
3.61619532e-01 4.08758074e-01 4.16636914e-01 -1.38509774e+00
-1.06602418e+00 6.02366269e-01 -6.23951375e-01 1.90284234e-02
3.08986098e-01 -4.55497384e-01 1.08489108e+00 -4.78166610e-01
5.38781881e-01 1.19158888e+00 9.65151548e-01 2.82826453e-01
-1.58077848e+00 -4.75489408e-01 -2.20718622e-01 2.96315938e-01
-1.80816555e+00 2.41139635e-01 1.11512697e+00 -4.79950339e-01
5.53956747e-01 3.49912286e-01 8.25500190e-01 3.42688352e-01
4.91358727e-01 1.27378300e-01 1.05463767e+00 -8.62008989e-01
1.34076938e-01 -1.78325251e-01 4.43670690e-01 1.20404470e+00
4.63199973e-01 -1.43740803e-01 -1.12173811e-01 -2.17792913e-01
6.94169343e-01 9.28703323e-02 -5.87566078e-01 -7.55762279e-01
-1.12070000e+00 1.06799889e+00 7.67998278e-01 5.55811644e-01
-8.16785470e-02 1.20650090e-01 1.42220020e-01 4.64925259e-01
8.02266777e-01 4.28989589e-01 4.32209224e-02 5.89430153e-01
-7.57413268e-01 9.24339965e-02 8.77049804e-01 8.79639328e-01
1.16946948e+00 -2.09978044e-01 1.08918868e-01 6.43337965e-01
8.88568908e-02 1.76415548e-01 2.82858729e-01 -7.51653671e-01
-9.59837716e-03 8.45364511e-01 -4.48969215e-01 -1.32515049e+00
-5.44396758e-01 -2.33926490e-01 -1.16129375e+00 1.97285041e-01
6.54591978e-01 1.48811355e-01 -6.66707993e-01 1.73983371e+00
4.23833996e-01 5.60566247e-01 -3.63426536e-01 7.20903456e-01
5.32085121e-01 5.22382140e-01 -3.58059704e-01 -2.80131966e-01
9.97350812e-01 -7.96690941e-01 -4.87863302e-01 2.80006289e-01
6.84354186e-01 -5.58182776e-01 1.07332122e+00 3.94568324e-01
-7.59682536e-01 -5.22437453e-01 -1.17448342e+00 -1.19725227e-01
-7.10150838e-01 -2.36712947e-01 3.65972549e-01 7.34520257e-01
-1.63380945e+00 1.17899954e+00 -5.57255983e-01 -6.56697690e-01
2.69741733e-02 3.14780295e-01 -5.42384565e-01 -1.89900428e-01
-1.21716356e+00 7.53788769e-01 5.53894341e-01 -9.27495211e-02
-1.20513104e-01 -7.06259131e-01 -1.01767147e+00 1.40226156e-01
3.05681914e-01 -5.24402559e-01 4.84916002e-01 -9.44462121e-01
-1.14500940e+00 9.04016972e-01 2.23182291e-01 -4.78645355e-01
3.66290301e-01 2.84153700e-01 -4.47435617e-01 1.18144648e-02
2.33053956e-02 1.95587486e-01 9.52612519e-01 -7.47448385e-01
-1.08023353e-01 -5.30586004e-01 3.42295170e-02 1.50177488e-02
-2.13369533e-01 -4.14316684e-01 -3.07552457e-01 -7.11988270e-01
1.65906549e-01 -1.11331737e+00 -3.22663367e-01 -2.28254065e-01
-5.03771007e-01 -3.62007380e-01 7.75720835e-01 -6.59931481e-01
1.24039125e+00 -2.17472196e+00 3.09232295e-01 7.01002479e-01
7.12494135e-01 1.78551376e-01 -2.18112275e-01 7.94606864e-01
-5.80482662e-01 1.89725772e-01 -5.42804956e-01 -7.32493326e-02
-3.65340672e-02 -4.32224907e-02 1.62496686e-01 9.86322641e-01
-1.69417143e-01 7.10504889e-01 -8.67908359e-01 -2.22877026e-01
2.87032127e-01 5.20487249e-01 -4.43364352e-01 -8.82815048e-02
1.17150307e-01 1.09042674e-01 -4.51406054e-02 1.88723490e-01
7.83586919e-01 -3.87499750e-01 2.30738044e-01 -3.17019552e-01
6.73739016e-02 -4.71156865e-01 -1.52387869e+00 1.61306202e+00
-6.33788109e-02 6.71649635e-01 1.36066929e-01 -1.47416997e+00
1.02136254e+00 -1.70854643e-01 7.02621520e-01 -3.30264330e-01
1.50386900e-01 9.28778648e-02 -1.17436878e-01 -7.05996156e-02
2.18776897e-01 -1.91932857e-01 1.01974472e-01 3.79700243e-01
-3.52849811e-03 -1.28580198e-01 4.07733440e-01 3.20764959e-01
1.30724978e+00 -3.04533899e-01 6.15722537e-01 -8.63700867e-01
6.08858466e-01 -2.75667965e-01 1.18902110e-01 4.64017093e-01
-1.50639459e-01 4.36790109e-01 4.48419362e-01 -5.22908866e-01
-8.79018962e-01 -1.16324294e+00 -1.41043365e-01 8.38427961e-01
3.63254130e-01 -5.36459386e-01 -1.08885109e+00 -4.98239368e-01
2.54242063e-01 5.50778270e-01 -8.86885047e-01 -5.91273069e-01
-2.93854058e-01 -5.03049195e-01 3.31655294e-01 3.12998779e-02
4.36615527e-01 -7.40491807e-01 1.57904923e-01 5.67878075e-02
-8.71993452e-02 -7.13189960e-01 -1.01644981e+00 4.20990363e-02
-7.22460032e-01 -1.49810147e+00 -5.30517936e-01 -9.24096048e-01
9.04795229e-01 4.58612353e-01 9.12256956e-01 4.38078016e-01
-4.42184269e-01 4.63517666e-01 -4.36335683e-01 -9.41613317e-02
-2.95292586e-01 -1.41163077e-02 -1.08731110e-02 2.30022863e-01
2.81061321e-01 -5.63466370e-01 -2.84720600e-01 2.24132225e-01
-1.06285954e+00 -7.01071322e-02 1.43210977e-01 6.46837771e-01
6.64861202e-01 4.05626416e-01 2.84744859e-01 -9.00820196e-01
9.57730532e-01 -3.87453556e-01 -7.56190658e-01 3.31652135e-01
-9.23028052e-01 3.66103113e-01 8.02119374e-01 -2.59234250e-01
-2.47134805e-01 4.62582335e-03 1.57706738e-01 -5.27079642e-01
2.19712511e-01 5.03910303e-01 -6.97028339e-02 -7.06650376e-01
7.71116376e-01 -2.45429985e-02 3.02626401e-01 -2.71611720e-01
6.19636238e-01 2.61207014e-01 3.16669405e-01 -3.22053730e-01
8.77099454e-01 4.21241730e-01 4.78751063e-01 -1.17403543e+00
-6.27927065e-01 -7.08090127e-01 -8.27168226e-01 -2.64654577e-01
8.90543103e-01 -1.89607367e-01 -5.38201809e-01 3.33121628e-01
-8.60419631e-01 -3.37103486e-01 -4.96455699e-01 6.29606366e-01
-5.81538200e-01 7.00085998e-01 -4.37447697e-01 -2.00142145e-01
-1.87172815e-01 -8.35890591e-01 7.07164705e-01 3.93756181e-02
3.01512890e-02 -1.52585220e+00 6.02499187e-01 2.47728778e-03
2.35015884e-01 6.89516246e-01 9.59403992e-01 -5.85234404e-01
-2.46327907e-01 -3.13599050e-01 -3.07381153e-01 4.47433412e-01
2.74119884e-01 -1.05502889e-01 -5.06106973e-01 -5.56011260e-01
-4.03851271e-02 2.42484033e-01 9.34931755e-01 7.15011299e-01
1.15270042e+00 -2.34586492e-01 -2.66690016e-01 8.69524181e-01
1.69820404e+00 -1.44230545e-01 4.98970062e-01 -5.60485832e-02
1.07884228e+00 4.66559231e-01 6.09977506e-02 1.57573983e-01
2.21808448e-01 4.95524555e-01 3.28072459e-01 -9.33052748e-02
-2.87847996e-01 -1.78114563e-01 9.31555703e-02 9.58993077e-01
-2.54382312e-01 -1.32406473e-01 -8.16807270e-01 2.92722195e-01
-1.88822687e+00 -9.64694142e-01 -3.02294016e-01 2.40623999e+00
5.18760383e-01 -2.24921644e-01 2.40495861e-01 3.34787816e-01
1.32352483e+00 2.56070584e-01 -2.94781506e-01 -4.48798746e-01
8.08708221e-02 4.44978654e-01 8.94617558e-01 9.89409089e-01
-8.74538541e-01 7.44871616e-01 5.70831060e+00 8.05541277e-01
-1.12798834e+00 -8.53952300e-03 2.96651661e-01 3.21578264e-01
-2.51447290e-01 -3.07415649e-02 -2.04083949e-01 2.67651618e-01
9.16225672e-01 -6.25653565e-01 9.34827268e-01 8.63756359e-01
8.11771303e-02 1.24900743e-01 -9.83807027e-01 1.04588497e+00
2.46113122e-01 -1.40746200e+00 -3.57538159e-03 2.21863598e-01
5.25453568e-01 4.78657372e-02 -3.12698483e-01 3.40231620e-02
3.24313313e-01 -1.13671219e+00 3.50287765e-01 5.13756871e-01
8.43835652e-01 -9.20061052e-01 5.81625640e-01 1.83473662e-01
-1.71652937e+00 3.97748709e-01 -4.11191583e-01 -8.49467423e-03
-2.71520078e-01 9.32773352e-01 -7.58472621e-01 8.55311275e-01
5.77235818e-01 8.31098080e-01 -6.18023813e-01 1.00576651e+00
7.55456239e-02 4.13463593e-01 -1.02313638e-01 -1.18508756e-01
-5.46194427e-02 -9.28858042e-01 5.74448586e-01 1.11057127e+00
4.45659369e-01 8.49839672e-03 4.28788960e-01 8.03879917e-01
-3.42168331e-01 4.87183213e-01 -9.57068801e-01 -1.86165541e-01
1.61385074e-01 1.30571902e+00 -1.18722141e+00 -1.40166551e-01
-4.47445810e-01 9.31271315e-01 5.55983841e-01 3.03070962e-01
-7.25552440e-01 -8.45714271e-01 3.83813530e-01 3.27416092e-01
6.59716278e-02 -4.93806779e-01 -9.51648951e-02 -1.07074857e+00
-6.60738796e-02 -3.67753655e-01 5.57221830e-01 -5.66969454e-01
-1.27968442e+00 5.03086925e-01 8.47716555e-02 -8.76684129e-01
2.29360953e-01 -6.65921926e-01 -6.22539580e-01 8.50146711e-01
-1.09224820e+00 -8.35708857e-01 -5.36345005e-01 9.87473547e-01
-5.30289859e-03 1.50717452e-01 6.02770507e-01 3.66904348e-01
-3.21411729e-01 3.92666280e-01 2.53107190e-01 1.89921036e-01
6.79394782e-01 -1.53662419e+00 3.38558435e-01 9.15691674e-01
3.67867798e-01 4.79522467e-01 5.68373263e-01 -7.47816026e-01
-1.46593583e+00 -1.18945503e+00 6.81743324e-01 -5.53215556e-02
9.11063731e-01 -3.41339350e-01 -1.03267777e+00 6.28119528e-01
1.11334600e-01 3.59522581e-01 5.48663616e-01 6.19722642e-02
-3.05578709e-01 -1.39088780e-01 -1.35906315e+00 5.12199342e-01
1.17664301e+00 -4.91394848e-01 -1.93678916e-01 4.73040283e-01
6.72848463e-01 -1.99910387e-01 -1.09685087e+00 1.26206517e-01
2.81299919e-01 -9.32358146e-01 7.67777145e-01 -3.76255542e-01
-2.34177917e-01 -3.80390376e-01 -9.57957953e-02 -1.81621337e+00
-7.34018922e-01 -8.27814221e-01 1.84647620e-01 7.08611965e-01
4.95217033e-02 -9.76083279e-01 7.71896005e-01 7.13680014e-02
-5.73499836e-02 -5.66324294e-01 -8.65951240e-01 -1.00811934e+00
1.13267489e-02 -2.33002901e-01 5.50228715e-01 1.23857188e+00
2.19220266e-01 1.73808306e-01 -1.84220582e-01 2.95726061e-01
9.78792191e-01 5.32679781e-02 6.12981796e-01 -1.54252112e+00
-7.55886436e-02 -6.53212786e-01 -9.39131200e-01 -2.87691116e-01
3.46166819e-01 -1.48337018e+00 -1.18358776e-01 -1.58097041e+00
-1.64945319e-01 -1.28696993e-01 -2.68828809e-01 2.09691286e-01
-2.48474609e-02 2.07547650e-01 9.69836190e-02 -5.16071543e-02
-2.85440236e-01 2.77886569e-01 1.31794202e+00 -1.24824271e-01
-2.27505311e-01 -5.49678318e-03 -6.30636096e-01 6.63521707e-01
1.04036534e+00 -4.22896981e-01 -5.49393713e-01 1.90211460e-01
7.84894153e-02 -1.74162954e-01 2.39774808e-01 -1.08825803e+00
1.50581941e-01 -2.33269334e-02 -1.97558403e-02 1.02702983e-01
-9.74727944e-02 -7.46969998e-01 4.14169729e-01 6.78046584e-01
-2.28558138e-01 1.68327868e-01 -7.00286850e-02 7.69159913e-01
-8.94051939e-02 -2.53999412e-01 1.28682697e+00 -1.39842201e-02
-6.14011526e-01 6.35569215e-01 2.15288952e-01 2.78774112e-01
1.24188101e+00 -2.87383586e-01 -2.45017469e-01 -2.95981914e-01
-9.69486237e-01 -1.51946843e-01 7.08007097e-01 1.11812875e-01
5.58115304e-01 -1.59381139e+00 -6.10626101e-01 3.23859125e-01
7.13355392e-02 -3.53421926e-01 -4.41084541e-02 9.08226848e-01
-8.82783771e-01 1.00129478e-01 -2.91642576e-01 -4.06342834e-01
-1.34604204e+00 7.68601477e-01 3.72325182e-01 -2.42439821e-01
-8.32392097e-01 6.36187136e-01 1.21937074e-01 -4.17278469e-01
8.49896967e-02 -5.74979186e-01 4.62642647e-02 -1.64932460e-01
3.85991603e-01 6.79435611e-01 1.25015855e-01 -5.69741666e-01
-3.42220038e-01 8.38437915e-01 2.29184493e-01 1.65777817e-01
1.20798874e+00 -2.11971290e-02 -5.84733188e-01 3.59073848e-01
1.57491219e+00 3.14970791e-01 -8.88085783e-01 -2.78884709e-01
7.13144764e-02 -4.99295115e-01 1.00144975e-01 2.52758395e-02
-1.22245252e+00 5.22852182e-01 4.54906702e-01 1.20242560e+00
9.88022685e-01 1.66640550e-01 5.45121610e-01 2.47138232e-01
3.04075003e-01 -9.50918078e-01 2.14896929e-02 3.63017380e-01
9.46082592e-01 -9.47923958e-01 1.09318413e-01 -5.89385629e-01
-1.66350096e-01 1.28985775e+00 1.21819496e-01 -7.88103402e-01
1.19143736e+00 -1.20899938e-01 -3.16766530e-01 -3.41030449e-01
-2.96596866e-02 -3.55341882e-01 6.21231854e-01 6.77453697e-01
3.92731160e-01 4.30388391e-01 -4.30335969e-01 -7.95191806e-03
-3.68661255e-01 -2.46634766e-01 6.57571852e-01 5.40120482e-01
-6.13371491e-01 -1.10637569e+00 -2.62832403e-01 6.81194365e-01
-8.11209530e-02 -1.25918478e-01 -6.14686966e-01 9.45679426e-01
1.51397079e-01 8.26349854e-01 -1.71063811e-01 -5.74272871e-01
3.04066271e-01 -6.10864498e-02 6.32998884e-01 -7.06048667e-01
7.59946182e-02 -2.12848321e-01 -9.44825709e-02 -6.96553946e-01
-4.84255463e-01 -4.62947518e-01 -1.18545616e+00 -7.77885914e-01
-3.67879599e-01 4.30238247e-01 5.20082057e-01 7.00896204e-01
1.92798868e-01 3.66637170e-01 6.66748166e-01 -7.44148195e-01
-2.85462737e-01 -6.57255232e-01 -1.23928750e+00 4.97705191e-01
3.31088662e-01 -6.40840769e-01 -6.74508274e-01 -1.02525122e-01]
|
[7.134591579437256, 5.50976037979126]
|
22ad219a-c50c-4f40-b9dd-e4a652316733
|
more-robust-schema-guided-dialogue-state
|
2303.09905
| null |
https://arxiv.org/abs/2303.09905v1
|
https://arxiv.org/pdf/2303.09905v1.pdf
|
More Robust Schema-Guided Dialogue State Tracking via Tree-Based Paraphrase Ranking
|
The schema-guided paradigm overcomes scalability issues inherent in building task-oriented dialogue (TOD) agents with static ontologies. Instead of operating on dialogue context alone, agents have access to hierarchical schemas containing task-relevant natural language descriptions. Fine-tuned language models excel at schema-guided dialogue state tracking (DST) but are sensitive to the writing style of the schemas. We explore methods for improving the robustness of DST models. We propose a framework for generating synthetic schemas which uses tree-based ranking to jointly optimise lexical diversity and semantic faithfulness. The generalisation of strong baselines is improved when augmenting their training data with prompts generated by our framework, as demonstrated by marked improvements in average joint goal accuracy (JGA) and schema sensitivity (SS) on the SGD-X benchmark.
|
['B. Byrne', 'W. Lin', 'B. H. Tseng', 'A. Coca']
|
2023-03-17
| null | null | null | null |
['dialogue-state-tracking']
|
['natural-language-processing']
|
[ 2.92423844e-01 8.94119620e-01 -1.60276070e-01 -6.29397392e-01
-9.39507544e-01 -8.39438677e-01 1.42309868e+00 1.64791420e-01
-5.30795753e-01 8.93376529e-01 9.02166963e-01 -8.46624821e-02
-6.03504293e-02 -7.15872765e-01 -2.51907766e-01 -2.49855164e-02
1.83483973e-01 1.07954001e+00 3.69367748e-01 -1.17098773e+00
3.42474878e-01 -2.63462007e-01 -1.27154016e+00 6.30718350e-01
1.13825703e+00 5.33032179e-01 3.37825008e-02 6.61336064e-01
-4.97495234e-01 1.26685226e+00 -8.17816973e-01 -6.88021600e-01
5.57589810e-03 -6.77809834e-01 -1.25283182e+00 1.63180962e-01
4.07996476e-01 -2.57686615e-01 -1.89461291e-01 9.31024432e-01
4.17516708e-01 2.45841235e-01 4.79122162e-01 -1.15245712e+00
-3.56674999e-01 1.03320777e+00 1.82118952e-01 -1.98984444e-02
9.63435948e-01 4.68550533e-01 1.15593600e+00 -2.59545267e-01
1.13825929e+00 1.71897089e+00 6.28317833e-01 1.07940149e+00
-1.58683407e+00 -1.31858900e-01 2.12788507e-01 -2.76123613e-01
-7.49184132e-01 -9.24049079e-01 5.89999855e-01 -2.27219656e-01
1.25591767e+00 2.91338652e-01 3.80399317e-01 1.72851539e+00
-1.43710554e-01 8.17838371e-01 1.36609054e+00 -4.54234332e-01
3.03455859e-01 5.14711797e-01 4.51163948e-02 7.17968822e-01
1.22338943e-01 8.46417993e-02 -8.25784087e-01 -2.93813348e-01
5.35569429e-01 -1.01501620e+00 8.25412851e-03 -7.41860390e-01
-1.42770362e+00 1.11481202e+00 -6.65422380e-02 1.26457885e-02
-1.82629034e-01 -3.38849008e-01 9.63085830e-01 7.20953703e-01
3.45636725e-01 1.45502067e+00 -5.63896596e-01 -3.70513350e-01
-4.12210375e-01 9.28276181e-01 1.19114089e+00 1.32881153e+00
2.26738349e-01 1.19853973e-01 -5.12663424e-01 1.17763853e+00
5.45360036e-02 3.57239902e-01 6.84722424e-01 -1.47114158e+00
7.32312560e-01 8.57665658e-01 3.42340589e-01 -4.52809453e-01
-7.10211754e-01 -8.74184519e-02 -3.48050386e-01 -1.31872341e-01
6.66280985e-01 -2.97675103e-01 -5.09335816e-01 2.07060981e+00
3.99252355e-01 -9.26501870e-01 8.95103157e-01 6.39704287e-01
8.65779757e-01 4.35452610e-01 6.16569877e-01 -1.34066433e-01
1.43086278e+00 -1.00581694e+00 -6.79635763e-01 -5.57056427e-01
1.03259015e+00 -2.74648070e-01 1.42948556e+00 2.21377864e-01
-1.36531913e+00 -2.22446516e-01 -8.70273471e-01 -2.47321159e-01
-3.58776957e-01 -4.75088298e-01 6.02164268e-01 6.78668499e-01
-1.14613700e+00 2.23620579e-01 -4.37400818e-01 -9.02924061e-01
-1.95191521e-02 1.42199965e-02 -2.26132840e-01 3.79154027e-01
-1.61053538e+00 1.27228236e+00 8.96821022e-01 -7.20663011e-01
-8.66680503e-01 -4.87812400e-01 -1.16137302e+00 -2.74156123e-01
7.89467692e-01 -1.01205945e+00 1.88929021e+00 -8.08030844e-01
-2.04838490e+00 9.84456241e-01 2.97974616e-01 -7.24814177e-01
7.19630837e-01 5.81468157e-02 -1.65307820e-01 -9.15728137e-03
3.49719018e-01 8.88170779e-01 3.04711789e-01 -1.05589581e+00
-7.61220694e-01 -9.69114974e-02 6.50268137e-01 8.13430429e-01
-1.30371720e-01 1.07103765e-01 -2.85591055e-02 -4.22922313e-01
-7.84772187e-02 -8.18369806e-01 -4.37593400e-01 -5.12094259e-01
-5.25581241e-01 -4.42707956e-01 2.40197152e-01 -4.31775302e-01
1.07950735e+00 -1.62697279e+00 3.30238998e-01 -1.98618591e-01
-1.93294704e-01 2.29569837e-01 -2.94006079e-01 6.94976330e-01
3.18467736e-01 6.17233030e-02 -1.10342771e-01 -7.33780414e-02
4.76995796e-01 1.80482075e-01 -3.02065313e-01 -3.69406730e-01
2.07709014e-01 8.45260084e-01 -1.11698639e+00 -5.91080010e-01
9.25617591e-02 -1.54117182e-01 -7.39927292e-01 4.57073122e-01
-1.03136635e+00 3.20173413e-01 -5.47065675e-01 1.51308611e-01
-7.76673555e-02 -2.20936194e-01 6.40443385e-01 1.84703693e-01
5.45539670e-02 1.25501883e+00 -8.44591916e-01 2.20854950e+00
-4.75984007e-01 1.47513106e-01 -9.21063870e-02 -4.49560672e-01
9.22299385e-01 4.13137287e-01 2.39997208e-02 -1.13894475e+00
-2.39320576e-01 -1.37836365e-02 -1.86436191e-01 -6.14254415e-01
7.13082671e-01 -2.24672616e-01 -6.07060075e-01 7.86480784e-01
2.17024505e-01 -5.39413571e-01 4.48648870e-01 6.94773793e-01
8.37257385e-01 3.17568392e-01 5.18633008e-01 -6.72134697e-01
4.57708597e-01 7.32548058e-01 1.59798622e-01 9.23345447e-01
-1.90610588e-02 -1.50683410e-02 8.13098252e-01 -4.13045377e-01
-1.47670519e+00 -8.03402424e-01 -1.14674442e-01 1.60011256e+00
-3.08348089e-02 -8.00757706e-01 -9.86884773e-01 -1.01021600e+00
-1.97127089e-01 1.46452439e+00 -4.51051474e-01 -2.61688888e-01
-5.04465520e-01 -6.01383746e-01 1.03884542e+00 2.10606173e-01
6.35910571e-01 -1.03205478e+00 -8.34932625e-01 4.96082872e-01
-5.06887615e-01 -1.25308800e+00 -2.23323315e-01 6.06392659e-02
-8.42818737e-01 -8.76406014e-01 -2.24739313e-01 -3.10257614e-01
5.38675897e-02 -1.49007291e-01 1.55608261e+00 -2.07111284e-01
3.76301438e-01 4.52239245e-01 -3.28876674e-01 -3.81372660e-01
-1.28672945e+00 4.34137195e-01 1.26350177e-02 -6.84352577e-01
1.94647104e-01 -1.12619728e-01 -2.07036063e-01 3.29175055e-01
-6.46107912e-01 5.99698305e-01 8.67973566e-02 9.80234206e-01
-2.50693083e-01 -3.83625716e-01 6.76251233e-01 -1.33695912e+00
1.23752403e+00 -4.10002321e-01 -4.16200817e-01 3.48748296e-01
-8.23439717e-01 5.35869122e-01 4.71843630e-01 -5.96820712e-02
-1.67772937e+00 -3.26039255e-01 -4.50769104e-02 3.57241929e-01
-3.38543832e-01 2.60446072e-01 -1.24168001e-01 4.15513813e-01
1.29612875e+00 1.65292770e-01 2.88958073e-01 -2.68659472e-01
6.46903276e-01 5.94637752e-01 5.60725212e-01 -1.16984236e+00
4.37187523e-01 -6.38007075e-02 -5.23967981e-01 -5.80859244e-01
-9.71328735e-01 -1.13936342e-01 -3.44674170e-01 -7.42462650e-03
6.51420891e-01 -9.63499129e-01 -4.98879224e-01 1.99494034e-01
-9.67449844e-01 -8.49987984e-01 -3.77119571e-01 -9.54754278e-02
-1.07553983e+00 2.40531430e-01 -6.31034255e-01 -7.85542190e-01
-3.57588619e-01 -9.53390896e-01 1.04159307e+00 -3.39220613e-02
-8.61550510e-01 -1.26365721e+00 2.45021269e-01 5.16794860e-01
5.17067432e-01 1.75339997e-01 1.18770838e+00 -1.33402884e+00
-2.99669921e-01 2.68987328e-01 -1.29565166e-03 -9.19199660e-02
-1.05532750e-01 -4.09466445e-01 -8.54760706e-01 -2.12714776e-01
-4.42674905e-02 -1.04447258e+00 3.46533149e-01 -1.64316729e-01
1.77689627e-01 -8.19869995e-01 -9.38836560e-02 2.28425846e-01
1.07387567e+00 1.71574041e-01 2.52882332e-01 9.74713743e-01
2.93161541e-01 1.16125250e+00 8.51335645e-01 4.11979586e-01
9.01978076e-01 1.09800529e+00 5.01128379e-03 2.97881007e-01
-1.39501497e-01 -5.89567125e-01 4.47748780e-01 1.88519195e-01
3.24565232e-01 -3.12866092e-01 -1.08811879e+00 5.24477005e-01
-1.84732282e+00 -9.55026746e-01 3.27421241e-02 1.91136098e+00
1.41493571e+00 3.38811696e-01 5.49242795e-01 -2.85632759e-01
4.16642070e-01 4.07470763e-01 -5.36785007e-01 -7.14808583e-01
-2.65099674e-01 -2.78061062e-01 2.00512335e-01 7.17128456e-01
-8.71340096e-01 1.41333437e+00 7.13723230e+00 3.26357782e-01
-5.41681588e-01 6.92868531e-02 3.02441299e-01 -6.92759305e-02
-5.36119998e-01 -7.61007667e-02 -7.21879840e-01 2.43059531e-01
1.22676027e+00 -4.76586878e-01 5.39422274e-01 8.13867450e-01
-3.17274369e-02 -2.23988798e-02 -1.28743827e+00 2.37404197e-01
-7.26841986e-02 -1.22793138e+00 2.96355963e-01 -2.04572856e-01
7.04065859e-01 -7.69822896e-02 -8.65788311e-02 6.79001033e-01
1.31223559e+00 -7.00049520e-01 8.55542839e-01 1.64713889e-01
7.02126861e-01 -4.34724122e-01 3.99266779e-01 5.19400418e-01
-3.82014543e-01 -4.01039496e-02 -1.82272539e-01 2.04519928e-02
9.73349344e-03 -3.59948128e-01 -1.36212301e+00 3.96345228e-01
2.80914724e-01 2.21382514e-01 -6.71122491e-01 3.29038531e-01
-2.51269042e-01 2.90784150e-01 -2.46408179e-01 -1.98204502e-01
5.05973995e-01 -1.31382083e-04 8.66239846e-01 1.40988231e+00
-3.61406565e-01 3.01255852e-01 1.95015252e-01 9.18651462e-01
1.39684275e-01 1.41560450e-01 -8.08568418e-01 -1.27742663e-01
6.97219014e-01 8.39944601e-01 -2.41373032e-01 -6.27707362e-01
-2.31311187e-01 6.97389901e-01 3.56774628e-01 2.27132678e-01
-3.25220883e-01 1.51056759e-02 6.48735523e-01 -1.15254126e-01
-2.10198328e-01 -1.68713294e-02 -4.13045824e-01 -1.25737226e+00
-4.12835896e-01 -1.62030995e+00 8.26680422e-01 -7.57151902e-01
-1.15726531e+00 6.45954132e-01 3.02517742e-01 -6.85635448e-01
-8.74850571e-01 -3.13328207e-01 -2.71627158e-01 6.51156723e-01
-1.10358667e+00 -1.10689735e+00 5.69397770e-03 4.93523210e-01
1.15284598e+00 -4.57799017e-01 1.13568223e+00 -4.69475627e-01
-4.29084301e-01 6.41455233e-01 -2.36294910e-01 5.53868860e-02
8.43284786e-01 -1.75963330e+00 1.07689238e+00 5.25451005e-01
-6.77996054e-02 7.37766743e-01 1.23662269e+00 -7.38526344e-01
-1.01015174e+00 -8.92085016e-01 1.03424776e+00 -1.01018488e+00
7.20537305e-01 -4.15776223e-01 -8.71442080e-01 8.63855302e-01
5.97393751e-01 -9.41591203e-01 4.46094155e-01 4.27339643e-01
-5.36809087e-01 3.59665662e-01 -1.26074350e+00 7.81057656e-01
1.29530573e+00 -6.69041991e-01 -1.21997976e+00 5.66340744e-01
8.80120456e-01 -7.96916306e-01 -9.66885269e-01 1.67852268e-01
2.83614695e-01 -1.13765311e+00 9.01901603e-01 -1.24173522e+00
4.38840687e-01 8.38651881e-02 -7.61262551e-02 -1.48313117e+00
-1.46899134e-01 -8.60117018e-01 2.05686972e-01 1.22247732e+00
5.46280742e-01 -5.03303766e-01 5.39213300e-01 1.12294447e+00
-1.50123790e-01 -2.15175346e-01 -5.36747515e-01 -6.52037561e-01
1.04914203e-01 -2.66529359e-02 8.60639632e-01 1.15029764e+00
6.68429136e-01 9.31496203e-01 -9.78382602e-02 -1.59496725e-01
6.18567586e-01 -3.12857091e-01 9.26357269e-01 -1.13926375e+00
-1.05211467e-01 -6.28591955e-01 1.56423941e-01 -8.08792472e-01
2.30958760e-01 -7.24156797e-01 -3.96555103e-03 -1.50431502e+00
-2.60189194e-02 -5.96765220e-01 9.55679268e-02 4.23168361e-01
-2.43582278e-01 -2.92254657e-01 2.02822804e-01 1.21245779e-01
-1.05592442e+00 4.50590223e-01 1.10506713e+00 2.69818716e-02
-4.07680184e-01 -3.06035727e-01 -1.00734091e+00 6.07319832e-01
8.37820292e-01 -3.11202735e-01 -9.36809063e-01 -4.53270465e-01
3.61751288e-01 5.04301786e-01 1.69148549e-01 -8.90928864e-01
3.32575917e-01 -4.01569009e-01 -2.49119624e-01 1.40988693e-01
2.44024783e-01 -2.05212250e-01 3.80903925e-03 2.32573137e-01
-1.31314468e+00 7.21488744e-02 2.63680339e-01 3.82453680e-01
1.32793501e-01 -2.15464294e-01 7.79619634e-01 -5.58890462e-01
-7.63947010e-01 -4.34449553e-01 -5.53695381e-01 7.42929995e-01
7.38056898e-01 -1.86913535e-01 -7.35370755e-01 -5.61983705e-01
-4.96849328e-01 4.88077313e-01 8.47774386e-01 6.13824666e-01
1.14656888e-01 -1.14028573e+00 -7.73885250e-01 3.04917190e-02
6.60048008e-01 -1.52761757e-03 -1.08329423e-01 2.87555486e-01
-2.37981543e-01 8.84576917e-01 -5.51591992e-01 -3.94691885e-01
-8.81470442e-01 4.34987783e-01 6.28877044e-01 -3.63619864e-01
-5.03685296e-01 9.04980421e-01 1.77600086e-01 -9.44154143e-01
2.59431541e-01 7.39479288e-02 -4.16400641e-01 7.33614638e-02
3.60821307e-01 1.94092333e-01 -4.35197875e-02 -4.27674174e-01
2.98886597e-02 -1.38663694e-01 -4.88251090e-01 -6.28659844e-01
1.01095057e+00 -2.93244600e-01 1.70621678e-01 3.88562649e-01
4.05134380e-01 -2.14575186e-01 -1.32055116e+00 -5.71669340e-01
5.97925723e-01 -1.55530602e-01 -1.45032138e-01 -1.50132811e+00
-1.52822465e-01 3.01566929e-01 9.71326008e-02 6.32341921e-01
4.46934402e-01 -5.37272654e-02 5.64688265e-01 8.26590359e-01
7.12729156e-01 -1.32097983e+00 2.27688715e-01 7.45013773e-01
9.08435702e-01 -1.19651961e+00 -1.58628121e-01 -1.09092250e-01
-1.45111871e+00 9.50836003e-01 9.83049631e-01 3.46751899e-01
-4.81956720e-01 1.75729562e-02 4.37064767e-01 -4.13237423e-01
-1.42459536e+00 -2.55113512e-01 -1.55148245e-02 7.39301205e-01
4.33815747e-01 -1.19017668e-01 -2.20787391e-01 3.25613141e-01
-6.66247189e-01 -3.94103825e-01 5.25910616e-01 8.67493629e-01
-5.00312150e-01 -1.24562526e+00 -5.00793122e-02 2.59374529e-01
-1.41623423e-01 -5.17537780e-02 -9.03736115e-01 9.14506197e-01
-5.89330375e-01 1.15328360e+00 -1.14307240e-01 -2.01532811e-01
7.45502412e-01 6.45414889e-01 3.53693098e-01 -8.00103605e-01
-9.07108188e-01 -2.77694285e-01 1.25686347e+00 -7.57963896e-01
-2.77905822e-01 -8.89125943e-01 -9.58114982e-01 -3.20548058e-01
-2.70944405e-02 4.50688064e-01 4.11417902e-01 8.20871055e-01
2.21332252e-01 2.07278579e-01 3.29382747e-01 -2.28383362e-01
-1.01494586e+00 -1.00790524e+00 3.23313735e-02 7.38521278e-01
2.00746864e-01 -6.17198467e-01 1.32395506e-01 -1.36488661e-01]
|
[12.770879745483398, 7.975526332855225]
|
17949131-1012-443f-86e7-2108bc42c6f4
|
targeted-adversarial-attacks-against-neural-1
|
2303.01068
| null |
https://arxiv.org/abs/2303.01068v1
|
https://arxiv.org/pdf/2303.01068v1.pdf
|
Targeted Adversarial Attacks against Neural Machine Translation
|
Neural Machine Translation (NMT) systems are used in various applications. However, it has been shown that they are vulnerable to very small perturbations of their inputs, known as adversarial attacks. In this paper, we propose a new targeted adversarial attack against NMT models. In particular, our goal is to insert a predefined target keyword into the translation of the adversarial sentence while maintaining similarity between the original sentence and the perturbed one in the source domain. To this aim, we propose an optimization problem, including an adversarial loss term and a similarity term. We use gradient projection in the embedding space to craft an adversarial sentence. Experimental results show that our attack outperforms Seq2Sick, the other targeted adversarial attack against NMT models, in terms of success rate and decrease in translation quality. Our attack succeeds in inserting a keyword into the translation for more than 75% of sentences while similarity with the original sentence stays preserved.
|
['Pascal Frossard', 'Ljiljana Dolamic', 'AmirHossein Dabiri Aghdam', 'Sahar Sadrizadeh']
|
2023-03-02
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 6.59215927e-01 7.28831589e-02 2.46583432e-01 -1.16569676e-01
-9.75098550e-01 -1.19264555e+00 7.32977450e-01 -1.27043784e-01
-4.08307374e-01 8.38806093e-01 -2.06520576e-02 -5.17767966e-01
4.66782123e-01 -6.83727264e-01 -1.22641540e+00 -7.74754941e-01
3.49293530e-01 3.37951422e-01 2.89067142e-02 -5.14174879e-01
1.31685495e-01 4.96526510e-01 -4.21200991e-01 1.82876155e-01
9.62241113e-01 4.07214254e-01 -3.51268798e-02 6.00806475e-01
3.55207175e-02 2.81751215e-01 -9.82031405e-01 -1.07099342e+00
8.04397821e-01 -6.36629939e-01 -6.78317070e-01 -2.84459531e-01
5.06721497e-01 2.77193896e-02 -5.43469548e-01 1.54792678e+00
7.32426047e-01 -1.58497930e-01 5.36892772e-01 -1.21578097e+00
-8.85282755e-01 7.23506927e-01 -3.23182285e-01 1.19164236e-01
2.62731135e-01 1.96764767e-01 7.23269343e-01 -9.96365368e-01
5.24049044e-01 1.27045465e+00 4.83862579e-01 9.02077377e-01
-1.25237715e+00 -8.04120660e-01 -2.98435800e-02 -7.33807590e-03
-1.13053811e+00 -4.37575489e-01 7.39291668e-01 -1.68795481e-01
7.81633794e-01 5.97161293e-01 1.24430984e-01 1.42642450e+00
8.72430265e-01 4.90060866e-01 1.16396201e+00 -5.83738148e-01
1.46115035e-01 3.61615956e-01 -5.12959480e-01 4.57995564e-01
-5.12316823e-02 1.05467260e-01 -2.46944129e-01 -2.89772063e-01
3.23132604e-01 -3.18892837e-01 -2.53467828e-01 -1.05505042e-01
-1.27705228e+00 8.14343989e-01 3.00224423e-01 3.83040816e-01
-2.49995202e-01 8.53557512e-02 4.72094983e-01 9.48659658e-01
5.85779011e-01 6.74822807e-01 -4.17538583e-01 3.19919025e-04
-5.25676370e-01 2.08901897e-01 6.51526093e-01 8.39702427e-01
4.28959548e-01 7.07944110e-02 -3.04214150e-01 7.14657605e-01
-9.82757807e-02 1.02689338e+00 6.84902430e-01 -4.75413203e-01
1.12317467e+00 3.51807714e-01 -8.10051709e-03 -1.10816431e+00
7.39642009e-02 -4.66329515e-01 -8.59170318e-01 2.53933370e-01
2.05170512e-01 -5.18014610e-01 -6.89364672e-01 1.97974527e+00
2.48163834e-01 1.38203561e-01 4.58961695e-01 8.29389989e-01
1.46986634e-01 9.44712639e-01 -3.82647127e-01 -3.21451485e-01
9.23141003e-01 -9.53350365e-01 -7.25774229e-01 -6.21452868e-01
5.06663620e-01 -1.30257905e+00 1.06993079e+00 -7.40560368e-02
-1.16992819e+00 -2.00916275e-01 -1.19662225e+00 2.72553414e-01
-3.05588633e-01 -2.23078385e-01 -3.75001431e-02 7.89130509e-01
-6.46851540e-01 6.25125051e-01 -6.18062139e-01 -1.83945343e-01
1.90342456e-01 5.74229777e-01 -6.45444930e-01 -7.83449411e-02
-1.60248470e+00 1.16343558e+00 4.53406014e-02 1.47572113e-03
-8.41298819e-01 -4.83370602e-01 -6.67519212e-01 -1.22697979e-01
1.16785787e-01 -7.11366177e-01 1.24159503e+00 -1.24079573e+00
-1.74124491e+00 7.09319890e-01 -1.90430768e-02 -7.41409123e-01
8.90359461e-01 -2.05382735e-01 -4.54040051e-01 -1.16390385e-01
1.16240174e-01 1.16546273e-01 1.21296263e+00 -9.36453164e-01
-1.86681733e-01 -4.07249898e-01 1.32283941e-01 2.44762316e-01
-6.34699643e-01 2.77725548e-01 -1.07439749e-01 -1.09060574e+00
-1.82256967e-01 -1.26230669e+00 -2.27600262e-01 -2.70312250e-01
-6.85957491e-01 2.37338513e-01 8.86050701e-01 -6.85291231e-01
1.00265980e+00 -2.08646369e+00 5.91329753e-01 9.10904706e-02
-7.91498497e-02 6.76093936e-01 -3.85327816e-01 7.44165063e-01
-1.09961085e-01 2.88578063e-01 -4.49766427e-01 -3.11140537e-01
3.52298096e-02 2.45915279e-01 -7.35442340e-01 5.34427762e-01
3.34507436e-01 1.05635762e+00 -8.79322231e-01 -1.89719722e-01
-1.93482921e-01 3.10954809e-01 -2.48603016e-01 3.02888155e-01
-8.53486285e-02 3.77608389e-01 -4.32012916e-01 1.74651280e-01
7.16548264e-01 5.50592303e-01 2.49865074e-02 1.24751419e-01
2.82089859e-01 2.41792306e-01 -6.04141891e-01 1.43036401e+00
-5.46717763e-01 6.94892466e-01 -1.82692781e-01 -9.16093409e-01
1.01128054e+00 5.72254062e-01 -1.54271796e-02 -4.62397426e-01
2.26376995e-01 4.40734059e-01 2.08695650e-01 -3.44901204e-01
1.89769000e-01 -4.05162513e-01 -3.17508578e-01 5.48003137e-01
-1.18964709e-01 -2.92467326e-01 -1.04799837e-01 1.19782433e-01
1.15334404e+00 -2.67727703e-01 -2.05393489e-02 3.41632552e-02
6.68893278e-01 -1.55878201e-01 5.54358959e-01 5.73815227e-01
-1.01138882e-01 5.54315388e-01 5.40496826e-01 -1.90347359e-01
-1.40909457e+00 -9.75395262e-01 3.14343482e-01 6.34040892e-01
3.03331204e-02 -1.37410328e-01 -1.14044249e+00 -1.16715634e+00
-1.11319870e-01 7.28162646e-01 -5.02983809e-01 -8.35684121e-01
-1.08584177e+00 -5.36754012e-01 1.00144684e+00 2.04113513e-01
6.64352357e-01 -8.90001237e-01 9.21843499e-02 1.33643225e-01
-3.69441807e-01 -1.21426094e+00 -1.17593014e+00 1.21432535e-01
-8.17105293e-01 -6.19347155e-01 -7.24597573e-01 -9.03235614e-01
8.04595470e-01 1.42329514e-01 7.29401588e-01 -3.85171294e-01
1.98810503e-01 -2.47049138e-01 -2.20227882e-01 -3.79583538e-01
-1.22727203e+00 1.66469142e-01 4.68182504e-01 2.50247806e-01
1.65657744e-01 -6.18274629e-01 -6.59523979e-02 4.77144003e-01
-1.15069592e+00 -1.79376572e-01 6.57233775e-01 8.17400575e-01
2.79114276e-01 1.62752077e-01 6.74535275e-01 -8.94381523e-01
1.02469933e+00 -3.54249179e-01 -5.48450232e-01 2.96464056e-01
-4.07720625e-01 1.94361702e-01 1.33486640e+00 -8.85554612e-01
-5.36285102e-01 -1.28797308e-01 -1.94328427e-01 -6.65443301e-01
3.86375636e-01 3.80137205e-01 -5.63031793e-01 -3.34429860e-01
8.71647835e-01 5.62245190e-01 4.35646810e-02 -4.19426471e-01
2.12216616e-01 6.89001262e-01 5.59602678e-01 -4.86445963e-01
1.71179903e+00 -2.58269701e-02 -1.45768030e-02 -2.50628054e-01
-4.71556455e-01 1.49166584e-01 -4.73550081e-01 7.14508817e-02
3.90200913e-01 -4.78183776e-01 -1.54631853e-01 6.07548654e-01
-1.41468310e+00 1.21623138e-02 5.76939024e-02 3.06745350e-01
-6.06002510e-01 3.90812576e-01 -3.80540818e-01 -3.65079910e-01
-7.12164521e-01 -1.39698422e+00 7.03199923e-01 -6.49666041e-02
-2.38007102e-02 -8.59102249e-01 2.17990458e-01 3.56005132e-01
4.46825624e-01 4.21316743e-01 1.00603402e+00 -9.53675449e-01
-3.16943556e-01 -7.47338593e-01 2.27980509e-01 8.29876900e-01
3.73511791e-01 -1.73313379e-01 -5.17028511e-01 -5.63926816e-01
4.28554833e-01 -2.31675096e-02 3.40359271e-01 -2.81259656e-01
5.61786413e-01 -8.04214537e-01 -9.73612592e-02 5.51368475e-01
1.24474669e+00 2.90971756e-01 6.07454419e-01 4.34219271e-01
5.54753661e-01 3.04143757e-01 6.42895341e-01 -2.37646133e-01
-4.90917176e-01 9.38085973e-01 4.02698547e-01 1.54597908e-01
1.98863178e-01 -3.01955223e-01 9.26749110e-01 9.04391766e-01
2.92758614e-01 -4.69134152e-01 -7.43244052e-01 3.35074931e-01
-1.82289612e+00 -9.04096842e-01 2.37467080e-01 2.36995363e+00
1.01126814e+00 4.97572660e-01 -1.73371166e-01 1.61033794e-01
8.81588101e-01 7.80589134e-02 -6.52283967e-01 -9.38269675e-01
-2.95204014e-01 1.79655522e-01 6.70322537e-01 6.67162299e-01
-1.03515947e+00 1.23798919e+00 5.64297056e+00 8.04059565e-01
-1.29283297e+00 1.97181404e-01 2.44137317e-01 -1.35492310e-01
-1.91707090e-01 -1.81063384e-01 -4.71295267e-01 6.64808333e-01
1.09800541e+00 -7.15951622e-01 6.11185849e-01 4.34144020e-01
3.22283596e-01 7.57965088e-01 -1.00667131e+00 5.02492785e-01
1.55582443e-01 -1.12274647e+00 2.77533293e-01 1.20575465e-02
7.64484763e-01 -1.90369964e-01 3.75486434e-01 2.66392738e-01
2.71100342e-01 -1.00495052e+00 6.27417386e-01 1.39639974e-01
7.21800327e-01 -1.11779654e+00 9.14360166e-01 5.05612195e-01
-5.28712749e-01 1.85177386e-01 -4.25182402e-01 7.13065714e-02
-8.11069086e-02 3.03205758e-01 -9.62918580e-01 6.10926330e-01
1.42871261e-01 1.83387190e-01 -2.86299586e-01 5.54065466e-01
-4.76971567e-01 7.09101975e-01 -1.43320277e-01 -2.20354497e-02
4.98176545e-01 -2.18355909e-01 1.09924281e+00 1.02942777e+00
2.47981966e-01 -2.78526425e-01 -3.50117534e-02 5.55011332e-01
-5.73597789e-01 4.06098694e-01 -1.06853700e+00 -1.97739601e-01
5.05009055e-01 7.39561737e-01 -5.40767983e-02 -7.64010772e-02
-9.33608487e-02 1.72902775e+00 2.89674908e-01 2.87549376e-01
-1.10637438e+00 -7.09413886e-01 1.05164039e+00 -2.30001181e-01
1.45547152e-01 -6.27612099e-02 -1.85608834e-01 -1.25512362e+00
5.97026885e-01 -1.49887586e+00 -2.63353020e-01 -4.85075951e-01
-1.14806283e+00 1.03298140e+00 -4.38296080e-01 -1.34779668e+00
-2.44273797e-01 -3.34501863e-01 -6.51595056e-01 1.29468453e+00
-9.27548766e-01 -1.17718494e+00 4.14338440e-01 6.50384367e-01
6.60268843e-01 -5.74070334e-01 1.00850224e+00 1.97313920e-01
-6.93728387e-01 1.14371097e+00 5.59428275e-01 4.48133975e-01
9.31116879e-01 -1.08324730e+00 1.02436388e+00 1.23751092e+00
2.73390144e-01 7.53292620e-01 1.20745659e+00 -6.67267025e-01
-1.47405338e+00 -1.46714807e+00 1.21657515e+00 -4.61347491e-01
7.33984530e-01 -5.92627227e-01 -8.13753128e-01 7.93668568e-01
3.70789409e-01 -1.94939122e-01 4.25456345e-01 -5.33265293e-01
-6.01544619e-01 -2.65974142e-02 -1.44112599e+00 1.01268053e+00
7.30052710e-01 -7.03523993e-01 -7.12231040e-01 7.52949178e-01
1.16672850e+00 -4.98286873e-01 -7.21128106e-01 2.86395967e-01
3.78137767e-01 -3.22121352e-01 1.02174175e+00 -1.08095407e+00
5.44756770e-01 -2.91044176e-01 -3.25582892e-01 -1.77400434e+00
-1.06541120e-01 -1.13575327e+00 -5.47313318e-03 1.08251917e+00
6.94962025e-01 -8.49299788e-01 7.20332503e-01 2.73527056e-01
-4.09758314e-02 -7.74658501e-01 -1.21465695e+00 -1.19222081e+00
7.52575457e-01 -6.78964406e-02 5.84880948e-01 8.72049630e-01
-1.14727981e-01 5.30772865e-01 -8.09304297e-01 4.91763383e-01
4.89012271e-01 -2.10845187e-01 8.84884655e-01 -5.57581544e-01
-3.02044183e-01 -2.19418392e-01 -5.12806058e-01 -7.27145314e-01
3.73767823e-01 -9.41369951e-01 1.54371774e-02 -1.07018495e+00
-9.69110653e-02 -1.23871658e-02 -2.16547787e-01 3.25800121e-01
-4.62398678e-01 4.78276044e-01 3.68053883e-01 1.68515429e-01
1.74639851e-01 5.90490103e-01 1.12852395e+00 -5.12640715e-01
-5.73057905e-02 4.76852030e-01 -5.02243757e-01 4.37879473e-01
1.19806373e+00 -9.79861021e-01 -2.94276685e-01 -6.68934524e-01
1.78822741e-01 -1.31040379e-01 1.52615728e-02 -6.81794643e-01
5.37005104e-02 -2.33844742e-01 -7.28599206e-02 -1.74430043e-01
3.39027315e-01 -9.26088333e-01 9.66966152e-02 5.60863793e-01
-4.42533314e-01 3.98181260e-01 3.42583060e-01 4.18879032e-01
-9.37371850e-02 -4.27095711e-01 9.90590274e-01 1.77920789e-01
1.45025209e-01 1.55653045e-01 -2.96410859e-01 -9.89967436e-02
1.07175803e+00 8.31870511e-02 -2.86108613e-01 -3.14070493e-01
-5.24215519e-01 1.09690763e-02 5.82264781e-01 7.61759162e-01
5.65775692e-01 -1.41352129e+00 -1.11781633e+00 1.63741156e-01
-6.58229217e-02 -5.17592490e-01 -2.36919060e-01 5.76176167e-01
-4.65413064e-01 4.25014108e-01 -2.15927556e-01 -5.94419204e-02
-1.67539704e+00 8.72017562e-01 5.13169706e-01 -3.05202335e-01
-4.07759935e-01 7.89289892e-01 -2.98163779e-02 -7.44632423e-01
4.50684279e-02 4.41254079e-02 3.04202020e-01 -4.72017109e-01
4.17662978e-01 4.40851077e-02 1.81245133e-01 -7.08165526e-01
-2.38767609e-01 5.23610473e-01 -4.41276759e-01 -1.66580886e-01
9.84417021e-01 -7.49343587e-03 -2.21634150e-01 1.73139721e-01
1.57355821e+00 4.65827674e-01 -7.22340822e-01 -5.37697375e-01
-1.94136217e-01 -6.71597600e-01 -3.40735197e-01 -8.33607554e-01
-8.33433807e-01 7.57557809e-01 5.31432271e-01 2.08752617e-01
1.08658159e+00 -4.30590093e-01 1.23052454e+00 5.77620804e-01
3.28612596e-01 -5.83406866e-01 -1.11353830e-01 5.19131780e-01
1.06092215e+00 -1.03511238e+00 -4.93034035e-01 -3.32099825e-01
-5.26773393e-01 8.72130692e-01 4.16520685e-01 -2.56800503e-01
8.95782784e-02 8.00470710e-02 5.20850480e-01 4.58483338e-01
-6.57675385e-01 5.21016836e-01 1.56887069e-01 5.31376123e-01
3.88926379e-02 3.40736322e-02 -4.45594698e-01 1.27107248e-01
-4.73967135e-01 -4.93439466e-01 5.28776467e-01 7.10895658e-01
-2.06640735e-01 -1.63464820e+00 -6.23284698e-01 -6.99093640e-02
-9.59276319e-01 -3.52377594e-01 -9.20556486e-01 4.14407462e-01
-8.60616565e-02 9.61174369e-01 -4.89590108e-01 -8.36179376e-01
5.65328300e-01 2.88619339e-01 3.51533890e-01 -5.50026953e-01
-1.00384784e+00 -2.81793684e-01 -5.38052293e-03 -3.47046703e-01
1.34556100e-01 -6.00025356e-01 -7.37865031e-01 -5.05888283e-01
-2.12499812e-01 4.66081828e-01 7.62644053e-01 9.43669975e-01
3.12694311e-01 3.27915341e-01 1.33673799e+00 -3.75932068e-01
-1.19779861e+00 -1.14610159e+00 -4.19534072e-02 3.88443232e-01
4.60331976e-01 -1.29420176e-01 -6.00883663e-01 -2.44145170e-02]
|
[6.052786827087402, 8.17674732208252]
|
c85551bd-e3d3-48e3-9464-eb1b61f35e22
|
second-order-democratic-aggregation
|
1808.07503
| null |
http://arxiv.org/abs/1808.07503v1
|
http://arxiv.org/pdf/1808.07503v1.pdf
|
Second-order Democratic Aggregation
|
Aggregated second-order features extracted from deep convolutional networks
have been shown to be effective for texture generation, fine-grained
recognition, material classification, and scene understanding. In this paper,
we study a class of orderless aggregation functions designed to minimize
interference or equalize contributions in the context of second-order features
and we show that they can be computed just as efficiently as their first-order
counterparts and they have favorable properties over aggregation by summation.
Another line of work has shown that matrix power normalization after
aggregation can significantly improve the generalization of second-order
representations. We show that matrix power normalization implicitly equalizes
contributions during aggregation thus establishing a connection between matrix
normalization techniques and prior work on minimizing interference. Based on
the analysis we present {\gamma}-democratic aggregators that interpolate
between sum ({\gamma}=1) and democratic pooling ({\gamma}=0) outperforming both
on several classification tasks. Moreover, unlike power normalization, the
{\gamma}-democratic aggregations can be computed in a low dimensional space by
sketching that allows the use of very high-dimensional second-order features.
This results in a state-of-the-art performance on several datasets.
|
['Tsung-Yu Lin', 'Subhransu Maji', 'Piotr Koniusz']
|
2018-08-22
|
second-order-democratic-aggregation-1
|
http://openaccess.thecvf.com/content_ECCV_2018/html/Tsung-Yu_Lin_Second-order_Democratic_Aggregation_ECCV_2018_paper.html
|
http://openaccess.thecvf.com/content_ECCV_2018/papers/Tsung-Yu_Lin_Second-order_Democratic_Aggregation_ECCV_2018_paper.pdf
|
eccv-2018-9
|
['material-classification']
|
['computer-vision']
|
[ 5.65498710e-01 1.52254596e-01 3.65418792e-02 -3.56787562e-01
-6.21431708e-01 -3.47178042e-01 1.05083585e+00 4.00606096e-01
-5.94298661e-01 6.42407060e-01 2.85095930e-01 -1.80261835e-01
-5.61125457e-01 -1.09242857e+00 -6.91330135e-01 -9.25399899e-01
-2.94060171e-01 8.80095735e-02 1.33374259e-01 -3.83483857e-01
2.81160891e-01 9.35889244e-01 -1.91489494e+00 7.99328208e-01
5.10394573e-01 1.67097878e+00 -2.45865673e-01 6.86284304e-01
-1.89022303e-01 6.05864465e-01 -6.51996136e-01 -3.61818939e-01
4.67457294e-01 -7.12959319e-02 -7.93798327e-01 -1.81234673e-01
1.10034633e+00 -1.47644162e-01 -2.75727570e-01 6.45300329e-01
3.07864964e-01 2.11801022e-01 9.54225779e-01 -1.08807552e+00
-6.97897196e-01 4.88205761e-01 -3.06332111e-01 4.04460728e-02
-1.08080938e-01 -2.73655564e-01 1.29481137e+00 -1.01440167e+00
3.56444120e-01 1.24202359e+00 7.54905164e-01 3.05952996e-01
-1.57139337e+00 -4.19737637e-01 8.68342072e-02 -4.54492643e-02
-1.31929350e+00 -3.80592436e-01 6.75073147e-01 -4.45509166e-01
1.16008949e+00 7.14504004e-01 4.36958462e-01 6.44637883e-01
1.72542080e-01 6.46980703e-01 1.05385113e+00 -6.82170987e-01
-6.11725561e-02 -1.32597163e-01 4.50267762e-01 7.99124420e-01
5.71200728e-01 -2.30687365e-01 -7.14533389e-01 -1.30695835e-01
9.08118725e-01 -7.03123137e-02 6.76786676e-02 -3.43210846e-01
-1.41298485e+00 8.69172871e-01 3.78955424e-01 5.39332688e-01
-4.34170753e-01 3.32790256e-01 3.08645010e-01 3.28973681e-01
6.68953776e-01 5.63507617e-01 -4.37447995e-01 1.64008051e-01
-7.25535214e-01 4.60422158e-01 8.02376568e-01 5.23328960e-01
1.07921350e+00 2.40898412e-02 -5.78044951e-01 9.40324843e-01
-3.00545096e-01 5.02695560e-01 6.15299530e-02 -9.49206114e-01
3.94681811e-01 6.98437572e-01 -1.18749015e-01 -1.05173254e+00
-7.07477033e-01 -5.85286260e-01 -1.40600812e+00 5.34472585e-01
7.27163970e-01 2.52256155e-01 -6.62269413e-01 1.80834162e+00
-8.55574533e-02 -4.17776465e-01 -5.62408715e-02 5.60039997e-01
7.64820755e-01 3.45791191e-01 -1.04442060e-01 -2.85275858e-02
1.43035042e+00 -6.53084338e-01 -4.09871340e-01 3.52713138e-01
5.97305536e-01 -7.80118227e-01 1.09522510e+00 6.52900517e-01
-8.41409147e-01 -4.51632977e-01 -1.18671834e+00 -4.15981412e-01
-6.16894901e-01 3.59142095e-01 1.20974207e+00 8.49310994e-01
-1.09308493e+00 1.19265854e+00 -5.42165399e-01 -5.57242669e-02
6.92291319e-01 7.49694049e-01 -7.37779021e-01 3.84910315e-01
-8.57586980e-01 7.33891606e-01 1.35137020e-02 4.46492285e-02
-1.74906716e-01 -8.63013089e-01 -6.66248202e-01 1.67742580e-01
9.78069380e-02 -5.34344137e-01 7.90414333e-01 -7.34004617e-01
-1.45858657e+00 8.32883179e-01 -1.49048567e-01 -6.40202165e-01
3.13203543e-01 -3.78358871e-01 -9.71558988e-02 1.52968153e-01
-1.87829688e-01 5.09963512e-01 7.13994265e-01 -7.86318541e-01
-5.42433798e-01 -4.52873796e-01 3.72146428e-01 -4.47830930e-03
-8.48362744e-01 -2.91850477e-01 2.91334670e-02 -8.89717877e-01
3.33516687e-01 -7.87444711e-01 -2.99871624e-01 1.13360777e-01
-4.37418252e-01 -2.91360795e-01 5.55790007e-01 -3.97795498e-01
1.31585276e+00 -2.05129361e+00 2.14389235e-01 3.56608570e-01
6.07258499e-01 1.58638790e-01 -2.32868269e-01 2.59223521e-01
-2.22409189e-01 2.37598464e-01 -6.75387979e-02 -2.24215552e-01
3.19631755e-01 2.35770899e-03 -2.45789483e-01 5.01855016e-01
6.28036797e-01 8.05621266e-01 -4.12363380e-01 -1.22903436e-01
4.05154645e-01 7.28001773e-01 -8.07492375e-01 -3.94314915e-01
7.87342042e-02 7.09010884e-02 -2.23052546e-01 3.50089312e-01
7.26855278e-01 -2.25565508e-01 3.63110676e-02 -8.89780104e-01
-1.67747974e-01 1.97808474e-01 -1.22925580e+00 1.33291364e+00
-6.10573590e-01 8.43109787e-01 -1.10191025e-01 -1.08855343e+00
9.01276469e-01 -1.79843396e-01 7.35689163e-01 -7.99823165e-01
-3.64989229e-02 2.69656897e-01 -1.71809772e-03 1.18963629e-01
6.07602179e-01 -4.47767936e-02 -2.69954860e-01 5.67699671e-01
2.41269469e-01 -8.99654403e-02 4.34473246e-01 1.49682537e-01
1.01434422e+00 -3.20642740e-01 1.89034447e-01 -7.46112525e-01
6.86394751e-01 -6.74633145e-01 7.22751245e-02 9.86636519e-01
4.61670250e-01 5.84430575e-01 8.21867764e-01 -7.13866949e-01
-1.10949779e+00 -1.08715057e+00 -1.98640943e-01 1.38447976e+00
-3.40960205e-01 -6.52080595e-01 -7.51075149e-01 -2.66684890e-01
6.21491931e-02 2.13325277e-01 -1.06392431e+00 -3.68755125e-02
-6.73118412e-01 -1.13296473e+00 6.29490852e-01 6.12432361e-01
5.99960387e-01 -6.27260566e-01 -6.12901270e-01 3.61965708e-02
1.02983683e-01 -1.01253033e+00 -2.27362320e-01 5.75390697e-01
-7.12319076e-01 -8.51373494e-01 -7.09078550e-01 -2.38907740e-01
6.41561508e-01 4.35951203e-02 1.10626090e+00 9.24010575e-02
-4.11836475e-01 1.66613370e-01 -1.54277220e-01 -4.65583295e-01
5.10475561e-02 3.40120524e-01 1.40785754e-01 2.37577394e-01
2.24513814e-01 -7.80900896e-01 -4.87114370e-01 2.01665238e-02
-1.17376304e+00 1.18699286e-03 7.07565546e-01 8.83794725e-01
5.28690815e-01 -1.74320012e-01 1.76626638e-01 -7.81325936e-01
8.30510616e-01 2.58862227e-01 -4.67338979e-01 5.08356057e-02
-1.82085797e-01 6.89112782e-01 7.20197082e-01 -4.51810658e-01
-6.43623769e-01 9.60098580e-02 -1.25633618e-02 -1.25160590e-01
1.93720460e-02 2.85581112e-01 -1.59445226e-01 -5.28915524e-01
8.53930056e-01 2.90322886e-03 -7.35525563e-02 -5.90487599e-01
5.81324518e-01 2.79794186e-01 3.74372900e-01 -9.10669625e-01
5.38087308e-01 7.03549504e-01 7.28361845e-01 -1.07250905e+00
-7.87635446e-01 -3.30218551e-04 -8.76959443e-01 1.64103091e-01
6.82707429e-01 -4.18623865e-01 -1.02937615e+00 5.56146920e-01
-1.12113094e+00 -1.81684017e-01 -6.28275812e-01 2.92545438e-01
-5.82725942e-01 3.86849850e-01 -4.65168953e-01 -8.47523630e-01
-3.88531059e-01 -9.38543618e-01 1.15200949e+00 -9.82245281e-02
-2.31688157e-01 -7.96567738e-01 -4.58586335e-01 -1.13691308e-01
5.71847320e-01 2.84528494e-01 1.18579519e+00 -5.20382464e-01
-5.43065369e-01 -1.26871705e-01 -4.84250456e-01 6.23550296e-01
4.24861982e-02 6.08613947e-03 -1.04262340e+00 3.44049744e-02
-3.06431800e-01 6.61252439e-02 1.37337887e+00 3.23763967e-01
1.27820122e+00 -5.59976041e-01 1.29873380e-01 5.96746743e-01
1.09808052e+00 -2.96397775e-01 7.25637078e-01 1.44102082e-01
7.88013875e-01 6.82332337e-01 -8.47297013e-02 5.98829985e-01
-9.28198397e-02 8.57248664e-01 1.75562590e-01 -2.72472233e-01
-3.74047101e-01 2.78811663e-01 -8.43545720e-02 5.03918648e-01
-5.13576031e-01 6.64572641e-02 -5.72074175e-01 2.64065266e-01
-1.66545677e+00 -9.40751553e-01 -4.42954957e-01 2.42494988e+00
5.62622488e-01 2.94056624e-01 1.33351728e-01 4.10887629e-01
3.43167663e-01 2.58655906e-01 -9.69977453e-02 -4.94302571e-01
-6.71616912e-01 1.09022498e+00 7.94644237e-01 6.80574954e-01
-1.28634357e+00 8.02980661e-01 6.75099516e+00 1.19440436e+00
-1.08276582e+00 5.39663061e-02 6.37999117e-01 -2.03796640e-01
-4.62240875e-01 -2.85128832e-01 -6.35446906e-01 1.03238411e-01
3.81755561e-01 -2.34863684e-02 4.39821690e-01 4.56556410e-01
-2.95908093e-01 -7.28132203e-02 -1.31035447e+00 1.11675537e+00
-1.45926714e-01 -1.67761707e+00 5.23092806e-01 3.75629872e-01
7.50218332e-01 -1.84457794e-01 3.00511092e-01 -7.08284229e-02
1.51493430e-01 -1.15374649e+00 7.51914263e-01 7.87083566e-01
8.47192943e-01 -7.29351103e-01 4.72543627e-01 -1.62203908e-01
-1.28459263e+00 -4.31471057e-02 -3.69040996e-01 -4.51625645e-01
-1.71448901e-01 1.24884272e+00 -1.90764338e-01 7.03265667e-01
3.76949459e-01 1.65428728e-01 -6.90214157e-01 5.81350505e-01
2.24799946e-01 6.13412000e-02 -8.19475949e-01 -2.33681306e-01
2.89335519e-01 -9.97071117e-02 3.44765544e-01 1.30530107e+00
2.42384762e-01 -5.23101985e-02 -2.12654233e-01 6.00851297e-01
-6.68506697e-02 2.63451904e-01 -3.65173936e-01 -8.29326957e-02
-6.81619346e-02 1.23385656e+00 -9.15824234e-01 -3.93375486e-01
-2.22048476e-01 8.29689562e-01 3.18288237e-01 3.37563068e-01
-4.97245699e-01 -7.38531888e-01 9.85672057e-01 2.19788700e-01
4.03212219e-01 -5.05665183e-01 -6.57282233e-01 -1.12898993e+00
3.46246928e-01 -4.58116591e-01 1.68831483e-01 -2.39057481e-01
-1.27431083e+00 6.96785152e-01 4.01785895e-02 -8.91382277e-01
-1.01183131e-01 -1.25779366e+00 -1.80173397e-01 9.09174144e-01
-1.02248108e+00 -1.06999385e+00 -2.43103877e-01 5.52352965e-01
1.78244878e-02 -1.65079951e-01 1.12344503e+00 4.83574539e-01
-1.61463678e-01 7.84341395e-01 5.76171428e-02 3.50460745e-02
1.96522415e-01 -1.11702394e+00 2.98756927e-01 5.84194303e-01
3.70117366e-01 6.39676750e-01 5.83397269e-01 -1.66618913e-01
-1.23990870e+00 -7.01214314e-01 8.65395308e-01 -3.28513414e-01
6.17571652e-01 -7.57714272e-01 -7.34036684e-01 1.85858995e-01
-1.11658573e-01 2.62583941e-01 6.70158863e-01 3.23640674e-01
-7.38433719e-01 -4.99455959e-01 -9.03587282e-01 5.79695702e-01
1.31052852e+00 -6.91473484e-01 -4.77775447e-02 4.34434414e-01
4.45459187e-01 1.00395763e-02 -9.26714480e-01 6.31793261e-01
1.04473114e+00 -1.06794083e+00 1.06764007e+00 -6.48048937e-01
4.40405101e-01 1.30489934e-02 -5.83674908e-01 -8.38957667e-01
-6.47597194e-01 -6.05498374e-01 -7.46017545e-02 7.85921693e-01
4.77860123e-01 -7.37345397e-01 8.41525614e-01 4.87334728e-01
1.16004758e-01 -9.01351213e-01 -1.05845964e+00 -8.37023199e-01
2.17142925e-01 -5.73039234e-01 4.46578532e-01 6.24397099e-01
-1.97777972e-01 1.79878414e-01 -1.55765474e-01 -3.00093263e-01
5.98429203e-01 -1.16112292e-01 7.55314231e-01 -1.57525682e+00
-2.07048744e-01 -1.03770602e+00 -7.53619790e-01 -1.00346243e+00
2.12963615e-02 -8.25757742e-01 -4.82186407e-01 -1.28952992e+00
6.02589585e-02 -5.19097507e-01 -5.53825080e-01 7.25282669e-01
3.56772542e-01 7.57640123e-01 2.65669644e-01 -5.63966818e-02
-3.18544865e-01 2.81761825e-01 1.05497909e+00 -2.39195406e-01
-3.22369374e-02 -2.98411220e-01 -9.12504494e-01 6.69794559e-01
6.19178295e-01 -8.18008333e-02 -9.59961489e-02 -3.16122860e-01
6.00568116e-01 -5.20689428e-01 5.18627226e-01 -1.15049160e+00
-3.01557221e-02 1.26271732e-02 5.37529945e-01 -4.15053666e-01
5.28727591e-01 -5.01951694e-01 3.90098616e-02 1.41877145e-01
-4.99210268e-01 -1.63816273e-01 4.13692892e-01 -1.02142468e-02
-6.69315159e-02 -8.22746977e-02 5.88406146e-01 -6.57128617e-02
-3.86069208e-01 1.35278076e-01 -3.80723834e-01 -2.64708042e-01
5.67016840e-01 -2.85963058e-01 -2.96082318e-01 -2.78023511e-01
-8.40827584e-01 -5.48314214e-01 3.48032340e-02 3.19446146e-01
3.27613324e-01 -1.63977122e+00 -7.57175803e-01 4.30342644e-01
-1.08748622e-01 -2.02751622e-01 2.70827860e-01 8.57529223e-01
-4.59480226e-01 6.68722630e-01 -4.79537785e-01 -5.58689058e-01
-1.19314921e+00 1.61123171e-01 1.52800694e-01 -6.55239284e-01
-3.01577300e-01 1.00983894e+00 6.61795914e-01 -2.06267089e-01
5.75751029e-02 -6.60882294e-01 -3.40396096e-03 3.62224281e-01
6.50856435e-01 3.63204271e-01 5.89655399e-01 -7.05080926e-01
-3.62429947e-01 6.73135638e-01 -1.35081977e-01 -2.71575928e-01
1.61693084e+00 2.54106164e-01 -4.59430367e-01 3.23033512e-01
1.52056420e+00 5.06812595e-02 -1.07732630e+00 -1.70197845e-01
-2.52960563e-01 -3.10283959e-01 4.89435308e-02 -5.09657085e-01
-9.75280404e-01 1.06301022e+00 4.83191788e-01 4.88724083e-01
1.18929911e+00 -1.34049794e-02 2.78576255e-01 7.50191450e-01
3.25254112e-01 -1.00426126e+00 2.21759617e-01 6.87677681e-01
1.08729219e+00 -6.41275406e-01 3.12555939e-01 -7.76419342e-01
-4.15610857e-02 1.34112799e+00 1.68838710e-01 -4.16804552e-01
6.87878788e-01 4.02861178e-01 -3.41743618e-01 -1.11522987e-01
-5.49481809e-01 -3.52132469e-01 6.64418578e-01 5.34298480e-01
6.04470849e-01 2.16285542e-01 -5.02623558e-01 5.75220168e-01
-4.23007607e-01 -2.30804935e-01 2.07121387e-01 6.93534553e-01
-3.62479687e-01 -1.26999056e+00 -3.19268882e-01 9.28022861e-01
-6.35407209e-01 -4.14343506e-01 -4.77905631e-01 6.78803384e-01
2.67886758e-01 5.56177318e-01 3.27119619e-01 -5.24419963e-01
3.07501704e-01 -5.88734560e-02 1.05046856e+00 -3.85584086e-01
-5.47252834e-01 -1.12830617e-01 1.49104491e-01 -6.99971199e-01
-4.60850507e-01 -3.99908394e-01 -8.99409890e-01 -3.34900528e-01
-1.94710851e-01 -7.99003243e-02 6.22860670e-01 9.15307820e-01
3.92649531e-01 6.32161856e-01 2.58522004e-01 -1.19123459e+00
-3.31926912e-01 -9.15321112e-01 -6.34348035e-01 5.71893275e-01
3.58255982e-01 -8.10181797e-01 -2.25784257e-01 -8.78359079e-02]
|
[9.011884689331055, 2.4234886169433594]
|
9714c1ca-0df8-49ae-b248-250b557ea2da
|
learning-a-pose-lexicon-for-semantic-action
|
1604.00147
| null |
http://arxiv.org/abs/1604.00147v1
|
http://arxiv.org/pdf/1604.00147v1.pdf
|
Learning a Pose Lexicon for Semantic Action Recognition
|
This paper presents a novel method for learning a pose lexicon comprising
semantic poses defined by textual instructions and their associated visual
poses defined by visual features. The proposed method simultaneously takes two
input streams, semantic poses and visual pose candidates, and statistically
learns a mapping between them to construct the lexicon. With the learned
lexicon, action recognition can be cast as the problem of finding the maximum
translation probability of a sequence of semantic poses given a stream of
visual pose candidates. Experiments evaluating pre-trained and zero-shot action
recognition conducted on MSRC-12 gesture and WorkoutSu-10 exercise datasets
were used to verify the efficacy of the proposed method.
|
['Lijuan Zhou', 'Philip Ogunbona', 'Wanqing Li']
|
2016-04-01
| null | null | null | null |
['zero-shot-action-recognition']
|
['computer-vision']
|
[ 5.38231492e-01 -2.54708111e-01 -4.22292948e-01 -5.63321352e-01
-8.63857627e-01 -4.87163097e-01 6.89749122e-01 -3.30346942e-01
-6.09571815e-01 4.06259418e-01 5.32670438e-01 3.55123490e-01
-3.09454888e-01 -2.81015635e-01 -7.40172267e-01 -5.49615443e-01
-2.78829336e-01 6.15047455e-01 3.56047213e-01 1.49819553e-01
4.65931892e-01 4.28153813e-01 -1.91547465e+00 4.88783121e-01
6.09457493e-03 1.09870958e+00 4.77040648e-01 1.12322617e+00
5.22733107e-03 8.94417167e-01 -6.98774397e-01 6.62882254e-02
3.23436677e-01 -5.92185616e-01 -9.24232841e-01 5.39309442e-01
5.80591917e-01 -3.43165636e-01 -6.12480044e-01 7.94852734e-01
5.25458813e-01 9.13227558e-01 7.46707678e-01 -1.21761012e+00
-9.50028226e-02 1.71797246e-01 -2.25836456e-01 3.36609125e-01
8.33648145e-01 3.60532999e-01 8.55502963e-01 -6.98410153e-01
8.63620400e-01 1.27636790e+00 1.54688001e-01 5.94736874e-01
-8.87504458e-01 -3.85462552e-01 1.84005886e-01 5.07646859e-01
-1.38214028e+00 -1.42624661e-01 7.53214478e-01 -4.51858044e-01
1.20273256e+00 3.06976616e-01 9.27629411e-01 1.33037066e+00
1.27831250e-01 7.34911323e-01 1.01980948e+00 -5.43338776e-01
4.62959439e-01 -3.86799276e-01 6.61418736e-02 7.72564530e-01
-2.13995814e-01 3.85555476e-02 -1.31107211e+00 5.26702106e-02
8.44554484e-01 4.22864556e-02 6.67666793e-02 -4.74836171e-01
-1.18473554e+00 4.83191848e-01 2.28835225e-01 1.37075678e-01
-3.68113577e-01 4.35589790e-01 2.22689867e-01 -6.61201701e-02
7.37751722e-02 1.91825926e-01 -2.81636924e-01 -4.63312745e-01
-5.19296646e-01 6.56648949e-02 5.40057182e-01 9.31127310e-01
6.32955909e-01 -3.26109491e-02 -2.84619451e-01 3.50649595e-01
4.90312099e-01 7.97171712e-01 5.49740314e-01 -8.90171766e-01
5.85510254e-01 6.30385995e-01 1.05175123e-01 -1.01688242e+00
-3.68309766e-01 1.76613316e-01 8.36534202e-02 -1.25017986e-02
4.47545856e-01 9.68327839e-03 -1.16958618e+00 1.40128827e+00
4.64766294e-01 5.11581421e-01 7.25049302e-02 1.28609896e+00
5.44489026e-01 5.49291611e-01 3.84594172e-01 -1.28674954e-01
1.30168033e+00 -4.92953122e-01 -6.77198887e-01 -6.30694389e-01
1.65476203e-01 -5.40589929e-01 9.94755447e-01 1.41328365e-01
-5.05815685e-01 -7.86676049e-01 -9.97444808e-01 1.13193288e-01
2.07477082e-02 5.94700456e-01 4.26242173e-01 4.75667536e-01
-5.54657578e-01 4.53246295e-01 -1.13039231e+00 -4.77814347e-01
1.95366427e-01 3.99764717e-01 -3.50146919e-01 1.99032471e-01
-8.03735614e-01 7.33230472e-01 5.53274393e-01 2.64116228e-01
-1.30083454e+00 -3.37333605e-02 -9.95333791e-01 -4.31322396e-01
5.60563684e-01 -6.48871735e-02 1.16797853e+00 -1.11616528e+00
-1.67960596e+00 7.52944946e-01 -1.69351682e-01 -3.24755430e-01
2.31126025e-01 -7.52680779e-01 -2.67949402e-01 5.76285660e-01
1.11684315e-02 7.74547458e-01 8.69332373e-01 -1.24803185e+00
-6.76961124e-01 -5.85661232e-01 9.63415578e-02 7.87087083e-01
-2.69968182e-01 -8.21759365e-03 -7.84881234e-01 -2.54208833e-01
2.59658426e-01 -9.83590305e-01 -1.03783742e-01 -9.58718657e-02
-3.49741280e-01 -2.01201797e-01 7.14361429e-01 -7.37256229e-01
7.38794684e-01 -2.02261400e+00 4.39814717e-01 3.79139572e-01
-4.99158949e-01 -6.30114228e-02 -2.20885858e-01 2.40217656e-01
1.16881952e-01 -5.38461447e-01 2.44099811e-01 -6.43396974e-02
-1.48112893e-01 4.47248250e-01 -4.42164212e-01 7.85313845e-01
5.46245724e-02 8.06156397e-01 -1.00783217e+00 -6.52119279e-01
7.19212890e-01 4.46729720e-01 -9.35782939e-02 4.86736745e-01
-6.29053116e-01 5.38523436e-01 -7.39558518e-01 5.12860537e-01
-2.36967430e-01 4.10097837e-02 5.89973629e-01 -3.97677422e-01
1.64780468e-01 -1.87170848e-01 -1.24847817e+00 1.92307019e+00
-8.89415592e-02 6.35674238e-01 -7.88373590e-01 -7.62613237e-01
1.11328530e+00 2.91119456e-01 7.45428503e-01 -5.70481658e-01
4.77196157e-01 -2.33810529e-01 -4.19082522e-01 -1.03416395e+00
4.88843858e-01 -2.16852546e-01 -4.20253545e-01 4.26594794e-01
3.93446863e-01 1.79940656e-01 -9.30444598e-02 -1.26198336e-01
1.12602699e+00 8.07195723e-01 3.20479304e-01 3.00083160e-01
3.63489807e-01 2.56697625e-01 2.81582296e-01 5.64277887e-01
-2.39950448e-01 4.71666396e-01 1.04963772e-01 -5.85378528e-01
-7.97255814e-01 -1.21471715e+00 5.65575838e-01 1.03387070e+00
4.71906096e-01 -4.04031038e-01 -5.89105070e-01 -9.27906394e-01
-4.08921003e-01 6.18500173e-01 -6.23358011e-01 -1.02760375e-01
-6.65391564e-01 -9.63909458e-03 6.95624113e-01 8.51060152e-01
4.45853800e-01 -1.31072521e+00 -1.35492301e+00 -2.82497704e-01
-4.20263708e-01 -1.29978740e+00 -5.90827763e-01 7.03755841e-02
-8.17140520e-01 -1.53151202e+00 -3.99974585e-01 -9.68598664e-01
8.97310734e-01 -1.07272491e-02 6.45437956e-01 -1.84715003e-01
-6.05736494e-01 8.69969487e-01 -5.54802537e-01 -1.55366495e-01
-2.21811146e-01 -4.49397385e-01 1.90717399e-01 4.41172659e-01
4.24534351e-01 -1.05070338e-01 -4.48037326e-01 3.69559646e-01
-6.15449071e-01 2.27078795e-04 5.84748268e-01 5.40746272e-01
1.00470543e+00 -1.45172924e-01 1.26318514e-01 -2.31268913e-01
2.54335642e-01 7.71269500e-02 -3.08850080e-01 5.65182388e-01
8.23292509e-03 2.70115554e-01 -2.01822817e-03 -7.55689561e-01
-1.20603251e+00 1.07364714e+00 4.45154876e-01 -8.22057068e-01
-4.67986554e-01 2.71795988e-01 -3.56235981e-01 2.57926822e-01
7.26077139e-01 3.64804178e-01 7.60343671e-02 -3.94856513e-01
6.24748111e-01 5.61043561e-01 9.59264517e-01 -6.28076434e-01
6.28845394e-01 5.16302347e-01 6.28431439e-02 -7.96618223e-01
-7.67072916e-01 -7.83118486e-01 -1.19409752e+00 -1.05597103e+00
1.40652704e+00 -7.81142831e-01 -7.02441037e-01 3.43648911e-01
-1.06059098e+00 -1.83406621e-01 -3.02373886e-01 8.29779387e-01
-9.79144692e-01 2.72234201e-01 2.90579293e-02 -1.00757432e+00
1.59995466e-01 -1.01452184e+00 1.43984294e+00 2.65518218e-01
-5.13206482e-01 -6.68067098e-01 2.10843086e-01 5.31437218e-01
-3.74381065e-01 3.43010306e-01 5.15907109e-01 -6.49896502e-01
-4.84344542e-01 -5.36786556e-01 3.06067318e-01 1.25557974e-01
2.94080496e-01 -3.58068734e-01 -7.70718157e-01 -1.13269836e-01
-1.23194650e-01 -6.96911573e-01 4.08765286e-01 2.58482665e-01
8.96458745e-01 -1.69630006e-01 -3.34190786e-01 3.17441821e-01
1.28836465e+00 4.80742365e-01 6.65742695e-01 1.90116405e-01
7.77012229e-01 5.68914592e-01 1.01749313e+00 5.80139637e-01
-1.20972246e-01 9.36042428e-01 3.09607387e-01 2.94443995e-01
-8.94420743e-02 -5.97325385e-01 5.21214962e-01 4.71724182e-01
-2.99603641e-01 -7.10303634e-02 -8.24326098e-01 9.30013210e-02
-1.88136971e+00 -1.16171145e+00 3.62375438e-01 2.07350183e+00
8.19300175e-01 2.13021673e-02 2.42530569e-01 -2.60670427e-02
6.88222587e-01 2.30944857e-01 -7.27575302e-01 1.47294462e-01
2.47076392e-01 3.48448664e-01 4.82887954e-01 3.96520197e-01
-1.27184486e+00 9.38527167e-01 6.51553535e+00 6.67663217e-01
-1.07789791e+00 -3.17541100e-02 2.58639932e-01 -3.64711583e-01
2.26170436e-01 -9.12357643e-02 -7.06474662e-01 -4.19067033e-03
8.45854998e-01 3.02729551e-02 2.73941875e-01 8.62807393e-01
2.54160255e-01 -3.40743303e-01 -1.20405126e+00 1.16795135e+00
5.79160333e-01 -1.11152840e+00 1.91824168e-01 -3.08107525e-01
3.88360977e-01 -2.23966449e-01 -8.60063434e-02 2.10940152e-01
6.40482828e-02 -9.82170343e-01 9.91173983e-01 8.13256323e-01
9.32807088e-01 -5.48560262e-01 1.42427117e-01 2.58802235e-01
-1.34413469e+00 -1.93400055e-01 -9.32268426e-02 -2.94045895e-01
3.96807566e-02 -4.97637779e-01 -1.00473356e+00 4.79930520e-01
3.68696958e-01 7.66097188e-01 -6.11547530e-01 8.44721735e-01
-5.44235587e-01 6.06467605e-01 -1.64751336e-01 -3.28276604e-01
-2.81148888e-02 2.07217902e-01 5.54516375e-01 8.81511509e-01
1.98316485e-01 1.35215104e-01 6.02244020e-01 3.42788905e-01
5.27282953e-01 -2.87405066e-02 -5.76001525e-01 -3.41754779e-02
3.66353065e-01 8.42560589e-01 -1.36069417e+00 -1.42282218e-01
-9.34163481e-02 1.21411824e+00 8.04691948e-03 4.60635155e-01
-8.83492768e-01 3.76996696e-02 6.15696788e-01 -1.28792584e-01
2.04545677e-01 -4.32043999e-01 -5.86979762e-02 -7.84353495e-01
-7.13011846e-02 -7.92866826e-01 4.92426276e-01 -9.23608243e-01
-7.76146352e-01 4.31129396e-01 5.90219080e-01 -1.77986872e+00
-4.39685643e-01 -6.76597953e-01 -3.37097943e-01 4.77645457e-01
-4.98745322e-01 -1.00525582e+00 -4.13709700e-01 6.58211470e-01
1.16783881e+00 -2.53950924e-01 9.67281401e-01 -1.35291010e-01
-2.38275051e-01 2.21632272e-01 -3.41998577e-01 2.69542873e-01
4.19362664e-01 -9.44145799e-01 1.32057130e-01 6.88823879e-01
8.51909041e-01 3.14060688e-01 7.86674738e-01 -1.10051870e+00
-1.68915009e+00 -1.05570769e+00 4.69499290e-01 -7.00328052e-01
2.86528617e-01 -2.57535517e-01 -2.71588802e-01 6.86586320e-01
-2.53034323e-01 -4.75928411e-02 4.72025484e-01 -3.89600515e-01
-1.30145654e-01 1.82102084e-01 -7.95628667e-01 5.25608599e-01
1.24532020e+00 -6.59512758e-01 -8.62593174e-01 4.37817216e-01
4.46874499e-01 -8.10606122e-01 -4.78071213e-01 9.34645832e-02
8.33783567e-01 -3.33870471e-01 1.05432844e+00 -8.50056767e-01
3.01409543e-01 -2.45074168e-01 -4.56062496e-01 -9.96417046e-01
1.96469814e-01 -3.19509149e-01 5.97694069e-02 7.23949313e-01
-1.10313185e-02 1.82716280e-01 9.16380286e-01 3.48662734e-01
1.58907115e-01 -5.46261907e-01 -1.26475549e+00 -9.88471866e-01
-7.74608016e-01 -8.14840853e-01 2.86530167e-01 2.14321166e-01
-3.97070497e-02 2.61453092e-01 -5.87568879e-01 3.11917603e-01
6.23309851e-01 -2.66770236e-02 8.60369742e-01 -7.59536743e-01
-4.16123152e-01 3.14624682e-02 -9.55547333e-01 -1.10476232e+00
4.02683794e-01 -7.91963100e-01 8.11073065e-01 -1.45525050e+00
3.07033926e-01 1.94580138e-01 -2.62666374e-01 7.00386763e-01
5.79601750e-02 2.04750776e-01 2.64373600e-01 3.35320741e-01
-9.23408091e-01 5.02301097e-01 9.79361415e-01 -2.34554604e-01
-2.05747575e-01 -9.30386633e-02 9.59995538e-02 6.85568631e-01
6.06611192e-01 -4.85841602e-01 -9.23625946e-01 -1.50187284e-01
-1.14615515e-01 4.15474117e-01 6.98009670e-01 -1.24244571e+00
9.48546380e-02 -5.14935970e-01 5.77444434e-01 -7.15620756e-01
6.21140480e-01 -7.75094926e-01 2.92070389e-01 5.41300356e-01
-8.13576996e-01 -8.80113393e-02 1.30578592e-01 1.08287585e+00
9.67568308e-02 -9.24522057e-02 1.95886984e-01 -4.66526896e-02
-1.49172938e+00 1.55637190e-02 -3.06461990e-01 7.51511157e-02
1.32007718e+00 -5.43595850e-01 -2.99568363e-02 -3.45789909e-01
-9.91528392e-01 9.16672423e-02 2.91536860e-02 8.35259080e-01
1.13380158e+00 -1.42999780e+00 -1.66385382e-01 3.02440673e-01
2.26856157e-01 -3.57113570e-01 1.94436401e-01 5.52548110e-01
-3.92575711e-01 3.13582718e-01 -5.24103165e-01 -7.66610146e-01
-1.72641122e+00 1.80277042e-02 3.42113793e-01 6.94950074e-02
-9.30511892e-01 8.85934711e-01 -1.69706851e-01 -1.57385066e-01
6.87684894e-01 -2.06271023e-01 -3.47520441e-01 -1.36684552e-01
4.53560799e-01 2.99440175e-01 -2.68553466e-01 -1.14014411e+00
-7.03883171e-01 6.29003942e-01 6.25600696e-01 -7.01600194e-01
1.05003488e+00 4.67854217e-02 5.49534857e-01 8.10607970e-01
1.11195040e+00 -3.82574856e-01 -1.56910658e+00 -1.33020282e-01
1.34246275e-01 -7.82989323e-01 -3.43843579e-01 -7.89511085e-01
-8.41809630e-01 5.50723553e-01 9.43390131e-01 -8.10018301e-01
8.50664914e-01 2.51788676e-01 4.67627048e-01 4.81527209e-01
6.89068556e-01 -1.31678009e+00 1.02883410e+00 3.90184164e-01
9.64872897e-01 -8.66757333e-01 -6.05259240e-02 -2.25991175e-01
-1.05635595e+00 1.16074729e+00 7.84431338e-01 -6.51209801e-02
4.78356183e-01 6.52463064e-02 2.86032468e-01 -4.54046905e-01
-4.66211200e-01 -4.69386935e-01 7.14393020e-01 8.04289222e-01
6.47920221e-02 2.68923610e-01 -1.27203465e-01 5.40916622e-01
-3.77068222e-02 9.46773291e-02 1.00864828e-01 1.21944952e+00
-6.63884580e-01 -8.15998673e-01 -4.63776767e-01 2.70767540e-01
4.94954921e-02 5.43886364e-01 -6.16621375e-01 6.62701011e-01
1.01245768e-01 8.56237650e-01 2.89505571e-01 -1.04037857e+00
5.07140875e-01 4.44531858e-01 8.56265545e-01 -9.06882465e-01
-3.00377727e-01 3.85187976e-02 1.41359001e-01 -9.30832982e-01
-6.28050745e-01 -9.36154783e-01 -1.76967835e+00 6.76788211e-01
-1.15193143e-01 -1.54428422e-01 5.57553411e-01 1.34306407e+00
-1.16742462e-01 5.24382472e-01 3.52267921e-01 -1.08872283e+00
-8.92325580e-01 -8.55977952e-01 -4.16901231e-01 8.48852336e-01
-1.87347382e-01 -9.79431510e-01 -2.74267644e-02 5.89576483e-01]
|
[8.32082462310791, 0.6803924441337585]
|
56b69c3b-3f7c-4a7b-ac95-8e6c40100e4a
|
uncertainty-aware-lidar-panoptic-segmentation
|
2210.04472
| null |
https://arxiv.org/abs/2210.04472v1
|
https://arxiv.org/pdf/2210.04472v1.pdf
|
Uncertainty-aware LiDAR Panoptic Segmentation
|
Modern autonomous systems often rely on LiDAR scanners, in particular for autonomous driving scenarios. In this context, reliable scene understanding is indispensable. Current learning-based methods typically try to achieve maximum performance for this task, while neglecting a proper estimation of the associated uncertainties. In this work, we introduce a novel approach for solving the task of uncertainty-aware panoptic segmentation using LiDAR point clouds. Our proposed EvLPSNet network is the first to solve this task efficiently in a sampling-free manner. It aims to predict per-point semantic and instance segmentations, together with per-point uncertainty estimates. Moreover, it incorporates methods for improving the performance by employing the predicted uncertainties. We provide several strong baselines combining state-of-the-art panoptic segmentation networks with sampling-free uncertainty estimation techniques. Extensive evaluations show that we achieve the best performance on uncertainty-aware panoptic segmentation quality and calibration compared to these baselines. We make our code available at: \url{https://github.com/kshitij3112/EvLPSNet}
|
['Wolfram Burgard', 'Daniel Büscher', 'Sajad Marvi', 'Kshitij Sirohi']
|
2022-10-10
| null | null | null | null |
['panoptic-segmentation']
|
['computer-vision']
|
[-1.37363598e-01 -1.28528491e-01 -2.16869533e-01 -7.70406961e-01
-1.21227109e+00 -4.88265753e-01 5.39096177e-01 6.20153919e-02
-4.21308398e-01 9.13020372e-01 -5.09021103e-01 -4.03290778e-01
-2.75082648e-01 -1.11203218e+00 -9.27601039e-01 -6.08312249e-01
1.15558974e-01 1.05576491e+00 4.95080978e-01 -1.41519669e-03
2.26644307e-01 5.75561106e-01 -1.72332156e+00 -5.70037782e-01
1.52908623e+00 1.21239090e+00 2.88916588e-01 4.64290470e-01
-2.85733074e-01 8.50477815e-02 -1.84766948e-01 -2.85739452e-01
6.33358598e-01 2.65097797e-01 -4.31714535e-01 -1.60238340e-01
8.06894064e-01 -3.36853534e-01 9.09309983e-02 1.33147418e+00
2.80135244e-01 1.62864059e-01 7.57466137e-01 -1.23913050e+00
1.74823001e-01 6.17959142e-01 -6.15386844e-01 1.35997497e-02
-4.92303669e-01 2.81461418e-01 9.83879447e-01 -8.76887441e-01
1.13263287e-01 1.04411495e+00 6.12052441e-01 5.98161146e-02
-1.41035938e+00 -1.08818746e+00 1.37669265e-01 2.15435967e-01
-1.68407655e+00 -4.31435585e-01 5.89723110e-01 -5.05106986e-01
5.70406079e-01 7.41455182e-02 4.47347730e-01 4.87494260e-01
1.14654258e-01 7.31309652e-01 1.10664153e+00 -2.68890541e-02
4.28063273e-01 -2.95561738e-02 2.62688905e-01 3.55378151e-01
5.61031699e-01 4.41476136e-01 -3.80185038e-01 6.42699301e-02
5.34414172e-01 -2.57790238e-01 -1.01584420e-01 -6.02298737e-01
-8.91060531e-01 9.03258145e-01 5.46547115e-01 -5.77078685e-02
-2.18190432e-01 4.23444122e-01 1.60241909e-02 9.37700178e-03
8.86543810e-01 3.46004069e-01 -6.13101721e-01 2.43550893e-02
-1.42717731e+00 3.43352437e-01 6.96868420e-01 1.08092976e+00
1.34266365e+00 8.29028487e-02 3.34997982e-01 7.10843086e-01
6.36385560e-01 1.07198393e+00 -2.43540674e-01 -1.23645806e+00
4.81385201e-01 2.86321759e-01 2.73606658e-01 -6.14115298e-01
-3.85707051e-01 -3.74644399e-01 -6.35488212e-01 5.18198371e-01
3.52919281e-01 -2.33817935e-01 -1.19872606e+00 1.46956873e+00
5.17975152e-01 5.20518482e-01 -3.19891050e-02 7.75137424e-01
4.65394676e-01 6.42908871e-01 1.51473969e-01 8.67235586e-02
1.08643639e+00 -6.85398698e-01 -4.55386698e-01 -3.72753501e-01
2.85214216e-01 -7.21872032e-01 8.77723932e-01 3.72011393e-01
-6.08870268e-01 -5.90785027e-01 -1.08795309e+00 1.80321932e-01
-3.79246414e-01 1.78867742e-01 4.09990311e-01 7.01545238e-01
-8.68960321e-01 8.82974625e-01 -1.06061697e+00 -9.79533046e-02
2.00435370e-01 3.68439227e-01 1.57881454e-01 8.68348330e-02
-1.10531616e+00 1.01706207e+00 6.16735935e-01 3.30414802e-01
-7.92688072e-01 -1.05315328e+00 -9.42510366e-01 -1.91196769e-01
7.73786426e-01 -5.35548627e-01 1.40198088e+00 -2.14653522e-01
-1.43078554e+00 3.71553451e-01 -6.37707710e-02 -7.34998226e-01
7.04914272e-01 -6.48910105e-01 -1.37779042e-01 -8.43091309e-03
2.27864072e-01 1.07744849e+00 9.18340206e-01 -1.54261971e+00
-7.78917134e-01 -3.43955070e-01 -2.18529344e-01 9.28014982e-03
2.90459752e-01 -5.95519602e-01 -4.04455096e-01 -1.98386267e-01
3.88241678e-01 -1.03337204e+00 -4.97509927e-01 2.19455212e-01
-4.42159534e-01 -1.35000095e-01 7.55552590e-01 -3.22095186e-01
7.83418357e-01 -1.82363093e+00 -2.29135841e-01 3.87110353e-01
7.12268725e-02 4.20826793e-01 9.21159834e-02 1.47065833e-01
4.79766905e-01 1.35281116e-01 -1.06927180e+00 -6.03551626e-01
1.75953239e-01 5.21351457e-01 -4.41639215e-01 3.42641801e-01
3.67048144e-01 8.12618554e-01 -7.58698046e-01 -7.94698894e-01
8.99283707e-01 4.91851658e-01 -2.44034484e-01 1.92493305e-01
-8.59703302e-01 6.84117675e-01 -5.77724040e-01 6.20592833e-01
1.30045414e+00 1.02551393e-01 -5.65065406e-02 -1.04240470e-01
-4.67468113e-01 1.30863130e-01 -1.48872149e+00 1.59968615e+00
-6.88969970e-01 4.62870359e-01 2.64614522e-01 -5.73708892e-01
1.05287564e+00 -1.09876104e-01 4.37063247e-01 -3.39740515e-01
1.64115936e-01 5.84690452e-01 -1.88299507e-01 -2.75216135e-03
7.64028966e-01 -1.61629915e-01 4.34298888e-02 5.30772377e-03
-3.49088348e-02 -1.19336414e+00 2.07034871e-01 9.22960043e-02
3.81198049e-01 5.18001556e-01 1.72057644e-01 -5.46001256e-01
5.57403624e-01 1.66525885e-01 6.79491580e-01 6.65592909e-01
-1.98984981e-01 6.14101589e-01 2.17367038e-02 -1.23660244e-01
-9.74254012e-01 -1.23023558e+00 -7.02140689e-01 3.45389456e-01
3.36445212e-01 -1.65319517e-01 -4.68209118e-01 -3.88930887e-01
4.59054589e-01 1.11702144e+00 -1.86068028e-01 3.15172791e-01
-2.09785283e-01 -6.35357976e-01 4.23125595e-01 4.22822177e-01
7.85983980e-01 -5.10863543e-01 -6.13843322e-01 2.03491807e-01
-1.61816582e-01 -1.12907398e+00 1.27033576e-01 1.25925899e-01
-1.16345000e+00 -1.03508556e+00 -3.96562219e-01 9.91806611e-02
1.97332293e-01 1.79230824e-01 1.08304298e+00 -1.73885450e-01
-1.19850896e-01 1.41012460e-01 -2.07977630e-02 -7.37179518e-01
-8.64297897e-02 1.52426407e-01 4.75051850e-02 -3.06229293e-01
4.23591584e-01 -8.64359617e-01 -4.87666339e-01 3.90954882e-01
-8.34817111e-01 -1.73548609e-01 6.73006356e-01 3.67566943e-01
9.37864006e-01 -3.05270939e-03 1.47224322e-01 -8.01593482e-01
1.22474529e-01 -4.04799432e-01 -1.40799356e+00 8.65677893e-02
-9.21113491e-01 2.18003720e-01 3.22064251e-01 2.02036470e-01
-1.17385018e+00 3.56407702e-01 -4.89794880e-01 -5.11286676e-01
-4.45594132e-01 3.94889623e-01 -1.18991412e-01 -2.70115644e-01
6.40039980e-01 3.32369581e-02 -2.43544087e-01 -5.43026984e-01
6.69791639e-01 5.77440023e-01 5.70470870e-01 -8.98618519e-01
1.04400110e+00 6.16770804e-01 2.39708170e-01 -9.66200590e-01
-9.28997815e-01 -7.79188573e-01 -1.02973807e+00 -3.06775928e-01
5.66903293e-01 -1.10029924e+00 -2.86357820e-01 4.36099708e-01
-9.18990016e-01 -4.47566807e-01 -4.78274077e-01 5.90125144e-01
-7.28018463e-01 5.89761257e-01 2.79665971e-03 -1.18127048e+00
-3.12857330e-01 -1.22115493e+00 1.31328225e+00 2.18126774e-01
2.97676027e-01 -7.52785027e-01 3.03159088e-01 2.35045001e-01
3.89146209e-01 1.75886482e-01 4.09644127e-01 -2.69990504e-01
-1.09067869e+00 1.56877160e-01 -5.94638050e-01 4.95662540e-01
-8.48321319e-02 4.19832528e-01 -1.14115775e+00 8.26838240e-02
-9.32567790e-02 -1.65758669e-01 1.27472484e+00 6.99174702e-01
1.05890703e+00 1.90473288e-01 -3.06401193e-01 6.57010436e-01
1.72975266e+00 -7.30004385e-02 4.96990919e-01 1.91242129e-01
5.41574121e-01 7.23443687e-01 9.53087866e-01 3.09896797e-01
5.06283104e-01 5.41570127e-01 9.13484097e-01 2.69273251e-01
7.72597417e-02 -1.57297447e-01 -2.28232250e-01 4.92440104e-01
5.04667833e-02 -2.28924200e-01 -1.30687392e+00 6.99496150e-01
-2.12310219e+00 -3.95412296e-01 -4.72070158e-01 2.29556537e+00
6.35616064e-01 2.85930067e-01 -3.00948977e-01 -1.71878606e-01
5.87761760e-01 4.07967746e-01 -7.33027637e-01 -4.57864478e-02
1.77553341e-01 2.69874275e-01 1.06593263e+00 8.52242470e-01
-1.18394113e+00 1.25709188e+00 5.34905958e+00 9.71917510e-01
-8.26481283e-01 2.17537239e-01 2.91973978e-01 1.47883266e-01
-4.20974880e-01 2.40165278e-01 -1.18438518e+00 3.02616596e-01
1.20750213e+00 4.40007672e-02 2.71971244e-02 8.41362059e-01
2.65214652e-01 -6.51109695e-01 -6.71587765e-01 6.91643357e-01
-4.80640918e-01 -1.23954201e+00 -1.40454307e-01 8.52222890e-02
8.54565859e-01 6.34943545e-01 -8.08698386e-02 1.26647636e-01
6.87634110e-01 -6.46561980e-01 6.31598055e-01 6.70623243e-01
6.52174652e-01 -8.21662843e-01 7.87393272e-01 4.20470357e-01
-1.29532099e+00 2.40180865e-01 -5.82789600e-01 1.58167943e-01
4.80329931e-01 1.34380734e+00 -6.39937997e-01 9.21094894e-01
7.22627819e-01 7.11991251e-01 -2.98869967e-01 1.40100718e+00
-6.76593423e-01 6.18991256e-01 -9.73068595e-01 3.16522121e-01
3.90275657e-01 -5.50237536e-01 7.18207419e-01 1.08307838e+00
4.64699984e-01 1.32301569e-01 1.67073160e-01 1.17053211e+00
-3.61056291e-02 5.84103800e-02 -7.10218012e-01 3.18157703e-01
5.98785520e-01 1.21660674e+00 -5.69588721e-01 -2.72605866e-01
-7.43570998e-02 1.56375274e-01 1.12635717e-01 3.58919427e-02
-7.41271257e-01 -1.85788184e-01 1.03480482e+00 7.09037706e-02
2.64368623e-01 -7.66656816e-01 -5.15142083e-01 -9.87655580e-01
-8.31261575e-02 -2.29903191e-01 1.44611657e-01 -8.36115003e-01
-1.09942973e+00 3.88920724e-01 5.06956637e-01 -1.18628383e+00
-2.70495504e-01 -5.46001852e-01 -5.08045614e-01 8.38057399e-01
-2.25583649e+00 -1.13598859e+00 -4.03634489e-01 1.79787695e-01
6.64481997e-01 3.83558065e-01 4.10125077e-01 1.56538293e-01
-2.15612248e-01 -1.26831859e-01 3.37411404e-01 -3.78575057e-01
6.73369944e-01 -1.36514521e+00 5.88413954e-01 1.14451170e+00
1.16557702e-01 1.81348488e-01 8.90817225e-01 -8.44967604e-01
-8.86670291e-01 -1.34963703e+00 6.30890191e-01 -3.73901308e-01
9.23800230e-01 -1.28364891e-01 -9.62300777e-01 5.65730810e-01
-1.20419152e-01 -5.74339516e-02 1.29245862e-01 7.38157928e-02
-1.62241146e-01 -3.17201972e-01 -1.17462242e+00 3.28869790e-01
7.51830518e-01 -3.14900845e-01 -5.61494052e-01 3.21558833e-01
8.53067696e-01 -4.88679737e-01 -8.39980423e-01 1.01116610e+00
4.57907766e-01 -9.85062659e-01 8.67517829e-01 1.61447406e-01
2.62913734e-01 -7.79527247e-01 -3.06219697e-01 -1.32910490e+00
1.95660621e-01 -2.36352637e-01 -2.32909415e-02 1.11443567e+00
5.56692898e-01 -8.60482693e-01 7.94186354e-01 6.44998312e-01
-2.47737661e-01 -2.37733513e-01 -1.23517859e+00 -1.06295025e+00
2.65960097e-01 -1.00331807e+00 7.82166064e-01 4.74404693e-01
-8.45057249e-01 -5.89229129e-02 -3.30048263e-01 7.72265255e-01
1.23101938e+00 3.14457804e-01 8.64632785e-01 -1.59807503e+00
1.55371457e-01 -3.25503737e-01 -2.88032979e-01 -1.01074266e+00
2.35104680e-01 -4.46030885e-01 5.90134561e-01 -1.51655519e+00
-3.75812262e-01 -9.92480695e-01 1.05466126e-02 1.80030242e-01
4.79492098e-02 1.53657049e-01 2.27446735e-01 3.76396030e-01
-2.17786238e-01 7.95152366e-01 8.74848664e-01 -1.91372812e-01
-1.89775392e-01 4.10654724e-01 -1.34690702e-01 7.03652859e-01
1.39637661e+00 -6.89699531e-01 -4.62123334e-01 -6.38416350e-01
5.32307327e-02 -1.68908849e-01 4.77370441e-01 -1.50536382e+00
4.19440299e-01 -3.71156126e-01 -1.56050444e-01 -1.28533161e+00
5.82130492e-01 -1.01061761e+00 2.61661172e-01 2.49856204e-01
1.36157677e-01 -3.04923415e-01 5.47817171e-01 7.01086223e-01
-2.75819391e-01 -5.01670182e-01 9.68575895e-01 -1.30372524e-01
-1.05999219e+00 5.21128476e-01 -1.31001592e-01 -7.74761885e-02
8.86517763e-01 1.73284605e-01 -3.62860531e-01 -2.17193678e-01
-5.85128486e-01 7.49539554e-01 4.42724824e-01 1.18400536e-01
4.10954654e-01 -7.56832123e-01 -7.31887579e-01 4.21232395e-02
2.39775851e-01 6.43090248e-01 2.86955833e-01 6.87271476e-01
-5.86637497e-01 5.91067910e-01 -2.52568368e-02 -1.12035930e+00
-8.02102923e-01 1.51822135e-01 4.79274720e-01 -1.63202658e-01
-4.62565690e-01 7.73433685e-01 1.37216225e-02 -9.75616157e-01
-5.10179773e-02 -6.40011430e-01 3.63251045e-02 6.93325922e-02
1.45173997e-01 3.57953221e-01 1.46444693e-01 -5.31417906e-01
-2.32645273e-01 8.27998757e-01 3.48447025e-01 -3.11492771e-01
1.21169019e+00 -3.43258202e-01 6.42351201e-03 6.07325077e-01
6.16811574e-01 -2.03699231e-01 -1.57115519e+00 -4.41313654e-01
1.67103708e-01 -7.05053747e-01 3.92922878e-01 -7.51148403e-01
-1.19083869e+00 1.23607206e+00 5.53324997e-01 -1.70230076e-01
7.86692441e-01 -1.65835187e-01 6.89831018e-01 4.56238955e-01
7.08681285e-01 -1.10200942e+00 -5.63941061e-01 4.49779183e-01
6.50577903e-01 -1.60138047e+00 3.19800228e-01 -6.42284334e-01
-4.14437085e-01 9.72488880e-01 5.53112864e-01 -1.39373645e-01
1.06787324e+00 1.54465213e-01 1.98655963e-01 -1.27263159e-01
-4.38874930e-01 -7.51722753e-01 3.56107235e-01 4.92395014e-01
-1.47764176e-01 3.69610637e-01 -1.14504628e-01 -7.05828145e-02
-1.75119519e-01 -3.49146202e-02 3.07805300e-01 7.21549034e-01
-9.04368937e-01 -1.21196437e+00 -4.22890365e-01 4.61881340e-01
1.40933469e-01 -2.11536959e-01 -3.72029678e-03 9.03532445e-01
3.67142819e-02 9.48544383e-01 1.38794944e-01 -2.10080460e-01
3.81797403e-01 9.82521623e-02 1.19944252e-01 -5.89022398e-01
-1.48737635e-02 3.78488563e-02 1.41999632e-01 -6.19389176e-01
-6.38580859e-01 -8.29549193e-01 -1.13330865e+00 -4.67652887e-01
-4.48621899e-01 3.42082918e-01 1.12937129e+00 7.53797412e-01
2.26027161e-01 1.75087139e-01 3.75604838e-01 -1.05056000e+00
-6.43575549e-01 -8.54848623e-01 -7.30189443e-01 -4.98591095e-01
3.33736002e-01 -9.56660807e-01 -4.56701279e-01 -4.81849134e-01]
|
[8.114096641540527, -2.5367250442504883]
|
750f390a-7f53-4d03-8a7f-bc9d150b647b
|
sar-image-despeckling-using-a-convolutional
|
1706.00552
| null |
http://arxiv.org/abs/1706.00552v2
|
http://arxiv.org/pdf/1706.00552v2.pdf
|
SAR Image Despeckling Using a Convolutional Neural Network
|
Synthetic Aperture Radar (SAR) images are often contaminated by a
multiplicative noise known as speckle. Speckle makes the processing and
interpretation of SAR images difficult. We propose a deep learning-based
approach called, Image Despeckling Convolutional Neural Network (ID-CNN), for
automatically removing speckle from the input noisy images. In particular,
ID-CNN uses a set of convolutional layers along with batch normalization and
rectified linear unit (ReLU) activation function and a component-wise division
residual layer to estimate speckle and it is trained in an end-to-end fashion
using a combination of Euclidean loss and Total Variation (TV) loss. Extensive
experiments on synthetic and real SAR images show that the proposed method
achieves significant improvements over the state-of-the-art speckle reduction
methods.
|
['He Zhang', 'Vishal M. Patel', 'Puyang Wang']
|
2017-06-02
| null | null | null | null |
['sar-image-despeckling']
|
['computer-vision']
|
[ 7.34205246e-01 -2.99911439e-01 7.38259792e-01 -7.32685983e-01
-9.15957093e-01 -2.48211369e-01 3.29028517e-01 -7.84046412e-01
-5.49310923e-01 6.13090098e-01 2.12790221e-01 1.25800807e-03
-3.13113779e-01 -6.47291183e-01 -6.52603030e-01 -1.13066387e+00
1.51535235e-02 -5.91677204e-02 -9.75498408e-02 -9.99374762e-02
8.72974470e-03 6.74395204e-01 -1.11251152e+00 1.68620601e-01
6.46312177e-01 1.25047064e+00 -7.38796070e-02 6.16032600e-01
3.67342502e-01 1.18229222e+00 -4.67682719e-01 -1.01748928e-02
7.40746200e-01 -6.28866851e-01 -3.81667942e-01 4.20834690e-01
7.73141563e-01 -5.33831835e-01 -7.27384031e-01 1.35342729e+00
7.99680352e-01 2.53593445e-01 2.86707938e-01 -1.25544950e-01
-1.14395452e+00 3.94085050e-01 -9.75910366e-01 6.29997909e-01
-6.34983897e-01 3.65523368e-01 2.29145125e-01 -1.16023242e+00
5.05939543e-01 1.22343028e+00 1.04033101e+00 2.91369975e-01
-1.29013336e+00 -4.09735918e-01 -5.46420038e-01 -1.50086522e-01
-1.46934938e+00 -5.21594763e-01 1.01726639e+00 -1.98809490e-01
5.12424886e-01 7.99730644e-02 2.06920937e-01 5.76531291e-01
3.68313193e-01 3.47376585e-01 1.40945244e+00 -2.96813875e-01
1.15407519e-01 -7.16399729e-01 3.25286716e-01 4.51263040e-01
4.84113991e-01 4.33592230e-01 -2.94844024e-02 2.18644574e-01
8.17393184e-01 -1.27521111e-02 -2.95363873e-01 7.88617805e-02
-9.42309141e-01 7.29219377e-01 9.10419524e-01 2.74968117e-01
-7.42725790e-01 2.78357238e-01 1.72749519e-01 5.20725012e-01
8.45103145e-01 3.07266802e-01 -2.09386915e-01 6.70683086e-01
-1.24542892e+00 3.56171578e-01 2.92723894e-01 3.10224921e-01
7.91103363e-01 7.66304672e-01 -5.57085395e-01 7.58354545e-01
1.55759990e-01 1.11253810e+00 1.81582421e-02 -9.42128539e-01
2.13423029e-01 2.20763862e-01 2.06258103e-01 -1.18561447e+00
-4.57577348e-01 -1.06873238e+00 -1.54043639e+00 6.63284183e-01
4.19637039e-02 -5.57012081e-01 -1.54166615e+00 1.33659840e+00
1.41992737e-02 3.61170352e-01 3.90670985e-01 1.44010580e+00
7.85004973e-01 4.83428955e-01 -1.45262897e-01 -2.71131605e-01
1.02176714e+00 -8.33531976e-01 -9.02935922e-01 -4.96952236e-01
2.35743672e-01 -9.43930149e-01 5.22774816e-01 2.95458496e-01
-1.01920390e+00 -7.66120374e-01 -1.09728980e+00 -9.60293114e-02
7.07551017e-02 4.52291608e-01 2.96740949e-01 5.22631109e-01
-7.92891622e-01 7.12198615e-01 -9.79374588e-01 2.81889349e-01
1.05047166e+00 1.21743806e-01 -1.37868598e-01 -3.71193647e-01
-8.76770437e-01 6.42527938e-01 8.59849900e-02 1.06641185e+00
-8.27983797e-01 -6.36239111e-01 -9.58355725e-01 -5.94880478e-03
1.57302901e-01 -6.23134673e-01 8.07263732e-01 -1.23588288e+00
-1.39788914e+00 7.82042027e-01 1.99739397e-01 -6.08594477e-01
3.97612125e-01 -6.44545734e-01 -6.12048268e-01 1.34298161e-01
-1.72589123e-01 -9.05219764e-02 1.38854480e+00 -1.22018301e+00
1.66598093e-02 -4.93677765e-01 -5.01254976e-01 -1.62313402e-01
4.57017243e-01 2.56109148e-01 1.47278711e-01 -1.05311966e+00
3.54050785e-01 -6.34604394e-01 -5.72503090e-01 -8.27325732e-02
-4.62360173e-01 3.95913392e-01 1.36175907e+00 -9.54140425e-01
8.83358777e-01 -2.20118713e+00 -1.83321893e-01 1.86468557e-01
1.85672924e-01 8.04191232e-01 -5.10057271e-01 -1.76697284e-01
-3.74436796e-01 -3.26314420e-01 -9.61684525e-01 -1.45894110e-01
-4.42652404e-01 -1.76801473e-01 -3.36084694e-01 8.11355293e-01
6.48371994e-01 9.53784704e-01 -6.27873600e-01 6.44247979e-02
4.48383540e-01 8.49966228e-01 -1.29243866e-01 2.50391006e-01
-1.66219696e-01 7.87648320e-01 -3.47536922e-01 6.82007134e-01
1.44007587e+00 -1.95216328e-01 -1.43233091e-01 -6.82266831e-01
-2.22872689e-01 -4.74694431e-01 -1.12031722e+00 1.33657253e+00
-1.65753901e-01 8.53139520e-01 3.18633318e-01 -9.28215444e-01
1.27609646e+00 -2.14807063e-01 -4.51628817e-03 -9.21886861e-01
5.33096790e-01 1.34503961e-01 6.29036874e-02 -6.27709329e-01
2.80577153e-01 -2.40098596e-01 1.84838459e-01 3.64940703e-01
-2.28830621e-01 6.29122183e-02 -1.05565041e-01 -2.63995618e-01
1.37080634e+00 -7.34708607e-02 4.42528352e-02 -2.49339759e-01
7.82649875e-01 3.83552276e-02 7.37426281e-01 7.93164611e-01
9.34948307e-03 9.59791720e-01 -9.30225104e-02 -9.46869493e-01
-1.32556820e+00 -1.09807539e+00 -5.96662760e-02 4.46814179e-01
-7.46682286e-02 6.64414704e-01 -7.40516722e-01 -4.09855872e-01
-3.34511667e-01 4.27058220e-01 -6.14853859e-01 7.14483559e-02
-1.26396191e+00 -1.16618228e+00 4.70061809e-01 4.92981195e-01
1.28001916e+00 -1.11817646e+00 -5.78363121e-01 2.32483029e-01
5.02467304e-02 -1.28291345e+00 -3.45406175e-01 2.10201919e-01
-8.48780811e-01 -1.02021587e+00 -6.35607421e-01 -7.26434708e-01
8.89484584e-01 5.52237451e-01 9.88941550e-01 -3.68332528e-02
-7.08507538e-01 -3.21703970e-01 -3.14813852e-01 -2.60314167e-01
-2.08067652e-02 -4.15994257e-01 -2.81564593e-01 4.95807528e-01
4.19826716e-01 -6.16361260e-01 -1.00128853e+00 -1.47488132e-01
-1.31804681e+00 -4.82413322e-01 1.21088600e+00 1.23613894e+00
7.73431063e-01 2.71777153e-01 1.99087888e-01 -1.09428906e+00
3.73975098e-01 -3.32343951e-02 -9.25630212e-01 -4.98155057e-02
-2.76085705e-01 4.75614630e-02 7.19465792e-01 5.47444485e-02
-1.40260029e+00 2.28376359e-01 4.30142358e-02 -7.58239925e-01
-2.12294117e-01 5.80133617e-01 9.31772515e-02 -4.21374857e-01
8.99539709e-01 3.88284743e-01 -1.55238668e-02 -6.09900713e-01
3.21124375e-01 7.30569422e-01 9.44927514e-01 4.03566658e-01
1.48061144e+00 8.18911612e-01 2.28783548e-01 -1.11818051e+00
-1.45989156e+00 -1.49398953e-01 -7.33614683e-01 -8.40526521e-02
1.03409863e+00 -1.15622759e+00 -2.82453567e-01 9.23510134e-01
-9.61525202e-01 -4.00580436e-01 -2.64465481e-01 6.27176940e-01
-2.06184238e-01 3.17577153e-01 -3.90027136e-01 -8.10264528e-01
-9.76941049e-01 -7.01148629e-01 1.05651295e+00 4.98260498e-01
5.17064452e-01 -5.89108348e-01 -7.83665255e-02 4.61897492e-01
1.01602101e+00 4.89471227e-01 6.45631313e-01 -2.14214101e-01
-8.36775661e-01 -1.62341416e-01 -8.32287550e-01 1.08090234e+00
1.99904405e-02 -4.47067708e-01 -8.95214081e-01 -5.28243661e-01
3.95149976e-01 -3.13931376e-01 1.45676279e+00 9.92160678e-01
1.12200749e+00 -2.91882545e-01 3.02852929e-01 1.06564784e+00
2.06296420e+00 -4.53814445e-03 1.28306103e+00 3.09601128e-01
5.72482288e-01 1.25141874e-01 3.83949339e-01 3.03325474e-01
-4.68591630e-01 1.49736896e-01 3.14340711e-01 -5.74782252e-01
-5.30176640e-01 6.57325923e-01 9.96349752e-03 4.63220686e-01
-1.39744475e-01 -1.38484791e-01 -6.32275164e-01 4.54025716e-01
-1.51155066e+00 -1.26692545e+00 -3.09604824e-01 1.77688110e+00
5.64983606e-01 -5.88671453e-02 -6.89260423e-01 4.97794561e-02
6.64030433e-01 7.81803370e-01 -5.90818167e-01 -5.76916225e-02
-6.85923994e-01 6.92409217e-01 9.37076449e-01 4.94999379e-01
-1.61436200e+00 9.53824580e-01 6.27296782e+00 6.30052507e-01
-1.34446549e+00 9.41391811e-02 8.24749768e-01 3.53373885e-01
3.10773939e-01 -3.34174603e-01 -2.80654818e-01 2.61389285e-01
5.85567474e-01 4.63151157e-01 4.51427907e-01 4.88242865e-01
5.55389345e-01 -5.49139529e-02 -1.52038455e-01 8.67247164e-01
3.09489936e-01 -1.34738970e+00 -5.23971021e-02 -4.51792806e-01
9.78409410e-01 2.74455458e-01 7.98080862e-02 2.79038604e-02
3.38784784e-01 -1.10037386e+00 3.41868430e-01 1.06887472e+00
7.49833405e-01 -8.53950858e-01 1.31536961e+00 -2.13056654e-01
-8.02683234e-01 -1.74379602e-01 -6.54496789e-01 -1.23564340e-01
2.55368173e-01 1.18509126e+00 -3.87189239e-02 6.68117464e-01
7.79070914e-01 5.46065032e-01 -4.59176242e-01 8.07616532e-01
-3.21514785e-01 7.17746198e-01 2.44738273e-02 6.55932963e-01
3.60080391e-01 -5.10119617e-01 7.62516677e-01 1.17960405e+00
3.12321126e-01 6.37236118e-01 9.49612062e-04 1.09069490e+00
-1.99920759e-01 -6.57978415e-01 -3.69698346e-01 -6.62955418e-02
9.65246707e-02 1.55419374e+00 -5.55953503e-01 -6.25167862e-02
-2.83591747e-01 8.48003626e-01 -1.73356324e-01 5.39173126e-01
-5.06957829e-01 -8.33061278e-01 4.83638197e-01 -8.65048096e-02
9.58886981e-01 -5.23547053e-01 -5.14257371e-01 -8.11964750e-01
1.35478124e-01 -6.70946360e-01 -6.26944378e-02 -1.05917573e+00
-1.59176052e+00 1.00029182e+00 -4.55287904e-01 -1.37964606e+00
3.16183865e-01 -5.84133923e-01 -7.50282228e-01 1.08317554e+00
-1.93043876e+00 -1.38160753e+00 -8.91294241e-01 4.36779231e-01
4.77128208e-01 -2.64871061e-01 3.40630174e-01 3.37817937e-01
-4.92905825e-01 1.16501510e-01 3.46730620e-01 7.28262126e-01
4.81605858e-01 -8.68782580e-01 5.35614133e-01 1.38905966e+00
-4.60957617e-01 2.07343131e-01 8.84212732e-01 -6.87253118e-01
-1.33212245e+00 -1.95591378e+00 6.57760859e-01 2.64077842e-01
4.29143637e-01 1.29996300e-01 -9.81246531e-01 6.97985172e-01
4.29275721e-01 5.75817704e-01 2.59280473e-01 -5.91796696e-01
-2.36175835e-01 -3.92845005e-01 -1.11908257e+00 1.87880889e-01
8.80479157e-01 -7.93192163e-02 -6.16961539e-01 2.47092664e-01
4.98234898e-01 -3.63962710e-01 -5.09469271e-01 8.12626243e-01
3.90389115e-02 -8.78197908e-01 1.08839917e+00 -3.95329148e-01
6.66163981e-01 -7.12787390e-01 -2.82153487e-01 -1.26855314e+00
-6.76555514e-01 -5.39266050e-01 1.12176493e-01 7.71794319e-01
9.92001370e-02 -2.62092352e-01 6.91827297e-01 -4.70374413e-02
-4.33645755e-01 -4.45173532e-01 -6.84875667e-01 -8.89330328e-01
-2.04950303e-01 2.70473361e-02 2.83408523e-01 8.82855058e-01
-1.38669372e+00 4.02595073e-01 -8.09310913e-01 6.76907241e-01
1.39961934e+00 1.88359544e-01 5.68679273e-01 -1.11624432e+00
9.19899419e-02 9.47827101e-02 -4.57181513e-01 -7.96022058e-01
-1.49183303e-01 -4.56907719e-01 4.14078295e-01 -1.35145032e+00
-2.40648258e-02 -4.70286980e-02 -2.23205864e-01 2.02636808e-01
-1.28457919e-01 9.18304861e-01 -7.02274591e-02 3.10272992e-01
-3.98008555e-01 6.61342442e-01 1.34995759e+00 -3.47670496e-01
-1.64258659e-01 -1.18894748e-01 -4.96233523e-01 8.43273759e-01
7.75940776e-01 -6.44238710e-01 2.86737323e-01 -1.00970948e+00
7.75217935e-02 -5.13308756e-02 7.48779416e-01 -1.28250635e+00
3.99705052e-01 9.48486570e-03 8.01169813e-01 -8.60679507e-01
1.48817509e-01 -6.19291842e-01 5.56388237e-02 4.68927652e-01
-2.02062935e-01 -2.62407303e-01 -7.54249692e-02 6.43622994e-01
-4.64195102e-01 6.06919751e-02 1.42069912e+00 -1.23996489e-01
-6.89666390e-01 3.05435002e-01 -2.87560225e-01 -1.35418810e-02
6.11161351e-01 -9.19845849e-02 -5.93287349e-01 -6.00819811e-02
-6.79827929e-01 -1.90082699e-01 -1.56027749e-01 -1.15830332e-01
8.12598169e-01 -1.25038910e+00 -1.30704343e+00 5.91711342e-01
-3.06819856e-01 4.77487268e-03 8.01346421e-01 1.06055319e+00
-1.01052463e+00 1.20070335e-02 -3.16416442e-01 -3.39801550e-01
-1.17130530e+00 1.38300136e-01 5.36125898e-01 -3.28162223e-01
-9.95159149e-01 8.54241133e-01 -1.17812745e-01 -3.59113961e-01
-1.27737477e-01 -1.37662217e-01 -1.85597483e-02 -5.31254888e-01
9.94437456e-01 4.42812592e-01 1.83249280e-01 -7.62555242e-01
-2.06981003e-01 7.44905889e-01 -7.92375430e-02 2.27580130e-01
1.84181869e+00 -1.97533473e-01 -3.75003040e-01 -2.21441567e-01
1.08649826e+00 -2.02836007e-01 -1.33483863e+00 -8.40744913e-01
-6.82547614e-02 -6.84920371e-01 6.76670969e-01 -9.11154985e-01
-1.53243744e+00 6.46610618e-01 1.17807436e+00 -1.89514980e-01
1.66707420e+00 -4.25299525e-01 1.03117096e+00 7.83176959e-01
-3.30456138e-01 -9.90112960e-01 9.45565253e-02 8.75069320e-01
8.65409434e-01 -1.43627059e+00 2.58859873e-01 -2.23806158e-01
-3.27491909e-01 1.14708507e+00 2.35822737e-01 -9.06302989e-01
9.64615643e-01 5.48320115e-01 6.43645167e-01 -5.40139973e-01
-2.08634272e-01 -4.53608960e-01 1.64861172e-01 4.12356973e-01
2.16558024e-01 -1.66769132e-01 -3.54016334e-01 4.39843625e-01
1.56951740e-01 2.09243879e-01 4.54693973e-01 9.50455308e-01
-8.09544504e-01 -5.96208632e-01 -6.31538868e-01 5.93272328e-01
-6.27970338e-01 -2.63394982e-01 -8.09612870e-02 5.35667002e-01
1.91321239e-01 9.41455603e-01 2.85309434e-01 -3.19146782e-01
5.11965573e-01 -4.63873029e-01 2.93359578e-01 -3.25559556e-01
-5.76165676e-01 2.43318334e-01 -2.38310456e-01 -4.64435130e-01
-8.35142672e-01 -2.26693392e-01 -8.04755509e-01 -2.83195406e-01
-3.20400149e-01 -2.75130153e-01 4.07617450e-01 1.06689274e+00
3.42571199e-01 8.74304116e-01 9.55024481e-01 -1.04339147e+00
-7.90086031e-01 -1.21548319e+00 -8.83370757e-01 4.21476811e-01
7.20808148e-01 -2.31035247e-01 -5.11632681e-01 4.12619084e-01]
|
[10.488554954528809, -2.266045570373535]
|
713eca9b-7ba6-4df1-9693-f9a27d924532
|
beyond-co2-emissions-the-overlooked-impact-of
|
2306.16668
| null |
https://arxiv.org/abs/2306.16668v1
|
https://arxiv.org/pdf/2306.16668v1.pdf
|
Beyond CO2 Emissions: The Overlooked Impact of Water Consumption of Information Retrieval Models
|
As in other fields of artificial intelligence, the information retrieval community has grown interested in investigating the power consumption associated with neural models, particularly models of search. This interest has become particularly relevant as the energy consumption of information retrieval models has risen with new neural models based on large language models, leading to an associated increase of CO2 emissions, albeit relatively low compared to fields such as natural language processing.
|
['Shengyao Zhuang', 'Harrisen Scells', 'Guido Zuccon']
|
2023-06-29
| null | null | null | null |
['retrieval', 'information-retrieval']
|
['methodology', 'natural-language-processing']
|
[ 1.76260293e-01 5.50569668e-02 -5.01707971e-01 1.43959597e-01
-3.56505632e-01 -4.57305640e-01 9.22895730e-01 6.90699637e-01
-7.89382577e-01 4.85807538e-01 3.04394979e-02 -5.35616577e-01
-5.08983076e-01 -9.02771115e-01 -3.76537442e-01 -4.29911375e-01
-3.00374806e-01 2.90460527e-01 -7.73513690e-02 7.27799833e-02
4.67556745e-01 7.43467212e-01 -1.64622700e+00 -4.80532050e-01
6.06914222e-01 1.19265068e+00 3.21926743e-01 2.66073942e-01
-1.78455487e-01 6.60049140e-01 -4.38175380e-01 -1.74781293e-01
1.00168809e-02 -1.90840334e-01 -9.48358417e-01 -6.89425528e-01
-9.97570902e-02 1.76621780e-01 -5.74696660e-01 1.32050371e+00
3.59124869e-01 6.76639438e-01 5.46110392e-01 -8.84777427e-01
-5.06422639e-01 3.11616778e-01 -3.75995427e-01 4.84111369e-01
-1.08776949e-01 -8.32873210e-02 9.88742232e-01 -5.56241453e-01
4.91877675e-01 1.33542490e+00 3.57806504e-01 3.38333547e-01
-9.47763920e-01 -4.37872529e-01 -3.92849185e-02 5.01726389e-01
-1.50383103e+00 -4.34857696e-01 7.73325920e-01 1.61405981e-01
1.81311822e+00 3.48133922e-01 8.68049502e-01 6.35912657e-01
5.50184369e-01 4.28087592e-01 6.72934949e-01 -8.98825943e-01
4.83210862e-01 4.84211057e-01 1.98800638e-01 5.04174113e-01
7.14458108e-01 3.05446804e-01 -5.00867188e-01 -3.84094529e-02
2.52895772e-01 -1.63752750e-01 1.76773667e-01 2.42402494e-01
-7.65966415e-01 9.94799852e-01 4.77536887e-01 7.38961041e-01
-5.88771462e-01 4.65434402e-01 4.37252998e-01 -9.92360264e-02
5.83842039e-01 8.81429434e-01 -1.67907327e-01 -3.16144019e-01
-1.05067742e+00 2.87195712e-01 8.56115639e-01 7.64235318e-01
5.50016642e-01 1.07686028e-01 3.31442833e-01 1.01114810e+00
3.47970009e-01 5.27602494e-01 7.15806484e-01 -1.04142582e+00
2.96653926e-01 8.06219220e-01 -2.37791374e-01 -1.09591162e+00
-6.16913617e-01 -4.46991473e-01 -1.09017277e+00 -3.19393098e-01
-2.92796582e-01 9.09309983e-02 -8.45630705e-01 1.43012524e+00
-2.15254039e-01 -4.94261473e-01 3.95859070e-02 5.22702157e-01
3.66726756e-01 1.09495664e+00 4.48107481e-01 -3.31090361e-01
1.38244605e+00 -9.37680483e-01 -8.64030600e-01 -7.16933548e-01
4.92028594e-01 -2.77644485e-01 3.89424413e-01 4.54498112e-01
-1.15826344e+00 -3.26940864e-01 -9.96030450e-01 -3.31479758e-01
-1.03009009e+00 -6.49538696e-01 1.09422374e+00 8.39852154e-01
-1.42072546e+00 3.12671930e-01 -7.98193097e-01 -8.54991972e-01
4.32318598e-01 4.67654854e-01 2.88256556e-01 -1.88719600e-01
-1.45642602e+00 1.54143178e+00 6.46230578e-01 3.82898748e-02
-4.88834858e-01 -4.21487689e-01 -7.90562928e-01 3.82336169e-01
3.08713317e-01 -5.63600302e-01 1.17224944e+00 -3.24154437e-01
-9.56554055e-01 3.82825345e-01 -3.72065574e-01 -6.75286472e-01
-2.32280597e-01 2.46396661e-01 -8.14987004e-01 7.19266385e-02
-4.82571274e-01 9.15314972e-01 2.72118360e-01 -8.84713829e-01
-4.04719740e-01 -4.43645328e-01 3.48948538e-02 3.03031474e-01
-8.72262478e-01 4.95008200e-01 -4.80010420e-01 -3.13176244e-01
-2.94366181e-01 -1.01222920e+00 -6.31241798e-01 2.65448261e-02
1.84358750e-02 -6.99348330e-01 9.44329143e-01 -4.65327233e-01
1.56930292e+00 -1.70053625e+00 -9.40846726e-02 3.95878583e-01
1.53747633e-01 1.99597210e-01 8.06366727e-02 3.09155464e-01
2.43815109e-01 7.97489882e-01 8.66162330e-02 2.37101659e-01
-6.87235221e-02 2.70778924e-01 -9.69898552e-02 3.11660860e-02
3.20512712e-01 1.00621414e+00 -8.83477271e-01 -4.79722202e-01
1.41174674e-01 5.36806762e-01 -2.71554112e-01 -1.19427986e-01
-9.27877724e-02 -5.55345595e-01 -6.32867754e-01 7.02652395e-01
-1.84765793e-02 -6.45517349e-01 -1.40808662e-02 2.86274910e-01
-3.44861805e-01 6.02650940e-01 -5.71735978e-01 1.79718208e+00
-6.98819041e-01 9.54694569e-01 -1.75760761e-02 -1.07341385e+00
5.39615214e-01 3.65837038e-01 6.81047499e-01 -1.24861407e+00
2.89612204e-01 1.09218411e-01 2.27973178e-01 -4.35252160e-01
9.53152716e-01 -4.69853505e-02 -3.76938619e-02 4.40914214e-01
-1.31552428e-01 -3.29797804e-01 5.11928320e-01 2.90298611e-01
1.06937695e+00 -1.72453731e-01 1.52117133e-01 -5.59743047e-01
8.55466500e-02 2.07611978e-01 -1.39297798e-01 5.07730722e-01
5.55463471e-02 -5.03065623e-02 -2.78898537e-01 -2.29271010e-01
-8.55957210e-01 -4.94901121e-01 -3.49937499e-01 5.95897257e-01
4.29136306e-01 -5.17818749e-01 -8.42381299e-01 1.33331209e-01
-2.35762373e-01 8.64825428e-01 -3.84363264e-01 -5.21143138e-01
-3.67353797e-01 -9.76183295e-01 5.37561893e-01 4.14794981e-01
6.18196845e-01 -1.13973296e+00 -1.25502431e+00 2.29214922e-01
-1.48615604e-02 -9.23921347e-01 -4.54819575e-02 9.82506454e-01
-1.28839839e+00 -5.15367210e-01 -4.32700604e-01 -4.82071102e-01
3.49531889e-01 3.46596204e-02 1.43168128e+00 2.82984376e-01
-8.28794301e-01 5.39328098e-01 -1.79651797e-01 -1.06993723e+00
-4.07388002e-01 4.81272101e-01 4.97066826e-02 -7.74149060e-01
7.72321701e-01 -1.73696473e-01 -8.46618295e-01 -2.30885118e-01
-1.29101837e+00 -2.61196971e-01 6.17308319e-01 1.82914257e-01
3.74430358e-01 6.38226569e-01 6.65775478e-01 -5.19406378e-01
1.15186977e+00 -8.10131013e-01 -7.13197410e-01 1.33951306e-01
-1.78018081e+00 2.84892708e-01 2.45690003e-01 -2.36100227e-01
-9.12703693e-01 -4.70516533e-01 1.11102879e-01 -1.12775434e-02
-2.66414672e-01 9.25820231e-01 1.52279273e-01 -6.73781857e-02
6.55837893e-01 2.74985194e-01 -2.06315681e-01 -3.77237946e-01
1.81104280e-02 7.06213415e-01 1.13237731e-01 -7.49864578e-02
8.49604666e-01 -1.30214944e-01 2.87398964e-01 -1.27818465e+00
-6.15088604e-02 -3.81783038e-01 -5.75812533e-02 -2.91915059e-01
9.58220601e-01 -5.52978992e-01 -5.78220487e-01 2.42425561e-01
-1.24431682e+00 5.31528629e-02 -3.31423998e-01 2.25410879e-01
-7.94376656e-02 4.39993516e-02 -5.43020189e-01 -1.18791783e+00
-5.44580698e-01 -9.66047764e-01 5.24850190e-01 5.53786099e-01
-5.90118408e-01 -1.29968870e+00 6.82541505e-02 1.67625308e-01
1.03844857e+00 -2.98207700e-01 1.53315032e+00 -6.89021289e-01
-7.23051310e-01 -7.02998042e-01 -3.64973880e-02 6.24565966e-02
1.03821494e-02 -3.61934483e-01 -1.03294671e+00 -2.99674906e-02
7.57596940e-02 -2.72797674e-01 7.04451382e-01 2.91915715e-01
1.39432728e+00 -3.93965274e-01 -9.49331105e-01 -2.02631593e-01
1.74063563e+00 9.29073393e-01 4.41415578e-01 4.45596009e-01
1.79879703e-02 3.85155380e-01 -2.62364931e-03 -1.18704371e-01
2.67393738e-01 6.06362164e-01 3.90589237e-01 -6.54430240e-02
1.15509130e-01 -2.20117807e-01 -7.68807232e-02 9.54906523e-01
-1.42103776e-01 -7.41460562e-01 -1.10339510e+00 6.81634605e-01
-1.43052781e+00 -8.79576862e-01 3.59914362e-01 2.09562325e+00
6.28139973e-01 1.79026261e-01 -3.12169641e-01 3.66339564e-01
3.09410781e-01 3.77724767e-01 -9.51118767e-01 -8.34847450e-01
4.51644920e-02 4.94040310e-01 5.48231363e-01 1.45903185e-01
-5.45969725e-01 4.07378405e-01 7.79229641e+00 8.35366964e-01
-9.91021156e-01 -8.91892388e-02 8.61841917e-01 -3.04981440e-01
-4.45405751e-01 -3.76981139e-01 -5.79938531e-01 5.32211781e-01
2.06891370e+00 -7.13147998e-01 9.46056306e-01 6.09390378e-01
5.29615819e-01 -6.43286347e-01 -1.02681291e+00 8.50124598e-01
6.86206594e-02 -1.24968255e+00 1.48118451e-01 5.79671323e-01
4.31191236e-01 3.08157355e-01 -1.76383644e-01 2.92254180e-01
1.10838592e-01 -1.26230311e+00 3.56771439e-01 3.29232067e-01
5.45797050e-01 -9.32306767e-01 4.87630725e-01 7.36364782e-01
-1.01805139e+00 -2.53768235e-01 -3.31304580e-01 -3.02354366e-01
1.40239909e-01 5.04031599e-01 -4.90532100e-01 3.87532681e-01
9.50840235e-01 1.55334398e-01 -4.76580113e-01 1.11978710e+00
3.23085308e-01 7.34310269e-01 -6.41290963e-01 -7.09900379e-01
2.53003836e-01 -2.71789044e-01 1.02864221e-01 1.04059708e+00
4.75026131e-01 1.57688335e-01 -3.33584726e-01 8.21151197e-01
-5.16856134e-01 1.60073146e-01 -8.34820986e-01 -6.49249613e-01
6.76856041e-01 1.20455313e+00 -8.71890426e-01 -3.08208346e-01
-4.72049385e-01 6.88934207e-01 8.16530362e-02 3.73783827e-01
-6.71829581e-01 -4.98989254e-01 2.04237685e-01 1.96747258e-01
-3.32666039e-01 -2.06506565e-01 -1.70291916e-01 -5.97468793e-01
-7.02329576e-02 -6.46506548e-01 6.19171523e-02 -8.59437346e-01
-6.59740686e-01 6.18058801e-01 4.23637718e-01 -5.62352657e-01
-6.48235083e-01 -4.53079462e-01 -3.38187039e-01 9.44975317e-01
-1.60773158e+00 -3.81416142e-01 7.17664212e-02 -1.38171380e-02
6.33092403e-01 1.07062966e-01 1.25342488e+00 3.81983906e-01
-4.29863542e-01 3.06021810e-01 1.73217401e-01 -4.42141503e-01
1.11348853e-02 -7.02509403e-01 3.42147410e-01 6.98990643e-01
2.61748880e-01 7.34147131e-01 5.45605481e-01 -1.73769772e-01
-1.88205218e+00 -8.40319991e-01 1.02631247e+00 -4.73113745e-01
5.24345040e-01 -1.40381619e-01 -6.49009705e-01 2.06858978e-01
6.33199096e-01 -3.25894117e-01 4.92338687e-01 -5.96461110e-02
3.18929069e-02 -2.03928515e-01 -1.07481897e+00 3.61767650e-01
9.45405543e-01 -6.15791202e-01 -1.92285344e-01 3.13277632e-01
5.12748897e-01 1.64815843e-01 -8.81839395e-01 4.30949092e-01
3.73817831e-01 -2.70184964e-01 8.91080618e-01 -4.74050820e-01
3.88535202e-01 2.36683458e-01 -7.91876670e-03 -1.21807611e+00
-4.96275336e-01 -2.28656530e-01 -4.71566617e-01 8.17962110e-01
7.29387999e-01 -4.91465211e-01 8.09217751e-01 1.45413816e+00
2.46265277e-01 -8.06798458e-01 -9.86145794e-01 -9.75899220e-01
2.69030392e-01 -5.29422045e-01 1.23357385e-01 5.66742063e-01
4.74857479e-01 6.53759062e-01 4.06173617e-02 -4.00301158e-01
3.47382158e-01 -1.33914307e-01 -2.01672077e-01 -1.80682862e+00
2.72552902e-03 -9.50587153e-01 -2.17765883e-01 -8.54630291e-01
-4.90419753e-02 -1.01304066e+00 5.19631058e-02 -1.80532002e+00
5.65728664e-01 -2.29850084e-01 -7.13859916e-01 4.47738141e-01
1.25502929e-01 4.02274907e-01 6.06364980e-02 9.45355669e-02
-5.66131473e-01 2.85897821e-01 5.81986129e-01 -5.92379093e-01
6.37822971e-02 -1.78748116e-01 -9.50430334e-01 5.37920713e-01
8.77515554e-01 -6.05438471e-01 -5.55510223e-01 -4.17889774e-01
9.75728333e-01 -2.08819374e-01 -9.97291058e-02 -1.23178518e+00
4.83372450e-01 -3.67035270e-02 3.23414981e-01 -2.87696034e-01
4.47667897e-01 -1.43872106e+00 3.10590148e-01 7.59794056e-01
-5.27987897e-01 3.86999726e-01 6.63198948e-01 5.97587943e-01
-2.99421668e-01 -5.48241735e-01 3.58818620e-01 -3.81959081e-01
-5.62265694e-01 8.82186666e-02 -1.03208005e+00 -3.47317696e-01
8.58150423e-01 -3.87188494e-01 -3.69070411e-01 -3.16491991e-01
-1.78432882e-01 2.18317106e-01 1.46219119e-01 8.99658203e-01
2.95588851e-01 -1.06998158e+00 1.52141094e-01 -2.87201554e-01
-3.75532322e-02 6.29977584e-02 -1.95615485e-01 3.98023963e-01
-1.67174280e-01 1.29603314e+00 3.27049166e-01 -5.63598098e-03
-9.10648286e-01 8.06757808e-01 2.98928618e-01 -5.63284695e-01
-1.57787383e-01 5.73392093e-01 -4.15193111e-01 2.39013016e-01
3.81065607e-01 -3.09543759e-01 -1.71091616e-01 2.50802338e-02
3.05670857e-01 8.08116555e-01 4.38891232e-01 -9.74397436e-02
-5.75947940e-01 1.47665948e-01 5.80483302e-02 1.05353676e-01
1.35978985e+00 1.96793571e-01 -3.49666178e-01 7.14424014e-01
1.19908595e+00 -4.87105101e-01 -2.20591038e-01 6.30215649e-03
4.33008909e-01 2.10135236e-01 7.55096078e-01 -1.19520724e+00
-9.18025911e-01 9.02953923e-01 9.67744410e-01 6.82649195e-01
1.28395081e+00 -6.97366595e-02 6.63980961e-01 9.63549197e-01
6.95841730e-01 -1.44668579e+00 -3.76748323e-01 5.63360691e-01
4.47670072e-01 -1.08830523e+00 3.03810626e-01 2.01500297e-01
3.06711972e-01 8.00262570e-01 2.44735554e-01 3.23945522e-01
9.65613484e-01 5.49434602e-01 -3.33771765e-01 -4.88799393e-01
-9.15527701e-01 1.41568244e-01 1.78114951e-01 2.71937698e-01
6.18118882e-01 -2.81071097e-01 -5.67021608e-01 2.27396205e-01
9.07009766e-02 4.15108144e-01 2.38161117e-01 9.18421984e-01
-5.49349725e-01 -1.13796806e+00 -1.43815264e-01 8.79215121e-01
-5.48542142e-01 -5.43234289e-01 -4.86966819e-01 6.30987823e-01
-3.32161337e-01 1.27730393e+00 2.41984963e-01 -2.00543210e-01
1.24198690e-01 5.09627938e-01 1.17771640e-01 -4.69379991e-01
-6.76926672e-01 -1.60443038e-01 2.71262497e-01 -4.41932291e-01
-4.40845162e-01 -3.58493805e-01 -1.32347953e+00 -1.91783413e-01
-5.59685886e-01 5.21907926e-01 1.23838651e+00 8.44400644e-01
4.51417714e-01 5.97319305e-01 2.37039536e-01 -9.91762340e-01
-3.58189285e-01 -1.16888285e+00 -4.26540613e-01 -1.39783949e-01
-1.05960190e-01 -2.94442087e-01 -3.47443044e-01 -1.94366768e-01]
|
[11.034448623657227, 7.672529697418213]
|
4ab956e4-e4ca-450b-9594-0f583d1aa75f
|
self-interpretable-time-series-prediction
|
2306.06024
| null |
https://arxiv.org/abs/2306.06024v3
|
https://arxiv.org/pdf/2306.06024v3.pdf
|
Self-Interpretable Time Series Prediction with Counterfactual Explanations
|
Interpretable time series prediction is crucial for safety-critical areas such as healthcare and autonomous driving. Most existing methods focus on interpreting predictions by assigning important scores to segments of time series. In this paper, we take a different and more challenging route and aim at developing a self-interpretable model, dubbed Counterfactual Time Series (CounTS), which generates counterfactual and actionable explanations for time series predictions. Specifically, we formalize the problem of time series counterfactual explanations, establish associated evaluation protocols, and propose a variational Bayesian deep learning model equipped with counterfactual inference capability of time series abduction, action, and prediction. Compared with state-of-the-art baselines, our self-interpretable model can generate better counterfactual explanations while maintaining comparable prediction accuracy.
|
['Hao Wang', 'Jingquan Yan']
|
2023-06-09
| null | null | null | null |
['counterfactual-inference', 'time-series-prediction']
|
['miscellaneous', 'time-series']
|
[ 5.96061826e-01 8.64182353e-01 -5.92203856e-01 -7.34335482e-01
-6.29761219e-01 -2.07043350e-01 9.96359646e-01 5.49703697e-03
6.49607033e-02 1.13625479e+00 8.93970191e-01 -8.73108625e-01
-3.26434970e-01 -6.70665920e-01 -9.01474476e-01 -2.58310109e-01
-2.98106492e-01 6.27622724e-01 -3.49287838e-01 -3.81074362e-02
2.82192588e-01 2.60067750e-02 -1.53069735e+00 4.79252964e-01
9.66082156e-01 9.23687935e-01 -2.19965264e-01 5.28513730e-01
6.99569508e-02 9.64603901e-01 -4.82486039e-01 -6.79703236e-01
-4.10281755e-02 -3.06509703e-01 -6.26133084e-01 -2.99403995e-01
7.35832436e-04 -5.35860598e-01 -2.33731195e-01 7.56060004e-01
-5.09445779e-02 1.55808911e-01 8.69407356e-01 -1.97810102e+00
-7.11384356e-01 1.16529465e+00 -2.17188284e-01 2.13352814e-01
8.82146731e-02 5.51642001e-01 1.04688931e+00 -2.58024365e-01
1.63608104e-01 1.51702583e+00 7.63627052e-01 8.08807075e-01
-1.35508478e+00 -6.84804082e-01 5.13524711e-01 5.34852743e-01
-3.62957090e-01 -1.16607651e-01 1.01059687e+00 -2.92770237e-01
1.04182410e+00 7.40700126e-01 7.07312882e-01 1.85941446e+00
7.40761638e-01 1.00496864e+00 9.57958698e-01 9.53625590e-02
5.85089147e-01 -5.09927034e-01 9.48622376e-02 1.30584657e-01
3.07302535e-01 6.80518687e-01 -3.04782450e-01 -3.09150457e-01
4.25263286e-01 5.91414094e-01 -1.99703932e-01 -1.92829043e-01
-1.70510197e+00 1.04822612e+00 2.90355474e-01 -1.69941187e-01
-8.66193473e-01 6.33288980e-01 4.67220575e-01 2.64665708e-02
8.56453478e-01 5.98908663e-01 -8.79086196e-01 -1.78257331e-01
-6.82887018e-01 7.53502607e-01 6.00081503e-01 5.93995035e-01
9.95590612e-02 2.55771458e-01 -7.58137584e-01 1.12049595e-01
3.95603329e-01 7.09044695e-01 7.54660606e-01 -1.27749252e+00
4.67204392e-01 3.19485068e-01 4.12184179e-01 -6.66301370e-01
-4.50317472e-01 -3.50499839e-01 -8.29218209e-01 1.23287842e-04
3.42160687e-02 -4.06783134e-01 -9.15733337e-01 1.85734987e+00
1.56959325e-01 1.06094766e+00 2.90677130e-01 8.35833669e-01
3.80750239e-01 7.17142761e-01 3.99428308e-01 -7.03537941e-01
1.04337347e+00 -6.83383942e-01 -9.68794823e-01 -1.91155344e-01
4.47024822e-01 -1.92307621e-01 9.01122689e-01 1.35137916e-01
-7.50683188e-01 -5.19568503e-01 -9.19328809e-01 1.18395977e-01
-1.82747841e-01 -4.60643202e-01 1.08504975e+00 3.66874665e-01
-5.68365514e-01 7.92053699e-01 -1.18635166e+00 3.86979848e-01
4.93719339e-01 1.53215379e-01 4.15646285e-02 3.84345204e-01
-1.37900269e+00 8.34157705e-01 5.60290039e-01 -1.18011206e-01
-1.00911772e+00 -1.28645682e+00 -1.14459991e+00 1.35305017e-01
2.66009659e-01 -1.14574707e+00 1.84999454e+00 -8.22585523e-01
-1.49080527e+00 2.58332133e-01 -3.73962730e-01 -1.45555270e+00
9.50209260e-01 -2.98052877e-01 -7.13707328e-01 -4.14115071e-01
2.51465499e-01 7.34954298e-01 8.90090108e-01 -8.62712026e-01
-6.10945880e-01 -3.68772209e-01 1.43547922e-01 -1.31649926e-01
3.24100584e-01 -5.65809369e-01 5.03264248e-01 -7.62757063e-01
-2.76076704e-01 -8.04604411e-01 -6.19802892e-01 -3.22378069e-01
-6.02024198e-01 -5.50194144e-01 8.16597521e-01 -5.15852094e-01
1.31452799e+00 -1.75876880e+00 -2.54617900e-01 -2.43818656e-01
3.79165530e-01 -1.97329342e-01 -2.00249646e-02 7.27379620e-02
-5.35338819e-01 1.79454759e-01 -4.68876123e-01 -4.27221030e-01
4.04005647e-01 4.04373914e-01 -1.25581598e+00 1.76864907e-01
2.90641099e-01 1.18173420e+00 -1.32849324e+00 -2.25126728e-01
5.77472866e-01 1.81308556e-02 -7.05637217e-01 3.59356217e-02
-8.70623708e-01 4.26260084e-01 -5.30028880e-01 1.26010105e-01
5.71067989e-01 -2.75570899e-01 2.26973206e-01 5.20049818e-02
-6.07583411e-02 6.09067440e-01 -7.42405236e-01 1.27765930e+00
-4.96620357e-01 7.04675317e-01 -1.15624607e+00 -1.16374671e+00
5.58384478e-01 5.16254008e-01 4.13798064e-01 -5.92642188e-01
-3.60529753e-03 9.35399160e-02 -1.12687550e-01 -7.40124047e-01
4.78304982e-01 -6.93101764e-01 -3.68845016e-01 6.06922507e-01
-3.98638457e-01 -1.13938846e-01 -4.03776586e-01 -2.00425774e-01
8.09431970e-01 4.33249861e-01 7.67998457e-01 -7.62404467e-04
1.38218671e-01 1.19998045e-01 7.84003377e-01 8.06546867e-01
-2.54725277e-01 5.61461806e-01 6.31411910e-01 -9.61018622e-01
-1.05474329e+00 -1.43514776e+00 -4.04527265e-04 6.53586984e-01
-4.94995564e-02 -5.56663945e-02 -4.28104222e-01 -1.04890084e+00
3.15151900e-01 1.91708302e+00 -8.14298511e-01 -4.61644948e-01
-4.70270008e-01 -8.42436552e-01 2.90271193e-01 8.31586123e-01
4.20239061e-01 -9.85680461e-01 -9.40817118e-01 3.27980101e-01
-6.43676579e-01 -4.67231601e-01 -3.83922637e-01 -2.50399023e-01
-1.11989963e+00 -9.78712320e-01 -1.93175018e-01 3.44917655e-01
1.00889765e-01 7.94107392e-02 1.34809291e+00 -5.60169220e-01
2.24284679e-01 1.17203504e-01 1.17048986e-01 -1.45791173e+00
-4.67575222e-01 -5.16303957e-01 2.07369134e-01 1.09172404e-01
5.86775780e-01 -6.44697964e-01 -8.84369373e-01 -8.91495571e-02
-8.71033013e-01 4.04918849e-01 3.20938021e-01 8.39183152e-01
7.21544027e-01 -2.93095082e-01 9.63079333e-01 -9.86235917e-01
7.20675707e-01 -9.32115316e-01 -4.73256886e-01 1.27275020e-01
-8.84962142e-01 4.79467392e-01 9.16928291e-01 -4.65161294e-01
-1.45568168e+00 -2.63695568e-01 1.26884682e-02 -4.64854121e-01
-2.76841134e-01 4.65828001e-01 4.46186922e-02 1.17006767e+00
5.91798902e-01 1.88265055e-01 4.60396670e-02 -1.68184206e-01
5.63471794e-01 7.32122302e-01 6.13866925e-01 -1.82670519e-01
3.74983698e-01 9.09753919e-01 4.19778116e-02 -1.25472993e-01
-1.18039083e+00 -1.73718408e-02 -1.41060323e-01 -1.62480071e-01
7.56594956e-01 -8.94271672e-01 -1.02700269e+00 -3.57290842e-02
-1.41803908e+00 -3.75785977e-01 -6.32219732e-01 7.93331981e-01
-1.20349312e+00 4.38982025e-02 1.44126797e-02 -1.17410231e+00
-2.06440955e-01 -7.38942146e-01 1.40262365e+00 -1.95997894e-01
-9.43116903e-01 -1.24834204e+00 2.47717187e-01 4.94701475e-01
1.65645659e-01 9.69536781e-01 8.48183632e-01 -7.13997424e-01
-5.99066436e-01 -5.06670177e-02 1.37249334e-02 -2.86210775e-02
-4.13693041e-02 -2.37679303e-01 -1.09985030e+00 2.33130902e-01
1.90320566e-01 1.62716389e-01 9.88717973e-01 8.42832625e-01
1.63793457e+00 -1.12574875e+00 -5.91126859e-01 3.41860473e-01
8.79253089e-01 3.47725242e-01 5.89081526e-01 1.12752914e-01
3.30447972e-01 6.85250044e-01 9.32833791e-01 6.89927280e-01
8.17733765e-01 2.74465919e-01 7.89448261e-01 2.54090935e-01
4.61305678e-01 -6.88999057e-01 3.67634833e-01 4.07450087e-02
-1.58053741e-01 -2.34709620e-01 -6.62448227e-01 8.96116138e-01
-2.48871040e+00 -1.52482247e+00 -3.61967772e-01 2.03933477e+00
6.05237186e-01 2.18361408e-01 1.16861574e-02 6.71199262e-02
3.52022022e-01 4.30153906e-01 -1.21359408e+00 -5.46132565e-01
2.11883590e-01 -1.54590622e-01 4.28091973e-01 4.56058413e-01
-1.13633704e+00 3.37154061e-01 6.69740105e+00 5.78886569e-01
-1.16274309e+00 1.57431066e-01 1.06405032e+00 -5.08632541e-01
-1.05470479e+00 -7.50596374e-02 -5.54602593e-02 8.55743229e-01
1.50913537e+00 -6.86827958e-01 1.53473645e-01 1.02950478e+00
8.50815535e-01 5.03591478e-01 -1.67723751e+00 8.09306264e-01
-3.11051011e-01 -1.92021513e+00 4.05193418e-01 -9.35035422e-02
8.58118355e-01 1.70507088e-01 2.21029118e-01 4.80338126e-01
7.77156532e-01 -1.25823772e+00 1.10743058e+00 9.29563403e-01
5.07464051e-01 -5.46743274e-01 8.37929726e-01 5.22872448e-01
-6.20546341e-01 -3.45619768e-01 -1.25540599e-01 -7.56724000e-01
3.65914464e-01 5.50098121e-01 -9.39218104e-01 6.32772326e-01
3.52407634e-01 9.82206225e-01 7.44158849e-02 5.89283168e-01
-3.89787614e-01 8.46477151e-01 -9.94647853e-03 -1.38012767e-01
1.55455217e-01 1.49961099e-01 6.82103574e-01 7.48088539e-01
4.58081871e-01 1.76865578e-01 -2.70671606e-01 1.28700280e+00
1.04606904e-01 -6.71716571e-01 -9.83048201e-01 1.90375298e-01
3.92711788e-01 4.79796290e-01 -1.54363334e-01 -7.70414472e-01
-2.65289664e-01 7.10694253e-01 3.07758003e-02 4.13015604e-01
-1.35623074e+00 2.02697203e-01 1.10698974e+00 -1.13503821e-03
-2.84260008e-02 2.84633964e-01 -8.78319800e-01 -1.36646271e+00
-7.40596503e-02 -6.61403000e-01 7.52734184e-01 -1.05484939e+00
-1.27851379e+00 2.23907635e-01 4.74136949e-01 -1.53223002e+00
-1.00678766e+00 -5.64796090e-01 -1.07221138e+00 5.95226407e-01
-1.49581337e+00 -1.09155715e+00 1.46378785e-01 1.63037658e-01
8.41562510e-01 1.35016397e-01 6.65589154e-01 -5.74306965e-01
-3.45488161e-01 2.00103581e-01 4.50286046e-02 -4.15565073e-01
1.44390956e-01 -1.52520263e+00 7.96181023e-01 8.43171656e-01
8.72811079e-02 7.29201317e-01 1.35455656e+00 -5.79681754e-01
-9.65103865e-01 -1.65540910e+00 1.02683830e+00 -9.10297871e-01
7.00277388e-01 1.68851167e-01 -5.77549696e-01 1.13849378e+00
1.29423544e-01 -5.02520144e-01 7.48528004e-01 1.73488736e-01
-2.14244798e-01 -5.93346171e-02 -1.18581963e+00 1.00237930e+00
1.11773181e+00 -3.60911101e-01 -1.43363667e+00 4.23994482e-01
1.36569607e+00 -2.77400821e-01 -4.29787040e-01 4.21723396e-01
6.32745147e-01 -9.36095715e-01 1.13912797e+00 -1.38384593e+00
1.04140651e+00 -1.51130438e-01 5.85539639e-02 -1.57354069e+00
-2.56497473e-01 -8.39525521e-01 -4.74458128e-01 5.57186007e-01
5.21146238e-01 -1.01044726e+00 6.02637112e-01 9.00769651e-01
-3.61557275e-01 -6.73882067e-01 -1.12690628e+00 -6.91027403e-01
-7.90137872e-02 -1.13321924e+00 1.51224124e+00 9.67888236e-01
2.97721863e-01 1.77698761e-01 -4.60142463e-01 3.42596650e-01
8.72383296e-01 7.06141233e-01 6.07105732e-01 -1.24518621e+00
-2.65965015e-01 -3.94583255e-01 -1.10034265e-01 -7.64927506e-01
5.31474531e-01 -6.11104071e-01 1.51394950e-02 -1.36872375e+00
1.39783636e-01 2.99789757e-01 -5.06433249e-01 4.05320853e-01
-5.11360943e-01 -9.62987617e-02 5.56964986e-02 -7.06403255e-02
-6.65603995e-01 9.69908535e-01 1.09199798e+00 -2.94296533e-01
4.28343900e-02 2.77734637e-01 -8.53822649e-01 9.80268180e-01
8.48622084e-01 -2.98946291e-01 -7.93165982e-01 -2.60273814e-01
4.35080193e-03 6.24563992e-01 9.22384560e-01 -4.92686212e-01
-1.36340067e-01 -8.78549159e-01 2.30181396e-01 -7.33125985e-01
1.17617562e-01 -5.57513833e-01 4.89849508e-01 7.63620079e-01
-7.03236938e-01 1.11331239e-01 2.02305838e-01 1.20122337e+00
-1.78272352e-01 2.42026299e-01 2.01308504e-01 -6.89062700e-02
-7.44344890e-01 3.58148634e-01 -3.94114882e-01 -1.72523543e-01
1.10934854e+00 -1.55622035e-01 -3.90915990e-01 -6.39485478e-01
-5.51452816e-01 5.56862235e-01 -1.28674597e-01 8.14223886e-01
7.47056365e-01 -1.44338286e+00 -8.75346303e-01 -3.27631459e-03
3.40035647e-01 -7.95516372e-02 5.23414731e-01 6.43191755e-01
-1.00369163e-01 8.42759490e-01 1.72921002e-03 -4.94849682e-01
-7.29476571e-01 7.46790707e-01 2.04331279e-01 -5.02172053e-01
-5.09458899e-01 2.88926065e-01 5.86376190e-01 -6.99478507e-01
-2.77009904e-01 -1.05146742e+00 -1.94476411e-01 -3.13617557e-01
6.30854130e-01 5.62572122e-01 -4.04726386e-01 -7.91903734e-02
-1.21187918e-01 -2.53076911e-01 2.45174915e-01 -1.13810435e-01
1.53215230e+00 -6.95377588e-02 2.76644826e-01 7.81456530e-01
8.37989807e-01 -6.98164225e-01 -1.70138252e+00 2.38914356e-01
2.16484904e-01 -3.07374865e-01 -1.27855018e-01 -8.67653370e-01
-2.51265466e-01 7.28350937e-01 2.97873110e-01 4.61923599e-01
9.87466931e-01 -4.28559668e-02 8.45332265e-01 2.35576719e-01
2.47298643e-01 -7.89444149e-01 -3.87735516e-01 1.62389979e-01
1.22725642e+00 -1.54981935e+00 -3.31525981e-01 -7.67724365e-02
-6.84629560e-01 7.87566960e-01 3.52771252e-01 -5.50379865e-02
4.15165603e-01 -2.28897512e-01 -1.27234012e-01 -1.10710748e-01
-1.34790742e+00 1.91820472e-01 5.16750634e-01 6.28108799e-01
4.81717825e-01 8.20027888e-01 -2.43112266e-01 1.05640614e+00
-7.26206422e-01 3.85927051e-01 4.82847393e-01 1.35321379e-01
7.20249265e-02 -4.00401235e-01 -2.87819177e-01 8.22227180e-01
-4.50564682e-01 -5.99937327e-02 1.36429248e-02 7.29216933e-01
2.58619748e-02 1.02782857e+00 4.44754153e-01 -2.02718943e-01
2.83019513e-01 1.80416644e-01 -1.06010400e-01 -3.23406905e-01
-9.27922968e-03 -4.25023139e-01 2.34582826e-01 -8.93721044e-01
-4.58122730e-01 -9.40210223e-01 -1.30023515e+00 -5.02603233e-01
2.05536261e-01 2.78633479e-02 2.59355783e-01 1.35577583e+00
6.66523516e-01 9.76169229e-01 4.96192873e-01 -6.93321705e-01
-9.71475363e-01 -1.04777730e+00 -1.48546189e-01 7.23226428e-01
9.44843173e-01 -5.47851264e-01 -3.02023739e-01 2.48365521e-01]
|
[8.704639434814453, 5.614113807678223]
|
af2eb9d1-47fe-4e12-966f-4df054d27142
|
3d-guided-weakly-supervised-semantic
|
2012.00242
| null |
https://arxiv.org/abs/2012.00242v1
|
https://arxiv.org/pdf/2012.00242v1.pdf
|
3D Guided Weakly Supervised Semantic Segmentation
|
Pixel-wise clean annotation is necessary for fully-supervised semantic segmentation, which is laborious and expensive to obtain. In this paper, we propose a weakly supervised 2D semantic segmentation model by incorporating sparse bounding box labels with available 3D information, which is much easier to obtain with advanced sensors. We manually labeled a subset of the 2D-3D Semantics(2D-3D-S) dataset with bounding boxes, and introduce our 2D-3D inference module to generate accurate pixel-wise segment proposal masks. Guided by 3D information, we first generate a point cloud of objects and calculate objectness probability score for each point. Then we project the point cloud with objectness probabilities back to 2D images followed by a refinement step to obtain segment proposals, which are treated as pseudo labels to train a semantic segmentation network. Our method works in a recursive manner to gradually refine the above-mentioned segment proposals. Extensive experimental results on the 2D-3D-S dataset show that the proposed method can generate accurate segment proposals when bounding box labels are available on only a small subset of training images. Performance comparison with recent state-of-the-art methods further illustrates the effectiveness of our method.
|
['Nick Barnes', 'Jing Zhang', 'Weixuan Sun']
|
2020-12-01
| null | null | null | null |
['2d-semantic-segmentation']
|
['computer-vision']
|
[ 4.68237668e-01 5.25942981e-01 -3.14342290e-01 -5.74432254e-01
-8.00657511e-01 -6.24309301e-01 3.26829374e-01 9.32579488e-02
-2.52307028e-01 2.97156632e-01 -3.61526668e-01 -1.00995056e-01
2.92982817e-01 -8.10130239e-01 -8.35128844e-01 -4.57552671e-01
2.10253417e-01 9.08916116e-01 9.71254587e-01 8.24553519e-02
2.35567510e-01 6.31438434e-01 -1.53314042e+00 -2.46320888e-01
1.03762531e+00 1.15652108e+00 6.16199851e-01 3.88546705e-01
-5.69339573e-01 5.49933538e-02 -2.84593344e-01 7.41941631e-02
5.32292783e-01 -2.22194940e-01 -9.13437426e-01 7.74824381e-01
2.31686652e-01 -4.87598717e-01 3.07322979e-01 1.15814006e+00
1.19734600e-01 -4.95193619e-03 6.25484526e-01 -9.63017046e-01
-9.54117849e-02 2.21347705e-01 -8.44208539e-01 -4.92392957e-01
1.52311400e-01 -2.58269664e-02 7.43108630e-01 -9.70816851e-01
5.44630170e-01 1.12042415e+00 7.07536459e-01 4.81972575e-01
-1.14014971e+00 -5.67259371e-01 4.64069396e-01 -4.66848284e-01
-1.41136003e+00 3.52057219e-02 1.17377615e+00 -3.75728905e-01
5.00919104e-01 -4.16843966e-02 7.39519775e-01 4.76175100e-01
-4.70539689e-01 1.10482955e+00 1.03586733e+00 -2.49007583e-01
5.06278276e-01 1.03390731e-01 1.66277781e-01 7.72492647e-01
2.17672110e-01 -2.92900056e-01 -8.56386423e-02 2.99562607e-02
1.21307671e+00 1.50298730e-01 9.94883627e-02 -7.50528634e-01
-1.25715661e+00 6.45453572e-01 6.27293229e-01 6.38011172e-02
-4.09509838e-01 8.97900611e-02 5.74880140e-03 -3.33310217e-01
8.26117814e-01 7.83344731e-02 -6.04976237e-01 2.87148684e-01
-1.19357681e+00 1.96522549e-01 4.80462879e-01 1.29072833e+00
1.19239879e+00 -3.13554466e-01 2.14747712e-01 8.74997854e-01
5.24234951e-01 6.22956812e-01 -1.54552367e-02 -1.30576086e+00
3.91684532e-01 8.61680388e-01 4.26347911e-01 -7.17761993e-01
-4.89836901e-01 -2.91371137e-01 -6.48151159e-01 2.18142390e-01
3.26936245e-01 1.83878869e-01 -1.52534878e+00 1.21399426e+00
8.48907232e-01 1.94598943e-01 -9.39076841e-02 1.16593051e+00
7.19507933e-01 5.89706004e-01 7.07199872e-02 8.32706466e-02
1.15557468e+00 -1.00164437e+00 -3.11582595e-01 -4.73320276e-01
5.87025583e-01 -5.44912398e-01 1.02198148e+00 1.33399501e-01
-1.04193032e+00 -6.30635798e-01 -8.06473315e-01 -6.99315369e-02
-4.83972654e-02 2.00016722e-01 5.96518874e-01 5.14242172e-01
-7.59919524e-01 4.16795611e-01 -1.12833273e+00 -1.94439098e-01
1.08561683e+00 2.34423012e-01 -1.88027546e-01 -6.65781796e-02
-7.12864995e-01 5.35333812e-01 7.78343678e-01 3.19540314e-02
-1.10278797e+00 -4.94631588e-01 -1.10027349e+00 -2.99014956e-01
6.66747153e-01 -5.92745423e-01 1.23217738e+00 -5.23576498e-01
-1.34922123e+00 1.24971938e+00 -3.20832491e-01 -1.98219612e-01
5.10435820e-01 -1.74652651e-01 2.95552552e-01 3.76055866e-01
5.31576455e-01 1.18273413e+00 7.30661929e-01 -1.87054014e+00
-8.51969540e-01 -5.54101884e-01 1.93358138e-02 4.22991097e-01
3.62642050e-01 -3.19468170e-01 -9.79395568e-01 -4.06291485e-01
1.01153910e+00 -8.52234900e-01 -7.01720834e-01 2.32970521e-01
-7.97258675e-01 -3.65599811e-01 9.42384303e-01 -4.79770094e-01
7.22207725e-01 -1.97526538e+00 -7.43017793e-02 4.05680895e-01
1.10246763e-01 1.49231598e-01 1.08767606e-01 -1.64912164e-01
3.67001206e-01 6.12443462e-02 -9.97549772e-01 -6.92618012e-01
-5.24757393e-02 3.83186340e-01 -4.06997688e-02 3.85943592e-01
4.25776511e-01 8.14397931e-01 -1.02885413e+00 -8.72149050e-01
5.76629639e-01 1.95400730e-01 -5.82682967e-01 2.56622106e-01
-7.74368942e-01 6.46608233e-01 -1.05328822e+00 1.05301332e+00
9.20575380e-01 -4.12429333e-01 -3.44479352e-01 -1.59833759e-01
-2.05028936e-01 2.32788712e-01 -1.19557917e+00 2.17106080e+00
-1.65425420e-01 9.54346657e-02 3.22034210e-02 -1.14459181e+00
1.16183686e+00 -1.89226437e-02 6.19889736e-01 -2.40023211e-01
1.62003458e-01 4.26225871e-01 -6.34143174e-01 -4.10636634e-01
3.23694676e-01 -2.54364252e-01 -3.40622395e-01 3.82949203e-01
-8.12256895e-03 -1.09577358e+00 -3.28736566e-02 7.86640216e-04
6.01500154e-01 6.66400075e-01 9.59742591e-02 2.97067985e-02
4.45149064e-01 4.84093875e-01 6.30398989e-01 6.80629790e-01
-2.42474660e-01 1.02237940e+00 2.99552888e-01 -2.73145139e-01
-1.12703288e+00 -1.13315558e+00 -2.99759656e-01 3.84145677e-01
8.44884396e-01 -4.40758243e-02 -1.06219661e+00 -1.00048566e+00
-1.10752732e-01 5.74500501e-01 -3.79799634e-01 2.17985630e-01
-4.90374327e-01 -4.60087568e-01 8.27747062e-02 6.60326600e-01
8.69871795e-01 -7.89802313e-01 -7.52233088e-01 1.50117144e-01
-1.82701245e-01 -1.23880303e+00 -3.77796531e-01 2.89357007e-01
-1.23625302e+00 -9.26243484e-01 -9.82179582e-01 -1.11114049e+00
1.12921309e+00 3.73688817e-01 9.43048358e-01 -1.88426825e-03
7.55679188e-03 1.16716750e-01 -4.14004236e-01 -4.45448756e-01
-1.68532953e-01 -8.40469822e-03 -3.13551217e-01 -1.89964861e-01
2.60275602e-01 -3.86277944e-01 -7.53006518e-01 6.22917652e-01
-8.04638386e-01 4.14709151e-01 6.29184425e-01 4.25145328e-01
1.36999333e+00 1.75901815e-01 2.48651251e-01 -1.05411756e+00
-1.35309041e-01 -2.40711197e-01 -8.12932730e-01 -1.43995419e-01
-2.12417006e-01 -1.98437780e-01 1.49285242e-01 -1.39201850e-01
-1.09074712e+00 7.36293018e-01 -2.51263171e-01 -6.57341123e-01
-7.05187738e-01 1.96318552e-01 -4.43215132e-01 1.94914520e-01
3.28391135e-01 1.78419188e-01 -3.48175839e-02 -6.70316994e-01
5.80168426e-01 7.09617257e-01 6.40809119e-01 -5.42956412e-01
9.33451653e-01 8.81270289e-01 -6.98982850e-02 -6.32986367e-01
-1.20169830e+00 -7.10856259e-01 -1.11158741e+00 -1.43325374e-01
1.17863226e+00 -9.57121730e-01 -1.75039575e-01 5.61510921e-01
-1.15000820e+00 -6.06623054e-01 -5.02936542e-01 4.25877035e-01
-6.97994173e-01 3.61564517e-01 -2.83694506e-01 -8.38854909e-01
-9.57414731e-02 -1.16396093e+00 1.82638299e+00 2.57685810e-01
-5.75993918e-02 -8.45341206e-01 -2.19221398e-01 6.25518799e-01
-2.56067485e-01 3.19328338e-01 5.75784445e-01 -4.61408705e-01
-7.57617116e-01 -2.48811394e-01 -5.37415147e-01 5.19360185e-01
8.99968520e-02 -4.03708249e-01 -8.50430369e-01 3.07306647e-01
2.31045851e-05 -2.24602789e-01 7.43834317e-01 5.74686885e-01
1.50103939e+00 1.71206385e-01 -6.32650375e-01 5.68821132e-01
1.22779083e+00 2.63132472e-02 2.40844637e-01 9.31698382e-02
1.00361383e+00 6.58763945e-01 1.19732118e+00 2.97602952e-01
6.16422117e-01 4.81016070e-01 6.91659987e-01 -3.37707281e-01
-1.97525874e-01 -6.27530754e-01 -2.24289536e-01 6.37332380e-01
8.82894546e-02 2.31747539e-03 -9.36843634e-01 7.87716389e-01
-1.70764184e+00 -3.15980881e-01 -3.57970804e-01 1.92986584e+00
9.51493740e-01 3.89737368e-01 1.64933622e-01 2.26263478e-01
8.72684717e-01 6.41375706e-02 -7.91767001e-01 2.41620436e-01
1.10256337e-01 1.80058837e-01 5.90261042e-01 4.88379210e-01
-1.26010776e+00 1.33998644e+00 5.51612425e+00 8.75824869e-01
-8.09077561e-01 8.50648805e-02 6.93631530e-01 2.48753190e-01
-4.45032477e-01 2.15154812e-01 -9.14841235e-01 4.82493192e-01
1.51258320e-01 4.54572469e-01 -1.32882953e-01 1.03211951e+00
3.23781967e-01 -4.66534972e-01 -1.00942600e+00 9.73736405e-01
1.64113156e-02 -1.10640156e+00 -6.56006336e-02 -1.00488715e-01
1.01891053e+00 6.59242719e-02 -3.76333952e-01 -8.57816041e-02
2.93810666e-01 -6.44181252e-01 9.65739727e-01 2.19287708e-01
7.56545544e-01 -5.45626938e-01 5.29504418e-01 7.70893693e-01
-1.14522183e+00 3.36944252e-01 -4.48894948e-01 1.41044006e-01
5.77492416e-01 1.08102369e+00 -8.43817174e-01 3.65643740e-01
7.47334421e-01 8.48479211e-01 -1.29029661e-01 9.63478565e-01
-5.56240380e-01 5.38146198e-01 -7.36258745e-01 1.57419607e-01
4.24028486e-01 -4.24589247e-01 4.88647968e-01 7.05209434e-01
2.99979210e-01 4.30101454e-01 4.86209363e-01 1.27080154e+00
-1.31959364e-01 -1.93368465e-01 -3.57215613e-01 3.57568592e-01
6.07146621e-01 1.14755905e+00 -1.52763474e+00 -5.51984012e-01
-2.05624312e-01 1.11367238e+00 2.99846567e-02 2.46537745e-01
-6.49430394e-01 -3.36871117e-01 1.32816821e-01 1.70893237e-01
5.14986098e-01 -3.90446752e-01 -6.40773237e-01 -9.27194297e-01
6.12739623e-02 -7.98068047e-02 6.40669465e-02 -1.06165123e+00
-1.03512526e+00 1.87900439e-01 2.83520758e-01 -1.27685177e+00
1.66648909e-01 -3.64804059e-01 -3.89716566e-01 7.51815498e-01
-1.63341236e+00 -1.22255516e+00 -4.87369388e-01 3.18847954e-01
7.74732471e-01 4.51354265e-01 4.92112815e-01 1.08447403e-01
-3.04239154e-01 -6.95547163e-02 -3.28476697e-01 1.31574512e-01
1.19559363e-01 -1.23963737e+00 5.75431585e-01 6.96947932e-01
1.10907517e-02 1.74113929e-01 3.90296400e-01 -1.00143754e+00
-7.16488898e-01 -1.42920923e+00 6.65016770e-01 -5.05049348e-01
1.63714319e-01 -4.21643913e-01 -8.05703521e-01 5.66317141e-01
-4.82400954e-01 2.19833910e-01 3.33770066e-01 -4.29089159e-01
6.49795979e-02 1.71114609e-01 -1.35995686e+00 4.04750466e-01
1.39398074e+00 -3.44128311e-01 -7.69021869e-01 3.97901207e-01
1.00695765e+00 -7.00377405e-01 -6.84636474e-01 6.27784669e-01
1.05328247e-01 -7.55304456e-01 1.07924259e+00 3.12675424e-02
3.22832763e-01 -7.11036503e-01 -9.85326618e-02 -1.05531669e+00
2.17640132e-01 -3.17019284e-01 1.68506458e-01 1.07410705e+00
3.04979861e-01 -4.06158030e-01 1.24203157e+00 4.78692502e-01
-5.71145415e-01 -7.14600503e-01 -8.79590869e-01 -6.82919145e-01
-1.20554723e-01 -8.89380097e-01 6.94264650e-01 7.61508942e-01
-4.62031722e-01 2.79523656e-02 2.65617281e-01 4.68797028e-01
1.01712763e+00 4.99109924e-01 8.22165728e-01 -1.33684635e+00
2.01504365e-01 -2.82767862e-01 -3.89868975e-01 -1.82175672e+00
1.64710164e-01 -8.39241922e-01 4.58106905e-01 -1.86174321e+00
6.74950257e-02 -1.00115764e+00 6.03798404e-02 5.20934820e-01
-2.05132008e-01 6.23458922e-01 -1.34306431e-01 4.02051538e-01
-7.13036835e-01 5.79147339e-01 1.55830920e+00 -8.64711553e-02
-4.14710581e-01 2.27503434e-01 -4.35197860e-01 1.13751209e+00
6.47842526e-01 -4.88479644e-01 -5.02726614e-01 -4.34605539e-01
-2.62922585e-01 -6.22512102e-02 4.51876283e-01 -8.64342213e-01
-5.95449731e-02 -2.01437309e-01 3.98430109e-01 -1.34677780e+00
5.01113772e-01 -1.01213431e+00 -2.05459684e-01 1.79513425e-01
-1.07442819e-01 -8.62471342e-01 -7.22420961e-02 5.62060535e-01
-6.90758303e-02 -3.71724069e-01 7.50556648e-01 -3.81107211e-01
-7.44533181e-01 6.05950832e-01 5.84611408e-02 6.85017370e-03
1.34498715e+00 -6.32446885e-01 3.60856801e-01 5.06965294e-02
-8.78802001e-01 5.26579082e-01 7.98131824e-01 2.05085501e-01
7.44322240e-01 -1.20691216e+00 -4.21985596e-01 3.12442064e-01
1.62285730e-01 1.29716992e+00 2.13940173e-01 6.59448862e-01
-7.70907164e-01 3.04831982e-01 1.62227675e-01 -1.32067955e+00
-8.69193912e-01 2.92616367e-01 2.40425989e-01 2.08496749e-01
-8.15191507e-01 1.02763915e+00 4.68708396e-01 -8.24284434e-01
1.38421372e-01 -7.67883539e-01 7.43792066e-03 -1.39240801e-01
-2.18892302e-02 1.44787639e-01 -1.78019598e-01 -6.69504941e-01
-3.82235765e-01 1.13823938e+00 1.30977616e-01 -1.28199294e-01
1.34334171e+00 -3.40001434e-01 1.16389282e-01 4.59970027e-01
9.99147892e-01 -4.56982732e-01 -1.78143346e+00 -3.24769825e-01
-1.49315253e-01 -5.76222301e-01 1.31692529e-01 -4.56646591e-01
-1.11946571e+00 9.26719487e-01 3.32428038e-01 4.67111655e-02
9.35277462e-01 6.81899011e-01 1.07598317e+00 -5.18610738e-02
6.48303926e-01 -1.22122717e+00 -9.13283881e-03 2.78117836e-01
4.18391556e-01 -1.41756213e+00 1.63821280e-02 -9.72732842e-01
-4.59704041e-01 7.31606483e-01 5.91991067e-01 -1.23738006e-01
7.22846866e-01 -1.10177942e-01 7.82872587e-02 -3.65197122e-01
8.70583281e-02 -4.89968121e-01 2.05045149e-01 7.18071759e-01
-1.62907436e-01 7.39628822e-02 8.51881411e-03 5.82176626e-01
-8.43779892e-02 -3.37880962e-02 1.58044338e-01 1.05853534e+00
-7.56019473e-01 -9.85659242e-01 -5.46145916e-01 2.87186146e-01
5.90942334e-03 3.27102453e-01 -2.72442430e-01 6.82822764e-01
3.02189708e-01 7.67646849e-01 3.41579735e-01 -8.35769922e-02
3.18618685e-01 -8.53755996e-02 3.99051696e-01 -1.03395283e+00
2.02245623e-01 4.43243772e-01 -1.34633198e-01 -5.93730927e-01
-6.56452298e-01 -7.26508737e-01 -2.00419354e+00 4.30500656e-01
-4.11171377e-01 -4.81026135e-02 1.03893805e+00 1.01335251e+00
2.34281123e-01 2.13316560e-01 6.84230804e-01 -1.45183897e+00
-1.30623832e-01 -6.61274731e-01 -5.76639473e-01 4.74714965e-01
2.04565033e-01 -7.44459867e-01 -3.21336895e-01 2.91269630e-01]
|
[8.030277252197266, -3.147217273712158]
|
ce254f58-7651-4495-b5d0-4bc06bf6eecb
|
omnivl-one-foundation-model-for-image
|
2209.07526
| null |
https://arxiv.org/abs/2209.07526v2
|
https://arxiv.org/pdf/2209.07526v2.pdf
|
OmniVL:One Foundation Model for Image-Language and Video-Language Tasks
|
This paper presents OmniVL, a new foundation model to support both image-language and video-language tasks using one universal architecture. It adopts a unified transformer-based visual encoder for both image and video inputs, and thus can perform joint image-language and video-language pretraining. We demonstrate, for the first time, such a paradigm benefits both image and video tasks, as opposed to the conventional one-directional transfer (e.g., use image-language to help video-language). To this end, we propose a decoupled joint pretraining of image-language and video-language to effectively decompose the vision-language modeling into spatial and temporal dimensions and obtain performance boost on both image and video tasks. Moreover, we introduce a novel unified vision-language contrastive (UniVLC) loss to leverage image-text, video-text, image-label (e.g., image classification), video-label (e.g., video action recognition) data together, so that both supervised and noisily supervised pretraining data are utilized as much as possible. Without incurring extra task-specific adaptors, OmniVL can simultaneously support visual only tasks (e.g., image classification, video action recognition), cross-modal alignment tasks (e.g., image/video-text retrieval), and multi-modal understanding and generation tasks (e.g., image/video question answering, captioning). We evaluate OmniVL on a wide range of downstream tasks and achieve state-of-the-art or competitive results with similar model size and data scale.
|
['Lu Yuan', 'Yu-Gang Jiang', 'Ce Liu', 'Yujia Xie', 'Yucheng Zhao', 'Luowei Zhou', 'Chong Luo', 'Zuxuan Wu', 'Dongdong Chen', 'Junke Wang']
|
2022-09-15
| null | null | null | null |
['video-text-retrieval', 'action-classification', 'video-question-answering']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 2.79042959e-01 -1.31364614e-01 -4.61565912e-01 -3.85482788e-01
-1.05569613e+00 -7.46622086e-01 7.75197029e-01 -2.79556066e-01
-7.14403570e-01 3.76200140e-01 3.34189236e-02 -4.53778923e-01
5.21136403e-01 -4.25367087e-01 -1.27909315e+00 -5.03286958e-01
5.19543767e-01 2.56298333e-01 1.35455757e-01 1.00088388e-01
-3.11519094e-02 1.49239063e-01 -1.67796266e+00 6.98754609e-01
3.45998347e-01 1.08413243e+00 6.66415215e-01 9.04724300e-01
-3.64939608e-02 1.20173168e+00 -7.01895654e-02 -3.67615759e-01
1.22177236e-01 -2.19463661e-01 -8.01761210e-01 5.43187082e-01
9.64904308e-01 -8.48815084e-01 -6.84349120e-01 9.21258748e-01
2.45740071e-01 -2.11184323e-02 5.60824037e-01 -1.58739603e+00
-1.14524674e+00 1.68551996e-01 -6.30186856e-01 1.56698003e-02
3.73742133e-01 6.61303282e-01 9.99705136e-01 -1.22217703e+00
7.31177449e-01 1.41925073e+00 1.20558009e-01 6.99408472e-01
-1.29596949e+00 -6.33583248e-01 4.14890081e-01 3.15566212e-01
-1.18154359e+00 -7.54768610e-01 5.50817072e-01 -5.79422832e-01
9.85321164e-01 -4.74426299e-02 4.00244296e-01 1.51748204e+00
1.03135087e-01 1.25752342e+00 8.52295101e-01 -2.13660046e-01
-1.02584101e-01 1.22926645e-01 -4.77890186e-02 8.80844414e-01
-1.21328190e-01 2.65875328e-02 -7.00552881e-01 4.31127548e-01
9.07059669e-01 2.35089630e-01 -2.84249634e-01 -2.60600775e-01
-1.32322192e+00 6.17833972e-01 1.36144683e-01 5.67147881e-02
-1.06781051e-01 5.23119688e-01 5.73056996e-01 4.29863781e-01
2.48560280e-01 -1.14819989e-01 -4.37215716e-01 -8.93926769e-02
-9.80316043e-01 -9.74079370e-02 2.86089927e-01 1.23378015e+00
9.88910139e-01 2.99474031e-01 -4.32624161e-01 7.39190340e-01
4.66610491e-01 1.00473142e+00 5.17471433e-01 -1.05811906e+00
7.04741538e-01 2.33673513e-01 -1.46390319e-01 -7.18418837e-01
-1.85768977e-01 5.56556024e-02 -8.80755007e-01 -4.55504796e-03
1.86112568e-01 1.01107016e-01 -1.25437522e+00 2.11650968e+00
-7.82788545e-03 3.47621113e-01 2.44098127e-01 1.06276047e+00
9.85600114e-01 9.17178214e-01 3.00808460e-01 -2.21364856e-01
1.57536292e+00 -1.41903329e+00 -4.55087721e-01 -6.36578500e-01
5.86052299e-01 -8.00891161e-01 1.51853681e+00 9.76921618e-02
-1.32328629e+00 -9.35232460e-01 -6.21056676e-01 -6.76630199e-01
-2.30418935e-01 3.60488445e-01 4.01045799e-01 1.63984776e-01
-1.38165367e+00 -1.47199482e-01 -8.78266394e-01 -4.28111374e-01
2.32500061e-01 1.45967588e-01 -7.25061953e-01 -4.79128182e-01
-1.02596760e+00 4.55999970e-01 3.37468863e-01 -2.54811674e-01
-1.41192544e+00 -5.78348696e-01 -1.13737774e+00 -3.15444340e-04
4.74395007e-01 -1.08019996e+00 1.22176862e+00 -1.40886497e+00
-1.29026663e+00 1.32568610e+00 -2.76695341e-01 -6.06430531e-01
3.92088830e-01 -2.72680521e-01 -2.24822626e-01 5.08835256e-01
2.66954482e-01 1.29157817e+00 1.15740705e+00 -1.24050105e+00
-6.45985603e-01 -2.62375325e-01 3.74907285e-01 3.82034987e-01
-6.48250639e-01 -3.49919014e-02 -1.22437394e+00 -6.18327379e-01
-2.94948846e-01 -1.01513827e+00 2.80079037e-01 4.04808342e-01
-2.49529585e-01 -1.09580927e-01 1.04558682e+00 -8.06561172e-01
8.36809158e-01 -2.28227186e+00 3.14388335e-01 -3.17649484e-01
2.81837016e-01 2.75240809e-01 -7.80035734e-01 2.29051337e-01
-1.10009603e-01 6.58446178e-02 -1.56303093e-01 -6.21997893e-01
-2.00612321e-01 4.25627291e-01 -3.97236884e-01 3.68232697e-01
3.14492494e-01 1.36090267e+00 -8.02481771e-01 -6.66031659e-01
4.41979349e-01 4.57380116e-01 -8.32684755e-01 4.60268229e-01
-5.27897418e-01 3.44892621e-01 -2.94169903e-01 5.62569857e-01
3.97751927e-01 -7.16270626e-01 6.82269782e-03 -5.75863779e-01
1.12799712e-01 -9.88432169e-02 -6.60788834e-01 1.99748945e+00
-9.29552138e-01 8.50411296e-01 3.36006492e-01 -1.08231628e+00
2.88914740e-01 3.98344517e-01 5.40617943e-01 -1.02204275e+00
-1.58170328e-01 5.04873469e-02 -4.74995643e-01 -7.78523505e-01
4.28335547e-01 9.23495963e-02 7.18865870e-03 5.23574710e-01
4.98938859e-01 1.46495849e-01 2.21948445e-01 5.37372947e-01
8.40032935e-01 3.73297781e-01 1.82269905e-02 1.67402670e-01
6.59156024e-01 -6.79675788e-02 1.19165264e-01 5.61290622e-01
-1.10478833e-01 6.54868424e-01 1.28376797e-01 1.76888090e-02
-1.01509702e+00 -1.15815759e+00 3.18216413e-01 1.50826776e+00
2.69488424e-01 -5.03907979e-01 -6.32036746e-01 -5.76644480e-01
7.76452571e-02 5.39558887e-01 -3.37150186e-01 -3.39266539e-01
-4.30269212e-01 -1.10182330e-01 5.77952862e-01 6.04560494e-01
5.17185092e-01 -9.44283187e-01 -2.87713587e-01 -6.19376823e-02
-4.34130520e-01 -1.79460251e+00 -9.47560310e-01 -1.77685976e-01
-7.16289639e-01 -8.24887931e-01 -7.56593347e-01 -1.03035903e+00
5.63163757e-01 9.22237217e-01 1.23261213e+00 5.11195250e-02
-1.74274355e-01 1.17926490e+00 -2.71826237e-01 3.46876204e-01
-3.14234108e-01 -2.62744814e-01 -9.71789286e-03 2.27619737e-01
8.38230774e-02 -4.87807572e-01 -6.45043314e-01 3.52681637e-01
-1.25549948e+00 5.12335539e-01 7.93986917e-01 1.04526985e+00
6.89419866e-01 -5.31687498e-01 2.81755626e-01 -5.07936895e-01
1.92805693e-01 -4.70206350e-01 -4.81570512e-01 5.73485196e-01
-3.81045640e-01 -5.00625782e-02 6.54502869e-01 -6.32170260e-01
-8.83829474e-01 1.18438276e-02 -4.56872098e-02 -1.24228907e+00
-2.23426104e-01 5.23685396e-01 -2.26782620e-01 3.13628055e-02
3.13551694e-01 7.86930978e-01 1.85778841e-01 -3.15904021e-01
8.36006999e-01 7.08211839e-01 8.28827620e-01 -4.89090562e-01
7.64159620e-01 5.56796193e-01 -2.72548616e-01 -8.57632101e-01
-8.40244651e-01 -6.34552896e-01 -3.80599827e-01 -3.10382098e-01
1.34050190e+00 -1.57174635e+00 -7.49035716e-01 5.65670788e-01
-1.40016711e+00 -4.77673680e-01 -7.91044459e-02 3.33531708e-01
-8.03507388e-01 5.77141464e-01 -7.32720792e-01 -4.62292641e-01
-2.98045546e-01 -1.36432886e+00 1.60567212e+00 -6.72666281e-02
3.01240772e-01 -8.44183683e-01 -5.12630165e-01 7.56009758e-01
1.57810166e-01 -3.18915933e-01 6.72189593e-01 -2.49673024e-01
-9.85138237e-01 1.37434572e-01 -8.19917977e-01 5.93325436e-01
-2.15632647e-01 -2.03031465e-01 -9.77815092e-01 -5.56503415e-01
-9.44190249e-02 -9.80442286e-01 1.03886652e+00 2.10648045e-01
1.36057770e+00 -2.39373520e-01 -1.26466036e-01 8.67147863e-01
1.21666253e+00 3.50021683e-02 6.56614184e-01 1.43980961e-02
1.11680174e+00 3.99888366e-01 6.49796426e-01 9.73325297e-02
7.20317185e-01 9.37691569e-01 3.59582275e-01 -3.40873122e-01
-5.99816442e-01 -4.71070260e-01 1.03903544e+00 8.44481051e-01
2.52101690e-01 -4.56816882e-01 -8.14401746e-01 4.78269547e-01
-2.04138899e+00 -9.10389066e-01 1.57734856e-01 1.94791603e+00
6.74610496e-01 -2.29847193e-01 -3.32500897e-02 -4.87465292e-01
5.60363114e-01 2.97948569e-01 -7.82324731e-01 6.90792799e-02
-1.94887444e-01 -2.29651824e-01 5.12226224e-01 4.43482220e-01
-1.18223870e+00 1.32401907e+00 5.28872776e+00 9.94616747e-01
-1.42523658e+00 4.31346238e-01 7.73935735e-01 -3.52911115e-01
-3.87146741e-01 -1.64800644e-01 -6.09892666e-01 3.99082601e-01
8.64306748e-01 -3.92289683e-02 5.41814446e-01 7.47857571e-01
2.49115452e-01 1.16278216e-01 -1.37238324e+00 1.50596499e+00
4.48615372e-01 -1.44926214e+00 5.72430313e-01 -7.30910376e-02
4.77438420e-01 2.44423464e-01 1.98628381e-01 3.92520398e-01
7.02387393e-02 -8.76869619e-01 1.11574209e+00 2.82380342e-01
1.48507524e+00 -9.76273939e-02 1.89647377e-01 1.48511156e-01
-1.30325198e+00 -1.37548342e-01 -1.75769981e-02 2.81602889e-01
5.15544891e-01 3.41379911e-01 -5.27718700e-02 5.29591382e-01
7.28249431e-01 1.19387364e+00 -4.83898968e-01 2.96963632e-01
-1.31302103e-01 5.17683148e-01 -1.44719869e-01 5.17489374e-01
5.32646120e-01 -1.63154125e-01 3.40154052e-01 1.15612388e+00
2.19825581e-01 -4.11419123e-02 6.14076316e-01 7.03216910e-01
-2.96224713e-01 1.12550497e-01 -7.42110848e-01 -3.27689320e-01
1.53636426e-01 1.09663177e+00 -3.40346128e-01 -4.96120036e-01
-1.01795840e+00 1.47704422e+00 3.18674833e-01 7.50475049e-01
-9.87886250e-01 2.16670170e-01 6.66811764e-01 1.08454145e-01
2.18925908e-01 -4.08518851e-01 2.71216393e-01 -1.57426262e+00
1.49437413e-01 -9.63455796e-01 3.62570733e-01 -1.21747887e+00
-1.19650292e+00 4.68483925e-01 -1.00740507e-01 -1.27701652e+00
-4.09035385e-01 -9.20455873e-01 -2.28908718e-01 5.72415233e-01
-1.69550765e+00 -1.63845468e+00 -3.08149844e-01 1.21293783e+00
8.63912225e-01 -2.76826799e-01 3.44110310e-01 6.65385723e-01
-4.45857108e-01 6.54770017e-01 -3.10013164e-02 2.67801315e-01
9.98865962e-01 -7.70546734e-01 1.68818220e-01 9.09208059e-01
5.00103951e-01 3.48077148e-01 7.94108808e-02 -3.61139685e-01
-1.95154583e+00 -1.46904647e+00 5.05786896e-01 -3.44331712e-01
8.72688770e-01 -5.24237216e-01 -5.30546546e-01 8.94899666e-01
2.17808604e-01 1.37197137e-01 4.12898868e-01 -2.66341209e-01
-6.88285887e-01 -1.21902511e-01 -6.44420147e-01 7.97692776e-01
1.21524453e+00 -1.20702207e+00 -1.53754771e-01 5.88326931e-01
1.16651535e+00 -2.34599382e-01 -5.82268119e-01 1.88357681e-01
5.15445113e-01 -7.94515669e-01 1.28678215e+00 -7.07559943e-01
8.14831316e-01 -2.60567874e-01 -4.75569248e-01 -7.05802381e-01
-1.57514781e-01 -4.81012076e-01 -1.95671931e-01 1.21460164e+00
2.94114232e-01 -3.82277161e-01 4.87341106e-01 3.97278577e-01
-2.78792590e-01 -5.40381134e-01 -8.01040888e-01 -7.70334959e-01
-2.08514333e-01 -5.83066285e-01 -1.16274796e-01 8.24542940e-01
-3.68629187e-01 6.58046365e-01 -8.06914389e-01 7.89893139e-03
4.49123591e-01 2.30879694e-01 8.75413537e-01 -5.41066647e-01
-5.91341317e-01 -3.69464517e-01 -2.89312243e-01 -1.66427362e+00
5.53573728e-01 -1.17540002e+00 -8.52097198e-02 -1.48965299e+00
3.71268779e-01 1.91688165e-02 -2.50108629e-01 6.99055910e-01
-5.33467047e-02 4.76371557e-01 5.89807451e-01 4.06857044e-01
-1.01332664e+00 5.78890622e-01 1.38590693e+00 -3.73431653e-01
2.04564169e-01 -5.55427313e-01 -6.47385895e-01 4.84747559e-01
3.35634261e-01 -3.82364653e-02 -7.54961669e-01 -9.96870637e-01
-9.33868513e-02 4.71769571e-01 7.22923040e-01 -7.85571396e-01
2.56291509e-01 -2.71985650e-01 1.07052878e-01 -2.64468342e-01
5.43488681e-01 -7.36215711e-01 -1.70079350e-01 2.70098478e-01
-4.35897589e-01 1.66439936e-01 2.88738251e-01 8.40344191e-01
-6.10425711e-01 2.19922572e-01 7.19817162e-01 -6.79721683e-02
-1.10237610e+00 7.05492377e-01 -3.22157443e-01 9.01924372e-02
1.07543552e+00 -1.05838314e-01 -6.13850951e-01 -6.88036382e-01
-6.58841133e-01 5.89813709e-01 6.81241333e-01 7.50899017e-01
8.77950251e-01 -1.32196331e+00 -6.64653838e-01 2.22709179e-01
5.52527487e-01 -1.80411756e-01 5.32584965e-01 9.83315170e-01
-2.23368481e-01 5.46824634e-01 -1.86442941e-01 -9.17393446e-01
-1.40043008e+00 9.29903150e-01 6.92364797e-02 -2.17837706e-01
-5.49863040e-01 7.78991997e-01 9.40002441e-01 -1.49262011e-01
3.22575748e-01 -1.29522458e-01 1.71364844e-01 -6.03061803e-02
6.06476128e-01 -2.08522558e-01 -2.97709852e-01 -8.99821758e-01
-2.82953918e-01 7.76156127e-01 -1.99677765e-01 -2.40559280e-01
8.75350177e-01 -4.26607430e-01 -4.97797020e-02 3.78256440e-01
1.60873103e+00 -5.50440788e-01 -1.27712798e+00 -3.58794838e-01
-4.10992265e-01 -2.61739969e-01 2.13971809e-01 -5.88231206e-01
-1.30654621e+00 1.24319708e+00 5.75516641e-01 -2.91691571e-01
1.34952867e+00 1.64472267e-01 9.54826236e-01 3.98223132e-01
2.83205718e-01 -9.63891327e-01 5.11506855e-01 5.16493201e-01
8.31929266e-01 -1.44813573e+00 -2.75996029e-01 -2.93324322e-01
-8.27170730e-01 8.63528550e-01 7.24461496e-01 2.53599614e-01
4.33404148e-01 1.59790367e-01 3.31480131e-02 -1.19858049e-02
-1.14993632e+00 -4.63953346e-01 4.12623465e-01 4.80132490e-01
4.38305914e-01 -1.11198694e-01 1.20648690e-01 3.00461471e-01
4.79301333e-01 1.53925642e-01 2.48064533e-01 7.26106644e-01
-1.94563687e-01 -1.02205217e+00 -1.58596516e-01 3.77522379e-01
-1.92242518e-01 -3.92636061e-01 -7.05778822e-02 6.26971543e-01
-6.62154481e-02 8.68342757e-01 2.75602728e-01 -4.62537795e-01
-6.09178469e-02 3.81407999e-02 5.31506956e-01 -5.81665933e-01
-2.40091369e-01 2.40004808e-01 -4.54827920e-02 -8.98487926e-01
-6.93466604e-01 -3.07733923e-01 -1.12457108e+00 -8.11021104e-02
5.62050119e-02 -2.59005666e-01 4.99802798e-01 9.03868735e-01
5.25112927e-01 3.17654967e-01 3.77329379e-01 -9.48112011e-01
-2.41683707e-01 -4.93047148e-01 -3.21740836e-01 8.23563933e-01
3.65348876e-01 -5.97565353e-01 -1.88140586e-01 6.33116126e-01]
|
[10.349349021911621, 0.9884293079376221]
|
0fc2a840-1ee0-4c13-b052-4dd29722aba8
|
spatial-econometrics-for-misaligned-data
|
2207.04082
| null |
https://arxiv.org/abs/2207.04082v1
|
https://arxiv.org/pdf/2207.04082v1.pdf
|
Spatial Econometrics for Misaligned Data
|
We produce methodology for regression analysis when the geographic locations of the independent and dependent variables do not coincide, in which case we speak of misaligned data. We develop and investigate two complementary methods for regression analysis with misaligned data that circumvent the need to estimate or specify the covariance of the regression errors. We carry out a detailed reanalysis of Maccini and Yang (2009) and find economically significant quantitative differences but sustain most qualitative conclusions.
|
['Guillaume Allaire Pouliot']
|
2022-07-08
| null | null | null | null |
['econometrics']
|
['miscellaneous']
|
[-3.29561859e-01 -1.78404614e-01 -6.14618897e-01 -3.54485989e-01
-4.69666362e-01 -7.58230925e-01 8.63517165e-01 -4.98143658e-02
-5.97540140e-01 9.74071085e-01 4.45303053e-01 -9.07844901e-01
-5.16886115e-01 -5.03888011e-01 -5.03402829e-01 -5.15866756e-01
5.01209535e-02 -5.41592166e-02 -4.44130361e-01 -7.51504526e-02
5.21268904e-01 3.32156569e-01 -9.20882225e-01 -7.19116151e-01
8.07865679e-01 4.81875092e-01 1.07149174e-02 7.12259829e-01
4.68913227e-01 7.39592433e-01 -4.83540803e-01 -3.51293594e-01
3.26889932e-01 -2.71110684e-01 -3.30033898e-01 -2.43599132e-01
-2.29146685e-02 -2.06945151e-01 -3.63149554e-01 5.88102520e-01
1.25220701e-01 -1.02097243e-02 1.24348879e+00 -1.48075259e+00
-8.80738497e-01 4.98338193e-01 -9.75414872e-01 4.20893848e-01
2.31073216e-01 -4.71902266e-02 8.02936375e-01 -7.85696387e-01
7.00506568e-01 1.13343763e+00 9.19041455e-01 -2.27029815e-01
-1.25822258e+00 -7.26577759e-01 5.37896752e-02 -2.23540038e-01
-1.55751514e+00 -6.59060478e-01 4.43795979e-01 -7.55466044e-01
6.86366022e-01 2.11884946e-01 2.30149880e-01 1.00911403e+00
4.70917851e-01 -6.32163584e-02 1.26870096e+00 -6.02338314e-01
-2.55570561e-01 2.34324813e-01 1.73029438e-01 2.04080716e-01
8.09991121e-01 4.36486900e-01 -2.43633419e-01 -2.43661374e-01
9.07190979e-01 -1.64651908e-02 2.09679410e-01 -1.60695612e-01
-1.10463643e+00 1.05922985e+00 1.43987089e-01 2.93742597e-01
-4.82083499e-01 5.67881279e-02 2.38415245e-02 4.52901810e-01
5.84624410e-01 4.41565037e-01 -6.45074308e-01 4.44894098e-02
-1.06430614e+00 3.66799772e-01 6.84257269e-01 7.93460608e-01
4.50371861e-01 1.62523732e-01 4.10673618e-01 4.99670953e-01
4.31240410e-01 8.99635851e-01 2.76117831e-01 -1.03463411e+00
7.67261088e-01 3.83002639e-01 4.60386664e-01 -1.36418235e+00
-7.63602138e-01 -9.53524709e-02 -6.45856977e-01 -6.77006915e-02
6.36985421e-01 -8.68512392e-01 -5.26396930e-01 1.72538972e+00
-1.21067472e-01 -4.53500956e-01 1.95636451e-01 6.30673289e-01
3.45753640e-01 5.04708886e-01 1.67032272e-01 -4.40680265e-01
9.37453389e-01 -5.05557835e-01 -7.30139732e-01 -3.27693462e-01
8.58466089e-01 -5.44835806e-01 7.15913117e-01 5.72502874e-02
-9.61779892e-01 -2.51977414e-01 -7.55388200e-01 1.30095221e-02
-5.79560995e-01 1.45117044e-01 7.49835730e-01 4.33911979e-01
-8.74122083e-01 3.25721204e-01 -8.96084070e-01 -3.75011563e-01
-9.15600434e-02 2.49024928e-01 -4.59593803e-01 5.06451190e-01
-9.48077917e-01 1.31137931e+00 -2.60759145e-01 3.52310568e-01
-3.21304463e-02 -6.17573857e-01 -8.98191988e-01 -2.62663305e-01
5.39839491e-02 -2.75170594e-01 9.64692771e-01 -9.17888999e-01
-7.86013901e-01 6.95127308e-01 -3.23090553e-01 -6.08473420e-02
4.83199567e-01 -7.83811510e-02 -5.86379945e-01 -5.42606890e-01
5.02396405e-01 -6.27625734e-02 2.49506965e-01 -1.11042213e+00
-6.89251900e-01 -7.05413878e-01 -5.00191748e-01 7.25226700e-02
4.50839326e-02 7.39683867e-01 -1.23055607e-01 -9.12899435e-01
1.45917684e-01 -7.86826432e-01 -3.56542110e-01 -7.74292946e-01
-4.17944759e-01 6.14026189e-02 2.71100610e-01 -9.80068147e-01
1.75588036e+00 -1.98070276e+00 3.59676868e-01 5.00768006e-01
9.71810147e-02 -7.12893069e-01 2.18738109e-01 6.25805497e-01
-3.96921247e-01 6.04899704e-01 -4.10947412e-01 -2.24541724e-01
1.96610112e-02 2.14171901e-01 -2.80800670e-01 1.13049078e+00
2.18786880e-01 6.97380185e-01 -3.58568907e-01 -4.85572308e-01
1.53426871e-01 -3.84794399e-02 -3.15475434e-01 -1.42661765e-01
8.99663866e-01 3.74712080e-01 -3.38852733e-01 6.78763926e-01
5.90016544e-01 2.29032815e-01 2.32724085e-01 1.68894842e-01
-1.09564662e+00 4.64896381e-01 -1.13690722e+00 9.82471704e-01
-4.81446922e-01 1.13331151e+00 2.99407125e-01 -1.01308393e+00
8.18450272e-01 1.06146060e-01 4.85350132e-01 -7.74649918e-01
-4.19681855e-02 2.03072011e-01 1.39777020e-01 -3.01846683e-01
5.59381962e-01 -1.16398744e-01 -5.99877477e-01 3.25059801e-01
-1.85350373e-01 -2.61122286e-01 1.24354184e-01 -2.02530459e-01
8.17786813e-01 3.66284341e-01 6.38747156e-01 -6.25765324e-01
1.31925166e-01 2.46650651e-01 5.42105198e-01 7.85913408e-01
1.53373927e-01 4.56280351e-01 9.78719771e-01 1.27306832e-02
-1.00827503e+00 -1.05070508e+00 -6.30949616e-01 9.44758415e-01
-3.44298840e-01 -2.51954973e-01 -1.18966222e-01 -3.24099511e-01
4.52757180e-01 7.27734864e-01 -9.82389808e-01 2.85906494e-01
-4.11950588e-01 -1.03919673e+00 4.55156326e-01 4.63084340e-01
-1.69163957e-01 -2.07070008e-01 -6.31821573e-01 -1.13124870e-01
2.02647507e-01 -5.71271360e-01 2.46585950e-01 4.39910114e-01
-7.86983907e-01 -1.03663433e+00 -5.11645913e-01 -4.94688213e-01
6.48923695e-01 7.13607147e-02 9.05967295e-01 1.18761575e-02
2.44151220e-01 7.03672692e-02 -1.28783450e-01 -6.49573863e-01
-1.79599926e-01 1.14396401e-01 1.89340800e-01 -6.85472548e-01
4.56401408e-01 -5.02170682e-01 -2.33144715e-01 4.27696913e-01
-4.57167566e-01 -2.46815696e-01 1.38954058e-01 5.38791895e-01
-2.33890722e-03 -2.84232534e-02 8.19488525e-01 -6.30327106e-01
5.74275851e-01 -1.01771867e+00 -1.05855632e+00 2.02990517e-01
-8.15306187e-01 -8.26927200e-02 5.09479940e-01 -1.64412037e-01
-9.42785144e-01 -4.95301217e-01 4.73789126e-01 2.81868160e-01
-5.81394299e-04 9.23060298e-01 4.33267087e-01 2.01276839e-01
3.77471596e-01 -7.32664943e-01 -1.75216317e-01 -5.02934873e-01
1.93444863e-01 9.24179852e-01 4.11075681e-01 -4.36856419e-01
1.15480399e+00 3.36286932e-01 9.58516300e-02 -6.29643917e-01
3.08833979e-02 -2.98573315e-01 -1.00203741e+00 -3.17104980e-02
6.51183426e-01 -1.19185948e+00 -4.23410952e-01 2.01867089e-01
-8.59101534e-01 -5.04982233e-01 3.09189223e-02 1.09699130e+00
-5.56787431e-01 -2.57562339e-01 -1.93125919e-01 -9.29805517e-01
5.71979582e-01 -1.03170729e+00 6.05697751e-01 2.30308741e-01
-8.48019838e-01 -1.44876778e+00 5.58242381e-01 -2.52277315e-01
2.83856183e-01 5.81152439e-01 9.03682947e-01 -5.80924928e-01
2.66533256e-01 -2.19936162e-01 -3.23541164e-01 -4.78237331e-01
4.12841797e-01 1.11120152e+00 -5.03847718e-01 -2.04118565e-02
-6.54410645e-02 1.86440572e-01 3.98838311e-01 8.79858375e-01
8.11484754e-01 -4.77481574e-01 -3.46466094e-01 7.48265505e-01
1.47885680e+00 3.54011685e-01 3.02368075e-01 6.61415577e-01
6.08204842e-01 1.28514087e+00 6.37590170e-01 5.60643911e-01
6.20422363e-01 4.15488690e-01 -2.38003924e-01 -4.44924682e-01
5.30195594e-01 -2.95215786e-01 1.15127139e-01 6.17118239e-01
-4.00291473e-01 -7.22920820e-02 -1.18624568e+00 6.96078241e-01
-1.82546115e+00 -8.16000819e-01 -6.58528328e-01 2.29575443e+00
6.65406346e-01 -2.43463054e-01 4.42473829e-01 -8.55057687e-02
4.44139600e-01 2.28473008e-01 6.54307902e-02 -6.87426209e-01
-2.07484558e-01 1.37971327e-01 1.44347596e+00 7.08600760e-01
-7.97907114e-01 6.94177985e-01 8.80601406e+00 -8.51516426e-03
-6.69366717e-01 -1.38188407e-01 5.95151305e-01 -2.90748715e-01
-4.05772060e-01 4.68678832e-01 -3.46057594e-01 3.16648066e-01
1.28408182e+00 -3.89704883e-01 1.84467927e-01 2.90042996e-01
7.77557194e-01 -7.39482105e-01 -8.88192773e-01 3.12198937e-01
-3.10351431e-01 -8.64283144e-01 -8.47845435e-01 4.21500474e-01
8.61033797e-01 1.01265676e-01 1.57546178e-01 1.39856249e-01
7.93453991e-01 -1.27105927e+00 9.26756442e-01 6.77944958e-01
6.69134378e-01 -8.11070502e-01 5.43809354e-01 1.11585423e-01
-6.82821989e-01 -2.90356457e-01 -2.42928192e-01 -7.76352167e-01
-1.36706084e-02 1.32520646e-01 -7.16649517e-02 5.69396317e-01
8.29784811e-01 9.11447883e-01 -6.61829948e-01 4.21356678e-01
-9.26822051e-03 6.91896796e-01 -3.90436143e-01 2.47984320e-01
2.71582484e-01 -9.92061019e-01 2.31717378e-01 1.01072526e+00
5.28514802e-01 -6.87918905e-03 -7.24501550e-01 7.83273160e-01
3.02845478e-01 9.78019908e-02 -1.22897458e+00 -1.87230170e-01
7.10730076e-01 6.94143951e-01 -5.96061289e-01 1.11735903e-01
-1.07400763e+00 2.22123891e-01 1.61440283e-01 7.02633560e-01
-7.84118652e-01 -3.83527458e-01 5.94528794e-01 1.43726179e-02
6.95791468e-02 -6.21939957e-01 -9.25454199e-01 -1.12376082e+00
-2.55148727e-02 -6.17072940e-01 3.33820730e-01 -7.36672640e-01
-1.05151153e+00 -5.22906005e-01 5.52655339e-01 -6.69249117e-01
-6.56380534e-01 -5.22611916e-01 -5.91168284e-01 1.25703323e+00
-1.18216956e+00 -6.78085506e-01 3.81294042e-01 2.21514598e-01
-1.76536694e-01 -9.88700539e-02 5.68786860e-01 5.92948589e-03
-9.86422837e-01 4.68686670e-02 8.95309448e-01 8.16110224e-02
9.36167598e-01 -1.31902206e+00 6.03568077e-01 8.30535054e-01
-3.84258866e-01 1.03459191e+00 8.56414437e-01 -9.30859447e-01
-1.16967285e+00 -4.34376270e-01 1.20844162e+00 -8.29946876e-01
1.18017042e+00 -2.87759572e-01 -3.62446308e-01 1.25776076e+00
2.65415400e-01 -5.77246249e-01 5.87227404e-01 6.32168710e-01
-9.80824381e-02 -1.02647461e-01 -9.40407634e-01 8.97127628e-01
6.52061403e-01 -4.11587030e-01 -6.38233721e-01 -1.63679868e-02
4.87812430e-01 2.08185613e-01 -1.35277200e+00 4.10267562e-01
9.22045887e-01 -9.77243960e-01 7.83527136e-01 -7.05069482e-01
4.38742816e-01 -2.39316560e-02 -3.35793972e-01 -1.29720759e+00
-2.60306925e-01 -3.85245472e-01 7.40119100e-01 1.48947263e+00
6.93193674e-01 -8.06134343e-01 1.88613445e-01 1.06468308e+00
2.55989641e-01 -3.15843821e-01 -9.52631831e-01 -5.68421483e-01
6.97573543e-01 -3.89539123e-01 5.93166888e-01 1.41921329e+00
3.32414746e-01 -1.08437382e-01 -2.18135878e-01 -1.00445047e-01
3.85922581e-01 -1.33440346e-02 8.78482401e-01 -1.18844676e+00
3.33829463e-01 -6.26746416e-01 -1.26825348e-01 -1.76170394e-01
2.57022560e-01 -1.86406076e-01 -2.53382623e-01 -1.26804364e+00
-1.20236753e-02 -5.52307725e-01 1.82342738e-01 3.02461952e-01
1.30074263e-01 -1.15208201e-01 -1.26636863e-01 2.27501675e-01
2.54186243e-01 1.80963978e-01 4.96358097e-01 4.45149392e-01
-5.31723380e-01 3.98435518e-02 -1.09882677e+00 7.11824536e-01
7.62985647e-01 -7.61149585e-01 -2.25228250e-01 -2.55013078e-01
6.43193126e-01 4.12367761e-01 4.54653621e-01 -4.22104657e-01
-2.02604353e-01 -1.13514578e+00 5.44308662e-01 -4.48455900e-01
-2.95999885e-01 -1.03923583e+00 4.70428050e-01 1.98322132e-01
-3.29149842e-01 9.30069566e-01 2.67039329e-01 1.77048922e-01
-2.74257548e-02 -3.35884035e-01 1.98824987e-01 2.84008801e-01
1.36727408e-01 -2.61457652e-01 -9.11916971e-01 -1.83753282e-01
7.93252885e-01 -1.04930244e-01 -4.20912504e-01 -5.62141776e-01
-2.56113231e-01 4.21081334e-01 1.01386511e+00 2.94678152e-01
7.46343657e-03 -1.26948619e+00 -8.28936756e-01 -9.13739875e-02
-1.05472311e-01 -5.48867762e-01 -3.12049985e-01 1.32522309e+00
-4.75807309e-01 7.43289173e-01 -5.16435057e-02 1.76378787e-01
-8.96621048e-01 5.57064891e-01 2.71835744e-01 -6.59831520e-03
-3.30298096e-02 1.57643527e-01 2.38654152e-01 -4.71052200e-01
-2.68581510e-01 -4.52691823e-01 -2.36045122e-01 5.03373742e-01
1.10301614e-01 8.73742104e-01 -2.53434330e-01 -9.32418287e-01
-5.87631822e-01 5.48362851e-01 4.96146619e-01 -6.14617467e-01
1.64623713e+00 -7.53495693e-01 -1.53381437e-01 1.14502263e+00
1.20070589e+00 4.54342723e-01 -1.01872551e+00 1.84191957e-01
2.26436362e-01 -3.79132152e-01 2.07042843e-02 -4.52778131e-01
-7.83637166e-01 2.83972681e-01 7.22661009e-03 2.89184660e-01
9.28674161e-01 -1.84756890e-01 -4.14355546e-01 4.65683714e-02
-3.07100922e-01 -1.27722335e+00 -9.74149227e-01 8.85837004e-02
8.08338583e-01 -1.15477109e+00 6.62111938e-01 1.03799209e-01
-6.25615537e-01 8.75888884e-01 3.82619560e-01 -2.99639970e-01
6.18951738e-01 3.03058654e-01 -2.31067799e-02 -2.10835561e-01
-6.22589648e-01 -4.78718020e-02 1.64757207e-01 7.24956334e-01
7.61021614e-01 5.82175776e-02 -1.07505751e+00 7.75006592e-01
-5.13220727e-01 -5.27032733e-01 6.97021127e-01 9.24986720e-01
2.74744362e-01 -7.90721357e-01 -7.93953359e-01 4.70249444e-01
-6.00118577e-01 -1.41850218e-01 -6.42889678e-01 1.77693462e+00
-5.51766396e-01 1.28524411e+00 6.25922024e-01 -2.65679181e-01
2.04266205e-01 6.96691647e-02 1.79246560e-01 -1.00847371e-01
-5.45265853e-01 3.34236234e-01 3.16667825e-01 -1.58764333e-01
-7.04561591e-01 -1.09981859e+00 -9.52056348e-01 -8.34452152e-01
-6.41207099e-01 8.81137401e-02 8.03779602e-01 1.17901850e+00
-1.33883536e-01 2.09306508e-01 1.17214358e+00 -5.80244124e-01
-1.90152615e-01 -1.13347900e+00 -7.09249079e-01 -1.57295063e-01
7.45379150e-01 -8.62074733e-01 -8.92887175e-01 -2.33665884e-01]
|
[7.878166675567627, 5.125662326812744]
|
7bb35561-a6e5-4d77-8fec-c037a4f3c937
|
machine-translation-from-signed-to-spoken
|
2202.03086
| null |
https://arxiv.org/abs/2202.03086v4
|
https://arxiv.org/pdf/2202.03086v4.pdf
|
Machine Translation from Signed to Spoken Languages: State of the Art and Challenges
|
Automatic translation from signed to spoken languages is an interdisciplinary research domain, lying on the intersection of computer vision, machine translation and linguistics. Nevertheless, research in this domain is performed mostly by computer scientists in isolation. As the domain is becoming increasingly popular - the majority of scientific papers on the topic of sign language translation have been published in the past three years - we provide an overview of the state of the art as well as some required background in the different related disciplines. We give a high-level introduction to sign language linguistics and machine translation to illustrate the requirements of automatic sign language translation. We present a systematic literature review to illustrate the state of the art in the domain and then, harking back to the requirements, lay out several challenges for future research. We find that significant advances have been made on the shoulders of spoken language machine translation research. However, current approaches are often not linguistically motivated or are not adapted to the different input modality of sign languages. We explore challenges related to the representation of sign language data, the collection of datasets, the need for interdisciplinary research and requirements for moving beyond research, towards applications. Based on our findings, we advocate for interdisciplinary research and to base future research on linguistic analysis of sign languages. Furthermore, the inclusion of deaf and hearing end users of sign language translation applications in use case identification, data collection and evaluation is of the utmost importance in the creation of useful sign language translation models. We recommend iterative, human-in-the-loop, design and development of sign language translation models.
|
['Joni Dambre', 'Mieke Van Herreweghe', 'Dimitar Shterionov', 'Mathieu De Coster']
|
2022-02-07
| null | null | null | null |
['sign-language-translation']
|
['computer-vision']
|
[ 2.77578294e-01 -8.22449327e-02 -4.87199724e-01 -4.41402525e-01
-8.06130648e-01 -6.58867776e-01 5.87941766e-01 -4.49666917e-01
-4.40890282e-01 3.43102306e-01 7.55866826e-01 -5.50666988e-01
-8.76436755e-02 -1.81053922e-01 -2.09211200e-01 -2.03404903e-01
4.60059822e-01 5.13469219e-01 4.95164916e-02 -4.70046759e-01
1.94038883e-01 5.11154234e-01 -1.91035926e+00 2.77554154e-01
6.08313620e-01 3.82171780e-01 -1.52593508e-01 5.26687860e-01
-5.33453703e-01 4.77078319e-01 -3.56788754e-01 -3.31896573e-01
2.78365612e-01 -9.36270714e-01 -9.20914292e-01 2.46575132e-01
5.37193716e-01 -4.12649781e-01 5.24548031e-02 6.87195122e-01
9.67202723e-01 -3.12745810e-01 4.82421637e-01 -1.33228934e+00
-8.87460291e-01 2.97536552e-01 -3.75511535e-02 -2.97892630e-01
6.50529027e-01 2.47067407e-01 7.36627996e-01 -1.01478744e+00
1.02320409e+00 1.14812660e+00 4.24004674e-01 9.40290391e-01
-6.88386440e-01 -5.11894047e-01 1.46664605e-01 1.03895269e-01
-1.05908453e+00 -5.82718670e-01 5.72277248e-01 -6.43836558e-01
9.39861000e-01 2.04679176e-01 7.85602987e-01 1.04965675e+00
-5.88479996e-01 9.88997340e-01 1.35623991e+00 -1.13340402e+00
2.46865284e-02 2.18242183e-01 3.15609962e-01 5.36442280e-01
1.86910331e-01 2.27386057e-01 -8.00751448e-01 -7.88892210e-02
6.89344108e-01 -5.84459901e-01 -1.32015869e-01 -4.39824045e-01
-1.28606319e+00 4.37278748e-01 -3.02425057e-01 8.03976119e-01
-3.11569184e-01 -1.09127924e-01 5.73200107e-01 7.15880215e-01
7.54243135e-02 -2.88936328e-02 -3.66317451e-01 -6.40944064e-01
-9.60740507e-01 2.50729054e-01 1.09145761e+00 9.92499411e-01
1.50693819e-01 -3.61179002e-02 -9.40990448e-02 9.72070634e-01
7.73474157e-01 8.62662852e-01 5.44683576e-01 -7.47487128e-01
2.88210064e-01 7.17110991e-01 3.22916098e-02 -2.95519888e-01
-6.96931258e-02 1.61243096e-01 1.79902352e-02 5.10333121e-01
5.66271961e-01 -1.45136371e-01 -1.20702446e+00 1.31883359e+00
4.86484729e-02 -3.76297116e-01 7.48376250e-02 1.10656309e+00
9.21705902e-01 4.92737219e-02 2.20608264e-01 -1.24704853e-01
1.40136325e+00 -6.15292788e-01 -8.10724676e-01 -2.74305671e-01
6.01816773e-01 -1.18888390e+00 1.23706317e+00 1.43256530e-01
-1.02807999e+00 -9.93338600e-02 -6.42607689e-01 -2.38015294e-01
-4.43182677e-01 4.40830141e-01 3.14371079e-01 8.99058998e-01
-1.09998071e+00 -1.32279247e-01 -7.84230709e-01 -1.22812188e+00
2.30442047e-01 4.06432658e-01 -3.48521650e-01 -1.25495270e-01
-8.92262757e-01 1.47074068e+00 -1.53288797e-01 2.23060504e-01
-6.55982643e-02 -2.18249202e-01 -7.04605579e-01 -7.71128356e-01
-2.26284433e-02 -6.48292542e-01 1.57438934e+00 -1.18322456e+00
-1.79781044e+00 1.40128028e+00 -4.62778181e-01 1.32902250e-01
7.34154701e-01 -1.87512711e-01 -6.00431621e-01 -2.21887246e-01
4.23101969e-02 2.90171832e-01 4.01929915e-01 -1.20165980e+00
-9.25124466e-01 -3.81243825e-01 -3.20046991e-01 2.29641259e-01
2.18244065e-02 9.99189258e-01 -4.23820257e-01 -4.38705742e-01
1.94216967e-01 -9.77975428e-01 1.73037454e-01 4.59265292e-01
4.42585588e-01 -1.68657109e-01 8.98857713e-01 -7.93300688e-01
1.28711665e+00 -2.01847386e+00 1.47497147e-01 1.37721360e-01
-3.19977254e-01 6.68594599e-01 -2.05357477e-01 8.11942160e-01
1.96459100e-01 4.54948246e-02 -3.32268357e-01 -1.49326161e-01
2.96840757e-01 3.91863376e-01 -3.99147838e-01 4.62341994e-01
-1.39947221e-01 1.02333772e+00 -9.38854635e-01 -3.56552988e-01
4.70471203e-01 6.62992895e-01 -2.39886399e-02 -1.06421314e-01
8.79490525e-02 3.92216921e-01 -3.59289378e-01 9.47757244e-01
3.28043759e-01 3.98844093e-01 1.80427045e-01 3.51736136e-02
-6.06475174e-01 3.38079214e-01 -9.66214001e-01 1.65555322e+00
-5.93841970e-01 1.05437410e+00 3.39765251e-01 -6.58641398e-01
9.49270010e-01 7.49242008e-01 4.45652843e-01 -6.84681535e-01
3.52103263e-01 1.17172265e+00 2.18834564e-01 -9.96505857e-01
1.30100444e-01 -4.00189847e-01 1.93797484e-01 8.67746711e-01
-2.59627670e-01 -2.80301422e-01 3.27590466e-01 -4.94618922e-01
5.06491125e-01 7.06318915e-01 3.39037389e-01 1.55283436e-01
6.25380516e-01 3.10424805e-01 5.59213646e-02 2.29869872e-01
-5.72744429e-01 5.88990748e-01 -6.92138597e-02 -3.02158058e-01
-8.18902552e-01 -6.79752469e-01 -3.93862128e-02 1.11981261e+00
-2.85349041e-01 -1.37060165e-01 -7.91798890e-01 -3.95936817e-01
-2.03396007e-01 6.51658058e-01 -2.08077997e-01 2.70093471e-01
-9.16485786e-01 -3.01803261e-01 8.52196038e-01 4.98192221e-01
3.30430657e-01 -1.49496436e+00 -1.05361307e+00 1.63021654e-01
-2.41430089e-01 -1.20609474e+00 -2.06798509e-01 -5.49072087e-01
-8.26399803e-01 -1.07492256e+00 -1.14319277e+00 -1.36640871e+00
6.24700427e-01 6.54175431e-02 7.15437293e-01 2.26875637e-02
-2.00997934e-01 9.92441058e-01 -7.70500541e-01 -6.31340563e-01
-8.55727196e-01 -1.56842396e-01 -9.84368101e-02 -2.40326852e-01
1.10226393e+00 -3.49081784e-01 -8.76151547e-02 5.43067515e-01
-9.40606415e-01 -1.73770234e-01 6.47399068e-01 6.16131365e-01
1.15597188e-01 -8.24097753e-01 1.94115877e-01 -1.91070646e-01
1.00277948e+00 1.54048890e-01 -2.64454097e-01 6.28933072e-01
-4.62199867e-01 2.52958443e-02 -1.49976000e-01 -4.85501170e-01
-7.41597176e-01 8.81315693e-02 -8.37515518e-02 5.90606332e-02
-4.21755672e-01 6.88634336e-01 -7.20250010e-02 -3.14104915e-01
8.22951078e-01 3.90166223e-01 6.07945442e-01 -4.32868570e-01
5.24237156e-01 1.32013369e+00 2.81517237e-01 -6.17926955e-01
5.25748610e-01 4.71883714e-01 -2.81106859e-01 -1.12709224e+00
9.67791863e-03 -5.63805580e-01 -8.74790072e-01 -5.46209216e-01
4.61959749e-01 -4.44036812e-01 -2.46295050e-01 7.46285141e-01
-1.14937186e+00 -2.94793963e-01 -5.26059449e-01 7.20470488e-01
-8.13203394e-01 3.55846554e-01 -8.16604719e-02 -1.06358433e+00
-3.38446021e-01 -1.19466782e+00 1.11568189e+00 1.25787659e-02
-7.89381564e-01 -7.52765954e-01 2.75842428e-01 6.58417702e-01
6.35963857e-01 1.51587039e-01 7.59764314e-01 -2.08862990e-01
-2.49865994e-01 -3.67232382e-01 -1.54363215e-01 3.39815348e-01
3.59408408e-01 1.41364962e-01 -7.38241553e-01 1.11936174e-01
-3.72325212e-01 -2.96341419e-01 3.19418192e-01 2.70731807e-01
-1.61952585e-01 5.12556136e-02 -7.90475011e-02 -7.43894652e-02
1.00382912e+00 1.76895440e-01 5.53319693e-01 3.50032181e-01
3.42887729e-01 1.20432472e+00 6.93165183e-01 5.34055047e-02
5.63407481e-01 8.12190711e-01 -2.35069469e-01 -3.79307866e-02
-7.03121781e-01 -2.68356085e-01 6.97714686e-01 1.00186026e+00
-5.83771884e-01 1.26869678e-01 -1.21184838e+00 9.31686997e-01
-1.97755122e+00 -8.40244949e-01 -2.56438822e-01 2.22603798e+00
6.37022913e-01 -5.04940212e-01 4.87345636e-01 3.11556458e-01
4.05379474e-01 -2.47497439e-01 -9.23963413e-02 -6.65543437e-01
-1.55558303e-01 3.43801230e-01 3.74692023e-01 7.08179772e-01
-7.05510378e-01 1.20272851e+00 6.58229399e+00 5.57186529e-02
-1.53474247e+00 8.74400958e-02 -2.80478984e-01 9.66872051e-02
-3.61203760e-01 2.62996912e-01 -6.35271013e-01 1.60091370e-01
5.61819196e-01 -1.33361936e-01 4.49328899e-01 5.39142430e-01
7.01602161e-01 5.35128526e-02 -1.00991213e+00 1.04472685e+00
4.62056488e-01 -7.60546446e-01 -3.29251364e-02 1.61923692e-01
4.08653498e-01 2.46663705e-01 -1.41114309e-01 5.97034730e-02
8.89993310e-02 -8.64511728e-01 9.32371855e-01 2.30902404e-01
9.75870669e-01 -1.14722634e-02 6.72185719e-01 1.90688714e-01
-1.00666392e+00 1.60910159e-01 2.66777366e-01 -2.78281957e-01
6.18472576e-01 -2.06896514e-01 -5.62691152e-01 2.69093096e-01
4.34753120e-01 4.64084238e-01 -7.55266994e-02 1.12355733e+00
-4.73685920e-01 6.84561193e-01 -5.14652073e-01 -5.64233124e-01
8.44754353e-02 -2.84438074e-01 4.44890350e-01 1.35814226e+00
4.61657465e-01 8.34309235e-02 4.41161692e-02 4.74618822e-01
5.50780833e-01 6.31265700e-01 -6.93558693e-01 -4.07987654e-01
1.45525292e-01 5.08492768e-01 -6.28263116e-01 -3.89690548e-02
-8.05111051e-01 9.96265531e-01 -2.82925308e-01 5.30016303e-01
-2.69861013e-01 -1.85369432e-01 7.46447086e-01 3.70583117e-01
-3.87785211e-03 -3.65240723e-01 -5.25338352e-01 -9.22877669e-01
4.92326796e-01 -1.35384238e+00 1.77889019e-01 -5.76215625e-01
-1.12348938e+00 4.44051117e-01 1.69065595e-01 -1.58774114e+00
-6.46151304e-01 -9.31214571e-01 -2.16107205e-01 1.02168822e+00
-1.42312920e+00 -1.75610507e+00 -1.33786127e-01 2.72565156e-01
5.10188222e-01 -1.24220967e-01 1.08318114e+00 4.43270028e-01
1.62321880e-01 3.73207986e-01 -1.79706842e-01 2.37664640e-01
7.49871671e-01 -5.08798361e-01 3.37918133e-01 7.10538447e-01
2.47885481e-01 6.83379531e-01 5.60244262e-01 -7.06139207e-01
-1.33176172e+00 -2.39044085e-01 1.78489017e+00 -5.91421843e-01
7.05835223e-01 -2.31932886e-02 -4.17054504e-01 5.21365821e-01
2.60252595e-01 -2.97886223e-01 8.35411072e-01 -1.41873628e-01
-2.99689114e-01 2.26610482e-01 -1.05148721e+00 8.72572303e-01
1.30441689e+00 -8.32832515e-01 -8.86000693e-01 1.96726888e-01
-9.17438418e-02 -3.12174767e-01 -3.95205438e-01 2.05893248e-01
1.25576973e+00 -5.10454535e-01 5.85525036e-01 -6.94220543e-01
3.70007791e-02 -5.19026220e-01 -2.36821055e-01 -8.00600290e-01
2.12074384e-01 -7.61456490e-01 3.60652119e-01 9.93346989e-01
5.08945227e-01 -6.16005063e-01 7.27403224e-01 1.08615470e+00
-2.22642384e-02 -4.35360163e-01 -9.49405551e-01 -6.80369258e-01
1.90777853e-01 -8.30475032e-01 2.54039437e-01 7.59958148e-01
3.17510277e-01 1.58747226e-01 -2.49522135e-01 -1.78749695e-01
2.72616208e-01 5.63283823e-02 1.03492439e+00 -1.18357885e+00
1.97608843e-01 -8.68411362e-01 -6.93028986e-01 -8.53670776e-01
-1.54808303e-02 -7.98718452e-01 1.18516237e-01 -2.15276265e+00
-3.06759655e-01 2.70307004e-01 3.05446148e-01 6.94128811e-01
3.71382117e-01 2.05452472e-01 4.35610592e-01 2.97854394e-01
1.96648657e-01 8.73442367e-02 1.22860742e+00 8.10212344e-02
-5.08358657e-01 2.08533660e-01 -5.56404293e-01 7.05003023e-01
6.03514373e-01 -9.17921588e-02 -2.51567781e-01 -6.27080977e-01
1.00051120e-01 -5.32704949e-01 3.06426078e-01 -5.79986691e-01
2.89619595e-01 -2.66411424e-01 -5.96345603e-01 -4.02366638e-01
3.88263091e-02 -1.14023507e+00 -7.39425793e-02 3.78510356e-01
-1.03987105e-01 -5.83930388e-02 1.65601268e-01 -1.80476904e-01
-4.38633502e-01 -2.97325969e-01 5.18277347e-01 -1.59423247e-01
-1.01191926e+00 -2.22654417e-01 -6.54593050e-01 5.03608584e-02
8.97895038e-01 -7.99138308e-01 1.01181969e-01 -5.61612129e-01
-6.95255160e-01 1.26491100e-01 5.13643742e-01 8.39217663e-01
5.39594531e-01 -1.19009757e+00 -8.73360634e-01 3.40964258e-01
5.17581880e-01 -5.48681974e-01 -2.81880140e-01 8.40410650e-01
-8.03143620e-01 5.06314278e-01 -2.68294126e-01 -2.66012669e-01
-1.68714869e+00 -2.70665288e-01 2.17941582e-01 3.46520931e-01
-4.15958881e-01 4.81819928e-01 -5.90877950e-01 -7.09152520e-01
4.41455841e-01 -4.59064931e-01 -1.45506144e-01 1.28690824e-01
3.68615210e-01 5.26318789e-01 1.39870504e-02 -1.32063055e+00
-5.51306307e-01 1.21610856e+00 3.84307325e-01 -9.47762728e-01
1.06543493e+00 -1.38019785e-01 -3.51955295e-01 5.47555864e-01
8.20945859e-01 2.97763813e-02 -5.14042318e-01 -2.09393337e-01
2.92194843e-01 -3.36752564e-01 -2.20849887e-01 -1.20928442e+00
-5.73765874e-01 8.08598757e-01 8.22438180e-01 -4.78342593e-01
1.00716186e+00 1.50494188e-01 5.38146794e-01 2.90744454e-01
6.15094185e-01 -1.47879481e+00 -5.48431158e-01 6.50106370e-01
1.23489439e+00 -1.11248648e+00 -9.43756923e-02 -4.40042555e-01
-7.45205700e-01 1.19414067e+00 5.60902618e-02 2.05514804e-01
8.55622292e-01 3.90315443e-01 1.09596825e+00 -1.75867349e-01
-2.56612580e-02 -5.74403405e-01 4.54285741e-01 8.76150966e-01
9.01762247e-01 5.71127199e-02 -1.25534368e+00 2.21801221e-01
-3.14444065e-01 8.13332558e-01 1.55566782e-01 1.47436225e+00
-2.80102998e-01 -1.87276673e+00 -6.12266719e-01 1.49180517e-01
-1.22320175e-01 5.05989864e-02 -1.13304508e+00 9.71858084e-01
2.46472925e-01 9.91278231e-01 -5.02718329e-01 -1.40311331e-01
9.14270461e-01 4.98868197e-01 6.85632586e-01 -5.34117341e-01
-6.95116699e-01 -6.34486750e-02 3.31172168e-01 -1.56501845e-01
-9.64138687e-01 -1.19218409e+00 -1.27647853e+00 9.38533321e-02
-1.62776709e-01 -2.47230634e-01 1.16052008e+00 1.21695745e+00
3.25192869e-01 -1.59806639e-01 -1.67015642e-01 -4.34226155e-01
-5.66440403e-01 -9.30717587e-01 -2.07870066e-01 2.72313774e-01
1.36151969e-01 -3.06643963e-01 -2.64468998e-01 3.20628315e-01]
|
[9.101709365844727, -6.406668663024902]
|
d8074122-2dc2-4663-ae31-d7f84e53116b
|
environmental-sound-classification-on-the
|
2103.03483
| null |
https://arxiv.org/abs/2103.03483v4
|
https://arxiv.org/pdf/2103.03483v4.pdf
|
Environmental Sound Classification on the Edge: A Pipeline for Deep Acoustic Networks on Extremely Resource-Constrained Devices
|
Significant efforts are being invested to bring state-of-the-art classification and recognition to edge devices with extreme resource constraints (memory, speed, and lack of GPU support). Here, we demonstrate the first deep network for acoustic recognition that is small, flexible and compression-friendly yet achieves state-of-the-art performance for raw audio classification. Rather than handcrafting a once-off solution, we present a generic pipeline that automatically converts a large deep convolutional network via compression and quantization into a network for resource-impoverished edge devices. After introducing ACDNet, which produces above state-of-the-art accuracy on ESC-10 (96.65%), ESC-50 (87.10%), UrbanSound8K (84.45%) and AudioEvent (92.57%), we describe the compression pipeline and show that it allows us to achieve 97.22% size reduction and 97.28% FLOP reduction while maintaining close to state-of-the-art accuracy 96.25%, 83.65%, 78.27% and 89.69% on these datasets. We describe a successful implementation on a standard off-the-shelf microcontroller and, beyond laboratory benchmarks, report successful tests on real-world datasets.
|
['Bernd Meyer', 'Ian Thomas West', 'Christoph Bergmeir', 'Md Mohaimenuzzaman']
|
2021-03-05
| null | null | null | null |
['environmental-sound-classification', 'sound-classification']
|
['audio', 'audio']
|
[ 4.74017747e-02 -8.41354951e-02 8.71095136e-02 -2.96781301e-01
-9.29910660e-01 -2.71716297e-01 -1.31049184e-02 -9.54051390e-02
-4.07175899e-01 3.14974040e-01 -1.36666983e-01 -4.62960362e-01
6.21235259e-02 -6.66765153e-01 -5.82867265e-01 -3.81668210e-01
-3.97219688e-01 8.12817551e-03 1.28517225e-01 7.06663653e-02
-1.95135161e-01 4.68927622e-01 -1.71231782e+00 4.70984370e-01
1.17802136e-01 1.72821867e+00 -1.29763439e-01 1.24193287e+00
3.06637913e-01 6.35557055e-01 -1.05694389e+00 -3.90665561e-01
2.31023580e-01 7.00400919e-02 -5.04210055e-01 -4.39587921e-01
6.92029595e-01 -6.02036953e-01 -6.69162631e-01 6.09838128e-01
1.15943599e+00 -4.38960791e-01 4.23156500e-01 -1.36707211e+00
-3.17405611e-01 8.08544815e-01 -4.65652972e-01 1.28888473e-01
1.52979866e-01 -7.88809452e-03 8.00153136e-01 -7.27370620e-01
9.20149013e-02 6.77914739e-01 1.27043295e+00 4.65976775e-01
-9.54668462e-01 -1.30594993e+00 -6.12969577e-01 2.22647116e-01
-1.79406834e+00 -1.02454722e+00 5.99796414e-01 1.20403811e-01
1.63651645e+00 2.26607084e-01 8.85290444e-01 9.53909218e-01
1.05409056e-01 5.05851626e-01 6.09499633e-01 -4.29822266e-01
4.54688072e-01 -1.87652424e-01 5.66607900e-02 5.99004567e-01
2.68716812e-01 -4.85335551e-02 -1.04091120e+00 -8.99171457e-02
4.34469193e-01 -4.34661925e-01 -1.73081636e-01 3.58474016e-01
-6.55762255e-01 2.61697233e-01 1.12615995e-01 1.60287768e-01
-2.21531704e-01 8.23507428e-01 7.52776682e-01 2.26902112e-01
1.88171446e-01 -5.94968582e-03 -7.22501218e-01 -1.02319419e+00
-1.27415168e+00 -7.05783218e-02 9.83435333e-01 1.19374645e+00
2.57393599e-01 7.79312909e-01 2.38491029e-01 8.32395136e-01
1.92356408e-01 6.97173178e-01 6.41507804e-01 -9.59082365e-01
3.95506680e-01 -1.01214461e-01 -3.29174966e-01 -7.41285503e-01
-6.84279263e-01 -7.76368916e-01 -1.02928674e+00 2.18878210e-01
1.03326626e-01 -3.03963482e-01 -8.84788692e-01 1.67215371e+00
-1.33371830e-01 3.49782795e-01 3.30750287e-01 6.32586479e-01
1.14756823e+00 6.21351182e-01 -1.60187349e-01 1.04256690e-01
1.30761230e+00 -6.47275865e-01 -4.57110137e-01 -1.61996052e-01
6.04877770e-01 -8.24771345e-01 1.05510199e+00 9.90458190e-01
-1.16733050e+00 -6.87368155e-01 -1.49948132e+00 4.77028340e-02
1.02438973e-02 4.79079574e-01 6.05295300e-01 1.20787954e+00
-1.32752395e+00 6.86708272e-01 -1.01547229e+00 -2.17453703e-01
5.77484488e-01 7.59114385e-01 -2.01770943e-02 2.16264918e-01
-8.89191985e-01 2.77604967e-01 8.33378807e-02 -2.69129008e-01
-9.26311135e-01 -8.60590994e-01 -3.71503353e-01 4.12962615e-01
-1.12681285e-01 -4.22095120e-01 1.53935564e+00 -6.47931933e-01
-1.93452716e+00 4.77566034e-01 2.48860002e-01 -8.33832860e-01
1.35811632e-02 -3.12115967e-01 -1.06413984e+00 1.90012857e-01
-5.40243447e-01 8.24871719e-01 7.71028042e-01 -4.32166040e-01
-9.23642933e-01 -8.34241360e-02 -2.55551577e-01 -2.57198095e-01
-8.56072903e-01 3.96523029e-02 -3.57581139e-01 -5.63703895e-01
-6.48343414e-02 -9.92029011e-01 4.04168785e-01 1.94847196e-01
-1.41192675e-01 1.02403708e-01 1.01080453e+00 -6.22732282e-01
1.35220647e+00 -2.33395958e+00 -5.85718632e-01 -2.86363978e-02
1.69450030e-01 7.03877091e-01 -3.79095748e-02 9.51836556e-02
-4.72973175e-02 1.05472021e-01 6.21597320e-02 -5.35550177e-01
1.27848804e-01 -1.05131388e-01 -4.97065745e-02 4.06540245e-01
-1.63504668e-02 5.03536463e-01 -2.06850544e-01 2.48554535e-02
2.95020938e-01 8.70444655e-01 -8.15911651e-01 -6.61329180e-02
5.08705854e-01 -2.58928418e-01 -1.96575709e-02 8.86912644e-01
7.17220187e-01 1.84819713e-01 3.41947041e-02 -6.00381613e-01
-3.14698555e-02 5.90863228e-01 -1.38683748e+00 1.77465880e+00
-7.74794519e-01 1.22190821e+00 4.41466540e-01 -8.13044190e-01
1.05619371e+00 4.01712149e-01 3.53863478e-01 -7.37952650e-01
4.39244479e-01 5.29896080e-01 1.29302949e-01 -2.98025072e-01
5.90923488e-01 1.62236139e-01 -3.05200905e-01 3.24424446e-01
4.28831667e-01 -3.42289031e-01 -2.86368221e-01 3.26953921e-03
1.36019671e+00 -4.30191338e-01 -4.07001600e-02 -3.42451155e-01
4.50591147e-02 -4.96635765e-01 3.54803801e-01 6.34378672e-01
-3.29839677e-01 7.08084404e-01 -5.17067325e-04 -3.77268404e-01
-1.07347989e+00 -8.97842348e-01 -3.07285041e-01 9.66820180e-01
-4.97692376e-01 -9.48301077e-01 -9.38792586e-01 2.57488281e-01
-2.42431492e-01 4.29843217e-01 3.97151373e-02 -2.71183074e-01
-3.65557522e-01 -7.06896305e-01 1.59980917e+00 7.23304093e-01
8.78545284e-01 -5.19796431e-01 -1.10902178e+00 2.87485391e-01
3.45301986e-01 -1.38049543e+00 -1.23217823e-02 6.21696889e-01
-6.57076716e-01 -4.73355263e-01 -2.57179946e-01 -8.58836651e-01
-7.53302500e-02 -1.35686085e-01 1.28268337e+00 -1.22444324e-01
-4.39901948e-01 3.53037387e-01 -2.47839227e-01 -6.50350928e-01
-2.46054709e-01 1.93530768e-01 4.87700850e-01 -4.71573859e-01
3.31692040e-01 -9.36128795e-01 -4.99733239e-01 6.47098199e-03
-5.14534056e-01 -2.51385182e-01 4.64345574e-01 5.12592375e-01
5.10974824e-01 7.78706521e-02 5.40462613e-01 -5.87529652e-02
4.71525609e-01 3.75271663e-02 -4.84790295e-01 -1.42310381e-01
-6.27987921e-01 -2.96105206e-01 8.38702619e-01 -5.25236964e-01
-3.09921652e-01 4.64210153e-01 -7.75351763e-01 -4.91882801e-01
-1.75162628e-01 7.40032271e-02 -1.11413948e-01 -3.74337465e-01
5.14336646e-01 7.92101473e-02 -2.69503415e-01 -3.00413370e-01
5.38899750e-02 1.51099682e+00 9.46922779e-01 -2.01013803e-01
2.30343923e-01 3.03326935e-01 -2.64180265e-02 -1.20148015e+00
-1.82548210e-01 -5.67182936e-02 -1.90658912e-01 -1.74316034e-01
5.37350357e-01 -1.42172897e+00 -9.93082821e-01 8.92308652e-01
-9.38912272e-01 -3.95208359e-01 -3.54820907e-01 5.28243959e-01
-4.15975720e-01 -9.01405327e-03 -7.57245183e-01 -8.18632960e-01
-1.16816878e+00 -1.03057051e+00 1.19251037e+00 2.03502849e-01
-3.63974273e-01 -3.14587176e-01 -3.57601732e-01 -8.81045386e-02
7.71082103e-01 -1.03630066e-01 2.49300763e-01 -5.22507608e-01
-2.85193622e-01 -3.77812654e-01 -5.43641001e-02 4.83245850e-01
-2.47811899e-01 1.51851729e-01 -1.60700858e+00 -3.22799712e-01
9.33754444e-03 -4.61244881e-01 6.11561775e-01 3.85921061e-01
1.49719393e+00 -1.04535613e-02 -1.04331449e-01 1.08415473e+00
1.14757562e+00 3.12963426e-01 8.23996127e-01 1.02287352e-01
4.09927487e-01 -3.61831963e-01 1.07775427e-01 8.74918640e-01
1.44939810e-01 6.77241027e-01 4.09475505e-01 2.02376917e-01
-5.05676985e-01 4.23597693e-02 3.92138392e-01 1.23930860e+00
3.14304262e-01 -3.90699029e-01 -8.62591743e-01 3.48677665e-01
-1.00375330e+00 -6.10522807e-01 -1.35551244e-01 1.93256605e+00
6.89161777e-01 5.63472033e-01 1.18670255e-01 9.24570382e-01
4.37846124e-01 -4.90451558e-03 -5.94489932e-01 -8.79415989e-01
-6.66098222e-02 9.02763665e-01 6.44933760e-01 1.97582290e-01
-1.06172490e+00 5.51712215e-01 6.52886152e+00 1.13466096e+00
-1.50195336e+00 2.00807229e-01 8.60260129e-01 -5.50906241e-01
3.61955404e-01 -7.24758804e-01 -9.64592695e-01 2.93254733e-01
1.88345242e+00 6.24777339e-02 4.33632374e-01 1.03551543e+00
-1.83065727e-01 1.80260196e-01 -1.16686916e+00 1.71990740e+00
-1.19528687e-02 -1.36139905e+00 -5.38041651e-01 -8.11798647e-02
3.14181238e-01 2.93700695e-01 1.54486060e-01 3.28203261e-01
-1.53515693e-02 -9.67589259e-01 1.03245461e+00 6.67694509e-02
1.58248568e+00 -1.15424061e+00 7.58863211e-01 6.91015497e-02
-1.56334543e+00 -2.01944023e-01 -4.97820377e-01 -3.18664372e-01
-1.17945947e-01 9.84694779e-01 -6.64562285e-01 2.33591031e-02
1.18844402e+00 4.34505701e-01 -3.80892247e-01 7.20837653e-01
2.02588052e-01 1.01079094e+00 -1.00173295e+00 -2.51546979e-01
2.42290050e-02 5.61131716e-01 6.60581961e-02 1.63027525e+00
6.69725299e-01 2.59531647e-01 -2.76550651e-01 1.45040795e-01
-3.19348603e-01 -2.70323962e-01 -2.24644646e-01 2.39508703e-01
8.71310234e-01 1.24384189e+00 -4.43886340e-01 -3.95430297e-01
-1.77650854e-01 7.39941061e-01 6.30415082e-02 -2.13650048e-01
-1.10570955e+00 -9.13461089e-01 9.16977346e-01 -2.26169050e-01
5.40774524e-01 -3.04344684e-01 -4.68153745e-01 -6.72520936e-01
8.29705000e-02 -9.32102382e-01 -4.90382686e-02 -8.02686453e-01
-6.99814498e-01 9.25309122e-01 -4.77821529e-01 -1.24818182e+00
-3.29206467e-01 -5.87774694e-01 -3.90830725e-01 5.27480423e-01
-1.18068218e+00 -7.09767938e-01 -6.33016229e-01 3.29545319e-01
3.49057049e-01 -4.38644111e-01 1.17228532e+00 8.98663640e-01
-4.15836781e-01 1.37997687e+00 2.76911199e-01 1.12629704e-01
4.06389952e-01 -7.44148254e-01 7.27792025e-01 5.52831709e-01
1.24683276e-01 8.38392675e-02 5.48174322e-01 -5.08423485e-02
-1.84534037e+00 -1.12968183e+00 5.32786548e-01 2.38849372e-01
6.25189304e-01 -7.40950644e-01 -4.74514425e-01 1.23948537e-01
2.61059433e-01 3.50432426e-01 8.21023941e-01 -1.10893227e-01
-3.27027082e-01 -7.30254650e-01 -1.31771457e+00 4.54090089e-01
1.22623038e+00 -8.39807987e-01 1.54143244e-01 3.18529387e-03
7.04493761e-01 -7.16733754e-01 -1.01906669e+00 3.10658604e-01
1.08685768e+00 -9.59598541e-01 8.60929191e-01 3.14821191e-02
1.45939425e-01 3.30861891e-03 -5.77538252e-01 -1.05604959e+00
-1.95954621e-01 -8.34973156e-01 -4.48167801e-01 1.42964530e+00
4.42524880e-01 -5.34093499e-01 1.07479668e+00 3.42030078e-01
-6.95932150e-01 -8.94062221e-01 -1.43556845e+00 -8.82157683e-01
-1.01963371e-01 -1.23285329e+00 6.80638909e-01 5.12976706e-01
2.00810641e-01 1.15943879e-01 -2.87061334e-01 1.92779317e-01
3.03586662e-01 -3.86808723e-01 6.75819099e-01 -9.33348417e-01
-5.24070501e-01 -2.14239642e-01 -1.04255176e+00 -8.73019814e-01
-2.03599095e-01 -4.41043496e-01 -1.21700533e-01 -1.00142407e+00
-3.07852566e-01 -3.99451315e-01 -2.02761725e-01 6.20502591e-01
7.54666328e-01 8.53375316e-01 2.17352033e-01 -2.69440323e-01
-4.90092695e-01 1.85537919e-01 3.26432943e-01 -2.48773053e-01
-9.71849859e-02 -2.83781469e-01 -6.33380353e-01 6.52517200e-01
1.12978530e+00 -4.73053843e-01 -3.22878599e-01 -6.35178506e-01
-2.18539406e-02 -1.01715438e-01 1.40678734e-01 -2.13305831e+00
3.68695021e-01 6.49030805e-01 4.40595627e-01 -4.61336434e-01
7.38274157e-01 -8.69535029e-01 3.76700521e-01 5.89642882e-01
-1.50964394e-01 1.51202247e-01 7.58536339e-01 3.11655793e-02
-1.48668364e-01 6.69165235e-03 8.67055357e-01 4.46687520e-01
-6.34910226e-01 6.37648255e-02 -5.04781723e-01 6.78822920e-02
6.59446478e-01 -1.93845913e-01 -4.31383431e-01 -5.43972731e-01
-5.95316827e-01 -3.90428901e-01 -3.90142277e-02 1.37619287e-01
6.12329423e-01 -1.29107451e+00 -5.95857680e-01 4.34414357e-01
-2.42363051e-01 -1.97516054e-01 2.12007597e-01 3.29994708e-01
-7.94476449e-01 5.92085779e-01 -2.12517679e-01 -8.27774942e-01
-1.53003097e+00 -9.06844512e-02 6.38172626e-01 4.83980924e-02
-6.39372408e-01 9.86944258e-01 -6.46157026e-01 1.55668482e-01
6.73930705e-01 -7.78411567e-01 2.74397045e-01 -1.46558493e-01
6.74074948e-01 6.30107105e-01 8.49399805e-01 -2.71696061e-01
-6.35654569e-01 7.85451531e-01 3.81592333e-01 -1.93797536e-02
1.35451424e+00 2.52754688e-01 4.88225549e-01 2.07732499e-01
1.55306053e+00 -5.06308302e-02 -1.07122159e+00 3.58462065e-01
-6.39562905e-01 -1.36962369e-01 5.49358010e-01 -7.67936230e-01
-1.48133564e+00 1.13330948e+00 1.38032138e+00 1.94671914e-01
1.69927168e+00 -2.17027619e-01 1.20992029e+00 5.98399758e-01
5.12854815e-01 -1.26236522e+00 -9.26153958e-02 5.36089540e-01
4.00655061e-01 -5.66256106e-01 1.81820393e-01 -3.11617434e-01
-8.49039946e-03 1.21462893e+00 3.17077994e-01 -1.04509912e-01
1.01412690e+00 1.30899549e+00 -2.03412846e-01 9.94565487e-02
-9.74256873e-01 1.29852608e-01 -1.70573801e-01 7.92336464e-01
5.58654726e-01 2.81492621e-01 3.78894694e-02 1.00210190e+00
-7.95948088e-01 3.38239253e-01 6.26531720e-01 1.08513689e+00
-4.73828614e-01 -6.08722806e-01 -2.37691239e-01 6.62162960e-01
-8.31478477e-01 -3.05700451e-01 -1.57664835e-01 5.48263967e-01
2.22114429e-01 1.09633017e+00 6.12382412e-01 -1.17919791e+00
3.57776910e-01 2.78011691e-02 2.43718892e-01 -1.05548434e-01
-8.13131809e-01 1.65052846e-01 5.29167354e-01 -4.79505062e-01
6.03654236e-03 -5.38372576e-01 -1.30414081e+00 -7.49293804e-01
-3.30487669e-01 -2.17556104e-01 1.19090772e+00 4.51574057e-01
9.20340598e-01 9.81728911e-01 3.20152611e-01 -8.72893572e-01
-3.22572380e-01 -9.55915749e-01 -7.39872634e-01 -2.98248082e-01
3.92672300e-01 -2.12921411e-01 -3.79558444e-01 -3.78514230e-02]
|
[14.583333969116211, 5.513881683349609]
|
62193e35-3427-4974-85a3-2c802dc736d9
|
textir-a-simple-framework-for-text-based
|
2302.14736
| null |
https://arxiv.org/abs/2302.14736v1
|
https://arxiv.org/pdf/2302.14736v1.pdf
|
TextIR: A Simple Framework for Text-based Editable Image Restoration
|
Most existing image restoration methods use neural networks to learn strong image-level priors from huge data to estimate the lost information. However, these works still struggle in cases when images have severe information deficits. Introducing external priors or using reference images to provide information also have limitations in the application domain. In contrast, text input is more readily available and provides information with higher flexibility. In this work, we design an effective framework that allows the user to control the restoration process of degraded images with text descriptions. We use the text-image feature compatibility of the CLIP to alleviate the difficulty of fusing text and image features. Our framework can be used for various image restoration tasks, including image inpainting, image super-resolution, and image colorization. Extensive experiments demonstrate the effectiveness of our method.
|
['Zhi Wang', 'Chun Yuan', 'Chao Dong', 'Shuzhao Xie', 'Cairong Wang', 'Yunpeng Bai']
|
2023-02-28
| null | null | null | null |
['colorization', 'image-inpainting']
|
['computer-vision', 'computer-vision']
|
[ 5.11227608e-01 -3.98501217e-01 -2.02751309e-01 -3.56954575e-01
-4.77951974e-01 -1.31215677e-01 1.31143823e-01 -4.52939779e-01
-2.30343372e-01 6.85339749e-01 4.02943075e-01 1.56090841e-01
9.23124328e-02 -7.55108237e-01 -5.84484577e-01 -8.00756454e-01
6.65825605e-01 -2.51903802e-01 1.86467007e-01 -3.15763593e-01
4.60105419e-01 4.73121792e-01 -1.55370331e+00 5.03296316e-01
1.25871110e+00 8.08006227e-01 7.57503688e-01 3.07874233e-01
-2.73402423e-01 7.50069022e-01 -6.60221815e-01 -9.69034433e-02
5.43226719e-01 -3.97934824e-01 -3.35550189e-01 4.95570630e-01
4.49504584e-01 -1.01797652e+00 -7.09107995e-01 1.41272819e+00
5.58315158e-01 1.70517594e-01 1.73123315e-01 -1.05018449e+00
-9.00248230e-01 3.10979009e-01 -8.61279845e-01 1.51299790e-01
5.01002491e-01 1.03797026e-01 2.93174803e-01 -1.00249255e+00
6.27265394e-01 1.39464080e+00 5.22910774e-01 4.31649119e-01
-1.16311443e+00 -7.22767651e-01 1.12236962e-01 4.03437674e-01
-1.07472861e+00 -7.11373091e-01 1.14322221e+00 -8.08898136e-02
4.80144769e-01 2.37441242e-01 3.78705084e-01 8.98145974e-01
1.05830118e-01 8.56680989e-01 1.26299131e+00 -5.29276729e-01
-4.84296195e-02 -2.12350860e-02 -1.33346707e-01 2.91794956e-01
8.48629400e-02 1.12366535e-01 -7.14306951e-01 -3.90704162e-03
1.20573759e+00 4.10292953e-01 -8.21707010e-01 1.33704813e-02
-1.04920208e+00 5.44941187e-01 3.01565230e-01 1.32629231e-01
-3.26388389e-01 -8.92670974e-02 7.55955353e-02 3.92630100e-01
4.99092817e-01 1.16269574e-01 -2.01775923e-01 2.29771137e-01
-1.12858295e+00 3.32081541e-02 3.87728095e-01 8.03206146e-01
7.65798211e-01 2.56945401e-01 -1.88547134e-01 1.21926320e+00
1.55964211e-01 4.63182658e-01 5.33317387e-01 -1.38696790e+00
4.93264675e-01 2.18152866e-01 3.38874817e-01 -1.13206720e+00
-4.26764740e-03 -1.04420342e-01 -1.19164562e+00 6.04585767e-01
1.18087262e-01 7.01421499e-02 -1.18893456e+00 1.45809937e+00
7.12954532e-03 5.06017804e-02 -3.94860767e-02 1.09205914e+00
6.44171536e-01 7.79704154e-01 -3.50132436e-01 -3.64353210e-01
1.02358758e+00 -1.05416310e+00 -1.29226816e+00 -4.02577013e-01
-2.50800222e-01 -1.06085873e+00 1.22139084e+00 5.01852214e-01
-1.25196028e+00 -8.31379831e-01 -1.13970542e+00 -3.76533031e-01
-7.57449418e-02 1.97536185e-01 3.45142573e-01 3.70557487e-01
-1.15449548e+00 7.17678785e-01 -6.75038695e-01 -1.37186036e-01
4.40784663e-01 2.64712334e-01 -4.19440657e-01 -6.01516008e-01
-1.08213007e+00 7.92454243e-01 3.70816976e-01 2.01960728e-01
-7.05763638e-01 -5.46760499e-01 -7.95592308e-01 2.09470876e-02
3.27696860e-01 -5.74015081e-01 7.78101921e-01 -1.22631514e+00
-1.68034351e+00 5.60482621e-01 -1.92107499e-01 -9.48719084e-02
6.54191256e-01 -4.40117687e-01 -3.27397048e-01 3.48746687e-01
5.85822649e-02 6.29357696e-01 1.34166348e+00 -1.46261930e+00
-4.08385307e-01 -2.41477445e-01 1.39312401e-01 3.87064427e-01
-3.39011580e-01 7.47841373e-02 -9.05426502e-01 -1.03558862e+00
4.68000025e-01 -3.64562660e-01 -2.03791872e-01 5.67107856e-01
-3.12258840e-01 4.37073708e-01 1.25686765e+00 -1.09538555e+00
8.50542963e-01 -2.32475376e+00 -2.48859487e-02 -2.04665512e-01
1.23619393e-01 2.30535924e-01 -3.12896192e-01 4.29295957e-01
5.14251553e-02 -4.79934067e-02 -2.97467947e-01 -3.91802281e-01
-4.01036590e-01 3.64835858e-01 -4.15435225e-01 3.37663025e-01
3.87984253e-02 4.92572844e-01 -4.74494994e-01 -5.75332701e-01
5.17036200e-01 7.92585075e-01 -4.97611195e-01 3.97068888e-01
-1.72982793e-02 5.37698030e-01 -3.09245914e-01 6.47662461e-01
1.13015819e+00 -8.25530738e-02 -1.21347897e-01 -5.65943956e-01
7.83921704e-02 -2.67220706e-01 -1.20235860e+00 1.91182876e+00
-4.20470774e-01 7.79686034e-01 5.91059566e-01 -8.04033756e-01
8.55339587e-01 3.37230116e-02 4.37769920e-01 -8.69757593e-01
-4.97915260e-02 -1.49305314e-01 -3.16047668e-01 -7.87945330e-01
5.75202048e-01 -6.51929155e-02 6.29604101e-01 3.88423741e-01
-1.77377060e-01 -1.77529439e-01 1.73313782e-01 4.98812012e-02
6.93696201e-01 3.79870236e-01 2.80445963e-02 1.37872010e-01
4.21081841e-01 -3.27905148e-01 7.02152729e-01 7.63572395e-01
-1.24470666e-01 1.15060043e+00 6.31985068e-02 -3.04745823e-01
-1.12071109e+00 -9.17978466e-01 -2.10286751e-01 8.97304296e-01
5.47177315e-01 -2.41215885e-01 -6.49530768e-01 -2.45508760e-01
-3.14616948e-01 4.94592816e-01 -3.29169422e-01 -5.98850474e-03
-6.13719165e-01 -7.16639698e-01 9.53195766e-02 4.52682912e-01
9.74975884e-01 -9.65776742e-01 -3.28073502e-01 1.20778836e-01
-7.35362172e-01 -1.13647437e+00 -7.60007977e-01 -1.34416863e-01
-1.04570687e+00 -8.74313354e-01 -8.47077847e-01 -9.55186188e-01
1.02964270e+00 8.00691307e-01 7.60128975e-01 3.27541262e-01
-3.72824192e-01 1.30839229e-01 -2.62565225e-01 6.71821460e-02
-2.94618994e-01 -6.11398458e-01 -1.96739942e-01 4.44008857e-02
-1.91739038e-01 -8.13667357e-01 -8.36831629e-01 3.54455918e-01
-1.38057065e+00 4.61506814e-01 6.38016403e-01 1.18170321e+00
5.42623281e-01 5.08446395e-01 2.99638331e-01 -6.34107947e-01
8.06216717e-01 -3.51632424e-02 -4.79517341e-01 1.74636096e-01
-4.86638576e-01 4.40854318e-02 6.86074674e-01 -5.84033430e-01
-1.61322594e+00 4.96689864e-02 -9.61366966e-02 -5.13655663e-01
-1.65308580e-01 3.74544352e-01 -4.45786208e-01 -2.75941998e-01
4.65173244e-01 4.05826837e-01 1.91171616e-01 -7.86269784e-01
2.56681770e-01 6.57605469e-01 8.21098387e-01 -4.98237371e-01
7.63488948e-01 6.50061071e-01 -2.25861669e-01 -6.05496943e-01
-7.10446715e-01 -1.92808017e-01 -5.83419085e-01 -9.65055078e-02
5.10718524e-01 -1.02252424e+00 -3.80029738e-01 6.67149365e-01
-1.16929734e+00 -2.38822788e-01 -1.11693159e-01 3.53567421e-01
-4.44024324e-01 8.68515611e-01 -9.82943475e-01 -4.30829227e-01
-3.62897903e-01 -1.25521886e+00 8.45453322e-01 4.16137308e-01
5.00210047e-01 -7.26470292e-01 -4.17015612e-01 5.42020619e-01
7.51669347e-01 1.85824126e-01 8.26649129e-01 1.91428021e-01
-7.82781243e-01 -9.93329212e-02 -5.52486002e-01 7.08754480e-01
4.01085109e-01 -2.48480260e-01 -9.47547376e-01 -3.70121300e-01
2.97006130e-01 -1.73190519e-01 9.87974048e-01 5.33626437e-01
1.44084501e+00 -4.46766734e-01 -1.82737723e-01 8.04541469e-01
1.65198064e+00 1.31734639e-01 1.12013483e+00 4.24716026e-01
6.13475323e-01 5.31207561e-01 6.83898330e-01 4.21786189e-01
5.63735180e-02 6.52237594e-01 3.76368195e-01 -4.50902730e-01
-6.07225239e-01 -1.82006568e-01 4.19476777e-01 5.86927295e-01
-1.28294609e-03 -2.57947534e-01 -5.43564677e-01 2.90133029e-01
-1.80842197e+00 -9.71274495e-01 1.30741969e-01 2.03211832e+00
1.11736929e+00 -8.62677842e-02 -5.69962800e-01 4.62648571e-02
9.54293609e-01 1.67050973e-01 -6.42151237e-01 2.78062701e-01
-3.53451431e-01 -1.12948678e-01 3.72182757e-01 6.47063613e-01
-8.61573219e-01 7.98593402e-01 7.17732430e+00 9.68419313e-01
-9.73447144e-01 6.43121228e-02 7.08195150e-01 -8.55100714e-03
-9.93129313e-02 1.08331107e-01 -3.37801218e-01 6.00606680e-01
2.25447714e-01 -9.53432452e-03 7.19855487e-01 5.06569982e-01
4.34813678e-01 -4.39924419e-01 -8.32695127e-01 1.36669421e+00
3.19800705e-01 -1.16785192e+00 2.26838216e-01 -2.64018416e-01
7.61283338e-01 -2.34805480e-01 9.26126242e-02 -2.21962318e-01
4.84235920e-02 -9.47286904e-01 4.44000959e-01 7.53832519e-01
8.66138399e-01 -5.12514949e-01 6.24695480e-01 3.13082457e-01
-6.82074487e-01 -7.42867067e-02 -6.11632228e-01 -1.01729542e-01
2.63151109e-01 6.97642326e-01 -3.02251220e-01 5.26734829e-01
8.64185810e-01 7.55210161e-01 -5.48132479e-01 1.16987336e+00
-2.53999978e-01 2.75812387e-01 -1.88170552e-01 7.63875902e-01
-3.54579508e-01 -4.72392380e-01 3.73396546e-01 8.73225272e-01
3.04052472e-01 1.95080623e-01 4.28653270e-01 9.29468691e-01
-8.94643962e-02 -2.13625302e-04 -4.88247842e-01 2.50804394e-01
2.46347278e-01 1.08565056e+00 -5.76276839e-01 -3.19107473e-01
-4.34718192e-01 1.37926877e+00 -1.08310744e-01 7.39881456e-01
-5.45060456e-01 -5.57511151e-01 4.39618289e-01 1.63122341e-01
5.52062206e-02 -2.58885235e-01 -2.73507416e-01 -1.36996841e+00
2.36029208e-01 -9.29171801e-01 3.17570157e-02 -1.42711687e+00
-1.24614179e+00 4.85816091e-01 -4.55150381e-02 -1.27254105e+00
1.46500003e-02 -4.29458320e-01 -5.80835044e-01 8.19596469e-01
-1.69026041e+00 -1.14384890e+00 -6.79003537e-01 9.47140396e-01
7.61222243e-01 -1.91603824e-02 4.48693484e-01 4.41073447e-01
-4.45123166e-01 2.72949457e-01 4.25708354e-01 8.36396739e-02
1.06818485e+00 -9.21244502e-01 -6.62847161e-02 1.18418694e+00
-2.81878650e-01 5.13567269e-01 8.22126269e-01 -8.02108347e-01
-1.43472695e+00 -7.32430756e-01 3.35975066e-02 1.01840504e-01
1.82718620e-01 -9.84988138e-02 -9.76529479e-01 4.89682525e-01
5.21015644e-01 1.11355238e-01 3.80881935e-01 -3.37444276e-01
-1.71272248e-01 -3.71778905e-01 -1.27151084e+00 7.20183015e-01
8.37426126e-01 -5.24798274e-01 -3.34273875e-01 3.85382980e-01
7.34579802e-01 -5.79809129e-01 -7.12655008e-01 2.72048950e-01
4.10121620e-01 -1.11250412e+00 1.19314229e+00 -9.68656130e-03
4.93423373e-01 -6.10269368e-01 -2.57479876e-01 -1.19006467e+00
-2.28149891e-01 -5.10480464e-01 1.72569245e-01 1.33405304e+00
-1.26809359e-01 -4.87499803e-01 5.16480327e-01 7.44684160e-01
2.94056139e-03 -2.22264275e-01 -5.72772145e-01 -3.22675437e-01
-3.25275719e-01 -1.41858846e-01 5.61053693e-01 8.84569824e-01
-2.51037091e-01 -5.92224970e-02 -9.62388337e-01 2.18656003e-01
8.65394056e-01 2.26821184e-01 6.05186105e-01 -1.03541505e+00
-2.17421100e-01 -9.59826261e-02 -8.68333206e-02 -1.15721679e+00
5.14534824e-02 -4.06273812e-01 2.60727167e-01 -1.73846757e+00
4.34014857e-01 -1.48080423e-01 -2.78527558e-01 5.46336830e-01
-2.08209410e-01 4.51935619e-01 3.02045196e-01 4.29200649e-01
-3.64208817e-01 5.98496020e-01 1.66376042e+00 -3.43673468e-01
-1.89049900e-01 -2.89419949e-01 -7.22914100e-01 7.99131930e-01
9.14201438e-01 -2.76682645e-01 -4.44796205e-01 -8.07843566e-01
-1.13806710e-01 3.00335348e-01 4.19359744e-01 -8.82995784e-01
4.29467827e-01 -4.11059022e-01 9.61027682e-01 -7.18807697e-01
5.29536307e-01 -9.33475435e-01 1.33795559e-01 1.85334623e-01
-2.50981748e-01 -1.53859749e-01 1.46165997e-01 6.39789522e-01
-5.31106412e-01 -3.74009818e-01 9.69741166e-01 -3.38948846e-01
-5.68706095e-01 2.91005850e-01 -2.68095583e-01 -3.95837545e-01
6.11774743e-01 -3.47690403e-01 -4.25842196e-01 -7.75798619e-01
-5.01282156e-01 3.32260653e-02 8.36750746e-01 4.42996383e-01
1.10327017e+00 -1.27310944e+00 -7.12360620e-01 4.41910744e-01
-1.88005030e-01 -1.51754960e-01 5.05427837e-01 6.28054917e-01
-6.34401798e-01 -1.61967635e-01 -6.51557148e-01 -4.67371285e-01
-1.09904206e+00 6.97776198e-01 2.38184154e-01 3.60489972e-02
-1.05479002e+00 2.93833166e-01 4.22568172e-01 7.60843158e-02
3.90596569e-01 -7.57125914e-02 -6.88580424e-02 -2.63652980e-01
9.09685135e-01 2.50772715e-01 -1.77983105e-01 -4.97751385e-01
1.96253404e-01 6.06087327e-01 -3.45109433e-01 -1.28050268e-01
1.43146169e+00 -7.34362245e-01 -4.77892429e-01 -2.31072493e-02
8.44335377e-01 7.50292912e-02 -1.71668589e+00 -4.43982929e-01
-4.32229370e-01 -1.06478631e+00 4.04557675e-01 -9.35512543e-01
-1.50717413e+00 8.67943943e-01 9.12703097e-01 -1.67018577e-01
1.80724967e+00 -5.04777312e-01 7.75658965e-01 2.65592813e-01
2.77373970e-01 -1.27088106e+00 3.71006101e-01 1.06358856e-01
1.20873058e+00 -1.31583047e+00 3.32286566e-01 -4.64937598e-01
-4.46347415e-01 1.46385205e+00 7.47604549e-01 3.00891744e-03
5.48866332e-01 4.09315109e-01 1.74303785e-01 1.79765627e-01
-4.28966403e-01 1.00146197e-02 7.42579112e-03 8.44350100e-01
1.74743012e-01 -3.88162971e-01 -2.88472265e-01 3.36697072e-01
2.05897391e-01 1.20374165e-01 9.48185086e-01 8.18757474e-01
-5.46783149e-01 -1.19352448e+00 -8.83139253e-01 2.94369787e-01
-5.60877502e-01 -3.30321878e-01 5.95280342e-02 3.49112213e-01
1.18101142e-01 1.17751610e+00 -3.11818451e-01 -6.11680709e-02
1.67391878e-02 -2.83386379e-01 4.52598959e-01 -2.38430917e-01
3.86266410e-02 4.86651629e-01 -3.43552858e-01 -6.78825915e-01
-6.30794585e-01 -1.94233015e-01 -9.63595331e-01 -2.07141325e-01
-3.87758702e-01 -3.15475076e-01 6.36411548e-01 5.94685376e-01
3.32925528e-01 3.87415975e-01 5.64103544e-01 -1.13829589e+00
-2.09844813e-01 -1.01885915e+00 -7.56852746e-01 6.03116632e-01
5.32907248e-01 -5.11978984e-01 -3.46148551e-01 5.47473252e-01]
|
[11.215758323669434, -1.9128596782684326]
|
83b39ae2-ec79-4ebe-877e-ad7b4278119d
|
convgru-in-fine-grained-pitching-action
|
2008.07819
| null |
https://arxiv.org/abs/2008.07819v1
|
https://arxiv.org/pdf/2008.07819v1.pdf
|
ConvGRU in Fine-grained Pitching Action Recognition for Action Outcome Prediction
|
Prediction of the action outcome is a new challenge for a robot collaboratively working with humans. With the impressive progress in video action recognition in recent years, fine-grained action recognition from video data turns into a new concern. Fine-grained action recognition detects subtle differences of actions in more specific granularity and is significant in many fields such as human-robot interaction, intelligent traffic management, sports training, health caring. Considering that the different outcomes are closely connected to the subtle differences in actions, fine-grained action recognition is a practical method for action outcome prediction. In this paper, we explore the performance of convolutional gate recurrent unit (ConvGRU) method on a fine-grained action recognition tasks: predicting outcomes of ball-pitching. Based on sequences of RGB images of human actions, the proposed approach achieved the performance of 79.17% accuracy, which exceeds the current state-of-the-art result. We also compared different network implementations and showed the influence of different image sampling methods, different fusion methods and pre-training, etc. Finally, we discussed the advantages and limitations of ConvGRU in such action outcome prediction and fine-grained action recognition tasks.
|
['Ou Ma', 'Lin Zhang', 'Xiumin Diao', 'Tianqi Ma']
|
2020-08-18
| null | null | null | null |
['fine-grained-action-recognition']
|
['computer-vision']
|
[ 5.21808386e-01 -3.42656404e-01 -1.20052293e-01 -3.91272038e-01
-5.34241974e-01 7.88246766e-02 6.87017858e-01 -1.57945946e-01
-4.69329745e-01 8.96106184e-01 6.54270589e-01 2.95056850e-01
-3.96330237e-01 -6.85827136e-01 -5.54753959e-01 -8.67300272e-01
-1.40759245e-01 3.24766338e-01 5.65799534e-01 -2.07792744e-01
6.17747486e-01 6.69990718e-01 -2.11377430e+00 9.50411916e-01
4.11757469e-01 1.37677419e+00 2.02659801e-01 9.02379155e-01
1.21078111e-01 1.60238218e+00 -8.17578256e-01 -7.01960102e-02
1.12007685e-01 -5.21341980e-01 -9.43704307e-01 5.10584973e-02
3.27289850e-01 -3.54309380e-01 -3.38650644e-01 7.99267352e-01
5.01167893e-01 4.34681475e-01 6.03760302e-01 -1.27637172e+00
-3.81841153e-01 3.21846992e-01 -2.65242636e-01 4.41450447e-01
4.79319751e-01 2.49188110e-01 6.49132192e-01 -4.38985705e-01
4.52255547e-01 1.46542394e+00 7.16317892e-01 5.14321864e-01
-5.42659640e-01 -5.50539792e-01 -6.77858666e-02 1.06885505e+00
-1.09574056e+00 -2.73845404e-01 1.88885942e-01 -4.68166471e-01
1.48865664e+00 2.05824301e-01 6.96545780e-01 1.03199303e+00
6.08014703e-01 7.41799295e-01 1.17370677e+00 -1.76507816e-01
9.15479586e-02 -6.23518229e-01 -1.28907531e-01 5.56112468e-01
-1.12886183e-01 1.97107613e-01 -6.55705273e-01 2.41462022e-01
9.03822958e-01 4.23428595e-01 5.52950278e-02 6.16152547e-02
-1.56857622e+00 5.97915411e-01 3.47676009e-01 5.00990987e-01
-7.43323982e-01 6.75617158e-01 7.28860736e-01 2.44618282e-01
1.67856753e-01 1.95305794e-01 -3.85253251e-01 -8.91584635e-01
-5.10777891e-01 2.56803840e-01 4.81083423e-01 4.52261090e-01
6.30358875e-01 7.41360188e-02 -7.06535697e-01 6.13055408e-01
-5.58646545e-02 2.87990838e-01 9.99663353e-01 -1.05969667e+00
3.64204854e-01 8.23755026e-01 1.68520242e-01 -9.34706748e-01
-5.70213079e-01 2.58942634e-01 -9.30580199e-01 5.64395785e-01
5.76466382e-01 1.05368398e-01 -9.93500292e-01 1.11029506e+00
2.73982882e-01 3.92061293e-01 1.67304091e-02 1.08067703e+00
7.67666221e-01 5.07391274e-01 4.56011355e-01 -2.83584967e-02
1.19762313e+00 -1.06789362e+00 -7.18553543e-01 2.41627898e-02
9.24695015e-01 -5.15903711e-01 7.63320804e-01 3.39938968e-01
-6.14599228e-01 -9.87249672e-01 -7.64886975e-01 1.74732819e-01
-4.04825032e-01 3.07560742e-01 9.09639299e-01 3.98690104e-01
-9.60945725e-01 1.08609366e+00 -8.64178419e-01 -7.33199656e-01
3.47154677e-01 5.03226221e-01 -6.79566085e-01 1.26659483e-01
-1.07930946e+00 1.07008553e+00 5.00312209e-01 1.34867191e-01
-9.03496683e-01 -1.59200132e-01 -7.25754321e-01 3.57227022e-04
4.35015112e-01 -2.93515891e-01 1.28658426e+00 -8.56988132e-01
-1.70607388e+00 6.78497970e-01 1.01871565e-01 -9.07806396e-01
2.58602053e-01 -3.42391580e-01 -4.51746553e-01 9.56108607e-03
4.43781950e-02 5.68726659e-01 7.57120371e-01 1.07471540e-03
-1.12988925e+00 -5.00762880e-01 1.67799383e-01 3.26533467e-01
2.57802099e-01 2.31457457e-01 1.96863003e-02 -5.74174643e-01
-1.69618666e-01 -9.36661541e-01 -4.10445869e-01 -2.90339857e-01
2.40296841e-01 -8.69413257e-01 7.98950970e-01 -3.48669797e-01
1.07871091e+00 -2.24300790e+00 1.22987702e-01 -4.08785671e-01
-7.56389275e-02 2.84033448e-01 -6.99702725e-02 4.94093448e-01
-5.48082590e-02 -2.25389272e-01 3.82028639e-01 7.57001415e-02
9.43439156e-02 2.49895439e-01 -3.16897742e-02 4.06531274e-01
2.16950580e-01 9.87289310e-01 -7.22159326e-01 -5.58355629e-01
9.26597059e-01 1.24062903e-01 -1.64865211e-01 2.61645377e-01
-6.50151521e-02 4.72933799e-01 -7.94417977e-01 7.20861435e-01
-3.70603800e-03 -1.73531950e-01 -1.33185804e-01 -4.00666863e-01
-1.79776862e-01 1.12092428e-01 -1.24757540e+00 1.59689784e+00
-2.08010122e-01 6.61493957e-01 -6.29897833e-01 -1.16849267e+00
9.48783517e-01 3.67528856e-01 6.34641826e-01 -1.00669420e+00
3.42905790e-01 1.92544803e-01 1.74795672e-01 -6.91603422e-01
6.26108885e-01 -1.48923963e-01 -3.69386345e-01 4.38781381e-01
1.55747801e-01 1.77804660e-02 3.01228911e-01 -3.37769419e-01
1.51587903e+00 4.35964316e-01 9.43011463e-01 1.41755641e-01
5.47017694e-01 1.39218271e-01 4.46076334e-01 8.27414095e-01
-8.48869681e-01 5.33926249e-01 8.22591782e-02 -1.09337664e+00
-7.19358087e-01 -5.24622738e-01 3.12843591e-01 1.32937515e+00
2.01202899e-01 -1.93864033e-01 -6.77088797e-01 -5.96869767e-01
-1.98362753e-01 1.96113557e-01 -8.38037372e-01 -2.96608090e-01
-7.20306516e-01 -4.32487428e-01 6.24513030e-01 8.12681794e-01
1.13412166e+00 -1.95472312e+00 -1.11330056e+00 3.34521621e-01
-2.05226332e-01 -1.17242813e+00 5.52971587e-02 2.05471143e-01
-7.58268893e-01 -1.16324687e+00 -6.19165242e-01 -4.66136158e-01
-1.49996191e-01 1.88546464e-01 9.11631584e-01 -4.65473570e-02
-3.92190039e-01 3.16290528e-01 -9.29079711e-01 -3.09210211e-01
-2.44494662e-01 -2.18513638e-01 1.07093014e-01 2.27132231e-01
6.59294486e-01 -3.17451358e-01 -6.44271970e-01 4.71564621e-01
-5.88389575e-01 4.26375860e-04 8.91420901e-01 8.34087253e-01
6.44983649e-01 1.91227630e-01 4.11124378e-01 -3.44784856e-01
3.94424260e-01 -6.96574152e-02 -1.09920815e-01 3.83366019e-01
3.65291983e-02 2.06754342e-01 4.82547194e-01 -4.66262043e-01
-1.14652848e+00 1.98353276e-01 -2.41438642e-01 -3.98071408e-01
-7.91316569e-01 1.01438545e-01 7.75650591e-02 -1.24181613e-01
5.59172750e-01 1.75269544e-01 -1.94281504e-01 -3.35311383e-01
9.45025831e-02 8.14007401e-01 4.89096373e-01 -8.35599825e-02
-1.56259596e-01 4.13708210e-01 8.29991549e-02 -7.16424823e-01
-6.00954950e-01 -4.29199904e-01 -7.49794245e-01 -6.46334827e-01
1.39309609e+00 -7.45805502e-01 -1.10239851e+00 1.10266471e+00
-1.06012714e+00 -5.84254146e-01 -4.86439407e-01 7.96932817e-01
-8.92091990e-01 1.33477181e-01 -5.63387096e-01 -5.81004798e-01
-1.56851932e-01 -1.19301748e+00 1.38180673e+00 3.00658196e-01
-2.81766534e-01 -6.60634339e-01 1.01533411e-02 5.23041606e-01
2.81514347e-01 4.56727713e-01 3.39006305e-01 -4.82928723e-01
-5.05998492e-01 -2.57891566e-01 -1.68850929e-01 1.90419704e-01
2.81008393e-01 -3.45212311e-01 -6.36614859e-01 2.24896818e-01
-2.22063046e-02 -6.20579720e-01 8.22831452e-01 4.48268890e-01
1.22393501e+00 2.93724444e-02 -3.25182468e-01 2.05965355e-01
8.94451678e-01 4.97814745e-01 1.19753587e+00 4.39303041e-01
7.77446389e-01 4.41230983e-01 1.26601410e+00 8.68580937e-01
5.11904769e-02 9.63206232e-01 5.34202218e-01 2.79839635e-01
-4.41928357e-01 1.42858522e-02 4.52489287e-01 1.19859613e-01
-7.92294919e-01 -8.91850069e-02 -5.64721406e-01 3.21412206e-01
-2.25795960e+00 -1.51853395e+00 -1.70139968e-01 1.82819521e+00
3.19986016e-01 1.57675538e-02 1.08664468e-01 2.02792004e-01
7.85978913e-01 3.30242783e-01 -5.65786958e-01 -7.29755759e-01
1.93712905e-01 3.51727217e-01 5.83534896e-01 -1.33102030e-01
-1.40068614e+00 1.18998694e+00 6.12016630e+00 1.27071559e+00
-1.21406126e+00 1.19871013e-01 5.45089960e-01 8.69327807e-04
6.30639255e-01 -4.71974343e-01 -8.71816278e-01 3.90867800e-01
9.33174014e-01 2.58939654e-01 8.60871151e-02 8.62627149e-01
2.48393908e-01 -4.08324838e-01 -9.92487669e-01 1.28210187e+00
7.67495856e-02 -1.44560802e+00 5.59297949e-02 -1.94597080e-01
5.63595891e-01 1.95862487e-01 -4.60839123e-01 5.12175977e-01
4.45133716e-01 -1.27179909e+00 6.60127878e-01 7.68873036e-01
6.74136817e-01 -5.12308478e-01 8.62911224e-01 2.33335987e-01
-1.35932207e+00 -3.69622052e-01 -3.94186229e-01 -8.49663377e-01
2.44542032e-01 1.92002758e-01 -6.77197397e-01 2.97402322e-01
1.12675238e+00 1.20562673e+00 -4.77507651e-01 8.59255910e-01
-1.75832018e-01 3.77753913e-01 -2.69021727e-02 -4.42455441e-01
3.96828800e-01 4.84934784e-02 6.10110536e-02 1.00103140e+00
4.80515867e-01 5.83798587e-01 5.46068996e-02 1.05253153e-01
4.59173024e-01 -2.61659324e-01 -6.48103178e-01 -2.41707772e-01
-9.68086794e-02 1.05085039e+00 -6.99107170e-01 -4.45233852e-01
-4.97254640e-01 1.19631350e+00 3.93326938e-01 -5.04160970e-02
-9.29273725e-01 -1.08375415e-01 9.49276268e-01 -1.73331231e-01
5.32470345e-01 -1.61641538e-01 -3.53348330e-02 -8.38830590e-01
-7.46013224e-02 -8.79678488e-01 5.27257681e-01 -1.01422942e+00
-8.22102308e-01 6.18334711e-01 -9.22482908e-02 -1.65756583e+00
-6.14921331e-01 -9.09774423e-01 -4.50738519e-01 4.04935896e-01
-7.31172025e-01 -1.05496526e+00 -4.93391663e-01 8.74049127e-01
9.75612998e-01 -2.06933767e-01 9.41609025e-01 1.48685470e-01
-4.08759445e-01 2.69041002e-01 -2.11024970e-01 -2.57485230e-02
4.80634362e-01 -7.75326431e-01 3.10998857e-01 6.17525101e-01
1.52023152e-01 -1.59953728e-01 5.21907628e-01 -6.55391574e-01
-1.14433062e+00 -1.05654466e+00 8.53760898e-01 -3.03401321e-01
6.03335202e-01 2.35235646e-01 -5.00540197e-01 4.97693241e-01
-1.82915367e-02 2.15634137e-01 3.33010167e-01 -1.55986503e-01
1.69181064e-01 -1.86346754e-01 -1.03162229e+00 5.34037709e-01
1.29016852e+00 -2.73758501e-01 -6.28953874e-01 2.09767789e-01
4.15705115e-01 -3.35268795e-01 -8.28812897e-01 6.68477356e-01
8.17076862e-01 -1.33310449e+00 8.07127118e-01 -6.91301525e-01
3.55395645e-01 -3.77264708e-01 -3.65405440e-01 -1.15420473e+00
-5.35517514e-01 -7.32657909e-02 3.56413610e-02 6.00790381e-01
-2.16137066e-01 -4.06864703e-01 8.41237903e-01 4.15365189e-01
-3.26124787e-01 -8.18961799e-01 -1.02359807e+00 -5.69345593e-01
-5.69511831e-01 -6.04213357e-01 5.77971637e-01 2.11964265e-01
1.04682865e-02 8.81251320e-02 -6.80388510e-01 -1.98512495e-01
1.49764970e-01 2.33883768e-01 8.83171320e-01 -1.07178724e+00
-3.27292949e-01 -3.44012588e-01 -1.44162285e+00 -9.38960493e-01
-5.40630147e-02 -2.24267900e-01 1.31884590e-01 -1.71561265e+00
1.02184109e-01 -7.74488300e-02 -4.51585263e-01 6.24800503e-01
1.28720496e-02 7.84237266e-01 3.59278709e-01 2.11542547e-01
-1.17840469e+00 5.25791466e-01 1.36657357e+00 -1.05009653e-01
9.81053934e-02 9.36351046e-02 -1.80560365e-01 8.40975702e-01
8.58051896e-01 -3.04784745e-01 -9.99410152e-02 -1.22548088e-01
-1.07160851e-01 2.96900272e-01 5.12504280e-01 -1.71859396e+00
2.19608590e-01 -3.88500482e-01 4.37794060e-01 -6.45268261e-01
5.33019900e-01 -6.70745730e-01 3.24358910e-01 6.71375334e-01
-1.65013626e-01 -1.54183298e-01 5.37861623e-02 4.06487674e-01
-5.61552048e-01 -5.37979864e-02 4.66441184e-01 -3.46130997e-01
-1.94717515e+00 1.96199074e-01 -6.48895264e-01 -2.99728692e-01
1.50350916e+00 -8.00069273e-01 -1.81703299e-01 -1.57248750e-01
-8.46015275e-01 -2.58218408e-01 -2.91078147e-02 6.51495993e-01
5.94048560e-01 -1.51066363e+00 -6.13182425e-01 1.56543389e-01
1.82131529e-01 -4.61015314e-01 6.61916435e-01 1.01395023e+00
-4.03758585e-01 6.14804983e-01 -8.56467962e-01 -5.73164225e-01
-1.59084320e+00 1.97579473e-01 3.03139806e-01 -5.93281806e-01
-6.06610835e-01 7.48343885e-01 2.64103949e-01 -1.69689834e-01
2.01461285e-01 -5.28919637e-01 -6.61219954e-01 -9.56293866e-02
7.96220839e-01 8.37729573e-01 1.34675235e-01 -8.47537756e-01
-4.38384742e-01 6.86805427e-01 2.49665752e-02 3.05858284e-01
1.18585837e+00 -2.80013066e-02 1.55539617e-01 5.48649490e-01
8.39294970e-01 -8.07136178e-01 -1.47266340e+00 1.10860042e-01
-3.77152041e-02 -4.37469304e-01 -3.07023376e-01 -8.54313791e-01
-8.84652734e-01 8.07743311e-01 8.94957900e-01 8.75907093e-02
1.25888896e+00 1.72147214e-01 6.53247297e-01 5.71029544e-01
9.59983587e-01 -1.19706941e+00 2.98520774e-01 9.00974095e-01
8.98210466e-01 -1.40582621e+00 -4.48510014e-02 -3.81278396e-02
-8.47132385e-01 1.07583785e+00 6.54014349e-01 -3.88972223e-01
3.92214090e-01 6.80760890e-02 6.54938444e-02 -2.61262387e-01
-6.18372321e-01 -4.46516991e-01 2.71510750e-01 5.43251038e-01
4.54579860e-01 3.49019498e-01 -3.60448778e-01 2.85882622e-01
-2.03267504e-02 5.66459656e-01 2.09875301e-01 8.74515653e-01
-6.90401614e-01 -8.65238070e-01 -3.98357391e-01 7.23380625e-01
-4.85296667e-01 8.24656039e-02 -3.28927875e-01 7.07453907e-01
4.86205399e-01 1.05759346e+00 1.53304309e-01 -7.52324522e-01
4.20702100e-01 -9.31150243e-02 7.08981216e-01 -4.47570235e-01
-6.71011627e-01 -4.94142979e-01 2.54527986e-01 -1.40089035e+00
-8.76877248e-01 -8.68103683e-01 -1.45121360e+00 -3.24199170e-01
2.98647601e-02 -1.67740852e-01 3.98347855e-01 1.42801011e+00
4.82851863e-01 6.75857306e-01 2.14314803e-01 -1.30091918e+00
-4.44500089e-01 -1.36426651e+00 -8.03847432e-01 5.50800681e-01
2.34823190e-02 -1.06916487e+00 -5.55890165e-02 1.36451468e-01]
|
[8.171290397644043, 0.5872477293014526]
|
515494df-165c-475d-b65b-d1ee25a101b0
|
temporal-cue-guided-video-highlight-detection
| null | null |
http://openaccess.thecvf.com//content/ICCV2021/html/Ye_Temporal_Cue_Guided_Video_Highlight_Detection_With_Low-Rank_Audio-Visual_Fusion_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Ye_Temporal_Cue_Guided_Video_Highlight_Detection_With_Low-Rank_Audio-Visual_Fusion_ICCV_2021_paper.pdf
|
Temporal Cue Guided Video Highlight Detection With Low-Rank Audio-Visual Fusion
|
Video highlight detection plays an increasingly important role in social media content filtering, however, it remains highly challenging to develop automated video highlight detection methods because of the lack of temporal annotations (i.e., where the highlight moments are in long videos) for supervised learning. In this paper, we propose a novel weakly supervised method that can learn to detect highlights by mining video characteristics with video level annotations (topic tags) only. Particularly, we exploit audio-visual features to enhance video representation and take temporal cues into account for improving detection performance. Our contributions are threefold: 1) we propose an audio-visual tensor fusion mechanism that efficiently models the complex association between two modalities while reducing the gap of the heterogeneity between the two modalities; 2) we introduce a novel hierarchical temporal context encoder to embed local temporal clues in between neighboring segments; 3) finally, we alleviate the gradient vanishing problem theoretically during model optimization with attention-gated instance aggregation. Extensive experiments on two benchmark datasets (YouTube Highlights and TVSum) have demonstrated our method outperforms other state-of-the-art methods with remarkable improvements.
|
['Guang Yang', 'Ping Li', 'Qi Bi', 'ZiRui Wang', 'Yuan Gao', 'Xiyue Shen', 'Qinghao Ye']
|
2021-01-01
| null | null | null |
iccv-2021-1
|
['highlight-detection']
|
['computer-vision']
|
[ 1.85100853e-01 -4.30536568e-01 -3.05254519e-01 -9.89029258e-02
-1.02596700e+00 -4.71725047e-01 3.49808365e-01 3.72096837e-01
-3.43351126e-01 4.31760460e-01 4.86550540e-01 2.37942189e-01
-1.26673743e-01 -1.99638650e-01 -7.89068997e-01 -7.15891004e-01
-4.99056786e-01 -6.02964103e-01 4.80654567e-01 9.22617018e-02
3.61154109e-01 -6.79244325e-02 -1.79526830e+00 7.25717127e-01
8.35982919e-01 1.46917868e+00 2.94232011e-01 5.85728347e-01
1.33072749e-01 1.36478817e+00 -1.45080149e-01 -1.02810197e-01
2.14156687e-01 -3.93318444e-01 -4.84140933e-01 4.33602631e-01
8.36728573e-01 -3.88028473e-01 -4.12087560e-01 9.16355908e-01
4.02481854e-01 4.06050354e-01 2.42222786e-01 -1.25069547e+00
-2.68656641e-01 6.78905308e-01 -1.00973725e+00 5.79282761e-01
3.10942292e-01 3.01425923e-02 1.33284640e+00 -1.10560334e+00
8.55083585e-01 8.74465466e-01 6.04719460e-01 5.39592409e-04
-9.56192136e-01 -4.13420051e-01 7.48586714e-01 6.65382862e-01
-1.33809793e+00 -1.90728143e-01 1.18473935e+00 -6.64911687e-01
4.20563400e-01 3.79261613e-01 7.91324496e-01 9.71003115e-01
-3.59385222e-01 1.24771428e+00 8.96204710e-01 -2.37913609e-01
-6.64219111e-02 -1.33105982e-02 -1.13863558e-01 9.24151182e-01
-3.72884601e-01 -2.99435973e-01 -1.21627486e+00 -2.03087449e-01
4.59641814e-01 2.72350758e-01 -3.59520912e-01 -3.24178725e-01
-1.55207860e+00 6.80381119e-01 3.48997205e-01 3.62310261e-01
-4.81639385e-01 2.10786790e-01 7.71870017e-01 1.77885145e-01
8.16532195e-01 1.16505191e-01 -3.71892214e-01 -2.83682108e-01
-1.27993011e+00 2.00934350e-01 2.21620888e-01 7.18260348e-01
7.13658035e-01 -4.79051061e-02 -5.20211995e-01 8.71236861e-01
-8.51604193e-02 8.17119628e-02 2.34374091e-01 -8.63370359e-01
6.14698172e-01 4.99012083e-01 9.85854492e-02 -1.24731445e+00
-1.72386691e-01 -4.64652002e-01 -5.02626836e-01 -3.49049926e-01
2.70440757e-01 -7.62229115e-02 -4.84089255e-01 1.56023133e+00
3.45994234e-01 8.41556489e-01 -5.68528414e-01 1.09622037e+00
8.93277168e-01 6.41300142e-01 1.59028098e-01 -5.05911410e-01
1.47842538e+00 -1.14485133e+00 -7.28812456e-01 2.81037420e-01
5.57445109e-01 -8.60420763e-01 1.05596173e+00 4.59344774e-01
-1.14973688e+00 -5.56163192e-01 -8.38207185e-01 -2.11669922e-01
-2.03297302e-01 5.40812552e-01 8.33666682e-01 6.09823912e-02
-7.19419599e-01 4.86353964e-01 -7.43449867e-01 -1.92698389e-01
4.58162546e-01 5.08871861e-02 -1.36596903e-01 8.48129243e-02
-9.89877582e-01 6.37081340e-02 3.38619292e-01 7.16797560e-02
-9.40434098e-01 -9.92489457e-01 -7.32847154e-01 -1.57902613e-02
9.18632865e-01 -3.12310576e-01 9.11093712e-01 -1.26750493e+00
-1.19409454e+00 6.20459616e-01 -4.50772285e-01 -4.08690661e-01
6.73687935e-01 -5.85064888e-01 -2.14327246e-01 5.79015434e-01
1.24136090e-01 6.78050399e-01 1.06669557e+00 -1.21153975e+00
-1.10113227e+00 -4.34128828e-02 1.84774980e-01 2.11111099e-01
-9.08073068e-01 2.40311474e-01 -9.98695135e-01 -1.10255337e+00
6.18987530e-02 -8.51082325e-01 -7.13857114e-02 1.77694306e-01
-5.56941092e-01 -3.05751562e-01 1.09102976e+00 -7.49079466e-01
1.60814345e+00 -2.34544563e+00 2.85570860e-01 1.51225209e-01
4.56420630e-01 -6.33347631e-02 2.50193067e-02 3.74233335e-01
1.69596076e-01 -1.49873704e-01 8.53845757e-03 -4.36864495e-01
-1.70016319e-01 -3.36239815e-01 -5.12295187e-01 3.26125264e-01
1.89688817e-01 6.39028251e-01 -1.05154204e+00 -8.74375165e-01
1.42691195e-01 5.72528064e-01 -7.32555747e-01 1.88447699e-01
-2.62403518e-01 4.80251282e-01 -4.65792030e-01 9.54011500e-01
4.01154757e-01 -3.53037626e-01 2.93693747e-02 -6.23468816e-01
-4.09501851e-01 2.03418406e-03 -1.19197452e+00 1.91169286e+00
-1.93508208e-01 8.75684619e-01 -3.45215872e-02 -8.07602406e-01
3.85200232e-01 2.75124967e-01 9.60388303e-01 -4.05225068e-01
6.04404174e-02 -2.62171999e-02 -4.75942314e-01 -8.74397397e-01
7.52076864e-01 3.09979796e-01 2.19228305e-02 1.73584685e-01
-3.54293771e-02 4.89638597e-01 5.25434256e-01 3.71569067e-01
1.02826202e+00 2.16165692e-01 -9.94738191e-02 3.82065885e-02
5.30120373e-01 -1.68647006e-01 8.34782839e-01 5.53911746e-01
-3.03823948e-01 7.26513982e-01 7.10787237e-01 -3.55267167e-01
-6.99756384e-01 -7.57655859e-01 1.81068555e-01 1.70311022e+00
1.89974159e-01 -8.76395285e-01 -5.54556489e-01 -7.86230564e-01
8.52957554e-03 -2.86620744e-02 -8.16475630e-01 2.27510884e-01
-7.11779773e-01 -5.63068748e-01 2.32996687e-01 6.06590629e-01
3.35617483e-01 -6.22991979e-01 -4.28274900e-01 2.58785393e-03
-5.71072102e-01 -1.43498039e+00 -9.16282594e-01 -9.37512219e-02
-9.08549666e-01 -1.09778404e+00 -9.36182857e-01 -7.98373520e-01
4.89362031e-01 7.88776040e-01 8.23185980e-01 1.43090084e-01
-3.72712821e-01 5.70031047e-01 -5.61338663e-01 -7.49083906e-02
3.67188066e-01 -2.89961156e-02 -2.10134014e-01 4.91603047e-01
2.00780958e-01 -4.94208187e-01 -9.05849397e-01 3.29818726e-01
-8.51122558e-01 1.89264551e-01 3.44823062e-01 7.52928972e-01
7.94394851e-01 3.38539481e-02 2.85814047e-01 -8.61028135e-01
3.76486242e-01 -4.95253950e-01 -3.50741446e-01 1.79048181e-01
-7.31344670e-02 -2.26204678e-01 3.61989617e-01 -6.48243666e-01
-9.97483909e-01 1.25121206e-01 5.40206611e-01 -8.41706693e-01
2.19898030e-01 6.54173017e-01 1.11022323e-01 -5.79297766e-02
2.13106021e-01 9.39152688e-02 -3.71497244e-01 -5.82109213e-01
3.12581569e-01 2.35814467e-01 4.04276699e-01 -5.06110191e-01
6.68669820e-01 8.84226680e-01 -4.93397340e-02 -9.24751520e-01
-1.03279817e+00 -8.61409128e-01 -6.55079544e-01 -5.96803784e-01
7.59358883e-01 -1.42191637e+00 -7.40631640e-01 2.03467101e-01
-8.53335381e-01 -1.25942364e-01 -4.79667559e-02 6.32591009e-01
-4.03191149e-01 7.49763131e-01 -7.96023726e-01 -8.35044384e-01
-1.68427065e-01 -9.81505215e-01 1.24072897e+00 1.09315462e-01
-2.87602432e-02 -6.92805767e-01 -1.33095905e-01 4.16471332e-01
1.25028327e-01 5.63620508e-01 5.48305750e-01 -7.52023235e-02
-8.99595201e-01 -3.17001045e-02 -5.09126484e-01 1.36268228e-01
-1.14185505e-01 3.23797315e-01 -1.00719404e+00 -1.46747857e-01
-4.21541989e-01 -2.10411385e-01 1.37141311e+00 4.83499318e-01
1.48903120e+00 -2.78282940e-01 -3.29462677e-01 5.16023636e-01
1.19418263e+00 -1.22345865e-01 2.09863365e-01 3.41508627e-01
1.03476119e+00 7.56208956e-01 9.82185483e-01 8.45874846e-01
4.49100167e-01 8.44265938e-01 3.53775173e-01 -5.43039516e-02
-2.10160121e-01 -2.95742363e-01 5.41932881e-01 1.02979994e+00
-5.08729756e-01 4.14159000e-02 -4.18180346e-01 6.77168310e-01
-2.27513146e+00 -1.14462459e+00 -9.26009864e-02 2.08561039e+00
6.12845898e-01 1.65019125e-01 5.65964341e-01 1.43637359e-01
7.97016144e-01 5.08651376e-01 -3.40977699e-01 3.73036295e-01
-3.14168364e-01 -3.56191844e-01 2.54335433e-01 1.26532197e-01
-1.63424122e+00 7.04354227e-01 4.85574198e+00 8.71578395e-01
-1.14554071e+00 3.84733200e-01 5.93096256e-01 -6.64247990e-01
-9.51603130e-02 -1.40729517e-01 -4.23007160e-01 6.70450568e-01
3.88322949e-01 1.94948629e-01 1.35458544e-01 7.48284936e-01
5.35169125e-01 -7.19677806e-02 -1.01052046e+00 1.27184331e+00
2.44178772e-01 -1.43179727e+00 -1.19953312e-01 -8.64344016e-02
1.00162899e+00 -1.18852481e-01 3.95278245e-01 6.32460862e-02
-4.22952443e-01 -3.49270046e-01 1.06451070e+00 5.93532562e-01
5.00955641e-01 -7.47269750e-01 3.80863845e-01 -2.02735931e-01
-1.71480441e+00 -3.45882684e-01 -1.14132866e-01 1.81217775e-01
1.81377202e-01 7.29459882e-01 -4.04479861e-01 4.19331491e-01
9.97523487e-01 1.24749756e+00 -4.96502787e-01 1.26957393e+00
-5.93718551e-02 6.92050993e-01 -2.03584820e-01 2.08191097e-01
3.40617418e-01 7.26080015e-02 6.73197687e-01 1.48253500e+00
2.56312788e-01 -7.98307955e-02 4.88053411e-01 4.08785701e-01
-1.74989536e-01 3.39306861e-01 -1.55473024e-01 -7.80138373e-02
2.40637898e-01 1.30351698e+00 -8.45660329e-01 -1.90227345e-01
-6.62193835e-01 9.64156210e-01 1.77827001e-01 4.00504112e-01
-1.05060303e+00 -1.74920887e-01 5.64984620e-01 2.33766392e-01
5.94820738e-01 -1.85313463e-01 4.37108129e-02 -1.49882519e+00
3.67222816e-01 -6.85089588e-01 7.07457840e-01 -5.91574728e-01
-1.03230000e+00 4.23001915e-01 -2.80400187e-01 -1.69293296e+00
5.28939739e-02 -1.94714502e-01 -5.91713667e-01 1.76295310e-01
-1.67768109e+00 -1.23827779e+00 -4.72954750e-01 7.02626824e-01
8.08648348e-01 1.02222599e-01 8.90135914e-02 7.69690692e-01
-6.29938960e-01 5.38595796e-01 1.70698743e-02 8.56617764e-02
9.54033852e-01 -1.04627478e+00 -3.42580378e-01 1.02595437e+00
2.89880753e-01 4.00374472e-01 5.77365816e-01 -5.90948224e-01
-1.51995695e+00 -1.13907862e+00 6.94980204e-01 -1.57022178e-01
8.27233255e-01 -3.01973432e-01 -8.05041790e-01 3.98618519e-01
2.07604066e-01 2.11771160e-01 9.23305750e-01 2.66015559e-01
-5.28996110e-01 -2.20251366e-01 -5.49100041e-01 5.34002304e-01
1.13468635e+00 -8.68307889e-01 -8.76675472e-02 4.62131321e-01
6.94956303e-01 -2.57501572e-01 -7.67167211e-01 3.35821152e-01
6.54642344e-01 -9.31382120e-01 8.83977175e-01 -4.44560051e-01
6.53179586e-01 -5.20037174e-01 -1.65085137e-01 -7.53931820e-01
-3.77790108e-02 -9.78184104e-01 -5.98112762e-01 1.42617083e+00
1.14586174e-01 1.24245703e-01 7.84233749e-01 1.92770377e-01
-4.07518633e-02 -9.42363381e-01 -7.93223202e-01 -4.62141842e-01
-9.03209388e-01 -5.26563108e-01 1.22974031e-01 1.06044018e+00
2.14359552e-01 -3.19787003e-02 -8.91432226e-01 1.60507321e-01
6.00461960e-01 4.04194951e-01 6.32732928e-01 -1.08243728e+00
-3.57344717e-01 -5.20919442e-01 -3.82169694e-01 -1.18284261e+00
-1.10982001e-01 -5.10661423e-01 -6.45827204e-02 -1.08081973e+00
6.42987311e-01 -1.09559186e-01 -6.91350877e-01 3.04388672e-01
-4.14191604e-01 4.84978735e-01 4.29872900e-01 4.42809224e-01
-1.39352095e+00 5.80604613e-01 1.05714560e+00 -1.21088542e-01
-2.32887715e-01 -2.35258371e-01 -4.98479486e-01 8.58143866e-01
1.70636341e-01 -2.24459499e-01 -3.32897663e-01 -3.01985741e-01
4.25083101e-01 4.28935885e-02 4.12542343e-01 -9.36830997e-01
2.52401590e-01 -3.09323822e-03 2.88717210e-01 -8.51176381e-01
4.00553793e-01 -6.39088392e-01 -4.06112105e-01 6.95097595e-02
-5.57269752e-01 -5.64107709e-02 3.28871645e-02 9.96768892e-01
-5.62942743e-01 1.92357749e-01 4.74941730e-01 9.55168158e-02
-1.02184248e+00 4.46962178e-01 -3.24763745e-01 1.09510440e-02
9.77351665e-01 1.04239009e-01 -1.62376598e-01 -3.97579789e-01
-7.50320792e-01 3.33463341e-01 3.21639240e-01 5.68826497e-01
5.96728444e-01 -1.35346508e+00 -5.75729251e-01 -2.53601313e-01
2.71576583e-01 -3.64840180e-01 6.87331796e-01 1.30857897e+00
-2.18290180e-01 2.21833959e-01 1.65292129e-01 -6.73148155e-01
-1.58252037e+00 6.43114567e-01 -2.00866476e-01 -1.13111608e-01
-5.94805300e-01 1.03971565e+00 3.84089917e-01 4.30295199e-01
6.10663474e-01 -3.83357495e-01 -3.37101012e-01 6.59484446e-01
8.49290431e-01 5.12012661e-01 -7.61640891e-02 -7.81474233e-01
-2.66140699e-01 5.95223367e-01 -2.72935241e-01 2.21976787e-01
1.39351118e+00 -3.86771977e-01 9.98275951e-02 5.19289911e-01
1.29767966e+00 1.70987010e-01 -1.57964242e+00 -5.55344939e-01
2.72914637e-02 -7.39822924e-01 1.84939310e-01 -2.46601298e-01
-1.29306102e+00 8.63142550e-01 6.17981672e-01 3.07431281e-01
1.35160053e+00 -2.94318553e-02 1.13130474e+00 1.90125868e-01
2.78773103e-02 -1.39797211e+00 5.06492376e-01 1.05199106e-01
7.27148652e-01 -1.39917135e+00 1.66584983e-01 -6.98881924e-01
-8.04741859e-01 1.13101566e+00 4.05059606e-01 -8.16295668e-02
6.11436009e-01 -1.12871908e-01 -1.88427627e-01 -2.65266806e-01
-8.59604239e-01 -5.14871001e-01 6.77056551e-01 5.60505427e-02
6.10114276e-01 -9.54163224e-02 -2.69725889e-01 5.04248023e-01
4.18512881e-01 -3.97281721e-02 2.47433007e-01 9.41992998e-01
-4.34936911e-01 -7.37333417e-01 -2.29745492e-01 3.39648426e-01
-8.96820426e-01 -1.27901986e-01 -2.15290859e-01 5.03498554e-01
2.88388014e-01 8.67495596e-01 -6.31047832e-03 -4.59566504e-01
2.72808503e-02 -1.52056694e-01 2.47152582e-01 -3.01481277e-01
-6.59496069e-01 7.24440575e-01 -6.21737633e-03 -7.89448202e-01
-7.93887913e-01 -9.49192107e-01 -8.47010791e-01 8.29836726e-02
-2.91550756e-01 1.93618298e-01 4.10222650e-01 7.19736338e-01
5.38107932e-01 6.30798459e-01 7.67592549e-01 -1.10390270e+00
-1.42394807e-02 -6.91206813e-01 -6.41981900e-01 6.30525768e-01
5.43198168e-01 -8.29539776e-01 -3.65169913e-01 4.00490671e-01]
|
[10.11257266998291, 0.47619274258613586]
|
559a2f8a-a6dc-4b4a-b701-89625489bb2b
|
pingan-vcgroup-s-solution-for-icdar-2021
|
2105.01848
| null |
https://arxiv.org/abs/2105.01848v1
|
https://arxiv.org/pdf/2105.01848v1.pdf
|
PingAn-VCGroup's Solution for ICDAR 2021 Competition on Scientific Literature Parsing Task B: Table Recognition to HTML
|
This paper presents our solution for ICDAR 2021 competition on scientific literature parsing taskB: table recognition to HTML. In our method, we divide the table content recognition task into foursub-tasks: table structure recognition, text line detection, text line recognition, and box assignment.Our table structure recognition algorithm is customized based on MASTER [1], a robust image textrecognition algorithm. PSENet [2] is used to detect each text line in the table image. For text linerecognition, our model is also built on MASTER. Finally, in the box assignment phase, we associatedthe text boxes detected by PSENet with the structure item reconstructed by table structure prediction,and fill the recognized content of the text line into the corresponding item. Our proposed methodachieves a 96.84% TEDS score on 9,115 validation samples in the development phase, and a 96.32%TEDS score on 9,064 samples in the final evaluation phase.
|
['Rong Xiao', 'Peng Gao', 'Dengyi Gu', 'Yihao Chen', 'Yelin He', 'Xianbiao Qi', 'Jiaquan Ye']
|
2021-05-05
| null | null | null | null |
['table-recognition', 'line-detection']
|
['computer-vision', 'computer-vision']
|
[ 3.42719346e-01 1.38027176e-01 -6.25078976e-01 -2.72471875e-01
-1.24009144e+00 -8.30218375e-01 2.54002482e-01 4.40319866e-01
-2.43721917e-01 5.19253671e-01 -2.50382423e-02 -2.96958894e-01
3.07588816e-01 -6.36997700e-01 -9.31180954e-01 -1.90673679e-01
6.01840138e-01 7.14353204e-01 3.47181827e-01 3.83987576e-01
4.49580014e-01 5.87521732e-01 -9.93369162e-01 9.75033700e-01
9.81129885e-01 1.29402471e+00 9.19137374e-02 8.32392812e-01
-7.32347488e-01 7.56974816e-01 -8.22246373e-01 -5.87685347e-01
-2.17965275e-01 -5.80073237e-01 -7.83436596e-01 4.72123921e-01
5.98340511e-01 -2.21351951e-01 -7.28017166e-02 8.30796599e-01
8.95535946e-02 -3.80846858e-01 8.15563679e-01 -8.40110421e-01
-5.16731888e-02 8.87834311e-01 -8.33600283e-01 2.06306189e-01
4.44004595e-01 -5.29139042e-01 9.63034868e-01 -1.32266939e+00
1.05184901e+00 9.46938336e-01 3.24955076e-01 3.87353301e-01
-9.05401468e-01 -8.95080388e-01 -2.12508012e-02 1.68151468e-01
-1.13290203e+00 -6.01146698e-01 5.14133990e-01 -4.22531426e-01
8.31500232e-01 3.82116914e-01 2.80016035e-01 6.78711116e-01
4.35427159e-01 1.29335904e+00 7.25170851e-01 -5.44132471e-01
2.31735989e-01 1.50662541e-01 4.13508773e-01 9.80850518e-01
5.66283405e-01 -7.76986897e-01 -6.64888918e-01 1.98879719e-01
7.18118131e-01 -3.40976149e-01 1.62394062e-01 -8.65321532e-02
-9.17247951e-01 3.21658254e-01 -1.43516749e-01 1.94742262e-01
-1.00605242e-01 -4.31948245e-01 5.46724796e-01 2.91240551e-02
3.99085283e-02 3.53674114e-01 -3.30644369e-01 4.84686419e-02
-1.07743311e+00 6.40479103e-03 8.37614000e-01 1.43920207e+00
4.35288966e-01 -2.18795002e-01 -4.09110755e-01 8.75613451e-01
2.57111311e-01 5.67485094e-01 4.20558542e-01 -5.36944032e-01
1.11478817e+00 8.49861443e-01 -8.10195282e-02 -7.69080579e-01
-6.07525766e-01 -1.75307393e-01 -9.95161116e-01 -3.89538586e-01
4.77608383e-01 -2.81990230e-01 -1.02561033e+00 7.48550594e-01
9.45357792e-03 -3.73560488e-01 -8.27796012e-03 4.01969671e-01
1.24662983e+00 7.03409433e-01 -8.47804099e-02 -2.78735667e-01
1.85383213e+00 -9.63877082e-01 -1.01262128e+00 -1.85266614e-01
9.19885993e-01 -1.02502131e+00 5.76931596e-01 8.40384066e-01
-1.20410693e+00 -5.74826837e-01 -1.17922866e+00 -3.45257342e-01
-2.72743523e-01 9.65121508e-01 4.67242777e-01 7.25798011e-01
-4.59380329e-01 1.86334461e-01 -6.24927878e-01 -2.20452532e-01
5.62084496e-01 1.60561606e-01 -4.64127511e-01 -6.99463338e-02
-6.01635695e-01 3.81987154e-01 4.21829075e-01 -1.10334970e-01
-2.62909442e-01 -4.58771050e-01 -6.78730309e-01 1.81548104e-01
5.64423919e-01 -3.64121079e-01 1.01928532e+00 -3.55250090e-01
-1.20421600e+00 1.31841922e+00 -2.95915276e-01 -3.89725417e-01
3.77311617e-01 -1.91112518e-01 -4.74572212e-01 2.54312396e-01
3.09712708e-01 4.65470344e-01 5.68988502e-01 -1.09657311e+00
-8.87802601e-01 -6.08678997e-01 -6.66466355e-01 1.98143438e-01
1.32554621e-01 1.07539766e-01 -1.27362347e+00 -8.94799173e-01
6.78144217e-01 -5.61050355e-01 3.42818826e-01 -2.17691436e-01
-1.04353738e+00 -3.11894536e-01 4.06259686e-01 -1.01888001e+00
1.41262710e+00 -2.08581829e+00 -3.06135207e-01 3.93844217e-01
2.66821682e-01 -3.36055197e-02 1.74349591e-01 1.58089310e-01
-2.09409297e-01 3.53243649e-01 3.38655722e-04 -2.21472859e-01
-7.86258429e-02 -4.38713461e-01 -5.04169226e-01 1.49714425e-01
4.81986329e-02 6.84064150e-01 -3.60343903e-01 -8.21486652e-01
-1.13950878e-01 -2.08505094e-01 -3.88776541e-01 2.03328520e-01
-2.33179629e-01 -1.69261470e-01 -6.09236896e-01 9.06617582e-01
8.11383069e-01 -2.46947706e-01 5.75167716e-01 -9.05235112e-02
1.93017945e-02 1.81178704e-01 -1.34079099e+00 1.57248926e+00
-4.46247756e-02 4.19445306e-01 -2.11457489e-03 -4.74008262e-01
1.03605175e+00 2.32828528e-01 3.95875394e-01 -7.01335132e-01
9.44469869e-02 3.49330306e-01 -1.22189827e-01 -4.94834542e-01
5.47849834e-01 6.14589095e-01 -3.64368290e-01 2.49939695e-01
-7.39534870e-02 -1.73207566e-01 6.34752333e-01 1.48832992e-01
1.05205226e+00 2.56762266e-01 2.87812799e-01 -1.51890755e-01
8.62114191e-01 2.90149659e-01 4.13779020e-01 8.75233769e-01
1.41672075e-01 4.51268703e-01 6.96775079e-01 -3.39524537e-01
-9.32778001e-01 -8.67407680e-01 -2.91484296e-01 7.96529233e-01
-1.60057366e-01 -6.06019735e-01 -1.03075838e+00 -9.21558678e-01
-1.61492765e-01 5.16078591e-01 -4.19349760e-01 2.45982215e-01
-6.78035915e-01 -7.15900719e-01 7.80581355e-01 7.20440209e-01
7.85927713e-01 -1.06520128e+00 -3.26064765e-01 8.88223201e-02
-2.47165605e-01 -1.39903200e+00 -7.61131763e-01 4.44605917e-01
-8.42496157e-01 -1.14349842e+00 -4.65848684e-01 -1.18455565e+00
7.23598659e-01 -1.32936999e-01 8.52775574e-01 -7.20189363e-02
-3.44546735e-01 -1.81324437e-01 -1.32219493e-01 -4.58711714e-01
-3.57152104e-01 1.73619002e-01 -5.71138442e-01 -2.36089692e-01
4.29245532e-01 3.71012717e-01 -1.10296078e-01 2.13619888e-01
-5.08553326e-01 3.54608953e-01 6.07920408e-01 7.11210310e-01
9.35564399e-01 1.94929272e-01 1.43701270e-01 -1.33885741e+00
2.80129462e-01 1.61770701e-01 -1.01103330e+00 5.70364535e-01
-6.20056510e-01 9.42112356e-02 7.62916923e-01 5.34496009e-02
-1.11055899e+00 4.54770595e-01 -3.97359163e-01 -1.84308477e-02
-2.46697947e-01 3.42121780e-01 -7.79472411e-01 3.59428972e-01
4.01269823e-01 4.48316842e-01 -4.94537950e-01 -6.99695408e-01
2.20034897e-01 8.99663568e-01 6.32512808e-01 -4.21336293e-01
5.45771420e-01 2.03823045e-01 2.14541536e-02 -7.40858078e-01
-7.35326231e-01 -6.63829744e-01 -7.86512911e-01 2.14872971e-01
1.07930803e+00 -7.47857988e-01 -8.67723703e-01 5.42078197e-01
-1.18575048e+00 2.55217291e-02 1.18414305e-01 2.90496767e-01
-3.67654800e-01 4.62216258e-01 -8.36560726e-01 -6.09562516e-01
-6.72875583e-01 -9.97230232e-01 1.14219952e+00 2.25313276e-01
-2.45440006e-01 -5.38975298e-01 -3.34933549e-01 6.87192082e-01
-5.04988194e-01 -1.46239936e-01 1.41844785e+00 -1.08573604e+00
-5.63798964e-01 -6.41767442e-01 -4.81049895e-01 -1.86999947e-01
-1.29188493e-01 -7.99525678e-02 -9.03787553e-01 3.06044251e-01
-2.16568515e-01 -1.92226753e-01 9.12598848e-01 3.27456504e-01
1.70466888e+00 6.48065731e-02 -5.76132357e-01 7.03653276e-01
1.25662017e+00 9.69041884e-01 7.29671836e-01 3.68924141e-01
7.17553079e-01 4.84311521e-01 6.47055447e-01 3.71767908e-01
2.96069682e-01 5.64953089e-01 -1.92530096e-01 -1.71633035e-01
-1.27949774e-01 -3.92308950e-01 6.64266497e-02 4.67724442e-01
3.16202343e-01 -7.53198385e-01 -1.06118250e+00 -8.12259316e-02
-1.40642536e+00 -6.96488976e-01 -3.96905363e-01 2.10159945e+00
7.86800623e-01 7.34235525e-01 1.44254625e-01 3.82820427e-01
6.31852090e-01 -2.60511220e-01 -5.05358279e-01 -4.18176740e-01
-1.41232714e-01 8.90055671e-02 6.30550623e-01 2.12373734e-01
-1.31571436e+00 1.16331935e+00 6.36454105e+00 9.41473246e-01
-7.53288567e-01 -4.93600547e-01 1.16410172e+00 5.59628248e-01
-2.12799553e-02 -3.88079375e-01 -1.42670286e+00 3.22080016e-01
6.88379884e-01 -2.47156564e-02 2.36129835e-02 7.16029823e-01
-9.60676372e-02 -2.98321068e-01 -1.09103692e+00 1.17665827e+00
4.50798839e-01 -1.60392153e+00 2.83503413e-01 -2.47536585e-01
2.17794493e-01 -7.29622066e-01 -1.48600504e-01 2.55332798e-01
-4.65269648e-02 -9.88577604e-01 5.96129358e-01 2.80188888e-01
1.01060176e+00 -7.34717309e-01 7.85495043e-01 2.39613816e-01
-1.18667269e+00 3.05612475e-01 -1.24072038e-01 8.12298715e-01
-4.49562311e-01 3.02251458e-01 -1.17288637e+00 5.98563552e-01
4.99202996e-01 5.55434048e-01 -7.53538668e-01 1.11112332e+00
-3.19551080e-01 6.67193294e-01 -6.69843853e-02 -5.70657104e-02
-2.75068283e-01 -2.52080083e-01 1.47945151e-01 1.44526219e+00
2.06444077e-02 -2.84074731e-02 2.26244658e-01 5.13245821e-01
-6.47189081e-01 6.38362586e-01 -2.20186844e-01 -2.61575252e-01
4.92191404e-01 1.14928508e+00 -1.51782405e+00 -7.00542271e-01
-3.62020910e-01 1.09935665e+00 1.00467615e-02 1.49077073e-01
-6.83909178e-01 -9.26099718e-01 -3.62874895e-01 -1.71481133e-01
4.88240898e-01 3.24169070e-01 -1.06684506e+00 -1.06803477e+00
1.77044719e-01 -1.05608463e+00 7.40740061e-01 -8.87524784e-01
-6.53761268e-01 6.51983023e-01 -3.30812305e-01 -8.88015926e-01
-1.37410285e-02 -7.68505871e-01 -3.57147366e-01 7.24229634e-01
-9.43012714e-01 -9.24277067e-01 -9.12741497e-02 2.27807403e-01
9.81586099e-01 -2.91973144e-01 7.93266296e-01 1.24546967e-01
-1.08238935e+00 9.63334322e-01 1.59607202e-01 6.95818841e-01
7.86227286e-01 -1.51763272e+00 5.65681040e-01 9.07436132e-01
-4.65493686e-02 5.88133991e-01 3.70495677e-01 -1.08536863e+00
-1.54002666e+00 -9.77645874e-01 1.19624901e+00 -5.00186622e-01
4.35983062e-01 -8.07935953e-01 -8.23401034e-01 7.23637462e-01
-5.47659919e-02 -3.25939566e-01 6.07396722e-01 -3.60466912e-02
-4.68252271e-01 9.60766710e-03 -1.10899532e+00 5.08666933e-01
5.67184210e-01 -2.56763071e-01 -4.85154957e-01 5.06714046e-01
1.50093064e-01 -9.40576315e-01 -9.33922291e-01 4.98406403e-02
6.37998819e-01 -3.59722048e-01 7.18483865e-01 -2.92918921e-01
6.57135606e-01 -1.68298796e-01 1.05069034e-01 -4.13673639e-01
-1.59183681e-01 -3.23058546e-01 -1.21798456e-01 1.02079296e+00
8.60898912e-01 9.94731579e-03 1.42313707e+00 3.91121000e-01
-2.16483340e-01 -5.63201308e-01 -7.29036689e-01 -5.16957462e-01
1.47046655e-01 -1.24333993e-01 5.75971425e-01 4.81655687e-01
2.20589146e-01 7.11237788e-01 -3.97557579e-02 -7.00201988e-02
3.45705271e-01 4.09450650e-01 7.23333299e-01 -1.06332433e+00
-1.98041111e-01 -5.79841316e-01 -5.05527370e-02 -1.23104703e+00
-8.35940167e-02 -9.97792661e-01 2.14238733e-01 -1.66033292e+00
3.78375232e-01 2.80718207e-02 8.77723470e-02 4.69877839e-01
-1.46110997e-01 1.26320096e-02 1.58158660e-01 6.86327666e-02
-8.26779246e-01 -1.96664199e-01 1.13563085e+00 -2.35133827e-01
-1.57042384e-01 8.11872855e-02 -5.17967463e-01 5.97711027e-01
6.19080067e-01 -4.24368709e-01 -1.35426462e-01 -8.57657641e-02
-4.92682159e-02 5.80124557e-01 -4.97446895e-01 -8.10024440e-01
3.09508502e-01 -1.95156708e-02 1.18585825e+00 -1.49449778e+00
1.94924846e-02 -6.29147768e-01 -3.13435793e-01 5.49358070e-01
-3.38978231e-01 1.93867072e-01 4.39750910e-01 2.82584637e-01
1.25470906e-01 -4.61084127e-01 4.56388831e-01 1.05958451e-02
-5.05475640e-01 1.79296106e-01 -7.37555385e-01 8.82875621e-02
6.59264386e-01 -3.52601618e-01 -5.66089690e-01 1.63909242e-01
-6.63929760e-01 2.82776982e-01 2.65110731e-01 1.66664585e-01
7.25768149e-01 -7.75633693e-01 -3.93652797e-01 3.98911506e-01
2.94501811e-01 5.99823371e-02 1.16421632e-01 5.14582694e-01
-8.52806985e-01 8.76014531e-01 7.32116029e-02 -3.96343142e-01
-1.54416835e+00 5.33753037e-01 1.33956999e-01 -6.00044012e-01
-6.81194484e-01 9.72490609e-01 2.70830452e-01 -1.05889969e-01
5.64574599e-01 -5.92824340e-01 -6.52039766e-01 1.76432729e-01
6.47454917e-01 1.36595041e-01 5.83649039e-01 -1.18133873e-01
-5.67699194e-01 5.59112072e-01 -6.53467417e-01 6.04085363e-02
9.02181387e-01 2.02255234e-01 1.13762744e-01 2.89971173e-01
8.05608571e-01 4.19277579e-01 -6.11563444e-01 1.45418823e-01
6.15552545e-01 2.20515728e-02 6.94882795e-02 -1.14094925e+00
-6.78425729e-01 5.17944276e-01 5.75354397e-01 -8.17072093e-02
1.00760138e+00 -2.65889436e-01 5.87513387e-01 6.07988000e-01
5.21338731e-02 -1.19946826e+00 -3.54249887e-02 5.12881994e-01
6.99222863e-01 -1.25146699e+00 2.22476900e-01 -1.02433407e+00
-6.08660638e-01 1.48884952e+00 7.40039647e-01 2.02730790e-01
3.41774344e-01 6.34902716e-01 -1.18771225e-01 -9.00117084e-02
-8.65823746e-01 1.34447396e-01 5.58650374e-01 2.46769965e-01
7.54881918e-01 -1.15978137e-01 -3.04652661e-01 8.92211616e-01
-2.24670887e-01 -2.53146999e-02 5.47184169e-01 1.05255842e+00
-4.85611349e-01 -9.36299562e-01 -4.59180832e-01 7.26713717e-01
-7.72708058e-01 -1.50543541e-01 -7.06797719e-01 7.54713893e-01
-7.01227784e-02 8.32352757e-01 3.12711686e-01 -2.76666105e-01
5.33548892e-01 1.09368183e-01 4.04239297e-01 -8.55253935e-01
-7.05660403e-01 7.33251929e-01 3.22071165e-01 -1.67716712e-01
2.71447748e-01 -7.70254612e-01 -1.65380895e+00 1.70566200e-03
-2.31363252e-01 3.44048083e-01 6.33656859e-01 9.37277138e-01
1.55023620e-01 5.86331725e-01 2.09519327e-01 2.74758935e-01
-3.44820134e-02 -7.94764400e-01 -4.84232247e-01 4.87042293e-02
-1.53832007e-02 -1.67408898e-01 3.30661178e-01 3.72143924e-01]
|
[11.697680473327637, 3.0489230155944824]
|
86e2f62a-7681-498d-a632-47dcc58f762d
|
train-short-test-long-attention-with-linear
|
2108.12409
| null |
https://arxiv.org/abs/2108.12409v2
|
https://arxiv.org/pdf/2108.12409v2.pdf
|
Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation
|
Since the introduction of the transformer model by Vaswani et al. (2017), a fundamental question has yet to be answered: how does a model achieve extrapolation at inference time for sequences that are longer than it saw during training? We first show that extrapolation can be enabled by simply changing the position representation method, though we find that current methods do not allow for efficient extrapolation. We therefore introduce a simpler and more efficient position method, Attention with Linear Biases (ALiBi). ALiBi does not add positional embeddings to word embeddings; instead, it biases query-key attention scores with a penalty that is proportional to their distance. We show that this method trains a 1.3 billion parameter model on input sequences of length 1024 that extrapolates to input sequences of length 2048, achieving the same perplexity as a sinusoidal position embedding model trained on inputs of length 2048 but training 11% faster and using 11% less memory. ALiBi's inductive bias towards recency also leads it to outperform multiple strong position methods on the WikiText-103 benchmark.
|
['Mike Lewis', 'Noah A. Smith', 'Ofir Press']
|
2021-08-27
|
train-short-test-long-attention-with-linear-1
|
https://openreview.net/forum?id=R8sQPpGCv0
|
https://openreview.net/pdf?id=R8sQPpGCv0
|
iclr-2022-4
|
['2048']
|
['playing-games']
|
[ 2.66915619e-01 2.92762309e-01 -2.76777983e-01 -2.51790941e-01
-7.70481825e-01 -7.43288219e-01 7.88598120e-01 2.60067701e-01
-1.08359015e+00 7.32891440e-01 5.88974893e-01 -9.41816032e-01
1.96012571e-01 -7.86417902e-01 -1.02513289e+00 -3.85576606e-01
-1.14398919e-01 5.96315086e-01 4.56191212e-01 -3.00947368e-01
3.66136551e-01 2.55555928e-01 -1.31202126e+00 3.36424649e-01
4.27564144e-01 5.84967613e-01 2.15282366e-01 1.14220178e+00
-1.26699328e-01 4.70070332e-01 -5.43147981e-01 -5.46318352e-01
1.79461807e-01 -2.61542618e-01 -1.08381033e+00 -6.28897727e-01
5.46418786e-01 -6.33693278e-01 -5.44387341e-01 6.06735289e-01
5.63983560e-01 2.15342298e-01 7.80155301e-01 -8.13155770e-01
-1.24385214e+00 8.75981987e-01 -3.28558505e-01 7.34918416e-01
1.27982959e-01 1.95048571e-01 1.47632563e+00 -1.14454067e+00
5.74602604e-01 1.07422531e+00 8.99304211e-01 5.15291393e-01
-1.31021190e+00 -4.29728657e-01 1.86405301e-01 3.47483933e-01
-1.10391450e+00 -1.06286965e-01 1.76002860e-01 -2.65831888e-01
1.73082542e+00 2.12620065e-01 6.01367176e-01 1.50333786e+00
1.96988136e-01 6.97545290e-01 6.81608081e-01 -6.66938186e-01
-1.05288088e-01 1.45443290e-01 2.11241096e-01 3.25026512e-01
4.46463108e-01 2.16550767e-01 -3.91113073e-01 -1.69746116e-01
5.07162213e-01 -1.58526868e-01 -3.81446660e-01 4.21583168e-02
-1.37307167e+00 1.06508565e+00 3.73622656e-01 3.40588182e-01
-2.78434575e-01 5.38367748e-01 6.11552060e-01 3.83326828e-01
3.95026147e-01 9.27653134e-01 -7.33728111e-01 -4.70703632e-01
-8.52165759e-01 3.29172820e-01 5.99000454e-01 8.83505881e-01
5.74133754e-01 -3.24842334e-02 -1.44416705e-01 6.45788372e-01
8.86916220e-02 3.49158317e-01 8.97664547e-01 -8.20230424e-01
4.94302958e-01 -4.74550528e-03 1.60340294e-01 -5.81187248e-01
-1.94019750e-01 -4.31932658e-01 -1.59986541e-01 7.64274923e-03
6.46163166e-01 -2.66488850e-01 -8.04338515e-01 1.93952107e+00
-1.39558807e-01 7.17412457e-02 -4.95622121e-02 6.74294233e-01
1.89998955e-01 1.08116424e+00 1.02384381e-01 2.63117075e-01
1.31965065e+00 -1.01174235e+00 -4.60452020e-01 -4.89739239e-01
1.11038554e+00 -7.75173128e-01 1.47099721e+00 3.45580488e-01
-1.06320322e+00 -5.22818029e-01 -1.33676302e+00 -4.77382034e-01
-5.93542695e-01 -2.82458097e-01 5.94810545e-01 6.08744681e-01
-1.25462270e+00 7.39423692e-01 -7.46377170e-01 -5.34308553e-01
8.07645619e-02 1.07716940e-01 -3.20944965e-01 1.98573589e-01
-1.65434587e+00 1.48761666e+00 4.44182634e-01 -1.75480649e-01
-5.37289679e-01 -1.00259817e+00 -7.22447634e-01 2.01576725e-01
-8.44369382e-02 -5.08123815e-01 1.36515284e+00 -8.21463406e-01
-1.34667015e+00 5.83184600e-01 -1.76923245e-01 -9.17740405e-01
1.87305957e-01 -6.13954902e-01 -2.76559234e-01 -8.02537054e-02
-2.61865497e-01 7.38994956e-01 3.91330302e-01 -6.36615813e-01
-5.68941832e-01 -3.16597223e-02 1.68538690e-01 4.87457179e-02
-7.25398481e-01 -9.97094214e-02 -6.32080510e-02 -5.29495060e-01
-4.72599566e-01 -8.64110589e-01 -8.12277570e-03 9.92715582e-02
6.92705885e-02 -4.33877975e-01 3.46189827e-01 -7.68124044e-01
1.43056548e+00 -1.95932877e+00 8.20217580e-02 -1.82347551e-01
7.69832656e-02 5.72938025e-01 -5.97224236e-01 8.93513918e-01
-2.86639333e-01 3.45542878e-01 -1.99952856e-01 -1.15933642e-01
2.56837994e-01 3.11738044e-01 -6.89886630e-01 3.43140453e-01
3.79850030e-01 1.05885708e+00 -1.16293550e+00 -2.32950345e-01
-1.14574641e-01 5.02540469e-01 -1.02214026e+00 1.47841066e-01
-1.61731601e-01 -3.29155505e-01 9.30217952e-02 -1.03718325e-01
4.39300478e-01 -2.71373212e-01 7.19541013e-02 -1.30919397e-01
-1.86218515e-01 7.51978040e-01 -5.44728279e-01 1.63173676e+00
-7.80814588e-01 1.14112556e+00 -6.49604499e-01 -8.43325377e-01
6.49363697e-01 2.97906190e-01 1.54449549e-02 -6.77460849e-01
-1.28313020e-01 3.89940828e-01 3.19774479e-01 -5.37245333e-01
7.69172490e-01 -3.18078101e-01 1.47690736e-02 6.70323193e-01
2.53890067e-01 5.63652366e-02 8.21627602e-02 3.17111641e-01
1.40063691e+00 2.67656982e-01 6.17785053e-03 -1.79727301e-01
1.12216301e-01 -1.53566688e-01 1.54106706e-01 8.19441557e-01
-5.63153997e-02 5.42043805e-01 5.06149709e-01 -5.18852353e-01
-1.78077602e+00 -1.14600062e+00 -2.42892280e-01 1.47418809e+00
-3.15878451e-01 -6.03965878e-01 -4.94428962e-01 -6.77633286e-01
1.46466434e-01 1.28715336e+00 -9.58669901e-01 -2.88366884e-01
-7.32025087e-01 -5.42230308e-01 8.72858286e-01 9.02118385e-01
-2.88607001e-01 -1.06260300e+00 -7.19717443e-01 3.02830219e-01
-1.46232828e-01 -6.78569198e-01 -6.79414272e-01 4.89371508e-01
-7.78732955e-01 -6.13108873e-01 -8.78939271e-01 -5.33535898e-01
4.50276434e-01 -1.69411480e-01 1.33291245e+00 -5.33181131e-02
-2.19502151e-01 2.01724902e-01 -5.19027412e-01 -3.85128379e-01
-3.05238724e-01 6.05696797e-01 -6.24863133e-02 -6.59540713e-01
8.53750527e-01 -6.09645665e-01 -7.49438822e-01 -6.41413629e-02
-1.04392338e+00 -2.48785496e-01 8.41265142e-01 1.17242110e+00
2.25112215e-02 -8.16799402e-01 7.10647821e-01 -9.33414459e-01
7.81395674e-01 -6.88031375e-01 -4.13000286e-01 4.60240841e-02
-7.44942963e-01 6.33960783e-01 7.78635681e-01 -6.07799590e-01
-7.00259387e-01 -4.64667499e-01 -7.10766673e-01 -1.30977154e-01
1.43211767e-01 4.40795630e-01 4.86491293e-01 4.18029934e-01
9.10792828e-01 3.54316562e-01 7.40309060e-02 -4.33306217e-01
8.50251973e-01 7.05069125e-01 1.77498668e-01 -6.00023687e-01
6.58743560e-01 1.03700850e-02 -4.41711783e-01 -6.68824792e-01
-7.74793625e-01 -2.47054800e-01 -4.20952827e-01 2.67789304e-01
7.55241990e-01 -6.81123495e-01 -5.92503130e-01 -8.11307132e-03
-1.45604992e+00 -7.02487648e-01 -4.26463008e-01 5.59388936e-01
-5.14961302e-01 3.67008507e-01 -8.23697567e-01 -5.08147717e-01
-3.47482413e-01 -6.95430040e-01 7.78089523e-01 -3.49776000e-01
-7.67590761e-01 -1.11769271e+00 3.69318634e-01 -2.92422682e-01
8.92399371e-01 -2.44567066e-01 1.14279401e+00 -9.82781351e-01
-2.74429172e-01 -1.13424003e-01 -1.74613193e-01 6.93383753e-01
-5.32754697e-02 -1.88143589e-02 -8.15104544e-01 -3.21412176e-01
-1.53199509e-01 -3.94476622e-01 1.02366138e+00 -1.27126023e-01
1.11791337e+00 -5.23408651e-01 -1.66083053e-01 5.53365171e-01
1.56429803e+00 9.08687245e-03 7.08093345e-01 5.61006546e-01
3.30249131e-01 3.15746933e-01 3.44159454e-01 1.81901753e-01
2.09747806e-01 4.00293708e-01 3.89118701e-01 2.27124825e-01
-2.51534045e-01 -5.54839134e-01 4.84999955e-01 1.03891385e+00
-7.06494600e-02 -3.84698510e-01 -7.81508327e-01 1.09689546e+00
-1.38124430e+00 -1.13886547e+00 -1.35301910e-02 2.36808133e+00
1.08855700e+00 4.66986954e-01 -8.50205719e-02 2.21929676e-03
2.04212204e-01 2.90907830e-01 -4.43157077e-01 -1.06682873e+00
7.37180263e-02 6.53990388e-01 8.93500447e-01 7.76397228e-01
-6.02532566e-01 7.47496724e-01 7.22342825e+00 5.80549121e-01
-1.02427340e+00 3.39598268e-01 4.01571482e-01 -3.21576804e-01
-7.38455892e-01 9.91420671e-02 -9.62465644e-01 6.26146615e-01
1.60061860e+00 -2.18063727e-01 3.81650358e-01 7.00500190e-01
-3.17904413e-01 1.47942469e-01 -1.43601096e+00 4.62532014e-01
2.07156941e-01 -1.27899623e+00 2.58498818e-01 4.68028896e-02
5.57526767e-01 3.84753495e-01 2.80565530e-01 5.95312417e-01
3.40671510e-01 -1.10660458e+00 6.12588704e-01 3.51259440e-01
6.66622877e-01 -7.11356282e-01 7.60104597e-01 3.76839876e-01
-5.54549873e-01 -1.10096321e-01 -7.45213568e-01 -1.61637858e-01
2.39629775e-01 3.04247499e-01 -1.05705655e+00 1.60680369e-01
3.43633473e-01 2.25009263e-01 -3.53946447e-01 8.15839410e-01
-2.89193183e-01 8.95661235e-01 -4.20372963e-01 -4.93669420e-01
5.53116202e-01 2.20186025e-01 2.68927008e-01 1.64742410e+00
6.13582671e-01 -3.09277803e-01 -6.65034533e-01 6.45722687e-01
-7.99172595e-02 -4.74367104e-02 -7.64499068e-01 -1.30531594e-01
7.49416292e-01 7.80857086e-01 5.39936423e-02 -4.23022479e-01
-6.38603926e-01 1.13236868e+00 6.78484917e-01 4.19852406e-01
-1.04218721e+00 -8.64161670e-01 6.22654915e-01 3.71265709e-01
8.56400311e-01 -4.19484437e-01 -7.18126213e-03 -8.28559995e-01
-1.47785455e-01 -7.87025213e-01 2.48136669e-01 -7.48761714e-01
-1.28907454e+00 7.55874217e-01 -5.05593680e-02 -8.01831901e-01
-4.50736254e-01 -9.59613919e-01 -4.82564032e-01 1.22320759e+00
-1.62385190e+00 -7.31431544e-01 2.98052192e-01 -4.94099706e-02
5.82512259e-01 2.45415330e-01 1.06284726e+00 4.46334213e-01
-1.23305790e-01 9.38023984e-01 2.44569495e-01 1.89338058e-01
8.27122271e-01 -1.42299843e+00 9.47160900e-01 6.39031053e-01
2.77276665e-01 1.05494106e+00 9.78763819e-01 -1.70066357e-01
-1.34572506e+00 -1.04548073e+00 1.40158796e+00 -9.48225498e-01
9.86834526e-01 -4.07302588e-01 -1.05902278e+00 1.25512159e+00
6.77264035e-01 -3.15801390e-02 7.24364102e-01 3.41112494e-01
-7.80859172e-01 6.57987222e-02 -5.41136265e-01 7.60274351e-01
1.12836695e+00 -6.21686935e-01 -1.11725008e+00 2.82288879e-01
1.12470126e+00 -2.13716105e-01 -7.73255646e-01 1.28088847e-01
6.41819775e-01 -7.10492373e-01 9.40026462e-01 -9.58498955e-01
6.48601115e-01 4.70789783e-02 -1.55974448e-01 -1.54274130e+00
-5.83113134e-01 -2.62319356e-01 -1.97664529e-01 8.02891970e-01
8.30974579e-01 -1.02836561e+00 4.73582745e-01 2.01706037e-01
-2.23954871e-01 -1.08716834e+00 -7.92256832e-01 -1.09939468e+00
8.58808696e-01 -4.50937003e-01 5.97470164e-01 6.22687161e-01
3.19861352e-01 4.79900539e-01 -2.29882926e-01 -1.14994198e-01
2.67612964e-01 -2.79107958e-01 4.89281356e-01 -8.11364472e-01
-4.90278959e-01 -5.47357500e-01 -2.86157280e-01 -1.71489692e+00
9.63031687e-03 -1.04040217e+00 5.65192439e-02 -1.39771152e+00
9.99424141e-03 -3.19306761e-01 -5.41884840e-01 3.55217844e-01
-2.60095477e-01 2.70757765e-01 -5.47314920e-02 -1.85905918e-01
-2.86524475e-01 6.02254212e-01 9.14198518e-01 3.97746526e-02
4.81945157e-01 -4.41540688e-01 -7.75747061e-01 3.77125055e-01
6.78156257e-01 -4.92367685e-01 -5.09344101e-01 -8.86990607e-01
6.65892005e-01 -3.63970757e-01 1.48656800e-01 -7.97098339e-01
1.34052351e-01 2.32258588e-01 3.80345106e-01 -4.77599323e-01
4.03433919e-01 -5.46306729e-01 -2.00795218e-01 4.77495968e-01
-7.82888532e-01 5.05608439e-01 4.19038802e-01 4.41014379e-01
5.00021800e-02 -6.67183757e-01 4.61647749e-01 -1.35798743e-02
-4.70275223e-01 3.31346989e-02 -5.85401595e-01 2.13234603e-01
7.48808205e-01 -6.58642054e-02 -3.56786698e-01 -2.86336303e-01
-5.21051884e-01 -3.07195541e-02 3.67251217e-01 4.34169292e-01
4.47101772e-01 -1.42592883e+00 -8.20854247e-01 1.67182714e-01
1.56061456e-01 -4.29343253e-01 -1.09312944e-01 6.17423534e-01
-6.10321045e-01 7.21121013e-01 -7.02163056e-02 -2.31452867e-01
-7.95836151e-01 8.14554393e-01 1.37723371e-01 -2.85451353e-01
-5.21232426e-01 1.19129384e+00 -3.74048352e-02 -4.41737831e-01
4.72453982e-02 -5.91193080e-01 2.02553049e-01 3.43039781e-02
6.47269726e-01 1.94379672e-01 1.05324864e-01 -1.61247537e-01
-3.09372962e-01 4.11986679e-01 -4.45599020e-01 -3.93325031e-01
1.40696299e+00 4.53778096e-02 1.61936149e-01 6.72461629e-01
1.52456796e+00 -1.28628165e-02 -1.18685079e+00 -2.76088208e-01
-4.17538593e-03 -4.86435652e-01 -2.97160417e-01 -6.79717183e-01
-4.47677612e-01 1.34667349e+00 3.41537923e-01 3.57224494e-01
5.74515581e-01 -1.12841256e-01 1.04555750e+00 4.44021612e-01
2.14545950e-01 -1.04061484e+00 1.84763119e-01 8.37465584e-01
9.34693933e-01 -7.68190742e-01 -2.16815576e-01 4.01616186e-01
-3.39840621e-01 1.15651965e+00 4.72267717e-01 -2.86956966e-01
4.00721371e-01 1.94042221e-01 -1.43999070e-01 6.24736920e-02
-1.18913758e+00 6.87612966e-02 -6.49441453e-03 5.27604699e-01
8.25431705e-01 -2.33145192e-01 -5.42622328e-01 2.37832353e-01
-3.03680629e-01 -1.00020096e-01 4.67570066e-01 7.10107744e-01
-7.08220184e-01 -1.24703729e+00 -1.06095418e-01 5.13374865e-01
-5.09175062e-01 -7.13569283e-01 1.88674256e-01 8.72208416e-01
-5.99974208e-02 4.73323137e-01 4.14497584e-01 -3.06604475e-01
2.23380685e-01 5.35187006e-01 6.11734152e-01 -6.38098180e-01
-3.84437233e-01 -6.23865604e-01 1.36325940e-01 -2.80380309e-01
2.66370386e-01 -5.28345883e-01 -9.64838028e-01 -3.81821722e-01
-3.02402735e-01 3.06638658e-01 6.66532040e-01 7.65703738e-01
4.26183671e-01 5.79352379e-01 3.11118841e-01 -7.28471279e-01
-1.12938082e+00 -1.23493469e+00 -4.24511731e-03 3.79091233e-01
4.56275880e-01 -2.72194237e-01 -5.53358972e-01 -9.56535786e-02]
|
[10.750683784484863, 8.581208229064941]
|
6b76968f-e8ca-4f1a-9f67-0f6e341f2a3e
|
higher-order-recurrent-space-time-transformer
|
2104.08665
| null |
https://arxiv.org/abs/2104.08665v3
|
https://arxiv.org/pdf/2104.08665v3.pdf
|
Higher Order Recurrent Space-Time Transformer for Video Action Prediction
|
Endowing visual agents with predictive capability is a key step towards video intelligence at scale. The predominant modeling paradigm for this is sequence learning, mostly implemented through LSTMs. Feed-forward Transformer architectures have replaced recurrent model designs in ML applications of language processing and also partly in computer vision. In this paper we investigate on the competitiveness of Transformer-style architectures for video predictive tasks. To do so we propose HORST, a novel higher order recurrent layer design whose core element is a spatial-temporal decomposition of self-attention for video. HORST achieves state of the art competitive performance on Something-Something early action recognition and EPIC-Kitchens action anticipation, showing evidence of predictive capability that we attribute to our recurrent higher order design of self-attention.
|
['Oswald Lanz', 'Cheng-Kuang Lee', 'Giuseppe Fiameni', 'Tsung-Ming Tai']
|
2021-04-17
| null | null | null | null |
['action-anticipation']
|
['computer-vision']
|
[ 1.31149039e-01 2.06484288e-01 -3.46932501e-01 -1.22961193e-01
-1.73907980e-01 -1.00869879e-01 1.22863555e+00 -4.03375626e-01
-2.17951015e-01 1.56784460e-01 9.18247640e-01 -2.99689323e-01
3.91599424e-02 -3.58216763e-01 -8.90145957e-01 -7.63645768e-01
-2.25207001e-01 4.84222025e-01 2.26508319e-01 -4.73438263e-01
1.58104733e-01 2.60690510e-01 -1.50092864e+00 1.32858276e+00
5.04935570e-02 1.10490131e+00 6.41621649e-01 1.09373128e+00
-5.53845093e-02 2.20078635e+00 1.36962458e-01 -3.11866373e-01
6.37831092e-02 -3.62263620e-01 -7.55928338e-01 1.16248257e-01
1.47726536e-01 -1.41557321e-01 -7.28239775e-01 3.53359610e-01
1.63565993e-01 1.53426021e-01 6.15503073e-01 -1.08268273e+00
-8.15826774e-01 1.04714644e+00 -3.65995139e-01 5.94123960e-01
4.22924340e-01 4.31172878e-01 1.16205943e+00 -8.85930896e-01
8.56784225e-01 1.39915121e+00 8.75197768e-01 7.58826852e-01
-1.18242145e+00 -3.75614464e-02 4.34537143e-01 9.70938146e-01
-8.42510104e-01 -7.14858532e-01 6.94036007e-01 -4.61715400e-01
1.84333682e+00 8.04180503e-02 1.08981681e+00 1.69446647e+00
4.99647230e-01 1.44308269e+00 8.76775146e-01 -2.67199725e-01
8.03226158e-02 -1.02209792e-01 -3.48114341e-01 8.30965161e-01
-6.40843213e-01 2.96172738e-01 -1.01257288e+00 4.22377229e-01
8.87586594e-01 2.20469728e-01 4.02923934e-02 -4.05887753e-01
-1.24217510e+00 7.38408148e-01 6.13842309e-01 4.89847571e-01
-8.65754604e-01 7.44115710e-01 9.36482489e-01 5.46684921e-01
5.46039701e-01 2.70998746e-01 -5.08862019e-01 -3.85134250e-01
-8.30067694e-01 -2.89236158e-01 4.18990850e-01 9.33319688e-01
8.20743367e-02 7.54225492e-01 -4.83663559e-01 5.20368278e-01
3.36012423e-01 8.65774751e-02 9.71232235e-01 -9.61158872e-01
3.05599779e-01 4.38193768e-01 -2.55793184e-01 -6.44146919e-01
-4.77010131e-01 -2.81874537e-01 -9.06740069e-01 1.57295257e-01
-3.36680859e-02 2.10753739e-01 -8.65222752e-01 1.51880801e+00
-4.67989564e-01 4.50763047e-01 3.46115679e-01 7.20847189e-01
5.71161270e-01 9.56314802e-01 6.23534024e-01 -3.81251425e-01
1.15846348e+00 -1.23759305e+00 -3.63325328e-01 -5.52935243e-01
4.75552410e-01 -3.63722444e-01 8.92831922e-01 2.99319059e-01
-1.23099804e+00 -7.85316765e-01 -7.23454177e-01 -1.87403217e-01
2.11222973e-02 3.05537641e-01 8.62711728e-01 4.46027406e-02
-1.58880782e+00 6.93226993e-01 -9.37718451e-01 -6.44671917e-01
3.32231969e-01 2.14233652e-01 -4.33081061e-01 3.60906124e-01
-9.37816799e-01 1.14797056e+00 3.13441396e-01 1.54645950e-01
-1.15977216e+00 -5.57249188e-01 -8.32381904e-01 2.45809853e-01
2.52660424e-01 -1.04637408e+00 1.43121183e+00 -1.62840021e+00
-1.70269740e+00 9.62447345e-01 -4.44827974e-01 -1.56631315e+00
5.44426739e-01 -4.01030689e-01 -1.33715436e-01 2.24784881e-01
-3.41556519e-01 7.97102571e-01 1.44805562e+00 -7.96313226e-01
-5.65038264e-01 -1.99108869e-01 -2.48019665e-01 1.82743534e-01
-1.83262050e-01 1.89662740e-01 -1.51771717e-02 -6.86190307e-01
-2.54866689e-01 -8.56590688e-01 -4.85840052e-01 -3.86932462e-01
1.78353280e-01 -6.32296085e-01 1.00198209e+00 -6.96526527e-01
9.08953846e-01 -1.85510552e+00 4.69257444e-01 -4.26958799e-01
2.23139659e-01 4.09121096e-01 -4.44069207e-01 6.61451340e-01
-4.52651083e-01 -1.46818206e-01 4.66765344e-01 -4.63532805e-01
1.09603606e-01 1.05243504e-01 -1.05937207e+00 2.00980365e-01
5.94031930e-01 1.43239439e+00 -7.14542389e-01 -3.30251306e-01
6.21004820e-01 5.22741258e-01 -4.10330385e-01 3.31341356e-01
-8.41072083e-01 1.32650703e-01 -3.89768958e-01 3.98675084e-01
-2.27058157e-01 -3.79725993e-01 2.78380513e-01 -1.62431285e-01
-3.71326983e-01 3.27832788e-01 -8.41012374e-02 1.79215562e+00
-4.63396579e-01 1.16423786e+00 -4.81667161e-01 -1.23129129e+00
7.49494731e-01 6.92599952e-01 4.89091069e-01 -1.13873720e+00
-5.71194626e-02 -1.90790698e-01 1.29490495e-02 -7.96229005e-01
6.04033232e-01 -2.95503706e-01 2.90892959e-01 1.51212975e-01
2.34780490e-01 2.47076660e-01 1.15057908e-01 2.55653977e-01
1.31274891e+00 8.39642048e-01 4.06241655e-01 -8.17413107e-02
5.11154056e-01 2.03842632e-02 2.29498789e-01 7.16472805e-01
-1.86491366e-02 3.08140516e-01 7.73301959e-01 -1.00495994e+00
-1.63104868e+00 -7.49445736e-01 6.20246053e-01 1.72984719e+00
-6.19426668e-01 -7.61498928e-01 -3.06764305e-01 -2.50327498e-01
-5.13878882e-01 7.70647347e-01 -9.51769352e-01 -3.37284595e-01
-9.20537531e-01 -2.75275677e-01 3.51824045e-01 9.83793557e-01
3.07158828e-01 -1.67911875e+00 -1.17800784e+00 4.20710385e-01
1.91583093e-02 -1.25489247e+00 -1.19959652e-01 4.45470601e-01
-9.94538784e-01 -6.08279407e-01 -7.08180785e-01 -8.27186108e-01
-1.01436689e-01 2.73974668e-02 1.58607256e+00 -2.61559099e-01
7.63052106e-02 6.59824848e-01 -4.25007164e-01 -6.32680506e-02
-5.05868733e-01 -7.21461028e-02 7.27710500e-02 5.22125401e-02
5.76810762e-02 -6.87150180e-01 -4.01847631e-01 -9.28126425e-02
-5.52891254e-01 5.96747935e-01 9.06354189e-01 8.58170569e-01
2.07432076e-01 -5.68763137e-01 2.48796821e-01 -4.77253854e-01
2.91105181e-01 -2.95942903e-01 -4.03183937e-01 4.43739295e-01
-1.44076079e-01 4.13120300e-01 8.93036664e-01 -5.39374888e-01
-1.21881199e+00 2.19418824e-01 -9.29504558e-02 -1.06607914e+00
1.09531440e-01 5.28438509e-01 4.46678907e-01 1.88393742e-01
3.56408328e-01 7.67620325e-01 -2.34555602e-02 -9.77984369e-02
3.92908812e-01 -5.39943911e-02 3.53980273e-01 -1.47256002e-01
1.38195738e-01 3.72671694e-01 1.55690789e-01 -1.00902498e+00
-6.24086201e-01 -1.69055894e-01 -4.55320269e-01 -4.42111075e-01
1.08722305e+00 -1.07686341e+00 -1.22852767e+00 4.32516605e-01
-1.24967611e+00 -9.14871395e-01 -5.36364734e-01 2.24058449e-01
-1.28020942e+00 -5.16846478e-02 -1.07279682e+00 -1.00814819e+00
-4.17791724e-01 -8.60540092e-01 8.89597535e-01 1.43578231e-01
-1.10480271e-01 -1.13554740e+00 1.40318200e-01 1.47031382e-01
5.84576070e-01 -1.93268016e-01 7.54809320e-01 -5.27919650e-01
-8.78872693e-01 1.98932618e-01 -7.67089427e-02 9.75142885e-03
-6.22852027e-01 1.05291456e-01 -9.88100648e-01 -7.21459016e-02
1.04715370e-01 -5.58965385e-01 1.44204104e+00 7.62220562e-01
9.64219570e-01 -3.62938792e-01 -2.11031660e-01 6.54140651e-01
1.21860278e+00 2.84334719e-01 9.79630291e-01 5.07738233e-01
6.90709889e-01 6.00294888e-01 3.70301336e-01 6.48727536e-01
3.45422298e-01 7.19051063e-01 3.94087553e-01 1.33085549e-01
-3.39073479e-01 -6.03326559e-01 1.11383498e+00 8.48546863e-01
-5.32829762e-01 -2.47274145e-01 -9.46022451e-01 4.23234195e-01
-2.17956233e+00 -1.69684601e+00 -6.27904236e-02 1.54106379e+00
3.18586141e-01 2.10852623e-01 2.15045229e-01 -2.64869928e-01
3.63547891e-01 4.12776262e-01 -3.64287317e-01 -6.91138029e-01
-2.72421241e-01 -2.99650989e-02 3.14147055e-01 1.39270097e-01
-1.17800450e+00 1.38172877e+00 6.43008471e+00 8.39792371e-01
-1.30915737e+00 5.57758361e-02 9.02882338e-01 -2.05338933e-02
-1.09148338e-01 -1.23698413e-01 -4.86737907e-01 1.90554455e-01
1.27435637e+00 6.27310872e-02 3.81844044e-01 7.88206995e-01
4.36758190e-01 -4.92013767e-02 -1.30052543e+00 9.48014736e-01
3.03752303e-01 -1.86327779e+00 2.71654814e-01 -1.90396696e-01
4.18959439e-01 5.46975613e-01 4.14242506e-01 6.32479131e-01
3.88362676e-01 -1.31310976e+00 1.09048164e+00 8.54027569e-01
4.29377198e-01 -4.33968991e-01 4.02796805e-01 2.38369569e-01
-1.37320483e+00 -6.94213390e-01 -2.97314376e-01 -2.47746319e-01
3.25105578e-01 -1.51827350e-01 -9.17678118e-01 6.57415912e-02
7.24233508e-01 1.39641500e+00 -6.17848992e-01 7.56596208e-01
-5.86307421e-02 5.85498929e-01 2.08340734e-02 -5.79484887e-02
6.20625019e-01 -5.55315763e-02 6.42184019e-01 1.37717974e+00
3.15190434e-01 -2.99469884e-02 -1.31098332e-03 5.65087795e-01
9.80092064e-02 -3.36074978e-01 -9.94980752e-01 -2.82080442e-01
-2.24313229e-01 9.24752474e-01 -7.02795267e-01 -5.15283406e-01
-2.84780204e-01 1.20632064e+00 5.01913369e-01 2.67100275e-01
-9.19947863e-01 5.73139012e-01 4.32746589e-01 5.02051413e-02
8.74233425e-01 -2.73181021e-01 -8.78652632e-02 -1.13193107e+00
-2.29676798e-01 -8.40498865e-01 3.65467310e-01 -1.32907951e+00
-9.20129061e-01 9.82416451e-01 -2.56267309e-01 -1.16570675e+00
-1.08039629e+00 -9.07898128e-01 -7.92400658e-01 1.72718525e-01
-1.03005254e+00 -1.70304871e+00 1.64025560e-01 7.04659283e-01
1.22579670e+00 -6.26662195e-01 7.77756870e-01 -2.31208146e-01
-3.72599959e-01 -1.04409298e-02 -1.54750213e-01 -9.23813730e-02
2.02914432e-01 -1.14092481e+00 7.23803997e-01 9.24571276e-01
7.39108562e-01 2.32237086e-01 1.14695764e+00 -3.74807745e-01
-1.69701731e+00 -8.33760202e-01 9.39738452e-01 -4.81810719e-01
1.17666852e+00 -2.50494391e-01 -5.61929107e-01 1.20009327e+00
7.63417780e-01 -3.61973457e-02 1.02576345e-01 -6.07286766e-02
-5.08743465e-01 -4.74492759e-02 -3.38880211e-01 8.32046390e-01
1.01207137e+00 -7.76078403e-01 -6.91175878e-01 2.96365976e-01
8.75118434e-01 3.29193585e-02 -5.38277805e-01 3.33393455e-01
5.72902083e-01 -1.35280371e+00 1.09031761e+00 -8.46947491e-01
7.48164952e-01 -3.93844731e-02 2.03190316e-02 -1.12728095e+00
-5.85004687e-01 -9.45973158e-01 -4.29008275e-01 8.54569674e-01
2.74000764e-01 -2.04465345e-01 1.00838029e+00 2.87858367e-01
-4.94579226e-01 -6.78504765e-01 -9.54903066e-01 -5.21977246e-01
-3.44850868e-01 -6.80550396e-01 -1.96296319e-01 5.32236874e-01
2.05438539e-01 6.87583804e-01 -9.25270677e-01 -5.70576787e-01
2.00815514e-01 -3.60042905e-03 3.46744061e-01 -6.92121267e-01
-4.34411407e-01 -8.01088750e-01 -5.84803641e-01 -1.07604384e+00
3.63968939e-01 -5.91171145e-01 -1.02376550e-01 -1.42677152e+00
3.24877203e-01 1.79246709e-01 -2.00700298e-01 4.50781882e-01
5.15784740e-01 1.82051986e-01 5.75053692e-01 1.65910527e-01
-1.04110134e+00 8.83063495e-01 8.30969810e-01 -2.15101346e-01
3.85452583e-02 -5.54501079e-02 -1.67552948e-01 8.45862865e-01
7.82426357e-01 -2.57327091e-02 -4.51474339e-01 -6.22630298e-01
5.06278455e-01 5.14210582e-01 6.95195377e-01 -1.22271538e+00
3.87630403e-01 -9.78013501e-02 5.44488430e-01 -4.29200679e-01
7.51772046e-01 -8.11991751e-01 9.00127813e-02 7.98702717e-01
-6.98616505e-01 5.14291763e-01 1.85676605e-01 5.21181822e-01
-2.00260028e-01 2.19997302e-01 4.45883393e-01 -4.83405262e-01
-1.39212704e+00 2.23406270e-01 -1.02494597e+00 -3.46274137e-01
9.48148847e-01 -2.38273904e-01 -3.95723343e-01 -5.60244560e-01
-8.98028851e-01 -2.53767762e-02 2.87922472e-01 5.87655783e-01
9.98640239e-01 -1.15513992e+00 -7.96813548e-01 8.76356587e-02
-4.01095906e-03 -8.97304952e-01 3.27175289e-01 1.24028170e+00
-3.50283027e-01 8.95635307e-01 -6.89257205e-01 -5.93015611e-01
-1.31781626e+00 1.06493282e+00 3.13456982e-01 -4.81792063e-01
-9.59905624e-01 1.06479108e+00 2.99919695e-01 4.35217917e-01
2.01699108e-01 -4.06719923e-01 -5.20636082e-01 1.22862302e-01
5.34700811e-01 1.53017440e-03 -4.29291576e-01 -6.38912320e-01
-3.00146341e-01 2.27888063e-01 -8.26678723e-02 -1.04009517e-01
1.70631528e+00 -8.92699733e-02 -2.02448159e-01 8.01667869e-01
7.75448143e-01 -6.58064663e-01 -1.68007660e+00 6.56909309e-03
4.42465931e-01 1.03010438e-01 -1.53447911e-01 -5.02825558e-01
-9.67430532e-01 1.00633717e+00 3.83224696e-01 2.99688697e-01
1.10599804e+00 2.63013303e-01 4.37362075e-01 5.87866306e-01
1.98525861e-01 -1.02032757e+00 4.35904443e-01 9.12955105e-01
1.14811063e+00 -1.07545352e+00 -2.37756804e-01 3.64706039e-01
-1.20205808e+00 1.28764057e+00 5.11190712e-01 -5.71243167e-01
1.56145856e-01 3.33355099e-01 -3.63669872e-01 -1.65550262e-01
-2.02902699e+00 -3.96832258e-01 3.36259365e-01 5.73264062e-01
6.64225399e-01 -1.47908881e-01 3.01193595e-01 2.45804563e-01
1.33504316e-01 1.41753554e-01 2.07600072e-01 4.96572107e-01
-3.27359438e-01 -8.22740078e-01 8.31539482e-02 2.82488257e-01
-2.65933603e-01 -4.84952211e-01 -2.02795580e-01 5.07883310e-01
-1.68299228e-01 6.10315084e-01 2.33203411e-01 -6.72620296e-01
-1.83307365e-01 1.05643488e-01 7.39941955e-01 -2.87151653e-02
-9.24506962e-01 6.14417605e-02 1.84775680e-01 -8.44034314e-01
-6.72093391e-01 -8.31546068e-01 -8.51884425e-01 -1.97592884e-01
4.17546511e-01 -1.03773586e-01 2.40678534e-01 9.80409145e-01
2.75460333e-01 7.13408291e-01 2.79521763e-01 -1.33203828e+00
-3.45191300e-01 -1.00357509e+00 -2.65279872e-04 5.09859920e-02
4.75777566e-01 -2.08983198e-01 2.55524725e-01 4.66956079e-01]
|
[8.649149894714355, 0.4187415540218353]
|
e9008755-1044-4bcc-b373-3b65c09cfba7
|
an-energy-management-system-model-with-power
|
2212.01910
| null |
https://arxiv.org/abs/2212.01910v1
|
https://arxiv.org/pdf/2212.01910v1.pdf
|
An energy management system model with power quality constraints for unbalanced multi-microgrids interacting in a local energy market
|
As multi-microgrids become readily available, some limited models have been proposed that study operational and power quality constraints with local energy markets independently. This paper proposes a convex optimization model of an energy management system with operational and power quality constraints and interactions in a Local Energy Market (LEM) for unbalanced microgrids (MGs). The LEM consists of a pre-dispatch step and an energy transactions step (ETS). The ETS combines the MGs' objectives while considering two strategies: minimize the cost of buyers or maximize the revenue of sellers. Our proposed model considers harmonic distortion and voltage limit power quality constraints in both steps. Moreover, we model operational constraints such as power flow, power balance, and distributed energy resources behaviors and capacities. We numerically evaluate the proposed model using three unbalanced MGs with residential, industrial, and commercial load profiles, where each microgrid manages its resources locally. Furthermore, we create two groups of cases to analyze the interactions in the local energy market. In the first group, the price of the DSO energy and the surplus from MGs to DSO are the same. The numerical results show that using the increasing revenue strategy promotes MGs to interact more while encouraging them to have high energy prices. When the reducing cost strategy is used, fewer energy interactions occur, and the price of MGs energy is encouraged to be lower.
|
['Diego Patino', 'César A. Uribe', 'Gabriel Ordóñez-Plata', 'Alejandro Garcés', 'Carlos Adrian Correa-Florez', 'Johanna Castellanos']
|
2022-12-04
| null | null | null | null |
['energy-management']
|
['time-series']
|
[-6.10754669e-01 6.12333827e-02 -2.60668635e-01 1.92749277e-01
-8.39938894e-02 -8.55993450e-01 2.27001473e-01 1.99707136e-01
1.24781281e-01 9.68610346e-01 -2.21398398e-01 -6.02253415e-02
-3.60904366e-01 -1.06294966e+00 -1.86866403e-01 -1.17731464e+00
-4.22261329e-03 2.08234221e-01 -5.52505404e-02 -1.77750081e-01
-5.39045408e-02 4.81276572e-01 -1.01774085e+00 -6.44158661e-01
1.11173356e+00 1.19106138e+00 3.53965938e-01 5.64079545e-02
3.49686742e-01 2.85537511e-01 -8.99676442e-01 1.91092074e-01
7.07312286e-01 -6.17816627e-01 -9.35331956e-02 1.38080180e-01
-1.11782217e+00 -6.19889200e-01 2.21456215e-01 1.26418555e+00
6.65679514e-01 2.73680478e-01 4.51329350e-01 -2.18642330e+00
-5.16107380e-01 7.01498151e-01 -7.82421112e-01 1.22059785e-01
-2.68973429e-02 1.03828631e-01 1.09196222e+00 -1.83177665e-01
1.64790779e-01 5.77784002e-01 -2.34311074e-01 1.88256189e-01
-9.68528152e-01 -6.66600943e-01 1.80742949e-01 3.32184136e-01
-1.34674037e+00 7.56406412e-02 8.28530967e-01 -5.26592433e-02
1.26013470e+00 4.42632914e-01 1.15838575e+00 -1.83781013e-01
4.34971064e-01 6.02109671e-01 8.71076345e-01 -3.70324016e-01
7.87984252e-01 2.70374775e-01 -2.51070783e-02 -3.72849584e-01
6.74966753e-01 -2.19189003e-01 -7.19521269e-02 -1.89138398e-01
1.99824989e-01 -1.45685926e-01 -5.36269128e-01 -5.66461146e-01
-6.03646517e-01 6.30093634e-01 3.35887223e-01 4.97403115e-01
-5.18594384e-01 -2.50617027e-01 3.43984775e-02 2.82214820e-01
3.22752118e-01 6.11457415e-02 -1.93405226e-01 1.70944095e-01
-7.70092010e-01 -3.14008258e-02 8.65506947e-01 1.12461400e+00
4.17876750e-01 3.98861736e-01 2.87564695e-01 5.16490638e-01
3.32275361e-01 5.41208863e-01 4.94851321e-01 -6.19315803e-01
9.16812941e-02 7.19458163e-01 6.55045331e-01 -6.23859942e-01
-3.20752412e-01 -6.61403537e-01 -7.28551626e-01 2.63486356e-01
6.55444041e-02 -4.35564399e-01 -6.01633918e-03 1.56004667e+00
1.61691934e-01 -4.35868919e-01 1.28205083e-02 9.20339227e-01
-1.40843214e-02 1.12634051e+00 -1.93375796e-01 -1.14459968e+00
1.18419528e+00 -7.14856684e-01 -1.07170987e+00 1.94027171e-01
3.98945093e-01 -5.66337645e-01 1.91196471e-01 9.11609456e-02
-1.79919279e+00 1.05246596e-01 -1.28503525e+00 4.23173994e-01
-5.42426348e-01 2.05997020e-01 -1.98836580e-01 4.52883750e-01
-1.21048963e+00 2.40774497e-01 -8.32173228e-01 -2.52234012e-01
-2.30946034e-01 4.52225059e-01 4.30228293e-01 6.33372903e-01
-1.06591916e+00 9.95700896e-01 3.75067860e-01 9.90533456e-02
-3.16943377e-01 -6.39110625e-01 -5.98877966e-01 6.43135548e-01
4.35471207e-01 -6.42651558e-01 1.18596792e+00 -6.88183129e-01
-1.44141233e+00 1.92080215e-01 2.38828510e-01 -3.93624812e-01
4.29603755e-01 3.76796305e-01 -4.57349300e-01 1.10066906e-01
1.21811219e-01 3.36934738e-02 -1.09686226e-01 -1.17526591e+00
-9.66475666e-01 -3.18546236e-01 4.27810326e-02 6.96847141e-01
-3.34447384e-01 -1.34050980e-01 3.38950992e-01 -4.15782839e-01
-2.96390235e-01 -6.25509381e-01 -2.08146006e-01 -5.80888033e-01
-3.73849034e-01 -5.75346231e-01 8.47984195e-01 -4.35578793e-01
1.19471693e+00 -1.86662555e+00 3.64727229e-01 4.29392576e-01
-2.60736465e-01 -2.63921797e-01 3.62206012e-01 9.02307749e-01
-2.25390177e-02 1.31901741e-01 8.85983631e-02 -1.05018094e-01
5.17736316e-01 3.79496276e-01 1.64146513e-01 2.72811502e-01
-2.38890767e-01 5.37602007e-01 -7.65267611e-01 1.73536018e-01
3.33936542e-01 -2.14726269e-01 -2.60955691e-01 2.06678540e-01
-2.99640894e-02 -1.11927480e-01 -5.63603103e-01 6.05071366e-01
7.77267039e-01 -1.75287887e-01 6.42995477e-01 -3.42927039e-01
-5.74779093e-01 -9.84517932e-02 -1.64924788e+00 8.28600407e-01
-5.77656507e-01 1.64622560e-01 7.66209483e-01 -1.19767845e+00
4.15936649e-01 2.25863069e-01 9.54308927e-01 -7.62466967e-01
2.03988135e-01 4.07905698e-01 1.05033085e-01 -2.37504356e-02
3.80626470e-01 -9.57855303e-03 9.65706706e-02 6.35379553e-01
1.04418911e-01 -1.14079170e-01 7.58362055e-01 2.38878518e-01
4.59507585e-01 -1.87048391e-01 6.54438019e-01 -9.60424960e-01
5.67348421e-01 -3.36587667e-01 9.52481031e-01 -1.78999931e-01
-5.71226440e-02 -3.30319256e-01 4.26398724e-01 3.98732483e-01
-8.37561667e-01 -1.09542513e+00 7.22729266e-02 5.59018493e-01
7.91343033e-01 5.75850494e-02 -6.12628460e-01 -1.66080192e-01
1.23458989e-01 1.20999825e+00 -1.57866068e-02 -1.24754831e-02
-2.85737872e-01 -1.29111564e+00 -7.46755004e-01 3.21770608e-01
4.82781470e-01 -4.58254576e-01 -9.29950237e-01 1.24055654e-01
9.75814536e-02 -7.61236787e-01 -7.26688743e-01 3.55841041e-01
-2.09339425e-01 -9.25337195e-01 -7.15712249e-01 -6.71895981e-01
1.01400995e+00 9.85360220e-02 7.50162661e-01 -1.68435693e-01
-4.63011526e-02 2.81715065e-01 -1.03635564e-01 -3.94330114e-01
-2.42251419e-02 7.07365125e-02 2.72622973e-01 -1.44477054e-01
4.99390848e-02 -7.14125514e-01 -8.91969383e-01 6.56194031e-01
-8.52789402e-01 1.41269550e-01 2.00882554e-01 4.79317218e-01
5.07410049e-01 8.46414506e-01 1.38455331e+00 1.26502231e-01
7.99949110e-01 -8.57740164e-01 -1.17897964e+00 3.53668839e-01
-9.38473225e-01 -3.29533964e-01 1.00451708e+00 -9.79866385e-02
-8.29933643e-01 -2.81827599e-01 3.76844108e-01 -1.26670953e-02
2.35752076e-01 1.09638363e-01 -8.47958505e-01 9.53150317e-02
-6.43544674e-01 3.61043721e-01 -5.99569678e-02 -4.33857441e-01
-1.29432343e-02 6.68316782e-01 2.46002972e-01 -3.93864900e-01
1.17372000e+00 7.68120065e-02 1.08989798e-01 -3.22240770e-01
2.48716146e-01 -2.57261693e-01 -6.29726499e-02 -5.12676060e-01
5.89558542e-01 -8.63978684e-01 -1.43196857e+00 5.15471876e-01
-5.25680602e-01 -1.00298926e-01 -4.32036459e-01 4.36384320e-01
-4.54670250e-01 2.83210039e-01 -2.94266343e-01 -1.08122313e+00
-3.46210569e-01 -1.24507785e+00 2.91821152e-01 9.27420974e-01
2.19276041e-01 -9.38751757e-01 -3.26640606e-01 -2.29315348e-02
5.81315756e-01 5.43622315e-01 9.33080077e-01 -3.60820144e-01
-9.40653205e-01 3.86845559e-01 1.59970269e-01 4.02490616e-01
6.28596008e-01 -1.14553533e-01 -6.51914999e-02 -9.63149786e-01
1.98234484e-01 2.13303685e-01 -3.90574813e-01 3.73450071e-01
5.62445939e-01 -6.74477100e-01 -4.27230060e-01 1.25974804e-01
2.26428676e+00 9.93545413e-01 -1.06941983e-02 4.75480914e-01
-1.79160208e-01 6.05519414e-01 5.25409937e-01 6.90607667e-01
7.63837516e-01 7.43080616e-01 7.32461274e-01 -3.12114917e-02
6.57290697e-01 2.87826747e-01 4.37352717e-01 9.31385994e-01
-8.61277524e-03 -8.08383465e-01 -2.77713537e-01 8.53001237e-01
-1.89714324e+00 -5.62316656e-01 2.12052763e-01 2.08683777e+00
4.98683304e-01 -5.95338978e-02 3.89042526e-01 2.24051163e-01
7.50005782e-01 -1.71901926e-01 -8.00655961e-01 -4.64364380e-01
-2.31979042e-01 -3.16998690e-01 8.12720537e-01 2.71378040e-01
-7.88128898e-02 -3.07135224e-01 5.00141001e+00 7.79010832e-01
-7.47445345e-01 7.94830322e-02 8.51139665e-01 -8.30439210e-01
-5.00768900e-01 -9.48645994e-02 -5.73801398e-01 1.13769197e+00
9.00152743e-01 -1.31857491e+00 7.74511397e-01 5.34592092e-01
9.35455322e-01 -7.44131386e-01 -8.97928655e-01 5.34053206e-01
-3.92946452e-01 -9.94547844e-01 -3.67961049e-01 4.47962582e-01
9.30764079e-01 -1.61784381e-01 -4.71742630e-01 -1.62105292e-01
4.02881533e-01 -2.81033486e-01 9.57946777e-01 3.41191024e-01
-8.58787447e-02 -1.41849470e+00 8.57518256e-01 5.02457440e-01
-1.53358519e+00 -4.91988033e-01 1.10976547e-01 1.21228419e-01
8.90995026e-01 5.72908401e-01 5.37691377e-02 1.31374633e+00
7.49494672e-01 2.06616387e-01 6.98039010e-02 7.71977007e-01
-4.07390147e-02 5.00592813e-02 -6.24228954e-01 -2.62649715e-01
-7.63258291e-03 -1.12039256e+00 4.19769078e-01 2.91596293e-01
4.51293021e-01 4.53760266e-01 3.97810072e-01 1.05222476e+00
-9.30400044e-02 1.88545808e-01 -1.28803596e-01 2.29328796e-01
7.39482343e-01 1.45429075e+00 -8.19564581e-01 -2.12548107e-01
-5.92852712e-01 5.12912750e-01 -5.79501092e-01 5.21602571e-01
-7.14699090e-01 -4.62252766e-01 7.67733872e-01 6.67102188e-02
-1.67405635e-01 -2.03414690e-02 -5.08528352e-01 -9.27423358e-01
1.90854594e-01 -2.07175508e-01 3.76004726e-01 -8.21930945e-01
-1.25900733e+00 -7.27751851e-02 3.91390264e-01 -1.12067389e+00
-6.84964180e-01 -1.86430156e-01 -1.05164492e+00 1.07833838e+00
-1.89578402e+00 -6.71691179e-01 1.37051404e-01 3.25241864e-01
5.32976806e-01 3.74676920e-02 1.15415528e-01 5.79081178e-01
-1.02907479e+00 2.08450273e-01 7.84495771e-01 -2.37634644e-01
-2.38329500e-01 -1.41521513e+00 -6.90444052e-01 1.02245891e+00
-8.15909863e-01 2.20985353e-01 7.01052368e-01 -4.06823546e-01
-1.45105791e+00 -5.76208234e-01 6.13451064e-01 7.34991610e-01
8.01914692e-01 -2.92583376e-01 -3.70133907e-01 3.34310055e-01
1.23976958e+00 -4.78237391e-01 5.38979769e-01 -9.02786374e-01
8.38001251e-01 -3.35971415e-01 -1.63881040e+00 5.39162099e-01
4.83659327e-01 4.64638062e-02 -1.34530872e-01 2.54027992e-01
1.48771137e-01 -4.68308181e-02 -1.19515920e+00 3.66663456e-01
2.23189294e-01 -3.60942364e-01 5.91231048e-01 2.32051924e-01
-3.38109493e-01 -6.73893332e-01 -2.27821931e-01 -1.96168685e+00
-6.41365871e-02 -8.71554136e-01 -1.49651319e-01 1.56855977e+00
1.55531153e-01 -1.09391308e+00 3.10738176e-01 6.13107562e-01
4.94466648e-02 -7.27454007e-01 -1.31964159e+00 -1.01039469e+00
2.00419605e-01 6.78493083e-01 9.95101035e-01 8.63664627e-01
7.23215640e-01 9.58456993e-02 1.41568050e-01 5.06001651e-01
8.61887217e-01 4.54201639e-01 6.63524568e-02 -6.76735222e-01
-1.98646132e-02 -6.99292600e-01 1.54243320e-01 -5.64158082e-01
-1.32336840e-01 -5.33762038e-01 -4.47702646e-01 -1.87743163e+00
4.51515205e-02 5.94492927e-02 -3.84625286e-01 4.53562319e-01
3.61890793e-01 -1.01253666e-01 5.85875571e-01 2.42826149e-01
4.04869467e-02 7.38434911e-01 6.95657730e-01 -1.37986436e-01
-2.90826946e-01 1.91098243e-01 -3.31012815e-01 5.22025108e-01
1.05161285e+00 1.76320016e-01 -7.66558588e-01 2.60039661e-02
9.52082053e-02 2.78688848e-01 -2.59546697e-01 -6.34072185e-01
2.99641699e-01 -4.38278764e-01 -9.75204706e-02 -9.44062412e-01
-7.44668171e-02 -1.40142202e+00 7.98868358e-01 6.73254550e-01
2.92914718e-01 4.02783602e-01 -1.88954070e-01 1.70293525e-01
-1.18228368e-01 -1.78768918e-01 8.02491367e-01 4.75119762e-02
-2.48362854e-01 -1.06532425e-01 -7.46636569e-01 -5.74044764e-01
1.83225739e+00 -1.81283772e-01 -5.06922245e-01 -4.93357748e-01
-9.39773977e-01 1.33107817e+00 6.53605938e-01 5.61906278e-01
-1.56807080e-01 -1.24780679e+00 -3.05462122e-01 -2.24033028e-01
-4.35290664e-01 -1.50853291e-01 1.31742433e-01 6.54299080e-01
-1.27472669e-01 5.03616691e-01 -3.55136901e-01 -2.84375250e-01
-8.73359263e-01 3.75432670e-01 8.47564280e-01 -3.81668687e-01
1.49658350e-02 -1.51045725e-01 4.34869416e-02 8.19264576e-02
-1.60410389e-01 -5.90044320e-01 -7.07761198e-02 6.68063343e-01
-1.05989818e-02 1.00766969e+00 8.09895173e-02 -5.21838129e-01
-3.15532476e-01 3.25590402e-01 6.08823478e-01 2.17422768e-01
1.39987314e+00 -7.44555354e-01 -2.29918793e-01 1.21542491e-01
9.08322096e-01 -7.03035807e-03 -8.28990459e-01 2.67876387e-01
-1.66266531e-01 -9.41723883e-02 1.20465860e-01 -9.89896476e-01
-1.54572201e+00 -1.28447607e-01 4.38896179e-01 1.09971929e+00
1.86270905e+00 -3.17064971e-01 5.35337150e-01 -4.19796854e-01
6.97706819e-01 -1.60797417e+00 -4.40556198e-01 -2.17143908e-01
5.77690661e-01 -4.64923203e-01 -1.26092479e-01 -3.49690288e-01
-1.39097586e-01 7.56116271e-01 5.82384229e-01 -5.64850718e-02
8.32837641e-01 5.33290207e-01 -5.85454226e-01 3.38359058e-01
-6.91934884e-01 7.78643489e-02 -3.35332870e-01 2.19513819e-01
-1.17476083e-01 3.71097773e-01 -1.22033608e+00 7.65841901e-01
-2.78387591e-02 -9.42318738e-02 1.09615266e+00 1.00077796e+00
-1.89010248e-01 -1.07483757e+00 -3.39225471e-01 3.67050231e-01
-4.70880836e-01 3.42043251e-01 2.12210581e-01 7.05855191e-01
4.09853369e-01 1.18750131e+00 4.03527021e-01 5.11236608e-01
6.29913867e-01 -1.64546683e-01 -2.78676134e-02 -1.58133730e-01
-4.75214988e-01 3.83830547e-01 -6.65891841e-02 -7.30140731e-02
-4.62845236e-01 -7.12961793e-01 -1.47062814e+00 -4.14572418e-01
-7.40713239e-01 8.24959874e-01 8.67555857e-01 8.12310755e-01
2.10600391e-01 6.53060138e-01 1.41944385e+00 -5.86020708e-01
-6.26643479e-01 -6.11451447e-01 -1.33505344e+00 -1.55188948e-01
-3.16741341e-03 -3.86429280e-01 -8.35402668e-01 -2.66124517e-01]
|
[5.643560409545898, 2.530181884765625]
|
8bc2241f-1661-4c3c-99e8-8a2a0c64799f
|
owl-observe-watch-listen-localizing-actions
|
2202.04947
| null |
https://arxiv.org/abs/2202.04947v3
|
https://arxiv.org/pdf/2202.04947v3.pdf
|
OWL (Observe, Watch, Listen): Audiovisual Temporal Context for Localizing Actions in Egocentric Videos
|
Egocentric videos capture sequences of human activities from a first-person perspective and can provide rich multimodal signals. However, most current localization methods use third-person videos and only incorporate visual information. In this work, we take a deep look into the effectiveness of audiovisual context in detecting actions in egocentric videos and introduce a simple-yet-effective approach via Observing, Watching, and Listening (OWL). OWL leverages audiovisual information and context for egocentric temporal action localization (TAL). We validate our approach in two large-scale datasets, EPIC-Kitchens, and HOMAGE. Extensive experiments demonstrate the relevance of the audiovisual temporal context. Namely, we boost the localization performance (mAP) over visual-only models by +2.23% and +3.35% in the above datasets.
|
['Bernard Ghanem', 'Chen Zhao', 'Fabian Caba Heilbron', 'Victor Escorcia', 'Merey Ramazanova']
|
2022-02-10
| null | null | null | null |
['action-localization']
|
['computer-vision']
|
[-5.31795919e-02 -4.44554627e-01 -3.06925565e-01 -2.56974310e-01
-9.53650057e-01 -6.64291084e-01 7.89430559e-01 -1.74174190e-01
-5.82355559e-01 5.59732854e-01 8.81753325e-01 5.49070060e-01
2.80978262e-01 -2.46715873e-01 -7.78705478e-01 -4.82372522e-01
-3.88391525e-01 -3.89019847e-01 3.21639508e-01 8.90819952e-02
6.51085377e-02 1.10979974e-01 -1.61978698e+00 6.28051460e-01
2.34827563e-01 9.10064340e-01 -2.06644777e-02 9.30137455e-01
4.16366011e-01 1.19844437e+00 -4.35138643e-01 -1.97702691e-01
5.02132401e-02 -4.61353958e-01 -5.37893176e-01 1.70280356e-02
7.27779567e-01 -6.96482599e-01 -7.76939094e-01 7.12086856e-01
7.69204140e-01 3.91537964e-01 3.93271059e-01 -1.35690856e+00
-2.75731653e-01 5.35474837e-01 -6.62430108e-01 5.18537760e-01
1.26626205e+00 3.80695999e-01 9.10134673e-01 -9.51759040e-01
7.76536226e-01 1.29520857e+00 6.98373556e-01 2.85915732e-01
-1.00387752e+00 -6.25843644e-01 2.65562922e-01 5.68082452e-01
-1.34576488e+00 -8.92090976e-01 8.21435928e-01 -5.18341422e-01
9.42409635e-01 1.14546455e-01 8.67855370e-01 1.82738113e+00
6.02826141e-02 1.10899150e+00 8.47887337e-01 -2.02143699e-01
1.18157938e-01 -1.46021336e-01 -3.68348926e-01 5.51625371e-01
-1.95104420e-01 7.46401027e-02 -1.29819393e+00 -4.92792465e-02
7.98114181e-01 1.68081582e-01 -2.88344026e-01 -3.39508146e-01
-1.70900786e+00 5.21956265e-01 3.66805464e-01 2.41123706e-01
-5.17051101e-01 6.69685721e-01 7.22015858e-01 8.66563898e-03
3.06273043e-01 2.56989181e-01 4.50677983e-02 -7.31011569e-01
-7.49349594e-01 1.22576877e-01 1.75900161e-01 1.02357578e+00
5.34764409e-01 1.04303166e-01 -5.35197318e-01 5.82204819e-01
1.05060533e-01 6.94907248e-01 3.76313508e-01 -1.31001151e+00
4.24141377e-01 2.18913600e-01 2.15854973e-01 -1.17594004e+00
-4.11840379e-01 -1.61772162e-01 -6.12727925e-02 -4.16651577e-01
4.99053001e-01 -1.55576408e-01 -4.35101330e-01 1.96635783e+00
4.49371606e-01 4.70359236e-01 -2.62837321e-01 1.22754276e+00
8.05257976e-01 3.25277716e-01 4.03580368e-01 -9.95352194e-02
1.54345191e+00 -8.66911352e-01 -7.94817924e-01 -2.38433972e-01
5.84350348e-01 -5.58650017e-01 1.11875892e+00 2.68389463e-01
-9.70104814e-01 -6.29394829e-01 -7.33990490e-01 6.05018698e-02
-1.04396358e-01 2.68061846e-01 6.74711049e-01 2.76760966e-01
-7.53260434e-01 1.64274693e-01 -8.94159138e-01 -8.00787032e-01
1.75372958e-01 -1.00174226e-01 -8.00550044e-01 -1.19046025e-01
-1.10891211e+00 4.37645793e-01 2.14781642e-01 -1.60610884e-01
-1.44415057e+00 -4.54136491e-01 -1.08522332e+00 -1.41047522e-01
5.38999557e-01 -5.67444742e-01 1.30318868e+00 -9.89432096e-01
-1.18719172e+00 5.72641015e-01 -4.12675798e-01 -4.57596004e-01
6.13996923e-01 -6.42530739e-01 -4.96200413e-01 1.05441320e+00
3.01329106e-01 7.75772572e-01 6.51020944e-01 -1.02344346e+00
-6.90207839e-01 -3.21642399e-01 3.74383867e-01 2.62389988e-01
-4.00289774e-01 1.75190836e-01 -6.38012946e-01 -7.60258973e-01
-5.39680608e-02 -7.59986401e-01 3.46783251e-01 1.19832575e-01
-3.18244882e-02 -1.59874628e-03 6.68388903e-01 -7.62900174e-01
1.19494534e+00 -2.37960458e+00 9.18702409e-02 -2.02898249e-01
9.73562226e-02 -3.91503751e-01 -1.72586769e-01 7.07362831e-01
1.25830874e-01 -2.79341877e-01 4.19805080e-01 -5.36543190e-01
1.54009834e-01 -6.88702390e-02 -1.91820577e-01 7.30839908e-01
-6.00239411e-02 1.08890963e+00 -1.31003714e+00 -7.60566652e-01
3.80467594e-01 7.15368509e-01 -7.93975651e-01 1.06385656e-01
1.26883537e-01 5.97327173e-01 -1.84604570e-01 1.12602532e+00
1.23999216e-01 8.64346884e-03 1.99729368e-01 -3.56129527e-01
-5.32922968e-02 2.10821219e-02 -8.74650776e-01 2.23185849e+00
-1.92415163e-01 1.07771719e+00 -1.86402932e-01 -5.22217333e-01
3.17874998e-01 4.58662659e-01 6.80628240e-01 -9.52362955e-01
1.07445635e-01 -2.17792884e-01 -3.81624520e-01 -9.19477522e-01
5.56265652e-01 5.09787351e-02 -2.94396698e-01 2.86717743e-01
3.99599105e-01 7.33120441e-01 1.44968808e-01 4.47938859e-01
1.23218048e+00 5.12480140e-01 3.80331844e-01 1.91830695e-01
3.94641638e-01 -1.72838077e-01 5.11361897e-01 8.36553276e-01
-8.49285901e-01 5.37548959e-01 5.44950604e-01 -1.36635348e-01
-5.72957397e-01 -1.09977257e+00 3.62630874e-01 1.65547693e+00
2.86261499e-01 -8.60278606e-01 -6.18602455e-01 -7.78219044e-01
-7.40973726e-02 5.72171748e-01 -8.26103866e-01 -1.76972017e-01
-5.75646877e-01 -1.04058273e-02 8.32950175e-01 8.57537687e-01
5.39863586e-01 -9.92032707e-01 -8.08365405e-01 -2.14362457e-01
-8.11195731e-01 -1.62171030e+00 -5.77640057e-01 -4.29363519e-01
-6.19780362e-01 -1.09204805e+00 -7.17368901e-01 -2.83474416e-01
2.55436808e-01 7.57958472e-01 7.24361897e-01 -5.47470748e-01
-2.97582388e-01 1.25218379e+00 -4.97737229e-01 5.70193939e-02
4.49625582e-01 -3.04648072e-01 4.71612394e-01 3.41272980e-01
5.94277918e-01 -6.50908113e-01 -7.78333187e-01 4.90993917e-01
-3.02351296e-01 -1.57063603e-01 3.87309879e-01 5.64041853e-01
3.92747670e-01 -4.65944529e-01 4.19031858e-01 -2.66064256e-01
8.40601772e-02 -6.61904633e-01 -6.67724758e-02 -6.71056807e-02
1.97159186e-01 -5.23331583e-01 2.63752043e-01 -6.27979517e-01
-1.09044552e+00 2.71702856e-01 3.23423833e-01 -8.18196833e-01
-4.16535884e-01 3.53458077e-01 -7.52059892e-02 -7.92150013e-03
6.56804919e-01 2.16023326e-01 -4.97807503e-01 -2.39311591e-01
4.14821863e-01 4.32484359e-01 9.42313313e-01 -4.96423960e-01
4.21344906e-01 9.64083016e-01 -3.18030804e-01 -9.49670255e-01
-5.23489714e-01 -8.02470922e-01 -6.54516816e-01 -8.80008936e-01
1.02818763e+00 -1.52610576e+00 -1.03750753e+00 1.20134413e-01
-8.56753349e-01 -2.24316105e-01 -6.73287883e-02 9.64573205e-01
-8.40041757e-01 7.04887986e-01 -4.30523247e-01 -9.68456209e-01
1.86116397e-01 -7.00840414e-01 1.28660321e+00 3.23886015e-02
-3.90617549e-01 -6.46847844e-01 6.88445121e-02 5.68975806e-01
9.22545567e-02 5.85195899e-01 -5.68998307e-02 -2.30850548e-01
-6.13695145e-01 -3.76844555e-01 -1.40146270e-01 -2.20903099e-01
-5.79693355e-02 -2.94232756e-01 -1.39220488e+00 -2.19094172e-01
-3.14119697e-01 -4.02539253e-01 9.03590560e-01 3.48011255e-01
8.45879257e-01 -2.00937152e-01 -4.32186276e-01 6.20171785e-01
1.03092623e+00 2.41255537e-02 5.36487699e-01 1.39799893e-01
6.76724970e-01 6.32822454e-01 1.01060796e+00 7.79976428e-01
4.08076465e-01 8.47658336e-01 4.96243954e-01 3.79943252e-01
-1.45948738e-01 -6.90778315e-01 8.22586596e-01 4.61771965e-01
-3.45073342e-01 -2.00916663e-01 -7.61910796e-01 8.12285244e-01
-2.10130072e+00 -1.54569125e+00 2.21516177e-01 1.93132794e+00
3.95566493e-01 -1.27708107e-01 4.84932959e-01 -8.14493820e-02
6.25975251e-01 3.72778982e-01 -2.92833418e-01 2.38036484e-01
-1.26112878e-01 -5.37473917e-01 2.96577096e-01 1.45936579e-01
-1.47388029e+00 9.54456210e-01 6.58266735e+00 5.36566079e-01
-8.91781449e-01 2.03292206e-01 -2.01999873e-01 -8.29106450e-01
1.14739627e-01 -1.44766912e-01 -3.99377108e-01 5.67497313e-01
8.04832160e-01 2.01873556e-01 3.79023403e-01 9.05599177e-01
5.26804149e-01 -4.55754340e-01 -1.35843098e+00 1.38137889e+00
5.00081003e-01 -9.38903928e-01 -2.27795035e-01 -1.06093876e-01
4.76540715e-01 6.40468821e-02 -1.82940662e-02 3.96533281e-01
-2.08632454e-01 -7.68757284e-01 9.53826725e-01 7.23244905e-01
7.37112641e-01 -7.68541276e-01 4.60692018e-01 1.33019000e-01
-1.47886086e+00 -3.47042173e-01 -1.10043213e-01 -1.71322197e-01
5.12561083e-01 -6.19932823e-02 -5.47267795e-01 2.87148088e-01
1.16618764e+00 1.17503667e+00 -6.97194755e-01 1.00526929e+00
-3.73438507e-01 5.64902246e-01 -2.30757564e-01 1.50577635e-01
2.29906976e-01 3.26244503e-01 8.01281691e-01 1.54274368e+00
2.24209175e-01 2.61436462e-01 1.94499671e-01 4.23546284e-01
1.56067595e-01 -3.65583748e-02 -8.93847525e-01 -2.83067584e-01
3.51846039e-01 1.06214643e+00 -4.72019106e-01 -1.99783102e-01
-4.89756107e-01 1.15097857e+00 1.37763903e-01 6.53780043e-01
-1.25380480e+00 -4.14710760e-01 7.61971593e-01 -3.83465551e-02
3.74247253e-01 -4.54338998e-01 2.97006458e-01 -1.41728652e+00
5.57955652e-02 -7.20511854e-01 5.57364106e-01 -1.13708603e+00
-8.81676912e-01 7.11515024e-02 1.86787769e-01 -1.50885558e+00
-5.10101020e-01 -2.97873139e-01 -2.77343154e-01 1.53914973e-01
-1.10189521e+00 -1.41649330e+00 -8.16212177e-01 8.45676303e-01
6.16537571e-01 -8.89240652e-02 4.49767113e-01 5.90679586e-01
-3.50931376e-01 6.72820330e-01 -2.30830774e-01 3.42533767e-01
1.23897457e+00 -9.11162198e-01 -1.41699205e-03 8.76153767e-01
3.14909786e-01 7.83598006e-01 7.54738152e-01 -4.87848401e-01
-1.69756174e+00 -7.70893812e-01 6.50471091e-01 -7.64046371e-01
7.44737804e-01 -6.08715177e-01 -2.68770784e-01 1.02149045e+00
2.20579147e-01 -1.33698866e-01 8.30334485e-01 8.30264613e-02
-5.64394295e-01 7.86092579e-02 -9.99037147e-01 7.38654971e-01
1.47764790e+00 -1.04194176e+00 -6.60012424e-01 9.69046876e-02
5.96524656e-01 -1.27803430e-01 -7.55906701e-01 2.30396912e-01
1.25104046e+00 -1.06974077e+00 1.21741426e+00 -6.52418852e-01
5.16744316e-01 -4.75465804e-01 -6.78440869e-01 -8.96558583e-01
-2.93431371e-01 -5.41064084e-01 -3.77273083e-01 1.22693169e+00
-3.70886892e-01 -3.87684643e-01 4.01972890e-01 5.23478910e-02
1.22039862e-01 -7.03675374e-02 -9.39439178e-01 -7.40598381e-01
-8.40540051e-01 -7.71808207e-01 2.93300062e-01 1.03391564e+00
4.22058761e-01 1.21475451e-01 -8.63213956e-01 1.10419914e-01
5.17180681e-01 -4.43381071e-02 1.12512553e+00 -7.32543826e-01
-2.58275330e-01 -1.83796715e-02 -8.61457348e-01 -1.13951492e+00
8.13937336e-02 -4.41406935e-01 -8.91418010e-02 -1.15603781e+00
3.62406850e-01 5.49967766e-01 -4.59834367e-01 3.38435262e-01
1.06464274e-01 6.83785975e-01 2.35672683e-01 1.41909182e-01
-1.23696899e+00 6.04774356e-01 7.21770346e-01 1.06260907e-02
1.04336828e-01 -4.66747046e-01 -4.17081028e-01 8.56441438e-01
5.07562459e-01 -2.75534630e-01 -3.90354663e-01 -4.91364479e-01
2.39000395e-01 1.01695783e-01 8.57439518e-01 -1.12491870e+00
3.62825841e-01 -1.36524603e-01 6.30485535e-01 -7.11043835e-01
8.08473170e-01 -7.01806426e-01 -1.03066087e-01 1.81628853e-01
-3.33573341e-01 -3.55568491e-02 1.84399351e-01 1.15154004e+00
-3.09122175e-01 4.61548775e-01 3.21235657e-01 -1.96021870e-01
-1.19540334e+00 -8.42679963e-02 -5.05179167e-01 1.49814785e-01
1.14063168e+00 -3.57389271e-01 -5.37968159e-01 -9.20472026e-01
-8.00470114e-01 3.62124950e-01 5.10437906e-01 5.41478693e-01
6.86560452e-01 -1.59085751e+00 -3.15766037e-01 2.17529684e-02
6.19569182e-01 -8.23143542e-01 6.68746769e-01 1.34002018e+00
-3.05405229e-01 5.45298696e-01 -3.96260351e-01 -9.85912442e-01
-1.38273525e+00 6.33936226e-01 1.74130663e-01 6.60613358e-01
-6.38295531e-01 7.74484634e-01 4.94233787e-01 -3.18419859e-02
4.44822848e-01 -1.16691478e-01 -3.57690245e-01 3.60111803e-01
7.31758773e-01 7.25542307e-01 -6.34238362e-01 -1.02765954e+00
-6.25793815e-01 6.59071207e-01 4.17235643e-01 -4.51239347e-01
1.02157819e+00 -3.85625958e-01 3.00076306e-01 7.17248559e-01
1.12015629e+00 2.40316257e-01 -1.41040945e+00 -2.75190890e-01
-2.86489964e-01 -8.68405104e-01 -9.82929841e-02 -6.82913959e-01
-7.59248555e-01 9.96349454e-01 5.26487172e-01 -1.81161657e-01
1.07753849e+00 1.72709554e-01 6.02553785e-01 4.86812919e-01
6.52380824e-01 -1.15038693e+00 5.25411963e-01 2.41600588e-01
7.81249464e-01 -1.30826664e+00 9.32606682e-02 -1.89351171e-01
-9.83772933e-01 8.31390381e-01 6.73326313e-01 4.89473948e-03
1.83945209e-01 -3.07493657e-01 -1.39929533e-01 -2.05993518e-01
-8.41972053e-01 -4.39794421e-01 2.86054164e-01 6.43595755e-01
2.48196036e-01 -6.48789927e-02 -6.16268814e-03 6.96801126e-01
1.05778500e-01 -3.55639867e-02 2.78759748e-01 1.07318115e+00
-3.11495185e-01 -1.55529797e-01 -3.35274667e-01 -2.07543701e-01
-4.75274146e-01 2.81342238e-01 -6.38771296e-01 8.54543030e-01
1.85064115e-02 1.09941936e+00 -4.13534082e-02 -5.73680818e-01
3.08197677e-01 1.98898137e-01 5.10925293e-01 -2.07854286e-01
-4.68500793e-01 4.23354536e-01 2.20044419e-01 -1.31903148e+00
-7.75155306e-01 -8.38441253e-01 -9.07681942e-01 -2.51052320e-01
3.74841504e-02 -1.46096885e-01 3.85806978e-01 7.26023138e-01
4.97712225e-01 5.07660568e-01 3.00724387e-01 -1.12751913e+00
-4.08823676e-02 -9.03524995e-01 -5.28137565e-01 3.65284711e-01
4.53888595e-01 -1.03322589e+00 -6.03335977e-01 1.48425803e-01]
|
[8.321980476379395, 0.6305134892463684]
|
749688e5-35ef-467c-9ad5-d516971bd7ee
|
selective-eye-gaze-augmentation-to-enhance
|
2012.03145
| null |
https://arxiv.org/abs/2012.03145v1
|
https://arxiv.org/pdf/2012.03145v1.pdf
|
Selective Eye-gaze Augmentation To Enhance Imitation Learning In Atari Games
|
This paper presents the selective use of eye-gaze information in learning human actions in Atari games. Vast evidence suggests that our eye movement convey a wealth of information about the direction of our attention and mental states and encode the information necessary to complete a task. Based on this evidence, we hypothesize that selective use of eye-gaze, as a clue for attention direction, will enhance the learning from demonstration. For this purpose, we propose a selective eye-gaze augmentation (SEA) network that learns when to use the eye-gaze information. The proposed network architecture consists of three sub-networks: gaze prediction, gating, and action prediction network. Using the prior 4 game frames, a gaze map is predicted by the gaze prediction network which is used for augmenting the input frame. The gating network will determine whether the predicted gaze map should be used in learning and is fed to the final network to predict the action at the current frame. To validate this approach, we use publicly available Atari Human Eye-Tracking And Demonstration (Atari-HEAD) dataset consists of 20 Atari games with 28 million human demonstrations and 328 million eye-gazes (over game frames) collected from four subjects. We demonstrate the efficacy of selective eye-gaze augmentation in comparison with state of the art Attention Guided Imitation Learning (AGIL), Behavior Cloning (BC). The results indicate that the selective augmentation approach (the SEA network) performs significantly better than the AGIL and BC. Moreover, to demonstrate the significance of selective use of gaze through the gating network, we compare our approach with the random selection of the gaze. Even in this case, the SEA network performs significantly better validating the advantage of selectively using the gaze in demonstration learning.
|
['Ehsan T. Esfahani', 'Hemanth Manjunatha', 'Chaitanya Thammineni']
|
2020-12-05
| null | null | null | null |
['eye-tracking']
|
['computer-vision']
|
[ 1.26259491e-01 9.18425769e-02 -6.87491149e-02 -4.62887660e-02
1.74684957e-01 -7.58289844e-02 5.63957393e-01 -6.21370852e-01
-8.73481810e-01 9.36801076e-01 1.63741753e-01 -3.63187529e-02
1.22237176e-01 -3.44205379e-01 -8.41561854e-01 -9.02830958e-01
9.77055654e-02 2.04637721e-01 4.05244499e-01 -4.31548208e-01
6.49507463e-01 2.68108517e-01 -2.20456076e+00 1.04914151e-01
6.62592649e-01 7.96037197e-01 6.46860063e-01 7.79360175e-01
3.82903874e-01 1.18206322e+00 -4.66206461e-01 2.30046570e-01
5.25269471e-02 -8.38604271e-01 -7.72639811e-01 -1.64560258e-01
4.28462863e-01 -7.37970293e-01 -2.92513132e-01 8.71440649e-01
5.38127303e-01 6.89681411e-01 3.79474610e-01 -1.68319488e+00
-4.46013868e-01 2.97190160e-01 -4.81668144e-01 8.36290359e-01
5.13200164e-01 8.20192277e-01 9.90585685e-01 -6.17587805e-01
8.97527516e-01 1.03396690e+00 1.02277339e-01 1.25785446e+00
-9.14157391e-01 -9.16318893e-01 3.12868267e-01 7.06498146e-01
-9.80493128e-01 -3.94091040e-01 7.98093617e-01 -4.64134336e-01
1.06624019e+00 1.67019349e-02 1.05113780e+00 1.46509373e+00
2.22031251e-01 9.34513032e-01 1.21820736e+00 -6.11202538e-01
4.40879799e-02 -1.74287513e-01 8.43659118e-02 7.29340255e-01
-1.20651715e-01 8.87149334e-01 -1.16661417e+00 3.89601350e-01
8.05415809e-01 3.28766331e-02 -7.68637717e-01 -2.75381535e-01
-1.15794671e+00 6.57467544e-01 6.35098815e-01 1.36942223e-01
-7.31232941e-01 4.09554094e-01 7.04152137e-02 3.73772293e-01
2.27462307e-01 5.39785743e-01 -3.08948070e-01 -6.14191949e-01
-6.38512015e-01 3.02863449e-01 3.98488462e-01 6.18900239e-01
8.52360368e-01 1.99129377e-02 -5.79200685e-01 3.09642434e-01
6.54767215e-01 5.78329444e-01 5.87957561e-01 -1.01666331e+00
3.04567873e-01 6.23981714e-01 9.63244289e-02 -4.64485079e-01
-5.83687901e-01 3.48591544e-02 7.74872750e-02 1.01164615e+00
6.59821689e-01 -2.73304820e-01 -9.05068576e-01 2.13262177e+00
2.77260035e-01 3.75146180e-01 2.90464498e-02 1.41889811e+00
8.98613870e-01 2.97453791e-01 4.30172265e-01 -1.27079980e-02
1.47259521e+00 -1.00350487e+00 -7.57234573e-01 -2.19737530e-01
7.99563944e-01 -3.02198917e-01 1.28023696e+00 3.37197125e-01
-8.50512087e-01 -6.46190166e-01 -8.66324842e-01 -1.21242039e-01
-9.67811495e-02 2.01559961e-01 3.79397869e-01 2.70223528e-01
-1.11919940e+00 5.17511845e-01 -7.22977459e-01 -5.21125019e-01
3.95888418e-01 7.13863611e-01 -3.28122586e-01 4.04001296e-01
-1.02002549e+00 1.07272792e+00 2.48015895e-01 -9.12636518e-02
-1.14771998e+00 -5.16707122e-01 -8.86690676e-01 2.34186783e-01
5.26008248e-01 -6.11283362e-01 1.11302590e+00 -1.28894055e+00
-1.92264712e+00 7.09163427e-01 -3.19551528e-01 -5.85426271e-01
2.03923419e-01 -5.40546000e-01 7.78525919e-02 2.61236548e-01
-1.89645946e-01 1.42142737e+00 1.01059377e+00 -8.38988125e-01
-1.02080262e+00 -3.79867524e-01 3.84909540e-01 4.42490488e-01
-4.14647497e-02 7.42249042e-02 -1.11499213e-01 -2.27902487e-01
-4.90285248e-01 -1.38933611e+00 4.10841852e-01 -1.18230768e-01
-1.25003263e-01 -7.79838145e-01 9.55680788e-01 -4.37828720e-01
1.23324132e+00 -2.16875696e+00 6.04304671e-01 -1.30271450e-01
4.34896260e-01 1.32667825e-01 -6.51611164e-02 9.45638344e-02
-3.27358156e-01 -3.45838964e-01 2.13212281e-01 -3.91885519e-01
-2.34214038e-01 1.14609003e-01 -1.87755674e-01 4.29805785e-01
9.79559869e-02 1.17590129e+00 -8.48486006e-01 -2.99526930e-01
3.73546392e-01 4.19033170e-01 -9.09513533e-01 4.78415400e-01
-2.32336089e-01 1.02170146e+00 -1.90654010e-01 2.14128450e-01
-1.33720785e-01 -2.93425798e-01 -2.76288599e-01 -8.61237273e-02
-2.49979168e-01 4.39491987e-01 -4.84972626e-01 1.60394740e+00
-1.22880496e-01 1.25867712e+00 -3.41648728e-01 -3.71717066e-01
6.37558341e-01 2.13096634e-01 2.03482136e-01 -1.05447400e+00
5.69664419e-01 -1.81327268e-01 8.49745452e-01 -8.41974080e-01
3.57804805e-01 9.59633812e-02 4.53690320e-01 8.61046791e-01
2.79930830e-01 3.46800625e-01 3.69117349e-01 5.69926202e-02
8.07537377e-01 8.31018984e-01 8.67366493e-02 -1.52084857e-01
7.42530227e-01 9.83935967e-03 1.78594321e-01 4.48812544e-01
-5.66420019e-01 3.50290090e-01 5.72901666e-01 -3.88289273e-01
-7.97168612e-01 -4.40511405e-01 6.12227678e-01 1.74904513e+00
1.36139274e-01 -1.73362762e-01 -1.01074803e+00 -7.65765071e-01
-3.25225890e-01 9.44402754e-01 -1.19851100e+00 -4.44970846e-01
-7.14611888e-01 -3.53164375e-02 2.44148076e-01 5.90051115e-01
3.80617440e-01 -2.12945938e+00 -1.37838316e+00 -4.51665848e-01
-1.86473906e-01 -6.45317793e-01 -6.05761409e-01 3.22414935e-02
-5.47235668e-01 -1.49731696e+00 -6.02290690e-01 -4.92281079e-01
6.95856810e-01 2.06888810e-01 7.23983943e-01 3.27493072e-01
1.40029088e-01 6.09262764e-01 -4.36578989e-01 -4.84807342e-01
-3.45573425e-01 1.33086145e-01 1.04924969e-01 -1.14240296e-01
7.15993702e-01 -3.08198780e-01 -7.30014622e-01 2.02758223e-01
-3.45272511e-01 1.49976656e-01 5.20458996e-01 9.30885911e-01
1.69996500e-01 -8.19425464e-01 2.59762913e-01 -6.93185210e-01
6.80838466e-01 -2.88506031e-01 -6.90980852e-01 5.88138076e-03
-5.77681243e-01 3.19502354e-01 3.87742482e-02 -5.74685872e-01
-1.06802380e+00 -1.44440636e-01 -3.89644504e-02 -7.14750886e-01
-2.62685627e-01 2.36246243e-01 3.78446698e-01 -1.04992010e-01
7.38753021e-01 1.37034148e-01 5.00884235e-01 -1.47991315e-01
1.42274782e-01 4.67274070e-01 2.10758388e-01 -1.99423730e-01
2.28324130e-01 3.34786385e-01 2.83103832e-03 -5.95750213e-01
-7.05373287e-01 -3.27888459e-01 -8.10858190e-01 -6.76852047e-01
1.28685927e+00 -7.68574178e-01 -1.51564705e+00 4.84523803e-01
-8.78530741e-01 -8.29974651e-01 -3.48625571e-01 7.56138861e-01
-9.06555712e-01 -3.19332108e-02 -1.77906409e-01 -8.03840399e-01
-3.18242222e-01 -1.45958591e+00 8.31617713e-01 5.29072881e-01
-4.27074522e-01 -8.86655986e-01 1.97828189e-01 2.16906115e-01
7.22386464e-02 -1.15611628e-01 4.70580697e-01 -7.14372993e-01
-5.22914886e-01 3.20295304e-01 -3.46974283e-02 6.71171844e-02
1.07756615e-01 -7.65533000e-02 -8.75350893e-01 -3.19026738e-01
-1.83970481e-01 -4.31364834e-01 7.18474925e-01 5.36955237e-01
7.13178456e-01 1.94206923e-01 -2.62595773e-01 4.80190933e-01
8.99560869e-01 2.91074723e-01 8.81082833e-01 5.23265183e-01
7.43881047e-01 6.29042327e-01 8.44633937e-01 2.08706543e-01
3.99673641e-01 8.76430392e-01 7.53324986e-01 2.76657313e-01
-4.25463438e-01 -1.16412766e-01 6.87345505e-01 1.49734110e-01
-6.95059001e-01 -1.66127145e-01 -7.43928432e-01 2.21345156e-01
-1.71939504e+00 -1.17977607e+00 -7.98177198e-02 2.20172358e+00
7.01814950e-01 1.16373897e-01 4.35654730e-01 -1.19077943e-01
4.76478487e-01 -9.02086869e-02 -6.51712239e-01 -3.35805774e-01
3.46441865e-01 3.12310874e-01 1.48604989e-01 4.21404421e-01
-7.90910482e-01 1.05622685e+00 6.04845333e+00 2.66423374e-01
-1.42819369e+00 1.91895097e-01 -3.35000940e-02 -3.35909784e-01
3.14757288e-01 -1.82797909e-01 -9.72082019e-01 5.46776950e-01
9.63521123e-01 1.81023031e-01 7.15124488e-01 7.25858331e-01
3.75384331e-01 -5.51467657e-01 -1.11795533e+00 9.31816518e-01
3.15392762e-01 -1.07116389e+00 -2.92483509e-01 2.86886930e-01
3.31801981e-01 3.18563789e-01 1.00319631e-01 4.61748332e-01
2.55380511e-01 -8.63714874e-01 6.94588602e-01 7.88783967e-01
7.47264743e-01 -4.55965579e-01 6.20976925e-01 5.06458521e-01
-6.62172139e-01 -4.80196774e-01 5.56327291e-02 -4.20337886e-01
1.41044892e-02 -8.30011845e-01 -8.08763623e-01 -8.64195973e-02
8.13997269e-01 7.99731016e-01 -6.39861763e-01 8.58650744e-01
-8.28539431e-01 7.06159472e-01 -1.14422284e-01 -2.79197991e-01
2.65553892e-01 8.50492064e-03 7.73773611e-01 5.20878732e-01
6.50856197e-02 2.08330169e-01 -2.46333376e-01 9.44849670e-01
1.46721110e-01 -2.47339934e-01 -4.17004734e-01 1.30607933e-01
2.40521416e-01 8.33097279e-01 -4.41401511e-01 -2.54112989e-01
-3.12702179e-01 8.21354926e-01 6.42440856e-01 4.79370236e-01
-9.47050095e-01 -5.25046922e-02 7.05813944e-01 1.32815883e-01
5.50703704e-01 2.93399114e-02 1.64086729e-01 -8.09405625e-01
-4.41055000e-01 -7.64397800e-01 2.39402339e-01 -1.36936355e+00
-5.51621199e-01 6.29275024e-01 7.85151049e-02 -1.29175615e+00
-6.47797227e-01 -7.47353196e-01 -6.47033632e-01 1.10661006e+00
-1.54086232e+00 -8.50757480e-01 -6.43753469e-01 7.66436338e-01
6.64831519e-01 -2.72762001e-01 5.08379638e-01 -1.02051742e-01
-5.82902670e-01 4.25245345e-01 -6.74053490e-01 4.33501005e-02
7.50271857e-01 -1.21393371e+00 9.69826896e-03 7.44591177e-01
8.67880657e-02 5.45664489e-01 7.71575749e-01 -5.87483704e-01
-9.63113964e-01 -5.16992152e-01 5.83197236e-01 -6.63475513e-01
3.07329297e-01 -5.09051308e-02 -8.18899632e-01 9.23140407e-01
5.69102466e-01 -7.62498379e-02 5.80044270e-01 4.57796976e-02
2.76042186e-02 2.36294776e-01 -8.58381331e-01 6.84163392e-01
9.63664412e-01 -4.02011514e-01 -9.97430623e-01 -2.43138283e-01
5.40305197e-01 -4.83695537e-01 -2.59005070e-01 2.41549127e-02
6.50372922e-01 -1.14713335e+00 6.36070848e-01 -7.91317821e-01
4.95663792e-01 -3.07795644e-01 2.99202323e-01 -1.43117046e+00
-3.04137915e-01 -3.86009783e-01 -3.03510398e-01 5.61689198e-01
1.05563745e-01 -6.15346909e-01 7.89119124e-01 4.51565504e-01
-1.78381801e-01 -5.70796728e-01 -9.40901160e-01 -2.23877937e-01
-2.74415433e-01 -6.28570244e-02 3.53241324e-01 4.99930620e-01
2.06958815e-01 5.39480686e-01 -3.80878240e-01 -2.24888995e-01
2.39747912e-01 -2.22599119e-01 1.20044756e+00 -1.28357768e+00
-7.76178166e-02 -4.22793865e-01 -3.38195711e-01 -1.29152048e+00
4.14887577e-01 -4.88853902e-01 3.10261726e-01 -1.16281021e+00
-4.32791971e-02 -7.92000592e-02 -3.47402155e-01 7.70054936e-01
-4.54951406e-01 4.39038426e-01 5.74722469e-01 4.49534178e-01
-6.30044878e-01 6.52613699e-01 1.61835277e+00 3.14999759e-01
-5.34920335e-01 6.89899698e-02 -3.42790693e-01 5.35476923e-01
6.27379298e-01 -4.61704284e-01 -4.45654511e-01 -8.00043270e-02
2.13209435e-01 -2.16529474e-01 6.86066031e-01 -1.08782947e+00
4.49259430e-01 -4.94124144e-02 2.16247380e-01 -5.81105769e-01
4.74735290e-01 -6.70184374e-01 -3.31196606e-01 6.83162093e-01
-3.61046284e-01 -1.98523849e-02 3.17494184e-01 4.78530705e-01
1.26581147e-01 -2.90447265e-01 5.65523684e-01 3.57223824e-02
-1.18419480e+00 6.26740009e-02 -4.06295151e-01 -3.44687328e-02
1.13312519e+00 -5.36454499e-01 -4.28851515e-01 -3.96976560e-01
-1.10615671e+00 3.31484646e-01 4.36171204e-01 6.18168533e-01
4.64111149e-01 -1.02248657e+00 -5.64967573e-01 5.43538570e-01
1.59330562e-01 -3.76675606e-01 1.52924404e-01 1.37593424e+00
-2.84458071e-01 4.67886209e-01 -7.29058981e-01 -8.69300485e-01
-1.51160407e+00 6.28686547e-01 4.28616405e-01 -7.39757577e-03
-7.03079224e-01 9.39442515e-01 5.43978393e-01 7.21586719e-02
-1.25884973e-02 -3.10673773e-01 -9.02703047e-01 6.45502796e-03
6.08210027e-01 3.86188984e-01 -3.20364237e-01 -1.07829297e+00
-1.42088264e-01 3.49307328e-01 -5.73861273e-03 -1.61066741e-01
1.24476767e+00 -3.33229065e-01 7.86864012e-02 5.02287030e-01
6.69864833e-01 -1.21482946e-01 -1.99285853e+00 3.90730798e-02
-3.59193653e-01 -3.56723398e-01 5.25083207e-02 -8.49616945e-01
-1.05721033e+00 1.05968308e+00 7.95452416e-01 -6.45036846e-02
1.05421901e+00 5.43711744e-02 2.58308709e-01 3.50724339e-01
2.60191351e-01 -8.27658653e-01 2.73466229e-01 5.97863317e-01
9.08050478e-01 -1.34455085e+00 -1.24150835e-01 -4.46019433e-02
-1.00174594e+00 9.53387380e-01 1.14284992e+00 -3.63100678e-01
6.03824079e-01 -3.35507900e-01 4.12135161e-02 -5.84161937e-01
-9.79681253e-01 -7.11439550e-01 3.86981189e-01 8.06816339e-01
3.09876710e-01 -2.87627131e-01 -1.01209179e-01 2.50957638e-01
-3.12657595e-01 3.10986876e-01 4.05435950e-01 7.92283237e-01
-4.50697154e-01 -6.54732943e-01 -2.67292500e-01 3.27118248e-01
-1.90215051e-01 -1.11317314e-01 -9.69151706e-02 1.04029703e+00
1.39749676e-01 6.91139460e-01 3.66118014e-01 -3.66847605e-01
1.80037305e-01 2.50519365e-01 7.74452627e-01 -8.54083180e-01
-7.79809117e-01 -2.25099057e-01 -2.62542903e-01 -7.68962443e-01
-8.43405008e-01 -8.25380683e-01 -1.39085531e+00 -1.48099944e-01
-5.94023943e-01 -7.30428025e-02 3.54148626e-01 1.23732340e+00
4.01939839e-01 8.59889507e-01 8.68425667e-02 -1.21293855e+00
-2.77117044e-01 -1.36082649e+00 -2.73933411e-01 5.77503622e-01
4.74393785e-01 -1.53703463e+00 -4.70320910e-01 1.58904731e-01]
|
[13.97118854522705, 0.05053410306572914]
|
605f95fa-c09f-419e-8136-4b17702c7ec1
|
chinese-native-language-identification
| null | null |
https://aclanthology.org/E14-4019
|
https://aclanthology.org/E14-4019.pdf
|
Chinese Native Language Identification
| null |
['Mark Dras', 'Shervin Malmasi']
|
2014-04-01
| null | null | null |
eacl-2014-4
|
['native-language-identification']
|
['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]
|
[-7.364411354064941, 3.527883529663086]
|
1edcb0da-2edf-40be-b642-b0b95ac550fd
|
improving-long-tailed-document-level-relation
|
2205.10511
| null |
https://arxiv.org/abs/2205.10511v1
|
https://arxiv.org/pdf/2205.10511v1.pdf
|
Improving Long Tailed Document-Level Relation Extraction via Easy Relation Augmentation and Contrastive Learning
|
Towards real-world information extraction scenario, research of relation extraction is advancing to document-level relation extraction(DocRE). Existing approaches for DocRE aim to extract relation by encoding various information sources in the long context by novel model architectures. However, the inherent long-tailed distribution problem of DocRE is overlooked by prior work. We argue that mitigating the long-tailed distribution problem is crucial for DocRE in the real-world scenario. Motivated by the long-tailed distribution problem, we propose an Easy Relation Augmentation(ERA) method for improving DocRE by enhancing the performance of tailed relations. In addition, we further propose a novel contrastive learning framework based on our ERA, i.e., ERACL, which can further improve the model performance on tailed relations and achieve competitive overall DocRE performance compared to the state-of-arts.
|
['Shouling Ji', 'Bo Long', 'Xuhong Zhang', 'Yiming Wu', 'Lingfei Wu', 'Tengfei Ma', 'Yangkai Du']
|
2022-05-21
| null | null | null | null |
['document-level-relation-extraction']
|
['natural-language-processing']
|
[ 2.91459620e-01 6.33452237e-01 -5.51279783e-01 -1.21867105e-01
-7.47093797e-01 -4.36562479e-01 7.50501633e-01 1.82616979e-01
-2.07328841e-01 9.17090595e-01 4.53255475e-01 -5.43993413e-01
-4.09467459e-01 -8.99760067e-01 -4.73737359e-01 -1.64028376e-01
-2.35594437e-01 8.06365609e-01 1.32905975e-01 -3.84752899e-01
-3.36098999e-01 4.32267785e-01 -9.99505520e-01 3.74361664e-01
8.28453064e-01 1.02200735e+00 -3.95457864e-01 4.00851548e-01
-2.68810809e-01 1.10579073e+00 -5.31754673e-01 -8.32144916e-01
1.98236912e-01 -3.52965057e-01 -1.28710461e+00 -2.03101560e-01
2.07335904e-01 -8.06019008e-02 -6.36037230e-01 8.19100559e-01
2.26604208e-01 -1.06949173e-01 9.39121902e-01 -1.37032700e+00
-6.54202282e-01 1.27090418e+00 -6.67248785e-01 2.29736388e-01
2.30467096e-01 -6.28581822e-01 1.40267527e+00 -7.80638814e-01
6.57912374e-01 1.15891862e+00 5.41722059e-01 1.53782561e-01
-1.02370834e+00 -7.59163499e-01 4.64179099e-01 3.82890046e-01
-1.50493288e+00 -4.56930876e-01 8.78396869e-01 -1.51779830e-01
1.36784220e+00 3.18445861e-01 3.75520378e-01 1.00179279e+00
1.78678602e-01 9.58607554e-01 9.10348773e-01 -4.64401931e-01
-2.83982098e-01 -9.84488502e-02 3.58138859e-01 4.27502751e-01
6.66252494e-01 -6.45931512e-02 -6.32452488e-01 5.31280935e-02
4.61739093e-01 -2.55308241e-01 -5.91368303e-02 -8.33418444e-02
-8.20745289e-01 4.75181848e-01 3.08440864e-01 2.43827134e-01
-2.14728326e-01 -8.11800659e-02 4.12297577e-01 3.02969843e-01
7.05817044e-01 7.40532935e-01 -1.10262895e+00 -2.41476800e-02
-8.14725578e-01 5.30779183e-01 1.11726213e+00 1.42316008e+00
4.00108933e-01 -4.21761543e-01 -3.50575328e-01 7.67128170e-01
1.66631773e-01 7.44710267e-02 1.52645960e-01 -4.54481393e-01
9.16881263e-01 9.29894805e-01 -3.61155123e-02 -9.28066313e-01
-5.90785861e-01 -1.16705787e+00 -1.18316329e+00 -5.11045992e-01
3.00155252e-01 -3.29511464e-01 -8.18265498e-01 1.52888024e+00
2.61184126e-01 1.91801131e-01 4.35227156e-01 4.16128039e-01
1.06933832e+00 3.94161493e-01 9.26907584e-02 -4.61260945e-01
1.60173869e+00 -1.18496346e+00 -1.00935483e+00 -5.32273948e-01
8.04718256e-01 -5.73607683e-01 6.19192541e-01 3.02105218e-01
-8.15365374e-01 -1.02335908e-01 -1.21698308e+00 -3.82690936e-01
-4.53919411e-01 1.28516957e-01 1.23418653e+00 3.47464889e-01
-4.43537712e-01 3.40030462e-01 -7.47895241e-01 -5.63604124e-02
6.46210372e-01 3.54723155e-01 -4.02724802e-01 4.53130007e-02
-1.54446876e+00 9.41016197e-01 6.61663294e-01 1.35483861e-01
-1.16999358e-01 -8.42927754e-01 -9.90861714e-01 3.43487293e-01
1.23382521e+00 -7.53548205e-01 1.18265581e+00 9.09058750e-02
-1.18006313e+00 5.80301046e-01 -2.29936749e-01 -7.76272893e-01
1.39804095e-01 -8.11340511e-01 -6.55118942e-01 -2.73444325e-01
-4.19072844e-02 9.54537019e-02 3.42182130e-01 -1.33053398e+00
-5.36170900e-01 -2.66295403e-01 1.21406563e-01 1.28629476e-01
-4.94120568e-01 7.86866769e-02 -4.92379278e-01 -6.58501387e-01
1.58065453e-01 -6.38730407e-01 -2.32195154e-01 -7.65986383e-01
-7.29417264e-01 -5.78319907e-01 8.36181045e-01 -5.89040935e-01
1.83349383e+00 -1.66497302e+00 -1.03645466e-01 6.58110529e-02
6.94228292e-01 5.88337660e-01 -5.75071573e-03 6.57223463e-01
-9.40096751e-02 1.95193231e-01 -6.46115318e-02 -2.97044337e-01
4.98873405e-02 1.47314772e-01 -3.50581735e-01 -3.00039470e-01
6.45428181e-01 1.38177323e+00 -9.71758544e-01 -7.79608488e-01
-3.54860961e-01 2.09564254e-01 -4.33126569e-01 3.59849781e-01
-2.42304891e-01 3.45285609e-02 -4.17123556e-01 6.18908167e-01
7.16385305e-01 -4.55951035e-01 6.36833012e-01 -4.66012836e-01
3.20121467e-01 7.52675116e-01 -9.99578059e-01 1.31652749e+00
-5.44602454e-01 2.76297599e-01 -3.47758502e-01 -7.75973916e-01
1.12741792e+00 3.85681838e-01 6.02295816e-01 -4.48656231e-01
1.56730458e-01 2.08512932e-01 4.02003646e-01 -2.76503861e-01
8.45665753e-01 -7.19959736e-02 -3.57914530e-02 1.70392558e-01
2.32953161e-01 -3.93357500e-02 3.67798120e-01 5.94199419e-01
1.27705121e+00 4.32570636e-01 6.77372813e-01 -1.30320936e-02
3.34155083e-01 -3.26018184e-01 7.81306088e-01 4.44829613e-01
1.90262020e-01 1.27871111e-01 8.78858089e-01 -1.26426682e-01
-5.78186572e-01 -6.15345240e-01 -1.89035431e-01 7.84574807e-01
7.79485423e-03 -1.12315440e+00 -4.76837188e-01 -1.29390216e+00
-3.01447026e-02 5.54087877e-01 -2.39782244e-01 -2.29120612e-01
-8.09681594e-01 -1.15821290e+00 6.61483288e-01 8.11373413e-01
5.19027174e-01 -5.89177191e-01 -1.13204345e-01 5.39141655e-01
-4.78887409e-01 -1.88343227e+00 -1.61168963e-01 5.08232355e-01
-5.90125501e-01 -8.72448206e-01 -2.72612721e-01 -3.78700435e-01
9.97964367e-02 -3.72695848e-02 1.65206051e+00 -2.48555019e-01
1.73913479e-01 -3.91284525e-01 -6.81432486e-01 -5.72420776e-01
-5.66528253e-02 6.88489795e-01 -3.15812945e-01 -3.44744802e-01
8.39829624e-01 -8.66318762e-01 -2.99000472e-01 8.95337611e-02
-6.93526924e-01 2.46868521e-01 8.02464545e-01 7.48975694e-01
3.96443129e-01 5.43311715e-01 1.09460056e+00 -1.59473157e+00
7.36079276e-01 -6.44959867e-01 -1.44030556e-01 4.45088506e-01
-1.30534220e+00 3.20037127e-01 4.93145406e-01 -2.11891621e-01
-1.11315358e+00 -3.18523824e-01 -3.12652625e-02 2.68474728e-01
5.83553538e-02 9.63800669e-01 -6.13911331e-01 3.82373601e-01
3.71128589e-01 -3.55352074e-01 -6.37388825e-01 -6.09971106e-01
5.94914913e-01 8.13465238e-01 7.98953295e-01 -7.96766102e-01
8.10543776e-01 8.47638026e-03 5.27829051e-01 -1.35801747e-01
-1.54428220e+00 -5.21583915e-01 -6.85616076e-01 4.74834800e-01
3.83072078e-01 -1.05861259e+00 -5.24292409e-01 2.57266909e-01
-1.14548385e+00 1.89789012e-02 -2.05427274e-01 2.60676831e-01
-1.90182552e-01 1.89407140e-01 -7.56882250e-01 -9.35346603e-01
-6.96510017e-01 -6.40920162e-01 1.16490531e+00 1.12131312e-01
-3.89296442e-01 -9.77061570e-01 7.40020350e-02 4.96014297e-01
3.92843746e-02 2.73159444e-01 1.15641522e+00 -1.00554121e+00
-5.28489470e-01 -2.79092938e-01 -6.80824161e-01 -1.25943217e-02
3.53610784e-01 -3.17496270e-01 -8.95595610e-01 1.16677940e-01
-4.25245821e-01 -1.40384078e-01 9.08289731e-01 -1.73975334e-01
9.80066955e-01 -4.72985566e-01 -6.72346711e-01 4.82321799e-01
1.05533576e+00 1.63228467e-01 7.27329552e-01 2.69838780e-01
1.08512771e+00 5.45943439e-01 8.94264162e-01 1.29970342e-01
1.11869776e+00 9.62743104e-01 1.62093878e-01 -2.14147210e-01
-4.69477952e-01 -4.48124915e-01 -6.56031594e-02 9.71551418e-01
-2.87402302e-01 -5.24265647e-01 -1.07657611e+00 7.61205196e-01
-2.02470422e+00 -4.28373992e-01 -2.28914112e-01 1.52938640e+00
1.39487398e+00 5.56760490e-01 -7.43119232e-03 5.33894658e-01
1.22632936e-01 1.03291541e-01 -1.19883828e-01 -4.16672170e-01
-2.39962369e-01 6.29527330e-01 4.15798217e-01 3.76075089e-01
-1.45602345e+00 1.40564740e+00 5.42257833e+00 9.98529136e-01
-7.05364823e-01 -6.84883744e-02 3.80867720e-01 3.18399936e-01
-2.64406323e-01 2.07083926e-01 -1.31081724e+00 1.15228742e-02
1.06783581e+00 -2.40620881e-01 -8.99288431e-02 6.68439269e-01
-3.78581882e-01 1.91219509e-01 -1.40638232e+00 8.29640925e-01
-4.39883955e-02 -1.28229725e+00 -7.09085390e-02 2.87331283e-01
6.92514479e-01 -3.69768709e-01 -3.29663485e-01 6.57973647e-01
5.67321599e-01 -1.14958394e+00 1.88838482e-01 4.50185150e-01
7.79243708e-01 -9.15867686e-01 1.28868747e+00 1.91379488e-01
-1.54415107e+00 1.10217065e-01 -9.59117524e-03 -2.18766838e-01
1.79140389e-01 1.08050597e+00 -9.87871826e-01 1.43608725e+00
2.08975822e-01 9.13138807e-01 -7.60494590e-01 6.92738295e-01
-4.13315415e-01 7.28469133e-01 -1.70480207e-01 2.94849932e-01
1.93407889e-02 3.66961062e-02 3.28153551e-01 1.48413742e+00
-7.34818727e-02 1.49300873e-01 2.14510933e-01 4.61222887e-01
-4.35075998e-01 -2.16610208e-02 -4.60339099e-01 -3.44042450e-01
6.44415796e-01 1.44458151e+00 -4.67902660e-01 -2.62380809e-01
-4.12275195e-01 6.22448504e-01 6.48695529e-01 2.83276528e-01
-4.17390645e-01 -5.09438574e-01 2.06351981e-01 -3.14884931e-02
2.44214043e-01 -2.53408045e-01 -6.30453229e-01 -1.20461702e+00
2.61450857e-01 -9.93630886e-01 6.46804035e-01 -1.76651999e-01
-1.38124895e+00 9.15934622e-01 2.44439393e-01 -9.67295945e-01
-4.96576250e-01 -4.74309415e-01 -1.37780577e-01 6.83943033e-01
-1.78828621e+00 -1.84144330e+00 2.45946556e-01 2.02628553e-01
3.13289672e-01 -1.07383654e-01 9.52094674e-01 6.54206574e-01
-6.55862868e-01 1.03624153e+00 -5.21715641e-01 2.99052477e-01
6.53167367e-01 -1.47875357e+00 7.03770518e-01 8.67044866e-01
5.35495818e-01 6.99982226e-01 3.51641119e-01 -8.08855355e-01
-1.16816854e+00 -1.29277277e+00 1.47059476e+00 -6.42032325e-01
8.02345514e-01 -4.55760241e-01 -8.14217567e-01 9.50289428e-01
8.61723348e-02 1.43208476e-02 6.62571013e-01 9.59012747e-01
-6.98604524e-01 -2.23059237e-01 -9.06625986e-01 6.47496402e-01
1.30895948e+00 -5.15143812e-01 -6.21486783e-01 1.10982239e-01
8.75708103e-01 -4.96298373e-01 -1.32592201e+00 1.01782143e+00
4.67982441e-01 -2.62277544e-01 9.43752050e-01 -9.23881292e-01
7.66419768e-01 -4.53891978e-02 -2.92594116e-02 -1.17816794e+00
-2.99922466e-01 -8.85669053e-01 -1.37938213e+00 1.73900509e+00
7.22926736e-01 -3.48554730e-01 7.44077504e-01 4.74754810e-01
-1.95151456e-02 -1.36390269e+00 -5.95659614e-01 -8.57366681e-01
2.75893033e-01 -3.42326313e-01 9.98830378e-01 8.56423795e-01
2.26269811e-01 1.14458561e+00 -6.00937903e-01 2.28559256e-01
3.80583555e-01 3.96075726e-01 7.77051449e-01 -1.36940014e+00
-4.34658349e-01 -7.93907270e-02 -2.72658944e-01 -1.26889241e+00
2.10773602e-01 -1.01982498e+00 -1.18866362e-01 -1.67565489e+00
3.79211336e-01 -7.38588989e-01 -5.83900690e-01 5.25275290e-01
-6.79225802e-01 9.17011648e-02 1.39370620e-01 -8.05533316e-04
-6.74366415e-01 5.53439081e-01 1.41126931e+00 3.76216546e-02
-7.46904090e-02 7.90200830e-02 -1.23292041e+00 6.90774858e-01
6.01896703e-01 -5.33032238e-01 -6.18366539e-01 -1.93782106e-01
5.27382314e-01 -1.84726361e-02 -7.79176429e-02 -6.48023903e-01
1.82282925e-01 -1.45786926e-01 8.53584707e-02 -8.15706909e-01
2.56222188e-01 -8.91811252e-01 -2.07776964e-01 -2.06021711e-01
-3.07774544e-01 -2.11603954e-01 6.28038449e-03 4.50010329e-01
-3.85623634e-01 1.43263340e-01 1.05115160e-01 2.23303884e-01
-2.45511800e-01 4.92280424e-01 9.56704021e-02 3.93084586e-01
4.98442441e-01 3.96914065e-01 -6.87792420e-01 -2.93636948e-01
-4.61252183e-01 3.02359253e-01 -4.07910198e-01 3.23259264e-01
3.06839854e-01 -1.34317601e+00 -8.69095981e-01 -6.74370751e-02
2.77345896e-01 3.20133805e-01 -3.18680465e-01 8.31844687e-01
2.16728598e-01 6.13942027e-01 4.60667700e-01 8.19105282e-02
-1.15260005e+00 5.87045968e-01 -5.97430952e-02 -1.48586285e+00
-7.18313277e-01 9.81308699e-01 -4.69322167e-02 -2.71758169e-01
-1.59022547e-02 -4.58938092e-01 -5.18500447e-01 5.77543564e-02
4.00843412e-01 1.20048702e-01 5.27699530e-01 -3.56292576e-01
-4.66976047e-01 2.14412510e-01 -6.83892667e-01 1.03486910e-01
1.35944510e+00 -7.95788467e-02 -2.37522379e-01 1.88599840e-01
7.74994493e-01 4.46404994e-01 -7.95921087e-01 -7.45529413e-01
8.21910739e-01 -2.02320918e-01 1.83686297e-02 -1.24843979e+00
-9.75005925e-01 5.39186180e-01 -4.26873490e-02 1.79563895e-01
1.16398239e+00 2.43868545e-01 1.40094340e+00 4.87444788e-01
4.06492352e-01 -7.79277682e-01 -3.25700700e-01 8.79742205e-01
7.66001523e-01 -1.12139153e+00 4.92043495e-01 -1.24855459e+00
-5.71308672e-01 6.93326771e-01 5.67153931e-01 1.66713655e-01
8.50857675e-01 1.02414334e+00 8.32170323e-02 -3.81083399e-01
-9.57725346e-01 -6.08386338e-01 6.84318721e-01 7.02597857e-01
7.80427456e-01 2.17478693e-01 -5.96145034e-01 1.18362498e+00
-5.83540857e-01 -1.47628449e-02 1.46316737e-01 8.19227755e-01
2.05513090e-01 -1.72367907e+00 2.85332888e-01 6.66073859e-01
-7.44791925e-01 -6.01628661e-01 -6.72998905e-01 6.57825947e-01
2.70927489e-01 1.16268885e+00 -4.43402499e-01 -5.65436780e-01
2.91205645e-01 2.58503824e-01 6.13984466e-01 -8.33743334e-01
-5.49370289e-01 1.40442759e-01 9.42860126e-01 -1.95203632e-01
-3.99262160e-01 -3.07626069e-01 -1.15176213e+00 2.10404377e-02
-6.59713805e-01 2.32160628e-01 1.76996171e-01 1.32033277e+00
3.94389391e-01 1.09314644e+00 2.88924098e-01 5.97046688e-02
-2.82196403e-01 -1.25303257e+00 -3.60755563e-01 2.04778343e-01
1.96814910e-01 -8.37167978e-01 2.06674099e-01 -8.15955177e-02]
|
[9.291796684265137, 8.626953125]
|
34b609ba-02f6-45c5-ab2f-37939f07a993
|
dense-prediction-transformer-for-scale
|
2210.01723
| null |
https://arxiv.org/abs/2210.01723v1
|
https://arxiv.org/pdf/2210.01723v1.pdf
|
Dense Prediction Transformer for Scale Estimation in Monocular Visual Odometry
|
Monocular visual odometry consists of the estimation of the position of an agent through images of a single camera, and it is applied in autonomous vehicles, medical robots, and augmented reality. However, monocular systems suffer from the scale ambiguity problem due to the lack of depth information in 2D frames. This paper contributes by showing an application of the dense prediction transformer model for scale estimation in monocular visual odometry systems. Experimental results show that the scale drift problem of monocular systems can be reduced through the accurate estimation of the depth map by this model, achieving competitive state-of-the-art performance on a visual odometry benchmark.
|
['Marcos R. O. A. Maximo', 'André O. Françani']
|
2022-10-04
| null | null | null | null |
['monocular-visual-odometry']
|
['robots']
|
[-3.29760462e-01 1.75575629e-01 -2.63430655e-01 -1.48228392e-01
1.26624495e-01 -2.92410910e-01 7.65155852e-01 -3.02187979e-01
-4.87596184e-01 7.39262402e-01 -1.97591454e-01 8.80747437e-02
4.38711047e-01 -3.93105537e-01 -6.84186280e-01 -6.11876070e-01
2.70075709e-01 9.60815787e-01 4.93641436e-01 -2.62348711e-01
1.97862178e-01 5.88816583e-01 -1.22562790e+00 -4.73022044e-01
4.54989970e-01 8.83270442e-01 6.37293994e-01 6.56552255e-01
4.40413475e-01 8.55471969e-01 -4.10447508e-01 1.63648371e-02
4.80939388e-01 -9.10002440e-02 -2.74463534e-01 3.73436391e-01
6.59284532e-01 -8.53007019e-01 -9.02954102e-01 1.44747972e+00
4.71837044e-01 -5.01864254e-01 4.22404438e-01 -1.50929749e+00
-5.15287519e-01 -4.11580771e-01 -4.89210248e-01 -3.87056842e-02
7.82239914e-01 4.09690775e-02 5.39857864e-01 -1.05398583e+00
1.31234407e+00 1.25455546e+00 6.38260722e-01 4.01474208e-01
-8.52206767e-01 -1.59167483e-01 -1.92408249e-01 4.55169618e-01
-1.34873116e+00 -2.65404701e-01 3.37964118e-01 -6.49005830e-01
1.09967422e+00 -6.13781452e-01 8.02220285e-01 4.43987519e-01
8.84141326e-01 3.97461355e-01 1.00816250e+00 -2.16624632e-01
2.11399198e-01 1.70708284e-01 -2.08714992e-01 1.01829159e+00
7.95207024e-01 5.77204585e-01 -7.03264117e-01 2.51075029e-01
1.04090929e+00 2.59024709e-01 -3.83513510e-01 -1.47267187e+00
-1.15564787e+00 8.45749080e-01 1.01774955e+00 -2.37319276e-01
-2.84538388e-01 2.06229955e-01 -1.01223262e-03 3.40532869e-01
1.94045603e-01 4.85571831e-01 -1.43070832e-01 -1.26027122e-01
-2.49417350e-01 1.90491423e-01 4.19795960e-01 1.39906907e+00
9.14724052e-01 1.37662485e-01 6.40826225e-01 2.41157800e-01
5.66663563e-01 1.09138846e+00 6.78490281e-01 -1.24368644e+00
2.97650427e-01 7.10010231e-01 4.47037429e-01 -6.99747503e-01
-8.83317173e-01 -1.77405208e-01 -5.93119681e-01 8.79134178e-01
2.03938738e-01 -2.38070130e-01 -1.15593398e+00 1.35819864e+00
3.69257659e-01 3.07487566e-02 4.20468032e-01 1.25373936e+00
8.64374936e-01 1.68983296e-01 -8.06901157e-01 -1.16713993e-01
1.19224763e+00 -1.00862288e+00 -8.89822602e-01 -9.63962615e-01
4.21443403e-01 -6.66745603e-01 1.15940370e-01 3.62986058e-01
-7.67103434e-01 -4.69918609e-01 -1.55493653e+00 -3.79002035e-01
-1.21725239e-01 1.13509838e-02 3.50950181e-01 4.83468056e-01
-1.12171090e+00 5.37999608e-02 -1.07736516e+00 -6.42780483e-01
-1.86633095e-01 5.20579994e-01 -6.28401279e-01 -2.51045763e-01
-9.72712874e-01 1.73463202e+00 2.01980621e-01 -2.22464994e-01
-5.73370576e-01 -4.06260252e-01 -1.41721928e+00 -4.86635178e-01
1.28806442e-01 -1.02590525e+00 1.31056607e+00 -3.01964849e-01
-1.72948742e+00 1.06859887e+00 -4.92214471e-01 -7.87248373e-01
6.52003288e-01 -3.15701127e-01 8.60821679e-02 4.09458697e-01
1.09796897e-01 8.61919880e-01 6.02178454e-01 -1.12214494e+00
-7.79240429e-01 -9.04008865e-01 1.55393690e-01 7.92745531e-01
5.90504169e-01 -7.62966812e-01 -5.44378996e-01 4.40028548e-01
1.05222118e+00 -1.48727131e+00 -4.38947588e-01 2.77989656e-01
3.24855559e-02 4.91332024e-01 8.06966305e-01 -3.37056339e-01
2.76168168e-01 -1.97148693e+00 2.39377931e-01 -2.78532773e-01
4.65126663e-01 -1.64795220e-01 4.66838002e-01 1.30304337e-01
4.17821079e-01 -9.89623666e-01 4.87707049e-01 -5.23838818e-01
-2.30668992e-01 4.00016904e-01 -1.25477418e-01 1.13078618e+00
-3.38383406e-01 8.50526452e-01 -1.00199103e+00 -2.12110460e-01
7.10372090e-01 5.68185210e-01 -2.87243843e-01 8.54002014e-02
1.68862864e-01 4.20317322e-01 4.66594920e-02 5.15527308e-01
7.88883269e-01 -4.15725023e-01 3.30996573e-01 1.04643926e-02
-2.41676852e-01 4.65026975e-01 -1.17135668e+00 1.85197282e+00
-2.29138091e-01 9.92779911e-01 -6.00681268e-02 -2.54584730e-01
9.73453343e-01 1.94551677e-01 4.07377452e-01 -1.00041270e+00
1.73672423e-01 5.60726583e-01 -5.06593846e-02 -1.30797457e-02
8.87698531e-01 -3.18112016e-01 5.93506880e-02 -2.21766949e-01
-1.12094983e-01 -1.00667775e+00 1.38399184e-01 2.62155347e-02
7.70042300e-01 1.14630215e-01 1.03806531e+00 8.00246745e-02
3.66118163e-01 2.10641891e-01 5.05281210e-01 3.59222591e-01
-4.83785838e-01 6.82306290e-01 1.24056377e-01 -5.54596066e-01
-1.35871398e+00 -1.19941151e+00 -3.95314664e-01 -2.12596729e-01
1.08012104e+00 9.74992961e-02 -9.52321291e-02 -1.40777677e-01
5.05885363e-01 -4.93111275e-02 -4.82358813e-01 -8.57318565e-02
-4.96224970e-01 -4.84731525e-01 4.04278375e-02 6.10407293e-01
5.42658746e-01 -3.34495127e-01 -1.20979238e+00 1.10289954e-01
-7.37521946e-02 -1.78027880e+00 -9.07115266e-02 2.46826530e-01
-9.85989153e-01 -1.16474795e+00 -7.81005681e-01 -7.94211924e-01
6.68650091e-01 9.78688896e-01 6.99735761e-01 -4.60979462e-01
7.42039010e-02 1.45072088e-01 1.52016923e-01 -4.77560878e-01
-2.70080626e-01 -2.22126067e-01 5.87576270e-01 -5.73852420e-01
3.93855512e-01 -1.46393493e-01 -6.32252157e-01 6.49834156e-01
-1.04812548e-01 1.12103231e-01 3.88980150e-01 8.73991430e-01
3.44912171e-01 -7.33253479e-01 1.74136773e-01 -2.51380146e-01
-2.54181355e-01 -7.49168992e-02 -1.30445027e+00 -5.10540605e-01
-9.37946498e-01 6.85787126e-02 9.90985110e-02 -3.44364755e-02
-6.49869502e-01 4.84422356e-01 2.85269171e-01 -4.83142287e-01
2.72521257e-01 1.00587822e-01 2.20437840e-01 -4.69885200e-01
8.35359454e-01 8.18169191e-02 7.00784564e-01 1.11428918e-02
1.03842758e-01 5.23727059e-01 7.07217395e-01 4.70076472e-01
7.56242931e-01 1.23549736e+00 6.69161439e-01 -9.87237811e-01
-3.88595521e-01 -8.91642570e-01 -8.00482750e-01 -3.58223505e-02
7.62278736e-01 -1.67108977e+00 -6.84139788e-01 5.58151484e-01
-1.23587513e+00 -5.32120354e-02 -1.78164959e-01 1.10944319e+00
-9.76475477e-01 5.54390192e-01 -6.31557822e-01 -3.73093903e-01
2.07206905e-02 -1.48697841e+00 1.26877677e+00 3.43586713e-01
-1.25467042e-02 -8.62674296e-01 4.96091396e-01 4.31134813e-02
-2.49016106e-01 -8.55595246e-02 1.14024200e-01 1.86065972e-01
-1.03111005e+00 -2.53762752e-01 -2.57160425e-01 -1.09276392e-01
1.74553003e-02 -5.41882455e-01 -9.72104907e-01 -5.83101690e-01
6.89672604e-02 -6.16565943e-02 5.67623556e-01 6.48375988e-01
-4.92185056e-01 4.04476404e-01 -6.41691029e-01 8.78305376e-01
1.56420100e+00 4.94164348e-01 7.06548512e-01 7.56210208e-01
7.35655487e-01 9.47530866e-02 8.78241122e-01 3.35688919e-01
7.86205947e-01 1.35315382e+00 9.59820271e-01 4.29239944e-02
-1.32268965e-01 -1.44547120e-01 2.88911492e-01 4.35857773e-01
1.40364215e-01 3.33706796e-01 -7.32605219e-01 5.79993904e-01
-1.95217919e+00 -4.98009831e-01 -1.77582875e-01 2.10778236e+00
2.42444590e-01 2.14562584e-02 -5.48894033e-02 -5.11343740e-02
3.35261792e-01 4.94883768e-02 -6.90596938e-01 -2.03539416e-01
-1.93492338e-01 -7.18383670e-01 1.07189727e+00 9.40748751e-01
-9.07986581e-01 8.76257122e-01 6.65075874e+00 -4.02201205e-01
-1.30503953e+00 5.35943396e-02 -4.04813439e-01 -6.47697523e-02
4.83668596e-01 -9.29018036e-02 -1.36679590e+00 1.63346305e-01
4.91874307e-01 -2.49065980e-01 1.89208582e-01 1.08970726e+00
-8.13297406e-02 -5.89140415e-01 -1.15338671e+00 1.44545472e+00
3.82766902e-01 -1.16094387e+00 -4.79294091e-01 5.19875646e-01
1.17558074e+00 7.58686364e-01 5.04496209e-02 -1.14484198e-01
3.18477631e-01 -4.75469798e-01 6.72174335e-01 6.04244880e-02
5.48013568e-01 -3.75950217e-01 9.98336077e-01 5.82619846e-01
-1.14908528e+00 -1.88935176e-01 -8.32823157e-01 -5.76893330e-01
4.69052255e-01 5.00277698e-01 -1.47543502e+00 4.99114633e-01
5.15474796e-01 1.02843153e+00 -2.63913393e-01 1.32043898e+00
-3.25971574e-01 -7.80938387e-01 -3.46202075e-01 2.32270464e-01
1.40360430e-01 -1.41721755e-01 6.80001259e-01 4.25156593e-01
5.10023534e-01 -2.18683764e-01 1.39978901e-01 2.70804942e-01
2.38539323e-01 -5.24345994e-01 -1.13809276e+00 8.14536512e-01
2.41085589e-01 6.90242767e-01 -2.12717786e-01 -3.57551157e-01
-5.55114150e-01 8.73675406e-01 3.58934999e-02 2.17586279e-01
-5.17896354e-01 -2.25886758e-02 1.03849292e+00 1.52557224e-01
4.33430225e-01 -6.33532226e-01 -5.22420824e-01 -1.35304797e+00
1.16679125e-01 -4.07877088e-01 -7.60820210e-02 -1.22688842e+00
-3.80608141e-01 5.63890755e-01 -2.11114556e-01 -1.83147800e+00
-9.67193961e-01 -1.08622658e+00 1.91759109e-01 6.23741388e-01
-1.89214802e+00 -6.68717384e-01 -7.70763755e-01 3.10667008e-01
3.64002049e-01 -1.41155824e-01 4.79125410e-01 1.42461970e-01
4.14401054e-01 -9.35505480e-02 2.46733978e-01 -2.18654409e-01
8.79568696e-01 -1.41330004e+00 7.19837725e-01 7.77696609e-01
-7.08599389e-02 2.64645576e-01 1.10976136e+00 -6.14224255e-01
-1.58941364e+00 -6.28824472e-01 1.01459897e+00 -7.54321873e-01
3.43222916e-01 -8.54417402e-03 -3.53505999e-01 1.00737309e+00
-3.52317393e-02 3.19508255e-01 -4.72756714e-01 -4.94432062e-01
-1.44690856e-01 -6.15855865e-02 -1.18149436e+00 4.45854396e-01
8.71788979e-01 -6.29901946e-01 -5.96262515e-01 1.15991183e-01
6.27571881e-01 -1.37756407e+00 -4.23319489e-01 3.22373509e-01
8.79387617e-01 -1.13801265e+00 1.12901711e+00 1.86114088e-01
-4.80832532e-02 -7.25617886e-01 -2.45044470e-01 -1.54203689e+00
-1.00798801e-01 -2.44508177e-01 -2.04517692e-01 1.74603201e-02
-1.15085833e-01 -8.56587410e-01 9.87489700e-01 -8.36817995e-02
-3.93268764e-02 -4.82356362e-02 -1.28108895e+00 -9.73143220e-01
-4.30070996e-01 8.91950428e-02 2.60710120e-01 4.27588165e-01
3.85106117e-01 6.08140707e-01 -4.98287857e-01 7.13497818e-01
9.29091215e-01 4.16773260e-02 1.19450355e+00 -1.32238364e+00
-1.25764430e-01 2.35435367e-01 -1.43516731e+00 -1.86574304e+00
-9.07070860e-02 -2.74438769e-01 2.90491283e-01 -1.61893916e+00
-3.24043864e-03 2.99747676e-01 5.28107584e-01 -4.45937276e-01
2.52914965e-01 5.71018934e-01 4.93140221e-01 4.85165119e-01
-4.21669126e-01 4.49866325e-01 1.43299794e+00 -4.95379753e-02
-1.54524371e-01 -8.08260143e-02 9.08415318e-02 1.06278682e+00
3.37221682e-01 -4.01580960e-01 -4.49099809e-01 -4.24584746e-01
4.26048160e-01 5.63431025e-01 6.34269953e-01 -1.25357854e+00
4.57193136e-01 1.99999493e-02 3.38050663e-01 -9.65869248e-01
8.89094055e-01 -1.00451350e+00 8.44437927e-02 1.20903587e+00
4.36461926e-01 3.71512026e-01 -1.13693193e-01 4.95313674e-01
-3.87296349e-01 -9.35778767e-02 1.08789456e+00 -6.97893873e-02
-1.20062697e+00 1.45685285e-01 -2.88312465e-01 -1.99461058e-01
8.87010098e-01 -5.52998364e-01 -9.15843189e-01 -5.77371299e-01
-4.79232997e-01 7.85313845e-02 1.16068017e+00 4.89053279e-01
7.74555624e-01 -1.36605608e+00 -2.19071507e-01 5.41908026e-01
4.49811727e-01 1.05850719e-01 -2.86462884e-02 1.01743042e+00
-1.12231636e+00 1.03420436e+00 -5.72635472e-01 -1.32572865e+00
-1.27780807e+00 5.97095907e-01 8.11093390e-01 -1.80489004e-01
-9.00972545e-01 2.46425688e-01 6.55327439e-01 -5.38907647e-01
6.84133684e-03 -6.01135790e-01 -7.19036534e-02 -4.62167591e-01
4.97164220e-01 4.39170480e-01 9.79611650e-02 -9.36928809e-01
-5.95680356e-01 1.07170367e+00 1.18233830e-01 -4.33367550e-01
6.66983366e-01 -9.00490642e-01 3.98773313e-01 4.21375185e-01
1.09787381e+00 -3.11843634e-01 -1.71260774e+00 -3.91406596e-01
-2.63303161e-01 -5.33888221e-01 -1.32955518e-02 -2.49804661e-01
-5.96386015e-01 8.83684158e-01 9.05321419e-01 -2.58518189e-01
4.67354447e-01 -1.45618930e-01 4.48459506e-01 6.70259893e-01
9.67105508e-01 -8.71882081e-01 3.67724411e-02 1.09242439e+00
6.57987952e-01 -1.60580862e+00 4.26099300e-01 -4.53138262e-01
-6.48656964e-01 9.19042945e-01 5.60665786e-01 -1.94550470e-01
6.31361604e-01 3.74082148e-01 6.21981680e-01 -1.33727510e-02
-7.04882860e-01 -2.99656808e-01 6.68331534e-02 8.96123827e-01
-1.25085592e-01 6.80944771e-02 -1.34085044e-02 -5.52087843e-01
-1.40449837e-01 1.09581023e-01 1.04838920e+00 1.04718542e+00
-8.73804986e-01 -6.18025482e-01 -5.50604820e-01 -4.27965969e-01
1.67668775e-01 2.91968048e-01 -2.33515501e-01 9.94458020e-01
-1.69106051e-01 7.72255301e-01 2.19757169e-01 -1.07433952e-01
4.74002689e-01 -3.00111502e-01 8.68033528e-01 -5.69936514e-01
1.16225198e-01 -5.67953438e-02 -7.01359883e-02 -7.00988710e-01
-3.02029729e-01 -6.30478442e-01 -1.55913723e+00 -2.64950573e-01
-5.55836260e-02 -2.38616154e-01 1.07565916e+00 9.01777744e-01
1.73265547e-01 1.98447570e-01 3.74613762e-01 -1.11138487e+00
-6.77078187e-01 -7.18188882e-01 -8.70575190e-01 9.15323570e-02
9.66317832e-01 -1.11051476e+00 -2.95812368e-01 -1.90846190e-01]
|
[7.937764644622803, -2.203077554702759]
|
9cba8817-77ec-48ae-a8fc-956cff03073d
|
depthwise-convolution-for-multi-agent
|
2203.02896
| null |
https://arxiv.org/abs/2203.02896v2
|
https://arxiv.org/pdf/2203.02896v2.pdf
|
Depthwise Convolution for Multi-Agent Communication with Enhanced Mean-Field Approximation
|
Multi-agent settings remain a fundamental challenge in the reinforcement learning (RL) domain due to the partial observability and the lack of accurate real-time interactions across agents. In this paper, we propose a new method based on local communication learning to tackle the multi-agent RL (MARL) challenge within a large number of agents coexisting. First, we design a new communication protocol that exploits the ability of depthwise convolution to efficiently extract local relations and learn local communication between neighboring agents. To facilitate multi-agent coordination, we explicitly learn the effect of joint actions by taking the policies of neighboring agents as inputs. Second, we introduce the mean-field approximation into our method to reduce the scale of agent interactions. To more effectively coordinate behaviors of neighboring agents, we enhance the mean-field approximation by a supervised policy rectification network (PRN) for rectifying real-time agent interactions and by a learnable compensation term for correcting the approximation bias. The proposed method enables efficient coordination as well as outperforms several baseline approaches on the adaptive traffic signal control (ATSC) task and the StarCraft II multi-agent challenge (SMAC).
|
['Daoyi Dong', 'Chunlin Chen', 'Zhi Wang', 'Donghan Xie']
|
2022-03-06
| null | null | null | null |
['smac-1', 'smac']
|
['playing-games', 'playing-games']
|
[-2.70042837e-01 4.06192839e-02 -1.34826452e-02 -2.90431362e-02
-8.29141319e-01 -5.29755712e-01 8.71107697e-01 6.62892417e-04
-7.05225825e-01 1.10568285e+00 2.90507942e-01 -7.83717334e-02
-4.28375721e-01 -6.26488090e-01 -9.05660570e-01 -8.73406529e-01
-4.16877717e-01 6.73192084e-01 3.40176821e-01 -6.51542425e-01
-1.71578079e-01 6.52508616e-01 -1.07371068e+00 9.75010246e-02
9.69081044e-01 8.06069434e-01 1.27938583e-01 8.52556407e-01
4.25269127e-01 1.43505287e+00 -9.12380278e-01 3.12729925e-01
4.24520463e-01 -2.83397406e-01 -6.10235870e-01 1.71459913e-01
1.65084854e-01 -4.94685799e-01 -5.67121625e-01 7.96638548e-01
6.47580206e-01 6.31435513e-01 2.76625693e-01 -1.63343263e+00
-3.32484812e-01 8.29834104e-01 -7.09236443e-01 2.39293948e-01
4.97215427e-03 8.01475763e-01 1.01618791e+00 -1.08444020e-02
4.37594712e-01 1.44782770e+00 4.76857960e-01 5.35296679e-01
-1.19308734e+00 -6.77832544e-01 8.04051399e-01 3.01387966e-01
-1.03537989e+00 -2.66818941e-01 6.72871470e-01 -1.99357957e-01
9.59145665e-01 -7.69178569e-02 5.31044185e-01 1.12529171e+00
-3.26747745e-02 7.91910410e-01 8.39522481e-01 -1.17922902e-01
3.99831563e-01 -3.12290281e-01 -2.80708492e-01 9.20276701e-01
-2.27372035e-01 3.83137941e-01 -3.97828877e-01 -3.94962728e-01
1.08274865e+00 4.47295122e-02 -1.84450284e-01 -6.58088401e-02
-1.42897701e+00 8.13327670e-01 6.75330997e-01 8.20037574e-02
-8.59630167e-01 8.11373711e-01 4.04262125e-01 6.32892311e-01
3.04844022e-01 5.97050428e-01 -3.80142421e-01 -1.67372562e-02
-1.80362061e-01 5.38511753e-01 5.67891717e-01 6.54253483e-01
7.86553741e-01 4.66661811e-01 -3.54264110e-01 6.56785190e-01
3.27449962e-02 6.52482092e-01 9.35063139e-02 -1.65851116e+00
6.48477912e-01 4.16339636e-01 5.40022075e-01 -8.89009178e-01
-6.14457667e-01 -4.92116898e-01 -8.41015279e-01 6.32886708e-01
4.83653069e-01 -7.63647556e-01 -2.84453511e-01 2.01886082e+00
5.26095808e-01 6.75344884e-01 2.87404567e-01 1.04327619e+00
8.56373738e-03 6.12182736e-01 -1.53869748e-01 -1.65458828e-01
9.85341907e-01 -1.44373560e+00 -5.61995983e-01 -1.75703168e-01
7.04839945e-01 -2.04527020e-01 6.55207336e-01 1.78992197e-01
-9.57526922e-01 -3.32460016e-01 -8.31484795e-01 4.72076476e-01
-2.19262689e-01 -4.67099212e-02 6.08922362e-01 -1.45160481e-01
-1.00910938e+00 5.11675835e-01 -1.00640917e+00 3.44324447e-02
4.47716922e-01 7.60431647e-01 -1.23477735e-01 1.15949236e-01
-1.19282568e+00 8.90165150e-01 -4.83115017e-02 -1.57247891e-03
-1.31212306e+00 -9.09453392e-01 -7.22540259e-01 -3.52049358e-02
9.35070097e-01 -6.60721183e-01 1.37109160e+00 -1.16007435e+00
-2.02879429e+00 -2.39414573e-01 3.60097528e-01 -8.88817012e-01
5.13229430e-01 -8.45487490e-02 -2.77522653e-01 3.85901392e-01
9.57592130e-02 7.09178448e-01 7.45154917e-01 -1.30491948e+00
-1.05005455e+00 -3.68821174e-02 6.65334940e-01 5.10647774e-01
6.55042529e-02 -3.51134777e-01 1.15708217e-01 -6.86283648e-01
-8.85282397e-01 -9.80024040e-01 -7.31311738e-01 -2.42184728e-01
5.21903262e-02 -4.10173088e-01 1.04132771e+00 -2.70105720e-01
7.00505316e-01 -2.05566168e+00 4.45916116e-01 3.52104425e-01
5.13429523e-01 3.02306324e-01 -8.30323994e-01 4.79740947e-01
3.40052247e-01 -5.22380948e-01 1.66178152e-01 -4.01303440e-01
3.03431660e-01 6.56063020e-01 -3.36935192e-01 5.91937840e-01
-5.63792232e-03 9.51213598e-01 -1.08755100e+00 -1.16347775e-01
2.17351183e-01 4.98362958e-01 -6.85127378e-01 2.60655552e-01
-6.69558108e-01 9.99494016e-01 -7.51989961e-01 7.69653022e-02
3.81458193e-01 -2.00412095e-01 1.43875450e-01 7.95297325e-02
-2.31913671e-01 6.56172484e-02 -1.27138627e+00 1.55424607e+00
-8.93544674e-01 3.92546296e-01 7.61055529e-01 -1.10632050e+00
5.60121357e-01 4.85246927e-01 9.67930853e-01 -9.11303461e-01
1.23680346e-01 3.51308100e-02 2.44471654e-01 -1.62706390e-01
1.51892364e-01 2.83892334e-01 -1.62750065e-01 8.44565094e-01
-7.17674941e-02 6.33189604e-02 3.09638232e-01 3.17672610e-01
1.50428569e+00 -2.08295017e-01 5.48300222e-02 -7.88738132e-02
6.67011738e-01 -1.79158859e-02 7.87049174e-01 1.05677390e+00
-3.49527717e-01 -1.74779296e-01 4.08710361e-01 -6.33904278e-01
-7.59839892e-01 -6.54219449e-01 8.16266835e-01 1.42423666e+00
1.65984213e-01 -2.02298537e-01 -5.98075330e-01 -8.42558563e-01
2.12603852e-01 3.96597117e-01 -5.58275998e-01 -1.54109016e-01
-1.05014455e+00 -5.98801374e-01 5.60587049e-01 4.94234115e-01
7.21661627e-01 -1.10762691e+00 -6.90004170e-01 5.69043040e-01
-1.53104246e-01 -1.43503344e+00 -9.71563160e-01 -5.10358019e-03
-1.62617072e-01 -1.12263227e+00 -4.56568718e-01 -5.53947628e-01
5.57984591e-01 1.91412553e-01 7.26044416e-01 1.81088001e-01
-1.13628618e-02 8.09503436e-01 -1.71721622e-01 -5.87081984e-02
-5.04459918e-01 1.69354185e-01 3.22386503e-01 3.24505061e-01
-2.95831591e-01 -8.58096063e-01 -5.06164610e-01 3.85409743e-01
-6.70899868e-01 -1.62194949e-02 6.86216533e-01 9.72780883e-01
2.14716330e-01 1.21357590e-01 8.64264131e-01 -5.68793893e-01
8.80106628e-01 -3.20886731e-01 -1.09135246e+00 1.89030886e-01
-1.71492413e-01 2.69887775e-01 1.22864902e+00 -8.49004567e-01
-9.79014099e-01 -7.19614401e-02 2.51221448e-01 -3.84133875e-01
-1.56005278e-01 3.77535671e-02 2.14158162e-01 -3.71772587e-01
4.49511230e-01 -5.59786055e-03 3.83922964e-01 -6.97384775e-02
5.17059267e-01 3.60482693e-01 4.48194444e-01 -9.22653854e-01
7.82394946e-01 6.05668545e-01 2.17096701e-01 -7.03924060e-01
-8.20083499e-01 -1.41726375e-01 -2.94138670e-01 -3.25048923e-01
6.22123837e-01 -9.90510464e-01 -1.63933969e+00 4.41321969e-01
-1.34333301e+00 -1.12590468e+00 -5.10524035e-01 6.24839723e-01
-7.64681756e-01 1.62561014e-01 -7.52879858e-01 -5.90728581e-01
-1.68625370e-01 -1.30240679e+00 8.39700937e-01 1.30318195e-01
2.12387905e-01 -1.33139992e+00 2.57056385e-01 2.17577606e-01
7.34955788e-01 1.02631204e-01 4.94785756e-01 -4.56109375e-01
-8.42215359e-01 2.30106443e-01 -2.02933867e-02 1.47196397e-01
5.09218723e-02 -5.15193105e-01 -3.85579050e-01 -6.24373198e-01
-3.60413551e-01 -4.53356326e-01 6.99342668e-01 4.63475078e-01
7.54773438e-01 -6.92146003e-01 -2.91807204e-01 3.57023686e-01
1.07356656e+00 3.11364233e-03 4.12464067e-02 2.39152327e-01
6.92622781e-01 4.91578192e-01 4.02931720e-01 7.69360304e-01
9.20242965e-01 7.31739044e-01 6.40389204e-01 -2.25114614e-01
-1.47915334e-01 -1.17162347e-01 6.15359604e-01 5.24600685e-01
-1.83148518e-01 -1.98405847e-01 -5.88764846e-01 2.27020472e-01
-2.48331785e+00 -1.02381349e+00 2.64849305e-01 1.78275013e+00
6.37526453e-01 -1.23276606e-01 4.17995483e-01 -4.41164553e-01
5.18714607e-01 2.51168132e-01 -7.44143307e-01 -5.45080900e-02
-1.17858745e-01 -2.34926492e-01 7.74667859e-01 1.13973629e+00
-1.04536211e+00 9.35687065e-01 5.92174292e+00 8.09390426e-01
-9.68805492e-01 2.73799717e-01 2.53909945e-01 -1.91311628e-01
1.42448857e-01 -1.42027035e-01 -6.59723878e-01 3.11090112e-01
8.68014097e-01 2.23924309e-01 1.18796504e+00 4.12397474e-01
6.22576535e-01 -4.41995263e-02 -9.15276825e-01 6.79232419e-01
-1.86106786e-01 -1.54624641e+00 -1.47719279e-01 2.07203433e-01
9.68287468e-01 3.15033942e-01 4.72728238e-02 4.98199403e-01
1.21281171e+00 -7.49867976e-01 4.92201269e-01 4.50215578e-01
2.40125582e-01 -7.77280509e-01 5.26148140e-01 5.27892947e-01
-1.30422938e+00 -6.22563601e-01 -7.84284696e-02 -3.68425786e-01
1.76738560e-01 -2.93976627e-02 -5.51205695e-01 4.76435006e-01
2.72932321e-01 8.13649178e-01 -1.16283283e-01 5.66289902e-01
-1.70593917e-01 3.35175753e-01 -5.68319619e-01 2.59373407e-03
7.73162246e-01 -2.65350610e-01 7.64907658e-01 7.02924371e-01
-2.31619373e-01 1.86271146e-01 8.34033072e-01 7.06184626e-01
-2.90795472e-02 -3.92942280e-01 -4.26408052e-01 1.92269787e-01
5.58885872e-01 1.24988353e+00 -1.75694302e-01 -3.52108717e-01
-5.99770010e-01 8.08983862e-01 7.59595931e-01 9.16964173e-01
-9.47445750e-01 -1.41449064e-01 1.04760218e+00 -2.17016354e-01
3.77970904e-01 -4.66114730e-01 3.45235050e-01 -9.89469349e-01
-2.21014440e-01 -1.03904104e+00 3.83403838e-01 -2.55071312e-01
-1.22985685e+00 5.25749147e-01 -2.96756625e-01 -1.06748796e+00
-4.68845904e-01 -2.77970582e-01 -8.15931857e-01 4.10630584e-01
-1.79822946e+00 -1.30003512e+00 5.17797656e-03 8.94878030e-01
4.03981060e-01 -6.11481667e-01 5.00256777e-01 4.14070874e-01
-6.84451818e-01 2.85640270e-01 1.71403587e-02 2.37149820e-01
5.81770837e-01 -1.19950187e+00 -1.61566794e-01 5.35798609e-01
-1.53657272e-01 1.96212575e-01 5.90485930e-01 -2.31782123e-01
-1.40351331e+00 -1.30754995e+00 1.38766006e-01 -4.29980420e-02
9.99082625e-01 -3.11743170e-01 -5.48416913e-01 1.01338422e+00
5.14232278e-01 4.01102036e-01 1.93225905e-01 -2.06108034e-01
-2.44927377e-01 -5.02951503e-01 -8.49002600e-01 8.01760197e-01
6.77393436e-01 -2.69934058e-01 -1.00653514e-01 3.80361319e-01
8.82376790e-01 -2.73116320e-01 -7.23451316e-01 -3.01704071e-02
1.33630618e-01 -4.53808337e-01 9.22296822e-01 -5.87999344e-01
-2.97602475e-01 -5.68571389e-01 3.36678065e-02 -1.94796109e+00
-1.01538375e-01 -1.28742802e+00 -1.23551056e-01 9.15799916e-01
2.97500134e-01 -1.00491190e+00 5.31165004e-01 2.54983097e-01
-1.66907385e-01 -5.33045411e-01 -1.05450535e+00 -6.96132958e-01
5.14024682e-02 -3.81201468e-02 6.62312746e-01 7.94778883e-01
8.24419111e-02 4.66571897e-01 -6.71403348e-01 5.66709101e-01
7.54836857e-01 6.63294317e-03 9.96670783e-01 -7.81741083e-01
-8.04722190e-01 -5.13880193e-01 -4.79115807e-02 -1.42799461e+00
6.81243777e-01 -6.42445862e-01 9.05799791e-02 -1.20955598e+00
-2.72183537e-01 -5.78554511e-01 -9.58760306e-02 4.54738259e-01
1.19784623e-01 -2.55428016e-01 4.79685336e-01 1.45600721e-01
-1.30307424e+00 9.43091929e-01 1.48339021e+00 -2.74930418e-01
-3.59444499e-01 1.70862332e-01 -4.25331295e-01 6.13088608e-01
8.90967548e-01 -3.19919795e-01 -3.74002695e-01 -7.83270419e-01
-5.34803830e-02 3.74994457e-01 6.79972708e-01 -8.67527544e-01
7.48008668e-01 -5.02429128e-01 -2.44250804e-01 -1.35642916e-01
4.64740068e-01 -9.10133302e-01 -3.78784776e-01 5.81400931e-01
-6.92709923e-01 1.87384307e-01 -7.28451014e-02 8.18698645e-01
-1.73722863e-01 3.81081760e-01 8.50038111e-01 -1.63875565e-01
-4.91331428e-01 4.81938601e-01 -7.45498180e-01 2.04012811e-01
1.15725911e+00 6.43070102e-01 -6.20915174e-01 -6.94346845e-01
-5.83760977e-01 1.22796345e+00 3.76050062e-02 1.12438135e-01
4.67918187e-01 -1.30440092e+00 -7.38505721e-01 9.92503613e-02
-3.25754374e-01 -4.15882990e-02 3.16663593e-01 1.15915215e+00
-1.83460400e-01 2.82536715e-01 -2.01014861e-01 -1.93741664e-01
-8.41365159e-01 2.65316427e-01 8.22213471e-01 -6.38772309e-01
-7.54962027e-01 3.96754414e-01 1.06088303e-01 -7.44936168e-01
4.98062521e-01 -3.19113523e-01 -2.13549152e-01 -3.88228208e-01
6.56646609e-01 6.81718767e-01 -5.09210467e-01 -4.48696315e-01
-7.69197121e-02 3.70424807e-01 -1.79531515e-01 -2.52248585e-01
1.53996944e+00 -3.17666978e-01 5.68545163e-02 5.06586116e-03
8.35745394e-01 -1.77793413e-01 -2.09352660e+00 -5.89822888e-01
-2.32188478e-01 -2.12550908e-02 9.73198786e-02 -8.40770006e-01
-1.31992936e+00 5.15991032e-01 1.01703353e-01 1.26658916e-01
7.52158344e-01 -8.50644708e-03 9.17157352e-01 7.21314311e-01
3.25003892e-01 -1.39014888e+00 4.15465623e-01 7.59392798e-01
8.72111142e-01 -1.25899565e+00 -3.37950617e-01 -7.03744739e-02
-8.46988916e-01 1.01728308e+00 8.38402033e-01 -5.10420978e-01
5.61049581e-01 5.72536170e-01 1.07536264e-01 -1.22066475e-01
-1.37201345e+00 -4.26631391e-01 -3.45536143e-01 8.84432495e-01
-3.38208318e-01 4.14735638e-02 2.60140866e-01 3.06973517e-01
3.66399288e-01 -2.51888007e-01 6.42260909e-01 7.60370970e-01
-3.70666146e-01 -9.99294937e-01 -1.90617457e-01 5.92489615e-02
-2.08866391e-02 3.32839161e-01 -6.18758909e-02 7.29202926e-01
-2.34930739e-02 1.13068497e+00 2.37569377e-01 -6.67279214e-02
4.64120269e-01 -6.24845147e-01 3.04145128e-01 -2.60606915e-01
-7.52502561e-01 1.78494439e-01 -3.66610400e-02 -8.05488110e-01
-5.41018188e-01 -4.95082110e-01 -1.62322497e+00 -4.93515819e-01
7.61282146e-02 3.03119451e-01 1.04873359e-01 1.17116916e+00
7.41234481e-01 9.07145977e-01 8.22860539e-01 -9.87840772e-01
-9.75694180e-01 -8.15861940e-01 -5.26515841e-01 2.84714639e-01
9.59838212e-01 -7.66506493e-01 -3.77890319e-01 -4.96333241e-01]
|
[3.769620656967163, 1.9734346866607666]
|
4b1ca1d3-61f3-4f0a-b772-29742d3f420b
|
sms-spam-detection-through-skip-gram
| null | null |
https://aclanthology.org/2021.findings-acl.367
|
https://aclanthology.org/2021.findings-acl.367.pdf
|
SMS Spam Detection Through Skip-gram Embeddings and Shallow Networks
| null |
['Ivan Rizzo Guilherme', 'João Paulo Papa', 'Daniel Carlos Guimarães Pedronette', 'Gustavo Sousa']
| null | null | null | null |
findings-acl-2021-8
|
['spam-detection']
|
['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]
|
[-7.40623664855957, 3.6529669761657715]
|
59a44fce-6501-4a7f-b922-dd468816a827
|
pod-positional-dependency-based-word
|
1911.03785
| null |
https://arxiv.org/abs/1911.03785v2
|
https://arxiv.org/pdf/1911.03785v2.pdf
|
PoD: Positional Dependency-Based Word Embedding for Aspect Term Extraction
|
Dependency context-based word embedding jointly learns the representations of word and dependency context, and has been proved effective in aspect term extraction. In this paper, we design the positional dependency-based word embedding (PoD) which considers both dependency context and positional context for aspect term extraction. Specifically, the positional context is modeled via relative position encoding. Besides, we enhance the dependency context by integrating more lexical information (e.g., POS tags) along dependency paths. Experiments on SemEval 2014/2015/2016 datasets show that our approach outperforms other embedding methods in aspect term extraction.
|
['Chenguang Wang', 'Ming Zhang', 'Yichun Yin']
|
2019-11-09
| null |
https://aclanthology.org/2020.coling-main.150
|
https://aclanthology.org/2020.coling-main.150.pdf
|
coling-2020-8
|
['aspect-term-extraction-and-sentiment']
|
['natural-language-processing']
|
[-2.35430017e-01 -8.40353072e-02 -8.93215299e-01 -3.97897601e-01
-4.52526122e-01 -5.24345577e-01 7.75570929e-01 7.41716921e-01
-7.37226963e-01 5.13266385e-01 1.04891050e+00 -3.76912594e-01
-1.35547206e-01 -1.03036964e+00 -1.28921688e-01 -5.43452024e-01
-1.65432394e-01 4.84004170e-02 5.42390458e-02 -3.34378749e-01
1.92281783e-01 1.76878765e-01 -1.04536343e+00 1.39597684e-01
5.43359101e-01 5.83861768e-01 1.32561699e-01 2.21253648e-01
-5.39493144e-01 3.33002359e-01 -5.54971814e-01 -4.70197082e-01
-3.05099487e-01 6.07823059e-02 -6.65101349e-01 -3.26299399e-01
-1.69758946e-01 -6.81975260e-02 -4.81997609e-01 7.87194729e-01
4.30262357e-01 4.70742285e-02 6.57912970e-01 -8.35108876e-01
-9.76948023e-01 8.90950918e-01 -6.72055125e-01 4.68184650e-01
4.34170753e-01 -1.10068113e-01 1.96044779e+00 -1.16244090e+00
6.79229617e-01 9.37473893e-01 4.90895092e-01 1.24353714e-01
-9.27419841e-01 -2.63189435e-01 5.70938170e-01 5.98216176e-01
-1.30135691e+00 2.26476699e-01 1.08303368e+00 -1.89659968e-01
1.68125427e+00 8.97605717e-02 1.03303611e+00 1.17461598e+00
3.98632824e-01 8.54997456e-01 6.95578814e-01 -6.10882521e-01
-1.67308122e-01 -1.48392364e-01 7.92494178e-01 3.51395398e-01
7.80535400e-01 3.19892429e-02 -3.92915756e-01 -3.12061340e-01
3.65353078e-01 1.34695128e-01 -1.54892042e-01 1.47620767e-01
-9.53231573e-01 1.02722526e+00 3.76759470e-01 6.52367055e-01
-4.15966868e-01 2.92929828e-01 4.32890981e-01 -9.84660313e-02
6.46546841e-01 3.55077386e-01 -8.40820372e-01 -1.11517124e-02
-4.98911679e-01 2.31497422e-01 5.31701684e-01 1.16637254e+00
7.37686098e-01 2.51004659e-03 -4.22573864e-01 7.57503450e-01
8.12380254e-01 5.16134679e-01 5.24161577e-01 -1.72471553e-01
4.48062629e-01 9.07312512e-01 -3.44683319e-01 -9.51423824e-01
-4.14968282e-01 -3.17619652e-01 -5.09102166e-01 -4.94862348e-01
-4.49666560e-01 -1.02935195e-01 -8.41078103e-01 1.71538353e+00
4.11444187e-01 1.17960267e-01 2.64022183e-02 5.45401990e-01
1.18654644e+00 7.40830243e-01 2.87685126e-01 -2.63071716e-01
2.01841688e+00 -9.10753846e-01 -1.27476013e+00 -5.24912119e-01
6.72213912e-01 -9.09614444e-01 9.50751424e-01 -2.67374426e-01
-6.74443901e-01 -7.30836838e-02 -1.26182556e+00 -2.32965827e-01
-9.13093030e-01 2.55596451e-02 1.04090559e+00 4.30270880e-01
-4.76135463e-01 1.49491325e-01 -7.40143478e-01 -8.76460373e-02
2.58737952e-01 1.34987190e-01 -5.19866049e-01 -3.00500780e-01
-1.65892220e+00 7.92200387e-01 5.46066165e-01 -1.77565798e-01
-1.69467628e-01 -6.33685470e-01 -1.52997887e+00 2.01124802e-01
4.47699308e-01 -7.09810138e-01 7.97307014e-01 -7.54152909e-02
-9.16138411e-01 5.38385212e-01 -4.86066669e-01 -2.44818389e-01
-6.56919956e-01 -4.13557738e-01 -6.38933539e-01 4.23814654e-02
3.73824686e-02 2.08178222e-01 7.36956060e-01 -9.41087246e-01
-4.07598644e-01 -4.41290051e-01 6.42227888e-01 1.54253304e-01
-7.61465311e-01 3.03221226e-01 -7.10988045e-01 -1.06839776e+00
-2.60538850e-02 -6.92569017e-01 -3.31646979e-01 -3.86323690e-01
-2.77160048e-01 -6.57080114e-01 7.59050012e-01 -8.25657904e-01
2.05439591e+00 -2.05324364e+00 2.13511735e-01 3.52771282e-02
1.73209116e-01 2.15699598e-01 -2.13896155e-01 8.06205988e-01
-6.84481338e-02 4.44012076e-01 -3.33175510e-01 -3.35742682e-01
2.44691044e-01 7.13687181e-01 -2.19876632e-01 3.40629704e-02
4.40628111e-01 1.32292449e+00 -9.28370953e-01 -6.86421394e-01
-7.28047565e-02 8.95773947e-01 -6.70539558e-01 2.64230426e-02
-1.73463985e-01 -2.78217554e-01 -5.75463712e-01 7.69075692e-01
6.04641557e-01 -1.37289194e-02 5.50980926e-01 -3.94775063e-01
9.43079665e-02 8.81663859e-01 -7.92930722e-01 1.60443938e+00
-8.71154904e-01 5.80696523e-01 -4.58260566e-01 -4.31148767e-01
6.71483696e-01 5.32264471e-01 3.07782620e-01 -5.71463168e-01
1.45059720e-01 -2.84812480e-01 -1.12979509e-01 -4.94855672e-01
8.26399803e-01 -1.74227536e-01 -3.66440326e-01 6.57676637e-01
3.94660503e-01 1.72537938e-02 4.12127942e-01 3.33696783e-01
1.08565569e+00 7.78307915e-02 8.59210968e-01 -1.47354066e-01
6.57939196e-01 -3.66809785e-01 7.91565955e-01 1.32558914e-03
-5.25737852e-02 4.48434085e-01 6.50506735e-01 -3.73147488e-01
-6.47344291e-01 -1.04794455e+00 -2.39195034e-01 1.05690122e+00
-9.12107900e-02 -1.53358436e+00 -1.10819057e-01 -1.27334237e+00
-3.36611010e-02 7.91885376e-01 -9.12524343e-01 -2.46207505e-01
-1.04808569e+00 -1.00371456e+00 3.35118622e-01 9.34399366e-01
-5.69860823e-03 -1.13597965e+00 -8.51851329e-02 4.26952630e-01
-8.31820294e-02 -1.22857273e+00 -6.67962372e-01 3.17458451e-01
-7.48874366e-01 -9.94798422e-01 -1.08172670e-01 -8.29597890e-01
4.69279468e-01 1.34722903e-01 1.42832637e+00 2.30417550e-01
-2.32987806e-01 3.80287766e-01 -8.14115465e-01 -2.90000647e-01
5.25763273e-01 2.32500315e-01 -1.74473867e-01 -5.61350346e-01
1.00654161e+00 -7.36915112e-01 -5.56020856e-01 -1.57896489e-01
-1.20724893e+00 -4.40449655e-01 7.14095652e-01 9.13816571e-01
8.12968433e-01 -1.72521442e-01 3.20367575e-01 -1.05902994e+00
8.34221423e-01 -6.55455887e-01 -1.37156427e-01 1.28433242e-01
-1.07244098e+00 2.61698425e-01 1.35513067e-01 -3.28015029e-01
-1.03265679e+00 -5.83823264e-01 -5.06411970e-01 2.17317969e-01
9.77859646e-02 1.06608701e+00 -5.61754167e-01 4.92191523e-01
-2.32168287e-02 9.22230035e-02 -8.65235448e-01 -7.05076337e-01
7.29017437e-01 4.91482109e-01 -1.04328610e-01 -6.65712774e-01
7.47286499e-01 2.42198944e-01 -8.59150663e-02 -7.99524367e-01
-8.90391648e-01 -4.46043462e-01 -6.77098036e-01 3.68492246e-01
1.03425121e+00 -7.59764433e-01 3.33930515e-02 -2.28751257e-01
-1.53312111e+00 5.81908643e-01 -4.82467562e-01 6.17429852e-01
7.32516870e-02 3.32530588e-01 -8.86626363e-01 -4.29451823e-01
-5.48047066e-01 -8.38371933e-01 9.96334910e-01 1.79019153e-01
-4.14841980e-01 -1.45161140e+00 4.93767083e-01 -1.07090026e-01
3.85208368e-01 1.40887797e-01 1.36781859e+00 -6.59394681e-01
-3.96275401e-01 -4.37203884e-01 -2.29570419e-01 1.59236357e-01
4.14319843e-01 -4.78222258e-02 -7.65124559e-01 5.22624813e-02
-2.60630757e-01 3.47297221e-01 1.10741663e+00 1.15234211e-01
5.29817700e-01 -4.54701632e-01 -3.23474735e-01 5.48223972e-01
1.49835908e+00 -3.34287174e-02 5.30678689e-01 4.12337840e-01
9.75549638e-01 4.88971323e-01 7.25951672e-01 5.01541257e-01
7.65020847e-01 6.27114356e-01 1.77630469e-01 4.67166692e-01
-3.23621631e-01 -2.63184667e-01 3.67478549e-01 1.41125047e+00
-9.01224986e-02 -1.36284590e-01 -6.30785823e-01 8.75920594e-01
-1.53924918e+00 -5.22297204e-01 -1.35111794e-01 1.66827023e+00
1.04533017e+00 2.81556219e-01 -3.45666170e-01 1.88255847e-01
2.24196970e-01 9.36126590e-01 7.69559070e-02 -6.86156571e-01
-2.94177800e-01 6.53854489e-01 2.00707391e-01 7.11103618e-01
-9.68465090e-01 1.18021429e+00 5.82701397e+00 8.07205200e-01
-4.13083494e-01 4.74753618e-01 -2.14842260e-01 5.70954904e-02
-1.11750305e+00 3.75033289e-01 -9.30302501e-01 1.98952794e-01
7.06012726e-01 -5.56201816e-01 -1.46948770e-01 8.05658877e-01
-3.22385877e-01 2.30985478e-01 -8.01927865e-01 6.41630113e-01
3.30812424e-01 -1.07980442e+00 4.07471538e-01 1.59276739e-01
5.77665746e-01 -2.48258337e-01 -8.27868506e-02 5.53813875e-01
-7.68906847e-02 -8.41598511e-01 3.28422152e-02 2.27528781e-01
5.69880068e-01 -1.06090450e+00 1.06420338e+00 -1.09155364e-01
-1.83619821e+00 2.53924161e-01 -3.88005018e-01 2.18944270e-02
5.30065060e-01 9.34414804e-01 -3.44065964e-01 9.84609127e-01
3.29855353e-01 1.11031246e+00 -6.90001905e-01 5.56962371e-01
-1.37614405e+00 6.75060272e-01 -3.24732810e-02 -1.74293056e-01
2.27408275e-01 -2.86740392e-01 7.36230135e-01 1.55103397e+00
-5.17248549e-02 4.06583026e-02 -1.14024282e-01 4.38852549e-01
-4.62994017e-02 3.87192309e-01 -7.47056901e-01 -3.99952710e-01
3.76433551e-01 1.40565205e+00 -3.75582516e-01 -2.05450103e-01
-8.15459013e-01 7.42616296e-01 3.52569640e-01 3.39287192e-01
-7.24474370e-01 -5.26970804e-01 1.20537782e+00 -2.66591996e-01
7.74987876e-01 -6.02466762e-01 -1.99241266e-01 -1.25661147e+00
2.29839072e-01 -2.87317306e-01 6.06198847e-01 -3.57386053e-01
-1.23548162e+00 8.55830848e-01 2.15432093e-01 -1.11074495e+00
-3.07765901e-01 -6.63259804e-01 -8.38492453e-01 7.69499183e-01
-1.98884189e+00 -1.58545446e+00 3.41713220e-01 7.42112398e-02
5.53882539e-01 1.16529785e-01 1.17463291e+00 5.22490263e-01
-5.76689363e-01 7.01305985e-01 -5.74326396e-01 2.83145279e-01
4.44105178e-01 -1.39332473e+00 5.61457813e-01 7.13657320e-01
6.91310048e-01 1.17241359e+00 4.37523574e-01 -6.99725151e-01
-1.47448409e+00 -1.04956603e+00 1.70430553e+00 -4.95620847e-01
8.07288826e-01 -2.68375635e-01 -9.45289493e-01 9.01496768e-01
6.60526454e-01 6.24902397e-02 1.30851424e+00 5.30564427e-01
-9.37565684e-01 1.21761255e-01 -8.09182763e-01 7.71838188e-01
1.00009012e+00 -7.52658367e-01 -1.26475203e+00 3.05455122e-02
1.49800062e+00 1.13574095e-01 -8.81226122e-01 6.75764322e-01
6.62212133e-01 -2.29770407e-01 1.11188722e+00 -7.65078723e-01
4.45091844e-01 -2.16340289e-01 -5.28989971e-01 -1.27658570e+00
-4.03412461e-01 -1.53731316e-01 -9.21474993e-01 1.68868041e+00
7.62286365e-01 -5.35179913e-01 3.50550830e-01 1.29733486e-02
1.13664448e-01 -1.24360788e+00 -8.78457546e-01 -7.02377737e-01
1.46365508e-01 -7.59231269e-01 9.34276283e-01 9.07777965e-01
1.48729637e-01 8.92585576e-01 -1.20407782e-01 1.49120063e-01
3.38554770e-01 3.54276419e-01 4.05181460e-02 -8.46498013e-01
-3.91712517e-01 -3.54448378e-01 -5.06426752e-01 -8.35760295e-01
4.31419879e-01 -9.29884255e-01 -3.90075356e-01 -1.81527197e+00
1.85649738e-01 1.40211843e-02 -9.05981600e-01 3.56691450e-01
-8.27632308e-01 1.93665922e-02 -6.65135980e-02 -2.96854436e-01
-5.83810449e-01 1.03117049e+00 9.32639182e-01 -4.15075213e-01
1.09568320e-01 -2.91560858e-01 -7.53007174e-01 6.91856921e-01
8.99899840e-01 -7.82125711e-01 -2.06607759e-01 -6.24997556e-01
7.24728227e-01 -4.06655461e-01 -3.62316787e-01 -4.53774869e-01
-4.01429124e-02 -8.61016065e-02 1.16554491e-01 -8.17994833e-01
4.10517812e-01 -9.20340419e-01 -5.12354910e-01 1.29782543e-01
-1.00971684e-01 4.41061884e-01 1.96659103e-01 7.30290532e-01
-4.20081258e-01 -3.94003898e-01 -3.35600562e-02 1.25575557e-01
-6.83608830e-01 4.46611196e-01 -3.58723730e-01 2.52075255e-01
6.94016159e-01 5.09496212e-01 -2.92553306e-01 1.84964482e-02
-5.20835638e-01 2.66792297e-01 1.18234813e-01 7.34376788e-01
8.66414011e-01 -1.86406910e+00 -5.29044271e-01 2.34421715e-01
7.24639893e-01 -3.12509209e-01 4.77887094e-02 5.41746855e-01
-3.23194638e-02 3.53761673e-01 3.77108455e-01 -3.53760980e-02
-1.47064567e+00 7.10647345e-01 -4.90917325e-01 -7.39305854e-01
-5.45221746e-01 8.05858552e-01 -1.38395563e-01 -4.76063043e-01
-7.31614158e-02 -5.20212352e-01 -9.11317110e-01 4.04431552e-01
8.43926787e-01 -2.84194294e-02 -1.03546008e-01 -6.79747581e-01
-6.62999809e-01 8.09847236e-01 -2.67312348e-01 -4.38296288e-01
1.46719563e+00 -9.22226254e-03 -4.79372501e-01 5.24646103e-01
1.35592520e+00 4.43071008e-01 -4.59403783e-01 -5.24319708e-01
5.63199699e-01 -5.23940623e-01 1.49710000e-01 -4.70335513e-01
-9.63891745e-01 8.52700114e-01 1.15107432e-01 1.00168502e-02
9.51661050e-01 2.57344753e-01 1.18304503e+00 2.47289523e-01
2.04335064e-01 -8.53878856e-01 -9.60351154e-02 9.50262547e-01
7.14357257e-01 -9.97576237e-01 4.21060652e-01 -6.06782734e-01
-4.88587350e-01 9.22720432e-01 4.69942361e-01 -1.50438115e-01
1.11952913e+00 4.42465216e-01 -5.53048700e-02 -3.51510704e-01
-9.48988557e-01 -7.43893385e-01 6.23938024e-01 4.82378244e-01
7.75513232e-01 2.84267575e-01 -1.17430699e+00 1.24616849e+00
-5.45281991e-02 -5.47293305e-01 1.03475407e-01 1.43031788e+00
-2.48651162e-01 -1.77148640e+00 1.73910439e-01 3.35474849e-01
-6.70539558e-01 -8.38654220e-01 -4.05344814e-01 7.42612004e-01
2.62643665e-01 9.06423688e-01 -6.31128401e-02 -4.55374002e-01
3.43789726e-01 2.68684179e-01 3.72220755e-01 -1.00218034e+00
-6.74539387e-01 1.37089074e-01 2.64494061e-01 -4.65848565e-01
-2.62165338e-01 -3.61409903e-01 -1.27635479e+00 2.08862517e-02
-4.47740644e-01 3.18214804e-01 8.26149642e-01 1.05636168e+00
3.40962052e-01 8.40712607e-01 5.97990215e-01 -3.09553742e-01
7.70445019e-02 -1.22836328e+00 -5.52527726e-01 1.32412687e-01
2.19880044e-01 -6.34778798e-01 -4.72727031e-01 -3.49951476e-01]
|
[10.550362586975098, 8.487650871276855]
|
2be0c558-d7ef-4978-a672-478776a445ee
|
curriculum-guided-abstractive-summarization-1
|
2302.01342
| null |
https://arxiv.org/abs/2302.01342v2
|
https://arxiv.org/pdf/2302.01342v2.pdf
|
Curriculum-Guided Abstractive Summarization
|
Recent Transformer-based summarization models have provided a promising approach to abstractive summarization. They go beyond sentence selection and extractive strategies to deal with more complicated tasks such as novel word generation and sentence paraphrasing. Nonetheless, these models have two shortcomings: (1) they often perform poorly in content selection, and (2) their training strategy is not quite efficient, which restricts model performance. In this paper, we explore two orthogonal ways to compensate for these pitfalls. First, we augment the Transformer network with a sentence cross-attention module in the decoder, encouraging more abstraction of salient content. Second, we include a curriculum learning approach to reweight the training samples, bringing about an efficient learning procedure. Our second approach to enhance the training strategy of Transformers networks makes stronger gains as compared to the first approach. We apply our model on extreme summarization dataset of Reddit TIFU posts. We further look into three cross-domain summarization datasets (Webis-TLDR-17, CNN/DM, and XSum), measuring the efficacy of curriculum learning when applied in summarization. Moreover, a human evaluation is conducted to show the efficacy of the proposed method in terms of qualitative criteria, namely, fluency, informativeness, and overall quality.
|
['Nazli Goharian', 'Franck Dernoncourt', 'Hanieh Deilamsalehy', 'Sajad Sotudeh']
|
2023-02-02
| null | null | null | null |
['abstractive-text-summarization', 'extreme-summarization']
|
['natural-language-processing', 'natural-language-processing']
|
[ 4.18750137e-01 3.99960399e-01 -2.29206488e-01 -1.35744482e-01
-9.76866126e-01 -3.52432907e-01 6.36772275e-01 4.56677943e-01
-5.75995147e-01 9.38460588e-01 1.01386631e+00 -1.52681828e-01
8.13898072e-02 -6.82308137e-01 -6.82511687e-01 -2.82036185e-01
4.84622806e-01 3.10538352e-01 1.74216837e-01 -5.87386131e-01
7.55646467e-01 2.48295311e-02 -1.25066817e+00 4.31059361e-01
1.57845521e+00 4.75489974e-01 3.84771049e-01 5.30092537e-01
-3.28774959e-01 8.93647552e-01 -9.83037174e-01 -7.55835652e-01
-2.05419540e-01 -9.31997776e-01 -1.04768825e+00 4.57080361e-03
5.06667733e-01 -4.70329165e-01 -3.74391139e-01 1.01555598e+00
8.17613363e-01 1.61879346e-01 7.40051925e-01 -5.39384425e-01
-8.46213341e-01 1.29103327e+00 -6.03487968e-01 2.75062084e-01
6.86650634e-01 1.14196099e-01 1.30715990e+00 -6.52004957e-01
6.02285028e-01 1.12848091e+00 4.69209850e-01 6.94279611e-01
-1.03585029e+00 -3.80859852e-01 1.86036691e-01 3.46644014e-01
-7.70648360e-01 -5.92054248e-01 9.95018482e-01 -1.49637640e-01
1.13492489e+00 3.72021556e-01 7.23927379e-01 1.36143219e+00
2.27269113e-01 1.32150388e+00 7.92935967e-01 -4.41848636e-01
2.01913685e-01 1.93896487e-01 2.71084219e-01 3.43559086e-01
2.41795361e-01 -2.93298721e-01 -5.76163590e-01 1.96606383e-01
2.74459541e-01 -2.93013453e-01 -3.60894799e-01 8.80702287e-02
-1.16533554e+00 9.20182467e-01 3.58821809e-01 4.85530704e-01
-4.84842837e-01 -5.44929840e-02 8.33406448e-01 3.62136006e-01
6.29752934e-01 1.05099106e+00 -2.29363769e-01 -1.84606031e-01
-1.30146337e+00 3.81010890e-01 6.61201119e-01 1.00709963e+00
3.77135247e-01 4.37458783e-01 -7.59673774e-01 1.03629947e+00
-5.96568249e-02 2.82648236e-01 8.86003315e-01 -7.35176206e-01
9.24132109e-01 5.46686709e-01 -2.25219011e-01 -8.14765215e-01
-3.17696750e-01 -7.69234538e-01 -1.02791929e+00 -2.99488068e-01
-7.17481151e-02 -2.31433466e-01 -6.47077084e-01 1.62108362e+00
-2.86691725e-01 -3.27121139e-01 2.77611196e-01 6.04837954e-01
1.29270124e+00 7.52366304e-01 5.11398055e-02 -2.68911928e-01
1.21582913e+00 -1.25940013e+00 -9.34103727e-01 -4.24372762e-01
6.77897453e-01 -6.15357637e-01 1.27259481e+00 3.09679866e-01
-1.58900762e+00 -5.08830547e-01 -1.10806847e+00 -2.52027661e-01
-6.83577284e-02 3.05335701e-01 3.14008921e-01 3.57898057e-01
-9.89797294e-01 7.85942554e-01 -3.45208287e-01 -5.00451088e-01
4.32677418e-01 3.49820629e-02 4.20987047e-02 2.40059957e-01
-1.29512072e+00 1.11426413e+00 7.34767914e-01 -2.18168288e-01
-5.19794226e-01 -6.54920101e-01 -9.16246057e-01 5.44005573e-01
2.48249114e-01 -9.93356824e-01 1.42823386e+00 -9.51212287e-01
-1.76287436e+00 6.14569306e-01 9.79052484e-03 -6.98362231e-01
3.87595952e-01 -4.26460922e-01 -1.05586432e-01 2.03845814e-01
1.89508319e-01 6.67575002e-01 5.99809408e-01 -1.01377535e+00
-5.63132942e-01 -5.16886413e-02 1.42058641e-01 4.85929877e-01
-6.74960971e-01 -1.06134206e-01 -6.45534024e-02 -1.01383042e+00
-2.88655043e-01 -3.82354498e-01 -2.52755046e-01 -8.46792817e-01
-8.06721568e-01 -2.50066459e-01 2.87519932e-01 -9.83218551e-01
1.74510419e+00 -1.96370327e+00 5.40153563e-01 -3.83106440e-01
7.74117038e-02 6.08695745e-01 -3.73344481e-01 7.91640759e-01
1.17015809e-01 3.04865897e-01 -4.45663095e-01 -4.69944417e-01
2.20707096e-02 -1.93099827e-01 -4.99254555e-01 -1.41535670e-01
4.37387913e-01 1.05891037e+00 -1.03154063e+00 -5.84793091e-01
1.36927962e-02 -3.46724838e-02 -8.00732493e-01 6.06784970e-02
-3.45148921e-01 3.89149249e-01 -4.53714907e-01 2.72313654e-01
3.56795222e-01 1.65387765e-01 -5.22452593e-03 -6.79540560e-02
-1.28365964e-01 8.33234727e-01 -6.46456599e-01 1.80696487e+00
-4.72714365e-01 5.96608102e-01 -4.78495419e-01 -1.10827065e+00
7.96496987e-01 3.03976178e-01 2.30626032e-01 -9.83299255e-01
1.83803946e-01 1.11227915e-01 -1.15514442e-01 -6.67194009e-01
1.21041632e+00 -8.36658627e-02 -2.06356883e-01 4.06354487e-01
2.64635503e-01 -3.51902485e-01 5.58571696e-01 5.04470825e-01
9.52945232e-01 8.58050957e-02 5.03614902e-01 -2.30683789e-01
5.24832070e-01 1.41992699e-02 4.43531305e-01 6.69030905e-01
9.45508927e-02 6.96121216e-01 7.43894339e-01 9.83009338e-02
-1.06745601e+00 -7.71231234e-01 2.40271270e-01 8.88526618e-01
-7.08132312e-02 -6.51870251e-01 -9.11258698e-01 -7.31444120e-01
-3.28901470e-01 1.27643180e+00 -3.77402425e-01 -4.50715661e-01
-7.65026510e-01 -6.11914456e-01 6.57132745e-01 6.60959363e-01
6.32342458e-01 -1.38990211e+00 -6.27841055e-01 3.83891702e-01
-6.66110516e-01 -9.40127611e-01 -6.16447330e-01 -1.63136423e-01
-1.19394076e+00 -6.24150336e-01 -7.90898919e-01 -7.58000612e-01
4.55239713e-01 2.31789201e-01 1.21970367e+00 -1.07713882e-02
2.28449106e-01 1.59460306e-01 -6.59172356e-01 -4.74443048e-01
-6.35069311e-01 6.65911078e-01 -2.82385767e-01 -3.60944778e-01
5.35609499e-02 -4.76881742e-01 -4.93194938e-01 -2.51825541e-01
-9.85210717e-01 3.13090503e-01 9.75786388e-01 9.96688724e-01
1.42627731e-01 -2.56846726e-01 1.13437045e+00 -9.95485902e-01
1.30701911e+00 -3.35016191e-01 3.38751972e-02 3.43484104e-01
-5.29774249e-01 8.83039534e-02 8.19628716e-01 -2.54899591e-01
-1.27015030e+00 -4.59379017e-01 -4.98479903e-01 1.18676610e-01
1.25289381e-01 8.69742870e-01 7.00146239e-03 5.08223891e-01
7.50945866e-01 5.41166306e-01 -5.49746267e-02 -4.66597855e-01
4.35602784e-01 6.47419930e-01 4.73264933e-01 -3.81449163e-01
7.21359789e-01 -6.98414743e-02 -4.54593062e-01 -9.13230002e-01
-9.49368119e-01 -2.05300957e-01 -3.99840474e-01 1.08226109e-02
7.47493625e-01 -7.66136229e-01 -3.61391842e-01 2.85331786e-01
-1.25492096e+00 -2.09363192e-01 -7.17413902e-01 3.83447230e-01
-6.34164691e-01 8.18721890e-01 -7.09477544e-01 -5.09058535e-01
-9.29461598e-01 -9.02617574e-01 8.51454556e-01 4.18412387e-01
-5.33204317e-01 -9.39039707e-01 2.15616021e-02 4.63179052e-01
6.42459393e-01 4.38481271e-02 1.16169536e+00 -8.37977469e-01
-1.13283217e-01 -7.36646056e-02 -1.05178505e-01 5.63424110e-01
-9.25195143e-02 -3.93551812e-02 -6.57990992e-01 -2.53537714e-01
1.46950148e-02 -4.74419475e-01 1.16782951e+00 3.95282447e-01
1.09239805e+00 -5.87818682e-01 4.62915897e-02 3.23936701e-01
1.01692021e+00 -6.75819516e-02 7.87044704e-01 5.15749633e-01
4.56760049e-01 6.34575188e-01 6.90014660e-01 4.48560953e-01
4.78247225e-01 4.81734097e-01 2.28510484e-01 -1.47622377e-01
-4.03412580e-01 -4.19653475e-01 6.17041707e-01 1.32293451e+00
-6.70381486e-02 -4.72744107e-01 -4.66144085e-01 6.15610361e-01
-1.84054112e+00 -1.09673619e+00 -1.09295726e-01 1.96376753e+00
1.03556919e+00 2.63131350e-01 1.57154948e-01 4.20460925e-02
4.49568152e-01 3.13525468e-01 -2.86824465e-01 -7.85450339e-01
-2.98891962e-01 1.84572801e-01 6.56110197e-02 2.36495703e-01
-7.41541386e-01 1.06061316e+00 6.04718113e+00 9.13256764e-01
-9.62580264e-01 -2.11241515e-03 4.52958643e-01 -2.18422815e-01
-5.85774958e-01 -1.78521872e-01 -6.25849962e-01 5.45966506e-01
7.84365535e-01 -4.66911584e-01 1.24094456e-01 5.52021861e-01
2.89006114e-01 -6.03022426e-02 -1.07337880e+00 5.47386408e-01
3.40364784e-01 -1.53311086e+00 5.15313983e-01 -3.23691726e-01
7.42935121e-01 -1.77235991e-01 1.75605845e-02 7.85078168e-01
-1.41658839e-02 -7.83182025e-01 8.27858865e-01 4.34499264e-01
5.65782607e-01 -8.10370207e-01 8.83607507e-01 4.91017580e-01
-5.94269454e-01 -7.44324178e-02 -4.19032454e-01 -1.33118220e-02
3.12781483e-01 4.18363392e-01 -7.15884387e-01 8.96885395e-01
2.99343526e-01 6.99699879e-01 -7.62678027e-01 1.16040850e+00
-5.43428242e-01 7.12740302e-01 2.46927634e-01 -3.22392851e-01
2.91974485e-01 -2.25080311e-01 8.69375348e-01 1.50937796e+00
3.12211156e-01 -1.13767445e-01 -1.72018066e-01 8.72038603e-01
-2.82160372e-01 3.43764961e-01 -5.17870724e-01 -1.05481781e-01
3.48345160e-01 1.05461025e+00 -2.90137023e-01 -4.78471071e-01
-2.76945472e-01 9.60997343e-01 4.15905923e-01 3.31981063e-01
-7.07168698e-01 -5.54047585e-01 9.69102159e-02 7.31509328e-02
2.45187595e-01 -5.18309101e-02 -5.12951672e-01 -1.36032927e+00
-4.28499654e-03 -1.13897192e+00 2.09546745e-01 -5.62545180e-01
-1.09527385e+00 5.90885460e-01 1.22169822e-01 -1.17187643e+00
-3.65088046e-01 -4.16833013e-02 -9.46086884e-01 6.88961208e-01
-1.66851914e+00 -1.10321093e+00 2.88950745e-03 9.56738591e-02
1.12675381e+00 -2.91403979e-01 6.18694544e-01 3.40517461e-01
-7.69081056e-01 7.63886690e-01 6.46560937e-02 -8.52027386e-02
8.26333880e-01 -1.39098060e+00 4.99668658e-01 1.00241768e+00
-1.69525504e-01 5.41152596e-01 8.65998149e-01 -5.67803383e-01
-1.17943287e+00 -1.07855892e+00 1.13452888e+00 -1.71392620e-01
4.10006285e-01 -1.59181714e-01 -8.93473983e-01 5.66778600e-01
7.61892855e-01 -1.11138797e+00 6.40624285e-01 1.72375977e-01
7.39695132e-02 3.30541208e-02 -8.85203004e-01 9.07719374e-01
8.55613291e-01 -4.84887287e-02 -1.10933149e+00 2.81048626e-01
9.12800431e-01 -3.34190249e-01 -6.88771784e-01 3.92634094e-01
1.80100054e-01 -8.81582141e-01 7.27238894e-01 -6.16987050e-01
1.18985248e+00 1.47220373e-01 1.81561410e-01 -1.83697748e+00
-5.06410599e-01 -6.23198092e-01 -2.68489420e-01 1.69740915e+00
5.18041670e-01 -5.46866179e-01 5.61402917e-01 7.47082382e-02
-6.14588857e-01 -9.34288025e-01 -6.94955468e-01 -6.28460288e-01
3.79807293e-01 8.01811814e-02 6.13796532e-01 6.80817604e-01
3.03021699e-01 9.94442523e-01 -4.38441873e-01 -6.15819752e-01
3.12600434e-01 1.85927190e-02 7.40330517e-01 -9.00631309e-01
-3.11793327e-01 -1.01099181e+00 1.22277766e-01 -1.26122117e+00
2.30234399e-01 -9.57699776e-01 -8.11338704e-03 -1.95433474e+00
4.11981612e-01 9.79790464e-03 -5.25487494e-03 2.32435271e-01
-5.52630186e-01 -1.90167278e-01 2.91131318e-01 7.24954754e-02
-5.88886559e-01 1.02659035e+00 1.44197071e+00 -3.05833757e-01
-2.54564047e-01 1.06371388e-01 -1.25405240e+00 4.97866660e-01
1.01284826e+00 -1.85469583e-01 -4.74402070e-01 -7.23118603e-01
2.76534170e-01 2.54146248e-01 2.66829897e-02 -8.97525132e-01
2.28578731e-01 4.93195355e-02 1.63561732e-01 -7.86753953e-01
5.43975234e-02 -2.08474860e-01 -4.58125830e-01 5.18887281e-01
-7.06390262e-01 1.16486065e-01 2.40848199e-01 3.67804915e-01
-4.19804156e-01 -7.08472848e-01 6.51215911e-01 -1.84404328e-01
-4.59702820e-01 -1.44973233e-01 -5.99867463e-01 3.06731105e-01
6.36650443e-01 -2.96517551e-01 -4.56423283e-01 -5.52119851e-01
-1.82614803e-01 4.25922453e-01 3.40771168e-01 4.02976125e-01
6.60075128e-01 -1.02685189e+00 -1.27876818e+00 -1.11444399e-01
9.22991782e-02 -3.95280272e-02 3.16005528e-01 9.01414692e-01
-4.16400969e-01 5.12790084e-01 -2.86046743e-01 -2.63312906e-01
-9.74409103e-01 4.95330513e-01 -1.60286576e-03 -6.76006079e-01
-8.70899379e-01 5.66257656e-01 1.20951816e-01 -3.76990765e-01
2.24168092e-01 -2.87187099e-01 -6.54665470e-01 4.06079024e-01
4.08497572e-01 5.25714874e-01 1.49989784e-01 -2.85599142e-01
9.57348105e-03 3.43374550e-01 -4.43866253e-01 -9.06360596e-02
1.31376553e+00 -1.53381988e-01 -7.46090859e-02 1.28228828e-01
8.41351509e-01 6.35464340e-02 -7.21497357e-01 -3.39481771e-01
2.38280714e-01 -1.08398482e-01 -1.64876029e-01 -8.93394053e-01
-7.13499546e-01 8.87039900e-01 -1.70507938e-01 2.85045117e-01
1.30950284e+00 -2.12715015e-01 1.06848001e+00 3.55767310e-01
-1.24086723e-01 -1.28311944e+00 3.94585788e-01 7.99881279e-01
1.16560781e+00 -1.04522204e+00 1.16832972e-01 -1.37237132e-01
-9.83995914e-01 1.17789984e+00 6.90384865e-01 -4.74263281e-02
-1.68660820e-01 -1.84757560e-01 -3.22674811e-01 -1.43622398e-01
-8.07613432e-01 -1.36263698e-01 2.50064343e-01 3.05812657e-01
7.50817657e-01 -1.08088985e-01 -9.10028577e-01 6.93718612e-01
-5.84636927e-01 -1.77877530e-01 8.95446539e-01 7.17738390e-01
-6.10599399e-01 -8.95590842e-01 2.56070681e-03 5.68102598e-01
-5.52163005e-01 -2.52662480e-01 -5.73340416e-01 6.10433221e-01
-4.33755636e-01 1.02419722e+00 -1.85426325e-01 -2.51644939e-01
7.25355148e-01 -3.53820398e-02 5.14219165e-01 -9.91192460e-01
-9.99255896e-01 -1.08133912e-01 3.78706217e-01 -1.45385057e-01
-1.75487638e-01 -6.37046099e-01 -9.98929739e-01 -2.58759201e-01
-4.20232922e-01 2.52903640e-01 4.87258583e-01 9.04909551e-01
3.69386077e-01 1.12264490e+00 6.16100311e-01 -7.30834365e-01
-1.06306398e+00 -1.35855055e+00 -2.95824289e-01 3.79009485e-01
1.67552114e-01 -1.35119379e-01 -1.18773028e-01 -2.11828172e-01]
|
[12.311898231506348, 9.356561660766602]
|
f4921c57-b1d6-4642-919a-c34b3bf3c4bc
|
fact-vs-opinion-the-role-of-argumentation
| null | null |
https://aclanthology.org/2020.coling-main.540
|
https://aclanthology.org/2020.coling-main.540.pdf
|
Fact vs. Opinion: the Role of Argumentation Features in News Classification
|
A 2018 study led by the Media Insight Project showed that most journalists think that a clearmarking of what is news reporting and what is commentary or opinion (e.g., editorial, op-ed)is essential for gaining public trust. We present an approach to classify news articles into newsstories (i.e., reporting of factual information) and opinion pieces using models that aim to sup-plement the article content representation with argumentation features. Our hypothesis is thatthe nature of argumentative discourse is important in distinguishing between news stories andopinion articles. We show that argumentation features outperform linguistic features used previ-ously and improve on fine-tuned transformer-based models when tested on data from publishersunseen in training. Automatically flagging opinion pieces vs. news stories can aid applicationssuch as fact-checking or event extraction.
|
['Daniel Preotiuc-Pietro', 'Smaranda Muresan', 'Tariq Alhindi']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['news-classification']
|
['natural-language-processing']
|
[-1.49426594e-01 6.63005829e-01 -9.98226702e-01 -1.30335033e-01
-1.14508319e+00 -1.05969167e+00 1.39968538e+00 1.21101665e+00
-5.23726642e-01 1.06604397e+00 1.27505803e+00 -8.72490525e-01
6.73145279e-02 -8.97102714e-01 -9.46097195e-01 -8.69037732e-02
3.72660518e-01 2.58636922e-01 1.56908423e-01 -5.68839133e-01
8.88344824e-01 -1.14189103e-01 -1.57548451e+00 1.16308486e+00
9.69409764e-01 9.92508948e-01 -4.03058171e-01 4.72435594e-01
-7.75046945e-01 1.69513929e+00 -1.15446496e+00 -8.00214410e-01
-4.32787418e-01 -5.31565428e-01 -1.04883456e+00 -1.61344826e-01
4.45433438e-01 1.37601674e-01 2.36269623e-01 8.90016735e-01
3.65064479e-02 -4.66031671e-01 7.16996610e-01 -9.71245408e-01
-8.12269986e-01 1.67070460e+00 -5.07709265e-01 8.72713089e-01
6.99857116e-01 -6.85890198e-01 1.45806646e+00 -6.26877546e-01
1.22406757e+00 1.16778994e+00 7.37284064e-01 -3.82176787e-01
-9.73913789e-01 -4.15430516e-01 3.74206424e-01 2.52772272e-01
-3.46275151e-01 -6.61777437e-01 6.96281493e-01 -8.98244500e-01
8.15179884e-01 5.68002939e-01 7.70974576e-01 1.37549031e+00
6.48878336e-01 8.12393367e-01 1.57568741e+00 -5.00539362e-01
3.48941870e-02 5.72011054e-01 4.69477385e-01 2.73789614e-01
7.01264441e-01 -5.12029767e-01 -1.01639402e+00 -6.07204854e-01
-7.22551122e-02 -3.96640509e-01 -3.25401962e-01 5.68188727e-01
-1.31737292e+00 1.22033679e+00 -2.57902797e-02 5.41636705e-01
-6.29856527e-01 -4.03002381e-01 9.13947165e-01 8.25690329e-01
9.94529009e-01 6.25807226e-01 -5.58179438e-01 -3.93568873e-01
-9.89656866e-01 4.40103978e-01 1.45560765e+00 3.18974614e-01
1.89542919e-01 -1.58976153e-01 -3.24159533e-01 8.52642238e-01
2.48119220e-01 4.31785315e-01 5.96050203e-01 -6.84084415e-01
7.18036652e-01 7.64828682e-01 3.74128848e-01 -1.34383059e+00
-3.13555241e-01 -5.23340404e-01 2.11396795e-02 -1.39148057e-01
2.55874187e-01 -3.80036950e-01 -4.04584289e-01 9.35711145e-01
1.29447922e-01 -5.09814143e-01 2.33782709e-01 4.71669853e-01
1.32466829e+00 7.67476976e-01 6.86158659e-03 -4.44773644e-01
1.79302800e+00 -3.80214125e-01 -1.04646492e+00 -1.66199535e-01
7.91007578e-01 -1.30587471e+00 6.78388834e-01 4.96067494e-01
-1.14057839e+00 1.60329770e-02 -1.13383412e+00 -1.45194873e-01
-5.29502928e-01 8.38867202e-02 4.69205201e-01 2.42040411e-01
-2.93261498e-01 3.71439815e-01 -3.52722436e-01 -2.69144565e-01
4.92456645e-01 -6.97110713e-01 -3.40974361e-01 5.83297193e-01
-1.32973027e+00 1.09738648e+00 3.60211968e-01 -4.63213831e-01
-1.09429568e-01 -6.75595224e-01 -5.89984655e-01 -3.36718857e-01
5.75103581e-01 -2.54830629e-01 1.48201549e+00 -1.03426814e+00
-1.13784063e+00 1.18348396e+00 -8.90608802e-02 -8.22623312e-01
4.94106025e-01 -4.07837361e-01 -8.44003856e-01 2.10196987e-01
6.27394557e-01 -3.94619167e-01 6.94491804e-01 -9.26626265e-01
-1.10538828e+00 -1.36327744e-01 5.02161562e-01 -1.74194034e-02
-2.07503080e-01 7.14702010e-01 5.10899901e-01 -9.05992091e-01
-2.14051250e-02 -5.16451657e-01 5.02874613e-01 -3.29537928e-01
-8.04899573e-01 -4.36365724e-01 9.25966442e-01 -1.00458193e+00
1.42335224e+00 -1.57923698e+00 -3.51446986e-01 7.35540614e-02
3.37902278e-01 -2.36220270e-01 5.80814719e-01 8.77643049e-01
7.47145414e-02 6.37962937e-01 4.59102184e-01 5.47395766e-01
1.04724526e-01 1.03338555e-01 -9.56961870e-01 5.04574060e-01
-7.30739459e-02 6.80417776e-01 -7.98410356e-01 -5.73643088e-01
-5.93153894e-01 2.72605121e-01 -2.81199664e-01 -3.18218023e-01
-4.67857927e-01 1.33427918e-01 -5.33749521e-01 5.65848410e-01
-7.93967247e-02 -5.35160899e-01 1.40319094e-01 -3.61447990e-01
-8.01418841e-01 1.39184082e+00 -7.51851320e-01 7.04250872e-01
-5.09333372e-01 1.28364289e+00 1.50754631e-01 -7.25727618e-01
6.20379031e-01 5.31671584e-01 4.80231792e-02 -5.82534552e-01
5.59253752e-01 3.32658321e-01 3.35295647e-02 -3.00760299e-01
5.23639023e-01 -4.85200658e-02 -2.73467571e-01 8.07453156e-01
-4.01770562e-01 6.80174083e-02 3.95856082e-01 4.87880975e-01
6.10029042e-01 -2.50568241e-01 7.39612401e-01 -5.80964565e-01
3.61880064e-01 5.55963576e-01 3.03182006e-01 7.09357023e-01
4.76725131e-01 1.77452177e-01 1.16911936e+00 -3.53410363e-01
-1.01409638e+00 -2.30905920e-01 -5.06487131e-01 1.16859460e+00
-2.83182204e-01 -8.48580301e-01 -3.67229551e-01 -8.31694365e-01
1.72061741e-01 1.42024660e+00 -1.05361569e+00 3.93078119e-01
-3.99738491e-01 -5.73690772e-01 2.36570895e-01 3.03695321e-01
2.35028222e-01 -6.06162310e-01 -8.33377659e-01 4.09369200e-01
-5.59045732e-01 -8.56696129e-01 -1.85158491e-01 2.94058233e-01
-2.81336099e-01 -1.31873345e+00 -2.28032738e-01 -3.02082539e-01
2.59655029e-01 -1.03881195e-01 1.28414559e+00 -7.03193918e-02
3.84213805e-01 8.86754021e-02 -7.73274422e-01 -9.51640904e-01
-9.41870272e-01 1.50589263e-02 -4.75286931e-01 -1.57302961e-01
2.69782573e-01 -3.37662011e-01 -2.03023002e-01 -5.40311672e-02
-7.67325759e-01 -2.82066949e-02 2.71325558e-01 8.35746348e-01
3.50088686e-01 -1.43077180e-01 5.47362387e-01 -1.60036314e+00
1.03972685e+00 -8.91095877e-01 -2.39513531e-01 1.49803385e-01
-7.47351408e-01 1.04911089e-01 3.01456183e-01 -3.46849322e-01
-1.03051364e+00 -1.20573592e+00 -1.95065930e-01 5.90203404e-01
2.69902676e-01 1.24093449e+00 3.45636904e-01 4.68572527e-01
9.81710434e-01 -3.18293571e-01 -1.94819923e-02 -2.35419616e-01
3.23966771e-01 7.90466666e-01 5.76450974e-02 -6.37952268e-01
4.95111823e-01 7.13836908e-01 -6.61902845e-01 -7.83136725e-01
-1.78323627e+00 -2.96245605e-01 1.55839726e-01 -2.66777068e-01
5.26495695e-01 -8.68634462e-01 -4.56370860e-01 -2.13179603e-01
-1.26108253e+00 4.24660951e-01 -5.22857726e-01 4.46005553e-01
-2.61896700e-01 -8.91420469e-02 -6.92315698e-01 -6.65378213e-01
-2.42882818e-01 -5.18675864e-01 6.06837988e-01 -1.07024319e-01
-7.48252869e-01 -9.96992111e-01 1.74246013e-01 5.77277899e-01
2.50684768e-01 6.74699962e-01 8.65693688e-01 -1.35054457e+00
1.68934122e-01 -4.17147070e-01 -9.10328329e-02 2.00543981e-02
-2.85788942e-02 4.72689837e-01 -8.32011223e-01 3.62962037e-01
2.45727286e-01 -4.23983127e-01 8.45509708e-01 3.32595706e-01
5.00070214e-01 -1.41746891e+00 -4.17591542e-01 -2.81450331e-01
1.05049586e+00 -2.46279407e-02 3.92420858e-01 1.02685428e+00
9.51937139e-02 5.76174796e-01 4.37993497e-01 5.19247949e-01
5.89318454e-01 3.56865197e-01 6.53317990e-03 4.59393680e-01
-7.94959962e-02 -4.35926318e-01 3.97673815e-01 8.00283015e-01
-1.29556537e-01 -4.26550686e-01 -8.39627266e-01 5.84899843e-01
-1.69761348e+00 -1.60731888e+00 -5.04604399e-01 1.61562014e+00
1.10597146e+00 8.33116353e-01 2.37004846e-01 3.64177823e-01
5.98716795e-01 4.99141306e-01 1.09707564e-01 -7.53580093e-01
-6.79442108e-01 -1.90880299e-01 5.12670577e-01 5.42010546e-01
-9.39313531e-01 2.87426770e-01 6.10542107e+00 4.79992598e-01
-1.03156948e+00 4.46854621e-01 6.30892932e-01 1.79215938e-01
-7.94208705e-01 1.90782368e-01 -1.01660502e+00 5.62866807e-01
1.12584531e+00 -6.72684014e-01 -5.24528623e-01 8.39290857e-01
1.97096527e-01 -3.68292511e-01 -7.02533960e-01 1.57692865e-01
4.12886024e-01 -2.07792306e+00 4.56084982e-02 9.11073666e-03
7.97138929e-01 7.28303120e-02 -1.52323484e-01 6.79141581e-02
6.42793596e-01 -4.82753664e-01 1.61374104e+00 2.95547605e-01
1.90320835e-01 -4.15218860e-01 1.06306732e+00 2.61146575e-01
-4.76362526e-01 -7.01208934e-02 7.16326460e-02 -2.01519579e-01
4.33251649e-01 9.97852325e-01 -7.29456842e-01 5.31867385e-01
7.53483355e-01 1.01718092e+00 -3.79716188e-01 8.07253599e-01
-5.53296149e-01 1.17769825e+00 -5.49391359e-02 -3.73021096e-01
2.01217219e-01 2.00854644e-01 8.78100991e-01 1.37267387e+00
2.56346524e-01 -5.64019121e-02 1.38188213e-01 5.15715182e-01
-1.66562587e-01 3.69523436e-01 -2.95615882e-01 -6.55283868e-01
3.93893629e-01 1.08785307e+00 -9.51302707e-01 -5.64167500e-01
-6.43793643e-01 2.30277866e-01 1.12133741e-01 1.43784150e-01
-6.77407682e-01 -2.97162086e-01 2.05155551e-01 6.30124092e-01
6.31853640e-01 2.74778545e-01 -1.91443577e-01 -1.21360576e+00
-1.04831392e-02 -9.21633780e-01 5.45367301e-01 -7.39774287e-01
-1.45394206e+00 8.01309764e-01 7.96052534e-03 -1.06044126e+00
-2.38905087e-01 -5.12373567e-01 -6.74764633e-01 5.01070797e-01
-1.70153773e+00 -9.52431917e-01 2.34320417e-01 -1.90680444e-01
3.15954000e-01 1.48999646e-01 3.90016526e-01 -7.24809095e-02
-1.14787310e-01 -3.00363321e-02 1.03828907e-01 3.48830491e-01
8.38057101e-01 -1.21092176e+00 -1.27896853e-02 3.76849264e-01
3.61676365e-01 6.41985118e-01 1.39792478e+00 -9.02689517e-01
-9.54525173e-01 -5.30195713e-01 1.47369659e+00 -7.65618324e-01
1.55802584e+00 9.38467905e-02 -7.51578689e-01 6.34938836e-01
9.59803998e-01 -7.72405446e-01 1.04289103e+00 6.10661566e-01
-9.33986604e-01 2.98801810e-01 -8.00020397e-01 3.38709652e-01
5.63054144e-01 -6.35246515e-01 -1.65430522e+00 6.43634737e-01
5.71002245e-01 -4.90354270e-01 -8.56918156e-01 -8.86303037e-02
5.76456487e-01 -7.81603396e-01 5.32221556e-01 -9.58393693e-01
8.91157568e-01 -9.78150591e-02 -7.05654696e-02 -1.31087697e+00
5.63024394e-02 -5.80787838e-01 -2.78409123e-01 1.33623660e+00
9.74870503e-01 -9.26892042e-01 2.90078431e-01 1.74454749e-01
-1.55106917e-01 -3.63203347e-01 -6.89458668e-01 -5.33039987e-01
1.12076283e-01 -3.71951550e-01 2.41929829e-01 1.34914351e+00
5.65859377e-01 6.40660107e-01 2.24903956e-01 -2.65154094e-01
1.23911491e-02 6.60492182e-01 2.77097434e-01 -1.72946239e+00
-5.13069071e-02 -7.06154048e-01 -2.21249804e-01 -4.36724305e-01
1.05945900e-01 -7.37447500e-01 -3.97359073e-01 -1.98899388e+00
2.13742822e-01 8.14520121e-02 2.07103393e-03 4.08546269e-01
1.21136658e-01 -8.71061981e-02 -2.86322564e-01 4.04009700e-01
-3.57207775e-01 3.95620205e-02 1.03670871e+00 -2.71807849e-01
8.07800218e-02 -7.51211345e-02 -1.39971721e+00 9.75153327e-01
5.81795454e-01 -6.24070585e-01 -4.44789194e-02 3.32992338e-02
1.32253158e+00 -1.08673185e-01 4.06236172e-01 -3.99098188e-01
1.65889874e-01 -3.48713011e-01 4.75877412e-02 -7.59908676e-01
-2.09244251e-01 -3.77624542e-01 -3.65979820e-02 1.71173394e-01
-8.57319653e-01 3.75718951e-01 1.18191175e-01 5.56844711e-01
-5.46765268e-01 -2.65043080e-01 2.12986380e-01 -1.41133636e-01
-4.98513617e-02 -4.88471240e-01 -7.87256658e-01 7.56755471e-01
5.17089784e-01 -2.01964632e-01 -1.17034292e+00 -6.75531983e-01
-5.27554572e-01 -2.18588531e-01 2.57550448e-01 3.44911188e-01
6.51527420e-02 -9.46435928e-01 -1.36842859e+00 -5.30201852e-01
2.15643421e-01 -8.29647779e-01 -2.06848398e-01 1.19402266e+00
-3.84596378e-01 5.25731027e-01 2.02284202e-01 1.65418964e-02
-1.13097620e+00 8.38341638e-02 -2.62500733e-01 -1.00240082e-01
-7.47182906e-01 7.97562420e-01 -4.70294386e-01 8.84249955e-02
-1.32294133e-01 -6.44778013e-01 -7.63727903e-01 1.06395328e+00
8.88356268e-01 3.29251051e-01 1.37786627e-01 -8.38191807e-01
-1.55348167e-01 -5.93778528e-02 -4.70190644e-01 -2.43704081e-01
1.52923477e+00 -8.31252784e-02 -4.91398305e-01 9.80811894e-01
9.04210627e-01 9.36876118e-01 -4.68262225e-01 -8.85430053e-02
3.01354378e-01 -2.13570818e-01 3.58347803e-01 -1.15180755e+00
-5.04517555e-01 1.24992043e-01 -3.92988801e-01 1.16827953e+00
1.91770256e-01 6.35543525e-01 4.62547094e-01 2.90143311e-01
-1.41503550e-02 -1.46156156e+00 -2.28308454e-01 4.22038913e-01
1.46149480e+00 -1.15948439e+00 5.37582457e-01 -3.72649103e-01
-8.50675762e-01 1.22711861e+00 -3.54540497e-02 3.23847495e-02
9.92560804e-01 3.44652891e-01 1.77941576e-01 -7.90604651e-01
-9.01904464e-01 1.80074558e-01 5.90632200e-01 -1.57533646e-01
1.05452085e+00 -4.02242728e-02 -1.01961219e+00 8.17192614e-01
-7.77464271e-01 -3.47493261e-01 8.49969327e-01 1.05123603e+00
-6.92454934e-01 -6.53303206e-01 -4.26090419e-01 1.02315044e+00
-1.17636991e+00 -1.30009979e-01 -6.94418252e-01 9.15865660e-01
2.09747199e-02 1.16048181e+00 6.47083446e-02 9.48373042e-03
1.83008254e-01 -5.34790121e-02 2.05302127e-02 -6.34247899e-01
-1.01097167e+00 3.52631770e-02 1.32623971e+00 -2.21735820e-01
-9.68804777e-01 -1.07101488e+00 -9.84770417e-01 -3.96373868e-01
-5.66255629e-01 9.91968691e-01 8.23383868e-01 1.16459000e+00
4.39714074e-01 5.01590312e-01 6.15999587e-02 9.11875218e-02
-2.85331041e-01 -7.73480952e-01 -9.12255347e-02 1.36599019e-01
4.88107234e-01 -6.69976771e-01 -5.70098519e-01 1.67620197e-01]
|
[9.037596702575684, 9.773233413696289]
|
6bc854b9-02e8-46bb-8e7b-fe4f609dae02
|
twitter-bot-detection-using-bidirectional
|
2002.01336
| null |
https://arxiv.org/abs/2002.01336v1
|
https://arxiv.org/pdf/2002.01336v1.pdf
|
Twitter Bot Detection Using Bidirectional Long Short-term Memory Neural Networks and Word Embeddings
|
Twitter is a web application playing dual roles of online social networking and micro-blogging. The popularity and open structure of Twitter have attracted a large number of automated programs, known as bots. Legitimate bots generate a large amount of benign contextual content, i.e., tweets delivering news and updating feeds, while malicious bots spread spam or malicious contents. To assist human users in identifying who they are interacting with, this paper focuses on the classification of human and spambot accounts on Twitter, by employing recurrent neural networks, specifically bidirectional Long Short-term Memory (BiLSTM), to efficiently capture features across tweets. To the best of our knowledge, our work is the first that develops a recurrent neural model with word embeddings to distinguish Twitter bots from human accounts, that requires no prior knowledge or assumption about users' profiles, friendship networks, or historical behavior on the target account. Moreover, our model does not require any handcrafted features. The preliminary simulation results are very encouraging. Experiments on the cresci-2017 dataset show that our approach can achieve competitive performance compared with existing state-of-the-art bot detection systems.
|
['Uyen Trang Nguyen', 'Feng Wei']
|
2020-02-03
| null | null | null | null |
['twitter-bot-detection']
|
['miscellaneous']
|
[-2.93568641e-01 -1.63668975e-01 -4.56253976e-01 4.43087816e-02
-9.24448669e-02 -4.42436874e-01 8.58628511e-01 4.56900112e-02
-4.75947022e-01 3.60759288e-01 -2.02015433e-02 -4.42689836e-01
7.00509965e-01 -9.97091830e-01 -2.62942076e-01 -5.27389169e-01
-8.33698586e-02 3.14719528e-01 7.51895905e-01 -3.82744044e-01
3.93116355e-01 2.76543528e-01 -1.05836236e+00 1.15752645e-01
5.45230746e-01 7.28183091e-01 3.05082798e-02 6.67564571e-01
-3.39780480e-01 1.46341741e+00 -7.53773332e-01 -7.14317918e-01
-1.29329056e-01 -1.94131628e-01 -7.61748791e-01 -1.63141742e-01
-1.69608250e-01 -5.82109272e-01 -8.33639622e-01 1.21750033e+00
3.45072865e-01 -6.73329756e-02 4.64589655e-01 -1.36362112e+00
-9.93257344e-01 8.22486639e-01 -7.15664685e-01 6.22243524e-01
-2.81849448e-02 4.03289586e-01 1.08883560e+00 -7.40672648e-01
5.08614004e-01 1.25659633e+00 9.42225218e-01 4.84476209e-01
-8.21848571e-01 -9.32248235e-01 2.46931687e-01 1.34760097e-01
-9.11602616e-01 -1.05498493e-01 6.42228007e-01 -7.50545621e-01
8.42837095e-01 -5.63643128e-02 4.46223587e-01 1.71963263e+00
1.27151862e-01 9.35706317e-01 6.40733242e-01 1.19779557e-01
-2.45820239e-01 5.32175004e-01 5.24644673e-01 8.21622849e-01
3.60087305e-01 -2.78487295e-01 -2.14698225e-01 -6.33376062e-01
3.67957860e-01 5.13399065e-01 1.79332927e-01 3.83018553e-01
-8.68452907e-01 1.44736588e+00 7.20302999e-01 5.87291598e-01
-5.04142284e-01 3.41451764e-01 9.17914033e-01 1.87067762e-01
8.67065072e-01 2.45730668e-01 2.60102749e-02 -9.86251310e-02
-6.45848215e-01 -5.40829450e-02 9.25159991e-01 6.92910254e-01
7.33789802e-01 2.57013708e-01 -2.16737706e-02 7.41500974e-01
3.52210104e-01 6.18134856e-01 1.05925977e+00 -3.19720894e-01
4.76493150e-01 7.73671508e-01 8.44768733e-02 -1.64160359e+00
-3.26946497e-01 -2.12404266e-01 -7.27901340e-01 -5.17691374e-01
4.16084617e-01 -4.30926204e-01 -4.22669381e-01 1.33887959e+00
9.36443731e-02 5.14629543e-01 -4.42878008e-01 3.54760140e-01
8.85277033e-01 7.73803055e-01 6.86571673e-02 3.16661806e-03
1.33008790e+00 -1.09884870e+00 -7.35375106e-01 -3.43887120e-01
5.09914577e-01 -5.19071460e-01 8.18758607e-01 -1.73006520e-01
-7.05535769e-01 -8.45045373e-02 -4.18110937e-01 2.19742611e-01
-1.02799594e+00 -1.70366615e-01 4.88385171e-01 7.38719523e-01
-6.29460752e-01 5.88468373e-01 -9.27804410e-01 -5.21893442e-01
6.16237223e-01 9.38240215e-02 1.78287446e-01 5.87036848e-01
-1.54360843e+00 8.57525408e-01 1.13492951e-01 -6.36057779e-02
-8.08549881e-01 -2.90106565e-01 -5.38038075e-01 -1.76423237e-01
2.88224280e-01 -6.90793842e-02 1.56227958e+00 -7.83568203e-01
-1.31741488e+00 8.36226225e-01 -2.28987858e-01 -7.48699963e-01
3.91567200e-01 3.51917632e-02 -3.91282916e-01 -2.73646843e-02
4.86019760e-01 2.95013702e-03 1.08651745e+00 -8.72525334e-01
-6.26992702e-01 -2.70739317e-01 6.20440878e-02 -5.20750642e-01
-1.10866189e+00 5.71354687e-01 -1.84004560e-01 -6.07961297e-01
-3.82040799e-01 -1.02733231e+00 -9.74608138e-02 -7.30707645e-01
-8.79820347e-01 -6.89373791e-01 1.40426278e+00 -7.03613400e-01
1.43104184e+00 -1.85117257e+00 -3.27352762e-01 2.40728576e-02
6.32829249e-01 7.24278688e-01 1.06872693e-01 7.60387838e-01
2.54364103e-01 5.43344259e-01 2.30776906e-01 -4.22817022e-01
3.60726416e-02 -1.00996412e-01 -8.04049313e-01 6.64960861e-01
1.02849223e-01 1.12638998e+00 -1.22288990e+00 -2.97988296e-01
-1.82375520e-01 4.11298782e-01 -3.15053016e-01 1.40235871e-01
-9.50254053e-02 3.75098258e-01 -8.69214535e-01 4.91724193e-01
2.55952269e-01 -7.95083940e-01 -1.64679229e-01 2.96178490e-01
-2.08830312e-01 6.53195739e-01 -3.63960147e-01 4.58754301e-01
-5.85280180e-01 1.02179754e+00 -7.03759957e-03 -7.44797468e-01
6.98423147e-01 2.87851512e-01 4.27508950e-01 -3.11082274e-01
7.62102842e-01 2.29287639e-01 -3.51328477e-02 -4.92357850e-01
5.24544537e-01 2.88904309e-01 -1.32032976e-01 1.00261962e+00
-2.17938110e-01 4.87948328e-01 2.79797912e-01 4.07082200e-01
1.33687699e+00 -7.66127050e-01 1.48637876e-01 2.54313707e-01
4.46680635e-01 -4.30854037e-02 2.27164268e-01 8.99795055e-01
-5.50202072e-01 5.56291975e-02 8.02270174e-01 -3.50270182e-01
-8.75599802e-01 -5.75637043e-01 1.78940341e-01 1.52126777e+00
7.78732821e-02 -2.50374019e-01 -7.11955369e-01 -9.01506901e-01
9.76976380e-02 4.81599152e-01 -5.50155878e-01 -4.72705588e-02
-9.42470074e-01 -9.38580334e-01 9.50566471e-01 1.42757028e-01
6.38271868e-01 -1.51298368e+00 -1.69817537e-01 4.04486984e-01
-3.58111203e-01 -1.28347731e+00 -7.40702391e-01 -3.09122711e-01
-4.57749069e-01 -1.38689506e+00 -5.78497529e-01 -8.47117126e-01
4.26800042e-01 6.75664306e-01 7.84964621e-01 1.13714702e-01
-5.97041957e-02 1.83503419e-01 -3.46786112e-01 -4.64604914e-01
-5.67005396e-01 5.12539744e-01 9.18324888e-02 3.43229026e-01
5.80915272e-01 -6.85979724e-01 -2.91117132e-01 5.41710436e-01
-7.99896777e-01 -4.25367355e-01 3.31604332e-01 7.08563328e-01
-3.94497246e-01 1.54248551e-01 7.81694651e-01 -1.46044683e+00
9.87940729e-01 -1.27899587e+00 -3.48744035e-01 -2.82966882e-01
-3.43070596e-01 -5.17638743e-01 1.05517375e+00 -8.15132499e-01
-6.12750292e-01 -3.83088946e-01 7.82331303e-02 -2.65149146e-01
1.94904074e-01 4.14532214e-01 4.96977687e-01 4.63155545e-02
7.47243047e-01 3.43506634e-01 -5.34339212e-02 -3.74920279e-01
1.93634599e-01 1.20374846e+00 2.45251339e-02 1.70380324e-01
1.12124813e+00 6.64470792e-01 -5.88808000e-01 -1.05329895e+00
-8.18816304e-01 -9.45961654e-01 -2.51914054e-01 -2.12352663e-01
7.81477213e-01 -6.02950037e-01 -1.28681421e+00 1.13829041e+00
-1.52219343e+00 -1.21805951e-01 3.22693199e-01 4.36195172e-02
-9.13162082e-02 4.39235061e-01 -1.39044011e+00 -8.90275359e-01
-5.23311257e-01 -9.93495882e-01 6.03525341e-01 5.54398745e-02
-2.51388788e-01 -1.23820710e+00 -1.00783445e-02 3.61421376e-01
8.00867140e-01 2.05286875e-01 4.55133826e-01 -1.25846112e+00
-4.79122937e-01 -8.19946289e-01 -5.95116198e-01 3.10979396e-01
3.46571922e-01 -1.46013005e-02 -1.00203466e+00 -1.28775269e-01
1.31820664e-02 -2.97846138e-01 9.70018864e-01 -9.91906971e-02
9.89719570e-01 -1.00151134e+00 -4.74448413e-01 -3.82879283e-03
8.50100577e-01 -1.04567558e-01 2.52913147e-01 3.23013037e-01
1.13953817e+00 4.89509314e-01 4.71447445e-02 5.40066004e-01
4.82923865e-01 3.41480523e-01 8.31507087e-01 2.51966804e-01
3.24615806e-01 -4.89019513e-01 7.45654225e-01 9.54755187e-01
1.28705770e-01 -3.60823810e-01 -1.03262722e+00 5.88674068e-01
-1.94956040e+00 -1.40807343e+00 -5.38312018e-01 1.60067618e+00
6.88901186e-01 1.58158958e-01 7.36736774e-01 -1.61696076e-01
1.38719761e+00 6.39370620e-01 -6.70647562e-01 -3.10165167e-01
3.41986358e-01 -1.37125477e-01 1.00042176e+00 5.24422050e-01
-1.37508321e+00 1.23534691e+00 5.46959591e+00 9.10366595e-01
-1.28432703e+00 5.49191952e-01 4.37405169e-01 1.21421441e-01
2.55088061e-01 -4.11079675e-01 -1.08855188e+00 9.67519164e-01
1.33026385e+00 -2.38815576e-01 5.93818724e-01 1.05715346e+00
4.84647006e-01 3.99420559e-01 -3.78951460e-01 6.08797014e-01
1.46025151e-01 -1.34703112e+00 -2.43019536e-01 2.43570954e-01
7.15009511e-01 6.51402652e-01 3.12441736e-01 4.94797945e-01
6.82168484e-01 -8.38104367e-01 5.32221019e-01 3.51444453e-01
1.68757632e-01 -5.01410067e-01 8.41695249e-01 7.50455737e-01
-1.03147233e+00 -4.18985367e-01 -3.42005104e-01 -1.97940785e-02
2.30680853e-01 6.51321471e-01 -9.89736080e-01 -2.19768986e-01
4.49859351e-01 1.21782815e+00 -6.43904865e-01 7.35755682e-01
-3.79794687e-01 1.07292771e+00 -7.06952438e-02 -6.78288102e-01
6.55000091e-01 3.17596160e-02 7.46327221e-01 1.36521804e+00
-7.95161501e-02 -3.34784925e-01 2.72124380e-01 9.07600343e-01
-4.64863896e-01 4.08048183e-02 -9.73672390e-01 -6.87786996e-01
4.31745946e-01 1.34283018e+00 -6.75542831e-01 -3.99808496e-01
-4.31928962e-01 8.90800714e-01 4.11980540e-01 2.16360822e-01
-1.07376730e+00 -5.42104483e-01 6.16406679e-01 5.12691617e-01
2.03760371e-01 -2.50454694e-01 -6.51138946e-02 -1.18604517e+00
-2.79514283e-01 -6.53714657e-01 1.82474151e-01 -3.14876497e-01
-1.88898003e+00 6.52508199e-01 -3.94785196e-01 -9.73661065e-01
-4.30407435e-01 -5.65299094e-01 -1.09988678e+00 7.63855696e-01
-1.43965781e+00 -1.23902535e+00 -9.31321681e-02 4.95997101e-01
4.26839381e-01 -4.87073362e-01 4.20364201e-01 3.20161581e-01
-9.57472026e-01 3.99300814e-01 3.33043844e-01 9.05717015e-01
3.91851544e-01 -9.38362837e-01 7.52276838e-01 5.61881125e-01
-3.05989921e-01 7.76817441e-01 6.80043995e-01 -8.09692264e-01
-1.11633301e+00 -1.58095193e+00 1.11256015e+00 -7.06475496e-01
1.67828679e+00 -3.47248077e-01 -7.48161554e-01 9.36481655e-01
1.15110382e-01 -9.10664443e-03 4.85900521e-01 -9.16698053e-02
-6.02127731e-01 2.55418837e-01 -1.13507760e+00 7.64820516e-01
7.31528342e-01 -8.12088013e-01 -9.61522609e-02 9.06291306e-01
7.33205557e-01 5.83017692e-02 -2.76821464e-01 -2.25886822e-01
3.74323487e-01 -9.04888988e-01 8.00670087e-01 -8.64542365e-01
4.55746740e-01 2.26145685e-01 3.80004048e-01 -1.12336802e+00
-2.82777220e-01 -7.60331631e-01 -3.62748146e-01 1.24171901e+00
3.05862665e-01 -1.26224232e+00 8.44410539e-01 2.07416013e-01
2.57548898e-01 -4.89388764e-01 -6.48671746e-01 -7.29297459e-01
8.35135132e-02 -2.67256379e-01 1.95940226e-01 1.11179733e+00
2.20018327e-01 4.33877200e-01 -7.09982514e-01 3.15179750e-02
4.27438289e-01 -1.54180333e-01 8.38295698e-01 -1.21536899e+00
-5.44256829e-02 -7.80922055e-01 -3.35767537e-01 -9.27553713e-01
6.15138829e-01 -7.86124468e-01 -2.99493492e-01 -1.24581790e+00
1.10648617e-01 -2.01582775e-01 6.49124058e-03 5.24538636e-01
3.27604786e-02 6.17560983e-01 5.84775917e-02 6.43779576e-01
-7.46173084e-01 3.62594008e-01 9.98954594e-01 -4.72821504e-01
-2.06092000e-01 6.21656001e-01 -7.13346004e-01 1.16134298e+00
1.16550231e+00 -8.24360132e-01 2.76683108e-03 -1.28652543e-01
3.00459415e-01 -3.67562801e-01 3.91739011e-01 -4.80434477e-01
3.67143363e-01 -4.72184792e-02 -3.96823674e-01 -3.74100775e-01
2.64470637e-01 -5.62392294e-01 -5.48419416e-01 7.26087749e-01
-2.07061321e-01 1.36560023e-01 -1.35479569e-01 1.00960243e+00
-2.97090597e-02 -4.20829177e-01 9.47443604e-01 -3.41083854e-01
-2.16019034e-01 5.16447842e-01 -1.08571625e+00 2.46903777e-01
9.77527738e-01 9.48582143e-02 -6.42482638e-01 -7.63900697e-01
-5.42235792e-01 1.22586027e-01 7.63997063e-02 5.86636901e-01
1.97312951e-01 -1.20146358e+00 -5.50904155e-01 -1.30621299e-01
-1.35611594e-01 -6.29396617e-01 -1.40854299e-01 9.27251816e-01
-3.91810328e-01 5.68117082e-01 3.00828755e-01 -2.75225639e-01
-1.15426981e+00 4.34336632e-01 1.73053101e-01 -5.76885104e-01
-4.75599527e-01 6.32505298e-01 -2.66683906e-01 -5.96937120e-01
3.11608255e-01 1.57464415e-01 -6.01820707e-01 5.85622191e-01
7.35836208e-01 7.25305915e-01 -2.13295504e-01 -1.07703865e+00
-1.48503482e-01 -1.81633592e-01 -3.53220195e-01 2.43925154e-01
1.25698888e+00 1.39004856e-01 -5.19206762e-01 7.26382732e-01
1.46577728e+00 1.90637961e-01 -4.13657695e-01 -5.61429441e-01
2.18127251e-01 -5.02829194e-01 -1.44011497e-01 -1.18379831e-01
-1.16518056e+00 9.41991627e-01 -9.60583705e-03 1.19461715e+00
2.69978732e-01 -7.44599402e-02 1.63217342e+00 6.42648757e-01
3.00797135e-01 -8.16978216e-01 5.99145353e-01 1.22128713e+00
4.43382323e-01 -1.35282874e+00 -5.64155221e-01 -3.20214301e-01
-5.65527439e-01 9.75388885e-01 7.17458725e-01 -4.61049139e-01
1.00171506e+00 -2.04184815e-01 -3.35846581e-02 -2.90931046e-01
-5.26589513e-01 -3.04708213e-01 -3.31873715e-01 2.95481443e-01
2.98425078e-01 -7.26456419e-02 -2.01669574e-01 4.00253415e-01
-1.57324582e-01 -5.18477321e-01 8.39170277e-01 7.13162541e-01
-8.16764951e-01 -8.28281760e-01 -2.19566405e-01 7.60967791e-01
-9.35250163e-01 -2.13719860e-01 -6.45493150e-01 5.42486727e-01
-2.50778377e-01 1.22862196e+00 2.57567782e-02 -6.04368806e-01
-2.04356499e-02 4.69112620e-02 -4.92260605e-01 -9.37815785e-01
-9.81873333e-01 -5.45938611e-01 -1.87041555e-02 -2.07650691e-01
-9.57897455e-02 -4.52150583e-01 -9.58683550e-01 -1.07875562e+00
-6.41646445e-01 2.56412119e-01 6.12584531e-01 8.14894736e-01
2.58007169e-01 2.78158277e-01 1.15247595e+00 -9.59345043e-01
-7.52589941e-01 -1.38486350e+00 -5.20864189e-01 4.47481066e-01
4.43482041e-01 -4.99725908e-01 -8.10890973e-01 -3.34512055e-01]
|
[8.131558418273926, 10.166800498962402]
|
28a6d1e0-2701-4912-8229-0d9e0cb22275
|
multi-view-masked-world-models-for-visual
|
2302.02408
| null |
https://arxiv.org/abs/2302.02408v2
|
https://arxiv.org/pdf/2302.02408v2.pdf
|
Multi-View Masked World Models for Visual Robotic Manipulation
|
Visual robotic manipulation research and applications often use multiple cameras, or views, to better perceive the world. How else can we utilize the richness of multi-view data? In this paper, we investigate how to learn good representations with multi-view data and utilize them for visual robotic manipulation. Specifically, we train a multi-view masked autoencoder which reconstructs pixels of randomly masked viewpoints and then learn a world model operating on the representations from the autoencoder. We demonstrate the effectiveness of our method in a range of scenarios, including multi-view control and single-view control with auxiliary cameras for representation learning. We also show that the multi-view masked autoencoder trained with multiple randomized viewpoints enables training a policy with strong viewpoint randomization and transferring the policy to solve real-robot tasks without camera calibration and an adaptation procedure. Video demonstrations are available at: https://sites.google.com/view/mv-mwm.
|
['Pieter Abbeel', 'Jinwoo Shin', 'Kimin Lee', 'Stephen James', 'Junsu Kim', 'Younggyo Seo']
|
2023-02-05
| null | null | null | null |
['camera-calibration']
|
['computer-vision']
|
[-1.56009957e-01 1.13710962e-01 -2.92623788e-01 -2.59631962e-01
-3.67373258e-01 -8.18239868e-01 4.68159169e-01 -8.23823929e-01
-5.43270968e-02 4.50898677e-01 2.63406396e-01 -2.88493070e-03
1.38366923e-01 -5.77589571e-01 -1.27765071e+00 -6.75846875e-01
3.47758025e-01 4.39005464e-01 -4.43209894e-02 -2.31872723e-01
1.33577615e-01 5.90309083e-01 -1.55555391e+00 4.22751516e-01
2.08392978e-01 5.35718977e-01 8.47284734e-01 9.88083839e-01
6.81066811e-01 1.20529068e+00 -3.07073206e-01 9.21939686e-02
7.54540503e-01 -5.65306135e-02 -3.84260565e-01 5.45853198e-01
3.48641872e-01 -1.05258548e+00 -9.49190557e-01 9.58214343e-01
2.76024222e-01 4.18324947e-01 3.64037424e-01 -1.28643346e+00
-8.92393410e-01 4.01639789e-01 -5.87309539e-01 -9.63860229e-02
5.46502829e-01 5.77838659e-01 7.31416702e-01 -7.22844362e-01
9.36895847e-01 1.36789334e+00 4.67385314e-02 8.24410617e-01
-1.10531151e+00 -4.48863983e-01 3.52089405e-01 2.76059687e-01
-8.45118523e-01 -4.20951337e-01 7.74844646e-01 -6.34030759e-01
8.51408184e-01 -3.20370674e-01 7.13809729e-01 1.64962661e+00
4.04908419e-01 7.83109248e-01 1.08496392e+00 -3.54160547e-01
-8.75049233e-02 1.16651647e-01 -3.25081140e-01 9.47718918e-01
1.49059519e-01 5.98600388e-01 -3.51179749e-01 7.02997223e-02
1.49125099e+00 5.24400055e-01 -5.23099840e-01 -1.06289339e+00
-1.60934401e+00 8.77369881e-01 5.95982075e-01 -1.62097454e-01
-6.77333117e-01 3.60640854e-01 -1.21819675e-02 3.60132605e-01
-7.17275729e-03 7.40415394e-01 -5.53953290e-01 1.48199424e-01
-1.82545204e-02 9.56931442e-04 5.46517074e-01 1.44959104e+00
8.37533534e-01 4.15859073e-01 3.56949598e-01 5.22868276e-01
1.08066417e-01 7.99030244e-01 4.27521110e-01 -1.84972346e+00
5.94177425e-01 3.50985765e-01 2.63041228e-01 -6.23586535e-01
-2.01311469e-01 3.14127982e-01 -4.73000079e-01 9.18806255e-01
3.30866367e-01 -3.33992690e-01 -1.08058059e+00 1.75760126e+00
2.83384562e-01 -2.24633217e-01 2.36505255e-01 1.02698350e+00
3.02880883e-01 5.72601616e-01 -4.96410340e-01 2.96241790e-01
1.01698279e+00 -1.15821993e+00 -6.60098493e-01 -4.75189596e-01
1.13109194e-01 -5.18856347e-01 1.00925148e+00 5.94572961e-01
-1.09733450e+00 -6.63822412e-01 -1.13503683e+00 -3.47273380e-01
-3.32168609e-01 4.50022072e-01 5.40543199e-01 -1.97697774e-01
-8.49206150e-01 4.78837848e-01 -1.40269244e+00 -4.49320197e-01
1.34424940e-01 3.54081631e-01 -9.54024613e-01 -3.64912182e-01
-6.06834590e-01 1.24738348e+00 3.89578909e-01 1.47147523e-02
-1.75198042e+00 -1.69953778e-01 -1.05288947e+00 -1.75528407e-01
6.37814820e-01 -1.07846797e+00 1.24801874e+00 -9.71441567e-01
-1.78750670e+00 6.41829908e-01 -8.58682320e-02 -1.01090707e-01
3.61342072e-01 -4.97215837e-01 8.15037116e-02 4.36381489e-01
3.33232023e-02 8.05941045e-01 1.23028064e+00 -1.83382142e+00
-4.03609037e-01 -6.56942844e-01 6.72038496e-01 3.63129467e-01
1.59543708e-01 -3.06873441e-01 -1.64304256e-01 -4.02333468e-01
2.73959637e-01 -1.33896673e+00 -3.27451944e-01 1.88002482e-01
-2.43884146e-01 2.69957751e-01 7.86959469e-01 -7.13653743e-01
7.11342841e-02 -1.99827290e+00 9.30730164e-01 -2.58179218e-01
1.72237039e-01 -2.92361289e-01 -2.14645699e-01 5.53912699e-01
-2.20322207e-01 -4.31464344e-01 2.69507408e-01 2.38943454e-02
-2.29770437e-01 4.82489228e-01 -3.46909195e-01 5.95194101e-01
-2.48036459e-02 7.56322205e-01 -7.69147158e-01 2.06836283e-01
5.62208951e-01 4.99295622e-01 -7.52418995e-01 6.56441152e-01
-3.88287634e-01 9.82234061e-01 -4.24513489e-01 4.05699015e-01
3.39304954e-01 -4.01846886e-01 2.95581430e-01 -2.53909796e-01
7.62066022e-02 -1.32963687e-01 -9.16355848e-01 2.04018760e+00
-6.26699567e-01 6.33548379e-01 5.35166979e-01 -9.60773170e-01
5.02334774e-01 3.09387594e-01 4.60632086e-01 -1.13436639e-01
3.70941490e-01 -1.66863278e-01 -9.52899456e-02 -1.05294788e+00
4.02053505e-01 -1.11994617e-01 7.67117068e-02 4.28235859e-01
4.29638386e-01 -4.93118376e-01 -2.14139417e-01 1.01827450e-01
8.84786367e-01 6.52345479e-01 4.25387383e-01 3.35515499e-01
5.05029671e-02 -2.71829456e-04 2.74091601e-01 5.19107342e-01
-5.15805818e-02 6.10069931e-01 2.58007973e-01 -5.16178846e-01
-1.33035457e+00 -1.13142717e+00 5.84940851e-01 1.01652670e+00
2.21419647e-01 4.57165129e-02 -3.58296961e-01 -5.89017391e-01
2.19215289e-01 6.71679974e-01 -6.76986754e-01 -2.74928868e-01
-7.31273711e-01 1.26324758e-01 -5.48211448e-02 7.76797593e-01
7.34695792e-02 -1.06253302e+00 -1.03070295e+00 -3.04677993e-01
-1.57479569e-01 -1.42720664e+00 -2.24428654e-01 4.14130121e-01
-9.05770361e-01 -1.28424752e+00 -5.93028784e-01 -7.34305918e-01
8.64622653e-01 7.92591751e-01 7.38520384e-01 -5.12678683e-01
-4.25349772e-02 1.05414736e+00 -2.98205972e-01 -2.87506610e-01
-5.74420452e-01 -2.86571145e-01 3.29329133e-01 -2.96276122e-01
-1.20640419e-01 -7.86102712e-01 -3.73429626e-01 1.17884398e-01
-7.82129526e-01 1.04556918e-01 7.22084820e-01 1.04551852e+00
3.04336399e-01 -5.62386513e-01 -4.52276058e-02 -5.01425982e-01
5.25886528e-02 -4.41506147e-01 -8.82008493e-01 1.61044136e-01
2.47486141e-02 1.11199759e-01 4.85630661e-01 -7.01691866e-01
-1.00312936e+00 5.14539778e-01 2.77107358e-01 -1.37741196e+00
-4.21059132e-01 7.65167177e-02 -2.55081594e-01 -3.38758714e-02
7.20013916e-01 7.82235414e-02 3.60150903e-01 -2.08833933e-01
9.06109631e-01 4.74871665e-01 4.60704952e-01 -6.43631220e-01
8.14887702e-01 8.94571602e-01 -3.77674848e-01 -6.08384430e-01
-4.77472216e-01 -2.18291715e-01 -9.24672127e-01 -3.11088979e-01
1.11461437e+00 -1.46504629e+00 -8.94672871e-01 2.54817158e-01
-1.25243390e+00 -6.66819870e-01 -2.89957553e-01 8.98899317e-01
-1.25191569e+00 8.85616913e-02 -4.89579052e-01 -4.58171040e-01
2.88938195e-01 -1.47100604e+00 1.23865640e+00 1.60346746e-01
2.03296736e-01 -6.53551519e-01 -1.51344195e-01 4.56705779e-01
-6.60748705e-02 1.57529652e-01 6.52368009e-01 -2.63857365e-01
-9.59547043e-01 -5.05303256e-02 1.99915424e-01 4.69519049e-01
3.67339045e-01 -2.10349813e-01 -1.11823916e+00 -5.98990858e-01
2.91948766e-01 -6.79329693e-01 6.81257248e-01 6.33074701e-01
1.30353928e+00 -2.23774031e-01 -2.63833970e-01 7.53415823e-01
1.45206213e+00 4.27839756e-01 4.07646269e-01 3.22109371e-01
8.59173536e-01 5.59829533e-01 3.93409342e-01 4.41983819e-01
3.70975584e-01 5.16823113e-01 8.84132385e-01 3.34918499e-01
1.48263454e-01 -4.83756095e-01 6.26776636e-01 8.31312597e-01
-3.11617821e-01 -2.22858012e-01 -5.89334011e-01 5.59673548e-01
-1.66178477e+00 -1.08103621e+00 6.41296327e-01 1.95468366e+00
2.22500622e-01 -2.51867026e-01 -5.00213504e-02 -4.04042810e-01
7.12051868e-01 3.39608192e-01 -9.17512119e-01 -2.24618256e-01
1.26741573e-01 -1.34711847e-01 6.66718185e-01 5.46359539e-01
-1.07342017e+00 9.38039780e-01 6.22538233e+00 -1.56928346e-01
-1.21493196e+00 1.15519024e-01 -1.79445848e-01 -1.58961162e-01
-3.13828774e-02 1.00117683e-01 -3.80419582e-01 -4.72418256e-02
4.06858355e-01 1.77985087e-01 1.13771605e+00 1.16515124e+00
4.06378098e-02 -3.62612791e-02 -1.31515515e+00 9.74367797e-01
3.60923052e-01 -1.04056156e+00 5.35230190e-02 2.17053160e-01
9.41304505e-01 4.72238868e-01 4.01731916e-02 2.44829133e-01
1.00729525e+00 -5.64679563e-01 7.67336607e-01 3.86811882e-01
5.30170381e-01 -1.82675466e-01 7.80783296e-02 6.44307554e-01
-6.42101169e-01 -7.36703694e-01 -4.38690037e-01 -1.43807039e-01
2.47378483e-01 -1.36718735e-01 -7.27419913e-01 4.88960892e-01
6.60034716e-01 1.07072890e+00 -1.22620620e-01 4.58064944e-01
-4.43270773e-01 -1.64442345e-01 -2.32354239e-01 2.43526220e-01
-1.03422843e-01 -3.11027676e-01 6.35520399e-01 2.22687647e-01
3.47827315e-01 1.73158422e-01 3.93050760e-01 8.75176430e-01
-1.24541754e-02 -7.87618637e-01 -1.32337391e+00 7.02392906e-02
3.44965607e-01 9.74785030e-01 -2.18305826e-01 -2.83856034e-01
-6.26571119e-01 1.05032051e+00 5.64700305e-01 8.21889162e-01
-7.90684044e-01 7.04189986e-02 7.35104620e-01 -2.16730043e-01
7.01707423e-01 -6.61209285e-01 2.13672087e-01 -1.68806112e+00
2.90419161e-02 -1.25871372e+00 -1.46981059e-02 -1.45799160e+00
-1.10770524e+00 3.89088273e-01 2.94853240e-01 -1.56022549e+00
-4.50782806e-01 -1.03955984e+00 -1.90164089e-01 3.95116955e-01
-1.30978072e+00 -1.34238923e+00 -5.26501656e-01 7.89656639e-01
8.82852852e-01 -3.62249523e-01 8.11754465e-01 -2.47652709e-01
-2.14763999e-01 -5.77235259e-02 1.50492132e-01 1.41805843e-01
8.38049948e-01 -1.19500721e+00 8.96351337e-02 5.58638394e-01
2.73478508e-01 8.06192338e-01 4.87743229e-01 -3.69474649e-01
-1.96780300e+00 -7.62340784e-01 -3.97143602e-01 -8.02245617e-01
6.02179110e-01 -2.44414017e-01 -5.01602232e-01 1.73626435e+00
7.09927857e-01 2.14113235e-01 1.41190007e-01 -2.53302455e-01
-5.05630910e-01 7.99011812e-02 -8.57126892e-01 5.75080335e-01
9.58569229e-01 -8.43592942e-01 -7.19216168e-01 3.03422660e-01
7.32408106e-01 -9.21678007e-01 -9.92346287e-01 2.83721268e-01
8.82932603e-01 -7.96065032e-01 1.16669691e+00 -9.81065452e-01
8.13069642e-01 -2.92633742e-01 -5.92141390e-01 -1.78195894e+00
-3.81949902e-01 -3.66496205e-01 -1.20554328e-01 4.44025040e-01
1.87226713e-01 -7.75111198e-01 4.91327286e-01 3.08864117e-01
-1.57814652e-01 -3.70323509e-01 -5.98343909e-01 -6.85165882e-01
2.23520651e-01 1.14667706e-01 2.15088055e-01 8.40394735e-01
-8.70343447e-02 1.80282429e-01 -5.50239563e-01 6.53500795e-01
3.62362474e-01 3.66765171e-01 1.26931107e+00 -7.79541194e-01
-6.92077100e-01 1.93381563e-01 -3.24768335e-01 -1.24830782e+00
4.33619112e-01 -5.76011002e-01 3.21229815e-01 -1.35815394e+00
2.85903126e-01 2.32301950e-01 7.96265975e-02 4.64912891e-01
-5.15152514e-03 -4.89254802e-01 5.64390481e-01 2.31936336e-01
-2.67166644e-01 7.66173542e-01 1.71277797e+00 -1.99876189e-01
3.21594141e-02 -1.23009525e-01 -4.40256834e-01 8.85584176e-01
9.06387150e-01 -3.07878554e-01 -5.63130260e-01 -1.08711517e+00
-1.00302793e-01 4.20350581e-01 6.06630623e-01 -9.10400689e-01
1.55560017e-01 -4.68437523e-01 5.81639349e-01 -2.80761480e-01
9.17483509e-01 -1.16456580e+00 5.70553429e-02 3.75832647e-01
-4.24377888e-01 5.22296727e-01 8.58911574e-02 1.02057803e+00
1.73430145e-02 6.32006628e-03 6.60007000e-01 -8.58370543e-01
-6.30006313e-01 1.33603647e-01 -4.99134898e-01 -2.33494475e-01
1.15762973e+00 1.27429068e-01 -5.91414750e-01 -7.03522861e-01
-1.06007636e+00 2.26631567e-01 1.01394892e+00 7.66478300e-01
7.28943646e-01 -1.24397802e+00 -3.01469833e-01 4.16820556e-01
9.76207256e-02 1.56672210e-01 1.54341474e-01 4.86011535e-01
-5.68878233e-01 1.70515463e-01 -6.37102664e-01 -7.49315739e-01
-9.88348424e-01 1.10926914e+00 3.85248095e-01 2.23884702e-01
-8.20331872e-01 5.30419767e-01 4.68070358e-01 -8.63624513e-01
1.28554985e-01 -4.68684107e-01 7.44501427e-02 -5.21121442e-01
1.97571591e-01 2.32451081e-01 -4.96794790e-01 -5.75827360e-01
1.38725951e-01 8.03134143e-01 1.50760457e-01 -4.28058624e-01
1.59364367e+00 -1.59441888e-01 4.35595214e-02 6.49878681e-01
1.21935642e+00 -3.64605077e-02 -1.80497694e+00 4.26243506e-02
-5.60490370e-01 -4.67326194e-01 -1.22452892e-01 -4.39841479e-01
-1.29104769e+00 1.18720067e+00 3.68152052e-01 -2.38757253e-01
9.07722414e-01 -3.77267748e-02 2.99917698e-01 9.62344825e-01
6.19275510e-01 -8.11689198e-01 4.71577317e-01 5.21397650e-01
1.34239709e+00 -1.48237133e+00 4.62250710e-02 -1.79151997e-01
-1.04206967e+00 1.13505554e+00 9.72045958e-01 -7.14245558e-01
7.44496524e-01 3.07036489e-01 1.78676054e-01 -3.07350218e-01
-1.00635183e+00 -2.01007605e-01 -2.95726925e-01 8.30126107e-01
-2.79546380e-01 2.87480559e-02 7.70679533e-01 4.96216826e-02
-1.51424408e-02 8.56563449e-03 8.73672009e-01 1.19608307e+00
-2.88825095e-01 -7.37414420e-01 -3.16341758e-01 4.58786497e-03
-9.41884238e-03 4.54772562e-01 -3.81609350e-01 9.10560071e-01
-1.66506052e-01 8.13647926e-01 -1.90148994e-01 -6.11099958e-01
4.45263475e-01 2.00124066e-02 1.08149683e+00 -7.65981495e-01
-2.15858996e-01 4.27305065e-02 -2.61927605e-01 -9.19375122e-01
-7.58516371e-01 -7.02831268e-01 -8.67536366e-01 -7.66297942e-03
-4.02359098e-01 -4.06612873e-01 6.66345417e-01 6.43578470e-01
4.00475979e-01 5.68198800e-01 8.14720452e-01 -1.50360405e+00
-9.54028666e-01 -9.32467461e-01 -3.98310155e-01 2.30104089e-01
9.06713009e-01 -9.65658605e-01 -3.79653484e-01 3.19041789e-01]
|
[4.631661415100098, 0.7069984674453735]
|
d669c8cf-e501-4c0d-8859-c06843343f3b
|
deep-exposure-fusion-with-deghosting-via
|
2004.09089
| null |
https://arxiv.org/abs/2004.09089v1
|
https://arxiv.org/pdf/2004.09089v1.pdf
|
Deep Exposure Fusion with Deghosting via Homography Estimation and Attention Learning
|
Modern cameras have limited dynamic ranges and often produce images with saturated or dark regions using a single exposure. Although the problem could be addressed by taking multiple images with different exposures, exposure fusion methods need to deal with ghosting artifacts and detail loss caused by camera motion or moving objects. This paper proposes a deep network for exposure fusion. For reducing the potential ghosting problem, our network only takes two images, an underexposed image and an overexposed one. Our network integrates together homography estimation for compensating camera motion, attention mechanism for correcting remaining misalignment and moving pixels, and adversarial learning for alleviating other remaining artifacts. Experiments on real-world photos taken using handheld mobile phones show that the proposed method can generate high-quality images with faithful detail and vivid color rendition in both dark and bright areas.
|
['Yung-Yu Chuang', 'Sheng-Yeh Chen']
|
2020-04-20
| null | null | null | null |
['homography-estimation']
|
['computer-vision']
|
[ 7.22232401e-01 -2.95892060e-01 3.64050448e-01 -1.64046437e-01
-4.36040193e-01 -6.35763764e-01 3.91196847e-01 -5.16616523e-01
-2.48314798e-01 8.54629219e-01 1.55120060e-01 -8.18353444e-02
3.80834758e-01 -6.90788865e-01 -7.53612936e-01 -6.75099790e-01
2.04816893e-01 -6.06244028e-01 2.63909817e-01 -1.32807821e-01
1.95155486e-01 1.73726395e-01 -1.30214047e+00 1.65212870e-01
1.15466344e+00 8.78582954e-01 3.06974143e-01 8.64102185e-01
1.69157863e-01 7.62817860e-01 -8.68644416e-01 -4.54883099e-01
7.65963197e-01 -4.74927604e-01 -3.03809017e-01 5.99684715e-01
1.00078821e+00 -1.10087705e+00 -7.11009145e-01 1.36300170e+00
5.76294959e-01 2.50499249e-01 1.18939884e-01 -1.32942927e+00
-1.00312197e+00 1.69683680e-01 -1.08914900e+00 3.16558778e-01
4.94392484e-01 6.46038234e-01 -9.77477133e-02 -6.51597917e-01
2.86761552e-01 1.24822831e+00 7.35936224e-01 5.45786440e-01
-1.09825361e+00 -8.55021060e-01 1.89753994e-02 8.74475464e-02
-1.32110345e+00 -5.58125734e-01 8.52122664e-01 -1.05308965e-01
4.19895828e-01 2.45893314e-01 5.12352347e-01 1.11241353e+00
4.66227174e-01 2.51142114e-01 1.33531582e+00 -3.38169605e-01
-1.55951723e-01 2.18947336e-01 -2.27415621e-01 5.36680818e-01
4.42370892e-01 3.43354732e-01 -1.62225798e-01 9.77665111e-02
1.06433737e+00 3.28942299e-01 -9.69550431e-01 1.29084647e-01
-1.13078642e+00 2.58274615e-01 3.24687809e-01 4.08014841e-02
-3.44931245e-01 2.09032297e-01 -9.11397859e-02 1.25136882e-01
1.14536323e-01 2.19108030e-01 -8.20057392e-02 3.24857831e-01
-8.92611086e-01 -8.62137005e-02 2.93673038e-01 1.05124950e+00
8.15581560e-01 4.66035485e-01 -1.69632867e-01 7.82717884e-01
1.07695960e-01 7.50240088e-01 4.24928039e-01 -1.11949146e+00
6.63395882e-01 5.79298437e-02 4.73016441e-01 -1.31365395e+00
9.81169846e-03 -3.12218703e-02 -1.18332624e+00 6.97564244e-01
3.61312956e-01 -5.44372082e-01 -1.01871443e+00 1.69475007e+00
2.49274775e-01 4.06436205e-01 9.62659642e-02 1.18590558e+00
6.16062343e-01 9.11799192e-01 -1.04891658e-01 -5.11468053e-01
1.30088615e+00 -1.12632418e+00 -1.27223599e+00 -5.22969782e-01
-3.73898923e-01 -1.01899564e+00 1.08055961e+00 6.10351920e-01
-1.58564949e+00 -8.47411752e-01 -1.46967781e+00 -2.94031620e-01
4.28273790e-02 -1.57836363e-01 1.56117022e-01 8.18890333e-01
-9.67817485e-01 5.34905612e-01 -3.04873586e-01 1.72500446e-01
3.79169255e-01 2.31055707e-01 -3.58373493e-01 -3.11578810e-01
-1.35395825e+00 6.84980571e-01 1.81263521e-01 2.61924505e-01
-8.25186193e-01 -6.58650398e-01 -8.15353692e-01 -3.24692056e-02
2.93056965e-01 -6.02520168e-01 1.02239227e+00 -1.47992003e+00
-1.59969890e+00 5.22940814e-01 2.26509809e-01 -1.75201148e-01
6.61800742e-01 -3.25879276e-01 -8.34788799e-01 1.46622673e-01
-1.95011854e-01 5.57801783e-01 1.36195433e+00 -1.60645282e+00
-5.51367164e-01 -1.65837094e-01 -4.76749521e-03 4.18563783e-01
-3.62815082e-01 -2.78663486e-01 -6.77532792e-01 -7.05486119e-01
-1.36025578e-01 -6.74381018e-01 -8.39769021e-02 1.70433640e-01
-3.25797051e-01 7.57548809e-01 1.24817348e+00 -1.13849640e+00
1.20360684e+00 -2.09319329e+00 -2.62473702e-01 -2.02983677e-01
1.45229116e-01 4.82022136e-01 -2.84019142e-01 1.30796567e-01
-1.97965935e-01 8.24288949e-02 -3.16361338e-01 -2.42758110e-01
-5.48201859e-01 -6.84466213e-02 -5.62173009e-01 5.68551004e-01
2.33510062e-02 5.43988287e-01 -9.40224826e-01 -4.53407615e-01
6.31714523e-01 8.07257354e-01 -6.36387244e-02 2.69442290e-01
1.43675610e-01 4.77462232e-01 2.04949141e-01 4.89321679e-01
1.51118898e+00 1.92053486e-02 6.97312085e-03 -4.47689861e-01
-4.21592128e-03 -3.98234189e-01 -1.41912341e+00 1.43701541e+00
-5.47979772e-01 1.06034553e+00 3.10266495e-01 -3.32903713e-02
6.42847836e-01 1.66136220e-01 1.47295326e-01 -9.70596433e-01
2.74338007e-01 -8.36674571e-02 -2.51559049e-01 -6.39922976e-01
8.27430248e-01 -2.75309652e-01 1.75708473e-01 3.47231120e-01
-3.57373446e-01 -7.11280033e-02 -2.01253995e-01 5.11463247e-02
7.72668898e-01 -1.18737034e-01 7.54797161e-02 1.68685645e-01
5.77969790e-01 -7.20051885e-01 7.58952677e-01 5.37929714e-01
-2.53295332e-01 1.11609411e+00 1.83738582e-02 -4.66139436e-01
-1.33890378e+00 -1.02135932e+00 1.33111849e-01 4.40150559e-01
8.99188936e-01 1.21768057e-01 -7.84845114e-01 -2.81984657e-01
-4.04048622e-01 4.92632926e-01 -4.13424999e-01 -1.87928319e-01
-8.34408402e-01 -5.45725465e-01 4.58820969e-01 3.28057647e-01
1.11463845e+00 -9.43825662e-01 -5.74897468e-01 9.91259143e-02
-4.37508553e-01 -1.19018137e+00 -1.09982991e+00 -3.59647274e-01
-4.96329904e-01 -1.05092192e+00 -1.07088900e+00 -7.06545115e-01
7.51883209e-01 9.31609929e-01 9.36939001e-01 2.58399785e-01
-3.50634634e-01 -5.66566549e-02 3.83438282e-02 -1.08612314e-01
-4.38789040e-01 -8.16674709e-01 -1.19217308e-02 1.93324924e-01
-4.85261202e-01 -4.92498338e-01 -1.16722333e+00 3.06343585e-01
-1.44410002e+00 2.67871618e-01 2.98370302e-01 7.54851997e-01
1.32150933e-01 4.93829697e-01 1.10543877e-01 -5.48283100e-01
6.08441293e-01 -1.59386784e-01 -7.07687318e-01 2.54293948e-01
-4.89823163e-01 -3.04351538e-01 9.11426127e-01 -8.21266115e-01
-1.61764050e+00 -1.18564032e-01 2.82967448e-01 -7.47702420e-01
-5.03162295e-02 -4.21811163e-01 -3.93335223e-01 -3.04930508e-01
5.81097841e-01 2.47844994e-01 3.37417834e-02 -1.72311172e-01
4.08911616e-01 6.28471076e-01 1.00539994e+00 -4.15131450e-03
1.03518534e+00 6.96860015e-01 -1.43051296e-01 -7.16594100e-01
-3.86741549e-01 1.83253318e-01 -9.33914036e-02 -3.54119778e-01
8.90436232e-01 -1.06681538e+00 -7.89239109e-01 9.40178812e-01
-1.29857671e+00 -2.80985057e-01 1.27304867e-01 8.14961791e-02
-1.17619567e-01 8.75444233e-01 -7.35204995e-01 -5.98484993e-01
-5.35076499e-01 -1.16360533e+00 8.20991933e-01 8.45127404e-01
3.00419599e-01 -7.67395377e-01 -6.94847703e-02 1.65668681e-01
5.32202780e-01 3.57032984e-01 4.54671264e-01 5.51069319e-01
-8.31731260e-01 -9.96385664e-02 -3.19754452e-01 6.41458809e-01
4.73533183e-01 2.97954120e-02 -1.00634158e+00 -4.29621339e-01
2.45829001e-01 5.92539907e-02 7.22147346e-01 3.06100130e-01
1.18795288e+00 -5.51771045e-01 -5.62831126e-02 9.97577429e-01
1.94427037e+00 6.65255308e-01 1.28828967e+00 2.30853170e-01
8.62575531e-01 3.32409263e-01 3.72244984e-01 2.56316543e-01
-1.65059548e-02 5.85293233e-01 4.93927360e-01 -6.36246741e-01
-4.92941201e-01 -1.96084216e-01 3.50313336e-01 1.93555117e-01
2.68709689e-01 -9.26378191e-01 -4.19942707e-01 3.83284688e-01
-1.44182384e+00 -1.29598904e+00 -3.46021861e-01 2.33623576e+00
7.55800664e-01 -1.27363324e-01 -2.11517602e-01 -2.75419801e-02
1.18187010e+00 3.16644639e-01 -6.26621008e-01 -2.49301016e-01
-6.86628997e-01 -2.06213161e-01 8.10258150e-01 7.57260084e-01
-9.70507801e-01 6.01349115e-01 6.57816505e+00 5.63761890e-01
-1.20429039e+00 3.03632151e-02 8.89452338e-01 -1.73139289e-01
-3.82410944e-01 -6.37488961e-02 -4.27712679e-01 9.93790746e-01
3.76772344e-01 9.33202282e-02 7.60960579e-01 2.92414218e-01
2.47127563e-01 -3.07034671e-01 -4.64736104e-01 1.34459329e+00
3.74222040e-01 -1.31931436e+00 -2.57148426e-02 -4.36591543e-02
1.16647434e+00 -5.10573208e-01 5.57357490e-01 -2.42198870e-01
1.59754455e-01 -8.61699104e-01 6.30215406e-01 8.00204515e-01
9.94973123e-01 -6.81248546e-01 4.70072031e-01 1.33724529e-02
-9.92810249e-01 -2.20134005e-01 -2.29143158e-01 6.40110392e-03
5.24206817e-01 3.13706964e-01 -1.71497509e-01 4.04807627e-01
6.01897836e-01 4.14215624e-01 -6.66225612e-01 9.01340961e-01
-1.33921638e-01 3.53179686e-02 -8.89710858e-02 5.94531775e-01
6.99717477e-02 -4.00643438e-01 5.39669514e-01 1.01014018e+00
5.37848532e-01 3.60234916e-01 -2.55657315e-01 8.21832359e-01
-1.20067529e-01 -5.71038067e-01 -7.49219298e-01 4.09961224e-01
5.50746918e-01 1.32459891e+00 -5.15093565e-01 -4.81300503e-01
-4.21419442e-01 1.62971866e+00 -3.06574553e-01 6.94581568e-01
-1.21963489e+00 -6.51378810e-01 6.23667717e-01 9.50362384e-02
2.00101063e-01 1.76524315e-02 -3.21112812e-01 -1.23223233e+00
1.26443550e-01 -9.99727190e-01 -2.77935788e-02 -1.45893776e+00
-1.05441833e+00 6.12974524e-01 -2.01049462e-01 -1.39726985e+00
2.64155328e-01 -1.80603191e-01 -9.73785937e-01 1.04075050e+00
-1.33927345e+00 -6.95185781e-01 -8.36496115e-01 6.32404745e-01
6.95644975e-01 1.14258919e-02 1.45822108e-01 5.82662463e-01
-7.12099373e-01 5.84824860e-01 2.37228140e-01 -1.99482873e-01
1.09046030e+00 -1.09284782e+00 2.40094125e-01 1.37725961e+00
-3.48767638e-01 4.22397375e-01 8.13289821e-01 -6.93882763e-01
-1.20726812e+00 -1.03289974e+00 2.06535384e-01 -1.44638434e-01
-1.00489289e-01 -3.91399525e-02 -1.12719154e+00 3.89800698e-01
9.91767287e-01 -1.00517213e-01 2.02419072e-01 -8.57976019e-01
-1.08410552e-01 -3.29107046e-01 -1.27800381e+00 7.42436171e-01
6.87426805e-01 -2.95771331e-01 -1.14594316e-02 9.75149721e-02
8.60657990e-01 -6.44395769e-01 -5.53549945e-01 1.54551834e-01
4.63039696e-01 -1.26886821e+00 1.15101981e+00 3.16515684e-01
5.88744104e-01 -6.95424438e-01 8.30910727e-02 -1.10648501e+00
-2.38462776e-01 -1.24786580e+00 5.38162887e-02 1.42673600e+00
-1.55152276e-01 -3.63746881e-01 4.17164147e-01 7.41848350e-01
-9.12339762e-02 -1.86371356e-01 -3.77001584e-01 -5.31400502e-01
-2.79561400e-01 1.07525662e-01 8.41531813e-01 1.01221740e+00
-4.69720632e-01 4.12860811e-01 -1.25154853e+00 5.28352559e-01
1.01960969e+00 -1.95450902e-01 9.15316105e-01 -5.58534801e-01
-2.28511885e-01 -5.09826280e-02 -3.72229815e-02 -9.45097983e-01
-2.15855584e-01 2.01401878e-02 1.51590928e-01 -1.22470915e+00
2.72596389e-01 9.52817127e-02 1.20383807e-01 9.15747061e-02
-5.44625819e-01 5.43772280e-01 3.20126683e-01 -9.00029577e-03
-2.97100067e-01 4.28068936e-01 1.63791060e+00 -7.59294406e-02
-2.34604612e-01 -4.61717129e-01 -7.83164680e-01 8.54701877e-01
6.69949770e-01 -1.07167169e-01 -5.23686588e-01 -8.46355855e-01
9.52911153e-02 4.95428801e-01 6.97977602e-01 -1.25738227e+00
2.97447443e-01 -2.03302994e-01 1.01948225e+00 -6.36804521e-01
3.54743510e-01 -9.37642932e-01 6.34187341e-01 3.69155943e-01
-7.29775503e-02 2.53508151e-01 4.33451474e-01 7.96769559e-01
-1.19955614e-01 -7.28161335e-02 1.18819296e+00 -1.01277471e-01
-7.18294322e-01 3.05982351e-01 -2.41458923e-01 -2.53710717e-01
1.05084789e+00 -6.32805943e-01 -7.29196131e-01 -7.73101270e-01
-3.68824005e-01 4.72694486e-02 9.54680502e-01 3.82174820e-01
9.31326389e-01 -1.40586877e+00 -4.25943077e-01 4.96994108e-01
-4.02907312e-01 -2.44211897e-01 1.03693700e+00 2.97747970e-01
-8.60401630e-01 -9.21787173e-02 -5.02263963e-01 -1.53604999e-01
-1.36049533e+00 9.15808141e-01 5.07760406e-01 1.21701933e-01
-6.45303428e-01 6.88737988e-01 3.87720883e-01 3.54889661e-01
9.82755423e-02 -5.62343933e-02 1.48875207e-01 -4.76453871e-01
8.84112537e-01 5.87842464e-01 -3.48918766e-01 -6.21344745e-01
-8.72957408e-02 6.80763900e-01 -1.16183870e-01 -2.73244023e-01
7.78562963e-01 -8.55863750e-01 1.32900938e-01 1.61879752e-02
1.30840242e+00 4.95335311e-02 -1.84446740e+00 -7.78440684e-02
-8.77231717e-01 -1.11294186e+00 1.90422714e-01 -9.03292239e-01
-1.25636208e+00 9.04777110e-01 1.07037652e+00 1.70985073e-01
1.66850019e+00 -7.90743351e-01 1.17690909e+00 -1.75284967e-01
-3.51324156e-02 -9.12104011e-01 4.69703615e-01 4.91790250e-02
8.08562875e-01 -1.41588843e+00 1.55306861e-01 -3.13857526e-01
-8.22971344e-01 9.98332679e-01 8.94370794e-01 -2.50635535e-01
2.40151063e-02 1.98725209e-01 2.03608945e-01 2.93090492e-01
-5.16441822e-01 2.35020757e-01 1.75985992e-01 8.15493464e-01
2.52176337e-02 -4.26261961e-01 -3.60473245e-02 2.87916102e-02
2.34439239e-01 -2.90261447e-01 9.40007031e-01 7.00410783e-01
-3.35768819e-01 -8.05289447e-01 -9.20681715e-01 -1.14215635e-01
-6.22076869e-01 -1.65536374e-01 -8.27415660e-02 7.76143909e-01
4.72525775e-01 1.16765428e+00 2.14258209e-01 -3.86152536e-01
1.26661330e-01 -5.00640690e-01 6.64478958e-01 -2.95750313e-02
-5.45837760e-01 3.22757840e-01 -2.88069040e-01 -5.76148927e-01
-2.96462238e-01 -1.92071453e-01 -8.31417084e-01 -5.45799732e-01
-3.00136000e-01 -4.27315354e-01 4.19022799e-01 4.07082111e-01
1.90337911e-01 6.80205643e-01 8.89056444e-01 -8.82584274e-01
-1.90697268e-01 -8.18615198e-01 -7.04782367e-01 7.02543378e-01
9.75143552e-01 -1.30374715e-01 -4.84218657e-01 4.75816101e-01]
|
[10.925236701965332, -2.1439049243927]
|
f9968bb6-de12-4916-9a34-0bde325defd9
|
utilizing-deep-learning-for-automated-tuning
|
2306.14349
| null |
https://arxiv.org/abs/2306.14349v1
|
https://arxiv.org/pdf/2306.14349v1.pdf
|
Utilizing deep learning for automated tuning of database management systems
|
Managing the configurations of a database system poses significant challenges due to the multitude of configuration knobs that impact various system aspects.The lack of standardization, independence, and universality among these knobs further complicates the task of determining the optimal settings.To address this issue, an automated solution leveraging supervised and unsupervised machine learning techniques was developed.This solution aims to identify influential knobs, analyze previously unseen workloads, and provide recommendations for knob settings.The effectiveness of this approach is demonstrated through the evaluation of a new tool called OtterTune [1] on three different database management systems (DBMSs).The results indicate that OtterTune's recommendations are comparable to or even surpass the configurations generated by existing tools or human experts.In this study, we build upon the automated technique introduced in the original OtterTune paper, utilizing previously collected training data to optimize new DBMS deployments.By employing supervised and unsupervised machine learning methods, we focus on improving latency prediction.Our approach expands upon the methods proposed in the original paper by incorporating GMM clustering to streamline metrics selection and combining ensemble models (such as RandomForest) with non-linear models (like neural networks) for more accurate prediction modeling.
|
['Rachana Acharya', 'Kajal Tiwari', 'Karthick Prasad Gunasekaran']
|
2023-06-25
| null | null | null | null |
['management']
|
['miscellaneous']
|
[-2.35911757e-01 -4.91360694e-01 -2.66234785e-01 -4.56756294e-01
-3.56559992e-01 -5.19077659e-01 1.72963962e-01 4.38693911e-01
-2.96530992e-01 5.83535731e-01 5.94959520e-02 -6.09672308e-01
-5.86595714e-01 -4.93422002e-01 -3.50090712e-01 -3.55265260e-01
-1.90326661e-01 8.94033551e-01 4.82174158e-01 -2.84422070e-01
8.63002539e-01 8.82037997e-01 -1.91197670e+00 5.68476617e-01
5.50289631e-01 8.67424726e-01 6.07036091e-02 9.18581247e-01
-3.90610248e-01 1.08608544e+00 -1.10770714e+00 7.19984854e-03
1.47233531e-01 1.09624760e-02 -5.71188211e-01 -2.44679078e-01
6.03591613e-02 -4.24862921e-01 3.77880447e-02 1.21071726e-01
4.42887336e-01 6.36149123e-02 4.64703649e-01 -1.36745811e+00
1.95794970e-01 7.66081989e-01 -2.86549658e-01 6.78750336e-01
1.20986640e-01 -9.12383944e-02 7.51513958e-01 -3.50018233e-01
3.81135583e-01 8.43601823e-01 4.82859880e-01 -9.44747925e-02
-1.42094290e+00 -4.17934000e-01 7.97083452e-02 6.34913206e-01
-1.63796937e+00 -6.44986808e-01 5.38340032e-01 -5.70804358e-01
1.43599832e+00 4.83639389e-01 1.89466417e-01 6.34569824e-01
2.53333330e-01 4.24617440e-01 7.44886935e-01 -9.21782672e-01
6.09978735e-01 5.33989549e-01 2.05624685e-01 8.01380649e-02
1.89442381e-01 3.72258611e-02 -5.94688535e-01 -5.49847484e-01
1.80811301e-01 -1.25353709e-01 1.23436883e-01 -4.52656537e-01
-9.04960155e-01 5.98773479e-01 -1.82328761e-01 4.81458724e-01
-3.80418777e-01 -7.72068501e-02 4.96054232e-01 1.90837726e-01
2.16464520e-01 7.94361353e-01 -7.00143397e-01 -5.96311212e-01
-1.42750669e+00 -1.18538400e-03 1.38165486e+00 9.42907274e-01
8.54537606e-01 -1.71791557e-02 -2.51887590e-01 7.38664925e-01
3.67491633e-01 1.40548557e-01 4.75258678e-01 -1.12874627e+00
2.04178199e-01 7.07109630e-01 5.16313426e-02 -1.04923165e+00
-4.52489704e-01 -3.50497425e-01 -2.28963047e-01 3.18162292e-02
1.11871287e-01 3.45887318e-02 -5.81350446e-01 1.19651175e+00
1.55855373e-01 -7.15043619e-02 -4.31267112e-01 4.94886696e-01
-1.58096611e-01 4.92082685e-01 2.05084193e-03 -3.17772776e-01
7.55845189e-01 -7.40825057e-01 -5.26802838e-01 2.83878356e-01
5.13796806e-01 -1.20993662e+00 9.70542073e-01 8.59041035e-01
-5.98116338e-01 -6.54495835e-01 -1.32217503e+00 6.87411726e-01
-3.69420290e-01 1.11379968e-02 5.79160392e-01 1.03816938e+00
-1.14853919e+00 7.37983584e-01 -1.15389478e+00 -6.39881849e-01
-1.96454570e-01 5.76431572e-01 3.69394869e-01 1.18118465e-01
-5.21191955e-01 1.00004482e+00 5.02348661e-01 -2.83135682e-01
-5.54731905e-01 -6.56899154e-01 -4.55298088e-03 1.52196884e-01
8.78397465e-01 -5.20356297e-01 1.40225244e+00 -5.11152744e-01
-1.57451332e+00 4.08688523e-02 4.89166826e-02 -3.92655492e-01
2.57508665e-01 -1.84160456e-01 -5.49368918e-01 -1.33734524e-01
-2.74554938e-01 -1.21847525e-01 5.65455377e-01 -1.29689670e+00
-6.29821181e-01 -3.75728071e-01 -1.61720157e-01 -2.41971761e-01
-6.78232670e-01 1.41561195e-01 -5.84295511e-01 -3.28182310e-01
-1.98223159e-01 -7.25478649e-01 -2.08196491e-01 -1.06555128e+00
-6.28496289e-01 6.15479276e-02 6.72278047e-01 -4.82121617e-01
2.13818312e+00 -1.71181536e+00 -8.90919641e-02 7.34265983e-01
1.70668796e-01 2.85581768e-01 1.00223862e-01 9.88181949e-01
2.58615553e-01 1.53594479e-01 3.29008073e-01 -2.95344144e-02
9.12149176e-02 3.78514051e-01 -3.71036269e-02 -1.19519234e-01
-1.51042864e-01 1.56964988e-01 -4.25715059e-01 -6.42441034e-01
5.50117016e-01 8.28906745e-02 -6.58002257e-01 5.92703760e-01
-4.54569161e-01 1.67478248e-02 -2.01208487e-01 7.96954393e-01
4.53952491e-01 -3.43869716e-01 4.93788362e-01 -2.95437664e-01
-3.24229568e-01 1.53050989e-01 -1.44733071e+00 9.35931027e-01
-6.06538415e-01 3.20455253e-01 -4.93235976e-01 -1.00672174e+00
1.04385257e+00 -1.06691159e-01 8.21801305e-01 -4.41773206e-01
3.62299709e-03 1.38222888e-01 3.70500833e-02 -7.91830122e-01
6.54973805e-01 4.55583602e-01 2.41613667e-02 8.80762756e-01
-6.48199469e-02 3.38217884e-01 2.84753382e-01 1.32608548e-01
1.36559522e+00 -1.78059377e-02 5.58285236e-01 -5.64274527e-02
3.37645918e-01 2.01898798e-01 3.34077388e-01 8.87584388e-01
-1.87218085e-01 4.07409698e-01 4.36099440e-01 -4.67755467e-01
-9.72943723e-01 -9.35401201e-01 7.70375207e-02 1.42960608e+00
-1.39416575e-01 -8.45220387e-01 -7.97721505e-01 -4.97640252e-01
2.11327031e-01 1.07215416e+00 -4.85871077e-01 1.39182322e-02
-4.64414477e-01 -6.19955838e-01 3.20447177e-01 3.30375493e-01
-1.52076660e-02 -8.11531186e-01 -9.03299153e-01 5.77941895e-01
-1.18970033e-02 -1.09131479e+00 -1.55735135e-01 3.41392368e-01
-8.86325121e-01 -1.20834887e+00 2.09601983e-01 2.79340148e-01
9.31873769e-02 1.55504078e-01 1.49157643e+00 2.16809466e-01
-6.14107072e-01 5.25180459e-01 -4.13691521e-01 -5.44778883e-01
-7.00182378e-01 5.90141892e-01 9.36589018e-02 -1.53828993e-01
7.89755404e-01 -7.13088751e-01 -5.64983487e-01 7.18160927e-01
-7.58472264e-01 -3.52115750e-01 6.94168985e-01 4.51931715e-01
6.26236320e-01 2.76483476e-01 4.83481675e-01 -1.16996586e+00
7.93068230e-01 -8.25793982e-01 -6.41665995e-01 5.75673759e-01
-1.56786823e+00 3.19629788e-01 5.77082872e-01 -4.81582284e-01
-8.09787869e-01 -1.78827956e-01 2.07793936e-01 -4.60945249e-01
-4.49393004e-01 3.97303998e-01 -8.88431817e-02 9.08103883e-02
8.27107549e-01 -1.33608997e-01 1.35585153e-02 -5.58063805e-01
3.70364152e-02 1.03323662e+00 2.27210686e-01 -6.52915776e-01
4.02584910e-01 1.28764495e-01 -3.70960891e-01 -7.10247040e-01
-3.22863221e-01 -6.93503737e-01 -7.60954440e-01 -4.89397794e-01
1.69933140e-01 -2.24551380e-01 -7.95845866e-01 -7.30710849e-02
-7.86176622e-01 -1.31964952e-01 -2.03469247e-02 4.08692867e-01
-5.23214042e-01 2.82842577e-01 -1.85332969e-01 -1.04548645e+00
-2.56639302e-01 -1.07280087e+00 6.98433578e-01 1.25794619e-01
-6.61985695e-01 -7.63771057e-01 4.26216125e-01 5.99516690e-01
1.08626509e+00 9.02547315e-02 1.12836468e+00 -1.11906636e+00
-8.16283941e-01 -3.59071106e-01 -9.32291895e-02 3.66495192e-01
7.67279565e-02 6.57588243e-01 -7.36202836e-01 -1.25367180e-01
-1.12646811e-01 2.03107819e-01 3.31308544e-01 1.43551081e-01
1.25927174e+00 -1.60857111e-01 -2.95690179e-01 3.58363301e-01
1.61895859e+00 4.26255673e-01 4.59813148e-01 8.75535607e-01
4.05528396e-01 5.92094421e-01 7.78128564e-01 9.65014756e-01
5.17817557e-01 1.02369428e+00 3.83587301e-01 3.27229470e-01
3.33301485e-01 -5.89575395e-02 3.03238302e-01 8.18568945e-01
-1.70930758e-01 -1.73320547e-01 -1.18200922e+00 1.96202949e-01
-1.86642730e+00 -7.69735694e-01 2.30127856e-01 2.51757812e+00
4.42234427e-01 3.78616750e-01 5.37829578e-01 3.82809669e-01
5.72702169e-01 -2.35284477e-01 -6.39543593e-01 -8.91902566e-01
4.29270536e-01 2.22948655e-01 5.72272778e-01 7.60978386e-02
-7.05688715e-01 7.04211712e-01 6.40529394e+00 7.09513903e-01
-9.88407493e-01 -1.69603229e-01 5.83718061e-01 -3.30828846e-01
-1.11543193e-01 2.64701307e-01 -8.00784945e-01 6.43117487e-01
1.67507136e+00 -2.37630025e-01 7.81551003e-01 8.26662481e-01
5.67886114e-01 -5.41071713e-01 -1.27507174e+00 8.33509386e-01
1.20953068e-01 -1.41437757e+00 -1.66310653e-01 1.88196465e-01
2.91616350e-01 1.15771420e-01 -3.63630056e-01 4.04809207e-01
2.13585392e-01 -8.21508229e-01 3.57970148e-01 8.43151867e-01
1.94412425e-01 -8.32712829e-01 9.85896051e-01 3.95095795e-01
-7.05393493e-01 -2.35039651e-01 -2.12753385e-01 2.21486250e-03
-3.98659892e-03 7.56931543e-01 -1.41907716e+00 5.49864829e-01
7.46847451e-01 -8.61485079e-02 -8.84979129e-01 1.22037554e+00
5.13248146e-01 8.27187777e-01 -6.19314015e-01 -3.21599722e-01
-3.46113920e-01 1.45328432e-01 2.30459824e-01 1.34583342e+00
3.29599470e-01 -2.95443535e-01 2.07840711e-01 6.19766355e-01
3.95688266e-01 5.56609631e-01 -1.64107695e-01 -4.77424897e-02
9.76161599e-01 1.13495374e+00 -7.78346717e-01 -2.51926109e-02
-1.80725530e-01 3.04110110e-01 7.80038610e-02 3.61876160e-01
-7.20262706e-01 -2.98320711e-01 3.79215658e-01 4.40448254e-01
2.01002091e-01 -2.43822828e-01 -5.07016718e-01 -6.70914114e-01
6.54559284e-02 -9.07962084e-01 5.53445995e-01 -1.91596121e-01
-1.10398722e+00 8.52626562e-01 3.53399158e-01 -1.14149964e+00
-6.91475153e-01 -3.46944183e-01 -5.62685788e-01 5.39001822e-01
-9.50930715e-01 -6.00986183e-01 -4.29419488e-01 2.73180366e-01
2.30482325e-01 -6.02477670e-01 7.36326098e-01 2.87664443e-01
-8.74208450e-01 7.74265766e-01 3.29908967e-01 -7.79252112e-01
1.03108454e+00 -1.22346497e+00 -1.45615712e-01 5.61353147e-01
1.97661296e-01 7.22447276e-01 8.36604059e-01 -3.73937905e-01
-1.37465227e+00 -7.95874834e-01 6.96177244e-01 -5.05014002e-01
6.30955219e-01 -1.37275130e-01 -8.18893075e-01 2.64619589e-01
1.09077461e-01 -3.99202079e-01 1.23493516e+00 4.54563558e-01
-3.24908227e-01 -5.39157152e-01 -9.47429657e-01 1.93549573e-01
4.25802141e-01 -2.54303813e-01 -2.01265544e-01 9.83879343e-02
2.85305321e-01 -6.82915226e-02 -1.14557767e+00 6.02475524e-01
6.37666762e-01 -1.55153728e+00 8.21825802e-01 -7.49494553e-01
3.67303528e-02 -2.34361365e-01 -6.06189609e-01 -9.90107238e-01
-1.35434428e-02 -7.35037744e-01 -5.69989383e-01 1.38055623e+00
5.52582383e-01 -3.66140544e-01 8.65823448e-01 8.08852792e-01
2.14488804e-01 -9.66283679e-01 -6.02872014e-01 -7.43613720e-01
-4.96227145e-01 -6.04312599e-01 7.71606445e-01 7.01505244e-01
-9.63802636e-02 2.36602321e-01 -1.30856171e-01 1.57141402e-01
5.79817951e-01 2.80110419e-01 1.25086534e+00 -1.03993392e+00
-9.06064808e-01 -5.00075758e-01 -5.02737820e-01 -5.69519341e-01
-1.94945484e-01 -3.85699481e-01 -2.43180409e-01 -9.91398871e-01
2.02545509e-01 -8.19065392e-01 -7.59611070e-01 2.00511679e-01
-4.52394150e-02 -2.51105845e-01 2.55048662e-01 5.74075639e-01
-9.33065891e-01 6.96368441e-02 1.38922408e-01 1.90080866e-01
-7.11992681e-01 5.39044023e-01 -7.21458614e-01 4.00845975e-01
9.90198195e-01 -4.10459936e-01 -4.94829535e-01 -2.87548631e-01
3.10838036e-02 -1.47253659e-03 -9.22817141e-02 -1.24025476e+00
4.81533796e-01 -1.46108985e-01 2.09387988e-01 -7.31482387e-01
-1.94221884e-01 -7.40549743e-01 4.98980522e-01 6.12071864e-02
-2.24921435e-01 3.89301538e-01 2.74442703e-01 4.59620267e-01
-1.62581488e-01 -2.66547143e-01 3.54580790e-01 3.07380795e-01
-6.98069870e-01 -3.62607278e-02 -4.88425642e-01 -1.08552024e-01
1.21058953e+00 -3.64704490e-01 -3.26510340e-01 -2.76738882e-01
-7.52196789e-01 2.10365579e-01 5.11081576e-01 3.59516114e-01
3.62557203e-01 -8.44777048e-01 -3.79741967e-01 1.08694592e-02
3.08297753e-01 -5.09276688e-01 1.12658404e-01 7.65121400e-01
-7.34627664e-01 5.70005178e-01 -1.58956125e-01 -7.54931808e-01
-1.21358502e+00 6.82383657e-01 1.66975766e-01 -4.94828880e-01
4.16469052e-02 2.14025781e-01 -8.66665483e-01 -1.48220122e-01
4.93923932e-01 1.38962895e-01 -9.28821713e-02 2.84443442e-02
3.58567357e-01 8.19581211e-01 6.26238525e-01 -3.11454218e-02
-3.84820759e-01 7.15825930e-02 -4.01053220e-01 4.50273678e-02
1.42618191e+00 -2.74762273e-01 -1.34463876e-01 7.11594760e-01
9.70561922e-01 6.91914260e-02 -7.62918353e-01 -2.27284163e-01
8.55353951e-01 -6.17741406e-01 6.06638491e-02 -8.97754192e-01
-8.71413350e-01 6.81989014e-01 6.82510495e-01 5.24438024e-01
1.39679396e+00 -1.41535297e-01 3.81149560e-01 4.99303162e-01
7.38896549e-01 -1.38283896e+00 -3.92251089e-02 7.79696554e-02
4.11499828e-01 -1.16691124e+00 8.34027380e-02 -1.63276032e-01
-4.13501382e-01 1.42399669e+00 7.97590435e-01 3.24687570e-01
8.55337858e-01 7.13275433e-01 1.90893263e-01 1.26812449e-02
-1.13900661e+00 3.76202911e-02 -1.33026496e-01 5.09099066e-01
4.52980340e-01 2.38656476e-01 -2.09605291e-01 6.74555361e-01
-1.82043180e-01 1.84351280e-01 5.85995018e-01 1.22298658e+00
-4.48263407e-01 -1.54986310e+00 -5.54886281e-01 8.25308502e-01
-1.74309626e-01 1.24258809e-01 -2.98067451e-01 6.80009723e-01
5.02593890e-02 1.12263918e+00 4.37471457e-02 -1.09006214e+00
6.43065274e-01 5.25159389e-02 3.53695422e-01 -5.63233733e-01
-6.71440363e-01 1.20748021e-01 2.08923161e-01 -7.70855486e-01
-4.15070504e-02 -7.42799699e-01 -7.10328519e-01 -5.87842166e-01
-3.01476717e-01 3.08519781e-01 1.03003287e+00 8.24567258e-01
7.65029848e-01 7.23760903e-01 1.11612761e+00 -6.70709968e-01
-1.04512322e+00 -8.87929797e-01 -3.75854939e-01 2.00625792e-01
-2.55262971e-01 -7.60826588e-01 -1.93571761e-01 6.73266798e-02]
|
[7.19570779800415, 3.1830203533172607]
|
52b3ca4e-9148-4595-8446-95799188edc9
|
generating-texture-for-3d-human-avatar-from-a
|
2305.00936
| null |
https://arxiv.org/abs/2305.00936v1
|
https://arxiv.org/pdf/2305.00936v1.pdf
|
Generating Texture for 3D Human Avatar from a Single Image using Sampling and Refinement Networks
|
There has been significant progress in generating an animatable 3D human avatar from a single image. However, recovering texture for the 3D human avatar from a single image has been relatively less addressed. Because the generated 3D human avatar reveals the occluded texture of the given image as it moves, it is critical to synthesize the occluded texture pattern that is unseen from the source image. To generate a plausible texture map for 3D human avatars, the occluded texture pattern needs to be synthesized with respect to the visible texture from the given image. Moreover, the generated texture should align with the surface of the target 3D mesh. In this paper, we propose a texture synthesis method for a 3D human avatar that incorporates geometry information. The proposed method consists of two convolutional networks for the sampling and refining process. The sampler network fills in the occluded regions of the source image and aligns the texture with the surface of the target 3D mesh using the geometry information. The sampled texture is further refined and adjusted by the refiner network. To maintain the clear details in the given image, both sampled and refined texture is blended to produce the final texture map. To effectively guide the sampler network to achieve its goal, we designed a curriculum learning scheme that starts from a simple sampling task and gradually progresses to the task where the alignment needs to be considered. We conducted experiments to show that our method outperforms previous methods qualitatively and quantitatively.
|
['Junyong Noh', 'Amirsaman Ashtari', 'Kwanggyoon Seo', 'Sihun Cha']
|
2023-05-01
| null | null | null | null |
['texture-synthesis']
|
['computer-vision']
|
[ 3.63267839e-01 4.99678671e-01 2.54890293e-01 -1.90123752e-01
-4.13812369e-01 -3.79679173e-01 3.69896322e-01 -4.14872617e-01
1.38363540e-01 2.94955790e-01 -1.07539557e-01 6.39647245e-02
4.79164094e-01 -1.00639522e+00 -8.32627714e-01 -8.64707172e-01
3.73622388e-01 8.35462749e-01 5.47226727e-01 -3.87859613e-01
-2.85211671e-02 5.22789240e-01 -1.57547832e+00 2.98823506e-01
5.84956050e-01 1.00518024e+00 4.72880840e-01 5.06602228e-01
-1.64361224e-01 6.63326025e-01 -5.78772306e-01 -1.22624915e-02
6.11633718e-01 -7.51993895e-01 -9.64762568e-01 6.04640067e-01
6.87239528e-01 -3.36494118e-01 -1.86628088e-01 1.01465404e+00
1.27448544e-01 1.86537251e-01 5.81378937e-01 -9.82957125e-01
-2.46778980e-01 1.42320499e-01 -7.65775919e-01 -3.63860309e-01
2.10790873e-01 1.92007884e-01 6.30108893e-01 -8.19245934e-01
1.13104355e+00 1.41945720e+00 3.16598684e-01 7.87194133e-01
-1.18993723e+00 -5.25937557e-01 1.21171825e-01 -1.24994665e-01
-1.11745787e+00 -1.22454301e-01 1.30467522e+00 -3.18722844e-01
2.63828129e-01 1.66020915e-01 1.21682203e+00 9.09604728e-01
2.59915859e-01 8.58017623e-01 1.15864789e+00 -4.56802934e-01
2.81870812e-01 -5.77603355e-02 -3.55667353e-01 9.58798409e-01
-2.05087036e-01 1.57167256e-01 -5.00284314e-01 2.77131110e-01
1.52083349e+00 -2.47114569e-01 -1.61751837e-01 -9.57288265e-01
-1.02808416e+00 4.38845754e-01 3.24791729e-01 -1.74099356e-01
-3.93457890e-01 9.88289565e-02 -3.96663323e-02 3.12358528e-01
7.28892922e-01 3.05478543e-01 -2.00642243e-01 -1.59605235e-01
-7.89427638e-01 4.57473397e-01 6.22461557e-01 1.01165843e+00
1.05661976e+00 1.79869294e-01 1.95990190e-01 7.79883027e-01
3.23603779e-01 6.90308332e-01 -3.61681342e-01 -1.14659655e+00
2.42353439e-01 8.72471690e-01 9.77521986e-02 -9.84183371e-01
-8.43850970e-02 -1.01906560e-01 -9.10243690e-01 8.89224112e-01
7.56676197e-01 -8.80494155e-03 -1.15718710e+00 1.48295391e+00
1.05260038e+00 -1.58919975e-01 -1.70624375e-01 1.33916569e+00
9.10239279e-01 7.33785152e-01 -3.79894555e-01 3.10541153e-01
1.26453471e+00 -1.13053381e+00 -5.51027954e-01 -2.94748366e-01
1.62657633e-01 -9.16603506e-01 1.19922876e+00 1.98389813e-01
-1.30611324e+00 -7.88777471e-01 -1.18725646e+00 -1.94868132e-01
3.53837878e-01 -1.52679294e-01 3.18462521e-01 1.70345500e-01
-8.47175181e-01 4.73675817e-01 -8.16429913e-01 -3.18347842e-01
4.23905790e-01 1.70532912e-02 -3.09997320e-01 -9.08232331e-02
-9.74334955e-01 7.05640972e-01 1.37958512e-01 2.30086252e-01
-1.06523645e+00 -5.07279873e-01 -9.22008634e-01 -1.28446385e-01
4.28643703e-01 -8.87083054e-01 1.28016055e+00 -1.48633504e+00
-2.02203465e+00 9.41516578e-01 -4.07175869e-02 1.84228405e-01
7.58950591e-01 1.74436346e-01 1.94455519e-01 3.02314639e-01
4.82638776e-02 8.70248258e-01 1.05581653e+00 -1.54769111e+00
-7.04834938e-01 -1.95095941e-01 1.53616160e-01 5.75239599e-01
4.73957717e-01 -4.53790516e-01 -7.23580301e-01 -7.15636075e-01
4.21743572e-01 -9.46535289e-01 -1.39287874e-01 3.83623153e-01
-2.22895011e-01 1.54698659e-02 1.21396375e+00 -4.37783927e-01
6.40569389e-01 -2.16586590e+00 3.59447837e-01 3.51452291e-01
3.15451592e-01 -2.34081000e-01 -2.26132289e-01 9.93011147e-02
1.41123369e-01 -2.97335535e-01 -1.52828142e-01 -3.53235364e-01
-2.87506223e-01 1.40948370e-01 -4.80985753e-02 3.63090336e-01
1.72597662e-01 7.37944484e-01 -8.35147738e-01 -5.07292211e-01
2.40119100e-01 5.30577838e-01 -6.87320471e-01 4.21749264e-01
-6.15997076e-01 8.15652847e-01 -6.62284851e-01 6.39731586e-01
8.00117016e-01 -2.46916413e-01 1.21738791e-01 -2.14105740e-01
-1.13853194e-01 1.13016970e-01 -1.20348048e+00 1.72143340e+00
-1.32396132e-01 4.78101164e-01 2.37696856e-01 -5.82663059e-01
1.34439492e+00 2.86483437e-01 4.93094862e-01 -8.67640555e-01
1.96610332e-01 2.04055190e-01 -1.92581102e-01 -4.52430010e-01
5.56121826e-01 -1.83184102e-01 -3.34980413e-02 7.40031958e-01
-3.88166159e-01 -7.28083134e-01 -1.72593027e-01 -1.74452499e-01
4.33490962e-01 7.00955093e-01 -1.81662768e-01 -3.57498705e-01
3.85266483e-01 4.92134482e-01 4.94094580e-01 5.28985381e-01
1.73120365e-01 9.75792527e-01 5.67303240e-01 -1.17411685e+00
-1.73747778e+00 -9.54657197e-01 1.45662546e-01 6.26375079e-01
7.28716850e-01 -9.84271895e-03 -1.10124242e+00 -6.07561946e-01
-4.11031485e-01 1.39751703e-01 -9.78815794e-01 1.18966594e-01
-8.12160611e-01 -1.27954274e-01 -4.61782143e-02 1.04767092e-01
9.17308033e-01 -1.35990894e+00 -9.20568824e-01 1.59013972e-01
-4.46035385e-01 -7.08361566e-01 -6.83597505e-01 -1.03758015e-02
-7.16490626e-01 -1.10823667e+00 -7.49540985e-01 -1.19010711e+00
9.43163693e-01 3.63191217e-01 1.03255665e+00 1.79372817e-01
7.79310465e-02 -7.51873925e-02 -3.00492167e-01 -2.08794788e-01
-5.94173908e-01 1.08103998e-01 -3.96526009e-01 1.79049462e-01
-1.78586721e-01 -3.74761313e-01 -5.74713945e-01 6.10672355e-01
-9.27643120e-01 1.13829553e+00 3.29709619e-01 9.42467749e-01
8.41666102e-01 1.86739713e-01 -1.49520207e-02 -7.86220372e-01
1.41846836e-01 2.20077168e-02 -6.12518728e-01 -8.12918320e-02
1.33549161e-02 9.52589139e-02 3.79811734e-01 -7.98992932e-01
-1.20121419e+00 3.33700240e-01 -1.94369182e-01 -5.35781920e-01
-1.43202871e-01 1.90846130e-01 -2.97406793e-01 4.72486988e-02
5.22233665e-01 1.17580883e-01 4.01620835e-01 -4.21079189e-01
5.46185412e-02 4.22676623e-01 3.68079692e-01 -8.17781687e-01
8.01778078e-01 7.02446640e-01 -7.71679506e-02 -6.79684103e-01
-1.03191888e+00 8.75002667e-02 -7.13530242e-01 -6.03825927e-01
9.91079628e-01 -6.09847784e-01 -6.68977678e-01 7.66963661e-01
-1.12574506e+00 -1.08151805e+00 -5.80581427e-01 1.29464462e-01
-9.00958300e-01 4.92339842e-02 -5.97540557e-01 -4.35110778e-01
-2.65222907e-01 -1.42853415e+00 1.36296237e+00 2.52959222e-01
-4.68804747e-01 -7.59387136e-01 1.02919742e-01 3.64775151e-01
1.76032260e-01 4.39059377e-01 1.03578019e+00 2.46395618e-01
-9.46645617e-01 1.49144933e-01 6.21728227e-02 -6.78180382e-02
3.47238064e-01 2.75766831e-02 -8.47225547e-01 -8.66704211e-02
-2.63163112e-02 -2.85928935e-01 3.46850574e-01 2.92896956e-01
8.33087146e-01 -1.89136237e-01 2.52508447e-02 7.37157226e-01
1.02459884e+00 3.57134074e-01 6.42156780e-01 7.00371921e-01
9.31889772e-01 7.78328121e-01 6.71816826e-01 7.04952553e-02
3.46115232e-01 8.11424553e-01 2.58570582e-01 -5.60466528e-01
-5.52783370e-01 -5.46642959e-01 1.20503642e-01 4.34282392e-01
-2.69367516e-01 -6.91833496e-02 -7.02719212e-01 2.36885011e-01
-1.82175076e+00 -8.46676648e-01 -4.76737432e-02 2.02925086e+00
9.29214656e-01 2.48105735e-01 9.89117548e-02 3.46135013e-02
6.11355782e-01 1.81138426e-01 -6.43440962e-01 -5.12403369e-01
-1.51223272e-01 1.94001138e-01 -1.45324860e-02 6.96753621e-01
-8.38326395e-01 1.29609418e+00 5.98963976e+00 7.30812967e-01
-1.34678090e+00 -4.09926832e-01 8.72056782e-01 9.38078016e-02
-5.33119202e-01 1.40903696e-01 -4.75304246e-01 1.54523268e-01
6.09172229e-03 2.12116286e-01 3.67648512e-01 7.69553900e-01
4.09271270e-01 -3.75731528e-01 -1.01810265e+00 9.23904240e-01
-1.10300466e-01 -1.28646064e+00 2.49318540e-01 4.77241389e-02
1.08833373e+00 -5.19725800e-01 6.62094057e-02 -1.06461853e-01
3.87313575e-01 -1.01676261e+00 1.27683949e+00 5.93266785e-01
9.83946204e-01 -8.35915506e-01 4.77558196e-01 2.96101719e-01
-1.34107459e+00 5.25830686e-01 -1.76126286e-01 -9.19246152e-02
1.68403342e-01 3.29018027e-01 -8.10942829e-01 1.95469394e-01
8.34554911e-01 5.29889703e-01 -2.96379626e-01 6.34523153e-01
-3.95595461e-01 2.01910272e-01 -1.82366237e-01 1.21099390e-01
3.40824187e-01 -4.32980418e-01 5.42576969e-01 6.28931940e-01
2.10411340e-01 1.26094773e-01 3.72279376e-01 1.22710586e+00
1.02903813e-01 1.56779051e-01 -4.58130240e-01 3.52620691e-01
1.90495789e-01 1.20191479e+00 -9.38866675e-01 -5.71581423e-01
-1.39307350e-01 1.06424379e+00 3.29460859e-01 5.26456296e-01
-6.38491511e-01 -1.52729869e-01 5.00370741e-01 4.78382647e-01
3.27257663e-01 -2.20004678e-01 -5.53247869e-01 -9.20829952e-01
-2.69894879e-02 -1.13546216e+00 -2.27715131e-02 -1.13935184e+00
-8.39757621e-01 8.50089252e-01 -1.94162101e-01 -1.34393811e+00
-4.00462672e-02 -5.90671264e-02 -3.93804371e-01 1.20673370e+00
-1.05822361e+00 -1.26009738e+00 -6.51363075e-01 4.27453071e-01
8.55788648e-01 1.66603163e-01 6.01434648e-01 -5.35883494e-02
-1.87480733e-01 2.12124601e-01 -4.89717871e-01 8.34578201e-02
5.56426048e-01 -9.87415433e-01 6.47959769e-01 5.50148189e-01
-3.14613789e-01 1.35177359e-01 6.98976815e-01 -1.02959573e+00
-1.22514260e+00 -8.77704084e-01 7.08776534e-01 -3.03645611e-01
2.49600679e-01 -5.30621946e-01 -1.05140901e+00 4.00125235e-01
1.07744701e-01 -2.56462038e-01 -1.44111574e-01 -2.88541496e-01
-8.32944512e-02 -1.44028366e-01 -1.03564370e+00 1.04473102e+00
9.45734084e-01 -2.37446293e-01 -3.49179417e-01 -1.86765894e-01
3.49431157e-01 -1.07851863e+00 -7.29367375e-01 3.51798028e-01
9.07063782e-01 -7.82425582e-01 6.45136893e-01 -1.41648486e-01
6.77470863e-01 -7.80920625e-01 1.95368946e-01 -1.40377355e+00
-2.92685658e-01 -7.08839774e-01 2.57849753e-01 6.97136819e-01
2.24219248e-01 -1.40003234e-01 1.28607452e+00 6.57458842e-01
-1.21656895e-01 -7.86270499e-01 -6.12623811e-01 -3.14372391e-01
8.03422406e-02 -8.12079087e-02 8.23516369e-01 8.17842185e-01
-2.09945917e-01 2.91842937e-01 -4.62485403e-01 7.33311623e-02
7.30077922e-01 6.30818486e-01 1.27266967e+00 -1.20640814e+00
-5.27152196e-02 -2.83399642e-01 -1.42313644e-01 -1.62690127e+00
-2.31885985e-01 -5.94993055e-01 2.78347045e-01 -1.51213825e+00
1.78015947e-01 -9.21164811e-01 5.63812375e-01 4.19134587e-01
-1.53699234e-01 4.74371940e-01 6.53066346e-03 2.17571482e-01
-2.47207806e-01 6.58739090e-01 2.34458733e+00 -1.90115839e-01
-4.67372924e-01 -1.06261842e-01 -4.77961957e-01 7.57372022e-01
7.66363919e-01 -3.42762291e-01 -4.49611336e-01 -5.14674366e-01
1.54958725e-01 4.05737191e-01 3.94324183e-01 -6.39247954e-01
3.96982059e-02 -4.32120144e-01 6.82304621e-01 -7.94327974e-01
4.20292228e-01 -9.09270823e-01 6.30957901e-01 2.77177602e-01
-2.21967399e-01 3.55837233e-02 -2.46031564e-02 1.35538921e-01
-4.44979630e-02 1.55848607e-01 1.06124759e+00 -2.86085606e-01
-4.47716206e-01 5.85099757e-01 -5.43128490e-01 -5.38965538e-02
8.52756381e-01 -7.08882332e-01 1.45722702e-01 -4.29990381e-01
-7.94057727e-01 9.95650887e-02 1.12546611e+00 4.50307161e-01
9.26746070e-01 -1.51699591e+00 -5.61320603e-01 7.02957571e-01
-4.85959044e-03 7.94478834e-01 4.03138399e-01 4.35022622e-01
-9.42849696e-01 -2.39751101e-01 -4.61609423e-01 -9.05841231e-01
-1.24449694e+00 3.75930369e-01 7.17028439e-01 1.09737173e-01
-1.19771981e+00 4.08234537e-01 6.95900023e-01 -5.76395452e-01
2.68307924e-01 -3.69890451e-01 -4.61998582e-02 -3.84346128e-01
4.20098901e-01 2.02452604e-04 -2.78274059e-01 -6.95870757e-01
2.46438175e-01 9.85334754e-01 1.29297420e-01 -2.99959928e-01
1.41897094e+00 -2.48320907e-01 -3.01927719e-02 4.53040063e-01
8.92414749e-01 8.21997002e-02 -2.00969219e+00 -2.91058987e-01
-5.58457434e-01 -5.41910172e-01 -2.54446775e-01 -4.35943723e-01
-1.32045197e+00 6.68879449e-01 2.96771824e-01 -1.07647724e-01
9.21307027e-01 4.93391305e-02 8.41042101e-01 6.95085004e-02
5.37469327e-01 -1.05002153e+00 3.91056597e-01 6.74754560e-01
7.67009795e-01 -8.43402326e-01 -1.82569414e-01 -6.59747660e-01
-6.82464480e-01 1.32709420e+00 8.93003702e-01 -3.16123664e-01
4.69295949e-01 4.30764347e-01 5.05096853e-01 -6.08522713e-01
-5.64882815e-01 7.70028606e-02 4.31994081e-01 6.02269948e-01
3.77355665e-02 -1.32803842e-01 3.72782439e-01 9.75093171e-02
-5.08738637e-01 -1.51499882e-01 5.37950218e-01 7.72467434e-01
-5.47499895e-01 -1.07659698e+00 -6.12550795e-01 -9.92943943e-02
1.29789701e-02 1.65405184e-01 -4.78344202e-01 8.35164368e-01
1.25427842e-01 6.72976911e-01 2.43090868e-01 -2.82208532e-01
5.04777193e-01 -2.06700206e-01 7.72182167e-01 -6.27628207e-01
-3.17521095e-01 5.30577540e-01 9.34153125e-02 -4.59018081e-01
-4.55777764e-01 -4.52103853e-01 -1.27615416e+00 -4.57813025e-01
3.45631056e-02 1.05004258e-01 3.12414169e-01 7.52409458e-01
-2.30931491e-02 4.76829022e-01 7.22014964e-01 -1.14566576e+00
3.94386828e-01 -6.86632514e-01 -6.01871550e-01 5.69235623e-01
3.22729379e-01 -7.57953644e-01 -1.95189387e-01 3.12145531e-01]
|
[9.480668067932129, -3.064960479736328]
|
6f7949ab-e28f-4b41-a0a0-51630763e1ad
|
enhanced-multi-level-features-for-very-high
|
2305.00679
| null |
https://arxiv.org/abs/2305.00679v2
|
https://arxiv.org/pdf/2305.00679v2.pdf
|
Enhanced Multi-level Features for Very High Resolution Remote Sensing Scene Classification
|
Very high-resolution (VHR) remote sensing (RS) scene classification is a challenging task due to the higher inter-class similarity and intra-class variability problems. Recently, the existing deep learning (DL)-based methods have shown great promise in VHR RS scene classification. However, they still provide an unstable classification performance. To address such a problem, we, in this letter, propose a novel DL-based approach. For this, we devise an enhanced VHR attention module (EAM), followed by the atrous spatial pyramid pooling (ASPP) and global average pooling (GAP). This procedure imparts the enhanced features from the corresponding level. Then, the multi-level feature fusion is performed. Experimental results on two widely-used VHR RS datasets show that the proposed approach yields a competitive and stable/robust classification performance with the least standard deviation of 0.001. Further, the highest overall accuracies on the AID and the NWPU datasets are 95.39% and 93.04%, respectively.
|
['Jagannath Aryal', 'Sumesh KC', 'Chiranjibi Sitaula']
|
2023-05-01
| null | null | null | null |
['scene-classification']
|
['computer-vision']
|
[ 3.03088218e-01 -5.94847739e-01 1.64876580e-01 -3.92973959e-01
-1.13106787e+00 -8.33221450e-02 7.35389411e-01 7.60766342e-02
-4.32807982e-01 8.26033056e-01 -1.32095441e-01 -5.39254770e-02
-3.17680359e-01 -9.71395969e-01 -4.16785091e-01 -1.10033989e+00
1.19592667e-01 -4.41236734e-01 2.52565801e-01 -3.06580245e-01
2.67989755e-01 9.19111550e-01 -1.78076863e+00 3.20267588e-01
1.33713484e+00 1.33384383e+00 5.67763567e-01 2.48457029e-01
4.10313681e-02 8.15675199e-01 -3.19326222e-01 1.87137783e-01
2.95327991e-01 -7.92607218e-02 -3.60062689e-01 1.32668472e-03
2.67414063e-01 -3.46758902e-01 -2.09948018e-01 9.95377183e-01
6.27937257e-01 4.10697013e-01 5.74246228e-01 -7.66651511e-01
-8.94581616e-01 4.27593738e-01 -1.15269697e+00 4.00652111e-01
-1.14113502e-01 8.96572508e-03 8.81189167e-01 -1.14856005e+00
1.56811625e-01 1.09700084e+00 6.88539207e-01 -1.91366449e-02
-1.01189888e+00 -7.30773211e-01 2.59005129e-01 3.55325729e-01
-1.89126086e+00 -3.75347972e-01 7.28261650e-01 -4.70962584e-01
6.97165787e-01 2.99500346e-01 3.08168948e-01 4.42372292e-01
1.70896411e-01 5.17854273e-01 1.54789782e+00 -2.20657443e-03
1.46814108e-01 4.10522595e-02 2.64114916e-01 3.47190857e-01
1.21111780e-01 -2.88192689e-01 -6.88171685e-02 1.46510839e-01
7.52181590e-01 2.84091860e-01 -5.49610376e-01 2.69377738e-01
-9.79842305e-01 8.42853487e-01 1.00558662e+00 6.11148596e-01
-8.28061223e-01 -3.03563446e-01 4.85387146e-02 -3.71092884e-03
4.95559335e-01 1.36670113e-01 -2.72111088e-01 5.92876554e-01
-7.15581357e-01 2.57068604e-01 3.91365997e-02 4.02712762e-01
8.81633639e-01 2.41002351e-01 -2.74174303e-01 1.28737485e+00
2.63005644e-01 5.73258579e-01 3.51345450e-01 -3.30060840e-01
4.54942852e-01 4.78001833e-01 1.75028309e-01 -1.43305600e+00
-4.02376294e-01 -6.41239822e-01 -1.28286421e+00 1.68463677e-01
-1.71281219e-01 1.27643188e-02 -9.84487295e-01 1.37473202e+00
4.88742515e-02 2.28010058e-01 4.19383347e-01 1.05659711e+00
1.19018853e+00 1.19417834e+00 4.82440144e-01 -5.96468858e-02
1.12752056e+00 -5.67636430e-01 -7.26052940e-01 -1.51338682e-01
1.02387689e-01 -6.40375972e-01 1.05550635e+00 -7.27590872e-03
-6.27744496e-01 -8.42734694e-01 -1.14829588e+00 5.70249632e-02
-4.71127480e-01 5.33898532e-01 6.06181860e-01 2.76678562e-01
-7.88239777e-01 6.39624774e-01 -6.39893055e-01 -3.51534009e-01
6.70086622e-01 -1.36815412e-02 -4.27264094e-01 -1.41885132e-01
-1.15010607e+00 6.65386081e-01 5.39199054e-01 7.71574855e-01
-5.97321868e-01 -7.60836482e-01 -7.00776935e-01 1.73218414e-01
4.77988459e-02 1.43564209e-01 7.59055614e-01 -6.08016253e-01
-1.35778093e+00 7.32363641e-01 6.67421445e-02 -1.17801964e-01
2.34758019e-01 -1.77708566e-02 -5.43194652e-01 7.24878237e-02
3.03881131e-02 3.03021461e-01 5.33533216e-01 -1.27754200e+00
-7.50814319e-01 -6.10018373e-01 -6.87093809e-02 4.75694984e-01
-2.24745899e-01 1.68444946e-01 8.44864771e-02 -5.31550288e-01
5.58762193e-01 -4.40747529e-01 -2.86049992e-01 -6.45292550e-02
-1.42681494e-01 -1.50129169e-01 6.36937678e-01 -8.84579837e-01
1.02535188e+00 -2.43806219e+00 -1.93357006e-01 1.15595616e-01
-8.16738792e-03 5.86167276e-01 -7.37223551e-02 5.88209145e-02
-1.08976983e-01 1.74711406e-01 -3.92596871e-01 2.36406341e-01
-3.45947951e-01 -1.74078390e-01 -2.33164728e-01 6.94948733e-01
4.14100528e-01 6.67080939e-01 -5.07382154e-01 -3.50099951e-01
5.86796463e-01 7.42684841e-01 -1.88785881e-01 3.39509636e-01
1.05481058e-01 5.53026319e-01 -6.57852292e-01 6.39965296e-01
1.23332608e+00 -1.38319463e-01 -2.59191602e-01 -4.20280099e-01
-6.65914714e-01 -3.33407432e-01 -1.32679081e+00 1.27667916e+00
-4.17766511e-01 4.23041195e-01 9.43794996e-02 -8.32780719e-01
1.58380890e+00 1.56019554e-02 3.10134947e-01 -9.47615743e-01
1.54101118e-01 4.16996390e-01 -1.37603998e-01 -5.50896406e-01
5.10456443e-01 -3.24911445e-01 2.46642277e-01 -2.56464899e-01
-3.76722515e-01 1.75282538e-01 -4.10836637e-01 -5.09774387e-01
5.22482693e-01 -1.31975606e-01 5.02935469e-01 -4.40671235e-01
9.25002038e-01 -1.63515657e-02 5.06662667e-01 6.68787479e-01
-2.65376478e-01 6.27554834e-01 -9.11170319e-02 -5.09293854e-01
-8.37416768e-01 -8.56903434e-01 -6.00604415e-01 7.73271382e-01
2.68332362e-01 3.48766267e-01 -1.64145529e-01 -1.53124826e-02
-9.00615007e-02 4.23435956e-01 -5.05288780e-01 1.12049922e-01
-4.80628908e-01 -1.32172549e+00 2.98706561e-01 5.29986024e-01
1.38194668e+00 -1.09021091e+00 -5.03975332e-01 2.20891103e-01
-1.59334794e-01 -1.11441386e+00 4.35048252e-01 -1.54762417e-01
-6.34707749e-01 -5.91916621e-01 -1.08392441e+00 -7.23890364e-01
1.71905026e-01 5.98250568e-01 4.74661112e-01 -2.50132680e-01
-2.61918038e-01 -4.64840591e-01 -4.62997377e-01 -1.41596541e-01
1.34008810e-01 4.87374440e-02 -3.38705629e-01 4.16899413e-01
4.15733486e-01 -4.17479187e-01 -5.80028594e-01 1.04976043e-01
-6.92462981e-01 -6.05390705e-02 8.43627453e-01 6.94595635e-01
8.25191736e-01 3.28571320e-01 6.49446428e-01 -4.22933728e-01
3.60367686e-01 -5.62064350e-01 -7.08128929e-01 2.08528444e-01
-2.14606002e-01 -5.12242138e-01 8.36608171e-01 8.87160469e-03
-1.34584761e+00 -6.41153008e-02 -5.43606281e-01 -1.86959788e-01
-4.90733027e-01 6.25543296e-01 -3.11122775e-01 -2.24669397e-01
6.78906262e-01 3.72312337e-01 -4.32551622e-01 -7.50022948e-01
-1.24962963e-02 1.28675485e+00 4.74140733e-01 -1.71110734e-01
7.52664328e-01 4.42237884e-01 -1.89827144e-01 -1.15317869e+00
-8.75103474e-01 -4.86286014e-01 -4.91629660e-01 9.78906639e-03
1.09526610e+00 -1.42436004e+00 -5.43204010e-01 8.92381847e-01
-9.02747095e-01 -1.30689263e-01 1.17260374e-01 6.34443700e-01
-2.18930364e-01 2.09003523e-01 -4.63864237e-01 -1.16342437e+00
-5.00031650e-01 -1.05048108e+00 1.05789292e+00 5.08346140e-01
5.11588991e-01 -3.11821967e-01 -2.67317563e-01 9.51339975e-02
7.05244303e-01 7.07290947e-01 7.41866350e-01 -3.61987233e-01
-5.35777092e-01 -1.41693905e-01 -9.87768054e-01 3.85085344e-01
2.00817019e-01 4.26194258e-03 -1.28938663e+00 -1.98746443e-01
-1.50994182e-01 -1.99182630e-01 1.02357101e+00 3.46353024e-01
1.54405928e+00 1.71224736e-02 3.76028083e-02 7.67545104e-01
2.05110192e+00 3.51829737e-01 8.01254869e-01 5.08950889e-01
7.68482924e-01 5.85911274e-01 8.57884765e-01 6.34001911e-01
1.50097445e-01 4.85077322e-01 3.59148383e-01 -3.74000251e-01
1.13314338e-01 3.62369977e-02 1.33479640e-01 4.21290696e-01
-3.19059998e-01 -6.09301999e-02 -1.02225959e+00 4.32706147e-01
-1.76404643e+00 -1.05200267e+00 -4.30290163e-01 2.05400324e+00
3.54484379e-01 -2.87854910e-01 -4.38851506e-01 2.83332646e-01
8.90169084e-01 7.03946233e-01 -4.46975738e-01 -9.06949639e-02
-4.86583769e-01 2.35668883e-01 6.20267034e-01 4.67545122e-01
-1.45192778e+00 9.47927594e-01 5.26855373e+00 9.61647034e-01
-1.27407813e+00 4.48170230e-02 5.38729072e-01 4.94543374e-01
-3.74318399e-02 -5.20476460e-01 -8.03386927e-01 2.56955981e-01
6.31184459e-01 4.77029383e-02 2.63371259e-01 7.04554558e-01
4.33831155e-01 -4.49248180e-02 -2.10804507e-01 1.26582527e+00
1.89811010e-02 -1.09055436e+00 2.15716347e-01 -1.91468358e-01
6.89096689e-01 1.47395447e-01 4.14716601e-02 3.81958872e-01
-1.03199303e-01 -1.00435972e+00 4.21590477e-01 5.95406353e-01
7.95947671e-01 -1.00202250e+00 1.07219470e+00 2.79179722e-01
-1.55891705e+00 -4.15429771e-01 -8.28325510e-01 4.15326506e-02
-1.88461095e-01 5.59613109e-01 -8.88425559e-02 8.52700710e-01
8.92323494e-01 9.19881701e-01 -5.98416865e-01 1.06935310e+00
-1.00306384e-01 2.33656853e-01 -1.40937567e-01 6.27024397e-02
4.09685850e-01 -2.69200474e-01 3.36158097e-01 9.86193836e-01
4.21398401e-01 5.36338210e-01 2.65488774e-01 7.89409399e-01
7.37994090e-02 6.13299966e-01 -5.43746710e-01 1.89465612e-01
3.41387421e-01 1.40057325e+00 -4.56142575e-01 -1.82880223e-01
-3.35298061e-01 7.18563855e-01 1.90947011e-01 4.54237461e-01
-9.31496680e-01 -6.91589117e-01 5.90678930e-01 -1.92144424e-01
3.20615143e-01 -8.87149945e-02 -2.86010772e-01 -1.17285192e+00
6.05395138e-02 -7.33755708e-01 3.60718131e-01 -9.14607286e-01
-1.13164127e+00 7.77960658e-01 -2.58877397e-01 -1.27467144e+00
5.59911072e-01 -4.78680670e-01 -5.96519232e-01 1.25383544e+00
-2.44654059e+00 -1.06081605e+00 -1.04836154e+00 5.16449749e-01
4.90347087e-01 -9.27401558e-02 6.97008789e-01 4.49734151e-01
-8.74824226e-01 3.49154353e-01 3.22626352e-01 7.26780742e-02
2.34143138e-01 -9.12295640e-01 -1.53604344e-01 9.45224047e-01
-5.56469440e-01 2.81710654e-01 4.65150356e-01 -3.01414937e-01
-1.10776365e+00 -1.39996850e+00 6.26998127e-01 2.87229270e-01
4.92334187e-01 -2.36488711e-02 -1.26195407e+00 4.86650348e-01
-2.10673958e-01 2.77113676e-01 5.18310845e-01 -3.06301624e-01
-3.19750667e-01 -5.70985317e-01 -1.47496462e+00 4.09855366e-01
6.25198722e-01 -4.67976332e-01 -5.55224895e-01 1.88316748e-01
5.35000324e-01 -6.09818213e-02 -1.21191823e+00 8.65272105e-01
5.50553024e-01 -9.71682549e-01 8.91475439e-01 -1.08829699e-01
5.72027981e-01 -6.95602477e-01 -7.30062962e-01 -1.03654289e+00
-6.57861590e-01 1.77877024e-01 4.56427813e-01 1.27449441e+00
2.54688442e-01 -7.74919212e-01 2.67221779e-01 3.24297458e-01
-2.43215531e-01 -5.76377392e-01 -5.93453407e-01 -6.90039575e-01
-6.44247979e-02 -6.70421869e-02 6.88960552e-01 1.11615109e+00
-5.81066251e-01 2.60396302e-02 -2.94375241e-01 6.96049392e-01
6.65513933e-01 5.38886249e-01 5.65708578e-01 -1.40223074e+00
-1.49883144e-02 -2.70937175e-01 -4.11769390e-01 -6.75946891e-01
-9.36453324e-03 -8.57591748e-01 3.82244080e-01 -1.68933225e+00
2.77053922e-01 -5.89463949e-01 -8.11319888e-01 4.76441532e-01
-3.93267572e-01 3.58939946e-01 2.04539038e-02 3.22621316e-01
-3.25185597e-01 9.43050444e-01 1.01249099e+00 5.06189093e-02
-2.85206884e-01 -1.58570841e-01 -5.70113778e-01 5.49671471e-01
1.22162604e+00 -1.68231457e-01 1.17788561e-01 -4.36640412e-01
-3.55472207e-01 -1.55211240e-01 3.33630621e-01 -1.31761742e+00
-1.28834352e-01 -1.93261161e-01 6.69416368e-01 -8.81482124e-01
2.43824154e-01 -6.75287485e-01 9.59823355e-02 4.44397300e-01
-9.78742465e-02 -3.02806795e-01 2.37638056e-01 3.63656998e-01
-4.71218675e-01 1.77403212e-01 1.39185798e+00 -1.52137473e-01
-1.24197018e+00 6.10120595e-01 -2.41146103e-01 -5.31607211e-01
1.03885579e+00 -3.92122716e-02 -4.47856188e-01 1.23338945e-01
-3.49296212e-01 1.60637826e-01 5.37005700e-02 3.30230743e-01
6.78766429e-01 -1.26889157e+00 -1.08979666e+00 2.83096492e-01
4.66512442e-01 3.14377137e-02 7.92879641e-01 7.66617775e-01
-5.16530216e-01 2.02177912e-01 -4.44278568e-01 -6.38948977e-01
-1.09792340e+00 2.29505599e-01 4.88843620e-01 6.35332093e-02
-6.61430240e-01 5.94262540e-01 6.37546107e-02 -5.03247261e-01
-2.12728769e-01 -1.17445119e-01 -8.30978751e-01 3.16125482e-01
8.15822721e-01 5.00655293e-01 1.19500734e-01 -9.32390332e-01
-4.90738481e-01 1.12298250e+00 1.35434642e-01 2.83635050e-01
1.67436481e+00 -1.33959353e-01 -1.07907817e-01 3.96213174e-01
1.16060233e+00 -2.66600519e-01 -1.13810229e+00 -3.08142781e-01
-2.27632314e-01 -8.19155872e-01 3.73801380e-01 -5.24172723e-01
-1.05549526e+00 1.00789416e+00 9.77966726e-01 1.82664230e-01
1.26014245e+00 -1.30568400e-01 6.42786205e-01 4.43012536e-01
2.52425313e-01 -7.48885393e-01 -4.62004334e-01 4.48195606e-01
1.04411542e+00 -1.43975532e+00 6.65643811e-02 -3.91781479e-01
-4.68102425e-01 9.17447746e-01 5.75530112e-01 -3.33561599e-01
6.97476387e-01 -1.62779659e-01 1.34215623e-01 -9.91412252e-02
-1.78829804e-01 -5.06712437e-01 7.15981051e-02 3.80413949e-01
4.37419176e-01 1.84239820e-01 -4.47383374e-01 4.33662564e-01
9.57493335e-02 -2.51996629e-02 2.94791669e-01 7.67547309e-01
-8.56958270e-01 -2.76179582e-01 -4.04591024e-01 4.37195837e-01
-6.27038717e-01 -8.71786699e-02 9.33856070e-02 6.31477773e-01
-1.30594876e-02 1.05573463e+00 3.54363322e-02 -3.96875322e-01
4.49723005e-01 -3.27782750e-01 4.86946292e-02 -3.88503015e-01
-4.18650717e-01 -1.96118429e-02 -2.49713346e-01 -4.90505517e-01
-5.60014307e-01 -5.04099846e-01 -1.23399627e+00 -2.51206428e-01
-2.85215050e-01 4.58539836e-02 6.77153111e-01 7.57259011e-01
1.99501097e-01 4.18053031e-01 1.05173421e+00 -8.10671508e-01
-4.94251937e-01 -1.01140511e+00 -1.00543535e+00 2.60701090e-01
5.46513319e-01 -5.99086523e-01 -4.30469692e-01 -4.94420499e-01]
|
[9.78081226348877, -1.3942631483078003]
|
eeb452ff-6e03-47d6-a1be-de13879f3556
|
differentially-private-data-generative-models
|
1812.02274
| null |
http://arxiv.org/abs/1812.02274v1
|
http://arxiv.org/pdf/1812.02274v1.pdf
|
Differentially Private Data Generative Models
|
Deep neural networks (DNNs) have recently been widely adopted in various
applications, and such success is largely due to a combination of algorithmic
breakthroughs, computation resource improvements, and access to a large amount
of data. However, the large-scale data collections required for deep learning
often contain sensitive information, therefore raising many privacy concerns.
Prior research has shown several successful attacks in inferring sensitive
training data information, such as model inversion, membership inference, and
generative adversarial networks (GAN) based leakage attacks against
collaborative deep learning. In this paper, to enable learning efficiency as
well as to generate data with privacy guarantees and high utility, we propose a
differentially private autoencoder-based generative model (DP-AuGM) and a
differentially private variational autoencoder-based generative model
(DP-VaeGM). We evaluate the robustness of two proposed models. We show that
DP-AuGM can effectively defend against the model inversion, membership
inference, and GAN-based attacks. We also show that DP-VaeGM is robust against
the membership inference attack. We conjecture that the key to defend against
the model inversion and GAN-based attacks is not due to differential privacy
but the perturbation of training data. Finally, we demonstrate that both
DP-AuGM and DP-VaeGM can be easily integrated with real-world machine learning
applications, such as machine learning as a service and federated learning,
which are otherwise threatened by the membership inference attack and the
GAN-based attack, respectively.
|
['Chong Xiang', 'Bo Li', 'Qingrong Chen', 'Minhui Xue', 'Dali Kaarfar', 'Nikita Borisov', 'Haojin Zhu']
|
2018-12-06
| null | null | null | null |
['membership-inference-attack']
|
['computer-vision']
|
[ 1.79149568e-01 2.46269181e-01 2.99008489e-01 -3.73978078e-01
-9.67544854e-01 -1.01837361e+00 6.03304684e-01 -2.60024786e-01
-3.01915169e-01 9.01750147e-01 -7.80154839e-02 -5.05193830e-01
3.97643298e-02 -1.16758108e+00 -1.03721392e+00 -1.23773885e+00
-3.18362713e-02 2.26091594e-01 -2.26028580e-02 1.32312447e-01
-1.46479324e-01 6.34740591e-01 -1.16999757e+00 4.90942262e-02
7.08510637e-01 1.08568537e+00 -5.56818068e-01 4.62917507e-01
3.34173620e-01 5.65486550e-01 -8.11835110e-01 -1.09251463e+00
7.83059120e-01 -4.03972864e-01 -6.47885144e-01 -4.71751332e-01
1.57806113e-01 -7.14790523e-01 -6.65420413e-01 1.44055438e+00
6.68629050e-01 -1.09557444e-02 3.84508997e-01 -2.03300023e+00
-5.96385419e-01 5.65487981e-01 -2.22030267e-01 -2.90387034e-01
-2.46797681e-01 2.60755986e-01 5.13171911e-01 -2.60989070e-01
2.26749644e-01 1.10465920e+00 6.35694802e-01 1.02130377e+00
-9.87305522e-01 -1.26702738e+00 -3.20950478e-01 -1.49554297e-01
-1.42422032e+00 -4.87889826e-01 7.16968060e-01 -3.08562130e-01
3.54133159e-01 6.28281593e-01 3.90534431e-01 1.56492519e+00
2.69725591e-01 6.80051088e-01 9.84723747e-01 2.37906314e-02
6.44924521e-01 2.61966199e-01 -3.16302508e-01 5.67003250e-01
4.67114985e-01 4.06353056e-01 -2.32305914e-01 -8.72552335e-01
8.31368446e-01 2.87805527e-01 -4.21216130e-01 -1.97946206e-01
-4.51214373e-01 1.11098719e+00 2.24624127e-01 -2.35046461e-01
-1.93640441e-01 2.76712686e-01 6.29709303e-01 4.02030051e-01
2.88114578e-01 1.25963122e-01 -5.15527666e-01 1.89901367e-01
-5.52182078e-01 3.48017514e-01 8.84377897e-01 1.06436324e+00
5.82467556e-01 2.05109254e-01 1.04277596e-01 2.96181470e-01
4.40950304e-01 4.88818794e-01 6.37613952e-01 -8.29465628e-01
4.17736411e-01 1.87119693e-01 -2.24365056e-01 -9.86791968e-01
1.61857396e-01 -2.98897505e-01 -1.24241936e+00 2.37954453e-01
2.43036002e-01 -8.15378606e-01 -4.71499115e-01 2.31536341e+00
5.40587306e-01 3.98682326e-01 6.33777857e-01 4.72901970e-01
5.00609457e-01 5.22751570e-01 -1.33673504e-01 9.42614228e-02
1.11912501e+00 -4.33286190e-01 -3.68737608e-01 3.33557874e-02
7.29668319e-01 -1.94000125e-01 6.19069874e-01 1.06430948e-01
-8.90781343e-01 -9.12937373e-02 -1.12603998e+00 -9.72259641e-02
-3.81572425e-01 -3.57343495e-01 6.41258478e-01 1.11189890e+00
-1.00199878e+00 3.67713243e-01 -9.39397454e-01 -9.79482159e-02
1.13692105e+00 5.21496832e-01 -6.65355027e-01 5.41215278e-02
-1.48281276e+00 2.83614993e-01 4.13519055e-01 3.51143852e-02
-1.20949292e+00 -7.01928675e-01 -7.18386531e-01 8.72764811e-02
6.01325259e-02 -9.10006404e-01 8.71591926e-01 -6.81726336e-01
-1.47955811e+00 8.35778117e-01 5.30745268e-01 -1.02196538e+00
8.86686206e-01 9.89320651e-02 -4.55505669e-01 4.48678397e-02
-3.65263909e-01 2.20695138e-01 9.61804450e-01 -1.13539422e+00
-4.81769681e-01 -8.35241973e-01 6.39825836e-02 -1.74203992e-01
-5.75719953e-01 -1.98956400e-01 2.04045743e-01 -5.77324390e-01
-2.00271755e-01 -9.28073108e-01 -1.22692794e-01 5.85308596e-02
-6.98400915e-01 2.39674345e-01 1.42377698e+00 -5.40922880e-01
8.02697480e-01 -2.36576986e+00 -2.48867244e-01 3.58234167e-01
2.52493590e-01 5.56117594e-01 -7.92583078e-03 3.21259469e-01
3.18582594e-01 4.38678652e-01 -3.46869141e-01 -5.86962700e-01
1.65572837e-01 5.78267395e-01 -9.12381828e-01 6.48205578e-01
-1.70846522e-01 9.45562720e-01 -5.93345284e-01 -4.27368693e-02
-1.12093337e-01 6.03834152e-01 -7.11373806e-01 3.42465520e-01
-2.07478523e-01 4.77282345e-01 -6.04296565e-01 5.42771280e-01
1.03487456e+00 7.33765867e-03 1.42288089e-01 -1.79025978e-02
5.05980015e-01 -1.41858503e-01 -7.54700005e-01 1.06635070e+00
-2.05057710e-01 3.72904241e-01 3.21203828e-01 -8.46803188e-01
8.63648295e-01 5.96348405e-01 5.93105052e-03 -1.96086556e-01
4.17658210e-01 1.49552584e-01 -2.17905432e-01 -1.32589936e-01
1.67702183e-01 -1.23564497e-01 -2.59361625e-01 6.07663870e-01
1.34892821e-01 2.28431910e-01 -8.50273907e-01 2.66673476e-01
1.00242734e+00 -3.04085940e-01 -4.35533561e-02 -2.23629326e-01
4.14549649e-01 -6.32023156e-01 8.10892045e-01 8.50736022e-01
-2.90775716e-01 3.62024724e-01 5.36029756e-01 -1.20304704e-01
-9.26621199e-01 -1.07131684e+00 1.38021223e-02 6.92142308e-01
-5.86064570e-02 -1.66193828e-01 -1.02644479e+00 -9.63360190e-01
1.50966898e-01 6.24729097e-01 -5.44053853e-01 -7.24312484e-01
-9.53033790e-02 -7.76460588e-01 1.37132037e+00 4.77083027e-01
9.75535512e-01 -5.76454639e-01 -2.89735138e-01 -1.58158422e-01
2.05011770e-01 -8.18600595e-01 -4.53672498e-01 1.15552368e-02
-7.78891981e-01 -9.90845859e-01 -3.04228812e-01 -4.26791161e-01
7.84382820e-01 -1.41583607e-01 6.03520691e-01 -2.48900652e-02
-5.32966070e-02 3.30928415e-01 6.71161935e-02 -6.58727467e-01
-8.11940670e-01 -1.72816236e-02 2.15060890e-01 3.63430381e-01
3.50193530e-01 -9.12016988e-01 -5.39480805e-01 1.40996352e-01
-1.37217855e+00 -2.70703286e-01 2.74813920e-01 8.13138306e-01
4.31908607e-01 3.32122952e-01 7.21094310e-01 -1.29914403e+00
6.60954952e-01 -6.61315739e-01 -9.23350036e-01 2.09615409e-01
-6.50197089e-01 -1.35745198e-01 9.66752410e-01 -3.92131478e-01
-9.37032759e-01 -9.51176360e-02 -5.39811373e-01 -8.45226586e-01
-2.40944311e-01 1.97452947e-01 -9.48143780e-01 -3.43642682e-01
4.47323382e-01 3.42935771e-01 2.12849066e-01 -2.99156278e-01
2.56147176e-01 7.69153059e-01 6.10155106e-01 -6.26859009e-01
1.10348678e+00 6.05610013e-01 3.57345730e-01 -5.36938369e-01
-5.31820655e-01 3.72684717e-01 6.74493387e-02 2.16521457e-01
6.59897327e-01 -1.03705323e+00 -9.03112829e-01 1.20069206e+00
-8.98905993e-01 -1.96659297e-01 -1.77790955e-01 3.01019788e-01
-4.76896584e-01 3.81684750e-01 -6.77385926e-01 -9.31384444e-01
-7.26396203e-01 -1.15596139e+00 5.78315794e-01 3.56533438e-01
3.65375280e-01 -1.20876372e+00 -2.99981683e-01 4.93099391e-01
4.86555934e-01 8.27678442e-01 8.83616686e-01 -1.32490528e+00
-7.06001401e-01 -5.88350713e-01 2.06160516e-01 7.13921428e-01
4.58696857e-02 -6.19294494e-02 -1.26962030e+00 -7.21120656e-01
6.27520025e-01 -4.98163551e-01 4.00425732e-01 8.20246711e-03
1.48013866e+00 -1.16042912e+00 2.18195282e-02 1.20119143e+00
1.46830440e+00 3.52858663e-01 8.48975956e-01 -4.89911512e-02
7.33637571e-01 2.42876187e-01 9.15351659e-02 6.12782180e-01
7.19417399e-03 6.02495112e-02 8.39513540e-01 8.20985362e-02
7.95904815e-01 -6.57609224e-01 2.73346663e-01 3.22413236e-01
3.53657007e-01 -3.91811639e-01 -4.30083156e-01 9.18768197e-02
-1.75239527e+00 -1.00842512e+00 1.47068322e-01 2.40134048e+00
8.74288797e-01 -1.37190238e-01 2.62468308e-02 -2.33774930e-02
5.29883087e-01 8.22931752e-02 -7.92271316e-01 -4.59088296e-01
-1.80260077e-01 2.53590226e-01 8.26331675e-01 1.70325026e-01
-1.06545496e+00 7.71418393e-01 5.62461567e+00 8.00973117e-01
-9.91318345e-01 2.88989276e-01 1.00758135e+00 -1.79004043e-01
-5.08138537e-01 -8.01784918e-02 -6.77336991e-01 6.39898658e-01
7.85474002e-01 -3.97478998e-01 4.82764542e-01 1.17659473e+00
-5.04356861e-01 5.63180029e-01 -1.16039860e+00 9.21303272e-01
-1.40457273e-01 -1.40036094e+00 1.46675736e-01 6.06835723e-01
6.82968855e-01 -7.31975678e-03 4.67133224e-01 3.46054822e-01
8.30723345e-01 -1.04261684e+00 4.21378225e-01 2.29856327e-01
6.69854045e-01 -1.40580213e+00 7.57883668e-01 5.60030282e-01
-6.00587845e-01 -6.83175549e-02 -6.14137828e-01 2.64012456e-01
-3.28256726e-01 3.98008317e-01 -3.83038819e-01 5.83337247e-01
5.25398731e-01 6.09388910e-02 -1.76099077e-01 3.66159767e-01
-2.62370825e-01 7.83094347e-01 -4.28665459e-01 1.43754855e-01
2.23813191e-01 -3.94694388e-01 5.68884552e-01 5.85169971e-01
3.32228214e-01 5.48364818e-02 -3.07908028e-01 1.15086424e+00
-6.43396318e-01 -3.31192076e-01 -7.41181314e-01 -1.30529389e-01
9.72797334e-01 1.02313435e+00 4.71252538e-02 -5.05603999e-02
-2.71193329e-02 1.16307807e+00 2.31089562e-01 3.98836643e-01
-8.10609698e-01 -2.45172083e-01 1.35262501e+00 -3.57010335e-01
1.16737843e-01 1.35758385e-01 -9.59715471e-02 -1.22918582e+00
-1.39831364e-01 -1.06016064e+00 6.28210306e-01 -2.88748115e-01
-1.63549173e+00 4.39014047e-01 -3.62593442e-01 -8.85393739e-01
-2.93620348e-01 -2.23154023e-01 -7.59161949e-01 1.07744980e+00
-1.02206588e+00 -1.11874378e+00 1.87635288e-01 1.11630487e+00
-3.55693668e-01 -5.73948741e-01 1.07246995e+00 1.04277000e-01
-8.48056495e-01 1.41450357e+00 6.70913577e-01 7.16627836e-01
3.04478437e-01 -7.98446238e-01 3.33653271e-01 1.26571095e+00
1.74517468e-01 8.96743655e-01 4.41293061e-01 -3.57053190e-01
-1.70612133e+00 -1.64926207e+00 2.40824118e-01 -3.08528721e-01
3.48540694e-01 -8.07443202e-01 -9.70684707e-01 1.10301316e+00
-1.09785929e-01 2.04330474e-01 1.14744461e+00 -3.89093906e-01
-6.46453977e-01 -2.55008191e-01 -2.03927279e+00 5.86943448e-01
6.15593374e-01 -9.54274178e-01 -1.61299691e-01 2.92870134e-01
9.78888988e-01 -4.18561041e-01 -1.01066709e+00 6.30310103e-02
4.61211056e-01 -1.11113799e+00 8.70910048e-01 -7.94936359e-01
1.78874344e-01 6.68830350e-02 -4.28790569e-01 -1.07512057e+00
1.30971387e-01 -1.03363883e+00 -3.57795149e-01 1.65030146e+00
3.16957384e-02 -1.34931552e+00 9.44684386e-01 1.13854551e+00
3.40727389e-01 -5.49649894e-01 -1.27958846e+00 -7.02883482e-01
4.29228336e-01 -1.61590293e-01 1.25961149e+00 1.26990306e+00
-5.49397290e-01 -2.31536567e-01 -6.66543365e-01 6.21439278e-01
9.19354796e-01 -1.68436572e-01 1.03753459e+00 -1.00923729e+00
-5.66209495e-01 -1.75112307e-01 -6.12676799e-01 -5.37262738e-01
2.17945322e-01 -8.69695187e-01 -3.96977007e-01 -8.41062307e-01
-1.34786725e-01 -3.61743152e-01 -4.66615796e-01 5.66328347e-01
9.02047157e-02 2.02591568e-01 1.17706574e-01 6.91214651e-02
1.14721477e-01 6.05240583e-01 7.75529742e-01 1.50771469e-01
8.41573849e-02 3.60517591e-01 -9.97695506e-01 5.23309469e-01
9.77784514e-01 -5.92564762e-01 -7.48602331e-01 -2.17608288e-01
5.52992038e-02 3.02829240e-02 6.93570852e-01 -9.17733490e-01
2.18492165e-01 -6.29360005e-02 2.59078264e-01 -1.85940087e-01
1.31997466e-01 -1.13754284e+00 6.24986768e-01 5.10489285e-01
-2.38111228e-01 -4.29409862e-01 7.26784915e-02 8.07464480e-01
1.56059787e-02 9.24114957e-02 9.20477331e-01 1.23132206e-02
-1.18822239e-01 9.93661344e-01 -1.17783323e-01 1.75376981e-01
1.15756404e+00 4.72348332e-02 -5.42801440e-01 -5.10604858e-01
-3.55375141e-01 1.40452206e-01 6.15531087e-01 2.14413926e-01
5.98877072e-01 -1.34315586e+00 -6.36780262e-01 7.69246638e-01
-1.18617252e-01 3.54564130e-01 4.49253768e-01 2.76930749e-01
-2.59107918e-01 -1.48888389e-02 -1.67809010e-01 -1.08417585e-01
-1.32033956e+00 6.26962304e-01 4.20955896e-01 3.41695398e-02
-5.34818709e-01 1.08229423e+00 3.45151186e-01 -4.53419268e-01
4.86341298e-01 -3.49019729e-02 5.78711987e-01 -3.76867324e-01
5.39392233e-01 3.17896217e-01 -1.85053676e-01 -4.73300159e-01
-2.12725431e-01 -2.00203061e-01 -1.13992169e-01 4.11961339e-02
1.01872110e+00 -2.79587563e-02 -1.25645310e-01 -1.14847437e-01
1.59765661e+00 4.67717415e-03 -1.36456442e+00 -2.61313140e-01
-6.60521626e-01 -5.30141890e-01 1.02312729e-01 -5.36334276e-01
-1.55742383e+00 9.14721370e-01 4.89838868e-01 2.66567200e-01
1.27605295e+00 -2.10884497e-01 1.13454783e+00 3.09240580e-01
6.87283576e-01 -6.66921675e-01 -4.24496740e-01 9.82776061e-02
4.98552650e-01 -9.09093261e-01 -3.18357587e-01 5.85501269e-02
-4.83830482e-01 5.89761674e-01 5.26330292e-01 5.71755692e-03
9.59595561e-01 5.73028982e-01 1.30160391e-01 2.83444345e-01
-6.90118849e-01 7.41854608e-01 -2.19941199e-01 8.40747654e-01
-4.37167704e-01 -8.06426350e-03 3.49563301e-01 1.07322156e+00
-3.11315030e-01 -1.91328943e-01 2.94821918e-01 9.31082189e-01
1.39772728e-01 -1.13721359e+00 -2.61229396e-01 4.79024768e-01
-8.41944635e-01 6.37082979e-02 -5.06007612e-01 4.81165141e-01
6.86322898e-02 7.19386995e-01 9.42232460e-02 -7.35182703e-01
-2.11685479e-01 9.37469453e-02 3.25484090e-02 -2.15249527e-02
-7.73715496e-01 -3.57463390e-01 -4.02374864e-01 -4.29373741e-01
2.85543706e-02 -4.90072072e-01 -1.12625813e+00 -8.86666000e-01
-2.83223003e-01 1.72636002e-01 5.40553689e-01 6.93097174e-01
7.05739319e-01 1.70371234e-01 1.07740819e+00 -9.50972661e-02
-1.17588830e+00 -5.51778555e-01 -9.40941811e-01 3.78624707e-01
3.15133899e-01 -9.79537740e-02 -8.64839494e-01 -4.54512745e-01]
|
[5.895511150360107, 7.011638164520264]
|
cd9f708e-409b-4444-9504-06d85ef2b882
|
a-fully-automated-latent-fingerprint-matcher
|
1406.6854
| null |
http://arxiv.org/abs/1406.6854v1
|
http://arxiv.org/pdf/1406.6854v1.pdf
|
A Fully Automated Latent Fingerprint Matcher with Embedded Self-learning Segmentation Module
|
Latent fingerprint has the practical value to identify the suspects who have
unintentionally left a trace of fingerprint in the crime scenes. However,
designing a fully automated latent fingerprint matcher is a very challenging
task as it needs to address many challenging issues including the separation of
overlapping structured patterns over the partial and poor quality latent
fingerprint image, and finding a match against a large background database that
would have different resolutions. Currently there is no fully automated latent
fingerprint matcher available to the public and most literature reports have
utilized a specialized latent fingerprint matcher COTS3 which is not accessible
to the public. This will make it infeasible to assess and compare the relevant
research work which is vital for this research community. In this study, we
target to develop a fully automated latent matcher for adaptive detection of
the region of interest and robust matching of latent prints. Unlike the
manually conducted matching procedure, the proposed latent matcher can run like
a sealed black box without any manual intervention. This matcher consists of
the following two modules: (i) the dictionary learning-based region of interest
(ROI) segmentation scheme; and (ii) the genetic algorithm-based minutiae set
matching unit. Experimental results on NIST SD27 latent fingerprint database
demonstrates that the proposed matcher outperforms the currently public
state-of-art latent fingerprint matcher.
|
['Xiuping Jia', 'Jinwei Xu', 'Jiankun Hu']
|
2014-06-26
| null | null | null | null |
['set-matching']
|
['computer-vision']
|
[ 6.30057812e-01 -3.83341402e-01 -3.81502867e-01 -4.35876250e-01
-6.52954936e-01 -9.48326051e-01 3.89974505e-01 -7.40615502e-02
-7.11823478e-02 4.46447551e-01 -3.60228062e-01 -5.59066474e-01
-1.65009633e-01 -7.88105726e-01 -5.03899038e-01 -6.23141646e-01
2.63781935e-01 7.21507072e-01 3.81465077e-01 3.62330645e-01
8.14440787e-01 1.06931674e+00 -1.33625042e+00 2.10855752e-01
8.28357816e-01 7.01884210e-01 1.79551855e-01 4.76021081e-01
-1.90105334e-01 1.79488972e-01 -2.89588243e-01 -4.48373348e-01
7.52385557e-01 -2.35061824e-01 -5.00936508e-01 -2.77143359e-01
1.30366790e+00 -5.91689587e-01 3.54752131e-02 1.15220094e+00
5.65407634e-01 -3.45846325e-01 5.33208132e-01 -8.79284322e-01
-2.77510792e-01 1.56868443e-01 -7.79729843e-01 1.94473699e-01
2.81526715e-01 -9.44998637e-02 2.98560679e-01 -7.36873507e-01
8.37457001e-01 1.27257037e+00 1.06236649e+00 2.84388363e-01
-1.41533875e+00 -1.35579538e+00 -5.85899234e-01 -6.12350367e-02
-1.35165691e+00 -5.94814718e-01 8.46765161e-01 -5.79330683e-01
5.34065127e-01 2.11241141e-01 2.56558299e-01 8.58082175e-01
4.26951468e-01 3.09994698e-01 1.55313122e+00 -5.41976869e-01
-7.74866715e-02 3.34216923e-01 2.10867286e-01 1.01374626e+00
6.57280028e-01 5.09792387e-01 -3.51486892e-01 -6.95427597e-01
1.00448906e+00 1.61560237e-01 3.42124254e-01 -4.57697719e-01
-7.11587846e-01 5.90990126e-01 -2.19444811e-01 3.79267722e-01
-3.52935821e-01 -9.14018042e-03 4.87906151e-02 2.26049304e-01
1.14207275e-01 3.05891037e-01 3.39050055e-01 1.08948983e-01
-1.61143625e+00 4.07794654e-01 5.97360373e-01 7.24109590e-01
8.57311189e-01 -3.23220819e-01 -2.30285123e-01 6.72342539e-01
7.95582950e-01 8.52676928e-01 -5.09838723e-02 -8.16280782e-01
6.24071777e-01 7.67362654e-01 1.84020534e-01 -1.48031771e+00
1.96889743e-01 -3.64471018e-01 -3.47344369e-01 2.73316741e-01
6.37216330e-01 1.03707887e-01 -1.07289815e+00 1.00658512e+00
3.17688048e-01 2.30554670e-01 -4.48647141e-01 4.64047700e-01
6.27831638e-01 4.02756304e-01 -1.31241843e-01 -4.65943590e-02
1.39710248e+00 -5.84018350e-01 -7.81454384e-01 -2.61924028e-01
-6.36236891e-02 -1.33914566e+00 4.10671443e-01 2.22736225e-01
-7.41563797e-01 -7.47932434e-01 -1.36963260e+00 2.49239743e-01
-3.72239709e-01 2.69347817e-01 7.35964298e-01 1.58848691e+00
-5.68601370e-01 4.66921508e-01 -7.57633150e-01 -6.84002459e-01
2.56213188e-01 6.43707395e-01 -3.73563915e-01 -2.59135842e-01
-7.88132250e-01 6.53391421e-01 1.44556925e-01 3.53433460e-01
-7.39433289e-01 -4.28890228e-01 -5.68738401e-01 -3.88342112e-01
2.02016875e-01 3.00621358e-03 5.20102322e-01 -3.72226775e-01
-1.11696506e+00 1.15638506e+00 -5.13242662e-01 -2.00385526e-02
6.55115247e-01 1.39134958e-01 -5.67926049e-01 1.89723924e-01
5.70683539e-01 2.49423563e-01 8.65237832e-01 -1.24907696e+00
-5.61258972e-01 -6.81094050e-01 -5.57586312e-01 -4.14383084e-01
1.99454516e-01 3.68450046e-01 -7.14192808e-01 -7.75437176e-01
7.02670932e-01 -9.51334536e-01 -1.79788638e-02 -2.13907599e-01
-1.82046309e-01 1.32950112e-01 9.69514787e-01 -9.05532897e-01
1.24358582e+00 -1.86343491e+00 -1.09622955e+00 9.43039358e-01
6.66641071e-02 5.17237067e-01 1.07141256e-01 4.96817946e-01
1.52193978e-01 3.61733660e-02 -7.38427117e-02 -1.25804106e-02
1.02951499e-02 -1.33873865e-01 -3.76964897e-01 8.64357591e-01
-3.88202518e-01 7.29961395e-01 -6.70071721e-01 -9.30652440e-01
2.30244547e-01 1.73076674e-01 -2.33862713e-01 1.59904137e-01
3.77081960e-01 2.57770896e-01 -6.71803713e-01 1.61942875e+00
1.24507558e+00 1.88507825e-01 2.37632006e-01 -3.01433235e-01
-3.75058472e-01 -1.48164868e-01 -1.60089564e+00 1.39573777e+00
3.11999232e-01 5.79064429e-01 6.78827614e-02 -5.46094596e-01
1.43920553e+00 3.50950599e-01 6.18292212e-01 -7.99893260e-01
-6.13418669e-02 5.98239064e-01 -3.44177544e-01 -5.65471053e-01
6.45748734e-01 -2.81882361e-02 8.96437913e-02 8.92392278e-01
-2.35350922e-01 5.61089873e-01 4.83247936e-02 -1.95561305e-01
9.85457182e-01 1.49040073e-01 -1.00701891e-01 -2.23464787e-01
3.89140666e-01 1.07582562e-01 8.27007532e-01 1.23619413e+00
-4.39469308e-01 3.70453745e-01 -6.09247871e-02 -4.41611171e-01
-1.05704224e+00 -1.13079262e+00 -3.78843129e-01 6.14206553e-01
5.29987216e-01 1.46784812e-01 -6.92674041e-01 -3.25984776e-01
3.88095289e-01 -1.18190721e-01 -4.84083086e-01 5.30283749e-01
-6.97644532e-01 -4.83078450e-01 1.16632831e+00 1.14313841e-01
8.72794151e-01 -8.91225219e-01 -4.66034293e-01 4.01215166e-01
-4.71359454e-02 -7.90402949e-01 -5.28205216e-01 -1.52227283e-01
-8.90848756e-01 -1.17225635e+00 -6.55492187e-01 -7.50564277e-01
8.94374549e-01 3.31107289e-01 4.62935954e-01 -1.73217177e-01
-6.44908071e-01 1.90415978e-01 -5.12317568e-02 -3.58791351e-01
-4.44939554e-01 -3.96769792e-02 -1.36615887e-01 4.63519990e-01
7.16465056e-01 -3.43998492e-01 -7.23023832e-01 6.56162918e-01
-4.86550212e-01 -2.48269588e-01 9.35295403e-01 4.30939168e-01
7.82222211e-01 8.06472637e-03 2.61807859e-01 -1.04517019e+00
7.02585638e-01 -3.04325700e-01 -8.52944314e-01 7.41469264e-01
-1.01119125e+00 1.62233654e-02 -2.22860709e-01 -2.81356066e-01
-1.20641601e+00 2.58007020e-01 1.97865412e-01 -5.89005984e-02
-1.38088778e-01 1.09094284e-01 -2.17865467e-01 -7.90761590e-01
4.98329759e-01 2.12087512e-01 -3.30897979e-02 -8.06328356e-01
-1.52088001e-01 1.06617010e+00 9.50287759e-01 -9.07946527e-01
1.04178202e+00 6.63468659e-01 1.30335316e-01 -4.46997166e-01
9.64026451e-02 -1.09170794e+00 -5.93445599e-01 -4.24740821e-01
7.66511142e-01 -1.98287845e-01 -8.96059275e-01 5.18789828e-01
-1.17668831e+00 3.44780326e-01 3.59897107e-01 4.33209598e-01
-2.05938131e-01 8.01238239e-01 -1.79552853e-01 -8.82887006e-01
-4.68115032e-01 -1.12747979e+00 8.28177154e-01 4.84012097e-01
-2.24268705e-01 -5.37927270e-01 5.45530975e-01 9.74586070e-01
4.92407411e-01 5.39718688e-01 6.30769491e-01 -3.15909833e-01
-9.31773126e-01 -7.64127672e-01 -3.23459357e-01 -1.38167918e-01
4.06636178e-01 4.66710553e-02 -1.11347318e+00 -2.21758351e-01
-8.37869570e-02 9.17092115e-02 6.48277938e-01 2.54937023e-01
5.90063930e-01 -3.97028923e-02 -8.10786188e-01 8.38661075e-01
1.59877920e+00 8.14881325e-01 8.02383840e-01 3.90890330e-01
5.09356081e-01 6.99736416e-01 8.50282550e-01 -1.30638508e-02
-1.73249945e-01 8.07611823e-01 -1.15191557e-01 -1.92022398e-01
9.74478722e-02 -5.54871202e-01 -1.76208385e-03 1.96131617e-02
-4.07929450e-01 9.65884104e-02 -1.07576442e+00 3.72491390e-01
-1.70842922e+00 -1.47154236e+00 -5.58101349e-02 2.38733649e+00
6.39868021e-01 -9.37130973e-02 -1.26634553e-01 3.25781219e-02
1.08246183e+00 -1.00574605e-01 -4.54737008e-01 -5.39841413e-01
1.79420829e-01 4.40679252e-01 1.12842274e+00 4.75388855e-01
-1.28801978e+00 8.85798931e-01 6.53183270e+00 8.74199927e-01
-1.16759682e+00 5.92430867e-02 1.33936703e-01 6.29712939e-01
-2.69165874e-01 3.40534627e-01 -1.26894259e+00 4.95717257e-01
4.60151345e-01 3.93190742e-01 2.36151576e-01 5.87778151e-01
1.72833800e-01 -3.22815001e-01 -8.59529316e-01 1.08617699e+00
-1.34160936e-01 -1.39107728e+00 -1.35516733e-01 5.16083121e-01
6.96493328e-01 -4.38318074e-01 5.15164025e-02 -2.83581167e-01
-2.26485163e-01 -9.68621731e-01 5.27889967e-01 8.41163158e-01
8.13254058e-01 -4.89826381e-01 5.92496395e-01 8.42566043e-02
-1.43822074e+00 2.68401682e-01 -3.46368492e-01 6.64615333e-01
-6.17149798e-03 2.40394339e-01 -1.23586380e+00 1.88940585e-01
4.34732437e-01 -6.53261691e-02 -6.47266865e-01 1.25380516e+00
2.93875694e-01 6.96213901e-01 -2.78615475e-01 3.02877933e-01
8.24778154e-02 -2.84181237e-01 4.53143299e-01 1.47202182e+00
2.26149991e-01 -2.07477421e-01 -3.69648151e-02 1.22074449e+00
2.14251116e-01 1.36520162e-01 -6.65078878e-01 -4.14128572e-01
7.32028902e-01 9.07741904e-01 -1.20408189e+00 3.56727317e-02
-3.06113839e-01 8.51812720e-01 -5.13372183e-01 3.00620645e-01
-4.72477704e-01 -5.25387943e-01 2.34644651e-01 5.64620256e-01
1.61502331e-01 -1.91931084e-01 -6.00154698e-01 -6.22376025e-01
1.30052060e-01 -8.40185046e-01 3.50377649e-01 -1.55954853e-01
-1.08907044e+00 2.35538408e-01 2.59714834e-02 -1.16382623e+00
-1.71078414e-01 -4.07434970e-01 -5.26296914e-01 1.38062119e+00
-1.07959032e+00 -1.55999124e+00 -3.78052384e-01 4.79811311e-01
5.29054515e-02 -6.77338719e-01 6.97541595e-01 6.76517189e-01
-2.86835641e-01 1.05462742e+00 2.04185918e-01 2.15475425e-01
9.07295585e-01 -8.00154805e-01 4.49176282e-01 1.14776838e+00
-2.82277763e-01 1.47642565e+00 5.05279303e-01 -1.49921620e+00
-1.50146317e+00 -6.72187746e-01 1.17643321e+00 -4.85693097e-01
4.14920710e-02 -3.46176356e-01 -5.58412731e-01 5.43488562e-01
-4.05411094e-01 -1.36096537e-01 8.34975958e-01 -6.34269267e-02
-2.25485548e-01 -4.12352622e-01 -1.60378623e+00 1.44154891e-01
6.75768793e-01 -8.69073331e-01 -4.67547864e-01 -1.62053913e-01
-4.79892045e-01 -1.79462656e-01 -7.64071703e-01 5.27760506e-01
1.62547731e+00 -8.68455350e-01 1.12395239e+00 4.25507780e-03
-1.91822022e-01 -6.36036158e-01 -8.56582969e-02 1.98904678e-01
-2.42449448e-01 -4.34671193e-01 2.57419467e-01 1.71304572e+00
1.24465071e-01 -7.51123190e-01 1.29018629e+00 9.14497554e-01
2.27999374e-01 -3.95053238e-01 -7.83323050e-01 -9.42147017e-01
-7.97307551e-01 -2.39473343e-01 5.63374460e-01 9.31648374e-01
-6.29253626e-01 -4.24900174e-01 -5.23598731e-01 3.87191087e-01
1.35204148e+00 5.11706889e-01 8.47146451e-01 -1.12611854e+00
-3.21544558e-01 -2.55936682e-01 -5.30078828e-01 -6.53723776e-01
-3.48744124e-01 -7.25666165e-01 -3.09393138e-01 -1.23039031e+00
4.72032994e-01 -1.09379947e+00 -1.20627984e-01 4.55459446e-01
1.30582646e-01 4.85367209e-01 -2.63718545e-01 5.44508517e-01
3.43311415e-03 -4.80011374e-01 1.05237472e+00 -4.52135742e-01
-2.93259472e-01 2.75237560e-01 -5.84352314e-01 4.71742719e-01
5.27867794e-01 -1.00026858e+00 -3.59957457e-01 1.00223377e-01
1.46825150e-01 8.28115046e-02 2.92861313e-01 -8.63762558e-01
6.28555894e-01 -4.15780067e-01 5.57560444e-01 -1.16271627e+00
4.94472645e-02 -8.47395301e-01 8.51610363e-01 5.35795033e-01
1.25220627e-01 -2.91013807e-01 9.98993069e-02 3.99893314e-01
-5.46848848e-02 -5.68283916e-01 6.82629526e-01 -8.86766240e-02
-8.21979105e-01 3.44987601e-01 -2.00530171e-01 -7.11995542e-01
9.59289551e-01 -1.38959157e+00 -3.76676798e-01 2.22957030e-01
-2.05008253e-01 -1.78945586e-01 6.67988896e-01 2.27455199e-01
5.14948130e-01 -1.30543745e+00 -4.98631746e-01 5.93033552e-01
2.07144156e-01 -7.26073563e-01 9.62212533e-02 3.81033063e-01
-1.19427216e+00 7.22945750e-01 -7.64986634e-01 -4.65578079e-01
-1.76292419e+00 1.78219393e-01 3.88087422e-01 -2.38992527e-01
-2.61625707e-01 6.07452869e-01 -1.60780340e-01 -4.03472275e-01
3.68430316e-01 5.81311211e-02 -1.41317040e-01 4.99458835e-02
4.15149003e-01 7.42175341e-01 -8.48446637e-02 -5.87541580e-01
-5.53781331e-01 1.01230609e+00 2.36115474e-02 -1.22441985e-01
9.32621062e-01 1.55148022e-02 -4.16479737e-01 -1.03656352e-01
8.84519398e-01 5.33981025e-01 -8.35302472e-01 -1.84372179e-02
1.02146931e-01 -9.36778009e-01 -3.25237274e-01 -6.75238132e-01
-6.42914116e-01 6.07877851e-01 1.42880619e+00 -1.72203586e-01
7.50379801e-01 -3.13270301e-01 7.29528964e-01 2.34119862e-01
6.76519275e-01 -1.24082887e+00 -4.23649877e-01 -7.32760653e-02
5.98052144e-01 -1.11722553e+00 1.96340606e-01 -3.46928269e-01
1.07860938e-01 1.17163825e+00 1.59319699e-01 5.09340912e-02
5.28659105e-01 2.60414392e-01 1.48484871e-01 -5.80837190e-01
2.67641604e-01 2.24902213e-01 6.27039254e-01 5.26029229e-01
4.22214180e-01 1.38465390e-01 -6.50363684e-01 3.47640127e-01
1.17498361e-01 2.61917412e-01 -9.14400965e-02 1.12341857e+00
-5.63804567e-01 -1.77665317e+00 -1.10916328e+00 5.07103860e-01
-7.92874098e-01 3.21917772e-01 -5.01685441e-01 4.20843482e-01
9.03876603e-01 9.37320888e-01 -3.77723187e-01 -3.78433615e-01
4.68194075e-02 -1.43121809e-01 7.72188604e-01 -4.03609037e-01
-6.20639205e-01 1.23942800e-01 -6.70237392e-02 -4.92499977e-01
-3.97408873e-01 -7.59696126e-01 -8.18615735e-01 -5.65384328e-01
-2.26606995e-01 6.72993585e-02 1.11904955e+00 8.56703758e-01
-8.47638547e-02 -3.50617260e-01 4.41184402e-01 -4.68526483e-01
-6.82150647e-02 -6.98882580e-01 -7.48041749e-01 4.84366491e-02
9.71987471e-02 -6.46086931e-01 3.38045537e-01 2.01737672e-01]
|
[12.937544822692871, 0.9756936430931091]
|
a6d15430-39bf-49dc-a288-f39bb42ce2e9
|
re-reproducing-learning-to-deceive-with
| null | null |
http://rescience.github.io/bibliography/Habacker_2021.html
|
https://zenodo.org/record/4834146/files/article.pdf
|
[Re] Reproducing Learning to Deceive With Attention-Based Explanations
|
Scope of Reproducibility
Based on the intuition that attention in neural networks is what the model focuses on, attention is now being used as an explanation for a modelsʼ prediction (see Galassi, Lippi, and Torroni1 for a survey). Pruthi et al.2 challenge the usage of attention-based explanation through a series of experiments using classification and sequence-to-sequence (seq2seq) models. They examine the modelʼs use of impermissible tokens, which are user-defined tokens that can introduce bias e.g. gendered pronouns. Across multiple datasets, the authors show that with the impermissible tokens removed the model accuracy drops, implying their usage in prediction. And then by penalising attention paid to the impermissible tokens but keeping them in, they train models that retain full accuracy hence must be using the impermissible tokens, but that does not show attention being paid to the impermissible tokens. As the paperʼs claims have such significant implications for the use of attention-based explanations, we seek to reproduce their results.
Methodology
Using the authorsʼ code, for classifiers we attempt to reproduce their embedding, BiLSTM, and BERT results across the occupation prediction, gender identify, and SST + wiki datasets. Further, we reimplemented BERT using HuggingFaceʼs transformer library [3] with restricted self-attention (information cannot flow between permissible and impermissible tokens). For seq2seq we used the authorsʼ code to reproduce results across Bigram Flip, Sequence Copy, Sequence Reverse, and English-German (En-De) machine translation datasets. We performed refactoring on the authorsʼ code aiming toward a more uniformly usable code style as well as porting across to PyTorch Lightning. All experiments were run in approximately 130 GPU hours on a computing cluster with nodes containing Titan RTX GPUs.
Results
We reproduced the authorsʼ results across all models and all available datasets, confirming their findings that attention-based explanations can be manipulated and that mod els can learn to deceive. We also replicated their BERT results using our reimplemented model. There was only one result not as strongly (> 1 S.D.) in their experimental direction.
What Was Easy
The authorsʼ methods were largely well described and easy to follow, and we could quickly produce the first results as their code worked straightaway with minor adjustments. They were also extremely responsive and helpful via email.
What Was Difficult
Re-implementing the BERT-based classification model to perform replicability, with further specification details on model architecture, penalty mechanism, and training procedure needed. Also, porting code across to PyTorch Lightning.
Communication With Original Authors
There was a continuous email chain with the authors for several weeks during the reproducibility work. They made additional code and datasets available per our requests, along with providing detailed responses and clarifications to our emailed questions. They encouraged the work and we wish to thank them for their time and support.
|
['Mathias Parisot', 'Ard Snijders', 'Rahel Habacker', 'Andrew Harrison']
|
2021-01-31
| null | null | null |
rc-2020
|
['occupation-prediction']
|
['natural-language-processing']
|
[ 2.71778405e-01 3.98361862e-01 -2.21312270e-01 -4.26216990e-01
-6.83978915e-01 -7.75387645e-01 7.23124385e-01 -1.76782951e-01
-6.43601537e-01 9.00821865e-01 2.44456202e-01 -9.25006390e-01
-3.06103323e-02 -3.77559900e-01 -7.72328138e-01 -6.73351169e-01
4.03996706e-02 5.39285183e-01 -1.70139104e-01 -4.25830245e-01
5.93784153e-01 2.58928359e-01 -1.65984654e+00 3.55820835e-01
5.61804414e-01 3.52965504e-01 -2.56566890e-02 8.62266660e-01
9.01442915e-02 6.87920153e-01 -6.68954253e-01 -8.67793620e-01
1.09612443e-01 -2.61913419e-01 -1.06790674e+00 -5.50648689e-01
7.37205625e-01 -2.33883440e-01 4.29580398e-02 7.66616344e-01
6.15030169e-01 -2.07200795e-02 6.37390375e-01 -1.38117528e+00
-1.01733232e+00 8.64725530e-01 -3.40430439e-01 4.77311015e-01
-8.30121487e-02 3.15444589e-01 1.19311225e+00 -8.39430332e-01
5.03834486e-01 1.35793650e+00 9.53131795e-01 1.00779235e+00
-1.42365170e+00 -9.11333084e-01 2.38063596e-02 2.45954394e-01
-1.05578637e+00 -4.09697592e-01 1.97029650e-01 -3.89438450e-01
1.47243285e+00 8.92908990e-01 5.06947637e-01 1.64543581e+00
5.25955915e-01 5.72179377e-01 1.01425767e+00 -4.80285525e-01
-1.42804265e-01 3.12449545e-01 2.31177598e-01 4.34372902e-01
9.47840735e-02 3.87189746e-01 -4.98192668e-01 -1.93215027e-01
3.21345896e-01 -2.75108755e-01 -6.18440211e-02 3.81585866e-01
-1.10102034e+00 9.86939371e-01 2.72131830e-01 4.37736273e-01
-3.77390534e-02 5.41323006e-01 6.24546528e-01 4.28642094e-01
3.98373127e-01 7.48058140e-01 -9.33567941e-01 -5.92409372e-01
-8.18292975e-01 1.06824547e-01 6.61562860e-01 6.91493809e-01
4.76211965e-01 2.67483175e-01 -2.04455554e-01 7.85443187e-01
1.94847852e-01 1.22548096e-01 9.24811721e-01 -8.83331001e-01
2.74731338e-01 7.25685507e-02 -1.88771546e-01 -9.24358130e-01
-4.27281141e-01 -5.79340100e-01 -8.39991927e-01 1.09377816e-01
6.88426137e-01 -3.52066398e-01 -9.61929739e-01 2.04650307e+00
-1.97382405e-01 -1.76169053e-01 -1.73054636e-04 7.84420252e-01
7.51052380e-01 5.09807706e-01 3.38275045e-01 2.73930967e-01
1.51186156e+00 -7.82920241e-01 -4.29686189e-01 -4.32466507e-01
1.01107085e+00 -6.55771613e-01 1.23996150e+00 3.77947748e-01
-9.54357266e-01 -4.96867359e-01 -1.00719404e+00 -2.92263508e-01
-3.64146262e-01 -1.46025330e-01 8.57195497e-01 8.02789271e-01
-1.21699595e+00 1.02108431e+00 -7.36486733e-01 -4.20289457e-01
2.21622527e-01 7.40407228e-01 -3.64672869e-01 3.22745562e-01
-1.38112462e+00 1.26359248e+00 8.08268040e-02 2.10778385e-01
-4.98940229e-01 -7.54045129e-01 -8.08720589e-01 1.23398192e-01
-5.12731746e-02 -4.96240616e-01 1.29356420e+00 -1.59185016e+00
-1.04963529e+00 9.84332085e-01 -2.76357174e-01 -5.03103912e-01
5.00797987e-01 3.29553708e-02 -2.75276512e-01 -5.19884229e-01
9.23029333e-02 9.76038933e-01 4.77267891e-01 -1.00619805e+00
-4.01801586e-01 -3.33016783e-01 1.27267838e-01 -2.55288500e-02
-2.13367462e-01 3.22667718e-01 2.64007628e-01 -5.93693495e-01
-1.66773707e-01 -1.10072613e+00 2.23473266e-01 -4.66310382e-01
-6.21627152e-01 -3.02101314e-01 3.01782578e-01 -7.43603230e-01
1.11096096e+00 -2.14744711e+00 -1.85877942e-02 2.52088793e-02
1.53763190e-01 2.29248673e-01 -1.78228348e-01 4.18412328e-01
-6.01664543e-01 9.07908916e-01 -3.57820094e-01 -3.56858790e-01
4.45723474e-01 4.06492949e-01 -4.20650542e-01 3.01953942e-01
3.33548725e-01 9.29157019e-01 -5.96790910e-01 -1.42138273e-01
-7.10234195e-02 6.89667940e-01 -8.37778568e-01 -1.61859185e-01
2.46938422e-01 9.13190767e-02 2.55233943e-01 2.99559742e-01
3.41086835e-01 -4.04921696e-02 1.98906273e-01 1.90168738e-01
-2.43843347e-01 5.60873449e-01 -7.01511502e-01 9.99571919e-01
-4.43651259e-01 9.11710799e-01 -2.24535391e-02 -8.62433076e-01
9.50173736e-01 3.00646424e-01 -2.74459600e-01 -4.58108902e-01
3.49265672e-02 6.15132391e-01 7.34022498e-01 -4.66279536e-01
6.67073607e-01 -5.05015254e-01 -3.62627842e-02 4.05951142e-01
-1.24760188e-01 4.07688171e-02 -8.85128528e-02 -3.37128490e-02
1.13160443e+00 1.67119235e-01 -1.70670822e-02 -5.19040942e-01
3.77817124e-01 1.14232823e-01 4.22630847e-01 1.02242899e+00
-1.97964221e-01 7.93118238e-01 8.06887984e-01 -5.46333134e-01
-1.13167858e+00 -6.40750051e-01 -3.31975698e-01 1.53658509e+00
-5.41647017e-01 -4.66390550e-01 -7.10774601e-01 -6.75377846e-01
-5.24339676e-02 1.23625362e+00 -1.06510043e+00 -2.15826422e-01
-5.65963566e-01 -8.12659621e-01 8.63941729e-01 4.62681919e-01
2.50924230e-01 -1.38972116e+00 -6.68239534e-01 3.97401862e-02
-4.81552482e-02 -2.84337878e-01 -4.79820162e-01 7.28895843e-01
-6.11882925e-01 -6.37100518e-01 -4.39884722e-01 -6.32184386e-01
5.41081667e-01 -3.95563602e-01 1.17232060e+00 7.22893834e-01
-6.68414608e-02 8.67267102e-02 -5.67418262e-02 -6.86336815e-01
-4.85832393e-01 3.55048656e-01 1.75122451e-03 -4.97029155e-01
4.73849654e-01 -5.91370046e-01 -3.13267618e-01 1.10276483e-01
-7.73975253e-01 2.01270729e-01 4.45081532e-01 1.37607801e+00
-1.00513650e-02 -7.25874245e-01 4.71008599e-01 -9.63939250e-01
6.27032578e-01 -5.76849401e-01 -4.27516848e-02 -1.21499866e-01
-7.88110852e-01 3.74487609e-01 7.80994773e-01 -4.34655041e-01
-5.21098495e-01 -4.29011285e-01 -5.03521085e-01 -3.29955429e-01
-1.72024325e-01 2.34014839e-01 1.38571709e-01 3.36114943e-01
5.93943596e-01 5.78330532e-02 1.08108848e-01 -2.55308479e-01
-3.90860699e-02 6.43119752e-01 3.12750101e-01 -7.24043131e-01
5.46404481e-01 -2.50770971e-02 -2.89108515e-01 -5.42360067e-01
-4.05194342e-01 3.44151616e-01 -3.73577774e-01 1.93629518e-01
8.49779844e-01 -4.67371136e-01 -9.23850775e-01 2.19953924e-01
-1.36907876e+00 -4.92165267e-01 -1.69819191e-01 5.66478074e-01
-4.14316297e-01 7.87552521e-02 -7.10669994e-01 -7.76837409e-01
-4.68303680e-01 -1.41870189e+00 7.86250830e-01 3.43123041e-02
-1.03072727e+00 -1.19276309e+00 -1.52621359e-01 1.94718376e-01
6.84799314e-01 -1.34197205e-01 1.29017186e+00 -1.20576453e+00
-3.36719379e-02 1.16473332e-01 -2.56114900e-01 2.88350761e-01
-1.88033551e-01 2.16119274e-01 -1.12458301e+00 -1.47680283e-01
-1.90088749e-01 -2.58105546e-01 8.95370424e-01 1.43257335e-01
1.31867707e+00 -6.01844311e-01 -2.47790478e-04 6.46602869e-01
1.04497838e+00 -1.15645770e-02 6.92742169e-01 7.53496408e-01
6.39935851e-01 8.22868586e-01 2.03264132e-01 -1.46269156e-02
3.02299798e-01 7.00321913e-01 4.41700876e-01 -7.64349550e-02
2.02597991e-01 -7.46129602e-02 6.51361704e-01 5.69356918e-01
-2.76154578e-01 -1.43367141e-01 -8.86618435e-01 6.84277296e-01
-1.63004982e+00 -1.27114952e+00 -5.16715646e-01 2.23641062e+00
9.02188301e-01 2.77702093e-01 -1.38362706e-01 3.56640434e-03
5.98778844e-01 -9.63220894e-02 -3.53053808e-01 -1.53694773e+00
7.11915195e-02 4.26138788e-01 5.14893949e-01 7.88040519e-01
-7.19432771e-01 7.74410665e-01 6.00806189e+00 9.09992337e-01
-1.22918141e+00 9.11650956e-02 9.94511247e-01 -2.74975479e-01
-7.09451675e-01 5.56938201e-02 -7.97477961e-01 6.46472216e-01
1.42350519e+00 -2.59181619e-01 4.20986265e-01 6.76201940e-01
1.45653829e-01 4.12556678e-02 -1.41439414e+00 7.02598274e-01
-3.49891335e-02 -1.20026612e+00 -1.49499744e-01 1.02623187e-01
2.47812569e-01 6.60835505e-02 1.99710175e-01 6.84198618e-01
4.23639566e-01 -1.56552315e+00 1.16248214e+00 1.40380919e-01
7.46526778e-01 -8.71278346e-01 1.12339020e+00 2.87955463e-01
-4.35987830e-01 -1.08007248e-02 -2.65266210e-01 -5.22765517e-01
-3.14305693e-01 1.12779483e-01 -9.37151968e-01 2.49659687e-01
7.03399003e-01 4.59599495e-01 -6.70267224e-01 3.42769325e-01
-3.47573161e-01 9.45421517e-01 -2.32740492e-01 -3.14871073e-01
4.18353587e-01 1.30573273e-01 4.43501532e-01 1.42935538e+00
4.02813733e-01 -1.15633607e-01 -5.88771045e-01 9.83968019e-01
1.20543100e-01 -1.07064053e-01 -4.94272351e-01 2.32178301e-01
3.09433550e-01 1.29146314e+00 -5.38054228e-01 -3.23773324e-01
-1.40172109e-01 9.92444277e-01 3.81785303e-01 1.53452158e-01
-1.04742265e+00 -4.65102851e-01 9.22799945e-01 3.90393436e-02
1.34563506e-01 6.39746338e-02 -5.64829111e-01 -8.79930139e-01
-1.65241241e-01 -1.15876603e+00 1.80880368e-01 -9.46440876e-01
-1.24905801e+00 6.25871241e-01 1.94489006e-02 -4.56272423e-01
-4.13248956e-01 -7.28038967e-01 -6.97626591e-01 1.29962766e+00
-1.09320498e+00 -9.27539825e-01 1.77037328e-01 -1.57629233e-02
4.58738625e-01 1.68230593e-01 1.01516807e+00 1.12139419e-01
-6.38079047e-01 1.00628662e+00 7.10076094e-02 2.66770512e-01
7.94050097e-01 -1.38547599e+00 7.11960554e-01 5.22146940e-01
1.75284177e-01 1.07926500e+00 9.04373467e-01 -4.61296439e-01
-1.04626274e+00 -7.07506716e-01 1.42421567e+00 -7.87029386e-01
5.57049453e-01 -5.40819824e-01 -9.54109192e-01 1.16098905e+00
5.40934861e-01 -3.87271196e-01 5.61163425e-01 3.32734525e-01
-3.10236335e-01 1.96378976e-01 -9.55708861e-01 8.61154437e-01
1.01182556e+00 -2.86745608e-01 -7.28311956e-01 1.81700990e-01
5.96562922e-01 -3.23073626e-01 -4.91730899e-01 1.76782340e-01
8.66996825e-01 -1.11143088e+00 6.63813293e-01 -9.56687689e-01
9.10234690e-01 2.27860481e-01 -1.38289928e-01 -1.38538313e+00
-4.16765839e-01 -4.67302293e-01 4.58460450e-01 1.37467074e+00
9.88802791e-01 -1.04460895e+00 5.11268795e-01 9.55300570e-01
-3.50711882e-01 -8.93200040e-01 -1.18722057e+00 -6.98554814e-01
6.22087896e-01 -7.48960972e-01 6.22738957e-01 1.10445142e+00
2.59512097e-01 3.20146441e-01 -2.70218581e-01 -2.54872084e-01
2.73651212e-01 -4.70502168e-01 5.50956964e-01 -1.01516819e+00
-4.39104170e-01 -5.22407472e-01 -1.75431460e-01 -4.22053009e-01
4.16474134e-01 -1.36168015e+00 -7.63077736e-02 -1.04309249e+00
3.32649142e-01 -4.13439780e-01 -1.06222674e-01 1.04568875e+00
-1.50102183e-01 5.82964540e-01 3.21470886e-01 2.24778786e-01
3.13217603e-02 2.22999156e-01 8.04584384e-01 -1.15133952e-02
1.67420611e-01 -4.07874137e-01 -1.20647693e+00 7.45193839e-01
9.42880034e-01 -6.97296321e-01 -5.60279656e-03 -5.64441085e-01
3.54462564e-01 -3.93523633e-01 7.11677551e-01 -6.28070593e-01
-1.67848885e-01 4.05834168e-02 5.67666113e-01 -5.56166843e-02
1.68466628e-01 -5.26965737e-01 2.66012013e-01 6.95235968e-01
-6.92271471e-01 4.32445467e-01 7.77167678e-01 -9.08694118e-02
3.55401695e-01 -4.76772010e-01 4.77465689e-01 -1.71467736e-01
-3.71770442e-01 -2.36773133e-01 -7.52631545e-01 5.27874678e-02
6.67506635e-01 -2.46605590e-01 -4.58753526e-01 -4.34676021e-01
-7.12413967e-01 -3.45673114e-02 5.09514332e-01 6.90752566e-01
1.07445329e-01 -1.07468009e+00 -8.40200007e-01 3.81639421e-01
-1.63464472e-01 -5.78614116e-01 3.56598869e-02 9.45681572e-01
-2.86261201e-01 5.12980103e-01 -2.28846788e-01 -3.82264555e-01
-1.46447694e+00 3.31322908e-01 4.77948844e-01 -6.16587996e-02
-2.83589959e-01 9.39680398e-01 1.37301564e-01 -8.55083227e-01
-7.74571449e-02 -4.38157767e-01 -3.90948355e-02 -6.44081309e-02
4.19764936e-01 2.18666330e-01 1.46519899e-01 -5.37482023e-01
-6.73007131e-01 2.80219674e-01 -1.85736120e-01 -3.11398327e-01
1.38588619e+00 9.60555524e-02 -1.92973912e-01 5.78914762e-01
1.18087792e+00 2.27369983e-02 -8.77048790e-01 5.13773441e-01
-1.14015438e-01 -1.57646626e-01 -2.22303361e-01 -1.14076710e+00
-6.74119473e-01 1.10023797e+00 3.70209306e-01 1.71027407e-01
5.58269560e-01 -2.06443071e-01 2.63554186e-01 -2.70000892e-04
1.79068968e-01 -7.96237230e-01 -5.68648577e-01 7.67887533e-01
1.04013562e+00 -1.04486430e+00 -5.56743965e-02 -5.64324223e-02
-4.99413431e-01 1.17018175e+00 8.46059084e-01 1.68252885e-01
-5.32530248e-06 1.94889694e-01 -2.97990777e-02 -7.51549453e-02
-1.05604672e+00 3.50787669e-01 5.72584476e-03 4.70152378e-01
9.09638047e-01 -4.42114845e-02 -6.34272456e-01 8.97443056e-01
-8.44057560e-01 -3.67168278e-01 7.08251774e-01 4.93855119e-01
-3.14369649e-02 -1.13627362e+00 -5.02183497e-01 5.83292603e-01
-8.25506032e-01 -6.81694984e-01 -4.65263784e-01 1.14659023e+00
2.74743438e-01 8.59596252e-01 3.89839679e-01 -6.74918175e-01
7.75714740e-02 4.27227020e-01 2.32087523e-01 -5.48556924e-01
-1.25346422e+00 -5.39866567e-01 3.73630702e-01 -2.33245552e-01
7.49794319e-02 -8.66848707e-01 -1.25911093e+00 -8.93565714e-01
-1.15745813e-01 3.10466141e-01 7.10880816e-01 9.19866204e-01
5.17267466e-01 5.92341661e-01 6.78642243e-02 -9.75076437e-01
-7.96987414e-01 -1.10543871e+00 -3.26904505e-01 2.55238086e-01
2.80661047e-01 -3.01185191e-01 -9.69263196e-01 -3.27766836e-01]
|
[10.572684288024902, 8.56872844696045]
|
432f7dd8-d70b-43f6-98c0-b15e70fecb61
|
joint-domain-adaptation-and-speech-bandwidth
|
2203.16614
| null |
https://arxiv.org/abs/2203.16614v1
|
https://arxiv.org/pdf/2203.16614v1.pdf
|
Joint domain adaptation and speech bandwidth extension using time-domain GANs for speaker verification
|
Speech systems developed for a particular choice of acoustic domain and sampling frequency do not translate easily to others. The usual practice is to learn domain adaptation and bandwidth extension models independently. Contrary to this, we propose to learn both tasks together. Particularly, we learn to map narrowband conversational telephone speech to wideband microphone speech. We developed parallel and non-parallel learning solutions which utilize both paired and unpaired data. First, we first discuss joint and disjoint training of multiple generative models for our tasks. Then, we propose a two-stage learning solution where we use a pre-trained domain adaptation system for pre-processing in bandwidth extension training. We evaluated our schemes on a Speaker Verification downstream task. We used the JHU-MIT experimental setup for NIST SRE21, which comprises SRE16, SRE-CTS Superset and SRE21. Our results provide the first evidence that learning both tasks is better than learning just one. On SRE16, our best system achieves 22% relative improvement in Equal Error Rate w.r.t. a direct learning baseline and 8% w.r.t. a strong bandwidth expansion system.
|
['Najim Dehak', 'Laureano Moro-Velázquez', 'Jesús Villalba', 'Saurabh Kataria']
|
2022-03-30
| null | null | null | null |
['bandwidth-extension', 'bandwidth-extension']
|
['audio', 'speech']
|
[ 2.89817780e-01 9.07206014e-02 3.37556228e-02 -7.35410154e-01
-1.82473850e+00 -6.05488002e-01 6.07116282e-01 -5.16145885e-01
-6.31317556e-01 7.78323412e-01 3.19616497e-01 -6.61925435e-01
2.06903368e-02 -1.19606309e-01 -6.61369920e-01 -8.01459968e-01
1.57727897e-01 5.23450851e-01 1.84711978e-01 -1.23151243e-01
-2.24862501e-01 2.44117618e-01 -1.12305951e+00 3.68242055e-01
5.51248729e-01 7.19965577e-01 4.78930175e-01 1.20824647e+00
1.56822175e-01 2.79814541e-01 -7.54285455e-01 -6.47493526e-02
1.33881450e-01 -4.97703224e-01 -9.46863413e-01 1.54041618e-01
3.86942416e-01 -3.44262123e-01 -4.12286431e-01 6.92168415e-01
1.05929744e+00 3.18442017e-01 6.25773489e-01 -8.56469929e-01
-2.27666959e-01 8.93209755e-01 -4.45665330e-01 2.84173995e-01
2.37322688e-01 -7.73886815e-02 7.64159203e-01 -1.02604127e+00
1.89453602e-01 1.07766831e+00 6.84467256e-01 8.82157385e-01
-1.36301327e+00 -7.75479078e-01 -2.62581874e-02 -6.70410693e-03
-1.50025332e+00 -1.29426026e+00 6.23229086e-01 -1.54848635e-01
1.25137830e+00 1.55160189e-01 -1.30509019e-01 1.35504043e+00
-4.68054682e-01 4.88122970e-01 9.74425197e-01 -7.14792073e-01
1.92860976e-01 3.82523477e-01 1.76223889e-01 1.08919442e-01
-3.47091079e-01 3.22612584e-01 -7.77983129e-01 -3.25654596e-01
4.20494050e-01 -8.84999335e-01 -4.92636591e-01 -1.04434103e-01
-9.52343345e-01 5.97505808e-01 -2.09008455e-01 2.82745868e-01
1.93927418e-02 -6.60435334e-02 4.27424729e-01 6.40197217e-01
4.16456074e-01 6.07772879e-02 -8.61359537e-01 -2.34206423e-01
-1.06325209e+00 -6.29130825e-02 1.04439974e+00 9.45650220e-01
5.15638530e-01 2.37946153e-01 1.12574898e-01 1.59527075e+00
4.36766863e-01 6.03541136e-01 6.28387272e-01 -7.29815543e-01
6.12157643e-01 -6.25877440e-01 -8.56338143e-02 2.50064135e-01
-3.26707751e-01 -3.83311301e-01 -6.00430608e-01 -1.58091441e-01
4.92328227e-01 -6.12140477e-01 -8.47129703e-01 1.94980478e+00
1.97188571e-01 4.94971812e-01 3.73754203e-01 6.37196362e-01
5.49778461e-01 7.98181772e-01 -4.36328985e-02 -3.92015517e-01
1.21164834e+00 -8.86842191e-01 -7.03451753e-01 -2.34021544e-01
6.81872368e-01 -1.08996379e+00 1.27548552e+00 6.39493525e-01
-1.09488201e+00 -8.79521847e-01 -1.22651720e+00 -5.58897369e-02
-4.09338363e-02 3.10526669e-01 4.97065522e-02 1.32776594e+00
-1.13474274e+00 3.11962038e-01 -6.31159782e-01 -1.77797630e-01
-1.17481112e-01 5.41967094e-01 -1.64317831e-01 -8.85896757e-03
-1.30406988e+00 7.19766438e-01 3.38784158e-01 -3.65308791e-01
-9.71774101e-01 -6.86160266e-01 -6.46823108e-01 1.26370937e-02
4.86896858e-02 -5.72339892e-01 1.95223773e+00 -5.11601329e-01
-2.12383723e+00 8.28921616e-01 -3.35031956e-01 -6.70092821e-01
2.44873732e-01 -2.99667031e-01 -9.41166043e-01 -1.74781322e-01
-3.75151187e-01 4.25661355e-01 9.16914344e-01 -1.19226301e+00
-6.40060782e-01 -1.45387053e-01 -3.47653747e-01 2.32661784e-01
-5.04593253e-01 1.14553019e-01 -3.27986240e-01 -5.07200956e-01
-5.58026657e-02 -8.74162316e-01 1.17279120e-01 -7.07207024e-01
-2.78218478e-01 -1.80999517e-01 9.37924981e-01 -1.05393755e+00
1.23794961e+00 -2.39070749e+00 -4.70637083e-02 3.52793455e-01
-3.92402500e-01 3.31381172e-01 -2.37146482e-01 1.92101642e-01
-2.63262928e-01 -1.13422126e-01 -1.97222531e-01 -8.04999709e-01
2.34838799e-01 1.56951785e-01 -4.71471071e-01 5.00222802e-01
7.46012991e-03 2.50937372e-01 -4.65645552e-01 -3.04612905e-01
6.83082193e-02 6.88276231e-01 -7.17419446e-01 3.76485348e-01
5.25999069e-02 5.55238903e-01 1.21752813e-01 2.49593705e-01
6.96979225e-01 1.43494830e-01 3.19968998e-01 -1.29189044e-01
3.18123251e-02 9.31598425e-01 -1.19576764e+00 2.10520315e+00
-8.62787426e-01 5.65418005e-01 4.85914171e-01 -1.01952457e+00
9.46301162e-01 8.86522055e-01 1.61081001e-01 -3.64530891e-01
-1.11017816e-01 6.50662422e-01 8.39346200e-02 -2.62927532e-01
3.42140436e-01 -4.80730802e-01 -4.97102737e-02 5.12952209e-01
6.29788280e-01 -5.08432090e-01 -3.34419906e-01 -7.83420280e-02
9.10682023e-01 -2.65449844e-02 4.00199145e-01 -1.81509942e-01
7.23758936e-01 -6.92058623e-01 2.59644836e-01 8.15881193e-01
-2.44038939e-01 7.14520395e-01 -2.06461288e-02 4.42436904e-01
-1.16993046e+00 -1.34715378e+00 -3.05192679e-01 1.67848313e+00
-3.87181967e-01 -2.04418063e-01 -8.91525447e-01 -7.03772783e-01
-3.45485151e-01 9.42128837e-01 1.30000621e-01 -1.51679486e-01
-7.82102346e-01 -6.08432949e-01 1.00634408e+00 4.93316501e-01
3.55348051e-01 -4.47995245e-01 3.87745351e-01 3.35854977e-01
-2.31180891e-01 -1.28810835e+00 -9.47269857e-01 6.52558923e-01
-7.24988759e-01 -4.33916241e-01 -9.21982706e-01 -1.12861931e+00
3.28009315e-02 2.08228990e-01 9.86190259e-01 -5.21011531e-01
1.89103276e-01 4.77213800e-01 -6.10165223e-02 -3.69825155e-01
-8.74056518e-01 4.34802055e-01 4.65256482e-01 -2.16907591e-01
4.72920626e-01 -7.41257727e-01 -2.45625332e-01 6.00932777e-01
-4.54516053e-01 -3.22140038e-01 4.86657500e-01 1.09487939e+00
2.87851900e-01 -1.91451520e-01 1.16132510e+00 -5.86569190e-01
7.30078936e-01 -3.08343977e-01 -4.58306760e-01 3.77464324e-01
-3.71096104e-01 5.45112342e-02 3.46493810e-01 -5.34800112e-01
-1.60230410e+00 3.49765092e-01 -8.16930890e-01 -8.06010440e-02
-5.92437446e-01 7.71571398e-02 -6.04892731e-01 2.28315368e-01
9.95323062e-01 2.92347759e-01 -3.00690711e-01 -7.63162136e-01
5.24928629e-01 1.47911131e+00 8.60094130e-01 -8.91004384e-01
5.85072279e-01 -2.12472621e-02 -8.39919567e-01 -1.27407312e+00
-3.32268149e-01 -8.71761203e-01 -5.07773995e-01 7.19197020e-02
5.62876046e-01 -1.23186696e+00 -4.13376927e-01 5.10728598e-01
-1.20330322e+00 -6.23336792e-01 -1.88873634e-01 8.71278346e-01
-8.95425975e-01 4.78570551e-01 -7.13673294e-01 -9.92678761e-01
-2.56744653e-01 -9.17056799e-01 1.10960400e+00 -1.04450479e-01
-2.88375378e-01 -9.87432718e-01 3.70730698e-01 5.51299453e-01
5.63493252e-01 -9.37180161e-01 5.47376275e-01 -1.08180857e+00
6.22445252e-03 9.51889753e-02 8.13731924e-02 7.94266880e-01
3.51635903e-01 -4.65642601e-01 -1.84206057e+00 -5.76649010e-01
4.28752869e-01 -5.55211663e-01 8.65189195e-01 3.90324712e-01
1.26017928e+00 -5.01225181e-02 -1.82616815e-01 5.58497071e-01
8.91435802e-01 4.50063080e-01 5.15229404e-01 -9.74003077e-02
2.12441593e-01 5.54466307e-01 2.44869873e-01 2.11500183e-01
1.38530836e-01 1.05850375e+00 -4.92803752e-01 6.88519143e-03
-5.57684302e-01 -1.66803256e-01 7.98748672e-01 1.31935954e+00
9.76561308e-02 -1.70042977e-01 -1.06786156e+00 5.32106400e-01
-1.28010976e+00 -8.44888389e-01 3.45202774e-01 2.42893338e+00
1.31712413e+00 2.25036889e-01 3.75064373e-01 1.69028327e-01
6.83818758e-01 -1.89138427e-01 -4.11305368e-01 -3.81213188e-01
-5.54912724e-02 5.37551880e-01 4.11297321e-01 9.36325729e-01
-1.02927482e+00 6.61950290e-01 6.50794411e+00 1.07076275e+00
-1.12275314e+00 5.07181644e-01 6.17798626e-01 -1.59791052e-01
-1.59171954e-01 -3.67929012e-01 -1.06326723e+00 1.58885494e-01
1.84049058e+00 -1.68221533e-01 5.35206497e-01 8.00522864e-01
1.93092123e-01 2.98780650e-01 -1.43989015e+00 1.03823578e+00
-1.41836315e-01 -8.29696953e-01 -4.06284302e-01 2.62492951e-02
6.67285681e-01 2.14801803e-01 2.27840796e-01 5.95847428e-01
3.30774963e-01 -9.54878628e-01 4.70603198e-01 -4.31782100e-03
1.12122011e+00 -6.25181735e-01 5.72165906e-01 4.33889091e-01
-1.07198524e+00 6.68517128e-02 -2.11552501e-01 3.42281491e-01
3.37668300e-01 4.13203061e-01 -1.39189911e+00 6.18021607e-01
3.92912239e-01 2.12120824e-02 1.82963297e-01 8.48633230e-01
-1.00508913e-01 1.09262300e+00 -4.95688409e-01 3.89029622e-01
-1.67190358e-01 2.55101949e-01 3.66152465e-01 1.77662981e+00
6.04600966e-01 -2.39015240e-02 -3.99502181e-02 2.78090149e-01
-1.53804988e-01 -6.99572489e-02 -4.53877360e-01 1.79390058e-01
7.13280857e-01 6.12698972e-01 -5.59537672e-02 -4.17097837e-01
-6.20289326e-01 9.43672299e-01 1.43954113e-01 6.38957918e-01
-8.26798499e-01 -4.41642165e-01 6.98274612e-01 -1.92009687e-01
2.94785112e-01 -3.59953701e-01 -2.76974082e-01 -1.09778953e+00
-3.24758440e-01 -1.03310037e+00 2.67755836e-01 -4.22638983e-01
-1.33783734e+00 5.65869153e-01 7.29901940e-02 -9.55307126e-01
-8.50099266e-01 -5.17795324e-01 -4.85871375e-01 1.36846304e+00
-1.54745257e+00 -9.03104544e-01 2.23079324e-01 8.02630663e-01
9.72494304e-01 -5.05756080e-01 1.17273271e+00 6.48156345e-01
-2.39526287e-01 1.01558828e+00 3.11379999e-01 3.16342711e-02
1.22747254e+00 -1.15543354e+00 6.49738610e-01 7.14185119e-01
4.77063328e-01 6.27117217e-01 7.07127392e-01 -1.87913492e-01
-8.94245625e-01 -7.37884402e-01 1.06191158e+00 -2.00449124e-01
5.84705830e-01 -6.87933326e-01 -1.14050913e+00 6.94311738e-01
3.77147645e-01 -3.41325283e-01 1.04185653e+00 4.25918996e-01
-4.78920370e-01 -2.75899649e-01 -9.52283382e-01 2.34130174e-01
9.87059951e-01 -1.08833694e+00 -8.02160978e-01 2.82252550e-01
9.12663043e-01 -4.68043029e-01 -8.33474636e-01 2.31707051e-01
4.38262105e-01 -6.95731521e-01 1.17126977e+00 -4.26637858e-01
-2.85300463e-01 -1.30418018e-01 -5.75658798e-01 -1.62618625e+00
6.18034154e-02 -8.01117480e-01 -1.32007003e-01 1.57095969e+00
7.51210392e-01 -6.16846561e-01 6.10729992e-01 2.86035687e-01
-4.58530962e-01 -1.12450048e-02 -1.16334820e+00 -1.19112360e+00
4.16461647e-01 -7.30260134e-01 4.75121856e-01 7.52817869e-01
1.59288913e-01 6.05381012e-01 -4.84870851e-01 3.31471145e-01
6.50741935e-01 -3.46053779e-01 7.90940642e-01 -7.40569651e-01
-8.24705422e-01 -2.56916106e-01 2.64556289e-01 -1.75585997e+00
1.94887102e-01 -6.62398338e-01 4.75323826e-01 -8.17779124e-01
-1.31672591e-01 -4.97345716e-01 -3.40672940e-01 2.12964550e-01
2.38010995e-02 -5.79320900e-02 -1.72588825e-01 -4.79357727e-02
-1.36873603e-01 4.18038845e-01 6.19983494e-01 -2.19249740e-01
-4.31655407e-01 4.74618256e-01 -4.85426694e-01 4.21797186e-01
1.16133761e+00 -2.66145766e-01 -7.56527126e-01 -3.82011414e-01
-4.96606886e-01 2.66334325e-01 2.87105087e-02 -1.05289185e+00
4.55554873e-01 1.54634461e-01 2.26570666e-02 -3.07657719e-01
6.14500999e-01 -6.02685392e-01 -1.18334077e-01 2.50515025e-02
-4.98384565e-01 -6.70456767e-01 4.03296769e-01 5.54306269e-01
-2.99435586e-01 -2.45040521e-01 9.03944016e-01 3.17304224e-01
-6.26329899e-01 -1.51783869e-01 -3.14751446e-01 -7.23528489e-02
3.85995090e-01 1.22343674e-01 -2.96152622e-01 -5.20349741e-01
-1.14818656e+00 -7.75430351e-02 -1.07440650e-01 4.32718843e-01
4.14205194e-01 -1.17247748e+00 -9.81172204e-01 4.41362292e-01
-6.34868443e-02 -2.37192050e-01 3.41459244e-01 6.33711696e-01
1.79879189e-01 6.09638393e-01 2.05104306e-01 -6.84048951e-01
-1.51315296e+00 1.58960044e-01 4.88943934e-01 -9.70885307e-02
-2.32992500e-01 1.16071379e+00 2.37213299e-01 -7.12683082e-01
5.69433451e-01 -1.88844711e-01 1.37160987e-01 -2.35743642e-01
5.54687977e-01 2.13877559e-01 4.73129421e-01 -4.14010823e-01
-3.33142579e-01 3.03158373e-01 -1.20108426e-01 -7.81416059e-01
1.05917239e+00 -4.76419806e-01 5.68056822e-01 6.10695243e-01
1.34891546e+00 2.49814823e-01 -1.20413721e+00 -5.52367151e-01
1.64927229e-01 -1.38589308e-01 1.04842886e-01 -9.09551620e-01
-5.77893853e-01 1.07223570e+00 7.43518114e-01 2.41092563e-01
1.19847393e+00 1.22455969e-01 6.62317455e-01 3.99080664e-01
2.56652415e-01 -1.28432322e+00 -2.99640149e-02 7.23027706e-01
8.89471292e-01 -1.20339298e+00 -3.40338141e-01 -2.57955790e-01
-4.69154388e-01 1.06546354e+00 4.15165126e-01 4.47778225e-01
7.86356091e-01 5.80347776e-01 2.16861531e-01 4.77624655e-01
-7.70950973e-01 -1.89399086e-02 3.49691093e-01 9.22443807e-01
8.00827622e-01 6.61575496e-02 -4.06507589e-02 7.62305379e-01
-5.39068401e-01 -1.47968888e-01 1.53555468e-01 5.68552315e-01
-5.93675256e-01 -1.43521941e+00 -6.18626177e-01 4.86681871e-02
-3.54701340e-01 -4.65192497e-02 -1.76309377e-01 5.42141974e-01
-2.23649710e-01 1.38226318e+00 -1.30059406e-01 -4.65465903e-01
3.55900347e-01 7.29010940e-01 4.69113350e-01 -6.63675129e-01
-3.25660914e-01 5.68723142e-01 5.50329328e-01 -1.25173524e-01
-3.72328639e-01 -8.54645848e-01 -1.11172557e+00 -1.50009736e-01
-3.94950062e-01 3.62811565e-01 9.96504426e-01 8.22983861e-01
-1.85947847e-02 6.69594347e-01 7.35142708e-01 -7.65749156e-01
-1.21361303e+00 -1.25326204e+00 -7.10843623e-01 3.49416956e-02
5.94210207e-01 -2.39481822e-01 -5.47452986e-01 3.46448362e-01]
|
[14.560944557189941, 6.482390880584717]
|
68fdbeec-a827-4c5a-ab4a-4f4f887bd0a5
|
local-network-community-detection-with
|
1601.05775
| null |
http://arxiv.org/abs/1601.05775v2
|
http://arxiv.org/pdf/1601.05775v2.pdf
|
Local Network Community Detection with Continuous Optimization of Conductance and Weighted Kernel K-Means
|
Local network community detection is the task of finding a single community
of nodes concentrated around few given seed nodes in a localized way.
Conductance is a popular objective function used in many algorithms for local
community detection. This paper studies a continuous relaxation of conductance.
We show that continuous optimization of this objective still leads to discrete
communities. We investigate the relation of conductance with weighted kernel
k-means for a single community, which leads to the introduction of a new
objective function, $\sigma$-conductance. Conductance is obtained by setting
$\sigma$ to $0$. Two algorithms, EMc and PGDc, are proposed to locally optimize
$\sigma$-conductance and automatically tune the parameter $\sigma$. They are
based on expectation maximization and projected gradient descent, respectively.
We prove locality and give performance guarantees for EMc and PGDc for a class
of dense and well separated communities centered around the seeds. Experiments
are conducted on networks with ground-truth communities, comparing to
state-of-the-art graph diffusion algorithms for conductance optimization. On
large graphs, results indicate that EMc and PGDc stay localized and produce
communities most similar to the ground, while graph diffusion algorithms
generate large communities of lower quality.
|
['Twan van Laarhoven', 'Elena Marchiori']
|
2016-01-21
| null | null | null | null |
['local-community-detection']
|
['graphs']
|
[-2.01209057e-02 1.92541450e-01 8.68346244e-02 2.20750749e-01
-4.32692230e-01 -6.62097037e-01 2.56485343e-01 7.77665079e-01
-4.13930446e-01 4.24184322e-01 -2.13217229e-01 -2.15171248e-01
-2.52638876e-01 -1.21755064e+00 -3.01360279e-01 -8.29007030e-01
-8.03091109e-01 5.58580279e-01 5.58787942e-01 -9.46312547e-02
4.16938871e-01 4.13546622e-01 -6.85234249e-01 -1.19167961e-01
1.00173914e+00 3.17524791e-01 1.20981678e-01 8.45765412e-01
-1.74072161e-01 3.77834618e-01 -4.52704221e-01 -3.79642725e-01
3.44245076e-01 -7.40884304e-01 -8.40642869e-01 3.12871963e-01
-2.61182105e-03 4.31744218e-01 -4.55130041e-02 1.33340490e+00
4.24514890e-01 -1.84416607e-01 6.18351877e-01 -1.32516658e+00
-7.19806671e-01 1.13075447e+00 -1.09124625e+00 2.85429657e-01
2.52746969e-01 -8.95200446e-02 1.23809743e+00 -6.34069443e-01
1.03096879e+00 1.22711527e+00 9.59509909e-01 2.81107426e-01
-1.70032060e+00 -3.86472672e-01 2.23839089e-01 -3.13620985e-01
-2.12065291e+00 -5.67135476e-02 7.70649672e-01 -7.35315681e-01
8.12762320e-01 4.45972607e-02 7.66732454e-01 2.17750654e-01
4.37587649e-02 3.58298779e-01 9.26898420e-01 -6.40909135e-01
5.54117501e-01 1.42868266e-01 -1.84144869e-01 9.18129921e-01
7.43984103e-01 -3.67509156e-01 -2.60027915e-01 -7.03288913e-01
8.32611024e-01 -3.27259690e-01 -4.07441527e-01 -8.04700732e-01
-1.14954007e+00 1.29295135e+00 8.33562791e-01 6.34481609e-01
-3.51025552e-01 3.83961529e-01 1.18783928e-01 2.95759022e-01
4.98975337e-01 2.60958612e-01 2.10324064e-01 3.26478601e-01
-1.10098910e+00 2.64149029e-02 1.16665614e+00 9.09337580e-01
1.02834022e+00 -2.85117984e-01 2.64835149e-01 5.91362774e-01
4.14046317e-01 6.59335017e-01 -2.26260155e-01 -8.35709810e-01
3.85135829e-01 1.09173226e+00 -9.65048820e-02 -1.65133262e+00
-1.70759708e-01 -4.84823674e-01 -1.32157695e+00 -5.25484197e-02
3.51607591e-01 -2.39887297e-01 -4.82201725e-01 1.58708453e+00
4.68658209e-01 6.84261993e-02 -6.98258430e-02 8.57868969e-01
2.60552973e-01 5.41484833e-01 -4.08764720e-01 -4.93020743e-01
6.85235620e-01 -9.33409631e-01 -3.44180167e-01 -6.38798475e-02
6.39995992e-01 -5.51472843e-01 5.47525108e-01 4.51296270e-02
-1.20424271e+00 8.48070905e-02 -8.12479675e-01 5.74226201e-01
-2.16564879e-01 -3.10757071e-01 4.77437526e-01 9.68797684e-01
-1.75371325e+00 7.29339540e-01 -7.37078846e-01 -6.63566053e-01
2.45790988e-01 3.24637860e-01 -1.11525103e-01 2.87328325e-02
-5.68372726e-01 2.53084898e-01 1.46801576e-01 7.53890350e-02
-8.23716044e-01 -2.55835056e-01 -5.25520861e-01 -5.48049621e-02
3.65793079e-01 -5.45656145e-01 3.16089034e-01 -8.04683149e-01
-8.80909443e-01 1.15885937e+00 -3.58124077e-02 -5.81282198e-01
3.75637114e-01 7.17290044e-01 -1.10586509e-01 5.84654391e-01
4.58082497e-01 5.57484269e-01 5.84446847e-01 -1.34681320e+00
-2.74496406e-01 -1.23566583e-01 -3.90293986e-01 -1.80143360e-02
-5.77469289e-01 1.03560545e-01 -2.26776585e-01 -2.78382838e-01
5.41541040e-01 -8.47845316e-01 -7.42234826e-01 -3.54070924e-02
-5.56182742e-01 -1.62383199e-01 5.12199998e-01 -3.23210001e-01
1.60791838e+00 -1.68488216e+00 4.03747439e-01 1.13015103e+00
8.79340410e-01 -1.93412066e-01 -2.98521399e-01 7.56264210e-01
1.69117257e-01 8.02852869e-01 -6.45562112e-01 -1.91211373e-01
-2.82984853e-01 -2.33637020e-02 3.75186592e-01 9.31571603e-01
2.94878427e-02 5.69530070e-01 -1.10187531e+00 -5.81286013e-01
-4.08139020e-01 3.09884310e-01 -8.17851186e-01 -3.07697594e-01
1.27959237e-01 -5.48531823e-02 -3.53483319e-01 5.96578181e-01
7.64923990e-01 -8.15070570e-01 6.15786850e-01 4.93165463e-01
-2.40374908e-01 -3.38624716e-01 -1.67277288e+00 1.43102324e+00
2.66555488e-01 3.36542964e-01 8.07700038e-01 -1.12526870e+00
1.30065048e+00 5.73418438e-02 7.01110899e-01 8.17875937e-02
1.10253878e-01 2.83880830e-01 9.39492509e-02 1.12686858e-01
2.62213767e-01 3.41937020e-02 1.18439719e-01 9.47363019e-01
-3.41920286e-01 -1.13812424e-01 5.91619253e-01 9.03393924e-01
1.65156484e+00 -8.89352560e-01 6.69615120e-02 -9.47036803e-01
4.35537845e-01 4.33533005e-02 3.28227073e-01 1.01245451e+00
-3.24773014e-01 4.77605522e-01 6.94500208e-01 7.96226189e-02
-1.14655077e+00 -1.13723552e+00 4.44844037e-01 5.63467979e-01
4.52503651e-01 -5.36437035e-01 -1.12859356e+00 -3.68959159e-01
1.24948911e-01 -1.62752047e-01 -6.11993015e-01 3.98055501e-02
-5.39432704e-01 -7.92541623e-01 3.19662780e-01 8.90269056e-02
3.15583587e-01 -8.09846342e-01 -2.98266858e-01 4.96601701e-01
-3.04842412e-01 -7.33989954e-01 -9.59290206e-01 2.62966976e-02
-1.10142267e+00 -1.14684582e+00 -8.32214177e-01 -9.83409762e-01
1.18744481e+00 5.87908149e-01 1.27531314e+00 8.44260693e-01
-4.63118076e-01 4.77895617e-01 -3.97729576e-01 4.04454410e-01
-4.87613201e-01 3.65583092e-01 -8.45850408e-02 1.82469487e-01
1.16114959e-01 -8.23840559e-01 -7.39411592e-01 2.69144118e-01
-6.28640056e-01 -3.52759033e-01 3.43729109e-01 5.72649002e-01
6.44242406e-01 3.42662096e-01 3.07391673e-01 -7.10857689e-01
1.18910182e+00 -5.92154443e-01 -6.75165534e-01 2.50003725e-01
-9.70888913e-01 7.74601251e-02 1.43624827e-01 -5.25556087e-01
-2.71269739e-01 -2.20383499e-02 5.12216330e-01 -2.53065407e-01
5.14115572e-01 7.85617411e-01 2.60959923e-01 -4.76714224e-01
9.53705490e-01 1.07709557e-01 1.04909003e-01 -1.70109868e-01
4.77347314e-01 4.77016121e-01 2.79717371e-02 -5.97340763e-01
8.99074137e-01 9.21675622e-01 -1.56638667e-01 -9.27619636e-01
-5.76648340e-02 -9.61140215e-01 -7.13906527e-01 -3.25895399e-01
3.90852988e-01 -7.25087583e-01 -5.61797202e-01 2.43251741e-01
-1.03031969e+00 -4.78606910e-01 -2.85386682e-01 3.29879969e-02
-2.86605299e-01 6.80754721e-01 -7.54541397e-01 -1.03974962e+00
-5.69738030e-01 -6.75048590e-01 7.54382849e-01 6.71458542e-02
2.85578016e-02 -1.24736464e+00 4.62984443e-01 -2.21206620e-01
4.39852774e-01 4.07508701e-01 6.16471171e-01 -2.22087950e-01
-7.10641801e-01 -1.77789569e-01 -4.12526429e-01 -9.78189241e-03
6.82115834e-03 3.46916765e-01 -1.72408268e-01 -8.71274352e-01
-4.54905987e-01 1.04513280e-01 8.95311058e-01 7.30737686e-01
4.10489291e-01 -4.93385583e-01 -7.93674529e-01 4.10832673e-01
1.90827370e+00 -7.36304373e-02 6.48584545e-01 1.47575885e-01
4.80744421e-01 4.38262492e-01 -1.30660117e-01 5.43315291e-01
2.04315379e-01 1.26554042e-01 4.93000239e-01 -1.54705942e-01
1.48004651e-01 -1.24003686e-01 6.36451617e-02 1.09884894e+00
-2.45321557e-01 -3.86854261e-01 -1.15797222e+00 1.08755136e+00
-1.85491645e+00 -1.07459152e+00 -5.17799258e-01 2.17684603e+00
9.46303368e-01 2.33331323e-01 5.10772705e-01 1.66346617e-02
1.40156507e+00 -1.70310400e-02 -1.88334942e-01 -2.34650433e-01
-3.53223592e-01 1.81351736e-01 6.20795250e-01 9.59889710e-01
-7.85918832e-01 8.81546378e-01 6.51963091e+00 6.89892054e-01
-8.06154966e-01 1.10750496e-01 5.17844319e-01 2.04761922e-01
-4.82261062e-01 3.72747004e-01 -6.59484267e-01 2.11158782e-01
6.08599663e-01 -4.96388763e-01 7.17478454e-01 8.40502322e-01
3.85330260e-01 -9.38121676e-02 -6.16841197e-01 7.49651551e-01
-1.09644860e-01 -1.50037372e+00 -2.76352733e-01 4.98973757e-01
1.38017142e+00 3.56790125e-02 -3.74705613e-01 -3.67984951e-01
8.28800440e-01 -8.38170290e-01 5.01326382e-01 2.50489920e-01
5.74387670e-01 -6.69901669e-01 2.27473035e-01 3.82344931e-01
-1.78848052e+00 -2.85198130e-02 -6.02295041e-01 8.96609724e-02
1.95359200e-01 1.08589494e+00 -9.14546192e-01 1.80456683e-01
6.93683624e-01 5.51270545e-01 -5.81634402e-01 1.30814278e+00
9.68840122e-02 4.64515656e-01 -7.62513816e-01 -5.13033986e-01
2.98946142e-01 -5.42468786e-01 8.39506328e-01 1.43274152e+00
3.62233013e-01 -8.42091218e-02 4.25364971e-01 1.36166894e+00
-2.06999421e-01 3.83931547e-01 -5.97357631e-01 -3.51138651e-01
7.43738234e-01 1.27952933e+00 -1.54334009e+00 -1.16476454e-01
1.03978835e-01 1.09062517e+00 5.56066334e-01 3.19406629e-01
-4.95047092e-01 -5.90280235e-01 2.67812729e-01 6.62387550e-01
4.04420257e-01 -5.15020549e-01 -1.39805660e-01 -9.40142214e-01
-8.77930149e-02 -5.92265844e-01 4.02032912e-01 -1.59726292e-01
-1.44268227e+00 5.16650975e-01 -2.20703378e-01 -6.96789861e-01
2.04333588e-01 -7.14154616e-02 -9.11128581e-01 8.22071016e-01
-1.07504678e+00 -6.97174907e-01 -3.48259866e-01 6.99992001e-01
-1.39404684e-01 1.18165717e-01 5.39737880e-01 9.38020125e-02
-3.48462105e-01 3.59916896e-01 3.72590929e-01 2.21742839e-01
2.13158444e-01 -1.37665880e+00 3.81033450e-01 1.22225392e+00
-2.84360768e-03 8.22555006e-01 5.28774261e-01 -1.07328129e+00
-1.03264618e+00 -9.16029930e-01 1.16500473e+00 1.75928608e-01
9.28386092e-01 -5.10574341e-01 -7.81428993e-01 1.34305879e-01
2.88649648e-01 3.31763849e-02 2.46678129e-01 -2.63854899e-02
-1.33555517e-01 1.02519378e-01 -1.41309607e+00 5.49796343e-01
1.28083348e+00 -3.94948006e-01 3.40710223e-01 4.59434658e-01
5.61080039e-01 2.82619864e-01 -9.31543946e-01 -5.77875860e-02
1.89457871e-02 -9.34970260e-01 9.97902155e-01 -1.93116620e-01
-2.10497435e-02 -4.13889617e-01 8.62824097e-02 -1.22813320e+00
-6.10196769e-01 -9.95268404e-01 1.32930398e-01 1.05026650e+00
6.17238998e-01 -5.52185237e-01 1.18486571e+00 -8.18562582e-02
4.52967674e-01 -5.66791356e-01 -8.54559660e-01 -8.47962976e-01
2.49810949e-01 2.64729977e-01 1.13251716e-01 1.19083881e+00
2.55287826e-01 3.33495408e-01 1.18556321e-01 2.16717526e-01
1.25028753e+00 1.16109900e-01 5.22398055e-01 -1.52557325e+00
-1.79035157e-01 -7.98607171e-01 -2.29792669e-01 -8.55570018e-01
-3.11874598e-01 -9.64478135e-01 5.19987941e-02 -1.76636422e+00
4.67782915e-01 -8.39936733e-01 2.12857276e-01 1.49123907e-01
1.68528128e-02 2.80122906e-01 -1.29686832e-01 2.48882458e-01
-6.90491319e-01 1.22855440e-01 1.23161077e+00 -3.77024740e-01
-6.42185509e-01 -8.29928666e-02 -7.18171597e-01 4.31674391e-01
8.15975785e-01 -6.71320975e-01 -2.79575169e-01 -1.29171878e-01
5.45401454e-01 2.73689367e-02 2.47757152e-01 -8.53356838e-01
6.74625635e-01 -2.01790944e-01 -2.44425118e-01 -4.98705506e-01
-7.15711042e-02 -4.89224195e-01 3.07149917e-01 7.70568252e-01
-2.38887087e-01 2.33004987e-01 -3.14780414e-01 8.92192543e-01
5.11346608e-02 -3.75017196e-01 9.34324861e-01 -3.60214442e-01
-2.57328987e-01 2.38398209e-01 -4.76897597e-01 3.17298621e-01
1.02674580e+00 -5.10169625e-01 -1.86973512e-01 -6.16413593e-01
-9.24081564e-01 3.61607075e-01 6.32533312e-01 -2.16032088e-01
7.39766657e-01 -1.11085999e+00 -1.18584180e+00 -1.68288380e-01
-3.56429182e-02 -1.72595963e-01 -2.12717310e-01 9.67295527e-01
-8.52476776e-01 -9.31890905e-02 4.55776930e-01 -8.65560412e-01
-1.42608714e+00 5.83439648e-01 5.62222838e-01 -5.18256366e-01
-3.63201648e-01 1.13843012e+00 -2.81414360e-01 -3.39409053e-01
1.05872408e-01 -1.55134767e-01 2.42827043e-01 -5.22363223e-02
2.10172951e-01 4.48523670e-01 -3.15794438e-01 -4.48016107e-01
-5.11449695e-01 5.75522780e-01 2.45768264e-01 -2.31131405e-01
1.29940331e+00 -4.33899403e-01 -7.83801138e-01 -1.47671819e-01
1.11506641e+00 2.55364805e-01 -8.88096631e-01 -2.02909902e-01
2.23557413e-01 -4.37495142e-01 -1.24663189e-01 -3.01919281e-01
-1.33868015e+00 5.10533810e-01 5.50865591e-01 8.62166226e-01
7.55516708e-01 2.15089172e-01 2.21261874e-01 2.20124632e-01
4.72621322e-01 -1.14274156e+00 4.01907712e-01 3.45220685e-01
5.85717916e-01 -8.66683662e-01 1.23446859e-01 -8.18375945e-01
-1.60914987e-01 9.17753816e-01 3.26093912e-01 -3.60503763e-01
9.29640830e-01 4.46184456e-01 -3.20229888e-01 -4.68748927e-01
-4.25443977e-01 -3.01377863e-01 -1.88571051e-01 7.30152607e-01
2.36976072e-01 1.88862145e-01 -4.87520188e-01 4.11584526e-02
1.69563204e-01 -4.50648218e-01 6.45184755e-01 9.44796920e-01
-9.58648562e-01 -1.11214447e+00 -4.88245875e-01 2.80356526e-01
-6.62325546e-02 -2.65397877e-01 -8.73937666e-01 6.23041034e-01
-5.52695319e-02 1.17841494e+00 -5.72372414e-02 -2.00023592e-01
-6.91541210e-02 -4.05278176e-01 3.13886791e-01 -6.01439953e-01
-6.68924451e-01 3.09503555e-01 -1.22450441e-01 -1.17890693e-01
-5.93986392e-01 -5.59526205e-01 -1.23185897e+00 -9.07702923e-01
-9.13051188e-01 7.13572562e-01 3.50395977e-01 3.42343420e-01
2.02258274e-01 4.79932316e-02 8.49778295e-01 -5.01650274e-01
-3.91601324e-01 -7.23232090e-01 -1.01857865e+00 1.67835310e-01
-3.41786370e-02 -1.83681071e-01 -8.72060478e-01 -6.79545512e-04]
|
[6.976230621337891, 5.2472076416015625]
|
4eb2b9e6-a64c-4a2b-abed-4a0fca887cd3
|
what-makes-imagenet-look-unlike-laion
|
2306.15769
| null |
https://arxiv.org/abs/2306.15769v1
|
https://arxiv.org/pdf/2306.15769v1.pdf
|
What Makes ImageNet Look Unlike LAION
|
ImageNet was famously created from Flickr image search results. What if we recreated ImageNet instead by searching the massive LAION dataset based on image captions alone? In this work, we carry out this counterfactual investigation. We find that the resulting ImageNet recreation, which we call LAIONet, looks distinctly unlike the original. Specifically, the intra-class similarity of images in the original ImageNet is dramatically higher than it is for LAIONet. Consequently, models trained on ImageNet perform significantly worse on LAIONet. We propose a rigorous explanation for the discrepancy in terms of a subtle, yet important, difference in two plausible causal data-generating processes for the respective datasets, that we support with systematic experimentation. In a nutshell, searching based on an image caption alone creates an information bottleneck that mitigates the selection bias otherwise present in image-based filtering. Our explanation formalizes a long-held intuition in the community that ImageNet images are stereotypical, unnatural, and overly simple representations of the class category. At the same time, it provides a simple and actionable takeaway for future dataset creation efforts.
|
['Moritz Hardt', 'Ali Shirali']
|
2023-06-27
| null | null | null | null |
['image-retrieval', 'image-captioning', 'selection-bias']
|
['computer-vision', 'computer-vision', 'natural-language-processing']
|
[ 5.22027731e-01 1.95259348e-01 -2.00031102e-01 -3.96068633e-01
-4.28148389e-01 -1.04460227e+00 1.03645802e+00 -6.51157508e-03
-5.36076128e-01 5.31209111e-01 6.03788257e-01 -6.24637008e-01
-2.52290487e-01 -7.57291973e-01 -9.92662549e-01 -4.55595821e-01
3.16392511e-01 1.07065670e-01 -9.39156339e-02 -3.13973755e-01
4.59822178e-01 2.08538085e-01 -1.95258081e+00 4.46536362e-01
7.08758473e-01 8.13159347e-01 2.53575981e-01 6.10373735e-01
4.74299490e-02 9.22946095e-01 -3.49109530e-01 -7.06732094e-01
7.18172133e-01 -5.27049780e-01 -8.20036530e-01 2.98932433e-01
1.27860379e+00 -2.74468452e-01 -5.86532474e-01 1.35797143e+00
2.47558460e-01 -9.75448489e-02 4.19194162e-01 -1.33682144e+00
-1.10151517e+00 5.40577888e-01 -5.26748598e-01 1.61893159e-01
2.24510625e-01 7.67031372e-01 1.14621329e+00 -6.12372458e-01
1.04412043e+00 1.52071464e+00 8.59304965e-01 3.19908261e-01
-1.51057696e+00 -5.45638800e-01 1.20009311e-01 -1.38243698e-02
-1.34132826e+00 -4.23238039e-01 5.24214804e-01 -6.48145974e-01
2.03144237e-01 8.86678517e-01 8.31991255e-01 1.08919585e+00
-2.87676416e-02 5.40403783e-01 1.41758645e+00 -4.11138862e-01
-1.83979914e-01 3.43896806e-01 -1.89154476e-01 6.05566502e-01
5.82964301e-01 3.12607497e-01 -2.36646116e-01 -4.22117352e-01
9.31005597e-01 -1.08794458e-02 -4.18780237e-01 -4.55028594e-01
-1.41119444e+00 9.17879045e-01 8.63199294e-01 3.31444055e-01
-4.43893045e-01 2.72886366e-01 1.78111881e-01 4.31535870e-01
4.01448935e-01 9.06343102e-01 -3.26358318e-01 1.96368635e-01
-1.00599623e+00 4.36503649e-01 5.41898608e-01 5.85769415e-01
6.95681512e-01 -2.75287181e-01 -2.31668726e-01 6.08976483e-01
7.36670867e-02 5.78714669e-01 4.08079505e-01 -1.18263066e+00
6.68367520e-02 4.41906184e-01 3.56260538e-01 -1.60071993e+00
8.02324638e-02 -3.99205089e-01 -5.37375569e-01 2.21954793e-01
6.25919044e-01 9.51333120e-02 -1.06321919e+00 1.90648437e+00
-6.96911737e-02 -2.21220538e-01 -3.17824423e-01 1.16187739e+00
5.99845648e-01 2.64412254e-01 3.61245781e-01 1.25157863e-01
1.46438479e+00 -6.48626626e-01 -4.51794893e-01 -3.77097577e-01
4.01565969e-01 -9.28185523e-01 1.56175697e+00 1.33660361e-02
-8.28096092e-01 -4.52997178e-01 -8.89164448e-01 1.16015732e-01
-5.47655642e-01 -5.07021695e-02 8.48342001e-01 5.98653555e-01
-1.19100559e+00 5.31653285e-01 6.23304658e-02 -1.00216532e+00
3.63571972e-01 -1.77058816e-01 -1.94170401e-01 -4.31913324e-02
-1.12178636e+00 7.81129599e-01 3.00316006e-01 -1.70237035e-01
-7.58047640e-01 -7.87597001e-01 -5.07685184e-01 -1.49387598e-01
4.55794603e-01 -6.41258478e-01 1.10902536e+00 -1.56174338e+00
-5.60675263e-01 1.37881804e+00 9.76679730e-04 -6.18128777e-01
7.29693413e-01 1.06621712e-01 -4.41973090e-01 1.68497577e-01
3.85562241e-01 1.07140827e+00 8.58108103e-01 -1.70555115e+00
-4.31281447e-01 -3.54024053e-01 2.28982225e-01 2.16283262e-01
-3.23868170e-02 -7.96066374e-02 -2.19644308e-01 -9.79629934e-01
-1.60626203e-01 -1.09070933e+00 -4.59827483e-01 1.99202657e-01
-3.44758242e-01 8.05987194e-02 4.48732525e-01 -4.97035146e-01
1.26041007e+00 -2.21841645e+00 -6.27881825e-01 2.62154698e-01
5.43780088e-01 1.68823019e-01 -4.52070862e-01 3.95743996e-01
-3.19536358e-01 9.46893096e-01 5.26848249e-02 5.44485331e-01
-5.91761880e-02 -9.09027159e-02 -6.15676880e-01 4.60057914e-01
-1.22908661e-02 1.16123176e+00 -1.16954613e+00 -5.53894937e-01
1.27491996e-01 1.15948118e-01 -5.47750533e-01 -3.02288949e-01
-2.71259040e-01 1.09013043e-01 -2.68871665e-01 4.34130788e-01
8.23674738e-01 -4.26832795e-01 2.51361579e-01 -5.10910451e-01
-4.74855214e-01 8.70006233e-02 -5.06415665e-01 1.48794603e+00
-7.07621500e-02 1.01522410e+00 -1.26045600e-01 -5.24277627e-01
5.20124555e-01 -2.70842873e-02 3.01143199e-01 -8.43655944e-01
3.93293649e-02 2.54976630e-01 1.84507072e-02 -5.46403587e-01
7.61950195e-01 -2.25596085e-01 1.13500664e-02 6.21783376e-01
-2.81536162e-01 -4.33001071e-01 1.22848414e-01 3.82706821e-01
1.10669076e+00 -2.00470656e-01 1.28210723e-01 -5.62044799e-01
7.90864825e-02 5.82910180e-01 1.49441898e-01 1.50546300e+00
-3.67326766e-01 8.57811034e-01 1.37117818e-01 -6.43146753e-01
-1.39646101e+00 -1.01388466e+00 -1.26228854e-01 7.64237344e-01
4.25050765e-01 -4.24523860e-01 -7.12296069e-01 -7.85377562e-01
1.53429881e-01 8.51502180e-01 -9.40275550e-01 -1.24197133e-01
-1.76532492e-01 -5.65872967e-01 7.86427498e-01 -2.40581080e-01
7.59726942e-01 -9.61846113e-01 -7.80907214e-01 -2.10287988e-01
-4.16529000e-01 -7.62547791e-01 -6.00797355e-01 -4.49260771e-01
-6.46797061e-01 -1.21976829e+00 -6.67127490e-01 -4.46674466e-01
6.05103850e-01 9.51363862e-01 1.56505001e+00 5.75986445e-01
-3.23563606e-01 5.67774296e-01 -2.00695649e-01 -4.60618585e-01
-4.94654179e-01 -1.98518172e-01 -2.58810192e-01 7.33158551e-03
6.11201763e-01 -3.05959284e-01 -8.69181871e-01 3.97359520e-01
-1.28496468e+00 1.54795468e-01 8.78978670e-01 6.72578096e-01
3.16742331e-01 9.75313708e-02 2.36435428e-01 -9.36917722e-01
7.91296721e-01 -6.10480607e-01 -3.58132154e-01 5.36334872e-01
-8.09349179e-01 2.14392077e-02 1.53438568e-01 -4.20296043e-01
-1.14278352e+00 -7.11950436e-02 3.28004420e-01 -4.76256162e-01
-2.01945901e-01 3.98769975e-01 3.66201371e-01 -1.72213867e-01
1.09765267e+00 2.61188835e-01 1.39786914e-01 -5.63519120e-01
7.15445518e-01 5.91456115e-01 5.41937411e-01 -2.51988649e-01
1.02176428e+00 9.59178627e-01 -4.88988400e-01 -7.78668404e-01
-8.57401431e-01 -2.11560220e-01 -1.77131191e-01 -4.22855973e-01
7.55058050e-01 -9.84329820e-01 -6.27488375e-01 1.88073501e-01
-1.08111238e+00 -2.39429682e-01 -5.32233894e-01 2.24030480e-01
-4.44946170e-01 1.00084633e-01 -7.91760162e-02 -5.22533000e-01
-3.45792715e-03 -8.70842636e-01 7.84069836e-01 5.95853925e-02
-3.64124000e-01 -7.59368360e-01 -3.39692496e-02 3.88450176e-01
6.99829340e-01 3.65156949e-01 6.95971489e-01 -5.16265750e-01
-8.17084789e-01 -2.66814500e-01 -7.79251993e-01 9.94855165e-02
1.63581550e-01 9.26801935e-02 -9.62155879e-01 -8.83914307e-02
6.05251491e-02 -3.94106239e-01 1.08037055e+00 1.25203386e-01
1.06938028e+00 -7.89813936e-01 -2.26561844e-01 2.03315675e-01
1.86297429e+00 -3.10992394e-02 7.42247045e-01 4.59547848e-01
4.50278431e-01 6.80303037e-01 5.33793926e-01 1.68754280e-01
2.75895268e-01 5.05783796e-01 4.06598657e-01 -3.28694612e-01
-4.21033561e-01 -7.77520120e-01 3.52460332e-02 6.98485300e-02
5.61107583e-02 -2.33180359e-01 -9.05383408e-01 8.12283039e-01
-1.82094777e+00 -1.35268283e+00 -2.16295287e-01 2.23571467e+00
7.54661143e-01 -4.51886095e-02 5.79052381e-02 -3.90700519e-01
8.87812078e-01 4.65349287e-01 -4.43384647e-01 -1.39902845e-01
-5.21627009e-01 -3.00768107e-01 1.04245484e+00 3.08533549e-01
-9.32578862e-01 7.42784202e-01 7.33790207e+00 8.65620136e-01
-1.10313702e+00 9.36513767e-02 8.95011604e-01 -1.13606684e-01
-7.86398351e-01 2.85137892e-01 -7.25204572e-02 4.18417126e-01
7.52865553e-01 -5.45191526e-01 6.05100155e-01 6.73308134e-01
3.95352721e-01 -2.65394092e-01 -9.43553329e-01 1.04877222e+00
8.70420560e-02 -1.77430332e+00 5.16551316e-01 2.58951157e-01
7.13271797e-01 3.43752533e-01 3.86265635e-01 -1.70488134e-01
6.62950695e-01 -1.08595741e+00 1.10587108e+00 3.87137473e-01
7.89849401e-01 -2.13288844e-01 3.77847999e-01 5.39952070e-02
-6.35038435e-01 -1.34394526e-01 -2.60645032e-01 -2.49478459e-01
-1.10248305e-01 6.48379624e-01 -6.56264722e-01 2.33581692e-01
7.62789369e-01 4.46887940e-01 -1.26014686e+00 1.12865663e+00
6.54645488e-02 5.66866755e-01 -1.28237158e-01 1.83827937e-01
4.31944788e-01 -4.80513014e-02 7.12842643e-01 1.04975724e+00
1.97364002e-01 -4.75933850e-02 -5.87607063e-02 1.00247812e+00
-2.35224158e-01 -1.30161285e-01 -1.21335924e+00 -4.04321253e-01
4.78658319e-01 1.02912176e+00 -8.02706003e-01 -4.42142278e-01
-3.38236779e-01 7.72603452e-01 8.31183866e-02 4.82763708e-01
-7.74002373e-01 -8.62776637e-02 6.87466145e-01 3.73546064e-01
2.08702218e-02 1.94601849e-01 -4.17447984e-01 -1.31654537e+00
-1.59296021e-01 -1.05884910e+00 2.42701069e-01 -1.14256263e+00
-1.30595791e+00 5.57719588e-01 5.16376942e-02 -1.02569878e+00
1.19224355e-01 -2.39225090e-01 -4.52489972e-01 5.52005470e-01
-1.14679313e+00 -9.03839111e-01 -5.03102660e-01 2.72897959e-01
4.36766088e-01 3.87180626e-01 4.35813904e-01 2.58169651e-01
-1.65517882e-01 3.56481642e-01 3.86783890e-02 -2.13794205e-02
7.60619402e-01 -9.08652186e-01 4.97526586e-01 7.94813812e-01
4.57530648e-01 7.76495278e-01 1.12826872e+00 -7.03399777e-01
-1.03021455e+00 -1.03684342e+00 8.84667993e-01 -6.87671959e-01
7.48957694e-01 -1.65475428e-01 -6.47701740e-01 6.07602835e-01
5.59050977e-01 -4.30128366e-01 3.38168323e-01 -1.35934621e-01
-6.11342072e-01 2.15101734e-01 -1.32255912e+00 9.69994783e-01
1.32519996e+00 -5.00226378e-01 -4.36296225e-01 5.08236706e-01
8.32098484e-01 2.17406452e-01 -4.11842942e-01 2.19654024e-01
7.41188467e-01 -1.20771110e+00 1.15028632e+00 -7.28347957e-01
6.14245117e-01 -2.37978369e-01 -3.31082046e-01 -1.19723630e+00
-4.84315902e-01 -4.85564739e-01 8.21363151e-01 1.14252925e+00
4.72673774e-01 -4.86525059e-01 7.57247686e-01 8.73077154e-01
3.92085135e-01 -3.22095573e-01 -4.63158816e-01 -8.46214294e-01
-1.28222764e-01 -3.84904891e-01 7.78348982e-01 1.14864480e+00
-5.03882706e-01 1.62072167e-01 -3.42575938e-01 -2.28709489e-01
8.07760954e-01 3.28434296e-02 8.79239142e-01 -1.15886056e+00
8.85717273e-02 -4.54206944e-01 -1.64088398e-01 -8.20151746e-01
-2.95340449e-01 -7.51793146e-01 -7.13275224e-02 -1.15117216e+00
5.29251397e-01 -5.92144966e-01 -1.35230780e-01 2.70126283e-01
-8.48700330e-02 7.69800842e-01 6.16639733e-01 6.31654978e-01
-5.37159383e-01 9.51624587e-02 1.31271553e+00 -1.94568411e-01
1.26467451e-01 -4.72718447e-01 -1.34385693e+00 8.80728662e-01
8.60229313e-01 -3.75000268e-01 -3.88049424e-01 -6.04985356e-01
5.64853966e-01 -5.20541310e-01 9.83125091e-01 -6.40910685e-01
-5.60285263e-02 -4.60561216e-01 1.55814692e-01 -3.05502862e-01
2.46255919e-02 -8.50762486e-01 5.48262715e-01 4.49218273e-01
-7.64209330e-01 4.97351326e-02 1.32652462e-01 7.05176651e-01
-1.29562110e-01 -1.18916050e-01 6.97835982e-01 -7.19171584e-01
-9.44220245e-01 5.08257896e-02 -4.76702064e-01 1.13724820e-01
6.80775523e-01 -3.33769321e-01 -5.91887534e-01 -7.41247416e-01
-3.10715199e-01 -2.55155358e-02 1.06443322e+00 5.78888476e-01
3.45047504e-01 -1.47100174e+00 -6.86532259e-01 4.51053120e-02
3.74419391e-01 -6.35908604e-01 1.27092108e-01 7.42120981e-01
-5.03879786e-01 4.62579489e-01 -1.64826736e-01 -3.31198543e-01
-1.06080258e+00 6.48720980e-01 2.30843931e-01 3.42110991e-02
-2.53873974e-01 6.69166565e-01 6.15028262e-01 -2.83307433e-01
-1.59753084e-01 -1.20622918e-01 2.39546955e-01 1.08722314e-01
4.47466761e-01 9.82999131e-02 -3.47517133e-01 -6.38872683e-01
-1.89665869e-01 1.62514731e-01 1.52136711e-02 -2.80102938e-01
1.16235638e+00 -5.51622450e-01 -1.99594632e-01 1.97614372e-01
1.10482037e+00 1.23942956e-01 -9.97916996e-01 -1.19432390e-01
-9.50845703e-03 -1.20483708e+00 -1.25901299e-02 -1.16829932e+00
-9.47222173e-01 3.58630151e-01 6.86783075e-01 7.27469862e-01
1.10609543e+00 2.05598280e-01 4.46930915e-01 3.37088257e-01
1.92839801e-01 -8.39664936e-01 -1.00076281e-01 -2.71804295e-02
1.27845323e+00 -1.42214370e+00 8.00277069e-02 -2.18087465e-01
-5.08642018e-01 5.05476058e-01 3.73541027e-01 -1.22839987e-01
4.34783936e-01 -5.54837942e-01 8.93292055e-02 -6.16520166e-01
-7.55899489e-01 -3.98478389e-01 1.42665833e-01 5.07309377e-01
3.57607603e-01 -2.11491226e-03 -6.04651690e-01 1.57325909e-01
-3.71420145e-01 1.29137382e-01 5.76711297e-01 5.90913713e-01
-4.29082453e-01 -6.43938959e-01 -5.84156096e-01 6.77858829e-01
-4.29562390e-01 -3.90382081e-01 -1.01170897e+00 1.09263897e+00
3.37505668e-01 9.71171319e-01 1.40984684e-01 -5.40397763e-01
1.59470513e-01 -2.40447238e-01 3.98728967e-01 -2.98374504e-01
-5.03800273e-01 -3.14444721e-01 1.75774068e-01 -6.81495786e-01
-5.62089801e-01 -4.77885187e-01 -3.89446527e-01 -5.29580057e-01
-2.12448686e-01 1.21118568e-01 7.00010478e-01 5.22041023e-01
6.80447936e-01 -1.62957922e-01 5.15696645e-01 -8.21337283e-01
-3.21851879e-01 -7.25258112e-01 -3.53684515e-01 1.10452890e+00
2.69395232e-01 -4.12237227e-01 -6.98582470e-01 1.10542476e-01]
|
[10.82689094543457, 1.5555078983306885]
|
33cbce65-d2e0-4d6e-8d91-d7a45722a09e
|
self-supervised-anomaly-detection-of-rogue
|
2305.05495
| null |
https://arxiv.org/abs/2305.05495v1
|
https://arxiv.org/pdf/2305.05495v1.pdf
|
Self-Supervised Anomaly Detection of Rogue Soil Moisture Sensors
|
IoT data is a central element in the successful digital transformation of agriculture. However, IoT data comes with its own set of challenges. E.g., the risk of data contamination due to rogue sensors. A sensor is considered rogue when it provides incorrect measurements over time. To ensure correct analytical results, an essential preprocessing step when working with IoT data is the detection of such rogue sensors. Existing methods assume that well-behaving sensors are known or that a large majority of the sensors is well-behaving. However, real-world data is often completely unlabeled and voluminous, calling for self-supervised methods that can detect rogue sensors without prior information. We present a self-supervised anomalous sensor detector based on a neural network with a contrastive loss, followed by DBSCAN. A core contribution of our paper is the use of Dynamic Time Warping in the negative sampling for the triplet loss. This novelty makes the use of triplet networks feasible for anomalous sensor detection. Our method shows promising results on a challenging dataset of soil moisture sensors deployed in multiple pear orchards.
|
['Estefanía Serral Asensio', 'Jan Diels', 'Bart Baesens', 'Boje Deforce']
|
2023-05-09
| null | null | null | null |
['self-supervised-anomaly-detection', 'supervised-anomaly-detection', 'dynamic-time-warping']
|
['computer-vision', 'computer-vision', 'time-series']
|
[ 6.01322651e-01 -1.92604974e-01 6.68242865e-04 -2.22974867e-01
-2.69906729e-01 -7.91081727e-01 3.36397648e-01 6.04738116e-01
-3.49600054e-02 5.09230137e-01 -3.57099503e-01 -2.56696671e-01
-1.12120986e-01 -1.00339961e+00 -9.37206626e-01 -1.13445055e+00
-4.17024165e-01 2.43406698e-01 3.03023577e-01 -2.73112476e-01
8.41347501e-03 8.14464748e-01 -1.67601931e+00 6.13682857e-03
8.42377245e-01 1.40635908e+00 -1.04683600e-01 6.04282975e-01
1.18930541e-01 2.27514997e-01 -7.11417377e-01 -4.81141582e-02
4.82609004e-01 -1.23293765e-01 -9.60622802e-02 -1.70106143e-01
-6.09529242e-02 -2.95818090e-01 2.89248288e-01 1.41279411e+00
5.03156006e-01 -3.45195502e-01 4.29491192e-01 -1.71908402e+00
1.13505041e-02 6.50647819e-01 -5.91120243e-01 1.95551917e-01
1.02261931e-01 1.22858532e-01 6.18695080e-01 -2.49493092e-01
1.64705619e-01 8.31693769e-01 9.67453301e-01 -4.91441116e-02
-1.39330411e+00 -7.78555036e-01 -1.56475808e-02 6.38183281e-02
-1.21215844e+00 -4.47241277e-01 9.12457705e-01 -3.45364392e-01
5.23107827e-01 2.60102302e-01 6.69002175e-01 1.43637097e+00
5.03391862e-01 3.59596670e-01 7.29526818e-01 -8.36458877e-02
6.58250153e-01 -2.44042724e-01 -7.32228532e-02 -8.83107632e-02
7.41554677e-01 2.15091884e-01 -1.65676400e-01 -6.15351319e-01
1.34397715e-01 4.64402437e-01 9.91537273e-02 -6.46533787e-01
-1.02024353e+00 6.15390420e-01 1.87973469e-01 4.19151783e-01
-7.72660196e-01 -1.16456121e-01 5.74844480e-01 4.70733047e-01
5.47859788e-01 3.54584426e-01 -6.04740739e-01 1.89230661e-04
-8.09949636e-01 -2.05444664e-01 6.84980929e-01 6.22130036e-01
4.20875639e-01 1.00083888e-01 3.35756898e-01 2.98772991e-01
1.54156283e-01 1.15829790e+00 3.80409718e-01 -5.77231467e-01
2.14961514e-01 7.27709830e-01 3.98878753e-01 -1.38052809e+00
-6.17403924e-01 -2.09796414e-01 -1.13471150e+00 7.23985434e-02
3.72480631e-01 -1.29270628e-01 -6.51917696e-01 1.55347419e+00
6.54638231e-01 2.25437090e-01 3.20775956e-01 3.75615060e-01
1.82600860e-02 5.41022539e-01 -1.92949772e-02 -4.57802653e-01
9.29780304e-01 1.18903056e-01 -1.01219201e+00 -5.82026877e-02
4.91802156e-01 -4.31855291e-01 6.44141853e-01 6.76390469e-01
-4.63834256e-01 -9.82435048e-02 -1.33971143e+00 7.61698782e-01
-8.43854725e-01 -3.99623096e-01 4.51876193e-01 8.47160161e-01
-3.70179236e-01 8.10707986e-01 -1.19832706e+00 -6.77196681e-01
6.05742872e-01 3.42682481e-01 -3.36317152e-01 1.30211422e-02
-9.48511839e-01 6.47614121e-01 4.88665640e-01 3.96251917e-01
-6.38221681e-01 -3.58298272e-01 -6.99735165e-01 -2.85353124e-01
4.81889278e-01 2.77716964e-01 7.29285479e-01 -6.84832931e-01
-1.08164561e+00 4.89113510e-01 2.78476417e-01 -7.77445734e-01
7.63233244e-01 -2.63773948e-01 -1.05076659e+00 2.92725265e-02
8.05374831e-02 -1.33666471e-01 1.30758858e+00 -1.07725644e+00
-5.47704220e-01 -8.25143456e-01 -7.76607156e-01 -5.71129501e-01
-5.75384855e-01 -2.87343621e-01 5.20484090e-01 -5.75053155e-01
5.94046175e-01 -9.94765043e-01 5.88431694e-02 -1.70670018e-01
-5.38540423e-01 2.10468516e-01 1.13357508e+00 -5.51990449e-01
1.00022125e+00 -2.32446361e+00 -5.69952369e-01 4.45801497e-01
4.87878360e-02 4.47053730e-01 -8.97506624e-02 5.96278131e-01
-3.63531351e-01 -1.40231371e-01 -5.78812778e-01 2.93251455e-01
-9.45218951e-02 2.53952593e-01 -7.51856983e-01 9.88921404e-01
1.42227024e-01 5.06205976e-01 -1.04206276e+00 7.53757283e-02
2.61217922e-01 2.72544563e-01 1.60398558e-01 1.41974883e-02
-2.22072318e-01 5.27839839e-01 -3.97346973e-01 9.83939111e-01
1.06260264e+00 7.07689524e-02 -9.96577740e-03 -3.68554354e-01
-2.60076046e-01 -3.64973933e-01 -1.30514860e+00 1.19683814e+00
-3.76895741e-02 1.95361286e-01 -4.69328798e-02 -1.29103112e+00
1.11716461e+00 3.22962105e-01 1.03010488e+00 -6.02334738e-01
2.79260844e-01 5.44793725e-01 -2.38140553e-01 -5.40545344e-01
1.78790942e-01 2.85848141e-01 -1.22977525e-01 3.95852208e-01
-4.34662759e-01 -2.44237017e-02 -7.76746944e-02 -2.13176027e-01
1.56225109e+00 4.03770879e-02 1.78326488e-01 -1.06910326e-01
1.32948026e-01 4.15567122e-02 8.23746562e-01 6.56706989e-01
-5.03603935e-01 3.66840005e-01 4.97981548e-01 -5.92860162e-01
-1.00314486e+00 -1.01528633e+00 -2.70358682e-01 6.85132742e-01
3.70250791e-02 8.64503160e-02 -2.38976911e-01 -5.49778461e-01
5.04676461e-01 7.27066755e-01 -4.70528483e-01 -4.91711318e-01
-2.98098177e-01 -1.09885693e+00 7.23730743e-01 3.15358669e-01
5.15587687e-01 -9.66565311e-01 -1.15612698e+00 4.34235722e-01
-1.47222966e-01 -1.19650936e+00 1.99385434e-01 7.42639422e-01
-1.07681847e+00 -1.32308817e+00 -2.78458208e-01 -2.40988787e-02
5.99648058e-01 3.73727113e-01 7.25635052e-01 -4.53070343e-01
-4.22743142e-01 1.38046563e-01 -4.16435391e-01 -8.54980886e-01
-6.03161573e-01 -1.92550376e-01 4.79798704e-01 2.71092892e-01
8.07756364e-01 -8.55114341e-01 -3.70281368e-01 4.43211168e-01
-1.16381776e+00 -8.54778767e-01 3.83758336e-01 4.91620302e-01
7.24335074e-01 5.05483687e-01 5.44973016e-01 -5.43378770e-01
2.44720474e-01 -8.06531191e-01 -1.06801188e+00 2.98964798e-01
-7.42428601e-01 -6.75259382e-02 6.64808512e-01 -6.19146287e-01
-3.47627252e-01 4.98207450e-01 3.19826782e-01 -2.69916415e-01
-3.14381480e-01 7.97237307e-02 -4.04753029e-01 1.69771403e-01
7.17416823e-01 -5.09396642e-02 9.80351642e-02 -4.08687115e-01
4.54700626e-02 8.17643344e-01 6.48690581e-01 -2.62313839e-02
1.02306056e+00 8.80095303e-01 2.63117462e-01 -1.24192584e+00
-4.41520005e-01 -4.68743682e-01 -5.93955994e-01 -2.38657549e-01
5.13198555e-01 -8.37825000e-01 -8.60599220e-01 9.00823116e-01
-9.19219673e-01 -1.85507573e-02 -5.41222036e-01 4.28557336e-01
-1.44951418e-01 3.00974160e-01 5.79043925e-02 -1.15205491e+00
-3.68328571e-01 -7.86292672e-01 1.02886260e+00 -1.72656909e-01
-1.28236175e-01 -6.00345969e-01 1.71679852e-03 7.55849034e-02
4.58494902e-01 1.03867614e+00 3.94294471e-01 -9.75837827e-01
-1.33478567e-01 -7.92578280e-01 1.51347861e-01 4.95669633e-01
6.12283647e-01 1.61701351e-01 -1.25068498e+00 -4.85701203e-01
4.05136257e-01 -1.48586616e-01 5.29010355e-01 4.01859075e-01
7.73346364e-01 -1.30441129e-01 -3.77061009e-01 5.52731693e-01
1.23378515e+00 2.93450266e-01 3.79443139e-01 2.30680659e-01
4.95521396e-01 5.23017287e-01 6.97082102e-01 6.58901274e-01
-4.20686118e-02 2.17461362e-01 1.08711720e+00 1.78637862e-01
8.12715590e-01 -6.35225549e-02 5.41962087e-01 5.14817536e-01
4.07551527e-01 -3.94167751e-01 -9.05377746e-01 6.22652590e-01
-1.94683731e+00 -8.82660449e-01 -2.36370087e-01 2.52235126e+00
2.84278661e-01 1.30734146e-01 1.24864124e-01 8.44331086e-01
9.92892027e-01 4.59303223e-02 -1.34198797e+00 -7.48698786e-02
-3.38979810e-01 -1.60765901e-01 1.12043154e+00 -1.62207097e-01
-1.39954829e+00 4.04609114e-01 5.44745111e+00 8.14981833e-02
-1.14990819e+00 -1.47655413e-01 1.90018132e-01 -1.84475090e-02
8.50578994e-02 -3.20695370e-01 -4.90812004e-01 7.81695902e-01
1.06463504e+00 1.81595922e-01 2.09240139e-01 7.91870236e-01
2.45948255e-01 -3.30762804e-01 -7.82184243e-01 8.81339729e-01
-6.16638316e-03 -5.48674524e-01 -4.55697209e-01 1.80255085e-01
3.27343851e-01 3.94164622e-01 -8.23135301e-02 -3.97246033e-01
5.33921719e-01 -4.93753493e-01 4.54143822e-01 3.98743063e-01
4.10749018e-01 -7.36214638e-01 7.92197049e-01 3.12824249e-01
-1.03735948e+00 -3.46586734e-01 -2.96349138e-01 1.04309030e-01
-8.44928920e-02 1.32589078e+00 -6.07548892e-01 4.25093800e-01
9.69486713e-01 9.31584299e-01 -3.54052454e-01 9.76711333e-01
-6.89863786e-02 5.57042360e-01 -1.10914934e+00 1.78288072e-01
-2.26629436e-01 -2.54263133e-01 7.16144979e-01 4.53441471e-01
6.97582722e-01 -2.30063111e-01 1.11094542e-01 5.98876178e-01
2.63659418e-01 -1.24455504e-01 -1.13626671e+00 -3.39252532e-01
6.91824198e-01 9.87871647e-01 -9.80211079e-01 3.21860760e-01
-1.75437815e-02 8.62402737e-01 -3.68455499e-01 1.60449535e-01
-5.01339734e-01 -4.45113927e-01 5.47160983e-01 3.97725999e-02
5.28630912e-01 -2.35556304e-01 -1.78457618e-01 -9.14362550e-01
4.09015864e-01 -7.62718797e-01 5.25868118e-01 -4.23606306e-01
-1.54831576e+00 1.32214442e-01 -2.84300238e-01 -1.61412883e+00
-2.10650638e-01 -3.77301365e-01 -4.14671957e-01 3.18343282e-01
-1.36813152e+00 -9.90187347e-01 -5.71841657e-01 5.78317821e-01
-2.22242385e-01 -1.09646223e-01 9.37757373e-01 1.44087657e-01
-6.45150423e-01 2.27581829e-01 4.89542305e-01 -9.31340158e-02
7.11386502e-01 -8.95533085e-01 4.91921365e-01 1.21660638e+00
-2.44966254e-01 8.83029178e-02 8.91429543e-01 -1.05474591e+00
-1.67959774e+00 -1.24070644e+00 5.76000988e-01 -3.23607147e-01
9.37682271e-01 -5.00708938e-01 -9.61846113e-01 6.02494717e-01
-5.22374928e-01 4.70853388e-01 5.93938529e-01 -3.47926676e-01
-2.30417252e-01 -6.36058748e-01 -1.71649146e+00 1.90002933e-01
7.64803171e-01 -2.26294413e-01 -3.41347009e-01 4.06040251e-01
4.18296129e-01 -3.38963754e-02 -9.41066742e-01 6.12877369e-01
5.87344348e-01 -8.89735818e-01 8.41265440e-01 -2.61192143e-01
-2.76791692e-01 -4.63734984e-01 -3.94057870e-01 -1.24541187e+00
3.08261663e-01 -8.35937798e-01 -2.27017418e-01 1.35897720e+00
-8.84579048e-02 -9.72810924e-01 5.26509821e-01 3.09856266e-01
4.34781730e-01 2.13079870e-01 -9.92503703e-01 -1.18390560e+00
-2.29443565e-01 -7.69917727e-01 1.01512921e+00 1.15426886e+00
-1.42964900e-01 -1.89435288e-01 -3.21446359e-01 6.94596767e-01
1.00755417e+00 -1.71783566e-01 6.30604625e-01 -1.95503569e+00
3.62415522e-01 1.26471028e-01 -6.45891666e-01 -1.45748585e-01
-2.07551241e-01 -4.33648407e-01 2.28268474e-01 -9.18698549e-01
-4.90667284e-01 -4.15932328e-01 -2.70380169e-01 5.58607221e-01
3.23009223e-01 2.15623364e-01 -3.05947602e-01 2.24562943e-01
-1.59178495e-01 4.29759622e-01 4.42623764e-01 -1.54010385e-01
-5.39569139e-01 3.78714323e-01 -1.96022034e-01 7.26588726e-01
1.03268242e+00 -8.06496084e-01 -2.68833876e-01 -1.51041061e-01
6.12537444e-01 -2.59568751e-01 3.94626319e-01 -1.31698608e+00
9.87600088e-02 -3.48404758e-02 3.22628230e-01 -9.53895628e-01
-2.81447824e-03 -1.62087274e+00 3.54273081e-01 8.21606576e-01
1.53867170e-01 3.37646127e-01 1.63155317e-01 8.48779321e-01
1.24606863e-01 8.78510699e-02 7.07685709e-01 1.28886849e-01
-3.27657729e-01 2.22115859e-01 -3.06226492e-01 -1.61697909e-01
1.23143959e+00 -1.84767500e-01 -1.87704638e-01 -3.71920206e-02
-5.25075376e-01 2.62877524e-01 5.30216634e-01 6.31376207e-01
3.38038146e-01 -1.12728035e+00 -5.14695525e-01 7.13802516e-01
5.12779415e-01 -5.67587540e-02 -1.50793180e-01 7.14280844e-01
-1.66715533e-01 -9.84028056e-02 -2.71704465e-01 -1.04526210e+00
-1.05885780e+00 7.47077227e-01 1.14779115e-01 1.73128948e-01
-6.84654713e-01 1.27182320e-01 -7.02137887e-01 -3.88715893e-01
3.85561049e-01 -6.45686805e-01 -3.34331617e-02 6.33633137e-01
5.67000568e-01 5.82909882e-01 4.95528191e-01 -5.57797015e-01
-5.63545167e-01 3.44722807e-01 2.71135300e-01 1.78596139e-01
1.67040944e+00 -1.01590812e-01 -2.37433881e-01 9.12330151e-01
9.13634241e-01 -3.05663794e-01 -1.12240994e+00 -1.36138931e-01
2.22144082e-01 -2.96312749e-01 4.58208099e-02 -5.53704441e-01
-1.07938457e+00 6.74975157e-01 1.32061064e+00 8.35856915e-01
1.18236911e+00 -5.19581676e-01 8.02274942e-01 7.03756452e-01
5.20068228e-01 -1.30752540e+00 -7.19946399e-02 2.99088418e-01
4.66111839e-01 -1.54139185e+00 -1.17236916e-02 -6.45736530e-02
-2.35754043e-01 9.20832276e-01 -1.43473595e-02 1.28030270e-01
9.38388228e-01 4.13325489e-01 1.69286311e-01 -1.26269117e-01
-1.09316833e-01 -1.05385363e-01 -5.35725534e-01 9.32315111e-01
-3.00094664e-01 2.47348860e-01 1.37546465e-01 1.66750267e-01
-7.57575184e-02 1.19306758e-01 4.68422085e-01 1.48604083e+00
-3.34241211e-01 -7.80411363e-01 -7.23196208e-01 5.25670230e-01
-4.97630924e-01 1.98680431e-01 -2.28317574e-01 3.65552396e-01
3.52299139e-02 1.29822731e+00 1.80323377e-01 -4.10232276e-01
5.58765531e-01 3.22708935e-01 -1.12146705e-01 -5.24189360e-02
-3.16633254e-01 -1.63721815e-01 -4.14235145e-01 -8.91669869e-01
-4.79706734e-01 -9.88944530e-01 -1.07654345e+00 -3.81594956e-01
-4.87277985e-01 -1.25873536e-01 1.14785159e+00 8.71359527e-01
4.75404769e-01 3.68234277e-01 1.08963919e+00 -5.27404368e-01
-8.89869392e-01 -7.00869739e-01 -9.43063378e-01 4.56311882e-01
8.79882395e-01 -5.84481239e-01 -6.83843911e-01 -3.41602005e-02]
|
[7.3027448654174805, 2.6799449920654297]
|
0d0feb87-84f9-41ef-a5dd-5d95c650ef05
|
self-attention-comparison-module-for-boosting
|
2012.11357
| null |
https://arxiv.org/abs/2012.11357v1
|
https://arxiv.org/pdf/2012.11357v1.pdf
|
Self-attention Comparison Module for Boosting Performance on Retrieval-based Open-Domain Dialog Systems
|
Since the pre-trained language models are widely used, retrieval-based open-domain dialog systems, have attracted considerable attention from researchers recently. Most of the previous works select a suitable response only according to the matching degree between the query and each individual candidate response. Although good performance has been achieved, these recent works ignore the comparison among the candidate responses, which could provide rich information for selecting the most appropriate response. Intuitively, better decisions could be made when the models can get access to the comparison information among all the candidate responses. In order to leverage the comparison information among the candidate responses, in this paper, we propose a novel and plug-in Self-attention Comparison Module for retrieval-based open-domain dialog systems, called SCM. Extensive experiment results demonstrate that our proposed self-attention comparison module effectively boosts the performance of the existing retrieval-based open-domain dialog systems. Besides, we have publicly released our source codes for future research.
|
['Heyan Huang', 'Wei Wei', 'Zhipeng Zhao', 'Xian-Ling Mao', 'Tian Lan']
|
2020-12-21
| null | null | null | null |
['open-domain-dialog']
|
['natural-language-processing']
|
[-2.43377760e-01 -1.19085193e-01 -3.97227794e-01 -5.52983880e-01
-9.69897270e-01 -4.93868917e-01 6.29783809e-01 1.31119668e-01
-4.73218292e-01 6.09258831e-01 4.19677585e-01 -1.38502136e-01
-1.11060016e-01 -6.62526548e-01 1.51270637e-02 -3.07264507e-01
5.77433586e-01 7.71449864e-01 6.39355004e-01 -6.16612256e-01
3.35624784e-01 -1.33879557e-01 -1.02980268e+00 1.65529847e-01
1.21693325e+00 9.53923047e-01 6.38208091e-01 1.93378076e-01
-6.26839697e-01 6.13271058e-01 -4.27564710e-01 -3.67905617e-01
-5.85505925e-03 -4.55856234e-01 -1.09561300e+00 -2.28095040e-01
-4.62039374e-02 -6.88532889e-01 -4.22049910e-01 9.35451031e-01
6.14674628e-01 5.56602716e-01 5.51653504e-01 -9.97352421e-01
-1.03805065e+00 6.79604352e-01 -1.67226240e-01 6.79412112e-02
6.47932768e-01 2.29971945e-01 1.32696462e+00 -9.31049824e-01
4.33186173e-01 1.51173079e+00 -2.07490921e-01 8.39101136e-01
-9.00125325e-01 -7.61790931e-01 3.48287106e-01 2.43773863e-01
-1.23827136e+00 -4.33285594e-01 8.61928880e-01 -9.80425403e-02
5.74410617e-01 1.35488033e-01 4.35380936e-02 9.12560165e-01
-2.27617458e-01 9.44174588e-01 9.11710501e-01 -2.59849429e-01
1.16513677e-01 4.77378756e-01 6.79637134e-01 3.11175436e-01
-2.51925915e-01 -2.93243378e-01 -3.78576249e-01 -4.42924708e-01
6.11701190e-01 2.73331404e-01 -4.17065710e-01 -1.91855934e-02
-9.40460145e-01 1.10583484e+00 7.12574482e-01 4.03056502e-01
-3.67787987e-01 -6.11103714e-01 2.34047368e-01 3.40778679e-01
5.25407374e-01 3.99488777e-01 -3.79976630e-01 5.24481572e-02
-4.12243009e-01 1.78289458e-01 8.29037428e-01 9.69098508e-01
9.24276114e-01 -4.66601640e-01 -5.14300585e-01 1.32419002e+00
6.85935795e-01 3.42577845e-01 7.95497179e-01 -7.10960388e-01
5.86650133e-01 1.10897720e+00 3.00068885e-01 -9.78281796e-01
-1.07529432e-01 9.43297371e-02 -5.65308809e-01 -5.37688076e-01
3.15938741e-01 -3.80975246e-01 -2.89012313e-01 1.68915558e+00
4.41686064e-01 -2.12713003e-01 3.18668932e-01 1.34879780e+00
1.21589363e+00 7.82297432e-01 8.59535933e-02 -1.13443680e-01
1.38361406e+00 -1.10095692e+00 -8.01883996e-01 -3.82226378e-01
3.52728724e-01 -9.66367424e-01 1.40547597e+00 -1.86575174e-01
-7.68369973e-01 -5.97250760e-01 -8.35103691e-01 -1.71674863e-01
-1.37896001e-01 3.92296650e-02 4.42511290e-01 2.02428535e-01
-9.38510835e-01 1.59860298e-01 -3.27937871e-01 -5.03747642e-01
-1.65963292e-01 4.58210796e-01 -1.38755394e-02 -1.51832148e-01
-1.71098340e+00 8.38970244e-01 5.44541143e-02 -2.20281575e-02
-5.22709310e-01 -1.42848760e-01 -6.24348342e-01 2.60222048e-01
7.07235754e-01 -4.27097946e-01 1.57704771e+00 -7.34219611e-01
-1.68821108e+00 6.21703565e-01 -4.21884418e-01 -2.28834152e-01
2.79399633e-01 -3.01179856e-01 -3.52135420e-01 1.17287226e-01
4.54272628e-02 7.63146639e-01 4.87980038e-01 -9.85097587e-01
-5.31292617e-01 -3.25433403e-01 4.97093678e-01 3.99259895e-01
-6.83562517e-01 2.43521348e-01 -7.39864051e-01 -2.77076274e-01
-5.77743798e-02 -9.07308161e-01 -3.09393644e-01 -1.90493211e-01
-2.54591346e-01 -8.95017922e-01 6.95282876e-01 -2.33025178e-01
1.36913300e+00 -1.95734644e+00 -5.25737042e-03 -2.57029146e-01
1.90864161e-01 4.55106616e-01 -2.49759242e-01 7.60387719e-01
5.28724492e-01 6.60592839e-02 4.93237861e-02 -1.16209410e-01
1.58487231e-01 1.50357531e-02 -4.96669888e-01 -2.16340218e-02
1.18462034e-01 6.44736171e-01 -7.57982731e-01 -7.05793142e-01
2.06423536e-01 1.50664791e-01 -4.50011611e-01 8.68373573e-01
-5.87145865e-01 5.51933587e-01 -1.19342816e+00 2.66044110e-01
6.81188583e-01 -5.84761322e-01 1.09420083e-01 1.04795061e-01
8.51154551e-02 6.11988366e-01 -8.39791000e-01 1.73135245e+00
-7.57892311e-01 2.37519339e-01 2.90842056e-01 -4.57608193e-01
1.26145005e+00 4.41661686e-01 2.03393638e-01 -8.87160778e-01
2.39117742e-01 2.01909840e-01 2.98289299e-01 -3.68489534e-01
7.65769422e-01 8.70876536e-02 -2.30830118e-01 8.27307880e-01
-1.01964474e-01 1.75216511e-01 -2.88318992e-02 6.27105713e-01
7.17967510e-01 -2.54467756e-01 2.76620656e-01 -1.36461809e-01
9.99424279e-01 -3.90094183e-02 3.02889824e-01 6.71263933e-01
-3.56405616e-01 4.23324496e-01 2.11909875e-01 -9.17251408e-02
-4.86446381e-01 -7.29883850e-01 -5.34549840e-02 1.49778020e+00
7.19960868e-01 -2.87161380e-01 -7.20997095e-01 -7.46334851e-01
-1.31081656e-01 3.98243755e-01 -1.15814172e-01 -3.05507332e-01
-4.02969837e-01 -1.71069786e-01 2.12733090e-01 3.71077061e-01
7.42701828e-01 -1.25507534e+00 -1.16092913e-01 2.83916682e-01
-4.67535377e-01 -8.86429310e-01 -8.78128231e-01 -1.42834872e-01
-7.55848706e-01 -8.89204502e-01 -1.03829432e+00 -9.47753131e-01
3.50258708e-01 8.37591410e-01 8.74298871e-01 4.36077118e-01
3.97583395e-01 3.41159701e-01 -6.85588658e-01 -2.61536986e-02
-2.98813075e-01 3.44377726e-01 2.40576696e-02 1.16403960e-01
8.52730751e-01 -6.93038329e-02 -7.26659298e-01 8.45796883e-01
-6.89730585e-01 -1.38751373e-01 5.43062866e-01 1.07523859e+00
2.01569259e-01 -6.01427019e-01 1.06549025e+00 -7.04269648e-01
1.19381940e+00 -7.39818871e-01 -4.90464538e-01 4.14456129e-01
-5.49532115e-01 3.93666953e-01 8.58261049e-01 -4.89091486e-01
-1.49301946e+00 -2.66739428e-01 -3.52038234e-01 -1.35281399e-01
-2.04377487e-01 6.30940855e-01 -3.66656780e-01 6.93499371e-02
3.18701446e-01 8.40041116e-02 -8.91192928e-02 -7.64888287e-01
3.46745014e-01 1.29734707e+00 8.86226520e-02 -7.20313787e-01
5.33783078e-01 6.77508786e-02 -8.79789531e-01 -5.49206913e-01
-8.22230995e-01 -9.57991421e-01 -4.08588797e-01 -9.00553390e-02
7.65215337e-01 -8.11906099e-01 -7.82761872e-01 2.34300330e-01
-1.45688117e+00 -9.20064449e-02 4.38278079e-01 3.65188032e-01
-2.60524396e-02 4.47508901e-01 -6.53073609e-01 -1.04897892e+00
-3.98671478e-01 -1.45839107e+00 1.03519189e+00 8.08109760e-01
-2.87764162e-01 -9.22647655e-01 8.65250900e-02 5.48761904e-01
6.24106407e-01 -8.49155188e-01 8.34184647e-01 -1.57988012e+00
-8.15662563e-01 -1.01320028e-01 -3.93100262e-01 -6.71085343e-02
3.08729172e-01 -4.92442757e-01 -9.25679862e-01 -9.19920206e-02
1.03353389e-01 -6.68129623e-01 8.01093876e-01 1.02799358e-02
6.19694054e-01 -2.68285424e-01 -3.83550674e-01 -1.74813986e-01
9.89320338e-01 3.95123631e-01 3.41725588e-01 2.13135958e-01
4.05874640e-01 7.60932922e-01 1.05575323e+00 5.17133176e-01
5.39291441e-01 8.36198628e-01 1.55055314e-01 5.73017411e-02
1.92505911e-01 -3.81772012e-01 2.54613787e-01 9.62012768e-01
5.37536323e-01 -5.09015501e-01 -5.36569953e-01 4.91564959e-01
-1.93624842e+00 -6.25913143e-01 1.73199400e-01 2.24306536e+00
8.97004187e-01 2.91279908e-02 2.45049391e-02 -4.56819206e-01
9.37639534e-01 4.16634709e-01 -7.70164192e-01 -1.75337076e-01
1.63120031e-01 9.12099108e-02 -1.14180595e-01 5.84377229e-01
-8.31176758e-01 1.09879529e+00 5.30453825e+00 8.99450064e-01
-9.56561029e-01 3.09557212e-03 5.10855496e-01 3.07828218e-01
-4.10717875e-01 2.73373514e-01 -1.01323354e+00 5.23849189e-01
7.70743608e-01 -4.79005426e-01 1.47169381e-01 9.33842838e-01
1.83710441e-01 -1.03786193e-01 -1.03358090e+00 6.26886010e-01
-4.20128405e-02 -7.69027889e-01 2.49365970e-01 -1.58063963e-01
3.52040857e-01 -2.58309513e-01 -3.30309942e-02 7.06223488e-01
5.95507741e-01 -6.43177271e-01 -2.80122217e-02 3.49487305e-01
2.77343363e-01 -5.82640231e-01 8.76598179e-01 6.80193603e-01
-1.27315843e+00 -4.06671455e-03 -5.90275347e-01 -6.77483082e-02
6.57600164e-02 -1.27052618e-02 -9.56852257e-01 6.62900984e-01
3.19670469e-01 3.96374941e-01 -4.30897951e-01 8.51286352e-01
-2.08550811e-01 5.79222918e-01 -1.78743914e-01 -7.08487749e-01
3.90705109e-01 -2.12088004e-01 3.48810703e-01 7.75191009e-01
3.81040238e-02 3.80944729e-01 6.86074257e-01 8.75953496e-01
-1.42874360e-01 6.39268100e-01 -2.87635922e-01 -4.49305810e-02
7.65509129e-01 1.21784985e+00 -3.47567648e-01 -3.73342842e-01
-7.00423717e-01 7.68241167e-01 3.42695236e-01 2.60137051e-01
-5.77514112e-01 -4.92552161e-01 5.27007282e-01 -1.24368362e-01
1.14177883e-01 -7.27665471e-03 6.32565096e-02 -1.26626635e+00
-2.56211665e-02 -8.69353592e-01 7.81881511e-01 -6.83913469e-01
-1.69253159e+00 7.64738262e-01 -1.83590963e-01 -1.29582238e+00
-4.03088808e-01 -1.64370283e-01 -7.83155978e-01 1.14942408e+00
-1.52781677e+00 -8.09863567e-01 -2.37041175e-01 5.07028401e-01
1.07068634e+00 -9.03042257e-02 8.62284660e-01 1.86959192e-01
-5.84495366e-01 6.17584765e-01 1.22959636e-01 3.31856817e-01
1.22898245e+00 -1.02730417e+00 2.15564370e-01 4.08983141e-01
-9.07145143e-02 1.26941240e+00 4.47801709e-01 -6.07021391e-01
-1.33573973e+00 -5.73746860e-01 9.48046029e-01 -3.31964605e-02
6.07169092e-01 -5.28842658e-02 -1.45150232e+00 3.60610634e-01
6.99635029e-01 -4.39960390e-01 6.93740189e-01 2.21809015e-01
-2.23027900e-01 -6.97319210e-02 -7.53365993e-01 7.61439025e-01
5.60771167e-01 -5.32033980e-01 -9.88267899e-01 2.30089426e-01
1.08638883e+00 -2.49409929e-01 -6.10319138e-01 1.96112350e-01
3.64358515e-01 -7.87824929e-01 8.34095657e-01 -4.97723848e-01
3.04435223e-01 -1.21988110e-01 6.41196361e-03 -1.26749420e+00
-3.02692000e-02 -5.54398477e-01 2.30798483e-01 1.58528268e+00
3.84369284e-01 -7.37452090e-01 4.62710410e-01 1.08915544e+00
-7.08763301e-03 -6.22020543e-01 -7.51077652e-01 -4.08138841e-01
1.30397156e-01 1.19450778e-01 7.32569695e-01 7.07318485e-01
4.03544277e-01 9.70151842e-01 -5.61118126e-01 -8.25513080e-02
-8.72951970e-02 7.39247143e-01 1.00946653e+00 -1.43521583e+00
-2.22695962e-01 -3.03232878e-01 1.93155184e-01 -2.18060064e+00
4.09817338e-01 -7.50131130e-01 2.34375611e-01 -1.58321822e+00
3.91987711e-01 -6.93155706e-01 -2.95758605e-01 3.08771312e-01
-7.78646410e-01 -4.01154488e-01 1.51042581e-01 5.52052438e-01
-1.05879676e+00 8.40233624e-01 1.31087267e+00 -1.64889976e-01
-4.40208435e-01 1.55196607e-01 -1.05281198e+00 4.44200605e-01
6.24082983e-01 -4.53213066e-01 -3.85428876e-01 -3.45098972e-01
-3.80511612e-01 5.84339559e-01 5.87265193e-03 -5.93736649e-01
6.59645021e-01 -4.24710125e-01 -8.10934678e-02 -6.26836002e-01
4.73643064e-01 -8.11170399e-01 -5.34961283e-01 1.80894539e-01
-1.00665498e+00 4.42823954e-02 -6.95116147e-02 7.62051523e-01
-4.07232791e-01 -5.28066456e-01 6.40585005e-01 -1.95352510e-01
-8.81370723e-01 4.52738613e-01 -3.16518456e-01 2.20935225e-01
5.46576202e-01 3.12041432e-01 -4.74941015e-01 -9.34137344e-01
-2.31574297e-01 8.82555306e-01 4.16832715e-01 7.57401288e-01
6.07062519e-01 -1.14002812e+00 -5.94019771e-01 -1.76677033e-01
3.16685408e-01 -9.97721404e-03 4.31283861e-01 3.82219315e-01
2.37342939e-01 8.05229485e-01 -1.18870167e-02 -3.82128268e-01
-1.30352759e+00 5.98377943e-01 -5.05718775e-02 -4.72638249e-01
-1.51733875e-01 6.59825027e-01 4.39820796e-01 -5.41963875e-01
2.88433105e-01 8.79982933e-02 -6.78872108e-01 -1.44979805e-01
6.96032763e-01 6.35616342e-03 -2.33600795e-01 -7.03094304e-01
-3.82337570e-01 3.23499292e-01 -4.74104315e-01 -3.50199342e-01
8.40894043e-01 -5.19939959e-01 7.45870396e-02 3.90018433e-01
1.18242800e+00 1.53247550e-01 -8.51817787e-01 -8.38187695e-01
3.93785872e-02 -5.40928364e-01 -1.94076076e-01 -6.71638310e-01
-8.67999554e-01 1.07423925e+00 2.59011030e-01 3.13446611e-01
1.00723779e+00 1.19850554e-01 1.24159765e+00 6.75633848e-01
4.01295274e-01 -9.29673254e-01 3.30042124e-01 7.53536701e-01
6.03259981e-01 -1.66707838e+00 -2.66582102e-01 -4.08495188e-01
-9.59048331e-01 9.28961992e-01 1.24471188e+00 4.34778333e-02
3.29088420e-01 -3.28772694e-01 5.10546744e-01 3.35541996e-03
-9.35377777e-01 -3.30081075e-01 3.53390932e-01 3.92835408e-01
7.52610207e-01 -3.12892288e-01 -7.37401187e-01 1.08797026e+00
1.90404013e-01 -3.44299585e-01 1.64944723e-01 7.79845238e-01
-7.99758136e-01 -1.49781311e+00 -2.70654410e-01 2.76767492e-01
-2.23144993e-01 -2.27533117e-01 -7.81022727e-01 4.23181266e-01
-6.43556893e-01 1.61023808e+00 -1.48462579e-01 -3.92843962e-01
2.68347621e-01 2.66307980e-01 -2.76748151e-01 -7.92619050e-01
-8.45323741e-01 1.74855947e-01 1.98118463e-02 -1.37623116e-01
-3.48821878e-01 -2.05458120e-01 -1.32950413e+00 -7.64153600e-02
-9.80880439e-01 6.52666986e-01 1.22217812e-01 9.75929916e-01
6.38594568e-01 1.04771696e-01 9.26625192e-01 -3.45354706e-01
-9.09716189e-01 -1.23832500e+00 -2.40250394e-01 4.06083673e-01
4.79155406e-03 -8.63317192e-01 -1.80034712e-01 -4.77278322e-01]
|
[12.559213638305664, 7.8484697341918945]
|
56db824b-57fd-410d-b2bf-491b67a882b3
|
learning-dynamic-hierarchical-models-for
|
1608.03474
| null |
http://arxiv.org/abs/1608.03474v1
|
http://arxiv.org/pdf/1608.03474v1.pdf
|
Learning Dynamic Hierarchical Models for Anytime Scene Labeling
|
With increasing demand for efficient image and video analysis, test-time cost
of scene parsing becomes critical for many large-scale or time-sensitive vision
applications. We propose a dynamic hierarchical model for anytime scene
labeling that allows us to achieve flexible trade-offs between efficiency and
accuracy in pixel-level prediction. In particular, our approach incorporates
the cost of feature computation and model inference, and optimizes the model
performance for any given test-time budget by learning a sequence of
image-adaptive hierarchical models. We formulate this anytime representation
learning as a Markov Decision Process with a discrete-continuous state-action
space. A high-quality policy of feature and model selection is learned based on
an approximate policy iteration method with action proposal mechanism. We
demonstrate the advantages of our dynamic non-myopic anytime scene parsing on
three semantic segmentation datasets, which achieves $90\%$ of the
state-of-the-art performances by using $15\%$ of their overall costs.
|
['Buyu Liu', 'Xuming He']
|
2016-08-11
| null | null | null | null |
['scene-labeling']
|
['computer-vision']
|
[ 5.86060882e-01 2.00536489e-01 -4.19536889e-01 -8.24457765e-01
-1.26794016e+00 -4.28762466e-01 2.60448270e-02 -7.41574019e-02
-5.92833579e-01 3.75895888e-01 -5.48289597e-01 -3.79021227e-01
-1.18991852e-01 -6.83682084e-01 -8.52295935e-01 -5.94103932e-01
3.44902575e-02 8.06204379e-01 7.01153159e-01 5.21095097e-01
4.12759483e-01 3.31982106e-01 -1.70238972e+00 2.71332413e-01
9.54314709e-01 1.41547120e+00 6.30302787e-01 1.14795673e+00
2.44582002e-03 6.38251364e-01 -2.61642933e-01 -4.71552312e-01
6.76950276e-01 -3.26630175e-01 -8.64940107e-01 7.80929029e-01
4.75090265e-01 -4.14096385e-01 1.47013448e-03 1.19328296e+00
3.44505280e-01 8.98203924e-02 2.65687138e-01 -1.30479050e+00
-8.56290832e-02 3.31933320e-01 -6.34563625e-01 1.32702544e-01
-1.45641416e-01 6.39810920e-01 9.58122075e-01 -3.24088126e-01
5.55935800e-01 1.22437179e+00 1.81878164e-01 5.47841728e-01
-1.31314194e+00 -3.97174001e-01 7.09825397e-01 4.39900696e-01
-1.09417772e+00 -3.38563353e-01 5.75860441e-01 -3.19279671e-01
1.10826421e+00 1.82331294e-01 8.59976590e-01 3.78384918e-01
3.04795355e-01 1.10917103e+00 1.25519645e+00 -3.61350983e-01
5.75989783e-01 -2.01601580e-01 6.08264878e-02 1.28834939e+00
-1.23627305e-01 -1.25403941e-01 -5.39289832e-01 -4.78461199e-02
8.67564976e-01 -1.42559394e-01 3.45539927e-01 -5.58950782e-01
-7.34612405e-01 7.04333723e-01 1.57821365e-02 -4.88123983e-01
-2.07979381e-01 5.06031454e-01 1.90025479e-01 1.92504793e-01
2.92756498e-01 3.34388316e-01 -8.97835493e-01 -3.31466436e-01
-1.08975983e+00 1.26425833e-01 5.47760487e-01 1.14779246e+00
8.00170004e-01 -7.43703470e-02 -3.49396288e-01 7.72557914e-01
3.24244738e-01 6.04511201e-01 2.20242605e-01 -1.81365418e+00
3.09830606e-01 6.33426845e-01 1.73545748e-01 -4.18627411e-01
-4.42689210e-01 -2.35260889e-01 -3.79840761e-01 1.95978135e-01
2.17691168e-01 1.84923317e-02 -1.45275724e+00 1.65404475e+00
7.70606279e-01 4.66683239e-01 -2.06786022e-01 6.56051099e-01
1.26046494e-01 7.05788136e-01 1.80906534e-01 -6.99883223e-01
1.25016987e+00 -1.22381306e+00 -3.05213422e-01 -4.68905360e-01
3.66764039e-01 -4.77056503e-01 1.17899799e+00 4.19683576e-01
-1.36308110e+00 -5.87995589e-01 -8.01151931e-01 -1.74648613e-01
1.82866544e-01 -1.31564170e-01 7.87978411e-01 6.82915449e-01
-1.03349495e+00 6.29834354e-01 -1.43400908e+00 -7.22045451e-02
5.00551701e-01 6.34522200e-01 1.88979834e-01 -1.95460528e-01
-3.50184888e-01 4.11371827e-01 1.88366830e-01 3.79095897e-02
-1.29042029e+00 -6.67405486e-01 -7.33484149e-01 2.81769242e-02
8.09322953e-01 -6.34500682e-01 1.67913270e+00 -7.94316053e-01
-1.70871067e+00 8.89500976e-01 -4.10300136e-01 -5.16048849e-01
6.56273067e-01 -2.30847195e-01 -4.14194353e-02 2.17358515e-01
4.25717145e-01 1.02514994e+00 9.53898549e-01 -9.73852515e-01
-1.18167150e+00 -3.90893728e-01 5.61391488e-02 3.67478877e-01
-3.47105041e-02 -9.65308771e-02 -1.16049755e+00 -1.98865190e-01
4.38347131e-01 -1.26064229e+00 -9.20178771e-01 2.23437503e-01
-3.73398274e-01 -9.14182439e-02 7.24923313e-01 -5.20611286e-01
1.08467841e+00 -1.88922298e+00 -1.41475484e-01 -4.23865393e-02
-1.77375734e-01 -1.50095969e-01 -2.52492040e-01 -1.32429913e-01
6.74594820e-01 5.35651706e-02 -4.49432075e-01 -4.30868179e-01
-1.33420825e-01 4.22189176e-01 -1.49738584e-02 3.08703959e-01
2.44589493e-01 8.24641407e-01 -7.38825977e-01 -1.04195333e+00
3.25080901e-01 -1.00546122e-01 -1.00252044e+00 3.24515224e-01
-7.34309137e-01 4.00146067e-01 -8.55820715e-01 9.68554080e-01
4.74730998e-01 -5.50129056e-01 3.46101612e-01 2.17185766e-02
-1.26080081e-01 -1.81771740e-01 -1.22513080e+00 1.84315956e+00
-2.49537870e-01 2.51130193e-01 5.36539555e-02 -9.72262740e-01
4.61461157e-01 -2.73049563e-01 7.07744837e-01 -8.62040401e-01
-7.21670464e-02 1.29229743e-02 -1.03009172e-01 -4.50708061e-01
4.57643509e-01 3.19580495e-01 -2.30214834e-01 1.12660728e-01
-2.46866513e-02 -6.37666583e-01 2.57446527e-01 -3.53181511e-02
1.36996388e+00 4.62558508e-01 1.85935482e-01 -2.81002015e-01
1.50079146e-01 2.83850044e-01 1.03151000e+00 1.09276152e+00
-4.75944489e-01 3.46863866e-01 5.29197097e-01 -4.56317216e-01
-9.49973106e-01 -9.51985180e-01 -2.90548820e-02 1.23679698e+00
3.90042990e-01 3.65909673e-02 -1.06064999e+00 -6.90695763e-01
-1.38695344e-01 6.74910903e-01 -3.15715283e-01 3.42603475e-02
-7.01553702e-01 -7.17057943e-01 -1.18289374e-01 4.51248050e-01
7.33088970e-01 -1.02087760e+00 -1.29523718e+00 3.62014085e-01
4.75896262e-02 -1.38057232e+00 -4.71646369e-01 3.70940745e-01
-1.15766978e+00 -9.85547543e-01 -1.57373115e-01 -6.83012247e-01
6.96063459e-01 1.62044391e-01 1.05539453e+00 -3.34882624e-02
-6.35502338e-01 4.69159991e-01 -1.34652033e-01 -2.88051546e-01
-6.22430025e-03 -1.23946384e-01 -2.62154669e-01 2.89106984e-02
-5.09094819e-02 -1.54013187e-01 -8.53750587e-01 3.21225733e-01
-5.79405725e-01 5.28154373e-01 4.82590944e-01 8.38357985e-01
1.47435629e+00 -3.09659112e-02 1.44435436e-01 -1.02132905e+00
-4.19204161e-02 1.46987170e-01 -1.32428682e+00 5.77525079e-01
-9.61966515e-01 2.37425417e-01 2.73299634e-01 -3.02312583e-01
-1.10339379e+00 6.60229027e-01 8.81694406e-02 -5.43952346e-01
4.59597856e-02 -4.32013609e-02 -2.14623615e-01 -2.10395873e-01
1.49932951e-01 2.73486137e-01 -3.68005872e-01 -1.52396336e-01
4.99175489e-01 2.32477829e-01 5.35991728e-01 -8.45169604e-01
3.88133764e-01 4.04489368e-01 2.29341879e-01 -4.74365383e-01
-1.06065524e+00 -3.73688579e-01 -5.95934629e-01 -3.25132281e-01
1.11675310e+00 -9.05800223e-01 -8.81973267e-01 4.88982469e-01
-8.72889280e-01 -8.71925771e-01 -5.45247614e-01 2.06347004e-01
-1.19067526e+00 2.11699650e-01 -6.02698922e-01 -9.63445604e-01
-1.57170668e-01 -1.62457848e+00 1.32534683e+00 4.42285597e-01
3.52764189e-01 -3.33072603e-01 -4.13902104e-01 6.61325216e-01
1.10571338e-02 1.12431534e-01 9.33263242e-01 -7.17393383e-02
-1.38152826e+00 9.83833447e-02 -3.21163893e-01 9.70426202e-02
-4.23185736e-01 -3.31353098e-02 -9.45110559e-01 -2.91814595e-01
-3.94392349e-02 -4.40679759e-01 8.36895347e-01 1.04395735e+00
1.78838074e+00 -2.81534314e-01 -3.47030073e-01 8.49495292e-01
1.61768770e+00 6.48725510e-01 2.86513031e-01 3.80709440e-01
7.11142600e-01 5.19713998e-01 1.25072718e+00 6.20651603e-01
5.69710255e-01 6.33602262e-01 5.64523578e-01 6.95753098e-02
2.11103614e-02 -2.64642596e-01 6.79317936e-02 4.47477371e-01
1.69475570e-01 -2.11579159e-01 -9.58506703e-01 6.40700698e-01
-2.20188856e+00 -5.62278807e-01 2.92159051e-01 2.23673940e+00
7.07418621e-01 4.83013839e-01 1.23999943e-03 -3.11230928e-01
6.93630874e-01 1.04492251e-02 -1.37486982e+00 -3.73451054e-01
2.72431910e-01 1.05580188e-01 9.23657298e-01 4.20960307e-01
-1.20022464e+00 1.44832563e+00 6.76880550e+00 8.84899735e-01
-9.78959441e-01 2.31173038e-01 1.31399584e+00 -3.42815638e-01
5.87956943e-02 1.74381077e-01 -9.33331609e-01 3.67821723e-01
1.08407867e+00 -1.39483452e-01 5.24190724e-01 1.29244661e+00
2.92451531e-01 -4.27118182e-01 -1.11782598e+00 1.02312911e+00
-2.91208476e-01 -1.39547789e+00 -2.00809628e-01 6.31375313e-02
8.82365644e-01 2.20802039e-01 -4.30495143e-02 1.80311546e-01
6.59089208e-01 -6.30478382e-01 9.49259460e-01 2.23983973e-01
8.15386355e-01 -6.97132051e-01 2.11578012e-01 3.50593477e-01
-1.23972607e+00 -5.31529844e-01 -3.24231237e-01 1.85081214e-01
4.29559678e-01 4.37535316e-01 -6.14657283e-01 2.53926758e-02
9.11419749e-01 2.84770370e-01 -4.93937671e-01 1.06737614e+00
-4.70110737e-02 6.28141403e-01 -2.81309783e-01 5.83673976e-02
2.32788444e-01 -1.69933438e-01 8.38392079e-02 1.06779885e+00
2.17516989e-01 4.44531441e-01 8.86184454e-01 3.96580726e-01
-1.34291286e-02 -1.50856867e-01 -1.65329859e-01 3.99093293e-02
5.68985820e-01 1.12662160e+00 -1.37966430e+00 -4.75425005e-01
-2.32983381e-01 1.02437794e+00 2.64777303e-01 8.56818557e-02
-1.01189983e+00 2.41073117e-01 5.90524971e-01 -5.87276630e-02
4.90963727e-01 -2.99697310e-01 -4.95186508e-01 -1.01974821e+00
6.85004145e-02 -5.62590718e-01 5.55012286e-01 -6.67675078e-01
-7.78536856e-01 3.52613062e-01 1.19239531e-01 -1.11069274e+00
-3.52280557e-01 -4.49901551e-01 -1.29057825e-01 2.97158092e-01
-1.54035115e+00 -1.13416529e+00 -9.12634656e-02 5.27136624e-01
1.28628504e+00 1.20128363e-01 5.05624413e-01 -6.71588704e-02
-4.92804438e-01 4.39005017e-01 -1.39861435e-01 -2.95494646e-01
1.26089752e-01 -1.16216576e+00 5.28759241e-01 1.05695689e+00
1.06599376e-01 -2.01952264e-01 6.10495448e-01 -5.60939133e-01
-1.64433408e+00 -1.20950007e+00 3.57923985e-01 -2.41110459e-01
4.31630611e-01 -2.89040059e-01 -4.31702554e-01 6.15579247e-01
-1.28313944e-01 3.78434837e-01 2.58881390e-01 -1.66492730e-01
-3.59107624e-03 -3.15247566e-01 -1.39351833e+00 5.82595110e-01
1.46632516e+00 -2.22119242e-01 1.09793156e-01 4.77325141e-01
1.15461934e+00 -6.45690441e-01 -7.32789695e-01 5.59214890e-01
5.29246747e-01 -8.17599058e-01 7.01955676e-01 -5.40662467e-01
1.50967389e-01 -1.62972152e-01 -4.14840043e-01 -6.26787722e-01
-4.89782184e-01 -5.84384620e-01 -2.00474232e-01 8.16114247e-01
5.28012574e-01 -2.86438376e-01 1.30607629e+00 1.15255415e+00
-2.81763822e-01 -1.00004387e+00 -1.21692228e+00 -6.15725517e-01
-4.42406893e-01 -6.46755457e-01 1.85289487e-01 4.06573027e-01
-7.50455678e-01 -8.42093676e-02 -2.35240623e-01 4.98859495e-01
9.72475231e-01 5.81363440e-01 5.80507457e-01 -9.13113296e-01
-6.19778097e-01 -1.67799816e-01 -4.38443184e-01 -1.26049817e+00
8.26542601e-02 -2.97164291e-01 4.25184697e-01 -1.39999807e+00
5.93090117e-01 -5.10093749e-01 -4.12788481e-01 6.07707739e-01
-8.04861188e-02 -1.65548339e-01 4.78166044e-01 8.78988653e-02
-1.24109137e+00 2.31702760e-01 1.33715320e+00 -9.62808728e-02
-2.49171615e-01 -3.60556543e-02 -4.48791951e-01 7.63244689e-01
4.65119243e-01 -6.12614632e-01 -7.44092762e-01 -6.48093700e-01
-1.76621169e-01 6.14391148e-01 7.75179192e-02 -1.01524448e+00
3.26166362e-01 -8.54505301e-01 2.40770161e-01 -6.39207304e-01
5.83074749e-01 -6.91520631e-01 -4.52971235e-02 7.51831293e-01
-4.47752833e-01 1.36035621e-01 -4.99096699e-02 8.01649988e-01
5.80952317e-02 -6.70042410e-02 1.12040627e+00 -4.10383075e-01
-1.29992187e+00 6.47691131e-01 -5.44096977e-02 1.15469560e-01
1.34056234e+00 -3.43135118e-01 -2.03888267e-02 7.37184882e-02
-6.90575063e-01 6.26929939e-01 5.03174305e-01 2.91246384e-01
5.56475937e-01 -7.66945541e-01 -2.90733367e-01 1.54835626e-01
-3.10341828e-02 3.28758359e-01 5.24271190e-01 4.56299514e-01
-4.72039491e-01 1.47532299e-01 -1.21361509e-01 -1.03429592e+00
-1.26378357e+00 3.78500909e-01 1.69643283e-01 -5.68090439e-01
-4.27443624e-01 1.07397449e+00 2.32497185e-01 -2.24223897e-01
1.54324248e-01 -3.28901768e-01 2.84956574e-01 -4.14313078e-01
2.15124115e-01 2.72673875e-01 -1.99514478e-01 -8.87545198e-02
-2.36074626e-01 6.56441748e-01 -6.83869496e-02 -4.11201179e-01
1.21707916e+00 -3.53199095e-01 1.82690069e-01 3.88393223e-01
7.23778188e-01 -6.26175404e-01 -2.16023421e+00 -3.29319164e-02
5.53149655e-02 -5.24543285e-01 2.39152327e-01 -9.53319907e-01
-1.09627533e+00 6.61352873e-01 1.03043711e+00 -3.86714116e-02
1.35564721e+00 1.46837845e-01 6.95010900e-01 4.36613411e-01
9.06443596e-01 -1.54145980e+00 1.20201245e-01 3.03416640e-01
2.96714872e-01 -1.47069204e+00 -2.50851177e-02 -5.55021763e-01
-7.79980838e-01 7.67215133e-01 9.85587597e-01 -1.13058887e-01
5.20497084e-01 3.87101978e-01 -5.11952862e-03 8.45115911e-03
-1.15474498e+00 -2.25935102e-01 3.53767686e-02 4.97670054e-01
-7.64051825e-02 3.60238761e-01 -1.69902533e-01 2.83371270e-01
1.30605310e-01 -6.15139194e-02 1.82187870e-01 1.05945218e+00
-7.55207181e-01 -9.99716878e-01 1.94217097e-02 7.69283712e-01
-4.30203229e-01 1.59997106e-01 1.61835238e-01 4.63836044e-01
1.19593881e-01 9.83861089e-01 1.63472742e-01 -2.19150454e-01
1.24274135e-01 -2.41249166e-02 5.72066903e-01 -6.85204744e-01
-1.39037430e-01 4.10032868e-01 8.45844150e-02 -1.28374088e+00
-4.06190276e-01 -9.45548952e-01 -1.45770550e+00 1.16754256e-01
-2.69202322e-01 -1.55828789e-01 8.83505821e-01 9.84182715e-01
6.10998690e-01 4.79823053e-01 7.59904325e-01 -8.64210248e-01
-5.61449826e-01 -3.86976928e-01 -4.91269439e-01 1.19284488e-01
4.38427739e-03 -4.97650415e-01 2.57114142e-01 3.17415327e-01]
|
[9.207330703735352, -0.05894557386636734]
|
14012e90-e36d-494e-9da0-98023678d1a8
|
pt-resnet-perspective-transformation-based
|
1910.13055
| null |
https://arxiv.org/abs/1910.13055v1
|
https://arxiv.org/pdf/1910.13055v1.pdf
|
PT-ResNet: Perspective Transformation-Based Residual Network for Semantic Road Image Segmentation
|
Semantic road region segmentation is a high-level task, which paves the way towards road scene understanding. This paper presents a residual network trained for semantic road segmentation. Firstly, we represent the projections of road disparities in the v-disparity map as a linear model, which can be estimated by optimizing the v-disparity map using dynamic programming. This linear model is then utilized to reduce the redundant information in the left and right road images. The right image is also transformed into the left perspective view, which greatly enhances the road surface similarity between the two images. Finally, the processed stereo images and their disparity maps are concatenated to create a set of 3D images, which are then utilized to train our neural network. The experimental results illustrate that our network achieves a maximum F1-measure of approximately 91.19% when analyzing the images from the KITTI road dataset.
|
['Yu-An Wang', 'Ming Liu', 'Weidong Zhang', 'Rui Fan', 'Peng Han', 'Lei Qiao', 'Ruiwen Yao', 'Ioannis Pitas']
|
2019-10-29
| null | null | null | null |
['road-scene-understanding', 'road-segementation']
|
['computer-vision', 'computer-vision']
|
[ 4.81828541e-01 2.33415306e-01 -1.78944558e-01 -7.47341812e-01
-3.91384542e-01 -2.77231455e-01 3.94056439e-01 -4.56871003e-01
-4.53973472e-01 2.15505809e-01 1.06327765e-01 -3.80770534e-01
2.01508135e-01 -1.10206783e+00 -7.69853354e-01 -5.11287391e-01
4.02692616e-01 1.09518953e-01 5.36396682e-01 -1.18060783e-01
5.23703516e-01 5.49878895e-01 -1.80527413e+00 4.61914651e-02
1.03829384e+00 9.09472287e-01 6.16152704e-01 4.76053387e-01
-4.84215170e-01 4.67223436e-01 -5.12416027e-02 -2.15774581e-01
3.50630999e-01 3.86355184e-02 -7.95957148e-01 1.37016147e-01
6.17571056e-01 -5.60765147e-01 -5.31544268e-01 1.05516827e+00
1.34021610e-01 1.96537569e-01 5.05948246e-01 -9.12000299e-01
-1.58760443e-01 1.72636092e-01 -6.83855653e-01 -2.25829072e-02
1.06851481e-01 -1.33067086e-01 7.09362507e-01 -9.81295228e-01
6.23373091e-01 1.44937277e+00 2.62972027e-01 1.24003053e-01
-8.71091127e-01 -5.63275278e-01 3.15966278e-01 1.69967592e-01
-1.27273762e+00 -1.70826897e-01 9.91133571e-01 -4.74905491e-01
7.04795063e-01 2.64638327e-02 7.74765611e-01 1.69291541e-01
8.78658593e-02 8.68428886e-01 1.07699811e+00 -1.51740447e-01
-2.29739308e-01 2.52552554e-02 3.20479244e-01 6.50023222e-01
5.19572012e-02 1.96029663e-01 -1.28807425e-01 7.23653376e-01
9.21492994e-01 4.74754646e-02 -1.92423359e-01 -4.06708062e-01
-1.00389576e+00 6.95703983e-01 8.30450952e-01 -4.39816061e-03
-4.50771570e-01 -1.18975580e-01 3.99733186e-02 -5.84548060e-03
4.37822014e-01 -6.30458221e-02 -2.23815680e-01 2.12493137e-01
-5.84977269e-01 4.11544517e-02 3.45827967e-01 7.38905013e-01
1.24783993e+00 9.82648283e-02 4.58076745e-01 1.12435710e+00
4.72185224e-01 7.33815491e-01 -6.13282137e-02 -1.25349808e+00
8.71499002e-01 9.10440505e-01 -2.33344138e-01 -1.28678644e+00
-4.74707216e-01 -9.75782797e-02 -5.84196091e-01 3.61775905e-01
4.39661443e-01 2.91909464e-02 -1.21675158e+00 1.35129201e+00
4.86203611e-01 1.76822275e-01 1.57879427e-01 9.65261936e-01
9.34724987e-01 8.75160456e-01 -1.68945968e-01 2.58950710e-01
1.11455929e+00 -1.13950789e+00 -3.82799327e-01 -6.59341633e-01
3.76657367e-01 -7.71409571e-01 9.26435292e-01 7.34888166e-02
-9.78911459e-01 -8.14719558e-01 -1.04398954e+00 -3.95607919e-01
-5.74056983e-01 2.48613149e-01 4.03452575e-01 3.35819125e-01
-1.03455377e+00 2.62665898e-01 -6.47447109e-01 -8.29608664e-02
5.24520278e-01 1.08429514e-01 -3.69967639e-01 -3.95543724e-01
-1.17837751e+00 9.48025942e-01 4.93404627e-01 4.59990501e-01
-6.20890260e-01 -5.00429690e-01 -1.20204139e+00 -7.98580796e-02
2.95453608e-01 -3.35501403e-01 8.13801169e-01 -5.29865146e-01
-1.40947819e+00 1.10950470e+00 -4.16060537e-01 -1.05427369e-01
4.96217489e-01 -1.91551909e-01 -3.32090616e-01 2.06589937e-01
3.56612951e-01 1.19160330e+00 7.36084342e-01 -1.54865515e+00
-1.13884366e+00 -4.94494975e-01 3.02792963e-04 7.08172441e-01
3.55912298e-01 -3.30865204e-01 -1.02683771e+00 -1.29214659e-01
8.45605731e-01 -9.26278055e-01 -3.85024428e-01 -2.51414329e-01
-5.69040895e-01 1.31865621e-01 9.31380630e-01 -9.11063552e-01
8.88864398e-01 -2.05690384e+00 -2.45972387e-02 5.28118253e-01
4.03659455e-02 1.29896998e-01 -1.97754562e-01 -9.56303701e-02
-2.59526759e-01 1.80370789e-02 -5.09462178e-01 5.53289466e-02
-5.30571222e-01 1.48851737e-01 -2.03316972e-01 1.85280681e-01
1.43696100e-01 8.44144225e-01 -6.91443503e-01 -4.29475278e-01
6.48522675e-01 4.92363930e-01 -2.10344061e-01 2.64196187e-01
2.26255208e-01 5.40916026e-01 -4.71460313e-01 2.89153457e-01
1.09964788e+00 2.60772049e-01 -1.30946308e-01 -2.26301268e-01
-4.18267220e-01 3.79708201e-01 -1.18855870e+00 1.30722284e+00
-5.77356398e-01 9.36855316e-01 -2.38225594e-01 -8.49880695e-01
1.24402523e+00 -3.11692625e-01 4.60131735e-01 -1.09507251e+00
1.27816141e-01 7.56097734e-02 -1.42776683e-01 -4.04047579e-01
7.10692883e-01 2.91928053e-01 3.97270843e-02 3.26925427e-01
-5.84419549e-01 -6.49741054e-01 2.63712823e-01 -8.47529098e-02
1.13119237e-01 2.31122673e-01 -2.18683422e-01 6.66348040e-02
8.78801525e-01 8.00357834e-02 3.98972481e-01 2.21189842e-01
1.38136297e-01 7.41999686e-01 4.56369877e-01 -7.32988775e-01
-1.14507222e+00 -1.21859109e+00 -1.76446036e-01 4.30242389e-01
8.23708832e-01 2.12637007e-01 -9.87130582e-01 -5.44972956e-01
-3.38582024e-02 6.87431812e-01 -3.69461060e-01 -7.93675631e-02
-8.06316614e-01 -3.85213196e-01 9.52629298e-02 6.22301877e-01
8.68388832e-01 -8.22865009e-01 -5.10229826e-01 7.99723528e-03
-5.30299306e-01 -1.31278789e+00 -2.82399565e-01 -2.48551980e-01
-9.62308347e-01 -1.16350961e+00 -6.06559634e-01 -1.09719539e+00
8.71163666e-01 9.64911938e-01 6.93900168e-01 -1.51981249e-01
-5.73858842e-02 -9.97838080e-02 2.10586727e-01 -2.53228903e-01
-1.07424960e-01 -1.29839376e-01 -3.24714959e-01 2.48818472e-02
3.61090004e-01 -4.06637847e-01 -6.26663148e-01 6.83191419e-01
-5.01769662e-01 6.14009202e-01 5.83725333e-01 5.77678502e-01
9.50592756e-01 1.59220278e-01 1.37486577e-01 -7.95737803e-01
1.71734154e-01 -1.52919620e-01 -9.62735057e-01 -2.71458626e-02
-3.66765916e-01 -1.85298070e-01 2.54787415e-01 4.34121713e-02
-1.42330694e+00 2.38747105e-01 -4.13976103e-01 -1.46443084e-01
-1.72287300e-01 2.96896368e-01 -3.42333168e-01 3.27244075e-03
2.16387987e-01 2.04723656e-01 1.26414850e-01 -2.00105116e-01
6.77938342e-01 7.74725139e-01 7.72511840e-01 -2.95116156e-02
8.03398788e-01 7.00143933e-01 -1.48377031e-01 -1.11730850e+00
-9.31762099e-01 -5.45574963e-01 -9.51138735e-01 -4.71532464e-01
1.25080442e+00 -1.00126243e+00 -4.76271182e-01 7.77819991e-01
-1.06803834e+00 -3.60315770e-01 -7.31480727e-03 5.75629115e-01
-5.79755723e-01 3.30927044e-01 -3.49581420e-01 -5.31913519e-01
-1.37166858e-01 -1.27129114e+00 9.27046776e-01 7.09833264e-01
1.29767314e-01 -8.23000550e-01 -2.12187454e-01 8.40188324e-01
6.41164705e-02 -4.90914546e-02 9.43384588e-01 -8.27820227e-02
-9.55671489e-01 -1.85105488e-01 -7.86188781e-01 4.86763269e-01
-3.18808444e-02 6.93151727e-02 -9.62843478e-01 3.30502063e-01
-1.45644799e-01 8.53898004e-02 1.03773963e+00 7.18530178e-01
1.22610438e+00 3.14379752e-01 -3.85552704e-01 9.19623137e-01
1.33431482e+00 4.48104292e-01 9.95309174e-01 5.03498912e-01
1.29284680e+00 1.15606570e+00 9.44075406e-01 -2.23573864e-01
8.11511993e-01 4.66221750e-01 5.05499184e-01 -5.20141959e-01
-2.27192596e-01 -5.48223555e-01 -1.29217982e-01 7.04860985e-01
-1.37813892e-02 4.65608537e-02 -1.04449701e+00 5.29454708e-01
-1.59091175e+00 -8.13542902e-01 -5.42365134e-01 2.13733816e+00
1.47766054e-01 2.07649797e-01 -2.26601258e-01 -6.71202391e-02
9.54887867e-01 3.27044100e-01 -6.05105042e-01 -4.72415507e-01
-9.95063260e-02 -1.72094002e-01 7.16534257e-01 8.02151144e-01
-1.11880684e+00 1.41986084e+00 6.40457678e+00 5.76525092e-01
-1.30019319e+00 -5.24958909e-01 7.55742431e-01 6.13794744e-01
-4.39421386e-01 9.30946097e-02 -7.57247865e-01 3.66339624e-01
7.20949829e-01 -1.74110383e-02 2.17554107e-01 9.49324787e-01
4.69143689e-01 -5.69283009e-01 -5.74975491e-01 9.68800068e-01
-9.10094008e-02 -1.06513798e+00 1.03764623e-01 2.20025763e-01
8.87805164e-01 3.62515867e-01 4.45320494e-02 5.14845587e-02
4.38482225e-01 -1.10028672e+00 4.29781884e-01 4.22252804e-01
8.24123561e-01 -9.62813199e-01 7.41601586e-01 3.78388494e-01
-1.36347461e+00 -1.40427919e-02 -4.66814846e-01 1.77901089e-01
5.60067832e-01 4.90632921e-01 -8.56047809e-01 5.58898687e-01
6.77546859e-01 1.00159943e+00 -3.95652592e-01 7.82362938e-01
-5.08036852e-01 1.48067459e-01 -3.09347183e-01 4.42143112e-01
3.35996777e-01 -9.19366777e-01 2.68700242e-01 6.98879004e-01
1.45534277e-01 -1.12526290e-01 1.09033331e-01 6.72349215e-01
-4.83210087e-02 6.73656985e-02 -8.29739988e-01 2.87015945e-01
3.92366171e-01 1.15558684e+00 -8.60573888e-01 -3.99255365e-01
-4.75808531e-01 7.14612663e-01 2.74569601e-01 4.87284720e-01
-7.11097479e-01 -4.28122640e-01 5.52954018e-01 -1.92885473e-02
2.87758440e-01 -8.32955986e-02 -7.10464776e-01 -9.35015678e-01
8.93005952e-02 -3.84343535e-01 -8.83301795e-02 -9.86714244e-01
-6.21930122e-01 5.05301535e-01 -3.70904580e-02 -1.23722422e+00
-1.10698417e-01 -5.30512631e-01 -7.28257775e-01 1.06253004e+00
-1.88095629e+00 -1.05540061e+00 -6.81471288e-01 2.82515317e-01
6.07065201e-01 6.59173578e-02 2.72355646e-01 2.63501644e-01
-6.73308551e-01 2.98355259e-02 5.45807257e-02 2.40403458e-01
2.17532694e-01 -9.50047076e-01 8.72040868e-01 1.06216872e+00
-2.32656181e-01 3.90192360e-01 1.34096727e-01 -6.90444589e-01
-7.97432780e-01 -1.23309433e+00 9.43738997e-01 -1.46818131e-01
1.39201880e-01 7.33808102e-03 -1.12100863e+00 4.11892921e-01
-3.22697401e-01 -2.71256804e-01 2.25662753e-01 -2.44911432e-01
-2.87042707e-01 -2.21940771e-01 -9.15267408e-01 8.30885470e-01
1.13960612e+00 -6.88703775e-01 -6.06751025e-01 -1.72219276e-01
6.15382195e-01 -5.81125081e-01 -7.04553366e-01 4.35886264e-01
7.32156098e-01 -1.05512488e+00 1.27076721e+00 -6.50016367e-02
7.06897616e-01 -4.95018601e-01 9.31533705e-03 -1.27875745e+00
1.18686043e-01 7.27082044e-02 6.43276095e-01 9.37307894e-01
4.76462305e-01 -7.20153868e-01 9.99779940e-01 4.72376227e-01
-3.34236413e-01 -8.17828298e-01 -6.39143169e-01 -3.49458784e-01
2.30696122e-03 -7.12470293e-01 5.14082491e-01 6.82145298e-01
-4.68116015e-01 4.12092447e-01 -1.16305195e-01 1.00515842e-01
6.04483843e-01 4.32702273e-01 1.05337477e+00 -1.14675272e+00
5.29006898e-01 -5.04739165e-01 -3.60885292e-01 -1.71937120e+00
2.51390427e-01 -7.96250343e-01 1.57125592e-01 -1.86279929e+00
-6.01263903e-02 -4.72313792e-01 1.75462663e-01 1.11281976e-01
-3.02316010e-01 2.43488327e-01 5.99415228e-03 2.30840325e-01
1.23923071e-01 5.19200504e-01 1.62373328e+00 -1.08320549e-01
-5.78335941e-01 2.44806781e-01 -6.35231137e-01 1.10851479e+00
9.35057878e-01 -1.62532642e-01 -5.89125514e-01 -4.69369143e-01
-1.66176274e-01 7.21617565e-02 7.48103485e-02 -7.89720297e-01
7.84398466e-02 -3.03275049e-01 3.59665602e-01 -1.36877906e+00
4.45432812e-01 -8.06332469e-01 -2.87736449e-02 3.18852663e-01
-2.29790404e-01 -1.91227794e-01 -1.77714620e-02 5.19492805e-01
-4.52799767e-01 5.39279841e-02 8.91126871e-01 -2.87638307e-02
-1.14738548e+00 4.35779691e-01 -1.63506508e-01 9.33288876e-03
1.06043732e+00 -8.37714553e-01 -1.92170709e-01 -4.18165505e-01
-5.02885938e-01 7.24402905e-01 5.67846835e-01 5.62979877e-01
1.07870281e+00 -1.04178226e+00 -5.51513851e-01 5.29644549e-01
1.05163150e-01 6.53321147e-01 5.78853428e-01 6.36877298e-01
-8.28412712e-01 3.89641196e-01 -4.41758424e-01 -9.05900896e-01
-1.25935268e+00 2.67955333e-01 3.96601468e-01 5.51243909e-02
-7.55250752e-01 5.87184012e-01 5.39188266e-01 -7.89042592e-01
-1.00934036e-01 -2.83162117e-01 -6.68788910e-01 1.12743229e-02
4.11909699e-01 5.39496779e-01 -2.05096036e-01 -1.19154882e+00
-2.62215793e-01 1.45176482e+00 -5.61084040e-02 -3.55428368e-01
1.26678562e+00 -5.59957206e-01 -2.98742726e-02 9.89656076e-02
1.54720962e+00 -8.92940164e-02 -1.51882553e+00 -2.09226862e-01
-1.77550077e-01 -7.28100657e-01 1.08168572e-01 -2.28331119e-01
-1.32325745e+00 1.27840340e+00 6.20114207e-01 -2.47901574e-01
9.69101131e-01 -1.03796497e-01 9.44422722e-01 3.22419614e-01
1.27564952e-01 -1.21831024e+00 -2.53319144e-01 8.17084253e-01
5.58873713e-01 -1.35530043e+00 -3.43825221e-02 -9.57126379e-01
-9.23420310e-01 1.19456458e+00 6.46453440e-01 -1.32763863e-01
4.48577106e-01 3.51490900e-02 4.47325081e-01 -1.66889429e-01
-7.19838291e-02 -5.86079299e-01 3.68741840e-01 7.91559339e-01
-3.18555720e-02 -1.17832553e-02 -8.17044228e-02 -1.50027096e-01
-4.70793903e-01 -2.64827937e-01 5.46495974e-01 4.82189327e-01
-8.03767025e-01 -7.97262132e-01 -2.20796764e-01 2.82460153e-01
3.99063230e-02 8.16423818e-02 -2.13911131e-01 8.19444060e-01
4.34951335e-02 9.13235486e-01 4.53652829e-01 -6.01347387e-01
6.14972711e-01 -2.53518313e-01 3.51812132e-02 -3.84962082e-01
1.46272108e-01 -6.40621781e-02 2.26015717e-01 -6.06642008e-01
-2.70669848e-01 -3.67206782e-01 -1.62259030e+00 -1.12601839e-01
-1.68977693e-01 -1.38433084e-01 1.03381622e+00 9.53170180e-01
8.93705636e-02 3.51929396e-01 1.07416523e+00 -1.00237310e+00
2.91614324e-01 -3.82629752e-01 -2.01418743e-01 3.53387892e-01
1.53405592e-01 -5.87639749e-01 -3.30882609e-01 8.74809623e-02]
|
[8.626986503601074, -1.7137805223464966]
|
d936bb9a-3dea-4b1e-ae48-baca170a2a57
|
conditioning-hierarchical-reinforcement
|
2302.10639
| null |
https://arxiv.org/abs/2302.10639v1
|
https://arxiv.org/pdf/2302.10639v1.pdf
|
Conditioning Hierarchical Reinforcement Learning on Flexible Constraints
|
Safety in goal directed Reinforcement Learning (RL) settings has typically been handled through constraints over trajectories and have demonstrated good performance in primarily short horizon tasks (goal is not too far away). In this paper, we are specifically interested in the problem of solving temporally extended decision making problems such as (1) robots that have to clean different areas in a house while avoiding slippery and unsafe areas (e.g., stairs) and retaining enough charge to move to a charging dock; (2) autonomous electric vehicles that have to reach a far away destination while having to optimize charging locations along the way; in the presence of complex safety constraints. Our key contribution is a (safety) Constrained Planning with Reinforcement Learning (CoP-RL) mechanism that combines a high-level constrained planning agent (which computes a reward maximizing path from a given start to a far away goal state while satisfying cost constraints) with a low-level goal conditioned RL agent (which estimates cost and reward values to move between nearby states). A major advantage of CoP-RL is that it can handle constraints on the cost value distribution (e.g., on Conditional Value at Risk, CVaR, and also on expected value). We perform extensive experiments with different types of safety constraints to demonstrate the utility of our approach over leading best approaches in constrained and hierarchical RL.
|
['Arunesh Sinha', 'Pradeep Varakantham', 'Yuxiao Lu']
|
2023-02-21
| null | null | null | null |
['hierarchical-reinforcement-learning']
|
['methodology']
|
[ 3.43845263e-02 4.87406611e-01 -3.41697901e-01 -2.94703186e-01
-1.08363760e+00 -5.39064407e-01 4.58118290e-01 4.99864012e-01
-6.01222038e-01 1.13772631e+00 -4.23061214e-02 -4.13522452e-01
-7.83617854e-01 -9.90853608e-01 -7.91930377e-01 -7.43702233e-01
-8.12036693e-01 7.71084845e-01 2.35848367e-01 -6.01016402e-01
2.14484453e-01 7.08501399e-01 -1.25204945e+00 -5.35145402e-01
8.22770178e-01 6.84437573e-01 5.01605630e-01 4.73960221e-01
6.17302299e-01 6.95153117e-01 -2.52398133e-01 4.07826096e-01
1.48733288e-01 -7.04990476e-02 -9.71376538e-01 1.10523865e-01
-6.36545002e-01 -4.80711162e-01 2.45764762e-01 8.39446008e-01
1.19136631e-01 7.19581187e-01 7.01713502e-01 -1.73655486e+00
1.22693442e-01 3.94662589e-01 -3.21094722e-01 -1.32522807e-01
2.99122930e-01 5.96191287e-01 8.26348186e-01 2.95028482e-02
2.43029520e-01 1.18987060e+00 1.24082275e-01 5.01908123e-01
-1.19721711e+00 -1.59459144e-01 7.57400036e-01 -1.36183370e-02
-1.26962984e+00 -4.76208655e-03 1.83924884e-01 -2.36960784e-01
1.31534457e+00 -3.58087383e-02 5.93457639e-01 6.93166792e-01
5.51849961e-01 4.37682539e-01 7.59310424e-01 -1.54861033e-01
9.62319672e-01 -3.39057326e-01 -1.71290144e-01 4.17739987e-01
1.10992752e-01 5.61479390e-01 1.26987502e-01 -1.39744028e-01
1.13228887e-01 -2.23133639e-01 1.56076312e-01 -6.33813024e-01
-9.69978809e-01 1.19280064e+00 4.49542075e-01 8.10837746e-02
-7.10278928e-01 4.44186926e-01 1.87990472e-01 5.91869503e-02
-2.09285036e-01 5.18646121e-01 -4.37397450e-01 -2.13668533e-02
-5.54911375e-01 8.65264118e-01 7.30398118e-01 1.31223536e+00
5.18790126e-01 9.73190069e-02 -3.50055903e-01 2.53544390e-01
3.48103315e-01 5.37732303e-01 -4.85672325e-01 -1.21909618e+00
8.33121181e-01 -3.47450259e-03 9.82868969e-01 -3.72962385e-01
-8.98571730e-01 1.91052064e-01 -3.79067361e-01 9.03736532e-01
3.53935570e-01 -7.26952076e-01 -6.72015250e-01 1.95942771e+00
1.61541596e-01 -1.84760302e-01 4.24072474e-01 8.77815425e-01
-3.48286450e-01 8.99465322e-01 5.77245772e-01 -4.61726278e-01
1.08880210e+00 -5.94617784e-01 -4.57888752e-01 -4.57577378e-01
6.59954548e-01 2.73250528e-02 6.26329064e-01 4.48060334e-01
-1.17916989e+00 3.12988758e-02 -1.15060043e+00 4.26916927e-01
-3.69362086e-01 -4.97743815e-01 3.99632245e-01 2.37312153e-01
-1.04044425e+00 9.16103005e-01 -8.81596267e-01 -3.62258971e-01
3.46754968e-01 5.90338588e-01 -1.41835675e-01 -1.62862614e-01
-1.29962599e+00 1.35558927e+00 4.45063889e-01 -1.54232802e-02
-1.58248639e+00 -2.68496692e-01 -1.11228514e+00 2.35902444e-01
9.53748286e-01 -2.54083127e-01 1.58229589e+00 -2.66644657e-01
-1.37370002e+00 -3.74327339e-02 4.26647812e-01 -5.49459994e-01
4.71866697e-01 -1.34560773e-02 -1.35968715e-01 -1.20819576e-01
4.60157394e-01 6.96178555e-01 4.37378109e-01 -1.08257186e+00
-9.65547085e-01 -3.57435226e-01 3.96270961e-01 6.77870393e-01
3.27800065e-01 -1.56163678e-01 9.95729566e-02 1.29467295e-02
-4.23822671e-01 -1.21548712e+00 -9.41970289e-01 -4.34313416e-01
-4.33146298e-01 -5.81220627e-01 2.39423826e-01 -3.35921705e-01
8.99133921e-01 -1.74583781e+00 2.57906795e-01 3.52587193e-01
-6.27107501e-01 -3.15184981e-01 -2.65303791e-01 7.28865206e-01
1.57912701e-01 -1.60372332e-01 -5.11527598e-01 -4.11776304e-02
3.15552086e-01 5.39326072e-01 -2.90084202e-02 5.79623640e-01
3.00042331e-01 5.30889988e-01 -1.35683358e+00 -1.55953452e-01
3.58697951e-01 -1.29030675e-01 -3.71836126e-01 1.94367871e-01
-6.29038036e-01 5.38382232e-01 -9.77860510e-01 5.15325010e-01
4.53515381e-01 3.68057400e-01 2.54674643e-01 6.31693363e-01
-6.25397444e-01 1.22070149e-01 -1.20448852e+00 1.52169001e+00
-6.46756649e-01 -1.31958444e-02 3.42253357e-01 -9.31311667e-01
6.32162631e-01 1.36736408e-01 7.12538242e-01 -8.52192998e-01
2.06382886e-01 -6.92939460e-02 -2.47084618e-01 -3.12524706e-01
6.12842202e-01 -4.11987543e-01 -8.12258303e-01 2.97667831e-01
-5.29047072e-01 -5.93523622e-01 9.24281627e-02 -8.92845821e-03
1.26168895e+00 4.03823346e-01 2.61887997e-01 -5.79462945e-01
3.33654881e-01 6.01504207e-01 5.51099181e-01 6.53807163e-01
-5.15644431e-01 -1.44508660e-01 8.11883271e-01 -5.19932434e-02
-8.19235504e-01 -1.03378141e+00 8.40630606e-02 1.03377080e+00
6.03209853e-01 1.49002701e-01 -5.58826387e-01 -5.53114533e-01
2.55342066e-01 1.48949230e+00 -4.34156120e-01 -1.99071854e-01
-6.50561750e-01 -4.42494303e-01 3.30593996e-02 6.66192591e-01
1.06645934e-01 -1.09109223e+00 -1.07959604e+00 3.98344487e-01
1.49755374e-01 -6.75548434e-01 -3.73589218e-01 7.78876066e-01
-5.45680642e-01 -1.08721185e+00 -4.28931028e-01 -6.29315376e-01
7.12365687e-01 -6.08363450e-02 7.87808180e-01 -3.58321369e-01
6.88440502e-02 5.46573877e-01 -1.75450534e-01 -6.46152735e-01
-1.56861052e-01 -1.45471141e-01 1.04378454e-01 -3.93297642e-01
-2.35754177e-02 -6.97532743e-02 -3.85479569e-01 4.35507774e-01
-5.68045378e-01 -2.25903943e-01 2.06944212e-01 5.15580893e-01
7.04989493e-01 7.31143355e-01 7.44230449e-01 -3.64536881e-01
8.25655520e-01 -8.53287101e-01 -1.25953567e+00 2.52911866e-01
-6.12966180e-01 2.99446940e-01 6.56618118e-01 -1.26356870e-01
-9.70648468e-01 3.80012721e-01 6.14447147e-03 1.17947042e-01
-1.29573509e-01 2.54732102e-01 -2.97831297e-01 2.99005389e-01
1.64450154e-01 -1.92352384e-01 9.40144211e-02 1.41437557e-02
2.85220295e-01 1.51751310e-01 2.53127724e-01 -8.75707150e-01
6.08916461e-01 1.53822571e-01 4.73903179e-01 -2.53963053e-01
-4.33428317e-01 -1.59340277e-01 -4.67737287e-01 -3.45052838e-01
1.04502964e+00 -8.34556580e-01 -1.34917450e+00 -5.07113012e-03
-6.38270497e-01 -1.12026489e+00 -3.66338313e-01 3.40412617e-01
-1.28200912e+00 3.59272175e-02 -1.54207483e-01 -1.25959086e+00
1.61006480e-01 -1.12256777e+00 8.79594982e-01 3.24672282e-01
-9.60892439e-02 -8.10458660e-01 1.08691806e-03 -7.42065012e-02
2.87160754e-01 7.78299809e-01 8.85650039e-01 -1.34500653e-01
-6.67368650e-01 1.60437450e-01 2.38346636e-01 -6.32443950e-02
-5.72435893e-02 -5.35821259e-01 -3.44146401e-01 -6.31819010e-01
-2.62712687e-01 -5.28117955e-01 2.98146427e-01 6.51835561e-01
8.35531831e-01 -3.90294105e-01 -5.85064769e-01 -3.91760990e-02
1.63731134e+00 8.54371548e-01 4.65842515e-01 6.75046504e-01
1.20699238e-02 9.50808764e-01 1.65878832e+00 7.43383706e-01
7.64663041e-01 7.78343141e-01 1.15424073e+00 2.97886193e-01
7.12872505e-01 -1.02385543e-01 4.90967005e-01 -3.88274938e-01
1.13922298e-01 -4.34841245e-01 -7.59760439e-01 8.74448776e-01
-2.22893238e+00 -8.99142206e-01 1.32841229e-01 2.38526583e+00
3.83934438e-01 2.92934328e-01 2.41462618e-01 -1.66885853e-01
5.12236238e-01 -7.56198838e-02 -9.95465338e-01 -9.68424976e-01
5.34861684e-01 -1.07462965e-01 1.11211574e+00 8.51791501e-01
-1.04697096e+00 8.17814052e-01 5.89578056e+00 5.81290305e-01
-4.08577293e-01 -1.08036682e-01 5.29213250e-01 -2.98040539e-01
-1.01514995e-01 1.22905158e-01 -8.06231737e-01 4.54471231e-01
1.01178575e+00 -1.68275520e-01 8.26593339e-01 9.53722239e-01
6.65176570e-01 -7.86884546e-01 -1.41508412e+00 3.72657299e-01
-7.48321652e-01 -7.73366988e-01 -7.18143106e-01 7.77374506e-02
6.67724848e-01 1.89496893e-02 -1.48939207e-01 6.52929127e-01
1.12272179e+00 -1.08280528e+00 1.02006960e+00 5.12545824e-01
5.92533648e-01 -1.45149207e+00 4.40431148e-01 8.18881452e-01
-1.17798769e+00 -5.90096116e-01 -4.90694270e-02 -1.38225660e-01
6.00347459e-01 1.06841717e-02 -7.63331950e-01 6.42533839e-01
6.62043750e-01 3.23206753e-01 3.33742559e-01 7.39893317e-01
-4.29265946e-01 -2.12827310e-01 -4.45606142e-01 -4.76039678e-01
1.01565015e+00 -3.22023362e-01 4.27618504e-01 7.13129759e-01
2.93038815e-01 3.54898721e-01 5.75374067e-01 8.28399122e-01
7.07169831e-01 -4.63209450e-01 -8.34085941e-01 3.14371794e-01
3.87361735e-01 1.05609894e+00 -7.87472904e-01 8.78822282e-02
-1.00604177e-01 5.69764793e-01 2.87163734e-01 3.07554930e-01
-1.09935570e+00 -4.84793127e-01 9.53489780e-01 -2.19092239e-02
2.91088969e-01 -4.11676407e-01 -1.03651695e-01 -1.83640495e-01
-2.55540520e-01 -2.22690910e-01 6.23036504e-01 -8.86260927e-01
-8.00540626e-01 2.72316098e-01 4.67663288e-01 -1.16428947e+00
-4.49536443e-01 -3.44484031e-01 -7.18120635e-01 1.02567923e+00
-2.03083158e+00 -6.02983356e-01 3.39712724e-02 7.71447599e-01
6.30938411e-01 7.20633939e-02 5.31030118e-01 -1.40610024e-01
-3.78581226e-01 -1.59633636e-01 7.63503313e-02 -4.67181921e-01
8.37424099e-02 -1.29193294e+00 2.44216584e-02 5.87861001e-01
-1.10059106e+00 -1.20047517e-02 6.80181265e-01 -8.25276613e-01
-1.61774313e+00 -1.39680052e+00 6.19409144e-01 -2.86836058e-01
4.71466273e-01 -1.94110006e-01 -3.63333225e-01 7.51604617e-01
8.60807598e-02 -3.13484818e-01 -1.97226450e-01 -2.44031385e-01
5.74410260e-01 3.06500029e-02 -1.53821695e+00 7.99145639e-01
7.24547207e-01 1.77006155e-01 -4.04627711e-01 4.20264244e-01
6.29751146e-01 -3.64537477e-01 -5.65808594e-01 3.22593242e-01
5.50201759e-02 -5.93996406e-01 6.88198209e-01 -6.12717450e-01
-1.40471458e-01 -3.83241296e-01 -1.41247183e-01 -1.61222935e+00
-6.30569041e-01 -7.59580374e-01 3.53839904e-01 8.77109706e-01
4.15198416e-01 -4.71602738e-01 6.52677715e-01 1.09825194e+00
-3.98492962e-01 -8.60315681e-01 -1.25173354e+00 -1.14643407e+00
2.50217587e-01 -5.26405036e-01 8.16959560e-01 3.71005714e-01
3.28942388e-01 1.83798801e-02 -3.33975643e-01 7.74323702e-01
8.01324427e-01 -1.61939085e-01 1.56203270e-01 -7.80767560e-01
1.69682637e-01 -2.42130920e-01 1.92243084e-01 -5.61582804e-01
3.64824295e-01 -6.62097573e-01 8.62807691e-01 -2.04033136e+00
-2.52204895e-01 -1.03233755e+00 -2.04279989e-01 8.04793775e-01
3.23694766e-01 -7.27524817e-01 2.94682294e-01 -2.80730098e-01
-8.43941569e-01 6.60171390e-01 1.17037594e+00 -2.13320091e-01
-4.67147022e-01 3.40812147e-01 -6.28618300e-01 5.11217237e-01
1.09012580e+00 -5.53758442e-01 -7.59190559e-01 -2.42130175e-01
2.51677513e-01 9.98859823e-01 1.58951834e-01 -8.93544257e-01
2.28840619e-01 -1.03280151e+00 -1.19891010e-01 -6.02963388e-01
3.92950714e-01 -1.21527541e+00 6.54851124e-02 8.81523192e-01
-4.66253847e-01 2.17903674e-01 3.13567042e-01 8.24066937e-01
3.00254852e-01 -6.01911068e-01 8.66913557e-01 -2.03188106e-01
-9.95519161e-01 2.82762885e-01 -9.47563887e-01 -2.94573288e-02
1.88050723e+00 1.79658592e-01 -1.96912717e-02 -4.06267077e-01
-8.75557423e-01 1.46375811e+00 1.48007393e-01 4.54682648e-01
4.71873939e-01 -1.02192748e+00 -3.99129033e-01 -1.79662168e-01
-3.49444114e-02 2.82882571e-01 8.04946125e-02 5.19514263e-01
-1.73032627e-01 5.98688722e-01 -3.48632604e-01 -2.97050148e-01
-7.26689219e-01 9.10522163e-01 4.72965091e-01 -4.59642828e-01
-6.66221201e-01 4.13265139e-01 -8.23245198e-02 -3.73903036e-01
4.68387812e-01 -5.39358854e-01 -3.70484978e-01 4.24548015e-02
2.82396495e-01 6.42246068e-01 -5.90793788e-02 -2.97843397e-01
-5.12622416e-01 3.56583029e-01 3.60896200e-01 -3.95378441e-01
1.40842831e+00 -2.20163658e-01 3.80295187e-01 3.06280646e-02
5.41604877e-01 -6.31052911e-01 -1.81743920e+00 3.26352417e-01
2.67575234e-01 -1.09113649e-01 7.58278295e-02 -9.49538648e-01
-5.47255576e-01 3.15791875e-01 4.31237578e-01 3.40843290e-01
8.63750696e-01 -5.59003986e-02 5.13583124e-01 4.17100996e-01
1.09816122e+00 -1.54636669e+00 -1.10254008e-02 6.19728148e-01
7.69234836e-01 -1.02273524e+00 -3.07214633e-02 -3.10664009e-02
-1.16331828e+00 8.08732808e-01 6.19034708e-01 -2.28616506e-01
4.52521831e-01 3.59467119e-01 -4.21584070e-01 -1.24383539e-01
-8.87836456e-01 -5.58984995e-01 -4.69227761e-01 9.02633071e-01
-5.04965365e-01 3.40929240e-01 -2.24895164e-01 3.00707996e-01
1.84953839e-01 -8.40163901e-02 7.29081631e-01 1.31402969e+00
-8.27038944e-01 -8.50650370e-01 -4.67016995e-01 2.08968073e-01
4.51393276e-02 5.11683583e-01 3.46862644e-01 9.25780952e-01
1.94963723e-01 1.35776675e+00 1.98616192e-01 3.03559929e-01
7.07286954e-01 -2.01296628e-01 4.13611710e-01 -6.82808101e-01
-4.53330249e-01 -9.37945917e-02 3.56244832e-01 -8.81141961e-01
-1.80180252e-01 -1.00004315e+00 -1.95079052e+00 2.39247233e-02
1.04106730e-02 2.65414000e-01 7.65295565e-01 9.30096805e-01
-8.94368975e-04 7.16648936e-01 9.90661204e-01 -1.09162736e+00
-8.88555646e-01 -2.71682709e-01 -9.09822702e-01 -1.50652617e-01
5.91409862e-01 -8.15083027e-01 -1.77970558e-01 -6.03663743e-01]
|
[4.607389450073242, 1.9782577753067017]
|
72f8ea7a-ddcd-4eee-a3fd-db6097826439
|
analysis-computational-complexity-reduction
|
2206.09071
| null |
https://arxiv.org/abs/2206.09071v1
|
https://arxiv.org/pdf/2206.09071v1.pdf
|
Analysis & Computational Complexity Reduction of Monocular and Stereo Depth Estimation Techniques
|
Accurate depth estimation with lowest compute and energy cost is a crucial requirement for unmanned and battery operated autonomous systems. Robotic applications require real time depth estimation for navigation and decision making under rapidly changing 3D surroundings. A high accuracy algorithm may provide the best depth estimation but may consume tremendous compute and energy resources. A general trade-off is to choose less accurate methods for initial depth estimate and a more accurate yet compute intensive method when needed. Previous work has shown this trade-off can be improved by developing a state-of-the-art method (AnyNet) to improve stereo depth estimation. We studied both the monocular and stereo vision depth estimation methods and investigated methods to reduce computational complexity of these methods. This was our baseline. Consequently, our experiments show reduction of monocular depth estimation model size by ~75% reduces accuracy by less than 2% (SSIM metric). Our experiments with the novel stereo vision method (AnyNet) show that accuracy of depth estimation does not degrade more than 3% (three pixel error metric) in spite of reduction in model size by ~20%. We have shown that smaller models can indeed perform competitively.
|
['Varo Ly', 'Rajeev Patwari']
|
2022-06-18
| null | null | null | null |
['stereo-depth-estimation']
|
['computer-vision']
|
[ 2.43758589e-01 5.76335788e-02 3.00496608e-01 -2.91418850e-01
-4.63039041e-01 -5.79969764e-01 5.88089764e-01 -1.52630433e-01
-9.08925235e-01 8.76940191e-01 -2.34922096e-01 -2.59189308e-01
2.91957378e-01 -1.02502370e+00 -4.92862850e-01 -5.64517140e-01
-6.92481995e-02 5.16291440e-01 7.39506245e-01 -1.72357112e-01
7.02504337e-01 7.36119747e-01 -1.85216010e+00 -2.65025854e-01
7.81718314e-01 1.19878316e+00 3.66403639e-01 1.02614772e+00
2.15860263e-01 5.28188646e-01 -4.69997048e-01 7.86565766e-02
7.46690154e-01 -2.05508277e-01 -7.38993466e-01 -4.37476896e-02
6.43553555e-01 -7.87368655e-01 -3.22687328e-01 9.97971058e-01
7.30830550e-01 3.71909253e-02 5.71263790e-01 -1.09413779e+00
5.28440773e-01 -2.77154446e-01 -7.14435399e-01 1.75315663e-01
5.65830827e-01 2.73714125e-01 3.79471749e-01 -6.26718879e-01
7.46126592e-01 1.01284122e+00 6.39581382e-01 6.46986067e-01
-8.90594602e-01 -4.60249484e-01 -1.97592303e-01 4.25951302e-01
-1.45942438e+00 -5.25610745e-01 2.78215021e-01 -1.13833629e-01
1.47757173e+00 6.64074570e-02 9.18607056e-01 3.76236171e-01
5.40517390e-01 4.46729928e-01 1.06756449e+00 -3.62587243e-01
5.68957090e-01 -3.75223570e-02 -1.87340111e-01 8.79980505e-01
7.01434553e-01 3.87223214e-01 -5.21142006e-01 2.35984996e-01
8.47568870e-01 -3.50661457e-01 -3.75727654e-01 -5.73205531e-01
-1.13819611e+00 6.51371479e-01 4.88808006e-01 -1.07132740e-01
-2.41311193e-01 4.71525848e-01 2.67088026e-01 4.04139847e-01
2.18940184e-01 5.90895414e-01 -3.68726313e-01 -6.34028494e-01
-8.55924547e-01 2.76676416e-01 8.91816616e-01 9.68447328e-01
1.04383671e+00 1.11679947e-02 6.43073082e-01 4.90075558e-01
2.49003649e-01 6.92187905e-01 5.63241839e-01 -1.40377200e+00
3.75860602e-01 5.46879232e-01 2.93172717e-01 -8.10626388e-01
-5.31694829e-01 -5.58317006e-02 -6.35993898e-01 9.90587890e-01
3.99249732e-01 -2.05534026e-01 -9.36262131e-01 1.07190859e+00
3.31918448e-01 -2.17858732e-01 1.69150323e-01 1.03901172e+00
5.68954349e-01 4.91813540e-01 -5.24445534e-01 -3.25566009e-02
9.02848601e-01 -8.57814252e-01 -2.13236555e-01 -7.55167127e-01
7.20699489e-01 -6.91187680e-01 5.93812585e-01 6.63378596e-01
-1.07717848e+00 -1.91332504e-01 -1.72755218e+00 -1.91263959e-01
-7.05879554e-02 -3.06861788e-01 6.81053996e-01 8.00988138e-01
-1.39892912e+00 6.36456788e-01 -1.01202381e+00 -7.08652735e-01
-1.78951840e-03 6.64234221e-01 -5.32925248e-01 -3.57682616e-01
-6.77130878e-01 1.25535226e+00 2.81514972e-01 -6.56845793e-02
-7.44904220e-01 -2.78177232e-01 -9.65016186e-01 -4.72512066e-01
2.10498661e-01 -9.49314356e-01 1.36965573e+00 -5.77239752e-01
-1.71292901e+00 9.76455927e-01 -3.38901490e-01 -8.08367848e-01
7.18591571e-01 -3.19753736e-01 2.88302898e-01 2.72615284e-01
-6.45568147e-02 1.09048069e+00 5.38779140e-01 -1.03759241e+00
-1.08441567e+00 -6.14222229e-01 3.36436331e-01 6.81683540e-01
1.58716023e-01 -7.98741698e-01 -4.27470833e-01 1.43744573e-01
8.52669001e-01 -1.19309735e+00 -3.58589709e-01 4.36518461e-01
2.49101445e-01 3.82358581e-01 8.36348176e-01 -2.85117269e-01
6.53766870e-01 -1.65238631e+00 -5.44732325e-02 -6.52472004e-02
1.99353337e-01 -8.57724398e-02 2.94844806e-01 2.97670990e-01
6.54296458e-01 -3.38456571e-01 -1.05824456e-01 -3.08676183e-01
-3.38826478e-01 4.04369831e-01 2.31395781e-01 8.22784185e-01
-4.32633072e-01 4.76259619e-01 -8.10461700e-01 -3.44255239e-01
8.16162348e-01 4.21539664e-01 -6.93906426e-01 3.96019183e-02
1.47161469e-01 2.23338231e-01 -1.09199258e-02 7.00747609e-01
8.30089092e-01 2.01017722e-01 4.19584773e-02 -2.13852674e-01
-3.39171380e-01 2.89194226e-01 -1.18963385e+00 1.96068239e+00
-6.97979212e-01 1.08334911e+00 2.58482039e-01 -5.94144106e-01
8.60396266e-01 -9.66511890e-02 3.76240730e-01 -1.08004475e+00
1.88100055e-01 5.76907754e-01 -6.00347109e-02 -1.84449941e-01
8.42215657e-01 -1.33875981e-01 1.05132960e-01 1.04739718e-01
-3.46320331e-01 -8.89733136e-01 -1.52498735e-02 1.85928959e-02
1.33723199e+00 2.22371340e-01 5.69051862e-01 -4.15613562e-01
3.95583183e-01 4.24197793e-01 4.18971777e-01 5.12373209e-01
-6.14790976e-01 6.02229476e-01 3.74611504e-02 -5.49267709e-01
-1.09921992e+00 -6.79658830e-01 -1.13378704e-01 4.16258514e-01
7.69322336e-01 -5.99390678e-02 -6.67705417e-01 -1.93888709e-01
-1.41585264e-02 3.45611960e-01 -2.92112619e-01 1.26647398e-01
-5.33065975e-01 -5.41168451e-01 2.34910980e-01 5.04768848e-01
1.07624769e+00 -4.50226516e-01 -1.57086003e+00 2.43227184e-01
-2.94193309e-02 -1.30690539e+00 9.60547104e-02 3.87635171e-01
-1.31648445e+00 -1.15236259e+00 -6.68732882e-01 -5.67579150e-01
6.08277798e-01 6.43693268e-01 1.14575303e+00 -5.23454063e-02
-1.66469425e-01 3.53174299e-01 -2.16325015e-01 -3.64297599e-01
1.19944468e-01 2.27831975e-02 1.15089133e-01 -8.52396071e-01
3.25795621e-01 -5.46635866e-01 -1.19654441e+00 4.13246691e-01
-5.79158425e-01 3.37452501e-01 4.42563146e-01 4.93074089e-01
4.96755540e-01 -1.36860749e-02 -1.60540104e-01 -3.89778972e-01
1.19190447e-01 1.95520446e-02 -1.09444535e+00 -5.11673093e-01
-8.76665831e-01 3.90540987e-01 2.40893245e-01 2.11931951e-02
-7.26714313e-01 5.59068322e-01 -1.51278600e-01 -7.37843812e-02
1.00083992e-01 9.34275165e-02 1.15109906e-01 -7.33853042e-01
8.55351746e-01 1.10195771e-01 1.67572692e-01 7.94874784e-03
3.42659354e-02 5.68743765e-01 6.79315269e-01 -7.94624612e-02
5.68796396e-01 1.05487466e+00 5.54566562e-01 -9.64964628e-01
-2.10766777e-01 -6.40410900e-01 -6.89619899e-01 -3.12448025e-01
5.28157711e-01 -1.18898118e+00 -7.15342462e-01 6.59529686e-01
-1.00629234e+00 -4.68493462e-01 7.57369995e-02 4.99930412e-01
-6.82098269e-01 5.92503726e-01 -4.09698218e-01 -9.01797712e-01
-3.91162813e-01 -1.22825813e+00 1.19028926e+00 1.49803057e-01
-1.94235742e-01 -9.28428054e-01 5.31342737e-02 2.11741254e-01
5.06872058e-01 3.73783171e-01 1.07585393e-01 3.91267061e-01
-9.36090708e-01 -2.32876986e-01 -3.96115631e-01 -8.75615478e-02
-7.39437714e-03 -3.74356121e-01 -1.06380165e+00 -4.07463223e-01
4.02650423e-02 -8.50711688e-02 8.90562296e-01 5.57223439e-01
4.52052951e-01 1.37020901e-01 -4.05459076e-01 8.14395070e-01
2.00846124e+00 4.93585497e-01 1.00234294e+00 7.92783141e-01
5.80980897e-01 4.18294400e-01 1.01432014e+00 4.14976627e-01
4.98428106e-01 8.95949423e-01 8.64430904e-01 2.06668928e-01
-1.53795749e-01 1.25451773e-01 3.70858610e-01 5.07013917e-01
-1.63134813e-01 -5.76068237e-02 -1.13646400e+00 5.73061287e-01
-1.75741398e+00 -6.27835393e-01 -1.56399354e-01 2.42126250e+00
4.71198976e-01 3.20118636e-01 -5.29926755e-02 4.92302328e-01
1.47329718e-01 -1.26984537e-01 -6.73448682e-01 -5.32813132e-01
1.92692906e-01 -1.58793293e-02 1.25763607e+00 8.64190817e-01
-8.71714056e-01 9.21463132e-01 6.60340023e+00 4.48781729e-01
-1.33126593e+00 -1.61841512e-01 2.75442004e-01 -3.09261531e-01
1.21655958e-02 -6.43923581e-02 -9.69462454e-01 2.89709300e-01
9.53201294e-01 -1.25248089e-01 3.52526546e-01 1.11277056e+00
2.89312750e-01 -1.21788418e+00 -1.06896520e+00 1.67705333e+00
8.80425200e-02 -1.09060395e+00 -2.75726587e-01 4.17019993e-01
7.84258604e-01 3.32129389e-01 -5.01283884e-01 -6.18923120e-02
1.45897418e-01 -8.40979576e-01 6.40552640e-01 1.25297502e-01
7.60830998e-01 -8.33360732e-01 1.09242702e+00 4.25930917e-01
-1.36722898e+00 1.68652758e-01 -6.81582093e-01 -5.91092646e-01
1.40625238e-01 6.11959100e-01 -9.59363043e-01 2.18538985e-01
8.02067161e-01 5.75608373e-01 -3.74934912e-01 1.23543727e+00
-4.56870496e-02 -2.54903644e-01 -6.49366438e-01 -2.08407193e-01
3.38957727e-01 -5.72693013e-02 4.25269306e-01 8.08057845e-01
5.06762505e-01 2.96144456e-01 -2.32959971e-01 1.86182559e-01
2.84053117e-01 -3.74143541e-01 -9.18104827e-01 6.04595423e-01
5.15547395e-01 9.04389381e-01 -7.83792973e-01 -3.96860629e-01
-3.31525236e-01 1.41516113e+00 -3.49416323e-02 -1.65647119e-01
-5.78662395e-01 -4.59550649e-01 8.40076089e-01 3.67080152e-01
6.91716969e-02 -7.83711493e-01 -7.20323324e-01 -1.06365323e+00
-1.87624656e-02 -4.08501118e-01 -1.21962130e-02 -8.67645562e-01
-5.16539752e-01 7.05344021e-01 -1.46957427e-01 -1.56443071e+00
-7.22781658e-01 -8.06510329e-01 -2.16338481e-03 7.25106359e-01
-1.75519133e+00 -4.65688050e-01 -1.10556293e+00 3.60036999e-01
8.08773398e-01 1.69074371e-01 6.02885306e-01 2.13476256e-01
1.12120256e-01 9.28463712e-02 -1.22255730e-02 -3.95274401e-01
4.69690233e-01 -1.27115333e+00 6.00063622e-01 9.65987921e-01
-4.13493872e-01 1.56338617e-01 9.32960570e-01 -5.74098289e-01
-1.82307982e+00 -6.80675030e-01 6.75280571e-01 -5.04214466e-01
2.00143740e-01 -1.97359637e-04 -3.20061833e-01 2.32018694e-01
-1.78485047e-02 -1.11074239e-01 -1.01785235e-01 -3.52256775e-01
7.05035776e-02 -1.15319997e-01 -1.41812909e+00 5.89330792e-01
1.29241443e+00 -3.60280544e-01 -2.46476293e-01 -1.82673872e-01
3.21347415e-01 -7.19682217e-01 -4.74765867e-01 6.22499883e-01
9.89704370e-01 -1.60124683e+00 8.89078915e-01 7.08121836e-01
2.73260415e-01 -4.95090872e-01 -3.42490762e-01 -1.10889268e+00
1.24851920e-01 -2.70994335e-01 -1.30104780e-01 2.55569220e-01
1.94403335e-01 -6.79955423e-01 1.29260886e+00 6.83522344e-01
-1.99391350e-01 -5.21371961e-01 -1.07058954e+00 -9.58280981e-01
-2.43517056e-01 -6.65959537e-01 8.74298140e-02 3.43185425e-01
-1.90079361e-02 2.36572444e-01 -7.67515302e-02 2.28580743e-01
7.58894682e-01 -6.31189123e-02 1.07861924e+00 -1.11776018e+00
1.59949511e-02 -2.98831522e-01 -1.18618727e+00 -1.54054153e+00
-4.63059455e-01 -1.70391187e-01 3.07828993e-01 -2.00457382e+00
-1.08243421e-01 -2.54353553e-01 4.42988634e-01 -2.44812723e-02
2.09048808e-01 5.09270668e-01 -9.35298130e-02 1.81071579e-01
-4.30185199e-01 2.84565359e-01 1.04695821e+00 8.57761055e-02
-1.20554678e-01 -3.10093522e-01 -1.84169620e-01 8.05214763e-01
6.71136141e-01 -2.13189557e-01 -6.35391355e-01 -5.79002321e-01
3.83449256e-01 1.07874542e-01 2.81149745e-01 -1.81175804e+00
4.43089306e-01 1.10581987e-01 3.68545264e-01 -7.60467947e-01
7.85272539e-01 -1.04410613e+00 5.36601022e-02 9.27367985e-01
5.18642128e-01 9.57047939e-02 4.29295003e-01 3.91010642e-01
-2.10491732e-01 -1.93360940e-01 1.05676317e+00 -4.18691456e-01
-1.45655715e+00 2.41243094e-02 -5.05163074e-01 -3.99297565e-01
1.04486918e+00 -1.08239317e+00 -2.18788013e-01 -6.64819479e-01
-2.81253964e-01 2.88932137e-02 1.07383728e+00 3.49309221e-02
7.09850848e-01 -9.44369316e-01 -4.36703056e-01 1.79477677e-01
1.56349689e-02 1.83257297e-01 -1.15418844e-01 5.20484388e-01
-1.59621179e+00 7.73255050e-01 -4.10048723e-01 -9.08770084e-01
-1.44938552e+00 -1.26854941e-01 6.56729043e-01 3.13834250e-02
-4.62740660e-01 1.03553724e+00 2.33557634e-02 -2.08700687e-01
2.57808775e-01 -5.59903860e-01 1.53474525e-01 -2.59553730e-01
4.99516338e-01 6.82090700e-01 3.54304373e-01 -2.48652488e-01
-6.10618055e-01 9.35860634e-01 3.09883773e-01 -5.37133098e-01
1.11444879e+00 -6.23073280e-01 1.75723150e-01 1.86277092e-01
1.04714227e+00 -2.48416767e-01 -1.56206429e+00 2.98359543e-01
-2.82463372e-01 -7.12164402e-01 5.96949935e-01 -5.17511308e-01
-9.02024508e-01 8.09760451e-01 9.83862221e-01 -3.62070687e-02
1.23529720e+00 -4.10349399e-01 8.04225445e-01 5.20736754e-01
1.31514192e+00 -1.04591072e+00 -2.15676963e-01 6.55655026e-01
6.49538338e-01 -1.58601153e+00 4.28948343e-01 -5.34868836e-01
-3.30024332e-01 1.12758803e+00 7.78299570e-01 -2.68246770e-01
4.95683908e-01 4.73668516e-01 1.91558763e-01 -2.24443048e-01
-6.02929652e-01 -3.85749519e-01 -5.61366938e-02 7.06654608e-01
2.50475824e-01 -2.20785677e-01 -3.39868665e-01 -4.26562786e-01
-4.74667937e-01 -5.79872876e-02 6.95221364e-01 1.25669682e+00
-9.13105905e-01 -8.04435551e-01 -4.27455276e-01 1.90826148e-01
-5.50603792e-02 -6.43642545e-02 -2.79717267e-01 8.61536384e-01
-1.20972112e-01 1.10379362e+00 2.26379201e-01 -5.66927314e-01
2.35398918e-01 -3.71407151e-01 8.21603358e-01 -4.49381620e-01
-1.79447800e-01 -2.44848371e-01 3.15315694e-01 -1.02302647e+00
-5.02225697e-01 -4.89842713e-01 -1.27979589e+00 -7.92350948e-01
-1.83323845e-01 -2.43055925e-01 1.23816073e+00 7.03241467e-01
3.25579315e-01 2.41642073e-01 3.42267424e-01 -1.33735001e+00
2.33211070e-02 -7.34300375e-01 -4.85583305e-01 -2.38431394e-01
2.50321150e-01 -6.47925615e-01 -6.45454228e-01 -8.24675784e-02]
|
[8.675045013427734, -2.3673534393310547]
|
bddc4316-e84c-4318-9a33-3cd695ac38de
|
190600638
|
1906.00638
| null |
https://arxiv.org/abs/1906.00638v1
|
https://arxiv.org/pdf/1906.00638v1.pdf
|
Federated Hierarchical Hybrid Networks for Clickbait Detection
|
Online media outlets adopt clickbait techniques to lure readers to click on articles in a bid to expand their reach and subsequently increase revenue through ad monetization. As the adverse effects of clickbait attract more and more attention, researchers have started to explore machine learning techniques to automatically detect clickbaits. Previous work on clickbait detection assumes that all the training data is available locally during training. In many real-world applications, however, training data is generally distributedly stored by different parties (e.g., different parties maintain data with different feature spaces), and the parties cannot share their data with each other due to data privacy issues. It is challenging to build models of high-quality federally for detecting clickbaits effectively without data sharing. In this paper, we propose a federated training framework, which is called federated hierarchical hybrid networks, to build clickbait detection models, where the titles and contents are stored by different parties, whose relationships must be exploited for clickbait detection. We empirically demonstrate that our approach is effective by comparing our approach to the state-of-the-art approaches using datasets from social media.
|
['Hankz Hankui Zhuo', 'Feng Liao', 'Xiaoling Huang', 'Yu Zhang']
|
2019-06-03
| null | null | null | null |
['clickbait-detection']
|
['natural-language-processing']
|
[-4.10592943e-01 -2.50217676e-01 -9.09952462e-01 -4.45173621e-01
-8.80949736e-01 -8.23469579e-01 6.47727728e-01 2.29296342e-01
-4.32985067e-01 7.24443674e-01 2.29542842e-03 -5.97253859e-01
-9.49656069e-02 -9.80767488e-01 -8.24525416e-01 -1.82032902e-02
1.13756582e-01 4.22618866e-01 4.95669156e-01 5.40521666e-02
2.71203578e-01 2.88389940e-02 -1.35558152e+00 4.50028300e-01
9.73859370e-01 1.22753155e+00 -2.43226126e-01 3.58605146e-01
-4.09140438e-01 9.42182660e-01 -4.02564168e-01 -8.86222541e-01
8.41579378e-01 -4.92850654e-02 -7.49893188e-01 -2.67188519e-01
7.44434953e-01 -6.51794851e-01 -6.46053910e-01 1.31010377e+00
-9.50913280e-02 -2.30243623e-01 1.86757833e-01 -1.44757414e+00
-8.23727071e-01 7.29184031e-01 -5.14477432e-01 1.29657313e-01
-3.45590599e-02 -1.45725394e-02 1.73183775e+00 -3.02424282e-01
7.09770739e-01 6.19631112e-01 3.84127945e-01 1.58849970e-01
-1.31837773e+00 -9.02095199e-01 2.21393034e-01 2.87422776e-01
-1.12813580e+00 -1.27601758e-01 5.56749701e-01 -5.08064210e-01
1.78879157e-01 1.10915565e+00 6.43163443e-01 9.69614208e-01
-3.06041807e-01 1.64341974e+00 1.30040467e+00 -5.84353916e-02
3.07097942e-01 7.19720483e-01 5.32670259e-01 1.21705316e-01
6.17954016e-01 4.00991067e-02 -5.40992677e-01 -6.23960972e-01
1.72815859e-01 5.30834615e-01 1.86613761e-02 -5.11409044e-01
-7.10420549e-01 1.30117714e+00 5.55773437e-01 3.06253076e-01
-2.85840094e-01 -6.05384260e-02 2.29019552e-01 4.21089292e-01
7.46508002e-01 4.86547589e-01 -5.68862081e-01 1.94431722e-01
-1.31528747e+00 6.20995581e-01 1.09589994e+00 9.95329142e-01
9.46341276e-01 -8.74733090e-01 -1.53782532e-01 6.92557275e-01
3.71219426e-01 2.48514280e-01 4.90045190e-01 -7.09082901e-01
8.26877832e-01 9.19696748e-01 3.56654733e-01 -1.07155323e+00
2.64345407e-01 -4.83196199e-01 -4.66709167e-01 -3.86752754e-01
4.81321871e-01 1.59178097e-02 -6.18305027e-01 1.19244230e+00
2.83658653e-01 -1.54429629e-01 -6.43477082e-01 9.82851088e-01
1.72409311e-01 2.65121043e-01 -1.19826481e-01 2.26477429e-01
1.39463520e+00 -8.85121405e-01 -8.35560799e-01 -2.32453719e-01
6.26056790e-01 -6.66710556e-01 7.40071297e-01 1.96809500e-01
-9.71832871e-01 1.51514843e-01 -8.21205020e-01 -9.61398706e-02
-8.93682718e-01 -1.85223699e-01 8.44099641e-01 7.93871582e-01
-2.42324576e-01 5.08483112e-01 -6.85579419e-01 -6.68549240e-02
1.00855362e+00 1.59296364e-01 1.05366357e-01 -4.31507127e-03
-1.38509881e+00 3.62916559e-01 4.97294553e-02 -1.86505720e-01
-6.52237117e-01 -9.30425584e-01 -3.79653722e-02 2.10040167e-01
8.61042082e-01 -3.87540013e-01 1.54275632e+00 -6.62328839e-01
-6.41568601e-01 6.58139348e-01 2.58852214e-01 -8.26674759e-01
1.00368023e+00 -2.60127902e-01 -5.59976697e-01 -2.06498772e-01
3.36125404e-01 -7.80254081e-02 7.84732997e-01 -8.27419877e-01
-1.15760458e+00 -6.42144024e-01 6.04656525e-02 -4.33281273e-01
-6.49216473e-01 1.20227605e-01 -4.42584008e-01 -3.23617935e-01
8.25545490e-02 -8.78743589e-01 -1.23022959e-01 -4.31584977e-02
-9.56907213e-01 -4.19526935e-01 9.59492803e-01 -7.11539149e-01
1.67316449e+00 -1.83794081e+00 -5.27165592e-01 7.18875885e-01
9.97670472e-01 5.42590261e-01 3.06734920e-01 4.94296014e-01
3.33903104e-01 8.93898964e-01 5.63235521e-01 4.46319915e-02
4.09660280e-01 -1.37622148e-01 -5.38934827e-01 2.94270188e-01
-4.80374545e-01 9.29400861e-01 -5.25147676e-01 -5.33651769e-01
-1.31477818e-01 -2.35733420e-01 -6.73428833e-01 1.87327892e-01
-4.31489706e-01 -2.06402794e-01 -1.26844370e+00 7.82680929e-01
8.74226928e-01 -8.02693188e-01 2.62473583e-01 -7.57089853e-02
-4.51917678e-01 6.59344018e-01 -9.49589014e-01 8.81670594e-01
-8.42747837e-02 5.93132198e-01 3.89280856e-01 -7.66119123e-01
4.45272535e-01 5.77459186e-02 5.45449495e-01 -1.00982845e+00
1.34575114e-01 5.56957245e-01 -3.38145703e-01 -4.63425577e-01
5.57996690e-01 5.13640404e-01 1.01206414e-01 8.91312242e-01
-5.53109884e-01 6.89199984e-01 2.61285752e-01 6.04806364e-01
1.39891183e+00 -6.37461245e-01 -1.01899067e-02 2.36038417e-01
3.41350198e-01 2.14337394e-01 2.79817671e-01 1.18740821e+00
-3.29980105e-01 8.90569985e-02 4.14195895e-01 -5.78684509e-01
-1.22825646e+00 -7.27767766e-01 -4.49099205e-02 1.38841748e+00
1.29869264e-02 -5.21416605e-01 -6.16148531e-01 -1.04108381e+00
5.56818962e-01 5.00853419e-01 -3.32613081e-01 1.79569185e-01
-4.10468400e-01 -3.10190439e-01 4.47341800e-01 1.89449236e-01
9.11042154e-01 -3.63992453e-01 -1.38626039e-01 1.79158136e-01
-3.10816437e-01 -8.02971423e-01 -7.46500552e-01 -8.07596892e-02
-4.56407517e-01 -1.28615487e+00 -6.15669429e-01 -3.27672035e-01
3.80853564e-01 5.18792987e-01 8.30591738e-01 1.02900930e-01
-3.71930987e-01 1.52046785e-01 -1.81220978e-01 -3.42384964e-01
-1.75604939e-01 5.99768758e-01 -3.88297290e-01 3.25227082e-01
8.58227611e-01 -2.73232311e-01 -9.87536907e-01 8.53051007e-01
-1.07403851e+00 -3.34268153e-01 4.09169078e-01 6.92295015e-01
1.33811668e-01 1.06474228e-01 2.83902526e-01 -1.30916190e+00
8.78550589e-01 -6.89684212e-01 -9.76319790e-01 4.60125864e-01
-7.17147052e-01 -1.30146801e-01 3.11337054e-01 -5.24187803e-01
-7.09427655e-01 -1.52306601e-01 3.52725923e-01 -2.98014879e-01
1.56644315e-01 7.26058304e-01 1.01565465e-01 -1.25918031e-01
5.60542941e-01 6.97744489e-02 4.65059131e-02 -8.70480180e-01
3.50705028e-01 1.03608096e+00 -1.57728165e-01 -6.33878186e-02
1.14763486e+00 4.25638586e-01 -5.42418122e-01 -3.38671833e-01
-1.12615669e+00 -8.28100920e-01 -5.65911494e-02 1.33450516e-02
4.51591879e-01 -7.52961755e-01 -1.23291659e+00 3.20130229e-01
-7.26664066e-01 4.50334907e-01 -6.97909072e-02 4.23318416e-01
-1.69439688e-02 1.48103058e-01 -6.65636897e-01 -8.93720090e-01
-1.10701181e-01 -7.14807808e-01 3.85056257e-01 6.01531789e-02
-9.81824100e-02 -7.00058222e-01 7.50525743e-02 1.04787576e+00
8.83255839e-01 -4.05306101e-01 7.15991914e-01 -1.51827240e+00
-1.45330787e+00 -1.06317461e+00 -5.06933391e-01 4.73674655e-01
5.80347255e-02 -3.53984118e-01 -7.15421200e-01 -1.53970510e-01
-1.15379982e-01 -3.91137064e-01 6.68758690e-01 1.17625989e-01
1.41959882e+00 -1.05947089e+00 -6.20311260e-01 1.12726048e-01
1.03477061e+00 -2.45996997e-01 5.17951071e-01 6.87547088e-01
4.93798941e-01 5.72164714e-01 5.69754004e-01 5.90737164e-01
3.91575843e-01 4.94630665e-01 6.44640386e-01 5.81400543e-02
4.26467955e-01 -8.29852760e-01 -3.23334992e-01 3.67674798e-01
1.60389677e-01 -2.25621790e-01 -4.25976396e-01 3.78470480e-01
-1.98236132e+00 -1.16018319e+00 -9.06287730e-02 2.29137516e+00
7.79894412e-01 5.33747748e-02 4.21527803e-01 -7.74661228e-02
1.01978457e+00 2.22498685e-01 -6.69531107e-01 1.08992867e-01
2.19792709e-01 -1.97716445e-01 1.42806685e+00 -5.30495262e-03
-1.12591052e+00 5.75924873e-01 5.73905563e+00 8.11616004e-01
-9.10789609e-01 1.93772435e-01 7.28612721e-01 -2.50742912e-01
-3.82368445e-01 1.17245875e-01 -1.03434741e+00 9.20652628e-01
7.21717656e-01 -4.56559271e-01 6.06385052e-01 1.26783514e+00
1.04724526e-01 3.26347560e-01 -1.05036402e+00 7.34318495e-01
-3.25222939e-01 -1.82315779e+00 -3.54786694e-01 7.80242085e-01
6.37659609e-01 3.91914040e-01 3.22603196e-01 4.04598534e-01
7.17625320e-01 -5.75833142e-01 5.14361799e-01 2.19244137e-01
1.18380204e-01 -2.70220071e-01 4.13266391e-01 6.46279812e-01
-8.12090039e-01 -4.18993801e-01 -8.75554830e-02 1.94435954e-01
-5.34492433e-02 6.38711870e-01 -8.67827952e-01 1.97288513e-01
5.53065479e-01 5.37194073e-01 -7.92145312e-01 1.41951180e+00
1.61552534e-01 9.10159349e-01 -5.31446755e-01 -6.70145273e-01
4.70163494e-01 -4.75813780e-04 3.21284205e-01 3.46454680e-01
1.27097428e-01 -4.18903619e-01 2.88012832e-01 1.17497444e+00
-6.22199476e-01 2.88618773e-01 -3.57325733e-01 -6.28788173e-01
6.09737635e-01 1.22574854e+00 -1.62536591e-01 -3.34518880e-01
-7.42050171e-01 4.86092001e-01 3.37174684e-01 2.53912717e-01
-8.88230145e-01 -3.95202637e-01 5.47905624e-01 9.00443137e-01
5.11382520e-01 -1.12149138e-02 7.47477487e-02 -1.46674562e+00
2.92439163e-01 -9.20776248e-01 6.51126564e-01 -4.17301327e-01
-1.87025082e+00 3.99412289e-02 -3.13990474e-01 -1.27342474e+00
6.05751537e-02 -3.01855505e-01 -3.85875493e-01 7.68037975e-01
-1.58582056e+00 -1.09961593e+00 -1.11885704e-02 6.78705513e-01
9.33785588e-02 -6.46150187e-02 2.18212292e-01 8.70729089e-01
-5.24118721e-01 5.97132385e-01 4.28276718e-01 4.99179214e-01
5.60853243e-01 -8.89951289e-01 2.77446419e-01 4.31720853e-01
2.83280015e-01 1.07283366e+00 4.79122937e-01 -8.70647252e-01
-1.46997070e+00 -1.07905293e+00 1.13412392e+00 -3.85285169e-01
1.38775170e+00 -6.48731172e-01 -8.67852688e-01 1.02393055e+00
2.59877648e-02 1.88773096e-01 8.94070327e-01 6.88252032e-01
-6.61905587e-01 -4.19793755e-01 -1.20905936e+00 2.96991765e-01
6.02297008e-01 -5.51100016e-01 -4.36625719e-01 8.27873170e-01
3.90914589e-01 9.97962281e-02 -6.26054406e-01 -3.00831079e-01
7.04435408e-01 -7.84550309e-01 6.72369897e-01 -1.18313992e+00
1.47789061e-01 2.34479140e-02 -8.11493322e-02 -9.24805164e-01
-2.66117960e-01 -6.52476192e-01 -2.19312534e-01 1.39068890e+00
5.85697174e-01 -9.67255116e-01 1.41280794e+00 1.38188434e+00
5.36641479e-01 -2.12268427e-01 -1.09753895e+00 -9.00341034e-01
-1.66759640e-01 -2.79899150e-01 1.06703520e+00 9.43860769e-01
-3.28564905e-02 1.70502171e-01 -6.25548661e-01 1.39527349e-03
6.47734404e-01 3.79299819e-01 1.06408334e+00 -1.33181548e+00
-2.52303928e-01 -3.21490914e-01 -2.13416368e-01 -1.40677094e+00
-1.80178538e-01 -9.64074612e-01 -4.25272286e-01 -1.04034030e+00
3.13705057e-01 -8.52056265e-01 -4.65532720e-01 5.82148850e-01
2.61281073e-01 1.10490881e-02 1.57923669e-01 6.61919057e-01
-1.06551123e+00 1.95433348e-01 1.09353781e+00 -4.66434836e-01
-2.22473189e-01 6.36335731e-01 -1.01424992e+00 4.90616947e-01
6.29011273e-01 -6.27838790e-01 -7.08374567e-03 -2.18080625e-01
6.08026683e-01 -1.71174556e-01 6.13179743e-01 -5.02098322e-01
5.71597159e-01 -1.70485184e-01 1.33474290e-01 -9.78890657e-01
1.85329337e-02 -1.26030672e+00 1.01574481e-01 2.27369174e-01
-9.55703735e-01 -4.07610685e-01 -7.03044295e-01 9.57113624e-01
-2.56874263e-01 -2.25972101e-01 3.00518334e-01 -2.69629598e-01
2.25608330e-02 5.17057121e-01 -3.08389276e-01 7.51271620e-02
9.51740205e-01 2.19800204e-01 -7.72915006e-01 -5.18952012e-01
-5.71578860e-01 4.73456532e-01 9.82103273e-02 4.36746299e-01
2.20659263e-02 -1.25608742e+00 -3.98382843e-01 3.10625821e-01
2.33348519e-01 -5.47328293e-01 5.16550243e-02 6.51736379e-01
-4.05505747e-02 7.86190987e-01 1.78306609e-01 -3.58659446e-01
-1.26105034e+00 5.98487198e-01 8.28780350e-04 -5.04079700e-01
-2.33652055e-01 6.09277487e-01 -1.64413512e-01 -2.77228892e-01
2.46882454e-01 -1.29247531e-01 3.50274175e-01 2.29022607e-01
6.17627800e-01 6.27491593e-01 2.29415953e-01 -3.58144753e-02
-1.19201936e-01 -3.85162681e-01 -9.45202887e-01 1.09413952e-01
1.31967425e+00 -1.89553529e-01 -1.33535773e-01 6.65501505e-02
1.51514292e+00 2.42702842e-01 -7.33466685e-01 -5.70074677e-01
1.63481921e-01 -1.22132790e+00 2.59435982e-01 -9.60418165e-01
-1.41001225e+00 2.45481521e-01 3.08147967e-01 1.16694200e+00
5.14774263e-01 2.28877246e-01 1.14456964e+00 4.75064903e-01
2.90531933e-01 -1.40664804e+00 -1.97172351e-02 -5.58760613e-02
2.40149751e-01 -1.33692026e+00 2.75767565e-01 -4.03227091e-01
-2.86321878e-01 5.12935519e-01 4.26404029e-01 1.67953165e-03
8.73441160e-01 -3.89753670e-01 -2.94851094e-01 -3.47064137e-01
-6.67530119e-01 2.05397263e-01 2.87646085e-01 -9.52442959e-02
7.23354379e-03 1.56461820e-01 -5.04311502e-01 7.64668107e-01
-5.32361530e-02 1.24656789e-01 2.20765844e-01 1.21193147e+00
-4.07566965e-01 -1.36086857e+00 -1.76802978e-01 1.34787810e+00
-8.62121463e-01 -6.50099665e-02 -5.56494296e-01 7.95279264e-01
-1.98664203e-01 7.49079049e-01 -7.73410276e-02 -2.85694629e-01
2.04751357e-01 4.94148843e-02 -3.55311424e-01 -3.07743877e-01
-7.80529916e-01 1.16639458e-01 2.45210975e-01 -5.68142951e-01
-2.61389494e-01 -6.35589361e-01 -2.95460492e-01 -6.74537003e-01
-8.74535024e-01 5.68561375e-01 1.09569240e+00 5.99998951e-01
6.57767713e-01 -5.28329350e-02 1.24254394e+00 2.06186429e-01
-1.01376975e+00 -5.62530100e-01 -8.92382979e-01 5.35960078e-01
3.63238305e-01 -3.83962482e-01 -7.70503938e-01 -4.67410982e-01]
|
[7.732699394226074, 9.775704383850098]
|
947cb963-1de1-47e0-b123-ffb65d1f9584
|
generalized-feedback-loop-for-joint-hand
|
1903.10883
| null |
http://arxiv.org/abs/1903.10883v1
|
http://arxiv.org/pdf/1903.10883v1.pdf
|
Generalized Feedback Loop for Joint Hand-Object Pose Estimation
|
We propose an approach to estimating the 3D pose of a hand, possibly handling
an object, given a depth image. We show that we can correct the mistakes made
by a Convolutional Neural Network trained to predict an estimate of the 3D pose
by using a feedback loop. The components of this feedback loop are also Deep
Networks, optimized using training data. This approach can be generalized to a
hand interacting with an object. Therefore, we jointly estimate the 3D pose of
the hand and the 3D pose of the object. Our approach performs en-par with
state-of-the-art methods for 3D hand pose estimation, and outperforms
state-of-the-art methods for joint hand-object pose estimation when using depth
images only. Also, our approach is efficient as our implementation runs in
real-time on a single GPU.
|
['Markus Oberweger', 'Vincent Lepetit', 'Paul Wohlhart']
|
2019-03-25
| null | null | null | null |
['hand-object-pose']
|
['computer-vision']
|
[-2.49702364e-01 3.30820382e-02 9.73639917e-03 -7.90815204e-02
-4.88826156e-01 -6.45046234e-01 1.90864027e-01 -1.61959410e-01
-5.95434189e-01 1.82081923e-01 4.99018058e-02 -3.94899165e-03
1.69765398e-01 -5.27312398e-01 -1.07744586e+00 -3.97676677e-01
4.27644141e-02 1.29184055e+00 3.97666484e-01 1.28649607e-01
4.47049499e-01 1.30566013e+00 -1.37229645e+00 3.47108692e-02
-8.93185362e-02 8.56162548e-01 3.94053787e-01 1.27041936e+00
1.64297312e-01 4.33461726e-01 -5.29345274e-01 -1.96739420e-01
4.87182438e-01 2.33312532e-01 -1.17664063e+00 4.57676083e-01
6.83911860e-01 -1.16999352e+00 -5.41685760e-01 5.35290897e-01
8.94719005e-01 -1.08835086e-01 6.28123641e-01 -9.79902208e-01
1.17380850e-01 5.90939447e-02 -5.09010732e-01 -2.00443447e-01
6.20449543e-01 2.37083927e-01 7.22208917e-01 -1.05566823e+00
9.00117755e-01 1.45239508e+00 6.98566198e-01 4.96525049e-01
-1.00345302e+00 -2.08020091e-01 3.33529651e-01 4.35650311e-02
-1.45313549e+00 -6.07774854e-02 5.89471579e-01 -7.93124676e-01
1.27920294e+00 -9.39474702e-02 9.80433226e-01 8.58443856e-01
6.88589290e-02 1.14998353e+00 5.84920108e-01 -7.31719255e-01
4.37133992e-03 -2.15322286e-01 -1.96637306e-02 9.08063650e-01
3.98430834e-03 1.32106962e-02 -5.91792405e-01 -2.59128571e-01
1.45667493e+00 7.44489208e-02 -1.89591393e-01 -6.52497351e-01
-1.12606359e+00 3.11055988e-01 4.02758867e-01 -7.60191381e-02
-6.26272380e-01 4.79790986e-01 2.36471519e-01 -1.12084679e-01
3.13670397e-01 3.83003168e-02 -9.14371252e-01 -2.57460147e-01
-6.85634375e-01 8.74112666e-01 1.02028596e+00 9.45515811e-01
4.28685635e-01 -5.35308897e-01 -2.40377635e-01 2.16730550e-01
5.34537673e-01 3.18981051e-01 -2.62205601e-01 -1.19901037e+00
3.57488692e-01 4.57954496e-01 4.72533762e-01 -4.03815061e-01
-6.56392872e-01 -2.61642724e-01 -1.33533388e-01 9.07110214e-01
8.54673684e-01 -1.49303183e-01 -9.58887994e-01 1.38282144e+00
7.25245833e-01 -2.84682870e-01 -5.56099772e-01 1.14181280e+00
4.39031094e-01 1.26658097e-01 -4.15992111e-01 2.91170061e-01
1.34484065e+00 -1.04516578e+00 -3.76878768e-01 -3.22869182e-01
1.70628116e-01 -9.16155815e-01 6.90374851e-01 6.82452381e-01
-1.31937528e+00 -5.20434976e-01 -6.76013589e-01 -2.14760721e-01
2.83793900e-02 4.20174539e-01 4.98308480e-01 4.27341223e-01
-9.66968775e-01 9.51403618e-01 -1.30863357e+00 -2.68160075e-01
4.03016746e-01 7.51075447e-01 -3.61965567e-01 2.03968987e-01
-3.49706769e-01 1.21334374e+00 3.94457638e-01 3.31012249e-01
-1.00805378e+00 -4.40904230e-01 -4.97099876e-01 -1.90091744e-01
4.48932141e-01 -8.78344715e-01 1.76434302e+00 -4.58578259e-01
-1.72777009e+00 1.08216715e+00 -3.62476170e-01 1.39309958e-01
9.32644486e-01 -8.33458662e-01 6.93111837e-01 2.41990745e-01
-1.36151746e-01 6.96035683e-01 1.04183531e+00 -1.30183136e+00
-3.08105409e-01 -9.37741458e-01 2.28232816e-01 1.57504126e-01
2.12607712e-01 1.15747988e-01 -8.05570006e-01 -3.91575545e-01
2.97340810e-01 -1.10110545e+00 -1.69380933e-01 8.25817525e-01
-5.54107904e-01 -4.16481912e-01 8.00273478e-01 -7.88131416e-01
5.02528489e-01 -1.71076262e+00 4.74506646e-01 2.42554605e-01
2.20011905e-01 2.91338176e-01 8.26469064e-02 2.40935788e-01
1.37370989e-01 -3.65638018e-01 1.40029743e-01 -7.43152857e-01
-3.22230207e-03 9.27079411e-04 -1.80403590e-02 5.41893303e-01
1.27328426e-01 8.12926710e-01 -7.62146711e-01 -3.90002251e-01
2.24058807e-01 9.31378365e-01 -7.12122917e-01 7.14193165e-01
-5.10408401e-01 6.88947618e-01 -4.82494950e-01 4.59002763e-01
6.72463119e-01 -3.72347832e-01 3.00996810e-01 -2.92137414e-01
-2.97155064e-02 3.77788275e-01 -1.50838244e+00 1.87740803e+00
-4.38747406e-01 3.68969768e-01 2.59120494e-01 -3.58282566e-01
4.70750481e-01 6.56638563e-01 2.44652689e-01 4.54739481e-01
5.91975212e-01 3.62474948e-01 -1.71709269e-01 -5.05956829e-01
9.29090306e-02 9.76312980e-02 6.33576870e-01 9.78301287e-01
3.20575863e-01 -3.29896986e-01 -2.33542100e-01 -1.39609814e-01
7.82778203e-01 7.14500606e-01 2.67772734e-01 1.44826263e-01
1.82398438e-01 -2.85332024e-01 -2.57422000e-01 5.65306604e-01
6.24803863e-02 7.51087487e-01 5.18495023e-01 -7.56695211e-01
-1.24561548e+00 -9.11297679e-01 5.51283285e-02 9.70691025e-01
-1.36977971e-01 -3.00126165e-01 -9.68804955e-01 -7.08234012e-01
3.28301281e-01 -1.78066511e-02 -7.13886619e-01 6.43877387e-01
-9.51640427e-01 5.06030433e-02 1.13170773e-01 1.11486781e+00
6.91643804e-02 -1.13998365e+00 -1.02713335e+00 2.91660696e-01
2.23203097e-02 -1.04747558e+00 -7.10313559e-01 2.67857432e-01
-1.14577246e+00 -1.33502185e+00 -9.52479124e-01 -8.51426542e-01
8.38693261e-01 1.27384504e-02 1.15447176e+00 2.54958868e-01
-5.12773633e-01 4.38349426e-01 2.02549081e-02 -5.74214339e-01
-1.47722587e-01 2.43733466e-01 1.28095046e-01 -4.21890587e-01
1.14511333e-01 -5.70143819e-01 -8.76309574e-01 3.04765791e-01
-3.25764954e-01 -2.21105456e-01 3.53325248e-01 5.87025702e-01
5.99960148e-01 -3.35970461e-01 -2.76265919e-01 -3.32756907e-01
1.69224307e-01 3.57899696e-01 -7.22113132e-01 7.97998682e-02
6.47296757e-03 2.68349022e-01 1.34473607e-01 -7.02349663e-01
-9.67191279e-01 9.05715227e-01 -4.41525728e-01 -6.68577433e-01
-3.44314605e-01 -2.15538070e-01 -2.52630170e-02 -3.52754354e-01
5.58293700e-01 -2.21819326e-01 1.20748855e-01 -8.63518417e-01
3.03755552e-01 7.09798336e-01 4.65825528e-01 -7.00387776e-01
5.56913018e-01 6.50783658e-01 2.69069284e-01 -4.04847801e-01
-9.72645402e-01 -4.37597036e-01 -1.47040856e+00 -3.11199367e-01
7.51832604e-01 -7.12947786e-01 -1.60976529e+00 9.71068621e-01
-1.98772120e+00 -5.20030856e-01 -4.12530974e-02 5.08637667e-01
-8.75250757e-01 1.97979063e-01 -6.38760626e-01 -1.09382570e+00
-4.62130457e-01 -1.24612749e+00 1.80836213e+00 -1.52223229e-01
-4.46053267e-01 -7.13546932e-01 -5.75435646e-02 2.21775994e-01
-4.94215228e-02 3.04689836e-02 5.28665066e-01 -1.99098617e-01
-8.30463707e-01 -5.23882389e-01 -1.30557671e-01 7.10581169e-02
-1.04169575e-02 -6.13693073e-02 -1.16503882e+00 -4.38069195e-01
-1.58188269e-01 -3.90227020e-01 4.22160506e-01 6.73870206e-01
1.38402152e+00 -2.57468432e-01 -4.57439750e-01 3.71001601e-01
1.06027091e+00 -1.99546054e-01 3.51950616e-01 2.57555962e-01
8.18888068e-01 6.73362255e-01 3.70066881e-01 6.50853515e-01
3.34205955e-01 9.05077577e-01 8.27149153e-01 2.87292689e-01
-1.57225162e-01 -1.57525256e-01 -1.30576417e-01 1.03134423e-01
-7.52185047e-01 -5.99639229e-02 -9.06813323e-01 3.82208169e-01
-1.66241384e+00 -4.38937992e-01 2.30104961e-02 2.13216400e+00
6.75653458e-01 1.96224779e-01 6.41822278e-01 3.79060894e-01
4.65429813e-01 -1.14076376e-01 -7.41836905e-01 3.38588730e-02
5.67115843e-01 5.09438217e-01 3.98346841e-01 7.71460831e-01
-1.00680268e+00 1.02699304e+00 7.34172726e+00 1.28314570e-01
-9.66151178e-01 5.79371229e-02 -3.89241502e-02 -2.11035833e-01
2.57143617e-01 -2.39022419e-01 -9.10720646e-01 -2.63718009e-01
3.87482569e-02 5.89093745e-01 5.27414560e-01 9.69781160e-01
-8.97329599e-02 -1.07515827e-01 -1.65309811e+00 9.54567313e-01
1.17953584e-01 -9.34843779e-01 -1.99184209e-01 1.31065443e-01
3.84026557e-01 -5.53340791e-03 -2.80358613e-01 -3.70518863e-01
1.03031874e-01 -7.89276242e-01 1.05163431e+00 4.86131430e-01
5.79144776e-01 -5.63037574e-01 6.40176773e-01 6.94716871e-01
-1.16109097e+00 1.33318335e-01 2.70569716e-02 -4.38364983e-01
2.74985671e-01 2.66366124e-01 -1.02785146e+00 -1.21235847e-01
8.55744660e-01 1.90089867e-01 -1.89570308e-01 9.83701289e-01
-6.23893082e-01 -6.49793968e-02 -4.11820799e-01 2.52575260e-02
-1.25819892e-01 4.46353287e-01 6.77016735e-01 9.26915050e-01
5.82674518e-02 2.69140840e-01 2.42879942e-01 7.90597022e-01
-6.35243719e-03 -2.27160200e-01 -2.35476241e-01 1.50802612e-01
2.07845852e-01 8.80907834e-01 -6.99921846e-01 -2.47581124e-01
1.02163166e-01 1.36153495e+00 5.54660738e-01 3.45872939e-02
-1.99719712e-01 -3.40666771e-01 6.17655993e-01 2.90150464e-01
6.70412064e-01 -6.39015675e-01 -3.89383197e-01 -9.89961147e-01
5.51492155e-01 -5.72546363e-01 -9.04814303e-02 -1.10601783e+00
-1.01503658e+00 5.60479641e-01 -1.83133796e-01 -9.23477352e-01
-4.25775230e-01 -1.27040005e+00 -3.45339328e-01 1.03454304e+00
-1.17673147e+00 -1.02930903e+00 -5.72449386e-01 4.93111461e-01
4.03028458e-01 1.83855310e-01 1.13269854e+00 -2.02350825e-01
5.55981509e-02 2.88200200e-01 -5.46616971e-01 3.19248319e-01
5.20394266e-01 -1.30242252e+00 8.43512297e-01 2.64649779e-01
1.14559688e-01 5.72558105e-01 5.67967534e-01 -5.52132845e-01
-1.58106065e+00 -4.61555988e-01 8.59821558e-01 -9.87702489e-01
3.00570637e-01 -5.24376154e-01 -6.00593090e-01 1.10488081e+00
-9.19987410e-02 2.30978712e-01 8.84295553e-02 3.23057026e-01
-5.03546119e-01 2.99797446e-01 -1.20329928e+00 3.54870975e-01
1.35731936e+00 -6.73080802e-01 -6.07940555e-01 7.68340886e-01
3.09897512e-01 -1.25985336e+00 -7.97510505e-01 5.37838088e-03
1.29519510e+00 -8.37365448e-01 1.24423969e+00 -7.20098078e-01
2.48484507e-01 -2.20826983e-01 1.31143928e-01 -1.03196371e+00
-2.28304923e-01 -5.57303190e-01 -6.26625538e-01 2.79791445e-01
-6.04337454e-02 -1.24314569e-01 1.27625811e+00 4.94900167e-01
3.68458539e-01 -8.36520851e-01 -8.84129763e-01 -5.82429230e-01
2.26158760e-02 -5.51174760e-01 4.92178977e-01 2.61205644e-03
-1.72674716e-01 1.31478831e-01 -3.11852336e-01 4.62666869e-01
8.42624426e-01 1.63905650e-01 1.07221568e+00 -1.47465420e+00
-5.17010391e-01 -3.48451763e-01 -6.07997656e-01 -1.55585217e+00
3.92473698e-01 -3.90358925e-01 2.14633152e-01 -1.56803501e+00
3.69002074e-01 -1.10730432e-01 3.83746862e-01 6.61286235e-01
-1.35797635e-01 3.14263821e-01 3.23747128e-01 2.25340545e-01
-9.44913328e-02 4.91616316e-02 1.60889876e+00 4.24500741e-02
-2.54701078e-01 3.68879616e-01 2.23907456e-02 1.01695263e+00
5.47192097e-01 -5.07247090e-01 2.96952307e-01 -7.76651561e-01
6.23818897e-02 2.90722072e-01 7.87247181e-01 -6.38732731e-01
1.84133410e-01 9.05833840e-02 7.13121951e-01 -1.06972170e+00
6.08582377e-01 -9.65618074e-01 -8.16835016e-02 7.47746646e-01
-1.94667503e-01 -1.88189045e-01 1.65027454e-01 1.58915177e-01
3.68945807e-01 -2.77752519e-01 6.76612377e-01 -3.87309790e-01
-1.36793673e-01 5.55388391e-01 -1.91521764e-01 -3.78186584e-01
7.07446575e-01 -3.21373492e-01 1.61436498e-01 -3.73832315e-01
-1.14849102e+00 -2.65660062e-02 3.13308567e-01 2.58850634e-01
5.07792830e-01 -1.18140733e+00 -6.52842999e-01 3.21381569e-01
-2.60792643e-01 5.39811850e-01 5.80179272e-03 5.12475848e-01
-8.17243874e-01 4.77795690e-01 -2.11416274e-01 -9.43281114e-01
-1.60018456e+00 3.61341059e-01 6.45315588e-01 -6.67341650e-02
-5.76737106e-01 1.13646853e+00 -8.15106407e-02 -7.45626271e-01
8.03763449e-01 -3.40896010e-01 3.32009882e-01 -2.40610778e-01
7.51643538e-01 5.48884153e-01 3.63427401e-01 -4.67370868e-01
-5.45069814e-01 1.06794250e+00 2.36374121e-02 -2.38525704e-01
1.48306811e+00 2.96720147e-01 -2.76063859e-01 2.68149287e-01
1.20724976e+00 -1.01691209e-01 -1.64705670e+00 -2.49574587e-01
-4.26362783e-01 -7.92247295e-01 -4.72135209e-02 -8.71299267e-01
-1.08189535e+00 1.15830684e+00 6.32952452e-01 -3.78854424e-01
6.78851962e-01 3.59065384e-01 5.52813113e-01 8.45308900e-01
5.89416146e-01 -9.02301729e-01 4.77970898e-01 7.01914072e-01
1.34453022e+00 -1.18502057e+00 1.43789634e-01 -5.83741963e-01
-1.46635413e-01 1.57070494e+00 5.99678636e-01 -3.76554430e-01
9.60316837e-01 5.72974086e-01 9.77845788e-02 -2.16382414e-01
-3.19022506e-01 -1.55839071e-01 5.15903890e-01 5.45174658e-01
4.93294477e-01 6.08237647e-02 2.95054972e-01 -4.79416549e-02
-1.42572090e-01 3.44943404e-01 5.50835207e-02 1.34249568e+00
-2.76883483e-01 -1.26654994e+00 -7.49657631e-01 8.47091153e-02
-4.58519340e-01 2.54819542e-01 -5.22081852e-01 6.86615109e-01
1.12801291e-01 3.45382333e-01 1.62849575e-01 -2.13207200e-01
7.28086770e-01 1.18495144e-01 1.46296787e+00 -7.31111050e-01
-6.13677084e-01 2.23586142e-01 -3.28432590e-01 -7.53261268e-01
-4.50388104e-01 -6.16991818e-01 -1.10608232e+00 -1.96202114e-01
-5.23343861e-01 -4.61830944e-01 8.06792140e-01 1.10515440e+00
1.81793317e-01 2.75249898e-01 3.77342403e-01 -1.98498392e+00
-8.25690508e-01 -1.02866578e+00 -6.66373014e-01 -1.27713576e-01
5.72172225e-01 -1.05651188e+00 -1.35384336e-01 -1.39411002e-01]
|
[6.567220211029053, -0.8325075507164001]
|
fe50083a-4dda-4978-b26c-fdef8739accb
|
large-scale-cloze-test-dataset-designed-by
| null | null |
https://openreview.net/forum?id=rJJzTyWCZ
|
https://openreview.net/pdf?id=rJJzTyWCZ
|
Large-scale Cloze Test Dataset Designed by Teachers
|
Cloze test is widely adopted in language exams to evaluate students' language proficiency. In this paper, we propose the first large-scale human-designed cloze test dataset CLOTH in which the questions were used in middle-school and high-school language exams. With the missing blanks carefully created by teachers and candidate choices purposely designed to be confusing, CLOTH requires a deeper language understanding and a wider attention span than previous automatically generated cloze datasets. We show humans outperform dedicated designed baseline models by a significant margin, even when the model is trained on sufficiently large external data. We investigate the source of the performance gap, trace model deficiencies to some distinct properties of CLOTH, and identify the limited ability of comprehending a long-term context to be the key bottleneck. In addition, we find that human-designed data leads to a larger gap between the model's performance and human performance when compared to automatically generated data.
|
['Qizhe Xie', 'Zihang Dai', 'Guokun Lai', 'Eduard Hovy']
|
2018-01-01
| null | null | null |
iclr-2018-1
|
['cloze-test']
|
['natural-language-processing']
|
[-2.60979056e-01 1.45986691e-01 -1.47545353e-01 -6.12611324e-02
-1.08626688e+00 -9.83504117e-01 2.55329967e-01 4.28648770e-01
-3.39933217e-01 6.17234111e-01 3.96764845e-01 -1.04336131e+00
-2.61513442e-01 -7.03098893e-01 -6.61663830e-01 8.16079080e-02
5.35231888e-01 1.52434334e-01 3.36326450e-01 -4.02273297e-01
3.84501666e-01 -2.18100771e-02 -1.59095204e+00 3.81947428e-01
1.55548227e+00 1.32429719e-01 3.52829456e-01 8.24757636e-01
-6.03696443e-02 1.14925754e+00 -9.05142307e-01 -5.81306040e-01
-1.25127465e-01 -8.44379246e-01 -9.25399244e-01 5.23720607e-02
1.09475422e+00 -1.99780375e-01 -9.82650965e-02 9.76535857e-01
5.65917671e-01 2.53673077e-01 3.47437143e-01 -6.18547976e-01
-1.27430105e+00 8.04827809e-01 -2.06536531e-01 3.93646896e-01
7.70466208e-01 5.57387352e-01 9.94708180e-01 -7.46054351e-01
6.84987068e-01 9.35515881e-01 6.56443059e-01 9.64011550e-01
-1.13298106e+00 -6.04358912e-01 2.78382957e-01 7.85814747e-02
-1.09255743e+00 -2.78357506e-01 4.88283604e-01 -7.20510900e-01
7.45537341e-01 3.11260641e-01 8.09366226e-01 1.16199231e+00
-9.65337828e-02 7.91650653e-01 1.48143923e+00 -8.15048635e-01
6.01071753e-02 4.64624584e-01 5.79790771e-01 7.43302703e-01
5.28892219e-01 -2.01772660e-01 -5.13674557e-01 6.77454099e-02
2.98586637e-01 -4.21572655e-01 -5.12824595e-01 -9.71355438e-02
-9.43645716e-01 8.04274559e-01 2.18558982e-01 2.52593040e-01
3.31553370e-01 -2.74494320e-01 1.06897831e-01 6.51226401e-01
-5.54806627e-02 1.29050291e+00 -4.13686365e-01 -5.26172280e-01
-9.43143010e-01 4.78245556e-01 7.66666532e-01 1.13420081e+00
4.92432192e-02 -1.53657943e-01 -5.24809062e-01 5.97873569e-01
1.13949172e-01 3.27979803e-01 8.98893058e-01 -6.50449574e-01
8.32815349e-01 8.23923588e-01 4.12150323e-02 -6.53908014e-01
-2.96183825e-02 -6.41588748e-01 -1.09044373e-01 4.77347635e-02
1.00862730e+00 -3.44105102e-02 -7.15099990e-01 1.70597696e+00
-4.95611168e-02 -8.53215903e-02 -9.36307162e-02 7.08537340e-01
1.28786218e+00 9.56837758e-02 3.76764685e-01 7.49069527e-02
1.42645454e+00 -1.11757290e+00 -8.83865654e-01 -5.23583531e-01
9.89108801e-01 -8.38085175e-01 2.10171700e+00 6.09948277e-01
-1.43998194e+00 -7.32472837e-01 -1.24697733e+00 -4.75285351e-01
-3.93143505e-01 -4.37326282e-02 4.64183778e-01 1.00750601e+00
-9.32129145e-01 5.20067751e-01 -3.96089792e-01 -1.24635950e-01
2.08404258e-01 -2.97868717e-02 -8.87207761e-02 -3.06414187e-01
-1.16659892e+00 8.26988995e-01 1.62175403e-03 -4.54316884e-01
-9.43402171e-01 -8.98862779e-01 -9.34601963e-01 1.79524571e-01
3.29442531e-01 -3.36631238e-01 1.58248663e+00 -4.43728507e-01
-1.40865910e+00 1.18046975e+00 -7.12952614e-02 1.03602588e-01
8.81126583e-01 -4.60720569e-01 -1.79653272e-01 -1.93791851e-01
2.15042070e-01 1.10574372e-01 1.77933022e-01 -8.91055584e-01
-4.30847943e-01 9.12024602e-02 2.69638542e-02 5.08968532e-02
-5.55259347e-01 9.69294161e-02 -4.23309594e-01 -5.23445427e-01
1.72397107e-01 -6.85641110e-01 6.28485084e-02 -6.82438612e-01
-2.99264640e-01 -6.91828072e-01 1.94707066e-01 -6.64943159e-01
2.10103703e+00 -1.95575964e+00 -2.24738240e-01 7.48026222e-02
6.36519551e-01 4.56206918e-01 -2.90408373e-01 3.82884800e-01
-9.80163552e-03 6.74870074e-01 4.23853546e-01 -1.99766830e-01
2.32553467e-01 -4.27300721e-01 -1.59636751e-01 2.76055872e-01
-5.55359665e-03 9.90885079e-01 -1.46045280e+00 -4.22783673e-01
-1.20346934e-01 -2.15705633e-01 -7.43877292e-01 5.16608179e-01
-2.44884029e-01 2.58031040e-01 -3.89950395e-01 5.91209650e-01
3.76325220e-01 -3.15484226e-01 -8.01976323e-02 7.11127639e-01
-5.14119938e-02 1.06916797e+00 -7.81478643e-01 1.56929767e+00
-4.73600268e-01 9.22996819e-01 -3.47496003e-01 -7.31767192e-02
7.37626076e-01 3.99511099e-01 -3.99897128e-01 -7.71482408e-01
1.45710126e-01 1.43814266e-01 4.54082340e-01 -8.83443415e-01
5.48486352e-01 1.52731752e-02 -6.97722584e-02 4.94153827e-01
-1.73839301e-01 -5.11459827e-01 6.16884708e-01 3.87733936e-01
1.33576763e+00 -6.27167448e-02 3.21607530e-01 -4.69051868e-01
3.62079233e-01 1.53788537e-01 2.99593598e-01 1.20156550e+00
-4.25859898e-01 5.18598974e-01 4.78589833e-01 -6.85891658e-02
-9.02864516e-01 -8.38831961e-01 -1.02788039e-01 1.51614523e+00
-1.96105734e-01 -8.98432255e-01 -9.40702260e-01 -9.23985481e-01
-8.54655802e-02 1.09701645e+00 -5.53912163e-01 -3.74670029e-01
-5.22041142e-01 -1.82258159e-01 5.77300847e-01 5.06162524e-01
2.17582211e-01 -1.05880606e+00 -5.46957493e-01 5.09930253e-02
-1.10194229e-01 -1.02913070e+00 -6.12160623e-01 1.56844892e-02
-6.67638898e-01 -1.00631630e+00 -3.29187751e-01 -1.12113738e+00
6.95464909e-01 3.50217760e-01 1.71794522e+00 7.48224258e-01
-1.87184792e-02 2.49744073e-01 -3.48243684e-01 -5.44766247e-01
-4.61036265e-01 2.67719924e-01 -2.28667527e-01 -9.95687008e-01
1.12646258e+00 -1.75254330e-01 -4.23801959e-01 1.45947114e-01
-6.76384985e-01 1.75095335e-01 4.74767864e-01 7.67187119e-01
-3.00943945e-02 -1.60359889e-01 4.29707617e-01 -1.13057745e+00
1.42222667e+00 -4.39338714e-01 -5.61513126e-01 3.96599799e-01
-7.01588392e-01 -1.01780258e-01 6.13986254e-01 -8.11633646e-01
-6.79216683e-01 -5.84508479e-01 -1.65902488e-02 -8.10719207e-02
-2.30717048e-01 5.70005536e-01 -2.22933978e-01 1.93536896e-02
1.12511849e+00 -1.07936434e-01 -4.89430189e-01 -6.67243779e-01
1.24505505e-01 6.38450742e-01 3.93322587e-01 -1.18143153e+00
8.65292847e-01 -4.27062839e-01 -5.16633034e-01 -3.72621030e-01
-1.42015541e+00 -4.29974377e-01 -5.07328272e-01 -1.66232809e-02
7.07150340e-01 -1.19747472e+00 -1.05613160e+00 4.85251285e-02
-9.90561366e-01 -7.71335959e-01 -1.02748789e-01 5.49133182e-01
-4.96840030e-02 -8.68509859e-02 -1.05384970e+00 -5.93761861e-01
1.17007948e-01 -1.29924500e+00 5.27514637e-01 4.62838322e-01
-6.73385859e-01 -1.04301667e+00 2.14367300e-01 8.80307555e-01
1.56238794e-01 -9.68284626e-03 1.10275459e+00 -9.15466011e-01
-4.82967913e-01 -1.71274602e-01 -2.99407952e-02 2.45560378e-01
-8.48722085e-02 -9.64982715e-03 -9.62989271e-01 -3.60065669e-01
-1.54679477e-01 -8.51636291e-01 6.78530633e-01 -1.31659448e-01
1.05743897e+00 -1.74193010e-01 7.05271438e-02 5.38608551e-01
1.22450793e+00 -1.06928810e-01 3.06516171e-01 6.56189322e-01
7.62567341e-01 6.20473325e-01 2.97047943e-01 -1.29783750e-01
4.54909682e-01 2.42097989e-01 -2.24932268e-01 2.68361121e-01
-4.10934210e-01 -9.31613982e-01 5.08372962e-01 1.30970573e+00
2.61055022e-01 -2.87702143e-01 -1.24294102e+00 9.82473493e-01
-1.28056073e+00 -7.40628302e-01 -4.73756909e-01 2.32885170e+00
1.33118141e+00 4.54797417e-01 -4.97190421e-03 4.86098789e-02
3.91334802e-01 -3.12069476e-01 -1.08085737e-01 -5.58289111e-01
7.65453875e-02 4.28023934e-01 2.47307718e-01 9.03726339e-01
-6.98155820e-01 1.23550820e+00 7.07662582e+00 7.70555258e-01
-7.93845773e-01 8.04530010e-02 3.47836137e-01 -2.31930807e-01
-6.99641943e-01 -7.43403584e-02 -1.05870974e+00 4.05800968e-01
9.50436711e-01 -2.04093918e-01 2.81142116e-01 7.34948039e-01
2.49369115e-01 1.12719245e-01 -1.28807390e+00 5.58429003e-01
1.21333562e-01 -8.44993174e-01 -1.52228788e-01 8.77024308e-02
1.18135905e+00 -4.86949295e-01 3.06335598e-01 6.58651352e-01
7.45423794e-01 -1.41993570e+00 9.83867347e-01 -8.04981869e-03
7.32901096e-01 -4.34216142e-01 5.02925992e-01 4.31849778e-01
-6.23676717e-01 -7.49692544e-02 -3.61065567e-01 -6.71286702e-01
-5.95347881e-01 2.61609465e-01 -9.62211728e-01 -1.84743464e-01
3.97703290e-01 2.19179541e-01 -1.41892147e+00 1.08013999e+00
-8.62777770e-01 1.23689425e+00 2.54865527e-01 -5.51627815e-01
4.33564365e-01 1.11240223e-01 2.38149598e-01 1.18246901e+00
3.10467035e-01 6.10012747e-02 2.61700153e-01 1.06869721e+00
-2.53696173e-01 2.60098159e-01 -5.26139915e-01 -1.96211755e-01
7.77356446e-01 9.76933420e-01 -3.40481877e-01 -2.89678752e-01
-6.49754107e-01 6.01252019e-01 6.73639059e-01 4.11694229e-01
-5.51129997e-01 -4.58514780e-01 4.42057610e-01 5.80715954e-01
-5.97744957e-02 -5.97558692e-02 -6.68938220e-01 -1.24068499e+00
-6.58355327e-03 -1.37315309e+00 2.79904217e-01 -8.18035305e-01
-1.35249174e+00 3.47536802e-01 -3.42895627e-01 -1.13319719e+00
-1.46910325e-01 -7.97707200e-01 -9.02218282e-01 1.18080211e+00
-1.41041780e+00 -7.48028278e-01 -5.96584797e-01 3.18649799e-01
5.70894539e-01 -2.00558137e-02 5.41585743e-01 1.41173735e-01
-6.95410848e-01 1.19289160e+00 -9.71029773e-02 4.70480472e-01
1.19003761e+00 -1.79356778e+00 4.26578492e-01 1.02492833e+00
-9.25091058e-02 1.17216885e+00 9.61233377e-01 -8.86745334e-01
-9.88902569e-01 -6.67039931e-01 1.42729974e+00 -1.30662906e+00
9.08687055e-01 -5.39562106e-01 -1.15071499e+00 6.52001262e-01
4.29671824e-01 -1.60801142e-01 1.04484665e+00 4.29363340e-01
-5.03721356e-01 3.82640779e-01 -8.16063881e-01 8.21199119e-01
1.15251780e+00 -6.45240307e-01 -1.20038807e+00 2.17096373e-01
8.84964406e-01 -8.09785783e-01 -6.83548689e-01 -2.03321017e-02
3.19338024e-01 -7.80899942e-01 6.14952922e-01 -1.20758688e+00
1.02360570e+00 -4.02045771e-02 3.86456370e-01 -1.35949218e+00
-5.52029550e-01 -7.95594871e-01 4.41580340e-02 1.32619226e+00
5.60645044e-01 -9.97845754e-02 8.28964531e-01 1.05040514e+00
-2.35062674e-01 -6.53411686e-01 -3.22661877e-01 -6.42154336e-01
7.23280013e-01 -3.21325749e-01 6.58112884e-01 1.23443913e+00
4.75977659e-01 4.86920387e-01 9.86334160e-02 -1.44391179e-01
3.98462355e-01 -6.39305636e-02 7.94603229e-01 -1.12610316e+00
-1.93146661e-01 -6.77037716e-01 9.85751674e-02 -1.10034955e+00
9.43965018e-02 -7.71972835e-01 6.39775768e-02 -1.18878758e+00
3.61505806e-01 -4.29982096e-01 -1.82890311e-01 3.31277162e-01
-1.00800657e+00 1.92507617e-02 3.06285203e-01 -3.07514459e-01
-9.00918007e-01 -3.31631862e-02 1.78677905e+00 1.20015576e-01
-4.04787928e-01 -8.46162438e-02 -1.34998369e+00 7.75145173e-01
6.25062466e-01 -1.69959903e-01 -5.97696066e-01 -6.55334711e-01
5.15658081e-01 -2.78745562e-01 5.94628751e-02 -9.61211026e-01
4.26410258e-01 -5.16255319e-01 4.96638179e-01 2.51281280e-02
-4.91593778e-01 -2.93995380e-01 -6.92426383e-01 1.09868057e-01
-8.20494473e-01 3.85767728e-01 4.93129998e-01 -5.40950522e-02
-2.47592181e-01 -5.77652395e-01 5.74291050e-01 -5.56742668e-01
-3.37565422e-01 -2.80610234e-01 -3.47600341e-01 7.23090589e-01
6.95661426e-01 -1.23306014e-01 -6.75585032e-01 -4.41454679e-01
-4.73530918e-01 3.56109947e-01 5.62864363e-01 5.40990949e-01
3.15415561e-01 -1.25210238e+00 -8.14891994e-01 1.15633562e-01
3.63468796e-01 9.96686965e-02 1.33047104e-01 4.09274548e-01
-7.10879862e-01 7.87439883e-01 1.54066309e-01 -1.23547591e-01
-1.30501795e+00 5.38034201e-01 1.15234070e-01 -6.01885676e-01
-4.23051149e-01 1.11995006e+00 -9.80417132e-02 -3.69410008e-01
4.57424939e-01 -6.16526663e-01 -2.26331607e-01 -1.06165417e-01
1.01836550e+00 3.50626439e-01 3.20022792e-01 -1.77043408e-01
2.69745260e-01 3.83044004e-01 -1.71111375e-01 5.84884770e-02
9.10535216e-01 -6.73299730e-02 3.19557637e-01 4.09024715e-01
7.13614702e-01 9.95688319e-01 -9.69209313e-01 -2.19878018e-01
3.26719701e-01 -5.61192989e-01 1.08167611e-03 -1.26568449e+00
-4.77179945e-01 9.63331461e-01 1.76041350e-01 1.88592374e-01
7.37691104e-01 -8.07468146e-02 5.70167124e-01 3.49515527e-01
9.26315144e-04 -1.16731703e+00 2.45536566e-01 6.59619570e-01
5.58186173e-01 -1.44121373e+00 2.43265871e-02 -4.10132140e-01
-5.30969799e-01 8.34587395e-01 1.51698291e+00 1.41156495e-01
6.25257269e-02 -1.24581354e-02 3.12728047e-01 -1.01199344e-01
-1.02171552e+00 -2.30187103e-01 7.54722297e-01 3.89463305e-01
1.20818079e+00 -5.68312174e-03 -4.71805036e-01 1.18353689e+00
-8.23929489e-01 -1.31688178e-01 8.08194339e-01 9.52755153e-01
-5.39399862e-01 -1.05066323e+00 -5.74277937e-01 3.02717716e-01
-5.97570419e-01 -5.00447214e-01 -7.34726369e-01 8.84651780e-01
2.46960327e-01 1.09908605e+00 -1.00567147e-01 -3.19550037e-01
4.79589075e-01 4.46530223e-01 5.33657670e-01 -1.14500391e+00
-1.11131704e+00 -2.82739669e-01 -1.03751257e-01 -2.64973342e-01
3.66798133e-01 -4.16766584e-01 -9.17584598e-01 -4.80127513e-01
-3.24825913e-01 4.21824753e-01 1.27415448e-01 9.64692950e-01
-3.96964848e-02 7.47039497e-01 4.26752008e-02 1.62087291e-01
-8.87269437e-01 -1.41537035e+00 -2.86101967e-01 8.53573918e-01
4.99877512e-01 -3.53390276e-01 -6.16760552e-01 -3.01391780e-02]
|
[10.172881126403809, 7.6978278160095215]
|
30426f57-8458-4d51-b9fa-db84aed0cd97
|
ernie-gen-an-enhanced-multi-flow-pre-training
|
2001.11314
| null |
https://arxiv.org/abs/2001.11314v3
|
https://arxiv.org/pdf/2001.11314v3.pdf
|
ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation
|
Current pre-training works in natural language generation pay little attention to the problem of exposure bias on downstream tasks. To address this issue, we propose an enhanced multi-flow sequence to sequence pre-training and fine-tuning framework named ERNIE-GEN, which bridges the discrepancy between training and inference with an infilling generation mechanism and a noise-aware generation method. To make generation closer to human writing patterns, this framework introduces a span-by-span generation flow that trains the model to predict semantically-complete spans consecutively rather than predicting word by word. Unlike existing pre-training methods, ERNIE-GEN incorporates multi-granularity target sampling to construct pre-training data, which enhances the correlation between encoder and decoder. Experimental results demonstrate that ERNIE-GEN achieves state-of-the-art results with a much smaller amount of pre-training data and parameters on a range of language generation tasks, including abstractive summarization (Gigaword and CNN/DailyMail), question generation (SQuAD), dialogue generation (Persona-Chat) and generative question answering (CoQA).
|
['Yu Sun', 'Hua Wu', 'Hao Tian', 'Yukun Li', 'Han Zhang', 'Dongling Xiao', 'Haifeng Wang']
|
2020-01-26
| null | null | null | null |
['generative-question-answering']
|
['natural-language-processing']
|
[ 4.57029015e-01 7.29945242e-01 1.40828833e-01 -4.07088906e-01
-1.15985894e+00 -5.18935442e-01 9.88461852e-01 -1.19091921e-01
-2.55571246e-01 1.21037745e+00 9.37767565e-01 -4.06364352e-01
3.99756640e-01 -1.12730110e+00 -6.45179629e-01 -7.08751231e-02
4.78667736e-01 7.04148948e-01 -6.33010417e-02 -9.12354529e-01
2.60108262e-01 -4.08623159e-01 -1.05510223e+00 7.86877751e-01
1.27465725e+00 5.08397937e-01 3.30846488e-01 1.11774182e+00
-2.67328858e-01 9.68064249e-01 -1.00836790e+00 -9.05993819e-01
-8.55836496e-02 -1.23121560e+00 -1.22390282e+00 -1.43887565e-01
3.83170813e-01 -5.52172542e-01 -2.87505776e-01 5.59365630e-01
9.76513326e-01 2.58078307e-01 6.66675448e-01 -8.76365781e-01
-1.24814200e+00 1.30203307e+00 -1.20471060e-01 2.30769873e-01
5.38504779e-01 6.17919505e-01 1.15988326e+00 -8.72657359e-01
6.54842675e-01 1.51754773e+00 6.10927582e-01 1.09048665e+00
-1.14664149e+00 -4.16914523e-01 2.49474738e-02 -1.04238585e-01
-8.16932321e-01 -6.51758611e-01 4.53285605e-01 -9.02785584e-02
1.27273238e+00 2.80784339e-01 5.06042004e-01 1.67919052e+00
3.20893642e-03 1.01433933e+00 5.80798447e-01 -5.02308547e-01
4.99729477e-02 -3.65891725e-01 -8.24192092e-02 6.05347216e-01
-1.83512911e-03 2.86772996e-02 -6.89985216e-01 -3.51109579e-02
5.45545220e-01 -6.44051492e-01 -3.62799674e-01 6.42614424e-01
-1.43554389e+00 1.06612813e+00 2.31963605e-01 9.59618166e-02
-3.56665581e-01 2.65948951e-01 5.81241786e-01 3.37608367e-01
7.19780982e-01 9.73480701e-01 -2.27505952e-01 -4.96510535e-01
-1.10266900e+00 7.11070359e-01 1.00599337e+00 1.39834058e+00
5.08317411e-01 4.34640974e-01 -1.28253388e+00 9.21984375e-01
-2.35887561e-02 2.98208445e-01 9.68683898e-01 -8.00881684e-01
1.02098930e+00 3.75215352e-01 -9.61863324e-02 -2.53707051e-01
-8.76363181e-03 -4.20406818e-01 -1.03709280e+00 -4.63362157e-01
3.68780375e-01 -6.87917709e-01 -9.04507339e-01 1.83515072e+00
5.50850369e-02 -2.05772474e-01 3.16395700e-01 6.85118139e-01
1.00704706e+00 1.00036836e+00 4.21333984e-02 -1.21210016e-01
1.43507063e+00 -1.61317611e+00 -7.28194237e-01 -4.16578263e-01
6.39895558e-01 -6.40459538e-01 1.38315952e+00 4.76323292e-02
-1.57184374e+00 -8.54941428e-01 -8.66071284e-01 -5.44352472e-01
-1.09461203e-01 3.95328775e-02 4.10469860e-01 3.48080695e-01
-1.04489434e+00 5.97936571e-01 -3.71554375e-01 -1.90470234e-01
2.11130783e-01 -3.59798849e-01 8.82972628e-02 5.81430793e-02
-1.60922337e+00 9.80125010e-01 7.30500698e-01 3.09997387e-02
-8.55726600e-01 -9.88565862e-01 -9.41157639e-01 1.27558842e-01
1.33740574e-01 -1.49770927e+00 1.80677497e+00 -7.38541186e-01
-1.90736306e+00 5.69545805e-01 -1.93021327e-01 -8.23074341e-01
6.80757225e-01 -4.20975775e-01 -2.09750146e-01 -8.45787898e-02
1.56551912e-01 9.98423040e-01 5.40935636e-01 -8.21455836e-01
-3.82889599e-01 1.50915146e-01 -3.02945096e-02 4.10518110e-01
-1.69865042e-02 -1.84065297e-01 -8.39343481e-03 -9.97583270e-01
-6.13975406e-01 -4.71794158e-01 -1.80975825e-01 -7.50530243e-01
-8.83039474e-01 -5.52047670e-01 3.61711651e-01 -9.47434545e-01
1.54731441e+00 -1.46725297e+00 1.05710752e-01 -4.73493427e-01
-1.69413965e-02 3.58236283e-01 -6.01304889e-01 1.04247117e+00
1.33706942e-01 1.01301156e-01 -4.53098565e-01 -6.06897652e-01
2.59398878e-01 3.60307433e-02 -7.44283080e-01 -4.36984330e-01
6.66172028e-01 1.56088805e+00 -1.25216866e+00 -4.92210925e-01
-3.83691102e-01 1.19083688e-01 -7.66812801e-01 6.90449059e-01
-7.79903114e-01 2.88724899e-01 -2.04597816e-01 3.07950675e-01
1.32979423e-01 -3.00489932e-01 -9.15358737e-02 1.60886496e-01
4.98885252e-02 1.02844763e+00 -4.35631901e-01 2.13214898e+00
-6.03944302e-01 4.66848612e-01 -4.59877908e-01 -6.34946823e-01
9.68711376e-01 5.78958392e-01 -3.69093150e-01 -7.31950104e-01
-4.26312387e-02 1.04181305e-01 6.79138079e-02 -5.16187489e-01
1.27165592e+00 -2.23145723e-01 -3.04166853e-01 9.08871591e-01
4.72664624e-01 -3.24781686e-01 7.20942795e-01 4.77414459e-01
1.07992065e+00 2.70993710e-01 3.33064824e-01 -5.98239079e-02
2.34731570e-01 -1.85580812e-02 2.37114534e-01 9.03264165e-01
3.30659717e-01 9.36002195e-01 4.88154799e-01 4.16434780e-02
-1.22661018e+00 -1.02570760e+00 4.79035705e-01 1.36545384e+00
-3.57372522e-01 -6.05216861e-01 -1.26959360e+00 -7.27585256e-01
-2.65674770e-01 1.40880620e+00 -5.35053670e-01 -4.55139488e-01
-8.44956815e-01 -6.71421468e-01 1.13294101e+00 6.56350255e-01
6.16786838e-01 -1.62763107e+00 -4.46217954e-01 6.87325299e-01
-7.20088661e-01 -8.81944954e-01 -9.17436302e-01 -3.44354659e-01
-8.20153773e-01 -5.48505604e-01 -8.49595964e-01 -6.86896026e-01
5.18660009e-01 -2.54108191e-01 1.79195404e+00 2.72104349e-02
-5.34649901e-02 -5.38015403e-02 -6.14784360e-01 -3.15214008e-01
-1.01675165e+00 8.01235855e-01 -5.48642278e-01 -3.29272956e-01
-3.07205264e-02 -5.31065285e-01 -6.91916645e-01 -6.50027618e-02
-8.83063257e-01 6.71146095e-01 7.44718671e-01 1.17375016e+00
1.24005198e-01 -8.38776886e-01 1.30782068e+00 -1.02839220e+00
1.46702123e+00 -5.66515028e-01 -7.01131895e-02 4.32740897e-01
-5.39395750e-01 3.04599762e-01 8.20790648e-01 -3.70202869e-01
-1.39357758e+00 -6.80265009e-01 -4.13521230e-01 1.59275144e-01
4.13151495e-02 5.95433533e-01 -1.26428351e-01 1.00109613e+00
9.28979218e-01 6.08374178e-01 -2.64777038e-02 -3.93297404e-01
9.99672055e-01 6.31284833e-01 8.53318214e-01 -6.46741807e-01
5.88110387e-01 -2.54471838e-01 -5.86773574e-01 -4.09754515e-01
-1.05324483e+00 1.13025539e-01 -2.50670254e-01 -3.20497751e-02
9.06431139e-01 -8.61511350e-01 -2.44522199e-01 3.74013871e-01
-1.63670707e+00 -7.59142578e-01 -7.94842422e-01 -6.56006783e-02
-6.64520383e-01 2.65487522e-01 -1.01109147e+00 -7.33873546e-01
-1.20800650e+00 -5.94607174e-01 1.17244148e+00 1.56042352e-01
-7.63967574e-01 -1.00239289e+00 2.61899531e-01 5.41526020e-01
6.92772329e-01 1.68135986e-01 9.55679774e-01 -7.30644107e-01
-5.18708646e-01 1.39521047e-01 -8.78603086e-02 4.98455673e-01
-2.63358895e-02 -3.68602991e-01 -7.36118495e-01 -1.17079038e-02
-2.56113589e-01 -6.55161798e-01 1.04421914e+00 -6.21745639e-05
1.09516788e+00 -8.45433354e-01 1.17631808e-01 3.66095275e-01
8.43724132e-01 -5.08512668e-02 9.47658241e-01 -1.00406662e-01
6.76923752e-01 6.62331045e-01 2.96105593e-01 4.58996087e-01
7.66447127e-01 3.99273694e-01 5.61058261e-02 -1.15519669e-02
-5.61257660e-01 -9.08894122e-01 5.29448986e-01 1.12515914e+00
2.60054208e-02 -8.68857503e-01 -6.12024605e-01 7.59160161e-01
-1.77380157e+00 -1.24919116e+00 -1.64735898e-01 1.66288269e+00
1.58943260e+00 -1.06711149e-01 1.50022537e-01 -2.06488520e-01
6.25053108e-01 3.37289512e-01 -4.25503790e-01 -6.05184674e-01
-1.20872773e-01 4.57762629e-01 -1.55402739e-02 6.28658712e-01
-5.13247192e-01 1.17770135e+00 6.33326960e+00 1.04121900e+00
-7.52286017e-01 1.92107767e-01 9.12352622e-01 -7.31121823e-02
-6.67434573e-01 -8.46340284e-02 -1.01109672e+00 6.96678698e-01
1.12355113e+00 -5.02156794e-01 3.91428202e-01 6.31994188e-01
1.88716248e-01 1.76922292e-01 -1.28532350e+00 4.86892402e-01
3.14777911e-01 -1.63905799e+00 5.09247899e-01 -1.94893762e-01
8.65134597e-01 -1.94559410e-01 -2.08081245e-01 7.23825812e-01
6.94595218e-01 -1.18712366e+00 8.80473793e-01 6.74283683e-01
8.85794342e-01 -5.89541554e-01 5.16920865e-01 6.29464269e-01
-7.17418492e-01 1.10090815e-01 -3.40600312e-01 -2.19240829e-01
9.19014394e-01 6.80865228e-01 -1.26548684e+00 7.26714075e-01
-1.00683108e-01 3.01428944e-01 -4.04316247e-01 5.96593559e-01
-5.97820461e-01 9.11856294e-01 1.80124283e-01 -2.89859504e-01
2.78236210e-01 3.38263810e-02 4.00626570e-01 1.54559839e+00
5.80510795e-01 8.93763751e-02 -1.80337563e-01 1.26553786e+00
-5.28261602e-01 -1.45407125e-01 -2.14566112e-01 -2.11321518e-01
5.65345168e-01 1.07342613e+00 -8.06236118e-02 -6.73704326e-01
4.15896773e-02 1.28723967e+00 4.55366611e-01 3.09655339e-01
-7.48981476e-01 -6.44203067e-01 2.77935445e-01 1.94098547e-01
2.90509999e-01 -8.71014372e-02 -2.92435318e-01 -1.10861599e+00
-1.57405823e-01 -1.11201382e+00 3.10436100e-01 -8.51260066e-01
-1.45539796e+00 8.21789742e-01 5.31798862e-02 -7.84233272e-01
-1.11063421e+00 -8.89898315e-02 -1.00848317e+00 1.15835297e+00
-1.44658101e+00 -1.29923165e+00 -1.80841312e-01 1.99261740e-01
1.12419009e+00 -3.09734136e-01 7.76737750e-01 -3.09136529e-02
-3.20081502e-01 8.21345329e-01 -3.14354450e-01 2.43983075e-01
7.82034039e-01 -1.38195395e+00 1.34325027e+00 9.22064900e-01
2.88199051e-03 5.52895606e-01 4.40399021e-01 -7.06394196e-01
-9.71772492e-01 -1.43351638e+00 1.41973281e+00 -6.12848163e-01
3.71766567e-01 -5.83100200e-01 -7.25897372e-01 5.65891206e-01
9.63381767e-01 -6.00668490e-01 6.73934162e-01 1.82883125e-02
-2.84849346e-01 2.74760336e-01 -6.80836141e-01 7.56799698e-01
1.28220856e+00 -4.02631819e-01 -9.20814157e-01 2.89693445e-01
1.25675690e+00 -6.47054732e-01 -6.13844752e-01 8.35604742e-02
2.70100087e-01 -8.36869180e-01 6.97313607e-01 -8.09748709e-01
1.31069863e+00 3.28997672e-02 2.59054422e-01 -1.90948606e+00
-2.62869567e-01 -1.16718102e+00 -4.84378964e-01 1.57163596e+00
8.73402953e-01 -4.46702719e-01 5.02706110e-01 1.55571580e-01
-7.50252545e-01 -8.75431716e-01 -5.78595102e-01 -6.33569896e-01
3.65986019e-01 -5.37014902e-02 1.03942978e+00 6.26315117e-01
5.52995540e-02 1.11467564e+00 -4.51716661e-01 -5.90446055e-01
1.79578707e-01 9.97756347e-02 9.28655088e-01 -5.94392717e-01
-7.11103737e-01 -4.80993599e-01 4.89435732e-01 -1.55795348e+00
1.03502385e-01 -1.05712903e+00 3.67933989e-01 -1.94881821e+00
-2.12069862e-02 3.67991216e-02 4.09983128e-01 3.64595145e-01
-7.95164764e-01 8.56110007e-02 1.86560363e-01 -1.50192663e-01
-6.06306732e-01 1.05251443e+00 1.61363900e+00 6.32585287e-02
-4.22243122e-03 -1.25822932e-01 -1.25547445e+00 5.62538616e-02
7.18732297e-01 -1.79843590e-01 -6.15648210e-01 -8.46070826e-01
3.71808290e-01 2.80299276e-01 2.12942258e-01 -7.41973579e-01
1.20850496e-01 9.89408568e-02 1.82064563e-01 -5.34457445e-01
1.24566913e-01 4.09317642e-01 -2.22621545e-01 2.85650939e-01
-9.94547725e-01 3.48584443e-01 3.48692061e-03 3.18254471e-01
-2.04947039e-01 -3.88511658e-01 4.23941046e-01 -4.08179611e-01
-2.62234986e-01 1.31220534e-01 -3.97297591e-01 9.36982036e-01
4.36645538e-01 -6.75844401e-02 -7.95899510e-01 -7.87269592e-01
-2.30777457e-01 4.69203621e-01 -7.19209090e-02 5.95752656e-01
5.19349813e-01 -1.37775862e+00 -1.36821687e+00 -4.13881242e-03
1.02743991e-01 3.50084066e-01 3.16157103e-01 3.42362195e-01
-3.95767033e-01 4.69488621e-01 1.83747604e-01 -7.99838640e-03
-7.50858843e-01 3.98971848e-02 1.52226076e-01 -9.27591860e-01
-4.91517305e-01 1.15975511e+00 8.18725675e-02 -6.56347811e-01
-1.25821874e-01 -3.46377045e-01 1.23211913e-01 1.07325368e-01
6.98007703e-01 4.66591328e-01 -3.40154842e-02 -8.88018757e-02
3.27054441e-01 -3.96359384e-01 -2.64703751e-01 -4.48544681e-01
9.83941078e-01 1.28693106e-02 -6.45932183e-02 1.82512522e-01
8.44987750e-01 -2.71041185e-01 -1.30681515e+00 -1.48517922e-01
-1.03453264e-01 3.08975112e-02 -4.30958837e-01 -1.29037392e+00
-5.70028901e-01 9.17441189e-01 -3.62251431e-01 2.69440264e-01
7.83028066e-01 -7.43042082e-02 1.46329463e+00 3.61302644e-01
-4.20920290e-02 -1.23480499e+00 5.72228968e-01 1.04459083e+00
1.43711472e+00 -7.76182711e-01 -4.23725188e-01 -6.29927143e-02
-1.02882135e+00 8.63278389e-01 9.37476993e-01 -8.90898332e-02
-3.67439613e-02 -1.24443248e-02 -1.35370240e-01 1.42910317e-01
-1.33628845e+00 -8.14967975e-02 3.15049022e-01 5.01419187e-01
8.17968249e-01 -4.72751400e-03 -4.26486671e-01 8.34761739e-01
-1.12870526e+00 -6.29793331e-02 4.68802780e-01 5.50242424e-01
-3.58016551e-01 -1.19430101e+00 4.16030660e-02 4.65102881e-01
-3.04074615e-01 -7.74359286e-01 -4.10046488e-01 5.66800177e-01
-1.43980101e-01 9.53413010e-01 2.28148699e-01 -1.59677818e-01
3.37840885e-01 4.51173693e-01 4.91322190e-01 -9.07216847e-01
-1.11431086e+00 -2.29060054e-01 7.19073772e-01 -2.60017961e-01
1.10411242e-01 -3.02378953e-01 -1.02249420e+00 -2.54913032e-01
-2.92234093e-01 3.27980220e-01 2.62589991e-01 9.03379977e-01
7.44858563e-01 9.69898760e-01 3.71364921e-01 -6.45214260e-01
-1.04126155e+00 -1.63187027e+00 -8.97594169e-02 4.02798563e-01
1.12615548e-01 1.78684741e-01 -3.93205881e-02 2.25967303e-01]
|
[11.930377960205078, 9.041038513183594]
|
e5c03297-e6af-4010-9fba-35d66ca0db24
|
an-effective-automatic-image-annotation-model
|
2001.10590
| null |
https://arxiv.org/abs/2001.10590v1
|
https://arxiv.org/pdf/2001.10590v1.pdf
|
An Effective Automatic Image Annotation Model Via Attention Model and Data Equilibrium
|
Nowadays, a huge number of images are available. However, retrieving a required image for an ordinary user is a challenging task in computer vision systems. During the past two decades, many types of research have been introduced to improve the performance of the automatic annotation of images, which are traditionally focused on content-based image retrieval. Although, recent research demonstrates that there is a semantic gap between content-based image retrieval and image semantics understandable by humans. As a result, existing research in this area has caused to bridge the semantic gap between low-level image features and high-level semantics. The conventional method of bridging the semantic gap is through the automatic image annotation (AIA) that extracts semantic features using machine learning techniques. In this paper, we propose a novel AIA model based on the deep learning feature extraction method. The proposed model has three phases, including a feature extractor, a tag generator, and an image annotator. First, the proposed model extracts automatically the high and low-level features based on dual-tree continues wavelet transform (DT-CWT), singular value decomposition, distribution of color ton, and the deep neural network. Moreover, the tag generator balances the dictionary of the annotated keywords by a new log-entropy auto-encoder (LEAE) and then describes these keywords by word embedding. Finally, the annotator works based on the long-short-term memory (LSTM) network in order to obtain the importance degree of specific features of the image. The experiments conducted on two benchmark datasets confirm that the superiority of the proposed model compared to the previous models in terms of performance criteria.
|
['Mostafa Rahimi', 'Milad Taleby Ahvanooey', 'Amir Vatani']
|
2020-01-26
| null | null | null | null |
['content-based-image-retrieval']
|
['computer-vision']
|
[ 3.21171671e-01 -2.02853978e-01 -2.47117057e-02 -3.61435354e-01
-5.16316712e-01 -8.94131064e-02 5.28992295e-01 2.92047143e-01
-7.06269145e-01 2.71748453e-01 -1.28796622e-01 9.75461397e-03
-3.37729193e-02 -8.66568029e-01 -3.86681050e-01 -8.39489102e-01
2.43025154e-01 9.83249545e-02 4.19557333e-01 1.45616094e-02
4.20087248e-01 2.46259689e-01 -1.77855289e+00 2.73330271e-01
7.09145367e-01 1.57533979e+00 5.58442831e-01 9.51538906e-02
-5.52012563e-01 7.03824937e-01 -4.64956492e-01 -3.47968042e-01
3.33485864e-02 -4.48914051e-01 -8.11700225e-01 1.79576740e-01
-3.49907726e-02 -3.28694671e-01 -3.46228749e-01 1.36027932e+00
4.67219830e-01 -6.04938306e-02 5.24505436e-01 -1.25455022e+00
-8.33024800e-01 2.88044602e-01 -4.06154394e-01 4.33440730e-02
7.27225691e-02 -8.46929178e-02 9.42433417e-01 -8.97223115e-01
3.80922556e-01 1.19141054e+00 4.94590372e-01 2.14063033e-01
-6.00933671e-01 -5.86534202e-01 -6.91934079e-02 6.84484959e-01
-1.61236501e+00 1.06212817e-01 9.36365366e-01 -5.87783396e-01
6.45456791e-01 1.40435249e-03 7.70267189e-01 7.08780706e-01
1.50596544e-01 6.86954677e-01 1.09596121e+00 -4.83102977e-01
4.56394292e-02 4.05411363e-01 7.60008097e-02 8.27008247e-01
1.55082136e-01 -2.54812807e-01 -2.75290042e-01 1.10399304e-02
4.84598249e-01 2.29998603e-01 -2.23274469e-01 -3.23261648e-01
-1.04740441e+00 8.76528323e-01 4.67705965e-01 7.19285965e-01
-4.92842793e-01 -6.53077215e-02 7.46616304e-01 -7.16772452e-02
3.75838995e-01 8.92501995e-02 -2.92778522e-01 1.50279269e-01
-9.91456389e-01 -1.52396679e-01 6.19652987e-01 8.08869600e-01
8.02666426e-01 -9.74311158e-02 -1.47253677e-01 7.77369499e-01
4.96319592e-01 4.82067317e-01 1.00953877e+00 -4.87474591e-01
9.01418477e-02 8.92073989e-01 -9.58890766e-02 -1.68987143e+00
-1.06564768e-01 -4.44012642e-01 -9.84608591e-01 -1.05537161e-01
-1.51050866e-01 3.21251839e-01 -8.57172132e-01 1.49255264e+00
2.07709402e-01 2.19160020e-02 -7.52942218e-03 1.09023988e+00
9.36209679e-01 9.05890346e-01 3.09625059e-01 -1.50832057e-01
1.73176837e+00 -9.30845320e-01 -1.05561244e+00 -8.15289989e-02
4.02188361e-01 -8.95992935e-01 1.08271027e+00 1.58343554e-01
-7.26215243e-01 -9.21591759e-01 -1.14660585e+00 -2.64846623e-01
-8.65547597e-01 5.72409749e-01 2.96956658e-01 2.98240691e-01
-9.15587604e-01 3.57263774e-01 -4.18169081e-01 -4.99378741e-01
4.01488602e-01 2.67929137e-01 -3.26646477e-01 1.82415042e-02
-1.56018567e+00 8.18099022e-01 9.97688949e-01 3.14827204e-01
-5.72229803e-01 -7.89576173e-02 -9.08173621e-01 2.33732224e-01
3.25364053e-01 -4.69428420e-01 8.70716810e-01 -1.36648858e+00
-1.20130062e+00 1.03941393e+00 4.44230177e-02 -3.79768133e-01
1.73919514e-01 -8.54728520e-02 -5.30784607e-01 4.69167620e-01
2.04911262e-01 7.19708562e-01 8.55498254e-01 -1.18399930e+00
-9.82369244e-01 -4.30693656e-01 -1.45450577e-01 2.20318899e-01
-9.36992109e-01 -6.35663569e-02 -6.26871943e-01 -6.51816189e-01
-2.64205150e-02 -7.13085771e-01 2.17489406e-01 6.55156896e-02
-1.95601374e-01 -4.72119719e-01 1.05492198e+00 -8.59548748e-01
1.33410287e+00 -2.35927224e+00 -1.49802696e-02 2.67691970e-01
-1.42869903e-02 5.87665439e-01 -6.47547394e-02 3.76315653e-01
3.21276858e-02 2.09770665e-01 -2.69049376e-01 5.31115569e-02
-6.80627301e-02 2.57307440e-01 -3.06570083e-01 1.35159582e-01
1.31690532e-01 7.15502262e-01 -8.30522180e-01 -1.02701747e+00
2.72469163e-01 5.33472538e-01 -1.62653655e-01 3.49819720e-01
-6.60289302e-02 2.01904327e-01 -7.43873954e-01 4.91391063e-01
5.91605008e-01 -2.88318843e-01 -4.41131555e-02 -6.45664155e-01
-1.80073634e-01 -3.05924624e-01 -1.01486528e+00 1.74297071e+00
-4.06904548e-01 3.52290660e-01 -2.37865344e-01 -1.29930460e+00
1.03129971e+00 3.82505566e-01 6.54979467e-01 -7.11454213e-01
4.63400424e-01 5.13009608e-01 -1.75809383e-01 -9.97213840e-01
2.96207964e-01 9.63948816e-02 2.89513413e-02 1.64445236e-01
1.49452180e-01 1.11053012e-01 2.29286969e-01 3.00514325e-02
6.44736409e-01 1.65682703e-01 2.74796784e-01 -7.01212138e-02
1.07689238e+00 2.00330049e-01 4.39097226e-01 2.97261983e-01
-4.53564897e-02 3.23766738e-01 6.61315769e-02 -6.35851920e-01
-1.16794264e+00 -4.16746080e-01 -9.04469341e-02 7.51675844e-01
5.46579421e-01 -3.32431674e-01 -9.01338160e-01 -5.12241602e-01
-1.20054096e-01 3.53687525e-01 -4.74360108e-01 -4.96986330e-01
-1.94008768e-01 -5.51708996e-01 4.20968443e-01 2.33857036e-01
1.24143565e+00 -1.26446998e+00 -7.83893943e-01 5.53703532e-02
-4.21940893e-01 -1.26787937e+00 -3.93299431e-01 -1.52810991e-01
-5.67706704e-01 -9.09650624e-01 -7.28678524e-01 -1.16942143e+00
7.17213035e-01 2.96290040e-01 6.64557338e-01 4.04502571e-01
-4.99770135e-01 2.06873804e-01 -7.50855982e-01 -3.48698944e-01
-1.10821515e-01 2.18378514e-01 -3.31627309e-01 3.39849710e-01
7.34116197e-01 -2.10019454e-01 -7.25950420e-01 2.08173886e-01
-1.29785049e+00 1.82460338e-01 1.08498609e+00 8.44927311e-01
7.61200786e-01 3.81088495e-01 4.25579399e-01 -5.27824402e-01
6.01904929e-01 -4.44334805e-01 -6.23098135e-01 3.09124529e-01
-7.85479903e-01 1.89527292e-02 6.82134211e-01 -2.78315097e-01
-1.01684332e+00 7.47944564e-02 -1.82133988e-01 -3.57369483e-01
-6.09538108e-02 8.52068126e-01 -2.84386069e-01 -1.15504547e-03
1.40027404e-01 7.55433083e-01 6.32376373e-02 -3.85135293e-01
2.34791100e-01 1.09894490e+00 3.39012295e-01 -2.41907194e-01
6.86087132e-01 3.83883834e-01 -1.78946957e-01 -6.05827987e-01
-9.76271391e-01 -5.67949772e-01 -5.86424649e-01 -2.29544863e-01
1.25490797e+00 -8.19059670e-01 -7.90899158e-01 6.13898993e-01
-1.26893830e+00 3.24021697e-01 -7.09718615e-02 6.17205679e-01
-3.96042854e-01 4.60894942e-01 -4.25925404e-01 -5.47575295e-01
-6.34156227e-01 -1.24358249e+00 1.08334160e+00 5.15041292e-01
2.21872598e-01 -7.08683968e-01 -3.91350478e-01 3.41477424e-01
3.98509562e-01 5.40403388e-02 1.11271322e+00 -7.83619940e-01
-4.96414036e-01 -2.86591649e-01 -5.89103639e-01 6.44378424e-01
9.28686261e-02 -2.92179942e-01 -8.01127493e-01 -1.23234726e-01
1.98767886e-01 -4.06850934e-01 8.29094827e-01 1.67781681e-01
1.63949096e+00 -3.77854407e-01 -2.73127079e-01 4.01416898e-01
1.72933817e+00 5.81803143e-01 6.16358697e-01 5.40772974e-01
6.36632681e-01 5.36097646e-01 7.17391729e-01 3.64759564e-01
4.71191555e-01 5.11436343e-01 4.86604661e-01 -1.23452589e-01
-6.60362914e-02 -2.97789812e-01 4.34555262e-02 1.22086191e+00
-3.78650776e-03 -6.43193573e-02 -8.33661258e-01 6.24358654e-01
-1.81323397e+00 -8.07794213e-01 1.02532953e-01 1.96309257e+00
1.02103400e+00 1.42711699e-01 -3.05568814e-01 3.29623401e-01
8.28806818e-01 8.45396966e-02 -5.22730768e-01 -2.27488860e-01
5.47916517e-02 -1.33678630e-01 4.71470505e-01 6.04570247e-02
-1.17827857e+00 8.76385450e-01 4.37605572e+00 1.17743182e+00
-1.23204732e+00 1.17416508e-01 5.81388056e-01 6.06864095e-01
-1.07603826e-01 -1.30583241e-03 -6.55519962e-01 7.34774411e-01
5.23978770e-01 -1.49328694e-01 2.53159881e-01 9.21414912e-01
3.18606682e-02 -3.08650862e-02 -6.96780145e-01 1.24642730e+00
4.39331442e-01 -9.84867394e-01 4.26175356e-01 -3.61717343e-02
3.64820331e-01 -3.45724910e-01 1.01740018e-01 2.31261328e-01
-4.02576268e-01 -7.45787919e-01 6.49072528e-01 7.38684535e-01
6.72174037e-01 -7.68840611e-01 1.28811789e+00 3.84091705e-01
-1.29491532e+00 -1.61203921e-01 -5.55778205e-01 3.34396362e-01
-2.34698653e-01 6.35132313e-01 -4.73677337e-01 6.91799343e-01
8.78955126e-01 6.61665142e-01 -6.50919199e-01 9.41452444e-01
-1.98613882e-01 3.39807063e-01 3.73950675e-02 -9.45306569e-02
3.78542989e-01 -3.11403841e-01 1.81160808e-01 1.16639507e+00
4.49650228e-01 1.91673085e-01 2.34439954e-01 7.54192472e-01
-1.60160124e-01 5.73564708e-01 -5.26800692e-01 -3.09933990e-01
3.36858571e-01 1.55500197e+00 -8.41259718e-01 -5.41817665e-01
-3.91496897e-01 1.12575984e+00 -1.86938662e-02 2.33108431e-01
-7.95149803e-01 -8.69066358e-01 -1.25734940e-01 -1.08217172e-01
3.82578611e-01 -5.13267405e-02 9.38875154e-02 -1.00475919e+00
2.17227161e-01 -8.31330657e-01 3.15295011e-01 -9.63545799e-01
-1.19820094e+00 6.25934482e-01 -1.00091949e-01 -1.26843441e+00
2.07337737e-02 -5.45677066e-01 -1.15577951e-01 8.98273051e-01
-1.85433543e+00 -1.39241672e+00 -6.41577721e-01 6.53898954e-01
6.91907048e-01 -1.82454199e-01 8.15300703e-01 6.29165053e-01
-3.45424920e-01 1.85164362e-01 5.42735532e-02 5.06604731e-01
5.64572990e-01 -8.63514364e-01 -3.13046694e-01 5.80089509e-01
5.00115566e-02 4.14680868e-01 3.52862448e-01 -5.62986791e-01
-1.22247410e+00 -9.97321904e-01 1.06808794e+00 3.75650227e-01
4.17884946e-01 -9.23870653e-02 -9.24422264e-01 3.38067621e-01
3.93405795e-01 2.33368665e-01 5.53846180e-01 -5.61094344e-01
-1.64825723e-01 -3.71001303e-01 -1.05039668e+00 2.38564312e-01
5.04286170e-01 -5.86358011e-01 -6.09898210e-01 2.13176340e-01
6.52016222e-01 7.23427301e-03 -8.90556097e-01 4.98284489e-01
5.73195636e-01 -6.35338247e-01 8.38386953e-01 -2.31621280e-01
5.35297275e-01 -4.70743805e-01 -1.14819065e-01 -1.05268526e+00
-1.30604625e-01 1.72266945e-01 2.73123115e-01 1.29173195e+00
-6.96174800e-02 -4.35307592e-01 3.08456689e-01 2.11875468e-01
4.54849787e-02 -8.37446272e-01 -5.72830200e-01 -5.22673190e-01
-3.43497962e-01 -3.51418345e-03 5.25376558e-01 8.79236460e-01
-2.71314710e-01 3.65280837e-01 -2.26252034e-01 3.92228970e-03
5.01487613e-01 2.70687491e-01 3.40874016e-01 -1.41491735e+00
1.57881781e-01 -4.14615929e-01 -6.80915833e-01 -7.74932802e-01
2.62685210e-01 -9.42473531e-01 -7.17202621e-03 -1.70328629e+00
4.94635344e-01 -3.07948768e-01 -5.92751324e-01 5.65195799e-01
-1.35285750e-01 2.14012772e-01 1.24254405e-01 4.55924690e-01
-6.73424423e-01 6.59830570e-01 1.22840309e+00 -4.00863081e-01
2.73077071e-01 -3.53022665e-01 -3.75004053e-01 8.74932647e-01
8.11959803e-01 -5.15890062e-01 -5.03973365e-01 -4.92929250e-01
2.86485016e-01 -1.97446167e-01 4.50206876e-01 -1.01381171e+00
3.90718132e-01 4.71063331e-02 4.58480567e-01 -5.91112077e-01
1.65150255e-01 -1.25047588e+00 -2.35291347e-01 5.62659979e-01
-3.64799380e-01 3.23783159e-02 4.29537818e-02 4.73576546e-01
-7.68336535e-01 -5.10578036e-01 6.21646821e-01 -3.15381676e-01
-1.09876180e+00 3.57133180e-01 -2.07902819e-01 -4.81817797e-02
1.09084654e+00 -1.88009158e-01 -5.15658315e-03 -1.90767348e-01
-4.51018631e-01 7.67121762e-02 7.56497160e-02 4.45468873e-01
8.71527433e-01 -1.46067715e+00 -4.35671568e-01 1.07719570e-01
3.88570905e-01 -1.63597763e-01 2.64400721e-01 6.95267618e-01
-6.42781734e-01 4.45833504e-01 -3.52577746e-01 -4.48576212e-01
-1.24633765e+00 6.42820895e-01 1.26376957e-01 -2.07272306e-01
-3.95119578e-01 4.70831811e-01 -1.19243748e-01 3.15370187e-02
2.21904159e-01 -7.19882846e-02 -6.70382977e-01 3.31410944e-01
3.68589967e-01 -8.61429200e-02 -7.20073655e-02 -1.00115263e+00
-2.25025743e-01 8.61601472e-01 1.10678412e-02 6.76205242e-03
1.17455685e+00 -2.81868458e-01 -5.85245073e-01 4.25363153e-01
1.54526579e+00 -4.88666415e-01 -6.48261547e-01 -4.39584821e-01
6.30451292e-02 -4.29209113e-01 3.12257081e-01 -6.55916274e-01
-1.23082626e+00 1.21059370e+00 9.28713083e-01 3.10291290e-01
1.39899993e+00 -2.36199304e-01 1.18661416e+00 3.77618551e-01
2.20588222e-01 -1.39186442e+00 2.04100952e-01 2.00339079e-01
6.66974604e-01 -1.24188280e+00 -3.86105068e-02 -1.44847631e-01
-4.45833385e-01 1.21011710e+00 5.41001797e-01 -5.88776581e-02
8.08566570e-01 -1.97983950e-01 -3.65871117e-02 -2.44574070e-01
-1.35751560e-01 -3.78850490e-01 3.81011605e-01 1.69087544e-01
1.97447285e-01 -2.81622916e-01 -7.54995227e-01 7.45176196e-01
2.31380034e-02 2.13297591e-01 1.78559218e-02 8.57415915e-01
-6.62086248e-01 -1.08166611e+00 -2.94817388e-01 2.37876147e-01
-5.63426793e-01 -1.30520448e-01 -1.23396508e-01 6.66438460e-01
5.65271378e-01 8.64447832e-01 5.24378717e-02 -4.64607775e-01
-2.35879347e-02 1.18754104e-01 1.83698505e-01 -3.45877290e-01
-2.37998590e-01 -2.96133608e-02 -4.39063579e-01 -2.79126257e-01
-8.08974624e-01 -1.76756248e-01 -1.40245438e+00 3.00494909e-01
-4.11011130e-01 3.92036557e-01 1.09463179e+00 1.00223982e+00
2.97643363e-01 5.46118021e-01 8.06780636e-01 -4.08433467e-01
-3.79941970e-01 -1.02883506e+00 -3.89330089e-01 7.74060786e-01
-7.32746422e-02 -6.23674333e-01 -1.99180156e-01 4.30114806e-01]
|
[11.050382614135742, 1.5528103113174438]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.